{ "best_metric": null, "best_model_checkpoint": null, "epoch": 100.0, "eval_steps": 200.0, "global_step": 40000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0025, "grad_norm": 46.75, "learning_rate": 1.666666666666667e-08, "loss": 10.0992, "step": 1 }, { "epoch": 0.005, "grad_norm": 49.5, "learning_rate": 3.333333333333334e-08, "loss": 10.2017, "step": 2 }, { "epoch": 0.0075, "grad_norm": 52.75, "learning_rate": 5.0000000000000004e-08, "loss": 8.9451, "step": 3 }, { "epoch": 0.01, "grad_norm": 37.75, "learning_rate": 6.666666666666668e-08, "loss": 9.075, "step": 4 }, { "epoch": 0.0125, "grad_norm": 46.25, "learning_rate": 8.333333333333334e-08, "loss": 10.078, "step": 5 }, { "epoch": 0.015, "grad_norm": 70.0, "learning_rate": 1.0000000000000001e-07, "loss": 9.855, "step": 6 }, { "epoch": 0.0175, "grad_norm": 51.75, "learning_rate": 1.1666666666666668e-07, "loss": 9.7825, "step": 7 }, { "epoch": 0.02, "grad_norm": 41.5, "learning_rate": 1.3333333333333336e-07, "loss": 9.2797, "step": 8 }, { "epoch": 0.0225, "grad_norm": 61.5, "learning_rate": 1.5000000000000002e-07, "loss": 9.9084, "step": 9 }, { "epoch": 0.025, "grad_norm": 42.75, "learning_rate": 1.6666666666666668e-07, "loss": 9.5687, "step": 10 }, { "epoch": 0.0275, "grad_norm": 164.0, "learning_rate": 1.8333333333333336e-07, "loss": 9.6469, "step": 11 }, { "epoch": 0.03, "grad_norm": 61.5, "learning_rate": 2.0000000000000002e-07, "loss": 9.2414, "step": 12 }, { "epoch": 0.0325, "grad_norm": 44.5, "learning_rate": 2.166666666666667e-07, "loss": 9.8255, "step": 13 }, { "epoch": 0.035, "grad_norm": 46.25, "learning_rate": 2.3333333333333336e-07, "loss": 9.709, "step": 14 }, { "epoch": 0.0375, "grad_norm": 43.75, "learning_rate": 2.5000000000000004e-07, "loss": 9.6571, "step": 15 }, { "epoch": 0.04, "grad_norm": 47.25, "learning_rate": 2.666666666666667e-07, "loss": 9.9408, "step": 16 }, { "epoch": 0.0425, "grad_norm": 42.0, "learning_rate": 2.8333333333333336e-07, "loss": 9.5288, "step": 17 }, { "epoch": 0.045, "grad_norm": 97.0, "learning_rate": 3.0000000000000004e-07, "loss": 10.2495, "step": 18 }, { "epoch": 0.0475, "grad_norm": 45.0, "learning_rate": 3.166666666666667e-07, "loss": 10.1329, "step": 19 }, { "epoch": 0.05, "grad_norm": 47.0, "learning_rate": 3.3333333333333335e-07, "loss": 9.8122, "step": 20 }, { "epoch": 0.0525, "grad_norm": 45.0, "learning_rate": 3.5000000000000004e-07, "loss": 9.9312, "step": 21 }, { "epoch": 0.055, "grad_norm": 46.5, "learning_rate": 3.666666666666667e-07, "loss": 9.4744, "step": 22 }, { "epoch": 0.0575, "grad_norm": 42.5, "learning_rate": 3.8333333333333335e-07, "loss": 9.419, "step": 23 }, { "epoch": 0.06, "grad_norm": 49.5, "learning_rate": 4.0000000000000003e-07, "loss": 9.6554, "step": 24 }, { "epoch": 0.0625, "grad_norm": 63.5, "learning_rate": 4.1666666666666667e-07, "loss": 9.6787, "step": 25 }, { "epoch": 0.065, "grad_norm": 40.0, "learning_rate": 4.333333333333334e-07, "loss": 9.3478, "step": 26 }, { "epoch": 0.0675, "grad_norm": 42.75, "learning_rate": 4.5000000000000003e-07, "loss": 9.3593, "step": 27 }, { "epoch": 0.07, "grad_norm": 44.75, "learning_rate": 4.666666666666667e-07, "loss": 9.9515, "step": 28 }, { "epoch": 0.0725, "grad_norm": 75.5, "learning_rate": 4.833333333333334e-07, "loss": 10.0245, "step": 29 }, { "epoch": 0.075, "grad_norm": 42.0, "learning_rate": 5.000000000000001e-07, "loss": 9.9094, "step": 30 }, { "epoch": 0.0775, "grad_norm": 75.5, "learning_rate": 5.166666666666667e-07, "loss": 9.3547, "step": 31 }, { "epoch": 0.08, "grad_norm": 53.75, "learning_rate": 5.333333333333335e-07, "loss": 9.4604, "step": 32 }, { "epoch": 0.0825, "grad_norm": 43.75, "learning_rate": 5.5e-07, "loss": 9.3116, "step": 33 }, { "epoch": 0.085, "grad_norm": 49.25, "learning_rate": 5.666666666666667e-07, "loss": 10.6544, "step": 34 }, { "epoch": 0.0875, "grad_norm": 54.25, "learning_rate": 5.833333333333334e-07, "loss": 9.8437, "step": 35 }, { "epoch": 0.09, "grad_norm": 57.5, "learning_rate": 6.000000000000001e-07, "loss": 10.8177, "step": 36 }, { "epoch": 0.0925, "grad_norm": 105.5, "learning_rate": 6.166666666666668e-07, "loss": 9.916, "step": 37 }, { "epoch": 0.095, "grad_norm": 41.25, "learning_rate": 6.333333333333334e-07, "loss": 9.0587, "step": 38 }, { "epoch": 0.0975, "grad_norm": 63.0, "learning_rate": 6.5e-07, "loss": 10.6416, "step": 39 }, { "epoch": 0.1, "grad_norm": 49.75, "learning_rate": 6.666666666666667e-07, "loss": 9.913, "step": 40 }, { "epoch": 0.1025, "grad_norm": 62.75, "learning_rate": 6.833333333333334e-07, "loss": 9.89, "step": 41 }, { "epoch": 0.105, "grad_norm": 44.25, "learning_rate": 7.000000000000001e-07, "loss": 10.2584, "step": 42 }, { "epoch": 0.1075, "grad_norm": 43.75, "learning_rate": 7.166666666666668e-07, "loss": 9.7427, "step": 43 }, { "epoch": 0.11, "grad_norm": 50.0, "learning_rate": 7.333333333333334e-07, "loss": 10.0036, "step": 44 }, { "epoch": 0.1125, "grad_norm": 38.5, "learning_rate": 7.5e-07, "loss": 9.5495, "step": 45 }, { "epoch": 0.115, "grad_norm": 47.5, "learning_rate": 7.666666666666667e-07, "loss": 10.1639, "step": 46 }, { "epoch": 0.1175, "grad_norm": 42.75, "learning_rate": 7.833333333333335e-07, "loss": 9.6134, "step": 47 }, { "epoch": 0.12, "grad_norm": 40.25, "learning_rate": 8.000000000000001e-07, "loss": 9.6019, "step": 48 }, { "epoch": 0.1225, "grad_norm": 45.5, "learning_rate": 8.166666666666668e-07, "loss": 9.4489, "step": 49 }, { "epoch": 0.125, "grad_norm": 51.25, "learning_rate": 8.333333333333333e-07, "loss": 9.7871, "step": 50 }, { "epoch": 0.1275, "grad_norm": 51.5, "learning_rate": 8.500000000000001e-07, "loss": 10.2177, "step": 51 }, { "epoch": 0.13, "grad_norm": 247.0, "learning_rate": 8.666666666666668e-07, "loss": 9.4197, "step": 52 }, { "epoch": 0.1325, "grad_norm": 47.75, "learning_rate": 8.833333333333334e-07, "loss": 10.4213, "step": 53 }, { "epoch": 0.135, "grad_norm": 38.75, "learning_rate": 9.000000000000001e-07, "loss": 9.4324, "step": 54 }, { "epoch": 0.1375, "grad_norm": 44.25, "learning_rate": 9.166666666666666e-07, "loss": 9.7703, "step": 55 }, { "epoch": 0.14, "grad_norm": 42.25, "learning_rate": 9.333333333333334e-07, "loss": 9.4987, "step": 56 }, { "epoch": 0.1425, "grad_norm": 58.0, "learning_rate": 9.500000000000001e-07, "loss": 10.2849, "step": 57 }, { "epoch": 0.145, "grad_norm": 57.25, "learning_rate": 9.666666666666668e-07, "loss": 9.8987, "step": 58 }, { "epoch": 0.1475, "grad_norm": 56.0, "learning_rate": 9.833333333333334e-07, "loss": 9.8406, "step": 59 }, { "epoch": 0.15, "grad_norm": 43.75, "learning_rate": 1.0000000000000002e-06, "loss": 9.6352, "step": 60 }, { "epoch": 0.1525, "grad_norm": 45.0, "learning_rate": 1.0166666666666667e-06, "loss": 9.3892, "step": 61 }, { "epoch": 0.155, "grad_norm": 40.5, "learning_rate": 1.0333333333333333e-06, "loss": 8.6607, "step": 62 }, { "epoch": 0.1575, "grad_norm": 53.0, "learning_rate": 1.0500000000000001e-06, "loss": 9.7663, "step": 63 }, { "epoch": 0.16, "grad_norm": 63.75, "learning_rate": 1.066666666666667e-06, "loss": 9.4073, "step": 64 }, { "epoch": 0.1625, "grad_norm": 47.0, "learning_rate": 1.0833333333333335e-06, "loss": 9.7578, "step": 65 }, { "epoch": 0.165, "grad_norm": 45.0, "learning_rate": 1.1e-06, "loss": 10.5495, "step": 66 }, { "epoch": 0.1675, "grad_norm": 43.5, "learning_rate": 1.1166666666666666e-06, "loss": 9.6429, "step": 67 }, { "epoch": 0.17, "grad_norm": 36.75, "learning_rate": 1.1333333333333334e-06, "loss": 8.8594, "step": 68 }, { "epoch": 0.1725, "grad_norm": 68.5, "learning_rate": 1.1500000000000002e-06, "loss": 9.3274, "step": 69 }, { "epoch": 0.175, "grad_norm": 67.5, "learning_rate": 1.1666666666666668e-06, "loss": 9.4872, "step": 70 }, { "epoch": 0.1775, "grad_norm": 41.25, "learning_rate": 1.1833333333333334e-06, "loss": 9.298, "step": 71 }, { "epoch": 0.18, "grad_norm": 42.75, "learning_rate": 1.2000000000000002e-06, "loss": 9.7142, "step": 72 }, { "epoch": 0.1825, "grad_norm": 52.5, "learning_rate": 1.2166666666666667e-06, "loss": 9.9272, "step": 73 }, { "epoch": 0.185, "grad_norm": 40.25, "learning_rate": 1.2333333333333335e-06, "loss": 9.5084, "step": 74 }, { "epoch": 0.1875, "grad_norm": 44.0, "learning_rate": 1.25e-06, "loss": 8.6144, "step": 75 }, { "epoch": 0.19, "grad_norm": 38.0, "learning_rate": 1.2666666666666669e-06, "loss": 9.2034, "step": 76 }, { "epoch": 0.1925, "grad_norm": 48.75, "learning_rate": 1.2833333333333335e-06, "loss": 9.7196, "step": 77 }, { "epoch": 0.195, "grad_norm": 40.25, "learning_rate": 1.3e-06, "loss": 9.2503, "step": 78 }, { "epoch": 0.1975, "grad_norm": 39.75, "learning_rate": 1.3166666666666666e-06, "loss": 9.3654, "step": 79 }, { "epoch": 0.2, "grad_norm": 41.5, "learning_rate": 1.3333333333333334e-06, "loss": 9.3666, "step": 80 }, { "epoch": 0.2025, "grad_norm": 143.0, "learning_rate": 1.3500000000000002e-06, "loss": 10.0513, "step": 81 }, { "epoch": 0.205, "grad_norm": 40.25, "learning_rate": 1.3666666666666668e-06, "loss": 9.6241, "step": 82 }, { "epoch": 0.2075, "grad_norm": 43.25, "learning_rate": 1.3833333333333336e-06, "loss": 9.1153, "step": 83 }, { "epoch": 0.21, "grad_norm": 60.75, "learning_rate": 1.4000000000000001e-06, "loss": 9.0998, "step": 84 }, { "epoch": 0.2125, "grad_norm": 42.25, "learning_rate": 1.4166666666666667e-06, "loss": 9.1439, "step": 85 }, { "epoch": 0.215, "grad_norm": 43.25, "learning_rate": 1.4333333333333335e-06, "loss": 9.3441, "step": 86 }, { "epoch": 0.2175, "grad_norm": 48.0, "learning_rate": 1.45e-06, "loss": 9.168, "step": 87 }, { "epoch": 0.22, "grad_norm": 43.5, "learning_rate": 1.4666666666666669e-06, "loss": 9.6726, "step": 88 }, { "epoch": 0.2225, "grad_norm": 52.25, "learning_rate": 1.4833333333333337e-06, "loss": 9.1626, "step": 89 }, { "epoch": 0.225, "grad_norm": 46.5, "learning_rate": 1.5e-06, "loss": 9.8785, "step": 90 }, { "epoch": 0.2275, "grad_norm": 40.75, "learning_rate": 1.5166666666666668e-06, "loss": 9.5296, "step": 91 }, { "epoch": 0.23, "grad_norm": 37.25, "learning_rate": 1.5333333333333334e-06, "loss": 9.2457, "step": 92 }, { "epoch": 0.2325, "grad_norm": 40.25, "learning_rate": 1.5500000000000002e-06, "loss": 9.513, "step": 93 }, { "epoch": 0.235, "grad_norm": 38.5, "learning_rate": 1.566666666666667e-06, "loss": 8.7753, "step": 94 }, { "epoch": 0.2375, "grad_norm": 44.0, "learning_rate": 1.5833333333333333e-06, "loss": 9.7964, "step": 95 }, { "epoch": 0.24, "grad_norm": 127.0, "learning_rate": 1.6000000000000001e-06, "loss": 9.4092, "step": 96 }, { "epoch": 0.2425, "grad_norm": 43.25, "learning_rate": 1.6166666666666667e-06, "loss": 9.5658, "step": 97 }, { "epoch": 0.245, "grad_norm": 43.0, "learning_rate": 1.6333333333333335e-06, "loss": 9.3599, "step": 98 }, { "epoch": 0.2475, "grad_norm": 49.75, "learning_rate": 1.6500000000000003e-06, "loss": 9.2605, "step": 99 }, { "epoch": 0.25, "grad_norm": 53.0, "learning_rate": 1.6666666666666667e-06, "loss": 9.4063, "step": 100 }, { "epoch": 0.2525, "grad_norm": 42.25, "learning_rate": 1.6833333333333335e-06, "loss": 9.8047, "step": 101 }, { "epoch": 0.255, "grad_norm": 52.5, "learning_rate": 1.7000000000000002e-06, "loss": 9.3019, "step": 102 }, { "epoch": 0.2575, "grad_norm": 45.5, "learning_rate": 1.7166666666666668e-06, "loss": 9.8228, "step": 103 }, { "epoch": 0.26, "grad_norm": 41.0, "learning_rate": 1.7333333333333336e-06, "loss": 9.6539, "step": 104 }, { "epoch": 0.2625, "grad_norm": 40.5, "learning_rate": 1.75e-06, "loss": 9.4999, "step": 105 }, { "epoch": 0.265, "grad_norm": 42.75, "learning_rate": 1.7666666666666668e-06, "loss": 10.0124, "step": 106 }, { "epoch": 0.2675, "grad_norm": 71.0, "learning_rate": 1.7833333333333336e-06, "loss": 9.3583, "step": 107 }, { "epoch": 0.27, "grad_norm": 46.75, "learning_rate": 1.8000000000000001e-06, "loss": 9.095, "step": 108 }, { "epoch": 0.2725, "grad_norm": 39.0, "learning_rate": 1.816666666666667e-06, "loss": 8.423, "step": 109 }, { "epoch": 0.275, "grad_norm": 220.0, "learning_rate": 1.8333333333333333e-06, "loss": 8.9617, "step": 110 }, { "epoch": 0.2775, "grad_norm": 43.0, "learning_rate": 1.85e-06, "loss": 8.7274, "step": 111 }, { "epoch": 0.28, "grad_norm": 71.0, "learning_rate": 1.8666666666666669e-06, "loss": 9.4153, "step": 112 }, { "epoch": 0.2825, "grad_norm": 57.5, "learning_rate": 1.8833333333333334e-06, "loss": 9.3256, "step": 113 }, { "epoch": 0.285, "grad_norm": 35.75, "learning_rate": 1.9000000000000002e-06, "loss": 8.985, "step": 114 }, { "epoch": 0.2875, "grad_norm": 37.25, "learning_rate": 1.916666666666667e-06, "loss": 9.3457, "step": 115 }, { "epoch": 0.29, "grad_norm": 42.0, "learning_rate": 1.9333333333333336e-06, "loss": 9.6603, "step": 116 }, { "epoch": 0.2925, "grad_norm": 48.5, "learning_rate": 1.9500000000000004e-06, "loss": 9.1723, "step": 117 }, { "epoch": 0.295, "grad_norm": 45.5, "learning_rate": 1.9666666666666668e-06, "loss": 9.4646, "step": 118 }, { "epoch": 0.2975, "grad_norm": 42.75, "learning_rate": 1.9833333333333335e-06, "loss": 9.7551, "step": 119 }, { "epoch": 0.3, "grad_norm": 39.0, "learning_rate": 2.0000000000000003e-06, "loss": 8.831, "step": 120 }, { "epoch": 0.3025, "grad_norm": 43.25, "learning_rate": 2.0166666666666667e-06, "loss": 9.21, "step": 121 }, { "epoch": 0.305, "grad_norm": 118.0, "learning_rate": 2.0333333333333335e-06, "loss": 8.7874, "step": 122 }, { "epoch": 0.3075, "grad_norm": 44.25, "learning_rate": 2.05e-06, "loss": 9.732, "step": 123 }, { "epoch": 0.31, "grad_norm": 37.0, "learning_rate": 2.0666666666666666e-06, "loss": 9.0394, "step": 124 }, { "epoch": 0.3125, "grad_norm": 68.5, "learning_rate": 2.0833333333333334e-06, "loss": 9.4056, "step": 125 }, { "epoch": 0.315, "grad_norm": 41.0, "learning_rate": 2.1000000000000002e-06, "loss": 9.32, "step": 126 }, { "epoch": 0.3175, "grad_norm": 37.75, "learning_rate": 2.116666666666667e-06, "loss": 9.1429, "step": 127 }, { "epoch": 0.32, "grad_norm": 51.75, "learning_rate": 2.133333333333334e-06, "loss": 9.5337, "step": 128 }, { "epoch": 0.3225, "grad_norm": 39.5, "learning_rate": 2.15e-06, "loss": 9.3533, "step": 129 }, { "epoch": 0.325, "grad_norm": 37.75, "learning_rate": 2.166666666666667e-06, "loss": 9.1808, "step": 130 }, { "epoch": 0.3275, "grad_norm": 42.75, "learning_rate": 2.1833333333333333e-06, "loss": 9.0486, "step": 131 }, { "epoch": 0.33, "grad_norm": 33.25, "learning_rate": 2.2e-06, "loss": 8.4126, "step": 132 }, { "epoch": 0.3325, "grad_norm": 39.5, "learning_rate": 2.216666666666667e-06, "loss": 9.1552, "step": 133 }, { "epoch": 0.335, "grad_norm": 46.25, "learning_rate": 2.2333333333333333e-06, "loss": 9.0268, "step": 134 }, { "epoch": 0.3375, "grad_norm": 46.0, "learning_rate": 2.25e-06, "loss": 9.4363, "step": 135 }, { "epoch": 0.34, "grad_norm": 53.75, "learning_rate": 2.266666666666667e-06, "loss": 8.6541, "step": 136 }, { "epoch": 0.3425, "grad_norm": 36.5, "learning_rate": 2.2833333333333336e-06, "loss": 8.8334, "step": 137 }, { "epoch": 0.345, "grad_norm": 39.0, "learning_rate": 2.3000000000000004e-06, "loss": 8.7937, "step": 138 }, { "epoch": 0.3475, "grad_norm": 41.5, "learning_rate": 2.316666666666667e-06, "loss": 8.5079, "step": 139 }, { "epoch": 0.35, "grad_norm": 37.5, "learning_rate": 2.3333333333333336e-06, "loss": 8.5844, "step": 140 }, { "epoch": 0.3525, "grad_norm": 41.75, "learning_rate": 2.35e-06, "loss": 9.3903, "step": 141 }, { "epoch": 0.355, "grad_norm": 38.5, "learning_rate": 2.3666666666666667e-06, "loss": 8.8346, "step": 142 }, { "epoch": 0.3575, "grad_norm": 38.75, "learning_rate": 2.3833333333333335e-06, "loss": 8.8785, "step": 143 }, { "epoch": 0.36, "grad_norm": 43.25, "learning_rate": 2.4000000000000003e-06, "loss": 8.9883, "step": 144 }, { "epoch": 0.3625, "grad_norm": 40.5, "learning_rate": 2.4166666666666667e-06, "loss": 9.0533, "step": 145 }, { "epoch": 0.365, "grad_norm": 54.75, "learning_rate": 2.4333333333333335e-06, "loss": 9.2187, "step": 146 }, { "epoch": 0.3675, "grad_norm": 40.5, "learning_rate": 2.4500000000000003e-06, "loss": 8.4211, "step": 147 }, { "epoch": 0.37, "grad_norm": 38.5, "learning_rate": 2.466666666666667e-06, "loss": 9.243, "step": 148 }, { "epoch": 0.3725, "grad_norm": 36.75, "learning_rate": 2.4833333333333334e-06, "loss": 8.7438, "step": 149 }, { "epoch": 0.375, "grad_norm": 37.75, "learning_rate": 2.5e-06, "loss": 8.5839, "step": 150 }, { "epoch": 0.3775, "grad_norm": 43.75, "learning_rate": 2.5166666666666666e-06, "loss": 8.7294, "step": 151 }, { "epoch": 0.38, "grad_norm": 34.75, "learning_rate": 2.5333333333333338e-06, "loss": 8.5796, "step": 152 }, { "epoch": 0.3825, "grad_norm": 42.75, "learning_rate": 2.55e-06, "loss": 8.3822, "step": 153 }, { "epoch": 0.385, "grad_norm": 55.0, "learning_rate": 2.566666666666667e-06, "loss": 8.6376, "step": 154 }, { "epoch": 0.3875, "grad_norm": 140.0, "learning_rate": 2.5833333333333337e-06, "loss": 8.5386, "step": 155 }, { "epoch": 0.39, "grad_norm": 41.25, "learning_rate": 2.6e-06, "loss": 8.5971, "step": 156 }, { "epoch": 0.3925, "grad_norm": 35.25, "learning_rate": 2.616666666666667e-06, "loss": 8.4711, "step": 157 }, { "epoch": 0.395, "grad_norm": 33.75, "learning_rate": 2.6333333333333332e-06, "loss": 7.8206, "step": 158 }, { "epoch": 0.3975, "grad_norm": 34.75, "learning_rate": 2.6500000000000005e-06, "loss": 8.4776, "step": 159 }, { "epoch": 0.4, "grad_norm": 38.75, "learning_rate": 2.666666666666667e-06, "loss": 8.402, "step": 160 }, { "epoch": 0.4025, "grad_norm": 34.75, "learning_rate": 2.683333333333333e-06, "loss": 7.8293, "step": 161 }, { "epoch": 0.405, "grad_norm": 41.0, "learning_rate": 2.7000000000000004e-06, "loss": 8.9062, "step": 162 }, { "epoch": 0.4075, "grad_norm": 43.25, "learning_rate": 2.7166666666666668e-06, "loss": 8.0534, "step": 163 }, { "epoch": 0.41, "grad_norm": 42.0, "learning_rate": 2.7333333333333336e-06, "loss": 8.2074, "step": 164 }, { "epoch": 0.4125, "grad_norm": 31.625, "learning_rate": 2.7500000000000004e-06, "loss": 8.0392, "step": 165 }, { "epoch": 0.415, "grad_norm": 71.0, "learning_rate": 2.766666666666667e-06, "loss": 7.893, "step": 166 }, { "epoch": 0.4175, "grad_norm": 44.75, "learning_rate": 2.7833333333333335e-06, "loss": 8.0257, "step": 167 }, { "epoch": 0.42, "grad_norm": 37.25, "learning_rate": 2.8000000000000003e-06, "loss": 8.3567, "step": 168 }, { "epoch": 0.4225, "grad_norm": 39.25, "learning_rate": 2.816666666666667e-06, "loss": 7.6085, "step": 169 }, { "epoch": 0.425, "grad_norm": 65.5, "learning_rate": 2.8333333333333335e-06, "loss": 8.0111, "step": 170 }, { "epoch": 0.4275, "grad_norm": 36.0, "learning_rate": 2.85e-06, "loss": 8.0696, "step": 171 }, { "epoch": 0.43, "grad_norm": 40.0, "learning_rate": 2.866666666666667e-06, "loss": 8.0667, "step": 172 }, { "epoch": 0.4325, "grad_norm": 51.0, "learning_rate": 2.8833333333333334e-06, "loss": 8.0656, "step": 173 }, { "epoch": 0.435, "grad_norm": 45.25, "learning_rate": 2.9e-06, "loss": 7.9804, "step": 174 }, { "epoch": 0.4375, "grad_norm": 46.5, "learning_rate": 2.916666666666667e-06, "loss": 8.0171, "step": 175 }, { "epoch": 0.44, "grad_norm": 45.25, "learning_rate": 2.9333333333333338e-06, "loss": 7.6725, "step": 176 }, { "epoch": 0.4425, "grad_norm": 50.25, "learning_rate": 2.95e-06, "loss": 8.0119, "step": 177 }, { "epoch": 0.445, "grad_norm": 63.75, "learning_rate": 2.9666666666666673e-06, "loss": 8.0369, "step": 178 }, { "epoch": 0.4475, "grad_norm": 52.25, "learning_rate": 2.9833333333333337e-06, "loss": 8.1075, "step": 179 }, { "epoch": 0.45, "grad_norm": 44.0, "learning_rate": 3e-06, "loss": 7.7709, "step": 180 }, { "epoch": 0.4525, "grad_norm": 40.75, "learning_rate": 3.0166666666666673e-06, "loss": 7.303, "step": 181 }, { "epoch": 0.455, "grad_norm": 104.5, "learning_rate": 3.0333333333333337e-06, "loss": 7.4257, "step": 182 }, { "epoch": 0.4575, "grad_norm": 35.75, "learning_rate": 3.05e-06, "loss": 7.162, "step": 183 }, { "epoch": 0.46, "grad_norm": 57.25, "learning_rate": 3.066666666666667e-06, "loss": 7.4971, "step": 184 }, { "epoch": 0.4625, "grad_norm": 38.25, "learning_rate": 3.0833333333333336e-06, "loss": 7.4069, "step": 185 }, { "epoch": 0.465, "grad_norm": 40.5, "learning_rate": 3.1000000000000004e-06, "loss": 7.6979, "step": 186 }, { "epoch": 0.4675, "grad_norm": 43.75, "learning_rate": 3.1166666666666668e-06, "loss": 7.2977, "step": 187 }, { "epoch": 0.47, "grad_norm": 52.0, "learning_rate": 3.133333333333334e-06, "loss": 7.4828, "step": 188 }, { "epoch": 0.4725, "grad_norm": 63.0, "learning_rate": 3.1500000000000003e-06, "loss": 7.5672, "step": 189 }, { "epoch": 0.475, "grad_norm": 37.0, "learning_rate": 3.1666666666666667e-06, "loss": 6.9435, "step": 190 }, { "epoch": 0.4775, "grad_norm": 41.0, "learning_rate": 3.183333333333334e-06, "loss": 6.764, "step": 191 }, { "epoch": 0.48, "grad_norm": 40.25, "learning_rate": 3.2000000000000003e-06, "loss": 7.3457, "step": 192 }, { "epoch": 0.4825, "grad_norm": 37.0, "learning_rate": 3.2166666666666666e-06, "loss": 6.9832, "step": 193 }, { "epoch": 0.485, "grad_norm": 43.0, "learning_rate": 3.2333333333333334e-06, "loss": 6.9403, "step": 194 }, { "epoch": 0.4875, "grad_norm": 44.0, "learning_rate": 3.2500000000000002e-06, "loss": 7.3005, "step": 195 }, { "epoch": 0.49, "grad_norm": 44.25, "learning_rate": 3.266666666666667e-06, "loss": 6.9415, "step": 196 }, { "epoch": 0.4925, "grad_norm": 62.0, "learning_rate": 3.2833333333333334e-06, "loss": 6.7639, "step": 197 }, { "epoch": 0.495, "grad_norm": 196.0, "learning_rate": 3.3000000000000006e-06, "loss": 6.6233, "step": 198 }, { "epoch": 0.4975, "grad_norm": 36.25, "learning_rate": 3.316666666666667e-06, "loss": 6.7272, "step": 199 }, { "epoch": 0.5, "grad_norm": 51.25, "learning_rate": 3.3333333333333333e-06, "loss": 6.8627, "step": 200 }, { "epoch": 0.5025, "grad_norm": 48.0, "learning_rate": 3.3500000000000005e-06, "loss": 6.8284, "step": 201 }, { "epoch": 0.505, "grad_norm": 37.0, "learning_rate": 3.366666666666667e-06, "loss": 6.3442, "step": 202 }, { "epoch": 0.5075, "grad_norm": 43.0, "learning_rate": 3.3833333333333333e-06, "loss": 6.7435, "step": 203 }, { "epoch": 0.51, "grad_norm": 67.5, "learning_rate": 3.4000000000000005e-06, "loss": 6.5693, "step": 204 }, { "epoch": 0.5125, "grad_norm": 36.5, "learning_rate": 3.416666666666667e-06, "loss": 6.3743, "step": 205 }, { "epoch": 0.515, "grad_norm": 129.0, "learning_rate": 3.4333333333333336e-06, "loss": 6.3781, "step": 206 }, { "epoch": 0.5175, "grad_norm": 36.25, "learning_rate": 3.45e-06, "loss": 6.68, "step": 207 }, { "epoch": 0.52, "grad_norm": 38.0, "learning_rate": 3.4666666666666672e-06, "loss": 6.262, "step": 208 }, { "epoch": 0.5225, "grad_norm": 71.5, "learning_rate": 3.4833333333333336e-06, "loss": 6.3261, "step": 209 }, { "epoch": 0.525, "grad_norm": 36.5, "learning_rate": 3.5e-06, "loss": 6.2317, "step": 210 }, { "epoch": 0.5275, "grad_norm": 46.5, "learning_rate": 3.516666666666667e-06, "loss": 6.2135, "step": 211 }, { "epoch": 0.53, "grad_norm": 41.75, "learning_rate": 3.5333333333333335e-06, "loss": 6.2779, "step": 212 }, { "epoch": 0.5325, "grad_norm": 38.75, "learning_rate": 3.5500000000000003e-06, "loss": 6.4138, "step": 213 }, { "epoch": 0.535, "grad_norm": 37.25, "learning_rate": 3.566666666666667e-06, "loss": 6.3177, "step": 214 }, { "epoch": 0.5375, "grad_norm": 39.25, "learning_rate": 3.5833333333333335e-06, "loss": 6.003, "step": 215 }, { "epoch": 0.54, "grad_norm": 36.25, "learning_rate": 3.6000000000000003e-06, "loss": 6.0333, "step": 216 }, { "epoch": 0.5425, "grad_norm": 39.5, "learning_rate": 3.616666666666667e-06, "loss": 5.9571, "step": 217 }, { "epoch": 0.545, "grad_norm": 33.5, "learning_rate": 3.633333333333334e-06, "loss": 5.8766, "step": 218 }, { "epoch": 0.5475, "grad_norm": 40.75, "learning_rate": 3.65e-06, "loss": 5.8508, "step": 219 }, { "epoch": 0.55, "grad_norm": 36.0, "learning_rate": 3.6666666666666666e-06, "loss": 6.0458, "step": 220 }, { "epoch": 0.5525, "grad_norm": 31.375, "learning_rate": 3.6833333333333338e-06, "loss": 5.6195, "step": 221 }, { "epoch": 0.555, "grad_norm": 39.5, "learning_rate": 3.7e-06, "loss": 5.8788, "step": 222 }, { "epoch": 0.5575, "grad_norm": 47.5, "learning_rate": 3.716666666666667e-06, "loss": 5.8811, "step": 223 }, { "epoch": 0.56, "grad_norm": 47.25, "learning_rate": 3.7333333333333337e-06, "loss": 6.1015, "step": 224 }, { "epoch": 0.5625, "grad_norm": 35.75, "learning_rate": 3.7500000000000005e-06, "loss": 5.6728, "step": 225 }, { "epoch": 0.565, "grad_norm": 37.25, "learning_rate": 3.766666666666667e-06, "loss": 5.8925, "step": 226 }, { "epoch": 0.5675, "grad_norm": 38.0, "learning_rate": 3.7833333333333337e-06, "loss": 5.6239, "step": 227 }, { "epoch": 0.57, "grad_norm": 38.75, "learning_rate": 3.8000000000000005e-06, "loss": 5.6675, "step": 228 }, { "epoch": 0.5725, "grad_norm": 35.25, "learning_rate": 3.816666666666667e-06, "loss": 5.5825, "step": 229 }, { "epoch": 0.575, "grad_norm": 40.0, "learning_rate": 3.833333333333334e-06, "loss": 5.9147, "step": 230 }, { "epoch": 0.5775, "grad_norm": 45.5, "learning_rate": 3.85e-06, "loss": 5.3883, "step": 231 }, { "epoch": 0.58, "grad_norm": 37.5, "learning_rate": 3.866666666666667e-06, "loss": 5.4497, "step": 232 }, { "epoch": 0.5825, "grad_norm": 224.0, "learning_rate": 3.883333333333333e-06, "loss": 5.5422, "step": 233 }, { "epoch": 0.585, "grad_norm": 38.0, "learning_rate": 3.900000000000001e-06, "loss": 5.426, "step": 234 }, { "epoch": 0.5875, "grad_norm": 31.625, "learning_rate": 3.916666666666667e-06, "loss": 5.4259, "step": 235 }, { "epoch": 0.59, "grad_norm": 39.75, "learning_rate": 3.9333333333333335e-06, "loss": 5.6086, "step": 236 }, { "epoch": 0.5925, "grad_norm": 45.0, "learning_rate": 3.95e-06, "loss": 5.6409, "step": 237 }, { "epoch": 0.595, "grad_norm": 40.5, "learning_rate": 3.966666666666667e-06, "loss": 5.2139, "step": 238 }, { "epoch": 0.5975, "grad_norm": 39.5, "learning_rate": 3.983333333333334e-06, "loss": 5.2797, "step": 239 }, { "epoch": 0.6, "grad_norm": 45.25, "learning_rate": 4.000000000000001e-06, "loss": 5.2778, "step": 240 }, { "epoch": 0.6025, "grad_norm": 35.25, "learning_rate": 4.0166666666666675e-06, "loss": 5.3007, "step": 241 }, { "epoch": 0.605, "grad_norm": 40.75, "learning_rate": 4.033333333333333e-06, "loss": 5.2972, "step": 242 }, { "epoch": 0.6075, "grad_norm": 40.75, "learning_rate": 4.05e-06, "loss": 5.1902, "step": 243 }, { "epoch": 0.61, "grad_norm": 37.25, "learning_rate": 4.066666666666667e-06, "loss": 5.4196, "step": 244 }, { "epoch": 0.6125, "grad_norm": 31.625, "learning_rate": 4.083333333333334e-06, "loss": 5.2485, "step": 245 }, { "epoch": 0.615, "grad_norm": 41.75, "learning_rate": 4.1e-06, "loss": 5.2654, "step": 246 }, { "epoch": 0.6175, "grad_norm": 40.25, "learning_rate": 4.116666666666667e-06, "loss": 5.2445, "step": 247 }, { "epoch": 0.62, "grad_norm": 42.0, "learning_rate": 4.133333333333333e-06, "loss": 5.0072, "step": 248 }, { "epoch": 0.6225, "grad_norm": 35.25, "learning_rate": 4.15e-06, "loss": 4.9919, "step": 249 }, { "epoch": 0.625, "grad_norm": 38.25, "learning_rate": 4.166666666666667e-06, "loss": 5.0436, "step": 250 }, { "epoch": 0.6275, "grad_norm": 39.0, "learning_rate": 4.183333333333334e-06, "loss": 5.1982, "step": 251 }, { "epoch": 0.63, "grad_norm": 38.75, "learning_rate": 4.2000000000000004e-06, "loss": 4.9891, "step": 252 }, { "epoch": 0.6325, "grad_norm": 41.0, "learning_rate": 4.216666666666667e-06, "loss": 5.1384, "step": 253 }, { "epoch": 0.635, "grad_norm": 45.0, "learning_rate": 4.233333333333334e-06, "loss": 5.1634, "step": 254 }, { "epoch": 0.6375, "grad_norm": 35.25, "learning_rate": 4.25e-06, "loss": 4.9578, "step": 255 }, { "epoch": 0.64, "grad_norm": 34.25, "learning_rate": 4.266666666666668e-06, "loss": 4.966, "step": 256 }, { "epoch": 0.6425, "grad_norm": 31.0, "learning_rate": 4.2833333333333335e-06, "loss": 5.0735, "step": 257 }, { "epoch": 0.645, "grad_norm": 426.0, "learning_rate": 4.3e-06, "loss": 4.8613, "step": 258 }, { "epoch": 0.6475, "grad_norm": 33.25, "learning_rate": 4.316666666666667e-06, "loss": 4.9193, "step": 259 }, { "epoch": 0.65, "grad_norm": 43.0, "learning_rate": 4.333333333333334e-06, "loss": 4.7374, "step": 260 }, { "epoch": 0.6525, "grad_norm": 45.25, "learning_rate": 4.350000000000001e-06, "loss": 4.9289, "step": 261 }, { "epoch": 0.655, "grad_norm": 38.5, "learning_rate": 4.366666666666667e-06, "loss": 4.8356, "step": 262 }, { "epoch": 0.6575, "grad_norm": 39.25, "learning_rate": 4.383333333333334e-06, "loss": 4.7654, "step": 263 }, { "epoch": 0.66, "grad_norm": 60.0, "learning_rate": 4.4e-06, "loss": 4.7299, "step": 264 }, { "epoch": 0.6625, "grad_norm": 37.5, "learning_rate": 4.416666666666667e-06, "loss": 5.1247, "step": 265 }, { "epoch": 0.665, "grad_norm": 37.25, "learning_rate": 4.433333333333334e-06, "loss": 4.6393, "step": 266 }, { "epoch": 0.6675, "grad_norm": 34.75, "learning_rate": 4.450000000000001e-06, "loss": 5.0079, "step": 267 }, { "epoch": 0.67, "grad_norm": 35.25, "learning_rate": 4.4666666666666665e-06, "loss": 4.6392, "step": 268 }, { "epoch": 0.6725, "grad_norm": 33.5, "learning_rate": 4.483333333333333e-06, "loss": 4.8432, "step": 269 }, { "epoch": 0.675, "grad_norm": 36.25, "learning_rate": 4.5e-06, "loss": 4.5323, "step": 270 }, { "epoch": 0.6775, "grad_norm": 33.5, "learning_rate": 4.516666666666667e-06, "loss": 4.6274, "step": 271 }, { "epoch": 0.68, "grad_norm": 39.75, "learning_rate": 4.533333333333334e-06, "loss": 4.5539, "step": 272 }, { "epoch": 0.6825, "grad_norm": 42.5, "learning_rate": 4.5500000000000005e-06, "loss": 4.7105, "step": 273 }, { "epoch": 0.685, "grad_norm": 36.5, "learning_rate": 4.566666666666667e-06, "loss": 4.6753, "step": 274 }, { "epoch": 0.6875, "grad_norm": 34.0, "learning_rate": 4.583333333333333e-06, "loss": 4.7576, "step": 275 }, { "epoch": 0.69, "grad_norm": 35.0, "learning_rate": 4.600000000000001e-06, "loss": 4.6467, "step": 276 }, { "epoch": 0.6925, "grad_norm": 33.75, "learning_rate": 4.616666666666667e-06, "loss": 4.523, "step": 277 }, { "epoch": 0.695, "grad_norm": 31.625, "learning_rate": 4.633333333333334e-06, "loss": 4.6887, "step": 278 }, { "epoch": 0.6975, "grad_norm": 42.75, "learning_rate": 4.65e-06, "loss": 4.6587, "step": 279 }, { "epoch": 0.7, "grad_norm": 37.75, "learning_rate": 4.666666666666667e-06, "loss": 4.5859, "step": 280 }, { "epoch": 0.7025, "grad_norm": 38.0, "learning_rate": 4.683333333333334e-06, "loss": 4.3794, "step": 281 }, { "epoch": 0.705, "grad_norm": 44.0, "learning_rate": 4.7e-06, "loss": 4.7972, "step": 282 }, { "epoch": 0.7075, "grad_norm": 34.5, "learning_rate": 4.7166666666666675e-06, "loss": 4.54, "step": 283 }, { "epoch": 0.71, "grad_norm": 36.0, "learning_rate": 4.7333333333333335e-06, "loss": 4.376, "step": 284 }, { "epoch": 0.7125, "grad_norm": 36.0, "learning_rate": 4.75e-06, "loss": 4.2452, "step": 285 }, { "epoch": 0.715, "grad_norm": 48.25, "learning_rate": 4.766666666666667e-06, "loss": 4.3917, "step": 286 }, { "epoch": 0.7175, "grad_norm": 34.75, "learning_rate": 4.783333333333334e-06, "loss": 4.4996, "step": 287 }, { "epoch": 0.72, "grad_norm": 34.5, "learning_rate": 4.800000000000001e-06, "loss": 4.4778, "step": 288 }, { "epoch": 0.7225, "grad_norm": 40.0, "learning_rate": 4.816666666666667e-06, "loss": 4.0101, "step": 289 }, { "epoch": 0.725, "grad_norm": 35.25, "learning_rate": 4.833333333333333e-06, "loss": 4.2483, "step": 290 }, { "epoch": 0.7275, "grad_norm": 36.5, "learning_rate": 4.85e-06, "loss": 4.3227, "step": 291 }, { "epoch": 0.73, "grad_norm": 44.5, "learning_rate": 4.866666666666667e-06, "loss": 4.2707, "step": 292 }, { "epoch": 0.7325, "grad_norm": 36.5, "learning_rate": 4.883333333333334e-06, "loss": 4.477, "step": 293 }, { "epoch": 0.735, "grad_norm": 38.0, "learning_rate": 4.9000000000000005e-06, "loss": 4.2562, "step": 294 }, { "epoch": 0.7375, "grad_norm": 36.75, "learning_rate": 4.9166666666666665e-06, "loss": 4.2401, "step": 295 }, { "epoch": 0.74, "grad_norm": 36.5, "learning_rate": 4.933333333333334e-06, "loss": 4.3451, "step": 296 }, { "epoch": 0.7425, "grad_norm": 33.75, "learning_rate": 4.95e-06, "loss": 4.438, "step": 297 }, { "epoch": 0.745, "grad_norm": 41.0, "learning_rate": 4.966666666666667e-06, "loss": 4.1959, "step": 298 }, { "epoch": 0.7475, "grad_norm": 34.75, "learning_rate": 4.983333333333334e-06, "loss": 4.0974, "step": 299 }, { "epoch": 0.75, "grad_norm": 38.0, "learning_rate": 5e-06, "loss": 3.9841, "step": 300 }, { "epoch": 0.7525, "grad_norm": 35.75, "learning_rate": 5.016666666666667e-06, "loss": 3.9552, "step": 301 }, { "epoch": 0.755, "grad_norm": 33.0, "learning_rate": 5.033333333333333e-06, "loss": 4.1007, "step": 302 }, { "epoch": 0.7575, "grad_norm": 39.25, "learning_rate": 5.050000000000001e-06, "loss": 4.287, "step": 303 }, { "epoch": 0.76, "grad_norm": 34.0, "learning_rate": 5.0666666666666676e-06, "loss": 4.016, "step": 304 }, { "epoch": 0.7625, "grad_norm": 37.0, "learning_rate": 5.0833333333333335e-06, "loss": 4.0867, "step": 305 }, { "epoch": 0.765, "grad_norm": 33.25, "learning_rate": 5.1e-06, "loss": 4.1524, "step": 306 }, { "epoch": 0.7675, "grad_norm": 38.25, "learning_rate": 5.116666666666668e-06, "loss": 4.1088, "step": 307 }, { "epoch": 0.77, "grad_norm": 29.875, "learning_rate": 5.133333333333334e-06, "loss": 4.0826, "step": 308 }, { "epoch": 0.7725, "grad_norm": 31.75, "learning_rate": 5.150000000000001e-06, "loss": 4.1249, "step": 309 }, { "epoch": 0.775, "grad_norm": 35.25, "learning_rate": 5.1666666666666675e-06, "loss": 4.0744, "step": 310 }, { "epoch": 0.7775, "grad_norm": 33.75, "learning_rate": 5.183333333333333e-06, "loss": 3.9247, "step": 311 }, { "epoch": 0.78, "grad_norm": 32.75, "learning_rate": 5.2e-06, "loss": 4.0388, "step": 312 }, { "epoch": 0.7825, "grad_norm": 36.75, "learning_rate": 5.216666666666666e-06, "loss": 3.7449, "step": 313 }, { "epoch": 0.785, "grad_norm": 36.0, "learning_rate": 5.233333333333334e-06, "loss": 3.918, "step": 314 }, { "epoch": 0.7875, "grad_norm": 33.25, "learning_rate": 5.2500000000000006e-06, "loss": 3.9551, "step": 315 }, { "epoch": 0.79, "grad_norm": 34.5, "learning_rate": 5.2666666666666665e-06, "loss": 3.6692, "step": 316 }, { "epoch": 0.7925, "grad_norm": 30.75, "learning_rate": 5.283333333333333e-06, "loss": 3.795, "step": 317 }, { "epoch": 0.795, "grad_norm": 34.0, "learning_rate": 5.300000000000001e-06, "loss": 3.9138, "step": 318 }, { "epoch": 0.7975, "grad_norm": 35.75, "learning_rate": 5.316666666666667e-06, "loss": 4.0661, "step": 319 }, { "epoch": 0.8, "grad_norm": 37.25, "learning_rate": 5.333333333333334e-06, "loss": 3.8719, "step": 320 }, { "epoch": 0.8025, "grad_norm": 39.0, "learning_rate": 5.3500000000000004e-06, "loss": 3.8021, "step": 321 }, { "epoch": 0.805, "grad_norm": 31.125, "learning_rate": 5.366666666666666e-06, "loss": 3.6683, "step": 322 }, { "epoch": 0.8075, "grad_norm": 35.5, "learning_rate": 5.383333333333334e-06, "loss": 3.7525, "step": 323 }, { "epoch": 0.81, "grad_norm": 31.875, "learning_rate": 5.400000000000001e-06, "loss": 3.852, "step": 324 }, { "epoch": 0.8125, "grad_norm": 34.5, "learning_rate": 5.416666666666667e-06, "loss": 3.7356, "step": 325 }, { "epoch": 0.815, "grad_norm": 36.0, "learning_rate": 5.4333333333333335e-06, "loss": 3.6034, "step": 326 }, { "epoch": 0.8175, "grad_norm": 24.0, "learning_rate": 5.450000000000001e-06, "loss": 3.9993, "step": 327 }, { "epoch": 0.82, "grad_norm": 35.5, "learning_rate": 5.466666666666667e-06, "loss": 4.3002, "step": 328 }, { "epoch": 0.8225, "grad_norm": 32.5, "learning_rate": 5.483333333333334e-06, "loss": 3.7443, "step": 329 }, { "epoch": 0.825, "grad_norm": 32.25, "learning_rate": 5.500000000000001e-06, "loss": 3.5392, "step": 330 }, { "epoch": 0.8275, "grad_norm": 75.5, "learning_rate": 5.516666666666667e-06, "loss": 3.6678, "step": 331 }, { "epoch": 0.83, "grad_norm": 28.125, "learning_rate": 5.533333333333334e-06, "loss": 3.9691, "step": 332 }, { "epoch": 0.8325, "grad_norm": 34.75, "learning_rate": 5.550000000000001e-06, "loss": 3.859, "step": 333 }, { "epoch": 0.835, "grad_norm": 29.625, "learning_rate": 5.566666666666667e-06, "loss": 3.2724, "step": 334 }, { "epoch": 0.8375, "grad_norm": 36.0, "learning_rate": 5.583333333333334e-06, "loss": 3.4493, "step": 335 }, { "epoch": 0.84, "grad_norm": 30.0, "learning_rate": 5.600000000000001e-06, "loss": 3.6837, "step": 336 }, { "epoch": 0.8425, "grad_norm": 41.0, "learning_rate": 5.6166666666666665e-06, "loss": 3.3929, "step": 337 }, { "epoch": 0.845, "grad_norm": 37.75, "learning_rate": 5.633333333333334e-06, "loss": 3.4388, "step": 338 }, { "epoch": 0.8475, "grad_norm": 35.25, "learning_rate": 5.65e-06, "loss": 3.4829, "step": 339 }, { "epoch": 0.85, "grad_norm": 41.5, "learning_rate": 5.666666666666667e-06, "loss": 3.4883, "step": 340 }, { "epoch": 0.8525, "grad_norm": 34.0, "learning_rate": 5.683333333333334e-06, "loss": 3.6224, "step": 341 }, { "epoch": 0.855, "grad_norm": 50.75, "learning_rate": 5.7e-06, "loss": 3.4448, "step": 342 }, { "epoch": 0.8575, "grad_norm": 34.25, "learning_rate": 5.716666666666667e-06, "loss": 3.4176, "step": 343 }, { "epoch": 0.86, "grad_norm": 31.125, "learning_rate": 5.733333333333334e-06, "loss": 3.2845, "step": 344 }, { "epoch": 0.8625, "grad_norm": 32.75, "learning_rate": 5.75e-06, "loss": 3.4523, "step": 345 }, { "epoch": 0.865, "grad_norm": 29.375, "learning_rate": 5.766666666666667e-06, "loss": 3.5548, "step": 346 }, { "epoch": 0.8675, "grad_norm": 37.5, "learning_rate": 5.7833333333333344e-06, "loss": 3.476, "step": 347 }, { "epoch": 0.87, "grad_norm": 31.875, "learning_rate": 5.8e-06, "loss": 3.4827, "step": 348 }, { "epoch": 0.8725, "grad_norm": 29.375, "learning_rate": 5.816666666666667e-06, "loss": 3.6004, "step": 349 }, { "epoch": 0.875, "grad_norm": 32.25, "learning_rate": 5.833333333333334e-06, "loss": 3.3331, "step": 350 }, { "epoch": 0.8775, "grad_norm": 30.5, "learning_rate": 5.85e-06, "loss": 3.4726, "step": 351 }, { "epoch": 0.88, "grad_norm": 35.0, "learning_rate": 5.8666666666666675e-06, "loss": 3.5024, "step": 352 }, { "epoch": 0.8825, "grad_norm": 66.0, "learning_rate": 5.883333333333334e-06, "loss": 3.4756, "step": 353 }, { "epoch": 0.885, "grad_norm": 30.0, "learning_rate": 5.9e-06, "loss": 3.4321, "step": 354 }, { "epoch": 0.8875, "grad_norm": 30.375, "learning_rate": 5.916666666666667e-06, "loss": 3.2764, "step": 355 }, { "epoch": 0.89, "grad_norm": 30.5, "learning_rate": 5.933333333333335e-06, "loss": 3.1743, "step": 356 }, { "epoch": 0.8925, "grad_norm": 25.25, "learning_rate": 5.950000000000001e-06, "loss": 3.5444, "step": 357 }, { "epoch": 0.895, "grad_norm": 32.25, "learning_rate": 5.966666666666667e-06, "loss": 3.2602, "step": 358 }, { "epoch": 0.8975, "grad_norm": 41.0, "learning_rate": 5.983333333333334e-06, "loss": 3.2278, "step": 359 }, { "epoch": 0.9, "grad_norm": 30.0, "learning_rate": 6e-06, "loss": 3.5181, "step": 360 }, { "epoch": 0.9025, "grad_norm": 27.5, "learning_rate": 6.016666666666667e-06, "loss": 3.3751, "step": 361 }, { "epoch": 0.905, "grad_norm": 28.375, "learning_rate": 6.033333333333335e-06, "loss": 3.4614, "step": 362 }, { "epoch": 0.9075, "grad_norm": 27.0, "learning_rate": 6.0500000000000005e-06, "loss": 3.4974, "step": 363 }, { "epoch": 0.91, "grad_norm": 66.0, "learning_rate": 6.066666666666667e-06, "loss": 3.44, "step": 364 }, { "epoch": 0.9125, "grad_norm": 25.875, "learning_rate": 6.083333333333333e-06, "loss": 3.5268, "step": 365 }, { "epoch": 0.915, "grad_norm": 26.375, "learning_rate": 6.1e-06, "loss": 3.1015, "step": 366 }, { "epoch": 0.9175, "grad_norm": 26.25, "learning_rate": 6.116666666666668e-06, "loss": 3.027, "step": 367 }, { "epoch": 0.92, "grad_norm": 26.375, "learning_rate": 6.133333333333334e-06, "loss": 3.2108, "step": 368 }, { "epoch": 0.9225, "grad_norm": 30.25, "learning_rate": 6.15e-06, "loss": 3.2478, "step": 369 }, { "epoch": 0.925, "grad_norm": 26.0, "learning_rate": 6.166666666666667e-06, "loss": 3.3125, "step": 370 }, { "epoch": 0.9275, "grad_norm": 30.75, "learning_rate": 6.183333333333333e-06, "loss": 2.8875, "step": 371 }, { "epoch": 0.93, "grad_norm": 26.0, "learning_rate": 6.200000000000001e-06, "loss": 3.2186, "step": 372 }, { "epoch": 0.9325, "grad_norm": 27.75, "learning_rate": 6.2166666666666676e-06, "loss": 3.1704, "step": 373 }, { "epoch": 0.935, "grad_norm": 28.875, "learning_rate": 6.2333333333333335e-06, "loss": 3.051, "step": 374 }, { "epoch": 0.9375, "grad_norm": 27.5, "learning_rate": 6.25e-06, "loss": 2.8795, "step": 375 }, { "epoch": 0.94, "grad_norm": 26.25, "learning_rate": 6.266666666666668e-06, "loss": 2.9556, "step": 376 }, { "epoch": 0.9425, "grad_norm": 23.625, "learning_rate": 6.283333333333334e-06, "loss": 3.0939, "step": 377 }, { "epoch": 0.945, "grad_norm": 20.375, "learning_rate": 6.300000000000001e-06, "loss": 3.2777, "step": 378 }, { "epoch": 0.9475, "grad_norm": 26.5, "learning_rate": 6.3166666666666675e-06, "loss": 3.0888, "step": 379 }, { "epoch": 0.95, "grad_norm": 25.125, "learning_rate": 6.333333333333333e-06, "loss": 3.1779, "step": 380 }, { "epoch": 0.9525, "grad_norm": 22.625, "learning_rate": 6.35e-06, "loss": 3.1534, "step": 381 }, { "epoch": 0.955, "grad_norm": 21.375, "learning_rate": 6.366666666666668e-06, "loss": 3.1749, "step": 382 }, { "epoch": 0.9575, "grad_norm": 22.125, "learning_rate": 6.383333333333334e-06, "loss": 3.0325, "step": 383 }, { "epoch": 0.96, "grad_norm": 22.875, "learning_rate": 6.4000000000000006e-06, "loss": 3.0529, "step": 384 }, { "epoch": 0.9625, "grad_norm": 23.625, "learning_rate": 6.416666666666667e-06, "loss": 2.6881, "step": 385 }, { "epoch": 0.965, "grad_norm": 20.75, "learning_rate": 6.433333333333333e-06, "loss": 3.4933, "step": 386 }, { "epoch": 0.9675, "grad_norm": 20.75, "learning_rate": 6.450000000000001e-06, "loss": 3.1473, "step": 387 }, { "epoch": 0.97, "grad_norm": 16.125, "learning_rate": 6.466666666666667e-06, "loss": 3.2876, "step": 388 }, { "epoch": 0.9725, "grad_norm": 22.875, "learning_rate": 6.483333333333334e-06, "loss": 2.9448, "step": 389 }, { "epoch": 0.975, "grad_norm": 23.0, "learning_rate": 6.5000000000000004e-06, "loss": 2.9583, "step": 390 }, { "epoch": 0.9775, "grad_norm": 32.25, "learning_rate": 6.516666666666666e-06, "loss": 3.1209, "step": 391 }, { "epoch": 0.98, "grad_norm": 23.5, "learning_rate": 6.533333333333334e-06, "loss": 2.9169, "step": 392 }, { "epoch": 0.9825, "grad_norm": 18.25, "learning_rate": 6.550000000000001e-06, "loss": 3.1089, "step": 393 }, { "epoch": 0.985, "grad_norm": 21.875, "learning_rate": 6.566666666666667e-06, "loss": 2.9523, "step": 394 }, { "epoch": 0.9875, "grad_norm": 20.125, "learning_rate": 6.5833333333333335e-06, "loss": 3.0562, "step": 395 }, { "epoch": 0.99, "grad_norm": 19.75, "learning_rate": 6.600000000000001e-06, "loss": 2.8529, "step": 396 }, { "epoch": 0.9925, "grad_norm": 19.5, "learning_rate": 6.616666666666667e-06, "loss": 3.0818, "step": 397 }, { "epoch": 0.995, "grad_norm": 20.75, "learning_rate": 6.633333333333334e-06, "loss": 2.8075, "step": 398 }, { "epoch": 0.9975, "grad_norm": 17.875, "learning_rate": 6.650000000000001e-06, "loss": 3.0522, "step": 399 }, { "epoch": 1.0, "grad_norm": 20.125, "learning_rate": 6.666666666666667e-06, "loss": 2.9551, "step": 400 }, { "epoch": 1.0025, "grad_norm": 19.25, "learning_rate": 6.683333333333334e-06, "loss": 2.8023, "step": 401 }, { "epoch": 1.005, "grad_norm": 20.375, "learning_rate": 6.700000000000001e-06, "loss": 2.8645, "step": 402 }, { "epoch": 1.0075, "grad_norm": 16.5, "learning_rate": 6.716666666666667e-06, "loss": 2.8933, "step": 403 }, { "epoch": 1.01, "grad_norm": 18.5, "learning_rate": 6.733333333333334e-06, "loss": 2.9214, "step": 404 }, { "epoch": 1.0125, "grad_norm": 18.25, "learning_rate": 6.750000000000001e-06, "loss": 2.8971, "step": 405 }, { "epoch": 1.015, "grad_norm": 18.25, "learning_rate": 6.7666666666666665e-06, "loss": 2.7445, "step": 406 }, { "epoch": 1.0175, "grad_norm": 16.375, "learning_rate": 6.783333333333334e-06, "loss": 2.7187, "step": 407 }, { "epoch": 1.02, "grad_norm": 19.5, "learning_rate": 6.800000000000001e-06, "loss": 2.6019, "step": 408 }, { "epoch": 1.0225, "grad_norm": 16.5, "learning_rate": 6.816666666666667e-06, "loss": 2.8554, "step": 409 }, { "epoch": 1.025, "grad_norm": 15.0625, "learning_rate": 6.833333333333334e-06, "loss": 2.7271, "step": 410 }, { "epoch": 1.0275, "grad_norm": 16.5, "learning_rate": 6.850000000000001e-06, "loss": 2.4462, "step": 411 }, { "epoch": 1.03, "grad_norm": 16.5, "learning_rate": 6.866666666666667e-06, "loss": 2.7865, "step": 412 }, { "epoch": 1.0325, "grad_norm": 21.875, "learning_rate": 6.883333333333334e-06, "loss": 2.7073, "step": 413 }, { "epoch": 1.035, "grad_norm": 15.5625, "learning_rate": 6.9e-06, "loss": 2.9631, "step": 414 }, { "epoch": 1.0375, "grad_norm": 21.125, "learning_rate": 6.916666666666667e-06, "loss": 2.9304, "step": 415 }, { "epoch": 1.04, "grad_norm": 14.1875, "learning_rate": 6.9333333333333344e-06, "loss": 2.8085, "step": 416 }, { "epoch": 1.0425, "grad_norm": 15.6875, "learning_rate": 6.95e-06, "loss": 2.6834, "step": 417 }, { "epoch": 1.045, "grad_norm": 101.0, "learning_rate": 6.966666666666667e-06, "loss": 2.7616, "step": 418 }, { "epoch": 1.0475, "grad_norm": 16.5, "learning_rate": 6.983333333333334e-06, "loss": 2.9399, "step": 419 }, { "epoch": 1.05, "grad_norm": 17.625, "learning_rate": 7e-06, "loss": 3.1528, "step": 420 }, { "epoch": 1.0525, "grad_norm": 14.375, "learning_rate": 7.0166666666666675e-06, "loss": 2.8435, "step": 421 }, { "epoch": 1.055, "grad_norm": 14.875, "learning_rate": 7.033333333333334e-06, "loss": 2.8945, "step": 422 }, { "epoch": 1.0575, "grad_norm": 14.0625, "learning_rate": 7.05e-06, "loss": 2.6239, "step": 423 }, { "epoch": 1.06, "grad_norm": 12.625, "learning_rate": 7.066666666666667e-06, "loss": 2.8568, "step": 424 }, { "epoch": 1.0625, "grad_norm": 13.75, "learning_rate": 7.083333333333335e-06, "loss": 2.9027, "step": 425 }, { "epoch": 1.065, "grad_norm": 13.3125, "learning_rate": 7.100000000000001e-06, "loss": 2.7098, "step": 426 }, { "epoch": 1.0675, "grad_norm": 13.5, "learning_rate": 7.116666666666667e-06, "loss": 2.6456, "step": 427 }, { "epoch": 1.07, "grad_norm": 13.125, "learning_rate": 7.133333333333334e-06, "loss": 2.4807, "step": 428 }, { "epoch": 1.0725, "grad_norm": 14.0625, "learning_rate": 7.15e-06, "loss": 2.4727, "step": 429 }, { "epoch": 1.075, "grad_norm": 14.1875, "learning_rate": 7.166666666666667e-06, "loss": 2.6801, "step": 430 }, { "epoch": 1.0775, "grad_norm": 14.3125, "learning_rate": 7.183333333333335e-06, "loss": 2.6607, "step": 431 }, { "epoch": 1.08, "grad_norm": 13.25, "learning_rate": 7.2000000000000005e-06, "loss": 2.6508, "step": 432 }, { "epoch": 1.0825, "grad_norm": 18.0, "learning_rate": 7.216666666666667e-06, "loss": 2.7242, "step": 433 }, { "epoch": 1.085, "grad_norm": 11.25, "learning_rate": 7.233333333333334e-06, "loss": 2.888, "step": 434 }, { "epoch": 1.0875, "grad_norm": 14.375, "learning_rate": 7.25e-06, "loss": 2.7127, "step": 435 }, { "epoch": 1.09, "grad_norm": 16.875, "learning_rate": 7.266666666666668e-06, "loss": 2.9125, "step": 436 }, { "epoch": 1.0925, "grad_norm": 11.0, "learning_rate": 7.2833333333333345e-06, "loss": 2.833, "step": 437 }, { "epoch": 1.095, "grad_norm": 31.5, "learning_rate": 7.3e-06, "loss": 2.574, "step": 438 }, { "epoch": 1.0975, "grad_norm": 10.875, "learning_rate": 7.316666666666667e-06, "loss": 2.5179, "step": 439 }, { "epoch": 1.1, "grad_norm": 11.6875, "learning_rate": 7.333333333333333e-06, "loss": 2.623, "step": 440 }, { "epoch": 1.1025, "grad_norm": 10.375, "learning_rate": 7.350000000000001e-06, "loss": 2.5319, "step": 441 }, { "epoch": 1.105, "grad_norm": 11.375, "learning_rate": 7.3666666666666676e-06, "loss": 2.812, "step": 442 }, { "epoch": 1.1075, "grad_norm": 13.8125, "learning_rate": 7.3833333333333335e-06, "loss": 2.571, "step": 443 }, { "epoch": 1.11, "grad_norm": 9.8125, "learning_rate": 7.4e-06, "loss": 2.5694, "step": 444 }, { "epoch": 1.1125, "grad_norm": 10.25, "learning_rate": 7.416666666666668e-06, "loss": 2.5861, "step": 445 }, { "epoch": 1.115, "grad_norm": 12.625, "learning_rate": 7.433333333333334e-06, "loss": 2.7714, "step": 446 }, { "epoch": 1.1175, "grad_norm": 10.9375, "learning_rate": 7.450000000000001e-06, "loss": 2.5376, "step": 447 }, { "epoch": 1.12, "grad_norm": 11.375, "learning_rate": 7.4666666666666675e-06, "loss": 2.6654, "step": 448 }, { "epoch": 1.1225, "grad_norm": 10.9375, "learning_rate": 7.483333333333333e-06, "loss": 2.7599, "step": 449 }, { "epoch": 1.125, "grad_norm": 12.875, "learning_rate": 7.500000000000001e-06, "loss": 2.2183, "step": 450 }, { "epoch": 1.1275, "grad_norm": 9.6875, "learning_rate": 7.516666666666668e-06, "loss": 2.4856, "step": 451 }, { "epoch": 1.13, "grad_norm": 9.6875, "learning_rate": 7.533333333333334e-06, "loss": 2.753, "step": 452 }, { "epoch": 1.1325, "grad_norm": 8.4375, "learning_rate": 7.5500000000000006e-06, "loss": 2.461, "step": 453 }, { "epoch": 1.135, "grad_norm": 9.4375, "learning_rate": 7.566666666666667e-06, "loss": 2.4456, "step": 454 }, { "epoch": 1.1375, "grad_norm": 9.0, "learning_rate": 7.583333333333333e-06, "loss": 2.6318, "step": 455 }, { "epoch": 1.1400000000000001, "grad_norm": 8.625, "learning_rate": 7.600000000000001e-06, "loss": 2.7061, "step": 456 }, { "epoch": 1.1425, "grad_norm": 10.0, "learning_rate": 7.616666666666668e-06, "loss": 2.8401, "step": 457 }, { "epoch": 1.145, "grad_norm": 10.125, "learning_rate": 7.633333333333334e-06, "loss": 2.4801, "step": 458 }, { "epoch": 1.1475, "grad_norm": 9.0, "learning_rate": 7.650000000000001e-06, "loss": 2.5855, "step": 459 }, { "epoch": 1.15, "grad_norm": 8.875, "learning_rate": 7.666666666666667e-06, "loss": 2.806, "step": 460 }, { "epoch": 1.1525, "grad_norm": 8.3125, "learning_rate": 7.683333333333333e-06, "loss": 2.3993, "step": 461 }, { "epoch": 1.155, "grad_norm": 7.6875, "learning_rate": 7.7e-06, "loss": 2.9418, "step": 462 }, { "epoch": 1.1575, "grad_norm": 9.6875, "learning_rate": 7.716666666666667e-06, "loss": 2.5935, "step": 463 }, { "epoch": 1.16, "grad_norm": 8.0625, "learning_rate": 7.733333333333334e-06, "loss": 2.3242, "step": 464 }, { "epoch": 1.1625, "grad_norm": 8.3125, "learning_rate": 7.75e-06, "loss": 2.7282, "step": 465 }, { "epoch": 1.165, "grad_norm": 6.96875, "learning_rate": 7.766666666666666e-06, "loss": 2.3787, "step": 466 }, { "epoch": 1.1675, "grad_norm": 7.625, "learning_rate": 7.783333333333334e-06, "loss": 2.4942, "step": 467 }, { "epoch": 1.17, "grad_norm": 7.96875, "learning_rate": 7.800000000000002e-06, "loss": 2.6135, "step": 468 }, { "epoch": 1.1724999999999999, "grad_norm": 8.9375, "learning_rate": 7.816666666666667e-06, "loss": 2.3168, "step": 469 }, { "epoch": 1.175, "grad_norm": 12.3125, "learning_rate": 7.833333333333333e-06, "loss": 2.4798, "step": 470 }, { "epoch": 1.1775, "grad_norm": 14.25, "learning_rate": 7.850000000000001e-06, "loss": 2.8695, "step": 471 }, { "epoch": 1.18, "grad_norm": 8.25, "learning_rate": 7.866666666666667e-06, "loss": 2.3098, "step": 472 }, { "epoch": 1.1825, "grad_norm": 7.4375, "learning_rate": 7.883333333333335e-06, "loss": 2.5796, "step": 473 }, { "epoch": 1.185, "grad_norm": 6.78125, "learning_rate": 7.9e-06, "loss": 2.6991, "step": 474 }, { "epoch": 1.1875, "grad_norm": 7.75, "learning_rate": 7.916666666666667e-06, "loss": 2.6741, "step": 475 }, { "epoch": 1.19, "grad_norm": 6.59375, "learning_rate": 7.933333333333334e-06, "loss": 2.4158, "step": 476 }, { "epoch": 1.1925, "grad_norm": 8.75, "learning_rate": 7.950000000000002e-06, "loss": 2.6731, "step": 477 }, { "epoch": 1.195, "grad_norm": 6.84375, "learning_rate": 7.966666666666668e-06, "loss": 2.4752, "step": 478 }, { "epoch": 1.1975, "grad_norm": 7.34375, "learning_rate": 7.983333333333334e-06, "loss": 2.4917, "step": 479 }, { "epoch": 1.2, "grad_norm": 8.9375, "learning_rate": 8.000000000000001e-06, "loss": 2.6133, "step": 480 }, { "epoch": 1.2025000000000001, "grad_norm": 8.75, "learning_rate": 8.016666666666667e-06, "loss": 2.3521, "step": 481 }, { "epoch": 1.205, "grad_norm": 6.65625, "learning_rate": 8.033333333333335e-06, "loss": 2.5572, "step": 482 }, { "epoch": 1.2075, "grad_norm": 8.5625, "learning_rate": 8.050000000000001e-06, "loss": 2.7403, "step": 483 }, { "epoch": 1.21, "grad_norm": 7.40625, "learning_rate": 8.066666666666667e-06, "loss": 2.4752, "step": 484 }, { "epoch": 1.2125, "grad_norm": 8.4375, "learning_rate": 8.083333333333334e-06, "loss": 2.3339, "step": 485 }, { "epoch": 1.215, "grad_norm": 8.875, "learning_rate": 8.1e-06, "loss": 2.6029, "step": 486 }, { "epoch": 1.2175, "grad_norm": 12.25, "learning_rate": 8.116666666666666e-06, "loss": 2.662, "step": 487 }, { "epoch": 1.22, "grad_norm": 7.0625, "learning_rate": 8.133333333333334e-06, "loss": 2.5579, "step": 488 }, { "epoch": 1.2225, "grad_norm": 11.0625, "learning_rate": 8.15e-06, "loss": 2.7507, "step": 489 }, { "epoch": 1.225, "grad_norm": 7.15625, "learning_rate": 8.166666666666668e-06, "loss": 2.3063, "step": 490 }, { "epoch": 1.2275, "grad_norm": 10.9375, "learning_rate": 8.183333333333333e-06, "loss": 2.4425, "step": 491 }, { "epoch": 1.23, "grad_norm": 6.46875, "learning_rate": 8.2e-06, "loss": 2.3957, "step": 492 }, { "epoch": 1.2325, "grad_norm": 7.34375, "learning_rate": 8.216666666666667e-06, "loss": 2.4771, "step": 493 }, { "epoch": 1.2349999999999999, "grad_norm": 6.5, "learning_rate": 8.233333333333335e-06, "loss": 2.296, "step": 494 }, { "epoch": 1.2375, "grad_norm": 6.15625, "learning_rate": 8.25e-06, "loss": 2.4717, "step": 495 }, { "epoch": 1.24, "grad_norm": 10.1875, "learning_rate": 8.266666666666667e-06, "loss": 2.7574, "step": 496 }, { "epoch": 1.2425, "grad_norm": 6.5, "learning_rate": 8.283333333333334e-06, "loss": 2.4877, "step": 497 }, { "epoch": 1.245, "grad_norm": 8.8125, "learning_rate": 8.3e-06, "loss": 2.3433, "step": 498 }, { "epoch": 1.2475, "grad_norm": 6.875, "learning_rate": 8.316666666666668e-06, "loss": 2.6452, "step": 499 }, { "epoch": 1.25, "grad_norm": 6.03125, "learning_rate": 8.333333333333334e-06, "loss": 2.4287, "step": 500 }, { "epoch": 1.2525, "grad_norm": 6.53125, "learning_rate": 8.35e-06, "loss": 2.6096, "step": 501 }, { "epoch": 1.255, "grad_norm": 6.1875, "learning_rate": 8.366666666666667e-06, "loss": 2.507, "step": 502 }, { "epoch": 1.2575, "grad_norm": 5.15625, "learning_rate": 8.383333333333335e-06, "loss": 2.4231, "step": 503 }, { "epoch": 1.26, "grad_norm": 5.90625, "learning_rate": 8.400000000000001e-06, "loss": 2.3531, "step": 504 }, { "epoch": 1.2625, "grad_norm": 7.71875, "learning_rate": 8.416666666666667e-06, "loss": 2.1631, "step": 505 }, { "epoch": 1.2650000000000001, "grad_norm": 7.59375, "learning_rate": 8.433333333333334e-06, "loss": 2.6385, "step": 506 }, { "epoch": 1.2675, "grad_norm": 5.78125, "learning_rate": 8.45e-06, "loss": 2.6483, "step": 507 }, { "epoch": 1.27, "grad_norm": 5.9375, "learning_rate": 8.466666666666668e-06, "loss": 2.595, "step": 508 }, { "epoch": 1.2725, "grad_norm": 5.78125, "learning_rate": 8.483333333333334e-06, "loss": 2.4391, "step": 509 }, { "epoch": 1.275, "grad_norm": 5.53125, "learning_rate": 8.5e-06, "loss": 2.5958, "step": 510 }, { "epoch": 1.2775, "grad_norm": 6.25, "learning_rate": 8.516666666666668e-06, "loss": 2.6982, "step": 511 }, { "epoch": 1.28, "grad_norm": 4.78125, "learning_rate": 8.533333333333335e-06, "loss": 2.5129, "step": 512 }, { "epoch": 1.2825, "grad_norm": 5.3125, "learning_rate": 8.550000000000001e-06, "loss": 2.278, "step": 513 }, { "epoch": 1.285, "grad_norm": 7.125, "learning_rate": 8.566666666666667e-06, "loss": 2.426, "step": 514 }, { "epoch": 1.2875, "grad_norm": 6.09375, "learning_rate": 8.583333333333333e-06, "loss": 2.7468, "step": 515 }, { "epoch": 1.29, "grad_norm": 7.59375, "learning_rate": 8.6e-06, "loss": 2.4357, "step": 516 }, { "epoch": 1.2925, "grad_norm": 7.125, "learning_rate": 8.616666666666668e-06, "loss": 2.4779, "step": 517 }, { "epoch": 1.295, "grad_norm": 4.78125, "learning_rate": 8.633333333333334e-06, "loss": 2.3729, "step": 518 }, { "epoch": 1.2974999999999999, "grad_norm": 9.125, "learning_rate": 8.65e-06, "loss": 2.6424, "step": 519 }, { "epoch": 1.3, "grad_norm": 7.40625, "learning_rate": 8.666666666666668e-06, "loss": 2.418, "step": 520 }, { "epoch": 1.3025, "grad_norm": 6.875, "learning_rate": 8.683333333333334e-06, "loss": 2.6512, "step": 521 }, { "epoch": 1.305, "grad_norm": 6.625, "learning_rate": 8.700000000000001e-06, "loss": 2.3187, "step": 522 }, { "epoch": 1.3075, "grad_norm": 6.59375, "learning_rate": 8.716666666666667e-06, "loss": 2.1587, "step": 523 }, { "epoch": 1.31, "grad_norm": 7.71875, "learning_rate": 8.733333333333333e-06, "loss": 2.4568, "step": 524 }, { "epoch": 1.3125, "grad_norm": 5.65625, "learning_rate": 8.750000000000001e-06, "loss": 2.3468, "step": 525 }, { "epoch": 1.315, "grad_norm": 5.03125, "learning_rate": 8.766666666666669e-06, "loss": 2.5339, "step": 526 }, { "epoch": 1.3175, "grad_norm": 8.1875, "learning_rate": 8.783333333333335e-06, "loss": 2.678, "step": 527 }, { "epoch": 1.32, "grad_norm": 7.09375, "learning_rate": 8.8e-06, "loss": 2.3985, "step": 528 }, { "epoch": 1.3225, "grad_norm": 4.875, "learning_rate": 8.816666666666668e-06, "loss": 2.4683, "step": 529 }, { "epoch": 1.325, "grad_norm": 7.5, "learning_rate": 8.833333333333334e-06, "loss": 2.525, "step": 530 }, { "epoch": 1.3275000000000001, "grad_norm": 4.78125, "learning_rate": 8.85e-06, "loss": 2.291, "step": 531 }, { "epoch": 1.33, "grad_norm": 7.1875, "learning_rate": 8.866666666666668e-06, "loss": 2.5368, "step": 532 }, { "epoch": 1.3325, "grad_norm": 13.0, "learning_rate": 8.883333333333334e-06, "loss": 2.4191, "step": 533 }, { "epoch": 1.335, "grad_norm": 6.59375, "learning_rate": 8.900000000000001e-06, "loss": 2.5024, "step": 534 }, { "epoch": 1.3375, "grad_norm": 5.15625, "learning_rate": 8.916666666666667e-06, "loss": 2.5444, "step": 535 }, { "epoch": 1.34, "grad_norm": 4.8125, "learning_rate": 8.933333333333333e-06, "loss": 2.3121, "step": 536 }, { "epoch": 1.3425, "grad_norm": 5.90625, "learning_rate": 8.95e-06, "loss": 2.3541, "step": 537 }, { "epoch": 1.345, "grad_norm": 7.3125, "learning_rate": 8.966666666666667e-06, "loss": 2.3902, "step": 538 }, { "epoch": 1.3475, "grad_norm": 5.4375, "learning_rate": 8.983333333333334e-06, "loss": 2.3287, "step": 539 }, { "epoch": 1.35, "grad_norm": 6.15625, "learning_rate": 9e-06, "loss": 2.2596, "step": 540 }, { "epoch": 1.3525, "grad_norm": 5.96875, "learning_rate": 9.016666666666666e-06, "loss": 2.4653, "step": 541 }, { "epoch": 1.355, "grad_norm": 4.5625, "learning_rate": 9.033333333333334e-06, "loss": 2.274, "step": 542 }, { "epoch": 1.3575, "grad_norm": 5.1875, "learning_rate": 9.050000000000001e-06, "loss": 2.4588, "step": 543 }, { "epoch": 1.3599999999999999, "grad_norm": 6.59375, "learning_rate": 9.066666666666667e-06, "loss": 2.5082, "step": 544 }, { "epoch": 1.3625, "grad_norm": 9.3125, "learning_rate": 9.083333333333333e-06, "loss": 2.3622, "step": 545 }, { "epoch": 1.365, "grad_norm": 5.25, "learning_rate": 9.100000000000001e-06, "loss": 2.0824, "step": 546 }, { "epoch": 1.3675, "grad_norm": 7.28125, "learning_rate": 9.116666666666667e-06, "loss": 2.5548, "step": 547 }, { "epoch": 1.37, "grad_norm": 6.4375, "learning_rate": 9.133333333333335e-06, "loss": 2.6988, "step": 548 }, { "epoch": 1.3725, "grad_norm": 6.09375, "learning_rate": 9.15e-06, "loss": 2.4931, "step": 549 }, { "epoch": 1.375, "grad_norm": 10.0625, "learning_rate": 9.166666666666666e-06, "loss": 2.2807, "step": 550 }, { "epoch": 1.3775, "grad_norm": 7.03125, "learning_rate": 9.183333333333334e-06, "loss": 2.3912, "step": 551 }, { "epoch": 1.38, "grad_norm": 4.625, "learning_rate": 9.200000000000002e-06, "loss": 2.283, "step": 552 }, { "epoch": 1.3825, "grad_norm": 4.625, "learning_rate": 9.216666666666668e-06, "loss": 2.3714, "step": 553 }, { "epoch": 1.385, "grad_norm": 6.71875, "learning_rate": 9.233333333333334e-06, "loss": 2.4449, "step": 554 }, { "epoch": 1.3875, "grad_norm": 6.4375, "learning_rate": 9.250000000000001e-06, "loss": 2.3218, "step": 555 }, { "epoch": 1.3900000000000001, "grad_norm": 5.03125, "learning_rate": 9.266666666666667e-06, "loss": 2.4162, "step": 556 }, { "epoch": 1.3925, "grad_norm": 4.9375, "learning_rate": 9.283333333333335e-06, "loss": 2.3885, "step": 557 }, { "epoch": 1.395, "grad_norm": 8.25, "learning_rate": 9.3e-06, "loss": 2.2674, "step": 558 }, { "epoch": 1.3975, "grad_norm": 6.75, "learning_rate": 9.316666666666667e-06, "loss": 2.4389, "step": 559 }, { "epoch": 1.4, "grad_norm": 5.84375, "learning_rate": 9.333333333333334e-06, "loss": 2.1591, "step": 560 }, { "epoch": 1.4025, "grad_norm": 6.46875, "learning_rate": 9.350000000000002e-06, "loss": 2.1581, "step": 561 }, { "epoch": 1.405, "grad_norm": 7.125, "learning_rate": 9.366666666666668e-06, "loss": 2.2492, "step": 562 }, { "epoch": 1.4075, "grad_norm": 7.15625, "learning_rate": 9.383333333333334e-06, "loss": 2.4621, "step": 563 }, { "epoch": 1.41, "grad_norm": 4.53125, "learning_rate": 9.4e-06, "loss": 2.3289, "step": 564 }, { "epoch": 1.4125, "grad_norm": 6.03125, "learning_rate": 9.416666666666667e-06, "loss": 2.2774, "step": 565 }, { "epoch": 1.415, "grad_norm": 4.21875, "learning_rate": 9.433333333333335e-06, "loss": 2.45, "step": 566 }, { "epoch": 1.4175, "grad_norm": 6.0625, "learning_rate": 9.450000000000001e-06, "loss": 2.4981, "step": 567 }, { "epoch": 1.42, "grad_norm": 5.53125, "learning_rate": 9.466666666666667e-06, "loss": 2.3383, "step": 568 }, { "epoch": 1.4224999999999999, "grad_norm": 6.28125, "learning_rate": 9.483333333333335e-06, "loss": 2.4441, "step": 569 }, { "epoch": 1.425, "grad_norm": 5.09375, "learning_rate": 9.5e-06, "loss": 2.024, "step": 570 }, { "epoch": 1.4275, "grad_norm": 5.03125, "learning_rate": 9.516666666666668e-06, "loss": 2.2529, "step": 571 }, { "epoch": 1.43, "grad_norm": 6.375, "learning_rate": 9.533333333333334e-06, "loss": 2.5773, "step": 572 }, { "epoch": 1.4325, "grad_norm": 7.40625, "learning_rate": 9.55e-06, "loss": 2.3121, "step": 573 }, { "epoch": 1.435, "grad_norm": 6.375, "learning_rate": 9.566666666666668e-06, "loss": 2.4891, "step": 574 }, { "epoch": 1.4375, "grad_norm": 6.65625, "learning_rate": 9.583333333333335e-06, "loss": 2.3392, "step": 575 }, { "epoch": 1.44, "grad_norm": 4.21875, "learning_rate": 9.600000000000001e-06, "loss": 2.1842, "step": 576 }, { "epoch": 1.4425, "grad_norm": 6.71875, "learning_rate": 9.616666666666667e-06, "loss": 2.1509, "step": 577 }, { "epoch": 1.445, "grad_norm": 8.0625, "learning_rate": 9.633333333333335e-06, "loss": 2.5773, "step": 578 }, { "epoch": 1.4475, "grad_norm": 4.0625, "learning_rate": 9.65e-06, "loss": 2.1325, "step": 579 }, { "epoch": 1.45, "grad_norm": 8.25, "learning_rate": 9.666666666666667e-06, "loss": 2.5568, "step": 580 }, { "epoch": 1.4525000000000001, "grad_norm": 7.03125, "learning_rate": 9.683333333333334e-06, "loss": 2.4501, "step": 581 }, { "epoch": 1.455, "grad_norm": 6.71875, "learning_rate": 9.7e-06, "loss": 2.236, "step": 582 }, { "epoch": 1.4575, "grad_norm": 5.25, "learning_rate": 9.716666666666668e-06, "loss": 2.2402, "step": 583 }, { "epoch": 1.46, "grad_norm": 6.125, "learning_rate": 9.733333333333334e-06, "loss": 2.2842, "step": 584 }, { "epoch": 1.4625, "grad_norm": 5.8125, "learning_rate": 9.75e-06, "loss": 2.3796, "step": 585 }, { "epoch": 1.465, "grad_norm": 7.46875, "learning_rate": 9.766666666666667e-06, "loss": 2.5746, "step": 586 }, { "epoch": 1.4675, "grad_norm": 4.9375, "learning_rate": 9.783333333333335e-06, "loss": 2.1611, "step": 587 }, { "epoch": 1.47, "grad_norm": 10.75, "learning_rate": 9.800000000000001e-06, "loss": 2.6208, "step": 588 }, { "epoch": 1.4725, "grad_norm": 5.65625, "learning_rate": 9.816666666666667e-06, "loss": 2.6307, "step": 589 }, { "epoch": 1.475, "grad_norm": 4.5, "learning_rate": 9.833333333333333e-06, "loss": 2.3625, "step": 590 }, { "epoch": 1.4775, "grad_norm": 43.25, "learning_rate": 9.85e-06, "loss": 2.2714, "step": 591 }, { "epoch": 1.48, "grad_norm": 6.59375, "learning_rate": 9.866666666666668e-06, "loss": 2.3563, "step": 592 }, { "epoch": 1.4825, "grad_norm": 5.28125, "learning_rate": 9.883333333333334e-06, "loss": 2.0076, "step": 593 }, { "epoch": 1.4849999999999999, "grad_norm": 4.46875, "learning_rate": 9.9e-06, "loss": 2.299, "step": 594 }, { "epoch": 1.4875, "grad_norm": 5.90625, "learning_rate": 9.916666666666668e-06, "loss": 2.3877, "step": 595 }, { "epoch": 1.49, "grad_norm": 7.34375, "learning_rate": 9.933333333333334e-06, "loss": 2.196, "step": 596 }, { "epoch": 1.4925, "grad_norm": 7.53125, "learning_rate": 9.950000000000001e-06, "loss": 2.5093, "step": 597 }, { "epoch": 1.495, "grad_norm": 5.09375, "learning_rate": 9.966666666666667e-06, "loss": 2.1674, "step": 598 }, { "epoch": 1.4975, "grad_norm": 7.875, "learning_rate": 9.983333333333333e-06, "loss": 2.1109, "step": 599 }, { "epoch": 1.5, "grad_norm": 17.375, "learning_rate": 1e-05, "loss": 2.5373, "step": 600 }, { "epoch": 1.5025, "grad_norm": 9.125, "learning_rate": 1.0016666666666668e-05, "loss": 2.4258, "step": 601 }, { "epoch": 1.505, "grad_norm": 4.6875, "learning_rate": 1.0033333333333334e-05, "loss": 2.2085, "step": 602 }, { "epoch": 1.5074999999999998, "grad_norm": 4.40625, "learning_rate": 1.005e-05, "loss": 2.1124, "step": 603 }, { "epoch": 1.51, "grad_norm": 4.78125, "learning_rate": 1.0066666666666666e-05, "loss": 2.1669, "step": 604 }, { "epoch": 1.5125, "grad_norm": 10.5, "learning_rate": 1.0083333333333334e-05, "loss": 2.4755, "step": 605 }, { "epoch": 1.5150000000000001, "grad_norm": 5.9375, "learning_rate": 1.0100000000000002e-05, "loss": 2.3747, "step": 606 }, { "epoch": 1.5175, "grad_norm": 3.890625, "learning_rate": 1.0116666666666667e-05, "loss": 2.3133, "step": 607 }, { "epoch": 1.52, "grad_norm": 4.03125, "learning_rate": 1.0133333333333335e-05, "loss": 1.9879, "step": 608 }, { "epoch": 1.5225, "grad_norm": 38.25, "learning_rate": 1.015e-05, "loss": 2.4532, "step": 609 }, { "epoch": 1.525, "grad_norm": 11.625, "learning_rate": 1.0166666666666667e-05, "loss": 2.4224, "step": 610 }, { "epoch": 1.5274999999999999, "grad_norm": 5.3125, "learning_rate": 1.0183333333333335e-05, "loss": 2.378, "step": 611 }, { "epoch": 1.53, "grad_norm": 4.96875, "learning_rate": 1.02e-05, "loss": 2.6499, "step": 612 }, { "epoch": 1.5325, "grad_norm": 8.5625, "learning_rate": 1.0216666666666668e-05, "loss": 2.5104, "step": 613 }, { "epoch": 1.5350000000000001, "grad_norm": 6.71875, "learning_rate": 1.0233333333333336e-05, "loss": 2.5348, "step": 614 }, { "epoch": 1.5375, "grad_norm": 5.09375, "learning_rate": 1.025e-05, "loss": 2.462, "step": 615 }, { "epoch": 1.54, "grad_norm": 4.59375, "learning_rate": 1.0266666666666668e-05, "loss": 2.2831, "step": 616 }, { "epoch": 1.5425, "grad_norm": 8.0625, "learning_rate": 1.0283333333333334e-05, "loss": 2.2952, "step": 617 }, { "epoch": 1.545, "grad_norm": 4.84375, "learning_rate": 1.0300000000000001e-05, "loss": 2.3241, "step": 618 }, { "epoch": 1.5474999999999999, "grad_norm": 4.5625, "learning_rate": 1.0316666666666669e-05, "loss": 2.0229, "step": 619 }, { "epoch": 1.55, "grad_norm": 9.1875, "learning_rate": 1.0333333333333335e-05, "loss": 2.3034, "step": 620 }, { "epoch": 1.5525, "grad_norm": 5.46875, "learning_rate": 1.0350000000000001e-05, "loss": 2.3536, "step": 621 }, { "epoch": 1.5550000000000002, "grad_norm": 5.625, "learning_rate": 1.0366666666666667e-05, "loss": 2.3762, "step": 622 }, { "epoch": 1.5575, "grad_norm": 3.765625, "learning_rate": 1.0383333333333334e-05, "loss": 2.3512, "step": 623 }, { "epoch": 1.56, "grad_norm": 6.5, "learning_rate": 1.04e-05, "loss": 2.399, "step": 624 }, { "epoch": 1.5625, "grad_norm": 5.0, "learning_rate": 1.0416666666666668e-05, "loss": 2.1594, "step": 625 }, { "epoch": 1.565, "grad_norm": 7.28125, "learning_rate": 1.0433333333333332e-05, "loss": 2.4941, "step": 626 }, { "epoch": 1.5675, "grad_norm": 5.5, "learning_rate": 1.045e-05, "loss": 2.4516, "step": 627 }, { "epoch": 1.5699999999999998, "grad_norm": 4.90625, "learning_rate": 1.0466666666666668e-05, "loss": 2.3817, "step": 628 }, { "epoch": 1.5725, "grad_norm": 7.59375, "learning_rate": 1.0483333333333333e-05, "loss": 2.292, "step": 629 }, { "epoch": 1.575, "grad_norm": 5.09375, "learning_rate": 1.0500000000000001e-05, "loss": 2.2381, "step": 630 }, { "epoch": 1.5775000000000001, "grad_norm": 5.28125, "learning_rate": 1.0516666666666669e-05, "loss": 2.4368, "step": 631 }, { "epoch": 1.58, "grad_norm": 7.8125, "learning_rate": 1.0533333333333333e-05, "loss": 2.2819, "step": 632 }, { "epoch": 1.5825, "grad_norm": 8.1875, "learning_rate": 1.055e-05, "loss": 2.4331, "step": 633 }, { "epoch": 1.585, "grad_norm": 8.4375, "learning_rate": 1.0566666666666667e-05, "loss": 2.2884, "step": 634 }, { "epoch": 1.5875, "grad_norm": 5.125, "learning_rate": 1.0583333333333334e-05, "loss": 2.3723, "step": 635 }, { "epoch": 1.5899999999999999, "grad_norm": 6.65625, "learning_rate": 1.0600000000000002e-05, "loss": 2.2401, "step": 636 }, { "epoch": 1.5925, "grad_norm": 7.9375, "learning_rate": 1.0616666666666668e-05, "loss": 2.308, "step": 637 }, { "epoch": 1.595, "grad_norm": 6.40625, "learning_rate": 1.0633333333333334e-05, "loss": 2.232, "step": 638 }, { "epoch": 1.5975000000000001, "grad_norm": 5.84375, "learning_rate": 1.065e-05, "loss": 2.3888, "step": 639 }, { "epoch": 1.6, "grad_norm": 5.3125, "learning_rate": 1.0666666666666667e-05, "loss": 2.2601, "step": 640 }, { "epoch": 1.6025, "grad_norm": 4.40625, "learning_rate": 1.0683333333333335e-05, "loss": 2.1007, "step": 641 }, { "epoch": 1.605, "grad_norm": 8.5625, "learning_rate": 1.0700000000000001e-05, "loss": 2.4201, "step": 642 }, { "epoch": 1.6075, "grad_norm": 6.125, "learning_rate": 1.0716666666666669e-05, "loss": 2.1391, "step": 643 }, { "epoch": 1.6099999999999999, "grad_norm": 4.78125, "learning_rate": 1.0733333333333333e-05, "loss": 2.0971, "step": 644 }, { "epoch": 1.6125, "grad_norm": 5.6875, "learning_rate": 1.075e-05, "loss": 2.5874, "step": 645 }, { "epoch": 1.615, "grad_norm": 4.25, "learning_rate": 1.0766666666666668e-05, "loss": 2.2886, "step": 646 }, { "epoch": 1.6175000000000002, "grad_norm": 3.453125, "learning_rate": 1.0783333333333334e-05, "loss": 2.2407, "step": 647 }, { "epoch": 1.62, "grad_norm": 8.125, "learning_rate": 1.0800000000000002e-05, "loss": 2.2976, "step": 648 }, { "epoch": 1.6225, "grad_norm": 6.59375, "learning_rate": 1.081666666666667e-05, "loss": 2.317, "step": 649 }, { "epoch": 1.625, "grad_norm": 12.375, "learning_rate": 1.0833333333333334e-05, "loss": 2.2563, "step": 650 }, { "epoch": 1.6275, "grad_norm": 8.25, "learning_rate": 1.0850000000000001e-05, "loss": 1.9967, "step": 651 }, { "epoch": 1.63, "grad_norm": 5.0, "learning_rate": 1.0866666666666667e-05, "loss": 2.1562, "step": 652 }, { "epoch": 1.6324999999999998, "grad_norm": 7.4375, "learning_rate": 1.0883333333333335e-05, "loss": 2.3256, "step": 653 }, { "epoch": 1.635, "grad_norm": 5.0625, "learning_rate": 1.0900000000000002e-05, "loss": 2.3647, "step": 654 }, { "epoch": 1.6375, "grad_norm": 6.875, "learning_rate": 1.0916666666666667e-05, "loss": 2.2289, "step": 655 }, { "epoch": 1.6400000000000001, "grad_norm": 5.4375, "learning_rate": 1.0933333333333334e-05, "loss": 2.1381, "step": 656 }, { "epoch": 1.6425, "grad_norm": 9.375, "learning_rate": 1.095e-05, "loss": 2.3215, "step": 657 }, { "epoch": 1.645, "grad_norm": 7.375, "learning_rate": 1.0966666666666668e-05, "loss": 2.2542, "step": 658 }, { "epoch": 1.6475, "grad_norm": 6.75, "learning_rate": 1.0983333333333335e-05, "loss": 2.2391, "step": 659 }, { "epoch": 1.65, "grad_norm": 4.28125, "learning_rate": 1.1000000000000001e-05, "loss": 2.1635, "step": 660 }, { "epoch": 1.6524999999999999, "grad_norm": 3.75, "learning_rate": 1.1016666666666667e-05, "loss": 2.1259, "step": 661 }, { "epoch": 1.655, "grad_norm": 7.53125, "learning_rate": 1.1033333333333333e-05, "loss": 2.2506, "step": 662 }, { "epoch": 1.6575, "grad_norm": 6.78125, "learning_rate": 1.1050000000000001e-05, "loss": 2.141, "step": 663 }, { "epoch": 1.6600000000000001, "grad_norm": 7.03125, "learning_rate": 1.1066666666666669e-05, "loss": 1.9565, "step": 664 }, { "epoch": 1.6625, "grad_norm": 4.53125, "learning_rate": 1.1083333333333335e-05, "loss": 2.0902, "step": 665 }, { "epoch": 1.665, "grad_norm": 5.03125, "learning_rate": 1.1100000000000002e-05, "loss": 2.2307, "step": 666 }, { "epoch": 1.6675, "grad_norm": 8.125, "learning_rate": 1.1116666666666666e-05, "loss": 2.3693, "step": 667 }, { "epoch": 1.67, "grad_norm": 3.796875, "learning_rate": 1.1133333333333334e-05, "loss": 2.1731, "step": 668 }, { "epoch": 1.6724999999999999, "grad_norm": 4.6875, "learning_rate": 1.1150000000000002e-05, "loss": 2.3818, "step": 669 }, { "epoch": 1.675, "grad_norm": 8.5625, "learning_rate": 1.1166666666666668e-05, "loss": 2.0544, "step": 670 }, { "epoch": 1.6775, "grad_norm": 4.90625, "learning_rate": 1.1183333333333335e-05, "loss": 2.1501, "step": 671 }, { "epoch": 1.6800000000000002, "grad_norm": 7.6875, "learning_rate": 1.1200000000000001e-05, "loss": 2.2736, "step": 672 }, { "epoch": 1.6825, "grad_norm": 4.75, "learning_rate": 1.1216666666666667e-05, "loss": 2.1984, "step": 673 }, { "epoch": 1.685, "grad_norm": 4.5625, "learning_rate": 1.1233333333333333e-05, "loss": 2.1837, "step": 674 }, { "epoch": 1.6875, "grad_norm": 8.0, "learning_rate": 1.125e-05, "loss": 2.3859, "step": 675 }, { "epoch": 1.69, "grad_norm": 10.6875, "learning_rate": 1.1266666666666668e-05, "loss": 2.2933, "step": 676 }, { "epoch": 1.6925, "grad_norm": 6.84375, "learning_rate": 1.1283333333333334e-05, "loss": 2.3159, "step": 677 }, { "epoch": 1.6949999999999998, "grad_norm": 10.0625, "learning_rate": 1.13e-05, "loss": 2.2536, "step": 678 }, { "epoch": 1.6975, "grad_norm": 5.9375, "learning_rate": 1.1316666666666666e-05, "loss": 2.4062, "step": 679 }, { "epoch": 1.7, "grad_norm": 7.8125, "learning_rate": 1.1333333333333334e-05, "loss": 2.3413, "step": 680 }, { "epoch": 1.7025000000000001, "grad_norm": 5.1875, "learning_rate": 1.1350000000000001e-05, "loss": 2.0114, "step": 681 }, { "epoch": 1.705, "grad_norm": 4.1875, "learning_rate": 1.1366666666666667e-05, "loss": 2.2618, "step": 682 }, { "epoch": 1.7075, "grad_norm": 8.6875, "learning_rate": 1.1383333333333335e-05, "loss": 2.4077, "step": 683 }, { "epoch": 1.71, "grad_norm": 7.46875, "learning_rate": 1.14e-05, "loss": 2.0117, "step": 684 }, { "epoch": 1.7125, "grad_norm": 4.84375, "learning_rate": 1.1416666666666667e-05, "loss": 2.0636, "step": 685 }, { "epoch": 1.7149999999999999, "grad_norm": 6.1875, "learning_rate": 1.1433333333333335e-05, "loss": 2.1709, "step": 686 }, { "epoch": 1.7175, "grad_norm": 11.6875, "learning_rate": 1.145e-05, "loss": 2.2053, "step": 687 }, { "epoch": 1.72, "grad_norm": 5.34375, "learning_rate": 1.1466666666666668e-05, "loss": 2.0969, "step": 688 }, { "epoch": 1.7225000000000001, "grad_norm": 15.25, "learning_rate": 1.1483333333333336e-05, "loss": 2.3095, "step": 689 }, { "epoch": 1.725, "grad_norm": 5.15625, "learning_rate": 1.15e-05, "loss": 2.3641, "step": 690 }, { "epoch": 1.7275, "grad_norm": 9.6875, "learning_rate": 1.1516666666666668e-05, "loss": 2.4563, "step": 691 }, { "epoch": 1.73, "grad_norm": 10.5625, "learning_rate": 1.1533333333333334e-05, "loss": 2.4586, "step": 692 }, { "epoch": 1.7325, "grad_norm": 6.1875, "learning_rate": 1.1550000000000001e-05, "loss": 2.2677, "step": 693 }, { "epoch": 1.7349999999999999, "grad_norm": 7.09375, "learning_rate": 1.1566666666666669e-05, "loss": 2.1204, "step": 694 }, { "epoch": 1.7375, "grad_norm": 15.625, "learning_rate": 1.1583333333333335e-05, "loss": 2.4552, "step": 695 }, { "epoch": 1.74, "grad_norm": 6.03125, "learning_rate": 1.16e-05, "loss": 2.1939, "step": 696 }, { "epoch": 1.7425000000000002, "grad_norm": 4.625, "learning_rate": 1.1616666666666667e-05, "loss": 2.0358, "step": 697 }, { "epoch": 1.745, "grad_norm": 6.28125, "learning_rate": 1.1633333333333334e-05, "loss": 2.2745, "step": 698 }, { "epoch": 1.7475, "grad_norm": 6.625, "learning_rate": 1.1650000000000002e-05, "loss": 2.2432, "step": 699 }, { "epoch": 1.75, "grad_norm": 5.0, "learning_rate": 1.1666666666666668e-05, "loss": 2.2842, "step": 700 }, { "epoch": 1.7525, "grad_norm": 5.25, "learning_rate": 1.1683333333333334e-05, "loss": 2.2688, "step": 701 }, { "epoch": 1.755, "grad_norm": 7.03125, "learning_rate": 1.17e-05, "loss": 2.2592, "step": 702 }, { "epoch": 1.7574999999999998, "grad_norm": 5.59375, "learning_rate": 1.1716666666666667e-05, "loss": 2.338, "step": 703 }, { "epoch": 1.76, "grad_norm": 6.96875, "learning_rate": 1.1733333333333335e-05, "loss": 2.0749, "step": 704 }, { "epoch": 1.7625, "grad_norm": 6.4375, "learning_rate": 1.1750000000000001e-05, "loss": 2.2665, "step": 705 }, { "epoch": 1.7650000000000001, "grad_norm": 7.15625, "learning_rate": 1.1766666666666669e-05, "loss": 2.2906, "step": 706 }, { "epoch": 1.7675, "grad_norm": 6.78125, "learning_rate": 1.1783333333333333e-05, "loss": 2.1754, "step": 707 }, { "epoch": 1.77, "grad_norm": 4.46875, "learning_rate": 1.18e-05, "loss": 2.1833, "step": 708 }, { "epoch": 1.7725, "grad_norm": 8.0625, "learning_rate": 1.1816666666666668e-05, "loss": 2.4032, "step": 709 }, { "epoch": 1.775, "grad_norm": 5.46875, "learning_rate": 1.1833333333333334e-05, "loss": 2.48, "step": 710 }, { "epoch": 1.7774999999999999, "grad_norm": 9.75, "learning_rate": 1.1850000000000002e-05, "loss": 2.2657, "step": 711 }, { "epoch": 1.78, "grad_norm": 4.34375, "learning_rate": 1.186666666666667e-05, "loss": 2.039, "step": 712 }, { "epoch": 1.7825, "grad_norm": 4.40625, "learning_rate": 1.1883333333333334e-05, "loss": 2.0582, "step": 713 }, { "epoch": 1.7850000000000001, "grad_norm": 6.0, "learning_rate": 1.1900000000000001e-05, "loss": 2.0916, "step": 714 }, { "epoch": 1.7875, "grad_norm": 4.125, "learning_rate": 1.1916666666666667e-05, "loss": 1.975, "step": 715 }, { "epoch": 1.79, "grad_norm": 7.3125, "learning_rate": 1.1933333333333335e-05, "loss": 2.2233, "step": 716 }, { "epoch": 1.7925, "grad_norm": 5.1875, "learning_rate": 1.195e-05, "loss": 2.2758, "step": 717 }, { "epoch": 1.795, "grad_norm": 8.375, "learning_rate": 1.1966666666666668e-05, "loss": 2.5058, "step": 718 }, { "epoch": 1.7974999999999999, "grad_norm": 12.25, "learning_rate": 1.1983333333333333e-05, "loss": 2.2842, "step": 719 }, { "epoch": 1.8, "grad_norm": 4.21875, "learning_rate": 1.2e-05, "loss": 2.2625, "step": 720 }, { "epoch": 1.8025, "grad_norm": 4.90625, "learning_rate": 1.2016666666666668e-05, "loss": 2.0808, "step": 721 }, { "epoch": 1.8050000000000002, "grad_norm": 4.375, "learning_rate": 1.2033333333333334e-05, "loss": 2.3278, "step": 722 }, { "epoch": 1.8075, "grad_norm": 4.75, "learning_rate": 1.2050000000000002e-05, "loss": 2.0731, "step": 723 }, { "epoch": 1.81, "grad_norm": 10.75, "learning_rate": 1.206666666666667e-05, "loss": 2.2009, "step": 724 }, { "epoch": 1.8125, "grad_norm": 4.75, "learning_rate": 1.2083333333333333e-05, "loss": 2.1008, "step": 725 }, { "epoch": 1.815, "grad_norm": 3.609375, "learning_rate": 1.2100000000000001e-05, "loss": 2.1235, "step": 726 }, { "epoch": 1.8175, "grad_norm": 5.28125, "learning_rate": 1.2116666666666667e-05, "loss": 2.1198, "step": 727 }, { "epoch": 1.8199999999999998, "grad_norm": 3.546875, "learning_rate": 1.2133333333333335e-05, "loss": 2.0826, "step": 728 }, { "epoch": 1.8225, "grad_norm": 5.53125, "learning_rate": 1.2150000000000002e-05, "loss": 2.2459, "step": 729 }, { "epoch": 1.825, "grad_norm": 4.59375, "learning_rate": 1.2166666666666667e-05, "loss": 2.1, "step": 730 }, { "epoch": 1.8275000000000001, "grad_norm": 4.375, "learning_rate": 1.2183333333333334e-05, "loss": 2.201, "step": 731 }, { "epoch": 1.83, "grad_norm": 5.59375, "learning_rate": 1.22e-05, "loss": 2.0224, "step": 732 }, { "epoch": 1.8325, "grad_norm": 6.5, "learning_rate": 1.2216666666666668e-05, "loss": 2.1819, "step": 733 }, { "epoch": 1.835, "grad_norm": 5.6875, "learning_rate": 1.2233333333333335e-05, "loss": 2.1169, "step": 734 }, { "epoch": 1.8375, "grad_norm": 4.65625, "learning_rate": 1.2250000000000001e-05, "loss": 2.2904, "step": 735 }, { "epoch": 1.8399999999999999, "grad_norm": 4.9375, "learning_rate": 1.2266666666666667e-05, "loss": 2.4535, "step": 736 }, { "epoch": 1.8425, "grad_norm": 5.625, "learning_rate": 1.2283333333333333e-05, "loss": 2.0729, "step": 737 }, { "epoch": 1.845, "grad_norm": 5.375, "learning_rate": 1.23e-05, "loss": 1.9899, "step": 738 }, { "epoch": 1.8475000000000001, "grad_norm": 4.15625, "learning_rate": 1.2316666666666668e-05, "loss": 1.9903, "step": 739 }, { "epoch": 1.85, "grad_norm": 4.71875, "learning_rate": 1.2333333333333334e-05, "loss": 2.1439, "step": 740 }, { "epoch": 1.8525, "grad_norm": 7.21875, "learning_rate": 1.2350000000000002e-05, "loss": 2.063, "step": 741 }, { "epoch": 1.855, "grad_norm": 4.59375, "learning_rate": 1.2366666666666666e-05, "loss": 2.1982, "step": 742 }, { "epoch": 1.8575, "grad_norm": 4.4375, "learning_rate": 1.2383333333333334e-05, "loss": 2.1986, "step": 743 }, { "epoch": 1.8599999999999999, "grad_norm": 7.03125, "learning_rate": 1.2400000000000002e-05, "loss": 2.4128, "step": 744 }, { "epoch": 1.8625, "grad_norm": 6.4375, "learning_rate": 1.2416666666666667e-05, "loss": 2.5702, "step": 745 }, { "epoch": 1.865, "grad_norm": 5.9375, "learning_rate": 1.2433333333333335e-05, "loss": 2.1336, "step": 746 }, { "epoch": 1.8675000000000002, "grad_norm": 8.125, "learning_rate": 1.2450000000000003e-05, "loss": 2.0909, "step": 747 }, { "epoch": 1.87, "grad_norm": 5.125, "learning_rate": 1.2466666666666667e-05, "loss": 2.2201, "step": 748 }, { "epoch": 1.8725, "grad_norm": 6.53125, "learning_rate": 1.2483333333333335e-05, "loss": 2.4892, "step": 749 }, { "epoch": 1.875, "grad_norm": 7.96875, "learning_rate": 1.25e-05, "loss": 2.179, "step": 750 }, { "epoch": 1.8775, "grad_norm": 4.28125, "learning_rate": 1.2516666666666668e-05, "loss": 2.1378, "step": 751 }, { "epoch": 1.88, "grad_norm": 4.5625, "learning_rate": 1.2533333333333336e-05, "loss": 2.11, "step": 752 }, { "epoch": 1.8824999999999998, "grad_norm": 8.5, "learning_rate": 1.255e-05, "loss": 2.2398, "step": 753 }, { "epoch": 1.885, "grad_norm": 4.65625, "learning_rate": 1.2566666666666668e-05, "loss": 2.1582, "step": 754 }, { "epoch": 1.8875, "grad_norm": 6.3125, "learning_rate": 1.2583333333333334e-05, "loss": 2.1649, "step": 755 }, { "epoch": 1.8900000000000001, "grad_norm": 5.1875, "learning_rate": 1.2600000000000001e-05, "loss": 2.3324, "step": 756 }, { "epoch": 1.8925, "grad_norm": 5.84375, "learning_rate": 1.2616666666666669e-05, "loss": 2.2249, "step": 757 }, { "epoch": 1.895, "grad_norm": 7.75, "learning_rate": 1.2633333333333335e-05, "loss": 2.1318, "step": 758 }, { "epoch": 1.8975, "grad_norm": 6.28125, "learning_rate": 1.2650000000000001e-05, "loss": 2.155, "step": 759 }, { "epoch": 1.9, "grad_norm": 15.6875, "learning_rate": 1.2666666666666667e-05, "loss": 2.34, "step": 760 }, { "epoch": 1.9024999999999999, "grad_norm": 5.78125, "learning_rate": 1.2683333333333334e-05, "loss": 2.0971, "step": 761 }, { "epoch": 1.905, "grad_norm": 6.875, "learning_rate": 1.27e-05, "loss": 2.2255, "step": 762 }, { "epoch": 1.9075, "grad_norm": 5.125, "learning_rate": 1.2716666666666668e-05, "loss": 2.1014, "step": 763 }, { "epoch": 1.9100000000000001, "grad_norm": 4.0625, "learning_rate": 1.2733333333333336e-05, "loss": 2.2357, "step": 764 }, { "epoch": 1.9125, "grad_norm": 4.875, "learning_rate": 1.275e-05, "loss": 2.1801, "step": 765 }, { "epoch": 1.915, "grad_norm": 5.46875, "learning_rate": 1.2766666666666668e-05, "loss": 2.016, "step": 766 }, { "epoch": 1.9175, "grad_norm": 5.03125, "learning_rate": 1.2783333333333333e-05, "loss": 2.1122, "step": 767 }, { "epoch": 1.92, "grad_norm": 6.65625, "learning_rate": 1.2800000000000001e-05, "loss": 2.0548, "step": 768 }, { "epoch": 1.9224999999999999, "grad_norm": 4.21875, "learning_rate": 1.2816666666666669e-05, "loss": 2.0895, "step": 769 }, { "epoch": 1.925, "grad_norm": 4.5625, "learning_rate": 1.2833333333333335e-05, "loss": 2.1282, "step": 770 }, { "epoch": 1.9275, "grad_norm": 5.59375, "learning_rate": 1.285e-05, "loss": 2.2298, "step": 771 }, { "epoch": 1.9300000000000002, "grad_norm": 4.6875, "learning_rate": 1.2866666666666667e-05, "loss": 2.3166, "step": 772 }, { "epoch": 1.9325, "grad_norm": 5.625, "learning_rate": 1.2883333333333334e-05, "loss": 2.2222, "step": 773 }, { "epoch": 1.935, "grad_norm": 5.96875, "learning_rate": 1.2900000000000002e-05, "loss": 2.286, "step": 774 }, { "epoch": 1.9375, "grad_norm": 6.09375, "learning_rate": 1.2916666666666668e-05, "loss": 2.2687, "step": 775 }, { "epoch": 1.94, "grad_norm": 7.0625, "learning_rate": 1.2933333333333334e-05, "loss": 2.0837, "step": 776 }, { "epoch": 1.9425, "grad_norm": 6.21875, "learning_rate": 1.295e-05, "loss": 2.1631, "step": 777 }, { "epoch": 1.9449999999999998, "grad_norm": 16.5, "learning_rate": 1.2966666666666667e-05, "loss": 2.1654, "step": 778 }, { "epoch": 1.9475, "grad_norm": 6.0625, "learning_rate": 1.2983333333333335e-05, "loss": 2.2467, "step": 779 }, { "epoch": 1.95, "grad_norm": 5.5625, "learning_rate": 1.3000000000000001e-05, "loss": 2.2724, "step": 780 }, { "epoch": 1.9525000000000001, "grad_norm": 8.8125, "learning_rate": 1.3016666666666669e-05, "loss": 2.1467, "step": 781 }, { "epoch": 1.955, "grad_norm": 7.6875, "learning_rate": 1.3033333333333333e-05, "loss": 2.1383, "step": 782 }, { "epoch": 1.9575, "grad_norm": 5.9375, "learning_rate": 1.305e-05, "loss": 2.2323, "step": 783 }, { "epoch": 1.96, "grad_norm": 7.59375, "learning_rate": 1.3066666666666668e-05, "loss": 2.2405, "step": 784 }, { "epoch": 1.9625, "grad_norm": 7.40625, "learning_rate": 1.3083333333333334e-05, "loss": 2.2214, "step": 785 }, { "epoch": 1.9649999999999999, "grad_norm": 6.875, "learning_rate": 1.3100000000000002e-05, "loss": 2.1078, "step": 786 }, { "epoch": 1.9675, "grad_norm": 5.90625, "learning_rate": 1.311666666666667e-05, "loss": 2.2157, "step": 787 }, { "epoch": 1.97, "grad_norm": 5.5625, "learning_rate": 1.3133333333333334e-05, "loss": 2.0693, "step": 788 }, { "epoch": 1.9725000000000001, "grad_norm": 7.25, "learning_rate": 1.3150000000000001e-05, "loss": 2.0793, "step": 789 }, { "epoch": 1.975, "grad_norm": 4.90625, "learning_rate": 1.3166666666666667e-05, "loss": 1.9154, "step": 790 }, { "epoch": 1.9775, "grad_norm": 4.21875, "learning_rate": 1.3183333333333335e-05, "loss": 2.0583, "step": 791 }, { "epoch": 1.98, "grad_norm": 5.875, "learning_rate": 1.3200000000000002e-05, "loss": 2.3508, "step": 792 }, { "epoch": 1.9825, "grad_norm": 4.375, "learning_rate": 1.3216666666666668e-05, "loss": 2.053, "step": 793 }, { "epoch": 1.9849999999999999, "grad_norm": 5.375, "learning_rate": 1.3233333333333334e-05, "loss": 2.0851, "step": 794 }, { "epoch": 1.9875, "grad_norm": 3.765625, "learning_rate": 1.325e-05, "loss": 2.3285, "step": 795 }, { "epoch": 1.99, "grad_norm": 3.3125, "learning_rate": 1.3266666666666668e-05, "loss": 1.7628, "step": 796 }, { "epoch": 1.9925000000000002, "grad_norm": 5.96875, "learning_rate": 1.3283333333333335e-05, "loss": 2.4103, "step": 797 }, { "epoch": 1.995, "grad_norm": 5.71875, "learning_rate": 1.3300000000000001e-05, "loss": 2.2174, "step": 798 }, { "epoch": 1.9975, "grad_norm": 4.34375, "learning_rate": 1.3316666666666669e-05, "loss": 2.0302, "step": 799 }, { "epoch": 2.0, "grad_norm": 4.9375, "learning_rate": 1.3333333333333333e-05, "loss": 2.1977, "step": 800 }, { "epoch": 2.0025, "grad_norm": 5.21875, "learning_rate": 1.3350000000000001e-05, "loss": 2.1019, "step": 801 }, { "epoch": 2.005, "grad_norm": 8.0, "learning_rate": 1.3366666666666669e-05, "loss": 1.8904, "step": 802 }, { "epoch": 2.0075, "grad_norm": 6.0625, "learning_rate": 1.3383333333333335e-05, "loss": 2.0538, "step": 803 }, { "epoch": 2.01, "grad_norm": 4.625, "learning_rate": 1.3400000000000002e-05, "loss": 2.1459, "step": 804 }, { "epoch": 2.0125, "grad_norm": 5.53125, "learning_rate": 1.3416666666666666e-05, "loss": 2.2285, "step": 805 }, { "epoch": 2.015, "grad_norm": 3.859375, "learning_rate": 1.3433333333333334e-05, "loss": 2.0201, "step": 806 }, { "epoch": 2.0175, "grad_norm": 10.25, "learning_rate": 1.3450000000000002e-05, "loss": 2.2488, "step": 807 }, { "epoch": 2.02, "grad_norm": 5.0625, "learning_rate": 1.3466666666666668e-05, "loss": 2.1743, "step": 808 }, { "epoch": 2.0225, "grad_norm": 4.125, "learning_rate": 1.3483333333333335e-05, "loss": 2.1689, "step": 809 }, { "epoch": 2.025, "grad_norm": 5.9375, "learning_rate": 1.3500000000000001e-05, "loss": 2.1496, "step": 810 }, { "epoch": 2.0275, "grad_norm": 4.46875, "learning_rate": 1.3516666666666667e-05, "loss": 2.0628, "step": 811 }, { "epoch": 2.03, "grad_norm": 4.15625, "learning_rate": 1.3533333333333333e-05, "loss": 2.1093, "step": 812 }, { "epoch": 2.0325, "grad_norm": 4.40625, "learning_rate": 1.355e-05, "loss": 2.0665, "step": 813 }, { "epoch": 2.035, "grad_norm": 5.5625, "learning_rate": 1.3566666666666668e-05, "loss": 2.2955, "step": 814 }, { "epoch": 2.0375, "grad_norm": 5.5, "learning_rate": 1.3583333333333334e-05, "loss": 2.0253, "step": 815 }, { "epoch": 2.04, "grad_norm": 6.6875, "learning_rate": 1.3600000000000002e-05, "loss": 2.2427, "step": 816 }, { "epoch": 2.0425, "grad_norm": 6.75, "learning_rate": 1.3616666666666666e-05, "loss": 2.0945, "step": 817 }, { "epoch": 2.045, "grad_norm": 6.90625, "learning_rate": 1.3633333333333334e-05, "loss": 2.0718, "step": 818 }, { "epoch": 2.0475, "grad_norm": 5.90625, "learning_rate": 1.3650000000000001e-05, "loss": 2.2301, "step": 819 }, { "epoch": 2.05, "grad_norm": 6.3125, "learning_rate": 1.3666666666666667e-05, "loss": 2.0974, "step": 820 }, { "epoch": 2.0525, "grad_norm": 6.375, "learning_rate": 1.3683333333333335e-05, "loss": 2.0239, "step": 821 }, { "epoch": 2.055, "grad_norm": 5.25, "learning_rate": 1.3700000000000003e-05, "loss": 2.0914, "step": 822 }, { "epoch": 2.0575, "grad_norm": 4.875, "learning_rate": 1.3716666666666667e-05, "loss": 2.2125, "step": 823 }, { "epoch": 2.06, "grad_norm": 7.40625, "learning_rate": 1.3733333333333335e-05, "loss": 2.1213, "step": 824 }, { "epoch": 2.0625, "grad_norm": 5.25, "learning_rate": 1.375e-05, "loss": 1.9855, "step": 825 }, { "epoch": 2.065, "grad_norm": 5.0625, "learning_rate": 1.3766666666666668e-05, "loss": 2.237, "step": 826 }, { "epoch": 2.0675, "grad_norm": 4.6875, "learning_rate": 1.3783333333333336e-05, "loss": 1.9571, "step": 827 }, { "epoch": 2.07, "grad_norm": 6.8125, "learning_rate": 1.38e-05, "loss": 2.2941, "step": 828 }, { "epoch": 2.0725, "grad_norm": 5.5625, "learning_rate": 1.3816666666666668e-05, "loss": 2.3408, "step": 829 }, { "epoch": 2.075, "grad_norm": 5.8125, "learning_rate": 1.3833333333333334e-05, "loss": 2.052, "step": 830 }, { "epoch": 2.0775, "grad_norm": 9.5625, "learning_rate": 1.3850000000000001e-05, "loss": 2.1599, "step": 831 }, { "epoch": 2.08, "grad_norm": 4.4375, "learning_rate": 1.3866666666666669e-05, "loss": 2.0871, "step": 832 }, { "epoch": 2.0825, "grad_norm": 4.65625, "learning_rate": 1.3883333333333335e-05, "loss": 2.0995, "step": 833 }, { "epoch": 2.085, "grad_norm": 4.28125, "learning_rate": 1.39e-05, "loss": 2.0482, "step": 834 }, { "epoch": 2.0875, "grad_norm": 5.46875, "learning_rate": 1.3916666666666667e-05, "loss": 2.0197, "step": 835 }, { "epoch": 2.09, "grad_norm": 6.46875, "learning_rate": 1.3933333333333334e-05, "loss": 2.2098, "step": 836 }, { "epoch": 2.0925, "grad_norm": 5.46875, "learning_rate": 1.3950000000000002e-05, "loss": 2.0465, "step": 837 }, { "epoch": 2.095, "grad_norm": 9.625, "learning_rate": 1.3966666666666668e-05, "loss": 2.2213, "step": 838 }, { "epoch": 2.0975, "grad_norm": 7.84375, "learning_rate": 1.3983333333333336e-05, "loss": 2.1863, "step": 839 }, { "epoch": 2.1, "grad_norm": 5.375, "learning_rate": 1.4e-05, "loss": 2.2154, "step": 840 }, { "epoch": 2.1025, "grad_norm": 4.75, "learning_rate": 1.4016666666666667e-05, "loss": 2.1147, "step": 841 }, { "epoch": 2.105, "grad_norm": 6.4375, "learning_rate": 1.4033333333333335e-05, "loss": 1.9826, "step": 842 }, { "epoch": 2.1075, "grad_norm": 8.4375, "learning_rate": 1.4050000000000001e-05, "loss": 2.3034, "step": 843 }, { "epoch": 2.11, "grad_norm": 5.34375, "learning_rate": 1.4066666666666669e-05, "loss": 2.1513, "step": 844 }, { "epoch": 2.1125, "grad_norm": 5.0, "learning_rate": 1.4083333333333336e-05, "loss": 2.2191, "step": 845 }, { "epoch": 2.115, "grad_norm": 4.9375, "learning_rate": 1.41e-05, "loss": 1.8972, "step": 846 }, { "epoch": 2.1175, "grad_norm": 11.0625, "learning_rate": 1.4116666666666668e-05, "loss": 2.2284, "step": 847 }, { "epoch": 2.12, "grad_norm": 10.8125, "learning_rate": 1.4133333333333334e-05, "loss": 2.3602, "step": 848 }, { "epoch": 2.1225, "grad_norm": 5.90625, "learning_rate": 1.4150000000000002e-05, "loss": 2.1054, "step": 849 }, { "epoch": 2.125, "grad_norm": 4.21875, "learning_rate": 1.416666666666667e-05, "loss": 1.9299, "step": 850 }, { "epoch": 2.1275, "grad_norm": 5.15625, "learning_rate": 1.4183333333333334e-05, "loss": 2.1605, "step": 851 }, { "epoch": 2.13, "grad_norm": 5.96875, "learning_rate": 1.4200000000000001e-05, "loss": 2.1391, "step": 852 }, { "epoch": 2.1325, "grad_norm": 5.1875, "learning_rate": 1.4216666666666667e-05, "loss": 2.1659, "step": 853 }, { "epoch": 2.135, "grad_norm": 6.09375, "learning_rate": 1.4233333333333335e-05, "loss": 2.0537, "step": 854 }, { "epoch": 2.1375, "grad_norm": 5.03125, "learning_rate": 1.425e-05, "loss": 1.9866, "step": 855 }, { "epoch": 2.14, "grad_norm": 5.0, "learning_rate": 1.4266666666666668e-05, "loss": 1.9441, "step": 856 }, { "epoch": 2.1425, "grad_norm": 3.8125, "learning_rate": 1.4283333333333333e-05, "loss": 2.175, "step": 857 }, { "epoch": 2.145, "grad_norm": 6.53125, "learning_rate": 1.43e-05, "loss": 2.1124, "step": 858 }, { "epoch": 2.1475, "grad_norm": 5.75, "learning_rate": 1.4316666666666668e-05, "loss": 1.9806, "step": 859 }, { "epoch": 2.15, "grad_norm": 6.09375, "learning_rate": 1.4333333333333334e-05, "loss": 2.1242, "step": 860 }, { "epoch": 2.1525, "grad_norm": 10.4375, "learning_rate": 1.4350000000000002e-05, "loss": 2.2417, "step": 861 }, { "epoch": 2.155, "grad_norm": 12.3125, "learning_rate": 1.436666666666667e-05, "loss": 2.0964, "step": 862 }, { "epoch": 2.1575, "grad_norm": 8.1875, "learning_rate": 1.4383333333333333e-05, "loss": 1.8285, "step": 863 }, { "epoch": 2.16, "grad_norm": 6.09375, "learning_rate": 1.4400000000000001e-05, "loss": 2.1551, "step": 864 }, { "epoch": 2.1625, "grad_norm": 5.3125, "learning_rate": 1.4416666666666667e-05, "loss": 1.8806, "step": 865 }, { "epoch": 2.165, "grad_norm": 5.0, "learning_rate": 1.4433333333333335e-05, "loss": 2.0913, "step": 866 }, { "epoch": 2.1675, "grad_norm": 3.78125, "learning_rate": 1.4450000000000002e-05, "loss": 2.0992, "step": 867 }, { "epoch": 2.17, "grad_norm": 5.71875, "learning_rate": 1.4466666666666668e-05, "loss": 2.0695, "step": 868 }, { "epoch": 2.1725, "grad_norm": 5.90625, "learning_rate": 1.4483333333333334e-05, "loss": 1.9942, "step": 869 }, { "epoch": 2.175, "grad_norm": 7.28125, "learning_rate": 1.45e-05, "loss": 2.1465, "step": 870 }, { "epoch": 2.1775, "grad_norm": 3.703125, "learning_rate": 1.4516666666666668e-05, "loss": 1.8938, "step": 871 }, { "epoch": 2.18, "grad_norm": 4.34375, "learning_rate": 1.4533333333333335e-05, "loss": 1.9392, "step": 872 }, { "epoch": 2.1825, "grad_norm": 4.90625, "learning_rate": 1.4550000000000001e-05, "loss": 2.0911, "step": 873 }, { "epoch": 2.185, "grad_norm": 5.34375, "learning_rate": 1.4566666666666669e-05, "loss": 1.9249, "step": 874 }, { "epoch": 2.1875, "grad_norm": 9.625, "learning_rate": 1.4583333333333333e-05, "loss": 2.1404, "step": 875 }, { "epoch": 2.19, "grad_norm": 7.375, "learning_rate": 1.46e-05, "loss": 2.1294, "step": 876 }, { "epoch": 2.1925, "grad_norm": 5.78125, "learning_rate": 1.4616666666666668e-05, "loss": 2.0514, "step": 877 }, { "epoch": 2.195, "grad_norm": 7.09375, "learning_rate": 1.4633333333333334e-05, "loss": 2.3339, "step": 878 }, { "epoch": 2.1975, "grad_norm": 30.5, "learning_rate": 1.4650000000000002e-05, "loss": 2.3157, "step": 879 }, { "epoch": 2.2, "grad_norm": 7.71875, "learning_rate": 1.4666666666666666e-05, "loss": 2.0141, "step": 880 }, { "epoch": 2.2025, "grad_norm": 5.59375, "learning_rate": 1.4683333333333334e-05, "loss": 2.1851, "step": 881 }, { "epoch": 2.205, "grad_norm": 4.96875, "learning_rate": 1.4700000000000002e-05, "loss": 2.2142, "step": 882 }, { "epoch": 2.2075, "grad_norm": 5.03125, "learning_rate": 1.4716666666666668e-05, "loss": 2.0211, "step": 883 }, { "epoch": 2.21, "grad_norm": 8.0, "learning_rate": 1.4733333333333335e-05, "loss": 2.193, "step": 884 }, { "epoch": 2.2125, "grad_norm": 5.0625, "learning_rate": 1.4750000000000003e-05, "loss": 2.0376, "step": 885 }, { "epoch": 2.215, "grad_norm": 4.53125, "learning_rate": 1.4766666666666667e-05, "loss": 1.9365, "step": 886 }, { "epoch": 2.2175, "grad_norm": 5.03125, "learning_rate": 1.4783333333333335e-05, "loss": 2.113, "step": 887 }, { "epoch": 2.22, "grad_norm": 4.1875, "learning_rate": 1.48e-05, "loss": 1.8859, "step": 888 }, { "epoch": 2.2225, "grad_norm": 4.9375, "learning_rate": 1.4816666666666668e-05, "loss": 1.7935, "step": 889 }, { "epoch": 2.225, "grad_norm": 5.40625, "learning_rate": 1.4833333333333336e-05, "loss": 2.0552, "step": 890 }, { "epoch": 2.2275, "grad_norm": 9.375, "learning_rate": 1.4850000000000002e-05, "loss": 2.116, "step": 891 }, { "epoch": 2.23, "grad_norm": 6.3125, "learning_rate": 1.4866666666666668e-05, "loss": 2.0169, "step": 892 }, { "epoch": 2.2325, "grad_norm": 5.96875, "learning_rate": 1.4883333333333334e-05, "loss": 1.931, "step": 893 }, { "epoch": 2.235, "grad_norm": 5.75, "learning_rate": 1.4900000000000001e-05, "loss": 2.0556, "step": 894 }, { "epoch": 2.2375, "grad_norm": 10.125, "learning_rate": 1.4916666666666669e-05, "loss": 2.1084, "step": 895 }, { "epoch": 2.24, "grad_norm": 6.03125, "learning_rate": 1.4933333333333335e-05, "loss": 1.9779, "step": 896 }, { "epoch": 2.2425, "grad_norm": 8.0, "learning_rate": 1.4950000000000003e-05, "loss": 1.967, "step": 897 }, { "epoch": 2.245, "grad_norm": 4.625, "learning_rate": 1.4966666666666667e-05, "loss": 2.2011, "step": 898 }, { "epoch": 2.2475, "grad_norm": 6.21875, "learning_rate": 1.4983333333333334e-05, "loss": 2.1506, "step": 899 }, { "epoch": 2.25, "grad_norm": 6.84375, "learning_rate": 1.5000000000000002e-05, "loss": 2.1612, "step": 900 }, { "epoch": 2.2525, "grad_norm": 5.40625, "learning_rate": 1.5016666666666668e-05, "loss": 2.2524, "step": 901 }, { "epoch": 2.255, "grad_norm": 4.875, "learning_rate": 1.5033333333333336e-05, "loss": 2.142, "step": 902 }, { "epoch": 2.2575, "grad_norm": 5.46875, "learning_rate": 1.505e-05, "loss": 2.0696, "step": 903 }, { "epoch": 2.26, "grad_norm": 6.3125, "learning_rate": 1.5066666666666668e-05, "loss": 1.9331, "step": 904 }, { "epoch": 2.2625, "grad_norm": 5.125, "learning_rate": 1.5083333333333333e-05, "loss": 1.869, "step": 905 }, { "epoch": 2.265, "grad_norm": 6.0, "learning_rate": 1.5100000000000001e-05, "loss": 2.1445, "step": 906 }, { "epoch": 2.2675, "grad_norm": 7.28125, "learning_rate": 1.5116666666666669e-05, "loss": 2.1843, "step": 907 }, { "epoch": 2.27, "grad_norm": 8.125, "learning_rate": 1.5133333333333335e-05, "loss": 2.2059, "step": 908 }, { "epoch": 2.2725, "grad_norm": 8.8125, "learning_rate": 1.515e-05, "loss": 2.0744, "step": 909 }, { "epoch": 2.275, "grad_norm": 6.875, "learning_rate": 1.5166666666666667e-05, "loss": 2.0109, "step": 910 }, { "epoch": 2.2775, "grad_norm": 6.4375, "learning_rate": 1.5183333333333334e-05, "loss": 2.2395, "step": 911 }, { "epoch": 2.2800000000000002, "grad_norm": 8.0, "learning_rate": 1.5200000000000002e-05, "loss": 1.9761, "step": 912 }, { "epoch": 2.2824999999999998, "grad_norm": 4.71875, "learning_rate": 1.5216666666666668e-05, "loss": 1.9362, "step": 913 }, { "epoch": 2.285, "grad_norm": 5.25, "learning_rate": 1.5233333333333335e-05, "loss": 2.1243, "step": 914 }, { "epoch": 2.2875, "grad_norm": 5.6875, "learning_rate": 1.525e-05, "loss": 2.0869, "step": 915 }, { "epoch": 2.29, "grad_norm": 5.25, "learning_rate": 1.5266666666666667e-05, "loss": 2.0226, "step": 916 }, { "epoch": 2.2925, "grad_norm": 5.5, "learning_rate": 1.5283333333333333e-05, "loss": 2.0657, "step": 917 }, { "epoch": 2.295, "grad_norm": 5.6875, "learning_rate": 1.5300000000000003e-05, "loss": 2.09, "step": 918 }, { "epoch": 2.2975, "grad_norm": 5.09375, "learning_rate": 1.531666666666667e-05, "loss": 2.0657, "step": 919 }, { "epoch": 2.3, "grad_norm": 6.375, "learning_rate": 1.5333333333333334e-05, "loss": 2.0831, "step": 920 }, { "epoch": 2.3025, "grad_norm": 4.34375, "learning_rate": 1.535e-05, "loss": 1.8523, "step": 921 }, { "epoch": 2.305, "grad_norm": 3.71875, "learning_rate": 1.5366666666666666e-05, "loss": 1.8564, "step": 922 }, { "epoch": 2.3075, "grad_norm": 6.65625, "learning_rate": 1.5383333333333336e-05, "loss": 2.1627, "step": 923 }, { "epoch": 2.31, "grad_norm": 6.875, "learning_rate": 1.54e-05, "loss": 2.0997, "step": 924 }, { "epoch": 2.3125, "grad_norm": 5.15625, "learning_rate": 1.5416666666666668e-05, "loss": 1.9972, "step": 925 }, { "epoch": 2.315, "grad_norm": 6.71875, "learning_rate": 1.5433333333333334e-05, "loss": 2.057, "step": 926 }, { "epoch": 2.3175, "grad_norm": 4.1875, "learning_rate": 1.545e-05, "loss": 1.8369, "step": 927 }, { "epoch": 2.32, "grad_norm": 5.4375, "learning_rate": 1.546666666666667e-05, "loss": 1.9189, "step": 928 }, { "epoch": 2.3225, "grad_norm": 4.96875, "learning_rate": 1.5483333333333335e-05, "loss": 2.0483, "step": 929 }, { "epoch": 2.325, "grad_norm": 4.96875, "learning_rate": 1.55e-05, "loss": 2.0435, "step": 930 }, { "epoch": 2.3275, "grad_norm": 5.6875, "learning_rate": 1.551666666666667e-05, "loss": 2.1837, "step": 931 }, { "epoch": 2.33, "grad_norm": 10.6875, "learning_rate": 1.5533333333333333e-05, "loss": 2.0218, "step": 932 }, { "epoch": 2.3325, "grad_norm": 4.71875, "learning_rate": 1.5550000000000002e-05, "loss": 1.9719, "step": 933 }, { "epoch": 2.335, "grad_norm": 5.90625, "learning_rate": 1.5566666666666668e-05, "loss": 2.2795, "step": 934 }, { "epoch": 2.3375, "grad_norm": 4.78125, "learning_rate": 1.5583333333333334e-05, "loss": 1.982, "step": 935 }, { "epoch": 2.34, "grad_norm": 3.984375, "learning_rate": 1.5600000000000003e-05, "loss": 1.9206, "step": 936 }, { "epoch": 2.3425000000000002, "grad_norm": 5.9375, "learning_rate": 1.561666666666667e-05, "loss": 1.9739, "step": 937 }, { "epoch": 2.3449999999999998, "grad_norm": 10.375, "learning_rate": 1.5633333333333335e-05, "loss": 2.0938, "step": 938 }, { "epoch": 2.3475, "grad_norm": 5.84375, "learning_rate": 1.565e-05, "loss": 1.9551, "step": 939 }, { "epoch": 2.35, "grad_norm": 4.75, "learning_rate": 1.5666666666666667e-05, "loss": 1.7354, "step": 940 }, { "epoch": 2.3525, "grad_norm": 16.625, "learning_rate": 1.5683333333333336e-05, "loss": 1.9141, "step": 941 }, { "epoch": 2.355, "grad_norm": 4.65625, "learning_rate": 1.5700000000000002e-05, "loss": 1.7929, "step": 942 }, { "epoch": 2.3575, "grad_norm": 4.34375, "learning_rate": 1.5716666666666668e-05, "loss": 1.7774, "step": 943 }, { "epoch": 2.36, "grad_norm": 14.875, "learning_rate": 1.5733333333333334e-05, "loss": 2.0986, "step": 944 }, { "epoch": 2.3625, "grad_norm": 6.78125, "learning_rate": 1.575e-05, "loss": 2.1457, "step": 945 }, { "epoch": 2.365, "grad_norm": 7.21875, "learning_rate": 1.576666666666667e-05, "loss": 2.0769, "step": 946 }, { "epoch": 2.3675, "grad_norm": 11.0, "learning_rate": 1.5783333333333335e-05, "loss": 2.1403, "step": 947 }, { "epoch": 2.37, "grad_norm": 4.65625, "learning_rate": 1.58e-05, "loss": 2.1666, "step": 948 }, { "epoch": 2.3725, "grad_norm": 7.96875, "learning_rate": 1.581666666666667e-05, "loss": 1.9914, "step": 949 }, { "epoch": 2.375, "grad_norm": 6.75, "learning_rate": 1.5833333333333333e-05, "loss": 2.1367, "step": 950 }, { "epoch": 2.3775, "grad_norm": 7.75, "learning_rate": 1.5850000000000002e-05, "loss": 2.0564, "step": 951 }, { "epoch": 2.38, "grad_norm": 5.1875, "learning_rate": 1.586666666666667e-05, "loss": 2.1247, "step": 952 }, { "epoch": 2.3825, "grad_norm": 6.46875, "learning_rate": 1.5883333333333334e-05, "loss": 2.0779, "step": 953 }, { "epoch": 2.385, "grad_norm": 5.625, "learning_rate": 1.5900000000000004e-05, "loss": 1.9608, "step": 954 }, { "epoch": 2.3875, "grad_norm": 6.0625, "learning_rate": 1.5916666666666666e-05, "loss": 2.0096, "step": 955 }, { "epoch": 2.39, "grad_norm": 5.375, "learning_rate": 1.5933333333333336e-05, "loss": 1.8496, "step": 956 }, { "epoch": 2.3925, "grad_norm": 4.25, "learning_rate": 1.595e-05, "loss": 1.905, "step": 957 }, { "epoch": 2.395, "grad_norm": 6.0625, "learning_rate": 1.5966666666666667e-05, "loss": 2.0353, "step": 958 }, { "epoch": 2.3975, "grad_norm": 5.125, "learning_rate": 1.5983333333333337e-05, "loss": 2.1178, "step": 959 }, { "epoch": 2.4, "grad_norm": 5.03125, "learning_rate": 1.6000000000000003e-05, "loss": 2.121, "step": 960 }, { "epoch": 2.4025, "grad_norm": 5.65625, "learning_rate": 1.601666666666667e-05, "loss": 1.9449, "step": 961 }, { "epoch": 2.4050000000000002, "grad_norm": 4.71875, "learning_rate": 1.6033333333333335e-05, "loss": 2.0081, "step": 962 }, { "epoch": 2.4074999999999998, "grad_norm": 5.15625, "learning_rate": 1.605e-05, "loss": 1.9294, "step": 963 }, { "epoch": 2.41, "grad_norm": 6.3125, "learning_rate": 1.606666666666667e-05, "loss": 1.9264, "step": 964 }, { "epoch": 2.4125, "grad_norm": 5.0, "learning_rate": 1.6083333333333336e-05, "loss": 2.1251, "step": 965 }, { "epoch": 2.415, "grad_norm": 6.59375, "learning_rate": 1.6100000000000002e-05, "loss": 2.0564, "step": 966 }, { "epoch": 2.4175, "grad_norm": 5.625, "learning_rate": 1.6116666666666668e-05, "loss": 2.0617, "step": 967 }, { "epoch": 2.42, "grad_norm": 5.1875, "learning_rate": 1.6133333333333334e-05, "loss": 1.8817, "step": 968 }, { "epoch": 2.4225, "grad_norm": 3.625, "learning_rate": 1.6150000000000003e-05, "loss": 1.9963, "step": 969 }, { "epoch": 2.425, "grad_norm": 4.84375, "learning_rate": 1.616666666666667e-05, "loss": 2.0236, "step": 970 }, { "epoch": 2.4275, "grad_norm": 6.28125, "learning_rate": 1.6183333333333335e-05, "loss": 2.2166, "step": 971 }, { "epoch": 2.43, "grad_norm": 5.78125, "learning_rate": 1.62e-05, "loss": 2.1028, "step": 972 }, { "epoch": 2.4325, "grad_norm": 4.46875, "learning_rate": 1.6216666666666667e-05, "loss": 1.9279, "step": 973 }, { "epoch": 2.435, "grad_norm": 7.875, "learning_rate": 1.6233333333333333e-05, "loss": 2.0919, "step": 974 }, { "epoch": 2.4375, "grad_norm": 5.6875, "learning_rate": 1.6250000000000002e-05, "loss": 2.1137, "step": 975 }, { "epoch": 2.44, "grad_norm": 7.5625, "learning_rate": 1.6266666666666668e-05, "loss": 1.9492, "step": 976 }, { "epoch": 2.4425, "grad_norm": 4.65625, "learning_rate": 1.6283333333333334e-05, "loss": 1.9736, "step": 977 }, { "epoch": 2.445, "grad_norm": 16.5, "learning_rate": 1.63e-05, "loss": 2.1259, "step": 978 }, { "epoch": 2.4475, "grad_norm": 6.28125, "learning_rate": 1.6316666666666666e-05, "loss": 2.127, "step": 979 }, { "epoch": 2.45, "grad_norm": 8.1875, "learning_rate": 1.6333333333333335e-05, "loss": 2.2834, "step": 980 }, { "epoch": 2.4525, "grad_norm": 5.4375, "learning_rate": 1.635e-05, "loss": 2.1537, "step": 981 }, { "epoch": 2.455, "grad_norm": 4.28125, "learning_rate": 1.6366666666666667e-05, "loss": 1.7982, "step": 982 }, { "epoch": 2.4575, "grad_norm": 5.34375, "learning_rate": 1.6383333333333336e-05, "loss": 2.0059, "step": 983 }, { "epoch": 2.46, "grad_norm": 7.09375, "learning_rate": 1.64e-05, "loss": 2.2235, "step": 984 }, { "epoch": 2.4625, "grad_norm": 5.71875, "learning_rate": 1.6416666666666668e-05, "loss": 2.0258, "step": 985 }, { "epoch": 2.465, "grad_norm": 6.34375, "learning_rate": 1.6433333333333334e-05, "loss": 1.9054, "step": 986 }, { "epoch": 2.4675000000000002, "grad_norm": 4.15625, "learning_rate": 1.645e-05, "loss": 2.1304, "step": 987 }, { "epoch": 2.4699999999999998, "grad_norm": 5.84375, "learning_rate": 1.646666666666667e-05, "loss": 2.2226, "step": 988 }, { "epoch": 2.4725, "grad_norm": 6.4375, "learning_rate": 1.6483333333333335e-05, "loss": 1.8324, "step": 989 }, { "epoch": 2.475, "grad_norm": 5.46875, "learning_rate": 1.65e-05, "loss": 2.1832, "step": 990 }, { "epoch": 2.4775, "grad_norm": 5.46875, "learning_rate": 1.6516666666666667e-05, "loss": 1.9992, "step": 991 }, { "epoch": 2.48, "grad_norm": 5.96875, "learning_rate": 1.6533333333333333e-05, "loss": 1.8744, "step": 992 }, { "epoch": 2.4825, "grad_norm": 8.6875, "learning_rate": 1.6550000000000002e-05, "loss": 1.7618, "step": 993 }, { "epoch": 2.485, "grad_norm": 11.25, "learning_rate": 1.656666666666667e-05, "loss": 1.7849, "step": 994 }, { "epoch": 2.4875, "grad_norm": 5.1875, "learning_rate": 1.6583333333333334e-05, "loss": 1.9482, "step": 995 }, { "epoch": 2.49, "grad_norm": 4.15625, "learning_rate": 1.66e-05, "loss": 1.8971, "step": 996 }, { "epoch": 2.4925, "grad_norm": 5.4375, "learning_rate": 1.6616666666666666e-05, "loss": 2.0077, "step": 997 }, { "epoch": 2.495, "grad_norm": 4.625, "learning_rate": 1.6633333333333336e-05, "loss": 1.8957, "step": 998 }, { "epoch": 2.4975, "grad_norm": 4.90625, "learning_rate": 1.665e-05, "loss": 1.9924, "step": 999 }, { "epoch": 2.5, "grad_norm": 6.28125, "learning_rate": 1.6666666666666667e-05, "loss": 1.9546, "step": 1000 }, { "epoch": 2.5025, "grad_norm": 6.46875, "learning_rate": 1.6683333333333333e-05, "loss": 1.9256, "step": 1001 }, { "epoch": 2.505, "grad_norm": 5.96875, "learning_rate": 1.67e-05, "loss": 1.9413, "step": 1002 }, { "epoch": 2.5075, "grad_norm": 4.875, "learning_rate": 1.671666666666667e-05, "loss": 1.9513, "step": 1003 }, { "epoch": 2.51, "grad_norm": 5.03125, "learning_rate": 1.6733333333333335e-05, "loss": 1.8286, "step": 1004 }, { "epoch": 2.5125, "grad_norm": 4.3125, "learning_rate": 1.675e-05, "loss": 1.7921, "step": 1005 }, { "epoch": 2.515, "grad_norm": 7.59375, "learning_rate": 1.676666666666667e-05, "loss": 2.0094, "step": 1006 }, { "epoch": 2.5175, "grad_norm": 6.375, "learning_rate": 1.6783333333333332e-05, "loss": 1.8933, "step": 1007 }, { "epoch": 2.52, "grad_norm": 5.0625, "learning_rate": 1.6800000000000002e-05, "loss": 1.917, "step": 1008 }, { "epoch": 2.5225, "grad_norm": 6.09375, "learning_rate": 1.6816666666666668e-05, "loss": 2.0524, "step": 1009 }, { "epoch": 2.525, "grad_norm": 3.890625, "learning_rate": 1.6833333333333334e-05, "loss": 1.8706, "step": 1010 }, { "epoch": 2.5275, "grad_norm": 4.28125, "learning_rate": 1.6850000000000003e-05, "loss": 1.8255, "step": 1011 }, { "epoch": 2.5300000000000002, "grad_norm": 5.03125, "learning_rate": 1.686666666666667e-05, "loss": 1.9479, "step": 1012 }, { "epoch": 2.5324999999999998, "grad_norm": 5.375, "learning_rate": 1.6883333333333335e-05, "loss": 1.9716, "step": 1013 }, { "epoch": 2.535, "grad_norm": 6.25, "learning_rate": 1.69e-05, "loss": 2.0299, "step": 1014 }, { "epoch": 2.5375, "grad_norm": 5.28125, "learning_rate": 1.6916666666666667e-05, "loss": 2.0556, "step": 1015 }, { "epoch": 2.54, "grad_norm": 7.78125, "learning_rate": 1.6933333333333336e-05, "loss": 2.0398, "step": 1016 }, { "epoch": 2.5425, "grad_norm": 6.40625, "learning_rate": 1.6950000000000002e-05, "loss": 1.9619, "step": 1017 }, { "epoch": 2.545, "grad_norm": 4.8125, "learning_rate": 1.6966666666666668e-05, "loss": 2.0708, "step": 1018 }, { "epoch": 2.5475, "grad_norm": 3.8125, "learning_rate": 1.6983333333333334e-05, "loss": 1.8095, "step": 1019 }, { "epoch": 2.55, "grad_norm": 5.59375, "learning_rate": 1.7e-05, "loss": 1.9441, "step": 1020 }, { "epoch": 2.5525, "grad_norm": 6.90625, "learning_rate": 1.701666666666667e-05, "loss": 2.1118, "step": 1021 }, { "epoch": 2.555, "grad_norm": 5.5625, "learning_rate": 1.7033333333333335e-05, "loss": 1.8776, "step": 1022 }, { "epoch": 2.5575, "grad_norm": 4.21875, "learning_rate": 1.705e-05, "loss": 1.9019, "step": 1023 }, { "epoch": 2.56, "grad_norm": 5.6875, "learning_rate": 1.706666666666667e-05, "loss": 2.0005, "step": 1024 }, { "epoch": 2.5625, "grad_norm": 7.21875, "learning_rate": 1.7083333333333333e-05, "loss": 2.0719, "step": 1025 }, { "epoch": 2.565, "grad_norm": 3.78125, "learning_rate": 1.7100000000000002e-05, "loss": 2.1498, "step": 1026 }, { "epoch": 2.5675, "grad_norm": 4.65625, "learning_rate": 1.7116666666666668e-05, "loss": 2.0234, "step": 1027 }, { "epoch": 2.57, "grad_norm": 7.3125, "learning_rate": 1.7133333333333334e-05, "loss": 2.0095, "step": 1028 }, { "epoch": 2.5725, "grad_norm": 4.96875, "learning_rate": 1.7150000000000004e-05, "loss": 1.7727, "step": 1029 }, { "epoch": 2.575, "grad_norm": 5.0625, "learning_rate": 1.7166666666666666e-05, "loss": 1.9208, "step": 1030 }, { "epoch": 2.5775, "grad_norm": 9.0, "learning_rate": 1.7183333333333335e-05, "loss": 2.1152, "step": 1031 }, { "epoch": 2.58, "grad_norm": 5.6875, "learning_rate": 1.72e-05, "loss": 1.9053, "step": 1032 }, { "epoch": 2.5825, "grad_norm": 6.03125, "learning_rate": 1.7216666666666667e-05, "loss": 1.8991, "step": 1033 }, { "epoch": 2.585, "grad_norm": 7.3125, "learning_rate": 1.7233333333333337e-05, "loss": 1.876, "step": 1034 }, { "epoch": 2.5875, "grad_norm": 5.5, "learning_rate": 1.7250000000000003e-05, "loss": 2.1581, "step": 1035 }, { "epoch": 2.59, "grad_norm": 5.21875, "learning_rate": 1.726666666666667e-05, "loss": 1.952, "step": 1036 }, { "epoch": 2.5925000000000002, "grad_norm": 17.375, "learning_rate": 1.7283333333333334e-05, "loss": 1.9064, "step": 1037 }, { "epoch": 2.5949999999999998, "grad_norm": 8.375, "learning_rate": 1.73e-05, "loss": 1.8947, "step": 1038 }, { "epoch": 2.5975, "grad_norm": 5.125, "learning_rate": 1.731666666666667e-05, "loss": 1.7269, "step": 1039 }, { "epoch": 2.6, "grad_norm": 7.78125, "learning_rate": 1.7333333333333336e-05, "loss": 1.9108, "step": 1040 }, { "epoch": 2.6025, "grad_norm": 4.90625, "learning_rate": 1.735e-05, "loss": 2.0378, "step": 1041 }, { "epoch": 2.605, "grad_norm": 12.0, "learning_rate": 1.7366666666666668e-05, "loss": 1.8386, "step": 1042 }, { "epoch": 2.6075, "grad_norm": 5.46875, "learning_rate": 1.7383333333333333e-05, "loss": 1.9788, "step": 1043 }, { "epoch": 2.61, "grad_norm": 7.25, "learning_rate": 1.7400000000000003e-05, "loss": 1.9596, "step": 1044 }, { "epoch": 2.6125, "grad_norm": 3.953125, "learning_rate": 1.741666666666667e-05, "loss": 1.8623, "step": 1045 }, { "epoch": 2.615, "grad_norm": 5.21875, "learning_rate": 1.7433333333333335e-05, "loss": 1.787, "step": 1046 }, { "epoch": 2.6175, "grad_norm": 6.9375, "learning_rate": 1.7450000000000004e-05, "loss": 1.967, "step": 1047 }, { "epoch": 2.62, "grad_norm": 6.03125, "learning_rate": 1.7466666666666667e-05, "loss": 1.9528, "step": 1048 }, { "epoch": 2.6225, "grad_norm": 5.46875, "learning_rate": 1.7483333333333336e-05, "loss": 1.8991, "step": 1049 }, { "epoch": 2.625, "grad_norm": 4.65625, "learning_rate": 1.7500000000000002e-05, "loss": 1.999, "step": 1050 }, { "epoch": 2.6275, "grad_norm": 7.53125, "learning_rate": 1.7516666666666668e-05, "loss": 1.8479, "step": 1051 }, { "epoch": 2.63, "grad_norm": 9.6875, "learning_rate": 1.7533333333333337e-05, "loss": 1.9734, "step": 1052 }, { "epoch": 2.6325, "grad_norm": 10.5, "learning_rate": 1.755e-05, "loss": 1.9231, "step": 1053 }, { "epoch": 2.635, "grad_norm": 7.8125, "learning_rate": 1.756666666666667e-05, "loss": 2.0457, "step": 1054 }, { "epoch": 2.6375, "grad_norm": 5.875, "learning_rate": 1.7583333333333335e-05, "loss": 1.9955, "step": 1055 }, { "epoch": 2.64, "grad_norm": 10.9375, "learning_rate": 1.76e-05, "loss": 1.9297, "step": 1056 }, { "epoch": 2.6425, "grad_norm": 4.5625, "learning_rate": 1.761666666666667e-05, "loss": 2.0702, "step": 1057 }, { "epoch": 2.645, "grad_norm": 7.53125, "learning_rate": 1.7633333333333336e-05, "loss": 2.0315, "step": 1058 }, { "epoch": 2.6475, "grad_norm": 6.46875, "learning_rate": 1.7650000000000002e-05, "loss": 2.0171, "step": 1059 }, { "epoch": 2.65, "grad_norm": 7.03125, "learning_rate": 1.7666666666666668e-05, "loss": 2.0794, "step": 1060 }, { "epoch": 2.6525, "grad_norm": 9.8125, "learning_rate": 1.7683333333333334e-05, "loss": 1.9506, "step": 1061 }, { "epoch": 2.6550000000000002, "grad_norm": 6.28125, "learning_rate": 1.77e-05, "loss": 1.6683, "step": 1062 }, { "epoch": 2.6574999999999998, "grad_norm": 6.0625, "learning_rate": 1.771666666666667e-05, "loss": 2.0721, "step": 1063 }, { "epoch": 2.66, "grad_norm": 8.5, "learning_rate": 1.7733333333333335e-05, "loss": 2.0603, "step": 1064 }, { "epoch": 2.6625, "grad_norm": 17.5, "learning_rate": 1.775e-05, "loss": 1.9729, "step": 1065 }, { "epoch": 2.665, "grad_norm": 7.09375, "learning_rate": 1.7766666666666667e-05, "loss": 1.897, "step": 1066 }, { "epoch": 2.6675, "grad_norm": 6.375, "learning_rate": 1.7783333333333333e-05, "loss": 2.068, "step": 1067 }, { "epoch": 2.67, "grad_norm": 3.9375, "learning_rate": 1.7800000000000002e-05, "loss": 1.7763, "step": 1068 }, { "epoch": 2.6725, "grad_norm": 4.1875, "learning_rate": 1.781666666666667e-05, "loss": 1.805, "step": 1069 }, { "epoch": 2.675, "grad_norm": 22.25, "learning_rate": 1.7833333333333334e-05, "loss": 2.0473, "step": 1070 }, { "epoch": 2.6775, "grad_norm": 5.3125, "learning_rate": 1.785e-05, "loss": 1.8418, "step": 1071 }, { "epoch": 2.68, "grad_norm": 7.03125, "learning_rate": 1.7866666666666666e-05, "loss": 1.9227, "step": 1072 }, { "epoch": 2.6825, "grad_norm": 6.1875, "learning_rate": 1.7883333333333335e-05, "loss": 1.8605, "step": 1073 }, { "epoch": 2.685, "grad_norm": 8.1875, "learning_rate": 1.79e-05, "loss": 2.0944, "step": 1074 }, { "epoch": 2.6875, "grad_norm": 6.53125, "learning_rate": 1.7916666666666667e-05, "loss": 1.9725, "step": 1075 }, { "epoch": 2.69, "grad_norm": 7.8125, "learning_rate": 1.7933333333333333e-05, "loss": 1.8923, "step": 1076 }, { "epoch": 2.6925, "grad_norm": 4.875, "learning_rate": 1.795e-05, "loss": 1.8971, "step": 1077 }, { "epoch": 2.695, "grad_norm": 7.5, "learning_rate": 1.796666666666667e-05, "loss": 2.1974, "step": 1078 }, { "epoch": 2.6975, "grad_norm": 7.90625, "learning_rate": 1.7983333333333335e-05, "loss": 1.9787, "step": 1079 }, { "epoch": 2.7, "grad_norm": 5.40625, "learning_rate": 1.8e-05, "loss": 2.048, "step": 1080 }, { "epoch": 2.7025, "grad_norm": 6.90625, "learning_rate": 1.801666666666667e-05, "loss": 1.8088, "step": 1081 }, { "epoch": 2.705, "grad_norm": 5.59375, "learning_rate": 1.8033333333333332e-05, "loss": 1.9584, "step": 1082 }, { "epoch": 2.7075, "grad_norm": 9.0, "learning_rate": 1.805e-05, "loss": 1.9621, "step": 1083 }, { "epoch": 2.71, "grad_norm": 5.6875, "learning_rate": 1.8066666666666668e-05, "loss": 2.0354, "step": 1084 }, { "epoch": 2.7125, "grad_norm": 11.3125, "learning_rate": 1.8083333333333334e-05, "loss": 1.9684, "step": 1085 }, { "epoch": 2.715, "grad_norm": 8.5625, "learning_rate": 1.8100000000000003e-05, "loss": 1.8195, "step": 1086 }, { "epoch": 2.7175000000000002, "grad_norm": 6.9375, "learning_rate": 1.811666666666667e-05, "loss": 2.0978, "step": 1087 }, { "epoch": 2.7199999999999998, "grad_norm": 8.0625, "learning_rate": 1.8133333333333335e-05, "loss": 1.8214, "step": 1088 }, { "epoch": 2.7225, "grad_norm": 5.21875, "learning_rate": 1.815e-05, "loss": 1.9825, "step": 1089 }, { "epoch": 2.725, "grad_norm": 5.96875, "learning_rate": 1.8166666666666667e-05, "loss": 1.918, "step": 1090 }, { "epoch": 2.7275, "grad_norm": 8.6875, "learning_rate": 1.8183333333333336e-05, "loss": 2.0887, "step": 1091 }, { "epoch": 2.73, "grad_norm": 5.09375, "learning_rate": 1.8200000000000002e-05, "loss": 2.0704, "step": 1092 }, { "epoch": 2.7325, "grad_norm": 8.1875, "learning_rate": 1.8216666666666668e-05, "loss": 2.2046, "step": 1093 }, { "epoch": 2.735, "grad_norm": 19.75, "learning_rate": 1.8233333333333334e-05, "loss": 1.8731, "step": 1094 }, { "epoch": 2.7375, "grad_norm": 5.53125, "learning_rate": 1.825e-05, "loss": 1.9398, "step": 1095 }, { "epoch": 2.74, "grad_norm": 13.5, "learning_rate": 1.826666666666667e-05, "loss": 1.6855, "step": 1096 }, { "epoch": 2.7425, "grad_norm": 5.40625, "learning_rate": 1.8283333333333335e-05, "loss": 1.694, "step": 1097 }, { "epoch": 2.745, "grad_norm": 7.34375, "learning_rate": 1.83e-05, "loss": 2.1006, "step": 1098 }, { "epoch": 2.7475, "grad_norm": 3.8125, "learning_rate": 1.831666666666667e-05, "loss": 1.8608, "step": 1099 }, { "epoch": 2.75, "grad_norm": 5.03125, "learning_rate": 1.8333333333333333e-05, "loss": 2.0132, "step": 1100 }, { "epoch": 2.7525, "grad_norm": 8.6875, "learning_rate": 1.8350000000000002e-05, "loss": 2.0259, "step": 1101 }, { "epoch": 2.755, "grad_norm": 5.75, "learning_rate": 1.8366666666666668e-05, "loss": 1.7211, "step": 1102 }, { "epoch": 2.7575, "grad_norm": 8.3125, "learning_rate": 1.8383333333333334e-05, "loss": 1.8318, "step": 1103 }, { "epoch": 2.76, "grad_norm": 5.1875, "learning_rate": 1.8400000000000003e-05, "loss": 1.7431, "step": 1104 }, { "epoch": 2.7625, "grad_norm": 6.03125, "learning_rate": 1.8416666666666666e-05, "loss": 1.8809, "step": 1105 }, { "epoch": 2.765, "grad_norm": 6.34375, "learning_rate": 1.8433333333333335e-05, "loss": 1.885, "step": 1106 }, { "epoch": 2.7675, "grad_norm": 5.21875, "learning_rate": 1.845e-05, "loss": 1.839, "step": 1107 }, { "epoch": 2.77, "grad_norm": 9.75, "learning_rate": 1.8466666666666667e-05, "loss": 2.0009, "step": 1108 }, { "epoch": 2.7725, "grad_norm": 6.03125, "learning_rate": 1.8483333333333337e-05, "loss": 1.9806, "step": 1109 }, { "epoch": 2.775, "grad_norm": 4.75, "learning_rate": 1.8500000000000002e-05, "loss": 1.9732, "step": 1110 }, { "epoch": 2.7775, "grad_norm": 6.78125, "learning_rate": 1.851666666666667e-05, "loss": 2.0954, "step": 1111 }, { "epoch": 2.7800000000000002, "grad_norm": 5.28125, "learning_rate": 1.8533333333333334e-05, "loss": 1.9871, "step": 1112 }, { "epoch": 2.7824999999999998, "grad_norm": 6.84375, "learning_rate": 1.855e-05, "loss": 1.9994, "step": 1113 }, { "epoch": 2.785, "grad_norm": 17.5, "learning_rate": 1.856666666666667e-05, "loss": 1.8177, "step": 1114 }, { "epoch": 2.7875, "grad_norm": 5.25, "learning_rate": 1.8583333333333336e-05, "loss": 1.8558, "step": 1115 }, { "epoch": 2.79, "grad_norm": 5.65625, "learning_rate": 1.86e-05, "loss": 2.0907, "step": 1116 }, { "epoch": 2.7925, "grad_norm": 7.375, "learning_rate": 1.8616666666666667e-05, "loss": 1.9366, "step": 1117 }, { "epoch": 2.795, "grad_norm": 4.21875, "learning_rate": 1.8633333333333333e-05, "loss": 1.8627, "step": 1118 }, { "epoch": 2.7975, "grad_norm": 7.5, "learning_rate": 1.8650000000000003e-05, "loss": 2.0305, "step": 1119 }, { "epoch": 2.8, "grad_norm": 6.8125, "learning_rate": 1.866666666666667e-05, "loss": 2.1552, "step": 1120 }, { "epoch": 2.8025, "grad_norm": 5.09375, "learning_rate": 1.8683333333333335e-05, "loss": 1.9721, "step": 1121 }, { "epoch": 2.805, "grad_norm": 6.15625, "learning_rate": 1.8700000000000004e-05, "loss": 2.161, "step": 1122 }, { "epoch": 2.8075, "grad_norm": 8.25, "learning_rate": 1.8716666666666666e-05, "loss": 1.8084, "step": 1123 }, { "epoch": 2.81, "grad_norm": 4.25, "learning_rate": 1.8733333333333336e-05, "loss": 1.907, "step": 1124 }, { "epoch": 2.8125, "grad_norm": 5.53125, "learning_rate": 1.8750000000000002e-05, "loss": 1.7191, "step": 1125 }, { "epoch": 2.815, "grad_norm": 9.0, "learning_rate": 1.8766666666666668e-05, "loss": 1.8629, "step": 1126 }, { "epoch": 2.8175, "grad_norm": 5.4375, "learning_rate": 1.8783333333333337e-05, "loss": 1.7733, "step": 1127 }, { "epoch": 2.82, "grad_norm": 6.1875, "learning_rate": 1.88e-05, "loss": 2.0752, "step": 1128 }, { "epoch": 2.8225, "grad_norm": 6.4375, "learning_rate": 1.881666666666667e-05, "loss": 1.9315, "step": 1129 }, { "epoch": 2.825, "grad_norm": 11.1875, "learning_rate": 1.8833333333333335e-05, "loss": 2.0606, "step": 1130 }, { "epoch": 2.8275, "grad_norm": 8.8125, "learning_rate": 1.885e-05, "loss": 1.79, "step": 1131 }, { "epoch": 2.83, "grad_norm": 6.65625, "learning_rate": 1.886666666666667e-05, "loss": 1.9338, "step": 1132 }, { "epoch": 2.8325, "grad_norm": 6.5625, "learning_rate": 1.8883333333333336e-05, "loss": 1.8693, "step": 1133 }, { "epoch": 2.835, "grad_norm": 9.3125, "learning_rate": 1.8900000000000002e-05, "loss": 1.8299, "step": 1134 }, { "epoch": 2.8375, "grad_norm": 4.96875, "learning_rate": 1.8916666666666668e-05, "loss": 1.8321, "step": 1135 }, { "epoch": 2.84, "grad_norm": 5.6875, "learning_rate": 1.8933333333333334e-05, "loss": 1.6887, "step": 1136 }, { "epoch": 2.8425000000000002, "grad_norm": 9.4375, "learning_rate": 1.8950000000000003e-05, "loss": 1.7844, "step": 1137 }, { "epoch": 2.8449999999999998, "grad_norm": 6.25, "learning_rate": 1.896666666666667e-05, "loss": 1.918, "step": 1138 }, { "epoch": 2.8475, "grad_norm": 9.875, "learning_rate": 1.8983333333333335e-05, "loss": 2.051, "step": 1139 }, { "epoch": 2.85, "grad_norm": 7.0625, "learning_rate": 1.9e-05, "loss": 2.0235, "step": 1140 }, { "epoch": 2.8525, "grad_norm": 8.6875, "learning_rate": 1.9016666666666667e-05, "loss": 1.7907, "step": 1141 }, { "epoch": 2.855, "grad_norm": 8.4375, "learning_rate": 1.9033333333333336e-05, "loss": 1.9738, "step": 1142 }, { "epoch": 2.8575, "grad_norm": 6.59375, "learning_rate": 1.9050000000000002e-05, "loss": 1.9504, "step": 1143 }, { "epoch": 2.86, "grad_norm": 6.28125, "learning_rate": 1.9066666666666668e-05, "loss": 1.8223, "step": 1144 }, { "epoch": 2.8625, "grad_norm": 5.71875, "learning_rate": 1.9083333333333338e-05, "loss": 1.9082, "step": 1145 }, { "epoch": 2.865, "grad_norm": 7.625, "learning_rate": 1.91e-05, "loss": 2.1068, "step": 1146 }, { "epoch": 2.8675, "grad_norm": 5.65625, "learning_rate": 1.911666666666667e-05, "loss": 1.9845, "step": 1147 }, { "epoch": 2.87, "grad_norm": 11.1875, "learning_rate": 1.9133333333333335e-05, "loss": 2.012, "step": 1148 }, { "epoch": 2.8725, "grad_norm": 7.90625, "learning_rate": 1.915e-05, "loss": 2.0073, "step": 1149 }, { "epoch": 2.875, "grad_norm": 14.0625, "learning_rate": 1.916666666666667e-05, "loss": 1.8509, "step": 1150 }, { "epoch": 2.8775, "grad_norm": 6.4375, "learning_rate": 1.9183333333333333e-05, "loss": 2.137, "step": 1151 }, { "epoch": 2.88, "grad_norm": 7.78125, "learning_rate": 1.9200000000000003e-05, "loss": 1.8292, "step": 1152 }, { "epoch": 2.8825, "grad_norm": 5.0625, "learning_rate": 1.921666666666667e-05, "loss": 1.9826, "step": 1153 }, { "epoch": 2.885, "grad_norm": 5.125, "learning_rate": 1.9233333333333334e-05, "loss": 1.8121, "step": 1154 }, { "epoch": 2.8875, "grad_norm": 5.21875, "learning_rate": 1.925e-05, "loss": 2.0655, "step": 1155 }, { "epoch": 2.89, "grad_norm": 8.0625, "learning_rate": 1.926666666666667e-05, "loss": 2.0406, "step": 1156 }, { "epoch": 2.8925, "grad_norm": 5.40625, "learning_rate": 1.9283333333333332e-05, "loss": 1.9152, "step": 1157 }, { "epoch": 2.895, "grad_norm": 6.0625, "learning_rate": 1.93e-05, "loss": 2.1359, "step": 1158 }, { "epoch": 2.8975, "grad_norm": 4.0625, "learning_rate": 1.9316666666666668e-05, "loss": 1.7634, "step": 1159 }, { "epoch": 2.9, "grad_norm": 6.375, "learning_rate": 1.9333333333333333e-05, "loss": 1.998, "step": 1160 }, { "epoch": 2.9025, "grad_norm": 6.90625, "learning_rate": 1.9350000000000003e-05, "loss": 1.9948, "step": 1161 }, { "epoch": 2.9050000000000002, "grad_norm": 6.8125, "learning_rate": 1.936666666666667e-05, "loss": 2.0025, "step": 1162 }, { "epoch": 2.9074999999999998, "grad_norm": 7.8125, "learning_rate": 1.9383333333333335e-05, "loss": 1.9168, "step": 1163 }, { "epoch": 2.91, "grad_norm": 9.875, "learning_rate": 1.94e-05, "loss": 1.97, "step": 1164 }, { "epoch": 2.9125, "grad_norm": 5.9375, "learning_rate": 1.9416666666666667e-05, "loss": 1.882, "step": 1165 }, { "epoch": 2.915, "grad_norm": 8.1875, "learning_rate": 1.9433333333333336e-05, "loss": 1.9268, "step": 1166 }, { "epoch": 2.9175, "grad_norm": 4.375, "learning_rate": 1.9450000000000002e-05, "loss": 1.7827, "step": 1167 }, { "epoch": 2.92, "grad_norm": 11.6875, "learning_rate": 1.9466666666666668e-05, "loss": 1.982, "step": 1168 }, { "epoch": 2.9225, "grad_norm": 6.0625, "learning_rate": 1.9483333333333334e-05, "loss": 1.814, "step": 1169 }, { "epoch": 2.925, "grad_norm": 5.65625, "learning_rate": 1.95e-05, "loss": 2.0255, "step": 1170 }, { "epoch": 2.9275, "grad_norm": 4.5, "learning_rate": 1.951666666666667e-05, "loss": 1.8823, "step": 1171 }, { "epoch": 2.93, "grad_norm": 5.0625, "learning_rate": 1.9533333333333335e-05, "loss": 1.8165, "step": 1172 }, { "epoch": 2.9325, "grad_norm": 4.21875, "learning_rate": 1.955e-05, "loss": 1.7301, "step": 1173 }, { "epoch": 2.935, "grad_norm": 5.59375, "learning_rate": 1.956666666666667e-05, "loss": 1.7541, "step": 1174 }, { "epoch": 2.9375, "grad_norm": 6.6875, "learning_rate": 1.9583333333333333e-05, "loss": 2.0039, "step": 1175 }, { "epoch": 2.94, "grad_norm": 7.90625, "learning_rate": 1.9600000000000002e-05, "loss": 1.9307, "step": 1176 }, { "epoch": 2.9425, "grad_norm": 5.6875, "learning_rate": 1.9616666666666668e-05, "loss": 2.0508, "step": 1177 }, { "epoch": 2.945, "grad_norm": 5.0, "learning_rate": 1.9633333333333334e-05, "loss": 1.9476, "step": 1178 }, { "epoch": 2.9475, "grad_norm": 5.25, "learning_rate": 1.9650000000000003e-05, "loss": 2.0636, "step": 1179 }, { "epoch": 2.95, "grad_norm": 4.03125, "learning_rate": 1.9666666666666666e-05, "loss": 1.7682, "step": 1180 }, { "epoch": 2.9525, "grad_norm": 7.78125, "learning_rate": 1.9683333333333335e-05, "loss": 2.068, "step": 1181 }, { "epoch": 2.955, "grad_norm": 4.3125, "learning_rate": 1.97e-05, "loss": 1.6164, "step": 1182 }, { "epoch": 2.9575, "grad_norm": 4.5, "learning_rate": 1.9716666666666667e-05, "loss": 2.0029, "step": 1183 }, { "epoch": 2.96, "grad_norm": 4.09375, "learning_rate": 1.9733333333333336e-05, "loss": 1.9279, "step": 1184 }, { "epoch": 2.9625, "grad_norm": 5.875, "learning_rate": 1.9750000000000002e-05, "loss": 1.9947, "step": 1185 }, { "epoch": 2.965, "grad_norm": 5.5, "learning_rate": 1.9766666666666668e-05, "loss": 1.9609, "step": 1186 }, { "epoch": 2.9675000000000002, "grad_norm": 5.6875, "learning_rate": 1.9783333333333334e-05, "loss": 1.7226, "step": 1187 }, { "epoch": 2.9699999999999998, "grad_norm": 5.59375, "learning_rate": 1.98e-05, "loss": 1.8813, "step": 1188 }, { "epoch": 2.9725, "grad_norm": 9.75, "learning_rate": 1.981666666666667e-05, "loss": 1.7862, "step": 1189 }, { "epoch": 2.975, "grad_norm": 6.1875, "learning_rate": 1.9833333333333335e-05, "loss": 1.8693, "step": 1190 }, { "epoch": 2.9775, "grad_norm": 9.3125, "learning_rate": 1.985e-05, "loss": 1.9065, "step": 1191 }, { "epoch": 2.98, "grad_norm": 6.59375, "learning_rate": 1.9866666666666667e-05, "loss": 1.9369, "step": 1192 }, { "epoch": 2.9825, "grad_norm": 5.71875, "learning_rate": 1.9883333333333333e-05, "loss": 1.8397, "step": 1193 }, { "epoch": 2.985, "grad_norm": 5.875, "learning_rate": 1.9900000000000003e-05, "loss": 1.6927, "step": 1194 }, { "epoch": 2.9875, "grad_norm": 6.15625, "learning_rate": 1.991666666666667e-05, "loss": 1.8641, "step": 1195 }, { "epoch": 2.99, "grad_norm": 5.0, "learning_rate": 1.9933333333333334e-05, "loss": 1.6801, "step": 1196 }, { "epoch": 2.9925, "grad_norm": 10.0625, "learning_rate": 1.9950000000000004e-05, "loss": 1.8714, "step": 1197 }, { "epoch": 2.995, "grad_norm": 5.625, "learning_rate": 1.9966666666666666e-05, "loss": 1.8282, "step": 1198 }, { "epoch": 2.9975, "grad_norm": 7.03125, "learning_rate": 1.9983333333333336e-05, "loss": 1.9291, "step": 1199 }, { "epoch": 3.0, "grad_norm": 5.90625, "learning_rate": 2e-05, "loss": 1.929, "step": 1200 }, { "epoch": 3.0025, "grad_norm": 10.6875, "learning_rate": 1.99999999672202e-05, "loss": 1.893, "step": 1201 }, { "epoch": 3.005, "grad_norm": 6.375, "learning_rate": 1.9999999868880802e-05, "loss": 1.8558, "step": 1202 }, { "epoch": 3.0075, "grad_norm": 5.34375, "learning_rate": 1.9999999704981804e-05, "loss": 1.7404, "step": 1203 }, { "epoch": 3.01, "grad_norm": 5.0625, "learning_rate": 1.9999999475523208e-05, "loss": 1.7598, "step": 1204 }, { "epoch": 3.0125, "grad_norm": 4.5625, "learning_rate": 1.9999999180505014e-05, "loss": 1.9448, "step": 1205 }, { "epoch": 3.015, "grad_norm": 5.875, "learning_rate": 1.999999881992723e-05, "loss": 1.9417, "step": 1206 }, { "epoch": 3.0175, "grad_norm": 5.6875, "learning_rate": 1.999999839378985e-05, "loss": 1.9082, "step": 1207 }, { "epoch": 3.02, "grad_norm": 6.40625, "learning_rate": 1.9999997902092877e-05, "loss": 1.9146, "step": 1208 }, { "epoch": 3.0225, "grad_norm": 5.78125, "learning_rate": 1.9999997344836325e-05, "loss": 1.768, "step": 1209 }, { "epoch": 3.025, "grad_norm": 6.625, "learning_rate": 1.999999672202019e-05, "loss": 1.8115, "step": 1210 }, { "epoch": 3.0275, "grad_norm": 6.375, "learning_rate": 1.9999996033644475e-05, "loss": 2.0157, "step": 1211 }, { "epoch": 3.03, "grad_norm": 3.78125, "learning_rate": 1.9999995279709183e-05, "loss": 1.6764, "step": 1212 }, { "epoch": 3.0325, "grad_norm": 5.0625, "learning_rate": 1.9999994460214324e-05, "loss": 1.8027, "step": 1213 }, { "epoch": 3.035, "grad_norm": 4.28125, "learning_rate": 1.9999993575159904e-05, "loss": 1.6113, "step": 1214 }, { "epoch": 3.0375, "grad_norm": 5.5, "learning_rate": 1.9999992624545923e-05, "loss": 1.9954, "step": 1215 }, { "epoch": 3.04, "grad_norm": 6.0625, "learning_rate": 1.9999991608372392e-05, "loss": 1.8524, "step": 1216 }, { "epoch": 3.0425, "grad_norm": 7.25, "learning_rate": 1.999999052663932e-05, "loss": 1.746, "step": 1217 }, { "epoch": 3.045, "grad_norm": 5.5, "learning_rate": 1.9999989379346705e-05, "loss": 1.8521, "step": 1218 }, { "epoch": 3.0475, "grad_norm": 10.0, "learning_rate": 1.999998816649456e-05, "loss": 1.8435, "step": 1219 }, { "epoch": 3.05, "grad_norm": 6.1875, "learning_rate": 1.9999986888082895e-05, "loss": 1.8641, "step": 1220 }, { "epoch": 3.0525, "grad_norm": 4.40625, "learning_rate": 1.9999985544111716e-05, "loss": 1.8216, "step": 1221 }, { "epoch": 3.055, "grad_norm": 4.84375, "learning_rate": 1.999998413458103e-05, "loss": 1.8234, "step": 1222 }, { "epoch": 3.0575, "grad_norm": 6.71875, "learning_rate": 1.999998265949085e-05, "loss": 1.9419, "step": 1223 }, { "epoch": 3.06, "grad_norm": 6.625, "learning_rate": 1.9999981118841182e-05, "loss": 1.8445, "step": 1224 }, { "epoch": 3.0625, "grad_norm": 4.65625, "learning_rate": 1.9999979512632042e-05, "loss": 1.6995, "step": 1225 }, { "epoch": 3.065, "grad_norm": 7.28125, "learning_rate": 1.9999977840863432e-05, "loss": 1.9492, "step": 1226 }, { "epoch": 3.0675, "grad_norm": 4.6875, "learning_rate": 1.999997610353537e-05, "loss": 1.5926, "step": 1227 }, { "epoch": 3.07, "grad_norm": 6.125, "learning_rate": 1.9999974300647867e-05, "loss": 1.9308, "step": 1228 }, { "epoch": 3.0725, "grad_norm": 5.625, "learning_rate": 1.9999972432200928e-05, "loss": 1.9836, "step": 1229 }, { "epoch": 3.075, "grad_norm": 5.84375, "learning_rate": 1.999997049819457e-05, "loss": 1.7627, "step": 1230 }, { "epoch": 3.0775, "grad_norm": 6.1875, "learning_rate": 1.999996849862881e-05, "loss": 1.8379, "step": 1231 }, { "epoch": 3.08, "grad_norm": 6.21875, "learning_rate": 1.999996643350365e-05, "loss": 1.8771, "step": 1232 }, { "epoch": 3.0825, "grad_norm": 5.84375, "learning_rate": 1.9999964302819117e-05, "loss": 1.9687, "step": 1233 }, { "epoch": 3.085, "grad_norm": 6.46875, "learning_rate": 1.9999962106575218e-05, "loss": 1.7232, "step": 1234 }, { "epoch": 3.0875, "grad_norm": 5.53125, "learning_rate": 1.999995984477196e-05, "loss": 1.7689, "step": 1235 }, { "epoch": 3.09, "grad_norm": 4.59375, "learning_rate": 1.9999957517409375e-05, "loss": 1.8905, "step": 1236 }, { "epoch": 3.0925, "grad_norm": 5.09375, "learning_rate": 1.9999955124487462e-05, "loss": 1.7699, "step": 1237 }, { "epoch": 3.095, "grad_norm": 5.375, "learning_rate": 1.9999952666006245e-05, "loss": 1.8356, "step": 1238 }, { "epoch": 3.0975, "grad_norm": 5.125, "learning_rate": 1.999995014196574e-05, "loss": 1.952, "step": 1239 }, { "epoch": 3.1, "grad_norm": 4.5625, "learning_rate": 1.999994755236596e-05, "loss": 1.9456, "step": 1240 }, { "epoch": 3.1025, "grad_norm": 5.4375, "learning_rate": 1.9999944897206925e-05, "loss": 1.787, "step": 1241 }, { "epoch": 3.105, "grad_norm": 5.53125, "learning_rate": 1.9999942176488655e-05, "loss": 1.9328, "step": 1242 }, { "epoch": 3.1075, "grad_norm": 6.03125, "learning_rate": 1.999993939021116e-05, "loss": 1.8243, "step": 1243 }, { "epoch": 3.11, "grad_norm": 5.625, "learning_rate": 1.999993653837446e-05, "loss": 1.9167, "step": 1244 }, { "epoch": 3.1125, "grad_norm": 6.0, "learning_rate": 1.999993362097858e-05, "loss": 1.7933, "step": 1245 }, { "epoch": 3.115, "grad_norm": 10.4375, "learning_rate": 1.9999930638023536e-05, "loss": 1.821, "step": 1246 }, { "epoch": 3.1175, "grad_norm": 4.34375, "learning_rate": 1.9999927589509346e-05, "loss": 1.6155, "step": 1247 }, { "epoch": 3.12, "grad_norm": 5.96875, "learning_rate": 1.999992447543603e-05, "loss": 1.7322, "step": 1248 }, { "epoch": 3.1225, "grad_norm": 5.375, "learning_rate": 1.999992129580361e-05, "loss": 1.7875, "step": 1249 }, { "epoch": 3.125, "grad_norm": 7.25, "learning_rate": 1.999991805061211e-05, "loss": 2.0122, "step": 1250 }, { "epoch": 3.1275, "grad_norm": 5.46875, "learning_rate": 1.999991473986154e-05, "loss": 1.6609, "step": 1251 }, { "epoch": 3.13, "grad_norm": 9.25, "learning_rate": 1.9999911363551933e-05, "loss": 1.9097, "step": 1252 }, { "epoch": 3.1325, "grad_norm": 6.90625, "learning_rate": 1.9999907921683307e-05, "loss": 1.961, "step": 1253 }, { "epoch": 3.135, "grad_norm": 7.25, "learning_rate": 1.9999904414255685e-05, "loss": 1.8295, "step": 1254 }, { "epoch": 3.1375, "grad_norm": 3.09375, "learning_rate": 1.999990084126909e-05, "loss": 1.648, "step": 1255 }, { "epoch": 3.14, "grad_norm": 6.09375, "learning_rate": 1.9999897202723546e-05, "loss": 2.0132, "step": 1256 }, { "epoch": 3.1425, "grad_norm": 5.1875, "learning_rate": 1.9999893498619076e-05, "loss": 1.7464, "step": 1257 }, { "epoch": 3.145, "grad_norm": 4.4375, "learning_rate": 1.9999889728955702e-05, "loss": 1.6837, "step": 1258 }, { "epoch": 3.1475, "grad_norm": 11.4375, "learning_rate": 1.9999885893733452e-05, "loss": 1.7472, "step": 1259 }, { "epoch": 3.15, "grad_norm": 5.0625, "learning_rate": 1.9999881992952353e-05, "loss": 1.8696, "step": 1260 }, { "epoch": 3.1525, "grad_norm": 6.09375, "learning_rate": 1.9999878026612425e-05, "loss": 1.8059, "step": 1261 }, { "epoch": 3.155, "grad_norm": 6.78125, "learning_rate": 1.9999873994713698e-05, "loss": 2.0424, "step": 1262 }, { "epoch": 3.1575, "grad_norm": 6.65625, "learning_rate": 1.9999869897256197e-05, "loss": 1.9482, "step": 1263 }, { "epoch": 3.16, "grad_norm": 6.15625, "learning_rate": 1.999986573423995e-05, "loss": 1.9045, "step": 1264 }, { "epoch": 3.1625, "grad_norm": 6.6875, "learning_rate": 1.999986150566498e-05, "loss": 1.8092, "step": 1265 }, { "epoch": 3.165, "grad_norm": 6.0625, "learning_rate": 1.999985721153132e-05, "loss": 1.8764, "step": 1266 }, { "epoch": 3.1675, "grad_norm": 6.0, "learning_rate": 1.9999852851838995e-05, "loss": 1.6738, "step": 1267 }, { "epoch": 3.17, "grad_norm": 6.21875, "learning_rate": 1.9999848426588038e-05, "loss": 1.8563, "step": 1268 }, { "epoch": 3.1725, "grad_norm": 5.4375, "learning_rate": 1.9999843935778475e-05, "loss": 1.9142, "step": 1269 }, { "epoch": 3.175, "grad_norm": 8.0, "learning_rate": 1.9999839379410334e-05, "loss": 1.8836, "step": 1270 }, { "epoch": 3.1775, "grad_norm": 8.875, "learning_rate": 1.9999834757483646e-05, "loss": 2.0093, "step": 1271 }, { "epoch": 3.18, "grad_norm": 5.21875, "learning_rate": 1.999983006999844e-05, "loss": 1.8991, "step": 1272 }, { "epoch": 3.1825, "grad_norm": 5.03125, "learning_rate": 1.999982531695475e-05, "loss": 1.7243, "step": 1273 }, { "epoch": 3.185, "grad_norm": 6.40625, "learning_rate": 1.9999820498352606e-05, "loss": 2.0485, "step": 1274 }, { "epoch": 3.1875, "grad_norm": 5.875, "learning_rate": 1.999981561419204e-05, "loss": 1.8308, "step": 1275 }, { "epoch": 3.19, "grad_norm": 6.25, "learning_rate": 1.999981066447308e-05, "loss": 1.8994, "step": 1276 }, { "epoch": 3.1925, "grad_norm": 6.15625, "learning_rate": 1.999980564919576e-05, "loss": 1.8738, "step": 1277 }, { "epoch": 3.195, "grad_norm": 4.78125, "learning_rate": 1.999980056836012e-05, "loss": 1.7078, "step": 1278 }, { "epoch": 3.1975, "grad_norm": 9.75, "learning_rate": 1.9999795421966184e-05, "loss": 1.7754, "step": 1279 }, { "epoch": 3.2, "grad_norm": 10.375, "learning_rate": 1.999979021001399e-05, "loss": 1.9362, "step": 1280 }, { "epoch": 3.2025, "grad_norm": 7.03125, "learning_rate": 1.9999784932503573e-05, "loss": 1.8958, "step": 1281 }, { "epoch": 3.205, "grad_norm": 5.53125, "learning_rate": 1.999977958943496e-05, "loss": 1.655, "step": 1282 }, { "epoch": 3.2075, "grad_norm": 9.0625, "learning_rate": 1.99997741808082e-05, "loss": 1.9573, "step": 1283 }, { "epoch": 3.21, "grad_norm": 6.53125, "learning_rate": 1.9999768706623318e-05, "loss": 1.7896, "step": 1284 }, { "epoch": 3.2125, "grad_norm": 5.75, "learning_rate": 1.9999763166880354e-05, "loss": 1.8975, "step": 1285 }, { "epoch": 3.215, "grad_norm": 9.6875, "learning_rate": 1.9999757561579338e-05, "loss": 2.1004, "step": 1286 }, { "epoch": 3.2175, "grad_norm": 6.3125, "learning_rate": 1.9999751890720315e-05, "loss": 1.8492, "step": 1287 }, { "epoch": 3.22, "grad_norm": 4.8125, "learning_rate": 1.9999746154303318e-05, "loss": 1.914, "step": 1288 }, { "epoch": 3.2225, "grad_norm": 6.53125, "learning_rate": 1.9999740352328386e-05, "loss": 1.8887, "step": 1289 }, { "epoch": 3.225, "grad_norm": 5.375, "learning_rate": 1.9999734484795557e-05, "loss": 1.8045, "step": 1290 }, { "epoch": 3.2275, "grad_norm": 7.09375, "learning_rate": 1.9999728551704865e-05, "loss": 1.8272, "step": 1291 }, { "epoch": 3.23, "grad_norm": 7.84375, "learning_rate": 1.9999722553056354e-05, "loss": 1.7735, "step": 1292 }, { "epoch": 3.2325, "grad_norm": 4.875, "learning_rate": 1.9999716488850068e-05, "loss": 1.7994, "step": 1293 }, { "epoch": 3.235, "grad_norm": 4.78125, "learning_rate": 1.9999710359086037e-05, "loss": 1.9198, "step": 1294 }, { "epoch": 3.2375, "grad_norm": 7.03125, "learning_rate": 1.99997041637643e-05, "loss": 1.7248, "step": 1295 }, { "epoch": 3.24, "grad_norm": 5.625, "learning_rate": 1.999969790288491e-05, "loss": 1.5721, "step": 1296 }, { "epoch": 3.2425, "grad_norm": 7.84375, "learning_rate": 1.99996915764479e-05, "loss": 1.9604, "step": 1297 }, { "epoch": 3.245, "grad_norm": 6.15625, "learning_rate": 1.9999685184453306e-05, "loss": 1.9334, "step": 1298 }, { "epoch": 3.2475, "grad_norm": 7.5625, "learning_rate": 1.9999678726901182e-05, "loss": 1.7484, "step": 1299 }, { "epoch": 3.25, "grad_norm": 7.625, "learning_rate": 1.9999672203791564e-05, "loss": 1.7856, "step": 1300 }, { "epoch": 3.2525, "grad_norm": 9.25, "learning_rate": 1.9999665615124493e-05, "loss": 1.8987, "step": 1301 }, { "epoch": 3.255, "grad_norm": 7.28125, "learning_rate": 1.9999658960900016e-05, "loss": 1.7618, "step": 1302 }, { "epoch": 3.2575, "grad_norm": 7.3125, "learning_rate": 1.9999652241118174e-05, "loss": 1.8424, "step": 1303 }, { "epoch": 3.26, "grad_norm": 3.90625, "learning_rate": 1.9999645455779014e-05, "loss": 1.6625, "step": 1304 }, { "epoch": 3.2625, "grad_norm": 7.34375, "learning_rate": 1.9999638604882577e-05, "loss": 1.8203, "step": 1305 }, { "epoch": 3.265, "grad_norm": 5.625, "learning_rate": 1.999963168842891e-05, "loss": 1.9853, "step": 1306 }, { "epoch": 3.2675, "grad_norm": 5.9375, "learning_rate": 1.9999624706418058e-05, "loss": 2.0641, "step": 1307 }, { "epoch": 3.27, "grad_norm": 4.96875, "learning_rate": 1.9999617658850067e-05, "loss": 1.8185, "step": 1308 }, { "epoch": 3.2725, "grad_norm": 11.25, "learning_rate": 1.999961054572498e-05, "loss": 1.6709, "step": 1309 }, { "epoch": 3.275, "grad_norm": 5.5, "learning_rate": 1.9999603367042848e-05, "loss": 1.8351, "step": 1310 }, { "epoch": 3.2775, "grad_norm": 4.34375, "learning_rate": 1.9999596122803714e-05, "loss": 1.8669, "step": 1311 }, { "epoch": 3.2800000000000002, "grad_norm": 5.71875, "learning_rate": 1.999958881300763e-05, "loss": 1.8311, "step": 1312 }, { "epoch": 3.2824999999999998, "grad_norm": 4.9375, "learning_rate": 1.9999581437654645e-05, "loss": 1.9144, "step": 1313 }, { "epoch": 3.285, "grad_norm": 6.1875, "learning_rate": 1.99995739967448e-05, "loss": 1.8694, "step": 1314 }, { "epoch": 3.2875, "grad_norm": 6.0625, "learning_rate": 1.999956649027815e-05, "loss": 1.7634, "step": 1315 }, { "epoch": 3.29, "grad_norm": 6.40625, "learning_rate": 1.9999558918254746e-05, "loss": 1.8373, "step": 1316 }, { "epoch": 3.2925, "grad_norm": 4.75, "learning_rate": 1.999955128067463e-05, "loss": 2.0522, "step": 1317 }, { "epoch": 3.295, "grad_norm": 9.75, "learning_rate": 1.9999543577537853e-05, "loss": 1.8544, "step": 1318 }, { "epoch": 3.2975, "grad_norm": 5.9375, "learning_rate": 1.9999535808844473e-05, "loss": 1.5974, "step": 1319 }, { "epoch": 3.3, "grad_norm": 13.3125, "learning_rate": 1.999952797459453e-05, "loss": 1.6903, "step": 1320 }, { "epoch": 3.3025, "grad_norm": 5.71875, "learning_rate": 1.999952007478809e-05, "loss": 1.6247, "step": 1321 }, { "epoch": 3.305, "grad_norm": 7.8125, "learning_rate": 1.9999512109425195e-05, "loss": 2.0017, "step": 1322 }, { "epoch": 3.3075, "grad_norm": 6.21875, "learning_rate": 1.9999504078505895e-05, "loss": 1.8568, "step": 1323 }, { "epoch": 3.31, "grad_norm": 5.0625, "learning_rate": 1.999949598203025e-05, "loss": 1.7641, "step": 1324 }, { "epoch": 3.3125, "grad_norm": 5.0, "learning_rate": 1.9999487819998307e-05, "loss": 1.6499, "step": 1325 }, { "epoch": 3.315, "grad_norm": 6.28125, "learning_rate": 1.9999479592410123e-05, "loss": 1.9493, "step": 1326 }, { "epoch": 3.3175, "grad_norm": 8.4375, "learning_rate": 1.999947129926575e-05, "loss": 1.762, "step": 1327 }, { "epoch": 3.32, "grad_norm": 6.34375, "learning_rate": 1.9999462940565242e-05, "loss": 1.7612, "step": 1328 }, { "epoch": 3.3225, "grad_norm": 8.5625, "learning_rate": 1.999945451630866e-05, "loss": 1.7599, "step": 1329 }, { "epoch": 3.325, "grad_norm": 27.75, "learning_rate": 1.999944602649605e-05, "loss": 1.9548, "step": 1330 }, { "epoch": 3.3275, "grad_norm": 5.34375, "learning_rate": 1.9999437471127475e-05, "loss": 1.9742, "step": 1331 }, { "epoch": 3.33, "grad_norm": 5.25, "learning_rate": 1.9999428850202985e-05, "loss": 1.8431, "step": 1332 }, { "epoch": 3.3325, "grad_norm": 7.40625, "learning_rate": 1.999942016372264e-05, "loss": 2.0707, "step": 1333 }, { "epoch": 3.335, "grad_norm": 7.1875, "learning_rate": 1.99994114116865e-05, "loss": 1.9062, "step": 1334 }, { "epoch": 3.3375, "grad_norm": 5.5, "learning_rate": 1.9999402594094617e-05, "loss": 1.6637, "step": 1335 }, { "epoch": 3.34, "grad_norm": 5.15625, "learning_rate": 1.999939371094705e-05, "loss": 1.8927, "step": 1336 }, { "epoch": 3.3425000000000002, "grad_norm": 6.1875, "learning_rate": 1.9999384762243858e-05, "loss": 1.97, "step": 1337 }, { "epoch": 3.3449999999999998, "grad_norm": 5.625, "learning_rate": 1.99993757479851e-05, "loss": 1.9023, "step": 1338 }, { "epoch": 3.3475, "grad_norm": 7.15625, "learning_rate": 1.9999366668170835e-05, "loss": 1.6447, "step": 1339 }, { "epoch": 3.35, "grad_norm": 9.8125, "learning_rate": 1.9999357522801125e-05, "loss": 1.9588, "step": 1340 }, { "epoch": 3.3525, "grad_norm": 8.3125, "learning_rate": 1.999934831187602e-05, "loss": 1.9374, "step": 1341 }, { "epoch": 3.355, "grad_norm": 6.15625, "learning_rate": 1.9999339035395593e-05, "loss": 1.8469, "step": 1342 }, { "epoch": 3.3575, "grad_norm": 8.9375, "learning_rate": 1.99993296933599e-05, "loss": 1.8164, "step": 1343 }, { "epoch": 3.36, "grad_norm": 7.40625, "learning_rate": 1.9999320285769e-05, "loss": 1.8471, "step": 1344 }, { "epoch": 3.3625, "grad_norm": 9.5, "learning_rate": 1.9999310812622956e-05, "loss": 1.9297, "step": 1345 }, { "epoch": 3.365, "grad_norm": 5.25, "learning_rate": 1.9999301273921832e-05, "loss": 1.9223, "step": 1346 }, { "epoch": 3.3675, "grad_norm": 4.78125, "learning_rate": 1.9999291669665687e-05, "loss": 1.6741, "step": 1347 }, { "epoch": 3.37, "grad_norm": 7.28125, "learning_rate": 1.9999281999854586e-05, "loss": 2.0234, "step": 1348 }, { "epoch": 3.3725, "grad_norm": 10.25, "learning_rate": 1.9999272264488593e-05, "loss": 1.7774, "step": 1349 }, { "epoch": 3.375, "grad_norm": 6.3125, "learning_rate": 1.9999262463567772e-05, "loss": 1.8516, "step": 1350 }, { "epoch": 3.3775, "grad_norm": 5.3125, "learning_rate": 1.9999252597092185e-05, "loss": 1.8844, "step": 1351 }, { "epoch": 3.38, "grad_norm": 5.65625, "learning_rate": 1.99992426650619e-05, "loss": 1.7647, "step": 1352 }, { "epoch": 3.3825, "grad_norm": 5.1875, "learning_rate": 1.9999232667476976e-05, "loss": 1.6803, "step": 1353 }, { "epoch": 3.385, "grad_norm": 7.0, "learning_rate": 1.999922260433749e-05, "loss": 1.8999, "step": 1354 }, { "epoch": 3.3875, "grad_norm": 6.71875, "learning_rate": 1.9999212475643497e-05, "loss": 1.8849, "step": 1355 }, { "epoch": 3.39, "grad_norm": 5.46875, "learning_rate": 1.9999202281395064e-05, "loss": 1.7411, "step": 1356 }, { "epoch": 3.3925, "grad_norm": 6.25, "learning_rate": 1.9999192021592267e-05, "loss": 1.8958, "step": 1357 }, { "epoch": 3.395, "grad_norm": 8.25, "learning_rate": 1.9999181696235165e-05, "loss": 1.954, "step": 1358 }, { "epoch": 3.3975, "grad_norm": 4.875, "learning_rate": 1.9999171305323824e-05, "loss": 1.6184, "step": 1359 }, { "epoch": 3.4, "grad_norm": 7.03125, "learning_rate": 1.999916084885832e-05, "loss": 1.7274, "step": 1360 }, { "epoch": 3.4025, "grad_norm": 7.09375, "learning_rate": 1.9999150326838718e-05, "loss": 1.8882, "step": 1361 }, { "epoch": 3.4050000000000002, "grad_norm": 5.5, "learning_rate": 1.9999139739265084e-05, "loss": 1.765, "step": 1362 }, { "epoch": 3.4074999999999998, "grad_norm": 5.875, "learning_rate": 1.9999129086137493e-05, "loss": 1.6507, "step": 1363 }, { "epoch": 3.41, "grad_norm": 4.84375, "learning_rate": 1.999911836745601e-05, "loss": 1.7585, "step": 1364 }, { "epoch": 3.4125, "grad_norm": 6.34375, "learning_rate": 1.9999107583220706e-05, "loss": 1.7813, "step": 1365 }, { "epoch": 3.415, "grad_norm": 6.03125, "learning_rate": 1.9999096733431655e-05, "loss": 1.9594, "step": 1366 }, { "epoch": 3.4175, "grad_norm": 7.59375, "learning_rate": 1.9999085818088922e-05, "loss": 1.7244, "step": 1367 }, { "epoch": 3.42, "grad_norm": 6.125, "learning_rate": 1.9999074837192588e-05, "loss": 1.9529, "step": 1368 }, { "epoch": 3.4225, "grad_norm": 4.21875, "learning_rate": 1.9999063790742717e-05, "loss": 1.7397, "step": 1369 }, { "epoch": 3.425, "grad_norm": 6.25, "learning_rate": 1.9999052678739385e-05, "loss": 1.8908, "step": 1370 }, { "epoch": 3.4275, "grad_norm": 9.375, "learning_rate": 1.9999041501182664e-05, "loss": 2.0681, "step": 1371 }, { "epoch": 3.43, "grad_norm": 6.25, "learning_rate": 1.9999030258072624e-05, "loss": 1.9083, "step": 1372 }, { "epoch": 3.4325, "grad_norm": 8.3125, "learning_rate": 1.9999018949409343e-05, "loss": 1.9515, "step": 1373 }, { "epoch": 3.435, "grad_norm": 7.1875, "learning_rate": 1.9999007575192893e-05, "loss": 1.9328, "step": 1374 }, { "epoch": 3.4375, "grad_norm": 5.25, "learning_rate": 1.9998996135423355e-05, "loss": 1.7643, "step": 1375 }, { "epoch": 3.44, "grad_norm": 6.40625, "learning_rate": 1.999898463010079e-05, "loss": 1.8631, "step": 1376 }, { "epoch": 3.4425, "grad_norm": 5.0625, "learning_rate": 1.999897305922529e-05, "loss": 1.7378, "step": 1377 }, { "epoch": 3.445, "grad_norm": 6.75, "learning_rate": 1.999896142279692e-05, "loss": 1.7258, "step": 1378 }, { "epoch": 3.4475, "grad_norm": 8.4375, "learning_rate": 1.9998949720815758e-05, "loss": 1.9329, "step": 1379 }, { "epoch": 3.45, "grad_norm": 5.625, "learning_rate": 1.999893795328188e-05, "loss": 1.8169, "step": 1380 }, { "epoch": 3.4525, "grad_norm": 5.5625, "learning_rate": 1.9998926120195368e-05, "loss": 1.9561, "step": 1381 }, { "epoch": 3.455, "grad_norm": 9.625, "learning_rate": 1.9998914221556295e-05, "loss": 2.0267, "step": 1382 }, { "epoch": 3.4575, "grad_norm": 4.09375, "learning_rate": 1.9998902257364742e-05, "loss": 1.6806, "step": 1383 }, { "epoch": 3.46, "grad_norm": 5.78125, "learning_rate": 1.9998890227620784e-05, "loss": 1.9261, "step": 1384 }, { "epoch": 3.4625, "grad_norm": 4.28125, "learning_rate": 1.9998878132324502e-05, "loss": 1.8032, "step": 1385 }, { "epoch": 3.465, "grad_norm": 4.3125, "learning_rate": 1.999886597147598e-05, "loss": 1.6907, "step": 1386 }, { "epoch": 3.4675000000000002, "grad_norm": 8.25, "learning_rate": 1.9998853745075287e-05, "loss": 1.981, "step": 1387 }, { "epoch": 3.4699999999999998, "grad_norm": 4.90625, "learning_rate": 1.999884145312251e-05, "loss": 1.6315, "step": 1388 }, { "epoch": 3.4725, "grad_norm": 6.84375, "learning_rate": 1.9998829095617726e-05, "loss": 1.8004, "step": 1389 }, { "epoch": 3.475, "grad_norm": 5.125, "learning_rate": 1.9998816672561022e-05, "loss": 1.623, "step": 1390 }, { "epoch": 3.4775, "grad_norm": 6.8125, "learning_rate": 1.9998804183952476e-05, "loss": 1.8164, "step": 1391 }, { "epoch": 3.48, "grad_norm": 6.5, "learning_rate": 1.9998791629792172e-05, "loss": 1.9598, "step": 1392 }, { "epoch": 3.4825, "grad_norm": 6.6875, "learning_rate": 1.999877901008019e-05, "loss": 1.8943, "step": 1393 }, { "epoch": 3.485, "grad_norm": 6.40625, "learning_rate": 1.9998766324816606e-05, "loss": 1.6305, "step": 1394 }, { "epoch": 3.4875, "grad_norm": 5.09375, "learning_rate": 1.9998753574001516e-05, "loss": 1.8788, "step": 1395 }, { "epoch": 3.49, "grad_norm": 34.25, "learning_rate": 1.9998740757634998e-05, "loss": 1.9651, "step": 1396 }, { "epoch": 3.4925, "grad_norm": 7.59375, "learning_rate": 1.999872787571713e-05, "loss": 1.9668, "step": 1397 }, { "epoch": 3.495, "grad_norm": 4.78125, "learning_rate": 1.9998714928248008e-05, "loss": 1.7346, "step": 1398 }, { "epoch": 3.4975, "grad_norm": 4.90625, "learning_rate": 1.999870191522771e-05, "loss": 1.7507, "step": 1399 }, { "epoch": 3.5, "grad_norm": 5.125, "learning_rate": 1.9998688836656322e-05, "loss": 1.7371, "step": 1400 }, { "epoch": 3.5025, "grad_norm": 6.71875, "learning_rate": 1.999867569253393e-05, "loss": 1.77, "step": 1401 }, { "epoch": 3.505, "grad_norm": 7.71875, "learning_rate": 1.999866248286062e-05, "loss": 1.7766, "step": 1402 }, { "epoch": 3.5075, "grad_norm": 6.78125, "learning_rate": 1.999864920763648e-05, "loss": 1.6516, "step": 1403 }, { "epoch": 3.51, "grad_norm": 4.9375, "learning_rate": 1.9998635866861597e-05, "loss": 1.7646, "step": 1404 }, { "epoch": 3.5125, "grad_norm": 8.375, "learning_rate": 1.9998622460536054e-05, "loss": 1.9261, "step": 1405 }, { "epoch": 3.515, "grad_norm": 6.65625, "learning_rate": 1.9998608988659943e-05, "loss": 1.8932, "step": 1406 }, { "epoch": 3.5175, "grad_norm": 15.3125, "learning_rate": 1.9998595451233355e-05, "loss": 1.7696, "step": 1407 }, { "epoch": 3.52, "grad_norm": 5.1875, "learning_rate": 1.999858184825637e-05, "loss": 1.5215, "step": 1408 }, { "epoch": 3.5225, "grad_norm": 6.5625, "learning_rate": 1.9998568179729086e-05, "loss": 1.7032, "step": 1409 }, { "epoch": 3.525, "grad_norm": 5.0625, "learning_rate": 1.999855444565159e-05, "loss": 1.8159, "step": 1410 }, { "epoch": 3.5275, "grad_norm": 4.59375, "learning_rate": 1.999854064602397e-05, "loss": 1.732, "step": 1411 }, { "epoch": 3.5300000000000002, "grad_norm": 6.21875, "learning_rate": 1.9998526780846314e-05, "loss": 1.754, "step": 1412 }, { "epoch": 3.5324999999999998, "grad_norm": 4.5625, "learning_rate": 1.999851285011872e-05, "loss": 1.7183, "step": 1413 }, { "epoch": 3.535, "grad_norm": 6.3125, "learning_rate": 1.9998498853841276e-05, "loss": 1.836, "step": 1414 }, { "epoch": 3.5375, "grad_norm": 7.84375, "learning_rate": 1.9998484792014073e-05, "loss": 1.8469, "step": 1415 }, { "epoch": 3.54, "grad_norm": 6.375, "learning_rate": 1.9998470664637205e-05, "loss": 1.6206, "step": 1416 }, { "epoch": 3.5425, "grad_norm": 15.0625, "learning_rate": 1.9998456471710763e-05, "loss": 1.8973, "step": 1417 }, { "epoch": 3.545, "grad_norm": 9.0625, "learning_rate": 1.999844221323484e-05, "loss": 1.8655, "step": 1418 }, { "epoch": 3.5475, "grad_norm": 5.0625, "learning_rate": 1.999842788920953e-05, "loss": 1.7477, "step": 1419 }, { "epoch": 3.55, "grad_norm": 5.3125, "learning_rate": 1.9998413499634927e-05, "loss": 1.7058, "step": 1420 }, { "epoch": 3.5525, "grad_norm": 7.15625, "learning_rate": 1.9998399044511123e-05, "loss": 1.8726, "step": 1421 }, { "epoch": 3.555, "grad_norm": 3.640625, "learning_rate": 1.9998384523838217e-05, "loss": 1.9592, "step": 1422 }, { "epoch": 3.5575, "grad_norm": 8.1875, "learning_rate": 1.99983699376163e-05, "loss": 1.8741, "step": 1423 }, { "epoch": 3.56, "grad_norm": 8.3125, "learning_rate": 1.9998355285845473e-05, "loss": 1.9137, "step": 1424 }, { "epoch": 3.5625, "grad_norm": 4.5625, "learning_rate": 1.9998340568525828e-05, "loss": 1.7031, "step": 1425 }, { "epoch": 3.565, "grad_norm": 7.5625, "learning_rate": 1.999832578565746e-05, "loss": 1.6793, "step": 1426 }, { "epoch": 3.5675, "grad_norm": 6.1875, "learning_rate": 1.9998310937240474e-05, "loss": 1.531, "step": 1427 }, { "epoch": 3.57, "grad_norm": 7.75, "learning_rate": 1.9998296023274958e-05, "loss": 1.8679, "step": 1428 }, { "epoch": 3.5725, "grad_norm": 6.625, "learning_rate": 1.999828104376101e-05, "loss": 1.8185, "step": 1429 }, { "epoch": 3.575, "grad_norm": 9.0625, "learning_rate": 1.999826599869874e-05, "loss": 1.8302, "step": 1430 }, { "epoch": 3.5775, "grad_norm": 5.59375, "learning_rate": 1.9998250888088232e-05, "loss": 1.6148, "step": 1431 }, { "epoch": 3.58, "grad_norm": 4.65625, "learning_rate": 1.9998235711929593e-05, "loss": 1.6564, "step": 1432 }, { "epoch": 3.5825, "grad_norm": 5.75, "learning_rate": 1.999822047022292e-05, "loss": 1.7402, "step": 1433 }, { "epoch": 3.585, "grad_norm": 5.8125, "learning_rate": 1.9998205162968314e-05, "loss": 1.8918, "step": 1434 }, { "epoch": 3.5875, "grad_norm": 7.90625, "learning_rate": 1.9998189790165873e-05, "loss": 1.7793, "step": 1435 }, { "epoch": 3.59, "grad_norm": 7.6875, "learning_rate": 1.9998174351815704e-05, "loss": 1.8247, "step": 1436 }, { "epoch": 3.5925000000000002, "grad_norm": 6.78125, "learning_rate": 1.99981588479179e-05, "loss": 1.7689, "step": 1437 }, { "epoch": 3.5949999999999998, "grad_norm": 5.28125, "learning_rate": 1.999814327847257e-05, "loss": 1.7379, "step": 1438 }, { "epoch": 3.5975, "grad_norm": 13.625, "learning_rate": 1.999812764347981e-05, "loss": 1.8967, "step": 1439 }, { "epoch": 3.6, "grad_norm": 7.84375, "learning_rate": 1.9998111942939727e-05, "loss": 1.7441, "step": 1440 }, { "epoch": 3.6025, "grad_norm": 3.75, "learning_rate": 1.9998096176852422e-05, "loss": 1.688, "step": 1441 }, { "epoch": 3.605, "grad_norm": 8.6875, "learning_rate": 1.9998080345218e-05, "loss": 1.9862, "step": 1442 }, { "epoch": 3.6075, "grad_norm": 7.84375, "learning_rate": 1.9998064448036562e-05, "loss": 1.8084, "step": 1443 }, { "epoch": 3.61, "grad_norm": 9.125, "learning_rate": 1.9998048485308213e-05, "loss": 1.7808, "step": 1444 }, { "epoch": 3.6125, "grad_norm": 9.375, "learning_rate": 1.9998032457033057e-05, "loss": 1.6826, "step": 1445 }, { "epoch": 3.615, "grad_norm": 7.25, "learning_rate": 1.9998016363211202e-05, "loss": 1.9048, "step": 1446 }, { "epoch": 3.6175, "grad_norm": 3.953125, "learning_rate": 1.9998000203842752e-05, "loss": 1.6748, "step": 1447 }, { "epoch": 3.62, "grad_norm": 7.0, "learning_rate": 1.9997983978927813e-05, "loss": 1.884, "step": 1448 }, { "epoch": 3.6225, "grad_norm": 9.125, "learning_rate": 1.999796768846649e-05, "loss": 1.9403, "step": 1449 }, { "epoch": 3.625, "grad_norm": 4.90625, "learning_rate": 1.999795133245889e-05, "loss": 1.7433, "step": 1450 }, { "epoch": 3.6275, "grad_norm": 6.59375, "learning_rate": 1.9997934910905124e-05, "loss": 1.8712, "step": 1451 }, { "epoch": 3.63, "grad_norm": 5.3125, "learning_rate": 1.9997918423805295e-05, "loss": 1.9213, "step": 1452 }, { "epoch": 3.6325, "grad_norm": 5.3125, "learning_rate": 1.999790187115951e-05, "loss": 1.8596, "step": 1453 }, { "epoch": 3.635, "grad_norm": 5.28125, "learning_rate": 1.9997885252967887e-05, "loss": 1.766, "step": 1454 }, { "epoch": 3.6375, "grad_norm": 6.3125, "learning_rate": 1.9997868569230524e-05, "loss": 1.6723, "step": 1455 }, { "epoch": 3.64, "grad_norm": 4.5, "learning_rate": 1.9997851819947537e-05, "loss": 1.6778, "step": 1456 }, { "epoch": 3.6425, "grad_norm": 5.90625, "learning_rate": 1.9997835005119036e-05, "loss": 1.6348, "step": 1457 }, { "epoch": 3.645, "grad_norm": 6.6875, "learning_rate": 1.9997818124745123e-05, "loss": 1.7351, "step": 1458 }, { "epoch": 3.6475, "grad_norm": 4.78125, "learning_rate": 1.9997801178825917e-05, "loss": 1.755, "step": 1459 }, { "epoch": 3.65, "grad_norm": 14.0, "learning_rate": 1.9997784167361526e-05, "loss": 1.835, "step": 1460 }, { "epoch": 3.6525, "grad_norm": 6.125, "learning_rate": 1.9997767090352063e-05, "loss": 1.7889, "step": 1461 }, { "epoch": 3.6550000000000002, "grad_norm": 6.0625, "learning_rate": 1.999774994779764e-05, "loss": 1.8466, "step": 1462 }, { "epoch": 3.6574999999999998, "grad_norm": 6.75, "learning_rate": 1.9997732739698366e-05, "loss": 1.8405, "step": 1463 }, { "epoch": 3.66, "grad_norm": 7.53125, "learning_rate": 1.9997715466054357e-05, "loss": 1.8527, "step": 1464 }, { "epoch": 3.6625, "grad_norm": 5.875, "learning_rate": 1.9997698126865726e-05, "loss": 1.8284, "step": 1465 }, { "epoch": 3.665, "grad_norm": 9.9375, "learning_rate": 1.9997680722132588e-05, "loss": 1.8911, "step": 1466 }, { "epoch": 3.6675, "grad_norm": 6.65625, "learning_rate": 1.999766325185505e-05, "loss": 1.9576, "step": 1467 }, { "epoch": 3.67, "grad_norm": 8.8125, "learning_rate": 1.9997645716033233e-05, "loss": 1.6502, "step": 1468 }, { "epoch": 3.6725, "grad_norm": 6.09375, "learning_rate": 1.9997628114667254e-05, "loss": 1.6611, "step": 1469 }, { "epoch": 3.675, "grad_norm": 5.75, "learning_rate": 1.9997610447757222e-05, "loss": 1.6742, "step": 1470 }, { "epoch": 3.6775, "grad_norm": 7.21875, "learning_rate": 1.999759271530326e-05, "loss": 1.7665, "step": 1471 }, { "epoch": 3.68, "grad_norm": 8.625, "learning_rate": 1.999757491730548e-05, "loss": 1.828, "step": 1472 }, { "epoch": 3.6825, "grad_norm": 12.5, "learning_rate": 1.9997557053763995e-05, "loss": 2.1816, "step": 1473 }, { "epoch": 3.685, "grad_norm": 4.9375, "learning_rate": 1.999753912467893e-05, "loss": 1.7451, "step": 1474 }, { "epoch": 3.6875, "grad_norm": 7.46875, "learning_rate": 1.9997521130050393e-05, "loss": 1.6121, "step": 1475 }, { "epoch": 3.69, "grad_norm": 10.625, "learning_rate": 1.9997503069878515e-05, "loss": 1.8427, "step": 1476 }, { "epoch": 3.6925, "grad_norm": 8.0, "learning_rate": 1.99974849441634e-05, "loss": 1.9005, "step": 1477 }, { "epoch": 3.695, "grad_norm": 8.375, "learning_rate": 1.9997466752905178e-05, "loss": 1.9205, "step": 1478 }, { "epoch": 3.6975, "grad_norm": 6.875, "learning_rate": 1.9997448496103962e-05, "loss": 1.94, "step": 1479 }, { "epoch": 3.7, "grad_norm": 5.0625, "learning_rate": 1.9997430173759876e-05, "loss": 1.8244, "step": 1480 }, { "epoch": 3.7025, "grad_norm": 6.21875, "learning_rate": 1.9997411785873034e-05, "loss": 1.8227, "step": 1481 }, { "epoch": 3.705, "grad_norm": 4.9375, "learning_rate": 1.9997393332443566e-05, "loss": 1.6223, "step": 1482 }, { "epoch": 3.7075, "grad_norm": 7.75, "learning_rate": 1.9997374813471582e-05, "loss": 1.9185, "step": 1483 }, { "epoch": 3.71, "grad_norm": 6.59375, "learning_rate": 1.9997356228957213e-05, "loss": 1.8623, "step": 1484 }, { "epoch": 3.7125, "grad_norm": 5.96875, "learning_rate": 1.9997337578900575e-05, "loss": 1.5856, "step": 1485 }, { "epoch": 3.715, "grad_norm": 5.875, "learning_rate": 1.9997318863301792e-05, "loss": 1.687, "step": 1486 }, { "epoch": 3.7175000000000002, "grad_norm": 6.8125, "learning_rate": 1.9997300082160985e-05, "loss": 1.8122, "step": 1487 }, { "epoch": 3.7199999999999998, "grad_norm": 5.625, "learning_rate": 1.999728123547828e-05, "loss": 1.6305, "step": 1488 }, { "epoch": 3.7225, "grad_norm": 7.625, "learning_rate": 1.99972623232538e-05, "loss": 1.8271, "step": 1489 }, { "epoch": 3.725, "grad_norm": 4.4375, "learning_rate": 1.9997243345487666e-05, "loss": 1.6786, "step": 1490 }, { "epoch": 3.7275, "grad_norm": 5.90625, "learning_rate": 1.9997224302180008e-05, "loss": 1.7659, "step": 1491 }, { "epoch": 3.73, "grad_norm": 4.8125, "learning_rate": 1.9997205193330943e-05, "loss": 1.6209, "step": 1492 }, { "epoch": 3.7325, "grad_norm": 6.3125, "learning_rate": 1.9997186018940606e-05, "loss": 1.7627, "step": 1493 }, { "epoch": 3.735, "grad_norm": 5.3125, "learning_rate": 1.9997166779009113e-05, "loss": 1.8959, "step": 1494 }, { "epoch": 3.7375, "grad_norm": 5.8125, "learning_rate": 1.99971474735366e-05, "loss": 1.8268, "step": 1495 }, { "epoch": 3.74, "grad_norm": 9.4375, "learning_rate": 1.9997128102523186e-05, "loss": 1.9583, "step": 1496 }, { "epoch": 3.7425, "grad_norm": 5.3125, "learning_rate": 1.9997108665968996e-05, "loss": 1.9799, "step": 1497 }, { "epoch": 3.745, "grad_norm": 5.125, "learning_rate": 1.9997089163874165e-05, "loss": 1.8317, "step": 1498 }, { "epoch": 3.7475, "grad_norm": 7.03125, "learning_rate": 1.9997069596238818e-05, "loss": 2.0116, "step": 1499 }, { "epoch": 3.75, "grad_norm": 6.53125, "learning_rate": 1.999704996306308e-05, "loss": 1.8035, "step": 1500 }, { "epoch": 3.7525, "grad_norm": 7.21875, "learning_rate": 1.9997030264347084e-05, "loss": 1.7798, "step": 1501 }, { "epoch": 3.755, "grad_norm": 7.40625, "learning_rate": 1.9997010500090962e-05, "loss": 1.6995, "step": 1502 }, { "epoch": 3.7575, "grad_norm": 19.125, "learning_rate": 1.9996990670294835e-05, "loss": 1.7571, "step": 1503 }, { "epoch": 3.76, "grad_norm": 7.21875, "learning_rate": 1.9996970774958836e-05, "loss": 1.9588, "step": 1504 }, { "epoch": 3.7625, "grad_norm": 8.1875, "learning_rate": 1.99969508140831e-05, "loss": 1.9389, "step": 1505 }, { "epoch": 3.765, "grad_norm": 6.90625, "learning_rate": 1.9996930787667752e-05, "loss": 1.7124, "step": 1506 }, { "epoch": 3.7675, "grad_norm": 5.96875, "learning_rate": 1.999691069571293e-05, "loss": 1.9817, "step": 1507 }, { "epoch": 3.77, "grad_norm": 10.25, "learning_rate": 1.9996890538218756e-05, "loss": 1.9642, "step": 1508 }, { "epoch": 3.7725, "grad_norm": 5.59375, "learning_rate": 1.999687031518537e-05, "loss": 1.6962, "step": 1509 }, { "epoch": 3.775, "grad_norm": 6.40625, "learning_rate": 1.9996850026612903e-05, "loss": 1.786, "step": 1510 }, { "epoch": 3.7775, "grad_norm": 9.125, "learning_rate": 1.9996829672501484e-05, "loss": 1.7104, "step": 1511 }, { "epoch": 3.7800000000000002, "grad_norm": 9.375, "learning_rate": 1.9996809252851254e-05, "loss": 1.9031, "step": 1512 }, { "epoch": 3.7824999999999998, "grad_norm": 7.59375, "learning_rate": 1.999678876766234e-05, "loss": 1.9096, "step": 1513 }, { "epoch": 3.785, "grad_norm": 33.5, "learning_rate": 1.9996768216934883e-05, "loss": 1.7586, "step": 1514 }, { "epoch": 3.7875, "grad_norm": 13.0625, "learning_rate": 1.999674760066901e-05, "loss": 1.6673, "step": 1515 }, { "epoch": 3.79, "grad_norm": 6.03125, "learning_rate": 1.9996726918864857e-05, "loss": 1.9368, "step": 1516 }, { "epoch": 3.7925, "grad_norm": 4.625, "learning_rate": 1.9996706171522567e-05, "loss": 1.9611, "step": 1517 }, { "epoch": 3.795, "grad_norm": 5.09375, "learning_rate": 1.9996685358642268e-05, "loss": 1.8891, "step": 1518 }, { "epoch": 3.7975, "grad_norm": 8.3125, "learning_rate": 1.9996664480224102e-05, "loss": 1.7419, "step": 1519 }, { "epoch": 3.8, "grad_norm": 6.90625, "learning_rate": 1.9996643536268202e-05, "loss": 1.6969, "step": 1520 }, { "epoch": 3.8025, "grad_norm": 6.71875, "learning_rate": 1.999662252677471e-05, "loss": 1.8065, "step": 1521 }, { "epoch": 3.805, "grad_norm": 8.9375, "learning_rate": 1.999660145174376e-05, "loss": 1.7327, "step": 1522 }, { "epoch": 3.8075, "grad_norm": 4.5, "learning_rate": 1.999658031117549e-05, "loss": 1.5646, "step": 1523 }, { "epoch": 3.81, "grad_norm": 10.875, "learning_rate": 1.9996559105070038e-05, "loss": 1.9675, "step": 1524 }, { "epoch": 3.8125, "grad_norm": 7.59375, "learning_rate": 1.9996537833427547e-05, "loss": 1.8708, "step": 1525 }, { "epoch": 3.815, "grad_norm": 5.8125, "learning_rate": 1.999651649624815e-05, "loss": 1.7491, "step": 1526 }, { "epoch": 3.8175, "grad_norm": 7.3125, "learning_rate": 1.9996495093531994e-05, "loss": 1.8178, "step": 1527 }, { "epoch": 3.82, "grad_norm": 8.75, "learning_rate": 1.999647362527922e-05, "loss": 1.9331, "step": 1528 }, { "epoch": 3.8225, "grad_norm": 7.84375, "learning_rate": 1.9996452091489963e-05, "loss": 1.7975, "step": 1529 }, { "epoch": 3.825, "grad_norm": 7.375, "learning_rate": 1.9996430492164363e-05, "loss": 1.9596, "step": 1530 }, { "epoch": 3.8275, "grad_norm": 6.8125, "learning_rate": 1.9996408827302564e-05, "loss": 1.9374, "step": 1531 }, { "epoch": 3.83, "grad_norm": 7.21875, "learning_rate": 1.999638709690471e-05, "loss": 1.7522, "step": 1532 }, { "epoch": 3.8325, "grad_norm": 9.375, "learning_rate": 1.9996365300970947e-05, "loss": 1.6418, "step": 1533 }, { "epoch": 3.835, "grad_norm": 8.125, "learning_rate": 1.9996343439501406e-05, "loss": 1.9129, "step": 1534 }, { "epoch": 3.8375, "grad_norm": 5.8125, "learning_rate": 1.9996321512496242e-05, "loss": 1.7624, "step": 1535 }, { "epoch": 3.84, "grad_norm": 5.15625, "learning_rate": 1.999629951995559e-05, "loss": 1.608, "step": 1536 }, { "epoch": 3.8425000000000002, "grad_norm": 5.125, "learning_rate": 1.9996277461879603e-05, "loss": 1.7599, "step": 1537 }, { "epoch": 3.8449999999999998, "grad_norm": 9.0, "learning_rate": 1.999625533826842e-05, "loss": 1.7447, "step": 1538 }, { "epoch": 3.8475, "grad_norm": 7.625, "learning_rate": 1.9996233149122187e-05, "loss": 1.6757, "step": 1539 }, { "epoch": 3.85, "grad_norm": 4.46875, "learning_rate": 1.9996210894441047e-05, "loss": 1.7555, "step": 1540 }, { "epoch": 3.8525, "grad_norm": 9.0, "learning_rate": 1.999618857422515e-05, "loss": 1.7566, "step": 1541 }, { "epoch": 3.855, "grad_norm": 5.65625, "learning_rate": 1.9996166188474643e-05, "loss": 1.7802, "step": 1542 }, { "epoch": 3.8575, "grad_norm": 5.5, "learning_rate": 1.9996143737189665e-05, "loss": 1.7707, "step": 1543 }, { "epoch": 3.86, "grad_norm": 9.5, "learning_rate": 1.9996121220370374e-05, "loss": 1.9189, "step": 1544 }, { "epoch": 3.8625, "grad_norm": 4.96875, "learning_rate": 1.999609863801691e-05, "loss": 1.758, "step": 1545 }, { "epoch": 3.865, "grad_norm": 5.75, "learning_rate": 1.999607599012942e-05, "loss": 1.7082, "step": 1546 }, { "epoch": 3.8675, "grad_norm": 4.65625, "learning_rate": 1.9996053276708064e-05, "loss": 1.6141, "step": 1547 }, { "epoch": 3.87, "grad_norm": 5.09375, "learning_rate": 1.9996030497752977e-05, "loss": 1.776, "step": 1548 }, { "epoch": 3.8725, "grad_norm": 6.40625, "learning_rate": 1.9996007653264317e-05, "loss": 1.6197, "step": 1549 }, { "epoch": 3.875, "grad_norm": 6.375, "learning_rate": 1.999598474324223e-05, "loss": 1.9554, "step": 1550 }, { "epoch": 3.8775, "grad_norm": 6.15625, "learning_rate": 1.9995961767686864e-05, "loss": 1.8441, "step": 1551 }, { "epoch": 3.88, "grad_norm": 7.15625, "learning_rate": 1.9995938726598374e-05, "loss": 1.8245, "step": 1552 }, { "epoch": 3.8825, "grad_norm": 5.0625, "learning_rate": 1.999591561997691e-05, "loss": 1.5678, "step": 1553 }, { "epoch": 3.885, "grad_norm": 4.3125, "learning_rate": 1.9995892447822624e-05, "loss": 1.8471, "step": 1554 }, { "epoch": 3.8875, "grad_norm": 7.875, "learning_rate": 1.999586921013567e-05, "loss": 1.6937, "step": 1555 }, { "epoch": 3.89, "grad_norm": 6.25, "learning_rate": 1.999584590691619e-05, "loss": 1.7083, "step": 1556 }, { "epoch": 3.8925, "grad_norm": 5.4375, "learning_rate": 1.9995822538164348e-05, "loss": 1.5665, "step": 1557 }, { "epoch": 3.895, "grad_norm": 6.75, "learning_rate": 1.9995799103880297e-05, "loss": 1.8389, "step": 1558 }, { "epoch": 3.8975, "grad_norm": 5.96875, "learning_rate": 1.9995775604064183e-05, "loss": 1.7076, "step": 1559 }, { "epoch": 3.9, "grad_norm": 8.1875, "learning_rate": 1.9995752038716166e-05, "loss": 1.8348, "step": 1560 }, { "epoch": 3.9025, "grad_norm": 5.78125, "learning_rate": 1.99957284078364e-05, "loss": 1.6774, "step": 1561 }, { "epoch": 3.9050000000000002, "grad_norm": 5.0625, "learning_rate": 1.9995704711425036e-05, "loss": 1.6839, "step": 1562 }, { "epoch": 3.9074999999999998, "grad_norm": 15.625, "learning_rate": 1.9995680949482234e-05, "loss": 1.859, "step": 1563 }, { "epoch": 3.91, "grad_norm": 7.375, "learning_rate": 1.999565712200814e-05, "loss": 1.7042, "step": 1564 }, { "epoch": 3.9125, "grad_norm": 6.96875, "learning_rate": 1.9995633229002925e-05, "loss": 1.9452, "step": 1565 }, { "epoch": 3.915, "grad_norm": 7.1875, "learning_rate": 1.999560927046674e-05, "loss": 1.7139, "step": 1566 }, { "epoch": 3.9175, "grad_norm": 6.96875, "learning_rate": 1.9995585246399737e-05, "loss": 1.7277, "step": 1567 }, { "epoch": 3.92, "grad_norm": 5.0625, "learning_rate": 1.999556115680208e-05, "loss": 1.6727, "step": 1568 }, { "epoch": 3.9225, "grad_norm": 5.75, "learning_rate": 1.999553700167392e-05, "loss": 1.7957, "step": 1569 }, { "epoch": 3.925, "grad_norm": 6.8125, "learning_rate": 1.999551278101542e-05, "loss": 1.7826, "step": 1570 }, { "epoch": 3.9275, "grad_norm": 4.84375, "learning_rate": 1.9995488494826743e-05, "loss": 1.6873, "step": 1571 }, { "epoch": 3.93, "grad_norm": 6.90625, "learning_rate": 1.999546414310804e-05, "loss": 2.1045, "step": 1572 }, { "epoch": 3.9325, "grad_norm": 5.625, "learning_rate": 1.9995439725859475e-05, "loss": 1.8251, "step": 1573 }, { "epoch": 3.935, "grad_norm": 4.9375, "learning_rate": 1.9995415243081206e-05, "loss": 1.6382, "step": 1574 }, { "epoch": 3.9375, "grad_norm": 5.84375, "learning_rate": 1.9995390694773396e-05, "loss": 1.7341, "step": 1575 }, { "epoch": 3.94, "grad_norm": 6.0625, "learning_rate": 1.9995366080936206e-05, "loss": 1.5863, "step": 1576 }, { "epoch": 3.9425, "grad_norm": 6.0625, "learning_rate": 1.9995341401569793e-05, "loss": 1.7767, "step": 1577 }, { "epoch": 3.945, "grad_norm": 9.1875, "learning_rate": 1.9995316656674322e-05, "loss": 1.8255, "step": 1578 }, { "epoch": 3.9475, "grad_norm": 6.65625, "learning_rate": 1.9995291846249954e-05, "loss": 1.9555, "step": 1579 }, { "epoch": 3.95, "grad_norm": 5.53125, "learning_rate": 1.9995266970296856e-05, "loss": 1.5834, "step": 1580 }, { "epoch": 3.9525, "grad_norm": 6.1875, "learning_rate": 1.9995242028815186e-05, "loss": 1.7005, "step": 1581 }, { "epoch": 3.955, "grad_norm": 5.0, "learning_rate": 1.999521702180511e-05, "loss": 1.842, "step": 1582 }, { "epoch": 3.9575, "grad_norm": 5.78125, "learning_rate": 1.9995191949266788e-05, "loss": 1.9002, "step": 1583 }, { "epoch": 3.96, "grad_norm": 5.1875, "learning_rate": 1.999516681120039e-05, "loss": 1.8775, "step": 1584 }, { "epoch": 3.9625, "grad_norm": 7.40625, "learning_rate": 1.9995141607606078e-05, "loss": 1.9813, "step": 1585 }, { "epoch": 3.965, "grad_norm": 16.375, "learning_rate": 1.999511633848402e-05, "loss": 1.8849, "step": 1586 }, { "epoch": 3.9675000000000002, "grad_norm": 5.0625, "learning_rate": 1.9995091003834376e-05, "loss": 1.743, "step": 1587 }, { "epoch": 3.9699999999999998, "grad_norm": 5.4375, "learning_rate": 1.9995065603657317e-05, "loss": 1.6352, "step": 1588 }, { "epoch": 3.9725, "grad_norm": 7.90625, "learning_rate": 1.9995040137953007e-05, "loss": 1.8006, "step": 1589 }, { "epoch": 3.975, "grad_norm": 8.9375, "learning_rate": 1.9995014606721615e-05, "loss": 1.8246, "step": 1590 }, { "epoch": 3.9775, "grad_norm": 6.09375, "learning_rate": 1.9994989009963307e-05, "loss": 1.6524, "step": 1591 }, { "epoch": 3.98, "grad_norm": 7.90625, "learning_rate": 1.999496334767825e-05, "loss": 1.7469, "step": 1592 }, { "epoch": 3.9825, "grad_norm": 9.625, "learning_rate": 1.9994937619866614e-05, "loss": 1.904, "step": 1593 }, { "epoch": 3.985, "grad_norm": 5.71875, "learning_rate": 1.9994911826528567e-05, "loss": 1.7921, "step": 1594 }, { "epoch": 3.9875, "grad_norm": 5.625, "learning_rate": 1.9994885967664276e-05, "loss": 2.0041, "step": 1595 }, { "epoch": 3.99, "grad_norm": 5.71875, "learning_rate": 1.9994860043273915e-05, "loss": 1.8936, "step": 1596 }, { "epoch": 3.9925, "grad_norm": 3.78125, "learning_rate": 1.999483405335765e-05, "loss": 1.6428, "step": 1597 }, { "epoch": 3.995, "grad_norm": 4.15625, "learning_rate": 1.9994807997915654e-05, "loss": 2.0505, "step": 1598 }, { "epoch": 3.9975, "grad_norm": 5.6875, "learning_rate": 1.99947818769481e-05, "loss": 1.996, "step": 1599 }, { "epoch": 4.0, "grad_norm": 6.59375, "learning_rate": 1.9994755690455154e-05, "loss": 1.7256, "step": 1600 }, { "epoch": 4.0025, "grad_norm": 5.5, "learning_rate": 1.9994729438436982e-05, "loss": 1.8179, "step": 1601 }, { "epoch": 4.005, "grad_norm": 5.46875, "learning_rate": 1.9994703120893772e-05, "loss": 1.6242, "step": 1602 }, { "epoch": 4.0075, "grad_norm": 6.3125, "learning_rate": 1.9994676737825685e-05, "loss": 1.8761, "step": 1603 }, { "epoch": 4.01, "grad_norm": 12.9375, "learning_rate": 1.9994650289232898e-05, "loss": 1.833, "step": 1604 }, { "epoch": 4.0125, "grad_norm": 12.4375, "learning_rate": 1.9994623775115587e-05, "loss": 1.6211, "step": 1605 }, { "epoch": 4.015, "grad_norm": 6.96875, "learning_rate": 1.999459719547392e-05, "loss": 1.6476, "step": 1606 }, { "epoch": 4.0175, "grad_norm": 7.5625, "learning_rate": 1.999457055030807e-05, "loss": 1.7587, "step": 1607 }, { "epoch": 4.02, "grad_norm": 5.40625, "learning_rate": 1.999454383961822e-05, "loss": 1.7415, "step": 1608 }, { "epoch": 4.0225, "grad_norm": 8.0625, "learning_rate": 1.9994517063404542e-05, "loss": 1.8577, "step": 1609 }, { "epoch": 4.025, "grad_norm": 6.4375, "learning_rate": 1.9994490221667207e-05, "loss": 1.795, "step": 1610 }, { "epoch": 4.0275, "grad_norm": 7.5, "learning_rate": 1.9994463314406394e-05, "loss": 1.84, "step": 1611 }, { "epoch": 4.03, "grad_norm": 6.28125, "learning_rate": 1.999443634162228e-05, "loss": 1.7784, "step": 1612 }, { "epoch": 4.0325, "grad_norm": 5.9375, "learning_rate": 1.9994409303315042e-05, "loss": 1.9264, "step": 1613 }, { "epoch": 4.035, "grad_norm": 7.90625, "learning_rate": 1.9994382199484855e-05, "loss": 1.6118, "step": 1614 }, { "epoch": 4.0375, "grad_norm": 7.59375, "learning_rate": 1.99943550301319e-05, "loss": 2.0118, "step": 1615 }, { "epoch": 4.04, "grad_norm": 5.59375, "learning_rate": 1.999432779525635e-05, "loss": 1.7644, "step": 1616 }, { "epoch": 4.0425, "grad_norm": 5.53125, "learning_rate": 1.9994300494858392e-05, "loss": 1.7699, "step": 1617 }, { "epoch": 4.045, "grad_norm": 7.84375, "learning_rate": 1.9994273128938197e-05, "loss": 1.993, "step": 1618 }, { "epoch": 4.0475, "grad_norm": 4.9375, "learning_rate": 1.9994245697495947e-05, "loss": 1.6812, "step": 1619 }, { "epoch": 4.05, "grad_norm": 17.125, "learning_rate": 1.9994218200531823e-05, "loss": 1.6043, "step": 1620 }, { "epoch": 4.0525, "grad_norm": 5.75, "learning_rate": 1.9994190638046004e-05, "loss": 1.4684, "step": 1621 }, { "epoch": 4.055, "grad_norm": 8.125, "learning_rate": 1.9994163010038673e-05, "loss": 1.8008, "step": 1622 }, { "epoch": 4.0575, "grad_norm": 4.03125, "learning_rate": 1.999413531651001e-05, "loss": 1.6734, "step": 1623 }, { "epoch": 4.06, "grad_norm": 7.25, "learning_rate": 1.999410755746019e-05, "loss": 1.7008, "step": 1624 }, { "epoch": 4.0625, "grad_norm": 5.875, "learning_rate": 1.9994079732889404e-05, "loss": 1.6428, "step": 1625 }, { "epoch": 4.065, "grad_norm": 4.3125, "learning_rate": 1.999405184279783e-05, "loss": 1.7013, "step": 1626 }, { "epoch": 4.0675, "grad_norm": 10.1875, "learning_rate": 1.9994023887185655e-05, "loss": 1.7765, "step": 1627 }, { "epoch": 4.07, "grad_norm": 6.625, "learning_rate": 1.9993995866053057e-05, "loss": 1.7012, "step": 1628 }, { "epoch": 4.0725, "grad_norm": 6.0625, "learning_rate": 1.9993967779400224e-05, "loss": 1.6963, "step": 1629 }, { "epoch": 4.075, "grad_norm": 7.65625, "learning_rate": 1.9993939627227337e-05, "loss": 1.8181, "step": 1630 }, { "epoch": 4.0775, "grad_norm": 7.34375, "learning_rate": 1.999391140953458e-05, "loss": 1.8265, "step": 1631 }, { "epoch": 4.08, "grad_norm": 4.25, "learning_rate": 1.9993883126322142e-05, "loss": 1.831, "step": 1632 }, { "epoch": 4.0825, "grad_norm": 7.03125, "learning_rate": 1.9993854777590207e-05, "loss": 1.9102, "step": 1633 }, { "epoch": 4.085, "grad_norm": 6.15625, "learning_rate": 1.999382636333896e-05, "loss": 1.7575, "step": 1634 }, { "epoch": 4.0875, "grad_norm": 5.78125, "learning_rate": 1.9993797883568585e-05, "loss": 1.7777, "step": 1635 }, { "epoch": 4.09, "grad_norm": 8.0, "learning_rate": 1.999376933827927e-05, "loss": 1.8184, "step": 1636 }, { "epoch": 4.0925, "grad_norm": 7.53125, "learning_rate": 1.999374072747121e-05, "loss": 1.8129, "step": 1637 }, { "epoch": 4.095, "grad_norm": 5.03125, "learning_rate": 1.999371205114458e-05, "loss": 1.7024, "step": 1638 }, { "epoch": 4.0975, "grad_norm": 7.5625, "learning_rate": 1.9993683309299574e-05, "loss": 1.8076, "step": 1639 }, { "epoch": 4.1, "grad_norm": 8.6875, "learning_rate": 1.999365450193638e-05, "loss": 1.6645, "step": 1640 }, { "epoch": 4.1025, "grad_norm": 6.9375, "learning_rate": 1.9993625629055188e-05, "loss": 1.9107, "step": 1641 }, { "epoch": 4.105, "grad_norm": 5.0, "learning_rate": 1.9993596690656185e-05, "loss": 1.8527, "step": 1642 }, { "epoch": 4.1075, "grad_norm": 4.40625, "learning_rate": 1.9993567686739562e-05, "loss": 1.6462, "step": 1643 }, { "epoch": 4.11, "grad_norm": 4.75, "learning_rate": 1.999353861730551e-05, "loss": 1.7245, "step": 1644 }, { "epoch": 4.1125, "grad_norm": 6.4375, "learning_rate": 1.999350948235422e-05, "loss": 1.6936, "step": 1645 }, { "epoch": 4.115, "grad_norm": 5.9375, "learning_rate": 1.999348028188588e-05, "loss": 1.8291, "step": 1646 }, { "epoch": 4.1175, "grad_norm": 7.09375, "learning_rate": 1.9993451015900686e-05, "loss": 1.6397, "step": 1647 }, { "epoch": 4.12, "grad_norm": 4.8125, "learning_rate": 1.9993421684398825e-05, "loss": 1.6451, "step": 1648 }, { "epoch": 4.1225, "grad_norm": 9.5625, "learning_rate": 1.999339228738049e-05, "loss": 1.7937, "step": 1649 }, { "epoch": 4.125, "grad_norm": 6.5625, "learning_rate": 1.9993362824845878e-05, "loss": 1.8048, "step": 1650 }, { "epoch": 4.1275, "grad_norm": 4.3125, "learning_rate": 1.9993333296795175e-05, "loss": 1.5757, "step": 1651 }, { "epoch": 4.13, "grad_norm": 5.875, "learning_rate": 1.9993303703228585e-05, "loss": 1.8085, "step": 1652 }, { "epoch": 4.1325, "grad_norm": 5.1875, "learning_rate": 1.9993274044146293e-05, "loss": 1.7596, "step": 1653 }, { "epoch": 4.135, "grad_norm": 5.09375, "learning_rate": 1.9993244319548493e-05, "loss": 2.0642, "step": 1654 }, { "epoch": 4.1375, "grad_norm": 6.125, "learning_rate": 1.9993214529435384e-05, "loss": 1.6592, "step": 1655 }, { "epoch": 4.14, "grad_norm": 5.25, "learning_rate": 1.999318467380716e-05, "loss": 1.9846, "step": 1656 }, { "epoch": 4.1425, "grad_norm": 4.84375, "learning_rate": 1.9993154752664022e-05, "loss": 1.7948, "step": 1657 }, { "epoch": 4.145, "grad_norm": 5.0625, "learning_rate": 1.9993124766006157e-05, "loss": 1.7191, "step": 1658 }, { "epoch": 4.1475, "grad_norm": 8.75, "learning_rate": 1.999309471383377e-05, "loss": 1.7247, "step": 1659 }, { "epoch": 4.15, "grad_norm": 4.875, "learning_rate": 1.999306459614705e-05, "loss": 1.7143, "step": 1660 }, { "epoch": 4.1525, "grad_norm": 4.40625, "learning_rate": 1.99930344129462e-05, "loss": 1.5448, "step": 1661 }, { "epoch": 4.155, "grad_norm": 15.25, "learning_rate": 1.999300416423142e-05, "loss": 1.647, "step": 1662 }, { "epoch": 4.1575, "grad_norm": 6.34375, "learning_rate": 1.9992973850002898e-05, "loss": 1.7402, "step": 1663 }, { "epoch": 4.16, "grad_norm": 7.40625, "learning_rate": 1.9992943470260845e-05, "loss": 1.9499, "step": 1664 }, { "epoch": 4.1625, "grad_norm": 6.875, "learning_rate": 1.9992913025005448e-05, "loss": 1.7469, "step": 1665 }, { "epoch": 4.165, "grad_norm": 7.34375, "learning_rate": 1.9992882514236918e-05, "loss": 1.641, "step": 1666 }, { "epoch": 4.1675, "grad_norm": 4.96875, "learning_rate": 1.9992851937955448e-05, "loss": 1.6694, "step": 1667 }, { "epoch": 4.17, "grad_norm": 6.5, "learning_rate": 1.999282129616124e-05, "loss": 1.7415, "step": 1668 }, { "epoch": 4.1725, "grad_norm": 12.125, "learning_rate": 1.99927905888545e-05, "loss": 1.8655, "step": 1669 }, { "epoch": 4.175, "grad_norm": 6.3125, "learning_rate": 1.999275981603542e-05, "loss": 1.9122, "step": 1670 }, { "epoch": 4.1775, "grad_norm": 5.90625, "learning_rate": 1.9992728977704208e-05, "loss": 1.8085, "step": 1671 }, { "epoch": 4.18, "grad_norm": 7.78125, "learning_rate": 1.9992698073861067e-05, "loss": 1.821, "step": 1672 }, { "epoch": 4.1825, "grad_norm": 6.46875, "learning_rate": 1.9992667104506192e-05, "loss": 1.6129, "step": 1673 }, { "epoch": 4.185, "grad_norm": 4.53125, "learning_rate": 1.9992636069639797e-05, "loss": 1.5462, "step": 1674 }, { "epoch": 4.1875, "grad_norm": 7.3125, "learning_rate": 1.9992604969262078e-05, "loss": 1.6177, "step": 1675 }, { "epoch": 4.19, "grad_norm": 6.1875, "learning_rate": 1.9992573803373242e-05, "loss": 1.6667, "step": 1676 }, { "epoch": 4.1925, "grad_norm": 5.46875, "learning_rate": 1.999254257197349e-05, "loss": 1.6425, "step": 1677 }, { "epoch": 4.195, "grad_norm": 5.71875, "learning_rate": 1.9992511275063025e-05, "loss": 1.7171, "step": 1678 }, { "epoch": 4.1975, "grad_norm": 5.3125, "learning_rate": 1.999247991264206e-05, "loss": 1.7065, "step": 1679 }, { "epoch": 4.2, "grad_norm": 8.3125, "learning_rate": 1.99924484847108e-05, "loss": 1.9919, "step": 1680 }, { "epoch": 4.2025, "grad_norm": 6.25, "learning_rate": 1.999241699126944e-05, "loss": 1.7144, "step": 1681 }, { "epoch": 4.205, "grad_norm": 4.34375, "learning_rate": 1.99923854323182e-05, "loss": 1.6712, "step": 1682 }, { "epoch": 4.2075, "grad_norm": 3.8125, "learning_rate": 1.999235380785728e-05, "loss": 1.6229, "step": 1683 }, { "epoch": 4.21, "grad_norm": 6.09375, "learning_rate": 1.9992322117886886e-05, "loss": 1.8437, "step": 1684 }, { "epoch": 4.2125, "grad_norm": 5.46875, "learning_rate": 1.9992290362407232e-05, "loss": 1.5479, "step": 1685 }, { "epoch": 4.215, "grad_norm": 6.03125, "learning_rate": 1.999225854141852e-05, "loss": 1.7029, "step": 1686 }, { "epoch": 4.2175, "grad_norm": 7.5625, "learning_rate": 1.9992226654920958e-05, "loss": 1.8283, "step": 1687 }, { "epoch": 4.22, "grad_norm": 9.125, "learning_rate": 1.9992194702914762e-05, "loss": 1.8701, "step": 1688 }, { "epoch": 4.2225, "grad_norm": 6.625, "learning_rate": 1.9992162685400136e-05, "loss": 1.6851, "step": 1689 }, { "epoch": 4.225, "grad_norm": 6.875, "learning_rate": 1.9992130602377292e-05, "loss": 1.7272, "step": 1690 }, { "epoch": 4.2275, "grad_norm": 6.9375, "learning_rate": 1.999209845384644e-05, "loss": 1.672, "step": 1691 }, { "epoch": 4.23, "grad_norm": 9.0, "learning_rate": 1.999206623980779e-05, "loss": 1.8337, "step": 1692 }, { "epoch": 4.2325, "grad_norm": 4.125, "learning_rate": 1.9992033960261548e-05, "loss": 1.796, "step": 1693 }, { "epoch": 4.235, "grad_norm": 5.40625, "learning_rate": 1.9992001615207938e-05, "loss": 1.6263, "step": 1694 }, { "epoch": 4.2375, "grad_norm": 6.53125, "learning_rate": 1.9991969204647164e-05, "loss": 1.8094, "step": 1695 }, { "epoch": 4.24, "grad_norm": 6.5625, "learning_rate": 1.9991936728579438e-05, "loss": 1.8338, "step": 1696 }, { "epoch": 4.2425, "grad_norm": 4.59375, "learning_rate": 1.9991904187004975e-05, "loss": 1.6902, "step": 1697 }, { "epoch": 4.245, "grad_norm": 6.3125, "learning_rate": 1.999187157992399e-05, "loss": 1.8425, "step": 1698 }, { "epoch": 4.2475, "grad_norm": 7.4375, "learning_rate": 1.9991838907336693e-05, "loss": 1.8375, "step": 1699 }, { "epoch": 4.25, "grad_norm": 10.5625, "learning_rate": 1.9991806169243302e-05, "loss": 1.8142, "step": 1700 }, { "epoch": 4.2525, "grad_norm": 7.34375, "learning_rate": 1.999177336564403e-05, "loss": 1.8007, "step": 1701 }, { "epoch": 4.255, "grad_norm": 5.09375, "learning_rate": 1.9991740496539086e-05, "loss": 1.7014, "step": 1702 }, { "epoch": 4.2575, "grad_norm": 5.9375, "learning_rate": 1.9991707561928697e-05, "loss": 1.7133, "step": 1703 }, { "epoch": 4.26, "grad_norm": 5.96875, "learning_rate": 1.999167456181307e-05, "loss": 1.8675, "step": 1704 }, { "epoch": 4.2625, "grad_norm": 5.34375, "learning_rate": 1.9991641496192428e-05, "loss": 1.6463, "step": 1705 }, { "epoch": 4.265, "grad_norm": 4.5625, "learning_rate": 1.999160836506698e-05, "loss": 1.7891, "step": 1706 }, { "epoch": 4.2675, "grad_norm": 5.96875, "learning_rate": 1.999157516843695e-05, "loss": 1.8437, "step": 1707 }, { "epoch": 4.27, "grad_norm": 5.90625, "learning_rate": 1.9991541906302552e-05, "loss": 1.5559, "step": 1708 }, { "epoch": 4.2725, "grad_norm": 6.46875, "learning_rate": 1.9991508578664004e-05, "loss": 1.8152, "step": 1709 }, { "epoch": 4.275, "grad_norm": 7.0625, "learning_rate": 1.999147518552153e-05, "loss": 1.772, "step": 1710 }, { "epoch": 4.2775, "grad_norm": 6.25, "learning_rate": 1.999144172687534e-05, "loss": 1.7693, "step": 1711 }, { "epoch": 4.28, "grad_norm": 6.25, "learning_rate": 1.999140820272566e-05, "loss": 1.9352, "step": 1712 }, { "epoch": 4.2825, "grad_norm": 4.28125, "learning_rate": 1.9991374613072705e-05, "loss": 1.6001, "step": 1713 }, { "epoch": 4.285, "grad_norm": 7.65625, "learning_rate": 1.9991340957916698e-05, "loss": 1.8564, "step": 1714 }, { "epoch": 4.2875, "grad_norm": 5.625, "learning_rate": 1.999130723725786e-05, "loss": 1.7919, "step": 1715 }, { "epoch": 4.29, "grad_norm": 5.84375, "learning_rate": 1.9991273451096414e-05, "loss": 1.7092, "step": 1716 }, { "epoch": 4.2925, "grad_norm": 5.15625, "learning_rate": 1.999123959943258e-05, "loss": 1.6699, "step": 1717 }, { "epoch": 4.295, "grad_norm": 5.21875, "learning_rate": 1.9991205682266573e-05, "loss": 1.5856, "step": 1718 }, { "epoch": 4.2975, "grad_norm": 7.25, "learning_rate": 1.999117169959863e-05, "loss": 1.9734, "step": 1719 }, { "epoch": 4.3, "grad_norm": 7.46875, "learning_rate": 1.9991137651428957e-05, "loss": 1.9337, "step": 1720 }, { "epoch": 4.3025, "grad_norm": 4.125, "learning_rate": 1.999110353775779e-05, "loss": 1.574, "step": 1721 }, { "epoch": 4.305, "grad_norm": 5.625, "learning_rate": 1.9991069358585348e-05, "loss": 1.8218, "step": 1722 }, { "epoch": 4.3075, "grad_norm": 3.671875, "learning_rate": 1.9991035113911855e-05, "loss": 1.6396, "step": 1723 }, { "epoch": 4.31, "grad_norm": 6.65625, "learning_rate": 1.999100080373753e-05, "loss": 1.5844, "step": 1724 }, { "epoch": 4.3125, "grad_norm": 7.59375, "learning_rate": 1.999096642806261e-05, "loss": 1.7716, "step": 1725 }, { "epoch": 4.315, "grad_norm": 5.15625, "learning_rate": 1.9990931986887313e-05, "loss": 1.7291, "step": 1726 }, { "epoch": 4.3175, "grad_norm": 7.46875, "learning_rate": 1.9990897480211865e-05, "loss": 1.8144, "step": 1727 }, { "epoch": 4.32, "grad_norm": 5.96875, "learning_rate": 1.9990862908036492e-05, "loss": 1.6891, "step": 1728 }, { "epoch": 4.3225, "grad_norm": 7.25, "learning_rate": 1.9990828270361423e-05, "loss": 1.6237, "step": 1729 }, { "epoch": 4.325, "grad_norm": 4.875, "learning_rate": 1.999079356718688e-05, "loss": 1.8167, "step": 1730 }, { "epoch": 4.3275, "grad_norm": 6.78125, "learning_rate": 1.9990758798513098e-05, "loss": 1.7383, "step": 1731 }, { "epoch": 4.33, "grad_norm": 5.9375, "learning_rate": 1.9990723964340303e-05, "loss": 1.7422, "step": 1732 }, { "epoch": 4.3325, "grad_norm": 5.9375, "learning_rate": 1.9990689064668717e-05, "loss": 1.6097, "step": 1733 }, { "epoch": 4.335, "grad_norm": 7.84375, "learning_rate": 1.9990654099498574e-05, "loss": 1.6901, "step": 1734 }, { "epoch": 4.3375, "grad_norm": 5.75, "learning_rate": 1.9990619068830104e-05, "loss": 1.8266, "step": 1735 }, { "epoch": 4.34, "grad_norm": 14.1875, "learning_rate": 1.9990583972663534e-05, "loss": 1.5578, "step": 1736 }, { "epoch": 4.3425, "grad_norm": 8.75, "learning_rate": 1.9990548810999097e-05, "loss": 1.5814, "step": 1737 }, { "epoch": 4.345, "grad_norm": 8.625, "learning_rate": 1.9990513583837022e-05, "loss": 1.623, "step": 1738 }, { "epoch": 4.3475, "grad_norm": 8.0625, "learning_rate": 1.9990478291177538e-05, "loss": 1.71, "step": 1739 }, { "epoch": 4.35, "grad_norm": 5.46875, "learning_rate": 1.999044293302088e-05, "loss": 1.8523, "step": 1740 }, { "epoch": 4.3525, "grad_norm": 6.40625, "learning_rate": 1.9990407509367274e-05, "loss": 1.8332, "step": 1741 }, { "epoch": 4.355, "grad_norm": 6.90625, "learning_rate": 1.999037202021696e-05, "loss": 1.7705, "step": 1742 }, { "epoch": 4.3575, "grad_norm": 5.21875, "learning_rate": 1.9990336465570164e-05, "loss": 1.654, "step": 1743 }, { "epoch": 4.36, "grad_norm": 4.28125, "learning_rate": 1.9990300845427123e-05, "loss": 1.4245, "step": 1744 }, { "epoch": 4.3625, "grad_norm": 6.625, "learning_rate": 1.999026515978807e-05, "loss": 1.7359, "step": 1745 }, { "epoch": 4.365, "grad_norm": 7.84375, "learning_rate": 1.999022940865324e-05, "loss": 1.7497, "step": 1746 }, { "epoch": 4.3675, "grad_norm": 5.5625, "learning_rate": 1.9990193592022865e-05, "loss": 1.5155, "step": 1747 }, { "epoch": 4.37, "grad_norm": 10.0625, "learning_rate": 1.999015770989718e-05, "loss": 1.8687, "step": 1748 }, { "epoch": 4.3725, "grad_norm": 8.5, "learning_rate": 1.999012176227642e-05, "loss": 1.6628, "step": 1749 }, { "epoch": 4.375, "grad_norm": 4.65625, "learning_rate": 1.999008574916082e-05, "loss": 1.4914, "step": 1750 }, { "epoch": 4.3775, "grad_norm": 5.59375, "learning_rate": 1.999004967055062e-05, "loss": 1.7017, "step": 1751 }, { "epoch": 4.38, "grad_norm": 6.875, "learning_rate": 1.9990013526446056e-05, "loss": 1.9149, "step": 1752 }, { "epoch": 4.3825, "grad_norm": 5.9375, "learning_rate": 1.998997731684736e-05, "loss": 1.7341, "step": 1753 }, { "epoch": 4.385, "grad_norm": 4.53125, "learning_rate": 1.9989941041754772e-05, "loss": 1.7279, "step": 1754 }, { "epoch": 4.3875, "grad_norm": 8.75, "learning_rate": 1.998990470116853e-05, "loss": 1.6953, "step": 1755 }, { "epoch": 4.39, "grad_norm": 7.21875, "learning_rate": 1.9989868295088876e-05, "loss": 1.8765, "step": 1756 }, { "epoch": 4.3925, "grad_norm": 8.25, "learning_rate": 1.9989831823516044e-05, "loss": 1.5683, "step": 1757 }, { "epoch": 4.395, "grad_norm": 5.875, "learning_rate": 1.9989795286450272e-05, "loss": 1.6711, "step": 1758 }, { "epoch": 4.3975, "grad_norm": 4.65625, "learning_rate": 1.9989758683891804e-05, "loss": 1.7548, "step": 1759 }, { "epoch": 4.4, "grad_norm": 4.78125, "learning_rate": 1.998972201584088e-05, "loss": 1.8355, "step": 1760 }, { "epoch": 4.4025, "grad_norm": 7.5, "learning_rate": 1.9989685282297736e-05, "loss": 1.7224, "step": 1761 }, { "epoch": 4.405, "grad_norm": 6.90625, "learning_rate": 1.9989648483262614e-05, "loss": 1.7033, "step": 1762 }, { "epoch": 4.4075, "grad_norm": 5.96875, "learning_rate": 1.9989611618735754e-05, "loss": 1.6738, "step": 1763 }, { "epoch": 4.41, "grad_norm": 4.59375, "learning_rate": 1.9989574688717405e-05, "loss": 1.8183, "step": 1764 }, { "epoch": 4.4125, "grad_norm": 5.71875, "learning_rate": 1.9989537693207803e-05, "loss": 1.6332, "step": 1765 }, { "epoch": 4.415, "grad_norm": 9.25, "learning_rate": 1.998950063220719e-05, "loss": 1.6599, "step": 1766 }, { "epoch": 4.4175, "grad_norm": 5.96875, "learning_rate": 1.9989463505715815e-05, "loss": 1.6919, "step": 1767 }, { "epoch": 4.42, "grad_norm": 4.84375, "learning_rate": 1.9989426313733912e-05, "loss": 1.8369, "step": 1768 }, { "epoch": 4.4225, "grad_norm": 5.375, "learning_rate": 1.998938905626173e-05, "loss": 1.6938, "step": 1769 }, { "epoch": 4.425, "grad_norm": 5.15625, "learning_rate": 1.9989351733299514e-05, "loss": 1.6806, "step": 1770 }, { "epoch": 4.4275, "grad_norm": 6.375, "learning_rate": 1.998931434484751e-05, "loss": 1.8868, "step": 1771 }, { "epoch": 4.43, "grad_norm": 6.5625, "learning_rate": 1.9989276890905957e-05, "loss": 1.6887, "step": 1772 }, { "epoch": 4.4325, "grad_norm": 4.03125, "learning_rate": 1.998923937147511e-05, "loss": 1.5476, "step": 1773 }, { "epoch": 4.435, "grad_norm": 5.75, "learning_rate": 1.9989201786555202e-05, "loss": 1.7342, "step": 1774 }, { "epoch": 4.4375, "grad_norm": 5.5625, "learning_rate": 1.9989164136146492e-05, "loss": 1.6151, "step": 1775 }, { "epoch": 4.44, "grad_norm": 4.9375, "learning_rate": 1.998912642024922e-05, "loss": 1.629, "step": 1776 }, { "epoch": 4.4425, "grad_norm": 5.21875, "learning_rate": 1.9989088638863635e-05, "loss": 1.7951, "step": 1777 }, { "epoch": 4.445, "grad_norm": 5.8125, "learning_rate": 1.9989050791989986e-05, "loss": 1.7679, "step": 1778 }, { "epoch": 4.4475, "grad_norm": 6.28125, "learning_rate": 1.998901287962852e-05, "loss": 1.6046, "step": 1779 }, { "epoch": 4.45, "grad_norm": 8.4375, "learning_rate": 1.9988974901779482e-05, "loss": 1.7931, "step": 1780 }, { "epoch": 4.4525, "grad_norm": 5.875, "learning_rate": 1.9988936858443128e-05, "loss": 1.6599, "step": 1781 }, { "epoch": 4.455, "grad_norm": 13.75, "learning_rate": 1.9988898749619702e-05, "loss": 1.6983, "step": 1782 }, { "epoch": 4.4575, "grad_norm": 6.125, "learning_rate": 1.9988860575309455e-05, "loss": 1.6511, "step": 1783 }, { "epoch": 4.46, "grad_norm": 8.0, "learning_rate": 1.9988822335512638e-05, "loss": 1.5615, "step": 1784 }, { "epoch": 4.4625, "grad_norm": 5.3125, "learning_rate": 1.9988784030229502e-05, "loss": 1.6263, "step": 1785 }, { "epoch": 4.465, "grad_norm": 6.40625, "learning_rate": 1.9988745659460297e-05, "loss": 1.707, "step": 1786 }, { "epoch": 4.4675, "grad_norm": 5.71875, "learning_rate": 1.9988707223205277e-05, "loss": 1.9336, "step": 1787 }, { "epoch": 4.47, "grad_norm": 7.5625, "learning_rate": 1.998866872146469e-05, "loss": 1.6179, "step": 1788 }, { "epoch": 4.4725, "grad_norm": 4.5, "learning_rate": 1.9988630154238793e-05, "loss": 1.6118, "step": 1789 }, { "epoch": 4.475, "grad_norm": 5.96875, "learning_rate": 1.9988591521527834e-05, "loss": 1.6182, "step": 1790 }, { "epoch": 4.4775, "grad_norm": 8.3125, "learning_rate": 1.998855282333207e-05, "loss": 1.7733, "step": 1791 }, { "epoch": 4.48, "grad_norm": 4.21875, "learning_rate": 1.998851405965175e-05, "loss": 1.8012, "step": 1792 }, { "epoch": 4.4825, "grad_norm": 6.21875, "learning_rate": 1.9988475230487137e-05, "loss": 1.6612, "step": 1793 }, { "epoch": 4.485, "grad_norm": 5.25, "learning_rate": 1.9988436335838475e-05, "loss": 1.7936, "step": 1794 }, { "epoch": 4.4875, "grad_norm": 5.75, "learning_rate": 1.998839737570603e-05, "loss": 1.9515, "step": 1795 }, { "epoch": 4.49, "grad_norm": 6.875, "learning_rate": 1.9988358350090045e-05, "loss": 2.0487, "step": 1796 }, { "epoch": 4.4925, "grad_norm": 5.8125, "learning_rate": 1.9988319258990788e-05, "loss": 1.6515, "step": 1797 }, { "epoch": 4.495, "grad_norm": 7.0625, "learning_rate": 1.9988280102408508e-05, "loss": 1.7926, "step": 1798 }, { "epoch": 4.4975, "grad_norm": 6.96875, "learning_rate": 1.9988240880343462e-05, "loss": 1.9384, "step": 1799 }, { "epoch": 4.5, "grad_norm": 6.53125, "learning_rate": 1.998820159279591e-05, "loss": 1.8389, "step": 1800 }, { "epoch": 4.5024999999999995, "grad_norm": 6.78125, "learning_rate": 1.9988162239766105e-05, "loss": 1.789, "step": 1801 }, { "epoch": 4.505, "grad_norm": 4.78125, "learning_rate": 1.9988122821254312e-05, "loss": 1.8203, "step": 1802 }, { "epoch": 4.5075, "grad_norm": 6.09375, "learning_rate": 1.9988083337260784e-05, "loss": 1.6777, "step": 1803 }, { "epoch": 4.51, "grad_norm": 8.5625, "learning_rate": 1.998804378778578e-05, "loss": 1.7382, "step": 1804 }, { "epoch": 4.5125, "grad_norm": 5.34375, "learning_rate": 1.998800417282956e-05, "loss": 1.6149, "step": 1805 }, { "epoch": 4.515, "grad_norm": 6.625, "learning_rate": 1.998796449239239e-05, "loss": 1.8611, "step": 1806 }, { "epoch": 4.5175, "grad_norm": 5.375, "learning_rate": 1.9987924746474518e-05, "loss": 1.6863, "step": 1807 }, { "epoch": 4.52, "grad_norm": 4.125, "learning_rate": 1.9987884935076213e-05, "loss": 1.4771, "step": 1808 }, { "epoch": 4.5225, "grad_norm": 5.28125, "learning_rate": 1.9987845058197735e-05, "loss": 1.7986, "step": 1809 }, { "epoch": 4.525, "grad_norm": 4.90625, "learning_rate": 1.9987805115839344e-05, "loss": 1.5022, "step": 1810 }, { "epoch": 4.5275, "grad_norm": 7.1875, "learning_rate": 1.99877651080013e-05, "loss": 1.7388, "step": 1811 }, { "epoch": 4.53, "grad_norm": 6.71875, "learning_rate": 1.9987725034683875e-05, "loss": 1.5908, "step": 1812 }, { "epoch": 4.5325, "grad_norm": 11.375, "learning_rate": 1.998768489588732e-05, "loss": 2.0186, "step": 1813 }, { "epoch": 4.535, "grad_norm": 5.65625, "learning_rate": 1.99876446916119e-05, "loss": 1.6375, "step": 1814 }, { "epoch": 4.5375, "grad_norm": 6.21875, "learning_rate": 1.9987604421857884e-05, "loss": 1.6396, "step": 1815 }, { "epoch": 4.54, "grad_norm": 6.40625, "learning_rate": 1.998756408662553e-05, "loss": 1.8631, "step": 1816 }, { "epoch": 4.5425, "grad_norm": 4.9375, "learning_rate": 1.9987523685915108e-05, "loss": 1.6553, "step": 1817 }, { "epoch": 4.545, "grad_norm": 6.90625, "learning_rate": 1.998748321972688e-05, "loss": 1.7447, "step": 1818 }, { "epoch": 4.5475, "grad_norm": 6.65625, "learning_rate": 1.9987442688061112e-05, "loss": 1.9006, "step": 1819 }, { "epoch": 4.55, "grad_norm": 7.625, "learning_rate": 1.998740209091807e-05, "loss": 1.7541, "step": 1820 }, { "epoch": 4.5525, "grad_norm": 9.8125, "learning_rate": 1.998736142829802e-05, "loss": 1.7049, "step": 1821 }, { "epoch": 4.555, "grad_norm": 6.71875, "learning_rate": 1.9987320700201225e-05, "loss": 1.7216, "step": 1822 }, { "epoch": 4.5575, "grad_norm": 6.65625, "learning_rate": 1.9987279906627956e-05, "loss": 1.7281, "step": 1823 }, { "epoch": 4.5600000000000005, "grad_norm": 7.03125, "learning_rate": 1.9987239047578482e-05, "loss": 1.6268, "step": 1824 }, { "epoch": 4.5625, "grad_norm": 6.625, "learning_rate": 1.9987198123053066e-05, "loss": 1.9653, "step": 1825 }, { "epoch": 4.5649999999999995, "grad_norm": 9.75, "learning_rate": 1.998715713305198e-05, "loss": 1.8606, "step": 1826 }, { "epoch": 4.5675, "grad_norm": 6.0625, "learning_rate": 1.9987116077575493e-05, "loss": 1.8516, "step": 1827 }, { "epoch": 4.57, "grad_norm": 8.6875, "learning_rate": 1.9987074956623872e-05, "loss": 1.7776, "step": 1828 }, { "epoch": 4.5725, "grad_norm": 8.75, "learning_rate": 1.9987033770197386e-05, "loss": 1.8993, "step": 1829 }, { "epoch": 4.575, "grad_norm": 14.0, "learning_rate": 1.9986992518296304e-05, "loss": 1.5792, "step": 1830 }, { "epoch": 4.5775, "grad_norm": 7.90625, "learning_rate": 1.9986951200920904e-05, "loss": 1.7186, "step": 1831 }, { "epoch": 4.58, "grad_norm": 6.25, "learning_rate": 1.998690981807145e-05, "loss": 1.8212, "step": 1832 }, { "epoch": 4.5825, "grad_norm": 6.75, "learning_rate": 1.998686836974821e-05, "loss": 1.7145, "step": 1833 }, { "epoch": 4.585, "grad_norm": 7.625, "learning_rate": 1.9986826855951467e-05, "loss": 1.6582, "step": 1834 }, { "epoch": 4.5875, "grad_norm": 9.125, "learning_rate": 1.9986785276681485e-05, "loss": 1.7654, "step": 1835 }, { "epoch": 4.59, "grad_norm": 6.84375, "learning_rate": 1.9986743631938536e-05, "loss": 1.8197, "step": 1836 }, { "epoch": 4.5925, "grad_norm": 5.8125, "learning_rate": 1.9986701921722895e-05, "loss": 1.6514, "step": 1837 }, { "epoch": 4.595, "grad_norm": 16.25, "learning_rate": 1.998666014603484e-05, "loss": 1.7014, "step": 1838 }, { "epoch": 4.5975, "grad_norm": 5.90625, "learning_rate": 1.9986618304874637e-05, "loss": 1.6608, "step": 1839 }, { "epoch": 4.6, "grad_norm": 5.5, "learning_rate": 1.9986576398242566e-05, "loss": 1.7695, "step": 1840 }, { "epoch": 4.6025, "grad_norm": 8.5, "learning_rate": 1.99865344261389e-05, "loss": 1.7803, "step": 1841 }, { "epoch": 4.605, "grad_norm": 6.5625, "learning_rate": 1.998649238856391e-05, "loss": 1.4631, "step": 1842 }, { "epoch": 4.6075, "grad_norm": 9.4375, "learning_rate": 1.998645028551788e-05, "loss": 1.7802, "step": 1843 }, { "epoch": 4.61, "grad_norm": 6.6875, "learning_rate": 1.9986408117001082e-05, "loss": 1.7446, "step": 1844 }, { "epoch": 4.6125, "grad_norm": 5.71875, "learning_rate": 1.998636588301379e-05, "loss": 1.7814, "step": 1845 }, { "epoch": 4.615, "grad_norm": 6.125, "learning_rate": 1.9986323583556283e-05, "loss": 1.5754, "step": 1846 }, { "epoch": 4.6175, "grad_norm": 5.375, "learning_rate": 1.998628121862884e-05, "loss": 1.4773, "step": 1847 }, { "epoch": 4.62, "grad_norm": 4.25, "learning_rate": 1.9986238788231733e-05, "loss": 1.5202, "step": 1848 }, { "epoch": 4.6225000000000005, "grad_norm": 10.5625, "learning_rate": 1.9986196292365247e-05, "loss": 1.7849, "step": 1849 }, { "epoch": 4.625, "grad_norm": 5.46875, "learning_rate": 1.9986153731029657e-05, "loss": 1.609, "step": 1850 }, { "epoch": 4.6274999999999995, "grad_norm": 5.0625, "learning_rate": 1.9986111104225244e-05, "loss": 1.7031, "step": 1851 }, { "epoch": 4.63, "grad_norm": 5.75, "learning_rate": 1.9986068411952283e-05, "loss": 1.7084, "step": 1852 }, { "epoch": 4.6325, "grad_norm": 5.59375, "learning_rate": 1.998602565421106e-05, "loss": 1.7137, "step": 1853 }, { "epoch": 4.635, "grad_norm": 6.25, "learning_rate": 1.9985982831001855e-05, "loss": 1.7226, "step": 1854 }, { "epoch": 4.6375, "grad_norm": 6.375, "learning_rate": 1.998593994232494e-05, "loss": 1.8972, "step": 1855 }, { "epoch": 4.64, "grad_norm": 6.9375, "learning_rate": 1.9985896988180607e-05, "loss": 1.7415, "step": 1856 }, { "epoch": 4.6425, "grad_norm": 5.21875, "learning_rate": 1.998585396856913e-05, "loss": 1.6035, "step": 1857 }, { "epoch": 4.645, "grad_norm": 5.59375, "learning_rate": 1.9985810883490795e-05, "loss": 1.6754, "step": 1858 }, { "epoch": 4.6475, "grad_norm": 5.75, "learning_rate": 1.9985767732945884e-05, "loss": 1.7813, "step": 1859 }, { "epoch": 4.65, "grad_norm": 5.03125, "learning_rate": 1.998572451693468e-05, "loss": 1.8112, "step": 1860 }, { "epoch": 4.6525, "grad_norm": 6.6875, "learning_rate": 1.9985681235457465e-05, "loss": 1.6376, "step": 1861 }, { "epoch": 4.655, "grad_norm": 8.1875, "learning_rate": 1.998563788851452e-05, "loss": 1.8701, "step": 1862 }, { "epoch": 4.6575, "grad_norm": 8.5, "learning_rate": 1.9985594476106138e-05, "loss": 1.7623, "step": 1863 }, { "epoch": 4.66, "grad_norm": 6.34375, "learning_rate": 1.9985550998232598e-05, "loss": 1.7222, "step": 1864 }, { "epoch": 4.6625, "grad_norm": 5.59375, "learning_rate": 1.9985507454894182e-05, "loss": 1.739, "step": 1865 }, { "epoch": 4.665, "grad_norm": 7.03125, "learning_rate": 1.998546384609118e-05, "loss": 1.4541, "step": 1866 }, { "epoch": 4.6675, "grad_norm": 5.71875, "learning_rate": 1.9985420171823877e-05, "loss": 1.573, "step": 1867 }, { "epoch": 4.67, "grad_norm": 4.8125, "learning_rate": 1.9985376432092558e-05, "loss": 1.6683, "step": 1868 }, { "epoch": 4.6725, "grad_norm": 6.125, "learning_rate": 1.998533262689751e-05, "loss": 1.596, "step": 1869 }, { "epoch": 4.675, "grad_norm": 4.96875, "learning_rate": 1.9985288756239022e-05, "loss": 1.5964, "step": 1870 }, { "epoch": 4.6775, "grad_norm": 5.78125, "learning_rate": 1.998524482011738e-05, "loss": 1.7531, "step": 1871 }, { "epoch": 4.68, "grad_norm": 6.25, "learning_rate": 1.9985200818532873e-05, "loss": 1.8204, "step": 1872 }, { "epoch": 4.6825, "grad_norm": 6.1875, "learning_rate": 1.9985156751485794e-05, "loss": 1.5371, "step": 1873 }, { "epoch": 4.6850000000000005, "grad_norm": 9.25, "learning_rate": 1.998511261897642e-05, "loss": 1.5911, "step": 1874 }, { "epoch": 4.6875, "grad_norm": 7.5, "learning_rate": 1.998506842100505e-05, "loss": 1.9063, "step": 1875 }, { "epoch": 4.6899999999999995, "grad_norm": 6.15625, "learning_rate": 1.9985024157571972e-05, "loss": 1.841, "step": 1876 }, { "epoch": 4.6925, "grad_norm": 6.4375, "learning_rate": 1.998497982867747e-05, "loss": 1.6285, "step": 1877 }, { "epoch": 4.695, "grad_norm": 7.03125, "learning_rate": 1.9984935434321847e-05, "loss": 1.9604, "step": 1878 }, { "epoch": 4.6975, "grad_norm": 6.75, "learning_rate": 1.9984890974505383e-05, "loss": 1.6578, "step": 1879 }, { "epoch": 4.7, "grad_norm": 4.59375, "learning_rate": 1.998484644922837e-05, "loss": 1.5279, "step": 1880 }, { "epoch": 4.7025, "grad_norm": 7.28125, "learning_rate": 1.998480185849111e-05, "loss": 1.9127, "step": 1881 }, { "epoch": 4.705, "grad_norm": 4.40625, "learning_rate": 1.998475720229388e-05, "loss": 1.5432, "step": 1882 }, { "epoch": 4.7075, "grad_norm": 6.21875, "learning_rate": 1.998471248063699e-05, "loss": 1.6271, "step": 1883 }, { "epoch": 4.71, "grad_norm": 8.4375, "learning_rate": 1.998466769352072e-05, "loss": 1.8162, "step": 1884 }, { "epoch": 4.7125, "grad_norm": 5.03125, "learning_rate": 1.998462284094537e-05, "loss": 1.8215, "step": 1885 }, { "epoch": 4.715, "grad_norm": 5.53125, "learning_rate": 1.998457792291123e-05, "loss": 1.7027, "step": 1886 }, { "epoch": 4.7175, "grad_norm": 7.125, "learning_rate": 1.9984532939418603e-05, "loss": 1.6585, "step": 1887 }, { "epoch": 4.72, "grad_norm": 5.4375, "learning_rate": 1.9984487890467773e-05, "loss": 1.6124, "step": 1888 }, { "epoch": 4.7225, "grad_norm": 6.6875, "learning_rate": 1.998444277605904e-05, "loss": 1.432, "step": 1889 }, { "epoch": 4.725, "grad_norm": 7.4375, "learning_rate": 1.9984397596192698e-05, "loss": 1.7605, "step": 1890 }, { "epoch": 4.7275, "grad_norm": 6.8125, "learning_rate": 1.9984352350869046e-05, "loss": 1.7098, "step": 1891 }, { "epoch": 4.73, "grad_norm": 8.0625, "learning_rate": 1.998430704008838e-05, "loss": 1.7349, "step": 1892 }, { "epoch": 4.7325, "grad_norm": 9.4375, "learning_rate": 1.9984261663850998e-05, "loss": 1.8596, "step": 1893 }, { "epoch": 4.735, "grad_norm": 7.46875, "learning_rate": 1.9984216222157197e-05, "loss": 1.7394, "step": 1894 }, { "epoch": 4.7375, "grad_norm": 5.34375, "learning_rate": 1.998417071500727e-05, "loss": 1.911, "step": 1895 }, { "epoch": 4.74, "grad_norm": 7.03125, "learning_rate": 1.998412514240152e-05, "loss": 1.6398, "step": 1896 }, { "epoch": 4.7425, "grad_norm": 6.375, "learning_rate": 1.998407950434025e-05, "loss": 1.4815, "step": 1897 }, { "epoch": 4.745, "grad_norm": 6.5625, "learning_rate": 1.9984033800823747e-05, "loss": 1.7589, "step": 1898 }, { "epoch": 4.7475000000000005, "grad_norm": 6.21875, "learning_rate": 1.9983988031852323e-05, "loss": 1.9458, "step": 1899 }, { "epoch": 4.75, "grad_norm": 15.8125, "learning_rate": 1.9983942197426272e-05, "loss": 1.6404, "step": 1900 }, { "epoch": 4.7524999999999995, "grad_norm": 4.5, "learning_rate": 1.9983896297545896e-05, "loss": 1.6219, "step": 1901 }, { "epoch": 4.755, "grad_norm": 5.25, "learning_rate": 1.9983850332211496e-05, "loss": 1.6674, "step": 1902 }, { "epoch": 4.7575, "grad_norm": 4.75, "learning_rate": 1.9983804301423367e-05, "loss": 1.7695, "step": 1903 }, { "epoch": 4.76, "grad_norm": 6.40625, "learning_rate": 1.9983758205181824e-05, "loss": 1.7501, "step": 1904 }, { "epoch": 4.7625, "grad_norm": 5.1875, "learning_rate": 1.9983712043487155e-05, "loss": 1.6146, "step": 1905 }, { "epoch": 4.765, "grad_norm": 10.125, "learning_rate": 1.9983665816339672e-05, "loss": 1.6007, "step": 1906 }, { "epoch": 4.7675, "grad_norm": 16.375, "learning_rate": 1.998361952373968e-05, "loss": 1.7449, "step": 1907 }, { "epoch": 4.77, "grad_norm": 6.6875, "learning_rate": 1.9983573165687473e-05, "loss": 1.6626, "step": 1908 }, { "epoch": 4.7725, "grad_norm": 5.0625, "learning_rate": 1.998352674218336e-05, "loss": 1.821, "step": 1909 }, { "epoch": 4.775, "grad_norm": 6.1875, "learning_rate": 1.9983480253227646e-05, "loss": 1.7159, "step": 1910 }, { "epoch": 4.7775, "grad_norm": 6.1875, "learning_rate": 1.9983433698820636e-05, "loss": 1.4693, "step": 1911 }, { "epoch": 4.78, "grad_norm": 5.71875, "learning_rate": 1.9983387078962634e-05, "loss": 1.7006, "step": 1912 }, { "epoch": 4.7825, "grad_norm": 5.625, "learning_rate": 1.9983340393653946e-05, "loss": 1.6962, "step": 1913 }, { "epoch": 4.785, "grad_norm": 5.96875, "learning_rate": 1.9983293642894876e-05, "loss": 1.7722, "step": 1914 }, { "epoch": 4.7875, "grad_norm": 5.34375, "learning_rate": 1.9983246826685733e-05, "loss": 1.5653, "step": 1915 }, { "epoch": 4.79, "grad_norm": 7.34375, "learning_rate": 1.9983199945026822e-05, "loss": 1.7452, "step": 1916 }, { "epoch": 4.7925, "grad_norm": 6.65625, "learning_rate": 1.9983152997918457e-05, "loss": 1.4682, "step": 1917 }, { "epoch": 4.795, "grad_norm": 11.75, "learning_rate": 1.9983105985360934e-05, "loss": 1.7328, "step": 1918 }, { "epoch": 4.7975, "grad_norm": 6.15625, "learning_rate": 1.998305890735457e-05, "loss": 1.6981, "step": 1919 }, { "epoch": 4.8, "grad_norm": 6.375, "learning_rate": 1.9983011763899674e-05, "loss": 1.8135, "step": 1920 }, { "epoch": 4.8025, "grad_norm": 8.75, "learning_rate": 1.998296455499655e-05, "loss": 1.8197, "step": 1921 }, { "epoch": 4.805, "grad_norm": 5.84375, "learning_rate": 1.998291728064551e-05, "loss": 1.6939, "step": 1922 }, { "epoch": 4.8075, "grad_norm": 6.6875, "learning_rate": 1.9982869940846864e-05, "loss": 1.9775, "step": 1923 }, { "epoch": 4.8100000000000005, "grad_norm": 9.5625, "learning_rate": 1.998282253560092e-05, "loss": 1.6647, "step": 1924 }, { "epoch": 4.8125, "grad_norm": 8.4375, "learning_rate": 1.9982775064907995e-05, "loss": 2.0888, "step": 1925 }, { "epoch": 4.8149999999999995, "grad_norm": 6.4375, "learning_rate": 1.9982727528768393e-05, "loss": 1.826, "step": 1926 }, { "epoch": 4.8175, "grad_norm": 6.40625, "learning_rate": 1.998267992718243e-05, "loss": 1.7523, "step": 1927 }, { "epoch": 4.82, "grad_norm": 9.375, "learning_rate": 1.9982632260150417e-05, "loss": 1.7044, "step": 1928 }, { "epoch": 4.8225, "grad_norm": 5.96875, "learning_rate": 1.9982584527672666e-05, "loss": 1.6015, "step": 1929 }, { "epoch": 4.825, "grad_norm": 8.125, "learning_rate": 1.998253672974949e-05, "loss": 1.6029, "step": 1930 }, { "epoch": 4.8275, "grad_norm": 5.1875, "learning_rate": 1.9982488866381204e-05, "loss": 1.8532, "step": 1931 }, { "epoch": 4.83, "grad_norm": 7.09375, "learning_rate": 1.9982440937568118e-05, "loss": 1.6994, "step": 1932 }, { "epoch": 4.8325, "grad_norm": 6.40625, "learning_rate": 1.9982392943310553e-05, "loss": 1.7026, "step": 1933 }, { "epoch": 4.835, "grad_norm": 5.0625, "learning_rate": 1.9982344883608816e-05, "loss": 1.705, "step": 1934 }, { "epoch": 4.8375, "grad_norm": 5.78125, "learning_rate": 1.9982296758463227e-05, "loss": 1.6333, "step": 1935 }, { "epoch": 4.84, "grad_norm": 6.28125, "learning_rate": 1.9982248567874098e-05, "loss": 1.6506, "step": 1936 }, { "epoch": 4.8425, "grad_norm": 5.1875, "learning_rate": 1.998220031184175e-05, "loss": 1.6492, "step": 1937 }, { "epoch": 4.845, "grad_norm": 4.5625, "learning_rate": 1.9982151990366493e-05, "loss": 1.5804, "step": 1938 }, { "epoch": 4.8475, "grad_norm": 6.78125, "learning_rate": 1.9982103603448645e-05, "loss": 1.5592, "step": 1939 }, { "epoch": 4.85, "grad_norm": 7.28125, "learning_rate": 1.998205515108853e-05, "loss": 1.7517, "step": 1940 }, { "epoch": 4.8525, "grad_norm": 6.03125, "learning_rate": 1.9982006633286462e-05, "loss": 1.8638, "step": 1941 }, { "epoch": 4.855, "grad_norm": 7.0625, "learning_rate": 1.9981958050042752e-05, "loss": 1.842, "step": 1942 }, { "epoch": 4.8575, "grad_norm": 7.53125, "learning_rate": 1.998190940135773e-05, "loss": 1.4984, "step": 1943 }, { "epoch": 4.86, "grad_norm": 6.71875, "learning_rate": 1.9981860687231708e-05, "loss": 1.6231, "step": 1944 }, { "epoch": 4.8625, "grad_norm": 6.34375, "learning_rate": 1.9981811907665005e-05, "loss": 1.7426, "step": 1945 }, { "epoch": 4.865, "grad_norm": 7.25, "learning_rate": 1.9981763062657942e-05, "loss": 1.6942, "step": 1946 }, { "epoch": 4.8675, "grad_norm": 7.03125, "learning_rate": 1.9981714152210842e-05, "loss": 1.8867, "step": 1947 }, { "epoch": 4.87, "grad_norm": 6.3125, "learning_rate": 1.9981665176324023e-05, "loss": 1.5244, "step": 1948 }, { "epoch": 4.8725000000000005, "grad_norm": 7.71875, "learning_rate": 1.9981616134997806e-05, "loss": 1.78, "step": 1949 }, { "epoch": 4.875, "grad_norm": 4.96875, "learning_rate": 1.9981567028232514e-05, "loss": 1.5652, "step": 1950 }, { "epoch": 4.8774999999999995, "grad_norm": 7.8125, "learning_rate": 1.9981517856028466e-05, "loss": 1.8166, "step": 1951 }, { "epoch": 4.88, "grad_norm": 5.5625, "learning_rate": 1.998146861838599e-05, "loss": 1.6123, "step": 1952 }, { "epoch": 4.8825, "grad_norm": 4.3125, "learning_rate": 1.9981419315305402e-05, "loss": 1.5402, "step": 1953 }, { "epoch": 4.885, "grad_norm": 9.0, "learning_rate": 1.9981369946787028e-05, "loss": 1.7077, "step": 1954 }, { "epoch": 4.8875, "grad_norm": 7.40625, "learning_rate": 1.9981320512831195e-05, "loss": 1.7829, "step": 1955 }, { "epoch": 4.89, "grad_norm": 5.34375, "learning_rate": 1.998127101343822e-05, "loss": 1.7385, "step": 1956 }, { "epoch": 4.8925, "grad_norm": 14.0, "learning_rate": 1.9981221448608437e-05, "loss": 1.8532, "step": 1957 }, { "epoch": 4.895, "grad_norm": 6.3125, "learning_rate": 1.998117181834216e-05, "loss": 1.6887, "step": 1958 }, { "epoch": 4.8975, "grad_norm": 5.96875, "learning_rate": 1.9981122122639726e-05, "loss": 1.5311, "step": 1959 }, { "epoch": 4.9, "grad_norm": 5.8125, "learning_rate": 1.998107236150145e-05, "loss": 1.8335, "step": 1960 }, { "epoch": 4.9025, "grad_norm": 7.59375, "learning_rate": 1.9981022534927667e-05, "loss": 1.662, "step": 1961 }, { "epoch": 4.905, "grad_norm": 5.59375, "learning_rate": 1.9980972642918696e-05, "loss": 1.6247, "step": 1962 }, { "epoch": 4.9075, "grad_norm": 5.375, "learning_rate": 1.9980922685474867e-05, "loss": 1.7162, "step": 1963 }, { "epoch": 4.91, "grad_norm": 7.5625, "learning_rate": 1.9980872662596514e-05, "loss": 1.8946, "step": 1964 }, { "epoch": 4.9125, "grad_norm": 6.53125, "learning_rate": 1.9980822574283954e-05, "loss": 1.6051, "step": 1965 }, { "epoch": 4.915, "grad_norm": 5.375, "learning_rate": 1.9980772420537523e-05, "loss": 1.6407, "step": 1966 }, { "epoch": 4.9175, "grad_norm": 5.0625, "learning_rate": 1.9980722201357547e-05, "loss": 1.7221, "step": 1967 }, { "epoch": 4.92, "grad_norm": 5.0625, "learning_rate": 1.9980671916744356e-05, "loss": 1.7622, "step": 1968 }, { "epoch": 4.9225, "grad_norm": 6.71875, "learning_rate": 1.9980621566698276e-05, "loss": 1.715, "step": 1969 }, { "epoch": 4.925, "grad_norm": 7.6875, "learning_rate": 1.9980571151219644e-05, "loss": 1.6411, "step": 1970 }, { "epoch": 4.9275, "grad_norm": 5.875, "learning_rate": 1.9980520670308784e-05, "loss": 1.7584, "step": 1971 }, { "epoch": 4.93, "grad_norm": 7.8125, "learning_rate": 1.998047012396603e-05, "loss": 1.7881, "step": 1972 }, { "epoch": 4.9325, "grad_norm": 11.4375, "learning_rate": 1.9980419512191717e-05, "loss": 1.8361, "step": 1973 }, { "epoch": 4.9350000000000005, "grad_norm": 5.59375, "learning_rate": 1.9980368834986172e-05, "loss": 1.6856, "step": 1974 }, { "epoch": 4.9375, "grad_norm": 9.1875, "learning_rate": 1.9980318092349727e-05, "loss": 1.7112, "step": 1975 }, { "epoch": 4.9399999999999995, "grad_norm": 7.4375, "learning_rate": 1.9980267284282718e-05, "loss": 1.7345, "step": 1976 }, { "epoch": 4.9425, "grad_norm": 9.5625, "learning_rate": 1.9980216410785473e-05, "loss": 1.8209, "step": 1977 }, { "epoch": 4.945, "grad_norm": 6.59375, "learning_rate": 1.998016547185833e-05, "loss": 1.6301, "step": 1978 }, { "epoch": 4.9475, "grad_norm": 5.25, "learning_rate": 1.9980114467501624e-05, "loss": 1.5407, "step": 1979 }, { "epoch": 4.95, "grad_norm": 6.625, "learning_rate": 1.9980063397715685e-05, "loss": 1.745, "step": 1980 }, { "epoch": 4.9525, "grad_norm": 6.125, "learning_rate": 1.9980012262500848e-05, "loss": 1.7561, "step": 1981 }, { "epoch": 4.955, "grad_norm": 4.96875, "learning_rate": 1.9979961061857453e-05, "loss": 1.7534, "step": 1982 }, { "epoch": 4.9575, "grad_norm": 6.46875, "learning_rate": 1.997990979578583e-05, "loss": 1.7283, "step": 1983 }, { "epoch": 4.96, "grad_norm": 6.65625, "learning_rate": 1.9979858464286317e-05, "loss": 1.662, "step": 1984 }, { "epoch": 4.9625, "grad_norm": 4.65625, "learning_rate": 1.9979807067359252e-05, "loss": 1.5507, "step": 1985 }, { "epoch": 4.965, "grad_norm": 6.3125, "learning_rate": 1.9979755605004973e-05, "loss": 1.6956, "step": 1986 }, { "epoch": 4.9675, "grad_norm": 4.875, "learning_rate": 1.9979704077223816e-05, "loss": 1.55, "step": 1987 }, { "epoch": 4.97, "grad_norm": 4.8125, "learning_rate": 1.9979652484016117e-05, "loss": 1.6655, "step": 1988 }, { "epoch": 4.9725, "grad_norm": 7.09375, "learning_rate": 1.9979600825382217e-05, "loss": 1.5711, "step": 1989 }, { "epoch": 4.975, "grad_norm": 7.0625, "learning_rate": 1.9979549101322452e-05, "loss": 1.7093, "step": 1990 }, { "epoch": 4.9775, "grad_norm": 6.3125, "learning_rate": 1.997949731183716e-05, "loss": 1.6561, "step": 1991 }, { "epoch": 4.98, "grad_norm": 8.5625, "learning_rate": 1.997944545692669e-05, "loss": 1.7395, "step": 1992 }, { "epoch": 4.9825, "grad_norm": 6.78125, "learning_rate": 1.997939353659137e-05, "loss": 1.6978, "step": 1993 }, { "epoch": 4.985, "grad_norm": 7.15625, "learning_rate": 1.9979341550831542e-05, "loss": 1.6291, "step": 1994 }, { "epoch": 4.9875, "grad_norm": 5.9375, "learning_rate": 1.9979289499647555e-05, "loss": 1.5289, "step": 1995 }, { "epoch": 4.99, "grad_norm": 4.9375, "learning_rate": 1.9979237383039745e-05, "loss": 1.5864, "step": 1996 }, { "epoch": 4.9925, "grad_norm": 6.875, "learning_rate": 1.9979185201008456e-05, "loss": 1.8068, "step": 1997 }, { "epoch": 4.995, "grad_norm": 5.25, "learning_rate": 1.9979132953554026e-05, "loss": 1.5882, "step": 1998 }, { "epoch": 4.9975000000000005, "grad_norm": 13.0625, "learning_rate": 1.9979080640676797e-05, "loss": 1.9283, "step": 1999 }, { "epoch": 5.0, "grad_norm": 6.625, "learning_rate": 1.997902826237712e-05, "loss": 1.5594, "step": 2000 }, { "epoch": 5.0025, "grad_norm": 6.5, "learning_rate": 1.997897581865533e-05, "loss": 1.7257, "step": 2001 }, { "epoch": 5.005, "grad_norm": 5.65625, "learning_rate": 1.9978923309511774e-05, "loss": 1.6539, "step": 2002 }, { "epoch": 5.0075, "grad_norm": 4.25, "learning_rate": 1.9978870734946792e-05, "loss": 1.7157, "step": 2003 }, { "epoch": 5.01, "grad_norm": 6.59375, "learning_rate": 1.997881809496074e-05, "loss": 1.6588, "step": 2004 }, { "epoch": 5.0125, "grad_norm": 5.71875, "learning_rate": 1.9978765389553953e-05, "loss": 1.7409, "step": 2005 }, { "epoch": 5.015, "grad_norm": 5.6875, "learning_rate": 1.997871261872678e-05, "loss": 1.7003, "step": 2006 }, { "epoch": 5.0175, "grad_norm": 4.84375, "learning_rate": 1.9978659782479567e-05, "loss": 1.6778, "step": 2007 }, { "epoch": 5.02, "grad_norm": 7.09375, "learning_rate": 1.9978606880812657e-05, "loss": 1.773, "step": 2008 }, { "epoch": 5.0225, "grad_norm": 4.5, "learning_rate": 1.99785539137264e-05, "loss": 1.371, "step": 2009 }, { "epoch": 5.025, "grad_norm": 9.8125, "learning_rate": 1.9978500881221144e-05, "loss": 1.7816, "step": 2010 }, { "epoch": 5.0275, "grad_norm": 7.84375, "learning_rate": 1.9978447783297235e-05, "loss": 1.7703, "step": 2011 }, { "epoch": 5.03, "grad_norm": 6.6875, "learning_rate": 1.9978394619955024e-05, "loss": 1.5581, "step": 2012 }, { "epoch": 5.0325, "grad_norm": 5.40625, "learning_rate": 1.9978341391194855e-05, "loss": 1.7605, "step": 2013 }, { "epoch": 5.035, "grad_norm": 8.4375, "learning_rate": 1.997828809701708e-05, "loss": 1.5119, "step": 2014 }, { "epoch": 5.0375, "grad_norm": 5.125, "learning_rate": 1.9978234737422047e-05, "loss": 1.5066, "step": 2015 }, { "epoch": 5.04, "grad_norm": 5.15625, "learning_rate": 1.9978181312410104e-05, "loss": 1.6073, "step": 2016 }, { "epoch": 5.0425, "grad_norm": 7.03125, "learning_rate": 1.9978127821981607e-05, "loss": 1.8618, "step": 2017 }, { "epoch": 5.045, "grad_norm": 6.3125, "learning_rate": 1.99780742661369e-05, "loss": 1.6815, "step": 2018 }, { "epoch": 5.0475, "grad_norm": 10.0, "learning_rate": 1.9978020644876343e-05, "loss": 1.9178, "step": 2019 }, { "epoch": 5.05, "grad_norm": 4.3125, "learning_rate": 1.9977966958200276e-05, "loss": 1.6154, "step": 2020 }, { "epoch": 5.0525, "grad_norm": 5.5, "learning_rate": 1.9977913206109062e-05, "loss": 1.7428, "step": 2021 }, { "epoch": 5.055, "grad_norm": 5.09375, "learning_rate": 1.9977859388603048e-05, "loss": 1.8399, "step": 2022 }, { "epoch": 5.0575, "grad_norm": 4.21875, "learning_rate": 1.9977805505682584e-05, "loss": 1.7634, "step": 2023 }, { "epoch": 5.06, "grad_norm": 5.4375, "learning_rate": 1.9977751557348028e-05, "loss": 1.6825, "step": 2024 }, { "epoch": 5.0625, "grad_norm": 4.96875, "learning_rate": 1.9977697543599727e-05, "loss": 1.8966, "step": 2025 }, { "epoch": 5.065, "grad_norm": 7.0625, "learning_rate": 1.9977643464438046e-05, "loss": 1.5552, "step": 2026 }, { "epoch": 5.0675, "grad_norm": 9.0, "learning_rate": 1.9977589319863334e-05, "loss": 1.8063, "step": 2027 }, { "epoch": 5.07, "grad_norm": 6.71875, "learning_rate": 1.9977535109875944e-05, "loss": 1.7327, "step": 2028 }, { "epoch": 5.0725, "grad_norm": 6.4375, "learning_rate": 1.9977480834476234e-05, "loss": 1.8906, "step": 2029 }, { "epoch": 5.075, "grad_norm": 6.90625, "learning_rate": 1.9977426493664555e-05, "loss": 1.6253, "step": 2030 }, { "epoch": 5.0775, "grad_norm": 9.625, "learning_rate": 1.9977372087441274e-05, "loss": 1.7295, "step": 2031 }, { "epoch": 5.08, "grad_norm": 5.0625, "learning_rate": 1.9977317615806738e-05, "loss": 1.4811, "step": 2032 }, { "epoch": 5.0825, "grad_norm": 7.03125, "learning_rate": 1.9977263078761306e-05, "loss": 1.5668, "step": 2033 }, { "epoch": 5.085, "grad_norm": 5.125, "learning_rate": 1.9977208476305333e-05, "loss": 1.5438, "step": 2034 }, { "epoch": 5.0875, "grad_norm": 8.5, "learning_rate": 1.9977153808439186e-05, "loss": 1.9419, "step": 2035 }, { "epoch": 5.09, "grad_norm": 6.875, "learning_rate": 1.9977099075163216e-05, "loss": 1.7359, "step": 2036 }, { "epoch": 5.0925, "grad_norm": 8.5, "learning_rate": 1.9977044276477784e-05, "loss": 1.461, "step": 2037 }, { "epoch": 5.095, "grad_norm": 3.78125, "learning_rate": 1.997698941238325e-05, "loss": 1.6583, "step": 2038 }, { "epoch": 5.0975, "grad_norm": 6.8125, "learning_rate": 1.997693448287997e-05, "loss": 1.6397, "step": 2039 }, { "epoch": 5.1, "grad_norm": 5.8125, "learning_rate": 1.997687948796831e-05, "loss": 1.6672, "step": 2040 }, { "epoch": 5.1025, "grad_norm": 5.625, "learning_rate": 1.9976824427648626e-05, "loss": 1.6267, "step": 2041 }, { "epoch": 5.105, "grad_norm": 7.28125, "learning_rate": 1.9976769301921276e-05, "loss": 1.6865, "step": 2042 }, { "epoch": 5.1075, "grad_norm": 5.375, "learning_rate": 1.9976714110786634e-05, "loss": 1.4902, "step": 2043 }, { "epoch": 5.11, "grad_norm": 5.8125, "learning_rate": 1.9976658854245046e-05, "loss": 1.6348, "step": 2044 }, { "epoch": 5.1125, "grad_norm": 8.875, "learning_rate": 1.9976603532296886e-05, "loss": 1.8354, "step": 2045 }, { "epoch": 5.115, "grad_norm": 10.875, "learning_rate": 1.997654814494251e-05, "loss": 1.8958, "step": 2046 }, { "epoch": 5.1175, "grad_norm": 8.4375, "learning_rate": 1.9976492692182286e-05, "loss": 1.8329, "step": 2047 }, { "epoch": 5.12, "grad_norm": 22.0, "learning_rate": 1.9976437174016575e-05, "loss": 1.6965, "step": 2048 }, { "epoch": 5.1225, "grad_norm": 10.875, "learning_rate": 1.9976381590445738e-05, "loss": 1.8016, "step": 2049 }, { "epoch": 5.125, "grad_norm": 7.5, "learning_rate": 1.9976325941470147e-05, "loss": 1.6241, "step": 2050 }, { "epoch": 5.1275, "grad_norm": 5.96875, "learning_rate": 1.997627022709016e-05, "loss": 1.5465, "step": 2051 }, { "epoch": 5.13, "grad_norm": 16.375, "learning_rate": 1.9976214447306145e-05, "loss": 1.6773, "step": 2052 }, { "epoch": 5.1325, "grad_norm": 5.75, "learning_rate": 1.9976158602118465e-05, "loss": 1.6853, "step": 2053 }, { "epoch": 5.135, "grad_norm": 5.84375, "learning_rate": 1.997610269152749e-05, "loss": 1.4048, "step": 2054 }, { "epoch": 5.1375, "grad_norm": 6.75, "learning_rate": 1.997604671553359e-05, "loss": 1.6169, "step": 2055 }, { "epoch": 5.14, "grad_norm": 5.3125, "learning_rate": 1.997599067413712e-05, "loss": 1.7889, "step": 2056 }, { "epoch": 5.1425, "grad_norm": 5.78125, "learning_rate": 1.9975934567338456e-05, "loss": 1.6459, "step": 2057 }, { "epoch": 5.145, "grad_norm": 4.34375, "learning_rate": 1.9975878395137966e-05, "loss": 1.4955, "step": 2058 }, { "epoch": 5.1475, "grad_norm": 4.1875, "learning_rate": 1.9975822157536012e-05, "loss": 1.6348, "step": 2059 }, { "epoch": 5.15, "grad_norm": 5.6875, "learning_rate": 1.9975765854532974e-05, "loss": 1.5879, "step": 2060 }, { "epoch": 5.1525, "grad_norm": 5.53125, "learning_rate": 1.997570948612921e-05, "loss": 1.6696, "step": 2061 }, { "epoch": 5.155, "grad_norm": 5.125, "learning_rate": 1.9975653052325095e-05, "loss": 1.6307, "step": 2062 }, { "epoch": 5.1575, "grad_norm": 6.46875, "learning_rate": 1.9975596553120996e-05, "loss": 1.6534, "step": 2063 }, { "epoch": 5.16, "grad_norm": 6.46875, "learning_rate": 1.997553998851729e-05, "loss": 1.601, "step": 2064 }, { "epoch": 5.1625, "grad_norm": 7.3125, "learning_rate": 1.9975483358514337e-05, "loss": 1.8723, "step": 2065 }, { "epoch": 5.165, "grad_norm": 6.125, "learning_rate": 1.997542666311252e-05, "loss": 1.6625, "step": 2066 }, { "epoch": 5.1675, "grad_norm": 4.25, "learning_rate": 1.99753699023122e-05, "loss": 1.7558, "step": 2067 }, { "epoch": 5.17, "grad_norm": 5.21875, "learning_rate": 1.9975313076113754e-05, "loss": 1.7058, "step": 2068 }, { "epoch": 5.1725, "grad_norm": 7.75, "learning_rate": 1.9975256184517557e-05, "loss": 1.6924, "step": 2069 }, { "epoch": 5.175, "grad_norm": 6.3125, "learning_rate": 1.9975199227523982e-05, "loss": 1.5497, "step": 2070 }, { "epoch": 5.1775, "grad_norm": 5.84375, "learning_rate": 1.9975142205133395e-05, "loss": 1.5969, "step": 2071 }, { "epoch": 5.18, "grad_norm": 6.125, "learning_rate": 1.997508511734618e-05, "loss": 1.657, "step": 2072 }, { "epoch": 5.1825, "grad_norm": 6.25, "learning_rate": 1.9975027964162704e-05, "loss": 1.672, "step": 2073 }, { "epoch": 5.185, "grad_norm": 7.34375, "learning_rate": 1.9974970745583343e-05, "loss": 1.7897, "step": 2074 }, { "epoch": 5.1875, "grad_norm": 7.5, "learning_rate": 1.9974913461608473e-05, "loss": 1.6729, "step": 2075 }, { "epoch": 5.19, "grad_norm": 6.59375, "learning_rate": 1.997485611223847e-05, "loss": 1.6553, "step": 2076 }, { "epoch": 5.1925, "grad_norm": 30.0, "learning_rate": 1.997479869747371e-05, "loss": 1.7724, "step": 2077 }, { "epoch": 5.195, "grad_norm": 4.875, "learning_rate": 1.9974741217314565e-05, "loss": 1.6413, "step": 2078 }, { "epoch": 5.1975, "grad_norm": 5.125, "learning_rate": 1.997468367176142e-05, "loss": 1.5176, "step": 2079 }, { "epoch": 5.2, "grad_norm": 6.1875, "learning_rate": 1.997462606081465e-05, "loss": 1.6169, "step": 2080 }, { "epoch": 5.2025, "grad_norm": 6.1875, "learning_rate": 1.9974568384474624e-05, "loss": 1.4823, "step": 2081 }, { "epoch": 5.205, "grad_norm": 6.0, "learning_rate": 1.997451064274173e-05, "loss": 1.543, "step": 2082 }, { "epoch": 5.2075, "grad_norm": 5.40625, "learning_rate": 1.9974452835616348e-05, "loss": 1.5888, "step": 2083 }, { "epoch": 5.21, "grad_norm": 7.15625, "learning_rate": 1.9974394963098846e-05, "loss": 1.9485, "step": 2084 }, { "epoch": 5.2125, "grad_norm": 20.375, "learning_rate": 1.9974337025189612e-05, "loss": 1.6815, "step": 2085 }, { "epoch": 5.215, "grad_norm": 5.375, "learning_rate": 1.9974279021889024e-05, "loss": 1.5866, "step": 2086 }, { "epoch": 5.2175, "grad_norm": 6.34375, "learning_rate": 1.9974220953197462e-05, "loss": 1.8156, "step": 2087 }, { "epoch": 5.22, "grad_norm": 7.3125, "learning_rate": 1.9974162819115306e-05, "loss": 1.8008, "step": 2088 }, { "epoch": 5.2225, "grad_norm": 4.78125, "learning_rate": 1.9974104619642934e-05, "loss": 1.6667, "step": 2089 }, { "epoch": 5.225, "grad_norm": 5.84375, "learning_rate": 1.9974046354780737e-05, "loss": 1.7227, "step": 2090 }, { "epoch": 5.2275, "grad_norm": 8.75, "learning_rate": 1.997398802452909e-05, "loss": 1.8229, "step": 2091 }, { "epoch": 5.23, "grad_norm": 5.15625, "learning_rate": 1.9973929628888378e-05, "loss": 1.6864, "step": 2092 }, { "epoch": 5.2325, "grad_norm": 7.4375, "learning_rate": 1.9973871167858978e-05, "loss": 1.6772, "step": 2093 }, { "epoch": 5.235, "grad_norm": 5.40625, "learning_rate": 1.997381264144128e-05, "loss": 1.6241, "step": 2094 }, { "epoch": 5.2375, "grad_norm": 5.125, "learning_rate": 1.9973754049635666e-05, "loss": 1.6507, "step": 2095 }, { "epoch": 5.24, "grad_norm": 7.90625, "learning_rate": 1.997369539244252e-05, "loss": 1.6407, "step": 2096 }, { "epoch": 5.2425, "grad_norm": 6.15625, "learning_rate": 1.9973636669862223e-05, "loss": 1.8205, "step": 2097 }, { "epoch": 5.245, "grad_norm": 7.375, "learning_rate": 1.997357788189517e-05, "loss": 1.9166, "step": 2098 }, { "epoch": 5.2475, "grad_norm": 5.25, "learning_rate": 1.9973519028541734e-05, "loss": 1.5236, "step": 2099 }, { "epoch": 5.25, "grad_norm": 8.5625, "learning_rate": 1.9973460109802306e-05, "loss": 1.7218, "step": 2100 }, { "epoch": 5.2525, "grad_norm": 5.25, "learning_rate": 1.9973401125677274e-05, "loss": 1.6568, "step": 2101 }, { "epoch": 5.255, "grad_norm": 9.25, "learning_rate": 1.9973342076167024e-05, "loss": 1.6099, "step": 2102 }, { "epoch": 5.2575, "grad_norm": 8.25, "learning_rate": 1.997328296127194e-05, "loss": 1.6016, "step": 2103 }, { "epoch": 5.26, "grad_norm": 7.59375, "learning_rate": 1.9973223780992414e-05, "loss": 1.8686, "step": 2104 }, { "epoch": 5.2625, "grad_norm": 6.3125, "learning_rate": 1.9973164535328833e-05, "loss": 1.6484, "step": 2105 }, { "epoch": 5.265, "grad_norm": 5.46875, "learning_rate": 1.9973105224281582e-05, "loss": 1.636, "step": 2106 }, { "epoch": 5.2675, "grad_norm": 4.46875, "learning_rate": 1.9973045847851056e-05, "loss": 1.533, "step": 2107 }, { "epoch": 5.27, "grad_norm": 7.21875, "learning_rate": 1.997298640603764e-05, "loss": 1.665, "step": 2108 }, { "epoch": 5.2725, "grad_norm": 7.625, "learning_rate": 1.9972926898841722e-05, "loss": 1.6432, "step": 2109 }, { "epoch": 5.275, "grad_norm": 6.3125, "learning_rate": 1.997286732626369e-05, "loss": 1.8012, "step": 2110 }, { "epoch": 5.2775, "grad_norm": 6.3125, "learning_rate": 1.9972807688303945e-05, "loss": 1.6416, "step": 2111 }, { "epoch": 5.28, "grad_norm": 4.65625, "learning_rate": 1.997274798496287e-05, "loss": 1.6723, "step": 2112 }, { "epoch": 5.2825, "grad_norm": 5.09375, "learning_rate": 1.9972688216240857e-05, "loss": 1.5814, "step": 2113 }, { "epoch": 5.285, "grad_norm": 5.0625, "learning_rate": 1.99726283821383e-05, "loss": 1.3726, "step": 2114 }, { "epoch": 5.2875, "grad_norm": 5.90625, "learning_rate": 1.9972568482655594e-05, "loss": 1.678, "step": 2115 }, { "epoch": 5.29, "grad_norm": 6.8125, "learning_rate": 1.9972508517793125e-05, "loss": 1.7649, "step": 2116 }, { "epoch": 5.2925, "grad_norm": 6.34375, "learning_rate": 1.997244848755129e-05, "loss": 1.8004, "step": 2117 }, { "epoch": 5.295, "grad_norm": 4.65625, "learning_rate": 1.9972388391930482e-05, "loss": 1.5679, "step": 2118 }, { "epoch": 5.2975, "grad_norm": 7.0, "learning_rate": 1.9972328230931094e-05, "loss": 1.5972, "step": 2119 }, { "epoch": 5.3, "grad_norm": 5.40625, "learning_rate": 1.997226800455352e-05, "loss": 1.5224, "step": 2120 }, { "epoch": 5.3025, "grad_norm": 6.90625, "learning_rate": 1.9972207712798158e-05, "loss": 1.8151, "step": 2121 }, { "epoch": 5.305, "grad_norm": 4.9375, "learning_rate": 1.9972147355665397e-05, "loss": 1.6037, "step": 2122 }, { "epoch": 5.3075, "grad_norm": 79.5, "learning_rate": 1.9972086933155642e-05, "loss": 1.8151, "step": 2123 }, { "epoch": 5.31, "grad_norm": 7.875, "learning_rate": 1.9972026445269285e-05, "loss": 1.7367, "step": 2124 }, { "epoch": 5.3125, "grad_norm": 5.6875, "learning_rate": 1.997196589200672e-05, "loss": 1.6445, "step": 2125 }, { "epoch": 5.315, "grad_norm": 7.28125, "learning_rate": 1.9971905273368345e-05, "loss": 1.9358, "step": 2126 }, { "epoch": 5.3175, "grad_norm": 6.625, "learning_rate": 1.997184458935456e-05, "loss": 1.7274, "step": 2127 }, { "epoch": 5.32, "grad_norm": 7.53125, "learning_rate": 1.9971783839965756e-05, "loss": 1.523, "step": 2128 }, { "epoch": 5.3225, "grad_norm": 6.0625, "learning_rate": 1.9971723025202337e-05, "loss": 1.6227, "step": 2129 }, { "epoch": 5.325, "grad_norm": 7.71875, "learning_rate": 1.9971662145064706e-05, "loss": 1.6045, "step": 2130 }, { "epoch": 5.3275, "grad_norm": 6.96875, "learning_rate": 1.9971601199553254e-05, "loss": 1.6539, "step": 2131 }, { "epoch": 5.33, "grad_norm": 5.34375, "learning_rate": 1.9971540188668386e-05, "loss": 1.7764, "step": 2132 }, { "epoch": 5.3325, "grad_norm": 5.15625, "learning_rate": 1.9971479112410494e-05, "loss": 1.7133, "step": 2133 }, { "epoch": 5.335, "grad_norm": 6.28125, "learning_rate": 1.9971417970779988e-05, "loss": 1.7716, "step": 2134 }, { "epoch": 5.3375, "grad_norm": 8.6875, "learning_rate": 1.9971356763777265e-05, "loss": 1.8087, "step": 2135 }, { "epoch": 5.34, "grad_norm": 7.21875, "learning_rate": 1.9971295491402725e-05, "loss": 1.6253, "step": 2136 }, { "epoch": 5.3425, "grad_norm": 7.3125, "learning_rate": 1.997123415365677e-05, "loss": 1.6287, "step": 2137 }, { "epoch": 5.345, "grad_norm": 6.21875, "learning_rate": 1.9971172750539804e-05, "loss": 1.4573, "step": 2138 }, { "epoch": 5.3475, "grad_norm": 7.0625, "learning_rate": 1.997111128205223e-05, "loss": 1.7682, "step": 2139 }, { "epoch": 5.35, "grad_norm": 6.03125, "learning_rate": 1.9971049748194448e-05, "loss": 1.6906, "step": 2140 }, { "epoch": 5.3525, "grad_norm": 6.625, "learning_rate": 1.997098814896686e-05, "loss": 1.7008, "step": 2141 }, { "epoch": 5.355, "grad_norm": 4.90625, "learning_rate": 1.997092648436988e-05, "loss": 1.6114, "step": 2142 }, { "epoch": 5.3575, "grad_norm": 7.09375, "learning_rate": 1.9970864754403898e-05, "loss": 1.676, "step": 2143 }, { "epoch": 5.36, "grad_norm": 5.96875, "learning_rate": 1.997080295906933e-05, "loss": 1.46, "step": 2144 }, { "epoch": 5.3625, "grad_norm": 6.71875, "learning_rate": 1.9970741098366574e-05, "loss": 1.6537, "step": 2145 }, { "epoch": 5.365, "grad_norm": 5.53125, "learning_rate": 1.997067917229604e-05, "loss": 1.5157, "step": 2146 }, { "epoch": 5.3675, "grad_norm": 5.3125, "learning_rate": 1.997061718085813e-05, "loss": 1.537, "step": 2147 }, { "epoch": 5.37, "grad_norm": 6.1875, "learning_rate": 1.9970555124053256e-05, "loss": 1.5885, "step": 2148 }, { "epoch": 5.3725, "grad_norm": 10.4375, "learning_rate": 1.9970493001881818e-05, "loss": 1.7094, "step": 2149 }, { "epoch": 5.375, "grad_norm": 6.78125, "learning_rate": 1.997043081434423e-05, "loss": 1.6174, "step": 2150 }, { "epoch": 5.3775, "grad_norm": 10.125, "learning_rate": 1.9970368561440895e-05, "loss": 1.6614, "step": 2151 }, { "epoch": 5.38, "grad_norm": 5.84375, "learning_rate": 1.9970306243172223e-05, "loss": 1.5953, "step": 2152 }, { "epoch": 5.3825, "grad_norm": 4.90625, "learning_rate": 1.997024385953862e-05, "loss": 1.6171, "step": 2153 }, { "epoch": 5.385, "grad_norm": 8.5, "learning_rate": 1.99701814105405e-05, "loss": 1.6505, "step": 2154 }, { "epoch": 5.3875, "grad_norm": 6.46875, "learning_rate": 1.9970118896178273e-05, "loss": 1.8278, "step": 2155 }, { "epoch": 5.39, "grad_norm": 2064.0, "learning_rate": 1.997005631645234e-05, "loss": 1.7972, "step": 2156 }, { "epoch": 5.3925, "grad_norm": 5.34375, "learning_rate": 1.996999367136312e-05, "loss": 1.6865, "step": 2157 }, { "epoch": 5.395, "grad_norm": 4.5, "learning_rate": 1.9969930960911017e-05, "loss": 1.5357, "step": 2158 }, { "epoch": 5.3975, "grad_norm": 6.5625, "learning_rate": 1.9969868185096448e-05, "loss": 1.8303, "step": 2159 }, { "epoch": 5.4, "grad_norm": 5.25, "learning_rate": 1.9969805343919822e-05, "loss": 1.6259, "step": 2160 }, { "epoch": 5.4025, "grad_norm": 5.9375, "learning_rate": 1.996974243738155e-05, "loss": 1.6001, "step": 2161 }, { "epoch": 5.405, "grad_norm": 6.5, "learning_rate": 1.9969679465482046e-05, "loss": 1.7144, "step": 2162 }, { "epoch": 5.4075, "grad_norm": 5.9375, "learning_rate": 1.9969616428221725e-05, "loss": 1.6664, "step": 2163 }, { "epoch": 5.41, "grad_norm": 6.84375, "learning_rate": 1.9969553325600994e-05, "loss": 1.6816, "step": 2164 }, { "epoch": 5.4125, "grad_norm": 9.0625, "learning_rate": 1.9969490157620274e-05, "loss": 1.8637, "step": 2165 }, { "epoch": 5.415, "grad_norm": 5.5, "learning_rate": 1.996942692427997e-05, "loss": 1.5715, "step": 2166 }, { "epoch": 5.4175, "grad_norm": 7.96875, "learning_rate": 1.9969363625580508e-05, "loss": 1.8171, "step": 2167 }, { "epoch": 5.42, "grad_norm": 6.0625, "learning_rate": 1.9969300261522292e-05, "loss": 1.7009, "step": 2168 }, { "epoch": 5.4225, "grad_norm": 6.5, "learning_rate": 1.9969236832105747e-05, "loss": 1.7389, "step": 2169 }, { "epoch": 5.425, "grad_norm": 6.21875, "learning_rate": 1.9969173337331283e-05, "loss": 1.6405, "step": 2170 }, { "epoch": 5.4275, "grad_norm": 13.1875, "learning_rate": 1.9969109777199315e-05, "loss": 1.6851, "step": 2171 }, { "epoch": 5.43, "grad_norm": 8.1875, "learning_rate": 1.9969046151710266e-05, "loss": 1.7082, "step": 2172 }, { "epoch": 5.4325, "grad_norm": 6.78125, "learning_rate": 1.9968982460864547e-05, "loss": 1.7967, "step": 2173 }, { "epoch": 5.435, "grad_norm": 5.4375, "learning_rate": 1.9968918704662576e-05, "loss": 1.5779, "step": 2174 }, { "epoch": 5.4375, "grad_norm": 4.15625, "learning_rate": 1.9968854883104776e-05, "loss": 1.5303, "step": 2175 }, { "epoch": 5.44, "grad_norm": 7.375, "learning_rate": 1.996879099619156e-05, "loss": 1.6608, "step": 2176 }, { "epoch": 5.4425, "grad_norm": 6.5625, "learning_rate": 1.9968727043923352e-05, "loss": 1.4983, "step": 2177 }, { "epoch": 5.445, "grad_norm": 6.4375, "learning_rate": 1.9968663026300566e-05, "loss": 1.6625, "step": 2178 }, { "epoch": 5.4475, "grad_norm": 6.0, "learning_rate": 1.9968598943323627e-05, "loss": 1.9112, "step": 2179 }, { "epoch": 5.45, "grad_norm": 5.28125, "learning_rate": 1.9968534794992947e-05, "loss": 1.5617, "step": 2180 }, { "epoch": 5.4525, "grad_norm": 5.15625, "learning_rate": 1.9968470581308955e-05, "loss": 1.6734, "step": 2181 }, { "epoch": 5.455, "grad_norm": 7.875, "learning_rate": 1.9968406302272072e-05, "loss": 1.7756, "step": 2182 }, { "epoch": 5.4575, "grad_norm": 5.1875, "learning_rate": 1.9968341957882715e-05, "loss": 1.5957, "step": 2183 }, { "epoch": 5.46, "grad_norm": 4.625, "learning_rate": 1.9968277548141302e-05, "loss": 1.5627, "step": 2184 }, { "epoch": 5.4625, "grad_norm": 6.34375, "learning_rate": 1.9968213073048262e-05, "loss": 1.6017, "step": 2185 }, { "epoch": 5.465, "grad_norm": 5.03125, "learning_rate": 1.9968148532604018e-05, "loss": 1.7494, "step": 2186 }, { "epoch": 5.4675, "grad_norm": 7.875, "learning_rate": 1.9968083926808993e-05, "loss": 1.6975, "step": 2187 }, { "epoch": 5.47, "grad_norm": 5.59375, "learning_rate": 1.9968019255663608e-05, "loss": 1.6363, "step": 2188 }, { "epoch": 5.4725, "grad_norm": 8.6875, "learning_rate": 1.9967954519168283e-05, "loss": 1.8022, "step": 2189 }, { "epoch": 5.475, "grad_norm": 5.625, "learning_rate": 1.996788971732345e-05, "loss": 1.4701, "step": 2190 }, { "epoch": 5.4775, "grad_norm": 5.84375, "learning_rate": 1.996782485012953e-05, "loss": 1.6917, "step": 2191 }, { "epoch": 5.48, "grad_norm": 6.09375, "learning_rate": 1.9967759917586953e-05, "loss": 1.6249, "step": 2192 }, { "epoch": 5.4825, "grad_norm": 7.0, "learning_rate": 1.9967694919696135e-05, "loss": 1.9177, "step": 2193 }, { "epoch": 5.485, "grad_norm": 6.90625, "learning_rate": 1.9967629856457515e-05, "loss": 1.6283, "step": 2194 }, { "epoch": 5.4875, "grad_norm": 4.6875, "learning_rate": 1.9967564727871505e-05, "loss": 1.5444, "step": 2195 }, { "epoch": 5.49, "grad_norm": 5.0, "learning_rate": 1.9967499533938544e-05, "loss": 1.7457, "step": 2196 }, { "epoch": 5.4925, "grad_norm": 4.6875, "learning_rate": 1.9967434274659054e-05, "loss": 1.5663, "step": 2197 }, { "epoch": 5.495, "grad_norm": 14.75, "learning_rate": 1.9967368950033462e-05, "loss": 1.7374, "step": 2198 }, { "epoch": 5.4975, "grad_norm": 6.5, "learning_rate": 1.9967303560062202e-05, "loss": 1.5843, "step": 2199 }, { "epoch": 5.5, "grad_norm": 5.5625, "learning_rate": 1.9967238104745695e-05, "loss": 1.7666, "step": 2200 }, { "epoch": 5.5024999999999995, "grad_norm": 7.0625, "learning_rate": 1.9967172584084377e-05, "loss": 1.7491, "step": 2201 }, { "epoch": 5.505, "grad_norm": 6.0625, "learning_rate": 1.996710699807867e-05, "loss": 1.7878, "step": 2202 }, { "epoch": 5.5075, "grad_norm": 6.53125, "learning_rate": 1.9967041346729013e-05, "loss": 1.6933, "step": 2203 }, { "epoch": 5.51, "grad_norm": 11.3125, "learning_rate": 1.996697563003583e-05, "loss": 1.6442, "step": 2204 }, { "epoch": 5.5125, "grad_norm": 9.125, "learning_rate": 1.9966909847999552e-05, "loss": 1.7076, "step": 2205 }, { "epoch": 5.515, "grad_norm": 8.625, "learning_rate": 1.9966844000620615e-05, "loss": 1.7292, "step": 2206 }, { "epoch": 5.5175, "grad_norm": 8.6875, "learning_rate": 1.9966778087899445e-05, "loss": 1.7068, "step": 2207 }, { "epoch": 5.52, "grad_norm": 6.75, "learning_rate": 1.9966712109836476e-05, "loss": 1.6329, "step": 2208 }, { "epoch": 5.5225, "grad_norm": 5.03125, "learning_rate": 1.9966646066432146e-05, "loss": 1.6043, "step": 2209 }, { "epoch": 5.525, "grad_norm": 8.1875, "learning_rate": 1.996657995768688e-05, "loss": 1.6171, "step": 2210 }, { "epoch": 5.5275, "grad_norm": 9.3125, "learning_rate": 1.9966513783601115e-05, "loss": 1.6347, "step": 2211 }, { "epoch": 5.53, "grad_norm": 6.90625, "learning_rate": 1.9966447544175285e-05, "loss": 1.8571, "step": 2212 }, { "epoch": 5.5325, "grad_norm": 8.125, "learning_rate": 1.996638123940982e-05, "loss": 1.559, "step": 2213 }, { "epoch": 5.535, "grad_norm": 5.875, "learning_rate": 1.9966314869305164e-05, "loss": 1.6102, "step": 2214 }, { "epoch": 5.5375, "grad_norm": 6.09375, "learning_rate": 1.9966248433861744e-05, "loss": 1.6365, "step": 2215 }, { "epoch": 5.54, "grad_norm": 6.46875, "learning_rate": 1.996618193308e-05, "loss": 1.5467, "step": 2216 }, { "epoch": 5.5425, "grad_norm": 8.75, "learning_rate": 1.996611536696036e-05, "loss": 1.65, "step": 2217 }, { "epoch": 5.545, "grad_norm": 5.96875, "learning_rate": 1.996604873550327e-05, "loss": 1.6633, "step": 2218 }, { "epoch": 5.5475, "grad_norm": 7.0625, "learning_rate": 1.9965982038709168e-05, "loss": 1.772, "step": 2219 }, { "epoch": 5.55, "grad_norm": 4.65625, "learning_rate": 1.996591527657848e-05, "loss": 1.4945, "step": 2220 }, { "epoch": 5.5525, "grad_norm": 5.84375, "learning_rate": 1.9965848449111654e-05, "loss": 1.7979, "step": 2221 }, { "epoch": 5.555, "grad_norm": 6.15625, "learning_rate": 1.996578155630912e-05, "loss": 1.6959, "step": 2222 }, { "epoch": 5.5575, "grad_norm": 4.96875, "learning_rate": 1.9965714598171326e-05, "loss": 1.5656, "step": 2223 }, { "epoch": 5.5600000000000005, "grad_norm": 9.0, "learning_rate": 1.9965647574698705e-05, "loss": 1.7652, "step": 2224 }, { "epoch": 5.5625, "grad_norm": 7.75, "learning_rate": 1.9965580485891695e-05, "loss": 1.6712, "step": 2225 }, { "epoch": 5.5649999999999995, "grad_norm": 5.5, "learning_rate": 1.996551333175074e-05, "loss": 1.6026, "step": 2226 }, { "epoch": 5.5675, "grad_norm": 6.21875, "learning_rate": 1.996544611227628e-05, "loss": 1.5881, "step": 2227 }, { "epoch": 5.57, "grad_norm": 7.125, "learning_rate": 1.996537882746875e-05, "loss": 1.5088, "step": 2228 }, { "epoch": 5.5725, "grad_norm": 7.71875, "learning_rate": 1.9965311477328596e-05, "loss": 1.6676, "step": 2229 }, { "epoch": 5.575, "grad_norm": 9.6875, "learning_rate": 1.9965244061856262e-05, "loss": 1.6279, "step": 2230 }, { "epoch": 5.5775, "grad_norm": 10.3125, "learning_rate": 1.9965176581052185e-05, "loss": 2.032, "step": 2231 }, { "epoch": 5.58, "grad_norm": 7.28125, "learning_rate": 1.9965109034916806e-05, "loss": 1.6648, "step": 2232 }, { "epoch": 5.5825, "grad_norm": 9.8125, "learning_rate": 1.9965041423450576e-05, "loss": 2.0315, "step": 2233 }, { "epoch": 5.585, "grad_norm": 6.53125, "learning_rate": 1.9964973746653928e-05, "loss": 1.7642, "step": 2234 }, { "epoch": 5.5875, "grad_norm": 6.0, "learning_rate": 1.9964906004527316e-05, "loss": 1.5243, "step": 2235 }, { "epoch": 5.59, "grad_norm": 6.28125, "learning_rate": 1.9964838197071173e-05, "loss": 1.7973, "step": 2236 }, { "epoch": 5.5925, "grad_norm": 5.59375, "learning_rate": 1.9964770324285954e-05, "loss": 1.6662, "step": 2237 }, { "epoch": 5.595, "grad_norm": 6.15625, "learning_rate": 1.99647023861721e-05, "loss": 1.6698, "step": 2238 }, { "epoch": 5.5975, "grad_norm": 6.0, "learning_rate": 1.9964634382730053e-05, "loss": 1.7823, "step": 2239 }, { "epoch": 5.6, "grad_norm": 6.6875, "learning_rate": 1.9964566313960265e-05, "loss": 1.653, "step": 2240 }, { "epoch": 5.6025, "grad_norm": 5.3125, "learning_rate": 1.996449817986318e-05, "loss": 1.647, "step": 2241 }, { "epoch": 5.605, "grad_norm": 5.71875, "learning_rate": 1.996442998043924e-05, "loss": 1.8186, "step": 2242 }, { "epoch": 5.6075, "grad_norm": 5.8125, "learning_rate": 1.9964361715688896e-05, "loss": 1.6244, "step": 2243 }, { "epoch": 5.61, "grad_norm": 6.28125, "learning_rate": 1.9964293385612595e-05, "loss": 1.5399, "step": 2244 }, { "epoch": 5.6125, "grad_norm": 6.84375, "learning_rate": 1.996422499021079e-05, "loss": 1.7224, "step": 2245 }, { "epoch": 5.615, "grad_norm": 5.6875, "learning_rate": 1.996415652948392e-05, "loss": 1.611, "step": 2246 }, { "epoch": 5.6175, "grad_norm": 6.1875, "learning_rate": 1.996408800343244e-05, "loss": 1.5723, "step": 2247 }, { "epoch": 5.62, "grad_norm": 4.375, "learning_rate": 1.9964019412056798e-05, "loss": 1.6972, "step": 2248 }, { "epoch": 5.6225000000000005, "grad_norm": 5.125, "learning_rate": 1.996395075535744e-05, "loss": 1.6266, "step": 2249 }, { "epoch": 5.625, "grad_norm": 4.75, "learning_rate": 1.9963882033334827e-05, "loss": 1.6452, "step": 2250 }, { "epoch": 5.6274999999999995, "grad_norm": 8.3125, "learning_rate": 1.9963813245989395e-05, "loss": 1.5842, "step": 2251 }, { "epoch": 5.63, "grad_norm": 4.5, "learning_rate": 1.9963744393321606e-05, "loss": 1.579, "step": 2252 }, { "epoch": 5.6325, "grad_norm": 6.5625, "learning_rate": 1.9963675475331907e-05, "loss": 1.4868, "step": 2253 }, { "epoch": 5.635, "grad_norm": 8.375, "learning_rate": 1.996360649202075e-05, "loss": 1.739, "step": 2254 }, { "epoch": 5.6375, "grad_norm": 6.46875, "learning_rate": 1.996353744338859e-05, "loss": 1.7206, "step": 2255 }, { "epoch": 5.64, "grad_norm": 6.5625, "learning_rate": 1.9963468329435872e-05, "loss": 1.8702, "step": 2256 }, { "epoch": 5.6425, "grad_norm": 9.0, "learning_rate": 1.9963399150163058e-05, "loss": 1.7472, "step": 2257 }, { "epoch": 5.645, "grad_norm": 6.59375, "learning_rate": 1.9963329905570593e-05, "loss": 1.7637, "step": 2258 }, { "epoch": 5.6475, "grad_norm": 7.03125, "learning_rate": 1.9963260595658942e-05, "loss": 1.7258, "step": 2259 }, { "epoch": 5.65, "grad_norm": 7.25, "learning_rate": 1.9963191220428552e-05, "loss": 1.6273, "step": 2260 }, { "epoch": 5.6525, "grad_norm": 8.0625, "learning_rate": 1.9963121779879876e-05, "loss": 1.7611, "step": 2261 }, { "epoch": 5.655, "grad_norm": 7.375, "learning_rate": 1.9963052274013374e-05, "loss": 1.6042, "step": 2262 }, { "epoch": 5.6575, "grad_norm": 7.09375, "learning_rate": 1.99629827028295e-05, "loss": 1.6404, "step": 2263 }, { "epoch": 5.66, "grad_norm": 7.75, "learning_rate": 1.996291306632871e-05, "loss": 1.5047, "step": 2264 }, { "epoch": 5.6625, "grad_norm": 7.0, "learning_rate": 1.9962843364511463e-05, "loss": 1.4837, "step": 2265 }, { "epoch": 5.665, "grad_norm": 7.0625, "learning_rate": 1.996277359737821e-05, "loss": 1.793, "step": 2266 }, { "epoch": 5.6675, "grad_norm": 8.3125, "learning_rate": 1.9962703764929413e-05, "loss": 1.488, "step": 2267 }, { "epoch": 5.67, "grad_norm": 8.1875, "learning_rate": 1.996263386716553e-05, "loss": 1.7524, "step": 2268 }, { "epoch": 5.6725, "grad_norm": 4.96875, "learning_rate": 1.9962563904087016e-05, "loss": 1.8305, "step": 2269 }, { "epoch": 5.675, "grad_norm": 6.375, "learning_rate": 1.9962493875694335e-05, "loss": 1.5811, "step": 2270 }, { "epoch": 5.6775, "grad_norm": 8.4375, "learning_rate": 1.9962423781987942e-05, "loss": 1.7489, "step": 2271 }, { "epoch": 5.68, "grad_norm": 5.46875, "learning_rate": 1.9962353622968296e-05, "loss": 1.6331, "step": 2272 }, { "epoch": 5.6825, "grad_norm": 5.90625, "learning_rate": 1.9962283398635857e-05, "loss": 1.6167, "step": 2273 }, { "epoch": 5.6850000000000005, "grad_norm": 6.71875, "learning_rate": 1.996221310899109e-05, "loss": 1.3987, "step": 2274 }, { "epoch": 5.6875, "grad_norm": 6.9375, "learning_rate": 1.996214275403445e-05, "loss": 1.6291, "step": 2275 }, { "epoch": 5.6899999999999995, "grad_norm": 5.96875, "learning_rate": 1.99620723337664e-05, "loss": 1.7655, "step": 2276 }, { "epoch": 5.6925, "grad_norm": 6.21875, "learning_rate": 1.9962001848187405e-05, "loss": 1.644, "step": 2277 }, { "epoch": 5.695, "grad_norm": 5.1875, "learning_rate": 1.9961931297297923e-05, "loss": 1.5232, "step": 2278 }, { "epoch": 5.6975, "grad_norm": 5.65625, "learning_rate": 1.996186068109842e-05, "loss": 1.5205, "step": 2279 }, { "epoch": 5.7, "grad_norm": 4.25, "learning_rate": 1.9961789999589357e-05, "loss": 1.7295, "step": 2280 }, { "epoch": 5.7025, "grad_norm": 4.71875, "learning_rate": 1.9961719252771194e-05, "loss": 1.6632, "step": 2281 }, { "epoch": 5.705, "grad_norm": 7.5, "learning_rate": 1.99616484406444e-05, "loss": 1.6127, "step": 2282 }, { "epoch": 5.7075, "grad_norm": 6.5, "learning_rate": 1.996157756320944e-05, "loss": 1.6887, "step": 2283 }, { "epoch": 5.71, "grad_norm": 8.25, "learning_rate": 1.9961506620466774e-05, "loss": 1.7595, "step": 2284 }, { "epoch": 5.7125, "grad_norm": 6.3125, "learning_rate": 1.9961435612416866e-05, "loss": 1.7872, "step": 2285 }, { "epoch": 5.715, "grad_norm": 4.96875, "learning_rate": 1.996136453906019e-05, "loss": 1.5654, "step": 2286 }, { "epoch": 5.7175, "grad_norm": 5.21875, "learning_rate": 1.9961293400397203e-05, "loss": 1.5782, "step": 2287 }, { "epoch": 5.72, "grad_norm": 6.84375, "learning_rate": 1.996122219642838e-05, "loss": 1.749, "step": 2288 }, { "epoch": 5.7225, "grad_norm": 6.03125, "learning_rate": 1.9961150927154178e-05, "loss": 1.6934, "step": 2289 }, { "epoch": 5.725, "grad_norm": 7.25, "learning_rate": 1.996107959257507e-05, "loss": 1.6233, "step": 2290 }, { "epoch": 5.7275, "grad_norm": 7.21875, "learning_rate": 1.9961008192691525e-05, "loss": 1.7, "step": 2291 }, { "epoch": 5.73, "grad_norm": 5.8125, "learning_rate": 1.9960936727504007e-05, "loss": 1.6506, "step": 2292 }, { "epoch": 5.7325, "grad_norm": 7.1875, "learning_rate": 1.9960865197012987e-05, "loss": 1.8225, "step": 2293 }, { "epoch": 5.735, "grad_norm": 8.5625, "learning_rate": 1.9960793601218932e-05, "loss": 1.5096, "step": 2294 }, { "epoch": 5.7375, "grad_norm": 6.65625, "learning_rate": 1.9960721940122316e-05, "loss": 1.7573, "step": 2295 }, { "epoch": 5.74, "grad_norm": 6.375, "learning_rate": 1.9960650213723604e-05, "loss": 1.4884, "step": 2296 }, { "epoch": 5.7425, "grad_norm": 8.5, "learning_rate": 1.9960578422023266e-05, "loss": 1.7483, "step": 2297 }, { "epoch": 5.745, "grad_norm": 6.6875, "learning_rate": 1.9960506565021777e-05, "loss": 1.5305, "step": 2298 }, { "epoch": 5.7475000000000005, "grad_norm": 5.625, "learning_rate": 1.9960434642719604e-05, "loss": 1.5794, "step": 2299 }, { "epoch": 5.75, "grad_norm": 5.90625, "learning_rate": 1.996036265511722e-05, "loss": 1.6421, "step": 2300 }, { "epoch": 5.7524999999999995, "grad_norm": 4.6875, "learning_rate": 1.9960290602215098e-05, "loss": 1.6582, "step": 2301 }, { "epoch": 5.755, "grad_norm": 7.5, "learning_rate": 1.996021848401371e-05, "loss": 1.6655, "step": 2302 }, { "epoch": 5.7575, "grad_norm": 6.34375, "learning_rate": 1.9960146300513525e-05, "loss": 1.613, "step": 2303 }, { "epoch": 5.76, "grad_norm": 5.28125, "learning_rate": 1.9960074051715022e-05, "loss": 1.7677, "step": 2304 }, { "epoch": 5.7625, "grad_norm": 6.8125, "learning_rate": 1.9960001737618674e-05, "loss": 1.5953, "step": 2305 }, { "epoch": 5.765, "grad_norm": 6.25, "learning_rate": 1.995992935822495e-05, "loss": 1.6447, "step": 2306 }, { "epoch": 5.7675, "grad_norm": 6.25, "learning_rate": 1.9959856913534325e-05, "loss": 1.6409, "step": 2307 }, { "epoch": 5.77, "grad_norm": 7.0, "learning_rate": 1.9959784403547283e-05, "loss": 1.573, "step": 2308 }, { "epoch": 5.7725, "grad_norm": 6.71875, "learning_rate": 1.9959711828264288e-05, "loss": 1.7593, "step": 2309 }, { "epoch": 5.775, "grad_norm": 6.78125, "learning_rate": 1.9959639187685826e-05, "loss": 1.8107, "step": 2310 }, { "epoch": 5.7775, "grad_norm": 8.625, "learning_rate": 1.9959566481812366e-05, "loss": 1.6768, "step": 2311 }, { "epoch": 5.78, "grad_norm": 6.40625, "learning_rate": 1.9959493710644385e-05, "loss": 1.6399, "step": 2312 }, { "epoch": 5.7825, "grad_norm": 5.46875, "learning_rate": 1.995942087418236e-05, "loss": 1.635, "step": 2313 }, { "epoch": 5.785, "grad_norm": 5.59375, "learning_rate": 1.9959347972426775e-05, "loss": 1.6774, "step": 2314 }, { "epoch": 5.7875, "grad_norm": 8.1875, "learning_rate": 1.99592750053781e-05, "loss": 1.6901, "step": 2315 }, { "epoch": 5.79, "grad_norm": 9.0625, "learning_rate": 1.9959201973036816e-05, "loss": 1.6693, "step": 2316 }, { "epoch": 5.7925, "grad_norm": 9.625, "learning_rate": 1.9959128875403408e-05, "loss": 1.7853, "step": 2317 }, { "epoch": 5.795, "grad_norm": 6.5625, "learning_rate": 1.9959055712478342e-05, "loss": 1.7881, "step": 2318 }, { "epoch": 5.7975, "grad_norm": 8.125, "learning_rate": 1.995898248426211e-05, "loss": 1.7552, "step": 2319 }, { "epoch": 5.8, "grad_norm": 5.3125, "learning_rate": 1.995890919075519e-05, "loss": 1.7212, "step": 2320 }, { "epoch": 5.8025, "grad_norm": 6.28125, "learning_rate": 1.9958835831958053e-05, "loss": 1.5146, "step": 2321 }, { "epoch": 5.805, "grad_norm": 7.53125, "learning_rate": 1.995876240787119e-05, "loss": 1.8392, "step": 2322 }, { "epoch": 5.8075, "grad_norm": 5.125, "learning_rate": 1.995868891849508e-05, "loss": 1.7768, "step": 2323 }, { "epoch": 5.8100000000000005, "grad_norm": 5.34375, "learning_rate": 1.9958615363830202e-05, "loss": 1.5047, "step": 2324 }, { "epoch": 5.8125, "grad_norm": 6.34375, "learning_rate": 1.995854174387704e-05, "loss": 1.6771, "step": 2325 }, { "epoch": 5.8149999999999995, "grad_norm": 5.21875, "learning_rate": 1.995846805863608e-05, "loss": 1.707, "step": 2326 }, { "epoch": 5.8175, "grad_norm": 5.375, "learning_rate": 1.9958394308107803e-05, "loss": 1.6909, "step": 2327 }, { "epoch": 5.82, "grad_norm": 8.625, "learning_rate": 1.9958320492292686e-05, "loss": 1.7297, "step": 2328 }, { "epoch": 5.8225, "grad_norm": 6.28125, "learning_rate": 1.995824661119122e-05, "loss": 1.6569, "step": 2329 }, { "epoch": 5.825, "grad_norm": 23.875, "learning_rate": 1.995817266480389e-05, "loss": 1.6874, "step": 2330 }, { "epoch": 5.8275, "grad_norm": 8.4375, "learning_rate": 1.9958098653131177e-05, "loss": 1.5928, "step": 2331 }, { "epoch": 5.83, "grad_norm": 6.125, "learning_rate": 1.995802457617357e-05, "loss": 1.5493, "step": 2332 }, { "epoch": 5.8325, "grad_norm": 8.25, "learning_rate": 1.9957950433931548e-05, "loss": 1.7138, "step": 2333 }, { "epoch": 5.835, "grad_norm": 7.59375, "learning_rate": 1.9957876226405607e-05, "loss": 1.6705, "step": 2334 }, { "epoch": 5.8375, "grad_norm": 6.0, "learning_rate": 1.9957801953596224e-05, "loss": 1.8065, "step": 2335 }, { "epoch": 5.84, "grad_norm": 7.65625, "learning_rate": 1.995772761550389e-05, "loss": 1.6713, "step": 2336 }, { "epoch": 5.8425, "grad_norm": 8.125, "learning_rate": 1.9957653212129093e-05, "loss": 1.6603, "step": 2337 }, { "epoch": 5.845, "grad_norm": 7.90625, "learning_rate": 1.995757874347232e-05, "loss": 1.6792, "step": 2338 }, { "epoch": 5.8475, "grad_norm": 7.21875, "learning_rate": 1.9957504209534058e-05, "loss": 1.7762, "step": 2339 }, { "epoch": 5.85, "grad_norm": 6.5625, "learning_rate": 1.9957429610314797e-05, "loss": 1.6014, "step": 2340 }, { "epoch": 5.8525, "grad_norm": 6.09375, "learning_rate": 1.9957354945815028e-05, "loss": 1.747, "step": 2341 }, { "epoch": 5.855, "grad_norm": 5.78125, "learning_rate": 1.9957280216035236e-05, "loss": 1.6766, "step": 2342 }, { "epoch": 5.8575, "grad_norm": 5.5625, "learning_rate": 1.9957205420975917e-05, "loss": 1.5917, "step": 2343 }, { "epoch": 5.86, "grad_norm": 5.25, "learning_rate": 1.9957130560637553e-05, "loss": 1.618, "step": 2344 }, { "epoch": 5.8625, "grad_norm": 6.96875, "learning_rate": 1.9957055635020644e-05, "loss": 1.7007, "step": 2345 }, { "epoch": 5.865, "grad_norm": 7.5625, "learning_rate": 1.9956980644125674e-05, "loss": 1.4824, "step": 2346 }, { "epoch": 5.8675, "grad_norm": 6.46875, "learning_rate": 1.995690558795314e-05, "loss": 1.7251, "step": 2347 }, { "epoch": 5.87, "grad_norm": 4.78125, "learning_rate": 1.995683046650353e-05, "loss": 1.6442, "step": 2348 }, { "epoch": 5.8725000000000005, "grad_norm": 4.53125, "learning_rate": 1.9956755279777336e-05, "loss": 1.6633, "step": 2349 }, { "epoch": 5.875, "grad_norm": 7.96875, "learning_rate": 1.9956680027775054e-05, "loss": 1.6649, "step": 2350 }, { "epoch": 5.8774999999999995, "grad_norm": 7.25, "learning_rate": 1.9956604710497174e-05, "loss": 1.7658, "step": 2351 }, { "epoch": 5.88, "grad_norm": 5.4375, "learning_rate": 1.9956529327944198e-05, "loss": 1.6133, "step": 2352 }, { "epoch": 5.8825, "grad_norm": 7.0625, "learning_rate": 1.995645388011661e-05, "loss": 1.6702, "step": 2353 }, { "epoch": 5.885, "grad_norm": 6.625, "learning_rate": 1.9956378367014908e-05, "loss": 1.8189, "step": 2354 }, { "epoch": 5.8875, "grad_norm": 7.84375, "learning_rate": 1.995630278863959e-05, "loss": 1.8147, "step": 2355 }, { "epoch": 5.89, "grad_norm": 5.875, "learning_rate": 1.995622714499115e-05, "loss": 1.5994, "step": 2356 }, { "epoch": 5.8925, "grad_norm": 6.96875, "learning_rate": 1.995615143607008e-05, "loss": 1.7996, "step": 2357 }, { "epoch": 5.895, "grad_norm": 4.59375, "learning_rate": 1.995607566187688e-05, "loss": 1.5038, "step": 2358 }, { "epoch": 5.8975, "grad_norm": 6.78125, "learning_rate": 1.9955999822412048e-05, "loss": 1.8393, "step": 2359 }, { "epoch": 5.9, "grad_norm": 6.0, "learning_rate": 1.995592391767608e-05, "loss": 1.8301, "step": 2360 }, { "epoch": 5.9025, "grad_norm": 7.4375, "learning_rate": 1.9955847947669475e-05, "loss": 1.6885, "step": 2361 }, { "epoch": 5.905, "grad_norm": 6.28125, "learning_rate": 1.9955771912392724e-05, "loss": 1.4929, "step": 2362 }, { "epoch": 5.9075, "grad_norm": 5.875, "learning_rate": 1.9955695811846337e-05, "loss": 1.6659, "step": 2363 }, { "epoch": 5.91, "grad_norm": 5.875, "learning_rate": 1.99556196460308e-05, "loss": 1.5354, "step": 2364 }, { "epoch": 5.9125, "grad_norm": 5.125, "learning_rate": 1.9955543414946624e-05, "loss": 1.5381, "step": 2365 }, { "epoch": 5.915, "grad_norm": 5.125, "learning_rate": 1.99554671185943e-05, "loss": 1.6477, "step": 2366 }, { "epoch": 5.9175, "grad_norm": 7.0625, "learning_rate": 1.995539075697434e-05, "loss": 1.7639, "step": 2367 }, { "epoch": 5.92, "grad_norm": 6.09375, "learning_rate": 1.9955314330087225e-05, "loss": 1.7777, "step": 2368 }, { "epoch": 5.9225, "grad_norm": 8.5, "learning_rate": 1.9955237837933475e-05, "loss": 1.6923, "step": 2369 }, { "epoch": 5.925, "grad_norm": 6.09375, "learning_rate": 1.9955161280513583e-05, "loss": 1.5481, "step": 2370 }, { "epoch": 5.9275, "grad_norm": 5.40625, "learning_rate": 1.995508465782805e-05, "loss": 1.6583, "step": 2371 }, { "epoch": 5.93, "grad_norm": 8.0, "learning_rate": 1.9955007969877384e-05, "loss": 1.6485, "step": 2372 }, { "epoch": 5.9325, "grad_norm": 5.34375, "learning_rate": 1.9954931216662082e-05, "loss": 1.5766, "step": 2373 }, { "epoch": 5.9350000000000005, "grad_norm": 5.65625, "learning_rate": 1.995485439818265e-05, "loss": 1.5442, "step": 2374 }, { "epoch": 5.9375, "grad_norm": 5.625, "learning_rate": 1.995477751443959e-05, "loss": 1.7264, "step": 2375 }, { "epoch": 5.9399999999999995, "grad_norm": 5.25, "learning_rate": 1.9954700565433406e-05, "loss": 1.5507, "step": 2376 }, { "epoch": 5.9425, "grad_norm": 5.75, "learning_rate": 1.9954623551164603e-05, "loss": 1.5599, "step": 2377 }, { "epoch": 5.945, "grad_norm": 6.0, "learning_rate": 1.995454647163369e-05, "loss": 1.522, "step": 2378 }, { "epoch": 5.9475, "grad_norm": 26.375, "learning_rate": 1.9954469326841166e-05, "loss": 1.5637, "step": 2379 }, { "epoch": 5.95, "grad_norm": 6.5625, "learning_rate": 1.995439211678754e-05, "loss": 1.5431, "step": 2380 }, { "epoch": 5.9525, "grad_norm": 5.15625, "learning_rate": 1.9954314841473316e-05, "loss": 1.4962, "step": 2381 }, { "epoch": 5.955, "grad_norm": 4.78125, "learning_rate": 1.9954237500899006e-05, "loss": 1.5824, "step": 2382 }, { "epoch": 5.9575, "grad_norm": 6.03125, "learning_rate": 1.995416009506511e-05, "loss": 1.6668, "step": 2383 }, { "epoch": 5.96, "grad_norm": 9.0625, "learning_rate": 1.9954082623972143e-05, "loss": 1.5807, "step": 2384 }, { "epoch": 5.9625, "grad_norm": 6.65625, "learning_rate": 1.99540050876206e-05, "loss": 1.6197, "step": 2385 }, { "epoch": 5.965, "grad_norm": 7.0, "learning_rate": 1.9953927486011008e-05, "loss": 1.8507, "step": 2386 }, { "epoch": 5.9675, "grad_norm": 5.46875, "learning_rate": 1.995384981914386e-05, "loss": 1.7187, "step": 2387 }, { "epoch": 5.97, "grad_norm": 5.3125, "learning_rate": 1.9953772087019674e-05, "loss": 1.5449, "step": 2388 }, { "epoch": 5.9725, "grad_norm": 6.375, "learning_rate": 1.9953694289638954e-05, "loss": 1.5597, "step": 2389 }, { "epoch": 5.975, "grad_norm": 5.4375, "learning_rate": 1.9953616427002213e-05, "loss": 1.6022, "step": 2390 }, { "epoch": 5.9775, "grad_norm": 5.53125, "learning_rate": 1.995353849910996e-05, "loss": 1.6854, "step": 2391 }, { "epoch": 5.98, "grad_norm": 7.71875, "learning_rate": 1.995346050596271e-05, "loss": 1.6136, "step": 2392 }, { "epoch": 5.9825, "grad_norm": 5.78125, "learning_rate": 1.995338244756097e-05, "loss": 1.6311, "step": 2393 }, { "epoch": 5.985, "grad_norm": 5.8125, "learning_rate": 1.9953304323905252e-05, "loss": 1.5857, "step": 2394 }, { "epoch": 5.9875, "grad_norm": 6.3125, "learning_rate": 1.995322613499607e-05, "loss": 1.6645, "step": 2395 }, { "epoch": 5.99, "grad_norm": 5.875, "learning_rate": 1.9953147880833935e-05, "loss": 1.6277, "step": 2396 }, { "epoch": 5.9925, "grad_norm": 4.28125, "learning_rate": 1.9953069561419362e-05, "loss": 1.5063, "step": 2397 }, { "epoch": 5.995, "grad_norm": 5.03125, "learning_rate": 1.9952991176752864e-05, "loss": 1.4545, "step": 2398 }, { "epoch": 5.9975000000000005, "grad_norm": 5.8125, "learning_rate": 1.995291272683495e-05, "loss": 1.7745, "step": 2399 }, { "epoch": 6.0, "grad_norm": 5.09375, "learning_rate": 1.995283421166614e-05, "loss": 1.5806, "step": 2400 }, { "epoch": 6.0025, "grad_norm": 6.1875, "learning_rate": 1.995275563124695e-05, "loss": 1.7182, "step": 2401 }, { "epoch": 6.005, "grad_norm": 16.625, "learning_rate": 1.995267698557789e-05, "loss": 1.5804, "step": 2402 }, { "epoch": 6.0075, "grad_norm": 7.0, "learning_rate": 1.9952598274659477e-05, "loss": 1.5914, "step": 2403 }, { "epoch": 6.01, "grad_norm": 6.4375, "learning_rate": 1.995251949849223e-05, "loss": 1.6294, "step": 2404 }, { "epoch": 6.0125, "grad_norm": 4.8125, "learning_rate": 1.9952440657076666e-05, "loss": 1.5567, "step": 2405 }, { "epoch": 6.015, "grad_norm": 5.5, "learning_rate": 1.9952361750413297e-05, "loss": 1.4998, "step": 2406 }, { "epoch": 6.0175, "grad_norm": 5.34375, "learning_rate": 1.995228277850264e-05, "loss": 1.6601, "step": 2407 }, { "epoch": 6.02, "grad_norm": 5.40625, "learning_rate": 1.995220374134522e-05, "loss": 1.6874, "step": 2408 }, { "epoch": 6.0225, "grad_norm": 5.0, "learning_rate": 1.9952124638941547e-05, "loss": 1.5459, "step": 2409 }, { "epoch": 6.025, "grad_norm": 6.09375, "learning_rate": 1.995204547129214e-05, "loss": 1.5348, "step": 2410 }, { "epoch": 6.0275, "grad_norm": 6.75, "learning_rate": 1.995196623839753e-05, "loss": 1.5543, "step": 2411 }, { "epoch": 6.03, "grad_norm": 7.4375, "learning_rate": 1.995188694025822e-05, "loss": 1.7017, "step": 2412 }, { "epoch": 6.0325, "grad_norm": 8.0625, "learning_rate": 1.9951807576874743e-05, "loss": 1.8353, "step": 2413 }, { "epoch": 6.035, "grad_norm": 6.6875, "learning_rate": 1.995172814824761e-05, "loss": 1.7617, "step": 2414 }, { "epoch": 6.0375, "grad_norm": 6.03125, "learning_rate": 1.995164865437735e-05, "loss": 1.7715, "step": 2415 }, { "epoch": 6.04, "grad_norm": 8.9375, "learning_rate": 1.9951569095264473e-05, "loss": 1.6316, "step": 2416 }, { "epoch": 6.0425, "grad_norm": 7.53125, "learning_rate": 1.9951489470909512e-05, "loss": 1.5548, "step": 2417 }, { "epoch": 6.045, "grad_norm": 6.5, "learning_rate": 1.995140978131298e-05, "loss": 1.6015, "step": 2418 }, { "epoch": 6.0475, "grad_norm": 4.78125, "learning_rate": 1.9951330026475406e-05, "loss": 1.6844, "step": 2419 }, { "epoch": 6.05, "grad_norm": 4.46875, "learning_rate": 1.995125020639731e-05, "loss": 1.5622, "step": 2420 }, { "epoch": 6.0525, "grad_norm": 8.625, "learning_rate": 1.9951170321079217e-05, "loss": 1.7783, "step": 2421 }, { "epoch": 6.055, "grad_norm": 4.65625, "learning_rate": 1.9951090370521653e-05, "loss": 1.5202, "step": 2422 }, { "epoch": 6.0575, "grad_norm": 7.34375, "learning_rate": 1.9951010354725137e-05, "loss": 1.7911, "step": 2423 }, { "epoch": 6.06, "grad_norm": 4.96875, "learning_rate": 1.995093027369019e-05, "loss": 1.5045, "step": 2424 }, { "epoch": 6.0625, "grad_norm": 5.25, "learning_rate": 1.9950850127417346e-05, "loss": 1.3256, "step": 2425 }, { "epoch": 6.065, "grad_norm": 6.90625, "learning_rate": 1.9950769915907128e-05, "loss": 1.7455, "step": 2426 }, { "epoch": 6.0675, "grad_norm": 10.125, "learning_rate": 1.995068963916006e-05, "loss": 1.7524, "step": 2427 }, { "epoch": 6.07, "grad_norm": 6.84375, "learning_rate": 1.9950609297176668e-05, "loss": 1.9136, "step": 2428 }, { "epoch": 6.0725, "grad_norm": 6.59375, "learning_rate": 1.995052888995748e-05, "loss": 1.6057, "step": 2429 }, { "epoch": 6.075, "grad_norm": 5.25, "learning_rate": 1.995044841750302e-05, "loss": 1.636, "step": 2430 }, { "epoch": 6.0775, "grad_norm": 7.96875, "learning_rate": 1.9950367879813825e-05, "loss": 1.6676, "step": 2431 }, { "epoch": 6.08, "grad_norm": 5.8125, "learning_rate": 1.995028727689041e-05, "loss": 1.6145, "step": 2432 }, { "epoch": 6.0825, "grad_norm": 6.75, "learning_rate": 1.9950206608733313e-05, "loss": 1.649, "step": 2433 }, { "epoch": 6.085, "grad_norm": 18.25, "learning_rate": 1.9950125875343057e-05, "loss": 1.8236, "step": 2434 }, { "epoch": 6.0875, "grad_norm": 6.625, "learning_rate": 1.9950045076720175e-05, "loss": 1.6602, "step": 2435 }, { "epoch": 6.09, "grad_norm": 6.875, "learning_rate": 1.99499642128652e-05, "loss": 1.5713, "step": 2436 }, { "epoch": 6.0925, "grad_norm": 5.5625, "learning_rate": 1.9949883283778653e-05, "loss": 1.6357, "step": 2437 }, { "epoch": 6.095, "grad_norm": 5.625, "learning_rate": 1.994980228946107e-05, "loss": 1.5789, "step": 2438 }, { "epoch": 6.0975, "grad_norm": 6.1875, "learning_rate": 1.9949721229912983e-05, "loss": 1.5324, "step": 2439 }, { "epoch": 6.1, "grad_norm": 5.96875, "learning_rate": 1.994964010513492e-05, "loss": 1.5239, "step": 2440 }, { "epoch": 6.1025, "grad_norm": 8.0625, "learning_rate": 1.9949558915127415e-05, "loss": 1.5533, "step": 2441 }, { "epoch": 6.105, "grad_norm": 7.0, "learning_rate": 1.9949477659891002e-05, "loss": 1.6764, "step": 2442 }, { "epoch": 6.1075, "grad_norm": 5.28125, "learning_rate": 1.994939633942621e-05, "loss": 1.6339, "step": 2443 }, { "epoch": 6.11, "grad_norm": 5.5, "learning_rate": 1.9949314953733573e-05, "loss": 1.6221, "step": 2444 }, { "epoch": 6.1125, "grad_norm": 8.375, "learning_rate": 1.9949233502813627e-05, "loss": 1.6147, "step": 2445 }, { "epoch": 6.115, "grad_norm": 12.0625, "learning_rate": 1.9949151986666903e-05, "loss": 1.5902, "step": 2446 }, { "epoch": 6.1175, "grad_norm": 8.4375, "learning_rate": 1.994907040529394e-05, "loss": 1.5247, "step": 2447 }, { "epoch": 6.12, "grad_norm": 10.375, "learning_rate": 1.9948988758695263e-05, "loss": 1.5483, "step": 2448 }, { "epoch": 6.1225, "grad_norm": 5.8125, "learning_rate": 1.994890704687142e-05, "loss": 1.6892, "step": 2449 }, { "epoch": 6.125, "grad_norm": 4.75, "learning_rate": 1.9948825269822934e-05, "loss": 1.5735, "step": 2450 }, { "epoch": 6.1275, "grad_norm": 5.375, "learning_rate": 1.9948743427550355e-05, "loss": 1.6538, "step": 2451 }, { "epoch": 6.13, "grad_norm": 6.625, "learning_rate": 1.994866152005421e-05, "loss": 1.4615, "step": 2452 }, { "epoch": 6.1325, "grad_norm": 6.71875, "learning_rate": 1.9948579547335035e-05, "loss": 1.772, "step": 2453 }, { "epoch": 6.135, "grad_norm": 5.4375, "learning_rate": 1.9948497509393376e-05, "loss": 1.7363, "step": 2454 }, { "epoch": 6.1375, "grad_norm": 6.0625, "learning_rate": 1.994841540622976e-05, "loss": 1.5122, "step": 2455 }, { "epoch": 6.14, "grad_norm": 6.3125, "learning_rate": 1.994833323784473e-05, "loss": 1.5651, "step": 2456 }, { "epoch": 6.1425, "grad_norm": 3.875, "learning_rate": 1.994825100423883e-05, "loss": 1.5857, "step": 2457 }, { "epoch": 6.145, "grad_norm": 5.78125, "learning_rate": 1.9948168705412595e-05, "loss": 1.5987, "step": 2458 }, { "epoch": 6.1475, "grad_norm": 6.15625, "learning_rate": 1.994808634136656e-05, "loss": 1.5117, "step": 2459 }, { "epoch": 6.15, "grad_norm": 6.625, "learning_rate": 1.9948003912101274e-05, "loss": 1.7787, "step": 2460 }, { "epoch": 6.1525, "grad_norm": 6.21875, "learning_rate": 1.9947921417617265e-05, "loss": 1.6271, "step": 2461 }, { "epoch": 6.155, "grad_norm": 6.5, "learning_rate": 1.9947838857915087e-05, "loss": 1.6583, "step": 2462 }, { "epoch": 6.1575, "grad_norm": 6.09375, "learning_rate": 1.9947756232995276e-05, "loss": 1.641, "step": 2463 }, { "epoch": 6.16, "grad_norm": 8.875, "learning_rate": 1.994767354285837e-05, "loss": 1.5251, "step": 2464 }, { "epoch": 6.1625, "grad_norm": 5.53125, "learning_rate": 1.9947590787504916e-05, "loss": 1.5258, "step": 2465 }, { "epoch": 6.165, "grad_norm": 7.5, "learning_rate": 1.9947507966935453e-05, "loss": 1.5954, "step": 2466 }, { "epoch": 6.1675, "grad_norm": 9.6875, "learning_rate": 1.994742508115053e-05, "loss": 1.5305, "step": 2467 }, { "epoch": 6.17, "grad_norm": 4.84375, "learning_rate": 1.994734213015068e-05, "loss": 1.5778, "step": 2468 }, { "epoch": 6.1725, "grad_norm": 10.3125, "learning_rate": 1.9947259113936457e-05, "loss": 1.6827, "step": 2469 }, { "epoch": 6.175, "grad_norm": 6.09375, "learning_rate": 1.99471760325084e-05, "loss": 1.5567, "step": 2470 }, { "epoch": 6.1775, "grad_norm": 7.71875, "learning_rate": 1.994709288586706e-05, "loss": 1.785, "step": 2471 }, { "epoch": 6.18, "grad_norm": 5.90625, "learning_rate": 1.9947009674012975e-05, "loss": 1.6665, "step": 2472 }, { "epoch": 6.1825, "grad_norm": 7.78125, "learning_rate": 1.9946926396946688e-05, "loss": 1.6536, "step": 2473 }, { "epoch": 6.185, "grad_norm": 6.25, "learning_rate": 1.994684305466875e-05, "loss": 1.6387, "step": 2474 }, { "epoch": 6.1875, "grad_norm": 5.71875, "learning_rate": 1.9946759647179712e-05, "loss": 1.7145, "step": 2475 }, { "epoch": 6.19, "grad_norm": 14.75, "learning_rate": 1.9946676174480115e-05, "loss": 1.6476, "step": 2476 }, { "epoch": 6.1925, "grad_norm": 5.375, "learning_rate": 1.9946592636570503e-05, "loss": 1.7471, "step": 2477 }, { "epoch": 6.195, "grad_norm": 10.375, "learning_rate": 1.994650903345143e-05, "loss": 1.5805, "step": 2478 }, { "epoch": 6.1975, "grad_norm": 8.25, "learning_rate": 1.9946425365123443e-05, "loss": 1.6505, "step": 2479 }, { "epoch": 6.2, "grad_norm": 6.8125, "learning_rate": 1.9946341631587086e-05, "loss": 1.6201, "step": 2480 }, { "epoch": 6.2025, "grad_norm": 6.5, "learning_rate": 1.9946257832842914e-05, "loss": 1.657, "step": 2481 }, { "epoch": 6.205, "grad_norm": 9.5, "learning_rate": 1.9946173968891473e-05, "loss": 1.6006, "step": 2482 }, { "epoch": 6.2075, "grad_norm": 5.1875, "learning_rate": 1.9946090039733314e-05, "loss": 1.5333, "step": 2483 }, { "epoch": 6.21, "grad_norm": 4.90625, "learning_rate": 1.9946006045368987e-05, "loss": 1.6235, "step": 2484 }, { "epoch": 6.2125, "grad_norm": 6.5625, "learning_rate": 1.994592198579904e-05, "loss": 1.7623, "step": 2485 }, { "epoch": 6.215, "grad_norm": 6.71875, "learning_rate": 1.994583786102403e-05, "loss": 1.977, "step": 2486 }, { "epoch": 6.2175, "grad_norm": 7.46875, "learning_rate": 1.9945753671044505e-05, "loss": 1.8593, "step": 2487 }, { "epoch": 6.22, "grad_norm": 7.25, "learning_rate": 1.994566941586101e-05, "loss": 1.6596, "step": 2488 }, { "epoch": 6.2225, "grad_norm": 5.3125, "learning_rate": 1.994558509547411e-05, "loss": 1.3774, "step": 2489 }, { "epoch": 6.225, "grad_norm": 4.125, "learning_rate": 1.9945500709884355e-05, "loss": 1.5641, "step": 2490 }, { "epoch": 6.2275, "grad_norm": 6.09375, "learning_rate": 1.994541625909229e-05, "loss": 1.5881, "step": 2491 }, { "epoch": 6.23, "grad_norm": 8.0625, "learning_rate": 1.9945331743098476e-05, "loss": 1.6555, "step": 2492 }, { "epoch": 6.2325, "grad_norm": 7.53125, "learning_rate": 1.9945247161903465e-05, "loss": 1.6499, "step": 2493 }, { "epoch": 6.235, "grad_norm": 10.0625, "learning_rate": 1.9945162515507812e-05, "loss": 1.776, "step": 2494 }, { "epoch": 6.2375, "grad_norm": 6.875, "learning_rate": 1.994507780391207e-05, "loss": 1.7354, "step": 2495 }, { "epoch": 6.24, "grad_norm": 6.03125, "learning_rate": 1.9944993027116798e-05, "loss": 1.579, "step": 2496 }, { "epoch": 6.2425, "grad_norm": 9.4375, "learning_rate": 1.994490818512255e-05, "loss": 1.7036, "step": 2497 }, { "epoch": 6.245, "grad_norm": 4.21875, "learning_rate": 1.9944823277929876e-05, "loss": 1.5682, "step": 2498 }, { "epoch": 6.2475, "grad_norm": 4.15625, "learning_rate": 1.9944738305539342e-05, "loss": 1.5251, "step": 2499 }, { "epoch": 6.25, "grad_norm": 6.21875, "learning_rate": 1.9944653267951507e-05, "loss": 1.6923, "step": 2500 }, { "epoch": 6.2525, "grad_norm": 5.96875, "learning_rate": 1.9944568165166916e-05, "loss": 1.7069, "step": 2501 }, { "epoch": 6.255, "grad_norm": 7.0625, "learning_rate": 1.9944482997186138e-05, "loss": 1.642, "step": 2502 }, { "epoch": 6.2575, "grad_norm": 8.6875, "learning_rate": 1.9944397764009726e-05, "loss": 1.5466, "step": 2503 }, { "epoch": 6.26, "grad_norm": 6.09375, "learning_rate": 1.9944312465638244e-05, "loss": 1.6203, "step": 2504 }, { "epoch": 6.2625, "grad_norm": 4.59375, "learning_rate": 1.994422710207224e-05, "loss": 1.5984, "step": 2505 }, { "epoch": 6.265, "grad_norm": 5.21875, "learning_rate": 1.9944141673312287e-05, "loss": 1.4255, "step": 2506 }, { "epoch": 6.2675, "grad_norm": 5.25, "learning_rate": 1.9944056179358936e-05, "loss": 1.6772, "step": 2507 }, { "epoch": 6.27, "grad_norm": 6.1875, "learning_rate": 1.9943970620212752e-05, "loss": 1.5071, "step": 2508 }, { "epoch": 6.2725, "grad_norm": 8.5, "learning_rate": 1.9943884995874295e-05, "loss": 1.7711, "step": 2509 }, { "epoch": 6.275, "grad_norm": 6.625, "learning_rate": 1.9943799306344123e-05, "loss": 1.6867, "step": 2510 }, { "epoch": 6.2775, "grad_norm": 14.5625, "learning_rate": 1.9943713551622806e-05, "loss": 1.5581, "step": 2511 }, { "epoch": 6.28, "grad_norm": 7.59375, "learning_rate": 1.9943627731710896e-05, "loss": 1.517, "step": 2512 }, { "epoch": 6.2825, "grad_norm": 5.78125, "learning_rate": 1.9943541846608965e-05, "loss": 1.5178, "step": 2513 }, { "epoch": 6.285, "grad_norm": 7.625, "learning_rate": 1.9943455896317566e-05, "loss": 1.476, "step": 2514 }, { "epoch": 6.2875, "grad_norm": 7.3125, "learning_rate": 1.994336988083727e-05, "loss": 1.4858, "step": 2515 }, { "epoch": 6.29, "grad_norm": 5.5625, "learning_rate": 1.9943283800168643e-05, "loss": 1.5265, "step": 2516 }, { "epoch": 6.2925, "grad_norm": 5.5, "learning_rate": 1.994319765431224e-05, "loss": 1.7281, "step": 2517 }, { "epoch": 6.295, "grad_norm": 7.25, "learning_rate": 1.9943111443268634e-05, "loss": 1.6568, "step": 2518 }, { "epoch": 6.2975, "grad_norm": 7.15625, "learning_rate": 1.9943025167038387e-05, "loss": 1.5828, "step": 2519 }, { "epoch": 6.3, "grad_norm": 4.34375, "learning_rate": 1.9942938825622064e-05, "loss": 1.3875, "step": 2520 }, { "epoch": 6.3025, "grad_norm": 13.6875, "learning_rate": 1.9942852419020233e-05, "loss": 1.6877, "step": 2521 }, { "epoch": 6.305, "grad_norm": 5.90625, "learning_rate": 1.994276594723346e-05, "loss": 1.588, "step": 2522 }, { "epoch": 6.3075, "grad_norm": 6.3125, "learning_rate": 1.994267941026231e-05, "loss": 1.6951, "step": 2523 }, { "epoch": 6.31, "grad_norm": 7.125, "learning_rate": 1.9942592808107354e-05, "loss": 1.5947, "step": 2524 }, { "epoch": 6.3125, "grad_norm": 9.3125, "learning_rate": 1.9942506140769155e-05, "loss": 1.6307, "step": 2525 }, { "epoch": 6.315, "grad_norm": 7.0, "learning_rate": 1.994241940824828e-05, "loss": 1.5962, "step": 2526 }, { "epoch": 6.3175, "grad_norm": 5.875, "learning_rate": 1.9942332610545304e-05, "loss": 1.54, "step": 2527 }, { "epoch": 6.32, "grad_norm": 11.1875, "learning_rate": 1.9942245747660797e-05, "loss": 1.6741, "step": 2528 }, { "epoch": 6.3225, "grad_norm": 7.65625, "learning_rate": 1.9942158819595323e-05, "loss": 1.7052, "step": 2529 }, { "epoch": 6.325, "grad_norm": 6.5625, "learning_rate": 1.994207182634945e-05, "loss": 1.5405, "step": 2530 }, { "epoch": 6.3275, "grad_norm": 7.3125, "learning_rate": 1.9941984767923756e-05, "loss": 1.7027, "step": 2531 }, { "epoch": 6.33, "grad_norm": 8.25, "learning_rate": 1.9941897644318805e-05, "loss": 1.665, "step": 2532 }, { "epoch": 6.3325, "grad_norm": 6.625, "learning_rate": 1.9941810455535173e-05, "loss": 1.5725, "step": 2533 }, { "epoch": 6.335, "grad_norm": 5.25, "learning_rate": 1.994172320157343e-05, "loss": 1.7165, "step": 2534 }, { "epoch": 6.3375, "grad_norm": 8.9375, "learning_rate": 1.994163588243415e-05, "loss": 1.5447, "step": 2535 }, { "epoch": 6.34, "grad_norm": 5.21875, "learning_rate": 1.9941548498117894e-05, "loss": 1.5489, "step": 2536 }, { "epoch": 6.3425, "grad_norm": 3.78125, "learning_rate": 1.9941461048625253e-05, "loss": 1.4943, "step": 2537 }, { "epoch": 6.345, "grad_norm": 6.875, "learning_rate": 1.9941373533956788e-05, "loss": 1.5871, "step": 2538 }, { "epoch": 6.3475, "grad_norm": 5.40625, "learning_rate": 1.9941285954113074e-05, "loss": 1.6278, "step": 2539 }, { "epoch": 6.35, "grad_norm": 4.1875, "learning_rate": 1.994119830909469e-05, "loss": 1.5643, "step": 2540 }, { "epoch": 6.3525, "grad_norm": 7.71875, "learning_rate": 1.9941110598902206e-05, "loss": 1.6705, "step": 2541 }, { "epoch": 6.355, "grad_norm": 7.59375, "learning_rate": 1.9941022823536197e-05, "loss": 1.6124, "step": 2542 }, { "epoch": 6.3575, "grad_norm": 6.1875, "learning_rate": 1.9940934982997243e-05, "loss": 1.712, "step": 2543 }, { "epoch": 6.36, "grad_norm": 8.375, "learning_rate": 1.9940847077285918e-05, "loss": 1.5428, "step": 2544 }, { "epoch": 6.3625, "grad_norm": 5.0625, "learning_rate": 1.9940759106402793e-05, "loss": 1.6726, "step": 2545 }, { "epoch": 6.365, "grad_norm": 15.6875, "learning_rate": 1.9940671070348456e-05, "loss": 1.5268, "step": 2546 }, { "epoch": 6.3675, "grad_norm": 4.875, "learning_rate": 1.994058296912347e-05, "loss": 1.4576, "step": 2547 }, { "epoch": 6.37, "grad_norm": 6.34375, "learning_rate": 1.994049480272842e-05, "loss": 1.652, "step": 2548 }, { "epoch": 6.3725, "grad_norm": 6.96875, "learning_rate": 1.994040657116389e-05, "loss": 1.9771, "step": 2549 }, { "epoch": 6.375, "grad_norm": 8.1875, "learning_rate": 1.994031827443045e-05, "loss": 1.5666, "step": 2550 }, { "epoch": 6.3775, "grad_norm": 7.5, "learning_rate": 1.9940229912528678e-05, "loss": 1.6592, "step": 2551 }, { "epoch": 6.38, "grad_norm": 5.71875, "learning_rate": 1.994014148545916e-05, "loss": 1.5353, "step": 2552 }, { "epoch": 6.3825, "grad_norm": 6.59375, "learning_rate": 1.9940052993222473e-05, "loss": 1.6721, "step": 2553 }, { "epoch": 6.385, "grad_norm": 8.3125, "learning_rate": 1.993996443581919e-05, "loss": 1.5641, "step": 2554 }, { "epoch": 6.3875, "grad_norm": 5.1875, "learning_rate": 1.9939875813249902e-05, "loss": 1.3885, "step": 2555 }, { "epoch": 6.39, "grad_norm": 6.25, "learning_rate": 1.9939787125515188e-05, "loss": 1.6581, "step": 2556 }, { "epoch": 6.3925, "grad_norm": 8.75, "learning_rate": 1.9939698372615627e-05, "loss": 1.624, "step": 2557 }, { "epoch": 6.395, "grad_norm": 9.25, "learning_rate": 1.99396095545518e-05, "loss": 1.5718, "step": 2558 }, { "epoch": 6.3975, "grad_norm": 4.59375, "learning_rate": 1.993952067132429e-05, "loss": 1.6476, "step": 2559 }, { "epoch": 6.4, "grad_norm": 7.1875, "learning_rate": 1.9939431722933678e-05, "loss": 1.5845, "step": 2560 }, { "epoch": 6.4025, "grad_norm": 5.5, "learning_rate": 1.9939342709380555e-05, "loss": 1.6986, "step": 2561 }, { "epoch": 6.405, "grad_norm": 15.125, "learning_rate": 1.9939253630665495e-05, "loss": 1.5778, "step": 2562 }, { "epoch": 6.4075, "grad_norm": 5.59375, "learning_rate": 1.9939164486789085e-05, "loss": 1.6244, "step": 2563 }, { "epoch": 6.41, "grad_norm": 7.75, "learning_rate": 1.9939075277751914e-05, "loss": 1.6796, "step": 2564 }, { "epoch": 6.4125, "grad_norm": 4.65625, "learning_rate": 1.993898600355456e-05, "loss": 1.6133, "step": 2565 }, { "epoch": 6.415, "grad_norm": 7.28125, "learning_rate": 1.9938896664197615e-05, "loss": 1.693, "step": 2566 }, { "epoch": 6.4175, "grad_norm": 6.40625, "learning_rate": 1.9938807259681655e-05, "loss": 1.5283, "step": 2567 }, { "epoch": 6.42, "grad_norm": 9.9375, "learning_rate": 1.993871779000728e-05, "loss": 1.6631, "step": 2568 }, { "epoch": 6.4225, "grad_norm": 5.46875, "learning_rate": 1.9938628255175064e-05, "loss": 1.4977, "step": 2569 }, { "epoch": 6.425, "grad_norm": 5.125, "learning_rate": 1.99385386551856e-05, "loss": 1.6255, "step": 2570 }, { "epoch": 6.4275, "grad_norm": 7.71875, "learning_rate": 1.9938448990039475e-05, "loss": 1.722, "step": 2571 }, { "epoch": 6.43, "grad_norm": 6.0625, "learning_rate": 1.9938359259737273e-05, "loss": 1.5412, "step": 2572 }, { "epoch": 6.4325, "grad_norm": 7.65625, "learning_rate": 1.9938269464279592e-05, "loss": 1.9026, "step": 2573 }, { "epoch": 6.435, "grad_norm": 7.28125, "learning_rate": 1.993817960366701e-05, "loss": 1.7908, "step": 2574 }, { "epoch": 6.4375, "grad_norm": 5.0, "learning_rate": 1.993808967790012e-05, "loss": 1.4305, "step": 2575 }, { "epoch": 6.44, "grad_norm": 11.5, "learning_rate": 1.993799968697951e-05, "loss": 1.6595, "step": 2576 }, { "epoch": 6.4425, "grad_norm": 18.625, "learning_rate": 1.9937909630905774e-05, "loss": 1.6315, "step": 2577 }, { "epoch": 6.445, "grad_norm": 7.65625, "learning_rate": 1.99378195096795e-05, "loss": 1.4081, "step": 2578 }, { "epoch": 6.4475, "grad_norm": 5.84375, "learning_rate": 1.993772932330128e-05, "loss": 1.5576, "step": 2579 }, { "epoch": 6.45, "grad_norm": 7.03125, "learning_rate": 1.9937639071771704e-05, "loss": 1.7545, "step": 2580 }, { "epoch": 6.4525, "grad_norm": 4.8125, "learning_rate": 1.993754875509136e-05, "loss": 1.5502, "step": 2581 }, { "epoch": 6.455, "grad_norm": 8.625, "learning_rate": 1.993745837326085e-05, "loss": 1.6171, "step": 2582 }, { "epoch": 6.4575, "grad_norm": 5.4375, "learning_rate": 1.9937367926280756e-05, "loss": 1.5852, "step": 2583 }, { "epoch": 6.46, "grad_norm": 9.5, "learning_rate": 1.9937277414151677e-05, "loss": 1.7411, "step": 2584 }, { "epoch": 6.4625, "grad_norm": 5.78125, "learning_rate": 1.9937186836874208e-05, "loss": 1.6018, "step": 2585 }, { "epoch": 6.465, "grad_norm": 6.96875, "learning_rate": 1.9937096194448937e-05, "loss": 1.5737, "step": 2586 }, { "epoch": 6.4675, "grad_norm": 5.5, "learning_rate": 1.993700548687646e-05, "loss": 1.4788, "step": 2587 }, { "epoch": 6.47, "grad_norm": 4.84375, "learning_rate": 1.9936914714157375e-05, "loss": 1.5535, "step": 2588 }, { "epoch": 6.4725, "grad_norm": 10.4375, "learning_rate": 1.9936823876292272e-05, "loss": 1.5411, "step": 2589 }, { "epoch": 6.475, "grad_norm": 5.625, "learning_rate": 1.993673297328175e-05, "loss": 1.633, "step": 2590 }, { "epoch": 6.4775, "grad_norm": 5.375, "learning_rate": 1.993664200512641e-05, "loss": 1.6639, "step": 2591 }, { "epoch": 6.48, "grad_norm": 5.84375, "learning_rate": 1.9936550971826835e-05, "loss": 1.5967, "step": 2592 }, { "epoch": 6.4825, "grad_norm": 8.1875, "learning_rate": 1.993645987338363e-05, "loss": 1.5768, "step": 2593 }, { "epoch": 6.485, "grad_norm": 5.34375, "learning_rate": 1.9936368709797394e-05, "loss": 1.6699, "step": 2594 }, { "epoch": 6.4875, "grad_norm": 5.90625, "learning_rate": 1.993627748106872e-05, "loss": 1.5608, "step": 2595 }, { "epoch": 6.49, "grad_norm": 5.53125, "learning_rate": 1.9936186187198214e-05, "loss": 1.495, "step": 2596 }, { "epoch": 6.4925, "grad_norm": 7.5625, "learning_rate": 1.9936094828186465e-05, "loss": 1.6656, "step": 2597 }, { "epoch": 6.495, "grad_norm": 5.71875, "learning_rate": 1.9936003404034077e-05, "loss": 1.7256, "step": 2598 }, { "epoch": 6.4975, "grad_norm": 8.3125, "learning_rate": 1.9935911914741647e-05, "loss": 1.6908, "step": 2599 }, { "epoch": 6.5, "grad_norm": 4.0625, "learning_rate": 1.993582036030978e-05, "loss": 1.5686, "step": 2600 }, { "epoch": 6.5024999999999995, "grad_norm": 6.5625, "learning_rate": 1.9935728740739067e-05, "loss": 1.5727, "step": 2601 }, { "epoch": 6.505, "grad_norm": 5.75, "learning_rate": 1.993563705603012e-05, "loss": 1.6423, "step": 2602 }, { "epoch": 6.5075, "grad_norm": 5.40625, "learning_rate": 1.9935545306183527e-05, "loss": 1.5192, "step": 2603 }, { "epoch": 6.51, "grad_norm": 5.9375, "learning_rate": 1.9935453491199902e-05, "loss": 1.606, "step": 2604 }, { "epoch": 6.5125, "grad_norm": 7.5, "learning_rate": 1.9935361611079837e-05, "loss": 1.6275, "step": 2605 }, { "epoch": 6.515, "grad_norm": 6.875, "learning_rate": 1.9935269665823944e-05, "loss": 1.6322, "step": 2606 }, { "epoch": 6.5175, "grad_norm": 4.34375, "learning_rate": 1.9935177655432816e-05, "loss": 1.5756, "step": 2607 }, { "epoch": 6.52, "grad_norm": 6.34375, "learning_rate": 1.9935085579907064e-05, "loss": 1.6671, "step": 2608 }, { "epoch": 6.5225, "grad_norm": 6.0625, "learning_rate": 1.9934993439247287e-05, "loss": 1.5715, "step": 2609 }, { "epoch": 6.525, "grad_norm": 5.5, "learning_rate": 1.9934901233454092e-05, "loss": 1.5304, "step": 2610 }, { "epoch": 6.5275, "grad_norm": 5.625, "learning_rate": 1.993480896252808e-05, "loss": 1.4555, "step": 2611 }, { "epoch": 6.53, "grad_norm": 7.5625, "learning_rate": 1.993471662646986e-05, "loss": 1.7201, "step": 2612 }, { "epoch": 6.5325, "grad_norm": 5.8125, "learning_rate": 1.9934624225280037e-05, "loss": 1.5005, "step": 2613 }, { "epoch": 6.535, "grad_norm": 7.875, "learning_rate": 1.993453175895921e-05, "loss": 1.5078, "step": 2614 }, { "epoch": 6.5375, "grad_norm": 6.625, "learning_rate": 1.993443922750799e-05, "loss": 1.548, "step": 2615 }, { "epoch": 6.54, "grad_norm": 5.46875, "learning_rate": 1.9934346630926988e-05, "loss": 1.6023, "step": 2616 }, { "epoch": 6.5425, "grad_norm": 4.875, "learning_rate": 1.9934253969216807e-05, "loss": 1.6764, "step": 2617 }, { "epoch": 6.545, "grad_norm": 7.28125, "learning_rate": 1.9934161242378054e-05, "loss": 1.7905, "step": 2618 }, { "epoch": 6.5475, "grad_norm": 5.40625, "learning_rate": 1.9934068450411337e-05, "loss": 1.4885, "step": 2619 }, { "epoch": 6.55, "grad_norm": 5.3125, "learning_rate": 1.9933975593317263e-05, "loss": 1.5758, "step": 2620 }, { "epoch": 6.5525, "grad_norm": 5.96875, "learning_rate": 1.9933882671096445e-05, "loss": 1.7165, "step": 2621 }, { "epoch": 6.555, "grad_norm": 26.625, "learning_rate": 1.993378968374949e-05, "loss": 1.5021, "step": 2622 }, { "epoch": 6.5575, "grad_norm": 11.125, "learning_rate": 1.9933696631277007e-05, "loss": 1.5223, "step": 2623 }, { "epoch": 6.5600000000000005, "grad_norm": 7.59375, "learning_rate": 1.9933603513679604e-05, "loss": 1.5076, "step": 2624 }, { "epoch": 6.5625, "grad_norm": 7.3125, "learning_rate": 1.9933510330957896e-05, "loss": 1.6293, "step": 2625 }, { "epoch": 6.5649999999999995, "grad_norm": 6.6875, "learning_rate": 1.9933417083112493e-05, "loss": 1.6622, "step": 2626 }, { "epoch": 6.5675, "grad_norm": 5.40625, "learning_rate": 1.9933323770144003e-05, "loss": 1.5041, "step": 2627 }, { "epoch": 6.57, "grad_norm": 8.3125, "learning_rate": 1.993323039205304e-05, "loss": 1.6912, "step": 2628 }, { "epoch": 6.5725, "grad_norm": 5.8125, "learning_rate": 1.9933136948840218e-05, "loss": 1.6171, "step": 2629 }, { "epoch": 6.575, "grad_norm": 6.90625, "learning_rate": 1.9933043440506147e-05, "loss": 1.6791, "step": 2630 }, { "epoch": 6.5775, "grad_norm": 14.0625, "learning_rate": 1.9932949867051437e-05, "loss": 1.6417, "step": 2631 }, { "epoch": 6.58, "grad_norm": 5.65625, "learning_rate": 1.9932856228476705e-05, "loss": 1.6064, "step": 2632 }, { "epoch": 6.5825, "grad_norm": 7.65625, "learning_rate": 1.9932762524782572e-05, "loss": 1.9382, "step": 2633 }, { "epoch": 6.585, "grad_norm": 6.03125, "learning_rate": 1.9932668755969644e-05, "loss": 1.656, "step": 2634 }, { "epoch": 6.5875, "grad_norm": 5.59375, "learning_rate": 1.9932574922038533e-05, "loss": 1.6755, "step": 2635 }, { "epoch": 6.59, "grad_norm": 6.96875, "learning_rate": 1.9932481022989857e-05, "loss": 1.5691, "step": 2636 }, { "epoch": 6.5925, "grad_norm": 6.03125, "learning_rate": 1.9932387058824236e-05, "loss": 1.6359, "step": 2637 }, { "epoch": 6.595, "grad_norm": 4.6875, "learning_rate": 1.9932293029542283e-05, "loss": 1.4749, "step": 2638 }, { "epoch": 6.5975, "grad_norm": 6.09375, "learning_rate": 1.9932198935144614e-05, "loss": 1.6157, "step": 2639 }, { "epoch": 6.6, "grad_norm": 7.71875, "learning_rate": 1.9932104775631847e-05, "loss": 1.7058, "step": 2640 }, { "epoch": 6.6025, "grad_norm": 6.0, "learning_rate": 1.9932010551004597e-05, "loss": 1.7331, "step": 2641 }, { "epoch": 6.605, "grad_norm": 6.09375, "learning_rate": 1.9931916261263486e-05, "loss": 1.5255, "step": 2642 }, { "epoch": 6.6075, "grad_norm": 7.75, "learning_rate": 1.9931821906409125e-05, "loss": 1.6556, "step": 2643 }, { "epoch": 6.61, "grad_norm": 6.71875, "learning_rate": 1.993172748644214e-05, "loss": 1.6415, "step": 2644 }, { "epoch": 6.6125, "grad_norm": 5.5, "learning_rate": 1.9931633001363148e-05, "loss": 1.6693, "step": 2645 }, { "epoch": 6.615, "grad_norm": 7.1875, "learning_rate": 1.9931538451172768e-05, "loss": 1.7422, "step": 2646 }, { "epoch": 6.6175, "grad_norm": 6.90625, "learning_rate": 1.9931443835871616e-05, "loss": 1.631, "step": 2647 }, { "epoch": 6.62, "grad_norm": 6.65625, "learning_rate": 1.9931349155460315e-05, "loss": 1.5837, "step": 2648 }, { "epoch": 6.6225000000000005, "grad_norm": 6.78125, "learning_rate": 1.9931254409939492e-05, "loss": 1.5759, "step": 2649 }, { "epoch": 6.625, "grad_norm": 6.28125, "learning_rate": 1.9931159599309757e-05, "loss": 1.8748, "step": 2650 }, { "epoch": 6.6274999999999995, "grad_norm": 4.90625, "learning_rate": 1.993106472357174e-05, "loss": 1.6551, "step": 2651 }, { "epoch": 6.63, "grad_norm": 6.15625, "learning_rate": 1.9930969782726058e-05, "loss": 1.6421, "step": 2652 }, { "epoch": 6.6325, "grad_norm": 7.96875, "learning_rate": 1.9930874776773338e-05, "loss": 1.662, "step": 2653 }, { "epoch": 6.635, "grad_norm": 5.59375, "learning_rate": 1.99307797057142e-05, "loss": 1.4222, "step": 2654 }, { "epoch": 6.6375, "grad_norm": 6.1875, "learning_rate": 1.9930684569549265e-05, "loss": 1.7082, "step": 2655 }, { "epoch": 6.64, "grad_norm": 6.0, "learning_rate": 1.993058936827916e-05, "loss": 1.6396, "step": 2656 }, { "epoch": 6.6425, "grad_norm": 7.75, "learning_rate": 1.9930494101904513e-05, "loss": 1.5576, "step": 2657 }, { "epoch": 6.645, "grad_norm": 6.96875, "learning_rate": 1.993039877042594e-05, "loss": 1.6254, "step": 2658 }, { "epoch": 6.6475, "grad_norm": 5.375, "learning_rate": 1.993030337384407e-05, "loss": 1.6171, "step": 2659 }, { "epoch": 6.65, "grad_norm": 6.65625, "learning_rate": 1.993020791215953e-05, "loss": 1.4852, "step": 2660 }, { "epoch": 6.6525, "grad_norm": 4.84375, "learning_rate": 1.993011238537294e-05, "loss": 1.673, "step": 2661 }, { "epoch": 6.655, "grad_norm": 4.8125, "learning_rate": 1.9930016793484935e-05, "loss": 1.6644, "step": 2662 }, { "epoch": 6.6575, "grad_norm": 5.75, "learning_rate": 1.9929921136496135e-05, "loss": 1.4732, "step": 2663 }, { "epoch": 6.66, "grad_norm": 5.71875, "learning_rate": 1.9929825414407175e-05, "loss": 1.6291, "step": 2664 }, { "epoch": 6.6625, "grad_norm": 5.375, "learning_rate": 1.992972962721867e-05, "loss": 1.6235, "step": 2665 }, { "epoch": 6.665, "grad_norm": 8.875, "learning_rate": 1.992963377493126e-05, "loss": 1.6839, "step": 2666 }, { "epoch": 6.6675, "grad_norm": 5.84375, "learning_rate": 1.9929537857545562e-05, "loss": 1.5602, "step": 2667 }, { "epoch": 6.67, "grad_norm": 5.65625, "learning_rate": 1.9929441875062218e-05, "loss": 1.7288, "step": 2668 }, { "epoch": 6.6725, "grad_norm": 4.875, "learning_rate": 1.9929345827481847e-05, "loss": 1.4522, "step": 2669 }, { "epoch": 6.675, "grad_norm": 8.1875, "learning_rate": 1.992924971480508e-05, "loss": 1.6058, "step": 2670 }, { "epoch": 6.6775, "grad_norm": 6.09375, "learning_rate": 1.9929153537032553e-05, "loss": 1.4995, "step": 2671 }, { "epoch": 6.68, "grad_norm": 8.75, "learning_rate": 1.9929057294164894e-05, "loss": 1.5933, "step": 2672 }, { "epoch": 6.6825, "grad_norm": 6.5625, "learning_rate": 1.992896098620273e-05, "loss": 1.5524, "step": 2673 }, { "epoch": 6.6850000000000005, "grad_norm": 6.375, "learning_rate": 1.9928864613146696e-05, "loss": 1.8097, "step": 2674 }, { "epoch": 6.6875, "grad_norm": 5.15625, "learning_rate": 1.992876817499742e-05, "loss": 1.5605, "step": 2675 }, { "epoch": 6.6899999999999995, "grad_norm": 5.71875, "learning_rate": 1.992867167175554e-05, "loss": 1.4835, "step": 2676 }, { "epoch": 6.6925, "grad_norm": 11.75, "learning_rate": 1.9928575103421685e-05, "loss": 1.5917, "step": 2677 }, { "epoch": 6.695, "grad_norm": 5.15625, "learning_rate": 1.992847846999649e-05, "loss": 1.7284, "step": 2678 }, { "epoch": 6.6975, "grad_norm": 5.78125, "learning_rate": 1.992838177148059e-05, "loss": 1.7771, "step": 2679 }, { "epoch": 6.7, "grad_norm": 6.5625, "learning_rate": 1.992828500787461e-05, "loss": 1.6904, "step": 2680 }, { "epoch": 6.7025, "grad_norm": 4.65625, "learning_rate": 1.9928188179179195e-05, "loss": 1.4857, "step": 2681 }, { "epoch": 6.705, "grad_norm": 6.46875, "learning_rate": 1.9928091285394973e-05, "loss": 1.5137, "step": 2682 }, { "epoch": 6.7075, "grad_norm": 6.9375, "learning_rate": 1.9927994326522585e-05, "loss": 1.5994, "step": 2683 }, { "epoch": 6.71, "grad_norm": 5.3125, "learning_rate": 1.992789730256266e-05, "loss": 1.5608, "step": 2684 }, { "epoch": 6.7125, "grad_norm": 7.40625, "learning_rate": 1.992780021351584e-05, "loss": 1.721, "step": 2685 }, { "epoch": 6.715, "grad_norm": 6.1875, "learning_rate": 1.992770305938276e-05, "loss": 1.5517, "step": 2686 }, { "epoch": 6.7175, "grad_norm": 6.625, "learning_rate": 1.992760584016405e-05, "loss": 1.5537, "step": 2687 }, { "epoch": 6.72, "grad_norm": 12.0, "learning_rate": 1.992750855586036e-05, "loss": 1.5353, "step": 2688 }, { "epoch": 6.7225, "grad_norm": 6.0, "learning_rate": 1.992741120647232e-05, "loss": 1.7271, "step": 2689 }, { "epoch": 6.725, "grad_norm": 5.09375, "learning_rate": 1.992731379200057e-05, "loss": 1.54, "step": 2690 }, { "epoch": 6.7275, "grad_norm": 6.5, "learning_rate": 1.9927216312445746e-05, "loss": 1.5479, "step": 2691 }, { "epoch": 6.73, "grad_norm": 6.25, "learning_rate": 1.992711876780849e-05, "loss": 1.5461, "step": 2692 }, { "epoch": 6.7325, "grad_norm": 7.96875, "learning_rate": 1.9927021158089438e-05, "loss": 1.5863, "step": 2693 }, { "epoch": 6.735, "grad_norm": 6.375, "learning_rate": 1.9926923483289238e-05, "loss": 1.4784, "step": 2694 }, { "epoch": 6.7375, "grad_norm": 7.25, "learning_rate": 1.992682574340852e-05, "loss": 1.6237, "step": 2695 }, { "epoch": 6.74, "grad_norm": 4.84375, "learning_rate": 1.9926727938447935e-05, "loss": 1.4897, "step": 2696 }, { "epoch": 6.7425, "grad_norm": 6.375, "learning_rate": 1.9926630068408114e-05, "loss": 1.6789, "step": 2697 }, { "epoch": 6.745, "grad_norm": 5.375, "learning_rate": 1.9926532133289705e-05, "loss": 1.5008, "step": 2698 }, { "epoch": 6.7475000000000005, "grad_norm": 9.4375, "learning_rate": 1.9926434133093348e-05, "loss": 1.4463, "step": 2699 }, { "epoch": 6.75, "grad_norm": 4.71875, "learning_rate": 1.9926336067819686e-05, "loss": 1.4644, "step": 2700 }, { "epoch": 6.7524999999999995, "grad_norm": 5.4375, "learning_rate": 1.9926237937469364e-05, "loss": 1.6008, "step": 2701 }, { "epoch": 6.755, "grad_norm": 6.8125, "learning_rate": 1.992613974204302e-05, "loss": 1.5444, "step": 2702 }, { "epoch": 6.7575, "grad_norm": 6.53125, "learning_rate": 1.9926041481541304e-05, "loss": 1.7017, "step": 2703 }, { "epoch": 6.76, "grad_norm": 5.25, "learning_rate": 1.9925943155964857e-05, "loss": 1.7438, "step": 2704 }, { "epoch": 6.7625, "grad_norm": 6.09375, "learning_rate": 1.992584476531432e-05, "loss": 1.6918, "step": 2705 }, { "epoch": 6.765, "grad_norm": 7.0625, "learning_rate": 1.992574630959035e-05, "loss": 1.6213, "step": 2706 }, { "epoch": 6.7675, "grad_norm": 5.25, "learning_rate": 1.9925647788793577e-05, "loss": 1.444, "step": 2707 }, { "epoch": 6.77, "grad_norm": 7.125, "learning_rate": 1.9925549202924657e-05, "loss": 1.6724, "step": 2708 }, { "epoch": 6.7725, "grad_norm": 5.1875, "learning_rate": 1.992545055198423e-05, "loss": 1.5918, "step": 2709 }, { "epoch": 6.775, "grad_norm": 7.59375, "learning_rate": 1.992535183597295e-05, "loss": 1.7591, "step": 2710 }, { "epoch": 6.7775, "grad_norm": 7.71875, "learning_rate": 1.9925253054891464e-05, "loss": 1.7575, "step": 2711 }, { "epoch": 6.78, "grad_norm": 6.625, "learning_rate": 1.9925154208740412e-05, "loss": 1.6363, "step": 2712 }, { "epoch": 6.7825, "grad_norm": 6.40625, "learning_rate": 1.9925055297520445e-05, "loss": 1.6346, "step": 2713 }, { "epoch": 6.785, "grad_norm": 4.4375, "learning_rate": 1.9924956321232213e-05, "loss": 1.4208, "step": 2714 }, { "epoch": 6.7875, "grad_norm": 7.53125, "learning_rate": 1.9924857279876368e-05, "loss": 1.4971, "step": 2715 }, { "epoch": 6.79, "grad_norm": 6.34375, "learning_rate": 1.9924758173453555e-05, "loss": 1.6568, "step": 2716 }, { "epoch": 6.7925, "grad_norm": 4.28125, "learning_rate": 1.9924659001964422e-05, "loss": 1.5706, "step": 2717 }, { "epoch": 6.795, "grad_norm": 7.53125, "learning_rate": 1.992455976540962e-05, "loss": 1.6531, "step": 2718 }, { "epoch": 6.7975, "grad_norm": 6.8125, "learning_rate": 1.9924460463789806e-05, "loss": 1.4057, "step": 2719 }, { "epoch": 6.8, "grad_norm": 6.375, "learning_rate": 1.9924361097105624e-05, "loss": 1.6633, "step": 2720 }, { "epoch": 6.8025, "grad_norm": 5.25, "learning_rate": 1.9924261665357728e-05, "loss": 1.5249, "step": 2721 }, { "epoch": 6.805, "grad_norm": 5.875, "learning_rate": 1.992416216854677e-05, "loss": 1.5782, "step": 2722 }, { "epoch": 6.8075, "grad_norm": 9.0, "learning_rate": 1.9924062606673404e-05, "loss": 1.8882, "step": 2723 }, { "epoch": 6.8100000000000005, "grad_norm": 6.25, "learning_rate": 1.9923962979738277e-05, "loss": 1.5877, "step": 2724 }, { "epoch": 6.8125, "grad_norm": 5.5, "learning_rate": 1.9923863287742045e-05, "loss": 1.7006, "step": 2725 }, { "epoch": 6.8149999999999995, "grad_norm": 4.8125, "learning_rate": 1.992376353068537e-05, "loss": 1.6055, "step": 2726 }, { "epoch": 6.8175, "grad_norm": 5.53125, "learning_rate": 1.992366370856889e-05, "loss": 1.6218, "step": 2727 }, { "epoch": 6.82, "grad_norm": 4.65625, "learning_rate": 1.9923563821393272e-05, "loss": 1.6274, "step": 2728 }, { "epoch": 6.8225, "grad_norm": 5.6875, "learning_rate": 1.9923463869159168e-05, "loss": 1.5916, "step": 2729 }, { "epoch": 6.825, "grad_norm": 5.71875, "learning_rate": 1.992336385186723e-05, "loss": 1.6631, "step": 2730 }, { "epoch": 6.8275, "grad_norm": 6.3125, "learning_rate": 1.9923263769518114e-05, "loss": 1.7524, "step": 2731 }, { "epoch": 6.83, "grad_norm": 5.59375, "learning_rate": 1.992316362211248e-05, "loss": 1.7034, "step": 2732 }, { "epoch": 6.8325, "grad_norm": 7.4375, "learning_rate": 1.9923063409650983e-05, "loss": 1.5816, "step": 2733 }, { "epoch": 6.835, "grad_norm": 6.375, "learning_rate": 1.9922963132134278e-05, "loss": 1.609, "step": 2734 }, { "epoch": 6.8375, "grad_norm": 7.46875, "learning_rate": 1.9922862789563024e-05, "loss": 1.7994, "step": 2735 }, { "epoch": 6.84, "grad_norm": 7.0625, "learning_rate": 1.992276238193788e-05, "loss": 1.5816, "step": 2736 }, { "epoch": 6.8425, "grad_norm": 6.65625, "learning_rate": 1.99226619092595e-05, "loss": 1.726, "step": 2737 }, { "epoch": 6.845, "grad_norm": 6.34375, "learning_rate": 1.9922561371528547e-05, "loss": 1.5998, "step": 2738 }, { "epoch": 6.8475, "grad_norm": 3.71875, "learning_rate": 1.9922460768745678e-05, "loss": 1.4834, "step": 2739 }, { "epoch": 6.85, "grad_norm": 5.8125, "learning_rate": 1.9922360100911553e-05, "loss": 1.7592, "step": 2740 }, { "epoch": 6.8525, "grad_norm": 7.1875, "learning_rate": 1.9922259368026836e-05, "loss": 1.7497, "step": 2741 }, { "epoch": 6.855, "grad_norm": 5.53125, "learning_rate": 1.992215857009218e-05, "loss": 1.4957, "step": 2742 }, { "epoch": 6.8575, "grad_norm": 8.875, "learning_rate": 1.992205770710825e-05, "loss": 1.4883, "step": 2743 }, { "epoch": 6.86, "grad_norm": 13.6875, "learning_rate": 1.9921956779075707e-05, "loss": 1.5049, "step": 2744 }, { "epoch": 6.8625, "grad_norm": 6.46875, "learning_rate": 1.9921855785995214e-05, "loss": 1.5146, "step": 2745 }, { "epoch": 6.865, "grad_norm": 9.5, "learning_rate": 1.992175472786743e-05, "loss": 1.7485, "step": 2746 }, { "epoch": 6.8675, "grad_norm": 5.03125, "learning_rate": 1.992165360469302e-05, "loss": 1.5824, "step": 2747 }, { "epoch": 6.87, "grad_norm": 5.46875, "learning_rate": 1.9921552416472644e-05, "loss": 1.6251, "step": 2748 }, { "epoch": 6.8725000000000005, "grad_norm": 7.40625, "learning_rate": 1.992145116320697e-05, "loss": 1.7992, "step": 2749 }, { "epoch": 6.875, "grad_norm": 5.59375, "learning_rate": 1.9921349844896655e-05, "loss": 1.6995, "step": 2750 }, { "epoch": 6.8774999999999995, "grad_norm": 5.71875, "learning_rate": 1.9921248461542374e-05, "loss": 1.646, "step": 2751 }, { "epoch": 6.88, "grad_norm": 3.6875, "learning_rate": 1.9921147013144782e-05, "loss": 1.4136, "step": 2752 }, { "epoch": 6.8825, "grad_norm": 4.34375, "learning_rate": 1.9921045499704547e-05, "loss": 1.7297, "step": 2753 }, { "epoch": 6.885, "grad_norm": 4.5625, "learning_rate": 1.9920943921222332e-05, "loss": 1.3985, "step": 2754 }, { "epoch": 6.8875, "grad_norm": 6.21875, "learning_rate": 1.992084227769881e-05, "loss": 1.5379, "step": 2755 }, { "epoch": 6.89, "grad_norm": 5.78125, "learning_rate": 1.992074056913464e-05, "loss": 1.7466, "step": 2756 }, { "epoch": 6.8925, "grad_norm": 7.0, "learning_rate": 1.9920638795530492e-05, "loss": 1.6517, "step": 2757 }, { "epoch": 6.895, "grad_norm": 6.46875, "learning_rate": 1.9920536956887034e-05, "loss": 1.8067, "step": 2758 }, { "epoch": 6.8975, "grad_norm": 8.3125, "learning_rate": 1.9920435053204934e-05, "loss": 1.5284, "step": 2759 }, { "epoch": 6.9, "grad_norm": 7.3125, "learning_rate": 1.992033308448486e-05, "loss": 1.7051, "step": 2760 }, { "epoch": 6.9025, "grad_norm": 6.78125, "learning_rate": 1.9920231050727473e-05, "loss": 1.5973, "step": 2761 }, { "epoch": 6.905, "grad_norm": 4.625, "learning_rate": 1.9920128951933454e-05, "loss": 1.5205, "step": 2762 }, { "epoch": 6.9075, "grad_norm": 5.25, "learning_rate": 1.9920026788103463e-05, "loss": 1.6846, "step": 2763 }, { "epoch": 6.91, "grad_norm": 5.4375, "learning_rate": 1.9919924559238173e-05, "loss": 1.5615, "step": 2764 }, { "epoch": 6.9125, "grad_norm": 5.3125, "learning_rate": 1.9919822265338257e-05, "loss": 1.6499, "step": 2765 }, { "epoch": 6.915, "grad_norm": 7.03125, "learning_rate": 1.9919719906404384e-05, "loss": 1.6777, "step": 2766 }, { "epoch": 6.9175, "grad_norm": 10.375, "learning_rate": 1.9919617482437222e-05, "loss": 1.6779, "step": 2767 }, { "epoch": 6.92, "grad_norm": 5.625, "learning_rate": 1.9919514993437445e-05, "loss": 1.76, "step": 2768 }, { "epoch": 6.9225, "grad_norm": 7.03125, "learning_rate": 1.9919412439405725e-05, "loss": 1.8519, "step": 2769 }, { "epoch": 6.925, "grad_norm": 8.3125, "learning_rate": 1.9919309820342735e-05, "loss": 1.7695, "step": 2770 }, { "epoch": 6.9275, "grad_norm": 6.09375, "learning_rate": 1.991920713624915e-05, "loss": 1.714, "step": 2771 }, { "epoch": 6.93, "grad_norm": 5.59375, "learning_rate": 1.9919104387125632e-05, "loss": 1.4481, "step": 2772 }, { "epoch": 6.9325, "grad_norm": 7.09375, "learning_rate": 1.991900157297287e-05, "loss": 1.5701, "step": 2773 }, { "epoch": 6.9350000000000005, "grad_norm": 6.46875, "learning_rate": 1.9918898693791525e-05, "loss": 1.573, "step": 2774 }, { "epoch": 6.9375, "grad_norm": 7.25, "learning_rate": 1.9918795749582277e-05, "loss": 1.4938, "step": 2775 }, { "epoch": 6.9399999999999995, "grad_norm": 6.0625, "learning_rate": 1.9918692740345804e-05, "loss": 1.7537, "step": 2776 }, { "epoch": 6.9425, "grad_norm": 6.71875, "learning_rate": 1.991858966608278e-05, "loss": 1.7087, "step": 2777 }, { "epoch": 6.945, "grad_norm": 5.34375, "learning_rate": 1.9918486526793877e-05, "loss": 1.6685, "step": 2778 }, { "epoch": 6.9475, "grad_norm": 5.0625, "learning_rate": 1.9918383322479774e-05, "loss": 1.7376, "step": 2779 }, { "epoch": 6.95, "grad_norm": 5.15625, "learning_rate": 1.9918280053141144e-05, "loss": 1.5171, "step": 2780 }, { "epoch": 6.9525, "grad_norm": 6.59375, "learning_rate": 1.991817671877867e-05, "loss": 1.5944, "step": 2781 }, { "epoch": 6.955, "grad_norm": 9.8125, "learning_rate": 1.9918073319393025e-05, "loss": 1.6743, "step": 2782 }, { "epoch": 6.9575, "grad_norm": 7.78125, "learning_rate": 1.9917969854984886e-05, "loss": 1.6242, "step": 2783 }, { "epoch": 6.96, "grad_norm": 6.5625, "learning_rate": 1.9917866325554936e-05, "loss": 1.5518, "step": 2784 }, { "epoch": 6.9625, "grad_norm": 6.71875, "learning_rate": 1.9917762731103854e-05, "loss": 1.6275, "step": 2785 }, { "epoch": 6.965, "grad_norm": 5.84375, "learning_rate": 1.991765907163231e-05, "loss": 1.5851, "step": 2786 }, { "epoch": 6.9675, "grad_norm": 6.6875, "learning_rate": 1.9917555347140994e-05, "loss": 1.5964, "step": 2787 }, { "epoch": 6.97, "grad_norm": 4.9375, "learning_rate": 1.991745155763058e-05, "loss": 1.5584, "step": 2788 }, { "epoch": 6.9725, "grad_norm": 6.6875, "learning_rate": 1.9917347703101756e-05, "loss": 1.7541, "step": 2789 }, { "epoch": 6.975, "grad_norm": 6.625, "learning_rate": 1.991724378355519e-05, "loss": 1.7905, "step": 2790 }, { "epoch": 6.9775, "grad_norm": 5.90625, "learning_rate": 1.9917139798991572e-05, "loss": 1.6978, "step": 2791 }, { "epoch": 6.98, "grad_norm": 6.21875, "learning_rate": 1.9917035749411585e-05, "loss": 1.4313, "step": 2792 }, { "epoch": 6.9825, "grad_norm": 8.375, "learning_rate": 1.991693163481591e-05, "loss": 1.6391, "step": 2793 }, { "epoch": 6.985, "grad_norm": 6.96875, "learning_rate": 1.991682745520523e-05, "loss": 1.6466, "step": 2794 }, { "epoch": 6.9875, "grad_norm": 8.0625, "learning_rate": 1.991672321058022e-05, "loss": 1.5696, "step": 2795 }, { "epoch": 6.99, "grad_norm": 5.15625, "learning_rate": 1.9916618900941567e-05, "loss": 1.3906, "step": 2796 }, { "epoch": 6.9925, "grad_norm": 7.625, "learning_rate": 1.9916514526289962e-05, "loss": 1.5615, "step": 2797 }, { "epoch": 6.995, "grad_norm": 7.0, "learning_rate": 1.9916410086626085e-05, "loss": 1.8294, "step": 2798 }, { "epoch": 6.9975000000000005, "grad_norm": 5.78125, "learning_rate": 1.991630558195062e-05, "loss": 1.7885, "step": 2799 }, { "epoch": 7.0, "grad_norm": 6.375, "learning_rate": 1.9916201012264255e-05, "loss": 1.7278, "step": 2800 }, { "epoch": 7.0025, "grad_norm": 5.4375, "learning_rate": 1.9916096377567667e-05, "loss": 1.6894, "step": 2801 }, { "epoch": 7.005, "grad_norm": 5.09375, "learning_rate": 1.991599167786155e-05, "loss": 1.7273, "step": 2802 }, { "epoch": 7.0075, "grad_norm": 6.8125, "learning_rate": 1.991588691314659e-05, "loss": 1.4862, "step": 2803 }, { "epoch": 7.01, "grad_norm": 6.6875, "learning_rate": 1.9915782083423473e-05, "loss": 1.7667, "step": 2804 }, { "epoch": 7.0125, "grad_norm": 5.71875, "learning_rate": 1.9915677188692883e-05, "loss": 1.6905, "step": 2805 }, { "epoch": 7.015, "grad_norm": 8.9375, "learning_rate": 1.9915572228955512e-05, "loss": 1.6212, "step": 2806 }, { "epoch": 7.0175, "grad_norm": 7.3125, "learning_rate": 1.9915467204212046e-05, "loss": 1.7642, "step": 2807 }, { "epoch": 7.02, "grad_norm": 8.9375, "learning_rate": 1.9915362114463172e-05, "loss": 1.5815, "step": 2808 }, { "epoch": 7.0225, "grad_norm": 5.28125, "learning_rate": 1.991525695970958e-05, "loss": 1.4415, "step": 2809 }, { "epoch": 7.025, "grad_norm": 7.5, "learning_rate": 1.9915151739951965e-05, "loss": 1.5293, "step": 2810 }, { "epoch": 7.0275, "grad_norm": 5.875, "learning_rate": 1.991504645519101e-05, "loss": 1.647, "step": 2811 }, { "epoch": 7.03, "grad_norm": 6.09375, "learning_rate": 1.9914941105427403e-05, "loss": 1.672, "step": 2812 }, { "epoch": 7.0325, "grad_norm": 6.28125, "learning_rate": 1.9914835690661846e-05, "loss": 1.5907, "step": 2813 }, { "epoch": 7.035, "grad_norm": 7.03125, "learning_rate": 1.991473021089502e-05, "loss": 1.627, "step": 2814 }, { "epoch": 7.0375, "grad_norm": 6.1875, "learning_rate": 1.9914624666127618e-05, "loss": 1.5824, "step": 2815 }, { "epoch": 7.04, "grad_norm": 4.46875, "learning_rate": 1.991451905636033e-05, "loss": 1.7197, "step": 2816 }, { "epoch": 7.0425, "grad_norm": 10.125, "learning_rate": 1.9914413381593858e-05, "loss": 1.6267, "step": 2817 }, { "epoch": 7.045, "grad_norm": 7.09375, "learning_rate": 1.9914307641828888e-05, "loss": 1.7384, "step": 2818 }, { "epoch": 7.0475, "grad_norm": 5.4375, "learning_rate": 1.9914201837066113e-05, "loss": 1.5632, "step": 2819 }, { "epoch": 7.05, "grad_norm": 7.0, "learning_rate": 1.9914095967306224e-05, "loss": 1.5583, "step": 2820 }, { "epoch": 7.0525, "grad_norm": 4.28125, "learning_rate": 1.9913990032549923e-05, "loss": 1.4964, "step": 2821 }, { "epoch": 7.055, "grad_norm": 7.125, "learning_rate": 1.99138840327979e-05, "loss": 1.6479, "step": 2822 }, { "epoch": 7.0575, "grad_norm": 6.21875, "learning_rate": 1.9913777968050848e-05, "loss": 1.6837, "step": 2823 }, { "epoch": 7.06, "grad_norm": 5.46875, "learning_rate": 1.9913671838309466e-05, "loss": 1.4943, "step": 2824 }, { "epoch": 7.0625, "grad_norm": 6.9375, "learning_rate": 1.9913565643574447e-05, "loss": 1.5568, "step": 2825 }, { "epoch": 7.065, "grad_norm": 5.78125, "learning_rate": 1.9913459383846488e-05, "loss": 1.4153, "step": 2826 }, { "epoch": 7.0675, "grad_norm": 5.78125, "learning_rate": 1.9913353059126287e-05, "loss": 1.5335, "step": 2827 }, { "epoch": 7.07, "grad_norm": 5.3125, "learning_rate": 1.991324666941454e-05, "loss": 1.5921, "step": 2828 }, { "epoch": 7.0725, "grad_norm": 5.15625, "learning_rate": 1.9913140214711943e-05, "loss": 1.534, "step": 2829 }, { "epoch": 7.075, "grad_norm": 7.4375, "learning_rate": 1.9913033695019194e-05, "loss": 1.5469, "step": 2830 }, { "epoch": 7.0775, "grad_norm": 5.71875, "learning_rate": 1.9912927110336996e-05, "loss": 1.6463, "step": 2831 }, { "epoch": 7.08, "grad_norm": 7.40625, "learning_rate": 1.9912820460666046e-05, "loss": 1.6036, "step": 2832 }, { "epoch": 7.0825, "grad_norm": 6.0625, "learning_rate": 1.991271374600704e-05, "loss": 1.6609, "step": 2833 }, { "epoch": 7.085, "grad_norm": 4.3125, "learning_rate": 1.9912606966360678e-05, "loss": 1.5196, "step": 2834 }, { "epoch": 7.0875, "grad_norm": 4.71875, "learning_rate": 1.991250012172766e-05, "loss": 1.4651, "step": 2835 }, { "epoch": 7.09, "grad_norm": 7.625, "learning_rate": 1.9912393212108692e-05, "loss": 1.6849, "step": 2836 }, { "epoch": 7.0925, "grad_norm": 8.9375, "learning_rate": 1.991228623750447e-05, "loss": 1.5408, "step": 2837 }, { "epoch": 7.095, "grad_norm": 8.5625, "learning_rate": 1.9912179197915694e-05, "loss": 1.5479, "step": 2838 }, { "epoch": 7.0975, "grad_norm": 7.0, "learning_rate": 1.991207209334307e-05, "loss": 1.7358, "step": 2839 }, { "epoch": 7.1, "grad_norm": 7.46875, "learning_rate": 1.9911964923787295e-05, "loss": 1.5706, "step": 2840 }, { "epoch": 7.1025, "grad_norm": 6.875, "learning_rate": 1.9911857689249077e-05, "loss": 1.6636, "step": 2841 }, { "epoch": 7.105, "grad_norm": 5.75, "learning_rate": 1.991175038972912e-05, "loss": 1.495, "step": 2842 }, { "epoch": 7.1075, "grad_norm": 7.78125, "learning_rate": 1.991164302522812e-05, "loss": 1.5992, "step": 2843 }, { "epoch": 7.11, "grad_norm": 6.6875, "learning_rate": 1.9911535595746783e-05, "loss": 1.7051, "step": 2844 }, { "epoch": 7.1125, "grad_norm": 7.15625, "learning_rate": 1.991142810128582e-05, "loss": 1.5891, "step": 2845 }, { "epoch": 7.115, "grad_norm": 9.3125, "learning_rate": 1.9911320541845928e-05, "loss": 1.6763, "step": 2846 }, { "epoch": 7.1175, "grad_norm": 4.71875, "learning_rate": 1.9911212917427812e-05, "loss": 1.4976, "step": 2847 }, { "epoch": 7.12, "grad_norm": 6.09375, "learning_rate": 1.9911105228032186e-05, "loss": 1.5176, "step": 2848 }, { "epoch": 7.1225, "grad_norm": 5.75, "learning_rate": 1.991099747365975e-05, "loss": 1.5248, "step": 2849 }, { "epoch": 7.125, "grad_norm": 4.65625, "learning_rate": 1.991088965431121e-05, "loss": 1.3067, "step": 2850 }, { "epoch": 7.1275, "grad_norm": 4.65625, "learning_rate": 1.991078176998727e-05, "loss": 1.3957, "step": 2851 }, { "epoch": 7.13, "grad_norm": 7.09375, "learning_rate": 1.9910673820688652e-05, "loss": 1.6308, "step": 2852 }, { "epoch": 7.1325, "grad_norm": 8.1875, "learning_rate": 1.9910565806416043e-05, "loss": 1.746, "step": 2853 }, { "epoch": 7.135, "grad_norm": 5.5625, "learning_rate": 1.9910457727170166e-05, "loss": 1.5986, "step": 2854 }, { "epoch": 7.1375, "grad_norm": 6.1875, "learning_rate": 1.9910349582951726e-05, "loss": 1.6655, "step": 2855 }, { "epoch": 7.14, "grad_norm": 8.8125, "learning_rate": 1.9910241373761426e-05, "loss": 1.5553, "step": 2856 }, { "epoch": 7.1425, "grad_norm": 4.875, "learning_rate": 1.9910133099599983e-05, "loss": 1.625, "step": 2857 }, { "epoch": 7.145, "grad_norm": 7.03125, "learning_rate": 1.9910024760468104e-05, "loss": 1.5982, "step": 2858 }, { "epoch": 7.1475, "grad_norm": 6.5625, "learning_rate": 1.9909916356366498e-05, "loss": 1.3535, "step": 2859 }, { "epoch": 7.15, "grad_norm": 6.75, "learning_rate": 1.990980788729588e-05, "loss": 1.711, "step": 2860 }, { "epoch": 7.1525, "grad_norm": 6.15625, "learning_rate": 1.9909699353256955e-05, "loss": 1.5804, "step": 2861 }, { "epoch": 7.155, "grad_norm": 7.40625, "learning_rate": 1.9909590754250438e-05, "loss": 1.5432, "step": 2862 }, { "epoch": 7.1575, "grad_norm": 5.125, "learning_rate": 1.9909482090277043e-05, "loss": 1.5938, "step": 2863 }, { "epoch": 7.16, "grad_norm": 6.34375, "learning_rate": 1.9909373361337475e-05, "loss": 1.849, "step": 2864 }, { "epoch": 7.1625, "grad_norm": 5.5625, "learning_rate": 1.9909264567432454e-05, "loss": 1.5152, "step": 2865 }, { "epoch": 7.165, "grad_norm": 6.5, "learning_rate": 1.9909155708562694e-05, "loss": 1.6984, "step": 2866 }, { "epoch": 7.1675, "grad_norm": 5.59375, "learning_rate": 1.99090467847289e-05, "loss": 1.5554, "step": 2867 }, { "epoch": 7.17, "grad_norm": 4.9375, "learning_rate": 1.9908937795931798e-05, "loss": 1.4313, "step": 2868 }, { "epoch": 7.1725, "grad_norm": 7.1875, "learning_rate": 1.9908828742172096e-05, "loss": 1.7755, "step": 2869 }, { "epoch": 7.175, "grad_norm": 5.625, "learning_rate": 1.9908719623450507e-05, "loss": 1.6553, "step": 2870 }, { "epoch": 7.1775, "grad_norm": 5.8125, "learning_rate": 1.9908610439767744e-05, "loss": 1.5398, "step": 2871 }, { "epoch": 7.18, "grad_norm": 5.09375, "learning_rate": 1.9908501191124535e-05, "loss": 1.5172, "step": 2872 }, { "epoch": 7.1825, "grad_norm": 7.65625, "learning_rate": 1.9908391877521586e-05, "loss": 1.7298, "step": 2873 }, { "epoch": 7.185, "grad_norm": 8.1875, "learning_rate": 1.9908282498959616e-05, "loss": 1.5914, "step": 2874 }, { "epoch": 7.1875, "grad_norm": 7.96875, "learning_rate": 1.9908173055439343e-05, "loss": 1.6862, "step": 2875 }, { "epoch": 7.19, "grad_norm": 5.9375, "learning_rate": 1.9908063546961482e-05, "loss": 1.6493, "step": 2876 }, { "epoch": 7.1925, "grad_norm": 7.375, "learning_rate": 1.990795397352676e-05, "loss": 1.6203, "step": 2877 }, { "epoch": 7.195, "grad_norm": 7.40625, "learning_rate": 1.990784433513588e-05, "loss": 1.5938, "step": 2878 }, { "epoch": 7.1975, "grad_norm": 5.53125, "learning_rate": 1.9907734631789572e-05, "loss": 1.5571, "step": 2879 }, { "epoch": 7.2, "grad_norm": 6.375, "learning_rate": 1.990762486348855e-05, "loss": 1.6116, "step": 2880 }, { "epoch": 7.2025, "grad_norm": 6.625, "learning_rate": 1.9907515030233536e-05, "loss": 1.5992, "step": 2881 }, { "epoch": 7.205, "grad_norm": 7.6875, "learning_rate": 1.9907405132025252e-05, "loss": 1.7266, "step": 2882 }, { "epoch": 7.2075, "grad_norm": 6.5625, "learning_rate": 1.990729516886442e-05, "loss": 1.7269, "step": 2883 }, { "epoch": 7.21, "grad_norm": 6.625, "learning_rate": 1.9907185140751753e-05, "loss": 1.5492, "step": 2884 }, { "epoch": 7.2125, "grad_norm": 6.125, "learning_rate": 1.9907075047687975e-05, "loss": 1.6049, "step": 2885 }, { "epoch": 7.215, "grad_norm": 6.40625, "learning_rate": 1.990696488967381e-05, "loss": 1.4678, "step": 2886 }, { "epoch": 7.2175, "grad_norm": 8.5, "learning_rate": 1.9906854666709985e-05, "loss": 1.7714, "step": 2887 }, { "epoch": 7.22, "grad_norm": 5.875, "learning_rate": 1.9906744378797214e-05, "loss": 1.5277, "step": 2888 }, { "epoch": 7.2225, "grad_norm": 5.65625, "learning_rate": 1.9906634025936223e-05, "loss": 1.5557, "step": 2889 }, { "epoch": 7.225, "grad_norm": 6.59375, "learning_rate": 1.9906523608127734e-05, "loss": 1.4142, "step": 2890 }, { "epoch": 7.2275, "grad_norm": 6.40625, "learning_rate": 1.9906413125372476e-05, "loss": 1.4044, "step": 2891 }, { "epoch": 7.23, "grad_norm": 7.375, "learning_rate": 1.9906302577671166e-05, "loss": 1.6297, "step": 2892 }, { "epoch": 7.2325, "grad_norm": 5.8125, "learning_rate": 1.9906191965024536e-05, "loss": 1.6579, "step": 2893 }, { "epoch": 7.235, "grad_norm": 6.0625, "learning_rate": 1.9906081287433306e-05, "loss": 1.5302, "step": 2894 }, { "epoch": 7.2375, "grad_norm": 6.8125, "learning_rate": 1.9905970544898206e-05, "loss": 1.5843, "step": 2895 }, { "epoch": 7.24, "grad_norm": 5.40625, "learning_rate": 1.990585973741996e-05, "loss": 1.6244, "step": 2896 }, { "epoch": 7.2425, "grad_norm": 16.375, "learning_rate": 1.9905748864999288e-05, "loss": 1.6561, "step": 2897 }, { "epoch": 7.245, "grad_norm": 5.71875, "learning_rate": 1.9905637927636927e-05, "loss": 1.6156, "step": 2898 }, { "epoch": 7.2475, "grad_norm": 5.71875, "learning_rate": 1.99055269253336e-05, "loss": 1.633, "step": 2899 }, { "epoch": 7.25, "grad_norm": 6.09375, "learning_rate": 1.9905415858090036e-05, "loss": 1.6687, "step": 2900 }, { "epoch": 7.2525, "grad_norm": 4.375, "learning_rate": 1.990530472590696e-05, "loss": 1.688, "step": 2901 }, { "epoch": 7.255, "grad_norm": 6.9375, "learning_rate": 1.9905193528785103e-05, "loss": 1.5702, "step": 2902 }, { "epoch": 7.2575, "grad_norm": 6.40625, "learning_rate": 1.9905082266725194e-05, "loss": 1.5301, "step": 2903 }, { "epoch": 7.26, "grad_norm": 8.0625, "learning_rate": 1.9904970939727962e-05, "loss": 1.6097, "step": 2904 }, { "epoch": 7.2625, "grad_norm": 10.625, "learning_rate": 1.9904859547794138e-05, "loss": 1.5068, "step": 2905 }, { "epoch": 7.265, "grad_norm": 6.5, "learning_rate": 1.990474809092445e-05, "loss": 1.6656, "step": 2906 }, { "epoch": 7.2675, "grad_norm": 6.96875, "learning_rate": 1.990463656911963e-05, "loss": 1.3985, "step": 2907 }, { "epoch": 7.27, "grad_norm": 6.46875, "learning_rate": 1.9904524982380408e-05, "loss": 1.6458, "step": 2908 }, { "epoch": 7.2725, "grad_norm": 6.375, "learning_rate": 1.990441333070752e-05, "loss": 1.5888, "step": 2909 }, { "epoch": 7.275, "grad_norm": 9.125, "learning_rate": 1.990430161410169e-05, "loss": 1.7257, "step": 2910 }, { "epoch": 7.2775, "grad_norm": 7.28125, "learning_rate": 1.9904189832563658e-05, "loss": 1.6646, "step": 2911 }, { "epoch": 7.28, "grad_norm": 5.8125, "learning_rate": 1.9904077986094153e-05, "loss": 1.7095, "step": 2912 }, { "epoch": 7.2825, "grad_norm": 7.84375, "learning_rate": 1.9903966074693908e-05, "loss": 1.582, "step": 2913 }, { "epoch": 7.285, "grad_norm": 5.4375, "learning_rate": 1.9903854098363658e-05, "loss": 1.5171, "step": 2914 }, { "epoch": 7.2875, "grad_norm": 5.75, "learning_rate": 1.9903742057104138e-05, "loss": 1.5538, "step": 2915 }, { "epoch": 7.29, "grad_norm": 7.46875, "learning_rate": 1.9903629950916083e-05, "loss": 1.5902, "step": 2916 }, { "epoch": 7.2925, "grad_norm": 6.0625, "learning_rate": 1.9903517779800222e-05, "loss": 1.7538, "step": 2917 }, { "epoch": 7.295, "grad_norm": 7.34375, "learning_rate": 1.9903405543757297e-05, "loss": 1.5115, "step": 2918 }, { "epoch": 7.2975, "grad_norm": 5.71875, "learning_rate": 1.9903293242788043e-05, "loss": 1.655, "step": 2919 }, { "epoch": 7.3, "grad_norm": 4.6875, "learning_rate": 1.9903180876893195e-05, "loss": 1.7908, "step": 2920 }, { "epoch": 7.3025, "grad_norm": 7.5625, "learning_rate": 1.9903068446073485e-05, "loss": 1.5528, "step": 2921 }, { "epoch": 7.305, "grad_norm": 7.0, "learning_rate": 1.9902955950329658e-05, "loss": 1.8233, "step": 2922 }, { "epoch": 7.3075, "grad_norm": 6.0, "learning_rate": 1.990284338966245e-05, "loss": 1.575, "step": 2923 }, { "epoch": 7.31, "grad_norm": 10.0, "learning_rate": 1.9902730764072593e-05, "loss": 1.686, "step": 2924 }, { "epoch": 7.3125, "grad_norm": 6.375, "learning_rate": 1.990261807356083e-05, "loss": 1.5786, "step": 2925 }, { "epoch": 7.315, "grad_norm": 6.09375, "learning_rate": 1.99025053181279e-05, "loss": 1.4659, "step": 2926 }, { "epoch": 7.3175, "grad_norm": 6.5, "learning_rate": 1.9902392497774543e-05, "loss": 1.587, "step": 2927 }, { "epoch": 7.32, "grad_norm": 8.125, "learning_rate": 1.9902279612501494e-05, "loss": 1.7794, "step": 2928 }, { "epoch": 7.3225, "grad_norm": 7.75, "learning_rate": 1.99021666623095e-05, "loss": 1.7559, "step": 2929 }, { "epoch": 7.325, "grad_norm": 7.75, "learning_rate": 1.9902053647199296e-05, "loss": 1.4228, "step": 2930 }, { "epoch": 7.3275, "grad_norm": 5.59375, "learning_rate": 1.990194056717162e-05, "loss": 1.4679, "step": 2931 }, { "epoch": 7.33, "grad_norm": 5.1875, "learning_rate": 1.9901827422227225e-05, "loss": 1.606, "step": 2932 }, { "epoch": 7.3325, "grad_norm": 6.96875, "learning_rate": 1.990171421236684e-05, "loss": 1.6675, "step": 2933 }, { "epoch": 7.335, "grad_norm": 9.875, "learning_rate": 1.9901600937591218e-05, "loss": 1.555, "step": 2934 }, { "epoch": 7.3375, "grad_norm": 6.15625, "learning_rate": 1.9901487597901093e-05, "loss": 1.4443, "step": 2935 }, { "epoch": 7.34, "grad_norm": 6.875, "learning_rate": 1.9901374193297212e-05, "loss": 1.6662, "step": 2936 }, { "epoch": 7.3425, "grad_norm": 7.03125, "learning_rate": 1.990126072378032e-05, "loss": 1.6237, "step": 2937 }, { "epoch": 7.345, "grad_norm": 8.1875, "learning_rate": 1.9901147189351156e-05, "loss": 1.5807, "step": 2938 }, { "epoch": 7.3475, "grad_norm": 11.5, "learning_rate": 1.9901033590010467e-05, "loss": 1.6948, "step": 2939 }, { "epoch": 7.35, "grad_norm": 7.375, "learning_rate": 1.9900919925759e-05, "loss": 1.602, "step": 2940 }, { "epoch": 7.3525, "grad_norm": 6.9375, "learning_rate": 1.99008061965975e-05, "loss": 1.7442, "step": 2941 }, { "epoch": 7.355, "grad_norm": 6.21875, "learning_rate": 1.9900692402526708e-05, "loss": 1.5577, "step": 2942 }, { "epoch": 7.3575, "grad_norm": 5.96875, "learning_rate": 1.990057854354737e-05, "loss": 1.5702, "step": 2943 }, { "epoch": 7.36, "grad_norm": 6.71875, "learning_rate": 1.9900464619660243e-05, "loss": 1.4855, "step": 2944 }, { "epoch": 7.3625, "grad_norm": 8.1875, "learning_rate": 1.990035063086606e-05, "loss": 1.6781, "step": 2945 }, { "epoch": 7.365, "grad_norm": 5.6875, "learning_rate": 1.990023657716558e-05, "loss": 1.5246, "step": 2946 }, { "epoch": 7.3675, "grad_norm": 6.46875, "learning_rate": 1.9900122458559537e-05, "loss": 1.4463, "step": 2947 }, { "epoch": 7.37, "grad_norm": 5.21875, "learning_rate": 1.9900008275048696e-05, "loss": 1.501, "step": 2948 }, { "epoch": 7.3725, "grad_norm": 6.65625, "learning_rate": 1.989989402663379e-05, "loss": 1.6188, "step": 2949 }, { "epoch": 7.375, "grad_norm": 14.5625, "learning_rate": 1.9899779713315577e-05, "loss": 1.5744, "step": 2950 }, { "epoch": 7.3775, "grad_norm": 6.125, "learning_rate": 1.9899665335094805e-05, "loss": 1.7814, "step": 2951 }, { "epoch": 7.38, "grad_norm": 5.1875, "learning_rate": 1.9899550891972224e-05, "loss": 1.4903, "step": 2952 }, { "epoch": 7.3825, "grad_norm": 7.21875, "learning_rate": 1.9899436383948583e-05, "loss": 1.4904, "step": 2953 }, { "epoch": 7.385, "grad_norm": 6.46875, "learning_rate": 1.9899321811024633e-05, "loss": 1.6785, "step": 2954 }, { "epoch": 7.3875, "grad_norm": 6.9375, "learning_rate": 1.9899207173201124e-05, "loss": 1.5739, "step": 2955 }, { "epoch": 7.39, "grad_norm": 5.875, "learning_rate": 1.989909247047881e-05, "loss": 1.6029, "step": 2956 }, { "epoch": 7.3925, "grad_norm": 6.03125, "learning_rate": 1.9898977702858444e-05, "loss": 1.6176, "step": 2957 }, { "epoch": 7.395, "grad_norm": 6.59375, "learning_rate": 1.9898862870340773e-05, "loss": 1.4651, "step": 2958 }, { "epoch": 7.3975, "grad_norm": 5.84375, "learning_rate": 1.9898747972926555e-05, "loss": 1.5897, "step": 2959 }, { "epoch": 7.4, "grad_norm": 5.8125, "learning_rate": 1.989863301061654e-05, "loss": 1.6948, "step": 2960 }, { "epoch": 7.4025, "grad_norm": 5.25, "learning_rate": 1.989851798341149e-05, "loss": 1.5121, "step": 2961 }, { "epoch": 7.405, "grad_norm": 4.09375, "learning_rate": 1.989840289131214e-05, "loss": 1.6561, "step": 2962 }, { "epoch": 7.4075, "grad_norm": 7.6875, "learning_rate": 1.9898287734319266e-05, "loss": 1.5239, "step": 2963 }, { "epoch": 7.41, "grad_norm": 6.21875, "learning_rate": 1.989817251243361e-05, "loss": 1.6205, "step": 2964 }, { "epoch": 7.4125, "grad_norm": 4.4375, "learning_rate": 1.9898057225655934e-05, "loss": 1.4645, "step": 2965 }, { "epoch": 7.415, "grad_norm": 6.5625, "learning_rate": 1.989794187398699e-05, "loss": 1.4711, "step": 2966 }, { "epoch": 7.4175, "grad_norm": 7.96875, "learning_rate": 1.989782645742753e-05, "loss": 1.7123, "step": 2967 }, { "epoch": 7.42, "grad_norm": 6.25, "learning_rate": 1.9897710975978323e-05, "loss": 1.5983, "step": 2968 }, { "epoch": 7.4225, "grad_norm": 5.96875, "learning_rate": 1.9897595429640116e-05, "loss": 1.5142, "step": 2969 }, { "epoch": 7.425, "grad_norm": 5.0625, "learning_rate": 1.989747981841367e-05, "loss": 1.3951, "step": 2970 }, { "epoch": 7.4275, "grad_norm": 6.09375, "learning_rate": 1.9897364142299738e-05, "loss": 1.551, "step": 2971 }, { "epoch": 7.43, "grad_norm": 6.6875, "learning_rate": 1.989724840129909e-05, "loss": 1.4792, "step": 2972 }, { "epoch": 7.4325, "grad_norm": 4.6875, "learning_rate": 1.989713259541247e-05, "loss": 1.4694, "step": 2973 }, { "epoch": 7.435, "grad_norm": 5.78125, "learning_rate": 1.989701672464065e-05, "loss": 1.4767, "step": 2974 }, { "epoch": 7.4375, "grad_norm": 6.0625, "learning_rate": 1.9896900788984383e-05, "loss": 1.6805, "step": 2975 }, { "epoch": 7.44, "grad_norm": 5.0, "learning_rate": 1.989678478844443e-05, "loss": 1.4961, "step": 2976 }, { "epoch": 7.4425, "grad_norm": 5.53125, "learning_rate": 1.989666872302155e-05, "loss": 1.5969, "step": 2977 }, { "epoch": 7.445, "grad_norm": 4.9375, "learning_rate": 1.9896552592716513e-05, "loss": 1.597, "step": 2978 }, { "epoch": 7.4475, "grad_norm": 7.25, "learning_rate": 1.989643639753007e-05, "loss": 1.3652, "step": 2979 }, { "epoch": 7.45, "grad_norm": 6.625, "learning_rate": 1.9896320137462984e-05, "loss": 1.7699, "step": 2980 }, { "epoch": 7.4525, "grad_norm": 6.65625, "learning_rate": 1.989620381251602e-05, "loss": 1.6925, "step": 2981 }, { "epoch": 7.455, "grad_norm": 6.21875, "learning_rate": 1.9896087422689937e-05, "loss": 1.7047, "step": 2982 }, { "epoch": 7.4575, "grad_norm": 7.1875, "learning_rate": 1.9895970967985507e-05, "loss": 1.4866, "step": 2983 }, { "epoch": 7.46, "grad_norm": 4.84375, "learning_rate": 1.9895854448403482e-05, "loss": 1.4675, "step": 2984 }, { "epoch": 7.4625, "grad_norm": 5.96875, "learning_rate": 1.9895737863944632e-05, "loss": 1.5853, "step": 2985 }, { "epoch": 7.465, "grad_norm": 6.75, "learning_rate": 1.9895621214609723e-05, "loss": 1.589, "step": 2986 }, { "epoch": 7.4675, "grad_norm": 5.15625, "learning_rate": 1.9895504500399517e-05, "loss": 1.6136, "step": 2987 }, { "epoch": 7.47, "grad_norm": 5.5625, "learning_rate": 1.9895387721314776e-05, "loss": 1.5339, "step": 2988 }, { "epoch": 7.4725, "grad_norm": 5.96875, "learning_rate": 1.9895270877356273e-05, "loss": 1.5699, "step": 2989 }, { "epoch": 7.475, "grad_norm": 7.15625, "learning_rate": 1.9895153968524767e-05, "loss": 1.6388, "step": 2990 }, { "epoch": 7.4775, "grad_norm": 4.65625, "learning_rate": 1.989503699482103e-05, "loss": 1.6672, "step": 2991 }, { "epoch": 7.48, "grad_norm": 5.6875, "learning_rate": 1.9894919956245825e-05, "loss": 1.6201, "step": 2992 }, { "epoch": 7.4825, "grad_norm": 7.9375, "learning_rate": 1.989480285279992e-05, "loss": 1.6083, "step": 2993 }, { "epoch": 7.485, "grad_norm": 5.59375, "learning_rate": 1.989468568448409e-05, "loss": 1.7022, "step": 2994 }, { "epoch": 7.4875, "grad_norm": 6.40625, "learning_rate": 1.989456845129909e-05, "loss": 1.5939, "step": 2995 }, { "epoch": 7.49, "grad_norm": 6.53125, "learning_rate": 1.9894451153245695e-05, "loss": 1.5957, "step": 2996 }, { "epoch": 7.4925, "grad_norm": 10.5, "learning_rate": 1.9894333790324678e-05, "loss": 1.7003, "step": 2997 }, { "epoch": 7.495, "grad_norm": 5.8125, "learning_rate": 1.9894216362536804e-05, "loss": 1.472, "step": 2998 }, { "epoch": 7.4975, "grad_norm": 6.59375, "learning_rate": 1.989409886988284e-05, "loss": 1.4982, "step": 2999 }, { "epoch": 7.5, "grad_norm": 5.53125, "learning_rate": 1.9893981312363563e-05, "loss": 1.7501, "step": 3000 }, { "epoch": 7.5024999999999995, "grad_norm": 11.875, "learning_rate": 1.989386368997974e-05, "loss": 1.5548, "step": 3001 }, { "epoch": 7.505, "grad_norm": 8.125, "learning_rate": 1.989374600273214e-05, "loss": 1.6297, "step": 3002 }, { "epoch": 7.5075, "grad_norm": 11.6875, "learning_rate": 1.9893628250621537e-05, "loss": 1.7917, "step": 3003 }, { "epoch": 7.51, "grad_norm": 5.34375, "learning_rate": 1.9893510433648705e-05, "loss": 1.4755, "step": 3004 }, { "epoch": 7.5125, "grad_norm": 6.15625, "learning_rate": 1.9893392551814416e-05, "loss": 1.5498, "step": 3005 }, { "epoch": 7.515, "grad_norm": 6.78125, "learning_rate": 1.989327460511944e-05, "loss": 1.5954, "step": 3006 }, { "epoch": 7.5175, "grad_norm": 6.1875, "learning_rate": 1.9893156593564552e-05, "loss": 1.548, "step": 3007 }, { "epoch": 7.52, "grad_norm": 7.71875, "learning_rate": 1.9893038517150526e-05, "loss": 1.5343, "step": 3008 }, { "epoch": 7.5225, "grad_norm": 5.84375, "learning_rate": 1.9892920375878133e-05, "loss": 1.5257, "step": 3009 }, { "epoch": 7.525, "grad_norm": 4.90625, "learning_rate": 1.989280216974815e-05, "loss": 1.6952, "step": 3010 }, { "epoch": 7.5275, "grad_norm": 8.75, "learning_rate": 1.9892683898761355e-05, "loss": 1.775, "step": 3011 }, { "epoch": 7.53, "grad_norm": 8.5625, "learning_rate": 1.9892565562918518e-05, "loss": 1.5357, "step": 3012 }, { "epoch": 7.5325, "grad_norm": 5.34375, "learning_rate": 1.9892447162220418e-05, "loss": 1.4315, "step": 3013 }, { "epoch": 7.535, "grad_norm": 6.0, "learning_rate": 1.9892328696667827e-05, "loss": 1.674, "step": 3014 }, { "epoch": 7.5375, "grad_norm": 7.875, "learning_rate": 1.989221016626153e-05, "loss": 1.6549, "step": 3015 }, { "epoch": 7.54, "grad_norm": 6.125, "learning_rate": 1.9892091571002295e-05, "loss": 1.7247, "step": 3016 }, { "epoch": 7.5425, "grad_norm": 7.15625, "learning_rate": 1.9891972910890906e-05, "loss": 1.7053, "step": 3017 }, { "epoch": 7.545, "grad_norm": 7.65625, "learning_rate": 1.9891854185928136e-05, "loss": 1.7078, "step": 3018 }, { "epoch": 7.5475, "grad_norm": 5.78125, "learning_rate": 1.9891735396114765e-05, "loss": 1.7068, "step": 3019 }, { "epoch": 7.55, "grad_norm": 7.28125, "learning_rate": 1.989161654145158e-05, "loss": 1.9428, "step": 3020 }, { "epoch": 7.5525, "grad_norm": 5.46875, "learning_rate": 1.9891497621939345e-05, "loss": 1.663, "step": 3021 }, { "epoch": 7.555, "grad_norm": 7.6875, "learning_rate": 1.989137863757885e-05, "loss": 1.5355, "step": 3022 }, { "epoch": 7.5575, "grad_norm": 5.375, "learning_rate": 1.9891259588370873e-05, "loss": 1.3783, "step": 3023 }, { "epoch": 7.5600000000000005, "grad_norm": 5.28125, "learning_rate": 1.9891140474316197e-05, "loss": 1.5953, "step": 3024 }, { "epoch": 7.5625, "grad_norm": 6.6875, "learning_rate": 1.9891021295415592e-05, "loss": 1.6109, "step": 3025 }, { "epoch": 7.5649999999999995, "grad_norm": 7.78125, "learning_rate": 1.9890902051669855e-05, "loss": 1.4762, "step": 3026 }, { "epoch": 7.5675, "grad_norm": 7.0, "learning_rate": 1.989078274307975e-05, "loss": 1.7069, "step": 3027 }, { "epoch": 7.57, "grad_norm": 6.59375, "learning_rate": 1.989066336964608e-05, "loss": 1.6334, "step": 3028 }, { "epoch": 7.5725, "grad_norm": 8.8125, "learning_rate": 1.9890543931369612e-05, "loss": 1.5932, "step": 3029 }, { "epoch": 7.575, "grad_norm": 6.34375, "learning_rate": 1.9890424428251135e-05, "loss": 1.6075, "step": 3030 }, { "epoch": 7.5775, "grad_norm": 6.125, "learning_rate": 1.9890304860291428e-05, "loss": 1.5621, "step": 3031 }, { "epoch": 7.58, "grad_norm": 5.71875, "learning_rate": 1.9890185227491285e-05, "loss": 1.5993, "step": 3032 }, { "epoch": 7.5825, "grad_norm": 5.25, "learning_rate": 1.9890065529851477e-05, "loss": 1.4645, "step": 3033 }, { "epoch": 7.585, "grad_norm": 5.6875, "learning_rate": 1.98899457673728e-05, "loss": 1.6818, "step": 3034 }, { "epoch": 7.5875, "grad_norm": 6.5625, "learning_rate": 1.988982594005603e-05, "loss": 1.6221, "step": 3035 }, { "epoch": 7.59, "grad_norm": 5.21875, "learning_rate": 1.9889706047901956e-05, "loss": 1.61, "step": 3036 }, { "epoch": 7.5925, "grad_norm": 7.0625, "learning_rate": 1.9889586090911367e-05, "loss": 1.6292, "step": 3037 }, { "epoch": 7.595, "grad_norm": 6.4375, "learning_rate": 1.9889466069085048e-05, "loss": 1.5908, "step": 3038 }, { "epoch": 7.5975, "grad_norm": 4.875, "learning_rate": 1.9889345982423784e-05, "loss": 1.6685, "step": 3039 }, { "epoch": 7.6, "grad_norm": 6.5, "learning_rate": 1.9889225830928365e-05, "loss": 1.4944, "step": 3040 }, { "epoch": 7.6025, "grad_norm": 5.40625, "learning_rate": 1.9889105614599577e-05, "loss": 1.588, "step": 3041 }, { "epoch": 7.605, "grad_norm": 5.9375, "learning_rate": 1.9888985333438207e-05, "loss": 1.6037, "step": 3042 }, { "epoch": 7.6075, "grad_norm": 6.75, "learning_rate": 1.988886498744505e-05, "loss": 1.4246, "step": 3043 }, { "epoch": 7.61, "grad_norm": 5.90625, "learning_rate": 1.9888744576620883e-05, "loss": 1.5966, "step": 3044 }, { "epoch": 7.6125, "grad_norm": 4.75, "learning_rate": 1.9888624100966504e-05, "loss": 1.4242, "step": 3045 }, { "epoch": 7.615, "grad_norm": 11.4375, "learning_rate": 1.9888503560482703e-05, "loss": 1.6074, "step": 3046 }, { "epoch": 7.6175, "grad_norm": 4.53125, "learning_rate": 1.9888382955170267e-05, "loss": 1.5962, "step": 3047 }, { "epoch": 7.62, "grad_norm": 6.28125, "learning_rate": 1.988826228502999e-05, "loss": 1.6208, "step": 3048 }, { "epoch": 7.6225000000000005, "grad_norm": 6.28125, "learning_rate": 1.988814155006266e-05, "loss": 1.5113, "step": 3049 }, { "epoch": 7.625, "grad_norm": 7.15625, "learning_rate": 1.9888020750269067e-05, "loss": 1.6649, "step": 3050 }, { "epoch": 7.6274999999999995, "grad_norm": 4.625, "learning_rate": 1.988789988565001e-05, "loss": 1.6053, "step": 3051 }, { "epoch": 7.63, "grad_norm": 6.4375, "learning_rate": 1.9887778956206274e-05, "loss": 1.3769, "step": 3052 }, { "epoch": 7.6325, "grad_norm": 6.3125, "learning_rate": 1.9887657961938657e-05, "loss": 1.5302, "step": 3053 }, { "epoch": 7.635, "grad_norm": 8.3125, "learning_rate": 1.9887536902847947e-05, "loss": 1.7564, "step": 3054 }, { "epoch": 7.6375, "grad_norm": 5.34375, "learning_rate": 1.9887415778934944e-05, "loss": 1.5372, "step": 3055 }, { "epoch": 7.64, "grad_norm": 6.125, "learning_rate": 1.9887294590200437e-05, "loss": 1.5454, "step": 3056 }, { "epoch": 7.6425, "grad_norm": 10.75, "learning_rate": 1.9887173336645223e-05, "loss": 1.6844, "step": 3057 }, { "epoch": 7.645, "grad_norm": 6.28125, "learning_rate": 1.9887052018270098e-05, "loss": 1.5396, "step": 3058 }, { "epoch": 7.6475, "grad_norm": 8.3125, "learning_rate": 1.988693063507585e-05, "loss": 1.6701, "step": 3059 }, { "epoch": 7.65, "grad_norm": 7.375, "learning_rate": 1.9886809187063285e-05, "loss": 1.4614, "step": 3060 }, { "epoch": 7.6525, "grad_norm": 7.5, "learning_rate": 1.9886687674233193e-05, "loss": 1.5174, "step": 3061 }, { "epoch": 7.655, "grad_norm": 7.5625, "learning_rate": 1.9886566096586375e-05, "loss": 1.5299, "step": 3062 }, { "epoch": 7.6575, "grad_norm": 6.90625, "learning_rate": 1.9886444454123623e-05, "loss": 1.723, "step": 3063 }, { "epoch": 7.66, "grad_norm": 5.875, "learning_rate": 1.9886322746845738e-05, "loss": 1.5168, "step": 3064 }, { "epoch": 7.6625, "grad_norm": 6.53125, "learning_rate": 1.9886200974753514e-05, "loss": 1.5884, "step": 3065 }, { "epoch": 7.665, "grad_norm": 6.03125, "learning_rate": 1.9886079137847756e-05, "loss": 1.6049, "step": 3066 }, { "epoch": 7.6675, "grad_norm": 6.90625, "learning_rate": 1.9885957236129257e-05, "loss": 1.6089, "step": 3067 }, { "epoch": 7.67, "grad_norm": 6.96875, "learning_rate": 1.9885835269598818e-05, "loss": 1.6882, "step": 3068 }, { "epoch": 7.6725, "grad_norm": 6.40625, "learning_rate": 1.9885713238257236e-05, "loss": 1.678, "step": 3069 }, { "epoch": 7.675, "grad_norm": 5.375, "learning_rate": 1.988559114210532e-05, "loss": 1.5338, "step": 3070 }, { "epoch": 7.6775, "grad_norm": 6.375, "learning_rate": 1.9885468981143857e-05, "loss": 1.4699, "step": 3071 }, { "epoch": 7.68, "grad_norm": 8.375, "learning_rate": 1.988534675537366e-05, "loss": 1.6183, "step": 3072 }, { "epoch": 7.6825, "grad_norm": 7.25, "learning_rate": 1.988522446479552e-05, "loss": 1.7785, "step": 3073 }, { "epoch": 7.6850000000000005, "grad_norm": 5.4375, "learning_rate": 1.988510210941025e-05, "loss": 1.6306, "step": 3074 }, { "epoch": 7.6875, "grad_norm": 5.65625, "learning_rate": 1.988497968921864e-05, "loss": 1.5119, "step": 3075 }, { "epoch": 7.6899999999999995, "grad_norm": 7.53125, "learning_rate": 1.9884857204221503e-05, "loss": 1.6736, "step": 3076 }, { "epoch": 7.6925, "grad_norm": 5.40625, "learning_rate": 1.9884734654419636e-05, "loss": 1.5732, "step": 3077 }, { "epoch": 7.695, "grad_norm": 5.6875, "learning_rate": 1.9884612039813847e-05, "loss": 1.4002, "step": 3078 }, { "epoch": 7.6975, "grad_norm": 10.25, "learning_rate": 1.9884489360404932e-05, "loss": 1.5385, "step": 3079 }, { "epoch": 7.7, "grad_norm": 7.75, "learning_rate": 1.9884366616193707e-05, "loss": 1.7761, "step": 3080 }, { "epoch": 7.7025, "grad_norm": 5.96875, "learning_rate": 1.9884243807180965e-05, "loss": 1.4799, "step": 3081 }, { "epoch": 7.705, "grad_norm": 4.71875, "learning_rate": 1.9884120933367515e-05, "loss": 1.4025, "step": 3082 }, { "epoch": 7.7075, "grad_norm": 5.0, "learning_rate": 1.988399799475417e-05, "loss": 1.4493, "step": 3083 }, { "epoch": 7.71, "grad_norm": 11.0, "learning_rate": 1.9883874991341724e-05, "loss": 1.6221, "step": 3084 }, { "epoch": 7.7125, "grad_norm": 4.25, "learning_rate": 1.988375192313099e-05, "loss": 1.6229, "step": 3085 }, { "epoch": 7.715, "grad_norm": 6.21875, "learning_rate": 1.9883628790122776e-05, "loss": 1.4721, "step": 3086 }, { "epoch": 7.7175, "grad_norm": 6.21875, "learning_rate": 1.9883505592317886e-05, "loss": 1.7473, "step": 3087 }, { "epoch": 7.72, "grad_norm": 8.1875, "learning_rate": 1.988338232971713e-05, "loss": 1.4579, "step": 3088 }, { "epoch": 7.7225, "grad_norm": 7.125, "learning_rate": 1.9883259002321312e-05, "loss": 1.7705, "step": 3089 }, { "epoch": 7.725, "grad_norm": 11.625, "learning_rate": 1.9883135610131244e-05, "loss": 1.8039, "step": 3090 }, { "epoch": 7.7275, "grad_norm": 5.125, "learning_rate": 1.9883012153147737e-05, "loss": 1.8094, "step": 3091 }, { "epoch": 7.73, "grad_norm": 9.5, "learning_rate": 1.9882888631371596e-05, "loss": 1.7108, "step": 3092 }, { "epoch": 7.7325, "grad_norm": 9.1875, "learning_rate": 1.9882765044803635e-05, "loss": 1.7746, "step": 3093 }, { "epoch": 7.735, "grad_norm": 7.03125, "learning_rate": 1.988264139344466e-05, "loss": 1.4771, "step": 3094 }, { "epoch": 7.7375, "grad_norm": 13.1875, "learning_rate": 1.988251767729548e-05, "loss": 1.4842, "step": 3095 }, { "epoch": 7.74, "grad_norm": 3.21875, "learning_rate": 1.9882393896356915e-05, "loss": 1.5289, "step": 3096 }, { "epoch": 7.7425, "grad_norm": 8.125, "learning_rate": 1.988227005062977e-05, "loss": 1.6265, "step": 3097 }, { "epoch": 7.745, "grad_norm": 7.65625, "learning_rate": 1.9882146140114855e-05, "loss": 1.7426, "step": 3098 }, { "epoch": 7.7475000000000005, "grad_norm": 5.96875, "learning_rate": 1.988202216481299e-05, "loss": 1.5302, "step": 3099 }, { "epoch": 7.75, "grad_norm": 6.28125, "learning_rate": 1.988189812472498e-05, "loss": 1.4817, "step": 3100 }, { "epoch": 7.7524999999999995, "grad_norm": 5.9375, "learning_rate": 1.9881774019851642e-05, "loss": 1.6481, "step": 3101 }, { "epoch": 7.755, "grad_norm": 5.84375, "learning_rate": 1.988164985019379e-05, "loss": 1.6116, "step": 3102 }, { "epoch": 7.7575, "grad_norm": 6.09375, "learning_rate": 1.9881525615752235e-05, "loss": 1.5402, "step": 3103 }, { "epoch": 7.76, "grad_norm": 6.5, "learning_rate": 1.9881401316527795e-05, "loss": 1.559, "step": 3104 }, { "epoch": 7.7625, "grad_norm": 5.4375, "learning_rate": 1.9881276952521282e-05, "loss": 1.5796, "step": 3105 }, { "epoch": 7.765, "grad_norm": 8.125, "learning_rate": 1.9881152523733513e-05, "loss": 1.4782, "step": 3106 }, { "epoch": 7.7675, "grad_norm": 5.21875, "learning_rate": 1.9881028030165305e-05, "loss": 1.6732, "step": 3107 }, { "epoch": 7.77, "grad_norm": 4.75, "learning_rate": 1.9880903471817473e-05, "loss": 1.3753, "step": 3108 }, { "epoch": 7.7725, "grad_norm": 5.1875, "learning_rate": 1.9880778848690832e-05, "loss": 1.5334, "step": 3109 }, { "epoch": 7.775, "grad_norm": 8.0, "learning_rate": 1.9880654160786198e-05, "loss": 1.7222, "step": 3110 }, { "epoch": 7.7775, "grad_norm": 5.75, "learning_rate": 1.9880529408104394e-05, "loss": 1.5509, "step": 3111 }, { "epoch": 7.78, "grad_norm": 7.21875, "learning_rate": 1.9880404590646233e-05, "loss": 1.5271, "step": 3112 }, { "epoch": 7.7825, "grad_norm": 7.21875, "learning_rate": 1.9880279708412535e-05, "loss": 1.4799, "step": 3113 }, { "epoch": 7.785, "grad_norm": 7.09375, "learning_rate": 1.988015476140412e-05, "loss": 1.5947, "step": 3114 }, { "epoch": 7.7875, "grad_norm": 4.9375, "learning_rate": 1.988002974962181e-05, "loss": 1.507, "step": 3115 }, { "epoch": 7.79, "grad_norm": 6.375, "learning_rate": 1.987990467306641e-05, "loss": 1.7412, "step": 3116 }, { "epoch": 7.7925, "grad_norm": 4.8125, "learning_rate": 1.9879779531738758e-05, "loss": 1.551, "step": 3117 }, { "epoch": 7.795, "grad_norm": 6.125, "learning_rate": 1.9879654325639665e-05, "loss": 1.5666, "step": 3118 }, { "epoch": 7.7975, "grad_norm": 5.84375, "learning_rate": 1.987952905476995e-05, "loss": 1.6113, "step": 3119 }, { "epoch": 7.8, "grad_norm": 8.0, "learning_rate": 1.987940371913044e-05, "loss": 1.7071, "step": 3120 }, { "epoch": 7.8025, "grad_norm": 12.875, "learning_rate": 1.9879278318721958e-05, "loss": 1.6111, "step": 3121 }, { "epoch": 7.805, "grad_norm": 6.9375, "learning_rate": 1.987915285354532e-05, "loss": 1.5776, "step": 3122 }, { "epoch": 7.8075, "grad_norm": 8.625, "learning_rate": 1.987902732360135e-05, "loss": 1.6812, "step": 3123 }, { "epoch": 7.8100000000000005, "grad_norm": 6.4375, "learning_rate": 1.9878901728890875e-05, "loss": 1.7122, "step": 3124 }, { "epoch": 7.8125, "grad_norm": 7.1875, "learning_rate": 1.9878776069414714e-05, "loss": 1.7798, "step": 3125 }, { "epoch": 7.8149999999999995, "grad_norm": 8.3125, "learning_rate": 1.9878650345173695e-05, "loss": 1.7873, "step": 3126 }, { "epoch": 7.8175, "grad_norm": 6.90625, "learning_rate": 1.9878524556168635e-05, "loss": 1.836, "step": 3127 }, { "epoch": 7.82, "grad_norm": 4.96875, "learning_rate": 1.9878398702400364e-05, "loss": 1.6687, "step": 3128 }, { "epoch": 7.8225, "grad_norm": 5.4375, "learning_rate": 1.9878272783869706e-05, "loss": 1.5844, "step": 3129 }, { "epoch": 7.825, "grad_norm": 6.28125, "learning_rate": 1.987814680057749e-05, "loss": 1.5742, "step": 3130 }, { "epoch": 7.8275, "grad_norm": 8.5625, "learning_rate": 1.987802075252454e-05, "loss": 1.6931, "step": 3131 }, { "epoch": 7.83, "grad_norm": 5.21875, "learning_rate": 1.9877894639711678e-05, "loss": 1.5099, "step": 3132 }, { "epoch": 7.8325, "grad_norm": 7.59375, "learning_rate": 1.9877768462139737e-05, "loss": 1.6425, "step": 3133 }, { "epoch": 7.835, "grad_norm": 7.5625, "learning_rate": 1.9877642219809535e-05, "loss": 1.4798, "step": 3134 }, { "epoch": 7.8375, "grad_norm": 8.6875, "learning_rate": 1.9877515912721913e-05, "loss": 1.7236, "step": 3135 }, { "epoch": 7.84, "grad_norm": 7.25, "learning_rate": 1.9877389540877686e-05, "loss": 1.5592, "step": 3136 }, { "epoch": 7.8425, "grad_norm": 7.125, "learning_rate": 1.9877263104277693e-05, "loss": 1.7352, "step": 3137 }, { "epoch": 7.845, "grad_norm": 6.96875, "learning_rate": 1.987713660292276e-05, "loss": 1.5021, "step": 3138 }, { "epoch": 7.8475, "grad_norm": 6.5, "learning_rate": 1.987701003681371e-05, "loss": 1.6148, "step": 3139 }, { "epoch": 7.85, "grad_norm": 6.0625, "learning_rate": 1.9876883405951378e-05, "loss": 1.6474, "step": 3140 }, { "epoch": 7.8525, "grad_norm": 5.9375, "learning_rate": 1.9876756710336594e-05, "loss": 1.6138, "step": 3141 }, { "epoch": 7.855, "grad_norm": 10.5625, "learning_rate": 1.987662994997019e-05, "loss": 1.5793, "step": 3142 }, { "epoch": 7.8575, "grad_norm": 8.5, "learning_rate": 1.9876503124852995e-05, "loss": 1.5913, "step": 3143 }, { "epoch": 7.86, "grad_norm": 5.15625, "learning_rate": 1.9876376234985838e-05, "loss": 1.4318, "step": 3144 }, { "epoch": 7.8625, "grad_norm": 8.625, "learning_rate": 1.9876249280369558e-05, "loss": 1.5874, "step": 3145 }, { "epoch": 7.865, "grad_norm": 6.4375, "learning_rate": 1.987612226100498e-05, "loss": 1.5833, "step": 3146 }, { "epoch": 7.8675, "grad_norm": 5.15625, "learning_rate": 1.987599517689294e-05, "loss": 1.6172, "step": 3147 }, { "epoch": 7.87, "grad_norm": 5.90625, "learning_rate": 1.987586802803427e-05, "loss": 1.665, "step": 3148 }, { "epoch": 7.8725000000000005, "grad_norm": 7.78125, "learning_rate": 1.987574081442981e-05, "loss": 1.7103, "step": 3149 }, { "epoch": 7.875, "grad_norm": 7.84375, "learning_rate": 1.987561353608038e-05, "loss": 1.6514, "step": 3150 }, { "epoch": 7.8774999999999995, "grad_norm": 4.65625, "learning_rate": 1.9875486192986826e-05, "loss": 1.4285, "step": 3151 }, { "epoch": 7.88, "grad_norm": 6.8125, "learning_rate": 1.9875358785149982e-05, "loss": 1.5871, "step": 3152 }, { "epoch": 7.8825, "grad_norm": 6.25, "learning_rate": 1.987523131257068e-05, "loss": 1.6137, "step": 3153 }, { "epoch": 7.885, "grad_norm": 7.46875, "learning_rate": 1.9875103775249755e-05, "loss": 1.7634, "step": 3154 }, { "epoch": 7.8875, "grad_norm": 5.875, "learning_rate": 1.9874976173188046e-05, "loss": 1.3975, "step": 3155 }, { "epoch": 7.89, "grad_norm": 6.21875, "learning_rate": 1.9874848506386392e-05, "loss": 1.6689, "step": 3156 }, { "epoch": 7.8925, "grad_norm": 8.6875, "learning_rate": 1.987472077484562e-05, "loss": 1.5772, "step": 3157 }, { "epoch": 7.895, "grad_norm": 7.46875, "learning_rate": 1.9874592978566575e-05, "loss": 1.6051, "step": 3158 }, { "epoch": 7.8975, "grad_norm": 5.9375, "learning_rate": 1.9874465117550096e-05, "loss": 1.5124, "step": 3159 }, { "epoch": 7.9, "grad_norm": 7.46875, "learning_rate": 1.987433719179702e-05, "loss": 1.5963, "step": 3160 }, { "epoch": 7.9025, "grad_norm": 5.0625, "learning_rate": 1.987420920130818e-05, "loss": 1.5207, "step": 3161 }, { "epoch": 7.905, "grad_norm": 6.4375, "learning_rate": 1.9874081146084423e-05, "loss": 1.509, "step": 3162 }, { "epoch": 7.9075, "grad_norm": 5.53125, "learning_rate": 1.9873953026126583e-05, "loss": 1.4729, "step": 3163 }, { "epoch": 7.91, "grad_norm": 5.3125, "learning_rate": 1.9873824841435505e-05, "loss": 1.5469, "step": 3164 }, { "epoch": 7.9125, "grad_norm": 6.125, "learning_rate": 1.9873696592012023e-05, "loss": 1.6414, "step": 3165 }, { "epoch": 7.915, "grad_norm": 11.3125, "learning_rate": 1.9873568277856986e-05, "loss": 1.7604, "step": 3166 }, { "epoch": 7.9175, "grad_norm": 6.5, "learning_rate": 1.9873439898971227e-05, "loss": 1.7062, "step": 3167 }, { "epoch": 7.92, "grad_norm": 7.21875, "learning_rate": 1.987331145535559e-05, "loss": 1.7984, "step": 3168 }, { "epoch": 7.9225, "grad_norm": 9.3125, "learning_rate": 1.987318294701092e-05, "loss": 1.5489, "step": 3169 }, { "epoch": 7.925, "grad_norm": 4.25, "learning_rate": 1.987305437393806e-05, "loss": 1.5055, "step": 3170 }, { "epoch": 7.9275, "grad_norm": 9.5, "learning_rate": 1.9872925736137848e-05, "loss": 1.7572, "step": 3171 }, { "epoch": 7.93, "grad_norm": 4.71875, "learning_rate": 1.987279703361113e-05, "loss": 1.724, "step": 3172 }, { "epoch": 7.9325, "grad_norm": 5.28125, "learning_rate": 1.987266826635875e-05, "loss": 1.646, "step": 3173 }, { "epoch": 7.9350000000000005, "grad_norm": 5.5, "learning_rate": 1.987253943438155e-05, "loss": 1.4519, "step": 3174 }, { "epoch": 7.9375, "grad_norm": 7.6875, "learning_rate": 1.9872410537680378e-05, "loss": 1.6875, "step": 3175 }, { "epoch": 7.9399999999999995, "grad_norm": 5.3125, "learning_rate": 1.9872281576256078e-05, "loss": 1.617, "step": 3176 }, { "epoch": 7.9425, "grad_norm": 6.625, "learning_rate": 1.9872152550109496e-05, "loss": 1.5778, "step": 3177 }, { "epoch": 7.945, "grad_norm": 5.9375, "learning_rate": 1.9872023459241476e-05, "loss": 1.5359, "step": 3178 }, { "epoch": 7.9475, "grad_norm": 7.0625, "learning_rate": 1.9871894303652867e-05, "loss": 1.6375, "step": 3179 }, { "epoch": 7.95, "grad_norm": 6.25, "learning_rate": 1.987176508334451e-05, "loss": 1.5766, "step": 3180 }, { "epoch": 7.9525, "grad_norm": 6.71875, "learning_rate": 1.9871635798317257e-05, "loss": 1.5709, "step": 3181 }, { "epoch": 7.955, "grad_norm": 6.625, "learning_rate": 1.9871506448571956e-05, "loss": 1.6252, "step": 3182 }, { "epoch": 7.9575, "grad_norm": 7.09375, "learning_rate": 1.9871377034109456e-05, "loss": 1.6754, "step": 3183 }, { "epoch": 7.96, "grad_norm": 5.96875, "learning_rate": 1.98712475549306e-05, "loss": 1.5205, "step": 3184 }, { "epoch": 7.9625, "grad_norm": 6.03125, "learning_rate": 1.9871118011036242e-05, "loss": 1.5446, "step": 3185 }, { "epoch": 7.965, "grad_norm": 6.09375, "learning_rate": 1.9870988402427227e-05, "loss": 1.5985, "step": 3186 }, { "epoch": 7.9675, "grad_norm": 5.0625, "learning_rate": 1.987085872910441e-05, "loss": 1.5278, "step": 3187 }, { "epoch": 7.97, "grad_norm": 7.34375, "learning_rate": 1.9870728991068636e-05, "loss": 1.6036, "step": 3188 }, { "epoch": 7.9725, "grad_norm": 5.34375, "learning_rate": 1.9870599188320757e-05, "loss": 1.5651, "step": 3189 }, { "epoch": 7.975, "grad_norm": 12.6875, "learning_rate": 1.987046932086163e-05, "loss": 1.5895, "step": 3190 }, { "epoch": 7.9775, "grad_norm": 7.625, "learning_rate": 1.9870339388692095e-05, "loss": 1.5654, "step": 3191 }, { "epoch": 7.98, "grad_norm": 5.4375, "learning_rate": 1.9870209391813013e-05, "loss": 1.4136, "step": 3192 }, { "epoch": 7.9825, "grad_norm": 5.0625, "learning_rate": 1.987007933022523e-05, "loss": 1.519, "step": 3193 }, { "epoch": 7.985, "grad_norm": 6.4375, "learning_rate": 1.9869949203929608e-05, "loss": 1.719, "step": 3194 }, { "epoch": 7.9875, "grad_norm": 5.78125, "learning_rate": 1.986981901292699e-05, "loss": 1.6324, "step": 3195 }, { "epoch": 7.99, "grad_norm": 11.0625, "learning_rate": 1.9869688757218233e-05, "loss": 1.5637, "step": 3196 }, { "epoch": 7.9925, "grad_norm": 7.4375, "learning_rate": 1.986955843680419e-05, "loss": 1.6867, "step": 3197 }, { "epoch": 7.995, "grad_norm": 8.4375, "learning_rate": 1.986942805168572e-05, "loss": 1.5477, "step": 3198 }, { "epoch": 7.9975000000000005, "grad_norm": 6.03125, "learning_rate": 1.9869297601863673e-05, "loss": 1.5873, "step": 3199 }, { "epoch": 8.0, "grad_norm": 5.8125, "learning_rate": 1.9869167087338908e-05, "loss": 1.5941, "step": 3200 }, { "epoch": 8.0025, "grad_norm": 4.625, "learning_rate": 1.9869036508112277e-05, "loss": 1.5326, "step": 3201 }, { "epoch": 8.005, "grad_norm": 6.59375, "learning_rate": 1.986890586418464e-05, "loss": 1.6143, "step": 3202 }, { "epoch": 8.0075, "grad_norm": 7.4375, "learning_rate": 1.986877515555685e-05, "loss": 1.5137, "step": 3203 }, { "epoch": 8.01, "grad_norm": 8.0625, "learning_rate": 1.9868644382229762e-05, "loss": 1.5395, "step": 3204 }, { "epoch": 8.0125, "grad_norm": 7.4375, "learning_rate": 1.986851354420424e-05, "loss": 1.7295, "step": 3205 }, { "epoch": 8.015, "grad_norm": 7.28125, "learning_rate": 1.9868382641481133e-05, "loss": 1.7365, "step": 3206 }, { "epoch": 8.0175, "grad_norm": 6.3125, "learning_rate": 1.9868251674061308e-05, "loss": 1.502, "step": 3207 }, { "epoch": 8.02, "grad_norm": 9.1875, "learning_rate": 1.9868120641945622e-05, "loss": 1.5911, "step": 3208 }, { "epoch": 8.0225, "grad_norm": 7.3125, "learning_rate": 1.9867989545134928e-05, "loss": 1.4776, "step": 3209 }, { "epoch": 8.025, "grad_norm": 6.0, "learning_rate": 1.986785838363009e-05, "loss": 1.6447, "step": 3210 }, { "epoch": 8.0275, "grad_norm": 6.53125, "learning_rate": 1.986772715743197e-05, "loss": 1.4991, "step": 3211 }, { "epoch": 8.03, "grad_norm": 11.4375, "learning_rate": 1.9867595866541423e-05, "loss": 1.5736, "step": 3212 }, { "epoch": 8.0325, "grad_norm": 8.5625, "learning_rate": 1.9867464510959314e-05, "loss": 1.6209, "step": 3213 }, { "epoch": 8.035, "grad_norm": 5.34375, "learning_rate": 1.98673330906865e-05, "loss": 1.4994, "step": 3214 }, { "epoch": 8.0375, "grad_norm": 6.8125, "learning_rate": 1.986720160572385e-05, "loss": 1.7304, "step": 3215 }, { "epoch": 8.04, "grad_norm": 8.25, "learning_rate": 1.9867070056072215e-05, "loss": 1.7559, "step": 3216 }, { "epoch": 8.0425, "grad_norm": 6.15625, "learning_rate": 1.9866938441732468e-05, "loss": 1.5967, "step": 3217 }, { "epoch": 8.045, "grad_norm": 5.3125, "learning_rate": 1.9866806762705465e-05, "loss": 1.5743, "step": 3218 }, { "epoch": 8.0475, "grad_norm": 7.25, "learning_rate": 1.9866675018992073e-05, "loss": 1.4798, "step": 3219 }, { "epoch": 8.05, "grad_norm": 6.59375, "learning_rate": 1.9866543210593154e-05, "loss": 1.4645, "step": 3220 }, { "epoch": 8.0525, "grad_norm": 6.0, "learning_rate": 1.9866411337509575e-05, "loss": 1.558, "step": 3221 }, { "epoch": 8.055, "grad_norm": 7.5, "learning_rate": 1.9866279399742196e-05, "loss": 1.719, "step": 3222 }, { "epoch": 8.0575, "grad_norm": 7.75, "learning_rate": 1.9866147397291884e-05, "loss": 1.7044, "step": 3223 }, { "epoch": 8.06, "grad_norm": 5.0625, "learning_rate": 1.9866015330159504e-05, "loss": 1.5308, "step": 3224 }, { "epoch": 8.0625, "grad_norm": 6.34375, "learning_rate": 1.986588319834592e-05, "loss": 1.4465, "step": 3225 }, { "epoch": 8.065, "grad_norm": 5.625, "learning_rate": 1.9865751001852005e-05, "loss": 1.3923, "step": 3226 }, { "epoch": 8.0675, "grad_norm": 5.65625, "learning_rate": 1.986561874067862e-05, "loss": 1.4787, "step": 3227 }, { "epoch": 8.07, "grad_norm": 6.125, "learning_rate": 1.9865486414826634e-05, "loss": 1.5317, "step": 3228 }, { "epoch": 8.0725, "grad_norm": 6.65625, "learning_rate": 1.9865354024296912e-05, "loss": 1.4933, "step": 3229 }, { "epoch": 8.075, "grad_norm": 6.40625, "learning_rate": 1.9865221569090326e-05, "loss": 1.5765, "step": 3230 }, { "epoch": 8.0775, "grad_norm": 6.84375, "learning_rate": 1.986508904920774e-05, "loss": 1.5454, "step": 3231 }, { "epoch": 8.08, "grad_norm": 5.125, "learning_rate": 1.9864956464650027e-05, "loss": 1.4412, "step": 3232 }, { "epoch": 8.0825, "grad_norm": 5.875, "learning_rate": 1.9864823815418052e-05, "loss": 1.4947, "step": 3233 }, { "epoch": 8.085, "grad_norm": 5.25, "learning_rate": 1.986469110151269e-05, "loss": 1.5319, "step": 3234 }, { "epoch": 8.0875, "grad_norm": 6.90625, "learning_rate": 1.9864558322934802e-05, "loss": 1.654, "step": 3235 }, { "epoch": 8.09, "grad_norm": 5.46875, "learning_rate": 1.986442547968527e-05, "loss": 1.5978, "step": 3236 }, { "epoch": 8.0925, "grad_norm": 6.75, "learning_rate": 1.9864292571764956e-05, "loss": 1.6882, "step": 3237 }, { "epoch": 8.095, "grad_norm": 7.03125, "learning_rate": 1.986415959917474e-05, "loss": 1.6626, "step": 3238 }, { "epoch": 8.0975, "grad_norm": 6.28125, "learning_rate": 1.986402656191548e-05, "loss": 1.5837, "step": 3239 }, { "epoch": 8.1, "grad_norm": 5.9375, "learning_rate": 1.986389345998806e-05, "loss": 1.5846, "step": 3240 }, { "epoch": 8.1025, "grad_norm": 4.78125, "learning_rate": 1.9863760293393353e-05, "loss": 1.4629, "step": 3241 }, { "epoch": 8.105, "grad_norm": 6.25, "learning_rate": 1.9863627062132226e-05, "loss": 1.4865, "step": 3242 }, { "epoch": 8.1075, "grad_norm": 7.21875, "learning_rate": 1.986349376620555e-05, "loss": 1.7011, "step": 3243 }, { "epoch": 8.11, "grad_norm": 5.03125, "learning_rate": 1.9863360405614213e-05, "loss": 1.5707, "step": 3244 }, { "epoch": 8.1125, "grad_norm": 5.6875, "learning_rate": 1.9863226980359075e-05, "loss": 1.5808, "step": 3245 }, { "epoch": 8.115, "grad_norm": 8.1875, "learning_rate": 1.986309349044101e-05, "loss": 1.61, "step": 3246 }, { "epoch": 8.1175, "grad_norm": 5.8125, "learning_rate": 1.9862959935860907e-05, "loss": 1.61, "step": 3247 }, { "epoch": 8.12, "grad_norm": 6.03125, "learning_rate": 1.986282631661963e-05, "loss": 1.5985, "step": 3248 }, { "epoch": 8.1225, "grad_norm": 6.40625, "learning_rate": 1.986269263271806e-05, "loss": 1.3612, "step": 3249 }, { "epoch": 8.125, "grad_norm": 6.0625, "learning_rate": 1.9862558884157067e-05, "loss": 1.5954, "step": 3250 }, { "epoch": 8.1275, "grad_norm": 5.1875, "learning_rate": 1.986242507093754e-05, "loss": 1.5323, "step": 3251 }, { "epoch": 8.13, "grad_norm": 8.125, "learning_rate": 1.9862291193060344e-05, "loss": 1.7517, "step": 3252 }, { "epoch": 8.1325, "grad_norm": 7.0625, "learning_rate": 1.9862157250526362e-05, "loss": 1.656, "step": 3253 }, { "epoch": 8.135, "grad_norm": 5.5, "learning_rate": 1.9862023243336472e-05, "loss": 1.4972, "step": 3254 }, { "epoch": 8.1375, "grad_norm": 8.625, "learning_rate": 1.9861889171491555e-05, "loss": 1.7283, "step": 3255 }, { "epoch": 8.14, "grad_norm": 6.90625, "learning_rate": 1.9861755034992483e-05, "loss": 1.4977, "step": 3256 }, { "epoch": 8.1425, "grad_norm": 11.0, "learning_rate": 1.9861620833840143e-05, "loss": 1.5053, "step": 3257 }, { "epoch": 8.145, "grad_norm": 6.3125, "learning_rate": 1.986148656803541e-05, "loss": 1.5348, "step": 3258 }, { "epoch": 8.1475, "grad_norm": 6.28125, "learning_rate": 1.9861352237579163e-05, "loss": 1.5555, "step": 3259 }, { "epoch": 8.15, "grad_norm": 7.71875, "learning_rate": 1.986121784247229e-05, "loss": 1.5565, "step": 3260 }, { "epoch": 8.1525, "grad_norm": 5.4375, "learning_rate": 1.9861083382715668e-05, "loss": 1.6481, "step": 3261 }, { "epoch": 8.155, "grad_norm": 9.125, "learning_rate": 1.986094885831017e-05, "loss": 1.6338, "step": 3262 }, { "epoch": 8.1575, "grad_norm": 7.375, "learning_rate": 1.9860814269256693e-05, "loss": 1.3824, "step": 3263 }, { "epoch": 8.16, "grad_norm": 7.59375, "learning_rate": 1.9860679615556112e-05, "loss": 1.5763, "step": 3264 }, { "epoch": 8.1625, "grad_norm": 7.6875, "learning_rate": 1.986054489720931e-05, "loss": 1.6455, "step": 3265 }, { "epoch": 8.165, "grad_norm": 6.875, "learning_rate": 1.986041011421717e-05, "loss": 1.6602, "step": 3266 }, { "epoch": 8.1675, "grad_norm": 6.9375, "learning_rate": 1.9860275266580573e-05, "loss": 1.7677, "step": 3267 }, { "epoch": 8.17, "grad_norm": 7.96875, "learning_rate": 1.9860140354300406e-05, "loss": 1.6716, "step": 3268 }, { "epoch": 8.1725, "grad_norm": 5.71875, "learning_rate": 1.9860005377377556e-05, "loss": 1.5506, "step": 3269 }, { "epoch": 8.175, "grad_norm": 9.6875, "learning_rate": 1.9859870335812904e-05, "loss": 1.6399, "step": 3270 }, { "epoch": 8.1775, "grad_norm": 4.46875, "learning_rate": 1.9859735229607337e-05, "loss": 1.5143, "step": 3271 }, { "epoch": 8.18, "grad_norm": 4.90625, "learning_rate": 1.985960005876174e-05, "loss": 1.5157, "step": 3272 }, { "epoch": 8.1825, "grad_norm": 4.6875, "learning_rate": 1.9859464823277e-05, "loss": 1.6068, "step": 3273 }, { "epoch": 8.185, "grad_norm": 8.125, "learning_rate": 1.9859329523154004e-05, "loss": 1.5246, "step": 3274 }, { "epoch": 8.1875, "grad_norm": 4.8125, "learning_rate": 1.985919415839364e-05, "loss": 1.6178, "step": 3275 }, { "epoch": 8.19, "grad_norm": 5.75, "learning_rate": 1.9859058728996788e-05, "loss": 1.4565, "step": 3276 }, { "epoch": 8.1925, "grad_norm": 8.4375, "learning_rate": 1.9858923234964344e-05, "loss": 1.5018, "step": 3277 }, { "epoch": 8.195, "grad_norm": 5.3125, "learning_rate": 1.9858787676297195e-05, "loss": 1.5433, "step": 3278 }, { "epoch": 8.1975, "grad_norm": 6.84375, "learning_rate": 1.9858652052996228e-05, "loss": 1.4782, "step": 3279 }, { "epoch": 8.2, "grad_norm": 10.75, "learning_rate": 1.9858516365062334e-05, "loss": 1.6052, "step": 3280 }, { "epoch": 8.2025, "grad_norm": 6.28125, "learning_rate": 1.98583806124964e-05, "loss": 1.645, "step": 3281 }, { "epoch": 8.205, "grad_norm": 5.25, "learning_rate": 1.985824479529932e-05, "loss": 1.5129, "step": 3282 }, { "epoch": 8.2075, "grad_norm": 6.28125, "learning_rate": 1.9858108913471973e-05, "loss": 1.4225, "step": 3283 }, { "epoch": 8.21, "grad_norm": 9.6875, "learning_rate": 1.9857972967015268e-05, "loss": 1.6666, "step": 3284 }, { "epoch": 8.2125, "grad_norm": 4.625, "learning_rate": 1.9857836955930085e-05, "loss": 1.4835, "step": 3285 }, { "epoch": 8.215, "grad_norm": 6.0, "learning_rate": 1.985770088021731e-05, "loss": 1.5246, "step": 3286 }, { "epoch": 8.2175, "grad_norm": 8.75, "learning_rate": 1.985756473987785e-05, "loss": 1.5315, "step": 3287 }, { "epoch": 8.22, "grad_norm": 6.53125, "learning_rate": 1.985742853491259e-05, "loss": 1.6024, "step": 3288 }, { "epoch": 8.2225, "grad_norm": 5.96875, "learning_rate": 1.985729226532242e-05, "loss": 1.7312, "step": 3289 }, { "epoch": 8.225, "grad_norm": 7.4375, "learning_rate": 1.9857155931108235e-05, "loss": 1.5602, "step": 3290 }, { "epoch": 8.2275, "grad_norm": 6.96875, "learning_rate": 1.9857019532270932e-05, "loss": 1.657, "step": 3291 }, { "epoch": 8.23, "grad_norm": 4.34375, "learning_rate": 1.9856883068811405e-05, "loss": 1.425, "step": 3292 }, { "epoch": 8.2325, "grad_norm": 4.28125, "learning_rate": 1.9856746540730545e-05, "loss": 1.4843, "step": 3293 }, { "epoch": 8.235, "grad_norm": 5.25, "learning_rate": 1.985660994802925e-05, "loss": 1.5297, "step": 3294 }, { "epoch": 8.2375, "grad_norm": 6.1875, "learning_rate": 1.9856473290708414e-05, "loss": 1.518, "step": 3295 }, { "epoch": 8.24, "grad_norm": 5.1875, "learning_rate": 1.9856336568768936e-05, "loss": 1.5326, "step": 3296 }, { "epoch": 8.2425, "grad_norm": 4.09375, "learning_rate": 1.9856199782211707e-05, "loss": 1.5321, "step": 3297 }, { "epoch": 8.245, "grad_norm": 5.25, "learning_rate": 1.9856062931037627e-05, "loss": 1.4276, "step": 3298 }, { "epoch": 8.2475, "grad_norm": 5.78125, "learning_rate": 1.9855926015247594e-05, "loss": 1.5808, "step": 3299 }, { "epoch": 8.25, "grad_norm": 5.15625, "learning_rate": 1.9855789034842504e-05, "loss": 1.4722, "step": 3300 }, { "epoch": 8.2525, "grad_norm": 6.25, "learning_rate": 1.9855651989823257e-05, "loss": 1.4474, "step": 3301 }, { "epoch": 8.255, "grad_norm": 5.125, "learning_rate": 1.9855514880190748e-05, "loss": 1.5482, "step": 3302 }, { "epoch": 8.2575, "grad_norm": 7.0625, "learning_rate": 1.9855377705945878e-05, "loss": 1.6959, "step": 3303 }, { "epoch": 8.26, "grad_norm": 8.375, "learning_rate": 1.9855240467089545e-05, "loss": 1.6075, "step": 3304 }, { "epoch": 8.2625, "grad_norm": 7.875, "learning_rate": 1.9855103163622653e-05, "loss": 1.5121, "step": 3305 }, { "epoch": 8.265, "grad_norm": 5.78125, "learning_rate": 1.9854965795546097e-05, "loss": 1.6214, "step": 3306 }, { "epoch": 8.2675, "grad_norm": 5.125, "learning_rate": 1.985482836286078e-05, "loss": 1.6794, "step": 3307 }, { "epoch": 8.27, "grad_norm": 6.75, "learning_rate": 1.9854690865567604e-05, "loss": 1.6729, "step": 3308 }, { "epoch": 8.2725, "grad_norm": 8.25, "learning_rate": 1.9854553303667464e-05, "loss": 1.5878, "step": 3309 }, { "epoch": 8.275, "grad_norm": 5.15625, "learning_rate": 1.985441567716127e-05, "loss": 1.6611, "step": 3310 }, { "epoch": 8.2775, "grad_norm": 6.65625, "learning_rate": 1.9854277986049923e-05, "loss": 1.7864, "step": 3311 }, { "epoch": 8.28, "grad_norm": 13.0, "learning_rate": 1.9854140230334323e-05, "loss": 1.5182, "step": 3312 }, { "epoch": 8.2825, "grad_norm": 6.96875, "learning_rate": 1.9854002410015374e-05, "loss": 1.4908, "step": 3313 }, { "epoch": 8.285, "grad_norm": 5.625, "learning_rate": 1.9853864525093977e-05, "loss": 1.6524, "step": 3314 }, { "epoch": 8.2875, "grad_norm": 6.5625, "learning_rate": 1.985372657557104e-05, "loss": 1.5796, "step": 3315 }, { "epoch": 8.29, "grad_norm": 5.71875, "learning_rate": 1.985358856144747e-05, "loss": 1.5614, "step": 3316 }, { "epoch": 8.2925, "grad_norm": 5.21875, "learning_rate": 1.985345048272416e-05, "loss": 1.5119, "step": 3317 }, { "epoch": 8.295, "grad_norm": 6.65625, "learning_rate": 1.9853312339402027e-05, "loss": 1.5584, "step": 3318 }, { "epoch": 8.2975, "grad_norm": 4.5, "learning_rate": 1.9853174131481977e-05, "loss": 1.4795, "step": 3319 }, { "epoch": 8.3, "grad_norm": 6.5, "learning_rate": 1.9853035858964907e-05, "loss": 1.5143, "step": 3320 }, { "epoch": 8.3025, "grad_norm": 4.25, "learning_rate": 1.985289752185173e-05, "loss": 1.5605, "step": 3321 }, { "epoch": 8.305, "grad_norm": 6.21875, "learning_rate": 1.9852759120143348e-05, "loss": 1.5366, "step": 3322 }, { "epoch": 8.3075, "grad_norm": 8.5, "learning_rate": 1.985262065384067e-05, "loss": 1.6416, "step": 3323 }, { "epoch": 8.31, "grad_norm": 5.15625, "learning_rate": 1.985248212294461e-05, "loss": 1.4982, "step": 3324 }, { "epoch": 8.3125, "grad_norm": 5.46875, "learning_rate": 1.9852343527456074e-05, "loss": 1.5258, "step": 3325 }, { "epoch": 8.315, "grad_norm": 5.75, "learning_rate": 1.9852204867375962e-05, "loss": 1.5888, "step": 3326 }, { "epoch": 8.3175, "grad_norm": 6.25, "learning_rate": 1.985206614270519e-05, "loss": 1.4989, "step": 3327 }, { "epoch": 8.32, "grad_norm": 5.1875, "learning_rate": 1.985192735344467e-05, "loss": 1.4656, "step": 3328 }, { "epoch": 8.3225, "grad_norm": 5.59375, "learning_rate": 1.9851788499595305e-05, "loss": 1.5909, "step": 3329 }, { "epoch": 8.325, "grad_norm": 6.53125, "learning_rate": 1.985164958115801e-05, "loss": 1.5307, "step": 3330 }, { "epoch": 8.3275, "grad_norm": 4.96875, "learning_rate": 1.9851510598133697e-05, "loss": 1.4714, "step": 3331 }, { "epoch": 8.33, "grad_norm": 6.46875, "learning_rate": 1.9851371550523272e-05, "loss": 1.613, "step": 3332 }, { "epoch": 8.3325, "grad_norm": 8.8125, "learning_rate": 1.985123243832765e-05, "loss": 1.6556, "step": 3333 }, { "epoch": 8.335, "grad_norm": 6.0, "learning_rate": 1.985109326154774e-05, "loss": 1.7274, "step": 3334 }, { "epoch": 8.3375, "grad_norm": 10.5, "learning_rate": 1.985095402018446e-05, "loss": 1.6142, "step": 3335 }, { "epoch": 8.34, "grad_norm": 7.09375, "learning_rate": 1.9850814714238718e-05, "loss": 1.5654, "step": 3336 }, { "epoch": 8.3425, "grad_norm": 6.9375, "learning_rate": 1.985067534371143e-05, "loss": 1.5061, "step": 3337 }, { "epoch": 8.345, "grad_norm": 6.59375, "learning_rate": 1.9850535908603506e-05, "loss": 1.6595, "step": 3338 }, { "epoch": 8.3475, "grad_norm": 5.8125, "learning_rate": 1.9850396408915866e-05, "loss": 1.513, "step": 3339 }, { "epoch": 8.35, "grad_norm": 7.96875, "learning_rate": 1.9850256844649422e-05, "loss": 1.8618, "step": 3340 }, { "epoch": 8.3525, "grad_norm": 6.71875, "learning_rate": 1.9850117215805087e-05, "loss": 1.5011, "step": 3341 }, { "epoch": 8.355, "grad_norm": 17.875, "learning_rate": 1.9849977522383777e-05, "loss": 1.7849, "step": 3342 }, { "epoch": 8.3575, "grad_norm": 9.0625, "learning_rate": 1.9849837764386407e-05, "loss": 1.6399, "step": 3343 }, { "epoch": 8.36, "grad_norm": 4.84375, "learning_rate": 1.98496979418139e-05, "loss": 1.4961, "step": 3344 }, { "epoch": 8.3625, "grad_norm": 7.78125, "learning_rate": 1.9849558054667163e-05, "loss": 1.6397, "step": 3345 }, { "epoch": 8.365, "grad_norm": 5.625, "learning_rate": 1.984941810294712e-05, "loss": 1.417, "step": 3346 }, { "epoch": 8.3675, "grad_norm": 5.4375, "learning_rate": 1.9849278086654686e-05, "loss": 1.5104, "step": 3347 }, { "epoch": 8.37, "grad_norm": 7.21875, "learning_rate": 1.9849138005790775e-05, "loss": 1.528, "step": 3348 }, { "epoch": 8.3725, "grad_norm": 7.6875, "learning_rate": 1.9848997860356312e-05, "loss": 1.5548, "step": 3349 }, { "epoch": 8.375, "grad_norm": 10.375, "learning_rate": 1.9848857650352213e-05, "loss": 1.6132, "step": 3350 }, { "epoch": 8.3775, "grad_norm": 6.875, "learning_rate": 1.98487173757794e-05, "loss": 1.4409, "step": 3351 }, { "epoch": 8.38, "grad_norm": 7.65625, "learning_rate": 1.984857703663879e-05, "loss": 1.5254, "step": 3352 }, { "epoch": 8.3825, "grad_norm": 6.78125, "learning_rate": 1.98484366329313e-05, "loss": 1.4386, "step": 3353 }, { "epoch": 8.385, "grad_norm": 6.03125, "learning_rate": 1.9848296164657852e-05, "loss": 1.4058, "step": 3354 }, { "epoch": 8.3875, "grad_norm": 8.6875, "learning_rate": 1.984815563181937e-05, "loss": 1.784, "step": 3355 }, { "epoch": 8.39, "grad_norm": 5.625, "learning_rate": 1.9848015034416776e-05, "loss": 1.4814, "step": 3356 }, { "epoch": 8.3925, "grad_norm": 8.625, "learning_rate": 1.9847874372450987e-05, "loss": 1.5298, "step": 3357 }, { "epoch": 8.395, "grad_norm": 7.0, "learning_rate": 1.984773364592293e-05, "loss": 1.5544, "step": 3358 }, { "epoch": 8.3975, "grad_norm": 6.21875, "learning_rate": 1.9847592854833525e-05, "loss": 1.5849, "step": 3359 }, { "epoch": 8.4, "grad_norm": 4.625, "learning_rate": 1.9847451999183692e-05, "loss": 1.5131, "step": 3360 }, { "epoch": 8.4025, "grad_norm": 5.875, "learning_rate": 1.984731107897436e-05, "loss": 1.4485, "step": 3361 }, { "epoch": 8.405, "grad_norm": 7.46875, "learning_rate": 1.9847170094206452e-05, "loss": 1.4711, "step": 3362 }, { "epoch": 8.4075, "grad_norm": 8.6875, "learning_rate": 1.984702904488089e-05, "loss": 1.5073, "step": 3363 }, { "epoch": 8.41, "grad_norm": 5.46875, "learning_rate": 1.98468879309986e-05, "loss": 1.5697, "step": 3364 }, { "epoch": 8.4125, "grad_norm": 5.9375, "learning_rate": 1.9846746752560507e-05, "loss": 1.6716, "step": 3365 }, { "epoch": 8.415, "grad_norm": 7.84375, "learning_rate": 1.9846605509567534e-05, "loss": 1.6456, "step": 3366 }, { "epoch": 8.4175, "grad_norm": 5.4375, "learning_rate": 1.9846464202020613e-05, "loss": 1.531, "step": 3367 }, { "epoch": 8.42, "grad_norm": 6.84375, "learning_rate": 1.9846322829920663e-05, "loss": 1.6502, "step": 3368 }, { "epoch": 8.4225, "grad_norm": 5.125, "learning_rate": 1.9846181393268617e-05, "loss": 1.5528, "step": 3369 }, { "epoch": 8.425, "grad_norm": 6.75, "learning_rate": 1.98460398920654e-05, "loss": 1.5888, "step": 3370 }, { "epoch": 8.4275, "grad_norm": 8.5625, "learning_rate": 1.9845898326311935e-05, "loss": 1.5927, "step": 3371 }, { "epoch": 8.43, "grad_norm": 6.25, "learning_rate": 1.984575669600916e-05, "loss": 1.4945, "step": 3372 }, { "epoch": 8.4325, "grad_norm": 6.0, "learning_rate": 1.9845615001157995e-05, "loss": 1.6035, "step": 3373 }, { "epoch": 8.435, "grad_norm": 5.625, "learning_rate": 1.9845473241759376e-05, "loss": 1.47, "step": 3374 }, { "epoch": 8.4375, "grad_norm": 6.46875, "learning_rate": 1.9845331417814223e-05, "loss": 1.6693, "step": 3375 }, { "epoch": 8.44, "grad_norm": 8.375, "learning_rate": 1.9845189529323473e-05, "loss": 1.4591, "step": 3376 }, { "epoch": 8.4425, "grad_norm": 5.84375, "learning_rate": 1.9845047576288057e-05, "loss": 1.7785, "step": 3377 }, { "epoch": 8.445, "grad_norm": 6.65625, "learning_rate": 1.98449055587089e-05, "loss": 1.542, "step": 3378 }, { "epoch": 8.4475, "grad_norm": 6.90625, "learning_rate": 1.984476347658694e-05, "loss": 1.51, "step": 3379 }, { "epoch": 8.45, "grad_norm": 7.40625, "learning_rate": 1.98446213299231e-05, "loss": 1.6231, "step": 3380 }, { "epoch": 8.4525, "grad_norm": 7.375, "learning_rate": 1.9844479118718317e-05, "loss": 1.7124, "step": 3381 }, { "epoch": 8.455, "grad_norm": 9.125, "learning_rate": 1.984433684297353e-05, "loss": 1.7218, "step": 3382 }, { "epoch": 8.4575, "grad_norm": 6.25, "learning_rate": 1.984419450268966e-05, "loss": 1.46, "step": 3383 }, { "epoch": 8.46, "grad_norm": 14.125, "learning_rate": 1.9844052097867638e-05, "loss": 1.4607, "step": 3384 }, { "epoch": 8.4625, "grad_norm": 5.75, "learning_rate": 1.9843909628508414e-05, "loss": 1.5903, "step": 3385 }, { "epoch": 8.465, "grad_norm": 5.5, "learning_rate": 1.9843767094612907e-05, "loss": 1.5837, "step": 3386 }, { "epoch": 8.4675, "grad_norm": 6.21875, "learning_rate": 1.9843624496182058e-05, "loss": 1.4416, "step": 3387 }, { "epoch": 8.47, "grad_norm": 4.875, "learning_rate": 1.98434818332168e-05, "loss": 1.62, "step": 3388 }, { "epoch": 8.4725, "grad_norm": 6.625, "learning_rate": 1.984333910571807e-05, "loss": 1.6292, "step": 3389 }, { "epoch": 8.475, "grad_norm": 6.40625, "learning_rate": 1.9843196313686803e-05, "loss": 1.5219, "step": 3390 }, { "epoch": 8.4775, "grad_norm": 5.15625, "learning_rate": 1.9843053457123937e-05, "loss": 1.4903, "step": 3391 }, { "epoch": 8.48, "grad_norm": 4.59375, "learning_rate": 1.98429105360304e-05, "loss": 1.5328, "step": 3392 }, { "epoch": 8.4825, "grad_norm": 8.375, "learning_rate": 1.9842767550407144e-05, "loss": 1.7211, "step": 3393 }, { "epoch": 8.485, "grad_norm": 7.125, "learning_rate": 1.984262450025509e-05, "loss": 1.7137, "step": 3394 }, { "epoch": 8.4875, "grad_norm": 5.5, "learning_rate": 1.984248138557519e-05, "loss": 1.5713, "step": 3395 }, { "epoch": 8.49, "grad_norm": 5.4375, "learning_rate": 1.9842338206368375e-05, "loss": 1.5303, "step": 3396 }, { "epoch": 8.4925, "grad_norm": 4.5, "learning_rate": 1.984219496263558e-05, "loss": 1.5384, "step": 3397 }, { "epoch": 8.495, "grad_norm": 8.625, "learning_rate": 1.9842051654377753e-05, "loss": 1.5814, "step": 3398 }, { "epoch": 8.4975, "grad_norm": 5.03125, "learning_rate": 1.9841908281595828e-05, "loss": 1.7109, "step": 3399 }, { "epoch": 8.5, "grad_norm": 7.21875, "learning_rate": 1.9841764844290744e-05, "loss": 1.5157, "step": 3400 }, { "epoch": 8.5025, "grad_norm": 7.4375, "learning_rate": 1.9841621342463444e-05, "loss": 1.6777, "step": 3401 }, { "epoch": 8.505, "grad_norm": 7.9375, "learning_rate": 1.984147777611487e-05, "loss": 1.6507, "step": 3402 }, { "epoch": 8.5075, "grad_norm": 7.0625, "learning_rate": 1.984133414524596e-05, "loss": 1.5847, "step": 3403 }, { "epoch": 8.51, "grad_norm": 6.625, "learning_rate": 1.9841190449857662e-05, "loss": 1.469, "step": 3404 }, { "epoch": 8.5125, "grad_norm": 13.375, "learning_rate": 1.984104668995091e-05, "loss": 1.5026, "step": 3405 }, { "epoch": 8.515, "grad_norm": 10.5625, "learning_rate": 1.9840902865526645e-05, "loss": 1.5374, "step": 3406 }, { "epoch": 8.5175, "grad_norm": 5.15625, "learning_rate": 1.984075897658582e-05, "loss": 1.5165, "step": 3407 }, { "epoch": 8.52, "grad_norm": 6.625, "learning_rate": 1.9840615023129372e-05, "loss": 1.4832, "step": 3408 }, { "epoch": 8.5225, "grad_norm": 8.4375, "learning_rate": 1.9840471005158246e-05, "loss": 1.3902, "step": 3409 }, { "epoch": 8.525, "grad_norm": 8.6875, "learning_rate": 1.9840326922673384e-05, "loss": 1.6307, "step": 3410 }, { "epoch": 8.5275, "grad_norm": 10.0625, "learning_rate": 1.9840182775675735e-05, "loss": 1.602, "step": 3411 }, { "epoch": 8.53, "grad_norm": 6.78125, "learning_rate": 1.984003856416624e-05, "loss": 1.4642, "step": 3412 }, { "epoch": 8.5325, "grad_norm": 3.890625, "learning_rate": 1.9839894288145846e-05, "loss": 1.4296, "step": 3413 }, { "epoch": 8.535, "grad_norm": 5.6875, "learning_rate": 1.98397499476155e-05, "loss": 1.6535, "step": 3414 }, { "epoch": 8.5375, "grad_norm": 9.5625, "learning_rate": 1.9839605542576146e-05, "loss": 1.6763, "step": 3415 }, { "epoch": 8.54, "grad_norm": 5.25, "learning_rate": 1.9839461073028733e-05, "loss": 1.5361, "step": 3416 }, { "epoch": 8.5425, "grad_norm": 5.8125, "learning_rate": 1.9839316538974205e-05, "loss": 1.5993, "step": 3417 }, { "epoch": 8.545, "grad_norm": 4.875, "learning_rate": 1.9839171940413514e-05, "loss": 1.4768, "step": 3418 }, { "epoch": 8.5475, "grad_norm": 9.25, "learning_rate": 1.98390272773476e-05, "loss": 1.5058, "step": 3419 }, { "epoch": 8.55, "grad_norm": 8.125, "learning_rate": 1.9838882549777426e-05, "loss": 1.5591, "step": 3420 }, { "epoch": 8.5525, "grad_norm": 8.0, "learning_rate": 1.9838737757703926e-05, "loss": 1.5373, "step": 3421 }, { "epoch": 8.555, "grad_norm": 5.5, "learning_rate": 1.9838592901128054e-05, "loss": 1.4629, "step": 3422 }, { "epoch": 8.5575, "grad_norm": 6.0625, "learning_rate": 1.9838447980050762e-05, "loss": 1.4507, "step": 3423 }, { "epoch": 8.56, "grad_norm": 5.6875, "learning_rate": 1.9838302994473e-05, "loss": 1.5444, "step": 3424 }, { "epoch": 8.5625, "grad_norm": 4.6875, "learning_rate": 1.9838157944395713e-05, "loss": 1.5984, "step": 3425 }, { "epoch": 8.565, "grad_norm": 5.78125, "learning_rate": 1.983801282981986e-05, "loss": 1.6027, "step": 3426 }, { "epoch": 8.5675, "grad_norm": 4.375, "learning_rate": 1.983786765074639e-05, "loss": 1.547, "step": 3427 }, { "epoch": 8.57, "grad_norm": 8.125, "learning_rate": 1.983772240717625e-05, "loss": 1.5315, "step": 3428 }, { "epoch": 8.5725, "grad_norm": 8.6875, "learning_rate": 1.98375770991104e-05, "loss": 1.5891, "step": 3429 }, { "epoch": 8.575, "grad_norm": 6.25, "learning_rate": 1.9837431726549784e-05, "loss": 1.4068, "step": 3430 }, { "epoch": 8.5775, "grad_norm": 6.8125, "learning_rate": 1.9837286289495362e-05, "loss": 1.5435, "step": 3431 }, { "epoch": 8.58, "grad_norm": 7.65625, "learning_rate": 1.9837140787948082e-05, "loss": 1.5646, "step": 3432 }, { "epoch": 8.5825, "grad_norm": 8.3125, "learning_rate": 1.9836995221908903e-05, "loss": 1.4579, "step": 3433 }, { "epoch": 8.585, "grad_norm": 7.8125, "learning_rate": 1.9836849591378777e-05, "loss": 1.5799, "step": 3434 }, { "epoch": 8.5875, "grad_norm": 11.25, "learning_rate": 1.9836703896358655e-05, "loss": 1.53, "step": 3435 }, { "epoch": 8.59, "grad_norm": 7.96875, "learning_rate": 1.98365581368495e-05, "loss": 1.6163, "step": 3436 }, { "epoch": 8.5925, "grad_norm": 6.5, "learning_rate": 1.983641231285226e-05, "loss": 1.6004, "step": 3437 }, { "epoch": 8.595, "grad_norm": 7.78125, "learning_rate": 1.98362664243679e-05, "loss": 1.4476, "step": 3438 }, { "epoch": 8.5975, "grad_norm": 5.03125, "learning_rate": 1.9836120471397365e-05, "loss": 1.3215, "step": 3439 }, { "epoch": 8.6, "grad_norm": 6.125, "learning_rate": 1.9835974453941623e-05, "loss": 1.465, "step": 3440 }, { "epoch": 8.6025, "grad_norm": 6.0, "learning_rate": 1.983582837200162e-05, "loss": 1.5889, "step": 3441 }, { "epoch": 8.605, "grad_norm": 6.46875, "learning_rate": 1.9835682225578323e-05, "loss": 1.6546, "step": 3442 }, { "epoch": 8.6075, "grad_norm": 6.65625, "learning_rate": 1.9835536014672686e-05, "loss": 1.6346, "step": 3443 }, { "epoch": 8.61, "grad_norm": 5.875, "learning_rate": 1.983538973928567e-05, "loss": 1.5205, "step": 3444 }, { "epoch": 8.6125, "grad_norm": 4.6875, "learning_rate": 1.9835243399418234e-05, "loss": 1.6416, "step": 3445 }, { "epoch": 8.615, "grad_norm": 7.09375, "learning_rate": 1.9835096995071333e-05, "loss": 1.6272, "step": 3446 }, { "epoch": 8.6175, "grad_norm": 6.8125, "learning_rate": 1.9834950526245933e-05, "loss": 1.5986, "step": 3447 }, { "epoch": 8.62, "grad_norm": 6.875, "learning_rate": 1.983480399294299e-05, "loss": 1.5455, "step": 3448 }, { "epoch": 8.6225, "grad_norm": 6.65625, "learning_rate": 1.9834657395163463e-05, "loss": 1.6505, "step": 3449 }, { "epoch": 8.625, "grad_norm": 6.28125, "learning_rate": 1.9834510732908314e-05, "loss": 1.5175, "step": 3450 }, { "epoch": 8.6275, "grad_norm": 5.625, "learning_rate": 1.983436400617851e-05, "loss": 1.5289, "step": 3451 }, { "epoch": 8.63, "grad_norm": 7.4375, "learning_rate": 1.9834217214975008e-05, "loss": 1.6415, "step": 3452 }, { "epoch": 8.6325, "grad_norm": 6.1875, "learning_rate": 1.983407035929877e-05, "loss": 1.5005, "step": 3453 }, { "epoch": 8.635, "grad_norm": 8.875, "learning_rate": 1.9833923439150763e-05, "loss": 1.5058, "step": 3454 }, { "epoch": 8.6375, "grad_norm": 6.4375, "learning_rate": 1.9833776454531948e-05, "loss": 1.6775, "step": 3455 }, { "epoch": 8.64, "grad_norm": 5.09375, "learning_rate": 1.9833629405443283e-05, "loss": 1.6132, "step": 3456 }, { "epoch": 8.6425, "grad_norm": 5.84375, "learning_rate": 1.9833482291885744e-05, "loss": 1.5123, "step": 3457 }, { "epoch": 8.645, "grad_norm": 5.15625, "learning_rate": 1.9833335113860284e-05, "loss": 1.5838, "step": 3458 }, { "epoch": 8.6475, "grad_norm": 6.53125, "learning_rate": 1.9833187871367875e-05, "loss": 1.4814, "step": 3459 }, { "epoch": 8.65, "grad_norm": 6.03125, "learning_rate": 1.983304056440948e-05, "loss": 1.518, "step": 3460 }, { "epoch": 8.6525, "grad_norm": 5.28125, "learning_rate": 1.983289319298606e-05, "loss": 1.4947, "step": 3461 }, { "epoch": 8.655, "grad_norm": 7.34375, "learning_rate": 1.9832745757098588e-05, "loss": 1.5482, "step": 3462 }, { "epoch": 8.6575, "grad_norm": 6.25, "learning_rate": 1.983259825674803e-05, "loss": 1.754, "step": 3463 }, { "epoch": 8.66, "grad_norm": 6.78125, "learning_rate": 1.9832450691935353e-05, "loss": 1.4817, "step": 3464 }, { "epoch": 8.6625, "grad_norm": 4.625, "learning_rate": 1.9832303062661517e-05, "loss": 1.5027, "step": 3465 }, { "epoch": 8.665, "grad_norm": 7.375, "learning_rate": 1.98321553689275e-05, "loss": 1.5432, "step": 3466 }, { "epoch": 8.6675, "grad_norm": 6.03125, "learning_rate": 1.9832007610734264e-05, "loss": 1.4704, "step": 3467 }, { "epoch": 8.67, "grad_norm": 6.6875, "learning_rate": 1.9831859788082778e-05, "loss": 1.615, "step": 3468 }, { "epoch": 8.6725, "grad_norm": 6.03125, "learning_rate": 1.9831711900974016e-05, "loss": 1.8524, "step": 3469 }, { "epoch": 8.675, "grad_norm": 6.21875, "learning_rate": 1.9831563949408945e-05, "loss": 1.61, "step": 3470 }, { "epoch": 8.6775, "grad_norm": 5.4375, "learning_rate": 1.983141593338853e-05, "loss": 1.4145, "step": 3471 }, { "epoch": 8.68, "grad_norm": 5.8125, "learning_rate": 1.983126785291375e-05, "loss": 1.5931, "step": 3472 }, { "epoch": 8.6825, "grad_norm": 5.84375, "learning_rate": 1.983111970798557e-05, "loss": 1.5976, "step": 3473 }, { "epoch": 8.685, "grad_norm": 6.28125, "learning_rate": 1.983097149860496e-05, "loss": 1.5339, "step": 3474 }, { "epoch": 8.6875, "grad_norm": 7.125, "learning_rate": 1.9830823224772894e-05, "loss": 1.6108, "step": 3475 }, { "epoch": 8.69, "grad_norm": 8.8125, "learning_rate": 1.983067488649035e-05, "loss": 1.5713, "step": 3476 }, { "epoch": 8.6925, "grad_norm": 7.1875, "learning_rate": 1.983052648375829e-05, "loss": 1.4424, "step": 3477 }, { "epoch": 8.695, "grad_norm": 5.75, "learning_rate": 1.983037801657769e-05, "loss": 1.525, "step": 3478 }, { "epoch": 8.6975, "grad_norm": 4.5625, "learning_rate": 1.983022948494953e-05, "loss": 1.6286, "step": 3479 }, { "epoch": 8.7, "grad_norm": 6.75, "learning_rate": 1.983008088887478e-05, "loss": 1.5629, "step": 3480 }, { "epoch": 8.7025, "grad_norm": 9.1875, "learning_rate": 1.982993222835441e-05, "loss": 1.732, "step": 3481 }, { "epoch": 8.705, "grad_norm": 8.5625, "learning_rate": 1.9829783503389395e-05, "loss": 1.4994, "step": 3482 }, { "epoch": 8.7075, "grad_norm": 6.03125, "learning_rate": 1.9829634713980717e-05, "loss": 1.37, "step": 3483 }, { "epoch": 8.71, "grad_norm": 4.9375, "learning_rate": 1.9829485860129347e-05, "loss": 1.4597, "step": 3484 }, { "epoch": 8.7125, "grad_norm": 12.75, "learning_rate": 1.9829336941836258e-05, "loss": 1.4881, "step": 3485 }, { "epoch": 8.715, "grad_norm": 8.8125, "learning_rate": 1.982918795910243e-05, "loss": 1.6677, "step": 3486 }, { "epoch": 8.7175, "grad_norm": 7.15625, "learning_rate": 1.982903891192884e-05, "loss": 1.5402, "step": 3487 }, { "epoch": 8.72, "grad_norm": 5.5, "learning_rate": 1.9828889800316467e-05, "loss": 1.5296, "step": 3488 }, { "epoch": 8.7225, "grad_norm": 6.40625, "learning_rate": 1.9828740624266282e-05, "loss": 1.4534, "step": 3489 }, { "epoch": 8.725, "grad_norm": 6.0, "learning_rate": 1.982859138377927e-05, "loss": 1.5996, "step": 3490 }, { "epoch": 8.7275, "grad_norm": 10.6875, "learning_rate": 1.98284420788564e-05, "loss": 1.2926, "step": 3491 }, { "epoch": 8.73, "grad_norm": 7.125, "learning_rate": 1.9828292709498665e-05, "loss": 1.6117, "step": 3492 }, { "epoch": 8.7325, "grad_norm": 5.875, "learning_rate": 1.982814327570703e-05, "loss": 1.462, "step": 3493 }, { "epoch": 8.735, "grad_norm": 4.90625, "learning_rate": 1.9827993777482484e-05, "loss": 1.5238, "step": 3494 }, { "epoch": 8.7375, "grad_norm": 8.8125, "learning_rate": 1.9827844214826e-05, "loss": 1.7285, "step": 3495 }, { "epoch": 8.74, "grad_norm": 7.21875, "learning_rate": 1.982769458773857e-05, "loss": 1.5789, "step": 3496 }, { "epoch": 8.7425, "grad_norm": 10.5, "learning_rate": 1.9827544896221163e-05, "loss": 1.3983, "step": 3497 }, { "epoch": 8.745, "grad_norm": 7.40625, "learning_rate": 1.9827395140274767e-05, "loss": 1.6409, "step": 3498 }, { "epoch": 8.7475, "grad_norm": 7.4375, "learning_rate": 1.9827245319900358e-05, "loss": 1.6345, "step": 3499 }, { "epoch": 8.75, "grad_norm": 7.96875, "learning_rate": 1.9827095435098926e-05, "loss": 1.4679, "step": 3500 }, { "epoch": 8.7525, "grad_norm": 5.40625, "learning_rate": 1.9826945485871446e-05, "loss": 1.4079, "step": 3501 }, { "epoch": 8.755, "grad_norm": 6.65625, "learning_rate": 1.9826795472218907e-05, "loss": 1.4507, "step": 3502 }, { "epoch": 8.7575, "grad_norm": 10.375, "learning_rate": 1.9826645394142293e-05, "loss": 1.4799, "step": 3503 }, { "epoch": 8.76, "grad_norm": 9.25, "learning_rate": 1.982649525164258e-05, "loss": 1.7477, "step": 3504 }, { "epoch": 8.7625, "grad_norm": 8.4375, "learning_rate": 1.9826345044720758e-05, "loss": 1.63, "step": 3505 }, { "epoch": 8.765, "grad_norm": 5.84375, "learning_rate": 1.9826194773377815e-05, "loss": 1.6885, "step": 3506 }, { "epoch": 8.7675, "grad_norm": 17.625, "learning_rate": 1.9826044437614727e-05, "loss": 1.6538, "step": 3507 }, { "epoch": 8.77, "grad_norm": 6.15625, "learning_rate": 1.9825894037432486e-05, "loss": 1.6581, "step": 3508 }, { "epoch": 8.7725, "grad_norm": 10.4375, "learning_rate": 1.9825743572832082e-05, "loss": 1.5294, "step": 3509 }, { "epoch": 8.775, "grad_norm": 6.65625, "learning_rate": 1.9825593043814493e-05, "loss": 1.5389, "step": 3510 }, { "epoch": 8.7775, "grad_norm": 6.15625, "learning_rate": 1.9825442450380706e-05, "loss": 1.6415, "step": 3511 }, { "epoch": 8.78, "grad_norm": 5.46875, "learning_rate": 1.9825291792531717e-05, "loss": 1.5465, "step": 3512 }, { "epoch": 8.7825, "grad_norm": 5.90625, "learning_rate": 1.9825141070268505e-05, "loss": 1.5985, "step": 3513 }, { "epoch": 8.785, "grad_norm": 8.5625, "learning_rate": 1.9824990283592062e-05, "loss": 1.473, "step": 3514 }, { "epoch": 8.7875, "grad_norm": 5.84375, "learning_rate": 1.9824839432503375e-05, "loss": 1.7883, "step": 3515 }, { "epoch": 8.79, "grad_norm": 7.90625, "learning_rate": 1.9824688517003433e-05, "loss": 1.7207, "step": 3516 }, { "epoch": 8.7925, "grad_norm": 8.125, "learning_rate": 1.982453753709323e-05, "loss": 1.6324, "step": 3517 }, { "epoch": 8.795, "grad_norm": 6.3125, "learning_rate": 1.9824386492773748e-05, "loss": 1.4924, "step": 3518 }, { "epoch": 8.7975, "grad_norm": 9.5625, "learning_rate": 1.9824235384045983e-05, "loss": 1.6994, "step": 3519 }, { "epoch": 8.8, "grad_norm": 7.71875, "learning_rate": 1.9824084210910924e-05, "loss": 1.6141, "step": 3520 }, { "epoch": 8.8025, "grad_norm": 5.375, "learning_rate": 1.9823932973369564e-05, "loss": 1.5696, "step": 3521 }, { "epoch": 8.805, "grad_norm": 5.65625, "learning_rate": 1.982378167142289e-05, "loss": 1.4358, "step": 3522 }, { "epoch": 8.8075, "grad_norm": 5.4375, "learning_rate": 1.9823630305071896e-05, "loss": 1.6568, "step": 3523 }, { "epoch": 8.81, "grad_norm": 6.625, "learning_rate": 1.982347887431758e-05, "loss": 1.5252, "step": 3524 }, { "epoch": 8.8125, "grad_norm": 5.25, "learning_rate": 1.9823327379160923e-05, "loss": 1.5382, "step": 3525 }, { "epoch": 8.815, "grad_norm": 6.9375, "learning_rate": 1.9823175819602926e-05, "loss": 1.4184, "step": 3526 }, { "epoch": 8.8175, "grad_norm": 6.71875, "learning_rate": 1.9823024195644586e-05, "loss": 1.5494, "step": 3527 }, { "epoch": 8.82, "grad_norm": 5.71875, "learning_rate": 1.982287250728689e-05, "loss": 1.5561, "step": 3528 }, { "epoch": 8.8225, "grad_norm": 7.46875, "learning_rate": 1.9822720754530834e-05, "loss": 1.5671, "step": 3529 }, { "epoch": 8.825, "grad_norm": 6.3125, "learning_rate": 1.982256893737741e-05, "loss": 1.4871, "step": 3530 }, { "epoch": 8.8275, "grad_norm": 7.9375, "learning_rate": 1.982241705582762e-05, "loss": 1.6598, "step": 3531 }, { "epoch": 8.83, "grad_norm": 7.65625, "learning_rate": 1.9822265109882463e-05, "loss": 1.5419, "step": 3532 }, { "epoch": 8.8325, "grad_norm": 7.125, "learning_rate": 1.9822113099542922e-05, "loss": 1.6822, "step": 3533 }, { "epoch": 8.835, "grad_norm": 8.75, "learning_rate": 1.9821961024810003e-05, "loss": 1.6202, "step": 3534 }, { "epoch": 8.8375, "grad_norm": 7.125, "learning_rate": 1.9821808885684703e-05, "loss": 1.6037, "step": 3535 }, { "epoch": 8.84, "grad_norm": 7.96875, "learning_rate": 1.9821656682168013e-05, "loss": 1.6291, "step": 3536 }, { "epoch": 8.8425, "grad_norm": 6.9375, "learning_rate": 1.9821504414260935e-05, "loss": 1.4453, "step": 3537 }, { "epoch": 8.845, "grad_norm": 5.40625, "learning_rate": 1.982135208196447e-05, "loss": 1.632, "step": 3538 }, { "epoch": 8.8475, "grad_norm": 7.46875, "learning_rate": 1.9821199685279612e-05, "loss": 1.5916, "step": 3539 }, { "epoch": 8.85, "grad_norm": 6.0, "learning_rate": 1.9821047224207362e-05, "loss": 1.6586, "step": 3540 }, { "epoch": 8.8525, "grad_norm": 6.5, "learning_rate": 1.982089469874872e-05, "loss": 1.4287, "step": 3541 }, { "epoch": 8.855, "grad_norm": 8.125, "learning_rate": 1.9820742108904683e-05, "loss": 1.6738, "step": 3542 }, { "epoch": 8.8575, "grad_norm": 4.84375, "learning_rate": 1.9820589454676256e-05, "loss": 1.4365, "step": 3543 }, { "epoch": 8.86, "grad_norm": 4.40625, "learning_rate": 1.9820436736064437e-05, "loss": 1.5237, "step": 3544 }, { "epoch": 8.8625, "grad_norm": 5.6875, "learning_rate": 1.9820283953070228e-05, "loss": 1.5817, "step": 3545 }, { "epoch": 8.865, "grad_norm": 6.90625, "learning_rate": 1.982013110569463e-05, "loss": 1.5946, "step": 3546 }, { "epoch": 8.8675, "grad_norm": 5.84375, "learning_rate": 1.9819978193938645e-05, "loss": 1.4738, "step": 3547 }, { "epoch": 8.87, "grad_norm": 9.0625, "learning_rate": 1.9819825217803275e-05, "loss": 1.5505, "step": 3548 }, { "epoch": 8.8725, "grad_norm": 6.46875, "learning_rate": 1.9819672177289525e-05, "loss": 1.554, "step": 3549 }, { "epoch": 8.875, "grad_norm": 4.375, "learning_rate": 1.9819519072398397e-05, "loss": 1.5069, "step": 3550 }, { "epoch": 8.8775, "grad_norm": 6.46875, "learning_rate": 1.98193659031309e-05, "loss": 1.4867, "step": 3551 }, { "epoch": 8.88, "grad_norm": 6.90625, "learning_rate": 1.9819212669488026e-05, "loss": 1.4966, "step": 3552 }, { "epoch": 8.8825, "grad_norm": 5.84375, "learning_rate": 1.981905937147079e-05, "loss": 1.6005, "step": 3553 }, { "epoch": 8.885, "grad_norm": 6.15625, "learning_rate": 1.981890600908019e-05, "loss": 1.459, "step": 3554 }, { "epoch": 8.8875, "grad_norm": 5.0625, "learning_rate": 1.981875258231724e-05, "loss": 1.3846, "step": 3555 }, { "epoch": 8.89, "grad_norm": 6.71875, "learning_rate": 1.981859909118294e-05, "loss": 1.3987, "step": 3556 }, { "epoch": 8.8925, "grad_norm": 5.15625, "learning_rate": 1.9818445535678296e-05, "loss": 1.4843, "step": 3557 }, { "epoch": 8.895, "grad_norm": 5.15625, "learning_rate": 1.9818291915804316e-05, "loss": 1.6291, "step": 3558 }, { "epoch": 8.8975, "grad_norm": 5.125, "learning_rate": 1.9818138231562006e-05, "loss": 1.4037, "step": 3559 }, { "epoch": 8.9, "grad_norm": 6.4375, "learning_rate": 1.9817984482952378e-05, "loss": 1.5223, "step": 3560 }, { "epoch": 8.9025, "grad_norm": 6.46875, "learning_rate": 1.981783066997643e-05, "loss": 1.6615, "step": 3561 }, { "epoch": 8.905, "grad_norm": 6.71875, "learning_rate": 1.9817676792635186e-05, "loss": 1.5884, "step": 3562 }, { "epoch": 8.9075, "grad_norm": 6.90625, "learning_rate": 1.981752285092964e-05, "loss": 1.7894, "step": 3563 }, { "epoch": 8.91, "grad_norm": 7.46875, "learning_rate": 1.981736884486081e-05, "loss": 1.4817, "step": 3564 }, { "epoch": 8.9125, "grad_norm": 5.8125, "learning_rate": 1.98172147744297e-05, "loss": 1.7092, "step": 3565 }, { "epoch": 8.915, "grad_norm": 5.59375, "learning_rate": 1.9817060639637324e-05, "loss": 1.5309, "step": 3566 }, { "epoch": 8.9175, "grad_norm": 5.84375, "learning_rate": 1.9816906440484695e-05, "loss": 1.551, "step": 3567 }, { "epoch": 8.92, "grad_norm": 6.21875, "learning_rate": 1.9816752176972815e-05, "loss": 1.5903, "step": 3568 }, { "epoch": 8.9225, "grad_norm": 5.21875, "learning_rate": 1.98165978491027e-05, "loss": 1.5278, "step": 3569 }, { "epoch": 8.925, "grad_norm": 6.4375, "learning_rate": 1.9816443456875365e-05, "loss": 1.7762, "step": 3570 }, { "epoch": 8.9275, "grad_norm": 7.78125, "learning_rate": 1.981628900029182e-05, "loss": 1.5446, "step": 3571 }, { "epoch": 8.93, "grad_norm": 6.59375, "learning_rate": 1.9816134479353076e-05, "loss": 1.6448, "step": 3572 }, { "epoch": 8.9325, "grad_norm": 4.5, "learning_rate": 1.9815979894060146e-05, "loss": 1.3747, "step": 3573 }, { "epoch": 8.935, "grad_norm": 5.5, "learning_rate": 1.9815825244414046e-05, "loss": 1.4895, "step": 3574 }, { "epoch": 8.9375, "grad_norm": 6.34375, "learning_rate": 1.9815670530415788e-05, "loss": 1.641, "step": 3575 }, { "epoch": 8.94, "grad_norm": 11.6875, "learning_rate": 1.9815515752066386e-05, "loss": 1.4655, "step": 3576 }, { "epoch": 8.9425, "grad_norm": 6.78125, "learning_rate": 1.9815360909366858e-05, "loss": 1.559, "step": 3577 }, { "epoch": 8.945, "grad_norm": 5.34375, "learning_rate": 1.9815206002318217e-05, "loss": 1.5767, "step": 3578 }, { "epoch": 8.9475, "grad_norm": 6.8125, "learning_rate": 1.981505103092148e-05, "loss": 1.5395, "step": 3579 }, { "epoch": 8.95, "grad_norm": 8.375, "learning_rate": 1.9814895995177653e-05, "loss": 1.6144, "step": 3580 }, { "epoch": 8.9525, "grad_norm": 4.375, "learning_rate": 1.9814740895087764e-05, "loss": 1.4178, "step": 3581 }, { "epoch": 8.955, "grad_norm": 5.53125, "learning_rate": 1.981458573065283e-05, "loss": 1.5252, "step": 3582 }, { "epoch": 8.9575, "grad_norm": 6.59375, "learning_rate": 1.9814430501873865e-05, "loss": 1.6945, "step": 3583 }, { "epoch": 8.96, "grad_norm": 7.4375, "learning_rate": 1.9814275208751882e-05, "loss": 1.3988, "step": 3584 }, { "epoch": 8.9625, "grad_norm": 4.875, "learning_rate": 1.9814119851287908e-05, "loss": 1.6161, "step": 3585 }, { "epoch": 8.965, "grad_norm": 5.53125, "learning_rate": 1.9813964429482952e-05, "loss": 1.4638, "step": 3586 }, { "epoch": 8.9675, "grad_norm": 7.25, "learning_rate": 1.981380894333804e-05, "loss": 1.6552, "step": 3587 }, { "epoch": 8.97, "grad_norm": 5.59375, "learning_rate": 1.981365339285419e-05, "loss": 1.5454, "step": 3588 }, { "epoch": 8.9725, "grad_norm": 7.40625, "learning_rate": 1.981349777803242e-05, "loss": 1.4973, "step": 3589 }, { "epoch": 8.975, "grad_norm": 5.34375, "learning_rate": 1.9813342098873753e-05, "loss": 1.5923, "step": 3590 }, { "epoch": 8.9775, "grad_norm": 6.28125, "learning_rate": 1.981318635537921e-05, "loss": 1.4482, "step": 3591 }, { "epoch": 8.98, "grad_norm": 4.5, "learning_rate": 1.9813030547549806e-05, "loss": 1.611, "step": 3592 }, { "epoch": 8.9825, "grad_norm": 6.09375, "learning_rate": 1.9812874675386566e-05, "loss": 1.5646, "step": 3593 }, { "epoch": 8.985, "grad_norm": 6.34375, "learning_rate": 1.9812718738890516e-05, "loss": 1.5937, "step": 3594 }, { "epoch": 8.9875, "grad_norm": 6.25, "learning_rate": 1.981256273806267e-05, "loss": 1.4874, "step": 3595 }, { "epoch": 8.99, "grad_norm": 6.46875, "learning_rate": 1.9812406672904058e-05, "loss": 1.451, "step": 3596 }, { "epoch": 8.9925, "grad_norm": 5.34375, "learning_rate": 1.98122505434157e-05, "loss": 1.4318, "step": 3597 }, { "epoch": 8.995, "grad_norm": 8.6875, "learning_rate": 1.9812094349598622e-05, "loss": 1.3709, "step": 3598 }, { "epoch": 8.9975, "grad_norm": 5.96875, "learning_rate": 1.9811938091453846e-05, "loss": 1.4888, "step": 3599 }, { "epoch": 9.0, "grad_norm": 5.6875, "learning_rate": 1.9811781768982392e-05, "loss": 1.5, "step": 3600 }, { "epoch": 9.0025, "grad_norm": 6.34375, "learning_rate": 1.9811625382185295e-05, "loss": 1.6302, "step": 3601 }, { "epoch": 9.005, "grad_norm": 5.71875, "learning_rate": 1.981146893106357e-05, "loss": 1.5553, "step": 3602 }, { "epoch": 9.0075, "grad_norm": 7.21875, "learning_rate": 1.9811312415618248e-05, "loss": 1.6179, "step": 3603 }, { "epoch": 9.01, "grad_norm": 4.71875, "learning_rate": 1.9811155835850355e-05, "loss": 1.451, "step": 3604 }, { "epoch": 9.0125, "grad_norm": 6.6875, "learning_rate": 1.981099919176092e-05, "loss": 1.5118, "step": 3605 }, { "epoch": 9.015, "grad_norm": 5.40625, "learning_rate": 1.9810842483350964e-05, "loss": 1.543, "step": 3606 }, { "epoch": 9.0175, "grad_norm": 5.4375, "learning_rate": 1.981068571062152e-05, "loss": 1.6597, "step": 3607 }, { "epoch": 9.02, "grad_norm": 5.15625, "learning_rate": 1.981052887357361e-05, "loss": 1.3637, "step": 3608 }, { "epoch": 9.0225, "grad_norm": 6.8125, "learning_rate": 1.9810371972208266e-05, "loss": 1.6858, "step": 3609 }, { "epoch": 9.025, "grad_norm": 5.71875, "learning_rate": 1.9810215006526517e-05, "loss": 1.514, "step": 3610 }, { "epoch": 9.0275, "grad_norm": 5.15625, "learning_rate": 1.9810057976529388e-05, "loss": 1.6293, "step": 3611 }, { "epoch": 9.03, "grad_norm": 5.3125, "learning_rate": 1.9809900882217915e-05, "loss": 1.4766, "step": 3612 }, { "epoch": 9.0325, "grad_norm": 4.625, "learning_rate": 1.9809743723593124e-05, "loss": 1.3898, "step": 3613 }, { "epoch": 9.035, "grad_norm": 8.9375, "learning_rate": 1.9809586500656045e-05, "loss": 1.6496, "step": 3614 }, { "epoch": 9.0375, "grad_norm": 5.5, "learning_rate": 1.9809429213407708e-05, "loss": 1.4348, "step": 3615 }, { "epoch": 9.04, "grad_norm": 9.0625, "learning_rate": 1.9809271861849147e-05, "loss": 1.5883, "step": 3616 }, { "epoch": 9.0425, "grad_norm": 5.9375, "learning_rate": 1.980911444598139e-05, "loss": 1.5031, "step": 3617 }, { "epoch": 9.045, "grad_norm": 7.65625, "learning_rate": 1.9808956965805475e-05, "loss": 1.6433, "step": 3618 }, { "epoch": 9.0475, "grad_norm": 7.0625, "learning_rate": 1.9808799421322428e-05, "loss": 1.4966, "step": 3619 }, { "epoch": 9.05, "grad_norm": 5.375, "learning_rate": 1.9808641812533286e-05, "loss": 1.4657, "step": 3620 }, { "epoch": 9.0525, "grad_norm": 7.0, "learning_rate": 1.980848413943908e-05, "loss": 1.6311, "step": 3621 }, { "epoch": 9.055, "grad_norm": 5.125, "learning_rate": 1.9808326402040844e-05, "loss": 1.5316, "step": 3622 }, { "epoch": 9.0575, "grad_norm": 8.75, "learning_rate": 1.9808168600339616e-05, "loss": 1.6646, "step": 3623 }, { "epoch": 9.06, "grad_norm": 5.40625, "learning_rate": 1.9808010734336425e-05, "loss": 1.5429, "step": 3624 }, { "epoch": 9.0625, "grad_norm": 6.3125, "learning_rate": 1.9807852804032306e-05, "loss": 1.6029, "step": 3625 }, { "epoch": 9.065, "grad_norm": 6.21875, "learning_rate": 1.9807694809428297e-05, "loss": 1.7041, "step": 3626 }, { "epoch": 9.0675, "grad_norm": 8.375, "learning_rate": 1.9807536750525437e-05, "loss": 1.665, "step": 3627 }, { "epoch": 9.07, "grad_norm": 6.125, "learning_rate": 1.9807378627324753e-05, "loss": 1.6041, "step": 3628 }, { "epoch": 9.0725, "grad_norm": 7.5625, "learning_rate": 1.9807220439827293e-05, "loss": 1.601, "step": 3629 }, { "epoch": 9.075, "grad_norm": 5.0, "learning_rate": 1.9807062188034086e-05, "loss": 1.421, "step": 3630 }, { "epoch": 9.0775, "grad_norm": 5.0, "learning_rate": 1.980690387194617e-05, "loss": 1.5968, "step": 3631 }, { "epoch": 9.08, "grad_norm": 5.6875, "learning_rate": 1.9806745491564588e-05, "loss": 1.3924, "step": 3632 }, { "epoch": 9.0825, "grad_norm": 13.1875, "learning_rate": 1.980658704689037e-05, "loss": 1.5443, "step": 3633 }, { "epoch": 9.085, "grad_norm": 7.0625, "learning_rate": 1.9806428537924564e-05, "loss": 1.5249, "step": 3634 }, { "epoch": 9.0875, "grad_norm": 5.0, "learning_rate": 1.9806269964668204e-05, "loss": 1.4929, "step": 3635 }, { "epoch": 9.09, "grad_norm": 5.125, "learning_rate": 1.9806111327122332e-05, "loss": 1.3847, "step": 3636 }, { "epoch": 9.0925, "grad_norm": 4.125, "learning_rate": 1.980595262528798e-05, "loss": 1.4244, "step": 3637 }, { "epoch": 9.095, "grad_norm": 6.34375, "learning_rate": 1.9805793859166203e-05, "loss": 1.5588, "step": 3638 }, { "epoch": 9.0975, "grad_norm": 6.90625, "learning_rate": 1.980563502875803e-05, "loss": 1.5575, "step": 3639 }, { "epoch": 9.1, "grad_norm": 5.75, "learning_rate": 1.980547613406451e-05, "loss": 1.6431, "step": 3640 }, { "epoch": 9.1025, "grad_norm": 6.0625, "learning_rate": 1.9805317175086672e-05, "loss": 1.6162, "step": 3641 }, { "epoch": 9.105, "grad_norm": 9.5, "learning_rate": 1.9805158151825573e-05, "loss": 1.5963, "step": 3642 }, { "epoch": 9.1075, "grad_norm": 10.75, "learning_rate": 1.980499906428225e-05, "loss": 1.6458, "step": 3643 }, { "epoch": 9.11, "grad_norm": 4.0625, "learning_rate": 1.9804839912457746e-05, "loss": 1.4808, "step": 3644 }, { "epoch": 9.1125, "grad_norm": 9.3125, "learning_rate": 1.9804680696353103e-05, "loss": 1.5939, "step": 3645 }, { "epoch": 9.115, "grad_norm": 5.375, "learning_rate": 1.9804521415969363e-05, "loss": 1.4613, "step": 3646 }, { "epoch": 9.1175, "grad_norm": 4.84375, "learning_rate": 1.9804362071307576e-05, "loss": 1.5272, "step": 3647 }, { "epoch": 9.12, "grad_norm": 4.71875, "learning_rate": 1.9804202662368782e-05, "loss": 1.3848, "step": 3648 }, { "epoch": 9.1225, "grad_norm": 6.0, "learning_rate": 1.980404318915403e-05, "loss": 1.6029, "step": 3649 }, { "epoch": 9.125, "grad_norm": 5.0625, "learning_rate": 1.980388365166436e-05, "loss": 1.5543, "step": 3650 }, { "epoch": 9.1275, "grad_norm": 7.65625, "learning_rate": 1.9803724049900822e-05, "loss": 1.62, "step": 3651 }, { "epoch": 9.13, "grad_norm": 5.5625, "learning_rate": 1.9803564383864465e-05, "loss": 1.4333, "step": 3652 }, { "epoch": 9.1325, "grad_norm": 6.125, "learning_rate": 1.980340465355633e-05, "loss": 1.5717, "step": 3653 }, { "epoch": 9.135, "grad_norm": 5.84375, "learning_rate": 1.9803244858977464e-05, "loss": 1.6266, "step": 3654 }, { "epoch": 9.1375, "grad_norm": 7.1875, "learning_rate": 1.9803085000128918e-05, "loss": 1.5012, "step": 3655 }, { "epoch": 9.14, "grad_norm": 6.4375, "learning_rate": 1.9802925077011742e-05, "loss": 1.5799, "step": 3656 }, { "epoch": 9.1425, "grad_norm": 7.46875, "learning_rate": 1.980276508962698e-05, "loss": 1.6068, "step": 3657 }, { "epoch": 9.145, "grad_norm": 30.625, "learning_rate": 1.980260503797568e-05, "loss": 1.5751, "step": 3658 }, { "epoch": 9.1475, "grad_norm": 8.125, "learning_rate": 1.9802444922058893e-05, "loss": 1.6486, "step": 3659 }, { "epoch": 9.15, "grad_norm": 6.21875, "learning_rate": 1.9802284741877674e-05, "loss": 1.473, "step": 3660 }, { "epoch": 9.1525, "grad_norm": 9.5, "learning_rate": 1.9802124497433065e-05, "loss": 1.6216, "step": 3661 }, { "epoch": 9.155, "grad_norm": 5.90625, "learning_rate": 1.9801964188726123e-05, "loss": 1.4815, "step": 3662 }, { "epoch": 9.1575, "grad_norm": 6.25, "learning_rate": 1.9801803815757894e-05, "loss": 1.5189, "step": 3663 }, { "epoch": 9.16, "grad_norm": 5.96875, "learning_rate": 1.980164337852943e-05, "loss": 1.38, "step": 3664 }, { "epoch": 9.1625, "grad_norm": 7.65625, "learning_rate": 1.9801482877041787e-05, "loss": 1.4287, "step": 3665 }, { "epoch": 9.165, "grad_norm": 6.6875, "learning_rate": 1.9801322311296012e-05, "loss": 1.5224, "step": 3666 }, { "epoch": 9.1675, "grad_norm": 3.59375, "learning_rate": 1.9801161681293163e-05, "loss": 1.38, "step": 3667 }, { "epoch": 9.17, "grad_norm": 6.46875, "learning_rate": 1.980100098703429e-05, "loss": 1.5387, "step": 3668 }, { "epoch": 9.1725, "grad_norm": 6.0625, "learning_rate": 1.9800840228520448e-05, "loss": 1.563, "step": 3669 }, { "epoch": 9.175, "grad_norm": 6.0625, "learning_rate": 1.9800679405752686e-05, "loss": 1.4776, "step": 3670 }, { "epoch": 9.1775, "grad_norm": 6.5625, "learning_rate": 1.9800518518732064e-05, "loss": 1.5461, "step": 3671 }, { "epoch": 9.18, "grad_norm": 6.03125, "learning_rate": 1.9800357567459633e-05, "loss": 1.5098, "step": 3672 }, { "epoch": 9.1825, "grad_norm": 5.96875, "learning_rate": 1.980019655193645e-05, "loss": 1.4769, "step": 3673 }, { "epoch": 9.185, "grad_norm": 4.53125, "learning_rate": 1.9800035472163574e-05, "loss": 1.585, "step": 3674 }, { "epoch": 9.1875, "grad_norm": 4.9375, "learning_rate": 1.9799874328142053e-05, "loss": 1.3896, "step": 3675 }, { "epoch": 9.19, "grad_norm": 5.125, "learning_rate": 1.979971311987295e-05, "loss": 1.5998, "step": 3676 }, { "epoch": 9.1925, "grad_norm": 4.84375, "learning_rate": 1.9799551847357322e-05, "loss": 1.5495, "step": 3677 }, { "epoch": 9.195, "grad_norm": 6.5, "learning_rate": 1.9799390510596222e-05, "loss": 1.5992, "step": 3678 }, { "epoch": 9.1975, "grad_norm": 4.5, "learning_rate": 1.9799229109590713e-05, "loss": 1.5345, "step": 3679 }, { "epoch": 9.2, "grad_norm": 12.3125, "learning_rate": 1.9799067644341844e-05, "loss": 1.4246, "step": 3680 }, { "epoch": 9.2025, "grad_norm": 6.78125, "learning_rate": 1.9798906114850686e-05, "loss": 1.3512, "step": 3681 }, { "epoch": 9.205, "grad_norm": 8.4375, "learning_rate": 1.9798744521118285e-05, "loss": 1.6817, "step": 3682 }, { "epoch": 9.2075, "grad_norm": 5.34375, "learning_rate": 1.9798582863145714e-05, "loss": 1.6616, "step": 3683 }, { "epoch": 9.21, "grad_norm": 7.15625, "learning_rate": 1.979842114093402e-05, "loss": 1.4397, "step": 3684 }, { "epoch": 9.2125, "grad_norm": 9.375, "learning_rate": 1.9798259354484272e-05, "loss": 1.6336, "step": 3685 }, { "epoch": 9.215, "grad_norm": 6.28125, "learning_rate": 1.9798097503797533e-05, "loss": 1.4688, "step": 3686 }, { "epoch": 9.2175, "grad_norm": 6.8125, "learning_rate": 1.979793558887485e-05, "loss": 1.5225, "step": 3687 }, { "epoch": 9.22, "grad_norm": 5.96875, "learning_rate": 1.9797773609717295e-05, "loss": 1.5035, "step": 3688 }, { "epoch": 9.2225, "grad_norm": 9.0625, "learning_rate": 1.9797611566325932e-05, "loss": 1.6873, "step": 3689 }, { "epoch": 9.225, "grad_norm": 6.0625, "learning_rate": 1.9797449458701818e-05, "loss": 1.5918, "step": 3690 }, { "epoch": 9.2275, "grad_norm": 7.1875, "learning_rate": 1.979728728684602e-05, "loss": 1.653, "step": 3691 }, { "epoch": 9.23, "grad_norm": 4.34375, "learning_rate": 1.9797125050759593e-05, "loss": 1.4499, "step": 3692 }, { "epoch": 9.2325, "grad_norm": 6.6875, "learning_rate": 1.9796962750443608e-05, "loss": 1.5704, "step": 3693 }, { "epoch": 9.235, "grad_norm": 4.4375, "learning_rate": 1.9796800385899127e-05, "loss": 1.5183, "step": 3694 }, { "epoch": 9.2375, "grad_norm": 6.9375, "learning_rate": 1.979663795712722e-05, "loss": 1.5748, "step": 3695 }, { "epoch": 9.24, "grad_norm": 5.5625, "learning_rate": 1.9796475464128943e-05, "loss": 1.4216, "step": 3696 }, { "epoch": 9.2425, "grad_norm": 7.3125, "learning_rate": 1.9796312906905364e-05, "loss": 1.4629, "step": 3697 }, { "epoch": 9.245, "grad_norm": 12.625, "learning_rate": 1.979615028545755e-05, "loss": 1.5169, "step": 3698 }, { "epoch": 9.2475, "grad_norm": 7.125, "learning_rate": 1.9795987599786567e-05, "loss": 1.6079, "step": 3699 }, { "epoch": 9.25, "grad_norm": 5.46875, "learning_rate": 1.9795824849893483e-05, "loss": 1.4393, "step": 3700 }, { "epoch": 9.2525, "grad_norm": 5.375, "learning_rate": 1.979566203577936e-05, "loss": 1.4599, "step": 3701 }, { "epoch": 9.255, "grad_norm": 4.84375, "learning_rate": 1.979549915744527e-05, "loss": 1.5548, "step": 3702 }, { "epoch": 9.2575, "grad_norm": 6.90625, "learning_rate": 1.979533621489228e-05, "loss": 1.594, "step": 3703 }, { "epoch": 9.26, "grad_norm": 5.0, "learning_rate": 1.979517320812146e-05, "loss": 1.6792, "step": 3704 }, { "epoch": 9.2625, "grad_norm": 7.21875, "learning_rate": 1.979501013713387e-05, "loss": 1.6416, "step": 3705 }, { "epoch": 9.265, "grad_norm": 4.96875, "learning_rate": 1.979484700193059e-05, "loss": 1.5123, "step": 3706 }, { "epoch": 9.2675, "grad_norm": 5.21875, "learning_rate": 1.9794683802512688e-05, "loss": 1.4753, "step": 3707 }, { "epoch": 9.27, "grad_norm": 6.1875, "learning_rate": 1.9794520538881222e-05, "loss": 1.5101, "step": 3708 }, { "epoch": 9.2725, "grad_norm": 6.4375, "learning_rate": 1.979435721103728e-05, "loss": 1.5293, "step": 3709 }, { "epoch": 9.275, "grad_norm": 6.21875, "learning_rate": 1.9794193818981918e-05, "loss": 1.5367, "step": 3710 }, { "epoch": 9.2775, "grad_norm": 6.03125, "learning_rate": 1.9794030362716218e-05, "loss": 1.4071, "step": 3711 }, { "epoch": 9.28, "grad_norm": 9.3125, "learning_rate": 1.9793866842241245e-05, "loss": 1.57, "step": 3712 }, { "epoch": 9.2825, "grad_norm": 10.375, "learning_rate": 1.9793703257558072e-05, "loss": 1.6763, "step": 3713 }, { "epoch": 9.285, "grad_norm": 7.75, "learning_rate": 1.9793539608667775e-05, "loss": 1.5183, "step": 3714 }, { "epoch": 9.2875, "grad_norm": 8.1875, "learning_rate": 1.979337589557142e-05, "loss": 1.6208, "step": 3715 }, { "epoch": 9.29, "grad_norm": 7.0, "learning_rate": 1.979321211827009e-05, "loss": 1.5379, "step": 3716 }, { "epoch": 9.2925, "grad_norm": 6.125, "learning_rate": 1.9793048276764847e-05, "loss": 1.4983, "step": 3717 }, { "epoch": 9.295, "grad_norm": 6.15625, "learning_rate": 1.9792884371056777e-05, "loss": 1.6995, "step": 3718 }, { "epoch": 9.2975, "grad_norm": 5.75, "learning_rate": 1.979272040114695e-05, "loss": 1.5283, "step": 3719 }, { "epoch": 9.3, "grad_norm": 4.6875, "learning_rate": 1.9792556367036432e-05, "loss": 1.4649, "step": 3720 }, { "epoch": 9.3025, "grad_norm": 5.8125, "learning_rate": 1.9792392268726314e-05, "loss": 1.5425, "step": 3721 }, { "epoch": 9.305, "grad_norm": 5.53125, "learning_rate": 1.979222810621766e-05, "loss": 1.3903, "step": 3722 }, { "epoch": 9.3075, "grad_norm": 5.34375, "learning_rate": 1.979206387951155e-05, "loss": 1.4538, "step": 3723 }, { "epoch": 9.31, "grad_norm": 8.375, "learning_rate": 1.9791899588609063e-05, "loss": 1.5198, "step": 3724 }, { "epoch": 9.3125, "grad_norm": 5.75, "learning_rate": 1.9791735233511274e-05, "loss": 1.6196, "step": 3725 }, { "epoch": 9.315, "grad_norm": 6.5, "learning_rate": 1.979157081421926e-05, "loss": 1.5759, "step": 3726 }, { "epoch": 9.3175, "grad_norm": 6.40625, "learning_rate": 1.97914063307341e-05, "loss": 1.5846, "step": 3727 }, { "epoch": 9.32, "grad_norm": 7.53125, "learning_rate": 1.9791241783056874e-05, "loss": 1.5787, "step": 3728 }, { "epoch": 9.3225, "grad_norm": 6.90625, "learning_rate": 1.9791077171188657e-05, "loss": 1.5457, "step": 3729 }, { "epoch": 9.325, "grad_norm": 5.4375, "learning_rate": 1.9790912495130524e-05, "loss": 1.4744, "step": 3730 }, { "epoch": 9.3275, "grad_norm": 14.8125, "learning_rate": 1.979074775488357e-05, "loss": 1.5706, "step": 3731 }, { "epoch": 9.33, "grad_norm": 11.4375, "learning_rate": 1.979058295044886e-05, "loss": 1.5112, "step": 3732 }, { "epoch": 9.3325, "grad_norm": 6.6875, "learning_rate": 1.979041808182748e-05, "loss": 1.5045, "step": 3733 }, { "epoch": 9.335, "grad_norm": 6.09375, "learning_rate": 1.9790253149020513e-05, "loss": 1.5434, "step": 3734 }, { "epoch": 9.3375, "grad_norm": 5.84375, "learning_rate": 1.979008815202904e-05, "loss": 1.4938, "step": 3735 }, { "epoch": 9.34, "grad_norm": 5.4375, "learning_rate": 1.9789923090854138e-05, "loss": 1.451, "step": 3736 }, { "epoch": 9.3425, "grad_norm": 6.96875, "learning_rate": 1.9789757965496892e-05, "loss": 1.4617, "step": 3737 }, { "epoch": 9.345, "grad_norm": 7.78125, "learning_rate": 1.9789592775958385e-05, "loss": 1.5371, "step": 3738 }, { "epoch": 9.3475, "grad_norm": 8.875, "learning_rate": 1.97894275222397e-05, "loss": 1.7169, "step": 3739 }, { "epoch": 9.35, "grad_norm": 8.0, "learning_rate": 1.9789262204341918e-05, "loss": 1.4495, "step": 3740 }, { "epoch": 9.3525, "grad_norm": 7.75, "learning_rate": 1.9789096822266126e-05, "loss": 1.7323, "step": 3741 }, { "epoch": 9.355, "grad_norm": 6.46875, "learning_rate": 1.9788931376013403e-05, "loss": 1.4882, "step": 3742 }, { "epoch": 9.3575, "grad_norm": 5.3125, "learning_rate": 1.9788765865584846e-05, "loss": 1.6254, "step": 3743 }, { "epoch": 9.36, "grad_norm": 5.71875, "learning_rate": 1.9788600290981525e-05, "loss": 1.6514, "step": 3744 }, { "epoch": 9.3625, "grad_norm": 5.4375, "learning_rate": 1.9788434652204536e-05, "loss": 1.644, "step": 3745 }, { "epoch": 9.365, "grad_norm": 6.5, "learning_rate": 1.9788268949254958e-05, "loss": 1.5501, "step": 3746 }, { "epoch": 9.3675, "grad_norm": 10.8125, "learning_rate": 1.9788103182133886e-05, "loss": 1.5744, "step": 3747 }, { "epoch": 9.37, "grad_norm": 6.84375, "learning_rate": 1.9787937350842394e-05, "loss": 1.5217, "step": 3748 }, { "epoch": 9.3725, "grad_norm": 7.65625, "learning_rate": 1.978777145538158e-05, "loss": 1.6108, "step": 3749 }, { "epoch": 9.375, "grad_norm": 6.84375, "learning_rate": 1.9787605495752528e-05, "loss": 1.6914, "step": 3750 }, { "epoch": 9.3775, "grad_norm": 5.375, "learning_rate": 1.9787439471956326e-05, "loss": 1.429, "step": 3751 }, { "epoch": 9.38, "grad_norm": 5.5, "learning_rate": 1.978727338399406e-05, "loss": 1.4067, "step": 3752 }, { "epoch": 9.3825, "grad_norm": 8.1875, "learning_rate": 1.9787107231866828e-05, "loss": 1.5448, "step": 3753 }, { "epoch": 9.385, "grad_norm": 5.125, "learning_rate": 1.9786941015575703e-05, "loss": 1.5059, "step": 3754 }, { "epoch": 9.3875, "grad_norm": 5.96875, "learning_rate": 1.9786774735121793e-05, "loss": 1.633, "step": 3755 }, { "epoch": 9.39, "grad_norm": 5.71875, "learning_rate": 1.9786608390506176e-05, "loss": 1.4719, "step": 3756 }, { "epoch": 9.3925, "grad_norm": 11.5, "learning_rate": 1.9786441981729946e-05, "loss": 1.4866, "step": 3757 }, { "epoch": 9.395, "grad_norm": 7.6875, "learning_rate": 1.9786275508794194e-05, "loss": 1.6748, "step": 3758 }, { "epoch": 9.3975, "grad_norm": 8.4375, "learning_rate": 1.978610897170001e-05, "loss": 1.7657, "step": 3759 }, { "epoch": 9.4, "grad_norm": 8.0625, "learning_rate": 1.978594237044849e-05, "loss": 1.5718, "step": 3760 }, { "epoch": 9.4025, "grad_norm": 6.0625, "learning_rate": 1.9785775705040723e-05, "loss": 1.6364, "step": 3761 }, { "epoch": 9.405, "grad_norm": 8.1875, "learning_rate": 1.97856089754778e-05, "loss": 1.6668, "step": 3762 }, { "epoch": 9.4075, "grad_norm": 6.03125, "learning_rate": 1.9785442181760816e-05, "loss": 1.4818, "step": 3763 }, { "epoch": 9.41, "grad_norm": 8.125, "learning_rate": 1.9785275323890867e-05, "loss": 1.6647, "step": 3764 }, { "epoch": 9.4125, "grad_norm": 7.375, "learning_rate": 1.9785108401869046e-05, "loss": 1.4844, "step": 3765 }, { "epoch": 9.415, "grad_norm": 4.8125, "learning_rate": 1.978494141569644e-05, "loss": 1.5466, "step": 3766 }, { "epoch": 9.4175, "grad_norm": 7.46875, "learning_rate": 1.9784774365374156e-05, "loss": 1.6679, "step": 3767 }, { "epoch": 9.42, "grad_norm": 5.1875, "learning_rate": 1.9784607250903278e-05, "loss": 1.4882, "step": 3768 }, { "epoch": 9.4225, "grad_norm": 7.71875, "learning_rate": 1.978444007228491e-05, "loss": 1.5317, "step": 3769 }, { "epoch": 9.425, "grad_norm": 5.9375, "learning_rate": 1.978427282952014e-05, "loss": 1.4908, "step": 3770 }, { "epoch": 9.4275, "grad_norm": 7.625, "learning_rate": 1.978410552261007e-05, "loss": 1.5807, "step": 3771 }, { "epoch": 9.43, "grad_norm": 5.4375, "learning_rate": 1.9783938151555803e-05, "loss": 1.481, "step": 3772 }, { "epoch": 9.4325, "grad_norm": 7.375, "learning_rate": 1.978377071635842e-05, "loss": 1.6017, "step": 3773 }, { "epoch": 9.435, "grad_norm": 8.3125, "learning_rate": 1.9783603217019033e-05, "loss": 1.4136, "step": 3774 }, { "epoch": 9.4375, "grad_norm": 5.84375, "learning_rate": 1.9783435653538733e-05, "loss": 1.488, "step": 3775 }, { "epoch": 9.44, "grad_norm": 9.5, "learning_rate": 1.9783268025918622e-05, "loss": 1.6979, "step": 3776 }, { "epoch": 9.4425, "grad_norm": 9.9375, "learning_rate": 1.9783100334159794e-05, "loss": 1.7132, "step": 3777 }, { "epoch": 9.445, "grad_norm": 5.0625, "learning_rate": 1.9782932578263353e-05, "loss": 1.5481, "step": 3778 }, { "epoch": 9.4475, "grad_norm": 8.5625, "learning_rate": 1.97827647582304e-05, "loss": 1.5591, "step": 3779 }, { "epoch": 9.45, "grad_norm": 6.0625, "learning_rate": 1.9782596874062028e-05, "loss": 1.5858, "step": 3780 }, { "epoch": 9.4525, "grad_norm": 8.5, "learning_rate": 1.9782428925759346e-05, "loss": 1.5389, "step": 3781 }, { "epoch": 9.455, "grad_norm": 7.5, "learning_rate": 1.978226091332345e-05, "loss": 1.7508, "step": 3782 }, { "epoch": 9.4575, "grad_norm": 6.78125, "learning_rate": 1.9782092836755442e-05, "loss": 1.6971, "step": 3783 }, { "epoch": 9.46, "grad_norm": 7.0, "learning_rate": 1.9781924696056426e-05, "loss": 1.4314, "step": 3784 }, { "epoch": 9.4625, "grad_norm": 6.125, "learning_rate": 1.9781756491227505e-05, "loss": 1.4881, "step": 3785 }, { "epoch": 9.465, "grad_norm": 6.9375, "learning_rate": 1.9781588222269778e-05, "loss": 1.478, "step": 3786 }, { "epoch": 9.4675, "grad_norm": 7.65625, "learning_rate": 1.978141988918435e-05, "loss": 1.7046, "step": 3787 }, { "epoch": 9.47, "grad_norm": 4.96875, "learning_rate": 1.9781251491972324e-05, "loss": 1.35, "step": 3788 }, { "epoch": 9.4725, "grad_norm": 5.96875, "learning_rate": 1.9781083030634806e-05, "loss": 1.6128, "step": 3789 }, { "epoch": 9.475, "grad_norm": 7.0625, "learning_rate": 1.97809145051729e-05, "loss": 1.5623, "step": 3790 }, { "epoch": 9.4775, "grad_norm": 6.46875, "learning_rate": 1.978074591558771e-05, "loss": 1.4778, "step": 3791 }, { "epoch": 9.48, "grad_norm": 6.4375, "learning_rate": 1.9780577261880336e-05, "loss": 1.4883, "step": 3792 }, { "epoch": 9.4825, "grad_norm": 6.8125, "learning_rate": 1.9780408544051894e-05, "loss": 1.4907, "step": 3793 }, { "epoch": 9.485, "grad_norm": 6.4375, "learning_rate": 1.9780239762103482e-05, "loss": 1.3745, "step": 3794 }, { "epoch": 9.4875, "grad_norm": 5.15625, "learning_rate": 1.978007091603621e-05, "loss": 1.4633, "step": 3795 }, { "epoch": 9.49, "grad_norm": 7.34375, "learning_rate": 1.9779902005851187e-05, "loss": 1.5658, "step": 3796 }, { "epoch": 9.4925, "grad_norm": 7.4375, "learning_rate": 1.9779733031549517e-05, "loss": 1.5196, "step": 3797 }, { "epoch": 9.495, "grad_norm": 5.40625, "learning_rate": 1.977956399313231e-05, "loss": 1.4671, "step": 3798 }, { "epoch": 9.4975, "grad_norm": 5.46875, "learning_rate": 1.977939489060067e-05, "loss": 1.501, "step": 3799 }, { "epoch": 9.5, "grad_norm": 5.6875, "learning_rate": 1.977922572395571e-05, "loss": 1.504, "step": 3800 }, { "epoch": 9.5025, "grad_norm": 6.15625, "learning_rate": 1.9779056493198535e-05, "loss": 1.4494, "step": 3801 }, { "epoch": 9.505, "grad_norm": 5.4375, "learning_rate": 1.977888719833026e-05, "loss": 1.7627, "step": 3802 }, { "epoch": 9.5075, "grad_norm": 6.0, "learning_rate": 1.977871783935199e-05, "loss": 1.5169, "step": 3803 }, { "epoch": 9.51, "grad_norm": 6.1875, "learning_rate": 1.9778548416264837e-05, "loss": 1.4626, "step": 3804 }, { "epoch": 9.5125, "grad_norm": 5.1875, "learning_rate": 1.977837892906991e-05, "loss": 1.3741, "step": 3805 }, { "epoch": 9.515, "grad_norm": 7.375, "learning_rate": 1.9778209377768327e-05, "loss": 1.7124, "step": 3806 }, { "epoch": 9.5175, "grad_norm": 5.90625, "learning_rate": 1.9778039762361192e-05, "loss": 1.6029, "step": 3807 }, { "epoch": 9.52, "grad_norm": 5.9375, "learning_rate": 1.977787008284962e-05, "loss": 1.5596, "step": 3808 }, { "epoch": 9.5225, "grad_norm": 8.125, "learning_rate": 1.9777700339234722e-05, "loss": 1.6009, "step": 3809 }, { "epoch": 9.525, "grad_norm": 7.75, "learning_rate": 1.9777530531517613e-05, "loss": 1.7224, "step": 3810 }, { "epoch": 9.5275, "grad_norm": 5.65625, "learning_rate": 1.9777360659699406e-05, "loss": 1.3758, "step": 3811 }, { "epoch": 9.53, "grad_norm": 6.71875, "learning_rate": 1.977719072378121e-05, "loss": 1.5458, "step": 3812 }, { "epoch": 9.5325, "grad_norm": 5.125, "learning_rate": 1.9777020723764146e-05, "loss": 1.6673, "step": 3813 }, { "epoch": 9.535, "grad_norm": 6.9375, "learning_rate": 1.9776850659649324e-05, "loss": 1.472, "step": 3814 }, { "epoch": 9.5375, "grad_norm": 6.65625, "learning_rate": 1.9776680531437865e-05, "loss": 1.6866, "step": 3815 }, { "epoch": 9.54, "grad_norm": 9.375, "learning_rate": 1.9776510339130874e-05, "loss": 1.5361, "step": 3816 }, { "epoch": 9.5425, "grad_norm": 9.125, "learning_rate": 1.9776340082729476e-05, "loss": 1.5173, "step": 3817 }, { "epoch": 9.545, "grad_norm": 9.125, "learning_rate": 1.977616976223478e-05, "loss": 1.7112, "step": 3818 }, { "epoch": 9.5475, "grad_norm": 7.34375, "learning_rate": 1.977599937764791e-05, "loss": 1.4603, "step": 3819 }, { "epoch": 9.55, "grad_norm": 5.90625, "learning_rate": 1.9775828928969976e-05, "loss": 1.6694, "step": 3820 }, { "epoch": 9.5525, "grad_norm": 5.71875, "learning_rate": 1.97756584162021e-05, "loss": 1.5077, "step": 3821 }, { "epoch": 9.555, "grad_norm": 5.71875, "learning_rate": 1.97754878393454e-05, "loss": 1.4187, "step": 3822 }, { "epoch": 9.5575, "grad_norm": 7.71875, "learning_rate": 1.977531719840099e-05, "loss": 1.5636, "step": 3823 }, { "epoch": 9.56, "grad_norm": 7.125, "learning_rate": 1.9775146493369996e-05, "loss": 1.509, "step": 3824 }, { "epoch": 9.5625, "grad_norm": 6.15625, "learning_rate": 1.977497572425353e-05, "loss": 1.6457, "step": 3825 }, { "epoch": 9.565, "grad_norm": 7.84375, "learning_rate": 1.9774804891052714e-05, "loss": 1.4037, "step": 3826 }, { "epoch": 9.5675, "grad_norm": 7.9375, "learning_rate": 1.977463399376867e-05, "loss": 1.6542, "step": 3827 }, { "epoch": 9.57, "grad_norm": 6.28125, "learning_rate": 1.9774463032402514e-05, "loss": 1.547, "step": 3828 }, { "epoch": 9.5725, "grad_norm": 5.625, "learning_rate": 1.977429200695537e-05, "loss": 1.5492, "step": 3829 }, { "epoch": 9.575, "grad_norm": 6.78125, "learning_rate": 1.977412091742836e-05, "loss": 1.5005, "step": 3830 }, { "epoch": 9.5775, "grad_norm": 6.90625, "learning_rate": 1.9773949763822604e-05, "loss": 1.5979, "step": 3831 }, { "epoch": 9.58, "grad_norm": 5.25, "learning_rate": 1.9773778546139228e-05, "loss": 1.4033, "step": 3832 }, { "epoch": 9.5825, "grad_norm": 7.59375, "learning_rate": 1.9773607264379345e-05, "loss": 1.699, "step": 3833 }, { "epoch": 9.585, "grad_norm": 7.78125, "learning_rate": 1.9773435918544085e-05, "loss": 1.7643, "step": 3834 }, { "epoch": 9.5875, "grad_norm": 6.53125, "learning_rate": 1.9773264508634575e-05, "loss": 1.2571, "step": 3835 }, { "epoch": 9.59, "grad_norm": 6.4375, "learning_rate": 1.9773093034651928e-05, "loss": 1.3856, "step": 3836 }, { "epoch": 9.5925, "grad_norm": 7.96875, "learning_rate": 1.9772921496597276e-05, "loss": 1.7006, "step": 3837 }, { "epoch": 9.595, "grad_norm": 6.90625, "learning_rate": 1.9772749894471744e-05, "loss": 1.6258, "step": 3838 }, { "epoch": 9.5975, "grad_norm": 5.59375, "learning_rate": 1.9772578228276454e-05, "loss": 1.52, "step": 3839 }, { "epoch": 9.6, "grad_norm": 5.5625, "learning_rate": 1.977240649801253e-05, "loss": 1.5514, "step": 3840 }, { "epoch": 9.6025, "grad_norm": 7.0625, "learning_rate": 1.9772234703681105e-05, "loss": 1.3579, "step": 3841 }, { "epoch": 9.605, "grad_norm": 6.875, "learning_rate": 1.9772062845283295e-05, "loss": 1.4296, "step": 3842 }, { "epoch": 9.6075, "grad_norm": 4.625, "learning_rate": 1.9771890922820236e-05, "loss": 1.5245, "step": 3843 }, { "epoch": 9.61, "grad_norm": 6.875, "learning_rate": 1.9771718936293047e-05, "loss": 1.6231, "step": 3844 }, { "epoch": 9.6125, "grad_norm": 6.09375, "learning_rate": 1.9771546885702862e-05, "loss": 1.5357, "step": 3845 }, { "epoch": 9.615, "grad_norm": 5.65625, "learning_rate": 1.977137477105081e-05, "loss": 1.5742, "step": 3846 }, { "epoch": 9.6175, "grad_norm": 6.0, "learning_rate": 1.977120259233801e-05, "loss": 1.554, "step": 3847 }, { "epoch": 9.62, "grad_norm": 5.34375, "learning_rate": 1.97710303495656e-05, "loss": 1.4108, "step": 3848 }, { "epoch": 9.6225, "grad_norm": 5.71875, "learning_rate": 1.9770858042734705e-05, "loss": 1.5033, "step": 3849 }, { "epoch": 9.625, "grad_norm": 7.6875, "learning_rate": 1.977068567184646e-05, "loss": 1.5481, "step": 3850 }, { "epoch": 9.6275, "grad_norm": 6.40625, "learning_rate": 1.977051323690198e-05, "loss": 1.6294, "step": 3851 }, { "epoch": 9.63, "grad_norm": 6.5625, "learning_rate": 1.9770340737902417e-05, "loss": 1.5565, "step": 3852 }, { "epoch": 9.6325, "grad_norm": 8.375, "learning_rate": 1.9770168174848886e-05, "loss": 1.4973, "step": 3853 }, { "epoch": 9.635, "grad_norm": 6.9375, "learning_rate": 1.9769995547742527e-05, "loss": 1.4966, "step": 3854 }, { "epoch": 9.6375, "grad_norm": 6.3125, "learning_rate": 1.9769822856584465e-05, "loss": 1.6032, "step": 3855 }, { "epoch": 9.64, "grad_norm": 6.53125, "learning_rate": 1.9769650101375835e-05, "loss": 1.3893, "step": 3856 }, { "epoch": 9.6425, "grad_norm": 7.0625, "learning_rate": 1.9769477282117772e-05, "loss": 1.6227, "step": 3857 }, { "epoch": 9.645, "grad_norm": 6.59375, "learning_rate": 1.9769304398811404e-05, "loss": 1.5436, "step": 3858 }, { "epoch": 9.6475, "grad_norm": 7.28125, "learning_rate": 1.976913145145787e-05, "loss": 1.4894, "step": 3859 }, { "epoch": 9.65, "grad_norm": 6.875, "learning_rate": 1.97689584400583e-05, "loss": 1.493, "step": 3860 }, { "epoch": 9.6525, "grad_norm": 6.40625, "learning_rate": 1.976878536461383e-05, "loss": 1.3917, "step": 3861 }, { "epoch": 9.655, "grad_norm": 7.84375, "learning_rate": 1.9768612225125598e-05, "loss": 1.4482, "step": 3862 }, { "epoch": 9.6575, "grad_norm": 6.25, "learning_rate": 1.9768439021594728e-05, "loss": 1.5764, "step": 3863 }, { "epoch": 9.66, "grad_norm": 8.5625, "learning_rate": 1.9768265754022366e-05, "loss": 1.8224, "step": 3864 }, { "epoch": 9.6625, "grad_norm": 5.875, "learning_rate": 1.9768092422409645e-05, "loss": 1.6305, "step": 3865 }, { "epoch": 9.665, "grad_norm": 11.5625, "learning_rate": 1.97679190267577e-05, "loss": 1.4706, "step": 3866 }, { "epoch": 9.6675, "grad_norm": 6.90625, "learning_rate": 1.9767745567067667e-05, "loss": 1.6576, "step": 3867 }, { "epoch": 9.67, "grad_norm": 7.90625, "learning_rate": 1.9767572043340687e-05, "loss": 1.5483, "step": 3868 }, { "epoch": 9.6725, "grad_norm": 9.25, "learning_rate": 1.9767398455577898e-05, "loss": 1.5404, "step": 3869 }, { "epoch": 9.675, "grad_norm": 4.0625, "learning_rate": 1.9767224803780435e-05, "loss": 1.5256, "step": 3870 }, { "epoch": 9.6775, "grad_norm": 5.15625, "learning_rate": 1.9767051087949435e-05, "loss": 1.4072, "step": 3871 }, { "epoch": 9.68, "grad_norm": 7.0625, "learning_rate": 1.9766877308086038e-05, "loss": 1.659, "step": 3872 }, { "epoch": 9.6825, "grad_norm": 6.40625, "learning_rate": 1.9766703464191382e-05, "loss": 1.5414, "step": 3873 }, { "epoch": 9.685, "grad_norm": 7.0, "learning_rate": 1.9766529556266615e-05, "loss": 1.5184, "step": 3874 }, { "epoch": 9.6875, "grad_norm": 6.6875, "learning_rate": 1.9766355584312866e-05, "loss": 1.4575, "step": 3875 }, { "epoch": 9.69, "grad_norm": 6.75, "learning_rate": 1.9766181548331283e-05, "loss": 1.4988, "step": 3876 }, { "epoch": 9.6925, "grad_norm": 10.1875, "learning_rate": 1.9766007448323005e-05, "loss": 1.5998, "step": 3877 }, { "epoch": 9.695, "grad_norm": 7.15625, "learning_rate": 1.9765833284289173e-05, "loss": 1.6364, "step": 3878 }, { "epoch": 9.6975, "grad_norm": 6.40625, "learning_rate": 1.976565905623093e-05, "loss": 1.4394, "step": 3879 }, { "epoch": 9.7, "grad_norm": 4.21875, "learning_rate": 1.9765484764149413e-05, "loss": 1.4541, "step": 3880 }, { "epoch": 9.7025, "grad_norm": 7.125, "learning_rate": 1.9765310408045772e-05, "loss": 1.5956, "step": 3881 }, { "epoch": 9.705, "grad_norm": 10.3125, "learning_rate": 1.9765135987921144e-05, "loss": 1.4832, "step": 3882 }, { "epoch": 9.7075, "grad_norm": 6.03125, "learning_rate": 1.976496150377668e-05, "loss": 1.4772, "step": 3883 }, { "epoch": 9.71, "grad_norm": 5.59375, "learning_rate": 1.9764786955613514e-05, "loss": 1.5193, "step": 3884 }, { "epoch": 9.7125, "grad_norm": 5.71875, "learning_rate": 1.97646123434328e-05, "loss": 1.5463, "step": 3885 }, { "epoch": 9.715, "grad_norm": 7.25, "learning_rate": 1.9764437667235677e-05, "loss": 1.6329, "step": 3886 }, { "epoch": 9.7175, "grad_norm": 4.625, "learning_rate": 1.976426292702329e-05, "loss": 1.6339, "step": 3887 }, { "epoch": 9.72, "grad_norm": 7.90625, "learning_rate": 1.9764088122796785e-05, "loss": 1.4904, "step": 3888 }, { "epoch": 9.7225, "grad_norm": 4.90625, "learning_rate": 1.976391325455731e-05, "loss": 1.4166, "step": 3889 }, { "epoch": 9.725, "grad_norm": 5.03125, "learning_rate": 1.9763738322306014e-05, "loss": 1.5159, "step": 3890 }, { "epoch": 9.7275, "grad_norm": 6.5, "learning_rate": 1.9763563326044032e-05, "loss": 1.5284, "step": 3891 }, { "epoch": 9.73, "grad_norm": 6.625, "learning_rate": 1.9763388265772527e-05, "loss": 1.3929, "step": 3892 }, { "epoch": 9.7325, "grad_norm": 9.625, "learning_rate": 1.9763213141492635e-05, "loss": 1.5215, "step": 3893 }, { "epoch": 9.735, "grad_norm": 10.9375, "learning_rate": 1.976303795320551e-05, "loss": 1.5147, "step": 3894 }, { "epoch": 9.7375, "grad_norm": 5.9375, "learning_rate": 1.97628627009123e-05, "loss": 1.4999, "step": 3895 }, { "epoch": 9.74, "grad_norm": 5.59375, "learning_rate": 1.9762687384614152e-05, "loss": 1.4291, "step": 3896 }, { "epoch": 9.7425, "grad_norm": 8.4375, "learning_rate": 1.9762512004312216e-05, "loss": 1.5625, "step": 3897 }, { "epoch": 9.745, "grad_norm": 6.65625, "learning_rate": 1.976233656000764e-05, "loss": 1.4544, "step": 3898 }, { "epoch": 9.7475, "grad_norm": 7.25, "learning_rate": 1.9762161051701582e-05, "loss": 1.4347, "step": 3899 }, { "epoch": 9.75, "grad_norm": 8.625, "learning_rate": 1.976198547939518e-05, "loss": 1.5015, "step": 3900 }, { "epoch": 9.7525, "grad_norm": 8.375, "learning_rate": 1.9761809843089598e-05, "loss": 1.4849, "step": 3901 }, { "epoch": 9.755, "grad_norm": 6.3125, "learning_rate": 1.9761634142785977e-05, "loss": 1.457, "step": 3902 }, { "epoch": 9.7575, "grad_norm": 4.65625, "learning_rate": 1.9761458378485474e-05, "loss": 1.5579, "step": 3903 }, { "epoch": 9.76, "grad_norm": 5.96875, "learning_rate": 1.976128255018924e-05, "loss": 1.4746, "step": 3904 }, { "epoch": 9.7625, "grad_norm": 5.0, "learning_rate": 1.976110665789843e-05, "loss": 1.5924, "step": 3905 }, { "epoch": 9.765, "grad_norm": 6.71875, "learning_rate": 1.9760930701614197e-05, "loss": 1.4384, "step": 3906 }, { "epoch": 9.7675, "grad_norm": 6.34375, "learning_rate": 1.9760754681337687e-05, "loss": 1.6185, "step": 3907 }, { "epoch": 9.77, "grad_norm": 8.125, "learning_rate": 1.9760578597070064e-05, "loss": 1.4603, "step": 3908 }, { "epoch": 9.7725, "grad_norm": 6.15625, "learning_rate": 1.9760402448812476e-05, "loss": 1.6053, "step": 3909 }, { "epoch": 9.775, "grad_norm": 5.25, "learning_rate": 1.9760226236566082e-05, "loss": 1.6073, "step": 3910 }, { "epoch": 9.7775, "grad_norm": 7.9375, "learning_rate": 1.9760049960332038e-05, "loss": 1.5184, "step": 3911 }, { "epoch": 9.78, "grad_norm": 8.875, "learning_rate": 1.9759873620111492e-05, "loss": 1.6716, "step": 3912 }, { "epoch": 9.7825, "grad_norm": 5.53125, "learning_rate": 1.975969721590561e-05, "loss": 1.6455, "step": 3913 }, { "epoch": 9.785, "grad_norm": 7.375, "learning_rate": 1.975952074771554e-05, "loss": 1.4998, "step": 3914 }, { "epoch": 9.7875, "grad_norm": 7.46875, "learning_rate": 1.975934421554244e-05, "loss": 1.5544, "step": 3915 }, { "epoch": 9.79, "grad_norm": 5.59375, "learning_rate": 1.9759167619387474e-05, "loss": 1.5591, "step": 3916 }, { "epoch": 9.7925, "grad_norm": 6.5, "learning_rate": 1.9758990959251797e-05, "loss": 1.5322, "step": 3917 }, { "epoch": 9.795, "grad_norm": 6.8125, "learning_rate": 1.9758814235136563e-05, "loss": 1.5222, "step": 3918 }, { "epoch": 9.7975, "grad_norm": 6.09375, "learning_rate": 1.9758637447042934e-05, "loss": 1.611, "step": 3919 }, { "epoch": 9.8, "grad_norm": 7.71875, "learning_rate": 1.9758460594972068e-05, "loss": 1.505, "step": 3920 }, { "epoch": 9.8025, "grad_norm": 8.5, "learning_rate": 1.9758283678925124e-05, "loss": 1.5283, "step": 3921 }, { "epoch": 9.805, "grad_norm": 6.4375, "learning_rate": 1.9758106698903266e-05, "loss": 1.5453, "step": 3922 }, { "epoch": 9.8075, "grad_norm": 7.0, "learning_rate": 1.9757929654907652e-05, "loss": 1.4343, "step": 3923 }, { "epoch": 9.81, "grad_norm": 6.46875, "learning_rate": 1.9757752546939434e-05, "loss": 1.7015, "step": 3924 }, { "epoch": 9.8125, "grad_norm": 5.84375, "learning_rate": 1.9757575374999785e-05, "loss": 1.4185, "step": 3925 }, { "epoch": 9.815, "grad_norm": 6.625, "learning_rate": 1.9757398139089863e-05, "loss": 1.5493, "step": 3926 }, { "epoch": 9.8175, "grad_norm": 5.3125, "learning_rate": 1.9757220839210834e-05, "loss": 1.4779, "step": 3927 }, { "epoch": 9.82, "grad_norm": 8.4375, "learning_rate": 1.975704347536385e-05, "loss": 1.5873, "step": 3928 }, { "epoch": 9.8225, "grad_norm": 9.0, "learning_rate": 1.975686604755008e-05, "loss": 1.6752, "step": 3929 }, { "epoch": 9.825, "grad_norm": 5.78125, "learning_rate": 1.9756688555770688e-05, "loss": 1.3717, "step": 3930 }, { "epoch": 9.8275, "grad_norm": 7.125, "learning_rate": 1.9756511000026834e-05, "loss": 1.5633, "step": 3931 }, { "epoch": 9.83, "grad_norm": 5.03125, "learning_rate": 1.9756333380319687e-05, "loss": 1.3772, "step": 3932 }, { "epoch": 9.8325, "grad_norm": 6.875, "learning_rate": 1.9756155696650405e-05, "loss": 1.3263, "step": 3933 }, { "epoch": 9.835, "grad_norm": 5.8125, "learning_rate": 1.975597794902016e-05, "loss": 1.5905, "step": 3934 }, { "epoch": 9.8375, "grad_norm": 10.125, "learning_rate": 1.975580013743011e-05, "loss": 1.5712, "step": 3935 }, { "epoch": 9.84, "grad_norm": 4.78125, "learning_rate": 1.975562226188143e-05, "loss": 1.4086, "step": 3936 }, { "epoch": 9.8425, "grad_norm": 5.8125, "learning_rate": 1.9755444322375276e-05, "loss": 1.6088, "step": 3937 }, { "epoch": 9.845, "grad_norm": 5.46875, "learning_rate": 1.975526631891282e-05, "loss": 1.6371, "step": 3938 }, { "epoch": 9.8475, "grad_norm": 8.75, "learning_rate": 1.9755088251495233e-05, "loss": 1.6963, "step": 3939 }, { "epoch": 9.85, "grad_norm": 5.1875, "learning_rate": 1.9754910120123675e-05, "loss": 1.4214, "step": 3940 }, { "epoch": 9.8525, "grad_norm": 5.28125, "learning_rate": 1.9754731924799317e-05, "loss": 1.4908, "step": 3941 }, { "epoch": 9.855, "grad_norm": 7.75, "learning_rate": 1.9754553665523325e-05, "loss": 1.46, "step": 3942 }, { "epoch": 9.8575, "grad_norm": 10.5, "learning_rate": 1.9754375342296873e-05, "loss": 1.4887, "step": 3943 }, { "epoch": 9.86, "grad_norm": 7.1875, "learning_rate": 1.9754196955121125e-05, "loss": 1.4785, "step": 3944 }, { "epoch": 9.8625, "grad_norm": 5.53125, "learning_rate": 1.975401850399725e-05, "loss": 1.4224, "step": 3945 }, { "epoch": 9.865, "grad_norm": 4.8125, "learning_rate": 1.9753839988926423e-05, "loss": 1.309, "step": 3946 }, { "epoch": 9.8675, "grad_norm": 3.34375, "learning_rate": 1.975366140990981e-05, "loss": 1.3119, "step": 3947 }, { "epoch": 9.87, "grad_norm": 8.875, "learning_rate": 1.975348276694858e-05, "loss": 1.6372, "step": 3948 }, { "epoch": 9.8725, "grad_norm": 8.125, "learning_rate": 1.9753304060043913e-05, "loss": 1.5988, "step": 3949 }, { "epoch": 9.875, "grad_norm": 7.8125, "learning_rate": 1.975312528919697e-05, "loss": 1.4917, "step": 3950 }, { "epoch": 9.8775, "grad_norm": 7.09375, "learning_rate": 1.9752946454408932e-05, "loss": 1.7288, "step": 3951 }, { "epoch": 9.88, "grad_norm": 6.375, "learning_rate": 1.9752767555680967e-05, "loss": 1.5744, "step": 3952 }, { "epoch": 9.8825, "grad_norm": 5.375, "learning_rate": 1.9752588593014246e-05, "loss": 1.541, "step": 3953 }, { "epoch": 9.885, "grad_norm": 5.59375, "learning_rate": 1.9752409566409945e-05, "loss": 1.4263, "step": 3954 }, { "epoch": 9.8875, "grad_norm": 9.375, "learning_rate": 1.9752230475869238e-05, "loss": 1.5591, "step": 3955 }, { "epoch": 9.89, "grad_norm": 6.0, "learning_rate": 1.97520513213933e-05, "loss": 1.586, "step": 3956 }, { "epoch": 9.8925, "grad_norm": 6.0, "learning_rate": 1.97518721029833e-05, "loss": 1.3313, "step": 3957 }, { "epoch": 9.895, "grad_norm": 7.5, "learning_rate": 1.9751692820640418e-05, "loss": 1.487, "step": 3958 }, { "epoch": 9.8975, "grad_norm": 16.5, "learning_rate": 1.9751513474365828e-05, "loss": 1.5415, "step": 3959 }, { "epoch": 9.9, "grad_norm": 5.25, "learning_rate": 1.9751334064160708e-05, "loss": 1.556, "step": 3960 }, { "epoch": 9.9025, "grad_norm": 7.65625, "learning_rate": 1.975115459002623e-05, "loss": 1.4996, "step": 3961 }, { "epoch": 9.905, "grad_norm": 5.96875, "learning_rate": 1.9750975051963573e-05, "loss": 1.4529, "step": 3962 }, { "epoch": 9.9075, "grad_norm": 6.5, "learning_rate": 1.9750795449973914e-05, "loss": 1.5275, "step": 3963 }, { "epoch": 9.91, "grad_norm": 5.65625, "learning_rate": 1.9750615784058427e-05, "loss": 1.5538, "step": 3964 }, { "epoch": 9.9125, "grad_norm": 4.03125, "learning_rate": 1.97504360542183e-05, "loss": 1.4296, "step": 3965 }, { "epoch": 9.915, "grad_norm": 7.3125, "learning_rate": 1.97502562604547e-05, "loss": 1.693, "step": 3966 }, { "epoch": 9.9175, "grad_norm": 6.625, "learning_rate": 1.975007640276881e-05, "loss": 1.5964, "step": 3967 }, { "epoch": 9.92, "grad_norm": 6.0, "learning_rate": 1.9749896481161807e-05, "loss": 1.6459, "step": 3968 }, { "epoch": 9.9225, "grad_norm": 3.765625, "learning_rate": 1.9749716495634877e-05, "loss": 1.5336, "step": 3969 }, { "epoch": 9.925, "grad_norm": 7.09375, "learning_rate": 1.9749536446189194e-05, "loss": 1.5571, "step": 3970 }, { "epoch": 9.9275, "grad_norm": 6.84375, "learning_rate": 1.9749356332825944e-05, "loss": 1.728, "step": 3971 }, { "epoch": 9.93, "grad_norm": 6.21875, "learning_rate": 1.97491761555463e-05, "loss": 1.5771, "step": 3972 }, { "epoch": 9.9325, "grad_norm": 5.59375, "learning_rate": 1.9748995914351446e-05, "loss": 1.5194, "step": 3973 }, { "epoch": 9.935, "grad_norm": 9.0, "learning_rate": 1.974881560924257e-05, "loss": 1.6628, "step": 3974 }, { "epoch": 9.9375, "grad_norm": 7.28125, "learning_rate": 1.9748635240220846e-05, "loss": 1.4743, "step": 3975 }, { "epoch": 9.94, "grad_norm": 4.6875, "learning_rate": 1.9748454807287458e-05, "loss": 1.3162, "step": 3976 }, { "epoch": 9.9425, "grad_norm": 5.34375, "learning_rate": 1.9748274310443593e-05, "loss": 1.347, "step": 3977 }, { "epoch": 9.945, "grad_norm": 6.59375, "learning_rate": 1.9748093749690433e-05, "loss": 1.5073, "step": 3978 }, { "epoch": 9.9475, "grad_norm": 7.0625, "learning_rate": 1.9747913125029154e-05, "loss": 1.3777, "step": 3979 }, { "epoch": 9.95, "grad_norm": 5.65625, "learning_rate": 1.9747732436460955e-05, "loss": 1.4066, "step": 3980 }, { "epoch": 9.9525, "grad_norm": 6.375, "learning_rate": 1.9747551683987006e-05, "loss": 1.5519, "step": 3981 }, { "epoch": 9.955, "grad_norm": 5.75, "learning_rate": 1.97473708676085e-05, "loss": 1.4023, "step": 3982 }, { "epoch": 9.9575, "grad_norm": 11.1875, "learning_rate": 1.974718998732662e-05, "loss": 1.5595, "step": 3983 }, { "epoch": 9.96, "grad_norm": 6.3125, "learning_rate": 1.9747009043142556e-05, "loss": 1.4161, "step": 3984 }, { "epoch": 9.9625, "grad_norm": 5.5, "learning_rate": 1.9746828035057486e-05, "loss": 1.3466, "step": 3985 }, { "epoch": 9.965, "grad_norm": 5.4375, "learning_rate": 1.9746646963072603e-05, "loss": 1.4454, "step": 3986 }, { "epoch": 9.9675, "grad_norm": 4.78125, "learning_rate": 1.9746465827189095e-05, "loss": 1.4963, "step": 3987 }, { "epoch": 9.97, "grad_norm": 6.34375, "learning_rate": 1.9746284627408143e-05, "loss": 1.471, "step": 3988 }, { "epoch": 9.9725, "grad_norm": 5.21875, "learning_rate": 1.9746103363730942e-05, "loss": 1.4295, "step": 3989 }, { "epoch": 9.975, "grad_norm": 5.15625, "learning_rate": 1.9745922036158678e-05, "loss": 1.4336, "step": 3990 }, { "epoch": 9.9775, "grad_norm": 10.0, "learning_rate": 1.9745740644692537e-05, "loss": 1.6913, "step": 3991 }, { "epoch": 9.98, "grad_norm": 9.8125, "learning_rate": 1.974555918933371e-05, "loss": 1.501, "step": 3992 }, { "epoch": 9.9825, "grad_norm": 5.90625, "learning_rate": 1.974537767008339e-05, "loss": 1.5452, "step": 3993 }, { "epoch": 9.985, "grad_norm": 6.09375, "learning_rate": 1.9745196086942763e-05, "loss": 1.3897, "step": 3994 }, { "epoch": 9.9875, "grad_norm": 5.5, "learning_rate": 1.974501443991302e-05, "loss": 1.5394, "step": 3995 }, { "epoch": 9.99, "grad_norm": 3.765625, "learning_rate": 1.974483272899535e-05, "loss": 1.4717, "step": 3996 }, { "epoch": 9.9925, "grad_norm": 6.84375, "learning_rate": 1.974465095419095e-05, "loss": 1.4787, "step": 3997 }, { "epoch": 9.995, "grad_norm": 5.6875, "learning_rate": 1.9744469115501005e-05, "loss": 1.5986, "step": 3998 }, { "epoch": 9.9975, "grad_norm": 6.875, "learning_rate": 1.9744287212926713e-05, "loss": 1.5454, "step": 3999 }, { "epoch": 10.0, "grad_norm": 5.75, "learning_rate": 1.9744105246469264e-05, "loss": 1.4399, "step": 4000 }, { "epoch": 10.0025, "grad_norm": 8.0625, "learning_rate": 1.9743923216129847e-05, "loss": 1.4981, "step": 4001 }, { "epoch": 10.005, "grad_norm": 8.4375, "learning_rate": 1.974374112190966e-05, "loss": 1.6981, "step": 4002 }, { "epoch": 10.0075, "grad_norm": 7.59375, "learning_rate": 1.97435589638099e-05, "loss": 1.5227, "step": 4003 }, { "epoch": 10.01, "grad_norm": 7.71875, "learning_rate": 1.974337674183176e-05, "loss": 1.6321, "step": 4004 }, { "epoch": 10.0125, "grad_norm": 8.8125, "learning_rate": 1.9743194455976424e-05, "loss": 1.7143, "step": 4005 }, { "epoch": 10.015, "grad_norm": 5.3125, "learning_rate": 1.97430121062451e-05, "loss": 1.5301, "step": 4006 }, { "epoch": 10.0175, "grad_norm": 6.15625, "learning_rate": 1.974282969263898e-05, "loss": 1.5076, "step": 4007 }, { "epoch": 10.02, "grad_norm": 5.6875, "learning_rate": 1.9742647215159255e-05, "loss": 1.5733, "step": 4008 }, { "epoch": 10.0225, "grad_norm": 4.28125, "learning_rate": 1.9742464673807125e-05, "loss": 1.3627, "step": 4009 }, { "epoch": 10.025, "grad_norm": 6.0625, "learning_rate": 1.9742282068583787e-05, "loss": 1.675, "step": 4010 }, { "epoch": 10.0275, "grad_norm": 5.78125, "learning_rate": 1.9742099399490437e-05, "loss": 1.5203, "step": 4011 }, { "epoch": 10.03, "grad_norm": 4.34375, "learning_rate": 1.9741916666528273e-05, "loss": 1.4296, "step": 4012 }, { "epoch": 10.0325, "grad_norm": 8.4375, "learning_rate": 1.9741733869698497e-05, "loss": 1.4927, "step": 4013 }, { "epoch": 10.035, "grad_norm": 6.6875, "learning_rate": 1.9741551009002297e-05, "loss": 1.5914, "step": 4014 }, { "epoch": 10.0375, "grad_norm": 5.875, "learning_rate": 1.9741368084440883e-05, "loss": 1.5366, "step": 4015 }, { "epoch": 10.04, "grad_norm": 6.84375, "learning_rate": 1.974118509601545e-05, "loss": 1.4963, "step": 4016 }, { "epoch": 10.0425, "grad_norm": 7.875, "learning_rate": 1.9741002043727197e-05, "loss": 1.5671, "step": 4017 }, { "epoch": 10.045, "grad_norm": 5.375, "learning_rate": 1.9740818927577322e-05, "loss": 1.6855, "step": 4018 }, { "epoch": 10.0475, "grad_norm": 8.8125, "learning_rate": 1.9740635747567028e-05, "loss": 1.4378, "step": 4019 }, { "epoch": 10.05, "grad_norm": 6.15625, "learning_rate": 1.9740452503697518e-05, "loss": 1.4641, "step": 4020 }, { "epoch": 10.0525, "grad_norm": 5.03125, "learning_rate": 1.974026919596999e-05, "loss": 1.4773, "step": 4021 }, { "epoch": 10.055, "grad_norm": 5.6875, "learning_rate": 1.9740085824385648e-05, "loss": 1.4521, "step": 4022 }, { "epoch": 10.0575, "grad_norm": 7.625, "learning_rate": 1.9739902388945694e-05, "loss": 1.4927, "step": 4023 }, { "epoch": 10.06, "grad_norm": 5.03125, "learning_rate": 1.9739718889651328e-05, "loss": 1.4295, "step": 4024 }, { "epoch": 10.0625, "grad_norm": 7.09375, "learning_rate": 1.9739535326503753e-05, "loss": 1.5673, "step": 4025 }, { "epoch": 10.065, "grad_norm": 7.03125, "learning_rate": 1.9739351699504177e-05, "loss": 1.6381, "step": 4026 }, { "epoch": 10.0675, "grad_norm": 5.3125, "learning_rate": 1.9739168008653798e-05, "loss": 1.5574, "step": 4027 }, { "epoch": 10.07, "grad_norm": 5.53125, "learning_rate": 1.973898425395383e-05, "loss": 1.4675, "step": 4028 }, { "epoch": 10.0725, "grad_norm": 6.625, "learning_rate": 1.9738800435405463e-05, "loss": 1.4347, "step": 4029 }, { "epoch": 10.075, "grad_norm": 5.03125, "learning_rate": 1.9738616553009913e-05, "loss": 1.5587, "step": 4030 }, { "epoch": 10.0775, "grad_norm": 7.09375, "learning_rate": 1.9738432606768384e-05, "loss": 1.3993, "step": 4031 }, { "epoch": 10.08, "grad_norm": 5.28125, "learning_rate": 1.9738248596682078e-05, "loss": 1.4362, "step": 4032 }, { "epoch": 10.0825, "grad_norm": 7.96875, "learning_rate": 1.9738064522752206e-05, "loss": 1.6853, "step": 4033 }, { "epoch": 10.085, "grad_norm": 4.4375, "learning_rate": 1.9737880384979973e-05, "loss": 1.3366, "step": 4034 }, { "epoch": 10.0875, "grad_norm": 6.40625, "learning_rate": 1.9737696183366585e-05, "loss": 1.4409, "step": 4035 }, { "epoch": 10.09, "grad_norm": 6.75, "learning_rate": 1.973751191791325e-05, "loss": 1.5065, "step": 4036 }, { "epoch": 10.0925, "grad_norm": 5.625, "learning_rate": 1.9737327588621178e-05, "loss": 1.5928, "step": 4037 }, { "epoch": 10.095, "grad_norm": 12.4375, "learning_rate": 1.9737143195491574e-05, "loss": 1.5415, "step": 4038 }, { "epoch": 10.0975, "grad_norm": 7.5, "learning_rate": 1.9736958738525648e-05, "loss": 1.5159, "step": 4039 }, { "epoch": 10.1, "grad_norm": 6.53125, "learning_rate": 1.9736774217724614e-05, "loss": 1.4185, "step": 4040 }, { "epoch": 10.1025, "grad_norm": 6.21875, "learning_rate": 1.9736589633089677e-05, "loss": 1.4883, "step": 4041 }, { "epoch": 10.105, "grad_norm": 10.5625, "learning_rate": 1.9736404984622046e-05, "loss": 1.5568, "step": 4042 }, { "epoch": 10.1075, "grad_norm": 7.125, "learning_rate": 1.9736220272322933e-05, "loss": 1.4979, "step": 4043 }, { "epoch": 10.11, "grad_norm": 7.40625, "learning_rate": 1.9736035496193552e-05, "loss": 1.5685, "step": 4044 }, { "epoch": 10.1125, "grad_norm": 6.59375, "learning_rate": 1.9735850656235112e-05, "loss": 1.4914, "step": 4045 }, { "epoch": 10.115, "grad_norm": 8.375, "learning_rate": 1.9735665752448822e-05, "loss": 1.5564, "step": 4046 }, { "epoch": 10.1175, "grad_norm": 4.78125, "learning_rate": 1.97354807848359e-05, "loss": 1.5093, "step": 4047 }, { "epoch": 10.12, "grad_norm": 5.875, "learning_rate": 1.973529575339755e-05, "loss": 1.6958, "step": 4048 }, { "epoch": 10.1225, "grad_norm": 5.5625, "learning_rate": 1.9735110658134996e-05, "loss": 1.3487, "step": 4049 }, { "epoch": 10.125, "grad_norm": 5.4375, "learning_rate": 1.9734925499049446e-05, "loss": 1.4093, "step": 4050 }, { "epoch": 10.1275, "grad_norm": 6.90625, "learning_rate": 1.973474027614211e-05, "loss": 1.4345, "step": 4051 }, { "epoch": 10.13, "grad_norm": 6.28125, "learning_rate": 1.9734554989414212e-05, "loss": 1.3676, "step": 4052 }, { "epoch": 10.1325, "grad_norm": 7.40625, "learning_rate": 1.973436963886696e-05, "loss": 1.4729, "step": 4053 }, { "epoch": 10.135, "grad_norm": 5.59375, "learning_rate": 1.9734184224501567e-05, "loss": 1.6726, "step": 4054 }, { "epoch": 10.1375, "grad_norm": 10.875, "learning_rate": 1.9733998746319255e-05, "loss": 1.5897, "step": 4055 }, { "epoch": 10.14, "grad_norm": 4.59375, "learning_rate": 1.9733813204321233e-05, "loss": 1.3682, "step": 4056 }, { "epoch": 10.1425, "grad_norm": 4.84375, "learning_rate": 1.973362759850872e-05, "loss": 1.4838, "step": 4057 }, { "epoch": 10.145, "grad_norm": 7.0, "learning_rate": 1.973344192888294e-05, "loss": 1.6071, "step": 4058 }, { "epoch": 10.1475, "grad_norm": 6.96875, "learning_rate": 1.97332561954451e-05, "loss": 1.4746, "step": 4059 }, { "epoch": 10.15, "grad_norm": 6.28125, "learning_rate": 1.9733070398196423e-05, "loss": 1.6267, "step": 4060 }, { "epoch": 10.1525, "grad_norm": 8.125, "learning_rate": 1.9732884537138125e-05, "loss": 1.456, "step": 4061 }, { "epoch": 10.155, "grad_norm": 5.84375, "learning_rate": 1.973269861227143e-05, "loss": 1.5517, "step": 4062 }, { "epoch": 10.1575, "grad_norm": 7.46875, "learning_rate": 1.973251262359755e-05, "loss": 1.4218, "step": 4063 }, { "epoch": 10.16, "grad_norm": 4.0625, "learning_rate": 1.9732326571117703e-05, "loss": 1.4354, "step": 4064 }, { "epoch": 10.1625, "grad_norm": 5.53125, "learning_rate": 1.9732140454833114e-05, "loss": 1.3225, "step": 4065 }, { "epoch": 10.165, "grad_norm": 9.1875, "learning_rate": 1.9731954274745004e-05, "loss": 1.6648, "step": 4066 }, { "epoch": 10.1675, "grad_norm": 6.625, "learning_rate": 1.9731768030854587e-05, "loss": 1.4658, "step": 4067 }, { "epoch": 10.17, "grad_norm": 7.15625, "learning_rate": 1.9731581723163094e-05, "loss": 1.4605, "step": 4068 }, { "epoch": 10.1725, "grad_norm": 9.0625, "learning_rate": 1.9731395351671737e-05, "loss": 1.6082, "step": 4069 }, { "epoch": 10.175, "grad_norm": 5.5, "learning_rate": 1.9731208916381744e-05, "loss": 1.5569, "step": 4070 }, { "epoch": 10.1775, "grad_norm": 6.875, "learning_rate": 1.9731022417294334e-05, "loss": 1.435, "step": 4071 }, { "epoch": 10.18, "grad_norm": 7.78125, "learning_rate": 1.9730835854410726e-05, "loss": 1.4894, "step": 4072 }, { "epoch": 10.1825, "grad_norm": 6.75, "learning_rate": 1.973064922773215e-05, "loss": 1.6216, "step": 4073 }, { "epoch": 10.185, "grad_norm": 6.75, "learning_rate": 1.9730462537259832e-05, "loss": 1.5642, "step": 4074 }, { "epoch": 10.1875, "grad_norm": 6.15625, "learning_rate": 1.9730275782994984e-05, "loss": 1.6464, "step": 4075 }, { "epoch": 10.19, "grad_norm": 6.3125, "learning_rate": 1.9730088964938842e-05, "loss": 1.567, "step": 4076 }, { "epoch": 10.1925, "grad_norm": 6.34375, "learning_rate": 1.9729902083092625e-05, "loss": 1.4528, "step": 4077 }, { "epoch": 10.195, "grad_norm": 5.53125, "learning_rate": 1.972971513745756e-05, "loss": 1.5602, "step": 4078 }, { "epoch": 10.1975, "grad_norm": 6.625, "learning_rate": 1.972952812803487e-05, "loss": 1.389, "step": 4079 }, { "epoch": 10.2, "grad_norm": 6.125, "learning_rate": 1.9729341054825783e-05, "loss": 1.4835, "step": 4080 }, { "epoch": 10.2025, "grad_norm": 4.78125, "learning_rate": 1.9729153917831528e-05, "loss": 1.675, "step": 4081 }, { "epoch": 10.205, "grad_norm": 8.125, "learning_rate": 1.9728966717053326e-05, "loss": 1.5417, "step": 4082 }, { "epoch": 10.2075, "grad_norm": 8.1875, "learning_rate": 1.972877945249241e-05, "loss": 1.6642, "step": 4083 }, { "epoch": 10.21, "grad_norm": 6.25, "learning_rate": 1.9728592124150003e-05, "loss": 1.4976, "step": 4084 }, { "epoch": 10.2125, "grad_norm": 6.5, "learning_rate": 1.9728404732027337e-05, "loss": 1.699, "step": 4085 }, { "epoch": 10.215, "grad_norm": 5.6875, "learning_rate": 1.9728217276125638e-05, "loss": 1.3552, "step": 4086 }, { "epoch": 10.2175, "grad_norm": 5.625, "learning_rate": 1.9728029756446135e-05, "loss": 1.4579, "step": 4087 }, { "epoch": 10.22, "grad_norm": 10.5, "learning_rate": 1.972784217299006e-05, "loss": 1.4712, "step": 4088 }, { "epoch": 10.2225, "grad_norm": 18.0, "learning_rate": 1.972765452575864e-05, "loss": 1.5773, "step": 4089 }, { "epoch": 10.225, "grad_norm": 5.9375, "learning_rate": 1.972746681475311e-05, "loss": 1.4899, "step": 4090 }, { "epoch": 10.2275, "grad_norm": 14.375, "learning_rate": 1.9727279039974693e-05, "loss": 1.6816, "step": 4091 }, { "epoch": 10.23, "grad_norm": 8.5, "learning_rate": 1.9727091201424623e-05, "loss": 1.5677, "step": 4092 }, { "epoch": 10.2325, "grad_norm": 6.0625, "learning_rate": 1.9726903299104133e-05, "loss": 1.4841, "step": 4093 }, { "epoch": 10.235, "grad_norm": 6.46875, "learning_rate": 1.9726715333014455e-05, "loss": 1.5604, "step": 4094 }, { "epoch": 10.2375, "grad_norm": 4.6875, "learning_rate": 1.972652730315682e-05, "loss": 1.7136, "step": 4095 }, { "epoch": 10.24, "grad_norm": 6.84375, "learning_rate": 1.9726339209532462e-05, "loss": 1.4195, "step": 4096 }, { "epoch": 10.2425, "grad_norm": 8.0625, "learning_rate": 1.9726151052142613e-05, "loss": 1.661, "step": 4097 }, { "epoch": 10.245, "grad_norm": 7.3125, "learning_rate": 1.9725962830988507e-05, "loss": 1.6983, "step": 4098 }, { "epoch": 10.2475, "grad_norm": 7.59375, "learning_rate": 1.972577454607138e-05, "loss": 1.5094, "step": 4099 }, { "epoch": 10.25, "grad_norm": 5.0625, "learning_rate": 1.972558619739246e-05, "loss": 1.4073, "step": 4100 }, { "epoch": 10.2525, "grad_norm": 7.125, "learning_rate": 1.972539778495299e-05, "loss": 1.5188, "step": 4101 }, { "epoch": 10.255, "grad_norm": 6.6875, "learning_rate": 1.97252093087542e-05, "loss": 1.5515, "step": 4102 }, { "epoch": 10.2575, "grad_norm": 6.21875, "learning_rate": 1.9725020768797325e-05, "loss": 1.5846, "step": 4103 }, { "epoch": 10.26, "grad_norm": 7.5625, "learning_rate": 1.9724832165083607e-05, "loss": 1.5535, "step": 4104 }, { "epoch": 10.2625, "grad_norm": 4.5625, "learning_rate": 1.9724643497614274e-05, "loss": 1.5567, "step": 4105 }, { "epoch": 10.265, "grad_norm": 8.25, "learning_rate": 1.9724454766390567e-05, "loss": 1.5332, "step": 4106 }, { "epoch": 10.2675, "grad_norm": 5.625, "learning_rate": 1.972426597141373e-05, "loss": 1.3552, "step": 4107 }, { "epoch": 10.27, "grad_norm": 6.03125, "learning_rate": 1.972407711268499e-05, "loss": 1.5395, "step": 4108 }, { "epoch": 10.2725, "grad_norm": 7.53125, "learning_rate": 1.972388819020559e-05, "loss": 1.5649, "step": 4109 }, { "epoch": 10.275, "grad_norm": 6.125, "learning_rate": 1.9723699203976768e-05, "loss": 1.5075, "step": 4110 }, { "epoch": 10.2775, "grad_norm": 6.90625, "learning_rate": 1.9723510153999764e-05, "loss": 1.4889, "step": 4111 }, { "epoch": 10.28, "grad_norm": 6.625, "learning_rate": 1.9723321040275816e-05, "loss": 1.5656, "step": 4112 }, { "epoch": 10.2825, "grad_norm": 8.1875, "learning_rate": 1.9723131862806163e-05, "loss": 1.3634, "step": 4113 }, { "epoch": 10.285, "grad_norm": 10.8125, "learning_rate": 1.972294262159205e-05, "loss": 1.6004, "step": 4114 }, { "epoch": 10.2875, "grad_norm": 8.1875, "learning_rate": 1.972275331663471e-05, "loss": 1.3782, "step": 4115 }, { "epoch": 10.29, "grad_norm": 7.1875, "learning_rate": 1.972256394793539e-05, "loss": 1.4834, "step": 4116 }, { "epoch": 10.2925, "grad_norm": 7.9375, "learning_rate": 1.9722374515495332e-05, "loss": 1.6621, "step": 4117 }, { "epoch": 10.295, "grad_norm": 5.71875, "learning_rate": 1.9722185019315775e-05, "loss": 1.5871, "step": 4118 }, { "epoch": 10.2975, "grad_norm": 5.6875, "learning_rate": 1.972199545939796e-05, "loss": 1.5362, "step": 4119 }, { "epoch": 10.3, "grad_norm": 7.78125, "learning_rate": 1.972180583574313e-05, "loss": 1.4769, "step": 4120 }, { "epoch": 10.3025, "grad_norm": 6.0625, "learning_rate": 1.9721616148352537e-05, "loss": 1.4601, "step": 4121 }, { "epoch": 10.305, "grad_norm": 7.09375, "learning_rate": 1.9721426397227412e-05, "loss": 1.4638, "step": 4122 }, { "epoch": 10.3075, "grad_norm": 6.65625, "learning_rate": 1.9721236582369004e-05, "loss": 1.5784, "step": 4123 }, { "epoch": 10.31, "grad_norm": 7.0625, "learning_rate": 1.9721046703778563e-05, "loss": 1.4167, "step": 4124 }, { "epoch": 10.3125, "grad_norm": 6.0625, "learning_rate": 1.9720856761457326e-05, "loss": 1.6787, "step": 4125 }, { "epoch": 10.315, "grad_norm": 7.03125, "learning_rate": 1.972066675540654e-05, "loss": 1.4746, "step": 4126 }, { "epoch": 10.3175, "grad_norm": 8.75, "learning_rate": 1.972047668562745e-05, "loss": 1.453, "step": 4127 }, { "epoch": 10.32, "grad_norm": 6.375, "learning_rate": 1.972028655212131e-05, "loss": 1.4429, "step": 4128 }, { "epoch": 10.3225, "grad_norm": 4.53125, "learning_rate": 1.9720096354889356e-05, "loss": 1.4211, "step": 4129 }, { "epoch": 10.325, "grad_norm": 6.25, "learning_rate": 1.971990609393284e-05, "loss": 1.3997, "step": 4130 }, { "epoch": 10.3275, "grad_norm": 6.3125, "learning_rate": 1.971971576925301e-05, "loss": 1.6105, "step": 4131 }, { "epoch": 10.33, "grad_norm": 9.6875, "learning_rate": 1.971952538085111e-05, "loss": 1.6307, "step": 4132 }, { "epoch": 10.3325, "grad_norm": 7.46875, "learning_rate": 1.9719334928728393e-05, "loss": 1.6577, "step": 4133 }, { "epoch": 10.335, "grad_norm": 5.28125, "learning_rate": 1.9719144412886104e-05, "loss": 1.5759, "step": 4134 }, { "epoch": 10.3375, "grad_norm": 5.375, "learning_rate": 1.9718953833325495e-05, "loss": 1.5686, "step": 4135 }, { "epoch": 10.34, "grad_norm": 6.8125, "learning_rate": 1.971876319004781e-05, "loss": 1.5411, "step": 4136 }, { "epoch": 10.3425, "grad_norm": 5.5625, "learning_rate": 1.9718572483054306e-05, "loss": 1.4874, "step": 4137 }, { "epoch": 10.345, "grad_norm": 6.125, "learning_rate": 1.9718381712346227e-05, "loss": 1.4739, "step": 4138 }, { "epoch": 10.3475, "grad_norm": 4.71875, "learning_rate": 1.9718190877924828e-05, "loss": 1.5236, "step": 4139 }, { "epoch": 10.35, "grad_norm": 6.25, "learning_rate": 1.9717999979791356e-05, "loss": 1.5028, "step": 4140 }, { "epoch": 10.3525, "grad_norm": 6.21875, "learning_rate": 1.9717809017947067e-05, "loss": 1.5879, "step": 4141 }, { "epoch": 10.355, "grad_norm": 7.125, "learning_rate": 1.971761799239321e-05, "loss": 1.51, "step": 4142 }, { "epoch": 10.3575, "grad_norm": 5.125, "learning_rate": 1.971742690313104e-05, "loss": 1.5189, "step": 4143 }, { "epoch": 10.36, "grad_norm": 5.46875, "learning_rate": 1.9717235750161808e-05, "loss": 1.3587, "step": 4144 }, { "epoch": 10.3625, "grad_norm": 8.1875, "learning_rate": 1.9717044533486766e-05, "loss": 1.4251, "step": 4145 }, { "epoch": 10.365, "grad_norm": 6.53125, "learning_rate": 1.971685325310717e-05, "loss": 1.556, "step": 4146 }, { "epoch": 10.3675, "grad_norm": 5.25, "learning_rate": 1.971666190902427e-05, "loss": 1.4129, "step": 4147 }, { "epoch": 10.37, "grad_norm": 11.1875, "learning_rate": 1.9716470501239327e-05, "loss": 1.4306, "step": 4148 }, { "epoch": 10.3725, "grad_norm": 5.65625, "learning_rate": 1.971627902975359e-05, "loss": 1.4625, "step": 4149 }, { "epoch": 10.375, "grad_norm": 6.1875, "learning_rate": 1.9716087494568318e-05, "loss": 1.3614, "step": 4150 }, { "epoch": 10.3775, "grad_norm": 6.0, "learning_rate": 1.9715895895684763e-05, "loss": 1.5853, "step": 4151 }, { "epoch": 10.38, "grad_norm": 5.5625, "learning_rate": 1.9715704233104188e-05, "loss": 1.4561, "step": 4152 }, { "epoch": 10.3825, "grad_norm": 6.875, "learning_rate": 1.971551250682784e-05, "loss": 1.6339, "step": 4153 }, { "epoch": 10.385, "grad_norm": 6.0, "learning_rate": 1.9715320716856982e-05, "loss": 1.5762, "step": 4154 }, { "epoch": 10.3875, "grad_norm": 6.0, "learning_rate": 1.9715128863192872e-05, "loss": 1.5785, "step": 4155 }, { "epoch": 10.39, "grad_norm": 6.5625, "learning_rate": 1.9714936945836764e-05, "loss": 1.4848, "step": 4156 }, { "epoch": 10.3925, "grad_norm": 6.5, "learning_rate": 1.971474496478992e-05, "loss": 1.468, "step": 4157 }, { "epoch": 10.395, "grad_norm": 8.125, "learning_rate": 1.9714552920053592e-05, "loss": 1.2734, "step": 4158 }, { "epoch": 10.3975, "grad_norm": 6.5625, "learning_rate": 1.971436081162905e-05, "loss": 1.4876, "step": 4159 }, { "epoch": 10.4, "grad_norm": 6.21875, "learning_rate": 1.9714168639517543e-05, "loss": 1.5033, "step": 4160 }, { "epoch": 10.4025, "grad_norm": 6.5625, "learning_rate": 1.9713976403720337e-05, "loss": 1.4511, "step": 4161 }, { "epoch": 10.405, "grad_norm": 6.6875, "learning_rate": 1.971378410423869e-05, "loss": 1.5016, "step": 4162 }, { "epoch": 10.4075, "grad_norm": 7.03125, "learning_rate": 1.9713591741073865e-05, "loss": 1.4234, "step": 4163 }, { "epoch": 10.41, "grad_norm": 7.34375, "learning_rate": 1.971339931422712e-05, "loss": 1.4709, "step": 4164 }, { "epoch": 10.4125, "grad_norm": 7.5, "learning_rate": 1.9713206823699716e-05, "loss": 1.4945, "step": 4165 }, { "epoch": 10.415, "grad_norm": 6.96875, "learning_rate": 1.971301426949292e-05, "loss": 1.4696, "step": 4166 }, { "epoch": 10.4175, "grad_norm": 8.0, "learning_rate": 1.971282165160799e-05, "loss": 1.3753, "step": 4167 }, { "epoch": 10.42, "grad_norm": 6.15625, "learning_rate": 1.971262897004619e-05, "loss": 1.5605, "step": 4168 }, { "epoch": 10.4225, "grad_norm": 8.3125, "learning_rate": 1.971243622480878e-05, "loss": 1.6512, "step": 4169 }, { "epoch": 10.425, "grad_norm": 7.3125, "learning_rate": 1.971224341589703e-05, "loss": 1.4448, "step": 4170 }, { "epoch": 10.4275, "grad_norm": 6.78125, "learning_rate": 1.97120505433122e-05, "loss": 1.6021, "step": 4171 }, { "epoch": 10.43, "grad_norm": 8.875, "learning_rate": 1.9711857607055553e-05, "loss": 1.3952, "step": 4172 }, { "epoch": 10.4325, "grad_norm": 5.25, "learning_rate": 1.971166460712836e-05, "loss": 1.4764, "step": 4173 }, { "epoch": 10.435, "grad_norm": 8.1875, "learning_rate": 1.971147154353188e-05, "loss": 1.7026, "step": 4174 }, { "epoch": 10.4375, "grad_norm": 6.4375, "learning_rate": 1.9711278416267385e-05, "loss": 1.4457, "step": 4175 }, { "epoch": 10.44, "grad_norm": 8.875, "learning_rate": 1.971108522533613e-05, "loss": 1.7021, "step": 4176 }, { "epoch": 10.4425, "grad_norm": 6.71875, "learning_rate": 1.9710891970739395e-05, "loss": 1.4999, "step": 4177 }, { "epoch": 10.445, "grad_norm": 5.0, "learning_rate": 1.9710698652478443e-05, "loss": 1.4255, "step": 4178 }, { "epoch": 10.4475, "grad_norm": 5.8125, "learning_rate": 1.971050527055453e-05, "loss": 1.4373, "step": 4179 }, { "epoch": 10.45, "grad_norm": 10.0, "learning_rate": 1.9710311824968942e-05, "loss": 1.5447, "step": 4180 }, { "epoch": 10.4525, "grad_norm": 5.28125, "learning_rate": 1.9710118315722934e-05, "loss": 1.5068, "step": 4181 }, { "epoch": 10.455, "grad_norm": 8.5625, "learning_rate": 1.970992474281778e-05, "loss": 1.712, "step": 4182 }, { "epoch": 10.4575, "grad_norm": 5.3125, "learning_rate": 1.9709731106254747e-05, "loss": 1.4287, "step": 4183 }, { "epoch": 10.46, "grad_norm": 6.25, "learning_rate": 1.9709537406035106e-05, "loss": 1.3846, "step": 4184 }, { "epoch": 10.4625, "grad_norm": 8.375, "learning_rate": 1.9709343642160128e-05, "loss": 1.5744, "step": 4185 }, { "epoch": 10.465, "grad_norm": 5.375, "learning_rate": 1.970914981463108e-05, "loss": 1.451, "step": 4186 }, { "epoch": 10.4675, "grad_norm": 6.96875, "learning_rate": 1.9708955923449235e-05, "loss": 1.4367, "step": 4187 }, { "epoch": 10.47, "grad_norm": 6.5, "learning_rate": 1.970876196861586e-05, "loss": 1.6182, "step": 4188 }, { "epoch": 10.4725, "grad_norm": 9.5, "learning_rate": 1.9708567950132235e-05, "loss": 1.7065, "step": 4189 }, { "epoch": 10.475, "grad_norm": 6.875, "learning_rate": 1.9708373867999624e-05, "loss": 1.5154, "step": 4190 }, { "epoch": 10.4775, "grad_norm": 6.125, "learning_rate": 1.9708179722219303e-05, "loss": 1.5226, "step": 4191 }, { "epoch": 10.48, "grad_norm": 7.375, "learning_rate": 1.9707985512792544e-05, "loss": 1.746, "step": 4192 }, { "epoch": 10.4825, "grad_norm": 7.0625, "learning_rate": 1.970779123972062e-05, "loss": 1.5299, "step": 4193 }, { "epoch": 10.485, "grad_norm": 6.84375, "learning_rate": 1.9707596903004806e-05, "loss": 1.614, "step": 4194 }, { "epoch": 10.4875, "grad_norm": 9.5, "learning_rate": 1.970740250264638e-05, "loss": 1.4429, "step": 4195 }, { "epoch": 10.49, "grad_norm": 4.5625, "learning_rate": 1.9707208038646605e-05, "loss": 1.3502, "step": 4196 }, { "epoch": 10.4925, "grad_norm": 4.625, "learning_rate": 1.9707013511006763e-05, "loss": 1.419, "step": 4197 }, { "epoch": 10.495, "grad_norm": 6.03125, "learning_rate": 1.970681891972813e-05, "loss": 1.5191, "step": 4198 }, { "epoch": 10.4975, "grad_norm": 14.1875, "learning_rate": 1.970662426481198e-05, "loss": 1.5198, "step": 4199 }, { "epoch": 10.5, "grad_norm": 9.625, "learning_rate": 1.9706429546259592e-05, "loss": 1.5092, "step": 4200 }, { "epoch": 10.5025, "grad_norm": 5.40625, "learning_rate": 1.970623476407224e-05, "loss": 1.7234, "step": 4201 }, { "epoch": 10.505, "grad_norm": 7.5, "learning_rate": 1.97060399182512e-05, "loss": 1.3716, "step": 4202 }, { "epoch": 10.5075, "grad_norm": 6.03125, "learning_rate": 1.9705845008797752e-05, "loss": 1.385, "step": 4203 }, { "epoch": 10.51, "grad_norm": 8.125, "learning_rate": 1.9705650035713172e-05, "loss": 1.687, "step": 4204 }, { "epoch": 10.5125, "grad_norm": 7.96875, "learning_rate": 1.9705454998998736e-05, "loss": 1.5265, "step": 4205 }, { "epoch": 10.515, "grad_norm": 6.875, "learning_rate": 1.970525989865573e-05, "loss": 1.5006, "step": 4206 }, { "epoch": 10.5175, "grad_norm": 8.625, "learning_rate": 1.9705064734685425e-05, "loss": 1.5247, "step": 4207 }, { "epoch": 10.52, "grad_norm": 5.34375, "learning_rate": 1.9704869507089105e-05, "loss": 1.4929, "step": 4208 }, { "epoch": 10.5225, "grad_norm": 6.40625, "learning_rate": 1.970467421586805e-05, "loss": 1.4047, "step": 4209 }, { "epoch": 10.525, "grad_norm": 7.71875, "learning_rate": 1.970447886102354e-05, "loss": 1.5775, "step": 4210 }, { "epoch": 10.5275, "grad_norm": 9.25, "learning_rate": 1.9704283442556853e-05, "loss": 1.5956, "step": 4211 }, { "epoch": 10.53, "grad_norm": 11.3125, "learning_rate": 1.970408796046927e-05, "loss": 1.7132, "step": 4212 }, { "epoch": 10.5325, "grad_norm": 6.46875, "learning_rate": 1.970389241476208e-05, "loss": 1.4078, "step": 4213 }, { "epoch": 10.535, "grad_norm": 6.65625, "learning_rate": 1.9703696805436557e-05, "loss": 1.5118, "step": 4214 }, { "epoch": 10.5375, "grad_norm": 6.09375, "learning_rate": 1.9703501132493986e-05, "loss": 1.3887, "step": 4215 }, { "epoch": 10.54, "grad_norm": 6.375, "learning_rate": 1.970330539593565e-05, "loss": 1.5583, "step": 4216 }, { "epoch": 10.5425, "grad_norm": 5.40625, "learning_rate": 1.970310959576283e-05, "loss": 1.5404, "step": 4217 }, { "epoch": 10.545, "grad_norm": 12.0, "learning_rate": 1.9702913731976813e-05, "loss": 1.5524, "step": 4218 }, { "epoch": 10.5475, "grad_norm": 5.71875, "learning_rate": 1.9702717804578884e-05, "loss": 1.5059, "step": 4219 }, { "epoch": 10.55, "grad_norm": 4.875, "learning_rate": 1.9702521813570322e-05, "loss": 1.6252, "step": 4220 }, { "epoch": 10.5525, "grad_norm": 9.5625, "learning_rate": 1.9702325758952418e-05, "loss": 1.4835, "step": 4221 }, { "epoch": 10.555, "grad_norm": 6.84375, "learning_rate": 1.9702129640726457e-05, "loss": 1.3466, "step": 4222 }, { "epoch": 10.5575, "grad_norm": 8.1875, "learning_rate": 1.9701933458893715e-05, "loss": 1.5288, "step": 4223 }, { "epoch": 10.56, "grad_norm": 7.46875, "learning_rate": 1.970173721345549e-05, "loss": 1.6823, "step": 4224 }, { "epoch": 10.5625, "grad_norm": 7.40625, "learning_rate": 1.9701540904413067e-05, "loss": 1.3999, "step": 4225 }, { "epoch": 10.565, "grad_norm": 5.34375, "learning_rate": 1.9701344531767722e-05, "loss": 1.3575, "step": 4226 }, { "epoch": 10.5675, "grad_norm": 7.5625, "learning_rate": 1.9701148095520757e-05, "loss": 1.6217, "step": 4227 }, { "epoch": 10.57, "grad_norm": 6.09375, "learning_rate": 1.970095159567345e-05, "loss": 1.5081, "step": 4228 }, { "epoch": 10.5725, "grad_norm": 8.75, "learning_rate": 1.9700755032227096e-05, "loss": 1.4289, "step": 4229 }, { "epoch": 10.575, "grad_norm": 6.65625, "learning_rate": 1.9700558405182978e-05, "loss": 1.4986, "step": 4230 }, { "epoch": 10.5775, "grad_norm": 8.125, "learning_rate": 1.9700361714542387e-05, "loss": 1.3692, "step": 4231 }, { "epoch": 10.58, "grad_norm": 8.3125, "learning_rate": 1.9700164960306612e-05, "loss": 1.4421, "step": 4232 }, { "epoch": 10.5825, "grad_norm": 4.21875, "learning_rate": 1.9699968142476946e-05, "loss": 1.3429, "step": 4233 }, { "epoch": 10.585, "grad_norm": 5.65625, "learning_rate": 1.9699771261054678e-05, "loss": 1.5455, "step": 4234 }, { "epoch": 10.5875, "grad_norm": 8.0, "learning_rate": 1.9699574316041094e-05, "loss": 1.4494, "step": 4235 }, { "epoch": 10.59, "grad_norm": 6.09375, "learning_rate": 1.969937730743749e-05, "loss": 1.54, "step": 4236 }, { "epoch": 10.5925, "grad_norm": 6.78125, "learning_rate": 1.9699180235245157e-05, "loss": 1.4162, "step": 4237 }, { "epoch": 10.595, "grad_norm": 6.09375, "learning_rate": 1.9698983099465387e-05, "loss": 1.2868, "step": 4238 }, { "epoch": 10.5975, "grad_norm": 7.5625, "learning_rate": 1.9698785900099474e-05, "loss": 1.6739, "step": 4239 }, { "epoch": 10.6, "grad_norm": 5.53125, "learning_rate": 1.9698588637148705e-05, "loss": 1.3495, "step": 4240 }, { "epoch": 10.6025, "grad_norm": 4.90625, "learning_rate": 1.9698391310614378e-05, "loss": 1.5591, "step": 4241 }, { "epoch": 10.605, "grad_norm": 6.34375, "learning_rate": 1.9698193920497787e-05, "loss": 1.5926, "step": 4242 }, { "epoch": 10.6075, "grad_norm": 8.375, "learning_rate": 1.9697996466800222e-05, "loss": 1.6126, "step": 4243 }, { "epoch": 10.61, "grad_norm": 7.5, "learning_rate": 1.9697798949522983e-05, "loss": 1.6016, "step": 4244 }, { "epoch": 10.6125, "grad_norm": 7.875, "learning_rate": 1.9697601368667362e-05, "loss": 1.5505, "step": 4245 }, { "epoch": 10.615, "grad_norm": 7.875, "learning_rate": 1.9697403724234654e-05, "loss": 1.5033, "step": 4246 }, { "epoch": 10.6175, "grad_norm": 8.4375, "learning_rate": 1.9697206016226154e-05, "loss": 1.5482, "step": 4247 }, { "epoch": 10.62, "grad_norm": 5.59375, "learning_rate": 1.9697008244643163e-05, "loss": 1.3788, "step": 4248 }, { "epoch": 10.6225, "grad_norm": 4.4375, "learning_rate": 1.969681040948697e-05, "loss": 1.4007, "step": 4249 }, { "epoch": 10.625, "grad_norm": 8.875, "learning_rate": 1.9696612510758878e-05, "loss": 1.3806, "step": 4250 }, { "epoch": 10.6275, "grad_norm": 5.9375, "learning_rate": 1.969641454846018e-05, "loss": 1.6486, "step": 4251 }, { "epoch": 10.63, "grad_norm": 6.6875, "learning_rate": 1.969621652259218e-05, "loss": 1.699, "step": 4252 }, { "epoch": 10.6325, "grad_norm": 6.25, "learning_rate": 1.969601843315617e-05, "loss": 1.473, "step": 4253 }, { "epoch": 10.635, "grad_norm": 7.84375, "learning_rate": 1.9695820280153454e-05, "loss": 1.5341, "step": 4254 }, { "epoch": 10.6375, "grad_norm": 6.09375, "learning_rate": 1.9695622063585325e-05, "loss": 1.6324, "step": 4255 }, { "epoch": 10.64, "grad_norm": 5.15625, "learning_rate": 1.9695423783453086e-05, "loss": 1.4273, "step": 4256 }, { "epoch": 10.6425, "grad_norm": 6.53125, "learning_rate": 1.9695225439758038e-05, "loss": 1.3997, "step": 4257 }, { "epoch": 10.645, "grad_norm": 8.0, "learning_rate": 1.969502703250148e-05, "loss": 1.6777, "step": 4258 }, { "epoch": 10.6475, "grad_norm": 5.53125, "learning_rate": 1.969482856168471e-05, "loss": 1.4651, "step": 4259 }, { "epoch": 10.65, "grad_norm": 8.4375, "learning_rate": 1.9694630027309035e-05, "loss": 1.4745, "step": 4260 }, { "epoch": 10.6525, "grad_norm": 5.46875, "learning_rate": 1.9694431429375753e-05, "loss": 1.5256, "step": 4261 }, { "epoch": 10.655, "grad_norm": 5.4375, "learning_rate": 1.9694232767886164e-05, "loss": 1.452, "step": 4262 }, { "epoch": 10.6575, "grad_norm": 6.15625, "learning_rate": 1.9694034042841576e-05, "loss": 1.414, "step": 4263 }, { "epoch": 10.66, "grad_norm": 5.71875, "learning_rate": 1.969383525424329e-05, "loss": 1.5797, "step": 4264 }, { "epoch": 10.6625, "grad_norm": 6.5625, "learning_rate": 1.9693636402092604e-05, "loss": 1.5329, "step": 4265 }, { "epoch": 10.665, "grad_norm": 4.84375, "learning_rate": 1.9693437486390826e-05, "loss": 1.492, "step": 4266 }, { "epoch": 10.6675, "grad_norm": 7.40625, "learning_rate": 1.9693238507139263e-05, "loss": 1.3822, "step": 4267 }, { "epoch": 10.67, "grad_norm": 6.5625, "learning_rate": 1.9693039464339216e-05, "loss": 1.5452, "step": 4268 }, { "epoch": 10.6725, "grad_norm": 7.9375, "learning_rate": 1.969284035799199e-05, "loss": 1.7231, "step": 4269 }, { "epoch": 10.675, "grad_norm": 5.6875, "learning_rate": 1.9692641188098886e-05, "loss": 1.4611, "step": 4270 }, { "epoch": 10.6775, "grad_norm": 5.53125, "learning_rate": 1.969244195466122e-05, "loss": 1.4935, "step": 4271 }, { "epoch": 10.68, "grad_norm": 5.84375, "learning_rate": 1.9692242657680286e-05, "loss": 1.5308, "step": 4272 }, { "epoch": 10.6825, "grad_norm": 6.4375, "learning_rate": 1.96920432971574e-05, "loss": 1.6031, "step": 4273 }, { "epoch": 10.685, "grad_norm": 6.5625, "learning_rate": 1.969184387309387e-05, "loss": 1.4791, "step": 4274 }, { "epoch": 10.6875, "grad_norm": 6.40625, "learning_rate": 1.9691644385490996e-05, "loss": 1.5847, "step": 4275 }, { "epoch": 10.69, "grad_norm": 7.09375, "learning_rate": 1.969144483435009e-05, "loss": 1.642, "step": 4276 }, { "epoch": 10.6925, "grad_norm": 7.625, "learning_rate": 1.9691245219672457e-05, "loss": 1.6311, "step": 4277 }, { "epoch": 10.695, "grad_norm": 9.0625, "learning_rate": 1.969104554145941e-05, "loss": 1.6462, "step": 4278 }, { "epoch": 10.6975, "grad_norm": 6.1875, "learning_rate": 1.9690845799712257e-05, "loss": 1.5271, "step": 4279 }, { "epoch": 10.7, "grad_norm": 4.96875, "learning_rate": 1.9690645994432307e-05, "loss": 1.5466, "step": 4280 }, { "epoch": 10.7025, "grad_norm": 6.90625, "learning_rate": 1.9690446125620867e-05, "loss": 1.6393, "step": 4281 }, { "epoch": 10.705, "grad_norm": 8.25, "learning_rate": 1.9690246193279253e-05, "loss": 1.5988, "step": 4282 }, { "epoch": 10.7075, "grad_norm": 6.75, "learning_rate": 1.969004619740877e-05, "loss": 1.3828, "step": 4283 }, { "epoch": 10.71, "grad_norm": 5.6875, "learning_rate": 1.9689846138010736e-05, "loss": 1.5312, "step": 4284 }, { "epoch": 10.7125, "grad_norm": 8.6875, "learning_rate": 1.9689646015086454e-05, "loss": 1.3977, "step": 4285 }, { "epoch": 10.715, "grad_norm": 4.75, "learning_rate": 1.9689445828637243e-05, "loss": 1.6186, "step": 4286 }, { "epoch": 10.7175, "grad_norm": 7.28125, "learning_rate": 1.968924557866441e-05, "loss": 1.6663, "step": 4287 }, { "epoch": 10.72, "grad_norm": 6.9375, "learning_rate": 1.9689045265169272e-05, "loss": 1.5785, "step": 4288 }, { "epoch": 10.7225, "grad_norm": 7.90625, "learning_rate": 1.9688844888153145e-05, "loss": 1.5336, "step": 4289 }, { "epoch": 10.725, "grad_norm": 4.71875, "learning_rate": 1.9688644447617336e-05, "loss": 1.6213, "step": 4290 }, { "epoch": 10.7275, "grad_norm": 5.5625, "learning_rate": 1.968844394356316e-05, "loss": 1.5129, "step": 4291 }, { "epoch": 10.73, "grad_norm": 6.9375, "learning_rate": 1.9688243375991935e-05, "loss": 1.5006, "step": 4292 }, { "epoch": 10.7325, "grad_norm": 5.125, "learning_rate": 1.9688042744904975e-05, "loss": 1.5511, "step": 4293 }, { "epoch": 10.735, "grad_norm": 4.71875, "learning_rate": 1.9687842050303592e-05, "loss": 1.429, "step": 4294 }, { "epoch": 10.7375, "grad_norm": 5.4375, "learning_rate": 1.9687641292189107e-05, "loss": 1.5112, "step": 4295 }, { "epoch": 10.74, "grad_norm": 4.875, "learning_rate": 1.968744047056283e-05, "loss": 1.461, "step": 4296 }, { "epoch": 10.7425, "grad_norm": 8.9375, "learning_rate": 1.9687239585426083e-05, "loss": 1.688, "step": 4297 }, { "epoch": 10.745, "grad_norm": 5.71875, "learning_rate": 1.9687038636780176e-05, "loss": 1.7419, "step": 4298 }, { "epoch": 10.7475, "grad_norm": 7.46875, "learning_rate": 1.968683762462644e-05, "loss": 1.6209, "step": 4299 }, { "epoch": 10.75, "grad_norm": 6.15625, "learning_rate": 1.9686636548966177e-05, "loss": 1.6015, "step": 4300 }, { "epoch": 10.7525, "grad_norm": 5.53125, "learning_rate": 1.9686435409800717e-05, "loss": 1.4167, "step": 4301 }, { "epoch": 10.755, "grad_norm": 6.5, "learning_rate": 1.9686234207131375e-05, "loss": 1.5041, "step": 4302 }, { "epoch": 10.7575, "grad_norm": 6.15625, "learning_rate": 1.9686032940959467e-05, "loss": 1.599, "step": 4303 }, { "epoch": 10.76, "grad_norm": 7.375, "learning_rate": 1.9685831611286312e-05, "loss": 1.5831, "step": 4304 }, { "epoch": 10.7625, "grad_norm": 10.25, "learning_rate": 1.9685630218113236e-05, "loss": 1.5526, "step": 4305 }, { "epoch": 10.765, "grad_norm": 4.84375, "learning_rate": 1.9685428761441555e-05, "loss": 1.5211, "step": 4306 }, { "epoch": 10.7675, "grad_norm": 6.09375, "learning_rate": 1.9685227241272588e-05, "loss": 1.5178, "step": 4307 }, { "epoch": 10.77, "grad_norm": 6.25, "learning_rate": 1.9685025657607663e-05, "loss": 1.6037, "step": 4308 }, { "epoch": 10.7725, "grad_norm": 6.75, "learning_rate": 1.9684824010448094e-05, "loss": 1.7036, "step": 4309 }, { "epoch": 10.775, "grad_norm": 4.84375, "learning_rate": 1.968462229979521e-05, "loss": 1.6404, "step": 4310 }, { "epoch": 10.7775, "grad_norm": 6.78125, "learning_rate": 1.9684420525650326e-05, "loss": 1.4726, "step": 4311 }, { "epoch": 10.78, "grad_norm": 12.25, "learning_rate": 1.9684218688014773e-05, "loss": 1.6719, "step": 4312 }, { "epoch": 10.7825, "grad_norm": 4.625, "learning_rate": 1.9684016786889864e-05, "loss": 1.5442, "step": 4313 }, { "epoch": 10.785, "grad_norm": 6.6875, "learning_rate": 1.9683814822276933e-05, "loss": 1.4468, "step": 4314 }, { "epoch": 10.7875, "grad_norm": 6.625, "learning_rate": 1.9683612794177297e-05, "loss": 1.5159, "step": 4315 }, { "epoch": 10.79, "grad_norm": 10.9375, "learning_rate": 1.9683410702592284e-05, "loss": 1.483, "step": 4316 }, { "epoch": 10.7925, "grad_norm": 5.75, "learning_rate": 1.9683208547523222e-05, "loss": 1.4718, "step": 4317 }, { "epoch": 10.795, "grad_norm": 7.96875, "learning_rate": 1.9683006328971426e-05, "loss": 1.4447, "step": 4318 }, { "epoch": 10.7975, "grad_norm": 5.4375, "learning_rate": 1.968280404693823e-05, "loss": 1.5121, "step": 4319 }, { "epoch": 10.8, "grad_norm": 4.3125, "learning_rate": 1.9682601701424958e-05, "loss": 1.4854, "step": 4320 }, { "epoch": 10.8025, "grad_norm": 6.96875, "learning_rate": 1.9682399292432938e-05, "loss": 1.6734, "step": 4321 }, { "epoch": 10.805, "grad_norm": 5.21875, "learning_rate": 1.9682196819963497e-05, "loss": 1.378, "step": 4322 }, { "epoch": 10.8075, "grad_norm": 4.96875, "learning_rate": 1.968199428401796e-05, "loss": 1.4283, "step": 4323 }, { "epoch": 10.81, "grad_norm": 6.875, "learning_rate": 1.9681791684597653e-05, "loss": 1.6558, "step": 4324 }, { "epoch": 10.8125, "grad_norm": 6.84375, "learning_rate": 1.968158902170391e-05, "loss": 1.6461, "step": 4325 }, { "epoch": 10.815, "grad_norm": 5.8125, "learning_rate": 1.9681386295338053e-05, "loss": 1.4537, "step": 4326 }, { "epoch": 10.8175, "grad_norm": 7.03125, "learning_rate": 1.968118350550142e-05, "loss": 1.5887, "step": 4327 }, { "epoch": 10.82, "grad_norm": 9.5625, "learning_rate": 1.9680980652195334e-05, "loss": 1.6708, "step": 4328 }, { "epoch": 10.8225, "grad_norm": 8.625, "learning_rate": 1.9680777735421124e-05, "loss": 1.5271, "step": 4329 }, { "epoch": 10.825, "grad_norm": 6.34375, "learning_rate": 1.9680574755180127e-05, "loss": 1.5296, "step": 4330 }, { "epoch": 10.8275, "grad_norm": 7.96875, "learning_rate": 1.9680371711473665e-05, "loss": 1.6403, "step": 4331 }, { "epoch": 10.83, "grad_norm": 8.0625, "learning_rate": 1.9680168604303074e-05, "loss": 1.5411, "step": 4332 }, { "epoch": 10.8325, "grad_norm": 8.875, "learning_rate": 1.9679965433669687e-05, "loss": 1.6419, "step": 4333 }, { "epoch": 10.835, "grad_norm": 5.6875, "learning_rate": 1.9679762199574834e-05, "loss": 1.5644, "step": 4334 }, { "epoch": 10.8375, "grad_norm": 8.4375, "learning_rate": 1.9679558902019845e-05, "loss": 1.5455, "step": 4335 }, { "epoch": 10.84, "grad_norm": 6.9375, "learning_rate": 1.9679355541006056e-05, "loss": 1.4899, "step": 4336 }, { "epoch": 10.8425, "grad_norm": 6.71875, "learning_rate": 1.9679152116534798e-05, "loss": 1.5712, "step": 4337 }, { "epoch": 10.845, "grad_norm": 6.8125, "learning_rate": 1.967894862860741e-05, "loss": 1.4201, "step": 4338 }, { "epoch": 10.8475, "grad_norm": 8.9375, "learning_rate": 1.967874507722522e-05, "loss": 1.4426, "step": 4339 }, { "epoch": 10.85, "grad_norm": 7.625, "learning_rate": 1.9678541462389564e-05, "loss": 1.4655, "step": 4340 }, { "epoch": 10.8525, "grad_norm": 4.71875, "learning_rate": 1.9678337784101776e-05, "loss": 1.525, "step": 4341 }, { "epoch": 10.855, "grad_norm": 4.3125, "learning_rate": 1.9678134042363198e-05, "loss": 1.5762, "step": 4342 }, { "epoch": 10.8575, "grad_norm": 6.34375, "learning_rate": 1.9677930237175157e-05, "loss": 1.5256, "step": 4343 }, { "epoch": 10.86, "grad_norm": 7.625, "learning_rate": 1.967772636853899e-05, "loss": 1.5029, "step": 4344 }, { "epoch": 10.8625, "grad_norm": 8.25, "learning_rate": 1.967752243645604e-05, "loss": 1.4592, "step": 4345 }, { "epoch": 10.865, "grad_norm": 6.78125, "learning_rate": 1.967731844092764e-05, "loss": 1.4674, "step": 4346 }, { "epoch": 10.8675, "grad_norm": 5.09375, "learning_rate": 1.967711438195513e-05, "loss": 1.5812, "step": 4347 }, { "epoch": 10.87, "grad_norm": 6.09375, "learning_rate": 1.967691025953984e-05, "loss": 1.472, "step": 4348 }, { "epoch": 10.8725, "grad_norm": 6.46875, "learning_rate": 1.9676706073683115e-05, "loss": 1.569, "step": 4349 }, { "epoch": 10.875, "grad_norm": 6.09375, "learning_rate": 1.9676501824386295e-05, "loss": 1.5743, "step": 4350 }, { "epoch": 10.8775, "grad_norm": 7.84375, "learning_rate": 1.967629751165071e-05, "loss": 1.556, "step": 4351 }, { "epoch": 10.88, "grad_norm": 6.25, "learning_rate": 1.9676093135477713e-05, "loss": 1.5614, "step": 4352 }, { "epoch": 10.8825, "grad_norm": 5.3125, "learning_rate": 1.967588869586863e-05, "loss": 1.52, "step": 4353 }, { "epoch": 10.885, "grad_norm": 8.5, "learning_rate": 1.9675684192824816e-05, "loss": 1.4091, "step": 4354 }, { "epoch": 10.8875, "grad_norm": 11.375, "learning_rate": 1.9675479626347597e-05, "loss": 1.5619, "step": 4355 }, { "epoch": 10.89, "grad_norm": 6.96875, "learning_rate": 1.9675274996438324e-05, "loss": 1.5491, "step": 4356 }, { "epoch": 10.8925, "grad_norm": 7.0625, "learning_rate": 1.9675070303098335e-05, "loss": 1.5505, "step": 4357 }, { "epoch": 10.895, "grad_norm": 7.15625, "learning_rate": 1.967486554632897e-05, "loss": 1.4521, "step": 4358 }, { "epoch": 10.8975, "grad_norm": 6.875, "learning_rate": 1.9674660726131578e-05, "loss": 1.445, "step": 4359 }, { "epoch": 10.9, "grad_norm": 8.625, "learning_rate": 1.9674455842507494e-05, "loss": 1.4245, "step": 4360 }, { "epoch": 10.9025, "grad_norm": 6.96875, "learning_rate": 1.9674250895458064e-05, "loss": 1.5501, "step": 4361 }, { "epoch": 10.905, "grad_norm": 4.90625, "learning_rate": 1.9674045884984636e-05, "loss": 1.3293, "step": 4362 }, { "epoch": 10.9075, "grad_norm": 6.78125, "learning_rate": 1.9673840811088546e-05, "loss": 1.5082, "step": 4363 }, { "epoch": 10.91, "grad_norm": 8.5625, "learning_rate": 1.9673635673771145e-05, "loss": 1.5127, "step": 4364 }, { "epoch": 10.9125, "grad_norm": 5.75, "learning_rate": 1.9673430473033775e-05, "loss": 1.4599, "step": 4365 }, { "epoch": 10.915, "grad_norm": 7.09375, "learning_rate": 1.967322520887778e-05, "loss": 1.5295, "step": 4366 }, { "epoch": 10.9175, "grad_norm": 8.25, "learning_rate": 1.9673019881304515e-05, "loss": 1.5813, "step": 4367 }, { "epoch": 10.92, "grad_norm": 5.8125, "learning_rate": 1.9672814490315312e-05, "loss": 1.5381, "step": 4368 }, { "epoch": 10.9225, "grad_norm": 6.4375, "learning_rate": 1.9672609035911527e-05, "loss": 1.5029, "step": 4369 }, { "epoch": 10.925, "grad_norm": 6.09375, "learning_rate": 1.9672403518094502e-05, "loss": 1.6913, "step": 4370 }, { "epoch": 10.9275, "grad_norm": 8.5, "learning_rate": 1.967219793686559e-05, "loss": 1.5456, "step": 4371 }, { "epoch": 10.93, "grad_norm": 4.96875, "learning_rate": 1.967199229222613e-05, "loss": 1.496, "step": 4372 }, { "epoch": 10.9325, "grad_norm": 7.9375, "learning_rate": 1.967178658417748e-05, "loss": 1.4962, "step": 4373 }, { "epoch": 10.935, "grad_norm": 12.5625, "learning_rate": 1.9671580812720983e-05, "loss": 1.5239, "step": 4374 }, { "epoch": 10.9375, "grad_norm": 7.1875, "learning_rate": 1.9671374977857987e-05, "loss": 1.5561, "step": 4375 }, { "epoch": 10.94, "grad_norm": 3.953125, "learning_rate": 1.967116907958985e-05, "loss": 1.4718, "step": 4376 }, { "epoch": 10.9425, "grad_norm": 5.0, "learning_rate": 1.967096311791791e-05, "loss": 1.3499, "step": 4377 }, { "epoch": 10.945, "grad_norm": 7.78125, "learning_rate": 1.9670757092843522e-05, "loss": 1.6355, "step": 4378 }, { "epoch": 10.9475, "grad_norm": 7.0625, "learning_rate": 1.967055100436804e-05, "loss": 1.5698, "step": 4379 }, { "epoch": 10.95, "grad_norm": 9.0, "learning_rate": 1.9670344852492814e-05, "loss": 1.5386, "step": 4380 }, { "epoch": 10.9525, "grad_norm": 14.5, "learning_rate": 1.967013863721919e-05, "loss": 1.3996, "step": 4381 }, { "epoch": 10.955, "grad_norm": 5.0, "learning_rate": 1.966993235854853e-05, "loss": 1.6212, "step": 4382 }, { "epoch": 10.9575, "grad_norm": 7.1875, "learning_rate": 1.9669726016482174e-05, "loss": 1.4449, "step": 4383 }, { "epoch": 10.96, "grad_norm": 7.90625, "learning_rate": 1.9669519611021485e-05, "loss": 1.5583, "step": 4384 }, { "epoch": 10.9625, "grad_norm": 5.53125, "learning_rate": 1.9669313142167813e-05, "loss": 1.5542, "step": 4385 }, { "epoch": 10.965, "grad_norm": 7.625, "learning_rate": 1.9669106609922507e-05, "loss": 1.5211, "step": 4386 }, { "epoch": 10.9675, "grad_norm": 8.625, "learning_rate": 1.966890001428693e-05, "loss": 1.5363, "step": 4387 }, { "epoch": 10.97, "grad_norm": 8.875, "learning_rate": 1.9668693355262428e-05, "loss": 1.6632, "step": 4388 }, { "epoch": 10.9725, "grad_norm": 7.9375, "learning_rate": 1.966848663285036e-05, "loss": 1.3382, "step": 4389 }, { "epoch": 10.975, "grad_norm": 6.0625, "learning_rate": 1.9668279847052082e-05, "loss": 1.6207, "step": 4390 }, { "epoch": 10.9775, "grad_norm": 6.125, "learning_rate": 1.9668072997868945e-05, "loss": 1.5093, "step": 4391 }, { "epoch": 10.98, "grad_norm": 8.25, "learning_rate": 1.9667866085302312e-05, "loss": 1.5972, "step": 4392 }, { "epoch": 10.9825, "grad_norm": 4.1875, "learning_rate": 1.966765910935353e-05, "loss": 1.5716, "step": 4393 }, { "epoch": 10.985, "grad_norm": 10.625, "learning_rate": 1.966745207002397e-05, "loss": 1.5288, "step": 4394 }, { "epoch": 10.9875, "grad_norm": 5.6875, "learning_rate": 1.9667244967314977e-05, "loss": 1.5876, "step": 4395 }, { "epoch": 10.99, "grad_norm": 9.4375, "learning_rate": 1.9667037801227914e-05, "loss": 1.5055, "step": 4396 }, { "epoch": 10.9925, "grad_norm": 6.1875, "learning_rate": 1.9666830571764134e-05, "loss": 1.6396, "step": 4397 }, { "epoch": 10.995, "grad_norm": 7.3125, "learning_rate": 1.9666623278925004e-05, "loss": 1.3945, "step": 4398 }, { "epoch": 10.9975, "grad_norm": 5.96875, "learning_rate": 1.9666415922711878e-05, "loss": 1.5028, "step": 4399 }, { "epoch": 11.0, "grad_norm": 6.34375, "learning_rate": 1.9666208503126115e-05, "loss": 1.4288, "step": 4400 }, { "epoch": 11.0025, "grad_norm": 8.5, "learning_rate": 1.9666001020169075e-05, "loss": 1.6593, "step": 4401 }, { "epoch": 11.005, "grad_norm": 5.28125, "learning_rate": 1.966579347384212e-05, "loss": 1.4602, "step": 4402 }, { "epoch": 11.0075, "grad_norm": 7.40625, "learning_rate": 1.966558586414661e-05, "loss": 1.4305, "step": 4403 }, { "epoch": 11.01, "grad_norm": 6.6875, "learning_rate": 1.9665378191083907e-05, "loss": 1.456, "step": 4404 }, { "epoch": 11.0125, "grad_norm": 6.5625, "learning_rate": 1.966517045465537e-05, "loss": 1.5448, "step": 4405 }, { "epoch": 11.015, "grad_norm": 6.3125, "learning_rate": 1.966496265486236e-05, "loss": 1.5432, "step": 4406 }, { "epoch": 11.0175, "grad_norm": 11.375, "learning_rate": 1.9664754791706245e-05, "loss": 1.4961, "step": 4407 }, { "epoch": 11.02, "grad_norm": 4.59375, "learning_rate": 1.9664546865188384e-05, "loss": 1.2276, "step": 4408 }, { "epoch": 11.0225, "grad_norm": 5.4375, "learning_rate": 1.966433887531014e-05, "loss": 1.3929, "step": 4409 }, { "epoch": 11.025, "grad_norm": 5.65625, "learning_rate": 1.9664130822072874e-05, "loss": 1.5318, "step": 4410 }, { "epoch": 11.0275, "grad_norm": 7.34375, "learning_rate": 1.9663922705477957e-05, "loss": 1.6076, "step": 4411 }, { "epoch": 11.03, "grad_norm": 5.8125, "learning_rate": 1.966371452552675e-05, "loss": 1.4652, "step": 4412 }, { "epoch": 11.0325, "grad_norm": 6.0, "learning_rate": 1.9663506282220616e-05, "loss": 1.4401, "step": 4413 }, { "epoch": 11.035, "grad_norm": 8.625, "learning_rate": 1.966329797556092e-05, "loss": 1.5762, "step": 4414 }, { "epoch": 11.0375, "grad_norm": 5.28125, "learning_rate": 1.966308960554903e-05, "loss": 1.5692, "step": 4415 }, { "epoch": 11.04, "grad_norm": 5.0, "learning_rate": 1.9662881172186313e-05, "loss": 1.5377, "step": 4416 }, { "epoch": 11.0425, "grad_norm": 7.53125, "learning_rate": 1.9662672675474136e-05, "loss": 1.4071, "step": 4417 }, { "epoch": 11.045, "grad_norm": 5.125, "learning_rate": 1.966246411541386e-05, "loss": 1.5311, "step": 4418 }, { "epoch": 11.0475, "grad_norm": 4.9375, "learning_rate": 1.9662255492006853e-05, "loss": 1.4999, "step": 4419 }, { "epoch": 11.05, "grad_norm": 6.15625, "learning_rate": 1.966204680525449e-05, "loss": 1.477, "step": 4420 }, { "epoch": 11.0525, "grad_norm": 5.34375, "learning_rate": 1.9661838055158134e-05, "loss": 1.5134, "step": 4421 }, { "epoch": 11.055, "grad_norm": 6.78125, "learning_rate": 1.9661629241719155e-05, "loss": 1.401, "step": 4422 }, { "epoch": 11.0575, "grad_norm": 8.875, "learning_rate": 1.9661420364938917e-05, "loss": 1.4455, "step": 4423 }, { "epoch": 11.06, "grad_norm": 9.9375, "learning_rate": 1.9661211424818798e-05, "loss": 1.5745, "step": 4424 }, { "epoch": 11.0625, "grad_norm": 6.6875, "learning_rate": 1.9661002421360164e-05, "loss": 1.6756, "step": 4425 }, { "epoch": 11.065, "grad_norm": 6.8125, "learning_rate": 1.966079335456438e-05, "loss": 1.5383, "step": 4426 }, { "epoch": 11.0675, "grad_norm": 7.15625, "learning_rate": 1.9660584224432827e-05, "loss": 1.4522, "step": 4427 }, { "epoch": 11.07, "grad_norm": 6.6875, "learning_rate": 1.9660375030966868e-05, "loss": 1.6101, "step": 4428 }, { "epoch": 11.0725, "grad_norm": 8.3125, "learning_rate": 1.9660165774167876e-05, "loss": 1.6206, "step": 4429 }, { "epoch": 11.075, "grad_norm": 7.03125, "learning_rate": 1.9659956454037223e-05, "loss": 1.5486, "step": 4430 }, { "epoch": 11.0775, "grad_norm": 7.34375, "learning_rate": 1.9659747070576285e-05, "loss": 1.4982, "step": 4431 }, { "epoch": 11.08, "grad_norm": 6.8125, "learning_rate": 1.9659537623786428e-05, "loss": 1.3723, "step": 4432 }, { "epoch": 11.0825, "grad_norm": 6.5625, "learning_rate": 1.9659328113669033e-05, "loss": 1.5761, "step": 4433 }, { "epoch": 11.085, "grad_norm": 10.6875, "learning_rate": 1.965911854022547e-05, "loss": 1.692, "step": 4434 }, { "epoch": 11.0875, "grad_norm": 6.125, "learning_rate": 1.965890890345711e-05, "loss": 1.618, "step": 4435 }, { "epoch": 11.09, "grad_norm": 6.59375, "learning_rate": 1.965869920336533e-05, "loss": 1.5995, "step": 4436 }, { "epoch": 11.0925, "grad_norm": 6.0, "learning_rate": 1.9658489439951503e-05, "loss": 1.4309, "step": 4437 }, { "epoch": 11.095, "grad_norm": 6.59375, "learning_rate": 1.9658279613217007e-05, "loss": 1.4997, "step": 4438 }, { "epoch": 11.0975, "grad_norm": 6.46875, "learning_rate": 1.9658069723163214e-05, "loss": 1.5807, "step": 4439 }, { "epoch": 11.1, "grad_norm": 6.3125, "learning_rate": 1.9657859769791506e-05, "loss": 1.3283, "step": 4440 }, { "epoch": 11.1025, "grad_norm": 5.03125, "learning_rate": 1.9657649753103255e-05, "loss": 1.5353, "step": 4441 }, { "epoch": 11.105, "grad_norm": 6.03125, "learning_rate": 1.965743967309984e-05, "loss": 1.6463, "step": 4442 }, { "epoch": 11.1075, "grad_norm": 5.65625, "learning_rate": 1.9657229529782637e-05, "loss": 1.3452, "step": 4443 }, { "epoch": 11.11, "grad_norm": 6.78125, "learning_rate": 1.965701932315302e-05, "loss": 1.4966, "step": 4444 }, { "epoch": 11.1125, "grad_norm": 11.125, "learning_rate": 1.965680905321237e-05, "loss": 1.5378, "step": 4445 }, { "epoch": 11.115, "grad_norm": 9.75, "learning_rate": 1.965659871996207e-05, "loss": 1.4484, "step": 4446 }, { "epoch": 11.1175, "grad_norm": 7.0625, "learning_rate": 1.9656388323403496e-05, "loss": 1.5921, "step": 4447 }, { "epoch": 11.12, "grad_norm": 8.8125, "learning_rate": 1.9656177863538025e-05, "loss": 1.5345, "step": 4448 }, { "epoch": 11.1225, "grad_norm": 7.09375, "learning_rate": 1.9655967340367036e-05, "loss": 1.4217, "step": 4449 }, { "epoch": 11.125, "grad_norm": 9.1875, "learning_rate": 1.9655756753891916e-05, "loss": 1.3749, "step": 4450 }, { "epoch": 11.1275, "grad_norm": 6.0, "learning_rate": 1.9655546104114038e-05, "loss": 1.5079, "step": 4451 }, { "epoch": 11.13, "grad_norm": 5.1875, "learning_rate": 1.9655335391034787e-05, "loss": 1.5187, "step": 4452 }, { "epoch": 11.1325, "grad_norm": 7.25, "learning_rate": 1.9655124614655543e-05, "loss": 1.4469, "step": 4453 }, { "epoch": 11.135, "grad_norm": 8.1875, "learning_rate": 1.965491377497769e-05, "loss": 1.5696, "step": 4454 }, { "epoch": 11.1375, "grad_norm": 6.625, "learning_rate": 1.965470287200261e-05, "loss": 1.4569, "step": 4455 }, { "epoch": 11.14, "grad_norm": 11.0, "learning_rate": 1.965449190573168e-05, "loss": 1.4541, "step": 4456 }, { "epoch": 11.1425, "grad_norm": 7.15625, "learning_rate": 1.965428087616629e-05, "loss": 1.5872, "step": 4457 }, { "epoch": 11.145, "grad_norm": 6.8125, "learning_rate": 1.965406978330782e-05, "loss": 1.5504, "step": 4458 }, { "epoch": 11.1475, "grad_norm": 7.03125, "learning_rate": 1.9653858627157655e-05, "loss": 1.6779, "step": 4459 }, { "epoch": 11.15, "grad_norm": 5.5625, "learning_rate": 1.965364740771718e-05, "loss": 1.4877, "step": 4460 }, { "epoch": 11.1525, "grad_norm": 5.125, "learning_rate": 1.9653436124987778e-05, "loss": 1.4672, "step": 4461 }, { "epoch": 11.155, "grad_norm": 9.9375, "learning_rate": 1.9653224778970834e-05, "loss": 1.746, "step": 4462 }, { "epoch": 11.1575, "grad_norm": 5.8125, "learning_rate": 1.965301336966774e-05, "loss": 1.7217, "step": 4463 }, { "epoch": 11.16, "grad_norm": 5.125, "learning_rate": 1.965280189707987e-05, "loss": 1.4062, "step": 4464 }, { "epoch": 11.1625, "grad_norm": 6.5625, "learning_rate": 1.965259036120862e-05, "loss": 1.5583, "step": 4465 }, { "epoch": 11.165, "grad_norm": 7.46875, "learning_rate": 1.965237876205537e-05, "loss": 1.5924, "step": 4466 }, { "epoch": 11.1675, "grad_norm": 6.28125, "learning_rate": 1.9652167099621518e-05, "loss": 1.7072, "step": 4467 }, { "epoch": 11.17, "grad_norm": 8.6875, "learning_rate": 1.965195537390844e-05, "loss": 1.5862, "step": 4468 }, { "epoch": 11.1725, "grad_norm": 7.8125, "learning_rate": 1.9651743584917532e-05, "loss": 1.5441, "step": 4469 }, { "epoch": 11.175, "grad_norm": 6.875, "learning_rate": 1.9651531732650175e-05, "loss": 1.6206, "step": 4470 }, { "epoch": 11.1775, "grad_norm": 6.53125, "learning_rate": 1.9651319817107766e-05, "loss": 1.3927, "step": 4471 }, { "epoch": 11.18, "grad_norm": 7.21875, "learning_rate": 1.965110783829169e-05, "loss": 1.3786, "step": 4472 }, { "epoch": 11.1825, "grad_norm": 6.09375, "learning_rate": 1.9650895796203335e-05, "loss": 1.6221, "step": 4473 }, { "epoch": 11.185, "grad_norm": 6.09375, "learning_rate": 1.9650683690844095e-05, "loss": 1.4199, "step": 4474 }, { "epoch": 11.1875, "grad_norm": 7.96875, "learning_rate": 1.965047152221536e-05, "loss": 1.5293, "step": 4475 }, { "epoch": 11.19, "grad_norm": 7.9375, "learning_rate": 1.965025929031852e-05, "loss": 1.6361, "step": 4476 }, { "epoch": 11.1925, "grad_norm": 6.59375, "learning_rate": 1.9650046995154963e-05, "loss": 1.4405, "step": 4477 }, { "epoch": 11.195, "grad_norm": 7.3125, "learning_rate": 1.9649834636726087e-05, "loss": 1.3631, "step": 4478 }, { "epoch": 11.1975, "grad_norm": 8.375, "learning_rate": 1.964962221503328e-05, "loss": 1.7498, "step": 4479 }, { "epoch": 11.2, "grad_norm": 6.34375, "learning_rate": 1.9649409730077934e-05, "loss": 1.5331, "step": 4480 }, { "epoch": 11.2025, "grad_norm": 6.75, "learning_rate": 1.9649197181861447e-05, "loss": 1.4796, "step": 4481 }, { "epoch": 11.205, "grad_norm": 5.375, "learning_rate": 1.9648984570385208e-05, "loss": 1.5385, "step": 4482 }, { "epoch": 11.2075, "grad_norm": 7.53125, "learning_rate": 1.9648771895650614e-05, "loss": 1.5241, "step": 4483 }, { "epoch": 11.21, "grad_norm": 11.1875, "learning_rate": 1.9648559157659054e-05, "loss": 1.566, "step": 4484 }, { "epoch": 11.2125, "grad_norm": 6.4375, "learning_rate": 1.964834635641193e-05, "loss": 1.7195, "step": 4485 }, { "epoch": 11.215, "grad_norm": 8.5, "learning_rate": 1.9648133491910633e-05, "loss": 1.3537, "step": 4486 }, { "epoch": 11.2175, "grad_norm": 4.78125, "learning_rate": 1.9647920564156557e-05, "loss": 1.5003, "step": 4487 }, { "epoch": 11.22, "grad_norm": 6.75, "learning_rate": 1.96477075731511e-05, "loss": 1.555, "step": 4488 }, { "epoch": 11.2225, "grad_norm": 6.1875, "learning_rate": 1.964749451889566e-05, "loss": 1.5442, "step": 4489 }, { "epoch": 11.225, "grad_norm": 5.84375, "learning_rate": 1.9647281401391633e-05, "loss": 1.5506, "step": 4490 }, { "epoch": 11.2275, "grad_norm": 8.125, "learning_rate": 1.964706822064041e-05, "loss": 1.6529, "step": 4491 }, { "epoch": 11.23, "grad_norm": 5.9375, "learning_rate": 1.96468549766434e-05, "loss": 1.4807, "step": 4492 }, { "epoch": 11.2325, "grad_norm": 7.65625, "learning_rate": 1.964664166940199e-05, "loss": 1.4934, "step": 4493 }, { "epoch": 11.235, "grad_norm": 5.21875, "learning_rate": 1.9646428298917587e-05, "loss": 1.5107, "step": 4494 }, { "epoch": 11.2375, "grad_norm": 7.90625, "learning_rate": 1.9646214865191585e-05, "loss": 1.6816, "step": 4495 }, { "epoch": 11.24, "grad_norm": 5.375, "learning_rate": 1.9646001368225382e-05, "loss": 1.5366, "step": 4496 }, { "epoch": 11.2425, "grad_norm": 9.5, "learning_rate": 1.9645787808020384e-05, "loss": 1.4322, "step": 4497 }, { "epoch": 11.245, "grad_norm": 6.59375, "learning_rate": 1.9645574184577982e-05, "loss": 1.4125, "step": 4498 }, { "epoch": 11.2475, "grad_norm": 8.5, "learning_rate": 1.9645360497899584e-05, "loss": 1.3044, "step": 4499 }, { "epoch": 11.25, "grad_norm": 5.8125, "learning_rate": 1.964514674798659e-05, "loss": 1.4774, "step": 4500 }, { "epoch": 11.2525, "grad_norm": 9.0, "learning_rate": 1.96449329348404e-05, "loss": 1.5635, "step": 4501 }, { "epoch": 11.255, "grad_norm": 10.0625, "learning_rate": 1.9644719058462414e-05, "loss": 1.5353, "step": 4502 }, { "epoch": 11.2575, "grad_norm": 5.15625, "learning_rate": 1.9644505118854035e-05, "loss": 1.3909, "step": 4503 }, { "epoch": 11.26, "grad_norm": 10.0625, "learning_rate": 1.9644291116016666e-05, "loss": 1.5476, "step": 4504 }, { "epoch": 11.2625, "grad_norm": 5.9375, "learning_rate": 1.9644077049951712e-05, "loss": 1.5002, "step": 4505 }, { "epoch": 11.265, "grad_norm": 6.3125, "learning_rate": 1.9643862920660578e-05, "loss": 1.6387, "step": 4506 }, { "epoch": 11.2675, "grad_norm": 6.96875, "learning_rate": 1.964364872814466e-05, "loss": 1.5367, "step": 4507 }, { "epoch": 11.27, "grad_norm": 8.1875, "learning_rate": 1.964343447240537e-05, "loss": 1.6195, "step": 4508 }, { "epoch": 11.2725, "grad_norm": 6.3125, "learning_rate": 1.9643220153444108e-05, "loss": 1.4621, "step": 4509 }, { "epoch": 11.275, "grad_norm": 7.0625, "learning_rate": 1.964300577126228e-05, "loss": 1.5907, "step": 4510 }, { "epoch": 11.2775, "grad_norm": 6.75, "learning_rate": 1.9642791325861293e-05, "loss": 1.654, "step": 4511 }, { "epoch": 11.28, "grad_norm": 6.4375, "learning_rate": 1.9642576817242553e-05, "loss": 1.6111, "step": 4512 }, { "epoch": 11.2825, "grad_norm": 6.09375, "learning_rate": 1.9642362245407464e-05, "loss": 1.507, "step": 4513 }, { "epoch": 11.285, "grad_norm": 5.1875, "learning_rate": 1.9642147610357432e-05, "loss": 1.52, "step": 4514 }, { "epoch": 11.2875, "grad_norm": 7.84375, "learning_rate": 1.964193291209387e-05, "loss": 1.52, "step": 4515 }, { "epoch": 11.29, "grad_norm": 6.03125, "learning_rate": 1.9641718150618177e-05, "loss": 1.438, "step": 4516 }, { "epoch": 11.2925, "grad_norm": 10.5, "learning_rate": 1.964150332593177e-05, "loss": 1.5462, "step": 4517 }, { "epoch": 11.295, "grad_norm": 4.0625, "learning_rate": 1.964128843803605e-05, "loss": 1.5531, "step": 4518 }, { "epoch": 11.2975, "grad_norm": 7.03125, "learning_rate": 1.9641073486932428e-05, "loss": 1.5667, "step": 4519 }, { "epoch": 11.3, "grad_norm": 6.40625, "learning_rate": 1.9640858472622316e-05, "loss": 1.4348, "step": 4520 }, { "epoch": 11.3025, "grad_norm": 7.25, "learning_rate": 1.964064339510712e-05, "loss": 1.5227, "step": 4521 }, { "epoch": 11.305, "grad_norm": 4.75, "learning_rate": 1.9640428254388256e-05, "loss": 1.404, "step": 4522 }, { "epoch": 11.3075, "grad_norm": 5.6875, "learning_rate": 1.964021305046713e-05, "loss": 1.3687, "step": 4523 }, { "epoch": 11.31, "grad_norm": 7.53125, "learning_rate": 1.963999778334515e-05, "loss": 1.5898, "step": 4524 }, { "epoch": 11.3125, "grad_norm": 11.875, "learning_rate": 1.963978245302373e-05, "loss": 1.5461, "step": 4525 }, { "epoch": 11.315, "grad_norm": 5.90625, "learning_rate": 1.9639567059504282e-05, "loss": 1.4345, "step": 4526 }, { "epoch": 11.3175, "grad_norm": 7.78125, "learning_rate": 1.963935160278822e-05, "loss": 1.3084, "step": 4527 }, { "epoch": 11.32, "grad_norm": 7.8125, "learning_rate": 1.9639136082876954e-05, "loss": 1.6331, "step": 4528 }, { "epoch": 11.3225, "grad_norm": 8.875, "learning_rate": 1.9638920499771896e-05, "loss": 1.5818, "step": 4529 }, { "epoch": 11.325, "grad_norm": 6.15625, "learning_rate": 1.9638704853474463e-05, "loss": 1.3888, "step": 4530 }, { "epoch": 11.3275, "grad_norm": 7.90625, "learning_rate": 1.9638489143986065e-05, "loss": 1.3811, "step": 4531 }, { "epoch": 11.33, "grad_norm": 7.0, "learning_rate": 1.9638273371308118e-05, "loss": 1.4402, "step": 4532 }, { "epoch": 11.3325, "grad_norm": 10.375, "learning_rate": 1.963805753544204e-05, "loss": 1.4307, "step": 4533 }, { "epoch": 11.335, "grad_norm": 7.625, "learning_rate": 1.963784163638924e-05, "loss": 1.5584, "step": 4534 }, { "epoch": 11.3375, "grad_norm": 6.0625, "learning_rate": 1.9637625674151132e-05, "loss": 1.4732, "step": 4535 }, { "epoch": 11.34, "grad_norm": 4.5625, "learning_rate": 1.9637409648729142e-05, "loss": 1.502, "step": 4536 }, { "epoch": 11.3425, "grad_norm": 5.78125, "learning_rate": 1.9637193560124677e-05, "loss": 1.4522, "step": 4537 }, { "epoch": 11.345, "grad_norm": 6.21875, "learning_rate": 1.9636977408339157e-05, "loss": 1.6069, "step": 4538 }, { "epoch": 11.3475, "grad_norm": 5.8125, "learning_rate": 1.9636761193373998e-05, "loss": 1.7047, "step": 4539 }, { "epoch": 11.35, "grad_norm": 6.84375, "learning_rate": 1.963654491523062e-05, "loss": 1.547, "step": 4540 }, { "epoch": 11.3525, "grad_norm": 7.75, "learning_rate": 1.9636328573910436e-05, "loss": 1.4389, "step": 4541 }, { "epoch": 11.355, "grad_norm": 6.5625, "learning_rate": 1.963611216941487e-05, "loss": 1.5542, "step": 4542 }, { "epoch": 11.3575, "grad_norm": 5.53125, "learning_rate": 1.963589570174534e-05, "loss": 1.5352, "step": 4543 }, { "epoch": 11.36, "grad_norm": 5.75, "learning_rate": 1.9635679170903258e-05, "loss": 1.7052, "step": 4544 }, { "epoch": 11.3625, "grad_norm": 6.40625, "learning_rate": 1.9635462576890056e-05, "loss": 1.4794, "step": 4545 }, { "epoch": 11.365, "grad_norm": 6.6875, "learning_rate": 1.9635245919707142e-05, "loss": 1.4471, "step": 4546 }, { "epoch": 11.3675, "grad_norm": 9.5, "learning_rate": 1.963502919935594e-05, "loss": 1.5887, "step": 4547 }, { "epoch": 11.37, "grad_norm": 4.46875, "learning_rate": 1.9634812415837876e-05, "loss": 1.36, "step": 4548 }, { "epoch": 11.3725, "grad_norm": 5.15625, "learning_rate": 1.963459556915437e-05, "loss": 1.2855, "step": 4549 }, { "epoch": 11.375, "grad_norm": 4.9375, "learning_rate": 1.9634378659306834e-05, "loss": 1.5837, "step": 4550 }, { "epoch": 11.3775, "grad_norm": 5.71875, "learning_rate": 1.96341616862967e-05, "loss": 1.5001, "step": 4551 }, { "epoch": 11.38, "grad_norm": 5.78125, "learning_rate": 1.963394465012539e-05, "loss": 1.5494, "step": 4552 }, { "epoch": 11.3825, "grad_norm": 5.9375, "learning_rate": 1.9633727550794323e-05, "loss": 1.5295, "step": 4553 }, { "epoch": 11.385, "grad_norm": 4.375, "learning_rate": 1.963351038830492e-05, "loss": 1.1664, "step": 4554 }, { "epoch": 11.3875, "grad_norm": 5.21875, "learning_rate": 1.963329316265861e-05, "loss": 1.6006, "step": 4555 }, { "epoch": 11.39, "grad_norm": 4.875, "learning_rate": 1.963307587385682e-05, "loss": 1.5782, "step": 4556 }, { "epoch": 11.3925, "grad_norm": 5.65625, "learning_rate": 1.9632858521900965e-05, "loss": 1.5396, "step": 4557 }, { "epoch": 11.395, "grad_norm": 7.8125, "learning_rate": 1.9632641106792475e-05, "loss": 1.3853, "step": 4558 }, { "epoch": 11.3975, "grad_norm": 6.125, "learning_rate": 1.963242362853278e-05, "loss": 1.5068, "step": 4559 }, { "epoch": 11.4, "grad_norm": 7.65625, "learning_rate": 1.9632206087123296e-05, "loss": 1.5246, "step": 4560 }, { "epoch": 11.4025, "grad_norm": 7.4375, "learning_rate": 1.963198848256546e-05, "loss": 1.5317, "step": 4561 }, { "epoch": 11.405, "grad_norm": 8.0, "learning_rate": 1.9631770814860686e-05, "loss": 1.6364, "step": 4562 }, { "epoch": 11.4075, "grad_norm": 8.25, "learning_rate": 1.9631553084010413e-05, "loss": 1.4643, "step": 4563 }, { "epoch": 11.41, "grad_norm": 9.1875, "learning_rate": 1.9631335290016063e-05, "loss": 1.3982, "step": 4564 }, { "epoch": 11.4125, "grad_norm": 7.03125, "learning_rate": 1.963111743287906e-05, "loss": 1.5872, "step": 4565 }, { "epoch": 11.415, "grad_norm": 6.84375, "learning_rate": 1.9630899512600842e-05, "loss": 1.5796, "step": 4566 }, { "epoch": 11.4175, "grad_norm": 5.53125, "learning_rate": 1.963068152918283e-05, "loss": 1.562, "step": 4567 }, { "epoch": 11.42, "grad_norm": 26.5, "learning_rate": 1.9630463482626454e-05, "loss": 1.5141, "step": 4568 }, { "epoch": 11.4225, "grad_norm": 5.6875, "learning_rate": 1.9630245372933147e-05, "loss": 1.4742, "step": 4569 }, { "epoch": 11.425, "grad_norm": 5.84375, "learning_rate": 1.9630027200104335e-05, "loss": 1.3639, "step": 4570 }, { "epoch": 11.4275, "grad_norm": 7.25, "learning_rate": 1.9629808964141454e-05, "loss": 1.4614, "step": 4571 }, { "epoch": 11.43, "grad_norm": 6.0625, "learning_rate": 1.9629590665045925e-05, "loss": 1.4389, "step": 4572 }, { "epoch": 11.4325, "grad_norm": 6.40625, "learning_rate": 1.9629372302819187e-05, "loss": 1.5528, "step": 4573 }, { "epoch": 11.435, "grad_norm": 7.59375, "learning_rate": 1.962915387746267e-05, "loss": 1.6219, "step": 4574 }, { "epoch": 11.4375, "grad_norm": 6.6875, "learning_rate": 1.9628935388977804e-05, "loss": 1.5618, "step": 4575 }, { "epoch": 11.44, "grad_norm": 7.25, "learning_rate": 1.962871683736603e-05, "loss": 1.5731, "step": 4576 }, { "epoch": 11.4425, "grad_norm": 6.53125, "learning_rate": 1.9628498222628764e-05, "loss": 1.5438, "step": 4577 }, { "epoch": 11.445, "grad_norm": 5.375, "learning_rate": 1.9628279544767457e-05, "loss": 1.5225, "step": 4578 }, { "epoch": 11.4475, "grad_norm": 7.3125, "learning_rate": 1.962806080378353e-05, "loss": 1.4143, "step": 4579 }, { "epoch": 11.45, "grad_norm": 5.96875, "learning_rate": 1.9627841999678422e-05, "loss": 1.479, "step": 4580 }, { "epoch": 11.4525, "grad_norm": 6.875, "learning_rate": 1.9627623132453565e-05, "loss": 1.4771, "step": 4581 }, { "epoch": 11.455, "grad_norm": 5.40625, "learning_rate": 1.96274042021104e-05, "loss": 1.3786, "step": 4582 }, { "epoch": 11.4575, "grad_norm": 6.34375, "learning_rate": 1.9627185208650357e-05, "loss": 1.4692, "step": 4583 }, { "epoch": 11.46, "grad_norm": 6.125, "learning_rate": 1.9626966152074876e-05, "loss": 1.6796, "step": 4584 }, { "epoch": 11.4625, "grad_norm": 7.96875, "learning_rate": 1.9626747032385385e-05, "loss": 1.4366, "step": 4585 }, { "epoch": 11.465, "grad_norm": 8.9375, "learning_rate": 1.962652784958333e-05, "loss": 1.5605, "step": 4586 }, { "epoch": 11.4675, "grad_norm": 11.0, "learning_rate": 1.962630860367014e-05, "loss": 1.4633, "step": 4587 }, { "epoch": 11.47, "grad_norm": 6.84375, "learning_rate": 1.9626089294647256e-05, "loss": 1.5132, "step": 4588 }, { "epoch": 11.4725, "grad_norm": 5.59375, "learning_rate": 1.962586992251612e-05, "loss": 1.3409, "step": 4589 }, { "epoch": 11.475, "grad_norm": 5.125, "learning_rate": 1.9625650487278164e-05, "loss": 1.4494, "step": 4590 }, { "epoch": 11.4775, "grad_norm": 7.65625, "learning_rate": 1.9625430988934827e-05, "loss": 1.4564, "step": 4591 }, { "epoch": 11.48, "grad_norm": 6.5625, "learning_rate": 1.962521142748755e-05, "loss": 1.4493, "step": 4592 }, { "epoch": 11.4825, "grad_norm": 8.0625, "learning_rate": 1.9624991802937772e-05, "loss": 1.626, "step": 4593 }, { "epoch": 11.485, "grad_norm": 5.96875, "learning_rate": 1.9624772115286933e-05, "loss": 1.6413, "step": 4594 }, { "epoch": 11.4875, "grad_norm": 5.9375, "learning_rate": 1.9624552364536472e-05, "loss": 1.4764, "step": 4595 }, { "epoch": 11.49, "grad_norm": 7.40625, "learning_rate": 1.9624332550687834e-05, "loss": 1.5276, "step": 4596 }, { "epoch": 11.4925, "grad_norm": 7.4375, "learning_rate": 1.9624112673742458e-05, "loss": 1.433, "step": 4597 }, { "epoch": 11.495, "grad_norm": 7.8125, "learning_rate": 1.962389273370178e-05, "loss": 1.5226, "step": 4598 }, { "epoch": 11.4975, "grad_norm": 5.5625, "learning_rate": 1.962367273056725e-05, "loss": 1.4501, "step": 4599 }, { "epoch": 11.5, "grad_norm": 6.34375, "learning_rate": 1.9623452664340305e-05, "loss": 1.3294, "step": 4600 }, { "epoch": 11.5025, "grad_norm": 6.1875, "learning_rate": 1.962323253502239e-05, "loss": 1.6223, "step": 4601 }, { "epoch": 11.505, "grad_norm": 7.625, "learning_rate": 1.9623012342614945e-05, "loss": 1.6322, "step": 4602 }, { "epoch": 11.5075, "grad_norm": 6.34375, "learning_rate": 1.9622792087119418e-05, "loss": 1.5378, "step": 4603 }, { "epoch": 11.51, "grad_norm": 6.46875, "learning_rate": 1.9622571768537254e-05, "loss": 1.515, "step": 4604 }, { "epoch": 11.5125, "grad_norm": 5.1875, "learning_rate": 1.9622351386869888e-05, "loss": 1.5153, "step": 4605 }, { "epoch": 11.515, "grad_norm": 5.9375, "learning_rate": 1.9622130942118777e-05, "loss": 1.5134, "step": 4606 }, { "epoch": 11.5175, "grad_norm": 6.28125, "learning_rate": 1.962191043428536e-05, "loss": 1.6474, "step": 4607 }, { "epoch": 11.52, "grad_norm": 6.5, "learning_rate": 1.9621689863371083e-05, "loss": 1.7254, "step": 4608 }, { "epoch": 11.5225, "grad_norm": 4.78125, "learning_rate": 1.962146922937739e-05, "loss": 1.2873, "step": 4609 }, { "epoch": 11.525, "grad_norm": 7.0625, "learning_rate": 1.9621248532305732e-05, "loss": 1.5331, "step": 4610 }, { "epoch": 11.5275, "grad_norm": 6.25, "learning_rate": 1.9621027772157553e-05, "loss": 1.4257, "step": 4611 }, { "epoch": 11.53, "grad_norm": 7.21875, "learning_rate": 1.9620806948934297e-05, "loss": 1.3443, "step": 4612 }, { "epoch": 11.5325, "grad_norm": 5.40625, "learning_rate": 1.9620586062637424e-05, "loss": 1.5596, "step": 4613 }, { "epoch": 11.535, "grad_norm": 5.3125, "learning_rate": 1.9620365113268366e-05, "loss": 1.4373, "step": 4614 }, { "epoch": 11.5375, "grad_norm": 7.4375, "learning_rate": 1.9620144100828585e-05, "loss": 1.4965, "step": 4615 }, { "epoch": 11.54, "grad_norm": 9.25, "learning_rate": 1.961992302531952e-05, "loss": 1.4974, "step": 4616 }, { "epoch": 11.5425, "grad_norm": 8.125, "learning_rate": 1.9619701886742625e-05, "loss": 1.5254, "step": 4617 }, { "epoch": 11.545, "grad_norm": 10.1875, "learning_rate": 1.9619480685099353e-05, "loss": 1.4915, "step": 4618 }, { "epoch": 11.5475, "grad_norm": 4.90625, "learning_rate": 1.9619259420391147e-05, "loss": 1.3769, "step": 4619 }, { "epoch": 11.55, "grad_norm": 5.1875, "learning_rate": 1.9619038092619465e-05, "loss": 1.58, "step": 4620 }, { "epoch": 11.5525, "grad_norm": 8.6875, "learning_rate": 1.961881670178575e-05, "loss": 1.4578, "step": 4621 }, { "epoch": 11.555, "grad_norm": 6.15625, "learning_rate": 1.961859524789146e-05, "loss": 1.5069, "step": 4622 }, { "epoch": 11.5575, "grad_norm": 5.125, "learning_rate": 1.9618373730938044e-05, "loss": 1.5551, "step": 4623 }, { "epoch": 11.56, "grad_norm": 7.125, "learning_rate": 1.9618152150926953e-05, "loss": 1.6356, "step": 4624 }, { "epoch": 11.5625, "grad_norm": 6.15625, "learning_rate": 1.9617930507859643e-05, "loss": 1.4786, "step": 4625 }, { "epoch": 11.565, "grad_norm": 6.84375, "learning_rate": 1.9617708801737568e-05, "loss": 1.4508, "step": 4626 }, { "epoch": 11.5675, "grad_norm": 5.0625, "learning_rate": 1.9617487032562173e-05, "loss": 1.324, "step": 4627 }, { "epoch": 11.57, "grad_norm": 7.0, "learning_rate": 1.9617265200334924e-05, "loss": 1.4545, "step": 4628 }, { "epoch": 11.5725, "grad_norm": 7.375, "learning_rate": 1.9617043305057266e-05, "loss": 1.5897, "step": 4629 }, { "epoch": 11.575, "grad_norm": 7.6875, "learning_rate": 1.961682134673066e-05, "loss": 1.4767, "step": 4630 }, { "epoch": 11.5775, "grad_norm": 7.875, "learning_rate": 1.9616599325356555e-05, "loss": 1.5028, "step": 4631 }, { "epoch": 11.58, "grad_norm": 8.0625, "learning_rate": 1.961637724093641e-05, "loss": 1.7297, "step": 4632 }, { "epoch": 11.5825, "grad_norm": 6.3125, "learning_rate": 1.961615509347168e-05, "loss": 1.5295, "step": 4633 }, { "epoch": 11.585, "grad_norm": 5.65625, "learning_rate": 1.961593288296382e-05, "loss": 1.45, "step": 4634 }, { "epoch": 11.5875, "grad_norm": 9.0, "learning_rate": 1.9615710609414293e-05, "loss": 1.3269, "step": 4635 }, { "epoch": 11.59, "grad_norm": 8.875, "learning_rate": 1.961548827282455e-05, "loss": 1.3689, "step": 4636 }, { "epoch": 11.5925, "grad_norm": 7.4375, "learning_rate": 1.9615265873196053e-05, "loss": 1.5677, "step": 4637 }, { "epoch": 11.595, "grad_norm": 7.03125, "learning_rate": 1.9615043410530258e-05, "loss": 1.6284, "step": 4638 }, { "epoch": 11.5975, "grad_norm": 9.0625, "learning_rate": 1.9614820884828617e-05, "loss": 1.5346, "step": 4639 }, { "epoch": 11.6, "grad_norm": 6.375, "learning_rate": 1.9614598296092603e-05, "loss": 1.5203, "step": 4640 }, { "epoch": 11.6025, "grad_norm": 5.0625, "learning_rate": 1.9614375644323663e-05, "loss": 1.3905, "step": 4641 }, { "epoch": 11.605, "grad_norm": 5.15625, "learning_rate": 1.961415292952326e-05, "loss": 1.4553, "step": 4642 }, { "epoch": 11.6075, "grad_norm": 4.875, "learning_rate": 1.9613930151692858e-05, "loss": 1.3723, "step": 4643 }, { "epoch": 11.61, "grad_norm": 7.25, "learning_rate": 1.9613707310833913e-05, "loss": 1.48, "step": 4644 }, { "epoch": 11.6125, "grad_norm": 8.625, "learning_rate": 1.961348440694789e-05, "loss": 1.5166, "step": 4645 }, { "epoch": 11.615, "grad_norm": 6.78125, "learning_rate": 1.9613261440036243e-05, "loss": 1.595, "step": 4646 }, { "epoch": 11.6175, "grad_norm": 8.0, "learning_rate": 1.961303841010044e-05, "loss": 1.5281, "step": 4647 }, { "epoch": 11.62, "grad_norm": 5.625, "learning_rate": 1.9612815317141947e-05, "loss": 1.5606, "step": 4648 }, { "epoch": 11.6225, "grad_norm": 10.125, "learning_rate": 1.9612592161162215e-05, "loss": 1.5063, "step": 4649 }, { "epoch": 11.625, "grad_norm": 6.78125, "learning_rate": 1.9612368942162717e-05, "loss": 1.5149, "step": 4650 }, { "epoch": 11.6275, "grad_norm": 5.34375, "learning_rate": 1.9612145660144914e-05, "loss": 1.4621, "step": 4651 }, { "epoch": 11.63, "grad_norm": 5.53125, "learning_rate": 1.9611922315110268e-05, "loss": 1.4472, "step": 4652 }, { "epoch": 11.6325, "grad_norm": 5.34375, "learning_rate": 1.9611698907060243e-05, "loss": 1.5499, "step": 4653 }, { "epoch": 11.635, "grad_norm": 6.59375, "learning_rate": 1.9611475435996303e-05, "loss": 1.5109, "step": 4654 }, { "epoch": 11.6375, "grad_norm": 8.1875, "learning_rate": 1.9611251901919916e-05, "loss": 1.4644, "step": 4655 }, { "epoch": 11.64, "grad_norm": 6.5, "learning_rate": 1.9611028304832547e-05, "loss": 1.8321, "step": 4656 }, { "epoch": 11.6425, "grad_norm": 5.03125, "learning_rate": 1.961080464473566e-05, "loss": 1.49, "step": 4657 }, { "epoch": 11.645, "grad_norm": 6.15625, "learning_rate": 1.9610580921630726e-05, "loss": 1.5702, "step": 4658 }, { "epoch": 11.6475, "grad_norm": 10.125, "learning_rate": 1.9610357135519204e-05, "loss": 1.3787, "step": 4659 }, { "epoch": 11.65, "grad_norm": 8.75, "learning_rate": 1.9610133286402565e-05, "loss": 1.5583, "step": 4660 }, { "epoch": 11.6525, "grad_norm": 6.84375, "learning_rate": 1.960990937428228e-05, "loss": 1.6561, "step": 4661 }, { "epoch": 11.655, "grad_norm": 5.5625, "learning_rate": 1.960968539915981e-05, "loss": 1.3977, "step": 4662 }, { "epoch": 11.6575, "grad_norm": 6.40625, "learning_rate": 1.960946136103663e-05, "loss": 1.45, "step": 4663 }, { "epoch": 11.66, "grad_norm": 9.0, "learning_rate": 1.96092372599142e-05, "loss": 1.607, "step": 4664 }, { "epoch": 11.6625, "grad_norm": 7.71875, "learning_rate": 1.9609013095794002e-05, "loss": 1.5539, "step": 4665 }, { "epoch": 11.665, "grad_norm": 5.5, "learning_rate": 1.9608788868677497e-05, "loss": 1.5099, "step": 4666 }, { "epoch": 11.6675, "grad_norm": 5.875, "learning_rate": 1.9608564578566155e-05, "loss": 1.6457, "step": 4667 }, { "epoch": 11.67, "grad_norm": 5.46875, "learning_rate": 1.9608340225461447e-05, "loss": 1.3879, "step": 4668 }, { "epoch": 11.6725, "grad_norm": 8.125, "learning_rate": 1.960811580936485e-05, "loss": 1.5274, "step": 4669 }, { "epoch": 11.675, "grad_norm": 7.0625, "learning_rate": 1.960789133027783e-05, "loss": 1.5302, "step": 4670 }, { "epoch": 11.6775, "grad_norm": 7.125, "learning_rate": 1.9607666788201853e-05, "loss": 1.5401, "step": 4671 }, { "epoch": 11.68, "grad_norm": 7.9375, "learning_rate": 1.9607442183138403e-05, "loss": 1.5527, "step": 4672 }, { "epoch": 11.6825, "grad_norm": 8.4375, "learning_rate": 1.9607217515088942e-05, "loss": 1.6409, "step": 4673 }, { "epoch": 11.685, "grad_norm": 8.875, "learning_rate": 1.960699278405495e-05, "loss": 1.7146, "step": 4674 }, { "epoch": 11.6875, "grad_norm": 6.4375, "learning_rate": 1.9606767990037896e-05, "loss": 1.4221, "step": 4675 }, { "epoch": 11.69, "grad_norm": 6.46875, "learning_rate": 1.9606543133039254e-05, "loss": 1.6265, "step": 4676 }, { "epoch": 11.6925, "grad_norm": 4.0, "learning_rate": 1.9606318213060504e-05, "loss": 1.2831, "step": 4677 }, { "epoch": 11.695, "grad_norm": 5.59375, "learning_rate": 1.9606093230103113e-05, "loss": 1.3091, "step": 4678 }, { "epoch": 11.6975, "grad_norm": 8.6875, "learning_rate": 1.9605868184168565e-05, "loss": 1.5424, "step": 4679 }, { "epoch": 11.7, "grad_norm": 10.375, "learning_rate": 1.9605643075258323e-05, "loss": 1.5547, "step": 4680 }, { "epoch": 11.7025, "grad_norm": 7.4375, "learning_rate": 1.9605417903373872e-05, "loss": 1.4873, "step": 4681 }, { "epoch": 11.705, "grad_norm": 8.6875, "learning_rate": 1.9605192668516682e-05, "loss": 1.828, "step": 4682 }, { "epoch": 11.7075, "grad_norm": 7.96875, "learning_rate": 1.960496737068824e-05, "loss": 1.6393, "step": 4683 }, { "epoch": 11.71, "grad_norm": 6.03125, "learning_rate": 1.960474200989001e-05, "loss": 1.399, "step": 4684 }, { "epoch": 11.7125, "grad_norm": 6.09375, "learning_rate": 1.9604516586123478e-05, "loss": 1.4728, "step": 4685 }, { "epoch": 11.715, "grad_norm": 6.34375, "learning_rate": 1.960429109939012e-05, "loss": 1.3886, "step": 4686 }, { "epoch": 11.7175, "grad_norm": 7.6875, "learning_rate": 1.9604065549691415e-05, "loss": 1.4111, "step": 4687 }, { "epoch": 11.72, "grad_norm": 7.15625, "learning_rate": 1.960383993702884e-05, "loss": 1.5628, "step": 4688 }, { "epoch": 11.7225, "grad_norm": 5.875, "learning_rate": 1.9603614261403873e-05, "loss": 1.5293, "step": 4689 }, { "epoch": 11.725, "grad_norm": 6.5, "learning_rate": 1.9603388522817998e-05, "loss": 1.601, "step": 4690 }, { "epoch": 11.7275, "grad_norm": 8.0, "learning_rate": 1.960316272127269e-05, "loss": 1.7151, "step": 4691 }, { "epoch": 11.73, "grad_norm": 5.40625, "learning_rate": 1.9602936856769432e-05, "loss": 1.5306, "step": 4692 }, { "epoch": 11.7325, "grad_norm": 6.46875, "learning_rate": 1.9602710929309705e-05, "loss": 1.7262, "step": 4693 }, { "epoch": 11.735, "grad_norm": 7.5625, "learning_rate": 1.960248493889499e-05, "loss": 1.5361, "step": 4694 }, { "epoch": 11.7375, "grad_norm": 6.84375, "learning_rate": 1.9602258885526765e-05, "loss": 1.5422, "step": 4695 }, { "epoch": 11.74, "grad_norm": 7.71875, "learning_rate": 1.9602032769206517e-05, "loss": 1.5041, "step": 4696 }, { "epoch": 11.7425, "grad_norm": 6.53125, "learning_rate": 1.9601806589935726e-05, "loss": 1.4911, "step": 4697 }, { "epoch": 11.745, "grad_norm": 7.71875, "learning_rate": 1.9601580347715875e-05, "loss": 1.4775, "step": 4698 }, { "epoch": 11.7475, "grad_norm": 7.375, "learning_rate": 1.960135404254845e-05, "loss": 1.6187, "step": 4699 }, { "epoch": 11.75, "grad_norm": 6.4375, "learning_rate": 1.960112767443493e-05, "loss": 1.4258, "step": 4700 }, { "epoch": 11.7525, "grad_norm": 6.40625, "learning_rate": 1.96009012433768e-05, "loss": 1.5046, "step": 4701 }, { "epoch": 11.755, "grad_norm": 5.84375, "learning_rate": 1.9600674749375546e-05, "loss": 1.433, "step": 4702 }, { "epoch": 11.7575, "grad_norm": 7.34375, "learning_rate": 1.9600448192432652e-05, "loss": 1.3409, "step": 4703 }, { "epoch": 11.76, "grad_norm": 6.875, "learning_rate": 1.9600221572549607e-05, "loss": 1.3575, "step": 4704 }, { "epoch": 11.7625, "grad_norm": 7.09375, "learning_rate": 1.9599994889727893e-05, "loss": 1.4256, "step": 4705 }, { "epoch": 11.765, "grad_norm": 8.25, "learning_rate": 1.9599768143968995e-05, "loss": 1.5084, "step": 4706 }, { "epoch": 11.7675, "grad_norm": 10.625, "learning_rate": 1.95995413352744e-05, "loss": 1.5032, "step": 4707 }, { "epoch": 11.77, "grad_norm": 6.0625, "learning_rate": 1.9599314463645597e-05, "loss": 1.322, "step": 4708 }, { "epoch": 11.7725, "grad_norm": 5.96875, "learning_rate": 1.9599087529084075e-05, "loss": 1.515, "step": 4709 }, { "epoch": 11.775, "grad_norm": 8.1875, "learning_rate": 1.959886053159132e-05, "loss": 1.6351, "step": 4710 }, { "epoch": 11.7775, "grad_norm": 8.5, "learning_rate": 1.9598633471168815e-05, "loss": 1.6974, "step": 4711 }, { "epoch": 11.78, "grad_norm": 6.8125, "learning_rate": 1.9598406347818056e-05, "loss": 1.5915, "step": 4712 }, { "epoch": 11.7825, "grad_norm": 7.25, "learning_rate": 1.959817916154053e-05, "loss": 1.4282, "step": 4713 }, { "epoch": 11.785, "grad_norm": 10.0625, "learning_rate": 1.9597951912337724e-05, "loss": 1.69, "step": 4714 }, { "epoch": 11.7875, "grad_norm": 7.375, "learning_rate": 1.9597724600211127e-05, "loss": 1.5899, "step": 4715 }, { "epoch": 11.79, "grad_norm": 9.3125, "learning_rate": 1.9597497225162233e-05, "loss": 1.4547, "step": 4716 }, { "epoch": 11.7925, "grad_norm": 7.8125, "learning_rate": 1.959726978719253e-05, "loss": 1.4809, "step": 4717 }, { "epoch": 11.795, "grad_norm": 7.875, "learning_rate": 1.9597042286303516e-05, "loss": 1.6822, "step": 4718 }, { "epoch": 11.7975, "grad_norm": 7.28125, "learning_rate": 1.9596814722496675e-05, "loss": 1.4603, "step": 4719 }, { "epoch": 11.8, "grad_norm": 7.125, "learning_rate": 1.9596587095773496e-05, "loss": 1.5039, "step": 4720 }, { "epoch": 11.8025, "grad_norm": 7.28125, "learning_rate": 1.9596359406135478e-05, "loss": 1.6068, "step": 4721 }, { "epoch": 11.805, "grad_norm": 10.25, "learning_rate": 1.9596131653584115e-05, "loss": 1.5621, "step": 4722 }, { "epoch": 11.8075, "grad_norm": 6.3125, "learning_rate": 1.9595903838120895e-05, "loss": 1.6147, "step": 4723 }, { "epoch": 11.81, "grad_norm": 6.96875, "learning_rate": 1.959567595974731e-05, "loss": 1.5952, "step": 4724 }, { "epoch": 11.8125, "grad_norm": 8.125, "learning_rate": 1.959544801846486e-05, "loss": 1.3966, "step": 4725 }, { "epoch": 11.815, "grad_norm": 5.625, "learning_rate": 1.959522001427504e-05, "loss": 1.6078, "step": 4726 }, { "epoch": 11.8175, "grad_norm": 5.375, "learning_rate": 1.9594991947179336e-05, "loss": 1.3721, "step": 4727 }, { "epoch": 11.82, "grad_norm": 8.125, "learning_rate": 1.9594763817179254e-05, "loss": 1.4951, "step": 4728 }, { "epoch": 11.8225, "grad_norm": 6.90625, "learning_rate": 1.959453562427628e-05, "loss": 1.5672, "step": 4729 }, { "epoch": 11.825, "grad_norm": 7.25, "learning_rate": 1.9594307368471918e-05, "loss": 1.5157, "step": 4730 }, { "epoch": 11.8275, "grad_norm": 9.0625, "learning_rate": 1.9594079049767656e-05, "loss": 1.641, "step": 4731 }, { "epoch": 11.83, "grad_norm": 6.65625, "learning_rate": 1.9593850668165002e-05, "loss": 1.5356, "step": 4732 }, { "epoch": 11.8325, "grad_norm": 9.375, "learning_rate": 1.959362222366544e-05, "loss": 1.5939, "step": 4733 }, { "epoch": 11.835, "grad_norm": 4.96875, "learning_rate": 1.959339371627048e-05, "loss": 1.4541, "step": 4734 }, { "epoch": 11.8375, "grad_norm": 11.75, "learning_rate": 1.9593165145981614e-05, "loss": 1.6645, "step": 4735 }, { "epoch": 11.84, "grad_norm": 6.0, "learning_rate": 1.959293651280034e-05, "loss": 1.5913, "step": 4736 }, { "epoch": 11.8425, "grad_norm": 6.15625, "learning_rate": 1.959270781672816e-05, "loss": 1.5418, "step": 4737 }, { "epoch": 11.845, "grad_norm": 6.4375, "learning_rate": 1.9592479057766568e-05, "loss": 1.4114, "step": 4738 }, { "epoch": 11.8475, "grad_norm": 11.3125, "learning_rate": 1.959225023591707e-05, "loss": 1.6344, "step": 4739 }, { "epoch": 11.85, "grad_norm": 5.9375, "learning_rate": 1.9592021351181163e-05, "loss": 1.5029, "step": 4740 }, { "epoch": 11.8525, "grad_norm": 7.3125, "learning_rate": 1.9591792403560346e-05, "loss": 1.4519, "step": 4741 }, { "epoch": 11.855, "grad_norm": 6.03125, "learning_rate": 1.9591563393056125e-05, "loss": 1.5109, "step": 4742 }, { "epoch": 11.8575, "grad_norm": 6.9375, "learning_rate": 1.9591334319669993e-05, "loss": 1.5628, "step": 4743 }, { "epoch": 11.86, "grad_norm": 8.1875, "learning_rate": 1.9591105183403463e-05, "loss": 1.6251, "step": 4744 }, { "epoch": 11.8625, "grad_norm": 7.0625, "learning_rate": 1.9590875984258025e-05, "loss": 1.6287, "step": 4745 }, { "epoch": 11.865, "grad_norm": 8.5, "learning_rate": 1.9590646722235193e-05, "loss": 1.6215, "step": 4746 }, { "epoch": 11.8675, "grad_norm": 6.6875, "learning_rate": 1.9590417397336465e-05, "loss": 1.665, "step": 4747 }, { "epoch": 11.87, "grad_norm": 7.03125, "learning_rate": 1.9590188009563342e-05, "loss": 1.4685, "step": 4748 }, { "epoch": 11.8725, "grad_norm": 7.90625, "learning_rate": 1.9589958558917332e-05, "loss": 1.4561, "step": 4749 }, { "epoch": 11.875, "grad_norm": 4.9375, "learning_rate": 1.9589729045399935e-05, "loss": 1.3767, "step": 4750 }, { "epoch": 11.8775, "grad_norm": 5.71875, "learning_rate": 1.958949946901266e-05, "loss": 1.5602, "step": 4751 }, { "epoch": 11.88, "grad_norm": 4.96875, "learning_rate": 1.958926982975701e-05, "loss": 1.4913, "step": 4752 }, { "epoch": 11.8825, "grad_norm": 10.3125, "learning_rate": 1.958904012763449e-05, "loss": 1.6171, "step": 4753 }, { "epoch": 11.885, "grad_norm": 7.46875, "learning_rate": 1.9588810362646608e-05, "loss": 1.5532, "step": 4754 }, { "epoch": 11.8875, "grad_norm": 6.59375, "learning_rate": 1.9588580534794864e-05, "loss": 1.4312, "step": 4755 }, { "epoch": 11.89, "grad_norm": 7.3125, "learning_rate": 1.9588350644080777e-05, "loss": 1.6336, "step": 4756 }, { "epoch": 11.8925, "grad_norm": 6.34375, "learning_rate": 1.9588120690505844e-05, "loss": 1.5706, "step": 4757 }, { "epoch": 11.895, "grad_norm": 4.78125, "learning_rate": 1.9587890674071573e-05, "loss": 1.5373, "step": 4758 }, { "epoch": 11.8975, "grad_norm": 6.6875, "learning_rate": 1.9587660594779477e-05, "loss": 1.5658, "step": 4759 }, { "epoch": 11.9, "grad_norm": 6.0625, "learning_rate": 1.958743045263106e-05, "loss": 1.3698, "step": 4760 }, { "epoch": 11.9025, "grad_norm": 6.65625, "learning_rate": 1.9587200247627837e-05, "loss": 1.6028, "step": 4761 }, { "epoch": 11.905, "grad_norm": 7.25, "learning_rate": 1.9586969979771307e-05, "loss": 1.5501, "step": 4762 }, { "epoch": 11.9075, "grad_norm": 6.875, "learning_rate": 1.958673964906299e-05, "loss": 1.6653, "step": 4763 }, { "epoch": 11.91, "grad_norm": 5.8125, "learning_rate": 1.958650925550439e-05, "loss": 1.4316, "step": 4764 }, { "epoch": 11.9125, "grad_norm": 7.0625, "learning_rate": 1.9586278799097017e-05, "loss": 1.4837, "step": 4765 }, { "epoch": 11.915, "grad_norm": 6.1875, "learning_rate": 1.9586048279842386e-05, "loss": 1.3962, "step": 4766 }, { "epoch": 11.9175, "grad_norm": 5.5, "learning_rate": 1.9585817697742006e-05, "loss": 1.5565, "step": 4767 }, { "epoch": 11.92, "grad_norm": 7.03125, "learning_rate": 1.958558705279739e-05, "loss": 1.4941, "step": 4768 }, { "epoch": 11.9225, "grad_norm": 5.8125, "learning_rate": 1.9585356345010045e-05, "loss": 1.4975, "step": 4769 }, { "epoch": 11.925, "grad_norm": 4.59375, "learning_rate": 1.958512557438149e-05, "loss": 1.3369, "step": 4770 }, { "epoch": 11.9275, "grad_norm": 8.0625, "learning_rate": 1.9584894740913234e-05, "loss": 1.6063, "step": 4771 }, { "epoch": 11.93, "grad_norm": 4.625, "learning_rate": 1.9584663844606792e-05, "loss": 1.4269, "step": 4772 }, { "epoch": 11.9325, "grad_norm": 5.40625, "learning_rate": 1.958443288546368e-05, "loss": 1.6537, "step": 4773 }, { "epoch": 11.935, "grad_norm": 7.8125, "learning_rate": 1.9584201863485404e-05, "loss": 1.5476, "step": 4774 }, { "epoch": 11.9375, "grad_norm": 6.65625, "learning_rate": 1.9583970778673487e-05, "loss": 1.4594, "step": 4775 }, { "epoch": 11.94, "grad_norm": 5.6875, "learning_rate": 1.9583739631029446e-05, "loss": 1.5135, "step": 4776 }, { "epoch": 11.9425, "grad_norm": 7.4375, "learning_rate": 1.9583508420554787e-05, "loss": 1.4778, "step": 4777 }, { "epoch": 11.945, "grad_norm": 6.40625, "learning_rate": 1.958327714725103e-05, "loss": 1.5112, "step": 4778 }, { "epoch": 11.9475, "grad_norm": 4.15625, "learning_rate": 1.958304581111969e-05, "loss": 1.446, "step": 4779 }, { "epoch": 11.95, "grad_norm": 5.375, "learning_rate": 1.9582814412162288e-05, "loss": 1.4731, "step": 4780 }, { "epoch": 11.9525, "grad_norm": 5.90625, "learning_rate": 1.9582582950380337e-05, "loss": 1.4091, "step": 4781 }, { "epoch": 11.955, "grad_norm": 7.3125, "learning_rate": 1.9582351425775358e-05, "loss": 1.5332, "step": 4782 }, { "epoch": 11.9575, "grad_norm": 5.125, "learning_rate": 1.9582119838348866e-05, "loss": 1.4113, "step": 4783 }, { "epoch": 11.96, "grad_norm": 6.78125, "learning_rate": 1.9581888188102375e-05, "loss": 1.5541, "step": 4784 }, { "epoch": 11.9625, "grad_norm": 8.0625, "learning_rate": 1.958165647503741e-05, "loss": 1.4476, "step": 4785 }, { "epoch": 11.965, "grad_norm": 7.875, "learning_rate": 1.9581424699155493e-05, "loss": 1.3616, "step": 4786 }, { "epoch": 11.9675, "grad_norm": 6.375, "learning_rate": 1.9581192860458138e-05, "loss": 1.627, "step": 4787 }, { "epoch": 11.97, "grad_norm": 6.8125, "learning_rate": 1.9580960958946865e-05, "loss": 1.5102, "step": 4788 }, { "epoch": 11.9725, "grad_norm": 6.09375, "learning_rate": 1.9580728994623194e-05, "loss": 1.4465, "step": 4789 }, { "epoch": 11.975, "grad_norm": 9.375, "learning_rate": 1.9580496967488648e-05, "loss": 1.6649, "step": 4790 }, { "epoch": 11.9775, "grad_norm": 14.9375, "learning_rate": 1.9580264877544745e-05, "loss": 1.5013, "step": 4791 }, { "epoch": 11.98, "grad_norm": 6.09375, "learning_rate": 1.958003272479301e-05, "loss": 1.5125, "step": 4792 }, { "epoch": 11.9825, "grad_norm": 7.75, "learning_rate": 1.9579800509234966e-05, "loss": 1.4021, "step": 4793 }, { "epoch": 11.985, "grad_norm": 7.0625, "learning_rate": 1.9579568230872135e-05, "loss": 1.4493, "step": 4794 }, { "epoch": 11.9875, "grad_norm": 7.78125, "learning_rate": 1.9579335889706033e-05, "loss": 1.5901, "step": 4795 }, { "epoch": 11.99, "grad_norm": 7.75, "learning_rate": 1.957910348573819e-05, "loss": 1.673, "step": 4796 }, { "epoch": 11.9925, "grad_norm": 39.5, "learning_rate": 1.957887101897013e-05, "loss": 1.4623, "step": 4797 }, { "epoch": 11.995, "grad_norm": 8.125, "learning_rate": 1.9578638489403375e-05, "loss": 1.674, "step": 4798 }, { "epoch": 11.9975, "grad_norm": 8.125, "learning_rate": 1.9578405897039444e-05, "loss": 1.4565, "step": 4799 }, { "epoch": 12.0, "grad_norm": 7.65625, "learning_rate": 1.957817324187987e-05, "loss": 1.486, "step": 4800 }, { "epoch": 12.0025, "grad_norm": 6.375, "learning_rate": 1.957794052392618e-05, "loss": 1.4413, "step": 4801 }, { "epoch": 12.005, "grad_norm": 6.75, "learning_rate": 1.957770774317989e-05, "loss": 1.5648, "step": 4802 }, { "epoch": 12.0075, "grad_norm": 7.5625, "learning_rate": 1.9577474899642532e-05, "loss": 1.5817, "step": 4803 }, { "epoch": 12.01, "grad_norm": 6.09375, "learning_rate": 1.9577241993315634e-05, "loss": 1.7118, "step": 4804 }, { "epoch": 12.0125, "grad_norm": 6.09375, "learning_rate": 1.957700902420072e-05, "loss": 1.5615, "step": 4805 }, { "epoch": 12.015, "grad_norm": 8.1875, "learning_rate": 1.9576775992299318e-05, "loss": 1.729, "step": 4806 }, { "epoch": 12.0175, "grad_norm": 7.40625, "learning_rate": 1.9576542897612953e-05, "loss": 1.5549, "step": 4807 }, { "epoch": 12.02, "grad_norm": 7.5625, "learning_rate": 1.957630974014316e-05, "loss": 1.5797, "step": 4808 }, { "epoch": 12.0225, "grad_norm": 7.25, "learning_rate": 1.9576076519891464e-05, "loss": 1.3112, "step": 4809 }, { "epoch": 12.025, "grad_norm": 6.25, "learning_rate": 1.9575843236859388e-05, "loss": 1.5209, "step": 4810 }, { "epoch": 12.0275, "grad_norm": 5.5, "learning_rate": 1.9575609891048473e-05, "loss": 1.4495, "step": 4811 }, { "epoch": 12.03, "grad_norm": 5.96875, "learning_rate": 1.957537648246024e-05, "loss": 1.4352, "step": 4812 }, { "epoch": 12.0325, "grad_norm": 6.875, "learning_rate": 1.957514301109622e-05, "loss": 1.4968, "step": 4813 }, { "epoch": 12.035, "grad_norm": 7.0, "learning_rate": 1.957490947695795e-05, "loss": 1.5367, "step": 4814 }, { "epoch": 12.0375, "grad_norm": 10.5, "learning_rate": 1.9574675880046953e-05, "loss": 1.4334, "step": 4815 }, { "epoch": 12.04, "grad_norm": 5.9375, "learning_rate": 1.9574442220364768e-05, "loss": 1.515, "step": 4816 }, { "epoch": 12.0425, "grad_norm": 8.625, "learning_rate": 1.9574208497912917e-05, "loss": 1.5373, "step": 4817 }, { "epoch": 12.045, "grad_norm": 7.40625, "learning_rate": 1.9573974712692944e-05, "loss": 1.5077, "step": 4818 }, { "epoch": 12.0475, "grad_norm": 7.03125, "learning_rate": 1.9573740864706372e-05, "loss": 1.4121, "step": 4819 }, { "epoch": 12.05, "grad_norm": 4.78125, "learning_rate": 1.957350695395474e-05, "loss": 1.3235, "step": 4820 }, { "epoch": 12.0525, "grad_norm": 6.03125, "learning_rate": 1.957327298043958e-05, "loss": 1.5821, "step": 4821 }, { "epoch": 12.055, "grad_norm": 5.625, "learning_rate": 1.9573038944162426e-05, "loss": 1.3942, "step": 4822 }, { "epoch": 12.0575, "grad_norm": 6.71875, "learning_rate": 1.957280484512481e-05, "loss": 1.4923, "step": 4823 }, { "epoch": 12.06, "grad_norm": 5.5, "learning_rate": 1.957257068332827e-05, "loss": 1.4832, "step": 4824 }, { "epoch": 12.0625, "grad_norm": 5.75, "learning_rate": 1.9572336458774336e-05, "loss": 1.4495, "step": 4825 }, { "epoch": 12.065, "grad_norm": 18.0, "learning_rate": 1.957210217146455e-05, "loss": 1.7258, "step": 4826 }, { "epoch": 12.0675, "grad_norm": 5.84375, "learning_rate": 1.9571867821400444e-05, "loss": 1.595, "step": 4827 }, { "epoch": 12.07, "grad_norm": 5.5, "learning_rate": 1.9571633408583558e-05, "loss": 1.5916, "step": 4828 }, { "epoch": 12.0725, "grad_norm": 5.25, "learning_rate": 1.9571398933015423e-05, "loss": 1.4636, "step": 4829 }, { "epoch": 12.075, "grad_norm": 5.4375, "learning_rate": 1.9571164394697585e-05, "loss": 1.502, "step": 4830 }, { "epoch": 12.0775, "grad_norm": 6.8125, "learning_rate": 1.957092979363157e-05, "loss": 1.4401, "step": 4831 }, { "epoch": 12.08, "grad_norm": 6.90625, "learning_rate": 1.9570695129818928e-05, "loss": 1.3157, "step": 4832 }, { "epoch": 12.0825, "grad_norm": 6.90625, "learning_rate": 1.9570460403261187e-05, "loss": 1.4141, "step": 4833 }, { "epoch": 12.085, "grad_norm": 5.75, "learning_rate": 1.957022561395989e-05, "loss": 1.3827, "step": 4834 }, { "epoch": 12.0875, "grad_norm": 7.09375, "learning_rate": 1.956999076191658e-05, "loss": 1.4738, "step": 4835 }, { "epoch": 12.09, "grad_norm": 6.03125, "learning_rate": 1.9569755847132796e-05, "loss": 1.4472, "step": 4836 }, { "epoch": 12.0925, "grad_norm": 7.71875, "learning_rate": 1.9569520869610073e-05, "loss": 1.3958, "step": 4837 }, { "epoch": 12.095, "grad_norm": 5.6875, "learning_rate": 1.956928582934995e-05, "loss": 1.5695, "step": 4838 }, { "epoch": 12.0975, "grad_norm": 7.6875, "learning_rate": 1.9569050726353977e-05, "loss": 1.4838, "step": 4839 }, { "epoch": 12.1, "grad_norm": 6.75, "learning_rate": 1.956881556062369e-05, "loss": 1.4738, "step": 4840 }, { "epoch": 12.1025, "grad_norm": 6.6875, "learning_rate": 1.956858033216063e-05, "loss": 1.4461, "step": 4841 }, { "epoch": 12.105, "grad_norm": 6.3125, "learning_rate": 1.9568345040966342e-05, "loss": 1.4438, "step": 4842 }, { "epoch": 12.1075, "grad_norm": 4.96875, "learning_rate": 1.9568109687042363e-05, "loss": 1.4774, "step": 4843 }, { "epoch": 12.11, "grad_norm": 5.34375, "learning_rate": 1.9567874270390242e-05, "loss": 1.3612, "step": 4844 }, { "epoch": 12.1125, "grad_norm": 8.6875, "learning_rate": 1.956763879101152e-05, "loss": 1.5403, "step": 4845 }, { "epoch": 12.115, "grad_norm": 6.4375, "learning_rate": 1.9567403248907743e-05, "loss": 1.4353, "step": 4846 }, { "epoch": 12.1175, "grad_norm": 8.0, "learning_rate": 1.9567167644080452e-05, "loss": 1.5985, "step": 4847 }, { "epoch": 12.12, "grad_norm": 5.8125, "learning_rate": 1.956693197653119e-05, "loss": 1.527, "step": 4848 }, { "epoch": 12.1225, "grad_norm": 5.375, "learning_rate": 1.9566696246261507e-05, "loss": 1.4052, "step": 4849 }, { "epoch": 12.125, "grad_norm": 4.59375, "learning_rate": 1.9566460453272945e-05, "loss": 1.607, "step": 4850 }, { "epoch": 12.1275, "grad_norm": 6.21875, "learning_rate": 1.9566224597567055e-05, "loss": 1.5354, "step": 4851 }, { "epoch": 12.13, "grad_norm": 6.78125, "learning_rate": 1.9565988679145375e-05, "loss": 1.5736, "step": 4852 }, { "epoch": 12.1325, "grad_norm": 8.0625, "learning_rate": 1.9565752698009457e-05, "loss": 1.3781, "step": 4853 }, { "epoch": 12.135, "grad_norm": 6.15625, "learning_rate": 1.9565516654160846e-05, "loss": 1.7319, "step": 4854 }, { "epoch": 12.1375, "grad_norm": 5.96875, "learning_rate": 1.9565280547601094e-05, "loss": 1.429, "step": 4855 }, { "epoch": 12.14, "grad_norm": 8.875, "learning_rate": 1.9565044378331745e-05, "loss": 1.4874, "step": 4856 }, { "epoch": 12.1425, "grad_norm": 5.1875, "learning_rate": 1.956480814635435e-05, "loss": 1.3799, "step": 4857 }, { "epoch": 12.145, "grad_norm": 6.53125, "learning_rate": 1.956457185167045e-05, "loss": 1.3723, "step": 4858 }, { "epoch": 12.1475, "grad_norm": 8.0625, "learning_rate": 1.95643354942816e-05, "loss": 1.4764, "step": 4859 }, { "epoch": 12.15, "grad_norm": 7.21875, "learning_rate": 1.956409907418935e-05, "loss": 1.4986, "step": 4860 }, { "epoch": 12.1525, "grad_norm": 7.75, "learning_rate": 1.9563862591395253e-05, "loss": 1.4289, "step": 4861 }, { "epoch": 12.155, "grad_norm": 5.46875, "learning_rate": 1.9563626045900852e-05, "loss": 1.5559, "step": 4862 }, { "epoch": 12.1575, "grad_norm": 11.75, "learning_rate": 1.9563389437707704e-05, "loss": 1.5081, "step": 4863 }, { "epoch": 12.16, "grad_norm": 6.65625, "learning_rate": 1.9563152766817356e-05, "loss": 1.4924, "step": 4864 }, { "epoch": 12.1625, "grad_norm": 6.0, "learning_rate": 1.9562916033231358e-05, "loss": 1.5273, "step": 4865 }, { "epoch": 12.165, "grad_norm": 8.3125, "learning_rate": 1.956267923695127e-05, "loss": 1.4539, "step": 4866 }, { "epoch": 12.1675, "grad_norm": 7.75, "learning_rate": 1.9562442377978636e-05, "loss": 1.4834, "step": 4867 }, { "epoch": 12.17, "grad_norm": 17.625, "learning_rate": 1.9562205456315017e-05, "loss": 1.6252, "step": 4868 }, { "epoch": 12.1725, "grad_norm": 4.8125, "learning_rate": 1.956196847196196e-05, "loss": 1.4424, "step": 4869 }, { "epoch": 12.175, "grad_norm": 6.3125, "learning_rate": 1.9561731424921014e-05, "loss": 1.4418, "step": 4870 }, { "epoch": 12.1775, "grad_norm": 4.71875, "learning_rate": 1.956149431519375e-05, "loss": 1.5794, "step": 4871 }, { "epoch": 12.18, "grad_norm": 5.9375, "learning_rate": 1.9561257142781706e-05, "loss": 1.5464, "step": 4872 }, { "epoch": 12.1825, "grad_norm": 6.15625, "learning_rate": 1.956101990768644e-05, "loss": 1.6015, "step": 4873 }, { "epoch": 12.185, "grad_norm": 6.21875, "learning_rate": 1.9560782609909514e-05, "loss": 1.5331, "step": 4874 }, { "epoch": 12.1875, "grad_norm": 7.46875, "learning_rate": 1.9560545249452477e-05, "loss": 1.4707, "step": 4875 }, { "epoch": 12.19, "grad_norm": 7.4375, "learning_rate": 1.9560307826316892e-05, "loss": 1.4907, "step": 4876 }, { "epoch": 12.1925, "grad_norm": 6.21875, "learning_rate": 1.956007034050431e-05, "loss": 1.4964, "step": 4877 }, { "epoch": 12.195, "grad_norm": 5.125, "learning_rate": 1.955983279201629e-05, "loss": 1.3385, "step": 4878 }, { "epoch": 12.1975, "grad_norm": 7.8125, "learning_rate": 1.9559595180854387e-05, "loss": 1.7346, "step": 4879 }, { "epoch": 12.2, "grad_norm": 6.03125, "learning_rate": 1.9559357507020163e-05, "loss": 1.4317, "step": 4880 }, { "epoch": 12.2025, "grad_norm": 6.90625, "learning_rate": 1.955911977051517e-05, "loss": 1.4155, "step": 4881 }, { "epoch": 12.205, "grad_norm": 6.0, "learning_rate": 1.9558881971340976e-05, "loss": 1.6439, "step": 4882 }, { "epoch": 12.2075, "grad_norm": 7.09375, "learning_rate": 1.955864410949913e-05, "loss": 1.5611, "step": 4883 }, { "epoch": 12.21, "grad_norm": 5.875, "learning_rate": 1.9558406184991195e-05, "loss": 1.43, "step": 4884 }, { "epoch": 12.2125, "grad_norm": 4.78125, "learning_rate": 1.9558168197818732e-05, "loss": 1.4892, "step": 4885 }, { "epoch": 12.215, "grad_norm": 7.5625, "learning_rate": 1.9557930147983303e-05, "loss": 1.4747, "step": 4886 }, { "epoch": 12.2175, "grad_norm": 6.1875, "learning_rate": 1.9557692035486468e-05, "loss": 1.6087, "step": 4887 }, { "epoch": 12.22, "grad_norm": 5.28125, "learning_rate": 1.955745386032978e-05, "loss": 1.4258, "step": 4888 }, { "epoch": 12.2225, "grad_norm": 8.25, "learning_rate": 1.955721562251481e-05, "loss": 1.6635, "step": 4889 }, { "epoch": 12.225, "grad_norm": 6.96875, "learning_rate": 1.955697732204312e-05, "loss": 1.4479, "step": 4890 }, { "epoch": 12.2275, "grad_norm": 7.6875, "learning_rate": 1.9556738958916265e-05, "loss": 1.5291, "step": 4891 }, { "epoch": 12.23, "grad_norm": 5.5, "learning_rate": 1.9556500533135817e-05, "loss": 1.5868, "step": 4892 }, { "epoch": 12.2325, "grad_norm": 5.625, "learning_rate": 1.955626204470333e-05, "loss": 1.4681, "step": 4893 }, { "epoch": 12.235, "grad_norm": 7.4375, "learning_rate": 1.955602349362037e-05, "loss": 1.6045, "step": 4894 }, { "epoch": 12.2375, "grad_norm": 4.46875, "learning_rate": 1.9555784879888504e-05, "loss": 1.6253, "step": 4895 }, { "epoch": 12.24, "grad_norm": 9.0625, "learning_rate": 1.9555546203509297e-05, "loss": 1.4616, "step": 4896 }, { "epoch": 12.2425, "grad_norm": 8.0, "learning_rate": 1.9555307464484308e-05, "loss": 1.4223, "step": 4897 }, { "epoch": 12.245, "grad_norm": 5.625, "learning_rate": 1.9555068662815107e-05, "loss": 1.324, "step": 4898 }, { "epoch": 12.2475, "grad_norm": 6.59375, "learning_rate": 1.955482979850326e-05, "loss": 1.5304, "step": 4899 }, { "epoch": 12.25, "grad_norm": 5.46875, "learning_rate": 1.955459087155033e-05, "loss": 1.4535, "step": 4900 }, { "epoch": 12.2525, "grad_norm": 5.3125, "learning_rate": 1.9554351881957884e-05, "loss": 1.6604, "step": 4901 }, { "epoch": 12.255, "grad_norm": 6.9375, "learning_rate": 1.9554112829727487e-05, "loss": 1.3465, "step": 4902 }, { "epoch": 12.2575, "grad_norm": 5.65625, "learning_rate": 1.9553873714860712e-05, "loss": 1.5107, "step": 4903 }, { "epoch": 12.26, "grad_norm": 8.1875, "learning_rate": 1.9553634537359123e-05, "loss": 1.5484, "step": 4904 }, { "epoch": 12.2625, "grad_norm": 6.375, "learning_rate": 1.9553395297224286e-05, "loss": 1.4097, "step": 4905 }, { "epoch": 12.265, "grad_norm": 8.0625, "learning_rate": 1.9553155994457773e-05, "loss": 1.4573, "step": 4906 }, { "epoch": 12.2675, "grad_norm": 6.90625, "learning_rate": 1.9552916629061155e-05, "loss": 1.6597, "step": 4907 }, { "epoch": 12.27, "grad_norm": 6.46875, "learning_rate": 1.9552677201035993e-05, "loss": 1.3856, "step": 4908 }, { "epoch": 12.2725, "grad_norm": 6.3125, "learning_rate": 1.9552437710383863e-05, "loss": 1.4358, "step": 4909 }, { "epoch": 12.275, "grad_norm": 4.625, "learning_rate": 1.9552198157106336e-05, "loss": 1.4379, "step": 4910 }, { "epoch": 12.2775, "grad_norm": 7.125, "learning_rate": 1.955195854120498e-05, "loss": 1.6751, "step": 4911 }, { "epoch": 12.28, "grad_norm": 5.09375, "learning_rate": 1.9551718862681363e-05, "loss": 1.2627, "step": 4912 }, { "epoch": 12.2825, "grad_norm": 6.96875, "learning_rate": 1.9551479121537058e-05, "loss": 1.6155, "step": 4913 }, { "epoch": 12.285, "grad_norm": 7.375, "learning_rate": 1.9551239317773643e-05, "loss": 1.3949, "step": 4914 }, { "epoch": 12.2875, "grad_norm": 6.6875, "learning_rate": 1.9550999451392684e-05, "loss": 1.5357, "step": 4915 }, { "epoch": 12.29, "grad_norm": 6.09375, "learning_rate": 1.9550759522395753e-05, "loss": 1.4839, "step": 4916 }, { "epoch": 12.2925, "grad_norm": 5.0, "learning_rate": 1.9550519530784425e-05, "loss": 1.3474, "step": 4917 }, { "epoch": 12.295, "grad_norm": 9.8125, "learning_rate": 1.9550279476560273e-05, "loss": 1.4673, "step": 4918 }, { "epoch": 12.2975, "grad_norm": 6.21875, "learning_rate": 1.955003935972487e-05, "loss": 1.3397, "step": 4919 }, { "epoch": 12.3, "grad_norm": 9.0, "learning_rate": 1.9549799180279793e-05, "loss": 1.5882, "step": 4920 }, { "epoch": 12.3025, "grad_norm": 8.0, "learning_rate": 1.9549558938226617e-05, "loss": 1.5565, "step": 4921 }, { "epoch": 12.305, "grad_norm": 14.5625, "learning_rate": 1.9549318633566914e-05, "loss": 1.3908, "step": 4922 }, { "epoch": 12.3075, "grad_norm": 5.90625, "learning_rate": 1.9549078266302258e-05, "loss": 1.5388, "step": 4923 }, { "epoch": 12.31, "grad_norm": 7.96875, "learning_rate": 1.9548837836434227e-05, "loss": 1.4626, "step": 4924 }, { "epoch": 12.3125, "grad_norm": 5.6875, "learning_rate": 1.9548597343964396e-05, "loss": 1.5203, "step": 4925 }, { "epoch": 12.315, "grad_norm": 6.9375, "learning_rate": 1.9548356788894348e-05, "loss": 1.7692, "step": 4926 }, { "epoch": 12.3175, "grad_norm": 5.28125, "learning_rate": 1.954811617122565e-05, "loss": 1.3832, "step": 4927 }, { "epoch": 12.32, "grad_norm": 7.125, "learning_rate": 1.9547875490959884e-05, "loss": 1.4957, "step": 4928 }, { "epoch": 12.3225, "grad_norm": 6.46875, "learning_rate": 1.9547634748098634e-05, "loss": 1.5673, "step": 4929 }, { "epoch": 12.325, "grad_norm": 11.75, "learning_rate": 1.9547393942643468e-05, "loss": 1.6238, "step": 4930 }, { "epoch": 12.3275, "grad_norm": 8.125, "learning_rate": 1.9547153074595972e-05, "loss": 1.4482, "step": 4931 }, { "epoch": 12.33, "grad_norm": 7.59375, "learning_rate": 1.954691214395772e-05, "loss": 1.3687, "step": 4932 }, { "epoch": 12.3325, "grad_norm": 6.65625, "learning_rate": 1.9546671150730293e-05, "loss": 1.4925, "step": 4933 }, { "epoch": 12.335, "grad_norm": 5.71875, "learning_rate": 1.9546430094915275e-05, "loss": 1.6681, "step": 4934 }, { "epoch": 12.3375, "grad_norm": 6.21875, "learning_rate": 1.9546188976514246e-05, "loss": 1.5719, "step": 4935 }, { "epoch": 12.34, "grad_norm": 4.65625, "learning_rate": 1.9545947795528777e-05, "loss": 1.4747, "step": 4936 }, { "epoch": 12.3425, "grad_norm": 6.0, "learning_rate": 1.954570655196046e-05, "loss": 1.4147, "step": 4937 }, { "epoch": 12.345, "grad_norm": 5.53125, "learning_rate": 1.9545465245810874e-05, "loss": 1.5161, "step": 4938 }, { "epoch": 12.3475, "grad_norm": 6.46875, "learning_rate": 1.9545223877081597e-05, "loss": 1.5363, "step": 4939 }, { "epoch": 12.35, "grad_norm": 6.90625, "learning_rate": 1.9544982445774217e-05, "loss": 1.5519, "step": 4940 }, { "epoch": 12.3525, "grad_norm": 5.1875, "learning_rate": 1.9544740951890312e-05, "loss": 1.4186, "step": 4941 }, { "epoch": 12.355, "grad_norm": 5.75, "learning_rate": 1.954449939543147e-05, "loss": 1.3624, "step": 4942 }, { "epoch": 12.3575, "grad_norm": 5.25, "learning_rate": 1.954425777639927e-05, "loss": 1.5371, "step": 4943 }, { "epoch": 12.36, "grad_norm": 7.96875, "learning_rate": 1.9544016094795294e-05, "loss": 1.5173, "step": 4944 }, { "epoch": 12.3625, "grad_norm": 6.875, "learning_rate": 1.9543774350621136e-05, "loss": 1.5001, "step": 4945 }, { "epoch": 12.365, "grad_norm": 5.59375, "learning_rate": 1.9543532543878372e-05, "loss": 1.5373, "step": 4946 }, { "epoch": 12.3675, "grad_norm": 4.96875, "learning_rate": 1.9543290674568595e-05, "loss": 1.4979, "step": 4947 }, { "epoch": 12.37, "grad_norm": 5.84375, "learning_rate": 1.9543048742693383e-05, "loss": 1.4614, "step": 4948 }, { "epoch": 12.3725, "grad_norm": 4.5625, "learning_rate": 1.9542806748254325e-05, "loss": 1.238, "step": 4949 }, { "epoch": 12.375, "grad_norm": 5.84375, "learning_rate": 1.954256469125301e-05, "loss": 1.409, "step": 4950 }, { "epoch": 12.3775, "grad_norm": 5.90625, "learning_rate": 1.9542322571691018e-05, "loss": 1.5162, "step": 4951 }, { "epoch": 12.38, "grad_norm": 7.25, "learning_rate": 1.9542080389569947e-05, "loss": 1.6633, "step": 4952 }, { "epoch": 12.3825, "grad_norm": 6.4375, "learning_rate": 1.9541838144891376e-05, "loss": 1.6233, "step": 4953 }, { "epoch": 12.385, "grad_norm": 6.0, "learning_rate": 1.9541595837656895e-05, "loss": 1.6169, "step": 4954 }, { "epoch": 12.3875, "grad_norm": 5.875, "learning_rate": 1.9541353467868095e-05, "loss": 1.3797, "step": 4955 }, { "epoch": 12.39, "grad_norm": 7.0625, "learning_rate": 1.9541111035526563e-05, "loss": 1.5143, "step": 4956 }, { "epoch": 12.3925, "grad_norm": 6.6875, "learning_rate": 1.9540868540633887e-05, "loss": 1.3405, "step": 4957 }, { "epoch": 12.395, "grad_norm": 6.21875, "learning_rate": 1.9540625983191663e-05, "loss": 1.2088, "step": 4958 }, { "epoch": 12.3975, "grad_norm": 7.78125, "learning_rate": 1.9540383363201477e-05, "loss": 1.4934, "step": 4959 }, { "epoch": 12.4, "grad_norm": 7.59375, "learning_rate": 1.9540140680664915e-05, "loss": 1.5703, "step": 4960 }, { "epoch": 12.4025, "grad_norm": 8.5, "learning_rate": 1.9539897935583575e-05, "loss": 1.4988, "step": 4961 }, { "epoch": 12.405, "grad_norm": 10.75, "learning_rate": 1.9539655127959044e-05, "loss": 1.5144, "step": 4962 }, { "epoch": 12.4075, "grad_norm": 6.34375, "learning_rate": 1.9539412257792916e-05, "loss": 1.3965, "step": 4963 }, { "epoch": 12.41, "grad_norm": 7.78125, "learning_rate": 1.9539169325086786e-05, "loss": 1.4064, "step": 4964 }, { "epoch": 12.4125, "grad_norm": 7.125, "learning_rate": 1.953892632984224e-05, "loss": 1.2743, "step": 4965 }, { "epoch": 12.415, "grad_norm": 9.75, "learning_rate": 1.9538683272060877e-05, "loss": 1.5388, "step": 4966 }, { "epoch": 12.4175, "grad_norm": 7.1875, "learning_rate": 1.9538440151744287e-05, "loss": 1.6926, "step": 4967 }, { "epoch": 12.42, "grad_norm": 5.3125, "learning_rate": 1.9538196968894067e-05, "loss": 1.4729, "step": 4968 }, { "epoch": 12.4225, "grad_norm": 5.46875, "learning_rate": 1.953795372351181e-05, "loss": 1.2987, "step": 4969 }, { "epoch": 12.425, "grad_norm": 9.0, "learning_rate": 1.9537710415599106e-05, "loss": 1.5378, "step": 4970 }, { "epoch": 12.4275, "grad_norm": 5.375, "learning_rate": 1.9537467045157556e-05, "loss": 1.4104, "step": 4971 }, { "epoch": 12.43, "grad_norm": 6.21875, "learning_rate": 1.9537223612188752e-05, "loss": 1.6316, "step": 4972 }, { "epoch": 12.4325, "grad_norm": 7.5, "learning_rate": 1.9536980116694297e-05, "loss": 1.5354, "step": 4973 }, { "epoch": 12.435, "grad_norm": 7.34375, "learning_rate": 1.953673655867578e-05, "loss": 1.3609, "step": 4974 }, { "epoch": 12.4375, "grad_norm": 6.59375, "learning_rate": 1.9536492938134796e-05, "loss": 1.5172, "step": 4975 }, { "epoch": 12.44, "grad_norm": 8.125, "learning_rate": 1.953624925507295e-05, "loss": 1.5339, "step": 4976 }, { "epoch": 12.4425, "grad_norm": 5.96875, "learning_rate": 1.9536005509491833e-05, "loss": 1.2975, "step": 4977 }, { "epoch": 12.445, "grad_norm": 4.96875, "learning_rate": 1.9535761701393045e-05, "loss": 1.5368, "step": 4978 }, { "epoch": 12.4475, "grad_norm": 7.78125, "learning_rate": 1.9535517830778188e-05, "loss": 1.4334, "step": 4979 }, { "epoch": 12.45, "grad_norm": 7.25, "learning_rate": 1.9535273897648857e-05, "loss": 1.4273, "step": 4980 }, { "epoch": 12.4525, "grad_norm": 5.90625, "learning_rate": 1.953502990200665e-05, "loss": 1.5074, "step": 4981 }, { "epoch": 12.455, "grad_norm": 5.15625, "learning_rate": 1.953478584385317e-05, "loss": 1.3339, "step": 4982 }, { "epoch": 12.4575, "grad_norm": 8.5, "learning_rate": 1.953454172319001e-05, "loss": 1.5628, "step": 4983 }, { "epoch": 12.46, "grad_norm": 5.5625, "learning_rate": 1.9534297540018787e-05, "loss": 1.3991, "step": 4984 }, { "epoch": 12.4625, "grad_norm": 5.59375, "learning_rate": 1.953405329434108e-05, "loss": 1.3613, "step": 4985 }, { "epoch": 12.465, "grad_norm": 8.25, "learning_rate": 1.953380898615851e-05, "loss": 1.5487, "step": 4986 }, { "epoch": 12.4675, "grad_norm": 5.125, "learning_rate": 1.9533564615472663e-05, "loss": 1.4091, "step": 4987 }, { "epoch": 12.47, "grad_norm": 5.25, "learning_rate": 1.9533320182285152e-05, "loss": 1.3672, "step": 4988 }, { "epoch": 12.4725, "grad_norm": 6.21875, "learning_rate": 1.9533075686597574e-05, "loss": 1.4085, "step": 4989 }, { "epoch": 12.475, "grad_norm": 5.375, "learning_rate": 1.9532831128411535e-05, "loss": 1.3136, "step": 4990 }, { "epoch": 12.4775, "grad_norm": 8.3125, "learning_rate": 1.9532586507728633e-05, "loss": 1.6796, "step": 4991 }, { "epoch": 12.48, "grad_norm": 6.34375, "learning_rate": 1.953234182455048e-05, "loss": 1.3796, "step": 4992 }, { "epoch": 12.4825, "grad_norm": 7.09375, "learning_rate": 1.953209707887867e-05, "loss": 1.582, "step": 4993 }, { "epoch": 12.485, "grad_norm": 4.21875, "learning_rate": 1.9531852270714817e-05, "loss": 1.4354, "step": 4994 }, { "epoch": 12.4875, "grad_norm": 6.46875, "learning_rate": 1.9531607400060522e-05, "loss": 1.5208, "step": 4995 }, { "epoch": 12.49, "grad_norm": 5.5, "learning_rate": 1.9531362466917388e-05, "loss": 1.5951, "step": 4996 }, { "epoch": 12.4925, "grad_norm": 5.34375, "learning_rate": 1.9531117471287024e-05, "loss": 1.4989, "step": 4997 }, { "epoch": 12.495, "grad_norm": 5.125, "learning_rate": 1.9530872413171033e-05, "loss": 1.4682, "step": 4998 }, { "epoch": 12.4975, "grad_norm": 5.15625, "learning_rate": 1.953062729257103e-05, "loss": 1.3826, "step": 4999 }, { "epoch": 12.5, "grad_norm": 7.4375, "learning_rate": 1.953038210948861e-05, "loss": 1.5513, "step": 5000 }, { "epoch": 12.5025, "grad_norm": 7.125, "learning_rate": 1.953013686392539e-05, "loss": 1.4476, "step": 5001 }, { "epoch": 12.505, "grad_norm": 7.28125, "learning_rate": 1.9529891555882973e-05, "loss": 1.5886, "step": 5002 }, { "epoch": 12.5075, "grad_norm": 8.625, "learning_rate": 1.952964618536297e-05, "loss": 1.5121, "step": 5003 }, { "epoch": 12.51, "grad_norm": 7.4375, "learning_rate": 1.9529400752366983e-05, "loss": 1.4721, "step": 5004 }, { "epoch": 12.5125, "grad_norm": 7.28125, "learning_rate": 1.952915525689663e-05, "loss": 1.4484, "step": 5005 }, { "epoch": 12.515, "grad_norm": 7.25, "learning_rate": 1.9528909698953512e-05, "loss": 1.4051, "step": 5006 }, { "epoch": 12.5175, "grad_norm": 8.125, "learning_rate": 1.952866407853925e-05, "loss": 1.4437, "step": 5007 }, { "epoch": 12.52, "grad_norm": 6.96875, "learning_rate": 1.9528418395655443e-05, "loss": 1.4674, "step": 5008 }, { "epoch": 12.5225, "grad_norm": 4.8125, "learning_rate": 1.9528172650303708e-05, "loss": 1.3993, "step": 5009 }, { "epoch": 12.525, "grad_norm": 6.78125, "learning_rate": 1.9527926842485652e-05, "loss": 1.4246, "step": 5010 }, { "epoch": 12.5275, "grad_norm": 8.5, "learning_rate": 1.952768097220289e-05, "loss": 1.3609, "step": 5011 }, { "epoch": 12.53, "grad_norm": 5.6875, "learning_rate": 1.9527435039457035e-05, "loss": 1.5379, "step": 5012 }, { "epoch": 12.5325, "grad_norm": 8.625, "learning_rate": 1.9527189044249697e-05, "loss": 1.4405, "step": 5013 }, { "epoch": 12.535, "grad_norm": 27.75, "learning_rate": 1.9526942986582485e-05, "loss": 1.4226, "step": 5014 }, { "epoch": 12.5375, "grad_norm": 6.15625, "learning_rate": 1.952669686645702e-05, "loss": 1.5884, "step": 5015 }, { "epoch": 12.54, "grad_norm": 7.65625, "learning_rate": 1.952645068387491e-05, "loss": 1.6064, "step": 5016 }, { "epoch": 12.5425, "grad_norm": 6.0, "learning_rate": 1.952620443883777e-05, "loss": 1.5616, "step": 5017 }, { "epoch": 12.545, "grad_norm": 6.15625, "learning_rate": 1.9525958131347215e-05, "loss": 1.5855, "step": 5018 }, { "epoch": 12.5475, "grad_norm": 8.125, "learning_rate": 1.952571176140486e-05, "loss": 1.4926, "step": 5019 }, { "epoch": 12.55, "grad_norm": 5.84375, "learning_rate": 1.9525465329012322e-05, "loss": 1.4297, "step": 5020 }, { "epoch": 12.5525, "grad_norm": 8.125, "learning_rate": 1.9525218834171213e-05, "loss": 1.5666, "step": 5021 }, { "epoch": 12.555, "grad_norm": 5.625, "learning_rate": 1.952497227688315e-05, "loss": 1.5289, "step": 5022 }, { "epoch": 12.5575, "grad_norm": 4.78125, "learning_rate": 1.952472565714975e-05, "loss": 1.4814, "step": 5023 }, { "epoch": 12.56, "grad_norm": 7.375, "learning_rate": 1.952447897497263e-05, "loss": 1.4879, "step": 5024 }, { "epoch": 12.5625, "grad_norm": 7.65625, "learning_rate": 1.9524232230353408e-05, "loss": 1.596, "step": 5025 }, { "epoch": 12.565, "grad_norm": 7.25, "learning_rate": 1.95239854232937e-05, "loss": 1.4235, "step": 5026 }, { "epoch": 12.5675, "grad_norm": 6.96875, "learning_rate": 1.9523738553795123e-05, "loss": 1.4541, "step": 5027 }, { "epoch": 12.57, "grad_norm": 7.4375, "learning_rate": 1.95234916218593e-05, "loss": 1.5562, "step": 5028 }, { "epoch": 12.5725, "grad_norm": 7.03125, "learning_rate": 1.9523244627487845e-05, "loss": 1.4395, "step": 5029 }, { "epoch": 12.575, "grad_norm": 5.75, "learning_rate": 1.952299757068238e-05, "loss": 1.56, "step": 5030 }, { "epoch": 12.5775, "grad_norm": 6.6875, "learning_rate": 1.9522750451444524e-05, "loss": 1.5618, "step": 5031 }, { "epoch": 12.58, "grad_norm": 7.40625, "learning_rate": 1.9522503269775897e-05, "loss": 1.5412, "step": 5032 }, { "epoch": 12.5825, "grad_norm": 7.3125, "learning_rate": 1.952225602567812e-05, "loss": 1.4795, "step": 5033 }, { "epoch": 12.585, "grad_norm": 4.90625, "learning_rate": 1.9522008719152814e-05, "loss": 1.4512, "step": 5034 }, { "epoch": 12.5875, "grad_norm": 5.90625, "learning_rate": 1.9521761350201598e-05, "loss": 1.5879, "step": 5035 }, { "epoch": 12.59, "grad_norm": 9.4375, "learning_rate": 1.95215139188261e-05, "loss": 1.5153, "step": 5036 }, { "epoch": 12.5925, "grad_norm": 5.5625, "learning_rate": 1.9521266425027932e-05, "loss": 1.4436, "step": 5037 }, { "epoch": 12.595, "grad_norm": 6.96875, "learning_rate": 1.9521018868808722e-05, "loss": 1.534, "step": 5038 }, { "epoch": 12.5975, "grad_norm": 5.5, "learning_rate": 1.95207712501701e-05, "loss": 1.4254, "step": 5039 }, { "epoch": 12.6, "grad_norm": 5.875, "learning_rate": 1.952052356911368e-05, "loss": 1.6051, "step": 5040 }, { "epoch": 12.6025, "grad_norm": 6.65625, "learning_rate": 1.9520275825641084e-05, "loss": 1.4525, "step": 5041 }, { "epoch": 12.605, "grad_norm": 5.84375, "learning_rate": 1.9520028019753946e-05, "loss": 1.5616, "step": 5042 }, { "epoch": 12.6075, "grad_norm": 8.25, "learning_rate": 1.9519780151453883e-05, "loss": 1.5668, "step": 5043 }, { "epoch": 12.61, "grad_norm": 7.0, "learning_rate": 1.951953222074252e-05, "loss": 1.3275, "step": 5044 }, { "epoch": 12.6125, "grad_norm": 7.4375, "learning_rate": 1.9519284227621488e-05, "loss": 1.5016, "step": 5045 }, { "epoch": 12.615, "grad_norm": 6.9375, "learning_rate": 1.9519036172092408e-05, "loss": 1.4963, "step": 5046 }, { "epoch": 12.6175, "grad_norm": 6.1875, "learning_rate": 1.951878805415691e-05, "loss": 1.5488, "step": 5047 }, { "epoch": 12.62, "grad_norm": 6.4375, "learning_rate": 1.9518539873816617e-05, "loss": 1.5193, "step": 5048 }, { "epoch": 12.6225, "grad_norm": 6.40625, "learning_rate": 1.9518291631073157e-05, "loss": 1.4709, "step": 5049 }, { "epoch": 12.625, "grad_norm": 7.8125, "learning_rate": 1.9518043325928157e-05, "loss": 1.5641, "step": 5050 }, { "epoch": 12.6275, "grad_norm": 7.15625, "learning_rate": 1.951779495838325e-05, "loss": 1.6679, "step": 5051 }, { "epoch": 12.63, "grad_norm": 6.75, "learning_rate": 1.9517546528440054e-05, "loss": 1.4067, "step": 5052 }, { "epoch": 12.6325, "grad_norm": 7.0625, "learning_rate": 1.951729803610021e-05, "loss": 1.7078, "step": 5053 }, { "epoch": 12.635, "grad_norm": 5.34375, "learning_rate": 1.9517049481365338e-05, "loss": 1.4688, "step": 5054 }, { "epoch": 12.6375, "grad_norm": 9.9375, "learning_rate": 1.951680086423707e-05, "loss": 1.6761, "step": 5055 }, { "epoch": 12.64, "grad_norm": 10.125, "learning_rate": 1.9516552184717036e-05, "loss": 1.4572, "step": 5056 }, { "epoch": 12.6425, "grad_norm": 6.5625, "learning_rate": 1.951630344280687e-05, "loss": 1.5329, "step": 5057 }, { "epoch": 12.645, "grad_norm": 8.1875, "learning_rate": 1.9516054638508196e-05, "loss": 1.4247, "step": 5058 }, { "epoch": 12.6475, "grad_norm": 10.375, "learning_rate": 1.951580577182265e-05, "loss": 1.5535, "step": 5059 }, { "epoch": 12.65, "grad_norm": 6.71875, "learning_rate": 1.9515556842751863e-05, "loss": 1.4989, "step": 5060 }, { "epoch": 12.6525, "grad_norm": 4.59375, "learning_rate": 1.9515307851297462e-05, "loss": 1.4439, "step": 5061 }, { "epoch": 12.655, "grad_norm": 6.21875, "learning_rate": 1.9515058797461087e-05, "loss": 1.418, "step": 5062 }, { "epoch": 12.6575, "grad_norm": 4.25, "learning_rate": 1.951480968124437e-05, "loss": 1.4339, "step": 5063 }, { "epoch": 12.66, "grad_norm": 6.0625, "learning_rate": 1.9514560502648936e-05, "loss": 1.4292, "step": 5064 }, { "epoch": 12.6625, "grad_norm": 7.78125, "learning_rate": 1.9514311261676427e-05, "loss": 1.6754, "step": 5065 }, { "epoch": 12.665, "grad_norm": 6.53125, "learning_rate": 1.9514061958328475e-05, "loss": 1.4008, "step": 5066 }, { "epoch": 12.6675, "grad_norm": 10.8125, "learning_rate": 1.951381259260671e-05, "loss": 1.6247, "step": 5067 }, { "epoch": 12.67, "grad_norm": 7.40625, "learning_rate": 1.9513563164512773e-05, "loss": 1.4334, "step": 5068 }, { "epoch": 12.6725, "grad_norm": 5.125, "learning_rate": 1.9513313674048295e-05, "loss": 1.4536, "step": 5069 }, { "epoch": 12.675, "grad_norm": 7.3125, "learning_rate": 1.9513064121214914e-05, "loss": 1.5933, "step": 5070 }, { "epoch": 12.6775, "grad_norm": 6.71875, "learning_rate": 1.951281450601427e-05, "loss": 1.4755, "step": 5071 }, { "epoch": 12.68, "grad_norm": 5.40625, "learning_rate": 1.951256482844799e-05, "loss": 1.5496, "step": 5072 }, { "epoch": 12.6825, "grad_norm": 7.15625, "learning_rate": 1.951231508851771e-05, "loss": 1.4251, "step": 5073 }, { "epoch": 12.685, "grad_norm": 10.3125, "learning_rate": 1.951206528622508e-05, "loss": 1.3652, "step": 5074 }, { "epoch": 12.6875, "grad_norm": 6.25, "learning_rate": 1.9511815421571733e-05, "loss": 1.5898, "step": 5075 }, { "epoch": 12.69, "grad_norm": 12.25, "learning_rate": 1.9511565494559298e-05, "loss": 1.7189, "step": 5076 }, { "epoch": 12.6925, "grad_norm": 5.8125, "learning_rate": 1.9511315505189425e-05, "loss": 1.5065, "step": 5077 }, { "epoch": 12.695, "grad_norm": 8.375, "learning_rate": 1.9511065453463748e-05, "loss": 1.3998, "step": 5078 }, { "epoch": 12.6975, "grad_norm": 8.6875, "learning_rate": 1.9510815339383904e-05, "loss": 1.6164, "step": 5079 }, { "epoch": 12.7, "grad_norm": 6.96875, "learning_rate": 1.9510565162951538e-05, "loss": 1.353, "step": 5080 }, { "epoch": 12.7025, "grad_norm": 7.46875, "learning_rate": 1.9510314924168284e-05, "loss": 1.6791, "step": 5081 }, { "epoch": 12.705, "grad_norm": 6.78125, "learning_rate": 1.951006462303579e-05, "loss": 1.4841, "step": 5082 }, { "epoch": 12.7075, "grad_norm": 7.28125, "learning_rate": 1.950981425955569e-05, "loss": 1.4017, "step": 5083 }, { "epoch": 12.71, "grad_norm": 7.71875, "learning_rate": 1.950956383372963e-05, "loss": 1.6599, "step": 5084 }, { "epoch": 12.7125, "grad_norm": 5.1875, "learning_rate": 1.9509313345559246e-05, "loss": 1.359, "step": 5085 }, { "epoch": 12.715, "grad_norm": 8.375, "learning_rate": 1.950906279504619e-05, "loss": 1.4424, "step": 5086 }, { "epoch": 12.7175, "grad_norm": 6.09375, "learning_rate": 1.9508812182192092e-05, "loss": 1.5649, "step": 5087 }, { "epoch": 12.72, "grad_norm": 6.4375, "learning_rate": 1.9508561506998613e-05, "loss": 1.4533, "step": 5088 }, { "epoch": 12.7225, "grad_norm": 8.25, "learning_rate": 1.9508310769467377e-05, "loss": 1.4853, "step": 5089 }, { "epoch": 12.725, "grad_norm": 7.75, "learning_rate": 1.9508059969600036e-05, "loss": 1.4462, "step": 5090 }, { "epoch": 12.7275, "grad_norm": 9.9375, "learning_rate": 1.9507809107398236e-05, "loss": 1.6906, "step": 5091 }, { "epoch": 12.73, "grad_norm": 10.1875, "learning_rate": 1.950755818286362e-05, "loss": 1.5437, "step": 5092 }, { "epoch": 12.7325, "grad_norm": 8.25, "learning_rate": 1.950730719599783e-05, "loss": 1.6537, "step": 5093 }, { "epoch": 12.735, "grad_norm": 7.53125, "learning_rate": 1.950705614680252e-05, "loss": 1.509, "step": 5094 }, { "epoch": 12.7375, "grad_norm": 6.25, "learning_rate": 1.9506805035279327e-05, "loss": 1.3437, "step": 5095 }, { "epoch": 12.74, "grad_norm": 8.5, "learning_rate": 1.95065538614299e-05, "loss": 1.5843, "step": 5096 }, { "epoch": 12.7425, "grad_norm": 7.46875, "learning_rate": 1.9506302625255888e-05, "loss": 1.4142, "step": 5097 }, { "epoch": 12.745, "grad_norm": 7.5, "learning_rate": 1.9506051326758933e-05, "loss": 1.3475, "step": 5098 }, { "epoch": 12.7475, "grad_norm": 9.8125, "learning_rate": 1.950579996594069e-05, "loss": 1.5794, "step": 5099 }, { "epoch": 12.75, "grad_norm": 6.03125, "learning_rate": 1.9505548542802805e-05, "loss": 1.5181, "step": 5100 }, { "epoch": 12.7525, "grad_norm": 4.625, "learning_rate": 1.950529705734692e-05, "loss": 1.3787, "step": 5101 }, { "epoch": 12.755, "grad_norm": 6.6875, "learning_rate": 1.9505045509574686e-05, "loss": 1.3701, "step": 5102 }, { "epoch": 12.7575, "grad_norm": 7.375, "learning_rate": 1.9504793899487757e-05, "loss": 1.2841, "step": 5103 }, { "epoch": 12.76, "grad_norm": 8.0, "learning_rate": 1.950454222708778e-05, "loss": 1.5107, "step": 5104 }, { "epoch": 12.7625, "grad_norm": 10.5625, "learning_rate": 1.9504290492376405e-05, "loss": 1.3883, "step": 5105 }, { "epoch": 12.765, "grad_norm": 6.40625, "learning_rate": 1.9504038695355278e-05, "loss": 1.5372, "step": 5106 }, { "epoch": 12.7675, "grad_norm": 6.15625, "learning_rate": 1.950378683602606e-05, "loss": 1.6102, "step": 5107 }, { "epoch": 12.77, "grad_norm": 6.4375, "learning_rate": 1.9503534914390393e-05, "loss": 1.5119, "step": 5108 }, { "epoch": 12.7725, "grad_norm": 14.9375, "learning_rate": 1.950328293044993e-05, "loss": 1.456, "step": 5109 }, { "epoch": 12.775, "grad_norm": 13.125, "learning_rate": 1.950303088420632e-05, "loss": 1.3957, "step": 5110 }, { "epoch": 12.7775, "grad_norm": 7.75, "learning_rate": 1.9502778775661228e-05, "loss": 1.676, "step": 5111 }, { "epoch": 12.78, "grad_norm": 5.40625, "learning_rate": 1.9502526604816293e-05, "loss": 1.4062, "step": 5112 }, { "epoch": 12.7825, "grad_norm": 5.5, "learning_rate": 1.950227437167318e-05, "loss": 1.4545, "step": 5113 }, { "epoch": 12.785, "grad_norm": 4.78125, "learning_rate": 1.9502022076233532e-05, "loss": 1.4187, "step": 5114 }, { "epoch": 12.7875, "grad_norm": 6.46875, "learning_rate": 1.9501769718499006e-05, "loss": 1.49, "step": 5115 }, { "epoch": 12.79, "grad_norm": 5.8125, "learning_rate": 1.950151729847126e-05, "loss": 1.5196, "step": 5116 }, { "epoch": 12.7925, "grad_norm": 4.84375, "learning_rate": 1.950126481615195e-05, "loss": 1.3443, "step": 5117 }, { "epoch": 12.795, "grad_norm": 9.4375, "learning_rate": 1.950101227154272e-05, "loss": 1.6987, "step": 5118 }, { "epoch": 12.7975, "grad_norm": 4.8125, "learning_rate": 1.950075966464524e-05, "loss": 1.3526, "step": 5119 }, { "epoch": 12.8, "grad_norm": 6.9375, "learning_rate": 1.950050699546116e-05, "loss": 1.3583, "step": 5120 }, { "epoch": 12.8025, "grad_norm": 20.125, "learning_rate": 1.9500254263992134e-05, "loss": 1.7173, "step": 5121 }, { "epoch": 12.805, "grad_norm": 5.03125, "learning_rate": 1.9500001470239823e-05, "loss": 1.4883, "step": 5122 }, { "epoch": 12.8075, "grad_norm": 7.9375, "learning_rate": 1.949974861420588e-05, "loss": 1.6003, "step": 5123 }, { "epoch": 12.81, "grad_norm": 6.625, "learning_rate": 1.949949569589197e-05, "loss": 1.5625, "step": 5124 }, { "epoch": 12.8125, "grad_norm": 6.84375, "learning_rate": 1.9499242715299743e-05, "loss": 1.4679, "step": 5125 }, { "epoch": 12.815, "grad_norm": 4.96875, "learning_rate": 1.949898967243086e-05, "loss": 1.2068, "step": 5126 }, { "epoch": 12.8175, "grad_norm": 7.625, "learning_rate": 1.9498736567286983e-05, "loss": 1.56, "step": 5127 }, { "epoch": 12.82, "grad_norm": 6.21875, "learning_rate": 1.9498483399869767e-05, "loss": 1.4504, "step": 5128 }, { "epoch": 12.8225, "grad_norm": 8.5625, "learning_rate": 1.9498230170180877e-05, "loss": 1.4844, "step": 5129 }, { "epoch": 12.825, "grad_norm": 6.21875, "learning_rate": 1.949797687822197e-05, "loss": 1.4897, "step": 5130 }, { "epoch": 12.8275, "grad_norm": 9.25, "learning_rate": 1.9497723523994704e-05, "loss": 1.583, "step": 5131 }, { "epoch": 12.83, "grad_norm": 6.625, "learning_rate": 1.949747010750074e-05, "loss": 1.4218, "step": 5132 }, { "epoch": 12.8325, "grad_norm": 8.4375, "learning_rate": 1.9497216628741746e-05, "loss": 1.5329, "step": 5133 }, { "epoch": 12.835, "grad_norm": 5.75, "learning_rate": 1.949696308771938e-05, "loss": 1.3558, "step": 5134 }, { "epoch": 12.8375, "grad_norm": 6.9375, "learning_rate": 1.9496709484435304e-05, "loss": 1.5838, "step": 5135 }, { "epoch": 12.84, "grad_norm": 9.4375, "learning_rate": 1.949645581889118e-05, "loss": 1.6307, "step": 5136 }, { "epoch": 12.8425, "grad_norm": 5.75, "learning_rate": 1.9496202091088673e-05, "loss": 1.4408, "step": 5137 }, { "epoch": 12.845, "grad_norm": 7.28125, "learning_rate": 1.9495948301029443e-05, "loss": 1.5044, "step": 5138 }, { "epoch": 12.8475, "grad_norm": 7.5, "learning_rate": 1.9495694448715156e-05, "loss": 1.5, "step": 5139 }, { "epoch": 12.85, "grad_norm": 6.46875, "learning_rate": 1.949544053414748e-05, "loss": 1.7015, "step": 5140 }, { "epoch": 12.8525, "grad_norm": 6.625, "learning_rate": 1.9495186557328073e-05, "loss": 1.491, "step": 5141 }, { "epoch": 12.855, "grad_norm": 8.125, "learning_rate": 1.9494932518258602e-05, "loss": 1.5505, "step": 5142 }, { "epoch": 12.8575, "grad_norm": 6.46875, "learning_rate": 1.9494678416940735e-05, "loss": 1.3131, "step": 5143 }, { "epoch": 12.86, "grad_norm": 5.375, "learning_rate": 1.9494424253376133e-05, "loss": 1.3595, "step": 5144 }, { "epoch": 12.8625, "grad_norm": 4.40625, "learning_rate": 1.949417002756647e-05, "loss": 1.3635, "step": 5145 }, { "epoch": 12.865, "grad_norm": 4.625, "learning_rate": 1.9493915739513408e-05, "loss": 1.374, "step": 5146 }, { "epoch": 12.8675, "grad_norm": 6.28125, "learning_rate": 1.949366138921861e-05, "loss": 1.5385, "step": 5147 }, { "epoch": 12.87, "grad_norm": 7.0, "learning_rate": 1.9493406976683752e-05, "loss": 1.574, "step": 5148 }, { "epoch": 12.8725, "grad_norm": 11.6875, "learning_rate": 1.9493152501910495e-05, "loss": 1.4474, "step": 5149 }, { "epoch": 12.875, "grad_norm": 6.40625, "learning_rate": 1.9492897964900512e-05, "loss": 1.3594, "step": 5150 }, { "epoch": 12.8775, "grad_norm": 7.53125, "learning_rate": 1.949264336565547e-05, "loss": 1.4519, "step": 5151 }, { "epoch": 12.88, "grad_norm": 4.5, "learning_rate": 1.9492388704177036e-05, "loss": 1.3687, "step": 5152 }, { "epoch": 12.8825, "grad_norm": 9.5, "learning_rate": 1.949213398046688e-05, "loss": 1.4337, "step": 5153 }, { "epoch": 12.885, "grad_norm": 5.21875, "learning_rate": 1.9491879194526678e-05, "loss": 1.5056, "step": 5154 }, { "epoch": 12.8875, "grad_norm": 7.53125, "learning_rate": 1.9491624346358095e-05, "loss": 1.3879, "step": 5155 }, { "epoch": 12.89, "grad_norm": 6.96875, "learning_rate": 1.9491369435962802e-05, "loss": 1.4289, "step": 5156 }, { "epoch": 12.8925, "grad_norm": 6.53125, "learning_rate": 1.9491114463342467e-05, "loss": 1.4742, "step": 5157 }, { "epoch": 12.895, "grad_norm": 8.125, "learning_rate": 1.9490859428498773e-05, "loss": 1.3774, "step": 5158 }, { "epoch": 12.8975, "grad_norm": 7.3125, "learning_rate": 1.9490604331433378e-05, "loss": 1.5854, "step": 5159 }, { "epoch": 12.9, "grad_norm": 16.125, "learning_rate": 1.9490349172147964e-05, "loss": 1.4218, "step": 5160 }, { "epoch": 12.9025, "grad_norm": 8.5, "learning_rate": 1.94900939506442e-05, "loss": 1.7124, "step": 5161 }, { "epoch": 12.905, "grad_norm": 7.0, "learning_rate": 1.948983866692376e-05, "loss": 1.5523, "step": 5162 }, { "epoch": 12.9075, "grad_norm": 7.375, "learning_rate": 1.9489583320988313e-05, "loss": 1.4317, "step": 5163 }, { "epoch": 12.91, "grad_norm": 7.21875, "learning_rate": 1.948932791283954e-05, "loss": 1.5758, "step": 5164 }, { "epoch": 12.9125, "grad_norm": 7.0625, "learning_rate": 1.948907244247912e-05, "loss": 1.6052, "step": 5165 }, { "epoch": 12.915, "grad_norm": 6.6875, "learning_rate": 1.9488816909908714e-05, "loss": 1.4415, "step": 5166 }, { "epoch": 12.9175, "grad_norm": 6.5625, "learning_rate": 1.9488561315130004e-05, "loss": 1.4132, "step": 5167 }, { "epoch": 12.92, "grad_norm": 7.5, "learning_rate": 1.9488305658144666e-05, "loss": 1.4686, "step": 5168 }, { "epoch": 12.9225, "grad_norm": 7.53125, "learning_rate": 1.9488049938954377e-05, "loss": 1.2728, "step": 5169 }, { "epoch": 12.925, "grad_norm": 10.0, "learning_rate": 1.9487794157560814e-05, "loss": 1.4951, "step": 5170 }, { "epoch": 12.9275, "grad_norm": 7.4375, "learning_rate": 1.9487538313965647e-05, "loss": 1.5081, "step": 5171 }, { "epoch": 12.93, "grad_norm": 10.75, "learning_rate": 1.9487282408170564e-05, "loss": 1.4725, "step": 5172 }, { "epoch": 12.9325, "grad_norm": 5.625, "learning_rate": 1.9487026440177235e-05, "loss": 1.494, "step": 5173 }, { "epoch": 12.935, "grad_norm": 5.71875, "learning_rate": 1.948677040998734e-05, "loss": 1.5817, "step": 5174 }, { "epoch": 12.9375, "grad_norm": 6.78125, "learning_rate": 1.9486514317602555e-05, "loss": 1.5893, "step": 5175 }, { "epoch": 12.94, "grad_norm": 5.8125, "learning_rate": 1.9486258163024567e-05, "loss": 1.3113, "step": 5176 }, { "epoch": 12.9425, "grad_norm": 12.3125, "learning_rate": 1.9486001946255046e-05, "loss": 1.5049, "step": 5177 }, { "epoch": 12.945, "grad_norm": 8.4375, "learning_rate": 1.948574566729568e-05, "loss": 1.5515, "step": 5178 }, { "epoch": 12.9475, "grad_norm": 6.0625, "learning_rate": 1.948548932614814e-05, "loss": 1.6066, "step": 5179 }, { "epoch": 12.95, "grad_norm": 5.1875, "learning_rate": 1.9485232922814117e-05, "loss": 1.3629, "step": 5180 }, { "epoch": 12.9525, "grad_norm": 5.15625, "learning_rate": 1.9484976457295285e-05, "loss": 1.3169, "step": 5181 }, { "epoch": 12.955, "grad_norm": 8.5625, "learning_rate": 1.9484719929593322e-05, "loss": 1.6306, "step": 5182 }, { "epoch": 12.9575, "grad_norm": 8.25, "learning_rate": 1.948446333970992e-05, "loss": 1.4879, "step": 5183 }, { "epoch": 12.96, "grad_norm": 5.75, "learning_rate": 1.9484206687646753e-05, "loss": 1.5174, "step": 5184 }, { "epoch": 12.9625, "grad_norm": 5.1875, "learning_rate": 1.948394997340551e-05, "loss": 1.326, "step": 5185 }, { "epoch": 12.965, "grad_norm": 6.78125, "learning_rate": 1.9483693196987868e-05, "loss": 1.5402, "step": 5186 }, { "epoch": 12.9675, "grad_norm": 6.75, "learning_rate": 1.9483436358395512e-05, "loss": 1.4483, "step": 5187 }, { "epoch": 12.97, "grad_norm": 7.875, "learning_rate": 1.948317945763013e-05, "loss": 1.3645, "step": 5188 }, { "epoch": 12.9725, "grad_norm": 4.96875, "learning_rate": 1.94829224946934e-05, "loss": 1.6069, "step": 5189 }, { "epoch": 12.975, "grad_norm": 6.1875, "learning_rate": 1.948266546958701e-05, "loss": 1.5428, "step": 5190 }, { "epoch": 12.9775, "grad_norm": 7.0, "learning_rate": 1.9482408382312644e-05, "loss": 1.4946, "step": 5191 }, { "epoch": 12.98, "grad_norm": 9.6875, "learning_rate": 1.948215123287199e-05, "loss": 1.705, "step": 5192 }, { "epoch": 12.9825, "grad_norm": 5.75, "learning_rate": 1.948189402126673e-05, "loss": 1.5449, "step": 5193 }, { "epoch": 12.985, "grad_norm": 8.5, "learning_rate": 1.9481636747498555e-05, "loss": 1.4359, "step": 5194 }, { "epoch": 12.9875, "grad_norm": 5.6875, "learning_rate": 1.9481379411569147e-05, "loss": 1.6774, "step": 5195 }, { "epoch": 12.99, "grad_norm": 5.625, "learning_rate": 1.94811220134802e-05, "loss": 1.5518, "step": 5196 }, { "epoch": 12.9925, "grad_norm": 7.15625, "learning_rate": 1.9480864553233386e-05, "loss": 1.3013, "step": 5197 }, { "epoch": 12.995, "grad_norm": 6.34375, "learning_rate": 1.948060703083041e-05, "loss": 1.482, "step": 5198 }, { "epoch": 12.9975, "grad_norm": 8.9375, "learning_rate": 1.9480349446272953e-05, "loss": 1.4388, "step": 5199 }, { "epoch": 13.0, "grad_norm": 6.0, "learning_rate": 1.9480091799562706e-05, "loss": 1.444, "step": 5200 }, { "epoch": 13.0025, "grad_norm": 5.53125, "learning_rate": 1.9479834090701354e-05, "loss": 1.4949, "step": 5201 }, { "epoch": 13.005, "grad_norm": 7.375, "learning_rate": 1.947957631969059e-05, "loss": 1.3859, "step": 5202 }, { "epoch": 13.0075, "grad_norm": 7.125, "learning_rate": 1.9479318486532104e-05, "loss": 1.5208, "step": 5203 }, { "epoch": 13.01, "grad_norm": 7.4375, "learning_rate": 1.9479060591227586e-05, "loss": 1.5068, "step": 5204 }, { "epoch": 13.0125, "grad_norm": 6.40625, "learning_rate": 1.947880263377872e-05, "loss": 1.3695, "step": 5205 }, { "epoch": 13.015, "grad_norm": 7.1875, "learning_rate": 1.9478544614187212e-05, "loss": 1.4285, "step": 5206 }, { "epoch": 13.0175, "grad_norm": 5.625, "learning_rate": 1.9478286532454737e-05, "loss": 1.3699, "step": 5207 }, { "epoch": 13.02, "grad_norm": 5.5625, "learning_rate": 1.9478028388583e-05, "loss": 1.5227, "step": 5208 }, { "epoch": 13.0225, "grad_norm": 6.3125, "learning_rate": 1.947777018257369e-05, "loss": 1.6093, "step": 5209 }, { "epoch": 13.025, "grad_norm": 6.46875, "learning_rate": 1.947751191442849e-05, "loss": 1.2917, "step": 5210 }, { "epoch": 13.0275, "grad_norm": 7.40625, "learning_rate": 1.947725358414911e-05, "loss": 1.4263, "step": 5211 }, { "epoch": 13.03, "grad_norm": 6.90625, "learning_rate": 1.947699519173723e-05, "loss": 1.4984, "step": 5212 }, { "epoch": 13.0325, "grad_norm": 4.40625, "learning_rate": 1.9476736737194547e-05, "loss": 1.5314, "step": 5213 }, { "epoch": 13.035, "grad_norm": 8.0, "learning_rate": 1.947647822052276e-05, "loss": 1.7036, "step": 5214 }, { "epoch": 13.0375, "grad_norm": 4.875, "learning_rate": 1.947621964172356e-05, "loss": 1.4581, "step": 5215 }, { "epoch": 13.04, "grad_norm": 7.46875, "learning_rate": 1.9475961000798645e-05, "loss": 1.6088, "step": 5216 }, { "epoch": 13.0425, "grad_norm": 6.03125, "learning_rate": 1.9475702297749705e-05, "loss": 1.3967, "step": 5217 }, { "epoch": 13.045, "grad_norm": 7.1875, "learning_rate": 1.947544353257844e-05, "loss": 1.4666, "step": 5218 }, { "epoch": 13.0475, "grad_norm": 8.3125, "learning_rate": 1.9475184705286553e-05, "loss": 1.5577, "step": 5219 }, { "epoch": 13.05, "grad_norm": 5.65625, "learning_rate": 1.947492581587573e-05, "loss": 1.5614, "step": 5220 }, { "epoch": 13.0525, "grad_norm": 5.71875, "learning_rate": 1.947466686434767e-05, "loss": 1.4417, "step": 5221 }, { "epoch": 13.055, "grad_norm": 6.4375, "learning_rate": 1.9474407850704076e-05, "loss": 1.366, "step": 5222 }, { "epoch": 13.0575, "grad_norm": 6.4375, "learning_rate": 1.9474148774946644e-05, "loss": 1.4701, "step": 5223 }, { "epoch": 13.06, "grad_norm": 5.40625, "learning_rate": 1.947388963707707e-05, "loss": 1.3639, "step": 5224 }, { "epoch": 13.0625, "grad_norm": 8.1875, "learning_rate": 1.9473630437097056e-05, "loss": 1.4768, "step": 5225 }, { "epoch": 13.065, "grad_norm": 6.0625, "learning_rate": 1.94733711750083e-05, "loss": 1.2532, "step": 5226 }, { "epoch": 13.0675, "grad_norm": 10.125, "learning_rate": 1.94731118508125e-05, "loss": 1.5804, "step": 5227 }, { "epoch": 13.07, "grad_norm": 6.0625, "learning_rate": 1.947285246451136e-05, "loss": 1.5985, "step": 5228 }, { "epoch": 13.0725, "grad_norm": 5.03125, "learning_rate": 1.9472593016106577e-05, "loss": 1.5073, "step": 5229 }, { "epoch": 13.075, "grad_norm": 6.65625, "learning_rate": 1.9472333505599855e-05, "loss": 1.5776, "step": 5230 }, { "epoch": 13.0775, "grad_norm": 9.4375, "learning_rate": 1.947207393299289e-05, "loss": 1.5994, "step": 5231 }, { "epoch": 13.08, "grad_norm": 5.4375, "learning_rate": 1.947181429828739e-05, "loss": 1.541, "step": 5232 }, { "epoch": 13.0825, "grad_norm": 7.78125, "learning_rate": 1.9471554601485054e-05, "loss": 1.4886, "step": 5233 }, { "epoch": 13.085, "grad_norm": 6.46875, "learning_rate": 1.9471294842587584e-05, "loss": 1.4183, "step": 5234 }, { "epoch": 13.0875, "grad_norm": 6.25, "learning_rate": 1.9471035021596683e-05, "loss": 1.3383, "step": 5235 }, { "epoch": 13.09, "grad_norm": 6.90625, "learning_rate": 1.9470775138514063e-05, "loss": 1.5527, "step": 5236 }, { "epoch": 13.0925, "grad_norm": 8.5625, "learning_rate": 1.9470515193341413e-05, "loss": 1.4274, "step": 5237 }, { "epoch": 13.095, "grad_norm": 7.3125, "learning_rate": 1.9470255186080446e-05, "loss": 1.46, "step": 5238 }, { "epoch": 13.0975, "grad_norm": 6.6875, "learning_rate": 1.9469995116732864e-05, "loss": 1.5938, "step": 5239 }, { "epoch": 13.1, "grad_norm": 6.09375, "learning_rate": 1.9469734985300373e-05, "loss": 1.5202, "step": 5240 }, { "epoch": 13.1025, "grad_norm": 5.90625, "learning_rate": 1.946947479178468e-05, "loss": 1.4756, "step": 5241 }, { "epoch": 13.105, "grad_norm": 5.65625, "learning_rate": 1.946921453618749e-05, "loss": 1.3636, "step": 5242 }, { "epoch": 13.1075, "grad_norm": 5.59375, "learning_rate": 1.9468954218510506e-05, "loss": 1.4959, "step": 5243 }, { "epoch": 13.11, "grad_norm": 4.6875, "learning_rate": 1.946869383875544e-05, "loss": 1.4701, "step": 5244 }, { "epoch": 13.1125, "grad_norm": 9.0625, "learning_rate": 1.946843339692399e-05, "loss": 1.5745, "step": 5245 }, { "epoch": 13.115, "grad_norm": 6.15625, "learning_rate": 1.9468172893017876e-05, "loss": 1.3402, "step": 5246 }, { "epoch": 13.1175, "grad_norm": 7.65625, "learning_rate": 1.9467912327038797e-05, "loss": 1.4988, "step": 5247 }, { "epoch": 13.12, "grad_norm": 8.5, "learning_rate": 1.9467651698988464e-05, "loss": 1.6245, "step": 5248 }, { "epoch": 13.1225, "grad_norm": 5.9375, "learning_rate": 1.9467391008868585e-05, "loss": 1.4263, "step": 5249 }, { "epoch": 13.125, "grad_norm": 5.34375, "learning_rate": 1.9467130256680867e-05, "loss": 1.4373, "step": 5250 }, { "epoch": 13.1275, "grad_norm": 6.5625, "learning_rate": 1.9466869442427026e-05, "loss": 1.3665, "step": 5251 }, { "epoch": 13.13, "grad_norm": 7.09375, "learning_rate": 1.9466608566108768e-05, "loss": 1.5553, "step": 5252 }, { "epoch": 13.1325, "grad_norm": 7.75, "learning_rate": 1.94663476277278e-05, "loss": 1.4344, "step": 5253 }, { "epoch": 13.135, "grad_norm": 5.34375, "learning_rate": 1.9466086627285836e-05, "loss": 1.5554, "step": 5254 }, { "epoch": 13.1375, "grad_norm": 7.46875, "learning_rate": 1.946582556478459e-05, "loss": 1.4387, "step": 5255 }, { "epoch": 13.14, "grad_norm": 8.125, "learning_rate": 1.9465564440225768e-05, "loss": 1.5511, "step": 5256 }, { "epoch": 13.1425, "grad_norm": 5.0, "learning_rate": 1.9465303253611086e-05, "loss": 1.4969, "step": 5257 }, { "epoch": 13.145, "grad_norm": 5.6875, "learning_rate": 1.9465042004942253e-05, "loss": 1.3807, "step": 5258 }, { "epoch": 13.1475, "grad_norm": 7.4375, "learning_rate": 1.9464780694220982e-05, "loss": 1.5988, "step": 5259 }, { "epoch": 13.15, "grad_norm": 6.46875, "learning_rate": 1.9464519321448988e-05, "loss": 1.4675, "step": 5260 }, { "epoch": 13.1525, "grad_norm": 6.53125, "learning_rate": 1.946425788662799e-05, "loss": 1.5778, "step": 5261 }, { "epoch": 13.155, "grad_norm": 6.53125, "learning_rate": 1.9463996389759688e-05, "loss": 1.4885, "step": 5262 }, { "epoch": 13.1575, "grad_norm": 5.5625, "learning_rate": 1.946373483084581e-05, "loss": 1.2785, "step": 5263 }, { "epoch": 13.16, "grad_norm": 4.46875, "learning_rate": 1.9463473209888063e-05, "loss": 1.5266, "step": 5264 }, { "epoch": 13.1625, "grad_norm": 5.34375, "learning_rate": 1.946321152688816e-05, "loss": 1.468, "step": 5265 }, { "epoch": 13.165, "grad_norm": 6.59375, "learning_rate": 1.9462949781847826e-05, "loss": 1.4749, "step": 5266 }, { "epoch": 13.1675, "grad_norm": 8.0625, "learning_rate": 1.9462687974768772e-05, "loss": 1.3942, "step": 5267 }, { "epoch": 13.17, "grad_norm": 6.34375, "learning_rate": 1.9462426105652713e-05, "loss": 1.5023, "step": 5268 }, { "epoch": 13.1725, "grad_norm": 5.375, "learning_rate": 1.946216417450137e-05, "loss": 1.5036, "step": 5269 }, { "epoch": 13.175, "grad_norm": 6.0625, "learning_rate": 1.9461902181316455e-05, "loss": 1.4081, "step": 5270 }, { "epoch": 13.1775, "grad_norm": 5.125, "learning_rate": 1.946164012609969e-05, "loss": 1.3905, "step": 5271 }, { "epoch": 13.18, "grad_norm": 11.0, "learning_rate": 1.9461378008852785e-05, "loss": 1.7027, "step": 5272 }, { "epoch": 13.1825, "grad_norm": 7.28125, "learning_rate": 1.946111582957747e-05, "loss": 1.5912, "step": 5273 }, { "epoch": 13.185, "grad_norm": 8.25, "learning_rate": 1.9460853588275454e-05, "loss": 1.5722, "step": 5274 }, { "epoch": 13.1875, "grad_norm": 4.53125, "learning_rate": 1.9460591284948463e-05, "loss": 1.3893, "step": 5275 }, { "epoch": 13.19, "grad_norm": 6.71875, "learning_rate": 1.9460328919598216e-05, "loss": 1.5556, "step": 5276 }, { "epoch": 13.1925, "grad_norm": 5.90625, "learning_rate": 1.9460066492226428e-05, "loss": 1.5262, "step": 5277 }, { "epoch": 13.195, "grad_norm": 8.625, "learning_rate": 1.9459804002834823e-05, "loss": 1.3484, "step": 5278 }, { "epoch": 13.1975, "grad_norm": 8.0625, "learning_rate": 1.9459541451425123e-05, "loss": 1.7256, "step": 5279 }, { "epoch": 13.2, "grad_norm": 23.25, "learning_rate": 1.9459278837999048e-05, "loss": 1.458, "step": 5280 }, { "epoch": 13.2025, "grad_norm": 7.3125, "learning_rate": 1.9459016162558316e-05, "loss": 1.4189, "step": 5281 }, { "epoch": 13.205, "grad_norm": 5.78125, "learning_rate": 1.9458753425104656e-05, "loss": 1.5425, "step": 5282 }, { "epoch": 13.2075, "grad_norm": 6.71875, "learning_rate": 1.9458490625639788e-05, "loss": 1.4371, "step": 5283 }, { "epoch": 13.21, "grad_norm": 6.71875, "learning_rate": 1.945822776416543e-05, "loss": 1.4368, "step": 5284 }, { "epoch": 13.2125, "grad_norm": 9.625, "learning_rate": 1.9457964840683314e-05, "loss": 1.3796, "step": 5285 }, { "epoch": 13.215, "grad_norm": 6.53125, "learning_rate": 1.9457701855195154e-05, "loss": 1.456, "step": 5286 }, { "epoch": 13.2175, "grad_norm": 7.90625, "learning_rate": 1.9457438807702682e-05, "loss": 1.7701, "step": 5287 }, { "epoch": 13.22, "grad_norm": 5.46875, "learning_rate": 1.945717569820762e-05, "loss": 1.5427, "step": 5288 }, { "epoch": 13.2225, "grad_norm": 6.3125, "learning_rate": 1.945691252671169e-05, "loss": 1.5239, "step": 5289 }, { "epoch": 13.225, "grad_norm": 6.125, "learning_rate": 1.9456649293216623e-05, "loss": 1.398, "step": 5290 }, { "epoch": 13.2275, "grad_norm": 5.28125, "learning_rate": 1.9456385997724142e-05, "loss": 1.517, "step": 5291 }, { "epoch": 13.23, "grad_norm": 6.03125, "learning_rate": 1.945612264023597e-05, "loss": 1.4855, "step": 5292 }, { "epoch": 13.2325, "grad_norm": 11.125, "learning_rate": 1.9455859220753838e-05, "loss": 1.4963, "step": 5293 }, { "epoch": 13.235, "grad_norm": 6.09375, "learning_rate": 1.945559573927947e-05, "loss": 1.5111, "step": 5294 }, { "epoch": 13.2375, "grad_norm": 6.125, "learning_rate": 1.9455332195814597e-05, "loss": 1.3125, "step": 5295 }, { "epoch": 13.24, "grad_norm": 5.3125, "learning_rate": 1.9455068590360943e-05, "loss": 1.3925, "step": 5296 }, { "epoch": 13.2425, "grad_norm": 9.4375, "learning_rate": 1.9454804922920235e-05, "loss": 1.5046, "step": 5297 }, { "epoch": 13.245, "grad_norm": 5.84375, "learning_rate": 1.945454119349421e-05, "loss": 1.3676, "step": 5298 }, { "epoch": 13.2475, "grad_norm": 6.125, "learning_rate": 1.9454277402084587e-05, "loss": 1.3916, "step": 5299 }, { "epoch": 13.25, "grad_norm": 11.875, "learning_rate": 1.9454013548693103e-05, "loss": 1.6141, "step": 5300 }, { "epoch": 13.2525, "grad_norm": 4.9375, "learning_rate": 1.945374963332148e-05, "loss": 1.4277, "step": 5301 }, { "epoch": 13.255, "grad_norm": 11.3125, "learning_rate": 1.9453485655971457e-05, "loss": 1.4988, "step": 5302 }, { "epoch": 13.2575, "grad_norm": 6.78125, "learning_rate": 1.945322161664476e-05, "loss": 1.4004, "step": 5303 }, { "epoch": 13.26, "grad_norm": 7.53125, "learning_rate": 1.945295751534312e-05, "loss": 1.4095, "step": 5304 }, { "epoch": 13.2625, "grad_norm": 9.375, "learning_rate": 1.9452693352068267e-05, "loss": 1.5175, "step": 5305 }, { "epoch": 13.265, "grad_norm": 8.0, "learning_rate": 1.9452429126821933e-05, "loss": 1.4639, "step": 5306 }, { "epoch": 13.2675, "grad_norm": 11.0, "learning_rate": 1.9452164839605855e-05, "loss": 1.5734, "step": 5307 }, { "epoch": 13.27, "grad_norm": 5.1875, "learning_rate": 1.9451900490421766e-05, "loss": 1.3991, "step": 5308 }, { "epoch": 13.2725, "grad_norm": 7.875, "learning_rate": 1.945163607927139e-05, "loss": 1.4971, "step": 5309 }, { "epoch": 13.275, "grad_norm": 5.53125, "learning_rate": 1.9451371606156468e-05, "loss": 1.5672, "step": 5310 }, { "epoch": 13.2775, "grad_norm": 9.5625, "learning_rate": 1.945110707107873e-05, "loss": 1.4984, "step": 5311 }, { "epoch": 13.28, "grad_norm": 7.21875, "learning_rate": 1.9450842474039914e-05, "loss": 1.5098, "step": 5312 }, { "epoch": 13.2825, "grad_norm": 5.65625, "learning_rate": 1.9450577815041753e-05, "loss": 1.4855, "step": 5313 }, { "epoch": 13.285, "grad_norm": 6.78125, "learning_rate": 1.945031309408598e-05, "loss": 1.4695, "step": 5314 }, { "epoch": 13.2875, "grad_norm": 5.90625, "learning_rate": 1.9450048311174338e-05, "loss": 1.524, "step": 5315 }, { "epoch": 13.29, "grad_norm": 7.5, "learning_rate": 1.9449783466308553e-05, "loss": 1.3735, "step": 5316 }, { "epoch": 13.2925, "grad_norm": 6.625, "learning_rate": 1.9449518559490368e-05, "loss": 1.5233, "step": 5317 }, { "epoch": 13.295, "grad_norm": 10.375, "learning_rate": 1.9449253590721514e-05, "loss": 1.3296, "step": 5318 }, { "epoch": 13.2975, "grad_norm": 5.125, "learning_rate": 1.9448988560003734e-05, "loss": 1.5846, "step": 5319 }, { "epoch": 13.3, "grad_norm": 5.25, "learning_rate": 1.9448723467338765e-05, "loss": 1.6346, "step": 5320 }, { "epoch": 13.3025, "grad_norm": 6.96875, "learning_rate": 1.944845831272834e-05, "loss": 1.4138, "step": 5321 }, { "epoch": 13.305, "grad_norm": 6.15625, "learning_rate": 1.9448193096174203e-05, "loss": 1.4778, "step": 5322 }, { "epoch": 13.3075, "grad_norm": 5.59375, "learning_rate": 1.9447927817678086e-05, "loss": 1.5514, "step": 5323 }, { "epoch": 13.31, "grad_norm": 4.4375, "learning_rate": 1.9447662477241737e-05, "loss": 1.4124, "step": 5324 }, { "epoch": 13.3125, "grad_norm": 4.78125, "learning_rate": 1.944739707486689e-05, "loss": 1.4435, "step": 5325 }, { "epoch": 13.315, "grad_norm": 7.375, "learning_rate": 1.944713161055528e-05, "loss": 1.4966, "step": 5326 }, { "epoch": 13.3175, "grad_norm": 6.71875, "learning_rate": 1.9446866084308657e-05, "loss": 1.5257, "step": 5327 }, { "epoch": 13.32, "grad_norm": 6.78125, "learning_rate": 1.944660049612876e-05, "loss": 1.5035, "step": 5328 }, { "epoch": 13.3225, "grad_norm": 4.46875, "learning_rate": 1.9446334846017328e-05, "loss": 1.3458, "step": 5329 }, { "epoch": 13.325, "grad_norm": 5.40625, "learning_rate": 1.94460691339761e-05, "loss": 1.317, "step": 5330 }, { "epoch": 13.3275, "grad_norm": 4.96875, "learning_rate": 1.9445803360006822e-05, "loss": 1.3672, "step": 5331 }, { "epoch": 13.33, "grad_norm": 7.625, "learning_rate": 1.9445537524111238e-05, "loss": 1.449, "step": 5332 }, { "epoch": 13.3325, "grad_norm": 7.375, "learning_rate": 1.9445271626291085e-05, "loss": 1.4319, "step": 5333 }, { "epoch": 13.335, "grad_norm": 6.9375, "learning_rate": 1.944500566654811e-05, "loss": 1.3643, "step": 5334 }, { "epoch": 13.3375, "grad_norm": 6.96875, "learning_rate": 1.9444739644884056e-05, "loss": 1.551, "step": 5335 }, { "epoch": 13.34, "grad_norm": 7.1875, "learning_rate": 1.9444473561300666e-05, "loss": 1.6379, "step": 5336 }, { "epoch": 13.3425, "grad_norm": 6.59375, "learning_rate": 1.9444207415799687e-05, "loss": 1.4012, "step": 5337 }, { "epoch": 13.345, "grad_norm": 6.75, "learning_rate": 1.944394120838286e-05, "loss": 1.6203, "step": 5338 }, { "epoch": 13.3475, "grad_norm": 4.875, "learning_rate": 1.9443674939051934e-05, "loss": 1.3678, "step": 5339 }, { "epoch": 13.35, "grad_norm": 6.21875, "learning_rate": 1.944340860780865e-05, "loss": 1.4689, "step": 5340 }, { "epoch": 13.3525, "grad_norm": 6.34375, "learning_rate": 1.9443142214654763e-05, "loss": 1.5156, "step": 5341 }, { "epoch": 13.355, "grad_norm": 6.875, "learning_rate": 1.9442875759592006e-05, "loss": 1.4618, "step": 5342 }, { "epoch": 13.3575, "grad_norm": 6.75, "learning_rate": 1.944260924262214e-05, "loss": 1.4111, "step": 5343 }, { "epoch": 13.36, "grad_norm": 9.6875, "learning_rate": 1.9442342663746903e-05, "loss": 1.5185, "step": 5344 }, { "epoch": 13.3625, "grad_norm": 6.65625, "learning_rate": 1.9442076022968043e-05, "loss": 1.5718, "step": 5345 }, { "epoch": 13.365, "grad_norm": 7.90625, "learning_rate": 1.9441809320287313e-05, "loss": 1.3266, "step": 5346 }, { "epoch": 13.3675, "grad_norm": 6.03125, "learning_rate": 1.944154255570646e-05, "loss": 1.4953, "step": 5347 }, { "epoch": 13.37, "grad_norm": 5.6875, "learning_rate": 1.944127572922723e-05, "loss": 1.4205, "step": 5348 }, { "epoch": 13.3725, "grad_norm": 5.90625, "learning_rate": 1.9441008840851373e-05, "loss": 1.347, "step": 5349 }, { "epoch": 13.375, "grad_norm": 6.3125, "learning_rate": 1.9440741890580643e-05, "loss": 1.4305, "step": 5350 }, { "epoch": 13.3775, "grad_norm": 9.5, "learning_rate": 1.9440474878416785e-05, "loss": 1.5384, "step": 5351 }, { "epoch": 13.38, "grad_norm": 8.1875, "learning_rate": 1.944020780436155e-05, "loss": 1.3776, "step": 5352 }, { "epoch": 13.3825, "grad_norm": 4.8125, "learning_rate": 1.9439940668416693e-05, "loss": 1.5777, "step": 5353 }, { "epoch": 13.385, "grad_norm": 6.90625, "learning_rate": 1.9439673470583962e-05, "loss": 1.5333, "step": 5354 }, { "epoch": 13.3875, "grad_norm": 7.25, "learning_rate": 1.9439406210865114e-05, "loss": 1.5879, "step": 5355 }, { "epoch": 13.39, "grad_norm": 6.65625, "learning_rate": 1.943913888926189e-05, "loss": 1.3752, "step": 5356 }, { "epoch": 13.3925, "grad_norm": 7.9375, "learning_rate": 1.943887150577605e-05, "loss": 1.5076, "step": 5357 }, { "epoch": 13.395, "grad_norm": 7.25, "learning_rate": 1.9438604060409348e-05, "loss": 1.3688, "step": 5358 }, { "epoch": 13.3975, "grad_norm": 7.125, "learning_rate": 1.9438336553163535e-05, "loss": 1.5441, "step": 5359 }, { "epoch": 13.4, "grad_norm": 5.71875, "learning_rate": 1.9438068984040366e-05, "loss": 1.4892, "step": 5360 }, { "epoch": 13.4025, "grad_norm": 6.53125, "learning_rate": 1.9437801353041596e-05, "loss": 1.498, "step": 5361 }, { "epoch": 13.405, "grad_norm": 8.0, "learning_rate": 1.9437533660168973e-05, "loss": 1.4997, "step": 5362 }, { "epoch": 13.4075, "grad_norm": 9.9375, "learning_rate": 1.943726590542426e-05, "loss": 1.4234, "step": 5363 }, { "epoch": 13.41, "grad_norm": 4.75, "learning_rate": 1.943699808880921e-05, "loss": 1.3117, "step": 5364 }, { "epoch": 13.4125, "grad_norm": 8.5, "learning_rate": 1.9436730210325577e-05, "loss": 1.5733, "step": 5365 }, { "epoch": 13.415, "grad_norm": 9.4375, "learning_rate": 1.9436462269975118e-05, "loss": 1.6403, "step": 5366 }, { "epoch": 13.4175, "grad_norm": 7.8125, "learning_rate": 1.9436194267759592e-05, "loss": 1.7082, "step": 5367 }, { "epoch": 13.42, "grad_norm": 7.34375, "learning_rate": 1.943592620368075e-05, "loss": 1.4602, "step": 5368 }, { "epoch": 13.4225, "grad_norm": 5.84375, "learning_rate": 1.9435658077740358e-05, "loss": 1.6309, "step": 5369 }, { "epoch": 13.425, "grad_norm": 6.9375, "learning_rate": 1.9435389889940167e-05, "loss": 1.5155, "step": 5370 }, { "epoch": 13.4275, "grad_norm": 7.03125, "learning_rate": 1.9435121640281936e-05, "loss": 1.5261, "step": 5371 }, { "epoch": 13.43, "grad_norm": 5.53125, "learning_rate": 1.9434853328767426e-05, "loss": 1.559, "step": 5372 }, { "epoch": 13.4325, "grad_norm": 4.53125, "learning_rate": 1.9434584955398397e-05, "loss": 1.2119, "step": 5373 }, { "epoch": 13.435, "grad_norm": 9.0, "learning_rate": 1.943431652017661e-05, "loss": 1.6344, "step": 5374 }, { "epoch": 13.4375, "grad_norm": 8.25, "learning_rate": 1.943404802310381e-05, "loss": 1.5103, "step": 5375 }, { "epoch": 13.44, "grad_norm": 8.75, "learning_rate": 1.943377946418178e-05, "loss": 1.5447, "step": 5376 }, { "epoch": 13.4425, "grad_norm": 7.84375, "learning_rate": 1.9433510843412264e-05, "loss": 1.4676, "step": 5377 }, { "epoch": 13.445, "grad_norm": 8.5, "learning_rate": 1.943324216079703e-05, "loss": 1.4063, "step": 5378 }, { "epoch": 13.4475, "grad_norm": 6.4375, "learning_rate": 1.9432973416337837e-05, "loss": 1.4529, "step": 5379 }, { "epoch": 13.45, "grad_norm": 8.125, "learning_rate": 1.9432704610036448e-05, "loss": 1.6172, "step": 5380 }, { "epoch": 13.4525, "grad_norm": 7.4375, "learning_rate": 1.9432435741894626e-05, "loss": 1.5788, "step": 5381 }, { "epoch": 13.455, "grad_norm": 7.375, "learning_rate": 1.943216681191413e-05, "loss": 1.6548, "step": 5382 }, { "epoch": 13.4575, "grad_norm": 5.59375, "learning_rate": 1.943189782009673e-05, "loss": 1.3267, "step": 5383 }, { "epoch": 13.46, "grad_norm": 5.125, "learning_rate": 1.943162876644418e-05, "loss": 1.3201, "step": 5384 }, { "epoch": 13.4625, "grad_norm": 4.8125, "learning_rate": 1.9431359650958253e-05, "loss": 1.5854, "step": 5385 }, { "epoch": 13.465, "grad_norm": 5.65625, "learning_rate": 1.943109047364071e-05, "loss": 1.4488, "step": 5386 }, { "epoch": 13.4675, "grad_norm": 7.40625, "learning_rate": 1.9430821234493314e-05, "loss": 1.3325, "step": 5387 }, { "epoch": 13.47, "grad_norm": 9.125, "learning_rate": 1.9430551933517833e-05, "loss": 1.3627, "step": 5388 }, { "epoch": 13.4725, "grad_norm": 4.78125, "learning_rate": 1.9430282570716028e-05, "loss": 1.414, "step": 5389 }, { "epoch": 13.475, "grad_norm": 38.25, "learning_rate": 1.943001314608967e-05, "loss": 1.4297, "step": 5390 }, { "epoch": 13.4775, "grad_norm": 6.59375, "learning_rate": 1.942974365964052e-05, "loss": 1.4914, "step": 5391 }, { "epoch": 13.48, "grad_norm": 5.03125, "learning_rate": 1.942947411137035e-05, "loss": 1.5411, "step": 5392 }, { "epoch": 13.4825, "grad_norm": 7.3125, "learning_rate": 1.9429204501280926e-05, "loss": 1.4731, "step": 5393 }, { "epoch": 13.485, "grad_norm": 7.03125, "learning_rate": 1.9428934829374014e-05, "loss": 1.4727, "step": 5394 }, { "epoch": 13.4875, "grad_norm": 7.21875, "learning_rate": 1.9428665095651385e-05, "loss": 1.4772, "step": 5395 }, { "epoch": 13.49, "grad_norm": 6.0625, "learning_rate": 1.9428395300114803e-05, "loss": 1.4983, "step": 5396 }, { "epoch": 13.4925, "grad_norm": 5.8125, "learning_rate": 1.9428125442766036e-05, "loss": 1.4202, "step": 5397 }, { "epoch": 13.495, "grad_norm": 5.09375, "learning_rate": 1.942785552360686e-05, "loss": 1.4269, "step": 5398 }, { "epoch": 13.4975, "grad_norm": 6.75, "learning_rate": 1.942758554263904e-05, "loss": 1.3856, "step": 5399 }, { "epoch": 13.5, "grad_norm": 6.09375, "learning_rate": 1.9427315499864345e-05, "loss": 1.3726, "step": 5400 }, { "epoch": 13.5025, "grad_norm": 6.0, "learning_rate": 1.9427045395284546e-05, "loss": 1.4028, "step": 5401 }, { "epoch": 13.505, "grad_norm": 5.21875, "learning_rate": 1.9426775228901415e-05, "loss": 1.5178, "step": 5402 }, { "epoch": 13.5075, "grad_norm": 4.84375, "learning_rate": 1.9426505000716723e-05, "loss": 1.3075, "step": 5403 }, { "epoch": 13.51, "grad_norm": 8.5625, "learning_rate": 1.9426234710732244e-05, "loss": 1.4062, "step": 5404 }, { "epoch": 13.5125, "grad_norm": 5.90625, "learning_rate": 1.9425964358949746e-05, "loss": 1.281, "step": 5405 }, { "epoch": 13.515, "grad_norm": 6.71875, "learning_rate": 1.9425693945371e-05, "loss": 1.4372, "step": 5406 }, { "epoch": 13.5175, "grad_norm": 5.625, "learning_rate": 1.9425423469997784e-05, "loss": 1.4833, "step": 5407 }, { "epoch": 13.52, "grad_norm": 6.71875, "learning_rate": 1.942515293283187e-05, "loss": 1.3581, "step": 5408 }, { "epoch": 13.5225, "grad_norm": 5.84375, "learning_rate": 1.942488233387503e-05, "loss": 1.3863, "step": 5409 }, { "epoch": 13.525, "grad_norm": 7.21875, "learning_rate": 1.9424611673129033e-05, "loss": 1.6391, "step": 5410 }, { "epoch": 13.5275, "grad_norm": 5.1875, "learning_rate": 1.9424340950595667e-05, "loss": 1.4808, "step": 5411 }, { "epoch": 13.53, "grad_norm": 6.84375, "learning_rate": 1.9424070166276695e-05, "loss": 1.4135, "step": 5412 }, { "epoch": 13.5325, "grad_norm": 5.53125, "learning_rate": 1.9423799320173896e-05, "loss": 1.2968, "step": 5413 }, { "epoch": 13.535, "grad_norm": 5.84375, "learning_rate": 1.942352841228905e-05, "loss": 1.5385, "step": 5414 }, { "epoch": 13.5375, "grad_norm": 5.96875, "learning_rate": 1.9423257442623924e-05, "loss": 1.404, "step": 5415 }, { "epoch": 13.54, "grad_norm": 4.53125, "learning_rate": 1.94229864111803e-05, "loss": 1.4073, "step": 5416 }, { "epoch": 13.5425, "grad_norm": 8.0625, "learning_rate": 1.9422715317959957e-05, "loss": 1.6237, "step": 5417 }, { "epoch": 13.545, "grad_norm": 7.21875, "learning_rate": 1.9422444162964666e-05, "loss": 1.2472, "step": 5418 }, { "epoch": 13.5475, "grad_norm": 7.125, "learning_rate": 1.942217294619621e-05, "loss": 1.5399, "step": 5419 }, { "epoch": 13.55, "grad_norm": 5.34375, "learning_rate": 1.9421901667656364e-05, "loss": 1.4146, "step": 5420 }, { "epoch": 13.5525, "grad_norm": 7.0625, "learning_rate": 1.942163032734691e-05, "loss": 1.4207, "step": 5421 }, { "epoch": 13.555, "grad_norm": 5.96875, "learning_rate": 1.942135892526962e-05, "loss": 1.4185, "step": 5422 }, { "epoch": 13.5575, "grad_norm": 7.125, "learning_rate": 1.9421087461426284e-05, "loss": 1.5817, "step": 5423 }, { "epoch": 13.56, "grad_norm": 10.625, "learning_rate": 1.9420815935818673e-05, "loss": 1.3583, "step": 5424 }, { "epoch": 13.5625, "grad_norm": 5.3125, "learning_rate": 1.942054434844857e-05, "loss": 1.4582, "step": 5425 }, { "epoch": 13.565, "grad_norm": 6.1875, "learning_rate": 1.942027269931775e-05, "loss": 1.3741, "step": 5426 }, { "epoch": 13.5675, "grad_norm": 5.84375, "learning_rate": 1.942000098842801e-05, "loss": 1.5199, "step": 5427 }, { "epoch": 13.57, "grad_norm": 6.59375, "learning_rate": 1.941972921578111e-05, "loss": 1.4898, "step": 5428 }, { "epoch": 13.5725, "grad_norm": 6.90625, "learning_rate": 1.941945738137885e-05, "loss": 1.5639, "step": 5429 }, { "epoch": 13.575, "grad_norm": 7.21875, "learning_rate": 1.9419185485223e-05, "loss": 1.5231, "step": 5430 }, { "epoch": 13.5775, "grad_norm": 8.5625, "learning_rate": 1.9418913527315347e-05, "loss": 1.5631, "step": 5431 }, { "epoch": 13.58, "grad_norm": 8.5, "learning_rate": 1.9418641507657673e-05, "loss": 1.5624, "step": 5432 }, { "epoch": 13.5825, "grad_norm": 6.875, "learning_rate": 1.9418369426251765e-05, "loss": 1.6031, "step": 5433 }, { "epoch": 13.585, "grad_norm": 6.625, "learning_rate": 1.9418097283099402e-05, "loss": 1.5563, "step": 5434 }, { "epoch": 13.5875, "grad_norm": 7.25, "learning_rate": 1.9417825078202372e-05, "loss": 1.5306, "step": 5435 }, { "epoch": 13.59, "grad_norm": 6.15625, "learning_rate": 1.9417552811562457e-05, "loss": 1.4437, "step": 5436 }, { "epoch": 13.5925, "grad_norm": 7.0625, "learning_rate": 1.941728048318144e-05, "loss": 1.4206, "step": 5437 }, { "epoch": 13.595, "grad_norm": 6.375, "learning_rate": 1.941700809306111e-05, "loss": 1.3807, "step": 5438 }, { "epoch": 13.5975, "grad_norm": 6.03125, "learning_rate": 1.9416735641203256e-05, "loss": 1.6043, "step": 5439 }, { "epoch": 13.6, "grad_norm": 8.625, "learning_rate": 1.9416463127609655e-05, "loss": 1.4443, "step": 5440 }, { "epoch": 13.6025, "grad_norm": 6.875, "learning_rate": 1.9416190552282102e-05, "loss": 1.617, "step": 5441 }, { "epoch": 13.605, "grad_norm": 6.0625, "learning_rate": 1.941591791522238e-05, "loss": 1.505, "step": 5442 }, { "epoch": 13.6075, "grad_norm": 5.96875, "learning_rate": 1.9415645216432274e-05, "loss": 1.3529, "step": 5443 }, { "epoch": 13.61, "grad_norm": 5.6875, "learning_rate": 1.9415372455913577e-05, "loss": 1.3413, "step": 5444 }, { "epoch": 13.6125, "grad_norm": 6.0625, "learning_rate": 1.9415099633668076e-05, "loss": 1.3047, "step": 5445 }, { "epoch": 13.615, "grad_norm": 7.8125, "learning_rate": 1.9414826749697556e-05, "loss": 1.4368, "step": 5446 }, { "epoch": 13.6175, "grad_norm": 8.3125, "learning_rate": 1.941455380400381e-05, "loss": 1.5045, "step": 5447 }, { "epoch": 13.62, "grad_norm": 5.09375, "learning_rate": 1.9414280796588625e-05, "loss": 1.4725, "step": 5448 }, { "epoch": 13.6225, "grad_norm": 6.90625, "learning_rate": 1.9414007727453793e-05, "loss": 1.6691, "step": 5449 }, { "epoch": 13.625, "grad_norm": 7.3125, "learning_rate": 1.9413734596601104e-05, "loss": 1.4737, "step": 5450 }, { "epoch": 13.6275, "grad_norm": 7.65625, "learning_rate": 1.9413461404032344e-05, "loss": 1.3999, "step": 5451 }, { "epoch": 13.63, "grad_norm": 7.4375, "learning_rate": 1.9413188149749312e-05, "loss": 1.5081, "step": 5452 }, { "epoch": 13.6325, "grad_norm": 5.09375, "learning_rate": 1.941291483375379e-05, "loss": 1.3425, "step": 5453 }, { "epoch": 13.635, "grad_norm": 6.90625, "learning_rate": 1.941264145604758e-05, "loss": 1.5362, "step": 5454 }, { "epoch": 13.6375, "grad_norm": 5.5625, "learning_rate": 1.941236801663247e-05, "loss": 1.6338, "step": 5455 }, { "epoch": 13.64, "grad_norm": 8.125, "learning_rate": 1.941209451551025e-05, "loss": 1.5652, "step": 5456 }, { "epoch": 13.6425, "grad_norm": 6.25, "learning_rate": 1.9411820952682714e-05, "loss": 1.405, "step": 5457 }, { "epoch": 13.645, "grad_norm": 5.65625, "learning_rate": 1.9411547328151657e-05, "loss": 1.3515, "step": 5458 }, { "epoch": 13.6475, "grad_norm": 11.25, "learning_rate": 1.9411273641918872e-05, "loss": 1.5836, "step": 5459 }, { "epoch": 13.65, "grad_norm": 8.0625, "learning_rate": 1.9410999893986157e-05, "loss": 1.6599, "step": 5460 }, { "epoch": 13.6525, "grad_norm": 6.59375, "learning_rate": 1.9410726084355298e-05, "loss": 1.5155, "step": 5461 }, { "epoch": 13.655, "grad_norm": 7.21875, "learning_rate": 1.94104522130281e-05, "loss": 1.4708, "step": 5462 }, { "epoch": 13.6575, "grad_norm": 7.03125, "learning_rate": 1.941017828000635e-05, "loss": 1.4084, "step": 5463 }, { "epoch": 13.66, "grad_norm": 5.375, "learning_rate": 1.940990428529185e-05, "loss": 1.5894, "step": 5464 }, { "epoch": 13.6625, "grad_norm": 7.46875, "learning_rate": 1.940963022888639e-05, "loss": 1.4709, "step": 5465 }, { "epoch": 13.665, "grad_norm": 9.0, "learning_rate": 1.9409356110791778e-05, "loss": 1.466, "step": 5466 }, { "epoch": 13.6675, "grad_norm": 9.0625, "learning_rate": 1.9409081931009798e-05, "loss": 1.4417, "step": 5467 }, { "epoch": 13.67, "grad_norm": 5.6875, "learning_rate": 1.9408807689542257e-05, "loss": 1.4897, "step": 5468 }, { "epoch": 13.6725, "grad_norm": 7.25, "learning_rate": 1.9408533386390948e-05, "loss": 1.6265, "step": 5469 }, { "epoch": 13.675, "grad_norm": 9.0, "learning_rate": 1.940825902155767e-05, "loss": 1.4237, "step": 5470 }, { "epoch": 13.6775, "grad_norm": 6.96875, "learning_rate": 1.9407984595044224e-05, "loss": 1.5412, "step": 5471 }, { "epoch": 13.68, "grad_norm": 5.0, "learning_rate": 1.9407710106852405e-05, "loss": 1.4182, "step": 5472 }, { "epoch": 13.6825, "grad_norm": 6.1875, "learning_rate": 1.9407435556984017e-05, "loss": 1.4425, "step": 5473 }, { "epoch": 13.685, "grad_norm": 6.5625, "learning_rate": 1.9407160945440856e-05, "loss": 1.4868, "step": 5474 }, { "epoch": 13.6875, "grad_norm": 7.1875, "learning_rate": 1.9406886272224724e-05, "loss": 1.5706, "step": 5475 }, { "epoch": 13.69, "grad_norm": 9.375, "learning_rate": 1.9406611537337425e-05, "loss": 1.5831, "step": 5476 }, { "epoch": 13.6925, "grad_norm": 9.6875, "learning_rate": 1.9406336740780758e-05, "loss": 1.4158, "step": 5477 }, { "epoch": 13.695, "grad_norm": 29.125, "learning_rate": 1.940606188255652e-05, "loss": 1.4287, "step": 5478 }, { "epoch": 13.6975, "grad_norm": 6.0625, "learning_rate": 1.940578696266652e-05, "loss": 1.3709, "step": 5479 }, { "epoch": 13.7, "grad_norm": 10.0625, "learning_rate": 1.9405511981112553e-05, "loss": 1.6651, "step": 5480 }, { "epoch": 13.7025, "grad_norm": 5.5, "learning_rate": 1.940523693789643e-05, "loss": 1.4032, "step": 5481 }, { "epoch": 13.705, "grad_norm": 7.59375, "learning_rate": 1.9404961833019947e-05, "loss": 1.5026, "step": 5482 }, { "epoch": 13.7075, "grad_norm": 9.5, "learning_rate": 1.9404686666484912e-05, "loss": 1.5684, "step": 5483 }, { "epoch": 13.71, "grad_norm": 7.375, "learning_rate": 1.9404411438293126e-05, "loss": 1.4491, "step": 5484 }, { "epoch": 13.7125, "grad_norm": 6.03125, "learning_rate": 1.9404136148446398e-05, "loss": 1.5568, "step": 5485 }, { "epoch": 13.715, "grad_norm": 9.75, "learning_rate": 1.9403860796946527e-05, "loss": 1.5746, "step": 5486 }, { "epoch": 13.7175, "grad_norm": 6.59375, "learning_rate": 1.940358538379532e-05, "loss": 1.5006, "step": 5487 }, { "epoch": 13.72, "grad_norm": 8.75, "learning_rate": 1.940330990899459e-05, "loss": 1.5546, "step": 5488 }, { "epoch": 13.7225, "grad_norm": 6.65625, "learning_rate": 1.940303437254613e-05, "loss": 1.4744, "step": 5489 }, { "epoch": 13.725, "grad_norm": 7.5, "learning_rate": 1.9402758774451756e-05, "loss": 1.501, "step": 5490 }, { "epoch": 13.7275, "grad_norm": 6.03125, "learning_rate": 1.9402483114713272e-05, "loss": 1.5581, "step": 5491 }, { "epoch": 13.73, "grad_norm": 7.8125, "learning_rate": 1.9402207393332483e-05, "loss": 1.5443, "step": 5492 }, { "epoch": 13.7325, "grad_norm": 6.5, "learning_rate": 1.94019316103112e-05, "loss": 1.4194, "step": 5493 }, { "epoch": 13.735, "grad_norm": 6.8125, "learning_rate": 1.9401655765651226e-05, "loss": 1.5271, "step": 5494 }, { "epoch": 13.7375, "grad_norm": 8.0, "learning_rate": 1.9401379859354376e-05, "loss": 1.5745, "step": 5495 }, { "epoch": 13.74, "grad_norm": 6.15625, "learning_rate": 1.9401103891422455e-05, "loss": 1.595, "step": 5496 }, { "epoch": 13.7425, "grad_norm": 8.375, "learning_rate": 1.9400827861857277e-05, "loss": 1.4862, "step": 5497 }, { "epoch": 13.745, "grad_norm": 7.03125, "learning_rate": 1.9400551770660643e-05, "loss": 1.5061, "step": 5498 }, { "epoch": 13.7475, "grad_norm": 5.84375, "learning_rate": 1.940027561783437e-05, "loss": 1.5355, "step": 5499 }, { "epoch": 13.75, "grad_norm": 5.3125, "learning_rate": 1.9399999403380266e-05, "loss": 1.2915, "step": 5500 }, { "epoch": 13.7525, "grad_norm": 7.53125, "learning_rate": 1.939972312730014e-05, "loss": 1.5003, "step": 5501 }, { "epoch": 13.755, "grad_norm": 5.375, "learning_rate": 1.9399446789595808e-05, "loss": 1.4523, "step": 5502 }, { "epoch": 13.7575, "grad_norm": 5.96875, "learning_rate": 1.939917039026908e-05, "loss": 1.5807, "step": 5503 }, { "epoch": 13.76, "grad_norm": 5.6875, "learning_rate": 1.9398893929321763e-05, "loss": 1.3377, "step": 5504 }, { "epoch": 13.7625, "grad_norm": 5.90625, "learning_rate": 1.9398617406755674e-05, "loss": 1.5522, "step": 5505 }, { "epoch": 13.765, "grad_norm": 8.0625, "learning_rate": 1.9398340822572628e-05, "loss": 1.5321, "step": 5506 }, { "epoch": 13.7675, "grad_norm": 6.4375, "learning_rate": 1.9398064176774432e-05, "loss": 1.6521, "step": 5507 }, { "epoch": 13.77, "grad_norm": 8.375, "learning_rate": 1.9397787469362908e-05, "loss": 1.7454, "step": 5508 }, { "epoch": 13.7725, "grad_norm": 4.625, "learning_rate": 1.9397510700339864e-05, "loss": 1.4838, "step": 5509 }, { "epoch": 13.775, "grad_norm": 5.875, "learning_rate": 1.9397233869707116e-05, "loss": 1.3387, "step": 5510 }, { "epoch": 13.7775, "grad_norm": 7.34375, "learning_rate": 1.9396956977466477e-05, "loss": 1.5094, "step": 5511 }, { "epoch": 13.78, "grad_norm": 4.40625, "learning_rate": 1.9396680023619767e-05, "loss": 1.4051, "step": 5512 }, { "epoch": 13.7825, "grad_norm": 7.125, "learning_rate": 1.9396403008168797e-05, "loss": 1.5775, "step": 5513 }, { "epoch": 13.785, "grad_norm": 4.8125, "learning_rate": 1.9396125931115384e-05, "loss": 1.4848, "step": 5514 }, { "epoch": 13.7875, "grad_norm": 6.375, "learning_rate": 1.9395848792461348e-05, "loss": 1.3226, "step": 5515 }, { "epoch": 13.79, "grad_norm": 12.0625, "learning_rate": 1.93955715922085e-05, "loss": 1.5212, "step": 5516 }, { "epoch": 13.7925, "grad_norm": 5.75, "learning_rate": 1.9395294330358663e-05, "loss": 1.3847, "step": 5517 }, { "epoch": 13.795, "grad_norm": 6.84375, "learning_rate": 1.939501700691365e-05, "loss": 1.4172, "step": 5518 }, { "epoch": 13.7975, "grad_norm": 5.40625, "learning_rate": 1.9394739621875285e-05, "loss": 1.4482, "step": 5519 }, { "epoch": 13.8, "grad_norm": 5.65625, "learning_rate": 1.9394462175245382e-05, "loss": 1.4631, "step": 5520 }, { "epoch": 13.8025, "grad_norm": 7.46875, "learning_rate": 1.939418466702576e-05, "loss": 1.4614, "step": 5521 }, { "epoch": 13.805, "grad_norm": 6.21875, "learning_rate": 1.9393907097218237e-05, "loss": 1.4523, "step": 5522 }, { "epoch": 13.8075, "grad_norm": 8.1875, "learning_rate": 1.9393629465824637e-05, "loss": 1.4286, "step": 5523 }, { "epoch": 13.81, "grad_norm": 9.0625, "learning_rate": 1.939335177284678e-05, "loss": 1.4139, "step": 5524 }, { "epoch": 13.8125, "grad_norm": 7.90625, "learning_rate": 1.939307401828648e-05, "loss": 1.3239, "step": 5525 }, { "epoch": 13.815, "grad_norm": 5.6875, "learning_rate": 1.939279620214557e-05, "loss": 1.3507, "step": 5526 }, { "epoch": 13.8175, "grad_norm": 6.90625, "learning_rate": 1.9392518324425857e-05, "loss": 1.3769, "step": 5527 }, { "epoch": 13.82, "grad_norm": 9.3125, "learning_rate": 1.9392240385129174e-05, "loss": 1.3567, "step": 5528 }, { "epoch": 13.8225, "grad_norm": 6.46875, "learning_rate": 1.939196238425734e-05, "loss": 1.3468, "step": 5529 }, { "epoch": 13.825, "grad_norm": 6.71875, "learning_rate": 1.9391684321812173e-05, "loss": 1.4283, "step": 5530 }, { "epoch": 13.8275, "grad_norm": 7.03125, "learning_rate": 1.9391406197795497e-05, "loss": 1.5659, "step": 5531 }, { "epoch": 13.83, "grad_norm": 6.9375, "learning_rate": 1.9391128012209144e-05, "loss": 1.4483, "step": 5532 }, { "epoch": 13.8325, "grad_norm": 8.375, "learning_rate": 1.9390849765054925e-05, "loss": 1.409, "step": 5533 }, { "epoch": 13.835, "grad_norm": 8.75, "learning_rate": 1.9390571456334678e-05, "loss": 1.5635, "step": 5534 }, { "epoch": 13.8375, "grad_norm": 8.1875, "learning_rate": 1.9390293086050213e-05, "loss": 1.5459, "step": 5535 }, { "epoch": 13.84, "grad_norm": 7.65625, "learning_rate": 1.939001465420337e-05, "loss": 1.3877, "step": 5536 }, { "epoch": 13.8425, "grad_norm": 7.15625, "learning_rate": 1.938973616079596e-05, "loss": 1.504, "step": 5537 }, { "epoch": 13.845, "grad_norm": 10.9375, "learning_rate": 1.938945760582982e-05, "loss": 1.6082, "step": 5538 }, { "epoch": 13.8475, "grad_norm": 6.65625, "learning_rate": 1.9389178989306768e-05, "loss": 1.5283, "step": 5539 }, { "epoch": 13.85, "grad_norm": 9.5625, "learning_rate": 1.9388900311228636e-05, "loss": 1.562, "step": 5540 }, { "epoch": 13.8525, "grad_norm": 8.3125, "learning_rate": 1.938862157159725e-05, "loss": 1.4315, "step": 5541 }, { "epoch": 13.855, "grad_norm": 5.65625, "learning_rate": 1.9388342770414437e-05, "loss": 1.4618, "step": 5542 }, { "epoch": 13.8575, "grad_norm": 6.8125, "learning_rate": 1.9388063907682025e-05, "loss": 1.6961, "step": 5543 }, { "epoch": 13.86, "grad_norm": 6.15625, "learning_rate": 1.9387784983401837e-05, "loss": 1.5254, "step": 5544 }, { "epoch": 13.8625, "grad_norm": 7.96875, "learning_rate": 1.938750599757571e-05, "loss": 1.5606, "step": 5545 }, { "epoch": 13.865, "grad_norm": 67.5, "learning_rate": 1.9387226950205468e-05, "loss": 1.5315, "step": 5546 }, { "epoch": 13.8675, "grad_norm": 7.3125, "learning_rate": 1.9386947841292945e-05, "loss": 1.464, "step": 5547 }, { "epoch": 13.87, "grad_norm": 6.03125, "learning_rate": 1.9386668670839965e-05, "loss": 1.5326, "step": 5548 }, { "epoch": 13.8725, "grad_norm": 7.65625, "learning_rate": 1.938638943884836e-05, "loss": 1.5685, "step": 5549 }, { "epoch": 13.875, "grad_norm": 5.9375, "learning_rate": 1.9386110145319962e-05, "loss": 1.3962, "step": 5550 }, { "epoch": 13.8775, "grad_norm": 5.0625, "learning_rate": 1.9385830790256602e-05, "loss": 1.3885, "step": 5551 }, { "epoch": 13.88, "grad_norm": 7.5625, "learning_rate": 1.9385551373660113e-05, "loss": 1.6528, "step": 5552 }, { "epoch": 13.8825, "grad_norm": 7.0625, "learning_rate": 1.9385271895532322e-05, "loss": 1.3889, "step": 5553 }, { "epoch": 13.885, "grad_norm": 7.40625, "learning_rate": 1.9384992355875065e-05, "loss": 1.4882, "step": 5554 }, { "epoch": 13.8875, "grad_norm": 5.875, "learning_rate": 1.9384712754690175e-05, "loss": 1.5293, "step": 5555 }, { "epoch": 13.89, "grad_norm": 6.0625, "learning_rate": 1.938443309197948e-05, "loss": 1.4983, "step": 5556 }, { "epoch": 13.8925, "grad_norm": 6.625, "learning_rate": 1.938415336774482e-05, "loss": 1.4679, "step": 5557 }, { "epoch": 13.895, "grad_norm": 6.90625, "learning_rate": 1.9383873581988026e-05, "loss": 1.4223, "step": 5558 }, { "epoch": 13.8975, "grad_norm": 7.09375, "learning_rate": 1.938359373471093e-05, "loss": 1.4688, "step": 5559 }, { "epoch": 13.9, "grad_norm": 6.5625, "learning_rate": 1.9383313825915372e-05, "loss": 1.5214, "step": 5560 }, { "epoch": 13.9025, "grad_norm": 8.25, "learning_rate": 1.9383033855603183e-05, "loss": 1.474, "step": 5561 }, { "epoch": 13.905, "grad_norm": 6.59375, "learning_rate": 1.93827538237762e-05, "loss": 1.5016, "step": 5562 }, { "epoch": 13.9075, "grad_norm": 6.71875, "learning_rate": 1.938247373043626e-05, "loss": 1.4656, "step": 5563 }, { "epoch": 13.91, "grad_norm": 6.0625, "learning_rate": 1.9382193575585193e-05, "loss": 1.3509, "step": 5564 }, { "epoch": 13.9125, "grad_norm": 6.5625, "learning_rate": 1.9381913359224844e-05, "loss": 1.3776, "step": 5565 }, { "epoch": 13.915, "grad_norm": 7.40625, "learning_rate": 1.9381633081357045e-05, "loss": 1.3116, "step": 5566 }, { "epoch": 13.9175, "grad_norm": 7.75, "learning_rate": 1.9381352741983636e-05, "loss": 1.542, "step": 5567 }, { "epoch": 13.92, "grad_norm": 9.0625, "learning_rate": 1.9381072341106453e-05, "loss": 1.5696, "step": 5568 }, { "epoch": 13.9225, "grad_norm": 7.125, "learning_rate": 1.9380791878727336e-05, "loss": 1.3846, "step": 5569 }, { "epoch": 13.925, "grad_norm": 11.6875, "learning_rate": 1.938051135484812e-05, "loss": 1.7445, "step": 5570 }, { "epoch": 13.9275, "grad_norm": 8.0, "learning_rate": 1.938023076947065e-05, "loss": 1.4658, "step": 5571 }, { "epoch": 13.93, "grad_norm": 5.78125, "learning_rate": 1.937995012259676e-05, "loss": 1.5774, "step": 5572 }, { "epoch": 13.9325, "grad_norm": 8.1875, "learning_rate": 1.9379669414228295e-05, "loss": 1.4046, "step": 5573 }, { "epoch": 13.935, "grad_norm": 5.125, "learning_rate": 1.937938864436709e-05, "loss": 1.4599, "step": 5574 }, { "epoch": 13.9375, "grad_norm": 8.9375, "learning_rate": 1.937910781301499e-05, "loss": 1.5133, "step": 5575 }, { "epoch": 13.94, "grad_norm": 8.0, "learning_rate": 1.9378826920173835e-05, "loss": 1.5434, "step": 5576 }, { "epoch": 13.9425, "grad_norm": 7.15625, "learning_rate": 1.9378545965845467e-05, "loss": 1.482, "step": 5577 }, { "epoch": 13.945, "grad_norm": 6.375, "learning_rate": 1.9378264950031725e-05, "loss": 1.5493, "step": 5578 }, { "epoch": 13.9475, "grad_norm": 5.6875, "learning_rate": 1.9377983872734454e-05, "loss": 1.4234, "step": 5579 }, { "epoch": 13.95, "grad_norm": 7.21875, "learning_rate": 1.9377702733955493e-05, "loss": 1.4522, "step": 5580 }, { "epoch": 13.9525, "grad_norm": 6.0625, "learning_rate": 1.9377421533696692e-05, "loss": 1.4868, "step": 5581 }, { "epoch": 13.955, "grad_norm": 7.25, "learning_rate": 1.9377140271959888e-05, "loss": 1.4205, "step": 5582 }, { "epoch": 13.9575, "grad_norm": 9.625, "learning_rate": 1.9376858948746933e-05, "loss": 1.5554, "step": 5583 }, { "epoch": 13.96, "grad_norm": 9.75, "learning_rate": 1.937657756405966e-05, "loss": 1.5093, "step": 5584 }, { "epoch": 13.9625, "grad_norm": 6.71875, "learning_rate": 1.937629611789992e-05, "loss": 1.3833, "step": 5585 }, { "epoch": 13.965, "grad_norm": 7.34375, "learning_rate": 1.9376014610269563e-05, "loss": 1.52, "step": 5586 }, { "epoch": 13.9675, "grad_norm": 8.8125, "learning_rate": 1.9375733041170424e-05, "loss": 1.5196, "step": 5587 }, { "epoch": 13.97, "grad_norm": 7.78125, "learning_rate": 1.9375451410604358e-05, "loss": 1.5834, "step": 5588 }, { "epoch": 13.9725, "grad_norm": 5.65625, "learning_rate": 1.9375169718573207e-05, "loss": 1.3338, "step": 5589 }, { "epoch": 13.975, "grad_norm": 9.1875, "learning_rate": 1.9374887965078817e-05, "loss": 1.5222, "step": 5590 }, { "epoch": 13.9775, "grad_norm": 7.84375, "learning_rate": 1.9374606150123036e-05, "loss": 1.4526, "step": 5591 }, { "epoch": 13.98, "grad_norm": 7.125, "learning_rate": 1.9374324273707717e-05, "loss": 1.3596, "step": 5592 }, { "epoch": 13.9825, "grad_norm": 8.25, "learning_rate": 1.93740423358347e-05, "loss": 1.4889, "step": 5593 }, { "epoch": 13.985, "grad_norm": 6.15625, "learning_rate": 1.9373760336505832e-05, "loss": 1.3949, "step": 5594 }, { "epoch": 13.9875, "grad_norm": 8.5625, "learning_rate": 1.9373478275722973e-05, "loss": 1.5417, "step": 5595 }, { "epoch": 13.99, "grad_norm": 5.9375, "learning_rate": 1.9373196153487962e-05, "loss": 1.4433, "step": 5596 }, { "epoch": 13.9925, "grad_norm": 5.625, "learning_rate": 1.9372913969802652e-05, "loss": 1.5092, "step": 5597 }, { "epoch": 13.995, "grad_norm": 6.75, "learning_rate": 1.9372631724668897e-05, "loss": 1.5604, "step": 5598 }, { "epoch": 13.9975, "grad_norm": 5.75, "learning_rate": 1.9372349418088542e-05, "loss": 1.5466, "step": 5599 }, { "epoch": 14.0, "grad_norm": 6.5625, "learning_rate": 1.937206705006344e-05, "loss": 1.4977, "step": 5600 }, { "epoch": 14.0025, "grad_norm": 6.0625, "learning_rate": 1.937178462059544e-05, "loss": 1.4447, "step": 5601 }, { "epoch": 14.005, "grad_norm": 6.25, "learning_rate": 1.9371502129686394e-05, "loss": 1.647, "step": 5602 }, { "epoch": 14.0075, "grad_norm": 6.25, "learning_rate": 1.9371219577338155e-05, "loss": 1.4425, "step": 5603 }, { "epoch": 14.01, "grad_norm": 9.375, "learning_rate": 1.9370936963552574e-05, "loss": 1.4275, "step": 5604 }, { "epoch": 14.0125, "grad_norm": 4.4375, "learning_rate": 1.937065428833151e-05, "loss": 1.4554, "step": 5605 }, { "epoch": 14.015, "grad_norm": 8.4375, "learning_rate": 1.9370371551676808e-05, "loss": 1.3466, "step": 5606 }, { "epoch": 14.0175, "grad_norm": 6.90625, "learning_rate": 1.9370088753590326e-05, "loss": 1.4422, "step": 5607 }, { "epoch": 14.02, "grad_norm": 6.78125, "learning_rate": 1.9369805894073917e-05, "loss": 1.4896, "step": 5608 }, { "epoch": 14.0225, "grad_norm": 6.46875, "learning_rate": 1.9369522973129436e-05, "loss": 1.4431, "step": 5609 }, { "epoch": 14.025, "grad_norm": 8.0625, "learning_rate": 1.936923999075874e-05, "loss": 1.5856, "step": 5610 }, { "epoch": 14.0275, "grad_norm": 7.25, "learning_rate": 1.936895694696368e-05, "loss": 1.6452, "step": 5611 }, { "epoch": 14.03, "grad_norm": 4.96875, "learning_rate": 1.936867384174611e-05, "loss": 1.4599, "step": 5612 }, { "epoch": 14.0325, "grad_norm": 8.125, "learning_rate": 1.936839067510789e-05, "loss": 1.5237, "step": 5613 }, { "epoch": 14.035, "grad_norm": 6.53125, "learning_rate": 1.936810744705088e-05, "loss": 1.3887, "step": 5614 }, { "epoch": 14.0375, "grad_norm": 5.8125, "learning_rate": 1.9367824157576924e-05, "loss": 1.5057, "step": 5615 }, { "epoch": 14.04, "grad_norm": 7.53125, "learning_rate": 1.9367540806687894e-05, "loss": 1.5419, "step": 5616 }, { "epoch": 14.0425, "grad_norm": 5.21875, "learning_rate": 1.936725739438564e-05, "loss": 1.4085, "step": 5617 }, { "epoch": 14.045, "grad_norm": 7.0, "learning_rate": 1.9366973920672022e-05, "loss": 1.5488, "step": 5618 }, { "epoch": 14.0475, "grad_norm": 6.5625, "learning_rate": 1.9366690385548897e-05, "loss": 1.3603, "step": 5619 }, { "epoch": 14.05, "grad_norm": 6.34375, "learning_rate": 1.9366406789018127e-05, "loss": 1.4945, "step": 5620 }, { "epoch": 14.0525, "grad_norm": 7.9375, "learning_rate": 1.9366123131081568e-05, "loss": 1.445, "step": 5621 }, { "epoch": 14.055, "grad_norm": 5.78125, "learning_rate": 1.936583941174108e-05, "loss": 1.4577, "step": 5622 }, { "epoch": 14.0575, "grad_norm": 7.96875, "learning_rate": 1.936555563099852e-05, "loss": 1.4885, "step": 5623 }, { "epoch": 14.06, "grad_norm": 14.8125, "learning_rate": 1.9365271788855756e-05, "loss": 1.5823, "step": 5624 }, { "epoch": 14.0625, "grad_norm": 6.46875, "learning_rate": 1.9364987885314645e-05, "loss": 1.5761, "step": 5625 }, { "epoch": 14.065, "grad_norm": 6.71875, "learning_rate": 1.9364703920377045e-05, "loss": 1.4544, "step": 5626 }, { "epoch": 14.0675, "grad_norm": 9.5, "learning_rate": 1.9364419894044823e-05, "loss": 1.5755, "step": 5627 }, { "epoch": 14.07, "grad_norm": 6.375, "learning_rate": 1.936413580631984e-05, "loss": 1.6228, "step": 5628 }, { "epoch": 14.0725, "grad_norm": 5.71875, "learning_rate": 1.936385165720396e-05, "loss": 1.4039, "step": 5629 }, { "epoch": 14.075, "grad_norm": 8.25, "learning_rate": 1.9363567446699035e-05, "loss": 1.5613, "step": 5630 }, { "epoch": 14.0775, "grad_norm": 7.21875, "learning_rate": 1.9363283174806944e-05, "loss": 1.454, "step": 5631 }, { "epoch": 14.08, "grad_norm": 7.4375, "learning_rate": 1.9362998841529542e-05, "loss": 1.5198, "step": 5632 }, { "epoch": 14.0825, "grad_norm": 5.90625, "learning_rate": 1.9362714446868693e-05, "loss": 1.5403, "step": 5633 }, { "epoch": 14.085, "grad_norm": 9.0, "learning_rate": 1.9362429990826262e-05, "loss": 1.3612, "step": 5634 }, { "epoch": 14.0875, "grad_norm": 14.0, "learning_rate": 1.9362145473404114e-05, "loss": 1.4308, "step": 5635 }, { "epoch": 14.09, "grad_norm": 8.9375, "learning_rate": 1.9361860894604116e-05, "loss": 1.4826, "step": 5636 }, { "epoch": 14.0925, "grad_norm": 6.3125, "learning_rate": 1.9361576254428134e-05, "loss": 1.423, "step": 5637 }, { "epoch": 14.095, "grad_norm": 8.5, "learning_rate": 1.936129155287803e-05, "loss": 1.6055, "step": 5638 }, { "epoch": 14.0975, "grad_norm": 9.5, "learning_rate": 1.9361006789955676e-05, "loss": 1.5899, "step": 5639 }, { "epoch": 14.1, "grad_norm": 8.0625, "learning_rate": 1.9360721965662934e-05, "loss": 1.6739, "step": 5640 }, { "epoch": 14.1025, "grad_norm": 7.84375, "learning_rate": 1.9360437080001672e-05, "loss": 1.4917, "step": 5641 }, { "epoch": 14.105, "grad_norm": 8.25, "learning_rate": 1.936015213297376e-05, "loss": 1.4681, "step": 5642 }, { "epoch": 14.1075, "grad_norm": 6.75, "learning_rate": 1.935986712458107e-05, "loss": 1.3708, "step": 5643 }, { "epoch": 14.11, "grad_norm": 6.0, "learning_rate": 1.9359582054825462e-05, "loss": 1.4349, "step": 5644 }, { "epoch": 14.1125, "grad_norm": 7.65625, "learning_rate": 1.935929692370881e-05, "loss": 1.3765, "step": 5645 }, { "epoch": 14.115, "grad_norm": 7.9375, "learning_rate": 1.9359011731232978e-05, "loss": 1.4772, "step": 5646 }, { "epoch": 14.1175, "grad_norm": 9.5, "learning_rate": 1.9358726477399844e-05, "loss": 1.4475, "step": 5647 }, { "epoch": 14.12, "grad_norm": 6.875, "learning_rate": 1.935844116221127e-05, "loss": 1.6314, "step": 5648 }, { "epoch": 14.1225, "grad_norm": 7.90625, "learning_rate": 1.9358155785669134e-05, "loss": 1.4655, "step": 5649 }, { "epoch": 14.125, "grad_norm": 6.21875, "learning_rate": 1.93578703477753e-05, "loss": 1.5667, "step": 5650 }, { "epoch": 14.1275, "grad_norm": 6.5625, "learning_rate": 1.9357584848531645e-05, "loss": 1.2811, "step": 5651 }, { "epoch": 14.13, "grad_norm": 6.71875, "learning_rate": 1.9357299287940037e-05, "loss": 1.5595, "step": 5652 }, { "epoch": 14.1325, "grad_norm": 4.78125, "learning_rate": 1.935701366600235e-05, "loss": 1.3306, "step": 5653 }, { "epoch": 14.135, "grad_norm": 5.21875, "learning_rate": 1.9356727982720454e-05, "loss": 1.2921, "step": 5654 }, { "epoch": 14.1375, "grad_norm": 7.625, "learning_rate": 1.9356442238096226e-05, "loss": 1.3933, "step": 5655 }, { "epoch": 14.14, "grad_norm": 6.78125, "learning_rate": 1.9356156432131533e-05, "loss": 1.5882, "step": 5656 }, { "epoch": 14.1425, "grad_norm": 6.3125, "learning_rate": 1.9355870564828258e-05, "loss": 1.5782, "step": 5657 }, { "epoch": 14.145, "grad_norm": 7.34375, "learning_rate": 1.935558463618827e-05, "loss": 1.4735, "step": 5658 }, { "epoch": 14.1475, "grad_norm": 12.5, "learning_rate": 1.9355298646213438e-05, "loss": 1.4706, "step": 5659 }, { "epoch": 14.15, "grad_norm": 7.5625, "learning_rate": 1.9355012594905645e-05, "loss": 1.4871, "step": 5660 }, { "epoch": 14.1525, "grad_norm": 8.625, "learning_rate": 1.9354726482266768e-05, "loss": 1.5114, "step": 5661 }, { "epoch": 14.155, "grad_norm": 5.59375, "learning_rate": 1.9354440308298676e-05, "loss": 1.2583, "step": 5662 }, { "epoch": 14.1575, "grad_norm": 5.5, "learning_rate": 1.9354154073003245e-05, "loss": 1.3933, "step": 5663 }, { "epoch": 14.16, "grad_norm": 7.46875, "learning_rate": 1.9353867776382357e-05, "loss": 1.653, "step": 5664 }, { "epoch": 14.1625, "grad_norm": 6.4375, "learning_rate": 1.9353581418437886e-05, "loss": 1.5968, "step": 5665 }, { "epoch": 14.165, "grad_norm": 7.0625, "learning_rate": 1.9353294999171706e-05, "loss": 1.6189, "step": 5666 }, { "epoch": 14.1675, "grad_norm": 7.1875, "learning_rate": 1.9353008518585704e-05, "loss": 1.448, "step": 5667 }, { "epoch": 14.17, "grad_norm": 7.625, "learning_rate": 1.9352721976681747e-05, "loss": 1.5576, "step": 5668 }, { "epoch": 14.1725, "grad_norm": 7.9375, "learning_rate": 1.9352435373461723e-05, "loss": 1.3961, "step": 5669 }, { "epoch": 14.175, "grad_norm": 9.875, "learning_rate": 1.9352148708927507e-05, "loss": 1.5461, "step": 5670 }, { "epoch": 14.1775, "grad_norm": 9.4375, "learning_rate": 1.9351861983080974e-05, "loss": 1.4572, "step": 5671 }, { "epoch": 14.18, "grad_norm": 6.53125, "learning_rate": 1.9351575195924014e-05, "loss": 1.6325, "step": 5672 }, { "epoch": 14.1825, "grad_norm": 5.3125, "learning_rate": 1.9351288347458498e-05, "loss": 1.5089, "step": 5673 }, { "epoch": 14.185, "grad_norm": 8.625, "learning_rate": 1.935100143768631e-05, "loss": 1.4822, "step": 5674 }, { "epoch": 14.1875, "grad_norm": 5.5625, "learning_rate": 1.935071446660933e-05, "loss": 1.3932, "step": 5675 }, { "epoch": 14.19, "grad_norm": 9.375, "learning_rate": 1.935042743422944e-05, "loss": 1.6476, "step": 5676 }, { "epoch": 14.1925, "grad_norm": 8.5, "learning_rate": 1.935014034054852e-05, "loss": 1.657, "step": 5677 }, { "epoch": 14.195, "grad_norm": 7.5625, "learning_rate": 1.934985318556846e-05, "loss": 1.4481, "step": 5678 }, { "epoch": 14.1975, "grad_norm": 7.125, "learning_rate": 1.9349565969291134e-05, "loss": 1.4521, "step": 5679 }, { "epoch": 14.2, "grad_norm": 5.6875, "learning_rate": 1.9349278691718426e-05, "loss": 1.4235, "step": 5680 }, { "epoch": 14.2025, "grad_norm": 6.53125, "learning_rate": 1.9348991352852224e-05, "loss": 1.5405, "step": 5681 }, { "epoch": 14.205, "grad_norm": 9.4375, "learning_rate": 1.9348703952694407e-05, "loss": 1.6564, "step": 5682 }, { "epoch": 14.2075, "grad_norm": 6.90625, "learning_rate": 1.934841649124686e-05, "loss": 1.3925, "step": 5683 }, { "epoch": 14.21, "grad_norm": 6.5625, "learning_rate": 1.934812896851147e-05, "loss": 1.4823, "step": 5684 }, { "epoch": 14.2125, "grad_norm": 7.96875, "learning_rate": 1.9347841384490122e-05, "loss": 1.5128, "step": 5685 }, { "epoch": 14.215, "grad_norm": 5.0625, "learning_rate": 1.9347553739184696e-05, "loss": 1.4185, "step": 5686 }, { "epoch": 14.2175, "grad_norm": 7.34375, "learning_rate": 1.9347266032597084e-05, "loss": 1.4683, "step": 5687 }, { "epoch": 14.22, "grad_norm": 5.59375, "learning_rate": 1.9346978264729173e-05, "loss": 1.4057, "step": 5688 }, { "epoch": 14.2225, "grad_norm": 4.84375, "learning_rate": 1.9346690435582842e-05, "loss": 1.3585, "step": 5689 }, { "epoch": 14.225, "grad_norm": 7.65625, "learning_rate": 1.9346402545159984e-05, "loss": 1.6861, "step": 5690 }, { "epoch": 14.2275, "grad_norm": 5.3125, "learning_rate": 1.9346114593462487e-05, "loss": 1.4298, "step": 5691 }, { "epoch": 14.23, "grad_norm": 6.03125, "learning_rate": 1.934582658049223e-05, "loss": 1.5588, "step": 5692 }, { "epoch": 14.2325, "grad_norm": 9.3125, "learning_rate": 1.9345538506251115e-05, "loss": 1.4616, "step": 5693 }, { "epoch": 14.235, "grad_norm": 5.125, "learning_rate": 1.934525037074102e-05, "loss": 1.446, "step": 5694 }, { "epoch": 14.2375, "grad_norm": 6.875, "learning_rate": 1.9344962173963838e-05, "loss": 1.4699, "step": 5695 }, { "epoch": 14.24, "grad_norm": 6.6875, "learning_rate": 1.934467391592146e-05, "loss": 1.4288, "step": 5696 }, { "epoch": 14.2425, "grad_norm": 6.96875, "learning_rate": 1.9344385596615774e-05, "loss": 1.2216, "step": 5697 }, { "epoch": 14.245, "grad_norm": 5.8125, "learning_rate": 1.9344097216048667e-05, "loss": 1.3808, "step": 5698 }, { "epoch": 14.2475, "grad_norm": 4.84375, "learning_rate": 1.9343808774222034e-05, "loss": 1.2871, "step": 5699 }, { "epoch": 14.25, "grad_norm": 7.625, "learning_rate": 1.9343520271137764e-05, "loss": 1.6261, "step": 5700 }, { "epoch": 14.2525, "grad_norm": 11.375, "learning_rate": 1.934323170679775e-05, "loss": 1.5892, "step": 5701 }, { "epoch": 14.255, "grad_norm": 7.65625, "learning_rate": 1.9342943081203885e-05, "loss": 1.4821, "step": 5702 }, { "epoch": 14.2575, "grad_norm": 8.625, "learning_rate": 1.9342654394358057e-05, "loss": 1.634, "step": 5703 }, { "epoch": 14.26, "grad_norm": 19.125, "learning_rate": 1.9342365646262158e-05, "loss": 1.5765, "step": 5704 }, { "epoch": 14.2625, "grad_norm": 11.375, "learning_rate": 1.9342076836918083e-05, "loss": 1.4961, "step": 5705 }, { "epoch": 14.265, "grad_norm": 5.90625, "learning_rate": 1.934178796632773e-05, "loss": 1.494, "step": 5706 }, { "epoch": 14.2675, "grad_norm": 10.875, "learning_rate": 1.934149903449299e-05, "loss": 1.5858, "step": 5707 }, { "epoch": 14.27, "grad_norm": 5.8125, "learning_rate": 1.9341210041415755e-05, "loss": 1.4368, "step": 5708 }, { "epoch": 14.2725, "grad_norm": 8.0625, "learning_rate": 1.9340920987097917e-05, "loss": 1.3886, "step": 5709 }, { "epoch": 14.275, "grad_norm": 6.96875, "learning_rate": 1.934063187154138e-05, "loss": 1.5576, "step": 5710 }, { "epoch": 14.2775, "grad_norm": 6.71875, "learning_rate": 1.934034269474803e-05, "loss": 1.3929, "step": 5711 }, { "epoch": 14.28, "grad_norm": 6.40625, "learning_rate": 1.9340053456719768e-05, "loss": 1.4174, "step": 5712 }, { "epoch": 14.2825, "grad_norm": 8.4375, "learning_rate": 1.933976415745849e-05, "loss": 1.3826, "step": 5713 }, { "epoch": 14.285, "grad_norm": 6.21875, "learning_rate": 1.9339474796966092e-05, "loss": 1.5321, "step": 5714 }, { "epoch": 14.2875, "grad_norm": 6.03125, "learning_rate": 1.933918537524447e-05, "loss": 1.3769, "step": 5715 }, { "epoch": 14.29, "grad_norm": 6.65625, "learning_rate": 1.9338895892295527e-05, "loss": 1.4133, "step": 5716 }, { "epoch": 14.2925, "grad_norm": 7.1875, "learning_rate": 1.933860634812115e-05, "loss": 1.4279, "step": 5717 }, { "epoch": 14.295, "grad_norm": 6.125, "learning_rate": 1.933831674272325e-05, "loss": 1.4014, "step": 5718 }, { "epoch": 14.2975, "grad_norm": 6.5625, "learning_rate": 1.9338027076103712e-05, "loss": 1.5601, "step": 5719 }, { "epoch": 14.3, "grad_norm": 6.09375, "learning_rate": 1.9337737348264448e-05, "loss": 1.5998, "step": 5720 }, { "epoch": 14.3025, "grad_norm": 8.0, "learning_rate": 1.933744755920735e-05, "loss": 1.5189, "step": 5721 }, { "epoch": 14.305, "grad_norm": 6.4375, "learning_rate": 1.933715770893432e-05, "loss": 1.6488, "step": 5722 }, { "epoch": 14.3075, "grad_norm": 10.25, "learning_rate": 1.9336867797447253e-05, "loss": 1.6213, "step": 5723 }, { "epoch": 14.31, "grad_norm": 8.0625, "learning_rate": 1.9336577824748057e-05, "loss": 1.4946, "step": 5724 }, { "epoch": 14.3125, "grad_norm": 8.625, "learning_rate": 1.933628779083863e-05, "loss": 1.3466, "step": 5725 }, { "epoch": 14.315, "grad_norm": 6.25, "learning_rate": 1.9335997695720877e-05, "loss": 1.3344, "step": 5726 }, { "epoch": 14.3175, "grad_norm": 6.9375, "learning_rate": 1.9335707539396694e-05, "loss": 1.5153, "step": 5727 }, { "epoch": 14.32, "grad_norm": 8.5, "learning_rate": 1.9335417321867988e-05, "loss": 1.4441, "step": 5728 }, { "epoch": 14.3225, "grad_norm": 6.09375, "learning_rate": 1.9335127043136658e-05, "loss": 1.4551, "step": 5729 }, { "epoch": 14.325, "grad_norm": 9.5, "learning_rate": 1.933483670320461e-05, "loss": 1.5162, "step": 5730 }, { "epoch": 14.3275, "grad_norm": 5.9375, "learning_rate": 1.9334546302073743e-05, "loss": 1.4918, "step": 5731 }, { "epoch": 14.33, "grad_norm": 8.0625, "learning_rate": 1.9334255839745966e-05, "loss": 1.513, "step": 5732 }, { "epoch": 14.3325, "grad_norm": 5.84375, "learning_rate": 1.9333965316223182e-05, "loss": 1.3981, "step": 5733 }, { "epoch": 14.335, "grad_norm": 8.0, "learning_rate": 1.9333674731507296e-05, "loss": 1.463, "step": 5734 }, { "epoch": 14.3375, "grad_norm": 7.375, "learning_rate": 1.9333384085600208e-05, "loss": 1.3078, "step": 5735 }, { "epoch": 14.34, "grad_norm": 9.8125, "learning_rate": 1.9333093378503832e-05, "loss": 1.3408, "step": 5736 }, { "epoch": 14.3425, "grad_norm": 5.90625, "learning_rate": 1.9332802610220067e-05, "loss": 1.4491, "step": 5737 }, { "epoch": 14.345, "grad_norm": 6.375, "learning_rate": 1.933251178075082e-05, "loss": 1.4829, "step": 5738 }, { "epoch": 14.3475, "grad_norm": 7.125, "learning_rate": 1.9332220890098002e-05, "loss": 1.4938, "step": 5739 }, { "epoch": 14.35, "grad_norm": 5.46875, "learning_rate": 1.9331929938263515e-05, "loss": 1.5431, "step": 5740 }, { "epoch": 14.3525, "grad_norm": 6.125, "learning_rate": 1.933163892524927e-05, "loss": 1.5513, "step": 5741 }, { "epoch": 14.355, "grad_norm": 7.1875, "learning_rate": 1.9331347851057175e-05, "loss": 1.4872, "step": 5742 }, { "epoch": 14.3575, "grad_norm": 6.71875, "learning_rate": 1.9331056715689133e-05, "loss": 1.4818, "step": 5743 }, { "epoch": 14.36, "grad_norm": 6.65625, "learning_rate": 1.9330765519147058e-05, "loss": 1.5153, "step": 5744 }, { "epoch": 14.3625, "grad_norm": 7.28125, "learning_rate": 1.933047426143286e-05, "loss": 1.5429, "step": 5745 }, { "epoch": 14.365, "grad_norm": 7.59375, "learning_rate": 1.9330182942548446e-05, "loss": 1.5624, "step": 5746 }, { "epoch": 14.3675, "grad_norm": 6.40625, "learning_rate": 1.9329891562495723e-05, "loss": 1.3678, "step": 5747 }, { "epoch": 14.37, "grad_norm": 9.9375, "learning_rate": 1.9329600121276607e-05, "loss": 1.4579, "step": 5748 }, { "epoch": 14.3725, "grad_norm": 6.96875, "learning_rate": 1.9329308618893005e-05, "loss": 1.5342, "step": 5749 }, { "epoch": 14.375, "grad_norm": 4.5, "learning_rate": 1.932901705534683e-05, "loss": 1.3361, "step": 5750 }, { "epoch": 14.3775, "grad_norm": 4.71875, "learning_rate": 1.932872543063999e-05, "loss": 1.4129, "step": 5751 }, { "epoch": 14.38, "grad_norm": 4.875, "learning_rate": 1.9328433744774403e-05, "loss": 1.5203, "step": 5752 }, { "epoch": 14.3825, "grad_norm": 5.9375, "learning_rate": 1.9328141997751975e-05, "loss": 1.3747, "step": 5753 }, { "epoch": 14.385, "grad_norm": 9.125, "learning_rate": 1.932785018957462e-05, "loss": 1.4648, "step": 5754 }, { "epoch": 14.3875, "grad_norm": 6.84375, "learning_rate": 1.9327558320244254e-05, "loss": 1.2562, "step": 5755 }, { "epoch": 14.39, "grad_norm": 7.125, "learning_rate": 1.9327266389762787e-05, "loss": 1.4052, "step": 5756 }, { "epoch": 14.3925, "grad_norm": 5.34375, "learning_rate": 1.932697439813214e-05, "loss": 1.4901, "step": 5757 }, { "epoch": 14.395, "grad_norm": 8.25, "learning_rate": 1.9326682345354218e-05, "loss": 1.5288, "step": 5758 }, { "epoch": 14.3975, "grad_norm": 8.4375, "learning_rate": 1.9326390231430944e-05, "loss": 1.5835, "step": 5759 }, { "epoch": 14.4, "grad_norm": 7.1875, "learning_rate": 1.9326098056364224e-05, "loss": 1.411, "step": 5760 }, { "epoch": 14.4025, "grad_norm": 5.78125, "learning_rate": 1.932580582015598e-05, "loss": 1.4585, "step": 5761 }, { "epoch": 14.405, "grad_norm": 6.375, "learning_rate": 1.9325513522808127e-05, "loss": 1.3625, "step": 5762 }, { "epoch": 14.4075, "grad_norm": 6.78125, "learning_rate": 1.932522116432258e-05, "loss": 1.3518, "step": 5763 }, { "epoch": 14.41, "grad_norm": 5.4375, "learning_rate": 1.9324928744701256e-05, "loss": 1.4393, "step": 5764 }, { "epoch": 14.4125, "grad_norm": 7.375, "learning_rate": 1.9324636263946074e-05, "loss": 1.509, "step": 5765 }, { "epoch": 14.415, "grad_norm": 4.65625, "learning_rate": 1.9324343722058946e-05, "loss": 1.4761, "step": 5766 }, { "epoch": 14.4175, "grad_norm": 8.5, "learning_rate": 1.9324051119041796e-05, "loss": 1.3402, "step": 5767 }, { "epoch": 14.42, "grad_norm": 4.78125, "learning_rate": 1.932375845489654e-05, "loss": 1.5303, "step": 5768 }, { "epoch": 14.4225, "grad_norm": 8.3125, "learning_rate": 1.93234657296251e-05, "loss": 1.5063, "step": 5769 }, { "epoch": 14.425, "grad_norm": 8.3125, "learning_rate": 1.9323172943229386e-05, "loss": 1.6759, "step": 5770 }, { "epoch": 14.4275, "grad_norm": 6.0625, "learning_rate": 1.9322880095711328e-05, "loss": 1.3286, "step": 5771 }, { "epoch": 14.43, "grad_norm": 9.625, "learning_rate": 1.9322587187072835e-05, "loss": 1.5119, "step": 5772 }, { "epoch": 14.4325, "grad_norm": 6.21875, "learning_rate": 1.932229421731584e-05, "loss": 1.5218, "step": 5773 }, { "epoch": 14.435, "grad_norm": 4.6875, "learning_rate": 1.9322001186442253e-05, "loss": 1.2838, "step": 5774 }, { "epoch": 14.4375, "grad_norm": 6.4375, "learning_rate": 1.9321708094454003e-05, "loss": 1.4904, "step": 5775 }, { "epoch": 14.44, "grad_norm": 4.40625, "learning_rate": 1.9321414941353006e-05, "loss": 1.4109, "step": 5776 }, { "epoch": 14.4425, "grad_norm": 5.1875, "learning_rate": 1.9321121727141184e-05, "loss": 1.4941, "step": 5777 }, { "epoch": 14.445, "grad_norm": 8.5625, "learning_rate": 1.9320828451820462e-05, "loss": 1.4025, "step": 5778 }, { "epoch": 14.4475, "grad_norm": 6.21875, "learning_rate": 1.9320535115392764e-05, "loss": 1.4762, "step": 5779 }, { "epoch": 14.45, "grad_norm": 5.84375, "learning_rate": 1.9320241717860007e-05, "loss": 1.3756, "step": 5780 }, { "epoch": 14.4525, "grad_norm": 6.0, "learning_rate": 1.931994825922412e-05, "loss": 1.4375, "step": 5781 }, { "epoch": 14.455, "grad_norm": 5.34375, "learning_rate": 1.9319654739487025e-05, "loss": 1.4263, "step": 5782 }, { "epoch": 14.4575, "grad_norm": 6.5625, "learning_rate": 1.9319361158650647e-05, "loss": 1.5259, "step": 5783 }, { "epoch": 14.46, "grad_norm": 5.4375, "learning_rate": 1.931906751671691e-05, "loss": 1.5043, "step": 5784 }, { "epoch": 14.4625, "grad_norm": 4.8125, "learning_rate": 1.9318773813687735e-05, "loss": 1.373, "step": 5785 }, { "epoch": 14.465, "grad_norm": 5.78125, "learning_rate": 1.9318480049565058e-05, "loss": 1.3912, "step": 5786 }, { "epoch": 14.4675, "grad_norm": 5.4375, "learning_rate": 1.9318186224350797e-05, "loss": 1.4333, "step": 5787 }, { "epoch": 14.47, "grad_norm": 7.1875, "learning_rate": 1.931789233804688e-05, "loss": 1.5466, "step": 5788 }, { "epoch": 14.4725, "grad_norm": 7.5625, "learning_rate": 1.931759839065523e-05, "loss": 1.4087, "step": 5789 }, { "epoch": 14.475, "grad_norm": 13.3125, "learning_rate": 1.9317304382177783e-05, "loss": 1.415, "step": 5790 }, { "epoch": 14.4775, "grad_norm": 8.1875, "learning_rate": 1.931701031261646e-05, "loss": 1.5984, "step": 5791 }, { "epoch": 14.48, "grad_norm": 8.875, "learning_rate": 1.931671618197319e-05, "loss": 1.4336, "step": 5792 }, { "epoch": 14.4825, "grad_norm": 10.75, "learning_rate": 1.93164219902499e-05, "loss": 1.3887, "step": 5793 }, { "epoch": 14.485, "grad_norm": 8.0625, "learning_rate": 1.9316127737448524e-05, "loss": 1.4282, "step": 5794 }, { "epoch": 14.4875, "grad_norm": 5.15625, "learning_rate": 1.931583342357098e-05, "loss": 1.467, "step": 5795 }, { "epoch": 14.49, "grad_norm": 6.40625, "learning_rate": 1.9315539048619212e-05, "loss": 1.5746, "step": 5796 }, { "epoch": 14.4925, "grad_norm": 6.125, "learning_rate": 1.931524461259514e-05, "loss": 1.4036, "step": 5797 }, { "epoch": 14.495, "grad_norm": 9.4375, "learning_rate": 1.9314950115500698e-05, "loss": 1.6811, "step": 5798 }, { "epoch": 14.4975, "grad_norm": 12.8125, "learning_rate": 1.9314655557337817e-05, "loss": 1.476, "step": 5799 }, { "epoch": 14.5, "grad_norm": 7.96875, "learning_rate": 1.9314360938108427e-05, "loss": 1.4055, "step": 5800 }, { "epoch": 14.5025, "grad_norm": 6.53125, "learning_rate": 1.9314066257814453e-05, "loss": 1.4472, "step": 5801 }, { "epoch": 14.505, "grad_norm": 5.8125, "learning_rate": 1.931377151645784e-05, "loss": 1.3985, "step": 5802 }, { "epoch": 14.5075, "grad_norm": 7.0625, "learning_rate": 1.9313476714040513e-05, "loss": 1.6953, "step": 5803 }, { "epoch": 14.51, "grad_norm": 5.21875, "learning_rate": 1.9313181850564402e-05, "loss": 1.4007, "step": 5804 }, { "epoch": 14.5125, "grad_norm": 5.46875, "learning_rate": 1.9312886926031446e-05, "loss": 1.4548, "step": 5805 }, { "epoch": 14.515, "grad_norm": 6.09375, "learning_rate": 1.9312591940443575e-05, "loss": 1.4447, "step": 5806 }, { "epoch": 14.5175, "grad_norm": 5.625, "learning_rate": 1.9312296893802726e-05, "loss": 1.5179, "step": 5807 }, { "epoch": 14.52, "grad_norm": 4.6875, "learning_rate": 1.931200178611083e-05, "loss": 1.4856, "step": 5808 }, { "epoch": 14.5225, "grad_norm": 6.6875, "learning_rate": 1.931170661736982e-05, "loss": 1.5997, "step": 5809 }, { "epoch": 14.525, "grad_norm": 5.53125, "learning_rate": 1.931141138758164e-05, "loss": 1.5004, "step": 5810 }, { "epoch": 14.5275, "grad_norm": 7.75, "learning_rate": 1.9311116096748214e-05, "loss": 1.617, "step": 5811 }, { "epoch": 14.53, "grad_norm": 4.1875, "learning_rate": 1.9310820744871486e-05, "loss": 1.4034, "step": 5812 }, { "epoch": 14.5325, "grad_norm": 7.375, "learning_rate": 1.931052533195339e-05, "loss": 1.4118, "step": 5813 }, { "epoch": 14.535, "grad_norm": 5.03125, "learning_rate": 1.931022985799586e-05, "loss": 1.3924, "step": 5814 }, { "epoch": 14.5375, "grad_norm": 6.84375, "learning_rate": 1.9309934323000838e-05, "loss": 1.4356, "step": 5815 }, { "epoch": 14.54, "grad_norm": 7.5, "learning_rate": 1.930963872697026e-05, "loss": 1.5098, "step": 5816 }, { "epoch": 14.5425, "grad_norm": 8.375, "learning_rate": 1.9309343069906062e-05, "loss": 1.3696, "step": 5817 }, { "epoch": 14.545, "grad_norm": 9.9375, "learning_rate": 1.9309047351810184e-05, "loss": 1.48, "step": 5818 }, { "epoch": 14.5475, "grad_norm": 5.625, "learning_rate": 1.9308751572684562e-05, "loss": 1.5547, "step": 5819 }, { "epoch": 14.55, "grad_norm": 5.625, "learning_rate": 1.930845573253114e-05, "loss": 1.225, "step": 5820 }, { "epoch": 14.5525, "grad_norm": 7.71875, "learning_rate": 1.9308159831351854e-05, "loss": 1.4829, "step": 5821 }, { "epoch": 14.555, "grad_norm": 9.9375, "learning_rate": 1.9307863869148643e-05, "loss": 1.612, "step": 5822 }, { "epoch": 14.5575, "grad_norm": 4.96875, "learning_rate": 1.930756784592345e-05, "loss": 1.489, "step": 5823 }, { "epoch": 14.56, "grad_norm": 5.25, "learning_rate": 1.9307271761678214e-05, "loss": 1.5704, "step": 5824 }, { "epoch": 14.5625, "grad_norm": 8.6875, "learning_rate": 1.9306975616414876e-05, "loss": 1.5454, "step": 5825 }, { "epoch": 14.565, "grad_norm": 6.0, "learning_rate": 1.930667941013538e-05, "loss": 1.4586, "step": 5826 }, { "epoch": 14.5675, "grad_norm": 7.6875, "learning_rate": 1.9306383142841663e-05, "loss": 1.4524, "step": 5827 }, { "epoch": 14.57, "grad_norm": 6.75, "learning_rate": 1.9306086814535678e-05, "loss": 1.453, "step": 5828 }, { "epoch": 14.5725, "grad_norm": 7.78125, "learning_rate": 1.9305790425219352e-05, "loss": 1.4694, "step": 5829 }, { "epoch": 14.575, "grad_norm": 6.625, "learning_rate": 1.9305493974894642e-05, "loss": 1.4994, "step": 5830 }, { "epoch": 14.5775, "grad_norm": 6.0, "learning_rate": 1.9305197463563484e-05, "loss": 1.6644, "step": 5831 }, { "epoch": 14.58, "grad_norm": 5.15625, "learning_rate": 1.9304900891227825e-05, "loss": 1.3365, "step": 5832 }, { "epoch": 14.5825, "grad_norm": 5.03125, "learning_rate": 1.9304604257889605e-05, "loss": 1.5282, "step": 5833 }, { "epoch": 14.585, "grad_norm": 8.8125, "learning_rate": 1.9304307563550773e-05, "loss": 1.5516, "step": 5834 }, { "epoch": 14.5875, "grad_norm": 8.125, "learning_rate": 1.9304010808213272e-05, "loss": 1.3859, "step": 5835 }, { "epoch": 14.59, "grad_norm": 8.25, "learning_rate": 1.9303713991879052e-05, "loss": 1.4832, "step": 5836 }, { "epoch": 14.5925, "grad_norm": 5.4375, "learning_rate": 1.9303417114550054e-05, "loss": 1.502, "step": 5837 }, { "epoch": 14.595, "grad_norm": 5.40625, "learning_rate": 1.9303120176228226e-05, "loss": 1.2145, "step": 5838 }, { "epoch": 14.5975, "grad_norm": 8.9375, "learning_rate": 1.9302823176915513e-05, "loss": 1.4768, "step": 5839 }, { "epoch": 14.6, "grad_norm": 6.21875, "learning_rate": 1.9302526116613863e-05, "loss": 1.5665, "step": 5840 }, { "epoch": 14.6025, "grad_norm": 10.5625, "learning_rate": 1.9302228995325224e-05, "loss": 1.679, "step": 5841 }, { "epoch": 14.605, "grad_norm": 5.28125, "learning_rate": 1.9301931813051548e-05, "loss": 1.443, "step": 5842 }, { "epoch": 14.6075, "grad_norm": 6.21875, "learning_rate": 1.9301634569794772e-05, "loss": 1.4104, "step": 5843 }, { "epoch": 14.61, "grad_norm": 5.5625, "learning_rate": 1.930133726555686e-05, "loss": 1.5461, "step": 5844 }, { "epoch": 14.6125, "grad_norm": 6.71875, "learning_rate": 1.930103990033975e-05, "loss": 1.4266, "step": 5845 }, { "epoch": 14.615, "grad_norm": 6.09375, "learning_rate": 1.930074247414539e-05, "loss": 1.5988, "step": 5846 }, { "epoch": 14.6175, "grad_norm": 6.125, "learning_rate": 1.9300444986975742e-05, "loss": 1.4414, "step": 5847 }, { "epoch": 14.62, "grad_norm": 5.0625, "learning_rate": 1.9300147438832747e-05, "loss": 1.5564, "step": 5848 }, { "epoch": 14.6225, "grad_norm": 8.125, "learning_rate": 1.9299849829718352e-05, "loss": 1.3927, "step": 5849 }, { "epoch": 14.625, "grad_norm": 5.75, "learning_rate": 1.929955215963452e-05, "loss": 1.3803, "step": 5850 }, { "epoch": 14.6275, "grad_norm": 5.8125, "learning_rate": 1.9299254428583195e-05, "loss": 1.2874, "step": 5851 }, { "epoch": 14.63, "grad_norm": 7.25, "learning_rate": 1.9298956636566326e-05, "loss": 1.4172, "step": 5852 }, { "epoch": 14.6325, "grad_norm": 8.5625, "learning_rate": 1.9298658783585874e-05, "loss": 1.4886, "step": 5853 }, { "epoch": 14.635, "grad_norm": 8.4375, "learning_rate": 1.9298360869643786e-05, "loss": 1.5537, "step": 5854 }, { "epoch": 14.6375, "grad_norm": 7.75, "learning_rate": 1.9298062894742014e-05, "loss": 1.5293, "step": 5855 }, { "epoch": 14.64, "grad_norm": 6.65625, "learning_rate": 1.9297764858882516e-05, "loss": 1.5212, "step": 5856 }, { "epoch": 14.6425, "grad_norm": 8.625, "learning_rate": 1.929746676206724e-05, "loss": 1.4801, "step": 5857 }, { "epoch": 14.645, "grad_norm": 7.0, "learning_rate": 1.929716860429815e-05, "loss": 1.6151, "step": 5858 }, { "epoch": 14.6475, "grad_norm": 6.875, "learning_rate": 1.9296870385577193e-05, "loss": 1.6031, "step": 5859 }, { "epoch": 14.65, "grad_norm": 5.53125, "learning_rate": 1.9296572105906323e-05, "loss": 1.6655, "step": 5860 }, { "epoch": 14.6525, "grad_norm": 7.1875, "learning_rate": 1.92962737652875e-05, "loss": 1.6584, "step": 5861 }, { "epoch": 14.655, "grad_norm": 6.53125, "learning_rate": 1.9295975363722678e-05, "loss": 1.4453, "step": 5862 }, { "epoch": 14.6575, "grad_norm": 6.5625, "learning_rate": 1.9295676901213813e-05, "loss": 1.3885, "step": 5863 }, { "epoch": 14.66, "grad_norm": 6.40625, "learning_rate": 1.9295378377762862e-05, "loss": 1.3357, "step": 5864 }, { "epoch": 14.6625, "grad_norm": 6.9375, "learning_rate": 1.9295079793371784e-05, "loss": 1.4394, "step": 5865 }, { "epoch": 14.665, "grad_norm": 6.125, "learning_rate": 1.9294781148042533e-05, "loss": 1.3034, "step": 5866 }, { "epoch": 14.6675, "grad_norm": 6.0625, "learning_rate": 1.929448244177707e-05, "loss": 1.4863, "step": 5867 }, { "epoch": 14.67, "grad_norm": 6.28125, "learning_rate": 1.9294183674577353e-05, "loss": 1.4353, "step": 5868 }, { "epoch": 14.6725, "grad_norm": 7.875, "learning_rate": 1.9293884846445338e-05, "loss": 1.4662, "step": 5869 }, { "epoch": 14.675, "grad_norm": 6.03125, "learning_rate": 1.9293585957382987e-05, "loss": 1.3956, "step": 5870 }, { "epoch": 14.6775, "grad_norm": 6.375, "learning_rate": 1.9293287007392258e-05, "loss": 1.453, "step": 5871 }, { "epoch": 14.68, "grad_norm": 4.71875, "learning_rate": 1.9292987996475113e-05, "loss": 1.4186, "step": 5872 }, { "epoch": 14.6825, "grad_norm": 8.6875, "learning_rate": 1.9292688924633508e-05, "loss": 1.4609, "step": 5873 }, { "epoch": 14.685, "grad_norm": 5.84375, "learning_rate": 1.9292389791869407e-05, "loss": 1.5464, "step": 5874 }, { "epoch": 14.6875, "grad_norm": 7.78125, "learning_rate": 1.9292090598184768e-05, "loss": 1.4688, "step": 5875 }, { "epoch": 14.69, "grad_norm": 6.0625, "learning_rate": 1.9291791343581557e-05, "loss": 1.3485, "step": 5876 }, { "epoch": 14.6925, "grad_norm": 9.4375, "learning_rate": 1.9291492028061735e-05, "loss": 1.567, "step": 5877 }, { "epoch": 14.695, "grad_norm": 6.09375, "learning_rate": 1.929119265162726e-05, "loss": 1.4135, "step": 5878 }, { "epoch": 14.6975, "grad_norm": 7.3125, "learning_rate": 1.92908932142801e-05, "loss": 1.5871, "step": 5879 }, { "epoch": 14.7, "grad_norm": 6.0, "learning_rate": 1.9290593716022218e-05, "loss": 1.3328, "step": 5880 }, { "epoch": 14.7025, "grad_norm": 7.59375, "learning_rate": 1.9290294156855575e-05, "loss": 1.5648, "step": 5881 }, { "epoch": 14.705, "grad_norm": 7.90625, "learning_rate": 1.928999453678213e-05, "loss": 1.4532, "step": 5882 }, { "epoch": 14.7075, "grad_norm": 7.15625, "learning_rate": 1.9289694855803858e-05, "loss": 1.4887, "step": 5883 }, { "epoch": 14.71, "grad_norm": 8.0625, "learning_rate": 1.9289395113922714e-05, "loss": 1.508, "step": 5884 }, { "epoch": 14.7125, "grad_norm": 4.5, "learning_rate": 1.928909531114067e-05, "loss": 1.2725, "step": 5885 }, { "epoch": 14.715, "grad_norm": 8.25, "learning_rate": 1.9288795447459693e-05, "loss": 1.4582, "step": 5886 }, { "epoch": 14.7175, "grad_norm": 9.5, "learning_rate": 1.9288495522881736e-05, "loss": 1.6181, "step": 5887 }, { "epoch": 14.72, "grad_norm": 7.625, "learning_rate": 1.928819553740878e-05, "loss": 1.5439, "step": 5888 }, { "epoch": 14.7225, "grad_norm": 5.9375, "learning_rate": 1.9287895491042782e-05, "loss": 1.3777, "step": 5889 }, { "epoch": 14.725, "grad_norm": 7.28125, "learning_rate": 1.9287595383785717e-05, "loss": 1.5539, "step": 5890 }, { "epoch": 14.7275, "grad_norm": 10.625, "learning_rate": 1.9287295215639546e-05, "loss": 1.3876, "step": 5891 }, { "epoch": 14.73, "grad_norm": 8.3125, "learning_rate": 1.928699498660624e-05, "loss": 1.4852, "step": 5892 }, { "epoch": 14.7325, "grad_norm": 7.03125, "learning_rate": 1.9286694696687767e-05, "loss": 1.4248, "step": 5893 }, { "epoch": 14.735, "grad_norm": 8.8125, "learning_rate": 1.9286394345886093e-05, "loss": 1.4591, "step": 5894 }, { "epoch": 14.7375, "grad_norm": 6.375, "learning_rate": 1.928609393420319e-05, "loss": 1.368, "step": 5895 }, { "epoch": 14.74, "grad_norm": 7.96875, "learning_rate": 1.928579346164103e-05, "loss": 1.4121, "step": 5896 }, { "epoch": 14.7425, "grad_norm": 6.4375, "learning_rate": 1.9285492928201575e-05, "loss": 1.4085, "step": 5897 }, { "epoch": 14.745, "grad_norm": 5.78125, "learning_rate": 1.9285192333886802e-05, "loss": 1.2822, "step": 5898 }, { "epoch": 14.7475, "grad_norm": 7.90625, "learning_rate": 1.928489167869868e-05, "loss": 1.4574, "step": 5899 }, { "epoch": 14.75, "grad_norm": 11.125, "learning_rate": 1.928459096263918e-05, "loss": 1.599, "step": 5900 }, { "epoch": 14.7525, "grad_norm": 8.9375, "learning_rate": 1.928429018571027e-05, "loss": 1.4837, "step": 5901 }, { "epoch": 14.755, "grad_norm": 10.4375, "learning_rate": 1.9283989347913928e-05, "loss": 1.4232, "step": 5902 }, { "epoch": 14.7575, "grad_norm": 4.90625, "learning_rate": 1.9283688449252123e-05, "loss": 1.3278, "step": 5903 }, { "epoch": 14.76, "grad_norm": 5.71875, "learning_rate": 1.9283387489726827e-05, "loss": 1.4572, "step": 5904 }, { "epoch": 14.7625, "grad_norm": 6.1875, "learning_rate": 1.9283086469340014e-05, "loss": 1.6176, "step": 5905 }, { "epoch": 14.765, "grad_norm": 7.3125, "learning_rate": 1.928278538809366e-05, "loss": 1.4696, "step": 5906 }, { "epoch": 14.7675, "grad_norm": 6.53125, "learning_rate": 1.9282484245989735e-05, "loss": 1.5972, "step": 5907 }, { "epoch": 14.77, "grad_norm": 9.1875, "learning_rate": 1.9282183043030216e-05, "loss": 1.625, "step": 5908 }, { "epoch": 14.7725, "grad_norm": 10.25, "learning_rate": 1.9281881779217074e-05, "loss": 1.4604, "step": 5909 }, { "epoch": 14.775, "grad_norm": 4.4375, "learning_rate": 1.928158045455229e-05, "loss": 1.4186, "step": 5910 }, { "epoch": 14.7775, "grad_norm": 12.4375, "learning_rate": 1.9281279069037834e-05, "loss": 1.4816, "step": 5911 }, { "epoch": 14.78, "grad_norm": 6.65625, "learning_rate": 1.928097762267568e-05, "loss": 1.4075, "step": 5912 }, { "epoch": 14.7825, "grad_norm": 8.75, "learning_rate": 1.9280676115467814e-05, "loss": 1.352, "step": 5913 }, { "epoch": 14.785, "grad_norm": 5.0625, "learning_rate": 1.9280374547416203e-05, "loss": 1.4787, "step": 5914 }, { "epoch": 14.7875, "grad_norm": 6.875, "learning_rate": 1.9280072918522828e-05, "loss": 1.5432, "step": 5915 }, { "epoch": 14.79, "grad_norm": 5.5, "learning_rate": 1.927977122878967e-05, "loss": 1.3489, "step": 5916 }, { "epoch": 14.7925, "grad_norm": 9.6875, "learning_rate": 1.92794694782187e-05, "loss": 1.5072, "step": 5917 }, { "epoch": 14.795, "grad_norm": 7.78125, "learning_rate": 1.9279167666811898e-05, "loss": 1.6422, "step": 5918 }, { "epoch": 14.7975, "grad_norm": 6.3125, "learning_rate": 1.927886579457125e-05, "loss": 1.3167, "step": 5919 }, { "epoch": 14.8, "grad_norm": 7.84375, "learning_rate": 1.9278563861498726e-05, "loss": 1.5285, "step": 5920 }, { "epoch": 14.8025, "grad_norm": 6.75, "learning_rate": 1.9278261867596307e-05, "loss": 1.3661, "step": 5921 }, { "epoch": 14.805, "grad_norm": 7.03125, "learning_rate": 1.9277959812865975e-05, "loss": 1.4488, "step": 5922 }, { "epoch": 14.8075, "grad_norm": 10.125, "learning_rate": 1.9277657697309712e-05, "loss": 1.4584, "step": 5923 }, { "epoch": 14.81, "grad_norm": 5.78125, "learning_rate": 1.9277355520929494e-05, "loss": 1.4437, "step": 5924 }, { "epoch": 14.8125, "grad_norm": 7.09375, "learning_rate": 1.9277053283727306e-05, "loss": 1.5347, "step": 5925 }, { "epoch": 14.815, "grad_norm": 6.84375, "learning_rate": 1.9276750985705127e-05, "loss": 1.5921, "step": 5926 }, { "epoch": 14.8175, "grad_norm": 5.6875, "learning_rate": 1.927644862686494e-05, "loss": 1.3091, "step": 5927 }, { "epoch": 14.82, "grad_norm": 4.84375, "learning_rate": 1.9276146207208727e-05, "loss": 1.4018, "step": 5928 }, { "epoch": 14.8225, "grad_norm": 8.8125, "learning_rate": 1.927584372673847e-05, "loss": 1.5917, "step": 5929 }, { "epoch": 14.825, "grad_norm": 6.84375, "learning_rate": 1.9275541185456154e-05, "loss": 1.4349, "step": 5930 }, { "epoch": 14.8275, "grad_norm": 5.6875, "learning_rate": 1.927523858336376e-05, "loss": 1.4973, "step": 5931 }, { "epoch": 14.83, "grad_norm": 9.75, "learning_rate": 1.9274935920463273e-05, "loss": 1.4213, "step": 5932 }, { "epoch": 14.8325, "grad_norm": 4.9375, "learning_rate": 1.9274633196756677e-05, "loss": 1.5552, "step": 5933 }, { "epoch": 14.835, "grad_norm": 6.84375, "learning_rate": 1.927433041224596e-05, "loss": 1.3805, "step": 5934 }, { "epoch": 14.8375, "grad_norm": 9.4375, "learning_rate": 1.9274027566933103e-05, "loss": 1.4213, "step": 5935 }, { "epoch": 14.84, "grad_norm": 7.71875, "learning_rate": 1.9273724660820086e-05, "loss": 1.3776, "step": 5936 }, { "epoch": 14.8425, "grad_norm": 5.53125, "learning_rate": 1.9273421693908908e-05, "loss": 1.265, "step": 5937 }, { "epoch": 14.845, "grad_norm": 10.9375, "learning_rate": 1.927311866620155e-05, "loss": 1.5738, "step": 5938 }, { "epoch": 14.8475, "grad_norm": 7.34375, "learning_rate": 1.927281557769999e-05, "loss": 1.5478, "step": 5939 }, { "epoch": 14.85, "grad_norm": 7.625, "learning_rate": 1.927251242840623e-05, "loss": 1.5967, "step": 5940 }, { "epoch": 14.8525, "grad_norm": 8.6875, "learning_rate": 1.9272209218322244e-05, "loss": 1.6397, "step": 5941 }, { "epoch": 14.855, "grad_norm": 5.21875, "learning_rate": 1.9271905947450026e-05, "loss": 1.4731, "step": 5942 }, { "epoch": 14.8575, "grad_norm": 7.96875, "learning_rate": 1.9271602615791566e-05, "loss": 1.6599, "step": 5943 }, { "epoch": 14.86, "grad_norm": 6.71875, "learning_rate": 1.927129922334885e-05, "loss": 1.5141, "step": 5944 }, { "epoch": 14.8625, "grad_norm": 6.46875, "learning_rate": 1.9270995770123865e-05, "loss": 1.3124, "step": 5945 }, { "epoch": 14.865, "grad_norm": 6.25, "learning_rate": 1.9270692256118602e-05, "loss": 1.3863, "step": 5946 }, { "epoch": 14.8675, "grad_norm": 7.6875, "learning_rate": 1.9270388681335056e-05, "loss": 1.6552, "step": 5947 }, { "epoch": 14.87, "grad_norm": 9.5625, "learning_rate": 1.927008504577521e-05, "loss": 1.41, "step": 5948 }, { "epoch": 14.8725, "grad_norm": 10.3125, "learning_rate": 1.9269781349441058e-05, "loss": 1.3871, "step": 5949 }, { "epoch": 14.875, "grad_norm": 6.9375, "learning_rate": 1.926947759233459e-05, "loss": 1.4515, "step": 5950 }, { "epoch": 14.8775, "grad_norm": 9.1875, "learning_rate": 1.9269173774457797e-05, "loss": 1.6441, "step": 5951 }, { "epoch": 14.88, "grad_norm": 9.0, "learning_rate": 1.9268869895812673e-05, "loss": 1.4281, "step": 5952 }, { "epoch": 14.8825, "grad_norm": 8.5, "learning_rate": 1.926856595640121e-05, "loss": 1.4006, "step": 5953 }, { "epoch": 14.885, "grad_norm": 6.09375, "learning_rate": 1.9268261956225397e-05, "loss": 1.3995, "step": 5954 }, { "epoch": 14.8875, "grad_norm": 6.84375, "learning_rate": 1.9267957895287233e-05, "loss": 1.3951, "step": 5955 }, { "epoch": 14.89, "grad_norm": 8.4375, "learning_rate": 1.9267653773588702e-05, "loss": 1.5581, "step": 5956 }, { "epoch": 14.8925, "grad_norm": 8.8125, "learning_rate": 1.926734959113181e-05, "loss": 1.6056, "step": 5957 }, { "epoch": 14.895, "grad_norm": 9.8125, "learning_rate": 1.926704534791854e-05, "loss": 1.3559, "step": 5958 }, { "epoch": 14.8975, "grad_norm": 8.3125, "learning_rate": 1.9266741043950895e-05, "loss": 1.5713, "step": 5959 }, { "epoch": 14.9, "grad_norm": 6.75, "learning_rate": 1.9266436679230866e-05, "loss": 1.4004, "step": 5960 }, { "epoch": 14.9025, "grad_norm": 8.625, "learning_rate": 1.926613225376045e-05, "loss": 1.4079, "step": 5961 }, { "epoch": 14.905, "grad_norm": 8.3125, "learning_rate": 1.9265827767541638e-05, "loss": 1.5048, "step": 5962 }, { "epoch": 14.9075, "grad_norm": 7.21875, "learning_rate": 1.9265523220576433e-05, "loss": 1.7038, "step": 5963 }, { "epoch": 14.91, "grad_norm": 5.25, "learning_rate": 1.926521861286683e-05, "loss": 1.4096, "step": 5964 }, { "epoch": 14.9125, "grad_norm": 10.0625, "learning_rate": 1.926491394441482e-05, "loss": 1.5318, "step": 5965 }, { "epoch": 14.915, "grad_norm": 5.96875, "learning_rate": 1.9264609215222405e-05, "loss": 1.5126, "step": 5966 }, { "epoch": 14.9175, "grad_norm": 5.75, "learning_rate": 1.9264304425291586e-05, "loss": 1.3737, "step": 5967 }, { "epoch": 14.92, "grad_norm": 4.25, "learning_rate": 1.9263999574624357e-05, "loss": 1.4054, "step": 5968 }, { "epoch": 14.9225, "grad_norm": 7.1875, "learning_rate": 1.9263694663222715e-05, "loss": 1.3694, "step": 5969 }, { "epoch": 14.925, "grad_norm": 6.375, "learning_rate": 1.9263389691088665e-05, "loss": 1.5284, "step": 5970 }, { "epoch": 14.9275, "grad_norm": 6.9375, "learning_rate": 1.92630846582242e-05, "loss": 1.4898, "step": 5971 }, { "epoch": 14.93, "grad_norm": 6.0, "learning_rate": 1.9262779564631326e-05, "loss": 1.3933, "step": 5972 }, { "epoch": 14.9325, "grad_norm": 5.125, "learning_rate": 1.926247441031204e-05, "loss": 1.3893, "step": 5973 }, { "epoch": 14.935, "grad_norm": 7.09375, "learning_rate": 1.926216919526834e-05, "loss": 1.5771, "step": 5974 }, { "epoch": 14.9375, "grad_norm": 9.125, "learning_rate": 1.9261863919502228e-05, "loss": 1.5956, "step": 5975 }, { "epoch": 14.94, "grad_norm": 7.0625, "learning_rate": 1.926155858301571e-05, "loss": 1.3892, "step": 5976 }, { "epoch": 14.9425, "grad_norm": 7.8125, "learning_rate": 1.926125318581078e-05, "loss": 1.5017, "step": 5977 }, { "epoch": 14.945, "grad_norm": 6.625, "learning_rate": 1.926094772788945e-05, "loss": 1.606, "step": 5978 }, { "epoch": 14.9475, "grad_norm": 10.125, "learning_rate": 1.9260642209253718e-05, "loss": 1.6582, "step": 5979 }, { "epoch": 14.95, "grad_norm": 5.96875, "learning_rate": 1.926033662990558e-05, "loss": 1.5435, "step": 5980 }, { "epoch": 14.9525, "grad_norm": 7.1875, "learning_rate": 1.9260030989847053e-05, "loss": 1.4714, "step": 5981 }, { "epoch": 14.955, "grad_norm": 5.4375, "learning_rate": 1.9259725289080127e-05, "loss": 1.3904, "step": 5982 }, { "epoch": 14.9575, "grad_norm": 6.0625, "learning_rate": 1.9259419527606814e-05, "loss": 1.3931, "step": 5983 }, { "epoch": 14.96, "grad_norm": 4.6875, "learning_rate": 1.925911370542912e-05, "loss": 1.5632, "step": 5984 }, { "epoch": 14.9625, "grad_norm": 5.0, "learning_rate": 1.9258807822549045e-05, "loss": 1.3646, "step": 5985 }, { "epoch": 14.965, "grad_norm": 7.5, "learning_rate": 1.9258501878968595e-05, "loss": 1.3828, "step": 5986 }, { "epoch": 14.9675, "grad_norm": 6.0625, "learning_rate": 1.925819587468978e-05, "loss": 1.5436, "step": 5987 }, { "epoch": 14.97, "grad_norm": 6.6875, "learning_rate": 1.9257889809714604e-05, "loss": 1.4926, "step": 5988 }, { "epoch": 14.9725, "grad_norm": 10.875, "learning_rate": 1.9257583684045072e-05, "loss": 1.4495, "step": 5989 }, { "epoch": 14.975, "grad_norm": 7.75, "learning_rate": 1.925727749768319e-05, "loss": 1.4053, "step": 5990 }, { "epoch": 14.9775, "grad_norm": 7.65625, "learning_rate": 1.925697125063097e-05, "loss": 1.5126, "step": 5991 }, { "epoch": 14.98, "grad_norm": 5.65625, "learning_rate": 1.9256664942890412e-05, "loss": 1.3196, "step": 5992 }, { "epoch": 14.9825, "grad_norm": 7.3125, "learning_rate": 1.9256358574463534e-05, "loss": 1.5982, "step": 5993 }, { "epoch": 14.985, "grad_norm": 6.78125, "learning_rate": 1.9256052145352337e-05, "loss": 1.4625, "step": 5994 }, { "epoch": 14.9875, "grad_norm": 5.53125, "learning_rate": 1.9255745655558834e-05, "loss": 1.3891, "step": 5995 }, { "epoch": 14.99, "grad_norm": 8.9375, "learning_rate": 1.925543910508503e-05, "loss": 1.5131, "step": 5996 }, { "epoch": 14.9925, "grad_norm": 7.15625, "learning_rate": 1.925513249393294e-05, "loss": 1.399, "step": 5997 }, { "epoch": 14.995, "grad_norm": 6.75, "learning_rate": 1.9254825822104574e-05, "loss": 1.4111, "step": 5998 }, { "epoch": 14.9975, "grad_norm": 10.0625, "learning_rate": 1.925451908960194e-05, "loss": 1.5263, "step": 5999 }, { "epoch": 15.0, "grad_norm": 7.59375, "learning_rate": 1.9254212296427043e-05, "loss": 1.3612, "step": 6000 }, { "epoch": 15.0025, "grad_norm": 6.5625, "learning_rate": 1.9253905442581904e-05, "loss": 1.3969, "step": 6001 }, { "epoch": 15.005, "grad_norm": 6.8125, "learning_rate": 1.925359852806853e-05, "loss": 1.6007, "step": 6002 }, { "epoch": 15.0075, "grad_norm": 5.125, "learning_rate": 1.9253291552888938e-05, "loss": 1.6041, "step": 6003 }, { "epoch": 15.01, "grad_norm": 9.125, "learning_rate": 1.9252984517045136e-05, "loss": 1.3649, "step": 6004 }, { "epoch": 15.0125, "grad_norm": 6.78125, "learning_rate": 1.9252677420539136e-05, "loss": 1.6032, "step": 6005 }, { "epoch": 15.015, "grad_norm": 7.125, "learning_rate": 1.9252370263372955e-05, "loss": 1.4581, "step": 6006 }, { "epoch": 15.0175, "grad_norm": 5.875, "learning_rate": 1.9252063045548604e-05, "loss": 1.4912, "step": 6007 }, { "epoch": 15.02, "grad_norm": 7.96875, "learning_rate": 1.9251755767068095e-05, "loss": 1.4695, "step": 6008 }, { "epoch": 15.0225, "grad_norm": 5.8125, "learning_rate": 1.925144842793345e-05, "loss": 1.4074, "step": 6009 }, { "epoch": 15.025, "grad_norm": 7.5, "learning_rate": 1.925114102814668e-05, "loss": 1.412, "step": 6010 }, { "epoch": 15.0275, "grad_norm": 6.75, "learning_rate": 1.9250833567709795e-05, "loss": 1.521, "step": 6011 }, { "epoch": 15.03, "grad_norm": 12.0625, "learning_rate": 1.9250526046624816e-05, "loss": 1.6161, "step": 6012 }, { "epoch": 15.0325, "grad_norm": 5.4375, "learning_rate": 1.9250218464893762e-05, "loss": 1.5438, "step": 6013 }, { "epoch": 15.035, "grad_norm": 6.75, "learning_rate": 1.9249910822518642e-05, "loss": 1.6973, "step": 6014 }, { "epoch": 15.0375, "grad_norm": 5.9375, "learning_rate": 1.924960311950148e-05, "loss": 1.5439, "step": 6015 }, { "epoch": 15.04, "grad_norm": 7.96875, "learning_rate": 1.9249295355844286e-05, "loss": 1.5667, "step": 6016 }, { "epoch": 15.0425, "grad_norm": 5.4375, "learning_rate": 1.9248987531549084e-05, "loss": 1.5242, "step": 6017 }, { "epoch": 15.045, "grad_norm": 6.40625, "learning_rate": 1.924867964661789e-05, "loss": 1.465, "step": 6018 }, { "epoch": 15.0475, "grad_norm": 5.25, "learning_rate": 1.9248371701052725e-05, "loss": 1.4306, "step": 6019 }, { "epoch": 15.05, "grad_norm": 6.9375, "learning_rate": 1.9248063694855603e-05, "loss": 1.285, "step": 6020 }, { "epoch": 15.0525, "grad_norm": 5.0625, "learning_rate": 1.9247755628028546e-05, "loss": 1.4479, "step": 6021 }, { "epoch": 15.055, "grad_norm": 5.21875, "learning_rate": 1.924744750057357e-05, "loss": 1.3868, "step": 6022 }, { "epoch": 15.0575, "grad_norm": 6.59375, "learning_rate": 1.92471393124927e-05, "loss": 1.2677, "step": 6023 }, { "epoch": 15.06, "grad_norm": 7.375, "learning_rate": 1.9246831063787956e-05, "loss": 1.6152, "step": 6024 }, { "epoch": 15.0625, "grad_norm": 5.625, "learning_rate": 1.924652275446136e-05, "loss": 1.5148, "step": 6025 }, { "epoch": 15.065, "grad_norm": 7.4375, "learning_rate": 1.9246214384514927e-05, "loss": 1.564, "step": 6026 }, { "epoch": 15.0675, "grad_norm": 6.96875, "learning_rate": 1.9245905953950682e-05, "loss": 1.3751, "step": 6027 }, { "epoch": 15.07, "grad_norm": 10.125, "learning_rate": 1.924559746277065e-05, "loss": 1.4471, "step": 6028 }, { "epoch": 15.0725, "grad_norm": 6.65625, "learning_rate": 1.924528891097685e-05, "loss": 1.5001, "step": 6029 }, { "epoch": 15.075, "grad_norm": 5.78125, "learning_rate": 1.9244980298571306e-05, "loss": 1.3623, "step": 6030 }, { "epoch": 15.0775, "grad_norm": 9.5, "learning_rate": 1.924467162555604e-05, "loss": 1.4976, "step": 6031 }, { "epoch": 15.08, "grad_norm": 7.5625, "learning_rate": 1.9244362891933077e-05, "loss": 1.5267, "step": 6032 }, { "epoch": 15.0825, "grad_norm": 7.375, "learning_rate": 1.9244054097704443e-05, "loss": 1.4891, "step": 6033 }, { "epoch": 15.085, "grad_norm": 4.84375, "learning_rate": 1.924374524287216e-05, "loss": 1.3957, "step": 6034 }, { "epoch": 15.0875, "grad_norm": 6.6875, "learning_rate": 1.924343632743825e-05, "loss": 1.3985, "step": 6035 }, { "epoch": 15.09, "grad_norm": 7.0625, "learning_rate": 1.9243127351404743e-05, "loss": 1.502, "step": 6036 }, { "epoch": 15.0925, "grad_norm": 6.03125, "learning_rate": 1.9242818314773663e-05, "loss": 1.5146, "step": 6037 }, { "epoch": 15.095, "grad_norm": 6.875, "learning_rate": 1.9242509217547035e-05, "loss": 1.5557, "step": 6038 }, { "epoch": 15.0975, "grad_norm": 7.5625, "learning_rate": 1.9242200059726892e-05, "loss": 1.4703, "step": 6039 }, { "epoch": 15.1, "grad_norm": 5.0625, "learning_rate": 1.924189084131525e-05, "loss": 1.3733, "step": 6040 }, { "epoch": 15.1025, "grad_norm": 5.3125, "learning_rate": 1.9241581562314142e-05, "loss": 1.4582, "step": 6041 }, { "epoch": 15.105, "grad_norm": 6.90625, "learning_rate": 1.9241272222725594e-05, "loss": 1.5836, "step": 6042 }, { "epoch": 15.1075, "grad_norm": 6.375, "learning_rate": 1.924096282255164e-05, "loss": 1.4072, "step": 6043 }, { "epoch": 15.11, "grad_norm": 6.0, "learning_rate": 1.92406533617943e-05, "loss": 1.3234, "step": 6044 }, { "epoch": 15.1125, "grad_norm": 5.8125, "learning_rate": 1.9240343840455608e-05, "loss": 1.4983, "step": 6045 }, { "epoch": 15.115, "grad_norm": 7.125, "learning_rate": 1.9240034258537585e-05, "loss": 1.5095, "step": 6046 }, { "epoch": 15.1175, "grad_norm": 6.71875, "learning_rate": 1.9239724616042276e-05, "loss": 1.3513, "step": 6047 }, { "epoch": 15.12, "grad_norm": 7.375, "learning_rate": 1.9239414912971697e-05, "loss": 1.3829, "step": 6048 }, { "epoch": 15.1225, "grad_norm": 6.28125, "learning_rate": 1.9239105149327884e-05, "loss": 1.435, "step": 6049 }, { "epoch": 15.125, "grad_norm": 7.90625, "learning_rate": 1.9238795325112867e-05, "loss": 1.5318, "step": 6050 }, { "epoch": 15.1275, "grad_norm": 7.6875, "learning_rate": 1.9238485440328683e-05, "loss": 1.545, "step": 6051 }, { "epoch": 15.13, "grad_norm": 6.34375, "learning_rate": 1.923817549497735e-05, "loss": 1.5324, "step": 6052 }, { "epoch": 15.1325, "grad_norm": 6.15625, "learning_rate": 1.9237865489060915e-05, "loss": 1.4106, "step": 6053 }, { "epoch": 15.135, "grad_norm": 9.4375, "learning_rate": 1.92375554225814e-05, "loss": 1.5964, "step": 6054 }, { "epoch": 15.1375, "grad_norm": 6.15625, "learning_rate": 1.9237245295540843e-05, "loss": 1.392, "step": 6055 }, { "epoch": 15.14, "grad_norm": 5.875, "learning_rate": 1.9236935107941272e-05, "loss": 1.4023, "step": 6056 }, { "epoch": 15.1425, "grad_norm": 8.1875, "learning_rate": 1.923662485978473e-05, "loss": 1.5382, "step": 6057 }, { "epoch": 15.145, "grad_norm": 9.75, "learning_rate": 1.923631455107324e-05, "loss": 1.5297, "step": 6058 }, { "epoch": 15.1475, "grad_norm": 5.1875, "learning_rate": 1.9236004181808844e-05, "loss": 1.3651, "step": 6059 }, { "epoch": 15.15, "grad_norm": 7.40625, "learning_rate": 1.923569375199357e-05, "loss": 1.3785, "step": 6060 }, { "epoch": 15.1525, "grad_norm": 9.25, "learning_rate": 1.9235383261629462e-05, "loss": 1.4923, "step": 6061 }, { "epoch": 15.155, "grad_norm": 5.84375, "learning_rate": 1.923507271071855e-05, "loss": 1.454, "step": 6062 }, { "epoch": 15.1575, "grad_norm": 8.5, "learning_rate": 1.923476209926287e-05, "loss": 1.4555, "step": 6063 }, { "epoch": 15.16, "grad_norm": 10.25, "learning_rate": 1.923445142726446e-05, "loss": 1.3495, "step": 6064 }, { "epoch": 15.1625, "grad_norm": 8.0, "learning_rate": 1.9234140694725356e-05, "loss": 1.6113, "step": 6065 }, { "epoch": 15.165, "grad_norm": 9.0625, "learning_rate": 1.9233829901647597e-05, "loss": 1.4141, "step": 6066 }, { "epoch": 15.1675, "grad_norm": 10.5625, "learning_rate": 1.9233519048033216e-05, "loss": 1.4496, "step": 6067 }, { "epoch": 15.17, "grad_norm": 5.78125, "learning_rate": 1.9233208133884253e-05, "loss": 1.4771, "step": 6068 }, { "epoch": 15.1725, "grad_norm": 6.15625, "learning_rate": 1.9232897159202748e-05, "loss": 1.4701, "step": 6069 }, { "epoch": 15.175, "grad_norm": 6.28125, "learning_rate": 1.923258612399074e-05, "loss": 1.5313, "step": 6070 }, { "epoch": 15.1775, "grad_norm": 6.5625, "learning_rate": 1.9232275028250264e-05, "loss": 1.4263, "step": 6071 }, { "epoch": 15.18, "grad_norm": 15.5, "learning_rate": 1.9231963871983367e-05, "loss": 1.351, "step": 6072 }, { "epoch": 15.1825, "grad_norm": 5.40625, "learning_rate": 1.923165265519208e-05, "loss": 1.443, "step": 6073 }, { "epoch": 15.185, "grad_norm": 5.375, "learning_rate": 1.923134137787845e-05, "loss": 1.369, "step": 6074 }, { "epoch": 15.1875, "grad_norm": 6.90625, "learning_rate": 1.9231030040044514e-05, "loss": 1.5964, "step": 6075 }, { "epoch": 15.19, "grad_norm": 5.3125, "learning_rate": 1.923071864169231e-05, "loss": 1.4387, "step": 6076 }, { "epoch": 15.1925, "grad_norm": 7.5, "learning_rate": 1.923040718282389e-05, "loss": 1.6893, "step": 6077 }, { "epoch": 15.195, "grad_norm": 7.96875, "learning_rate": 1.923009566344129e-05, "loss": 1.5453, "step": 6078 }, { "epoch": 15.1975, "grad_norm": 4.875, "learning_rate": 1.9229784083546552e-05, "loss": 1.3132, "step": 6079 }, { "epoch": 15.2, "grad_norm": 6.34375, "learning_rate": 1.922947244314172e-05, "loss": 1.4451, "step": 6080 }, { "epoch": 15.2025, "grad_norm": 7.375, "learning_rate": 1.9229160742228828e-05, "loss": 1.3895, "step": 6081 }, { "epoch": 15.205, "grad_norm": 5.75, "learning_rate": 1.9228848980809935e-05, "loss": 1.4773, "step": 6082 }, { "epoch": 15.2075, "grad_norm": 5.375, "learning_rate": 1.9228537158887076e-05, "loss": 1.5303, "step": 6083 }, { "epoch": 15.21, "grad_norm": 8.0625, "learning_rate": 1.92282252764623e-05, "loss": 1.2885, "step": 6084 }, { "epoch": 15.2125, "grad_norm": 7.5, "learning_rate": 1.9227913333537642e-05, "loss": 1.3404, "step": 6085 }, { "epoch": 15.215, "grad_norm": 6.875, "learning_rate": 1.9227601330115155e-05, "loss": 1.6017, "step": 6086 }, { "epoch": 15.2175, "grad_norm": 5.6875, "learning_rate": 1.9227289266196883e-05, "loss": 1.534, "step": 6087 }, { "epoch": 15.22, "grad_norm": 6.625, "learning_rate": 1.9226977141784873e-05, "loss": 1.4913, "step": 6088 }, { "epoch": 15.2225, "grad_norm": 6.90625, "learning_rate": 1.9226664956881172e-05, "loss": 1.6013, "step": 6089 }, { "epoch": 15.225, "grad_norm": 6.0, "learning_rate": 1.9226352711487823e-05, "loss": 1.4778, "step": 6090 }, { "epoch": 15.2275, "grad_norm": 6.34375, "learning_rate": 1.9226040405606875e-05, "loss": 1.5438, "step": 6091 }, { "epoch": 15.23, "grad_norm": 7.59375, "learning_rate": 1.9225728039240375e-05, "loss": 1.5227, "step": 6092 }, { "epoch": 15.2325, "grad_norm": 8.3125, "learning_rate": 1.922541561239037e-05, "loss": 1.3108, "step": 6093 }, { "epoch": 15.235, "grad_norm": 8.875, "learning_rate": 1.922510312505891e-05, "loss": 1.4143, "step": 6094 }, { "epoch": 15.2375, "grad_norm": 7.21875, "learning_rate": 1.9224790577248046e-05, "loss": 1.5019, "step": 6095 }, { "epoch": 15.24, "grad_norm": 8.875, "learning_rate": 1.922447796895982e-05, "loss": 1.5707, "step": 6096 }, { "epoch": 15.2425, "grad_norm": 7.875, "learning_rate": 1.922416530019629e-05, "loss": 1.5341, "step": 6097 }, { "epoch": 15.245, "grad_norm": 7.46875, "learning_rate": 1.9223852570959498e-05, "loss": 1.3999, "step": 6098 }, { "epoch": 15.2475, "grad_norm": 12.5, "learning_rate": 1.9223539781251498e-05, "loss": 1.5515, "step": 6099 }, { "epoch": 15.25, "grad_norm": 7.25, "learning_rate": 1.922322693107434e-05, "loss": 1.4618, "step": 6100 }, { "epoch": 15.2525, "grad_norm": 7.09375, "learning_rate": 1.9222914020430076e-05, "loss": 1.5059, "step": 6101 }, { "epoch": 15.255, "grad_norm": 9.3125, "learning_rate": 1.9222601049320757e-05, "loss": 1.5533, "step": 6102 }, { "epoch": 15.2575, "grad_norm": 7.65625, "learning_rate": 1.9222288017748435e-05, "loss": 1.4268, "step": 6103 }, { "epoch": 15.26, "grad_norm": 7.90625, "learning_rate": 1.922197492571516e-05, "loss": 1.6984, "step": 6104 }, { "epoch": 15.2625, "grad_norm": 5.90625, "learning_rate": 1.9221661773222987e-05, "loss": 1.5723, "step": 6105 }, { "epoch": 15.265, "grad_norm": 6.8125, "learning_rate": 1.9221348560273968e-05, "loss": 1.4793, "step": 6106 }, { "epoch": 15.2675, "grad_norm": 9.75, "learning_rate": 1.922103528687016e-05, "loss": 1.4476, "step": 6107 }, { "epoch": 15.27, "grad_norm": 5.90625, "learning_rate": 1.922072195301361e-05, "loss": 1.2871, "step": 6108 }, { "epoch": 15.2725, "grad_norm": 5.0, "learning_rate": 1.922040855870638e-05, "loss": 1.5063, "step": 6109 }, { "epoch": 15.275, "grad_norm": 6.4375, "learning_rate": 1.9220095103950518e-05, "loss": 1.3394, "step": 6110 }, { "epoch": 15.2775, "grad_norm": 5.375, "learning_rate": 1.921978158874808e-05, "loss": 1.5747, "step": 6111 }, { "epoch": 15.28, "grad_norm": 7.5625, "learning_rate": 1.9219468013101123e-05, "loss": 1.4115, "step": 6112 }, { "epoch": 15.2825, "grad_norm": 9.0, "learning_rate": 1.9219154377011703e-05, "loss": 1.3323, "step": 6113 }, { "epoch": 15.285, "grad_norm": 5.125, "learning_rate": 1.921884068048188e-05, "loss": 1.4518, "step": 6114 }, { "epoch": 15.2875, "grad_norm": 5.5, "learning_rate": 1.9218526923513704e-05, "loss": 1.5234, "step": 6115 }, { "epoch": 15.29, "grad_norm": 8.4375, "learning_rate": 1.9218213106109234e-05, "loss": 1.3894, "step": 6116 }, { "epoch": 15.2925, "grad_norm": 7.3125, "learning_rate": 1.9217899228270526e-05, "loss": 1.3745, "step": 6117 }, { "epoch": 15.295, "grad_norm": 8.75, "learning_rate": 1.9217585289999642e-05, "loss": 1.5794, "step": 6118 }, { "epoch": 15.2975, "grad_norm": 6.3125, "learning_rate": 1.9217271291298636e-05, "loss": 1.4003, "step": 6119 }, { "epoch": 15.3, "grad_norm": 9.0, "learning_rate": 1.9216957232169567e-05, "loss": 1.4916, "step": 6120 }, { "epoch": 15.3025, "grad_norm": 9.625, "learning_rate": 1.92166431126145e-05, "loss": 1.3943, "step": 6121 }, { "epoch": 15.305, "grad_norm": 6.625, "learning_rate": 1.9216328932635484e-05, "loss": 1.5746, "step": 6122 }, { "epoch": 15.3075, "grad_norm": 8.5, "learning_rate": 1.9216014692234588e-05, "loss": 1.3298, "step": 6123 }, { "epoch": 15.31, "grad_norm": 6.25, "learning_rate": 1.9215700391413867e-05, "loss": 1.4063, "step": 6124 }, { "epoch": 15.3125, "grad_norm": 6.15625, "learning_rate": 1.9215386030175383e-05, "loss": 1.4796, "step": 6125 }, { "epoch": 15.315, "grad_norm": 6.8125, "learning_rate": 1.9215071608521198e-05, "loss": 1.4708, "step": 6126 }, { "epoch": 15.3175, "grad_norm": 9.4375, "learning_rate": 1.9214757126453373e-05, "loss": 1.3932, "step": 6127 }, { "epoch": 15.32, "grad_norm": 10.25, "learning_rate": 1.9214442583973965e-05, "loss": 1.5025, "step": 6128 }, { "epoch": 15.3225, "grad_norm": 5.65625, "learning_rate": 1.9214127981085046e-05, "loss": 1.5821, "step": 6129 }, { "epoch": 15.325, "grad_norm": 6.53125, "learning_rate": 1.921381331778867e-05, "loss": 1.4345, "step": 6130 }, { "epoch": 15.3275, "grad_norm": 3.921875, "learning_rate": 1.92134985940869e-05, "loss": 1.3764, "step": 6131 }, { "epoch": 15.33, "grad_norm": 5.96875, "learning_rate": 1.9213183809981806e-05, "loss": 1.4163, "step": 6132 }, { "epoch": 15.3325, "grad_norm": 7.625, "learning_rate": 1.9212868965475444e-05, "loss": 1.5173, "step": 6133 }, { "epoch": 15.335, "grad_norm": 4.5625, "learning_rate": 1.9212554060569883e-05, "loss": 1.3774, "step": 6134 }, { "epoch": 15.3375, "grad_norm": 8.0625, "learning_rate": 1.921223909526719e-05, "loss": 1.4975, "step": 6135 }, { "epoch": 15.34, "grad_norm": 7.5, "learning_rate": 1.9211924069569422e-05, "loss": 1.4494, "step": 6136 }, { "epoch": 15.3425, "grad_norm": 6.46875, "learning_rate": 1.9211608983478648e-05, "loss": 1.427, "step": 6137 }, { "epoch": 15.345, "grad_norm": 6.28125, "learning_rate": 1.9211293836996936e-05, "loss": 1.4576, "step": 6138 }, { "epoch": 15.3475, "grad_norm": 6.03125, "learning_rate": 1.921097863012635e-05, "loss": 1.3452, "step": 6139 }, { "epoch": 15.35, "grad_norm": 8.9375, "learning_rate": 1.9210663362868956e-05, "loss": 1.5536, "step": 6140 }, { "epoch": 15.3525, "grad_norm": 6.53125, "learning_rate": 1.921034803522682e-05, "loss": 1.3597, "step": 6141 }, { "epoch": 15.355, "grad_norm": 12.5625, "learning_rate": 1.9210032647202013e-05, "loss": 1.406, "step": 6142 }, { "epoch": 15.3575, "grad_norm": 5.5625, "learning_rate": 1.9209717198796603e-05, "loss": 1.3807, "step": 6143 }, { "epoch": 15.36, "grad_norm": 9.0, "learning_rate": 1.920940169001265e-05, "loss": 1.4598, "step": 6144 }, { "epoch": 15.3625, "grad_norm": 5.96875, "learning_rate": 1.9209086120852235e-05, "loss": 1.4571, "step": 6145 }, { "epoch": 15.365, "grad_norm": 5.8125, "learning_rate": 1.9208770491317415e-05, "loss": 1.444, "step": 6146 }, { "epoch": 15.3675, "grad_norm": 5.125, "learning_rate": 1.9208454801410267e-05, "loss": 1.5162, "step": 6147 }, { "epoch": 15.37, "grad_norm": 7.3125, "learning_rate": 1.920813905113285e-05, "loss": 1.4591, "step": 6148 }, { "epoch": 15.3725, "grad_norm": 5.3125, "learning_rate": 1.9207823240487252e-05, "loss": 1.3291, "step": 6149 }, { "epoch": 15.375, "grad_norm": 6.84375, "learning_rate": 1.920750736947553e-05, "loss": 1.3864, "step": 6150 }, { "epoch": 15.3775, "grad_norm": 4.78125, "learning_rate": 1.920719143809976e-05, "loss": 1.4279, "step": 6151 }, { "epoch": 15.38, "grad_norm": 6.59375, "learning_rate": 1.9206875446362005e-05, "loss": 1.4364, "step": 6152 }, { "epoch": 15.3825, "grad_norm": 8.1875, "learning_rate": 1.9206559394264345e-05, "loss": 1.4817, "step": 6153 }, { "epoch": 15.385, "grad_norm": 8.0, "learning_rate": 1.920624328180885e-05, "loss": 1.4515, "step": 6154 }, { "epoch": 15.3875, "grad_norm": 6.75, "learning_rate": 1.9205927108997598e-05, "loss": 1.6602, "step": 6155 }, { "epoch": 15.39, "grad_norm": 7.84375, "learning_rate": 1.920561087583265e-05, "loss": 1.5665, "step": 6156 }, { "epoch": 15.3925, "grad_norm": 6.5, "learning_rate": 1.920529458231609e-05, "loss": 1.4365, "step": 6157 }, { "epoch": 15.395, "grad_norm": 7.59375, "learning_rate": 1.9204978228449983e-05, "loss": 1.4016, "step": 6158 }, { "epoch": 15.3975, "grad_norm": 7.0625, "learning_rate": 1.920466181423641e-05, "loss": 1.4737, "step": 6159 }, { "epoch": 15.4, "grad_norm": 7.34375, "learning_rate": 1.9204345339677442e-05, "loss": 1.4553, "step": 6160 }, { "epoch": 15.4025, "grad_norm": 6.4375, "learning_rate": 1.9204028804775152e-05, "loss": 1.5748, "step": 6161 }, { "epoch": 15.405, "grad_norm": 8.0625, "learning_rate": 1.920371220953162e-05, "loss": 1.4267, "step": 6162 }, { "epoch": 15.4075, "grad_norm": 5.71875, "learning_rate": 1.920339555394892e-05, "loss": 1.3898, "step": 6163 }, { "epoch": 15.41, "grad_norm": 8.75, "learning_rate": 1.9203078838029125e-05, "loss": 1.4026, "step": 6164 }, { "epoch": 15.4125, "grad_norm": 7.875, "learning_rate": 1.9202762061774314e-05, "loss": 1.2376, "step": 6165 }, { "epoch": 15.415, "grad_norm": 8.0625, "learning_rate": 1.9202445225186563e-05, "loss": 1.3321, "step": 6166 }, { "epoch": 15.4175, "grad_norm": 6.34375, "learning_rate": 1.920212832826795e-05, "loss": 1.3922, "step": 6167 }, { "epoch": 15.42, "grad_norm": 8.5, "learning_rate": 1.9201811371020553e-05, "loss": 1.3442, "step": 6168 }, { "epoch": 15.4225, "grad_norm": 4.6875, "learning_rate": 1.9201494353446445e-05, "loss": 1.5218, "step": 6169 }, { "epoch": 15.425, "grad_norm": 7.0625, "learning_rate": 1.920117727554771e-05, "loss": 1.3581, "step": 6170 }, { "epoch": 15.4275, "grad_norm": 6.53125, "learning_rate": 1.9200860137326428e-05, "loss": 1.6557, "step": 6171 }, { "epoch": 15.43, "grad_norm": 8.125, "learning_rate": 1.920054293878467e-05, "loss": 1.7016, "step": 6172 }, { "epoch": 15.4325, "grad_norm": 6.125, "learning_rate": 1.9200225679924526e-05, "loss": 1.3177, "step": 6173 }, { "epoch": 15.435, "grad_norm": 8.375, "learning_rate": 1.9199908360748067e-05, "loss": 1.5084, "step": 6174 }, { "epoch": 15.4375, "grad_norm": 5.78125, "learning_rate": 1.919959098125738e-05, "loss": 1.6215, "step": 6175 }, { "epoch": 15.44, "grad_norm": 8.3125, "learning_rate": 1.919927354145454e-05, "loss": 1.5539, "step": 6176 }, { "epoch": 15.4425, "grad_norm": 8.0625, "learning_rate": 1.9198956041341628e-05, "loss": 1.2985, "step": 6177 }, { "epoch": 15.445, "grad_norm": 7.0625, "learning_rate": 1.9198638480920733e-05, "loss": 1.4426, "step": 6178 }, { "epoch": 15.4475, "grad_norm": 8.125, "learning_rate": 1.9198320860193932e-05, "loss": 1.6155, "step": 6179 }, { "epoch": 15.45, "grad_norm": 6.53125, "learning_rate": 1.9198003179163308e-05, "loss": 1.4352, "step": 6180 }, { "epoch": 15.4525, "grad_norm": 7.03125, "learning_rate": 1.919768543783094e-05, "loss": 1.4223, "step": 6181 }, { "epoch": 15.455, "grad_norm": 10.625, "learning_rate": 1.9197367636198915e-05, "loss": 1.5627, "step": 6182 }, { "epoch": 15.4575, "grad_norm": 5.25, "learning_rate": 1.9197049774269317e-05, "loss": 1.4244, "step": 6183 }, { "epoch": 15.46, "grad_norm": 4.9375, "learning_rate": 1.919673185204423e-05, "loss": 1.4888, "step": 6184 }, { "epoch": 15.4625, "grad_norm": 7.9375, "learning_rate": 1.9196413869525735e-05, "loss": 1.5158, "step": 6185 }, { "epoch": 15.465, "grad_norm": 5.40625, "learning_rate": 1.919609582671592e-05, "loss": 1.3531, "step": 6186 }, { "epoch": 15.4675, "grad_norm": 7.40625, "learning_rate": 1.9195777723616866e-05, "loss": 1.3576, "step": 6187 }, { "epoch": 15.47, "grad_norm": 7.84375, "learning_rate": 1.9195459560230665e-05, "loss": 1.5941, "step": 6188 }, { "epoch": 15.4725, "grad_norm": 6.625, "learning_rate": 1.9195141336559398e-05, "loss": 1.3605, "step": 6189 }, { "epoch": 15.475, "grad_norm": 7.71875, "learning_rate": 1.919482305260515e-05, "loss": 1.4545, "step": 6190 }, { "epoch": 15.4775, "grad_norm": 7.5, "learning_rate": 1.9194504708370013e-05, "loss": 1.5061, "step": 6191 }, { "epoch": 15.48, "grad_norm": 7.03125, "learning_rate": 1.919418630385607e-05, "loss": 1.4341, "step": 6192 }, { "epoch": 15.4825, "grad_norm": 5.71875, "learning_rate": 1.919386783906541e-05, "loss": 1.369, "step": 6193 }, { "epoch": 15.485, "grad_norm": 7.71875, "learning_rate": 1.9193549314000117e-05, "loss": 1.4057, "step": 6194 }, { "epoch": 15.4875, "grad_norm": 7.625, "learning_rate": 1.9193230728662285e-05, "loss": 1.4869, "step": 6195 }, { "epoch": 15.49, "grad_norm": 7.46875, "learning_rate": 1.9192912083054003e-05, "loss": 1.5386, "step": 6196 }, { "epoch": 15.4925, "grad_norm": 11.125, "learning_rate": 1.9192593377177355e-05, "loss": 1.4579, "step": 6197 }, { "epoch": 15.495, "grad_norm": 10.4375, "learning_rate": 1.919227461103443e-05, "loss": 1.4573, "step": 6198 }, { "epoch": 15.4975, "grad_norm": 7.125, "learning_rate": 1.9191955784627326e-05, "loss": 1.5262, "step": 6199 }, { "epoch": 15.5, "grad_norm": 5.25, "learning_rate": 1.9191636897958123e-05, "loss": 1.3123, "step": 6200 }, { "epoch": 15.5025, "grad_norm": 5.5, "learning_rate": 1.919131795102892e-05, "loss": 1.2024, "step": 6201 }, { "epoch": 15.505, "grad_norm": 5.9375, "learning_rate": 1.91909989438418e-05, "loss": 1.5237, "step": 6202 }, { "epoch": 15.5075, "grad_norm": 6.8125, "learning_rate": 1.919067987639886e-05, "loss": 1.5028, "step": 6203 }, { "epoch": 15.51, "grad_norm": 7.96875, "learning_rate": 1.9190360748702192e-05, "loss": 1.5562, "step": 6204 }, { "epoch": 15.5125, "grad_norm": 6.59375, "learning_rate": 1.919004156075389e-05, "loss": 1.5046, "step": 6205 }, { "epoch": 15.515, "grad_norm": 6.71875, "learning_rate": 1.918972231255604e-05, "loss": 1.4656, "step": 6206 }, { "epoch": 15.5175, "grad_norm": 5.875, "learning_rate": 1.9189403004110737e-05, "loss": 1.5753, "step": 6207 }, { "epoch": 15.52, "grad_norm": 7.03125, "learning_rate": 1.9189083635420077e-05, "loss": 1.3638, "step": 6208 }, { "epoch": 15.5225, "grad_norm": 7.28125, "learning_rate": 1.918876420648615e-05, "loss": 1.4555, "step": 6209 }, { "epoch": 15.525, "grad_norm": 7.875, "learning_rate": 1.9188444717311055e-05, "loss": 1.4586, "step": 6210 }, { "epoch": 15.5275, "grad_norm": 8.8125, "learning_rate": 1.9188125167896884e-05, "loss": 1.4476, "step": 6211 }, { "epoch": 15.53, "grad_norm": 7.21875, "learning_rate": 1.918780555824573e-05, "loss": 1.4828, "step": 6212 }, { "epoch": 15.5325, "grad_norm": 6.90625, "learning_rate": 1.9187485888359693e-05, "loss": 1.3844, "step": 6213 }, { "epoch": 15.535, "grad_norm": 7.0, "learning_rate": 1.9187166158240868e-05, "loss": 1.39, "step": 6214 }, { "epoch": 15.5375, "grad_norm": 8.25, "learning_rate": 1.9186846367891345e-05, "loss": 1.3869, "step": 6215 }, { "epoch": 15.54, "grad_norm": 9.3125, "learning_rate": 1.9186526517313227e-05, "loss": 1.4131, "step": 6216 }, { "epoch": 15.5425, "grad_norm": 8.125, "learning_rate": 1.9186206606508607e-05, "loss": 1.3499, "step": 6217 }, { "epoch": 15.545, "grad_norm": 13.0625, "learning_rate": 1.9185886635479588e-05, "loss": 1.5713, "step": 6218 }, { "epoch": 15.5475, "grad_norm": 6.28125, "learning_rate": 1.918556660422826e-05, "loss": 1.5131, "step": 6219 }, { "epoch": 15.55, "grad_norm": 7.09375, "learning_rate": 1.9185246512756727e-05, "loss": 1.5827, "step": 6220 }, { "epoch": 15.5525, "grad_norm": 9.1875, "learning_rate": 1.9184926361067083e-05, "loss": 1.4397, "step": 6221 }, { "epoch": 15.555, "grad_norm": 7.21875, "learning_rate": 1.918460614916143e-05, "loss": 1.4373, "step": 6222 }, { "epoch": 15.5575, "grad_norm": 5.90625, "learning_rate": 1.918428587704187e-05, "loss": 1.4038, "step": 6223 }, { "epoch": 15.56, "grad_norm": 7.65625, "learning_rate": 1.9183965544710495e-05, "loss": 1.5932, "step": 6224 }, { "epoch": 15.5625, "grad_norm": 5.6875, "learning_rate": 1.918364515216941e-05, "loss": 1.5401, "step": 6225 }, { "epoch": 15.565, "grad_norm": 7.71875, "learning_rate": 1.918332469942072e-05, "loss": 1.3605, "step": 6226 }, { "epoch": 15.5675, "grad_norm": 12.875, "learning_rate": 1.9183004186466517e-05, "loss": 1.6858, "step": 6227 }, { "epoch": 15.57, "grad_norm": 6.3125, "learning_rate": 1.9182683613308905e-05, "loss": 1.4815, "step": 6228 }, { "epoch": 15.5725, "grad_norm": 6.59375, "learning_rate": 1.9182362979949988e-05, "loss": 1.4236, "step": 6229 }, { "epoch": 15.575, "grad_norm": 7.125, "learning_rate": 1.9182042286391867e-05, "loss": 1.5501, "step": 6230 }, { "epoch": 15.5775, "grad_norm": 7.53125, "learning_rate": 1.9181721532636642e-05, "loss": 1.422, "step": 6231 }, { "epoch": 15.58, "grad_norm": 6.9375, "learning_rate": 1.918140071868642e-05, "loss": 1.4781, "step": 6232 }, { "epoch": 15.5825, "grad_norm": 7.875, "learning_rate": 1.91810798445433e-05, "loss": 1.4402, "step": 6233 }, { "epoch": 15.585, "grad_norm": 6.46875, "learning_rate": 1.9180758910209394e-05, "loss": 1.5411, "step": 6234 }, { "epoch": 15.5875, "grad_norm": 7.09375, "learning_rate": 1.9180437915686795e-05, "loss": 1.5394, "step": 6235 }, { "epoch": 15.59, "grad_norm": 8.3125, "learning_rate": 1.9180116860977613e-05, "loss": 1.4839, "step": 6236 }, { "epoch": 15.5925, "grad_norm": 6.625, "learning_rate": 1.9179795746083952e-05, "loss": 1.4796, "step": 6237 }, { "epoch": 15.595, "grad_norm": 4.6875, "learning_rate": 1.917947457100792e-05, "loss": 1.3395, "step": 6238 }, { "epoch": 15.5975, "grad_norm": 7.5, "learning_rate": 1.9179153335751618e-05, "loss": 1.5637, "step": 6239 }, { "epoch": 15.6, "grad_norm": 7.09375, "learning_rate": 1.9178832040317153e-05, "loss": 1.3814, "step": 6240 }, { "epoch": 15.6025, "grad_norm": 8.5625, "learning_rate": 1.9178510684706635e-05, "loss": 1.6348, "step": 6241 }, { "epoch": 15.605, "grad_norm": 7.125, "learning_rate": 1.9178189268922165e-05, "loss": 1.4295, "step": 6242 }, { "epoch": 15.6075, "grad_norm": 6.59375, "learning_rate": 1.9177867792965856e-05, "loss": 1.4576, "step": 6243 }, { "epoch": 15.61, "grad_norm": 6.0, "learning_rate": 1.9177546256839814e-05, "loss": 1.5596, "step": 6244 }, { "epoch": 15.6125, "grad_norm": 7.34375, "learning_rate": 1.9177224660546144e-05, "loss": 1.525, "step": 6245 }, { "epoch": 15.615, "grad_norm": 8.0, "learning_rate": 1.917690300408696e-05, "loss": 1.4176, "step": 6246 }, { "epoch": 15.6175, "grad_norm": 6.96875, "learning_rate": 1.917658128746436e-05, "loss": 1.447, "step": 6247 }, { "epoch": 15.62, "grad_norm": 7.1875, "learning_rate": 1.9176259510680463e-05, "loss": 1.4352, "step": 6248 }, { "epoch": 15.6225, "grad_norm": 5.0, "learning_rate": 1.9175937673737377e-05, "loss": 1.3618, "step": 6249 }, { "epoch": 15.625, "grad_norm": 7.6875, "learning_rate": 1.9175615776637212e-05, "loss": 1.4003, "step": 6250 }, { "epoch": 15.6275, "grad_norm": 5.59375, "learning_rate": 1.917529381938208e-05, "loss": 1.4275, "step": 6251 }, { "epoch": 15.63, "grad_norm": 7.46875, "learning_rate": 1.9174971801974084e-05, "loss": 1.5458, "step": 6252 }, { "epoch": 15.6325, "grad_norm": 9.0625, "learning_rate": 1.9174649724415338e-05, "loss": 1.5064, "step": 6253 }, { "epoch": 15.635, "grad_norm": 6.40625, "learning_rate": 1.917432758670796e-05, "loss": 1.3091, "step": 6254 }, { "epoch": 15.6375, "grad_norm": 7.34375, "learning_rate": 1.9174005388854056e-05, "loss": 1.2337, "step": 6255 }, { "epoch": 15.64, "grad_norm": 11.9375, "learning_rate": 1.9173683130855737e-05, "loss": 1.391, "step": 6256 }, { "epoch": 15.6425, "grad_norm": 6.625, "learning_rate": 1.9173360812715122e-05, "loss": 1.358, "step": 6257 }, { "epoch": 15.645, "grad_norm": 7.875, "learning_rate": 1.917303843443432e-05, "loss": 1.6038, "step": 6258 }, { "epoch": 15.6475, "grad_norm": 6.40625, "learning_rate": 1.9172715996015446e-05, "loss": 1.4377, "step": 6259 }, { "epoch": 15.65, "grad_norm": 7.0625, "learning_rate": 1.917239349746061e-05, "loss": 1.4022, "step": 6260 }, { "epoch": 15.6525, "grad_norm": 5.0625, "learning_rate": 1.9172070938771935e-05, "loss": 1.418, "step": 6261 }, { "epoch": 15.655, "grad_norm": 11.0, "learning_rate": 1.9171748319951523e-05, "loss": 1.662, "step": 6262 }, { "epoch": 15.6575, "grad_norm": 6.78125, "learning_rate": 1.91714256410015e-05, "loss": 1.5453, "step": 6263 }, { "epoch": 15.66, "grad_norm": 6.96875, "learning_rate": 1.9171102901923978e-05, "loss": 1.5148, "step": 6264 }, { "epoch": 15.6625, "grad_norm": 7.96875, "learning_rate": 1.917078010272107e-05, "loss": 1.4975, "step": 6265 }, { "epoch": 15.665, "grad_norm": 5.0625, "learning_rate": 1.9170457243394898e-05, "loss": 1.4363, "step": 6266 }, { "epoch": 15.6675, "grad_norm": 6.78125, "learning_rate": 1.9170134323947572e-05, "loss": 1.5514, "step": 6267 }, { "epoch": 15.67, "grad_norm": 6.875, "learning_rate": 1.9169811344381215e-05, "loss": 1.4826, "step": 6268 }, { "epoch": 15.6725, "grad_norm": 8.375, "learning_rate": 1.9169488304697938e-05, "loss": 1.5855, "step": 6269 }, { "epoch": 15.675, "grad_norm": 8.625, "learning_rate": 1.9169165204899866e-05, "loss": 1.4916, "step": 6270 }, { "epoch": 15.6775, "grad_norm": 6.9375, "learning_rate": 1.9168842044989115e-05, "loss": 1.4835, "step": 6271 }, { "epoch": 15.68, "grad_norm": 6.59375, "learning_rate": 1.9168518824967797e-05, "loss": 1.5838, "step": 6272 }, { "epoch": 15.6825, "grad_norm": 4.65625, "learning_rate": 1.9168195544838043e-05, "loss": 1.332, "step": 6273 }, { "epoch": 15.685, "grad_norm": 5.5625, "learning_rate": 1.916787220460196e-05, "loss": 1.4975, "step": 6274 }, { "epoch": 15.6875, "grad_norm": 8.9375, "learning_rate": 1.9167548804261677e-05, "loss": 1.6198, "step": 6275 }, { "epoch": 15.69, "grad_norm": 6.59375, "learning_rate": 1.916722534381931e-05, "loss": 1.4171, "step": 6276 }, { "epoch": 15.6925, "grad_norm": 10.0, "learning_rate": 1.916690182327698e-05, "loss": 1.4696, "step": 6277 }, { "epoch": 15.695, "grad_norm": 6.4375, "learning_rate": 1.916657824263681e-05, "loss": 1.5326, "step": 6278 }, { "epoch": 15.6975, "grad_norm": 7.46875, "learning_rate": 1.916625460190092e-05, "loss": 1.6316, "step": 6279 }, { "epoch": 15.7, "grad_norm": 5.84375, "learning_rate": 1.916593090107143e-05, "loss": 1.5583, "step": 6280 }, { "epoch": 15.7025, "grad_norm": 6.15625, "learning_rate": 1.9165607140150465e-05, "loss": 1.3293, "step": 6281 }, { "epoch": 15.705, "grad_norm": 7.3125, "learning_rate": 1.9165283319140145e-05, "loss": 1.501, "step": 6282 }, { "epoch": 15.7075, "grad_norm": 8.8125, "learning_rate": 1.9164959438042596e-05, "loss": 1.3629, "step": 6283 }, { "epoch": 15.71, "grad_norm": 7.15625, "learning_rate": 1.9164635496859937e-05, "loss": 1.4441, "step": 6284 }, { "epoch": 15.7125, "grad_norm": 5.65625, "learning_rate": 1.9164311495594297e-05, "loss": 1.4935, "step": 6285 }, { "epoch": 15.715, "grad_norm": 5.4375, "learning_rate": 1.9163987434247798e-05, "loss": 1.4128, "step": 6286 }, { "epoch": 15.7175, "grad_norm": 11.875, "learning_rate": 1.9163663312822562e-05, "loss": 1.313, "step": 6287 }, { "epoch": 15.72, "grad_norm": 6.46875, "learning_rate": 1.916333913132072e-05, "loss": 1.3457, "step": 6288 }, { "epoch": 15.7225, "grad_norm": 12.0, "learning_rate": 1.9163014889744388e-05, "loss": 1.5597, "step": 6289 }, { "epoch": 15.725, "grad_norm": 7.6875, "learning_rate": 1.91626905880957e-05, "loss": 1.46, "step": 6290 }, { "epoch": 15.7275, "grad_norm": 7.65625, "learning_rate": 1.9162366226376777e-05, "loss": 1.4632, "step": 6291 }, { "epoch": 15.73, "grad_norm": 5.0625, "learning_rate": 1.9162041804589752e-05, "loss": 1.4818, "step": 6292 }, { "epoch": 15.7325, "grad_norm": 8.25, "learning_rate": 1.916171732273674e-05, "loss": 1.4134, "step": 6293 }, { "epoch": 15.735, "grad_norm": 5.8125, "learning_rate": 1.9161392780819884e-05, "loss": 1.4377, "step": 6294 }, { "epoch": 15.7375, "grad_norm": 6.96875, "learning_rate": 1.91610681788413e-05, "loss": 1.3702, "step": 6295 }, { "epoch": 15.74, "grad_norm": 6.5625, "learning_rate": 1.916074351680312e-05, "loss": 1.4189, "step": 6296 }, { "epoch": 15.7425, "grad_norm": 5.78125, "learning_rate": 1.9160418794707473e-05, "loss": 1.4955, "step": 6297 }, { "epoch": 15.745, "grad_norm": 8.1875, "learning_rate": 1.9160094012556484e-05, "loss": 1.3685, "step": 6298 }, { "epoch": 15.7475, "grad_norm": 6.5625, "learning_rate": 1.915976917035229e-05, "loss": 1.3599, "step": 6299 }, { "epoch": 15.75, "grad_norm": 5.78125, "learning_rate": 1.9159444268097012e-05, "loss": 1.558, "step": 6300 }, { "epoch": 15.7525, "grad_norm": 6.28125, "learning_rate": 1.9159119305792785e-05, "loss": 1.4466, "step": 6301 }, { "epoch": 15.755, "grad_norm": 5.5625, "learning_rate": 1.9158794283441738e-05, "loss": 1.2665, "step": 6302 }, { "epoch": 15.7575, "grad_norm": 7.84375, "learning_rate": 1.9158469201046002e-05, "loss": 1.3648, "step": 6303 }, { "epoch": 15.76, "grad_norm": 6.5, "learning_rate": 1.915814405860771e-05, "loss": 1.2865, "step": 6304 }, { "epoch": 15.7625, "grad_norm": 6.28125, "learning_rate": 1.915781885612899e-05, "loss": 1.5068, "step": 6305 }, { "epoch": 15.765, "grad_norm": 7.0, "learning_rate": 1.9157493593611977e-05, "loss": 1.397, "step": 6306 }, { "epoch": 15.7675, "grad_norm": 5.5625, "learning_rate": 1.9157168271058803e-05, "loss": 1.4846, "step": 6307 }, { "epoch": 15.77, "grad_norm": 6.5625, "learning_rate": 1.91568428884716e-05, "loss": 1.2562, "step": 6308 }, { "epoch": 15.7725, "grad_norm": 6.90625, "learning_rate": 1.91565174458525e-05, "loss": 1.3381, "step": 6309 }, { "epoch": 15.775, "grad_norm": 7.75, "learning_rate": 1.915619194320364e-05, "loss": 1.5769, "step": 6310 }, { "epoch": 15.7775, "grad_norm": 5.78125, "learning_rate": 1.915586638052715e-05, "loss": 1.5144, "step": 6311 }, { "epoch": 15.78, "grad_norm": 5.5, "learning_rate": 1.9155540757825168e-05, "loss": 1.5097, "step": 6312 }, { "epoch": 15.7825, "grad_norm": 6.6875, "learning_rate": 1.9155215075099825e-05, "loss": 1.5633, "step": 6313 }, { "epoch": 15.785, "grad_norm": 8.4375, "learning_rate": 1.9154889332353264e-05, "loss": 1.5882, "step": 6314 }, { "epoch": 15.7875, "grad_norm": 9.875, "learning_rate": 1.9154563529587608e-05, "loss": 1.4716, "step": 6315 }, { "epoch": 15.79, "grad_norm": 7.03125, "learning_rate": 1.9154237666805005e-05, "loss": 1.5082, "step": 6316 }, { "epoch": 15.7925, "grad_norm": 8.5, "learning_rate": 1.9153911744007584e-05, "loss": 1.4351, "step": 6317 }, { "epoch": 15.795, "grad_norm": 5.375, "learning_rate": 1.9153585761197483e-05, "loss": 1.4759, "step": 6318 }, { "epoch": 15.7975, "grad_norm": 6.4375, "learning_rate": 1.915325971837684e-05, "loss": 1.4533, "step": 6319 }, { "epoch": 15.8, "grad_norm": 11.625, "learning_rate": 1.91529336155478e-05, "loss": 1.5322, "step": 6320 }, { "epoch": 15.8025, "grad_norm": 6.65625, "learning_rate": 1.9152607452712482e-05, "loss": 1.3842, "step": 6321 }, { "epoch": 15.805, "grad_norm": 6.40625, "learning_rate": 1.9152281229873044e-05, "loss": 1.5567, "step": 6322 }, { "epoch": 15.8075, "grad_norm": 7.78125, "learning_rate": 1.9151954947031614e-05, "loss": 1.4711, "step": 6323 }, { "epoch": 15.81, "grad_norm": 7.84375, "learning_rate": 1.915162860419033e-05, "loss": 1.4378, "step": 6324 }, { "epoch": 15.8125, "grad_norm": 5.21875, "learning_rate": 1.915130220135134e-05, "loss": 1.4503, "step": 6325 }, { "epoch": 15.815, "grad_norm": 6.875, "learning_rate": 1.9150975738516776e-05, "loss": 1.3335, "step": 6326 }, { "epoch": 15.8175, "grad_norm": 6.09375, "learning_rate": 1.9150649215688785e-05, "loss": 1.5902, "step": 6327 }, { "epoch": 15.82, "grad_norm": 6.875, "learning_rate": 1.9150322632869498e-05, "loss": 1.4499, "step": 6328 }, { "epoch": 15.8225, "grad_norm": 8.875, "learning_rate": 1.9149995990061065e-05, "loss": 1.4015, "step": 6329 }, { "epoch": 15.825, "grad_norm": 5.59375, "learning_rate": 1.9149669287265626e-05, "loss": 1.3185, "step": 6330 }, { "epoch": 15.8275, "grad_norm": 6.625, "learning_rate": 1.9149342524485323e-05, "loss": 1.4708, "step": 6331 }, { "epoch": 15.83, "grad_norm": 6.75, "learning_rate": 1.914901570172229e-05, "loss": 1.478, "step": 6332 }, { "epoch": 15.8325, "grad_norm": 11.625, "learning_rate": 1.914868881897868e-05, "loss": 1.5662, "step": 6333 }, { "epoch": 15.835, "grad_norm": 6.0, "learning_rate": 1.9148361876256632e-05, "loss": 1.5593, "step": 6334 }, { "epoch": 15.8375, "grad_norm": 5.5, "learning_rate": 1.9148034873558292e-05, "loss": 1.3431, "step": 6335 }, { "epoch": 15.84, "grad_norm": 5.9375, "learning_rate": 1.9147707810885798e-05, "loss": 1.4734, "step": 6336 }, { "epoch": 15.8425, "grad_norm": 7.21875, "learning_rate": 1.9147380688241298e-05, "loss": 1.4006, "step": 6337 }, { "epoch": 15.845, "grad_norm": 7.0, "learning_rate": 1.914705350562694e-05, "loss": 1.4527, "step": 6338 }, { "epoch": 15.8475, "grad_norm": 5.0, "learning_rate": 1.914672626304486e-05, "loss": 1.368, "step": 6339 }, { "epoch": 15.85, "grad_norm": 5.78125, "learning_rate": 1.9146398960497213e-05, "loss": 1.5481, "step": 6340 }, { "epoch": 15.8525, "grad_norm": 7.15625, "learning_rate": 1.914607159798614e-05, "loss": 1.4874, "step": 6341 }, { "epoch": 15.855, "grad_norm": 7.28125, "learning_rate": 1.9145744175513784e-05, "loss": 1.4222, "step": 6342 }, { "epoch": 15.8575, "grad_norm": 4.6875, "learning_rate": 1.91454166930823e-05, "loss": 1.4234, "step": 6343 }, { "epoch": 15.86, "grad_norm": 6.9375, "learning_rate": 1.9145089150693823e-05, "loss": 1.5045, "step": 6344 }, { "epoch": 15.8625, "grad_norm": 6.0625, "learning_rate": 1.9144761548350512e-05, "loss": 1.446, "step": 6345 }, { "epoch": 15.865, "grad_norm": 6.46875, "learning_rate": 1.914443388605451e-05, "loss": 1.3647, "step": 6346 }, { "epoch": 15.8675, "grad_norm": 9.1875, "learning_rate": 1.9144106163807964e-05, "loss": 1.5332, "step": 6347 }, { "epoch": 15.87, "grad_norm": 6.09375, "learning_rate": 1.9143778381613026e-05, "loss": 1.3347, "step": 6348 }, { "epoch": 15.8725, "grad_norm": 5.84375, "learning_rate": 1.9143450539471842e-05, "loss": 1.4514, "step": 6349 }, { "epoch": 15.875, "grad_norm": 6.5, "learning_rate": 1.9143122637386567e-05, "loss": 1.5193, "step": 6350 }, { "epoch": 15.8775, "grad_norm": 7.34375, "learning_rate": 1.9142794675359338e-05, "loss": 1.6878, "step": 6351 }, { "epoch": 15.88, "grad_norm": 4.375, "learning_rate": 1.9142466653392317e-05, "loss": 1.4105, "step": 6352 }, { "epoch": 15.8825, "grad_norm": 7.71875, "learning_rate": 1.9142138571487652e-05, "loss": 1.6776, "step": 6353 }, { "epoch": 15.885, "grad_norm": 4.90625, "learning_rate": 1.914181042964749e-05, "loss": 1.6314, "step": 6354 }, { "epoch": 15.8875, "grad_norm": 6.71875, "learning_rate": 1.9141482227873984e-05, "loss": 1.407, "step": 6355 }, { "epoch": 15.89, "grad_norm": 5.78125, "learning_rate": 1.9141153966169287e-05, "loss": 1.5261, "step": 6356 }, { "epoch": 15.8925, "grad_norm": 8.0625, "learning_rate": 1.914082564453555e-05, "loss": 1.3914, "step": 6357 }, { "epoch": 15.895, "grad_norm": 6.6875, "learning_rate": 1.9140497262974927e-05, "loss": 1.3615, "step": 6358 }, { "epoch": 15.8975, "grad_norm": 8.125, "learning_rate": 1.914016882148957e-05, "loss": 1.3578, "step": 6359 }, { "epoch": 15.9, "grad_norm": 7.3125, "learning_rate": 1.913984032008163e-05, "loss": 1.4618, "step": 6360 }, { "epoch": 15.9025, "grad_norm": 5.09375, "learning_rate": 1.9139511758753265e-05, "loss": 1.5216, "step": 6361 }, { "epoch": 15.905, "grad_norm": 6.25, "learning_rate": 1.9139183137506622e-05, "loss": 1.503, "step": 6362 }, { "epoch": 15.9075, "grad_norm": 9.125, "learning_rate": 1.9138854456343864e-05, "loss": 1.5475, "step": 6363 }, { "epoch": 15.91, "grad_norm": 5.46875, "learning_rate": 1.913852571526714e-05, "loss": 1.3636, "step": 6364 }, { "epoch": 15.9125, "grad_norm": 6.78125, "learning_rate": 1.913819691427861e-05, "loss": 1.4725, "step": 6365 }, { "epoch": 15.915, "grad_norm": 6.1875, "learning_rate": 1.9137868053380427e-05, "loss": 1.499, "step": 6366 }, { "epoch": 15.9175, "grad_norm": 8.6875, "learning_rate": 1.9137539132574745e-05, "loss": 1.5623, "step": 6367 }, { "epoch": 15.92, "grad_norm": 6.34375, "learning_rate": 1.913721015186372e-05, "loss": 1.4312, "step": 6368 }, { "epoch": 15.9225, "grad_norm": 7.09375, "learning_rate": 1.9136881111249515e-05, "loss": 1.3559, "step": 6369 }, { "epoch": 15.925, "grad_norm": 8.5625, "learning_rate": 1.913655201073428e-05, "loss": 1.4308, "step": 6370 }, { "epoch": 15.9275, "grad_norm": 7.1875, "learning_rate": 1.9136222850320175e-05, "loss": 1.4615, "step": 6371 }, { "epoch": 15.93, "grad_norm": 7.3125, "learning_rate": 1.9135893630009362e-05, "loss": 1.5728, "step": 6372 }, { "epoch": 15.9325, "grad_norm": 7.4375, "learning_rate": 1.9135564349803996e-05, "loss": 1.402, "step": 6373 }, { "epoch": 15.935, "grad_norm": 9.9375, "learning_rate": 1.9135235009706234e-05, "loss": 1.4784, "step": 6374 }, { "epoch": 15.9375, "grad_norm": 8.3125, "learning_rate": 1.9134905609718235e-05, "loss": 1.6742, "step": 6375 }, { "epoch": 15.94, "grad_norm": 8.3125, "learning_rate": 1.9134576149842164e-05, "loss": 1.6725, "step": 6376 }, { "epoch": 15.9425, "grad_norm": 5.5625, "learning_rate": 1.9134246630080172e-05, "loss": 1.4148, "step": 6377 }, { "epoch": 15.945, "grad_norm": 6.84375, "learning_rate": 1.913391705043443e-05, "loss": 1.3791, "step": 6378 }, { "epoch": 15.9475, "grad_norm": 6.21875, "learning_rate": 1.913358741090709e-05, "loss": 1.3281, "step": 6379 }, { "epoch": 15.95, "grad_norm": 6.96875, "learning_rate": 1.9133257711500318e-05, "loss": 1.4457, "step": 6380 }, { "epoch": 15.9525, "grad_norm": 6.53125, "learning_rate": 1.9132927952216277e-05, "loss": 1.4683, "step": 6381 }, { "epoch": 15.955, "grad_norm": 4.71875, "learning_rate": 1.9132598133057117e-05, "loss": 1.4063, "step": 6382 }, { "epoch": 15.9575, "grad_norm": 5.78125, "learning_rate": 1.913226825402502e-05, "loss": 1.2423, "step": 6383 }, { "epoch": 15.96, "grad_norm": 6.46875, "learning_rate": 1.913193831512213e-05, "loss": 1.5237, "step": 6384 }, { "epoch": 15.9625, "grad_norm": 7.6875, "learning_rate": 1.913160831635062e-05, "loss": 1.3484, "step": 6385 }, { "epoch": 15.965, "grad_norm": 6.40625, "learning_rate": 1.913127825771265e-05, "loss": 1.3759, "step": 6386 }, { "epoch": 15.9675, "grad_norm": 7.78125, "learning_rate": 1.9130948139210388e-05, "loss": 1.4733, "step": 6387 }, { "epoch": 15.97, "grad_norm": 5.375, "learning_rate": 1.9130617960845996e-05, "loss": 1.5325, "step": 6388 }, { "epoch": 15.9725, "grad_norm": 7.53125, "learning_rate": 1.9130287722621636e-05, "loss": 1.4425, "step": 6389 }, { "epoch": 15.975, "grad_norm": 8.875, "learning_rate": 1.9129957424539473e-05, "loss": 1.6691, "step": 6390 }, { "epoch": 15.9775, "grad_norm": 5.4375, "learning_rate": 1.9129627066601677e-05, "loss": 1.2822, "step": 6391 }, { "epoch": 15.98, "grad_norm": 7.4375, "learning_rate": 1.912929664881041e-05, "loss": 1.6341, "step": 6392 }, { "epoch": 15.9825, "grad_norm": 6.28125, "learning_rate": 1.9128966171167845e-05, "loss": 1.3262, "step": 6393 }, { "epoch": 15.985, "grad_norm": 5.96875, "learning_rate": 1.9128635633676137e-05, "loss": 1.3064, "step": 6394 }, { "epoch": 15.9875, "grad_norm": 6.53125, "learning_rate": 1.912830503633746e-05, "loss": 1.3342, "step": 6395 }, { "epoch": 15.99, "grad_norm": 4.71875, "learning_rate": 1.9127974379153983e-05, "loss": 1.4747, "step": 6396 }, { "epoch": 15.9925, "grad_norm": 7.84375, "learning_rate": 1.9127643662127868e-05, "loss": 1.4501, "step": 6397 }, { "epoch": 15.995, "grad_norm": 5.0, "learning_rate": 1.9127312885261287e-05, "loss": 1.4622, "step": 6398 }, { "epoch": 15.9975, "grad_norm": 5.4375, "learning_rate": 1.912698204855641e-05, "loss": 1.3448, "step": 6399 }, { "epoch": 16.0, "grad_norm": 6.1875, "learning_rate": 1.9126651152015404e-05, "loss": 1.5853, "step": 6400 }, { "epoch": 16.0025, "grad_norm": 6.15625, "learning_rate": 1.9126320195640438e-05, "loss": 1.3116, "step": 6401 }, { "epoch": 16.005, "grad_norm": 6.0, "learning_rate": 1.912598917943368e-05, "loss": 1.444, "step": 6402 }, { "epoch": 16.0075, "grad_norm": 7.9375, "learning_rate": 1.9125658103397302e-05, "loss": 1.5169, "step": 6403 }, { "epoch": 16.01, "grad_norm": 8.25, "learning_rate": 1.9125326967533476e-05, "loss": 1.519, "step": 6404 }, { "epoch": 16.0125, "grad_norm": 5.59375, "learning_rate": 1.9124995771844374e-05, "loss": 1.4676, "step": 6405 }, { "epoch": 16.015, "grad_norm": 5.25, "learning_rate": 1.912466451633216e-05, "loss": 1.3908, "step": 6406 }, { "epoch": 16.0175, "grad_norm": 10.5, "learning_rate": 1.9124333200999014e-05, "loss": 1.4723, "step": 6407 }, { "epoch": 16.02, "grad_norm": 6.09375, "learning_rate": 1.9124001825847103e-05, "loss": 1.3562, "step": 6408 }, { "epoch": 16.0225, "grad_norm": 6.0, "learning_rate": 1.9123670390878603e-05, "loss": 1.3142, "step": 6409 }, { "epoch": 16.025, "grad_norm": 8.125, "learning_rate": 1.912333889609568e-05, "loss": 1.3688, "step": 6410 }, { "epoch": 16.0275, "grad_norm": 8.9375, "learning_rate": 1.912300734150052e-05, "loss": 1.5095, "step": 6411 }, { "epoch": 16.03, "grad_norm": 6.4375, "learning_rate": 1.9122675727095285e-05, "loss": 1.4016, "step": 6412 }, { "epoch": 16.0325, "grad_norm": 8.375, "learning_rate": 1.9122344052882153e-05, "loss": 1.4524, "step": 6413 }, { "epoch": 16.035, "grad_norm": 6.96875, "learning_rate": 1.91220123188633e-05, "loss": 1.3449, "step": 6414 }, { "epoch": 16.0375, "grad_norm": 6.75, "learning_rate": 1.9121680525040896e-05, "loss": 1.3224, "step": 6415 }, { "epoch": 16.04, "grad_norm": 7.09375, "learning_rate": 1.912134867141712e-05, "loss": 1.5515, "step": 6416 }, { "epoch": 16.0425, "grad_norm": 6.9375, "learning_rate": 1.9121016757994147e-05, "loss": 1.6283, "step": 6417 }, { "epoch": 16.045, "grad_norm": 6.5, "learning_rate": 1.9120684784774156e-05, "loss": 1.4443, "step": 6418 }, { "epoch": 16.0475, "grad_norm": 9.3125, "learning_rate": 1.9120352751759318e-05, "loss": 1.4095, "step": 6419 }, { "epoch": 16.05, "grad_norm": 7.96875, "learning_rate": 1.9120020658951814e-05, "loss": 1.6157, "step": 6420 }, { "epoch": 16.0525, "grad_norm": 5.9375, "learning_rate": 1.911968850635382e-05, "loss": 1.3796, "step": 6421 }, { "epoch": 16.055, "grad_norm": 6.5, "learning_rate": 1.911935629396751e-05, "loss": 1.5455, "step": 6422 }, { "epoch": 16.0575, "grad_norm": 6.6875, "learning_rate": 1.9119024021795066e-05, "loss": 1.4471, "step": 6423 }, { "epoch": 16.06, "grad_norm": 5.59375, "learning_rate": 1.9118691689838668e-05, "loss": 1.3976, "step": 6424 }, { "epoch": 16.0625, "grad_norm": 6.15625, "learning_rate": 1.9118359298100494e-05, "loss": 1.4282, "step": 6425 }, { "epoch": 16.065, "grad_norm": 10.125, "learning_rate": 1.911802684658272e-05, "loss": 1.5221, "step": 6426 }, { "epoch": 16.0675, "grad_norm": 10.875, "learning_rate": 1.911769433528752e-05, "loss": 1.3471, "step": 6427 }, { "epoch": 16.07, "grad_norm": 6.78125, "learning_rate": 1.9117361764217087e-05, "loss": 1.5411, "step": 6428 }, { "epoch": 16.0725, "grad_norm": 5.90625, "learning_rate": 1.9117029133373594e-05, "loss": 1.3939, "step": 6429 }, { "epoch": 16.075, "grad_norm": 5.8125, "learning_rate": 1.9116696442759224e-05, "loss": 1.2993, "step": 6430 }, { "epoch": 16.0775, "grad_norm": 7.34375, "learning_rate": 1.9116363692376153e-05, "loss": 1.4837, "step": 6431 }, { "epoch": 16.08, "grad_norm": 4.75, "learning_rate": 1.911603088222657e-05, "loss": 1.33, "step": 6432 }, { "epoch": 16.0825, "grad_norm": 5.65625, "learning_rate": 1.9115698012312652e-05, "loss": 1.3738, "step": 6433 }, { "epoch": 16.085, "grad_norm": 7.1875, "learning_rate": 1.9115365082636583e-05, "loss": 1.3492, "step": 6434 }, { "epoch": 16.0875, "grad_norm": 5.71875, "learning_rate": 1.9115032093200546e-05, "loss": 1.446, "step": 6435 }, { "epoch": 16.09, "grad_norm": 13.5625, "learning_rate": 1.9114699044006725e-05, "loss": 1.5244, "step": 6436 }, { "epoch": 16.0925, "grad_norm": 6.875, "learning_rate": 1.9114365935057298e-05, "loss": 1.4069, "step": 6437 }, { "epoch": 16.095, "grad_norm": 12.5, "learning_rate": 1.9114032766354453e-05, "loss": 1.4738, "step": 6438 }, { "epoch": 16.0975, "grad_norm": 6.875, "learning_rate": 1.9113699537900376e-05, "loss": 1.4865, "step": 6439 }, { "epoch": 16.1, "grad_norm": 5.34375, "learning_rate": 1.911336624969725e-05, "loss": 1.5166, "step": 6440 }, { "epoch": 16.1025, "grad_norm": 8.875, "learning_rate": 1.9113032901747256e-05, "loss": 1.5168, "step": 6441 }, { "epoch": 16.105, "grad_norm": 7.1875, "learning_rate": 1.9112699494052588e-05, "loss": 1.2836, "step": 6442 }, { "epoch": 16.1075, "grad_norm": 7.65625, "learning_rate": 1.9112366026615422e-05, "loss": 1.4815, "step": 6443 }, { "epoch": 16.11, "grad_norm": 4.78125, "learning_rate": 1.911203249943795e-05, "loss": 1.343, "step": 6444 }, { "epoch": 16.1125, "grad_norm": 6.65625, "learning_rate": 1.911169891252236e-05, "loss": 1.4477, "step": 6445 }, { "epoch": 16.115, "grad_norm": 8.875, "learning_rate": 1.9111365265870834e-05, "loss": 1.49, "step": 6446 }, { "epoch": 16.1175, "grad_norm": 8.0, "learning_rate": 1.9111031559485565e-05, "loss": 1.3621, "step": 6447 }, { "epoch": 16.12, "grad_norm": 6.40625, "learning_rate": 1.9110697793368733e-05, "loss": 1.44, "step": 6448 }, { "epoch": 16.1225, "grad_norm": 5.34375, "learning_rate": 1.9110363967522535e-05, "loss": 1.3222, "step": 6449 }, { "epoch": 16.125, "grad_norm": 5.25, "learning_rate": 1.9110030081949157e-05, "loss": 1.4512, "step": 6450 }, { "epoch": 16.1275, "grad_norm": 8.1875, "learning_rate": 1.9109696136650782e-05, "loss": 1.3656, "step": 6451 }, { "epoch": 16.13, "grad_norm": 13.0, "learning_rate": 1.9109362131629604e-05, "loss": 1.3637, "step": 6452 }, { "epoch": 16.1325, "grad_norm": 7.6875, "learning_rate": 1.9109028066887813e-05, "loss": 1.3489, "step": 6453 }, { "epoch": 16.135, "grad_norm": 7.4375, "learning_rate": 1.91086939424276e-05, "loss": 1.5567, "step": 6454 }, { "epoch": 16.1375, "grad_norm": 7.40625, "learning_rate": 1.910835975825115e-05, "loss": 1.4154, "step": 6455 }, { "epoch": 16.14, "grad_norm": 7.65625, "learning_rate": 1.910802551436066e-05, "loss": 1.4081, "step": 6456 }, { "epoch": 16.1425, "grad_norm": 7.6875, "learning_rate": 1.9107691210758323e-05, "loss": 1.3125, "step": 6457 }, { "epoch": 16.145, "grad_norm": 6.0625, "learning_rate": 1.9107356847446325e-05, "loss": 1.5832, "step": 6458 }, { "epoch": 16.1475, "grad_norm": 5.125, "learning_rate": 1.9107022424426857e-05, "loss": 1.2758, "step": 6459 }, { "epoch": 16.15, "grad_norm": 7.25, "learning_rate": 1.910668794170212e-05, "loss": 1.5978, "step": 6460 }, { "epoch": 16.1525, "grad_norm": 7.5, "learning_rate": 1.9106353399274294e-05, "loss": 1.4725, "step": 6461 }, { "epoch": 16.155, "grad_norm": 6.625, "learning_rate": 1.9106018797145585e-05, "loss": 1.4635, "step": 6462 }, { "epoch": 16.1575, "grad_norm": 7.875, "learning_rate": 1.9105684135318177e-05, "loss": 1.4521, "step": 6463 }, { "epoch": 16.16, "grad_norm": 6.25, "learning_rate": 1.9105349413794272e-05, "loss": 1.4021, "step": 6464 }, { "epoch": 16.1625, "grad_norm": 6.625, "learning_rate": 1.910501463257606e-05, "loss": 1.5134, "step": 6465 }, { "epoch": 16.165, "grad_norm": 8.9375, "learning_rate": 1.9104679791665733e-05, "loss": 1.4421, "step": 6466 }, { "epoch": 16.1675, "grad_norm": 6.6875, "learning_rate": 1.9104344891065497e-05, "loss": 1.3534, "step": 6467 }, { "epoch": 16.17, "grad_norm": 6.34375, "learning_rate": 1.9104009930777533e-05, "loss": 1.3616, "step": 6468 }, { "epoch": 16.1725, "grad_norm": 5.84375, "learning_rate": 1.910367491080405e-05, "loss": 1.4021, "step": 6469 }, { "epoch": 16.175, "grad_norm": 4.3125, "learning_rate": 1.910333983114723e-05, "loss": 1.3452, "step": 6470 }, { "epoch": 16.1775, "grad_norm": 6.15625, "learning_rate": 1.9103004691809285e-05, "loss": 1.3777, "step": 6471 }, { "epoch": 16.18, "grad_norm": 7.25, "learning_rate": 1.9102669492792406e-05, "loss": 1.5926, "step": 6472 }, { "epoch": 16.1825, "grad_norm": 5.4375, "learning_rate": 1.9102334234098786e-05, "loss": 1.4775, "step": 6473 }, { "epoch": 16.185, "grad_norm": 7.53125, "learning_rate": 1.9101998915730632e-05, "loss": 1.3624, "step": 6474 }, { "epoch": 16.1875, "grad_norm": 6.25, "learning_rate": 1.9101663537690134e-05, "loss": 1.5765, "step": 6475 }, { "epoch": 16.19, "grad_norm": 6.6875, "learning_rate": 1.9101328099979496e-05, "loss": 1.4616, "step": 6476 }, { "epoch": 16.1925, "grad_norm": 8.125, "learning_rate": 1.9100992602600913e-05, "loss": 1.6086, "step": 6477 }, { "epoch": 16.195, "grad_norm": 6.90625, "learning_rate": 1.9100657045556587e-05, "loss": 1.5592, "step": 6478 }, { "epoch": 16.1975, "grad_norm": 9.75, "learning_rate": 1.9100321428848717e-05, "loss": 1.5124, "step": 6479 }, { "epoch": 16.2, "grad_norm": 6.59375, "learning_rate": 1.9099985752479505e-05, "loss": 1.4833, "step": 6480 }, { "epoch": 16.2025, "grad_norm": 6.59375, "learning_rate": 1.909965001645115e-05, "loss": 1.4709, "step": 6481 }, { "epoch": 16.205, "grad_norm": 8.5625, "learning_rate": 1.9099314220765856e-05, "loss": 1.4237, "step": 6482 }, { "epoch": 16.2075, "grad_norm": 6.03125, "learning_rate": 1.909897836542582e-05, "loss": 1.1728, "step": 6483 }, { "epoch": 16.21, "grad_norm": 5.3125, "learning_rate": 1.9098642450433248e-05, "loss": 1.2738, "step": 6484 }, { "epoch": 16.2125, "grad_norm": 6.0625, "learning_rate": 1.909830647579034e-05, "loss": 1.3186, "step": 6485 }, { "epoch": 16.215, "grad_norm": 4.9375, "learning_rate": 1.9097970441499295e-05, "loss": 1.4519, "step": 6486 }, { "epoch": 16.2175, "grad_norm": 5.34375, "learning_rate": 1.9097634347562322e-05, "loss": 1.4074, "step": 6487 }, { "epoch": 16.22, "grad_norm": 6.9375, "learning_rate": 1.9097298193981623e-05, "loss": 1.4264, "step": 6488 }, { "epoch": 16.2225, "grad_norm": 8.4375, "learning_rate": 1.9096961980759402e-05, "loss": 1.5399, "step": 6489 }, { "epoch": 16.225, "grad_norm": 6.53125, "learning_rate": 1.9096625707897856e-05, "loss": 1.3265, "step": 6490 }, { "epoch": 16.2275, "grad_norm": 5.9375, "learning_rate": 1.9096289375399204e-05, "loss": 1.4205, "step": 6491 }, { "epoch": 16.23, "grad_norm": 6.0625, "learning_rate": 1.909595298326564e-05, "loss": 1.3567, "step": 6492 }, { "epoch": 16.2325, "grad_norm": 5.96875, "learning_rate": 1.909561653149937e-05, "loss": 1.4181, "step": 6493 }, { "epoch": 16.235, "grad_norm": 10.625, "learning_rate": 1.9095280020102604e-05, "loss": 1.4613, "step": 6494 }, { "epoch": 16.2375, "grad_norm": 4.75, "learning_rate": 1.9094943449077546e-05, "loss": 1.4028, "step": 6495 }, { "epoch": 16.24, "grad_norm": 6.84375, "learning_rate": 1.9094606818426403e-05, "loss": 1.583, "step": 6496 }, { "epoch": 16.2425, "grad_norm": 7.59375, "learning_rate": 1.9094270128151385e-05, "loss": 1.5989, "step": 6497 }, { "epoch": 16.245, "grad_norm": 6.375, "learning_rate": 1.909393337825469e-05, "loss": 1.5069, "step": 6498 }, { "epoch": 16.2475, "grad_norm": 7.8125, "learning_rate": 1.9093596568738536e-05, "loss": 1.36, "step": 6499 }, { "epoch": 16.25, "grad_norm": 6.15625, "learning_rate": 1.9093259699605125e-05, "loss": 1.442, "step": 6500 }, { "epoch": 16.2525, "grad_norm": 6.0625, "learning_rate": 1.9092922770856667e-05, "loss": 1.3319, "step": 6501 }, { "epoch": 16.255, "grad_norm": 7.1875, "learning_rate": 1.909258578249537e-05, "loss": 1.625, "step": 6502 }, { "epoch": 16.2575, "grad_norm": 6.8125, "learning_rate": 1.9092248734523443e-05, "loss": 1.4172, "step": 6503 }, { "epoch": 16.26, "grad_norm": 9.5, "learning_rate": 1.9091911626943104e-05, "loss": 1.4298, "step": 6504 }, { "epoch": 16.2625, "grad_norm": 6.75, "learning_rate": 1.9091574459756552e-05, "loss": 1.495, "step": 6505 }, { "epoch": 16.265, "grad_norm": 8.3125, "learning_rate": 1.9091237232966003e-05, "loss": 1.4016, "step": 6506 }, { "epoch": 16.2675, "grad_norm": 6.375, "learning_rate": 1.909089994657366e-05, "loss": 1.4606, "step": 6507 }, { "epoch": 16.27, "grad_norm": 7.625, "learning_rate": 1.9090562600581748e-05, "loss": 1.4527, "step": 6508 }, { "epoch": 16.2725, "grad_norm": 7.21875, "learning_rate": 1.909022519499247e-05, "loss": 1.4147, "step": 6509 }, { "epoch": 16.275, "grad_norm": 5.375, "learning_rate": 1.908988772980804e-05, "loss": 1.448, "step": 6510 }, { "epoch": 16.2775, "grad_norm": 7.5625, "learning_rate": 1.908955020503067e-05, "loss": 1.4765, "step": 6511 }, { "epoch": 16.28, "grad_norm": 6.9375, "learning_rate": 1.908921262066257e-05, "loss": 1.396, "step": 6512 }, { "epoch": 16.2825, "grad_norm": 8.0625, "learning_rate": 1.9088874976705955e-05, "loss": 1.3346, "step": 6513 }, { "epoch": 16.285, "grad_norm": 7.53125, "learning_rate": 1.9088537273163043e-05, "loss": 1.6004, "step": 6514 }, { "epoch": 16.2875, "grad_norm": 6.5625, "learning_rate": 1.9088199510036043e-05, "loss": 1.3691, "step": 6515 }, { "epoch": 16.29, "grad_norm": 9.25, "learning_rate": 1.908786168732717e-05, "loss": 1.4131, "step": 6516 }, { "epoch": 16.2925, "grad_norm": 5.4375, "learning_rate": 1.908752380503864e-05, "loss": 1.5111, "step": 6517 }, { "epoch": 16.295, "grad_norm": 7.84375, "learning_rate": 1.9087185863172667e-05, "loss": 1.4675, "step": 6518 }, { "epoch": 16.2975, "grad_norm": 7.34375, "learning_rate": 1.908684786173147e-05, "loss": 1.5996, "step": 6519 }, { "epoch": 16.3, "grad_norm": 7.0625, "learning_rate": 1.908650980071726e-05, "loss": 1.2579, "step": 6520 }, { "epoch": 16.3025, "grad_norm": 6.5, "learning_rate": 1.908617168013225e-05, "loss": 1.4487, "step": 6521 }, { "epoch": 16.305, "grad_norm": 9.4375, "learning_rate": 1.908583349997867e-05, "loss": 1.6972, "step": 6522 }, { "epoch": 16.3075, "grad_norm": 5.25, "learning_rate": 1.9085495260258725e-05, "loss": 1.3323, "step": 6523 }, { "epoch": 16.31, "grad_norm": 8.4375, "learning_rate": 1.908515696097464e-05, "loss": 1.4157, "step": 6524 }, { "epoch": 16.3125, "grad_norm": 5.5625, "learning_rate": 1.9084818602128627e-05, "loss": 1.38, "step": 6525 }, { "epoch": 16.315, "grad_norm": 5.78125, "learning_rate": 1.908448018372291e-05, "loss": 1.5452, "step": 6526 }, { "epoch": 16.3175, "grad_norm": 9.4375, "learning_rate": 1.90841417057597e-05, "loss": 1.622, "step": 6527 }, { "epoch": 16.32, "grad_norm": 6.8125, "learning_rate": 1.9083803168241225e-05, "loss": 1.4717, "step": 6528 }, { "epoch": 16.3225, "grad_norm": 6.96875, "learning_rate": 1.9083464571169694e-05, "loss": 1.4426, "step": 6529 }, { "epoch": 16.325, "grad_norm": 7.8125, "learning_rate": 1.9083125914547337e-05, "loss": 1.4458, "step": 6530 }, { "epoch": 16.3275, "grad_norm": 6.6875, "learning_rate": 1.908278719837637e-05, "loss": 1.3754, "step": 6531 }, { "epoch": 16.33, "grad_norm": 7.875, "learning_rate": 1.908244842265901e-05, "loss": 1.4969, "step": 6532 }, { "epoch": 16.3325, "grad_norm": 6.5, "learning_rate": 1.9082109587397487e-05, "loss": 1.4818, "step": 6533 }, { "epoch": 16.335, "grad_norm": 8.75, "learning_rate": 1.908177069259401e-05, "loss": 1.4432, "step": 6534 }, { "epoch": 16.3375, "grad_norm": 7.96875, "learning_rate": 1.9081431738250815e-05, "loss": 1.3186, "step": 6535 }, { "epoch": 16.34, "grad_norm": 8.125, "learning_rate": 1.9081092724370112e-05, "loss": 1.3439, "step": 6536 }, { "epoch": 16.3425, "grad_norm": 7.34375, "learning_rate": 1.9080753650954134e-05, "loss": 1.4293, "step": 6537 }, { "epoch": 16.345, "grad_norm": 7.875, "learning_rate": 1.9080414518005092e-05, "loss": 1.5787, "step": 6538 }, { "epoch": 16.3475, "grad_norm": 6.84375, "learning_rate": 1.908007532552522e-05, "loss": 1.5544, "step": 6539 }, { "epoch": 16.35, "grad_norm": 5.25, "learning_rate": 1.9079736073516735e-05, "loss": 1.3982, "step": 6540 }, { "epoch": 16.3525, "grad_norm": 6.375, "learning_rate": 1.9079396761981866e-05, "loss": 1.3732, "step": 6541 }, { "epoch": 16.355, "grad_norm": 7.34375, "learning_rate": 1.9079057390922836e-05, "loss": 1.481, "step": 6542 }, { "epoch": 16.3575, "grad_norm": 8.75, "learning_rate": 1.9078717960341863e-05, "loss": 1.5707, "step": 6543 }, { "epoch": 16.36, "grad_norm": 7.3125, "learning_rate": 1.9078378470241183e-05, "loss": 1.3861, "step": 6544 }, { "epoch": 16.3625, "grad_norm": 6.8125, "learning_rate": 1.9078038920623017e-05, "loss": 1.4486, "step": 6545 }, { "epoch": 16.365, "grad_norm": 8.5, "learning_rate": 1.907769931148959e-05, "loss": 1.5701, "step": 6546 }, { "epoch": 16.3675, "grad_norm": 7.96875, "learning_rate": 1.9077359642843132e-05, "loss": 1.4121, "step": 6547 }, { "epoch": 16.37, "grad_norm": 6.09375, "learning_rate": 1.9077019914685864e-05, "loss": 1.5191, "step": 6548 }, { "epoch": 16.3725, "grad_norm": 4.78125, "learning_rate": 1.907668012702002e-05, "loss": 1.3762, "step": 6549 }, { "epoch": 16.375, "grad_norm": 6.0625, "learning_rate": 1.907634027984782e-05, "loss": 1.3487, "step": 6550 }, { "epoch": 16.3775, "grad_norm": 6.09375, "learning_rate": 1.90760003731715e-05, "loss": 1.3006, "step": 6551 }, { "epoch": 16.38, "grad_norm": 8.125, "learning_rate": 1.9075660406993285e-05, "loss": 1.3232, "step": 6552 }, { "epoch": 16.3825, "grad_norm": 8.375, "learning_rate": 1.90753203813154e-05, "loss": 1.4474, "step": 6553 }, { "epoch": 16.385, "grad_norm": 6.84375, "learning_rate": 1.907498029614008e-05, "loss": 1.3569, "step": 6554 }, { "epoch": 16.3875, "grad_norm": 6.5, "learning_rate": 1.9074640151469553e-05, "loss": 1.5245, "step": 6555 }, { "epoch": 16.39, "grad_norm": 5.78125, "learning_rate": 1.907429994730605e-05, "loss": 1.3858, "step": 6556 }, { "epoch": 16.3925, "grad_norm": 8.0625, "learning_rate": 1.9073959683651796e-05, "loss": 1.3995, "step": 6557 }, { "epoch": 16.395, "grad_norm": 5.75, "learning_rate": 1.9073619360509025e-05, "loss": 1.3534, "step": 6558 }, { "epoch": 16.3975, "grad_norm": 8.1875, "learning_rate": 1.9073278977879973e-05, "loss": 1.5081, "step": 6559 }, { "epoch": 16.4, "grad_norm": 4.625, "learning_rate": 1.9072938535766864e-05, "loss": 1.3808, "step": 6560 }, { "epoch": 16.4025, "grad_norm": 6.375, "learning_rate": 1.9072598034171934e-05, "loss": 1.449, "step": 6561 }, { "epoch": 16.405, "grad_norm": 9.8125, "learning_rate": 1.9072257473097414e-05, "loss": 1.4428, "step": 6562 }, { "epoch": 16.4075, "grad_norm": 5.75, "learning_rate": 1.907191685254554e-05, "loss": 1.5447, "step": 6563 }, { "epoch": 16.41, "grad_norm": 6.25, "learning_rate": 1.9071576172518534e-05, "loss": 1.4495, "step": 6564 }, { "epoch": 16.4125, "grad_norm": 6.96875, "learning_rate": 1.9071235433018645e-05, "loss": 1.3849, "step": 6565 }, { "epoch": 16.415, "grad_norm": 8.5625, "learning_rate": 1.9070894634048096e-05, "loss": 1.3241, "step": 6566 }, { "epoch": 16.4175, "grad_norm": 5.5, "learning_rate": 1.9070553775609125e-05, "loss": 1.2876, "step": 6567 }, { "epoch": 16.42, "grad_norm": 8.3125, "learning_rate": 1.907021285770397e-05, "loss": 1.391, "step": 6568 }, { "epoch": 16.4225, "grad_norm": 7.25, "learning_rate": 1.9069871880334856e-05, "loss": 1.4607, "step": 6569 }, { "epoch": 16.425, "grad_norm": 5.28125, "learning_rate": 1.906953084350403e-05, "loss": 1.4412, "step": 6570 }, { "epoch": 16.4275, "grad_norm": 6.40625, "learning_rate": 1.9069189747213723e-05, "loss": 1.309, "step": 6571 }, { "epoch": 16.43, "grad_norm": 6.21875, "learning_rate": 1.906884859146617e-05, "loss": 1.486, "step": 6572 }, { "epoch": 16.4325, "grad_norm": 7.125, "learning_rate": 1.9068507376263607e-05, "loss": 1.3103, "step": 6573 }, { "epoch": 16.435, "grad_norm": 6.8125, "learning_rate": 1.9068166101608275e-05, "loss": 1.4402, "step": 6574 }, { "epoch": 16.4375, "grad_norm": 7.9375, "learning_rate": 1.9067824767502404e-05, "loss": 1.448, "step": 6575 }, { "epoch": 16.44, "grad_norm": 7.0, "learning_rate": 1.9067483373948245e-05, "loss": 1.4698, "step": 6576 }, { "epoch": 16.4425, "grad_norm": 6.46875, "learning_rate": 1.906714192094802e-05, "loss": 1.3913, "step": 6577 }, { "epoch": 16.445, "grad_norm": 5.75, "learning_rate": 1.9066800408503982e-05, "loss": 1.3728, "step": 6578 }, { "epoch": 16.4475, "grad_norm": 7.875, "learning_rate": 1.9066458836618358e-05, "loss": 1.6052, "step": 6579 }, { "epoch": 16.45, "grad_norm": 6.65625, "learning_rate": 1.9066117205293393e-05, "loss": 1.4425, "step": 6580 }, { "epoch": 16.4525, "grad_norm": 6.25, "learning_rate": 1.9065775514531327e-05, "loss": 1.459, "step": 6581 }, { "epoch": 16.455, "grad_norm": 7.0625, "learning_rate": 1.9065433764334402e-05, "loss": 1.4412, "step": 6582 }, { "epoch": 16.4575, "grad_norm": 6.8125, "learning_rate": 1.9065091954704855e-05, "loss": 1.279, "step": 6583 }, { "epoch": 16.46, "grad_norm": 6.9375, "learning_rate": 1.906475008564493e-05, "loss": 1.5707, "step": 6584 }, { "epoch": 16.4625, "grad_norm": 6.46875, "learning_rate": 1.9064408157156862e-05, "loss": 1.6534, "step": 6585 }, { "epoch": 16.465, "grad_norm": 7.4375, "learning_rate": 1.9064066169242895e-05, "loss": 1.3495, "step": 6586 }, { "epoch": 16.4675, "grad_norm": 6.96875, "learning_rate": 1.9063724121905277e-05, "loss": 1.6399, "step": 6587 }, { "epoch": 16.47, "grad_norm": 5.9375, "learning_rate": 1.906338201514625e-05, "loss": 1.3994, "step": 6588 }, { "epoch": 16.4725, "grad_norm": 6.90625, "learning_rate": 1.9063039848968045e-05, "loss": 1.3442, "step": 6589 }, { "epoch": 16.475, "grad_norm": 6.3125, "learning_rate": 1.906269762337292e-05, "loss": 1.6424, "step": 6590 }, { "epoch": 16.4775, "grad_norm": 5.6875, "learning_rate": 1.906235533836311e-05, "loss": 1.5063, "step": 6591 }, { "epoch": 16.48, "grad_norm": 9.3125, "learning_rate": 1.906201299394086e-05, "loss": 1.4628, "step": 6592 }, { "epoch": 16.4825, "grad_norm": 8.5625, "learning_rate": 1.9061670590108415e-05, "loss": 1.542, "step": 6593 }, { "epoch": 16.485, "grad_norm": 6.6875, "learning_rate": 1.906132812686802e-05, "loss": 1.3885, "step": 6594 }, { "epoch": 16.4875, "grad_norm": 8.0625, "learning_rate": 1.9060985604221922e-05, "loss": 1.1728, "step": 6595 }, { "epoch": 16.49, "grad_norm": 6.9375, "learning_rate": 1.9060643022172364e-05, "loss": 1.5127, "step": 6596 }, { "epoch": 16.4925, "grad_norm": 10.9375, "learning_rate": 1.9060300380721595e-05, "loss": 1.5904, "step": 6597 }, { "epoch": 16.495, "grad_norm": 7.8125, "learning_rate": 1.905995767987186e-05, "loss": 1.5275, "step": 6598 }, { "epoch": 16.4975, "grad_norm": 8.5, "learning_rate": 1.90596149196254e-05, "loss": 1.4159, "step": 6599 }, { "epoch": 16.5, "grad_norm": 6.34375, "learning_rate": 1.905927209998447e-05, "loss": 1.3286, "step": 6600 }, { "epoch": 16.5025, "grad_norm": 6.3125, "learning_rate": 1.9058929220951313e-05, "loss": 1.3424, "step": 6601 }, { "epoch": 16.505, "grad_norm": 9.5625, "learning_rate": 1.905858628252818e-05, "loss": 1.5841, "step": 6602 }, { "epoch": 16.5075, "grad_norm": 6.4375, "learning_rate": 1.905824328471732e-05, "loss": 1.4247, "step": 6603 }, { "epoch": 16.51, "grad_norm": 5.625, "learning_rate": 1.9057900227520977e-05, "loss": 1.4443, "step": 6604 }, { "epoch": 16.5125, "grad_norm": 9.875, "learning_rate": 1.90575571109414e-05, "loss": 1.6592, "step": 6605 }, { "epoch": 16.515, "grad_norm": 7.3125, "learning_rate": 1.9057213934980846e-05, "loss": 1.4722, "step": 6606 }, { "epoch": 16.5175, "grad_norm": 6.09375, "learning_rate": 1.905687069964156e-05, "loss": 1.3243, "step": 6607 }, { "epoch": 16.52, "grad_norm": 5.3125, "learning_rate": 1.9056527404925788e-05, "loss": 1.4273, "step": 6608 }, { "epoch": 16.5225, "grad_norm": 9.4375, "learning_rate": 1.905618405083579e-05, "loss": 1.3915, "step": 6609 }, { "epoch": 16.525, "grad_norm": 6.75, "learning_rate": 1.9055840637373805e-05, "loss": 1.5119, "step": 6610 }, { "epoch": 16.5275, "grad_norm": 6.40625, "learning_rate": 1.9055497164542095e-05, "loss": 1.3413, "step": 6611 }, { "epoch": 16.53, "grad_norm": 5.15625, "learning_rate": 1.9055153632342907e-05, "loss": 1.3271, "step": 6612 }, { "epoch": 16.5325, "grad_norm": 6.375, "learning_rate": 1.9054810040778497e-05, "loss": 1.4864, "step": 6613 }, { "epoch": 16.535, "grad_norm": 5.46875, "learning_rate": 1.905446638985111e-05, "loss": 1.46, "step": 6614 }, { "epoch": 16.5375, "grad_norm": 6.03125, "learning_rate": 1.905412267956301e-05, "loss": 1.4992, "step": 6615 }, { "epoch": 16.54, "grad_norm": 6.9375, "learning_rate": 1.905377890991644e-05, "loss": 1.3932, "step": 6616 }, { "epoch": 16.5425, "grad_norm": 6.6875, "learning_rate": 1.9053435080913658e-05, "loss": 1.5368, "step": 6617 }, { "epoch": 16.545, "grad_norm": 7.75, "learning_rate": 1.9053091192556916e-05, "loss": 1.7336, "step": 6618 }, { "epoch": 16.5475, "grad_norm": 9.375, "learning_rate": 1.9052747244848473e-05, "loss": 1.4755, "step": 6619 }, { "epoch": 16.55, "grad_norm": 8.6875, "learning_rate": 1.905240323779058e-05, "loss": 1.3682, "step": 6620 }, { "epoch": 16.5525, "grad_norm": 10.3125, "learning_rate": 1.9052059171385497e-05, "loss": 1.4943, "step": 6621 }, { "epoch": 16.555, "grad_norm": 8.5625, "learning_rate": 1.9051715045635472e-05, "loss": 1.4199, "step": 6622 }, { "epoch": 16.5575, "grad_norm": 10.25, "learning_rate": 1.9051370860542767e-05, "loss": 1.5331, "step": 6623 }, { "epoch": 16.56, "grad_norm": 7.65625, "learning_rate": 1.9051026616109637e-05, "loss": 1.3545, "step": 6624 }, { "epoch": 16.5625, "grad_norm": 6.15625, "learning_rate": 1.905068231233834e-05, "loss": 1.3854, "step": 6625 }, { "epoch": 16.565, "grad_norm": 4.4375, "learning_rate": 1.905033794923113e-05, "loss": 1.4473, "step": 6626 }, { "epoch": 16.5675, "grad_norm": 7.09375, "learning_rate": 1.9049993526790268e-05, "loss": 1.2501, "step": 6627 }, { "epoch": 16.57, "grad_norm": 6.09375, "learning_rate": 1.9049649045018008e-05, "loss": 1.4231, "step": 6628 }, { "epoch": 16.5725, "grad_norm": 10.25, "learning_rate": 1.9049304503916616e-05, "loss": 1.5532, "step": 6629 }, { "epoch": 16.575, "grad_norm": 5.71875, "learning_rate": 1.904895990348834e-05, "loss": 1.3533, "step": 6630 }, { "epoch": 16.5775, "grad_norm": 7.5, "learning_rate": 1.904861524373545e-05, "loss": 1.5188, "step": 6631 }, { "epoch": 16.58, "grad_norm": 6.25, "learning_rate": 1.9048270524660197e-05, "loss": 1.3968, "step": 6632 }, { "epoch": 16.5825, "grad_norm": 7.34375, "learning_rate": 1.9047925746264845e-05, "loss": 1.3237, "step": 6633 }, { "epoch": 16.585, "grad_norm": 5.25, "learning_rate": 1.9047580908551655e-05, "loss": 1.449, "step": 6634 }, { "epoch": 16.5875, "grad_norm": 7.625, "learning_rate": 1.904723601152289e-05, "loss": 1.4076, "step": 6635 }, { "epoch": 16.59, "grad_norm": 5.8125, "learning_rate": 1.90468910551808e-05, "loss": 1.4023, "step": 6636 }, { "epoch": 16.5925, "grad_norm": 9.4375, "learning_rate": 1.904654603952766e-05, "loss": 1.4536, "step": 6637 }, { "epoch": 16.595, "grad_norm": 8.625, "learning_rate": 1.9046200964565722e-05, "loss": 1.3288, "step": 6638 }, { "epoch": 16.5975, "grad_norm": 6.90625, "learning_rate": 1.9045855830297254e-05, "loss": 1.3536, "step": 6639 }, { "epoch": 16.6, "grad_norm": 6.1875, "learning_rate": 1.904551063672452e-05, "loss": 1.446, "step": 6640 }, { "epoch": 16.6025, "grad_norm": 5.15625, "learning_rate": 1.9045165383849778e-05, "loss": 1.3344, "step": 6641 }, { "epoch": 16.605, "grad_norm": 7.9375, "learning_rate": 1.9044820071675297e-05, "loss": 1.5148, "step": 6642 }, { "epoch": 16.6075, "grad_norm": 7.34375, "learning_rate": 1.9044474700203332e-05, "loss": 1.4059, "step": 6643 }, { "epoch": 16.61, "grad_norm": 4.03125, "learning_rate": 1.9044129269436155e-05, "loss": 1.5035, "step": 6644 }, { "epoch": 16.6125, "grad_norm": 7.25, "learning_rate": 1.904378377937603e-05, "loss": 1.5779, "step": 6645 }, { "epoch": 16.615, "grad_norm": 5.28125, "learning_rate": 1.904343823002522e-05, "loss": 1.4592, "step": 6646 }, { "epoch": 16.6175, "grad_norm": 6.46875, "learning_rate": 1.904309262138599e-05, "loss": 1.4611, "step": 6647 }, { "epoch": 16.62, "grad_norm": 4.9375, "learning_rate": 1.9042746953460607e-05, "loss": 1.4444, "step": 6648 }, { "epoch": 16.6225, "grad_norm": 8.6875, "learning_rate": 1.9042401226251335e-05, "loss": 1.4681, "step": 6649 }, { "epoch": 16.625, "grad_norm": 13.5625, "learning_rate": 1.9042055439760447e-05, "loss": 1.5032, "step": 6650 }, { "epoch": 16.6275, "grad_norm": 6.25, "learning_rate": 1.90417095939902e-05, "loss": 1.3978, "step": 6651 }, { "epoch": 16.63, "grad_norm": 8.75, "learning_rate": 1.904136368894287e-05, "loss": 1.4024, "step": 6652 }, { "epoch": 16.6325, "grad_norm": 7.8125, "learning_rate": 1.904101772462072e-05, "loss": 1.4864, "step": 6653 }, { "epoch": 16.635, "grad_norm": 6.1875, "learning_rate": 1.9040671701026022e-05, "loss": 1.3876, "step": 6654 }, { "epoch": 16.6375, "grad_norm": 7.46875, "learning_rate": 1.904032561816104e-05, "loss": 1.6685, "step": 6655 }, { "epoch": 16.64, "grad_norm": 6.3125, "learning_rate": 1.9039979476028044e-05, "loss": 1.5782, "step": 6656 }, { "epoch": 16.6425, "grad_norm": 6.40625, "learning_rate": 1.9039633274629308e-05, "loss": 1.516, "step": 6657 }, { "epoch": 16.645, "grad_norm": 7.59375, "learning_rate": 1.9039287013967092e-05, "loss": 1.5531, "step": 6658 }, { "epoch": 16.6475, "grad_norm": 6.96875, "learning_rate": 1.903894069404368e-05, "loss": 1.3778, "step": 6659 }, { "epoch": 16.65, "grad_norm": 7.5625, "learning_rate": 1.9038594314861328e-05, "loss": 1.304, "step": 6660 }, { "epoch": 16.6525, "grad_norm": 7.625, "learning_rate": 1.9038247876422316e-05, "loss": 1.4928, "step": 6661 }, { "epoch": 16.655, "grad_norm": 6.34375, "learning_rate": 1.9037901378728914e-05, "loss": 1.3666, "step": 6662 }, { "epoch": 16.6575, "grad_norm": 6.5625, "learning_rate": 1.903755482178339e-05, "loss": 1.3427, "step": 6663 }, { "epoch": 16.66, "grad_norm": 8.5, "learning_rate": 1.903720820558802e-05, "loss": 1.3981, "step": 6664 }, { "epoch": 16.6625, "grad_norm": 7.5, "learning_rate": 1.903686153014507e-05, "loss": 1.4838, "step": 6665 }, { "epoch": 16.665, "grad_norm": 5.5, "learning_rate": 1.903651479545682e-05, "loss": 1.2547, "step": 6666 }, { "epoch": 16.6675, "grad_norm": 6.625, "learning_rate": 1.9036168001525542e-05, "loss": 1.4943, "step": 6667 }, { "epoch": 16.67, "grad_norm": 7.28125, "learning_rate": 1.9035821148353507e-05, "loss": 1.3963, "step": 6668 }, { "epoch": 16.6725, "grad_norm": 5.21875, "learning_rate": 1.9035474235942996e-05, "loss": 1.4949, "step": 6669 }, { "epoch": 16.675, "grad_norm": 7.0, "learning_rate": 1.903512726429627e-05, "loss": 1.5844, "step": 6670 }, { "epoch": 16.6775, "grad_norm": 7.46875, "learning_rate": 1.9034780233415614e-05, "loss": 1.4869, "step": 6671 }, { "epoch": 16.68, "grad_norm": 6.90625, "learning_rate": 1.90344331433033e-05, "loss": 1.5914, "step": 6672 }, { "epoch": 16.6825, "grad_norm": 6.0, "learning_rate": 1.9034085993961603e-05, "loss": 1.3819, "step": 6673 }, { "epoch": 16.685, "grad_norm": 6.78125, "learning_rate": 1.90337387853928e-05, "loss": 1.4552, "step": 6674 }, { "epoch": 16.6875, "grad_norm": 5.5625, "learning_rate": 1.9033391517599166e-05, "loss": 1.6099, "step": 6675 }, { "epoch": 16.69, "grad_norm": 7.09375, "learning_rate": 1.903304419058298e-05, "loss": 1.3741, "step": 6676 }, { "epoch": 16.6925, "grad_norm": 5.6875, "learning_rate": 1.903269680434652e-05, "loss": 1.3241, "step": 6677 }, { "epoch": 16.695, "grad_norm": 6.96875, "learning_rate": 1.903234935889206e-05, "loss": 1.4812, "step": 6678 }, { "epoch": 16.6975, "grad_norm": 5.5, "learning_rate": 1.9032001854221878e-05, "loss": 1.5279, "step": 6679 }, { "epoch": 16.7, "grad_norm": 7.40625, "learning_rate": 1.9031654290338256e-05, "loss": 1.3964, "step": 6680 }, { "epoch": 16.7025, "grad_norm": 10.625, "learning_rate": 1.9031306667243467e-05, "loss": 1.7617, "step": 6681 }, { "epoch": 16.705, "grad_norm": 6.84375, "learning_rate": 1.903095898493979e-05, "loss": 1.4683, "step": 6682 }, { "epoch": 16.7075, "grad_norm": 6.15625, "learning_rate": 1.9030611243429514e-05, "loss": 1.4038, "step": 6683 }, { "epoch": 16.71, "grad_norm": 5.40625, "learning_rate": 1.9030263442714908e-05, "loss": 1.3803, "step": 6684 }, { "epoch": 16.7125, "grad_norm": 8.8125, "learning_rate": 1.9029915582798258e-05, "loss": 1.5738, "step": 6685 }, { "epoch": 16.715, "grad_norm": 6.71875, "learning_rate": 1.902956766368184e-05, "loss": 1.5196, "step": 6686 }, { "epoch": 16.7175, "grad_norm": 6.6875, "learning_rate": 1.902921968536794e-05, "loss": 1.3448, "step": 6687 }, { "epoch": 16.72, "grad_norm": 6.21875, "learning_rate": 1.9028871647858836e-05, "loss": 1.395, "step": 6688 }, { "epoch": 16.7225, "grad_norm": 7.9375, "learning_rate": 1.9028523551156812e-05, "loss": 1.3624, "step": 6689 }, { "epoch": 16.725, "grad_norm": 6.0625, "learning_rate": 1.9028175395264147e-05, "loss": 1.3613, "step": 6690 }, { "epoch": 16.7275, "grad_norm": 8.8125, "learning_rate": 1.9027827180183127e-05, "loss": 1.4099, "step": 6691 }, { "epoch": 16.73, "grad_norm": 6.3125, "learning_rate": 1.9027478905916035e-05, "loss": 1.4641, "step": 6692 }, { "epoch": 16.7325, "grad_norm": 7.78125, "learning_rate": 1.902713057246515e-05, "loss": 1.4219, "step": 6693 }, { "epoch": 16.735, "grad_norm": 8.0625, "learning_rate": 1.9026782179832758e-05, "loss": 1.6312, "step": 6694 }, { "epoch": 16.7375, "grad_norm": 8.1875, "learning_rate": 1.9026433728021144e-05, "loss": 1.3709, "step": 6695 }, { "epoch": 16.74, "grad_norm": 5.3125, "learning_rate": 1.9026085217032592e-05, "loss": 1.4249, "step": 6696 }, { "epoch": 16.7425, "grad_norm": 9.4375, "learning_rate": 1.9025736646869384e-05, "loss": 1.548, "step": 6697 }, { "epoch": 16.745, "grad_norm": 5.6875, "learning_rate": 1.902538801753381e-05, "loss": 1.3902, "step": 6698 }, { "epoch": 16.7475, "grad_norm": 9.8125, "learning_rate": 1.9025039329028154e-05, "loss": 1.438, "step": 6699 }, { "epoch": 16.75, "grad_norm": 5.34375, "learning_rate": 1.90246905813547e-05, "loss": 1.3281, "step": 6700 }, { "epoch": 16.7525, "grad_norm": 6.5625, "learning_rate": 1.9024341774515738e-05, "loss": 1.3961, "step": 6701 }, { "epoch": 16.755, "grad_norm": 5.4375, "learning_rate": 1.902399290851355e-05, "loss": 1.4356, "step": 6702 }, { "epoch": 16.7575, "grad_norm": 9.375, "learning_rate": 1.9023643983350425e-05, "loss": 1.3827, "step": 6703 }, { "epoch": 16.76, "grad_norm": 10.9375, "learning_rate": 1.9023294999028654e-05, "loss": 1.5778, "step": 6704 }, { "epoch": 16.7625, "grad_norm": 6.46875, "learning_rate": 1.902294595555052e-05, "loss": 1.4668, "step": 6705 }, { "epoch": 16.765, "grad_norm": 6.03125, "learning_rate": 1.9022596852918314e-05, "loss": 1.5957, "step": 6706 }, { "epoch": 16.7675, "grad_norm": 5.28125, "learning_rate": 1.9022247691134325e-05, "loss": 1.3752, "step": 6707 }, { "epoch": 16.77, "grad_norm": 7.75, "learning_rate": 1.902189847020084e-05, "loss": 1.3562, "step": 6708 }, { "epoch": 16.7725, "grad_norm": 6.3125, "learning_rate": 1.902154919012015e-05, "loss": 1.7195, "step": 6709 }, { "epoch": 16.775, "grad_norm": 5.1875, "learning_rate": 1.9021199850894544e-05, "loss": 1.427, "step": 6710 }, { "epoch": 16.7775, "grad_norm": 5.625, "learning_rate": 1.9020850452526315e-05, "loss": 1.4314, "step": 6711 }, { "epoch": 16.78, "grad_norm": 5.59375, "learning_rate": 1.9020500995017747e-05, "loss": 1.3888, "step": 6712 }, { "epoch": 16.7825, "grad_norm": 8.125, "learning_rate": 1.9020151478371138e-05, "loss": 1.5664, "step": 6713 }, { "epoch": 16.785, "grad_norm": 6.90625, "learning_rate": 1.9019801902588778e-05, "loss": 1.4695, "step": 6714 }, { "epoch": 16.7875, "grad_norm": 7.5, "learning_rate": 1.9019452267672957e-05, "loss": 1.528, "step": 6715 }, { "epoch": 16.79, "grad_norm": 7.5, "learning_rate": 1.9019102573625966e-05, "loss": 1.5267, "step": 6716 }, { "epoch": 16.7925, "grad_norm": 11.1875, "learning_rate": 1.9018752820450103e-05, "loss": 1.5203, "step": 6717 }, { "epoch": 16.795, "grad_norm": 6.875, "learning_rate": 1.9018403008147656e-05, "loss": 1.4647, "step": 6718 }, { "epoch": 16.7975, "grad_norm": 8.4375, "learning_rate": 1.901805313672092e-05, "loss": 1.4584, "step": 6719 }, { "epoch": 16.8, "grad_norm": 4.25, "learning_rate": 1.9017703206172187e-05, "loss": 1.4482, "step": 6720 }, { "epoch": 16.8025, "grad_norm": 5.375, "learning_rate": 1.9017353216503752e-05, "loss": 1.5641, "step": 6721 }, { "epoch": 16.805, "grad_norm": 10.3125, "learning_rate": 1.901700316771791e-05, "loss": 1.4105, "step": 6722 }, { "epoch": 16.8075, "grad_norm": 8.0, "learning_rate": 1.901665305981696e-05, "loss": 1.3225, "step": 6723 }, { "epoch": 16.81, "grad_norm": 7.1875, "learning_rate": 1.901630289280319e-05, "loss": 1.4838, "step": 6724 }, { "epoch": 16.8125, "grad_norm": 5.78125, "learning_rate": 1.9015952666678902e-05, "loss": 1.4538, "step": 6725 }, { "epoch": 16.815, "grad_norm": 8.0625, "learning_rate": 1.9015602381446387e-05, "loss": 1.3529, "step": 6726 }, { "epoch": 16.8175, "grad_norm": 7.21875, "learning_rate": 1.9015252037107944e-05, "loss": 1.3386, "step": 6727 }, { "epoch": 16.82, "grad_norm": 7.125, "learning_rate": 1.901490163366587e-05, "loss": 1.4632, "step": 6728 }, { "epoch": 16.8225, "grad_norm": 9.8125, "learning_rate": 1.9014551171122456e-05, "loss": 1.4074, "step": 6729 }, { "epoch": 16.825, "grad_norm": 7.25, "learning_rate": 1.9014200649480014e-05, "loss": 1.6085, "step": 6730 }, { "epoch": 16.8275, "grad_norm": 182.0, "learning_rate": 1.9013850068740828e-05, "loss": 1.2767, "step": 6731 }, { "epoch": 16.83, "grad_norm": 8.0625, "learning_rate": 1.9013499428907202e-05, "loss": 1.3527, "step": 6732 }, { "epoch": 16.8325, "grad_norm": 6.25, "learning_rate": 1.9013148729981438e-05, "loss": 1.5332, "step": 6733 }, { "epoch": 16.835, "grad_norm": 7.90625, "learning_rate": 1.901279797196583e-05, "loss": 1.4597, "step": 6734 }, { "epoch": 16.8375, "grad_norm": 9.625, "learning_rate": 1.9012447154862674e-05, "loss": 1.3553, "step": 6735 }, { "epoch": 16.84, "grad_norm": 5.625, "learning_rate": 1.9012096278674283e-05, "loss": 1.3699, "step": 6736 }, { "epoch": 16.8425, "grad_norm": 5.03125, "learning_rate": 1.901174534340294e-05, "loss": 1.4331, "step": 6737 }, { "epoch": 16.845, "grad_norm": 13.875, "learning_rate": 1.9011394349050964e-05, "loss": 1.5607, "step": 6738 }, { "epoch": 16.8475, "grad_norm": 8.0625, "learning_rate": 1.9011043295620647e-05, "loss": 1.5638, "step": 6739 }, { "epoch": 16.85, "grad_norm": 15.8125, "learning_rate": 1.9010692183114285e-05, "loss": 1.4051, "step": 6740 }, { "epoch": 16.8525, "grad_norm": 6.75, "learning_rate": 1.901034101153419e-05, "loss": 1.5008, "step": 6741 }, { "epoch": 16.855, "grad_norm": 7.0625, "learning_rate": 1.9009989780882655e-05, "loss": 1.5226, "step": 6742 }, { "epoch": 16.8575, "grad_norm": 6.78125, "learning_rate": 1.900963849116199e-05, "loss": 1.3677, "step": 6743 }, { "epoch": 16.86, "grad_norm": 5.90625, "learning_rate": 1.90092871423745e-05, "loss": 1.6253, "step": 6744 }, { "epoch": 16.8625, "grad_norm": 7.09375, "learning_rate": 1.9008935734522482e-05, "loss": 1.4371, "step": 6745 }, { "epoch": 16.865, "grad_norm": 8.8125, "learning_rate": 1.9008584267608238e-05, "loss": 1.5847, "step": 6746 }, { "epoch": 16.8675, "grad_norm": 7.125, "learning_rate": 1.900823274163408e-05, "loss": 1.443, "step": 6747 }, { "epoch": 16.87, "grad_norm": 6.28125, "learning_rate": 1.9007881156602307e-05, "loss": 1.4818, "step": 6748 }, { "epoch": 16.8725, "grad_norm": 8.875, "learning_rate": 1.9007529512515225e-05, "loss": 1.6387, "step": 6749 }, { "epoch": 16.875, "grad_norm": 7.5625, "learning_rate": 1.900717780937514e-05, "loss": 1.355, "step": 6750 }, { "epoch": 16.8775, "grad_norm": 5.71875, "learning_rate": 1.900682604718436e-05, "loss": 1.6168, "step": 6751 }, { "epoch": 16.88, "grad_norm": 7.875, "learning_rate": 1.900647422594519e-05, "loss": 1.4675, "step": 6752 }, { "epoch": 16.8825, "grad_norm": 7.0, "learning_rate": 1.900612234565993e-05, "loss": 1.4408, "step": 6753 }, { "epoch": 16.885, "grad_norm": 5.90625, "learning_rate": 1.9005770406330895e-05, "loss": 1.4638, "step": 6754 }, { "epoch": 16.8875, "grad_norm": 7.375, "learning_rate": 1.900541840796039e-05, "loss": 1.4665, "step": 6755 }, { "epoch": 16.89, "grad_norm": 5.96875, "learning_rate": 1.9005066350550724e-05, "loss": 1.5314, "step": 6756 }, { "epoch": 16.8925, "grad_norm": 12.9375, "learning_rate": 1.90047142341042e-05, "loss": 1.5721, "step": 6757 }, { "epoch": 16.895, "grad_norm": 5.84375, "learning_rate": 1.900436205862313e-05, "loss": 1.4685, "step": 6758 }, { "epoch": 16.8975, "grad_norm": 6.3125, "learning_rate": 1.9004009824109824e-05, "loss": 1.4483, "step": 6759 }, { "epoch": 16.9, "grad_norm": 8.5625, "learning_rate": 1.900365753056659e-05, "loss": 1.3985, "step": 6760 }, { "epoch": 16.9025, "grad_norm": 6.34375, "learning_rate": 1.9003305177995736e-05, "loss": 1.5247, "step": 6761 }, { "epoch": 16.905, "grad_norm": 9.1875, "learning_rate": 1.9002952766399578e-05, "loss": 1.4895, "step": 6762 }, { "epoch": 16.9075, "grad_norm": 6.53125, "learning_rate": 1.900260029578042e-05, "loss": 1.5513, "step": 6763 }, { "epoch": 16.91, "grad_norm": 4.78125, "learning_rate": 1.900224776614057e-05, "loss": 1.4558, "step": 6764 }, { "epoch": 16.9125, "grad_norm": 4.28125, "learning_rate": 1.900189517748235e-05, "loss": 1.3386, "step": 6765 }, { "epoch": 16.915, "grad_norm": 7.6875, "learning_rate": 1.900154252980806e-05, "loss": 1.5193, "step": 6766 }, { "epoch": 16.9175, "grad_norm": 7.5625, "learning_rate": 1.900118982312002e-05, "loss": 1.5419, "step": 6767 }, { "epoch": 16.92, "grad_norm": 7.03125, "learning_rate": 1.900083705742054e-05, "loss": 1.2757, "step": 6768 }, { "epoch": 16.9225, "grad_norm": 7.46875, "learning_rate": 1.9000484232711933e-05, "loss": 1.5116, "step": 6769 }, { "epoch": 16.925, "grad_norm": 8.4375, "learning_rate": 1.900013134899651e-05, "loss": 1.513, "step": 6770 }, { "epoch": 16.9275, "grad_norm": 7.03125, "learning_rate": 1.899977840627659e-05, "loss": 1.4499, "step": 6771 }, { "epoch": 16.93, "grad_norm": 5.0, "learning_rate": 1.899942540455448e-05, "loss": 1.4804, "step": 6772 }, { "epoch": 16.9325, "grad_norm": 6.5625, "learning_rate": 1.8999072343832495e-05, "loss": 1.5117, "step": 6773 }, { "epoch": 16.935, "grad_norm": 4.375, "learning_rate": 1.8998719224112955e-05, "loss": 1.4651, "step": 6774 }, { "epoch": 16.9375, "grad_norm": 6.53125, "learning_rate": 1.8998366045398172e-05, "loss": 1.4719, "step": 6775 }, { "epoch": 16.94, "grad_norm": 6.5, "learning_rate": 1.899801280769046e-05, "loss": 1.4239, "step": 6776 }, { "epoch": 16.9425, "grad_norm": 6.375, "learning_rate": 1.899765951099213e-05, "loss": 1.4531, "step": 6777 }, { "epoch": 16.945, "grad_norm": 8.5625, "learning_rate": 1.8997306155305515e-05, "loss": 1.3796, "step": 6778 }, { "epoch": 16.9475, "grad_norm": 6.59375, "learning_rate": 1.8996952740632917e-05, "loss": 1.4086, "step": 6779 }, { "epoch": 16.95, "grad_norm": 6.53125, "learning_rate": 1.8996599266976658e-05, "loss": 1.5206, "step": 6780 }, { "epoch": 16.9525, "grad_norm": 7.21875, "learning_rate": 1.8996245734339048e-05, "loss": 1.5312, "step": 6781 }, { "epoch": 16.955, "grad_norm": 7.75, "learning_rate": 1.8995892142722416e-05, "loss": 1.572, "step": 6782 }, { "epoch": 16.9575, "grad_norm": 5.53125, "learning_rate": 1.8995538492129075e-05, "loss": 1.366, "step": 6783 }, { "epoch": 16.96, "grad_norm": 6.84375, "learning_rate": 1.8995184782561343e-05, "loss": 1.6279, "step": 6784 }, { "epoch": 16.9625, "grad_norm": 10.3125, "learning_rate": 1.899483101402154e-05, "loss": 1.4325, "step": 6785 }, { "epoch": 16.965, "grad_norm": 6.84375, "learning_rate": 1.8994477186511986e-05, "loss": 1.3664, "step": 6786 }, { "epoch": 16.9675, "grad_norm": 4.5625, "learning_rate": 1.8994123300034995e-05, "loss": 1.4237, "step": 6787 }, { "epoch": 16.97, "grad_norm": 6.71875, "learning_rate": 1.8993769354592897e-05, "loss": 1.5411, "step": 6788 }, { "epoch": 16.9725, "grad_norm": 9.0625, "learning_rate": 1.8993415350188006e-05, "loss": 1.4828, "step": 6789 }, { "epoch": 16.975, "grad_norm": 8.0, "learning_rate": 1.899306128682264e-05, "loss": 1.3622, "step": 6790 }, { "epoch": 16.9775, "grad_norm": 5.59375, "learning_rate": 1.899270716449913e-05, "loss": 1.3567, "step": 6791 }, { "epoch": 16.98, "grad_norm": 6.3125, "learning_rate": 1.8992352983219785e-05, "loss": 1.4672, "step": 6792 }, { "epoch": 16.9825, "grad_norm": 6.28125, "learning_rate": 1.899199874298694e-05, "loss": 1.4255, "step": 6793 }, { "epoch": 16.985, "grad_norm": 7.34375, "learning_rate": 1.8991644443802906e-05, "loss": 1.3607, "step": 6794 }, { "epoch": 16.9875, "grad_norm": 8.5625, "learning_rate": 1.899129008567001e-05, "loss": 1.359, "step": 6795 }, { "epoch": 16.99, "grad_norm": 6.28125, "learning_rate": 1.8990935668590583e-05, "loss": 1.4365, "step": 6796 }, { "epoch": 16.9925, "grad_norm": 7.0, "learning_rate": 1.8990581192566936e-05, "loss": 1.2896, "step": 6797 }, { "epoch": 16.995, "grad_norm": 6.9375, "learning_rate": 1.89902266576014e-05, "loss": 1.5792, "step": 6798 }, { "epoch": 16.9975, "grad_norm": 5.1875, "learning_rate": 1.89898720636963e-05, "loss": 1.3014, "step": 6799 }, { "epoch": 17.0, "grad_norm": 7.75, "learning_rate": 1.8989517410853956e-05, "loss": 1.5444, "step": 6800 }, { "epoch": 17.0025, "grad_norm": 8.9375, "learning_rate": 1.8989162699076695e-05, "loss": 1.3565, "step": 6801 }, { "epoch": 17.005, "grad_norm": 5.59375, "learning_rate": 1.8988807928366843e-05, "loss": 1.4349, "step": 6802 }, { "epoch": 17.0075, "grad_norm": 8.375, "learning_rate": 1.8988453098726726e-05, "loss": 1.517, "step": 6803 }, { "epoch": 17.01, "grad_norm": 8.0625, "learning_rate": 1.8988098210158673e-05, "loss": 1.7391, "step": 6804 }, { "epoch": 17.0125, "grad_norm": 9.125, "learning_rate": 1.8987743262665007e-05, "loss": 1.2287, "step": 6805 }, { "epoch": 17.015, "grad_norm": 7.0625, "learning_rate": 1.8987388256248054e-05, "loss": 1.5509, "step": 6806 }, { "epoch": 17.0175, "grad_norm": 5.90625, "learning_rate": 1.8987033190910144e-05, "loss": 1.3177, "step": 6807 }, { "epoch": 17.02, "grad_norm": 8.9375, "learning_rate": 1.8986678066653604e-05, "loss": 1.5559, "step": 6808 }, { "epoch": 17.0225, "grad_norm": 4.5, "learning_rate": 1.898632288348076e-05, "loss": 1.2983, "step": 6809 }, { "epoch": 17.025, "grad_norm": 9.0625, "learning_rate": 1.8985967641393945e-05, "loss": 1.4578, "step": 6810 }, { "epoch": 17.0275, "grad_norm": 6.25, "learning_rate": 1.8985612340395487e-05, "loss": 1.4085, "step": 6811 }, { "epoch": 17.03, "grad_norm": 7.71875, "learning_rate": 1.8985256980487713e-05, "loss": 1.386, "step": 6812 }, { "epoch": 17.0325, "grad_norm": 5.40625, "learning_rate": 1.8984901561672956e-05, "loss": 1.3966, "step": 6813 }, { "epoch": 17.035, "grad_norm": 7.03125, "learning_rate": 1.898454608395354e-05, "loss": 1.3808, "step": 6814 }, { "epoch": 17.0375, "grad_norm": 5.375, "learning_rate": 1.89841905473318e-05, "loss": 1.434, "step": 6815 }, { "epoch": 17.04, "grad_norm": 6.6875, "learning_rate": 1.8983834951810068e-05, "loss": 1.5742, "step": 6816 }, { "epoch": 17.0425, "grad_norm": 11.375, "learning_rate": 1.8983479297390674e-05, "loss": 1.3248, "step": 6817 }, { "epoch": 17.045, "grad_norm": 6.09375, "learning_rate": 1.8983123584075947e-05, "loss": 1.499, "step": 6818 }, { "epoch": 17.0475, "grad_norm": 6.1875, "learning_rate": 1.8982767811868222e-05, "loss": 1.3973, "step": 6819 }, { "epoch": 17.05, "grad_norm": 8.4375, "learning_rate": 1.898241198076983e-05, "loss": 1.4141, "step": 6820 }, { "epoch": 17.0525, "grad_norm": 6.4375, "learning_rate": 1.8982056090783106e-05, "loss": 1.3555, "step": 6821 }, { "epoch": 17.055, "grad_norm": 7.84375, "learning_rate": 1.898170014191038e-05, "loss": 1.3763, "step": 6822 }, { "epoch": 17.0575, "grad_norm": 7.4375, "learning_rate": 1.898134413415399e-05, "loss": 1.3046, "step": 6823 }, { "epoch": 17.06, "grad_norm": 6.25, "learning_rate": 1.8980988067516265e-05, "loss": 1.3221, "step": 6824 }, { "epoch": 17.0625, "grad_norm": 10.4375, "learning_rate": 1.8980631941999544e-05, "loss": 1.566, "step": 6825 }, { "epoch": 17.065, "grad_norm": 9.375, "learning_rate": 1.8980275757606157e-05, "loss": 1.5129, "step": 6826 }, { "epoch": 17.0675, "grad_norm": 5.9375, "learning_rate": 1.8979919514338443e-05, "loss": 1.4758, "step": 6827 }, { "epoch": 17.07, "grad_norm": 5.90625, "learning_rate": 1.8979563212198737e-05, "loss": 1.4059, "step": 6828 }, { "epoch": 17.0725, "grad_norm": 8.4375, "learning_rate": 1.897920685118937e-05, "loss": 1.4973, "step": 6829 }, { "epoch": 17.075, "grad_norm": 8.0625, "learning_rate": 1.8978850431312686e-05, "loss": 1.4234, "step": 6830 }, { "epoch": 17.0775, "grad_norm": 5.625, "learning_rate": 1.897849395257102e-05, "loss": 1.5533, "step": 6831 }, { "epoch": 17.08, "grad_norm": 7.21875, "learning_rate": 1.89781374149667e-05, "loss": 1.3931, "step": 6832 }, { "epoch": 17.0825, "grad_norm": 6.28125, "learning_rate": 1.8977780818502074e-05, "loss": 1.3796, "step": 6833 }, { "epoch": 17.085, "grad_norm": 5.4375, "learning_rate": 1.8977424163179476e-05, "loss": 1.3753, "step": 6834 }, { "epoch": 17.0875, "grad_norm": 4.96875, "learning_rate": 1.8977067449001243e-05, "loss": 1.5227, "step": 6835 }, { "epoch": 17.09, "grad_norm": 6.59375, "learning_rate": 1.8976710675969715e-05, "loss": 1.4095, "step": 6836 }, { "epoch": 17.0925, "grad_norm": 6.5, "learning_rate": 1.8976353844087233e-05, "loss": 1.4488, "step": 6837 }, { "epoch": 17.095, "grad_norm": 4.46875, "learning_rate": 1.8975996953356136e-05, "loss": 1.2662, "step": 6838 }, { "epoch": 17.0975, "grad_norm": 6.65625, "learning_rate": 1.897564000377876e-05, "loss": 1.4847, "step": 6839 }, { "epoch": 17.1, "grad_norm": 6.25, "learning_rate": 1.8975282995357448e-05, "loss": 1.4662, "step": 6840 }, { "epoch": 17.1025, "grad_norm": 4.84375, "learning_rate": 1.8974925928094538e-05, "loss": 1.3902, "step": 6841 }, { "epoch": 17.105, "grad_norm": 5.5, "learning_rate": 1.8974568801992372e-05, "loss": 1.3079, "step": 6842 }, { "epoch": 17.1075, "grad_norm": 5.53125, "learning_rate": 1.8974211617053296e-05, "loss": 1.3442, "step": 6843 }, { "epoch": 17.11, "grad_norm": 5.875, "learning_rate": 1.8973854373279646e-05, "loss": 1.3701, "step": 6844 }, { "epoch": 17.1125, "grad_norm": 6.28125, "learning_rate": 1.8973497070673765e-05, "loss": 1.5053, "step": 6845 }, { "epoch": 17.115, "grad_norm": 8.375, "learning_rate": 1.8973139709237996e-05, "loss": 1.4384, "step": 6846 }, { "epoch": 17.1175, "grad_norm": 6.1875, "learning_rate": 1.8972782288974685e-05, "loss": 1.3131, "step": 6847 }, { "epoch": 17.12, "grad_norm": 9.3125, "learning_rate": 1.897242480988617e-05, "loss": 1.5444, "step": 6848 }, { "epoch": 17.1225, "grad_norm": 4.34375, "learning_rate": 1.89720672719748e-05, "loss": 1.4756, "step": 6849 }, { "epoch": 17.125, "grad_norm": 6.375, "learning_rate": 1.897170967524291e-05, "loss": 1.4453, "step": 6850 }, { "epoch": 17.1275, "grad_norm": 6.15625, "learning_rate": 1.8971352019692853e-05, "loss": 1.4608, "step": 6851 }, { "epoch": 17.13, "grad_norm": 8.375, "learning_rate": 1.8970994305326973e-05, "loss": 1.4745, "step": 6852 }, { "epoch": 17.1325, "grad_norm": 11.25, "learning_rate": 1.897063653214761e-05, "loss": 1.2996, "step": 6853 }, { "epoch": 17.135, "grad_norm": 9.75, "learning_rate": 1.897027870015712e-05, "loss": 1.5142, "step": 6854 }, { "epoch": 17.1375, "grad_norm": 7.6875, "learning_rate": 1.8969920809357837e-05, "loss": 1.4651, "step": 6855 }, { "epoch": 17.14, "grad_norm": 6.40625, "learning_rate": 1.896956285975211e-05, "loss": 1.3967, "step": 6856 }, { "epoch": 17.1425, "grad_norm": 7.0, "learning_rate": 1.896920485134229e-05, "loss": 1.5475, "step": 6857 }, { "epoch": 17.145, "grad_norm": 6.09375, "learning_rate": 1.8968846784130725e-05, "loss": 1.335, "step": 6858 }, { "epoch": 17.1475, "grad_norm": 5.65625, "learning_rate": 1.8968488658119755e-05, "loss": 1.4022, "step": 6859 }, { "epoch": 17.15, "grad_norm": 6.21875, "learning_rate": 1.8968130473311732e-05, "loss": 1.3154, "step": 6860 }, { "epoch": 17.1525, "grad_norm": 6.1875, "learning_rate": 1.8967772229709008e-05, "loss": 1.5229, "step": 6861 }, { "epoch": 17.155, "grad_norm": 6.21875, "learning_rate": 1.8967413927313923e-05, "loss": 1.3347, "step": 6862 }, { "epoch": 17.1575, "grad_norm": 5.59375, "learning_rate": 1.896705556612884e-05, "loss": 1.3795, "step": 6863 }, { "epoch": 17.16, "grad_norm": 6.34375, "learning_rate": 1.8966697146156092e-05, "loss": 1.3841, "step": 6864 }, { "epoch": 17.1625, "grad_norm": 5.84375, "learning_rate": 1.8966338667398037e-05, "loss": 1.3253, "step": 6865 }, { "epoch": 17.165, "grad_norm": 7.65625, "learning_rate": 1.896598012985703e-05, "loss": 1.5056, "step": 6866 }, { "epoch": 17.1675, "grad_norm": 6.78125, "learning_rate": 1.896562153353541e-05, "loss": 1.3051, "step": 6867 }, { "epoch": 17.17, "grad_norm": 6.90625, "learning_rate": 1.8965262878435534e-05, "loss": 1.4725, "step": 6868 }, { "epoch": 17.1725, "grad_norm": 7.40625, "learning_rate": 1.8964904164559758e-05, "loss": 1.291, "step": 6869 }, { "epoch": 17.175, "grad_norm": 7.5625, "learning_rate": 1.8964545391910426e-05, "loss": 1.4013, "step": 6870 }, { "epoch": 17.1775, "grad_norm": 6.9375, "learning_rate": 1.8964186560489893e-05, "loss": 1.443, "step": 6871 }, { "epoch": 17.18, "grad_norm": 10.5, "learning_rate": 1.8963827670300512e-05, "loss": 1.5589, "step": 6872 }, { "epoch": 17.1825, "grad_norm": 8.375, "learning_rate": 1.8963468721344633e-05, "loss": 1.4241, "step": 6873 }, { "epoch": 17.185, "grad_norm": 6.59375, "learning_rate": 1.8963109713624614e-05, "loss": 1.4717, "step": 6874 }, { "epoch": 17.1875, "grad_norm": 6.78125, "learning_rate": 1.8962750647142808e-05, "loss": 1.2947, "step": 6875 }, { "epoch": 17.19, "grad_norm": 10.5625, "learning_rate": 1.8962391521901565e-05, "loss": 1.5041, "step": 6876 }, { "epoch": 17.1925, "grad_norm": 8.625, "learning_rate": 1.8962032337903245e-05, "loss": 1.4901, "step": 6877 }, { "epoch": 17.195, "grad_norm": 5.78125, "learning_rate": 1.8961673095150197e-05, "loss": 1.4705, "step": 6878 }, { "epoch": 17.1975, "grad_norm": 6.65625, "learning_rate": 1.896131379364478e-05, "loss": 1.3787, "step": 6879 }, { "epoch": 17.2, "grad_norm": 7.96875, "learning_rate": 1.896095443338935e-05, "loss": 1.4266, "step": 6880 }, { "epoch": 17.2025, "grad_norm": 6.28125, "learning_rate": 1.8960595014386257e-05, "loss": 1.2463, "step": 6881 }, { "epoch": 17.205, "grad_norm": 6.53125, "learning_rate": 1.8960235536637864e-05, "loss": 1.4843, "step": 6882 }, { "epoch": 17.2075, "grad_norm": 5.15625, "learning_rate": 1.8959876000146525e-05, "loss": 1.3264, "step": 6883 }, { "epoch": 17.21, "grad_norm": 7.09375, "learning_rate": 1.8959516404914597e-05, "loss": 1.3308, "step": 6884 }, { "epoch": 17.2125, "grad_norm": 6.84375, "learning_rate": 1.895915675094444e-05, "loss": 1.3765, "step": 6885 }, { "epoch": 17.215, "grad_norm": 4.25, "learning_rate": 1.895879703823841e-05, "loss": 1.4859, "step": 6886 }, { "epoch": 17.2175, "grad_norm": 7.21875, "learning_rate": 1.895843726679886e-05, "loss": 1.4142, "step": 6887 }, { "epoch": 17.22, "grad_norm": 11.3125, "learning_rate": 1.895807743662816e-05, "loss": 1.4259, "step": 6888 }, { "epoch": 17.2225, "grad_norm": 5.0625, "learning_rate": 1.895771754772866e-05, "loss": 1.3045, "step": 6889 }, { "epoch": 17.225, "grad_norm": 5.875, "learning_rate": 1.8957357600102723e-05, "loss": 1.4447, "step": 6890 }, { "epoch": 17.2275, "grad_norm": 7.71875, "learning_rate": 1.8956997593752705e-05, "loss": 1.3758, "step": 6891 }, { "epoch": 17.23, "grad_norm": 8.6875, "learning_rate": 1.8956637528680973e-05, "loss": 1.4421, "step": 6892 }, { "epoch": 17.2325, "grad_norm": 5.6875, "learning_rate": 1.895627740488988e-05, "loss": 1.4867, "step": 6893 }, { "epoch": 17.235, "grad_norm": 7.125, "learning_rate": 1.8955917222381798e-05, "loss": 1.3758, "step": 6894 }, { "epoch": 17.2375, "grad_norm": 5.5, "learning_rate": 1.8955556981159074e-05, "loss": 1.3442, "step": 6895 }, { "epoch": 17.24, "grad_norm": 5.28125, "learning_rate": 1.895519668122408e-05, "loss": 1.5778, "step": 6896 }, { "epoch": 17.2425, "grad_norm": 5.28125, "learning_rate": 1.8954836322579174e-05, "loss": 1.3128, "step": 6897 }, { "epoch": 17.245, "grad_norm": 8.75, "learning_rate": 1.895447590522672e-05, "loss": 1.5473, "step": 6898 }, { "epoch": 17.2475, "grad_norm": 5.84375, "learning_rate": 1.895411542916908e-05, "loss": 1.3114, "step": 6899 }, { "epoch": 17.25, "grad_norm": 7.875, "learning_rate": 1.8953754894408617e-05, "loss": 1.2715, "step": 6900 }, { "epoch": 17.2525, "grad_norm": 5.53125, "learning_rate": 1.8953394300947697e-05, "loss": 1.468, "step": 6901 }, { "epoch": 17.255, "grad_norm": 7.96875, "learning_rate": 1.895303364878868e-05, "loss": 1.458, "step": 6902 }, { "epoch": 17.2575, "grad_norm": 7.4375, "learning_rate": 1.8952672937933934e-05, "loss": 1.3358, "step": 6903 }, { "epoch": 17.26, "grad_norm": 9.0625, "learning_rate": 1.895231216838582e-05, "loss": 1.4321, "step": 6904 }, { "epoch": 17.2625, "grad_norm": 7.34375, "learning_rate": 1.8951951340146707e-05, "loss": 1.4779, "step": 6905 }, { "epoch": 17.265, "grad_norm": 6.3125, "learning_rate": 1.895159045321896e-05, "loss": 1.549, "step": 6906 }, { "epoch": 17.2675, "grad_norm": 7.875, "learning_rate": 1.8951229507604946e-05, "loss": 1.6005, "step": 6907 }, { "epoch": 17.27, "grad_norm": 8.9375, "learning_rate": 1.8950868503307026e-05, "loss": 1.4539, "step": 6908 }, { "epoch": 17.2725, "grad_norm": 6.9375, "learning_rate": 1.8950507440327574e-05, "loss": 1.363, "step": 6909 }, { "epoch": 17.275, "grad_norm": 7.15625, "learning_rate": 1.8950146318668952e-05, "loss": 1.5653, "step": 6910 }, { "epoch": 17.2775, "grad_norm": 7.15625, "learning_rate": 1.8949785138333526e-05, "loss": 1.4123, "step": 6911 }, { "epoch": 17.28, "grad_norm": 6.03125, "learning_rate": 1.894942389932367e-05, "loss": 1.2696, "step": 6912 }, { "epoch": 17.2825, "grad_norm": 5.09375, "learning_rate": 1.894906260164175e-05, "loss": 1.4687, "step": 6913 }, { "epoch": 17.285, "grad_norm": 8.25, "learning_rate": 1.894870124529013e-05, "loss": 1.5873, "step": 6914 }, { "epoch": 17.2875, "grad_norm": 6.03125, "learning_rate": 1.8948339830271185e-05, "loss": 1.343, "step": 6915 }, { "epoch": 17.29, "grad_norm": 6.5, "learning_rate": 1.8947978356587283e-05, "loss": 1.5204, "step": 6916 }, { "epoch": 17.2925, "grad_norm": 8.4375, "learning_rate": 1.894761682424079e-05, "loss": 1.4805, "step": 6917 }, { "epoch": 17.295, "grad_norm": 6.0, "learning_rate": 1.894725523323408e-05, "loss": 1.4899, "step": 6918 }, { "epoch": 17.2975, "grad_norm": 5.65625, "learning_rate": 1.8946893583569526e-05, "loss": 1.5507, "step": 6919 }, { "epoch": 17.3, "grad_norm": 6.21875, "learning_rate": 1.8946531875249496e-05, "loss": 1.1597, "step": 6920 }, { "epoch": 17.3025, "grad_norm": 7.625, "learning_rate": 1.8946170108276357e-05, "loss": 1.4454, "step": 6921 }, { "epoch": 17.305, "grad_norm": 8.1875, "learning_rate": 1.8945808282652485e-05, "loss": 1.4521, "step": 6922 }, { "epoch": 17.3075, "grad_norm": 9.875, "learning_rate": 1.894544639838025e-05, "loss": 1.4624, "step": 6923 }, { "epoch": 17.31, "grad_norm": 6.125, "learning_rate": 1.8945084455462032e-05, "loss": 1.4774, "step": 6924 }, { "epoch": 17.3125, "grad_norm": 7.71875, "learning_rate": 1.8944722453900195e-05, "loss": 1.4243, "step": 6925 }, { "epoch": 17.315, "grad_norm": 5.0625, "learning_rate": 1.8944360393697117e-05, "loss": 1.3095, "step": 6926 }, { "epoch": 17.3175, "grad_norm": 5.90625, "learning_rate": 1.8943998274855173e-05, "loss": 1.3158, "step": 6927 }, { "epoch": 17.32, "grad_norm": 9.25, "learning_rate": 1.8943636097376728e-05, "loss": 1.5355, "step": 6928 }, { "epoch": 17.3225, "grad_norm": 7.71875, "learning_rate": 1.8943273861264163e-05, "loss": 1.6419, "step": 6929 }, { "epoch": 17.325, "grad_norm": 7.5, "learning_rate": 1.8942911566519858e-05, "loss": 1.5257, "step": 6930 }, { "epoch": 17.3275, "grad_norm": 5.1875, "learning_rate": 1.8942549213146176e-05, "loss": 1.4545, "step": 6931 }, { "epoch": 17.33, "grad_norm": 9.875, "learning_rate": 1.8942186801145505e-05, "loss": 1.4361, "step": 6932 }, { "epoch": 17.3325, "grad_norm": 6.71875, "learning_rate": 1.8941824330520212e-05, "loss": 1.4859, "step": 6933 }, { "epoch": 17.335, "grad_norm": 5.53125, "learning_rate": 1.8941461801272678e-05, "loss": 1.3293, "step": 6934 }, { "epoch": 17.3375, "grad_norm": 7.1875, "learning_rate": 1.894109921340528e-05, "loss": 1.4961, "step": 6935 }, { "epoch": 17.34, "grad_norm": 6.125, "learning_rate": 1.8940736566920386e-05, "loss": 1.4908, "step": 6936 }, { "epoch": 17.3425, "grad_norm": 5.90625, "learning_rate": 1.894037386182039e-05, "loss": 1.5882, "step": 6937 }, { "epoch": 17.345, "grad_norm": 6.125, "learning_rate": 1.8940011098107654e-05, "loss": 1.3097, "step": 6938 }, { "epoch": 17.3475, "grad_norm": 5.1875, "learning_rate": 1.8939648275784565e-05, "loss": 1.3104, "step": 6939 }, { "epoch": 17.35, "grad_norm": 6.375, "learning_rate": 1.89392853948535e-05, "loss": 1.5134, "step": 6940 }, { "epoch": 17.3525, "grad_norm": 4.75, "learning_rate": 1.893892245531684e-05, "loss": 1.4281, "step": 6941 }, { "epoch": 17.355, "grad_norm": 4.8125, "learning_rate": 1.893855945717696e-05, "loss": 1.3554, "step": 6942 }, { "epoch": 17.3575, "grad_norm": 6.28125, "learning_rate": 1.8938196400436242e-05, "loss": 1.4385, "step": 6943 }, { "epoch": 17.36, "grad_norm": 5.6875, "learning_rate": 1.8937833285097067e-05, "loss": 1.3665, "step": 6944 }, { "epoch": 17.3625, "grad_norm": 6.84375, "learning_rate": 1.8937470111161813e-05, "loss": 1.5121, "step": 6945 }, { "epoch": 17.365, "grad_norm": 6.40625, "learning_rate": 1.893710687863286e-05, "loss": 1.5727, "step": 6946 }, { "epoch": 17.3675, "grad_norm": 8.0, "learning_rate": 1.8936743587512598e-05, "loss": 1.5144, "step": 6947 }, { "epoch": 17.37, "grad_norm": 6.84375, "learning_rate": 1.8936380237803398e-05, "loss": 1.4452, "step": 6948 }, { "epoch": 17.3725, "grad_norm": 7.15625, "learning_rate": 1.893601682950765e-05, "loss": 1.5303, "step": 6949 }, { "epoch": 17.375, "grad_norm": 8.0625, "learning_rate": 1.893565336262773e-05, "loss": 1.4677, "step": 6950 }, { "epoch": 17.3775, "grad_norm": 9.125, "learning_rate": 1.8935289837166028e-05, "loss": 1.4143, "step": 6951 }, { "epoch": 17.38, "grad_norm": 5.78125, "learning_rate": 1.8934926253124922e-05, "loss": 1.488, "step": 6952 }, { "epoch": 17.3825, "grad_norm": 6.71875, "learning_rate": 1.8934562610506796e-05, "loss": 1.3237, "step": 6953 }, { "epoch": 17.385, "grad_norm": 7.03125, "learning_rate": 1.8934198909314037e-05, "loss": 1.4469, "step": 6954 }, { "epoch": 17.3875, "grad_norm": 8.375, "learning_rate": 1.8933835149549024e-05, "loss": 1.5763, "step": 6955 }, { "epoch": 17.39, "grad_norm": 5.90625, "learning_rate": 1.893347133121415e-05, "loss": 1.3796, "step": 6956 }, { "epoch": 17.3925, "grad_norm": 6.1875, "learning_rate": 1.893310745431179e-05, "loss": 1.3628, "step": 6957 }, { "epoch": 17.395, "grad_norm": 5.0625, "learning_rate": 1.8932743518844337e-05, "loss": 1.4344, "step": 6958 }, { "epoch": 17.3975, "grad_norm": 7.25, "learning_rate": 1.8932379524814175e-05, "loss": 1.4012, "step": 6959 }, { "epoch": 17.4, "grad_norm": 5.90625, "learning_rate": 1.8932015472223692e-05, "loss": 1.3777, "step": 6960 }, { "epoch": 17.4025, "grad_norm": 7.71875, "learning_rate": 1.8931651361075275e-05, "loss": 1.5028, "step": 6961 }, { "epoch": 17.405, "grad_norm": 5.0625, "learning_rate": 1.8931287191371303e-05, "loss": 1.4502, "step": 6962 }, { "epoch": 17.4075, "grad_norm": 10.125, "learning_rate": 1.8930922963114173e-05, "loss": 1.4046, "step": 6963 }, { "epoch": 17.41, "grad_norm": 5.875, "learning_rate": 1.893055867630627e-05, "loss": 1.3351, "step": 6964 }, { "epoch": 17.4125, "grad_norm": 9.4375, "learning_rate": 1.893019433094998e-05, "loss": 1.447, "step": 6965 }, { "epoch": 17.415, "grad_norm": 7.9375, "learning_rate": 1.8929829927047697e-05, "loss": 1.4404, "step": 6966 }, { "epoch": 17.4175, "grad_norm": 5.09375, "learning_rate": 1.8929465464601803e-05, "loss": 1.4224, "step": 6967 }, { "epoch": 17.42, "grad_norm": 4.9375, "learning_rate": 1.892910094361469e-05, "loss": 1.4623, "step": 6968 }, { "epoch": 17.4225, "grad_norm": 5.5625, "learning_rate": 1.892873636408875e-05, "loss": 1.4666, "step": 6969 }, { "epoch": 17.425, "grad_norm": 5.21875, "learning_rate": 1.892837172602637e-05, "loss": 1.3071, "step": 6970 }, { "epoch": 17.4275, "grad_norm": 8.9375, "learning_rate": 1.8928007029429944e-05, "loss": 1.489, "step": 6971 }, { "epoch": 17.43, "grad_norm": 6.3125, "learning_rate": 1.8927642274301862e-05, "loss": 1.3938, "step": 6972 }, { "epoch": 17.4325, "grad_norm": 7.5625, "learning_rate": 1.8927277460644514e-05, "loss": 1.5223, "step": 6973 }, { "epoch": 17.435, "grad_norm": 7.84375, "learning_rate": 1.892691258846029e-05, "loss": 1.5749, "step": 6974 }, { "epoch": 17.4375, "grad_norm": 5.6875, "learning_rate": 1.8926547657751585e-05, "loss": 1.4099, "step": 6975 }, { "epoch": 17.44, "grad_norm": 5.6875, "learning_rate": 1.8926182668520794e-05, "loss": 1.6292, "step": 6976 }, { "epoch": 17.4425, "grad_norm": 5.46875, "learning_rate": 1.8925817620770302e-05, "loss": 1.4195, "step": 6977 }, { "epoch": 17.445, "grad_norm": 5.375, "learning_rate": 1.8925452514502512e-05, "loss": 1.3852, "step": 6978 }, { "epoch": 17.4475, "grad_norm": 6.5625, "learning_rate": 1.892508734971981e-05, "loss": 1.6285, "step": 6979 }, { "epoch": 17.45, "grad_norm": 6.21875, "learning_rate": 1.892472212642459e-05, "loss": 1.3814, "step": 6980 }, { "epoch": 17.4525, "grad_norm": 9.75, "learning_rate": 1.8924356844619253e-05, "loss": 1.3247, "step": 6981 }, { "epoch": 17.455, "grad_norm": 5.9375, "learning_rate": 1.8923991504306187e-05, "loss": 1.3814, "step": 6982 }, { "epoch": 17.4575, "grad_norm": 6.875, "learning_rate": 1.892362610548779e-05, "loss": 1.3628, "step": 6983 }, { "epoch": 17.46, "grad_norm": 8.3125, "learning_rate": 1.8923260648166457e-05, "loss": 1.2635, "step": 6984 }, { "epoch": 17.4625, "grad_norm": 8.4375, "learning_rate": 1.8922895132344586e-05, "loss": 1.4718, "step": 6985 }, { "epoch": 17.465, "grad_norm": 7.9375, "learning_rate": 1.892252955802457e-05, "loss": 1.4595, "step": 6986 }, { "epoch": 17.4675, "grad_norm": 6.4375, "learning_rate": 1.8922163925208808e-05, "loss": 1.5095, "step": 6987 }, { "epoch": 17.47, "grad_norm": 6.40625, "learning_rate": 1.89217982338997e-05, "loss": 1.3008, "step": 6988 }, { "epoch": 17.4725, "grad_norm": 9.0, "learning_rate": 1.8921432484099633e-05, "loss": 1.5441, "step": 6989 }, { "epoch": 17.475, "grad_norm": 6.4375, "learning_rate": 1.8921066675811018e-05, "loss": 1.4959, "step": 6990 }, { "epoch": 17.4775, "grad_norm": 6.78125, "learning_rate": 1.8920700809036242e-05, "loss": 1.4288, "step": 6991 }, { "epoch": 17.48, "grad_norm": 6.25, "learning_rate": 1.892033488377771e-05, "loss": 1.3535, "step": 6992 }, { "epoch": 17.4825, "grad_norm": 8.625, "learning_rate": 1.891996890003782e-05, "loss": 1.581, "step": 6993 }, { "epoch": 17.485, "grad_norm": 7.78125, "learning_rate": 1.891960285781897e-05, "loss": 1.3578, "step": 6994 }, { "epoch": 17.4875, "grad_norm": 5.625, "learning_rate": 1.8919236757123563e-05, "loss": 1.4336, "step": 6995 }, { "epoch": 17.49, "grad_norm": 8.875, "learning_rate": 1.8918870597953996e-05, "loss": 1.4674, "step": 6996 }, { "epoch": 17.4925, "grad_norm": 5.9375, "learning_rate": 1.8918504380312667e-05, "loss": 1.4475, "step": 6997 }, { "epoch": 17.495, "grad_norm": 6.0625, "learning_rate": 1.891813810420198e-05, "loss": 1.3922, "step": 6998 }, { "epoch": 17.4975, "grad_norm": 8.1875, "learning_rate": 1.8917771769624342e-05, "loss": 1.4692, "step": 6999 }, { "epoch": 17.5, "grad_norm": 5.90625, "learning_rate": 1.8917405376582144e-05, "loss": 1.5025, "step": 7000 }, { "epoch": 17.5025, "grad_norm": 6.03125, "learning_rate": 1.8917038925077797e-05, "loss": 1.6136, "step": 7001 }, { "epoch": 17.505, "grad_norm": 5.78125, "learning_rate": 1.8916672415113698e-05, "loss": 1.4459, "step": 7002 }, { "epoch": 17.5075, "grad_norm": 4.375, "learning_rate": 1.891630584669225e-05, "loss": 1.4537, "step": 7003 }, { "epoch": 17.51, "grad_norm": 6.34375, "learning_rate": 1.891593921981586e-05, "loss": 1.5639, "step": 7004 }, { "epoch": 17.5125, "grad_norm": 5.25, "learning_rate": 1.891557253448693e-05, "loss": 1.4242, "step": 7005 }, { "epoch": 17.515, "grad_norm": 4.75, "learning_rate": 1.891520579070786e-05, "loss": 1.3517, "step": 7006 }, { "epoch": 17.5175, "grad_norm": 7.09375, "learning_rate": 1.891483898848106e-05, "loss": 1.3562, "step": 7007 }, { "epoch": 17.52, "grad_norm": 6.375, "learning_rate": 1.891447212780893e-05, "loss": 1.3369, "step": 7008 }, { "epoch": 17.5225, "grad_norm": 8.1875, "learning_rate": 1.891410520869388e-05, "loss": 1.4845, "step": 7009 }, { "epoch": 17.525, "grad_norm": 5.625, "learning_rate": 1.8913738231138312e-05, "loss": 1.3942, "step": 7010 }, { "epoch": 17.5275, "grad_norm": 7.375, "learning_rate": 1.8913371195144632e-05, "loss": 1.3933, "step": 7011 }, { "epoch": 17.53, "grad_norm": 9.4375, "learning_rate": 1.8913004100715247e-05, "loss": 1.4282, "step": 7012 }, { "epoch": 17.5325, "grad_norm": 8.3125, "learning_rate": 1.8912636947852566e-05, "loss": 1.5883, "step": 7013 }, { "epoch": 17.535, "grad_norm": 6.125, "learning_rate": 1.8912269736558994e-05, "loss": 1.4379, "step": 7014 }, { "epoch": 17.5375, "grad_norm": 5.125, "learning_rate": 1.8911902466836937e-05, "loss": 1.4478, "step": 7015 }, { "epoch": 17.54, "grad_norm": 6.09375, "learning_rate": 1.89115351386888e-05, "loss": 1.381, "step": 7016 }, { "epoch": 17.5425, "grad_norm": 5.78125, "learning_rate": 1.8911167752117003e-05, "loss": 1.3835, "step": 7017 }, { "epoch": 17.545, "grad_norm": 4.9375, "learning_rate": 1.8910800307123944e-05, "loss": 1.443, "step": 7018 }, { "epoch": 17.5475, "grad_norm": 5.96875, "learning_rate": 1.8910432803712032e-05, "loss": 1.4351, "step": 7019 }, { "epoch": 17.55, "grad_norm": 5.125, "learning_rate": 1.891006524188368e-05, "loss": 1.4937, "step": 7020 }, { "epoch": 17.5525, "grad_norm": 7.90625, "learning_rate": 1.89096976216413e-05, "loss": 1.3601, "step": 7021 }, { "epoch": 17.555, "grad_norm": 5.5, "learning_rate": 1.8909329942987297e-05, "loss": 1.3786, "step": 7022 }, { "epoch": 17.5575, "grad_norm": 7.53125, "learning_rate": 1.890896220592408e-05, "loss": 1.4173, "step": 7023 }, { "epoch": 17.56, "grad_norm": 11.75, "learning_rate": 1.8908594410454068e-05, "loss": 1.3542, "step": 7024 }, { "epoch": 17.5625, "grad_norm": 5.34375, "learning_rate": 1.8908226556579665e-05, "loss": 1.404, "step": 7025 }, { "epoch": 17.565, "grad_norm": 8.625, "learning_rate": 1.8907858644303286e-05, "loss": 1.5386, "step": 7026 }, { "epoch": 17.5675, "grad_norm": 8.1875, "learning_rate": 1.8907490673627343e-05, "loss": 1.5012, "step": 7027 }, { "epoch": 17.57, "grad_norm": 6.0, "learning_rate": 1.8907122644554245e-05, "loss": 1.4334, "step": 7028 }, { "epoch": 17.5725, "grad_norm": 6.1875, "learning_rate": 1.890675455708641e-05, "loss": 1.3589, "step": 7029 }, { "epoch": 17.575, "grad_norm": 6.28125, "learning_rate": 1.890638641122625e-05, "loss": 1.5169, "step": 7030 }, { "epoch": 17.5775, "grad_norm": 6.4375, "learning_rate": 1.8906018206976174e-05, "loss": 1.3698, "step": 7031 }, { "epoch": 17.58, "grad_norm": 6.5, "learning_rate": 1.8905649944338596e-05, "loss": 1.4147, "step": 7032 }, { "epoch": 17.5825, "grad_norm": 7.5, "learning_rate": 1.8905281623315938e-05, "loss": 1.4798, "step": 7033 }, { "epoch": 17.585, "grad_norm": 5.90625, "learning_rate": 1.8904913243910613e-05, "loss": 1.3523, "step": 7034 }, { "epoch": 17.5875, "grad_norm": 6.84375, "learning_rate": 1.890454480612503e-05, "loss": 1.4448, "step": 7035 }, { "epoch": 17.59, "grad_norm": 6.3125, "learning_rate": 1.8904176309961606e-05, "loss": 1.352, "step": 7036 }, { "epoch": 17.5925, "grad_norm": 5.8125, "learning_rate": 1.890380775542276e-05, "loss": 1.329, "step": 7037 }, { "epoch": 17.595, "grad_norm": 7.96875, "learning_rate": 1.8903439142510907e-05, "loss": 1.502, "step": 7038 }, { "epoch": 17.5975, "grad_norm": 7.8125, "learning_rate": 1.890307047122846e-05, "loss": 1.573, "step": 7039 }, { "epoch": 17.6, "grad_norm": 5.15625, "learning_rate": 1.8902701741577844e-05, "loss": 1.4396, "step": 7040 }, { "epoch": 17.6025, "grad_norm": 8.0, "learning_rate": 1.8902332953561467e-05, "loss": 1.3433, "step": 7041 }, { "epoch": 17.605, "grad_norm": 5.875, "learning_rate": 1.8901964107181755e-05, "loss": 1.2808, "step": 7042 }, { "epoch": 17.6075, "grad_norm": 11.125, "learning_rate": 1.890159520244112e-05, "loss": 1.4581, "step": 7043 }, { "epoch": 17.61, "grad_norm": 6.15625, "learning_rate": 1.8901226239341986e-05, "loss": 1.3774, "step": 7044 }, { "epoch": 17.6125, "grad_norm": 6.375, "learning_rate": 1.8900857217886767e-05, "loss": 1.555, "step": 7045 }, { "epoch": 17.615, "grad_norm": 6.71875, "learning_rate": 1.8900488138077884e-05, "loss": 1.4566, "step": 7046 }, { "epoch": 17.6175, "grad_norm": 5.90625, "learning_rate": 1.8900118999917757e-05, "loss": 1.6833, "step": 7047 }, { "epoch": 17.62, "grad_norm": 5.71875, "learning_rate": 1.8899749803408808e-05, "loss": 1.375, "step": 7048 }, { "epoch": 17.6225, "grad_norm": 5.5, "learning_rate": 1.889938054855345e-05, "loss": 1.4145, "step": 7049 }, { "epoch": 17.625, "grad_norm": 6.84375, "learning_rate": 1.8899011235354118e-05, "loss": 1.3491, "step": 7050 }, { "epoch": 17.6275, "grad_norm": 5.9375, "learning_rate": 1.889864186381322e-05, "loss": 1.5161, "step": 7051 }, { "epoch": 17.63, "grad_norm": 7.625, "learning_rate": 1.889827243393318e-05, "loss": 1.4703, "step": 7052 }, { "epoch": 17.6325, "grad_norm": 5.25, "learning_rate": 1.8897902945716425e-05, "loss": 1.5068, "step": 7053 }, { "epoch": 17.635, "grad_norm": 6.65625, "learning_rate": 1.889753339916537e-05, "loss": 1.4116, "step": 7054 }, { "epoch": 17.6375, "grad_norm": 8.6875, "learning_rate": 1.8897163794282447e-05, "loss": 1.4321, "step": 7055 }, { "epoch": 17.64, "grad_norm": 6.75, "learning_rate": 1.8896794131070073e-05, "loss": 1.4989, "step": 7056 }, { "epoch": 17.6425, "grad_norm": 5.15625, "learning_rate": 1.8896424409530672e-05, "loss": 1.3328, "step": 7057 }, { "epoch": 17.645, "grad_norm": 7.09375, "learning_rate": 1.889605462966667e-05, "loss": 1.5176, "step": 7058 }, { "epoch": 17.6475, "grad_norm": 7.78125, "learning_rate": 1.8895684791480487e-05, "loss": 1.4256, "step": 7059 }, { "epoch": 17.65, "grad_norm": 4.75, "learning_rate": 1.889531489497455e-05, "loss": 1.3311, "step": 7060 }, { "epoch": 17.6525, "grad_norm": 13.75, "learning_rate": 1.889494494015129e-05, "loss": 1.4652, "step": 7061 }, { "epoch": 17.655, "grad_norm": 5.59375, "learning_rate": 1.8894574927013124e-05, "loss": 1.3242, "step": 7062 }, { "epoch": 17.6575, "grad_norm": 7.9375, "learning_rate": 1.8894204855562478e-05, "loss": 1.4929, "step": 7063 }, { "epoch": 17.66, "grad_norm": 6.34375, "learning_rate": 1.8893834725801782e-05, "loss": 1.4921, "step": 7064 }, { "epoch": 17.6625, "grad_norm": 7.1875, "learning_rate": 1.8893464537733462e-05, "loss": 1.6442, "step": 7065 }, { "epoch": 17.665, "grad_norm": 7.6875, "learning_rate": 1.8893094291359948e-05, "loss": 1.5718, "step": 7066 }, { "epoch": 17.6675, "grad_norm": 7.9375, "learning_rate": 1.889272398668366e-05, "loss": 1.4041, "step": 7067 }, { "epoch": 17.67, "grad_norm": 8.3125, "learning_rate": 1.8892353623707033e-05, "loss": 1.3767, "step": 7068 }, { "epoch": 17.6725, "grad_norm": 9.3125, "learning_rate": 1.8891983202432484e-05, "loss": 1.4289, "step": 7069 }, { "epoch": 17.675, "grad_norm": 6.6875, "learning_rate": 1.8891612722862457e-05, "loss": 1.4037, "step": 7070 }, { "epoch": 17.6775, "grad_norm": 6.78125, "learning_rate": 1.8891242184999366e-05, "loss": 1.2494, "step": 7071 }, { "epoch": 17.68, "grad_norm": 5.125, "learning_rate": 1.8890871588845653e-05, "loss": 1.4377, "step": 7072 }, { "epoch": 17.6825, "grad_norm": 6.8125, "learning_rate": 1.889050093440374e-05, "loss": 1.5647, "step": 7073 }, { "epoch": 17.685, "grad_norm": 7.25, "learning_rate": 1.8890130221676057e-05, "loss": 1.3084, "step": 7074 }, { "epoch": 17.6875, "grad_norm": 5.4375, "learning_rate": 1.8889759450665036e-05, "loss": 1.3284, "step": 7075 }, { "epoch": 17.69, "grad_norm": 9.5, "learning_rate": 1.8889388621373107e-05, "loss": 1.3959, "step": 7076 }, { "epoch": 17.6925, "grad_norm": 8.6875, "learning_rate": 1.8889017733802702e-05, "loss": 1.3657, "step": 7077 }, { "epoch": 17.695, "grad_norm": 7.9375, "learning_rate": 1.8888646787956255e-05, "loss": 1.5204, "step": 7078 }, { "epoch": 17.6975, "grad_norm": 4.59375, "learning_rate": 1.8888275783836197e-05, "loss": 1.3512, "step": 7079 }, { "epoch": 17.7, "grad_norm": 6.6875, "learning_rate": 1.8887904721444955e-05, "loss": 1.4463, "step": 7080 }, { "epoch": 17.7025, "grad_norm": 7.71875, "learning_rate": 1.8887533600784967e-05, "loss": 1.383, "step": 7081 }, { "epoch": 17.705, "grad_norm": 5.90625, "learning_rate": 1.888716242185866e-05, "loss": 1.4142, "step": 7082 }, { "epoch": 17.7075, "grad_norm": 5.75, "learning_rate": 1.888679118466848e-05, "loss": 1.2656, "step": 7083 }, { "epoch": 17.71, "grad_norm": 7.59375, "learning_rate": 1.8886419889216846e-05, "loss": 1.4355, "step": 7084 }, { "epoch": 17.7125, "grad_norm": 12.125, "learning_rate": 1.88860485355062e-05, "loss": 1.6211, "step": 7085 }, { "epoch": 17.715, "grad_norm": 5.59375, "learning_rate": 1.8885677123538978e-05, "loss": 1.5319, "step": 7086 }, { "epoch": 17.7175, "grad_norm": 6.6875, "learning_rate": 1.888530565331761e-05, "loss": 1.4063, "step": 7087 }, { "epoch": 17.72, "grad_norm": 6.5, "learning_rate": 1.8884934124844534e-05, "loss": 1.4116, "step": 7088 }, { "epoch": 17.7225, "grad_norm": 9.125, "learning_rate": 1.8884562538122188e-05, "loss": 1.47, "step": 7089 }, { "epoch": 17.725, "grad_norm": 8.25, "learning_rate": 1.8884190893153e-05, "loss": 1.3564, "step": 7090 }, { "epoch": 17.7275, "grad_norm": 7.09375, "learning_rate": 1.888381918993942e-05, "loss": 1.4304, "step": 7091 }, { "epoch": 17.73, "grad_norm": 5.25, "learning_rate": 1.888344742848387e-05, "loss": 1.2338, "step": 7092 }, { "epoch": 17.7325, "grad_norm": 7.5625, "learning_rate": 1.8883075608788797e-05, "loss": 1.5643, "step": 7093 }, { "epoch": 17.735, "grad_norm": 6.4375, "learning_rate": 1.8882703730856632e-05, "loss": 1.4389, "step": 7094 }, { "epoch": 17.7375, "grad_norm": 5.5625, "learning_rate": 1.888233179468982e-05, "loss": 1.3737, "step": 7095 }, { "epoch": 17.74, "grad_norm": 5.71875, "learning_rate": 1.88819598002908e-05, "loss": 1.3646, "step": 7096 }, { "epoch": 17.7425, "grad_norm": 7.1875, "learning_rate": 1.8881587747662e-05, "loss": 1.4232, "step": 7097 }, { "epoch": 17.745, "grad_norm": 7.03125, "learning_rate": 1.888121563680587e-05, "loss": 1.4944, "step": 7098 }, { "epoch": 17.7475, "grad_norm": 7.59375, "learning_rate": 1.8880843467724846e-05, "loss": 1.2953, "step": 7099 }, { "epoch": 17.75, "grad_norm": 7.84375, "learning_rate": 1.8880471240421365e-05, "loss": 1.4183, "step": 7100 }, { "epoch": 17.7525, "grad_norm": 7.4375, "learning_rate": 1.888009895489787e-05, "loss": 1.4446, "step": 7101 }, { "epoch": 17.755, "grad_norm": 6.5625, "learning_rate": 1.8879726611156805e-05, "loss": 1.3135, "step": 7102 }, { "epoch": 17.7575, "grad_norm": 5.8125, "learning_rate": 1.8879354209200606e-05, "loss": 1.5336, "step": 7103 }, { "epoch": 17.76, "grad_norm": 6.15625, "learning_rate": 1.8878981749031718e-05, "loss": 1.3915, "step": 7104 }, { "epoch": 17.7625, "grad_norm": 6.90625, "learning_rate": 1.8878609230652578e-05, "loss": 1.5589, "step": 7105 }, { "epoch": 17.765, "grad_norm": 8.125, "learning_rate": 1.8878236654065635e-05, "loss": 1.4689, "step": 7106 }, { "epoch": 17.7675, "grad_norm": 6.15625, "learning_rate": 1.8877864019273322e-05, "loss": 1.4835, "step": 7107 }, { "epoch": 17.77, "grad_norm": 7.21875, "learning_rate": 1.8877491326278093e-05, "loss": 1.2325, "step": 7108 }, { "epoch": 17.7725, "grad_norm": 6.65625, "learning_rate": 1.8877118575082383e-05, "loss": 1.3673, "step": 7109 }, { "epoch": 17.775, "grad_norm": 6.75, "learning_rate": 1.887674576568864e-05, "loss": 1.4003, "step": 7110 }, { "epoch": 17.7775, "grad_norm": 8.1875, "learning_rate": 1.8876372898099308e-05, "loss": 1.618, "step": 7111 }, { "epoch": 17.78, "grad_norm": 7.875, "learning_rate": 1.8875999972316826e-05, "loss": 1.3919, "step": 7112 }, { "epoch": 17.7825, "grad_norm": 6.6875, "learning_rate": 1.8875626988343647e-05, "loss": 1.5039, "step": 7113 }, { "epoch": 17.785, "grad_norm": 8.4375, "learning_rate": 1.887525394618221e-05, "loss": 1.5038, "step": 7114 }, { "epoch": 17.7875, "grad_norm": 6.09375, "learning_rate": 1.8874880845834968e-05, "loss": 1.4023, "step": 7115 }, { "epoch": 17.79, "grad_norm": 5.59375, "learning_rate": 1.887450768730436e-05, "loss": 1.4996, "step": 7116 }, { "epoch": 17.7925, "grad_norm": 6.625, "learning_rate": 1.8874134470592836e-05, "loss": 1.4442, "step": 7117 }, { "epoch": 17.795, "grad_norm": 6.59375, "learning_rate": 1.8873761195702838e-05, "loss": 1.476, "step": 7118 }, { "epoch": 17.7975, "grad_norm": 5.3125, "learning_rate": 1.8873387862636818e-05, "loss": 1.3565, "step": 7119 }, { "epoch": 17.8, "grad_norm": 8.9375, "learning_rate": 1.8873014471397225e-05, "loss": 1.3261, "step": 7120 }, { "epoch": 17.8025, "grad_norm": 6.59375, "learning_rate": 1.88726410219865e-05, "loss": 1.3174, "step": 7121 }, { "epoch": 17.805, "grad_norm": 5.1875, "learning_rate": 1.88722675144071e-05, "loss": 1.2202, "step": 7122 }, { "epoch": 17.8075, "grad_norm": 6.40625, "learning_rate": 1.8871893948661467e-05, "loss": 1.4738, "step": 7123 }, { "epoch": 17.81, "grad_norm": 6.65625, "learning_rate": 1.8871520324752054e-05, "loss": 1.5252, "step": 7124 }, { "epoch": 17.8125, "grad_norm": 7.25, "learning_rate": 1.8871146642681304e-05, "loss": 1.4521, "step": 7125 }, { "epoch": 17.815, "grad_norm": 7.09375, "learning_rate": 1.8870772902451676e-05, "loss": 1.3832, "step": 7126 }, { "epoch": 17.8175, "grad_norm": 8.25, "learning_rate": 1.8870399104065616e-05, "loss": 1.4797, "step": 7127 }, { "epoch": 17.82, "grad_norm": 7.71875, "learning_rate": 1.887002524752557e-05, "loss": 1.5414, "step": 7128 }, { "epoch": 17.8225, "grad_norm": 6.03125, "learning_rate": 1.8869651332833996e-05, "loss": 1.3758, "step": 7129 }, { "epoch": 17.825, "grad_norm": 8.25, "learning_rate": 1.8869277359993346e-05, "loss": 1.5425, "step": 7130 }, { "epoch": 17.8275, "grad_norm": 7.53125, "learning_rate": 1.8868903329006065e-05, "loss": 1.5078, "step": 7131 }, { "epoch": 17.83, "grad_norm": 8.25, "learning_rate": 1.8868529239874608e-05, "loss": 1.3776, "step": 7132 }, { "epoch": 17.8325, "grad_norm": 7.9375, "learning_rate": 1.886815509260143e-05, "loss": 1.2852, "step": 7133 }, { "epoch": 17.835, "grad_norm": 7.0625, "learning_rate": 1.886778088718898e-05, "loss": 1.578, "step": 7134 }, { "epoch": 17.8375, "grad_norm": 7.90625, "learning_rate": 1.8867406623639718e-05, "loss": 1.4537, "step": 7135 }, { "epoch": 17.84, "grad_norm": 7.8125, "learning_rate": 1.886703230195609e-05, "loss": 1.4603, "step": 7136 }, { "epoch": 17.8425, "grad_norm": 5.78125, "learning_rate": 1.886665792214055e-05, "loss": 1.4618, "step": 7137 }, { "epoch": 17.845, "grad_norm": 6.90625, "learning_rate": 1.886628348419556e-05, "loss": 1.5867, "step": 7138 }, { "epoch": 17.8475, "grad_norm": 7.8125, "learning_rate": 1.886590898812357e-05, "loss": 1.5068, "step": 7139 }, { "epoch": 17.85, "grad_norm": 6.0, "learning_rate": 1.8865534433927034e-05, "loss": 1.2002, "step": 7140 }, { "epoch": 17.8525, "grad_norm": 5.5, "learning_rate": 1.886515982160841e-05, "loss": 1.5347, "step": 7141 }, { "epoch": 17.855, "grad_norm": 6.34375, "learning_rate": 1.886478515117015e-05, "loss": 1.3746, "step": 7142 }, { "epoch": 17.8575, "grad_norm": 6.6875, "learning_rate": 1.8864410422614718e-05, "loss": 1.392, "step": 7143 }, { "epoch": 17.86, "grad_norm": 7.03125, "learning_rate": 1.8864035635944564e-05, "loss": 1.4683, "step": 7144 }, { "epoch": 17.8625, "grad_norm": 6.1875, "learning_rate": 1.8863660791162147e-05, "loss": 1.4868, "step": 7145 }, { "epoch": 17.865, "grad_norm": 6.3125, "learning_rate": 1.8863285888269925e-05, "loss": 1.4865, "step": 7146 }, { "epoch": 17.8675, "grad_norm": 6.6875, "learning_rate": 1.8862910927270355e-05, "loss": 1.3462, "step": 7147 }, { "epoch": 17.87, "grad_norm": 5.40625, "learning_rate": 1.8862535908165897e-05, "loss": 1.2962, "step": 7148 }, { "epoch": 17.8725, "grad_norm": 9.5, "learning_rate": 1.8862160830959007e-05, "loss": 1.3814, "step": 7149 }, { "epoch": 17.875, "grad_norm": 7.40625, "learning_rate": 1.8861785695652142e-05, "loss": 1.4359, "step": 7150 }, { "epoch": 17.8775, "grad_norm": 6.125, "learning_rate": 1.886141050224777e-05, "loss": 1.3388, "step": 7151 }, { "epoch": 17.88, "grad_norm": 7.28125, "learning_rate": 1.8861035250748343e-05, "loss": 1.2334, "step": 7152 }, { "epoch": 17.8825, "grad_norm": 10.0625, "learning_rate": 1.886065994115632e-05, "loss": 1.4811, "step": 7153 }, { "epoch": 17.885, "grad_norm": 7.84375, "learning_rate": 1.886028457347417e-05, "loss": 1.4706, "step": 7154 }, { "epoch": 17.8875, "grad_norm": 6.625, "learning_rate": 1.885990914770435e-05, "loss": 1.3235, "step": 7155 }, { "epoch": 17.89, "grad_norm": 5.59375, "learning_rate": 1.8859533663849318e-05, "loss": 1.4393, "step": 7156 }, { "epoch": 17.8925, "grad_norm": 5.84375, "learning_rate": 1.8859158121911533e-05, "loss": 1.3565, "step": 7157 }, { "epoch": 17.895, "grad_norm": 6.8125, "learning_rate": 1.885878252189347e-05, "loss": 1.5137, "step": 7158 }, { "epoch": 17.8975, "grad_norm": 6.375, "learning_rate": 1.8858406863797574e-05, "loss": 1.5587, "step": 7159 }, { "epoch": 17.9, "grad_norm": 7.0625, "learning_rate": 1.8858031147626326e-05, "loss": 1.3603, "step": 7160 }, { "epoch": 17.9025, "grad_norm": 4.53125, "learning_rate": 1.8857655373382177e-05, "loss": 1.4493, "step": 7161 }, { "epoch": 17.905, "grad_norm": 5.90625, "learning_rate": 1.8857279541067593e-05, "loss": 1.5318, "step": 7162 }, { "epoch": 17.9075, "grad_norm": 6.34375, "learning_rate": 1.885690365068504e-05, "loss": 1.4834, "step": 7163 }, { "epoch": 17.91, "grad_norm": 7.46875, "learning_rate": 1.8856527702236983e-05, "loss": 1.3982, "step": 7164 }, { "epoch": 17.9125, "grad_norm": 7.0, "learning_rate": 1.8856151695725883e-05, "loss": 1.4313, "step": 7165 }, { "epoch": 17.915, "grad_norm": 5.90625, "learning_rate": 1.88557756311542e-05, "loss": 1.4352, "step": 7166 }, { "epoch": 17.9175, "grad_norm": 7.375, "learning_rate": 1.8855399508524415e-05, "loss": 1.392, "step": 7167 }, { "epoch": 17.92, "grad_norm": 7.71875, "learning_rate": 1.8855023327838984e-05, "loss": 1.1833, "step": 7168 }, { "epoch": 17.9225, "grad_norm": 7.6875, "learning_rate": 1.8854647089100374e-05, "loss": 1.625, "step": 7169 }, { "epoch": 17.925, "grad_norm": 8.25, "learning_rate": 1.885427079231105e-05, "loss": 1.4463, "step": 7170 }, { "epoch": 17.9275, "grad_norm": 5.71875, "learning_rate": 1.8853894437473482e-05, "loss": 1.4535, "step": 7171 }, { "epoch": 17.93, "grad_norm": 7.8125, "learning_rate": 1.8853518024590136e-05, "loss": 1.4714, "step": 7172 }, { "epoch": 17.9325, "grad_norm": 8.25, "learning_rate": 1.8853141553663482e-05, "loss": 1.2797, "step": 7173 }, { "epoch": 17.935, "grad_norm": 9.3125, "learning_rate": 1.885276502469599e-05, "loss": 1.5915, "step": 7174 }, { "epoch": 17.9375, "grad_norm": 10.0, "learning_rate": 1.8852388437690113e-05, "loss": 1.3474, "step": 7175 }, { "epoch": 17.94, "grad_norm": 8.9375, "learning_rate": 1.885201179264834e-05, "loss": 1.3477, "step": 7176 }, { "epoch": 17.9425, "grad_norm": 6.25, "learning_rate": 1.885163508957313e-05, "loss": 1.2792, "step": 7177 }, { "epoch": 17.945, "grad_norm": 6.59375, "learning_rate": 1.8851258328466958e-05, "loss": 1.2538, "step": 7178 }, { "epoch": 17.9475, "grad_norm": 10.9375, "learning_rate": 1.8850881509332286e-05, "loss": 1.5522, "step": 7179 }, { "epoch": 17.95, "grad_norm": 6.0, "learning_rate": 1.885050463217159e-05, "loss": 1.2452, "step": 7180 }, { "epoch": 17.9525, "grad_norm": 5.875, "learning_rate": 1.885012769698734e-05, "loss": 1.3756, "step": 7181 }, { "epoch": 17.955, "grad_norm": 8.125, "learning_rate": 1.8849750703782007e-05, "loss": 1.5142, "step": 7182 }, { "epoch": 17.9575, "grad_norm": 7.59375, "learning_rate": 1.8849373652558064e-05, "loss": 1.5644, "step": 7183 }, { "epoch": 17.96, "grad_norm": 6.0, "learning_rate": 1.8848996543317982e-05, "loss": 1.515, "step": 7184 }, { "epoch": 17.9625, "grad_norm": 4.96875, "learning_rate": 1.884861937606423e-05, "loss": 1.3454, "step": 7185 }, { "epoch": 17.965, "grad_norm": 8.6875, "learning_rate": 1.8848242150799283e-05, "loss": 1.3662, "step": 7186 }, { "epoch": 17.9675, "grad_norm": 5.09375, "learning_rate": 1.884786486752562e-05, "loss": 1.3472, "step": 7187 }, { "epoch": 17.97, "grad_norm": 5.9375, "learning_rate": 1.8847487526245703e-05, "loss": 1.5017, "step": 7188 }, { "epoch": 17.9725, "grad_norm": 7.90625, "learning_rate": 1.8847110126962014e-05, "loss": 1.4391, "step": 7189 }, { "epoch": 17.975, "grad_norm": 9.6875, "learning_rate": 1.8846732669677026e-05, "loss": 1.4785, "step": 7190 }, { "epoch": 17.9775, "grad_norm": 8.4375, "learning_rate": 1.884635515439321e-05, "loss": 1.4335, "step": 7191 }, { "epoch": 17.98, "grad_norm": 7.71875, "learning_rate": 1.8845977581113048e-05, "loss": 1.3422, "step": 7192 }, { "epoch": 17.9825, "grad_norm": 6.46875, "learning_rate": 1.8845599949839005e-05, "loss": 1.3932, "step": 7193 }, { "epoch": 17.985, "grad_norm": 11.5, "learning_rate": 1.884522226057357e-05, "loss": 1.4339, "step": 7194 }, { "epoch": 17.9875, "grad_norm": 5.4375, "learning_rate": 1.8844844513319205e-05, "loss": 1.4447, "step": 7195 }, { "epoch": 17.99, "grad_norm": 6.96875, "learning_rate": 1.8844466708078398e-05, "loss": 1.5184, "step": 7196 }, { "epoch": 17.9925, "grad_norm": 7.375, "learning_rate": 1.8844088844853615e-05, "loss": 1.5416, "step": 7197 }, { "epoch": 17.995, "grad_norm": 6.28125, "learning_rate": 1.8843710923647346e-05, "loss": 1.4168, "step": 7198 }, { "epoch": 17.9975, "grad_norm": 6.71875, "learning_rate": 1.8843332944462062e-05, "loss": 1.4205, "step": 7199 }, { "epoch": 18.0, "grad_norm": 6.875, "learning_rate": 1.8842954907300236e-05, "loss": 1.4714, "step": 7200 }, { "epoch": 18.0025, "grad_norm": 6.84375, "learning_rate": 1.8842576812164357e-05, "loss": 1.4095, "step": 7201 }, { "epoch": 18.005, "grad_norm": 6.15625, "learning_rate": 1.8842198659056894e-05, "loss": 1.2766, "step": 7202 }, { "epoch": 18.0075, "grad_norm": 5.96875, "learning_rate": 1.8841820447980335e-05, "loss": 1.404, "step": 7203 }, { "epoch": 18.01, "grad_norm": 6.625, "learning_rate": 1.884144217893715e-05, "loss": 1.4837, "step": 7204 }, { "epoch": 18.0125, "grad_norm": 7.0, "learning_rate": 1.8841063851929825e-05, "loss": 1.4228, "step": 7205 }, { "epoch": 18.015, "grad_norm": 5.71875, "learning_rate": 1.884068546696084e-05, "loss": 1.4204, "step": 7206 }, { "epoch": 18.0175, "grad_norm": 6.6875, "learning_rate": 1.884030702403268e-05, "loss": 1.3668, "step": 7207 }, { "epoch": 18.02, "grad_norm": 7.96875, "learning_rate": 1.883992852314781e-05, "loss": 1.5727, "step": 7208 }, { "epoch": 18.0225, "grad_norm": 7.125, "learning_rate": 1.8839549964308734e-05, "loss": 1.4864, "step": 7209 }, { "epoch": 18.025, "grad_norm": 6.71875, "learning_rate": 1.8839171347517914e-05, "loss": 1.3474, "step": 7210 }, { "epoch": 18.0275, "grad_norm": 8.1875, "learning_rate": 1.8838792672777844e-05, "loss": 1.4563, "step": 7211 }, { "epoch": 18.03, "grad_norm": 7.71875, "learning_rate": 1.8838413940091e-05, "loss": 1.4119, "step": 7212 }, { "epoch": 18.0325, "grad_norm": 8.1875, "learning_rate": 1.8838035149459872e-05, "loss": 1.5637, "step": 7213 }, { "epoch": 18.035, "grad_norm": 8.1875, "learning_rate": 1.8837656300886937e-05, "loss": 1.6384, "step": 7214 }, { "epoch": 18.0375, "grad_norm": 15.375, "learning_rate": 1.883727739437468e-05, "loss": 1.4677, "step": 7215 }, { "epoch": 18.04, "grad_norm": 6.25, "learning_rate": 1.8836898429925586e-05, "loss": 1.409, "step": 7216 }, { "epoch": 18.0425, "grad_norm": 5.75, "learning_rate": 1.883651940754214e-05, "loss": 1.4139, "step": 7217 }, { "epoch": 18.045, "grad_norm": 7.3125, "learning_rate": 1.883614032722683e-05, "loss": 1.4709, "step": 7218 }, { "epoch": 18.0475, "grad_norm": 5.90625, "learning_rate": 1.8835761188982134e-05, "loss": 1.4986, "step": 7219 }, { "epoch": 18.05, "grad_norm": 8.1875, "learning_rate": 1.883538199281054e-05, "loss": 1.4781, "step": 7220 }, { "epoch": 18.0525, "grad_norm": 8.5, "learning_rate": 1.8835002738714537e-05, "loss": 1.5023, "step": 7221 }, { "epoch": 18.055, "grad_norm": 7.875, "learning_rate": 1.883462342669661e-05, "loss": 1.4164, "step": 7222 }, { "epoch": 18.0575, "grad_norm": 7.125, "learning_rate": 1.8834244056759245e-05, "loss": 1.4392, "step": 7223 }, { "epoch": 18.06, "grad_norm": 6.21875, "learning_rate": 1.883386462890493e-05, "loss": 1.3989, "step": 7224 }, { "epoch": 18.0625, "grad_norm": 6.78125, "learning_rate": 1.883348514313615e-05, "loss": 1.4485, "step": 7225 }, { "epoch": 18.065, "grad_norm": 7.8125, "learning_rate": 1.8833105599455396e-05, "loss": 1.3665, "step": 7226 }, { "epoch": 18.0675, "grad_norm": 7.875, "learning_rate": 1.8832725997865156e-05, "loss": 1.4978, "step": 7227 }, { "epoch": 18.07, "grad_norm": 6.59375, "learning_rate": 1.883234633836792e-05, "loss": 1.51, "step": 7228 }, { "epoch": 18.0725, "grad_norm": 6.59375, "learning_rate": 1.883196662096617e-05, "loss": 1.4669, "step": 7229 }, { "epoch": 18.075, "grad_norm": 6.625, "learning_rate": 1.8831586845662403e-05, "loss": 1.5407, "step": 7230 }, { "epoch": 18.0775, "grad_norm": 6.0625, "learning_rate": 1.883120701245911e-05, "loss": 1.394, "step": 7231 }, { "epoch": 18.08, "grad_norm": 7.125, "learning_rate": 1.883082712135877e-05, "loss": 1.3857, "step": 7232 }, { "epoch": 18.0825, "grad_norm": 8.8125, "learning_rate": 1.883044717236388e-05, "loss": 1.4493, "step": 7233 }, { "epoch": 18.085, "grad_norm": 5.375, "learning_rate": 1.8830067165476937e-05, "loss": 1.3924, "step": 7234 }, { "epoch": 18.0875, "grad_norm": 7.21875, "learning_rate": 1.8829687100700428e-05, "loss": 1.5315, "step": 7235 }, { "epoch": 18.09, "grad_norm": 6.90625, "learning_rate": 1.8829306978036837e-05, "loss": 1.5514, "step": 7236 }, { "epoch": 18.0925, "grad_norm": 8.375, "learning_rate": 1.882892679748867e-05, "loss": 1.4213, "step": 7237 }, { "epoch": 18.095, "grad_norm": 6.0, "learning_rate": 1.8828546559058408e-05, "loss": 1.3008, "step": 7238 }, { "epoch": 18.0975, "grad_norm": 6.03125, "learning_rate": 1.8828166262748547e-05, "loss": 1.4633, "step": 7239 }, { "epoch": 18.1, "grad_norm": 5.3125, "learning_rate": 1.8827785908561585e-05, "loss": 1.4303, "step": 7240 }, { "epoch": 18.1025, "grad_norm": 6.78125, "learning_rate": 1.882740549650001e-05, "loss": 1.5122, "step": 7241 }, { "epoch": 18.105, "grad_norm": 6.59375, "learning_rate": 1.882702502656632e-05, "loss": 1.4197, "step": 7242 }, { "epoch": 18.1075, "grad_norm": 7.71875, "learning_rate": 1.8826644498763005e-05, "loss": 1.4264, "step": 7243 }, { "epoch": 18.11, "grad_norm": 6.59375, "learning_rate": 1.882626391309256e-05, "loss": 1.482, "step": 7244 }, { "epoch": 18.1125, "grad_norm": 7.0, "learning_rate": 1.8825883269557487e-05, "loss": 1.3943, "step": 7245 }, { "epoch": 18.115, "grad_norm": 6.28125, "learning_rate": 1.8825502568160272e-05, "loss": 1.3313, "step": 7246 }, { "epoch": 18.1175, "grad_norm": 6.15625, "learning_rate": 1.8825121808903418e-05, "loss": 1.365, "step": 7247 }, { "epoch": 18.12, "grad_norm": 7.28125, "learning_rate": 1.8824740991789417e-05, "loss": 1.4579, "step": 7248 }, { "epoch": 18.1225, "grad_norm": 7.09375, "learning_rate": 1.8824360116820764e-05, "loss": 1.4321, "step": 7249 }, { "epoch": 18.125, "grad_norm": 9.1875, "learning_rate": 1.8823979183999965e-05, "loss": 1.5017, "step": 7250 }, { "epoch": 18.1275, "grad_norm": 8.9375, "learning_rate": 1.8823598193329512e-05, "loss": 1.3567, "step": 7251 }, { "epoch": 18.13, "grad_norm": 7.0, "learning_rate": 1.88232171448119e-05, "loss": 1.4858, "step": 7252 }, { "epoch": 18.1325, "grad_norm": 7.0625, "learning_rate": 1.8822836038449628e-05, "loss": 1.3917, "step": 7253 }, { "epoch": 18.135, "grad_norm": 7.21875, "learning_rate": 1.8822454874245198e-05, "loss": 1.6612, "step": 7254 }, { "epoch": 18.1375, "grad_norm": 5.84375, "learning_rate": 1.8822073652201105e-05, "loss": 1.2097, "step": 7255 }, { "epoch": 18.14, "grad_norm": 7.03125, "learning_rate": 1.8821692372319852e-05, "loss": 1.4051, "step": 7256 }, { "epoch": 18.1425, "grad_norm": 7.5, "learning_rate": 1.8821311034603934e-05, "loss": 1.5266, "step": 7257 }, { "epoch": 18.145, "grad_norm": 8.3125, "learning_rate": 1.882092963905586e-05, "loss": 1.4565, "step": 7258 }, { "epoch": 18.1475, "grad_norm": 9.5625, "learning_rate": 1.8820548185678118e-05, "loss": 1.3643, "step": 7259 }, { "epoch": 18.15, "grad_norm": 6.1875, "learning_rate": 1.8820166674473217e-05, "loss": 1.3332, "step": 7260 }, { "epoch": 18.1525, "grad_norm": 7.125, "learning_rate": 1.8819785105443657e-05, "loss": 1.4153, "step": 7261 }, { "epoch": 18.155, "grad_norm": 6.15625, "learning_rate": 1.8819403478591938e-05, "loss": 1.553, "step": 7262 }, { "epoch": 18.1575, "grad_norm": 8.375, "learning_rate": 1.8819021793920564e-05, "loss": 1.4282, "step": 7263 }, { "epoch": 18.16, "grad_norm": 7.5, "learning_rate": 1.8818640051432036e-05, "loss": 1.4943, "step": 7264 }, { "epoch": 18.1625, "grad_norm": 5.03125, "learning_rate": 1.8818258251128854e-05, "loss": 1.3993, "step": 7265 }, { "epoch": 18.165, "grad_norm": 9.1875, "learning_rate": 1.8817876393013526e-05, "loss": 1.4184, "step": 7266 }, { "epoch": 18.1675, "grad_norm": 4.46875, "learning_rate": 1.8817494477088554e-05, "loss": 1.3017, "step": 7267 }, { "epoch": 18.17, "grad_norm": 6.40625, "learning_rate": 1.881711250335644e-05, "loss": 1.3286, "step": 7268 }, { "epoch": 18.1725, "grad_norm": 6.3125, "learning_rate": 1.8816730471819687e-05, "loss": 1.4748, "step": 7269 }, { "epoch": 18.175, "grad_norm": 7.375, "learning_rate": 1.8816348382480807e-05, "loss": 1.2973, "step": 7270 }, { "epoch": 18.1775, "grad_norm": 7.96875, "learning_rate": 1.8815966235342298e-05, "loss": 1.6308, "step": 7271 }, { "epoch": 18.18, "grad_norm": 6.71875, "learning_rate": 1.8815584030406663e-05, "loss": 1.4952, "step": 7272 }, { "epoch": 18.1825, "grad_norm": 6.625, "learning_rate": 1.8815201767676418e-05, "loss": 1.5867, "step": 7273 }, { "epoch": 18.185, "grad_norm": 6.25, "learning_rate": 1.881481944715406e-05, "loss": 1.472, "step": 7274 }, { "epoch": 18.1875, "grad_norm": 6.46875, "learning_rate": 1.8814437068842096e-05, "loss": 1.4366, "step": 7275 }, { "epoch": 18.19, "grad_norm": 5.53125, "learning_rate": 1.8814054632743038e-05, "loss": 1.3568, "step": 7276 }, { "epoch": 18.1925, "grad_norm": 6.34375, "learning_rate": 1.8813672138859392e-05, "loss": 1.4814, "step": 7277 }, { "epoch": 18.195, "grad_norm": 4.875, "learning_rate": 1.8813289587193663e-05, "loss": 1.4223, "step": 7278 }, { "epoch": 18.1975, "grad_norm": 12.875, "learning_rate": 1.8812906977748356e-05, "loss": 1.5958, "step": 7279 }, { "epoch": 18.2, "grad_norm": 5.65625, "learning_rate": 1.881252431052599e-05, "loss": 1.4669, "step": 7280 }, { "epoch": 18.2025, "grad_norm": 6.4375, "learning_rate": 1.8812141585529063e-05, "loss": 1.5271, "step": 7281 }, { "epoch": 18.205, "grad_norm": 5.96875, "learning_rate": 1.881175880276009e-05, "loss": 1.3896, "step": 7282 }, { "epoch": 18.2075, "grad_norm": 8.375, "learning_rate": 1.881137596222158e-05, "loss": 1.418, "step": 7283 }, { "epoch": 18.21, "grad_norm": 5.625, "learning_rate": 1.881099306391604e-05, "loss": 1.3723, "step": 7284 }, { "epoch": 18.2125, "grad_norm": 10.125, "learning_rate": 1.8810610107845983e-05, "loss": 1.4936, "step": 7285 }, { "epoch": 18.215, "grad_norm": 5.5, "learning_rate": 1.8810227094013917e-05, "loss": 1.3893, "step": 7286 }, { "epoch": 18.2175, "grad_norm": 7.1875, "learning_rate": 1.8809844022422356e-05, "loss": 1.4708, "step": 7287 }, { "epoch": 18.22, "grad_norm": 5.59375, "learning_rate": 1.880946089307381e-05, "loss": 1.4918, "step": 7288 }, { "epoch": 18.2225, "grad_norm": 27.0, "learning_rate": 1.8809077705970794e-05, "loss": 1.5879, "step": 7289 }, { "epoch": 18.225, "grad_norm": 8.875, "learning_rate": 1.8808694461115814e-05, "loss": 1.431, "step": 7290 }, { "epoch": 18.2275, "grad_norm": 7.25, "learning_rate": 1.8808311158511382e-05, "loss": 1.3864, "step": 7291 }, { "epoch": 18.23, "grad_norm": 7.78125, "learning_rate": 1.880792779816002e-05, "loss": 1.5057, "step": 7292 }, { "epoch": 18.2325, "grad_norm": 5.53125, "learning_rate": 1.8807544380064232e-05, "loss": 1.342, "step": 7293 }, { "epoch": 18.235, "grad_norm": 10.625, "learning_rate": 1.880716090422654e-05, "loss": 1.4883, "step": 7294 }, { "epoch": 18.2375, "grad_norm": 7.90625, "learning_rate": 1.8806777370649452e-05, "loss": 1.3381, "step": 7295 }, { "epoch": 18.24, "grad_norm": 5.71875, "learning_rate": 1.8806393779335483e-05, "loss": 1.5275, "step": 7296 }, { "epoch": 18.2425, "grad_norm": 7.3125, "learning_rate": 1.880601013028715e-05, "loss": 1.5114, "step": 7297 }, { "epoch": 18.245, "grad_norm": 6.0625, "learning_rate": 1.8805626423506966e-05, "loss": 1.3832, "step": 7298 }, { "epoch": 18.2475, "grad_norm": 5.84375, "learning_rate": 1.8805242658997446e-05, "loss": 1.4475, "step": 7299 }, { "epoch": 18.25, "grad_norm": 9.0625, "learning_rate": 1.880485883676111e-05, "loss": 1.3901, "step": 7300 }, { "epoch": 18.2525, "grad_norm": 6.5625, "learning_rate": 1.880447495680047e-05, "loss": 1.4355, "step": 7301 }, { "epoch": 18.255, "grad_norm": 8.75, "learning_rate": 1.8804091019118046e-05, "loss": 1.4325, "step": 7302 }, { "epoch": 18.2575, "grad_norm": 7.46875, "learning_rate": 1.8803707023716353e-05, "loss": 1.3909, "step": 7303 }, { "epoch": 18.26, "grad_norm": 5.46875, "learning_rate": 1.8803322970597907e-05, "loss": 1.4636, "step": 7304 }, { "epoch": 18.2625, "grad_norm": 6.15625, "learning_rate": 1.880293885976523e-05, "loss": 1.3076, "step": 7305 }, { "epoch": 18.265, "grad_norm": 6.78125, "learning_rate": 1.880255469122084e-05, "loss": 1.4635, "step": 7306 }, { "epoch": 18.2675, "grad_norm": 5.875, "learning_rate": 1.8802170464967253e-05, "loss": 1.4067, "step": 7307 }, { "epoch": 18.27, "grad_norm": 4.75, "learning_rate": 1.8801786181006985e-05, "loss": 1.3857, "step": 7308 }, { "epoch": 18.2725, "grad_norm": 20.625, "learning_rate": 1.8801401839342564e-05, "loss": 1.4606, "step": 7309 }, { "epoch": 18.275, "grad_norm": 7.0, "learning_rate": 1.88010174399765e-05, "loss": 1.4169, "step": 7310 }, { "epoch": 18.2775, "grad_norm": 9.8125, "learning_rate": 1.880063298291132e-05, "loss": 1.5681, "step": 7311 }, { "epoch": 18.28, "grad_norm": 6.03125, "learning_rate": 1.8800248468149545e-05, "loss": 1.2463, "step": 7312 }, { "epoch": 18.2825, "grad_norm": 6.5625, "learning_rate": 1.8799863895693687e-05, "loss": 1.4022, "step": 7313 }, { "epoch": 18.285, "grad_norm": 6.59375, "learning_rate": 1.879947926554628e-05, "loss": 1.6216, "step": 7314 }, { "epoch": 18.2875, "grad_norm": 6.96875, "learning_rate": 1.8799094577709834e-05, "loss": 1.2867, "step": 7315 }, { "epoch": 18.29, "grad_norm": 5.90625, "learning_rate": 1.879870983218688e-05, "loss": 1.3446, "step": 7316 }, { "epoch": 18.2925, "grad_norm": 12.0, "learning_rate": 1.8798325028979932e-05, "loss": 1.4038, "step": 7317 }, { "epoch": 18.295, "grad_norm": 5.90625, "learning_rate": 1.879794016809152e-05, "loss": 1.2792, "step": 7318 }, { "epoch": 18.2975, "grad_norm": 6.65625, "learning_rate": 1.8797555249524168e-05, "loss": 1.4695, "step": 7319 }, { "epoch": 18.3, "grad_norm": 5.625, "learning_rate": 1.879717027328039e-05, "loss": 1.2792, "step": 7320 }, { "epoch": 18.3025, "grad_norm": 7.0625, "learning_rate": 1.8796785239362717e-05, "loss": 1.5213, "step": 7321 }, { "epoch": 18.305, "grad_norm": 10.625, "learning_rate": 1.879640014777367e-05, "loss": 1.5246, "step": 7322 }, { "epoch": 18.3075, "grad_norm": 7.5625, "learning_rate": 1.879601499851578e-05, "loss": 1.5171, "step": 7323 }, { "epoch": 18.31, "grad_norm": 7.0, "learning_rate": 1.879562979159157e-05, "loss": 1.4401, "step": 7324 }, { "epoch": 18.3125, "grad_norm": 6.25, "learning_rate": 1.8795244527003557e-05, "loss": 1.3833, "step": 7325 }, { "epoch": 18.315, "grad_norm": 6.6875, "learning_rate": 1.8794859204754274e-05, "loss": 1.2536, "step": 7326 }, { "epoch": 18.3175, "grad_norm": 6.4375, "learning_rate": 1.8794473824846248e-05, "loss": 1.289, "step": 7327 }, { "epoch": 18.32, "grad_norm": 6.0, "learning_rate": 1.8794088387282e-05, "loss": 1.3949, "step": 7328 }, { "epoch": 18.3225, "grad_norm": 6.46875, "learning_rate": 1.8793702892064066e-05, "loss": 1.4166, "step": 7329 }, { "epoch": 18.325, "grad_norm": 5.28125, "learning_rate": 1.8793317339194965e-05, "loss": 1.4806, "step": 7330 }, { "epoch": 18.3275, "grad_norm": 7.90625, "learning_rate": 1.8792931728677226e-05, "loss": 1.4451, "step": 7331 }, { "epoch": 18.33, "grad_norm": 6.5, "learning_rate": 1.879254606051338e-05, "loss": 1.602, "step": 7332 }, { "epoch": 18.3325, "grad_norm": 7.28125, "learning_rate": 1.8792160334705953e-05, "loss": 1.3589, "step": 7333 }, { "epoch": 18.335, "grad_norm": 7.8125, "learning_rate": 1.8791774551257476e-05, "loss": 1.3686, "step": 7334 }, { "epoch": 18.3375, "grad_norm": 5.5, "learning_rate": 1.8791388710170477e-05, "loss": 1.4207, "step": 7335 }, { "epoch": 18.34, "grad_norm": 8.3125, "learning_rate": 1.8791002811447483e-05, "loss": 1.4156, "step": 7336 }, { "epoch": 18.3425, "grad_norm": 6.6875, "learning_rate": 1.879061685509103e-05, "loss": 1.4703, "step": 7337 }, { "epoch": 18.345, "grad_norm": 6.34375, "learning_rate": 1.879023084110364e-05, "loss": 1.507, "step": 7338 }, { "epoch": 18.3475, "grad_norm": 9.9375, "learning_rate": 1.878984476948785e-05, "loss": 1.4428, "step": 7339 }, { "epoch": 18.35, "grad_norm": 5.96875, "learning_rate": 1.8789458640246193e-05, "loss": 1.3468, "step": 7340 }, { "epoch": 18.3525, "grad_norm": 5.53125, "learning_rate": 1.878907245338119e-05, "loss": 1.353, "step": 7341 }, { "epoch": 18.355, "grad_norm": 6.6875, "learning_rate": 1.8788686208895388e-05, "loss": 1.5819, "step": 7342 }, { "epoch": 18.3575, "grad_norm": 7.78125, "learning_rate": 1.8788299906791306e-05, "loss": 1.305, "step": 7343 }, { "epoch": 18.36, "grad_norm": 6.4375, "learning_rate": 1.8787913547071485e-05, "loss": 1.3391, "step": 7344 }, { "epoch": 18.3625, "grad_norm": 6.8125, "learning_rate": 1.878752712973845e-05, "loss": 1.4244, "step": 7345 }, { "epoch": 18.365, "grad_norm": 6.65625, "learning_rate": 1.878714065479474e-05, "loss": 1.3672, "step": 7346 }, { "epoch": 18.3675, "grad_norm": 4.96875, "learning_rate": 1.878675412224289e-05, "loss": 1.2492, "step": 7347 }, { "epoch": 18.37, "grad_norm": 8.125, "learning_rate": 1.8786367532085426e-05, "loss": 1.3751, "step": 7348 }, { "epoch": 18.3725, "grad_norm": 8.4375, "learning_rate": 1.8785980884324892e-05, "loss": 1.5526, "step": 7349 }, { "epoch": 18.375, "grad_norm": 7.625, "learning_rate": 1.878559417896382e-05, "loss": 1.4725, "step": 7350 }, { "epoch": 18.3775, "grad_norm": 8.3125, "learning_rate": 1.878520741600474e-05, "loss": 1.2918, "step": 7351 }, { "epoch": 18.38, "grad_norm": 7.9375, "learning_rate": 1.8784820595450198e-05, "loss": 1.4665, "step": 7352 }, { "epoch": 18.3825, "grad_norm": 7.21875, "learning_rate": 1.878443371730272e-05, "loss": 1.5785, "step": 7353 }, { "epoch": 18.385, "grad_norm": 5.53125, "learning_rate": 1.8784046781564847e-05, "loss": 1.5375, "step": 7354 }, { "epoch": 18.3875, "grad_norm": 7.09375, "learning_rate": 1.8783659788239115e-05, "loss": 1.5015, "step": 7355 }, { "epoch": 18.39, "grad_norm": 5.96875, "learning_rate": 1.878327273732806e-05, "loss": 1.4015, "step": 7356 }, { "epoch": 18.3925, "grad_norm": 7.40625, "learning_rate": 1.878288562883422e-05, "loss": 1.2349, "step": 7357 }, { "epoch": 18.395, "grad_norm": 7.40625, "learning_rate": 1.8782498462760133e-05, "loss": 1.4095, "step": 7358 }, { "epoch": 18.3975, "grad_norm": 7.375, "learning_rate": 1.878211123910834e-05, "loss": 1.5305, "step": 7359 }, { "epoch": 18.4, "grad_norm": 7.53125, "learning_rate": 1.8781723957881374e-05, "loss": 1.5857, "step": 7360 }, { "epoch": 18.4025, "grad_norm": 9.1875, "learning_rate": 1.878133661908178e-05, "loss": 1.4779, "step": 7361 }, { "epoch": 18.405, "grad_norm": 6.15625, "learning_rate": 1.8780949222712093e-05, "loss": 1.4147, "step": 7362 }, { "epoch": 18.4075, "grad_norm": 5.34375, "learning_rate": 1.8780561768774853e-05, "loss": 1.4005, "step": 7363 }, { "epoch": 18.41, "grad_norm": 8.125, "learning_rate": 1.8780174257272603e-05, "loss": 1.6427, "step": 7364 }, { "epoch": 18.4125, "grad_norm": 7.6875, "learning_rate": 1.877978668820788e-05, "loss": 1.2754, "step": 7365 }, { "epoch": 18.415, "grad_norm": 6.78125, "learning_rate": 1.877939906158323e-05, "loss": 1.3994, "step": 7366 }, { "epoch": 18.4175, "grad_norm": 6.4375, "learning_rate": 1.8779011377401192e-05, "loss": 1.3854, "step": 7367 }, { "epoch": 18.42, "grad_norm": 7.03125, "learning_rate": 1.8778623635664305e-05, "loss": 1.4012, "step": 7368 }, { "epoch": 18.4225, "grad_norm": 9.3125, "learning_rate": 1.877823583637511e-05, "loss": 1.3358, "step": 7369 }, { "epoch": 18.425, "grad_norm": 6.3125, "learning_rate": 1.8777847979536155e-05, "loss": 1.515, "step": 7370 }, { "epoch": 18.4275, "grad_norm": 8.4375, "learning_rate": 1.8777460065149978e-05, "loss": 1.6763, "step": 7371 }, { "epoch": 18.43, "grad_norm": 10.8125, "learning_rate": 1.8777072093219127e-05, "loss": 1.483, "step": 7372 }, { "epoch": 18.4325, "grad_norm": 7.40625, "learning_rate": 1.877668406374614e-05, "loss": 1.2754, "step": 7373 }, { "epoch": 18.435, "grad_norm": 5.90625, "learning_rate": 1.8776295976733564e-05, "loss": 1.4039, "step": 7374 }, { "epoch": 18.4375, "grad_norm": 9.4375, "learning_rate": 1.8775907832183945e-05, "loss": 1.6495, "step": 7375 }, { "epoch": 18.44, "grad_norm": 6.90625, "learning_rate": 1.8775519630099822e-05, "loss": 1.413, "step": 7376 }, { "epoch": 18.4425, "grad_norm": 5.0625, "learning_rate": 1.8775131370483745e-05, "loss": 1.3973, "step": 7377 }, { "epoch": 18.445, "grad_norm": 6.84375, "learning_rate": 1.8774743053338257e-05, "loss": 1.3749, "step": 7378 }, { "epoch": 18.4475, "grad_norm": 9.8125, "learning_rate": 1.8774354678665906e-05, "loss": 1.4535, "step": 7379 }, { "epoch": 18.45, "grad_norm": 6.5625, "learning_rate": 1.8773966246469238e-05, "loss": 1.3373, "step": 7380 }, { "epoch": 18.4525, "grad_norm": 7.65625, "learning_rate": 1.87735777567508e-05, "loss": 1.4303, "step": 7381 }, { "epoch": 18.455, "grad_norm": 7.0, "learning_rate": 1.877318920951313e-05, "loss": 1.4093, "step": 7382 }, { "epoch": 18.4575, "grad_norm": 9.875, "learning_rate": 1.877280060475879e-05, "loss": 1.4749, "step": 7383 }, { "epoch": 18.46, "grad_norm": 8.1875, "learning_rate": 1.8772411942490315e-05, "loss": 1.3492, "step": 7384 }, { "epoch": 18.4625, "grad_norm": 7.03125, "learning_rate": 1.877202322271026e-05, "loss": 1.414, "step": 7385 }, { "epoch": 18.465, "grad_norm": 9.8125, "learning_rate": 1.8771634445421174e-05, "loss": 1.4199, "step": 7386 }, { "epoch": 18.4675, "grad_norm": 7.5, "learning_rate": 1.87712456106256e-05, "loss": 1.4269, "step": 7387 }, { "epoch": 18.47, "grad_norm": 8.1875, "learning_rate": 1.8770856718326094e-05, "loss": 1.4607, "step": 7388 }, { "epoch": 18.4725, "grad_norm": 6.21875, "learning_rate": 1.8770467768525203e-05, "loss": 1.4438, "step": 7389 }, { "epoch": 18.475, "grad_norm": 5.375, "learning_rate": 1.8770078761225476e-05, "loss": 1.5113, "step": 7390 }, { "epoch": 18.4775, "grad_norm": 6.90625, "learning_rate": 1.876968969642946e-05, "loss": 1.3975, "step": 7391 }, { "epoch": 18.48, "grad_norm": 9.375, "learning_rate": 1.876930057413971e-05, "loss": 1.524, "step": 7392 }, { "epoch": 18.4825, "grad_norm": 7.625, "learning_rate": 1.8768911394358777e-05, "loss": 1.5285, "step": 7393 }, { "epoch": 18.485, "grad_norm": 6.5, "learning_rate": 1.8768522157089214e-05, "loss": 1.4868, "step": 7394 }, { "epoch": 18.4875, "grad_norm": 7.65625, "learning_rate": 1.876813286233357e-05, "loss": 1.3876, "step": 7395 }, { "epoch": 18.49, "grad_norm": 6.0, "learning_rate": 1.8767743510094395e-05, "loss": 1.3062, "step": 7396 }, { "epoch": 18.4925, "grad_norm": 11.0625, "learning_rate": 1.8767354100374248e-05, "loss": 1.681, "step": 7397 }, { "epoch": 18.495, "grad_norm": 6.90625, "learning_rate": 1.8766964633175675e-05, "loss": 1.4436, "step": 7398 }, { "epoch": 18.4975, "grad_norm": 8.3125, "learning_rate": 1.8766575108501233e-05, "loss": 1.3745, "step": 7399 }, { "epoch": 18.5, "grad_norm": 5.90625, "learning_rate": 1.876618552635348e-05, "loss": 1.5815, "step": 7400 }, { "epoch": 18.5025, "grad_norm": 6.9375, "learning_rate": 1.8765795886734963e-05, "loss": 1.5307, "step": 7401 }, { "epoch": 18.505, "grad_norm": 9.5, "learning_rate": 1.8765406189648235e-05, "loss": 1.525, "step": 7402 }, { "epoch": 18.5075, "grad_norm": 8.1875, "learning_rate": 1.876501643509586e-05, "loss": 1.3216, "step": 7403 }, { "epoch": 18.51, "grad_norm": 7.1875, "learning_rate": 1.8764626623080385e-05, "loss": 1.4123, "step": 7404 }, { "epoch": 18.5125, "grad_norm": 6.21875, "learning_rate": 1.876423675360437e-05, "loss": 1.4273, "step": 7405 }, { "epoch": 18.515, "grad_norm": 8.8125, "learning_rate": 1.876384682667037e-05, "loss": 1.56, "step": 7406 }, { "epoch": 18.5175, "grad_norm": 11.875, "learning_rate": 1.876345684228094e-05, "loss": 1.3891, "step": 7407 }, { "epoch": 18.52, "grad_norm": 7.78125, "learning_rate": 1.8763066800438638e-05, "loss": 1.3871, "step": 7408 }, { "epoch": 18.5225, "grad_norm": 5.5625, "learning_rate": 1.876267670114602e-05, "loss": 1.3125, "step": 7409 }, { "epoch": 18.525, "grad_norm": 10.875, "learning_rate": 1.8762286544405643e-05, "loss": 1.488, "step": 7410 }, { "epoch": 18.5275, "grad_norm": 8.5625, "learning_rate": 1.876189633022007e-05, "loss": 1.2253, "step": 7411 }, { "epoch": 18.53, "grad_norm": 8.5625, "learning_rate": 1.876150605859185e-05, "loss": 1.3115, "step": 7412 }, { "epoch": 18.5325, "grad_norm": 7.53125, "learning_rate": 1.8761115729523555e-05, "loss": 1.6032, "step": 7413 }, { "epoch": 18.535, "grad_norm": 8.625, "learning_rate": 1.876072534301773e-05, "loss": 1.2067, "step": 7414 }, { "epoch": 18.5375, "grad_norm": 9.0625, "learning_rate": 1.876033489907694e-05, "loss": 1.5829, "step": 7415 }, { "epoch": 18.54, "grad_norm": 5.25, "learning_rate": 1.8759944397703748e-05, "loss": 1.3947, "step": 7416 }, { "epoch": 18.5425, "grad_norm": 6.5, "learning_rate": 1.875955383890071e-05, "loss": 1.4302, "step": 7417 }, { "epoch": 18.545, "grad_norm": 6.09375, "learning_rate": 1.8759163222670383e-05, "loss": 1.4295, "step": 7418 }, { "epoch": 18.5475, "grad_norm": 4.46875, "learning_rate": 1.875877254901534e-05, "loss": 1.1812, "step": 7419 }, { "epoch": 18.55, "grad_norm": 7.03125, "learning_rate": 1.8758381817938126e-05, "loss": 1.5031, "step": 7420 }, { "epoch": 18.5525, "grad_norm": 5.15625, "learning_rate": 1.8757991029441318e-05, "loss": 1.5288, "step": 7421 }, { "epoch": 18.555, "grad_norm": 6.34375, "learning_rate": 1.875760018352747e-05, "loss": 1.485, "step": 7422 }, { "epoch": 18.5575, "grad_norm": 6.46875, "learning_rate": 1.8757209280199145e-05, "loss": 1.4473, "step": 7423 }, { "epoch": 18.56, "grad_norm": 5.96875, "learning_rate": 1.875681831945891e-05, "loss": 1.2772, "step": 7424 }, { "epoch": 18.5625, "grad_norm": 5.9375, "learning_rate": 1.8756427301309317e-05, "loss": 1.2657, "step": 7425 }, { "epoch": 18.565, "grad_norm": 9.5625, "learning_rate": 1.8756036225752942e-05, "loss": 1.4123, "step": 7426 }, { "epoch": 18.5675, "grad_norm": 9.125, "learning_rate": 1.8755645092792344e-05, "loss": 1.3395, "step": 7427 }, { "epoch": 18.57, "grad_norm": 8.0625, "learning_rate": 1.8755253902430082e-05, "loss": 1.4528, "step": 7428 }, { "epoch": 18.5725, "grad_norm": 6.53125, "learning_rate": 1.875486265466873e-05, "loss": 1.423, "step": 7429 }, { "epoch": 18.575, "grad_norm": 6.46875, "learning_rate": 1.875447134951085e-05, "loss": 1.416, "step": 7430 }, { "epoch": 18.5775, "grad_norm": 6.5625, "learning_rate": 1.8754079986959005e-05, "loss": 1.3579, "step": 7431 }, { "epoch": 18.58, "grad_norm": 5.875, "learning_rate": 1.875368856701576e-05, "loss": 1.4023, "step": 7432 }, { "epoch": 18.5825, "grad_norm": 6.09375, "learning_rate": 1.8753297089683686e-05, "loss": 1.3975, "step": 7433 }, { "epoch": 18.585, "grad_norm": 9.0, "learning_rate": 1.8752905554965347e-05, "loss": 1.3476, "step": 7434 }, { "epoch": 18.5875, "grad_norm": 7.375, "learning_rate": 1.8752513962863306e-05, "loss": 1.4832, "step": 7435 }, { "epoch": 18.59, "grad_norm": 5.03125, "learning_rate": 1.8752122313380137e-05, "loss": 1.5491, "step": 7436 }, { "epoch": 18.5925, "grad_norm": 5.21875, "learning_rate": 1.8751730606518403e-05, "loss": 1.2209, "step": 7437 }, { "epoch": 18.595, "grad_norm": 7.9375, "learning_rate": 1.8751338842280674e-05, "loss": 1.5742, "step": 7438 }, { "epoch": 18.5975, "grad_norm": 6.59375, "learning_rate": 1.8750947020669517e-05, "loss": 1.3287, "step": 7439 }, { "epoch": 18.6, "grad_norm": 10.8125, "learning_rate": 1.87505551416875e-05, "loss": 1.4686, "step": 7440 }, { "epoch": 18.6025, "grad_norm": 7.78125, "learning_rate": 1.8750163205337196e-05, "loss": 1.348, "step": 7441 }, { "epoch": 18.605, "grad_norm": 7.0625, "learning_rate": 1.8749771211621172e-05, "loss": 1.3946, "step": 7442 }, { "epoch": 18.6075, "grad_norm": 6.75, "learning_rate": 1.8749379160541996e-05, "loss": 1.4296, "step": 7443 }, { "epoch": 18.61, "grad_norm": 6.5625, "learning_rate": 1.874898705210224e-05, "loss": 1.3558, "step": 7444 }, { "epoch": 18.6125, "grad_norm": 6.03125, "learning_rate": 1.874859488630448e-05, "loss": 1.341, "step": 7445 }, { "epoch": 18.615, "grad_norm": 5.65625, "learning_rate": 1.874820266315128e-05, "loss": 1.2645, "step": 7446 }, { "epoch": 18.6175, "grad_norm": 4.40625, "learning_rate": 1.874781038264521e-05, "loss": 1.2825, "step": 7447 }, { "epoch": 18.62, "grad_norm": 7.78125, "learning_rate": 1.8747418044788848e-05, "loss": 1.3938, "step": 7448 }, { "epoch": 18.6225, "grad_norm": 6.96875, "learning_rate": 1.8747025649584763e-05, "loss": 1.5657, "step": 7449 }, { "epoch": 18.625, "grad_norm": 8.3125, "learning_rate": 1.8746633197035525e-05, "loss": 1.3316, "step": 7450 }, { "epoch": 18.6275, "grad_norm": 7.625, "learning_rate": 1.8746240687143714e-05, "loss": 1.2748, "step": 7451 }, { "epoch": 18.63, "grad_norm": 6.0, "learning_rate": 1.8745848119911897e-05, "loss": 1.3747, "step": 7452 }, { "epoch": 18.6325, "grad_norm": 7.84375, "learning_rate": 1.8745455495342654e-05, "loss": 1.4549, "step": 7453 }, { "epoch": 18.635, "grad_norm": 6.375, "learning_rate": 1.8745062813438547e-05, "loss": 1.4128, "step": 7454 }, { "epoch": 18.6375, "grad_norm": 12.75, "learning_rate": 1.874467007420216e-05, "loss": 1.3552, "step": 7455 }, { "epoch": 18.64, "grad_norm": 7.40625, "learning_rate": 1.874427727763607e-05, "loss": 1.6148, "step": 7456 }, { "epoch": 18.6425, "grad_norm": 5.90625, "learning_rate": 1.8743884423742846e-05, "loss": 1.318, "step": 7457 }, { "epoch": 18.645, "grad_norm": 9.1875, "learning_rate": 1.8743491512525067e-05, "loss": 1.382, "step": 7458 }, { "epoch": 18.6475, "grad_norm": 5.71875, "learning_rate": 1.8743098543985306e-05, "loss": 1.3873, "step": 7459 }, { "epoch": 18.65, "grad_norm": 8.75, "learning_rate": 1.874270551812614e-05, "loss": 1.5497, "step": 7460 }, { "epoch": 18.6525, "grad_norm": 6.25, "learning_rate": 1.874231243495015e-05, "loss": 1.216, "step": 7461 }, { "epoch": 18.655, "grad_norm": 6.6875, "learning_rate": 1.8741919294459908e-05, "loss": 1.4728, "step": 7462 }, { "epoch": 18.6575, "grad_norm": 8.9375, "learning_rate": 1.874152609665799e-05, "loss": 1.344, "step": 7463 }, { "epoch": 18.66, "grad_norm": 6.375, "learning_rate": 1.874113284154698e-05, "loss": 1.3894, "step": 7464 }, { "epoch": 18.6625, "grad_norm": 4.8125, "learning_rate": 1.8740739529129453e-05, "loss": 1.3955, "step": 7465 }, { "epoch": 18.665, "grad_norm": 8.625, "learning_rate": 1.8740346159407985e-05, "loss": 1.4181, "step": 7466 }, { "epoch": 18.6675, "grad_norm": 8.3125, "learning_rate": 1.873995273238516e-05, "loss": 1.4719, "step": 7467 }, { "epoch": 18.67, "grad_norm": 6.53125, "learning_rate": 1.8739559248063553e-05, "loss": 1.431, "step": 7468 }, { "epoch": 18.6725, "grad_norm": 9.5, "learning_rate": 1.8739165706445742e-05, "loss": 1.4915, "step": 7469 }, { "epoch": 18.675, "grad_norm": 6.84375, "learning_rate": 1.8738772107534317e-05, "loss": 1.4773, "step": 7470 }, { "epoch": 18.6775, "grad_norm": 6.03125, "learning_rate": 1.873837845133185e-05, "loss": 1.5405, "step": 7471 }, { "epoch": 18.68, "grad_norm": 7.53125, "learning_rate": 1.873798473784092e-05, "loss": 1.4223, "step": 7472 }, { "epoch": 18.6825, "grad_norm": 12.5625, "learning_rate": 1.8737590967064114e-05, "loss": 1.3868, "step": 7473 }, { "epoch": 18.685, "grad_norm": 6.90625, "learning_rate": 1.873719713900401e-05, "loss": 1.5725, "step": 7474 }, { "epoch": 18.6875, "grad_norm": 8.1875, "learning_rate": 1.873680325366319e-05, "loss": 1.3901, "step": 7475 }, { "epoch": 18.69, "grad_norm": 5.75, "learning_rate": 1.8736409311044244e-05, "loss": 1.499, "step": 7476 }, { "epoch": 18.6925, "grad_norm": 6.375, "learning_rate": 1.8736015311149743e-05, "loss": 1.5646, "step": 7477 }, { "epoch": 18.695, "grad_norm": 7.96875, "learning_rate": 1.873562125398228e-05, "loss": 1.4778, "step": 7478 }, { "epoch": 18.6975, "grad_norm": 7.96875, "learning_rate": 1.8735227139544426e-05, "loss": 1.4956, "step": 7479 }, { "epoch": 18.7, "grad_norm": 5.3125, "learning_rate": 1.8734832967838775e-05, "loss": 1.4534, "step": 7480 }, { "epoch": 18.7025, "grad_norm": 8.4375, "learning_rate": 1.873443873886791e-05, "loss": 1.4883, "step": 7481 }, { "epoch": 18.705, "grad_norm": 6.71875, "learning_rate": 1.8734044452634418e-05, "loss": 1.5056, "step": 7482 }, { "epoch": 18.7075, "grad_norm": 7.0625, "learning_rate": 1.8733650109140873e-05, "loss": 1.543, "step": 7483 }, { "epoch": 18.71, "grad_norm": 7.375, "learning_rate": 1.8733255708389872e-05, "loss": 1.4069, "step": 7484 }, { "epoch": 18.7125, "grad_norm": 6.96875, "learning_rate": 1.8732861250383994e-05, "loss": 1.431, "step": 7485 }, { "epoch": 18.715, "grad_norm": 8.375, "learning_rate": 1.873246673512583e-05, "loss": 1.4625, "step": 7486 }, { "epoch": 18.7175, "grad_norm": 5.46875, "learning_rate": 1.873207216261796e-05, "loss": 1.4435, "step": 7487 }, { "epoch": 18.72, "grad_norm": 6.75, "learning_rate": 1.8731677532862975e-05, "loss": 1.2551, "step": 7488 }, { "epoch": 18.7225, "grad_norm": 6.8125, "learning_rate": 1.873128284586346e-05, "loss": 1.466, "step": 7489 }, { "epoch": 18.725, "grad_norm": 8.75, "learning_rate": 1.8730888101622006e-05, "loss": 1.4889, "step": 7490 }, { "epoch": 18.7275, "grad_norm": 8.6875, "learning_rate": 1.8730493300141202e-05, "loss": 1.5026, "step": 7491 }, { "epoch": 18.73, "grad_norm": 8.4375, "learning_rate": 1.873009844142363e-05, "loss": 1.3709, "step": 7492 }, { "epoch": 18.7325, "grad_norm": 9.6875, "learning_rate": 1.872970352547188e-05, "loss": 1.4857, "step": 7493 }, { "epoch": 18.735, "grad_norm": 8.625, "learning_rate": 1.8729308552288547e-05, "loss": 1.554, "step": 7494 }, { "epoch": 18.7375, "grad_norm": 9.625, "learning_rate": 1.8728913521876215e-05, "loss": 1.5676, "step": 7495 }, { "epoch": 18.74, "grad_norm": 5.625, "learning_rate": 1.8728518434237476e-05, "loss": 1.3142, "step": 7496 }, { "epoch": 18.7425, "grad_norm": 7.75, "learning_rate": 1.872812328937492e-05, "loss": 1.4007, "step": 7497 }, { "epoch": 18.745, "grad_norm": 4.25, "learning_rate": 1.8727728087291136e-05, "loss": 1.4007, "step": 7498 }, { "epoch": 18.7475, "grad_norm": 8.25, "learning_rate": 1.8727332827988713e-05, "loss": 1.2888, "step": 7499 }, { "epoch": 18.75, "grad_norm": 10.5625, "learning_rate": 1.8726937511470247e-05, "loss": 1.4409, "step": 7500 }, { "epoch": 18.7525, "grad_norm": 6.4375, "learning_rate": 1.8726542137738327e-05, "loss": 1.2997, "step": 7501 }, { "epoch": 18.755, "grad_norm": 8.25, "learning_rate": 1.8726146706795553e-05, "loss": 1.4985, "step": 7502 }, { "epoch": 18.7575, "grad_norm": 6.03125, "learning_rate": 1.8725751218644505e-05, "loss": 1.4902, "step": 7503 }, { "epoch": 18.76, "grad_norm": 7.03125, "learning_rate": 1.872535567328778e-05, "loss": 1.3752, "step": 7504 }, { "epoch": 18.7625, "grad_norm": 6.96875, "learning_rate": 1.8724960070727974e-05, "loss": 1.3833, "step": 7505 }, { "epoch": 18.765, "grad_norm": 6.75, "learning_rate": 1.872456441096768e-05, "loss": 1.5005, "step": 7506 }, { "epoch": 18.7675, "grad_norm": 9.3125, "learning_rate": 1.8724168694009485e-05, "loss": 1.587, "step": 7507 }, { "epoch": 18.77, "grad_norm": 8.125, "learning_rate": 1.8723772919855997e-05, "loss": 1.5302, "step": 7508 }, { "epoch": 18.7725, "grad_norm": 7.15625, "learning_rate": 1.8723377088509798e-05, "loss": 1.3059, "step": 7509 }, { "epoch": 18.775, "grad_norm": 8.375, "learning_rate": 1.872298119997349e-05, "loss": 1.5597, "step": 7510 }, { "epoch": 18.7775, "grad_norm": 7.15625, "learning_rate": 1.8722585254249663e-05, "loss": 1.5361, "step": 7511 }, { "epoch": 18.78, "grad_norm": 4.5625, "learning_rate": 1.872218925134092e-05, "loss": 1.3274, "step": 7512 }, { "epoch": 18.7825, "grad_norm": 8.625, "learning_rate": 1.8721793191249853e-05, "loss": 1.409, "step": 7513 }, { "epoch": 18.785, "grad_norm": 8.5625, "learning_rate": 1.8721397073979056e-05, "loss": 1.3467, "step": 7514 }, { "epoch": 18.7875, "grad_norm": 6.0625, "learning_rate": 1.8721000899531134e-05, "loss": 1.4323, "step": 7515 }, { "epoch": 18.79, "grad_norm": 7.125, "learning_rate": 1.8720604667908673e-05, "loss": 1.4633, "step": 7516 }, { "epoch": 18.7925, "grad_norm": 6.28125, "learning_rate": 1.872020837911428e-05, "loss": 1.5393, "step": 7517 }, { "epoch": 18.795, "grad_norm": 7.875, "learning_rate": 1.871981203315055e-05, "loss": 1.3156, "step": 7518 }, { "epoch": 18.7975, "grad_norm": 6.9375, "learning_rate": 1.871941563002008e-05, "loss": 1.4391, "step": 7519 }, { "epoch": 18.8, "grad_norm": 9.4375, "learning_rate": 1.871901916972547e-05, "loss": 1.5385, "step": 7520 }, { "epoch": 18.8025, "grad_norm": 6.0625, "learning_rate": 1.8718622652269323e-05, "loss": 1.3853, "step": 7521 }, { "epoch": 18.805, "grad_norm": 6.34375, "learning_rate": 1.871822607765423e-05, "loss": 1.4143, "step": 7522 }, { "epoch": 18.8075, "grad_norm": 8.5, "learning_rate": 1.8717829445882797e-05, "loss": 1.6698, "step": 7523 }, { "epoch": 18.81, "grad_norm": 6.5, "learning_rate": 1.8717432756957626e-05, "loss": 1.2829, "step": 7524 }, { "epoch": 18.8125, "grad_norm": 8.6875, "learning_rate": 1.871703601088131e-05, "loss": 1.3777, "step": 7525 }, { "epoch": 18.815, "grad_norm": 7.53125, "learning_rate": 1.871663920765646e-05, "loss": 1.341, "step": 7526 }, { "epoch": 18.8175, "grad_norm": 7.1875, "learning_rate": 1.8716242347285667e-05, "loss": 1.5682, "step": 7527 }, { "epoch": 18.82, "grad_norm": 7.0625, "learning_rate": 1.871584542977154e-05, "loss": 1.4054, "step": 7528 }, { "epoch": 18.8225, "grad_norm": 5.75, "learning_rate": 1.871544845511668e-05, "loss": 1.4424, "step": 7529 }, { "epoch": 18.825, "grad_norm": 6.78125, "learning_rate": 1.8715051423323688e-05, "loss": 1.435, "step": 7530 }, { "epoch": 18.8275, "grad_norm": 5.9375, "learning_rate": 1.8714654334395167e-05, "loss": 1.3036, "step": 7531 }, { "epoch": 18.83, "grad_norm": 7.75, "learning_rate": 1.871425718833372e-05, "loss": 1.4797, "step": 7532 }, { "epoch": 18.8325, "grad_norm": 5.4375, "learning_rate": 1.8713859985141956e-05, "loss": 1.3129, "step": 7533 }, { "epoch": 18.835, "grad_norm": 5.28125, "learning_rate": 1.871346272482247e-05, "loss": 1.5417, "step": 7534 }, { "epoch": 18.8375, "grad_norm": 6.9375, "learning_rate": 1.8713065407377873e-05, "loss": 1.3905, "step": 7535 }, { "epoch": 18.84, "grad_norm": 6.34375, "learning_rate": 1.8712668032810767e-05, "loss": 1.3844, "step": 7536 }, { "epoch": 18.8425, "grad_norm": 7.4375, "learning_rate": 1.871227060112376e-05, "loss": 1.4012, "step": 7537 }, { "epoch": 18.845, "grad_norm": 7.09375, "learning_rate": 1.8711873112319454e-05, "loss": 1.4658, "step": 7538 }, { "epoch": 18.8475, "grad_norm": 5.78125, "learning_rate": 1.871147556640046e-05, "loss": 1.3672, "step": 7539 }, { "epoch": 18.85, "grad_norm": 6.6875, "learning_rate": 1.8711077963369377e-05, "loss": 1.2905, "step": 7540 }, { "epoch": 18.8525, "grad_norm": 8.5625, "learning_rate": 1.8710680303228815e-05, "loss": 1.5237, "step": 7541 }, { "epoch": 18.855, "grad_norm": 5.0, "learning_rate": 1.871028258598138e-05, "loss": 1.2856, "step": 7542 }, { "epoch": 18.8575, "grad_norm": 5.25, "learning_rate": 1.8709884811629685e-05, "loss": 1.4362, "step": 7543 }, { "epoch": 18.86, "grad_norm": 7.0, "learning_rate": 1.870948698017633e-05, "loss": 1.3221, "step": 7544 }, { "epoch": 18.8625, "grad_norm": 7.5625, "learning_rate": 1.870908909162393e-05, "loss": 1.4739, "step": 7545 }, { "epoch": 18.865, "grad_norm": 7.1875, "learning_rate": 1.8708691145975095e-05, "loss": 1.4136, "step": 7546 }, { "epoch": 18.8675, "grad_norm": 8.0625, "learning_rate": 1.870829314323242e-05, "loss": 1.4667, "step": 7547 }, { "epoch": 18.87, "grad_norm": 7.28125, "learning_rate": 1.870789508339853e-05, "loss": 1.4244, "step": 7548 }, { "epoch": 18.8725, "grad_norm": 8.25, "learning_rate": 1.8707496966476023e-05, "loss": 1.4543, "step": 7549 }, { "epoch": 18.875, "grad_norm": 6.75, "learning_rate": 1.870709879246752e-05, "loss": 1.2935, "step": 7550 }, { "epoch": 18.8775, "grad_norm": 6.4375, "learning_rate": 1.870670056137562e-05, "loss": 1.4812, "step": 7551 }, { "epoch": 18.88, "grad_norm": 6.625, "learning_rate": 1.870630227320294e-05, "loss": 1.4446, "step": 7552 }, { "epoch": 18.8825, "grad_norm": 5.40625, "learning_rate": 1.8705903927952097e-05, "loss": 1.3356, "step": 7553 }, { "epoch": 18.885, "grad_norm": 5.65625, "learning_rate": 1.8705505525625695e-05, "loss": 1.4453, "step": 7554 }, { "epoch": 18.8875, "grad_norm": 8.125, "learning_rate": 1.8705107066226342e-05, "loss": 1.4438, "step": 7555 }, { "epoch": 18.89, "grad_norm": 14.9375, "learning_rate": 1.8704708549756657e-05, "loss": 1.4122, "step": 7556 }, { "epoch": 18.8925, "grad_norm": 7.0625, "learning_rate": 1.8704309976219256e-05, "loss": 1.5976, "step": 7557 }, { "epoch": 18.895, "grad_norm": 5.625, "learning_rate": 1.870391134561674e-05, "loss": 1.4867, "step": 7558 }, { "epoch": 18.8975, "grad_norm": 8.3125, "learning_rate": 1.8703512657951735e-05, "loss": 1.4869, "step": 7559 }, { "epoch": 18.9, "grad_norm": 7.59375, "learning_rate": 1.8703113913226847e-05, "loss": 1.6165, "step": 7560 }, { "epoch": 18.9025, "grad_norm": 5.71875, "learning_rate": 1.870271511144469e-05, "loss": 1.4656, "step": 7561 }, { "epoch": 18.905, "grad_norm": 6.9375, "learning_rate": 1.8702316252607887e-05, "loss": 1.4102, "step": 7562 }, { "epoch": 18.9075, "grad_norm": 6.34375, "learning_rate": 1.870191733671904e-05, "loss": 1.5185, "step": 7563 }, { "epoch": 18.91, "grad_norm": 7.625, "learning_rate": 1.8701518363780776e-05, "loss": 1.6437, "step": 7564 }, { "epoch": 18.9125, "grad_norm": 6.15625, "learning_rate": 1.8701119333795703e-05, "loss": 1.4613, "step": 7565 }, { "epoch": 18.915, "grad_norm": 5.6875, "learning_rate": 1.8700720246766444e-05, "loss": 1.3258, "step": 7566 }, { "epoch": 18.9175, "grad_norm": 6.6875, "learning_rate": 1.8700321102695608e-05, "loss": 1.3216, "step": 7567 }, { "epoch": 18.92, "grad_norm": 7.40625, "learning_rate": 1.8699921901585814e-05, "loss": 1.4808, "step": 7568 }, { "epoch": 18.9225, "grad_norm": 6.46875, "learning_rate": 1.8699522643439684e-05, "loss": 1.4871, "step": 7569 }, { "epoch": 18.925, "grad_norm": 7.5625, "learning_rate": 1.8699123328259828e-05, "loss": 1.357, "step": 7570 }, { "epoch": 18.9275, "grad_norm": 7.625, "learning_rate": 1.8698723956048866e-05, "loss": 1.565, "step": 7571 }, { "epoch": 18.93, "grad_norm": 7.28125, "learning_rate": 1.8698324526809423e-05, "loss": 1.5037, "step": 7572 }, { "epoch": 18.9325, "grad_norm": 8.4375, "learning_rate": 1.869792504054411e-05, "loss": 1.5315, "step": 7573 }, { "epoch": 18.935, "grad_norm": 6.375, "learning_rate": 1.869752549725555e-05, "loss": 1.3535, "step": 7574 }, { "epoch": 18.9375, "grad_norm": 6.25, "learning_rate": 1.869712589694636e-05, "loss": 1.5152, "step": 7575 }, { "epoch": 18.94, "grad_norm": 7.1875, "learning_rate": 1.869672623961916e-05, "loss": 1.4904, "step": 7576 }, { "epoch": 18.9425, "grad_norm": 8.25, "learning_rate": 1.8696326525276573e-05, "loss": 1.3896, "step": 7577 }, { "epoch": 18.945, "grad_norm": 5.8125, "learning_rate": 1.8695926753921214e-05, "loss": 1.4402, "step": 7578 }, { "epoch": 18.9475, "grad_norm": 7.4375, "learning_rate": 1.869552692555571e-05, "loss": 1.447, "step": 7579 }, { "epoch": 18.95, "grad_norm": 11.875, "learning_rate": 1.8695127040182678e-05, "loss": 1.4689, "step": 7580 }, { "epoch": 18.9525, "grad_norm": 7.21875, "learning_rate": 1.8694727097804743e-05, "loss": 1.4757, "step": 7581 }, { "epoch": 18.955, "grad_norm": 6.21875, "learning_rate": 1.869432709842452e-05, "loss": 1.45, "step": 7582 }, { "epoch": 18.9575, "grad_norm": 6.09375, "learning_rate": 1.8693927042044643e-05, "loss": 1.415, "step": 7583 }, { "epoch": 18.96, "grad_norm": 7.0, "learning_rate": 1.8693526928667724e-05, "loss": 1.3291, "step": 7584 }, { "epoch": 18.9625, "grad_norm": 6.84375, "learning_rate": 1.869312675829639e-05, "loss": 1.3491, "step": 7585 }, { "epoch": 18.965, "grad_norm": 9.0, "learning_rate": 1.8692726530933268e-05, "loss": 1.2903, "step": 7586 }, { "epoch": 18.9675, "grad_norm": 6.4375, "learning_rate": 1.8692326246580973e-05, "loss": 1.5509, "step": 7587 }, { "epoch": 18.97, "grad_norm": 4.9375, "learning_rate": 1.869192590524214e-05, "loss": 1.3708, "step": 7588 }, { "epoch": 18.9725, "grad_norm": 5.9375, "learning_rate": 1.8691525506919387e-05, "loss": 1.4308, "step": 7589 }, { "epoch": 18.975, "grad_norm": 4.8125, "learning_rate": 1.869112505161534e-05, "loss": 1.6094, "step": 7590 }, { "epoch": 18.9775, "grad_norm": 8.5625, "learning_rate": 1.8690724539332626e-05, "loss": 1.3581, "step": 7591 }, { "epoch": 18.98, "grad_norm": 10.125, "learning_rate": 1.8690323970073874e-05, "loss": 1.5204, "step": 7592 }, { "epoch": 18.9825, "grad_norm": 6.8125, "learning_rate": 1.86899233438417e-05, "loss": 1.2387, "step": 7593 }, { "epoch": 18.985, "grad_norm": 6.8125, "learning_rate": 1.868952266063874e-05, "loss": 1.4481, "step": 7594 }, { "epoch": 18.9875, "grad_norm": 5.75, "learning_rate": 1.868912192046761e-05, "loss": 1.5102, "step": 7595 }, { "epoch": 18.99, "grad_norm": 6.8125, "learning_rate": 1.8688721123330952e-05, "loss": 1.5221, "step": 7596 }, { "epoch": 18.9925, "grad_norm": 7.53125, "learning_rate": 1.8688320269231384e-05, "loss": 1.4053, "step": 7597 }, { "epoch": 18.995, "grad_norm": 6.90625, "learning_rate": 1.8687919358171537e-05, "loss": 1.5201, "step": 7598 }, { "epoch": 18.9975, "grad_norm": 9.4375, "learning_rate": 1.8687518390154037e-05, "loss": 1.4614, "step": 7599 }, { "epoch": 19.0, "grad_norm": 7.875, "learning_rate": 1.8687117365181514e-05, "loss": 1.5595, "step": 7600 }, { "epoch": 19.0025, "grad_norm": 8.25, "learning_rate": 1.8686716283256595e-05, "loss": 1.287, "step": 7601 }, { "epoch": 19.005, "grad_norm": 8.0, "learning_rate": 1.8686315144381914e-05, "loss": 1.3572, "step": 7602 }, { "epoch": 19.0075, "grad_norm": 6.0, "learning_rate": 1.8685913948560097e-05, "loss": 1.321, "step": 7603 }, { "epoch": 19.01, "grad_norm": 8.3125, "learning_rate": 1.8685512695793777e-05, "loss": 1.3277, "step": 7604 }, { "epoch": 19.0125, "grad_norm": 9.125, "learning_rate": 1.8685111386085585e-05, "loss": 1.4684, "step": 7605 }, { "epoch": 19.015, "grad_norm": 8.3125, "learning_rate": 1.8684710019438147e-05, "loss": 1.3413, "step": 7606 }, { "epoch": 19.0175, "grad_norm": 6.8125, "learning_rate": 1.86843085958541e-05, "loss": 1.3866, "step": 7607 }, { "epoch": 19.02, "grad_norm": 10.0, "learning_rate": 1.8683907115336073e-05, "loss": 1.347, "step": 7608 }, { "epoch": 19.0225, "grad_norm": 8.375, "learning_rate": 1.86835055778867e-05, "loss": 1.4929, "step": 7609 }, { "epoch": 19.025, "grad_norm": 8.1875, "learning_rate": 1.868310398350861e-05, "loss": 1.3168, "step": 7610 }, { "epoch": 19.0275, "grad_norm": 9.0625, "learning_rate": 1.868270233220444e-05, "loss": 1.367, "step": 7611 }, { "epoch": 19.03, "grad_norm": 8.375, "learning_rate": 1.8682300623976815e-05, "loss": 1.3136, "step": 7612 }, { "epoch": 19.0325, "grad_norm": 7.5625, "learning_rate": 1.868189885882838e-05, "loss": 1.4278, "step": 7613 }, { "epoch": 19.035, "grad_norm": 6.15625, "learning_rate": 1.868149703676176e-05, "loss": 1.389, "step": 7614 }, { "epoch": 19.0375, "grad_norm": 6.09375, "learning_rate": 1.8681095157779595e-05, "loss": 1.5455, "step": 7615 }, { "epoch": 19.04, "grad_norm": 7.96875, "learning_rate": 1.868069322188452e-05, "loss": 1.4969, "step": 7616 }, { "epoch": 19.0425, "grad_norm": 8.5, "learning_rate": 1.8680291229079163e-05, "loss": 1.3548, "step": 7617 }, { "epoch": 19.045, "grad_norm": 9.3125, "learning_rate": 1.8679889179366166e-05, "loss": 1.3726, "step": 7618 }, { "epoch": 19.0475, "grad_norm": 15.6875, "learning_rate": 1.8679487072748164e-05, "loss": 1.3527, "step": 7619 }, { "epoch": 19.05, "grad_norm": 7.125, "learning_rate": 1.867908490922779e-05, "loss": 1.3767, "step": 7620 }, { "epoch": 19.0525, "grad_norm": 5.78125, "learning_rate": 1.8678682688807686e-05, "loss": 1.2122, "step": 7621 }, { "epoch": 19.055, "grad_norm": 5.40625, "learning_rate": 1.8678280411490482e-05, "loss": 1.2838, "step": 7622 }, { "epoch": 19.0575, "grad_norm": 7.3125, "learning_rate": 1.867787807727882e-05, "loss": 1.3497, "step": 7623 }, { "epoch": 19.06, "grad_norm": 7.34375, "learning_rate": 1.8677475686175336e-05, "loss": 1.3786, "step": 7624 }, { "epoch": 19.0625, "grad_norm": 11.5, "learning_rate": 1.8677073238182667e-05, "loss": 1.4104, "step": 7625 }, { "epoch": 19.065, "grad_norm": 5.28125, "learning_rate": 1.8676670733303458e-05, "loss": 1.4029, "step": 7626 }, { "epoch": 19.0675, "grad_norm": 7.28125, "learning_rate": 1.867626817154034e-05, "loss": 1.4814, "step": 7627 }, { "epoch": 19.07, "grad_norm": 8.6875, "learning_rate": 1.8675865552895955e-05, "loss": 1.4138, "step": 7628 }, { "epoch": 19.0725, "grad_norm": 8.3125, "learning_rate": 1.8675462877372946e-05, "loss": 1.5226, "step": 7629 }, { "epoch": 19.075, "grad_norm": 5.9375, "learning_rate": 1.8675060144973945e-05, "loss": 1.3096, "step": 7630 }, { "epoch": 19.0775, "grad_norm": 4.78125, "learning_rate": 1.8674657355701598e-05, "loss": 1.4927, "step": 7631 }, { "epoch": 19.08, "grad_norm": 11.1875, "learning_rate": 1.8674254509558544e-05, "loss": 1.2739, "step": 7632 }, { "epoch": 19.0825, "grad_norm": 5.875, "learning_rate": 1.8673851606547426e-05, "loss": 1.4479, "step": 7633 }, { "epoch": 19.085, "grad_norm": 6.09375, "learning_rate": 1.8673448646670883e-05, "loss": 1.4086, "step": 7634 }, { "epoch": 19.0875, "grad_norm": 11.0625, "learning_rate": 1.8673045629931558e-05, "loss": 1.3416, "step": 7635 }, { "epoch": 19.09, "grad_norm": 9.875, "learning_rate": 1.8672642556332093e-05, "loss": 1.4374, "step": 7636 }, { "epoch": 19.0925, "grad_norm": 9.125, "learning_rate": 1.8672239425875132e-05, "loss": 1.4363, "step": 7637 }, { "epoch": 19.095, "grad_norm": 6.28125, "learning_rate": 1.8671836238563314e-05, "loss": 1.2964, "step": 7638 }, { "epoch": 19.0975, "grad_norm": 8.9375, "learning_rate": 1.8671432994399285e-05, "loss": 1.2974, "step": 7639 }, { "epoch": 19.1, "grad_norm": 6.78125, "learning_rate": 1.867102969338569e-05, "loss": 1.164, "step": 7640 }, { "epoch": 19.1025, "grad_norm": 7.625, "learning_rate": 1.867062633552517e-05, "loss": 1.453, "step": 7641 }, { "epoch": 19.105, "grad_norm": 8.3125, "learning_rate": 1.8670222920820372e-05, "loss": 1.3833, "step": 7642 }, { "epoch": 19.1075, "grad_norm": 7.09375, "learning_rate": 1.866981944927394e-05, "loss": 1.4441, "step": 7643 }, { "epoch": 19.11, "grad_norm": 10.125, "learning_rate": 1.8669415920888517e-05, "loss": 1.4583, "step": 7644 }, { "epoch": 19.1125, "grad_norm": 8.75, "learning_rate": 1.866901233566675e-05, "loss": 1.3471, "step": 7645 }, { "epoch": 19.115, "grad_norm": 7.34375, "learning_rate": 1.8668608693611284e-05, "loss": 1.4776, "step": 7646 }, { "epoch": 19.1175, "grad_norm": 7.375, "learning_rate": 1.866820499472477e-05, "loss": 1.3643, "step": 7647 }, { "epoch": 19.12, "grad_norm": 4.21875, "learning_rate": 1.8667801239009845e-05, "loss": 1.3883, "step": 7648 }, { "epoch": 19.1225, "grad_norm": 8.0625, "learning_rate": 1.8667397426469168e-05, "loss": 1.4393, "step": 7649 }, { "epoch": 19.125, "grad_norm": 5.0625, "learning_rate": 1.8666993557105377e-05, "loss": 1.3375, "step": 7650 }, { "epoch": 19.1275, "grad_norm": 6.6875, "learning_rate": 1.8666589630921125e-05, "loss": 1.2778, "step": 7651 }, { "epoch": 19.13, "grad_norm": 6.1875, "learning_rate": 1.866618564791906e-05, "loss": 1.2662, "step": 7652 }, { "epoch": 19.1325, "grad_norm": 6.90625, "learning_rate": 1.8665781608101824e-05, "loss": 1.3033, "step": 7653 }, { "epoch": 19.135, "grad_norm": 6.53125, "learning_rate": 1.866537751147207e-05, "loss": 1.2326, "step": 7654 }, { "epoch": 19.1375, "grad_norm": 5.96875, "learning_rate": 1.8664973358032453e-05, "loss": 1.3995, "step": 7655 }, { "epoch": 19.14, "grad_norm": 7.40625, "learning_rate": 1.8664569147785615e-05, "loss": 1.5403, "step": 7656 }, { "epoch": 19.1425, "grad_norm": 7.6875, "learning_rate": 1.8664164880734204e-05, "loss": 1.4348, "step": 7657 }, { "epoch": 19.145, "grad_norm": 6.25, "learning_rate": 1.866376055688088e-05, "loss": 1.4152, "step": 7658 }, { "epoch": 19.1475, "grad_norm": 5.875, "learning_rate": 1.8663356176228286e-05, "loss": 1.7017, "step": 7659 }, { "epoch": 19.15, "grad_norm": 7.4375, "learning_rate": 1.8662951738779077e-05, "loss": 1.4704, "step": 7660 }, { "epoch": 19.1525, "grad_norm": 6.78125, "learning_rate": 1.8662547244535906e-05, "loss": 1.4783, "step": 7661 }, { "epoch": 19.155, "grad_norm": 7.21875, "learning_rate": 1.8662142693501416e-05, "loss": 1.437, "step": 7662 }, { "epoch": 19.1575, "grad_norm": 5.75, "learning_rate": 1.866173808567827e-05, "loss": 1.3763, "step": 7663 }, { "epoch": 19.16, "grad_norm": 7.0, "learning_rate": 1.866133342106911e-05, "loss": 1.6056, "step": 7664 }, { "epoch": 19.1625, "grad_norm": 5.53125, "learning_rate": 1.86609286996766e-05, "loss": 1.4645, "step": 7665 }, { "epoch": 19.165, "grad_norm": 7.15625, "learning_rate": 1.8660523921503388e-05, "loss": 1.3097, "step": 7666 }, { "epoch": 19.1675, "grad_norm": 8.1875, "learning_rate": 1.8660119086552127e-05, "loss": 1.4492, "step": 7667 }, { "epoch": 19.17, "grad_norm": 5.15625, "learning_rate": 1.865971419482547e-05, "loss": 1.3507, "step": 7668 }, { "epoch": 19.1725, "grad_norm": 12.6875, "learning_rate": 1.8659309246326072e-05, "loss": 1.5077, "step": 7669 }, { "epoch": 19.175, "grad_norm": 5.40625, "learning_rate": 1.8658904241056594e-05, "loss": 1.2867, "step": 7670 }, { "epoch": 19.1775, "grad_norm": 7.25, "learning_rate": 1.8658499179019683e-05, "loss": 1.4719, "step": 7671 }, { "epoch": 19.18, "grad_norm": 7.0625, "learning_rate": 1.8658094060218e-05, "loss": 1.4876, "step": 7672 }, { "epoch": 19.1825, "grad_norm": 8.25, "learning_rate": 1.8657688884654197e-05, "loss": 1.379, "step": 7673 }, { "epoch": 19.185, "grad_norm": 9.9375, "learning_rate": 1.865728365233093e-05, "loss": 1.4839, "step": 7674 }, { "epoch": 19.1875, "grad_norm": 5.65625, "learning_rate": 1.865687836325086e-05, "loss": 1.366, "step": 7675 }, { "epoch": 19.19, "grad_norm": 8.125, "learning_rate": 1.8656473017416644e-05, "loss": 1.3753, "step": 7676 }, { "epoch": 19.1925, "grad_norm": 6.40625, "learning_rate": 1.8656067614830936e-05, "loss": 1.2625, "step": 7677 }, { "epoch": 19.195, "grad_norm": 6.25, "learning_rate": 1.8655662155496395e-05, "loss": 1.4856, "step": 7678 }, { "epoch": 19.1975, "grad_norm": 9.3125, "learning_rate": 1.865525663941568e-05, "loss": 1.5642, "step": 7679 }, { "epoch": 19.2, "grad_norm": 9.375, "learning_rate": 1.865485106659145e-05, "loss": 1.3271, "step": 7680 }, { "epoch": 19.2025, "grad_norm": 8.1875, "learning_rate": 1.8654445437026356e-05, "loss": 1.4132, "step": 7681 }, { "epoch": 19.205, "grad_norm": 12.75, "learning_rate": 1.865403975072307e-05, "loss": 1.4413, "step": 7682 }, { "epoch": 19.2075, "grad_norm": 7.09375, "learning_rate": 1.8653634007684242e-05, "loss": 1.2563, "step": 7683 }, { "epoch": 19.21, "grad_norm": 6.59375, "learning_rate": 1.865322820791254e-05, "loss": 1.5479, "step": 7684 }, { "epoch": 19.2125, "grad_norm": 6.8125, "learning_rate": 1.865282235141062e-05, "loss": 1.523, "step": 7685 }, { "epoch": 19.215, "grad_norm": 7.96875, "learning_rate": 1.865241643818114e-05, "loss": 1.3415, "step": 7686 }, { "epoch": 19.2175, "grad_norm": 7.09375, "learning_rate": 1.8652010468226763e-05, "loss": 1.2902, "step": 7687 }, { "epoch": 19.22, "grad_norm": 6.96875, "learning_rate": 1.8651604441550154e-05, "loss": 1.538, "step": 7688 }, { "epoch": 19.2225, "grad_norm": 6.90625, "learning_rate": 1.8651198358153973e-05, "loss": 1.4216, "step": 7689 }, { "epoch": 19.225, "grad_norm": 10.8125, "learning_rate": 1.8650792218040883e-05, "loss": 1.3665, "step": 7690 }, { "epoch": 19.2275, "grad_norm": 6.28125, "learning_rate": 1.8650386021213542e-05, "loss": 1.3957, "step": 7691 }, { "epoch": 19.23, "grad_norm": 8.5, "learning_rate": 1.864997976767462e-05, "loss": 1.4552, "step": 7692 }, { "epoch": 19.2325, "grad_norm": 8.5625, "learning_rate": 1.864957345742677e-05, "loss": 1.2959, "step": 7693 }, { "epoch": 19.235, "grad_norm": 7.28125, "learning_rate": 1.864916709047267e-05, "loss": 1.344, "step": 7694 }, { "epoch": 19.2375, "grad_norm": 5.78125, "learning_rate": 1.8648760666814976e-05, "loss": 1.5788, "step": 7695 }, { "epoch": 19.24, "grad_norm": 5.34375, "learning_rate": 1.864835418645635e-05, "loss": 1.4519, "step": 7696 }, { "epoch": 19.2425, "grad_norm": 5.59375, "learning_rate": 1.864794764939946e-05, "loss": 1.5163, "step": 7697 }, { "epoch": 19.245, "grad_norm": 8.1875, "learning_rate": 1.864754105564697e-05, "loss": 1.5177, "step": 7698 }, { "epoch": 19.2475, "grad_norm": 5.375, "learning_rate": 1.8647134405201554e-05, "loss": 1.2274, "step": 7699 }, { "epoch": 19.25, "grad_norm": 8.0, "learning_rate": 1.8646727698065865e-05, "loss": 1.3766, "step": 7700 }, { "epoch": 19.2525, "grad_norm": 7.21875, "learning_rate": 1.8646320934242578e-05, "loss": 1.4996, "step": 7701 }, { "epoch": 19.255, "grad_norm": 6.21875, "learning_rate": 1.8645914113734355e-05, "loss": 1.4611, "step": 7702 }, { "epoch": 19.2575, "grad_norm": 5.46875, "learning_rate": 1.8645507236543863e-05, "loss": 1.6177, "step": 7703 }, { "epoch": 19.26, "grad_norm": 7.25, "learning_rate": 1.8645100302673775e-05, "loss": 1.4304, "step": 7704 }, { "epoch": 19.2625, "grad_norm": 4.65625, "learning_rate": 1.8644693312126754e-05, "loss": 1.4861, "step": 7705 }, { "epoch": 19.265, "grad_norm": 4.84375, "learning_rate": 1.864428626490547e-05, "loss": 1.4172, "step": 7706 }, { "epoch": 19.2675, "grad_norm": 6.5, "learning_rate": 1.8643879161012593e-05, "loss": 1.3828, "step": 7707 }, { "epoch": 19.27, "grad_norm": 7.5625, "learning_rate": 1.8643472000450786e-05, "loss": 1.5729, "step": 7708 }, { "epoch": 19.2725, "grad_norm": 7.5625, "learning_rate": 1.8643064783222726e-05, "loss": 1.4455, "step": 7709 }, { "epoch": 19.275, "grad_norm": 8.75, "learning_rate": 1.8642657509331078e-05, "loss": 1.4563, "step": 7710 }, { "epoch": 19.2775, "grad_norm": 9.4375, "learning_rate": 1.864225017877851e-05, "loss": 1.6542, "step": 7711 }, { "epoch": 19.28, "grad_norm": 5.46875, "learning_rate": 1.86418427915677e-05, "loss": 1.5162, "step": 7712 }, { "epoch": 19.2825, "grad_norm": 6.125, "learning_rate": 1.8641435347701312e-05, "loss": 1.4441, "step": 7713 }, { "epoch": 19.285, "grad_norm": 9.1875, "learning_rate": 1.8641027847182022e-05, "loss": 1.299, "step": 7714 }, { "epoch": 19.2875, "grad_norm": 9.875, "learning_rate": 1.86406202900125e-05, "loss": 1.5075, "step": 7715 }, { "epoch": 19.29, "grad_norm": 6.65625, "learning_rate": 1.8640212676195415e-05, "loss": 1.6123, "step": 7716 }, { "epoch": 19.2925, "grad_norm": 6.59375, "learning_rate": 1.863980500573344e-05, "loss": 1.4845, "step": 7717 }, { "epoch": 19.295, "grad_norm": 5.15625, "learning_rate": 1.8639397278629253e-05, "loss": 1.2844, "step": 7718 }, { "epoch": 19.2975, "grad_norm": 8.0625, "learning_rate": 1.863898949488552e-05, "loss": 1.5153, "step": 7719 }, { "epoch": 19.3, "grad_norm": 9.0, "learning_rate": 1.863858165450492e-05, "loss": 1.4888, "step": 7720 }, { "epoch": 19.3025, "grad_norm": 16.375, "learning_rate": 1.863817375749012e-05, "loss": 1.4853, "step": 7721 }, { "epoch": 19.305, "grad_norm": 5.46875, "learning_rate": 1.8637765803843805e-05, "loss": 1.2377, "step": 7722 }, { "epoch": 19.3075, "grad_norm": 7.15625, "learning_rate": 1.863735779356864e-05, "loss": 1.4048, "step": 7723 }, { "epoch": 19.31, "grad_norm": 7.40625, "learning_rate": 1.86369497266673e-05, "loss": 1.4406, "step": 7724 }, { "epoch": 19.3125, "grad_norm": 7.125, "learning_rate": 1.8636541603142467e-05, "loss": 1.528, "step": 7725 }, { "epoch": 19.315, "grad_norm": 5.59375, "learning_rate": 1.8636133422996812e-05, "loss": 1.297, "step": 7726 }, { "epoch": 19.3175, "grad_norm": 11.5, "learning_rate": 1.8635725186233008e-05, "loss": 1.5591, "step": 7727 }, { "epoch": 19.32, "grad_norm": 5.03125, "learning_rate": 1.863531689285374e-05, "loss": 1.2536, "step": 7728 }, { "epoch": 19.3225, "grad_norm": 7.09375, "learning_rate": 1.863490854286168e-05, "loss": 1.6162, "step": 7729 }, { "epoch": 19.325, "grad_norm": 6.75, "learning_rate": 1.8634500136259503e-05, "loss": 1.3229, "step": 7730 }, { "epoch": 19.3275, "grad_norm": 5.09375, "learning_rate": 1.863409167304989e-05, "loss": 1.2359, "step": 7731 }, { "epoch": 19.33, "grad_norm": 8.8125, "learning_rate": 1.863368315323552e-05, "loss": 1.4493, "step": 7732 }, { "epoch": 19.3325, "grad_norm": 8.5, "learning_rate": 1.863327457681906e-05, "loss": 1.4877, "step": 7733 }, { "epoch": 19.335, "grad_norm": 6.1875, "learning_rate": 1.8632865943803206e-05, "loss": 1.426, "step": 7734 }, { "epoch": 19.3375, "grad_norm": 13.1875, "learning_rate": 1.8632457254190627e-05, "loss": 1.4693, "step": 7735 }, { "epoch": 19.34, "grad_norm": 7.15625, "learning_rate": 1.8632048507984e-05, "loss": 1.5704, "step": 7736 }, { "epoch": 19.3425, "grad_norm": 7.9375, "learning_rate": 1.8631639705186014e-05, "loss": 1.3824, "step": 7737 }, { "epoch": 19.345, "grad_norm": 5.875, "learning_rate": 1.8631230845799336e-05, "loss": 1.4277, "step": 7738 }, { "epoch": 19.3475, "grad_norm": 6.5625, "learning_rate": 1.863082192982666e-05, "loss": 1.3954, "step": 7739 }, { "epoch": 19.35, "grad_norm": 7.03125, "learning_rate": 1.863041295727066e-05, "loss": 1.4083, "step": 7740 }, { "epoch": 19.3525, "grad_norm": 7.65625, "learning_rate": 1.8630003928134015e-05, "loss": 1.4835, "step": 7741 }, { "epoch": 19.355, "grad_norm": 9.25, "learning_rate": 1.8629594842419415e-05, "loss": 1.4937, "step": 7742 }, { "epoch": 19.3575, "grad_norm": 6.03125, "learning_rate": 1.862918570012953e-05, "loss": 1.2173, "step": 7743 }, { "epoch": 19.36, "grad_norm": 6.34375, "learning_rate": 1.8628776501267052e-05, "loss": 1.3647, "step": 7744 }, { "epoch": 19.3625, "grad_norm": 6.375, "learning_rate": 1.862836724583466e-05, "loss": 1.2212, "step": 7745 }, { "epoch": 19.365, "grad_norm": 5.0, "learning_rate": 1.8627957933835038e-05, "loss": 1.404, "step": 7746 }, { "epoch": 19.3675, "grad_norm": 5.0, "learning_rate": 1.8627548565270868e-05, "loss": 1.4696, "step": 7747 }, { "epoch": 19.37, "grad_norm": 8.0625, "learning_rate": 1.862713914014484e-05, "loss": 1.461, "step": 7748 }, { "epoch": 19.3725, "grad_norm": 6.71875, "learning_rate": 1.8626729658459627e-05, "loss": 1.5934, "step": 7749 }, { "epoch": 19.375, "grad_norm": 7.5, "learning_rate": 1.8626320120217922e-05, "loss": 1.4451, "step": 7750 }, { "epoch": 19.3775, "grad_norm": 6.875, "learning_rate": 1.862591052542241e-05, "loss": 1.4768, "step": 7751 }, { "epoch": 19.38, "grad_norm": 7.34375, "learning_rate": 1.862550087407577e-05, "loss": 1.6555, "step": 7752 }, { "epoch": 19.3825, "grad_norm": 8.875, "learning_rate": 1.8625091166180693e-05, "loss": 1.3806, "step": 7753 }, { "epoch": 19.385, "grad_norm": 6.21875, "learning_rate": 1.8624681401739862e-05, "loss": 1.4123, "step": 7754 }, { "epoch": 19.3875, "grad_norm": 21.375, "learning_rate": 1.8624271580755965e-05, "loss": 1.3358, "step": 7755 }, { "epoch": 19.39, "grad_norm": 7.6875, "learning_rate": 1.862386170323169e-05, "loss": 1.5182, "step": 7756 }, { "epoch": 19.3925, "grad_norm": 11.5, "learning_rate": 1.8623451769169723e-05, "loss": 1.5086, "step": 7757 }, { "epoch": 19.395, "grad_norm": 7.75, "learning_rate": 1.862304177857275e-05, "loss": 1.4312, "step": 7758 }, { "epoch": 19.3975, "grad_norm": 4.90625, "learning_rate": 1.862263173144346e-05, "loss": 1.3937, "step": 7759 }, { "epoch": 19.4, "grad_norm": 12.25, "learning_rate": 1.862222162778454e-05, "loss": 1.4966, "step": 7760 }, { "epoch": 19.4025, "grad_norm": 9.8125, "learning_rate": 1.8621811467598686e-05, "loss": 1.4522, "step": 7761 }, { "epoch": 19.405, "grad_norm": 8.375, "learning_rate": 1.8621401250888574e-05, "loss": 1.5054, "step": 7762 }, { "epoch": 19.4075, "grad_norm": 4.90625, "learning_rate": 1.8620990977656906e-05, "loss": 1.5526, "step": 7763 }, { "epoch": 19.41, "grad_norm": 7.78125, "learning_rate": 1.8620580647906365e-05, "loss": 1.2621, "step": 7764 }, { "epoch": 19.4125, "grad_norm": 9.5625, "learning_rate": 1.8620170261639637e-05, "loss": 1.473, "step": 7765 }, { "epoch": 19.415, "grad_norm": 6.9375, "learning_rate": 1.861975981885942e-05, "loss": 1.3947, "step": 7766 }, { "epoch": 19.4175, "grad_norm": 6.96875, "learning_rate": 1.8619349319568408e-05, "loss": 1.4337, "step": 7767 }, { "epoch": 19.42, "grad_norm": 8.125, "learning_rate": 1.8618938763769283e-05, "loss": 1.38, "step": 7768 }, { "epoch": 19.4225, "grad_norm": 5.1875, "learning_rate": 1.861852815146474e-05, "loss": 1.3495, "step": 7769 }, { "epoch": 19.425, "grad_norm": 6.8125, "learning_rate": 1.8618117482657473e-05, "loss": 1.4985, "step": 7770 }, { "epoch": 19.4275, "grad_norm": 6.90625, "learning_rate": 1.861770675735017e-05, "loss": 1.4703, "step": 7771 }, { "epoch": 19.43, "grad_norm": 6.6875, "learning_rate": 1.861729597554553e-05, "loss": 1.5362, "step": 7772 }, { "epoch": 19.4325, "grad_norm": 5.96875, "learning_rate": 1.861688513724624e-05, "loss": 1.2452, "step": 7773 }, { "epoch": 19.435, "grad_norm": 6.375, "learning_rate": 1.8616474242454995e-05, "loss": 1.4487, "step": 7774 }, { "epoch": 19.4375, "grad_norm": 5.375, "learning_rate": 1.8616063291174492e-05, "loss": 1.4039, "step": 7775 }, { "epoch": 19.44, "grad_norm": 5.03125, "learning_rate": 1.861565228340742e-05, "loss": 1.338, "step": 7776 }, { "epoch": 19.4425, "grad_norm": 6.28125, "learning_rate": 1.8615241219156476e-05, "loss": 1.4195, "step": 7777 }, { "epoch": 19.445, "grad_norm": 5.40625, "learning_rate": 1.861483009842436e-05, "loss": 1.3565, "step": 7778 }, { "epoch": 19.4475, "grad_norm": 10.125, "learning_rate": 1.8614418921213757e-05, "loss": 1.3761, "step": 7779 }, { "epoch": 19.45, "grad_norm": 7.125, "learning_rate": 1.8614007687527374e-05, "loss": 1.3863, "step": 7780 }, { "epoch": 19.4525, "grad_norm": 7.125, "learning_rate": 1.8613596397367896e-05, "loss": 1.4391, "step": 7781 }, { "epoch": 19.455, "grad_norm": 8.6875, "learning_rate": 1.8613185050738028e-05, "loss": 1.4633, "step": 7782 }, { "epoch": 19.4575, "grad_norm": 6.71875, "learning_rate": 1.861277364764046e-05, "loss": 1.4353, "step": 7783 }, { "epoch": 19.46, "grad_norm": 4.65625, "learning_rate": 1.86123621880779e-05, "loss": 1.4214, "step": 7784 }, { "epoch": 19.4625, "grad_norm": 5.09375, "learning_rate": 1.861195067205303e-05, "loss": 1.3498, "step": 7785 }, { "epoch": 19.465, "grad_norm": 7.09375, "learning_rate": 1.861153909956856e-05, "loss": 1.4522, "step": 7786 }, { "epoch": 19.4675, "grad_norm": 5.3125, "learning_rate": 1.8611127470627185e-05, "loss": 1.3709, "step": 7787 }, { "epoch": 19.47, "grad_norm": 5.84375, "learning_rate": 1.8610715785231603e-05, "loss": 1.3749, "step": 7788 }, { "epoch": 19.4725, "grad_norm": 7.03125, "learning_rate": 1.8610304043384514e-05, "loss": 1.4133, "step": 7789 }, { "epoch": 19.475, "grad_norm": 7.84375, "learning_rate": 1.860989224508861e-05, "loss": 1.5951, "step": 7790 }, { "epoch": 19.4775, "grad_norm": 4.28125, "learning_rate": 1.8609480390346603e-05, "loss": 1.3017, "step": 7791 }, { "epoch": 19.48, "grad_norm": 6.625, "learning_rate": 1.8609068479161182e-05, "loss": 1.3992, "step": 7792 }, { "epoch": 19.4825, "grad_norm": 6.4375, "learning_rate": 1.860865651153506e-05, "loss": 1.5116, "step": 7793 }, { "epoch": 19.485, "grad_norm": 7.34375, "learning_rate": 1.8608244487470923e-05, "loss": 1.3044, "step": 7794 }, { "epoch": 19.4875, "grad_norm": 7.3125, "learning_rate": 1.8607832406971483e-05, "loss": 1.3984, "step": 7795 }, { "epoch": 19.49, "grad_norm": 9.0, "learning_rate": 1.860742027003944e-05, "loss": 1.3802, "step": 7796 }, { "epoch": 19.4925, "grad_norm": 6.25, "learning_rate": 1.860700807667749e-05, "loss": 1.2793, "step": 7797 }, { "epoch": 19.495, "grad_norm": 8.125, "learning_rate": 1.8606595826888344e-05, "loss": 1.514, "step": 7798 }, { "epoch": 19.4975, "grad_norm": 7.3125, "learning_rate": 1.8606183520674695e-05, "loss": 1.5712, "step": 7799 }, { "epoch": 19.5, "grad_norm": 7.75, "learning_rate": 1.8605771158039253e-05, "loss": 1.4552, "step": 7800 }, { "epoch": 19.5025, "grad_norm": 7.375, "learning_rate": 1.860535873898472e-05, "loss": 1.4617, "step": 7801 }, { "epoch": 19.505, "grad_norm": 7.46875, "learning_rate": 1.86049462635138e-05, "loss": 1.3336, "step": 7802 }, { "epoch": 19.5075, "grad_norm": 5.65625, "learning_rate": 1.86045337316292e-05, "loss": 1.195, "step": 7803 }, { "epoch": 19.51, "grad_norm": 9.125, "learning_rate": 1.8604121143333615e-05, "loss": 1.5694, "step": 7804 }, { "epoch": 19.5125, "grad_norm": 8.0, "learning_rate": 1.860370849862976e-05, "loss": 1.4105, "step": 7805 }, { "epoch": 19.515, "grad_norm": 4.8125, "learning_rate": 1.8603295797520336e-05, "loss": 1.5146, "step": 7806 }, { "epoch": 19.5175, "grad_norm": 7.75, "learning_rate": 1.8602883040008047e-05, "loss": 1.6262, "step": 7807 }, { "epoch": 19.52, "grad_norm": 10.6875, "learning_rate": 1.8602470226095602e-05, "loss": 1.5439, "step": 7808 }, { "epoch": 19.5225, "grad_norm": 7.15625, "learning_rate": 1.860205735578571e-05, "loss": 1.3128, "step": 7809 }, { "epoch": 19.525, "grad_norm": 7.0, "learning_rate": 1.860164442908107e-05, "loss": 1.4695, "step": 7810 }, { "epoch": 19.5275, "grad_norm": 7.4375, "learning_rate": 1.8601231445984394e-05, "loss": 1.5466, "step": 7811 }, { "epoch": 19.53, "grad_norm": 8.0625, "learning_rate": 1.8600818406498388e-05, "loss": 1.3749, "step": 7812 }, { "epoch": 19.5325, "grad_norm": 7.71875, "learning_rate": 1.860040531062576e-05, "loss": 1.4952, "step": 7813 }, { "epoch": 19.535, "grad_norm": 7.71875, "learning_rate": 1.8599992158369225e-05, "loss": 1.5777, "step": 7814 }, { "epoch": 19.5375, "grad_norm": 5.4375, "learning_rate": 1.8599578949731478e-05, "loss": 1.412, "step": 7815 }, { "epoch": 19.54, "grad_norm": 6.125, "learning_rate": 1.8599165684715238e-05, "loss": 1.427, "step": 7816 }, { "epoch": 19.5425, "grad_norm": 7.40625, "learning_rate": 1.8598752363323217e-05, "loss": 1.3813, "step": 7817 }, { "epoch": 19.545, "grad_norm": 7.21875, "learning_rate": 1.8598338985558113e-05, "loss": 1.371, "step": 7818 }, { "epoch": 19.5475, "grad_norm": 5.8125, "learning_rate": 1.8597925551422645e-05, "loss": 1.3086, "step": 7819 }, { "epoch": 19.55, "grad_norm": 6.84375, "learning_rate": 1.8597512060919523e-05, "loss": 1.2718, "step": 7820 }, { "epoch": 19.5525, "grad_norm": 9.75, "learning_rate": 1.859709851405145e-05, "loss": 1.4451, "step": 7821 }, { "epoch": 19.555, "grad_norm": 6.78125, "learning_rate": 1.8596684910821153e-05, "loss": 1.4006, "step": 7822 }, { "epoch": 19.5575, "grad_norm": 7.90625, "learning_rate": 1.8596271251231325e-05, "loss": 1.4629, "step": 7823 }, { "epoch": 19.56, "grad_norm": 6.6875, "learning_rate": 1.8595857535284692e-05, "loss": 1.4561, "step": 7824 }, { "epoch": 19.5625, "grad_norm": 7.4375, "learning_rate": 1.8595443762983958e-05, "loss": 1.4525, "step": 7825 }, { "epoch": 19.565, "grad_norm": 8.0625, "learning_rate": 1.8595029934331842e-05, "loss": 1.4761, "step": 7826 }, { "epoch": 19.5675, "grad_norm": 11.0625, "learning_rate": 1.859461604933105e-05, "loss": 1.3533, "step": 7827 }, { "epoch": 19.57, "grad_norm": 7.6875, "learning_rate": 1.85942021079843e-05, "loss": 1.4886, "step": 7828 }, { "epoch": 19.5725, "grad_norm": 8.375, "learning_rate": 1.8593788110294303e-05, "loss": 1.4341, "step": 7829 }, { "epoch": 19.575, "grad_norm": 13.375, "learning_rate": 1.859337405626378e-05, "loss": 1.3604, "step": 7830 }, { "epoch": 19.5775, "grad_norm": 8.5625, "learning_rate": 1.8592959945895435e-05, "loss": 1.4397, "step": 7831 }, { "epoch": 19.58, "grad_norm": 6.5, "learning_rate": 1.8592545779191993e-05, "loss": 1.3866, "step": 7832 }, { "epoch": 19.5825, "grad_norm": 5.84375, "learning_rate": 1.8592131556156164e-05, "loss": 1.3197, "step": 7833 }, { "epoch": 19.585, "grad_norm": 7.3125, "learning_rate": 1.8591717276790665e-05, "loss": 1.5539, "step": 7834 }, { "epoch": 19.5875, "grad_norm": 4.5, "learning_rate": 1.8591302941098207e-05, "loss": 1.2941, "step": 7835 }, { "epoch": 19.59, "grad_norm": 8.125, "learning_rate": 1.8590888549081514e-05, "loss": 1.5764, "step": 7836 }, { "epoch": 19.5925, "grad_norm": 5.4375, "learning_rate": 1.85904741007433e-05, "loss": 1.5146, "step": 7837 }, { "epoch": 19.595, "grad_norm": 6.40625, "learning_rate": 1.8590059596086282e-05, "loss": 1.4908, "step": 7838 }, { "epoch": 19.5975, "grad_norm": 5.84375, "learning_rate": 1.8589645035113174e-05, "loss": 1.4616, "step": 7839 }, { "epoch": 19.6, "grad_norm": 6.09375, "learning_rate": 1.85892304178267e-05, "loss": 1.2954, "step": 7840 }, { "epoch": 19.6025, "grad_norm": 7.0, "learning_rate": 1.858881574422957e-05, "loss": 1.4149, "step": 7841 }, { "epoch": 19.605, "grad_norm": 9.625, "learning_rate": 1.8588401014324512e-05, "loss": 1.4573, "step": 7842 }, { "epoch": 19.6075, "grad_norm": 5.125, "learning_rate": 1.858798622811424e-05, "loss": 1.2971, "step": 7843 }, { "epoch": 19.61, "grad_norm": 5.625, "learning_rate": 1.858757138560148e-05, "loss": 1.2954, "step": 7844 }, { "epoch": 19.6125, "grad_norm": 7.78125, "learning_rate": 1.8587156486788934e-05, "loss": 1.4292, "step": 7845 }, { "epoch": 19.615, "grad_norm": 5.28125, "learning_rate": 1.8586741531679344e-05, "loss": 1.3405, "step": 7846 }, { "epoch": 19.6175, "grad_norm": 8.875, "learning_rate": 1.8586326520275412e-05, "loss": 1.4324, "step": 7847 }, { "epoch": 19.62, "grad_norm": 7.5, "learning_rate": 1.858591145257987e-05, "loss": 1.5482, "step": 7848 }, { "epoch": 19.6225, "grad_norm": 6.4375, "learning_rate": 1.8585496328595437e-05, "loss": 1.4119, "step": 7849 }, { "epoch": 19.625, "grad_norm": 5.4375, "learning_rate": 1.858508114832483e-05, "loss": 1.52, "step": 7850 }, { "epoch": 19.6275, "grad_norm": 9.625, "learning_rate": 1.858466591177078e-05, "loss": 1.3074, "step": 7851 }, { "epoch": 19.63, "grad_norm": 7.125, "learning_rate": 1.8584250618936e-05, "loss": 1.395, "step": 7852 }, { "epoch": 19.6325, "grad_norm": 8.3125, "learning_rate": 1.8583835269823222e-05, "loss": 1.4396, "step": 7853 }, { "epoch": 19.635, "grad_norm": 6.625, "learning_rate": 1.8583419864435155e-05, "loss": 1.4537, "step": 7854 }, { "epoch": 19.6375, "grad_norm": 9.8125, "learning_rate": 1.8583004402774537e-05, "loss": 1.5027, "step": 7855 }, { "epoch": 19.64, "grad_norm": 7.9375, "learning_rate": 1.8582588884844086e-05, "loss": 1.4242, "step": 7856 }, { "epoch": 19.6425, "grad_norm": 7.53125, "learning_rate": 1.858217331064652e-05, "loss": 1.3305, "step": 7857 }, { "epoch": 19.645, "grad_norm": 4.90625, "learning_rate": 1.8581757680184575e-05, "loss": 1.6129, "step": 7858 }, { "epoch": 19.6475, "grad_norm": 7.71875, "learning_rate": 1.858134199346097e-05, "loss": 1.3549, "step": 7859 }, { "epoch": 19.65, "grad_norm": 5.75, "learning_rate": 1.8580926250478425e-05, "loss": 1.4316, "step": 7860 }, { "epoch": 19.6525, "grad_norm": 8.9375, "learning_rate": 1.8580510451239676e-05, "loss": 1.5481, "step": 7861 }, { "epoch": 19.655, "grad_norm": 8.5, "learning_rate": 1.8580094595747442e-05, "loss": 1.42, "step": 7862 }, { "epoch": 19.6575, "grad_norm": 6.59375, "learning_rate": 1.857967868400445e-05, "loss": 1.4427, "step": 7863 }, { "epoch": 19.66, "grad_norm": 8.875, "learning_rate": 1.857926271601343e-05, "loss": 1.5943, "step": 7864 }, { "epoch": 19.6625, "grad_norm": 6.0, "learning_rate": 1.8578846691777105e-05, "loss": 1.2814, "step": 7865 }, { "epoch": 19.665, "grad_norm": 7.53125, "learning_rate": 1.8578430611298203e-05, "loss": 1.5083, "step": 7866 }, { "epoch": 19.6675, "grad_norm": 6.71875, "learning_rate": 1.8578014474579457e-05, "loss": 1.5341, "step": 7867 }, { "epoch": 19.67, "grad_norm": 7.25, "learning_rate": 1.857759828162359e-05, "loss": 1.4151, "step": 7868 }, { "epoch": 19.6725, "grad_norm": 10.9375, "learning_rate": 1.8577182032433332e-05, "loss": 1.3437, "step": 7869 }, { "epoch": 19.675, "grad_norm": 7.125, "learning_rate": 1.8576765727011413e-05, "loss": 1.5235, "step": 7870 }, { "epoch": 19.6775, "grad_norm": 10.5, "learning_rate": 1.857634936536056e-05, "loss": 1.3832, "step": 7871 }, { "epoch": 19.68, "grad_norm": 8.375, "learning_rate": 1.8575932947483503e-05, "loss": 1.6126, "step": 7872 }, { "epoch": 19.6825, "grad_norm": 7.96875, "learning_rate": 1.8575516473382973e-05, "loss": 1.3577, "step": 7873 }, { "epoch": 19.685, "grad_norm": 7.9375, "learning_rate": 1.85750999430617e-05, "loss": 1.3864, "step": 7874 }, { "epoch": 19.6875, "grad_norm": 5.1875, "learning_rate": 1.8574683356522416e-05, "loss": 1.4881, "step": 7875 }, { "epoch": 19.69, "grad_norm": 6.25, "learning_rate": 1.857426671376785e-05, "loss": 1.4251, "step": 7876 }, { "epoch": 19.6925, "grad_norm": 6.6875, "learning_rate": 1.8573850014800737e-05, "loss": 1.6146, "step": 7877 }, { "epoch": 19.695, "grad_norm": 5.375, "learning_rate": 1.8573433259623805e-05, "loss": 1.4955, "step": 7878 }, { "epoch": 19.6975, "grad_norm": 7.03125, "learning_rate": 1.8573016448239783e-05, "loss": 1.5588, "step": 7879 }, { "epoch": 19.7, "grad_norm": 7.4375, "learning_rate": 1.8572599580651415e-05, "loss": 1.3047, "step": 7880 }, { "epoch": 19.7025, "grad_norm": 8.375, "learning_rate": 1.8572182656861425e-05, "loss": 1.438, "step": 7881 }, { "epoch": 19.705, "grad_norm": 5.5625, "learning_rate": 1.8571765676872547e-05, "loss": 1.3763, "step": 7882 }, { "epoch": 19.7075, "grad_norm": 11.25, "learning_rate": 1.8571348640687516e-05, "loss": 1.5307, "step": 7883 }, { "epoch": 19.71, "grad_norm": 5.4375, "learning_rate": 1.857093154830907e-05, "loss": 1.4238, "step": 7884 }, { "epoch": 19.7125, "grad_norm": 6.375, "learning_rate": 1.8570514399739936e-05, "loss": 1.6024, "step": 7885 }, { "epoch": 19.715, "grad_norm": 7.09375, "learning_rate": 1.8570097194982853e-05, "loss": 1.238, "step": 7886 }, { "epoch": 19.7175, "grad_norm": 6.59375, "learning_rate": 1.856967993404056e-05, "loss": 1.4691, "step": 7887 }, { "epoch": 19.72, "grad_norm": 7.1875, "learning_rate": 1.8569262616915784e-05, "loss": 1.4109, "step": 7888 }, { "epoch": 19.7225, "grad_norm": 6.34375, "learning_rate": 1.8568845243611267e-05, "loss": 1.3737, "step": 7889 }, { "epoch": 19.725, "grad_norm": 8.25, "learning_rate": 1.8568427814129743e-05, "loss": 1.4214, "step": 7890 }, { "epoch": 19.7275, "grad_norm": 7.75, "learning_rate": 1.8568010328473946e-05, "loss": 1.2872, "step": 7891 }, { "epoch": 19.73, "grad_norm": 6.09375, "learning_rate": 1.856759278664662e-05, "loss": 1.5601, "step": 7892 }, { "epoch": 19.7325, "grad_norm": 7.21875, "learning_rate": 1.85671751886505e-05, "loss": 1.2933, "step": 7893 }, { "epoch": 19.735, "grad_norm": 6.46875, "learning_rate": 1.856675753448832e-05, "loss": 1.4216, "step": 7894 }, { "epoch": 19.7375, "grad_norm": 5.28125, "learning_rate": 1.856633982416282e-05, "loss": 1.3592, "step": 7895 }, { "epoch": 19.74, "grad_norm": 6.4375, "learning_rate": 1.8565922057676738e-05, "loss": 1.4018, "step": 7896 }, { "epoch": 19.7425, "grad_norm": 7.0625, "learning_rate": 1.8565504235032813e-05, "loss": 1.4086, "step": 7897 }, { "epoch": 19.745, "grad_norm": 7.25, "learning_rate": 1.8565086356233788e-05, "loss": 1.51, "step": 7898 }, { "epoch": 19.7475, "grad_norm": 5.71875, "learning_rate": 1.8564668421282398e-05, "loss": 1.4276, "step": 7899 }, { "epoch": 19.75, "grad_norm": 9.0, "learning_rate": 1.8564250430181387e-05, "loss": 1.4832, "step": 7900 }, { "epoch": 19.7525, "grad_norm": 8.1875, "learning_rate": 1.856383238293349e-05, "loss": 1.5623, "step": 7901 }, { "epoch": 19.755, "grad_norm": 5.875, "learning_rate": 1.8563414279541453e-05, "loss": 1.496, "step": 7902 }, { "epoch": 19.7575, "grad_norm": 8.4375, "learning_rate": 1.856299612000801e-05, "loss": 1.2896, "step": 7903 }, { "epoch": 19.76, "grad_norm": 6.65625, "learning_rate": 1.8562577904335913e-05, "loss": 1.4247, "step": 7904 }, { "epoch": 19.7625, "grad_norm": 8.6875, "learning_rate": 1.8562159632527896e-05, "loss": 1.4754, "step": 7905 }, { "epoch": 19.765, "grad_norm": 7.65625, "learning_rate": 1.85617413045867e-05, "loss": 1.4413, "step": 7906 }, { "epoch": 19.7675, "grad_norm": 5.90625, "learning_rate": 1.8561322920515076e-05, "loss": 1.4074, "step": 7907 }, { "epoch": 19.77, "grad_norm": 6.625, "learning_rate": 1.8560904480315757e-05, "loss": 1.5099, "step": 7908 }, { "epoch": 19.7725, "grad_norm": 7.1875, "learning_rate": 1.8560485983991495e-05, "loss": 1.3649, "step": 7909 }, { "epoch": 19.775, "grad_norm": 6.90625, "learning_rate": 1.8560067431545023e-05, "loss": 1.3983, "step": 7910 }, { "epoch": 19.7775, "grad_norm": 7.28125, "learning_rate": 1.8559648822979098e-05, "loss": 1.5486, "step": 7911 }, { "epoch": 19.78, "grad_norm": 7.6875, "learning_rate": 1.8559230158296454e-05, "loss": 1.3923, "step": 7912 }, { "epoch": 19.7825, "grad_norm": 6.625, "learning_rate": 1.8558811437499842e-05, "loss": 1.4713, "step": 7913 }, { "epoch": 19.785, "grad_norm": 5.6875, "learning_rate": 1.8558392660592003e-05, "loss": 1.3323, "step": 7914 }, { "epoch": 19.7875, "grad_norm": 7.6875, "learning_rate": 1.8557973827575683e-05, "loss": 1.4141, "step": 7915 }, { "epoch": 19.79, "grad_norm": 7.8125, "learning_rate": 1.855755493845363e-05, "loss": 1.4254, "step": 7916 }, { "epoch": 19.7925, "grad_norm": 6.1875, "learning_rate": 1.8557135993228592e-05, "loss": 1.3941, "step": 7917 }, { "epoch": 19.795, "grad_norm": 9.0, "learning_rate": 1.855671699190331e-05, "loss": 1.399, "step": 7918 }, { "epoch": 19.7975, "grad_norm": 6.625, "learning_rate": 1.8556297934480536e-05, "loss": 1.4862, "step": 7919 }, { "epoch": 19.8, "grad_norm": 6.25, "learning_rate": 1.8555878820963014e-05, "loss": 1.3948, "step": 7920 }, { "epoch": 19.8025, "grad_norm": 8.0625, "learning_rate": 1.8555459651353493e-05, "loss": 1.4102, "step": 7921 }, { "epoch": 19.805, "grad_norm": 6.15625, "learning_rate": 1.855504042565472e-05, "loss": 1.4163, "step": 7922 }, { "epoch": 19.8075, "grad_norm": 5.78125, "learning_rate": 1.8554621143869445e-05, "loss": 1.2664, "step": 7923 }, { "epoch": 19.81, "grad_norm": 8.25, "learning_rate": 1.8554201806000417e-05, "loss": 1.3859, "step": 7924 }, { "epoch": 19.8125, "grad_norm": 5.90625, "learning_rate": 1.8553782412050384e-05, "loss": 1.3797, "step": 7925 }, { "epoch": 19.815, "grad_norm": 9.6875, "learning_rate": 1.8553362962022095e-05, "loss": 1.4841, "step": 7926 }, { "epoch": 19.8175, "grad_norm": 4.84375, "learning_rate": 1.85529434559183e-05, "loss": 1.3812, "step": 7927 }, { "epoch": 19.82, "grad_norm": 5.65625, "learning_rate": 1.855252389374175e-05, "loss": 1.2801, "step": 7928 }, { "epoch": 19.8225, "grad_norm": 8.0, "learning_rate": 1.85521042754952e-05, "loss": 1.3297, "step": 7929 }, { "epoch": 19.825, "grad_norm": 7.03125, "learning_rate": 1.855168460118139e-05, "loss": 1.3155, "step": 7930 }, { "epoch": 19.8275, "grad_norm": 7.875, "learning_rate": 1.855126487080308e-05, "loss": 1.3871, "step": 7931 }, { "epoch": 19.83, "grad_norm": 6.71875, "learning_rate": 1.8550845084363023e-05, "loss": 1.3833, "step": 7932 }, { "epoch": 19.8325, "grad_norm": 6.6875, "learning_rate": 1.8550425241863962e-05, "loss": 1.4558, "step": 7933 }, { "epoch": 19.835, "grad_norm": 7.125, "learning_rate": 1.855000534330866e-05, "loss": 1.4494, "step": 7934 }, { "epoch": 19.8375, "grad_norm": 5.84375, "learning_rate": 1.8549585388699864e-05, "loss": 1.3623, "step": 7935 }, { "epoch": 19.84, "grad_norm": 7.875, "learning_rate": 1.8549165378040328e-05, "loss": 1.4268, "step": 7936 }, { "epoch": 19.8425, "grad_norm": 5.71875, "learning_rate": 1.8548745311332808e-05, "loss": 1.3624, "step": 7937 }, { "epoch": 19.845, "grad_norm": 9.375, "learning_rate": 1.8548325188580053e-05, "loss": 1.4484, "step": 7938 }, { "epoch": 19.8475, "grad_norm": 6.40625, "learning_rate": 1.8547905009784822e-05, "loss": 1.516, "step": 7939 }, { "epoch": 19.85, "grad_norm": 5.71875, "learning_rate": 1.8547484774949865e-05, "loss": 1.491, "step": 7940 }, { "epoch": 19.8525, "grad_norm": 11.125, "learning_rate": 1.8547064484077942e-05, "loss": 1.5539, "step": 7941 }, { "epoch": 19.855, "grad_norm": 7.96875, "learning_rate": 1.854664413717181e-05, "loss": 1.4021, "step": 7942 }, { "epoch": 19.8575, "grad_norm": 7.15625, "learning_rate": 1.8546223734234215e-05, "loss": 1.6018, "step": 7943 }, { "epoch": 19.86, "grad_norm": 6.15625, "learning_rate": 1.8545803275267923e-05, "loss": 1.5833, "step": 7944 }, { "epoch": 19.8625, "grad_norm": 8.75, "learning_rate": 1.8545382760275683e-05, "loss": 1.4269, "step": 7945 }, { "epoch": 19.865, "grad_norm": 6.96875, "learning_rate": 1.8544962189260258e-05, "loss": 1.3476, "step": 7946 }, { "epoch": 19.8675, "grad_norm": 7.03125, "learning_rate": 1.8544541562224404e-05, "loss": 1.5525, "step": 7947 }, { "epoch": 19.87, "grad_norm": 7.0, "learning_rate": 1.8544120879170874e-05, "loss": 1.5446, "step": 7948 }, { "epoch": 19.8725, "grad_norm": 6.34375, "learning_rate": 1.854370014010243e-05, "loss": 1.3247, "step": 7949 }, { "epoch": 19.875, "grad_norm": 5.625, "learning_rate": 1.8543279345021834e-05, "loss": 1.2661, "step": 7950 }, { "epoch": 19.8775, "grad_norm": 9.125, "learning_rate": 1.854285849393184e-05, "loss": 1.5431, "step": 7951 }, { "epoch": 19.88, "grad_norm": 7.1875, "learning_rate": 1.8542437586835202e-05, "loss": 1.3818, "step": 7952 }, { "epoch": 19.8825, "grad_norm": 5.84375, "learning_rate": 1.854201662373469e-05, "loss": 1.3293, "step": 7953 }, { "epoch": 19.885, "grad_norm": 5.46875, "learning_rate": 1.8541595604633055e-05, "loss": 1.2582, "step": 7954 }, { "epoch": 19.8875, "grad_norm": 8.6875, "learning_rate": 1.8541174529533066e-05, "loss": 1.5242, "step": 7955 }, { "epoch": 19.89, "grad_norm": 7.28125, "learning_rate": 1.8540753398437473e-05, "loss": 1.4072, "step": 7956 }, { "epoch": 19.8925, "grad_norm": 7.71875, "learning_rate": 1.8540332211349046e-05, "loss": 1.506, "step": 7957 }, { "epoch": 19.895, "grad_norm": 8.6875, "learning_rate": 1.8539910968270544e-05, "loss": 1.3582, "step": 7958 }, { "epoch": 19.8975, "grad_norm": 9.0, "learning_rate": 1.8539489669204723e-05, "loss": 1.3812, "step": 7959 }, { "epoch": 19.9, "grad_norm": 8.125, "learning_rate": 1.8539068314154355e-05, "loss": 1.5702, "step": 7960 }, { "epoch": 19.9025, "grad_norm": 6.25, "learning_rate": 1.8538646903122193e-05, "loss": 1.5981, "step": 7961 }, { "epoch": 19.905, "grad_norm": 8.625, "learning_rate": 1.8538225436111003e-05, "loss": 1.5715, "step": 7962 }, { "epoch": 19.9075, "grad_norm": 7.46875, "learning_rate": 1.8537803913123552e-05, "loss": 1.7527, "step": 7963 }, { "epoch": 19.91, "grad_norm": 8.875, "learning_rate": 1.8537382334162596e-05, "loss": 1.3859, "step": 7964 }, { "epoch": 19.9125, "grad_norm": 6.25, "learning_rate": 1.853696069923091e-05, "loss": 1.3745, "step": 7965 }, { "epoch": 19.915, "grad_norm": 6.46875, "learning_rate": 1.853653900833124e-05, "loss": 1.4561, "step": 7966 }, { "epoch": 19.9175, "grad_norm": 6.84375, "learning_rate": 1.8536117261466372e-05, "loss": 1.512, "step": 7967 }, { "epoch": 19.92, "grad_norm": 7.46875, "learning_rate": 1.8535695458639056e-05, "loss": 1.3865, "step": 7968 }, { "epoch": 19.9225, "grad_norm": 4.71875, "learning_rate": 1.8535273599852065e-05, "loss": 1.3349, "step": 7969 }, { "epoch": 19.925, "grad_norm": 7.3125, "learning_rate": 1.853485168510816e-05, "loss": 1.4583, "step": 7970 }, { "epoch": 19.9275, "grad_norm": 7.25, "learning_rate": 1.8534429714410108e-05, "loss": 1.26, "step": 7971 }, { "epoch": 19.93, "grad_norm": 8.0, "learning_rate": 1.8534007687760682e-05, "loss": 1.4648, "step": 7972 }, { "epoch": 19.9325, "grad_norm": 5.1875, "learning_rate": 1.8533585605162635e-05, "loss": 1.3008, "step": 7973 }, { "epoch": 19.935, "grad_norm": 6.0, "learning_rate": 1.8533163466618744e-05, "loss": 1.4126, "step": 7974 }, { "epoch": 19.9375, "grad_norm": 7.28125, "learning_rate": 1.853274127213178e-05, "loss": 1.3644, "step": 7975 }, { "epoch": 19.94, "grad_norm": 6.3125, "learning_rate": 1.8532319021704502e-05, "loss": 1.468, "step": 7976 }, { "epoch": 19.9425, "grad_norm": 9.0625, "learning_rate": 1.853189671533968e-05, "loss": 1.3958, "step": 7977 }, { "epoch": 19.945, "grad_norm": 7.59375, "learning_rate": 1.853147435304009e-05, "loss": 1.3107, "step": 7978 }, { "epoch": 19.9475, "grad_norm": 8.0625, "learning_rate": 1.853105193480849e-05, "loss": 1.4235, "step": 7979 }, { "epoch": 19.95, "grad_norm": 7.96875, "learning_rate": 1.8530629460647658e-05, "loss": 1.4978, "step": 7980 }, { "epoch": 19.9525, "grad_norm": 7.375, "learning_rate": 1.853020693056036e-05, "loss": 1.4343, "step": 7981 }, { "epoch": 19.955, "grad_norm": 7.0, "learning_rate": 1.8529784344549367e-05, "loss": 1.4588, "step": 7982 }, { "epoch": 19.9575, "grad_norm": 7.875, "learning_rate": 1.852936170261745e-05, "loss": 1.3742, "step": 7983 }, { "epoch": 19.96, "grad_norm": 7.03125, "learning_rate": 1.8528939004767377e-05, "loss": 1.4078, "step": 7984 }, { "epoch": 19.9625, "grad_norm": 7.5, "learning_rate": 1.8528516251001925e-05, "loss": 1.3673, "step": 7985 }, { "epoch": 19.965, "grad_norm": 6.375, "learning_rate": 1.8528093441323855e-05, "loss": 1.5483, "step": 7986 }, { "epoch": 19.9675, "grad_norm": 6.78125, "learning_rate": 1.852767057573595e-05, "loss": 1.29, "step": 7987 }, { "epoch": 19.97, "grad_norm": 5.8125, "learning_rate": 1.8527247654240977e-05, "loss": 1.4143, "step": 7988 }, { "epoch": 19.9725, "grad_norm": 8.8125, "learning_rate": 1.8526824676841713e-05, "loss": 1.5082, "step": 7989 }, { "epoch": 19.975, "grad_norm": 7.34375, "learning_rate": 1.8526401643540924e-05, "loss": 1.2523, "step": 7990 }, { "epoch": 19.9775, "grad_norm": 8.625, "learning_rate": 1.8525978554341385e-05, "loss": 1.4513, "step": 7991 }, { "epoch": 19.98, "grad_norm": 7.3125, "learning_rate": 1.8525555409245877e-05, "loss": 1.3996, "step": 7992 }, { "epoch": 19.9825, "grad_norm": 7.21875, "learning_rate": 1.8525132208257166e-05, "loss": 1.596, "step": 7993 }, { "epoch": 19.985, "grad_norm": 7.125, "learning_rate": 1.852470895137803e-05, "loss": 1.4422, "step": 7994 }, { "epoch": 19.9875, "grad_norm": 7.65625, "learning_rate": 1.8524285638611245e-05, "loss": 1.5519, "step": 7995 }, { "epoch": 19.99, "grad_norm": 5.0, "learning_rate": 1.852386226995958e-05, "loss": 1.346, "step": 7996 }, { "epoch": 19.9925, "grad_norm": 8.4375, "learning_rate": 1.8523438845425818e-05, "loss": 1.2982, "step": 7997 }, { "epoch": 19.995, "grad_norm": 6.03125, "learning_rate": 1.8523015365012734e-05, "loss": 1.3841, "step": 7998 }, { "epoch": 19.9975, "grad_norm": 7.46875, "learning_rate": 1.85225918287231e-05, "loss": 1.5145, "step": 7999 }, { "epoch": 20.0, "grad_norm": 9.0625, "learning_rate": 1.8522168236559693e-05, "loss": 1.7695, "step": 8000 }, { "epoch": 20.0025, "grad_norm": 7.59375, "learning_rate": 1.8521744588525298e-05, "loss": 1.4799, "step": 8001 }, { "epoch": 20.005, "grad_norm": 6.5625, "learning_rate": 1.852132088462268e-05, "loss": 1.4544, "step": 8002 }, { "epoch": 20.0075, "grad_norm": 5.875, "learning_rate": 1.8520897124854628e-05, "loss": 1.353, "step": 8003 }, { "epoch": 20.01, "grad_norm": 8.125, "learning_rate": 1.8520473309223912e-05, "loss": 1.4038, "step": 8004 }, { "epoch": 20.0125, "grad_norm": 5.46875, "learning_rate": 1.8520049437733315e-05, "loss": 1.3845, "step": 8005 }, { "epoch": 20.015, "grad_norm": 7.0, "learning_rate": 1.8519625510385614e-05, "loss": 1.4637, "step": 8006 }, { "epoch": 20.0175, "grad_norm": 5.75, "learning_rate": 1.8519201527183592e-05, "loss": 1.329, "step": 8007 }, { "epoch": 20.02, "grad_norm": 11.1875, "learning_rate": 1.8518777488130024e-05, "loss": 1.3558, "step": 8008 }, { "epoch": 20.0225, "grad_norm": 5.03125, "learning_rate": 1.851835339322769e-05, "loss": 1.3056, "step": 8009 }, { "epoch": 20.025, "grad_norm": 7.75, "learning_rate": 1.8517929242479377e-05, "loss": 1.4832, "step": 8010 }, { "epoch": 20.0275, "grad_norm": 6.53125, "learning_rate": 1.8517505035887855e-05, "loss": 1.3569, "step": 8011 }, { "epoch": 20.03, "grad_norm": 7.28125, "learning_rate": 1.8517080773455913e-05, "loss": 1.2659, "step": 8012 }, { "epoch": 20.0325, "grad_norm": 8.6875, "learning_rate": 1.851665645518633e-05, "loss": 1.3715, "step": 8013 }, { "epoch": 20.035, "grad_norm": 6.21875, "learning_rate": 1.851623208108189e-05, "loss": 1.3883, "step": 8014 }, { "epoch": 20.0375, "grad_norm": 7.3125, "learning_rate": 1.8515807651145373e-05, "loss": 1.5051, "step": 8015 }, { "epoch": 20.04, "grad_norm": 7.09375, "learning_rate": 1.851538316537956e-05, "loss": 1.6318, "step": 8016 }, { "epoch": 20.0425, "grad_norm": 8.8125, "learning_rate": 1.8514958623787236e-05, "loss": 1.4824, "step": 8017 }, { "epoch": 20.045, "grad_norm": 7.375, "learning_rate": 1.8514534026371185e-05, "loss": 1.446, "step": 8018 }, { "epoch": 20.0475, "grad_norm": 8.5625, "learning_rate": 1.851410937313419e-05, "loss": 1.4115, "step": 8019 }, { "epoch": 20.05, "grad_norm": 5.5625, "learning_rate": 1.8513684664079033e-05, "loss": 1.3157, "step": 8020 }, { "epoch": 20.0525, "grad_norm": 6.0625, "learning_rate": 1.85132598992085e-05, "loss": 1.3976, "step": 8021 }, { "epoch": 20.055, "grad_norm": 5.53125, "learning_rate": 1.8512835078525378e-05, "loss": 1.4435, "step": 8022 }, { "epoch": 20.0575, "grad_norm": 6.0625, "learning_rate": 1.851241020203245e-05, "loss": 1.3173, "step": 8023 }, { "epoch": 20.06, "grad_norm": 4.46875, "learning_rate": 1.8511985269732496e-05, "loss": 1.4832, "step": 8024 }, { "epoch": 20.0625, "grad_norm": 4.15625, "learning_rate": 1.8511560281628312e-05, "loss": 1.4348, "step": 8025 }, { "epoch": 20.065, "grad_norm": 7.5625, "learning_rate": 1.851113523772268e-05, "loss": 1.3985, "step": 8026 }, { "epoch": 20.0675, "grad_norm": 6.6875, "learning_rate": 1.8510710138018384e-05, "loss": 1.3494, "step": 8027 }, { "epoch": 20.07, "grad_norm": 5.3125, "learning_rate": 1.8510284982518213e-05, "loss": 1.4899, "step": 8028 }, { "epoch": 20.0725, "grad_norm": 6.59375, "learning_rate": 1.8509859771224957e-05, "loss": 1.5117, "step": 8029 }, { "epoch": 20.075, "grad_norm": 5.6875, "learning_rate": 1.85094345041414e-05, "loss": 1.5463, "step": 8030 }, { "epoch": 20.0775, "grad_norm": 9.25, "learning_rate": 1.8509009181270326e-05, "loss": 1.3649, "step": 8031 }, { "epoch": 20.08, "grad_norm": 9.75, "learning_rate": 1.8508583802614534e-05, "loss": 1.3517, "step": 8032 }, { "epoch": 20.0825, "grad_norm": 9.5, "learning_rate": 1.8508158368176804e-05, "loss": 1.4862, "step": 8033 }, { "epoch": 20.085, "grad_norm": 8.5625, "learning_rate": 1.8507732877959926e-05, "loss": 1.3629, "step": 8034 }, { "epoch": 20.0875, "grad_norm": 8.4375, "learning_rate": 1.8507307331966697e-05, "loss": 1.3086, "step": 8035 }, { "epoch": 20.09, "grad_norm": 7.90625, "learning_rate": 1.85068817301999e-05, "loss": 1.5243, "step": 8036 }, { "epoch": 20.0925, "grad_norm": 7.34375, "learning_rate": 1.8506456072662328e-05, "loss": 1.3947, "step": 8037 }, { "epoch": 20.095, "grad_norm": 5.90625, "learning_rate": 1.8506030359356764e-05, "loss": 1.2669, "step": 8038 }, { "epoch": 20.0975, "grad_norm": 8.3125, "learning_rate": 1.850560459028601e-05, "loss": 1.4368, "step": 8039 }, { "epoch": 20.1, "grad_norm": 8.4375, "learning_rate": 1.8505178765452853e-05, "loss": 1.5347, "step": 8040 }, { "epoch": 20.1025, "grad_norm": 8.875, "learning_rate": 1.8504752884860085e-05, "loss": 1.2903, "step": 8041 }, { "epoch": 20.105, "grad_norm": 6.1875, "learning_rate": 1.85043269485105e-05, "loss": 1.4082, "step": 8042 }, { "epoch": 20.1075, "grad_norm": 8.4375, "learning_rate": 1.850390095640688e-05, "loss": 1.5364, "step": 8043 }, { "epoch": 20.11, "grad_norm": 8.125, "learning_rate": 1.850347490855203e-05, "loss": 1.278, "step": 8044 }, { "epoch": 20.1125, "grad_norm": 6.5, "learning_rate": 1.850304880494874e-05, "loss": 1.5941, "step": 8045 }, { "epoch": 20.115, "grad_norm": 5.375, "learning_rate": 1.8502622645599803e-05, "loss": 1.3842, "step": 8046 }, { "epoch": 20.1175, "grad_norm": 7.34375, "learning_rate": 1.850219643050801e-05, "loss": 1.3751, "step": 8047 }, { "epoch": 20.12, "grad_norm": 5.34375, "learning_rate": 1.8501770159676157e-05, "loss": 1.5792, "step": 8048 }, { "epoch": 20.1225, "grad_norm": 9.3125, "learning_rate": 1.850134383310704e-05, "loss": 1.4503, "step": 8049 }, { "epoch": 20.125, "grad_norm": 6.3125, "learning_rate": 1.850091745080345e-05, "loss": 1.2972, "step": 8050 }, { "epoch": 20.1275, "grad_norm": 6.3125, "learning_rate": 1.850049101276819e-05, "loss": 1.4265, "step": 8051 }, { "epoch": 20.13, "grad_norm": 7.625, "learning_rate": 1.8500064519004047e-05, "loss": 1.516, "step": 8052 }, { "epoch": 20.1325, "grad_norm": 6.53125, "learning_rate": 1.8499637969513826e-05, "loss": 1.4529, "step": 8053 }, { "epoch": 20.135, "grad_norm": 6.59375, "learning_rate": 1.8499211364300316e-05, "loss": 1.3225, "step": 8054 }, { "epoch": 20.1375, "grad_norm": 7.46875, "learning_rate": 1.8498784703366316e-05, "loss": 1.3593, "step": 8055 }, { "epoch": 20.14, "grad_norm": 7.0625, "learning_rate": 1.8498357986714624e-05, "loss": 1.4383, "step": 8056 }, { "epoch": 20.1425, "grad_norm": 6.8125, "learning_rate": 1.8497931214348036e-05, "loss": 1.2909, "step": 8057 }, { "epoch": 20.145, "grad_norm": 4.46875, "learning_rate": 1.8497504386269353e-05, "loss": 1.3938, "step": 8058 }, { "epoch": 20.1475, "grad_norm": 7.25, "learning_rate": 1.849707750248137e-05, "loss": 1.3835, "step": 8059 }, { "epoch": 20.15, "grad_norm": 6.3125, "learning_rate": 1.8496650562986888e-05, "loss": 1.3544, "step": 8060 }, { "epoch": 20.1525, "grad_norm": 5.28125, "learning_rate": 1.8496223567788703e-05, "loss": 1.1885, "step": 8061 }, { "epoch": 20.155, "grad_norm": 8.3125, "learning_rate": 1.8495796516889617e-05, "loss": 1.556, "step": 8062 }, { "epoch": 20.1575, "grad_norm": 8.8125, "learning_rate": 1.849536941029243e-05, "loss": 1.298, "step": 8063 }, { "epoch": 20.16, "grad_norm": 7.6875, "learning_rate": 1.849494224799994e-05, "loss": 1.3292, "step": 8064 }, { "epoch": 20.1625, "grad_norm": 11.25, "learning_rate": 1.849451503001495e-05, "loss": 1.3037, "step": 8065 }, { "epoch": 20.165, "grad_norm": 6.6875, "learning_rate": 1.849408775634026e-05, "loss": 1.4408, "step": 8066 }, { "epoch": 20.1675, "grad_norm": 6.0, "learning_rate": 1.849366042697867e-05, "loss": 1.4079, "step": 8067 }, { "epoch": 20.17, "grad_norm": 7.53125, "learning_rate": 1.8493233041932977e-05, "loss": 1.2229, "step": 8068 }, { "epoch": 20.1725, "grad_norm": 8.125, "learning_rate": 1.8492805601205996e-05, "loss": 1.5774, "step": 8069 }, { "epoch": 20.175, "grad_norm": 5.9375, "learning_rate": 1.8492378104800516e-05, "loss": 1.4921, "step": 8070 }, { "epoch": 20.1775, "grad_norm": 6.375, "learning_rate": 1.8491950552719347e-05, "loss": 1.3545, "step": 8071 }, { "epoch": 20.18, "grad_norm": 12.125, "learning_rate": 1.849152294496529e-05, "loss": 1.4147, "step": 8072 }, { "epoch": 20.1825, "grad_norm": 6.53125, "learning_rate": 1.8491095281541147e-05, "loss": 1.3163, "step": 8073 }, { "epoch": 20.185, "grad_norm": 6.0625, "learning_rate": 1.8490667562449727e-05, "loss": 1.4223, "step": 8074 }, { "epoch": 20.1875, "grad_norm": 6.625, "learning_rate": 1.8490239787693825e-05, "loss": 1.4399, "step": 8075 }, { "epoch": 20.19, "grad_norm": 6.75, "learning_rate": 1.8489811957276254e-05, "loss": 1.3384, "step": 8076 }, { "epoch": 20.1925, "grad_norm": 6.40625, "learning_rate": 1.8489384071199813e-05, "loss": 1.4071, "step": 8077 }, { "epoch": 20.195, "grad_norm": 8.6875, "learning_rate": 1.8488956129467314e-05, "loss": 1.4301, "step": 8078 }, { "epoch": 20.1975, "grad_norm": 5.84375, "learning_rate": 1.8488528132081555e-05, "loss": 1.3956, "step": 8079 }, { "epoch": 20.2, "grad_norm": 6.21875, "learning_rate": 1.8488100079045345e-05, "loss": 1.4097, "step": 8080 }, { "epoch": 20.2025, "grad_norm": 5.8125, "learning_rate": 1.8487671970361493e-05, "loss": 1.393, "step": 8081 }, { "epoch": 20.205, "grad_norm": 6.03125, "learning_rate": 1.84872438060328e-05, "loss": 1.4425, "step": 8082 }, { "epoch": 20.2075, "grad_norm": 6.34375, "learning_rate": 1.848681558606208e-05, "loss": 1.3261, "step": 8083 }, { "epoch": 20.21, "grad_norm": 5.625, "learning_rate": 1.8486387310452135e-05, "loss": 1.3283, "step": 8084 }, { "epoch": 20.2125, "grad_norm": 5.6875, "learning_rate": 1.8485958979205775e-05, "loss": 1.482, "step": 8085 }, { "epoch": 20.215, "grad_norm": 4.0, "learning_rate": 1.8485530592325806e-05, "loss": 1.3604, "step": 8086 }, { "epoch": 20.2175, "grad_norm": 6.90625, "learning_rate": 1.8485102149815038e-05, "loss": 1.3389, "step": 8087 }, { "epoch": 20.22, "grad_norm": 7.21875, "learning_rate": 1.8484673651676283e-05, "loss": 1.458, "step": 8088 }, { "epoch": 20.2225, "grad_norm": 7.09375, "learning_rate": 1.8484245097912345e-05, "loss": 1.6424, "step": 8089 }, { "epoch": 20.225, "grad_norm": 8.4375, "learning_rate": 1.8483816488526033e-05, "loss": 1.3591, "step": 8090 }, { "epoch": 20.2275, "grad_norm": 7.3125, "learning_rate": 1.8483387823520164e-05, "loss": 1.4887, "step": 8091 }, { "epoch": 20.23, "grad_norm": 8.875, "learning_rate": 1.848295910289754e-05, "loss": 1.34, "step": 8092 }, { "epoch": 20.2325, "grad_norm": 6.8125, "learning_rate": 1.8482530326660977e-05, "loss": 1.1942, "step": 8093 }, { "epoch": 20.235, "grad_norm": 5.5, "learning_rate": 1.8482101494813286e-05, "loss": 1.3764, "step": 8094 }, { "epoch": 20.2375, "grad_norm": 6.90625, "learning_rate": 1.8481672607357277e-05, "loss": 1.3529, "step": 8095 }, { "epoch": 20.24, "grad_norm": 8.1875, "learning_rate": 1.848124366429576e-05, "loss": 1.4672, "step": 8096 }, { "epoch": 20.2425, "grad_norm": 8.1875, "learning_rate": 1.848081466563155e-05, "loss": 1.39, "step": 8097 }, { "epoch": 20.245, "grad_norm": 6.6875, "learning_rate": 1.848038561136746e-05, "loss": 1.4744, "step": 8098 }, { "epoch": 20.2475, "grad_norm": 7.8125, "learning_rate": 1.8479956501506296e-05, "loss": 1.2809, "step": 8099 }, { "epoch": 20.25, "grad_norm": 4.78125, "learning_rate": 1.847952733605088e-05, "loss": 1.2289, "step": 8100 }, { "epoch": 20.2525, "grad_norm": 6.46875, "learning_rate": 1.8479098115004023e-05, "loss": 1.5806, "step": 8101 }, { "epoch": 20.255, "grad_norm": 11.25, "learning_rate": 1.8478668838368538e-05, "loss": 1.3352, "step": 8102 }, { "epoch": 20.2575, "grad_norm": 7.21875, "learning_rate": 1.8478239506147237e-05, "loss": 1.452, "step": 8103 }, { "epoch": 20.26, "grad_norm": 6.75, "learning_rate": 1.847781011834294e-05, "loss": 1.5047, "step": 8104 }, { "epoch": 20.2625, "grad_norm": 7.5, "learning_rate": 1.8477380674958457e-05, "loss": 1.4568, "step": 8105 }, { "epoch": 20.265, "grad_norm": 7.5625, "learning_rate": 1.8476951175996606e-05, "loss": 1.2282, "step": 8106 }, { "epoch": 20.2675, "grad_norm": 5.9375, "learning_rate": 1.8476521621460205e-05, "loss": 1.2979, "step": 8107 }, { "epoch": 20.27, "grad_norm": 8.625, "learning_rate": 1.8476092011352063e-05, "loss": 1.413, "step": 8108 }, { "epoch": 20.2725, "grad_norm": 6.0, "learning_rate": 1.8475662345675002e-05, "loss": 1.4952, "step": 8109 }, { "epoch": 20.275, "grad_norm": 6.1875, "learning_rate": 1.847523262443184e-05, "loss": 1.4321, "step": 8110 }, { "epoch": 20.2775, "grad_norm": 8.1875, "learning_rate": 1.847480284762539e-05, "loss": 1.3682, "step": 8111 }, { "epoch": 20.28, "grad_norm": 6.84375, "learning_rate": 1.8474373015258472e-05, "loss": 1.2815, "step": 8112 }, { "epoch": 20.2825, "grad_norm": 8.5625, "learning_rate": 1.8473943127333908e-05, "loss": 1.4338, "step": 8113 }, { "epoch": 20.285, "grad_norm": 7.4375, "learning_rate": 1.847351318385451e-05, "loss": 1.4297, "step": 8114 }, { "epoch": 20.2875, "grad_norm": 7.03125, "learning_rate": 1.8473083184823095e-05, "loss": 1.4991, "step": 8115 }, { "epoch": 20.29, "grad_norm": 6.40625, "learning_rate": 1.8472653130242485e-05, "loss": 1.5251, "step": 8116 }, { "epoch": 20.2925, "grad_norm": 6.4375, "learning_rate": 1.8472223020115507e-05, "loss": 1.3357, "step": 8117 }, { "epoch": 20.295, "grad_norm": 8.375, "learning_rate": 1.8471792854444966e-05, "loss": 1.3978, "step": 8118 }, { "epoch": 20.2975, "grad_norm": 5.46875, "learning_rate": 1.8471362633233693e-05, "loss": 1.4341, "step": 8119 }, { "epoch": 20.3, "grad_norm": 8.5, "learning_rate": 1.847093235648451e-05, "loss": 1.4217, "step": 8120 }, { "epoch": 20.3025, "grad_norm": 9.25, "learning_rate": 1.847050202420023e-05, "loss": 1.429, "step": 8121 }, { "epoch": 20.305, "grad_norm": 6.28125, "learning_rate": 1.8470071636383675e-05, "loss": 1.1818, "step": 8122 }, { "epoch": 20.3075, "grad_norm": 7.375, "learning_rate": 1.8469641193037673e-05, "loss": 1.3046, "step": 8123 }, { "epoch": 20.31, "grad_norm": 6.40625, "learning_rate": 1.8469210694165042e-05, "loss": 1.351, "step": 8124 }, { "epoch": 20.3125, "grad_norm": 9.0625, "learning_rate": 1.8468780139768602e-05, "loss": 1.3861, "step": 8125 }, { "epoch": 20.315, "grad_norm": 6.9375, "learning_rate": 1.846834952985118e-05, "loss": 1.4855, "step": 8126 }, { "epoch": 20.3175, "grad_norm": 8.0, "learning_rate": 1.84679188644156e-05, "loss": 1.41, "step": 8127 }, { "epoch": 20.32, "grad_norm": 6.96875, "learning_rate": 1.846748814346468e-05, "loss": 1.3646, "step": 8128 }, { "epoch": 20.3225, "grad_norm": 6.1875, "learning_rate": 1.8467057367001247e-05, "loss": 1.3514, "step": 8129 }, { "epoch": 20.325, "grad_norm": 7.46875, "learning_rate": 1.8466626535028122e-05, "loss": 1.5293, "step": 8130 }, { "epoch": 20.3275, "grad_norm": 5.34375, "learning_rate": 1.8466195647548136e-05, "loss": 1.5087, "step": 8131 }, { "epoch": 20.33, "grad_norm": 6.34375, "learning_rate": 1.846576470456411e-05, "loss": 1.3746, "step": 8132 }, { "epoch": 20.3325, "grad_norm": 8.1875, "learning_rate": 1.8465333706078867e-05, "loss": 1.378, "step": 8133 }, { "epoch": 20.335, "grad_norm": 6.75, "learning_rate": 1.8464902652095235e-05, "loss": 1.2759, "step": 8134 }, { "epoch": 20.3375, "grad_norm": 7.125, "learning_rate": 1.846447154261604e-05, "loss": 1.5522, "step": 8135 }, { "epoch": 20.34, "grad_norm": 5.90625, "learning_rate": 1.846404037764411e-05, "loss": 1.4255, "step": 8136 }, { "epoch": 20.3425, "grad_norm": 7.0, "learning_rate": 1.8463609157182273e-05, "loss": 1.4204, "step": 8137 }, { "epoch": 20.345, "grad_norm": 4.59375, "learning_rate": 1.8463177881233346e-05, "loss": 1.3803, "step": 8138 }, { "epoch": 20.3475, "grad_norm": 9.3125, "learning_rate": 1.8462746549800166e-05, "loss": 1.5039, "step": 8139 }, { "epoch": 20.35, "grad_norm": 7.46875, "learning_rate": 1.8462315162885563e-05, "loss": 1.5175, "step": 8140 }, { "epoch": 20.3525, "grad_norm": 6.65625, "learning_rate": 1.8461883720492358e-05, "loss": 1.4601, "step": 8141 }, { "epoch": 20.355, "grad_norm": 5.84375, "learning_rate": 1.846145222262338e-05, "loss": 1.5459, "step": 8142 }, { "epoch": 20.3575, "grad_norm": 7.34375, "learning_rate": 1.846102066928146e-05, "loss": 1.5707, "step": 8143 }, { "epoch": 20.36, "grad_norm": 7.0625, "learning_rate": 1.846058906046943e-05, "loss": 1.3972, "step": 8144 }, { "epoch": 20.3625, "grad_norm": 8.4375, "learning_rate": 1.846015739619011e-05, "loss": 1.3909, "step": 8145 }, { "epoch": 20.365, "grad_norm": 7.375, "learning_rate": 1.8459725676446343e-05, "loss": 1.367, "step": 8146 }, { "epoch": 20.3675, "grad_norm": 6.5625, "learning_rate": 1.845929390124095e-05, "loss": 1.3255, "step": 8147 }, { "epoch": 20.37, "grad_norm": 7.5625, "learning_rate": 1.8458862070576767e-05, "loss": 1.3786, "step": 8148 }, { "epoch": 20.3725, "grad_norm": 6.625, "learning_rate": 1.8458430184456618e-05, "loss": 1.3759, "step": 8149 }, { "epoch": 20.375, "grad_norm": 7.65625, "learning_rate": 1.8457998242883346e-05, "loss": 1.4766, "step": 8150 }, { "epoch": 20.3775, "grad_norm": 8.9375, "learning_rate": 1.845756624585977e-05, "loss": 1.6443, "step": 8151 }, { "epoch": 20.38, "grad_norm": 4.6875, "learning_rate": 1.845713419338873e-05, "loss": 1.4196, "step": 8152 }, { "epoch": 20.3825, "grad_norm": 6.5625, "learning_rate": 1.845670208547306e-05, "loss": 1.3849, "step": 8153 }, { "epoch": 20.385, "grad_norm": 8.0, "learning_rate": 1.8456269922115588e-05, "loss": 1.3634, "step": 8154 }, { "epoch": 20.3875, "grad_norm": 9.0625, "learning_rate": 1.845583770331915e-05, "loss": 1.2882, "step": 8155 }, { "epoch": 20.39, "grad_norm": 8.6875, "learning_rate": 1.8455405429086576e-05, "loss": 1.3897, "step": 8156 }, { "epoch": 20.3925, "grad_norm": 8.9375, "learning_rate": 1.84549730994207e-05, "loss": 1.5924, "step": 8157 }, { "epoch": 20.395, "grad_norm": 4.53125, "learning_rate": 1.8454540714324362e-05, "loss": 1.4341, "step": 8158 }, { "epoch": 20.3975, "grad_norm": 7.5625, "learning_rate": 1.8454108273800394e-05, "loss": 1.3943, "step": 8159 }, { "epoch": 20.4, "grad_norm": 6.1875, "learning_rate": 1.8453675777851627e-05, "loss": 1.4041, "step": 8160 }, { "epoch": 20.4025, "grad_norm": 9.375, "learning_rate": 1.8453243226480905e-05, "loss": 1.3718, "step": 8161 }, { "epoch": 20.405, "grad_norm": 5.125, "learning_rate": 1.8452810619691054e-05, "loss": 1.4363, "step": 8162 }, { "epoch": 20.4075, "grad_norm": 9.375, "learning_rate": 1.8452377957484915e-05, "loss": 1.4772, "step": 8163 }, { "epoch": 20.41, "grad_norm": 7.78125, "learning_rate": 1.8451945239865326e-05, "loss": 1.4797, "step": 8164 }, { "epoch": 20.4125, "grad_norm": 7.75, "learning_rate": 1.8451512466835125e-05, "loss": 1.3932, "step": 8165 }, { "epoch": 20.415, "grad_norm": 9.875, "learning_rate": 1.845107963839714e-05, "loss": 1.6426, "step": 8166 }, { "epoch": 20.4175, "grad_norm": 5.03125, "learning_rate": 1.845064675455422e-05, "loss": 1.4457, "step": 8167 }, { "epoch": 20.42, "grad_norm": 7.84375, "learning_rate": 1.84502138153092e-05, "loss": 1.4152, "step": 8168 }, { "epoch": 20.4225, "grad_norm": 6.5, "learning_rate": 1.844978082066491e-05, "loss": 1.373, "step": 8169 }, { "epoch": 20.425, "grad_norm": 9.25, "learning_rate": 1.8449347770624197e-05, "loss": 1.4418, "step": 8170 }, { "epoch": 20.4275, "grad_norm": 7.03125, "learning_rate": 1.8448914665189902e-05, "loss": 1.4218, "step": 8171 }, { "epoch": 20.43, "grad_norm": 6.59375, "learning_rate": 1.8448481504364856e-05, "loss": 1.2863, "step": 8172 }, { "epoch": 20.4325, "grad_norm": 9.9375, "learning_rate": 1.8448048288151906e-05, "loss": 1.459, "step": 8173 }, { "epoch": 20.435, "grad_norm": 6.625, "learning_rate": 1.844761501655389e-05, "loss": 1.4424, "step": 8174 }, { "epoch": 20.4375, "grad_norm": 7.84375, "learning_rate": 1.8447181689573646e-05, "loss": 1.3978, "step": 8175 }, { "epoch": 20.44, "grad_norm": 9.0625, "learning_rate": 1.844674830721402e-05, "loss": 1.4238, "step": 8176 }, { "epoch": 20.4425, "grad_norm": 7.4375, "learning_rate": 1.8446314869477845e-05, "loss": 1.4493, "step": 8177 }, { "epoch": 20.445, "grad_norm": 6.90625, "learning_rate": 1.844588137636797e-05, "loss": 1.2872, "step": 8178 }, { "epoch": 20.4475, "grad_norm": 6.78125, "learning_rate": 1.8445447827887238e-05, "loss": 1.3808, "step": 8179 }, { "epoch": 20.45, "grad_norm": 6.46875, "learning_rate": 1.8445014224038485e-05, "loss": 1.3103, "step": 8180 }, { "epoch": 20.4525, "grad_norm": 5.53125, "learning_rate": 1.8444580564824556e-05, "loss": 1.4778, "step": 8181 }, { "epoch": 20.455, "grad_norm": 31.125, "learning_rate": 1.8444146850248295e-05, "loss": 1.4379, "step": 8182 }, { "epoch": 20.4575, "grad_norm": 5.75, "learning_rate": 1.8443713080312547e-05, "loss": 1.376, "step": 8183 }, { "epoch": 20.46, "grad_norm": 10.125, "learning_rate": 1.8443279255020153e-05, "loss": 1.4609, "step": 8184 }, { "epoch": 20.4625, "grad_norm": 5.59375, "learning_rate": 1.8442845374373955e-05, "loss": 1.3011, "step": 8185 }, { "epoch": 20.465, "grad_norm": 6.4375, "learning_rate": 1.8442411438376808e-05, "loss": 1.3714, "step": 8186 }, { "epoch": 20.4675, "grad_norm": 5.8125, "learning_rate": 1.8441977447031542e-05, "loss": 1.3868, "step": 8187 }, { "epoch": 20.47, "grad_norm": 9.25, "learning_rate": 1.8441543400341012e-05, "loss": 1.6013, "step": 8188 }, { "epoch": 20.4725, "grad_norm": 7.6875, "learning_rate": 1.8441109298308063e-05, "loss": 1.4054, "step": 8189 }, { "epoch": 20.475, "grad_norm": 12.3125, "learning_rate": 1.844067514093554e-05, "loss": 1.7258, "step": 8190 }, { "epoch": 20.4775, "grad_norm": 6.28125, "learning_rate": 1.8440240928226286e-05, "loss": 1.3448, "step": 8191 }, { "epoch": 20.48, "grad_norm": 7.9375, "learning_rate": 1.843980666018315e-05, "loss": 1.515, "step": 8192 }, { "epoch": 20.4825, "grad_norm": 5.15625, "learning_rate": 1.8439372336808982e-05, "loss": 1.2656, "step": 8193 }, { "epoch": 20.485, "grad_norm": 5.75, "learning_rate": 1.8438937958106626e-05, "loss": 1.4485, "step": 8194 }, { "epoch": 20.4875, "grad_norm": 7.15625, "learning_rate": 1.8438503524078927e-05, "loss": 1.3666, "step": 8195 }, { "epoch": 20.49, "grad_norm": 8.9375, "learning_rate": 1.8438069034728738e-05, "loss": 1.6086, "step": 8196 }, { "epoch": 20.4925, "grad_norm": 14.0625, "learning_rate": 1.843763449005891e-05, "loss": 1.4131, "step": 8197 }, { "epoch": 20.495, "grad_norm": 5.34375, "learning_rate": 1.8437199890072285e-05, "loss": 1.4736, "step": 8198 }, { "epoch": 20.4975, "grad_norm": 7.0625, "learning_rate": 1.8436765234771716e-05, "loss": 1.3884, "step": 8199 }, { "epoch": 20.5, "grad_norm": 4.875, "learning_rate": 1.8436330524160048e-05, "loss": 1.2556, "step": 8200 }, { "epoch": 20.5025, "grad_norm": 4.90625, "learning_rate": 1.8435895758240137e-05, "loss": 1.3416, "step": 8201 }, { "epoch": 20.505, "grad_norm": 5.09375, "learning_rate": 1.843546093701483e-05, "loss": 1.4237, "step": 8202 }, { "epoch": 20.5075, "grad_norm": 7.96875, "learning_rate": 1.843502606048698e-05, "loss": 1.5144, "step": 8203 }, { "epoch": 20.51, "grad_norm": 7.625, "learning_rate": 1.843459112865944e-05, "loss": 1.4, "step": 8204 }, { "epoch": 20.5125, "grad_norm": 17.875, "learning_rate": 1.8434156141535052e-05, "loss": 1.5007, "step": 8205 }, { "epoch": 20.515, "grad_norm": 5.75, "learning_rate": 1.843372109911668e-05, "loss": 1.2738, "step": 8206 }, { "epoch": 20.5175, "grad_norm": 5.8125, "learning_rate": 1.8433286001407166e-05, "loss": 1.5163, "step": 8207 }, { "epoch": 20.52, "grad_norm": 7.28125, "learning_rate": 1.8432850848409367e-05, "loss": 1.3646, "step": 8208 }, { "epoch": 20.5225, "grad_norm": 7.15625, "learning_rate": 1.8432415640126137e-05, "loss": 1.3891, "step": 8209 }, { "epoch": 20.525, "grad_norm": 8.6875, "learning_rate": 1.8431980376560322e-05, "loss": 1.5873, "step": 8210 }, { "epoch": 20.5275, "grad_norm": 6.78125, "learning_rate": 1.8431545057714786e-05, "loss": 1.3264, "step": 8211 }, { "epoch": 20.53, "grad_norm": 5.875, "learning_rate": 1.843110968359238e-05, "loss": 1.3528, "step": 8212 }, { "epoch": 20.5325, "grad_norm": 8.625, "learning_rate": 1.843067425419595e-05, "loss": 1.495, "step": 8213 }, { "epoch": 20.535, "grad_norm": 6.21875, "learning_rate": 1.8430238769528364e-05, "loss": 1.3184, "step": 8214 }, { "epoch": 20.5375, "grad_norm": 5.15625, "learning_rate": 1.8429803229592464e-05, "loss": 1.3758, "step": 8215 }, { "epoch": 20.54, "grad_norm": 7.4375, "learning_rate": 1.8429367634391116e-05, "loss": 1.5175, "step": 8216 }, { "epoch": 20.5425, "grad_norm": 5.40625, "learning_rate": 1.8428931983927167e-05, "loss": 1.4861, "step": 8217 }, { "epoch": 20.545, "grad_norm": 7.375, "learning_rate": 1.842849627820348e-05, "loss": 1.3445, "step": 8218 }, { "epoch": 20.5475, "grad_norm": 10.4375, "learning_rate": 1.8428060517222906e-05, "loss": 1.4676, "step": 8219 }, { "epoch": 20.55, "grad_norm": 6.125, "learning_rate": 1.8427624700988308e-05, "loss": 1.4702, "step": 8220 }, { "epoch": 20.5525, "grad_norm": 7.6875, "learning_rate": 1.8427188829502536e-05, "loss": 1.3808, "step": 8221 }, { "epoch": 20.555, "grad_norm": 6.875, "learning_rate": 1.8426752902768454e-05, "loss": 1.29, "step": 8222 }, { "epoch": 20.5575, "grad_norm": 6.9375, "learning_rate": 1.8426316920788914e-05, "loss": 1.4461, "step": 8223 }, { "epoch": 20.56, "grad_norm": 10.125, "learning_rate": 1.8425880883566784e-05, "loss": 1.4224, "step": 8224 }, { "epoch": 20.5625, "grad_norm": 6.625, "learning_rate": 1.8425444791104912e-05, "loss": 1.3846, "step": 8225 }, { "epoch": 20.565, "grad_norm": 8.9375, "learning_rate": 1.8425008643406158e-05, "loss": 1.3186, "step": 8226 }, { "epoch": 20.5675, "grad_norm": 7.15625, "learning_rate": 1.842457244047339e-05, "loss": 1.2367, "step": 8227 }, { "epoch": 20.57, "grad_norm": 6.59375, "learning_rate": 1.842413618230946e-05, "loss": 1.3973, "step": 8228 }, { "epoch": 20.5725, "grad_norm": 6.96875, "learning_rate": 1.842369986891723e-05, "loss": 1.4485, "step": 8229 }, { "epoch": 20.575, "grad_norm": 7.71875, "learning_rate": 1.8423263500299562e-05, "loss": 1.4905, "step": 8230 }, { "epoch": 20.5775, "grad_norm": 11.0, "learning_rate": 1.8422827076459313e-05, "loss": 1.454, "step": 8231 }, { "epoch": 20.58, "grad_norm": 6.96875, "learning_rate": 1.842239059739935e-05, "loss": 1.493, "step": 8232 }, { "epoch": 20.5825, "grad_norm": 9.125, "learning_rate": 1.842195406312253e-05, "loss": 1.4954, "step": 8233 }, { "epoch": 20.585, "grad_norm": 6.34375, "learning_rate": 1.8421517473631713e-05, "loss": 1.3877, "step": 8234 }, { "epoch": 20.5875, "grad_norm": 6.71875, "learning_rate": 1.842108082892977e-05, "loss": 1.2204, "step": 8235 }, { "epoch": 20.59, "grad_norm": 5.8125, "learning_rate": 1.8420644129019555e-05, "loss": 1.2831, "step": 8236 }, { "epoch": 20.5925, "grad_norm": 6.625, "learning_rate": 1.8420207373903934e-05, "loss": 1.2493, "step": 8237 }, { "epoch": 20.595, "grad_norm": 7.625, "learning_rate": 1.841977056358577e-05, "loss": 1.4209, "step": 8238 }, { "epoch": 20.5975, "grad_norm": 6.71875, "learning_rate": 1.8419333698067928e-05, "loss": 1.4113, "step": 8239 }, { "epoch": 20.6, "grad_norm": 8.375, "learning_rate": 1.8418896777353272e-05, "loss": 1.5048, "step": 8240 }, { "epoch": 20.6025, "grad_norm": 8.6875, "learning_rate": 1.8418459801444663e-05, "loss": 1.4142, "step": 8241 }, { "epoch": 20.605, "grad_norm": 5.625, "learning_rate": 1.8418022770344972e-05, "loss": 1.4402, "step": 8242 }, { "epoch": 20.6075, "grad_norm": 8.875, "learning_rate": 1.841758568405706e-05, "loss": 1.5147, "step": 8243 }, { "epoch": 20.61, "grad_norm": 10.1875, "learning_rate": 1.8417148542583788e-05, "loss": 1.5031, "step": 8244 }, { "epoch": 20.6125, "grad_norm": 7.0, "learning_rate": 1.841671134592803e-05, "loss": 1.3829, "step": 8245 }, { "epoch": 20.615, "grad_norm": 4.65625, "learning_rate": 1.8416274094092652e-05, "loss": 1.4115, "step": 8246 }, { "epoch": 20.6175, "grad_norm": 6.9375, "learning_rate": 1.8415836787080517e-05, "loss": 1.4512, "step": 8247 }, { "epoch": 20.62, "grad_norm": 6.75, "learning_rate": 1.8415399424894492e-05, "loss": 1.3685, "step": 8248 }, { "epoch": 20.6225, "grad_norm": 5.5625, "learning_rate": 1.8414962007537443e-05, "loss": 1.4327, "step": 8249 }, { "epoch": 20.625, "grad_norm": 6.59375, "learning_rate": 1.8414524535012244e-05, "loss": 1.438, "step": 8250 }, { "epoch": 20.6275, "grad_norm": 8.4375, "learning_rate": 1.8414087007321754e-05, "loss": 1.3499, "step": 8251 }, { "epoch": 20.63, "grad_norm": 6.15625, "learning_rate": 1.841364942446885e-05, "loss": 1.4542, "step": 8252 }, { "epoch": 20.6325, "grad_norm": 6.71875, "learning_rate": 1.8413211786456395e-05, "loss": 1.356, "step": 8253 }, { "epoch": 20.635, "grad_norm": 10.3125, "learning_rate": 1.8412774093287262e-05, "loss": 1.5756, "step": 8254 }, { "epoch": 20.6375, "grad_norm": 5.8125, "learning_rate": 1.841233634496432e-05, "loss": 1.4696, "step": 8255 }, { "epoch": 20.64, "grad_norm": 7.5625, "learning_rate": 1.8411898541490433e-05, "loss": 1.4333, "step": 8256 }, { "epoch": 20.6425, "grad_norm": 7.46875, "learning_rate": 1.8411460682868477e-05, "loss": 1.3553, "step": 8257 }, { "epoch": 20.645, "grad_norm": 6.59375, "learning_rate": 1.8411022769101326e-05, "loss": 1.4818, "step": 8258 }, { "epoch": 20.6475, "grad_norm": 4.28125, "learning_rate": 1.841058480019184e-05, "loss": 1.2455, "step": 8259 }, { "epoch": 20.65, "grad_norm": 7.375, "learning_rate": 1.84101467761429e-05, "loss": 1.4011, "step": 8260 }, { "epoch": 20.6525, "grad_norm": 5.5625, "learning_rate": 1.8409708696957374e-05, "loss": 1.467, "step": 8261 }, { "epoch": 20.655, "grad_norm": 8.5625, "learning_rate": 1.8409270562638137e-05, "loss": 1.4568, "step": 8262 }, { "epoch": 20.6575, "grad_norm": 7.03125, "learning_rate": 1.8408832373188056e-05, "loss": 1.3545, "step": 8263 }, { "epoch": 20.66, "grad_norm": 4.9375, "learning_rate": 1.8408394128610005e-05, "loss": 1.4467, "step": 8264 }, { "epoch": 20.6625, "grad_norm": 8.4375, "learning_rate": 1.840795582890686e-05, "loss": 1.5817, "step": 8265 }, { "epoch": 20.665, "grad_norm": 8.4375, "learning_rate": 1.8407517474081494e-05, "loss": 1.3538, "step": 8266 }, { "epoch": 20.6675, "grad_norm": 6.0625, "learning_rate": 1.8407079064136777e-05, "loss": 1.3925, "step": 8267 }, { "epoch": 20.67, "grad_norm": 7.8125, "learning_rate": 1.8406640599075585e-05, "loss": 1.4881, "step": 8268 }, { "epoch": 20.6725, "grad_norm": 6.03125, "learning_rate": 1.84062020789008e-05, "loss": 1.3947, "step": 8269 }, { "epoch": 20.675, "grad_norm": 6.90625, "learning_rate": 1.8405763503615285e-05, "loss": 1.4886, "step": 8270 }, { "epoch": 20.6775, "grad_norm": 7.4375, "learning_rate": 1.8405324873221922e-05, "loss": 1.3803, "step": 8271 }, { "epoch": 20.68, "grad_norm": 8.0625, "learning_rate": 1.840488618772359e-05, "loss": 1.4658, "step": 8272 }, { "epoch": 20.6825, "grad_norm": 6.15625, "learning_rate": 1.8404447447123155e-05, "loss": 1.3594, "step": 8273 }, { "epoch": 20.685, "grad_norm": 7.375, "learning_rate": 1.8404008651423503e-05, "loss": 1.4276, "step": 8274 }, { "epoch": 20.6875, "grad_norm": 9.5625, "learning_rate": 1.84035698006275e-05, "loss": 1.333, "step": 8275 }, { "epoch": 20.69, "grad_norm": 15.375, "learning_rate": 1.8403130894738038e-05, "loss": 1.6342, "step": 8276 }, { "epoch": 20.6925, "grad_norm": 7.4375, "learning_rate": 1.840269193375798e-05, "loss": 1.2973, "step": 8277 }, { "epoch": 20.695, "grad_norm": 5.4375, "learning_rate": 1.8402252917690213e-05, "loss": 1.3974, "step": 8278 }, { "epoch": 20.6975, "grad_norm": 6.78125, "learning_rate": 1.8401813846537612e-05, "loss": 1.336, "step": 8279 }, { "epoch": 20.7, "grad_norm": 4.78125, "learning_rate": 1.8401374720303054e-05, "loss": 1.3067, "step": 8280 }, { "epoch": 20.7025, "grad_norm": 6.5625, "learning_rate": 1.840093553898942e-05, "loss": 1.4065, "step": 8281 }, { "epoch": 20.705, "grad_norm": 5.6875, "learning_rate": 1.840049630259959e-05, "loss": 1.4112, "step": 8282 }, { "epoch": 20.7075, "grad_norm": 6.25, "learning_rate": 1.8400057011136443e-05, "loss": 1.3609, "step": 8283 }, { "epoch": 20.71, "grad_norm": 6.1875, "learning_rate": 1.8399617664602856e-05, "loss": 1.4295, "step": 8284 }, { "epoch": 20.7125, "grad_norm": 7.125, "learning_rate": 1.8399178263001712e-05, "loss": 1.3596, "step": 8285 }, { "epoch": 20.715, "grad_norm": 6.1875, "learning_rate": 1.8398738806335892e-05, "loss": 1.3434, "step": 8286 }, { "epoch": 20.7175, "grad_norm": 7.90625, "learning_rate": 1.8398299294608278e-05, "loss": 1.4142, "step": 8287 }, { "epoch": 20.72, "grad_norm": 6.28125, "learning_rate": 1.8397859727821747e-05, "loss": 1.3623, "step": 8288 }, { "epoch": 20.7225, "grad_norm": 7.0, "learning_rate": 1.8397420105979186e-05, "loss": 1.4179, "step": 8289 }, { "epoch": 20.725, "grad_norm": 6.34375, "learning_rate": 1.8396980429083475e-05, "loss": 1.3712, "step": 8290 }, { "epoch": 20.7275, "grad_norm": 7.96875, "learning_rate": 1.8396540697137496e-05, "loss": 1.4225, "step": 8291 }, { "epoch": 20.73, "grad_norm": 6.4375, "learning_rate": 1.839610091014413e-05, "loss": 1.6505, "step": 8292 }, { "epoch": 20.7325, "grad_norm": 8.625, "learning_rate": 1.8395661068106264e-05, "loss": 1.3377, "step": 8293 }, { "epoch": 20.735, "grad_norm": 5.1875, "learning_rate": 1.839522117102678e-05, "loss": 1.227, "step": 8294 }, { "epoch": 20.7375, "grad_norm": 7.5, "learning_rate": 1.8394781218908564e-05, "loss": 1.3664, "step": 8295 }, { "epoch": 20.74, "grad_norm": 5.5, "learning_rate": 1.8394341211754496e-05, "loss": 1.6667, "step": 8296 }, { "epoch": 20.7425, "grad_norm": 9.5625, "learning_rate": 1.839390114956746e-05, "loss": 1.317, "step": 8297 }, { "epoch": 20.745, "grad_norm": 6.96875, "learning_rate": 1.839346103235035e-05, "loss": 1.3098, "step": 8298 }, { "epoch": 20.7475, "grad_norm": 6.375, "learning_rate": 1.839302086010604e-05, "loss": 1.554, "step": 8299 }, { "epoch": 20.75, "grad_norm": 6.625, "learning_rate": 1.8392580632837423e-05, "loss": 1.3427, "step": 8300 }, { "epoch": 20.7525, "grad_norm": 6.15625, "learning_rate": 1.8392140350547386e-05, "loss": 1.3585, "step": 8301 }, { "epoch": 20.755, "grad_norm": 9.5, "learning_rate": 1.8391700013238812e-05, "loss": 1.4424, "step": 8302 }, { "epoch": 20.7575, "grad_norm": 5.625, "learning_rate": 1.8391259620914584e-05, "loss": 1.3671, "step": 8303 }, { "epoch": 20.76, "grad_norm": 5.78125, "learning_rate": 1.83908191735776e-05, "loss": 1.3109, "step": 8304 }, { "epoch": 20.7625, "grad_norm": 5.40625, "learning_rate": 1.8390378671230736e-05, "loss": 1.4149, "step": 8305 }, { "epoch": 20.765, "grad_norm": 7.15625, "learning_rate": 1.8389938113876887e-05, "loss": 1.4518, "step": 8306 }, { "epoch": 20.7675, "grad_norm": 7.4375, "learning_rate": 1.838949750151894e-05, "loss": 1.503, "step": 8307 }, { "epoch": 20.77, "grad_norm": 6.65625, "learning_rate": 1.838905683415978e-05, "loss": 1.3252, "step": 8308 }, { "epoch": 20.7725, "grad_norm": 7.625, "learning_rate": 1.8388616111802307e-05, "loss": 1.525, "step": 8309 }, { "epoch": 20.775, "grad_norm": 6.9375, "learning_rate": 1.8388175334449397e-05, "loss": 1.2357, "step": 8310 }, { "epoch": 20.7775, "grad_norm": 8.25, "learning_rate": 1.8387734502103943e-05, "loss": 1.4585, "step": 8311 }, { "epoch": 20.78, "grad_norm": 5.5, "learning_rate": 1.8387293614768843e-05, "loss": 1.294, "step": 8312 }, { "epoch": 20.7825, "grad_norm": 9.25, "learning_rate": 1.838685267244698e-05, "loss": 1.3903, "step": 8313 }, { "epoch": 20.785, "grad_norm": 5.96875, "learning_rate": 1.8386411675141242e-05, "loss": 1.4059, "step": 8314 }, { "epoch": 20.7875, "grad_norm": 9.1875, "learning_rate": 1.838597062285453e-05, "loss": 1.399, "step": 8315 }, { "epoch": 20.79, "grad_norm": 9.0625, "learning_rate": 1.8385529515589726e-05, "loss": 1.4318, "step": 8316 }, { "epoch": 20.7925, "grad_norm": 7.625, "learning_rate": 1.838508835334973e-05, "loss": 1.3383, "step": 8317 }, { "epoch": 20.795, "grad_norm": 6.375, "learning_rate": 1.8384647136137424e-05, "loss": 1.3787, "step": 8318 }, { "epoch": 20.7975, "grad_norm": 6.78125, "learning_rate": 1.8384205863955716e-05, "loss": 1.3362, "step": 8319 }, { "epoch": 20.8, "grad_norm": 8.5625, "learning_rate": 1.8383764536807486e-05, "loss": 1.3967, "step": 8320 }, { "epoch": 20.8025, "grad_norm": 7.03125, "learning_rate": 1.838332315469563e-05, "loss": 1.4709, "step": 8321 }, { "epoch": 20.805, "grad_norm": 6.4375, "learning_rate": 1.838288171762304e-05, "loss": 1.2705, "step": 8322 }, { "epoch": 20.8075, "grad_norm": 5.40625, "learning_rate": 1.8382440225592617e-05, "loss": 1.4062, "step": 8323 }, { "epoch": 20.81, "grad_norm": 5.15625, "learning_rate": 1.838199867860725e-05, "loss": 1.3761, "step": 8324 }, { "epoch": 20.8125, "grad_norm": 5.46875, "learning_rate": 1.8381557076669837e-05, "loss": 1.4295, "step": 8325 }, { "epoch": 20.815, "grad_norm": 8.75, "learning_rate": 1.8381115419783268e-05, "loss": 1.3597, "step": 8326 }, { "epoch": 20.8175, "grad_norm": 6.40625, "learning_rate": 1.8380673707950443e-05, "loss": 1.3853, "step": 8327 }, { "epoch": 20.82, "grad_norm": 5.90625, "learning_rate": 1.8380231941174258e-05, "loss": 1.3797, "step": 8328 }, { "epoch": 20.8225, "grad_norm": 7.59375, "learning_rate": 1.8379790119457608e-05, "loss": 1.4438, "step": 8329 }, { "epoch": 20.825, "grad_norm": 5.0, "learning_rate": 1.8379348242803387e-05, "loss": 1.4335, "step": 8330 }, { "epoch": 20.8275, "grad_norm": 4.5, "learning_rate": 1.8378906311214495e-05, "loss": 1.4358, "step": 8331 }, { "epoch": 20.83, "grad_norm": 7.5625, "learning_rate": 1.837846432469383e-05, "loss": 1.6325, "step": 8332 }, { "epoch": 20.8325, "grad_norm": 6.40625, "learning_rate": 1.8378022283244287e-05, "loss": 1.4849, "step": 8333 }, { "epoch": 20.835, "grad_norm": 5.84375, "learning_rate": 1.8377580186868764e-05, "loss": 1.3522, "step": 8334 }, { "epoch": 20.8375, "grad_norm": 8.0625, "learning_rate": 1.837713803557016e-05, "loss": 1.3428, "step": 8335 }, { "epoch": 20.84, "grad_norm": 8.5, "learning_rate": 1.8376695829351378e-05, "loss": 1.3681, "step": 8336 }, { "epoch": 20.8425, "grad_norm": 6.71875, "learning_rate": 1.8376253568215312e-05, "loss": 1.4015, "step": 8337 }, { "epoch": 20.845, "grad_norm": 10.375, "learning_rate": 1.837581125216486e-05, "loss": 1.4117, "step": 8338 }, { "epoch": 20.8475, "grad_norm": 10.75, "learning_rate": 1.837536888120293e-05, "loss": 1.5397, "step": 8339 }, { "epoch": 20.85, "grad_norm": 7.40625, "learning_rate": 1.837492645533241e-05, "loss": 1.4786, "step": 8340 }, { "epoch": 20.8525, "grad_norm": 9.6875, "learning_rate": 1.8374483974556212e-05, "loss": 1.4446, "step": 8341 }, { "epoch": 20.855, "grad_norm": 6.875, "learning_rate": 1.837404143887723e-05, "loss": 1.5297, "step": 8342 }, { "epoch": 20.8575, "grad_norm": 6.4375, "learning_rate": 1.8373598848298368e-05, "loss": 1.3793, "step": 8343 }, { "epoch": 20.86, "grad_norm": 7.6875, "learning_rate": 1.8373156202822532e-05, "loss": 1.4884, "step": 8344 }, { "epoch": 20.8625, "grad_norm": 7.0625, "learning_rate": 1.8372713502452612e-05, "loss": 1.3633, "step": 8345 }, { "epoch": 20.865, "grad_norm": 7.09375, "learning_rate": 1.8372270747191522e-05, "loss": 1.5696, "step": 8346 }, { "epoch": 20.8675, "grad_norm": 8.8125, "learning_rate": 1.8371827937042158e-05, "loss": 1.4849, "step": 8347 }, { "epoch": 20.87, "grad_norm": 6.59375, "learning_rate": 1.8371385072007423e-05, "loss": 1.4837, "step": 8348 }, { "epoch": 20.8725, "grad_norm": 6.9375, "learning_rate": 1.8370942152090223e-05, "loss": 1.301, "step": 8349 }, { "epoch": 20.875, "grad_norm": 5.53125, "learning_rate": 1.8370499177293463e-05, "loss": 1.3644, "step": 8350 }, { "epoch": 20.8775, "grad_norm": 7.3125, "learning_rate": 1.8370056147620047e-05, "loss": 1.3315, "step": 8351 }, { "epoch": 20.88, "grad_norm": 7.0, "learning_rate": 1.8369613063072875e-05, "loss": 1.3693, "step": 8352 }, { "epoch": 20.8825, "grad_norm": 6.8125, "learning_rate": 1.8369169923654855e-05, "loss": 1.4145, "step": 8353 }, { "epoch": 20.885, "grad_norm": 10.625, "learning_rate": 1.8368726729368894e-05, "loss": 1.3673, "step": 8354 }, { "epoch": 20.8875, "grad_norm": 9.6875, "learning_rate": 1.8368283480217894e-05, "loss": 1.6355, "step": 8355 }, { "epoch": 20.89, "grad_norm": 8.5, "learning_rate": 1.836784017620476e-05, "loss": 1.5823, "step": 8356 }, { "epoch": 20.8925, "grad_norm": 5.5625, "learning_rate": 1.8367396817332403e-05, "loss": 1.4114, "step": 8357 }, { "epoch": 20.895, "grad_norm": 13.0, "learning_rate": 1.836695340360373e-05, "loss": 1.4548, "step": 8358 }, { "epoch": 20.8975, "grad_norm": 9.4375, "learning_rate": 1.836650993502164e-05, "loss": 1.4455, "step": 8359 }, { "epoch": 20.9, "grad_norm": 6.25, "learning_rate": 1.836606641158905e-05, "loss": 1.3875, "step": 8360 }, { "epoch": 20.9025, "grad_norm": 10.375, "learning_rate": 1.836562283330886e-05, "loss": 1.6819, "step": 8361 }, { "epoch": 20.905, "grad_norm": 5.78125, "learning_rate": 1.8365179200183983e-05, "loss": 1.4377, "step": 8362 }, { "epoch": 20.9075, "grad_norm": 5.875, "learning_rate": 1.8364735512217326e-05, "loss": 1.4665, "step": 8363 }, { "epoch": 20.91, "grad_norm": 7.125, "learning_rate": 1.83642917694118e-05, "loss": 1.5279, "step": 8364 }, { "epoch": 20.9125, "grad_norm": 6.46875, "learning_rate": 1.836384797177031e-05, "loss": 1.2024, "step": 8365 }, { "epoch": 20.915, "grad_norm": 8.25, "learning_rate": 1.8363404119295767e-05, "loss": 1.5781, "step": 8366 }, { "epoch": 20.9175, "grad_norm": 6.59375, "learning_rate": 1.836296021199108e-05, "loss": 1.3661, "step": 8367 }, { "epoch": 20.92, "grad_norm": 6.59375, "learning_rate": 1.8362516249859164e-05, "loss": 1.5194, "step": 8368 }, { "epoch": 20.9225, "grad_norm": 4.8125, "learning_rate": 1.8362072232902922e-05, "loss": 1.3465, "step": 8369 }, { "epoch": 20.925, "grad_norm": 7.03125, "learning_rate": 1.836162816112527e-05, "loss": 1.5568, "step": 8370 }, { "epoch": 20.9275, "grad_norm": 9.4375, "learning_rate": 1.836118403452912e-05, "loss": 1.4153, "step": 8371 }, { "epoch": 20.93, "grad_norm": 5.78125, "learning_rate": 1.8360739853117383e-05, "loss": 1.3963, "step": 8372 }, { "epoch": 20.9325, "grad_norm": 7.21875, "learning_rate": 1.8360295616892968e-05, "loss": 1.4916, "step": 8373 }, { "epoch": 20.935, "grad_norm": 6.59375, "learning_rate": 1.8359851325858792e-05, "loss": 1.1721, "step": 8374 }, { "epoch": 20.9375, "grad_norm": 6.625, "learning_rate": 1.8359406980017763e-05, "loss": 1.4261, "step": 8375 }, { "epoch": 20.94, "grad_norm": 8.125, "learning_rate": 1.8358962579372797e-05, "loss": 1.3988, "step": 8376 }, { "epoch": 20.9425, "grad_norm": 9.0625, "learning_rate": 1.8358518123926805e-05, "loss": 1.3972, "step": 8377 }, { "epoch": 20.945, "grad_norm": 7.03125, "learning_rate": 1.8358073613682705e-05, "loss": 1.3989, "step": 8378 }, { "epoch": 20.9475, "grad_norm": 8.0, "learning_rate": 1.835762904864341e-05, "loss": 1.5183, "step": 8379 }, { "epoch": 20.95, "grad_norm": 4.15625, "learning_rate": 1.835718442881183e-05, "loss": 1.2745, "step": 8380 }, { "epoch": 20.9525, "grad_norm": 9.25, "learning_rate": 1.8356739754190886e-05, "loss": 1.4018, "step": 8381 }, { "epoch": 20.955, "grad_norm": 6.40625, "learning_rate": 1.8356295024783488e-05, "loss": 1.4921, "step": 8382 }, { "epoch": 20.9575, "grad_norm": 7.0, "learning_rate": 1.8355850240592554e-05, "loss": 1.3509, "step": 8383 }, { "epoch": 20.96, "grad_norm": 5.65625, "learning_rate": 1.8355405401621e-05, "loss": 1.4341, "step": 8384 }, { "epoch": 20.9625, "grad_norm": 5.71875, "learning_rate": 1.8354960507871746e-05, "loss": 1.3305, "step": 8385 }, { "epoch": 20.965, "grad_norm": 7.03125, "learning_rate": 1.8354515559347703e-05, "loss": 1.45, "step": 8386 }, { "epoch": 20.9675, "grad_norm": 7.21875, "learning_rate": 1.8354070556051792e-05, "loss": 1.3635, "step": 8387 }, { "epoch": 20.97, "grad_norm": 5.5625, "learning_rate": 1.8353625497986928e-05, "loss": 1.3917, "step": 8388 }, { "epoch": 20.9725, "grad_norm": 6.90625, "learning_rate": 1.835318038515603e-05, "loss": 1.5286, "step": 8389 }, { "epoch": 20.975, "grad_norm": 5.34375, "learning_rate": 1.8352735217562015e-05, "loss": 1.2874, "step": 8390 }, { "epoch": 20.9775, "grad_norm": 5.5625, "learning_rate": 1.8352289995207802e-05, "loss": 1.4643, "step": 8391 }, { "epoch": 20.98, "grad_norm": 6.28125, "learning_rate": 1.835184471809631e-05, "loss": 1.4682, "step": 8392 }, { "epoch": 20.9825, "grad_norm": 7.78125, "learning_rate": 1.835139938623046e-05, "loss": 1.4616, "step": 8393 }, { "epoch": 20.985, "grad_norm": 7.90625, "learning_rate": 1.8350953999613165e-05, "loss": 1.3626, "step": 8394 }, { "epoch": 20.9875, "grad_norm": 4.84375, "learning_rate": 1.8350508558247355e-05, "loss": 1.2967, "step": 8395 }, { "epoch": 20.99, "grad_norm": 8.5625, "learning_rate": 1.835006306213594e-05, "loss": 1.2955, "step": 8396 }, { "epoch": 20.9925, "grad_norm": 6.90625, "learning_rate": 1.834961751128185e-05, "loss": 1.4204, "step": 8397 }, { "epoch": 20.995, "grad_norm": 7.53125, "learning_rate": 1.8349171905688e-05, "loss": 1.3591, "step": 8398 }, { "epoch": 20.9975, "grad_norm": 5.875, "learning_rate": 1.8348726245357317e-05, "loss": 1.3897, "step": 8399 }, { "epoch": 21.0, "grad_norm": 9.3125, "learning_rate": 1.8348280530292712e-05, "loss": 1.4437, "step": 8400 }, { "epoch": 21.0025, "grad_norm": 6.84375, "learning_rate": 1.8347834760497117e-05, "loss": 1.3413, "step": 8401 }, { "epoch": 21.005, "grad_norm": 6.5, "learning_rate": 1.8347388935973454e-05, "loss": 1.614, "step": 8402 }, { "epoch": 21.0075, "grad_norm": 6.53125, "learning_rate": 1.834694305672464e-05, "loss": 1.4881, "step": 8403 }, { "epoch": 21.01, "grad_norm": 6.28125, "learning_rate": 1.8346497122753603e-05, "loss": 1.3409, "step": 8404 }, { "epoch": 21.0125, "grad_norm": 6.71875, "learning_rate": 1.8346051134063262e-05, "loss": 1.4165, "step": 8405 }, { "epoch": 21.015, "grad_norm": 6.125, "learning_rate": 1.834560509065655e-05, "loss": 1.4154, "step": 8406 }, { "epoch": 21.0175, "grad_norm": 7.46875, "learning_rate": 1.834515899253638e-05, "loss": 1.3973, "step": 8407 }, { "epoch": 21.02, "grad_norm": 6.21875, "learning_rate": 1.8344712839705682e-05, "loss": 1.419, "step": 8408 }, { "epoch": 21.0225, "grad_norm": 7.0625, "learning_rate": 1.834426663216738e-05, "loss": 1.6504, "step": 8409 }, { "epoch": 21.025, "grad_norm": 7.21875, "learning_rate": 1.83438203699244e-05, "loss": 1.2524, "step": 8410 }, { "epoch": 21.0275, "grad_norm": 7.3125, "learning_rate": 1.834337405297967e-05, "loss": 1.3406, "step": 8411 }, { "epoch": 21.03, "grad_norm": 7.21875, "learning_rate": 1.8342927681336112e-05, "loss": 1.453, "step": 8412 }, { "epoch": 21.0325, "grad_norm": 7.03125, "learning_rate": 1.834248125499665e-05, "loss": 1.4101, "step": 8413 }, { "epoch": 21.035, "grad_norm": 6.5625, "learning_rate": 1.834203477396422e-05, "loss": 1.5042, "step": 8414 }, { "epoch": 21.0375, "grad_norm": 8.25, "learning_rate": 1.8341588238241743e-05, "loss": 1.254, "step": 8415 }, { "epoch": 21.04, "grad_norm": 6.4375, "learning_rate": 1.834114164783215e-05, "loss": 1.4003, "step": 8416 }, { "epoch": 21.0425, "grad_norm": 7.15625, "learning_rate": 1.834069500273836e-05, "loss": 1.4331, "step": 8417 }, { "epoch": 21.045, "grad_norm": 7.8125, "learning_rate": 1.834024830296331e-05, "loss": 1.3161, "step": 8418 }, { "epoch": 21.0475, "grad_norm": 7.34375, "learning_rate": 1.8339801548509927e-05, "loss": 1.3217, "step": 8419 }, { "epoch": 21.05, "grad_norm": 7.5, "learning_rate": 1.8339354739381138e-05, "loss": 1.623, "step": 8420 }, { "epoch": 21.0525, "grad_norm": 6.625, "learning_rate": 1.8338907875579876e-05, "loss": 1.235, "step": 8421 }, { "epoch": 21.055, "grad_norm": 6.28125, "learning_rate": 1.8338460957109064e-05, "loss": 1.3218, "step": 8422 }, { "epoch": 21.0575, "grad_norm": 7.21875, "learning_rate": 1.8338013983971637e-05, "loss": 1.3839, "step": 8423 }, { "epoch": 21.06, "grad_norm": 7.53125, "learning_rate": 1.8337566956170523e-05, "loss": 1.3958, "step": 8424 }, { "epoch": 21.0625, "grad_norm": 6.71875, "learning_rate": 1.8337119873708654e-05, "loss": 1.4476, "step": 8425 }, { "epoch": 21.065, "grad_norm": 8.75, "learning_rate": 1.833667273658896e-05, "loss": 1.3906, "step": 8426 }, { "epoch": 21.0675, "grad_norm": 6.09375, "learning_rate": 1.8336225544814374e-05, "loss": 1.5189, "step": 8427 }, { "epoch": 21.07, "grad_norm": 6.75, "learning_rate": 1.8335778298387828e-05, "loss": 1.4722, "step": 8428 }, { "epoch": 21.0725, "grad_norm": 11.0625, "learning_rate": 1.833533099731225e-05, "loss": 1.2928, "step": 8429 }, { "epoch": 21.075, "grad_norm": 7.65625, "learning_rate": 1.8334883641590577e-05, "loss": 1.2822, "step": 8430 }, { "epoch": 21.0775, "grad_norm": 13.125, "learning_rate": 1.8334436231225744e-05, "loss": 1.5816, "step": 8431 }, { "epoch": 21.08, "grad_norm": 4.625, "learning_rate": 1.8333988766220676e-05, "loss": 1.3194, "step": 8432 }, { "epoch": 21.0825, "grad_norm": 7.90625, "learning_rate": 1.8333541246578313e-05, "loss": 1.3548, "step": 8433 }, { "epoch": 21.085, "grad_norm": 6.375, "learning_rate": 1.8333093672301587e-05, "loss": 1.3362, "step": 8434 }, { "epoch": 21.0875, "grad_norm": 7.34375, "learning_rate": 1.833264604339343e-05, "loss": 1.3817, "step": 8435 }, { "epoch": 21.09, "grad_norm": 8.9375, "learning_rate": 1.8332198359856782e-05, "loss": 1.374, "step": 8436 }, { "epoch": 21.0925, "grad_norm": 6.9375, "learning_rate": 1.8331750621694576e-05, "loss": 1.2798, "step": 8437 }, { "epoch": 21.095, "grad_norm": 8.25, "learning_rate": 1.8331302828909746e-05, "loss": 1.7214, "step": 8438 }, { "epoch": 21.0975, "grad_norm": 6.0625, "learning_rate": 1.8330854981505223e-05, "loss": 1.3876, "step": 8439 }, { "epoch": 21.1, "grad_norm": 9.8125, "learning_rate": 1.833040707948395e-05, "loss": 1.3406, "step": 8440 }, { "epoch": 21.1025, "grad_norm": 5.78125, "learning_rate": 1.8329959122848865e-05, "loss": 1.33, "step": 8441 }, { "epoch": 21.105, "grad_norm": 4.65625, "learning_rate": 1.83295111116029e-05, "loss": 1.4952, "step": 8442 }, { "epoch": 21.1075, "grad_norm": 13.875, "learning_rate": 1.8329063045748994e-05, "loss": 1.3795, "step": 8443 }, { "epoch": 21.11, "grad_norm": 5.625, "learning_rate": 1.832861492529008e-05, "loss": 1.2837, "step": 8444 }, { "epoch": 21.1125, "grad_norm": 7.1875, "learning_rate": 1.8328166750229105e-05, "loss": 1.2916, "step": 8445 }, { "epoch": 21.115, "grad_norm": 5.03125, "learning_rate": 1.8327718520569e-05, "loss": 1.3233, "step": 8446 }, { "epoch": 21.1175, "grad_norm": 11.1875, "learning_rate": 1.8327270236312707e-05, "loss": 1.3787, "step": 8447 }, { "epoch": 21.12, "grad_norm": 7.375, "learning_rate": 1.832682189746316e-05, "loss": 1.4562, "step": 8448 }, { "epoch": 21.1225, "grad_norm": 6.9375, "learning_rate": 1.8326373504023304e-05, "loss": 1.4816, "step": 8449 }, { "epoch": 21.125, "grad_norm": 6.15625, "learning_rate": 1.8325925055996076e-05, "loss": 1.4055, "step": 8450 }, { "epoch": 21.1275, "grad_norm": 19.125, "learning_rate": 1.832547655338442e-05, "loss": 1.3159, "step": 8451 }, { "epoch": 21.13, "grad_norm": 6.09375, "learning_rate": 1.8325027996191267e-05, "loss": 1.1755, "step": 8452 }, { "epoch": 21.1325, "grad_norm": 8.875, "learning_rate": 1.8324579384419567e-05, "loss": 1.5426, "step": 8453 }, { "epoch": 21.135, "grad_norm": 5.65625, "learning_rate": 1.8324130718072257e-05, "loss": 1.3353, "step": 8454 }, { "epoch": 21.1375, "grad_norm": 5.21875, "learning_rate": 1.8323681997152278e-05, "loss": 1.3188, "step": 8455 }, { "epoch": 21.14, "grad_norm": 8.3125, "learning_rate": 1.8323233221662574e-05, "loss": 1.4835, "step": 8456 }, { "epoch": 21.1425, "grad_norm": 6.5625, "learning_rate": 1.832278439160609e-05, "loss": 1.3806, "step": 8457 }, { "epoch": 21.145, "grad_norm": 6.5, "learning_rate": 1.8322335506985758e-05, "loss": 1.4868, "step": 8458 }, { "epoch": 21.1475, "grad_norm": 5.21875, "learning_rate": 1.832188656780453e-05, "loss": 1.3732, "step": 8459 }, { "epoch": 21.15, "grad_norm": 7.8125, "learning_rate": 1.8321437574065347e-05, "loss": 1.319, "step": 8460 }, { "epoch": 21.1525, "grad_norm": 11.25, "learning_rate": 1.8320988525771152e-05, "loss": 1.4944, "step": 8461 }, { "epoch": 21.155, "grad_norm": 6.65625, "learning_rate": 1.832053942292489e-05, "loss": 1.3347, "step": 8462 }, { "epoch": 21.1575, "grad_norm": 6.15625, "learning_rate": 1.8320090265529505e-05, "loss": 1.3892, "step": 8463 }, { "epoch": 21.16, "grad_norm": 5.3125, "learning_rate": 1.831964105358794e-05, "loss": 1.4831, "step": 8464 }, { "epoch": 21.1625, "grad_norm": 7.5625, "learning_rate": 1.831919178710314e-05, "loss": 1.4904, "step": 8465 }, { "epoch": 21.165, "grad_norm": 7.34375, "learning_rate": 1.8318742466078056e-05, "loss": 1.3766, "step": 8466 }, { "epoch": 21.1675, "grad_norm": 7.0625, "learning_rate": 1.8318293090515623e-05, "loss": 1.5255, "step": 8467 }, { "epoch": 21.17, "grad_norm": 6.21875, "learning_rate": 1.83178436604188e-05, "loss": 1.3032, "step": 8468 }, { "epoch": 21.1725, "grad_norm": 5.78125, "learning_rate": 1.8317394175790523e-05, "loss": 1.395, "step": 8469 }, { "epoch": 21.175, "grad_norm": 8.125, "learning_rate": 1.8316944636633742e-05, "loss": 1.5516, "step": 8470 }, { "epoch": 21.1775, "grad_norm": 9.4375, "learning_rate": 1.8316495042951404e-05, "loss": 1.4172, "step": 8471 }, { "epoch": 21.18, "grad_norm": 6.25, "learning_rate": 1.831604539474646e-05, "loss": 1.5172, "step": 8472 }, { "epoch": 21.1825, "grad_norm": 8.6875, "learning_rate": 1.8315595692021853e-05, "loss": 1.2286, "step": 8473 }, { "epoch": 21.185, "grad_norm": 7.125, "learning_rate": 1.8315145934780536e-05, "loss": 1.2756, "step": 8474 }, { "epoch": 21.1875, "grad_norm": 4.59375, "learning_rate": 1.8314696123025456e-05, "loss": 1.5697, "step": 8475 }, { "epoch": 21.19, "grad_norm": 5.15625, "learning_rate": 1.8314246256759556e-05, "loss": 1.2208, "step": 8476 }, { "epoch": 21.1925, "grad_norm": 6.78125, "learning_rate": 1.8313796335985793e-05, "loss": 1.4897, "step": 8477 }, { "epoch": 21.195, "grad_norm": 7.0625, "learning_rate": 1.8313346360707113e-05, "loss": 1.4472, "step": 8478 }, { "epoch": 21.1975, "grad_norm": 9.375, "learning_rate": 1.831289633092647e-05, "loss": 1.5554, "step": 8479 }, { "epoch": 21.2, "grad_norm": 6.59375, "learning_rate": 1.831244624664681e-05, "loss": 1.397, "step": 8480 }, { "epoch": 21.2025, "grad_norm": 9.5625, "learning_rate": 1.8311996107871085e-05, "loss": 1.3863, "step": 8481 }, { "epoch": 21.205, "grad_norm": 7.34375, "learning_rate": 1.831154591460224e-05, "loss": 1.2799, "step": 8482 }, { "epoch": 21.2075, "grad_norm": 7.875, "learning_rate": 1.831109566684324e-05, "loss": 1.3893, "step": 8483 }, { "epoch": 21.21, "grad_norm": 8.25, "learning_rate": 1.8310645364597028e-05, "loss": 1.4101, "step": 8484 }, { "epoch": 21.2125, "grad_norm": 7.75, "learning_rate": 1.8310195007866557e-05, "loss": 1.519, "step": 8485 }, { "epoch": 21.215, "grad_norm": 9.0625, "learning_rate": 1.830974459665478e-05, "loss": 1.561, "step": 8486 }, { "epoch": 21.2175, "grad_norm": 12.875, "learning_rate": 1.830929413096465e-05, "loss": 1.5195, "step": 8487 }, { "epoch": 21.22, "grad_norm": 5.40625, "learning_rate": 1.8308843610799122e-05, "loss": 1.3563, "step": 8488 }, { "epoch": 21.2225, "grad_norm": 6.40625, "learning_rate": 1.8308393036161144e-05, "loss": 1.2242, "step": 8489 }, { "epoch": 21.225, "grad_norm": 7.625, "learning_rate": 1.8307942407053673e-05, "loss": 1.2904, "step": 8490 }, { "epoch": 21.2275, "grad_norm": 10.25, "learning_rate": 1.830749172347967e-05, "loss": 1.4237, "step": 8491 }, { "epoch": 21.23, "grad_norm": 9.0, "learning_rate": 1.830704098544208e-05, "loss": 1.3237, "step": 8492 }, { "epoch": 21.2325, "grad_norm": 10.6875, "learning_rate": 1.830659019294386e-05, "loss": 1.4003, "step": 8493 }, { "epoch": 21.235, "grad_norm": 7.875, "learning_rate": 1.8306139345987972e-05, "loss": 1.3505, "step": 8494 }, { "epoch": 21.2375, "grad_norm": 7.46875, "learning_rate": 1.8305688444577362e-05, "loss": 1.3422, "step": 8495 }, { "epoch": 21.24, "grad_norm": 7.71875, "learning_rate": 1.8305237488714995e-05, "loss": 1.2929, "step": 8496 }, { "epoch": 21.2425, "grad_norm": 7.84375, "learning_rate": 1.830478647840382e-05, "loss": 1.3655, "step": 8497 }, { "epoch": 21.245, "grad_norm": 4.4375, "learning_rate": 1.8304335413646798e-05, "loss": 1.3992, "step": 8498 }, { "epoch": 21.2475, "grad_norm": 5.875, "learning_rate": 1.8303884294446886e-05, "loss": 1.3086, "step": 8499 }, { "epoch": 21.25, "grad_norm": 7.9375, "learning_rate": 1.8303433120807043e-05, "loss": 1.3836, "step": 8500 }, { "epoch": 21.2525, "grad_norm": 6.90625, "learning_rate": 1.830298189273022e-05, "loss": 1.352, "step": 8501 }, { "epoch": 21.255, "grad_norm": 8.0625, "learning_rate": 1.8302530610219383e-05, "loss": 1.4298, "step": 8502 }, { "epoch": 21.2575, "grad_norm": 10.625, "learning_rate": 1.8302079273277486e-05, "loss": 1.4159, "step": 8503 }, { "epoch": 21.26, "grad_norm": 8.9375, "learning_rate": 1.8301627881907494e-05, "loss": 1.3889, "step": 8504 }, { "epoch": 21.2625, "grad_norm": 6.625, "learning_rate": 1.8301176436112356e-05, "loss": 1.2422, "step": 8505 }, { "epoch": 21.265, "grad_norm": 7.5625, "learning_rate": 1.830072493589504e-05, "loss": 1.4141, "step": 8506 }, { "epoch": 21.2675, "grad_norm": 7.28125, "learning_rate": 1.8300273381258506e-05, "loss": 1.3699, "step": 8507 }, { "epoch": 21.27, "grad_norm": 6.5625, "learning_rate": 1.8299821772205706e-05, "loss": 1.3501, "step": 8508 }, { "epoch": 21.2725, "grad_norm": 6.28125, "learning_rate": 1.829937010873961e-05, "loss": 1.5573, "step": 8509 }, { "epoch": 21.275, "grad_norm": 12.3125, "learning_rate": 1.8298918390863175e-05, "loss": 1.5645, "step": 8510 }, { "epoch": 21.2775, "grad_norm": 6.96875, "learning_rate": 1.8298466618579364e-05, "loss": 1.4301, "step": 8511 }, { "epoch": 21.28, "grad_norm": 8.4375, "learning_rate": 1.8298014791891138e-05, "loss": 1.4631, "step": 8512 }, { "epoch": 21.2825, "grad_norm": 8.8125, "learning_rate": 1.829756291080146e-05, "loss": 1.4834, "step": 8513 }, { "epoch": 21.285, "grad_norm": 7.375, "learning_rate": 1.829711097531329e-05, "loss": 1.4718, "step": 8514 }, { "epoch": 21.2875, "grad_norm": 6.90625, "learning_rate": 1.829665898542959e-05, "loss": 1.5089, "step": 8515 }, { "epoch": 21.29, "grad_norm": 5.625, "learning_rate": 1.8296206941153333e-05, "loss": 1.4593, "step": 8516 }, { "epoch": 21.2925, "grad_norm": 5.40625, "learning_rate": 1.8295754842487466e-05, "loss": 1.4448, "step": 8517 }, { "epoch": 21.295, "grad_norm": 6.875, "learning_rate": 1.829530268943497e-05, "loss": 1.373, "step": 8518 }, { "epoch": 21.2975, "grad_norm": 8.0625, "learning_rate": 1.82948504819988e-05, "loss": 1.3728, "step": 8519 }, { "epoch": 21.3, "grad_norm": 6.1875, "learning_rate": 1.829439822018192e-05, "loss": 1.1748, "step": 8520 }, { "epoch": 21.3025, "grad_norm": 6.1875, "learning_rate": 1.8293945903987298e-05, "loss": 1.4208, "step": 8521 }, { "epoch": 21.305, "grad_norm": 6.28125, "learning_rate": 1.8293493533417895e-05, "loss": 1.4864, "step": 8522 }, { "epoch": 21.3075, "grad_norm": 5.21875, "learning_rate": 1.8293041108476685e-05, "loss": 1.3674, "step": 8523 }, { "epoch": 21.31, "grad_norm": 6.8125, "learning_rate": 1.829258862916663e-05, "loss": 1.2538, "step": 8524 }, { "epoch": 21.3125, "grad_norm": 7.625, "learning_rate": 1.8292136095490692e-05, "loss": 1.3829, "step": 8525 }, { "epoch": 21.315, "grad_norm": 7.71875, "learning_rate": 1.829168350745184e-05, "loss": 1.4672, "step": 8526 }, { "epoch": 21.3175, "grad_norm": 6.25, "learning_rate": 1.829123086505305e-05, "loss": 1.3249, "step": 8527 }, { "epoch": 21.32, "grad_norm": 6.78125, "learning_rate": 1.829077816829728e-05, "loss": 1.2608, "step": 8528 }, { "epoch": 21.3225, "grad_norm": 7.625, "learning_rate": 1.82903254171875e-05, "loss": 1.4385, "step": 8529 }, { "epoch": 21.325, "grad_norm": 8.3125, "learning_rate": 1.8289872611726672e-05, "loss": 1.5277, "step": 8530 }, { "epoch": 21.3275, "grad_norm": 6.59375, "learning_rate": 1.8289419751917778e-05, "loss": 1.3812, "step": 8531 }, { "epoch": 21.33, "grad_norm": 9.0, "learning_rate": 1.828896683776378e-05, "loss": 1.5522, "step": 8532 }, { "epoch": 21.3325, "grad_norm": 6.625, "learning_rate": 1.8288513869267642e-05, "loss": 1.4929, "step": 8533 }, { "epoch": 21.335, "grad_norm": 9.9375, "learning_rate": 1.8288060846432342e-05, "loss": 1.3279, "step": 8534 }, { "epoch": 21.3375, "grad_norm": 6.34375, "learning_rate": 1.8287607769260845e-05, "loss": 1.2643, "step": 8535 }, { "epoch": 21.34, "grad_norm": 6.28125, "learning_rate": 1.8287154637756125e-05, "loss": 1.5358, "step": 8536 }, { "epoch": 21.3425, "grad_norm": 9.0625, "learning_rate": 1.8286701451921153e-05, "loss": 1.3832, "step": 8537 }, { "epoch": 21.345, "grad_norm": 7.6875, "learning_rate": 1.8286248211758893e-05, "loss": 1.5751, "step": 8538 }, { "epoch": 21.3475, "grad_norm": 5.8125, "learning_rate": 1.8285794917272325e-05, "loss": 1.3939, "step": 8539 }, { "epoch": 21.35, "grad_norm": 5.3125, "learning_rate": 1.8285341568464416e-05, "loss": 1.431, "step": 8540 }, { "epoch": 21.3525, "grad_norm": 9.0625, "learning_rate": 1.8284888165338137e-05, "loss": 1.175, "step": 8541 }, { "epoch": 21.355, "grad_norm": 7.5, "learning_rate": 1.8284434707896466e-05, "loss": 1.3774, "step": 8542 }, { "epoch": 21.3575, "grad_norm": 7.90625, "learning_rate": 1.828398119614237e-05, "loss": 1.3872, "step": 8543 }, { "epoch": 21.36, "grad_norm": 5.375, "learning_rate": 1.8283527630078827e-05, "loss": 1.2495, "step": 8544 }, { "epoch": 21.3625, "grad_norm": 6.75, "learning_rate": 1.8283074009708807e-05, "loss": 1.4309, "step": 8545 }, { "epoch": 21.365, "grad_norm": 6.03125, "learning_rate": 1.8282620335035285e-05, "loss": 1.5748, "step": 8546 }, { "epoch": 21.3675, "grad_norm": 5.0, "learning_rate": 1.8282166606061235e-05, "loss": 1.4132, "step": 8547 }, { "epoch": 21.37, "grad_norm": 10.0625, "learning_rate": 1.8281712822789632e-05, "loss": 1.6013, "step": 8548 }, { "epoch": 21.3725, "grad_norm": 7.625, "learning_rate": 1.8281258985223454e-05, "loss": 1.358, "step": 8549 }, { "epoch": 21.375, "grad_norm": 6.03125, "learning_rate": 1.8280805093365674e-05, "loss": 1.4833, "step": 8550 }, { "epoch": 21.3775, "grad_norm": 9.5625, "learning_rate": 1.828035114721926e-05, "loss": 1.2795, "step": 8551 }, { "epoch": 21.38, "grad_norm": 7.125, "learning_rate": 1.8279897146787204e-05, "loss": 1.3953, "step": 8552 }, { "epoch": 21.3825, "grad_norm": 6.5625, "learning_rate": 1.827944309207247e-05, "loss": 1.3649, "step": 8553 }, { "epoch": 21.385, "grad_norm": 6.0625, "learning_rate": 1.8278988983078036e-05, "loss": 1.4845, "step": 8554 }, { "epoch": 21.3875, "grad_norm": 6.625, "learning_rate": 1.8278534819806887e-05, "loss": 1.5175, "step": 8555 }, { "epoch": 21.39, "grad_norm": 5.84375, "learning_rate": 1.827808060226199e-05, "loss": 1.4735, "step": 8556 }, { "epoch": 21.3925, "grad_norm": 7.90625, "learning_rate": 1.8277626330446327e-05, "loss": 1.5123, "step": 8557 }, { "epoch": 21.395, "grad_norm": 6.96875, "learning_rate": 1.8277172004362882e-05, "loss": 1.4283, "step": 8558 }, { "epoch": 21.3975, "grad_norm": 7.21875, "learning_rate": 1.8276717624014628e-05, "loss": 1.439, "step": 8559 }, { "epoch": 21.4, "grad_norm": 6.625, "learning_rate": 1.827626318940454e-05, "loss": 1.3891, "step": 8560 }, { "epoch": 21.4025, "grad_norm": 7.9375, "learning_rate": 1.8275808700535608e-05, "loss": 1.5891, "step": 8561 }, { "epoch": 21.405, "grad_norm": 5.65625, "learning_rate": 1.82753541574108e-05, "loss": 1.3406, "step": 8562 }, { "epoch": 21.4075, "grad_norm": 7.9375, "learning_rate": 1.8274899560033103e-05, "loss": 1.4454, "step": 8563 }, { "epoch": 21.41, "grad_norm": 6.28125, "learning_rate": 1.8274444908405495e-05, "loss": 1.3635, "step": 8564 }, { "epoch": 21.4125, "grad_norm": 6.90625, "learning_rate": 1.827399020253096e-05, "loss": 1.3574, "step": 8565 }, { "epoch": 21.415, "grad_norm": 10.8125, "learning_rate": 1.8273535442412474e-05, "loss": 1.4881, "step": 8566 }, { "epoch": 21.4175, "grad_norm": 8.5, "learning_rate": 1.827308062805302e-05, "loss": 1.3689, "step": 8567 }, { "epoch": 21.42, "grad_norm": 8.8125, "learning_rate": 1.8272625759455582e-05, "loss": 1.2904, "step": 8568 }, { "epoch": 21.4225, "grad_norm": 15.4375, "learning_rate": 1.8272170836623138e-05, "loss": 1.4401, "step": 8569 }, { "epoch": 21.425, "grad_norm": 5.59375, "learning_rate": 1.827171585955868e-05, "loss": 1.3765, "step": 8570 }, { "epoch": 21.4275, "grad_norm": 6.53125, "learning_rate": 1.8271260828265177e-05, "loss": 1.5397, "step": 8571 }, { "epoch": 21.43, "grad_norm": 5.9375, "learning_rate": 1.827080574274562e-05, "loss": 1.2696, "step": 8572 }, { "epoch": 21.4325, "grad_norm": 6.96875, "learning_rate": 1.8270350603002993e-05, "loss": 1.399, "step": 8573 }, { "epoch": 21.435, "grad_norm": 5.4375, "learning_rate": 1.8269895409040276e-05, "loss": 1.4548, "step": 8574 }, { "epoch": 21.4375, "grad_norm": 9.0625, "learning_rate": 1.826944016086046e-05, "loss": 1.5154, "step": 8575 }, { "epoch": 21.44, "grad_norm": 6.71875, "learning_rate": 1.8268984858466524e-05, "loss": 1.335, "step": 8576 }, { "epoch": 21.4425, "grad_norm": 6.71875, "learning_rate": 1.826852950186145e-05, "loss": 1.4201, "step": 8577 }, { "epoch": 21.445, "grad_norm": 7.6875, "learning_rate": 1.826807409104823e-05, "loss": 1.4636, "step": 8578 }, { "epoch": 21.4475, "grad_norm": 7.6875, "learning_rate": 1.8267618626029848e-05, "loss": 1.3527, "step": 8579 }, { "epoch": 21.45, "grad_norm": 6.125, "learning_rate": 1.8267163106809288e-05, "loss": 1.2383, "step": 8580 }, { "epoch": 21.4525, "grad_norm": 7.5625, "learning_rate": 1.826670753338954e-05, "loss": 1.4208, "step": 8581 }, { "epoch": 21.455, "grad_norm": 7.71875, "learning_rate": 1.8266251905773585e-05, "loss": 1.3631, "step": 8582 }, { "epoch": 21.4575, "grad_norm": 5.46875, "learning_rate": 1.8265796223964415e-05, "loss": 1.2653, "step": 8583 }, { "epoch": 21.46, "grad_norm": 4.96875, "learning_rate": 1.8265340487965017e-05, "loss": 1.5056, "step": 8584 }, { "epoch": 21.4625, "grad_norm": 8.875, "learning_rate": 1.8264884697778374e-05, "loss": 1.3454, "step": 8585 }, { "epoch": 21.465, "grad_norm": 6.1875, "learning_rate": 1.826442885340748e-05, "loss": 1.5098, "step": 8586 }, { "epoch": 21.4675, "grad_norm": 7.90625, "learning_rate": 1.8263972954855322e-05, "loss": 1.3493, "step": 8587 }, { "epoch": 21.47, "grad_norm": 15.25, "learning_rate": 1.826351700212489e-05, "loss": 1.5161, "step": 8588 }, { "epoch": 21.4725, "grad_norm": 7.53125, "learning_rate": 1.8263060995219168e-05, "loss": 1.5118, "step": 8589 }, { "epoch": 21.475, "grad_norm": 7.09375, "learning_rate": 1.8262604934141146e-05, "loss": 1.4376, "step": 8590 }, { "epoch": 21.4775, "grad_norm": 6.59375, "learning_rate": 1.826214881889382e-05, "loss": 1.4856, "step": 8591 }, { "epoch": 21.48, "grad_norm": 5.25, "learning_rate": 1.8261692649480174e-05, "loss": 1.413, "step": 8592 }, { "epoch": 21.4825, "grad_norm": 7.0, "learning_rate": 1.8261236425903205e-05, "loss": 1.265, "step": 8593 }, { "epoch": 21.485, "grad_norm": 6.21875, "learning_rate": 1.8260780148165902e-05, "loss": 1.5819, "step": 8594 }, { "epoch": 21.4875, "grad_norm": 5.28125, "learning_rate": 1.8260323816271253e-05, "loss": 1.4612, "step": 8595 }, { "epoch": 21.49, "grad_norm": 7.5, "learning_rate": 1.825986743022225e-05, "loss": 1.3659, "step": 8596 }, { "epoch": 21.4925, "grad_norm": 9.3125, "learning_rate": 1.8259410990021888e-05, "loss": 1.4045, "step": 8597 }, { "epoch": 21.495, "grad_norm": 6.0, "learning_rate": 1.825895449567316e-05, "loss": 1.4718, "step": 8598 }, { "epoch": 21.4975, "grad_norm": 9.1875, "learning_rate": 1.8258497947179052e-05, "loss": 1.3966, "step": 8599 }, { "epoch": 21.5, "grad_norm": 8.4375, "learning_rate": 1.8258041344542567e-05, "loss": 1.3657, "step": 8600 }, { "epoch": 21.5025, "grad_norm": 5.84375, "learning_rate": 1.8257584687766693e-05, "loss": 1.4182, "step": 8601 }, { "epoch": 21.505, "grad_norm": 7.75, "learning_rate": 1.825712797685442e-05, "loss": 1.4207, "step": 8602 }, { "epoch": 21.5075, "grad_norm": 6.46875, "learning_rate": 1.825667121180875e-05, "loss": 1.39, "step": 8603 }, { "epoch": 21.51, "grad_norm": 4.28125, "learning_rate": 1.825621439263267e-05, "loss": 1.3347, "step": 8604 }, { "epoch": 21.5125, "grad_norm": 5.8125, "learning_rate": 1.8255757519329186e-05, "loss": 1.2101, "step": 8605 }, { "epoch": 21.515, "grad_norm": 5.125, "learning_rate": 1.825530059190128e-05, "loss": 1.1688, "step": 8606 }, { "epoch": 21.5175, "grad_norm": 6.125, "learning_rate": 1.8254843610351956e-05, "loss": 1.3393, "step": 8607 }, { "epoch": 21.52, "grad_norm": 18.25, "learning_rate": 1.8254386574684205e-05, "loss": 1.4417, "step": 8608 }, { "epoch": 21.5225, "grad_norm": 7.875, "learning_rate": 1.825392948490103e-05, "loss": 1.4908, "step": 8609 }, { "epoch": 21.525, "grad_norm": 7.40625, "learning_rate": 1.825347234100542e-05, "loss": 1.3879, "step": 8610 }, { "epoch": 21.5275, "grad_norm": 6.0625, "learning_rate": 1.825301514300038e-05, "loss": 1.4233, "step": 8611 }, { "epoch": 21.53, "grad_norm": 6.875, "learning_rate": 1.82525578908889e-05, "loss": 1.5744, "step": 8612 }, { "epoch": 21.5325, "grad_norm": 6.5, "learning_rate": 1.825210058467398e-05, "loss": 1.4648, "step": 8613 }, { "epoch": 21.535, "grad_norm": 7.375, "learning_rate": 1.825164322435862e-05, "loss": 1.377, "step": 8614 }, { "epoch": 21.5375, "grad_norm": 7.09375, "learning_rate": 1.8251185809945818e-05, "loss": 1.4631, "step": 8615 }, { "epoch": 21.54, "grad_norm": 7.15625, "learning_rate": 1.825072834143857e-05, "loss": 1.2872, "step": 8616 }, { "epoch": 21.5425, "grad_norm": 4.8125, "learning_rate": 1.8250270818839877e-05, "loss": 1.4708, "step": 8617 }, { "epoch": 21.545, "grad_norm": 9.0625, "learning_rate": 1.8249813242152742e-05, "loss": 1.5206, "step": 8618 }, { "epoch": 21.5475, "grad_norm": 4.84375, "learning_rate": 1.8249355611380157e-05, "loss": 1.4126, "step": 8619 }, { "epoch": 21.55, "grad_norm": 13.0, "learning_rate": 1.824889792652513e-05, "loss": 1.4322, "step": 8620 }, { "epoch": 21.5525, "grad_norm": 7.09375, "learning_rate": 1.8248440187590658e-05, "loss": 1.3857, "step": 8621 }, { "epoch": 21.555, "grad_norm": 12.1875, "learning_rate": 1.824798239457974e-05, "loss": 1.492, "step": 8622 }, { "epoch": 21.5575, "grad_norm": 6.6875, "learning_rate": 1.8247524547495382e-05, "loss": 1.3884, "step": 8623 }, { "epoch": 21.56, "grad_norm": 5.25, "learning_rate": 1.824706664634058e-05, "loss": 1.2859, "step": 8624 }, { "epoch": 21.5625, "grad_norm": 7.5625, "learning_rate": 1.8246608691118343e-05, "loss": 1.405, "step": 8625 }, { "epoch": 21.565, "grad_norm": 9.5625, "learning_rate": 1.8246150681831668e-05, "loss": 1.5718, "step": 8626 }, { "epoch": 21.5675, "grad_norm": 6.375, "learning_rate": 1.8245692618483558e-05, "loss": 1.3443, "step": 8627 }, { "epoch": 21.57, "grad_norm": 7.375, "learning_rate": 1.824523450107702e-05, "loss": 1.4677, "step": 8628 }, { "epoch": 21.5725, "grad_norm": 6.875, "learning_rate": 1.824477632961505e-05, "loss": 1.3287, "step": 8629 }, { "epoch": 21.575, "grad_norm": 7.625, "learning_rate": 1.824431810410066e-05, "loss": 1.3289, "step": 8630 }, { "epoch": 21.5775, "grad_norm": 8.5, "learning_rate": 1.8243859824536848e-05, "loss": 1.3794, "step": 8631 }, { "epoch": 21.58, "grad_norm": 8.3125, "learning_rate": 1.8243401490926623e-05, "loss": 1.309, "step": 8632 }, { "epoch": 21.5825, "grad_norm": 7.0, "learning_rate": 1.8242943103272986e-05, "loss": 1.2807, "step": 8633 }, { "epoch": 21.585, "grad_norm": 6.65625, "learning_rate": 1.8242484661578946e-05, "loss": 1.3781, "step": 8634 }, { "epoch": 21.5875, "grad_norm": 5.4375, "learning_rate": 1.8242026165847506e-05, "loss": 1.4583, "step": 8635 }, { "epoch": 21.59, "grad_norm": 6.96875, "learning_rate": 1.8241567616081674e-05, "loss": 1.3487, "step": 8636 }, { "epoch": 21.5925, "grad_norm": 7.09375, "learning_rate": 1.824110901228445e-05, "loss": 1.4858, "step": 8637 }, { "epoch": 21.595, "grad_norm": 5.46875, "learning_rate": 1.8240650354458846e-05, "loss": 1.254, "step": 8638 }, { "epoch": 21.5975, "grad_norm": 6.34375, "learning_rate": 1.824019164260787e-05, "loss": 1.4394, "step": 8639 }, { "epoch": 21.6, "grad_norm": 5.3125, "learning_rate": 1.8239732876734525e-05, "loss": 1.4511, "step": 8640 }, { "epoch": 21.6025, "grad_norm": 7.75, "learning_rate": 1.8239274056841822e-05, "loss": 1.3736, "step": 8641 }, { "epoch": 21.605, "grad_norm": 7.5625, "learning_rate": 1.823881518293277e-05, "loss": 1.5846, "step": 8642 }, { "epoch": 21.6075, "grad_norm": 5.59375, "learning_rate": 1.8238356255010373e-05, "loss": 1.287, "step": 8643 }, { "epoch": 21.61, "grad_norm": 5.4375, "learning_rate": 1.8237897273077644e-05, "loss": 1.447, "step": 8644 }, { "epoch": 21.6125, "grad_norm": 8.8125, "learning_rate": 1.8237438237137587e-05, "loss": 1.3488, "step": 8645 }, { "epoch": 21.615, "grad_norm": 10.5, "learning_rate": 1.823697914719322e-05, "loss": 1.3223, "step": 8646 }, { "epoch": 21.6175, "grad_norm": 8.75, "learning_rate": 1.8236520003247543e-05, "loss": 1.2833, "step": 8647 }, { "epoch": 21.62, "grad_norm": 8.3125, "learning_rate": 1.823606080530357e-05, "loss": 1.3964, "step": 8648 }, { "epoch": 21.6225, "grad_norm": 5.90625, "learning_rate": 1.8235601553364315e-05, "loss": 1.2747, "step": 8649 }, { "epoch": 21.625, "grad_norm": 8.125, "learning_rate": 1.8235142247432784e-05, "loss": 1.3371, "step": 8650 }, { "epoch": 21.6275, "grad_norm": 6.34375, "learning_rate": 1.823468288751199e-05, "loss": 1.4009, "step": 8651 }, { "epoch": 21.63, "grad_norm": 4.75, "learning_rate": 1.8234223473604946e-05, "loss": 1.2476, "step": 8652 }, { "epoch": 21.6325, "grad_norm": 7.3125, "learning_rate": 1.823376400571466e-05, "loss": 1.4904, "step": 8653 }, { "epoch": 21.635, "grad_norm": 6.96875, "learning_rate": 1.8233304483844147e-05, "loss": 1.4746, "step": 8654 }, { "epoch": 21.6375, "grad_norm": 5.96875, "learning_rate": 1.8232844907996422e-05, "loss": 1.4131, "step": 8655 }, { "epoch": 21.64, "grad_norm": 5.0, "learning_rate": 1.823238527817449e-05, "loss": 1.2587, "step": 8656 }, { "epoch": 21.6425, "grad_norm": 8.3125, "learning_rate": 1.8231925594381375e-05, "loss": 1.5985, "step": 8657 }, { "epoch": 21.645, "grad_norm": 7.4375, "learning_rate": 1.8231465856620084e-05, "loss": 1.5534, "step": 8658 }, { "epoch": 21.6475, "grad_norm": 7.5625, "learning_rate": 1.8231006064893632e-05, "loss": 1.2371, "step": 8659 }, { "epoch": 21.65, "grad_norm": 8.9375, "learning_rate": 1.8230546219205032e-05, "loss": 1.4077, "step": 8660 }, { "epoch": 21.6525, "grad_norm": 8.1875, "learning_rate": 1.82300863195573e-05, "loss": 1.4773, "step": 8661 }, { "epoch": 21.655, "grad_norm": 5.15625, "learning_rate": 1.8229626365953453e-05, "loss": 1.2762, "step": 8662 }, { "epoch": 21.6575, "grad_norm": 6.15625, "learning_rate": 1.8229166358396505e-05, "loss": 1.3723, "step": 8663 }, { "epoch": 21.66, "grad_norm": 7.15625, "learning_rate": 1.822870629688947e-05, "loss": 1.5061, "step": 8664 }, { "epoch": 21.6625, "grad_norm": 7.90625, "learning_rate": 1.8228246181435367e-05, "loss": 1.5598, "step": 8665 }, { "epoch": 21.665, "grad_norm": 6.53125, "learning_rate": 1.8227786012037213e-05, "loss": 1.4728, "step": 8666 }, { "epoch": 21.6675, "grad_norm": 7.59375, "learning_rate": 1.822732578869802e-05, "loss": 1.2981, "step": 8667 }, { "epoch": 21.67, "grad_norm": 6.65625, "learning_rate": 1.8226865511420812e-05, "loss": 1.538, "step": 8668 }, { "epoch": 21.6725, "grad_norm": 8.9375, "learning_rate": 1.82264051802086e-05, "loss": 1.5067, "step": 8669 }, { "epoch": 21.675, "grad_norm": 7.28125, "learning_rate": 1.8225944795064404e-05, "loss": 1.4439, "step": 8670 }, { "epoch": 21.6775, "grad_norm": 8.0, "learning_rate": 1.8225484355991247e-05, "loss": 1.5959, "step": 8671 }, { "epoch": 21.68, "grad_norm": 8.875, "learning_rate": 1.822502386299214e-05, "loss": 1.4265, "step": 8672 }, { "epoch": 21.6825, "grad_norm": 4.25, "learning_rate": 1.8224563316070108e-05, "loss": 1.2379, "step": 8673 }, { "epoch": 21.685, "grad_norm": 8.25, "learning_rate": 1.8224102715228166e-05, "loss": 1.4607, "step": 8674 }, { "epoch": 21.6875, "grad_norm": 6.40625, "learning_rate": 1.8223642060469338e-05, "loss": 1.3345, "step": 8675 }, { "epoch": 21.69, "grad_norm": 6.09375, "learning_rate": 1.8223181351796642e-05, "loss": 1.2571, "step": 8676 }, { "epoch": 21.6925, "grad_norm": 5.375, "learning_rate": 1.8222720589213095e-05, "loss": 1.3683, "step": 8677 }, { "epoch": 21.695, "grad_norm": 8.0625, "learning_rate": 1.8222259772721725e-05, "loss": 1.4536, "step": 8678 }, { "epoch": 21.6975, "grad_norm": 7.78125, "learning_rate": 1.8221798902325547e-05, "loss": 1.4752, "step": 8679 }, { "epoch": 21.7, "grad_norm": 6.65625, "learning_rate": 1.822133797802758e-05, "loss": 1.4509, "step": 8680 }, { "epoch": 21.7025, "grad_norm": 9.125, "learning_rate": 1.8220876999830855e-05, "loss": 1.3559, "step": 8681 }, { "epoch": 21.705, "grad_norm": 5.25, "learning_rate": 1.822041596773839e-05, "loss": 1.4662, "step": 8682 }, { "epoch": 21.7075, "grad_norm": 7.4375, "learning_rate": 1.8219954881753206e-05, "loss": 1.3303, "step": 8683 }, { "epoch": 21.71, "grad_norm": 4.9375, "learning_rate": 1.8219493741878325e-05, "loss": 1.355, "step": 8684 }, { "epoch": 21.7125, "grad_norm": 8.25, "learning_rate": 1.8219032548116776e-05, "loss": 1.5304, "step": 8685 }, { "epoch": 21.715, "grad_norm": 6.5625, "learning_rate": 1.8218571300471575e-05, "loss": 1.2438, "step": 8686 }, { "epoch": 21.7175, "grad_norm": 7.0, "learning_rate": 1.821810999894575e-05, "loss": 1.3273, "step": 8687 }, { "epoch": 21.72, "grad_norm": 6.375, "learning_rate": 1.8217648643542326e-05, "loss": 1.4554, "step": 8688 }, { "epoch": 21.7225, "grad_norm": 6.09375, "learning_rate": 1.8217187234264322e-05, "loss": 1.3504, "step": 8689 }, { "epoch": 21.725, "grad_norm": 8.1875, "learning_rate": 1.8216725771114773e-05, "loss": 1.6098, "step": 8690 }, { "epoch": 21.7275, "grad_norm": 6.625, "learning_rate": 1.8216264254096696e-05, "loss": 1.38, "step": 8691 }, { "epoch": 21.73, "grad_norm": 5.0625, "learning_rate": 1.821580268321312e-05, "loss": 1.2596, "step": 8692 }, { "epoch": 21.7325, "grad_norm": 9.625, "learning_rate": 1.821534105846707e-05, "loss": 1.4689, "step": 8693 }, { "epoch": 21.735, "grad_norm": 9.4375, "learning_rate": 1.821487937986157e-05, "loss": 1.3138, "step": 8694 }, { "epoch": 21.7375, "grad_norm": 8.6875, "learning_rate": 1.8214417647399655e-05, "loss": 1.3558, "step": 8695 }, { "epoch": 21.74, "grad_norm": 7.40625, "learning_rate": 1.8213955861084342e-05, "loss": 1.434, "step": 8696 }, { "epoch": 21.7425, "grad_norm": 8.375, "learning_rate": 1.8213494020918666e-05, "loss": 1.519, "step": 8697 }, { "epoch": 21.745, "grad_norm": 7.15625, "learning_rate": 1.821303212690565e-05, "loss": 1.4356, "step": 8698 }, { "epoch": 21.7475, "grad_norm": 8.8125, "learning_rate": 1.8212570179048325e-05, "loss": 1.3764, "step": 8699 }, { "epoch": 21.75, "grad_norm": 6.59375, "learning_rate": 1.8212108177349722e-05, "loss": 1.5914, "step": 8700 }, { "epoch": 21.7525, "grad_norm": 6.5, "learning_rate": 1.821164612181286e-05, "loss": 1.5264, "step": 8701 }, { "epoch": 21.755, "grad_norm": 7.9375, "learning_rate": 1.821118401244078e-05, "loss": 1.3653, "step": 8702 }, { "epoch": 21.7575, "grad_norm": 7.96875, "learning_rate": 1.8210721849236503e-05, "loss": 1.3946, "step": 8703 }, { "epoch": 21.76, "grad_norm": 10.0625, "learning_rate": 1.8210259632203063e-05, "loss": 1.4, "step": 8704 }, { "epoch": 21.7625, "grad_norm": 5.1875, "learning_rate": 1.8209797361343488e-05, "loss": 1.3838, "step": 8705 }, { "epoch": 21.765, "grad_norm": 9.4375, "learning_rate": 1.820933503666081e-05, "loss": 1.5457, "step": 8706 }, { "epoch": 21.7675, "grad_norm": 6.46875, "learning_rate": 1.8208872658158062e-05, "loss": 1.3768, "step": 8707 }, { "epoch": 21.77, "grad_norm": 6.3125, "learning_rate": 1.8208410225838276e-05, "loss": 1.5291, "step": 8708 }, { "epoch": 21.7725, "grad_norm": 7.96875, "learning_rate": 1.820794773970448e-05, "loss": 1.4534, "step": 8709 }, { "epoch": 21.775, "grad_norm": 7.15625, "learning_rate": 1.82074851997597e-05, "loss": 1.3509, "step": 8710 }, { "epoch": 21.7775, "grad_norm": 11.0, "learning_rate": 1.820702260600698e-05, "loss": 1.3785, "step": 8711 }, { "epoch": 21.78, "grad_norm": 9.0625, "learning_rate": 1.820655995844935e-05, "loss": 1.3208, "step": 8712 }, { "epoch": 21.7825, "grad_norm": 9.5625, "learning_rate": 1.8206097257089838e-05, "loss": 1.3616, "step": 8713 }, { "epoch": 21.785, "grad_norm": 8.5625, "learning_rate": 1.8205634501931484e-05, "loss": 1.527, "step": 8714 }, { "epoch": 21.7875, "grad_norm": 6.90625, "learning_rate": 1.8205171692977316e-05, "loss": 1.5325, "step": 8715 }, { "epoch": 21.79, "grad_norm": 6.375, "learning_rate": 1.8204708830230372e-05, "loss": 1.4044, "step": 8716 }, { "epoch": 21.7925, "grad_norm": 6.0, "learning_rate": 1.820424591369369e-05, "loss": 1.4534, "step": 8717 }, { "epoch": 21.795, "grad_norm": 7.03125, "learning_rate": 1.820378294337029e-05, "loss": 1.4189, "step": 8718 }, { "epoch": 21.7975, "grad_norm": 6.5, "learning_rate": 1.8203319919263225e-05, "loss": 1.4902, "step": 8719 }, { "epoch": 21.8, "grad_norm": 5.59375, "learning_rate": 1.8202856841375517e-05, "loss": 1.3627, "step": 8720 }, { "epoch": 21.8025, "grad_norm": 5.625, "learning_rate": 1.8202393709710212e-05, "loss": 1.3284, "step": 8721 }, { "epoch": 21.805, "grad_norm": 5.75, "learning_rate": 1.8201930524270343e-05, "loss": 1.4205, "step": 8722 }, { "epoch": 21.8075, "grad_norm": 6.09375, "learning_rate": 1.8201467285058945e-05, "loss": 1.4396, "step": 8723 }, { "epoch": 21.81, "grad_norm": 6.65625, "learning_rate": 1.8201003992079054e-05, "loss": 1.4543, "step": 8724 }, { "epoch": 21.8125, "grad_norm": 6.375, "learning_rate": 1.820054064533371e-05, "loss": 1.4712, "step": 8725 }, { "epoch": 21.815, "grad_norm": 5.9375, "learning_rate": 1.8200077244825948e-05, "loss": 1.2631, "step": 8726 }, { "epoch": 21.8175, "grad_norm": 5.6875, "learning_rate": 1.819961379055881e-05, "loss": 1.3938, "step": 8727 }, { "epoch": 21.82, "grad_norm": 6.0, "learning_rate": 1.8199150282535334e-05, "loss": 1.2205, "step": 8728 }, { "epoch": 21.8225, "grad_norm": 7.125, "learning_rate": 1.819868672075855e-05, "loss": 1.3781, "step": 8729 }, { "epoch": 21.825, "grad_norm": 8.1875, "learning_rate": 1.8198223105231512e-05, "loss": 1.4386, "step": 8730 }, { "epoch": 21.8275, "grad_norm": 6.71875, "learning_rate": 1.8197759435957248e-05, "loss": 1.6819, "step": 8731 }, { "epoch": 21.83, "grad_norm": 5.65625, "learning_rate": 1.8197295712938798e-05, "loss": 1.5938, "step": 8732 }, { "epoch": 21.8325, "grad_norm": 7.78125, "learning_rate": 1.8196831936179207e-05, "loss": 1.3381, "step": 8733 }, { "epoch": 21.835, "grad_norm": 9.75, "learning_rate": 1.8196368105681516e-05, "loss": 1.3318, "step": 8734 }, { "epoch": 21.8375, "grad_norm": 6.5, "learning_rate": 1.8195904221448762e-05, "loss": 1.5678, "step": 8735 }, { "epoch": 21.84, "grad_norm": 6.375, "learning_rate": 1.819544028348399e-05, "loss": 1.412, "step": 8736 }, { "epoch": 21.8425, "grad_norm": 6.8125, "learning_rate": 1.819497629179024e-05, "loss": 1.5295, "step": 8737 }, { "epoch": 21.845, "grad_norm": 7.5625, "learning_rate": 1.8194512246370547e-05, "loss": 1.31, "step": 8738 }, { "epoch": 21.8475, "grad_norm": 6.4375, "learning_rate": 1.8194048147227964e-05, "loss": 1.4428, "step": 8739 }, { "epoch": 21.85, "grad_norm": 7.53125, "learning_rate": 1.819358399436553e-05, "loss": 1.3942, "step": 8740 }, { "epoch": 21.8525, "grad_norm": 6.8125, "learning_rate": 1.8193119787786288e-05, "loss": 1.4969, "step": 8741 }, { "epoch": 21.855, "grad_norm": 6.46875, "learning_rate": 1.8192655527493278e-05, "loss": 1.4793, "step": 8742 }, { "epoch": 21.8575, "grad_norm": 6.96875, "learning_rate": 1.819219121348955e-05, "loss": 1.3819, "step": 8743 }, { "epoch": 21.86, "grad_norm": 7.25, "learning_rate": 1.819172684577814e-05, "loss": 1.4611, "step": 8744 }, { "epoch": 21.8625, "grad_norm": 4.6875, "learning_rate": 1.8191262424362096e-05, "loss": 1.5236, "step": 8745 }, { "epoch": 21.865, "grad_norm": 10.25, "learning_rate": 1.8190797949244468e-05, "loss": 1.3344, "step": 8746 }, { "epoch": 21.8675, "grad_norm": 5.46875, "learning_rate": 1.8190333420428294e-05, "loss": 1.386, "step": 8747 }, { "epoch": 21.87, "grad_norm": 6.8125, "learning_rate": 1.818986883791662e-05, "loss": 1.3349, "step": 8748 }, { "epoch": 21.8725, "grad_norm": 7.75, "learning_rate": 1.8189404201712494e-05, "loss": 1.4574, "step": 8749 }, { "epoch": 21.875, "grad_norm": 5.59375, "learning_rate": 1.8188939511818965e-05, "loss": 1.5883, "step": 8750 }, { "epoch": 21.8775, "grad_norm": 7.65625, "learning_rate": 1.8188474768239075e-05, "loss": 1.4287, "step": 8751 }, { "epoch": 21.88, "grad_norm": 5.4375, "learning_rate": 1.818800997097587e-05, "loss": 1.33, "step": 8752 }, { "epoch": 21.8825, "grad_norm": 6.21875, "learning_rate": 1.81875451200324e-05, "loss": 1.4244, "step": 8753 }, { "epoch": 21.885, "grad_norm": 8.4375, "learning_rate": 1.8187080215411713e-05, "loss": 1.2855, "step": 8754 }, { "epoch": 21.8875, "grad_norm": 8.5, "learning_rate": 1.8186615257116855e-05, "loss": 1.3817, "step": 8755 }, { "epoch": 21.89, "grad_norm": 6.28125, "learning_rate": 1.8186150245150876e-05, "loss": 1.5216, "step": 8756 }, { "epoch": 21.8925, "grad_norm": 7.25, "learning_rate": 1.8185685179516826e-05, "loss": 1.286, "step": 8757 }, { "epoch": 21.895, "grad_norm": 6.71875, "learning_rate": 1.8185220060217745e-05, "loss": 1.3475, "step": 8758 }, { "epoch": 21.8975, "grad_norm": 6.71875, "learning_rate": 1.8184754887256692e-05, "loss": 1.3837, "step": 8759 }, { "epoch": 21.9, "grad_norm": 7.90625, "learning_rate": 1.8184289660636715e-05, "loss": 1.3828, "step": 8760 }, { "epoch": 21.9025, "grad_norm": 7.0625, "learning_rate": 1.8183824380360863e-05, "loss": 1.2654, "step": 8761 }, { "epoch": 21.905, "grad_norm": 7.625, "learning_rate": 1.8183359046432182e-05, "loss": 1.4994, "step": 8762 }, { "epoch": 21.9075, "grad_norm": 6.59375, "learning_rate": 1.8182893658853732e-05, "loss": 1.3374, "step": 8763 }, { "epoch": 21.91, "grad_norm": 4.46875, "learning_rate": 1.8182428217628555e-05, "loss": 1.3603, "step": 8764 }, { "epoch": 21.9125, "grad_norm": 7.75, "learning_rate": 1.8181962722759705e-05, "loss": 1.4115, "step": 8765 }, { "epoch": 21.915, "grad_norm": 5.6875, "learning_rate": 1.8181497174250236e-05, "loss": 1.2816, "step": 8766 }, { "epoch": 21.9175, "grad_norm": 7.78125, "learning_rate": 1.81810315721032e-05, "loss": 1.3792, "step": 8767 }, { "epoch": 21.92, "grad_norm": 6.125, "learning_rate": 1.8180565916321646e-05, "loss": 1.3955, "step": 8768 }, { "epoch": 21.9225, "grad_norm": 6.53125, "learning_rate": 1.818010020690863e-05, "loss": 1.3952, "step": 8769 }, { "epoch": 21.925, "grad_norm": 5.46875, "learning_rate": 1.8179634443867207e-05, "loss": 1.3885, "step": 8770 }, { "epoch": 21.9275, "grad_norm": 10.0625, "learning_rate": 1.8179168627200426e-05, "loss": 1.4919, "step": 8771 }, { "epoch": 21.93, "grad_norm": 7.21875, "learning_rate": 1.817870275691134e-05, "loss": 1.4613, "step": 8772 }, { "epoch": 21.9325, "grad_norm": 7.71875, "learning_rate": 1.8178236833003012e-05, "loss": 1.4674, "step": 8773 }, { "epoch": 21.935, "grad_norm": 6.53125, "learning_rate": 1.8177770855478488e-05, "loss": 1.3341, "step": 8774 }, { "epoch": 21.9375, "grad_norm": 5.4375, "learning_rate": 1.8177304824340823e-05, "loss": 1.4146, "step": 8775 }, { "epoch": 21.94, "grad_norm": 8.0625, "learning_rate": 1.8176838739593078e-05, "loss": 1.2629, "step": 8776 }, { "epoch": 21.9425, "grad_norm": 7.09375, "learning_rate": 1.8176372601238302e-05, "loss": 1.5335, "step": 8777 }, { "epoch": 21.945, "grad_norm": 6.96875, "learning_rate": 1.8175906409279557e-05, "loss": 1.5027, "step": 8778 }, { "epoch": 21.9475, "grad_norm": 4.75, "learning_rate": 1.81754401637199e-05, "loss": 1.3705, "step": 8779 }, { "epoch": 21.95, "grad_norm": 6.875, "learning_rate": 1.817497386456238e-05, "loss": 1.5392, "step": 8780 }, { "epoch": 21.9525, "grad_norm": 4.59375, "learning_rate": 1.817450751181006e-05, "loss": 1.4846, "step": 8781 }, { "epoch": 21.955, "grad_norm": 7.625, "learning_rate": 1.8174041105465995e-05, "loss": 1.5408, "step": 8782 }, { "epoch": 21.9575, "grad_norm": 10.25, "learning_rate": 1.8173574645533242e-05, "loss": 1.5029, "step": 8783 }, { "epoch": 21.96, "grad_norm": 6.03125, "learning_rate": 1.817310813201486e-05, "loss": 1.3957, "step": 8784 }, { "epoch": 21.9625, "grad_norm": 8.0, "learning_rate": 1.8172641564913915e-05, "loss": 1.2865, "step": 8785 }, { "epoch": 21.965, "grad_norm": 8.5625, "learning_rate": 1.8172174944233455e-05, "loss": 1.372, "step": 8786 }, { "epoch": 21.9675, "grad_norm": 5.53125, "learning_rate": 1.817170826997654e-05, "loss": 1.393, "step": 8787 }, { "epoch": 21.97, "grad_norm": 8.625, "learning_rate": 1.8171241542146235e-05, "loss": 1.4285, "step": 8788 }, { "epoch": 21.9725, "grad_norm": 8.0, "learning_rate": 1.81707747607456e-05, "loss": 1.46, "step": 8789 }, { "epoch": 21.975, "grad_norm": 5.0625, "learning_rate": 1.817030792577769e-05, "loss": 1.2813, "step": 8790 }, { "epoch": 21.9775, "grad_norm": 9.125, "learning_rate": 1.816984103724557e-05, "loss": 1.4736, "step": 8791 }, { "epoch": 21.98, "grad_norm": 6.75, "learning_rate": 1.8169374095152298e-05, "loss": 1.2994, "step": 8792 }, { "epoch": 21.9825, "grad_norm": 7.15625, "learning_rate": 1.8168907099500934e-05, "loss": 1.3665, "step": 8793 }, { "epoch": 21.985, "grad_norm": 6.84375, "learning_rate": 1.8168440050294546e-05, "loss": 1.3999, "step": 8794 }, { "epoch": 21.9875, "grad_norm": 6.5, "learning_rate": 1.8167972947536188e-05, "loss": 1.3733, "step": 8795 }, { "epoch": 21.99, "grad_norm": 8.875, "learning_rate": 1.816750579122893e-05, "loss": 1.4468, "step": 8796 }, { "epoch": 21.9925, "grad_norm": 8.4375, "learning_rate": 1.816703858137583e-05, "loss": 1.3405, "step": 8797 }, { "epoch": 21.995, "grad_norm": 10.3125, "learning_rate": 1.816657131797995e-05, "loss": 1.5514, "step": 8798 }, { "epoch": 21.9975, "grad_norm": 7.0, "learning_rate": 1.8166104001044354e-05, "loss": 1.4281, "step": 8799 }, { "epoch": 22.0, "grad_norm": 6.84375, "learning_rate": 1.816563663057211e-05, "loss": 1.4346, "step": 8800 }, { "epoch": 22.0025, "grad_norm": 4.625, "learning_rate": 1.816516920656628e-05, "loss": 1.3802, "step": 8801 }, { "epoch": 22.005, "grad_norm": 6.21875, "learning_rate": 1.8164701729029923e-05, "loss": 1.309, "step": 8802 }, { "epoch": 22.0075, "grad_norm": 9.5, "learning_rate": 1.8164234197966112e-05, "loss": 1.3109, "step": 8803 }, { "epoch": 22.01, "grad_norm": 7.125, "learning_rate": 1.816376661337791e-05, "loss": 1.5308, "step": 8804 }, { "epoch": 22.0125, "grad_norm": 8.3125, "learning_rate": 1.8163298975268376e-05, "loss": 1.3156, "step": 8805 }, { "epoch": 22.015, "grad_norm": 6.9375, "learning_rate": 1.816283128364058e-05, "loss": 1.3776, "step": 8806 }, { "epoch": 22.0175, "grad_norm": 8.3125, "learning_rate": 1.8162363538497594e-05, "loss": 1.3807, "step": 8807 }, { "epoch": 22.02, "grad_norm": 6.65625, "learning_rate": 1.8161895739842476e-05, "loss": 1.5553, "step": 8808 }, { "epoch": 22.0225, "grad_norm": 6.5625, "learning_rate": 1.8161427887678297e-05, "loss": 1.3216, "step": 8809 }, { "epoch": 22.025, "grad_norm": 6.59375, "learning_rate": 1.816095998200812e-05, "loss": 1.4836, "step": 8810 }, { "epoch": 22.0275, "grad_norm": 7.09375, "learning_rate": 1.816049202283502e-05, "loss": 1.4616, "step": 8811 }, { "epoch": 22.03, "grad_norm": 7.125, "learning_rate": 1.8160024010162062e-05, "loss": 1.5018, "step": 8812 }, { "epoch": 22.0325, "grad_norm": 6.21875, "learning_rate": 1.8159555943992312e-05, "loss": 1.4922, "step": 8813 }, { "epoch": 22.035, "grad_norm": 7.78125, "learning_rate": 1.8159087824328837e-05, "loss": 1.4325, "step": 8814 }, { "epoch": 22.0375, "grad_norm": 8.8125, "learning_rate": 1.8158619651174713e-05, "loss": 1.5741, "step": 8815 }, { "epoch": 22.04, "grad_norm": 5.6875, "learning_rate": 1.8158151424533002e-05, "loss": 1.3945, "step": 8816 }, { "epoch": 22.0425, "grad_norm": 11.125, "learning_rate": 1.8157683144406775e-05, "loss": 1.4687, "step": 8817 }, { "epoch": 22.045, "grad_norm": 6.5625, "learning_rate": 1.8157214810799105e-05, "loss": 1.3468, "step": 8818 }, { "epoch": 22.0475, "grad_norm": 12.125, "learning_rate": 1.8156746423713064e-05, "loss": 1.185, "step": 8819 }, { "epoch": 22.05, "grad_norm": 7.21875, "learning_rate": 1.815627798315172e-05, "loss": 1.2419, "step": 8820 }, { "epoch": 22.0525, "grad_norm": 7.1875, "learning_rate": 1.815580948911814e-05, "loss": 1.3399, "step": 8821 }, { "epoch": 22.055, "grad_norm": 6.9375, "learning_rate": 1.81553409416154e-05, "loss": 1.4811, "step": 8822 }, { "epoch": 22.0575, "grad_norm": 6.4375, "learning_rate": 1.8154872340646573e-05, "loss": 1.406, "step": 8823 }, { "epoch": 22.06, "grad_norm": 7.1875, "learning_rate": 1.815440368621473e-05, "loss": 1.4894, "step": 8824 }, { "epoch": 22.0625, "grad_norm": 9.5625, "learning_rate": 1.815393497832294e-05, "loss": 1.4391, "step": 8825 }, { "epoch": 22.065, "grad_norm": 6.78125, "learning_rate": 1.815346621697428e-05, "loss": 1.477, "step": 8826 }, { "epoch": 22.0675, "grad_norm": 7.21875, "learning_rate": 1.8152997402171825e-05, "loss": 1.5765, "step": 8827 }, { "epoch": 22.07, "grad_norm": 7.03125, "learning_rate": 1.8152528533918642e-05, "loss": 1.3273, "step": 8828 }, { "epoch": 22.0725, "grad_norm": 5.625, "learning_rate": 1.8152059612217806e-05, "loss": 1.3308, "step": 8829 }, { "epoch": 22.075, "grad_norm": 6.40625, "learning_rate": 1.8151590637072396e-05, "loss": 1.457, "step": 8830 }, { "epoch": 22.0775, "grad_norm": 8.875, "learning_rate": 1.8151121608485485e-05, "loss": 1.5068, "step": 8831 }, { "epoch": 22.08, "grad_norm": 6.65625, "learning_rate": 1.8150652526460146e-05, "loss": 1.3037, "step": 8832 }, { "epoch": 22.0825, "grad_norm": 8.375, "learning_rate": 1.8150183390999453e-05, "loss": 1.2276, "step": 8833 }, { "epoch": 22.085, "grad_norm": 7.75, "learning_rate": 1.8149714202106488e-05, "loss": 1.3761, "step": 8834 }, { "epoch": 22.0875, "grad_norm": 6.0, "learning_rate": 1.814924495978432e-05, "loss": 1.4544, "step": 8835 }, { "epoch": 22.09, "grad_norm": 5.03125, "learning_rate": 1.814877566403603e-05, "loss": 1.2886, "step": 8836 }, { "epoch": 22.0925, "grad_norm": 9.375, "learning_rate": 1.814830631486469e-05, "loss": 1.5222, "step": 8837 }, { "epoch": 22.095, "grad_norm": 5.5625, "learning_rate": 1.814783691227338e-05, "loss": 1.4929, "step": 8838 }, { "epoch": 22.0975, "grad_norm": 6.71875, "learning_rate": 1.8147367456265178e-05, "loss": 1.5259, "step": 8839 }, { "epoch": 22.1, "grad_norm": 11.125, "learning_rate": 1.8146897946843162e-05, "loss": 1.3946, "step": 8840 }, { "epoch": 22.1025, "grad_norm": 6.59375, "learning_rate": 1.814642838401041e-05, "loss": 1.2983, "step": 8841 }, { "epoch": 22.105, "grad_norm": 8.125, "learning_rate": 1.8145958767769997e-05, "loss": 1.4276, "step": 8842 }, { "epoch": 22.1075, "grad_norm": 5.90625, "learning_rate": 1.8145489098125007e-05, "loss": 1.4303, "step": 8843 }, { "epoch": 22.11, "grad_norm": 6.59375, "learning_rate": 1.8145019375078517e-05, "loss": 1.3789, "step": 8844 }, { "epoch": 22.1125, "grad_norm": 7.96875, "learning_rate": 1.8144549598633603e-05, "loss": 1.5522, "step": 8845 }, { "epoch": 22.115, "grad_norm": 10.1875, "learning_rate": 1.8144079768793348e-05, "loss": 1.4574, "step": 8846 }, { "epoch": 22.1175, "grad_norm": 7.4375, "learning_rate": 1.8143609885560832e-05, "loss": 1.3628, "step": 8847 }, { "epoch": 22.12, "grad_norm": 6.96875, "learning_rate": 1.8143139948939138e-05, "loss": 1.3507, "step": 8848 }, { "epoch": 22.1225, "grad_norm": 8.375, "learning_rate": 1.8142669958931343e-05, "loss": 1.4153, "step": 8849 }, { "epoch": 22.125, "grad_norm": 6.0, "learning_rate": 1.814219991554053e-05, "loss": 1.365, "step": 8850 }, { "epoch": 22.1275, "grad_norm": 7.53125, "learning_rate": 1.8141729818769777e-05, "loss": 1.4718, "step": 8851 }, { "epoch": 22.13, "grad_norm": 12.75, "learning_rate": 1.8141259668622174e-05, "loss": 1.2847, "step": 8852 }, { "epoch": 22.1325, "grad_norm": 4.8125, "learning_rate": 1.814078946510079e-05, "loss": 1.5587, "step": 8853 }, { "epoch": 22.135, "grad_norm": 10.8125, "learning_rate": 1.8140319208208727e-05, "loss": 1.3919, "step": 8854 }, { "epoch": 22.1375, "grad_norm": 9.0, "learning_rate": 1.813984889794905e-05, "loss": 1.4806, "step": 8855 }, { "epoch": 22.14, "grad_norm": 6.78125, "learning_rate": 1.813937853432485e-05, "loss": 1.5352, "step": 8856 }, { "epoch": 22.1425, "grad_norm": 7.21875, "learning_rate": 1.8138908117339212e-05, "loss": 1.6046, "step": 8857 }, { "epoch": 22.145, "grad_norm": 4.65625, "learning_rate": 1.8138437646995217e-05, "loss": 1.4109, "step": 8858 }, { "epoch": 22.1475, "grad_norm": 6.90625, "learning_rate": 1.8137967123295953e-05, "loss": 1.543, "step": 8859 }, { "epoch": 22.15, "grad_norm": 6.5, "learning_rate": 1.81374965462445e-05, "loss": 1.3345, "step": 8860 }, { "epoch": 22.1525, "grad_norm": 5.5625, "learning_rate": 1.8137025915843943e-05, "loss": 1.4577, "step": 8861 }, { "epoch": 22.155, "grad_norm": 7.1875, "learning_rate": 1.8136555232097374e-05, "loss": 1.4273, "step": 8862 }, { "epoch": 22.1575, "grad_norm": 7.90625, "learning_rate": 1.8136084495007874e-05, "loss": 1.3617, "step": 8863 }, { "epoch": 22.16, "grad_norm": 8.75, "learning_rate": 1.8135613704578525e-05, "loss": 1.4612, "step": 8864 }, { "epoch": 22.1625, "grad_norm": 7.15625, "learning_rate": 1.8135142860812422e-05, "loss": 1.3806, "step": 8865 }, { "epoch": 22.165, "grad_norm": 7.5, "learning_rate": 1.813467196371265e-05, "loss": 1.2258, "step": 8866 }, { "epoch": 22.1675, "grad_norm": 7.0625, "learning_rate": 1.813420101328229e-05, "loss": 1.32, "step": 8867 }, { "epoch": 22.17, "grad_norm": 6.6875, "learning_rate": 1.8133730009524436e-05, "loss": 1.4041, "step": 8868 }, { "epoch": 22.1725, "grad_norm": 7.15625, "learning_rate": 1.813325895244217e-05, "loss": 1.4509, "step": 8869 }, { "epoch": 22.175, "grad_norm": 5.84375, "learning_rate": 1.8132787842038588e-05, "loss": 1.3778, "step": 8870 }, { "epoch": 22.1775, "grad_norm": 9.75, "learning_rate": 1.8132316678316774e-05, "loss": 1.478, "step": 8871 }, { "epoch": 22.18, "grad_norm": 6.34375, "learning_rate": 1.8131845461279813e-05, "loss": 1.3623, "step": 8872 }, { "epoch": 22.1825, "grad_norm": 8.0, "learning_rate": 1.81313741909308e-05, "loss": 1.4108, "step": 8873 }, { "epoch": 22.185, "grad_norm": 8.1875, "learning_rate": 1.8130902867272828e-05, "loss": 1.268, "step": 8874 }, { "epoch": 22.1875, "grad_norm": 8.375, "learning_rate": 1.813043149030898e-05, "loss": 1.5642, "step": 8875 }, { "epoch": 22.19, "grad_norm": 7.03125, "learning_rate": 1.8129960060042345e-05, "loss": 1.2983, "step": 8876 }, { "epoch": 22.1925, "grad_norm": 5.53125, "learning_rate": 1.812948857647602e-05, "loss": 1.4978, "step": 8877 }, { "epoch": 22.195, "grad_norm": 7.75, "learning_rate": 1.8129017039613094e-05, "loss": 1.2621, "step": 8878 }, { "epoch": 22.1975, "grad_norm": 7.15625, "learning_rate": 1.8128545449456655e-05, "loss": 1.421, "step": 8879 }, { "epoch": 22.2, "grad_norm": 6.6875, "learning_rate": 1.81280738060098e-05, "loss": 1.3063, "step": 8880 }, { "epoch": 22.2025, "grad_norm": 7.625, "learning_rate": 1.8127602109275617e-05, "loss": 1.4646, "step": 8881 }, { "epoch": 22.205, "grad_norm": 7.09375, "learning_rate": 1.81271303592572e-05, "loss": 1.4103, "step": 8882 }, { "epoch": 22.2075, "grad_norm": 6.59375, "learning_rate": 1.8126658555957643e-05, "loss": 1.496, "step": 8883 }, { "epoch": 22.21, "grad_norm": 6.0625, "learning_rate": 1.8126186699380035e-05, "loss": 1.3847, "step": 8884 }, { "epoch": 22.2125, "grad_norm": 7.1875, "learning_rate": 1.8125714789527474e-05, "loss": 1.3859, "step": 8885 }, { "epoch": 22.215, "grad_norm": 5.25, "learning_rate": 1.8125242826403052e-05, "loss": 1.327, "step": 8886 }, { "epoch": 22.2175, "grad_norm": 7.0, "learning_rate": 1.8124770810009863e-05, "loss": 1.3529, "step": 8887 }, { "epoch": 22.22, "grad_norm": 6.21875, "learning_rate": 1.8124298740351004e-05, "loss": 1.3287, "step": 8888 }, { "epoch": 22.2225, "grad_norm": 6.09375, "learning_rate": 1.812382661742956e-05, "loss": 1.5072, "step": 8889 }, { "epoch": 22.225, "grad_norm": 6.78125, "learning_rate": 1.812335444124864e-05, "loss": 1.2469, "step": 8890 }, { "epoch": 22.2275, "grad_norm": 4.75, "learning_rate": 1.8122882211811332e-05, "loss": 1.2638, "step": 8891 }, { "epoch": 22.23, "grad_norm": 9.5, "learning_rate": 1.8122409929120735e-05, "loss": 1.4488, "step": 8892 }, { "epoch": 22.2325, "grad_norm": 9.3125, "learning_rate": 1.8121937593179942e-05, "loss": 1.4249, "step": 8893 }, { "epoch": 22.235, "grad_norm": 6.71875, "learning_rate": 1.812146520399205e-05, "loss": 1.4181, "step": 8894 }, { "epoch": 22.2375, "grad_norm": 8.5, "learning_rate": 1.812099276156016e-05, "loss": 1.5628, "step": 8895 }, { "epoch": 22.24, "grad_norm": 9.0625, "learning_rate": 1.8120520265887364e-05, "loss": 1.3114, "step": 8896 }, { "epoch": 22.2425, "grad_norm": 5.21875, "learning_rate": 1.8120047716976762e-05, "loss": 1.2944, "step": 8897 }, { "epoch": 22.245, "grad_norm": 5.40625, "learning_rate": 1.8119575114831454e-05, "loss": 1.2827, "step": 8898 }, { "epoch": 22.2475, "grad_norm": 5.09375, "learning_rate": 1.8119102459454536e-05, "loss": 1.4422, "step": 8899 }, { "epoch": 22.25, "grad_norm": 8.3125, "learning_rate": 1.8118629750849106e-05, "loss": 1.3959, "step": 8900 }, { "epoch": 22.2525, "grad_norm": 7.53125, "learning_rate": 1.8118156989018266e-05, "loss": 1.4348, "step": 8901 }, { "epoch": 22.255, "grad_norm": 8.9375, "learning_rate": 1.811768417396511e-05, "loss": 1.6523, "step": 8902 }, { "epoch": 22.2575, "grad_norm": 5.75, "learning_rate": 1.8117211305692747e-05, "loss": 1.4004, "step": 8903 }, { "epoch": 22.26, "grad_norm": 9.6875, "learning_rate": 1.8116738384204268e-05, "loss": 1.4554, "step": 8904 }, { "epoch": 22.2625, "grad_norm": 8.375, "learning_rate": 1.8116265409502777e-05, "loss": 1.4653, "step": 8905 }, { "epoch": 22.265, "grad_norm": 7.28125, "learning_rate": 1.8115792381591373e-05, "loss": 1.456, "step": 8906 }, { "epoch": 22.2675, "grad_norm": 9.25, "learning_rate": 1.8115319300473162e-05, "loss": 1.4269, "step": 8907 }, { "epoch": 22.27, "grad_norm": 8.5, "learning_rate": 1.8114846166151243e-05, "loss": 1.4417, "step": 8908 }, { "epoch": 22.2725, "grad_norm": 8.8125, "learning_rate": 1.8114372978628715e-05, "loss": 1.2683, "step": 8909 }, { "epoch": 22.275, "grad_norm": 6.78125, "learning_rate": 1.8113899737908683e-05, "loss": 1.3735, "step": 8910 }, { "epoch": 22.2775, "grad_norm": 7.21875, "learning_rate": 1.811342644399425e-05, "loss": 1.3079, "step": 8911 }, { "epoch": 22.28, "grad_norm": 8.25, "learning_rate": 1.8112953096888517e-05, "loss": 1.2586, "step": 8912 }, { "epoch": 22.2825, "grad_norm": 5.21875, "learning_rate": 1.811247969659459e-05, "loss": 1.8165, "step": 8913 }, { "epoch": 22.285, "grad_norm": 7.125, "learning_rate": 1.8112006243115568e-05, "loss": 1.4155, "step": 8914 }, { "epoch": 22.2875, "grad_norm": 5.875, "learning_rate": 1.8111532736454556e-05, "loss": 1.4595, "step": 8915 }, { "epoch": 22.29, "grad_norm": 5.59375, "learning_rate": 1.8111059176614665e-05, "loss": 1.4549, "step": 8916 }, { "epoch": 22.2925, "grad_norm": 7.90625, "learning_rate": 1.8110585563598994e-05, "loss": 1.3381, "step": 8917 }, { "epoch": 22.295, "grad_norm": 7.78125, "learning_rate": 1.8110111897410647e-05, "loss": 1.4638, "step": 8918 }, { "epoch": 22.2975, "grad_norm": 7.96875, "learning_rate": 1.810963817805273e-05, "loss": 1.5027, "step": 8919 }, { "epoch": 22.3, "grad_norm": 9.75, "learning_rate": 1.810916440552835e-05, "loss": 1.2623, "step": 8920 }, { "epoch": 22.3025, "grad_norm": 6.40625, "learning_rate": 1.8108690579840613e-05, "loss": 1.3878, "step": 8921 }, { "epoch": 22.305, "grad_norm": 5.6875, "learning_rate": 1.8108216700992627e-05, "loss": 1.3238, "step": 8922 }, { "epoch": 22.3075, "grad_norm": 7.0, "learning_rate": 1.8107742768987492e-05, "loss": 1.519, "step": 8923 }, { "epoch": 22.31, "grad_norm": 7.125, "learning_rate": 1.8107268783828326e-05, "loss": 1.419, "step": 8924 }, { "epoch": 22.3125, "grad_norm": 5.6875, "learning_rate": 1.8106794745518225e-05, "loss": 1.4024, "step": 8925 }, { "epoch": 22.315, "grad_norm": 6.125, "learning_rate": 1.8106320654060307e-05, "loss": 1.4253, "step": 8926 }, { "epoch": 22.3175, "grad_norm": 5.1875, "learning_rate": 1.810584650945767e-05, "loss": 1.4703, "step": 8927 }, { "epoch": 22.32, "grad_norm": 11.375, "learning_rate": 1.810537231171343e-05, "loss": 1.419, "step": 8928 }, { "epoch": 22.3225, "grad_norm": 12.1875, "learning_rate": 1.8104898060830692e-05, "loss": 1.52, "step": 8929 }, { "epoch": 22.325, "grad_norm": 13.75, "learning_rate": 1.810442375681257e-05, "loss": 1.4004, "step": 8930 }, { "epoch": 22.3275, "grad_norm": 8.375, "learning_rate": 1.810394939966217e-05, "loss": 1.4173, "step": 8931 }, { "epoch": 22.33, "grad_norm": 7.65625, "learning_rate": 1.81034749893826e-05, "loss": 1.5902, "step": 8932 }, { "epoch": 22.3325, "grad_norm": 6.8125, "learning_rate": 1.8103000525976973e-05, "loss": 1.5098, "step": 8933 }, { "epoch": 22.335, "grad_norm": 6.21875, "learning_rate": 1.8102526009448398e-05, "loss": 1.5137, "step": 8934 }, { "epoch": 22.3375, "grad_norm": 12.5625, "learning_rate": 1.8102051439799987e-05, "loss": 1.5876, "step": 8935 }, { "epoch": 22.34, "grad_norm": 5.3125, "learning_rate": 1.8101576817034853e-05, "loss": 1.2474, "step": 8936 }, { "epoch": 22.3425, "grad_norm": 6.28125, "learning_rate": 1.81011021411561e-05, "loss": 1.5403, "step": 8937 }, { "epoch": 22.345, "grad_norm": 6.84375, "learning_rate": 1.8100627412166854e-05, "loss": 1.4165, "step": 8938 }, { "epoch": 22.3475, "grad_norm": 6.46875, "learning_rate": 1.810015263007021e-05, "loss": 1.2176, "step": 8939 }, { "epoch": 22.35, "grad_norm": 6.90625, "learning_rate": 1.8099677794869297e-05, "loss": 1.4218, "step": 8940 }, { "epoch": 22.3525, "grad_norm": 9.125, "learning_rate": 1.8099202906567217e-05, "loss": 1.5699, "step": 8941 }, { "epoch": 22.355, "grad_norm": 8.125, "learning_rate": 1.8098727965167088e-05, "loss": 1.4333, "step": 8942 }, { "epoch": 22.3575, "grad_norm": 5.6875, "learning_rate": 1.809825297067202e-05, "loss": 1.3316, "step": 8943 }, { "epoch": 22.36, "grad_norm": 6.59375, "learning_rate": 1.809777792308513e-05, "loss": 1.5719, "step": 8944 }, { "epoch": 22.3625, "grad_norm": 7.59375, "learning_rate": 1.8097302822409536e-05, "loss": 1.3855, "step": 8945 }, { "epoch": 22.365, "grad_norm": 5.5625, "learning_rate": 1.8096827668648344e-05, "loss": 1.353, "step": 8946 }, { "epoch": 22.3675, "grad_norm": 6.40625, "learning_rate": 1.8096352461804677e-05, "loss": 1.4596, "step": 8947 }, { "epoch": 22.37, "grad_norm": 8.5, "learning_rate": 1.8095877201881647e-05, "loss": 1.4651, "step": 8948 }, { "epoch": 22.3725, "grad_norm": 7.46875, "learning_rate": 1.8095401888882367e-05, "loss": 1.4154, "step": 8949 }, { "epoch": 22.375, "grad_norm": 5.59375, "learning_rate": 1.8094926522809958e-05, "loss": 1.3965, "step": 8950 }, { "epoch": 22.3775, "grad_norm": 9.5, "learning_rate": 1.8094451103667536e-05, "loss": 1.4343, "step": 8951 }, { "epoch": 22.38, "grad_norm": 6.375, "learning_rate": 1.8093975631458215e-05, "loss": 1.4719, "step": 8952 }, { "epoch": 22.3825, "grad_norm": 6.75, "learning_rate": 1.8093500106185116e-05, "loss": 1.2938, "step": 8953 }, { "epoch": 22.385, "grad_norm": 8.5625, "learning_rate": 1.809302452785135e-05, "loss": 1.3982, "step": 8954 }, { "epoch": 22.3875, "grad_norm": 6.65625, "learning_rate": 1.8092548896460042e-05, "loss": 1.3232, "step": 8955 }, { "epoch": 22.39, "grad_norm": 6.6875, "learning_rate": 1.8092073212014307e-05, "loss": 1.3193, "step": 8956 }, { "epoch": 22.3925, "grad_norm": 8.0625, "learning_rate": 1.8091597474517266e-05, "loss": 1.5037, "step": 8957 }, { "epoch": 22.395, "grad_norm": 6.3125, "learning_rate": 1.809112168397203e-05, "loss": 1.2433, "step": 8958 }, { "epoch": 22.3975, "grad_norm": 5.8125, "learning_rate": 1.809064584038173e-05, "loss": 1.1355, "step": 8959 }, { "epoch": 22.4, "grad_norm": 11.1875, "learning_rate": 1.8090169943749477e-05, "loss": 1.3785, "step": 8960 }, { "epoch": 22.4025, "grad_norm": 7.25, "learning_rate": 1.8089693994078394e-05, "loss": 1.3997, "step": 8961 }, { "epoch": 22.405, "grad_norm": 7.09375, "learning_rate": 1.80892179913716e-05, "loss": 1.4644, "step": 8962 }, { "epoch": 22.4075, "grad_norm": 7.65625, "learning_rate": 1.8088741935632215e-05, "loss": 1.5044, "step": 8963 }, { "epoch": 22.41, "grad_norm": 7.90625, "learning_rate": 1.8088265826863363e-05, "loss": 1.3732, "step": 8964 }, { "epoch": 22.4125, "grad_norm": 5.9375, "learning_rate": 1.8087789665068166e-05, "loss": 1.2654, "step": 8965 }, { "epoch": 22.415, "grad_norm": 5.09375, "learning_rate": 1.8087313450249742e-05, "loss": 1.2127, "step": 8966 }, { "epoch": 22.4175, "grad_norm": 6.1875, "learning_rate": 1.8086837182411217e-05, "loss": 1.387, "step": 8967 }, { "epoch": 22.42, "grad_norm": 6.125, "learning_rate": 1.8086360861555707e-05, "loss": 1.3935, "step": 8968 }, { "epoch": 22.4225, "grad_norm": 6.0, "learning_rate": 1.808588448768634e-05, "loss": 1.3275, "step": 8969 }, { "epoch": 22.425, "grad_norm": 76.5, "learning_rate": 1.808540806080624e-05, "loss": 1.3325, "step": 8970 }, { "epoch": 22.4275, "grad_norm": 7.53125, "learning_rate": 1.8084931580918524e-05, "loss": 1.2981, "step": 8971 }, { "epoch": 22.43, "grad_norm": 6.15625, "learning_rate": 1.8084455048026327e-05, "loss": 1.3892, "step": 8972 }, { "epoch": 22.4325, "grad_norm": 7.84375, "learning_rate": 1.8083978462132758e-05, "loss": 1.5012, "step": 8973 }, { "epoch": 22.435, "grad_norm": 9.875, "learning_rate": 1.8083501823240954e-05, "loss": 1.3767, "step": 8974 }, { "epoch": 22.4375, "grad_norm": 5.375, "learning_rate": 1.808302513135404e-05, "loss": 1.34, "step": 8975 }, { "epoch": 22.44, "grad_norm": 7.59375, "learning_rate": 1.808254838647513e-05, "loss": 1.4348, "step": 8976 }, { "epoch": 22.4425, "grad_norm": 7.75, "learning_rate": 1.8082071588607357e-05, "loss": 1.5602, "step": 8977 }, { "epoch": 22.445, "grad_norm": 6.5625, "learning_rate": 1.8081594737753845e-05, "loss": 1.4142, "step": 8978 }, { "epoch": 22.4475, "grad_norm": 5.28125, "learning_rate": 1.8081117833917728e-05, "loss": 1.408, "step": 8979 }, { "epoch": 22.45, "grad_norm": 4.40625, "learning_rate": 1.808064087710212e-05, "loss": 1.2446, "step": 8980 }, { "epoch": 22.4525, "grad_norm": 7.0625, "learning_rate": 1.8080163867310156e-05, "loss": 1.6205, "step": 8981 }, { "epoch": 22.455, "grad_norm": 9.3125, "learning_rate": 1.807968680454496e-05, "loss": 1.3092, "step": 8982 }, { "epoch": 22.4575, "grad_norm": 5.875, "learning_rate": 1.8079209688809662e-05, "loss": 1.4478, "step": 8983 }, { "epoch": 22.46, "grad_norm": 7.9375, "learning_rate": 1.8078732520107386e-05, "loss": 1.3263, "step": 8984 }, { "epoch": 22.4625, "grad_norm": 7.875, "learning_rate": 1.8078255298441264e-05, "loss": 1.3952, "step": 8985 }, { "epoch": 22.465, "grad_norm": 8.3125, "learning_rate": 1.8077778023814424e-05, "loss": 1.265, "step": 8986 }, { "epoch": 22.4675, "grad_norm": 8.6875, "learning_rate": 1.8077300696229997e-05, "loss": 1.3286, "step": 8987 }, { "epoch": 22.47, "grad_norm": 7.96875, "learning_rate": 1.8076823315691106e-05, "loss": 1.3606, "step": 8988 }, { "epoch": 22.4725, "grad_norm": 8.125, "learning_rate": 1.8076345882200886e-05, "loss": 1.4778, "step": 8989 }, { "epoch": 22.475, "grad_norm": 7.375, "learning_rate": 1.8075868395762467e-05, "loss": 1.4108, "step": 8990 }, { "epoch": 22.4775, "grad_norm": 8.0, "learning_rate": 1.8075390856378976e-05, "loss": 1.4555, "step": 8991 }, { "epoch": 22.48, "grad_norm": 5.28125, "learning_rate": 1.8074913264053547e-05, "loss": 1.4368, "step": 8992 }, { "epoch": 22.4825, "grad_norm": 7.0625, "learning_rate": 1.8074435618789308e-05, "loss": 1.3655, "step": 8993 }, { "epoch": 22.485, "grad_norm": 5.90625, "learning_rate": 1.8073957920589393e-05, "loss": 1.4124, "step": 8994 }, { "epoch": 22.4875, "grad_norm": 7.84375, "learning_rate": 1.807348016945693e-05, "loss": 1.386, "step": 8995 }, { "epoch": 22.49, "grad_norm": 6.1875, "learning_rate": 1.807300236539506e-05, "loss": 1.3588, "step": 8996 }, { "epoch": 22.4925, "grad_norm": 5.15625, "learning_rate": 1.8072524508406908e-05, "loss": 1.3464, "step": 8997 }, { "epoch": 22.495, "grad_norm": 7.28125, "learning_rate": 1.8072046598495606e-05, "loss": 1.4208, "step": 8998 }, { "epoch": 22.4975, "grad_norm": 4.5625, "learning_rate": 1.8071568635664292e-05, "loss": 1.3421, "step": 8999 }, { "epoch": 22.5, "grad_norm": 5.28125, "learning_rate": 1.8071090619916095e-05, "loss": 1.4227, "step": 9000 }, { "epoch": 22.5025, "grad_norm": 6.0625, "learning_rate": 1.807061255125415e-05, "loss": 1.5121, "step": 9001 }, { "epoch": 22.505, "grad_norm": 7.0625, "learning_rate": 1.8070134429681598e-05, "loss": 1.4366, "step": 9002 }, { "epoch": 22.5075, "grad_norm": 7.71875, "learning_rate": 1.806965625520156e-05, "loss": 1.3841, "step": 9003 }, { "epoch": 22.51, "grad_norm": 6.5, "learning_rate": 1.806917802781718e-05, "loss": 1.4308, "step": 9004 }, { "epoch": 22.5125, "grad_norm": 7.65625, "learning_rate": 1.8068699747531597e-05, "loss": 1.5433, "step": 9005 }, { "epoch": 22.515, "grad_norm": 6.59375, "learning_rate": 1.8068221414347937e-05, "loss": 1.4063, "step": 9006 }, { "epoch": 22.5175, "grad_norm": 7.625, "learning_rate": 1.8067743028269343e-05, "loss": 1.3519, "step": 9007 }, { "epoch": 22.52, "grad_norm": 6.125, "learning_rate": 1.8067264589298945e-05, "loss": 1.4619, "step": 9008 }, { "epoch": 22.5225, "grad_norm": 6.21875, "learning_rate": 1.8066786097439885e-05, "loss": 1.3196, "step": 9009 }, { "epoch": 22.525, "grad_norm": 7.09375, "learning_rate": 1.80663075526953e-05, "loss": 1.5033, "step": 9010 }, { "epoch": 22.5275, "grad_norm": 7.21875, "learning_rate": 1.806582895506832e-05, "loss": 1.3666, "step": 9011 }, { "epoch": 22.53, "grad_norm": 6.75, "learning_rate": 1.8065350304562095e-05, "loss": 1.4916, "step": 9012 }, { "epoch": 22.5325, "grad_norm": 7.90625, "learning_rate": 1.806487160117975e-05, "loss": 1.5378, "step": 9013 }, { "epoch": 22.535, "grad_norm": 9.0625, "learning_rate": 1.8064392844924436e-05, "loss": 1.321, "step": 9014 }, { "epoch": 22.5375, "grad_norm": 7.0, "learning_rate": 1.8063914035799278e-05, "loss": 1.3684, "step": 9015 }, { "epoch": 22.54, "grad_norm": 7.125, "learning_rate": 1.806343517380743e-05, "loss": 1.4357, "step": 9016 }, { "epoch": 22.5425, "grad_norm": 8.1875, "learning_rate": 1.806295625895202e-05, "loss": 1.346, "step": 9017 }, { "epoch": 22.545, "grad_norm": 9.0, "learning_rate": 1.8062477291236188e-05, "loss": 1.2827, "step": 9018 }, { "epoch": 22.5475, "grad_norm": 8.125, "learning_rate": 1.8061998270663082e-05, "loss": 1.511, "step": 9019 }, { "epoch": 22.55, "grad_norm": 6.34375, "learning_rate": 1.8061519197235835e-05, "loss": 1.4909, "step": 9020 }, { "epoch": 22.5525, "grad_norm": 5.1875, "learning_rate": 1.806104007095759e-05, "loss": 1.4067, "step": 9021 }, { "epoch": 22.555, "grad_norm": 6.34375, "learning_rate": 1.8060560891831495e-05, "loss": 1.3274, "step": 9022 }, { "epoch": 22.5575, "grad_norm": 5.21875, "learning_rate": 1.806008165986068e-05, "loss": 1.3266, "step": 9023 }, { "epoch": 22.56, "grad_norm": 6.0, "learning_rate": 1.8059602375048294e-05, "loss": 1.3338, "step": 9024 }, { "epoch": 22.5625, "grad_norm": 7.90625, "learning_rate": 1.8059123037397478e-05, "loss": 1.3798, "step": 9025 }, { "epoch": 22.565, "grad_norm": 6.15625, "learning_rate": 1.8058643646911372e-05, "loss": 1.4709, "step": 9026 }, { "epoch": 22.5675, "grad_norm": 8.4375, "learning_rate": 1.805816420359312e-05, "loss": 1.4511, "step": 9027 }, { "epoch": 22.57, "grad_norm": 7.5625, "learning_rate": 1.805768470744587e-05, "loss": 1.4358, "step": 9028 }, { "epoch": 22.5725, "grad_norm": 6.09375, "learning_rate": 1.805720515847276e-05, "loss": 1.3873, "step": 9029 }, { "epoch": 22.575, "grad_norm": 6.34375, "learning_rate": 1.8056725556676932e-05, "loss": 1.4334, "step": 9030 }, { "epoch": 22.5775, "grad_norm": 9.0625, "learning_rate": 1.805624590206154e-05, "loss": 1.3338, "step": 9031 }, { "epoch": 22.58, "grad_norm": 8.3125, "learning_rate": 1.8055766194629717e-05, "loss": 1.3694, "step": 9032 }, { "epoch": 22.5825, "grad_norm": 7.125, "learning_rate": 1.8055286434384616e-05, "loss": 1.433, "step": 9033 }, { "epoch": 22.585, "grad_norm": 7.6875, "learning_rate": 1.805480662132938e-05, "loss": 1.5662, "step": 9034 }, { "epoch": 22.5875, "grad_norm": 6.71875, "learning_rate": 1.805432675546715e-05, "loss": 1.3954, "step": 9035 }, { "epoch": 22.59, "grad_norm": 7.15625, "learning_rate": 1.8053846836801076e-05, "loss": 1.3739, "step": 9036 }, { "epoch": 22.5925, "grad_norm": 6.1875, "learning_rate": 1.8053366865334306e-05, "loss": 1.2654, "step": 9037 }, { "epoch": 22.595, "grad_norm": 5.375, "learning_rate": 1.8052886841069987e-05, "loss": 1.3353, "step": 9038 }, { "epoch": 22.5975, "grad_norm": 7.53125, "learning_rate": 1.8052406764011264e-05, "loss": 1.216, "step": 9039 }, { "epoch": 22.6, "grad_norm": 5.28125, "learning_rate": 1.8051926634161282e-05, "loss": 1.4564, "step": 9040 }, { "epoch": 22.6025, "grad_norm": 7.3125, "learning_rate": 1.8051446451523195e-05, "loss": 1.3993, "step": 9041 }, { "epoch": 22.605, "grad_norm": 5.5625, "learning_rate": 1.8050966216100145e-05, "loss": 1.3594, "step": 9042 }, { "epoch": 22.6075, "grad_norm": 9.9375, "learning_rate": 1.805048592789528e-05, "loss": 1.3945, "step": 9043 }, { "epoch": 22.61, "grad_norm": 4.4375, "learning_rate": 1.8050005586911757e-05, "loss": 1.5515, "step": 9044 }, { "epoch": 22.6125, "grad_norm": 4.875, "learning_rate": 1.8049525193152717e-05, "loss": 1.3407, "step": 9045 }, { "epoch": 22.615, "grad_norm": 9.3125, "learning_rate": 1.804904474662131e-05, "loss": 1.3036, "step": 9046 }, { "epoch": 22.6175, "grad_norm": 7.125, "learning_rate": 1.804856424732069e-05, "loss": 1.4422, "step": 9047 }, { "epoch": 22.62, "grad_norm": 5.0, "learning_rate": 1.8048083695254006e-05, "loss": 1.2607, "step": 9048 }, { "epoch": 22.6225, "grad_norm": 8.875, "learning_rate": 1.8047603090424407e-05, "loss": 1.3375, "step": 9049 }, { "epoch": 22.625, "grad_norm": 5.65625, "learning_rate": 1.804712243283504e-05, "loss": 1.4551, "step": 9050 }, { "epoch": 22.6275, "grad_norm": 5.5, "learning_rate": 1.804664172248906e-05, "loss": 1.5333, "step": 9051 }, { "epoch": 22.63, "grad_norm": 13.625, "learning_rate": 1.8046160959389625e-05, "loss": 1.437, "step": 9052 }, { "epoch": 22.6325, "grad_norm": 6.25, "learning_rate": 1.8045680143539878e-05, "loss": 1.36, "step": 9053 }, { "epoch": 22.635, "grad_norm": 6.90625, "learning_rate": 1.8045199274942977e-05, "loss": 1.4266, "step": 9054 }, { "epoch": 22.6375, "grad_norm": 7.6875, "learning_rate": 1.8044718353602065e-05, "loss": 1.5772, "step": 9055 }, { "epoch": 22.64, "grad_norm": 6.53125, "learning_rate": 1.8044237379520305e-05, "loss": 1.3687, "step": 9056 }, { "epoch": 22.6425, "grad_norm": 5.75, "learning_rate": 1.8043756352700844e-05, "loss": 1.3813, "step": 9057 }, { "epoch": 22.645, "grad_norm": 5.46875, "learning_rate": 1.8043275273146843e-05, "loss": 1.371, "step": 9058 }, { "epoch": 22.6475, "grad_norm": 5.34375, "learning_rate": 1.8042794140861447e-05, "loss": 1.2825, "step": 9059 }, { "epoch": 22.65, "grad_norm": 4.46875, "learning_rate": 1.804231295584782e-05, "loss": 1.5017, "step": 9060 }, { "epoch": 22.6525, "grad_norm": 7.125, "learning_rate": 1.8041831718109107e-05, "loss": 1.4579, "step": 9061 }, { "epoch": 22.655, "grad_norm": 8.25, "learning_rate": 1.8041350427648466e-05, "loss": 1.3539, "step": 9062 }, { "epoch": 22.6575, "grad_norm": 9.25, "learning_rate": 1.8040869084469053e-05, "loss": 1.444, "step": 9063 }, { "epoch": 22.66, "grad_norm": 6.9375, "learning_rate": 1.8040387688574027e-05, "loss": 1.374, "step": 9064 }, { "epoch": 22.6625, "grad_norm": 8.0625, "learning_rate": 1.803990623996654e-05, "loss": 1.4576, "step": 9065 }, { "epoch": 22.665, "grad_norm": 4.53125, "learning_rate": 1.803942473864975e-05, "loss": 1.5506, "step": 9066 }, { "epoch": 22.6675, "grad_norm": 9.6875, "learning_rate": 1.803894318462681e-05, "loss": 1.4362, "step": 9067 }, { "epoch": 22.67, "grad_norm": 5.96875, "learning_rate": 1.8038461577900883e-05, "loss": 1.2829, "step": 9068 }, { "epoch": 22.6725, "grad_norm": 5.375, "learning_rate": 1.8037979918475126e-05, "loss": 1.3869, "step": 9069 }, { "epoch": 22.675, "grad_norm": 7.21875, "learning_rate": 1.8037498206352692e-05, "loss": 1.5451, "step": 9070 }, { "epoch": 22.6775, "grad_norm": 10.0, "learning_rate": 1.8037016441536738e-05, "loss": 1.4803, "step": 9071 }, { "epoch": 22.68, "grad_norm": 6.21875, "learning_rate": 1.8036534624030428e-05, "loss": 1.4723, "step": 9072 }, { "epoch": 22.6825, "grad_norm": 7.0625, "learning_rate": 1.803605275383692e-05, "loss": 1.3739, "step": 9073 }, { "epoch": 22.685, "grad_norm": 7.59375, "learning_rate": 1.8035570830959374e-05, "loss": 1.4179, "step": 9074 }, { "epoch": 22.6875, "grad_norm": 4.40625, "learning_rate": 1.803508885540094e-05, "loss": 1.3269, "step": 9075 }, { "epoch": 22.69, "grad_norm": 6.6875, "learning_rate": 1.8034606827164795e-05, "loss": 1.412, "step": 9076 }, { "epoch": 22.6925, "grad_norm": 7.46875, "learning_rate": 1.803412474625408e-05, "loss": 1.4354, "step": 9077 }, { "epoch": 22.695, "grad_norm": 7.53125, "learning_rate": 1.803364261267197e-05, "loss": 1.4004, "step": 9078 }, { "epoch": 22.6975, "grad_norm": 5.71875, "learning_rate": 1.8033160426421616e-05, "loss": 1.3921, "step": 9079 }, { "epoch": 22.7, "grad_norm": 9.0, "learning_rate": 1.8032678187506187e-05, "loss": 1.1913, "step": 9080 }, { "epoch": 22.7025, "grad_norm": 6.71875, "learning_rate": 1.8032195895928842e-05, "loss": 1.2529, "step": 9081 }, { "epoch": 22.705, "grad_norm": 8.625, "learning_rate": 1.803171355169274e-05, "loss": 1.4244, "step": 9082 }, { "epoch": 22.7075, "grad_norm": 5.40625, "learning_rate": 1.8031231154801047e-05, "loss": 1.3679, "step": 9083 }, { "epoch": 22.71, "grad_norm": 7.21875, "learning_rate": 1.8030748705256923e-05, "loss": 1.3965, "step": 9084 }, { "epoch": 22.7125, "grad_norm": 5.125, "learning_rate": 1.803026620306353e-05, "loss": 1.2912, "step": 9085 }, { "epoch": 22.715, "grad_norm": 6.96875, "learning_rate": 1.8029783648224037e-05, "loss": 1.4978, "step": 9086 }, { "epoch": 22.7175, "grad_norm": 7.5, "learning_rate": 1.80293010407416e-05, "loss": 1.5036, "step": 9087 }, { "epoch": 22.72, "grad_norm": 6.375, "learning_rate": 1.802881838061939e-05, "loss": 1.3663, "step": 9088 }, { "epoch": 22.7225, "grad_norm": 6.90625, "learning_rate": 1.8028335667860567e-05, "loss": 1.372, "step": 9089 }, { "epoch": 22.725, "grad_norm": 8.625, "learning_rate": 1.8027852902468295e-05, "loss": 1.4555, "step": 9090 }, { "epoch": 22.7275, "grad_norm": 6.28125, "learning_rate": 1.802737008444574e-05, "loss": 1.4329, "step": 9091 }, { "epoch": 22.73, "grad_norm": 5.25, "learning_rate": 1.8026887213796072e-05, "loss": 1.5022, "step": 9092 }, { "epoch": 22.7325, "grad_norm": 7.375, "learning_rate": 1.802640429052245e-05, "loss": 1.3542, "step": 9093 }, { "epoch": 22.735, "grad_norm": 6.875, "learning_rate": 1.8025921314628044e-05, "loss": 1.3651, "step": 9094 }, { "epoch": 22.7375, "grad_norm": 6.96875, "learning_rate": 1.8025438286116016e-05, "loss": 1.3274, "step": 9095 }, { "epoch": 22.74, "grad_norm": 25.125, "learning_rate": 1.8024955204989537e-05, "loss": 1.2268, "step": 9096 }, { "epoch": 22.7425, "grad_norm": 5.65625, "learning_rate": 1.8024472071251775e-05, "loss": 1.3721, "step": 9097 }, { "epoch": 22.745, "grad_norm": 6.4375, "learning_rate": 1.8023988884905894e-05, "loss": 1.3617, "step": 9098 }, { "epoch": 22.7475, "grad_norm": 6.0625, "learning_rate": 1.8023505645955066e-05, "loss": 1.4444, "step": 9099 }, { "epoch": 22.75, "grad_norm": 6.5, "learning_rate": 1.802302235440245e-05, "loss": 1.4508, "step": 9100 }, { "epoch": 22.7525, "grad_norm": 7.9375, "learning_rate": 1.8022539010251222e-05, "loss": 1.3752, "step": 9101 }, { "epoch": 22.755, "grad_norm": 7.78125, "learning_rate": 1.8022055613504553e-05, "loss": 1.3502, "step": 9102 }, { "epoch": 22.7575, "grad_norm": 9.3125, "learning_rate": 1.8021572164165607e-05, "loss": 1.4297, "step": 9103 }, { "epoch": 22.76, "grad_norm": 9.625, "learning_rate": 1.802108866223755e-05, "loss": 1.6004, "step": 9104 }, { "epoch": 22.7625, "grad_norm": 6.25, "learning_rate": 1.802060510772356e-05, "loss": 1.6435, "step": 9105 }, { "epoch": 22.765, "grad_norm": 14.9375, "learning_rate": 1.8020121500626807e-05, "loss": 1.4272, "step": 9106 }, { "epoch": 22.7675, "grad_norm": 11.4375, "learning_rate": 1.801963784095046e-05, "loss": 1.4456, "step": 9107 }, { "epoch": 22.77, "grad_norm": 6.65625, "learning_rate": 1.801915412869768e-05, "loss": 1.4875, "step": 9108 }, { "epoch": 22.7725, "grad_norm": 9.5625, "learning_rate": 1.801867036387165e-05, "loss": 1.5709, "step": 9109 }, { "epoch": 22.775, "grad_norm": 5.78125, "learning_rate": 1.801818654647554e-05, "loss": 1.3477, "step": 9110 }, { "epoch": 22.7775, "grad_norm": 9.25, "learning_rate": 1.8017702676512517e-05, "loss": 1.4283, "step": 9111 }, { "epoch": 22.78, "grad_norm": 8.5625, "learning_rate": 1.8017218753985758e-05, "loss": 1.3695, "step": 9112 }, { "epoch": 22.7825, "grad_norm": 7.8125, "learning_rate": 1.8016734778898436e-05, "loss": 1.5009, "step": 9113 }, { "epoch": 22.785, "grad_norm": 5.65625, "learning_rate": 1.801625075125372e-05, "loss": 1.2937, "step": 9114 }, { "epoch": 22.7875, "grad_norm": 10.9375, "learning_rate": 1.8015766671054783e-05, "loss": 1.3712, "step": 9115 }, { "epoch": 22.79, "grad_norm": 6.78125, "learning_rate": 1.80152825383048e-05, "loss": 1.2882, "step": 9116 }, { "epoch": 22.7925, "grad_norm": 8.625, "learning_rate": 1.8014798353006948e-05, "loss": 1.4248, "step": 9117 }, { "epoch": 22.795, "grad_norm": 8.25, "learning_rate": 1.8014314115164398e-05, "loss": 1.4655, "step": 9118 }, { "epoch": 22.7975, "grad_norm": 6.4375, "learning_rate": 1.801382982478032e-05, "loss": 1.4227, "step": 9119 }, { "epoch": 22.8, "grad_norm": 6.78125, "learning_rate": 1.8013345481857903e-05, "loss": 1.5522, "step": 9120 }, { "epoch": 22.8025, "grad_norm": 5.34375, "learning_rate": 1.801286108640031e-05, "loss": 1.4715, "step": 9121 }, { "epoch": 22.805, "grad_norm": 10.125, "learning_rate": 1.801237663841072e-05, "loss": 1.4085, "step": 9122 }, { "epoch": 22.8075, "grad_norm": 7.84375, "learning_rate": 1.801189213789231e-05, "loss": 1.3779, "step": 9123 }, { "epoch": 22.81, "grad_norm": 4.84375, "learning_rate": 1.8011407584848255e-05, "loss": 1.2916, "step": 9124 }, { "epoch": 22.8125, "grad_norm": 7.8125, "learning_rate": 1.801092297928173e-05, "loss": 1.3898, "step": 9125 }, { "epoch": 22.815, "grad_norm": 6.125, "learning_rate": 1.801043832119592e-05, "loss": 1.4825, "step": 9126 }, { "epoch": 22.8175, "grad_norm": 6.625, "learning_rate": 1.8009953610593993e-05, "loss": 1.4828, "step": 9127 }, { "epoch": 22.82, "grad_norm": 6.59375, "learning_rate": 1.8009468847479134e-05, "loss": 1.4299, "step": 9128 }, { "epoch": 22.8225, "grad_norm": 7.65625, "learning_rate": 1.8008984031854517e-05, "loss": 1.5014, "step": 9129 }, { "epoch": 22.825, "grad_norm": 6.8125, "learning_rate": 1.800849916372332e-05, "loss": 1.3225, "step": 9130 }, { "epoch": 22.8275, "grad_norm": 7.9375, "learning_rate": 1.800801424308872e-05, "loss": 1.3006, "step": 9131 }, { "epoch": 22.83, "grad_norm": 8.0625, "learning_rate": 1.8007529269953902e-05, "loss": 1.3396, "step": 9132 }, { "epoch": 22.8325, "grad_norm": 6.5, "learning_rate": 1.800704424432204e-05, "loss": 1.5406, "step": 9133 }, { "epoch": 22.835, "grad_norm": 6.8125, "learning_rate": 1.8006559166196324e-05, "loss": 1.3594, "step": 9134 }, { "epoch": 22.8375, "grad_norm": 5.875, "learning_rate": 1.800607403557992e-05, "loss": 1.369, "step": 9135 }, { "epoch": 22.84, "grad_norm": 10.25, "learning_rate": 1.8005588852476018e-05, "loss": 1.4556, "step": 9136 }, { "epoch": 22.8425, "grad_norm": 6.15625, "learning_rate": 1.800510361688779e-05, "loss": 1.402, "step": 9137 }, { "epoch": 22.845, "grad_norm": 6.28125, "learning_rate": 1.800461832881843e-05, "loss": 1.3972, "step": 9138 }, { "epoch": 22.8475, "grad_norm": 7.53125, "learning_rate": 1.800413298827111e-05, "loss": 1.5727, "step": 9139 }, { "epoch": 22.85, "grad_norm": 9.1875, "learning_rate": 1.8003647595249016e-05, "loss": 1.4946, "step": 9140 }, { "epoch": 22.8525, "grad_norm": 9.0, "learning_rate": 1.8003162149755324e-05, "loss": 1.5288, "step": 9141 }, { "epoch": 22.855, "grad_norm": 11.125, "learning_rate": 1.8002676651793222e-05, "loss": 1.2559, "step": 9142 }, { "epoch": 22.8575, "grad_norm": 5.75, "learning_rate": 1.8002191101365897e-05, "loss": 1.3714, "step": 9143 }, { "epoch": 22.86, "grad_norm": 6.375, "learning_rate": 1.8001705498476526e-05, "loss": 1.3812, "step": 9144 }, { "epoch": 22.8625, "grad_norm": 6.1875, "learning_rate": 1.800121984312829e-05, "loss": 1.3381, "step": 9145 }, { "epoch": 22.865, "grad_norm": 6.8125, "learning_rate": 1.800073413532438e-05, "loss": 1.4268, "step": 9146 }, { "epoch": 22.8675, "grad_norm": 7.5, "learning_rate": 1.8000248375067977e-05, "loss": 1.4363, "step": 9147 }, { "epoch": 22.87, "grad_norm": 6.125, "learning_rate": 1.7999762562362266e-05, "loss": 1.538, "step": 9148 }, { "epoch": 22.8725, "grad_norm": 7.0625, "learning_rate": 1.799927669721043e-05, "loss": 1.4401, "step": 9149 }, { "epoch": 22.875, "grad_norm": 8.0, "learning_rate": 1.799879077961566e-05, "loss": 1.3918, "step": 9150 }, { "epoch": 22.8775, "grad_norm": 7.75, "learning_rate": 1.7998304809581136e-05, "loss": 1.3911, "step": 9151 }, { "epoch": 22.88, "grad_norm": 6.40625, "learning_rate": 1.7997818787110043e-05, "loss": 1.3001, "step": 9152 }, { "epoch": 22.8825, "grad_norm": 6.9375, "learning_rate": 1.7997332712205572e-05, "loss": 1.3529, "step": 9153 }, { "epoch": 22.885, "grad_norm": 6.0, "learning_rate": 1.799684658487091e-05, "loss": 1.3799, "step": 9154 }, { "epoch": 22.8875, "grad_norm": 6.96875, "learning_rate": 1.7996360405109238e-05, "loss": 1.4626, "step": 9155 }, { "epoch": 22.89, "grad_norm": 10.1875, "learning_rate": 1.799587417292375e-05, "loss": 1.3955, "step": 9156 }, { "epoch": 22.8925, "grad_norm": 6.5, "learning_rate": 1.799538788831763e-05, "loss": 1.3884, "step": 9157 }, { "epoch": 22.895, "grad_norm": 6.8125, "learning_rate": 1.7994901551294062e-05, "loss": 1.2994, "step": 9158 }, { "epoch": 22.8975, "grad_norm": 6.75, "learning_rate": 1.7994415161856246e-05, "loss": 1.4437, "step": 9159 }, { "epoch": 22.9, "grad_norm": 5.6875, "learning_rate": 1.799392872000736e-05, "loss": 1.3961, "step": 9160 }, { "epoch": 22.9025, "grad_norm": 6.71875, "learning_rate": 1.79934422257506e-05, "loss": 1.4509, "step": 9161 }, { "epoch": 22.905, "grad_norm": 7.53125, "learning_rate": 1.7992955679089152e-05, "loss": 1.2775, "step": 9162 }, { "epoch": 22.9075, "grad_norm": 11.875, "learning_rate": 1.799246908002621e-05, "loss": 1.4915, "step": 9163 }, { "epoch": 22.91, "grad_norm": 6.75, "learning_rate": 1.7991982428564953e-05, "loss": 1.3373, "step": 9164 }, { "epoch": 22.9125, "grad_norm": 5.75, "learning_rate": 1.7991495724708584e-05, "loss": 1.3202, "step": 9165 }, { "epoch": 22.915, "grad_norm": 8.1875, "learning_rate": 1.799100896846029e-05, "loss": 1.557, "step": 9166 }, { "epoch": 22.9175, "grad_norm": 6.1875, "learning_rate": 1.7990522159823255e-05, "loss": 1.5265, "step": 9167 }, { "epoch": 22.92, "grad_norm": 7.71875, "learning_rate": 1.7990035298800682e-05, "loss": 1.383, "step": 9168 }, { "epoch": 22.9225, "grad_norm": 9.25, "learning_rate": 1.7989548385395755e-05, "loss": 1.3224, "step": 9169 }, { "epoch": 22.925, "grad_norm": 6.3125, "learning_rate": 1.798906141961167e-05, "loss": 1.4833, "step": 9170 }, { "epoch": 22.9275, "grad_norm": 6.90625, "learning_rate": 1.7988574401451616e-05, "loss": 1.3308, "step": 9171 }, { "epoch": 22.93, "grad_norm": 6.75, "learning_rate": 1.7988087330918787e-05, "loss": 1.2436, "step": 9172 }, { "epoch": 22.9325, "grad_norm": 7.21875, "learning_rate": 1.798760020801638e-05, "loss": 1.4021, "step": 9173 }, { "epoch": 22.935, "grad_norm": 8.25, "learning_rate": 1.7987113032747586e-05, "loss": 1.4247, "step": 9174 }, { "epoch": 22.9375, "grad_norm": 6.59375, "learning_rate": 1.7986625805115597e-05, "loss": 1.482, "step": 9175 }, { "epoch": 22.94, "grad_norm": 8.25, "learning_rate": 1.798613852512361e-05, "loss": 1.3943, "step": 9176 }, { "epoch": 22.9425, "grad_norm": 7.8125, "learning_rate": 1.7985651192774818e-05, "loss": 1.3399, "step": 9177 }, { "epoch": 22.945, "grad_norm": 5.78125, "learning_rate": 1.7985163808072414e-05, "loss": 1.4281, "step": 9178 }, { "epoch": 22.9475, "grad_norm": 11.1875, "learning_rate": 1.79846763710196e-05, "loss": 1.4473, "step": 9179 }, { "epoch": 22.95, "grad_norm": 6.71875, "learning_rate": 1.7984188881619563e-05, "loss": 1.3372, "step": 9180 }, { "epoch": 22.9525, "grad_norm": 6.9375, "learning_rate": 1.7983701339875505e-05, "loss": 1.3191, "step": 9181 }, { "epoch": 22.955, "grad_norm": 6.71875, "learning_rate": 1.7983213745790622e-05, "loss": 1.3759, "step": 9182 }, { "epoch": 22.9575, "grad_norm": 5.46875, "learning_rate": 1.7982726099368106e-05, "loss": 1.3636, "step": 9183 }, { "epoch": 22.96, "grad_norm": 7.625, "learning_rate": 1.798223840061116e-05, "loss": 1.2533, "step": 9184 }, { "epoch": 22.9625, "grad_norm": 4.875, "learning_rate": 1.7981750649522975e-05, "loss": 1.4279, "step": 9185 }, { "epoch": 22.965, "grad_norm": 7.4375, "learning_rate": 1.7981262846106753e-05, "loss": 1.5234, "step": 9186 }, { "epoch": 22.9675, "grad_norm": 5.78125, "learning_rate": 1.7980774990365695e-05, "loss": 1.2439, "step": 9187 }, { "epoch": 22.97, "grad_norm": 6.625, "learning_rate": 1.7980287082302996e-05, "loss": 1.3654, "step": 9188 }, { "epoch": 22.9725, "grad_norm": 7.03125, "learning_rate": 1.797979912192185e-05, "loss": 1.5448, "step": 9189 }, { "epoch": 22.975, "grad_norm": 8.6875, "learning_rate": 1.797931110922546e-05, "loss": 1.5606, "step": 9190 }, { "epoch": 22.9775, "grad_norm": 12.1875, "learning_rate": 1.797882304421703e-05, "loss": 1.4606, "step": 9191 }, { "epoch": 22.98, "grad_norm": 7.09375, "learning_rate": 1.7978334926899748e-05, "loss": 1.3821, "step": 9192 }, { "epoch": 22.9825, "grad_norm": 9.625, "learning_rate": 1.797784675727683e-05, "loss": 1.345, "step": 9193 }, { "epoch": 22.985, "grad_norm": 8.875, "learning_rate": 1.7977358535351464e-05, "loss": 1.1405, "step": 9194 }, { "epoch": 22.9875, "grad_norm": 8.625, "learning_rate": 1.7976870261126853e-05, "loss": 1.4327, "step": 9195 }, { "epoch": 22.99, "grad_norm": 7.03125, "learning_rate": 1.79763819346062e-05, "loss": 1.481, "step": 9196 }, { "epoch": 22.9925, "grad_norm": 7.8125, "learning_rate": 1.797589355579271e-05, "loss": 1.4537, "step": 9197 }, { "epoch": 22.995, "grad_norm": 8.25, "learning_rate": 1.797540512468958e-05, "loss": 1.445, "step": 9198 }, { "epoch": 22.9975, "grad_norm": 9.1875, "learning_rate": 1.797491664130001e-05, "loss": 1.4874, "step": 9199 }, { "epoch": 23.0, "grad_norm": 5.40625, "learning_rate": 1.797442810562721e-05, "loss": 1.2839, "step": 9200 }, { "epoch": 23.0025, "grad_norm": 7.71875, "learning_rate": 1.7973939517674375e-05, "loss": 1.3608, "step": 9201 }, { "epoch": 23.005, "grad_norm": 6.53125, "learning_rate": 1.7973450877444713e-05, "loss": 1.4543, "step": 9202 }, { "epoch": 23.0075, "grad_norm": 5.15625, "learning_rate": 1.7972962184941426e-05, "loss": 1.3478, "step": 9203 }, { "epoch": 23.01, "grad_norm": 9.625, "learning_rate": 1.7972473440167716e-05, "loss": 1.4657, "step": 9204 }, { "epoch": 23.0125, "grad_norm": 8.9375, "learning_rate": 1.7971984643126794e-05, "loss": 1.364, "step": 9205 }, { "epoch": 23.015, "grad_norm": 6.75, "learning_rate": 1.7971495793821855e-05, "loss": 1.4401, "step": 9206 }, { "epoch": 23.0175, "grad_norm": 7.34375, "learning_rate": 1.7971006892256112e-05, "loss": 1.1945, "step": 9207 }, { "epoch": 23.02, "grad_norm": 7.15625, "learning_rate": 1.7970517938432766e-05, "loss": 1.4517, "step": 9208 }, { "epoch": 23.0225, "grad_norm": 5.28125, "learning_rate": 1.7970028932355022e-05, "loss": 1.3677, "step": 9209 }, { "epoch": 23.025, "grad_norm": 7.1875, "learning_rate": 1.7969539874026086e-05, "loss": 1.4345, "step": 9210 }, { "epoch": 23.0275, "grad_norm": 9.0625, "learning_rate": 1.796905076344917e-05, "loss": 1.5214, "step": 9211 }, { "epoch": 23.03, "grad_norm": 7.65625, "learning_rate": 1.796856160062747e-05, "loss": 1.5762, "step": 9212 }, { "epoch": 23.0325, "grad_norm": 7.96875, "learning_rate": 1.7968072385564205e-05, "loss": 1.4542, "step": 9213 }, { "epoch": 23.035, "grad_norm": 5.15625, "learning_rate": 1.796758311826257e-05, "loss": 1.4281, "step": 9214 }, { "epoch": 23.0375, "grad_norm": 5.71875, "learning_rate": 1.7967093798725784e-05, "loss": 1.4273, "step": 9215 }, { "epoch": 23.04, "grad_norm": 7.0625, "learning_rate": 1.796660442695705e-05, "loss": 1.2945, "step": 9216 }, { "epoch": 23.0425, "grad_norm": 19.25, "learning_rate": 1.7966115002959575e-05, "loss": 1.5314, "step": 9217 }, { "epoch": 23.045, "grad_norm": 10.9375, "learning_rate": 1.7965625526736567e-05, "loss": 1.4432, "step": 9218 }, { "epoch": 23.0475, "grad_norm": 5.65625, "learning_rate": 1.796513599829124e-05, "loss": 1.5013, "step": 9219 }, { "epoch": 23.05, "grad_norm": 7.8125, "learning_rate": 1.79646464176268e-05, "loss": 1.2389, "step": 9220 }, { "epoch": 23.0525, "grad_norm": 5.96875, "learning_rate": 1.7964156784746454e-05, "loss": 1.3723, "step": 9221 }, { "epoch": 23.055, "grad_norm": 8.6875, "learning_rate": 1.7963667099653418e-05, "loss": 1.4417, "step": 9222 }, { "epoch": 23.0575, "grad_norm": 7.90625, "learning_rate": 1.7963177362350894e-05, "loss": 1.323, "step": 9223 }, { "epoch": 23.06, "grad_norm": 7.3125, "learning_rate": 1.7962687572842102e-05, "loss": 1.5351, "step": 9224 }, { "epoch": 23.0625, "grad_norm": 6.03125, "learning_rate": 1.796219773113025e-05, "loss": 1.3393, "step": 9225 }, { "epoch": 23.065, "grad_norm": 8.125, "learning_rate": 1.7961707837218546e-05, "loss": 1.3244, "step": 9226 }, { "epoch": 23.0675, "grad_norm": 6.15625, "learning_rate": 1.79612178911102e-05, "loss": 1.3449, "step": 9227 }, { "epoch": 23.07, "grad_norm": 5.21875, "learning_rate": 1.7960727892808437e-05, "loss": 1.4254, "step": 9228 }, { "epoch": 23.0725, "grad_norm": 5.65625, "learning_rate": 1.7960237842316453e-05, "loss": 1.365, "step": 9229 }, { "epoch": 23.075, "grad_norm": 8.25, "learning_rate": 1.7959747739637473e-05, "loss": 1.573, "step": 9230 }, { "epoch": 23.0775, "grad_norm": 5.5625, "learning_rate": 1.7959257584774703e-05, "loss": 1.4639, "step": 9231 }, { "epoch": 23.08, "grad_norm": 6.28125, "learning_rate": 1.795876737773136e-05, "loss": 1.2944, "step": 9232 }, { "epoch": 23.0825, "grad_norm": 6.3125, "learning_rate": 1.7958277118510654e-05, "loss": 1.3781, "step": 9233 }, { "epoch": 23.085, "grad_norm": 6.71875, "learning_rate": 1.7957786807115806e-05, "loss": 1.3171, "step": 9234 }, { "epoch": 23.0875, "grad_norm": 5.4375, "learning_rate": 1.7957296443550023e-05, "loss": 1.4086, "step": 9235 }, { "epoch": 23.09, "grad_norm": 7.59375, "learning_rate": 1.795680602781652e-05, "loss": 1.3053, "step": 9236 }, { "epoch": 23.0925, "grad_norm": 4.53125, "learning_rate": 1.7956315559918523e-05, "loss": 1.1254, "step": 9237 }, { "epoch": 23.095, "grad_norm": 7.34375, "learning_rate": 1.7955825039859233e-05, "loss": 1.4155, "step": 9238 }, { "epoch": 23.0975, "grad_norm": 7.21875, "learning_rate": 1.795533446764188e-05, "loss": 1.3178, "step": 9239 }, { "epoch": 23.1, "grad_norm": 7.4375, "learning_rate": 1.7954843843269665e-05, "loss": 1.4477, "step": 9240 }, { "epoch": 23.1025, "grad_norm": 6.0625, "learning_rate": 1.7954353166745812e-05, "loss": 1.348, "step": 9241 }, { "epoch": 23.105, "grad_norm": 8.4375, "learning_rate": 1.795386243807354e-05, "loss": 1.3075, "step": 9242 }, { "epoch": 23.1075, "grad_norm": 7.8125, "learning_rate": 1.7953371657256066e-05, "loss": 1.4429, "step": 9243 }, { "epoch": 23.11, "grad_norm": 7.9375, "learning_rate": 1.7952880824296605e-05, "loss": 1.3217, "step": 9244 }, { "epoch": 23.1125, "grad_norm": 5.46875, "learning_rate": 1.7952389939198376e-05, "loss": 1.5036, "step": 9245 }, { "epoch": 23.115, "grad_norm": 6.65625, "learning_rate": 1.7951899001964595e-05, "loss": 1.4114, "step": 9246 }, { "epoch": 23.1175, "grad_norm": 6.78125, "learning_rate": 1.7951408012598484e-05, "loss": 1.3529, "step": 9247 }, { "epoch": 23.12, "grad_norm": 5.4375, "learning_rate": 1.795091697110326e-05, "loss": 1.3727, "step": 9248 }, { "epoch": 23.1225, "grad_norm": 7.5625, "learning_rate": 1.7950425877482137e-05, "loss": 1.4703, "step": 9249 }, { "epoch": 23.125, "grad_norm": 9.625, "learning_rate": 1.7949934731738348e-05, "loss": 1.2818, "step": 9250 }, { "epoch": 23.1275, "grad_norm": 9.375, "learning_rate": 1.79494435338751e-05, "loss": 1.3875, "step": 9251 }, { "epoch": 23.13, "grad_norm": 6.8125, "learning_rate": 1.794895228389562e-05, "loss": 1.3616, "step": 9252 }, { "epoch": 23.1325, "grad_norm": 4.5, "learning_rate": 1.794846098180313e-05, "loss": 1.3531, "step": 9253 }, { "epoch": 23.135, "grad_norm": 8.25, "learning_rate": 1.7947969627600846e-05, "loss": 1.3542, "step": 9254 }, { "epoch": 23.1375, "grad_norm": 4.75, "learning_rate": 1.7947478221291986e-05, "loss": 1.367, "step": 9255 }, { "epoch": 23.14, "grad_norm": 8.5625, "learning_rate": 1.7946986762879785e-05, "loss": 1.2547, "step": 9256 }, { "epoch": 23.1425, "grad_norm": 7.78125, "learning_rate": 1.794649525236745e-05, "loss": 1.2862, "step": 9257 }, { "epoch": 23.145, "grad_norm": 8.25, "learning_rate": 1.7946003689758216e-05, "loss": 1.2823, "step": 9258 }, { "epoch": 23.1475, "grad_norm": 7.75, "learning_rate": 1.7945512075055297e-05, "loss": 1.2881, "step": 9259 }, { "epoch": 23.15, "grad_norm": 6.125, "learning_rate": 1.794502040826192e-05, "loss": 1.3047, "step": 9260 }, { "epoch": 23.1525, "grad_norm": 6.34375, "learning_rate": 1.7944528689381306e-05, "loss": 1.3288, "step": 9261 }, { "epoch": 23.155, "grad_norm": 6.625, "learning_rate": 1.7944036918416677e-05, "loss": 1.3581, "step": 9262 }, { "epoch": 23.1575, "grad_norm": 8.625, "learning_rate": 1.7943545095371263e-05, "loss": 1.3813, "step": 9263 }, { "epoch": 23.16, "grad_norm": 8.4375, "learning_rate": 1.7943053220248284e-05, "loss": 1.5026, "step": 9264 }, { "epoch": 23.1625, "grad_norm": 5.75, "learning_rate": 1.7942561293050967e-05, "loss": 1.3503, "step": 9265 }, { "epoch": 23.165, "grad_norm": 7.1875, "learning_rate": 1.7942069313782536e-05, "loss": 1.3679, "step": 9266 }, { "epoch": 23.1675, "grad_norm": 6.5625, "learning_rate": 1.7941577282446214e-05, "loss": 1.4419, "step": 9267 }, { "epoch": 23.17, "grad_norm": 8.875, "learning_rate": 1.794108519904523e-05, "loss": 1.3145, "step": 9268 }, { "epoch": 23.1725, "grad_norm": 5.0, "learning_rate": 1.7940593063582813e-05, "loss": 1.2799, "step": 9269 }, { "epoch": 23.175, "grad_norm": 6.0, "learning_rate": 1.7940100876062184e-05, "loss": 1.4278, "step": 9270 }, { "epoch": 23.1775, "grad_norm": 5.0625, "learning_rate": 1.793960863648657e-05, "loss": 1.4141, "step": 9271 }, { "epoch": 23.18, "grad_norm": 9.25, "learning_rate": 1.79391163448592e-05, "loss": 1.4018, "step": 9272 }, { "epoch": 23.1825, "grad_norm": 8.375, "learning_rate": 1.79386240011833e-05, "loss": 1.5609, "step": 9273 }, { "epoch": 23.185, "grad_norm": 6.3125, "learning_rate": 1.79381316054621e-05, "loss": 1.489, "step": 9274 }, { "epoch": 23.1875, "grad_norm": 9.25, "learning_rate": 1.7937639157698825e-05, "loss": 1.4549, "step": 9275 }, { "epoch": 23.19, "grad_norm": 5.1875, "learning_rate": 1.7937146657896708e-05, "loss": 1.2509, "step": 9276 }, { "epoch": 23.1925, "grad_norm": 4.875, "learning_rate": 1.7936654106058975e-05, "loss": 1.4009, "step": 9277 }, { "epoch": 23.195, "grad_norm": 8.125, "learning_rate": 1.7936161502188853e-05, "loss": 1.6539, "step": 9278 }, { "epoch": 23.1975, "grad_norm": 6.3125, "learning_rate": 1.7935668846289575e-05, "loss": 1.37, "step": 9279 }, { "epoch": 23.2, "grad_norm": 5.40625, "learning_rate": 1.793517613836437e-05, "loss": 1.2731, "step": 9280 }, { "epoch": 23.2025, "grad_norm": 5.3125, "learning_rate": 1.7934683378416465e-05, "loss": 1.4113, "step": 9281 }, { "epoch": 23.205, "grad_norm": 6.6875, "learning_rate": 1.79341905664491e-05, "loss": 1.4229, "step": 9282 }, { "epoch": 23.2075, "grad_norm": 7.84375, "learning_rate": 1.7933697702465493e-05, "loss": 1.3346, "step": 9283 }, { "epoch": 23.21, "grad_norm": 6.375, "learning_rate": 1.7933204786468883e-05, "loss": 1.3364, "step": 9284 }, { "epoch": 23.2125, "grad_norm": 8.0, "learning_rate": 1.79327118184625e-05, "loss": 1.3676, "step": 9285 }, { "epoch": 23.215, "grad_norm": 7.90625, "learning_rate": 1.7932218798449577e-05, "loss": 1.5385, "step": 9286 }, { "epoch": 23.2175, "grad_norm": 6.5625, "learning_rate": 1.7931725726433343e-05, "loss": 1.306, "step": 9287 }, { "epoch": 23.22, "grad_norm": 8.125, "learning_rate": 1.7931232602417035e-05, "loss": 1.5151, "step": 9288 }, { "epoch": 23.2225, "grad_norm": 7.25, "learning_rate": 1.793073942640388e-05, "loss": 1.3059, "step": 9289 }, { "epoch": 23.225, "grad_norm": 9.1875, "learning_rate": 1.7930246198397118e-05, "loss": 1.3266, "step": 9290 }, { "epoch": 23.2275, "grad_norm": 8.8125, "learning_rate": 1.7929752918399977e-05, "loss": 1.3212, "step": 9291 }, { "epoch": 23.23, "grad_norm": 8.5625, "learning_rate": 1.7929259586415694e-05, "loss": 1.4004, "step": 9292 }, { "epoch": 23.2325, "grad_norm": 5.21875, "learning_rate": 1.7928766202447502e-05, "loss": 1.3195, "step": 9293 }, { "epoch": 23.235, "grad_norm": 7.5, "learning_rate": 1.792827276649864e-05, "loss": 1.4102, "step": 9294 }, { "epoch": 23.2375, "grad_norm": 5.0625, "learning_rate": 1.7927779278572332e-05, "loss": 1.3185, "step": 9295 }, { "epoch": 23.24, "grad_norm": 8.875, "learning_rate": 1.7927285738671825e-05, "loss": 1.3547, "step": 9296 }, { "epoch": 23.2425, "grad_norm": 6.34375, "learning_rate": 1.7926792146800348e-05, "loss": 1.3378, "step": 9297 }, { "epoch": 23.245, "grad_norm": 10.3125, "learning_rate": 1.792629850296114e-05, "loss": 1.4322, "step": 9298 }, { "epoch": 23.2475, "grad_norm": 8.0, "learning_rate": 1.7925804807157436e-05, "loss": 1.3376, "step": 9299 }, { "epoch": 23.25, "grad_norm": 8.9375, "learning_rate": 1.7925311059392472e-05, "loss": 1.4935, "step": 9300 }, { "epoch": 23.2525, "grad_norm": 7.21875, "learning_rate": 1.7924817259669488e-05, "loss": 1.2795, "step": 9301 }, { "epoch": 23.255, "grad_norm": 6.8125, "learning_rate": 1.792432340799172e-05, "loss": 1.6452, "step": 9302 }, { "epoch": 23.2575, "grad_norm": 5.5, "learning_rate": 1.79238295043624e-05, "loss": 1.1848, "step": 9303 }, { "epoch": 23.26, "grad_norm": 6.65625, "learning_rate": 1.7923335548784773e-05, "loss": 1.3634, "step": 9304 }, { "epoch": 23.2625, "grad_norm": 5.03125, "learning_rate": 1.7922841541262076e-05, "loss": 1.4836, "step": 9305 }, { "epoch": 23.265, "grad_norm": 8.125, "learning_rate": 1.7922347481797546e-05, "loss": 1.2215, "step": 9306 }, { "epoch": 23.2675, "grad_norm": 12.1875, "learning_rate": 1.7921853370394426e-05, "loss": 1.4861, "step": 9307 }, { "epoch": 23.27, "grad_norm": 6.875, "learning_rate": 1.792135920705595e-05, "loss": 1.4566, "step": 9308 }, { "epoch": 23.2725, "grad_norm": 7.9375, "learning_rate": 1.7920864991785357e-05, "loss": 1.4636, "step": 9309 }, { "epoch": 23.275, "grad_norm": 4.75, "learning_rate": 1.7920370724585895e-05, "loss": 1.4545, "step": 9310 }, { "epoch": 23.2775, "grad_norm": 7.03125, "learning_rate": 1.7919876405460796e-05, "loss": 1.3637, "step": 9311 }, { "epoch": 23.28, "grad_norm": 7.40625, "learning_rate": 1.7919382034413306e-05, "loss": 1.4248, "step": 9312 }, { "epoch": 23.2825, "grad_norm": 5.65625, "learning_rate": 1.7918887611446664e-05, "loss": 1.3338, "step": 9313 }, { "epoch": 23.285, "grad_norm": 4.8125, "learning_rate": 1.7918393136564112e-05, "loss": 1.2792, "step": 9314 }, { "epoch": 23.2875, "grad_norm": 7.40625, "learning_rate": 1.7917898609768893e-05, "loss": 1.3949, "step": 9315 }, { "epoch": 23.29, "grad_norm": 7.25, "learning_rate": 1.7917404031064245e-05, "loss": 1.3336, "step": 9316 }, { "epoch": 23.2925, "grad_norm": 5.96875, "learning_rate": 1.7916909400453412e-05, "loss": 1.4391, "step": 9317 }, { "epoch": 23.295, "grad_norm": 8.4375, "learning_rate": 1.7916414717939642e-05, "loss": 1.4582, "step": 9318 }, { "epoch": 23.2975, "grad_norm": 5.53125, "learning_rate": 1.7915919983526172e-05, "loss": 1.2769, "step": 9319 }, { "epoch": 23.3, "grad_norm": 5.65625, "learning_rate": 1.7915425197216246e-05, "loss": 1.2958, "step": 9320 }, { "epoch": 23.3025, "grad_norm": 8.75, "learning_rate": 1.791493035901311e-05, "loss": 1.3538, "step": 9321 }, { "epoch": 23.305, "grad_norm": 7.0625, "learning_rate": 1.791443546892001e-05, "loss": 1.3856, "step": 9322 }, { "epoch": 23.3075, "grad_norm": 6.9375, "learning_rate": 1.791394052694018e-05, "loss": 1.4354, "step": 9323 }, { "epoch": 23.31, "grad_norm": 7.46875, "learning_rate": 1.791344553307688e-05, "loss": 1.2958, "step": 9324 }, { "epoch": 23.3125, "grad_norm": 7.1875, "learning_rate": 1.7912950487333345e-05, "loss": 1.3979, "step": 9325 }, { "epoch": 23.315, "grad_norm": 6.59375, "learning_rate": 1.7912455389712824e-05, "loss": 1.3436, "step": 9326 }, { "epoch": 23.3175, "grad_norm": 6.0625, "learning_rate": 1.7911960240218562e-05, "loss": 1.4399, "step": 9327 }, { "epoch": 23.32, "grad_norm": 7.1875, "learning_rate": 1.7911465038853805e-05, "loss": 1.3611, "step": 9328 }, { "epoch": 23.3225, "grad_norm": 9.625, "learning_rate": 1.79109697856218e-05, "loss": 1.5135, "step": 9329 }, { "epoch": 23.325, "grad_norm": 10.5625, "learning_rate": 1.7910474480525795e-05, "loss": 1.4451, "step": 9330 }, { "epoch": 23.3275, "grad_norm": 6.8125, "learning_rate": 1.790997912356903e-05, "loss": 1.5393, "step": 9331 }, { "epoch": 23.33, "grad_norm": 9.375, "learning_rate": 1.7909483714754765e-05, "loss": 1.4071, "step": 9332 }, { "epoch": 23.3325, "grad_norm": 5.625, "learning_rate": 1.790898825408624e-05, "loss": 1.292, "step": 9333 }, { "epoch": 23.335, "grad_norm": 7.875, "learning_rate": 1.7908492741566704e-05, "loss": 1.4912, "step": 9334 }, { "epoch": 23.3375, "grad_norm": 8.0625, "learning_rate": 1.7907997177199405e-05, "loss": 1.3797, "step": 9335 }, { "epoch": 23.34, "grad_norm": 7.625, "learning_rate": 1.7907501560987594e-05, "loss": 1.461, "step": 9336 }, { "epoch": 23.3425, "grad_norm": 13.0, "learning_rate": 1.790700589293452e-05, "loss": 1.3587, "step": 9337 }, { "epoch": 23.345, "grad_norm": 6.84375, "learning_rate": 1.790651017304343e-05, "loss": 1.4699, "step": 9338 }, { "epoch": 23.3475, "grad_norm": 5.78125, "learning_rate": 1.7906014401317576e-05, "loss": 1.3682, "step": 9339 }, { "epoch": 23.35, "grad_norm": 4.6875, "learning_rate": 1.7905518577760207e-05, "loss": 1.6577, "step": 9340 }, { "epoch": 23.3525, "grad_norm": 5.3125, "learning_rate": 1.7905022702374576e-05, "loss": 1.4474, "step": 9341 }, { "epoch": 23.355, "grad_norm": 7.125, "learning_rate": 1.7904526775163935e-05, "loss": 1.4641, "step": 9342 }, { "epoch": 23.3575, "grad_norm": 9.0625, "learning_rate": 1.790403079613153e-05, "loss": 1.4926, "step": 9343 }, { "epoch": 23.36, "grad_norm": 5.90625, "learning_rate": 1.7903534765280616e-05, "loss": 1.3743, "step": 9344 }, { "epoch": 23.3625, "grad_norm": 5.875, "learning_rate": 1.7903038682614442e-05, "loss": 1.5569, "step": 9345 }, { "epoch": 23.365, "grad_norm": 6.75, "learning_rate": 1.7902542548136262e-05, "loss": 1.4065, "step": 9346 }, { "epoch": 23.3675, "grad_norm": 6.6875, "learning_rate": 1.7902046361849336e-05, "loss": 1.5044, "step": 9347 }, { "epoch": 23.37, "grad_norm": 6.40625, "learning_rate": 1.7901550123756906e-05, "loss": 1.4377, "step": 9348 }, { "epoch": 23.3725, "grad_norm": 6.84375, "learning_rate": 1.790105383386223e-05, "loss": 1.4316, "step": 9349 }, { "epoch": 23.375, "grad_norm": 7.0625, "learning_rate": 1.790055749216856e-05, "loss": 1.3142, "step": 9350 }, { "epoch": 23.3775, "grad_norm": 6.1875, "learning_rate": 1.7900061098679154e-05, "loss": 1.418, "step": 9351 }, { "epoch": 23.38, "grad_norm": 7.34375, "learning_rate": 1.789956465339726e-05, "loss": 1.2668, "step": 9352 }, { "epoch": 23.3825, "grad_norm": 5.25, "learning_rate": 1.789906815632614e-05, "loss": 1.3125, "step": 9353 }, { "epoch": 23.385, "grad_norm": 8.5, "learning_rate": 1.789857160746904e-05, "loss": 1.4741, "step": 9354 }, { "epoch": 23.3875, "grad_norm": 7.8125, "learning_rate": 1.7898075006829226e-05, "loss": 1.3821, "step": 9355 }, { "epoch": 23.39, "grad_norm": 6.78125, "learning_rate": 1.7897578354409945e-05, "loss": 1.3107, "step": 9356 }, { "epoch": 23.3925, "grad_norm": 6.625, "learning_rate": 1.7897081650214457e-05, "loss": 1.458, "step": 9357 }, { "epoch": 23.395, "grad_norm": 7.0, "learning_rate": 1.7896584894246022e-05, "loss": 1.4663, "step": 9358 }, { "epoch": 23.3975, "grad_norm": 12.4375, "learning_rate": 1.7896088086507885e-05, "loss": 1.4606, "step": 9359 }, { "epoch": 23.4, "grad_norm": 9.3125, "learning_rate": 1.7895591227003316e-05, "loss": 1.563, "step": 9360 }, { "epoch": 23.4025, "grad_norm": 6.40625, "learning_rate": 1.7895094315735564e-05, "loss": 1.3856, "step": 9361 }, { "epoch": 23.405, "grad_norm": 7.21875, "learning_rate": 1.7894597352707892e-05, "loss": 1.4136, "step": 9362 }, { "epoch": 23.4075, "grad_norm": 8.6875, "learning_rate": 1.7894100337923556e-05, "loss": 1.444, "step": 9363 }, { "epoch": 23.41, "grad_norm": 6.125, "learning_rate": 1.789360327138581e-05, "loss": 1.563, "step": 9364 }, { "epoch": 23.4125, "grad_norm": 9.375, "learning_rate": 1.7893106153097922e-05, "loss": 1.2218, "step": 9365 }, { "epoch": 23.415, "grad_norm": 8.375, "learning_rate": 1.7892608983063143e-05, "loss": 1.4101, "step": 9366 }, { "epoch": 23.4175, "grad_norm": 6.5625, "learning_rate": 1.7892111761284737e-05, "loss": 1.362, "step": 9367 }, { "epoch": 23.42, "grad_norm": 7.4375, "learning_rate": 1.789161448776596e-05, "loss": 1.3216, "step": 9368 }, { "epoch": 23.4225, "grad_norm": 7.0, "learning_rate": 1.7891117162510073e-05, "loss": 1.4436, "step": 9369 }, { "epoch": 23.425, "grad_norm": 8.375, "learning_rate": 1.789061978552034e-05, "loss": 1.5074, "step": 9370 }, { "epoch": 23.4275, "grad_norm": 6.5, "learning_rate": 1.789012235680002e-05, "loss": 1.2461, "step": 9371 }, { "epoch": 23.43, "grad_norm": 4.59375, "learning_rate": 1.788962487635237e-05, "loss": 1.4746, "step": 9372 }, { "epoch": 23.4325, "grad_norm": 6.8125, "learning_rate": 1.788912734418066e-05, "loss": 1.3023, "step": 9373 }, { "epoch": 23.435, "grad_norm": 7.5625, "learning_rate": 1.7888629760288147e-05, "loss": 1.289, "step": 9374 }, { "epoch": 23.4375, "grad_norm": 7.3125, "learning_rate": 1.788813212467809e-05, "loss": 1.4507, "step": 9375 }, { "epoch": 23.44, "grad_norm": 7.03125, "learning_rate": 1.7887634437353754e-05, "loss": 1.3567, "step": 9376 }, { "epoch": 23.4425, "grad_norm": 8.6875, "learning_rate": 1.7887136698318402e-05, "loss": 1.3508, "step": 9377 }, { "epoch": 23.445, "grad_norm": 7.8125, "learning_rate": 1.78866389075753e-05, "loss": 1.4384, "step": 9378 }, { "epoch": 23.4475, "grad_norm": 7.65625, "learning_rate": 1.7886141065127706e-05, "loss": 1.4176, "step": 9379 }, { "epoch": 23.45, "grad_norm": 10.6875, "learning_rate": 1.788564317097889e-05, "loss": 1.3703, "step": 9380 }, { "epoch": 23.4525, "grad_norm": 6.15625, "learning_rate": 1.7885145225132114e-05, "loss": 1.4709, "step": 9381 }, { "epoch": 23.455, "grad_norm": 6.15625, "learning_rate": 1.7884647227590638e-05, "loss": 1.3469, "step": 9382 }, { "epoch": 23.4575, "grad_norm": 8.1875, "learning_rate": 1.7884149178357726e-05, "loss": 1.3646, "step": 9383 }, { "epoch": 23.46, "grad_norm": 6.03125, "learning_rate": 1.7883651077436656e-05, "loss": 1.2471, "step": 9384 }, { "epoch": 23.4625, "grad_norm": 10.0625, "learning_rate": 1.788315292483068e-05, "loss": 1.5356, "step": 9385 }, { "epoch": 23.465, "grad_norm": 8.3125, "learning_rate": 1.7882654720543073e-05, "loss": 1.4974, "step": 9386 }, { "epoch": 23.4675, "grad_norm": 4.9375, "learning_rate": 1.7882156464577093e-05, "loss": 1.3979, "step": 9387 }, { "epoch": 23.47, "grad_norm": 10.0, "learning_rate": 1.7881658156936012e-05, "loss": 1.3758, "step": 9388 }, { "epoch": 23.4725, "grad_norm": 9.375, "learning_rate": 1.7881159797623095e-05, "loss": 1.515, "step": 9389 }, { "epoch": 23.475, "grad_norm": 5.1875, "learning_rate": 1.7880661386641612e-05, "loss": 1.3938, "step": 9390 }, { "epoch": 23.4775, "grad_norm": 6.0625, "learning_rate": 1.788016292399483e-05, "loss": 1.406, "step": 9391 }, { "epoch": 23.48, "grad_norm": 7.4375, "learning_rate": 1.7879664409686007e-05, "loss": 1.4964, "step": 9392 }, { "epoch": 23.4825, "grad_norm": 8.0625, "learning_rate": 1.7879165843718427e-05, "loss": 1.4321, "step": 9393 }, { "epoch": 23.485, "grad_norm": 6.78125, "learning_rate": 1.787866722609535e-05, "loss": 1.4744, "step": 9394 }, { "epoch": 23.4875, "grad_norm": 5.78125, "learning_rate": 1.7878168556820042e-05, "loss": 1.2209, "step": 9395 }, { "epoch": 23.49, "grad_norm": 6.53125, "learning_rate": 1.787766983589578e-05, "loss": 1.4645, "step": 9396 }, { "epoch": 23.4925, "grad_norm": 8.125, "learning_rate": 1.7877171063325826e-05, "loss": 1.3738, "step": 9397 }, { "epoch": 23.495, "grad_norm": 5.78125, "learning_rate": 1.787667223911346e-05, "loss": 1.4102, "step": 9398 }, { "epoch": 23.4975, "grad_norm": 4.96875, "learning_rate": 1.787617336326194e-05, "loss": 1.3859, "step": 9399 }, { "epoch": 23.5, "grad_norm": 8.125, "learning_rate": 1.7875674435774546e-05, "loss": 1.4654, "step": 9400 }, { "epoch": 23.5025, "grad_norm": 7.5625, "learning_rate": 1.7875175456654547e-05, "loss": 1.387, "step": 9401 }, { "epoch": 23.505, "grad_norm": 6.59375, "learning_rate": 1.787467642590521e-05, "loss": 1.3199, "step": 9402 }, { "epoch": 23.5075, "grad_norm": 6.09375, "learning_rate": 1.7874177343529807e-05, "loss": 1.3322, "step": 9403 }, { "epoch": 23.51, "grad_norm": 6.75, "learning_rate": 1.7873678209531615e-05, "loss": 1.3614, "step": 9404 }, { "epoch": 23.5125, "grad_norm": 9.9375, "learning_rate": 1.7873179023913905e-05, "loss": 1.4982, "step": 9405 }, { "epoch": 23.515, "grad_norm": 8.4375, "learning_rate": 1.7872679786679947e-05, "loss": 1.3656, "step": 9406 }, { "epoch": 23.5175, "grad_norm": 6.34375, "learning_rate": 1.7872180497833013e-05, "loss": 1.2994, "step": 9407 }, { "epoch": 23.52, "grad_norm": 5.71875, "learning_rate": 1.7871681157376382e-05, "loss": 1.4789, "step": 9408 }, { "epoch": 23.5225, "grad_norm": 6.53125, "learning_rate": 1.7871181765313327e-05, "loss": 1.4893, "step": 9409 }, { "epoch": 23.525, "grad_norm": 10.625, "learning_rate": 1.7870682321647113e-05, "loss": 1.454, "step": 9410 }, { "epoch": 23.5275, "grad_norm": 8.1875, "learning_rate": 1.7870182826381025e-05, "loss": 1.3687, "step": 9411 }, { "epoch": 23.53, "grad_norm": 7.0, "learning_rate": 1.7869683279518334e-05, "loss": 1.4056, "step": 9412 }, { "epoch": 23.5325, "grad_norm": 6.8125, "learning_rate": 1.786918368106231e-05, "loss": 1.5556, "step": 9413 }, { "epoch": 23.535, "grad_norm": 5.65625, "learning_rate": 1.7868684031016236e-05, "loss": 1.478, "step": 9414 }, { "epoch": 23.5375, "grad_norm": 6.3125, "learning_rate": 1.7868184329383384e-05, "loss": 1.4066, "step": 9415 }, { "epoch": 23.54, "grad_norm": 11.8125, "learning_rate": 1.7867684576167028e-05, "loss": 1.51, "step": 9416 }, { "epoch": 23.5425, "grad_norm": 7.90625, "learning_rate": 1.7867184771370452e-05, "loss": 1.3611, "step": 9417 }, { "epoch": 23.545, "grad_norm": 6.34375, "learning_rate": 1.7866684914996922e-05, "loss": 1.3793, "step": 9418 }, { "epoch": 23.5475, "grad_norm": 6.1875, "learning_rate": 1.7866185007049724e-05, "loss": 1.4083, "step": 9419 }, { "epoch": 23.55, "grad_norm": 6.0625, "learning_rate": 1.786568504753213e-05, "loss": 1.3994, "step": 9420 }, { "epoch": 23.5525, "grad_norm": 8.1875, "learning_rate": 1.786518503644742e-05, "loss": 1.4048, "step": 9421 }, { "epoch": 23.555, "grad_norm": 6.59375, "learning_rate": 1.7864684973798873e-05, "loss": 1.5205, "step": 9422 }, { "epoch": 23.5575, "grad_norm": 5.40625, "learning_rate": 1.7864184859589764e-05, "loss": 1.2004, "step": 9423 }, { "epoch": 23.56, "grad_norm": 5.84375, "learning_rate": 1.7863684693823375e-05, "loss": 1.3822, "step": 9424 }, { "epoch": 23.5625, "grad_norm": 7.3125, "learning_rate": 1.7863184476502984e-05, "loss": 1.351, "step": 9425 }, { "epoch": 23.565, "grad_norm": 6.21875, "learning_rate": 1.786268420763187e-05, "loss": 1.4766, "step": 9426 }, { "epoch": 23.5675, "grad_norm": 7.46875, "learning_rate": 1.786218388721331e-05, "loss": 1.5037, "step": 9427 }, { "epoch": 23.57, "grad_norm": 9.4375, "learning_rate": 1.7861683515250588e-05, "loss": 1.3356, "step": 9428 }, { "epoch": 23.5725, "grad_norm": 9.5, "learning_rate": 1.7861183091746988e-05, "loss": 1.426, "step": 9429 }, { "epoch": 23.575, "grad_norm": 6.6875, "learning_rate": 1.7860682616705784e-05, "loss": 1.3845, "step": 9430 }, { "epoch": 23.5775, "grad_norm": 50.75, "learning_rate": 1.7860182090130258e-05, "loss": 1.3812, "step": 9431 }, { "epoch": 23.58, "grad_norm": 7.15625, "learning_rate": 1.7859681512023694e-05, "loss": 1.4746, "step": 9432 }, { "epoch": 23.5825, "grad_norm": 6.1875, "learning_rate": 1.785918088238937e-05, "loss": 1.2454, "step": 9433 }, { "epoch": 23.585, "grad_norm": 7.03125, "learning_rate": 1.7858680201230574e-05, "loss": 1.4117, "step": 9434 }, { "epoch": 23.5875, "grad_norm": 14.25, "learning_rate": 1.7858179468550583e-05, "loss": 1.4078, "step": 9435 }, { "epoch": 23.59, "grad_norm": 6.71875, "learning_rate": 1.7857678684352684e-05, "loss": 1.4391, "step": 9436 }, { "epoch": 23.5925, "grad_norm": 7.375, "learning_rate": 1.7857177848640158e-05, "loss": 1.467, "step": 9437 }, { "epoch": 23.595, "grad_norm": 5.625, "learning_rate": 1.7856676961416286e-05, "loss": 1.3535, "step": 9438 }, { "epoch": 23.5975, "grad_norm": 9.0, "learning_rate": 1.7856176022684353e-05, "loss": 1.3893, "step": 9439 }, { "epoch": 23.6, "grad_norm": 5.59375, "learning_rate": 1.7855675032447648e-05, "loss": 1.3588, "step": 9440 }, { "epoch": 23.6025, "grad_norm": 5.8125, "learning_rate": 1.7855173990709448e-05, "loss": 1.5033, "step": 9441 }, { "epoch": 23.605, "grad_norm": 5.3125, "learning_rate": 1.7854672897473048e-05, "loss": 1.2536, "step": 9442 }, { "epoch": 23.6075, "grad_norm": 7.71875, "learning_rate": 1.785417175274172e-05, "loss": 1.5506, "step": 9443 }, { "epoch": 23.61, "grad_norm": 5.5625, "learning_rate": 1.7853670556518762e-05, "loss": 1.5088, "step": 9444 }, { "epoch": 23.6125, "grad_norm": 7.0, "learning_rate": 1.785316930880745e-05, "loss": 1.4108, "step": 9445 }, { "epoch": 23.615, "grad_norm": 6.0625, "learning_rate": 1.7852668009611075e-05, "loss": 1.3808, "step": 9446 }, { "epoch": 23.6175, "grad_norm": 7.0625, "learning_rate": 1.7852166658932922e-05, "loss": 1.3617, "step": 9447 }, { "epoch": 23.62, "grad_norm": 7.21875, "learning_rate": 1.7851665256776284e-05, "loss": 1.5002, "step": 9448 }, { "epoch": 23.6225, "grad_norm": 9.625, "learning_rate": 1.785116380314444e-05, "loss": 1.4077, "step": 9449 }, { "epoch": 23.625, "grad_norm": 7.03125, "learning_rate": 1.7850662298040676e-05, "loss": 1.2677, "step": 9450 }, { "epoch": 23.6275, "grad_norm": 6.9375, "learning_rate": 1.785016074146829e-05, "loss": 1.3981, "step": 9451 }, { "epoch": 23.63, "grad_norm": 5.28125, "learning_rate": 1.784965913343056e-05, "loss": 1.3328, "step": 9452 }, { "epoch": 23.6325, "grad_norm": 9.125, "learning_rate": 1.784915747393078e-05, "loss": 1.3013, "step": 9453 }, { "epoch": 23.635, "grad_norm": 9.8125, "learning_rate": 1.784865576297224e-05, "loss": 1.4627, "step": 9454 }, { "epoch": 23.6375, "grad_norm": 7.15625, "learning_rate": 1.7848154000558226e-05, "loss": 1.3778, "step": 9455 }, { "epoch": 23.64, "grad_norm": 5.28125, "learning_rate": 1.7847652186692025e-05, "loss": 1.309, "step": 9456 }, { "epoch": 23.6425, "grad_norm": 7.1875, "learning_rate": 1.7847150321376935e-05, "loss": 1.4529, "step": 9457 }, { "epoch": 23.645, "grad_norm": 6.28125, "learning_rate": 1.784664840461624e-05, "loss": 1.3839, "step": 9458 }, { "epoch": 23.6475, "grad_norm": 5.9375, "learning_rate": 1.784614643641323e-05, "loss": 1.2925, "step": 9459 }, { "epoch": 23.65, "grad_norm": 5.75, "learning_rate": 1.78456444167712e-05, "loss": 1.3535, "step": 9460 }, { "epoch": 23.6525, "grad_norm": 5.25, "learning_rate": 1.784514234569344e-05, "loss": 1.5022, "step": 9461 }, { "epoch": 23.655, "grad_norm": 6.4375, "learning_rate": 1.7844640223183235e-05, "loss": 1.5097, "step": 9462 }, { "epoch": 23.6575, "grad_norm": 8.6875, "learning_rate": 1.7844138049243888e-05, "loss": 1.3834, "step": 9463 }, { "epoch": 23.66, "grad_norm": 9.8125, "learning_rate": 1.7843635823878687e-05, "loss": 1.3608, "step": 9464 }, { "epoch": 23.6625, "grad_norm": 4.78125, "learning_rate": 1.784313354709092e-05, "loss": 1.4286, "step": 9465 }, { "epoch": 23.665, "grad_norm": 8.0625, "learning_rate": 1.7842631218883887e-05, "loss": 1.4459, "step": 9466 }, { "epoch": 23.6675, "grad_norm": 5.78125, "learning_rate": 1.7842128839260874e-05, "loss": 1.3337, "step": 9467 }, { "epoch": 23.67, "grad_norm": 7.375, "learning_rate": 1.7841626408225177e-05, "loss": 1.5757, "step": 9468 }, { "epoch": 23.6725, "grad_norm": 5.625, "learning_rate": 1.7841123925780093e-05, "loss": 1.5291, "step": 9469 }, { "epoch": 23.675, "grad_norm": 6.0625, "learning_rate": 1.7840621391928915e-05, "loss": 1.4077, "step": 9470 }, { "epoch": 23.6775, "grad_norm": 7.53125, "learning_rate": 1.784011880667494e-05, "loss": 1.1986, "step": 9471 }, { "epoch": 23.68, "grad_norm": 6.1875, "learning_rate": 1.7839616170021452e-05, "loss": 1.2656, "step": 9472 }, { "epoch": 23.6825, "grad_norm": 6.40625, "learning_rate": 1.7839113481971758e-05, "loss": 1.421, "step": 9473 }, { "epoch": 23.685, "grad_norm": 6.65625, "learning_rate": 1.7838610742529152e-05, "loss": 1.374, "step": 9474 }, { "epoch": 23.6875, "grad_norm": 6.5625, "learning_rate": 1.7838107951696926e-05, "loss": 1.5463, "step": 9475 }, { "epoch": 23.69, "grad_norm": 6.25, "learning_rate": 1.783760510947838e-05, "loss": 1.2441, "step": 9476 }, { "epoch": 23.6925, "grad_norm": 8.0625, "learning_rate": 1.7837102215876804e-05, "loss": 1.3144, "step": 9477 }, { "epoch": 23.695, "grad_norm": 5.78125, "learning_rate": 1.7836599270895504e-05, "loss": 1.3184, "step": 9478 }, { "epoch": 23.6975, "grad_norm": 5.625, "learning_rate": 1.783609627453777e-05, "loss": 1.2642, "step": 9479 }, { "epoch": 23.7, "grad_norm": 7.78125, "learning_rate": 1.7835593226806902e-05, "loss": 1.5931, "step": 9480 }, { "epoch": 23.7025, "grad_norm": 7.65625, "learning_rate": 1.78350901277062e-05, "loss": 1.4291, "step": 9481 }, { "epoch": 23.705, "grad_norm": 6.90625, "learning_rate": 1.783458697723896e-05, "loss": 1.3767, "step": 9482 }, { "epoch": 23.7075, "grad_norm": 8.9375, "learning_rate": 1.783408377540848e-05, "loss": 1.3798, "step": 9483 }, { "epoch": 23.71, "grad_norm": 6.4375, "learning_rate": 1.7833580522218062e-05, "loss": 1.3674, "step": 9484 }, { "epoch": 23.7125, "grad_norm": 9.5, "learning_rate": 1.7833077217671005e-05, "loss": 1.445, "step": 9485 }, { "epoch": 23.715, "grad_norm": 5.9375, "learning_rate": 1.7832573861770604e-05, "loss": 1.3295, "step": 9486 }, { "epoch": 23.7175, "grad_norm": 8.625, "learning_rate": 1.7832070454520166e-05, "loss": 1.4164, "step": 9487 }, { "epoch": 23.72, "grad_norm": 7.78125, "learning_rate": 1.7831566995922983e-05, "loss": 1.227, "step": 9488 }, { "epoch": 23.7225, "grad_norm": 6.28125, "learning_rate": 1.7831063485982363e-05, "loss": 1.414, "step": 9489 }, { "epoch": 23.725, "grad_norm": 7.46875, "learning_rate": 1.7830559924701606e-05, "loss": 1.3699, "step": 9490 }, { "epoch": 23.7275, "grad_norm": 5.15625, "learning_rate": 1.7830056312084004e-05, "loss": 1.2416, "step": 9491 }, { "epoch": 23.73, "grad_norm": 9.3125, "learning_rate": 1.7829552648132872e-05, "loss": 1.3481, "step": 9492 }, { "epoch": 23.7325, "grad_norm": 7.5625, "learning_rate": 1.7829048932851506e-05, "loss": 1.4516, "step": 9493 }, { "epoch": 23.735, "grad_norm": 5.75, "learning_rate": 1.7828545166243206e-05, "loss": 1.5437, "step": 9494 }, { "epoch": 23.7375, "grad_norm": 8.125, "learning_rate": 1.7828041348311283e-05, "loss": 1.413, "step": 9495 }, { "epoch": 23.74, "grad_norm": 7.5, "learning_rate": 1.7827537479059026e-05, "loss": 1.5666, "step": 9496 }, { "epoch": 23.7425, "grad_norm": 6.28125, "learning_rate": 1.7827033558489755e-05, "loss": 1.5698, "step": 9497 }, { "epoch": 23.745, "grad_norm": 8.4375, "learning_rate": 1.782652958660676e-05, "loss": 1.3081, "step": 9498 }, { "epoch": 23.7475, "grad_norm": 5.09375, "learning_rate": 1.782602556341335e-05, "loss": 1.3453, "step": 9499 }, { "epoch": 23.75, "grad_norm": 6.3125, "learning_rate": 1.7825521488912833e-05, "loss": 1.4519, "step": 9500 }, { "epoch": 23.7525, "grad_norm": 5.46875, "learning_rate": 1.7825017363108506e-05, "loss": 1.3331, "step": 9501 }, { "epoch": 23.755, "grad_norm": 9.6875, "learning_rate": 1.782451318600368e-05, "loss": 1.3033, "step": 9502 }, { "epoch": 23.7575, "grad_norm": 5.90625, "learning_rate": 1.782400895760166e-05, "loss": 1.269, "step": 9503 }, { "epoch": 23.76, "grad_norm": 28.625, "learning_rate": 1.782350467790575e-05, "loss": 1.5021, "step": 9504 }, { "epoch": 23.7625, "grad_norm": 9.8125, "learning_rate": 1.7823000346919256e-05, "loss": 1.5348, "step": 9505 }, { "epoch": 23.765, "grad_norm": 6.0625, "learning_rate": 1.782249596464548e-05, "loss": 1.4281, "step": 9506 }, { "epoch": 23.7675, "grad_norm": 6.71875, "learning_rate": 1.7821991531087742e-05, "loss": 1.4743, "step": 9507 }, { "epoch": 23.77, "grad_norm": 7.40625, "learning_rate": 1.7821487046249334e-05, "loss": 1.4163, "step": 9508 }, { "epoch": 23.7725, "grad_norm": 6.9375, "learning_rate": 1.7820982510133573e-05, "loss": 1.3362, "step": 9509 }, { "epoch": 23.775, "grad_norm": 7.0, "learning_rate": 1.7820477922743764e-05, "loss": 1.3592, "step": 9510 }, { "epoch": 23.7775, "grad_norm": 5.875, "learning_rate": 1.7819973284083214e-05, "loss": 1.2888, "step": 9511 }, { "epoch": 23.78, "grad_norm": 6.0, "learning_rate": 1.7819468594155235e-05, "loss": 1.38, "step": 9512 }, { "epoch": 23.7825, "grad_norm": 7.375, "learning_rate": 1.781896385296313e-05, "loss": 1.5289, "step": 9513 }, { "epoch": 23.785, "grad_norm": 8.125, "learning_rate": 1.7818459060510208e-05, "loss": 1.3631, "step": 9514 }, { "epoch": 23.7875, "grad_norm": 7.59375, "learning_rate": 1.7817954216799786e-05, "loss": 1.5484, "step": 9515 }, { "epoch": 23.79, "grad_norm": 7.71875, "learning_rate": 1.7817449321835166e-05, "loss": 1.3568, "step": 9516 }, { "epoch": 23.7925, "grad_norm": 4.9375, "learning_rate": 1.781694437561966e-05, "loss": 1.296, "step": 9517 }, { "epoch": 23.795, "grad_norm": 7.8125, "learning_rate": 1.7816439378156583e-05, "loss": 1.5859, "step": 9518 }, { "epoch": 23.7975, "grad_norm": 7.28125, "learning_rate": 1.781593432944924e-05, "loss": 1.3118, "step": 9519 }, { "epoch": 23.8, "grad_norm": 10.5, "learning_rate": 1.7815429229500946e-05, "loss": 1.5257, "step": 9520 }, { "epoch": 23.8025, "grad_norm": 8.3125, "learning_rate": 1.7814924078315006e-05, "loss": 1.3422, "step": 9521 }, { "epoch": 23.805, "grad_norm": 7.0625, "learning_rate": 1.7814418875894744e-05, "loss": 1.2977, "step": 9522 }, { "epoch": 23.8075, "grad_norm": 7.46875, "learning_rate": 1.781391362224346e-05, "loss": 1.4016, "step": 9523 }, { "epoch": 23.81, "grad_norm": 7.8125, "learning_rate": 1.781340831736447e-05, "loss": 1.3985, "step": 9524 }, { "epoch": 23.8125, "grad_norm": 8.9375, "learning_rate": 1.7812902961261092e-05, "loss": 1.4854, "step": 9525 }, { "epoch": 23.815, "grad_norm": 10.75, "learning_rate": 1.781239755393663e-05, "loss": 1.3858, "step": 9526 }, { "epoch": 23.8175, "grad_norm": 6.28125, "learning_rate": 1.7811892095394405e-05, "loss": 1.4947, "step": 9527 }, { "epoch": 23.82, "grad_norm": 6.65625, "learning_rate": 1.7811386585637728e-05, "loss": 1.3724, "step": 9528 }, { "epoch": 23.8225, "grad_norm": 9.0625, "learning_rate": 1.7810881024669916e-05, "loss": 1.4077, "step": 9529 }, { "epoch": 23.825, "grad_norm": 9.0625, "learning_rate": 1.7810375412494278e-05, "loss": 1.5824, "step": 9530 }, { "epoch": 23.8275, "grad_norm": 8.5625, "learning_rate": 1.7809869749114128e-05, "loss": 1.6273, "step": 9531 }, { "epoch": 23.83, "grad_norm": 5.9375, "learning_rate": 1.780936403453279e-05, "loss": 1.3461, "step": 9532 }, { "epoch": 23.8325, "grad_norm": 5.75, "learning_rate": 1.780885826875357e-05, "loss": 1.4659, "step": 9533 }, { "epoch": 23.835, "grad_norm": 8.0625, "learning_rate": 1.7808352451779787e-05, "loss": 1.3577, "step": 9534 }, { "epoch": 23.8375, "grad_norm": 9.5625, "learning_rate": 1.7807846583614762e-05, "loss": 1.3841, "step": 9535 }, { "epoch": 23.84, "grad_norm": 6.875, "learning_rate": 1.78073406642618e-05, "loss": 1.4106, "step": 9536 }, { "epoch": 23.8425, "grad_norm": 8.5, "learning_rate": 1.7806834693724233e-05, "loss": 1.4702, "step": 9537 }, { "epoch": 23.845, "grad_norm": 5.46875, "learning_rate": 1.7806328672005365e-05, "loss": 1.3428, "step": 9538 }, { "epoch": 23.8475, "grad_norm": 8.5625, "learning_rate": 1.7805822599108523e-05, "loss": 1.3253, "step": 9539 }, { "epoch": 23.85, "grad_norm": 6.25, "learning_rate": 1.7805316475037016e-05, "loss": 1.2697, "step": 9540 }, { "epoch": 23.8525, "grad_norm": 5.0625, "learning_rate": 1.780481029979417e-05, "loss": 1.5709, "step": 9541 }, { "epoch": 23.855, "grad_norm": 5.8125, "learning_rate": 1.7804304073383298e-05, "loss": 1.3951, "step": 9542 }, { "epoch": 23.8575, "grad_norm": 8.4375, "learning_rate": 1.780379779580772e-05, "loss": 1.5259, "step": 9543 }, { "epoch": 23.86, "grad_norm": 5.875, "learning_rate": 1.780329146707076e-05, "loss": 1.2918, "step": 9544 }, { "epoch": 23.8625, "grad_norm": 5.71875, "learning_rate": 1.780278508717573e-05, "loss": 1.3407, "step": 9545 }, { "epoch": 23.865, "grad_norm": 6.78125, "learning_rate": 1.7802278656125955e-05, "loss": 1.6022, "step": 9546 }, { "epoch": 23.8675, "grad_norm": 12.375, "learning_rate": 1.780177217392475e-05, "loss": 1.3867, "step": 9547 }, { "epoch": 23.87, "grad_norm": 6.4375, "learning_rate": 1.7801265640575442e-05, "loss": 1.3738, "step": 9548 }, { "epoch": 23.8725, "grad_norm": 8.5625, "learning_rate": 1.780075905608135e-05, "loss": 1.2618, "step": 9549 }, { "epoch": 23.875, "grad_norm": 8.0625, "learning_rate": 1.7800252420445788e-05, "loss": 1.5348, "step": 9550 }, { "epoch": 23.8775, "grad_norm": 6.75, "learning_rate": 1.779974573367209e-05, "loss": 1.2695, "step": 9551 }, { "epoch": 23.88, "grad_norm": 7.34375, "learning_rate": 1.779923899576357e-05, "loss": 1.2326, "step": 9552 }, { "epoch": 23.8825, "grad_norm": 6.78125, "learning_rate": 1.7798732206723545e-05, "loss": 1.3258, "step": 9553 }, { "epoch": 23.885, "grad_norm": 5.15625, "learning_rate": 1.779822536655535e-05, "loss": 1.422, "step": 9554 }, { "epoch": 23.8875, "grad_norm": 6.3125, "learning_rate": 1.77977184752623e-05, "loss": 1.3656, "step": 9555 }, { "epoch": 23.89, "grad_norm": 13.3125, "learning_rate": 1.779721153284772e-05, "loss": 1.4944, "step": 9556 }, { "epoch": 23.8925, "grad_norm": 5.71875, "learning_rate": 1.7796704539314933e-05, "loss": 1.3319, "step": 9557 }, { "epoch": 23.895, "grad_norm": 7.0, "learning_rate": 1.779619749466726e-05, "loss": 1.4794, "step": 9558 }, { "epoch": 23.8975, "grad_norm": 7.90625, "learning_rate": 1.779569039890803e-05, "loss": 1.217, "step": 9559 }, { "epoch": 23.9, "grad_norm": 6.9375, "learning_rate": 1.7795183252040568e-05, "loss": 1.386, "step": 9560 }, { "epoch": 23.9025, "grad_norm": 9.3125, "learning_rate": 1.7794676054068195e-05, "loss": 1.5413, "step": 9561 }, { "epoch": 23.905, "grad_norm": 6.34375, "learning_rate": 1.7794168804994237e-05, "loss": 1.3573, "step": 9562 }, { "epoch": 23.9075, "grad_norm": 7.875, "learning_rate": 1.7793661504822023e-05, "loss": 1.3916, "step": 9563 }, { "epoch": 23.91, "grad_norm": 14.6875, "learning_rate": 1.7793154153554875e-05, "loss": 1.5788, "step": 9564 }, { "epoch": 23.9125, "grad_norm": 7.0, "learning_rate": 1.7792646751196118e-05, "loss": 1.4335, "step": 9565 }, { "epoch": 23.915, "grad_norm": 11.375, "learning_rate": 1.779213929774908e-05, "loss": 1.4339, "step": 9566 }, { "epoch": 23.9175, "grad_norm": 6.96875, "learning_rate": 1.779163179321709e-05, "loss": 1.4198, "step": 9567 }, { "epoch": 23.92, "grad_norm": 4.40625, "learning_rate": 1.7791124237603477e-05, "loss": 1.2634, "step": 9568 }, { "epoch": 23.9225, "grad_norm": 7.9375, "learning_rate": 1.779061663091156e-05, "loss": 1.3613, "step": 9569 }, { "epoch": 23.925, "grad_norm": 6.0625, "learning_rate": 1.7790108973144677e-05, "loss": 1.3484, "step": 9570 }, { "epoch": 23.9275, "grad_norm": 6.34375, "learning_rate": 1.7789601264306148e-05, "loss": 1.315, "step": 9571 }, { "epoch": 23.93, "grad_norm": 4.5, "learning_rate": 1.7789093504399307e-05, "loss": 1.5749, "step": 9572 }, { "epoch": 23.9325, "grad_norm": 7.0, "learning_rate": 1.778858569342748e-05, "loss": 1.3654, "step": 9573 }, { "epoch": 23.935, "grad_norm": 10.0, "learning_rate": 1.7788077831394e-05, "loss": 1.3666, "step": 9574 }, { "epoch": 23.9375, "grad_norm": 9.875, "learning_rate": 1.7787569918302185e-05, "loss": 1.4781, "step": 9575 }, { "epoch": 23.94, "grad_norm": 8.6875, "learning_rate": 1.778706195415538e-05, "loss": 1.4353, "step": 9576 }, { "epoch": 23.9425, "grad_norm": 8.125, "learning_rate": 1.7786553938956905e-05, "loss": 1.3702, "step": 9577 }, { "epoch": 23.945, "grad_norm": 6.71875, "learning_rate": 1.77860458727101e-05, "loss": 1.3524, "step": 9578 }, { "epoch": 23.9475, "grad_norm": 7.71875, "learning_rate": 1.7785537755418283e-05, "loss": 1.4235, "step": 9579 }, { "epoch": 23.95, "grad_norm": 9.25, "learning_rate": 1.7785029587084793e-05, "loss": 1.4374, "step": 9580 }, { "epoch": 23.9525, "grad_norm": 5.75, "learning_rate": 1.7784521367712964e-05, "loss": 1.1944, "step": 9581 }, { "epoch": 23.955, "grad_norm": 10.625, "learning_rate": 1.7784013097306122e-05, "loss": 1.4837, "step": 9582 }, { "epoch": 23.9575, "grad_norm": 5.0625, "learning_rate": 1.77835047758676e-05, "loss": 1.4442, "step": 9583 }, { "epoch": 23.96, "grad_norm": 8.8125, "learning_rate": 1.7782996403400737e-05, "loss": 1.5247, "step": 9584 }, { "epoch": 23.9625, "grad_norm": 8.1875, "learning_rate": 1.7782487979908855e-05, "loss": 1.4238, "step": 9585 }, { "epoch": 23.965, "grad_norm": 7.3125, "learning_rate": 1.77819795053953e-05, "loss": 1.3547, "step": 9586 }, { "epoch": 23.9675, "grad_norm": 9.5, "learning_rate": 1.7781470979863394e-05, "loss": 1.3992, "step": 9587 }, { "epoch": 23.97, "grad_norm": 5.5, "learning_rate": 1.7780962403316477e-05, "loss": 1.3726, "step": 9588 }, { "epoch": 23.9725, "grad_norm": 5.78125, "learning_rate": 1.778045377575788e-05, "loss": 1.3707, "step": 9589 }, { "epoch": 23.975, "grad_norm": 9.25, "learning_rate": 1.7779945097190944e-05, "loss": 1.6758, "step": 9590 }, { "epoch": 23.9775, "grad_norm": 5.90625, "learning_rate": 1.7779436367618994e-05, "loss": 1.4154, "step": 9591 }, { "epoch": 23.98, "grad_norm": 6.375, "learning_rate": 1.7778927587045373e-05, "loss": 1.3527, "step": 9592 }, { "epoch": 23.9825, "grad_norm": 6.6875, "learning_rate": 1.7778418755473414e-05, "loss": 1.403, "step": 9593 }, { "epoch": 23.985, "grad_norm": 8.75, "learning_rate": 1.7777909872906452e-05, "loss": 1.5058, "step": 9594 }, { "epoch": 23.9875, "grad_norm": 9.3125, "learning_rate": 1.7777400939347826e-05, "loss": 1.444, "step": 9595 }, { "epoch": 23.99, "grad_norm": 7.15625, "learning_rate": 1.777689195480087e-05, "loss": 1.4025, "step": 9596 }, { "epoch": 23.9925, "grad_norm": 11.8125, "learning_rate": 1.7776382919268922e-05, "loss": 1.4664, "step": 9597 }, { "epoch": 23.995, "grad_norm": 7.25, "learning_rate": 1.777587383275532e-05, "loss": 1.362, "step": 9598 }, { "epoch": 23.9975, "grad_norm": 7.3125, "learning_rate": 1.7775364695263397e-05, "loss": 1.3927, "step": 9599 }, { "epoch": 24.0, "grad_norm": 7.09375, "learning_rate": 1.7774855506796497e-05, "loss": 1.613, "step": 9600 }, { "epoch": 24.0025, "grad_norm": 8.0625, "learning_rate": 1.777434626735795e-05, "loss": 1.4324, "step": 9601 }, { "epoch": 24.005, "grad_norm": 6.4375, "learning_rate": 1.7773836976951106e-05, "loss": 1.2278, "step": 9602 }, { "epoch": 24.0075, "grad_norm": 5.3125, "learning_rate": 1.7773327635579295e-05, "loss": 1.3258, "step": 9603 }, { "epoch": 24.01, "grad_norm": 7.0, "learning_rate": 1.777281824324586e-05, "loss": 1.3369, "step": 9604 }, { "epoch": 24.0125, "grad_norm": 7.75, "learning_rate": 1.777230879995414e-05, "loss": 1.3076, "step": 9605 }, { "epoch": 24.015, "grad_norm": 6.78125, "learning_rate": 1.7771799305707474e-05, "loss": 1.4597, "step": 9606 }, { "epoch": 24.0175, "grad_norm": 6.03125, "learning_rate": 1.7771289760509206e-05, "loss": 1.4401, "step": 9607 }, { "epoch": 24.02, "grad_norm": 7.09375, "learning_rate": 1.7770780164362665e-05, "loss": 1.2831, "step": 9608 }, { "epoch": 24.0225, "grad_norm": 7.3125, "learning_rate": 1.777027051727121e-05, "loss": 1.301, "step": 9609 }, { "epoch": 24.025, "grad_norm": 9.375, "learning_rate": 1.7769760819238165e-05, "loss": 1.4567, "step": 9610 }, { "epoch": 24.0275, "grad_norm": 5.625, "learning_rate": 1.7769251070266884e-05, "loss": 1.3765, "step": 9611 }, { "epoch": 24.03, "grad_norm": 6.96875, "learning_rate": 1.77687412703607e-05, "loss": 1.4406, "step": 9612 }, { "epoch": 24.0325, "grad_norm": 7.9375, "learning_rate": 1.7768231419522963e-05, "loss": 1.4744, "step": 9613 }, { "epoch": 24.035, "grad_norm": 5.4375, "learning_rate": 1.7767721517757007e-05, "loss": 1.37, "step": 9614 }, { "epoch": 24.0375, "grad_norm": 6.03125, "learning_rate": 1.7767211565066184e-05, "loss": 1.3619, "step": 9615 }, { "epoch": 24.04, "grad_norm": 7.34375, "learning_rate": 1.776670156145383e-05, "loss": 1.3069, "step": 9616 }, { "epoch": 24.0425, "grad_norm": 8.9375, "learning_rate": 1.776619150692329e-05, "loss": 1.4484, "step": 9617 }, { "epoch": 24.045, "grad_norm": 6.125, "learning_rate": 1.7765681401477916e-05, "loss": 1.275, "step": 9618 }, { "epoch": 24.0475, "grad_norm": 7.0625, "learning_rate": 1.776517124512104e-05, "loss": 1.2354, "step": 9619 }, { "epoch": 24.05, "grad_norm": 7.0625, "learning_rate": 1.7764661037856013e-05, "loss": 1.4316, "step": 9620 }, { "epoch": 24.0525, "grad_norm": 4.375, "learning_rate": 1.776415077968618e-05, "loss": 1.4086, "step": 9621 }, { "epoch": 24.055, "grad_norm": 9.0625, "learning_rate": 1.7763640470614883e-05, "loss": 1.4125, "step": 9622 }, { "epoch": 24.0575, "grad_norm": 7.125, "learning_rate": 1.7763130110645473e-05, "loss": 1.4833, "step": 9623 }, { "epoch": 24.06, "grad_norm": 6.3125, "learning_rate": 1.776261969978129e-05, "loss": 1.4821, "step": 9624 }, { "epoch": 24.0625, "grad_norm": 5.40625, "learning_rate": 1.7762109238025682e-05, "loss": 1.288, "step": 9625 }, { "epoch": 24.065, "grad_norm": 7.6875, "learning_rate": 1.7761598725381998e-05, "loss": 1.3565, "step": 9626 }, { "epoch": 24.0675, "grad_norm": 7.0625, "learning_rate": 1.7761088161853582e-05, "loss": 1.2572, "step": 9627 }, { "epoch": 24.07, "grad_norm": 7.6875, "learning_rate": 1.7760577547443785e-05, "loss": 1.3026, "step": 9628 }, { "epoch": 24.0725, "grad_norm": 8.5625, "learning_rate": 1.776006688215595e-05, "loss": 1.5685, "step": 9629 }, { "epoch": 24.075, "grad_norm": 9.0625, "learning_rate": 1.775955616599343e-05, "loss": 1.3357, "step": 9630 }, { "epoch": 24.0775, "grad_norm": 8.1875, "learning_rate": 1.775904539895957e-05, "loss": 1.3577, "step": 9631 }, { "epoch": 24.08, "grad_norm": 6.6875, "learning_rate": 1.775853458105772e-05, "loss": 1.5091, "step": 9632 }, { "epoch": 24.0825, "grad_norm": 5.53125, "learning_rate": 1.7758023712291223e-05, "loss": 1.3602, "step": 9633 }, { "epoch": 24.085, "grad_norm": 6.15625, "learning_rate": 1.7757512792663434e-05, "loss": 1.3432, "step": 9634 }, { "epoch": 24.0875, "grad_norm": 6.40625, "learning_rate": 1.7757001822177703e-05, "loss": 1.4467, "step": 9635 }, { "epoch": 24.09, "grad_norm": 8.9375, "learning_rate": 1.7756490800837377e-05, "loss": 1.4107, "step": 9636 }, { "epoch": 24.0925, "grad_norm": 5.0625, "learning_rate": 1.775597972864581e-05, "loss": 1.1258, "step": 9637 }, { "epoch": 24.095, "grad_norm": 4.28125, "learning_rate": 1.7755468605606347e-05, "loss": 1.4127, "step": 9638 }, { "epoch": 24.0975, "grad_norm": 7.40625, "learning_rate": 1.7754957431722347e-05, "loss": 1.2765, "step": 9639 }, { "epoch": 24.1, "grad_norm": 5.96875, "learning_rate": 1.7754446206997152e-05, "loss": 1.3819, "step": 9640 }, { "epoch": 24.1025, "grad_norm": 14.3125, "learning_rate": 1.775393493143412e-05, "loss": 1.2545, "step": 9641 }, { "epoch": 24.105, "grad_norm": 6.03125, "learning_rate": 1.77534236050366e-05, "loss": 1.439, "step": 9642 }, { "epoch": 24.1075, "grad_norm": 6.59375, "learning_rate": 1.775291222780795e-05, "loss": 1.2607, "step": 9643 }, { "epoch": 24.11, "grad_norm": 7.1875, "learning_rate": 1.7752400799751514e-05, "loss": 1.4179, "step": 9644 }, { "epoch": 24.1125, "grad_norm": 7.125, "learning_rate": 1.7751889320870644e-05, "loss": 1.4284, "step": 9645 }, { "epoch": 24.115, "grad_norm": 5.875, "learning_rate": 1.7751377791168703e-05, "loss": 1.3143, "step": 9646 }, { "epoch": 24.1175, "grad_norm": 9.0625, "learning_rate": 1.7750866210649037e-05, "loss": 1.3186, "step": 9647 }, { "epoch": 24.12, "grad_norm": 13.1875, "learning_rate": 1.7750354579315004e-05, "loss": 1.2756, "step": 9648 }, { "epoch": 24.1225, "grad_norm": 8.0625, "learning_rate": 1.7749842897169958e-05, "loss": 1.4933, "step": 9649 }, { "epoch": 24.125, "grad_norm": 6.71875, "learning_rate": 1.774933116421725e-05, "loss": 1.3403, "step": 9650 }, { "epoch": 24.1275, "grad_norm": 8.375, "learning_rate": 1.7748819380460238e-05, "loss": 1.3526, "step": 9651 }, { "epoch": 24.13, "grad_norm": 9.9375, "learning_rate": 1.7748307545902275e-05, "loss": 1.4172, "step": 9652 }, { "epoch": 24.1325, "grad_norm": 6.90625, "learning_rate": 1.7747795660546717e-05, "loss": 1.2655, "step": 9653 }, { "epoch": 24.135, "grad_norm": 6.625, "learning_rate": 1.774728372439692e-05, "loss": 1.5254, "step": 9654 }, { "epoch": 24.1375, "grad_norm": 6.90625, "learning_rate": 1.7746771737456244e-05, "loss": 1.2924, "step": 9655 }, { "epoch": 24.14, "grad_norm": 9.0625, "learning_rate": 1.774625969972804e-05, "loss": 1.5253, "step": 9656 }, { "epoch": 24.1425, "grad_norm": 10.5, "learning_rate": 1.7745747611215672e-05, "loss": 1.4403, "step": 9657 }, { "epoch": 24.145, "grad_norm": 8.0625, "learning_rate": 1.774523547192249e-05, "loss": 1.5011, "step": 9658 }, { "epoch": 24.1475, "grad_norm": 8.5, "learning_rate": 1.7744723281851852e-05, "loss": 1.3142, "step": 9659 }, { "epoch": 24.15, "grad_norm": 7.4375, "learning_rate": 1.774421104100712e-05, "loss": 1.4095, "step": 9660 }, { "epoch": 24.1525, "grad_norm": 5.4375, "learning_rate": 1.774369874939165e-05, "loss": 1.3408, "step": 9661 }, { "epoch": 24.155, "grad_norm": 5.9375, "learning_rate": 1.7743186407008803e-05, "loss": 1.3516, "step": 9662 }, { "epoch": 24.1575, "grad_norm": 8.0, "learning_rate": 1.7742674013861932e-05, "loss": 1.3748, "step": 9663 }, { "epoch": 24.16, "grad_norm": 7.6875, "learning_rate": 1.77421615699544e-05, "loss": 1.4944, "step": 9664 }, { "epoch": 24.1625, "grad_norm": 6.96875, "learning_rate": 1.7741649075289567e-05, "loss": 1.4743, "step": 9665 }, { "epoch": 24.165, "grad_norm": 9.125, "learning_rate": 1.7741136529870795e-05, "loss": 1.4243, "step": 9666 }, { "epoch": 24.1675, "grad_norm": 10.25, "learning_rate": 1.774062393370144e-05, "loss": 1.4338, "step": 9667 }, { "epoch": 24.17, "grad_norm": 6.8125, "learning_rate": 1.7740111286784864e-05, "loss": 1.2406, "step": 9668 }, { "epoch": 24.1725, "grad_norm": 6.9375, "learning_rate": 1.7739598589124432e-05, "loss": 1.3994, "step": 9669 }, { "epoch": 24.175, "grad_norm": 6.6875, "learning_rate": 1.7739085840723497e-05, "loss": 1.2411, "step": 9670 }, { "epoch": 24.1775, "grad_norm": 6.5625, "learning_rate": 1.7738573041585424e-05, "loss": 1.4604, "step": 9671 }, { "epoch": 24.18, "grad_norm": 5.8125, "learning_rate": 1.773806019171358e-05, "loss": 1.3975, "step": 9672 }, { "epoch": 24.1825, "grad_norm": 9.3125, "learning_rate": 1.773754729111132e-05, "loss": 1.4465, "step": 9673 }, { "epoch": 24.185, "grad_norm": 6.125, "learning_rate": 1.773703433978201e-05, "loss": 1.4872, "step": 9674 }, { "epoch": 24.1875, "grad_norm": 5.84375, "learning_rate": 1.7736521337729013e-05, "loss": 1.2901, "step": 9675 }, { "epoch": 24.19, "grad_norm": 6.34375, "learning_rate": 1.7736008284955693e-05, "loss": 1.4561, "step": 9676 }, { "epoch": 24.1925, "grad_norm": 6.1875, "learning_rate": 1.773549518146541e-05, "loss": 1.4194, "step": 9677 }, { "epoch": 24.195, "grad_norm": 7.0625, "learning_rate": 1.7734982027261535e-05, "loss": 1.5604, "step": 9678 }, { "epoch": 24.1975, "grad_norm": 12.0, "learning_rate": 1.7734468822347422e-05, "loss": 1.4742, "step": 9679 }, { "epoch": 24.2, "grad_norm": 8.875, "learning_rate": 1.7733955566726438e-05, "loss": 1.2807, "step": 9680 }, { "epoch": 24.2025, "grad_norm": 6.5625, "learning_rate": 1.7733442260401957e-05, "loss": 1.4664, "step": 9681 }, { "epoch": 24.205, "grad_norm": 6.8125, "learning_rate": 1.7732928903377336e-05, "loss": 1.4582, "step": 9682 }, { "epoch": 24.2075, "grad_norm": 8.8125, "learning_rate": 1.7732415495655946e-05, "loss": 1.3338, "step": 9683 }, { "epoch": 24.21, "grad_norm": 8.875, "learning_rate": 1.7731902037241143e-05, "loss": 1.5011, "step": 9684 }, { "epoch": 24.2125, "grad_norm": 5.71875, "learning_rate": 1.7731388528136303e-05, "loss": 1.4291, "step": 9685 }, { "epoch": 24.215, "grad_norm": 7.15625, "learning_rate": 1.7730874968344788e-05, "loss": 1.4174, "step": 9686 }, { "epoch": 24.2175, "grad_norm": 7.78125, "learning_rate": 1.773036135786997e-05, "loss": 1.4357, "step": 9687 }, { "epoch": 24.22, "grad_norm": 5.875, "learning_rate": 1.772984769671521e-05, "loss": 1.3136, "step": 9688 }, { "epoch": 24.2225, "grad_norm": 7.625, "learning_rate": 1.7729333984883877e-05, "loss": 1.494, "step": 9689 }, { "epoch": 24.225, "grad_norm": 5.21875, "learning_rate": 1.772882022237934e-05, "loss": 1.2866, "step": 9690 }, { "epoch": 24.2275, "grad_norm": 7.15625, "learning_rate": 1.772830640920497e-05, "loss": 1.2598, "step": 9691 }, { "epoch": 24.23, "grad_norm": 5.96875, "learning_rate": 1.7727792545364123e-05, "loss": 1.3097, "step": 9692 }, { "epoch": 24.2325, "grad_norm": 10.0625, "learning_rate": 1.7727278630860187e-05, "loss": 1.5479, "step": 9693 }, { "epoch": 24.235, "grad_norm": 5.3125, "learning_rate": 1.7726764665696518e-05, "loss": 1.3303, "step": 9694 }, { "epoch": 24.2375, "grad_norm": 8.8125, "learning_rate": 1.7726250649876488e-05, "loss": 1.4393, "step": 9695 }, { "epoch": 24.24, "grad_norm": 9.9375, "learning_rate": 1.772573658340347e-05, "loss": 1.4703, "step": 9696 }, { "epoch": 24.2425, "grad_norm": 7.0, "learning_rate": 1.772522246628083e-05, "loss": 1.3305, "step": 9697 }, { "epoch": 24.245, "grad_norm": 7.53125, "learning_rate": 1.7724708298511943e-05, "loss": 1.3264, "step": 9698 }, { "epoch": 24.2475, "grad_norm": 11.25, "learning_rate": 1.7724194080100178e-05, "loss": 1.4288, "step": 9699 }, { "epoch": 24.25, "grad_norm": 5.96875, "learning_rate": 1.7723679811048904e-05, "loss": 1.2144, "step": 9700 }, { "epoch": 24.2525, "grad_norm": 8.0625, "learning_rate": 1.7723165491361493e-05, "loss": 1.4209, "step": 9701 }, { "epoch": 24.255, "grad_norm": 7.4375, "learning_rate": 1.7722651121041318e-05, "loss": 1.3237, "step": 9702 }, { "epoch": 24.2575, "grad_norm": 23.375, "learning_rate": 1.7722136700091758e-05, "loss": 1.5155, "step": 9703 }, { "epoch": 24.26, "grad_norm": 9.125, "learning_rate": 1.772162222851617e-05, "loss": 1.4422, "step": 9704 }, { "epoch": 24.2625, "grad_norm": 7.875, "learning_rate": 1.772110770631794e-05, "loss": 1.3308, "step": 9705 }, { "epoch": 24.265, "grad_norm": 6.53125, "learning_rate": 1.7720593133500434e-05, "loss": 1.3915, "step": 9706 }, { "epoch": 24.2675, "grad_norm": 5.375, "learning_rate": 1.7720078510067035e-05, "loss": 1.4089, "step": 9707 }, { "epoch": 24.27, "grad_norm": 6.03125, "learning_rate": 1.7719563836021107e-05, "loss": 1.2389, "step": 9708 }, { "epoch": 24.2725, "grad_norm": 5.9375, "learning_rate": 1.7719049111366023e-05, "loss": 1.4439, "step": 9709 }, { "epoch": 24.275, "grad_norm": 7.25, "learning_rate": 1.7718534336105167e-05, "loss": 1.2804, "step": 9710 }, { "epoch": 24.2775, "grad_norm": 7.625, "learning_rate": 1.7718019510241908e-05, "loss": 1.2704, "step": 9711 }, { "epoch": 24.28, "grad_norm": 7.21875, "learning_rate": 1.7717504633779618e-05, "loss": 1.367, "step": 9712 }, { "epoch": 24.2825, "grad_norm": 9.1875, "learning_rate": 1.771698970672168e-05, "loss": 1.3871, "step": 9713 }, { "epoch": 24.285, "grad_norm": 6.21875, "learning_rate": 1.7716474729071467e-05, "loss": 1.4729, "step": 9714 }, { "epoch": 24.2875, "grad_norm": 7.4375, "learning_rate": 1.7715959700832352e-05, "loss": 1.6153, "step": 9715 }, { "epoch": 24.29, "grad_norm": 7.28125, "learning_rate": 1.7715444622007715e-05, "loss": 1.483, "step": 9716 }, { "epoch": 24.2925, "grad_norm": 6.09375, "learning_rate": 1.7714929492600933e-05, "loss": 1.5018, "step": 9717 }, { "epoch": 24.295, "grad_norm": 10.375, "learning_rate": 1.771441431261538e-05, "loss": 1.467, "step": 9718 }, { "epoch": 24.2975, "grad_norm": 6.15625, "learning_rate": 1.7713899082054437e-05, "loss": 1.2953, "step": 9719 }, { "epoch": 24.3, "grad_norm": 4.84375, "learning_rate": 1.771338380092148e-05, "loss": 1.3142, "step": 9720 }, { "epoch": 24.3025, "grad_norm": 7.3125, "learning_rate": 1.7712868469219885e-05, "loss": 1.5048, "step": 9721 }, { "epoch": 24.305, "grad_norm": 7.34375, "learning_rate": 1.7712353086953036e-05, "loss": 1.2861, "step": 9722 }, { "epoch": 24.3075, "grad_norm": 5.96875, "learning_rate": 1.7711837654124307e-05, "loss": 1.3443, "step": 9723 }, { "epoch": 24.31, "grad_norm": 7.21875, "learning_rate": 1.771132217073708e-05, "loss": 1.3458, "step": 9724 }, { "epoch": 24.3125, "grad_norm": 7.46875, "learning_rate": 1.7710806636794734e-05, "loss": 1.4181, "step": 9725 }, { "epoch": 24.315, "grad_norm": 6.125, "learning_rate": 1.7710291052300647e-05, "loss": 1.3424, "step": 9726 }, { "epoch": 24.3175, "grad_norm": 6.875, "learning_rate": 1.77097754172582e-05, "loss": 1.2992, "step": 9727 }, { "epoch": 24.32, "grad_norm": 8.0625, "learning_rate": 1.7709259731670774e-05, "loss": 1.3306, "step": 9728 }, { "epoch": 24.3225, "grad_norm": 10.9375, "learning_rate": 1.7708743995541753e-05, "loss": 1.3511, "step": 9729 }, { "epoch": 24.325, "grad_norm": 6.40625, "learning_rate": 1.770822820887451e-05, "loss": 1.3546, "step": 9730 }, { "epoch": 24.3275, "grad_norm": 7.09375, "learning_rate": 1.7707712371672436e-05, "loss": 1.4725, "step": 9731 }, { "epoch": 24.33, "grad_norm": 5.03125, "learning_rate": 1.7707196483938903e-05, "loss": 1.278, "step": 9732 }, { "epoch": 24.3325, "grad_norm": 5.9375, "learning_rate": 1.7706680545677303e-05, "loss": 1.3552, "step": 9733 }, { "epoch": 24.335, "grad_norm": 11.125, "learning_rate": 1.770616455689101e-05, "loss": 1.3445, "step": 9734 }, { "epoch": 24.3375, "grad_norm": 7.625, "learning_rate": 1.7705648517583416e-05, "loss": 1.3889, "step": 9735 }, { "epoch": 24.34, "grad_norm": 6.71875, "learning_rate": 1.7705132427757895e-05, "loss": 1.2612, "step": 9736 }, { "epoch": 24.3425, "grad_norm": 8.625, "learning_rate": 1.7704616287417832e-05, "loss": 1.5261, "step": 9737 }, { "epoch": 24.345, "grad_norm": 7.9375, "learning_rate": 1.7704100096566617e-05, "loss": 1.4243, "step": 9738 }, { "epoch": 24.3475, "grad_norm": 6.0, "learning_rate": 1.770358385520763e-05, "loss": 1.5275, "step": 9739 }, { "epoch": 24.35, "grad_norm": 4.8125, "learning_rate": 1.7703067563344252e-05, "loss": 1.2423, "step": 9740 }, { "epoch": 24.3525, "grad_norm": 5.21875, "learning_rate": 1.7702551220979872e-05, "loss": 1.3531, "step": 9741 }, { "epoch": 24.355, "grad_norm": 8.125, "learning_rate": 1.7702034828117875e-05, "loss": 1.2753, "step": 9742 }, { "epoch": 24.3575, "grad_norm": 4.8125, "learning_rate": 1.7701518384761646e-05, "loss": 1.1757, "step": 9743 }, { "epoch": 24.36, "grad_norm": 6.96875, "learning_rate": 1.770100189091457e-05, "loss": 1.4238, "step": 9744 }, { "epoch": 24.3625, "grad_norm": 7.875, "learning_rate": 1.7700485346580037e-05, "loss": 1.4937, "step": 9745 }, { "epoch": 24.365, "grad_norm": 7.78125, "learning_rate": 1.7699968751761428e-05, "loss": 1.4135, "step": 9746 }, { "epoch": 24.3675, "grad_norm": 11.3125, "learning_rate": 1.7699452106462135e-05, "loss": 1.3164, "step": 9747 }, { "epoch": 24.37, "grad_norm": 5.8125, "learning_rate": 1.7698935410685537e-05, "loss": 1.4972, "step": 9748 }, { "epoch": 24.3725, "grad_norm": 7.1875, "learning_rate": 1.7698418664435028e-05, "loss": 1.4348, "step": 9749 }, { "epoch": 24.375, "grad_norm": 6.4375, "learning_rate": 1.7697901867713997e-05, "loss": 1.3339, "step": 9750 }, { "epoch": 24.3775, "grad_norm": 7.4375, "learning_rate": 1.7697385020525824e-05, "loss": 1.4081, "step": 9751 }, { "epoch": 24.38, "grad_norm": 8.625, "learning_rate": 1.769686812287391e-05, "loss": 1.3926, "step": 9752 }, { "epoch": 24.3825, "grad_norm": 5.59375, "learning_rate": 1.769635117476163e-05, "loss": 1.371, "step": 9753 }, { "epoch": 24.385, "grad_norm": 6.71875, "learning_rate": 1.7695834176192384e-05, "loss": 1.3676, "step": 9754 }, { "epoch": 24.3875, "grad_norm": 6.75, "learning_rate": 1.7695317127169555e-05, "loss": 1.5912, "step": 9755 }, { "epoch": 24.39, "grad_norm": 7.71875, "learning_rate": 1.7694800027696536e-05, "loss": 1.5894, "step": 9756 }, { "epoch": 24.3925, "grad_norm": 9.0, "learning_rate": 1.7694282877776715e-05, "loss": 1.2981, "step": 9757 }, { "epoch": 24.395, "grad_norm": 5.9375, "learning_rate": 1.7693765677413486e-05, "loss": 1.2673, "step": 9758 }, { "epoch": 24.3975, "grad_norm": 6.15625, "learning_rate": 1.7693248426610234e-05, "loss": 1.496, "step": 9759 }, { "epoch": 24.4, "grad_norm": 5.5, "learning_rate": 1.7692731125370355e-05, "loss": 1.3293, "step": 9760 }, { "epoch": 24.4025, "grad_norm": 6.96875, "learning_rate": 1.7692213773697237e-05, "loss": 1.2769, "step": 9761 }, { "epoch": 24.405, "grad_norm": 6.25, "learning_rate": 1.7691696371594273e-05, "loss": 1.2742, "step": 9762 }, { "epoch": 24.4075, "grad_norm": 9.625, "learning_rate": 1.7691178919064857e-05, "loss": 1.3607, "step": 9763 }, { "epoch": 24.41, "grad_norm": 14.1875, "learning_rate": 1.769066141611238e-05, "loss": 1.3422, "step": 9764 }, { "epoch": 24.4125, "grad_norm": 7.6875, "learning_rate": 1.7690143862740234e-05, "loss": 1.4232, "step": 9765 }, { "epoch": 24.415, "grad_norm": 5.21875, "learning_rate": 1.7689626258951814e-05, "loss": 1.4416, "step": 9766 }, { "epoch": 24.4175, "grad_norm": 8.1875, "learning_rate": 1.7689108604750508e-05, "loss": 1.4439, "step": 9767 }, { "epoch": 24.42, "grad_norm": 6.96875, "learning_rate": 1.7688590900139715e-05, "loss": 1.3202, "step": 9768 }, { "epoch": 24.4225, "grad_norm": 7.875, "learning_rate": 1.768807314512283e-05, "loss": 1.4779, "step": 9769 }, { "epoch": 24.425, "grad_norm": 5.71875, "learning_rate": 1.7687555339703245e-05, "loss": 1.4252, "step": 9770 }, { "epoch": 24.4275, "grad_norm": 7.625, "learning_rate": 1.7687037483884352e-05, "loss": 1.3227, "step": 9771 }, { "epoch": 24.43, "grad_norm": 5.59375, "learning_rate": 1.768651957766955e-05, "loss": 1.3404, "step": 9772 }, { "epoch": 24.4325, "grad_norm": 7.9375, "learning_rate": 1.7686001621062235e-05, "loss": 1.5461, "step": 9773 }, { "epoch": 24.435, "grad_norm": 7.59375, "learning_rate": 1.76854836140658e-05, "loss": 1.2635, "step": 9774 }, { "epoch": 24.4375, "grad_norm": 5.09375, "learning_rate": 1.768496555668364e-05, "loss": 1.3927, "step": 9775 }, { "epoch": 24.44, "grad_norm": 6.90625, "learning_rate": 1.7684447448919156e-05, "loss": 1.4394, "step": 9776 }, { "epoch": 24.4425, "grad_norm": 6.9375, "learning_rate": 1.768392929077574e-05, "loss": 1.4186, "step": 9777 }, { "epoch": 24.445, "grad_norm": 4.75, "learning_rate": 1.768341108225679e-05, "loss": 1.3042, "step": 9778 }, { "epoch": 24.4475, "grad_norm": 6.90625, "learning_rate": 1.768289282336571e-05, "loss": 1.539, "step": 9779 }, { "epoch": 24.45, "grad_norm": 6.84375, "learning_rate": 1.768237451410589e-05, "loss": 1.3099, "step": 9780 }, { "epoch": 24.4525, "grad_norm": 8.375, "learning_rate": 1.7681856154480728e-05, "loss": 1.4893, "step": 9781 }, { "epoch": 24.455, "grad_norm": 9.5625, "learning_rate": 1.7681337744493626e-05, "loss": 1.3761, "step": 9782 }, { "epoch": 24.4575, "grad_norm": 6.09375, "learning_rate": 1.768081928414798e-05, "loss": 1.394, "step": 9783 }, { "epoch": 24.46, "grad_norm": 9.375, "learning_rate": 1.768030077344719e-05, "loss": 1.3532, "step": 9784 }, { "epoch": 24.4625, "grad_norm": 8.5, "learning_rate": 1.7679782212394657e-05, "loss": 1.5268, "step": 9785 }, { "epoch": 24.465, "grad_norm": 9.6875, "learning_rate": 1.7679263600993777e-05, "loss": 1.508, "step": 9786 }, { "epoch": 24.4675, "grad_norm": 8.75, "learning_rate": 1.7678744939247955e-05, "loss": 1.6186, "step": 9787 }, { "epoch": 24.47, "grad_norm": 5.84375, "learning_rate": 1.7678226227160586e-05, "loss": 1.1641, "step": 9788 }, { "epoch": 24.4725, "grad_norm": 6.21875, "learning_rate": 1.7677707464735073e-05, "loss": 1.3113, "step": 9789 }, { "epoch": 24.475, "grad_norm": 7.03125, "learning_rate": 1.767718865197482e-05, "loss": 1.2237, "step": 9790 }, { "epoch": 24.4775, "grad_norm": 7.96875, "learning_rate": 1.7676669788883222e-05, "loss": 1.3643, "step": 9791 }, { "epoch": 24.48, "grad_norm": 8.4375, "learning_rate": 1.7676150875463688e-05, "loss": 1.2597, "step": 9792 }, { "epoch": 24.4825, "grad_norm": 11.1875, "learning_rate": 1.7675631911719612e-05, "loss": 1.4939, "step": 9793 }, { "epoch": 24.485, "grad_norm": 5.90625, "learning_rate": 1.76751128976544e-05, "loss": 1.4272, "step": 9794 }, { "epoch": 24.4875, "grad_norm": 7.28125, "learning_rate": 1.767459383327146e-05, "loss": 1.3291, "step": 9795 }, { "epoch": 24.49, "grad_norm": 6.75, "learning_rate": 1.7674074718574187e-05, "loss": 1.3371, "step": 9796 }, { "epoch": 24.4925, "grad_norm": 7.3125, "learning_rate": 1.7673555553565985e-05, "loss": 1.3144, "step": 9797 }, { "epoch": 24.495, "grad_norm": 6.40625, "learning_rate": 1.767303633825026e-05, "loss": 1.4293, "step": 9798 }, { "epoch": 24.4975, "grad_norm": 5.875, "learning_rate": 1.767251707263042e-05, "loss": 1.1788, "step": 9799 }, { "epoch": 24.5, "grad_norm": 7.21875, "learning_rate": 1.767199775670986e-05, "loss": 1.4642, "step": 9800 }, { "epoch": 24.5025, "grad_norm": 5.625, "learning_rate": 1.7671478390491996e-05, "loss": 1.4067, "step": 9801 }, { "epoch": 24.505, "grad_norm": 7.46875, "learning_rate": 1.767095897398022e-05, "loss": 1.4227, "step": 9802 }, { "epoch": 24.5075, "grad_norm": 9.375, "learning_rate": 1.767043950717795e-05, "loss": 1.3771, "step": 9803 }, { "epoch": 24.51, "grad_norm": 7.125, "learning_rate": 1.7669919990088583e-05, "loss": 1.4858, "step": 9804 }, { "epoch": 24.5125, "grad_norm": 7.59375, "learning_rate": 1.7669400422715527e-05, "loss": 1.4688, "step": 9805 }, { "epoch": 24.515, "grad_norm": 4.8125, "learning_rate": 1.766888080506219e-05, "loss": 1.5269, "step": 9806 }, { "epoch": 24.5175, "grad_norm": 6.40625, "learning_rate": 1.7668361137131976e-05, "loss": 1.3123, "step": 9807 }, { "epoch": 24.52, "grad_norm": 6.03125, "learning_rate": 1.7667841418928292e-05, "loss": 1.4266, "step": 9808 }, { "epoch": 24.5225, "grad_norm": 7.15625, "learning_rate": 1.766732165045455e-05, "loss": 1.4138, "step": 9809 }, { "epoch": 24.525, "grad_norm": 6.40625, "learning_rate": 1.766680183171415e-05, "loss": 1.4604, "step": 9810 }, { "epoch": 24.5275, "grad_norm": 6.90625, "learning_rate": 1.7666281962710504e-05, "loss": 1.3661, "step": 9811 }, { "epoch": 24.53, "grad_norm": 8.6875, "learning_rate": 1.7665762043447022e-05, "loss": 1.4736, "step": 9812 }, { "epoch": 24.5325, "grad_norm": 6.0, "learning_rate": 1.7665242073927114e-05, "loss": 1.3256, "step": 9813 }, { "epoch": 24.535, "grad_norm": 7.53125, "learning_rate": 1.7664722054154185e-05, "loss": 1.4293, "step": 9814 }, { "epoch": 24.5375, "grad_norm": 5.6875, "learning_rate": 1.7664201984131642e-05, "loss": 1.44, "step": 9815 }, { "epoch": 24.54, "grad_norm": 7.6875, "learning_rate": 1.7663681863862895e-05, "loss": 1.4352, "step": 9816 }, { "epoch": 24.5425, "grad_norm": 7.875, "learning_rate": 1.766316169335136e-05, "loss": 1.4278, "step": 9817 }, { "epoch": 24.545, "grad_norm": 8.1875, "learning_rate": 1.7662641472600445e-05, "loss": 1.4218, "step": 9818 }, { "epoch": 24.5475, "grad_norm": 7.09375, "learning_rate": 1.766212120161356e-05, "loss": 1.502, "step": 9819 }, { "epoch": 24.55, "grad_norm": 7.5625, "learning_rate": 1.7661600880394113e-05, "loss": 1.2525, "step": 9820 }, { "epoch": 24.5525, "grad_norm": 6.71875, "learning_rate": 1.7661080508945514e-05, "loss": 1.4027, "step": 9821 }, { "epoch": 24.555, "grad_norm": 7.34375, "learning_rate": 1.7660560087271182e-05, "loss": 1.5469, "step": 9822 }, { "epoch": 24.5575, "grad_norm": 6.5, "learning_rate": 1.7660039615374522e-05, "loss": 1.5368, "step": 9823 }, { "epoch": 24.56, "grad_norm": 5.375, "learning_rate": 1.765951909325895e-05, "loss": 1.3401, "step": 9824 }, { "epoch": 24.5625, "grad_norm": 7.84375, "learning_rate": 1.7658998520927878e-05, "loss": 1.5424, "step": 9825 }, { "epoch": 24.565, "grad_norm": 6.8125, "learning_rate": 1.7658477898384717e-05, "loss": 1.3746, "step": 9826 }, { "epoch": 24.5675, "grad_norm": 7.15625, "learning_rate": 1.7657957225632882e-05, "loss": 1.443, "step": 9827 }, { "epoch": 24.57, "grad_norm": 7.1875, "learning_rate": 1.7657436502675785e-05, "loss": 1.4987, "step": 9828 }, { "epoch": 24.5725, "grad_norm": 6.4375, "learning_rate": 1.765691572951684e-05, "loss": 1.3299, "step": 9829 }, { "epoch": 24.575, "grad_norm": 6.125, "learning_rate": 1.7656394906159465e-05, "loss": 1.3318, "step": 9830 }, { "epoch": 24.5775, "grad_norm": 6.34375, "learning_rate": 1.765587403260707e-05, "loss": 1.4103, "step": 9831 }, { "epoch": 24.58, "grad_norm": 7.65625, "learning_rate": 1.7655353108863068e-05, "loss": 1.3137, "step": 9832 }, { "epoch": 24.5825, "grad_norm": 7.28125, "learning_rate": 1.7654832134930885e-05, "loss": 1.4163, "step": 9833 }, { "epoch": 24.585, "grad_norm": 7.53125, "learning_rate": 1.7654311110813924e-05, "loss": 1.3363, "step": 9834 }, { "epoch": 24.5875, "grad_norm": 8.75, "learning_rate": 1.7653790036515603e-05, "loss": 1.5857, "step": 9835 }, { "epoch": 24.59, "grad_norm": 7.96875, "learning_rate": 1.7653268912039346e-05, "loss": 1.4403, "step": 9836 }, { "epoch": 24.5925, "grad_norm": 7.4375, "learning_rate": 1.765274773738856e-05, "loss": 1.3241, "step": 9837 }, { "epoch": 24.595, "grad_norm": 6.0625, "learning_rate": 1.7652226512566668e-05, "loss": 1.348, "step": 9838 }, { "epoch": 24.5975, "grad_norm": 8.75, "learning_rate": 1.7651705237577085e-05, "loss": 1.2749, "step": 9839 }, { "epoch": 24.6, "grad_norm": 8.1875, "learning_rate": 1.7651183912423228e-05, "loss": 1.5209, "step": 9840 }, { "epoch": 24.6025, "grad_norm": 7.53125, "learning_rate": 1.7650662537108518e-05, "loss": 1.398, "step": 9841 }, { "epoch": 24.605, "grad_norm": 16.875, "learning_rate": 1.765014111163637e-05, "loss": 1.3771, "step": 9842 }, { "epoch": 24.6075, "grad_norm": 6.5625, "learning_rate": 1.76496196360102e-05, "loss": 1.5444, "step": 9843 }, { "epoch": 24.61, "grad_norm": 6.5, "learning_rate": 1.764909811023343e-05, "loss": 1.3701, "step": 9844 }, { "epoch": 24.6125, "grad_norm": 8.8125, "learning_rate": 1.7648576534309484e-05, "loss": 1.3514, "step": 9845 }, { "epoch": 24.615, "grad_norm": 6.0, "learning_rate": 1.7648054908241773e-05, "loss": 1.355, "step": 9846 }, { "epoch": 24.6175, "grad_norm": 7.1875, "learning_rate": 1.764753323203372e-05, "loss": 1.2102, "step": 9847 }, { "epoch": 24.62, "grad_norm": 8.4375, "learning_rate": 1.7647011505688745e-05, "loss": 1.3473, "step": 9848 }, { "epoch": 24.6225, "grad_norm": 8.5625, "learning_rate": 1.764648972921027e-05, "loss": 1.3599, "step": 9849 }, { "epoch": 24.625, "grad_norm": 7.65625, "learning_rate": 1.764596790260171e-05, "loss": 1.3744, "step": 9850 }, { "epoch": 24.6275, "grad_norm": 6.21875, "learning_rate": 1.7645446025866494e-05, "loss": 1.385, "step": 9851 }, { "epoch": 24.63, "grad_norm": 5.21875, "learning_rate": 1.7644924099008038e-05, "loss": 1.3004, "step": 9852 }, { "epoch": 24.6325, "grad_norm": 6.1875, "learning_rate": 1.764440212202977e-05, "loss": 1.3947, "step": 9853 }, { "epoch": 24.635, "grad_norm": 5.28125, "learning_rate": 1.7643880094935105e-05, "loss": 1.3943, "step": 9854 }, { "epoch": 24.6375, "grad_norm": 6.96875, "learning_rate": 1.7643358017727464e-05, "loss": 1.3058, "step": 9855 }, { "epoch": 24.64, "grad_norm": 5.875, "learning_rate": 1.764283589041028e-05, "loss": 1.3715, "step": 9856 }, { "epoch": 24.6425, "grad_norm": 8.0, "learning_rate": 1.7642313712986966e-05, "loss": 1.4135, "step": 9857 }, { "epoch": 24.645, "grad_norm": 6.4375, "learning_rate": 1.764179148546095e-05, "loss": 1.2601, "step": 9858 }, { "epoch": 24.6475, "grad_norm": 6.0625, "learning_rate": 1.7641269207835653e-05, "loss": 1.4818, "step": 9859 }, { "epoch": 24.65, "grad_norm": 6.71875, "learning_rate": 1.7640746880114505e-05, "loss": 1.4559, "step": 9860 }, { "epoch": 24.6525, "grad_norm": 6.125, "learning_rate": 1.7640224502300922e-05, "loss": 1.3961, "step": 9861 }, { "epoch": 24.655, "grad_norm": 12.3125, "learning_rate": 1.7639702074398335e-05, "loss": 1.5281, "step": 9862 }, { "epoch": 24.6575, "grad_norm": 9.0, "learning_rate": 1.7639179596410167e-05, "loss": 1.4886, "step": 9863 }, { "epoch": 24.66, "grad_norm": 9.75, "learning_rate": 1.7638657068339845e-05, "loss": 1.4152, "step": 9864 }, { "epoch": 24.6625, "grad_norm": 5.09375, "learning_rate": 1.7638134490190787e-05, "loss": 1.5037, "step": 9865 }, { "epoch": 24.665, "grad_norm": 7.1875, "learning_rate": 1.763761186196643e-05, "loss": 1.5853, "step": 9866 }, { "epoch": 24.6675, "grad_norm": 7.65625, "learning_rate": 1.7637089183670195e-05, "loss": 1.6685, "step": 9867 }, { "epoch": 24.67, "grad_norm": 7.4375, "learning_rate": 1.763656645530551e-05, "loss": 1.3761, "step": 9868 }, { "epoch": 24.6725, "grad_norm": 9.0625, "learning_rate": 1.7636043676875795e-05, "loss": 1.3226, "step": 9869 }, { "epoch": 24.675, "grad_norm": 7.34375, "learning_rate": 1.7635520848384492e-05, "loss": 1.3388, "step": 9870 }, { "epoch": 24.6775, "grad_norm": 7.25, "learning_rate": 1.7634997969835013e-05, "loss": 1.3924, "step": 9871 }, { "epoch": 24.68, "grad_norm": 7.625, "learning_rate": 1.7634475041230796e-05, "loss": 1.4341, "step": 9872 }, { "epoch": 24.6825, "grad_norm": 6.15625, "learning_rate": 1.7633952062575263e-05, "loss": 1.3434, "step": 9873 }, { "epoch": 24.685, "grad_norm": 9.125, "learning_rate": 1.763342903387185e-05, "loss": 1.4533, "step": 9874 }, { "epoch": 24.6875, "grad_norm": 6.875, "learning_rate": 1.763290595512398e-05, "loss": 1.448, "step": 9875 }, { "epoch": 24.69, "grad_norm": 10.625, "learning_rate": 1.7632382826335082e-05, "loss": 1.3359, "step": 9876 }, { "epoch": 24.6925, "grad_norm": 6.46875, "learning_rate": 1.7631859647508588e-05, "loss": 1.5144, "step": 9877 }, { "epoch": 24.695, "grad_norm": 6.28125, "learning_rate": 1.763133641864793e-05, "loss": 1.467, "step": 9878 }, { "epoch": 24.6975, "grad_norm": 5.8125, "learning_rate": 1.7630813139756536e-05, "loss": 1.2281, "step": 9879 }, { "epoch": 24.7, "grad_norm": 10.375, "learning_rate": 1.7630289810837836e-05, "loss": 1.4831, "step": 9880 }, { "epoch": 24.7025, "grad_norm": 6.8125, "learning_rate": 1.7629766431895258e-05, "loss": 1.4193, "step": 9881 }, { "epoch": 24.705, "grad_norm": 7.6875, "learning_rate": 1.762924300293224e-05, "loss": 1.4044, "step": 9882 }, { "epoch": 24.7075, "grad_norm": 8.875, "learning_rate": 1.7628719523952207e-05, "loss": 1.3382, "step": 9883 }, { "epoch": 24.71, "grad_norm": 8.6875, "learning_rate": 1.7628195994958596e-05, "loss": 1.4962, "step": 9884 }, { "epoch": 24.7125, "grad_norm": 6.46875, "learning_rate": 1.762767241595484e-05, "loss": 1.35, "step": 9885 }, { "epoch": 24.715, "grad_norm": 6.15625, "learning_rate": 1.7627148786944366e-05, "loss": 1.4299, "step": 9886 }, { "epoch": 24.7175, "grad_norm": 8.125, "learning_rate": 1.7626625107930612e-05, "loss": 1.4101, "step": 9887 }, { "epoch": 24.72, "grad_norm": 5.65625, "learning_rate": 1.7626101378917004e-05, "loss": 1.419, "step": 9888 }, { "epoch": 24.7225, "grad_norm": 5.625, "learning_rate": 1.7625577599906985e-05, "loss": 1.3298, "step": 9889 }, { "epoch": 24.725, "grad_norm": 6.09375, "learning_rate": 1.762505377090398e-05, "loss": 1.3875, "step": 9890 }, { "epoch": 24.7275, "grad_norm": 6.625, "learning_rate": 1.7624529891911436e-05, "loss": 1.3305, "step": 9891 }, { "epoch": 24.73, "grad_norm": 7.71875, "learning_rate": 1.762400596293277e-05, "loss": 1.3753, "step": 9892 }, { "epoch": 24.7325, "grad_norm": 9.1875, "learning_rate": 1.7623481983971432e-05, "loss": 1.5708, "step": 9893 }, { "epoch": 24.735, "grad_norm": 5.34375, "learning_rate": 1.7622957955030847e-05, "loss": 1.3369, "step": 9894 }, { "epoch": 24.7375, "grad_norm": 6.0625, "learning_rate": 1.762243387611446e-05, "loss": 1.3611, "step": 9895 }, { "epoch": 24.74, "grad_norm": 7.65625, "learning_rate": 1.76219097472257e-05, "loss": 1.3732, "step": 9896 }, { "epoch": 24.7425, "grad_norm": 8.875, "learning_rate": 1.7621385568368e-05, "loss": 1.4861, "step": 9897 }, { "epoch": 24.745, "grad_norm": 8.375, "learning_rate": 1.7620861339544805e-05, "loss": 1.4531, "step": 9898 }, { "epoch": 24.7475, "grad_norm": 5.34375, "learning_rate": 1.7620337060759546e-05, "loss": 1.4663, "step": 9899 }, { "epoch": 24.75, "grad_norm": 10.125, "learning_rate": 1.7619812732015664e-05, "loss": 1.5194, "step": 9900 }, { "epoch": 24.7525, "grad_norm": 9.3125, "learning_rate": 1.7619288353316592e-05, "loss": 1.5369, "step": 9901 }, { "epoch": 24.755, "grad_norm": 8.1875, "learning_rate": 1.7618763924665777e-05, "loss": 1.2786, "step": 9902 }, { "epoch": 24.7575, "grad_norm": 4.625, "learning_rate": 1.7618239446066642e-05, "loss": 1.2911, "step": 9903 }, { "epoch": 24.76, "grad_norm": 6.28125, "learning_rate": 1.761771491752264e-05, "loss": 1.3401, "step": 9904 }, { "epoch": 24.7625, "grad_norm": 7.5625, "learning_rate": 1.7617190339037203e-05, "loss": 1.2804, "step": 9905 }, { "epoch": 24.765, "grad_norm": 10.375, "learning_rate": 1.761666571061377e-05, "loss": 1.4551, "step": 9906 }, { "epoch": 24.7675, "grad_norm": 6.5625, "learning_rate": 1.761614103225578e-05, "loss": 1.4667, "step": 9907 }, { "epoch": 24.77, "grad_norm": 7.03125, "learning_rate": 1.7615616303966675e-05, "loss": 1.3284, "step": 9908 }, { "epoch": 24.7725, "grad_norm": 8.375, "learning_rate": 1.7615091525749897e-05, "loss": 1.4124, "step": 9909 }, { "epoch": 24.775, "grad_norm": 6.84375, "learning_rate": 1.761456669760888e-05, "loss": 1.3705, "step": 9910 }, { "epoch": 24.7775, "grad_norm": 7.71875, "learning_rate": 1.7614041819547074e-05, "loss": 1.5, "step": 9911 }, { "epoch": 24.78, "grad_norm": 10.8125, "learning_rate": 1.7613516891567907e-05, "loss": 1.5369, "step": 9912 }, { "epoch": 24.7825, "grad_norm": 5.9375, "learning_rate": 1.761299191367483e-05, "loss": 1.4272, "step": 9913 }, { "epoch": 24.785, "grad_norm": 7.34375, "learning_rate": 1.7612466885871286e-05, "loss": 1.4994, "step": 9914 }, { "epoch": 24.7875, "grad_norm": 9.4375, "learning_rate": 1.7611941808160713e-05, "loss": 1.2268, "step": 9915 }, { "epoch": 24.79, "grad_norm": 7.25, "learning_rate": 1.761141668054655e-05, "loss": 1.4285, "step": 9916 }, { "epoch": 24.7925, "grad_norm": 6.34375, "learning_rate": 1.7610891503032247e-05, "loss": 1.2953, "step": 9917 }, { "epoch": 24.795, "grad_norm": 7.4375, "learning_rate": 1.7610366275621245e-05, "loss": 1.5778, "step": 9918 }, { "epoch": 24.7975, "grad_norm": 8.5, "learning_rate": 1.7609840998316984e-05, "loss": 1.4539, "step": 9919 }, { "epoch": 24.8, "grad_norm": 7.53125, "learning_rate": 1.7609315671122912e-05, "loss": 1.3018, "step": 9920 }, { "epoch": 24.8025, "grad_norm": 9.25, "learning_rate": 1.760879029404247e-05, "loss": 1.4677, "step": 9921 }, { "epoch": 24.805, "grad_norm": 6.375, "learning_rate": 1.7608264867079104e-05, "loss": 1.4286, "step": 9922 }, { "epoch": 24.8075, "grad_norm": 9.375, "learning_rate": 1.7607739390236255e-05, "loss": 1.5683, "step": 9923 }, { "epoch": 24.81, "grad_norm": 6.90625, "learning_rate": 1.7607213863517376e-05, "loss": 1.4709, "step": 9924 }, { "epoch": 24.8125, "grad_norm": 6.375, "learning_rate": 1.7606688286925905e-05, "loss": 1.4012, "step": 9925 }, { "epoch": 24.815, "grad_norm": 8.6875, "learning_rate": 1.7606162660465288e-05, "loss": 1.3535, "step": 9926 }, { "epoch": 24.8175, "grad_norm": 7.875, "learning_rate": 1.7605636984138973e-05, "loss": 1.3167, "step": 9927 }, { "epoch": 24.82, "grad_norm": 8.6875, "learning_rate": 1.7605111257950408e-05, "loss": 1.5306, "step": 9928 }, { "epoch": 24.8225, "grad_norm": 7.90625, "learning_rate": 1.7604585481903037e-05, "loss": 1.4847, "step": 9929 }, { "epoch": 24.825, "grad_norm": 16.125, "learning_rate": 1.7604059656000313e-05, "loss": 1.466, "step": 9930 }, { "epoch": 24.8275, "grad_norm": 7.6875, "learning_rate": 1.7603533780245672e-05, "loss": 1.3725, "step": 9931 }, { "epoch": 24.83, "grad_norm": 7.34375, "learning_rate": 1.760300785464257e-05, "loss": 1.4557, "step": 9932 }, { "epoch": 24.8325, "grad_norm": 5.09375, "learning_rate": 1.7602481879194454e-05, "loss": 1.3981, "step": 9933 }, { "epoch": 24.835, "grad_norm": 7.3125, "learning_rate": 1.7601955853904765e-05, "loss": 1.5203, "step": 9934 }, { "epoch": 24.8375, "grad_norm": 8.25, "learning_rate": 1.7601429778776966e-05, "loss": 1.4276, "step": 9935 }, { "epoch": 24.84, "grad_norm": 6.40625, "learning_rate": 1.760090365381449e-05, "loss": 1.4953, "step": 9936 }, { "epoch": 24.8425, "grad_norm": 34.75, "learning_rate": 1.76003774790208e-05, "loss": 1.2677, "step": 9937 }, { "epoch": 24.845, "grad_norm": 4.8125, "learning_rate": 1.7599851254399336e-05, "loss": 1.3754, "step": 9938 }, { "epoch": 24.8475, "grad_norm": 4.78125, "learning_rate": 1.7599324979953554e-05, "loss": 1.4025, "step": 9939 }, { "epoch": 24.85, "grad_norm": 7.71875, "learning_rate": 1.75987986556869e-05, "loss": 1.4297, "step": 9940 }, { "epoch": 24.8525, "grad_norm": 5.40625, "learning_rate": 1.7598272281602823e-05, "loss": 1.4887, "step": 9941 }, { "epoch": 24.855, "grad_norm": 4.375, "learning_rate": 1.7597745857704782e-05, "loss": 1.2763, "step": 9942 }, { "epoch": 24.8575, "grad_norm": 14.0, "learning_rate": 1.759721938399622e-05, "loss": 1.618, "step": 9943 }, { "epoch": 24.86, "grad_norm": 5.625, "learning_rate": 1.7596692860480593e-05, "loss": 1.2099, "step": 9944 }, { "epoch": 24.8625, "grad_norm": 6.53125, "learning_rate": 1.759616628716135e-05, "loss": 1.3817, "step": 9945 }, { "epoch": 24.865, "grad_norm": 6.8125, "learning_rate": 1.7595639664041946e-05, "loss": 1.2776, "step": 9946 }, { "epoch": 24.8675, "grad_norm": 6.625, "learning_rate": 1.7595112991125835e-05, "loss": 1.4342, "step": 9947 }, { "epoch": 24.87, "grad_norm": 7.03125, "learning_rate": 1.759458626841646e-05, "loss": 1.1798, "step": 9948 }, { "epoch": 24.8725, "grad_norm": 12.5, "learning_rate": 1.7594059495917292e-05, "loss": 1.5444, "step": 9949 }, { "epoch": 24.875, "grad_norm": 8.3125, "learning_rate": 1.7593532673631765e-05, "loss": 1.4131, "step": 9950 }, { "epoch": 24.8775, "grad_norm": 6.875, "learning_rate": 1.7593005801563344e-05, "loss": 1.304, "step": 9951 }, { "epoch": 24.88, "grad_norm": 6.90625, "learning_rate": 1.759247887971548e-05, "loss": 1.2648, "step": 9952 }, { "epoch": 24.8825, "grad_norm": 8.0625, "learning_rate": 1.7591951908091632e-05, "loss": 1.258, "step": 9953 }, { "epoch": 24.885, "grad_norm": 8.5625, "learning_rate": 1.7591424886695246e-05, "loss": 1.4283, "step": 9954 }, { "epoch": 24.8875, "grad_norm": 6.3125, "learning_rate": 1.759089781552979e-05, "loss": 1.4099, "step": 9955 }, { "epoch": 24.89, "grad_norm": 10.5, "learning_rate": 1.75903706945987e-05, "loss": 1.4756, "step": 9956 }, { "epoch": 24.8925, "grad_norm": 8.375, "learning_rate": 1.7589843523905453e-05, "loss": 1.4373, "step": 9957 }, { "epoch": 24.895, "grad_norm": 5.0625, "learning_rate": 1.7589316303453495e-05, "loss": 1.28, "step": 9958 }, { "epoch": 24.8975, "grad_norm": 5.84375, "learning_rate": 1.7588789033246277e-05, "loss": 1.5681, "step": 9959 }, { "epoch": 24.9, "grad_norm": 8.5625, "learning_rate": 1.758826171328727e-05, "loss": 1.3519, "step": 9960 }, { "epoch": 24.9025, "grad_norm": 6.03125, "learning_rate": 1.7587734343579915e-05, "loss": 1.3754, "step": 9961 }, { "epoch": 24.905, "grad_norm": 6.875, "learning_rate": 1.758720692412768e-05, "loss": 1.4705, "step": 9962 }, { "epoch": 24.9075, "grad_norm": 9.0, "learning_rate": 1.7586679454934025e-05, "loss": 1.5157, "step": 9963 }, { "epoch": 24.91, "grad_norm": 9.9375, "learning_rate": 1.75861519360024e-05, "loss": 1.3194, "step": 9964 }, { "epoch": 24.9125, "grad_norm": 5.46875, "learning_rate": 1.7585624367336263e-05, "loss": 1.4523, "step": 9965 }, { "epoch": 24.915, "grad_norm": 11.875, "learning_rate": 1.7585096748939076e-05, "loss": 1.3828, "step": 9966 }, { "epoch": 24.9175, "grad_norm": 7.125, "learning_rate": 1.75845690808143e-05, "loss": 1.5194, "step": 9967 }, { "epoch": 24.92, "grad_norm": 7.65625, "learning_rate": 1.7584041362965397e-05, "loss": 1.3257, "step": 9968 }, { "epoch": 24.9225, "grad_norm": 10.375, "learning_rate": 1.7583513595395817e-05, "loss": 1.3995, "step": 9969 }, { "epoch": 24.925, "grad_norm": 6.53125, "learning_rate": 1.7582985778109026e-05, "loss": 1.2022, "step": 9970 }, { "epoch": 24.9275, "grad_norm": 5.46875, "learning_rate": 1.7582457911108485e-05, "loss": 1.3926, "step": 9971 }, { "epoch": 24.93, "grad_norm": 6.1875, "learning_rate": 1.7581929994397654e-05, "loss": 1.3231, "step": 9972 }, { "epoch": 24.9325, "grad_norm": 6.4375, "learning_rate": 1.758140202797999e-05, "loss": 1.4071, "step": 9973 }, { "epoch": 24.935, "grad_norm": 6.78125, "learning_rate": 1.758087401185896e-05, "loss": 1.4899, "step": 9974 }, { "epoch": 24.9375, "grad_norm": 6.0, "learning_rate": 1.7580345946038025e-05, "loss": 1.4392, "step": 9975 }, { "epoch": 24.94, "grad_norm": 8.625, "learning_rate": 1.7579817830520644e-05, "loss": 1.5367, "step": 9976 }, { "epoch": 24.9425, "grad_norm": 6.0, "learning_rate": 1.7579289665310278e-05, "loss": 1.4434, "step": 9977 }, { "epoch": 24.945, "grad_norm": 7.28125, "learning_rate": 1.75787614504104e-05, "loss": 1.437, "step": 9978 }, { "epoch": 24.9475, "grad_norm": 8.0, "learning_rate": 1.7578233185824457e-05, "loss": 1.4502, "step": 9979 }, { "epoch": 24.95, "grad_norm": 8.3125, "learning_rate": 1.7577704871555924e-05, "loss": 1.3699, "step": 9980 }, { "epoch": 24.9525, "grad_norm": 9.125, "learning_rate": 1.757717650760826e-05, "loss": 1.4011, "step": 9981 }, { "epoch": 24.955, "grad_norm": 9.4375, "learning_rate": 1.7576648093984928e-05, "loss": 1.517, "step": 9982 }, { "epoch": 24.9575, "grad_norm": 5.34375, "learning_rate": 1.75761196306894e-05, "loss": 1.4563, "step": 9983 }, { "epoch": 24.96, "grad_norm": 8.25, "learning_rate": 1.7575591117725132e-05, "loss": 1.4663, "step": 9984 }, { "epoch": 24.9625, "grad_norm": 5.9375, "learning_rate": 1.757506255509559e-05, "loss": 1.2502, "step": 9985 }, { "epoch": 24.965, "grad_norm": 7.5, "learning_rate": 1.7574533942804245e-05, "loss": 1.3546, "step": 9986 }, { "epoch": 24.9675, "grad_norm": 5.84375, "learning_rate": 1.7574005280854555e-05, "loss": 1.3652, "step": 9987 }, { "epoch": 24.97, "grad_norm": 5.71875, "learning_rate": 1.757347656924999e-05, "loss": 1.3087, "step": 9988 }, { "epoch": 24.9725, "grad_norm": 6.71875, "learning_rate": 1.7572947807994016e-05, "loss": 1.3193, "step": 9989 }, { "epoch": 24.975, "grad_norm": 9.6875, "learning_rate": 1.7572418997090103e-05, "loss": 1.4462, "step": 9990 }, { "epoch": 24.9775, "grad_norm": 5.1875, "learning_rate": 1.7571890136541707e-05, "loss": 1.358, "step": 9991 }, { "epoch": 24.98, "grad_norm": 8.4375, "learning_rate": 1.7571361226352305e-05, "loss": 1.4295, "step": 9992 }, { "epoch": 24.9825, "grad_norm": 7.0, "learning_rate": 1.7570832266525367e-05, "loss": 1.3266, "step": 9993 }, { "epoch": 24.985, "grad_norm": 6.0625, "learning_rate": 1.7570303257064347e-05, "loss": 1.2312, "step": 9994 }, { "epoch": 24.9875, "grad_norm": 6.40625, "learning_rate": 1.7569774197972726e-05, "loss": 1.3577, "step": 9995 }, { "epoch": 24.99, "grad_norm": 6.59375, "learning_rate": 1.756924508925397e-05, "loss": 1.4862, "step": 9996 }, { "epoch": 24.9925, "grad_norm": 7.34375, "learning_rate": 1.7568715930911545e-05, "loss": 1.2796, "step": 9997 }, { "epoch": 24.995, "grad_norm": 7.75, "learning_rate": 1.756818672294892e-05, "loss": 1.1972, "step": 9998 }, { "epoch": 24.9975, "grad_norm": 7.625, "learning_rate": 1.7567657465369568e-05, "loss": 1.4723, "step": 9999 }, { "epoch": 25.0, "grad_norm": 7.71875, "learning_rate": 1.7567128158176955e-05, "loss": 1.514, "step": 10000 }, { "epoch": 25.0025, "grad_norm": 6.34375, "learning_rate": 1.756659880137455e-05, "loss": 1.4009, "step": 10001 }, { "epoch": 25.005, "grad_norm": 9.25, "learning_rate": 1.756606939496583e-05, "loss": 1.3517, "step": 10002 }, { "epoch": 25.0075, "grad_norm": 6.8125, "learning_rate": 1.756553993895426e-05, "loss": 1.3601, "step": 10003 }, { "epoch": 25.01, "grad_norm": 6.875, "learning_rate": 1.7565010433343316e-05, "loss": 1.4103, "step": 10004 }, { "epoch": 25.0125, "grad_norm": 6.75, "learning_rate": 1.756448087813646e-05, "loss": 1.2611, "step": 10005 }, { "epoch": 25.015, "grad_norm": 6.5, "learning_rate": 1.7563951273337175e-05, "loss": 1.5207, "step": 10006 }, { "epoch": 25.0175, "grad_norm": 7.25, "learning_rate": 1.756342161894893e-05, "loss": 1.4025, "step": 10007 }, { "epoch": 25.02, "grad_norm": 7.8125, "learning_rate": 1.756289191497519e-05, "loss": 1.4596, "step": 10008 }, { "epoch": 25.0225, "grad_norm": 6.625, "learning_rate": 1.7562362161419434e-05, "loss": 1.4037, "step": 10009 }, { "epoch": 25.025, "grad_norm": 8.8125, "learning_rate": 1.7561832358285138e-05, "loss": 1.3336, "step": 10010 }, { "epoch": 25.0275, "grad_norm": 5.0, "learning_rate": 1.7561302505575768e-05, "loss": 1.3446, "step": 10011 }, { "epoch": 25.03, "grad_norm": 8.0625, "learning_rate": 1.75607726032948e-05, "loss": 1.374, "step": 10012 }, { "epoch": 25.0325, "grad_norm": 6.34375, "learning_rate": 1.7560242651445718e-05, "loss": 1.2433, "step": 10013 }, { "epoch": 25.035, "grad_norm": 8.5625, "learning_rate": 1.755971265003198e-05, "loss": 1.2695, "step": 10014 }, { "epoch": 25.0375, "grad_norm": 8.25, "learning_rate": 1.7559182599057073e-05, "loss": 1.399, "step": 10015 }, { "epoch": 25.04, "grad_norm": 7.28125, "learning_rate": 1.7558652498524464e-05, "loss": 1.4469, "step": 10016 }, { "epoch": 25.0425, "grad_norm": 7.5, "learning_rate": 1.755812234843763e-05, "loss": 1.3415, "step": 10017 }, { "epoch": 25.045, "grad_norm": 6.125, "learning_rate": 1.7557592148800054e-05, "loss": 1.3212, "step": 10018 }, { "epoch": 25.0475, "grad_norm": 8.0625, "learning_rate": 1.7557061899615203e-05, "loss": 1.4876, "step": 10019 }, { "epoch": 25.05, "grad_norm": 6.21875, "learning_rate": 1.7556531600886554e-05, "loss": 1.3109, "step": 10020 }, { "epoch": 25.0525, "grad_norm": 8.125, "learning_rate": 1.755600125261759e-05, "loss": 1.5014, "step": 10021 }, { "epoch": 25.055, "grad_norm": 6.90625, "learning_rate": 1.7555470854811782e-05, "loss": 1.3719, "step": 10022 }, { "epoch": 25.0575, "grad_norm": 6.78125, "learning_rate": 1.7554940407472612e-05, "loss": 1.4781, "step": 10023 }, { "epoch": 25.06, "grad_norm": 6.90625, "learning_rate": 1.7554409910603553e-05, "loss": 1.3618, "step": 10024 }, { "epoch": 25.0625, "grad_norm": 8.875, "learning_rate": 1.7553879364208088e-05, "loss": 1.4458, "step": 10025 }, { "epoch": 25.065, "grad_norm": 8.25, "learning_rate": 1.7553348768289686e-05, "loss": 1.378, "step": 10026 }, { "epoch": 25.0675, "grad_norm": 8.4375, "learning_rate": 1.7552818122851838e-05, "loss": 1.4132, "step": 10027 }, { "epoch": 25.07, "grad_norm": 10.375, "learning_rate": 1.7552287427898015e-05, "loss": 1.5488, "step": 10028 }, { "epoch": 25.0725, "grad_norm": 7.625, "learning_rate": 1.7551756683431694e-05, "loss": 1.5277, "step": 10029 }, { "epoch": 25.075, "grad_norm": 7.71875, "learning_rate": 1.7551225889456366e-05, "loss": 1.4152, "step": 10030 }, { "epoch": 25.0775, "grad_norm": 6.625, "learning_rate": 1.7550695045975496e-05, "loss": 1.2831, "step": 10031 }, { "epoch": 25.08, "grad_norm": 6.6875, "learning_rate": 1.7550164152992573e-05, "loss": 1.3868, "step": 10032 }, { "epoch": 25.0825, "grad_norm": 5.59375, "learning_rate": 1.7549633210511078e-05, "loss": 1.2439, "step": 10033 }, { "epoch": 25.085, "grad_norm": 12.6875, "learning_rate": 1.7549102218534487e-05, "loss": 1.4603, "step": 10034 }, { "epoch": 25.0875, "grad_norm": 6.25, "learning_rate": 1.754857117706629e-05, "loss": 1.375, "step": 10035 }, { "epoch": 25.09, "grad_norm": 8.5625, "learning_rate": 1.7548040086109957e-05, "loss": 1.2863, "step": 10036 }, { "epoch": 25.0925, "grad_norm": 6.28125, "learning_rate": 1.7547508945668975e-05, "loss": 1.315, "step": 10037 }, { "epoch": 25.095, "grad_norm": 6.15625, "learning_rate": 1.7546977755746828e-05, "loss": 1.3314, "step": 10038 }, { "epoch": 25.0975, "grad_norm": 12.375, "learning_rate": 1.7546446516346996e-05, "loss": 1.5406, "step": 10039 }, { "epoch": 25.1, "grad_norm": 5.4375, "learning_rate": 1.7545915227472967e-05, "loss": 1.3246, "step": 10040 }, { "epoch": 25.1025, "grad_norm": 7.15625, "learning_rate": 1.7545383889128215e-05, "loss": 1.3973, "step": 10041 }, { "epoch": 25.105, "grad_norm": 8.4375, "learning_rate": 1.7544852501316227e-05, "loss": 1.4118, "step": 10042 }, { "epoch": 25.1075, "grad_norm": 6.90625, "learning_rate": 1.754432106404049e-05, "loss": 1.5407, "step": 10043 }, { "epoch": 25.11, "grad_norm": 6.09375, "learning_rate": 1.7543789577304487e-05, "loss": 1.4496, "step": 10044 }, { "epoch": 25.1125, "grad_norm": 6.71875, "learning_rate": 1.75432580411117e-05, "loss": 1.524, "step": 10045 }, { "epoch": 25.115, "grad_norm": 6.65625, "learning_rate": 1.7542726455465616e-05, "loss": 1.2566, "step": 10046 }, { "epoch": 25.1175, "grad_norm": 7.3125, "learning_rate": 1.754219482036972e-05, "loss": 1.3451, "step": 10047 }, { "epoch": 25.12, "grad_norm": 7.78125, "learning_rate": 1.7541663135827493e-05, "loss": 1.4113, "step": 10048 }, { "epoch": 25.1225, "grad_norm": 17.375, "learning_rate": 1.7541131401842427e-05, "loss": 1.3257, "step": 10049 }, { "epoch": 25.125, "grad_norm": 7.625, "learning_rate": 1.754059961841801e-05, "loss": 1.5421, "step": 10050 }, { "epoch": 25.1275, "grad_norm": 7.75, "learning_rate": 1.7540067785557716e-05, "loss": 1.4931, "step": 10051 }, { "epoch": 25.13, "grad_norm": 7.75, "learning_rate": 1.753953590326504e-05, "loss": 1.481, "step": 10052 }, { "epoch": 25.1325, "grad_norm": 5.9375, "learning_rate": 1.753900397154347e-05, "loss": 1.2153, "step": 10053 }, { "epoch": 25.135, "grad_norm": 7.875, "learning_rate": 1.753847199039649e-05, "loss": 1.4051, "step": 10054 }, { "epoch": 25.1375, "grad_norm": 7.84375, "learning_rate": 1.7537939959827595e-05, "loss": 1.5006, "step": 10055 }, { "epoch": 25.14, "grad_norm": 4.59375, "learning_rate": 1.7537407879840266e-05, "loss": 1.1927, "step": 10056 }, { "epoch": 25.1425, "grad_norm": 8.125, "learning_rate": 1.7536875750437987e-05, "loss": 1.4176, "step": 10057 }, { "epoch": 25.145, "grad_norm": 7.9375, "learning_rate": 1.7536343571624258e-05, "loss": 1.4415, "step": 10058 }, { "epoch": 25.1475, "grad_norm": 5.96875, "learning_rate": 1.753581134340256e-05, "loss": 1.4238, "step": 10059 }, { "epoch": 25.15, "grad_norm": 6.96875, "learning_rate": 1.753527906577638e-05, "loss": 1.3079, "step": 10060 }, { "epoch": 25.1525, "grad_norm": 7.4375, "learning_rate": 1.7534746738749217e-05, "loss": 1.5064, "step": 10061 }, { "epoch": 25.155, "grad_norm": 5.375, "learning_rate": 1.7534214362324554e-05, "loss": 1.3254, "step": 10062 }, { "epoch": 25.1575, "grad_norm": 5.1875, "learning_rate": 1.7533681936505887e-05, "loss": 1.2884, "step": 10063 }, { "epoch": 25.16, "grad_norm": 5.25, "learning_rate": 1.75331494612967e-05, "loss": 1.5997, "step": 10064 }, { "epoch": 25.1625, "grad_norm": 5.34375, "learning_rate": 1.7532616936700484e-05, "loss": 1.3358, "step": 10065 }, { "epoch": 25.165, "grad_norm": 10.6875, "learning_rate": 1.7532084362720735e-05, "loss": 1.4279, "step": 10066 }, { "epoch": 25.1675, "grad_norm": 5.28125, "learning_rate": 1.7531551739360944e-05, "loss": 1.2686, "step": 10067 }, { "epoch": 25.17, "grad_norm": 6.46875, "learning_rate": 1.75310190666246e-05, "loss": 1.4399, "step": 10068 }, { "epoch": 25.1725, "grad_norm": 6.9375, "learning_rate": 1.7530486344515197e-05, "loss": 1.3973, "step": 10069 }, { "epoch": 25.175, "grad_norm": 8.3125, "learning_rate": 1.7529953573036225e-05, "loss": 1.406, "step": 10070 }, { "epoch": 25.1775, "grad_norm": 8.5625, "learning_rate": 1.7529420752191176e-05, "loss": 1.4861, "step": 10071 }, { "epoch": 25.18, "grad_norm": 10.1875, "learning_rate": 1.752888788198355e-05, "loss": 1.4654, "step": 10072 }, { "epoch": 25.1825, "grad_norm": 8.375, "learning_rate": 1.7528354962416836e-05, "loss": 1.1977, "step": 10073 }, { "epoch": 25.185, "grad_norm": 4.90625, "learning_rate": 1.7527821993494528e-05, "loss": 1.2138, "step": 10074 }, { "epoch": 25.1875, "grad_norm": 6.8125, "learning_rate": 1.7527288975220123e-05, "loss": 1.4517, "step": 10075 }, { "epoch": 25.19, "grad_norm": 6.78125, "learning_rate": 1.752675590759711e-05, "loss": 1.3578, "step": 10076 }, { "epoch": 25.1925, "grad_norm": 8.9375, "learning_rate": 1.7526222790628985e-05, "loss": 1.6112, "step": 10077 }, { "epoch": 25.195, "grad_norm": 7.0, "learning_rate": 1.7525689624319248e-05, "loss": 1.4602, "step": 10078 }, { "epoch": 25.1975, "grad_norm": 6.25, "learning_rate": 1.752515640867139e-05, "loss": 1.5397, "step": 10079 }, { "epoch": 25.2, "grad_norm": 6.875, "learning_rate": 1.7524623143688905e-05, "loss": 1.3486, "step": 10080 }, { "epoch": 25.2025, "grad_norm": 5.875, "learning_rate": 1.7524089829375292e-05, "loss": 1.4107, "step": 10081 }, { "epoch": 25.205, "grad_norm": 8.5, "learning_rate": 1.7523556465734048e-05, "loss": 1.4661, "step": 10082 }, { "epoch": 25.2075, "grad_norm": 7.4375, "learning_rate": 1.752302305276867e-05, "loss": 1.5901, "step": 10083 }, { "epoch": 25.21, "grad_norm": 9.25, "learning_rate": 1.7522489590482653e-05, "loss": 1.3697, "step": 10084 }, { "epoch": 25.2125, "grad_norm": 8.875, "learning_rate": 1.7521956078879495e-05, "loss": 1.3829, "step": 10085 }, { "epoch": 25.215, "grad_norm": 9.125, "learning_rate": 1.7521422517962695e-05, "loss": 1.5417, "step": 10086 }, { "epoch": 25.2175, "grad_norm": 6.1875, "learning_rate": 1.752088890773575e-05, "loss": 1.387, "step": 10087 }, { "epoch": 25.22, "grad_norm": 6.21875, "learning_rate": 1.7520355248202157e-05, "loss": 1.4534, "step": 10088 }, { "epoch": 25.2225, "grad_norm": 6.0625, "learning_rate": 1.751982153936542e-05, "loss": 1.4749, "step": 10089 }, { "epoch": 25.225, "grad_norm": 6.125, "learning_rate": 1.7519287781229032e-05, "loss": 1.4152, "step": 10090 }, { "epoch": 25.2275, "grad_norm": 8.625, "learning_rate": 1.751875397379649e-05, "loss": 1.3296, "step": 10091 }, { "epoch": 25.23, "grad_norm": 5.15625, "learning_rate": 1.75182201170713e-05, "loss": 1.4158, "step": 10092 }, { "epoch": 25.2325, "grad_norm": 8.9375, "learning_rate": 1.751768621105696e-05, "loss": 1.4459, "step": 10093 }, { "epoch": 25.235, "grad_norm": 4.96875, "learning_rate": 1.751715225575697e-05, "loss": 1.4352, "step": 10094 }, { "epoch": 25.2375, "grad_norm": 5.75, "learning_rate": 1.7516618251174834e-05, "loss": 1.3469, "step": 10095 }, { "epoch": 25.24, "grad_norm": 8.1875, "learning_rate": 1.7516084197314044e-05, "loss": 1.3276, "step": 10096 }, { "epoch": 25.2425, "grad_norm": 7.9375, "learning_rate": 1.7515550094178112e-05, "loss": 1.3139, "step": 10097 }, { "epoch": 25.245, "grad_norm": 6.375, "learning_rate": 1.751501594177053e-05, "loss": 1.3057, "step": 10098 }, { "epoch": 25.2475, "grad_norm": 4.59375, "learning_rate": 1.7514481740094805e-05, "loss": 1.2592, "step": 10099 }, { "epoch": 25.25, "grad_norm": 7.75, "learning_rate": 1.7513947489154443e-05, "loss": 1.3485, "step": 10100 }, { "epoch": 25.2525, "grad_norm": 6.5, "learning_rate": 1.7513413188952938e-05, "loss": 1.5221, "step": 10101 }, { "epoch": 25.255, "grad_norm": 5.9375, "learning_rate": 1.7512878839493798e-05, "loss": 1.3742, "step": 10102 }, { "epoch": 25.2575, "grad_norm": 10.25, "learning_rate": 1.7512344440780525e-05, "loss": 1.4787, "step": 10103 }, { "epoch": 25.26, "grad_norm": 10.5, "learning_rate": 1.751180999281662e-05, "loss": 1.2956, "step": 10104 }, { "epoch": 25.2625, "grad_norm": 7.5625, "learning_rate": 1.7511275495605592e-05, "loss": 1.3572, "step": 10105 }, { "epoch": 25.265, "grad_norm": 11.3125, "learning_rate": 1.7510740949150942e-05, "loss": 1.3506, "step": 10106 }, { "epoch": 25.2675, "grad_norm": 11.0, "learning_rate": 1.7510206353456173e-05, "loss": 1.4351, "step": 10107 }, { "epoch": 25.27, "grad_norm": 13.5, "learning_rate": 1.750967170852479e-05, "loss": 1.276, "step": 10108 }, { "epoch": 25.2725, "grad_norm": 7.65625, "learning_rate": 1.7509137014360305e-05, "loss": 1.4432, "step": 10109 }, { "epoch": 25.275, "grad_norm": 6.78125, "learning_rate": 1.750860227096622e-05, "loss": 1.3519, "step": 10110 }, { "epoch": 25.2775, "grad_norm": 7.125, "learning_rate": 1.750806747834603e-05, "loss": 1.4106, "step": 10111 }, { "epoch": 25.28, "grad_norm": 7.125, "learning_rate": 1.7507532636503256e-05, "loss": 1.3885, "step": 10112 }, { "epoch": 25.2825, "grad_norm": 7.09375, "learning_rate": 1.75069977454414e-05, "loss": 1.4433, "step": 10113 }, { "epoch": 25.285, "grad_norm": 6.28125, "learning_rate": 1.750646280516396e-05, "loss": 1.3386, "step": 10114 }, { "epoch": 25.2875, "grad_norm": 6.46875, "learning_rate": 1.7505927815674458e-05, "loss": 1.4967, "step": 10115 }, { "epoch": 25.29, "grad_norm": 6.34375, "learning_rate": 1.7505392776976392e-05, "loss": 1.3434, "step": 10116 }, { "epoch": 25.2925, "grad_norm": 6.09375, "learning_rate": 1.7504857689073268e-05, "loss": 1.2609, "step": 10117 }, { "epoch": 25.295, "grad_norm": 8.5, "learning_rate": 1.75043225519686e-05, "loss": 1.4646, "step": 10118 }, { "epoch": 25.2975, "grad_norm": 5.21875, "learning_rate": 1.7503787365665893e-05, "loss": 1.3033, "step": 10119 }, { "epoch": 25.3, "grad_norm": 5.875, "learning_rate": 1.7503252130168657e-05, "loss": 1.2677, "step": 10120 }, { "epoch": 25.3025, "grad_norm": 7.0625, "learning_rate": 1.7502716845480398e-05, "loss": 1.3498, "step": 10121 }, { "epoch": 25.305, "grad_norm": 9.9375, "learning_rate": 1.750218151160463e-05, "loss": 1.3996, "step": 10122 }, { "epoch": 25.3075, "grad_norm": 7.34375, "learning_rate": 1.7501646128544857e-05, "loss": 1.4556, "step": 10123 }, { "epoch": 25.31, "grad_norm": 7.59375, "learning_rate": 1.7501110696304598e-05, "loss": 1.3849, "step": 10124 }, { "epoch": 25.3125, "grad_norm": 7.34375, "learning_rate": 1.7500575214887354e-05, "loss": 1.4784, "step": 10125 }, { "epoch": 25.315, "grad_norm": 10.25, "learning_rate": 1.7500039684296636e-05, "loss": 1.3889, "step": 10126 }, { "epoch": 25.3175, "grad_norm": 8.3125, "learning_rate": 1.7499504104535963e-05, "loss": 1.4127, "step": 10127 }, { "epoch": 25.32, "grad_norm": 6.53125, "learning_rate": 1.749896847560884e-05, "loss": 1.3469, "step": 10128 }, { "epoch": 25.3225, "grad_norm": 7.5625, "learning_rate": 1.749843279751878e-05, "loss": 1.3888, "step": 10129 }, { "epoch": 25.325, "grad_norm": 8.3125, "learning_rate": 1.7497897070269293e-05, "loss": 1.2728, "step": 10130 }, { "epoch": 25.3275, "grad_norm": 7.90625, "learning_rate": 1.749736129386389e-05, "loss": 1.3656, "step": 10131 }, { "epoch": 25.33, "grad_norm": 5.90625, "learning_rate": 1.7496825468306092e-05, "loss": 1.2931, "step": 10132 }, { "epoch": 25.3325, "grad_norm": 7.75, "learning_rate": 1.7496289593599406e-05, "loss": 1.2232, "step": 10133 }, { "epoch": 25.335, "grad_norm": 8.875, "learning_rate": 1.7495753669747343e-05, "loss": 1.4109, "step": 10134 }, { "epoch": 25.3375, "grad_norm": 5.25, "learning_rate": 1.749521769675342e-05, "loss": 1.4525, "step": 10135 }, { "epoch": 25.34, "grad_norm": 7.1875, "learning_rate": 1.7494681674621148e-05, "loss": 1.4291, "step": 10136 }, { "epoch": 25.3425, "grad_norm": 6.15625, "learning_rate": 1.7494145603354042e-05, "loss": 1.4515, "step": 10137 }, { "epoch": 25.345, "grad_norm": 6.8125, "learning_rate": 1.749360948295562e-05, "loss": 1.3212, "step": 10138 }, { "epoch": 25.3475, "grad_norm": 7.375, "learning_rate": 1.7493073313429394e-05, "loss": 1.344, "step": 10139 }, { "epoch": 25.35, "grad_norm": 7.46875, "learning_rate": 1.749253709477888e-05, "loss": 1.4682, "step": 10140 }, { "epoch": 25.3525, "grad_norm": 6.1875, "learning_rate": 1.749200082700759e-05, "loss": 1.4212, "step": 10141 }, { "epoch": 25.355, "grad_norm": 9.25, "learning_rate": 1.7491464510119045e-05, "loss": 1.3158, "step": 10142 }, { "epoch": 25.3575, "grad_norm": 5.4375, "learning_rate": 1.7490928144116756e-05, "loss": 1.4055, "step": 10143 }, { "epoch": 25.36, "grad_norm": 7.9375, "learning_rate": 1.7490391729004242e-05, "loss": 1.4906, "step": 10144 }, { "epoch": 25.3625, "grad_norm": 6.34375, "learning_rate": 1.748985526478502e-05, "loss": 1.4476, "step": 10145 }, { "epoch": 25.365, "grad_norm": 9.625, "learning_rate": 1.748931875146261e-05, "loss": 1.6576, "step": 10146 }, { "epoch": 25.3675, "grad_norm": 8.125, "learning_rate": 1.7488782189040524e-05, "loss": 1.3419, "step": 10147 }, { "epoch": 25.37, "grad_norm": 5.8125, "learning_rate": 1.748824557752228e-05, "loss": 1.4046, "step": 10148 }, { "epoch": 25.3725, "grad_norm": 7.5625, "learning_rate": 1.7487708916911397e-05, "loss": 1.3711, "step": 10149 }, { "epoch": 25.375, "grad_norm": 6.375, "learning_rate": 1.7487172207211395e-05, "loss": 1.4051, "step": 10150 }, { "epoch": 25.3775, "grad_norm": 7.03125, "learning_rate": 1.7486635448425794e-05, "loss": 1.3855, "step": 10151 }, { "epoch": 25.38, "grad_norm": 10.375, "learning_rate": 1.7486098640558105e-05, "loss": 1.6563, "step": 10152 }, { "epoch": 25.3825, "grad_norm": 6.4375, "learning_rate": 1.748556178361186e-05, "loss": 1.2478, "step": 10153 }, { "epoch": 25.385, "grad_norm": 6.9375, "learning_rate": 1.7485024877590567e-05, "loss": 1.185, "step": 10154 }, { "epoch": 25.3875, "grad_norm": 8.375, "learning_rate": 1.748448792249775e-05, "loss": 1.6084, "step": 10155 }, { "epoch": 25.39, "grad_norm": 6.0, "learning_rate": 1.7483950918336933e-05, "loss": 1.31, "step": 10156 }, { "epoch": 25.3925, "grad_norm": 6.125, "learning_rate": 1.748341386511163e-05, "loss": 1.2905, "step": 10157 }, { "epoch": 25.395, "grad_norm": 9.4375, "learning_rate": 1.7482876762825366e-05, "loss": 1.595, "step": 10158 }, { "epoch": 25.3975, "grad_norm": 6.71875, "learning_rate": 1.7482339611481664e-05, "loss": 1.2764, "step": 10159 }, { "epoch": 25.4, "grad_norm": 6.1875, "learning_rate": 1.748180241108404e-05, "loss": 1.3296, "step": 10160 }, { "epoch": 25.4025, "grad_norm": 5.78125, "learning_rate": 1.7481265161636023e-05, "loss": 1.3637, "step": 10161 }, { "epoch": 25.405, "grad_norm": 6.65625, "learning_rate": 1.7480727863141126e-05, "loss": 1.4013, "step": 10162 }, { "epoch": 25.4075, "grad_norm": 7.65625, "learning_rate": 1.748019051560288e-05, "loss": 1.4142, "step": 10163 }, { "epoch": 25.41, "grad_norm": 11.1875, "learning_rate": 1.74796531190248e-05, "loss": 1.5205, "step": 10164 }, { "epoch": 25.4125, "grad_norm": 23.875, "learning_rate": 1.7479115673410415e-05, "loss": 1.6383, "step": 10165 }, { "epoch": 25.415, "grad_norm": 7.0625, "learning_rate": 1.747857817876325e-05, "loss": 1.3942, "step": 10166 }, { "epoch": 25.4175, "grad_norm": 16.75, "learning_rate": 1.7478040635086822e-05, "loss": 1.3934, "step": 10167 }, { "epoch": 25.42, "grad_norm": 7.84375, "learning_rate": 1.747750304238466e-05, "loss": 1.4463, "step": 10168 }, { "epoch": 25.4225, "grad_norm": 7.1875, "learning_rate": 1.747696540066029e-05, "loss": 1.3854, "step": 10169 }, { "epoch": 25.425, "grad_norm": 8.9375, "learning_rate": 1.747642770991723e-05, "loss": 1.2779, "step": 10170 }, { "epoch": 25.4275, "grad_norm": 4.90625, "learning_rate": 1.747588997015901e-05, "loss": 1.3202, "step": 10171 }, { "epoch": 25.43, "grad_norm": 7.40625, "learning_rate": 1.747535218138916e-05, "loss": 1.5483, "step": 10172 }, { "epoch": 25.4325, "grad_norm": 8.125, "learning_rate": 1.7474814343611193e-05, "loss": 1.2216, "step": 10173 }, { "epoch": 25.435, "grad_norm": 4.96875, "learning_rate": 1.7474276456828647e-05, "loss": 1.4109, "step": 10174 }, { "epoch": 25.4375, "grad_norm": 4.5, "learning_rate": 1.7473738521045042e-05, "loss": 1.1548, "step": 10175 }, { "epoch": 25.44, "grad_norm": 7.0, "learning_rate": 1.7473200536263905e-05, "loss": 1.3989, "step": 10176 }, { "epoch": 25.4425, "grad_norm": 12.3125, "learning_rate": 1.7472662502488766e-05, "loss": 1.5804, "step": 10177 }, { "epoch": 25.445, "grad_norm": 6.40625, "learning_rate": 1.7472124419723152e-05, "loss": 1.4299, "step": 10178 }, { "epoch": 25.4475, "grad_norm": 6.21875, "learning_rate": 1.7471586287970584e-05, "loss": 1.4295, "step": 10179 }, { "epoch": 25.45, "grad_norm": 7.9375, "learning_rate": 1.74710481072346e-05, "loss": 1.5673, "step": 10180 }, { "epoch": 25.4525, "grad_norm": 6.96875, "learning_rate": 1.7470509877518723e-05, "loss": 1.407, "step": 10181 }, { "epoch": 25.455, "grad_norm": 6.6875, "learning_rate": 1.746997159882648e-05, "loss": 1.4507, "step": 10182 }, { "epoch": 25.4575, "grad_norm": 8.0, "learning_rate": 1.7469433271161406e-05, "loss": 1.3836, "step": 10183 }, { "epoch": 25.46, "grad_norm": 8.1875, "learning_rate": 1.746889489452702e-05, "loss": 1.3026, "step": 10184 }, { "epoch": 25.4625, "grad_norm": 5.125, "learning_rate": 1.7468356468926865e-05, "loss": 1.5271, "step": 10185 }, { "epoch": 25.465, "grad_norm": 8.4375, "learning_rate": 1.746781799436446e-05, "loss": 1.4423, "step": 10186 }, { "epoch": 25.4675, "grad_norm": 7.21875, "learning_rate": 1.746727947084334e-05, "loss": 1.3241, "step": 10187 }, { "epoch": 25.47, "grad_norm": 7.03125, "learning_rate": 1.7466740898367033e-05, "loss": 1.3562, "step": 10188 }, { "epoch": 25.4725, "grad_norm": 7.375, "learning_rate": 1.7466202276939077e-05, "loss": 1.677, "step": 10189 }, { "epoch": 25.475, "grad_norm": 11.6875, "learning_rate": 1.746566360656299e-05, "loss": 1.4035, "step": 10190 }, { "epoch": 25.4775, "grad_norm": 4.78125, "learning_rate": 1.7465124887242315e-05, "loss": 1.437, "step": 10191 }, { "epoch": 25.48, "grad_norm": 9.75, "learning_rate": 1.746458611898058e-05, "loss": 1.3204, "step": 10192 }, { "epoch": 25.4825, "grad_norm": 7.75, "learning_rate": 1.7464047301781318e-05, "loss": 1.3457, "step": 10193 }, { "epoch": 25.485, "grad_norm": 8.6875, "learning_rate": 1.7463508435648057e-05, "loss": 1.3229, "step": 10194 }, { "epoch": 25.4875, "grad_norm": 7.1875, "learning_rate": 1.7462969520584336e-05, "loss": 1.3157, "step": 10195 }, { "epoch": 25.49, "grad_norm": 7.625, "learning_rate": 1.7462430556593687e-05, "loss": 1.4094, "step": 10196 }, { "epoch": 25.4925, "grad_norm": 8.125, "learning_rate": 1.7461891543679642e-05, "loss": 1.3545, "step": 10197 }, { "epoch": 25.495, "grad_norm": 9.8125, "learning_rate": 1.746135248184573e-05, "loss": 1.4696, "step": 10198 }, { "epoch": 25.4975, "grad_norm": 7.4375, "learning_rate": 1.7460813371095495e-05, "loss": 1.3248, "step": 10199 }, { "epoch": 25.5, "grad_norm": 8.5625, "learning_rate": 1.7460274211432463e-05, "loss": 1.3893, "step": 10200 }, { "epoch": 25.5025, "grad_norm": 5.625, "learning_rate": 1.745973500286017e-05, "loss": 1.2493, "step": 10201 }, { "epoch": 25.505, "grad_norm": 7.15625, "learning_rate": 1.7459195745382156e-05, "loss": 1.4883, "step": 10202 }, { "epoch": 25.5075, "grad_norm": 7.9375, "learning_rate": 1.7458656439001953e-05, "loss": 1.3107, "step": 10203 }, { "epoch": 25.51, "grad_norm": 7.375, "learning_rate": 1.7458117083723096e-05, "loss": 1.1634, "step": 10204 }, { "epoch": 25.5125, "grad_norm": 8.9375, "learning_rate": 1.7457577679549123e-05, "loss": 1.3971, "step": 10205 }, { "epoch": 25.515, "grad_norm": 6.46875, "learning_rate": 1.7457038226483565e-05, "loss": 1.2114, "step": 10206 }, { "epoch": 25.5175, "grad_norm": 8.625, "learning_rate": 1.745649872452997e-05, "loss": 1.417, "step": 10207 }, { "epoch": 25.52, "grad_norm": 5.5, "learning_rate": 1.7455959173691863e-05, "loss": 1.2602, "step": 10208 }, { "epoch": 25.5225, "grad_norm": 7.53125, "learning_rate": 1.7455419573972786e-05, "loss": 1.4388, "step": 10209 }, { "epoch": 25.525, "grad_norm": 11.6875, "learning_rate": 1.745487992537628e-05, "loss": 1.3897, "step": 10210 }, { "epoch": 25.5275, "grad_norm": 5.5, "learning_rate": 1.7454340227905876e-05, "loss": 1.3736, "step": 10211 }, { "epoch": 25.53, "grad_norm": 6.8125, "learning_rate": 1.7453800481565118e-05, "loss": 1.2594, "step": 10212 }, { "epoch": 25.5325, "grad_norm": 9.6875, "learning_rate": 1.745326068635754e-05, "loss": 1.3909, "step": 10213 }, { "epoch": 25.535, "grad_norm": 7.28125, "learning_rate": 1.7452720842286684e-05, "loss": 1.2425, "step": 10214 }, { "epoch": 25.5375, "grad_norm": 8.0, "learning_rate": 1.7452180949356086e-05, "loss": 1.4224, "step": 10215 }, { "epoch": 25.54, "grad_norm": 6.71875, "learning_rate": 1.7451641007569296e-05, "loss": 1.242, "step": 10216 }, { "epoch": 25.5425, "grad_norm": 9.5625, "learning_rate": 1.745110101692984e-05, "loss": 1.6545, "step": 10217 }, { "epoch": 25.545, "grad_norm": 7.34375, "learning_rate": 1.745056097744126e-05, "loss": 1.3971, "step": 10218 }, { "epoch": 25.5475, "grad_norm": 10.0625, "learning_rate": 1.745002088910711e-05, "loss": 1.4212, "step": 10219 }, { "epoch": 25.55, "grad_norm": 9.3125, "learning_rate": 1.7449480751930915e-05, "loss": 1.3678, "step": 10220 }, { "epoch": 25.5525, "grad_norm": 7.5, "learning_rate": 1.744894056591622e-05, "loss": 1.4443, "step": 10221 }, { "epoch": 25.555, "grad_norm": 8.3125, "learning_rate": 1.7448400331066575e-05, "loss": 1.3173, "step": 10222 }, { "epoch": 25.5575, "grad_norm": 7.40625, "learning_rate": 1.7447860047385513e-05, "loss": 1.4167, "step": 10223 }, { "epoch": 25.56, "grad_norm": 8.375, "learning_rate": 1.744731971487658e-05, "loss": 1.3381, "step": 10224 }, { "epoch": 25.5625, "grad_norm": 4.8125, "learning_rate": 1.7446779333543312e-05, "loss": 1.2889, "step": 10225 }, { "epoch": 25.565, "grad_norm": 6.1875, "learning_rate": 1.7446238903389258e-05, "loss": 1.4408, "step": 10226 }, { "epoch": 25.5675, "grad_norm": 7.75, "learning_rate": 1.7445698424417963e-05, "loss": 1.344, "step": 10227 }, { "epoch": 25.57, "grad_norm": 6.03125, "learning_rate": 1.7445157896632967e-05, "loss": 1.4606, "step": 10228 }, { "epoch": 25.5725, "grad_norm": 6.0625, "learning_rate": 1.744461732003781e-05, "loss": 1.4953, "step": 10229 }, { "epoch": 25.575, "grad_norm": 7.78125, "learning_rate": 1.7444076694636042e-05, "loss": 1.283, "step": 10230 }, { "epoch": 25.5775, "grad_norm": 7.71875, "learning_rate": 1.7443536020431206e-05, "loss": 1.4026, "step": 10231 }, { "epoch": 25.58, "grad_norm": 6.1875, "learning_rate": 1.7442995297426846e-05, "loss": 1.3403, "step": 10232 }, { "epoch": 25.5825, "grad_norm": 5.8125, "learning_rate": 1.7442454525626504e-05, "loss": 1.3969, "step": 10233 }, { "epoch": 25.585, "grad_norm": 6.09375, "learning_rate": 1.744191370503373e-05, "loss": 1.2682, "step": 10234 }, { "epoch": 25.5875, "grad_norm": 7.15625, "learning_rate": 1.744137283565207e-05, "loss": 1.4021, "step": 10235 }, { "epoch": 25.59, "grad_norm": 8.0625, "learning_rate": 1.7440831917485064e-05, "loss": 1.6113, "step": 10236 }, { "epoch": 25.5925, "grad_norm": 6.53125, "learning_rate": 1.744029095053626e-05, "loss": 1.4355, "step": 10237 }, { "epoch": 25.595, "grad_norm": 5.90625, "learning_rate": 1.743974993480921e-05, "loss": 1.3302, "step": 10238 }, { "epoch": 25.5975, "grad_norm": 7.65625, "learning_rate": 1.7439208870307457e-05, "loss": 1.4532, "step": 10239 }, { "epoch": 25.6, "grad_norm": 5.0625, "learning_rate": 1.7438667757034547e-05, "loss": 1.2854, "step": 10240 }, { "epoch": 25.6025, "grad_norm": 10.0, "learning_rate": 1.7438126594994028e-05, "loss": 1.3879, "step": 10241 }, { "epoch": 25.605, "grad_norm": 7.8125, "learning_rate": 1.7437585384189448e-05, "loss": 1.4865, "step": 10242 }, { "epoch": 25.6075, "grad_norm": 6.96875, "learning_rate": 1.7437044124624357e-05, "loss": 1.3683, "step": 10243 }, { "epoch": 25.61, "grad_norm": 6.125, "learning_rate": 1.7436502816302304e-05, "loss": 1.312, "step": 10244 }, { "epoch": 25.6125, "grad_norm": 6.0, "learning_rate": 1.7435961459226834e-05, "loss": 1.2643, "step": 10245 }, { "epoch": 25.615, "grad_norm": 8.5625, "learning_rate": 1.74354200534015e-05, "loss": 1.3992, "step": 10246 }, { "epoch": 25.6175, "grad_norm": 5.28125, "learning_rate": 1.743487859882985e-05, "loss": 1.3115, "step": 10247 }, { "epoch": 25.62, "grad_norm": 7.5, "learning_rate": 1.743433709551543e-05, "loss": 1.5374, "step": 10248 }, { "epoch": 25.6225, "grad_norm": 5.53125, "learning_rate": 1.74337955434618e-05, "loss": 1.2413, "step": 10249 }, { "epoch": 25.625, "grad_norm": 6.375, "learning_rate": 1.7433253942672497e-05, "loss": 1.2642, "step": 10250 }, { "epoch": 25.6275, "grad_norm": 9.3125, "learning_rate": 1.7432712293151082e-05, "loss": 1.3785, "step": 10251 }, { "epoch": 25.63, "grad_norm": 8.6875, "learning_rate": 1.7432170594901102e-05, "loss": 1.3407, "step": 10252 }, { "epoch": 25.6325, "grad_norm": 7.6875, "learning_rate": 1.7431628847926107e-05, "loss": 1.3563, "step": 10253 }, { "epoch": 25.635, "grad_norm": 7.40625, "learning_rate": 1.7431087052229655e-05, "loss": 1.4424, "step": 10254 }, { "epoch": 25.6375, "grad_norm": 7.5625, "learning_rate": 1.743054520781529e-05, "loss": 1.2128, "step": 10255 }, { "epoch": 25.64, "grad_norm": 8.3125, "learning_rate": 1.743000331468657e-05, "loss": 1.4611, "step": 10256 }, { "epoch": 25.6425, "grad_norm": 5.65625, "learning_rate": 1.7429461372847044e-05, "loss": 1.3099, "step": 10257 }, { "epoch": 25.645, "grad_norm": 6.90625, "learning_rate": 1.7428919382300268e-05, "loss": 1.3321, "step": 10258 }, { "epoch": 25.6475, "grad_norm": 6.65625, "learning_rate": 1.7428377343049795e-05, "loss": 1.2471, "step": 10259 }, { "epoch": 25.65, "grad_norm": 7.5625, "learning_rate": 1.7427835255099173e-05, "loss": 1.5404, "step": 10260 }, { "epoch": 25.6525, "grad_norm": 8.9375, "learning_rate": 1.7427293118451963e-05, "loss": 1.3106, "step": 10261 }, { "epoch": 25.655, "grad_norm": 7.0, "learning_rate": 1.742675093311172e-05, "loss": 1.3117, "step": 10262 }, { "epoch": 25.6575, "grad_norm": 7.21875, "learning_rate": 1.742620869908199e-05, "loss": 1.2582, "step": 10263 }, { "epoch": 25.66, "grad_norm": 9.0, "learning_rate": 1.7425666416366335e-05, "loss": 1.4379, "step": 10264 }, { "epoch": 25.6625, "grad_norm": 5.15625, "learning_rate": 1.7425124084968307e-05, "loss": 1.1917, "step": 10265 }, { "epoch": 25.665, "grad_norm": 8.5, "learning_rate": 1.742458170489146e-05, "loss": 1.4571, "step": 10266 }, { "epoch": 25.6675, "grad_norm": 7.8125, "learning_rate": 1.742403927613936e-05, "loss": 1.4017, "step": 10267 }, { "epoch": 25.67, "grad_norm": 7.125, "learning_rate": 1.742349679871555e-05, "loss": 1.4492, "step": 10268 }, { "epoch": 25.6725, "grad_norm": 4.71875, "learning_rate": 1.7422954272623595e-05, "loss": 1.4109, "step": 10269 }, { "epoch": 25.675, "grad_norm": 5.40625, "learning_rate": 1.7422411697867045e-05, "loss": 1.3913, "step": 10270 }, { "epoch": 25.6775, "grad_norm": 5.5625, "learning_rate": 1.7421869074449465e-05, "loss": 1.5483, "step": 10271 }, { "epoch": 25.68, "grad_norm": 8.0, "learning_rate": 1.7421326402374406e-05, "loss": 1.3673, "step": 10272 }, { "epoch": 25.6825, "grad_norm": 6.25, "learning_rate": 1.742078368164543e-05, "loss": 1.4428, "step": 10273 }, { "epoch": 25.685, "grad_norm": 5.4375, "learning_rate": 1.7420240912266088e-05, "loss": 1.4943, "step": 10274 }, { "epoch": 25.6875, "grad_norm": 7.59375, "learning_rate": 1.7419698094239947e-05, "loss": 1.3745, "step": 10275 }, { "epoch": 25.69, "grad_norm": 7.15625, "learning_rate": 1.7419155227570564e-05, "loss": 1.5082, "step": 10276 }, { "epoch": 25.6925, "grad_norm": 12.4375, "learning_rate": 1.7418612312261497e-05, "loss": 1.417, "step": 10277 }, { "epoch": 25.695, "grad_norm": 9.375, "learning_rate": 1.74180693483163e-05, "loss": 1.467, "step": 10278 }, { "epoch": 25.6975, "grad_norm": 6.09375, "learning_rate": 1.741752633573854e-05, "loss": 1.2169, "step": 10279 }, { "epoch": 25.7, "grad_norm": 5.8125, "learning_rate": 1.7416983274531777e-05, "loss": 1.3627, "step": 10280 }, { "epoch": 25.7025, "grad_norm": 8.0, "learning_rate": 1.741644016469956e-05, "loss": 1.4111, "step": 10281 }, { "epoch": 25.705, "grad_norm": 8.125, "learning_rate": 1.7415897006245464e-05, "loss": 1.3826, "step": 10282 }, { "epoch": 25.7075, "grad_norm": 6.90625, "learning_rate": 1.7415353799173045e-05, "loss": 1.3882, "step": 10283 }, { "epoch": 25.71, "grad_norm": 8.75, "learning_rate": 1.7414810543485862e-05, "loss": 1.2867, "step": 10284 }, { "epoch": 25.7125, "grad_norm": 13.6875, "learning_rate": 1.7414267239187478e-05, "loss": 1.6255, "step": 10285 }, { "epoch": 25.715, "grad_norm": 6.34375, "learning_rate": 1.7413723886281454e-05, "loss": 1.3951, "step": 10286 }, { "epoch": 25.7175, "grad_norm": 8.875, "learning_rate": 1.7413180484771352e-05, "loss": 1.4043, "step": 10287 }, { "epoch": 25.72, "grad_norm": 5.5625, "learning_rate": 1.7412637034660735e-05, "loss": 1.34, "step": 10288 }, { "epoch": 25.7225, "grad_norm": 8.5, "learning_rate": 1.741209353595317e-05, "loss": 1.2568, "step": 10289 }, { "epoch": 25.725, "grad_norm": 8.1875, "learning_rate": 1.7411549988652213e-05, "loss": 1.2777, "step": 10290 }, { "epoch": 25.7275, "grad_norm": 6.0625, "learning_rate": 1.7411006392761432e-05, "loss": 1.4756, "step": 10291 }, { "epoch": 25.73, "grad_norm": 7.25, "learning_rate": 1.7410462748284387e-05, "loss": 1.3048, "step": 10292 }, { "epoch": 25.7325, "grad_norm": 6.78125, "learning_rate": 1.740991905522465e-05, "loss": 1.3048, "step": 10293 }, { "epoch": 25.735, "grad_norm": 6.4375, "learning_rate": 1.7409375313585774e-05, "loss": 1.5822, "step": 10294 }, { "epoch": 25.7375, "grad_norm": 5.40625, "learning_rate": 1.7408831523371334e-05, "loss": 1.4261, "step": 10295 }, { "epoch": 25.74, "grad_norm": 8.3125, "learning_rate": 1.740828768458489e-05, "loss": 1.2953, "step": 10296 }, { "epoch": 25.7425, "grad_norm": 7.21875, "learning_rate": 1.740774379723001e-05, "loss": 1.4513, "step": 10297 }, { "epoch": 25.745, "grad_norm": 7.03125, "learning_rate": 1.740719986131026e-05, "loss": 1.4052, "step": 10298 }, { "epoch": 25.7475, "grad_norm": 6.1875, "learning_rate": 1.7406655876829195e-05, "loss": 1.3426, "step": 10299 }, { "epoch": 25.75, "grad_norm": 5.5625, "learning_rate": 1.74061118437904e-05, "loss": 1.5445, "step": 10300 }, { "epoch": 25.7525, "grad_norm": 9.125, "learning_rate": 1.740556776219743e-05, "loss": 1.2931, "step": 10301 }, { "epoch": 25.755, "grad_norm": 7.71875, "learning_rate": 1.740502363205385e-05, "loss": 1.3108, "step": 10302 }, { "epoch": 25.7575, "grad_norm": 8.0, "learning_rate": 1.7404479453363236e-05, "loss": 1.4537, "step": 10303 }, { "epoch": 25.76, "grad_norm": 8.5, "learning_rate": 1.740393522612915e-05, "loss": 1.3449, "step": 10304 }, { "epoch": 25.7625, "grad_norm": 7.5, "learning_rate": 1.7403390950355157e-05, "loss": 1.3511, "step": 10305 }, { "epoch": 25.765, "grad_norm": 9.6875, "learning_rate": 1.7402846626044834e-05, "loss": 1.4795, "step": 10306 }, { "epoch": 25.7675, "grad_norm": 7.1875, "learning_rate": 1.7402302253201743e-05, "loss": 1.5471, "step": 10307 }, { "epoch": 25.77, "grad_norm": 6.375, "learning_rate": 1.7401757831829454e-05, "loss": 1.3419, "step": 10308 }, { "epoch": 25.7725, "grad_norm": 7.0, "learning_rate": 1.7401213361931538e-05, "loss": 1.2739, "step": 10309 }, { "epoch": 25.775, "grad_norm": 4.90625, "learning_rate": 1.740066884351156e-05, "loss": 1.4906, "step": 10310 }, { "epoch": 25.7775, "grad_norm": 5.25, "learning_rate": 1.7400124276573098e-05, "loss": 1.6022, "step": 10311 }, { "epoch": 25.78, "grad_norm": 8.4375, "learning_rate": 1.7399579661119713e-05, "loss": 1.4063, "step": 10312 }, { "epoch": 25.7825, "grad_norm": 6.5625, "learning_rate": 1.7399034997154985e-05, "loss": 1.4981, "step": 10313 }, { "epoch": 25.785, "grad_norm": 12.75, "learning_rate": 1.739849028468248e-05, "loss": 1.4386, "step": 10314 }, { "epoch": 25.7875, "grad_norm": 7.9375, "learning_rate": 1.7397945523705765e-05, "loss": 1.2701, "step": 10315 }, { "epoch": 25.79, "grad_norm": 8.25, "learning_rate": 1.7397400714228414e-05, "loss": 1.3288, "step": 10316 }, { "epoch": 25.7925, "grad_norm": 5.75, "learning_rate": 1.7396855856254e-05, "loss": 1.4229, "step": 10317 }, { "epoch": 25.795, "grad_norm": 7.125, "learning_rate": 1.73963109497861e-05, "loss": 1.4374, "step": 10318 }, { "epoch": 25.7975, "grad_norm": 7.75, "learning_rate": 1.739576599482828e-05, "loss": 1.2476, "step": 10319 }, { "epoch": 25.8, "grad_norm": 9.375, "learning_rate": 1.739522099138411e-05, "loss": 1.4207, "step": 10320 }, { "epoch": 25.8025, "grad_norm": 8.4375, "learning_rate": 1.7394675939457167e-05, "loss": 1.4382, "step": 10321 }, { "epoch": 25.805, "grad_norm": 5.5625, "learning_rate": 1.739413083905103e-05, "loss": 1.3563, "step": 10322 }, { "epoch": 25.8075, "grad_norm": 6.875, "learning_rate": 1.7393585690169264e-05, "loss": 1.3155, "step": 10323 }, { "epoch": 25.81, "grad_norm": 6.09375, "learning_rate": 1.7393040492815445e-05, "loss": 1.2489, "step": 10324 }, { "epoch": 25.8125, "grad_norm": 7.34375, "learning_rate": 1.739249524699315e-05, "loss": 1.4586, "step": 10325 }, { "epoch": 25.815, "grad_norm": 7.5, "learning_rate": 1.739194995270595e-05, "loss": 1.3741, "step": 10326 }, { "epoch": 25.8175, "grad_norm": 5.5, "learning_rate": 1.7391404609957426e-05, "loss": 1.3347, "step": 10327 }, { "epoch": 25.82, "grad_norm": 7.1875, "learning_rate": 1.7390859218751145e-05, "loss": 1.3649, "step": 10328 }, { "epoch": 25.8225, "grad_norm": 7.25, "learning_rate": 1.7390313779090687e-05, "loss": 1.2971, "step": 10329 }, { "epoch": 25.825, "grad_norm": 9.1875, "learning_rate": 1.738976829097963e-05, "loss": 1.3547, "step": 10330 }, { "epoch": 25.8275, "grad_norm": 5.96875, "learning_rate": 1.7389222754421546e-05, "loss": 1.3115, "step": 10331 }, { "epoch": 25.83, "grad_norm": 7.5, "learning_rate": 1.7388677169420015e-05, "loss": 1.1852, "step": 10332 }, { "epoch": 25.8325, "grad_norm": 5.59375, "learning_rate": 1.7388131535978612e-05, "loss": 1.3489, "step": 10333 }, { "epoch": 25.835, "grad_norm": 8.375, "learning_rate": 1.7387585854100914e-05, "loss": 1.4889, "step": 10334 }, { "epoch": 25.8375, "grad_norm": 6.5, "learning_rate": 1.7387040123790498e-05, "loss": 1.3213, "step": 10335 }, { "epoch": 25.84, "grad_norm": 7.09375, "learning_rate": 1.7386494345050944e-05, "loss": 1.3254, "step": 10336 }, { "epoch": 25.8425, "grad_norm": 6.5625, "learning_rate": 1.7385948517885827e-05, "loss": 1.3243, "step": 10337 }, { "epoch": 25.845, "grad_norm": 6.1875, "learning_rate": 1.7385402642298728e-05, "loss": 1.3731, "step": 10338 }, { "epoch": 25.8475, "grad_norm": 8.8125, "learning_rate": 1.7384856718293223e-05, "loss": 1.3809, "step": 10339 }, { "epoch": 25.85, "grad_norm": 15.1875, "learning_rate": 1.7384310745872896e-05, "loss": 1.5919, "step": 10340 }, { "epoch": 25.8525, "grad_norm": 7.8125, "learning_rate": 1.738376472504132e-05, "loss": 1.3251, "step": 10341 }, { "epoch": 25.855, "grad_norm": 8.9375, "learning_rate": 1.7383218655802082e-05, "loss": 1.4508, "step": 10342 }, { "epoch": 25.8575, "grad_norm": 5.875, "learning_rate": 1.7382672538158753e-05, "loss": 1.5408, "step": 10343 }, { "epoch": 25.86, "grad_norm": 7.5, "learning_rate": 1.738212637211492e-05, "loss": 1.3559, "step": 10344 }, { "epoch": 25.8625, "grad_norm": 7.40625, "learning_rate": 1.738158015767416e-05, "loss": 1.3769, "step": 10345 }, { "epoch": 25.865, "grad_norm": 9.0625, "learning_rate": 1.738103389484006e-05, "loss": 1.3513, "step": 10346 }, { "epoch": 25.8675, "grad_norm": 6.34375, "learning_rate": 1.73804875836162e-05, "loss": 1.374, "step": 10347 }, { "epoch": 25.87, "grad_norm": 9.125, "learning_rate": 1.7379941224006152e-05, "loss": 1.2269, "step": 10348 }, { "epoch": 25.8725, "grad_norm": 7.21875, "learning_rate": 1.7379394816013508e-05, "loss": 1.3497, "step": 10349 }, { "epoch": 25.875, "grad_norm": 9.125, "learning_rate": 1.7378848359641846e-05, "loss": 1.4812, "step": 10350 }, { "epoch": 25.8775, "grad_norm": 4.625, "learning_rate": 1.737830185489475e-05, "loss": 1.4484, "step": 10351 }, { "epoch": 25.88, "grad_norm": 7.375, "learning_rate": 1.73777553017758e-05, "loss": 1.3953, "step": 10352 }, { "epoch": 25.8825, "grad_norm": 7.625, "learning_rate": 1.7377208700288584e-05, "loss": 1.3195, "step": 10353 }, { "epoch": 25.885, "grad_norm": 9.5, "learning_rate": 1.7376662050436684e-05, "loss": 1.3053, "step": 10354 }, { "epoch": 25.8875, "grad_norm": 6.0, "learning_rate": 1.737611535222368e-05, "loss": 1.323, "step": 10355 }, { "epoch": 25.89, "grad_norm": 22.875, "learning_rate": 1.737556860565316e-05, "loss": 1.448, "step": 10356 }, { "epoch": 25.8925, "grad_norm": 7.4375, "learning_rate": 1.7375021810728708e-05, "loss": 1.3546, "step": 10357 }, { "epoch": 25.895, "grad_norm": 9.375, "learning_rate": 1.7374474967453907e-05, "loss": 1.485, "step": 10358 }, { "epoch": 25.8975, "grad_norm": 9.3125, "learning_rate": 1.7373928075832345e-05, "loss": 1.4156, "step": 10359 }, { "epoch": 25.9, "grad_norm": 9.3125, "learning_rate": 1.7373381135867605e-05, "loss": 1.3659, "step": 10360 }, { "epoch": 25.9025, "grad_norm": 6.65625, "learning_rate": 1.737283414756327e-05, "loss": 1.5146, "step": 10361 }, { "epoch": 25.905, "grad_norm": 6.4375, "learning_rate": 1.7372287110922933e-05, "loss": 1.37, "step": 10362 }, { "epoch": 25.9075, "grad_norm": 10.0, "learning_rate": 1.7371740025950177e-05, "loss": 1.3547, "step": 10363 }, { "epoch": 25.91, "grad_norm": 7.84375, "learning_rate": 1.7371192892648587e-05, "loss": 1.3611, "step": 10364 }, { "epoch": 25.9125, "grad_norm": 8.0625, "learning_rate": 1.7370645711021752e-05, "loss": 1.3698, "step": 10365 }, { "epoch": 25.915, "grad_norm": 7.3125, "learning_rate": 1.7370098481073256e-05, "loss": 1.3635, "step": 10366 }, { "epoch": 25.9175, "grad_norm": 5.90625, "learning_rate": 1.736955120280669e-05, "loss": 1.4687, "step": 10367 }, { "epoch": 25.92, "grad_norm": 8.5, "learning_rate": 1.7369003876225644e-05, "loss": 1.2999, "step": 10368 }, { "epoch": 25.9225, "grad_norm": 6.4375, "learning_rate": 1.7368456501333703e-05, "loss": 1.3077, "step": 10369 }, { "epoch": 25.925, "grad_norm": 8.4375, "learning_rate": 1.7367909078134454e-05, "loss": 1.3868, "step": 10370 }, { "epoch": 25.9275, "grad_norm": 5.84375, "learning_rate": 1.736736160663149e-05, "loss": 1.4111, "step": 10371 }, { "epoch": 25.93, "grad_norm": 8.0625, "learning_rate": 1.7366814086828395e-05, "loss": 1.4853, "step": 10372 }, { "epoch": 25.9325, "grad_norm": 5.53125, "learning_rate": 1.7366266518728768e-05, "loss": 1.4222, "step": 10373 }, { "epoch": 25.935, "grad_norm": 6.9375, "learning_rate": 1.7365718902336187e-05, "loss": 1.3377, "step": 10374 }, { "epoch": 25.9375, "grad_norm": 6.75, "learning_rate": 1.736517123765425e-05, "loss": 1.3891, "step": 10375 }, { "epoch": 25.94, "grad_norm": 6.59375, "learning_rate": 1.7364623524686542e-05, "loss": 1.389, "step": 10376 }, { "epoch": 25.9425, "grad_norm": 7.0, "learning_rate": 1.736407576343666e-05, "loss": 1.4535, "step": 10377 }, { "epoch": 25.945, "grad_norm": 8.9375, "learning_rate": 1.736352795390819e-05, "loss": 1.4618, "step": 10378 }, { "epoch": 25.9475, "grad_norm": 8.125, "learning_rate": 1.7362980096104726e-05, "loss": 1.4919, "step": 10379 }, { "epoch": 25.95, "grad_norm": 6.625, "learning_rate": 1.7362432190029862e-05, "loss": 1.4622, "step": 10380 }, { "epoch": 25.9525, "grad_norm": 9.8125, "learning_rate": 1.7361884235687182e-05, "loss": 1.3302, "step": 10381 }, { "epoch": 25.955, "grad_norm": 12.8125, "learning_rate": 1.736133623308029e-05, "loss": 1.4884, "step": 10382 }, { "epoch": 25.9575, "grad_norm": 6.65625, "learning_rate": 1.7360788182212767e-05, "loss": 1.4929, "step": 10383 }, { "epoch": 25.96, "grad_norm": 10.0625, "learning_rate": 1.7360240083088213e-05, "loss": 1.3886, "step": 10384 }, { "epoch": 25.9625, "grad_norm": 6.1875, "learning_rate": 1.735969193571022e-05, "loss": 1.3564, "step": 10385 }, { "epoch": 25.965, "grad_norm": 5.84375, "learning_rate": 1.7359143740082382e-05, "loss": 1.4459, "step": 10386 }, { "epoch": 25.9675, "grad_norm": 5.875, "learning_rate": 1.735859549620829e-05, "loss": 1.2525, "step": 10387 }, { "epoch": 25.97, "grad_norm": 5.40625, "learning_rate": 1.7358047204091542e-05, "loss": 1.29, "step": 10388 }, { "epoch": 25.9725, "grad_norm": 6.96875, "learning_rate": 1.7357498863735728e-05, "loss": 1.2905, "step": 10389 }, { "epoch": 25.975, "grad_norm": 7.125, "learning_rate": 1.735695047514445e-05, "loss": 1.4124, "step": 10390 }, { "epoch": 25.9775, "grad_norm": 7.875, "learning_rate": 1.7356402038321296e-05, "loss": 1.3996, "step": 10391 }, { "epoch": 25.98, "grad_norm": 8.4375, "learning_rate": 1.7355853553269865e-05, "loss": 1.4247, "step": 10392 }, { "epoch": 25.9825, "grad_norm": 7.25, "learning_rate": 1.7355305019993752e-05, "loss": 1.3472, "step": 10393 }, { "epoch": 25.985, "grad_norm": 11.1875, "learning_rate": 1.7354756438496554e-05, "loss": 1.258, "step": 10394 }, { "epoch": 25.9875, "grad_norm": 5.53125, "learning_rate": 1.735420780878187e-05, "loss": 1.4148, "step": 10395 }, { "epoch": 25.99, "grad_norm": 7.25, "learning_rate": 1.735365913085329e-05, "loss": 1.3517, "step": 10396 }, { "epoch": 25.9925, "grad_norm": 4.8125, "learning_rate": 1.7353110404714418e-05, "loss": 1.3284, "step": 10397 }, { "epoch": 25.995, "grad_norm": 7.21875, "learning_rate": 1.735256163036885e-05, "loss": 1.2765, "step": 10398 }, { "epoch": 25.9975, "grad_norm": 6.46875, "learning_rate": 1.735201280782018e-05, "loss": 1.3588, "step": 10399 }, { "epoch": 26.0, "grad_norm": 7.65625, "learning_rate": 1.7351463937072008e-05, "loss": 1.4549, "step": 10400 }, { "epoch": 26.0025, "grad_norm": 7.1875, "learning_rate": 1.735091501812793e-05, "loss": 1.2624, "step": 10401 }, { "epoch": 26.005, "grad_norm": 5.40625, "learning_rate": 1.735036605099155e-05, "loss": 1.3197, "step": 10402 }, { "epoch": 26.0075, "grad_norm": 7.46875, "learning_rate": 1.7349817035666465e-05, "loss": 1.2853, "step": 10403 }, { "epoch": 26.01, "grad_norm": 6.5625, "learning_rate": 1.734926797215627e-05, "loss": 1.2926, "step": 10404 }, { "epoch": 26.0125, "grad_norm": 6.78125, "learning_rate": 1.7348718860464574e-05, "loss": 1.4599, "step": 10405 }, { "epoch": 26.015, "grad_norm": 6.0, "learning_rate": 1.7348169700594967e-05, "loss": 1.3582, "step": 10406 }, { "epoch": 26.0175, "grad_norm": 8.125, "learning_rate": 1.7347620492551052e-05, "loss": 1.5611, "step": 10407 }, { "epoch": 26.02, "grad_norm": 6.6875, "learning_rate": 1.7347071236336435e-05, "loss": 1.5408, "step": 10408 }, { "epoch": 26.0225, "grad_norm": 7.15625, "learning_rate": 1.734652193195471e-05, "loss": 1.3445, "step": 10409 }, { "epoch": 26.025, "grad_norm": 6.84375, "learning_rate": 1.7345972579409486e-05, "loss": 1.3459, "step": 10410 }, { "epoch": 26.0275, "grad_norm": 5.1875, "learning_rate": 1.7345423178704356e-05, "loss": 1.3731, "step": 10411 }, { "epoch": 26.03, "grad_norm": 6.03125, "learning_rate": 1.7344873729842927e-05, "loss": 1.5295, "step": 10412 }, { "epoch": 26.0325, "grad_norm": 9.8125, "learning_rate": 1.7344324232828796e-05, "loss": 1.4149, "step": 10413 }, { "epoch": 26.035, "grad_norm": 5.5625, "learning_rate": 1.7343774687665576e-05, "loss": 1.2962, "step": 10414 }, { "epoch": 26.0375, "grad_norm": 7.9375, "learning_rate": 1.7343225094356857e-05, "loss": 1.4758, "step": 10415 }, { "epoch": 26.04, "grad_norm": 5.53125, "learning_rate": 1.734267545290625e-05, "loss": 1.2716, "step": 10416 }, { "epoch": 26.0425, "grad_norm": 6.0, "learning_rate": 1.7342125763317357e-05, "loss": 1.3384, "step": 10417 }, { "epoch": 26.045, "grad_norm": 6.6875, "learning_rate": 1.7341576025593783e-05, "loss": 1.4086, "step": 10418 }, { "epoch": 26.0475, "grad_norm": 6.96875, "learning_rate": 1.7341026239739127e-05, "loss": 1.4428, "step": 10419 }, { "epoch": 26.05, "grad_norm": 6.28125, "learning_rate": 1.7340476405757e-05, "loss": 1.3927, "step": 10420 }, { "epoch": 26.0525, "grad_norm": 8.6875, "learning_rate": 1.7339926523651002e-05, "loss": 1.3494, "step": 10421 }, { "epoch": 26.055, "grad_norm": 6.25, "learning_rate": 1.7339376593424737e-05, "loss": 1.3423, "step": 10422 }, { "epoch": 26.0575, "grad_norm": 8.8125, "learning_rate": 1.7338826615081816e-05, "loss": 1.3666, "step": 10423 }, { "epoch": 26.06, "grad_norm": 10.75, "learning_rate": 1.733827658862584e-05, "loss": 1.3078, "step": 10424 }, { "epoch": 26.0625, "grad_norm": 7.875, "learning_rate": 1.733772651406042e-05, "loss": 1.4103, "step": 10425 }, { "epoch": 26.065, "grad_norm": 6.375, "learning_rate": 1.7337176391389153e-05, "loss": 1.4502, "step": 10426 }, { "epoch": 26.0675, "grad_norm": 6.9375, "learning_rate": 1.7336626220615654e-05, "loss": 1.4357, "step": 10427 }, { "epoch": 26.07, "grad_norm": 11.75, "learning_rate": 1.7336076001743525e-05, "loss": 1.2809, "step": 10428 }, { "epoch": 26.0725, "grad_norm": 6.40625, "learning_rate": 1.7335525734776378e-05, "loss": 1.3125, "step": 10429 }, { "epoch": 26.075, "grad_norm": 12.0625, "learning_rate": 1.7334975419717816e-05, "loss": 1.4021, "step": 10430 }, { "epoch": 26.0775, "grad_norm": 5.34375, "learning_rate": 1.733442505657145e-05, "loss": 1.4196, "step": 10431 }, { "epoch": 26.08, "grad_norm": 12.5, "learning_rate": 1.7333874645340886e-05, "loss": 1.2392, "step": 10432 }, { "epoch": 26.0825, "grad_norm": 14.0, "learning_rate": 1.733332418602973e-05, "loss": 1.4205, "step": 10433 }, { "epoch": 26.085, "grad_norm": 9.0625, "learning_rate": 1.73327736786416e-05, "loss": 1.625, "step": 10434 }, { "epoch": 26.0875, "grad_norm": 7.03125, "learning_rate": 1.7332223123180098e-05, "loss": 1.3362, "step": 10435 }, { "epoch": 26.09, "grad_norm": 8.0625, "learning_rate": 1.7331672519648834e-05, "loss": 1.4609, "step": 10436 }, { "epoch": 26.0925, "grad_norm": 7.125, "learning_rate": 1.7331121868051418e-05, "loss": 1.2785, "step": 10437 }, { "epoch": 26.095, "grad_norm": 7.46875, "learning_rate": 1.7330571168391458e-05, "loss": 1.4873, "step": 10438 }, { "epoch": 26.0975, "grad_norm": 9.0, "learning_rate": 1.7330020420672567e-05, "loss": 1.3784, "step": 10439 }, { "epoch": 26.1, "grad_norm": 9.0625, "learning_rate": 1.732946962489836e-05, "loss": 1.4003, "step": 10440 }, { "epoch": 26.1025, "grad_norm": 7.53125, "learning_rate": 1.7328918781072444e-05, "loss": 1.4907, "step": 10441 }, { "epoch": 26.105, "grad_norm": 8.875, "learning_rate": 1.732836788919842e-05, "loss": 1.5022, "step": 10442 }, { "epoch": 26.1075, "grad_norm": 7.96875, "learning_rate": 1.732781694927992e-05, "loss": 1.4072, "step": 10443 }, { "epoch": 26.11, "grad_norm": 11.6875, "learning_rate": 1.732726596132054e-05, "loss": 1.2511, "step": 10444 }, { "epoch": 26.1125, "grad_norm": 5.65625, "learning_rate": 1.73267149253239e-05, "loss": 1.192, "step": 10445 }, { "epoch": 26.115, "grad_norm": 10.875, "learning_rate": 1.732616384129361e-05, "loss": 1.4858, "step": 10446 }, { "epoch": 26.1175, "grad_norm": 8.25, "learning_rate": 1.7325612709233282e-05, "loss": 1.1654, "step": 10447 }, { "epoch": 26.12, "grad_norm": 6.875, "learning_rate": 1.7325061529146528e-05, "loss": 1.2672, "step": 10448 }, { "epoch": 26.1225, "grad_norm": 7.09375, "learning_rate": 1.7324510301036967e-05, "loss": 1.3471, "step": 10449 }, { "epoch": 26.125, "grad_norm": 4.65625, "learning_rate": 1.732395902490821e-05, "loss": 1.3693, "step": 10450 }, { "epoch": 26.1275, "grad_norm": 11.125, "learning_rate": 1.732340770076387e-05, "loss": 1.5287, "step": 10451 }, { "epoch": 26.13, "grad_norm": 5.53125, "learning_rate": 1.732285632860756e-05, "loss": 1.3129, "step": 10452 }, { "epoch": 26.1325, "grad_norm": 7.65625, "learning_rate": 1.73223049084429e-05, "loss": 1.3211, "step": 10453 }, { "epoch": 26.135, "grad_norm": 5.4375, "learning_rate": 1.73217534402735e-05, "loss": 1.3729, "step": 10454 }, { "epoch": 26.1375, "grad_norm": 6.78125, "learning_rate": 1.7321201924102976e-05, "loss": 1.3933, "step": 10455 }, { "epoch": 26.14, "grad_norm": 7.625, "learning_rate": 1.732065035993495e-05, "loss": 1.4328, "step": 10456 }, { "epoch": 26.1425, "grad_norm": 6.46875, "learning_rate": 1.732009874777303e-05, "loss": 1.4041, "step": 10457 }, { "epoch": 26.145, "grad_norm": 4.875, "learning_rate": 1.7319547087620836e-05, "loss": 1.3184, "step": 10458 }, { "epoch": 26.1475, "grad_norm": 7.6875, "learning_rate": 1.7318995379481983e-05, "loss": 1.3463, "step": 10459 }, { "epoch": 26.15, "grad_norm": 7.1875, "learning_rate": 1.7318443623360092e-05, "loss": 1.3731, "step": 10460 }, { "epoch": 26.1525, "grad_norm": 5.09375, "learning_rate": 1.7317891819258775e-05, "loss": 1.2953, "step": 10461 }, { "epoch": 26.155, "grad_norm": 9.375, "learning_rate": 1.7317339967181655e-05, "loss": 1.2662, "step": 10462 }, { "epoch": 26.1575, "grad_norm": 6.0, "learning_rate": 1.7316788067132345e-05, "loss": 1.3751, "step": 10463 }, { "epoch": 26.16, "grad_norm": 8.1875, "learning_rate": 1.7316236119114466e-05, "loss": 1.4246, "step": 10464 }, { "epoch": 26.1625, "grad_norm": 6.25, "learning_rate": 1.731568412313163e-05, "loss": 1.5256, "step": 10465 }, { "epoch": 26.165, "grad_norm": 8.1875, "learning_rate": 1.7315132079187467e-05, "loss": 1.4921, "step": 10466 }, { "epoch": 26.1675, "grad_norm": 6.75, "learning_rate": 1.7314579987285592e-05, "loss": 1.2177, "step": 10467 }, { "epoch": 26.17, "grad_norm": 5.9375, "learning_rate": 1.731402784742962e-05, "loss": 1.3682, "step": 10468 }, { "epoch": 26.1725, "grad_norm": 11.6875, "learning_rate": 1.731347565962318e-05, "loss": 1.3965, "step": 10469 }, { "epoch": 26.175, "grad_norm": 7.8125, "learning_rate": 1.731292342386988e-05, "loss": 1.373, "step": 10470 }, { "epoch": 26.1775, "grad_norm": 6.1875, "learning_rate": 1.731237114017335e-05, "loss": 1.4818, "step": 10471 }, { "epoch": 26.18, "grad_norm": 8.0, "learning_rate": 1.7311818808537206e-05, "loss": 1.5267, "step": 10472 }, { "epoch": 26.1825, "grad_norm": 6.9375, "learning_rate": 1.731126642896507e-05, "loss": 1.3014, "step": 10473 }, { "epoch": 26.185, "grad_norm": 6.3125, "learning_rate": 1.7310714001460564e-05, "loss": 1.3797, "step": 10474 }, { "epoch": 26.1875, "grad_norm": 10.125, "learning_rate": 1.731016152602731e-05, "loss": 1.4513, "step": 10475 }, { "epoch": 26.19, "grad_norm": 5.9375, "learning_rate": 1.7309609002668932e-05, "loss": 1.3809, "step": 10476 }, { "epoch": 26.1925, "grad_norm": 10.0, "learning_rate": 1.7309056431389045e-05, "loss": 1.4168, "step": 10477 }, { "epoch": 26.195, "grad_norm": 9.0, "learning_rate": 1.730850381219128e-05, "loss": 1.3652, "step": 10478 }, { "epoch": 26.1975, "grad_norm": 6.59375, "learning_rate": 1.7307951145079255e-05, "loss": 1.3492, "step": 10479 }, { "epoch": 26.2, "grad_norm": 6.84375, "learning_rate": 1.7307398430056595e-05, "loss": 1.3221, "step": 10480 }, { "epoch": 26.2025, "grad_norm": 9.0625, "learning_rate": 1.7306845667126925e-05, "loss": 1.4939, "step": 10481 }, { "epoch": 26.205, "grad_norm": 5.46875, "learning_rate": 1.7306292856293866e-05, "loss": 1.3583, "step": 10482 }, { "epoch": 26.2075, "grad_norm": 5.8125, "learning_rate": 1.7305739997561045e-05, "loss": 1.6079, "step": 10483 }, { "epoch": 26.21, "grad_norm": 8.625, "learning_rate": 1.7305187090932083e-05, "loss": 1.2423, "step": 10484 }, { "epoch": 26.2125, "grad_norm": 5.3125, "learning_rate": 1.7304634136410604e-05, "loss": 1.5262, "step": 10485 }, { "epoch": 26.215, "grad_norm": 5.78125, "learning_rate": 1.730408113400024e-05, "loss": 1.4501, "step": 10486 }, { "epoch": 26.2175, "grad_norm": 6.90625, "learning_rate": 1.7303528083704614e-05, "loss": 1.3145, "step": 10487 }, { "epoch": 26.22, "grad_norm": 7.84375, "learning_rate": 1.7302974985527346e-05, "loss": 1.4105, "step": 10488 }, { "epoch": 26.2225, "grad_norm": 7.0, "learning_rate": 1.7302421839472067e-05, "loss": 1.2257, "step": 10489 }, { "epoch": 26.225, "grad_norm": 5.75, "learning_rate": 1.73018686455424e-05, "loss": 1.4336, "step": 10490 }, { "epoch": 26.2275, "grad_norm": 5.5625, "learning_rate": 1.7301315403741982e-05, "loss": 1.3508, "step": 10491 }, { "epoch": 26.23, "grad_norm": 8.0625, "learning_rate": 1.7300762114074428e-05, "loss": 1.456, "step": 10492 }, { "epoch": 26.2325, "grad_norm": 5.96875, "learning_rate": 1.730020877654337e-05, "loss": 1.4334, "step": 10493 }, { "epoch": 26.235, "grad_norm": 8.25, "learning_rate": 1.7299655391152432e-05, "loss": 1.4354, "step": 10494 }, { "epoch": 26.2375, "grad_norm": 6.875, "learning_rate": 1.729910195790525e-05, "loss": 1.4233, "step": 10495 }, { "epoch": 26.24, "grad_norm": 7.03125, "learning_rate": 1.7298548476805446e-05, "loss": 1.5557, "step": 10496 }, { "epoch": 26.2425, "grad_norm": 7.0, "learning_rate": 1.729799494785665e-05, "loss": 1.5663, "step": 10497 }, { "epoch": 26.245, "grad_norm": 10.1875, "learning_rate": 1.7297441371062495e-05, "loss": 1.4042, "step": 10498 }, { "epoch": 26.2475, "grad_norm": 7.25, "learning_rate": 1.72968877464266e-05, "loss": 1.3614, "step": 10499 }, { "epoch": 26.25, "grad_norm": 9.875, "learning_rate": 1.7296334073952606e-05, "loss": 1.3157, "step": 10500 }, { "epoch": 26.2525, "grad_norm": 8.1875, "learning_rate": 1.7295780353644133e-05, "loss": 1.4639, "step": 10501 }, { "epoch": 26.255, "grad_norm": 7.15625, "learning_rate": 1.729522658550482e-05, "loss": 1.3985, "step": 10502 }, { "epoch": 26.2575, "grad_norm": 9.8125, "learning_rate": 1.729467276953829e-05, "loss": 1.3538, "step": 10503 }, { "epoch": 26.26, "grad_norm": 7.28125, "learning_rate": 1.7294118905748184e-05, "loss": 1.3956, "step": 10504 }, { "epoch": 26.2625, "grad_norm": 5.6875, "learning_rate": 1.729356499413812e-05, "loss": 1.4203, "step": 10505 }, { "epoch": 26.265, "grad_norm": 5.9375, "learning_rate": 1.7293011034711734e-05, "loss": 1.3841, "step": 10506 }, { "epoch": 26.2675, "grad_norm": 7.59375, "learning_rate": 1.7292457027472667e-05, "loss": 1.4466, "step": 10507 }, { "epoch": 26.27, "grad_norm": 8.25, "learning_rate": 1.729190297242454e-05, "loss": 1.4038, "step": 10508 }, { "epoch": 26.2725, "grad_norm": 8.6875, "learning_rate": 1.7291348869570986e-05, "loss": 1.4124, "step": 10509 }, { "epoch": 26.275, "grad_norm": 9.3125, "learning_rate": 1.7290794718915644e-05, "loss": 1.5334, "step": 10510 }, { "epoch": 26.2775, "grad_norm": 5.71875, "learning_rate": 1.7290240520462144e-05, "loss": 1.2541, "step": 10511 }, { "epoch": 26.28, "grad_norm": 8.75, "learning_rate": 1.7289686274214116e-05, "loss": 1.3445, "step": 10512 }, { "epoch": 26.2825, "grad_norm": 5.625, "learning_rate": 1.72891319801752e-05, "loss": 1.3154, "step": 10513 }, { "epoch": 26.285, "grad_norm": 6.1875, "learning_rate": 1.7288577638349023e-05, "loss": 1.4458, "step": 10514 }, { "epoch": 26.2875, "grad_norm": 6.78125, "learning_rate": 1.7288023248739225e-05, "loss": 1.204, "step": 10515 }, { "epoch": 26.29, "grad_norm": 5.96875, "learning_rate": 1.7287468811349437e-05, "loss": 1.3095, "step": 10516 }, { "epoch": 26.2925, "grad_norm": 5.9375, "learning_rate": 1.7286914326183298e-05, "loss": 1.2814, "step": 10517 }, { "epoch": 26.295, "grad_norm": 6.9375, "learning_rate": 1.7286359793244436e-05, "loss": 1.3534, "step": 10518 }, { "epoch": 26.2975, "grad_norm": 7.125, "learning_rate": 1.7285805212536494e-05, "loss": 1.4149, "step": 10519 }, { "epoch": 26.3, "grad_norm": 4.78125, "learning_rate": 1.72852505840631e-05, "loss": 1.2787, "step": 10520 }, { "epoch": 26.3025, "grad_norm": 9.75, "learning_rate": 1.72846959078279e-05, "loss": 1.4609, "step": 10521 }, { "epoch": 26.305, "grad_norm": 7.8125, "learning_rate": 1.7284141183834523e-05, "loss": 1.2612, "step": 10522 }, { "epoch": 26.3075, "grad_norm": 6.59375, "learning_rate": 1.7283586412086604e-05, "loss": 1.3447, "step": 10523 }, { "epoch": 26.31, "grad_norm": 7.21875, "learning_rate": 1.7283031592587792e-05, "loss": 1.2844, "step": 10524 }, { "epoch": 26.3125, "grad_norm": 8.0, "learning_rate": 1.7282476725341713e-05, "loss": 1.4499, "step": 10525 }, { "epoch": 26.315, "grad_norm": 5.96875, "learning_rate": 1.7281921810352008e-05, "loss": 1.2938, "step": 10526 }, { "epoch": 26.3175, "grad_norm": 5.6875, "learning_rate": 1.7281366847622313e-05, "loss": 1.3775, "step": 10527 }, { "epoch": 26.32, "grad_norm": 6.5625, "learning_rate": 1.7280811837156268e-05, "loss": 1.3246, "step": 10528 }, { "epoch": 26.3225, "grad_norm": 7.78125, "learning_rate": 1.7280256778957513e-05, "loss": 1.4047, "step": 10529 }, { "epoch": 26.325, "grad_norm": 7.125, "learning_rate": 1.727970167302969e-05, "loss": 1.5076, "step": 10530 }, { "epoch": 26.3275, "grad_norm": 8.4375, "learning_rate": 1.727914651937643e-05, "loss": 1.462, "step": 10531 }, { "epoch": 26.33, "grad_norm": 7.9375, "learning_rate": 1.7278591318001376e-05, "loss": 1.3746, "step": 10532 }, { "epoch": 26.3325, "grad_norm": 8.25, "learning_rate": 1.727803606890817e-05, "loss": 1.4718, "step": 10533 }, { "epoch": 26.335, "grad_norm": 4.96875, "learning_rate": 1.727748077210045e-05, "loss": 1.1724, "step": 10534 }, { "epoch": 26.3375, "grad_norm": 6.375, "learning_rate": 1.7276925427581858e-05, "loss": 1.4742, "step": 10535 }, { "epoch": 26.34, "grad_norm": 8.8125, "learning_rate": 1.7276370035356037e-05, "loss": 1.4288, "step": 10536 }, { "epoch": 26.3425, "grad_norm": 7.28125, "learning_rate": 1.727581459542662e-05, "loss": 1.4314, "step": 10537 }, { "epoch": 26.345, "grad_norm": 4.875, "learning_rate": 1.7275259107797257e-05, "loss": 1.3473, "step": 10538 }, { "epoch": 26.3475, "grad_norm": 8.1875, "learning_rate": 1.7274703572471585e-05, "loss": 1.4064, "step": 10539 }, { "epoch": 26.35, "grad_norm": 6.28125, "learning_rate": 1.7274147989453246e-05, "loss": 1.3447, "step": 10540 }, { "epoch": 26.3525, "grad_norm": 6.5, "learning_rate": 1.727359235874589e-05, "loss": 1.3336, "step": 10541 }, { "epoch": 26.355, "grad_norm": 6.65625, "learning_rate": 1.7273036680353147e-05, "loss": 1.4625, "step": 10542 }, { "epoch": 26.3575, "grad_norm": 5.96875, "learning_rate": 1.7272480954278668e-05, "loss": 1.3885, "step": 10543 }, { "epoch": 26.36, "grad_norm": 7.34375, "learning_rate": 1.7271925180526094e-05, "loss": 1.2542, "step": 10544 }, { "epoch": 26.3625, "grad_norm": 5.96875, "learning_rate": 1.7271369359099073e-05, "loss": 1.3922, "step": 10545 }, { "epoch": 26.365, "grad_norm": 7.40625, "learning_rate": 1.7270813490001245e-05, "loss": 1.3701, "step": 10546 }, { "epoch": 26.3675, "grad_norm": 10.0, "learning_rate": 1.7270257573236254e-05, "loss": 1.3222, "step": 10547 }, { "epoch": 26.37, "grad_norm": 6.8125, "learning_rate": 1.7269701608807744e-05, "loss": 1.3178, "step": 10548 }, { "epoch": 26.3725, "grad_norm": 5.0625, "learning_rate": 1.726914559671936e-05, "loss": 1.2638, "step": 10549 }, { "epoch": 26.375, "grad_norm": 7.96875, "learning_rate": 1.726858953697475e-05, "loss": 1.3848, "step": 10550 }, { "epoch": 26.3775, "grad_norm": 6.25, "learning_rate": 1.726803342957756e-05, "loss": 1.4694, "step": 10551 }, { "epoch": 26.38, "grad_norm": 7.65625, "learning_rate": 1.7267477274531432e-05, "loss": 1.4347, "step": 10552 }, { "epoch": 26.3825, "grad_norm": 5.71875, "learning_rate": 1.726692107184001e-05, "loss": 1.224, "step": 10553 }, { "epoch": 26.385, "grad_norm": 9.0, "learning_rate": 1.726636482150695e-05, "loss": 1.4803, "step": 10554 }, { "epoch": 26.3875, "grad_norm": 7.59375, "learning_rate": 1.7265808523535894e-05, "loss": 1.2908, "step": 10555 }, { "epoch": 26.39, "grad_norm": 5.71875, "learning_rate": 1.7265252177930483e-05, "loss": 1.2601, "step": 10556 }, { "epoch": 26.3925, "grad_norm": 8.125, "learning_rate": 1.7264695784694372e-05, "loss": 1.4793, "step": 10557 }, { "epoch": 26.395, "grad_norm": 8.5, "learning_rate": 1.7264139343831205e-05, "loss": 1.457, "step": 10558 }, { "epoch": 26.3975, "grad_norm": 8.1875, "learning_rate": 1.726358285534463e-05, "loss": 1.5093, "step": 10559 }, { "epoch": 26.4, "grad_norm": 6.5, "learning_rate": 1.72630263192383e-05, "loss": 1.3911, "step": 10560 }, { "epoch": 26.4025, "grad_norm": 7.90625, "learning_rate": 1.726246973551586e-05, "loss": 1.2404, "step": 10561 }, { "epoch": 26.405, "grad_norm": 8.1875, "learning_rate": 1.726191310418096e-05, "loss": 1.3883, "step": 10562 }, { "epoch": 26.4075, "grad_norm": 8.5625, "learning_rate": 1.7261356425237245e-05, "loss": 1.4086, "step": 10563 }, { "epoch": 26.41, "grad_norm": 7.53125, "learning_rate": 1.7260799698688367e-05, "loss": 1.4848, "step": 10564 }, { "epoch": 26.4125, "grad_norm": 6.625, "learning_rate": 1.726024292453798e-05, "loss": 1.3285, "step": 10565 }, { "epoch": 26.415, "grad_norm": 11.875, "learning_rate": 1.7259686102789727e-05, "loss": 1.5421, "step": 10566 }, { "epoch": 26.4175, "grad_norm": 7.5625, "learning_rate": 1.7259129233447266e-05, "loss": 1.2535, "step": 10567 }, { "epoch": 26.42, "grad_norm": 6.03125, "learning_rate": 1.725857231651424e-05, "loss": 1.3635, "step": 10568 }, { "epoch": 26.4225, "grad_norm": 9.625, "learning_rate": 1.7258015351994307e-05, "loss": 1.4765, "step": 10569 }, { "epoch": 26.425, "grad_norm": 9.5, "learning_rate": 1.7257458339891117e-05, "loss": 1.3913, "step": 10570 }, { "epoch": 26.4275, "grad_norm": 5.84375, "learning_rate": 1.725690128020832e-05, "loss": 1.3764, "step": 10571 }, { "epoch": 26.43, "grad_norm": 9.5625, "learning_rate": 1.725634417294957e-05, "loss": 1.555, "step": 10572 }, { "epoch": 26.4325, "grad_norm": 6.875, "learning_rate": 1.7255787018118515e-05, "loss": 1.4695, "step": 10573 }, { "epoch": 26.435, "grad_norm": 8.375, "learning_rate": 1.7255229815718813e-05, "loss": 1.2897, "step": 10574 }, { "epoch": 26.4375, "grad_norm": 9.3125, "learning_rate": 1.7254672565754115e-05, "loss": 1.4878, "step": 10575 }, { "epoch": 26.44, "grad_norm": 4.59375, "learning_rate": 1.7254115268228073e-05, "loss": 1.4217, "step": 10576 }, { "epoch": 26.4425, "grad_norm": 6.84375, "learning_rate": 1.725355792314434e-05, "loss": 1.3788, "step": 10577 }, { "epoch": 26.445, "grad_norm": 5.875, "learning_rate": 1.7253000530506573e-05, "loss": 1.3252, "step": 10578 }, { "epoch": 26.4475, "grad_norm": 7.03125, "learning_rate": 1.7252443090318424e-05, "loss": 1.3708, "step": 10579 }, { "epoch": 26.45, "grad_norm": 6.9375, "learning_rate": 1.7251885602583547e-05, "loss": 1.4297, "step": 10580 }, { "epoch": 26.4525, "grad_norm": 8.5, "learning_rate": 1.7251328067305602e-05, "loss": 1.4558, "step": 10581 }, { "epoch": 26.455, "grad_norm": 9.0, "learning_rate": 1.7250770484488235e-05, "loss": 1.3947, "step": 10582 }, { "epoch": 26.4575, "grad_norm": 8.4375, "learning_rate": 1.725021285413511e-05, "loss": 1.5051, "step": 10583 }, { "epoch": 26.46, "grad_norm": 7.96875, "learning_rate": 1.7249655176249883e-05, "loss": 1.2884, "step": 10584 }, { "epoch": 26.4625, "grad_norm": 8.8125, "learning_rate": 1.7249097450836202e-05, "loss": 1.6129, "step": 10585 }, { "epoch": 26.465, "grad_norm": 7.0625, "learning_rate": 1.724853967789773e-05, "loss": 1.3271, "step": 10586 }, { "epoch": 26.4675, "grad_norm": 6.84375, "learning_rate": 1.7247981857438123e-05, "loss": 1.3838, "step": 10587 }, { "epoch": 26.47, "grad_norm": 7.59375, "learning_rate": 1.7247423989461036e-05, "loss": 1.3926, "step": 10588 }, { "epoch": 26.4725, "grad_norm": 10.4375, "learning_rate": 1.7246866073970126e-05, "loss": 1.5296, "step": 10589 }, { "epoch": 26.475, "grad_norm": 7.90625, "learning_rate": 1.724630811096905e-05, "loss": 1.3601, "step": 10590 }, { "epoch": 26.4775, "grad_norm": 7.96875, "learning_rate": 1.7245750100461472e-05, "loss": 1.3165, "step": 10591 }, { "epoch": 26.48, "grad_norm": 8.1875, "learning_rate": 1.724519204245105e-05, "loss": 1.4397, "step": 10592 }, { "epoch": 26.4825, "grad_norm": 5.5625, "learning_rate": 1.7244633936941433e-05, "loss": 1.3573, "step": 10593 }, { "epoch": 26.485, "grad_norm": 8.0625, "learning_rate": 1.724407578393629e-05, "loss": 1.5936, "step": 10594 }, { "epoch": 26.4875, "grad_norm": 11.0, "learning_rate": 1.7243517583439267e-05, "loss": 1.5523, "step": 10595 }, { "epoch": 26.49, "grad_norm": 5.0625, "learning_rate": 1.724295933545404e-05, "loss": 1.2483, "step": 10596 }, { "epoch": 26.4925, "grad_norm": 5.9375, "learning_rate": 1.724240103998426e-05, "loss": 1.458, "step": 10597 }, { "epoch": 26.495, "grad_norm": 6.875, "learning_rate": 1.724184269703359e-05, "loss": 1.3127, "step": 10598 }, { "epoch": 26.4975, "grad_norm": 6.65625, "learning_rate": 1.724128430660569e-05, "loss": 1.3508, "step": 10599 }, { "epoch": 26.5, "grad_norm": 8.3125, "learning_rate": 1.7240725868704218e-05, "loss": 1.3765, "step": 10600 }, { "epoch": 26.5025, "grad_norm": 9.25, "learning_rate": 1.7240167383332835e-05, "loss": 1.3042, "step": 10601 }, { "epoch": 26.505, "grad_norm": 7.0625, "learning_rate": 1.7239608850495207e-05, "loss": 1.4463, "step": 10602 }, { "epoch": 26.5075, "grad_norm": 8.3125, "learning_rate": 1.723905027019499e-05, "loss": 1.4614, "step": 10603 }, { "epoch": 26.51, "grad_norm": 8.5625, "learning_rate": 1.7238491642435855e-05, "loss": 1.5442, "step": 10604 }, { "epoch": 26.5125, "grad_norm": 6.90625, "learning_rate": 1.7237932967221455e-05, "loss": 1.3393, "step": 10605 }, { "epoch": 26.515, "grad_norm": 5.5, "learning_rate": 1.7237374244555455e-05, "loss": 1.3907, "step": 10606 }, { "epoch": 26.5175, "grad_norm": 7.21875, "learning_rate": 1.723681547444152e-05, "loss": 1.3826, "step": 10607 }, { "epoch": 26.52, "grad_norm": 5.71875, "learning_rate": 1.723625665688331e-05, "loss": 1.3732, "step": 10608 }, { "epoch": 26.5225, "grad_norm": 5.8125, "learning_rate": 1.7235697791884492e-05, "loss": 1.2764, "step": 10609 }, { "epoch": 26.525, "grad_norm": 5.0625, "learning_rate": 1.7235138879448733e-05, "loss": 1.4503, "step": 10610 }, { "epoch": 26.5275, "grad_norm": 7.3125, "learning_rate": 1.723457991957969e-05, "loss": 1.4437, "step": 10611 }, { "epoch": 26.53, "grad_norm": 6.75, "learning_rate": 1.723402091228103e-05, "loss": 1.4597, "step": 10612 }, { "epoch": 26.5325, "grad_norm": 9.4375, "learning_rate": 1.723346185755642e-05, "loss": 1.3435, "step": 10613 }, { "epoch": 26.535, "grad_norm": 8.25, "learning_rate": 1.7232902755409523e-05, "loss": 1.2999, "step": 10614 }, { "epoch": 26.5375, "grad_norm": 5.65625, "learning_rate": 1.7232343605844004e-05, "loss": 1.3886, "step": 10615 }, { "epoch": 26.54, "grad_norm": 6.0, "learning_rate": 1.723178440886353e-05, "loss": 1.3212, "step": 10616 }, { "epoch": 26.5425, "grad_norm": 7.4375, "learning_rate": 1.723122516447177e-05, "loss": 1.497, "step": 10617 }, { "epoch": 26.545, "grad_norm": 4.90625, "learning_rate": 1.7230665872672382e-05, "loss": 1.2264, "step": 10618 }, { "epoch": 26.5475, "grad_norm": 8.3125, "learning_rate": 1.723010653346904e-05, "loss": 1.4035, "step": 10619 }, { "epoch": 26.55, "grad_norm": 6.90625, "learning_rate": 1.722954714686541e-05, "loss": 1.5158, "step": 10620 }, { "epoch": 26.5525, "grad_norm": 7.28125, "learning_rate": 1.7228987712865156e-05, "loss": 1.2696, "step": 10621 }, { "epoch": 26.555, "grad_norm": 6.625, "learning_rate": 1.722842823147195e-05, "loss": 1.3794, "step": 10622 }, { "epoch": 26.5575, "grad_norm": 8.1875, "learning_rate": 1.722786870268946e-05, "loss": 1.4246, "step": 10623 }, { "epoch": 26.56, "grad_norm": 10.8125, "learning_rate": 1.7227309126521347e-05, "loss": 1.3911, "step": 10624 }, { "epoch": 26.5625, "grad_norm": 6.5625, "learning_rate": 1.7226749502971288e-05, "loss": 1.5537, "step": 10625 }, { "epoch": 26.565, "grad_norm": 9.875, "learning_rate": 1.7226189832042945e-05, "loss": 1.4995, "step": 10626 }, { "epoch": 26.5675, "grad_norm": 6.28125, "learning_rate": 1.7225630113739993e-05, "loss": 1.4401, "step": 10627 }, { "epoch": 26.57, "grad_norm": 7.4375, "learning_rate": 1.72250703480661e-05, "loss": 1.3972, "step": 10628 }, { "epoch": 26.5725, "grad_norm": 8.4375, "learning_rate": 1.722451053502493e-05, "loss": 1.4228, "step": 10629 }, { "epoch": 26.575, "grad_norm": 6.6875, "learning_rate": 1.7223950674620163e-05, "loss": 1.3383, "step": 10630 }, { "epoch": 26.5775, "grad_norm": 6.21875, "learning_rate": 1.7223390766855463e-05, "loss": 1.3187, "step": 10631 }, { "epoch": 26.58, "grad_norm": 6.5, "learning_rate": 1.7222830811734502e-05, "loss": 1.3692, "step": 10632 }, { "epoch": 26.5825, "grad_norm": 6.65625, "learning_rate": 1.7222270809260954e-05, "loss": 1.4271, "step": 10633 }, { "epoch": 26.585, "grad_norm": 7.28125, "learning_rate": 1.7221710759438487e-05, "loss": 1.3886, "step": 10634 }, { "epoch": 26.5875, "grad_norm": 6.09375, "learning_rate": 1.722115066227077e-05, "loss": 1.3385, "step": 10635 }, { "epoch": 26.59, "grad_norm": 9.5, "learning_rate": 1.722059051776148e-05, "loss": 1.181, "step": 10636 }, { "epoch": 26.5925, "grad_norm": 6.875, "learning_rate": 1.7220030325914287e-05, "loss": 1.5116, "step": 10637 }, { "epoch": 26.595, "grad_norm": 7.5, "learning_rate": 1.7219470086732865e-05, "loss": 1.4823, "step": 10638 }, { "epoch": 26.5975, "grad_norm": 5.84375, "learning_rate": 1.7218909800220885e-05, "loss": 1.2771, "step": 10639 }, { "epoch": 26.6, "grad_norm": 6.90625, "learning_rate": 1.7218349466382024e-05, "loss": 1.3544, "step": 10640 }, { "epoch": 26.6025, "grad_norm": 24.375, "learning_rate": 1.721778908521995e-05, "loss": 1.3806, "step": 10641 }, { "epoch": 26.605, "grad_norm": 6.1875, "learning_rate": 1.7217228656738337e-05, "loss": 1.2607, "step": 10642 }, { "epoch": 26.6075, "grad_norm": 8.3125, "learning_rate": 1.7216668180940867e-05, "loss": 1.3165, "step": 10643 }, { "epoch": 26.61, "grad_norm": 7.59375, "learning_rate": 1.7216107657831206e-05, "loss": 1.4345, "step": 10644 }, { "epoch": 26.6125, "grad_norm": 6.71875, "learning_rate": 1.7215547087413034e-05, "loss": 1.4091, "step": 10645 }, { "epoch": 26.615, "grad_norm": 9.625, "learning_rate": 1.7214986469690024e-05, "loss": 1.429, "step": 10646 }, { "epoch": 26.6175, "grad_norm": 7.65625, "learning_rate": 1.7214425804665847e-05, "loss": 1.437, "step": 10647 }, { "epoch": 26.62, "grad_norm": 6.875, "learning_rate": 1.7213865092344188e-05, "loss": 1.4864, "step": 10648 }, { "epoch": 26.6225, "grad_norm": 7.5625, "learning_rate": 1.721330433272872e-05, "loss": 1.3985, "step": 10649 }, { "epoch": 26.625, "grad_norm": 6.625, "learning_rate": 1.721274352582311e-05, "loss": 1.2636, "step": 10650 }, { "epoch": 26.6275, "grad_norm": 7.90625, "learning_rate": 1.7212182671631047e-05, "loss": 1.334, "step": 10651 }, { "epoch": 26.63, "grad_norm": 5.5, "learning_rate": 1.72116217701562e-05, "loss": 1.3849, "step": 10652 }, { "epoch": 26.6325, "grad_norm": 8.125, "learning_rate": 1.7211060821402255e-05, "loss": 1.3796, "step": 10653 }, { "epoch": 26.635, "grad_norm": 7.0, "learning_rate": 1.721049982537288e-05, "loss": 1.3418, "step": 10654 }, { "epoch": 26.6375, "grad_norm": 7.09375, "learning_rate": 1.7209938782071754e-05, "loss": 1.295, "step": 10655 }, { "epoch": 26.64, "grad_norm": 7.78125, "learning_rate": 1.7209377691502565e-05, "loss": 1.4327, "step": 10656 }, { "epoch": 26.6425, "grad_norm": 7.3125, "learning_rate": 1.720881655366898e-05, "loss": 1.3273, "step": 10657 }, { "epoch": 26.645, "grad_norm": 6.0, "learning_rate": 1.720825536857468e-05, "loss": 1.6345, "step": 10658 }, { "epoch": 26.6475, "grad_norm": 9.0, "learning_rate": 1.720769413622335e-05, "loss": 1.4385, "step": 10659 }, { "epoch": 26.65, "grad_norm": 7.71875, "learning_rate": 1.7207132856618668e-05, "loss": 1.2804, "step": 10660 }, { "epoch": 26.6525, "grad_norm": 5.875, "learning_rate": 1.7206571529764305e-05, "loss": 1.3048, "step": 10661 }, { "epoch": 26.655, "grad_norm": 5.03125, "learning_rate": 1.720601015566395e-05, "loss": 1.5277, "step": 10662 }, { "epoch": 26.6575, "grad_norm": 7.40625, "learning_rate": 1.720544873432128e-05, "loss": 1.4183, "step": 10663 }, { "epoch": 26.66, "grad_norm": 7.59375, "learning_rate": 1.7204887265739978e-05, "loss": 1.4283, "step": 10664 }, { "epoch": 26.6625, "grad_norm": 5.90625, "learning_rate": 1.7204325749923722e-05, "loss": 1.3292, "step": 10665 }, { "epoch": 26.665, "grad_norm": 4.84375, "learning_rate": 1.7203764186876198e-05, "loss": 1.337, "step": 10666 }, { "epoch": 26.6675, "grad_norm": 7.0, "learning_rate": 1.720320257660108e-05, "loss": 1.3601, "step": 10667 }, { "epoch": 26.67, "grad_norm": 6.03125, "learning_rate": 1.7202640919102057e-05, "loss": 1.4765, "step": 10668 }, { "epoch": 26.6725, "grad_norm": 7.625, "learning_rate": 1.7202079214382807e-05, "loss": 1.4429, "step": 10669 }, { "epoch": 26.675, "grad_norm": 10.0, "learning_rate": 1.7201517462447014e-05, "loss": 1.3313, "step": 10670 }, { "epoch": 26.6775, "grad_norm": 7.75, "learning_rate": 1.7200955663298362e-05, "loss": 1.3715, "step": 10671 }, { "epoch": 26.68, "grad_norm": 6.9375, "learning_rate": 1.720039381694053e-05, "loss": 1.4618, "step": 10672 }, { "epoch": 26.6825, "grad_norm": 9.4375, "learning_rate": 1.7199831923377205e-05, "loss": 1.272, "step": 10673 }, { "epoch": 26.685, "grad_norm": 6.96875, "learning_rate": 1.719926998261207e-05, "loss": 1.2863, "step": 10674 }, { "epoch": 26.6875, "grad_norm": 6.90625, "learning_rate": 1.7198707994648812e-05, "loss": 1.2198, "step": 10675 }, { "epoch": 26.69, "grad_norm": 5.71875, "learning_rate": 1.7198145959491113e-05, "loss": 1.6707, "step": 10676 }, { "epoch": 26.6925, "grad_norm": 6.96875, "learning_rate": 1.7197583877142654e-05, "loss": 1.4351, "step": 10677 }, { "epoch": 26.695, "grad_norm": 8.625, "learning_rate": 1.7197021747607124e-05, "loss": 1.3875, "step": 10678 }, { "epoch": 26.6975, "grad_norm": 6.8125, "learning_rate": 1.7196459570888207e-05, "loss": 1.5142, "step": 10679 }, { "epoch": 26.7, "grad_norm": 10.3125, "learning_rate": 1.719589734698959e-05, "loss": 1.383, "step": 10680 }, { "epoch": 26.7025, "grad_norm": 6.6875, "learning_rate": 1.7195335075914955e-05, "loss": 1.3071, "step": 10681 }, { "epoch": 26.705, "grad_norm": 9.75, "learning_rate": 1.7194772757667993e-05, "loss": 1.377, "step": 10682 }, { "epoch": 26.7075, "grad_norm": 7.0, "learning_rate": 1.7194210392252392e-05, "loss": 1.4676, "step": 10683 }, { "epoch": 26.71, "grad_norm": 5.53125, "learning_rate": 1.719364797967183e-05, "loss": 1.422, "step": 10684 }, { "epoch": 26.7125, "grad_norm": 6.46875, "learning_rate": 1.7193085519930008e-05, "loss": 1.2579, "step": 10685 }, { "epoch": 26.715, "grad_norm": 5.375, "learning_rate": 1.7192523013030597e-05, "loss": 1.3883, "step": 10686 }, { "epoch": 26.7175, "grad_norm": 7.71875, "learning_rate": 1.7191960458977296e-05, "loss": 1.5452, "step": 10687 }, { "epoch": 26.72, "grad_norm": 6.25, "learning_rate": 1.7191397857773787e-05, "loss": 1.4711, "step": 10688 }, { "epoch": 26.7225, "grad_norm": 9.5, "learning_rate": 1.7190835209423767e-05, "loss": 1.5817, "step": 10689 }, { "epoch": 26.725, "grad_norm": 6.15625, "learning_rate": 1.7190272513930916e-05, "loss": 1.5133, "step": 10690 }, { "epoch": 26.7275, "grad_norm": 6.75, "learning_rate": 1.7189709771298925e-05, "loss": 1.4097, "step": 10691 }, { "epoch": 26.73, "grad_norm": 8.3125, "learning_rate": 1.7189146981531486e-05, "loss": 1.4157, "step": 10692 }, { "epoch": 26.7325, "grad_norm": 5.96875, "learning_rate": 1.7188584144632288e-05, "loss": 1.3593, "step": 10693 }, { "epoch": 26.735, "grad_norm": 5.8125, "learning_rate": 1.718802126060502e-05, "loss": 1.4489, "step": 10694 }, { "epoch": 26.7375, "grad_norm": 9.1875, "learning_rate": 1.718745832945337e-05, "loss": 1.4622, "step": 10695 }, { "epoch": 26.74, "grad_norm": 7.25, "learning_rate": 1.718689535118103e-05, "loss": 1.3677, "step": 10696 }, { "epoch": 26.7425, "grad_norm": 6.53125, "learning_rate": 1.7186332325791693e-05, "loss": 1.4428, "step": 10697 }, { "epoch": 26.745, "grad_norm": 7.78125, "learning_rate": 1.7185769253289054e-05, "loss": 1.4472, "step": 10698 }, { "epoch": 26.7475, "grad_norm": 4.96875, "learning_rate": 1.7185206133676794e-05, "loss": 1.3712, "step": 10699 }, { "epoch": 26.75, "grad_norm": 6.71875, "learning_rate": 1.718464296695861e-05, "loss": 1.3322, "step": 10700 }, { "epoch": 26.7525, "grad_norm": 6.21875, "learning_rate": 1.7184079753138196e-05, "loss": 1.2956, "step": 10701 }, { "epoch": 26.755, "grad_norm": 5.71875, "learning_rate": 1.718351649221924e-05, "loss": 1.3052, "step": 10702 }, { "epoch": 26.7575, "grad_norm": 8.1875, "learning_rate": 1.718295318420544e-05, "loss": 1.5108, "step": 10703 }, { "epoch": 26.76, "grad_norm": 7.0625, "learning_rate": 1.7182389829100484e-05, "loss": 1.3522, "step": 10704 }, { "epoch": 26.7625, "grad_norm": 8.125, "learning_rate": 1.718182642690807e-05, "loss": 1.3735, "step": 10705 }, { "epoch": 26.765, "grad_norm": 5.125, "learning_rate": 1.718126297763189e-05, "loss": 1.3307, "step": 10706 }, { "epoch": 26.7675, "grad_norm": 6.4375, "learning_rate": 1.7180699481275637e-05, "loss": 1.3517, "step": 10707 }, { "epoch": 26.77, "grad_norm": 8.75, "learning_rate": 1.7180135937843005e-05, "loss": 1.4834, "step": 10708 }, { "epoch": 26.7725, "grad_norm": 5.90625, "learning_rate": 1.717957234733769e-05, "loss": 1.4553, "step": 10709 }, { "epoch": 26.775, "grad_norm": 7.9375, "learning_rate": 1.7179008709763382e-05, "loss": 1.3363, "step": 10710 }, { "epoch": 26.7775, "grad_norm": 9.375, "learning_rate": 1.7178445025123784e-05, "loss": 1.464, "step": 10711 }, { "epoch": 26.78, "grad_norm": 6.25, "learning_rate": 1.7177881293422586e-05, "loss": 1.4223, "step": 10712 }, { "epoch": 26.7825, "grad_norm": 6.71875, "learning_rate": 1.7177317514663486e-05, "loss": 1.468, "step": 10713 }, { "epoch": 26.785, "grad_norm": 7.96875, "learning_rate": 1.7176753688850177e-05, "loss": 1.4606, "step": 10714 }, { "epoch": 26.7875, "grad_norm": 7.125, "learning_rate": 1.717618981598636e-05, "loss": 1.3545, "step": 10715 }, { "epoch": 26.79, "grad_norm": 7.46875, "learning_rate": 1.7175625896075732e-05, "loss": 1.3539, "step": 10716 }, { "epoch": 26.7925, "grad_norm": 6.75, "learning_rate": 1.7175061929121984e-05, "loss": 1.4082, "step": 10717 }, { "epoch": 26.795, "grad_norm": 7.25, "learning_rate": 1.7174497915128824e-05, "loss": 1.4671, "step": 10718 }, { "epoch": 26.7975, "grad_norm": 8.75, "learning_rate": 1.7173933854099935e-05, "loss": 1.3969, "step": 10719 }, { "epoch": 26.8, "grad_norm": 6.5625, "learning_rate": 1.7173369746039026e-05, "loss": 1.4788, "step": 10720 }, { "epoch": 26.8025, "grad_norm": 7.75, "learning_rate": 1.717280559094979e-05, "loss": 1.4419, "step": 10721 }, { "epoch": 26.805, "grad_norm": 6.40625, "learning_rate": 1.717224138883593e-05, "loss": 1.3497, "step": 10722 }, { "epoch": 26.8075, "grad_norm": 6.6875, "learning_rate": 1.717167713970114e-05, "loss": 1.4851, "step": 10723 }, { "epoch": 26.81, "grad_norm": 8.3125, "learning_rate": 1.7171112843549126e-05, "loss": 1.4028, "step": 10724 }, { "epoch": 26.8125, "grad_norm": 6.0625, "learning_rate": 1.717054850038358e-05, "loss": 1.2209, "step": 10725 }, { "epoch": 26.815, "grad_norm": 10.9375, "learning_rate": 1.716998411020821e-05, "loss": 1.4431, "step": 10726 }, { "epoch": 26.8175, "grad_norm": 6.75, "learning_rate": 1.7169419673026705e-05, "loss": 1.3818, "step": 10727 }, { "epoch": 26.82, "grad_norm": 8.0625, "learning_rate": 1.716885518884277e-05, "loss": 1.3731, "step": 10728 }, { "epoch": 26.8225, "grad_norm": 7.125, "learning_rate": 1.7168290657660113e-05, "loss": 1.4015, "step": 10729 }, { "epoch": 26.825, "grad_norm": 10.125, "learning_rate": 1.7167726079482427e-05, "loss": 1.2202, "step": 10730 }, { "epoch": 26.8275, "grad_norm": 12.125, "learning_rate": 1.7167161454313413e-05, "loss": 1.3755, "step": 10731 }, { "epoch": 26.83, "grad_norm": 6.84375, "learning_rate": 1.7166596782156778e-05, "loss": 1.4625, "step": 10732 }, { "epoch": 26.8325, "grad_norm": 6.5625, "learning_rate": 1.716603206301622e-05, "loss": 1.2879, "step": 10733 }, { "epoch": 26.835, "grad_norm": 5.875, "learning_rate": 1.7165467296895446e-05, "loss": 1.3298, "step": 10734 }, { "epoch": 26.8375, "grad_norm": 7.96875, "learning_rate": 1.7164902483798153e-05, "loss": 1.3458, "step": 10735 }, { "epoch": 26.84, "grad_norm": 5.625, "learning_rate": 1.7164337623728044e-05, "loss": 1.3168, "step": 10736 }, { "epoch": 26.8425, "grad_norm": 6.03125, "learning_rate": 1.7163772716688827e-05, "loss": 1.2321, "step": 10737 }, { "epoch": 26.845, "grad_norm": 6.6875, "learning_rate": 1.7163207762684205e-05, "loss": 1.4171, "step": 10738 }, { "epoch": 26.8475, "grad_norm": 8.875, "learning_rate": 1.7162642761717873e-05, "loss": 1.3125, "step": 10739 }, { "epoch": 26.85, "grad_norm": 10.25, "learning_rate": 1.7162077713793547e-05, "loss": 1.3309, "step": 10740 }, { "epoch": 26.8525, "grad_norm": 6.84375, "learning_rate": 1.7161512618914924e-05, "loss": 1.4121, "step": 10741 }, { "epoch": 26.855, "grad_norm": 7.625, "learning_rate": 1.7160947477085713e-05, "loss": 1.4577, "step": 10742 }, { "epoch": 26.8575, "grad_norm": 7.28125, "learning_rate": 1.7160382288309612e-05, "loss": 1.4711, "step": 10743 }, { "epoch": 26.86, "grad_norm": 6.53125, "learning_rate": 1.7159817052590332e-05, "loss": 1.5715, "step": 10744 }, { "epoch": 26.8625, "grad_norm": 8.125, "learning_rate": 1.7159251769931583e-05, "loss": 1.3729, "step": 10745 }, { "epoch": 26.865, "grad_norm": 6.9375, "learning_rate": 1.7158686440337065e-05, "loss": 1.4687, "step": 10746 }, { "epoch": 26.8675, "grad_norm": 8.5625, "learning_rate": 1.7158121063810483e-05, "loss": 1.4608, "step": 10747 }, { "epoch": 26.87, "grad_norm": 10.0, "learning_rate": 1.7157555640355546e-05, "loss": 1.3705, "step": 10748 }, { "epoch": 26.8725, "grad_norm": 8.6875, "learning_rate": 1.715699016997596e-05, "loss": 1.2378, "step": 10749 }, { "epoch": 26.875, "grad_norm": 6.59375, "learning_rate": 1.7156424652675433e-05, "loss": 1.4438, "step": 10750 }, { "epoch": 26.8775, "grad_norm": 5.78125, "learning_rate": 1.715585908845767e-05, "loss": 1.3259, "step": 10751 }, { "epoch": 26.88, "grad_norm": 7.09375, "learning_rate": 1.7155293477326385e-05, "loss": 1.4031, "step": 10752 }, { "epoch": 26.8825, "grad_norm": 7.5625, "learning_rate": 1.715472781928528e-05, "loss": 1.4522, "step": 10753 }, { "epoch": 26.885, "grad_norm": 5.6875, "learning_rate": 1.715416211433807e-05, "loss": 1.3803, "step": 10754 }, { "epoch": 26.8875, "grad_norm": 6.8125, "learning_rate": 1.7153596362488455e-05, "loss": 1.3672, "step": 10755 }, { "epoch": 26.89, "grad_norm": 6.03125, "learning_rate": 1.715303056374015e-05, "loss": 1.3978, "step": 10756 }, { "epoch": 26.8925, "grad_norm": 4.71875, "learning_rate": 1.7152464718096864e-05, "loss": 1.217, "step": 10757 }, { "epoch": 26.895, "grad_norm": 6.46875, "learning_rate": 1.7151898825562304e-05, "loss": 1.4457, "step": 10758 }, { "epoch": 26.8975, "grad_norm": 10.375, "learning_rate": 1.715133288614018e-05, "loss": 1.3707, "step": 10759 }, { "epoch": 26.9, "grad_norm": 6.5, "learning_rate": 1.7150766899834205e-05, "loss": 1.4931, "step": 10760 }, { "epoch": 26.9025, "grad_norm": 6.625, "learning_rate": 1.7150200866648093e-05, "loss": 1.3461, "step": 10761 }, { "epoch": 26.905, "grad_norm": 9.0, "learning_rate": 1.7149634786585544e-05, "loss": 1.5449, "step": 10762 }, { "epoch": 26.9075, "grad_norm": 6.15625, "learning_rate": 1.714906865965028e-05, "loss": 1.2979, "step": 10763 }, { "epoch": 26.91, "grad_norm": 5.75, "learning_rate": 1.7148502485846005e-05, "loss": 1.3677, "step": 10764 }, { "epoch": 26.9125, "grad_norm": 10.1875, "learning_rate": 1.7147936265176435e-05, "loss": 1.4522, "step": 10765 }, { "epoch": 26.915, "grad_norm": 8.6875, "learning_rate": 1.714736999764528e-05, "loss": 1.3844, "step": 10766 }, { "epoch": 26.9175, "grad_norm": 5.0, "learning_rate": 1.7146803683256252e-05, "loss": 1.311, "step": 10767 }, { "epoch": 26.92, "grad_norm": 8.1875, "learning_rate": 1.714623732201307e-05, "loss": 1.1919, "step": 10768 }, { "epoch": 26.9225, "grad_norm": 9.375, "learning_rate": 1.7145670913919438e-05, "loss": 1.3908, "step": 10769 }, { "epoch": 26.925, "grad_norm": 6.46875, "learning_rate": 1.7145104458979074e-05, "loss": 1.3478, "step": 10770 }, { "epoch": 26.9275, "grad_norm": 10.0625, "learning_rate": 1.714453795719569e-05, "loss": 1.434, "step": 10771 }, { "epoch": 26.93, "grad_norm": 7.90625, "learning_rate": 1.7143971408573e-05, "loss": 1.3583, "step": 10772 }, { "epoch": 26.9325, "grad_norm": 5.1875, "learning_rate": 1.7143404813114726e-05, "loss": 1.2904, "step": 10773 }, { "epoch": 26.935, "grad_norm": 9.3125, "learning_rate": 1.714283817082457e-05, "loss": 1.3944, "step": 10774 }, { "epoch": 26.9375, "grad_norm": 7.375, "learning_rate": 1.7142271481706256e-05, "loss": 1.3079, "step": 10775 }, { "epoch": 26.94, "grad_norm": 5.34375, "learning_rate": 1.7141704745763494e-05, "loss": 1.3721, "step": 10776 }, { "epoch": 26.9425, "grad_norm": 6.3125, "learning_rate": 1.7141137963e-05, "loss": 1.2072, "step": 10777 }, { "epoch": 26.945, "grad_norm": 6.96875, "learning_rate": 1.7140571133419494e-05, "loss": 1.4704, "step": 10778 }, { "epoch": 26.9475, "grad_norm": 9.1875, "learning_rate": 1.714000425702569e-05, "loss": 1.3348, "step": 10779 }, { "epoch": 26.95, "grad_norm": 7.15625, "learning_rate": 1.7139437333822303e-05, "loss": 1.4262, "step": 10780 }, { "epoch": 26.9525, "grad_norm": 9.4375, "learning_rate": 1.713887036381305e-05, "loss": 1.2682, "step": 10781 }, { "epoch": 26.955, "grad_norm": 7.59375, "learning_rate": 1.713830334700165e-05, "loss": 1.3963, "step": 10782 }, { "epoch": 26.9575, "grad_norm": 7.125, "learning_rate": 1.7137736283391815e-05, "loss": 1.3887, "step": 10783 }, { "epoch": 26.96, "grad_norm": 6.1875, "learning_rate": 1.713716917298727e-05, "loss": 1.5034, "step": 10784 }, { "epoch": 26.9625, "grad_norm": 8.0625, "learning_rate": 1.7136602015791727e-05, "loss": 1.4261, "step": 10785 }, { "epoch": 26.965, "grad_norm": 8.0625, "learning_rate": 1.713603481180891e-05, "loss": 1.33, "step": 10786 }, { "epoch": 26.9675, "grad_norm": 7.53125, "learning_rate": 1.713546756104253e-05, "loss": 1.2932, "step": 10787 }, { "epoch": 26.97, "grad_norm": 5.6875, "learning_rate": 1.713490026349631e-05, "loss": 1.3147, "step": 10788 }, { "epoch": 26.9725, "grad_norm": 9.75, "learning_rate": 1.7134332919173978e-05, "loss": 1.4199, "step": 10789 }, { "epoch": 26.975, "grad_norm": 7.125, "learning_rate": 1.7133765528079238e-05, "loss": 1.3607, "step": 10790 }, { "epoch": 26.9775, "grad_norm": 6.96875, "learning_rate": 1.7133198090215817e-05, "loss": 1.5489, "step": 10791 }, { "epoch": 26.98, "grad_norm": 9.25, "learning_rate": 1.7132630605587433e-05, "loss": 1.4457, "step": 10792 }, { "epoch": 26.9825, "grad_norm": 9.25, "learning_rate": 1.7132063074197813e-05, "loss": 1.3767, "step": 10793 }, { "epoch": 26.985, "grad_norm": 7.8125, "learning_rate": 1.7131495496050673e-05, "loss": 1.4044, "step": 10794 }, { "epoch": 26.9875, "grad_norm": 6.75, "learning_rate": 1.713092787114973e-05, "loss": 1.2495, "step": 10795 }, { "epoch": 26.99, "grad_norm": 7.34375, "learning_rate": 1.713036019949871e-05, "loss": 1.2833, "step": 10796 }, { "epoch": 26.9925, "grad_norm": 5.375, "learning_rate": 1.7129792481101336e-05, "loss": 1.2723, "step": 10797 }, { "epoch": 26.995, "grad_norm": 10.75, "learning_rate": 1.712922471596133e-05, "loss": 1.5695, "step": 10798 }, { "epoch": 26.9975, "grad_norm": 7.34375, "learning_rate": 1.7128656904082402e-05, "loss": 1.2283, "step": 10799 }, { "epoch": 27.0, "grad_norm": 6.6875, "learning_rate": 1.7128089045468294e-05, "loss": 1.3138, "step": 10800 }, { "epoch": 27.0025, "grad_norm": 6.0625, "learning_rate": 1.7127521140122718e-05, "loss": 1.4094, "step": 10801 }, { "epoch": 27.005, "grad_norm": 6.1875, "learning_rate": 1.7126953188049397e-05, "loss": 1.3336, "step": 10802 }, { "epoch": 27.0075, "grad_norm": 7.84375, "learning_rate": 1.7126385189252055e-05, "loss": 1.3392, "step": 10803 }, { "epoch": 27.01, "grad_norm": 6.0625, "learning_rate": 1.712581714373442e-05, "loss": 1.3482, "step": 10804 }, { "epoch": 27.0125, "grad_norm": 6.09375, "learning_rate": 1.712524905150021e-05, "loss": 1.4778, "step": 10805 }, { "epoch": 27.015, "grad_norm": 5.6875, "learning_rate": 1.7124680912553152e-05, "loss": 1.3702, "step": 10806 }, { "epoch": 27.0175, "grad_norm": 6.15625, "learning_rate": 1.7124112726896974e-05, "loss": 1.2782, "step": 10807 }, { "epoch": 27.02, "grad_norm": 5.5625, "learning_rate": 1.71235444945354e-05, "loss": 1.2287, "step": 10808 }, { "epoch": 27.0225, "grad_norm": 12.4375, "learning_rate": 1.7122976215472147e-05, "loss": 1.4074, "step": 10809 }, { "epoch": 27.025, "grad_norm": 6.90625, "learning_rate": 1.712240788971095e-05, "loss": 1.4292, "step": 10810 }, { "epoch": 27.0275, "grad_norm": 8.875, "learning_rate": 1.7121839517255532e-05, "loss": 1.4527, "step": 10811 }, { "epoch": 27.03, "grad_norm": 7.1875, "learning_rate": 1.712127109810962e-05, "loss": 1.3972, "step": 10812 }, { "epoch": 27.0325, "grad_norm": 8.4375, "learning_rate": 1.7120702632276938e-05, "loss": 1.3365, "step": 10813 }, { "epoch": 27.035, "grad_norm": 8.625, "learning_rate": 1.7120134119761215e-05, "loss": 1.405, "step": 10814 }, { "epoch": 27.0375, "grad_norm": 10.0, "learning_rate": 1.711956556056618e-05, "loss": 1.3363, "step": 10815 }, { "epoch": 27.04, "grad_norm": 6.34375, "learning_rate": 1.7118996954695553e-05, "loss": 1.4813, "step": 10816 }, { "epoch": 27.0425, "grad_norm": 8.75, "learning_rate": 1.7118428302153073e-05, "loss": 1.3759, "step": 10817 }, { "epoch": 27.045, "grad_norm": 7.40625, "learning_rate": 1.711785960294246e-05, "loss": 1.3866, "step": 10818 }, { "epoch": 27.0475, "grad_norm": 10.3125, "learning_rate": 1.711729085706744e-05, "loss": 1.367, "step": 10819 }, { "epoch": 27.05, "grad_norm": 6.9375, "learning_rate": 1.711672206453175e-05, "loss": 1.1923, "step": 10820 }, { "epoch": 27.0525, "grad_norm": 5.5, "learning_rate": 1.7116153225339114e-05, "loss": 1.2064, "step": 10821 }, { "epoch": 27.055, "grad_norm": 9.375, "learning_rate": 1.711558433949326e-05, "loss": 1.4272, "step": 10822 }, { "epoch": 27.0575, "grad_norm": 8.125, "learning_rate": 1.7115015406997924e-05, "loss": 1.3889, "step": 10823 }, { "epoch": 27.06, "grad_norm": 7.28125, "learning_rate": 1.711444642785683e-05, "loss": 1.2524, "step": 10824 }, { "epoch": 27.0625, "grad_norm": 7.53125, "learning_rate": 1.711387740207371e-05, "loss": 1.3297, "step": 10825 }, { "epoch": 27.065, "grad_norm": 8.0, "learning_rate": 1.711330832965229e-05, "loss": 1.4875, "step": 10826 }, { "epoch": 27.0675, "grad_norm": 7.65625, "learning_rate": 1.711273921059631e-05, "loss": 1.4138, "step": 10827 }, { "epoch": 27.07, "grad_norm": 6.9375, "learning_rate": 1.7112170044909497e-05, "loss": 1.4832, "step": 10828 }, { "epoch": 27.0725, "grad_norm": 6.96875, "learning_rate": 1.711160083259558e-05, "loss": 1.3395, "step": 10829 }, { "epoch": 27.075, "grad_norm": 8.4375, "learning_rate": 1.7111031573658296e-05, "loss": 1.3059, "step": 10830 }, { "epoch": 27.0775, "grad_norm": 7.1875, "learning_rate": 1.7110462268101365e-05, "loss": 1.2839, "step": 10831 }, { "epoch": 27.08, "grad_norm": 11.75, "learning_rate": 1.7109892915928535e-05, "loss": 1.4259, "step": 10832 }, { "epoch": 27.0825, "grad_norm": 4.21875, "learning_rate": 1.7109323517143533e-05, "loss": 1.2688, "step": 10833 }, { "epoch": 27.085, "grad_norm": 9.8125, "learning_rate": 1.7108754071750086e-05, "loss": 1.3525, "step": 10834 }, { "epoch": 27.0875, "grad_norm": 7.375, "learning_rate": 1.710818457975193e-05, "loss": 1.3444, "step": 10835 }, { "epoch": 27.09, "grad_norm": 6.46875, "learning_rate": 1.7107615041152807e-05, "loss": 1.1965, "step": 10836 }, { "epoch": 27.0925, "grad_norm": 7.78125, "learning_rate": 1.710704545595644e-05, "loss": 1.2026, "step": 10837 }, { "epoch": 27.095, "grad_norm": 7.28125, "learning_rate": 1.710647582416657e-05, "loss": 1.3927, "step": 10838 }, { "epoch": 27.0975, "grad_norm": 5.71875, "learning_rate": 1.7105906145786925e-05, "loss": 1.2665, "step": 10839 }, { "epoch": 27.1, "grad_norm": 7.34375, "learning_rate": 1.7105336420821247e-05, "loss": 1.451, "step": 10840 }, { "epoch": 27.1025, "grad_norm": 6.34375, "learning_rate": 1.7104766649273264e-05, "loss": 1.4198, "step": 10841 }, { "epoch": 27.105, "grad_norm": 9.5, "learning_rate": 1.710419683114672e-05, "loss": 1.3536, "step": 10842 }, { "epoch": 27.1075, "grad_norm": 5.21875, "learning_rate": 1.7103626966445346e-05, "loss": 1.3388, "step": 10843 }, { "epoch": 27.11, "grad_norm": 5.46875, "learning_rate": 1.7103057055172875e-05, "loss": 1.3508, "step": 10844 }, { "epoch": 27.1125, "grad_norm": 6.96875, "learning_rate": 1.7102487097333046e-05, "loss": 1.2759, "step": 10845 }, { "epoch": 27.115, "grad_norm": 7.84375, "learning_rate": 1.7101917092929597e-05, "loss": 1.4514, "step": 10846 }, { "epoch": 27.1175, "grad_norm": 7.90625, "learning_rate": 1.7101347041966265e-05, "loss": 1.5653, "step": 10847 }, { "epoch": 27.12, "grad_norm": 9.375, "learning_rate": 1.7100776944446783e-05, "loss": 1.2737, "step": 10848 }, { "epoch": 27.1225, "grad_norm": 5.9375, "learning_rate": 1.7100206800374892e-05, "loss": 1.5387, "step": 10849 }, { "epoch": 27.125, "grad_norm": 6.21875, "learning_rate": 1.709963660975433e-05, "loss": 1.2993, "step": 10850 }, { "epoch": 27.1275, "grad_norm": 5.4375, "learning_rate": 1.7099066372588832e-05, "loss": 1.4265, "step": 10851 }, { "epoch": 27.13, "grad_norm": 7.65625, "learning_rate": 1.7098496088882143e-05, "loss": 1.3606, "step": 10852 }, { "epoch": 27.1325, "grad_norm": 6.96875, "learning_rate": 1.7097925758637998e-05, "loss": 1.3616, "step": 10853 }, { "epoch": 27.135, "grad_norm": 7.90625, "learning_rate": 1.7097355381860136e-05, "loss": 1.2947, "step": 10854 }, { "epoch": 27.1375, "grad_norm": 6.78125, "learning_rate": 1.7096784958552294e-05, "loss": 1.4105, "step": 10855 }, { "epoch": 27.14, "grad_norm": 7.78125, "learning_rate": 1.709621448871821e-05, "loss": 1.4804, "step": 10856 }, { "epoch": 27.1425, "grad_norm": 6.6875, "learning_rate": 1.7095643972361633e-05, "loss": 1.3697, "step": 10857 }, { "epoch": 27.145, "grad_norm": 8.5, "learning_rate": 1.70950734094863e-05, "loss": 1.3762, "step": 10858 }, { "epoch": 27.1475, "grad_norm": 6.75, "learning_rate": 1.7094502800095944e-05, "loss": 1.3724, "step": 10859 }, { "epoch": 27.15, "grad_norm": 7.1875, "learning_rate": 1.709393214419431e-05, "loss": 1.4472, "step": 10860 }, { "epoch": 27.1525, "grad_norm": 6.875, "learning_rate": 1.709336144178515e-05, "loss": 1.4205, "step": 10861 }, { "epoch": 27.155, "grad_norm": 9.8125, "learning_rate": 1.709279069287219e-05, "loss": 1.2744, "step": 10862 }, { "epoch": 27.1575, "grad_norm": 7.65625, "learning_rate": 1.7092219897459177e-05, "loss": 1.3401, "step": 10863 }, { "epoch": 27.16, "grad_norm": 5.875, "learning_rate": 1.709164905554986e-05, "loss": 1.3097, "step": 10864 }, { "epoch": 27.1625, "grad_norm": 8.1875, "learning_rate": 1.709107816714797e-05, "loss": 1.3647, "step": 10865 }, { "epoch": 27.165, "grad_norm": 9.6875, "learning_rate": 1.7090507232257255e-05, "loss": 1.2778, "step": 10866 }, { "epoch": 27.1675, "grad_norm": 9.3125, "learning_rate": 1.708993625088146e-05, "loss": 1.5761, "step": 10867 }, { "epoch": 27.17, "grad_norm": 5.96875, "learning_rate": 1.708936522302433e-05, "loss": 1.3957, "step": 10868 }, { "epoch": 27.1725, "grad_norm": 7.71875, "learning_rate": 1.7088794148689602e-05, "loss": 1.3821, "step": 10869 }, { "epoch": 27.175, "grad_norm": 7.46875, "learning_rate": 1.7088223027881024e-05, "loss": 1.4782, "step": 10870 }, { "epoch": 27.1775, "grad_norm": 7.53125, "learning_rate": 1.708765186060234e-05, "loss": 1.3489, "step": 10871 }, { "epoch": 27.18, "grad_norm": 5.84375, "learning_rate": 1.7087080646857293e-05, "loss": 1.4969, "step": 10872 }, { "epoch": 27.1825, "grad_norm": 13.6875, "learning_rate": 1.7086509386649627e-05, "loss": 1.3227, "step": 10873 }, { "epoch": 27.185, "grad_norm": 7.40625, "learning_rate": 1.7085938079983094e-05, "loss": 1.3173, "step": 10874 }, { "epoch": 27.1875, "grad_norm": 8.8125, "learning_rate": 1.708536672686143e-05, "loss": 1.4193, "step": 10875 }, { "epoch": 27.19, "grad_norm": 7.1875, "learning_rate": 1.7084795327288387e-05, "loss": 1.285, "step": 10876 }, { "epoch": 27.1925, "grad_norm": 5.84375, "learning_rate": 1.708422388126771e-05, "loss": 1.3017, "step": 10877 }, { "epoch": 27.195, "grad_norm": 8.875, "learning_rate": 1.7083652388803145e-05, "loss": 1.3966, "step": 10878 }, { "epoch": 27.1975, "grad_norm": 8.75, "learning_rate": 1.7083080849898434e-05, "loss": 1.558, "step": 10879 }, { "epoch": 27.2, "grad_norm": 11.25, "learning_rate": 1.7082509264557333e-05, "loss": 1.2062, "step": 10880 }, { "epoch": 27.2025, "grad_norm": 9.0, "learning_rate": 1.7081937632783584e-05, "loss": 1.2997, "step": 10881 }, { "epoch": 27.205, "grad_norm": 8.625, "learning_rate": 1.7081365954580934e-05, "loss": 1.3684, "step": 10882 }, { "epoch": 27.2075, "grad_norm": 7.8125, "learning_rate": 1.7080794229953132e-05, "loss": 1.3917, "step": 10883 }, { "epoch": 27.21, "grad_norm": 7.53125, "learning_rate": 1.7080222458903928e-05, "loss": 1.3445, "step": 10884 }, { "epoch": 27.2125, "grad_norm": 8.9375, "learning_rate": 1.707965064143707e-05, "loss": 1.3268, "step": 10885 }, { "epoch": 27.215, "grad_norm": 12.3125, "learning_rate": 1.70790787775563e-05, "loss": 1.2441, "step": 10886 }, { "epoch": 27.2175, "grad_norm": 7.28125, "learning_rate": 1.7078506867265382e-05, "loss": 1.5937, "step": 10887 }, { "epoch": 27.22, "grad_norm": 7.8125, "learning_rate": 1.707793491056805e-05, "loss": 1.3075, "step": 10888 }, { "epoch": 27.2225, "grad_norm": 10.5, "learning_rate": 1.707736290746806e-05, "loss": 1.3183, "step": 10889 }, { "epoch": 27.225, "grad_norm": 9.25, "learning_rate": 1.7076790857969164e-05, "loss": 1.4574, "step": 10890 }, { "epoch": 27.2275, "grad_norm": 8.9375, "learning_rate": 1.7076218762075106e-05, "loss": 1.4004, "step": 10891 }, { "epoch": 27.23, "grad_norm": 6.5625, "learning_rate": 1.7075646619789646e-05, "loss": 1.4414, "step": 10892 }, { "epoch": 27.2325, "grad_norm": 4.0625, "learning_rate": 1.707507443111653e-05, "loss": 1.3113, "step": 10893 }, { "epoch": 27.235, "grad_norm": 7.5625, "learning_rate": 1.7074502196059507e-05, "loss": 1.4139, "step": 10894 }, { "epoch": 27.2375, "grad_norm": 7.96875, "learning_rate": 1.707392991462233e-05, "loss": 1.4041, "step": 10895 }, { "epoch": 27.24, "grad_norm": 6.9375, "learning_rate": 1.7073357586808753e-05, "loss": 1.349, "step": 10896 }, { "epoch": 27.2425, "grad_norm": 5.96875, "learning_rate": 1.7072785212622528e-05, "loss": 1.4108, "step": 10897 }, { "epoch": 27.245, "grad_norm": 6.0625, "learning_rate": 1.7072212792067402e-05, "loss": 1.4293, "step": 10898 }, { "epoch": 27.2475, "grad_norm": 5.53125, "learning_rate": 1.7071640325147133e-05, "loss": 1.3675, "step": 10899 }, { "epoch": 27.25, "grad_norm": 5.9375, "learning_rate": 1.7071067811865477e-05, "loss": 1.5285, "step": 10900 }, { "epoch": 27.2525, "grad_norm": 8.625, "learning_rate": 1.707049525222618e-05, "loss": 1.305, "step": 10901 }, { "epoch": 27.255, "grad_norm": 6.125, "learning_rate": 1.7069922646233e-05, "loss": 1.5045, "step": 10902 }, { "epoch": 27.2575, "grad_norm": 5.09375, "learning_rate": 1.706934999388969e-05, "loss": 1.381, "step": 10903 }, { "epoch": 27.26, "grad_norm": 7.125, "learning_rate": 1.7068777295200007e-05, "loss": 1.3239, "step": 10904 }, { "epoch": 27.2625, "grad_norm": 5.90625, "learning_rate": 1.7068204550167697e-05, "loss": 1.3345, "step": 10905 }, { "epoch": 27.265, "grad_norm": 5.75, "learning_rate": 1.7067631758796526e-05, "loss": 1.3434, "step": 10906 }, { "epoch": 27.2675, "grad_norm": 5.5, "learning_rate": 1.706705892109024e-05, "loss": 1.2383, "step": 10907 }, { "epoch": 27.27, "grad_norm": 6.125, "learning_rate": 1.70664860370526e-05, "loss": 1.3289, "step": 10908 }, { "epoch": 27.2725, "grad_norm": 7.375, "learning_rate": 1.706591310668736e-05, "loss": 1.3933, "step": 10909 }, { "epoch": 27.275, "grad_norm": 5.1875, "learning_rate": 1.706534012999828e-05, "loss": 1.2357, "step": 10910 }, { "epoch": 27.2775, "grad_norm": 6.84375, "learning_rate": 1.7064767106989114e-05, "loss": 1.5527, "step": 10911 }, { "epoch": 27.28, "grad_norm": 5.40625, "learning_rate": 1.706419403766361e-05, "loss": 1.2739, "step": 10912 }, { "epoch": 27.2825, "grad_norm": 6.9375, "learning_rate": 1.7063620922025536e-05, "loss": 1.4316, "step": 10913 }, { "epoch": 27.285, "grad_norm": 6.8125, "learning_rate": 1.7063047760078647e-05, "loss": 1.5004, "step": 10914 }, { "epoch": 27.2875, "grad_norm": 11.25, "learning_rate": 1.70624745518267e-05, "loss": 1.3928, "step": 10915 }, { "epoch": 27.29, "grad_norm": 6.65625, "learning_rate": 1.706190129727345e-05, "loss": 1.3737, "step": 10916 }, { "epoch": 27.2925, "grad_norm": 7.625, "learning_rate": 1.706132799642266e-05, "loss": 1.3372, "step": 10917 }, { "epoch": 27.295, "grad_norm": 5.5, "learning_rate": 1.7060754649278085e-05, "loss": 1.6055, "step": 10918 }, { "epoch": 27.2975, "grad_norm": 7.25, "learning_rate": 1.7060181255843484e-05, "loss": 1.3124, "step": 10919 }, { "epoch": 27.3, "grad_norm": 8.6875, "learning_rate": 1.705960781612262e-05, "loss": 1.4936, "step": 10920 }, { "epoch": 27.3025, "grad_norm": 6.59375, "learning_rate": 1.705903433011925e-05, "loss": 1.4181, "step": 10921 }, { "epoch": 27.305, "grad_norm": 11.125, "learning_rate": 1.7058460797837128e-05, "loss": 1.5315, "step": 10922 }, { "epoch": 27.3075, "grad_norm": 8.6875, "learning_rate": 1.7057887219280024e-05, "loss": 1.4477, "step": 10923 }, { "epoch": 27.31, "grad_norm": 7.96875, "learning_rate": 1.7057313594451696e-05, "loss": 1.4498, "step": 10924 }, { "epoch": 27.3125, "grad_norm": 6.34375, "learning_rate": 1.7056739923355897e-05, "loss": 1.4947, "step": 10925 }, { "epoch": 27.315, "grad_norm": 6.375, "learning_rate": 1.7056166205996395e-05, "loss": 1.4359, "step": 10926 }, { "epoch": 27.3175, "grad_norm": 5.625, "learning_rate": 1.705559244237695e-05, "loss": 1.5033, "step": 10927 }, { "epoch": 27.32, "grad_norm": 10.4375, "learning_rate": 1.7055018632501326e-05, "loss": 1.4013, "step": 10928 }, { "epoch": 27.3225, "grad_norm": 7.53125, "learning_rate": 1.705444477637328e-05, "loss": 1.266, "step": 10929 }, { "epoch": 27.325, "grad_norm": 4.84375, "learning_rate": 1.7053870873996575e-05, "loss": 1.3735, "step": 10930 }, { "epoch": 27.3275, "grad_norm": 7.90625, "learning_rate": 1.7053296925374978e-05, "loss": 1.4185, "step": 10931 }, { "epoch": 27.33, "grad_norm": 5.625, "learning_rate": 1.7052722930512246e-05, "loss": 1.3025, "step": 10932 }, { "epoch": 27.3325, "grad_norm": 6.21875, "learning_rate": 1.7052148889412147e-05, "loss": 1.3577, "step": 10933 }, { "epoch": 27.335, "grad_norm": 8.875, "learning_rate": 1.705157480207844e-05, "loss": 1.3803, "step": 10934 }, { "epoch": 27.3375, "grad_norm": 6.6875, "learning_rate": 1.7051000668514892e-05, "loss": 1.3379, "step": 10935 }, { "epoch": 27.34, "grad_norm": 6.84375, "learning_rate": 1.7050426488725264e-05, "loss": 1.4437, "step": 10936 }, { "epoch": 27.3425, "grad_norm": 8.25, "learning_rate": 1.7049852262713323e-05, "loss": 1.4566, "step": 10937 }, { "epoch": 27.345, "grad_norm": 7.25, "learning_rate": 1.7049277990482832e-05, "loss": 1.3469, "step": 10938 }, { "epoch": 27.3475, "grad_norm": 7.875, "learning_rate": 1.704870367203756e-05, "loss": 1.2961, "step": 10939 }, { "epoch": 27.35, "grad_norm": 8.5, "learning_rate": 1.7048129307381266e-05, "loss": 1.497, "step": 10940 }, { "epoch": 27.3525, "grad_norm": 4.4375, "learning_rate": 1.7047554896517717e-05, "loss": 1.3085, "step": 10941 }, { "epoch": 27.355, "grad_norm": 6.875, "learning_rate": 1.704698043945068e-05, "loss": 1.457, "step": 10942 }, { "epoch": 27.3575, "grad_norm": 7.15625, "learning_rate": 1.7046405936183928e-05, "loss": 1.4604, "step": 10943 }, { "epoch": 27.36, "grad_norm": 5.875, "learning_rate": 1.7045831386721213e-05, "loss": 1.326, "step": 10944 }, { "epoch": 27.3625, "grad_norm": 5.96875, "learning_rate": 1.7045256791066313e-05, "loss": 1.3914, "step": 10945 }, { "epoch": 27.365, "grad_norm": 7.53125, "learning_rate": 1.704468214922299e-05, "loss": 1.1142, "step": 10946 }, { "epoch": 27.3675, "grad_norm": 8.75, "learning_rate": 1.704410746119501e-05, "loss": 1.3464, "step": 10947 }, { "epoch": 27.37, "grad_norm": 5.78125, "learning_rate": 1.704353272698615e-05, "loss": 1.3664, "step": 10948 }, { "epoch": 27.3725, "grad_norm": 7.09375, "learning_rate": 1.7042957946600168e-05, "loss": 1.2724, "step": 10949 }, { "epoch": 27.375, "grad_norm": 7.21875, "learning_rate": 1.7042383120040837e-05, "loss": 1.176, "step": 10950 }, { "epoch": 27.3775, "grad_norm": 8.9375, "learning_rate": 1.704180824731192e-05, "loss": 1.3184, "step": 10951 }, { "epoch": 27.38, "grad_norm": 9.5625, "learning_rate": 1.7041233328417194e-05, "loss": 1.3641, "step": 10952 }, { "epoch": 27.3825, "grad_norm": 7.125, "learning_rate": 1.7040658363360424e-05, "loss": 1.5878, "step": 10953 }, { "epoch": 27.385, "grad_norm": 6.4375, "learning_rate": 1.7040083352145376e-05, "loss": 1.526, "step": 10954 }, { "epoch": 27.3875, "grad_norm": 5.96875, "learning_rate": 1.7039508294775824e-05, "loss": 1.4633, "step": 10955 }, { "epoch": 27.39, "grad_norm": 6.875, "learning_rate": 1.703893319125554e-05, "loss": 1.391, "step": 10956 }, { "epoch": 27.3925, "grad_norm": 7.53125, "learning_rate": 1.7038358041588292e-05, "loss": 1.4478, "step": 10957 }, { "epoch": 27.395, "grad_norm": 8.3125, "learning_rate": 1.703778284577785e-05, "loss": 1.4126, "step": 10958 }, { "epoch": 27.3975, "grad_norm": 7.1875, "learning_rate": 1.703720760382798e-05, "loss": 1.4024, "step": 10959 }, { "epoch": 27.4, "grad_norm": 6.125, "learning_rate": 1.7036632315742464e-05, "loss": 1.378, "step": 10960 }, { "epoch": 27.4025, "grad_norm": 8.125, "learning_rate": 1.7036056981525066e-05, "loss": 1.4007, "step": 10961 }, { "epoch": 27.405, "grad_norm": 7.84375, "learning_rate": 1.7035481601179562e-05, "loss": 1.5392, "step": 10962 }, { "epoch": 27.4075, "grad_norm": 9.8125, "learning_rate": 1.703490617470972e-05, "loss": 1.4976, "step": 10963 }, { "epoch": 27.41, "grad_norm": 8.8125, "learning_rate": 1.7034330702119314e-05, "loss": 1.4952, "step": 10964 }, { "epoch": 27.4125, "grad_norm": 5.40625, "learning_rate": 1.7033755183412123e-05, "loss": 1.328, "step": 10965 }, { "epoch": 27.415, "grad_norm": 5.84375, "learning_rate": 1.703317961859191e-05, "loss": 1.2631, "step": 10966 }, { "epoch": 27.4175, "grad_norm": 8.5, "learning_rate": 1.703260400766245e-05, "loss": 1.3515, "step": 10967 }, { "epoch": 27.42, "grad_norm": 5.96875, "learning_rate": 1.7032028350627525e-05, "loss": 1.3118, "step": 10968 }, { "epoch": 27.4225, "grad_norm": 7.25, "learning_rate": 1.70314526474909e-05, "loss": 1.5311, "step": 10969 }, { "epoch": 27.425, "grad_norm": 7.4375, "learning_rate": 1.7030876898256354e-05, "loss": 1.5801, "step": 10970 }, { "epoch": 27.4275, "grad_norm": 5.9375, "learning_rate": 1.703030110292766e-05, "loss": 1.1621, "step": 10971 }, { "epoch": 27.43, "grad_norm": 23.125, "learning_rate": 1.7029725261508594e-05, "loss": 1.3216, "step": 10972 }, { "epoch": 27.4325, "grad_norm": 11.0, "learning_rate": 1.7029149374002932e-05, "loss": 1.4229, "step": 10973 }, { "epoch": 27.435, "grad_norm": 10.125, "learning_rate": 1.7028573440414447e-05, "loss": 1.39, "step": 10974 }, { "epoch": 27.4375, "grad_norm": 5.8125, "learning_rate": 1.7027997460746914e-05, "loss": 1.3578, "step": 10975 }, { "epoch": 27.44, "grad_norm": 7.375, "learning_rate": 1.7027421435004114e-05, "loss": 1.3813, "step": 10976 }, { "epoch": 27.4425, "grad_norm": 5.125, "learning_rate": 1.7026845363189816e-05, "loss": 1.3588, "step": 10977 }, { "epoch": 27.445, "grad_norm": 5.34375, "learning_rate": 1.7026269245307804e-05, "loss": 1.3197, "step": 10978 }, { "epoch": 27.4475, "grad_norm": 7.21875, "learning_rate": 1.702569308136185e-05, "loss": 1.3027, "step": 10979 }, { "epoch": 27.45, "grad_norm": 5.8125, "learning_rate": 1.7025116871355737e-05, "loss": 1.271, "step": 10980 }, { "epoch": 27.4525, "grad_norm": 6.15625, "learning_rate": 1.7024540615293236e-05, "loss": 1.622, "step": 10981 }, { "epoch": 27.455, "grad_norm": 7.15625, "learning_rate": 1.702396431317813e-05, "loss": 1.3562, "step": 10982 }, { "epoch": 27.4575, "grad_norm": 8.3125, "learning_rate": 1.7023387965014193e-05, "loss": 1.4658, "step": 10983 }, { "epoch": 27.46, "grad_norm": 7.46875, "learning_rate": 1.7022811570805206e-05, "loss": 1.4826, "step": 10984 }, { "epoch": 27.4625, "grad_norm": 6.40625, "learning_rate": 1.7022235130554947e-05, "loss": 1.5083, "step": 10985 }, { "epoch": 27.465, "grad_norm": 8.1875, "learning_rate": 1.7021658644267197e-05, "loss": 1.3524, "step": 10986 }, { "epoch": 27.4675, "grad_norm": 6.0, "learning_rate": 1.702108211194573e-05, "loss": 1.3567, "step": 10987 }, { "epoch": 27.47, "grad_norm": 9.9375, "learning_rate": 1.7020505533594332e-05, "loss": 1.4135, "step": 10988 }, { "epoch": 27.4725, "grad_norm": 7.3125, "learning_rate": 1.7019928909216783e-05, "loss": 1.446, "step": 10989 }, { "epoch": 27.475, "grad_norm": 6.3125, "learning_rate": 1.701935223881686e-05, "loss": 1.392, "step": 10990 }, { "epoch": 27.4775, "grad_norm": 5.3125, "learning_rate": 1.701877552239834e-05, "loss": 1.3998, "step": 10991 }, { "epoch": 27.48, "grad_norm": 5.5, "learning_rate": 1.7018198759965018e-05, "loss": 1.406, "step": 10992 }, { "epoch": 27.4825, "grad_norm": 9.0625, "learning_rate": 1.7017621951520657e-05, "loss": 1.4034, "step": 10993 }, { "epoch": 27.485, "grad_norm": 5.15625, "learning_rate": 1.701704509706905e-05, "loss": 1.2672, "step": 10994 }, { "epoch": 27.4875, "grad_norm": 7.25, "learning_rate": 1.7016468196613977e-05, "loss": 1.4549, "step": 10995 }, { "epoch": 27.49, "grad_norm": 7.1875, "learning_rate": 1.701589125015922e-05, "loss": 1.333, "step": 10996 }, { "epoch": 27.4925, "grad_norm": 6.34375, "learning_rate": 1.701531425770856e-05, "loss": 1.3856, "step": 10997 }, { "epoch": 27.495, "grad_norm": 6.1875, "learning_rate": 1.7014737219265777e-05, "loss": 1.388, "step": 10998 }, { "epoch": 27.4975, "grad_norm": 10.3125, "learning_rate": 1.7014160134834663e-05, "loss": 1.4895, "step": 10999 }, { "epoch": 27.5, "grad_norm": 5.5, "learning_rate": 1.7013583004418994e-05, "loss": 1.4435, "step": 11000 }, { "epoch": 27.5025, "grad_norm": 9.0625, "learning_rate": 1.701300582802255e-05, "loss": 1.3872, "step": 11001 }, { "epoch": 27.505, "grad_norm": 7.25, "learning_rate": 1.701242860564913e-05, "loss": 1.5093, "step": 11002 }, { "epoch": 27.5075, "grad_norm": 7.8125, "learning_rate": 1.7011851337302504e-05, "loss": 1.4431, "step": 11003 }, { "epoch": 27.51, "grad_norm": 5.625, "learning_rate": 1.701127402298646e-05, "loss": 1.3058, "step": 11004 }, { "epoch": 27.5125, "grad_norm": 8.6875, "learning_rate": 1.7010696662704787e-05, "loss": 1.2904, "step": 11005 }, { "epoch": 27.515, "grad_norm": 7.25, "learning_rate": 1.7010119256461265e-05, "loss": 1.31, "step": 11006 }, { "epoch": 27.5175, "grad_norm": 7.40625, "learning_rate": 1.700954180425968e-05, "loss": 1.2876, "step": 11007 }, { "epoch": 27.52, "grad_norm": 6.5, "learning_rate": 1.7008964306103823e-05, "loss": 1.4452, "step": 11008 }, { "epoch": 27.5225, "grad_norm": 6.40625, "learning_rate": 1.7008386761997476e-05, "loss": 1.2656, "step": 11009 }, { "epoch": 27.525, "grad_norm": 8.3125, "learning_rate": 1.7007809171944423e-05, "loss": 1.3937, "step": 11010 }, { "epoch": 27.5275, "grad_norm": 9.1875, "learning_rate": 1.7007231535948456e-05, "loss": 1.6112, "step": 11011 }, { "epoch": 27.53, "grad_norm": 4.875, "learning_rate": 1.7006653854013362e-05, "loss": 1.2409, "step": 11012 }, { "epoch": 27.5325, "grad_norm": 6.84375, "learning_rate": 1.7006076126142922e-05, "loss": 1.4158, "step": 11013 }, { "epoch": 27.535, "grad_norm": 8.0, "learning_rate": 1.7005498352340928e-05, "loss": 1.3549, "step": 11014 }, { "epoch": 27.5375, "grad_norm": 7.34375, "learning_rate": 1.7004920532611168e-05, "loss": 1.4427, "step": 11015 }, { "epoch": 27.54, "grad_norm": 7.8125, "learning_rate": 1.7004342666957426e-05, "loss": 1.4141, "step": 11016 }, { "epoch": 27.5425, "grad_norm": 6.96875, "learning_rate": 1.7003764755383497e-05, "loss": 1.3634, "step": 11017 }, { "epoch": 27.545, "grad_norm": 8.5625, "learning_rate": 1.7003186797893166e-05, "loss": 1.4219, "step": 11018 }, { "epoch": 27.5475, "grad_norm": 7.625, "learning_rate": 1.700260879449022e-05, "loss": 1.4183, "step": 11019 }, { "epoch": 27.55, "grad_norm": 9.5625, "learning_rate": 1.7002030745178455e-05, "loss": 1.5827, "step": 11020 }, { "epoch": 27.5525, "grad_norm": 8.3125, "learning_rate": 1.7001452649961654e-05, "loss": 1.4947, "step": 11021 }, { "epoch": 27.555, "grad_norm": 5.875, "learning_rate": 1.700087450884361e-05, "loss": 1.5231, "step": 11022 }, { "epoch": 27.5575, "grad_norm": 6.71875, "learning_rate": 1.7000296321828115e-05, "loss": 1.3957, "step": 11023 }, { "epoch": 27.56, "grad_norm": 21.5, "learning_rate": 1.6999718088918956e-05, "loss": 1.2792, "step": 11024 }, { "epoch": 27.5625, "grad_norm": 9.25, "learning_rate": 1.6999139810119927e-05, "loss": 1.4718, "step": 11025 }, { "epoch": 27.565, "grad_norm": 9.8125, "learning_rate": 1.6998561485434813e-05, "loss": 1.535, "step": 11026 }, { "epoch": 27.5675, "grad_norm": 6.46875, "learning_rate": 1.6997983114867414e-05, "loss": 1.3631, "step": 11027 }, { "epoch": 27.57, "grad_norm": 5.96875, "learning_rate": 1.6997404698421517e-05, "loss": 1.5197, "step": 11028 }, { "epoch": 27.5725, "grad_norm": 7.125, "learning_rate": 1.6996826236100914e-05, "loss": 1.3573, "step": 11029 }, { "epoch": 27.575, "grad_norm": 7.46875, "learning_rate": 1.69962477279094e-05, "loss": 1.5771, "step": 11030 }, { "epoch": 27.5775, "grad_norm": 8.9375, "learning_rate": 1.699566917385076e-05, "loss": 1.4041, "step": 11031 }, { "epoch": 27.58, "grad_norm": 15.1875, "learning_rate": 1.69950905739288e-05, "loss": 1.4343, "step": 11032 }, { "epoch": 27.5825, "grad_norm": 6.28125, "learning_rate": 1.6994511928147304e-05, "loss": 1.2015, "step": 11033 }, { "epoch": 27.585, "grad_norm": 5.59375, "learning_rate": 1.6993933236510068e-05, "loss": 1.2816, "step": 11034 }, { "epoch": 27.5875, "grad_norm": 8.9375, "learning_rate": 1.6993354499020882e-05, "loss": 1.3437, "step": 11035 }, { "epoch": 27.59, "grad_norm": 10.6875, "learning_rate": 1.6992775715683544e-05, "loss": 1.4197, "step": 11036 }, { "epoch": 27.5925, "grad_norm": 7.375, "learning_rate": 1.6992196886501852e-05, "loss": 1.4327, "step": 11037 }, { "epoch": 27.595, "grad_norm": 8.1875, "learning_rate": 1.6991618011479598e-05, "loss": 1.3501, "step": 11038 }, { "epoch": 27.5975, "grad_norm": 7.03125, "learning_rate": 1.6991039090620573e-05, "loss": 1.2878, "step": 11039 }, { "epoch": 27.6, "grad_norm": 7.0625, "learning_rate": 1.6990460123928577e-05, "loss": 1.2868, "step": 11040 }, { "epoch": 27.6025, "grad_norm": 9.0, "learning_rate": 1.6989881111407404e-05, "loss": 1.4456, "step": 11041 }, { "epoch": 27.605, "grad_norm": 7.84375, "learning_rate": 1.6989302053060846e-05, "loss": 1.5117, "step": 11042 }, { "epoch": 27.6075, "grad_norm": 7.40625, "learning_rate": 1.698872294889271e-05, "loss": 1.5097, "step": 11043 }, { "epoch": 27.61, "grad_norm": 5.84375, "learning_rate": 1.6988143798906783e-05, "loss": 1.419, "step": 11044 }, { "epoch": 27.6125, "grad_norm": 8.0, "learning_rate": 1.6987564603106864e-05, "loss": 1.3937, "step": 11045 }, { "epoch": 27.615, "grad_norm": 6.25, "learning_rate": 1.6986985361496753e-05, "loss": 1.346, "step": 11046 }, { "epoch": 27.6175, "grad_norm": 6.375, "learning_rate": 1.6986406074080245e-05, "loss": 1.2932, "step": 11047 }, { "epoch": 27.62, "grad_norm": 6.25, "learning_rate": 1.698582674086114e-05, "loss": 1.2998, "step": 11048 }, { "epoch": 27.6225, "grad_norm": 7.3125, "learning_rate": 1.698524736184323e-05, "loss": 1.3744, "step": 11049 }, { "epoch": 27.625, "grad_norm": 5.96875, "learning_rate": 1.698466793703032e-05, "loss": 1.3909, "step": 11050 }, { "epoch": 27.6275, "grad_norm": 8.4375, "learning_rate": 1.6984088466426207e-05, "loss": 1.4797, "step": 11051 }, { "epoch": 27.63, "grad_norm": 10.6875, "learning_rate": 1.6983508950034687e-05, "loss": 1.4292, "step": 11052 }, { "epoch": 27.6325, "grad_norm": 8.0625, "learning_rate": 1.6982929387859567e-05, "loss": 1.361, "step": 11053 }, { "epoch": 27.635, "grad_norm": 7.53125, "learning_rate": 1.6982349779904637e-05, "loss": 1.4934, "step": 11054 }, { "epoch": 27.6375, "grad_norm": 6.1875, "learning_rate": 1.69817701261737e-05, "loss": 1.3906, "step": 11055 }, { "epoch": 27.64, "grad_norm": 5.78125, "learning_rate": 1.698119042667056e-05, "loss": 1.3729, "step": 11056 }, { "epoch": 27.6425, "grad_norm": 7.75, "learning_rate": 1.6980610681399018e-05, "loss": 1.4269, "step": 11057 }, { "epoch": 27.645, "grad_norm": 7.71875, "learning_rate": 1.698003089036287e-05, "loss": 1.3742, "step": 11058 }, { "epoch": 27.6475, "grad_norm": 6.9375, "learning_rate": 1.6979451053565915e-05, "loss": 1.461, "step": 11059 }, { "epoch": 27.65, "grad_norm": 8.125, "learning_rate": 1.6978871171011963e-05, "loss": 1.3009, "step": 11060 }, { "epoch": 27.6525, "grad_norm": 7.84375, "learning_rate": 1.6978291242704808e-05, "loss": 1.4405, "step": 11061 }, { "epoch": 27.655, "grad_norm": 8.875, "learning_rate": 1.6977711268648255e-05, "loss": 1.5611, "step": 11062 }, { "epoch": 27.6575, "grad_norm": 4.6875, "learning_rate": 1.6977131248846105e-05, "loss": 1.392, "step": 11063 }, { "epoch": 27.66, "grad_norm": 6.78125, "learning_rate": 1.6976551183302164e-05, "loss": 1.277, "step": 11064 }, { "epoch": 27.6625, "grad_norm": 5.71875, "learning_rate": 1.6975971072020232e-05, "loss": 1.4187, "step": 11065 }, { "epoch": 27.665, "grad_norm": 7.59375, "learning_rate": 1.6975390915004112e-05, "loss": 1.2724, "step": 11066 }, { "epoch": 27.6675, "grad_norm": 9.4375, "learning_rate": 1.697481071225761e-05, "loss": 1.295, "step": 11067 }, { "epoch": 27.67, "grad_norm": 8.4375, "learning_rate": 1.6974230463784528e-05, "loss": 1.3401, "step": 11068 }, { "epoch": 27.6725, "grad_norm": 6.90625, "learning_rate": 1.697365016958867e-05, "loss": 1.4503, "step": 11069 }, { "epoch": 27.675, "grad_norm": 10.0, "learning_rate": 1.6973069829673837e-05, "loss": 1.4116, "step": 11070 }, { "epoch": 27.6775, "grad_norm": 6.34375, "learning_rate": 1.697248944404384e-05, "loss": 1.5382, "step": 11071 }, { "epoch": 27.68, "grad_norm": 5.90625, "learning_rate": 1.6971909012702483e-05, "loss": 1.2581, "step": 11072 }, { "epoch": 27.6825, "grad_norm": 7.75, "learning_rate": 1.6971328535653568e-05, "loss": 1.2731, "step": 11073 }, { "epoch": 27.685, "grad_norm": 9.125, "learning_rate": 1.6970748012900903e-05, "loss": 1.5453, "step": 11074 }, { "epoch": 27.6875, "grad_norm": 7.21875, "learning_rate": 1.697016744444829e-05, "loss": 1.3915, "step": 11075 }, { "epoch": 27.69, "grad_norm": 7.96875, "learning_rate": 1.696958683029954e-05, "loss": 1.4315, "step": 11076 }, { "epoch": 27.6925, "grad_norm": 6.375, "learning_rate": 1.6969006170458458e-05, "loss": 1.3875, "step": 11077 }, { "epoch": 27.695, "grad_norm": 11.0625, "learning_rate": 1.696842546492885e-05, "loss": 1.4669, "step": 11078 }, { "epoch": 27.6975, "grad_norm": 8.125, "learning_rate": 1.6967844713714526e-05, "loss": 1.4111, "step": 11079 }, { "epoch": 27.7, "grad_norm": 7.4375, "learning_rate": 1.696726391681929e-05, "loss": 1.4253, "step": 11080 }, { "epoch": 27.7025, "grad_norm": 5.4375, "learning_rate": 1.6966683074246946e-05, "loss": 1.1776, "step": 11081 }, { "epoch": 27.705, "grad_norm": 8.8125, "learning_rate": 1.6966102186001312e-05, "loss": 1.3601, "step": 11082 }, { "epoch": 27.7075, "grad_norm": 7.15625, "learning_rate": 1.6965521252086186e-05, "loss": 1.4004, "step": 11083 }, { "epoch": 27.71, "grad_norm": 6.09375, "learning_rate": 1.6964940272505387e-05, "loss": 1.4328, "step": 11084 }, { "epoch": 27.7125, "grad_norm": 5.46875, "learning_rate": 1.6964359247262715e-05, "loss": 1.3555, "step": 11085 }, { "epoch": 27.715, "grad_norm": 6.40625, "learning_rate": 1.6963778176361982e-05, "loss": 1.4881, "step": 11086 }, { "epoch": 27.7175, "grad_norm": 7.65625, "learning_rate": 1.6963197059806996e-05, "loss": 1.4356, "step": 11087 }, { "epoch": 27.72, "grad_norm": 7.375, "learning_rate": 1.6962615897601573e-05, "loss": 1.3728, "step": 11088 }, { "epoch": 27.7225, "grad_norm": 5.4375, "learning_rate": 1.6962034689749514e-05, "loss": 1.2227, "step": 11089 }, { "epoch": 27.725, "grad_norm": 6.46875, "learning_rate": 1.696145343625464e-05, "loss": 1.2858, "step": 11090 }, { "epoch": 27.7275, "grad_norm": 8.875, "learning_rate": 1.6960872137120755e-05, "loss": 1.3414, "step": 11091 }, { "epoch": 27.73, "grad_norm": 7.46875, "learning_rate": 1.6960290792351668e-05, "loss": 1.5122, "step": 11092 }, { "epoch": 27.7325, "grad_norm": 6.34375, "learning_rate": 1.6959709401951195e-05, "loss": 1.4021, "step": 11093 }, { "epoch": 27.735, "grad_norm": 9.6875, "learning_rate": 1.6959127965923144e-05, "loss": 1.318, "step": 11094 }, { "epoch": 27.7375, "grad_norm": 5.53125, "learning_rate": 1.695854648427133e-05, "loss": 1.3948, "step": 11095 }, { "epoch": 27.74, "grad_norm": 6.375, "learning_rate": 1.6957964956999563e-05, "loss": 1.5425, "step": 11096 }, { "epoch": 27.7425, "grad_norm": 7.375, "learning_rate": 1.695738338411166e-05, "loss": 1.3939, "step": 11097 }, { "epoch": 27.745, "grad_norm": 5.65625, "learning_rate": 1.695680176561142e-05, "loss": 1.3702, "step": 11098 }, { "epoch": 27.7475, "grad_norm": 9.5, "learning_rate": 1.6956220101502675e-05, "loss": 1.3031, "step": 11099 }, { "epoch": 27.75, "grad_norm": 5.03125, "learning_rate": 1.695563839178923e-05, "loss": 1.3307, "step": 11100 }, { "epoch": 27.7525, "grad_norm": 5.8125, "learning_rate": 1.695505663647489e-05, "loss": 1.2659, "step": 11101 }, { "epoch": 27.755, "grad_norm": 7.0, "learning_rate": 1.6954474835563485e-05, "loss": 1.1333, "step": 11102 }, { "epoch": 27.7575, "grad_norm": 7.4375, "learning_rate": 1.6953892989058816e-05, "loss": 1.4743, "step": 11103 }, { "epoch": 27.76, "grad_norm": 5.71875, "learning_rate": 1.6953311096964706e-05, "loss": 1.3669, "step": 11104 }, { "epoch": 27.7625, "grad_norm": 9.5625, "learning_rate": 1.6952729159284963e-05, "loss": 1.2977, "step": 11105 }, { "epoch": 27.765, "grad_norm": 7.28125, "learning_rate": 1.695214717602341e-05, "loss": 1.3673, "step": 11106 }, { "epoch": 27.7675, "grad_norm": 5.9375, "learning_rate": 1.6951565147183856e-05, "loss": 1.3738, "step": 11107 }, { "epoch": 27.77, "grad_norm": 5.8125, "learning_rate": 1.6950983072770123e-05, "loss": 1.568, "step": 11108 }, { "epoch": 27.7725, "grad_norm": 5.03125, "learning_rate": 1.6950400952786017e-05, "loss": 1.2459, "step": 11109 }, { "epoch": 27.775, "grad_norm": 7.71875, "learning_rate": 1.6949818787235366e-05, "loss": 1.4418, "step": 11110 }, { "epoch": 27.7775, "grad_norm": 7.15625, "learning_rate": 1.6949236576121977e-05, "loss": 1.405, "step": 11111 }, { "epoch": 27.78, "grad_norm": 6.96875, "learning_rate": 1.6948654319449674e-05, "loss": 1.4345, "step": 11112 }, { "epoch": 27.7825, "grad_norm": 7.78125, "learning_rate": 1.694807201722227e-05, "loss": 1.5288, "step": 11113 }, { "epoch": 27.785, "grad_norm": 8.75, "learning_rate": 1.6947489669443582e-05, "loss": 1.3672, "step": 11114 }, { "epoch": 27.7875, "grad_norm": 9.375, "learning_rate": 1.6946907276117434e-05, "loss": 1.4951, "step": 11115 }, { "epoch": 27.79, "grad_norm": 5.71875, "learning_rate": 1.6946324837247636e-05, "loss": 1.2635, "step": 11116 }, { "epoch": 27.7925, "grad_norm": 6.59375, "learning_rate": 1.6945742352838013e-05, "loss": 1.472, "step": 11117 }, { "epoch": 27.795, "grad_norm": 8.4375, "learning_rate": 1.6945159822892382e-05, "loss": 1.2937, "step": 11118 }, { "epoch": 27.7975, "grad_norm": 5.625, "learning_rate": 1.6944577247414557e-05, "loss": 1.359, "step": 11119 }, { "epoch": 27.8, "grad_norm": 6.84375, "learning_rate": 1.6943994626408365e-05, "loss": 1.4597, "step": 11120 }, { "epoch": 27.8025, "grad_norm": 5.09375, "learning_rate": 1.6943411959877622e-05, "loss": 1.4815, "step": 11121 }, { "epoch": 27.805, "grad_norm": 8.1875, "learning_rate": 1.6942829247826146e-05, "loss": 1.4371, "step": 11122 }, { "epoch": 27.8075, "grad_norm": 7.90625, "learning_rate": 1.6942246490257758e-05, "loss": 1.2126, "step": 11123 }, { "epoch": 27.81, "grad_norm": 7.96875, "learning_rate": 1.694166368717628e-05, "loss": 1.3345, "step": 11124 }, { "epoch": 27.8125, "grad_norm": 9.3125, "learning_rate": 1.6941080838585537e-05, "loss": 1.3596, "step": 11125 }, { "epoch": 27.815, "grad_norm": 9.375, "learning_rate": 1.6940497944489346e-05, "loss": 1.4332, "step": 11126 }, { "epoch": 27.8175, "grad_norm": 6.25, "learning_rate": 1.6939915004891525e-05, "loss": 1.3836, "step": 11127 }, { "epoch": 27.82, "grad_norm": 5.6875, "learning_rate": 1.69393320197959e-05, "loss": 1.3063, "step": 11128 }, { "epoch": 27.8225, "grad_norm": 5.28125, "learning_rate": 1.6938748989206288e-05, "loss": 1.2666, "step": 11129 }, { "epoch": 27.825, "grad_norm": 7.90625, "learning_rate": 1.6938165913126522e-05, "loss": 1.4281, "step": 11130 }, { "epoch": 27.8275, "grad_norm": 9.3125, "learning_rate": 1.6937582791560413e-05, "loss": 1.4386, "step": 11131 }, { "epoch": 27.83, "grad_norm": 9.3125, "learning_rate": 1.6936999624511793e-05, "loss": 1.3605, "step": 11132 }, { "epoch": 27.8325, "grad_norm": 4.625, "learning_rate": 1.693641641198448e-05, "loss": 1.3397, "step": 11133 }, { "epoch": 27.835, "grad_norm": 7.625, "learning_rate": 1.6935833153982294e-05, "loss": 1.3853, "step": 11134 }, { "epoch": 27.8375, "grad_norm": 7.84375, "learning_rate": 1.6935249850509066e-05, "loss": 1.4113, "step": 11135 }, { "epoch": 27.84, "grad_norm": 9.5, "learning_rate": 1.6934666501568618e-05, "loss": 1.2152, "step": 11136 }, { "epoch": 27.8425, "grad_norm": 9.75, "learning_rate": 1.6934083107164777e-05, "loss": 1.424, "step": 11137 }, { "epoch": 27.845, "grad_norm": 8.0625, "learning_rate": 1.6933499667301362e-05, "loss": 1.4617, "step": 11138 }, { "epoch": 27.8475, "grad_norm": 7.5, "learning_rate": 1.69329161819822e-05, "loss": 1.277, "step": 11139 }, { "epoch": 27.85, "grad_norm": 6.125, "learning_rate": 1.6932332651211115e-05, "loss": 1.3146, "step": 11140 }, { "epoch": 27.8525, "grad_norm": 7.15625, "learning_rate": 1.6931749074991937e-05, "loss": 1.4, "step": 11141 }, { "epoch": 27.855, "grad_norm": 8.75, "learning_rate": 1.6931165453328492e-05, "loss": 1.3706, "step": 11142 }, { "epoch": 27.8575, "grad_norm": 8.3125, "learning_rate": 1.69305817862246e-05, "loss": 1.4791, "step": 11143 }, { "epoch": 27.86, "grad_norm": 7.6875, "learning_rate": 1.692999807368409e-05, "loss": 1.446, "step": 11144 }, { "epoch": 27.8625, "grad_norm": 6.09375, "learning_rate": 1.692941431571079e-05, "loss": 1.3634, "step": 11145 }, { "epoch": 27.865, "grad_norm": 6.09375, "learning_rate": 1.692883051230853e-05, "loss": 1.4867, "step": 11146 }, { "epoch": 27.8675, "grad_norm": 6.875, "learning_rate": 1.6928246663481135e-05, "loss": 1.3881, "step": 11147 }, { "epoch": 27.87, "grad_norm": 5.875, "learning_rate": 1.692766276923243e-05, "loss": 1.3232, "step": 11148 }, { "epoch": 27.8725, "grad_norm": 6.75, "learning_rate": 1.692707882956625e-05, "loss": 1.5044, "step": 11149 }, { "epoch": 27.875, "grad_norm": 8.9375, "learning_rate": 1.6926494844486412e-05, "loss": 1.3961, "step": 11150 }, { "epoch": 27.8775, "grad_norm": 6.90625, "learning_rate": 1.6925910813996752e-05, "loss": 1.3941, "step": 11151 }, { "epoch": 27.88, "grad_norm": 7.9375, "learning_rate": 1.69253267381011e-05, "loss": 1.3635, "step": 11152 }, { "epoch": 27.8825, "grad_norm": 8.6875, "learning_rate": 1.692474261680328e-05, "loss": 1.4278, "step": 11153 }, { "epoch": 27.885, "grad_norm": 6.3125, "learning_rate": 1.692415845010713e-05, "loss": 1.4465, "step": 11154 }, { "epoch": 27.8875, "grad_norm": 7.34375, "learning_rate": 1.692357423801647e-05, "loss": 1.4469, "step": 11155 }, { "epoch": 27.89, "grad_norm": 6.28125, "learning_rate": 1.6922989980535135e-05, "loss": 1.3567, "step": 11156 }, { "epoch": 27.8925, "grad_norm": 9.875, "learning_rate": 1.6922405677666954e-05, "loss": 1.2946, "step": 11157 }, { "epoch": 27.895, "grad_norm": 6.75, "learning_rate": 1.692182132941576e-05, "loss": 1.2494, "step": 11158 }, { "epoch": 27.8975, "grad_norm": 6.5, "learning_rate": 1.692123693578538e-05, "loss": 1.3292, "step": 11159 }, { "epoch": 27.9, "grad_norm": 7.71875, "learning_rate": 1.692065249677965e-05, "loss": 1.415, "step": 11160 }, { "epoch": 27.9025, "grad_norm": 8.375, "learning_rate": 1.69200680124024e-05, "loss": 1.506, "step": 11161 }, { "epoch": 27.905, "grad_norm": 6.40625, "learning_rate": 1.691948348265746e-05, "loss": 1.2183, "step": 11162 }, { "epoch": 27.9075, "grad_norm": 7.46875, "learning_rate": 1.6918898907548663e-05, "loss": 1.305, "step": 11163 }, { "epoch": 27.91, "grad_norm": 8.625, "learning_rate": 1.691831428707984e-05, "loss": 1.3971, "step": 11164 }, { "epoch": 27.9125, "grad_norm": 6.40625, "learning_rate": 1.691772962125483e-05, "loss": 1.3408, "step": 11165 }, { "epoch": 27.915, "grad_norm": 6.84375, "learning_rate": 1.6917144910077455e-05, "loss": 1.3001, "step": 11166 }, { "epoch": 27.9175, "grad_norm": 6.21875, "learning_rate": 1.691656015355156e-05, "loss": 1.4113, "step": 11167 }, { "epoch": 27.92, "grad_norm": 9.0, "learning_rate": 1.6915975351680968e-05, "loss": 1.6087, "step": 11168 }, { "epoch": 27.9225, "grad_norm": 15.625, "learning_rate": 1.6915390504469523e-05, "loss": 1.2334, "step": 11169 }, { "epoch": 27.925, "grad_norm": 7.75, "learning_rate": 1.6914805611921056e-05, "loss": 1.402, "step": 11170 }, { "epoch": 27.9275, "grad_norm": 7.34375, "learning_rate": 1.6914220674039396e-05, "loss": 1.3851, "step": 11171 }, { "epoch": 27.93, "grad_norm": 5.6875, "learning_rate": 1.6913635690828382e-05, "loss": 1.1981, "step": 11172 }, { "epoch": 27.9325, "grad_norm": 5.75, "learning_rate": 1.6913050662291852e-05, "loss": 1.4487, "step": 11173 }, { "epoch": 27.935, "grad_norm": 11.375, "learning_rate": 1.6912465588433638e-05, "loss": 1.419, "step": 11174 }, { "epoch": 27.9375, "grad_norm": 6.0, "learning_rate": 1.6911880469257576e-05, "loss": 1.3378, "step": 11175 }, { "epoch": 27.94, "grad_norm": 9.75, "learning_rate": 1.6911295304767497e-05, "loss": 1.3918, "step": 11176 }, { "epoch": 27.9425, "grad_norm": 6.625, "learning_rate": 1.6910710094967248e-05, "loss": 1.308, "step": 11177 }, { "epoch": 27.945, "grad_norm": 8.75, "learning_rate": 1.6910124839860664e-05, "loss": 1.5384, "step": 11178 }, { "epoch": 27.9475, "grad_norm": 6.0625, "learning_rate": 1.6909539539451572e-05, "loss": 1.2546, "step": 11179 }, { "epoch": 27.95, "grad_norm": 8.75, "learning_rate": 1.6908954193743816e-05, "loss": 1.5274, "step": 11180 }, { "epoch": 27.9525, "grad_norm": 25.125, "learning_rate": 1.6908368802741233e-05, "loss": 1.442, "step": 11181 }, { "epoch": 27.955, "grad_norm": 7.5625, "learning_rate": 1.690778336644766e-05, "loss": 1.3384, "step": 11182 }, { "epoch": 27.9575, "grad_norm": 4.90625, "learning_rate": 1.6907197884866936e-05, "loss": 1.2466, "step": 11183 }, { "epoch": 27.96, "grad_norm": 7.0625, "learning_rate": 1.69066123580029e-05, "loss": 1.5325, "step": 11184 }, { "epoch": 27.9625, "grad_norm": 6.1875, "learning_rate": 1.6906026785859393e-05, "loss": 1.2942, "step": 11185 }, { "epoch": 27.965, "grad_norm": 8.1875, "learning_rate": 1.6905441168440245e-05, "loss": 1.2748, "step": 11186 }, { "epoch": 27.9675, "grad_norm": 11.1875, "learning_rate": 1.6904855505749304e-05, "loss": 1.5138, "step": 11187 }, { "epoch": 27.97, "grad_norm": 6.84375, "learning_rate": 1.6904269797790404e-05, "loss": 1.3226, "step": 11188 }, { "epoch": 27.9725, "grad_norm": 33.75, "learning_rate": 1.6903684044567392e-05, "loss": 1.539, "step": 11189 }, { "epoch": 27.975, "grad_norm": 7.25, "learning_rate": 1.69030982460841e-05, "loss": 1.4043, "step": 11190 }, { "epoch": 27.9775, "grad_norm": 11.125, "learning_rate": 1.6902512402344373e-05, "loss": 1.2922, "step": 11191 }, { "epoch": 27.98, "grad_norm": 6.9375, "learning_rate": 1.6901926513352052e-05, "loss": 1.4431, "step": 11192 }, { "epoch": 27.9825, "grad_norm": 7.5, "learning_rate": 1.6901340579110974e-05, "loss": 1.4115, "step": 11193 }, { "epoch": 27.985, "grad_norm": 6.15625, "learning_rate": 1.6900754599624988e-05, "loss": 1.2847, "step": 11194 }, { "epoch": 27.9875, "grad_norm": 4.75, "learning_rate": 1.6900168574897928e-05, "loss": 1.3986, "step": 11195 }, { "epoch": 27.99, "grad_norm": 5.0625, "learning_rate": 1.6899582504933637e-05, "loss": 1.4777, "step": 11196 }, { "epoch": 27.9925, "grad_norm": 7.09375, "learning_rate": 1.6898996389735964e-05, "loss": 1.357, "step": 11197 }, { "epoch": 27.995, "grad_norm": 6.8125, "learning_rate": 1.6898410229308744e-05, "loss": 1.3116, "step": 11198 }, { "epoch": 27.9975, "grad_norm": 5.75, "learning_rate": 1.6897824023655826e-05, "loss": 1.4506, "step": 11199 }, { "epoch": 28.0, "grad_norm": 6.5, "learning_rate": 1.6897237772781046e-05, "loss": 1.3732, "step": 11200 }, { "epoch": 28.0025, "grad_norm": 8.0, "learning_rate": 1.689665147668825e-05, "loss": 1.3143, "step": 11201 }, { "epoch": 28.005, "grad_norm": 6.84375, "learning_rate": 1.6896065135381286e-05, "loss": 1.3426, "step": 11202 }, { "epoch": 28.0075, "grad_norm": 7.15625, "learning_rate": 1.689547874886399e-05, "loss": 1.3159, "step": 11203 }, { "epoch": 28.01, "grad_norm": 11.0625, "learning_rate": 1.6894892317140216e-05, "loss": 1.3747, "step": 11204 }, { "epoch": 28.0125, "grad_norm": 6.15625, "learning_rate": 1.6894305840213806e-05, "loss": 1.3496, "step": 11205 }, { "epoch": 28.015, "grad_norm": 7.40625, "learning_rate": 1.6893719318088596e-05, "loss": 1.4792, "step": 11206 }, { "epoch": 28.0175, "grad_norm": 7.09375, "learning_rate": 1.6893132750768442e-05, "loss": 1.4579, "step": 11207 }, { "epoch": 28.02, "grad_norm": 9.5, "learning_rate": 1.6892546138257187e-05, "loss": 1.3655, "step": 11208 }, { "epoch": 28.0225, "grad_norm": 7.15625, "learning_rate": 1.6891959480558672e-05, "loss": 1.4629, "step": 11209 }, { "epoch": 28.025, "grad_norm": 7.625, "learning_rate": 1.689137277767675e-05, "loss": 1.3677, "step": 11210 }, { "epoch": 28.0275, "grad_norm": 10.9375, "learning_rate": 1.689078602961526e-05, "loss": 1.464, "step": 11211 }, { "epoch": 28.03, "grad_norm": 4.9375, "learning_rate": 1.6890199236378056e-05, "loss": 1.3247, "step": 11212 }, { "epoch": 28.0325, "grad_norm": 8.0625, "learning_rate": 1.688961239796898e-05, "loss": 1.505, "step": 11213 }, { "epoch": 28.035, "grad_norm": 8.875, "learning_rate": 1.688902551439188e-05, "loss": 1.4511, "step": 11214 }, { "epoch": 28.0375, "grad_norm": 6.71875, "learning_rate": 1.6888438585650604e-05, "loss": 1.3937, "step": 11215 }, { "epoch": 28.04, "grad_norm": 9.125, "learning_rate": 1.6887851611749005e-05, "loss": 1.3486, "step": 11216 }, { "epoch": 28.0425, "grad_norm": 12.125, "learning_rate": 1.6887264592690923e-05, "loss": 1.3804, "step": 11217 }, { "epoch": 28.045, "grad_norm": 8.625, "learning_rate": 1.6886677528480206e-05, "loss": 1.2826, "step": 11218 }, { "epoch": 28.0475, "grad_norm": 6.96875, "learning_rate": 1.688609041912071e-05, "loss": 1.4186, "step": 11219 }, { "epoch": 28.05, "grad_norm": 5.875, "learning_rate": 1.6885503264616282e-05, "loss": 1.3768, "step": 11220 }, { "epoch": 28.0525, "grad_norm": 8.3125, "learning_rate": 1.688491606497077e-05, "loss": 1.3887, "step": 11221 }, { "epoch": 28.055, "grad_norm": 7.65625, "learning_rate": 1.6884328820188024e-05, "loss": 1.1489, "step": 11222 }, { "epoch": 28.0575, "grad_norm": 8.625, "learning_rate": 1.6883741530271893e-05, "loss": 1.4026, "step": 11223 }, { "epoch": 28.06, "grad_norm": 7.21875, "learning_rate": 1.688315419522623e-05, "loss": 1.473, "step": 11224 }, { "epoch": 28.0625, "grad_norm": 5.28125, "learning_rate": 1.6882566815054882e-05, "loss": 1.5178, "step": 11225 }, { "epoch": 28.065, "grad_norm": 7.5, "learning_rate": 1.68819793897617e-05, "loss": 1.4141, "step": 11226 }, { "epoch": 28.0675, "grad_norm": 8.6875, "learning_rate": 1.688139191935054e-05, "loss": 1.4107, "step": 11227 }, { "epoch": 28.07, "grad_norm": 5.21875, "learning_rate": 1.6880804403825245e-05, "loss": 1.3204, "step": 11228 }, { "epoch": 28.0725, "grad_norm": 5.8125, "learning_rate": 1.6880216843189673e-05, "loss": 1.3978, "step": 11229 }, { "epoch": 28.075, "grad_norm": 7.65625, "learning_rate": 1.6879629237447678e-05, "loss": 1.3483, "step": 11230 }, { "epoch": 28.0775, "grad_norm": 9.25, "learning_rate": 1.6879041586603104e-05, "loss": 1.4644, "step": 11231 }, { "epoch": 28.08, "grad_norm": 6.46875, "learning_rate": 1.6878453890659815e-05, "loss": 1.287, "step": 11232 }, { "epoch": 28.0825, "grad_norm": 6.03125, "learning_rate": 1.6877866149621655e-05, "loss": 1.53, "step": 11233 }, { "epoch": 28.085, "grad_norm": 6.71875, "learning_rate": 1.6877278363492475e-05, "loss": 1.2671, "step": 11234 }, { "epoch": 28.0875, "grad_norm": 8.875, "learning_rate": 1.6876690532276137e-05, "loss": 1.3583, "step": 11235 }, { "epoch": 28.09, "grad_norm": 9.375, "learning_rate": 1.6876102655976492e-05, "loss": 1.3865, "step": 11236 }, { "epoch": 28.0925, "grad_norm": 5.15625, "learning_rate": 1.6875514734597394e-05, "loss": 1.4315, "step": 11237 }, { "epoch": 28.095, "grad_norm": 7.21875, "learning_rate": 1.6874926768142695e-05, "loss": 1.41, "step": 11238 }, { "epoch": 28.0975, "grad_norm": 8.125, "learning_rate": 1.6874338756616252e-05, "loss": 1.5553, "step": 11239 }, { "epoch": 28.1, "grad_norm": 6.34375, "learning_rate": 1.6873750700021917e-05, "loss": 1.3843, "step": 11240 }, { "epoch": 28.1025, "grad_norm": 6.25, "learning_rate": 1.6873162598363548e-05, "loss": 1.3171, "step": 11241 }, { "epoch": 28.105, "grad_norm": 7.8125, "learning_rate": 1.6872574451645002e-05, "loss": 1.4731, "step": 11242 }, { "epoch": 28.1075, "grad_norm": 6.90625, "learning_rate": 1.687198625987013e-05, "loss": 1.382, "step": 11243 }, { "epoch": 28.11, "grad_norm": 6.125, "learning_rate": 1.6871398023042793e-05, "loss": 1.1672, "step": 11244 }, { "epoch": 28.1125, "grad_norm": 6.25, "learning_rate": 1.687080974116685e-05, "loss": 1.2757, "step": 11245 }, { "epoch": 28.115, "grad_norm": 6.40625, "learning_rate": 1.6870221414246143e-05, "loss": 1.2889, "step": 11246 }, { "epoch": 28.1175, "grad_norm": 5.125, "learning_rate": 1.6869633042284544e-05, "loss": 1.5769, "step": 11247 }, { "epoch": 28.12, "grad_norm": 9.375, "learning_rate": 1.686904462528591e-05, "loss": 1.3346, "step": 11248 }, { "epoch": 28.1225, "grad_norm": 17.375, "learning_rate": 1.686845616325409e-05, "loss": 1.3993, "step": 11249 }, { "epoch": 28.125, "grad_norm": 8.25, "learning_rate": 1.6867867656192946e-05, "loss": 1.3712, "step": 11250 }, { "epoch": 28.1275, "grad_norm": 6.90625, "learning_rate": 1.686727910410634e-05, "loss": 1.2997, "step": 11251 }, { "epoch": 28.13, "grad_norm": 10.5, "learning_rate": 1.6866690506998125e-05, "loss": 1.409, "step": 11252 }, { "epoch": 28.1325, "grad_norm": 7.21875, "learning_rate": 1.686610186487216e-05, "loss": 1.4004, "step": 11253 }, { "epoch": 28.135, "grad_norm": 6.90625, "learning_rate": 1.686551317773231e-05, "loss": 1.3097, "step": 11254 }, { "epoch": 28.1375, "grad_norm": 9.875, "learning_rate": 1.6864924445582428e-05, "loss": 1.3187, "step": 11255 }, { "epoch": 28.14, "grad_norm": 7.4375, "learning_rate": 1.6864335668426373e-05, "loss": 1.2529, "step": 11256 }, { "epoch": 28.1425, "grad_norm": 6.34375, "learning_rate": 1.6863746846268014e-05, "loss": 1.2836, "step": 11257 }, { "epoch": 28.145, "grad_norm": 5.5, "learning_rate": 1.68631579791112e-05, "loss": 1.4167, "step": 11258 }, { "epoch": 28.1475, "grad_norm": 7.84375, "learning_rate": 1.68625690669598e-05, "loss": 1.4656, "step": 11259 }, { "epoch": 28.15, "grad_norm": 5.96875, "learning_rate": 1.686198010981767e-05, "loss": 1.4475, "step": 11260 }, { "epoch": 28.1525, "grad_norm": 9.3125, "learning_rate": 1.6861391107688675e-05, "loss": 1.2379, "step": 11261 }, { "epoch": 28.155, "grad_norm": 7.78125, "learning_rate": 1.6860802060576674e-05, "loss": 1.4323, "step": 11262 }, { "epoch": 28.1575, "grad_norm": 7.90625, "learning_rate": 1.6860212968485526e-05, "loss": 1.325, "step": 11263 }, { "epoch": 28.16, "grad_norm": 6.34375, "learning_rate": 1.68596238314191e-05, "loss": 1.2269, "step": 11264 }, { "epoch": 28.1625, "grad_norm": 6.125, "learning_rate": 1.6859034649381257e-05, "loss": 1.3099, "step": 11265 }, { "epoch": 28.165, "grad_norm": 7.6875, "learning_rate": 1.685844542237585e-05, "loss": 1.3675, "step": 11266 }, { "epoch": 28.1675, "grad_norm": 6.78125, "learning_rate": 1.6857856150406756e-05, "loss": 1.3225, "step": 11267 }, { "epoch": 28.17, "grad_norm": 4.84375, "learning_rate": 1.6857266833477827e-05, "loss": 1.3029, "step": 11268 }, { "epoch": 28.1725, "grad_norm": 8.0, "learning_rate": 1.6856677471592933e-05, "loss": 1.3353, "step": 11269 }, { "epoch": 28.175, "grad_norm": 6.28125, "learning_rate": 1.6856088064755936e-05, "loss": 1.3676, "step": 11270 }, { "epoch": 28.1775, "grad_norm": 8.125, "learning_rate": 1.6855498612970703e-05, "loss": 1.3744, "step": 11271 }, { "epoch": 28.18, "grad_norm": 8.9375, "learning_rate": 1.685490911624109e-05, "loss": 1.594, "step": 11272 }, { "epoch": 28.1825, "grad_norm": 7.21875, "learning_rate": 1.685431957457097e-05, "loss": 1.4414, "step": 11273 }, { "epoch": 28.185, "grad_norm": 5.34375, "learning_rate": 1.6853729987964204e-05, "loss": 1.3269, "step": 11274 }, { "epoch": 28.1875, "grad_norm": 7.53125, "learning_rate": 1.685314035642466e-05, "loss": 1.3035, "step": 11275 }, { "epoch": 28.19, "grad_norm": 6.65625, "learning_rate": 1.68525506799562e-05, "loss": 1.3952, "step": 11276 }, { "epoch": 28.1925, "grad_norm": 11.625, "learning_rate": 1.6851960958562695e-05, "loss": 1.4042, "step": 11277 }, { "epoch": 28.195, "grad_norm": 6.90625, "learning_rate": 1.6851371192248005e-05, "loss": 1.2506, "step": 11278 }, { "epoch": 28.1975, "grad_norm": 7.5, "learning_rate": 1.6850781381016e-05, "loss": 1.3803, "step": 11279 }, { "epoch": 28.2, "grad_norm": 7.40625, "learning_rate": 1.6850191524870548e-05, "loss": 1.397, "step": 11280 }, { "epoch": 28.2025, "grad_norm": 6.25, "learning_rate": 1.684960162381551e-05, "loss": 1.3372, "step": 11281 }, { "epoch": 28.205, "grad_norm": 6.71875, "learning_rate": 1.6849011677854766e-05, "loss": 1.4737, "step": 11282 }, { "epoch": 28.2075, "grad_norm": 7.15625, "learning_rate": 1.684842168699217e-05, "loss": 1.3251, "step": 11283 }, { "epoch": 28.21, "grad_norm": 6.4375, "learning_rate": 1.6847831651231598e-05, "loss": 1.3195, "step": 11284 }, { "epoch": 28.2125, "grad_norm": 6.78125, "learning_rate": 1.6847241570576913e-05, "loss": 1.3123, "step": 11285 }, { "epoch": 28.215, "grad_norm": 7.5625, "learning_rate": 1.6846651445031986e-05, "loss": 1.3379, "step": 11286 }, { "epoch": 28.2175, "grad_norm": 4.9375, "learning_rate": 1.684606127460069e-05, "loss": 1.3213, "step": 11287 }, { "epoch": 28.22, "grad_norm": 6.65625, "learning_rate": 1.684547105928689e-05, "loss": 1.3874, "step": 11288 }, { "epoch": 28.2225, "grad_norm": 6.625, "learning_rate": 1.6844880799094454e-05, "loss": 1.294, "step": 11289 }, { "epoch": 28.225, "grad_norm": 6.53125, "learning_rate": 1.684429049402725e-05, "loss": 1.4162, "step": 11290 }, { "epoch": 28.2275, "grad_norm": 7.5625, "learning_rate": 1.6843700144089153e-05, "loss": 1.3411, "step": 11291 }, { "epoch": 28.23, "grad_norm": 8.25, "learning_rate": 1.6843109749284037e-05, "loss": 1.3396, "step": 11292 }, { "epoch": 28.2325, "grad_norm": 7.28125, "learning_rate": 1.6842519309615762e-05, "loss": 1.2746, "step": 11293 }, { "epoch": 28.235, "grad_norm": 8.8125, "learning_rate": 1.6841928825088206e-05, "loss": 1.319, "step": 11294 }, { "epoch": 28.2375, "grad_norm": 6.46875, "learning_rate": 1.6841338295705238e-05, "loss": 1.2838, "step": 11295 }, { "epoch": 28.24, "grad_norm": 5.78125, "learning_rate": 1.6840747721470733e-05, "loss": 1.421, "step": 11296 }, { "epoch": 28.2425, "grad_norm": 9.5, "learning_rate": 1.6840157102388554e-05, "loss": 1.609, "step": 11297 }, { "epoch": 28.245, "grad_norm": 11.4375, "learning_rate": 1.683956643846258e-05, "loss": 1.3378, "step": 11298 }, { "epoch": 28.2475, "grad_norm": 6.125, "learning_rate": 1.6838975729696686e-05, "loss": 1.3658, "step": 11299 }, { "epoch": 28.25, "grad_norm": 7.59375, "learning_rate": 1.6838384976094738e-05, "loss": 1.4587, "step": 11300 }, { "epoch": 28.2525, "grad_norm": 6.625, "learning_rate": 1.6837794177660614e-05, "loss": 1.4127, "step": 11301 }, { "epoch": 28.255, "grad_norm": 7.09375, "learning_rate": 1.6837203334398183e-05, "loss": 1.3781, "step": 11302 }, { "epoch": 28.2575, "grad_norm": 6.9375, "learning_rate": 1.6836612446311316e-05, "loss": 1.4129, "step": 11303 }, { "epoch": 28.26, "grad_norm": 6.53125, "learning_rate": 1.68360215134039e-05, "loss": 1.2569, "step": 11304 }, { "epoch": 28.2625, "grad_norm": 7.28125, "learning_rate": 1.6835430535679797e-05, "loss": 1.3123, "step": 11305 }, { "epoch": 28.265, "grad_norm": 7.25, "learning_rate": 1.6834839513142883e-05, "loss": 1.4032, "step": 11306 }, { "epoch": 28.2675, "grad_norm": 9.6875, "learning_rate": 1.6834248445797035e-05, "loss": 1.2937, "step": 11307 }, { "epoch": 28.27, "grad_norm": 10.625, "learning_rate": 1.6833657333646132e-05, "loss": 1.3657, "step": 11308 }, { "epoch": 28.2725, "grad_norm": 6.40625, "learning_rate": 1.683306617669404e-05, "loss": 1.3785, "step": 11309 }, { "epoch": 28.275, "grad_norm": 8.5, "learning_rate": 1.683247497494464e-05, "loss": 1.3612, "step": 11310 }, { "epoch": 28.2775, "grad_norm": 6.3125, "learning_rate": 1.6831883728401813e-05, "loss": 1.4827, "step": 11311 }, { "epoch": 28.28, "grad_norm": 9.125, "learning_rate": 1.6831292437069425e-05, "loss": 1.3219, "step": 11312 }, { "epoch": 28.2825, "grad_norm": 7.90625, "learning_rate": 1.6830701100951356e-05, "loss": 1.4448, "step": 11313 }, { "epoch": 28.285, "grad_norm": 6.09375, "learning_rate": 1.6830109720051488e-05, "loss": 1.2371, "step": 11314 }, { "epoch": 28.2875, "grad_norm": 6.40625, "learning_rate": 1.682951829437369e-05, "loss": 1.4149, "step": 11315 }, { "epoch": 28.29, "grad_norm": 6.5, "learning_rate": 1.6828926823921845e-05, "loss": 1.3386, "step": 11316 }, { "epoch": 28.2925, "grad_norm": 4.5625, "learning_rate": 1.682833530869983e-05, "loss": 1.23, "step": 11317 }, { "epoch": 28.295, "grad_norm": 9.25, "learning_rate": 1.6827743748711523e-05, "loss": 1.3346, "step": 11318 }, { "epoch": 28.2975, "grad_norm": 7.84375, "learning_rate": 1.6827152143960797e-05, "loss": 1.361, "step": 11319 }, { "epoch": 28.3, "grad_norm": 6.90625, "learning_rate": 1.682656049445154e-05, "loss": 1.5394, "step": 11320 }, { "epoch": 28.3025, "grad_norm": 4.6875, "learning_rate": 1.682596880018762e-05, "loss": 1.3643, "step": 11321 }, { "epoch": 28.305, "grad_norm": 6.1875, "learning_rate": 1.6825377061172926e-05, "loss": 1.2961, "step": 11322 }, { "epoch": 28.3075, "grad_norm": 8.4375, "learning_rate": 1.6824785277411335e-05, "loss": 1.4583, "step": 11323 }, { "epoch": 28.31, "grad_norm": 9.8125, "learning_rate": 1.6824193448906718e-05, "loss": 1.6108, "step": 11324 }, { "epoch": 28.3125, "grad_norm": 8.25, "learning_rate": 1.6823601575662963e-05, "loss": 1.3168, "step": 11325 }, { "epoch": 28.315, "grad_norm": 7.75, "learning_rate": 1.6823009657683952e-05, "loss": 1.3611, "step": 11326 }, { "epoch": 28.3175, "grad_norm": 7.3125, "learning_rate": 1.6822417694973562e-05, "loss": 1.3499, "step": 11327 }, { "epoch": 28.32, "grad_norm": 6.8125, "learning_rate": 1.6821825687535675e-05, "loss": 1.3795, "step": 11328 }, { "epoch": 28.3225, "grad_norm": 7.875, "learning_rate": 1.6821233635374174e-05, "loss": 1.3437, "step": 11329 }, { "epoch": 28.325, "grad_norm": 9.3125, "learning_rate": 1.6820641538492934e-05, "loss": 1.2926, "step": 11330 }, { "epoch": 28.3275, "grad_norm": 12.3125, "learning_rate": 1.6820049396895847e-05, "loss": 1.326, "step": 11331 }, { "epoch": 28.33, "grad_norm": 7.09375, "learning_rate": 1.681945721058678e-05, "loss": 1.4354, "step": 11332 }, { "epoch": 28.3325, "grad_norm": 5.59375, "learning_rate": 1.681886497956963e-05, "loss": 1.2588, "step": 11333 }, { "epoch": 28.335, "grad_norm": 7.15625, "learning_rate": 1.6818272703848273e-05, "loss": 1.5246, "step": 11334 }, { "epoch": 28.3375, "grad_norm": 5.625, "learning_rate": 1.6817680383426593e-05, "loss": 1.2686, "step": 11335 }, { "epoch": 28.34, "grad_norm": 6.75, "learning_rate": 1.6817088018308477e-05, "loss": 1.4904, "step": 11336 }, { "epoch": 28.3425, "grad_norm": 7.5, "learning_rate": 1.6816495608497797e-05, "loss": 1.3543, "step": 11337 }, { "epoch": 28.345, "grad_norm": 7.0625, "learning_rate": 1.681590315399845e-05, "loss": 1.4961, "step": 11338 }, { "epoch": 28.3475, "grad_norm": 8.5, "learning_rate": 1.6815310654814316e-05, "loss": 1.4122, "step": 11339 }, { "epoch": 28.35, "grad_norm": 11.875, "learning_rate": 1.6814718110949274e-05, "loss": 1.4649, "step": 11340 }, { "epoch": 28.3525, "grad_norm": 8.8125, "learning_rate": 1.6814125522407215e-05, "loss": 1.4245, "step": 11341 }, { "epoch": 28.355, "grad_norm": 8.4375, "learning_rate": 1.6813532889192022e-05, "loss": 1.4518, "step": 11342 }, { "epoch": 28.3575, "grad_norm": 6.21875, "learning_rate": 1.681294021130758e-05, "loss": 1.5928, "step": 11343 }, { "epoch": 28.36, "grad_norm": 6.1875, "learning_rate": 1.6812347488757774e-05, "loss": 1.2176, "step": 11344 }, { "epoch": 28.3625, "grad_norm": 8.3125, "learning_rate": 1.6811754721546488e-05, "loss": 1.4399, "step": 11345 }, { "epoch": 28.365, "grad_norm": 8.3125, "learning_rate": 1.6811161909677616e-05, "loss": 1.3422, "step": 11346 }, { "epoch": 28.3675, "grad_norm": 7.3125, "learning_rate": 1.6810569053155034e-05, "loss": 1.2563, "step": 11347 }, { "epoch": 28.37, "grad_norm": 6.375, "learning_rate": 1.6809976151982635e-05, "loss": 1.3535, "step": 11348 }, { "epoch": 28.3725, "grad_norm": 6.5, "learning_rate": 1.6809383206164308e-05, "loss": 1.4754, "step": 11349 }, { "epoch": 28.375, "grad_norm": 6.9375, "learning_rate": 1.6808790215703933e-05, "loss": 1.3632, "step": 11350 }, { "epoch": 28.3775, "grad_norm": 7.625, "learning_rate": 1.6808197180605406e-05, "loss": 1.2608, "step": 11351 }, { "epoch": 28.38, "grad_norm": 6.15625, "learning_rate": 1.6807604100872604e-05, "loss": 1.3386, "step": 11352 }, { "epoch": 28.3825, "grad_norm": 9.875, "learning_rate": 1.680701097650943e-05, "loss": 1.4707, "step": 11353 }, { "epoch": 28.385, "grad_norm": 7.4375, "learning_rate": 1.680641780751976e-05, "loss": 1.3515, "step": 11354 }, { "epoch": 28.3875, "grad_norm": 7.875, "learning_rate": 1.6805824593907488e-05, "loss": 1.479, "step": 11355 }, { "epoch": 28.39, "grad_norm": 6.34375, "learning_rate": 1.6805231335676505e-05, "loss": 1.3048, "step": 11356 }, { "epoch": 28.3925, "grad_norm": 6.8125, "learning_rate": 1.680463803283069e-05, "loss": 1.3334, "step": 11357 }, { "epoch": 28.395, "grad_norm": 5.90625, "learning_rate": 1.680404468537395e-05, "loss": 1.3382, "step": 11358 }, { "epoch": 28.3975, "grad_norm": 8.125, "learning_rate": 1.680345129331016e-05, "loss": 1.3716, "step": 11359 }, { "epoch": 28.4, "grad_norm": 5.78125, "learning_rate": 1.6802857856643214e-05, "loss": 1.4011, "step": 11360 }, { "epoch": 28.4025, "grad_norm": 7.84375, "learning_rate": 1.6802264375377008e-05, "loss": 1.4019, "step": 11361 }, { "epoch": 28.405, "grad_norm": 7.59375, "learning_rate": 1.6801670849515428e-05, "loss": 1.3092, "step": 11362 }, { "epoch": 28.4075, "grad_norm": 6.4375, "learning_rate": 1.6801077279062365e-05, "loss": 1.3527, "step": 11363 }, { "epoch": 28.41, "grad_norm": 7.78125, "learning_rate": 1.680048366402171e-05, "loss": 1.4241, "step": 11364 }, { "epoch": 28.4125, "grad_norm": 7.3125, "learning_rate": 1.6799890004397362e-05, "loss": 1.2384, "step": 11365 }, { "epoch": 28.415, "grad_norm": 12.8125, "learning_rate": 1.6799296300193206e-05, "loss": 1.3105, "step": 11366 }, { "epoch": 28.4175, "grad_norm": 9.5625, "learning_rate": 1.6798702551413127e-05, "loss": 1.3468, "step": 11367 }, { "epoch": 28.42, "grad_norm": 6.8125, "learning_rate": 1.6798108758061036e-05, "loss": 1.2843, "step": 11368 }, { "epoch": 28.4225, "grad_norm": 8.8125, "learning_rate": 1.6797514920140813e-05, "loss": 1.4019, "step": 11369 }, { "epoch": 28.425, "grad_norm": 10.625, "learning_rate": 1.679692103765635e-05, "loss": 1.4978, "step": 11370 }, { "epoch": 28.4275, "grad_norm": 5.78125, "learning_rate": 1.679632711061155e-05, "loss": 1.2563, "step": 11371 }, { "epoch": 28.43, "grad_norm": 7.9375, "learning_rate": 1.67957331390103e-05, "loss": 1.3724, "step": 11372 }, { "epoch": 28.4325, "grad_norm": 4.8125, "learning_rate": 1.6795139122856496e-05, "loss": 1.3099, "step": 11373 }, { "epoch": 28.435, "grad_norm": 7.03125, "learning_rate": 1.679454506215403e-05, "loss": 1.3592, "step": 11374 }, { "epoch": 28.4375, "grad_norm": 6.1875, "learning_rate": 1.67939509569068e-05, "loss": 1.5071, "step": 11375 }, { "epoch": 28.44, "grad_norm": 9.375, "learning_rate": 1.6793356807118695e-05, "loss": 1.364, "step": 11376 }, { "epoch": 28.4425, "grad_norm": 8.5, "learning_rate": 1.679276261279362e-05, "loss": 1.3596, "step": 11377 }, { "epoch": 28.445, "grad_norm": 8.875, "learning_rate": 1.6792168373935465e-05, "loss": 1.3054, "step": 11378 }, { "epoch": 28.4475, "grad_norm": 6.3125, "learning_rate": 1.6791574090548123e-05, "loss": 1.2137, "step": 11379 }, { "epoch": 28.45, "grad_norm": 6.09375, "learning_rate": 1.6790979762635497e-05, "loss": 1.421, "step": 11380 }, { "epoch": 28.4525, "grad_norm": 6.15625, "learning_rate": 1.6790385390201475e-05, "loss": 1.3444, "step": 11381 }, { "epoch": 28.455, "grad_norm": 9.25, "learning_rate": 1.6789790973249963e-05, "loss": 1.4752, "step": 11382 }, { "epoch": 28.4575, "grad_norm": 5.8125, "learning_rate": 1.678919651178485e-05, "loss": 1.3058, "step": 11383 }, { "epoch": 28.46, "grad_norm": 6.78125, "learning_rate": 1.6788602005810035e-05, "loss": 1.4975, "step": 11384 }, { "epoch": 28.4625, "grad_norm": 10.375, "learning_rate": 1.678800745532942e-05, "loss": 1.4179, "step": 11385 }, { "epoch": 28.465, "grad_norm": 7.90625, "learning_rate": 1.6787412860346896e-05, "loss": 1.4364, "step": 11386 }, { "epoch": 28.4675, "grad_norm": 6.8125, "learning_rate": 1.6786818220866368e-05, "loss": 1.4967, "step": 11387 }, { "epoch": 28.47, "grad_norm": 5.90625, "learning_rate": 1.678622353689173e-05, "loss": 1.353, "step": 11388 }, { "epoch": 28.4725, "grad_norm": 10.5625, "learning_rate": 1.6785628808426884e-05, "loss": 1.3879, "step": 11389 }, { "epoch": 28.475, "grad_norm": 7.15625, "learning_rate": 1.6785034035475724e-05, "loss": 1.2798, "step": 11390 }, { "epoch": 28.4775, "grad_norm": 5.0625, "learning_rate": 1.6784439218042155e-05, "loss": 1.37, "step": 11391 }, { "epoch": 28.48, "grad_norm": 5.59375, "learning_rate": 1.6783844356130073e-05, "loss": 1.452, "step": 11392 }, { "epoch": 28.4825, "grad_norm": 6.53125, "learning_rate": 1.6783249449743378e-05, "loss": 1.3366, "step": 11393 }, { "epoch": 28.485, "grad_norm": 5.28125, "learning_rate": 1.6782654498885972e-05, "loss": 1.3808, "step": 11394 }, { "epoch": 28.4875, "grad_norm": 6.9375, "learning_rate": 1.6782059503561756e-05, "loss": 1.2744, "step": 11395 }, { "epoch": 28.49, "grad_norm": 9.9375, "learning_rate": 1.6781464463774628e-05, "loss": 1.2828, "step": 11396 }, { "epoch": 28.4925, "grad_norm": 9.3125, "learning_rate": 1.678086937952849e-05, "loss": 1.4447, "step": 11397 }, { "epoch": 28.495, "grad_norm": 7.375, "learning_rate": 1.6780274250827243e-05, "loss": 1.2808, "step": 11398 }, { "epoch": 28.4975, "grad_norm": 9.0625, "learning_rate": 1.6779679077674792e-05, "loss": 1.4249, "step": 11399 }, { "epoch": 28.5, "grad_norm": 7.1875, "learning_rate": 1.6779083860075032e-05, "loss": 1.4301, "step": 11400 }, { "epoch": 28.5025, "grad_norm": 7.59375, "learning_rate": 1.6778488598031874e-05, "loss": 1.5004, "step": 11401 }, { "epoch": 28.505, "grad_norm": 8.25, "learning_rate": 1.6777893291549213e-05, "loss": 1.2148, "step": 11402 }, { "epoch": 28.5075, "grad_norm": 8.8125, "learning_rate": 1.6777297940630957e-05, "loss": 1.4007, "step": 11403 }, { "epoch": 28.51, "grad_norm": 7.34375, "learning_rate": 1.6776702545281004e-05, "loss": 1.2669, "step": 11404 }, { "epoch": 28.5125, "grad_norm": 7.71875, "learning_rate": 1.6776107105503262e-05, "loss": 1.3376, "step": 11405 }, { "epoch": 28.515, "grad_norm": 8.625, "learning_rate": 1.6775511621301634e-05, "loss": 1.1965, "step": 11406 }, { "epoch": 28.5175, "grad_norm": 8.3125, "learning_rate": 1.6774916092680022e-05, "loss": 1.4905, "step": 11407 }, { "epoch": 28.52, "grad_norm": 6.5, "learning_rate": 1.677432051964233e-05, "loss": 1.4617, "step": 11408 }, { "epoch": 28.5225, "grad_norm": 8.5, "learning_rate": 1.677372490219247e-05, "loss": 1.2291, "step": 11409 }, { "epoch": 28.525, "grad_norm": 5.4375, "learning_rate": 1.6773129240334333e-05, "loss": 1.4145, "step": 11410 }, { "epoch": 28.5275, "grad_norm": 6.59375, "learning_rate": 1.6772533534071833e-05, "loss": 1.2337, "step": 11411 }, { "epoch": 28.53, "grad_norm": 5.75, "learning_rate": 1.6771937783408873e-05, "loss": 1.4231, "step": 11412 }, { "epoch": 28.5325, "grad_norm": 7.03125, "learning_rate": 1.6771341988349365e-05, "loss": 1.2488, "step": 11413 }, { "epoch": 28.535, "grad_norm": 6.5, "learning_rate": 1.677074614889721e-05, "loss": 1.3749, "step": 11414 }, { "epoch": 28.5375, "grad_norm": 5.8125, "learning_rate": 1.6770150265056305e-05, "loss": 1.283, "step": 11415 }, { "epoch": 28.54, "grad_norm": 7.625, "learning_rate": 1.6769554336830577e-05, "loss": 1.3596, "step": 11416 }, { "epoch": 28.5425, "grad_norm": 7.875, "learning_rate": 1.6768958364223917e-05, "loss": 1.4121, "step": 11417 }, { "epoch": 28.545, "grad_norm": 8.375, "learning_rate": 1.6768362347240238e-05, "loss": 1.5132, "step": 11418 }, { "epoch": 28.5475, "grad_norm": 10.4375, "learning_rate": 1.676776628588344e-05, "loss": 1.299, "step": 11419 }, { "epoch": 28.55, "grad_norm": 6.53125, "learning_rate": 1.6767170180157442e-05, "loss": 1.2603, "step": 11420 }, { "epoch": 28.5525, "grad_norm": 7.75, "learning_rate": 1.6766574030066148e-05, "loss": 1.3318, "step": 11421 }, { "epoch": 28.555, "grad_norm": 10.375, "learning_rate": 1.6765977835613468e-05, "loss": 1.2758, "step": 11422 }, { "epoch": 28.5575, "grad_norm": 9.75, "learning_rate": 1.6765381596803303e-05, "loss": 1.384, "step": 11423 }, { "epoch": 28.56, "grad_norm": 6.40625, "learning_rate": 1.6764785313639568e-05, "loss": 1.4878, "step": 11424 }, { "epoch": 28.5625, "grad_norm": 6.0625, "learning_rate": 1.6764188986126174e-05, "loss": 1.165, "step": 11425 }, { "epoch": 28.565, "grad_norm": 6.75, "learning_rate": 1.6763592614267024e-05, "loss": 1.3055, "step": 11426 }, { "epoch": 28.5675, "grad_norm": 6.34375, "learning_rate": 1.676299619806603e-05, "loss": 1.4773, "step": 11427 }, { "epoch": 28.57, "grad_norm": 6.34375, "learning_rate": 1.676239973752711e-05, "loss": 1.17, "step": 11428 }, { "epoch": 28.5725, "grad_norm": 10.5625, "learning_rate": 1.6761803232654164e-05, "loss": 1.5064, "step": 11429 }, { "epoch": 28.575, "grad_norm": 5.8125, "learning_rate": 1.6761206683451106e-05, "loss": 1.354, "step": 11430 }, { "epoch": 28.5775, "grad_norm": 7.34375, "learning_rate": 1.6760610089921852e-05, "loss": 1.4192, "step": 11431 }, { "epoch": 28.58, "grad_norm": 9.125, "learning_rate": 1.6760013452070304e-05, "loss": 1.3495, "step": 11432 }, { "epoch": 28.5825, "grad_norm": 5.3125, "learning_rate": 1.675941676990038e-05, "loss": 1.2408, "step": 11433 }, { "epoch": 28.585, "grad_norm": 8.0, "learning_rate": 1.675882004341599e-05, "loss": 1.3683, "step": 11434 }, { "epoch": 28.5875, "grad_norm": 7.78125, "learning_rate": 1.6758223272621045e-05, "loss": 1.1825, "step": 11435 }, { "epoch": 28.59, "grad_norm": 7.71875, "learning_rate": 1.675762645751946e-05, "loss": 1.429, "step": 11436 }, { "epoch": 28.5925, "grad_norm": 6.90625, "learning_rate": 1.6757029598115146e-05, "loss": 1.2699, "step": 11437 }, { "epoch": 28.595, "grad_norm": 7.03125, "learning_rate": 1.675643269441202e-05, "loss": 1.3517, "step": 11438 }, { "epoch": 28.5975, "grad_norm": 9.8125, "learning_rate": 1.6755835746413987e-05, "loss": 1.5533, "step": 11439 }, { "epoch": 28.6, "grad_norm": 7.0, "learning_rate": 1.6755238754124965e-05, "loss": 1.1986, "step": 11440 }, { "epoch": 28.6025, "grad_norm": 7.625, "learning_rate": 1.6754641717548873e-05, "loss": 1.4661, "step": 11441 }, { "epoch": 28.605, "grad_norm": 8.125, "learning_rate": 1.6754044636689614e-05, "loss": 1.3393, "step": 11442 }, { "epoch": 28.6075, "grad_norm": 5.59375, "learning_rate": 1.675344751155111e-05, "loss": 1.3724, "step": 11443 }, { "epoch": 28.61, "grad_norm": 5.5, "learning_rate": 1.675285034213728e-05, "loss": 1.3582, "step": 11444 }, { "epoch": 28.6125, "grad_norm": 8.625, "learning_rate": 1.6752253128452025e-05, "loss": 1.3972, "step": 11445 }, { "epoch": 28.615, "grad_norm": 4.75, "learning_rate": 1.6751655870499275e-05, "loss": 1.3719, "step": 11446 }, { "epoch": 28.6175, "grad_norm": 5.9375, "learning_rate": 1.6751058568282937e-05, "loss": 1.2459, "step": 11447 }, { "epoch": 28.62, "grad_norm": 6.9375, "learning_rate": 1.675046122180693e-05, "loss": 1.3122, "step": 11448 }, { "epoch": 28.6225, "grad_norm": 8.0625, "learning_rate": 1.6749863831075168e-05, "loss": 1.3011, "step": 11449 }, { "epoch": 28.625, "grad_norm": 6.6875, "learning_rate": 1.674926639609157e-05, "loss": 1.3097, "step": 11450 }, { "epoch": 28.6275, "grad_norm": 8.5625, "learning_rate": 1.674866891686005e-05, "loss": 1.3447, "step": 11451 }, { "epoch": 28.63, "grad_norm": 10.6875, "learning_rate": 1.6748071393384527e-05, "loss": 1.3449, "step": 11452 }, { "epoch": 28.6325, "grad_norm": 6.59375, "learning_rate": 1.674747382566892e-05, "loss": 1.3876, "step": 11453 }, { "epoch": 28.635, "grad_norm": 6.625, "learning_rate": 1.674687621371714e-05, "loss": 1.4321, "step": 11454 }, { "epoch": 28.6375, "grad_norm": 7.59375, "learning_rate": 1.6746278557533114e-05, "loss": 1.4289, "step": 11455 }, { "epoch": 28.64, "grad_norm": 7.875, "learning_rate": 1.6745680857120757e-05, "loss": 1.523, "step": 11456 }, { "epoch": 28.6425, "grad_norm": 5.8125, "learning_rate": 1.6745083112483983e-05, "loss": 1.4109, "step": 11457 }, { "epoch": 28.645, "grad_norm": 7.90625, "learning_rate": 1.6744485323626716e-05, "loss": 1.4162, "step": 11458 }, { "epoch": 28.6475, "grad_norm": 7.34375, "learning_rate": 1.6743887490552872e-05, "loss": 1.3022, "step": 11459 }, { "epoch": 28.65, "grad_norm": 10.4375, "learning_rate": 1.674328961326637e-05, "loss": 1.2946, "step": 11460 }, { "epoch": 28.6525, "grad_norm": 10.75, "learning_rate": 1.6742691691771137e-05, "loss": 1.3275, "step": 11461 }, { "epoch": 28.655, "grad_norm": 6.0625, "learning_rate": 1.674209372607108e-05, "loss": 1.3758, "step": 11462 }, { "epoch": 28.6575, "grad_norm": 7.875, "learning_rate": 1.674149571617013e-05, "loss": 1.4464, "step": 11463 }, { "epoch": 28.66, "grad_norm": 6.9375, "learning_rate": 1.6740897662072203e-05, "loss": 1.4202, "step": 11464 }, { "epoch": 28.6625, "grad_norm": 6.28125, "learning_rate": 1.674029956378122e-05, "loss": 1.4396, "step": 11465 }, { "epoch": 28.665, "grad_norm": 6.1875, "learning_rate": 1.6739701421301104e-05, "loss": 1.3279, "step": 11466 }, { "epoch": 28.6675, "grad_norm": 8.5625, "learning_rate": 1.6739103234635775e-05, "loss": 1.398, "step": 11467 }, { "epoch": 28.67, "grad_norm": 9.4375, "learning_rate": 1.6738505003789157e-05, "loss": 1.4269, "step": 11468 }, { "epoch": 28.6725, "grad_norm": 7.03125, "learning_rate": 1.6737906728765163e-05, "loss": 1.3809, "step": 11469 }, { "epoch": 28.675, "grad_norm": 7.375, "learning_rate": 1.6737308409567727e-05, "loss": 1.3467, "step": 11470 }, { "epoch": 28.6775, "grad_norm": 6.59375, "learning_rate": 1.6736710046200768e-05, "loss": 1.2545, "step": 11471 }, { "epoch": 28.68, "grad_norm": 5.3125, "learning_rate": 1.6736111638668203e-05, "loss": 1.368, "step": 11472 }, { "epoch": 28.6825, "grad_norm": 6.5, "learning_rate": 1.6735513186973963e-05, "loss": 1.5138, "step": 11473 }, { "epoch": 28.685, "grad_norm": 8.5, "learning_rate": 1.6734914691121967e-05, "loss": 1.5212, "step": 11474 }, { "epoch": 28.6875, "grad_norm": 7.0625, "learning_rate": 1.6734316151116143e-05, "loss": 1.3409, "step": 11475 }, { "epoch": 28.69, "grad_norm": 8.5625, "learning_rate": 1.673371756696041e-05, "loss": 1.2222, "step": 11476 }, { "epoch": 28.6925, "grad_norm": 10.125, "learning_rate": 1.6733118938658693e-05, "loss": 1.3374, "step": 11477 }, { "epoch": 28.695, "grad_norm": 7.3125, "learning_rate": 1.6732520266214914e-05, "loss": 1.4542, "step": 11478 }, { "epoch": 28.6975, "grad_norm": 5.5625, "learning_rate": 1.6731921549633007e-05, "loss": 1.3891, "step": 11479 }, { "epoch": 28.7, "grad_norm": 8.375, "learning_rate": 1.6731322788916892e-05, "loss": 1.3646, "step": 11480 }, { "epoch": 28.7025, "grad_norm": 7.375, "learning_rate": 1.6730723984070493e-05, "loss": 1.4111, "step": 11481 }, { "epoch": 28.705, "grad_norm": 7.40625, "learning_rate": 1.6730125135097736e-05, "loss": 1.3174, "step": 11482 }, { "epoch": 28.7075, "grad_norm": 10.125, "learning_rate": 1.672952624200255e-05, "loss": 1.3639, "step": 11483 }, { "epoch": 28.71, "grad_norm": 4.59375, "learning_rate": 1.6728927304788854e-05, "loss": 1.4092, "step": 11484 }, { "epoch": 28.7125, "grad_norm": 7.90625, "learning_rate": 1.6728328323460585e-05, "loss": 1.4695, "step": 11485 }, { "epoch": 28.715, "grad_norm": 8.5625, "learning_rate": 1.6727729298021663e-05, "loss": 1.457, "step": 11486 }, { "epoch": 28.7175, "grad_norm": 6.28125, "learning_rate": 1.6727130228476018e-05, "loss": 1.5929, "step": 11487 }, { "epoch": 28.72, "grad_norm": 7.03125, "learning_rate": 1.6726531114827572e-05, "loss": 1.422, "step": 11488 }, { "epoch": 28.7225, "grad_norm": 10.0, "learning_rate": 1.672593195708026e-05, "loss": 1.2911, "step": 11489 }, { "epoch": 28.725, "grad_norm": 5.25, "learning_rate": 1.672533275523801e-05, "loss": 1.4274, "step": 11490 }, { "epoch": 28.7275, "grad_norm": 6.9375, "learning_rate": 1.6724733509304746e-05, "loss": 1.4077, "step": 11491 }, { "epoch": 28.73, "grad_norm": 8.5, "learning_rate": 1.6724134219284397e-05, "loss": 1.2659, "step": 11492 }, { "epoch": 28.7325, "grad_norm": 6.96875, "learning_rate": 1.6723534885180895e-05, "loss": 1.3865, "step": 11493 }, { "epoch": 28.735, "grad_norm": 6.90625, "learning_rate": 1.6722935506998165e-05, "loss": 1.3028, "step": 11494 }, { "epoch": 28.7375, "grad_norm": 6.96875, "learning_rate": 1.672233608474014e-05, "loss": 1.4796, "step": 11495 }, { "epoch": 28.74, "grad_norm": 9.375, "learning_rate": 1.672173661841075e-05, "loss": 1.3258, "step": 11496 }, { "epoch": 28.7425, "grad_norm": 8.375, "learning_rate": 1.6721137108013922e-05, "loss": 1.3607, "step": 11497 }, { "epoch": 28.745, "grad_norm": 12.9375, "learning_rate": 1.672053755355359e-05, "loss": 1.4272, "step": 11498 }, { "epoch": 28.7475, "grad_norm": 8.625, "learning_rate": 1.6719937955033677e-05, "loss": 1.2636, "step": 11499 }, { "epoch": 28.75, "grad_norm": 4.625, "learning_rate": 1.6719338312458123e-05, "loss": 1.3546, "step": 11500 }, { "epoch": 28.7525, "grad_norm": 7.1875, "learning_rate": 1.6718738625830863e-05, "loss": 1.3569, "step": 11501 }, { "epoch": 28.755, "grad_norm": 8.5, "learning_rate": 1.6718138895155812e-05, "loss": 1.3123, "step": 11502 }, { "epoch": 28.7575, "grad_norm": 5.5, "learning_rate": 1.6717539120436915e-05, "loss": 1.3559, "step": 11503 }, { "epoch": 28.76, "grad_norm": 6.46875, "learning_rate": 1.6716939301678098e-05, "loss": 1.4308, "step": 11504 }, { "epoch": 28.7625, "grad_norm": 5.34375, "learning_rate": 1.67163394388833e-05, "loss": 1.206, "step": 11505 }, { "epoch": 28.765, "grad_norm": 9.9375, "learning_rate": 1.6715739532056448e-05, "loss": 1.3063, "step": 11506 }, { "epoch": 28.7675, "grad_norm": 7.5, "learning_rate": 1.6715139581201474e-05, "loss": 1.2839, "step": 11507 }, { "epoch": 28.77, "grad_norm": 6.46875, "learning_rate": 1.6714539586322316e-05, "loss": 1.4724, "step": 11508 }, { "epoch": 28.7725, "grad_norm": 5.75, "learning_rate": 1.6713939547422902e-05, "loss": 1.5276, "step": 11509 }, { "epoch": 28.775, "grad_norm": 5.875, "learning_rate": 1.671333946450717e-05, "loss": 1.3255, "step": 11510 }, { "epoch": 28.7775, "grad_norm": 6.0625, "learning_rate": 1.6712739337579055e-05, "loss": 1.3711, "step": 11511 }, { "epoch": 28.78, "grad_norm": 7.75, "learning_rate": 1.671213916664249e-05, "loss": 1.3282, "step": 11512 }, { "epoch": 28.7825, "grad_norm": 10.9375, "learning_rate": 1.6711538951701406e-05, "loss": 1.2679, "step": 11513 }, { "epoch": 28.785, "grad_norm": 8.375, "learning_rate": 1.6710938692759743e-05, "loss": 1.4721, "step": 11514 }, { "epoch": 28.7875, "grad_norm": 7.6875, "learning_rate": 1.6710338389821437e-05, "loss": 1.4082, "step": 11515 }, { "epoch": 28.79, "grad_norm": 7.1875, "learning_rate": 1.670973804289042e-05, "loss": 1.2342, "step": 11516 }, { "epoch": 28.7925, "grad_norm": 8.5625, "learning_rate": 1.6709137651970625e-05, "loss": 1.4269, "step": 11517 }, { "epoch": 28.795, "grad_norm": 7.09375, "learning_rate": 1.6708537217065995e-05, "loss": 1.3622, "step": 11518 }, { "epoch": 28.7975, "grad_norm": 9.4375, "learning_rate": 1.6707936738180467e-05, "loss": 1.3788, "step": 11519 }, { "epoch": 28.8, "grad_norm": 7.09375, "learning_rate": 1.6707336215317968e-05, "loss": 1.3339, "step": 11520 }, { "epoch": 28.8025, "grad_norm": 7.1875, "learning_rate": 1.6706735648482447e-05, "loss": 1.3018, "step": 11521 }, { "epoch": 28.805, "grad_norm": 6.875, "learning_rate": 1.670613503767783e-05, "loss": 1.38, "step": 11522 }, { "epoch": 28.8075, "grad_norm": 5.21875, "learning_rate": 1.670553438290806e-05, "loss": 1.4592, "step": 11523 }, { "epoch": 28.81, "grad_norm": 6.0, "learning_rate": 1.6704933684177077e-05, "loss": 1.3593, "step": 11524 }, { "epoch": 28.8125, "grad_norm": 6.375, "learning_rate": 1.670433294148882e-05, "loss": 1.4727, "step": 11525 }, { "epoch": 28.815, "grad_norm": 7.59375, "learning_rate": 1.670373215484722e-05, "loss": 1.5166, "step": 11526 }, { "epoch": 28.8175, "grad_norm": 5.34375, "learning_rate": 1.6703131324256224e-05, "loss": 1.329, "step": 11527 }, { "epoch": 28.82, "grad_norm": 6.09375, "learning_rate": 1.6702530449719763e-05, "loss": 1.4253, "step": 11528 }, { "epoch": 28.8225, "grad_norm": 7.03125, "learning_rate": 1.6701929531241784e-05, "loss": 1.3269, "step": 11529 }, { "epoch": 28.825, "grad_norm": 6.34375, "learning_rate": 1.670132856882622e-05, "loss": 1.306, "step": 11530 }, { "epoch": 28.8275, "grad_norm": 7.125, "learning_rate": 1.6700727562477017e-05, "loss": 1.3504, "step": 11531 }, { "epoch": 28.83, "grad_norm": 7.03125, "learning_rate": 1.670012651219811e-05, "loss": 1.3582, "step": 11532 }, { "epoch": 28.8325, "grad_norm": 6.625, "learning_rate": 1.6699525417993446e-05, "loss": 1.3284, "step": 11533 }, { "epoch": 28.835, "grad_norm": 6.53125, "learning_rate": 1.6698924279866956e-05, "loss": 1.3366, "step": 11534 }, { "epoch": 28.8375, "grad_norm": 5.90625, "learning_rate": 1.669832309782259e-05, "loss": 1.3091, "step": 11535 }, { "epoch": 28.84, "grad_norm": 6.125, "learning_rate": 1.6697721871864286e-05, "loss": 1.321, "step": 11536 }, { "epoch": 28.8425, "grad_norm": 7.875, "learning_rate": 1.6697120601995983e-05, "loss": 1.6555, "step": 11537 }, { "epoch": 28.845, "grad_norm": 8.125, "learning_rate": 1.6696519288221626e-05, "loss": 1.5494, "step": 11538 }, { "epoch": 28.8475, "grad_norm": 11.4375, "learning_rate": 1.6695917930545155e-05, "loss": 1.5635, "step": 11539 }, { "epoch": 28.85, "grad_norm": 7.25, "learning_rate": 1.6695316528970517e-05, "loss": 1.3945, "step": 11540 }, { "epoch": 28.8525, "grad_norm": 5.71875, "learning_rate": 1.6694715083501652e-05, "loss": 1.4015, "step": 11541 }, { "epoch": 28.855, "grad_norm": 8.375, "learning_rate": 1.66941135941425e-05, "loss": 1.3433, "step": 11542 }, { "epoch": 28.8575, "grad_norm": 6.90625, "learning_rate": 1.6693512060897012e-05, "loss": 1.4386, "step": 11543 }, { "epoch": 28.86, "grad_norm": 10.125, "learning_rate": 1.6692910483769124e-05, "loss": 1.3215, "step": 11544 }, { "epoch": 28.8625, "grad_norm": 6.21875, "learning_rate": 1.669230886276278e-05, "loss": 1.3264, "step": 11545 }, { "epoch": 28.865, "grad_norm": 6.625, "learning_rate": 1.669170719788193e-05, "loss": 1.3429, "step": 11546 }, { "epoch": 28.8675, "grad_norm": 8.3125, "learning_rate": 1.6691105489130516e-05, "loss": 1.2996, "step": 11547 }, { "epoch": 28.87, "grad_norm": 7.5625, "learning_rate": 1.669050373651248e-05, "loss": 1.427, "step": 11548 }, { "epoch": 28.8725, "grad_norm": 5.1875, "learning_rate": 1.6689901940031768e-05, "loss": 1.2163, "step": 11549 }, { "epoch": 28.875, "grad_norm": 7.09375, "learning_rate": 1.6689300099692332e-05, "loss": 1.3227, "step": 11550 }, { "epoch": 28.8775, "grad_norm": 5.78125, "learning_rate": 1.668869821549811e-05, "loss": 1.3882, "step": 11551 }, { "epoch": 28.88, "grad_norm": 6.28125, "learning_rate": 1.6688096287453048e-05, "loss": 1.3685, "step": 11552 }, { "epoch": 28.8825, "grad_norm": 7.125, "learning_rate": 1.6687494315561094e-05, "loss": 1.3234, "step": 11553 }, { "epoch": 28.885, "grad_norm": 5.375, "learning_rate": 1.6686892299826196e-05, "loss": 1.4615, "step": 11554 }, { "epoch": 28.8875, "grad_norm": 9.1875, "learning_rate": 1.6686290240252302e-05, "loss": 1.525, "step": 11555 }, { "epoch": 28.89, "grad_norm": 9.0, "learning_rate": 1.6685688136843355e-05, "loss": 1.4752, "step": 11556 }, { "epoch": 28.8925, "grad_norm": 7.875, "learning_rate": 1.6685085989603303e-05, "loss": 1.3868, "step": 11557 }, { "epoch": 28.895, "grad_norm": 7.15625, "learning_rate": 1.6684483798536096e-05, "loss": 1.4594, "step": 11558 }, { "epoch": 28.8975, "grad_norm": 8.8125, "learning_rate": 1.6683881563645677e-05, "loss": 1.3518, "step": 11559 }, { "epoch": 28.9, "grad_norm": 6.3125, "learning_rate": 1.6683279284936004e-05, "loss": 1.3959, "step": 11560 }, { "epoch": 28.9025, "grad_norm": 5.78125, "learning_rate": 1.6682676962411015e-05, "loss": 1.2924, "step": 11561 }, { "epoch": 28.905, "grad_norm": 6.1875, "learning_rate": 1.6682074596074666e-05, "loss": 1.4069, "step": 11562 }, { "epoch": 28.9075, "grad_norm": 19.0, "learning_rate": 1.6681472185930904e-05, "loss": 1.5281, "step": 11563 }, { "epoch": 28.91, "grad_norm": 7.21875, "learning_rate": 1.6680869731983674e-05, "loss": 1.419, "step": 11564 }, { "epoch": 28.9125, "grad_norm": 7.46875, "learning_rate": 1.6680267234236933e-05, "loss": 1.3991, "step": 11565 }, { "epoch": 28.915, "grad_norm": 6.0, "learning_rate": 1.6679664692694627e-05, "loss": 1.4279, "step": 11566 }, { "epoch": 28.9175, "grad_norm": 9.4375, "learning_rate": 1.6679062107360705e-05, "loss": 1.4064, "step": 11567 }, { "epoch": 28.92, "grad_norm": 8.5, "learning_rate": 1.6678459478239116e-05, "loss": 1.3745, "step": 11568 }, { "epoch": 28.9225, "grad_norm": 9.3125, "learning_rate": 1.6677856805333818e-05, "loss": 1.3108, "step": 11569 }, { "epoch": 28.925, "grad_norm": 11.1875, "learning_rate": 1.667725408864876e-05, "loss": 1.3408, "step": 11570 }, { "epoch": 28.9275, "grad_norm": 7.09375, "learning_rate": 1.6676651328187888e-05, "loss": 1.2921, "step": 11571 }, { "epoch": 28.93, "grad_norm": 5.96875, "learning_rate": 1.6676048523955156e-05, "loss": 1.5131, "step": 11572 }, { "epoch": 28.9325, "grad_norm": 6.5, "learning_rate": 1.667544567595452e-05, "loss": 1.332, "step": 11573 }, { "epoch": 28.935, "grad_norm": 8.4375, "learning_rate": 1.667484278418993e-05, "loss": 1.5918, "step": 11574 }, { "epoch": 28.9375, "grad_norm": 6.71875, "learning_rate": 1.6674239848665336e-05, "loss": 1.4342, "step": 11575 }, { "epoch": 28.94, "grad_norm": 6.03125, "learning_rate": 1.667363686938469e-05, "loss": 1.4175, "step": 11576 }, { "epoch": 28.9425, "grad_norm": 7.0, "learning_rate": 1.6673033846351955e-05, "loss": 1.2735, "step": 11577 }, { "epoch": 28.945, "grad_norm": 6.1875, "learning_rate": 1.667243077957107e-05, "loss": 1.1959, "step": 11578 }, { "epoch": 28.9475, "grad_norm": 8.375, "learning_rate": 1.6671827669045998e-05, "loss": 1.4205, "step": 11579 }, { "epoch": 28.95, "grad_norm": 13.5, "learning_rate": 1.6671224514780692e-05, "loss": 1.4054, "step": 11580 }, { "epoch": 28.9525, "grad_norm": 5.84375, "learning_rate": 1.6670621316779104e-05, "loss": 1.4602, "step": 11581 }, { "epoch": 28.955, "grad_norm": 5.5, "learning_rate": 1.667001807504519e-05, "loss": 1.228, "step": 11582 }, { "epoch": 28.9575, "grad_norm": 9.25, "learning_rate": 1.6669414789582905e-05, "loss": 1.431, "step": 11583 }, { "epoch": 28.96, "grad_norm": 9.0, "learning_rate": 1.6668811460396202e-05, "loss": 1.2841, "step": 11584 }, { "epoch": 28.9625, "grad_norm": 25.5, "learning_rate": 1.666820808748904e-05, "loss": 1.3875, "step": 11585 }, { "epoch": 28.965, "grad_norm": 7.03125, "learning_rate": 1.666760467086537e-05, "loss": 1.3802, "step": 11586 }, { "epoch": 28.9675, "grad_norm": 5.375, "learning_rate": 1.666700121052915e-05, "loss": 1.3316, "step": 11587 }, { "epoch": 28.97, "grad_norm": 6.40625, "learning_rate": 1.6666397706484342e-05, "loss": 1.5466, "step": 11588 }, { "epoch": 28.9725, "grad_norm": 6.15625, "learning_rate": 1.666579415873489e-05, "loss": 1.25, "step": 11589 }, { "epoch": 28.975, "grad_norm": 5.5, "learning_rate": 1.6665190567284765e-05, "loss": 1.4724, "step": 11590 }, { "epoch": 28.9775, "grad_norm": 6.5625, "learning_rate": 1.6664586932137913e-05, "loss": 1.4428, "step": 11591 }, { "epoch": 28.98, "grad_norm": 6.75, "learning_rate": 1.66639832532983e-05, "loss": 1.3736, "step": 11592 }, { "epoch": 28.9825, "grad_norm": 8.125, "learning_rate": 1.6663379530769876e-05, "loss": 1.248, "step": 11593 }, { "epoch": 28.985, "grad_norm": 9.5625, "learning_rate": 1.6662775764556605e-05, "loss": 1.293, "step": 11594 }, { "epoch": 28.9875, "grad_norm": 6.5, "learning_rate": 1.6662171954662444e-05, "loss": 1.3137, "step": 11595 }, { "epoch": 28.99, "grad_norm": 8.0625, "learning_rate": 1.6661568101091345e-05, "loss": 1.3497, "step": 11596 }, { "epoch": 28.9925, "grad_norm": 4.78125, "learning_rate": 1.6660964203847274e-05, "loss": 1.2668, "step": 11597 }, { "epoch": 28.995, "grad_norm": 7.0625, "learning_rate": 1.6660360262934187e-05, "loss": 1.4481, "step": 11598 }, { "epoch": 28.9975, "grad_norm": 7.125, "learning_rate": 1.665975627835605e-05, "loss": 1.3987, "step": 11599 }, { "epoch": 29.0, "grad_norm": 8.625, "learning_rate": 1.665915225011681e-05, "loss": 1.3362, "step": 11600 }, { "epoch": 29.0025, "grad_norm": 7.53125, "learning_rate": 1.665854817822044e-05, "loss": 1.2441, "step": 11601 }, { "epoch": 29.005, "grad_norm": 7.0625, "learning_rate": 1.6657944062670896e-05, "loss": 1.2949, "step": 11602 }, { "epoch": 29.0075, "grad_norm": 7.375, "learning_rate": 1.6657339903472132e-05, "loss": 1.437, "step": 11603 }, { "epoch": 29.01, "grad_norm": 7.1875, "learning_rate": 1.6656735700628115e-05, "loss": 1.3861, "step": 11604 }, { "epoch": 29.0125, "grad_norm": 8.1875, "learning_rate": 1.6656131454142806e-05, "loss": 1.442, "step": 11605 }, { "epoch": 29.015, "grad_norm": 6.0625, "learning_rate": 1.6655527164020165e-05, "loss": 1.3267, "step": 11606 }, { "epoch": 29.0175, "grad_norm": 6.25, "learning_rate": 1.6654922830264157e-05, "loss": 1.2599, "step": 11607 }, { "epoch": 29.02, "grad_norm": 10.25, "learning_rate": 1.665431845287874e-05, "loss": 1.4608, "step": 11608 }, { "epoch": 29.0225, "grad_norm": 6.0, "learning_rate": 1.6653714031867874e-05, "loss": 1.4817, "step": 11609 }, { "epoch": 29.025, "grad_norm": 7.65625, "learning_rate": 1.6653109567235528e-05, "loss": 1.2775, "step": 11610 }, { "epoch": 29.0275, "grad_norm": 7.65625, "learning_rate": 1.665250505898566e-05, "loss": 1.4663, "step": 11611 }, { "epoch": 29.03, "grad_norm": 5.3125, "learning_rate": 1.665190050712224e-05, "loss": 1.1626, "step": 11612 }, { "epoch": 29.0325, "grad_norm": 7.0, "learning_rate": 1.6651295911649222e-05, "loss": 1.1035, "step": 11613 }, { "epoch": 29.035, "grad_norm": 8.375, "learning_rate": 1.6650691272570576e-05, "loss": 1.4886, "step": 11614 }, { "epoch": 29.0375, "grad_norm": 8.75, "learning_rate": 1.6650086589890265e-05, "loss": 1.276, "step": 11615 }, { "epoch": 29.04, "grad_norm": 7.65625, "learning_rate": 1.664948186361225e-05, "loss": 1.3241, "step": 11616 }, { "epoch": 29.0425, "grad_norm": 7.59375, "learning_rate": 1.66488770937405e-05, "loss": 1.2981, "step": 11617 }, { "epoch": 29.045, "grad_norm": 7.34375, "learning_rate": 1.6648272280278975e-05, "loss": 1.3814, "step": 11618 }, { "epoch": 29.0475, "grad_norm": 7.71875, "learning_rate": 1.6647667423231647e-05, "loss": 1.3174, "step": 11619 }, { "epoch": 29.05, "grad_norm": 5.84375, "learning_rate": 1.6647062522602474e-05, "loss": 1.4956, "step": 11620 }, { "epoch": 29.0525, "grad_norm": 6.96875, "learning_rate": 1.6646457578395424e-05, "loss": 1.141, "step": 11621 }, { "epoch": 29.055, "grad_norm": 7.40625, "learning_rate": 1.664585259061447e-05, "loss": 1.4123, "step": 11622 }, { "epoch": 29.0575, "grad_norm": 5.90625, "learning_rate": 1.664524755926357e-05, "loss": 1.2531, "step": 11623 }, { "epoch": 29.06, "grad_norm": 5.40625, "learning_rate": 1.6644642484346688e-05, "loss": 1.4061, "step": 11624 }, { "epoch": 29.0625, "grad_norm": 8.6875, "learning_rate": 1.66440373658678e-05, "loss": 1.4014, "step": 11625 }, { "epoch": 29.065, "grad_norm": 7.96875, "learning_rate": 1.6643432203830867e-05, "loss": 1.445, "step": 11626 }, { "epoch": 29.0675, "grad_norm": 10.9375, "learning_rate": 1.6642826998239862e-05, "loss": 1.474, "step": 11627 }, { "epoch": 29.07, "grad_norm": 7.21875, "learning_rate": 1.664222174909875e-05, "loss": 1.4116, "step": 11628 }, { "epoch": 29.0725, "grad_norm": 11.25, "learning_rate": 1.664161645641149e-05, "loss": 1.3636, "step": 11629 }, { "epoch": 29.075, "grad_norm": 6.875, "learning_rate": 1.6641011120182065e-05, "loss": 1.3189, "step": 11630 }, { "epoch": 29.0775, "grad_norm": 6.65625, "learning_rate": 1.6640405740414435e-05, "loss": 1.4008, "step": 11631 }, { "epoch": 29.08, "grad_norm": 9.5, "learning_rate": 1.663980031711257e-05, "loss": 1.4018, "step": 11632 }, { "epoch": 29.0825, "grad_norm": 8.25, "learning_rate": 1.663919485028044e-05, "loss": 1.4582, "step": 11633 }, { "epoch": 29.085, "grad_norm": 6.0, "learning_rate": 1.6638589339922013e-05, "loss": 1.4058, "step": 11634 }, { "epoch": 29.0875, "grad_norm": 5.125, "learning_rate": 1.663798378604126e-05, "loss": 1.2213, "step": 11635 }, { "epoch": 29.09, "grad_norm": 6.40625, "learning_rate": 1.6637378188642156e-05, "loss": 1.3194, "step": 11636 }, { "epoch": 29.0925, "grad_norm": 8.0625, "learning_rate": 1.663677254772866e-05, "loss": 1.4109, "step": 11637 }, { "epoch": 29.095, "grad_norm": 7.96875, "learning_rate": 1.6636166863304754e-05, "loss": 1.3831, "step": 11638 }, { "epoch": 29.0975, "grad_norm": 6.53125, "learning_rate": 1.6635561135374402e-05, "loss": 1.3783, "step": 11639 }, { "epoch": 29.1, "grad_norm": 8.375, "learning_rate": 1.6634955363941573e-05, "loss": 1.4928, "step": 11640 }, { "epoch": 29.1025, "grad_norm": 8.0625, "learning_rate": 1.6634349549010248e-05, "loss": 1.3934, "step": 11641 }, { "epoch": 29.105, "grad_norm": 6.96875, "learning_rate": 1.6633743690584387e-05, "loss": 1.4539, "step": 11642 }, { "epoch": 29.1075, "grad_norm": 7.6875, "learning_rate": 1.6633137788667976e-05, "loss": 1.3146, "step": 11643 }, { "epoch": 29.11, "grad_norm": 5.65625, "learning_rate": 1.663253184326497e-05, "loss": 1.3307, "step": 11644 }, { "epoch": 29.1125, "grad_norm": 7.125, "learning_rate": 1.6631925854379356e-05, "loss": 1.4225, "step": 11645 }, { "epoch": 29.115, "grad_norm": 8.25, "learning_rate": 1.66313198220151e-05, "loss": 1.2883, "step": 11646 }, { "epoch": 29.1175, "grad_norm": 10.3125, "learning_rate": 1.6630713746176174e-05, "loss": 1.4893, "step": 11647 }, { "epoch": 29.12, "grad_norm": 6.625, "learning_rate": 1.6630107626866558e-05, "loss": 1.5524, "step": 11648 }, { "epoch": 29.1225, "grad_norm": 6.40625, "learning_rate": 1.662950146409022e-05, "loss": 1.3344, "step": 11649 }, { "epoch": 29.125, "grad_norm": 8.4375, "learning_rate": 1.6628895257851136e-05, "loss": 1.5477, "step": 11650 }, { "epoch": 29.1275, "grad_norm": 4.875, "learning_rate": 1.662828900815328e-05, "loss": 1.4505, "step": 11651 }, { "epoch": 29.13, "grad_norm": 7.28125, "learning_rate": 1.6627682715000627e-05, "loss": 1.3737, "step": 11652 }, { "epoch": 29.1325, "grad_norm": 7.125, "learning_rate": 1.6627076378397153e-05, "loss": 1.2979, "step": 11653 }, { "epoch": 29.135, "grad_norm": 8.1875, "learning_rate": 1.6626469998346824e-05, "loss": 1.3474, "step": 11654 }, { "epoch": 29.1375, "grad_norm": 6.25, "learning_rate": 1.6625863574853633e-05, "loss": 1.2848, "step": 11655 }, { "epoch": 29.14, "grad_norm": 9.3125, "learning_rate": 1.662525710792154e-05, "loss": 1.3492, "step": 11656 }, { "epoch": 29.1425, "grad_norm": 13.625, "learning_rate": 1.6624650597554524e-05, "loss": 1.4564, "step": 11657 }, { "epoch": 29.145, "grad_norm": 6.8125, "learning_rate": 1.6624044043756573e-05, "loss": 1.3655, "step": 11658 }, { "epoch": 29.1475, "grad_norm": 11.0, "learning_rate": 1.6623437446531645e-05, "loss": 1.5212, "step": 11659 }, { "epoch": 29.15, "grad_norm": 7.71875, "learning_rate": 1.662283080588373e-05, "loss": 1.3056, "step": 11660 }, { "epoch": 29.1525, "grad_norm": 7.40625, "learning_rate": 1.66222241218168e-05, "loss": 1.2485, "step": 11661 }, { "epoch": 29.155, "grad_norm": 7.0625, "learning_rate": 1.6621617394334833e-05, "loss": 1.4566, "step": 11662 }, { "epoch": 29.1575, "grad_norm": 6.34375, "learning_rate": 1.6621010623441812e-05, "loss": 1.2183, "step": 11663 }, { "epoch": 29.16, "grad_norm": 7.71875, "learning_rate": 1.6620403809141707e-05, "loss": 1.3315, "step": 11664 }, { "epoch": 29.1625, "grad_norm": 6.0, "learning_rate": 1.6619796951438498e-05, "loss": 1.2464, "step": 11665 }, { "epoch": 29.165, "grad_norm": 8.0, "learning_rate": 1.6619190050336164e-05, "loss": 1.4551, "step": 11666 }, { "epoch": 29.1675, "grad_norm": 4.21875, "learning_rate": 1.6618583105838692e-05, "loss": 1.3054, "step": 11667 }, { "epoch": 29.17, "grad_norm": 5.46875, "learning_rate": 1.6617976117950046e-05, "loss": 1.3352, "step": 11668 }, { "epoch": 29.1725, "grad_norm": 5.875, "learning_rate": 1.6617369086674218e-05, "loss": 1.1917, "step": 11669 }, { "epoch": 29.175, "grad_norm": 8.8125, "learning_rate": 1.6616762012015182e-05, "loss": 1.4218, "step": 11670 }, { "epoch": 29.1775, "grad_norm": 6.28125, "learning_rate": 1.6616154893976922e-05, "loss": 1.5353, "step": 11671 }, { "epoch": 29.18, "grad_norm": 8.125, "learning_rate": 1.661554773256341e-05, "loss": 1.5872, "step": 11672 }, { "epoch": 29.1825, "grad_norm": 7.40625, "learning_rate": 1.6614940527778635e-05, "loss": 1.2408, "step": 11673 }, { "epoch": 29.185, "grad_norm": 8.625, "learning_rate": 1.6614333279626574e-05, "loss": 1.3525, "step": 11674 }, { "epoch": 29.1875, "grad_norm": 9.3125, "learning_rate": 1.661372598811121e-05, "loss": 1.5009, "step": 11675 }, { "epoch": 29.19, "grad_norm": 10.1875, "learning_rate": 1.661311865323652e-05, "loss": 1.2566, "step": 11676 }, { "epoch": 29.1925, "grad_norm": 10.0, "learning_rate": 1.6612511275006493e-05, "loss": 1.4537, "step": 11677 }, { "epoch": 29.195, "grad_norm": 9.9375, "learning_rate": 1.6611903853425103e-05, "loss": 1.3804, "step": 11678 }, { "epoch": 29.1975, "grad_norm": 6.09375, "learning_rate": 1.6611296388496337e-05, "loss": 1.3837, "step": 11679 }, { "epoch": 29.2, "grad_norm": 6.5625, "learning_rate": 1.6610688880224178e-05, "loss": 1.4445, "step": 11680 }, { "epoch": 29.2025, "grad_norm": 7.09375, "learning_rate": 1.6610081328612602e-05, "loss": 1.2857, "step": 11681 }, { "epoch": 29.205, "grad_norm": 6.90625, "learning_rate": 1.66094737336656e-05, "loss": 1.522, "step": 11682 }, { "epoch": 29.2075, "grad_norm": 9.5625, "learning_rate": 1.6608866095387158e-05, "loss": 1.5109, "step": 11683 }, { "epoch": 29.21, "grad_norm": 6.625, "learning_rate": 1.660825841378125e-05, "loss": 1.4206, "step": 11684 }, { "epoch": 29.2125, "grad_norm": 8.0625, "learning_rate": 1.6607650688851865e-05, "loss": 1.3739, "step": 11685 }, { "epoch": 29.215, "grad_norm": 6.375, "learning_rate": 1.6607042920602984e-05, "loss": 1.2477, "step": 11686 }, { "epoch": 29.2175, "grad_norm": 5.1875, "learning_rate": 1.6606435109038595e-05, "loss": 1.4282, "step": 11687 }, { "epoch": 29.22, "grad_norm": 7.75, "learning_rate": 1.660582725416268e-05, "loss": 1.3257, "step": 11688 }, { "epoch": 29.2225, "grad_norm": 7.59375, "learning_rate": 1.6605219355979233e-05, "loss": 1.3778, "step": 11689 }, { "epoch": 29.225, "grad_norm": 9.25, "learning_rate": 1.660461141449223e-05, "loss": 1.3122, "step": 11690 }, { "epoch": 29.2275, "grad_norm": 9.0625, "learning_rate": 1.6604003429705652e-05, "loss": 1.406, "step": 11691 }, { "epoch": 29.23, "grad_norm": 6.5, "learning_rate": 1.6603395401623498e-05, "loss": 1.1439, "step": 11692 }, { "epoch": 29.2325, "grad_norm": 6.5, "learning_rate": 1.6602787330249744e-05, "loss": 1.4262, "step": 11693 }, { "epoch": 29.235, "grad_norm": 6.875, "learning_rate": 1.6602179215588385e-05, "loss": 1.4625, "step": 11694 }, { "epoch": 29.2375, "grad_norm": 7.40625, "learning_rate": 1.6601571057643405e-05, "loss": 1.51, "step": 11695 }, { "epoch": 29.24, "grad_norm": 6.9375, "learning_rate": 1.6600962856418782e-05, "loss": 1.3596, "step": 11696 }, { "epoch": 29.2425, "grad_norm": 7.53125, "learning_rate": 1.6600354611918518e-05, "loss": 1.3344, "step": 11697 }, { "epoch": 29.245, "grad_norm": 6.125, "learning_rate": 1.6599746324146585e-05, "loss": 1.3778, "step": 11698 }, { "epoch": 29.2475, "grad_norm": 5.15625, "learning_rate": 1.6599137993106987e-05, "loss": 1.3081, "step": 11699 }, { "epoch": 29.25, "grad_norm": 9.5625, "learning_rate": 1.65985296188037e-05, "loss": 1.4391, "step": 11700 }, { "epoch": 29.2525, "grad_norm": 8.125, "learning_rate": 1.6597921201240717e-05, "loss": 1.4202, "step": 11701 }, { "epoch": 29.255, "grad_norm": 8.8125, "learning_rate": 1.659731274042203e-05, "loss": 1.2937, "step": 11702 }, { "epoch": 29.2575, "grad_norm": 8.8125, "learning_rate": 1.659670423635162e-05, "loss": 1.4259, "step": 11703 }, { "epoch": 29.26, "grad_norm": 6.09375, "learning_rate": 1.6596095689033484e-05, "loss": 1.3712, "step": 11704 }, { "epoch": 29.2625, "grad_norm": 6.96875, "learning_rate": 1.6595487098471607e-05, "loss": 1.4598, "step": 11705 }, { "epoch": 29.265, "grad_norm": 7.09375, "learning_rate": 1.6594878464669982e-05, "loss": 1.4424, "step": 11706 }, { "epoch": 29.2675, "grad_norm": 6.15625, "learning_rate": 1.6594269787632597e-05, "loss": 1.1942, "step": 11707 }, { "epoch": 29.27, "grad_norm": 9.125, "learning_rate": 1.6593661067363446e-05, "loss": 1.4837, "step": 11708 }, { "epoch": 29.2725, "grad_norm": 7.78125, "learning_rate": 1.659305230386651e-05, "loss": 1.2808, "step": 11709 }, { "epoch": 29.275, "grad_norm": 7.875, "learning_rate": 1.6592443497145793e-05, "loss": 1.4495, "step": 11710 }, { "epoch": 29.2775, "grad_norm": 7.15625, "learning_rate": 1.659183464720528e-05, "loss": 1.2945, "step": 11711 }, { "epoch": 29.28, "grad_norm": 7.28125, "learning_rate": 1.6591225754048963e-05, "loss": 1.3652, "step": 11712 }, { "epoch": 29.2825, "grad_norm": 7.5, "learning_rate": 1.659061681768083e-05, "loss": 1.6745, "step": 11713 }, { "epoch": 29.285, "grad_norm": 6.625, "learning_rate": 1.659000783810488e-05, "loss": 1.2679, "step": 11714 }, { "epoch": 29.2875, "grad_norm": 8.6875, "learning_rate": 1.65893988153251e-05, "loss": 1.2692, "step": 11715 }, { "epoch": 29.29, "grad_norm": 6.46875, "learning_rate": 1.6588789749345487e-05, "loss": 1.5131, "step": 11716 }, { "epoch": 29.2925, "grad_norm": 6.75, "learning_rate": 1.658818064017003e-05, "loss": 1.424, "step": 11717 }, { "epoch": 29.295, "grad_norm": 10.125, "learning_rate": 1.6587571487802728e-05, "loss": 1.3856, "step": 11718 }, { "epoch": 29.2975, "grad_norm": 5.90625, "learning_rate": 1.6586962292247568e-05, "loss": 1.2257, "step": 11719 }, { "epoch": 29.3, "grad_norm": 6.28125, "learning_rate": 1.6586353053508548e-05, "loss": 1.4109, "step": 11720 }, { "epoch": 29.3025, "grad_norm": 10.1875, "learning_rate": 1.658574377158966e-05, "loss": 1.3694, "step": 11721 }, { "epoch": 29.305, "grad_norm": 7.40625, "learning_rate": 1.65851344464949e-05, "loss": 1.3601, "step": 11722 }, { "epoch": 29.3075, "grad_norm": 7.1875, "learning_rate": 1.6584525078228262e-05, "loss": 1.3521, "step": 11723 }, { "epoch": 29.31, "grad_norm": 6.3125, "learning_rate": 1.658391566679374e-05, "loss": 1.2262, "step": 11724 }, { "epoch": 29.3125, "grad_norm": 5.1875, "learning_rate": 1.658330621219533e-05, "loss": 1.2455, "step": 11725 }, { "epoch": 29.315, "grad_norm": 6.375, "learning_rate": 1.6582696714437033e-05, "loss": 1.4199, "step": 11726 }, { "epoch": 29.3175, "grad_norm": 7.03125, "learning_rate": 1.6582087173522834e-05, "loss": 1.2916, "step": 11727 }, { "epoch": 29.32, "grad_norm": 6.21875, "learning_rate": 1.6581477589456737e-05, "loss": 1.404, "step": 11728 }, { "epoch": 29.3225, "grad_norm": 4.125, "learning_rate": 1.6580867962242735e-05, "loss": 1.2074, "step": 11729 }, { "epoch": 29.325, "grad_norm": 5.09375, "learning_rate": 1.658025829188483e-05, "loss": 1.2149, "step": 11730 }, { "epoch": 29.3275, "grad_norm": 8.5, "learning_rate": 1.657964857838701e-05, "loss": 1.3997, "step": 11731 }, { "epoch": 29.33, "grad_norm": 7.03125, "learning_rate": 1.6579038821753282e-05, "loss": 1.3744, "step": 11732 }, { "epoch": 29.3325, "grad_norm": 6.28125, "learning_rate": 1.6578429021987633e-05, "loss": 1.4286, "step": 11733 }, { "epoch": 29.335, "grad_norm": 6.5, "learning_rate": 1.6577819179094072e-05, "loss": 1.4024, "step": 11734 }, { "epoch": 29.3375, "grad_norm": 10.4375, "learning_rate": 1.6577209293076587e-05, "loss": 1.2585, "step": 11735 }, { "epoch": 29.34, "grad_norm": 4.84375, "learning_rate": 1.6576599363939185e-05, "loss": 1.3427, "step": 11736 }, { "epoch": 29.3425, "grad_norm": 9.0625, "learning_rate": 1.657598939168586e-05, "loss": 1.4308, "step": 11737 }, { "epoch": 29.345, "grad_norm": 6.34375, "learning_rate": 1.657537937632061e-05, "loss": 1.3191, "step": 11738 }, { "epoch": 29.3475, "grad_norm": 9.125, "learning_rate": 1.6574769317847437e-05, "loss": 1.4445, "step": 11739 }, { "epoch": 29.35, "grad_norm": 8.9375, "learning_rate": 1.657415921627034e-05, "loss": 1.4384, "step": 11740 }, { "epoch": 29.3525, "grad_norm": 7.40625, "learning_rate": 1.657354907159332e-05, "loss": 1.1941, "step": 11741 }, { "epoch": 29.355, "grad_norm": 10.0, "learning_rate": 1.657293888382037e-05, "loss": 1.3868, "step": 11742 }, { "epoch": 29.3575, "grad_norm": 5.6875, "learning_rate": 1.65723286529555e-05, "loss": 1.4847, "step": 11743 }, { "epoch": 29.36, "grad_norm": 8.8125, "learning_rate": 1.6571718379002705e-05, "loss": 1.2911, "step": 11744 }, { "epoch": 29.3625, "grad_norm": 6.03125, "learning_rate": 1.657110806196599e-05, "loss": 1.2653, "step": 11745 }, { "epoch": 29.365, "grad_norm": 6.25, "learning_rate": 1.6570497701849347e-05, "loss": 1.3698, "step": 11746 }, { "epoch": 29.3675, "grad_norm": 9.125, "learning_rate": 1.6569887298656787e-05, "loss": 1.3181, "step": 11747 }, { "epoch": 29.37, "grad_norm": 7.40625, "learning_rate": 1.656927685239231e-05, "loss": 1.4076, "step": 11748 }, { "epoch": 29.3725, "grad_norm": 5.46875, "learning_rate": 1.656866636305992e-05, "loss": 1.4042, "step": 11749 }, { "epoch": 29.375, "grad_norm": 7.0625, "learning_rate": 1.656805583066361e-05, "loss": 1.4153, "step": 11750 }, { "epoch": 29.3775, "grad_norm": 7.8125, "learning_rate": 1.656744525520739e-05, "loss": 1.3673, "step": 11751 }, { "epoch": 29.38, "grad_norm": 6.78125, "learning_rate": 1.6566834636695264e-05, "loss": 1.3337, "step": 11752 }, { "epoch": 29.3825, "grad_norm": 9.8125, "learning_rate": 1.6566223975131234e-05, "loss": 1.4331, "step": 11753 }, { "epoch": 29.385, "grad_norm": 5.15625, "learning_rate": 1.65656132705193e-05, "loss": 1.183, "step": 11754 }, { "epoch": 29.3875, "grad_norm": 8.375, "learning_rate": 1.656500252286347e-05, "loss": 1.3788, "step": 11755 }, { "epoch": 29.39, "grad_norm": 7.59375, "learning_rate": 1.6564391732167743e-05, "loss": 1.3991, "step": 11756 }, { "epoch": 29.3925, "grad_norm": 7.34375, "learning_rate": 1.656378089843613e-05, "loss": 1.19, "step": 11757 }, { "epoch": 29.395, "grad_norm": 9.4375, "learning_rate": 1.656317002167263e-05, "loss": 1.3739, "step": 11758 }, { "epoch": 29.3975, "grad_norm": 6.9375, "learning_rate": 1.656255910188125e-05, "loss": 1.268, "step": 11759 }, { "epoch": 29.4, "grad_norm": 7.03125, "learning_rate": 1.6561948139065997e-05, "loss": 1.1956, "step": 11760 }, { "epoch": 29.4025, "grad_norm": 5.5625, "learning_rate": 1.656133713323087e-05, "loss": 1.525, "step": 11761 }, { "epoch": 29.405, "grad_norm": 7.53125, "learning_rate": 1.6560726084379884e-05, "loss": 1.3942, "step": 11762 }, { "epoch": 29.4075, "grad_norm": 11.0625, "learning_rate": 1.656011499251704e-05, "loss": 1.3795, "step": 11763 }, { "epoch": 29.41, "grad_norm": 8.875, "learning_rate": 1.6559503857646342e-05, "loss": 1.4357, "step": 11764 }, { "epoch": 29.4125, "grad_norm": 6.78125, "learning_rate": 1.6558892679771804e-05, "loss": 1.3559, "step": 11765 }, { "epoch": 29.415, "grad_norm": 6.4375, "learning_rate": 1.6558281458897424e-05, "loss": 1.456, "step": 11766 }, { "epoch": 29.4175, "grad_norm": 6.75, "learning_rate": 1.6557670195027212e-05, "loss": 1.3024, "step": 11767 }, { "epoch": 29.42, "grad_norm": 7.9375, "learning_rate": 1.655705888816518e-05, "loss": 1.2339, "step": 11768 }, { "epoch": 29.4225, "grad_norm": 8.6875, "learning_rate": 1.655644753831533e-05, "loss": 1.4921, "step": 11769 }, { "epoch": 29.425, "grad_norm": 7.9375, "learning_rate": 1.6555836145481673e-05, "loss": 1.5667, "step": 11770 }, { "epoch": 29.4275, "grad_norm": 6.4375, "learning_rate": 1.6555224709668218e-05, "loss": 1.4081, "step": 11771 }, { "epoch": 29.43, "grad_norm": 6.4375, "learning_rate": 1.655461323087897e-05, "loss": 1.3186, "step": 11772 }, { "epoch": 29.4325, "grad_norm": 5.59375, "learning_rate": 1.655400170911794e-05, "loss": 1.3588, "step": 11773 }, { "epoch": 29.435, "grad_norm": 7.8125, "learning_rate": 1.655339014438914e-05, "loss": 1.3703, "step": 11774 }, { "epoch": 29.4375, "grad_norm": 8.375, "learning_rate": 1.655277853669657e-05, "loss": 1.3246, "step": 11775 }, { "epoch": 29.44, "grad_norm": 7.53125, "learning_rate": 1.6552166886044253e-05, "loss": 1.2441, "step": 11776 }, { "epoch": 29.4425, "grad_norm": 6.875, "learning_rate": 1.655155519243619e-05, "loss": 1.2821, "step": 11777 }, { "epoch": 29.445, "grad_norm": 6.40625, "learning_rate": 1.6550943455876394e-05, "loss": 1.3421, "step": 11778 }, { "epoch": 29.4475, "grad_norm": 8.9375, "learning_rate": 1.655033167636887e-05, "loss": 1.3323, "step": 11779 }, { "epoch": 29.45, "grad_norm": 7.34375, "learning_rate": 1.654971985391764e-05, "loss": 1.3367, "step": 11780 }, { "epoch": 29.4525, "grad_norm": 6.125, "learning_rate": 1.6549107988526707e-05, "loss": 1.3958, "step": 11781 }, { "epoch": 29.455, "grad_norm": 8.5, "learning_rate": 1.6548496080200083e-05, "loss": 1.4352, "step": 11782 }, { "epoch": 29.4575, "grad_norm": 8.375, "learning_rate": 1.654788412894178e-05, "loss": 1.3893, "step": 11783 }, { "epoch": 29.46, "grad_norm": 6.8125, "learning_rate": 1.6547272134755813e-05, "loss": 1.3584, "step": 11784 }, { "epoch": 29.4625, "grad_norm": 7.09375, "learning_rate": 1.654666009764619e-05, "loss": 1.4373, "step": 11785 }, { "epoch": 29.465, "grad_norm": 6.59375, "learning_rate": 1.654604801761693e-05, "loss": 1.4145, "step": 11786 }, { "epoch": 29.4675, "grad_norm": 6.71875, "learning_rate": 1.6545435894672033e-05, "loss": 1.332, "step": 11787 }, { "epoch": 29.47, "grad_norm": 8.75, "learning_rate": 1.6544823728815528e-05, "loss": 1.3439, "step": 11788 }, { "epoch": 29.4725, "grad_norm": 6.96875, "learning_rate": 1.6544211520051416e-05, "loss": 1.3181, "step": 11789 }, { "epoch": 29.475, "grad_norm": 8.1875, "learning_rate": 1.6543599268383713e-05, "loss": 1.399, "step": 11790 }, { "epoch": 29.4775, "grad_norm": 8.4375, "learning_rate": 1.6542986973816438e-05, "loss": 1.2896, "step": 11791 }, { "epoch": 29.48, "grad_norm": 6.6875, "learning_rate": 1.6542374636353605e-05, "loss": 1.5339, "step": 11792 }, { "epoch": 29.4825, "grad_norm": 8.25, "learning_rate": 1.6541762255999222e-05, "loss": 1.4655, "step": 11793 }, { "epoch": 29.485, "grad_norm": 10.4375, "learning_rate": 1.654114983275731e-05, "loss": 1.3851, "step": 11794 }, { "epoch": 29.4875, "grad_norm": 8.3125, "learning_rate": 1.654053736663188e-05, "loss": 1.4676, "step": 11795 }, { "epoch": 29.49, "grad_norm": 7.71875, "learning_rate": 1.6539924857626947e-05, "loss": 1.34, "step": 11796 }, { "epoch": 29.4925, "grad_norm": 8.75, "learning_rate": 1.653931230574653e-05, "loss": 1.6308, "step": 11797 }, { "epoch": 29.495, "grad_norm": 6.1875, "learning_rate": 1.6538699710994643e-05, "loss": 1.3606, "step": 11798 }, { "epoch": 29.4975, "grad_norm": 7.46875, "learning_rate": 1.6538087073375304e-05, "loss": 1.3419, "step": 11799 }, { "epoch": 29.5, "grad_norm": 5.96875, "learning_rate": 1.6537474392892527e-05, "loss": 1.2944, "step": 11800 }, { "epoch": 29.5025, "grad_norm": 11.3125, "learning_rate": 1.653686166955033e-05, "loss": 1.2508, "step": 11801 }, { "epoch": 29.505, "grad_norm": 9.0, "learning_rate": 1.6536248903352728e-05, "loss": 1.4874, "step": 11802 }, { "epoch": 29.5075, "grad_norm": 7.03125, "learning_rate": 1.653563609430374e-05, "loss": 1.3599, "step": 11803 }, { "epoch": 29.51, "grad_norm": 7.03125, "learning_rate": 1.6535023242407385e-05, "loss": 1.2824, "step": 11804 }, { "epoch": 29.5125, "grad_norm": 7.25, "learning_rate": 1.6534410347667678e-05, "loss": 1.2343, "step": 11805 }, { "epoch": 29.515, "grad_norm": 7.125, "learning_rate": 1.6533797410088636e-05, "loss": 1.3216, "step": 11806 }, { "epoch": 29.5175, "grad_norm": 7.1875, "learning_rate": 1.6533184429674284e-05, "loss": 1.2255, "step": 11807 }, { "epoch": 29.52, "grad_norm": 9.1875, "learning_rate": 1.653257140642863e-05, "loss": 1.3371, "step": 11808 }, { "epoch": 29.5225, "grad_norm": 7.625, "learning_rate": 1.6531958340355703e-05, "loss": 1.3851, "step": 11809 }, { "epoch": 29.525, "grad_norm": 10.5, "learning_rate": 1.653134523145952e-05, "loss": 1.3468, "step": 11810 }, { "epoch": 29.5275, "grad_norm": 5.15625, "learning_rate": 1.6530732079744098e-05, "loss": 1.5984, "step": 11811 }, { "epoch": 29.53, "grad_norm": 7.65625, "learning_rate": 1.6530118885213458e-05, "loss": 1.2225, "step": 11812 }, { "epoch": 29.5325, "grad_norm": 6.6875, "learning_rate": 1.6529505647871617e-05, "loss": 1.2345, "step": 11813 }, { "epoch": 29.535, "grad_norm": 7.96875, "learning_rate": 1.65288923677226e-05, "loss": 1.4752, "step": 11814 }, { "epoch": 29.5375, "grad_norm": 7.25, "learning_rate": 1.652827904477043e-05, "loss": 1.2696, "step": 11815 }, { "epoch": 29.54, "grad_norm": 9.3125, "learning_rate": 1.652766567901912e-05, "loss": 1.4449, "step": 11816 }, { "epoch": 29.5425, "grad_norm": 6.96875, "learning_rate": 1.6527052270472696e-05, "loss": 1.4518, "step": 11817 }, { "epoch": 29.545, "grad_norm": 9.8125, "learning_rate": 1.6526438819135174e-05, "loss": 1.3798, "step": 11818 }, { "epoch": 29.5475, "grad_norm": 7.96875, "learning_rate": 1.6525825325010586e-05, "loss": 1.5078, "step": 11819 }, { "epoch": 29.55, "grad_norm": 6.4375, "learning_rate": 1.6525211788102946e-05, "loss": 1.2356, "step": 11820 }, { "epoch": 29.5525, "grad_norm": 5.90625, "learning_rate": 1.652459820841628e-05, "loss": 1.2553, "step": 11821 }, { "epoch": 29.555, "grad_norm": 7.15625, "learning_rate": 1.652398458595461e-05, "loss": 1.1981, "step": 11822 }, { "epoch": 29.5575, "grad_norm": 9.0625, "learning_rate": 1.6523370920721954e-05, "loss": 1.3913, "step": 11823 }, { "epoch": 29.56, "grad_norm": 6.9375, "learning_rate": 1.6522757212722346e-05, "loss": 1.4113, "step": 11824 }, { "epoch": 29.5625, "grad_norm": 8.0, "learning_rate": 1.6522143461959796e-05, "loss": 1.1466, "step": 11825 }, { "epoch": 29.565, "grad_norm": 6.4375, "learning_rate": 1.6521529668438338e-05, "loss": 1.3083, "step": 11826 }, { "epoch": 29.5675, "grad_norm": 5.5, "learning_rate": 1.652091583216199e-05, "loss": 1.3295, "step": 11827 }, { "epoch": 29.57, "grad_norm": 10.0, "learning_rate": 1.6520301953134782e-05, "loss": 1.26, "step": 11828 }, { "epoch": 29.5725, "grad_norm": 10.5625, "learning_rate": 1.6519688031360732e-05, "loss": 1.2903, "step": 11829 }, { "epoch": 29.575, "grad_norm": 7.71875, "learning_rate": 1.651907406684387e-05, "loss": 1.542, "step": 11830 }, { "epoch": 29.5775, "grad_norm": 7.96875, "learning_rate": 1.651846005958822e-05, "loss": 1.3354, "step": 11831 }, { "epoch": 29.58, "grad_norm": 11.8125, "learning_rate": 1.6517846009597804e-05, "loss": 1.4008, "step": 11832 }, { "epoch": 29.5825, "grad_norm": 7.9375, "learning_rate": 1.651723191687665e-05, "loss": 1.4101, "step": 11833 }, { "epoch": 29.585, "grad_norm": 8.875, "learning_rate": 1.651661778142879e-05, "loss": 1.2186, "step": 11834 }, { "epoch": 29.5875, "grad_norm": 8.3125, "learning_rate": 1.651600360325824e-05, "loss": 1.4975, "step": 11835 }, { "epoch": 29.59, "grad_norm": 6.71875, "learning_rate": 1.6515389382369034e-05, "loss": 1.3643, "step": 11836 }, { "epoch": 29.5925, "grad_norm": 5.6875, "learning_rate": 1.6514775118765195e-05, "loss": 1.2554, "step": 11837 }, { "epoch": 29.595, "grad_norm": 7.40625, "learning_rate": 1.651416081245075e-05, "loss": 1.4794, "step": 11838 }, { "epoch": 29.5975, "grad_norm": 6.90625, "learning_rate": 1.6513546463429726e-05, "loss": 1.2972, "step": 11839 }, { "epoch": 29.6, "grad_norm": 6.625, "learning_rate": 1.6512932071706153e-05, "loss": 1.3272, "step": 11840 }, { "epoch": 29.6025, "grad_norm": 6.46875, "learning_rate": 1.651231763728406e-05, "loss": 1.5623, "step": 11841 }, { "epoch": 29.605, "grad_norm": 9.3125, "learning_rate": 1.651170316016747e-05, "loss": 1.5999, "step": 11842 }, { "epoch": 29.6075, "grad_norm": 9.3125, "learning_rate": 1.6511088640360417e-05, "loss": 1.3435, "step": 11843 }, { "epoch": 29.61, "grad_norm": 6.125, "learning_rate": 1.651047407786693e-05, "loss": 1.4223, "step": 11844 }, { "epoch": 29.6125, "grad_norm": 5.375, "learning_rate": 1.6509859472691033e-05, "loss": 1.4239, "step": 11845 }, { "epoch": 29.615, "grad_norm": 7.6875, "learning_rate": 1.6509244824836755e-05, "loss": 1.4165, "step": 11846 }, { "epoch": 29.6175, "grad_norm": 9.1875, "learning_rate": 1.650863013430813e-05, "loss": 1.3571, "step": 11847 }, { "epoch": 29.62, "grad_norm": 6.21875, "learning_rate": 1.650801540110919e-05, "loss": 1.2762, "step": 11848 }, { "epoch": 29.6225, "grad_norm": 7.59375, "learning_rate": 1.6507400625243957e-05, "loss": 1.524, "step": 11849 }, { "epoch": 29.625, "grad_norm": 8.5, "learning_rate": 1.6506785806716464e-05, "loss": 1.2433, "step": 11850 }, { "epoch": 29.6275, "grad_norm": 5.53125, "learning_rate": 1.650617094553075e-05, "loss": 1.3165, "step": 11851 }, { "epoch": 29.63, "grad_norm": 9.4375, "learning_rate": 1.6505556041690836e-05, "loss": 1.3397, "step": 11852 }, { "epoch": 29.6325, "grad_norm": 7.125, "learning_rate": 1.6504941095200758e-05, "loss": 1.2668, "step": 11853 }, { "epoch": 29.635, "grad_norm": 14.6875, "learning_rate": 1.650432610606455e-05, "loss": 1.6015, "step": 11854 }, { "epoch": 29.6375, "grad_norm": 9.5, "learning_rate": 1.6503711074286232e-05, "loss": 1.4759, "step": 11855 }, { "epoch": 29.64, "grad_norm": 6.0625, "learning_rate": 1.650309599986985e-05, "loss": 1.399, "step": 11856 }, { "epoch": 29.6425, "grad_norm": 5.4375, "learning_rate": 1.650248088281943e-05, "loss": 1.3192, "step": 11857 }, { "epoch": 29.645, "grad_norm": 9.5625, "learning_rate": 1.6501865723139006e-05, "loss": 1.358, "step": 11858 }, { "epoch": 29.6475, "grad_norm": 7.03125, "learning_rate": 1.650125052083261e-05, "loss": 1.3985, "step": 11859 }, { "epoch": 29.65, "grad_norm": 9.0, "learning_rate": 1.6500635275904274e-05, "loss": 1.4289, "step": 11860 }, { "epoch": 29.6525, "grad_norm": 7.96875, "learning_rate": 1.6500019988358036e-05, "loss": 1.3024, "step": 11861 }, { "epoch": 29.655, "grad_norm": 8.25, "learning_rate": 1.6499404658197927e-05, "loss": 1.4505, "step": 11862 }, { "epoch": 29.6575, "grad_norm": 6.0625, "learning_rate": 1.6498789285427977e-05, "loss": 1.5377, "step": 11863 }, { "epoch": 29.66, "grad_norm": 6.5625, "learning_rate": 1.6498173870052227e-05, "loss": 1.4808, "step": 11864 }, { "epoch": 29.6625, "grad_norm": 5.3125, "learning_rate": 1.649755841207471e-05, "loss": 1.4626, "step": 11865 }, { "epoch": 29.665, "grad_norm": 5.90625, "learning_rate": 1.649694291149946e-05, "loss": 1.2017, "step": 11866 }, { "epoch": 29.6675, "grad_norm": 5.875, "learning_rate": 1.6496327368330512e-05, "loss": 1.2341, "step": 11867 }, { "epoch": 29.67, "grad_norm": 5.1875, "learning_rate": 1.6495711782571902e-05, "loss": 1.5105, "step": 11868 }, { "epoch": 29.6725, "grad_norm": 13.75, "learning_rate": 1.6495096154227664e-05, "loss": 1.316, "step": 11869 }, { "epoch": 29.675, "grad_norm": 6.6875, "learning_rate": 1.6494480483301836e-05, "loss": 1.3244, "step": 11870 }, { "epoch": 29.6775, "grad_norm": 7.65625, "learning_rate": 1.6493864769798457e-05, "loss": 1.4197, "step": 11871 }, { "epoch": 29.68, "grad_norm": 7.59375, "learning_rate": 1.6493249013721558e-05, "loss": 1.3439, "step": 11872 }, { "epoch": 29.6825, "grad_norm": 7.40625, "learning_rate": 1.649263321507518e-05, "loss": 1.3624, "step": 11873 }, { "epoch": 29.685, "grad_norm": 4.8125, "learning_rate": 1.6492017373863355e-05, "loss": 1.4209, "step": 11874 }, { "epoch": 29.6875, "grad_norm": 6.28125, "learning_rate": 1.6491401490090125e-05, "loss": 1.3105, "step": 11875 }, { "epoch": 29.69, "grad_norm": 6.90625, "learning_rate": 1.649078556375953e-05, "loss": 1.305, "step": 11876 }, { "epoch": 29.6925, "grad_norm": 8.75, "learning_rate": 1.64901695948756e-05, "loss": 1.3482, "step": 11877 }, { "epoch": 29.695, "grad_norm": 6.09375, "learning_rate": 1.648955358344238e-05, "loss": 1.3147, "step": 11878 }, { "epoch": 29.6975, "grad_norm": 6.96875, "learning_rate": 1.6488937529463905e-05, "loss": 1.4485, "step": 11879 }, { "epoch": 29.7, "grad_norm": 5.375, "learning_rate": 1.6488321432944218e-05, "loss": 1.4728, "step": 11880 }, { "epoch": 29.7025, "grad_norm": 6.21875, "learning_rate": 1.6487705293887356e-05, "loss": 1.5094, "step": 11881 }, { "epoch": 29.705, "grad_norm": 6.03125, "learning_rate": 1.6487089112297355e-05, "loss": 1.2794, "step": 11882 }, { "epoch": 29.7075, "grad_norm": 5.34375, "learning_rate": 1.6486472888178256e-05, "loss": 1.2263, "step": 11883 }, { "epoch": 29.71, "grad_norm": 6.0625, "learning_rate": 1.6485856621534102e-05, "loss": 1.4184, "step": 11884 }, { "epoch": 29.7125, "grad_norm": 5.6875, "learning_rate": 1.648524031236893e-05, "loss": 1.3005, "step": 11885 }, { "epoch": 29.715, "grad_norm": 5.625, "learning_rate": 1.6484623960686786e-05, "loss": 1.3711, "step": 11886 }, { "epoch": 29.7175, "grad_norm": 6.6875, "learning_rate": 1.6484007566491704e-05, "loss": 1.4936, "step": 11887 }, { "epoch": 29.72, "grad_norm": 6.28125, "learning_rate": 1.6483391129787725e-05, "loss": 1.3213, "step": 11888 }, { "epoch": 29.7225, "grad_norm": 7.78125, "learning_rate": 1.6482774650578898e-05, "loss": 1.275, "step": 11889 }, { "epoch": 29.725, "grad_norm": 6.5625, "learning_rate": 1.6482158128869257e-05, "loss": 1.2059, "step": 11890 }, { "epoch": 29.7275, "grad_norm": 8.8125, "learning_rate": 1.6481541564662852e-05, "loss": 1.4781, "step": 11891 }, { "epoch": 29.73, "grad_norm": 6.46875, "learning_rate": 1.648092495796372e-05, "loss": 1.343, "step": 11892 }, { "epoch": 29.7325, "grad_norm": 8.8125, "learning_rate": 1.6480308308775894e-05, "loss": 1.3539, "step": 11893 }, { "epoch": 29.735, "grad_norm": 5.96875, "learning_rate": 1.6479691617103434e-05, "loss": 1.1724, "step": 11894 }, { "epoch": 29.7375, "grad_norm": 7.40625, "learning_rate": 1.647907488295037e-05, "loss": 1.5493, "step": 11895 }, { "epoch": 29.74, "grad_norm": 9.125, "learning_rate": 1.6478458106320755e-05, "loss": 1.4478, "step": 11896 }, { "epoch": 29.7425, "grad_norm": 6.78125, "learning_rate": 1.6477841287218622e-05, "loss": 1.2926, "step": 11897 }, { "epoch": 29.745, "grad_norm": 10.0, "learning_rate": 1.6477224425648026e-05, "loss": 1.2976, "step": 11898 }, { "epoch": 29.7475, "grad_norm": 7.375, "learning_rate": 1.6476607521613006e-05, "loss": 1.3169, "step": 11899 }, { "epoch": 29.75, "grad_norm": 6.8125, "learning_rate": 1.6475990575117603e-05, "loss": 1.1995, "step": 11900 }, { "epoch": 29.7525, "grad_norm": 7.125, "learning_rate": 1.6475373586165866e-05, "loss": 1.3028, "step": 11901 }, { "epoch": 29.755, "grad_norm": 8.0625, "learning_rate": 1.647475655476184e-05, "loss": 1.3268, "step": 11902 }, { "epoch": 29.7575, "grad_norm": 9.25, "learning_rate": 1.6474139480909568e-05, "loss": 1.4048, "step": 11903 }, { "epoch": 29.76, "grad_norm": 5.9375, "learning_rate": 1.64735223646131e-05, "loss": 1.3278, "step": 11904 }, { "epoch": 29.7625, "grad_norm": 5.15625, "learning_rate": 1.6472905205876474e-05, "loss": 1.4944, "step": 11905 }, { "epoch": 29.765, "grad_norm": 8.5, "learning_rate": 1.6472288004703742e-05, "loss": 1.3751, "step": 11906 }, { "epoch": 29.7675, "grad_norm": 7.8125, "learning_rate": 1.647167076109895e-05, "loss": 1.3445, "step": 11907 }, { "epoch": 29.77, "grad_norm": 8.5, "learning_rate": 1.647105347506614e-05, "loss": 1.2903, "step": 11908 }, { "epoch": 29.7725, "grad_norm": 8.75, "learning_rate": 1.6470436146609364e-05, "loss": 1.2999, "step": 11909 }, { "epoch": 29.775, "grad_norm": 7.59375, "learning_rate": 1.646981877573267e-05, "loss": 1.498, "step": 11910 }, { "epoch": 29.7775, "grad_norm": 7.03125, "learning_rate": 1.64692013624401e-05, "loss": 1.4206, "step": 11911 }, { "epoch": 29.78, "grad_norm": 8.0, "learning_rate": 1.646858390673571e-05, "loss": 1.5048, "step": 11912 }, { "epoch": 29.7825, "grad_norm": 9.0625, "learning_rate": 1.6467966408623536e-05, "loss": 1.3634, "step": 11913 }, { "epoch": 29.785, "grad_norm": 7.78125, "learning_rate": 1.646734886810764e-05, "loss": 1.2648, "step": 11914 }, { "epoch": 29.7875, "grad_norm": 7.84375, "learning_rate": 1.6466731285192057e-05, "loss": 1.4796, "step": 11915 }, { "epoch": 29.79, "grad_norm": 9.75, "learning_rate": 1.6466113659880845e-05, "loss": 1.3564, "step": 11916 }, { "epoch": 29.7925, "grad_norm": 6.59375, "learning_rate": 1.646549599217805e-05, "loss": 1.4023, "step": 11917 }, { "epoch": 29.795, "grad_norm": 9.8125, "learning_rate": 1.6464878282087723e-05, "loss": 1.4308, "step": 11918 }, { "epoch": 29.7975, "grad_norm": 6.65625, "learning_rate": 1.6464260529613915e-05, "loss": 1.2812, "step": 11919 }, { "epoch": 29.8, "grad_norm": 7.4375, "learning_rate": 1.646364273476067e-05, "loss": 1.3227, "step": 11920 }, { "epoch": 29.8025, "grad_norm": 4.15625, "learning_rate": 1.6463024897532042e-05, "loss": 1.3617, "step": 11921 }, { "epoch": 29.805, "grad_norm": 6.375, "learning_rate": 1.6462407017932083e-05, "loss": 1.2906, "step": 11922 }, { "epoch": 29.8075, "grad_norm": 8.8125, "learning_rate": 1.646178909596484e-05, "loss": 1.4726, "step": 11923 }, { "epoch": 29.81, "grad_norm": 5.6875, "learning_rate": 1.646117113163437e-05, "loss": 1.255, "step": 11924 }, { "epoch": 29.8125, "grad_norm": 8.75, "learning_rate": 1.646055312494472e-05, "loss": 1.2147, "step": 11925 }, { "epoch": 29.815, "grad_norm": 7.09375, "learning_rate": 1.645993507589994e-05, "loss": 1.3978, "step": 11926 }, { "epoch": 29.8175, "grad_norm": 7.34375, "learning_rate": 1.6459316984504084e-05, "loss": 1.3004, "step": 11927 }, { "epoch": 29.82, "grad_norm": 8.8125, "learning_rate": 1.6458698850761203e-05, "loss": 1.2566, "step": 11928 }, { "epoch": 29.8225, "grad_norm": 9.0, "learning_rate": 1.6458080674675355e-05, "loss": 1.349, "step": 11929 }, { "epoch": 29.825, "grad_norm": 7.09375, "learning_rate": 1.6457462456250586e-05, "loss": 1.3681, "step": 11930 }, { "epoch": 29.8275, "grad_norm": 8.125, "learning_rate": 1.6456844195490954e-05, "loss": 1.4376, "step": 11931 }, { "epoch": 29.83, "grad_norm": 9.25, "learning_rate": 1.6456225892400505e-05, "loss": 1.2634, "step": 11932 }, { "epoch": 29.8325, "grad_norm": 5.53125, "learning_rate": 1.64556075469833e-05, "loss": 1.2246, "step": 11933 }, { "epoch": 29.835, "grad_norm": 8.4375, "learning_rate": 1.645498915924339e-05, "loss": 1.3374, "step": 11934 }, { "epoch": 29.8375, "grad_norm": 9.3125, "learning_rate": 1.645437072918483e-05, "loss": 1.333, "step": 11935 }, { "epoch": 29.84, "grad_norm": 9.8125, "learning_rate": 1.6453752256811676e-05, "loss": 1.4268, "step": 11936 }, { "epoch": 29.8425, "grad_norm": 9.625, "learning_rate": 1.6453133742127978e-05, "loss": 1.4324, "step": 11937 }, { "epoch": 29.845, "grad_norm": 7.40625, "learning_rate": 1.645251518513779e-05, "loss": 1.2753, "step": 11938 }, { "epoch": 29.8475, "grad_norm": 5.25, "learning_rate": 1.645189658584518e-05, "loss": 1.3088, "step": 11939 }, { "epoch": 29.85, "grad_norm": 5.59375, "learning_rate": 1.6451277944254186e-05, "loss": 1.3053, "step": 11940 }, { "epoch": 29.8525, "grad_norm": 9.25, "learning_rate": 1.6450659260368874e-05, "loss": 1.2928, "step": 11941 }, { "epoch": 29.855, "grad_norm": 9.9375, "learning_rate": 1.6450040534193298e-05, "loss": 1.546, "step": 11942 }, { "epoch": 29.8575, "grad_norm": 6.5625, "learning_rate": 1.6449421765731516e-05, "loss": 1.4807, "step": 11943 }, { "epoch": 29.86, "grad_norm": 6.6875, "learning_rate": 1.644880295498758e-05, "loss": 1.2517, "step": 11944 }, { "epoch": 29.8625, "grad_norm": 7.09375, "learning_rate": 1.6448184101965554e-05, "loss": 1.316, "step": 11945 }, { "epoch": 29.865, "grad_norm": 6.5, "learning_rate": 1.644756520666949e-05, "loss": 1.3035, "step": 11946 }, { "epoch": 29.8675, "grad_norm": 6.53125, "learning_rate": 1.6446946269103444e-05, "loss": 1.3909, "step": 11947 }, { "epoch": 29.87, "grad_norm": 4.71875, "learning_rate": 1.6446327289271478e-05, "loss": 1.3055, "step": 11948 }, { "epoch": 29.8725, "grad_norm": 10.6875, "learning_rate": 1.644570826717765e-05, "loss": 1.5058, "step": 11949 }, { "epoch": 29.875, "grad_norm": 7.53125, "learning_rate": 1.644508920282601e-05, "loss": 1.3895, "step": 11950 }, { "epoch": 29.8775, "grad_norm": 5.75, "learning_rate": 1.6444470096220628e-05, "loss": 1.1513, "step": 11951 }, { "epoch": 29.88, "grad_norm": 6.53125, "learning_rate": 1.644385094736556e-05, "loss": 1.2602, "step": 11952 }, { "epoch": 29.8825, "grad_norm": 6.6875, "learning_rate": 1.644323175626486e-05, "loss": 1.4353, "step": 11953 }, { "epoch": 29.885, "grad_norm": 8.125, "learning_rate": 1.644261252292259e-05, "loss": 1.3257, "step": 11954 }, { "epoch": 29.8875, "grad_norm": 6.78125, "learning_rate": 1.644199324734281e-05, "loss": 1.355, "step": 11955 }, { "epoch": 29.89, "grad_norm": 7.21875, "learning_rate": 1.6441373929529583e-05, "loss": 1.229, "step": 11956 }, { "epoch": 29.8925, "grad_norm": 8.4375, "learning_rate": 1.6440754569486964e-05, "loss": 1.3862, "step": 11957 }, { "epoch": 29.895, "grad_norm": 9.125, "learning_rate": 1.6440135167219016e-05, "loss": 1.5081, "step": 11958 }, { "epoch": 29.8975, "grad_norm": 7.625, "learning_rate": 1.6439515722729804e-05, "loss": 1.4132, "step": 11959 }, { "epoch": 29.9, "grad_norm": 6.6875, "learning_rate": 1.6438896236023374e-05, "loss": 1.4297, "step": 11960 }, { "epoch": 29.9025, "grad_norm": 5.4375, "learning_rate": 1.6438276707103805e-05, "loss": 1.202, "step": 11961 }, { "epoch": 29.905, "grad_norm": 6.21875, "learning_rate": 1.643765713597515e-05, "loss": 1.467, "step": 11962 }, { "epoch": 29.9075, "grad_norm": 6.28125, "learning_rate": 1.6437037522641473e-05, "loss": 1.2022, "step": 11963 }, { "epoch": 29.91, "grad_norm": 7.875, "learning_rate": 1.643641786710684e-05, "loss": 1.4572, "step": 11964 }, { "epoch": 29.9125, "grad_norm": 7.5, "learning_rate": 1.6435798169375304e-05, "loss": 1.3531, "step": 11965 }, { "epoch": 29.915, "grad_norm": 6.84375, "learning_rate": 1.6435178429450937e-05, "loss": 1.3504, "step": 11966 }, { "epoch": 29.9175, "grad_norm": 7.5625, "learning_rate": 1.643455864733779e-05, "loss": 1.3698, "step": 11967 }, { "epoch": 29.92, "grad_norm": 7.4375, "learning_rate": 1.6433938823039942e-05, "loss": 1.2753, "step": 11968 }, { "epoch": 29.9225, "grad_norm": 5.6875, "learning_rate": 1.6433318956561443e-05, "loss": 1.3451, "step": 11969 }, { "epoch": 29.925, "grad_norm": 6.90625, "learning_rate": 1.6432699047906364e-05, "loss": 1.3941, "step": 11970 }, { "epoch": 29.9275, "grad_norm": 16.0, "learning_rate": 1.643207909707877e-05, "loss": 1.4431, "step": 11971 }, { "epoch": 29.93, "grad_norm": 5.75, "learning_rate": 1.6431459104082715e-05, "loss": 1.5211, "step": 11972 }, { "epoch": 29.9325, "grad_norm": 7.3125, "learning_rate": 1.643083906892228e-05, "loss": 1.4158, "step": 11973 }, { "epoch": 29.935, "grad_norm": 6.8125, "learning_rate": 1.643021899160152e-05, "loss": 1.4107, "step": 11974 }, { "epoch": 29.9375, "grad_norm": 8.0, "learning_rate": 1.64295988721245e-05, "loss": 1.4671, "step": 11975 }, { "epoch": 29.94, "grad_norm": 8.375, "learning_rate": 1.6428978710495286e-05, "loss": 1.3163, "step": 11976 }, { "epoch": 29.9425, "grad_norm": 7.3125, "learning_rate": 1.6428358506717947e-05, "loss": 1.3682, "step": 11977 }, { "epoch": 29.945, "grad_norm": 7.5, "learning_rate": 1.6427738260796548e-05, "loss": 1.3666, "step": 11978 }, { "epoch": 29.9475, "grad_norm": 7.28125, "learning_rate": 1.642711797273515e-05, "loss": 1.2442, "step": 11979 }, { "epoch": 29.95, "grad_norm": 8.0, "learning_rate": 1.6426497642537826e-05, "loss": 1.4242, "step": 11980 }, { "epoch": 29.9525, "grad_norm": 7.8125, "learning_rate": 1.642587727020864e-05, "loss": 1.367, "step": 11981 }, { "epoch": 29.955, "grad_norm": 6.15625, "learning_rate": 1.642525685575166e-05, "loss": 1.4269, "step": 11982 }, { "epoch": 29.9575, "grad_norm": 7.9375, "learning_rate": 1.6424636399170955e-05, "loss": 1.3566, "step": 11983 }, { "epoch": 29.96, "grad_norm": 8.5, "learning_rate": 1.642401590047059e-05, "loss": 1.4216, "step": 11984 }, { "epoch": 29.9625, "grad_norm": 7.40625, "learning_rate": 1.642339535965463e-05, "loss": 1.528, "step": 11985 }, { "epoch": 29.965, "grad_norm": 6.46875, "learning_rate": 1.6422774776727153e-05, "loss": 1.5048, "step": 11986 }, { "epoch": 29.9675, "grad_norm": 9.0, "learning_rate": 1.6422154151692218e-05, "loss": 1.3401, "step": 11987 }, { "epoch": 29.97, "grad_norm": 10.0625, "learning_rate": 1.64215334845539e-05, "loss": 1.3805, "step": 11988 }, { "epoch": 29.9725, "grad_norm": 9.875, "learning_rate": 1.6420912775316264e-05, "loss": 1.4497, "step": 11989 }, { "epoch": 29.975, "grad_norm": 5.3125, "learning_rate": 1.6420292023983376e-05, "loss": 1.298, "step": 11990 }, { "epoch": 29.9775, "grad_norm": 7.375, "learning_rate": 1.6419671230559318e-05, "loss": 1.2912, "step": 11991 }, { "epoch": 29.98, "grad_norm": 8.375, "learning_rate": 1.6419050395048147e-05, "loss": 1.2335, "step": 11992 }, { "epoch": 29.9825, "grad_norm": 7.1875, "learning_rate": 1.6418429517453943e-05, "loss": 1.3979, "step": 11993 }, { "epoch": 29.985, "grad_norm": 6.5, "learning_rate": 1.6417808597780772e-05, "loss": 1.3947, "step": 11994 }, { "epoch": 29.9875, "grad_norm": 5.71875, "learning_rate": 1.64171876360327e-05, "loss": 1.4458, "step": 11995 }, { "epoch": 29.99, "grad_norm": 7.0, "learning_rate": 1.6416566632213803e-05, "loss": 1.3357, "step": 11996 }, { "epoch": 29.9925, "grad_norm": 5.3125, "learning_rate": 1.6415945586328156e-05, "loss": 1.2898, "step": 11997 }, { "epoch": 29.995, "grad_norm": 6.40625, "learning_rate": 1.6415324498379825e-05, "loss": 1.4017, "step": 11998 }, { "epoch": 29.9975, "grad_norm": 6.90625, "learning_rate": 1.6414703368372883e-05, "loss": 1.3803, "step": 11999 }, { "epoch": 30.0, "grad_norm": 5.59375, "learning_rate": 1.6414082196311402e-05, "loss": 1.4482, "step": 12000 }, { "epoch": 30.0025, "grad_norm": 10.375, "learning_rate": 1.6413460982199455e-05, "loss": 1.4177, "step": 12001 }, { "epoch": 30.005, "grad_norm": 6.34375, "learning_rate": 1.641283972604111e-05, "loss": 1.3419, "step": 12002 }, { "epoch": 30.0075, "grad_norm": 7.375, "learning_rate": 1.6412218427840453e-05, "loss": 1.3448, "step": 12003 }, { "epoch": 30.01, "grad_norm": 6.75, "learning_rate": 1.6411597087601542e-05, "loss": 1.3541, "step": 12004 }, { "epoch": 30.0125, "grad_norm": 7.125, "learning_rate": 1.6410975705328463e-05, "loss": 1.1956, "step": 12005 }, { "epoch": 30.015, "grad_norm": 7.65625, "learning_rate": 1.6410354281025276e-05, "loss": 1.3395, "step": 12006 }, { "epoch": 30.0175, "grad_norm": 5.3125, "learning_rate": 1.640973281469607e-05, "loss": 1.532, "step": 12007 }, { "epoch": 30.02, "grad_norm": 8.1875, "learning_rate": 1.640911130634491e-05, "loss": 1.3058, "step": 12008 }, { "epoch": 30.0225, "grad_norm": 8.625, "learning_rate": 1.640848975597587e-05, "loss": 1.3612, "step": 12009 }, { "epoch": 30.025, "grad_norm": 7.125, "learning_rate": 1.6407868163593028e-05, "loss": 1.2384, "step": 12010 }, { "epoch": 30.0275, "grad_norm": 8.25, "learning_rate": 1.6407246529200463e-05, "loss": 1.2544, "step": 12011 }, { "epoch": 30.03, "grad_norm": 6.21875, "learning_rate": 1.6406624852802243e-05, "loss": 1.4645, "step": 12012 }, { "epoch": 30.0325, "grad_norm": 7.15625, "learning_rate": 1.6406003134402445e-05, "loss": 1.2595, "step": 12013 }, { "epoch": 30.035, "grad_norm": 6.1875, "learning_rate": 1.6405381374005153e-05, "loss": 1.3678, "step": 12014 }, { "epoch": 30.0375, "grad_norm": 7.59375, "learning_rate": 1.6404759571614432e-05, "loss": 1.2935, "step": 12015 }, { "epoch": 30.04, "grad_norm": 8.8125, "learning_rate": 1.6404137727234366e-05, "loss": 1.3817, "step": 12016 }, { "epoch": 30.0425, "grad_norm": 8.25, "learning_rate": 1.6403515840869025e-05, "loss": 1.4384, "step": 12017 }, { "epoch": 30.045, "grad_norm": 7.1875, "learning_rate": 1.6402893912522495e-05, "loss": 1.3867, "step": 12018 }, { "epoch": 30.0475, "grad_norm": 7.4375, "learning_rate": 1.640227194219885e-05, "loss": 1.2863, "step": 12019 }, { "epoch": 30.05, "grad_norm": 5.75, "learning_rate": 1.640164992990216e-05, "loss": 1.4293, "step": 12020 }, { "epoch": 30.0525, "grad_norm": 6.78125, "learning_rate": 1.6401027875636516e-05, "loss": 1.3866, "step": 12021 }, { "epoch": 30.055, "grad_norm": 8.8125, "learning_rate": 1.6400405779405986e-05, "loss": 1.4457, "step": 12022 }, { "epoch": 30.0575, "grad_norm": 5.46875, "learning_rate": 1.6399783641214653e-05, "loss": 1.3886, "step": 12023 }, { "epoch": 30.06, "grad_norm": 6.96875, "learning_rate": 1.6399161461066593e-05, "loss": 1.2333, "step": 12024 }, { "epoch": 30.0625, "grad_norm": 8.8125, "learning_rate": 1.6398539238965886e-05, "loss": 1.4532, "step": 12025 }, { "epoch": 30.065, "grad_norm": 6.40625, "learning_rate": 1.639791697491661e-05, "loss": 1.4136, "step": 12026 }, { "epoch": 30.0675, "grad_norm": 6.75, "learning_rate": 1.639729466892285e-05, "loss": 1.3079, "step": 12027 }, { "epoch": 30.07, "grad_norm": 6.9375, "learning_rate": 1.6396672320988684e-05, "loss": 1.3199, "step": 12028 }, { "epoch": 30.0725, "grad_norm": 11.75, "learning_rate": 1.6396049931118187e-05, "loss": 1.4138, "step": 12029 }, { "epoch": 30.075, "grad_norm": 6.875, "learning_rate": 1.6395427499315442e-05, "loss": 1.373, "step": 12030 }, { "epoch": 30.0775, "grad_norm": 6.09375, "learning_rate": 1.6394805025584533e-05, "loss": 1.2025, "step": 12031 }, { "epoch": 30.08, "grad_norm": 6.6875, "learning_rate": 1.639418250992954e-05, "loss": 1.2294, "step": 12032 }, { "epoch": 30.0825, "grad_norm": 8.5625, "learning_rate": 1.6393559952354538e-05, "loss": 1.3609, "step": 12033 }, { "epoch": 30.085, "grad_norm": 24.375, "learning_rate": 1.639293735286361e-05, "loss": 1.3131, "step": 12034 }, { "epoch": 30.0875, "grad_norm": 9.8125, "learning_rate": 1.6392314711460845e-05, "loss": 1.3265, "step": 12035 }, { "epoch": 30.09, "grad_norm": 6.40625, "learning_rate": 1.6391692028150323e-05, "loss": 1.3397, "step": 12036 }, { "epoch": 30.0925, "grad_norm": 5.25, "learning_rate": 1.639106930293612e-05, "loss": 1.4595, "step": 12037 }, { "epoch": 30.095, "grad_norm": 7.15625, "learning_rate": 1.6390446535822323e-05, "loss": 1.4021, "step": 12038 }, { "epoch": 30.0975, "grad_norm": 4.78125, "learning_rate": 1.6389823726813015e-05, "loss": 1.3143, "step": 12039 }, { "epoch": 30.1, "grad_norm": 8.6875, "learning_rate": 1.638920087591228e-05, "loss": 1.2532, "step": 12040 }, { "epoch": 30.1025, "grad_norm": 7.125, "learning_rate": 1.6388577983124197e-05, "loss": 1.45, "step": 12041 }, { "epoch": 30.105, "grad_norm": 6.40625, "learning_rate": 1.638795504845285e-05, "loss": 1.2544, "step": 12042 }, { "epoch": 30.1075, "grad_norm": 8.375, "learning_rate": 1.638733207190233e-05, "loss": 1.2434, "step": 12043 }, { "epoch": 30.11, "grad_norm": 7.875, "learning_rate": 1.6386709053476715e-05, "loss": 1.4225, "step": 12044 }, { "epoch": 30.1125, "grad_norm": 8.375, "learning_rate": 1.638608599318009e-05, "loss": 1.3407, "step": 12045 }, { "epoch": 30.115, "grad_norm": 6.6875, "learning_rate": 1.6385462891016543e-05, "loss": 1.3357, "step": 12046 }, { "epoch": 30.1175, "grad_norm": 8.125, "learning_rate": 1.6384839746990153e-05, "loss": 1.2682, "step": 12047 }, { "epoch": 30.12, "grad_norm": 8.4375, "learning_rate": 1.6384216561105014e-05, "loss": 1.3438, "step": 12048 }, { "epoch": 30.1225, "grad_norm": 6.1875, "learning_rate": 1.6383593333365203e-05, "loss": 1.1801, "step": 12049 }, { "epoch": 30.125, "grad_norm": 6.53125, "learning_rate": 1.638297006377481e-05, "loss": 1.2369, "step": 12050 }, { "epoch": 30.1275, "grad_norm": 6.375, "learning_rate": 1.6382346752337923e-05, "loss": 1.4684, "step": 12051 }, { "epoch": 30.13, "grad_norm": 7.09375, "learning_rate": 1.6381723399058623e-05, "loss": 1.4078, "step": 12052 }, { "epoch": 30.1325, "grad_norm": 9.1875, "learning_rate": 1.6381100003941005e-05, "loss": 1.4133, "step": 12053 }, { "epoch": 30.135, "grad_norm": 9.125, "learning_rate": 1.6380476566989145e-05, "loss": 1.4073, "step": 12054 }, { "epoch": 30.1375, "grad_norm": 6.96875, "learning_rate": 1.6379853088207138e-05, "loss": 1.3553, "step": 12055 }, { "epoch": 30.14, "grad_norm": 7.28125, "learning_rate": 1.6379229567599072e-05, "loss": 1.4272, "step": 12056 }, { "epoch": 30.1425, "grad_norm": 6.75, "learning_rate": 1.6378606005169028e-05, "loss": 1.4281, "step": 12057 }, { "epoch": 30.145, "grad_norm": 7.4375, "learning_rate": 1.63779824009211e-05, "loss": 1.4138, "step": 12058 }, { "epoch": 30.1475, "grad_norm": 9.8125, "learning_rate": 1.6377358754859373e-05, "loss": 1.5384, "step": 12059 }, { "epoch": 30.15, "grad_norm": 8.0, "learning_rate": 1.637673506698794e-05, "loss": 1.5312, "step": 12060 }, { "epoch": 30.1525, "grad_norm": 7.84375, "learning_rate": 1.637611133731089e-05, "loss": 1.3028, "step": 12061 }, { "epoch": 30.155, "grad_norm": 6.625, "learning_rate": 1.63754875658323e-05, "loss": 1.4127, "step": 12062 }, { "epoch": 30.1575, "grad_norm": 6.3125, "learning_rate": 1.6374863752556275e-05, "loss": 1.1879, "step": 12063 }, { "epoch": 30.16, "grad_norm": 6.53125, "learning_rate": 1.63742398974869e-05, "loss": 1.3024, "step": 12064 }, { "epoch": 30.1625, "grad_norm": 7.25, "learning_rate": 1.6373616000628257e-05, "loss": 1.3441, "step": 12065 }, { "epoch": 30.165, "grad_norm": 13.375, "learning_rate": 1.6372992061984447e-05, "loss": 1.6145, "step": 12066 }, { "epoch": 30.1675, "grad_norm": 7.5625, "learning_rate": 1.6372368081559557e-05, "loss": 1.4241, "step": 12067 }, { "epoch": 30.17, "grad_norm": 9.375, "learning_rate": 1.6371744059357675e-05, "loss": 1.3014, "step": 12068 }, { "epoch": 30.1725, "grad_norm": 12.125, "learning_rate": 1.6371119995382892e-05, "loss": 1.4023, "step": 12069 }, { "epoch": 30.175, "grad_norm": 7.625, "learning_rate": 1.6370495889639302e-05, "loss": 1.4003, "step": 12070 }, { "epoch": 30.1775, "grad_norm": 9.375, "learning_rate": 1.6369871742131e-05, "loss": 1.2902, "step": 12071 }, { "epoch": 30.18, "grad_norm": 6.9375, "learning_rate": 1.636924755286207e-05, "loss": 1.4077, "step": 12072 }, { "epoch": 30.1825, "grad_norm": 5.9375, "learning_rate": 1.636862332183661e-05, "loss": 1.2662, "step": 12073 }, { "epoch": 30.185, "grad_norm": 7.46875, "learning_rate": 1.636799904905871e-05, "loss": 1.4898, "step": 12074 }, { "epoch": 30.1875, "grad_norm": 8.625, "learning_rate": 1.636737473453246e-05, "loss": 1.4725, "step": 12075 }, { "epoch": 30.19, "grad_norm": 7.25, "learning_rate": 1.636675037826196e-05, "loss": 1.3809, "step": 12076 }, { "epoch": 30.1925, "grad_norm": 6.75, "learning_rate": 1.6366125980251298e-05, "loss": 1.2453, "step": 12077 }, { "epoch": 30.195, "grad_norm": 8.5, "learning_rate": 1.636550154050457e-05, "loss": 1.3477, "step": 12078 }, { "epoch": 30.1975, "grad_norm": 9.5, "learning_rate": 1.6364877059025864e-05, "loss": 1.483, "step": 12079 }, { "epoch": 30.2, "grad_norm": 8.4375, "learning_rate": 1.6364252535819284e-05, "loss": 1.4424, "step": 12080 }, { "epoch": 30.2025, "grad_norm": 7.09375, "learning_rate": 1.6363627970888914e-05, "loss": 1.3189, "step": 12081 }, { "epoch": 30.205, "grad_norm": 8.125, "learning_rate": 1.636300336423886e-05, "loss": 1.3967, "step": 12082 }, { "epoch": 30.2075, "grad_norm": 5.84375, "learning_rate": 1.6362378715873206e-05, "loss": 1.3173, "step": 12083 }, { "epoch": 30.21, "grad_norm": 6.46875, "learning_rate": 1.636175402579605e-05, "loss": 1.3502, "step": 12084 }, { "epoch": 30.2125, "grad_norm": 7.84375, "learning_rate": 1.6361129294011496e-05, "loss": 1.4983, "step": 12085 }, { "epoch": 30.215, "grad_norm": 7.125, "learning_rate": 1.6360504520523628e-05, "loss": 1.3588, "step": 12086 }, { "epoch": 30.2175, "grad_norm": 6.6875, "learning_rate": 1.6359879705336548e-05, "loss": 1.2267, "step": 12087 }, { "epoch": 30.22, "grad_norm": 8.1875, "learning_rate": 1.6359254848454352e-05, "loss": 1.5021, "step": 12088 }, { "epoch": 30.2225, "grad_norm": 6.6875, "learning_rate": 1.6358629949881133e-05, "loss": 1.4493, "step": 12089 }, { "epoch": 30.225, "grad_norm": 5.1875, "learning_rate": 1.6358005009620992e-05, "loss": 1.4735, "step": 12090 }, { "epoch": 30.2275, "grad_norm": 6.84375, "learning_rate": 1.6357380027678027e-05, "loss": 1.4874, "step": 12091 }, { "epoch": 30.23, "grad_norm": 9.25, "learning_rate": 1.635675500405633e-05, "loss": 1.2574, "step": 12092 }, { "epoch": 30.2325, "grad_norm": 6.40625, "learning_rate": 1.6356129938760005e-05, "loss": 1.3755, "step": 12093 }, { "epoch": 30.235, "grad_norm": 5.8125, "learning_rate": 1.635550483179314e-05, "loss": 1.3765, "step": 12094 }, { "epoch": 30.2375, "grad_norm": 10.1875, "learning_rate": 1.6354879683159842e-05, "loss": 1.2993, "step": 12095 }, { "epoch": 30.24, "grad_norm": 8.375, "learning_rate": 1.635425449286421e-05, "loss": 1.3343, "step": 12096 }, { "epoch": 30.2425, "grad_norm": 7.4375, "learning_rate": 1.635362926091034e-05, "loss": 1.2806, "step": 12097 }, { "epoch": 30.245, "grad_norm": 7.65625, "learning_rate": 1.635300398730233e-05, "loss": 1.2353, "step": 12098 }, { "epoch": 30.2475, "grad_norm": 7.28125, "learning_rate": 1.6352378672044274e-05, "loss": 1.3951, "step": 12099 }, { "epoch": 30.25, "grad_norm": 5.15625, "learning_rate": 1.6351753315140285e-05, "loss": 1.3563, "step": 12100 }, { "epoch": 30.2525, "grad_norm": 7.15625, "learning_rate": 1.6351127916594455e-05, "loss": 1.4959, "step": 12101 }, { "epoch": 30.255, "grad_norm": 7.25, "learning_rate": 1.6350502476410883e-05, "loss": 1.4345, "step": 12102 }, { "epoch": 30.2575, "grad_norm": 7.34375, "learning_rate": 1.634987699459367e-05, "loss": 1.3019, "step": 12103 }, { "epoch": 30.26, "grad_norm": 8.6875, "learning_rate": 1.6349251471146917e-05, "loss": 1.4841, "step": 12104 }, { "epoch": 30.2625, "grad_norm": 7.21875, "learning_rate": 1.6348625906074728e-05, "loss": 1.3749, "step": 12105 }, { "epoch": 30.265, "grad_norm": 10.25, "learning_rate": 1.6348000299381203e-05, "loss": 1.4173, "step": 12106 }, { "epoch": 30.2675, "grad_norm": 9.25, "learning_rate": 1.634737465107044e-05, "loss": 1.4406, "step": 12107 }, { "epoch": 30.27, "grad_norm": 6.8125, "learning_rate": 1.6346748961146542e-05, "loss": 1.2262, "step": 12108 }, { "epoch": 30.2725, "grad_norm": 9.375, "learning_rate": 1.6346123229613612e-05, "loss": 1.474, "step": 12109 }, { "epoch": 30.275, "grad_norm": 10.4375, "learning_rate": 1.6345497456475752e-05, "loss": 1.3041, "step": 12110 }, { "epoch": 30.2775, "grad_norm": 6.15625, "learning_rate": 1.6344871641737064e-05, "loss": 1.1061, "step": 12111 }, { "epoch": 30.28, "grad_norm": 6.875, "learning_rate": 1.6344245785401653e-05, "loss": 1.2977, "step": 12112 }, { "epoch": 30.2825, "grad_norm": 7.1875, "learning_rate": 1.634361988747362e-05, "loss": 1.4132, "step": 12113 }, { "epoch": 30.285, "grad_norm": 6.65625, "learning_rate": 1.634299394795707e-05, "loss": 1.3835, "step": 12114 }, { "epoch": 30.2875, "grad_norm": 7.5625, "learning_rate": 1.63423679668561e-05, "loss": 1.2996, "step": 12115 }, { "epoch": 30.29, "grad_norm": 7.0625, "learning_rate": 1.6341741944174824e-05, "loss": 1.3287, "step": 12116 }, { "epoch": 30.2925, "grad_norm": 7.40625, "learning_rate": 1.634111587991734e-05, "loss": 1.3408, "step": 12117 }, { "epoch": 30.295, "grad_norm": 12.75, "learning_rate": 1.6340489774087758e-05, "loss": 1.2792, "step": 12118 }, { "epoch": 30.2975, "grad_norm": 5.96875, "learning_rate": 1.6339863626690172e-05, "loss": 1.2869, "step": 12119 }, { "epoch": 30.3, "grad_norm": 6.6875, "learning_rate": 1.63392374377287e-05, "loss": 1.3181, "step": 12120 }, { "epoch": 30.3025, "grad_norm": 6.84375, "learning_rate": 1.633861120720744e-05, "loss": 1.2864, "step": 12121 }, { "epoch": 30.305, "grad_norm": 33.75, "learning_rate": 1.6337984935130497e-05, "loss": 1.2328, "step": 12122 }, { "epoch": 30.3075, "grad_norm": 7.25, "learning_rate": 1.633735862150198e-05, "loss": 1.2754, "step": 12123 }, { "epoch": 30.31, "grad_norm": 11.375, "learning_rate": 1.633673226632599e-05, "loss": 1.3652, "step": 12124 }, { "epoch": 30.3125, "grad_norm": 7.0625, "learning_rate": 1.633610586960664e-05, "loss": 1.4217, "step": 12125 }, { "epoch": 30.315, "grad_norm": 8.0625, "learning_rate": 1.6335479431348034e-05, "loss": 1.45, "step": 12126 }, { "epoch": 30.3175, "grad_norm": 8.875, "learning_rate": 1.6334852951554276e-05, "loss": 1.3961, "step": 12127 }, { "epoch": 30.32, "grad_norm": 5.1875, "learning_rate": 1.6334226430229475e-05, "loss": 1.357, "step": 12128 }, { "epoch": 30.3225, "grad_norm": 6.5, "learning_rate": 1.633359986737774e-05, "loss": 1.4267, "step": 12129 }, { "epoch": 30.325, "grad_norm": 6.875, "learning_rate": 1.633297326300318e-05, "loss": 1.2608, "step": 12130 }, { "epoch": 30.3275, "grad_norm": 6.0625, "learning_rate": 1.63323466171099e-05, "loss": 1.1885, "step": 12131 }, { "epoch": 30.33, "grad_norm": 6.65625, "learning_rate": 1.6331719929702004e-05, "loss": 1.3706, "step": 12132 }, { "epoch": 30.3325, "grad_norm": 6.90625, "learning_rate": 1.633109320078361e-05, "loss": 1.3047, "step": 12133 }, { "epoch": 30.335, "grad_norm": 6.3125, "learning_rate": 1.6330466430358822e-05, "loss": 1.2684, "step": 12134 }, { "epoch": 30.3375, "grad_norm": 7.46875, "learning_rate": 1.6329839618431752e-05, "loss": 1.41, "step": 12135 }, { "epoch": 30.34, "grad_norm": 8.875, "learning_rate": 1.6329212765006503e-05, "loss": 1.3347, "step": 12136 }, { "epoch": 30.3425, "grad_norm": 7.25, "learning_rate": 1.632858587008719e-05, "loss": 1.3152, "step": 12137 }, { "epoch": 30.345, "grad_norm": 7.78125, "learning_rate": 1.632795893367792e-05, "loss": 1.2311, "step": 12138 }, { "epoch": 30.3475, "grad_norm": 6.1875, "learning_rate": 1.6327331955782803e-05, "loss": 1.4478, "step": 12139 }, { "epoch": 30.35, "grad_norm": 9.0, "learning_rate": 1.6326704936405953e-05, "loss": 1.4232, "step": 12140 }, { "epoch": 30.3525, "grad_norm": 6.75, "learning_rate": 1.632607787555148e-05, "loss": 1.4458, "step": 12141 }, { "epoch": 30.355, "grad_norm": 9.3125, "learning_rate": 1.6325450773223494e-05, "loss": 1.6546, "step": 12142 }, { "epoch": 30.3575, "grad_norm": 5.71875, "learning_rate": 1.6324823629426103e-05, "loss": 1.4862, "step": 12143 }, { "epoch": 30.36, "grad_norm": 7.09375, "learning_rate": 1.632419644416342e-05, "loss": 1.4039, "step": 12144 }, { "epoch": 30.3625, "grad_norm": 7.0, "learning_rate": 1.6323569217439564e-05, "loss": 1.3211, "step": 12145 }, { "epoch": 30.365, "grad_norm": 6.625, "learning_rate": 1.6322941949258638e-05, "loss": 1.3772, "step": 12146 }, { "epoch": 30.3675, "grad_norm": 4.875, "learning_rate": 1.6322314639624756e-05, "loss": 1.2026, "step": 12147 }, { "epoch": 30.37, "grad_norm": 13.5625, "learning_rate": 1.6321687288542035e-05, "loss": 1.351, "step": 12148 }, { "epoch": 30.3725, "grad_norm": 6.375, "learning_rate": 1.6321059896014582e-05, "loss": 1.4749, "step": 12149 }, { "epoch": 30.375, "grad_norm": 8.6875, "learning_rate": 1.6320432462046516e-05, "loss": 1.3967, "step": 12150 }, { "epoch": 30.3775, "grad_norm": 5.53125, "learning_rate": 1.6319804986641948e-05, "loss": 1.3138, "step": 12151 }, { "epoch": 30.38, "grad_norm": 6.28125, "learning_rate": 1.631917746980499e-05, "loss": 1.4474, "step": 12152 }, { "epoch": 30.3825, "grad_norm": 7.5, "learning_rate": 1.6318549911539755e-05, "loss": 1.3741, "step": 12153 }, { "epoch": 30.385, "grad_norm": 6.46875, "learning_rate": 1.6317922311850364e-05, "loss": 1.3989, "step": 12154 }, { "epoch": 30.3875, "grad_norm": 5.4375, "learning_rate": 1.6317294670740923e-05, "loss": 1.6204, "step": 12155 }, { "epoch": 30.39, "grad_norm": 7.28125, "learning_rate": 1.6316666988215558e-05, "loss": 1.508, "step": 12156 }, { "epoch": 30.3925, "grad_norm": 9.25, "learning_rate": 1.631603926427837e-05, "loss": 1.3729, "step": 12157 }, { "epoch": 30.395, "grad_norm": 7.34375, "learning_rate": 1.6315411498933485e-05, "loss": 1.59, "step": 12158 }, { "epoch": 30.3975, "grad_norm": 6.0, "learning_rate": 1.6314783692185013e-05, "loss": 1.4689, "step": 12159 }, { "epoch": 30.4, "grad_norm": 5.875, "learning_rate": 1.6314155844037074e-05, "loss": 1.3262, "step": 12160 }, { "epoch": 30.4025, "grad_norm": 8.5, "learning_rate": 1.631352795449378e-05, "loss": 1.4968, "step": 12161 }, { "epoch": 30.405, "grad_norm": 8.25, "learning_rate": 1.6312900023559247e-05, "loss": 1.3603, "step": 12162 }, { "epoch": 30.4075, "grad_norm": 9.1875, "learning_rate": 1.63122720512376e-05, "loss": 1.4484, "step": 12163 }, { "epoch": 30.41, "grad_norm": 7.25, "learning_rate": 1.6311644037532947e-05, "loss": 1.2695, "step": 12164 }, { "epoch": 30.4125, "grad_norm": 6.46875, "learning_rate": 1.6311015982449407e-05, "loss": 1.4929, "step": 12165 }, { "epoch": 30.415, "grad_norm": 6.6875, "learning_rate": 1.63103878859911e-05, "loss": 1.5605, "step": 12166 }, { "epoch": 30.4175, "grad_norm": 5.9375, "learning_rate": 1.630975974816214e-05, "loss": 1.2772, "step": 12167 }, { "epoch": 30.42, "grad_norm": 29.75, "learning_rate": 1.630913156896665e-05, "loss": 1.3763, "step": 12168 }, { "epoch": 30.4225, "grad_norm": 7.9375, "learning_rate": 1.6308503348408747e-05, "loss": 1.4353, "step": 12169 }, { "epoch": 30.425, "grad_norm": 7.125, "learning_rate": 1.630787508649255e-05, "loss": 1.4873, "step": 12170 }, { "epoch": 30.4275, "grad_norm": 8.125, "learning_rate": 1.6307246783222175e-05, "loss": 1.5044, "step": 12171 }, { "epoch": 30.43, "grad_norm": 6.65625, "learning_rate": 1.6306618438601737e-05, "loss": 1.3093, "step": 12172 }, { "epoch": 30.4325, "grad_norm": 8.9375, "learning_rate": 1.6305990052635368e-05, "loss": 1.4021, "step": 12173 }, { "epoch": 30.435, "grad_norm": 7.03125, "learning_rate": 1.6305361625327177e-05, "loss": 1.2661, "step": 12174 }, { "epoch": 30.4375, "grad_norm": 7.5, "learning_rate": 1.630473315668129e-05, "loss": 1.3362, "step": 12175 }, { "epoch": 30.44, "grad_norm": 6.40625, "learning_rate": 1.6304104646701818e-05, "loss": 1.2468, "step": 12176 }, { "epoch": 30.4425, "grad_norm": 8.125, "learning_rate": 1.63034760953929e-05, "loss": 1.3211, "step": 12177 }, { "epoch": 30.445, "grad_norm": 8.1875, "learning_rate": 1.6302847502758633e-05, "loss": 1.4014, "step": 12178 }, { "epoch": 30.4475, "grad_norm": 7.03125, "learning_rate": 1.6302218868803158e-05, "loss": 1.3778, "step": 12179 }, { "epoch": 30.45, "grad_norm": 8.1875, "learning_rate": 1.6301590193530585e-05, "loss": 1.2812, "step": 12180 }, { "epoch": 30.4525, "grad_norm": 16.25, "learning_rate": 1.6300961476945037e-05, "loss": 1.4261, "step": 12181 }, { "epoch": 30.455, "grad_norm": 7.625, "learning_rate": 1.630033271905064e-05, "loss": 1.4224, "step": 12182 }, { "epoch": 30.4575, "grad_norm": 7.1875, "learning_rate": 1.6299703919851517e-05, "loss": 1.3979, "step": 12183 }, { "epoch": 30.46, "grad_norm": 7.59375, "learning_rate": 1.6299075079351784e-05, "loss": 1.3705, "step": 12184 }, { "epoch": 30.4625, "grad_norm": 7.3125, "learning_rate": 1.6298446197555567e-05, "loss": 1.3547, "step": 12185 }, { "epoch": 30.465, "grad_norm": 5.78125, "learning_rate": 1.6297817274466988e-05, "loss": 1.375, "step": 12186 }, { "epoch": 30.4675, "grad_norm": 7.875, "learning_rate": 1.6297188310090174e-05, "loss": 1.4045, "step": 12187 }, { "epoch": 30.47, "grad_norm": 9.4375, "learning_rate": 1.6296559304429242e-05, "loss": 1.2902, "step": 12188 }, { "epoch": 30.4725, "grad_norm": 7.21875, "learning_rate": 1.629593025748832e-05, "loss": 1.3366, "step": 12189 }, { "epoch": 30.475, "grad_norm": 7.0, "learning_rate": 1.629530116927153e-05, "loss": 1.2133, "step": 12190 }, { "epoch": 30.4775, "grad_norm": 7.625, "learning_rate": 1.6294672039782998e-05, "loss": 1.3844, "step": 12191 }, { "epoch": 30.48, "grad_norm": 6.65625, "learning_rate": 1.629404286902685e-05, "loss": 1.402, "step": 12192 }, { "epoch": 30.4825, "grad_norm": 6.96875, "learning_rate": 1.6293413657007212e-05, "loss": 1.3286, "step": 12193 }, { "epoch": 30.485, "grad_norm": 10.0, "learning_rate": 1.62927844037282e-05, "loss": 1.4391, "step": 12194 }, { "epoch": 30.4875, "grad_norm": 6.34375, "learning_rate": 1.6292155109193948e-05, "loss": 1.3044, "step": 12195 }, { "epoch": 30.49, "grad_norm": 6.875, "learning_rate": 1.6291525773408576e-05, "loss": 1.3748, "step": 12196 }, { "epoch": 30.4925, "grad_norm": 11.3125, "learning_rate": 1.6290896396376216e-05, "loss": 1.4179, "step": 12197 }, { "epoch": 30.495, "grad_norm": 8.25, "learning_rate": 1.629026697810099e-05, "loss": 1.3008, "step": 12198 }, { "epoch": 30.4975, "grad_norm": 9.8125, "learning_rate": 1.6289637518587023e-05, "loss": 1.5512, "step": 12199 }, { "epoch": 30.5, "grad_norm": 6.5625, "learning_rate": 1.6289008017838447e-05, "loss": 1.2296, "step": 12200 }, { "epoch": 30.5025, "grad_norm": 7.3125, "learning_rate": 1.6288378475859384e-05, "loss": 1.3919, "step": 12201 }, { "epoch": 30.505, "grad_norm": 7.09375, "learning_rate": 1.6287748892653963e-05, "loss": 1.2288, "step": 12202 }, { "epoch": 30.5075, "grad_norm": 6.90625, "learning_rate": 1.6287119268226316e-05, "loss": 1.4905, "step": 12203 }, { "epoch": 30.51, "grad_norm": 7.34375, "learning_rate": 1.6286489602580564e-05, "loss": 1.4028, "step": 12204 }, { "epoch": 30.5125, "grad_norm": 6.84375, "learning_rate": 1.6285859895720835e-05, "loss": 1.2677, "step": 12205 }, { "epoch": 30.515, "grad_norm": 5.84375, "learning_rate": 1.6285230147651266e-05, "loss": 1.1805, "step": 12206 }, { "epoch": 30.5175, "grad_norm": 8.9375, "learning_rate": 1.6284600358375977e-05, "loss": 1.3278, "step": 12207 }, { "epoch": 30.52, "grad_norm": 15.6875, "learning_rate": 1.62839705278991e-05, "loss": 1.3366, "step": 12208 }, { "epoch": 30.5225, "grad_norm": 6.40625, "learning_rate": 1.6283340656224762e-05, "loss": 1.4276, "step": 12209 }, { "epoch": 30.525, "grad_norm": 10.25, "learning_rate": 1.6282710743357097e-05, "loss": 1.3304, "step": 12210 }, { "epoch": 30.5275, "grad_norm": 7.6875, "learning_rate": 1.6282080789300228e-05, "loss": 1.2975, "step": 12211 }, { "epoch": 30.53, "grad_norm": 6.3125, "learning_rate": 1.628145079405829e-05, "loss": 1.4045, "step": 12212 }, { "epoch": 30.5325, "grad_norm": 8.375, "learning_rate": 1.6280820757635413e-05, "loss": 1.4676, "step": 12213 }, { "epoch": 30.535, "grad_norm": 8.8125, "learning_rate": 1.628019068003573e-05, "loss": 1.4612, "step": 12214 }, { "epoch": 30.5375, "grad_norm": 7.40625, "learning_rate": 1.627956056126336e-05, "loss": 1.3097, "step": 12215 }, { "epoch": 30.54, "grad_norm": 5.78125, "learning_rate": 1.6278930401322445e-05, "loss": 1.2903, "step": 12216 }, { "epoch": 30.5425, "grad_norm": 5.8125, "learning_rate": 1.6278300200217118e-05, "loss": 1.2083, "step": 12217 }, { "epoch": 30.545, "grad_norm": 9.75, "learning_rate": 1.6277669957951502e-05, "loss": 1.3135, "step": 12218 }, { "epoch": 30.5475, "grad_norm": 5.5625, "learning_rate": 1.6277039674529734e-05, "loss": 1.3879, "step": 12219 }, { "epoch": 30.55, "grad_norm": 10.5625, "learning_rate": 1.6276409349955945e-05, "loss": 1.5361, "step": 12220 }, { "epoch": 30.5525, "grad_norm": 9.3125, "learning_rate": 1.627577898423427e-05, "loss": 1.336, "step": 12221 }, { "epoch": 30.555, "grad_norm": 7.4375, "learning_rate": 1.6275148577368835e-05, "loss": 1.434, "step": 12222 }, { "epoch": 30.5575, "grad_norm": 6.6875, "learning_rate": 1.6274518129363782e-05, "loss": 1.3843, "step": 12223 }, { "epoch": 30.56, "grad_norm": 6.53125, "learning_rate": 1.627388764022323e-05, "loss": 1.2878, "step": 12224 }, { "epoch": 30.5625, "grad_norm": 6.375, "learning_rate": 1.627325710995133e-05, "loss": 1.1277, "step": 12225 }, { "epoch": 30.565, "grad_norm": 12.3125, "learning_rate": 1.6272626538552206e-05, "loss": 1.4593, "step": 12226 }, { "epoch": 30.5675, "grad_norm": 8.125, "learning_rate": 1.6271995926029993e-05, "loss": 1.2708, "step": 12227 }, { "epoch": 30.57, "grad_norm": 7.84375, "learning_rate": 1.6271365272388827e-05, "loss": 1.2135, "step": 12228 }, { "epoch": 30.5725, "grad_norm": 9.25, "learning_rate": 1.6270734577632837e-05, "loss": 1.2948, "step": 12229 }, { "epoch": 30.575, "grad_norm": 7.9375, "learning_rate": 1.6270103841766168e-05, "loss": 1.4451, "step": 12230 }, { "epoch": 30.5775, "grad_norm": 8.0, "learning_rate": 1.6269473064792945e-05, "loss": 1.2666, "step": 12231 }, { "epoch": 30.58, "grad_norm": 8.5625, "learning_rate": 1.6268842246717307e-05, "loss": 1.5455, "step": 12232 }, { "epoch": 30.5825, "grad_norm": 7.96875, "learning_rate": 1.6268211387543392e-05, "loss": 1.5016, "step": 12233 }, { "epoch": 30.585, "grad_norm": 8.6875, "learning_rate": 1.6267580487275336e-05, "loss": 1.405, "step": 12234 }, { "epoch": 30.5875, "grad_norm": 6.59375, "learning_rate": 1.6266949545917267e-05, "loss": 1.253, "step": 12235 }, { "epoch": 30.59, "grad_norm": 5.96875, "learning_rate": 1.626631856347333e-05, "loss": 1.3547, "step": 12236 }, { "epoch": 30.5925, "grad_norm": 7.15625, "learning_rate": 1.626568753994766e-05, "loss": 1.3663, "step": 12237 }, { "epoch": 30.595, "grad_norm": 7.6875, "learning_rate": 1.6265056475344394e-05, "loss": 1.3265, "step": 12238 }, { "epoch": 30.5975, "grad_norm": 9.0625, "learning_rate": 1.6264425369667667e-05, "loss": 1.3145, "step": 12239 }, { "epoch": 30.6, "grad_norm": 8.625, "learning_rate": 1.626379422292162e-05, "loss": 1.2957, "step": 12240 }, { "epoch": 30.6025, "grad_norm": 11.9375, "learning_rate": 1.626316303511039e-05, "loss": 1.6326, "step": 12241 }, { "epoch": 30.605, "grad_norm": 8.1875, "learning_rate": 1.626253180623811e-05, "loss": 1.4713, "step": 12242 }, { "epoch": 30.6075, "grad_norm": 6.71875, "learning_rate": 1.6261900536308927e-05, "loss": 1.3765, "step": 12243 }, { "epoch": 30.61, "grad_norm": 5.09375, "learning_rate": 1.6261269225326968e-05, "loss": 1.2598, "step": 12244 }, { "epoch": 30.6125, "grad_norm": 7.03125, "learning_rate": 1.6260637873296382e-05, "loss": 1.2825, "step": 12245 }, { "epoch": 30.615, "grad_norm": 7.21875, "learning_rate": 1.62600064802213e-05, "loss": 1.3858, "step": 12246 }, { "epoch": 30.6175, "grad_norm": 5.375, "learning_rate": 1.6259375046105877e-05, "loss": 1.3672, "step": 12247 }, { "epoch": 30.62, "grad_norm": 7.46875, "learning_rate": 1.6258743570954235e-05, "loss": 1.489, "step": 12248 }, { "epoch": 30.6225, "grad_norm": 5.15625, "learning_rate": 1.6258112054770518e-05, "loss": 1.4841, "step": 12249 }, { "epoch": 30.625, "grad_norm": 9.9375, "learning_rate": 1.6257480497558873e-05, "loss": 1.2309, "step": 12250 }, { "epoch": 30.6275, "grad_norm": 5.90625, "learning_rate": 1.6256848899323433e-05, "loss": 1.1864, "step": 12251 }, { "epoch": 30.63, "grad_norm": 7.8125, "learning_rate": 1.6256217260068346e-05, "loss": 1.3547, "step": 12252 }, { "epoch": 30.6325, "grad_norm": 7.46875, "learning_rate": 1.6255585579797745e-05, "loss": 1.357, "step": 12253 }, { "epoch": 30.635, "grad_norm": 7.3125, "learning_rate": 1.6254953858515777e-05, "loss": 1.4468, "step": 12254 }, { "epoch": 30.6375, "grad_norm": 7.46875, "learning_rate": 1.6254322096226583e-05, "loss": 1.4763, "step": 12255 }, { "epoch": 30.64, "grad_norm": 5.6875, "learning_rate": 1.6253690292934303e-05, "loss": 1.3633, "step": 12256 }, { "epoch": 30.6425, "grad_norm": 7.46875, "learning_rate": 1.6253058448643076e-05, "loss": 1.3208, "step": 12257 }, { "epoch": 30.645, "grad_norm": 7.125, "learning_rate": 1.6252426563357054e-05, "loss": 1.2014, "step": 12258 }, { "epoch": 30.6475, "grad_norm": 12.0625, "learning_rate": 1.625179463708037e-05, "loss": 1.5204, "step": 12259 }, { "epoch": 30.65, "grad_norm": 14.375, "learning_rate": 1.6251162669817172e-05, "loss": 1.3261, "step": 12260 }, { "epoch": 30.6525, "grad_norm": 9.375, "learning_rate": 1.62505306615716e-05, "loss": 1.412, "step": 12261 }, { "epoch": 30.655, "grad_norm": 7.84375, "learning_rate": 1.62498986123478e-05, "loss": 1.4469, "step": 12262 }, { "epoch": 30.6575, "grad_norm": 9.1875, "learning_rate": 1.6249266522149917e-05, "loss": 1.4938, "step": 12263 }, { "epoch": 30.66, "grad_norm": 8.0625, "learning_rate": 1.624863439098209e-05, "loss": 1.4163, "step": 12264 }, { "epoch": 30.6625, "grad_norm": 6.375, "learning_rate": 1.6248002218848468e-05, "loss": 1.2106, "step": 12265 }, { "epoch": 30.665, "grad_norm": 7.75, "learning_rate": 1.624737000575319e-05, "loss": 1.3385, "step": 12266 }, { "epoch": 30.6675, "grad_norm": 8.8125, "learning_rate": 1.6246737751700408e-05, "loss": 1.471, "step": 12267 }, { "epoch": 30.67, "grad_norm": 8.375, "learning_rate": 1.6246105456694263e-05, "loss": 1.2601, "step": 12268 }, { "epoch": 30.6725, "grad_norm": 6.125, "learning_rate": 1.6245473120738897e-05, "loss": 1.3303, "step": 12269 }, { "epoch": 30.675, "grad_norm": 5.9375, "learning_rate": 1.6244840743838463e-05, "loss": 1.2787, "step": 12270 }, { "epoch": 30.6775, "grad_norm": 7.03125, "learning_rate": 1.62442083259971e-05, "loss": 1.3764, "step": 12271 }, { "epoch": 30.68, "grad_norm": 7.71875, "learning_rate": 1.624357586721896e-05, "loss": 1.297, "step": 12272 }, { "epoch": 30.6825, "grad_norm": 9.8125, "learning_rate": 1.6242943367508183e-05, "loss": 1.479, "step": 12273 }, { "epoch": 30.685, "grad_norm": 8.6875, "learning_rate": 1.624231082686892e-05, "loss": 1.492, "step": 12274 }, { "epoch": 30.6875, "grad_norm": 6.53125, "learning_rate": 1.6241678245305323e-05, "loss": 1.4365, "step": 12275 }, { "epoch": 30.69, "grad_norm": 6.71875, "learning_rate": 1.6241045622821526e-05, "loss": 1.4766, "step": 12276 }, { "epoch": 30.6925, "grad_norm": 9.625, "learning_rate": 1.6240412959421687e-05, "loss": 1.5412, "step": 12277 }, { "epoch": 30.695, "grad_norm": 6.75, "learning_rate": 1.623978025510995e-05, "loss": 1.4099, "step": 12278 }, { "epoch": 30.6975, "grad_norm": 7.78125, "learning_rate": 1.6239147509890463e-05, "loss": 1.4608, "step": 12279 }, { "epoch": 30.7, "grad_norm": 5.96875, "learning_rate": 1.6238514723767372e-05, "loss": 1.3626, "step": 12280 }, { "epoch": 30.7025, "grad_norm": 7.0, "learning_rate": 1.623788189674483e-05, "loss": 1.4611, "step": 12281 }, { "epoch": 30.705, "grad_norm": 7.03125, "learning_rate": 1.6237249028826986e-05, "loss": 1.3334, "step": 12282 }, { "epoch": 30.7075, "grad_norm": 8.0, "learning_rate": 1.6236616120017986e-05, "loss": 1.2574, "step": 12283 }, { "epoch": 30.71, "grad_norm": 8.1875, "learning_rate": 1.623598317032198e-05, "loss": 1.3769, "step": 12284 }, { "epoch": 30.7125, "grad_norm": 5.3125, "learning_rate": 1.623535017974312e-05, "loss": 1.5066, "step": 12285 }, { "epoch": 30.715, "grad_norm": 8.6875, "learning_rate": 1.6234717148285547e-05, "loss": 1.3243, "step": 12286 }, { "epoch": 30.7175, "grad_norm": 6.71875, "learning_rate": 1.6234084075953425e-05, "loss": 1.2885, "step": 12287 }, { "epoch": 30.72, "grad_norm": 6.875, "learning_rate": 1.6233450962750895e-05, "loss": 1.2381, "step": 12288 }, { "epoch": 30.7225, "grad_norm": 9.0625, "learning_rate": 1.6232817808682106e-05, "loss": 1.3407, "step": 12289 }, { "epoch": 30.725, "grad_norm": 5.96875, "learning_rate": 1.6232184613751217e-05, "loss": 1.4431, "step": 12290 }, { "epoch": 30.7275, "grad_norm": 7.53125, "learning_rate": 1.6231551377962376e-05, "loss": 1.6342, "step": 12291 }, { "epoch": 30.73, "grad_norm": 8.625, "learning_rate": 1.6230918101319733e-05, "loss": 1.3193, "step": 12292 }, { "epoch": 30.7325, "grad_norm": 6.96875, "learning_rate": 1.623028478382744e-05, "loss": 1.3318, "step": 12293 }, { "epoch": 30.735, "grad_norm": 5.9375, "learning_rate": 1.6229651425489644e-05, "loss": 1.3127, "step": 12294 }, { "epoch": 30.7375, "grad_norm": 7.875, "learning_rate": 1.622901802631051e-05, "loss": 1.2822, "step": 12295 }, { "epoch": 30.74, "grad_norm": 6.53125, "learning_rate": 1.6228384586294178e-05, "loss": 1.4927, "step": 12296 }, { "epoch": 30.7425, "grad_norm": 6.46875, "learning_rate": 1.6227751105444807e-05, "loss": 1.4139, "step": 12297 }, { "epoch": 30.745, "grad_norm": 5.65625, "learning_rate": 1.622711758376655e-05, "loss": 1.2826, "step": 12298 }, { "epoch": 30.7475, "grad_norm": 5.90625, "learning_rate": 1.6226484021263564e-05, "loss": 1.3063, "step": 12299 }, { "epoch": 30.75, "grad_norm": 5.8125, "learning_rate": 1.622585041793999e-05, "loss": 1.4762, "step": 12300 }, { "epoch": 30.7525, "grad_norm": 5.875, "learning_rate": 1.6225216773799992e-05, "loss": 1.4938, "step": 12301 }, { "epoch": 30.755, "grad_norm": 6.8125, "learning_rate": 1.6224583088847726e-05, "loss": 1.5697, "step": 12302 }, { "epoch": 30.7575, "grad_norm": 8.0625, "learning_rate": 1.6223949363087337e-05, "loss": 1.4014, "step": 12303 }, { "epoch": 30.76, "grad_norm": 5.34375, "learning_rate": 1.622331559652299e-05, "loss": 1.349, "step": 12304 }, { "epoch": 30.7625, "grad_norm": 7.65625, "learning_rate": 1.6222681789158833e-05, "loss": 1.3845, "step": 12305 }, { "epoch": 30.765, "grad_norm": 6.46875, "learning_rate": 1.622204794099902e-05, "loss": 1.3529, "step": 12306 }, { "epoch": 30.7675, "grad_norm": 11.375, "learning_rate": 1.6221414052047713e-05, "loss": 1.4018, "step": 12307 }, { "epoch": 30.77, "grad_norm": 7.40625, "learning_rate": 1.6220780122309065e-05, "loss": 1.228, "step": 12308 }, { "epoch": 30.7725, "grad_norm": 6.34375, "learning_rate": 1.6220146151787227e-05, "loss": 1.3301, "step": 12309 }, { "epoch": 30.775, "grad_norm": 7.40625, "learning_rate": 1.6219512140486363e-05, "loss": 1.425, "step": 12310 }, { "epoch": 30.7775, "grad_norm": 8.1875, "learning_rate": 1.6218878088410627e-05, "loss": 1.2864, "step": 12311 }, { "epoch": 30.78, "grad_norm": 8.9375, "learning_rate": 1.6218243995564177e-05, "loss": 1.3714, "step": 12312 }, { "epoch": 30.7825, "grad_norm": 8.3125, "learning_rate": 1.6217609861951163e-05, "loss": 1.4405, "step": 12313 }, { "epoch": 30.785, "grad_norm": 6.40625, "learning_rate": 1.621697568757575e-05, "loss": 1.5074, "step": 12314 }, { "epoch": 30.7875, "grad_norm": 6.40625, "learning_rate": 1.6216341472442094e-05, "loss": 1.3436, "step": 12315 }, { "epoch": 30.79, "grad_norm": 6.84375, "learning_rate": 1.621570721655435e-05, "loss": 1.5334, "step": 12316 }, { "epoch": 30.7925, "grad_norm": 6.40625, "learning_rate": 1.621507291991668e-05, "loss": 1.2748, "step": 12317 }, { "epoch": 30.795, "grad_norm": 10.9375, "learning_rate": 1.6214438582533237e-05, "loss": 1.4268, "step": 12318 }, { "epoch": 30.7975, "grad_norm": 6.59375, "learning_rate": 1.621380420440819e-05, "loss": 1.4258, "step": 12319 }, { "epoch": 30.8, "grad_norm": 7.25, "learning_rate": 1.6213169785545688e-05, "loss": 1.3508, "step": 12320 }, { "epoch": 30.8025, "grad_norm": 6.875, "learning_rate": 1.6212535325949893e-05, "loss": 1.3011, "step": 12321 }, { "epoch": 30.805, "grad_norm": 10.75, "learning_rate": 1.6211900825624965e-05, "loss": 1.2344, "step": 12322 }, { "epoch": 30.8075, "grad_norm": 5.125, "learning_rate": 1.6211266284575066e-05, "loss": 1.3542, "step": 12323 }, { "epoch": 30.81, "grad_norm": 7.15625, "learning_rate": 1.6210631702804353e-05, "loss": 1.5322, "step": 12324 }, { "epoch": 30.8125, "grad_norm": 8.3125, "learning_rate": 1.6209997080316983e-05, "loss": 1.4902, "step": 12325 }, { "epoch": 30.815, "grad_norm": 8.125, "learning_rate": 1.6209362417117127e-05, "loss": 1.1876, "step": 12326 }, { "epoch": 30.8175, "grad_norm": 8.1875, "learning_rate": 1.6208727713208935e-05, "loss": 1.3217, "step": 12327 }, { "epoch": 30.82, "grad_norm": 9.25, "learning_rate": 1.6208092968596574e-05, "loss": 1.4337, "step": 12328 }, { "epoch": 30.8225, "grad_norm": 9.8125, "learning_rate": 1.6207458183284203e-05, "loss": 1.3888, "step": 12329 }, { "epoch": 30.825, "grad_norm": 7.25, "learning_rate": 1.6206823357275985e-05, "loss": 1.4455, "step": 12330 }, { "epoch": 30.8275, "grad_norm": 8.4375, "learning_rate": 1.620618849057608e-05, "loss": 1.3961, "step": 12331 }, { "epoch": 30.83, "grad_norm": 8.3125, "learning_rate": 1.6205553583188653e-05, "loss": 1.3591, "step": 12332 }, { "epoch": 30.8325, "grad_norm": 19.5, "learning_rate": 1.6204918635117864e-05, "loss": 1.416, "step": 12333 }, { "epoch": 30.835, "grad_norm": 6.03125, "learning_rate": 1.620428364636788e-05, "loss": 1.2848, "step": 12334 }, { "epoch": 30.8375, "grad_norm": 6.28125, "learning_rate": 1.620364861694286e-05, "loss": 1.3959, "step": 12335 }, { "epoch": 30.84, "grad_norm": 7.09375, "learning_rate": 1.6203013546846967e-05, "loss": 1.3518, "step": 12336 }, { "epoch": 30.8425, "grad_norm": 7.5625, "learning_rate": 1.6202378436084365e-05, "loss": 1.6797, "step": 12337 }, { "epoch": 30.845, "grad_norm": 9.0625, "learning_rate": 1.6201743284659217e-05, "loss": 1.4769, "step": 12338 }, { "epoch": 30.8475, "grad_norm": 6.71875, "learning_rate": 1.6201108092575692e-05, "loss": 1.2988, "step": 12339 }, { "epoch": 30.85, "grad_norm": 6.1875, "learning_rate": 1.6200472859837946e-05, "loss": 1.4369, "step": 12340 }, { "epoch": 30.8525, "grad_norm": 5.78125, "learning_rate": 1.6199837586450147e-05, "loss": 1.4966, "step": 12341 }, { "epoch": 30.855, "grad_norm": 8.5625, "learning_rate": 1.6199202272416466e-05, "loss": 1.2245, "step": 12342 }, { "epoch": 30.8575, "grad_norm": 7.5625, "learning_rate": 1.619856691774106e-05, "loss": 1.2907, "step": 12343 }, { "epoch": 30.86, "grad_norm": 6.03125, "learning_rate": 1.61979315224281e-05, "loss": 1.447, "step": 12344 }, { "epoch": 30.8625, "grad_norm": 14.4375, "learning_rate": 1.619729608648175e-05, "loss": 1.3603, "step": 12345 }, { "epoch": 30.865, "grad_norm": 4.8125, "learning_rate": 1.619666060990617e-05, "loss": 1.2689, "step": 12346 }, { "epoch": 30.8675, "grad_norm": 8.375, "learning_rate": 1.6196025092705534e-05, "loss": 1.0622, "step": 12347 }, { "epoch": 30.87, "grad_norm": 7.40625, "learning_rate": 1.6195389534884004e-05, "loss": 1.32, "step": 12348 }, { "epoch": 30.8725, "grad_norm": 6.0625, "learning_rate": 1.619475393644575e-05, "loss": 1.403, "step": 12349 }, { "epoch": 30.875, "grad_norm": 7.59375, "learning_rate": 1.6194118297394935e-05, "loss": 1.225, "step": 12350 }, { "epoch": 30.8775, "grad_norm": 8.25, "learning_rate": 1.619348261773573e-05, "loss": 1.3276, "step": 12351 }, { "epoch": 30.88, "grad_norm": 5.875, "learning_rate": 1.61928468974723e-05, "loss": 1.4014, "step": 12352 }, { "epoch": 30.8825, "grad_norm": 7.90625, "learning_rate": 1.619221113660881e-05, "loss": 1.3192, "step": 12353 }, { "epoch": 30.885, "grad_norm": 8.25, "learning_rate": 1.6191575335149437e-05, "loss": 1.3673, "step": 12354 }, { "epoch": 30.8875, "grad_norm": 5.625, "learning_rate": 1.6190939493098344e-05, "loss": 1.3415, "step": 12355 }, { "epoch": 30.89, "grad_norm": 5.625, "learning_rate": 1.6190303610459696e-05, "loss": 1.3207, "step": 12356 }, { "epoch": 30.8925, "grad_norm": 6.90625, "learning_rate": 1.6189667687237666e-05, "loss": 1.4485, "step": 12357 }, { "epoch": 30.895, "grad_norm": 13.25, "learning_rate": 1.6189031723436422e-05, "loss": 1.3441, "step": 12358 }, { "epoch": 30.8975, "grad_norm": 9.3125, "learning_rate": 1.618839571906014e-05, "loss": 1.321, "step": 12359 }, { "epoch": 30.9, "grad_norm": 7.8125, "learning_rate": 1.6187759674112972e-05, "loss": 1.2082, "step": 12360 }, { "epoch": 30.9025, "grad_norm": 9.0625, "learning_rate": 1.6187123588599108e-05, "loss": 1.3637, "step": 12361 }, { "epoch": 30.905, "grad_norm": 5.65625, "learning_rate": 1.6186487462522703e-05, "loss": 1.2637, "step": 12362 }, { "epoch": 30.9075, "grad_norm": 6.375, "learning_rate": 1.618585129588794e-05, "loss": 1.4335, "step": 12363 }, { "epoch": 30.91, "grad_norm": 6.53125, "learning_rate": 1.618521508869898e-05, "loss": 1.3203, "step": 12364 }, { "epoch": 30.9125, "grad_norm": 8.75, "learning_rate": 1.618457884096e-05, "loss": 1.2999, "step": 12365 }, { "epoch": 30.915, "grad_norm": 8.4375, "learning_rate": 1.6183942552675165e-05, "loss": 1.6178, "step": 12366 }, { "epoch": 30.9175, "grad_norm": 9.9375, "learning_rate": 1.618330622384865e-05, "loss": 1.3325, "step": 12367 }, { "epoch": 30.92, "grad_norm": 4.53125, "learning_rate": 1.618266985448463e-05, "loss": 1.3842, "step": 12368 }, { "epoch": 30.9225, "grad_norm": 8.6875, "learning_rate": 1.6182033444587274e-05, "loss": 1.3034, "step": 12369 }, { "epoch": 30.925, "grad_norm": 6.875, "learning_rate": 1.618139699416075e-05, "loss": 1.5733, "step": 12370 }, { "epoch": 30.9275, "grad_norm": 7.84375, "learning_rate": 1.618076050320924e-05, "loss": 1.4469, "step": 12371 }, { "epoch": 30.93, "grad_norm": 9.9375, "learning_rate": 1.6180123971736906e-05, "loss": 1.4589, "step": 12372 }, { "epoch": 30.9325, "grad_norm": 5.46875, "learning_rate": 1.617948739974793e-05, "loss": 1.3071, "step": 12373 }, { "epoch": 30.935, "grad_norm": 7.84375, "learning_rate": 1.617885078724648e-05, "loss": 1.5289, "step": 12374 }, { "epoch": 30.9375, "grad_norm": 5.75, "learning_rate": 1.6178214134236733e-05, "loss": 1.2136, "step": 12375 }, { "epoch": 30.94, "grad_norm": 6.90625, "learning_rate": 1.6177577440722863e-05, "loss": 1.4508, "step": 12376 }, { "epoch": 30.9425, "grad_norm": 6.65625, "learning_rate": 1.617694070670904e-05, "loss": 1.3714, "step": 12377 }, { "epoch": 30.945, "grad_norm": 9.375, "learning_rate": 1.617630393219944e-05, "loss": 1.3615, "step": 12378 }, { "epoch": 30.9475, "grad_norm": 11.1875, "learning_rate": 1.617566711719824e-05, "loss": 1.2818, "step": 12379 }, { "epoch": 30.95, "grad_norm": 8.375, "learning_rate": 1.6175030261709615e-05, "loss": 1.2491, "step": 12380 }, { "epoch": 30.9525, "grad_norm": 8.9375, "learning_rate": 1.6174393365737737e-05, "loss": 1.443, "step": 12381 }, { "epoch": 30.955, "grad_norm": 6.6875, "learning_rate": 1.6173756429286784e-05, "loss": 1.4887, "step": 12382 }, { "epoch": 30.9575, "grad_norm": 8.5625, "learning_rate": 1.6173119452360932e-05, "loss": 1.4533, "step": 12383 }, { "epoch": 30.96, "grad_norm": 7.25, "learning_rate": 1.6172482434964353e-05, "loss": 1.2551, "step": 12384 }, { "epoch": 30.9625, "grad_norm": 7.75, "learning_rate": 1.6171845377101226e-05, "loss": 1.4623, "step": 12385 }, { "epoch": 30.965, "grad_norm": 5.9375, "learning_rate": 1.617120827877573e-05, "loss": 1.4326, "step": 12386 }, { "epoch": 30.9675, "grad_norm": 6.0, "learning_rate": 1.617057113999204e-05, "loss": 1.4407, "step": 12387 }, { "epoch": 30.97, "grad_norm": 8.3125, "learning_rate": 1.616993396075433e-05, "loss": 1.241, "step": 12388 }, { "epoch": 30.9725, "grad_norm": 9.25, "learning_rate": 1.616929674106678e-05, "loss": 1.2193, "step": 12389 }, { "epoch": 30.975, "grad_norm": 5.34375, "learning_rate": 1.6168659480933572e-05, "loss": 1.4172, "step": 12390 }, { "epoch": 30.9775, "grad_norm": 6.78125, "learning_rate": 1.6168022180358876e-05, "loss": 1.5078, "step": 12391 }, { "epoch": 30.98, "grad_norm": 8.125, "learning_rate": 1.6167384839346872e-05, "loss": 1.4322, "step": 12392 }, { "epoch": 30.9825, "grad_norm": 9.0625, "learning_rate": 1.616674745790174e-05, "loss": 1.3251, "step": 12393 }, { "epoch": 30.985, "grad_norm": 6.5, "learning_rate": 1.616611003602766e-05, "loss": 1.2905, "step": 12394 }, { "epoch": 30.9875, "grad_norm": 8.0625, "learning_rate": 1.6165472573728812e-05, "loss": 1.4453, "step": 12395 }, { "epoch": 30.99, "grad_norm": 7.09375, "learning_rate": 1.6164835071009364e-05, "loss": 1.3975, "step": 12396 }, { "epoch": 30.9925, "grad_norm": 7.15625, "learning_rate": 1.6164197527873513e-05, "loss": 1.1876, "step": 12397 }, { "epoch": 30.995, "grad_norm": 7.59375, "learning_rate": 1.6163559944325425e-05, "loss": 1.289, "step": 12398 }, { "epoch": 30.9975, "grad_norm": 9.125, "learning_rate": 1.6162922320369285e-05, "loss": 1.233, "step": 12399 }, { "epoch": 31.0, "grad_norm": 7.9375, "learning_rate": 1.6162284656009276e-05, "loss": 1.3618, "step": 12400 }, { "epoch": 31.0025, "grad_norm": 6.96875, "learning_rate": 1.616164695124957e-05, "loss": 1.4271, "step": 12401 }, { "epoch": 31.005, "grad_norm": 6.0, "learning_rate": 1.616100920609436e-05, "loss": 1.282, "step": 12402 }, { "epoch": 31.0075, "grad_norm": 5.6875, "learning_rate": 1.6160371420547812e-05, "loss": 1.3206, "step": 12403 }, { "epoch": 31.01, "grad_norm": 8.25, "learning_rate": 1.615973359461412e-05, "loss": 1.2809, "step": 12404 }, { "epoch": 31.0125, "grad_norm": 6.34375, "learning_rate": 1.6159095728297463e-05, "loss": 1.3037, "step": 12405 }, { "epoch": 31.015, "grad_norm": 7.46875, "learning_rate": 1.615845782160202e-05, "loss": 1.3475, "step": 12406 }, { "epoch": 31.0175, "grad_norm": 6.3125, "learning_rate": 1.6157819874531972e-05, "loss": 1.3493, "step": 12407 }, { "epoch": 31.02, "grad_norm": 7.46875, "learning_rate": 1.6157181887091503e-05, "loss": 1.3689, "step": 12408 }, { "epoch": 31.0225, "grad_norm": 14.5, "learning_rate": 1.6156543859284796e-05, "loss": 1.3427, "step": 12409 }, { "epoch": 31.025, "grad_norm": 5.96875, "learning_rate": 1.615590579111604e-05, "loss": 1.4506, "step": 12410 }, { "epoch": 31.0275, "grad_norm": 8.0, "learning_rate": 1.6155267682589402e-05, "loss": 1.5013, "step": 12411 }, { "epoch": 31.03, "grad_norm": 10.375, "learning_rate": 1.6154629533709082e-05, "loss": 1.5902, "step": 12412 }, { "epoch": 31.0325, "grad_norm": 7.25, "learning_rate": 1.6153991344479254e-05, "loss": 1.5117, "step": 12413 }, { "epoch": 31.035, "grad_norm": 7.90625, "learning_rate": 1.6153353114904107e-05, "loss": 1.2593, "step": 12414 }, { "epoch": 31.0375, "grad_norm": 6.0, "learning_rate": 1.6152714844987824e-05, "loss": 1.2717, "step": 12415 }, { "epoch": 31.04, "grad_norm": 6.5, "learning_rate": 1.6152076534734585e-05, "loss": 1.465, "step": 12416 }, { "epoch": 31.0425, "grad_norm": 13.25, "learning_rate": 1.615143818414858e-05, "loss": 1.2403, "step": 12417 }, { "epoch": 31.045, "grad_norm": 12.6875, "learning_rate": 1.6150799793233994e-05, "loss": 1.4324, "step": 12418 }, { "epoch": 31.0475, "grad_norm": 6.4375, "learning_rate": 1.615016136199501e-05, "loss": 1.3527, "step": 12419 }, { "epoch": 31.05, "grad_norm": 11.4375, "learning_rate": 1.6149522890435815e-05, "loss": 1.437, "step": 12420 }, { "epoch": 31.0525, "grad_norm": 5.78125, "learning_rate": 1.6148884378560593e-05, "loss": 1.2655, "step": 12421 }, { "epoch": 31.055, "grad_norm": 5.28125, "learning_rate": 1.614824582637353e-05, "loss": 1.2649, "step": 12422 }, { "epoch": 31.0575, "grad_norm": 7.875, "learning_rate": 1.6147607233878814e-05, "loss": 1.4004, "step": 12423 }, { "epoch": 31.06, "grad_norm": 8.3125, "learning_rate": 1.6146968601080633e-05, "loss": 1.3833, "step": 12424 }, { "epoch": 31.0625, "grad_norm": 11.875, "learning_rate": 1.614632992798317e-05, "loss": 1.4675, "step": 12425 }, { "epoch": 31.065, "grad_norm": 8.375, "learning_rate": 1.6145691214590612e-05, "loss": 1.3949, "step": 12426 }, { "epoch": 31.0675, "grad_norm": 5.59375, "learning_rate": 1.6145052460907155e-05, "loss": 1.4434, "step": 12427 }, { "epoch": 31.07, "grad_norm": 7.78125, "learning_rate": 1.614441366693697e-05, "loss": 1.4979, "step": 12428 }, { "epoch": 31.0725, "grad_norm": 7.1875, "learning_rate": 1.6143774832684265e-05, "loss": 1.2399, "step": 12429 }, { "epoch": 31.075, "grad_norm": 9.5625, "learning_rate": 1.614313595815321e-05, "loss": 1.4727, "step": 12430 }, { "epoch": 31.0775, "grad_norm": 8.25, "learning_rate": 1.6142497043348007e-05, "loss": 1.3742, "step": 12431 }, { "epoch": 31.08, "grad_norm": 6.15625, "learning_rate": 1.6141858088272838e-05, "loss": 1.2269, "step": 12432 }, { "epoch": 31.0825, "grad_norm": 7.15625, "learning_rate": 1.6141219092931894e-05, "loss": 1.38, "step": 12433 }, { "epoch": 31.085, "grad_norm": 7.78125, "learning_rate": 1.614058005732936e-05, "loss": 1.3805, "step": 12434 }, { "epoch": 31.0875, "grad_norm": 6.78125, "learning_rate": 1.613994098146943e-05, "loss": 1.2232, "step": 12435 }, { "epoch": 31.09, "grad_norm": 9.0, "learning_rate": 1.6139301865356292e-05, "loss": 1.3578, "step": 12436 }, { "epoch": 31.0925, "grad_norm": 6.75, "learning_rate": 1.613866270899414e-05, "loss": 1.3169, "step": 12437 }, { "epoch": 31.095, "grad_norm": 6.4375, "learning_rate": 1.6138023512387157e-05, "loss": 1.2247, "step": 12438 }, { "epoch": 31.0975, "grad_norm": 6.53125, "learning_rate": 1.6137384275539542e-05, "loss": 1.2591, "step": 12439 }, { "epoch": 31.1, "grad_norm": 6.53125, "learning_rate": 1.6136744998455477e-05, "loss": 1.3908, "step": 12440 }, { "epoch": 31.1025, "grad_norm": 7.5, "learning_rate": 1.613610568113916e-05, "loss": 1.1193, "step": 12441 }, { "epoch": 31.105, "grad_norm": 7.71875, "learning_rate": 1.6135466323594777e-05, "loss": 1.3449, "step": 12442 }, { "epoch": 31.1075, "grad_norm": 7.71875, "learning_rate": 1.6134826925826525e-05, "loss": 1.2623, "step": 12443 }, { "epoch": 31.11, "grad_norm": 6.46875, "learning_rate": 1.6134187487838594e-05, "loss": 1.2729, "step": 12444 }, { "epoch": 31.1125, "grad_norm": 7.5625, "learning_rate": 1.613354800963517e-05, "loss": 1.2131, "step": 12445 }, { "epoch": 31.115, "grad_norm": 7.125, "learning_rate": 1.6132908491220455e-05, "loss": 1.46, "step": 12446 }, { "epoch": 31.1175, "grad_norm": 10.125, "learning_rate": 1.6132268932598637e-05, "loss": 1.3182, "step": 12447 }, { "epoch": 31.12, "grad_norm": 6.8125, "learning_rate": 1.6131629333773908e-05, "loss": 1.2622, "step": 12448 }, { "epoch": 31.1225, "grad_norm": 7.34375, "learning_rate": 1.6130989694750466e-05, "loss": 1.3524, "step": 12449 }, { "epoch": 31.125, "grad_norm": 8.4375, "learning_rate": 1.6130350015532498e-05, "loss": 1.3134, "step": 12450 }, { "epoch": 31.1275, "grad_norm": 6.59375, "learning_rate": 1.61297102961242e-05, "loss": 1.5346, "step": 12451 }, { "epoch": 31.13, "grad_norm": 7.25, "learning_rate": 1.6129070536529767e-05, "loss": 1.3492, "step": 12452 }, { "epoch": 31.1325, "grad_norm": 7.65625, "learning_rate": 1.612843073675339e-05, "loss": 1.1735, "step": 12453 }, { "epoch": 31.135, "grad_norm": 7.78125, "learning_rate": 1.6127790896799275e-05, "loss": 1.2564, "step": 12454 }, { "epoch": 31.1375, "grad_norm": 5.34375, "learning_rate": 1.6127151016671597e-05, "loss": 1.4122, "step": 12455 }, { "epoch": 31.14, "grad_norm": 7.0625, "learning_rate": 1.612651109637457e-05, "loss": 1.4615, "step": 12456 }, { "epoch": 31.1425, "grad_norm": 5.9375, "learning_rate": 1.612587113591238e-05, "loss": 1.4038, "step": 12457 }, { "epoch": 31.145, "grad_norm": 7.53125, "learning_rate": 1.6125231135289225e-05, "loss": 1.409, "step": 12458 }, { "epoch": 31.1475, "grad_norm": 10.125, "learning_rate": 1.6124591094509297e-05, "loss": 1.3369, "step": 12459 }, { "epoch": 31.15, "grad_norm": 6.25, "learning_rate": 1.6123951013576796e-05, "loss": 1.4332, "step": 12460 }, { "epoch": 31.1525, "grad_norm": 8.3125, "learning_rate": 1.6123310892495915e-05, "loss": 1.4728, "step": 12461 }, { "epoch": 31.155, "grad_norm": 10.1875, "learning_rate": 1.6122670731270857e-05, "loss": 1.3689, "step": 12462 }, { "epoch": 31.1575, "grad_norm": 9.125, "learning_rate": 1.6122030529905812e-05, "loss": 1.5153, "step": 12463 }, { "epoch": 31.16, "grad_norm": 8.25, "learning_rate": 1.612139028840498e-05, "loss": 1.3843, "step": 12464 }, { "epoch": 31.1625, "grad_norm": 7.875, "learning_rate": 1.612075000677256e-05, "loss": 1.4927, "step": 12465 }, { "epoch": 31.165, "grad_norm": 12.3125, "learning_rate": 1.6120109685012746e-05, "loss": 1.4384, "step": 12466 }, { "epoch": 31.1675, "grad_norm": 5.28125, "learning_rate": 1.6119469323129736e-05, "loss": 1.4947, "step": 12467 }, { "epoch": 31.17, "grad_norm": 6.125, "learning_rate": 1.6118828921127736e-05, "loss": 1.4261, "step": 12468 }, { "epoch": 31.1725, "grad_norm": 6.0, "learning_rate": 1.6118188479010933e-05, "loss": 1.2972, "step": 12469 }, { "epoch": 31.175, "grad_norm": 7.84375, "learning_rate": 1.6117547996783532e-05, "loss": 1.3808, "step": 12470 }, { "epoch": 31.1775, "grad_norm": 7.59375, "learning_rate": 1.6116907474449734e-05, "loss": 1.3365, "step": 12471 }, { "epoch": 31.18, "grad_norm": 6.75, "learning_rate": 1.6116266912013734e-05, "loss": 1.2132, "step": 12472 }, { "epoch": 31.1825, "grad_norm": 5.3125, "learning_rate": 1.6115626309479733e-05, "loss": 1.2035, "step": 12473 }, { "epoch": 31.185, "grad_norm": 11.25, "learning_rate": 1.611498566685193e-05, "loss": 1.3653, "step": 12474 }, { "epoch": 31.1875, "grad_norm": 6.0, "learning_rate": 1.6114344984134523e-05, "loss": 1.27, "step": 12475 }, { "epoch": 31.19, "grad_norm": 6.25, "learning_rate": 1.611370426133172e-05, "loss": 1.2635, "step": 12476 }, { "epoch": 31.1925, "grad_norm": 6.4375, "learning_rate": 1.6113063498447714e-05, "loss": 1.4911, "step": 12477 }, { "epoch": 31.195, "grad_norm": 6.15625, "learning_rate": 1.611242269548671e-05, "loss": 1.2876, "step": 12478 }, { "epoch": 31.1975, "grad_norm": 7.59375, "learning_rate": 1.6111781852452906e-05, "loss": 1.3737, "step": 12479 }, { "epoch": 31.2, "grad_norm": 9.9375, "learning_rate": 1.6111140969350504e-05, "loss": 1.3587, "step": 12480 }, { "epoch": 31.2025, "grad_norm": 7.875, "learning_rate": 1.6110500046183707e-05, "loss": 1.3676, "step": 12481 }, { "epoch": 31.205, "grad_norm": 12.375, "learning_rate": 1.6109859082956717e-05, "loss": 1.2763, "step": 12482 }, { "epoch": 31.2075, "grad_norm": 5.40625, "learning_rate": 1.6109218079673732e-05, "loss": 1.4391, "step": 12483 }, { "epoch": 31.21, "grad_norm": 9.625, "learning_rate": 1.610857703633896e-05, "loss": 1.3832, "step": 12484 }, { "epoch": 31.2125, "grad_norm": 9.1875, "learning_rate": 1.6107935952956604e-05, "loss": 1.2348, "step": 12485 }, { "epoch": 31.215, "grad_norm": 7.78125, "learning_rate": 1.6107294829530862e-05, "loss": 1.2451, "step": 12486 }, { "epoch": 31.2175, "grad_norm": 10.375, "learning_rate": 1.610665366606594e-05, "loss": 1.3608, "step": 12487 }, { "epoch": 31.22, "grad_norm": 10.0, "learning_rate": 1.6106012462566042e-05, "loss": 1.4009, "step": 12488 }, { "epoch": 31.2225, "grad_norm": 5.9375, "learning_rate": 1.6105371219035365e-05, "loss": 1.3476, "step": 12489 }, { "epoch": 31.225, "grad_norm": 5.625, "learning_rate": 1.6104729935478126e-05, "loss": 1.2859, "step": 12490 }, { "epoch": 31.2275, "grad_norm": 9.5, "learning_rate": 1.610408861189852e-05, "loss": 1.3122, "step": 12491 }, { "epoch": 31.23, "grad_norm": 5.5625, "learning_rate": 1.6103447248300754e-05, "loss": 1.3331, "step": 12492 }, { "epoch": 31.2325, "grad_norm": 8.3125, "learning_rate": 1.6102805844689026e-05, "loss": 1.3393, "step": 12493 }, { "epoch": 31.235, "grad_norm": 9.625, "learning_rate": 1.610216440106755e-05, "loss": 1.4622, "step": 12494 }, { "epoch": 31.2375, "grad_norm": 5.28125, "learning_rate": 1.610152291744053e-05, "loss": 1.2586, "step": 12495 }, { "epoch": 31.24, "grad_norm": 9.25, "learning_rate": 1.610088139381217e-05, "loss": 1.2757, "step": 12496 }, { "epoch": 31.2425, "grad_norm": 7.59375, "learning_rate": 1.6100239830186677e-05, "loss": 1.4584, "step": 12497 }, { "epoch": 31.245, "grad_norm": 6.75, "learning_rate": 1.6099598226568254e-05, "loss": 1.3884, "step": 12498 }, { "epoch": 31.2475, "grad_norm": 9.4375, "learning_rate": 1.6098956582961108e-05, "loss": 1.2936, "step": 12499 }, { "epoch": 31.25, "grad_norm": 5.78125, "learning_rate": 1.6098314899369446e-05, "loss": 1.5188, "step": 12500 }, { "epoch": 31.2525, "grad_norm": 8.625, "learning_rate": 1.609767317579748e-05, "loss": 1.4344, "step": 12501 }, { "epoch": 31.255, "grad_norm": 7.625, "learning_rate": 1.6097031412249407e-05, "loss": 1.2894, "step": 12502 }, { "epoch": 31.2575, "grad_norm": 8.0, "learning_rate": 1.6096389608729447e-05, "loss": 1.4987, "step": 12503 }, { "epoch": 31.26, "grad_norm": 7.9375, "learning_rate": 1.6095747765241797e-05, "loss": 1.3953, "step": 12504 }, { "epoch": 31.2625, "grad_norm": 4.9375, "learning_rate": 1.6095105881790668e-05, "loss": 1.5524, "step": 12505 }, { "epoch": 31.265, "grad_norm": 4.65625, "learning_rate": 1.609446395838027e-05, "loss": 1.32, "step": 12506 }, { "epoch": 31.2675, "grad_norm": 7.15625, "learning_rate": 1.609382199501481e-05, "loss": 1.3771, "step": 12507 }, { "epoch": 31.27, "grad_norm": 6.6875, "learning_rate": 1.6093179991698493e-05, "loss": 1.2536, "step": 12508 }, { "epoch": 31.2725, "grad_norm": 7.875, "learning_rate": 1.6092537948435537e-05, "loss": 1.3681, "step": 12509 }, { "epoch": 31.275, "grad_norm": 11.125, "learning_rate": 1.6091895865230143e-05, "loss": 1.2691, "step": 12510 }, { "epoch": 31.2775, "grad_norm": 7.78125, "learning_rate": 1.6091253742086525e-05, "loss": 1.4165, "step": 12511 }, { "epoch": 31.28, "grad_norm": 5.4375, "learning_rate": 1.609061157900889e-05, "loss": 1.4506, "step": 12512 }, { "epoch": 31.2825, "grad_norm": 7.3125, "learning_rate": 1.608996937600145e-05, "loss": 1.2565, "step": 12513 }, { "epoch": 31.285, "grad_norm": 6.5, "learning_rate": 1.6089327133068414e-05, "loss": 1.5791, "step": 12514 }, { "epoch": 31.2875, "grad_norm": 7.3125, "learning_rate": 1.6088684850213993e-05, "loss": 1.3297, "step": 12515 }, { "epoch": 31.29, "grad_norm": 10.75, "learning_rate": 1.60880425274424e-05, "loss": 1.4662, "step": 12516 }, { "epoch": 31.2925, "grad_norm": 10.625, "learning_rate": 1.608740016475784e-05, "loss": 1.4003, "step": 12517 }, { "epoch": 31.295, "grad_norm": 5.0, "learning_rate": 1.6086757762164534e-05, "loss": 1.2521, "step": 12518 }, { "epoch": 31.2975, "grad_norm": 7.84375, "learning_rate": 1.6086115319666683e-05, "loss": 1.4034, "step": 12519 }, { "epoch": 31.3, "grad_norm": 7.59375, "learning_rate": 1.6085472837268504e-05, "loss": 1.4281, "step": 12520 }, { "epoch": 31.3025, "grad_norm": 9.3125, "learning_rate": 1.608483031497421e-05, "loss": 1.3839, "step": 12521 }, { "epoch": 31.305, "grad_norm": 10.5, "learning_rate": 1.6084187752788008e-05, "loss": 1.3947, "step": 12522 }, { "epoch": 31.3075, "grad_norm": 5.28125, "learning_rate": 1.6083545150714116e-05, "loss": 1.3241, "step": 12523 }, { "epoch": 31.31, "grad_norm": 6.78125, "learning_rate": 1.608290250875675e-05, "loss": 1.5925, "step": 12524 }, { "epoch": 31.3125, "grad_norm": 6.53125, "learning_rate": 1.6082259826920114e-05, "loss": 1.2653, "step": 12525 }, { "epoch": 31.315, "grad_norm": 6.0, "learning_rate": 1.608161710520843e-05, "loss": 1.4435, "step": 12526 }, { "epoch": 31.3175, "grad_norm": 8.125, "learning_rate": 1.6080974343625904e-05, "loss": 1.5276, "step": 12527 }, { "epoch": 31.32, "grad_norm": 5.75, "learning_rate": 1.6080331542176754e-05, "loss": 1.3388, "step": 12528 }, { "epoch": 31.3225, "grad_norm": 6.5625, "learning_rate": 1.6079688700865196e-05, "loss": 1.4609, "step": 12529 }, { "epoch": 31.325, "grad_norm": 7.75, "learning_rate": 1.607904581969544e-05, "loss": 1.3855, "step": 12530 }, { "epoch": 31.3275, "grad_norm": 5.78125, "learning_rate": 1.6078402898671705e-05, "loss": 1.3666, "step": 12531 }, { "epoch": 31.33, "grad_norm": 6.53125, "learning_rate": 1.60777599377982e-05, "loss": 1.3482, "step": 12532 }, { "epoch": 31.3325, "grad_norm": 7.53125, "learning_rate": 1.6077116937079146e-05, "loss": 1.3475, "step": 12533 }, { "epoch": 31.335, "grad_norm": 6.625, "learning_rate": 1.6076473896518757e-05, "loss": 1.4349, "step": 12534 }, { "epoch": 31.3375, "grad_norm": 7.75, "learning_rate": 1.6075830816121247e-05, "loss": 1.2032, "step": 12535 }, { "epoch": 31.34, "grad_norm": 5.375, "learning_rate": 1.6075187695890837e-05, "loss": 1.4197, "step": 12536 }, { "epoch": 31.3425, "grad_norm": 9.75, "learning_rate": 1.6074544535831735e-05, "loss": 1.4381, "step": 12537 }, { "epoch": 31.345, "grad_norm": 11.9375, "learning_rate": 1.607390133594817e-05, "loss": 1.3102, "step": 12538 }, { "epoch": 31.3475, "grad_norm": 6.03125, "learning_rate": 1.607325809624434e-05, "loss": 1.3714, "step": 12539 }, { "epoch": 31.35, "grad_norm": 7.875, "learning_rate": 1.607261481672448e-05, "loss": 1.4095, "step": 12540 }, { "epoch": 31.3525, "grad_norm": 4.40625, "learning_rate": 1.60719714973928e-05, "loss": 1.3908, "step": 12541 }, { "epoch": 31.355, "grad_norm": 6.09375, "learning_rate": 1.6071328138253514e-05, "loss": 1.2814, "step": 12542 }, { "epoch": 31.3575, "grad_norm": 6.09375, "learning_rate": 1.6070684739310846e-05, "loss": 1.4563, "step": 12543 }, { "epoch": 31.36, "grad_norm": 6.375, "learning_rate": 1.6070041300569014e-05, "loss": 1.2601, "step": 12544 }, { "epoch": 31.3625, "grad_norm": 6.53125, "learning_rate": 1.606939782203223e-05, "loss": 1.4494, "step": 12545 }, { "epoch": 31.365, "grad_norm": 8.0, "learning_rate": 1.606875430370472e-05, "loss": 1.3322, "step": 12546 }, { "epoch": 31.3675, "grad_norm": 7.03125, "learning_rate": 1.6068110745590696e-05, "loss": 1.4182, "step": 12547 }, { "epoch": 31.37, "grad_norm": 7.6875, "learning_rate": 1.6067467147694382e-05, "loss": 1.1511, "step": 12548 }, { "epoch": 31.3725, "grad_norm": 9.75, "learning_rate": 1.6066823510019998e-05, "loss": 1.4449, "step": 12549 }, { "epoch": 31.375, "grad_norm": 7.40625, "learning_rate": 1.6066179832571762e-05, "loss": 1.3961, "step": 12550 }, { "epoch": 31.3775, "grad_norm": 5.625, "learning_rate": 1.6065536115353892e-05, "loss": 1.2475, "step": 12551 }, { "epoch": 31.38, "grad_norm": 8.375, "learning_rate": 1.6064892358370608e-05, "loss": 1.3449, "step": 12552 }, { "epoch": 31.3825, "grad_norm": 13.75, "learning_rate": 1.606424856162614e-05, "loss": 1.25, "step": 12553 }, { "epoch": 31.385, "grad_norm": 9.375, "learning_rate": 1.6063604725124697e-05, "loss": 1.3057, "step": 12554 }, { "epoch": 31.3875, "grad_norm": 6.0, "learning_rate": 1.6062960848870503e-05, "loss": 1.2237, "step": 12555 }, { "epoch": 31.39, "grad_norm": 6.5625, "learning_rate": 1.6062316932867777e-05, "loss": 1.3692, "step": 12556 }, { "epoch": 31.3925, "grad_norm": 6.78125, "learning_rate": 1.6061672977120748e-05, "loss": 1.2534, "step": 12557 }, { "epoch": 31.395, "grad_norm": 5.28125, "learning_rate": 1.6061028981633632e-05, "loss": 1.2041, "step": 12558 }, { "epoch": 31.3975, "grad_norm": 11.625, "learning_rate": 1.6060384946410657e-05, "loss": 1.4764, "step": 12559 }, { "epoch": 31.4, "grad_norm": 7.65625, "learning_rate": 1.6059740871456035e-05, "loss": 1.2466, "step": 12560 }, { "epoch": 31.4025, "grad_norm": 5.34375, "learning_rate": 1.6059096756774e-05, "loss": 1.3052, "step": 12561 }, { "epoch": 31.405, "grad_norm": 5.6875, "learning_rate": 1.6058452602368762e-05, "loss": 1.4078, "step": 12562 }, { "epoch": 31.4075, "grad_norm": 7.4375, "learning_rate": 1.6057808408244557e-05, "loss": 1.4433, "step": 12563 }, { "epoch": 31.41, "grad_norm": 8.125, "learning_rate": 1.6057164174405603e-05, "loss": 1.3412, "step": 12564 }, { "epoch": 31.4125, "grad_norm": 9.4375, "learning_rate": 1.6056519900856117e-05, "loss": 1.3169, "step": 12565 }, { "epoch": 31.415, "grad_norm": 9.5, "learning_rate": 1.6055875587600333e-05, "loss": 1.3728, "step": 12566 }, { "epoch": 31.4175, "grad_norm": 8.4375, "learning_rate": 1.605523123464247e-05, "loss": 1.5932, "step": 12567 }, { "epoch": 31.42, "grad_norm": 8.4375, "learning_rate": 1.6054586841986754e-05, "loss": 1.3888, "step": 12568 }, { "epoch": 31.4225, "grad_norm": 7.09375, "learning_rate": 1.6053942409637412e-05, "loss": 1.3438, "step": 12569 }, { "epoch": 31.425, "grad_norm": 7.09375, "learning_rate": 1.6053297937598658e-05, "loss": 1.5193, "step": 12570 }, { "epoch": 31.4275, "grad_norm": 7.59375, "learning_rate": 1.6052653425874733e-05, "loss": 1.2002, "step": 12571 }, { "epoch": 31.43, "grad_norm": 6.65625, "learning_rate": 1.605200887446985e-05, "loss": 1.3478, "step": 12572 }, { "epoch": 31.4325, "grad_norm": 8.625, "learning_rate": 1.605136428338824e-05, "loss": 1.5104, "step": 12573 }, { "epoch": 31.435, "grad_norm": 9.1875, "learning_rate": 1.605071965263413e-05, "loss": 1.3168, "step": 12574 }, { "epoch": 31.4375, "grad_norm": 10.25, "learning_rate": 1.6050074982211738e-05, "loss": 1.4088, "step": 12575 }, { "epoch": 31.44, "grad_norm": 5.96875, "learning_rate": 1.60494302721253e-05, "loss": 1.3873, "step": 12576 }, { "epoch": 31.4425, "grad_norm": 9.75, "learning_rate": 1.604878552237904e-05, "loss": 1.3977, "step": 12577 }, { "epoch": 31.445, "grad_norm": 6.84375, "learning_rate": 1.6048140732977185e-05, "loss": 1.5371, "step": 12578 }, { "epoch": 31.4475, "grad_norm": 12.0625, "learning_rate": 1.604749590392396e-05, "loss": 1.3435, "step": 12579 }, { "epoch": 31.45, "grad_norm": 8.1875, "learning_rate": 1.6046851035223594e-05, "loss": 1.377, "step": 12580 }, { "epoch": 31.4525, "grad_norm": 6.25, "learning_rate": 1.6046206126880316e-05, "loss": 1.3556, "step": 12581 }, { "epoch": 31.455, "grad_norm": 9.4375, "learning_rate": 1.604556117889835e-05, "loss": 1.4405, "step": 12582 }, { "epoch": 31.4575, "grad_norm": 6.53125, "learning_rate": 1.6044916191281927e-05, "loss": 1.3612, "step": 12583 }, { "epoch": 31.46, "grad_norm": 7.53125, "learning_rate": 1.6044271164035278e-05, "loss": 1.2851, "step": 12584 }, { "epoch": 31.4625, "grad_norm": 7.71875, "learning_rate": 1.604362609716263e-05, "loss": 1.437, "step": 12585 }, { "epoch": 31.465, "grad_norm": 8.6875, "learning_rate": 1.6042980990668208e-05, "loss": 1.3806, "step": 12586 }, { "epoch": 31.4675, "grad_norm": 5.9375, "learning_rate": 1.6042335844556243e-05, "loss": 1.2769, "step": 12587 }, { "epoch": 31.47, "grad_norm": 5.71875, "learning_rate": 1.604169065883097e-05, "loss": 1.3836, "step": 12588 }, { "epoch": 31.4725, "grad_norm": 7.28125, "learning_rate": 1.6041045433496613e-05, "loss": 1.1773, "step": 12589 }, { "epoch": 31.475, "grad_norm": 4.21875, "learning_rate": 1.6040400168557405e-05, "loss": 1.3588, "step": 12590 }, { "epoch": 31.4775, "grad_norm": 8.0, "learning_rate": 1.6039754864017573e-05, "loss": 1.3385, "step": 12591 }, { "epoch": 31.48, "grad_norm": 8.375, "learning_rate": 1.603910951988135e-05, "loss": 1.4193, "step": 12592 }, { "epoch": 31.4825, "grad_norm": 8.5625, "learning_rate": 1.6038464136152968e-05, "loss": 1.241, "step": 12593 }, { "epoch": 31.485, "grad_norm": 8.1875, "learning_rate": 1.6037818712836657e-05, "loss": 1.2476, "step": 12594 }, { "epoch": 31.4875, "grad_norm": 5.46875, "learning_rate": 1.6037173249936647e-05, "loss": 1.429, "step": 12595 }, { "epoch": 31.49, "grad_norm": 7.34375, "learning_rate": 1.6036527747457172e-05, "loss": 1.3021, "step": 12596 }, { "epoch": 31.4925, "grad_norm": 7.3125, "learning_rate": 1.603588220540246e-05, "loss": 1.3919, "step": 12597 }, { "epoch": 31.495, "grad_norm": 9.0, "learning_rate": 1.603523662377675e-05, "loss": 1.4303, "step": 12598 }, { "epoch": 31.4975, "grad_norm": 9.0, "learning_rate": 1.6034591002584266e-05, "loss": 1.4977, "step": 12599 }, { "epoch": 31.5, "grad_norm": 9.0625, "learning_rate": 1.603394534182925e-05, "loss": 1.2271, "step": 12600 }, { "epoch": 31.5025, "grad_norm": 8.1875, "learning_rate": 1.6033299641515925e-05, "loss": 1.487, "step": 12601 }, { "epoch": 31.505, "grad_norm": 8.1875, "learning_rate": 1.603265390164853e-05, "loss": 1.3893, "step": 12602 }, { "epoch": 31.5075, "grad_norm": 5.84375, "learning_rate": 1.60320081222313e-05, "loss": 1.2825, "step": 12603 }, { "epoch": 31.51, "grad_norm": 6.625, "learning_rate": 1.6031362303268464e-05, "loss": 1.3789, "step": 12604 }, { "epoch": 31.5125, "grad_norm": 8.4375, "learning_rate": 1.6030716444764257e-05, "loss": 1.3335, "step": 12605 }, { "epoch": 31.515, "grad_norm": 5.40625, "learning_rate": 1.6030070546722917e-05, "loss": 1.3234, "step": 12606 }, { "epoch": 31.5175, "grad_norm": 8.5625, "learning_rate": 1.6029424609148675e-05, "loss": 1.3238, "step": 12607 }, { "epoch": 31.52, "grad_norm": 6.9375, "learning_rate": 1.602877863204576e-05, "loss": 1.3269, "step": 12608 }, { "epoch": 31.5225, "grad_norm": 8.25, "learning_rate": 1.6028132615418423e-05, "loss": 1.349, "step": 12609 }, { "epoch": 31.525, "grad_norm": 7.03125, "learning_rate": 1.6027486559270885e-05, "loss": 1.1765, "step": 12610 }, { "epoch": 31.5275, "grad_norm": 7.78125, "learning_rate": 1.602684046360739e-05, "loss": 1.4348, "step": 12611 }, { "epoch": 31.53, "grad_norm": 8.3125, "learning_rate": 1.6026194328432166e-05, "loss": 1.3413, "step": 12612 }, { "epoch": 31.5325, "grad_norm": 6.09375, "learning_rate": 1.6025548153749455e-05, "loss": 1.263, "step": 12613 }, { "epoch": 31.535, "grad_norm": 6.6875, "learning_rate": 1.6024901939563493e-05, "loss": 1.4609, "step": 12614 }, { "epoch": 31.5375, "grad_norm": 10.75, "learning_rate": 1.6024255685878514e-05, "loss": 1.1664, "step": 12615 }, { "epoch": 31.54, "grad_norm": 7.84375, "learning_rate": 1.6023609392698753e-05, "loss": 1.4851, "step": 12616 }, { "epoch": 31.5425, "grad_norm": 6.125, "learning_rate": 1.6022963060028454e-05, "loss": 1.2502, "step": 12617 }, { "epoch": 31.545, "grad_norm": 12.0, "learning_rate": 1.6022316687871846e-05, "loss": 1.424, "step": 12618 }, { "epoch": 31.5475, "grad_norm": 6.375, "learning_rate": 1.6021670276233176e-05, "loss": 1.3731, "step": 12619 }, { "epoch": 31.55, "grad_norm": 8.9375, "learning_rate": 1.6021023825116672e-05, "loss": 1.5457, "step": 12620 }, { "epoch": 31.5525, "grad_norm": 5.84375, "learning_rate": 1.602037733452658e-05, "loss": 1.4283, "step": 12621 }, { "epoch": 31.555, "grad_norm": 10.0, "learning_rate": 1.601973080446713e-05, "loss": 1.6133, "step": 12622 }, { "epoch": 31.5575, "grad_norm": 7.71875, "learning_rate": 1.601908423494257e-05, "loss": 1.4272, "step": 12623 }, { "epoch": 31.56, "grad_norm": 9.125, "learning_rate": 1.6018437625957135e-05, "loss": 1.3778, "step": 12624 }, { "epoch": 31.5625, "grad_norm": 5.84375, "learning_rate": 1.6017790977515063e-05, "loss": 1.3662, "step": 12625 }, { "epoch": 31.565, "grad_norm": 6.59375, "learning_rate": 1.6017144289620592e-05, "loss": 1.2878, "step": 12626 }, { "epoch": 31.5675, "grad_norm": 8.6875, "learning_rate": 1.6016497562277967e-05, "loss": 1.4945, "step": 12627 }, { "epoch": 31.57, "grad_norm": 8.5625, "learning_rate": 1.601585079549142e-05, "loss": 1.2863, "step": 12628 }, { "epoch": 31.5725, "grad_norm": 5.25, "learning_rate": 1.6015203989265202e-05, "loss": 1.3629, "step": 12629 }, { "epoch": 31.575, "grad_norm": 5.65625, "learning_rate": 1.6014557143603543e-05, "loss": 1.2659, "step": 12630 }, { "epoch": 31.5775, "grad_norm": 6.25, "learning_rate": 1.601391025851069e-05, "loss": 1.2767, "step": 12631 }, { "epoch": 31.58, "grad_norm": 7.40625, "learning_rate": 1.601326333399088e-05, "loss": 1.2891, "step": 12632 }, { "epoch": 31.5825, "grad_norm": 8.1875, "learning_rate": 1.6012616370048363e-05, "loss": 1.2901, "step": 12633 }, { "epoch": 31.585, "grad_norm": 9.125, "learning_rate": 1.6011969366687367e-05, "loss": 1.3779, "step": 12634 }, { "epoch": 31.5875, "grad_norm": 8.875, "learning_rate": 1.6011322323912142e-05, "loss": 1.4481, "step": 12635 }, { "epoch": 31.59, "grad_norm": 6.28125, "learning_rate": 1.601067524172693e-05, "loss": 1.3689, "step": 12636 }, { "epoch": 31.5925, "grad_norm": 7.09375, "learning_rate": 1.6010028120135968e-05, "loss": 1.2354, "step": 12637 }, { "epoch": 31.595, "grad_norm": 10.0, "learning_rate": 1.6009380959143506e-05, "loss": 1.3299, "step": 12638 }, { "epoch": 31.5975, "grad_norm": 6.71875, "learning_rate": 1.600873375875378e-05, "loss": 1.3456, "step": 12639 }, { "epoch": 31.6, "grad_norm": 7.6875, "learning_rate": 1.6008086518971037e-05, "loss": 1.3518, "step": 12640 }, { "epoch": 31.6025, "grad_norm": 5.375, "learning_rate": 1.6007439239799522e-05, "loss": 1.3885, "step": 12641 }, { "epoch": 31.605, "grad_norm": 6.84375, "learning_rate": 1.600679192124347e-05, "loss": 1.362, "step": 12642 }, { "epoch": 31.6075, "grad_norm": 6.8125, "learning_rate": 1.6006144563307135e-05, "loss": 1.3162, "step": 12643 }, { "epoch": 31.61, "grad_norm": 9.0625, "learning_rate": 1.6005497165994754e-05, "loss": 1.4441, "step": 12644 }, { "epoch": 31.6125, "grad_norm": 8.125, "learning_rate": 1.6004849729310574e-05, "loss": 1.2952, "step": 12645 }, { "epoch": 31.615, "grad_norm": 12.5625, "learning_rate": 1.6004202253258844e-05, "loss": 1.3397, "step": 12646 }, { "epoch": 31.6175, "grad_norm": 9.0, "learning_rate": 1.6003554737843798e-05, "loss": 1.3585, "step": 12647 }, { "epoch": 31.62, "grad_norm": 5.75, "learning_rate": 1.6002907183069693e-05, "loss": 1.2712, "step": 12648 }, { "epoch": 31.6225, "grad_norm": 10.6875, "learning_rate": 1.6002259588940764e-05, "loss": 1.1804, "step": 12649 }, { "epoch": 31.625, "grad_norm": 7.4375, "learning_rate": 1.6001611955461265e-05, "loss": 1.3413, "step": 12650 }, { "epoch": 31.6275, "grad_norm": 17.0, "learning_rate": 1.6000964282635436e-05, "loss": 1.4965, "step": 12651 }, { "epoch": 31.63, "grad_norm": 10.875, "learning_rate": 1.6000316570467525e-05, "loss": 1.333, "step": 12652 }, { "epoch": 31.6325, "grad_norm": 5.90625, "learning_rate": 1.599966881896178e-05, "loss": 1.3528, "step": 12653 }, { "epoch": 31.635, "grad_norm": 6.8125, "learning_rate": 1.5999021028122447e-05, "loss": 1.4583, "step": 12654 }, { "epoch": 31.6375, "grad_norm": 6.03125, "learning_rate": 1.599837319795377e-05, "loss": 1.3559, "step": 12655 }, { "epoch": 31.64, "grad_norm": 17.875, "learning_rate": 1.599772532846e-05, "loss": 1.3215, "step": 12656 }, { "epoch": 31.6425, "grad_norm": 8.0, "learning_rate": 1.599707741964538e-05, "loss": 1.3547, "step": 12657 }, { "epoch": 31.645, "grad_norm": 6.59375, "learning_rate": 1.5996429471514162e-05, "loss": 1.2696, "step": 12658 }, { "epoch": 31.6475, "grad_norm": 6.875, "learning_rate": 1.5995781484070593e-05, "loss": 1.3597, "step": 12659 }, { "epoch": 31.65, "grad_norm": 6.28125, "learning_rate": 1.599513345731892e-05, "loss": 1.3212, "step": 12660 }, { "epoch": 31.6525, "grad_norm": 6.78125, "learning_rate": 1.5994485391263393e-05, "loss": 1.2468, "step": 12661 }, { "epoch": 31.655, "grad_norm": 11.25, "learning_rate": 1.599383728590826e-05, "loss": 1.4204, "step": 12662 }, { "epoch": 31.6575, "grad_norm": 7.1875, "learning_rate": 1.5993189141257768e-05, "loss": 1.3765, "step": 12663 }, { "epoch": 31.66, "grad_norm": 7.40625, "learning_rate": 1.599254095731617e-05, "loss": 1.3345, "step": 12664 }, { "epoch": 31.6625, "grad_norm": 11.625, "learning_rate": 1.5991892734087714e-05, "loss": 1.5506, "step": 12665 }, { "epoch": 31.665, "grad_norm": 6.78125, "learning_rate": 1.5991244471576646e-05, "loss": 1.4592, "step": 12666 }, { "epoch": 31.6675, "grad_norm": 14.0, "learning_rate": 1.5990596169787217e-05, "loss": 1.3726, "step": 12667 }, { "epoch": 31.67, "grad_norm": 5.46875, "learning_rate": 1.5989947828723682e-05, "loss": 1.1946, "step": 12668 }, { "epoch": 31.6725, "grad_norm": 8.3125, "learning_rate": 1.598929944839029e-05, "loss": 1.4154, "step": 12669 }, { "epoch": 31.675, "grad_norm": 6.375, "learning_rate": 1.5988651028791288e-05, "loss": 1.4716, "step": 12670 }, { "epoch": 31.6775, "grad_norm": 8.9375, "learning_rate": 1.5988002569930935e-05, "loss": 1.3805, "step": 12671 }, { "epoch": 31.68, "grad_norm": 7.21875, "learning_rate": 1.598735407181347e-05, "loss": 1.3841, "step": 12672 }, { "epoch": 31.6825, "grad_norm": 10.0625, "learning_rate": 1.5986705534443154e-05, "loss": 1.3161, "step": 12673 }, { "epoch": 31.685, "grad_norm": 9.5625, "learning_rate": 1.5986056957824236e-05, "loss": 1.3079, "step": 12674 }, { "epoch": 31.6875, "grad_norm": 5.15625, "learning_rate": 1.598540834196097e-05, "loss": 1.2853, "step": 12675 }, { "epoch": 31.69, "grad_norm": 6.0, "learning_rate": 1.5984759686857602e-05, "loss": 1.3392, "step": 12676 }, { "epoch": 31.6925, "grad_norm": 8.375, "learning_rate": 1.5984110992518394e-05, "loss": 1.3981, "step": 12677 }, { "epoch": 31.695, "grad_norm": 11.875, "learning_rate": 1.5983462258947587e-05, "loss": 1.4299, "step": 12678 }, { "epoch": 31.6975, "grad_norm": 6.0, "learning_rate": 1.5982813486149445e-05, "loss": 1.2233, "step": 12679 }, { "epoch": 31.7, "grad_norm": 7.25, "learning_rate": 1.598216467412822e-05, "loss": 1.463, "step": 12680 }, { "epoch": 31.7025, "grad_norm": 8.0, "learning_rate": 1.5981515822888162e-05, "loss": 1.3103, "step": 12681 }, { "epoch": 31.705, "grad_norm": 6.34375, "learning_rate": 1.598086693243352e-05, "loss": 1.244, "step": 12682 }, { "epoch": 31.7075, "grad_norm": 7.28125, "learning_rate": 1.598021800276856e-05, "loss": 1.3679, "step": 12683 }, { "epoch": 31.71, "grad_norm": 7.5, "learning_rate": 1.5979569033897527e-05, "loss": 1.4431, "step": 12684 }, { "epoch": 31.7125, "grad_norm": 8.5625, "learning_rate": 1.597892002582468e-05, "loss": 1.4027, "step": 12685 }, { "epoch": 31.715, "grad_norm": 5.84375, "learning_rate": 1.597827097855427e-05, "loss": 1.3694, "step": 12686 }, { "epoch": 31.7175, "grad_norm": 9.0625, "learning_rate": 1.597762189209056e-05, "loss": 1.5276, "step": 12687 }, { "epoch": 31.72, "grad_norm": 10.3125, "learning_rate": 1.5976972766437796e-05, "loss": 1.4197, "step": 12688 }, { "epoch": 31.7225, "grad_norm": 10.5, "learning_rate": 1.5976323601600236e-05, "loss": 1.3971, "step": 12689 }, { "epoch": 31.725, "grad_norm": 6.53125, "learning_rate": 1.597567439758214e-05, "loss": 1.4712, "step": 12690 }, { "epoch": 31.7275, "grad_norm": 7.5625, "learning_rate": 1.5975025154387764e-05, "loss": 1.304, "step": 12691 }, { "epoch": 31.73, "grad_norm": 7.75, "learning_rate": 1.597437587202136e-05, "loss": 1.4464, "step": 12692 }, { "epoch": 31.7325, "grad_norm": 6.84375, "learning_rate": 1.5973726550487187e-05, "loss": 1.388, "step": 12693 }, { "epoch": 31.735, "grad_norm": 8.875, "learning_rate": 1.59730771897895e-05, "loss": 1.3681, "step": 12694 }, { "epoch": 31.7375, "grad_norm": 7.0625, "learning_rate": 1.597242778993256e-05, "loss": 1.3347, "step": 12695 }, { "epoch": 31.74, "grad_norm": 7.9375, "learning_rate": 1.5971778350920622e-05, "loss": 1.3169, "step": 12696 }, { "epoch": 31.7425, "grad_norm": 6.25, "learning_rate": 1.5971128872757945e-05, "loss": 1.3783, "step": 12697 }, { "epoch": 31.745, "grad_norm": 5.59375, "learning_rate": 1.5970479355448786e-05, "loss": 1.5315, "step": 12698 }, { "epoch": 31.7475, "grad_norm": 8.875, "learning_rate": 1.5969829798997404e-05, "loss": 1.1996, "step": 12699 }, { "epoch": 31.75, "grad_norm": 6.65625, "learning_rate": 1.5969180203408052e-05, "loss": 1.3038, "step": 12700 }, { "epoch": 31.7525, "grad_norm": 9.0625, "learning_rate": 1.5968530568685e-05, "loss": 1.4715, "step": 12701 }, { "epoch": 31.755, "grad_norm": 6.375, "learning_rate": 1.5967880894832496e-05, "loss": 1.2734, "step": 12702 }, { "epoch": 31.7575, "grad_norm": 12.625, "learning_rate": 1.5967231181854806e-05, "loss": 1.3929, "step": 12703 }, { "epoch": 31.76, "grad_norm": 6.0, "learning_rate": 1.596658142975618e-05, "loss": 1.4008, "step": 12704 }, { "epoch": 31.7625, "grad_norm": 7.03125, "learning_rate": 1.5965931638540894e-05, "loss": 1.2852, "step": 12705 }, { "epoch": 31.765, "grad_norm": 10.0, "learning_rate": 1.59652818082132e-05, "loss": 1.3113, "step": 12706 }, { "epoch": 31.7675, "grad_norm": 7.21875, "learning_rate": 1.596463193877735e-05, "loss": 1.2629, "step": 12707 }, { "epoch": 31.77, "grad_norm": 7.59375, "learning_rate": 1.596398203023762e-05, "loss": 1.3341, "step": 12708 }, { "epoch": 31.7725, "grad_norm": 9.6875, "learning_rate": 1.596333208259825e-05, "loss": 1.3027, "step": 12709 }, { "epoch": 31.775, "grad_norm": 12.5, "learning_rate": 1.5962682095863523e-05, "loss": 1.3657, "step": 12710 }, { "epoch": 31.7775, "grad_norm": 6.03125, "learning_rate": 1.596203207003769e-05, "loss": 1.3076, "step": 12711 }, { "epoch": 31.78, "grad_norm": 7.78125, "learning_rate": 1.596138200512501e-05, "loss": 1.4858, "step": 12712 }, { "epoch": 31.7825, "grad_norm": 9.75, "learning_rate": 1.596073190112975e-05, "loss": 1.357, "step": 12713 }, { "epoch": 31.785, "grad_norm": 5.15625, "learning_rate": 1.5960081758056173e-05, "loss": 1.2936, "step": 12714 }, { "epoch": 31.7875, "grad_norm": 8.5, "learning_rate": 1.5959431575908533e-05, "loss": 1.4222, "step": 12715 }, { "epoch": 31.79, "grad_norm": 7.0625, "learning_rate": 1.59587813546911e-05, "loss": 1.3459, "step": 12716 }, { "epoch": 31.7925, "grad_norm": 5.8125, "learning_rate": 1.5958131094408134e-05, "loss": 1.2559, "step": 12717 }, { "epoch": 31.795, "grad_norm": 6.34375, "learning_rate": 1.5957480795063904e-05, "loss": 1.4351, "step": 12718 }, { "epoch": 31.7975, "grad_norm": 5.90625, "learning_rate": 1.5956830456662663e-05, "loss": 1.5401, "step": 12719 }, { "epoch": 31.8, "grad_norm": 4.96875, "learning_rate": 1.5956180079208684e-05, "loss": 1.3382, "step": 12720 }, { "epoch": 31.8025, "grad_norm": 9.3125, "learning_rate": 1.5955529662706224e-05, "loss": 1.4239, "step": 12721 }, { "epoch": 31.805, "grad_norm": 9.3125, "learning_rate": 1.595487920715955e-05, "loss": 1.3119, "step": 12722 }, { "epoch": 31.8075, "grad_norm": 7.75, "learning_rate": 1.595422871257293e-05, "loss": 1.5106, "step": 12723 }, { "epoch": 31.81, "grad_norm": 6.15625, "learning_rate": 1.5953578178950617e-05, "loss": 1.3345, "step": 12724 }, { "epoch": 31.8125, "grad_norm": 7.1875, "learning_rate": 1.5952927606296888e-05, "loss": 1.4215, "step": 12725 }, { "epoch": 31.815, "grad_norm": 8.625, "learning_rate": 1.5952276994616008e-05, "loss": 1.4337, "step": 12726 }, { "epoch": 31.8175, "grad_norm": 6.125, "learning_rate": 1.5951626343912232e-05, "loss": 1.4784, "step": 12727 }, { "epoch": 31.82, "grad_norm": 6.625, "learning_rate": 1.5950975654189832e-05, "loss": 1.3389, "step": 12728 }, { "epoch": 31.8225, "grad_norm": 7.09375, "learning_rate": 1.5950324925453078e-05, "loss": 1.3282, "step": 12729 }, { "epoch": 31.825, "grad_norm": 8.0625, "learning_rate": 1.5949674157706228e-05, "loss": 1.3251, "step": 12730 }, { "epoch": 31.8275, "grad_norm": 8.75, "learning_rate": 1.5949023350953555e-05, "loss": 1.5383, "step": 12731 }, { "epoch": 31.83, "grad_norm": 5.90625, "learning_rate": 1.5948372505199325e-05, "loss": 1.258, "step": 12732 }, { "epoch": 31.8325, "grad_norm": 7.0625, "learning_rate": 1.59477216204478e-05, "loss": 1.2674, "step": 12733 }, { "epoch": 31.835, "grad_norm": 7.0625, "learning_rate": 1.5947070696703247e-05, "loss": 1.3491, "step": 12734 }, { "epoch": 31.8375, "grad_norm": 8.375, "learning_rate": 1.594641973396994e-05, "loss": 1.2883, "step": 12735 }, { "epoch": 31.84, "grad_norm": 7.65625, "learning_rate": 1.5945768732252144e-05, "loss": 1.4605, "step": 12736 }, { "epoch": 31.8425, "grad_norm": 10.0, "learning_rate": 1.5945117691554125e-05, "loss": 1.4156, "step": 12737 }, { "epoch": 31.845, "grad_norm": 8.3125, "learning_rate": 1.594446661188015e-05, "loss": 1.3343, "step": 12738 }, { "epoch": 31.8475, "grad_norm": 8.3125, "learning_rate": 1.5943815493234494e-05, "loss": 1.4118, "step": 12739 }, { "epoch": 31.85, "grad_norm": 8.25, "learning_rate": 1.5943164335621418e-05, "loss": 1.3496, "step": 12740 }, { "epoch": 31.8525, "grad_norm": 6.59375, "learning_rate": 1.5942513139045196e-05, "loss": 1.4001, "step": 12741 }, { "epoch": 31.855, "grad_norm": 8.125, "learning_rate": 1.5941861903510096e-05, "loss": 1.368, "step": 12742 }, { "epoch": 31.8575, "grad_norm": 7.4375, "learning_rate": 1.594121062902039e-05, "loss": 1.2566, "step": 12743 }, { "epoch": 31.86, "grad_norm": 6.90625, "learning_rate": 1.5940559315580338e-05, "loss": 1.3657, "step": 12744 }, { "epoch": 31.8625, "grad_norm": 6.75, "learning_rate": 1.593990796319422e-05, "loss": 1.3967, "step": 12745 }, { "epoch": 31.865, "grad_norm": 6.84375, "learning_rate": 1.5939256571866306e-05, "loss": 1.4565, "step": 12746 }, { "epoch": 31.8675, "grad_norm": 5.59375, "learning_rate": 1.593860514160086e-05, "loss": 1.3066, "step": 12747 }, { "epoch": 31.87, "grad_norm": 7.4375, "learning_rate": 1.5937953672402157e-05, "loss": 1.3059, "step": 12748 }, { "epoch": 31.8725, "grad_norm": 5.84375, "learning_rate": 1.5937302164274467e-05, "loss": 1.4863, "step": 12749 }, { "epoch": 31.875, "grad_norm": 8.8125, "learning_rate": 1.5936650617222063e-05, "loss": 1.4986, "step": 12750 }, { "epoch": 31.8775, "grad_norm": 7.6875, "learning_rate": 1.5935999031249216e-05, "loss": 1.3128, "step": 12751 }, { "epoch": 31.88, "grad_norm": 8.3125, "learning_rate": 1.5935347406360192e-05, "loss": 1.3636, "step": 12752 }, { "epoch": 31.8825, "grad_norm": 7.03125, "learning_rate": 1.5934695742559275e-05, "loss": 1.3646, "step": 12753 }, { "epoch": 31.885, "grad_norm": 7.8125, "learning_rate": 1.5934044039850726e-05, "loss": 1.5339, "step": 12754 }, { "epoch": 31.8875, "grad_norm": 7.0625, "learning_rate": 1.593339229823882e-05, "loss": 1.3595, "step": 12755 }, { "epoch": 31.89, "grad_norm": 6.9375, "learning_rate": 1.5932740517727835e-05, "loss": 1.4332, "step": 12756 }, { "epoch": 31.8925, "grad_norm": 7.375, "learning_rate": 1.593208869832204e-05, "loss": 1.3956, "step": 12757 }, { "epoch": 31.895, "grad_norm": 9.25, "learning_rate": 1.5931436840025708e-05, "loss": 1.3035, "step": 12758 }, { "epoch": 31.8975, "grad_norm": 5.5, "learning_rate": 1.593078494284311e-05, "loss": 1.4693, "step": 12759 }, { "epoch": 31.9, "grad_norm": 5.84375, "learning_rate": 1.593013300677853e-05, "loss": 1.2816, "step": 12760 }, { "epoch": 31.9025, "grad_norm": 7.5625, "learning_rate": 1.5929481031836233e-05, "loss": 1.3134, "step": 12761 }, { "epoch": 31.905, "grad_norm": 6.4375, "learning_rate": 1.5928829018020496e-05, "loss": 1.4402, "step": 12762 }, { "epoch": 31.9075, "grad_norm": 9.3125, "learning_rate": 1.592817696533559e-05, "loss": 1.3748, "step": 12763 }, { "epoch": 31.91, "grad_norm": 8.875, "learning_rate": 1.5927524873785797e-05, "loss": 1.4766, "step": 12764 }, { "epoch": 31.9125, "grad_norm": 6.5625, "learning_rate": 1.5926872743375388e-05, "loss": 1.4762, "step": 12765 }, { "epoch": 31.915, "grad_norm": 7.28125, "learning_rate": 1.5926220574108637e-05, "loss": 1.4738, "step": 12766 }, { "epoch": 31.9175, "grad_norm": 7.59375, "learning_rate": 1.592556836598982e-05, "loss": 1.4806, "step": 12767 }, { "epoch": 31.92, "grad_norm": 8.9375, "learning_rate": 1.5924916119023214e-05, "loss": 1.4983, "step": 12768 }, { "epoch": 31.9225, "grad_norm": 6.28125, "learning_rate": 1.5924263833213095e-05, "loss": 1.3924, "step": 12769 }, { "epoch": 31.925, "grad_norm": 6.78125, "learning_rate": 1.592361150856374e-05, "loss": 1.221, "step": 12770 }, { "epoch": 31.9275, "grad_norm": 6.28125, "learning_rate": 1.5922959145079425e-05, "loss": 1.2656, "step": 12771 }, { "epoch": 31.93, "grad_norm": 8.0625, "learning_rate": 1.5922306742764426e-05, "loss": 1.4176, "step": 12772 }, { "epoch": 31.9325, "grad_norm": 8.75, "learning_rate": 1.5921654301623023e-05, "loss": 1.342, "step": 12773 }, { "epoch": 31.935, "grad_norm": 6.1875, "learning_rate": 1.592100182165949e-05, "loss": 1.227, "step": 12774 }, { "epoch": 31.9375, "grad_norm": 12.0625, "learning_rate": 1.592034930287811e-05, "loss": 1.3702, "step": 12775 }, { "epoch": 31.94, "grad_norm": 6.375, "learning_rate": 1.5919696745283154e-05, "loss": 1.3848, "step": 12776 }, { "epoch": 31.9425, "grad_norm": 6.96875, "learning_rate": 1.59190441488789e-05, "loss": 1.3505, "step": 12777 }, { "epoch": 31.945, "grad_norm": 7.65625, "learning_rate": 1.5918391513669635e-05, "loss": 1.2655, "step": 12778 }, { "epoch": 31.9475, "grad_norm": 7.3125, "learning_rate": 1.5917738839659623e-05, "loss": 1.5151, "step": 12779 }, { "epoch": 31.95, "grad_norm": 10.5, "learning_rate": 1.591708612685316e-05, "loss": 1.4139, "step": 12780 }, { "epoch": 31.9525, "grad_norm": 6.53125, "learning_rate": 1.5916433375254514e-05, "loss": 1.5127, "step": 12781 }, { "epoch": 31.955, "grad_norm": 7.28125, "learning_rate": 1.591578058486797e-05, "loss": 1.2602, "step": 12782 }, { "epoch": 31.9575, "grad_norm": 8.625, "learning_rate": 1.5915127755697805e-05, "loss": 1.3051, "step": 12783 }, { "epoch": 31.96, "grad_norm": 8.5625, "learning_rate": 1.5914474887748297e-05, "loss": 1.3317, "step": 12784 }, { "epoch": 31.9625, "grad_norm": 7.09375, "learning_rate": 1.5913821981023728e-05, "loss": 1.3378, "step": 12785 }, { "epoch": 31.965, "grad_norm": 5.6875, "learning_rate": 1.591316903552838e-05, "loss": 1.2887, "step": 12786 }, { "epoch": 31.9675, "grad_norm": 10.75, "learning_rate": 1.591251605126653e-05, "loss": 1.4119, "step": 12787 }, { "epoch": 31.97, "grad_norm": 6.75, "learning_rate": 1.591186302824247e-05, "loss": 1.4431, "step": 12788 }, { "epoch": 31.9725, "grad_norm": 8.6875, "learning_rate": 1.5911209966460465e-05, "loss": 1.3764, "step": 12789 }, { "epoch": 31.975, "grad_norm": 4.03125, "learning_rate": 1.5910556865924804e-05, "loss": 1.4618, "step": 12790 }, { "epoch": 31.9775, "grad_norm": 8.1875, "learning_rate": 1.5909903726639767e-05, "loss": 1.3751, "step": 12791 }, { "epoch": 31.98, "grad_norm": 8.625, "learning_rate": 1.5909250548609644e-05, "loss": 1.313, "step": 12792 }, { "epoch": 31.9825, "grad_norm": 8.3125, "learning_rate": 1.5908597331838707e-05, "loss": 1.344, "step": 12793 }, { "epoch": 31.985, "grad_norm": 7.75, "learning_rate": 1.5907944076331242e-05, "loss": 1.477, "step": 12794 }, { "epoch": 31.9875, "grad_norm": 6.75, "learning_rate": 1.5907290782091536e-05, "loss": 1.3353, "step": 12795 }, { "epoch": 31.99, "grad_norm": 7.0625, "learning_rate": 1.5906637449123864e-05, "loss": 1.3889, "step": 12796 }, { "epoch": 31.9925, "grad_norm": 9.875, "learning_rate": 1.5905984077432512e-05, "loss": 1.2574, "step": 12797 }, { "epoch": 31.995, "grad_norm": 8.5, "learning_rate": 1.590533066702177e-05, "loss": 1.3582, "step": 12798 }, { "epoch": 31.9975, "grad_norm": 8.375, "learning_rate": 1.5904677217895913e-05, "loss": 1.3656, "step": 12799 }, { "epoch": 32.0, "grad_norm": 8.125, "learning_rate": 1.5904023730059227e-05, "loss": 1.4417, "step": 12800 }, { "epoch": 32.0025, "grad_norm": 6.90625, "learning_rate": 1.5903370203516e-05, "loss": 1.2338, "step": 12801 }, { "epoch": 32.005, "grad_norm": 11.125, "learning_rate": 1.5902716638270513e-05, "loss": 1.4303, "step": 12802 }, { "epoch": 32.0075, "grad_norm": 6.65625, "learning_rate": 1.5902063034327054e-05, "loss": 1.3395, "step": 12803 }, { "epoch": 32.01, "grad_norm": 7.65625, "learning_rate": 1.5901409391689906e-05, "loss": 1.502, "step": 12804 }, { "epoch": 32.0125, "grad_norm": 7.5, "learning_rate": 1.590075571036335e-05, "loss": 1.5821, "step": 12805 }, { "epoch": 32.015, "grad_norm": 8.0625, "learning_rate": 1.590010199035168e-05, "loss": 1.3166, "step": 12806 }, { "epoch": 32.0175, "grad_norm": 8.125, "learning_rate": 1.5899448231659175e-05, "loss": 1.2981, "step": 12807 }, { "epoch": 32.02, "grad_norm": 7.96875, "learning_rate": 1.5898794434290124e-05, "loss": 1.3305, "step": 12808 }, { "epoch": 32.0225, "grad_norm": 6.9375, "learning_rate": 1.5898140598248815e-05, "loss": 1.3402, "step": 12809 }, { "epoch": 32.025, "grad_norm": 8.0625, "learning_rate": 1.589748672353953e-05, "loss": 1.3354, "step": 12810 }, { "epoch": 32.0275, "grad_norm": 9.3125, "learning_rate": 1.5896832810166557e-05, "loss": 1.4603, "step": 12811 }, { "epoch": 32.03, "grad_norm": 7.3125, "learning_rate": 1.5896178858134187e-05, "loss": 1.3992, "step": 12812 }, { "epoch": 32.0325, "grad_norm": 8.0, "learning_rate": 1.5895524867446705e-05, "loss": 1.4614, "step": 12813 }, { "epoch": 32.035, "grad_norm": 10.0, "learning_rate": 1.5894870838108395e-05, "loss": 1.5142, "step": 12814 }, { "epoch": 32.0375, "grad_norm": 8.4375, "learning_rate": 1.589421677012355e-05, "loss": 1.4791, "step": 12815 }, { "epoch": 32.04, "grad_norm": 6.0, "learning_rate": 1.589356266349645e-05, "loss": 1.412, "step": 12816 }, { "epoch": 32.0425, "grad_norm": 11.125, "learning_rate": 1.5892908518231396e-05, "loss": 1.4322, "step": 12817 }, { "epoch": 32.045, "grad_norm": 6.0625, "learning_rate": 1.5892254334332667e-05, "loss": 1.3518, "step": 12818 }, { "epoch": 32.0475, "grad_norm": 8.5625, "learning_rate": 1.5891600111804556e-05, "loss": 1.3543, "step": 12819 }, { "epoch": 32.05, "grad_norm": 8.6875, "learning_rate": 1.5890945850651347e-05, "loss": 1.3568, "step": 12820 }, { "epoch": 32.0525, "grad_norm": 8.875, "learning_rate": 1.5890291550877335e-05, "loss": 1.4407, "step": 12821 }, { "epoch": 32.055, "grad_norm": 8.6875, "learning_rate": 1.5889637212486806e-05, "loss": 1.5025, "step": 12822 }, { "epoch": 32.0575, "grad_norm": 6.84375, "learning_rate": 1.5888982835484053e-05, "loss": 1.4017, "step": 12823 }, { "epoch": 32.06, "grad_norm": 7.625, "learning_rate": 1.5888328419873363e-05, "loss": 1.3724, "step": 12824 }, { "epoch": 32.0625, "grad_norm": 7.625, "learning_rate": 1.5887673965659027e-05, "loss": 1.3036, "step": 12825 }, { "epoch": 32.065, "grad_norm": 7.59375, "learning_rate": 1.5887019472845336e-05, "loss": 1.2599, "step": 12826 }, { "epoch": 32.0675, "grad_norm": 7.21875, "learning_rate": 1.5886364941436584e-05, "loss": 1.5041, "step": 12827 }, { "epoch": 32.07, "grad_norm": 6.46875, "learning_rate": 1.5885710371437056e-05, "loss": 1.4216, "step": 12828 }, { "epoch": 32.0725, "grad_norm": 6.25, "learning_rate": 1.5885055762851047e-05, "loss": 1.3611, "step": 12829 }, { "epoch": 32.075, "grad_norm": 7.53125, "learning_rate": 1.588440111568285e-05, "loss": 1.264, "step": 12830 }, { "epoch": 32.0775, "grad_norm": 10.5625, "learning_rate": 1.5883746429936747e-05, "loss": 1.2605, "step": 12831 }, { "epoch": 32.08, "grad_norm": 8.625, "learning_rate": 1.5883091705617045e-05, "loss": 1.307, "step": 12832 }, { "epoch": 32.0825, "grad_norm": 5.34375, "learning_rate": 1.588243694272803e-05, "loss": 1.4303, "step": 12833 }, { "epoch": 32.085, "grad_norm": 5.40625, "learning_rate": 1.588178214127399e-05, "loss": 1.4811, "step": 12834 }, { "epoch": 32.0875, "grad_norm": 4.9375, "learning_rate": 1.5881127301259224e-05, "loss": 1.3946, "step": 12835 }, { "epoch": 32.09, "grad_norm": 7.96875, "learning_rate": 1.5880472422688023e-05, "loss": 1.3945, "step": 12836 }, { "epoch": 32.0925, "grad_norm": 8.0625, "learning_rate": 1.587981750556468e-05, "loss": 1.4507, "step": 12837 }, { "epoch": 32.095, "grad_norm": 6.40625, "learning_rate": 1.587916254989349e-05, "loss": 1.3955, "step": 12838 }, { "epoch": 32.0975, "grad_norm": 7.9375, "learning_rate": 1.587850755567874e-05, "loss": 1.4345, "step": 12839 }, { "epoch": 32.1, "grad_norm": 8.9375, "learning_rate": 1.5877852522924733e-05, "loss": 1.4751, "step": 12840 }, { "epoch": 32.1025, "grad_norm": 10.25, "learning_rate": 1.587719745163576e-05, "loss": 1.2895, "step": 12841 }, { "epoch": 32.105, "grad_norm": 5.96875, "learning_rate": 1.587654234181612e-05, "loss": 1.2817, "step": 12842 }, { "epoch": 32.1075, "grad_norm": 7.84375, "learning_rate": 1.5875887193470097e-05, "loss": 1.3651, "step": 12843 }, { "epoch": 32.11, "grad_norm": 7.59375, "learning_rate": 1.5875232006601993e-05, "loss": 1.2738, "step": 12844 }, { "epoch": 32.1125, "grad_norm": 8.1875, "learning_rate": 1.5874576781216105e-05, "loss": 1.3299, "step": 12845 }, { "epoch": 32.115, "grad_norm": 5.875, "learning_rate": 1.5873921517316727e-05, "loss": 1.2991, "step": 12846 }, { "epoch": 32.1175, "grad_norm": 4.78125, "learning_rate": 1.5873266214908155e-05, "loss": 1.1519, "step": 12847 }, { "epoch": 32.12, "grad_norm": 8.1875, "learning_rate": 1.5872610873994685e-05, "loss": 1.4252, "step": 12848 }, { "epoch": 32.1225, "grad_norm": 6.875, "learning_rate": 1.5871955494580614e-05, "loss": 1.398, "step": 12849 }, { "epoch": 32.125, "grad_norm": 7.65625, "learning_rate": 1.5871300076670236e-05, "loss": 1.3991, "step": 12850 }, { "epoch": 32.1275, "grad_norm": 7.75, "learning_rate": 1.587064462026785e-05, "loss": 1.2662, "step": 12851 }, { "epoch": 32.13, "grad_norm": 8.9375, "learning_rate": 1.586998912537775e-05, "loss": 1.2958, "step": 12852 }, { "epoch": 32.1325, "grad_norm": 8.9375, "learning_rate": 1.586933359200424e-05, "loss": 1.3832, "step": 12853 }, { "epoch": 32.135, "grad_norm": 5.21875, "learning_rate": 1.5868678020151614e-05, "loss": 1.3427, "step": 12854 }, { "epoch": 32.1375, "grad_norm": 6.53125, "learning_rate": 1.5868022409824167e-05, "loss": 1.4143, "step": 12855 }, { "epoch": 32.14, "grad_norm": 11.625, "learning_rate": 1.5867366761026198e-05, "loss": 1.4603, "step": 12856 }, { "epoch": 32.1425, "grad_norm": 6.84375, "learning_rate": 1.5866711073762013e-05, "loss": 1.3825, "step": 12857 }, { "epoch": 32.145, "grad_norm": 5.78125, "learning_rate": 1.5866055348035905e-05, "loss": 1.4461, "step": 12858 }, { "epoch": 32.1475, "grad_norm": 7.125, "learning_rate": 1.5865399583852166e-05, "loss": 1.5515, "step": 12859 }, { "epoch": 32.15, "grad_norm": 5.40625, "learning_rate": 1.586474378121511e-05, "loss": 1.3219, "step": 12860 }, { "epoch": 32.1525, "grad_norm": 8.9375, "learning_rate": 1.586408794012902e-05, "loss": 1.4253, "step": 12861 }, { "epoch": 32.155, "grad_norm": 5.4375, "learning_rate": 1.5863432060598213e-05, "loss": 1.3365, "step": 12862 }, { "epoch": 32.1575, "grad_norm": 8.0625, "learning_rate": 1.5862776142626976e-05, "loss": 1.3839, "step": 12863 }, { "epoch": 32.16, "grad_norm": 6.3125, "learning_rate": 1.5862120186219614e-05, "loss": 1.4679, "step": 12864 }, { "epoch": 32.1625, "grad_norm": 5.84375, "learning_rate": 1.5861464191380425e-05, "loss": 1.2636, "step": 12865 }, { "epoch": 32.165, "grad_norm": 6.78125, "learning_rate": 1.5860808158113713e-05, "loss": 1.4237, "step": 12866 }, { "epoch": 32.1675, "grad_norm": 9.25, "learning_rate": 1.5860152086423778e-05, "loss": 1.3054, "step": 12867 }, { "epoch": 32.17, "grad_norm": 9.125, "learning_rate": 1.5859495976314918e-05, "loss": 1.3558, "step": 12868 }, { "epoch": 32.1725, "grad_norm": 7.5, "learning_rate": 1.585883982779144e-05, "loss": 1.4117, "step": 12869 }, { "epoch": 32.175, "grad_norm": 5.0, "learning_rate": 1.585818364085764e-05, "loss": 1.3191, "step": 12870 }, { "epoch": 32.1775, "grad_norm": 5.25, "learning_rate": 1.5857527415517822e-05, "loss": 1.29, "step": 12871 }, { "epoch": 32.18, "grad_norm": 6.4375, "learning_rate": 1.585687115177629e-05, "loss": 1.3264, "step": 12872 }, { "epoch": 32.1825, "grad_norm": 7.46875, "learning_rate": 1.5856214849637344e-05, "loss": 1.4525, "step": 12873 }, { "epoch": 32.185, "grad_norm": 7.65625, "learning_rate": 1.585555850910529e-05, "loss": 1.2871, "step": 12874 }, { "epoch": 32.1875, "grad_norm": 9.875, "learning_rate": 1.5854902130184426e-05, "loss": 1.4101, "step": 12875 }, { "epoch": 32.19, "grad_norm": 7.0, "learning_rate": 1.585424571287906e-05, "loss": 1.4962, "step": 12876 }, { "epoch": 32.1925, "grad_norm": 8.125, "learning_rate": 1.585358925719349e-05, "loss": 1.3611, "step": 12877 }, { "epoch": 32.195, "grad_norm": 7.96875, "learning_rate": 1.5852932763132026e-05, "loss": 1.3675, "step": 12878 }, { "epoch": 32.1975, "grad_norm": 6.5625, "learning_rate": 1.585227623069897e-05, "loss": 1.2227, "step": 12879 }, { "epoch": 32.2, "grad_norm": 8.4375, "learning_rate": 1.5851619659898623e-05, "loss": 1.4006, "step": 12880 }, { "epoch": 32.2025, "grad_norm": 7.75, "learning_rate": 1.585096305073529e-05, "loss": 1.2576, "step": 12881 }, { "epoch": 32.205, "grad_norm": 6.5625, "learning_rate": 1.5850306403213283e-05, "loss": 1.4514, "step": 12882 }, { "epoch": 32.2075, "grad_norm": 7.875, "learning_rate": 1.5849649717336898e-05, "loss": 1.4651, "step": 12883 }, { "epoch": 32.21, "grad_norm": 6.0625, "learning_rate": 1.5848992993110445e-05, "loss": 1.3967, "step": 12884 }, { "epoch": 32.2125, "grad_norm": 6.46875, "learning_rate": 1.5848336230538225e-05, "loss": 1.4248, "step": 12885 }, { "epoch": 32.215, "grad_norm": 8.375, "learning_rate": 1.584767942962455e-05, "loss": 1.3291, "step": 12886 }, { "epoch": 32.2175, "grad_norm": 5.9375, "learning_rate": 1.584702259037372e-05, "loss": 1.3484, "step": 12887 }, { "epoch": 32.22, "grad_norm": 7.96875, "learning_rate": 1.584636571279005e-05, "loss": 1.439, "step": 12888 }, { "epoch": 32.2225, "grad_norm": 7.84375, "learning_rate": 1.584570879687783e-05, "loss": 1.1688, "step": 12889 }, { "epoch": 32.225, "grad_norm": 6.71875, "learning_rate": 1.5845051842641384e-05, "loss": 1.4415, "step": 12890 }, { "epoch": 32.2275, "grad_norm": 10.5625, "learning_rate": 1.584439485008501e-05, "loss": 1.5884, "step": 12891 }, { "epoch": 32.23, "grad_norm": 6.625, "learning_rate": 1.5843737819213017e-05, "loss": 1.3267, "step": 12892 }, { "epoch": 32.2325, "grad_norm": 7.96875, "learning_rate": 1.5843080750029717e-05, "loss": 1.2654, "step": 12893 }, { "epoch": 32.235, "grad_norm": 6.21875, "learning_rate": 1.584242364253941e-05, "loss": 1.3082, "step": 12894 }, { "epoch": 32.2375, "grad_norm": 6.75, "learning_rate": 1.5841766496746407e-05, "loss": 1.4736, "step": 12895 }, { "epoch": 32.24, "grad_norm": 7.40625, "learning_rate": 1.5841109312655017e-05, "loss": 1.183, "step": 12896 }, { "epoch": 32.2425, "grad_norm": 7.46875, "learning_rate": 1.5840452090269545e-05, "loss": 1.3334, "step": 12897 }, { "epoch": 32.245, "grad_norm": 7.09375, "learning_rate": 1.5839794829594308e-05, "loss": 1.4221, "step": 12898 }, { "epoch": 32.2475, "grad_norm": 7.4375, "learning_rate": 1.5839137530633605e-05, "loss": 1.3847, "step": 12899 }, { "epoch": 32.25, "grad_norm": 5.53125, "learning_rate": 1.5838480193391753e-05, "loss": 1.3583, "step": 12900 }, { "epoch": 32.2525, "grad_norm": 8.0, "learning_rate": 1.583782281787306e-05, "loss": 1.3106, "step": 12901 }, { "epoch": 32.255, "grad_norm": 7.71875, "learning_rate": 1.5837165404081834e-05, "loss": 1.2846, "step": 12902 }, { "epoch": 32.2575, "grad_norm": 7.625, "learning_rate": 1.5836507952022382e-05, "loss": 1.3437, "step": 12903 }, { "epoch": 32.26, "grad_norm": 7.53125, "learning_rate": 1.583585046169902e-05, "loss": 1.4228, "step": 12904 }, { "epoch": 32.2625, "grad_norm": 9.9375, "learning_rate": 1.5835192933116054e-05, "loss": 1.51, "step": 12905 }, { "epoch": 32.265, "grad_norm": 8.625, "learning_rate": 1.5834535366277797e-05, "loss": 1.2413, "step": 12906 }, { "epoch": 32.2675, "grad_norm": 6.6875, "learning_rate": 1.583387776118856e-05, "loss": 1.3738, "step": 12907 }, { "epoch": 32.27, "grad_norm": 7.6875, "learning_rate": 1.5833220117852655e-05, "loss": 1.2824, "step": 12908 }, { "epoch": 32.2725, "grad_norm": 5.6875, "learning_rate": 1.5832562436274393e-05, "loss": 1.3269, "step": 12909 }, { "epoch": 32.275, "grad_norm": 6.40625, "learning_rate": 1.5831904716458083e-05, "loss": 1.2948, "step": 12910 }, { "epoch": 32.2775, "grad_norm": 7.375, "learning_rate": 1.583124695840804e-05, "loss": 1.4737, "step": 12911 }, { "epoch": 32.28, "grad_norm": 5.46875, "learning_rate": 1.5830589162128574e-05, "loss": 1.2953, "step": 12912 }, { "epoch": 32.2825, "grad_norm": 8.5, "learning_rate": 1.5829931327623996e-05, "loss": 1.2996, "step": 12913 }, { "epoch": 32.285, "grad_norm": 9.0, "learning_rate": 1.5829273454898628e-05, "loss": 1.3632, "step": 12914 }, { "epoch": 32.2875, "grad_norm": 9.0, "learning_rate": 1.5828615543956773e-05, "loss": 1.391, "step": 12915 }, { "epoch": 32.29, "grad_norm": 5.46875, "learning_rate": 1.582795759480275e-05, "loss": 1.3491, "step": 12916 }, { "epoch": 32.2925, "grad_norm": 9.1875, "learning_rate": 1.5827299607440866e-05, "loss": 1.3588, "step": 12917 }, { "epoch": 32.295, "grad_norm": 8.25, "learning_rate": 1.5826641581875443e-05, "loss": 1.2651, "step": 12918 }, { "epoch": 32.2975, "grad_norm": 5.53125, "learning_rate": 1.582598351811079e-05, "loss": 1.1887, "step": 12919 }, { "epoch": 32.3, "grad_norm": 7.75, "learning_rate": 1.582532541615122e-05, "loss": 1.2503, "step": 12920 }, { "epoch": 32.3025, "grad_norm": 12.25, "learning_rate": 1.5824667276001052e-05, "loss": 1.3024, "step": 12921 }, { "epoch": 32.305, "grad_norm": 9.3125, "learning_rate": 1.5824009097664597e-05, "loss": 1.3165, "step": 12922 }, { "epoch": 32.3075, "grad_norm": 6.40625, "learning_rate": 1.5823350881146174e-05, "loss": 1.2732, "step": 12923 }, { "epoch": 32.31, "grad_norm": 10.4375, "learning_rate": 1.5822692626450096e-05, "loss": 1.3674, "step": 12924 }, { "epoch": 32.3125, "grad_norm": 8.25, "learning_rate": 1.5822034333580675e-05, "loss": 1.3135, "step": 12925 }, { "epoch": 32.315, "grad_norm": 11.25, "learning_rate": 1.582137600254223e-05, "loss": 1.2902, "step": 12926 }, { "epoch": 32.3175, "grad_norm": 6.4375, "learning_rate": 1.5820717633339077e-05, "loss": 1.376, "step": 12927 }, { "epoch": 32.32, "grad_norm": 7.9375, "learning_rate": 1.582005922597553e-05, "loss": 1.4497, "step": 12928 }, { "epoch": 32.3225, "grad_norm": 6.875, "learning_rate": 1.5819400780455913e-05, "loss": 1.2132, "step": 12929 }, { "epoch": 32.325, "grad_norm": 6.5, "learning_rate": 1.581874229678454e-05, "loss": 1.4181, "step": 12930 }, { "epoch": 32.3275, "grad_norm": 7.65625, "learning_rate": 1.5818083774965714e-05, "loss": 1.2476, "step": 12931 }, { "epoch": 32.33, "grad_norm": 8.25, "learning_rate": 1.5817425215003773e-05, "loss": 1.3889, "step": 12932 }, { "epoch": 32.3325, "grad_norm": 7.0625, "learning_rate": 1.581676661690302e-05, "loss": 1.3493, "step": 12933 }, { "epoch": 32.335, "grad_norm": 5.90625, "learning_rate": 1.5816107980667782e-05, "loss": 1.2535, "step": 12934 }, { "epoch": 32.3375, "grad_norm": 9.75, "learning_rate": 1.5815449306302372e-05, "loss": 1.4422, "step": 12935 }, { "epoch": 32.34, "grad_norm": 6.5625, "learning_rate": 1.581479059381111e-05, "loss": 1.4921, "step": 12936 }, { "epoch": 32.3425, "grad_norm": 6.78125, "learning_rate": 1.5814131843198307e-05, "loss": 1.3239, "step": 12937 }, { "epoch": 32.345, "grad_norm": 6.5625, "learning_rate": 1.5813473054468294e-05, "loss": 1.203, "step": 12938 }, { "epoch": 32.3475, "grad_norm": 7.34375, "learning_rate": 1.581281422762538e-05, "loss": 1.2029, "step": 12939 }, { "epoch": 32.35, "grad_norm": 7.15625, "learning_rate": 1.5812155362673895e-05, "loss": 1.5043, "step": 12940 }, { "epoch": 32.3525, "grad_norm": 7.78125, "learning_rate": 1.5811496459618148e-05, "loss": 1.2125, "step": 12941 }, { "epoch": 32.355, "grad_norm": 6.34375, "learning_rate": 1.5810837518462462e-05, "loss": 1.2854, "step": 12942 }, { "epoch": 32.3575, "grad_norm": 10.875, "learning_rate": 1.581017853921116e-05, "loss": 1.365, "step": 12943 }, { "epoch": 32.36, "grad_norm": 5.375, "learning_rate": 1.580951952186856e-05, "loss": 1.3153, "step": 12944 }, { "epoch": 32.3625, "grad_norm": 7.25, "learning_rate": 1.580886046643898e-05, "loss": 1.3485, "step": 12945 }, { "epoch": 32.365, "grad_norm": 9.0, "learning_rate": 1.5808201372926745e-05, "loss": 1.3898, "step": 12946 }, { "epoch": 32.3675, "grad_norm": 8.25, "learning_rate": 1.5807542241336172e-05, "loss": 1.3937, "step": 12947 }, { "epoch": 32.37, "grad_norm": 7.53125, "learning_rate": 1.5806883071671587e-05, "loss": 1.4506, "step": 12948 }, { "epoch": 32.3725, "grad_norm": 4.8125, "learning_rate": 1.580622386393731e-05, "loss": 1.348, "step": 12949 }, { "epoch": 32.375, "grad_norm": 6.5625, "learning_rate": 1.580556461813766e-05, "loss": 1.1919, "step": 12950 }, { "epoch": 32.3775, "grad_norm": 6.71875, "learning_rate": 1.580490533427696e-05, "loss": 1.3786, "step": 12951 }, { "epoch": 32.38, "grad_norm": 6.5, "learning_rate": 1.5804246012359535e-05, "loss": 1.4168, "step": 12952 }, { "epoch": 32.3825, "grad_norm": 9.25, "learning_rate": 1.5803586652389703e-05, "loss": 1.4318, "step": 12953 }, { "epoch": 32.385, "grad_norm": 6.9375, "learning_rate": 1.5802927254371788e-05, "loss": 1.3765, "step": 12954 }, { "epoch": 32.3875, "grad_norm": 5.28125, "learning_rate": 1.580226781831012e-05, "loss": 1.3283, "step": 12955 }, { "epoch": 32.39, "grad_norm": 5.1875, "learning_rate": 1.5801608344209012e-05, "loss": 1.3034, "step": 12956 }, { "epoch": 32.3925, "grad_norm": 9.375, "learning_rate": 1.5800948832072795e-05, "loss": 1.4525, "step": 12957 }, { "epoch": 32.395, "grad_norm": 6.90625, "learning_rate": 1.5800289281905784e-05, "loss": 1.3767, "step": 12958 }, { "epoch": 32.3975, "grad_norm": 8.6875, "learning_rate": 1.5799629693712313e-05, "loss": 1.1442, "step": 12959 }, { "epoch": 32.4, "grad_norm": 11.375, "learning_rate": 1.57989700674967e-05, "loss": 1.4363, "step": 12960 }, { "epoch": 32.4025, "grad_norm": 7.84375, "learning_rate": 1.5798310403263274e-05, "loss": 1.3424, "step": 12961 }, { "epoch": 32.405, "grad_norm": 4.96875, "learning_rate": 1.5797650701016352e-05, "loss": 1.4988, "step": 12962 }, { "epoch": 32.4075, "grad_norm": 7.90625, "learning_rate": 1.5796990960760267e-05, "loss": 1.3858, "step": 12963 }, { "epoch": 32.41, "grad_norm": 10.25, "learning_rate": 1.579633118249934e-05, "loss": 1.422, "step": 12964 }, { "epoch": 32.4125, "grad_norm": 6.875, "learning_rate": 1.5795671366237903e-05, "loss": 1.1838, "step": 12965 }, { "epoch": 32.415, "grad_norm": 6.4375, "learning_rate": 1.5795011511980273e-05, "loss": 1.4693, "step": 12966 }, { "epoch": 32.4175, "grad_norm": 8.4375, "learning_rate": 1.5794351619730777e-05, "loss": 1.2747, "step": 12967 }, { "epoch": 32.42, "grad_norm": 6.65625, "learning_rate": 1.5793691689493743e-05, "loss": 1.3447, "step": 12968 }, { "epoch": 32.4225, "grad_norm": 7.40625, "learning_rate": 1.5793031721273502e-05, "loss": 1.3138, "step": 12969 }, { "epoch": 32.425, "grad_norm": 9.0, "learning_rate": 1.5792371715074378e-05, "loss": 1.3659, "step": 12970 }, { "epoch": 32.4275, "grad_norm": 7.1875, "learning_rate": 1.5791711670900693e-05, "loss": 1.3877, "step": 12971 }, { "epoch": 32.43, "grad_norm": 7.9375, "learning_rate": 1.5791051588756777e-05, "loss": 1.3917, "step": 12972 }, { "epoch": 32.4325, "grad_norm": 7.875, "learning_rate": 1.5790391468646964e-05, "loss": 1.5288, "step": 12973 }, { "epoch": 32.435, "grad_norm": 8.0, "learning_rate": 1.5789731310575574e-05, "loss": 1.3563, "step": 12974 }, { "epoch": 32.4375, "grad_norm": 7.3125, "learning_rate": 1.5789071114546937e-05, "loss": 1.4968, "step": 12975 }, { "epoch": 32.44, "grad_norm": 6.375, "learning_rate": 1.578841088056538e-05, "loss": 1.2645, "step": 12976 }, { "epoch": 32.4425, "grad_norm": 8.75, "learning_rate": 1.5787750608635234e-05, "loss": 1.2603, "step": 12977 }, { "epoch": 32.445, "grad_norm": 7.1875, "learning_rate": 1.5787090298760826e-05, "loss": 1.2615, "step": 12978 }, { "epoch": 32.4475, "grad_norm": 7.4375, "learning_rate": 1.5786429950946486e-05, "loss": 1.4738, "step": 12979 }, { "epoch": 32.45, "grad_norm": 4.6875, "learning_rate": 1.5785769565196543e-05, "loss": 1.3898, "step": 12980 }, { "epoch": 32.4525, "grad_norm": 7.9375, "learning_rate": 1.5785109141515326e-05, "loss": 1.3726, "step": 12981 }, { "epoch": 32.455, "grad_norm": 11.0, "learning_rate": 1.5784448679907166e-05, "loss": 1.4691, "step": 12982 }, { "epoch": 32.4575, "grad_norm": 7.375, "learning_rate": 1.5783788180376388e-05, "loss": 1.1961, "step": 12983 }, { "epoch": 32.46, "grad_norm": 6.21875, "learning_rate": 1.578312764292733e-05, "loss": 1.4094, "step": 12984 }, { "epoch": 32.4625, "grad_norm": 7.625, "learning_rate": 1.5782467067564318e-05, "loss": 1.4889, "step": 12985 }, { "epoch": 32.465, "grad_norm": 7.1875, "learning_rate": 1.5781806454291683e-05, "loss": 1.306, "step": 12986 }, { "epoch": 32.4675, "grad_norm": 13.5625, "learning_rate": 1.5781145803113757e-05, "loss": 1.4124, "step": 12987 }, { "epoch": 32.47, "grad_norm": 6.8125, "learning_rate": 1.5780485114034868e-05, "loss": 1.4362, "step": 12988 }, { "epoch": 32.4725, "grad_norm": 7.4375, "learning_rate": 1.5779824387059352e-05, "loss": 1.2779, "step": 12989 }, { "epoch": 32.475, "grad_norm": 8.75, "learning_rate": 1.577916362219154e-05, "loss": 1.4606, "step": 12990 }, { "epoch": 32.4775, "grad_norm": 8.3125, "learning_rate": 1.577850281943576e-05, "loss": 1.2371, "step": 12991 }, { "epoch": 32.48, "grad_norm": 11.1875, "learning_rate": 1.5777841978796348e-05, "loss": 1.3012, "step": 12992 }, { "epoch": 32.4825, "grad_norm": 6.78125, "learning_rate": 1.5777181100277635e-05, "loss": 1.3498, "step": 12993 }, { "epoch": 32.485, "grad_norm": 7.625, "learning_rate": 1.5776520183883953e-05, "loss": 1.3343, "step": 12994 }, { "epoch": 32.4875, "grad_norm": 7.65625, "learning_rate": 1.5775859229619634e-05, "loss": 1.4086, "step": 12995 }, { "epoch": 32.49, "grad_norm": 5.59375, "learning_rate": 1.5775198237489016e-05, "loss": 1.364, "step": 12996 }, { "epoch": 32.4925, "grad_norm": 6.25, "learning_rate": 1.577453720749643e-05, "loss": 1.4918, "step": 12997 }, { "epoch": 32.495, "grad_norm": 8.3125, "learning_rate": 1.5773876139646208e-05, "loss": 1.3043, "step": 12998 }, { "epoch": 32.4975, "grad_norm": 6.15625, "learning_rate": 1.5773215033942685e-05, "loss": 1.3042, "step": 12999 }, { "epoch": 32.5, "grad_norm": 8.3125, "learning_rate": 1.5772553890390196e-05, "loss": 1.2745, "step": 13000 }, { "epoch": 32.5025, "grad_norm": 9.1875, "learning_rate": 1.5771892708993074e-05, "loss": 1.3162, "step": 13001 }, { "epoch": 32.505, "grad_norm": 7.03125, "learning_rate": 1.5771231489755657e-05, "loss": 1.3541, "step": 13002 }, { "epoch": 32.5075, "grad_norm": 9.875, "learning_rate": 1.577057023268227e-05, "loss": 1.2995, "step": 13003 }, { "epoch": 32.51, "grad_norm": 5.90625, "learning_rate": 1.5769908937777262e-05, "loss": 1.3424, "step": 13004 }, { "epoch": 32.5125, "grad_norm": 9.125, "learning_rate": 1.5769247605044965e-05, "loss": 1.4532, "step": 13005 }, { "epoch": 32.515, "grad_norm": 8.125, "learning_rate": 1.5768586234489706e-05, "loss": 1.5336, "step": 13006 }, { "epoch": 32.5175, "grad_norm": 6.375, "learning_rate": 1.5767924826115825e-05, "loss": 1.3002, "step": 13007 }, { "epoch": 32.52, "grad_norm": 6.78125, "learning_rate": 1.5767263379927663e-05, "loss": 1.4501, "step": 13008 }, { "epoch": 32.5225, "grad_norm": 6.125, "learning_rate": 1.5766601895929552e-05, "loss": 1.4766, "step": 13009 }, { "epoch": 32.525, "grad_norm": 9.4375, "learning_rate": 1.576594037412583e-05, "loss": 1.3961, "step": 13010 }, { "epoch": 32.5275, "grad_norm": 7.78125, "learning_rate": 1.5765278814520834e-05, "loss": 1.546, "step": 13011 }, { "epoch": 32.53, "grad_norm": 7.5625, "learning_rate": 1.57646172171189e-05, "loss": 1.5196, "step": 13012 }, { "epoch": 32.5325, "grad_norm": 5.8125, "learning_rate": 1.5763955581924368e-05, "loss": 1.3027, "step": 13013 }, { "epoch": 32.535, "grad_norm": 7.34375, "learning_rate": 1.5763293908941572e-05, "loss": 1.347, "step": 13014 }, { "epoch": 32.5375, "grad_norm": 7.46875, "learning_rate": 1.576263219817485e-05, "loss": 1.4043, "step": 13015 }, { "epoch": 32.54, "grad_norm": 7.15625, "learning_rate": 1.576197044962854e-05, "loss": 1.2982, "step": 13016 }, { "epoch": 32.5425, "grad_norm": 5.75, "learning_rate": 1.5761308663306988e-05, "loss": 1.2585, "step": 13017 }, { "epoch": 32.545, "grad_norm": 7.28125, "learning_rate": 1.5760646839214527e-05, "loss": 1.2994, "step": 13018 }, { "epoch": 32.5475, "grad_norm": 8.4375, "learning_rate": 1.5759984977355493e-05, "loss": 1.338, "step": 13019 }, { "epoch": 32.55, "grad_norm": 5.40625, "learning_rate": 1.5759323077734233e-05, "loss": 1.6001, "step": 13020 }, { "epoch": 32.5525, "grad_norm": 5.75, "learning_rate": 1.5758661140355072e-05, "loss": 1.2984, "step": 13021 }, { "epoch": 32.555, "grad_norm": 10.125, "learning_rate": 1.575799916522237e-05, "loss": 1.37, "step": 13022 }, { "epoch": 32.5575, "grad_norm": 6.15625, "learning_rate": 1.5757337152340445e-05, "loss": 1.3164, "step": 13023 }, { "epoch": 32.56, "grad_norm": 6.84375, "learning_rate": 1.5756675101713657e-05, "loss": 1.2799, "step": 13024 }, { "epoch": 32.5625, "grad_norm": 6.78125, "learning_rate": 1.575601301334633e-05, "loss": 1.2848, "step": 13025 }, { "epoch": 32.565, "grad_norm": 7.3125, "learning_rate": 1.5755350887242816e-05, "loss": 1.4329, "step": 13026 }, { "epoch": 32.5675, "grad_norm": 5.3125, "learning_rate": 1.5754688723407452e-05, "loss": 1.3234, "step": 13027 }, { "epoch": 32.57, "grad_norm": 6.15625, "learning_rate": 1.575402652184458e-05, "loss": 1.2452, "step": 13028 }, { "epoch": 32.5725, "grad_norm": 8.9375, "learning_rate": 1.575336428255854e-05, "loss": 1.5272, "step": 13029 }, { "epoch": 32.575, "grad_norm": 5.09375, "learning_rate": 1.5752702005553674e-05, "loss": 1.3192, "step": 13030 }, { "epoch": 32.5775, "grad_norm": 6.3125, "learning_rate": 1.5752039690834323e-05, "loss": 1.3929, "step": 13031 }, { "epoch": 32.58, "grad_norm": 7.96875, "learning_rate": 1.575137733840483e-05, "loss": 1.4711, "step": 13032 }, { "epoch": 32.5825, "grad_norm": 5.78125, "learning_rate": 1.5750714948269535e-05, "loss": 1.3262, "step": 13033 }, { "epoch": 32.585, "grad_norm": 5.1875, "learning_rate": 1.575005252043279e-05, "loss": 1.337, "step": 13034 }, { "epoch": 32.5875, "grad_norm": 7.1875, "learning_rate": 1.5749390054898924e-05, "loss": 1.2514, "step": 13035 }, { "epoch": 32.59, "grad_norm": 7.4375, "learning_rate": 1.574872755167229e-05, "loss": 1.323, "step": 13036 }, { "epoch": 32.5925, "grad_norm": 10.1875, "learning_rate": 1.5748065010757224e-05, "loss": 1.3409, "step": 13037 }, { "epoch": 32.595, "grad_norm": 6.46875, "learning_rate": 1.574740243215808e-05, "loss": 1.2741, "step": 13038 }, { "epoch": 32.5975, "grad_norm": 6.46875, "learning_rate": 1.574673981587919e-05, "loss": 1.2997, "step": 13039 }, { "epoch": 32.6, "grad_norm": 9.0, "learning_rate": 1.5746077161924905e-05, "loss": 1.3657, "step": 13040 }, { "epoch": 32.6025, "grad_norm": 5.125, "learning_rate": 1.5745414470299572e-05, "loss": 1.1808, "step": 13041 }, { "epoch": 32.605, "grad_norm": 7.21875, "learning_rate": 1.5744751741007527e-05, "loss": 1.4098, "step": 13042 }, { "epoch": 32.6075, "grad_norm": 8.6875, "learning_rate": 1.5744088974053125e-05, "loss": 1.353, "step": 13043 }, { "epoch": 32.61, "grad_norm": 6.625, "learning_rate": 1.57434261694407e-05, "loss": 1.1907, "step": 13044 }, { "epoch": 32.6125, "grad_norm": 7.125, "learning_rate": 1.5742763327174607e-05, "loss": 1.2533, "step": 13045 }, { "epoch": 32.615, "grad_norm": 5.5625, "learning_rate": 1.5742100447259186e-05, "loss": 1.5548, "step": 13046 }, { "epoch": 32.6175, "grad_norm": 8.5625, "learning_rate": 1.574143752969878e-05, "loss": 1.3888, "step": 13047 }, { "epoch": 32.62, "grad_norm": 7.90625, "learning_rate": 1.5740774574497747e-05, "loss": 1.4503, "step": 13048 }, { "epoch": 32.6225, "grad_norm": 6.53125, "learning_rate": 1.574011158166042e-05, "loss": 1.4311, "step": 13049 }, { "epoch": 32.625, "grad_norm": 4.75, "learning_rate": 1.573944855119115e-05, "loss": 1.313, "step": 13050 }, { "epoch": 32.6275, "grad_norm": 7.75, "learning_rate": 1.573878548309429e-05, "loss": 1.407, "step": 13051 }, { "epoch": 32.63, "grad_norm": 9.8125, "learning_rate": 1.5738122377374182e-05, "loss": 1.4608, "step": 13052 }, { "epoch": 32.6325, "grad_norm": 7.59375, "learning_rate": 1.5737459234035167e-05, "loss": 1.4703, "step": 13053 }, { "epoch": 32.635, "grad_norm": 6.25, "learning_rate": 1.5736796053081604e-05, "loss": 1.2565, "step": 13054 }, { "epoch": 32.6375, "grad_norm": 5.6875, "learning_rate": 1.5736132834517833e-05, "loss": 1.3299, "step": 13055 }, { "epoch": 32.64, "grad_norm": 6.59375, "learning_rate": 1.573546957834821e-05, "loss": 1.3286, "step": 13056 }, { "epoch": 32.6425, "grad_norm": 6.8125, "learning_rate": 1.5734806284577073e-05, "loss": 1.4322, "step": 13057 }, { "epoch": 32.645, "grad_norm": 10.5625, "learning_rate": 1.573414295320878e-05, "loss": 1.6126, "step": 13058 }, { "epoch": 32.6475, "grad_norm": 6.3125, "learning_rate": 1.5733479584247667e-05, "loss": 1.4222, "step": 13059 }, { "epoch": 32.65, "grad_norm": 5.9375, "learning_rate": 1.5732816177698097e-05, "loss": 1.1651, "step": 13060 }, { "epoch": 32.6525, "grad_norm": 8.125, "learning_rate": 1.5732152733564414e-05, "loss": 1.3064, "step": 13061 }, { "epoch": 32.655, "grad_norm": 8.1875, "learning_rate": 1.5731489251850966e-05, "loss": 1.4587, "step": 13062 }, { "epoch": 32.6575, "grad_norm": 9.6875, "learning_rate": 1.5730825732562103e-05, "loss": 1.3473, "step": 13063 }, { "epoch": 32.66, "grad_norm": 7.53125, "learning_rate": 1.573016217570218e-05, "loss": 1.1912, "step": 13064 }, { "epoch": 32.6625, "grad_norm": 10.8125, "learning_rate": 1.5729498581275534e-05, "loss": 1.4006, "step": 13065 }, { "epoch": 32.665, "grad_norm": 7.0625, "learning_rate": 1.5728834949286535e-05, "loss": 1.2946, "step": 13066 }, { "epoch": 32.6675, "grad_norm": 7.46875, "learning_rate": 1.5728171279739516e-05, "loss": 1.1247, "step": 13067 }, { "epoch": 32.67, "grad_norm": 8.125, "learning_rate": 1.5727507572638837e-05, "loss": 1.34, "step": 13068 }, { "epoch": 32.6725, "grad_norm": 6.46875, "learning_rate": 1.572684382798885e-05, "loss": 1.3039, "step": 13069 }, { "epoch": 32.675, "grad_norm": 8.0625, "learning_rate": 1.5726180045793902e-05, "loss": 1.3135, "step": 13070 }, { "epoch": 32.6775, "grad_norm": 8.375, "learning_rate": 1.572551622605835e-05, "loss": 1.3158, "step": 13071 }, { "epoch": 32.68, "grad_norm": 6.59375, "learning_rate": 1.572485236878654e-05, "loss": 1.2565, "step": 13072 }, { "epoch": 32.6825, "grad_norm": 7.125, "learning_rate": 1.5724188473982824e-05, "loss": 1.2764, "step": 13073 }, { "epoch": 32.685, "grad_norm": 7.5, "learning_rate": 1.572352454165156e-05, "loss": 1.4148, "step": 13074 }, { "epoch": 32.6875, "grad_norm": 19.375, "learning_rate": 1.5722860571797098e-05, "loss": 1.5842, "step": 13075 }, { "epoch": 32.69, "grad_norm": 9.625, "learning_rate": 1.572219656442379e-05, "loss": 1.4374, "step": 13076 }, { "epoch": 32.6925, "grad_norm": 8.375, "learning_rate": 1.5721532519535996e-05, "loss": 1.1786, "step": 13077 }, { "epoch": 32.695, "grad_norm": 7.4375, "learning_rate": 1.5720868437138057e-05, "loss": 1.3117, "step": 13078 }, { "epoch": 32.6975, "grad_norm": 8.25, "learning_rate": 1.5720204317234337e-05, "loss": 1.4647, "step": 13079 }, { "epoch": 32.7, "grad_norm": 5.53125, "learning_rate": 1.5719540159829185e-05, "loss": 1.3164, "step": 13080 }, { "epoch": 32.7025, "grad_norm": 6.40625, "learning_rate": 1.571887596492696e-05, "loss": 1.351, "step": 13081 }, { "epoch": 32.705, "grad_norm": 6.34375, "learning_rate": 1.571821173253201e-05, "loss": 1.3059, "step": 13082 }, { "epoch": 32.7075, "grad_norm": 6.5625, "learning_rate": 1.5717547462648693e-05, "loss": 1.1451, "step": 13083 }, { "epoch": 32.71, "grad_norm": 9.8125, "learning_rate": 1.5716883155281365e-05, "loss": 1.4127, "step": 13084 }, { "epoch": 32.7125, "grad_norm": 6.25, "learning_rate": 1.5716218810434374e-05, "loss": 1.3487, "step": 13085 }, { "epoch": 32.715, "grad_norm": 6.625, "learning_rate": 1.5715554428112088e-05, "loss": 1.2483, "step": 13086 }, { "epoch": 32.7175, "grad_norm": 8.0, "learning_rate": 1.5714890008318857e-05, "loss": 1.433, "step": 13087 }, { "epoch": 32.72, "grad_norm": 4.0625, "learning_rate": 1.5714225551059027e-05, "loss": 1.3347, "step": 13088 }, { "epoch": 32.7225, "grad_norm": 5.1875, "learning_rate": 1.571356105633697e-05, "loss": 1.3395, "step": 13089 }, { "epoch": 32.725, "grad_norm": 6.5625, "learning_rate": 1.571289652415703e-05, "loss": 1.3345, "step": 13090 }, { "epoch": 32.7275, "grad_norm": 6.1875, "learning_rate": 1.5712231954523576e-05, "loss": 1.2284, "step": 13091 }, { "epoch": 32.73, "grad_norm": 11.125, "learning_rate": 1.5711567347440955e-05, "loss": 1.3565, "step": 13092 }, { "epoch": 32.7325, "grad_norm": 6.40625, "learning_rate": 1.5710902702913525e-05, "loss": 1.4029, "step": 13093 }, { "epoch": 32.735, "grad_norm": 7.4375, "learning_rate": 1.571023802094565e-05, "loss": 1.3878, "step": 13094 }, { "epoch": 32.7375, "grad_norm": 7.875, "learning_rate": 1.5709573301541675e-05, "loss": 1.2758, "step": 13095 }, { "epoch": 32.74, "grad_norm": 5.375, "learning_rate": 1.5708908544705973e-05, "loss": 1.351, "step": 13096 }, { "epoch": 32.7425, "grad_norm": 8.8125, "learning_rate": 1.5708243750442894e-05, "loss": 1.2795, "step": 13097 }, { "epoch": 32.745, "grad_norm": 7.90625, "learning_rate": 1.5707578918756795e-05, "loss": 1.4229, "step": 13098 }, { "epoch": 32.7475, "grad_norm": 6.625, "learning_rate": 1.570691404965204e-05, "loss": 1.2709, "step": 13099 }, { "epoch": 32.75, "grad_norm": 5.28125, "learning_rate": 1.5706249143132982e-05, "loss": 1.3499, "step": 13100 }, { "epoch": 32.7525, "grad_norm": 7.8125, "learning_rate": 1.5705584199203986e-05, "loss": 1.3011, "step": 13101 }, { "epoch": 32.755, "grad_norm": 9.25, "learning_rate": 1.5704919217869408e-05, "loss": 1.2343, "step": 13102 }, { "epoch": 32.7575, "grad_norm": 7.5, "learning_rate": 1.5704254199133603e-05, "loss": 1.4525, "step": 13103 }, { "epoch": 32.76, "grad_norm": 5.375, "learning_rate": 1.570358914300094e-05, "loss": 1.3134, "step": 13104 }, { "epoch": 32.7625, "grad_norm": 6.84375, "learning_rate": 1.5702924049475775e-05, "loss": 1.2354, "step": 13105 }, { "epoch": 32.765, "grad_norm": 9.5625, "learning_rate": 1.5702258918562466e-05, "loss": 1.4592, "step": 13106 }, { "epoch": 32.7675, "grad_norm": 5.9375, "learning_rate": 1.570159375026538e-05, "loss": 1.2331, "step": 13107 }, { "epoch": 32.77, "grad_norm": 8.0, "learning_rate": 1.5700928544588867e-05, "loss": 1.3787, "step": 13108 }, { "epoch": 32.7725, "grad_norm": 5.375, "learning_rate": 1.57002633015373e-05, "loss": 1.2167, "step": 13109 }, { "epoch": 32.775, "grad_norm": 7.6875, "learning_rate": 1.569959802111503e-05, "loss": 1.368, "step": 13110 }, { "epoch": 32.7775, "grad_norm": 9.9375, "learning_rate": 1.569893270332643e-05, "loss": 1.3994, "step": 13111 }, { "epoch": 32.78, "grad_norm": 8.3125, "learning_rate": 1.5698267348175852e-05, "loss": 1.3287, "step": 13112 }, { "epoch": 32.7825, "grad_norm": 6.1875, "learning_rate": 1.569760195566766e-05, "loss": 1.3663, "step": 13113 }, { "epoch": 32.785, "grad_norm": 6.40625, "learning_rate": 1.5696936525806216e-05, "loss": 1.2335, "step": 13114 }, { "epoch": 32.7875, "grad_norm": 7.5625, "learning_rate": 1.5696271058595887e-05, "loss": 1.4164, "step": 13115 }, { "epoch": 32.79, "grad_norm": 8.25, "learning_rate": 1.5695605554041035e-05, "loss": 1.1196, "step": 13116 }, { "epoch": 32.7925, "grad_norm": 9.4375, "learning_rate": 1.569494001214602e-05, "loss": 1.2379, "step": 13117 }, { "epoch": 32.795, "grad_norm": 6.78125, "learning_rate": 1.56942744329152e-05, "loss": 1.3121, "step": 13118 }, { "epoch": 32.7975, "grad_norm": 8.25, "learning_rate": 1.5693608816352953e-05, "loss": 1.4289, "step": 13119 }, { "epoch": 32.8, "grad_norm": 7.5625, "learning_rate": 1.5692943162463628e-05, "loss": 1.548, "step": 13120 }, { "epoch": 32.8025, "grad_norm": 6.34375, "learning_rate": 1.5692277471251597e-05, "loss": 1.362, "step": 13121 }, { "epoch": 32.805, "grad_norm": 6.4375, "learning_rate": 1.5691611742721227e-05, "loss": 1.4318, "step": 13122 }, { "epoch": 32.8075, "grad_norm": 5.5, "learning_rate": 1.5690945976876873e-05, "loss": 1.2765, "step": 13123 }, { "epoch": 32.81, "grad_norm": 7.59375, "learning_rate": 1.5690280173722907e-05, "loss": 1.5034, "step": 13124 }, { "epoch": 32.8125, "grad_norm": 7.3125, "learning_rate": 1.568961433326369e-05, "loss": 1.2519, "step": 13125 }, { "epoch": 32.815, "grad_norm": 9.5, "learning_rate": 1.5688948455503595e-05, "loss": 1.3641, "step": 13126 }, { "epoch": 32.8175, "grad_norm": 9.625, "learning_rate": 1.5688282540446978e-05, "loss": 1.4814, "step": 13127 }, { "epoch": 32.82, "grad_norm": 8.875, "learning_rate": 1.5687616588098207e-05, "loss": 1.3093, "step": 13128 }, { "epoch": 32.8225, "grad_norm": 6.5, "learning_rate": 1.568695059846165e-05, "loss": 1.445, "step": 13129 }, { "epoch": 32.825, "grad_norm": 6.9375, "learning_rate": 1.5686284571541673e-05, "loss": 1.4322, "step": 13130 }, { "epoch": 32.8275, "grad_norm": 6.53125, "learning_rate": 1.5685618507342642e-05, "loss": 1.3313, "step": 13131 }, { "epoch": 32.83, "grad_norm": 7.5625, "learning_rate": 1.5684952405868923e-05, "loss": 1.2343, "step": 13132 }, { "epoch": 32.8325, "grad_norm": 7.71875, "learning_rate": 1.568428626712488e-05, "loss": 1.4804, "step": 13133 }, { "epoch": 32.835, "grad_norm": 9.625, "learning_rate": 1.5683620091114892e-05, "loss": 1.2308, "step": 13134 }, { "epoch": 32.8375, "grad_norm": 9.75, "learning_rate": 1.568295387784331e-05, "loss": 1.4694, "step": 13135 }, { "epoch": 32.84, "grad_norm": 7.84375, "learning_rate": 1.5682287627314513e-05, "loss": 1.3305, "step": 13136 }, { "epoch": 32.8425, "grad_norm": 6.40625, "learning_rate": 1.568162133953287e-05, "loss": 1.3402, "step": 13137 }, { "epoch": 32.845, "grad_norm": 7.90625, "learning_rate": 1.568095501450274e-05, "loss": 1.3411, "step": 13138 }, { "epoch": 32.8475, "grad_norm": 6.84375, "learning_rate": 1.5680288652228495e-05, "loss": 1.2966, "step": 13139 }, { "epoch": 32.85, "grad_norm": 9.6875, "learning_rate": 1.5679622252714507e-05, "loss": 1.6541, "step": 13140 }, { "epoch": 32.8525, "grad_norm": 6.40625, "learning_rate": 1.567895581596514e-05, "loss": 1.3597, "step": 13141 }, { "epoch": 32.855, "grad_norm": 6.84375, "learning_rate": 1.5678289341984774e-05, "loss": 1.3695, "step": 13142 }, { "epoch": 32.8575, "grad_norm": 9.375, "learning_rate": 1.5677622830777762e-05, "loss": 1.3948, "step": 13143 }, { "epoch": 32.86, "grad_norm": 8.6875, "learning_rate": 1.5676956282348488e-05, "loss": 1.3677, "step": 13144 }, { "epoch": 32.8625, "grad_norm": 5.0625, "learning_rate": 1.567628969670131e-05, "loss": 1.2658, "step": 13145 }, { "epoch": 32.865, "grad_norm": 8.5625, "learning_rate": 1.5675623073840608e-05, "loss": 1.3237, "step": 13146 }, { "epoch": 32.8675, "grad_norm": 11.0625, "learning_rate": 1.5674956413770744e-05, "loss": 1.4421, "step": 13147 }, { "epoch": 32.87, "grad_norm": 11.25, "learning_rate": 1.56742897164961e-05, "loss": 1.327, "step": 13148 }, { "epoch": 32.8725, "grad_norm": 5.90625, "learning_rate": 1.5673622982021033e-05, "loss": 1.2224, "step": 13149 }, { "epoch": 32.875, "grad_norm": 6.9375, "learning_rate": 1.5672956210349923e-05, "loss": 1.1958, "step": 13150 }, { "epoch": 32.8775, "grad_norm": 7.46875, "learning_rate": 1.567228940148714e-05, "loss": 1.4402, "step": 13151 }, { "epoch": 32.88, "grad_norm": 6.625, "learning_rate": 1.5671622555437055e-05, "loss": 1.346, "step": 13152 }, { "epoch": 32.8825, "grad_norm": 8.6875, "learning_rate": 1.567095567220404e-05, "loss": 1.3973, "step": 13153 }, { "epoch": 32.885, "grad_norm": 7.625, "learning_rate": 1.5670288751792463e-05, "loss": 1.4404, "step": 13154 }, { "epoch": 32.8875, "grad_norm": 8.125, "learning_rate": 1.56696217942067e-05, "loss": 1.4466, "step": 13155 }, { "epoch": 32.89, "grad_norm": 7.09375, "learning_rate": 1.566895479945113e-05, "loss": 1.4191, "step": 13156 }, { "epoch": 32.8925, "grad_norm": 8.4375, "learning_rate": 1.566828776753011e-05, "loss": 1.4245, "step": 13157 }, { "epoch": 32.895, "grad_norm": 6.4375, "learning_rate": 1.5667620698448027e-05, "loss": 1.3431, "step": 13158 }, { "epoch": 32.8975, "grad_norm": 8.3125, "learning_rate": 1.566695359220925e-05, "loss": 1.3255, "step": 13159 }, { "epoch": 32.9, "grad_norm": 9.6875, "learning_rate": 1.5666286448818152e-05, "loss": 1.4172, "step": 13160 }, { "epoch": 32.9025, "grad_norm": 8.0625, "learning_rate": 1.5665619268279105e-05, "loss": 1.4902, "step": 13161 }, { "epoch": 32.905, "grad_norm": 6.875, "learning_rate": 1.566495205059649e-05, "loss": 1.4741, "step": 13162 }, { "epoch": 32.9075, "grad_norm": 5.90625, "learning_rate": 1.566428479577467e-05, "loss": 1.3515, "step": 13163 }, { "epoch": 32.91, "grad_norm": 6.78125, "learning_rate": 1.5663617503818026e-05, "loss": 1.2462, "step": 13164 }, { "epoch": 32.9125, "grad_norm": 7.4375, "learning_rate": 1.5662950174730932e-05, "loss": 1.3341, "step": 13165 }, { "epoch": 32.915, "grad_norm": 10.0625, "learning_rate": 1.5662282808517768e-05, "loss": 1.4715, "step": 13166 }, { "epoch": 32.9175, "grad_norm": 6.4375, "learning_rate": 1.56616154051829e-05, "loss": 1.3161, "step": 13167 }, { "epoch": 32.92, "grad_norm": 7.0, "learning_rate": 1.566094796473071e-05, "loss": 1.2406, "step": 13168 }, { "epoch": 32.9225, "grad_norm": 10.625, "learning_rate": 1.5660280487165574e-05, "loss": 1.4306, "step": 13169 }, { "epoch": 32.925, "grad_norm": 8.75, "learning_rate": 1.565961297249186e-05, "loss": 1.4377, "step": 13170 }, { "epoch": 32.9275, "grad_norm": 6.8125, "learning_rate": 1.565894542071395e-05, "loss": 1.4156, "step": 13171 }, { "epoch": 32.93, "grad_norm": 7.40625, "learning_rate": 1.5658277831836227e-05, "loss": 1.3525, "step": 13172 }, { "epoch": 32.9325, "grad_norm": 7.125, "learning_rate": 1.5657610205863055e-05, "loss": 1.4283, "step": 13173 }, { "epoch": 32.935, "grad_norm": 8.125, "learning_rate": 1.565694254279882e-05, "loss": 1.4787, "step": 13174 }, { "epoch": 32.9375, "grad_norm": 7.4375, "learning_rate": 1.5656274842647895e-05, "loss": 1.3444, "step": 13175 }, { "epoch": 32.94, "grad_norm": 10.875, "learning_rate": 1.565560710541466e-05, "loss": 1.1984, "step": 13176 }, { "epoch": 32.9425, "grad_norm": 6.875, "learning_rate": 1.5654939331103487e-05, "loss": 1.3734, "step": 13177 }, { "epoch": 32.945, "grad_norm": 7.71875, "learning_rate": 1.565427151971876e-05, "loss": 1.4015, "step": 13178 }, { "epoch": 32.9475, "grad_norm": 8.5, "learning_rate": 1.5653603671264856e-05, "loss": 1.3478, "step": 13179 }, { "epoch": 32.95, "grad_norm": 5.90625, "learning_rate": 1.565293578574615e-05, "loss": 1.264, "step": 13180 }, { "epoch": 32.9525, "grad_norm": 6.5625, "learning_rate": 1.5652267863167026e-05, "loss": 1.4751, "step": 13181 }, { "epoch": 32.955, "grad_norm": 7.28125, "learning_rate": 1.565159990353186e-05, "loss": 1.4575, "step": 13182 }, { "epoch": 32.9575, "grad_norm": 11.0625, "learning_rate": 1.565093190684503e-05, "loss": 1.4227, "step": 13183 }, { "epoch": 32.96, "grad_norm": 6.15625, "learning_rate": 1.565026387311092e-05, "loss": 1.2496, "step": 13184 }, { "epoch": 32.9625, "grad_norm": 5.6875, "learning_rate": 1.5649595802333904e-05, "loss": 1.2497, "step": 13185 }, { "epoch": 32.965, "grad_norm": 8.0, "learning_rate": 1.5648927694518362e-05, "loss": 1.2672, "step": 13186 }, { "epoch": 32.9675, "grad_norm": 7.1875, "learning_rate": 1.5648259549668677e-05, "loss": 1.2847, "step": 13187 }, { "epoch": 32.97, "grad_norm": 5.125, "learning_rate": 1.564759136778923e-05, "loss": 1.2404, "step": 13188 }, { "epoch": 32.9725, "grad_norm": 9.9375, "learning_rate": 1.5646923148884398e-05, "loss": 1.5186, "step": 13189 }, { "epoch": 32.975, "grad_norm": 8.875, "learning_rate": 1.5646254892958566e-05, "loss": 1.2506, "step": 13190 }, { "epoch": 32.9775, "grad_norm": 9.9375, "learning_rate": 1.564558660001611e-05, "loss": 1.3922, "step": 13191 }, { "epoch": 32.98, "grad_norm": 4.5625, "learning_rate": 1.5644918270061418e-05, "loss": 1.3634, "step": 13192 }, { "epoch": 32.9825, "grad_norm": 8.625, "learning_rate": 1.5644249903098864e-05, "loss": 1.4539, "step": 13193 }, { "epoch": 32.985, "grad_norm": 7.96875, "learning_rate": 1.5643581499132835e-05, "loss": 1.5047, "step": 13194 }, { "epoch": 32.9875, "grad_norm": 9.875, "learning_rate": 1.564291305816771e-05, "loss": 1.5785, "step": 13195 }, { "epoch": 32.99, "grad_norm": 5.1875, "learning_rate": 1.5642244580207877e-05, "loss": 1.4502, "step": 13196 }, { "epoch": 32.9925, "grad_norm": 6.78125, "learning_rate": 1.5641576065257713e-05, "loss": 1.379, "step": 13197 }, { "epoch": 32.995, "grad_norm": 6.03125, "learning_rate": 1.56409075133216e-05, "loss": 1.273, "step": 13198 }, { "epoch": 32.9975, "grad_norm": 4.9375, "learning_rate": 1.5640238924403925e-05, "loss": 1.3213, "step": 13199 }, { "epoch": 33.0, "grad_norm": 9.75, "learning_rate": 1.5639570298509067e-05, "loss": 1.3812, "step": 13200 }, { "epoch": 33.0025, "grad_norm": 8.8125, "learning_rate": 1.563890163564141e-05, "loss": 1.3811, "step": 13201 }, { "epoch": 33.005, "grad_norm": 6.34375, "learning_rate": 1.5638232935805344e-05, "loss": 1.3598, "step": 13202 }, { "epoch": 33.0075, "grad_norm": 8.625, "learning_rate": 1.5637564199005247e-05, "loss": 1.5839, "step": 13203 }, { "epoch": 33.01, "grad_norm": 5.125, "learning_rate": 1.5636895425245507e-05, "loss": 1.3645, "step": 13204 }, { "epoch": 33.0125, "grad_norm": 5.25, "learning_rate": 1.5636226614530503e-05, "loss": 1.2155, "step": 13205 }, { "epoch": 33.015, "grad_norm": 11.1875, "learning_rate": 1.5635557766864625e-05, "loss": 1.566, "step": 13206 }, { "epoch": 33.0175, "grad_norm": 7.34375, "learning_rate": 1.5634888882252254e-05, "loss": 1.4169, "step": 13207 }, { "epoch": 33.02, "grad_norm": 6.59375, "learning_rate": 1.5634219960697778e-05, "loss": 1.2325, "step": 13208 }, { "epoch": 33.0225, "grad_norm": 6.875, "learning_rate": 1.563355100220558e-05, "loss": 1.294, "step": 13209 }, { "epoch": 33.025, "grad_norm": 7.53125, "learning_rate": 1.563288200678005e-05, "loss": 1.3663, "step": 13210 }, { "epoch": 33.0275, "grad_norm": 5.8125, "learning_rate": 1.5632212974425566e-05, "loss": 1.4746, "step": 13211 }, { "epoch": 33.03, "grad_norm": 11.3125, "learning_rate": 1.563154390514652e-05, "loss": 1.3119, "step": 13212 }, { "epoch": 33.0325, "grad_norm": 7.3125, "learning_rate": 1.5630874798947302e-05, "loss": 1.4604, "step": 13213 }, { "epoch": 33.035, "grad_norm": 6.0, "learning_rate": 1.5630205655832293e-05, "loss": 1.3931, "step": 13214 }, { "epoch": 33.0375, "grad_norm": 9.375, "learning_rate": 1.562953647580588e-05, "loss": 1.4117, "step": 13215 }, { "epoch": 33.04, "grad_norm": 8.625, "learning_rate": 1.562886725887245e-05, "loss": 1.3712, "step": 13216 }, { "epoch": 33.0425, "grad_norm": 7.6875, "learning_rate": 1.5628198005036392e-05, "loss": 1.3065, "step": 13217 }, { "epoch": 33.045, "grad_norm": 6.9375, "learning_rate": 1.5627528714302097e-05, "loss": 1.2921, "step": 13218 }, { "epoch": 33.0475, "grad_norm": 10.25, "learning_rate": 1.5626859386673944e-05, "loss": 1.3148, "step": 13219 }, { "epoch": 33.05, "grad_norm": 6.78125, "learning_rate": 1.5626190022156328e-05, "loss": 1.3544, "step": 13220 }, { "epoch": 33.0525, "grad_norm": 8.625, "learning_rate": 1.5625520620753634e-05, "loss": 1.3893, "step": 13221 }, { "epoch": 33.055, "grad_norm": 6.5, "learning_rate": 1.5624851182470253e-05, "loss": 1.2805, "step": 13222 }, { "epoch": 33.0575, "grad_norm": 6.3125, "learning_rate": 1.5624181707310573e-05, "loss": 1.2397, "step": 13223 }, { "epoch": 33.06, "grad_norm": 6.5625, "learning_rate": 1.5623512195278984e-05, "loss": 1.2259, "step": 13224 }, { "epoch": 33.0625, "grad_norm": 6.09375, "learning_rate": 1.5622842646379872e-05, "loss": 0.9844, "step": 13225 }, { "epoch": 33.065, "grad_norm": 7.625, "learning_rate": 1.5622173060617625e-05, "loss": 1.2388, "step": 13226 }, { "epoch": 33.0675, "grad_norm": 7.1875, "learning_rate": 1.5621503437996645e-05, "loss": 1.3147, "step": 13227 }, { "epoch": 33.07, "grad_norm": 10.4375, "learning_rate": 1.5620833778521306e-05, "loss": 1.6086, "step": 13228 }, { "epoch": 33.0725, "grad_norm": 9.4375, "learning_rate": 1.562016408219601e-05, "loss": 1.3253, "step": 13229 }, { "epoch": 33.075, "grad_norm": 6.4375, "learning_rate": 1.5619494349025142e-05, "loss": 1.2877, "step": 13230 }, { "epoch": 33.0775, "grad_norm": 6.03125, "learning_rate": 1.5618824579013094e-05, "loss": 1.36, "step": 13231 }, { "epoch": 33.08, "grad_norm": 7.28125, "learning_rate": 1.5618154772164257e-05, "loss": 1.383, "step": 13232 }, { "epoch": 33.0825, "grad_norm": 6.375, "learning_rate": 1.561748492848302e-05, "loss": 1.2306, "step": 13233 }, { "epoch": 33.085, "grad_norm": 8.5, "learning_rate": 1.5616815047973778e-05, "loss": 1.3845, "step": 13234 }, { "epoch": 33.0875, "grad_norm": 6.28125, "learning_rate": 1.561614513064092e-05, "loss": 1.4053, "step": 13235 }, { "epoch": 33.09, "grad_norm": 10.6875, "learning_rate": 1.5615475176488843e-05, "loss": 1.3721, "step": 13236 }, { "epoch": 33.0925, "grad_norm": 8.8125, "learning_rate": 1.5614805185521932e-05, "loss": 1.4176, "step": 13237 }, { "epoch": 33.095, "grad_norm": 7.65625, "learning_rate": 1.5614135157744588e-05, "loss": 1.4343, "step": 13238 }, { "epoch": 33.0975, "grad_norm": 6.96875, "learning_rate": 1.5613465093161196e-05, "loss": 1.2115, "step": 13239 }, { "epoch": 33.1, "grad_norm": 8.75, "learning_rate": 1.5612794991776147e-05, "loss": 1.379, "step": 13240 }, { "epoch": 33.1025, "grad_norm": 4.78125, "learning_rate": 1.5612124853593846e-05, "loss": 1.4089, "step": 13241 }, { "epoch": 33.105, "grad_norm": 6.75, "learning_rate": 1.5611454678618673e-05, "loss": 1.3751, "step": 13242 }, { "epoch": 33.1075, "grad_norm": 10.8125, "learning_rate": 1.5610784466855032e-05, "loss": 1.4638, "step": 13243 }, { "epoch": 33.11, "grad_norm": 6.625, "learning_rate": 1.561011421830731e-05, "loss": 1.2201, "step": 13244 }, { "epoch": 33.1125, "grad_norm": 9.25, "learning_rate": 1.5609443932979904e-05, "loss": 1.4383, "step": 13245 }, { "epoch": 33.115, "grad_norm": 6.5625, "learning_rate": 1.560877361087721e-05, "loss": 1.4698, "step": 13246 }, { "epoch": 33.1175, "grad_norm": 7.28125, "learning_rate": 1.5608103252003613e-05, "loss": 1.1298, "step": 13247 }, { "epoch": 33.12, "grad_norm": 6.125, "learning_rate": 1.5607432856363523e-05, "loss": 1.3494, "step": 13248 }, { "epoch": 33.1225, "grad_norm": 6.21875, "learning_rate": 1.5606762423961326e-05, "loss": 1.4532, "step": 13249 }, { "epoch": 33.125, "grad_norm": 7.5625, "learning_rate": 1.560609195480142e-05, "loss": 1.3799, "step": 13250 }, { "epoch": 33.1275, "grad_norm": 7.375, "learning_rate": 1.56054214488882e-05, "loss": 1.32, "step": 13251 }, { "epoch": 33.13, "grad_norm": 7.90625, "learning_rate": 1.5604750906226056e-05, "loss": 1.321, "step": 13252 }, { "epoch": 33.1325, "grad_norm": 8.1875, "learning_rate": 1.5604080326819393e-05, "loss": 1.4815, "step": 13253 }, { "epoch": 33.135, "grad_norm": 9.375, "learning_rate": 1.5603409710672605e-05, "loss": 1.2596, "step": 13254 }, { "epoch": 33.1375, "grad_norm": 7.9375, "learning_rate": 1.5602739057790085e-05, "loss": 1.4105, "step": 13255 }, { "epoch": 33.14, "grad_norm": 8.1875, "learning_rate": 1.5602068368176233e-05, "loss": 1.2402, "step": 13256 }, { "epoch": 33.1425, "grad_norm": 9.875, "learning_rate": 1.5601397641835446e-05, "loss": 1.3952, "step": 13257 }, { "epoch": 33.145, "grad_norm": 6.0625, "learning_rate": 1.5600726878772118e-05, "loss": 1.3407, "step": 13258 }, { "epoch": 33.1475, "grad_norm": 7.59375, "learning_rate": 1.560005607899065e-05, "loss": 1.4952, "step": 13259 }, { "epoch": 33.15, "grad_norm": 7.65625, "learning_rate": 1.5599385242495437e-05, "loss": 1.3573, "step": 13260 }, { "epoch": 33.1525, "grad_norm": 7.09375, "learning_rate": 1.559871436929088e-05, "loss": 1.2876, "step": 13261 }, { "epoch": 33.155, "grad_norm": 5.78125, "learning_rate": 1.5598043459381376e-05, "loss": 1.2982, "step": 13262 }, { "epoch": 33.1575, "grad_norm": 10.0625, "learning_rate": 1.559737251277132e-05, "loss": 1.4482, "step": 13263 }, { "epoch": 33.16, "grad_norm": 6.3125, "learning_rate": 1.559670152946512e-05, "loss": 1.4049, "step": 13264 }, { "epoch": 33.1625, "grad_norm": 5.5625, "learning_rate": 1.5596030509467163e-05, "loss": 1.3219, "step": 13265 }, { "epoch": 33.165, "grad_norm": 5.84375, "learning_rate": 1.5595359452781856e-05, "loss": 1.1335, "step": 13266 }, { "epoch": 33.1675, "grad_norm": 8.1875, "learning_rate": 1.559468835941359e-05, "loss": 1.3616, "step": 13267 }, { "epoch": 33.17, "grad_norm": 8.875, "learning_rate": 1.559401722936678e-05, "loss": 1.3534, "step": 13268 }, { "epoch": 33.1725, "grad_norm": 7.78125, "learning_rate": 1.5593346062645816e-05, "loss": 1.3467, "step": 13269 }, { "epoch": 33.175, "grad_norm": 6.75, "learning_rate": 1.5592674859255093e-05, "loss": 1.4163, "step": 13270 }, { "epoch": 33.1775, "grad_norm": 7.46875, "learning_rate": 1.5592003619199025e-05, "loss": 1.3498, "step": 13271 }, { "epoch": 33.18, "grad_norm": 9.4375, "learning_rate": 1.5591332342482002e-05, "loss": 1.4047, "step": 13272 }, { "epoch": 33.1825, "grad_norm": 23.625, "learning_rate": 1.5590661029108426e-05, "loss": 1.3669, "step": 13273 }, { "epoch": 33.185, "grad_norm": 6.8125, "learning_rate": 1.55899896790827e-05, "loss": 1.2631, "step": 13274 }, { "epoch": 33.1875, "grad_norm": 6.75, "learning_rate": 1.5589318292409227e-05, "loss": 1.3386, "step": 13275 }, { "epoch": 33.19, "grad_norm": 7.0, "learning_rate": 1.558864686909241e-05, "loss": 1.3352, "step": 13276 }, { "epoch": 33.1925, "grad_norm": 6.78125, "learning_rate": 1.558797540913664e-05, "loss": 1.2544, "step": 13277 }, { "epoch": 33.195, "grad_norm": 10.375, "learning_rate": 1.5587303912546335e-05, "loss": 1.3859, "step": 13278 }, { "epoch": 33.1975, "grad_norm": 6.21875, "learning_rate": 1.5586632379325882e-05, "loss": 1.3337, "step": 13279 }, { "epoch": 33.2, "grad_norm": 6.96875, "learning_rate": 1.5585960809479698e-05, "loss": 1.3091, "step": 13280 }, { "epoch": 33.2025, "grad_norm": 7.46875, "learning_rate": 1.5585289203012172e-05, "loss": 1.4827, "step": 13281 }, { "epoch": 33.205, "grad_norm": 6.34375, "learning_rate": 1.5584617559927718e-05, "loss": 1.344, "step": 13282 }, { "epoch": 33.2075, "grad_norm": 6.84375, "learning_rate": 1.5583945880230735e-05, "loss": 1.2823, "step": 13283 }, { "epoch": 33.21, "grad_norm": 8.4375, "learning_rate": 1.558327416392562e-05, "loss": 1.3621, "step": 13284 }, { "epoch": 33.2125, "grad_norm": 7.21875, "learning_rate": 1.5582602411016786e-05, "loss": 1.4131, "step": 13285 }, { "epoch": 33.215, "grad_norm": 7.59375, "learning_rate": 1.558193062150864e-05, "loss": 1.3679, "step": 13286 }, { "epoch": 33.2175, "grad_norm": 6.4375, "learning_rate": 1.5581258795405572e-05, "loss": 1.4947, "step": 13287 }, { "epoch": 33.22, "grad_norm": 6.40625, "learning_rate": 1.5580586932712e-05, "loss": 1.1845, "step": 13288 }, { "epoch": 33.2225, "grad_norm": 7.125, "learning_rate": 1.557991503343232e-05, "loss": 1.3959, "step": 13289 }, { "epoch": 33.225, "grad_norm": 9.3125, "learning_rate": 1.557924309757094e-05, "loss": 1.307, "step": 13290 }, { "epoch": 33.2275, "grad_norm": 7.71875, "learning_rate": 1.557857112513227e-05, "loss": 1.3908, "step": 13291 }, { "epoch": 33.23, "grad_norm": 7.625, "learning_rate": 1.5577899116120708e-05, "loss": 1.5252, "step": 13292 }, { "epoch": 33.2325, "grad_norm": 6.75, "learning_rate": 1.557722707054066e-05, "loss": 1.3731, "step": 13293 }, { "epoch": 33.235, "grad_norm": 7.25, "learning_rate": 1.5576554988396536e-05, "loss": 1.2858, "step": 13294 }, { "epoch": 33.2375, "grad_norm": 9.0625, "learning_rate": 1.5575882869692747e-05, "loss": 1.3999, "step": 13295 }, { "epoch": 33.24, "grad_norm": 6.0, "learning_rate": 1.5575210714433687e-05, "loss": 1.4125, "step": 13296 }, { "epoch": 33.2425, "grad_norm": 7.125, "learning_rate": 1.557453852262377e-05, "loss": 1.4228, "step": 13297 }, { "epoch": 33.245, "grad_norm": 7.0, "learning_rate": 1.5573866294267398e-05, "loss": 1.2809, "step": 13298 }, { "epoch": 33.2475, "grad_norm": 7.9375, "learning_rate": 1.557319402936899e-05, "loss": 1.4214, "step": 13299 }, { "epoch": 33.25, "grad_norm": 6.8125, "learning_rate": 1.5572521727932937e-05, "loss": 1.2509, "step": 13300 }, { "epoch": 33.2525, "grad_norm": 8.625, "learning_rate": 1.5571849389963655e-05, "loss": 1.3208, "step": 13301 }, { "epoch": 33.255, "grad_norm": 6.34375, "learning_rate": 1.5571177015465556e-05, "loss": 1.2308, "step": 13302 }, { "epoch": 33.2575, "grad_norm": 6.71875, "learning_rate": 1.557050460444304e-05, "loss": 1.3529, "step": 13303 }, { "epoch": 33.26, "grad_norm": 6.46875, "learning_rate": 1.556983215690052e-05, "loss": 1.3804, "step": 13304 }, { "epoch": 33.2625, "grad_norm": 7.71875, "learning_rate": 1.5569159672842405e-05, "loss": 1.2268, "step": 13305 }, { "epoch": 33.265, "grad_norm": 12.1875, "learning_rate": 1.5568487152273098e-05, "loss": 1.4291, "step": 13306 }, { "epoch": 33.2675, "grad_norm": 7.15625, "learning_rate": 1.5567814595197016e-05, "loss": 1.4497, "step": 13307 }, { "epoch": 33.27, "grad_norm": 6.6875, "learning_rate": 1.556714200161856e-05, "loss": 1.2848, "step": 13308 }, { "epoch": 33.2725, "grad_norm": 6.53125, "learning_rate": 1.556646937154215e-05, "loss": 1.437, "step": 13309 }, { "epoch": 33.275, "grad_norm": 9.875, "learning_rate": 1.5565796704972185e-05, "loss": 1.3273, "step": 13310 }, { "epoch": 33.2775, "grad_norm": 6.5625, "learning_rate": 1.556512400191308e-05, "loss": 1.2939, "step": 13311 }, { "epoch": 33.28, "grad_norm": 7.03125, "learning_rate": 1.5564451262369247e-05, "loss": 1.3047, "step": 13312 }, { "epoch": 33.2825, "grad_norm": 7.09375, "learning_rate": 1.556377848634509e-05, "loss": 1.373, "step": 13313 }, { "epoch": 33.285, "grad_norm": 11.9375, "learning_rate": 1.5563105673845027e-05, "loss": 1.482, "step": 13314 }, { "epoch": 33.2875, "grad_norm": 8.5, "learning_rate": 1.5562432824873467e-05, "loss": 1.2576, "step": 13315 }, { "epoch": 33.29, "grad_norm": 6.96875, "learning_rate": 1.5561759939434818e-05, "loss": 1.431, "step": 13316 }, { "epoch": 33.2925, "grad_norm": 5.03125, "learning_rate": 1.5561087017533494e-05, "loss": 1.3847, "step": 13317 }, { "epoch": 33.295, "grad_norm": 5.375, "learning_rate": 1.5560414059173904e-05, "loss": 1.1132, "step": 13318 }, { "epoch": 33.2975, "grad_norm": 9.875, "learning_rate": 1.5559741064360462e-05, "loss": 1.3918, "step": 13319 }, { "epoch": 33.3, "grad_norm": 9.625, "learning_rate": 1.5559068033097583e-05, "loss": 1.1684, "step": 13320 }, { "epoch": 33.3025, "grad_norm": 7.0, "learning_rate": 1.5558394965389674e-05, "loss": 1.434, "step": 13321 }, { "epoch": 33.305, "grad_norm": 9.125, "learning_rate": 1.555772186124115e-05, "loss": 1.3608, "step": 13322 }, { "epoch": 33.3075, "grad_norm": 7.78125, "learning_rate": 1.555704872065643e-05, "loss": 1.377, "step": 13323 }, { "epoch": 33.31, "grad_norm": 9.3125, "learning_rate": 1.5556375543639912e-05, "loss": 1.3404, "step": 13324 }, { "epoch": 33.3125, "grad_norm": 7.9375, "learning_rate": 1.5555702330196024e-05, "loss": 1.3455, "step": 13325 }, { "epoch": 33.315, "grad_norm": 5.25, "learning_rate": 1.555502908032917e-05, "loss": 1.4207, "step": 13326 }, { "epoch": 33.3175, "grad_norm": 8.375, "learning_rate": 1.555435579404377e-05, "loss": 1.4444, "step": 13327 }, { "epoch": 33.32, "grad_norm": 6.59375, "learning_rate": 1.5553682471344237e-05, "loss": 1.2581, "step": 13328 }, { "epoch": 33.3225, "grad_norm": 7.25, "learning_rate": 1.5553009112234982e-05, "loss": 1.485, "step": 13329 }, { "epoch": 33.325, "grad_norm": 7.0625, "learning_rate": 1.555233571672042e-05, "loss": 1.477, "step": 13330 }, { "epoch": 33.3275, "grad_norm": 5.90625, "learning_rate": 1.555166228480497e-05, "loss": 1.389, "step": 13331 }, { "epoch": 33.33, "grad_norm": 7.0, "learning_rate": 1.5550988816493043e-05, "loss": 1.3165, "step": 13332 }, { "epoch": 33.3325, "grad_norm": 8.875, "learning_rate": 1.555031531178906e-05, "loss": 1.5133, "step": 13333 }, { "epoch": 33.335, "grad_norm": 5.65625, "learning_rate": 1.5549641770697426e-05, "loss": 1.2562, "step": 13334 }, { "epoch": 33.3375, "grad_norm": 8.0, "learning_rate": 1.5548968193222563e-05, "loss": 1.3451, "step": 13335 }, { "epoch": 33.34, "grad_norm": 6.28125, "learning_rate": 1.554829457936889e-05, "loss": 1.3382, "step": 13336 }, { "epoch": 33.3425, "grad_norm": 9.375, "learning_rate": 1.5547620929140816e-05, "loss": 1.5193, "step": 13337 }, { "epoch": 33.345, "grad_norm": 8.25, "learning_rate": 1.5546947242542762e-05, "loss": 1.2057, "step": 13338 }, { "epoch": 33.3475, "grad_norm": 7.90625, "learning_rate": 1.5546273519579144e-05, "loss": 1.1753, "step": 13339 }, { "epoch": 33.35, "grad_norm": 6.9375, "learning_rate": 1.554559976025438e-05, "loss": 1.5085, "step": 13340 }, { "epoch": 33.3525, "grad_norm": 7.09375, "learning_rate": 1.5544925964572886e-05, "loss": 1.3865, "step": 13341 }, { "epoch": 33.355, "grad_norm": 5.46875, "learning_rate": 1.554425213253908e-05, "loss": 1.4876, "step": 13342 }, { "epoch": 33.3575, "grad_norm": 9.9375, "learning_rate": 1.5543578264157377e-05, "loss": 1.3256, "step": 13343 }, { "epoch": 33.36, "grad_norm": 6.78125, "learning_rate": 1.5542904359432198e-05, "loss": 1.5126, "step": 13344 }, { "epoch": 33.3625, "grad_norm": 6.34375, "learning_rate": 1.5542230418367957e-05, "loss": 1.4261, "step": 13345 }, { "epoch": 33.365, "grad_norm": 8.5625, "learning_rate": 1.5541556440969083e-05, "loss": 1.5171, "step": 13346 }, { "epoch": 33.3675, "grad_norm": 7.90625, "learning_rate": 1.5540882427239976e-05, "loss": 1.4854, "step": 13347 }, { "epoch": 33.37, "grad_norm": 8.0625, "learning_rate": 1.5540208377185072e-05, "loss": 1.2795, "step": 13348 }, { "epoch": 33.3725, "grad_norm": 7.625, "learning_rate": 1.553953429080878e-05, "loss": 1.2542, "step": 13349 }, { "epoch": 33.375, "grad_norm": 7.09375, "learning_rate": 1.5538860168115527e-05, "loss": 1.1888, "step": 13350 }, { "epoch": 33.3775, "grad_norm": 5.6875, "learning_rate": 1.553818600910973e-05, "loss": 1.2738, "step": 13351 }, { "epoch": 33.38, "grad_norm": 5.1875, "learning_rate": 1.55375118137958e-05, "loss": 1.41, "step": 13352 }, { "epoch": 33.3825, "grad_norm": 7.09375, "learning_rate": 1.5536837582178167e-05, "loss": 1.3451, "step": 13353 }, { "epoch": 33.385, "grad_norm": 8.3125, "learning_rate": 1.5536163314261252e-05, "loss": 1.4398, "step": 13354 }, { "epoch": 33.3875, "grad_norm": 8.875, "learning_rate": 1.5535489010049467e-05, "loss": 1.3711, "step": 13355 }, { "epoch": 33.39, "grad_norm": 9.3125, "learning_rate": 1.553481466954724e-05, "loss": 1.2712, "step": 13356 }, { "epoch": 33.3925, "grad_norm": 6.78125, "learning_rate": 1.5534140292758988e-05, "loss": 1.5837, "step": 13357 }, { "epoch": 33.395, "grad_norm": 7.21875, "learning_rate": 1.5533465879689134e-05, "loss": 1.2976, "step": 13358 }, { "epoch": 33.3975, "grad_norm": 14.0625, "learning_rate": 1.5532791430342102e-05, "loss": 1.3778, "step": 13359 }, { "epoch": 33.4, "grad_norm": 9.125, "learning_rate": 1.5532116944722308e-05, "loss": 1.1621, "step": 13360 }, { "epoch": 33.4025, "grad_norm": 8.5, "learning_rate": 1.553144242283418e-05, "loss": 1.5021, "step": 13361 }, { "epoch": 33.405, "grad_norm": 8.625, "learning_rate": 1.5530767864682132e-05, "loss": 1.3691, "step": 13362 }, { "epoch": 33.4075, "grad_norm": 7.6875, "learning_rate": 1.5530093270270595e-05, "loss": 1.2616, "step": 13363 }, { "epoch": 33.41, "grad_norm": 6.34375, "learning_rate": 1.5529418639603987e-05, "loss": 1.5563, "step": 13364 }, { "epoch": 33.4125, "grad_norm": 9.8125, "learning_rate": 1.5528743972686728e-05, "loss": 1.3617, "step": 13365 }, { "epoch": 33.415, "grad_norm": 7.8125, "learning_rate": 1.5528069269523248e-05, "loss": 1.3256, "step": 13366 }, { "epoch": 33.4175, "grad_norm": 7.25, "learning_rate": 1.552739453011797e-05, "loss": 1.1932, "step": 13367 }, { "epoch": 33.42, "grad_norm": 8.5625, "learning_rate": 1.552671975447531e-05, "loss": 1.3425, "step": 13368 }, { "epoch": 33.4225, "grad_norm": 7.90625, "learning_rate": 1.5526044942599697e-05, "loss": 1.2604, "step": 13369 }, { "epoch": 33.425, "grad_norm": 6.90625, "learning_rate": 1.5525370094495555e-05, "loss": 1.6493, "step": 13370 }, { "epoch": 33.4275, "grad_norm": 10.625, "learning_rate": 1.552469521016731e-05, "loss": 1.4441, "step": 13371 }, { "epoch": 33.43, "grad_norm": 7.96875, "learning_rate": 1.5524020289619383e-05, "loss": 1.3144, "step": 13372 }, { "epoch": 33.4325, "grad_norm": 6.0625, "learning_rate": 1.55233453328562e-05, "loss": 1.4034, "step": 13373 }, { "epoch": 33.435, "grad_norm": 5.25, "learning_rate": 1.5522670339882183e-05, "loss": 1.3779, "step": 13374 }, { "epoch": 33.4375, "grad_norm": 9.125, "learning_rate": 1.5521995310701762e-05, "loss": 1.4317, "step": 13375 }, { "epoch": 33.44, "grad_norm": 7.84375, "learning_rate": 1.5521320245319364e-05, "loss": 1.2414, "step": 13376 }, { "epoch": 33.4425, "grad_norm": 8.625, "learning_rate": 1.5520645143739406e-05, "loss": 1.3839, "step": 13377 }, { "epoch": 33.445, "grad_norm": 6.40625, "learning_rate": 1.5519970005966322e-05, "loss": 1.3731, "step": 13378 }, { "epoch": 33.4475, "grad_norm": 5.53125, "learning_rate": 1.5519294832004536e-05, "loss": 1.332, "step": 13379 }, { "epoch": 33.45, "grad_norm": 8.125, "learning_rate": 1.5518619621858474e-05, "loss": 1.2818, "step": 13380 }, { "epoch": 33.4525, "grad_norm": 7.21875, "learning_rate": 1.5517944375532565e-05, "loss": 1.3436, "step": 13381 }, { "epoch": 33.455, "grad_norm": 8.8125, "learning_rate": 1.551726909303123e-05, "loss": 1.4031, "step": 13382 }, { "epoch": 33.4575, "grad_norm": 5.78125, "learning_rate": 1.55165937743589e-05, "loss": 1.3042, "step": 13383 }, { "epoch": 33.46, "grad_norm": 9.3125, "learning_rate": 1.551591841952e-05, "loss": 1.3891, "step": 13384 }, { "epoch": 33.4625, "grad_norm": 7.28125, "learning_rate": 1.5515243028518963e-05, "loss": 1.3126, "step": 13385 }, { "epoch": 33.465, "grad_norm": 7.25, "learning_rate": 1.5514567601360214e-05, "loss": 1.2449, "step": 13386 }, { "epoch": 33.4675, "grad_norm": 6.03125, "learning_rate": 1.5513892138048178e-05, "loss": 1.3978, "step": 13387 }, { "epoch": 33.47, "grad_norm": 9.5625, "learning_rate": 1.5513216638587286e-05, "loss": 1.2813, "step": 13388 }, { "epoch": 33.4725, "grad_norm": 9.625, "learning_rate": 1.5512541102981967e-05, "loss": 1.3154, "step": 13389 }, { "epoch": 33.475, "grad_norm": 9.75, "learning_rate": 1.5511865531236653e-05, "loss": 1.359, "step": 13390 }, { "epoch": 33.4775, "grad_norm": 7.125, "learning_rate": 1.5511189923355764e-05, "loss": 1.2957, "step": 13391 }, { "epoch": 33.48, "grad_norm": 6.375, "learning_rate": 1.5510514279343736e-05, "loss": 1.3001, "step": 13392 }, { "epoch": 33.4825, "grad_norm": 6.5625, "learning_rate": 1.5509838599204997e-05, "loss": 1.3029, "step": 13393 }, { "epoch": 33.485, "grad_norm": 6.90625, "learning_rate": 1.5509162882943973e-05, "loss": 1.3753, "step": 13394 }, { "epoch": 33.4875, "grad_norm": 6.4375, "learning_rate": 1.55084871305651e-05, "loss": 1.2702, "step": 13395 }, { "epoch": 33.49, "grad_norm": 9.6875, "learning_rate": 1.5507811342072807e-05, "loss": 1.5471, "step": 13396 }, { "epoch": 33.4925, "grad_norm": 8.75, "learning_rate": 1.5507135517471525e-05, "loss": 1.3352, "step": 13397 }, { "epoch": 33.495, "grad_norm": 7.46875, "learning_rate": 1.5506459656765684e-05, "loss": 1.3564, "step": 13398 }, { "epoch": 33.4975, "grad_norm": 8.1875, "learning_rate": 1.550578375995971e-05, "loss": 1.5097, "step": 13399 }, { "epoch": 33.5, "grad_norm": 5.90625, "learning_rate": 1.5505107827058038e-05, "loss": 1.2886, "step": 13400 }, { "epoch": 33.5025, "grad_norm": 9.0625, "learning_rate": 1.55044318580651e-05, "loss": 1.3377, "step": 13401 }, { "epoch": 33.505, "grad_norm": 9.4375, "learning_rate": 1.5503755852985328e-05, "loss": 1.405, "step": 13402 }, { "epoch": 33.5075, "grad_norm": 5.59375, "learning_rate": 1.550307981182315e-05, "loss": 1.3274, "step": 13403 }, { "epoch": 33.51, "grad_norm": 7.375, "learning_rate": 1.5502403734583002e-05, "loss": 1.3495, "step": 13404 }, { "epoch": 33.5125, "grad_norm": 6.1875, "learning_rate": 1.5501727621269317e-05, "loss": 1.3308, "step": 13405 }, { "epoch": 33.515, "grad_norm": 9.5, "learning_rate": 1.5501051471886523e-05, "loss": 1.3746, "step": 13406 }, { "epoch": 33.5175, "grad_norm": 10.0625, "learning_rate": 1.550037528643906e-05, "loss": 1.2267, "step": 13407 }, { "epoch": 33.52, "grad_norm": 8.375, "learning_rate": 1.5499699064931354e-05, "loss": 1.3626, "step": 13408 }, { "epoch": 33.5225, "grad_norm": 4.875, "learning_rate": 1.549902280736784e-05, "loss": 1.4646, "step": 13409 }, { "epoch": 33.525, "grad_norm": 10.0, "learning_rate": 1.5498346513752956e-05, "loss": 1.2819, "step": 13410 }, { "epoch": 33.5275, "grad_norm": 8.0625, "learning_rate": 1.5497670184091132e-05, "loss": 1.5074, "step": 13411 }, { "epoch": 33.53, "grad_norm": 6.375, "learning_rate": 1.54969938183868e-05, "loss": 1.3862, "step": 13412 }, { "epoch": 33.5325, "grad_norm": 8.625, "learning_rate": 1.5496317416644396e-05, "loss": 1.1748, "step": 13413 }, { "epoch": 33.535, "grad_norm": 7.1875, "learning_rate": 1.5495640978868358e-05, "loss": 1.2682, "step": 13414 }, { "epoch": 33.5375, "grad_norm": 8.0625, "learning_rate": 1.5494964505063115e-05, "loss": 1.3305, "step": 13415 }, { "epoch": 33.54, "grad_norm": 10.4375, "learning_rate": 1.5494287995233107e-05, "loss": 1.3449, "step": 13416 }, { "epoch": 33.5425, "grad_norm": 6.6875, "learning_rate": 1.5493611449382763e-05, "loss": 1.2839, "step": 13417 }, { "epoch": 33.545, "grad_norm": 6.84375, "learning_rate": 1.5492934867516523e-05, "loss": 1.1955, "step": 13418 }, { "epoch": 33.5475, "grad_norm": 6.25, "learning_rate": 1.5492258249638827e-05, "loss": 1.2934, "step": 13419 }, { "epoch": 33.55, "grad_norm": 8.75, "learning_rate": 1.5491581595754102e-05, "loss": 1.4467, "step": 13420 }, { "epoch": 33.5525, "grad_norm": 6.34375, "learning_rate": 1.5490904905866787e-05, "loss": 1.3468, "step": 13421 }, { "epoch": 33.555, "grad_norm": 9.125, "learning_rate": 1.549022817998132e-05, "loss": 1.3962, "step": 13422 }, { "epoch": 33.5575, "grad_norm": 7.9375, "learning_rate": 1.5489551418102135e-05, "loss": 1.3524, "step": 13423 }, { "epoch": 33.56, "grad_norm": 8.4375, "learning_rate": 1.5488874620233674e-05, "loss": 1.5379, "step": 13424 }, { "epoch": 33.5625, "grad_norm": 7.6875, "learning_rate": 1.5488197786380367e-05, "loss": 1.4085, "step": 13425 }, { "epoch": 33.565, "grad_norm": 8.1875, "learning_rate": 1.5487520916546656e-05, "loss": 1.2866, "step": 13426 }, { "epoch": 33.5675, "grad_norm": 7.90625, "learning_rate": 1.5486844010736982e-05, "loss": 1.3842, "step": 13427 }, { "epoch": 33.57, "grad_norm": 8.875, "learning_rate": 1.5486167068955774e-05, "loss": 1.3634, "step": 13428 }, { "epoch": 33.5725, "grad_norm": 11.0, "learning_rate": 1.5485490091207473e-05, "loss": 1.3817, "step": 13429 }, { "epoch": 33.575, "grad_norm": 6.4375, "learning_rate": 1.548481307749652e-05, "loss": 1.4325, "step": 13430 }, { "epoch": 33.5775, "grad_norm": 6.34375, "learning_rate": 1.548413602782735e-05, "loss": 1.3436, "step": 13431 }, { "epoch": 33.58, "grad_norm": 10.125, "learning_rate": 1.5483458942204407e-05, "loss": 1.3016, "step": 13432 }, { "epoch": 33.5825, "grad_norm": 7.125, "learning_rate": 1.5482781820632128e-05, "loss": 1.4515, "step": 13433 }, { "epoch": 33.585, "grad_norm": 8.0625, "learning_rate": 1.548210466311495e-05, "loss": 1.4053, "step": 13434 }, { "epoch": 33.5875, "grad_norm": 7.28125, "learning_rate": 1.548142746965731e-05, "loss": 1.4375, "step": 13435 }, { "epoch": 33.59, "grad_norm": 9.0, "learning_rate": 1.5480750240263653e-05, "loss": 1.5177, "step": 13436 }, { "epoch": 33.5925, "grad_norm": 6.9375, "learning_rate": 1.5480072974938413e-05, "loss": 1.4537, "step": 13437 }, { "epoch": 33.595, "grad_norm": 7.40625, "learning_rate": 1.5479395673686038e-05, "loss": 1.2483, "step": 13438 }, { "epoch": 33.5975, "grad_norm": 5.5, "learning_rate": 1.5478718336510964e-05, "loss": 1.2416, "step": 13439 }, { "epoch": 33.6, "grad_norm": 9.1875, "learning_rate": 1.547804096341763e-05, "loss": 1.4841, "step": 13440 }, { "epoch": 33.6025, "grad_norm": 7.375, "learning_rate": 1.547736355441048e-05, "loss": 1.4125, "step": 13441 }, { "epoch": 33.605, "grad_norm": 11.125, "learning_rate": 1.5476686109493956e-05, "loss": 1.2413, "step": 13442 }, { "epoch": 33.6075, "grad_norm": 6.5625, "learning_rate": 1.5476008628672493e-05, "loss": 1.4468, "step": 13443 }, { "epoch": 33.61, "grad_norm": 6.90625, "learning_rate": 1.5475331111950538e-05, "loss": 1.377, "step": 13444 }, { "epoch": 33.6125, "grad_norm": 5.46875, "learning_rate": 1.5474653559332532e-05, "loss": 1.3994, "step": 13445 }, { "epoch": 33.615, "grad_norm": 9.625, "learning_rate": 1.5473975970822916e-05, "loss": 1.5759, "step": 13446 }, { "epoch": 33.6175, "grad_norm": 8.0625, "learning_rate": 1.5473298346426132e-05, "loss": 1.0808, "step": 13447 }, { "epoch": 33.62, "grad_norm": 6.0, "learning_rate": 1.5472620686146625e-05, "loss": 1.4381, "step": 13448 }, { "epoch": 33.6225, "grad_norm": 7.59375, "learning_rate": 1.547194298998883e-05, "loss": 1.3333, "step": 13449 }, { "epoch": 33.625, "grad_norm": 6.90625, "learning_rate": 1.5471265257957202e-05, "loss": 1.3536, "step": 13450 }, { "epoch": 33.6275, "grad_norm": 8.1875, "learning_rate": 1.5470587490056175e-05, "loss": 1.4193, "step": 13451 }, { "epoch": 33.63, "grad_norm": 6.5625, "learning_rate": 1.5469909686290197e-05, "loss": 1.3419, "step": 13452 }, { "epoch": 33.6325, "grad_norm": 6.59375, "learning_rate": 1.5469231846663708e-05, "loss": 1.3647, "step": 13453 }, { "epoch": 33.635, "grad_norm": 6.4375, "learning_rate": 1.5468553971181158e-05, "loss": 1.4462, "step": 13454 }, { "epoch": 33.6375, "grad_norm": 9.125, "learning_rate": 1.5467876059846983e-05, "loss": 1.2716, "step": 13455 }, { "epoch": 33.64, "grad_norm": 7.65625, "learning_rate": 1.5467198112665632e-05, "loss": 1.3571, "step": 13456 }, { "epoch": 33.6425, "grad_norm": 6.4375, "learning_rate": 1.546652012964155e-05, "loss": 1.3592, "step": 13457 }, { "epoch": 33.645, "grad_norm": 8.1875, "learning_rate": 1.546584211077918e-05, "loss": 1.4928, "step": 13458 }, { "epoch": 33.6475, "grad_norm": 6.21875, "learning_rate": 1.546516405608297e-05, "loss": 1.2644, "step": 13459 }, { "epoch": 33.65, "grad_norm": 6.1875, "learning_rate": 1.546448596555736e-05, "loss": 1.2719, "step": 13460 }, { "epoch": 33.6525, "grad_norm": 7.375, "learning_rate": 1.5463807839206798e-05, "loss": 1.3144, "step": 13461 }, { "epoch": 33.655, "grad_norm": 5.84375, "learning_rate": 1.5463129677035732e-05, "loss": 1.3533, "step": 13462 }, { "epoch": 33.6575, "grad_norm": 5.71875, "learning_rate": 1.5462451479048607e-05, "loss": 1.3944, "step": 13463 }, { "epoch": 33.66, "grad_norm": 7.3125, "learning_rate": 1.5461773245249866e-05, "loss": 1.4562, "step": 13464 }, { "epoch": 33.6625, "grad_norm": 6.34375, "learning_rate": 1.5461094975643964e-05, "loss": 1.6013, "step": 13465 }, { "epoch": 33.665, "grad_norm": 4.9375, "learning_rate": 1.5460416670235337e-05, "loss": 1.3017, "step": 13466 }, { "epoch": 33.6675, "grad_norm": 6.9375, "learning_rate": 1.5459738329028438e-05, "loss": 1.3651, "step": 13467 }, { "epoch": 33.67, "grad_norm": 7.3125, "learning_rate": 1.5459059952027716e-05, "loss": 1.3512, "step": 13468 }, { "epoch": 33.6725, "grad_norm": 6.40625, "learning_rate": 1.545838153923761e-05, "loss": 1.2381, "step": 13469 }, { "epoch": 33.675, "grad_norm": 8.5, "learning_rate": 1.5457703090662576e-05, "loss": 1.4026, "step": 13470 }, { "epoch": 33.6775, "grad_norm": 6.34375, "learning_rate": 1.5457024606307058e-05, "loss": 1.3838, "step": 13471 }, { "epoch": 33.68, "grad_norm": 4.78125, "learning_rate": 1.5456346086175508e-05, "loss": 1.4228, "step": 13472 }, { "epoch": 33.6825, "grad_norm": 7.46875, "learning_rate": 1.5455667530272368e-05, "loss": 1.3729, "step": 13473 }, { "epoch": 33.685, "grad_norm": 6.03125, "learning_rate": 1.5454988938602092e-05, "loss": 1.4356, "step": 13474 }, { "epoch": 33.6875, "grad_norm": 8.4375, "learning_rate": 1.5454310311169126e-05, "loss": 1.2604, "step": 13475 }, { "epoch": 33.69, "grad_norm": 7.09375, "learning_rate": 1.545363164797792e-05, "loss": 1.3298, "step": 13476 }, { "epoch": 33.6925, "grad_norm": 7.8125, "learning_rate": 1.5452952949032924e-05, "loss": 1.4103, "step": 13477 }, { "epoch": 33.695, "grad_norm": 8.6875, "learning_rate": 1.5452274214338584e-05, "loss": 1.3255, "step": 13478 }, { "epoch": 33.6975, "grad_norm": 8.1875, "learning_rate": 1.5451595443899355e-05, "loss": 1.3351, "step": 13479 }, { "epoch": 33.7, "grad_norm": 9.25, "learning_rate": 1.5450916637719683e-05, "loss": 1.2871, "step": 13480 }, { "epoch": 33.7025, "grad_norm": 7.78125, "learning_rate": 1.545023779580402e-05, "loss": 1.3983, "step": 13481 }, { "epoch": 33.705, "grad_norm": 9.375, "learning_rate": 1.5449558918156817e-05, "loss": 1.4214, "step": 13482 }, { "epoch": 33.7075, "grad_norm": 6.75, "learning_rate": 1.5448880004782523e-05, "loss": 1.4124, "step": 13483 }, { "epoch": 33.71, "grad_norm": 7.65625, "learning_rate": 1.544820105568559e-05, "loss": 1.2796, "step": 13484 }, { "epoch": 33.7125, "grad_norm": 9.5625, "learning_rate": 1.544752207087047e-05, "loss": 1.3954, "step": 13485 }, { "epoch": 33.715, "grad_norm": 9.3125, "learning_rate": 1.5446843050341608e-05, "loss": 1.3037, "step": 13486 }, { "epoch": 33.7175, "grad_norm": 9.4375, "learning_rate": 1.5446163994103468e-05, "loss": 1.3826, "step": 13487 }, { "epoch": 33.72, "grad_norm": 6.65625, "learning_rate": 1.5445484902160494e-05, "loss": 1.3502, "step": 13488 }, { "epoch": 33.7225, "grad_norm": 6.78125, "learning_rate": 1.5444805774517134e-05, "loss": 1.324, "step": 13489 }, { "epoch": 33.725, "grad_norm": 11.0, "learning_rate": 1.544412661117785e-05, "loss": 1.3918, "step": 13490 }, { "epoch": 33.7275, "grad_norm": 7.34375, "learning_rate": 1.5443447412147086e-05, "loss": 1.5234, "step": 13491 }, { "epoch": 33.73, "grad_norm": 8.3125, "learning_rate": 1.5442768177429302e-05, "loss": 1.2784, "step": 13492 }, { "epoch": 33.7325, "grad_norm": 6.3125, "learning_rate": 1.5442088907028944e-05, "loss": 1.17, "step": 13493 }, { "epoch": 33.735, "grad_norm": 7.6875, "learning_rate": 1.544140960095047e-05, "loss": 1.4352, "step": 13494 }, { "epoch": 33.7375, "grad_norm": 6.21875, "learning_rate": 1.5440730259198334e-05, "loss": 1.3429, "step": 13495 }, { "epoch": 33.74, "grad_norm": 6.625, "learning_rate": 1.544005088177699e-05, "loss": 1.5187, "step": 13496 }, { "epoch": 33.7425, "grad_norm": 6.28125, "learning_rate": 1.5439371468690886e-05, "loss": 1.3709, "step": 13497 }, { "epoch": 33.745, "grad_norm": 9.25, "learning_rate": 1.5438692019944485e-05, "loss": 1.4364, "step": 13498 }, { "epoch": 33.7475, "grad_norm": 5.53125, "learning_rate": 1.5438012535542232e-05, "loss": 1.1784, "step": 13499 }, { "epoch": 33.75, "grad_norm": 9.4375, "learning_rate": 1.5437333015488586e-05, "loss": 1.4746, "step": 13500 }, { "epoch": 33.7525, "grad_norm": 7.46875, "learning_rate": 1.5436653459788006e-05, "loss": 1.3454, "step": 13501 }, { "epoch": 33.755, "grad_norm": 7.1875, "learning_rate": 1.543597386844494e-05, "loss": 1.4809, "step": 13502 }, { "epoch": 33.7575, "grad_norm": 8.0625, "learning_rate": 1.5435294241463853e-05, "loss": 1.2699, "step": 13503 }, { "epoch": 33.76, "grad_norm": 6.875, "learning_rate": 1.543461457884919e-05, "loss": 1.2842, "step": 13504 }, { "epoch": 33.7625, "grad_norm": 9.3125, "learning_rate": 1.543393488060541e-05, "loss": 1.4349, "step": 13505 }, { "epoch": 33.765, "grad_norm": 11.3125, "learning_rate": 1.543325514673697e-05, "loss": 1.3611, "step": 13506 }, { "epoch": 33.7675, "grad_norm": 6.6875, "learning_rate": 1.543257537724833e-05, "loss": 1.562, "step": 13507 }, { "epoch": 33.77, "grad_norm": 7.59375, "learning_rate": 1.5431895572143942e-05, "loss": 1.4585, "step": 13508 }, { "epoch": 33.7725, "grad_norm": 8.25, "learning_rate": 1.5431215731428264e-05, "loss": 1.4698, "step": 13509 }, { "epoch": 33.775, "grad_norm": 6.46875, "learning_rate": 1.5430535855105754e-05, "loss": 1.243, "step": 13510 }, { "epoch": 33.7775, "grad_norm": 8.8125, "learning_rate": 1.5429855943180866e-05, "loss": 1.3827, "step": 13511 }, { "epoch": 33.78, "grad_norm": 7.71875, "learning_rate": 1.542917599565806e-05, "loss": 1.4634, "step": 13512 }, { "epoch": 33.7825, "grad_norm": 5.28125, "learning_rate": 1.5428496012541795e-05, "loss": 1.2782, "step": 13513 }, { "epoch": 33.785, "grad_norm": 6.15625, "learning_rate": 1.5427815993836526e-05, "loss": 1.3419, "step": 13514 }, { "epoch": 33.7875, "grad_norm": 11.875, "learning_rate": 1.5427135939546715e-05, "loss": 1.265, "step": 13515 }, { "epoch": 33.79, "grad_norm": 6.625, "learning_rate": 1.5426455849676814e-05, "loss": 1.3654, "step": 13516 }, { "epoch": 33.7925, "grad_norm": 7.84375, "learning_rate": 1.5425775724231285e-05, "loss": 1.4288, "step": 13517 }, { "epoch": 33.795, "grad_norm": 8.75, "learning_rate": 1.542509556321459e-05, "loss": 1.3187, "step": 13518 }, { "epoch": 33.7975, "grad_norm": 14.6875, "learning_rate": 1.542441536663119e-05, "loss": 1.561, "step": 13519 }, { "epoch": 33.8, "grad_norm": 6.96875, "learning_rate": 1.5423735134485537e-05, "loss": 1.5246, "step": 13520 }, { "epoch": 33.8025, "grad_norm": 9.9375, "learning_rate": 1.542305486678209e-05, "loss": 1.4159, "step": 13521 }, { "epoch": 33.805, "grad_norm": 5.90625, "learning_rate": 1.5422374563525316e-05, "loss": 1.3474, "step": 13522 }, { "epoch": 33.8075, "grad_norm": 9.0625, "learning_rate": 1.542169422471967e-05, "loss": 1.3644, "step": 13523 }, { "epoch": 33.81, "grad_norm": 9.875, "learning_rate": 1.5421013850369612e-05, "loss": 1.4817, "step": 13524 }, { "epoch": 33.8125, "grad_norm": 6.59375, "learning_rate": 1.542033344047961e-05, "loss": 1.4533, "step": 13525 }, { "epoch": 33.815, "grad_norm": 7.90625, "learning_rate": 1.541965299505411e-05, "loss": 1.3852, "step": 13526 }, { "epoch": 33.8175, "grad_norm": 9.0, "learning_rate": 1.5418972514097587e-05, "loss": 1.3223, "step": 13527 }, { "epoch": 33.82, "grad_norm": 6.28125, "learning_rate": 1.5418291997614497e-05, "loss": 1.4731, "step": 13528 }, { "epoch": 33.8225, "grad_norm": 7.25, "learning_rate": 1.54176114456093e-05, "loss": 1.4196, "step": 13529 }, { "epoch": 33.825, "grad_norm": 5.40625, "learning_rate": 1.541693085808646e-05, "loss": 1.2804, "step": 13530 }, { "epoch": 33.8275, "grad_norm": 11.75, "learning_rate": 1.541625023505044e-05, "loss": 1.3248, "step": 13531 }, { "epoch": 33.83, "grad_norm": 5.46875, "learning_rate": 1.5415569576505695e-05, "loss": 1.436, "step": 13532 }, { "epoch": 33.8325, "grad_norm": 5.3125, "learning_rate": 1.5414888882456696e-05, "loss": 1.314, "step": 13533 }, { "epoch": 33.835, "grad_norm": 6.65625, "learning_rate": 1.5414208152907903e-05, "loss": 1.3152, "step": 13534 }, { "epoch": 33.8375, "grad_norm": 6.03125, "learning_rate": 1.5413527387863777e-05, "loss": 1.2146, "step": 13535 }, { "epoch": 33.84, "grad_norm": 7.78125, "learning_rate": 1.541284658732878e-05, "loss": 1.4404, "step": 13536 }, { "epoch": 33.8425, "grad_norm": 9.875, "learning_rate": 1.5412165751307384e-05, "loss": 1.3786, "step": 13537 }, { "epoch": 33.845, "grad_norm": 9.1875, "learning_rate": 1.5411484879804038e-05, "loss": 1.2276, "step": 13538 }, { "epoch": 33.8475, "grad_norm": 10.6875, "learning_rate": 1.541080397282322e-05, "loss": 1.5466, "step": 13539 }, { "epoch": 33.85, "grad_norm": 6.15625, "learning_rate": 1.5410123030369387e-05, "loss": 1.3568, "step": 13540 }, { "epoch": 33.8525, "grad_norm": 8.5625, "learning_rate": 1.5409442052447003e-05, "loss": 1.2699, "step": 13541 }, { "epoch": 33.855, "grad_norm": 7.8125, "learning_rate": 1.540876103906053e-05, "loss": 1.3832, "step": 13542 }, { "epoch": 33.8575, "grad_norm": 6.84375, "learning_rate": 1.5408079990214436e-05, "loss": 1.3323, "step": 13543 }, { "epoch": 33.86, "grad_norm": 7.3125, "learning_rate": 1.5407398905913188e-05, "loss": 1.417, "step": 13544 }, { "epoch": 33.8625, "grad_norm": 7.78125, "learning_rate": 1.5406717786161252e-05, "loss": 1.4816, "step": 13545 }, { "epoch": 33.865, "grad_norm": 6.125, "learning_rate": 1.5406036630963085e-05, "loss": 1.449, "step": 13546 }, { "epoch": 33.8675, "grad_norm": 8.6875, "learning_rate": 1.5405355440323163e-05, "loss": 1.3909, "step": 13547 }, { "epoch": 33.87, "grad_norm": 6.28125, "learning_rate": 1.5404674214245946e-05, "loss": 1.3738, "step": 13548 }, { "epoch": 33.8725, "grad_norm": 5.53125, "learning_rate": 1.54039929527359e-05, "loss": 1.3511, "step": 13549 }, { "epoch": 33.875, "grad_norm": 7.125, "learning_rate": 1.5403311655797494e-05, "loss": 1.3308, "step": 13550 }, { "epoch": 33.8775, "grad_norm": 8.75, "learning_rate": 1.540263032343519e-05, "loss": 1.4074, "step": 13551 }, { "epoch": 33.88, "grad_norm": 3.953125, "learning_rate": 1.540194895565346e-05, "loss": 1.349, "step": 13552 }, { "epoch": 33.8825, "grad_norm": 7.0, "learning_rate": 1.540126755245677e-05, "loss": 1.237, "step": 13553 }, { "epoch": 33.885, "grad_norm": 8.1875, "learning_rate": 1.540058611384958e-05, "loss": 1.2623, "step": 13554 }, { "epoch": 33.8875, "grad_norm": 7.53125, "learning_rate": 1.5399904639836368e-05, "loss": 1.3781, "step": 13555 }, { "epoch": 33.89, "grad_norm": 7.46875, "learning_rate": 1.5399223130421603e-05, "loss": 1.29, "step": 13556 }, { "epoch": 33.8925, "grad_norm": 9.125, "learning_rate": 1.5398541585609737e-05, "loss": 1.4169, "step": 13557 }, { "epoch": 33.895, "grad_norm": 5.875, "learning_rate": 1.5397860005405254e-05, "loss": 1.281, "step": 13558 }, { "epoch": 33.8975, "grad_norm": 6.65625, "learning_rate": 1.5397178389812615e-05, "loss": 1.3791, "step": 13559 }, { "epoch": 33.9, "grad_norm": 9.4375, "learning_rate": 1.5396496738836292e-05, "loss": 1.3755, "step": 13560 }, { "epoch": 33.9025, "grad_norm": 13.375, "learning_rate": 1.5395815052480752e-05, "loss": 1.2489, "step": 13561 }, { "epoch": 33.905, "grad_norm": 8.125, "learning_rate": 1.5395133330750467e-05, "loss": 1.1416, "step": 13562 }, { "epoch": 33.9075, "grad_norm": 7.53125, "learning_rate": 1.5394451573649898e-05, "loss": 1.4265, "step": 13563 }, { "epoch": 33.91, "grad_norm": 7.5625, "learning_rate": 1.5393769781183525e-05, "loss": 1.2766, "step": 13564 }, { "epoch": 33.9125, "grad_norm": 8.75, "learning_rate": 1.539308795335581e-05, "loss": 1.4203, "step": 13565 }, { "epoch": 33.915, "grad_norm": 8.375, "learning_rate": 1.539240609017123e-05, "loss": 1.4352, "step": 13566 }, { "epoch": 33.9175, "grad_norm": 5.4375, "learning_rate": 1.5391724191634248e-05, "loss": 1.3841, "step": 13567 }, { "epoch": 33.92, "grad_norm": 7.9375, "learning_rate": 1.5391042257749338e-05, "loss": 1.4927, "step": 13568 }, { "epoch": 33.9225, "grad_norm": 7.5, "learning_rate": 1.5390360288520972e-05, "loss": 1.3739, "step": 13569 }, { "epoch": 33.925, "grad_norm": 6.8125, "learning_rate": 1.538967828395362e-05, "loss": 1.3485, "step": 13570 }, { "epoch": 33.9275, "grad_norm": 6.40625, "learning_rate": 1.538899624405175e-05, "loss": 1.3796, "step": 13571 }, { "epoch": 33.93, "grad_norm": 7.78125, "learning_rate": 1.5388314168819838e-05, "loss": 1.3541, "step": 13572 }, { "epoch": 33.9325, "grad_norm": 6.90625, "learning_rate": 1.5387632058262357e-05, "loss": 1.2251, "step": 13573 }, { "epoch": 33.935, "grad_norm": 5.96875, "learning_rate": 1.538694991238377e-05, "loss": 1.2757, "step": 13574 }, { "epoch": 33.9375, "grad_norm": 6.1875, "learning_rate": 1.538626773118856e-05, "loss": 1.2612, "step": 13575 }, { "epoch": 33.94, "grad_norm": 10.375, "learning_rate": 1.5385585514681193e-05, "loss": 1.613, "step": 13576 }, { "epoch": 33.9425, "grad_norm": 9.3125, "learning_rate": 1.5384903262866142e-05, "loss": 1.4698, "step": 13577 }, { "epoch": 33.945, "grad_norm": 9.125, "learning_rate": 1.538422097574788e-05, "loss": 1.2554, "step": 13578 }, { "epoch": 33.9475, "grad_norm": 8.875, "learning_rate": 1.538353865333088e-05, "loss": 1.2259, "step": 13579 }, { "epoch": 33.95, "grad_norm": 8.375, "learning_rate": 1.5382856295619622e-05, "loss": 1.3744, "step": 13580 }, { "epoch": 33.9525, "grad_norm": 8.3125, "learning_rate": 1.5382173902618566e-05, "loss": 1.319, "step": 13581 }, { "epoch": 33.955, "grad_norm": 6.78125, "learning_rate": 1.5381491474332196e-05, "loss": 1.3664, "step": 13582 }, { "epoch": 33.9575, "grad_norm": 7.84375, "learning_rate": 1.5380809010764987e-05, "loss": 1.3821, "step": 13583 }, { "epoch": 33.96, "grad_norm": 5.5625, "learning_rate": 1.5380126511921404e-05, "loss": 1.4456, "step": 13584 }, { "epoch": 33.9625, "grad_norm": 7.4375, "learning_rate": 1.5379443977805932e-05, "loss": 1.4054, "step": 13585 }, { "epoch": 33.965, "grad_norm": 8.1875, "learning_rate": 1.5378761408423037e-05, "loss": 1.4327, "step": 13586 }, { "epoch": 33.9675, "grad_norm": 9.5, "learning_rate": 1.5378078803777198e-05, "loss": 1.3264, "step": 13587 }, { "epoch": 33.97, "grad_norm": 8.125, "learning_rate": 1.537739616387289e-05, "loss": 1.3724, "step": 13588 }, { "epoch": 33.9725, "grad_norm": 11.875, "learning_rate": 1.537671348871459e-05, "loss": 1.5121, "step": 13589 }, { "epoch": 33.975, "grad_norm": 7.375, "learning_rate": 1.5376030778306766e-05, "loss": 1.2908, "step": 13590 }, { "epoch": 33.9775, "grad_norm": 5.4375, "learning_rate": 1.5375348032653902e-05, "loss": 1.2897, "step": 13591 }, { "epoch": 33.98, "grad_norm": 8.5, "learning_rate": 1.5374665251760474e-05, "loss": 1.2967, "step": 13592 }, { "epoch": 33.9825, "grad_norm": 8.5, "learning_rate": 1.5373982435630955e-05, "loss": 1.2376, "step": 13593 }, { "epoch": 33.985, "grad_norm": 6.90625, "learning_rate": 1.537329958426982e-05, "loss": 1.2906, "step": 13594 }, { "epoch": 33.9875, "grad_norm": 6.6875, "learning_rate": 1.537261669768155e-05, "loss": 1.4246, "step": 13595 }, { "epoch": 33.99, "grad_norm": 6.0625, "learning_rate": 1.5371933775870617e-05, "loss": 1.3592, "step": 13596 }, { "epoch": 33.9925, "grad_norm": 8.25, "learning_rate": 1.53712508188415e-05, "loss": 1.2209, "step": 13597 }, { "epoch": 33.995, "grad_norm": 8.4375, "learning_rate": 1.5370567826598682e-05, "loss": 1.2875, "step": 13598 }, { "epoch": 33.9975, "grad_norm": 7.0625, "learning_rate": 1.5369884799146637e-05, "loss": 1.4085, "step": 13599 }, { "epoch": 34.0, "grad_norm": 7.34375, "learning_rate": 1.536920173648984e-05, "loss": 1.3394, "step": 13600 }, { "epoch": 34.0025, "grad_norm": 8.25, "learning_rate": 1.536851863863277e-05, "loss": 1.2577, "step": 13601 }, { "epoch": 34.005, "grad_norm": 7.21875, "learning_rate": 1.536783550557991e-05, "loss": 1.4924, "step": 13602 }, { "epoch": 34.0075, "grad_norm": 8.875, "learning_rate": 1.5367152337335733e-05, "loss": 1.2939, "step": 13603 }, { "epoch": 34.01, "grad_norm": 7.75, "learning_rate": 1.536646913390472e-05, "loss": 1.312, "step": 13604 }, { "epoch": 34.0125, "grad_norm": 7.53125, "learning_rate": 1.536578589529135e-05, "loss": 1.3589, "step": 13605 }, { "epoch": 34.015, "grad_norm": 7.90625, "learning_rate": 1.53651026215001e-05, "loss": 1.3644, "step": 13606 }, { "epoch": 34.0175, "grad_norm": 8.5, "learning_rate": 1.5364419312535456e-05, "loss": 1.4982, "step": 13607 }, { "epoch": 34.02, "grad_norm": 5.75, "learning_rate": 1.5363735968401892e-05, "loss": 1.3434, "step": 13608 }, { "epoch": 34.0225, "grad_norm": 5.1875, "learning_rate": 1.536305258910389e-05, "loss": 1.2904, "step": 13609 }, { "epoch": 34.025, "grad_norm": 4.90625, "learning_rate": 1.536236917464593e-05, "loss": 1.3483, "step": 13610 }, { "epoch": 34.0275, "grad_norm": 12.25, "learning_rate": 1.5361685725032493e-05, "loss": 1.2389, "step": 13611 }, { "epoch": 34.03, "grad_norm": 7.34375, "learning_rate": 1.5361002240268058e-05, "loss": 1.4098, "step": 13612 }, { "epoch": 34.0325, "grad_norm": 5.65625, "learning_rate": 1.5360318720357105e-05, "loss": 1.3822, "step": 13613 }, { "epoch": 34.035, "grad_norm": 7.40625, "learning_rate": 1.5359635165304118e-05, "loss": 1.441, "step": 13614 }, { "epoch": 34.0375, "grad_norm": 7.6875, "learning_rate": 1.535895157511358e-05, "loss": 1.2318, "step": 13615 }, { "epoch": 34.04, "grad_norm": 5.78125, "learning_rate": 1.5358267949789968e-05, "loss": 1.358, "step": 13616 }, { "epoch": 34.0425, "grad_norm": 5.53125, "learning_rate": 1.5357584289337767e-05, "loss": 1.3213, "step": 13617 }, { "epoch": 34.045, "grad_norm": 6.15625, "learning_rate": 1.5356900593761455e-05, "loss": 1.3498, "step": 13618 }, { "epoch": 34.0475, "grad_norm": 9.25, "learning_rate": 1.535621686306552e-05, "loss": 1.3328, "step": 13619 }, { "epoch": 34.05, "grad_norm": 8.3125, "learning_rate": 1.535553309725444e-05, "loss": 1.3037, "step": 13620 }, { "epoch": 34.0525, "grad_norm": 7.15625, "learning_rate": 1.53548492963327e-05, "loss": 1.3063, "step": 13621 }, { "epoch": 34.055, "grad_norm": 6.90625, "learning_rate": 1.535416546030478e-05, "loss": 1.3497, "step": 13622 }, { "epoch": 34.0575, "grad_norm": 7.0625, "learning_rate": 1.535348158917517e-05, "loss": 1.4426, "step": 13623 }, { "epoch": 34.06, "grad_norm": 7.1875, "learning_rate": 1.5352797682948345e-05, "loss": 1.4991, "step": 13624 }, { "epoch": 34.0625, "grad_norm": 14.1875, "learning_rate": 1.5352113741628795e-05, "loss": 1.3447, "step": 13625 }, { "epoch": 34.065, "grad_norm": 5.53125, "learning_rate": 1.5351429765221e-05, "loss": 1.303, "step": 13626 }, { "epoch": 34.0675, "grad_norm": 8.625, "learning_rate": 1.5350745753729444e-05, "loss": 1.2634, "step": 13627 }, { "epoch": 34.07, "grad_norm": 8.625, "learning_rate": 1.5350061707158616e-05, "loss": 1.4756, "step": 13628 }, { "epoch": 34.0725, "grad_norm": 7.78125, "learning_rate": 1.5349377625512994e-05, "loss": 1.3386, "step": 13629 }, { "epoch": 34.075, "grad_norm": 8.0625, "learning_rate": 1.534869350879707e-05, "loss": 1.2317, "step": 13630 }, { "epoch": 34.0775, "grad_norm": 7.8125, "learning_rate": 1.5348009357015325e-05, "loss": 1.476, "step": 13631 }, { "epoch": 34.08, "grad_norm": 7.9375, "learning_rate": 1.5347325170172246e-05, "loss": 1.3279, "step": 13632 }, { "epoch": 34.0825, "grad_norm": 6.3125, "learning_rate": 1.5346640948272313e-05, "loss": 1.4481, "step": 13633 }, { "epoch": 34.085, "grad_norm": 9.1875, "learning_rate": 1.534595669132002e-05, "loss": 1.439, "step": 13634 }, { "epoch": 34.0875, "grad_norm": 8.5625, "learning_rate": 1.5345272399319843e-05, "loss": 1.3426, "step": 13635 }, { "epoch": 34.09, "grad_norm": 7.8125, "learning_rate": 1.534458807227628e-05, "loss": 1.399, "step": 13636 }, { "epoch": 34.0925, "grad_norm": 7.65625, "learning_rate": 1.5343903710193808e-05, "loss": 1.431, "step": 13637 }, { "epoch": 34.095, "grad_norm": 7.21875, "learning_rate": 1.534321931307692e-05, "loss": 1.2981, "step": 13638 }, { "epoch": 34.0975, "grad_norm": 7.375, "learning_rate": 1.53425348809301e-05, "loss": 1.3452, "step": 13639 }, { "epoch": 34.1, "grad_norm": 7.53125, "learning_rate": 1.5341850413757834e-05, "loss": 1.4955, "step": 13640 }, { "epoch": 34.1025, "grad_norm": 6.75, "learning_rate": 1.534116591156461e-05, "loss": 1.3728, "step": 13641 }, { "epoch": 34.105, "grad_norm": 7.53125, "learning_rate": 1.5340481374354913e-05, "loss": 1.3474, "step": 13642 }, { "epoch": 34.1075, "grad_norm": 6.375, "learning_rate": 1.533979680213324e-05, "loss": 1.3257, "step": 13643 }, { "epoch": 34.11, "grad_norm": 6.15625, "learning_rate": 1.5339112194904066e-05, "loss": 1.2032, "step": 13644 }, { "epoch": 34.1125, "grad_norm": 6.4375, "learning_rate": 1.5338427552671894e-05, "loss": 1.1028, "step": 13645 }, { "epoch": 34.115, "grad_norm": 7.875, "learning_rate": 1.5337742875441198e-05, "loss": 1.2761, "step": 13646 }, { "epoch": 34.1175, "grad_norm": 8.75, "learning_rate": 1.5337058163216477e-05, "loss": 1.4332, "step": 13647 }, { "epoch": 34.12, "grad_norm": 10.1875, "learning_rate": 1.533637341600221e-05, "loss": 1.5123, "step": 13648 }, { "epoch": 34.1225, "grad_norm": 5.5, "learning_rate": 1.5335688633802903e-05, "loss": 1.3402, "step": 13649 }, { "epoch": 34.125, "grad_norm": 6.1875, "learning_rate": 1.5335003816623027e-05, "loss": 1.3122, "step": 13650 }, { "epoch": 34.1275, "grad_norm": 10.125, "learning_rate": 1.533431896446708e-05, "loss": 1.5386, "step": 13651 }, { "epoch": 34.13, "grad_norm": 7.53125, "learning_rate": 1.5333634077339558e-05, "loss": 1.4651, "step": 13652 }, { "epoch": 34.1325, "grad_norm": 6.53125, "learning_rate": 1.5332949155244938e-05, "loss": 1.4177, "step": 13653 }, { "epoch": 34.135, "grad_norm": 6.1875, "learning_rate": 1.533226419818772e-05, "loss": 1.1983, "step": 13654 }, { "epoch": 34.1375, "grad_norm": 8.1875, "learning_rate": 1.5331579206172387e-05, "loss": 1.3537, "step": 13655 }, { "epoch": 34.14, "grad_norm": 9.5, "learning_rate": 1.5330894179203436e-05, "loss": 1.3059, "step": 13656 }, { "epoch": 34.1425, "grad_norm": 7.21875, "learning_rate": 1.5330209117285358e-05, "loss": 1.3932, "step": 13657 }, { "epoch": 34.145, "grad_norm": 6.40625, "learning_rate": 1.532952402042264e-05, "loss": 1.3394, "step": 13658 }, { "epoch": 34.1475, "grad_norm": 7.75, "learning_rate": 1.532883888861978e-05, "loss": 1.2054, "step": 13659 }, { "epoch": 34.15, "grad_norm": 6.78125, "learning_rate": 1.532815372188126e-05, "loss": 1.5094, "step": 13660 }, { "epoch": 34.1525, "grad_norm": 9.125, "learning_rate": 1.532746852021158e-05, "loss": 1.4122, "step": 13661 }, { "epoch": 34.155, "grad_norm": 7.875, "learning_rate": 1.532678328361523e-05, "loss": 1.3035, "step": 13662 }, { "epoch": 34.1575, "grad_norm": 8.3125, "learning_rate": 1.53260980120967e-05, "loss": 1.2137, "step": 13663 }, { "epoch": 34.16, "grad_norm": 6.53125, "learning_rate": 1.532541270566049e-05, "loss": 1.2623, "step": 13664 }, { "epoch": 34.1625, "grad_norm": 6.0, "learning_rate": 1.532472736431108e-05, "loss": 1.4199, "step": 13665 }, { "epoch": 34.165, "grad_norm": 5.53125, "learning_rate": 1.5324041988052976e-05, "loss": 1.3173, "step": 13666 }, { "epoch": 34.1675, "grad_norm": 5.8125, "learning_rate": 1.5323356576890663e-05, "loss": 1.3281, "step": 13667 }, { "epoch": 34.17, "grad_norm": 11.0, "learning_rate": 1.5322671130828635e-05, "loss": 1.3047, "step": 13668 }, { "epoch": 34.1725, "grad_norm": 8.875, "learning_rate": 1.5321985649871394e-05, "loss": 1.2032, "step": 13669 }, { "epoch": 34.175, "grad_norm": 7.5, "learning_rate": 1.5321300134023423e-05, "loss": 1.4336, "step": 13670 }, { "epoch": 34.1775, "grad_norm": 8.625, "learning_rate": 1.5320614583289227e-05, "loss": 1.5473, "step": 13671 }, { "epoch": 34.18, "grad_norm": 5.625, "learning_rate": 1.531992899767329e-05, "loss": 1.2234, "step": 13672 }, { "epoch": 34.1825, "grad_norm": 7.1875, "learning_rate": 1.531924337718011e-05, "loss": 1.4008, "step": 13673 }, { "epoch": 34.185, "grad_norm": 5.09375, "learning_rate": 1.5318557721814188e-05, "loss": 1.1617, "step": 13674 }, { "epoch": 34.1875, "grad_norm": 7.625, "learning_rate": 1.5317872031580012e-05, "loss": 1.38, "step": 13675 }, { "epoch": 34.19, "grad_norm": 7.3125, "learning_rate": 1.5317186306482082e-05, "loss": 1.3909, "step": 13676 }, { "epoch": 34.1925, "grad_norm": 6.6875, "learning_rate": 1.5316500546524887e-05, "loss": 1.2782, "step": 13677 }, { "epoch": 34.195, "grad_norm": 9.25, "learning_rate": 1.5315814751712933e-05, "loss": 1.3511, "step": 13678 }, { "epoch": 34.1975, "grad_norm": 10.0625, "learning_rate": 1.5315128922050703e-05, "loss": 1.3306, "step": 13679 }, { "epoch": 34.2, "grad_norm": 7.3125, "learning_rate": 1.5314443057542703e-05, "loss": 1.4012, "step": 13680 }, { "epoch": 34.2025, "grad_norm": 8.5625, "learning_rate": 1.531375715819343e-05, "loss": 1.6195, "step": 13681 }, { "epoch": 34.205, "grad_norm": 7.15625, "learning_rate": 1.5313071224007375e-05, "loss": 1.4259, "step": 13682 }, { "epoch": 34.2075, "grad_norm": 7.0, "learning_rate": 1.5312385254989038e-05, "loss": 1.3308, "step": 13683 }, { "epoch": 34.21, "grad_norm": 7.78125, "learning_rate": 1.5311699251142916e-05, "loss": 1.261, "step": 13684 }, { "epoch": 34.2125, "grad_norm": 5.6875, "learning_rate": 1.5311013212473507e-05, "loss": 1.288, "step": 13685 }, { "epoch": 34.215, "grad_norm": 8.125, "learning_rate": 1.5310327138985308e-05, "loss": 1.4109, "step": 13686 }, { "epoch": 34.2175, "grad_norm": 7.21875, "learning_rate": 1.5309641030682822e-05, "loss": 1.3372, "step": 13687 }, { "epoch": 34.22, "grad_norm": 8.9375, "learning_rate": 1.530895488757053e-05, "loss": 1.3123, "step": 13688 }, { "epoch": 34.2225, "grad_norm": 7.9375, "learning_rate": 1.530826870965295e-05, "loss": 1.2535, "step": 13689 }, { "epoch": 34.225, "grad_norm": 7.34375, "learning_rate": 1.530758249693457e-05, "loss": 1.3703, "step": 13690 }, { "epoch": 34.2275, "grad_norm": 8.875, "learning_rate": 1.5306896249419896e-05, "loss": 1.2266, "step": 13691 }, { "epoch": 34.23, "grad_norm": 16.375, "learning_rate": 1.530620996711342e-05, "loss": 1.3765, "step": 13692 }, { "epoch": 34.2325, "grad_norm": 6.0625, "learning_rate": 1.530552365001964e-05, "loss": 1.2255, "step": 13693 }, { "epoch": 34.235, "grad_norm": 7.15625, "learning_rate": 1.5304837298143062e-05, "loss": 1.4386, "step": 13694 }, { "epoch": 34.2375, "grad_norm": 7.09375, "learning_rate": 1.5304150911488185e-05, "loss": 1.3131, "step": 13695 }, { "epoch": 34.24, "grad_norm": 6.96875, "learning_rate": 1.5303464490059506e-05, "loss": 1.3221, "step": 13696 }, { "epoch": 34.2425, "grad_norm": 5.90625, "learning_rate": 1.5302778033861523e-05, "loss": 1.2931, "step": 13697 }, { "epoch": 34.245, "grad_norm": 8.9375, "learning_rate": 1.5302091542898743e-05, "loss": 1.4149, "step": 13698 }, { "epoch": 34.2475, "grad_norm": 8.375, "learning_rate": 1.5301405017175662e-05, "loss": 1.2019, "step": 13699 }, { "epoch": 34.25, "grad_norm": 10.4375, "learning_rate": 1.530071845669678e-05, "loss": 1.3168, "step": 13700 }, { "epoch": 34.2525, "grad_norm": 7.5625, "learning_rate": 1.5300031861466603e-05, "loss": 1.2284, "step": 13701 }, { "epoch": 34.255, "grad_norm": 7.1875, "learning_rate": 1.529934523148963e-05, "loss": 1.3915, "step": 13702 }, { "epoch": 34.2575, "grad_norm": 9.5, "learning_rate": 1.5298658566770357e-05, "loss": 1.3589, "step": 13703 }, { "epoch": 34.26, "grad_norm": 7.125, "learning_rate": 1.5297971867313293e-05, "loss": 1.3787, "step": 13704 }, { "epoch": 34.2625, "grad_norm": 10.0625, "learning_rate": 1.5297285133122936e-05, "loss": 1.3049, "step": 13705 }, { "epoch": 34.265, "grad_norm": 7.8125, "learning_rate": 1.529659836420379e-05, "loss": 1.2627, "step": 13706 }, { "epoch": 34.2675, "grad_norm": 7.5, "learning_rate": 1.529591156056036e-05, "loss": 1.3133, "step": 13707 }, { "epoch": 34.27, "grad_norm": 6.3125, "learning_rate": 1.5295224722197145e-05, "loss": 1.4872, "step": 13708 }, { "epoch": 34.2725, "grad_norm": 8.75, "learning_rate": 1.5294537849118646e-05, "loss": 1.2915, "step": 13709 }, { "epoch": 34.275, "grad_norm": 6.03125, "learning_rate": 1.529385094132937e-05, "loss": 1.3665, "step": 13710 }, { "epoch": 34.2775, "grad_norm": 7.40625, "learning_rate": 1.529316399883382e-05, "loss": 1.313, "step": 13711 }, { "epoch": 34.28, "grad_norm": 6.78125, "learning_rate": 1.5292477021636498e-05, "loss": 1.2699, "step": 13712 }, { "epoch": 34.2825, "grad_norm": 8.125, "learning_rate": 1.5291790009741906e-05, "loss": 1.2936, "step": 13713 }, { "epoch": 34.285, "grad_norm": 7.3125, "learning_rate": 1.5291102963154553e-05, "loss": 1.4668, "step": 13714 }, { "epoch": 34.2875, "grad_norm": 7.0, "learning_rate": 1.529041588187894e-05, "loss": 1.3611, "step": 13715 }, { "epoch": 34.29, "grad_norm": 7.1875, "learning_rate": 1.5289728765919575e-05, "loss": 1.4579, "step": 13716 }, { "epoch": 34.2925, "grad_norm": 8.4375, "learning_rate": 1.5289041615280956e-05, "loss": 1.3114, "step": 13717 }, { "epoch": 34.295, "grad_norm": 8.0625, "learning_rate": 1.5288354429967593e-05, "loss": 1.3024, "step": 13718 }, { "epoch": 34.2975, "grad_norm": 7.34375, "learning_rate": 1.5287667209983993e-05, "loss": 1.327, "step": 13719 }, { "epoch": 34.3, "grad_norm": 7.375, "learning_rate": 1.5286979955334655e-05, "loss": 1.3321, "step": 13720 }, { "epoch": 34.3025, "grad_norm": 9.5, "learning_rate": 1.528629266602409e-05, "loss": 1.4263, "step": 13721 }, { "epoch": 34.305, "grad_norm": 7.53125, "learning_rate": 1.52856053420568e-05, "loss": 1.3828, "step": 13722 }, { "epoch": 34.3075, "grad_norm": 8.5625, "learning_rate": 1.5284917983437293e-05, "loss": 1.3053, "step": 13723 }, { "epoch": 34.31, "grad_norm": 7.8125, "learning_rate": 1.5284230590170074e-05, "loss": 1.3195, "step": 13724 }, { "epoch": 34.3125, "grad_norm": 5.8125, "learning_rate": 1.5283543162259652e-05, "loss": 1.3996, "step": 13725 }, { "epoch": 34.315, "grad_norm": 7.46875, "learning_rate": 1.5282855699710535e-05, "loss": 1.3273, "step": 13726 }, { "epoch": 34.3175, "grad_norm": 6.75, "learning_rate": 1.5282168202527228e-05, "loss": 1.4725, "step": 13727 }, { "epoch": 34.32, "grad_norm": 9.5625, "learning_rate": 1.528148067071423e-05, "loss": 1.3495, "step": 13728 }, { "epoch": 34.3225, "grad_norm": 6.53125, "learning_rate": 1.5280793104276063e-05, "loss": 1.3188, "step": 13729 }, { "epoch": 34.325, "grad_norm": 9.125, "learning_rate": 1.5280105503217226e-05, "loss": 1.3392, "step": 13730 }, { "epoch": 34.3275, "grad_norm": 9.1875, "learning_rate": 1.527941786754223e-05, "loss": 1.3423, "step": 13731 }, { "epoch": 34.33, "grad_norm": 7.6875, "learning_rate": 1.5278730197255578e-05, "loss": 1.2281, "step": 13732 }, { "epoch": 34.3325, "grad_norm": 7.90625, "learning_rate": 1.5278042492361787e-05, "loss": 1.3173, "step": 13733 }, { "epoch": 34.335, "grad_norm": 12.375, "learning_rate": 1.5277354752865355e-05, "loss": 1.436, "step": 13734 }, { "epoch": 34.3375, "grad_norm": 6.53125, "learning_rate": 1.52766669787708e-05, "loss": 1.33, "step": 13735 }, { "epoch": 34.34, "grad_norm": 7.03125, "learning_rate": 1.5275979170082627e-05, "loss": 1.235, "step": 13736 }, { "epoch": 34.3425, "grad_norm": 7.6875, "learning_rate": 1.5275291326805346e-05, "loss": 1.3289, "step": 13737 }, { "epoch": 34.345, "grad_norm": 6.5625, "learning_rate": 1.5274603448943467e-05, "loss": 1.3392, "step": 13738 }, { "epoch": 34.3475, "grad_norm": 7.625, "learning_rate": 1.5273915536501494e-05, "loss": 1.1768, "step": 13739 }, { "epoch": 34.35, "grad_norm": 7.625, "learning_rate": 1.5273227589483945e-05, "loss": 1.4131, "step": 13740 }, { "epoch": 34.3525, "grad_norm": 8.3125, "learning_rate": 1.527253960789533e-05, "loss": 1.3924, "step": 13741 }, { "epoch": 34.355, "grad_norm": 5.03125, "learning_rate": 1.5271851591740152e-05, "loss": 1.2302, "step": 13742 }, { "epoch": 34.3575, "grad_norm": 9.8125, "learning_rate": 1.5271163541022928e-05, "loss": 1.2673, "step": 13743 }, { "epoch": 34.36, "grad_norm": 6.40625, "learning_rate": 1.5270475455748165e-05, "loss": 1.3752, "step": 13744 }, { "epoch": 34.3625, "grad_norm": 6.25, "learning_rate": 1.5269787335920377e-05, "loss": 1.2651, "step": 13745 }, { "epoch": 34.365, "grad_norm": 6.8125, "learning_rate": 1.5269099181544077e-05, "loss": 1.1717, "step": 13746 }, { "epoch": 34.3675, "grad_norm": 6.15625, "learning_rate": 1.5268410992623767e-05, "loss": 1.3193, "step": 13747 }, { "epoch": 34.37, "grad_norm": 8.8125, "learning_rate": 1.526772276916397e-05, "loss": 1.4698, "step": 13748 }, { "epoch": 34.3725, "grad_norm": 7.0625, "learning_rate": 1.5267034511169196e-05, "loss": 1.2379, "step": 13749 }, { "epoch": 34.375, "grad_norm": 6.625, "learning_rate": 1.526634621864395e-05, "loss": 1.2737, "step": 13750 }, { "epoch": 34.3775, "grad_norm": 7.375, "learning_rate": 1.5265657891592746e-05, "loss": 1.2384, "step": 13751 }, { "epoch": 34.38, "grad_norm": 6.59375, "learning_rate": 1.5264969530020105e-05, "loss": 1.2904, "step": 13752 }, { "epoch": 34.3825, "grad_norm": 6.96875, "learning_rate": 1.5264281133930536e-05, "loss": 1.2758, "step": 13753 }, { "epoch": 34.385, "grad_norm": 5.75, "learning_rate": 1.526359270332854e-05, "loss": 1.3809, "step": 13754 }, { "epoch": 34.3875, "grad_norm": 6.9375, "learning_rate": 1.5262904238218652e-05, "loss": 1.346, "step": 13755 }, { "epoch": 34.39, "grad_norm": 16.375, "learning_rate": 1.526221573860537e-05, "loss": 1.3407, "step": 13756 }, { "epoch": 34.3925, "grad_norm": 7.625, "learning_rate": 1.5261527204493215e-05, "loss": 1.4092, "step": 13757 }, { "epoch": 34.395, "grad_norm": 6.5625, "learning_rate": 1.5260838635886694e-05, "loss": 1.365, "step": 13758 }, { "epoch": 34.3975, "grad_norm": 7.8125, "learning_rate": 1.526015003279033e-05, "loss": 1.3873, "step": 13759 }, { "epoch": 34.4, "grad_norm": 4.125, "learning_rate": 1.5259461395208628e-05, "loss": 1.321, "step": 13760 }, { "epoch": 34.4025, "grad_norm": 5.4375, "learning_rate": 1.5258772723146114e-05, "loss": 1.333, "step": 13761 }, { "epoch": 34.405, "grad_norm": 7.96875, "learning_rate": 1.5258084016607289e-05, "loss": 1.3601, "step": 13762 }, { "epoch": 34.4075, "grad_norm": 7.4375, "learning_rate": 1.5257395275596681e-05, "loss": 1.426, "step": 13763 }, { "epoch": 34.41, "grad_norm": 5.8125, "learning_rate": 1.5256706500118798e-05, "loss": 1.294, "step": 13764 }, { "epoch": 34.4125, "grad_norm": 6.59375, "learning_rate": 1.5256017690178157e-05, "loss": 1.3038, "step": 13765 }, { "epoch": 34.415, "grad_norm": 6.8125, "learning_rate": 1.5255328845779275e-05, "loss": 1.4205, "step": 13766 }, { "epoch": 34.4175, "grad_norm": 10.75, "learning_rate": 1.5254639966926668e-05, "loss": 1.4532, "step": 13767 }, { "epoch": 34.42, "grad_norm": 7.0625, "learning_rate": 1.5253951053624852e-05, "loss": 1.4466, "step": 13768 }, { "epoch": 34.4225, "grad_norm": 8.3125, "learning_rate": 1.5253262105878342e-05, "loss": 1.5191, "step": 13769 }, { "epoch": 34.425, "grad_norm": 7.15625, "learning_rate": 1.5252573123691654e-05, "loss": 1.4072, "step": 13770 }, { "epoch": 34.4275, "grad_norm": 13.3125, "learning_rate": 1.5251884107069308e-05, "loss": 1.3131, "step": 13771 }, { "epoch": 34.43, "grad_norm": 8.875, "learning_rate": 1.5251195056015822e-05, "loss": 1.4539, "step": 13772 }, { "epoch": 34.4325, "grad_norm": 5.75, "learning_rate": 1.5250505970535709e-05, "loss": 1.4653, "step": 13773 }, { "epoch": 34.435, "grad_norm": 8.25, "learning_rate": 1.5249816850633493e-05, "loss": 1.5028, "step": 13774 }, { "epoch": 34.4375, "grad_norm": 6.75, "learning_rate": 1.5249127696313682e-05, "loss": 1.2906, "step": 13775 }, { "epoch": 34.44, "grad_norm": 7.46875, "learning_rate": 1.5248438507580806e-05, "loss": 1.3842, "step": 13776 }, { "epoch": 34.4425, "grad_norm": 8.9375, "learning_rate": 1.524774928443937e-05, "loss": 1.605, "step": 13777 }, { "epoch": 34.445, "grad_norm": 7.3125, "learning_rate": 1.5247060026893907e-05, "loss": 1.3078, "step": 13778 }, { "epoch": 34.4475, "grad_norm": 10.0, "learning_rate": 1.5246370734948923e-05, "loss": 1.4174, "step": 13779 }, { "epoch": 34.45, "grad_norm": 10.1875, "learning_rate": 1.5245681408608946e-05, "loss": 1.3533, "step": 13780 }, { "epoch": 34.4525, "grad_norm": 7.5, "learning_rate": 1.524499204787849e-05, "loss": 1.2816, "step": 13781 }, { "epoch": 34.455, "grad_norm": 9.25, "learning_rate": 1.5244302652762074e-05, "loss": 1.3621, "step": 13782 }, { "epoch": 34.4575, "grad_norm": 8.1875, "learning_rate": 1.5243613223264224e-05, "loss": 1.2387, "step": 13783 }, { "epoch": 34.46, "grad_norm": 8.1875, "learning_rate": 1.5242923759389454e-05, "loss": 1.4441, "step": 13784 }, { "epoch": 34.4625, "grad_norm": 7.46875, "learning_rate": 1.5242234261142283e-05, "loss": 1.1864, "step": 13785 }, { "epoch": 34.465, "grad_norm": 7.3125, "learning_rate": 1.5241544728527236e-05, "loss": 1.3128, "step": 13786 }, { "epoch": 34.4675, "grad_norm": 7.34375, "learning_rate": 1.5240855161548833e-05, "loss": 1.3879, "step": 13787 }, { "epoch": 34.47, "grad_norm": 8.6875, "learning_rate": 1.524016556021159e-05, "loss": 1.4166, "step": 13788 }, { "epoch": 34.4725, "grad_norm": 10.375, "learning_rate": 1.5239475924520035e-05, "loss": 1.4355, "step": 13789 }, { "epoch": 34.475, "grad_norm": 7.84375, "learning_rate": 1.5238786254478681e-05, "loss": 1.2948, "step": 13790 }, { "epoch": 34.4775, "grad_norm": 7.34375, "learning_rate": 1.5238096550092061e-05, "loss": 1.2875, "step": 13791 }, { "epoch": 34.48, "grad_norm": 7.4375, "learning_rate": 1.5237406811364682e-05, "loss": 1.3316, "step": 13792 }, { "epoch": 34.4825, "grad_norm": 12.625, "learning_rate": 1.5236717038301079e-05, "loss": 1.4104, "step": 13793 }, { "epoch": 34.485, "grad_norm": 12.125, "learning_rate": 1.5236027230905766e-05, "loss": 1.3686, "step": 13794 }, { "epoch": 34.4875, "grad_norm": 7.53125, "learning_rate": 1.5235337389183266e-05, "loss": 1.3032, "step": 13795 }, { "epoch": 34.49, "grad_norm": 8.1875, "learning_rate": 1.5234647513138106e-05, "loss": 1.4368, "step": 13796 }, { "epoch": 34.4925, "grad_norm": 6.0, "learning_rate": 1.5233957602774808e-05, "loss": 1.2828, "step": 13797 }, { "epoch": 34.495, "grad_norm": 6.375, "learning_rate": 1.523326765809789e-05, "loss": 1.2703, "step": 13798 }, { "epoch": 34.4975, "grad_norm": 9.25, "learning_rate": 1.5232577679111882e-05, "loss": 1.3657, "step": 13799 }, { "epoch": 34.5, "grad_norm": 8.5625, "learning_rate": 1.52318876658213e-05, "loss": 1.3838, "step": 13800 }, { "epoch": 34.5025, "grad_norm": 8.1875, "learning_rate": 1.5231197618230677e-05, "loss": 1.394, "step": 13801 }, { "epoch": 34.505, "grad_norm": 6.15625, "learning_rate": 1.5230507536344527e-05, "loss": 1.42, "step": 13802 }, { "epoch": 34.5075, "grad_norm": 9.25, "learning_rate": 1.5229817420167386e-05, "loss": 1.3872, "step": 13803 }, { "epoch": 34.51, "grad_norm": 6.625, "learning_rate": 1.5229127269703763e-05, "loss": 1.4895, "step": 13804 }, { "epoch": 34.5125, "grad_norm": 7.09375, "learning_rate": 1.5228437084958193e-05, "loss": 1.3468, "step": 13805 }, { "epoch": 34.515, "grad_norm": 6.46875, "learning_rate": 1.5227746865935203e-05, "loss": 1.2916, "step": 13806 }, { "epoch": 34.5175, "grad_norm": 7.9375, "learning_rate": 1.5227056612639307e-05, "loss": 1.3806, "step": 13807 }, { "epoch": 34.52, "grad_norm": 6.8125, "learning_rate": 1.5226366325075042e-05, "loss": 1.3481, "step": 13808 }, { "epoch": 34.5225, "grad_norm": 7.84375, "learning_rate": 1.5225676003246928e-05, "loss": 1.2887, "step": 13809 }, { "epoch": 34.525, "grad_norm": 9.0625, "learning_rate": 1.5224985647159489e-05, "loss": 1.2081, "step": 13810 }, { "epoch": 34.5275, "grad_norm": 8.8125, "learning_rate": 1.5224295256817256e-05, "loss": 1.395, "step": 13811 }, { "epoch": 34.53, "grad_norm": 6.875, "learning_rate": 1.522360483222475e-05, "loss": 1.4517, "step": 13812 }, { "epoch": 34.5325, "grad_norm": 7.125, "learning_rate": 1.5222914373386502e-05, "loss": 1.3355, "step": 13813 }, { "epoch": 34.535, "grad_norm": 12.1875, "learning_rate": 1.5222223880307037e-05, "loss": 1.5611, "step": 13814 }, { "epoch": 34.5375, "grad_norm": 8.0625, "learning_rate": 1.5221533352990876e-05, "loss": 1.2901, "step": 13815 }, { "epoch": 34.54, "grad_norm": 6.125, "learning_rate": 1.5220842791442558e-05, "loss": 1.4228, "step": 13816 }, { "epoch": 34.5425, "grad_norm": 6.28125, "learning_rate": 1.5220152195666597e-05, "loss": 1.5, "step": 13817 }, { "epoch": 34.545, "grad_norm": 9.125, "learning_rate": 1.5219461565667532e-05, "loss": 1.4186, "step": 13818 }, { "epoch": 34.5475, "grad_norm": 7.9375, "learning_rate": 1.5218770901449885e-05, "loss": 1.3758, "step": 13819 }, { "epoch": 34.55, "grad_norm": 6.0625, "learning_rate": 1.5218080203018181e-05, "loss": 1.3169, "step": 13820 }, { "epoch": 34.5525, "grad_norm": 6.09375, "learning_rate": 1.5217389470376955e-05, "loss": 1.2746, "step": 13821 }, { "epoch": 34.555, "grad_norm": 7.28125, "learning_rate": 1.5216698703530735e-05, "loss": 1.176, "step": 13822 }, { "epoch": 34.5575, "grad_norm": 10.8125, "learning_rate": 1.5216007902484044e-05, "loss": 1.5986, "step": 13823 }, { "epoch": 34.56, "grad_norm": 9.3125, "learning_rate": 1.5215317067241415e-05, "loss": 1.358, "step": 13824 }, { "epoch": 34.5625, "grad_norm": 5.875, "learning_rate": 1.5214626197807373e-05, "loss": 1.2213, "step": 13825 }, { "epoch": 34.565, "grad_norm": 6.25, "learning_rate": 1.5213935294186453e-05, "loss": 1.2694, "step": 13826 }, { "epoch": 34.5675, "grad_norm": 6.84375, "learning_rate": 1.5213244356383184e-05, "loss": 1.3609, "step": 13827 }, { "epoch": 34.57, "grad_norm": 6.4375, "learning_rate": 1.5212553384402088e-05, "loss": 1.4295, "step": 13828 }, { "epoch": 34.5725, "grad_norm": 7.03125, "learning_rate": 1.5211862378247708e-05, "loss": 1.3867, "step": 13829 }, { "epoch": 34.575, "grad_norm": 5.59375, "learning_rate": 1.521117133792456e-05, "loss": 1.3112, "step": 13830 }, { "epoch": 34.5775, "grad_norm": 5.75, "learning_rate": 1.5210480263437184e-05, "loss": 1.2435, "step": 13831 }, { "epoch": 34.58, "grad_norm": 9.0625, "learning_rate": 1.5209789154790107e-05, "loss": 1.4561, "step": 13832 }, { "epoch": 34.5825, "grad_norm": 5.6875, "learning_rate": 1.5209098011987862e-05, "loss": 1.2483, "step": 13833 }, { "epoch": 34.585, "grad_norm": 6.125, "learning_rate": 1.520840683503498e-05, "loss": 1.3506, "step": 13834 }, { "epoch": 34.5875, "grad_norm": 7.71875, "learning_rate": 1.5207715623935988e-05, "loss": 1.3311, "step": 13835 }, { "epoch": 34.59, "grad_norm": 6.5, "learning_rate": 1.5207024378695423e-05, "loss": 1.385, "step": 13836 }, { "epoch": 34.5925, "grad_norm": 8.25, "learning_rate": 1.5206333099317814e-05, "loss": 1.4481, "step": 13837 }, { "epoch": 34.595, "grad_norm": 9.6875, "learning_rate": 1.5205641785807693e-05, "loss": 1.3666, "step": 13838 }, { "epoch": 34.5975, "grad_norm": 6.78125, "learning_rate": 1.5204950438169594e-05, "loss": 1.2422, "step": 13839 }, { "epoch": 34.6, "grad_norm": 7.15625, "learning_rate": 1.5204259056408046e-05, "loss": 1.2508, "step": 13840 }, { "epoch": 34.6025, "grad_norm": 6.25, "learning_rate": 1.5203567640527586e-05, "loss": 1.3531, "step": 13841 }, { "epoch": 34.605, "grad_norm": 8.3125, "learning_rate": 1.5202876190532746e-05, "loss": 1.3323, "step": 13842 }, { "epoch": 34.6075, "grad_norm": 9.9375, "learning_rate": 1.5202184706428056e-05, "loss": 1.3859, "step": 13843 }, { "epoch": 34.61, "grad_norm": 4.65625, "learning_rate": 1.520149318821805e-05, "loss": 1.4167, "step": 13844 }, { "epoch": 34.6125, "grad_norm": 7.03125, "learning_rate": 1.5200801635907266e-05, "loss": 1.2171, "step": 13845 }, { "epoch": 34.615, "grad_norm": 7.5625, "learning_rate": 1.5200110049500232e-05, "loss": 1.2059, "step": 13846 }, { "epoch": 34.6175, "grad_norm": 9.125, "learning_rate": 1.5199418429001483e-05, "loss": 1.4182, "step": 13847 }, { "epoch": 34.62, "grad_norm": 6.53125, "learning_rate": 1.519872677441556e-05, "loss": 1.3428, "step": 13848 }, { "epoch": 34.6225, "grad_norm": 7.0, "learning_rate": 1.519803508574699e-05, "loss": 1.3559, "step": 13849 }, { "epoch": 34.625, "grad_norm": 8.375, "learning_rate": 1.5197343363000308e-05, "loss": 1.4928, "step": 13850 }, { "epoch": 34.6275, "grad_norm": 6.1875, "learning_rate": 1.5196651606180053e-05, "loss": 1.2373, "step": 13851 }, { "epoch": 34.63, "grad_norm": 8.5, "learning_rate": 1.5195959815290758e-05, "loss": 1.4026, "step": 13852 }, { "epoch": 34.6325, "grad_norm": 6.78125, "learning_rate": 1.5195267990336957e-05, "loss": 1.2763, "step": 13853 }, { "epoch": 34.635, "grad_norm": 8.3125, "learning_rate": 1.5194576131323188e-05, "loss": 1.372, "step": 13854 }, { "epoch": 34.6375, "grad_norm": 9.9375, "learning_rate": 1.5193884238253985e-05, "loss": 1.4705, "step": 13855 }, { "epoch": 34.64, "grad_norm": 5.28125, "learning_rate": 1.5193192311133884e-05, "loss": 1.3138, "step": 13856 }, { "epoch": 34.6425, "grad_norm": 6.03125, "learning_rate": 1.5192500349967424e-05, "loss": 1.2014, "step": 13857 }, { "epoch": 34.645, "grad_norm": 8.9375, "learning_rate": 1.519180835475914e-05, "loss": 1.5109, "step": 13858 }, { "epoch": 34.6475, "grad_norm": 10.0, "learning_rate": 1.5191116325513569e-05, "loss": 1.2739, "step": 13859 }, { "epoch": 34.65, "grad_norm": 7.4375, "learning_rate": 1.5190424262235241e-05, "loss": 1.2837, "step": 13860 }, { "epoch": 34.6525, "grad_norm": 9.25, "learning_rate": 1.5189732164928706e-05, "loss": 1.4983, "step": 13861 }, { "epoch": 34.655, "grad_norm": 9.25, "learning_rate": 1.518904003359849e-05, "loss": 1.1976, "step": 13862 }, { "epoch": 34.6575, "grad_norm": 7.5625, "learning_rate": 1.5188347868249137e-05, "loss": 1.2791, "step": 13863 }, { "epoch": 34.66, "grad_norm": 9.5625, "learning_rate": 1.5187655668885187e-05, "loss": 1.3508, "step": 13864 }, { "epoch": 34.6625, "grad_norm": 5.5, "learning_rate": 1.518696343551117e-05, "loss": 1.3498, "step": 13865 }, { "epoch": 34.665, "grad_norm": 7.125, "learning_rate": 1.5186271168131628e-05, "loss": 1.4105, "step": 13866 }, { "epoch": 34.6675, "grad_norm": 9.75, "learning_rate": 1.5185578866751099e-05, "loss": 1.4925, "step": 13867 }, { "epoch": 34.67, "grad_norm": 9.625, "learning_rate": 1.5184886531374123e-05, "loss": 1.4245, "step": 13868 }, { "epoch": 34.6725, "grad_norm": 9.875, "learning_rate": 1.518419416200524e-05, "loss": 1.5153, "step": 13869 }, { "epoch": 34.675, "grad_norm": 7.3125, "learning_rate": 1.5183501758648987e-05, "loss": 1.2732, "step": 13870 }, { "epoch": 34.6775, "grad_norm": 8.125, "learning_rate": 1.5182809321309903e-05, "loss": 1.4238, "step": 13871 }, { "epoch": 34.68, "grad_norm": 7.5, "learning_rate": 1.5182116849992528e-05, "loss": 1.3334, "step": 13872 }, { "epoch": 34.6825, "grad_norm": 7.0, "learning_rate": 1.5181424344701404e-05, "loss": 1.5107, "step": 13873 }, { "epoch": 34.685, "grad_norm": 5.90625, "learning_rate": 1.5180731805441066e-05, "loss": 1.3565, "step": 13874 }, { "epoch": 34.6875, "grad_norm": 6.40625, "learning_rate": 1.5180039232216062e-05, "loss": 1.2522, "step": 13875 }, { "epoch": 34.69, "grad_norm": 6.59375, "learning_rate": 1.5179346625030929e-05, "loss": 1.3705, "step": 13876 }, { "epoch": 34.6925, "grad_norm": 5.59375, "learning_rate": 1.51786539838902e-05, "loss": 1.3918, "step": 13877 }, { "epoch": 34.695, "grad_norm": 9.8125, "learning_rate": 1.517796130879843e-05, "loss": 1.3447, "step": 13878 }, { "epoch": 34.6975, "grad_norm": 10.6875, "learning_rate": 1.5177268599760147e-05, "loss": 1.4296, "step": 13879 }, { "epoch": 34.7, "grad_norm": 8.625, "learning_rate": 1.5176575856779904e-05, "loss": 1.2812, "step": 13880 }, { "epoch": 34.7025, "grad_norm": 7.6875, "learning_rate": 1.5175883079862232e-05, "loss": 1.2223, "step": 13881 }, { "epoch": 34.705, "grad_norm": 7.0, "learning_rate": 1.5175190269011682e-05, "loss": 1.3092, "step": 13882 }, { "epoch": 34.7075, "grad_norm": 7.25, "learning_rate": 1.5174497424232788e-05, "loss": 1.2619, "step": 13883 }, { "epoch": 34.71, "grad_norm": 5.75, "learning_rate": 1.5173804545530097e-05, "loss": 1.3649, "step": 13884 }, { "epoch": 34.7125, "grad_norm": 6.875, "learning_rate": 1.517311163290815e-05, "loss": 1.4059, "step": 13885 }, { "epoch": 34.715, "grad_norm": 10.0, "learning_rate": 1.5172418686371493e-05, "loss": 1.3088, "step": 13886 }, { "epoch": 34.7175, "grad_norm": 7.03125, "learning_rate": 1.5171725705924662e-05, "loss": 1.4681, "step": 13887 }, { "epoch": 34.72, "grad_norm": 4.75, "learning_rate": 1.5171032691572207e-05, "loss": 1.3119, "step": 13888 }, { "epoch": 34.7225, "grad_norm": 7.78125, "learning_rate": 1.5170339643318665e-05, "loss": 1.4294, "step": 13889 }, { "epoch": 34.725, "grad_norm": 4.9375, "learning_rate": 1.516964656116859e-05, "loss": 1.3632, "step": 13890 }, { "epoch": 34.7275, "grad_norm": 7.25, "learning_rate": 1.5168953445126513e-05, "loss": 1.4327, "step": 13891 }, { "epoch": 34.73, "grad_norm": 5.1875, "learning_rate": 1.5168260295196988e-05, "loss": 1.3807, "step": 13892 }, { "epoch": 34.7325, "grad_norm": 6.40625, "learning_rate": 1.5167567111384551e-05, "loss": 1.3252, "step": 13893 }, { "epoch": 34.735, "grad_norm": 7.0625, "learning_rate": 1.5166873893693754e-05, "loss": 1.3218, "step": 13894 }, { "epoch": 34.7375, "grad_norm": 7.40625, "learning_rate": 1.5166180642129136e-05, "loss": 1.4317, "step": 13895 }, { "epoch": 34.74, "grad_norm": 7.1875, "learning_rate": 1.5165487356695247e-05, "loss": 1.4232, "step": 13896 }, { "epoch": 34.7425, "grad_norm": 7.34375, "learning_rate": 1.5164794037396628e-05, "loss": 1.2901, "step": 13897 }, { "epoch": 34.745, "grad_norm": 7.59375, "learning_rate": 1.5164100684237829e-05, "loss": 1.3296, "step": 13898 }, { "epoch": 34.7475, "grad_norm": 10.625, "learning_rate": 1.5163407297223387e-05, "loss": 1.404, "step": 13899 }, { "epoch": 34.75, "grad_norm": 9.375, "learning_rate": 1.516271387635786e-05, "loss": 1.4382, "step": 13900 }, { "epoch": 34.7525, "grad_norm": 10.1875, "learning_rate": 1.5162020421645784e-05, "loss": 1.3727, "step": 13901 }, { "epoch": 34.755, "grad_norm": 7.65625, "learning_rate": 1.5161326933091708e-05, "loss": 1.3903, "step": 13902 }, { "epoch": 34.7575, "grad_norm": 8.75, "learning_rate": 1.5160633410700181e-05, "loss": 1.3945, "step": 13903 }, { "epoch": 34.76, "grad_norm": 10.3125, "learning_rate": 1.5159939854475743e-05, "loss": 1.4338, "step": 13904 }, { "epoch": 34.7625, "grad_norm": 6.84375, "learning_rate": 1.515924626442295e-05, "loss": 1.2441, "step": 13905 }, { "epoch": 34.765, "grad_norm": 7.125, "learning_rate": 1.5158552640546347e-05, "loss": 1.2581, "step": 13906 }, { "epoch": 34.7675, "grad_norm": 8.0, "learning_rate": 1.5157858982850475e-05, "loss": 1.3502, "step": 13907 }, { "epoch": 34.77, "grad_norm": 8.8125, "learning_rate": 1.515716529133989e-05, "loss": 1.3682, "step": 13908 }, { "epoch": 34.7725, "grad_norm": 6.625, "learning_rate": 1.5156471566019134e-05, "loss": 1.2934, "step": 13909 }, { "epoch": 34.775, "grad_norm": 8.4375, "learning_rate": 1.5155777806892756e-05, "loss": 1.3308, "step": 13910 }, { "epoch": 34.7775, "grad_norm": 12.0, "learning_rate": 1.5155084013965304e-05, "loss": 1.454, "step": 13911 }, { "epoch": 34.78, "grad_norm": 9.75, "learning_rate": 1.5154390187241328e-05, "loss": 1.435, "step": 13912 }, { "epoch": 34.7825, "grad_norm": 5.96875, "learning_rate": 1.5153696326725378e-05, "loss": 1.2726, "step": 13913 }, { "epoch": 34.785, "grad_norm": 9.0625, "learning_rate": 1.5153002432422e-05, "loss": 1.2246, "step": 13914 }, { "epoch": 34.7875, "grad_norm": 9.125, "learning_rate": 1.5152308504335742e-05, "loss": 1.2521, "step": 13915 }, { "epoch": 34.79, "grad_norm": 12.625, "learning_rate": 1.515161454247116e-05, "loss": 1.5105, "step": 13916 }, { "epoch": 34.7925, "grad_norm": 10.125, "learning_rate": 1.5150920546832798e-05, "loss": 1.4538, "step": 13917 }, { "epoch": 34.795, "grad_norm": 6.5, "learning_rate": 1.5150226517425206e-05, "loss": 1.3764, "step": 13918 }, { "epoch": 34.7975, "grad_norm": 6.28125, "learning_rate": 1.5149532454252933e-05, "loss": 1.29, "step": 13919 }, { "epoch": 34.8, "grad_norm": 7.53125, "learning_rate": 1.5148838357320537e-05, "loss": 1.3218, "step": 13920 }, { "epoch": 34.8025, "grad_norm": 10.5625, "learning_rate": 1.5148144226632559e-05, "loss": 1.4563, "step": 13921 }, { "epoch": 34.805, "grad_norm": 7.75, "learning_rate": 1.5147450062193554e-05, "loss": 1.2626, "step": 13922 }, { "epoch": 34.8075, "grad_norm": 7.875, "learning_rate": 1.5146755864008071e-05, "loss": 1.2758, "step": 13923 }, { "epoch": 34.81, "grad_norm": 9.8125, "learning_rate": 1.5146061632080665e-05, "loss": 1.4477, "step": 13924 }, { "epoch": 34.8125, "grad_norm": 8.9375, "learning_rate": 1.5145367366415882e-05, "loss": 1.4188, "step": 13925 }, { "epoch": 34.815, "grad_norm": 7.125, "learning_rate": 1.5144673067018279e-05, "loss": 1.4061, "step": 13926 }, { "epoch": 34.8175, "grad_norm": 7.40625, "learning_rate": 1.5143978733892404e-05, "loss": 1.5118, "step": 13927 }, { "epoch": 34.82, "grad_norm": 5.6875, "learning_rate": 1.5143284367042809e-05, "loss": 1.4314, "step": 13928 }, { "epoch": 34.8225, "grad_norm": 7.25, "learning_rate": 1.514258996647405e-05, "loss": 1.3146, "step": 13929 }, { "epoch": 34.825, "grad_norm": 9.0, "learning_rate": 1.5141895532190677e-05, "loss": 1.4395, "step": 13930 }, { "epoch": 34.8275, "grad_norm": 16.125, "learning_rate": 1.5141201064197241e-05, "loss": 1.4286, "step": 13931 }, { "epoch": 34.83, "grad_norm": 7.15625, "learning_rate": 1.5140506562498297e-05, "loss": 1.3498, "step": 13932 }, { "epoch": 34.8325, "grad_norm": 7.15625, "learning_rate": 1.51398120270984e-05, "loss": 1.5686, "step": 13933 }, { "epoch": 34.835, "grad_norm": 7.0625, "learning_rate": 1.5139117458002096e-05, "loss": 1.3455, "step": 13934 }, { "epoch": 34.8375, "grad_norm": 8.3125, "learning_rate": 1.513842285521395e-05, "loss": 1.4128, "step": 13935 }, { "epoch": 34.84, "grad_norm": 7.71875, "learning_rate": 1.5137728218738504e-05, "loss": 1.3959, "step": 13936 }, { "epoch": 34.8425, "grad_norm": 7.25, "learning_rate": 1.513703354858032e-05, "loss": 1.1833, "step": 13937 }, { "epoch": 34.845, "grad_norm": 7.875, "learning_rate": 1.5136338844743947e-05, "loss": 1.2474, "step": 13938 }, { "epoch": 34.8475, "grad_norm": 6.28125, "learning_rate": 1.5135644107233946e-05, "loss": 1.5194, "step": 13939 }, { "epoch": 34.85, "grad_norm": 12.9375, "learning_rate": 1.5134949336054866e-05, "loss": 1.3464, "step": 13940 }, { "epoch": 34.8525, "grad_norm": 6.8125, "learning_rate": 1.5134254531211264e-05, "loss": 1.2999, "step": 13941 }, { "epoch": 34.855, "grad_norm": 8.375, "learning_rate": 1.5133559692707693e-05, "loss": 1.2842, "step": 13942 }, { "epoch": 34.8575, "grad_norm": 9.4375, "learning_rate": 1.5132864820548712e-05, "loss": 1.3344, "step": 13943 }, { "epoch": 34.86, "grad_norm": 6.78125, "learning_rate": 1.5132169914738875e-05, "loss": 1.469, "step": 13944 }, { "epoch": 34.8625, "grad_norm": 7.78125, "learning_rate": 1.5131474975282734e-05, "loss": 1.3469, "step": 13945 }, { "epoch": 34.865, "grad_norm": 7.6875, "learning_rate": 1.5130780002184853e-05, "loss": 1.3708, "step": 13946 }, { "epoch": 34.8675, "grad_norm": 8.25, "learning_rate": 1.5130084995449778e-05, "loss": 1.2832, "step": 13947 }, { "epoch": 34.87, "grad_norm": 6.4375, "learning_rate": 1.5129389955082077e-05, "loss": 1.4118, "step": 13948 }, { "epoch": 34.8725, "grad_norm": 6.625, "learning_rate": 1.5128694881086296e-05, "loss": 1.332, "step": 13949 }, { "epoch": 34.875, "grad_norm": 5.59375, "learning_rate": 1.5127999773467001e-05, "loss": 1.3869, "step": 13950 }, { "epoch": 34.8775, "grad_norm": 8.625, "learning_rate": 1.5127304632228744e-05, "loss": 1.3044, "step": 13951 }, { "epoch": 34.88, "grad_norm": 9.125, "learning_rate": 1.512660945737608e-05, "loss": 1.3959, "step": 13952 }, { "epoch": 34.8825, "grad_norm": 10.625, "learning_rate": 1.5125914248913571e-05, "loss": 1.3853, "step": 13953 }, { "epoch": 34.885, "grad_norm": 8.75, "learning_rate": 1.5125219006845772e-05, "loss": 1.2828, "step": 13954 }, { "epoch": 34.8875, "grad_norm": 7.21875, "learning_rate": 1.5124523731177245e-05, "loss": 1.3998, "step": 13955 }, { "epoch": 34.89, "grad_norm": 8.4375, "learning_rate": 1.5123828421912545e-05, "loss": 1.4192, "step": 13956 }, { "epoch": 34.8925, "grad_norm": 7.28125, "learning_rate": 1.5123133079056229e-05, "loss": 1.2903, "step": 13957 }, { "epoch": 34.895, "grad_norm": 6.40625, "learning_rate": 1.512243770261286e-05, "loss": 1.4008, "step": 13958 }, { "epoch": 34.8975, "grad_norm": 7.53125, "learning_rate": 1.5121742292586992e-05, "loss": 1.2834, "step": 13959 }, { "epoch": 34.9, "grad_norm": 7.625, "learning_rate": 1.512104684898319e-05, "loss": 1.3008, "step": 13960 }, { "epoch": 34.9025, "grad_norm": 7.40625, "learning_rate": 1.5120351371806006e-05, "loss": 1.1776, "step": 13961 }, { "epoch": 34.905, "grad_norm": 7.84375, "learning_rate": 1.5119655861060003e-05, "loss": 1.4516, "step": 13962 }, { "epoch": 34.9075, "grad_norm": 6.40625, "learning_rate": 1.5118960316749745e-05, "loss": 1.5709, "step": 13963 }, { "epoch": 34.91, "grad_norm": 18.625, "learning_rate": 1.5118264738879785e-05, "loss": 1.3316, "step": 13964 }, { "epoch": 34.9125, "grad_norm": 9.5, "learning_rate": 1.5117569127454688e-05, "loss": 1.3147, "step": 13965 }, { "epoch": 34.915, "grad_norm": 7.65625, "learning_rate": 1.511687348247901e-05, "loss": 1.4191, "step": 13966 }, { "epoch": 34.9175, "grad_norm": 5.75, "learning_rate": 1.5116177803957315e-05, "loss": 1.452, "step": 13967 }, { "epoch": 34.92, "grad_norm": 5.4375, "learning_rate": 1.5115482091894164e-05, "loss": 1.3637, "step": 13968 }, { "epoch": 34.9225, "grad_norm": 5.9375, "learning_rate": 1.5114786346294118e-05, "loss": 1.2922, "step": 13969 }, { "epoch": 34.925, "grad_norm": 6.40625, "learning_rate": 1.5114090567161734e-05, "loss": 1.3099, "step": 13970 }, { "epoch": 34.9275, "grad_norm": 7.875, "learning_rate": 1.5113394754501581e-05, "loss": 1.5531, "step": 13971 }, { "epoch": 34.93, "grad_norm": 7.9375, "learning_rate": 1.5112698908318213e-05, "loss": 1.3569, "step": 13972 }, { "epoch": 34.9325, "grad_norm": 7.96875, "learning_rate": 1.51120030286162e-05, "loss": 1.4139, "step": 13973 }, { "epoch": 34.935, "grad_norm": 7.59375, "learning_rate": 1.5111307115400094e-05, "loss": 1.5768, "step": 13974 }, { "epoch": 34.9375, "grad_norm": 7.90625, "learning_rate": 1.5110611168674467e-05, "loss": 1.2583, "step": 13975 }, { "epoch": 34.94, "grad_norm": 7.3125, "learning_rate": 1.5109915188443877e-05, "loss": 1.2611, "step": 13976 }, { "epoch": 34.9425, "grad_norm": 7.0, "learning_rate": 1.5109219174712888e-05, "loss": 1.4458, "step": 13977 }, { "epoch": 34.945, "grad_norm": 4.53125, "learning_rate": 1.510852312748606e-05, "loss": 1.315, "step": 13978 }, { "epoch": 34.9475, "grad_norm": 7.71875, "learning_rate": 1.5107827046767962e-05, "loss": 1.2925, "step": 13979 }, { "epoch": 34.95, "grad_norm": 6.0625, "learning_rate": 1.5107130932563151e-05, "loss": 1.2006, "step": 13980 }, { "epoch": 34.9525, "grad_norm": 6.03125, "learning_rate": 1.5106434784876199e-05, "loss": 1.3902, "step": 13981 }, { "epoch": 34.955, "grad_norm": 8.375, "learning_rate": 1.5105738603711661e-05, "loss": 1.2594, "step": 13982 }, { "epoch": 34.9575, "grad_norm": 7.6875, "learning_rate": 1.5105042389074105e-05, "loss": 1.287, "step": 13983 }, { "epoch": 34.96, "grad_norm": 6.9375, "learning_rate": 1.5104346140968096e-05, "loss": 1.584, "step": 13984 }, { "epoch": 34.9625, "grad_norm": 8.8125, "learning_rate": 1.51036498593982e-05, "loss": 1.1896, "step": 13985 }, { "epoch": 34.965, "grad_norm": 9.1875, "learning_rate": 1.5102953544368979e-05, "loss": 1.3807, "step": 13986 }, { "epoch": 34.9675, "grad_norm": 7.625, "learning_rate": 1.5102257195884997e-05, "loss": 1.256, "step": 13987 }, { "epoch": 34.97, "grad_norm": 7.8125, "learning_rate": 1.5101560813950822e-05, "loss": 1.3661, "step": 13988 }, { "epoch": 34.9725, "grad_norm": 9.1875, "learning_rate": 1.5100864398571017e-05, "loss": 1.447, "step": 13989 }, { "epoch": 34.975, "grad_norm": 6.09375, "learning_rate": 1.510016794975015e-05, "loss": 1.3732, "step": 13990 }, { "epoch": 34.9775, "grad_norm": 8.625, "learning_rate": 1.5099471467492788e-05, "loss": 1.2822, "step": 13991 }, { "epoch": 34.98, "grad_norm": 5.46875, "learning_rate": 1.5098774951803492e-05, "loss": 1.2687, "step": 13992 }, { "epoch": 34.9825, "grad_norm": 8.4375, "learning_rate": 1.5098078402686836e-05, "loss": 1.3133, "step": 13993 }, { "epoch": 34.985, "grad_norm": 7.5625, "learning_rate": 1.5097381820147376e-05, "loss": 1.2615, "step": 13994 }, { "epoch": 34.9875, "grad_norm": 7.34375, "learning_rate": 1.5096685204189689e-05, "loss": 1.49, "step": 13995 }, { "epoch": 34.99, "grad_norm": 8.1875, "learning_rate": 1.5095988554818335e-05, "loss": 1.2056, "step": 13996 }, { "epoch": 34.9925, "grad_norm": 11.0, "learning_rate": 1.5095291872037886e-05, "loss": 1.3601, "step": 13997 }, { "epoch": 34.995, "grad_norm": 8.3125, "learning_rate": 1.5094595155852908e-05, "loss": 1.3635, "step": 13998 }, { "epoch": 34.9975, "grad_norm": 7.71875, "learning_rate": 1.5093898406267966e-05, "loss": 1.3539, "step": 13999 }, { "epoch": 35.0, "grad_norm": 7.625, "learning_rate": 1.5093201623287631e-05, "loss": 1.3588, "step": 14000 }, { "epoch": 35.0025, "grad_norm": 4.78125, "learning_rate": 1.5092504806916468e-05, "loss": 1.4091, "step": 14001 }, { "epoch": 35.005, "grad_norm": 7.3125, "learning_rate": 1.509180795715905e-05, "loss": 1.2483, "step": 14002 }, { "epoch": 35.0075, "grad_norm": 6.9375, "learning_rate": 1.5091111074019945e-05, "loss": 1.5076, "step": 14003 }, { "epoch": 35.01, "grad_norm": 7.96875, "learning_rate": 1.5090414157503715e-05, "loss": 1.2508, "step": 14004 }, { "epoch": 35.0125, "grad_norm": 5.875, "learning_rate": 1.5089717207614935e-05, "loss": 1.4086, "step": 14005 }, { "epoch": 35.015, "grad_norm": 7.59375, "learning_rate": 1.508902022435817e-05, "loss": 1.4257, "step": 14006 }, { "epoch": 35.0175, "grad_norm": 7.5625, "learning_rate": 1.5088323207737995e-05, "loss": 1.3044, "step": 14007 }, { "epoch": 35.02, "grad_norm": 6.1875, "learning_rate": 1.5087626157758979e-05, "loss": 1.4035, "step": 14008 }, { "epoch": 35.0225, "grad_norm": 7.1875, "learning_rate": 1.5086929074425687e-05, "loss": 1.397, "step": 14009 }, { "epoch": 35.025, "grad_norm": 7.5, "learning_rate": 1.5086231957742689e-05, "loss": 1.3366, "step": 14010 }, { "epoch": 35.0275, "grad_norm": 9.875, "learning_rate": 1.5085534807714562e-05, "loss": 1.434, "step": 14011 }, { "epoch": 35.03, "grad_norm": 6.4375, "learning_rate": 1.5084837624345869e-05, "loss": 1.2795, "step": 14012 }, { "epoch": 35.0325, "grad_norm": 7.5, "learning_rate": 1.5084140407641185e-05, "loss": 1.6179, "step": 14013 }, { "epoch": 35.035, "grad_norm": 10.3125, "learning_rate": 1.5083443157605082e-05, "loss": 1.3408, "step": 14014 }, { "epoch": 35.0375, "grad_norm": 6.53125, "learning_rate": 1.5082745874242126e-05, "loss": 1.4454, "step": 14015 }, { "epoch": 35.04, "grad_norm": 14.0, "learning_rate": 1.5082048557556892e-05, "loss": 1.2919, "step": 14016 }, { "epoch": 35.0425, "grad_norm": 5.6875, "learning_rate": 1.5081351207553953e-05, "loss": 1.393, "step": 14017 }, { "epoch": 35.045, "grad_norm": 7.71875, "learning_rate": 1.508065382423788e-05, "loss": 1.3211, "step": 14018 }, { "epoch": 35.0475, "grad_norm": 7.78125, "learning_rate": 1.507995640761324e-05, "loss": 1.4592, "step": 14019 }, { "epoch": 35.05, "grad_norm": 15.6875, "learning_rate": 1.507925895768461e-05, "loss": 1.3066, "step": 14020 }, { "epoch": 35.0525, "grad_norm": 8.3125, "learning_rate": 1.507856147445656e-05, "loss": 1.3848, "step": 14021 }, { "epoch": 35.055, "grad_norm": 6.84375, "learning_rate": 1.5077863957933669e-05, "loss": 1.2539, "step": 14022 }, { "epoch": 35.0575, "grad_norm": 8.375, "learning_rate": 1.5077166408120505e-05, "loss": 1.3381, "step": 14023 }, { "epoch": 35.06, "grad_norm": 10.125, "learning_rate": 1.5076468825021637e-05, "loss": 1.3411, "step": 14024 }, { "epoch": 35.0625, "grad_norm": 8.875, "learning_rate": 1.5075771208641645e-05, "loss": 1.4002, "step": 14025 }, { "epoch": 35.065, "grad_norm": 4.59375, "learning_rate": 1.5075073558985098e-05, "loss": 1.2817, "step": 14026 }, { "epoch": 35.0675, "grad_norm": 6.875, "learning_rate": 1.5074375876056574e-05, "loss": 1.2948, "step": 14027 }, { "epoch": 35.07, "grad_norm": 7.375, "learning_rate": 1.5073678159860646e-05, "loss": 1.2044, "step": 14028 }, { "epoch": 35.0725, "grad_norm": 7.84375, "learning_rate": 1.5072980410401883e-05, "loss": 1.3547, "step": 14029 }, { "epoch": 35.075, "grad_norm": 11.0, "learning_rate": 1.5072282627684867e-05, "loss": 1.2816, "step": 14030 }, { "epoch": 35.0775, "grad_norm": 7.1875, "learning_rate": 1.5071584811714166e-05, "loss": 1.4106, "step": 14031 }, { "epoch": 35.08, "grad_norm": 6.6875, "learning_rate": 1.507088696249436e-05, "loss": 1.3481, "step": 14032 }, { "epoch": 35.0825, "grad_norm": 8.0625, "learning_rate": 1.5070189080030023e-05, "loss": 1.2768, "step": 14033 }, { "epoch": 35.085, "grad_norm": 6.96875, "learning_rate": 1.5069491164325727e-05, "loss": 1.3517, "step": 14034 }, { "epoch": 35.0875, "grad_norm": 8.25, "learning_rate": 1.506879321538605e-05, "loss": 1.587, "step": 14035 }, { "epoch": 35.09, "grad_norm": 8.0, "learning_rate": 1.5068095233215569e-05, "loss": 1.4412, "step": 14036 }, { "epoch": 35.0925, "grad_norm": 8.5, "learning_rate": 1.5067397217818858e-05, "loss": 1.4379, "step": 14037 }, { "epoch": 35.095, "grad_norm": 6.09375, "learning_rate": 1.5066699169200494e-05, "loss": 1.3698, "step": 14038 }, { "epoch": 35.0975, "grad_norm": 7.40625, "learning_rate": 1.506600108736505e-05, "loss": 1.4166, "step": 14039 }, { "epoch": 35.1, "grad_norm": 8.25, "learning_rate": 1.5065302972317108e-05, "loss": 1.2578, "step": 14040 }, { "epoch": 35.1025, "grad_norm": 7.375, "learning_rate": 1.5064604824061242e-05, "loss": 1.4567, "step": 14041 }, { "epoch": 35.105, "grad_norm": 5.5, "learning_rate": 1.5063906642602031e-05, "loss": 1.4112, "step": 14042 }, { "epoch": 35.1075, "grad_norm": 13.5625, "learning_rate": 1.5063208427944052e-05, "loss": 1.2751, "step": 14043 }, { "epoch": 35.11, "grad_norm": 10.0, "learning_rate": 1.5062510180091874e-05, "loss": 1.4115, "step": 14044 }, { "epoch": 35.1125, "grad_norm": 7.5625, "learning_rate": 1.506181189905009e-05, "loss": 1.3699, "step": 14045 }, { "epoch": 35.115, "grad_norm": 7.59375, "learning_rate": 1.5061113584823264e-05, "loss": 1.3165, "step": 14046 }, { "epoch": 35.1175, "grad_norm": 5.875, "learning_rate": 1.5060415237415985e-05, "loss": 1.1896, "step": 14047 }, { "epoch": 35.12, "grad_norm": 4.90625, "learning_rate": 1.505971685683282e-05, "loss": 1.4079, "step": 14048 }, { "epoch": 35.1225, "grad_norm": 5.90625, "learning_rate": 1.505901844307836e-05, "loss": 1.3375, "step": 14049 }, { "epoch": 35.125, "grad_norm": 9.625, "learning_rate": 1.5058319996157172e-05, "loss": 1.2708, "step": 14050 }, { "epoch": 35.1275, "grad_norm": 9.8125, "learning_rate": 1.5057621516073845e-05, "loss": 1.3238, "step": 14051 }, { "epoch": 35.13, "grad_norm": 6.75, "learning_rate": 1.505692300283295e-05, "loss": 1.2942, "step": 14052 }, { "epoch": 35.1325, "grad_norm": 6.84375, "learning_rate": 1.5056224456439071e-05, "loss": 1.378, "step": 14053 }, { "epoch": 35.135, "grad_norm": 6.65625, "learning_rate": 1.5055525876896788e-05, "loss": 1.3679, "step": 14054 }, { "epoch": 35.1375, "grad_norm": 6.90625, "learning_rate": 1.505482726421068e-05, "loss": 1.3414, "step": 14055 }, { "epoch": 35.14, "grad_norm": 8.5, "learning_rate": 1.5054128618385324e-05, "loss": 1.4027, "step": 14056 }, { "epoch": 35.1425, "grad_norm": 9.625, "learning_rate": 1.5053429939425309e-05, "loss": 1.4072, "step": 14057 }, { "epoch": 35.145, "grad_norm": 10.0, "learning_rate": 1.5052731227335207e-05, "loss": 1.2496, "step": 14058 }, { "epoch": 35.1475, "grad_norm": 9.8125, "learning_rate": 1.5052032482119597e-05, "loss": 1.4549, "step": 14059 }, { "epoch": 35.15, "grad_norm": 9.5, "learning_rate": 1.5051333703783069e-05, "loss": 1.3308, "step": 14060 }, { "epoch": 35.1525, "grad_norm": 7.9375, "learning_rate": 1.5050634892330193e-05, "loss": 1.2946, "step": 14061 }, { "epoch": 35.155, "grad_norm": 7.625, "learning_rate": 1.5049936047765565e-05, "loss": 1.4445, "step": 14062 }, { "epoch": 35.1575, "grad_norm": 9.6875, "learning_rate": 1.5049237170093753e-05, "loss": 1.4079, "step": 14063 }, { "epoch": 35.16, "grad_norm": 6.4375, "learning_rate": 1.5048538259319347e-05, "loss": 1.5159, "step": 14064 }, { "epoch": 35.1625, "grad_norm": 6.09375, "learning_rate": 1.5047839315446923e-05, "loss": 1.2963, "step": 14065 }, { "epoch": 35.165, "grad_norm": 5.34375, "learning_rate": 1.504714033848107e-05, "loss": 1.2295, "step": 14066 }, { "epoch": 35.1675, "grad_norm": 6.09375, "learning_rate": 1.5046441328426364e-05, "loss": 1.3369, "step": 14067 }, { "epoch": 35.17, "grad_norm": 6.59375, "learning_rate": 1.5045742285287395e-05, "loss": 1.2781, "step": 14068 }, { "epoch": 35.1725, "grad_norm": 7.6875, "learning_rate": 1.5045043209068738e-05, "loss": 1.3123, "step": 14069 }, { "epoch": 35.175, "grad_norm": 8.0625, "learning_rate": 1.5044344099774978e-05, "loss": 1.2507, "step": 14070 }, { "epoch": 35.1775, "grad_norm": 8.0625, "learning_rate": 1.5043644957410703e-05, "loss": 1.2982, "step": 14071 }, { "epoch": 35.18, "grad_norm": 8.375, "learning_rate": 1.5042945781980494e-05, "loss": 1.3232, "step": 14072 }, { "epoch": 35.1825, "grad_norm": 9.0, "learning_rate": 1.5042246573488935e-05, "loss": 1.4251, "step": 14073 }, { "epoch": 35.185, "grad_norm": 12.625, "learning_rate": 1.5041547331940604e-05, "loss": 1.3381, "step": 14074 }, { "epoch": 35.1875, "grad_norm": 8.0625, "learning_rate": 1.5040848057340097e-05, "loss": 1.4656, "step": 14075 }, { "epoch": 35.19, "grad_norm": 6.6875, "learning_rate": 1.5040148749691983e-05, "loss": 1.3416, "step": 14076 }, { "epoch": 35.1925, "grad_norm": 9.375, "learning_rate": 1.5039449409000866e-05, "loss": 1.3053, "step": 14077 }, { "epoch": 35.195, "grad_norm": 8.625, "learning_rate": 1.5038750035271315e-05, "loss": 1.2939, "step": 14078 }, { "epoch": 35.1975, "grad_norm": 6.5, "learning_rate": 1.503805062850792e-05, "loss": 1.4938, "step": 14079 }, { "epoch": 35.2, "grad_norm": 8.3125, "learning_rate": 1.5037351188715265e-05, "loss": 1.4506, "step": 14080 }, { "epoch": 35.2025, "grad_norm": 7.125, "learning_rate": 1.5036651715897941e-05, "loss": 1.3427, "step": 14081 }, { "epoch": 35.205, "grad_norm": 6.84375, "learning_rate": 1.5035952210060529e-05, "loss": 1.513, "step": 14082 }, { "epoch": 35.2075, "grad_norm": 6.21875, "learning_rate": 1.5035252671207614e-05, "loss": 1.4069, "step": 14083 }, { "epoch": 35.21, "grad_norm": 6.96875, "learning_rate": 1.5034553099343785e-05, "loss": 1.2291, "step": 14084 }, { "epoch": 35.2125, "grad_norm": 6.9375, "learning_rate": 1.503385349447363e-05, "loss": 1.2479, "step": 14085 }, { "epoch": 35.215, "grad_norm": 7.3125, "learning_rate": 1.5033153856601728e-05, "loss": 1.3028, "step": 14086 }, { "epoch": 35.2175, "grad_norm": 4.875, "learning_rate": 1.5032454185732675e-05, "loss": 1.3463, "step": 14087 }, { "epoch": 35.22, "grad_norm": 9.5625, "learning_rate": 1.5031754481871053e-05, "loss": 1.3118, "step": 14088 }, { "epoch": 35.2225, "grad_norm": 10.0625, "learning_rate": 1.5031054745021446e-05, "loss": 1.3746, "step": 14089 }, { "epoch": 35.225, "grad_norm": 6.625, "learning_rate": 1.503035497518845e-05, "loss": 1.3626, "step": 14090 }, { "epoch": 35.2275, "grad_norm": 7.875, "learning_rate": 1.5029655172376644e-05, "loss": 1.3556, "step": 14091 }, { "epoch": 35.23, "grad_norm": 5.78125, "learning_rate": 1.5028955336590625e-05, "loss": 1.2188, "step": 14092 }, { "epoch": 35.2325, "grad_norm": 10.1875, "learning_rate": 1.5028255467834974e-05, "loss": 1.4186, "step": 14093 }, { "epoch": 35.235, "grad_norm": 9.875, "learning_rate": 1.502755556611428e-05, "loss": 1.435, "step": 14094 }, { "epoch": 35.2375, "grad_norm": 8.9375, "learning_rate": 1.5026855631433135e-05, "loss": 1.5061, "step": 14095 }, { "epoch": 35.24, "grad_norm": 5.6875, "learning_rate": 1.5026155663796123e-05, "loss": 1.3882, "step": 14096 }, { "epoch": 35.2425, "grad_norm": 6.8125, "learning_rate": 1.5025455663207837e-05, "loss": 1.2636, "step": 14097 }, { "epoch": 35.245, "grad_norm": 7.40625, "learning_rate": 1.5024755629672867e-05, "loss": 1.3725, "step": 14098 }, { "epoch": 35.2475, "grad_norm": 9.0, "learning_rate": 1.5024055563195796e-05, "loss": 1.3877, "step": 14099 }, { "epoch": 35.25, "grad_norm": 9.8125, "learning_rate": 1.5023355463781221e-05, "loss": 1.5103, "step": 14100 }, { "epoch": 35.2525, "grad_norm": 7.59375, "learning_rate": 1.5022655331433726e-05, "loss": 1.1333, "step": 14101 }, { "epoch": 35.255, "grad_norm": 6.53125, "learning_rate": 1.5021955166157906e-05, "loss": 1.4275, "step": 14102 }, { "epoch": 35.2575, "grad_norm": 6.15625, "learning_rate": 1.5021254967958348e-05, "loss": 1.1352, "step": 14103 }, { "epoch": 35.26, "grad_norm": 5.53125, "learning_rate": 1.5020554736839645e-05, "loss": 1.1423, "step": 14104 }, { "epoch": 35.2625, "grad_norm": 6.6875, "learning_rate": 1.5019854472806384e-05, "loss": 1.2714, "step": 14105 }, { "epoch": 35.265, "grad_norm": 6.9375, "learning_rate": 1.5019154175863158e-05, "loss": 1.3585, "step": 14106 }, { "epoch": 35.2675, "grad_norm": 7.5625, "learning_rate": 1.5018453846014558e-05, "loss": 1.3628, "step": 14107 }, { "epoch": 35.27, "grad_norm": 6.90625, "learning_rate": 1.5017753483265173e-05, "loss": 1.3865, "step": 14108 }, { "epoch": 35.2725, "grad_norm": 8.0, "learning_rate": 1.5017053087619602e-05, "loss": 1.248, "step": 14109 }, { "epoch": 35.275, "grad_norm": 7.34375, "learning_rate": 1.501635265908243e-05, "loss": 1.2878, "step": 14110 }, { "epoch": 35.2775, "grad_norm": 6.09375, "learning_rate": 1.501565219765825e-05, "loss": 1.3076, "step": 14111 }, { "epoch": 35.28, "grad_norm": 7.34375, "learning_rate": 1.5014951703351655e-05, "loss": 1.4135, "step": 14112 }, { "epoch": 35.2825, "grad_norm": 7.15625, "learning_rate": 1.5014251176167238e-05, "loss": 1.3752, "step": 14113 }, { "epoch": 35.285, "grad_norm": 6.59375, "learning_rate": 1.5013550616109588e-05, "loss": 1.2204, "step": 14114 }, { "epoch": 35.2875, "grad_norm": 6.8125, "learning_rate": 1.5012850023183307e-05, "loss": 1.4128, "step": 14115 }, { "epoch": 35.29, "grad_norm": 7.21875, "learning_rate": 1.5012149397392977e-05, "loss": 1.4796, "step": 14116 }, { "epoch": 35.2925, "grad_norm": 10.1875, "learning_rate": 1.5011448738743198e-05, "loss": 1.3812, "step": 14117 }, { "epoch": 35.295, "grad_norm": 8.0, "learning_rate": 1.5010748047238556e-05, "loss": 1.3804, "step": 14118 }, { "epoch": 35.2975, "grad_norm": 6.03125, "learning_rate": 1.501004732288366e-05, "loss": 1.3141, "step": 14119 }, { "epoch": 35.3, "grad_norm": 5.03125, "learning_rate": 1.5009346565683088e-05, "loss": 1.2657, "step": 14120 }, { "epoch": 35.3025, "grad_norm": 9.125, "learning_rate": 1.5008645775641441e-05, "loss": 1.373, "step": 14121 }, { "epoch": 35.305, "grad_norm": 8.25, "learning_rate": 1.5007944952763314e-05, "loss": 1.4129, "step": 14122 }, { "epoch": 35.3075, "grad_norm": 6.5, "learning_rate": 1.5007244097053299e-05, "loss": 1.3925, "step": 14123 }, { "epoch": 35.31, "grad_norm": 8.0, "learning_rate": 1.5006543208515993e-05, "loss": 1.3493, "step": 14124 }, { "epoch": 35.3125, "grad_norm": 5.4375, "learning_rate": 1.500584228715599e-05, "loss": 1.3054, "step": 14125 }, { "epoch": 35.315, "grad_norm": 6.5, "learning_rate": 1.5005141332977884e-05, "loss": 1.2807, "step": 14126 }, { "epoch": 35.3175, "grad_norm": 7.25, "learning_rate": 1.5004440345986275e-05, "loss": 1.2764, "step": 14127 }, { "epoch": 35.32, "grad_norm": 7.9375, "learning_rate": 1.500373932618575e-05, "loss": 1.5579, "step": 14128 }, { "epoch": 35.3225, "grad_norm": 9.125, "learning_rate": 1.5003038273580916e-05, "loss": 1.3474, "step": 14129 }, { "epoch": 35.325, "grad_norm": 8.0625, "learning_rate": 1.5002337188176362e-05, "loss": 1.402, "step": 14130 }, { "epoch": 35.3275, "grad_norm": 7.5, "learning_rate": 1.5001636069976682e-05, "loss": 1.4266, "step": 14131 }, { "epoch": 35.33, "grad_norm": 7.3125, "learning_rate": 1.500093491898648e-05, "loss": 1.2756, "step": 14132 }, { "epoch": 35.3325, "grad_norm": 5.71875, "learning_rate": 1.5000233735210345e-05, "loss": 1.3732, "step": 14133 }, { "epoch": 35.335, "grad_norm": 7.78125, "learning_rate": 1.4999532518652882e-05, "loss": 1.3773, "step": 14134 }, { "epoch": 35.3375, "grad_norm": 6.5, "learning_rate": 1.4998831269318684e-05, "loss": 1.3703, "step": 14135 }, { "epoch": 35.34, "grad_norm": 6.90625, "learning_rate": 1.4998129987212344e-05, "loss": 1.4182, "step": 14136 }, { "epoch": 35.3425, "grad_norm": 5.34375, "learning_rate": 1.4997428672338465e-05, "loss": 1.3765, "step": 14137 }, { "epoch": 35.345, "grad_norm": 5.46875, "learning_rate": 1.4996727324701646e-05, "loss": 1.5608, "step": 14138 }, { "epoch": 35.3475, "grad_norm": 9.5625, "learning_rate": 1.499602594430648e-05, "loss": 1.3289, "step": 14139 }, { "epoch": 35.35, "grad_norm": 6.34375, "learning_rate": 1.4995324531157569e-05, "loss": 1.3888, "step": 14140 }, { "epoch": 35.3525, "grad_norm": 8.5625, "learning_rate": 1.4994623085259509e-05, "loss": 1.441, "step": 14141 }, { "epoch": 35.355, "grad_norm": 8.9375, "learning_rate": 1.4993921606616902e-05, "loss": 1.5064, "step": 14142 }, { "epoch": 35.3575, "grad_norm": 7.21875, "learning_rate": 1.4993220095234344e-05, "loss": 1.2633, "step": 14143 }, { "epoch": 35.36, "grad_norm": 7.0, "learning_rate": 1.4992518551116436e-05, "loss": 1.224, "step": 14144 }, { "epoch": 35.3625, "grad_norm": 8.0625, "learning_rate": 1.4991816974267775e-05, "loss": 1.3389, "step": 14145 }, { "epoch": 35.365, "grad_norm": 9.25, "learning_rate": 1.4991115364692958e-05, "loss": 1.398, "step": 14146 }, { "epoch": 35.3675, "grad_norm": 10.6875, "learning_rate": 1.4990413722396594e-05, "loss": 1.3394, "step": 14147 }, { "epoch": 35.37, "grad_norm": 8.5, "learning_rate": 1.4989712047383276e-05, "loss": 1.3692, "step": 14148 }, { "epoch": 35.3725, "grad_norm": 5.84375, "learning_rate": 1.4989010339657605e-05, "loss": 1.2401, "step": 14149 }, { "epoch": 35.375, "grad_norm": 6.78125, "learning_rate": 1.4988308599224182e-05, "loss": 1.2416, "step": 14150 }, { "epoch": 35.3775, "grad_norm": 10.3125, "learning_rate": 1.4987606826087606e-05, "loss": 1.4543, "step": 14151 }, { "epoch": 35.38, "grad_norm": 7.9375, "learning_rate": 1.4986905020252482e-05, "loss": 1.4839, "step": 14152 }, { "epoch": 35.3825, "grad_norm": 10.375, "learning_rate": 1.4986203181723406e-05, "loss": 1.3378, "step": 14153 }, { "epoch": 35.385, "grad_norm": 10.3125, "learning_rate": 1.4985501310504982e-05, "loss": 1.3945, "step": 14154 }, { "epoch": 35.3875, "grad_norm": 8.125, "learning_rate": 1.4984799406601812e-05, "loss": 1.3702, "step": 14155 }, { "epoch": 35.39, "grad_norm": 6.25, "learning_rate": 1.4984097470018496e-05, "loss": 1.3038, "step": 14156 }, { "epoch": 35.3925, "grad_norm": 4.96875, "learning_rate": 1.4983395500759635e-05, "loss": 1.3094, "step": 14157 }, { "epoch": 35.395, "grad_norm": 8.5, "learning_rate": 1.4982693498829833e-05, "loss": 1.5075, "step": 14158 }, { "epoch": 35.3975, "grad_norm": 8.3125, "learning_rate": 1.4981991464233692e-05, "loss": 1.5188, "step": 14159 }, { "epoch": 35.4, "grad_norm": 15.25, "learning_rate": 1.4981289396975818e-05, "loss": 1.3295, "step": 14160 }, { "epoch": 35.4025, "grad_norm": 9.25, "learning_rate": 1.4980587297060805e-05, "loss": 1.2488, "step": 14161 }, { "epoch": 35.405, "grad_norm": 7.78125, "learning_rate": 1.4979885164493264e-05, "loss": 1.1978, "step": 14162 }, { "epoch": 35.4075, "grad_norm": 7.5, "learning_rate": 1.497918299927779e-05, "loss": 1.3358, "step": 14163 }, { "epoch": 35.41, "grad_norm": 7.46875, "learning_rate": 1.4978480801419e-05, "loss": 1.4005, "step": 14164 }, { "epoch": 35.4125, "grad_norm": 6.4375, "learning_rate": 1.4977778570921485e-05, "loss": 1.275, "step": 14165 }, { "epoch": 35.415, "grad_norm": 7.25, "learning_rate": 1.497707630778985e-05, "loss": 1.3021, "step": 14166 }, { "epoch": 35.4175, "grad_norm": 7.15625, "learning_rate": 1.4976374012028705e-05, "loss": 1.3376, "step": 14167 }, { "epoch": 35.42, "grad_norm": 5.15625, "learning_rate": 1.497567168364265e-05, "loss": 1.2306, "step": 14168 }, { "epoch": 35.4225, "grad_norm": 7.71875, "learning_rate": 1.497496932263629e-05, "loss": 1.379, "step": 14169 }, { "epoch": 35.425, "grad_norm": 8.3125, "learning_rate": 1.4974266929014234e-05, "loss": 1.3876, "step": 14170 }, { "epoch": 35.4275, "grad_norm": 7.71875, "learning_rate": 1.497356450278108e-05, "loss": 1.3005, "step": 14171 }, { "epoch": 35.43, "grad_norm": 6.9375, "learning_rate": 1.4972862043941436e-05, "loss": 1.4325, "step": 14172 }, { "epoch": 35.4325, "grad_norm": 7.28125, "learning_rate": 1.4972159552499908e-05, "loss": 1.3155, "step": 14173 }, { "epoch": 35.435, "grad_norm": 6.65625, "learning_rate": 1.4971457028461101e-05, "loss": 1.3193, "step": 14174 }, { "epoch": 35.4375, "grad_norm": 7.4375, "learning_rate": 1.497075447182962e-05, "loss": 1.2157, "step": 14175 }, { "epoch": 35.44, "grad_norm": 9.375, "learning_rate": 1.4970051882610073e-05, "loss": 1.5392, "step": 14176 }, { "epoch": 35.4425, "grad_norm": 7.65625, "learning_rate": 1.4969349260807065e-05, "loss": 1.1613, "step": 14177 }, { "epoch": 35.445, "grad_norm": 5.4375, "learning_rate": 1.4968646606425197e-05, "loss": 1.2803, "step": 14178 }, { "epoch": 35.4475, "grad_norm": 6.96875, "learning_rate": 1.4967943919469088e-05, "loss": 1.3535, "step": 14179 }, { "epoch": 35.45, "grad_norm": 7.65625, "learning_rate": 1.4967241199943332e-05, "loss": 1.3083, "step": 14180 }, { "epoch": 35.4525, "grad_norm": 7.40625, "learning_rate": 1.4966538447852545e-05, "loss": 1.3042, "step": 14181 }, { "epoch": 35.455, "grad_norm": 8.9375, "learning_rate": 1.496583566320133e-05, "loss": 1.3148, "step": 14182 }, { "epoch": 35.4575, "grad_norm": 6.0625, "learning_rate": 1.4965132845994293e-05, "loss": 1.4046, "step": 14183 }, { "epoch": 35.46, "grad_norm": 6.0625, "learning_rate": 1.4964429996236045e-05, "loss": 1.2766, "step": 14184 }, { "epoch": 35.4625, "grad_norm": 8.0, "learning_rate": 1.4963727113931194e-05, "loss": 1.3665, "step": 14185 }, { "epoch": 35.465, "grad_norm": 8.5, "learning_rate": 1.4963024199084343e-05, "loss": 1.1791, "step": 14186 }, { "epoch": 35.4675, "grad_norm": 8.6875, "learning_rate": 1.4962321251700109e-05, "loss": 1.2207, "step": 14187 }, { "epoch": 35.47, "grad_norm": 7.96875, "learning_rate": 1.4961618271783091e-05, "loss": 1.3019, "step": 14188 }, { "epoch": 35.4725, "grad_norm": 5.34375, "learning_rate": 1.4960915259337907e-05, "loss": 1.3242, "step": 14189 }, { "epoch": 35.475, "grad_norm": 8.125, "learning_rate": 1.4960212214369156e-05, "loss": 1.2333, "step": 14190 }, { "epoch": 35.4775, "grad_norm": 6.75, "learning_rate": 1.4959509136881454e-05, "loss": 1.4097, "step": 14191 }, { "epoch": 35.48, "grad_norm": 9.375, "learning_rate": 1.4958806026879411e-05, "loss": 1.31, "step": 14192 }, { "epoch": 35.4825, "grad_norm": 7.40625, "learning_rate": 1.495810288436763e-05, "loss": 1.391, "step": 14193 }, { "epoch": 35.485, "grad_norm": 7.5, "learning_rate": 1.4957399709350729e-05, "loss": 1.3239, "step": 14194 }, { "epoch": 35.4875, "grad_norm": 7.8125, "learning_rate": 1.495669650183331e-05, "loss": 1.4379, "step": 14195 }, { "epoch": 35.49, "grad_norm": 8.0625, "learning_rate": 1.4955993261819988e-05, "loss": 1.4072, "step": 14196 }, { "epoch": 35.4925, "grad_norm": 5.53125, "learning_rate": 1.4955289989315374e-05, "loss": 1.3536, "step": 14197 }, { "epoch": 35.495, "grad_norm": 5.375, "learning_rate": 1.4954586684324077e-05, "loss": 1.3834, "step": 14198 }, { "epoch": 35.4975, "grad_norm": 9.5, "learning_rate": 1.4953883346850706e-05, "loss": 1.3627, "step": 14199 }, { "epoch": 35.5, "grad_norm": 6.3125, "learning_rate": 1.4953179976899878e-05, "loss": 1.2603, "step": 14200 }, { "epoch": 35.5025, "grad_norm": 3.9375, "learning_rate": 1.4952476574476195e-05, "loss": 1.2526, "step": 14201 }, { "epoch": 35.505, "grad_norm": 7.0, "learning_rate": 1.495177313958428e-05, "loss": 1.3378, "step": 14202 }, { "epoch": 35.5075, "grad_norm": 9.0, "learning_rate": 1.4951069672228734e-05, "loss": 1.4012, "step": 14203 }, { "epoch": 35.51, "grad_norm": 7.125, "learning_rate": 1.4950366172414175e-05, "loss": 1.3855, "step": 14204 }, { "epoch": 35.5125, "grad_norm": 7.375, "learning_rate": 1.4949662640145212e-05, "loss": 1.4572, "step": 14205 }, { "epoch": 35.515, "grad_norm": 7.78125, "learning_rate": 1.494895907542646e-05, "loss": 1.4437, "step": 14206 }, { "epoch": 35.5175, "grad_norm": 6.8125, "learning_rate": 1.494825547826253e-05, "loss": 1.3308, "step": 14207 }, { "epoch": 35.52, "grad_norm": 5.625, "learning_rate": 1.4947551848658036e-05, "loss": 1.3329, "step": 14208 }, { "epoch": 35.5225, "grad_norm": 6.75, "learning_rate": 1.4946848186617588e-05, "loss": 1.3062, "step": 14209 }, { "epoch": 35.525, "grad_norm": 7.6875, "learning_rate": 1.4946144492145801e-05, "loss": 1.4451, "step": 14210 }, { "epoch": 35.5275, "grad_norm": 6.71875, "learning_rate": 1.4945440765247292e-05, "loss": 1.2861, "step": 14211 }, { "epoch": 35.53, "grad_norm": 9.1875, "learning_rate": 1.4944737005926668e-05, "loss": 1.417, "step": 14212 }, { "epoch": 35.5325, "grad_norm": 8.5625, "learning_rate": 1.4944033214188547e-05, "loss": 1.3097, "step": 14213 }, { "epoch": 35.535, "grad_norm": 9.75, "learning_rate": 1.4943329390037545e-05, "loss": 1.4134, "step": 14214 }, { "epoch": 35.5375, "grad_norm": 7.3125, "learning_rate": 1.494262553347827e-05, "loss": 1.366, "step": 14215 }, { "epoch": 35.54, "grad_norm": 6.34375, "learning_rate": 1.4941921644515338e-05, "loss": 1.3104, "step": 14216 }, { "epoch": 35.5425, "grad_norm": 7.59375, "learning_rate": 1.4941217723153372e-05, "loss": 1.4185, "step": 14217 }, { "epoch": 35.545, "grad_norm": 6.09375, "learning_rate": 1.4940513769396974e-05, "loss": 1.2237, "step": 14218 }, { "epoch": 35.5475, "grad_norm": 6.71875, "learning_rate": 1.493980978325077e-05, "loss": 1.2645, "step": 14219 }, { "epoch": 35.55, "grad_norm": 5.03125, "learning_rate": 1.4939105764719369e-05, "loss": 1.2317, "step": 14220 }, { "epoch": 35.5525, "grad_norm": 7.46875, "learning_rate": 1.4938401713807387e-05, "loss": 1.2552, "step": 14221 }, { "epoch": 35.555, "grad_norm": 8.875, "learning_rate": 1.4937697630519441e-05, "loss": 1.3822, "step": 14222 }, { "epoch": 35.5575, "grad_norm": 11.4375, "learning_rate": 1.4936993514860148e-05, "loss": 1.5071, "step": 14223 }, { "epoch": 35.56, "grad_norm": 6.5, "learning_rate": 1.4936289366834123e-05, "loss": 1.4348, "step": 14224 }, { "epoch": 35.5625, "grad_norm": 7.0, "learning_rate": 1.4935585186445982e-05, "loss": 1.4869, "step": 14225 }, { "epoch": 35.565, "grad_norm": 9.6875, "learning_rate": 1.4934880973700342e-05, "loss": 1.2518, "step": 14226 }, { "epoch": 35.5675, "grad_norm": 7.5625, "learning_rate": 1.493417672860182e-05, "loss": 1.4886, "step": 14227 }, { "epoch": 35.57, "grad_norm": 5.875, "learning_rate": 1.4933472451155034e-05, "loss": 1.325, "step": 14228 }, { "epoch": 35.5725, "grad_norm": 4.9375, "learning_rate": 1.4932768141364598e-05, "loss": 1.2058, "step": 14229 }, { "epoch": 35.575, "grad_norm": 6.09375, "learning_rate": 1.4932063799235135e-05, "loss": 1.357, "step": 14230 }, { "epoch": 35.5775, "grad_norm": 9.875, "learning_rate": 1.4931359424771255e-05, "loss": 1.3179, "step": 14231 }, { "epoch": 35.58, "grad_norm": 7.90625, "learning_rate": 1.4930655017977583e-05, "loss": 1.3526, "step": 14232 }, { "epoch": 35.5825, "grad_norm": 8.75, "learning_rate": 1.4929950578858729e-05, "loss": 1.195, "step": 14233 }, { "epoch": 35.585, "grad_norm": 6.84375, "learning_rate": 1.4929246107419323e-05, "loss": 1.2678, "step": 14234 }, { "epoch": 35.5875, "grad_norm": 7.09375, "learning_rate": 1.4928541603663971e-05, "loss": 1.301, "step": 14235 }, { "epoch": 35.59, "grad_norm": 6.53125, "learning_rate": 1.4927837067597301e-05, "loss": 1.3814, "step": 14236 }, { "epoch": 35.5925, "grad_norm": 6.5625, "learning_rate": 1.4927132499223928e-05, "loss": 1.6668, "step": 14237 }, { "epoch": 35.595, "grad_norm": 6.34375, "learning_rate": 1.492642789854847e-05, "loss": 1.3675, "step": 14238 }, { "epoch": 35.5975, "grad_norm": 6.8125, "learning_rate": 1.4925723265575549e-05, "loss": 1.3168, "step": 14239 }, { "epoch": 35.6, "grad_norm": 7.0, "learning_rate": 1.4925018600309784e-05, "loss": 1.3868, "step": 14240 }, { "epoch": 35.6025, "grad_norm": 15.1875, "learning_rate": 1.4924313902755793e-05, "loss": 1.5122, "step": 14241 }, { "epoch": 35.605, "grad_norm": 8.875, "learning_rate": 1.4923609172918198e-05, "loss": 1.2504, "step": 14242 }, { "epoch": 35.6075, "grad_norm": 8.5, "learning_rate": 1.4922904410801618e-05, "loss": 1.2937, "step": 14243 }, { "epoch": 35.61, "grad_norm": 8.8125, "learning_rate": 1.4922199616410675e-05, "loss": 1.5183, "step": 14244 }, { "epoch": 35.6125, "grad_norm": 7.6875, "learning_rate": 1.4921494789749984e-05, "loss": 1.5017, "step": 14245 }, { "epoch": 35.615, "grad_norm": 8.3125, "learning_rate": 1.4920789930824175e-05, "loss": 1.4754, "step": 14246 }, { "epoch": 35.6175, "grad_norm": 7.71875, "learning_rate": 1.492008503963786e-05, "loss": 1.2075, "step": 14247 }, { "epoch": 35.62, "grad_norm": 7.0, "learning_rate": 1.4919380116195665e-05, "loss": 1.3017, "step": 14248 }, { "epoch": 35.6225, "grad_norm": 8.5625, "learning_rate": 1.4918675160502215e-05, "loss": 1.3288, "step": 14249 }, { "epoch": 35.625, "grad_norm": 5.6875, "learning_rate": 1.4917970172562122e-05, "loss": 1.2679, "step": 14250 }, { "epoch": 35.6275, "grad_norm": 6.75, "learning_rate": 1.4917265152380019e-05, "loss": 1.328, "step": 14251 }, { "epoch": 35.63, "grad_norm": 7.40625, "learning_rate": 1.4916560099960519e-05, "loss": 1.3961, "step": 14252 }, { "epoch": 35.6325, "grad_norm": 6.28125, "learning_rate": 1.4915855015308248e-05, "loss": 1.4171, "step": 14253 }, { "epoch": 35.635, "grad_norm": 6.1875, "learning_rate": 1.4915149898427828e-05, "loss": 1.3867, "step": 14254 }, { "epoch": 35.6375, "grad_norm": 7.28125, "learning_rate": 1.4914444749323881e-05, "loss": 1.471, "step": 14255 }, { "epoch": 35.64, "grad_norm": 26.125, "learning_rate": 1.4913739568001034e-05, "loss": 1.2624, "step": 14256 }, { "epoch": 35.6425, "grad_norm": 7.125, "learning_rate": 1.4913034354463907e-05, "loss": 1.2738, "step": 14257 }, { "epoch": 35.645, "grad_norm": 6.8125, "learning_rate": 1.4912329108717121e-05, "loss": 1.3099, "step": 14258 }, { "epoch": 35.6475, "grad_norm": 8.125, "learning_rate": 1.4911623830765303e-05, "loss": 1.5304, "step": 14259 }, { "epoch": 35.65, "grad_norm": 8.9375, "learning_rate": 1.4910918520613074e-05, "loss": 1.35, "step": 14260 }, { "epoch": 35.6525, "grad_norm": 8.6875, "learning_rate": 1.4910213178265062e-05, "loss": 1.4492, "step": 14261 }, { "epoch": 35.655, "grad_norm": 6.875, "learning_rate": 1.4909507803725887e-05, "loss": 1.3932, "step": 14262 }, { "epoch": 35.6575, "grad_norm": 6.40625, "learning_rate": 1.4908802397000177e-05, "loss": 1.3233, "step": 14263 }, { "epoch": 35.66, "grad_norm": 6.15625, "learning_rate": 1.4908096958092554e-05, "loss": 1.4197, "step": 14264 }, { "epoch": 35.6625, "grad_norm": 7.375, "learning_rate": 1.4907391487007644e-05, "loss": 1.2697, "step": 14265 }, { "epoch": 35.665, "grad_norm": 8.375, "learning_rate": 1.4906685983750072e-05, "loss": 1.2986, "step": 14266 }, { "epoch": 35.6675, "grad_norm": 7.0, "learning_rate": 1.4905980448324462e-05, "loss": 1.3612, "step": 14267 }, { "epoch": 35.67, "grad_norm": 6.28125, "learning_rate": 1.4905274880735441e-05, "loss": 1.4588, "step": 14268 }, { "epoch": 35.6725, "grad_norm": 11.25, "learning_rate": 1.4904569280987635e-05, "loss": 1.4533, "step": 14269 }, { "epoch": 35.675, "grad_norm": 7.46875, "learning_rate": 1.4903863649085668e-05, "loss": 1.2727, "step": 14270 }, { "epoch": 35.6775, "grad_norm": 6.09375, "learning_rate": 1.4903157985034166e-05, "loss": 1.3063, "step": 14271 }, { "epoch": 35.68, "grad_norm": 7.125, "learning_rate": 1.4902452288837761e-05, "loss": 1.4314, "step": 14272 }, { "epoch": 35.6825, "grad_norm": 8.4375, "learning_rate": 1.4901746560501072e-05, "loss": 1.4542, "step": 14273 }, { "epoch": 35.685, "grad_norm": 7.25, "learning_rate": 1.490104080002873e-05, "loss": 1.3552, "step": 14274 }, { "epoch": 35.6875, "grad_norm": 6.4375, "learning_rate": 1.4900335007425358e-05, "loss": 1.367, "step": 14275 }, { "epoch": 35.69, "grad_norm": 13.6875, "learning_rate": 1.4899629182695587e-05, "loss": 1.4858, "step": 14276 }, { "epoch": 35.6925, "grad_norm": 7.5625, "learning_rate": 1.4898923325844044e-05, "loss": 1.2562, "step": 14277 }, { "epoch": 35.695, "grad_norm": 6.21875, "learning_rate": 1.4898217436875354e-05, "loss": 1.3154, "step": 14278 }, { "epoch": 35.6975, "grad_norm": 7.6875, "learning_rate": 1.489751151579415e-05, "loss": 1.4825, "step": 14279 }, { "epoch": 35.7, "grad_norm": 8.1875, "learning_rate": 1.4896805562605052e-05, "loss": 1.2409, "step": 14280 }, { "epoch": 35.7025, "grad_norm": 6.9375, "learning_rate": 1.4896099577312695e-05, "loss": 1.3584, "step": 14281 }, { "epoch": 35.705, "grad_norm": 8.25, "learning_rate": 1.4895393559921705e-05, "loss": 1.4113, "step": 14282 }, { "epoch": 35.7075, "grad_norm": 7.71875, "learning_rate": 1.4894687510436712e-05, "loss": 1.3127, "step": 14283 }, { "epoch": 35.71, "grad_norm": 6.5625, "learning_rate": 1.489398142886234e-05, "loss": 1.2512, "step": 14284 }, { "epoch": 35.7125, "grad_norm": 7.03125, "learning_rate": 1.4893275315203223e-05, "loss": 1.5028, "step": 14285 }, { "epoch": 35.715, "grad_norm": 9.0625, "learning_rate": 1.489256916946399e-05, "loss": 1.5093, "step": 14286 }, { "epoch": 35.7175, "grad_norm": 6.625, "learning_rate": 1.4891862991649273e-05, "loss": 1.3598, "step": 14287 }, { "epoch": 35.72, "grad_norm": 5.96875, "learning_rate": 1.4891156781763692e-05, "loss": 1.311, "step": 14288 }, { "epoch": 35.7225, "grad_norm": 6.75, "learning_rate": 1.4890450539811886e-05, "loss": 1.3103, "step": 14289 }, { "epoch": 35.725, "grad_norm": 7.75, "learning_rate": 1.4889744265798478e-05, "loss": 1.5828, "step": 14290 }, { "epoch": 35.7275, "grad_norm": 9.0625, "learning_rate": 1.4889037959728108e-05, "loss": 1.6455, "step": 14291 }, { "epoch": 35.73, "grad_norm": 7.1875, "learning_rate": 1.4888331621605397e-05, "loss": 1.1676, "step": 14292 }, { "epoch": 35.7325, "grad_norm": 7.8125, "learning_rate": 1.4887625251434981e-05, "loss": 1.4358, "step": 14293 }, { "epoch": 35.735, "grad_norm": 6.875, "learning_rate": 1.4886918849221488e-05, "loss": 1.3169, "step": 14294 }, { "epoch": 35.7375, "grad_norm": 9.0, "learning_rate": 1.4886212414969551e-05, "loss": 1.3193, "step": 14295 }, { "epoch": 35.74, "grad_norm": 6.9375, "learning_rate": 1.4885505948683801e-05, "loss": 1.5861, "step": 14296 }, { "epoch": 35.7425, "grad_norm": 12.875, "learning_rate": 1.4884799450368869e-05, "loss": 1.3441, "step": 14297 }, { "epoch": 35.745, "grad_norm": 6.375, "learning_rate": 1.488409292002939e-05, "loss": 1.2128, "step": 14298 }, { "epoch": 35.7475, "grad_norm": 7.0625, "learning_rate": 1.4883386357669991e-05, "loss": 1.3906, "step": 14299 }, { "epoch": 35.75, "grad_norm": 8.375, "learning_rate": 1.4882679763295307e-05, "loss": 1.4229, "step": 14300 }, { "epoch": 35.7525, "grad_norm": 8.375, "learning_rate": 1.4881973136909969e-05, "loss": 1.2772, "step": 14301 }, { "epoch": 35.755, "grad_norm": 6.375, "learning_rate": 1.4881266478518612e-05, "loss": 1.2375, "step": 14302 }, { "epoch": 35.7575, "grad_norm": 9.3125, "learning_rate": 1.4880559788125865e-05, "loss": 1.3995, "step": 14303 }, { "epoch": 35.76, "grad_norm": 8.0625, "learning_rate": 1.4879853065736366e-05, "loss": 1.3047, "step": 14304 }, { "epoch": 35.7625, "grad_norm": 7.125, "learning_rate": 1.4879146311354741e-05, "loss": 1.3651, "step": 14305 }, { "epoch": 35.765, "grad_norm": 9.9375, "learning_rate": 1.4878439524985635e-05, "loss": 1.4069, "step": 14306 }, { "epoch": 35.7675, "grad_norm": 6.46875, "learning_rate": 1.4877732706633673e-05, "loss": 1.292, "step": 14307 }, { "epoch": 35.77, "grad_norm": 6.6875, "learning_rate": 1.4877025856303486e-05, "loss": 1.3377, "step": 14308 }, { "epoch": 35.7725, "grad_norm": 8.375, "learning_rate": 1.4876318973999718e-05, "loss": 1.3869, "step": 14309 }, { "epoch": 35.775, "grad_norm": 6.9375, "learning_rate": 1.4875612059726993e-05, "loss": 1.1801, "step": 14310 }, { "epoch": 35.7775, "grad_norm": 7.34375, "learning_rate": 1.4874905113489953e-05, "loss": 1.3243, "step": 14311 }, { "epoch": 35.78, "grad_norm": 5.96875, "learning_rate": 1.4874198135293232e-05, "loss": 1.4481, "step": 14312 }, { "epoch": 35.7825, "grad_norm": 10.9375, "learning_rate": 1.4873491125141459e-05, "loss": 1.4825, "step": 14313 }, { "epoch": 35.785, "grad_norm": 6.1875, "learning_rate": 1.4872784083039275e-05, "loss": 1.2989, "step": 14314 }, { "epoch": 35.7875, "grad_norm": 7.03125, "learning_rate": 1.4872077008991314e-05, "loss": 1.2563, "step": 14315 }, { "epoch": 35.79, "grad_norm": 7.4375, "learning_rate": 1.4871369903002211e-05, "loss": 1.3413, "step": 14316 }, { "epoch": 35.7925, "grad_norm": 9.6875, "learning_rate": 1.4870662765076602e-05, "loss": 1.3494, "step": 14317 }, { "epoch": 35.795, "grad_norm": 6.875, "learning_rate": 1.486995559521912e-05, "loss": 1.501, "step": 14318 }, { "epoch": 35.7975, "grad_norm": 5.625, "learning_rate": 1.486924839343441e-05, "loss": 1.186, "step": 14319 }, { "epoch": 35.8, "grad_norm": 6.125, "learning_rate": 1.4868541159727097e-05, "loss": 1.3429, "step": 14320 }, { "epoch": 35.8025, "grad_norm": 6.15625, "learning_rate": 1.4867833894101827e-05, "loss": 1.4801, "step": 14321 }, { "epoch": 35.805, "grad_norm": 7.09375, "learning_rate": 1.4867126596563231e-05, "loss": 1.2476, "step": 14322 }, { "epoch": 35.8075, "grad_norm": 6.9375, "learning_rate": 1.4866419267115945e-05, "loss": 1.2501, "step": 14323 }, { "epoch": 35.81, "grad_norm": 7.1875, "learning_rate": 1.4865711905764613e-05, "loss": 1.2828, "step": 14324 }, { "epoch": 35.8125, "grad_norm": 7.125, "learning_rate": 1.4865004512513865e-05, "loss": 1.3777, "step": 14325 }, { "epoch": 35.815, "grad_norm": 9.8125, "learning_rate": 1.4864297087368344e-05, "loss": 1.4531, "step": 14326 }, { "epoch": 35.8175, "grad_norm": 9.625, "learning_rate": 1.486358963033269e-05, "loss": 1.2525, "step": 14327 }, { "epoch": 35.82, "grad_norm": 9.625, "learning_rate": 1.486288214141153e-05, "loss": 1.5037, "step": 14328 }, { "epoch": 35.8225, "grad_norm": 6.875, "learning_rate": 1.4862174620609516e-05, "loss": 1.3153, "step": 14329 }, { "epoch": 35.825, "grad_norm": 8.0, "learning_rate": 1.4861467067931273e-05, "loss": 1.3552, "step": 14330 }, { "epoch": 35.8275, "grad_norm": 8.375, "learning_rate": 1.486075948338145e-05, "loss": 1.2977, "step": 14331 }, { "epoch": 35.83, "grad_norm": 9.5625, "learning_rate": 1.4860051866964682e-05, "loss": 1.4468, "step": 14332 }, { "epoch": 35.8325, "grad_norm": 7.8125, "learning_rate": 1.485934421868561e-05, "loss": 1.4667, "step": 14333 }, { "epoch": 35.835, "grad_norm": 8.9375, "learning_rate": 1.485863653854887e-05, "loss": 1.417, "step": 14334 }, { "epoch": 35.8375, "grad_norm": 4.5625, "learning_rate": 1.4857928826559103e-05, "loss": 1.3759, "step": 14335 }, { "epoch": 35.84, "grad_norm": 7.03125, "learning_rate": 1.485722108272095e-05, "loss": 1.4026, "step": 14336 }, { "epoch": 35.8425, "grad_norm": 7.3125, "learning_rate": 1.4856513307039047e-05, "loss": 1.456, "step": 14337 }, { "epoch": 35.845, "grad_norm": 8.375, "learning_rate": 1.485580549951804e-05, "loss": 1.2525, "step": 14338 }, { "epoch": 35.8475, "grad_norm": 6.96875, "learning_rate": 1.4855097660162567e-05, "loss": 1.3286, "step": 14339 }, { "epoch": 35.85, "grad_norm": 5.6875, "learning_rate": 1.4854389788977266e-05, "loss": 1.4358, "step": 14340 }, { "epoch": 35.8525, "grad_norm": 8.375, "learning_rate": 1.4853681885966782e-05, "loss": 1.3349, "step": 14341 }, { "epoch": 35.855, "grad_norm": 7.03125, "learning_rate": 1.4852973951135754e-05, "loss": 1.3254, "step": 14342 }, { "epoch": 35.8575, "grad_norm": 7.21875, "learning_rate": 1.4852265984488818e-05, "loss": 1.1807, "step": 14343 }, { "epoch": 35.86, "grad_norm": 7.4375, "learning_rate": 1.4851557986030626e-05, "loss": 1.3897, "step": 14344 }, { "epoch": 35.8625, "grad_norm": 8.0, "learning_rate": 1.4850849955765809e-05, "loss": 1.3007, "step": 14345 }, { "epoch": 35.865, "grad_norm": 9.3125, "learning_rate": 1.4850141893699019e-05, "loss": 1.4399, "step": 14346 }, { "epoch": 35.8675, "grad_norm": 8.9375, "learning_rate": 1.4849433799834892e-05, "loss": 1.201, "step": 14347 }, { "epoch": 35.87, "grad_norm": 9.5625, "learning_rate": 1.4848725674178069e-05, "loss": 1.4235, "step": 14348 }, { "epoch": 35.8725, "grad_norm": 6.59375, "learning_rate": 1.4848017516733196e-05, "loss": 1.332, "step": 14349 }, { "epoch": 35.875, "grad_norm": 6.9375, "learning_rate": 1.484730932750491e-05, "loss": 1.1822, "step": 14350 }, { "epoch": 35.8775, "grad_norm": 6.40625, "learning_rate": 1.4846601106497863e-05, "loss": 1.278, "step": 14351 }, { "epoch": 35.88, "grad_norm": 7.71875, "learning_rate": 1.4845892853716692e-05, "loss": 1.475, "step": 14352 }, { "epoch": 35.8825, "grad_norm": 7.3125, "learning_rate": 1.484518456916604e-05, "loss": 1.282, "step": 14353 }, { "epoch": 35.885, "grad_norm": 5.46875, "learning_rate": 1.4844476252850552e-05, "loss": 1.2603, "step": 14354 }, { "epoch": 35.8875, "grad_norm": 9.5, "learning_rate": 1.4843767904774873e-05, "loss": 1.5626, "step": 14355 }, { "epoch": 35.89, "grad_norm": 12.5625, "learning_rate": 1.4843059524943644e-05, "loss": 1.4372, "step": 14356 }, { "epoch": 35.8925, "grad_norm": 8.125, "learning_rate": 1.484235111336151e-05, "loss": 1.3455, "step": 14357 }, { "epoch": 35.895, "grad_norm": 7.1875, "learning_rate": 1.4841642670033118e-05, "loss": 1.3824, "step": 14358 }, { "epoch": 35.8975, "grad_norm": 6.6875, "learning_rate": 1.4840934194963112e-05, "loss": 1.3017, "step": 14359 }, { "epoch": 35.9, "grad_norm": 5.53125, "learning_rate": 1.4840225688156132e-05, "loss": 1.3604, "step": 14360 }, { "epoch": 35.9025, "grad_norm": 6.71875, "learning_rate": 1.4839517149616827e-05, "loss": 1.4375, "step": 14361 }, { "epoch": 35.905, "grad_norm": 6.09375, "learning_rate": 1.4838808579349839e-05, "loss": 1.4514, "step": 14362 }, { "epoch": 35.9075, "grad_norm": 9.9375, "learning_rate": 1.4838099977359818e-05, "loss": 1.419, "step": 14363 }, { "epoch": 35.91, "grad_norm": 10.5, "learning_rate": 1.4837391343651409e-05, "loss": 1.4596, "step": 14364 }, { "epoch": 35.9125, "grad_norm": 11.4375, "learning_rate": 1.4836682678229253e-05, "loss": 1.3374, "step": 14365 }, { "epoch": 35.915, "grad_norm": 7.3125, "learning_rate": 1.4835973981098e-05, "loss": 1.3965, "step": 14366 }, { "epoch": 35.9175, "grad_norm": 7.625, "learning_rate": 1.4835265252262295e-05, "loss": 1.3598, "step": 14367 }, { "epoch": 35.92, "grad_norm": 5.5625, "learning_rate": 1.4834556491726781e-05, "loss": 1.2655, "step": 14368 }, { "epoch": 35.9225, "grad_norm": 10.25, "learning_rate": 1.4833847699496114e-05, "loss": 1.2878, "step": 14369 }, { "epoch": 35.925, "grad_norm": 6.125, "learning_rate": 1.4833138875574931e-05, "loss": 1.2515, "step": 14370 }, { "epoch": 35.9275, "grad_norm": 7.78125, "learning_rate": 1.4832430019967884e-05, "loss": 1.236, "step": 14371 }, { "epoch": 35.93, "grad_norm": 6.71875, "learning_rate": 1.483172113267962e-05, "loss": 1.2378, "step": 14372 }, { "epoch": 35.9325, "grad_norm": 9.375, "learning_rate": 1.4831012213714783e-05, "loss": 1.3027, "step": 14373 }, { "epoch": 35.935, "grad_norm": 5.78125, "learning_rate": 1.4830303263078028e-05, "loss": 1.3176, "step": 14374 }, { "epoch": 35.9375, "grad_norm": 8.6875, "learning_rate": 1.4829594280773993e-05, "loss": 1.2682, "step": 14375 }, { "epoch": 35.94, "grad_norm": 6.4375, "learning_rate": 1.4828885266807335e-05, "loss": 1.2669, "step": 14376 }, { "epoch": 35.9425, "grad_norm": 9.5625, "learning_rate": 1.4828176221182695e-05, "loss": 1.2584, "step": 14377 }, { "epoch": 35.945, "grad_norm": 11.3125, "learning_rate": 1.4827467143904728e-05, "loss": 1.3699, "step": 14378 }, { "epoch": 35.9475, "grad_norm": 11.5, "learning_rate": 1.482675803497808e-05, "loss": 1.3401, "step": 14379 }, { "epoch": 35.95, "grad_norm": 4.71875, "learning_rate": 1.4826048894407396e-05, "loss": 1.1125, "step": 14380 }, { "epoch": 35.9525, "grad_norm": 7.65625, "learning_rate": 1.4825339722197332e-05, "loss": 1.4091, "step": 14381 }, { "epoch": 35.955, "grad_norm": 6.875, "learning_rate": 1.4824630518352531e-05, "loss": 1.3002, "step": 14382 }, { "epoch": 35.9575, "grad_norm": 6.40625, "learning_rate": 1.4823921282877647e-05, "loss": 1.2585, "step": 14383 }, { "epoch": 35.96, "grad_norm": 6.75, "learning_rate": 1.482321201577733e-05, "loss": 1.3715, "step": 14384 }, { "epoch": 35.9625, "grad_norm": 8.375, "learning_rate": 1.4822502717056224e-05, "loss": 1.3981, "step": 14385 }, { "epoch": 35.965, "grad_norm": 8.625, "learning_rate": 1.4821793386718986e-05, "loss": 1.2849, "step": 14386 }, { "epoch": 35.9675, "grad_norm": 10.5, "learning_rate": 1.4821084024770263e-05, "loss": 1.4577, "step": 14387 }, { "epoch": 35.97, "grad_norm": 6.0, "learning_rate": 1.4820374631214704e-05, "loss": 1.3351, "step": 14388 }, { "epoch": 35.9725, "grad_norm": 6.875, "learning_rate": 1.4819665206056965e-05, "loss": 1.3611, "step": 14389 }, { "epoch": 35.975, "grad_norm": 7.625, "learning_rate": 1.4818955749301691e-05, "loss": 1.4867, "step": 14390 }, { "epoch": 35.9775, "grad_norm": 6.03125, "learning_rate": 1.481824626095354e-05, "loss": 1.5308, "step": 14391 }, { "epoch": 35.98, "grad_norm": 6.28125, "learning_rate": 1.4817536741017153e-05, "loss": 1.4111, "step": 14392 }, { "epoch": 35.9825, "grad_norm": 8.375, "learning_rate": 1.4816827189497194e-05, "loss": 1.3962, "step": 14393 }, { "epoch": 35.985, "grad_norm": 6.8125, "learning_rate": 1.4816117606398304e-05, "loss": 1.2279, "step": 14394 }, { "epoch": 35.9875, "grad_norm": 8.75, "learning_rate": 1.4815407991725144e-05, "loss": 1.5031, "step": 14395 }, { "epoch": 35.99, "grad_norm": 8.0, "learning_rate": 1.4814698345482359e-05, "loss": 1.2883, "step": 14396 }, { "epoch": 35.9925, "grad_norm": 9.4375, "learning_rate": 1.4813988667674603e-05, "loss": 1.3261, "step": 14397 }, { "epoch": 35.995, "grad_norm": 7.8125, "learning_rate": 1.4813278958306535e-05, "loss": 1.407, "step": 14398 }, { "epoch": 35.9975, "grad_norm": 5.71875, "learning_rate": 1.48125692173828e-05, "loss": 1.4101, "step": 14399 }, { "epoch": 36.0, "grad_norm": 8.9375, "learning_rate": 1.4811859444908053e-05, "loss": 1.3618, "step": 14400 }, { "epoch": 36.0025, "grad_norm": 6.15625, "learning_rate": 1.4811149640886948e-05, "loss": 1.4352, "step": 14401 }, { "epoch": 36.005, "grad_norm": 6.65625, "learning_rate": 1.4810439805324137e-05, "loss": 1.3411, "step": 14402 }, { "epoch": 36.0075, "grad_norm": 6.59375, "learning_rate": 1.480972993822428e-05, "loss": 1.41, "step": 14403 }, { "epoch": 36.01, "grad_norm": 8.875, "learning_rate": 1.4809020039592024e-05, "loss": 1.3572, "step": 14404 }, { "epoch": 36.0125, "grad_norm": 6.59375, "learning_rate": 1.4808310109432023e-05, "loss": 1.3008, "step": 14405 }, { "epoch": 36.015, "grad_norm": 6.46875, "learning_rate": 1.4807600147748935e-05, "loss": 1.3873, "step": 14406 }, { "epoch": 36.0175, "grad_norm": 6.59375, "learning_rate": 1.480689015454741e-05, "loss": 1.38, "step": 14407 }, { "epoch": 36.02, "grad_norm": 10.9375, "learning_rate": 1.4806180129832108e-05, "loss": 1.294, "step": 14408 }, { "epoch": 36.0225, "grad_norm": 12.9375, "learning_rate": 1.4805470073607682e-05, "loss": 1.1827, "step": 14409 }, { "epoch": 36.025, "grad_norm": 5.9375, "learning_rate": 1.4804759985878787e-05, "loss": 1.3146, "step": 14410 }, { "epoch": 36.0275, "grad_norm": 27.75, "learning_rate": 1.4804049866650076e-05, "loss": 1.3309, "step": 14411 }, { "epoch": 36.03, "grad_norm": 6.21875, "learning_rate": 1.4803339715926206e-05, "loss": 1.2771, "step": 14412 }, { "epoch": 36.0325, "grad_norm": 9.8125, "learning_rate": 1.4802629533711834e-05, "loss": 1.3391, "step": 14413 }, { "epoch": 36.035, "grad_norm": 9.875, "learning_rate": 1.4801919320011617e-05, "loss": 1.3386, "step": 14414 }, { "epoch": 36.0375, "grad_norm": 7.46875, "learning_rate": 1.4801209074830205e-05, "loss": 1.3351, "step": 14415 }, { "epoch": 36.04, "grad_norm": 12.1875, "learning_rate": 1.4800498798172263e-05, "loss": 1.3767, "step": 14416 }, { "epoch": 36.0425, "grad_norm": 8.875, "learning_rate": 1.4799788490042439e-05, "loss": 1.207, "step": 14417 }, { "epoch": 36.045, "grad_norm": 6.96875, "learning_rate": 1.4799078150445396e-05, "loss": 1.3025, "step": 14418 }, { "epoch": 36.0475, "grad_norm": 5.90625, "learning_rate": 1.479836777938579e-05, "loss": 1.4054, "step": 14419 }, { "epoch": 36.05, "grad_norm": 6.28125, "learning_rate": 1.4797657376868273e-05, "loss": 1.3209, "step": 14420 }, { "epoch": 36.0525, "grad_norm": 9.125, "learning_rate": 1.4796946942897508e-05, "loss": 1.3822, "step": 14421 }, { "epoch": 36.055, "grad_norm": 8.6875, "learning_rate": 1.4796236477478152e-05, "loss": 1.2943, "step": 14422 }, { "epoch": 36.0575, "grad_norm": 10.625, "learning_rate": 1.479552598061486e-05, "loss": 1.2204, "step": 14423 }, { "epoch": 36.06, "grad_norm": 8.125, "learning_rate": 1.479481545231229e-05, "loss": 1.3512, "step": 14424 }, { "epoch": 36.0625, "grad_norm": 8.5, "learning_rate": 1.4794104892575106e-05, "loss": 1.3834, "step": 14425 }, { "epoch": 36.065, "grad_norm": 6.875, "learning_rate": 1.479339430140796e-05, "loss": 1.2975, "step": 14426 }, { "epoch": 36.0675, "grad_norm": 8.0, "learning_rate": 1.479268367881551e-05, "loss": 1.2341, "step": 14427 }, { "epoch": 36.07, "grad_norm": 5.15625, "learning_rate": 1.4791973024802422e-05, "loss": 1.2888, "step": 14428 }, { "epoch": 36.0725, "grad_norm": 7.09375, "learning_rate": 1.4791262339373348e-05, "loss": 1.446, "step": 14429 }, { "epoch": 36.075, "grad_norm": 8.0625, "learning_rate": 1.4790551622532951e-05, "loss": 1.3016, "step": 14430 }, { "epoch": 36.0775, "grad_norm": 6.09375, "learning_rate": 1.478984087428589e-05, "loss": 1.3464, "step": 14431 }, { "epoch": 36.08, "grad_norm": 5.71875, "learning_rate": 1.478913009463682e-05, "loss": 1.2616, "step": 14432 }, { "epoch": 36.0825, "grad_norm": 13.75, "learning_rate": 1.4788419283590411e-05, "loss": 1.3753, "step": 14433 }, { "epoch": 36.085, "grad_norm": 7.96875, "learning_rate": 1.4787708441151314e-05, "loss": 1.35, "step": 14434 }, { "epoch": 36.0875, "grad_norm": 10.25, "learning_rate": 1.478699756732419e-05, "loss": 1.3603, "step": 14435 }, { "epoch": 36.09, "grad_norm": 7.8125, "learning_rate": 1.4786286662113704e-05, "loss": 1.2845, "step": 14436 }, { "epoch": 36.0925, "grad_norm": 7.21875, "learning_rate": 1.4785575725524512e-05, "loss": 1.3068, "step": 14437 }, { "epoch": 36.095, "grad_norm": 6.71875, "learning_rate": 1.4784864757561278e-05, "loss": 1.3171, "step": 14438 }, { "epoch": 36.0975, "grad_norm": 7.5625, "learning_rate": 1.4784153758228662e-05, "loss": 1.3213, "step": 14439 }, { "epoch": 36.1, "grad_norm": 6.71875, "learning_rate": 1.4783442727531328e-05, "loss": 1.2186, "step": 14440 }, { "epoch": 36.1025, "grad_norm": 8.0, "learning_rate": 1.4782731665473933e-05, "loss": 1.2575, "step": 14441 }, { "epoch": 36.105, "grad_norm": 6.03125, "learning_rate": 1.478202057206114e-05, "loss": 1.3348, "step": 14442 }, { "epoch": 36.1075, "grad_norm": 8.125, "learning_rate": 1.4781309447297611e-05, "loss": 1.2876, "step": 14443 }, { "epoch": 36.11, "grad_norm": 7.96875, "learning_rate": 1.4780598291188011e-05, "loss": 1.2193, "step": 14444 }, { "epoch": 36.1125, "grad_norm": 9.375, "learning_rate": 1.4779887103736996e-05, "loss": 1.3884, "step": 14445 }, { "epoch": 36.115, "grad_norm": 6.90625, "learning_rate": 1.4779175884949238e-05, "loss": 1.3077, "step": 14446 }, { "epoch": 36.1175, "grad_norm": 7.0, "learning_rate": 1.4778464634829389e-05, "loss": 1.4855, "step": 14447 }, { "epoch": 36.12, "grad_norm": 6.875, "learning_rate": 1.4777753353382121e-05, "loss": 1.3702, "step": 14448 }, { "epoch": 36.1225, "grad_norm": 6.71875, "learning_rate": 1.477704204061209e-05, "loss": 1.3554, "step": 14449 }, { "epoch": 36.125, "grad_norm": 8.8125, "learning_rate": 1.4776330696523964e-05, "loss": 1.3759, "step": 14450 }, { "epoch": 36.1275, "grad_norm": 7.40625, "learning_rate": 1.4775619321122404e-05, "loss": 1.2512, "step": 14451 }, { "epoch": 36.13, "grad_norm": 9.375, "learning_rate": 1.4774907914412076e-05, "loss": 1.4908, "step": 14452 }, { "epoch": 36.1325, "grad_norm": 8.625, "learning_rate": 1.4774196476397642e-05, "loss": 1.331, "step": 14453 }, { "epoch": 36.135, "grad_norm": 7.03125, "learning_rate": 1.477348500708377e-05, "loss": 1.4019, "step": 14454 }, { "epoch": 36.1375, "grad_norm": 6.8125, "learning_rate": 1.4772773506475116e-05, "loss": 1.3772, "step": 14455 }, { "epoch": 36.14, "grad_norm": 6.75, "learning_rate": 1.4772061974576353e-05, "loss": 1.3334, "step": 14456 }, { "epoch": 36.1425, "grad_norm": 8.3125, "learning_rate": 1.4771350411392143e-05, "loss": 1.3112, "step": 14457 }, { "epoch": 36.145, "grad_norm": 8.375, "learning_rate": 1.4770638816927148e-05, "loss": 1.2519, "step": 14458 }, { "epoch": 36.1475, "grad_norm": 76.5, "learning_rate": 1.476992719118604e-05, "loss": 1.4258, "step": 14459 }, { "epoch": 36.15, "grad_norm": 6.0625, "learning_rate": 1.4769215534173476e-05, "loss": 1.4049, "step": 14460 }, { "epoch": 36.1525, "grad_norm": 7.53125, "learning_rate": 1.4768503845894131e-05, "loss": 1.4717, "step": 14461 }, { "epoch": 36.155, "grad_norm": 13.4375, "learning_rate": 1.476779212635266e-05, "loss": 1.4265, "step": 14462 }, { "epoch": 36.1575, "grad_norm": 8.375, "learning_rate": 1.476708037555374e-05, "loss": 1.3491, "step": 14463 }, { "epoch": 36.16, "grad_norm": 11.6875, "learning_rate": 1.4766368593502028e-05, "loss": 1.5173, "step": 14464 }, { "epoch": 36.1625, "grad_norm": 7.09375, "learning_rate": 1.4765656780202193e-05, "loss": 1.3651, "step": 14465 }, { "epoch": 36.165, "grad_norm": 7.0625, "learning_rate": 1.4764944935658908e-05, "loss": 1.4205, "step": 14466 }, { "epoch": 36.1675, "grad_norm": 7.03125, "learning_rate": 1.4764233059876829e-05, "loss": 1.4083, "step": 14467 }, { "epoch": 36.17, "grad_norm": 5.59375, "learning_rate": 1.4763521152860632e-05, "loss": 1.1707, "step": 14468 }, { "epoch": 36.1725, "grad_norm": 7.125, "learning_rate": 1.476280921461498e-05, "loss": 1.3411, "step": 14469 }, { "epoch": 36.175, "grad_norm": 9.1875, "learning_rate": 1.4762097245144543e-05, "loss": 1.336, "step": 14470 }, { "epoch": 36.1775, "grad_norm": 7.9375, "learning_rate": 1.4761385244453984e-05, "loss": 1.3433, "step": 14471 }, { "epoch": 36.18, "grad_norm": 8.75, "learning_rate": 1.4760673212547975e-05, "loss": 1.4498, "step": 14472 }, { "epoch": 36.1825, "grad_norm": 6.5, "learning_rate": 1.4759961149431185e-05, "loss": 1.423, "step": 14473 }, { "epoch": 36.185, "grad_norm": 5.34375, "learning_rate": 1.4759249055108279e-05, "loss": 1.479, "step": 14474 }, { "epoch": 36.1875, "grad_norm": 6.59375, "learning_rate": 1.4758536929583926e-05, "loss": 1.2873, "step": 14475 }, { "epoch": 36.19, "grad_norm": 6.46875, "learning_rate": 1.4757824772862797e-05, "loss": 1.3068, "step": 14476 }, { "epoch": 36.1925, "grad_norm": 8.625, "learning_rate": 1.4757112584949558e-05, "loss": 1.3179, "step": 14477 }, { "epoch": 36.195, "grad_norm": 5.28125, "learning_rate": 1.4756400365848878e-05, "loss": 1.2994, "step": 14478 }, { "epoch": 36.1975, "grad_norm": 6.90625, "learning_rate": 1.475568811556543e-05, "loss": 1.4384, "step": 14479 }, { "epoch": 36.2, "grad_norm": 6.96875, "learning_rate": 1.4754975834103877e-05, "loss": 1.2704, "step": 14480 }, { "epoch": 36.2025, "grad_norm": 5.5625, "learning_rate": 1.4754263521468896e-05, "loss": 1.3237, "step": 14481 }, { "epoch": 36.205, "grad_norm": 8.625, "learning_rate": 1.4753551177665154e-05, "loss": 1.2543, "step": 14482 }, { "epoch": 36.2075, "grad_norm": 7.34375, "learning_rate": 1.4752838802697318e-05, "loss": 1.3662, "step": 14483 }, { "epoch": 36.21, "grad_norm": 7.59375, "learning_rate": 1.4752126396570064e-05, "loss": 1.3513, "step": 14484 }, { "epoch": 36.2125, "grad_norm": 7.21875, "learning_rate": 1.4751413959288057e-05, "loss": 1.4841, "step": 14485 }, { "epoch": 36.215, "grad_norm": 9.1875, "learning_rate": 1.4750701490855974e-05, "loss": 1.2893, "step": 14486 }, { "epoch": 36.2175, "grad_norm": 7.3125, "learning_rate": 1.4749988991278477e-05, "loss": 1.2451, "step": 14487 }, { "epoch": 36.22, "grad_norm": 6.9375, "learning_rate": 1.4749276460560247e-05, "loss": 1.381, "step": 14488 }, { "epoch": 36.2225, "grad_norm": 8.9375, "learning_rate": 1.4748563898705945e-05, "loss": 1.353, "step": 14489 }, { "epoch": 36.225, "grad_norm": 6.125, "learning_rate": 1.4747851305720255e-05, "loss": 1.4034, "step": 14490 }, { "epoch": 36.2275, "grad_norm": 10.125, "learning_rate": 1.4747138681607838e-05, "loss": 1.447, "step": 14491 }, { "epoch": 36.23, "grad_norm": 11.875, "learning_rate": 1.474642602637337e-05, "loss": 1.3371, "step": 14492 }, { "epoch": 36.2325, "grad_norm": 7.09375, "learning_rate": 1.4745713340021523e-05, "loss": 1.5065, "step": 14493 }, { "epoch": 36.235, "grad_norm": 7.03125, "learning_rate": 1.4745000622556968e-05, "loss": 1.2703, "step": 14494 }, { "epoch": 36.2375, "grad_norm": 13.125, "learning_rate": 1.4744287873984382e-05, "loss": 1.4291, "step": 14495 }, { "epoch": 36.24, "grad_norm": 6.625, "learning_rate": 1.474357509430843e-05, "loss": 1.3832, "step": 14496 }, { "epoch": 36.2425, "grad_norm": 6.875, "learning_rate": 1.4742862283533795e-05, "loss": 1.3182, "step": 14497 }, { "epoch": 36.245, "grad_norm": 5.90625, "learning_rate": 1.4742149441665142e-05, "loss": 1.2504, "step": 14498 }, { "epoch": 36.2475, "grad_norm": 6.3125, "learning_rate": 1.4741436568707146e-05, "loss": 1.3459, "step": 14499 }, { "epoch": 36.25, "grad_norm": 8.0625, "learning_rate": 1.4740723664664483e-05, "loss": 1.397, "step": 14500 }, { "epoch": 36.2525, "grad_norm": 6.34375, "learning_rate": 1.4740010729541826e-05, "loss": 1.2799, "step": 14501 }, { "epoch": 36.255, "grad_norm": 8.75, "learning_rate": 1.4739297763343846e-05, "loss": 1.2207, "step": 14502 }, { "epoch": 36.2575, "grad_norm": 7.84375, "learning_rate": 1.4738584766075221e-05, "loss": 1.4158, "step": 14503 }, { "epoch": 36.26, "grad_norm": 6.59375, "learning_rate": 1.473787173774062e-05, "loss": 1.3554, "step": 14504 }, { "epoch": 36.2625, "grad_norm": 8.6875, "learning_rate": 1.4737158678344727e-05, "loss": 1.3165, "step": 14505 }, { "epoch": 36.265, "grad_norm": 11.9375, "learning_rate": 1.4736445587892209e-05, "loss": 1.3837, "step": 14506 }, { "epoch": 36.2675, "grad_norm": 11.375, "learning_rate": 1.4735732466387745e-05, "loss": 1.4096, "step": 14507 }, { "epoch": 36.27, "grad_norm": 12.25, "learning_rate": 1.4735019313836008e-05, "loss": 1.4232, "step": 14508 }, { "epoch": 36.2725, "grad_norm": 5.34375, "learning_rate": 1.4734306130241672e-05, "loss": 1.298, "step": 14509 }, { "epoch": 36.275, "grad_norm": 8.75, "learning_rate": 1.4733592915609416e-05, "loss": 1.4385, "step": 14510 }, { "epoch": 36.2775, "grad_norm": 7.875, "learning_rate": 1.4732879669943913e-05, "loss": 1.4381, "step": 14511 }, { "epoch": 36.28, "grad_norm": 6.3125, "learning_rate": 1.473216639324984e-05, "loss": 1.4058, "step": 14512 }, { "epoch": 36.2825, "grad_norm": 8.625, "learning_rate": 1.4731453085531875e-05, "loss": 1.339, "step": 14513 }, { "epoch": 36.285, "grad_norm": 5.8125, "learning_rate": 1.4730739746794693e-05, "loss": 1.3409, "step": 14514 }, { "epoch": 36.2875, "grad_norm": 5.90625, "learning_rate": 1.473002637704297e-05, "loss": 1.2753, "step": 14515 }, { "epoch": 36.29, "grad_norm": 5.875, "learning_rate": 1.4729312976281385e-05, "loss": 1.2875, "step": 14516 }, { "epoch": 36.2925, "grad_norm": 10.3125, "learning_rate": 1.4728599544514607e-05, "loss": 1.2382, "step": 14517 }, { "epoch": 36.295, "grad_norm": 6.78125, "learning_rate": 1.472788608174733e-05, "loss": 1.3683, "step": 14518 }, { "epoch": 36.2975, "grad_norm": 7.6875, "learning_rate": 1.472717258798421e-05, "loss": 1.4737, "step": 14519 }, { "epoch": 36.3, "grad_norm": 8.25, "learning_rate": 1.4726459063229946e-05, "loss": 1.4177, "step": 14520 }, { "epoch": 36.3025, "grad_norm": 7.125, "learning_rate": 1.4725745507489197e-05, "loss": 1.2845, "step": 14521 }, { "epoch": 36.305, "grad_norm": 7.25, "learning_rate": 1.4725031920766654e-05, "loss": 1.2367, "step": 14522 }, { "epoch": 36.3075, "grad_norm": 9.25, "learning_rate": 1.472431830306699e-05, "loss": 1.3619, "step": 14523 }, { "epoch": 36.31, "grad_norm": 9.1875, "learning_rate": 1.4723604654394885e-05, "loss": 1.4286, "step": 14524 }, { "epoch": 36.3125, "grad_norm": 6.53125, "learning_rate": 1.4722890974755014e-05, "loss": 1.322, "step": 14525 }, { "epoch": 36.315, "grad_norm": 7.9375, "learning_rate": 1.4722177264152063e-05, "loss": 1.5963, "step": 14526 }, { "epoch": 36.3175, "grad_norm": 7.15625, "learning_rate": 1.4721463522590704e-05, "loss": 1.4306, "step": 14527 }, { "epoch": 36.32, "grad_norm": 7.5, "learning_rate": 1.472074975007562e-05, "loss": 1.417, "step": 14528 }, { "epoch": 36.3225, "grad_norm": 7.03125, "learning_rate": 1.4720035946611488e-05, "loss": 1.3947, "step": 14529 }, { "epoch": 36.325, "grad_norm": 9.75, "learning_rate": 1.4719322112202992e-05, "loss": 1.409, "step": 14530 }, { "epoch": 36.3275, "grad_norm": 7.53125, "learning_rate": 1.4718608246854808e-05, "loss": 1.3356, "step": 14531 }, { "epoch": 36.33, "grad_norm": 7.25, "learning_rate": 1.4717894350571616e-05, "loss": 1.4114, "step": 14532 }, { "epoch": 36.3325, "grad_norm": 5.03125, "learning_rate": 1.47171804233581e-05, "loss": 1.4579, "step": 14533 }, { "epoch": 36.335, "grad_norm": 9.875, "learning_rate": 1.4716466465218932e-05, "loss": 1.3588, "step": 14534 }, { "epoch": 36.3375, "grad_norm": 6.53125, "learning_rate": 1.4715752476158807e-05, "loss": 1.3775, "step": 14535 }, { "epoch": 36.34, "grad_norm": 5.96875, "learning_rate": 1.4715038456182394e-05, "loss": 1.3149, "step": 14536 }, { "epoch": 36.3425, "grad_norm": 8.5625, "learning_rate": 1.4714324405294376e-05, "loss": 1.3203, "step": 14537 }, { "epoch": 36.345, "grad_norm": 5.625, "learning_rate": 1.4713610323499436e-05, "loss": 1.1985, "step": 14538 }, { "epoch": 36.3475, "grad_norm": 8.125, "learning_rate": 1.4712896210802257e-05, "loss": 1.3879, "step": 14539 }, { "epoch": 36.35, "grad_norm": 6.1875, "learning_rate": 1.4712182067207516e-05, "loss": 1.1173, "step": 14540 }, { "epoch": 36.3525, "grad_norm": 5.40625, "learning_rate": 1.4711467892719901e-05, "loss": 1.3807, "step": 14541 }, { "epoch": 36.355, "grad_norm": 6.125, "learning_rate": 1.4710753687344089e-05, "loss": 1.4744, "step": 14542 }, { "epoch": 36.3575, "grad_norm": 5.59375, "learning_rate": 1.4710039451084768e-05, "loss": 1.3335, "step": 14543 }, { "epoch": 36.36, "grad_norm": 5.15625, "learning_rate": 1.4709325183946613e-05, "loss": 1.2916, "step": 14544 }, { "epoch": 36.3625, "grad_norm": 6.0625, "learning_rate": 1.4708610885934312e-05, "loss": 1.3862, "step": 14545 }, { "epoch": 36.365, "grad_norm": 8.625, "learning_rate": 1.4707896557052545e-05, "loss": 1.334, "step": 14546 }, { "epoch": 36.3675, "grad_norm": 8.6875, "learning_rate": 1.4707182197305998e-05, "loss": 1.3028, "step": 14547 }, { "epoch": 36.37, "grad_norm": 11.3125, "learning_rate": 1.4706467806699354e-05, "loss": 1.4218, "step": 14548 }, { "epoch": 36.3725, "grad_norm": 9.4375, "learning_rate": 1.4705753385237293e-05, "loss": 1.457, "step": 14549 }, { "epoch": 36.375, "grad_norm": 5.8125, "learning_rate": 1.4705038932924502e-05, "loss": 1.4481, "step": 14550 }, { "epoch": 36.3775, "grad_norm": 9.4375, "learning_rate": 1.4704324449765665e-05, "loss": 1.3479, "step": 14551 }, { "epoch": 36.38, "grad_norm": 6.875, "learning_rate": 1.4703609935765463e-05, "loss": 1.4593, "step": 14552 }, { "epoch": 36.3825, "grad_norm": 8.0625, "learning_rate": 1.4702895390928584e-05, "loss": 1.5651, "step": 14553 }, { "epoch": 36.385, "grad_norm": 6.875, "learning_rate": 1.470218081525971e-05, "loss": 1.2165, "step": 14554 }, { "epoch": 36.3875, "grad_norm": 8.1875, "learning_rate": 1.4701466208763528e-05, "loss": 1.3183, "step": 14555 }, { "epoch": 36.39, "grad_norm": 5.625, "learning_rate": 1.4700751571444724e-05, "loss": 1.3124, "step": 14556 }, { "epoch": 36.3925, "grad_norm": 14.6875, "learning_rate": 1.4700036903307977e-05, "loss": 1.3721, "step": 14557 }, { "epoch": 36.395, "grad_norm": 7.59375, "learning_rate": 1.4699322204357978e-05, "loss": 1.3359, "step": 14558 }, { "epoch": 36.3975, "grad_norm": 7.59375, "learning_rate": 1.4698607474599409e-05, "loss": 1.3672, "step": 14559 }, { "epoch": 36.4, "grad_norm": 6.03125, "learning_rate": 1.4697892714036959e-05, "loss": 1.2993, "step": 14560 }, { "epoch": 36.4025, "grad_norm": 5.84375, "learning_rate": 1.4697177922675311e-05, "loss": 1.3408, "step": 14561 }, { "epoch": 36.405, "grad_norm": 9.4375, "learning_rate": 1.4696463100519154e-05, "loss": 1.4198, "step": 14562 }, { "epoch": 36.4075, "grad_norm": 6.0625, "learning_rate": 1.4695748247573172e-05, "loss": 1.2036, "step": 14563 }, { "epoch": 36.41, "grad_norm": 5.71875, "learning_rate": 1.4695033363842053e-05, "loss": 1.2302, "step": 14564 }, { "epoch": 36.4125, "grad_norm": 8.6875, "learning_rate": 1.4694318449330482e-05, "loss": 1.3619, "step": 14565 }, { "epoch": 36.415, "grad_norm": 8.0625, "learning_rate": 1.4693603504043148e-05, "loss": 1.3362, "step": 14566 }, { "epoch": 36.4175, "grad_norm": 5.84375, "learning_rate": 1.4692888527984737e-05, "loss": 1.3232, "step": 14567 }, { "epoch": 36.42, "grad_norm": 8.5625, "learning_rate": 1.4692173521159936e-05, "loss": 1.3134, "step": 14568 }, { "epoch": 36.4225, "grad_norm": 7.78125, "learning_rate": 1.4691458483573434e-05, "loss": 1.2629, "step": 14569 }, { "epoch": 36.425, "grad_norm": 9.3125, "learning_rate": 1.4690743415229917e-05, "loss": 1.361, "step": 14570 }, { "epoch": 36.4275, "grad_norm": 10.75, "learning_rate": 1.4690028316134077e-05, "loss": 1.373, "step": 14571 }, { "epoch": 36.43, "grad_norm": 8.125, "learning_rate": 1.4689313186290594e-05, "loss": 1.2225, "step": 14572 }, { "epoch": 36.4325, "grad_norm": 7.21875, "learning_rate": 1.4688598025704167e-05, "loss": 1.3103, "step": 14573 }, { "epoch": 36.435, "grad_norm": 5.84375, "learning_rate": 1.4687882834379474e-05, "loss": 1.3177, "step": 14574 }, { "epoch": 36.4375, "grad_norm": 9.3125, "learning_rate": 1.4687167612321212e-05, "loss": 1.3163, "step": 14575 }, { "epoch": 36.44, "grad_norm": 6.28125, "learning_rate": 1.4686452359534067e-05, "loss": 1.3295, "step": 14576 }, { "epoch": 36.4425, "grad_norm": 8.8125, "learning_rate": 1.4685737076022725e-05, "loss": 1.2377, "step": 14577 }, { "epoch": 36.445, "grad_norm": 5.8125, "learning_rate": 1.468502176179188e-05, "loss": 1.207, "step": 14578 }, { "epoch": 36.4475, "grad_norm": 9.375, "learning_rate": 1.4684306416846219e-05, "loss": 1.3677, "step": 14579 }, { "epoch": 36.45, "grad_norm": 8.6875, "learning_rate": 1.4683591041190433e-05, "loss": 1.3243, "step": 14580 }, { "epoch": 36.4525, "grad_norm": 8.5625, "learning_rate": 1.468287563482921e-05, "loss": 1.2762, "step": 14581 }, { "epoch": 36.455, "grad_norm": 8.375, "learning_rate": 1.4682160197767243e-05, "loss": 1.3116, "step": 14582 }, { "epoch": 36.4575, "grad_norm": 5.03125, "learning_rate": 1.4681444730009222e-05, "loss": 1.4871, "step": 14583 }, { "epoch": 36.46, "grad_norm": 7.96875, "learning_rate": 1.4680729231559835e-05, "loss": 1.5689, "step": 14584 }, { "epoch": 36.4625, "grad_norm": 7.15625, "learning_rate": 1.4680013702423777e-05, "loss": 1.1972, "step": 14585 }, { "epoch": 36.465, "grad_norm": 6.53125, "learning_rate": 1.4679298142605735e-05, "loss": 1.4517, "step": 14586 }, { "epoch": 36.4675, "grad_norm": 7.875, "learning_rate": 1.4678582552110401e-05, "loss": 1.2147, "step": 14587 }, { "epoch": 36.47, "grad_norm": 6.375, "learning_rate": 1.467786693094247e-05, "loss": 1.2491, "step": 14588 }, { "epoch": 36.4725, "grad_norm": 8.5, "learning_rate": 1.4677151279106624e-05, "loss": 1.4236, "step": 14589 }, { "epoch": 36.475, "grad_norm": 7.0, "learning_rate": 1.467643559660757e-05, "loss": 1.3659, "step": 14590 }, { "epoch": 36.4775, "grad_norm": 7.03125, "learning_rate": 1.4675719883449986e-05, "loss": 1.4426, "step": 14591 }, { "epoch": 36.48, "grad_norm": 7.4375, "learning_rate": 1.467500413963857e-05, "loss": 1.3522, "step": 14592 }, { "epoch": 36.4825, "grad_norm": 7.625, "learning_rate": 1.4674288365178016e-05, "loss": 1.3664, "step": 14593 }, { "epoch": 36.485, "grad_norm": 8.9375, "learning_rate": 1.467357256007301e-05, "loss": 1.2685, "step": 14594 }, { "epoch": 36.4875, "grad_norm": 5.8125, "learning_rate": 1.4672856724328256e-05, "loss": 1.3869, "step": 14595 }, { "epoch": 36.49, "grad_norm": 7.71875, "learning_rate": 1.4672140857948436e-05, "loss": 1.4064, "step": 14596 }, { "epoch": 36.4925, "grad_norm": 7.1875, "learning_rate": 1.4671424960938247e-05, "loss": 1.4103, "step": 14597 }, { "epoch": 36.495, "grad_norm": 6.59375, "learning_rate": 1.4670709033302384e-05, "loss": 1.2984, "step": 14598 }, { "epoch": 36.4975, "grad_norm": 8.125, "learning_rate": 1.466999307504554e-05, "loss": 1.3838, "step": 14599 }, { "epoch": 36.5, "grad_norm": 7.84375, "learning_rate": 1.4669277086172406e-05, "loss": 1.4551, "step": 14600 }, { "epoch": 36.5025, "grad_norm": 7.6875, "learning_rate": 1.466856106668768e-05, "loss": 1.3975, "step": 14601 }, { "epoch": 36.505, "grad_norm": 6.40625, "learning_rate": 1.4667845016596056e-05, "loss": 1.278, "step": 14602 }, { "epoch": 36.5075, "grad_norm": 6.125, "learning_rate": 1.4667128935902227e-05, "loss": 1.3446, "step": 14603 }, { "epoch": 36.51, "grad_norm": 5.84375, "learning_rate": 1.4666412824610883e-05, "loss": 1.1808, "step": 14604 }, { "epoch": 36.5125, "grad_norm": 6.4375, "learning_rate": 1.466569668272673e-05, "loss": 1.2867, "step": 14605 }, { "epoch": 36.515, "grad_norm": 8.125, "learning_rate": 1.466498051025445e-05, "loss": 1.2059, "step": 14606 }, { "epoch": 36.5175, "grad_norm": 8.3125, "learning_rate": 1.4664264307198748e-05, "loss": 1.3552, "step": 14607 }, { "epoch": 36.52, "grad_norm": 7.96875, "learning_rate": 1.4663548073564316e-05, "loss": 1.2922, "step": 14608 }, { "epoch": 36.5225, "grad_norm": 9.6875, "learning_rate": 1.4662831809355847e-05, "loss": 1.3599, "step": 14609 }, { "epoch": 36.525, "grad_norm": 12.4375, "learning_rate": 1.4662115514578039e-05, "loss": 1.3956, "step": 14610 }, { "epoch": 36.5275, "grad_norm": 7.875, "learning_rate": 1.466139918923559e-05, "loss": 1.4235, "step": 14611 }, { "epoch": 36.53, "grad_norm": 6.59375, "learning_rate": 1.4660682833333195e-05, "loss": 1.3415, "step": 14612 }, { "epoch": 36.5325, "grad_norm": 7.46875, "learning_rate": 1.465996644687555e-05, "loss": 1.3691, "step": 14613 }, { "epoch": 36.535, "grad_norm": 11.3125, "learning_rate": 1.465925002986735e-05, "loss": 1.341, "step": 14614 }, { "epoch": 36.5375, "grad_norm": 7.0, "learning_rate": 1.4658533582313295e-05, "loss": 1.4364, "step": 14615 }, { "epoch": 36.54, "grad_norm": 8.25, "learning_rate": 1.4657817104218075e-05, "loss": 1.3495, "step": 14616 }, { "epoch": 36.5425, "grad_norm": 8.625, "learning_rate": 1.46571005955864e-05, "loss": 1.4988, "step": 14617 }, { "epoch": 36.545, "grad_norm": 6.78125, "learning_rate": 1.4656384056422957e-05, "loss": 1.1953, "step": 14618 }, { "epoch": 36.5475, "grad_norm": 9.1875, "learning_rate": 1.4655667486732444e-05, "loss": 1.4538, "step": 14619 }, { "epoch": 36.55, "grad_norm": 10.75, "learning_rate": 1.4654950886519563e-05, "loss": 1.3377, "step": 14620 }, { "epoch": 36.5525, "grad_norm": 9.0, "learning_rate": 1.4654234255789007e-05, "loss": 1.2584, "step": 14621 }, { "epoch": 36.555, "grad_norm": 7.4375, "learning_rate": 1.4653517594545483e-05, "loss": 1.3047, "step": 14622 }, { "epoch": 36.5575, "grad_norm": 9.3125, "learning_rate": 1.4652800902793683e-05, "loss": 1.282, "step": 14623 }, { "epoch": 36.56, "grad_norm": 8.75, "learning_rate": 1.4652084180538304e-05, "loss": 1.3718, "step": 14624 }, { "epoch": 36.5625, "grad_norm": 10.5, "learning_rate": 1.4651367427784049e-05, "loss": 1.2695, "step": 14625 }, { "epoch": 36.565, "grad_norm": 8.25, "learning_rate": 1.4650650644535613e-05, "loss": 1.4833, "step": 14626 }, { "epoch": 36.5675, "grad_norm": 10.3125, "learning_rate": 1.46499338307977e-05, "loss": 1.4592, "step": 14627 }, { "epoch": 36.57, "grad_norm": 10.0625, "learning_rate": 1.4649216986575007e-05, "loss": 1.3518, "step": 14628 }, { "epoch": 36.5725, "grad_norm": 6.46875, "learning_rate": 1.4648500111872233e-05, "loss": 1.3251, "step": 14629 }, { "epoch": 36.575, "grad_norm": 10.4375, "learning_rate": 1.464778320669408e-05, "loss": 1.3858, "step": 14630 }, { "epoch": 36.5775, "grad_norm": 8.25, "learning_rate": 1.464706627104524e-05, "loss": 1.3326, "step": 14631 }, { "epoch": 36.58, "grad_norm": 6.8125, "learning_rate": 1.4646349304930426e-05, "loss": 1.2713, "step": 14632 }, { "epoch": 36.5825, "grad_norm": 15.6875, "learning_rate": 1.464563230835433e-05, "loss": 1.2137, "step": 14633 }, { "epoch": 36.585, "grad_norm": 7.375, "learning_rate": 1.4644915281321655e-05, "loss": 1.2239, "step": 14634 }, { "epoch": 36.5875, "grad_norm": 7.46875, "learning_rate": 1.4644198223837098e-05, "loss": 1.3336, "step": 14635 }, { "epoch": 36.59, "grad_norm": 7.0, "learning_rate": 1.4643481135905367e-05, "loss": 1.3828, "step": 14636 }, { "epoch": 36.5925, "grad_norm": 7.40625, "learning_rate": 1.4642764017531157e-05, "loss": 1.3198, "step": 14637 }, { "epoch": 36.595, "grad_norm": 7.46875, "learning_rate": 1.4642046868719173e-05, "loss": 1.3927, "step": 14638 }, { "epoch": 36.5975, "grad_norm": 7.75, "learning_rate": 1.4641329689474115e-05, "loss": 1.3518, "step": 14639 }, { "epoch": 36.6, "grad_norm": 6.625, "learning_rate": 1.4640612479800686e-05, "loss": 1.2608, "step": 14640 }, { "epoch": 36.6025, "grad_norm": 18.0, "learning_rate": 1.4639895239703588e-05, "loss": 1.3981, "step": 14641 }, { "epoch": 36.605, "grad_norm": 9.75, "learning_rate": 1.4639177969187522e-05, "loss": 1.1584, "step": 14642 }, { "epoch": 36.6075, "grad_norm": 9.8125, "learning_rate": 1.463846066825719e-05, "loss": 1.3908, "step": 14643 }, { "epoch": 36.61, "grad_norm": 6.4375, "learning_rate": 1.4637743336917294e-05, "loss": 1.3652, "step": 14644 }, { "epoch": 36.6125, "grad_norm": 8.1875, "learning_rate": 1.4637025975172543e-05, "loss": 1.2894, "step": 14645 }, { "epoch": 36.615, "grad_norm": 7.375, "learning_rate": 1.463630858302763e-05, "loss": 1.4083, "step": 14646 }, { "epoch": 36.6175, "grad_norm": 7.0625, "learning_rate": 1.4635591160487268e-05, "loss": 1.4051, "step": 14647 }, { "epoch": 36.62, "grad_norm": 6.5625, "learning_rate": 1.4634873707556154e-05, "loss": 1.3015, "step": 14648 }, { "epoch": 36.6225, "grad_norm": 6.4375, "learning_rate": 1.463415622423899e-05, "loss": 1.3304, "step": 14649 }, { "epoch": 36.625, "grad_norm": 7.5625, "learning_rate": 1.4633438710540488e-05, "loss": 1.3303, "step": 14650 }, { "epoch": 36.6275, "grad_norm": 8.25, "learning_rate": 1.4632721166465348e-05, "loss": 1.4045, "step": 14651 }, { "epoch": 36.63, "grad_norm": 7.78125, "learning_rate": 1.4632003592018269e-05, "loss": 1.2446, "step": 14652 }, { "epoch": 36.6325, "grad_norm": 8.9375, "learning_rate": 1.4631285987203963e-05, "loss": 1.3209, "step": 14653 }, { "epoch": 36.635, "grad_norm": 6.8125, "learning_rate": 1.4630568352027131e-05, "loss": 1.4568, "step": 14654 }, { "epoch": 36.6375, "grad_norm": 6.03125, "learning_rate": 1.4629850686492475e-05, "loss": 1.3459, "step": 14655 }, { "epoch": 36.64, "grad_norm": 8.8125, "learning_rate": 1.4629132990604706e-05, "loss": 1.3866, "step": 14656 }, { "epoch": 36.6425, "grad_norm": 9.5, "learning_rate": 1.4628415264368527e-05, "loss": 1.1917, "step": 14657 }, { "epoch": 36.645, "grad_norm": 5.28125, "learning_rate": 1.4627697507788643e-05, "loss": 1.381, "step": 14658 }, { "epoch": 36.6475, "grad_norm": 8.5, "learning_rate": 1.4626979720869756e-05, "loss": 1.4196, "step": 14659 }, { "epoch": 36.65, "grad_norm": 5.65625, "learning_rate": 1.4626261903616579e-05, "loss": 1.2536, "step": 14660 }, { "epoch": 36.6525, "grad_norm": 7.03125, "learning_rate": 1.462554405603381e-05, "loss": 1.2528, "step": 14661 }, { "epoch": 36.655, "grad_norm": 5.59375, "learning_rate": 1.4624826178126164e-05, "loss": 1.3806, "step": 14662 }, { "epoch": 36.6575, "grad_norm": 5.6875, "learning_rate": 1.462410826989834e-05, "loss": 1.1821, "step": 14663 }, { "epoch": 36.66, "grad_norm": 9.3125, "learning_rate": 1.4623390331355048e-05, "loss": 1.2581, "step": 14664 }, { "epoch": 36.6625, "grad_norm": 8.9375, "learning_rate": 1.4622672362500994e-05, "loss": 1.2821, "step": 14665 }, { "epoch": 36.665, "grad_norm": 6.375, "learning_rate": 1.4621954363340884e-05, "loss": 1.2641, "step": 14666 }, { "epoch": 36.6675, "grad_norm": 6.3125, "learning_rate": 1.4621236333879427e-05, "loss": 1.1292, "step": 14667 }, { "epoch": 36.67, "grad_norm": 7.9375, "learning_rate": 1.462051827412133e-05, "loss": 1.1519, "step": 14668 }, { "epoch": 36.6725, "grad_norm": 6.28125, "learning_rate": 1.4619800184071298e-05, "loss": 1.3086, "step": 14669 }, { "epoch": 36.675, "grad_norm": 6.28125, "learning_rate": 1.461908206373404e-05, "loss": 1.3084, "step": 14670 }, { "epoch": 36.6775, "grad_norm": 7.375, "learning_rate": 1.461836391311427e-05, "loss": 1.4916, "step": 14671 }, { "epoch": 36.68, "grad_norm": 6.28125, "learning_rate": 1.4617645732216686e-05, "loss": 1.325, "step": 14672 }, { "epoch": 36.6825, "grad_norm": 6.5, "learning_rate": 1.4616927521046005e-05, "loss": 1.2798, "step": 14673 }, { "epoch": 36.685, "grad_norm": 7.4375, "learning_rate": 1.4616209279606928e-05, "loss": 1.423, "step": 14674 }, { "epoch": 36.6875, "grad_norm": 10.125, "learning_rate": 1.4615491007904172e-05, "loss": 1.3798, "step": 14675 }, { "epoch": 36.69, "grad_norm": 5.09375, "learning_rate": 1.4614772705942437e-05, "loss": 1.3386, "step": 14676 }, { "epoch": 36.6925, "grad_norm": 10.875, "learning_rate": 1.461405437372644e-05, "loss": 1.3419, "step": 14677 }, { "epoch": 36.695, "grad_norm": 7.875, "learning_rate": 1.4613336011260888e-05, "loss": 1.3539, "step": 14678 }, { "epoch": 36.6975, "grad_norm": 5.96875, "learning_rate": 1.4612617618550488e-05, "loss": 1.3063, "step": 14679 }, { "epoch": 36.7, "grad_norm": 7.1875, "learning_rate": 1.4611899195599952e-05, "loss": 1.186, "step": 14680 }, { "epoch": 36.7025, "grad_norm": 7.34375, "learning_rate": 1.4611180742413989e-05, "loss": 1.3632, "step": 14681 }, { "epoch": 36.705, "grad_norm": 9.5, "learning_rate": 1.4610462258997311e-05, "loss": 1.3684, "step": 14682 }, { "epoch": 36.7075, "grad_norm": 7.8125, "learning_rate": 1.4609743745354625e-05, "loss": 1.3595, "step": 14683 }, { "epoch": 36.71, "grad_norm": 8.8125, "learning_rate": 1.4609025201490643e-05, "loss": 1.3777, "step": 14684 }, { "epoch": 36.7125, "grad_norm": 9.875, "learning_rate": 1.460830662741008e-05, "loss": 1.3127, "step": 14685 }, { "epoch": 36.715, "grad_norm": 6.03125, "learning_rate": 1.4607588023117638e-05, "loss": 1.5111, "step": 14686 }, { "epoch": 36.7175, "grad_norm": 4.4375, "learning_rate": 1.460686938861804e-05, "loss": 1.1596, "step": 14687 }, { "epoch": 36.72, "grad_norm": 9.6875, "learning_rate": 1.4606150723915984e-05, "loss": 1.4938, "step": 14688 }, { "epoch": 36.7225, "grad_norm": 6.09375, "learning_rate": 1.4605432029016191e-05, "loss": 1.4791, "step": 14689 }, { "epoch": 36.725, "grad_norm": 9.8125, "learning_rate": 1.460471330392337e-05, "loss": 1.3358, "step": 14690 }, { "epoch": 36.7275, "grad_norm": 7.21875, "learning_rate": 1.460399454864223e-05, "loss": 1.3955, "step": 14691 }, { "epoch": 36.73, "grad_norm": 10.8125, "learning_rate": 1.4603275763177491e-05, "loss": 1.2016, "step": 14692 }, { "epoch": 36.7325, "grad_norm": 8.0, "learning_rate": 1.4602556947533856e-05, "loss": 1.2263, "step": 14693 }, { "epoch": 36.735, "grad_norm": 6.71875, "learning_rate": 1.4601838101716043e-05, "loss": 1.3556, "step": 14694 }, { "epoch": 36.7375, "grad_norm": 7.9375, "learning_rate": 1.4601119225728766e-05, "loss": 1.3554, "step": 14695 }, { "epoch": 36.74, "grad_norm": 4.9375, "learning_rate": 1.4600400319576734e-05, "loss": 1.3196, "step": 14696 }, { "epoch": 36.7425, "grad_norm": 6.90625, "learning_rate": 1.4599681383264661e-05, "loss": 1.319, "step": 14697 }, { "epoch": 36.745, "grad_norm": 8.1875, "learning_rate": 1.4598962416797262e-05, "loss": 1.3244, "step": 14698 }, { "epoch": 36.7475, "grad_norm": 6.375, "learning_rate": 1.4598243420179248e-05, "loss": 1.3702, "step": 14699 }, { "epoch": 36.75, "grad_norm": 8.8125, "learning_rate": 1.4597524393415336e-05, "loss": 1.2695, "step": 14700 }, { "epoch": 36.7525, "grad_norm": 4.96875, "learning_rate": 1.4596805336510236e-05, "loss": 1.3149, "step": 14701 }, { "epoch": 36.755, "grad_norm": 6.125, "learning_rate": 1.4596086249468666e-05, "loss": 1.3619, "step": 14702 }, { "epoch": 36.7575, "grad_norm": 5.125, "learning_rate": 1.4595367132295337e-05, "loss": 1.3759, "step": 14703 }, { "epoch": 36.76, "grad_norm": 5.75, "learning_rate": 1.4594647984994966e-05, "loss": 1.3086, "step": 14704 }, { "epoch": 36.7625, "grad_norm": 8.0625, "learning_rate": 1.4593928807572267e-05, "loss": 1.5277, "step": 14705 }, { "epoch": 36.765, "grad_norm": 11.0625, "learning_rate": 1.4593209600031952e-05, "loss": 1.3605, "step": 14706 }, { "epoch": 36.7675, "grad_norm": 4.71875, "learning_rate": 1.4592490362378739e-05, "loss": 1.3087, "step": 14707 }, { "epoch": 36.77, "grad_norm": 10.75, "learning_rate": 1.4591771094617345e-05, "loss": 1.2639, "step": 14708 }, { "epoch": 36.7725, "grad_norm": 15.9375, "learning_rate": 1.4591051796752483e-05, "loss": 1.4646, "step": 14709 }, { "epoch": 36.775, "grad_norm": 6.34375, "learning_rate": 1.4590332468788868e-05, "loss": 1.2308, "step": 14710 }, { "epoch": 36.7775, "grad_norm": 12.375, "learning_rate": 1.458961311073122e-05, "loss": 1.424, "step": 14711 }, { "epoch": 36.78, "grad_norm": 7.5, "learning_rate": 1.4588893722584247e-05, "loss": 1.4807, "step": 14712 }, { "epoch": 36.7825, "grad_norm": 7.125, "learning_rate": 1.4588174304352678e-05, "loss": 1.3681, "step": 14713 }, { "epoch": 36.785, "grad_norm": 5.9375, "learning_rate": 1.4587454856041215e-05, "loss": 1.1529, "step": 14714 }, { "epoch": 36.7875, "grad_norm": 8.75, "learning_rate": 1.4586735377654586e-05, "loss": 1.3536, "step": 14715 }, { "epoch": 36.79, "grad_norm": 10.5625, "learning_rate": 1.4586015869197499e-05, "loss": 1.5557, "step": 14716 }, { "epoch": 36.7925, "grad_norm": 6.6875, "learning_rate": 1.458529633067468e-05, "loss": 1.288, "step": 14717 }, { "epoch": 36.795, "grad_norm": 7.84375, "learning_rate": 1.4584576762090839e-05, "loss": 1.254, "step": 14718 }, { "epoch": 36.7975, "grad_norm": 6.34375, "learning_rate": 1.4583857163450696e-05, "loss": 1.3579, "step": 14719 }, { "epoch": 36.8, "grad_norm": 5.21875, "learning_rate": 1.4583137534758968e-05, "loss": 1.3819, "step": 14720 }, { "epoch": 36.8025, "grad_norm": 8.6875, "learning_rate": 1.4582417876020376e-05, "loss": 1.398, "step": 14721 }, { "epoch": 36.805, "grad_norm": 8.5625, "learning_rate": 1.4581698187239637e-05, "loss": 1.4144, "step": 14722 }, { "epoch": 36.8075, "grad_norm": 7.71875, "learning_rate": 1.4580978468421466e-05, "loss": 1.5223, "step": 14723 }, { "epoch": 36.81, "grad_norm": 7.21875, "learning_rate": 1.4580258719570584e-05, "loss": 1.2122, "step": 14724 }, { "epoch": 36.8125, "grad_norm": 7.90625, "learning_rate": 1.4579538940691707e-05, "loss": 1.1375, "step": 14725 }, { "epoch": 36.815, "grad_norm": 7.375, "learning_rate": 1.4578819131789557e-05, "loss": 1.4532, "step": 14726 }, { "epoch": 36.8175, "grad_norm": 6.875, "learning_rate": 1.4578099292868854e-05, "loss": 1.4379, "step": 14727 }, { "epoch": 36.82, "grad_norm": 5.59375, "learning_rate": 1.4577379423934313e-05, "loss": 1.4007, "step": 14728 }, { "epoch": 36.8225, "grad_norm": 15.75, "learning_rate": 1.4576659524990657e-05, "loss": 1.3314, "step": 14729 }, { "epoch": 36.825, "grad_norm": 8.25, "learning_rate": 1.4575939596042605e-05, "loss": 1.4312, "step": 14730 }, { "epoch": 36.8275, "grad_norm": 8.0, "learning_rate": 1.4575219637094873e-05, "loss": 1.4634, "step": 14731 }, { "epoch": 36.83, "grad_norm": 5.78125, "learning_rate": 1.457449964815219e-05, "loss": 1.4151, "step": 14732 }, { "epoch": 36.8325, "grad_norm": 11.875, "learning_rate": 1.4573779629219265e-05, "loss": 1.4675, "step": 14733 }, { "epoch": 36.835, "grad_norm": 9.4375, "learning_rate": 1.4573059580300826e-05, "loss": 1.3276, "step": 14734 }, { "epoch": 36.8375, "grad_norm": 6.40625, "learning_rate": 1.4572339501401591e-05, "loss": 1.3983, "step": 14735 }, { "epoch": 36.84, "grad_norm": 6.53125, "learning_rate": 1.4571619392526279e-05, "loss": 1.4292, "step": 14736 }, { "epoch": 36.8425, "grad_norm": 7.59375, "learning_rate": 1.4570899253679615e-05, "loss": 1.3392, "step": 14737 }, { "epoch": 36.845, "grad_norm": 6.6875, "learning_rate": 1.4570179084866319e-05, "loss": 1.3296, "step": 14738 }, { "epoch": 36.8475, "grad_norm": 11.1875, "learning_rate": 1.4569458886091112e-05, "loss": 1.4605, "step": 14739 }, { "epoch": 36.85, "grad_norm": 11.4375, "learning_rate": 1.4568738657358715e-05, "loss": 1.429, "step": 14740 }, { "epoch": 36.8525, "grad_norm": 5.96875, "learning_rate": 1.4568018398673849e-05, "loss": 1.3784, "step": 14741 }, { "epoch": 36.855, "grad_norm": 6.25, "learning_rate": 1.456729811004124e-05, "loss": 1.3778, "step": 14742 }, { "epoch": 36.8575, "grad_norm": 6.46875, "learning_rate": 1.4566577791465603e-05, "loss": 1.2363, "step": 14743 }, { "epoch": 36.86, "grad_norm": 6.875, "learning_rate": 1.4565857442951668e-05, "loss": 1.3293, "step": 14744 }, { "epoch": 36.8625, "grad_norm": 5.84375, "learning_rate": 1.4565137064504156e-05, "loss": 1.3598, "step": 14745 }, { "epoch": 36.865, "grad_norm": 6.21875, "learning_rate": 1.4564416656127782e-05, "loss": 1.322, "step": 14746 }, { "epoch": 36.8675, "grad_norm": 9.4375, "learning_rate": 1.4563696217827282e-05, "loss": 1.4609, "step": 14747 }, { "epoch": 36.87, "grad_norm": 6.125, "learning_rate": 1.4562975749607366e-05, "loss": 1.3752, "step": 14748 }, { "epoch": 36.8725, "grad_norm": 7.28125, "learning_rate": 1.456225525147277e-05, "loss": 1.2912, "step": 14749 }, { "epoch": 36.875, "grad_norm": 7.71875, "learning_rate": 1.4561534723428205e-05, "loss": 1.3457, "step": 14750 }, { "epoch": 36.8775, "grad_norm": 8.4375, "learning_rate": 1.4560814165478406e-05, "loss": 1.3801, "step": 14751 }, { "epoch": 36.88, "grad_norm": 8.3125, "learning_rate": 1.456009357762809e-05, "loss": 1.2225, "step": 14752 }, { "epoch": 36.8825, "grad_norm": 8.125, "learning_rate": 1.4559372959881981e-05, "loss": 1.3005, "step": 14753 }, { "epoch": 36.885, "grad_norm": 7.59375, "learning_rate": 1.4558652312244807e-05, "loss": 1.38, "step": 14754 }, { "epoch": 36.8875, "grad_norm": 6.65625, "learning_rate": 1.4557931634721291e-05, "loss": 1.3354, "step": 14755 }, { "epoch": 36.89, "grad_norm": 9.125, "learning_rate": 1.4557210927316157e-05, "loss": 1.3089, "step": 14756 }, { "epoch": 36.8925, "grad_norm": 7.6875, "learning_rate": 1.4556490190034132e-05, "loss": 1.4451, "step": 14757 }, { "epoch": 36.895, "grad_norm": 7.3125, "learning_rate": 1.4555769422879938e-05, "loss": 1.4099, "step": 14758 }, { "epoch": 36.8975, "grad_norm": 7.40625, "learning_rate": 1.4555048625858303e-05, "loss": 1.3587, "step": 14759 }, { "epoch": 36.9, "grad_norm": 8.6875, "learning_rate": 1.455432779897395e-05, "loss": 1.3861, "step": 14760 }, { "epoch": 36.9025, "grad_norm": 6.34375, "learning_rate": 1.4553606942231606e-05, "loss": 1.496, "step": 14761 }, { "epoch": 36.905, "grad_norm": 7.28125, "learning_rate": 1.4552886055636e-05, "loss": 1.4276, "step": 14762 }, { "epoch": 36.9075, "grad_norm": 12.4375, "learning_rate": 1.4552165139191852e-05, "loss": 1.3376, "step": 14763 }, { "epoch": 36.91, "grad_norm": 8.8125, "learning_rate": 1.4551444192903895e-05, "loss": 1.3975, "step": 14764 }, { "epoch": 36.9125, "grad_norm": 9.5, "learning_rate": 1.455072321677685e-05, "loss": 1.4769, "step": 14765 }, { "epoch": 36.915, "grad_norm": 6.46875, "learning_rate": 1.4550002210815443e-05, "loss": 1.234, "step": 14766 }, { "epoch": 36.9175, "grad_norm": 9.5625, "learning_rate": 1.4549281175024408e-05, "loss": 1.37, "step": 14767 }, { "epoch": 36.92, "grad_norm": 11.1875, "learning_rate": 1.4548560109408465e-05, "loss": 1.3985, "step": 14768 }, { "epoch": 36.9225, "grad_norm": 8.0625, "learning_rate": 1.4547839013972345e-05, "loss": 1.2023, "step": 14769 }, { "epoch": 36.925, "grad_norm": 5.8125, "learning_rate": 1.4547117888720776e-05, "loss": 1.3546, "step": 14770 }, { "epoch": 36.9275, "grad_norm": 6.78125, "learning_rate": 1.4546396733658481e-05, "loss": 1.4016, "step": 14771 }, { "epoch": 36.93, "grad_norm": 7.375, "learning_rate": 1.4545675548790192e-05, "loss": 1.407, "step": 14772 }, { "epoch": 36.9325, "grad_norm": 8.6875, "learning_rate": 1.4544954334120636e-05, "loss": 1.3271, "step": 14773 }, { "epoch": 36.935, "grad_norm": 11.75, "learning_rate": 1.4544233089654542e-05, "loss": 1.4269, "step": 14774 }, { "epoch": 36.9375, "grad_norm": 11.25, "learning_rate": 1.4543511815396638e-05, "loss": 1.3767, "step": 14775 }, { "epoch": 36.94, "grad_norm": 8.125, "learning_rate": 1.4542790511351652e-05, "loss": 1.3496, "step": 14776 }, { "epoch": 36.9425, "grad_norm": 8.8125, "learning_rate": 1.4542069177524312e-05, "loss": 1.3544, "step": 14777 }, { "epoch": 36.945, "grad_norm": 8.1875, "learning_rate": 1.454134781391935e-05, "loss": 1.3891, "step": 14778 }, { "epoch": 36.9475, "grad_norm": 7.28125, "learning_rate": 1.4540626420541493e-05, "loss": 1.3281, "step": 14779 }, { "epoch": 36.95, "grad_norm": 9.4375, "learning_rate": 1.4539904997395468e-05, "loss": 1.2993, "step": 14780 }, { "epoch": 36.9525, "grad_norm": 8.4375, "learning_rate": 1.453918354448601e-05, "loss": 1.2568, "step": 14781 }, { "epoch": 36.955, "grad_norm": 5.46875, "learning_rate": 1.4538462061817846e-05, "loss": 1.4267, "step": 14782 }, { "epoch": 36.9575, "grad_norm": 8.875, "learning_rate": 1.453774054939571e-05, "loss": 1.2015, "step": 14783 }, { "epoch": 36.96, "grad_norm": 9.6875, "learning_rate": 1.4537019007224324e-05, "loss": 1.5107, "step": 14784 }, { "epoch": 36.9625, "grad_norm": 6.625, "learning_rate": 1.4536297435308426e-05, "loss": 1.2966, "step": 14785 }, { "epoch": 36.965, "grad_norm": 5.09375, "learning_rate": 1.4535575833652739e-05, "loss": 1.2038, "step": 14786 }, { "epoch": 36.9675, "grad_norm": 10.9375, "learning_rate": 1.4534854202262003e-05, "loss": 1.2654, "step": 14787 }, { "epoch": 36.97, "grad_norm": 6.09375, "learning_rate": 1.453413254114094e-05, "loss": 1.3148, "step": 14788 }, { "epoch": 36.9725, "grad_norm": 7.40625, "learning_rate": 1.453341085029429e-05, "loss": 1.378, "step": 14789 }, { "epoch": 36.975, "grad_norm": 6.53125, "learning_rate": 1.4532689129726778e-05, "loss": 1.3068, "step": 14790 }, { "epoch": 36.9775, "grad_norm": 10.0625, "learning_rate": 1.4531967379443137e-05, "loss": 1.417, "step": 14791 }, { "epoch": 36.98, "grad_norm": 8.0625, "learning_rate": 1.4531245599448099e-05, "loss": 1.2037, "step": 14792 }, { "epoch": 36.9825, "grad_norm": 8.875, "learning_rate": 1.4530523789746395e-05, "loss": 1.2429, "step": 14793 }, { "epoch": 36.985, "grad_norm": 8.875, "learning_rate": 1.4529801950342761e-05, "loss": 1.3782, "step": 14794 }, { "epoch": 36.9875, "grad_norm": 6.4375, "learning_rate": 1.4529080081241925e-05, "loss": 1.4151, "step": 14795 }, { "epoch": 36.99, "grad_norm": 6.875, "learning_rate": 1.4528358182448621e-05, "loss": 1.2398, "step": 14796 }, { "epoch": 36.9925, "grad_norm": 7.0625, "learning_rate": 1.4527636253967586e-05, "loss": 1.265, "step": 14797 }, { "epoch": 36.995, "grad_norm": 5.625, "learning_rate": 1.4526914295803543e-05, "loss": 1.3484, "step": 14798 }, { "epoch": 36.9975, "grad_norm": 7.75, "learning_rate": 1.4526192307961235e-05, "loss": 1.3651, "step": 14799 }, { "epoch": 37.0, "grad_norm": 8.375, "learning_rate": 1.4525470290445392e-05, "loss": 1.3206, "step": 14800 }, { "epoch": 37.0025, "grad_norm": 9.5625, "learning_rate": 1.4524748243260742e-05, "loss": 1.4399, "step": 14801 }, { "epoch": 37.005, "grad_norm": 6.34375, "learning_rate": 1.4524026166412028e-05, "loss": 1.3561, "step": 14802 }, { "epoch": 37.0075, "grad_norm": 8.4375, "learning_rate": 1.4523304059903977e-05, "loss": 1.4413, "step": 14803 }, { "epoch": 37.01, "grad_norm": 7.375, "learning_rate": 1.4522581923741329e-05, "loss": 1.1882, "step": 14804 }, { "epoch": 37.0125, "grad_norm": 7.6875, "learning_rate": 1.4521859757928812e-05, "loss": 1.3606, "step": 14805 }, { "epoch": 37.015, "grad_norm": 8.6875, "learning_rate": 1.4521137562471163e-05, "loss": 1.3051, "step": 14806 }, { "epoch": 37.0175, "grad_norm": 11.0, "learning_rate": 1.4520415337373118e-05, "loss": 1.4295, "step": 14807 }, { "epoch": 37.02, "grad_norm": 7.8125, "learning_rate": 1.4519693082639413e-05, "loss": 1.3915, "step": 14808 }, { "epoch": 37.0225, "grad_norm": 8.125, "learning_rate": 1.4518970798274777e-05, "loss": 1.3397, "step": 14809 }, { "epoch": 37.025, "grad_norm": 7.375, "learning_rate": 1.451824848428395e-05, "loss": 1.4624, "step": 14810 }, { "epoch": 37.0275, "grad_norm": 8.3125, "learning_rate": 1.451752614067167e-05, "loss": 1.3377, "step": 14811 }, { "epoch": 37.03, "grad_norm": 7.5, "learning_rate": 1.4516803767442666e-05, "loss": 1.3492, "step": 14812 }, { "epoch": 37.0325, "grad_norm": 11.6875, "learning_rate": 1.451608136460168e-05, "loss": 1.3349, "step": 14813 }, { "epoch": 37.035, "grad_norm": 4.875, "learning_rate": 1.4515358932153443e-05, "loss": 1.2561, "step": 14814 }, { "epoch": 37.0375, "grad_norm": 7.0625, "learning_rate": 1.4514636470102696e-05, "loss": 1.2657, "step": 14815 }, { "epoch": 37.04, "grad_norm": 7.0, "learning_rate": 1.4513913978454169e-05, "loss": 1.3681, "step": 14816 }, { "epoch": 37.0425, "grad_norm": 10.125, "learning_rate": 1.4513191457212607e-05, "loss": 1.3436, "step": 14817 }, { "epoch": 37.045, "grad_norm": 8.0, "learning_rate": 1.4512468906382738e-05, "loss": 1.2634, "step": 14818 }, { "epoch": 37.0475, "grad_norm": 7.3125, "learning_rate": 1.451174632596931e-05, "loss": 1.3758, "step": 14819 }, { "epoch": 37.05, "grad_norm": 9.5, "learning_rate": 1.4511023715977048e-05, "loss": 1.3831, "step": 14820 }, { "epoch": 37.0525, "grad_norm": 10.25, "learning_rate": 1.4510301076410695e-05, "loss": 1.4275, "step": 14821 }, { "epoch": 37.055, "grad_norm": 7.375, "learning_rate": 1.4509578407274992e-05, "loss": 1.3713, "step": 14822 }, { "epoch": 37.0575, "grad_norm": 8.8125, "learning_rate": 1.4508855708574671e-05, "loss": 1.3189, "step": 14823 }, { "epoch": 37.06, "grad_norm": 6.28125, "learning_rate": 1.4508132980314473e-05, "loss": 1.432, "step": 14824 }, { "epoch": 37.0625, "grad_norm": 6.78125, "learning_rate": 1.4507410222499136e-05, "loss": 1.2664, "step": 14825 }, { "epoch": 37.065, "grad_norm": 15.375, "learning_rate": 1.4506687435133397e-05, "loss": 1.3794, "step": 14826 }, { "epoch": 37.0675, "grad_norm": 9.0, "learning_rate": 1.4505964618221997e-05, "loss": 1.3003, "step": 14827 }, { "epoch": 37.07, "grad_norm": 6.9375, "learning_rate": 1.4505241771769672e-05, "loss": 1.3814, "step": 14828 }, { "epoch": 37.0725, "grad_norm": 7.3125, "learning_rate": 1.4504518895781161e-05, "loss": 1.3121, "step": 14829 }, { "epoch": 37.075, "grad_norm": 5.375, "learning_rate": 1.4503795990261207e-05, "loss": 1.1914, "step": 14830 }, { "epoch": 37.0775, "grad_norm": 8.125, "learning_rate": 1.4503073055214545e-05, "loss": 1.3428, "step": 14831 }, { "epoch": 37.08, "grad_norm": 7.0625, "learning_rate": 1.4502350090645919e-05, "loss": 1.514, "step": 14832 }, { "epoch": 37.0825, "grad_norm": 7.625, "learning_rate": 1.4501627096560063e-05, "loss": 1.3596, "step": 14833 }, { "epoch": 37.085, "grad_norm": 9.0625, "learning_rate": 1.4500904072961722e-05, "loss": 1.3002, "step": 14834 }, { "epoch": 37.0875, "grad_norm": 7.0, "learning_rate": 1.4500181019855633e-05, "loss": 1.3125, "step": 14835 }, { "epoch": 37.09, "grad_norm": 6.71875, "learning_rate": 1.4499457937246537e-05, "loss": 1.4122, "step": 14836 }, { "epoch": 37.0925, "grad_norm": 8.75, "learning_rate": 1.4498734825139174e-05, "loss": 1.4611, "step": 14837 }, { "epoch": 37.095, "grad_norm": 7.125, "learning_rate": 1.4498011683538286e-05, "loss": 1.3708, "step": 14838 }, { "epoch": 37.0975, "grad_norm": 6.5625, "learning_rate": 1.4497288512448612e-05, "loss": 1.3156, "step": 14839 }, { "epoch": 37.1, "grad_norm": 6.21875, "learning_rate": 1.4496565311874902e-05, "loss": 1.5315, "step": 14840 }, { "epoch": 37.1025, "grad_norm": 6.84375, "learning_rate": 1.4495842081821881e-05, "loss": 1.4315, "step": 14841 }, { "epoch": 37.105, "grad_norm": 7.3125, "learning_rate": 1.4495118822294304e-05, "loss": 1.2726, "step": 14842 }, { "epoch": 37.1075, "grad_norm": 6.375, "learning_rate": 1.4494395533296903e-05, "loss": 1.3361, "step": 14843 }, { "epoch": 37.11, "grad_norm": 6.375, "learning_rate": 1.4493672214834432e-05, "loss": 1.3997, "step": 14844 }, { "epoch": 37.1125, "grad_norm": 7.8125, "learning_rate": 1.4492948866911617e-05, "loss": 1.3733, "step": 14845 }, { "epoch": 37.115, "grad_norm": 7.90625, "learning_rate": 1.4492225489533219e-05, "loss": 1.3516, "step": 14846 }, { "epoch": 37.1175, "grad_norm": 8.875, "learning_rate": 1.4491502082703964e-05, "loss": 1.1783, "step": 14847 }, { "epoch": 37.12, "grad_norm": 5.71875, "learning_rate": 1.4490778646428601e-05, "loss": 1.2949, "step": 14848 }, { "epoch": 37.1225, "grad_norm": 6.40625, "learning_rate": 1.4490055180711874e-05, "loss": 1.4669, "step": 14849 }, { "epoch": 37.125, "grad_norm": 7.53125, "learning_rate": 1.4489331685558525e-05, "loss": 1.3846, "step": 14850 }, { "epoch": 37.1275, "grad_norm": 11.5625, "learning_rate": 1.4488608160973298e-05, "loss": 1.4325, "step": 14851 }, { "epoch": 37.13, "grad_norm": 5.75, "learning_rate": 1.4487884606960934e-05, "loss": 1.2581, "step": 14852 }, { "epoch": 37.1325, "grad_norm": 5.6875, "learning_rate": 1.448716102352618e-05, "loss": 1.3069, "step": 14853 }, { "epoch": 37.135, "grad_norm": 8.5625, "learning_rate": 1.4486437410673777e-05, "loss": 1.4238, "step": 14854 }, { "epoch": 37.1375, "grad_norm": 7.8125, "learning_rate": 1.4485713768408469e-05, "loss": 1.321, "step": 14855 }, { "epoch": 37.14, "grad_norm": 7.6875, "learning_rate": 1.4484990096735002e-05, "loss": 1.57, "step": 14856 }, { "epoch": 37.1425, "grad_norm": 7.46875, "learning_rate": 1.448426639565812e-05, "loss": 1.3599, "step": 14857 }, { "epoch": 37.145, "grad_norm": 7.125, "learning_rate": 1.4483542665182566e-05, "loss": 1.1631, "step": 14858 }, { "epoch": 37.1475, "grad_norm": 11.625, "learning_rate": 1.4482818905313085e-05, "loss": 1.3387, "step": 14859 }, { "epoch": 37.15, "grad_norm": 5.65625, "learning_rate": 1.4482095116054421e-05, "loss": 1.2767, "step": 14860 }, { "epoch": 37.1525, "grad_norm": 7.4375, "learning_rate": 1.4481371297411325e-05, "loss": 1.345, "step": 14861 }, { "epoch": 37.155, "grad_norm": 7.3125, "learning_rate": 1.4480647449388538e-05, "loss": 1.4171, "step": 14862 }, { "epoch": 37.1575, "grad_norm": 6.5625, "learning_rate": 1.4479923571990803e-05, "loss": 1.3724, "step": 14863 }, { "epoch": 37.16, "grad_norm": 8.0, "learning_rate": 1.4479199665222869e-05, "loss": 1.2438, "step": 14864 }, { "epoch": 37.1625, "grad_norm": 7.4375, "learning_rate": 1.4478475729089483e-05, "loss": 1.297, "step": 14865 }, { "epoch": 37.165, "grad_norm": 9.3125, "learning_rate": 1.4477751763595387e-05, "loss": 1.3478, "step": 14866 }, { "epoch": 37.1675, "grad_norm": 7.96875, "learning_rate": 1.447702776874533e-05, "loss": 1.4253, "step": 14867 }, { "epoch": 37.17, "grad_norm": 7.90625, "learning_rate": 1.447630374454406e-05, "loss": 1.3227, "step": 14868 }, { "epoch": 37.1725, "grad_norm": 5.25, "learning_rate": 1.447557969099632e-05, "loss": 1.5449, "step": 14869 }, { "epoch": 37.175, "grad_norm": 10.4375, "learning_rate": 1.447485560810686e-05, "loss": 1.354, "step": 14870 }, { "epoch": 37.1775, "grad_norm": 6.71875, "learning_rate": 1.4474131495880423e-05, "loss": 1.3, "step": 14871 }, { "epoch": 37.18, "grad_norm": 7.8125, "learning_rate": 1.4473407354321763e-05, "loss": 1.2731, "step": 14872 }, { "epoch": 37.1825, "grad_norm": 8.0625, "learning_rate": 1.4472683183435618e-05, "loss": 1.4217, "step": 14873 }, { "epoch": 37.185, "grad_norm": 8.0625, "learning_rate": 1.4471958983226748e-05, "loss": 1.2393, "step": 14874 }, { "epoch": 37.1875, "grad_norm": 5.65625, "learning_rate": 1.4471234753699887e-05, "loss": 1.4164, "step": 14875 }, { "epoch": 37.19, "grad_norm": 8.9375, "learning_rate": 1.4470510494859796e-05, "loss": 1.2512, "step": 14876 }, { "epoch": 37.1925, "grad_norm": 12.9375, "learning_rate": 1.4469786206711213e-05, "loss": 1.2808, "step": 14877 }, { "epoch": 37.195, "grad_norm": 7.125, "learning_rate": 1.4469061889258892e-05, "loss": 1.2942, "step": 14878 }, { "epoch": 37.1975, "grad_norm": 11.6875, "learning_rate": 1.4468337542507579e-05, "loss": 1.3883, "step": 14879 }, { "epoch": 37.2, "grad_norm": 10.375, "learning_rate": 1.4467613166462024e-05, "loss": 1.3307, "step": 14880 }, { "epoch": 37.2025, "grad_norm": 6.1875, "learning_rate": 1.4466888761126977e-05, "loss": 1.2864, "step": 14881 }, { "epoch": 37.205, "grad_norm": 7.8125, "learning_rate": 1.4466164326507186e-05, "loss": 1.4383, "step": 14882 }, { "epoch": 37.2075, "grad_norm": 7.09375, "learning_rate": 1.44654398626074e-05, "loss": 1.3883, "step": 14883 }, { "epoch": 37.21, "grad_norm": 8.75, "learning_rate": 1.4464715369432368e-05, "loss": 1.2868, "step": 14884 }, { "epoch": 37.2125, "grad_norm": 7.65625, "learning_rate": 1.4463990846986842e-05, "loss": 1.508, "step": 14885 }, { "epoch": 37.215, "grad_norm": 6.28125, "learning_rate": 1.4463266295275568e-05, "loss": 1.3461, "step": 14886 }, { "epoch": 37.2175, "grad_norm": 9.625, "learning_rate": 1.4462541714303304e-05, "loss": 1.254, "step": 14887 }, { "epoch": 37.22, "grad_norm": 9.625, "learning_rate": 1.4461817104074788e-05, "loss": 1.5085, "step": 14888 }, { "epoch": 37.2225, "grad_norm": 5.65625, "learning_rate": 1.4461092464594786e-05, "loss": 1.4526, "step": 14889 }, { "epoch": 37.225, "grad_norm": 6.0, "learning_rate": 1.4460367795868033e-05, "loss": 1.2786, "step": 14890 }, { "epoch": 37.2275, "grad_norm": 9.5, "learning_rate": 1.4459643097899293e-05, "loss": 1.2666, "step": 14891 }, { "epoch": 37.23, "grad_norm": 8.1875, "learning_rate": 1.4458918370693306e-05, "loss": 1.3151, "step": 14892 }, { "epoch": 37.2325, "grad_norm": 7.65625, "learning_rate": 1.4458193614254831e-05, "loss": 1.2528, "step": 14893 }, { "epoch": 37.235, "grad_norm": 6.90625, "learning_rate": 1.445746882858862e-05, "loss": 1.3304, "step": 14894 }, { "epoch": 37.2375, "grad_norm": 11.6875, "learning_rate": 1.4456744013699418e-05, "loss": 1.4227, "step": 14895 }, { "epoch": 37.24, "grad_norm": 9.6875, "learning_rate": 1.445601916959198e-05, "loss": 1.4459, "step": 14896 }, { "epoch": 37.2425, "grad_norm": 7.6875, "learning_rate": 1.4455294296271063e-05, "loss": 1.4714, "step": 14897 }, { "epoch": 37.245, "grad_norm": 6.125, "learning_rate": 1.445456939374141e-05, "loss": 1.2185, "step": 14898 }, { "epoch": 37.2475, "grad_norm": 9.3125, "learning_rate": 1.445384446200778e-05, "loss": 1.3174, "step": 14899 }, { "epoch": 37.25, "grad_norm": 6.65625, "learning_rate": 1.4453119501074924e-05, "loss": 1.2449, "step": 14900 }, { "epoch": 37.2525, "grad_norm": 9.5625, "learning_rate": 1.4452394510947593e-05, "loss": 1.3707, "step": 14901 }, { "epoch": 37.255, "grad_norm": 8.4375, "learning_rate": 1.4451669491630546e-05, "loss": 1.3804, "step": 14902 }, { "epoch": 37.2575, "grad_norm": 8.4375, "learning_rate": 1.445094444312853e-05, "loss": 1.3581, "step": 14903 }, { "epoch": 37.26, "grad_norm": 5.71875, "learning_rate": 1.44502193654463e-05, "loss": 1.2122, "step": 14904 }, { "epoch": 37.2625, "grad_norm": 7.1875, "learning_rate": 1.444949425858861e-05, "loss": 1.404, "step": 14905 }, { "epoch": 37.265, "grad_norm": 6.25, "learning_rate": 1.4448769122560214e-05, "loss": 1.3041, "step": 14906 }, { "epoch": 37.2675, "grad_norm": 6.21875, "learning_rate": 1.4448043957365865e-05, "loss": 1.4571, "step": 14907 }, { "epoch": 37.27, "grad_norm": 7.8125, "learning_rate": 1.4447318763010319e-05, "loss": 1.3691, "step": 14908 }, { "epoch": 37.2725, "grad_norm": 9.25, "learning_rate": 1.4446593539498328e-05, "loss": 1.4345, "step": 14909 }, { "epoch": 37.275, "grad_norm": 9.75, "learning_rate": 1.4445868286834647e-05, "loss": 1.4192, "step": 14910 }, { "epoch": 37.2775, "grad_norm": 8.625, "learning_rate": 1.4445143005024035e-05, "loss": 1.4432, "step": 14911 }, { "epoch": 37.28, "grad_norm": 10.4375, "learning_rate": 1.4444417694071242e-05, "loss": 1.4764, "step": 14912 }, { "epoch": 37.2825, "grad_norm": 15.6875, "learning_rate": 1.444369235398102e-05, "loss": 1.2774, "step": 14913 }, { "epoch": 37.285, "grad_norm": 6.96875, "learning_rate": 1.4442966984758134e-05, "loss": 1.3282, "step": 14914 }, { "epoch": 37.2875, "grad_norm": 7.1875, "learning_rate": 1.444224158640733e-05, "loss": 1.2631, "step": 14915 }, { "epoch": 37.29, "grad_norm": 7.90625, "learning_rate": 1.4441516158933374e-05, "loss": 1.2261, "step": 14916 }, { "epoch": 37.2925, "grad_norm": 9.3125, "learning_rate": 1.444079070234101e-05, "loss": 1.3349, "step": 14917 }, { "epoch": 37.295, "grad_norm": 7.375, "learning_rate": 1.4440065216635002e-05, "loss": 1.2944, "step": 14918 }, { "epoch": 37.2975, "grad_norm": 7.09375, "learning_rate": 1.4439339701820103e-05, "loss": 1.3292, "step": 14919 }, { "epoch": 37.3, "grad_norm": 11.0625, "learning_rate": 1.4438614157901073e-05, "loss": 1.5142, "step": 14920 }, { "epoch": 37.3025, "grad_norm": 4.90625, "learning_rate": 1.4437888584882664e-05, "loss": 1.4853, "step": 14921 }, { "epoch": 37.305, "grad_norm": 8.25, "learning_rate": 1.4437162982769636e-05, "loss": 1.528, "step": 14922 }, { "epoch": 37.3075, "grad_norm": 7.125, "learning_rate": 1.4436437351566745e-05, "loss": 1.3494, "step": 14923 }, { "epoch": 37.31, "grad_norm": 9.6875, "learning_rate": 1.4435711691278747e-05, "loss": 1.3148, "step": 14924 }, { "epoch": 37.3125, "grad_norm": 8.5, "learning_rate": 1.4434986001910403e-05, "loss": 1.3527, "step": 14925 }, { "epoch": 37.315, "grad_norm": 5.09375, "learning_rate": 1.4434260283466466e-05, "loss": 1.3332, "step": 14926 }, { "epoch": 37.3175, "grad_norm": 9.5, "learning_rate": 1.44335345359517e-05, "loss": 1.3841, "step": 14927 }, { "epoch": 37.32, "grad_norm": 6.3125, "learning_rate": 1.4432808759370853e-05, "loss": 1.2912, "step": 14928 }, { "epoch": 37.3225, "grad_norm": 6.71875, "learning_rate": 1.4432082953728695e-05, "loss": 1.3336, "step": 14929 }, { "epoch": 37.325, "grad_norm": 7.65625, "learning_rate": 1.4431357119029972e-05, "loss": 1.4182, "step": 14930 }, { "epoch": 37.3275, "grad_norm": 5.34375, "learning_rate": 1.4430631255279456e-05, "loss": 1.3106, "step": 14931 }, { "epoch": 37.33, "grad_norm": 10.0625, "learning_rate": 1.4429905362481896e-05, "loss": 1.3463, "step": 14932 }, { "epoch": 37.3325, "grad_norm": 9.6875, "learning_rate": 1.4429179440642052e-05, "loss": 1.3394, "step": 14933 }, { "epoch": 37.335, "grad_norm": 9.4375, "learning_rate": 1.4428453489764685e-05, "loss": 1.2236, "step": 14934 }, { "epoch": 37.3375, "grad_norm": 8.125, "learning_rate": 1.4427727509854557e-05, "loss": 1.3653, "step": 14935 }, { "epoch": 37.34, "grad_norm": 6.25, "learning_rate": 1.4427001500916422e-05, "loss": 1.3265, "step": 14936 }, { "epoch": 37.3425, "grad_norm": 5.75, "learning_rate": 1.4426275462955044e-05, "loss": 1.255, "step": 14937 }, { "epoch": 37.345, "grad_norm": 8.375, "learning_rate": 1.442554939597518e-05, "loss": 1.363, "step": 14938 }, { "epoch": 37.3475, "grad_norm": 5.375, "learning_rate": 1.4424823299981591e-05, "loss": 1.2406, "step": 14939 }, { "epoch": 37.35, "grad_norm": 9.4375, "learning_rate": 1.4424097174979038e-05, "loss": 1.2962, "step": 14940 }, { "epoch": 37.3525, "grad_norm": 6.03125, "learning_rate": 1.442337102097228e-05, "loss": 1.2476, "step": 14941 }, { "epoch": 37.355, "grad_norm": 7.46875, "learning_rate": 1.4422644837966081e-05, "loss": 1.3547, "step": 14942 }, { "epoch": 37.3575, "grad_norm": 6.71875, "learning_rate": 1.44219186259652e-05, "loss": 1.2782, "step": 14943 }, { "epoch": 37.36, "grad_norm": 6.09375, "learning_rate": 1.4421192384974396e-05, "loss": 1.2397, "step": 14944 }, { "epoch": 37.3625, "grad_norm": 10.6875, "learning_rate": 1.4420466114998428e-05, "loss": 1.4087, "step": 14945 }, { "epoch": 37.365, "grad_norm": 11.0625, "learning_rate": 1.4419739816042067e-05, "loss": 1.4594, "step": 14946 }, { "epoch": 37.3675, "grad_norm": 6.1875, "learning_rate": 1.4419013488110065e-05, "loss": 1.4331, "step": 14947 }, { "epoch": 37.37, "grad_norm": 6.75, "learning_rate": 1.4418287131207189e-05, "loss": 1.1873, "step": 14948 }, { "epoch": 37.3725, "grad_norm": 6.0, "learning_rate": 1.44175607453382e-05, "loss": 1.3616, "step": 14949 }, { "epoch": 37.375, "grad_norm": 8.5625, "learning_rate": 1.4416834330507857e-05, "loss": 1.5604, "step": 14950 }, { "epoch": 37.3775, "grad_norm": 11.125, "learning_rate": 1.4416107886720925e-05, "loss": 1.317, "step": 14951 }, { "epoch": 37.38, "grad_norm": 6.59375, "learning_rate": 1.4415381413982168e-05, "loss": 1.4729, "step": 14952 }, { "epoch": 37.3825, "grad_norm": 7.25, "learning_rate": 1.4414654912296346e-05, "loss": 1.3116, "step": 14953 }, { "epoch": 37.385, "grad_norm": 10.625, "learning_rate": 1.4413928381668222e-05, "loss": 1.2608, "step": 14954 }, { "epoch": 37.3875, "grad_norm": 8.4375, "learning_rate": 1.4413201822102564e-05, "loss": 1.3869, "step": 14955 }, { "epoch": 37.39, "grad_norm": 7.125, "learning_rate": 1.441247523360413e-05, "loss": 1.501, "step": 14956 }, { "epoch": 37.3925, "grad_norm": 7.8125, "learning_rate": 1.4411748616177686e-05, "loss": 1.1973, "step": 14957 }, { "epoch": 37.395, "grad_norm": 6.625, "learning_rate": 1.441102196982799e-05, "loss": 1.3527, "step": 14958 }, { "epoch": 37.3975, "grad_norm": 9.625, "learning_rate": 1.4410295294559815e-05, "loss": 1.342, "step": 14959 }, { "epoch": 37.4, "grad_norm": 9.9375, "learning_rate": 1.4409568590377918e-05, "loss": 1.2914, "step": 14960 }, { "epoch": 37.4025, "grad_norm": 6.78125, "learning_rate": 1.4408841857287068e-05, "loss": 1.2204, "step": 14961 }, { "epoch": 37.405, "grad_norm": 6.875, "learning_rate": 1.4408115095292023e-05, "loss": 1.3474, "step": 14962 }, { "epoch": 37.4075, "grad_norm": 9.3125, "learning_rate": 1.4407388304397559e-05, "loss": 1.4661, "step": 14963 }, { "epoch": 37.41, "grad_norm": 6.625, "learning_rate": 1.4406661484608429e-05, "loss": 1.2827, "step": 14964 }, { "epoch": 37.4125, "grad_norm": 4.9375, "learning_rate": 1.4405934635929401e-05, "loss": 1.294, "step": 14965 }, { "epoch": 37.415, "grad_norm": 6.65625, "learning_rate": 1.4405207758365245e-05, "loss": 1.5483, "step": 14966 }, { "epoch": 37.4175, "grad_norm": 7.09375, "learning_rate": 1.4404480851920721e-05, "loss": 1.335, "step": 14967 }, { "epoch": 37.42, "grad_norm": 5.4375, "learning_rate": 1.44037539166006e-05, "loss": 1.3137, "step": 14968 }, { "epoch": 37.4225, "grad_norm": 9.125, "learning_rate": 1.4403026952409642e-05, "loss": 1.4131, "step": 14969 }, { "epoch": 37.425, "grad_norm": 8.0625, "learning_rate": 1.4402299959352614e-05, "loss": 1.2804, "step": 14970 }, { "epoch": 37.4275, "grad_norm": 7.125, "learning_rate": 1.4401572937434286e-05, "loss": 1.2187, "step": 14971 }, { "epoch": 37.43, "grad_norm": 8.6875, "learning_rate": 1.4400845886659423e-05, "loss": 1.4237, "step": 14972 }, { "epoch": 37.4325, "grad_norm": 11.3125, "learning_rate": 1.4400118807032788e-05, "loss": 1.4186, "step": 14973 }, { "epoch": 37.435, "grad_norm": 6.375, "learning_rate": 1.4399391698559153e-05, "loss": 1.2932, "step": 14974 }, { "epoch": 37.4375, "grad_norm": 7.53125, "learning_rate": 1.4398664561243278e-05, "loss": 1.331, "step": 14975 }, { "epoch": 37.44, "grad_norm": 5.8125, "learning_rate": 1.439793739508994e-05, "loss": 1.2679, "step": 14976 }, { "epoch": 37.4425, "grad_norm": 9.1875, "learning_rate": 1.4397210200103894e-05, "loss": 1.2496, "step": 14977 }, { "epoch": 37.445, "grad_norm": 6.71875, "learning_rate": 1.4396482976289918e-05, "loss": 1.3086, "step": 14978 }, { "epoch": 37.4475, "grad_norm": 5.96875, "learning_rate": 1.4395755723652775e-05, "loss": 1.1428, "step": 14979 }, { "epoch": 37.45, "grad_norm": 7.125, "learning_rate": 1.4395028442197231e-05, "loss": 1.3478, "step": 14980 }, { "epoch": 37.4525, "grad_norm": 6.75, "learning_rate": 1.4394301131928061e-05, "loss": 1.4, "step": 14981 }, { "epoch": 37.455, "grad_norm": 8.8125, "learning_rate": 1.4393573792850023e-05, "loss": 1.3475, "step": 14982 }, { "epoch": 37.4575, "grad_norm": 7.125, "learning_rate": 1.4392846424967895e-05, "loss": 1.1521, "step": 14983 }, { "epoch": 37.46, "grad_norm": 9.75, "learning_rate": 1.4392119028286443e-05, "loss": 1.3487, "step": 14984 }, { "epoch": 37.4625, "grad_norm": 6.28125, "learning_rate": 1.4391391602810429e-05, "loss": 1.4115, "step": 14985 }, { "epoch": 37.465, "grad_norm": 9.875, "learning_rate": 1.4390664148544633e-05, "loss": 1.3394, "step": 14986 }, { "epoch": 37.4675, "grad_norm": 9.625, "learning_rate": 1.4389936665493813e-05, "loss": 1.312, "step": 14987 }, { "epoch": 37.47, "grad_norm": 7.28125, "learning_rate": 1.4389209153662748e-05, "loss": 1.206, "step": 14988 }, { "epoch": 37.4725, "grad_norm": 6.84375, "learning_rate": 1.4388481613056204e-05, "loss": 1.2856, "step": 14989 }, { "epoch": 37.475, "grad_norm": 5.8125, "learning_rate": 1.4387754043678947e-05, "loss": 1.3252, "step": 14990 }, { "epoch": 37.4775, "grad_norm": 11.0, "learning_rate": 1.4387026445535751e-05, "loss": 1.4484, "step": 14991 }, { "epoch": 37.48, "grad_norm": 6.59375, "learning_rate": 1.4386298818631388e-05, "loss": 1.4561, "step": 14992 }, { "epoch": 37.4825, "grad_norm": 7.125, "learning_rate": 1.4385571162970622e-05, "loss": 1.1651, "step": 14993 }, { "epoch": 37.485, "grad_norm": 4.46875, "learning_rate": 1.4384843478558229e-05, "loss": 1.1376, "step": 14994 }, { "epoch": 37.4875, "grad_norm": 7.4375, "learning_rate": 1.4384115765398976e-05, "loss": 1.2902, "step": 14995 }, { "epoch": 37.49, "grad_norm": 7.875, "learning_rate": 1.4383388023497635e-05, "loss": 1.467, "step": 14996 }, { "epoch": 37.4925, "grad_norm": 6.6875, "learning_rate": 1.438266025285898e-05, "loss": 1.3725, "step": 14997 }, { "epoch": 37.495, "grad_norm": 8.5, "learning_rate": 1.4381932453487777e-05, "loss": 1.3612, "step": 14998 }, { "epoch": 37.4975, "grad_norm": 8.125, "learning_rate": 1.4381204625388802e-05, "loss": 1.3929, "step": 14999 }, { "epoch": 37.5, "grad_norm": 7.53125, "learning_rate": 1.4380476768566825e-05, "loss": 1.268, "step": 15000 }, { "epoch": 37.5025, "grad_norm": 8.5625, "learning_rate": 1.4379748883026618e-05, "loss": 1.2178, "step": 15001 }, { "epoch": 37.505, "grad_norm": 8.375, "learning_rate": 1.437902096877295e-05, "loss": 1.2976, "step": 15002 }, { "epoch": 37.5075, "grad_norm": 7.3125, "learning_rate": 1.4378293025810598e-05, "loss": 1.1252, "step": 15003 }, { "epoch": 37.51, "grad_norm": 7.75, "learning_rate": 1.4377565054144332e-05, "loss": 1.2348, "step": 15004 }, { "epoch": 37.5125, "grad_norm": 7.40625, "learning_rate": 1.4376837053778925e-05, "loss": 1.4559, "step": 15005 }, { "epoch": 37.515, "grad_norm": 7.71875, "learning_rate": 1.4376109024719146e-05, "loss": 1.3357, "step": 15006 }, { "epoch": 37.5175, "grad_norm": 9.0625, "learning_rate": 1.4375380966969777e-05, "loss": 1.1802, "step": 15007 }, { "epoch": 37.52, "grad_norm": 6.84375, "learning_rate": 1.437465288053558e-05, "loss": 1.4328, "step": 15008 }, { "epoch": 37.5225, "grad_norm": 4.5625, "learning_rate": 1.4373924765421337e-05, "loss": 1.3014, "step": 15009 }, { "epoch": 37.525, "grad_norm": 9.625, "learning_rate": 1.4373196621631817e-05, "loss": 1.2861, "step": 15010 }, { "epoch": 37.5275, "grad_norm": 6.3125, "learning_rate": 1.4372468449171795e-05, "loss": 1.2709, "step": 15011 }, { "epoch": 37.53, "grad_norm": 7.15625, "learning_rate": 1.4371740248046046e-05, "loss": 1.3489, "step": 15012 }, { "epoch": 37.5325, "grad_norm": 9.75, "learning_rate": 1.4371012018259343e-05, "loss": 1.2193, "step": 15013 }, { "epoch": 37.535, "grad_norm": 6.75, "learning_rate": 1.437028375981646e-05, "loss": 1.3415, "step": 15014 }, { "epoch": 37.5375, "grad_norm": 7.625, "learning_rate": 1.436955547272217e-05, "loss": 1.311, "step": 15015 }, { "epoch": 37.54, "grad_norm": 8.125, "learning_rate": 1.436882715698125e-05, "loss": 1.3156, "step": 15016 }, { "epoch": 37.5425, "grad_norm": 7.1875, "learning_rate": 1.4368098812598473e-05, "loss": 1.3377, "step": 15017 }, { "epoch": 37.545, "grad_norm": 8.6875, "learning_rate": 1.4367370439578619e-05, "loss": 1.324, "step": 15018 }, { "epoch": 37.5475, "grad_norm": 6.96875, "learning_rate": 1.4366642037926455e-05, "loss": 1.5083, "step": 15019 }, { "epoch": 37.55, "grad_norm": 11.25, "learning_rate": 1.4365913607646762e-05, "loss": 1.3341, "step": 15020 }, { "epoch": 37.5525, "grad_norm": 9.625, "learning_rate": 1.4365185148744313e-05, "loss": 1.3183, "step": 15021 }, { "epoch": 37.555, "grad_norm": 10.25, "learning_rate": 1.4364456661223884e-05, "loss": 1.3406, "step": 15022 }, { "epoch": 37.5575, "grad_norm": 9.3125, "learning_rate": 1.4363728145090255e-05, "loss": 1.3601, "step": 15023 }, { "epoch": 37.56, "grad_norm": 7.375, "learning_rate": 1.4362999600348198e-05, "loss": 1.5215, "step": 15024 }, { "epoch": 37.5625, "grad_norm": 6.5625, "learning_rate": 1.4362271027002491e-05, "loss": 1.3641, "step": 15025 }, { "epoch": 37.565, "grad_norm": 6.5625, "learning_rate": 1.4361542425057905e-05, "loss": 1.3579, "step": 15026 }, { "epoch": 37.5675, "grad_norm": 6.40625, "learning_rate": 1.4360813794519226e-05, "loss": 1.5751, "step": 15027 }, { "epoch": 37.57, "grad_norm": 6.34375, "learning_rate": 1.4360085135391224e-05, "loss": 1.2947, "step": 15028 }, { "epoch": 37.5725, "grad_norm": 37.75, "learning_rate": 1.4359356447678683e-05, "loss": 1.1827, "step": 15029 }, { "epoch": 37.575, "grad_norm": 7.28125, "learning_rate": 1.4358627731386369e-05, "loss": 1.4961, "step": 15030 }, { "epoch": 37.5775, "grad_norm": 7.9375, "learning_rate": 1.435789898651907e-05, "loss": 1.3888, "step": 15031 }, { "epoch": 37.58, "grad_norm": 5.9375, "learning_rate": 1.4357170213081556e-05, "loss": 1.374, "step": 15032 }, { "epoch": 37.5825, "grad_norm": 8.125, "learning_rate": 1.4356441411078614e-05, "loss": 1.4031, "step": 15033 }, { "epoch": 37.585, "grad_norm": 7.1875, "learning_rate": 1.4355712580515013e-05, "loss": 1.2586, "step": 15034 }, { "epoch": 37.5875, "grad_norm": 6.59375, "learning_rate": 1.4354983721395534e-05, "loss": 1.1765, "step": 15035 }, { "epoch": 37.59, "grad_norm": 7.96875, "learning_rate": 1.4354254833724958e-05, "loss": 1.2177, "step": 15036 }, { "epoch": 37.5925, "grad_norm": 6.4375, "learning_rate": 1.435352591750806e-05, "loss": 1.2844, "step": 15037 }, { "epoch": 37.595, "grad_norm": 7.875, "learning_rate": 1.4352796972749621e-05, "loss": 1.4231, "step": 15038 }, { "epoch": 37.5975, "grad_norm": 6.84375, "learning_rate": 1.435206799945442e-05, "loss": 1.1986, "step": 15039 }, { "epoch": 37.6, "grad_norm": 7.90625, "learning_rate": 1.4351338997627233e-05, "loss": 1.3491, "step": 15040 }, { "epoch": 37.6025, "grad_norm": 8.1875, "learning_rate": 1.4350609967272844e-05, "loss": 1.268, "step": 15041 }, { "epoch": 37.605, "grad_norm": 7.15625, "learning_rate": 1.4349880908396027e-05, "loss": 1.4462, "step": 15042 }, { "epoch": 37.6075, "grad_norm": 6.625, "learning_rate": 1.434915182100157e-05, "loss": 1.1855, "step": 15043 }, { "epoch": 37.61, "grad_norm": 7.9375, "learning_rate": 1.4348422705094245e-05, "loss": 1.3934, "step": 15044 }, { "epoch": 37.6125, "grad_norm": 7.625, "learning_rate": 1.4347693560678832e-05, "loss": 1.3083, "step": 15045 }, { "epoch": 37.615, "grad_norm": 7.5625, "learning_rate": 1.4346964387760121e-05, "loss": 1.2492, "step": 15046 }, { "epoch": 37.6175, "grad_norm": 7.40625, "learning_rate": 1.4346235186342878e-05, "loss": 1.3216, "step": 15047 }, { "epoch": 37.62, "grad_norm": 8.5625, "learning_rate": 1.4345505956431896e-05, "loss": 1.3201, "step": 15048 }, { "epoch": 37.6225, "grad_norm": 5.8125, "learning_rate": 1.4344776698031949e-05, "loss": 1.3865, "step": 15049 }, { "epoch": 37.625, "grad_norm": 7.15625, "learning_rate": 1.434404741114782e-05, "loss": 1.4042, "step": 15050 }, { "epoch": 37.6275, "grad_norm": 5.53125, "learning_rate": 1.434331809578429e-05, "loss": 1.3129, "step": 15051 }, { "epoch": 37.63, "grad_norm": 6.34375, "learning_rate": 1.4342588751946141e-05, "loss": 1.1843, "step": 15052 }, { "epoch": 37.6325, "grad_norm": 8.375, "learning_rate": 1.434185937963815e-05, "loss": 1.4305, "step": 15053 }, { "epoch": 37.635, "grad_norm": 7.21875, "learning_rate": 1.434112997886511e-05, "loss": 1.394, "step": 15054 }, { "epoch": 37.6375, "grad_norm": 8.4375, "learning_rate": 1.4340400549631788e-05, "loss": 1.252, "step": 15055 }, { "epoch": 37.64, "grad_norm": 6.34375, "learning_rate": 1.433967109194298e-05, "loss": 1.2993, "step": 15056 }, { "epoch": 37.6425, "grad_norm": 8.0625, "learning_rate": 1.4338941605803456e-05, "loss": 1.2875, "step": 15057 }, { "epoch": 37.645, "grad_norm": 8.4375, "learning_rate": 1.4338212091218009e-05, "loss": 1.5066, "step": 15058 }, { "epoch": 37.6475, "grad_norm": 7.4375, "learning_rate": 1.4337482548191414e-05, "loss": 1.4196, "step": 15059 }, { "epoch": 37.65, "grad_norm": 6.25, "learning_rate": 1.433675297672846e-05, "loss": 1.3898, "step": 15060 }, { "epoch": 37.6525, "grad_norm": 7.84375, "learning_rate": 1.4336023376833926e-05, "loss": 1.317, "step": 15061 }, { "epoch": 37.655, "grad_norm": 5.5625, "learning_rate": 1.4335293748512596e-05, "loss": 1.3078, "step": 15062 }, { "epoch": 37.6575, "grad_norm": 5.84375, "learning_rate": 1.4334564091769254e-05, "loss": 1.4135, "step": 15063 }, { "epoch": 37.66, "grad_norm": 7.9375, "learning_rate": 1.4333834406608682e-05, "loss": 1.3256, "step": 15064 }, { "epoch": 37.6625, "grad_norm": 3.96875, "learning_rate": 1.4333104693035663e-05, "loss": 1.2754, "step": 15065 }, { "epoch": 37.665, "grad_norm": 9.1875, "learning_rate": 1.4332374951054987e-05, "loss": 1.3235, "step": 15066 }, { "epoch": 37.6675, "grad_norm": 7.125, "learning_rate": 1.4331645180671435e-05, "loss": 1.2795, "step": 15067 }, { "epoch": 37.67, "grad_norm": 7.9375, "learning_rate": 1.4330915381889788e-05, "loss": 1.3259, "step": 15068 }, { "epoch": 37.6725, "grad_norm": 6.75, "learning_rate": 1.4330185554714834e-05, "loss": 1.249, "step": 15069 }, { "epoch": 37.675, "grad_norm": 8.0, "learning_rate": 1.4329455699151356e-05, "loss": 1.491, "step": 15070 }, { "epoch": 37.6775, "grad_norm": 8.625, "learning_rate": 1.4328725815204143e-05, "loss": 1.3319, "step": 15071 }, { "epoch": 37.68, "grad_norm": 6.78125, "learning_rate": 1.4327995902877972e-05, "loss": 1.2693, "step": 15072 }, { "epoch": 37.6825, "grad_norm": 8.25, "learning_rate": 1.4327265962177637e-05, "loss": 1.3627, "step": 15073 }, { "epoch": 37.685, "grad_norm": 6.59375, "learning_rate": 1.4326535993107917e-05, "loss": 1.3523, "step": 15074 }, { "epoch": 37.6875, "grad_norm": 6.78125, "learning_rate": 1.43258059956736e-05, "loss": 1.3553, "step": 15075 }, { "epoch": 37.69, "grad_norm": 10.875, "learning_rate": 1.4325075969879473e-05, "loss": 1.3692, "step": 15076 }, { "epoch": 37.6925, "grad_norm": 8.625, "learning_rate": 1.432434591573032e-05, "loss": 1.4086, "step": 15077 }, { "epoch": 37.695, "grad_norm": 7.03125, "learning_rate": 1.4323615833230929e-05, "loss": 1.3284, "step": 15078 }, { "epoch": 37.6975, "grad_norm": 5.0, "learning_rate": 1.4322885722386086e-05, "loss": 1.3753, "step": 15079 }, { "epoch": 37.7, "grad_norm": 7.46875, "learning_rate": 1.4322155583200577e-05, "loss": 1.3037, "step": 15080 }, { "epoch": 37.7025, "grad_norm": 7.65625, "learning_rate": 1.432142541567919e-05, "loss": 1.2207, "step": 15081 }, { "epoch": 37.705, "grad_norm": 8.5, "learning_rate": 1.4320695219826708e-05, "loss": 1.4286, "step": 15082 }, { "epoch": 37.7075, "grad_norm": 6.46875, "learning_rate": 1.4319964995647922e-05, "loss": 1.3452, "step": 15083 }, { "epoch": 37.71, "grad_norm": 11.0, "learning_rate": 1.4319234743147619e-05, "loss": 1.2979, "step": 15084 }, { "epoch": 37.7125, "grad_norm": 6.8125, "learning_rate": 1.4318504462330587e-05, "loss": 1.3658, "step": 15085 }, { "epoch": 37.715, "grad_norm": 6.53125, "learning_rate": 1.4317774153201613e-05, "loss": 1.4777, "step": 15086 }, { "epoch": 37.7175, "grad_norm": 8.3125, "learning_rate": 1.4317043815765481e-05, "loss": 1.3473, "step": 15087 }, { "epoch": 37.72, "grad_norm": 8.1875, "learning_rate": 1.4316313450026986e-05, "loss": 1.3022, "step": 15088 }, { "epoch": 37.7225, "grad_norm": 6.8125, "learning_rate": 1.431558305599091e-05, "loss": 1.2045, "step": 15089 }, { "epoch": 37.725, "grad_norm": 7.03125, "learning_rate": 1.4314852633662046e-05, "loss": 1.3713, "step": 15090 }, { "epoch": 37.7275, "grad_norm": 7.28125, "learning_rate": 1.4314122183045181e-05, "loss": 1.2651, "step": 15091 }, { "epoch": 37.73, "grad_norm": 9.625, "learning_rate": 1.43133917041451e-05, "loss": 1.469, "step": 15092 }, { "epoch": 37.7325, "grad_norm": 10.0625, "learning_rate": 1.43126611969666e-05, "loss": 1.4775, "step": 15093 }, { "epoch": 37.735, "grad_norm": 7.625, "learning_rate": 1.4311930661514464e-05, "loss": 1.3981, "step": 15094 }, { "epoch": 37.7375, "grad_norm": 8.4375, "learning_rate": 1.4311200097793484e-05, "loss": 1.2534, "step": 15095 }, { "epoch": 37.74, "grad_norm": 9.875, "learning_rate": 1.4310469505808447e-05, "loss": 1.3324, "step": 15096 }, { "epoch": 37.7425, "grad_norm": 8.75, "learning_rate": 1.4309738885564146e-05, "loss": 1.3576, "step": 15097 }, { "epoch": 37.745, "grad_norm": 4.53125, "learning_rate": 1.4309008237065367e-05, "loss": 1.3789, "step": 15098 }, { "epoch": 37.7475, "grad_norm": 8.5625, "learning_rate": 1.4308277560316906e-05, "loss": 1.553, "step": 15099 }, { "epoch": 37.75, "grad_norm": 7.09375, "learning_rate": 1.4307546855323549e-05, "loss": 1.2254, "step": 15100 }, { "epoch": 37.7525, "grad_norm": 6.40625, "learning_rate": 1.4306816122090086e-05, "loss": 1.2763, "step": 15101 }, { "epoch": 37.755, "grad_norm": 9.0625, "learning_rate": 1.4306085360621307e-05, "loss": 1.3963, "step": 15102 }, { "epoch": 37.7575, "grad_norm": 6.625, "learning_rate": 1.430535457092201e-05, "loss": 1.4417, "step": 15103 }, { "epoch": 37.76, "grad_norm": 7.3125, "learning_rate": 1.4304623752996974e-05, "loss": 1.1641, "step": 15104 }, { "epoch": 37.7625, "grad_norm": 7.65625, "learning_rate": 1.4303892906851002e-05, "loss": 1.4137, "step": 15105 }, { "epoch": 37.765, "grad_norm": 7.125, "learning_rate": 1.430316203248888e-05, "loss": 1.2345, "step": 15106 }, { "epoch": 37.7675, "grad_norm": 9.75, "learning_rate": 1.4302431129915398e-05, "loss": 1.376, "step": 15107 }, { "epoch": 37.77, "grad_norm": 7.3125, "learning_rate": 1.430170019913535e-05, "loss": 1.3113, "step": 15108 }, { "epoch": 37.7725, "grad_norm": 6.78125, "learning_rate": 1.4300969240153527e-05, "loss": 1.2537, "step": 15109 }, { "epoch": 37.775, "grad_norm": 7.03125, "learning_rate": 1.4300238252974722e-05, "loss": 1.3878, "step": 15110 }, { "epoch": 37.7775, "grad_norm": 10.5625, "learning_rate": 1.429950723760373e-05, "loss": 1.2387, "step": 15111 }, { "epoch": 37.78, "grad_norm": 15.3125, "learning_rate": 1.4298776194045337e-05, "loss": 1.3506, "step": 15112 }, { "epoch": 37.7825, "grad_norm": 5.34375, "learning_rate": 1.429804512230434e-05, "loss": 1.3163, "step": 15113 }, { "epoch": 37.785, "grad_norm": 8.5625, "learning_rate": 1.4297314022385532e-05, "loss": 1.4915, "step": 15114 }, { "epoch": 37.7875, "grad_norm": 8.6875, "learning_rate": 1.4296582894293705e-05, "loss": 1.5787, "step": 15115 }, { "epoch": 37.79, "grad_norm": 7.75, "learning_rate": 1.4295851738033655e-05, "loss": 1.3772, "step": 15116 }, { "epoch": 37.7925, "grad_norm": 7.875, "learning_rate": 1.4295120553610166e-05, "loss": 1.3786, "step": 15117 }, { "epoch": 37.795, "grad_norm": 8.875, "learning_rate": 1.4294389341028048e-05, "loss": 1.297, "step": 15118 }, { "epoch": 37.7975, "grad_norm": 9.0, "learning_rate": 1.4293658100292076e-05, "loss": 1.3435, "step": 15119 }, { "epoch": 37.8, "grad_norm": 7.6875, "learning_rate": 1.429292683140706e-05, "loss": 1.3377, "step": 15120 }, { "epoch": 37.8025, "grad_norm": 9.3125, "learning_rate": 1.4292195534377787e-05, "loss": 1.2394, "step": 15121 }, { "epoch": 37.805, "grad_norm": 8.6875, "learning_rate": 1.4291464209209047e-05, "loss": 1.3349, "step": 15122 }, { "epoch": 37.8075, "grad_norm": 7.0, "learning_rate": 1.4290732855905644e-05, "loss": 1.4292, "step": 15123 }, { "epoch": 37.81, "grad_norm": 7.0, "learning_rate": 1.4290001474472366e-05, "loss": 1.3274, "step": 15124 }, { "epoch": 37.8125, "grad_norm": 7.0625, "learning_rate": 1.4289270064914012e-05, "loss": 1.362, "step": 15125 }, { "epoch": 37.815, "grad_norm": 5.84375, "learning_rate": 1.4288538627235373e-05, "loss": 1.2456, "step": 15126 }, { "epoch": 37.8175, "grad_norm": 7.0625, "learning_rate": 1.4287807161441246e-05, "loss": 1.4287, "step": 15127 }, { "epoch": 37.82, "grad_norm": 8.5625, "learning_rate": 1.428707566753643e-05, "loss": 1.2506, "step": 15128 }, { "epoch": 37.8225, "grad_norm": 10.75, "learning_rate": 1.4286344145525713e-05, "loss": 1.4058, "step": 15129 }, { "epoch": 37.825, "grad_norm": 5.46875, "learning_rate": 1.42856125954139e-05, "loss": 1.3124, "step": 15130 }, { "epoch": 37.8275, "grad_norm": 8.625, "learning_rate": 1.4284881017205778e-05, "loss": 1.3312, "step": 15131 }, { "epoch": 37.83, "grad_norm": 9.3125, "learning_rate": 1.428414941090615e-05, "loss": 1.389, "step": 15132 }, { "epoch": 37.8325, "grad_norm": 8.3125, "learning_rate": 1.428341777651981e-05, "loss": 1.2315, "step": 15133 }, { "epoch": 37.835, "grad_norm": 10.5, "learning_rate": 1.4282686114051551e-05, "loss": 1.4194, "step": 15134 }, { "epoch": 37.8375, "grad_norm": 7.09375, "learning_rate": 1.4281954423506176e-05, "loss": 1.5616, "step": 15135 }, { "epoch": 37.84, "grad_norm": 7.9375, "learning_rate": 1.428122270488848e-05, "loss": 1.3072, "step": 15136 }, { "epoch": 37.8425, "grad_norm": 7.6875, "learning_rate": 1.428049095820326e-05, "loss": 1.3709, "step": 15137 }, { "epoch": 37.845, "grad_norm": 7.0625, "learning_rate": 1.4279759183455311e-05, "loss": 1.3617, "step": 15138 }, { "epoch": 37.8475, "grad_norm": 6.28125, "learning_rate": 1.4279027380649431e-05, "loss": 1.5244, "step": 15139 }, { "epoch": 37.85, "grad_norm": 8.1875, "learning_rate": 1.4278295549790419e-05, "loss": 1.3168, "step": 15140 }, { "epoch": 37.8525, "grad_norm": 8.9375, "learning_rate": 1.4277563690883076e-05, "loss": 1.3724, "step": 15141 }, { "epoch": 37.855, "grad_norm": 6.46875, "learning_rate": 1.4276831803932192e-05, "loss": 1.2649, "step": 15142 }, { "epoch": 37.8575, "grad_norm": 7.4375, "learning_rate": 1.4276099888942573e-05, "loss": 1.3947, "step": 15143 }, { "epoch": 37.86, "grad_norm": 9.25, "learning_rate": 1.4275367945919013e-05, "loss": 1.412, "step": 15144 }, { "epoch": 37.8625, "grad_norm": 6.90625, "learning_rate": 1.4274635974866314e-05, "loss": 1.305, "step": 15145 }, { "epoch": 37.865, "grad_norm": 9.1875, "learning_rate": 1.4273903975789269e-05, "loss": 1.4286, "step": 15146 }, { "epoch": 37.8675, "grad_norm": 8.875, "learning_rate": 1.4273171948692683e-05, "loss": 1.3426, "step": 15147 }, { "epoch": 37.87, "grad_norm": 12.75, "learning_rate": 1.427243989358135e-05, "loss": 1.2628, "step": 15148 }, { "epoch": 37.8725, "grad_norm": 7.125, "learning_rate": 1.4271707810460074e-05, "loss": 1.2481, "step": 15149 }, { "epoch": 37.875, "grad_norm": 8.5625, "learning_rate": 1.4270975699333653e-05, "loss": 1.424, "step": 15150 }, { "epoch": 37.8775, "grad_norm": 7.59375, "learning_rate": 1.4270243560206885e-05, "loss": 1.37, "step": 15151 }, { "epoch": 37.88, "grad_norm": 6.875, "learning_rate": 1.4269511393084572e-05, "loss": 1.4749, "step": 15152 }, { "epoch": 37.8825, "grad_norm": 6.125, "learning_rate": 1.4268779197971512e-05, "loss": 1.3039, "step": 15153 }, { "epoch": 37.885, "grad_norm": 9.625, "learning_rate": 1.426804697487251e-05, "loss": 1.3801, "step": 15154 }, { "epoch": 37.8875, "grad_norm": 6.65625, "learning_rate": 1.4267314723792359e-05, "loss": 1.2889, "step": 15155 }, { "epoch": 37.89, "grad_norm": 10.3125, "learning_rate": 1.4266582444735866e-05, "loss": 1.3741, "step": 15156 }, { "epoch": 37.8925, "grad_norm": 10.9375, "learning_rate": 1.4265850137707826e-05, "loss": 1.3577, "step": 15157 }, { "epoch": 37.895, "grad_norm": 6.3125, "learning_rate": 1.4265117802713046e-05, "loss": 1.4117, "step": 15158 }, { "epoch": 37.8975, "grad_norm": 8.4375, "learning_rate": 1.4264385439756324e-05, "loss": 1.2755, "step": 15159 }, { "epoch": 37.9, "grad_norm": 6.84375, "learning_rate": 1.4263653048842461e-05, "loss": 1.5802, "step": 15160 }, { "epoch": 37.9025, "grad_norm": 8.125, "learning_rate": 1.426292062997626e-05, "loss": 1.344, "step": 15161 }, { "epoch": 37.905, "grad_norm": 6.25, "learning_rate": 1.4262188183162519e-05, "loss": 1.4165, "step": 15162 }, { "epoch": 37.9075, "grad_norm": 6.59375, "learning_rate": 1.4261455708406047e-05, "loss": 1.4803, "step": 15163 }, { "epoch": 37.91, "grad_norm": 7.4375, "learning_rate": 1.426072320571164e-05, "loss": 1.3483, "step": 15164 }, { "epoch": 37.9125, "grad_norm": 7.9375, "learning_rate": 1.4259990675084102e-05, "loss": 1.3667, "step": 15165 }, { "epoch": 37.915, "grad_norm": 6.53125, "learning_rate": 1.4259258116528234e-05, "loss": 1.4542, "step": 15166 }, { "epoch": 37.9175, "grad_norm": 6.5625, "learning_rate": 1.4258525530048843e-05, "loss": 1.3631, "step": 15167 }, { "epoch": 37.92, "grad_norm": 6.78125, "learning_rate": 1.4257792915650728e-05, "loss": 1.1826, "step": 15168 }, { "epoch": 37.9225, "grad_norm": 6.21875, "learning_rate": 1.4257060273338693e-05, "loss": 1.3114, "step": 15169 }, { "epoch": 37.925, "grad_norm": 8.0625, "learning_rate": 1.4256327603117542e-05, "loss": 1.4044, "step": 15170 }, { "epoch": 37.9275, "grad_norm": 9.0, "learning_rate": 1.4255594904992078e-05, "loss": 1.203, "step": 15171 }, { "epoch": 37.93, "grad_norm": 9.5625, "learning_rate": 1.4254862178967098e-05, "loss": 1.297, "step": 15172 }, { "epoch": 37.9325, "grad_norm": 7.625, "learning_rate": 1.425412942504742e-05, "loss": 1.4981, "step": 15173 }, { "epoch": 37.935, "grad_norm": 10.6875, "learning_rate": 1.4253396643237835e-05, "loss": 1.4192, "step": 15174 }, { "epoch": 37.9375, "grad_norm": 6.78125, "learning_rate": 1.4252663833543157e-05, "loss": 1.3324, "step": 15175 }, { "epoch": 37.94, "grad_norm": 4.71875, "learning_rate": 1.4251930995968179e-05, "loss": 1.2066, "step": 15176 }, { "epoch": 37.9425, "grad_norm": 10.8125, "learning_rate": 1.4251198130517714e-05, "loss": 1.5333, "step": 15177 }, { "epoch": 37.945, "grad_norm": 7.875, "learning_rate": 1.4250465237196565e-05, "loss": 1.1507, "step": 15178 }, { "epoch": 37.9475, "grad_norm": 6.5625, "learning_rate": 1.4249732316009535e-05, "loss": 1.2607, "step": 15179 }, { "epoch": 37.95, "grad_norm": 5.40625, "learning_rate": 1.424899936696143e-05, "loss": 1.2847, "step": 15180 }, { "epoch": 37.9525, "grad_norm": 7.53125, "learning_rate": 1.4248266390057055e-05, "loss": 1.4501, "step": 15181 }, { "epoch": 37.955, "grad_norm": 8.75, "learning_rate": 1.4247533385301216e-05, "loss": 1.3229, "step": 15182 }, { "epoch": 37.9575, "grad_norm": 9.5, "learning_rate": 1.4246800352698718e-05, "loss": 1.2273, "step": 15183 }, { "epoch": 37.96, "grad_norm": 7.4375, "learning_rate": 1.4246067292254367e-05, "loss": 1.409, "step": 15184 }, { "epoch": 37.9625, "grad_norm": 9.6875, "learning_rate": 1.424533420397297e-05, "loss": 1.4103, "step": 15185 }, { "epoch": 37.965, "grad_norm": 7.34375, "learning_rate": 1.4244601087859331e-05, "loss": 1.4278, "step": 15186 }, { "epoch": 37.9675, "grad_norm": 5.875, "learning_rate": 1.4243867943918255e-05, "loss": 1.2742, "step": 15187 }, { "epoch": 37.97, "grad_norm": 5.5, "learning_rate": 1.4243134772154554e-05, "loss": 1.4147, "step": 15188 }, { "epoch": 37.9725, "grad_norm": 6.59375, "learning_rate": 1.4242401572573026e-05, "loss": 1.184, "step": 15189 }, { "epoch": 37.975, "grad_norm": 7.1875, "learning_rate": 1.4241668345178488e-05, "loss": 1.3966, "step": 15190 }, { "epoch": 37.9775, "grad_norm": 6.09375, "learning_rate": 1.424093508997574e-05, "loss": 1.3398, "step": 15191 }, { "epoch": 37.98, "grad_norm": 6.53125, "learning_rate": 1.4240201806969594e-05, "loss": 1.3314, "step": 15192 }, { "epoch": 37.9825, "grad_norm": 6.625, "learning_rate": 1.4239468496164852e-05, "loss": 1.4994, "step": 15193 }, { "epoch": 37.985, "grad_norm": 6.8125, "learning_rate": 1.4238735157566323e-05, "loss": 1.408, "step": 15194 }, { "epoch": 37.9875, "grad_norm": 5.15625, "learning_rate": 1.4238001791178816e-05, "loss": 1.2089, "step": 15195 }, { "epoch": 37.99, "grad_norm": 6.6875, "learning_rate": 1.4237268397007145e-05, "loss": 1.3468, "step": 15196 }, { "epoch": 37.9925, "grad_norm": 11.125, "learning_rate": 1.4236534975056104e-05, "loss": 1.4223, "step": 15197 }, { "epoch": 37.995, "grad_norm": 6.15625, "learning_rate": 1.4235801525330514e-05, "loss": 1.4533, "step": 15198 }, { "epoch": 37.9975, "grad_norm": 8.625, "learning_rate": 1.4235068047835177e-05, "loss": 1.3826, "step": 15199 }, { "epoch": 38.0, "grad_norm": 7.4375, "learning_rate": 1.4234334542574906e-05, "loss": 1.4156, "step": 15200 }, { "epoch": 38.0025, "grad_norm": 8.5, "learning_rate": 1.4233601009554503e-05, "loss": 1.2307, "step": 15201 }, { "epoch": 38.005, "grad_norm": 9.25, "learning_rate": 1.4232867448778783e-05, "loss": 1.4454, "step": 15202 }, { "epoch": 38.0075, "grad_norm": 6.875, "learning_rate": 1.4232133860252556e-05, "loss": 1.2034, "step": 15203 }, { "epoch": 38.01, "grad_norm": 6.375, "learning_rate": 1.4231400243980623e-05, "loss": 1.4327, "step": 15204 }, { "epoch": 38.0125, "grad_norm": 6.3125, "learning_rate": 1.4230666599967803e-05, "loss": 1.3674, "step": 15205 }, { "epoch": 38.015, "grad_norm": 9.5, "learning_rate": 1.4229932928218899e-05, "loss": 1.4181, "step": 15206 }, { "epoch": 38.0175, "grad_norm": 5.125, "learning_rate": 1.4229199228738728e-05, "loss": 1.2911, "step": 15207 }, { "epoch": 38.02, "grad_norm": 11.5, "learning_rate": 1.4228465501532096e-05, "loss": 1.439, "step": 15208 }, { "epoch": 38.0225, "grad_norm": 6.21875, "learning_rate": 1.4227731746603812e-05, "loss": 1.3547, "step": 15209 }, { "epoch": 38.025, "grad_norm": 8.375, "learning_rate": 1.4226997963958686e-05, "loss": 1.4086, "step": 15210 }, { "epoch": 38.0275, "grad_norm": 5.90625, "learning_rate": 1.4226264153601535e-05, "loss": 1.4497, "step": 15211 }, { "epoch": 38.03, "grad_norm": 7.3125, "learning_rate": 1.422553031553716e-05, "loss": 1.3506, "step": 15212 }, { "epoch": 38.0325, "grad_norm": 5.8125, "learning_rate": 1.4224796449770384e-05, "loss": 1.5253, "step": 15213 }, { "epoch": 38.035, "grad_norm": 8.1875, "learning_rate": 1.4224062556306004e-05, "loss": 1.4069, "step": 15214 }, { "epoch": 38.0375, "grad_norm": 9.3125, "learning_rate": 1.4223328635148844e-05, "loss": 1.2768, "step": 15215 }, { "epoch": 38.04, "grad_norm": 8.125, "learning_rate": 1.4222594686303707e-05, "loss": 1.3621, "step": 15216 }, { "epoch": 38.0425, "grad_norm": 8.1875, "learning_rate": 1.4221860709775413e-05, "loss": 1.5483, "step": 15217 }, { "epoch": 38.045, "grad_norm": 10.8125, "learning_rate": 1.4221126705568766e-05, "loss": 1.1363, "step": 15218 }, { "epoch": 38.0475, "grad_norm": 8.5625, "learning_rate": 1.4220392673688581e-05, "loss": 1.3727, "step": 15219 }, { "epoch": 38.05, "grad_norm": 7.25, "learning_rate": 1.4219658614139674e-05, "loss": 1.4081, "step": 15220 }, { "epoch": 38.0525, "grad_norm": 6.34375, "learning_rate": 1.421892452692685e-05, "loss": 1.4374, "step": 15221 }, { "epoch": 38.055, "grad_norm": 5.5625, "learning_rate": 1.4218190412054927e-05, "loss": 1.277, "step": 15222 }, { "epoch": 38.0575, "grad_norm": 5.9375, "learning_rate": 1.4217456269528715e-05, "loss": 1.439, "step": 15223 }, { "epoch": 38.06, "grad_norm": 7.34375, "learning_rate": 1.421672209935303e-05, "loss": 1.2812, "step": 15224 }, { "epoch": 38.0625, "grad_norm": 7.15625, "learning_rate": 1.4215987901532684e-05, "loss": 1.3572, "step": 15225 }, { "epoch": 38.065, "grad_norm": 6.3125, "learning_rate": 1.4215253676072488e-05, "loss": 1.324, "step": 15226 }, { "epoch": 38.0675, "grad_norm": 6.8125, "learning_rate": 1.421451942297726e-05, "loss": 1.3444, "step": 15227 }, { "epoch": 38.07, "grad_norm": 8.5, "learning_rate": 1.4213785142251813e-05, "loss": 1.4667, "step": 15228 }, { "epoch": 38.0725, "grad_norm": 6.4375, "learning_rate": 1.4213050833900953e-05, "loss": 1.1498, "step": 15229 }, { "epoch": 38.075, "grad_norm": 8.25, "learning_rate": 1.4212316497929505e-05, "loss": 1.2016, "step": 15230 }, { "epoch": 38.0775, "grad_norm": 5.5625, "learning_rate": 1.4211582134342277e-05, "loss": 1.3383, "step": 15231 }, { "epoch": 38.08, "grad_norm": 6.25, "learning_rate": 1.4210847743144087e-05, "loss": 1.2344, "step": 15232 }, { "epoch": 38.0825, "grad_norm": 6.65625, "learning_rate": 1.4210113324339748e-05, "loss": 1.3259, "step": 15233 }, { "epoch": 38.085, "grad_norm": 11.25, "learning_rate": 1.4209378877934071e-05, "loss": 1.3005, "step": 15234 }, { "epoch": 38.0875, "grad_norm": 6.65625, "learning_rate": 1.4208644403931878e-05, "loss": 1.3502, "step": 15235 }, { "epoch": 38.09, "grad_norm": 11.875, "learning_rate": 1.4207909902337978e-05, "loss": 1.5301, "step": 15236 }, { "epoch": 38.0925, "grad_norm": 5.6875, "learning_rate": 1.4207175373157189e-05, "loss": 1.3601, "step": 15237 }, { "epoch": 38.095, "grad_norm": 9.1875, "learning_rate": 1.4206440816394327e-05, "loss": 1.3485, "step": 15238 }, { "epoch": 38.0975, "grad_norm": 8.6875, "learning_rate": 1.4205706232054208e-05, "loss": 1.2668, "step": 15239 }, { "epoch": 38.1, "grad_norm": 6.78125, "learning_rate": 1.4204971620141648e-05, "loss": 1.5562, "step": 15240 }, { "epoch": 38.1025, "grad_norm": 7.1875, "learning_rate": 1.4204236980661462e-05, "loss": 1.3627, "step": 15241 }, { "epoch": 38.105, "grad_norm": 8.9375, "learning_rate": 1.4203502313618464e-05, "loss": 1.4334, "step": 15242 }, { "epoch": 38.1075, "grad_norm": 7.375, "learning_rate": 1.4202767619017479e-05, "loss": 1.2613, "step": 15243 }, { "epoch": 38.11, "grad_norm": 8.4375, "learning_rate": 1.420203289686331e-05, "loss": 1.5562, "step": 15244 }, { "epoch": 38.1125, "grad_norm": 7.46875, "learning_rate": 1.4201298147160789e-05, "loss": 1.3434, "step": 15245 }, { "epoch": 38.115, "grad_norm": 8.625, "learning_rate": 1.4200563369914719e-05, "loss": 1.2341, "step": 15246 }, { "epoch": 38.1175, "grad_norm": 7.0, "learning_rate": 1.4199828565129927e-05, "loss": 1.59, "step": 15247 }, { "epoch": 38.12, "grad_norm": 6.59375, "learning_rate": 1.4199093732811227e-05, "loss": 1.4075, "step": 15248 }, { "epoch": 38.1225, "grad_norm": 8.5625, "learning_rate": 1.4198358872963436e-05, "loss": 1.4008, "step": 15249 }, { "epoch": 38.125, "grad_norm": 7.28125, "learning_rate": 1.4197623985591373e-05, "loss": 1.3224, "step": 15250 }, { "epoch": 38.1275, "grad_norm": 11.0, "learning_rate": 1.4196889070699856e-05, "loss": 1.41, "step": 15251 }, { "epoch": 38.13, "grad_norm": 4.65625, "learning_rate": 1.4196154128293698e-05, "loss": 1.2429, "step": 15252 }, { "epoch": 38.1325, "grad_norm": 9.0, "learning_rate": 1.4195419158377726e-05, "loss": 1.2316, "step": 15253 }, { "epoch": 38.135, "grad_norm": 10.9375, "learning_rate": 1.419468416095675e-05, "loss": 1.4047, "step": 15254 }, { "epoch": 38.1375, "grad_norm": 5.25, "learning_rate": 1.4193949136035596e-05, "loss": 1.2847, "step": 15255 }, { "epoch": 38.14, "grad_norm": 7.5625, "learning_rate": 1.4193214083619075e-05, "loss": 1.3428, "step": 15256 }, { "epoch": 38.1425, "grad_norm": 7.90625, "learning_rate": 1.4192479003712014e-05, "loss": 1.4237, "step": 15257 }, { "epoch": 38.145, "grad_norm": 11.25, "learning_rate": 1.419174389631923e-05, "loss": 1.3369, "step": 15258 }, { "epoch": 38.1475, "grad_norm": 9.875, "learning_rate": 1.4191008761445537e-05, "loss": 1.4901, "step": 15259 }, { "epoch": 38.15, "grad_norm": 8.5625, "learning_rate": 1.4190273599095761e-05, "loss": 1.2943, "step": 15260 }, { "epoch": 38.1525, "grad_norm": 10.4375, "learning_rate": 1.4189538409274715e-05, "loss": 1.5203, "step": 15261 }, { "epoch": 38.155, "grad_norm": 9.0, "learning_rate": 1.4188803191987228e-05, "loss": 1.3785, "step": 15262 }, { "epoch": 38.1575, "grad_norm": 6.875, "learning_rate": 1.418806794723811e-05, "loss": 1.2982, "step": 15263 }, { "epoch": 38.16, "grad_norm": 8.375, "learning_rate": 1.4187332675032189e-05, "loss": 1.2814, "step": 15264 }, { "epoch": 38.1625, "grad_norm": 7.84375, "learning_rate": 1.4186597375374283e-05, "loss": 1.3195, "step": 15265 }, { "epoch": 38.165, "grad_norm": 7.59375, "learning_rate": 1.418586204826921e-05, "loss": 1.4371, "step": 15266 }, { "epoch": 38.1675, "grad_norm": 6.71875, "learning_rate": 1.4185126693721793e-05, "loss": 1.1843, "step": 15267 }, { "epoch": 38.17, "grad_norm": 5.875, "learning_rate": 1.4184391311736856e-05, "loss": 1.2979, "step": 15268 }, { "epoch": 38.1725, "grad_norm": 7.625, "learning_rate": 1.4183655902319213e-05, "loss": 1.5491, "step": 15269 }, { "epoch": 38.175, "grad_norm": 7.65625, "learning_rate": 1.4182920465473692e-05, "loss": 1.3734, "step": 15270 }, { "epoch": 38.1775, "grad_norm": 10.625, "learning_rate": 1.4182185001205111e-05, "loss": 1.3867, "step": 15271 }, { "epoch": 38.18, "grad_norm": 6.875, "learning_rate": 1.4181449509518292e-05, "loss": 1.2813, "step": 15272 }, { "epoch": 38.1825, "grad_norm": 7.0, "learning_rate": 1.418071399041806e-05, "loss": 1.2829, "step": 15273 }, { "epoch": 38.185, "grad_norm": 8.875, "learning_rate": 1.417997844390923e-05, "loss": 1.3348, "step": 15274 }, { "epoch": 38.1875, "grad_norm": 6.71875, "learning_rate": 1.4179242869996632e-05, "loss": 1.2974, "step": 15275 }, { "epoch": 38.19, "grad_norm": 6.25, "learning_rate": 1.4178507268685082e-05, "loss": 1.206, "step": 15276 }, { "epoch": 38.1925, "grad_norm": 8.375, "learning_rate": 1.417777163997941e-05, "loss": 1.3777, "step": 15277 }, { "epoch": 38.195, "grad_norm": 8.625, "learning_rate": 1.4177035983884431e-05, "loss": 1.4352, "step": 15278 }, { "epoch": 38.1975, "grad_norm": 7.8125, "learning_rate": 1.4176300300404973e-05, "loss": 1.4121, "step": 15279 }, { "epoch": 38.2, "grad_norm": 8.375, "learning_rate": 1.4175564589545853e-05, "loss": 1.2881, "step": 15280 }, { "epoch": 38.2025, "grad_norm": 6.3125, "learning_rate": 1.4174828851311904e-05, "loss": 1.235, "step": 15281 }, { "epoch": 38.205, "grad_norm": 6.46875, "learning_rate": 1.4174093085707941e-05, "loss": 1.3211, "step": 15282 }, { "epoch": 38.2075, "grad_norm": 6.34375, "learning_rate": 1.4173357292738794e-05, "loss": 1.3368, "step": 15283 }, { "epoch": 38.21, "grad_norm": 7.8125, "learning_rate": 1.417262147240928e-05, "loss": 1.3236, "step": 15284 }, { "epoch": 38.2125, "grad_norm": 8.5625, "learning_rate": 1.4171885624724229e-05, "loss": 1.3713, "step": 15285 }, { "epoch": 38.215, "grad_norm": 7.875, "learning_rate": 1.417114974968846e-05, "loss": 1.2967, "step": 15286 }, { "epoch": 38.2175, "grad_norm": 7.28125, "learning_rate": 1.4170413847306806e-05, "loss": 1.3082, "step": 15287 }, { "epoch": 38.22, "grad_norm": 7.28125, "learning_rate": 1.416967791758408e-05, "loss": 1.2728, "step": 15288 }, { "epoch": 38.2225, "grad_norm": 9.0, "learning_rate": 1.4168941960525113e-05, "loss": 1.4342, "step": 15289 }, { "epoch": 38.225, "grad_norm": 7.03125, "learning_rate": 1.416820597613473e-05, "loss": 1.4245, "step": 15290 }, { "epoch": 38.2275, "grad_norm": 9.125, "learning_rate": 1.4167469964417758e-05, "loss": 1.3855, "step": 15291 }, { "epoch": 38.23, "grad_norm": 9.875, "learning_rate": 1.4166733925379017e-05, "loss": 1.5102, "step": 15292 }, { "epoch": 38.2325, "grad_norm": 8.125, "learning_rate": 1.4165997859023335e-05, "loss": 1.3372, "step": 15293 }, { "epoch": 38.235, "grad_norm": 6.25, "learning_rate": 1.4165261765355539e-05, "loss": 1.2745, "step": 15294 }, { "epoch": 38.2375, "grad_norm": 6.9375, "learning_rate": 1.4164525644380453e-05, "loss": 1.2525, "step": 15295 }, { "epoch": 38.24, "grad_norm": 7.75, "learning_rate": 1.4163789496102902e-05, "loss": 1.3076, "step": 15296 }, { "epoch": 38.2425, "grad_norm": 5.84375, "learning_rate": 1.4163053320527716e-05, "loss": 1.2124, "step": 15297 }, { "epoch": 38.245, "grad_norm": 6.125, "learning_rate": 1.416231711765972e-05, "loss": 1.3192, "step": 15298 }, { "epoch": 38.2475, "grad_norm": 7.59375, "learning_rate": 1.4161580887503736e-05, "loss": 1.4189, "step": 15299 }, { "epoch": 38.25, "grad_norm": 10.5, "learning_rate": 1.4160844630064596e-05, "loss": 1.3517, "step": 15300 }, { "epoch": 38.2525, "grad_norm": 14.4375, "learning_rate": 1.4160108345347125e-05, "loss": 1.3967, "step": 15301 }, { "epoch": 38.255, "grad_norm": 6.25, "learning_rate": 1.4159372033356151e-05, "loss": 1.3881, "step": 15302 }, { "epoch": 38.2575, "grad_norm": 8.6875, "learning_rate": 1.4158635694096501e-05, "loss": 1.3469, "step": 15303 }, { "epoch": 38.26, "grad_norm": 8.75, "learning_rate": 1.4157899327572999e-05, "loss": 1.3147, "step": 15304 }, { "epoch": 38.2625, "grad_norm": 6.125, "learning_rate": 1.4157162933790478e-05, "loss": 1.3157, "step": 15305 }, { "epoch": 38.265, "grad_norm": 7.25, "learning_rate": 1.415642651275376e-05, "loss": 1.2969, "step": 15306 }, { "epoch": 38.2675, "grad_norm": 6.21875, "learning_rate": 1.415569006446768e-05, "loss": 1.3621, "step": 15307 }, { "epoch": 38.27, "grad_norm": 7.90625, "learning_rate": 1.4154953588937059e-05, "loss": 1.2443, "step": 15308 }, { "epoch": 38.2725, "grad_norm": 8.0, "learning_rate": 1.4154217086166731e-05, "loss": 1.3838, "step": 15309 }, { "epoch": 38.275, "grad_norm": 7.09375, "learning_rate": 1.4153480556161521e-05, "loss": 1.3001, "step": 15310 }, { "epoch": 38.2775, "grad_norm": 10.125, "learning_rate": 1.415274399892626e-05, "loss": 1.1397, "step": 15311 }, { "epoch": 38.28, "grad_norm": 10.1875, "learning_rate": 1.4152007414465771e-05, "loss": 1.4397, "step": 15312 }, { "epoch": 38.2825, "grad_norm": 7.40625, "learning_rate": 1.4151270802784894e-05, "loss": 1.4004, "step": 15313 }, { "epoch": 38.285, "grad_norm": 7.09375, "learning_rate": 1.4150534163888447e-05, "loss": 1.1687, "step": 15314 }, { "epoch": 38.2875, "grad_norm": 4.4375, "learning_rate": 1.4149797497781266e-05, "loss": 1.3369, "step": 15315 }, { "epoch": 38.29, "grad_norm": 9.1875, "learning_rate": 1.4149060804468178e-05, "loss": 1.3303, "step": 15316 }, { "epoch": 38.2925, "grad_norm": 8.9375, "learning_rate": 1.4148324083954015e-05, "loss": 1.3964, "step": 15317 }, { "epoch": 38.295, "grad_norm": 7.90625, "learning_rate": 1.4147587336243604e-05, "loss": 1.2691, "step": 15318 }, { "epoch": 38.2975, "grad_norm": 9.875, "learning_rate": 1.4146850561341777e-05, "loss": 1.4082, "step": 15319 }, { "epoch": 38.3, "grad_norm": 9.4375, "learning_rate": 1.4146113759253362e-05, "loss": 1.4024, "step": 15320 }, { "epoch": 38.3025, "grad_norm": 7.40625, "learning_rate": 1.4145376929983192e-05, "loss": 1.3169, "step": 15321 }, { "epoch": 38.305, "grad_norm": 5.8125, "learning_rate": 1.4144640073536095e-05, "loss": 1.1548, "step": 15322 }, { "epoch": 38.3075, "grad_norm": 6.21875, "learning_rate": 1.4143903189916908e-05, "loss": 1.2839, "step": 15323 }, { "epoch": 38.31, "grad_norm": 6.84375, "learning_rate": 1.4143166279130454e-05, "loss": 1.2805, "step": 15324 }, { "epoch": 38.3125, "grad_norm": 8.6875, "learning_rate": 1.4142429341181569e-05, "loss": 1.3835, "step": 15325 }, { "epoch": 38.315, "grad_norm": 6.9375, "learning_rate": 1.4141692376075082e-05, "loss": 1.4702, "step": 15326 }, { "epoch": 38.3175, "grad_norm": 11.1875, "learning_rate": 1.4140955383815827e-05, "loss": 1.4509, "step": 15327 }, { "epoch": 38.32, "grad_norm": 10.4375, "learning_rate": 1.4140218364408634e-05, "loss": 1.4053, "step": 15328 }, { "epoch": 38.3225, "grad_norm": 7.53125, "learning_rate": 1.4139481317858333e-05, "loss": 1.2128, "step": 15329 }, { "epoch": 38.325, "grad_norm": 8.5, "learning_rate": 1.4138744244169762e-05, "loss": 1.6083, "step": 15330 }, { "epoch": 38.3275, "grad_norm": 6.78125, "learning_rate": 1.4138007143347743e-05, "loss": 1.3554, "step": 15331 }, { "epoch": 38.33, "grad_norm": 4.9375, "learning_rate": 1.4137270015397122e-05, "loss": 1.4022, "step": 15332 }, { "epoch": 38.3325, "grad_norm": 6.21875, "learning_rate": 1.4136532860322718e-05, "loss": 1.4341, "step": 15333 }, { "epoch": 38.335, "grad_norm": 6.53125, "learning_rate": 1.4135795678129376e-05, "loss": 1.2964, "step": 15334 }, { "epoch": 38.3375, "grad_norm": 9.25, "learning_rate": 1.4135058468821918e-05, "loss": 1.5793, "step": 15335 }, { "epoch": 38.34, "grad_norm": 6.90625, "learning_rate": 1.4134321232405185e-05, "loss": 1.3116, "step": 15336 }, { "epoch": 38.3425, "grad_norm": 7.28125, "learning_rate": 1.4133583968884008e-05, "loss": 1.352, "step": 15337 }, { "epoch": 38.345, "grad_norm": 5.34375, "learning_rate": 1.4132846678263219e-05, "loss": 1.2283, "step": 15338 }, { "epoch": 38.3475, "grad_norm": 7.5625, "learning_rate": 1.4132109360547652e-05, "loss": 1.4677, "step": 15339 }, { "epoch": 38.35, "grad_norm": 9.5625, "learning_rate": 1.4131372015742141e-05, "loss": 1.3425, "step": 15340 }, { "epoch": 38.3525, "grad_norm": 8.5, "learning_rate": 1.413063464385152e-05, "loss": 1.5059, "step": 15341 }, { "epoch": 38.355, "grad_norm": 7.5, "learning_rate": 1.4129897244880625e-05, "loss": 1.4378, "step": 15342 }, { "epoch": 38.3575, "grad_norm": 7.09375, "learning_rate": 1.4129159818834288e-05, "loss": 1.2699, "step": 15343 }, { "epoch": 38.36, "grad_norm": 10.3125, "learning_rate": 1.4128422365717346e-05, "loss": 1.4227, "step": 15344 }, { "epoch": 38.3625, "grad_norm": 8.75, "learning_rate": 1.4127684885534633e-05, "loss": 1.5452, "step": 15345 }, { "epoch": 38.365, "grad_norm": 8.8125, "learning_rate": 1.4126947378290978e-05, "loss": 1.2985, "step": 15346 }, { "epoch": 38.3675, "grad_norm": 8.875, "learning_rate": 1.4126209843991228e-05, "loss": 1.3591, "step": 15347 }, { "epoch": 38.37, "grad_norm": 7.0, "learning_rate": 1.4125472282640204e-05, "loss": 1.3047, "step": 15348 }, { "epoch": 38.3725, "grad_norm": 7.21875, "learning_rate": 1.4124734694242754e-05, "loss": 1.3929, "step": 15349 }, { "epoch": 38.375, "grad_norm": 7.59375, "learning_rate": 1.4123997078803708e-05, "loss": 1.3496, "step": 15350 }, { "epoch": 38.3775, "grad_norm": 7.5625, "learning_rate": 1.4123259436327903e-05, "loss": 1.3353, "step": 15351 }, { "epoch": 38.38, "grad_norm": 9.125, "learning_rate": 1.4122521766820172e-05, "loss": 1.1917, "step": 15352 }, { "epoch": 38.3825, "grad_norm": 7.9375, "learning_rate": 1.4121784070285354e-05, "loss": 1.448, "step": 15353 }, { "epoch": 38.385, "grad_norm": 8.375, "learning_rate": 1.4121046346728282e-05, "loss": 1.4034, "step": 15354 }, { "epoch": 38.3875, "grad_norm": 8.875, "learning_rate": 1.41203085961538e-05, "loss": 1.2956, "step": 15355 }, { "epoch": 38.39, "grad_norm": 9.625, "learning_rate": 1.4119570818566735e-05, "loss": 1.3389, "step": 15356 }, { "epoch": 38.3925, "grad_norm": 6.15625, "learning_rate": 1.411883301397193e-05, "loss": 1.3259, "step": 15357 }, { "epoch": 38.395, "grad_norm": 8.5, "learning_rate": 1.4118095182374221e-05, "loss": 1.3696, "step": 15358 }, { "epoch": 38.3975, "grad_norm": 10.125, "learning_rate": 1.4117357323778446e-05, "loss": 1.2792, "step": 15359 }, { "epoch": 38.4, "grad_norm": 6.5, "learning_rate": 1.411661943818944e-05, "loss": 1.3189, "step": 15360 }, { "epoch": 38.4025, "grad_norm": 9.1875, "learning_rate": 1.4115881525612043e-05, "loss": 1.3835, "step": 15361 }, { "epoch": 38.405, "grad_norm": 8.5625, "learning_rate": 1.411514358605109e-05, "loss": 1.1851, "step": 15362 }, { "epoch": 38.4075, "grad_norm": 8.6875, "learning_rate": 1.4114405619511419e-05, "loss": 1.3069, "step": 15363 }, { "epoch": 38.41, "grad_norm": 7.875, "learning_rate": 1.4113667625997872e-05, "loss": 1.3347, "step": 15364 }, { "epoch": 38.4125, "grad_norm": 6.5, "learning_rate": 1.4112929605515285e-05, "loss": 1.2397, "step": 15365 }, { "epoch": 38.415, "grad_norm": 9.125, "learning_rate": 1.4112191558068493e-05, "loss": 1.4689, "step": 15366 }, { "epoch": 38.4175, "grad_norm": 7.625, "learning_rate": 1.411145348366234e-05, "loss": 1.2514, "step": 15367 }, { "epoch": 38.42, "grad_norm": 5.6875, "learning_rate": 1.4110715382301661e-05, "loss": 1.2658, "step": 15368 }, { "epoch": 38.4225, "grad_norm": 6.0625, "learning_rate": 1.4109977253991299e-05, "loss": 1.4653, "step": 15369 }, { "epoch": 38.425, "grad_norm": 7.09375, "learning_rate": 1.410923909873609e-05, "loss": 1.3202, "step": 15370 }, { "epoch": 38.4275, "grad_norm": 6.90625, "learning_rate": 1.4108500916540871e-05, "loss": 1.3498, "step": 15371 }, { "epoch": 38.43, "grad_norm": 7.1875, "learning_rate": 1.4107762707410486e-05, "loss": 1.4436, "step": 15372 }, { "epoch": 38.4325, "grad_norm": 7.96875, "learning_rate": 1.4107024471349772e-05, "loss": 1.356, "step": 15373 }, { "epoch": 38.435, "grad_norm": 6.46875, "learning_rate": 1.4106286208363574e-05, "loss": 1.3362, "step": 15374 }, { "epoch": 38.4375, "grad_norm": 7.28125, "learning_rate": 1.4105547918456726e-05, "loss": 1.3905, "step": 15375 }, { "epoch": 38.44, "grad_norm": 10.0, "learning_rate": 1.4104809601634069e-05, "loss": 1.339, "step": 15376 }, { "epoch": 38.4425, "grad_norm": 8.5, "learning_rate": 1.4104071257900445e-05, "loss": 1.4, "step": 15377 }, { "epoch": 38.445, "grad_norm": 8.4375, "learning_rate": 1.4103332887260692e-05, "loss": 1.3518, "step": 15378 }, { "epoch": 38.4475, "grad_norm": 13.875, "learning_rate": 1.4102594489719654e-05, "loss": 1.3804, "step": 15379 }, { "epoch": 38.45, "grad_norm": 7.5625, "learning_rate": 1.4101856065282174e-05, "loss": 1.3684, "step": 15380 }, { "epoch": 38.4525, "grad_norm": 7.96875, "learning_rate": 1.4101117613953086e-05, "loss": 1.5427, "step": 15381 }, { "epoch": 38.455, "grad_norm": 8.75, "learning_rate": 1.4100379135737237e-05, "loss": 1.1999, "step": 15382 }, { "epoch": 38.4575, "grad_norm": 8.875, "learning_rate": 1.4099640630639464e-05, "loss": 1.2915, "step": 15383 }, { "epoch": 38.46, "grad_norm": 8.375, "learning_rate": 1.4098902098664614e-05, "loss": 1.3142, "step": 15384 }, { "epoch": 38.4625, "grad_norm": 8.9375, "learning_rate": 1.4098163539817527e-05, "loss": 1.233, "step": 15385 }, { "epoch": 38.465, "grad_norm": 7.8125, "learning_rate": 1.4097424954103039e-05, "loss": 1.3708, "step": 15386 }, { "epoch": 38.4675, "grad_norm": 5.71875, "learning_rate": 1.4096686341526e-05, "loss": 1.3047, "step": 15387 }, { "epoch": 38.47, "grad_norm": 8.3125, "learning_rate": 1.4095947702091248e-05, "loss": 1.5336, "step": 15388 }, { "epoch": 38.4725, "grad_norm": 7.15625, "learning_rate": 1.4095209035803627e-05, "loss": 1.2106, "step": 15389 }, { "epoch": 38.475, "grad_norm": 7.5625, "learning_rate": 1.409447034266798e-05, "loss": 1.2526, "step": 15390 }, { "epoch": 38.4775, "grad_norm": 7.96875, "learning_rate": 1.4093731622689151e-05, "loss": 1.412, "step": 15391 }, { "epoch": 38.48, "grad_norm": 4.90625, "learning_rate": 1.409299287587198e-05, "loss": 1.5914, "step": 15392 }, { "epoch": 38.4825, "grad_norm": 6.71875, "learning_rate": 1.409225410222131e-05, "loss": 1.3326, "step": 15393 }, { "epoch": 38.485, "grad_norm": 6.46875, "learning_rate": 1.4091515301741987e-05, "loss": 1.3191, "step": 15394 }, { "epoch": 38.4875, "grad_norm": 11.0625, "learning_rate": 1.4090776474438853e-05, "loss": 1.2716, "step": 15395 }, { "epoch": 38.49, "grad_norm": 7.8125, "learning_rate": 1.4090037620316752e-05, "loss": 1.3774, "step": 15396 }, { "epoch": 38.4925, "grad_norm": 6.21875, "learning_rate": 1.4089298739380528e-05, "loss": 1.2573, "step": 15397 }, { "epoch": 38.495, "grad_norm": 6.65625, "learning_rate": 1.4088559831635027e-05, "loss": 1.4115, "step": 15398 }, { "epoch": 38.4975, "grad_norm": 5.125, "learning_rate": 1.4087820897085089e-05, "loss": 1.3377, "step": 15399 }, { "epoch": 38.5, "grad_norm": 7.25, "learning_rate": 1.4087081935735565e-05, "loss": 1.4048, "step": 15400 }, { "epoch": 38.5025, "grad_norm": 6.6875, "learning_rate": 1.408634294759129e-05, "loss": 1.3953, "step": 15401 }, { "epoch": 38.505, "grad_norm": 10.125, "learning_rate": 1.4085603932657117e-05, "loss": 1.3769, "step": 15402 }, { "epoch": 38.5075, "grad_norm": 7.03125, "learning_rate": 1.4084864890937886e-05, "loss": 1.4992, "step": 15403 }, { "epoch": 38.51, "grad_norm": 11.625, "learning_rate": 1.4084125822438447e-05, "loss": 1.5182, "step": 15404 }, { "epoch": 38.5125, "grad_norm": 8.0625, "learning_rate": 1.4083386727163641e-05, "loss": 1.3394, "step": 15405 }, { "epoch": 38.515, "grad_norm": 7.0625, "learning_rate": 1.4082647605118316e-05, "loss": 1.31, "step": 15406 }, { "epoch": 38.5175, "grad_norm": 9.625, "learning_rate": 1.4081908456307316e-05, "loss": 1.4921, "step": 15407 }, { "epoch": 38.52, "grad_norm": 9.875, "learning_rate": 1.4081169280735488e-05, "loss": 1.5675, "step": 15408 }, { "epoch": 38.5225, "grad_norm": 6.09375, "learning_rate": 1.4080430078407677e-05, "loss": 1.3903, "step": 15409 }, { "epoch": 38.525, "grad_norm": 7.625, "learning_rate": 1.4079690849328731e-05, "loss": 1.3664, "step": 15410 }, { "epoch": 38.5275, "grad_norm": 7.1875, "learning_rate": 1.4078951593503492e-05, "loss": 1.2527, "step": 15411 }, { "epoch": 38.53, "grad_norm": 8.4375, "learning_rate": 1.407821231093681e-05, "loss": 1.3643, "step": 15412 }, { "epoch": 38.5325, "grad_norm": 8.5625, "learning_rate": 1.4077473001633533e-05, "loss": 1.4722, "step": 15413 }, { "epoch": 38.535, "grad_norm": 8.375, "learning_rate": 1.4076733665598506e-05, "loss": 1.4521, "step": 15414 }, { "epoch": 38.5375, "grad_norm": 6.5, "learning_rate": 1.4075994302836578e-05, "loss": 1.191, "step": 15415 }, { "epoch": 38.54, "grad_norm": 6.84375, "learning_rate": 1.407525491335259e-05, "loss": 1.1996, "step": 15416 }, { "epoch": 38.5425, "grad_norm": 7.875, "learning_rate": 1.4074515497151397e-05, "loss": 1.3196, "step": 15417 }, { "epoch": 38.545, "grad_norm": 6.375, "learning_rate": 1.4073776054237842e-05, "loss": 1.3326, "step": 15418 }, { "epoch": 38.5475, "grad_norm": 7.40625, "learning_rate": 1.4073036584616774e-05, "loss": 1.3818, "step": 15419 }, { "epoch": 38.55, "grad_norm": 8.625, "learning_rate": 1.4072297088293043e-05, "loss": 1.4042, "step": 15420 }, { "epoch": 38.5525, "grad_norm": 7.15625, "learning_rate": 1.4071557565271495e-05, "loss": 1.368, "step": 15421 }, { "epoch": 38.555, "grad_norm": 7.1875, "learning_rate": 1.4070818015556977e-05, "loss": 1.3628, "step": 15422 }, { "epoch": 38.5575, "grad_norm": 6.21875, "learning_rate": 1.4070078439154343e-05, "loss": 1.2379, "step": 15423 }, { "epoch": 38.56, "grad_norm": 4.21875, "learning_rate": 1.4069338836068434e-05, "loss": 1.3321, "step": 15424 }, { "epoch": 38.5625, "grad_norm": 8.125, "learning_rate": 1.4068599206304104e-05, "loss": 1.2577, "step": 15425 }, { "epoch": 38.565, "grad_norm": 7.125, "learning_rate": 1.4067859549866196e-05, "loss": 1.4657, "step": 15426 }, { "epoch": 38.5675, "grad_norm": 7.5, "learning_rate": 1.406711986675957e-05, "loss": 1.404, "step": 15427 }, { "epoch": 38.57, "grad_norm": 6.6875, "learning_rate": 1.4066380156989063e-05, "loss": 1.1605, "step": 15428 }, { "epoch": 38.5725, "grad_norm": 5.90625, "learning_rate": 1.4065640420559537e-05, "loss": 1.2813, "step": 15429 }, { "epoch": 38.575, "grad_norm": 6.4375, "learning_rate": 1.406490065747583e-05, "loss": 1.3379, "step": 15430 }, { "epoch": 38.5775, "grad_norm": 8.3125, "learning_rate": 1.4064160867742798e-05, "loss": 1.3071, "step": 15431 }, { "epoch": 38.58, "grad_norm": 7.875, "learning_rate": 1.406342105136529e-05, "loss": 1.1584, "step": 15432 }, { "epoch": 38.5825, "grad_norm": 7.15625, "learning_rate": 1.4062681208348156e-05, "loss": 1.3459, "step": 15433 }, { "epoch": 38.585, "grad_norm": 7.4375, "learning_rate": 1.4061941338696248e-05, "loss": 1.4528, "step": 15434 }, { "epoch": 38.5875, "grad_norm": 6.96875, "learning_rate": 1.4061201442414414e-05, "loss": 1.3196, "step": 15435 }, { "epoch": 38.59, "grad_norm": 5.6875, "learning_rate": 1.4060461519507506e-05, "loss": 1.4301, "step": 15436 }, { "epoch": 38.5925, "grad_norm": 5.84375, "learning_rate": 1.4059721569980375e-05, "loss": 1.5135, "step": 15437 }, { "epoch": 38.595, "grad_norm": 7.90625, "learning_rate": 1.4058981593837872e-05, "loss": 1.2954, "step": 15438 }, { "epoch": 38.5975, "grad_norm": 8.1875, "learning_rate": 1.4058241591084847e-05, "loss": 1.5459, "step": 15439 }, { "epoch": 38.6, "grad_norm": 9.6875, "learning_rate": 1.4057501561726157e-05, "loss": 1.3108, "step": 15440 }, { "epoch": 38.6025, "grad_norm": 7.75, "learning_rate": 1.4056761505766643e-05, "loss": 1.4341, "step": 15441 }, { "epoch": 38.605, "grad_norm": 7.46875, "learning_rate": 1.4056021423211165e-05, "loss": 1.2563, "step": 15442 }, { "epoch": 38.6075, "grad_norm": 6.9375, "learning_rate": 1.4055281314064572e-05, "loss": 1.2662, "step": 15443 }, { "epoch": 38.61, "grad_norm": 8.5, "learning_rate": 1.405454117833172e-05, "loss": 1.3809, "step": 15444 }, { "epoch": 38.6125, "grad_norm": 12.0, "learning_rate": 1.4053801016017456e-05, "loss": 1.3935, "step": 15445 }, { "epoch": 38.615, "grad_norm": 5.46875, "learning_rate": 1.4053060827126636e-05, "loss": 1.4131, "step": 15446 }, { "epoch": 38.6175, "grad_norm": 5.75, "learning_rate": 1.405232061166411e-05, "loss": 1.2232, "step": 15447 }, { "epoch": 38.62, "grad_norm": 17.75, "learning_rate": 1.4051580369634734e-05, "loss": 1.4654, "step": 15448 }, { "epoch": 38.6225, "grad_norm": 6.8125, "learning_rate": 1.4050840101043357e-05, "loss": 1.3212, "step": 15449 }, { "epoch": 38.625, "grad_norm": 6.15625, "learning_rate": 1.4050099805894837e-05, "loss": 1.3115, "step": 15450 }, { "epoch": 38.6275, "grad_norm": 7.71875, "learning_rate": 1.4049359484194024e-05, "loss": 1.411, "step": 15451 }, { "epoch": 38.63, "grad_norm": 6.71875, "learning_rate": 1.4048619135945769e-05, "loss": 1.3773, "step": 15452 }, { "epoch": 38.6325, "grad_norm": 8.875, "learning_rate": 1.4047878761154934e-05, "loss": 1.3105, "step": 15453 }, { "epoch": 38.635, "grad_norm": 8.375, "learning_rate": 1.4047138359826364e-05, "loss": 1.3207, "step": 15454 }, { "epoch": 38.6375, "grad_norm": 9.5, "learning_rate": 1.4046397931964917e-05, "loss": 1.3762, "step": 15455 }, { "epoch": 38.64, "grad_norm": 8.25, "learning_rate": 1.404565747757545e-05, "loss": 1.3225, "step": 15456 }, { "epoch": 38.6425, "grad_norm": 8.5, "learning_rate": 1.4044916996662814e-05, "loss": 1.4701, "step": 15457 }, { "epoch": 38.645, "grad_norm": 8.0, "learning_rate": 1.404417648923186e-05, "loss": 1.3609, "step": 15458 }, { "epoch": 38.6475, "grad_norm": 9.75, "learning_rate": 1.4043435955287452e-05, "loss": 1.3379, "step": 15459 }, { "epoch": 38.65, "grad_norm": 6.09375, "learning_rate": 1.4042695394834435e-05, "loss": 1.3319, "step": 15460 }, { "epoch": 38.6525, "grad_norm": 8.875, "learning_rate": 1.4041954807877673e-05, "loss": 1.3398, "step": 15461 }, { "epoch": 38.655, "grad_norm": 7.5, "learning_rate": 1.4041214194422015e-05, "loss": 1.2823, "step": 15462 }, { "epoch": 38.6575, "grad_norm": 7.15625, "learning_rate": 1.4040473554472321e-05, "loss": 1.1103, "step": 15463 }, { "epoch": 38.66, "grad_norm": 6.09375, "learning_rate": 1.403973288803344e-05, "loss": 1.2371, "step": 15464 }, { "epoch": 38.6625, "grad_norm": 7.53125, "learning_rate": 1.4038992195110233e-05, "loss": 1.4219, "step": 15465 }, { "epoch": 38.665, "grad_norm": 8.6875, "learning_rate": 1.4038251475707556e-05, "loss": 1.2862, "step": 15466 }, { "epoch": 38.6675, "grad_norm": 4.34375, "learning_rate": 1.4037510729830262e-05, "loss": 1.4167, "step": 15467 }, { "epoch": 38.67, "grad_norm": 11.0, "learning_rate": 1.4036769957483212e-05, "loss": 1.4733, "step": 15468 }, { "epoch": 38.6725, "grad_norm": 7.125, "learning_rate": 1.4036029158671257e-05, "loss": 1.291, "step": 15469 }, { "epoch": 38.675, "grad_norm": 14.875, "learning_rate": 1.4035288333399257e-05, "loss": 1.4911, "step": 15470 }, { "epoch": 38.6775, "grad_norm": 5.65625, "learning_rate": 1.403454748167207e-05, "loss": 1.4133, "step": 15471 }, { "epoch": 38.68, "grad_norm": 5.71875, "learning_rate": 1.403380660349455e-05, "loss": 1.3491, "step": 15472 }, { "epoch": 38.6825, "grad_norm": 13.25, "learning_rate": 1.4033065698871552e-05, "loss": 1.2195, "step": 15473 }, { "epoch": 38.685, "grad_norm": 8.125, "learning_rate": 1.4032324767807943e-05, "loss": 1.396, "step": 15474 }, { "epoch": 38.6875, "grad_norm": 6.53125, "learning_rate": 1.4031583810308568e-05, "loss": 1.3453, "step": 15475 }, { "epoch": 38.69, "grad_norm": 4.78125, "learning_rate": 1.4030842826378297e-05, "loss": 1.3775, "step": 15476 }, { "epoch": 38.6925, "grad_norm": 7.3125, "learning_rate": 1.403010181602198e-05, "loss": 1.4249, "step": 15477 }, { "epoch": 38.695, "grad_norm": 6.1875, "learning_rate": 1.4029360779244474e-05, "loss": 1.3615, "step": 15478 }, { "epoch": 38.6975, "grad_norm": 6.84375, "learning_rate": 1.4028619716050642e-05, "loss": 1.2634, "step": 15479 }, { "epoch": 38.7, "grad_norm": 6.53125, "learning_rate": 1.4027878626445339e-05, "loss": 1.2476, "step": 15480 }, { "epoch": 38.7025, "grad_norm": 7.1875, "learning_rate": 1.4027137510433426e-05, "loss": 1.2753, "step": 15481 }, { "epoch": 38.705, "grad_norm": 6.625, "learning_rate": 1.4026396368019758e-05, "loss": 1.365, "step": 15482 }, { "epoch": 38.7075, "grad_norm": 9.1875, "learning_rate": 1.40256551992092e-05, "loss": 1.412, "step": 15483 }, { "epoch": 38.71, "grad_norm": 7.78125, "learning_rate": 1.4024914004006607e-05, "loss": 1.1998, "step": 15484 }, { "epoch": 38.7125, "grad_norm": 8.3125, "learning_rate": 1.4024172782416838e-05, "loss": 1.481, "step": 15485 }, { "epoch": 38.715, "grad_norm": 8.625, "learning_rate": 1.4023431534444754e-05, "loss": 1.3451, "step": 15486 }, { "epoch": 38.7175, "grad_norm": 5.59375, "learning_rate": 1.4022690260095215e-05, "loss": 1.4559, "step": 15487 }, { "epoch": 38.72, "grad_norm": 7.21875, "learning_rate": 1.4021948959373075e-05, "loss": 1.3107, "step": 15488 }, { "epoch": 38.7225, "grad_norm": 8.8125, "learning_rate": 1.4021207632283205e-05, "loss": 1.2858, "step": 15489 }, { "epoch": 38.725, "grad_norm": 7.875, "learning_rate": 1.4020466278830452e-05, "loss": 1.3753, "step": 15490 }, { "epoch": 38.7275, "grad_norm": 7.25, "learning_rate": 1.4019724899019689e-05, "loss": 1.3428, "step": 15491 }, { "epoch": 38.73, "grad_norm": 6.0625, "learning_rate": 1.4018983492855767e-05, "loss": 1.2936, "step": 15492 }, { "epoch": 38.7325, "grad_norm": 6.90625, "learning_rate": 1.401824206034355e-05, "loss": 1.2918, "step": 15493 }, { "epoch": 38.735, "grad_norm": 13.3125, "learning_rate": 1.4017500601487902e-05, "loss": 1.3848, "step": 15494 }, { "epoch": 38.7375, "grad_norm": 6.4375, "learning_rate": 1.4016759116293677e-05, "loss": 1.3676, "step": 15495 }, { "epoch": 38.74, "grad_norm": 5.09375, "learning_rate": 1.4016017604765742e-05, "loss": 1.2616, "step": 15496 }, { "epoch": 38.7425, "grad_norm": 7.78125, "learning_rate": 1.4015276066908956e-05, "loss": 1.3792, "step": 15497 }, { "epoch": 38.745, "grad_norm": 8.0625, "learning_rate": 1.4014534502728178e-05, "loss": 1.3603, "step": 15498 }, { "epoch": 38.7475, "grad_norm": 8.0625, "learning_rate": 1.401379291222828e-05, "loss": 1.2673, "step": 15499 }, { "epoch": 38.75, "grad_norm": 5.9375, "learning_rate": 1.4013051295414108e-05, "loss": 1.3931, "step": 15500 }, { "epoch": 38.7525, "grad_norm": 8.125, "learning_rate": 1.4012309652290536e-05, "loss": 1.3194, "step": 15501 }, { "epoch": 38.755, "grad_norm": 6.1875, "learning_rate": 1.4011567982862423e-05, "loss": 1.2213, "step": 15502 }, { "epoch": 38.7575, "grad_norm": 7.5, "learning_rate": 1.4010826287134629e-05, "loss": 1.2775, "step": 15503 }, { "epoch": 38.76, "grad_norm": 8.125, "learning_rate": 1.4010084565112018e-05, "loss": 1.3191, "step": 15504 }, { "epoch": 38.7625, "grad_norm": 7.96875, "learning_rate": 1.4009342816799456e-05, "loss": 1.4315, "step": 15505 }, { "epoch": 38.765, "grad_norm": 8.875, "learning_rate": 1.4008601042201802e-05, "loss": 1.3135, "step": 15506 }, { "epoch": 38.7675, "grad_norm": 5.375, "learning_rate": 1.4007859241323918e-05, "loss": 1.2312, "step": 15507 }, { "epoch": 38.77, "grad_norm": 6.0, "learning_rate": 1.400711741417067e-05, "loss": 1.2543, "step": 15508 }, { "epoch": 38.7725, "grad_norm": 6.9375, "learning_rate": 1.4006375560746921e-05, "loss": 1.2587, "step": 15509 }, { "epoch": 38.775, "grad_norm": 9.6875, "learning_rate": 1.4005633681057536e-05, "loss": 1.3576, "step": 15510 }, { "epoch": 38.7775, "grad_norm": 6.78125, "learning_rate": 1.4004891775107374e-05, "loss": 1.4017, "step": 15511 }, { "epoch": 38.78, "grad_norm": 5.96875, "learning_rate": 1.4004149842901305e-05, "loss": 1.4369, "step": 15512 }, { "epoch": 38.7825, "grad_norm": 9.0625, "learning_rate": 1.4003407884444185e-05, "loss": 1.3031, "step": 15513 }, { "epoch": 38.785, "grad_norm": 7.84375, "learning_rate": 1.4002665899740889e-05, "loss": 1.2924, "step": 15514 }, { "epoch": 38.7875, "grad_norm": 8.25, "learning_rate": 1.400192388879627e-05, "loss": 1.4642, "step": 15515 }, { "epoch": 38.79, "grad_norm": 7.3125, "learning_rate": 1.4001181851615203e-05, "loss": 1.3196, "step": 15516 }, { "epoch": 38.7925, "grad_norm": 15.4375, "learning_rate": 1.4000439788202544e-05, "loss": 1.2121, "step": 15517 }, { "epoch": 38.795, "grad_norm": 5.6875, "learning_rate": 1.3999697698563162e-05, "loss": 1.4406, "step": 15518 }, { "epoch": 38.7975, "grad_norm": 4.8125, "learning_rate": 1.3998955582701923e-05, "loss": 1.1781, "step": 15519 }, { "epoch": 38.8, "grad_norm": 5.96875, "learning_rate": 1.3998213440623691e-05, "loss": 1.3744, "step": 15520 }, { "epoch": 38.8025, "grad_norm": 6.84375, "learning_rate": 1.3997471272333334e-05, "loss": 1.2734, "step": 15521 }, { "epoch": 38.805, "grad_norm": 6.125, "learning_rate": 1.3996729077835713e-05, "loss": 1.2707, "step": 15522 }, { "epoch": 38.8075, "grad_norm": 7.21875, "learning_rate": 1.3995986857135697e-05, "loss": 1.363, "step": 15523 }, { "epoch": 38.81, "grad_norm": 6.25, "learning_rate": 1.399524461023815e-05, "loss": 1.2415, "step": 15524 }, { "epoch": 38.8125, "grad_norm": 6.4375, "learning_rate": 1.399450233714794e-05, "loss": 1.2522, "step": 15525 }, { "epoch": 38.815, "grad_norm": 5.625, "learning_rate": 1.3993760037869932e-05, "loss": 1.3414, "step": 15526 }, { "epoch": 38.8175, "grad_norm": 9.1875, "learning_rate": 1.3993017712408995e-05, "loss": 1.3306, "step": 15527 }, { "epoch": 38.82, "grad_norm": 7.21875, "learning_rate": 1.3992275360769989e-05, "loss": 1.3247, "step": 15528 }, { "epoch": 38.8225, "grad_norm": 7.96875, "learning_rate": 1.3991532982957792e-05, "loss": 1.3718, "step": 15529 }, { "epoch": 38.825, "grad_norm": 6.34375, "learning_rate": 1.3990790578977258e-05, "loss": 1.2707, "step": 15530 }, { "epoch": 38.8275, "grad_norm": 7.53125, "learning_rate": 1.3990048148833267e-05, "loss": 1.397, "step": 15531 }, { "epoch": 38.83, "grad_norm": 6.0, "learning_rate": 1.3989305692530675e-05, "loss": 1.4114, "step": 15532 }, { "epoch": 38.8325, "grad_norm": 10.125, "learning_rate": 1.3988563210074358e-05, "loss": 1.2574, "step": 15533 }, { "epoch": 38.835, "grad_norm": 7.65625, "learning_rate": 1.3987820701469176e-05, "loss": 1.3302, "step": 15534 }, { "epoch": 38.8375, "grad_norm": 7.71875, "learning_rate": 1.3987078166720003e-05, "loss": 1.4212, "step": 15535 }, { "epoch": 38.84, "grad_norm": 6.5, "learning_rate": 1.3986335605831707e-05, "loss": 1.1347, "step": 15536 }, { "epoch": 38.8425, "grad_norm": 8.0, "learning_rate": 1.3985593018809152e-05, "loss": 1.5881, "step": 15537 }, { "epoch": 38.845, "grad_norm": 7.6875, "learning_rate": 1.398485040565721e-05, "loss": 1.4895, "step": 15538 }, { "epoch": 38.8475, "grad_norm": 7.1875, "learning_rate": 1.3984107766380748e-05, "loss": 1.3991, "step": 15539 }, { "epoch": 38.85, "grad_norm": 6.46875, "learning_rate": 1.3983365100984633e-05, "loss": 1.3757, "step": 15540 }, { "epoch": 38.8525, "grad_norm": 7.59375, "learning_rate": 1.3982622409473739e-05, "loss": 1.2172, "step": 15541 }, { "epoch": 38.855, "grad_norm": 6.03125, "learning_rate": 1.3981879691852932e-05, "loss": 1.4815, "step": 15542 }, { "epoch": 38.8575, "grad_norm": 7.21875, "learning_rate": 1.3981136948127076e-05, "loss": 1.3837, "step": 15543 }, { "epoch": 38.86, "grad_norm": 7.0625, "learning_rate": 1.3980394178301051e-05, "loss": 1.1901, "step": 15544 }, { "epoch": 38.8625, "grad_norm": 8.25, "learning_rate": 1.3979651382379717e-05, "loss": 1.3976, "step": 15545 }, { "epoch": 38.865, "grad_norm": 5.9375, "learning_rate": 1.3978908560367953e-05, "loss": 1.3436, "step": 15546 }, { "epoch": 38.8675, "grad_norm": 5.0, "learning_rate": 1.3978165712270619e-05, "loss": 1.3048, "step": 15547 }, { "epoch": 38.87, "grad_norm": 6.46875, "learning_rate": 1.3977422838092592e-05, "loss": 1.4025, "step": 15548 }, { "epoch": 38.8725, "grad_norm": 7.59375, "learning_rate": 1.3976679937838737e-05, "loss": 1.2594, "step": 15549 }, { "epoch": 38.875, "grad_norm": 7.1875, "learning_rate": 1.3975937011513931e-05, "loss": 1.3917, "step": 15550 }, { "epoch": 38.8775, "grad_norm": 5.9375, "learning_rate": 1.3975194059123039e-05, "loss": 1.2926, "step": 15551 }, { "epoch": 38.88, "grad_norm": 9.25, "learning_rate": 1.3974451080670934e-05, "loss": 1.357, "step": 15552 }, { "epoch": 38.8825, "grad_norm": 6.25, "learning_rate": 1.3973708076162489e-05, "loss": 1.2304, "step": 15553 }, { "epoch": 38.885, "grad_norm": 5.6875, "learning_rate": 1.397296504560257e-05, "loss": 1.2339, "step": 15554 }, { "epoch": 38.8875, "grad_norm": 8.875, "learning_rate": 1.3972221988996053e-05, "loss": 1.4645, "step": 15555 }, { "epoch": 38.89, "grad_norm": 9.375, "learning_rate": 1.3971478906347806e-05, "loss": 1.248, "step": 15556 }, { "epoch": 38.8925, "grad_norm": 7.78125, "learning_rate": 1.3970735797662706e-05, "loss": 1.4818, "step": 15557 }, { "epoch": 38.895, "grad_norm": 9.1875, "learning_rate": 1.3969992662945616e-05, "loss": 1.3394, "step": 15558 }, { "epoch": 38.8975, "grad_norm": 10.125, "learning_rate": 1.3969249502201417e-05, "loss": 1.4205, "step": 15559 }, { "epoch": 38.9, "grad_norm": 5.78125, "learning_rate": 1.3968506315434973e-05, "loss": 1.3523, "step": 15560 }, { "epoch": 38.9025, "grad_norm": 4.90625, "learning_rate": 1.3967763102651165e-05, "loss": 1.3286, "step": 15561 }, { "epoch": 38.905, "grad_norm": 6.21875, "learning_rate": 1.3967019863854858e-05, "loss": 1.2129, "step": 15562 }, { "epoch": 38.9075, "grad_norm": 6.3125, "learning_rate": 1.396627659905093e-05, "loss": 1.4247, "step": 15563 }, { "epoch": 38.91, "grad_norm": 8.5, "learning_rate": 1.3965533308244248e-05, "loss": 1.3648, "step": 15564 }, { "epoch": 38.9125, "grad_norm": 9.125, "learning_rate": 1.3964789991439688e-05, "loss": 1.5093, "step": 15565 }, { "epoch": 38.915, "grad_norm": 11.75, "learning_rate": 1.3964046648642127e-05, "loss": 1.367, "step": 15566 }, { "epoch": 38.9175, "grad_norm": 6.09375, "learning_rate": 1.3963303279856435e-05, "loss": 1.255, "step": 15567 }, { "epoch": 38.92, "grad_norm": 7.0, "learning_rate": 1.3962559885087482e-05, "loss": 1.4768, "step": 15568 }, { "epoch": 38.9225, "grad_norm": 7.0, "learning_rate": 1.3961816464340148e-05, "loss": 1.3893, "step": 15569 }, { "epoch": 38.925, "grad_norm": 7.90625, "learning_rate": 1.39610730176193e-05, "loss": 1.2771, "step": 15570 }, { "epoch": 38.9275, "grad_norm": 7.53125, "learning_rate": 1.396032954492982e-05, "loss": 1.4682, "step": 15571 }, { "epoch": 38.93, "grad_norm": 8.125, "learning_rate": 1.3959586046276572e-05, "loss": 1.4952, "step": 15572 }, { "epoch": 38.9325, "grad_norm": 8.4375, "learning_rate": 1.3958842521664444e-05, "loss": 1.1633, "step": 15573 }, { "epoch": 38.935, "grad_norm": 10.375, "learning_rate": 1.3958098971098297e-05, "loss": 1.2847, "step": 15574 }, { "epoch": 38.9375, "grad_norm": 7.375, "learning_rate": 1.3957355394583014e-05, "loss": 1.3207, "step": 15575 }, { "epoch": 38.94, "grad_norm": 6.25, "learning_rate": 1.3956611792123469e-05, "loss": 1.2966, "step": 15576 }, { "epoch": 38.9425, "grad_norm": 8.0625, "learning_rate": 1.395586816372453e-05, "loss": 1.4284, "step": 15577 }, { "epoch": 38.945, "grad_norm": 7.46875, "learning_rate": 1.3955124509391083e-05, "loss": 1.2583, "step": 15578 }, { "epoch": 38.9475, "grad_norm": 5.65625, "learning_rate": 1.3954380829127996e-05, "loss": 1.3798, "step": 15579 }, { "epoch": 38.95, "grad_norm": 8.1875, "learning_rate": 1.3953637122940147e-05, "loss": 1.2568, "step": 15580 }, { "epoch": 38.9525, "grad_norm": 8.125, "learning_rate": 1.3952893390832412e-05, "loss": 1.4141, "step": 15581 }, { "epoch": 38.955, "grad_norm": 8.3125, "learning_rate": 1.3952149632809664e-05, "loss": 1.2638, "step": 15582 }, { "epoch": 38.9575, "grad_norm": 7.78125, "learning_rate": 1.395140584887678e-05, "loss": 1.4431, "step": 15583 }, { "epoch": 38.96, "grad_norm": 8.625, "learning_rate": 1.3950662039038643e-05, "loss": 1.3862, "step": 15584 }, { "epoch": 38.9625, "grad_norm": 6.75, "learning_rate": 1.3949918203300117e-05, "loss": 1.3641, "step": 15585 }, { "epoch": 38.965, "grad_norm": 10.5, "learning_rate": 1.3949174341666092e-05, "loss": 1.3059, "step": 15586 }, { "epoch": 38.9675, "grad_norm": 6.59375, "learning_rate": 1.394843045414143e-05, "loss": 1.2495, "step": 15587 }, { "epoch": 38.97, "grad_norm": 10.5, "learning_rate": 1.3947686540731024e-05, "loss": 1.3455, "step": 15588 }, { "epoch": 38.9725, "grad_norm": 5.46875, "learning_rate": 1.3946942601439739e-05, "loss": 1.3993, "step": 15589 }, { "epoch": 38.975, "grad_norm": 8.0, "learning_rate": 1.394619863627246e-05, "loss": 1.3562, "step": 15590 }, { "epoch": 38.9775, "grad_norm": 9.1875, "learning_rate": 1.3945454645234056e-05, "loss": 1.3804, "step": 15591 }, { "epoch": 38.98, "grad_norm": 6.46875, "learning_rate": 1.3944710628329409e-05, "loss": 1.4733, "step": 15592 }, { "epoch": 38.9825, "grad_norm": 7.03125, "learning_rate": 1.39439665855634e-05, "loss": 1.4424, "step": 15593 }, { "epoch": 38.985, "grad_norm": 6.53125, "learning_rate": 1.3943222516940902e-05, "loss": 1.272, "step": 15594 }, { "epoch": 38.9875, "grad_norm": 6.5, "learning_rate": 1.3942478422466797e-05, "loss": 1.138, "step": 15595 }, { "epoch": 38.99, "grad_norm": 6.625, "learning_rate": 1.394173430214596e-05, "loss": 1.3476, "step": 15596 }, { "epoch": 38.9925, "grad_norm": 8.875, "learning_rate": 1.3940990155983269e-05, "loss": 1.353, "step": 15597 }, { "epoch": 38.995, "grad_norm": 4.84375, "learning_rate": 1.3940245983983605e-05, "loss": 1.2643, "step": 15598 }, { "epoch": 38.9975, "grad_norm": 8.5, "learning_rate": 1.3939501786151848e-05, "loss": 1.452, "step": 15599 }, { "epoch": 39.0, "grad_norm": 7.59375, "learning_rate": 1.3938757562492873e-05, "loss": 1.4256, "step": 15600 }, { "epoch": 39.0025, "grad_norm": 8.375, "learning_rate": 1.3938013313011563e-05, "loss": 1.2232, "step": 15601 }, { "epoch": 39.005, "grad_norm": 6.3125, "learning_rate": 1.3937269037712793e-05, "loss": 1.4092, "step": 15602 }, { "epoch": 39.0075, "grad_norm": 6.53125, "learning_rate": 1.3936524736601447e-05, "loss": 1.1305, "step": 15603 }, { "epoch": 39.01, "grad_norm": 6.53125, "learning_rate": 1.39357804096824e-05, "loss": 1.3619, "step": 15604 }, { "epoch": 39.0125, "grad_norm": 9.3125, "learning_rate": 1.3935036056960536e-05, "loss": 1.3405, "step": 15605 }, { "epoch": 39.015, "grad_norm": 5.34375, "learning_rate": 1.3934291678440733e-05, "loss": 1.2529, "step": 15606 }, { "epoch": 39.0175, "grad_norm": 7.03125, "learning_rate": 1.393354727412787e-05, "loss": 1.4237, "step": 15607 }, { "epoch": 39.02, "grad_norm": 5.3125, "learning_rate": 1.393280284402683e-05, "loss": 1.4126, "step": 15608 }, { "epoch": 39.0225, "grad_norm": 108.5, "learning_rate": 1.3932058388142489e-05, "loss": 1.4125, "step": 15609 }, { "epoch": 39.025, "grad_norm": 5.40625, "learning_rate": 1.3931313906479733e-05, "loss": 1.2361, "step": 15610 }, { "epoch": 39.0275, "grad_norm": 9.0625, "learning_rate": 1.3930569399043439e-05, "loss": 1.3782, "step": 15611 }, { "epoch": 39.03, "grad_norm": 9.375, "learning_rate": 1.392982486583849e-05, "loss": 1.3658, "step": 15612 }, { "epoch": 39.0325, "grad_norm": 7.625, "learning_rate": 1.392908030686977e-05, "loss": 1.3195, "step": 15613 }, { "epoch": 39.035, "grad_norm": 7.15625, "learning_rate": 1.3928335722142152e-05, "loss": 1.2574, "step": 15614 }, { "epoch": 39.0375, "grad_norm": 6.625, "learning_rate": 1.3927591111660524e-05, "loss": 1.2102, "step": 15615 }, { "epoch": 39.04, "grad_norm": 7.03125, "learning_rate": 1.3926846475429767e-05, "loss": 1.1988, "step": 15616 }, { "epoch": 39.0425, "grad_norm": 8.8125, "learning_rate": 1.3926101813454757e-05, "loss": 1.2708, "step": 15617 }, { "epoch": 39.045, "grad_norm": 8.6875, "learning_rate": 1.3925357125740387e-05, "loss": 1.5237, "step": 15618 }, { "epoch": 39.0475, "grad_norm": 7.4375, "learning_rate": 1.3924612412291527e-05, "loss": 1.3507, "step": 15619 }, { "epoch": 39.05, "grad_norm": 9.0625, "learning_rate": 1.3923867673113067e-05, "loss": 1.4616, "step": 15620 }, { "epoch": 39.0525, "grad_norm": 6.78125, "learning_rate": 1.3923122908209889e-05, "loss": 1.3695, "step": 15621 }, { "epoch": 39.055, "grad_norm": 9.1875, "learning_rate": 1.3922378117586874e-05, "loss": 1.4429, "step": 15622 }, { "epoch": 39.0575, "grad_norm": 7.5, "learning_rate": 1.3921633301248905e-05, "loss": 1.3111, "step": 15623 }, { "epoch": 39.06, "grad_norm": 6.09375, "learning_rate": 1.3920888459200863e-05, "loss": 1.401, "step": 15624 }, { "epoch": 39.0625, "grad_norm": 6.65625, "learning_rate": 1.3920143591447635e-05, "loss": 1.3848, "step": 15625 }, { "epoch": 39.065, "grad_norm": 7.46875, "learning_rate": 1.3919398697994102e-05, "loss": 1.2914, "step": 15626 }, { "epoch": 39.0675, "grad_norm": 13.9375, "learning_rate": 1.3918653778845148e-05, "loss": 1.3629, "step": 15627 }, { "epoch": 39.07, "grad_norm": 6.0, "learning_rate": 1.3917908834005656e-05, "loss": 1.2771, "step": 15628 }, { "epoch": 39.0725, "grad_norm": 9.125, "learning_rate": 1.3917163863480514e-05, "loss": 1.2701, "step": 15629 }, { "epoch": 39.075, "grad_norm": 10.375, "learning_rate": 1.3916418867274597e-05, "loss": 1.4577, "step": 15630 }, { "epoch": 39.0775, "grad_norm": 8.0625, "learning_rate": 1.39156738453928e-05, "loss": 1.4383, "step": 15631 }, { "epoch": 39.08, "grad_norm": 8.875, "learning_rate": 1.3914928797839996e-05, "loss": 1.3501, "step": 15632 }, { "epoch": 39.0825, "grad_norm": 6.875, "learning_rate": 1.391418372462108e-05, "loss": 1.4794, "step": 15633 }, { "epoch": 39.085, "grad_norm": 8.5625, "learning_rate": 1.3913438625740931e-05, "loss": 1.4195, "step": 15634 }, { "epoch": 39.0875, "grad_norm": 8.75, "learning_rate": 1.3912693501204435e-05, "loss": 1.3462, "step": 15635 }, { "epoch": 39.09, "grad_norm": 8.0625, "learning_rate": 1.3911948351016475e-05, "loss": 1.2522, "step": 15636 }, { "epoch": 39.0925, "grad_norm": 7.53125, "learning_rate": 1.3911203175181943e-05, "loss": 1.5392, "step": 15637 }, { "epoch": 39.095, "grad_norm": 8.6875, "learning_rate": 1.3910457973705716e-05, "loss": 1.3444, "step": 15638 }, { "epoch": 39.0975, "grad_norm": 7.84375, "learning_rate": 1.3909712746592683e-05, "loss": 1.4097, "step": 15639 }, { "epoch": 39.1, "grad_norm": 9.125, "learning_rate": 1.390896749384773e-05, "loss": 1.1943, "step": 15640 }, { "epoch": 39.1025, "grad_norm": 7.84375, "learning_rate": 1.3908222215475744e-05, "loss": 1.2791, "step": 15641 }, { "epoch": 39.105, "grad_norm": 6.21875, "learning_rate": 1.3907476911481605e-05, "loss": 1.4119, "step": 15642 }, { "epoch": 39.1075, "grad_norm": 7.34375, "learning_rate": 1.390673158187021e-05, "loss": 1.2594, "step": 15643 }, { "epoch": 39.11, "grad_norm": 6.84375, "learning_rate": 1.3905986226646435e-05, "loss": 1.3136, "step": 15644 }, { "epoch": 39.1125, "grad_norm": 10.125, "learning_rate": 1.3905240845815171e-05, "loss": 1.5474, "step": 15645 }, { "epoch": 39.115, "grad_norm": 9.875, "learning_rate": 1.3904495439381304e-05, "loss": 1.3146, "step": 15646 }, { "epoch": 39.1175, "grad_norm": 9.375, "learning_rate": 1.390375000734972e-05, "loss": 1.4109, "step": 15647 }, { "epoch": 39.12, "grad_norm": 8.3125, "learning_rate": 1.3903004549725313e-05, "loss": 1.3692, "step": 15648 }, { "epoch": 39.1225, "grad_norm": 6.53125, "learning_rate": 1.390225906651296e-05, "loss": 1.2809, "step": 15649 }, { "epoch": 39.125, "grad_norm": 7.1875, "learning_rate": 1.3901513557717554e-05, "loss": 1.3461, "step": 15650 }, { "epoch": 39.1275, "grad_norm": 6.84375, "learning_rate": 1.390076802334398e-05, "loss": 1.173, "step": 15651 }, { "epoch": 39.13, "grad_norm": 6.90625, "learning_rate": 1.3900022463397127e-05, "loss": 1.2718, "step": 15652 }, { "epoch": 39.1325, "grad_norm": 6.0, "learning_rate": 1.3899276877881884e-05, "loss": 1.3365, "step": 15653 }, { "epoch": 39.135, "grad_norm": 6.75, "learning_rate": 1.389853126680314e-05, "loss": 1.4273, "step": 15654 }, { "epoch": 39.1375, "grad_norm": 9.4375, "learning_rate": 1.3897785630165774e-05, "loss": 1.2017, "step": 15655 }, { "epoch": 39.14, "grad_norm": 6.96875, "learning_rate": 1.3897039967974688e-05, "loss": 1.1944, "step": 15656 }, { "epoch": 39.1425, "grad_norm": 11.0625, "learning_rate": 1.3896294280234762e-05, "loss": 1.4215, "step": 15657 }, { "epoch": 39.145, "grad_norm": 6.34375, "learning_rate": 1.3895548566950887e-05, "loss": 1.258, "step": 15658 }, { "epoch": 39.1475, "grad_norm": 14.5625, "learning_rate": 1.389480282812795e-05, "loss": 1.4343, "step": 15659 }, { "epoch": 39.15, "grad_norm": 8.375, "learning_rate": 1.3894057063770841e-05, "loss": 1.4484, "step": 15660 }, { "epoch": 39.1525, "grad_norm": 14.125, "learning_rate": 1.389331127388445e-05, "loss": 1.4828, "step": 15661 }, { "epoch": 39.155, "grad_norm": 7.96875, "learning_rate": 1.3892565458473668e-05, "loss": 1.4886, "step": 15662 }, { "epoch": 39.1575, "grad_norm": 10.25, "learning_rate": 1.3891819617543384e-05, "loss": 1.2972, "step": 15663 }, { "epoch": 39.16, "grad_norm": 7.34375, "learning_rate": 1.3891073751098481e-05, "loss": 1.3777, "step": 15664 }, { "epoch": 39.1625, "grad_norm": 6.09375, "learning_rate": 1.3890327859143857e-05, "loss": 1.2323, "step": 15665 }, { "epoch": 39.165, "grad_norm": 5.84375, "learning_rate": 1.38895819416844e-05, "loss": 1.1329, "step": 15666 }, { "epoch": 39.1675, "grad_norm": 7.03125, "learning_rate": 1.3888835998724999e-05, "loss": 1.305, "step": 15667 }, { "epoch": 39.17, "grad_norm": 5.1875, "learning_rate": 1.3888090030270545e-05, "loss": 1.3381, "step": 15668 }, { "epoch": 39.1725, "grad_norm": 8.1875, "learning_rate": 1.388734403632593e-05, "loss": 1.3966, "step": 15669 }, { "epoch": 39.175, "grad_norm": 7.6875, "learning_rate": 1.3886598016896039e-05, "loss": 1.3202, "step": 15670 }, { "epoch": 39.1775, "grad_norm": 6.4375, "learning_rate": 1.3885851971985768e-05, "loss": 1.1664, "step": 15671 }, { "epoch": 39.18, "grad_norm": 5.125, "learning_rate": 1.3885105901600006e-05, "loss": 1.2704, "step": 15672 }, { "epoch": 39.1825, "grad_norm": 4.875, "learning_rate": 1.388435980574365e-05, "loss": 1.3932, "step": 15673 }, { "epoch": 39.185, "grad_norm": 6.21875, "learning_rate": 1.3883613684421584e-05, "loss": 1.2025, "step": 15674 }, { "epoch": 39.1875, "grad_norm": 8.4375, "learning_rate": 1.38828675376387e-05, "loss": 1.2551, "step": 15675 }, { "epoch": 39.19, "grad_norm": 9.5, "learning_rate": 1.3882121365399894e-05, "loss": 1.2848, "step": 15676 }, { "epoch": 39.1925, "grad_norm": 7.0625, "learning_rate": 1.3881375167710056e-05, "loss": 1.3242, "step": 15677 }, { "epoch": 39.195, "grad_norm": 5.0625, "learning_rate": 1.3880628944574077e-05, "loss": 1.402, "step": 15678 }, { "epoch": 39.1975, "grad_norm": 6.625, "learning_rate": 1.3879882695996849e-05, "loss": 1.2768, "step": 15679 }, { "epoch": 39.2, "grad_norm": 8.25, "learning_rate": 1.3879136421983265e-05, "loss": 1.2817, "step": 15680 }, { "epoch": 39.2025, "grad_norm": 6.71875, "learning_rate": 1.387839012253822e-05, "loss": 1.3383, "step": 15681 }, { "epoch": 39.205, "grad_norm": 6.9375, "learning_rate": 1.3877643797666605e-05, "loss": 1.2513, "step": 15682 }, { "epoch": 39.2075, "grad_norm": 8.9375, "learning_rate": 1.387689744737331e-05, "loss": 1.5567, "step": 15683 }, { "epoch": 39.21, "grad_norm": 10.9375, "learning_rate": 1.3876151071663233e-05, "loss": 1.307, "step": 15684 }, { "epoch": 39.2125, "grad_norm": 6.9375, "learning_rate": 1.3875404670541261e-05, "loss": 1.3986, "step": 15685 }, { "epoch": 39.215, "grad_norm": 10.1875, "learning_rate": 1.3874658244012296e-05, "loss": 1.394, "step": 15686 }, { "epoch": 39.2175, "grad_norm": 9.0625, "learning_rate": 1.3873911792081224e-05, "loss": 1.1638, "step": 15687 }, { "epoch": 39.22, "grad_norm": 6.1875, "learning_rate": 1.3873165314752941e-05, "loss": 1.4076, "step": 15688 }, { "epoch": 39.2225, "grad_norm": 8.1875, "learning_rate": 1.387241881203234e-05, "loss": 1.403, "step": 15689 }, { "epoch": 39.225, "grad_norm": 9.4375, "learning_rate": 1.3871672283924317e-05, "loss": 1.1867, "step": 15690 }, { "epoch": 39.2275, "grad_norm": 10.75, "learning_rate": 1.387092573043377e-05, "loss": 1.3107, "step": 15691 }, { "epoch": 39.23, "grad_norm": 6.1875, "learning_rate": 1.3870179151565581e-05, "loss": 1.1887, "step": 15692 }, { "epoch": 39.2325, "grad_norm": 8.1875, "learning_rate": 1.3869432547324658e-05, "loss": 1.2847, "step": 15693 }, { "epoch": 39.235, "grad_norm": 5.34375, "learning_rate": 1.3868685917715887e-05, "loss": 1.2882, "step": 15694 }, { "epoch": 39.2375, "grad_norm": 9.25, "learning_rate": 1.3867939262744169e-05, "loss": 1.3848, "step": 15695 }, { "epoch": 39.24, "grad_norm": 9.875, "learning_rate": 1.3867192582414393e-05, "loss": 1.3889, "step": 15696 }, { "epoch": 39.2425, "grad_norm": 8.8125, "learning_rate": 1.3866445876731459e-05, "loss": 1.2977, "step": 15697 }, { "epoch": 39.245, "grad_norm": 6.09375, "learning_rate": 1.3865699145700257e-05, "loss": 1.4425, "step": 15698 }, { "epoch": 39.2475, "grad_norm": 6.6875, "learning_rate": 1.386495238932569e-05, "loss": 1.3315, "step": 15699 }, { "epoch": 39.25, "grad_norm": 5.6875, "learning_rate": 1.3864205607612648e-05, "loss": 1.3836, "step": 15700 }, { "epoch": 39.2525, "grad_norm": 6.53125, "learning_rate": 1.3863458800566031e-05, "loss": 1.3286, "step": 15701 }, { "epoch": 39.255, "grad_norm": 5.125, "learning_rate": 1.3862711968190731e-05, "loss": 1.2347, "step": 15702 }, { "epoch": 39.2575, "grad_norm": 8.8125, "learning_rate": 1.3861965110491648e-05, "loss": 1.3319, "step": 15703 }, { "epoch": 39.26, "grad_norm": 6.40625, "learning_rate": 1.3861218227473674e-05, "loss": 1.4253, "step": 15704 }, { "epoch": 39.2625, "grad_norm": 8.0, "learning_rate": 1.3860471319141709e-05, "loss": 1.2871, "step": 15705 }, { "epoch": 39.265, "grad_norm": 7.5625, "learning_rate": 1.3859724385500648e-05, "loss": 1.3048, "step": 15706 }, { "epoch": 39.2675, "grad_norm": 7.96875, "learning_rate": 1.3858977426555388e-05, "loss": 1.4464, "step": 15707 }, { "epoch": 39.27, "grad_norm": 8.0625, "learning_rate": 1.3858230442310828e-05, "loss": 1.5222, "step": 15708 }, { "epoch": 39.2725, "grad_norm": 7.5, "learning_rate": 1.3857483432771861e-05, "loss": 1.3376, "step": 15709 }, { "epoch": 39.275, "grad_norm": 8.9375, "learning_rate": 1.3856736397943386e-05, "loss": 1.3935, "step": 15710 }, { "epoch": 39.2775, "grad_norm": 6.375, "learning_rate": 1.3855989337830306e-05, "loss": 1.4609, "step": 15711 }, { "epoch": 39.28, "grad_norm": 9.3125, "learning_rate": 1.3855242252437511e-05, "loss": 1.563, "step": 15712 }, { "epoch": 39.2825, "grad_norm": 5.8125, "learning_rate": 1.3854495141769904e-05, "loss": 1.4087, "step": 15713 }, { "epoch": 39.285, "grad_norm": 5.375, "learning_rate": 1.385374800583238e-05, "loss": 1.2989, "step": 15714 }, { "epoch": 39.2875, "grad_norm": 5.59375, "learning_rate": 1.3853000844629838e-05, "loss": 1.3502, "step": 15715 }, { "epoch": 39.29, "grad_norm": 7.25, "learning_rate": 1.3852253658167178e-05, "loss": 1.5029, "step": 15716 }, { "epoch": 39.2925, "grad_norm": 9.125, "learning_rate": 1.3851506446449294e-05, "loss": 1.4144, "step": 15717 }, { "epoch": 39.295, "grad_norm": 7.75, "learning_rate": 1.3850759209481093e-05, "loss": 1.4113, "step": 15718 }, { "epoch": 39.2975, "grad_norm": 7.6875, "learning_rate": 1.3850011947267463e-05, "loss": 1.3282, "step": 15719 }, { "epoch": 39.3, "grad_norm": 21.75, "learning_rate": 1.3849264659813314e-05, "loss": 1.4661, "step": 15720 }, { "epoch": 39.3025, "grad_norm": 8.25, "learning_rate": 1.3848517347123536e-05, "loss": 1.2472, "step": 15721 }, { "epoch": 39.305, "grad_norm": 10.0625, "learning_rate": 1.3847770009203034e-05, "loss": 1.3112, "step": 15722 }, { "epoch": 39.3075, "grad_norm": 7.09375, "learning_rate": 1.3847022646056706e-05, "loss": 1.3502, "step": 15723 }, { "epoch": 39.31, "grad_norm": 9.4375, "learning_rate": 1.384627525768945e-05, "loss": 1.4826, "step": 15724 }, { "epoch": 39.3125, "grad_norm": 7.0, "learning_rate": 1.3845527844106168e-05, "loss": 1.4296, "step": 15725 }, { "epoch": 39.315, "grad_norm": 9.8125, "learning_rate": 1.3844780405311762e-05, "loss": 1.3018, "step": 15726 }, { "epoch": 39.3175, "grad_norm": 11.0625, "learning_rate": 1.3844032941311125e-05, "loss": 1.486, "step": 15727 }, { "epoch": 39.32, "grad_norm": 8.0, "learning_rate": 1.3843285452109166e-05, "loss": 1.235, "step": 15728 }, { "epoch": 39.3225, "grad_norm": 5.65625, "learning_rate": 1.3842537937710778e-05, "loss": 1.2281, "step": 15729 }, { "epoch": 39.325, "grad_norm": 15.4375, "learning_rate": 1.3841790398120869e-05, "loss": 1.4741, "step": 15730 }, { "epoch": 39.3275, "grad_norm": 6.65625, "learning_rate": 1.384104283334433e-05, "loss": 1.3383, "step": 15731 }, { "epoch": 39.33, "grad_norm": 6.40625, "learning_rate": 1.3840295243386073e-05, "loss": 1.6041, "step": 15732 }, { "epoch": 39.3325, "grad_norm": 6.75, "learning_rate": 1.3839547628250993e-05, "loss": 1.3048, "step": 15733 }, { "epoch": 39.335, "grad_norm": 8.0625, "learning_rate": 1.3838799987943992e-05, "loss": 1.2786, "step": 15734 }, { "epoch": 39.3375, "grad_norm": 5.84375, "learning_rate": 1.383805232246997e-05, "loss": 1.3942, "step": 15735 }, { "epoch": 39.34, "grad_norm": 9.375, "learning_rate": 1.3837304631833832e-05, "loss": 1.2989, "step": 15736 }, { "epoch": 39.3425, "grad_norm": 8.375, "learning_rate": 1.383655691604048e-05, "loss": 1.4698, "step": 15737 }, { "epoch": 39.345, "grad_norm": 7.15625, "learning_rate": 1.3835809175094812e-05, "loss": 1.3925, "step": 15738 }, { "epoch": 39.3475, "grad_norm": 7.84375, "learning_rate": 1.3835061409001735e-05, "loss": 1.4007, "step": 15739 }, { "epoch": 39.35, "grad_norm": 6.4375, "learning_rate": 1.3834313617766146e-05, "loss": 1.2525, "step": 15740 }, { "epoch": 39.3525, "grad_norm": 7.59375, "learning_rate": 1.3833565801392954e-05, "loss": 1.5008, "step": 15741 }, { "epoch": 39.355, "grad_norm": 6.03125, "learning_rate": 1.3832817959887056e-05, "loss": 1.3361, "step": 15742 }, { "epoch": 39.3575, "grad_norm": 8.4375, "learning_rate": 1.3832070093253358e-05, "loss": 1.3209, "step": 15743 }, { "epoch": 39.36, "grad_norm": 9.125, "learning_rate": 1.3831322201496757e-05, "loss": 1.4612, "step": 15744 }, { "epoch": 39.3625, "grad_norm": 6.65625, "learning_rate": 1.3830574284622167e-05, "loss": 1.4245, "step": 15745 }, { "epoch": 39.365, "grad_norm": 9.1875, "learning_rate": 1.3829826342634482e-05, "loss": 1.2401, "step": 15746 }, { "epoch": 39.3675, "grad_norm": 8.25, "learning_rate": 1.3829078375538609e-05, "loss": 1.37, "step": 15747 }, { "epoch": 39.37, "grad_norm": 8.4375, "learning_rate": 1.3828330383339453e-05, "loss": 1.2902, "step": 15748 }, { "epoch": 39.3725, "grad_norm": 6.40625, "learning_rate": 1.3827582366041914e-05, "loss": 1.3485, "step": 15749 }, { "epoch": 39.375, "grad_norm": 9.875, "learning_rate": 1.3826834323650899e-05, "loss": 1.4434, "step": 15750 }, { "epoch": 39.3775, "grad_norm": 6.28125, "learning_rate": 1.3826086256171311e-05, "loss": 1.267, "step": 15751 }, { "epoch": 39.38, "grad_norm": 7.90625, "learning_rate": 1.3825338163608055e-05, "loss": 1.3941, "step": 15752 }, { "epoch": 39.3825, "grad_norm": 9.75, "learning_rate": 1.3824590045966036e-05, "loss": 1.4134, "step": 15753 }, { "epoch": 39.385, "grad_norm": 9.125, "learning_rate": 1.3823841903250157e-05, "loss": 1.3722, "step": 15754 }, { "epoch": 39.3875, "grad_norm": 7.21875, "learning_rate": 1.3823093735465322e-05, "loss": 1.359, "step": 15755 }, { "epoch": 39.39, "grad_norm": 7.90625, "learning_rate": 1.3822345542616443e-05, "loss": 1.4076, "step": 15756 }, { "epoch": 39.3925, "grad_norm": 5.1875, "learning_rate": 1.3821597324708412e-05, "loss": 1.2199, "step": 15757 }, { "epoch": 39.395, "grad_norm": 5.34375, "learning_rate": 1.3820849081746147e-05, "loss": 1.2135, "step": 15758 }, { "epoch": 39.3975, "grad_norm": 8.6875, "learning_rate": 1.3820100813734546e-05, "loss": 1.3908, "step": 15759 }, { "epoch": 39.4, "grad_norm": 10.25, "learning_rate": 1.3819352520678519e-05, "loss": 1.3464, "step": 15760 }, { "epoch": 39.4025, "grad_norm": 7.09375, "learning_rate": 1.3818604202582969e-05, "loss": 1.2696, "step": 15761 }, { "epoch": 39.405, "grad_norm": 7.15625, "learning_rate": 1.3817855859452802e-05, "loss": 1.4566, "step": 15762 }, { "epoch": 39.4075, "grad_norm": 6.6875, "learning_rate": 1.3817107491292926e-05, "loss": 1.2588, "step": 15763 }, { "epoch": 39.41, "grad_norm": 6.6875, "learning_rate": 1.3816359098108243e-05, "loss": 1.327, "step": 15764 }, { "epoch": 39.4125, "grad_norm": 11.9375, "learning_rate": 1.3815610679903665e-05, "loss": 1.3621, "step": 15765 }, { "epoch": 39.415, "grad_norm": 5.71875, "learning_rate": 1.3814862236684094e-05, "loss": 1.442, "step": 15766 }, { "epoch": 39.4175, "grad_norm": 7.65625, "learning_rate": 1.381411376845444e-05, "loss": 1.2286, "step": 15767 }, { "epoch": 39.42, "grad_norm": 6.65625, "learning_rate": 1.381336527521961e-05, "loss": 1.3693, "step": 15768 }, { "epoch": 39.4225, "grad_norm": 5.40625, "learning_rate": 1.381261675698451e-05, "loss": 1.344, "step": 15769 }, { "epoch": 39.425, "grad_norm": 7.84375, "learning_rate": 1.3811868213754043e-05, "loss": 1.3418, "step": 15770 }, { "epoch": 39.4275, "grad_norm": 6.75, "learning_rate": 1.3811119645533126e-05, "loss": 1.3718, "step": 15771 }, { "epoch": 39.43, "grad_norm": 8.125, "learning_rate": 1.3810371052326656e-05, "loss": 1.4267, "step": 15772 }, { "epoch": 39.4325, "grad_norm": 5.65625, "learning_rate": 1.3809622434139552e-05, "loss": 1.2276, "step": 15773 }, { "epoch": 39.435, "grad_norm": 13.5, "learning_rate": 1.3808873790976709e-05, "loss": 1.3325, "step": 15774 }, { "epoch": 39.4375, "grad_norm": 5.375, "learning_rate": 1.3808125122843048e-05, "loss": 1.3082, "step": 15775 }, { "epoch": 39.44, "grad_norm": 7.90625, "learning_rate": 1.3807376429743467e-05, "loss": 1.3595, "step": 15776 }, { "epoch": 39.4425, "grad_norm": 7.3125, "learning_rate": 1.380662771168288e-05, "loss": 1.4395, "step": 15777 }, { "epoch": 39.445, "grad_norm": 7.625, "learning_rate": 1.3805878968666193e-05, "loss": 1.3967, "step": 15778 }, { "epoch": 39.4475, "grad_norm": 8.3125, "learning_rate": 1.3805130200698317e-05, "loss": 1.4118, "step": 15779 }, { "epoch": 39.45, "grad_norm": 6.34375, "learning_rate": 1.380438140778416e-05, "loss": 1.1146, "step": 15780 }, { "epoch": 39.4525, "grad_norm": 6.625, "learning_rate": 1.380363258992863e-05, "loss": 1.3791, "step": 15781 }, { "epoch": 39.455, "grad_norm": 10.4375, "learning_rate": 1.3802883747136636e-05, "loss": 1.508, "step": 15782 }, { "epoch": 39.4575, "grad_norm": 9.4375, "learning_rate": 1.3802134879413092e-05, "loss": 1.2057, "step": 15783 }, { "epoch": 39.46, "grad_norm": 7.65625, "learning_rate": 1.38013859867629e-05, "loss": 1.3202, "step": 15784 }, { "epoch": 39.4625, "grad_norm": 7.375, "learning_rate": 1.3800637069190971e-05, "loss": 1.3798, "step": 15785 }, { "epoch": 39.465, "grad_norm": 8.6875, "learning_rate": 1.3799888126702223e-05, "loss": 1.2586, "step": 15786 }, { "epoch": 39.4675, "grad_norm": 8.3125, "learning_rate": 1.3799139159301557e-05, "loss": 1.3372, "step": 15787 }, { "epoch": 39.47, "grad_norm": 10.0625, "learning_rate": 1.3798390166993892e-05, "loss": 1.2769, "step": 15788 }, { "epoch": 39.4725, "grad_norm": 7.15625, "learning_rate": 1.3797641149784126e-05, "loss": 1.3115, "step": 15789 }, { "epoch": 39.475, "grad_norm": 7.1875, "learning_rate": 1.3796892107677183e-05, "loss": 1.488, "step": 15790 }, { "epoch": 39.4775, "grad_norm": 7.25, "learning_rate": 1.3796143040677965e-05, "loss": 1.3334, "step": 15791 }, { "epoch": 39.48, "grad_norm": 5.40625, "learning_rate": 1.3795393948791382e-05, "loss": 1.4683, "step": 15792 }, { "epoch": 39.4825, "grad_norm": 11.375, "learning_rate": 1.3794644832022353e-05, "loss": 1.3188, "step": 15793 }, { "epoch": 39.485, "grad_norm": 7.03125, "learning_rate": 1.379389569037578e-05, "loss": 1.5306, "step": 15794 }, { "epoch": 39.4875, "grad_norm": 7.21875, "learning_rate": 1.3793146523856582e-05, "loss": 1.2713, "step": 15795 }, { "epoch": 39.49, "grad_norm": 6.5625, "learning_rate": 1.3792397332469667e-05, "loss": 1.3443, "step": 15796 }, { "epoch": 39.4925, "grad_norm": 7.625, "learning_rate": 1.3791648116219944e-05, "loss": 1.3963, "step": 15797 }, { "epoch": 39.495, "grad_norm": 6.78125, "learning_rate": 1.379089887511233e-05, "loss": 1.1881, "step": 15798 }, { "epoch": 39.4975, "grad_norm": 9.3125, "learning_rate": 1.3790149609151733e-05, "loss": 1.3625, "step": 15799 }, { "epoch": 39.5, "grad_norm": 9.625, "learning_rate": 1.378940031834307e-05, "loss": 1.3719, "step": 15800 }, { "epoch": 39.5025, "grad_norm": 6.65625, "learning_rate": 1.3788651002691247e-05, "loss": 1.5423, "step": 15801 }, { "epoch": 39.505, "grad_norm": 8.6875, "learning_rate": 1.3787901662201181e-05, "loss": 1.298, "step": 15802 }, { "epoch": 39.5075, "grad_norm": 11.625, "learning_rate": 1.3787152296877783e-05, "loss": 1.3529, "step": 15803 }, { "epoch": 39.51, "grad_norm": 7.21875, "learning_rate": 1.3786402906725966e-05, "loss": 1.3413, "step": 15804 }, { "epoch": 39.5125, "grad_norm": 5.25, "learning_rate": 1.3785653491750643e-05, "loss": 1.3208, "step": 15805 }, { "epoch": 39.515, "grad_norm": 9.8125, "learning_rate": 1.3784904051956727e-05, "loss": 1.1916, "step": 15806 }, { "epoch": 39.5175, "grad_norm": 6.78125, "learning_rate": 1.3784154587349132e-05, "loss": 1.3603, "step": 15807 }, { "epoch": 39.52, "grad_norm": 11.3125, "learning_rate": 1.3783405097932772e-05, "loss": 1.2663, "step": 15808 }, { "epoch": 39.5225, "grad_norm": 8.5625, "learning_rate": 1.3782655583712557e-05, "loss": 1.2674, "step": 15809 }, { "epoch": 39.525, "grad_norm": 7.125, "learning_rate": 1.3781906044693404e-05, "loss": 1.4348, "step": 15810 }, { "epoch": 39.5275, "grad_norm": 7.875, "learning_rate": 1.3781156480880227e-05, "loss": 1.3338, "step": 15811 }, { "epoch": 39.53, "grad_norm": 6.96875, "learning_rate": 1.378040689227794e-05, "loss": 1.4259, "step": 15812 }, { "epoch": 39.5325, "grad_norm": 8.875, "learning_rate": 1.3779657278891456e-05, "loss": 1.368, "step": 15813 }, { "epoch": 39.535, "grad_norm": 7.78125, "learning_rate": 1.377890764072569e-05, "loss": 1.4117, "step": 15814 }, { "epoch": 39.5375, "grad_norm": 8.8125, "learning_rate": 1.3778157977785558e-05, "loss": 1.3629, "step": 15815 }, { "epoch": 39.54, "grad_norm": 6.8125, "learning_rate": 1.377740829007597e-05, "loss": 1.3227, "step": 15816 }, { "epoch": 39.5425, "grad_norm": 6.28125, "learning_rate": 1.377665857760185e-05, "loss": 1.4933, "step": 15817 }, { "epoch": 39.545, "grad_norm": 8.8125, "learning_rate": 1.3775908840368104e-05, "loss": 1.321, "step": 15818 }, { "epoch": 39.5475, "grad_norm": 8.625, "learning_rate": 1.3775159078379651e-05, "loss": 1.3987, "step": 15819 }, { "epoch": 39.55, "grad_norm": 8.375, "learning_rate": 1.3774409291641407e-05, "loss": 1.2893, "step": 15820 }, { "epoch": 39.5525, "grad_norm": 5.5625, "learning_rate": 1.3773659480158286e-05, "loss": 1.2607, "step": 15821 }, { "epoch": 39.555, "grad_norm": 6.65625, "learning_rate": 1.3772909643935203e-05, "loss": 1.2355, "step": 15822 }, { "epoch": 39.5575, "grad_norm": 6.5625, "learning_rate": 1.3772159782977076e-05, "loss": 1.3294, "step": 15823 }, { "epoch": 39.56, "grad_norm": 8.25, "learning_rate": 1.3771409897288823e-05, "loss": 1.2852, "step": 15824 }, { "epoch": 39.5625, "grad_norm": 6.875, "learning_rate": 1.3770659986875354e-05, "loss": 1.5375, "step": 15825 }, { "epoch": 39.565, "grad_norm": 4.46875, "learning_rate": 1.3769910051741593e-05, "loss": 1.3561, "step": 15826 }, { "epoch": 39.5675, "grad_norm": 7.4375, "learning_rate": 1.376916009189245e-05, "loss": 1.2591, "step": 15827 }, { "epoch": 39.57, "grad_norm": 7.03125, "learning_rate": 1.3768410107332846e-05, "loss": 1.416, "step": 15828 }, { "epoch": 39.5725, "grad_norm": 6.46875, "learning_rate": 1.3767660098067692e-05, "loss": 1.2574, "step": 15829 }, { "epoch": 39.575, "grad_norm": 7.125, "learning_rate": 1.3766910064101915e-05, "loss": 1.3058, "step": 15830 }, { "epoch": 39.5775, "grad_norm": 7.15625, "learning_rate": 1.376616000544042e-05, "loss": 1.3209, "step": 15831 }, { "epoch": 39.58, "grad_norm": 8.9375, "learning_rate": 1.3765409922088137e-05, "loss": 1.4291, "step": 15832 }, { "epoch": 39.5825, "grad_norm": 9.4375, "learning_rate": 1.3764659814049974e-05, "loss": 1.3685, "step": 15833 }, { "epoch": 39.585, "grad_norm": 9.0, "learning_rate": 1.3763909681330853e-05, "loss": 1.4227, "step": 15834 }, { "epoch": 39.5875, "grad_norm": 8.4375, "learning_rate": 1.3763159523935689e-05, "loss": 1.3256, "step": 15835 }, { "epoch": 39.59, "grad_norm": 6.96875, "learning_rate": 1.3762409341869404e-05, "loss": 1.4758, "step": 15836 }, { "epoch": 39.5925, "grad_norm": 7.25, "learning_rate": 1.3761659135136914e-05, "loss": 1.4825, "step": 15837 }, { "epoch": 39.595, "grad_norm": 6.1875, "learning_rate": 1.3760908903743136e-05, "loss": 1.2922, "step": 15838 }, { "epoch": 39.5975, "grad_norm": 5.9375, "learning_rate": 1.3760158647692987e-05, "loss": 1.2905, "step": 15839 }, { "epoch": 39.6, "grad_norm": 8.1875, "learning_rate": 1.3759408366991391e-05, "loss": 1.3256, "step": 15840 }, { "epoch": 39.6025, "grad_norm": 7.71875, "learning_rate": 1.3758658061643264e-05, "loss": 1.2804, "step": 15841 }, { "epoch": 39.605, "grad_norm": 6.9375, "learning_rate": 1.3757907731653526e-05, "loss": 1.4303, "step": 15842 }, { "epoch": 39.6075, "grad_norm": 5.03125, "learning_rate": 1.3757157377027097e-05, "loss": 1.31, "step": 15843 }, { "epoch": 39.61, "grad_norm": 8.625, "learning_rate": 1.375640699776889e-05, "loss": 1.3037, "step": 15844 }, { "epoch": 39.6125, "grad_norm": 14.125, "learning_rate": 1.3755656593883833e-05, "loss": 1.533, "step": 15845 }, { "epoch": 39.615, "grad_norm": 5.5625, "learning_rate": 1.3754906165376838e-05, "loss": 1.3347, "step": 15846 }, { "epoch": 39.6175, "grad_norm": 5.09375, "learning_rate": 1.3754155712252832e-05, "loss": 1.2114, "step": 15847 }, { "epoch": 39.62, "grad_norm": 6.78125, "learning_rate": 1.375340523451673e-05, "loss": 1.3382, "step": 15848 }, { "epoch": 39.6225, "grad_norm": 8.0625, "learning_rate": 1.3752654732173454e-05, "loss": 1.5563, "step": 15849 }, { "epoch": 39.625, "grad_norm": 7.125, "learning_rate": 1.3751904205227922e-05, "loss": 1.328, "step": 15850 }, { "epoch": 39.6275, "grad_norm": 5.90625, "learning_rate": 1.3751153653685059e-05, "loss": 1.2309, "step": 15851 }, { "epoch": 39.63, "grad_norm": 7.59375, "learning_rate": 1.375040307754978e-05, "loss": 1.3419, "step": 15852 }, { "epoch": 39.6325, "grad_norm": 8.875, "learning_rate": 1.3749652476827011e-05, "loss": 1.2735, "step": 15853 }, { "epoch": 39.635, "grad_norm": 8.75, "learning_rate": 1.3748901851521667e-05, "loss": 1.4166, "step": 15854 }, { "epoch": 39.6375, "grad_norm": 7.3125, "learning_rate": 1.3748151201638672e-05, "loss": 1.296, "step": 15855 }, { "epoch": 39.64, "grad_norm": 9.0625, "learning_rate": 1.3747400527182952e-05, "loss": 1.4127, "step": 15856 }, { "epoch": 39.6425, "grad_norm": 8.4375, "learning_rate": 1.3746649828159422e-05, "loss": 1.3521, "step": 15857 }, { "epoch": 39.645, "grad_norm": 8.6875, "learning_rate": 1.3745899104573008e-05, "loss": 1.1894, "step": 15858 }, { "epoch": 39.6475, "grad_norm": 9.1875, "learning_rate": 1.3745148356428626e-05, "loss": 1.3246, "step": 15859 }, { "epoch": 39.65, "grad_norm": 6.8125, "learning_rate": 1.3744397583731204e-05, "loss": 1.2862, "step": 15860 }, { "epoch": 39.6525, "grad_norm": 7.09375, "learning_rate": 1.3743646786485657e-05, "loss": 1.2369, "step": 15861 }, { "epoch": 39.655, "grad_norm": 11.6875, "learning_rate": 1.3742895964696916e-05, "loss": 1.466, "step": 15862 }, { "epoch": 39.6575, "grad_norm": 6.96875, "learning_rate": 1.3742145118369898e-05, "loss": 1.4846, "step": 15863 }, { "epoch": 39.66, "grad_norm": 7.90625, "learning_rate": 1.3741394247509524e-05, "loss": 1.3262, "step": 15864 }, { "epoch": 39.6625, "grad_norm": 6.65625, "learning_rate": 1.3740643352120722e-05, "loss": 1.2167, "step": 15865 }, { "epoch": 39.665, "grad_norm": 8.9375, "learning_rate": 1.373989243220841e-05, "loss": 1.3624, "step": 15866 }, { "epoch": 39.6675, "grad_norm": 7.21875, "learning_rate": 1.3739141487777512e-05, "loss": 1.2354, "step": 15867 }, { "epoch": 39.67, "grad_norm": 7.5, "learning_rate": 1.3738390518832955e-05, "loss": 1.3379, "step": 15868 }, { "epoch": 39.6725, "grad_norm": 6.90625, "learning_rate": 1.3737639525379658e-05, "loss": 1.3436, "step": 15869 }, { "epoch": 39.675, "grad_norm": 7.4375, "learning_rate": 1.3736888507422545e-05, "loss": 1.3918, "step": 15870 }, { "epoch": 39.6775, "grad_norm": 5.84375, "learning_rate": 1.3736137464966541e-05, "loss": 1.2545, "step": 15871 }, { "epoch": 39.68, "grad_norm": 6.8125, "learning_rate": 1.373538639801657e-05, "loss": 1.4298, "step": 15872 }, { "epoch": 39.6825, "grad_norm": 5.625, "learning_rate": 1.3734635306577556e-05, "loss": 1.3988, "step": 15873 }, { "epoch": 39.685, "grad_norm": 5.09375, "learning_rate": 1.3733884190654418e-05, "loss": 1.4105, "step": 15874 }, { "epoch": 39.6875, "grad_norm": 7.78125, "learning_rate": 1.3733133050252087e-05, "loss": 1.439, "step": 15875 }, { "epoch": 39.69, "grad_norm": 8.9375, "learning_rate": 1.3732381885375488e-05, "loss": 1.5028, "step": 15876 }, { "epoch": 39.6925, "grad_norm": 6.46875, "learning_rate": 1.373163069602954e-05, "loss": 1.3609, "step": 15877 }, { "epoch": 39.695, "grad_norm": 7.625, "learning_rate": 1.373087948221917e-05, "loss": 1.3504, "step": 15878 }, { "epoch": 39.6975, "grad_norm": 11.8125, "learning_rate": 1.3730128243949306e-05, "loss": 1.37, "step": 15879 }, { "epoch": 39.7, "grad_norm": 7.15625, "learning_rate": 1.3729376981224869e-05, "loss": 1.2727, "step": 15880 }, { "epoch": 39.7025, "grad_norm": 7.0, "learning_rate": 1.3728625694050783e-05, "loss": 1.3603, "step": 15881 }, { "epoch": 39.705, "grad_norm": 6.46875, "learning_rate": 1.372787438243198e-05, "loss": 1.3458, "step": 15882 }, { "epoch": 39.7075, "grad_norm": 6.65625, "learning_rate": 1.3727123046373382e-05, "loss": 1.3988, "step": 15883 }, { "epoch": 39.71, "grad_norm": 11.0625, "learning_rate": 1.3726371685879911e-05, "loss": 1.2724, "step": 15884 }, { "epoch": 39.7125, "grad_norm": 7.375, "learning_rate": 1.3725620300956499e-05, "loss": 1.3185, "step": 15885 }, { "epoch": 39.715, "grad_norm": 8.25, "learning_rate": 1.3724868891608067e-05, "loss": 1.3791, "step": 15886 }, { "epoch": 39.7175, "grad_norm": 7.84375, "learning_rate": 1.3724117457839547e-05, "loss": 1.3773, "step": 15887 }, { "epoch": 39.72, "grad_norm": 6.875, "learning_rate": 1.3723365999655859e-05, "loss": 1.4, "step": 15888 }, { "epoch": 39.7225, "grad_norm": 6.09375, "learning_rate": 1.3722614517061936e-05, "loss": 1.2648, "step": 15889 }, { "epoch": 39.725, "grad_norm": 11.625, "learning_rate": 1.37218630100627e-05, "loss": 1.4277, "step": 15890 }, { "epoch": 39.7275, "grad_norm": 6.46875, "learning_rate": 1.3721111478663075e-05, "loss": 1.2786, "step": 15891 }, { "epoch": 39.73, "grad_norm": 5.0, "learning_rate": 1.3720359922867996e-05, "loss": 1.2432, "step": 15892 }, { "epoch": 39.7325, "grad_norm": 7.0, "learning_rate": 1.3719608342682387e-05, "loss": 1.4968, "step": 15893 }, { "epoch": 39.735, "grad_norm": 8.1875, "learning_rate": 1.3718856738111171e-05, "loss": 1.3851, "step": 15894 }, { "epoch": 39.7375, "grad_norm": 8.0625, "learning_rate": 1.3718105109159281e-05, "loss": 1.2916, "step": 15895 }, { "epoch": 39.74, "grad_norm": 9.5, "learning_rate": 1.3717353455831643e-05, "loss": 1.3845, "step": 15896 }, { "epoch": 39.7425, "grad_norm": 8.0625, "learning_rate": 1.3716601778133183e-05, "loss": 1.3137, "step": 15897 }, { "epoch": 39.745, "grad_norm": 6.4375, "learning_rate": 1.3715850076068836e-05, "loss": 1.0778, "step": 15898 }, { "epoch": 39.7475, "grad_norm": 5.78125, "learning_rate": 1.3715098349643516e-05, "loss": 1.3827, "step": 15899 }, { "epoch": 39.75, "grad_norm": 7.46875, "learning_rate": 1.3714346598862168e-05, "loss": 1.3644, "step": 15900 }, { "epoch": 39.7525, "grad_norm": 13.4375, "learning_rate": 1.3713594823729706e-05, "loss": 1.2142, "step": 15901 }, { "epoch": 39.755, "grad_norm": 7.5, "learning_rate": 1.3712843024251068e-05, "loss": 1.3144, "step": 15902 }, { "epoch": 39.7575, "grad_norm": 8.875, "learning_rate": 1.371209120043118e-05, "loss": 1.422, "step": 15903 }, { "epoch": 39.76, "grad_norm": 6.1875, "learning_rate": 1.3711339352274969e-05, "loss": 1.3802, "step": 15904 }, { "epoch": 39.7625, "grad_norm": 8.625, "learning_rate": 1.3710587479787364e-05, "loss": 1.4628, "step": 15905 }, { "epoch": 39.765, "grad_norm": 7.5, "learning_rate": 1.37098355829733e-05, "loss": 1.4137, "step": 15906 }, { "epoch": 39.7675, "grad_norm": 7.96875, "learning_rate": 1.3709083661837702e-05, "loss": 1.4613, "step": 15907 }, { "epoch": 39.77, "grad_norm": 7.71875, "learning_rate": 1.3708331716385497e-05, "loss": 1.3269, "step": 15908 }, { "epoch": 39.7725, "grad_norm": 6.9375, "learning_rate": 1.370757974662162e-05, "loss": 1.3161, "step": 15909 }, { "epoch": 39.775, "grad_norm": 8.9375, "learning_rate": 1.3706827752550997e-05, "loss": 1.3324, "step": 15910 }, { "epoch": 39.7775, "grad_norm": 7.25, "learning_rate": 1.370607573417856e-05, "loss": 1.2614, "step": 15911 }, { "epoch": 39.78, "grad_norm": 8.4375, "learning_rate": 1.370532369150924e-05, "loss": 1.3684, "step": 15912 }, { "epoch": 39.7825, "grad_norm": 5.6875, "learning_rate": 1.3704571624547968e-05, "loss": 1.2939, "step": 15913 }, { "epoch": 39.785, "grad_norm": 6.9375, "learning_rate": 1.3703819533299666e-05, "loss": 1.4351, "step": 15914 }, { "epoch": 39.7875, "grad_norm": 7.625, "learning_rate": 1.3703067417769278e-05, "loss": 1.2529, "step": 15915 }, { "epoch": 39.79, "grad_norm": 9.25, "learning_rate": 1.3702315277961724e-05, "loss": 1.2068, "step": 15916 }, { "epoch": 39.7925, "grad_norm": 5.65625, "learning_rate": 1.3701563113881943e-05, "loss": 1.1611, "step": 15917 }, { "epoch": 39.795, "grad_norm": 8.75, "learning_rate": 1.370081092553486e-05, "loss": 1.2192, "step": 15918 }, { "epoch": 39.7975, "grad_norm": 7.40625, "learning_rate": 1.3700058712925409e-05, "loss": 1.5846, "step": 15919 }, { "epoch": 39.8, "grad_norm": 4.375, "learning_rate": 1.3699306476058523e-05, "loss": 1.2187, "step": 15920 }, { "epoch": 39.8025, "grad_norm": 6.59375, "learning_rate": 1.3698554214939129e-05, "loss": 1.3251, "step": 15921 }, { "epoch": 39.805, "grad_norm": 8.875, "learning_rate": 1.3697801929572162e-05, "loss": 1.4072, "step": 15922 }, { "epoch": 39.8075, "grad_norm": 8.6875, "learning_rate": 1.3697049619962554e-05, "loss": 1.29, "step": 15923 }, { "epoch": 39.81, "grad_norm": 5.8125, "learning_rate": 1.3696297286115237e-05, "loss": 1.2895, "step": 15924 }, { "epoch": 39.8125, "grad_norm": 7.84375, "learning_rate": 1.369554492803514e-05, "loss": 1.3837, "step": 15925 }, { "epoch": 39.815, "grad_norm": 13.5, "learning_rate": 1.3694792545727202e-05, "loss": 1.4475, "step": 15926 }, { "epoch": 39.8175, "grad_norm": 8.1875, "learning_rate": 1.3694040139196352e-05, "loss": 1.5397, "step": 15927 }, { "epoch": 39.82, "grad_norm": 7.4375, "learning_rate": 1.3693287708447523e-05, "loss": 1.4871, "step": 15928 }, { "epoch": 39.8225, "grad_norm": 6.875, "learning_rate": 1.3692535253485643e-05, "loss": 1.3771, "step": 15929 }, { "epoch": 39.825, "grad_norm": 7.875, "learning_rate": 1.3691782774315654e-05, "loss": 1.2071, "step": 15930 }, { "epoch": 39.8275, "grad_norm": 8.125, "learning_rate": 1.3691030270942482e-05, "loss": 1.3603, "step": 15931 }, { "epoch": 39.83, "grad_norm": 7.15625, "learning_rate": 1.3690277743371065e-05, "loss": 1.3441, "step": 15932 }, { "epoch": 39.8325, "grad_norm": 8.0625, "learning_rate": 1.3689525191606334e-05, "loss": 1.4887, "step": 15933 }, { "epoch": 39.835, "grad_norm": 8.1875, "learning_rate": 1.3688772615653225e-05, "loss": 1.4717, "step": 15934 }, { "epoch": 39.8375, "grad_norm": 5.5625, "learning_rate": 1.368802001551667e-05, "loss": 1.4936, "step": 15935 }, { "epoch": 39.84, "grad_norm": 8.1875, "learning_rate": 1.3687267391201604e-05, "loss": 1.42, "step": 15936 }, { "epoch": 39.8425, "grad_norm": 10.4375, "learning_rate": 1.3686514742712959e-05, "loss": 1.4317, "step": 15937 }, { "epoch": 39.845, "grad_norm": 9.625, "learning_rate": 1.368576207005567e-05, "loss": 1.3651, "step": 15938 }, { "epoch": 39.8475, "grad_norm": 4.9375, "learning_rate": 1.3685009373234675e-05, "loss": 1.2025, "step": 15939 }, { "epoch": 39.85, "grad_norm": 6.15625, "learning_rate": 1.3684256652254906e-05, "loss": 1.2407, "step": 15940 }, { "epoch": 39.8525, "grad_norm": 7.59375, "learning_rate": 1.3683503907121294e-05, "loss": 1.3957, "step": 15941 }, { "epoch": 39.855, "grad_norm": 9.0, "learning_rate": 1.3682751137838781e-05, "loss": 1.2455, "step": 15942 }, { "epoch": 39.8575, "grad_norm": 7.0625, "learning_rate": 1.3681998344412297e-05, "loss": 1.3352, "step": 15943 }, { "epoch": 39.86, "grad_norm": 8.1875, "learning_rate": 1.3681245526846782e-05, "loss": 1.2941, "step": 15944 }, { "epoch": 39.8625, "grad_norm": 8.125, "learning_rate": 1.3680492685147167e-05, "loss": 1.2971, "step": 15945 }, { "epoch": 39.865, "grad_norm": 8.125, "learning_rate": 1.3679739819318389e-05, "loss": 1.3631, "step": 15946 }, { "epoch": 39.8675, "grad_norm": 6.96875, "learning_rate": 1.3678986929365384e-05, "loss": 1.2389, "step": 15947 }, { "epoch": 39.87, "grad_norm": 9.1875, "learning_rate": 1.3678234015293089e-05, "loss": 1.2279, "step": 15948 }, { "epoch": 39.8725, "grad_norm": 7.9375, "learning_rate": 1.3677481077106438e-05, "loss": 1.4158, "step": 15949 }, { "epoch": 39.875, "grad_norm": 8.5, "learning_rate": 1.3676728114810367e-05, "loss": 1.3374, "step": 15950 }, { "epoch": 39.8775, "grad_norm": 9.0625, "learning_rate": 1.3675975128409814e-05, "loss": 1.2506, "step": 15951 }, { "epoch": 39.88, "grad_norm": 6.625, "learning_rate": 1.3675222117909716e-05, "loss": 1.2705, "step": 15952 }, { "epoch": 39.8825, "grad_norm": 6.625, "learning_rate": 1.3674469083315009e-05, "loss": 1.4293, "step": 15953 }, { "epoch": 39.885, "grad_norm": 6.34375, "learning_rate": 1.367371602463063e-05, "loss": 1.5225, "step": 15954 }, { "epoch": 39.8875, "grad_norm": 6.28125, "learning_rate": 1.3672962941861518e-05, "loss": 1.3811, "step": 15955 }, { "epoch": 39.89, "grad_norm": 9.0625, "learning_rate": 1.3672209835012602e-05, "loss": 1.3962, "step": 15956 }, { "epoch": 39.8925, "grad_norm": 7.5, "learning_rate": 1.3671456704088832e-05, "loss": 1.3346, "step": 15957 }, { "epoch": 39.895, "grad_norm": 8.875, "learning_rate": 1.3670703549095133e-05, "loss": 1.5516, "step": 15958 }, { "epoch": 39.8975, "grad_norm": 4.6875, "learning_rate": 1.3669950370036453e-05, "loss": 1.2353, "step": 15959 }, { "epoch": 39.9, "grad_norm": 12.0, "learning_rate": 1.3669197166917723e-05, "loss": 1.4358, "step": 15960 }, { "epoch": 39.9025, "grad_norm": 9.625, "learning_rate": 1.3668443939743884e-05, "loss": 1.3959, "step": 15961 }, { "epoch": 39.905, "grad_norm": 6.40625, "learning_rate": 1.3667690688519873e-05, "loss": 1.2496, "step": 15962 }, { "epoch": 39.9075, "grad_norm": 5.5, "learning_rate": 1.3666937413250627e-05, "loss": 1.2983, "step": 15963 }, { "epoch": 39.91, "grad_norm": 8.375, "learning_rate": 1.3666184113941086e-05, "loss": 1.3317, "step": 15964 }, { "epoch": 39.9125, "grad_norm": 6.9375, "learning_rate": 1.366543079059619e-05, "loss": 1.3617, "step": 15965 }, { "epoch": 39.915, "grad_norm": 8.25, "learning_rate": 1.3664677443220878e-05, "loss": 1.3292, "step": 15966 }, { "epoch": 39.9175, "grad_norm": 5.75, "learning_rate": 1.3663924071820085e-05, "loss": 1.3651, "step": 15967 }, { "epoch": 39.92, "grad_norm": 6.25, "learning_rate": 1.366317067639875e-05, "loss": 1.2816, "step": 15968 }, { "epoch": 39.9225, "grad_norm": 8.0, "learning_rate": 1.366241725696182e-05, "loss": 1.2892, "step": 15969 }, { "epoch": 39.925, "grad_norm": 5.5, "learning_rate": 1.366166381351423e-05, "loss": 1.3076, "step": 15970 }, { "epoch": 39.9275, "grad_norm": 6.6875, "learning_rate": 1.3660910346060912e-05, "loss": 1.284, "step": 15971 }, { "epoch": 39.93, "grad_norm": 6.65625, "learning_rate": 1.3660156854606819e-05, "loss": 1.4801, "step": 15972 }, { "epoch": 39.9325, "grad_norm": 8.4375, "learning_rate": 1.3659403339156878e-05, "loss": 1.2154, "step": 15973 }, { "epoch": 39.935, "grad_norm": 6.6875, "learning_rate": 1.365864979971604e-05, "loss": 1.3335, "step": 15974 }, { "epoch": 39.9375, "grad_norm": 10.625, "learning_rate": 1.3657896236289237e-05, "loss": 1.4092, "step": 15975 }, { "epoch": 39.94, "grad_norm": 9.3125, "learning_rate": 1.3657142648881414e-05, "loss": 1.2614, "step": 15976 }, { "epoch": 39.9425, "grad_norm": 8.125, "learning_rate": 1.365638903749751e-05, "loss": 1.4008, "step": 15977 }, { "epoch": 39.945, "grad_norm": 5.28125, "learning_rate": 1.3655635402142467e-05, "loss": 1.3514, "step": 15978 }, { "epoch": 39.9475, "grad_norm": 7.46875, "learning_rate": 1.3654881742821222e-05, "loss": 1.2655, "step": 15979 }, { "epoch": 39.95, "grad_norm": 8.4375, "learning_rate": 1.365412805953872e-05, "loss": 1.2847, "step": 15980 }, { "epoch": 39.9525, "grad_norm": 9.9375, "learning_rate": 1.3653374352299901e-05, "loss": 1.4298, "step": 15981 }, { "epoch": 39.955, "grad_norm": 8.0625, "learning_rate": 1.3652620621109707e-05, "loss": 1.323, "step": 15982 }, { "epoch": 39.9575, "grad_norm": 6.375, "learning_rate": 1.3651866865973076e-05, "loss": 1.4188, "step": 15983 }, { "epoch": 39.96, "grad_norm": 7.46875, "learning_rate": 1.3651113086894951e-05, "loss": 1.3582, "step": 15984 }, { "epoch": 39.9625, "grad_norm": 9.6875, "learning_rate": 1.3650359283880279e-05, "loss": 1.3417, "step": 15985 }, { "epoch": 39.965, "grad_norm": 5.875, "learning_rate": 1.3649605456933994e-05, "loss": 1.3431, "step": 15986 }, { "epoch": 39.9675, "grad_norm": 5.03125, "learning_rate": 1.3648851606061044e-05, "loss": 1.3679, "step": 15987 }, { "epoch": 39.97, "grad_norm": 7.8125, "learning_rate": 1.3648097731266368e-05, "loss": 1.3591, "step": 15988 }, { "epoch": 39.9725, "grad_norm": 8.1875, "learning_rate": 1.364734383255491e-05, "loss": 1.3945, "step": 15989 }, { "epoch": 39.975, "grad_norm": 6.0, "learning_rate": 1.3646589909931609e-05, "loss": 1.3277, "step": 15990 }, { "epoch": 39.9775, "grad_norm": 6.96875, "learning_rate": 1.3645835963401415e-05, "loss": 1.3794, "step": 15991 }, { "epoch": 39.98, "grad_norm": 6.71875, "learning_rate": 1.3645081992969262e-05, "loss": 1.243, "step": 15992 }, { "epoch": 39.9825, "grad_norm": 7.6875, "learning_rate": 1.36443279986401e-05, "loss": 1.248, "step": 15993 }, { "epoch": 39.985, "grad_norm": 7.59375, "learning_rate": 1.3643573980418866e-05, "loss": 1.2312, "step": 15994 }, { "epoch": 39.9875, "grad_norm": 5.78125, "learning_rate": 1.364281993831051e-05, "loss": 1.1899, "step": 15995 }, { "epoch": 39.99, "grad_norm": 7.125, "learning_rate": 1.3642065872319971e-05, "loss": 1.2909, "step": 15996 }, { "epoch": 39.9925, "grad_norm": 9.9375, "learning_rate": 1.3641311782452195e-05, "loss": 1.2407, "step": 15997 }, { "epoch": 39.995, "grad_norm": 7.9375, "learning_rate": 1.3640557668712124e-05, "loss": 1.1471, "step": 15998 }, { "epoch": 39.9975, "grad_norm": 6.90625, "learning_rate": 1.3639803531104702e-05, "loss": 1.3248, "step": 15999 }, { "epoch": 40.0, "grad_norm": 7.5, "learning_rate": 1.3639049369634878e-05, "loss": 1.2605, "step": 16000 }, { "epoch": 40.0025, "grad_norm": 8.5, "learning_rate": 1.3638295184307585e-05, "loss": 1.2082, "step": 16001 }, { "epoch": 40.005, "grad_norm": 7.125, "learning_rate": 1.363754097512778e-05, "loss": 1.2314, "step": 16002 }, { "epoch": 40.0075, "grad_norm": 7.75, "learning_rate": 1.3636786742100398e-05, "loss": 1.388, "step": 16003 }, { "epoch": 40.01, "grad_norm": 7.53125, "learning_rate": 1.363603248523039e-05, "loss": 1.3409, "step": 16004 }, { "epoch": 40.0125, "grad_norm": 10.5, "learning_rate": 1.36352782045227e-05, "loss": 1.2974, "step": 16005 }, { "epoch": 40.015, "grad_norm": 5.96875, "learning_rate": 1.3634523899982269e-05, "loss": 1.2972, "step": 16006 }, { "epoch": 40.0175, "grad_norm": 8.1875, "learning_rate": 1.3633769571614043e-05, "loss": 1.263, "step": 16007 }, { "epoch": 40.02, "grad_norm": 7.96875, "learning_rate": 1.3633015219422974e-05, "loss": 1.1937, "step": 16008 }, { "epoch": 40.0225, "grad_norm": 9.9375, "learning_rate": 1.3632260843413998e-05, "loss": 1.4946, "step": 16009 }, { "epoch": 40.025, "grad_norm": 8.6875, "learning_rate": 1.363150644359207e-05, "loss": 1.3656, "step": 16010 }, { "epoch": 40.0275, "grad_norm": 7.3125, "learning_rate": 1.3630752019962126e-05, "loss": 1.1975, "step": 16011 }, { "epoch": 40.03, "grad_norm": 5.6875, "learning_rate": 1.362999757252912e-05, "loss": 1.2605, "step": 16012 }, { "epoch": 40.0325, "grad_norm": 5.5, "learning_rate": 1.3629243101297992e-05, "loss": 1.2635, "step": 16013 }, { "epoch": 40.035, "grad_norm": 7.75, "learning_rate": 1.3628488606273697e-05, "loss": 1.385, "step": 16014 }, { "epoch": 40.0375, "grad_norm": 7.15625, "learning_rate": 1.3627734087461173e-05, "loss": 1.5067, "step": 16015 }, { "epoch": 40.04, "grad_norm": 8.6875, "learning_rate": 1.3626979544865369e-05, "loss": 1.2597, "step": 16016 }, { "epoch": 40.0425, "grad_norm": 10.625, "learning_rate": 1.3626224978491233e-05, "loss": 1.3719, "step": 16017 }, { "epoch": 40.045, "grad_norm": 7.125, "learning_rate": 1.3625470388343712e-05, "loss": 1.4344, "step": 16018 }, { "epoch": 40.0475, "grad_norm": 6.59375, "learning_rate": 1.3624715774427747e-05, "loss": 1.2421, "step": 16019 }, { "epoch": 40.05, "grad_norm": 8.4375, "learning_rate": 1.3623961136748296e-05, "loss": 1.2745, "step": 16020 }, { "epoch": 40.0525, "grad_norm": 10.75, "learning_rate": 1.36232064753103e-05, "loss": 1.4656, "step": 16021 }, { "epoch": 40.055, "grad_norm": 9.6875, "learning_rate": 1.3622451790118704e-05, "loss": 1.3175, "step": 16022 }, { "epoch": 40.0575, "grad_norm": 6.46875, "learning_rate": 1.3621697081178464e-05, "loss": 1.2942, "step": 16023 }, { "epoch": 40.06, "grad_norm": 6.125, "learning_rate": 1.3620942348494518e-05, "loss": 1.5398, "step": 16024 }, { "epoch": 40.0625, "grad_norm": 7.625, "learning_rate": 1.3620187592071825e-05, "loss": 1.3467, "step": 16025 }, { "epoch": 40.065, "grad_norm": 6.96875, "learning_rate": 1.361943281191532e-05, "loss": 1.2365, "step": 16026 }, { "epoch": 40.0675, "grad_norm": 6.78125, "learning_rate": 1.3618678008029964e-05, "loss": 1.2208, "step": 16027 }, { "epoch": 40.07, "grad_norm": 10.5625, "learning_rate": 1.3617923180420696e-05, "loss": 1.3237, "step": 16028 }, { "epoch": 40.0725, "grad_norm": 7.59375, "learning_rate": 1.3617168329092471e-05, "loss": 1.2169, "step": 16029 }, { "epoch": 40.075, "grad_norm": 6.65625, "learning_rate": 1.3616413454050234e-05, "loss": 1.5683, "step": 16030 }, { "epoch": 40.0775, "grad_norm": 10.3125, "learning_rate": 1.3615658555298936e-05, "loss": 1.4017, "step": 16031 }, { "epoch": 40.08, "grad_norm": 7.03125, "learning_rate": 1.3614903632843523e-05, "loss": 1.3296, "step": 16032 }, { "epoch": 40.0825, "grad_norm": 7.75, "learning_rate": 1.3614148686688948e-05, "loss": 1.4937, "step": 16033 }, { "epoch": 40.085, "grad_norm": 5.96875, "learning_rate": 1.361339371684016e-05, "loss": 1.1395, "step": 16034 }, { "epoch": 40.0875, "grad_norm": 7.65625, "learning_rate": 1.3612638723302104e-05, "loss": 1.3879, "step": 16035 }, { "epoch": 40.09, "grad_norm": 6.90625, "learning_rate": 1.3611883706079735e-05, "loss": 1.304, "step": 16036 }, { "epoch": 40.0925, "grad_norm": 8.4375, "learning_rate": 1.3611128665178002e-05, "loss": 1.2826, "step": 16037 }, { "epoch": 40.095, "grad_norm": 6.21875, "learning_rate": 1.3610373600601854e-05, "loss": 1.3307, "step": 16038 }, { "epoch": 40.0975, "grad_norm": 14.1875, "learning_rate": 1.360961851235624e-05, "loss": 1.3946, "step": 16039 }, { "epoch": 40.1, "grad_norm": 5.75, "learning_rate": 1.3608863400446113e-05, "loss": 1.3489, "step": 16040 }, { "epoch": 40.1025, "grad_norm": 14.3125, "learning_rate": 1.3608108264876418e-05, "loss": 1.3642, "step": 16041 }, { "epoch": 40.105, "grad_norm": 7.84375, "learning_rate": 1.3607353105652115e-05, "loss": 1.3915, "step": 16042 }, { "epoch": 40.1075, "grad_norm": 9.8125, "learning_rate": 1.3606597922778145e-05, "loss": 1.2904, "step": 16043 }, { "epoch": 40.11, "grad_norm": 5.6875, "learning_rate": 1.3605842716259465e-05, "loss": 1.1421, "step": 16044 }, { "epoch": 40.1125, "grad_norm": 7.46875, "learning_rate": 1.3605087486101025e-05, "loss": 1.342, "step": 16045 }, { "epoch": 40.115, "grad_norm": 9.1875, "learning_rate": 1.3604332232307775e-05, "loss": 1.3583, "step": 16046 }, { "epoch": 40.1175, "grad_norm": 6.8125, "learning_rate": 1.3603576954884667e-05, "loss": 1.3695, "step": 16047 }, { "epoch": 40.12, "grad_norm": 7.5, "learning_rate": 1.3602821653836654e-05, "loss": 1.3103, "step": 16048 }, { "epoch": 40.1225, "grad_norm": 7.34375, "learning_rate": 1.3602066329168686e-05, "loss": 1.3457, "step": 16049 }, { "epoch": 40.125, "grad_norm": 6.25, "learning_rate": 1.3601310980885714e-05, "loss": 1.4407, "step": 16050 }, { "epoch": 40.1275, "grad_norm": 8.4375, "learning_rate": 1.3600555608992691e-05, "loss": 1.3275, "step": 16051 }, { "epoch": 40.13, "grad_norm": 7.5, "learning_rate": 1.3599800213494571e-05, "loss": 1.4626, "step": 16052 }, { "epoch": 40.1325, "grad_norm": 8.25, "learning_rate": 1.3599044794396306e-05, "loss": 1.4206, "step": 16053 }, { "epoch": 40.135, "grad_norm": 5.53125, "learning_rate": 1.3598289351702845e-05, "loss": 1.1752, "step": 16054 }, { "epoch": 40.1375, "grad_norm": 7.0, "learning_rate": 1.3597533885419142e-05, "loss": 1.4085, "step": 16055 }, { "epoch": 40.14, "grad_norm": 5.25, "learning_rate": 1.3596778395550154e-05, "loss": 1.3001, "step": 16056 }, { "epoch": 40.1425, "grad_norm": 7.15625, "learning_rate": 1.359602288210083e-05, "loss": 1.4292, "step": 16057 }, { "epoch": 40.145, "grad_norm": 7.84375, "learning_rate": 1.3595267345076122e-05, "loss": 1.4136, "step": 16058 }, { "epoch": 40.1475, "grad_norm": 9.875, "learning_rate": 1.3594511784480989e-05, "loss": 1.2211, "step": 16059 }, { "epoch": 40.15, "grad_norm": 5.875, "learning_rate": 1.3593756200320373e-05, "loss": 1.3691, "step": 16060 }, { "epoch": 40.1525, "grad_norm": 8.75, "learning_rate": 1.3593000592599244e-05, "loss": 1.2465, "step": 16061 }, { "epoch": 40.155, "grad_norm": 6.53125, "learning_rate": 1.3592244961322543e-05, "loss": 1.2483, "step": 16062 }, { "epoch": 40.1575, "grad_norm": 7.6875, "learning_rate": 1.3591489306495226e-05, "loss": 1.3407, "step": 16063 }, { "epoch": 40.16, "grad_norm": 6.0, "learning_rate": 1.3590733628122253e-05, "loss": 1.1705, "step": 16064 }, { "epoch": 40.1625, "grad_norm": 8.25, "learning_rate": 1.358997792620857e-05, "loss": 1.4666, "step": 16065 }, { "epoch": 40.165, "grad_norm": 7.125, "learning_rate": 1.3589222200759138e-05, "loss": 1.4004, "step": 16066 }, { "epoch": 40.1675, "grad_norm": 4.9375, "learning_rate": 1.3588466451778908e-05, "loss": 1.2174, "step": 16067 }, { "epoch": 40.17, "grad_norm": 5.59375, "learning_rate": 1.3587710679272833e-05, "loss": 1.333, "step": 16068 }, { "epoch": 40.1725, "grad_norm": 7.78125, "learning_rate": 1.3586954883245875e-05, "loss": 1.301, "step": 16069 }, { "epoch": 40.175, "grad_norm": 7.625, "learning_rate": 1.358619906370298e-05, "loss": 1.2468, "step": 16070 }, { "epoch": 40.1775, "grad_norm": 7.1875, "learning_rate": 1.3585443220649111e-05, "loss": 1.2892, "step": 16071 }, { "epoch": 40.18, "grad_norm": 11.1875, "learning_rate": 1.3584687354089222e-05, "loss": 1.5002, "step": 16072 }, { "epoch": 40.1825, "grad_norm": 6.09375, "learning_rate": 1.3583931464028259e-05, "loss": 1.298, "step": 16073 }, { "epoch": 40.185, "grad_norm": 9.9375, "learning_rate": 1.358317555047119e-05, "loss": 1.362, "step": 16074 }, { "epoch": 40.1875, "grad_norm": 7.15625, "learning_rate": 1.3582419613422962e-05, "loss": 1.302, "step": 16075 }, { "epoch": 40.19, "grad_norm": 5.625, "learning_rate": 1.3581663652888536e-05, "loss": 1.2344, "step": 16076 }, { "epoch": 40.1925, "grad_norm": 9.5, "learning_rate": 1.3580907668872868e-05, "loss": 1.481, "step": 16077 }, { "epoch": 40.195, "grad_norm": 7.3125, "learning_rate": 1.3580151661380911e-05, "loss": 1.3618, "step": 16078 }, { "epoch": 40.1975, "grad_norm": 5.21875, "learning_rate": 1.3579395630417622e-05, "loss": 1.31, "step": 16079 }, { "epoch": 40.2, "grad_norm": 7.6875, "learning_rate": 1.357863957598796e-05, "loss": 1.2328, "step": 16080 }, { "epoch": 40.2025, "grad_norm": 10.3125, "learning_rate": 1.357788349809688e-05, "loss": 1.4305, "step": 16081 }, { "epoch": 40.205, "grad_norm": 5.90625, "learning_rate": 1.3577127396749341e-05, "loss": 1.5005, "step": 16082 }, { "epoch": 40.2075, "grad_norm": 8.625, "learning_rate": 1.3576371271950292e-05, "loss": 1.3563, "step": 16083 }, { "epoch": 40.21, "grad_norm": 9.25, "learning_rate": 1.3575615123704703e-05, "loss": 1.3249, "step": 16084 }, { "epoch": 40.2125, "grad_norm": 6.46875, "learning_rate": 1.357485895201752e-05, "loss": 1.2036, "step": 16085 }, { "epoch": 40.215, "grad_norm": 6.65625, "learning_rate": 1.3574102756893708e-05, "loss": 1.5295, "step": 16086 }, { "epoch": 40.2175, "grad_norm": 5.3125, "learning_rate": 1.3573346538338217e-05, "loss": 1.2927, "step": 16087 }, { "epoch": 40.22, "grad_norm": 5.90625, "learning_rate": 1.3572590296356012e-05, "loss": 1.347, "step": 16088 }, { "epoch": 40.2225, "grad_norm": 4.875, "learning_rate": 1.3571834030952047e-05, "loss": 1.2725, "step": 16089 }, { "epoch": 40.225, "grad_norm": 5.75, "learning_rate": 1.357107774213128e-05, "loss": 1.3533, "step": 16090 }, { "epoch": 40.2275, "grad_norm": 8.375, "learning_rate": 1.3570321429898674e-05, "loss": 1.4945, "step": 16091 }, { "epoch": 40.23, "grad_norm": 8.25, "learning_rate": 1.356956509425918e-05, "loss": 1.2757, "step": 16092 }, { "epoch": 40.2325, "grad_norm": 6.59375, "learning_rate": 1.3568808735217762e-05, "loss": 1.2237, "step": 16093 }, { "epoch": 40.235, "grad_norm": 8.625, "learning_rate": 1.3568052352779375e-05, "loss": 1.4216, "step": 16094 }, { "epoch": 40.2375, "grad_norm": 11.625, "learning_rate": 1.356729594694898e-05, "loss": 1.3634, "step": 16095 }, { "epoch": 40.24, "grad_norm": 8.5, "learning_rate": 1.3566539517731536e-05, "loss": 1.4251, "step": 16096 }, { "epoch": 40.2425, "grad_norm": 8.4375, "learning_rate": 1.3565783065132004e-05, "loss": 1.3526, "step": 16097 }, { "epoch": 40.245, "grad_norm": 7.40625, "learning_rate": 1.3565026589155336e-05, "loss": 1.383, "step": 16098 }, { "epoch": 40.2475, "grad_norm": 6.8125, "learning_rate": 1.3564270089806502e-05, "loss": 1.41, "step": 16099 }, { "epoch": 40.25, "grad_norm": 9.125, "learning_rate": 1.356351356709045e-05, "loss": 1.275, "step": 16100 }, { "epoch": 40.2525, "grad_norm": 7.53125, "learning_rate": 1.3562757021012152e-05, "loss": 1.3273, "step": 16101 }, { "epoch": 40.255, "grad_norm": 7.0, "learning_rate": 1.3562000451576556e-05, "loss": 1.3749, "step": 16102 }, { "epoch": 40.2575, "grad_norm": 6.65625, "learning_rate": 1.356124385878863e-05, "loss": 1.283, "step": 16103 }, { "epoch": 40.26, "grad_norm": 5.1875, "learning_rate": 1.3560487242653331e-05, "loss": 1.3668, "step": 16104 }, { "epoch": 40.2625, "grad_norm": 8.5625, "learning_rate": 1.355973060317562e-05, "loss": 1.3434, "step": 16105 }, { "epoch": 40.265, "grad_norm": 7.59375, "learning_rate": 1.3558973940360458e-05, "loss": 1.2916, "step": 16106 }, { "epoch": 40.2675, "grad_norm": 6.1875, "learning_rate": 1.3558217254212807e-05, "loss": 1.2294, "step": 16107 }, { "epoch": 40.27, "grad_norm": 7.625, "learning_rate": 1.3557460544737624e-05, "loss": 1.3326, "step": 16108 }, { "epoch": 40.2725, "grad_norm": 5.0, "learning_rate": 1.355670381193987e-05, "loss": 1.2318, "step": 16109 }, { "epoch": 40.275, "grad_norm": 7.59375, "learning_rate": 1.3555947055824511e-05, "loss": 1.3691, "step": 16110 }, { "epoch": 40.2775, "grad_norm": 5.75, "learning_rate": 1.3555190276396506e-05, "loss": 1.281, "step": 16111 }, { "epoch": 40.28, "grad_norm": 9.4375, "learning_rate": 1.3554433473660818e-05, "loss": 1.1737, "step": 16112 }, { "epoch": 40.2825, "grad_norm": 5.40625, "learning_rate": 1.35536766476224e-05, "loss": 1.2863, "step": 16113 }, { "epoch": 40.285, "grad_norm": 6.21875, "learning_rate": 1.3552919798286225e-05, "loss": 1.4057, "step": 16114 }, { "epoch": 40.2875, "grad_norm": 7.53125, "learning_rate": 1.3552162925657246e-05, "loss": 1.356, "step": 16115 }, { "epoch": 40.29, "grad_norm": 6.71875, "learning_rate": 1.3551406029740435e-05, "loss": 1.4378, "step": 16116 }, { "epoch": 40.2925, "grad_norm": 9.0, "learning_rate": 1.3550649110540743e-05, "loss": 1.3983, "step": 16117 }, { "epoch": 40.295, "grad_norm": 7.8125, "learning_rate": 1.354989216806314e-05, "loss": 1.3268, "step": 16118 }, { "epoch": 40.2975, "grad_norm": 7.875, "learning_rate": 1.3549135202312584e-05, "loss": 1.4031, "step": 16119 }, { "epoch": 40.3, "grad_norm": 9.0625, "learning_rate": 1.3548378213294042e-05, "loss": 1.2913, "step": 16120 }, { "epoch": 40.3025, "grad_norm": 6.09375, "learning_rate": 1.3547621201012473e-05, "loss": 1.3437, "step": 16121 }, { "epoch": 40.305, "grad_norm": 7.71875, "learning_rate": 1.3546864165472842e-05, "loss": 1.284, "step": 16122 }, { "epoch": 40.3075, "grad_norm": 6.9375, "learning_rate": 1.3546107106680108e-05, "loss": 1.4645, "step": 16123 }, { "epoch": 40.31, "grad_norm": 7.03125, "learning_rate": 1.354535002463924e-05, "loss": 1.2788, "step": 16124 }, { "epoch": 40.3125, "grad_norm": 8.125, "learning_rate": 1.3544592919355203e-05, "loss": 1.3659, "step": 16125 }, { "epoch": 40.315, "grad_norm": 6.71875, "learning_rate": 1.3543835790832953e-05, "loss": 1.3043, "step": 16126 }, { "epoch": 40.3175, "grad_norm": 7.28125, "learning_rate": 1.3543078639077459e-05, "loss": 1.3015, "step": 16127 }, { "epoch": 40.32, "grad_norm": 7.21875, "learning_rate": 1.354232146409368e-05, "loss": 1.2372, "step": 16128 }, { "epoch": 40.3225, "grad_norm": 5.78125, "learning_rate": 1.3541564265886589e-05, "loss": 1.3769, "step": 16129 }, { "epoch": 40.325, "grad_norm": 9.5, "learning_rate": 1.3540807044461137e-05, "loss": 1.4468, "step": 16130 }, { "epoch": 40.3275, "grad_norm": 7.84375, "learning_rate": 1.3540049799822302e-05, "loss": 1.3865, "step": 16131 }, { "epoch": 40.33, "grad_norm": 8.9375, "learning_rate": 1.3539292531975038e-05, "loss": 1.3115, "step": 16132 }, { "epoch": 40.3325, "grad_norm": 8.875, "learning_rate": 1.3538535240924317e-05, "loss": 1.3923, "step": 16133 }, { "epoch": 40.335, "grad_norm": 11.3125, "learning_rate": 1.35377779266751e-05, "loss": 1.3285, "step": 16134 }, { "epoch": 40.3375, "grad_norm": 5.84375, "learning_rate": 1.353702058923235e-05, "loss": 1.3308, "step": 16135 }, { "epoch": 40.34, "grad_norm": 7.1875, "learning_rate": 1.3536263228601036e-05, "loss": 1.3939, "step": 16136 }, { "epoch": 40.3425, "grad_norm": 9.0, "learning_rate": 1.3535505844786121e-05, "loss": 1.3452, "step": 16137 }, { "epoch": 40.345, "grad_norm": 6.78125, "learning_rate": 1.3534748437792573e-05, "loss": 1.1991, "step": 16138 }, { "epoch": 40.3475, "grad_norm": 10.25, "learning_rate": 1.3533991007625353e-05, "loss": 1.3962, "step": 16139 }, { "epoch": 40.35, "grad_norm": 7.40625, "learning_rate": 1.3533233554289433e-05, "loss": 1.3291, "step": 16140 }, { "epoch": 40.3525, "grad_norm": 7.84375, "learning_rate": 1.3532476077789772e-05, "loss": 1.2893, "step": 16141 }, { "epoch": 40.355, "grad_norm": 7.6875, "learning_rate": 1.3531718578131342e-05, "loss": 1.3625, "step": 16142 }, { "epoch": 40.3575, "grad_norm": 6.4375, "learning_rate": 1.3530961055319103e-05, "loss": 1.3984, "step": 16143 }, { "epoch": 40.36, "grad_norm": 6.625, "learning_rate": 1.353020350935803e-05, "loss": 1.3219, "step": 16144 }, { "epoch": 40.3625, "grad_norm": 5.625, "learning_rate": 1.3529445940253076e-05, "loss": 1.2538, "step": 16145 }, { "epoch": 40.365, "grad_norm": 7.96875, "learning_rate": 1.3528688348009223e-05, "loss": 1.2897, "step": 16146 }, { "epoch": 40.3675, "grad_norm": 7.3125, "learning_rate": 1.3527930732631427e-05, "loss": 1.2654, "step": 16147 }, { "epoch": 40.37, "grad_norm": 7.15625, "learning_rate": 1.3527173094124659e-05, "loss": 1.2301, "step": 16148 }, { "epoch": 40.3725, "grad_norm": 8.0, "learning_rate": 1.3526415432493885e-05, "loss": 1.4536, "step": 16149 }, { "epoch": 40.375, "grad_norm": 9.5625, "learning_rate": 1.3525657747744073e-05, "loss": 1.4082, "step": 16150 }, { "epoch": 40.3775, "grad_norm": 8.25, "learning_rate": 1.3524900039880188e-05, "loss": 1.481, "step": 16151 }, { "epoch": 40.38, "grad_norm": 7.6875, "learning_rate": 1.3524142308907205e-05, "loss": 1.444, "step": 16152 }, { "epoch": 40.3825, "grad_norm": 4.71875, "learning_rate": 1.3523384554830078e-05, "loss": 1.2249, "step": 16153 }, { "epoch": 40.385, "grad_norm": 7.25, "learning_rate": 1.352262677765379e-05, "loss": 1.4412, "step": 16154 }, { "epoch": 40.3875, "grad_norm": 5.875, "learning_rate": 1.3521868977383297e-05, "loss": 1.3776, "step": 16155 }, { "epoch": 40.39, "grad_norm": 10.3125, "learning_rate": 1.3521111154023573e-05, "loss": 1.484, "step": 16156 }, { "epoch": 40.3925, "grad_norm": 7.40625, "learning_rate": 1.3520353307579588e-05, "loss": 1.3312, "step": 16157 }, { "epoch": 40.395, "grad_norm": 8.9375, "learning_rate": 1.3519595438056306e-05, "loss": 1.3953, "step": 16158 }, { "epoch": 40.3975, "grad_norm": 7.96875, "learning_rate": 1.3518837545458696e-05, "loss": 1.4302, "step": 16159 }, { "epoch": 40.4, "grad_norm": 6.6875, "learning_rate": 1.3518079629791725e-05, "loss": 1.287, "step": 16160 }, { "epoch": 40.4025, "grad_norm": 7.125, "learning_rate": 1.3517321691060372e-05, "loss": 1.3525, "step": 16161 }, { "epoch": 40.405, "grad_norm": 6.875, "learning_rate": 1.3516563729269593e-05, "loss": 1.3356, "step": 16162 }, { "epoch": 40.4075, "grad_norm": 6.78125, "learning_rate": 1.3515805744424366e-05, "loss": 1.3001, "step": 16163 }, { "epoch": 40.41, "grad_norm": 5.1875, "learning_rate": 1.3515047736529656e-05, "loss": 1.4601, "step": 16164 }, { "epoch": 40.4125, "grad_norm": 6.34375, "learning_rate": 1.3514289705590433e-05, "loss": 1.2518, "step": 16165 }, { "epoch": 40.415, "grad_norm": 8.375, "learning_rate": 1.3513531651611669e-05, "loss": 1.4792, "step": 16166 }, { "epoch": 40.4175, "grad_norm": 12.875, "learning_rate": 1.3512773574598332e-05, "loss": 1.3256, "step": 16167 }, { "epoch": 40.42, "grad_norm": 7.75, "learning_rate": 1.3512015474555389e-05, "loss": 1.3493, "step": 16168 }, { "epoch": 40.4225, "grad_norm": 9.4375, "learning_rate": 1.3511257351487819e-05, "loss": 1.3723, "step": 16169 }, { "epoch": 40.425, "grad_norm": 6.9375, "learning_rate": 1.351049920540058e-05, "loss": 1.3098, "step": 16170 }, { "epoch": 40.4275, "grad_norm": 6.34375, "learning_rate": 1.3509741036298654e-05, "loss": 1.3727, "step": 16171 }, { "epoch": 40.43, "grad_norm": 6.5625, "learning_rate": 1.3508982844187003e-05, "loss": 1.3873, "step": 16172 }, { "epoch": 40.4325, "grad_norm": 8.75, "learning_rate": 1.35082246290706e-05, "loss": 1.2241, "step": 16173 }, { "epoch": 40.435, "grad_norm": 9.8125, "learning_rate": 1.350746639095442e-05, "loss": 1.4038, "step": 16174 }, { "epoch": 40.4375, "grad_norm": 5.9375, "learning_rate": 1.3506708129843426e-05, "loss": 1.312, "step": 16175 }, { "epoch": 40.44, "grad_norm": 6.625, "learning_rate": 1.3505949845742599e-05, "loss": 1.377, "step": 16176 }, { "epoch": 40.4425, "grad_norm": 5.46875, "learning_rate": 1.35051915386569e-05, "loss": 1.137, "step": 16177 }, { "epoch": 40.445, "grad_norm": 8.1875, "learning_rate": 1.350443320859131e-05, "loss": 1.3799, "step": 16178 }, { "epoch": 40.4475, "grad_norm": 8.125, "learning_rate": 1.3503674855550795e-05, "loss": 1.5111, "step": 16179 }, { "epoch": 40.45, "grad_norm": 7.0625, "learning_rate": 1.3502916479540327e-05, "loss": 1.1903, "step": 16180 }, { "epoch": 40.4525, "grad_norm": 6.75, "learning_rate": 1.3502158080564879e-05, "loss": 1.1359, "step": 16181 }, { "epoch": 40.455, "grad_norm": 7.625, "learning_rate": 1.3501399658629421e-05, "loss": 1.3114, "step": 16182 }, { "epoch": 40.4575, "grad_norm": 8.375, "learning_rate": 1.350064121373893e-05, "loss": 1.283, "step": 16183 }, { "epoch": 40.46, "grad_norm": 6.8125, "learning_rate": 1.3499882745898376e-05, "loss": 1.284, "step": 16184 }, { "epoch": 40.4625, "grad_norm": 9.4375, "learning_rate": 1.3499124255112727e-05, "loss": 1.4973, "step": 16185 }, { "epoch": 40.465, "grad_norm": 6.46875, "learning_rate": 1.3498365741386963e-05, "loss": 1.3806, "step": 16186 }, { "epoch": 40.4675, "grad_norm": 6.21875, "learning_rate": 1.3497607204726049e-05, "loss": 1.2426, "step": 16187 }, { "epoch": 40.47, "grad_norm": 8.375, "learning_rate": 1.3496848645134967e-05, "loss": 1.2789, "step": 16188 }, { "epoch": 40.4725, "grad_norm": 9.125, "learning_rate": 1.3496090062618683e-05, "loss": 1.226, "step": 16189 }, { "epoch": 40.475, "grad_norm": 11.875, "learning_rate": 1.3495331457182173e-05, "loss": 1.3375, "step": 16190 }, { "epoch": 40.4775, "grad_norm": 7.21875, "learning_rate": 1.3494572828830409e-05, "loss": 1.2542, "step": 16191 }, { "epoch": 40.48, "grad_norm": 9.25, "learning_rate": 1.3493814177568365e-05, "loss": 1.392, "step": 16192 }, { "epoch": 40.4825, "grad_norm": 8.5625, "learning_rate": 1.3493055503401013e-05, "loss": 1.2149, "step": 16193 }, { "epoch": 40.485, "grad_norm": 8.4375, "learning_rate": 1.3492296806333333e-05, "loss": 1.331, "step": 16194 }, { "epoch": 40.4875, "grad_norm": 6.6875, "learning_rate": 1.3491538086370293e-05, "loss": 1.4169, "step": 16195 }, { "epoch": 40.49, "grad_norm": 6.8125, "learning_rate": 1.349077934351687e-05, "loss": 1.228, "step": 16196 }, { "epoch": 40.4925, "grad_norm": 7.46875, "learning_rate": 1.3490020577778035e-05, "loss": 1.2926, "step": 16197 }, { "epoch": 40.495, "grad_norm": 6.65625, "learning_rate": 1.3489261789158767e-05, "loss": 1.2536, "step": 16198 }, { "epoch": 40.4975, "grad_norm": 6.46875, "learning_rate": 1.3488502977664039e-05, "loss": 1.358, "step": 16199 }, { "epoch": 40.5, "grad_norm": 7.40625, "learning_rate": 1.3487744143298822e-05, "loss": 1.5971, "step": 16200 }, { "epoch": 40.5025, "grad_norm": 7.65625, "learning_rate": 1.3486985286068099e-05, "loss": 1.5241, "step": 16201 }, { "epoch": 40.505, "grad_norm": 6.34375, "learning_rate": 1.3486226405976835e-05, "loss": 1.2773, "step": 16202 }, { "epoch": 40.5075, "grad_norm": 7.84375, "learning_rate": 1.3485467503030014e-05, "loss": 1.362, "step": 16203 }, { "epoch": 40.51, "grad_norm": 10.0, "learning_rate": 1.3484708577232606e-05, "loss": 1.3885, "step": 16204 }, { "epoch": 40.5125, "grad_norm": 9.25, "learning_rate": 1.3483949628589588e-05, "loss": 1.4208, "step": 16205 }, { "epoch": 40.515, "grad_norm": 8.125, "learning_rate": 1.3483190657105934e-05, "loss": 1.3232, "step": 16206 }, { "epoch": 40.5175, "grad_norm": 7.625, "learning_rate": 1.3482431662786621e-05, "loss": 1.3933, "step": 16207 }, { "epoch": 40.52, "grad_norm": 6.78125, "learning_rate": 1.3481672645636627e-05, "loss": 1.4419, "step": 16208 }, { "epoch": 40.5225, "grad_norm": 7.5625, "learning_rate": 1.3480913605660928e-05, "loss": 1.4283, "step": 16209 }, { "epoch": 40.525, "grad_norm": 9.6875, "learning_rate": 1.3480154542864496e-05, "loss": 1.2611, "step": 16210 }, { "epoch": 40.5275, "grad_norm": 6.53125, "learning_rate": 1.3479395457252312e-05, "loss": 1.3624, "step": 16211 }, { "epoch": 40.53, "grad_norm": 7.03125, "learning_rate": 1.347863634882935e-05, "loss": 1.2659, "step": 16212 }, { "epoch": 40.5325, "grad_norm": 5.8125, "learning_rate": 1.3477877217600588e-05, "loss": 1.2854, "step": 16213 }, { "epoch": 40.535, "grad_norm": 7.90625, "learning_rate": 1.3477118063571003e-05, "loss": 1.2878, "step": 16214 }, { "epoch": 40.5375, "grad_norm": 5.625, "learning_rate": 1.3476358886745566e-05, "loss": 1.2977, "step": 16215 }, { "epoch": 40.54, "grad_norm": 10.1875, "learning_rate": 1.3475599687129265e-05, "loss": 1.3307, "step": 16216 }, { "epoch": 40.5425, "grad_norm": 8.9375, "learning_rate": 1.3474840464727069e-05, "loss": 1.4036, "step": 16217 }, { "epoch": 40.545, "grad_norm": 7.25, "learning_rate": 1.347408121954396e-05, "loss": 1.3238, "step": 16218 }, { "epoch": 40.5475, "grad_norm": 7.125, "learning_rate": 1.347332195158491e-05, "loss": 1.2419, "step": 16219 }, { "epoch": 40.55, "grad_norm": 4.96875, "learning_rate": 1.3472562660854902e-05, "loss": 1.3284, "step": 16220 }, { "epoch": 40.5525, "grad_norm": 8.0625, "learning_rate": 1.3471803347358913e-05, "loss": 1.4484, "step": 16221 }, { "epoch": 40.555, "grad_norm": 7.46875, "learning_rate": 1.347104401110192e-05, "loss": 1.4231, "step": 16222 }, { "epoch": 40.5575, "grad_norm": 9.125, "learning_rate": 1.34702846520889e-05, "loss": 1.3613, "step": 16223 }, { "epoch": 40.56, "grad_norm": 8.625, "learning_rate": 1.3469525270324835e-05, "loss": 1.3428, "step": 16224 }, { "epoch": 40.5625, "grad_norm": 8.125, "learning_rate": 1.3468765865814696e-05, "loss": 1.4881, "step": 16225 }, { "epoch": 40.565, "grad_norm": 7.125, "learning_rate": 1.3468006438563474e-05, "loss": 1.2195, "step": 16226 }, { "epoch": 40.5675, "grad_norm": 7.375, "learning_rate": 1.3467246988576135e-05, "loss": 1.2141, "step": 16227 }, { "epoch": 40.57, "grad_norm": 8.375, "learning_rate": 1.3466487515857668e-05, "loss": 1.3679, "step": 16228 }, { "epoch": 40.5725, "grad_norm": 5.875, "learning_rate": 1.3465728020413046e-05, "loss": 1.4744, "step": 16229 }, { "epoch": 40.575, "grad_norm": 6.6875, "learning_rate": 1.3464968502247246e-05, "loss": 1.4082, "step": 16230 }, { "epoch": 40.5775, "grad_norm": 9.25, "learning_rate": 1.3464208961365255e-05, "loss": 1.3577, "step": 16231 }, { "epoch": 40.58, "grad_norm": 7.34375, "learning_rate": 1.3463449397772045e-05, "loss": 1.518, "step": 16232 }, { "epoch": 40.5825, "grad_norm": 8.1875, "learning_rate": 1.3462689811472604e-05, "loss": 1.3269, "step": 16233 }, { "epoch": 40.585, "grad_norm": 11.0625, "learning_rate": 1.3461930202471907e-05, "loss": 1.3305, "step": 16234 }, { "epoch": 40.5875, "grad_norm": 7.96875, "learning_rate": 1.346117057077493e-05, "loss": 1.4574, "step": 16235 }, { "epoch": 40.59, "grad_norm": 6.46875, "learning_rate": 1.3460410916386661e-05, "loss": 1.2897, "step": 16236 }, { "epoch": 40.5925, "grad_norm": 8.0, "learning_rate": 1.3459651239312075e-05, "loss": 1.3884, "step": 16237 }, { "epoch": 40.595, "grad_norm": 6.34375, "learning_rate": 1.3458891539556153e-05, "loss": 1.3288, "step": 16238 }, { "epoch": 40.5975, "grad_norm": 6.15625, "learning_rate": 1.345813181712388e-05, "loss": 1.289, "step": 16239 }, { "epoch": 40.6, "grad_norm": 9.6875, "learning_rate": 1.345737207202023e-05, "loss": 1.3679, "step": 16240 }, { "epoch": 40.6025, "grad_norm": 7.25, "learning_rate": 1.345661230425019e-05, "loss": 1.3096, "step": 16241 }, { "epoch": 40.605, "grad_norm": 9.625, "learning_rate": 1.3455852513818732e-05, "loss": 1.2349, "step": 16242 }, { "epoch": 40.6075, "grad_norm": 8.0625, "learning_rate": 1.345509270073085e-05, "loss": 1.5261, "step": 16243 }, { "epoch": 40.61, "grad_norm": 8.25, "learning_rate": 1.3454332864991515e-05, "loss": 1.484, "step": 16244 }, { "epoch": 40.6125, "grad_norm": 5.28125, "learning_rate": 1.345357300660571e-05, "loss": 1.3626, "step": 16245 }, { "epoch": 40.615, "grad_norm": 5.59375, "learning_rate": 1.3452813125578423e-05, "loss": 1.2278, "step": 16246 }, { "epoch": 40.6175, "grad_norm": 7.1875, "learning_rate": 1.3452053221914629e-05, "loss": 1.2747, "step": 16247 }, { "epoch": 40.62, "grad_norm": 6.84375, "learning_rate": 1.3451293295619312e-05, "loss": 1.3739, "step": 16248 }, { "epoch": 40.6225, "grad_norm": 9.6875, "learning_rate": 1.3450533346697456e-05, "loss": 1.4011, "step": 16249 }, { "epoch": 40.625, "grad_norm": 6.78125, "learning_rate": 1.344977337515404e-05, "loss": 1.416, "step": 16250 }, { "epoch": 40.6275, "grad_norm": 7.09375, "learning_rate": 1.3449013380994048e-05, "loss": 1.3214, "step": 16251 }, { "epoch": 40.63, "grad_norm": 8.0, "learning_rate": 1.3448253364222462e-05, "loss": 1.2787, "step": 16252 }, { "epoch": 40.6325, "grad_norm": 5.28125, "learning_rate": 1.3447493324844264e-05, "loss": 1.3205, "step": 16253 }, { "epoch": 40.635, "grad_norm": 6.375, "learning_rate": 1.3446733262864444e-05, "loss": 1.2709, "step": 16254 }, { "epoch": 40.6375, "grad_norm": 7.5625, "learning_rate": 1.344597317828797e-05, "loss": 1.3309, "step": 16255 }, { "epoch": 40.64, "grad_norm": 7.5625, "learning_rate": 1.3445213071119841e-05, "loss": 1.2314, "step": 16256 }, { "epoch": 40.6425, "grad_norm": 9.1875, "learning_rate": 1.3444452941365026e-05, "loss": 1.4653, "step": 16257 }, { "epoch": 40.645, "grad_norm": 10.25, "learning_rate": 1.3443692789028522e-05, "loss": 1.3123, "step": 16258 }, { "epoch": 40.6475, "grad_norm": 6.28125, "learning_rate": 1.3442932614115303e-05, "loss": 1.2593, "step": 16259 }, { "epoch": 40.65, "grad_norm": 9.0625, "learning_rate": 1.3442172416630355e-05, "loss": 1.3503, "step": 16260 }, { "epoch": 40.6525, "grad_norm": 9.0, "learning_rate": 1.3441412196578662e-05, "loss": 1.3068, "step": 16261 }, { "epoch": 40.655, "grad_norm": 8.4375, "learning_rate": 1.344065195396521e-05, "loss": 1.4778, "step": 16262 }, { "epoch": 40.6575, "grad_norm": 6.96875, "learning_rate": 1.3439891688794979e-05, "loss": 1.3272, "step": 16263 }, { "epoch": 40.66, "grad_norm": 8.6875, "learning_rate": 1.3439131401072958e-05, "loss": 1.5228, "step": 16264 }, { "epoch": 40.6625, "grad_norm": 7.8125, "learning_rate": 1.3438371090804128e-05, "loss": 1.316, "step": 16265 }, { "epoch": 40.665, "grad_norm": 6.15625, "learning_rate": 1.3437610757993476e-05, "loss": 1.3399, "step": 16266 }, { "epoch": 40.6675, "grad_norm": 5.3125, "learning_rate": 1.3436850402645982e-05, "loss": 1.2427, "step": 16267 }, { "epoch": 40.67, "grad_norm": 7.78125, "learning_rate": 1.3436090024766638e-05, "loss": 1.299, "step": 16268 }, { "epoch": 40.6725, "grad_norm": 5.84375, "learning_rate": 1.3435329624360428e-05, "loss": 1.3601, "step": 16269 }, { "epoch": 40.675, "grad_norm": 6.1875, "learning_rate": 1.3434569201432326e-05, "loss": 1.2779, "step": 16270 }, { "epoch": 40.6775, "grad_norm": 8.9375, "learning_rate": 1.3433808755987334e-05, "loss": 1.4594, "step": 16271 }, { "epoch": 40.68, "grad_norm": 10.4375, "learning_rate": 1.3433048288030424e-05, "loss": 1.401, "step": 16272 }, { "epoch": 40.6825, "grad_norm": 7.96875, "learning_rate": 1.343228779756659e-05, "loss": 1.3826, "step": 16273 }, { "epoch": 40.685, "grad_norm": 9.375, "learning_rate": 1.3431527284600811e-05, "loss": 1.4935, "step": 16274 }, { "epoch": 40.6875, "grad_norm": 7.71875, "learning_rate": 1.343076674913808e-05, "loss": 1.2536, "step": 16275 }, { "epoch": 40.69, "grad_norm": 7.15625, "learning_rate": 1.3430006191183378e-05, "loss": 1.4082, "step": 16276 }, { "epoch": 40.6925, "grad_norm": 6.0, "learning_rate": 1.3429245610741692e-05, "loss": 1.441, "step": 16277 }, { "epoch": 40.695, "grad_norm": 8.8125, "learning_rate": 1.342848500781801e-05, "loss": 1.3305, "step": 16278 }, { "epoch": 40.6975, "grad_norm": 8.25, "learning_rate": 1.3427724382417316e-05, "loss": 1.3636, "step": 16279 }, { "epoch": 40.7, "grad_norm": 5.40625, "learning_rate": 1.3426963734544601e-05, "loss": 1.2342, "step": 16280 }, { "epoch": 40.7025, "grad_norm": 8.5, "learning_rate": 1.3426203064204845e-05, "loss": 1.4855, "step": 16281 }, { "epoch": 40.705, "grad_norm": 7.9375, "learning_rate": 1.3425442371403041e-05, "loss": 1.3885, "step": 16282 }, { "epoch": 40.7075, "grad_norm": 8.6875, "learning_rate": 1.3424681656144174e-05, "loss": 1.3141, "step": 16283 }, { "epoch": 40.71, "grad_norm": 5.21875, "learning_rate": 1.3423920918433234e-05, "loss": 1.2785, "step": 16284 }, { "epoch": 40.7125, "grad_norm": 8.4375, "learning_rate": 1.3423160158275199e-05, "loss": 1.5067, "step": 16285 }, { "epoch": 40.715, "grad_norm": 7.875, "learning_rate": 1.3422399375675068e-05, "loss": 1.4542, "step": 16286 }, { "epoch": 40.7175, "grad_norm": 7.0, "learning_rate": 1.342163857063782e-05, "loss": 1.4251, "step": 16287 }, { "epoch": 40.72, "grad_norm": 8.5, "learning_rate": 1.342087774316845e-05, "loss": 1.2284, "step": 16288 }, { "epoch": 40.7225, "grad_norm": 5.3125, "learning_rate": 1.3420116893271941e-05, "loss": 1.3244, "step": 16289 }, { "epoch": 40.725, "grad_norm": 9.125, "learning_rate": 1.3419356020953283e-05, "loss": 1.255, "step": 16290 }, { "epoch": 40.7275, "grad_norm": 7.8125, "learning_rate": 1.3418595126217464e-05, "loss": 1.2807, "step": 16291 }, { "epoch": 40.73, "grad_norm": 7.71875, "learning_rate": 1.3417834209069472e-05, "loss": 1.3498, "step": 16292 }, { "epoch": 40.7325, "grad_norm": 5.5625, "learning_rate": 1.3417073269514293e-05, "loss": 1.4507, "step": 16293 }, { "epoch": 40.735, "grad_norm": 8.8125, "learning_rate": 1.3416312307556923e-05, "loss": 1.468, "step": 16294 }, { "epoch": 40.7375, "grad_norm": 9.0625, "learning_rate": 1.341555132320234e-05, "loss": 1.3556, "step": 16295 }, { "epoch": 40.74, "grad_norm": 7.875, "learning_rate": 1.3414790316455546e-05, "loss": 1.3364, "step": 16296 }, { "epoch": 40.7425, "grad_norm": 8.0, "learning_rate": 1.3414029287321517e-05, "loss": 1.4398, "step": 16297 }, { "epoch": 40.745, "grad_norm": 6.59375, "learning_rate": 1.3413268235805253e-05, "loss": 1.2565, "step": 16298 }, { "epoch": 40.7475, "grad_norm": 6.34375, "learning_rate": 1.3412507161911737e-05, "loss": 1.2729, "step": 16299 }, { "epoch": 40.75, "grad_norm": 8.75, "learning_rate": 1.3411746065645961e-05, "loss": 1.3052, "step": 16300 }, { "epoch": 40.7525, "grad_norm": 11.0625, "learning_rate": 1.3410984947012917e-05, "loss": 1.4906, "step": 16301 }, { "epoch": 40.755, "grad_norm": 7.34375, "learning_rate": 1.3410223806017588e-05, "loss": 1.2442, "step": 16302 }, { "epoch": 40.7575, "grad_norm": 5.46875, "learning_rate": 1.3409462642664972e-05, "loss": 1.3428, "step": 16303 }, { "epoch": 40.76, "grad_norm": 6.59375, "learning_rate": 1.3408701456960052e-05, "loss": 1.1803, "step": 16304 }, { "epoch": 40.7625, "grad_norm": 12.875, "learning_rate": 1.3407940248907825e-05, "loss": 1.392, "step": 16305 }, { "epoch": 40.765, "grad_norm": 7.8125, "learning_rate": 1.3407179018513277e-05, "loss": 1.2806, "step": 16306 }, { "epoch": 40.7675, "grad_norm": 8.5625, "learning_rate": 1.3406417765781399e-05, "loss": 1.328, "step": 16307 }, { "epoch": 40.77, "grad_norm": 6.0625, "learning_rate": 1.340565649071718e-05, "loss": 1.2051, "step": 16308 }, { "epoch": 40.7725, "grad_norm": 8.9375, "learning_rate": 1.3404895193325618e-05, "loss": 1.457, "step": 16309 }, { "epoch": 40.775, "grad_norm": 5.125, "learning_rate": 1.3404133873611696e-05, "loss": 1.318, "step": 16310 }, { "epoch": 40.7775, "grad_norm": 5.96875, "learning_rate": 1.3403372531580411e-05, "loss": 1.1742, "step": 16311 }, { "epoch": 40.78, "grad_norm": 10.4375, "learning_rate": 1.3402611167236748e-05, "loss": 1.4191, "step": 16312 }, { "epoch": 40.7825, "grad_norm": 5.1875, "learning_rate": 1.3401849780585708e-05, "loss": 1.4668, "step": 16313 }, { "epoch": 40.785, "grad_norm": 9.125, "learning_rate": 1.3401088371632272e-05, "loss": 1.3394, "step": 16314 }, { "epoch": 40.7875, "grad_norm": 11.125, "learning_rate": 1.340032694038144e-05, "loss": 1.1746, "step": 16315 }, { "epoch": 40.79, "grad_norm": 8.3125, "learning_rate": 1.33995654868382e-05, "loss": 1.384, "step": 16316 }, { "epoch": 40.7925, "grad_norm": 6.96875, "learning_rate": 1.3398804011007543e-05, "loss": 1.263, "step": 16317 }, { "epoch": 40.795, "grad_norm": 10.9375, "learning_rate": 1.3398042512894465e-05, "loss": 1.4612, "step": 16318 }, { "epoch": 40.7975, "grad_norm": 7.21875, "learning_rate": 1.3397280992503954e-05, "loss": 1.4684, "step": 16319 }, { "epoch": 40.8, "grad_norm": 6.25, "learning_rate": 1.3396519449841006e-05, "loss": 1.4754, "step": 16320 }, { "epoch": 40.8025, "grad_norm": 7.4375, "learning_rate": 1.3395757884910611e-05, "loss": 1.327, "step": 16321 }, { "epoch": 40.805, "grad_norm": 7.6875, "learning_rate": 1.3394996297717766e-05, "loss": 1.354, "step": 16322 }, { "epoch": 40.8075, "grad_norm": 6.0, "learning_rate": 1.3394234688267461e-05, "loss": 1.276, "step": 16323 }, { "epoch": 40.81, "grad_norm": 9.375, "learning_rate": 1.3393473056564685e-05, "loss": 1.3738, "step": 16324 }, { "epoch": 40.8125, "grad_norm": 14.0, "learning_rate": 1.339271140261444e-05, "loss": 1.2733, "step": 16325 }, { "epoch": 40.815, "grad_norm": 6.3125, "learning_rate": 1.3391949726421714e-05, "loss": 1.1855, "step": 16326 }, { "epoch": 40.8175, "grad_norm": 10.8125, "learning_rate": 1.33911880279915e-05, "loss": 1.5618, "step": 16327 }, { "epoch": 40.82, "grad_norm": 5.34375, "learning_rate": 1.3390426307328794e-05, "loss": 1.388, "step": 16328 }, { "epoch": 40.8225, "grad_norm": 9.125, "learning_rate": 1.3389664564438585e-05, "loss": 1.3646, "step": 16329 }, { "epoch": 40.825, "grad_norm": 6.8125, "learning_rate": 1.3388902799325875e-05, "loss": 1.3411, "step": 16330 }, { "epoch": 40.8275, "grad_norm": 8.8125, "learning_rate": 1.3388141011995654e-05, "loss": 1.3533, "step": 16331 }, { "epoch": 40.83, "grad_norm": 6.6875, "learning_rate": 1.3387379202452917e-05, "loss": 1.4495, "step": 16332 }, { "epoch": 40.8325, "grad_norm": 7.875, "learning_rate": 1.3386617370702654e-05, "loss": 1.4055, "step": 16333 }, { "epoch": 40.835, "grad_norm": 7.5, "learning_rate": 1.3385855516749863e-05, "loss": 1.2749, "step": 16334 }, { "epoch": 40.8375, "grad_norm": 8.25, "learning_rate": 1.3385093640599543e-05, "loss": 1.2583, "step": 16335 }, { "epoch": 40.84, "grad_norm": 12.25, "learning_rate": 1.338433174225668e-05, "loss": 1.4011, "step": 16336 }, { "epoch": 40.8425, "grad_norm": 10.3125, "learning_rate": 1.3383569821726276e-05, "loss": 1.3944, "step": 16337 }, { "epoch": 40.845, "grad_norm": 9.375, "learning_rate": 1.3382807879013322e-05, "loss": 1.5385, "step": 16338 }, { "epoch": 40.8475, "grad_norm": 7.53125, "learning_rate": 1.3382045914122817e-05, "loss": 1.2864, "step": 16339 }, { "epoch": 40.85, "grad_norm": 6.0625, "learning_rate": 1.3381283927059751e-05, "loss": 1.4183, "step": 16340 }, { "epoch": 40.8525, "grad_norm": 8.4375, "learning_rate": 1.338052191782913e-05, "loss": 1.2167, "step": 16341 }, { "epoch": 40.855, "grad_norm": 6.5625, "learning_rate": 1.3379759886435935e-05, "loss": 1.2256, "step": 16342 }, { "epoch": 40.8575, "grad_norm": 8.875, "learning_rate": 1.3378997832885175e-05, "loss": 1.2054, "step": 16343 }, { "epoch": 40.86, "grad_norm": 6.75, "learning_rate": 1.3378235757181834e-05, "loss": 1.5865, "step": 16344 }, { "epoch": 40.8625, "grad_norm": 6.84375, "learning_rate": 1.337747365933092e-05, "loss": 1.3625, "step": 16345 }, { "epoch": 40.865, "grad_norm": 5.65625, "learning_rate": 1.3376711539337421e-05, "loss": 1.3652, "step": 16346 }, { "epoch": 40.8675, "grad_norm": 6.03125, "learning_rate": 1.3375949397206339e-05, "loss": 1.409, "step": 16347 }, { "epoch": 40.87, "grad_norm": 7.9375, "learning_rate": 1.3375187232942666e-05, "loss": 1.335, "step": 16348 }, { "epoch": 40.8725, "grad_norm": 8.625, "learning_rate": 1.33744250465514e-05, "loss": 1.2954, "step": 16349 }, { "epoch": 40.875, "grad_norm": 4.90625, "learning_rate": 1.3373662838037538e-05, "loss": 1.2931, "step": 16350 }, { "epoch": 40.8775, "grad_norm": 9.25, "learning_rate": 1.337290060740608e-05, "loss": 1.5326, "step": 16351 }, { "epoch": 40.88, "grad_norm": 6.9375, "learning_rate": 1.3372138354662018e-05, "loss": 1.3138, "step": 16352 }, { "epoch": 40.8825, "grad_norm": 7.5625, "learning_rate": 1.3371376079810354e-05, "loss": 1.3082, "step": 16353 }, { "epoch": 40.885, "grad_norm": 6.90625, "learning_rate": 1.337061378285608e-05, "loss": 1.1873, "step": 16354 }, { "epoch": 40.8875, "grad_norm": 7.0, "learning_rate": 1.3369851463804201e-05, "loss": 1.2253, "step": 16355 }, { "epoch": 40.89, "grad_norm": 6.03125, "learning_rate": 1.336908912265971e-05, "loss": 1.3686, "step": 16356 }, { "epoch": 40.8925, "grad_norm": 7.78125, "learning_rate": 1.3368326759427604e-05, "loss": 1.2113, "step": 16357 }, { "epoch": 40.895, "grad_norm": 5.6875, "learning_rate": 1.3367564374112884e-05, "loss": 1.2809, "step": 16358 }, { "epoch": 40.8975, "grad_norm": 6.5625, "learning_rate": 1.3366801966720544e-05, "loss": 1.366, "step": 16359 }, { "epoch": 40.9, "grad_norm": 7.9375, "learning_rate": 1.3366039537255589e-05, "loss": 1.4079, "step": 16360 }, { "epoch": 40.9025, "grad_norm": 5.625, "learning_rate": 1.3365277085723012e-05, "loss": 1.2492, "step": 16361 }, { "epoch": 40.905, "grad_norm": 8.4375, "learning_rate": 1.3364514612127813e-05, "loss": 1.1772, "step": 16362 }, { "epoch": 40.9075, "grad_norm": 7.375, "learning_rate": 1.3363752116474988e-05, "loss": 1.1466, "step": 16363 }, { "epoch": 40.91, "grad_norm": 8.5, "learning_rate": 1.3362989598769543e-05, "loss": 1.2249, "step": 16364 }, { "epoch": 40.9125, "grad_norm": 8.0625, "learning_rate": 1.3362227059016468e-05, "loss": 1.2205, "step": 16365 }, { "epoch": 40.915, "grad_norm": 7.0, "learning_rate": 1.3361464497220773e-05, "loss": 1.4177, "step": 16366 }, { "epoch": 40.9175, "grad_norm": 6.59375, "learning_rate": 1.3360701913387447e-05, "loss": 1.241, "step": 16367 }, { "epoch": 40.92, "grad_norm": 8.6875, "learning_rate": 1.3359939307521494e-05, "loss": 1.3601, "step": 16368 }, { "epoch": 40.9225, "grad_norm": 7.40625, "learning_rate": 1.3359176679627912e-05, "loss": 1.3541, "step": 16369 }, { "epoch": 40.925, "grad_norm": 6.6875, "learning_rate": 1.3358414029711706e-05, "loss": 1.3464, "step": 16370 }, { "epoch": 40.9275, "grad_norm": 6.40625, "learning_rate": 1.3357651357777872e-05, "loss": 1.2781, "step": 16371 }, { "epoch": 40.93, "grad_norm": 8.6875, "learning_rate": 1.3356888663831406e-05, "loss": 1.4552, "step": 16372 }, { "epoch": 40.9325, "grad_norm": 8.25, "learning_rate": 1.3356125947877316e-05, "loss": 1.3825, "step": 16373 }, { "epoch": 40.935, "grad_norm": 6.6875, "learning_rate": 1.3355363209920595e-05, "loss": 1.3575, "step": 16374 }, { "epoch": 40.9375, "grad_norm": 9.125, "learning_rate": 1.335460044996625e-05, "loss": 1.3095, "step": 16375 }, { "epoch": 40.94, "grad_norm": 7.03125, "learning_rate": 1.335383766801928e-05, "loss": 1.2556, "step": 16376 }, { "epoch": 40.9425, "grad_norm": 6.84375, "learning_rate": 1.335307486408468e-05, "loss": 1.2977, "step": 16377 }, { "epoch": 40.945, "grad_norm": 8.625, "learning_rate": 1.3352312038167459e-05, "loss": 1.3869, "step": 16378 }, { "epoch": 40.9475, "grad_norm": 6.53125, "learning_rate": 1.3351549190272612e-05, "loss": 1.1852, "step": 16379 }, { "epoch": 40.95, "grad_norm": 5.46875, "learning_rate": 1.3350786320405145e-05, "loss": 1.3581, "step": 16380 }, { "epoch": 40.9525, "grad_norm": 11.3125, "learning_rate": 1.3350023428570057e-05, "loss": 1.3386, "step": 16381 }, { "epoch": 40.955, "grad_norm": 9.1875, "learning_rate": 1.3349260514772346e-05, "loss": 1.368, "step": 16382 }, { "epoch": 40.9575, "grad_norm": 5.5625, "learning_rate": 1.3348497579017021e-05, "loss": 1.2335, "step": 16383 }, { "epoch": 40.96, "grad_norm": 7.9375, "learning_rate": 1.3347734621309076e-05, "loss": 1.316, "step": 16384 }, { "epoch": 40.9625, "grad_norm": 8.25, "learning_rate": 1.3346971641653523e-05, "loss": 1.3627, "step": 16385 }, { "epoch": 40.965, "grad_norm": 8.5625, "learning_rate": 1.3346208640055354e-05, "loss": 1.4453, "step": 16386 }, { "epoch": 40.9675, "grad_norm": 5.84375, "learning_rate": 1.3345445616519574e-05, "loss": 1.4362, "step": 16387 }, { "epoch": 40.97, "grad_norm": 4.75, "learning_rate": 1.3344682571051187e-05, "loss": 1.3417, "step": 16388 }, { "epoch": 40.9725, "grad_norm": 7.1875, "learning_rate": 1.3343919503655194e-05, "loss": 1.3733, "step": 16389 }, { "epoch": 40.975, "grad_norm": 10.5, "learning_rate": 1.3343156414336599e-05, "loss": 1.3675, "step": 16390 }, { "epoch": 40.9775, "grad_norm": 7.625, "learning_rate": 1.3342393303100407e-05, "loss": 1.1704, "step": 16391 }, { "epoch": 40.98, "grad_norm": 6.375, "learning_rate": 1.3341630169951616e-05, "loss": 1.3898, "step": 16392 }, { "epoch": 40.9825, "grad_norm": 6.1875, "learning_rate": 1.3340867014895231e-05, "loss": 1.3661, "step": 16393 }, { "epoch": 40.985, "grad_norm": 7.5, "learning_rate": 1.3340103837936255e-05, "loss": 1.5433, "step": 16394 }, { "epoch": 40.9875, "grad_norm": 12.3125, "learning_rate": 1.3339340639079695e-05, "loss": 1.4006, "step": 16395 }, { "epoch": 40.99, "grad_norm": 7.625, "learning_rate": 1.3338577418330552e-05, "loss": 1.2817, "step": 16396 }, { "epoch": 40.9925, "grad_norm": 7.21875, "learning_rate": 1.3337814175693823e-05, "loss": 1.3433, "step": 16397 }, { "epoch": 40.995, "grad_norm": 6.96875, "learning_rate": 1.3337050911174525e-05, "loss": 1.4562, "step": 16398 }, { "epoch": 40.9975, "grad_norm": 6.8125, "learning_rate": 1.3336287624777649e-05, "loss": 1.3405, "step": 16399 }, { "epoch": 41.0, "grad_norm": 16.25, "learning_rate": 1.3335524316508208e-05, "loss": 1.1818, "step": 16400 }, { "epoch": 41.0025, "grad_norm": 7.25, "learning_rate": 1.3334760986371202e-05, "loss": 1.2412, "step": 16401 }, { "epoch": 41.005, "grad_norm": 7.125, "learning_rate": 1.3333997634371638e-05, "loss": 1.4039, "step": 16402 }, { "epoch": 41.0075, "grad_norm": 7.34375, "learning_rate": 1.3333234260514516e-05, "loss": 1.1583, "step": 16403 }, { "epoch": 41.01, "grad_norm": 4.34375, "learning_rate": 1.3332470864804845e-05, "loss": 1.3965, "step": 16404 }, { "epoch": 41.0125, "grad_norm": 5.625, "learning_rate": 1.3331707447247629e-05, "loss": 1.3179, "step": 16405 }, { "epoch": 41.015, "grad_norm": 7.3125, "learning_rate": 1.3330944007847872e-05, "loss": 1.422, "step": 16406 }, { "epoch": 41.0175, "grad_norm": 6.875, "learning_rate": 1.3330180546610578e-05, "loss": 1.3566, "step": 16407 }, { "epoch": 41.02, "grad_norm": 5.9375, "learning_rate": 1.3329417063540756e-05, "loss": 1.3651, "step": 16408 }, { "epoch": 41.0225, "grad_norm": 7.375, "learning_rate": 1.3328653558643407e-05, "loss": 1.2933, "step": 16409 }, { "epoch": 41.025, "grad_norm": 6.5, "learning_rate": 1.3327890031923538e-05, "loss": 1.4958, "step": 16410 }, { "epoch": 41.0275, "grad_norm": 7.0625, "learning_rate": 1.332712648338616e-05, "loss": 1.4009, "step": 16411 }, { "epoch": 41.03, "grad_norm": 8.3125, "learning_rate": 1.3326362913036266e-05, "loss": 1.1815, "step": 16412 }, { "epoch": 41.0325, "grad_norm": 10.0, "learning_rate": 1.3325599320878876e-05, "loss": 1.4886, "step": 16413 }, { "epoch": 41.035, "grad_norm": 8.5, "learning_rate": 1.3324835706918985e-05, "loss": 1.2606, "step": 16414 }, { "epoch": 41.0375, "grad_norm": 7.875, "learning_rate": 1.3324072071161611e-05, "loss": 1.2844, "step": 16415 }, { "epoch": 41.04, "grad_norm": 8.125, "learning_rate": 1.3323308413611748e-05, "loss": 1.3957, "step": 16416 }, { "epoch": 41.0425, "grad_norm": 7.96875, "learning_rate": 1.3322544734274412e-05, "loss": 1.3218, "step": 16417 }, { "epoch": 41.045, "grad_norm": 7.125, "learning_rate": 1.3321781033154602e-05, "loss": 1.289, "step": 16418 }, { "epoch": 41.0475, "grad_norm": 6.4375, "learning_rate": 1.332101731025733e-05, "loss": 1.345, "step": 16419 }, { "epoch": 41.05, "grad_norm": 8.625, "learning_rate": 1.3320253565587602e-05, "loss": 1.3933, "step": 16420 }, { "epoch": 41.0525, "grad_norm": 6.75, "learning_rate": 1.3319489799150422e-05, "loss": 1.4284, "step": 16421 }, { "epoch": 41.055, "grad_norm": 4.9375, "learning_rate": 1.3318726010950802e-05, "loss": 1.3688, "step": 16422 }, { "epoch": 41.0575, "grad_norm": 6.8125, "learning_rate": 1.3317962200993745e-05, "loss": 1.2858, "step": 16423 }, { "epoch": 41.06, "grad_norm": 8.4375, "learning_rate": 1.3317198369284264e-05, "loss": 1.3461, "step": 16424 }, { "epoch": 41.0625, "grad_norm": 5.28125, "learning_rate": 1.331643451582736e-05, "loss": 1.4341, "step": 16425 }, { "epoch": 41.065, "grad_norm": 8.4375, "learning_rate": 1.3315670640628047e-05, "loss": 1.3323, "step": 16426 }, { "epoch": 41.0675, "grad_norm": 7.125, "learning_rate": 1.3314906743691328e-05, "loss": 1.3042, "step": 16427 }, { "epoch": 41.07, "grad_norm": 7.6875, "learning_rate": 1.3314142825022213e-05, "loss": 1.419, "step": 16428 }, { "epoch": 41.0725, "grad_norm": 5.65625, "learning_rate": 1.331337888462571e-05, "loss": 1.2252, "step": 16429 }, { "epoch": 41.075, "grad_norm": 6.09375, "learning_rate": 1.331261492250683e-05, "loss": 1.2454, "step": 16430 }, { "epoch": 41.0775, "grad_norm": 8.0625, "learning_rate": 1.3311850938670576e-05, "loss": 1.5126, "step": 16431 }, { "epoch": 41.08, "grad_norm": 9.9375, "learning_rate": 1.3311086933121961e-05, "loss": 1.3619, "step": 16432 }, { "epoch": 41.0825, "grad_norm": 5.375, "learning_rate": 1.3310322905865993e-05, "loss": 1.2449, "step": 16433 }, { "epoch": 41.085, "grad_norm": 5.625, "learning_rate": 1.3309558856907683e-05, "loss": 1.4368, "step": 16434 }, { "epoch": 41.0875, "grad_norm": 8.9375, "learning_rate": 1.3308794786252032e-05, "loss": 1.4516, "step": 16435 }, { "epoch": 41.09, "grad_norm": 7.84375, "learning_rate": 1.330803069390406e-05, "loss": 1.5358, "step": 16436 }, { "epoch": 41.0925, "grad_norm": 6.5, "learning_rate": 1.3307266579868767e-05, "loss": 1.4152, "step": 16437 }, { "epoch": 41.095, "grad_norm": 7.96875, "learning_rate": 1.330650244415117e-05, "loss": 1.2786, "step": 16438 }, { "epoch": 41.0975, "grad_norm": 8.375, "learning_rate": 1.3305738286756273e-05, "loss": 1.2961, "step": 16439 }, { "epoch": 41.1, "grad_norm": 10.0, "learning_rate": 1.3304974107689088e-05, "loss": 1.2931, "step": 16440 }, { "epoch": 41.1025, "grad_norm": 6.5625, "learning_rate": 1.3304209906954626e-05, "loss": 1.358, "step": 16441 }, { "epoch": 41.105, "grad_norm": 11.1875, "learning_rate": 1.3303445684557896e-05, "loss": 1.3427, "step": 16442 }, { "epoch": 41.1075, "grad_norm": 6.8125, "learning_rate": 1.3302681440503911e-05, "loss": 1.4475, "step": 16443 }, { "epoch": 41.11, "grad_norm": 5.65625, "learning_rate": 1.3301917174797673e-05, "loss": 1.2627, "step": 16444 }, { "epoch": 41.1125, "grad_norm": 7.84375, "learning_rate": 1.3301152887444202e-05, "loss": 1.3056, "step": 16445 }, { "epoch": 41.115, "grad_norm": 7.4375, "learning_rate": 1.3300388578448503e-05, "loss": 1.4494, "step": 16446 }, { "epoch": 41.1175, "grad_norm": 9.375, "learning_rate": 1.329962424781559e-05, "loss": 1.3686, "step": 16447 }, { "epoch": 41.12, "grad_norm": 7.90625, "learning_rate": 1.3298859895550473e-05, "loss": 1.3974, "step": 16448 }, { "epoch": 41.1225, "grad_norm": 10.875, "learning_rate": 1.3298095521658161e-05, "loss": 1.2572, "step": 16449 }, { "epoch": 41.125, "grad_norm": 8.1875, "learning_rate": 1.3297331126143667e-05, "loss": 1.3277, "step": 16450 }, { "epoch": 41.1275, "grad_norm": 9.75, "learning_rate": 1.3296566709012003e-05, "loss": 1.2294, "step": 16451 }, { "epoch": 41.13, "grad_norm": 7.90625, "learning_rate": 1.3295802270268177e-05, "loss": 1.3928, "step": 16452 }, { "epoch": 41.1325, "grad_norm": 6.3125, "learning_rate": 1.3295037809917205e-05, "loss": 1.3379, "step": 16453 }, { "epoch": 41.135, "grad_norm": 5.375, "learning_rate": 1.3294273327964098e-05, "loss": 1.3809, "step": 16454 }, { "epoch": 41.1375, "grad_norm": 8.9375, "learning_rate": 1.3293508824413868e-05, "loss": 1.4084, "step": 16455 }, { "epoch": 41.14, "grad_norm": 8.0625, "learning_rate": 1.329274429927152e-05, "loss": 1.429, "step": 16456 }, { "epoch": 41.1425, "grad_norm": 6.40625, "learning_rate": 1.3291979752542078e-05, "loss": 1.412, "step": 16457 }, { "epoch": 41.145, "grad_norm": 9.0625, "learning_rate": 1.3291215184230545e-05, "loss": 1.3979, "step": 16458 }, { "epoch": 41.1475, "grad_norm": 7.40625, "learning_rate": 1.3290450594341939e-05, "loss": 1.3158, "step": 16459 }, { "epoch": 41.15, "grad_norm": 8.1875, "learning_rate": 1.328968598288127e-05, "loss": 1.4337, "step": 16460 }, { "epoch": 41.1525, "grad_norm": 30.125, "learning_rate": 1.3288921349853552e-05, "loss": 1.3822, "step": 16461 }, { "epoch": 41.155, "grad_norm": 7.4375, "learning_rate": 1.3288156695263797e-05, "loss": 1.3062, "step": 16462 }, { "epoch": 41.1575, "grad_norm": 9.375, "learning_rate": 1.3287392019117018e-05, "loss": 1.3448, "step": 16463 }, { "epoch": 41.16, "grad_norm": 8.125, "learning_rate": 1.3286627321418229e-05, "loss": 1.2816, "step": 16464 }, { "epoch": 41.1625, "grad_norm": 10.0, "learning_rate": 1.328586260217244e-05, "loss": 1.2087, "step": 16465 }, { "epoch": 41.165, "grad_norm": 5.375, "learning_rate": 1.3285097861384668e-05, "loss": 1.2158, "step": 16466 }, { "epoch": 41.1675, "grad_norm": 6.25, "learning_rate": 1.328433309905993e-05, "loss": 1.3042, "step": 16467 }, { "epoch": 41.17, "grad_norm": 25.875, "learning_rate": 1.3283568315203232e-05, "loss": 1.3984, "step": 16468 }, { "epoch": 41.1725, "grad_norm": 8.9375, "learning_rate": 1.328280350981959e-05, "loss": 1.38, "step": 16469 }, { "epoch": 41.175, "grad_norm": 7.625, "learning_rate": 1.3282038682914025e-05, "loss": 1.2916, "step": 16470 }, { "epoch": 41.1775, "grad_norm": 15.375, "learning_rate": 1.328127383449154e-05, "loss": 1.3712, "step": 16471 }, { "epoch": 41.18, "grad_norm": 6.09375, "learning_rate": 1.3280508964557162e-05, "loss": 1.3202, "step": 16472 }, { "epoch": 41.1825, "grad_norm": 5.8125, "learning_rate": 1.3279744073115892e-05, "loss": 1.4868, "step": 16473 }, { "epoch": 41.185, "grad_norm": 8.5625, "learning_rate": 1.3278979160172753e-05, "loss": 1.3753, "step": 16474 }, { "epoch": 41.1875, "grad_norm": 10.25, "learning_rate": 1.327821422573276e-05, "loss": 1.4147, "step": 16475 }, { "epoch": 41.19, "grad_norm": 5.59375, "learning_rate": 1.3277449269800924e-05, "loss": 1.383, "step": 16476 }, { "epoch": 41.1925, "grad_norm": 6.25, "learning_rate": 1.327668429238226e-05, "loss": 1.3711, "step": 16477 }, { "epoch": 41.195, "grad_norm": 7.28125, "learning_rate": 1.3275919293481788e-05, "loss": 1.368, "step": 16478 }, { "epoch": 41.1975, "grad_norm": 8.3125, "learning_rate": 1.3275154273104519e-05, "loss": 1.1783, "step": 16479 }, { "epoch": 41.2, "grad_norm": 7.5, "learning_rate": 1.3274389231255466e-05, "loss": 1.2992, "step": 16480 }, { "epoch": 41.2025, "grad_norm": 6.875, "learning_rate": 1.3273624167939654e-05, "loss": 1.3393, "step": 16481 }, { "epoch": 41.205, "grad_norm": 7.25, "learning_rate": 1.327285908316209e-05, "loss": 1.5201, "step": 16482 }, { "epoch": 41.2075, "grad_norm": 10.375, "learning_rate": 1.3272093976927796e-05, "loss": 1.5736, "step": 16483 }, { "epoch": 41.21, "grad_norm": 8.125, "learning_rate": 1.327132884924178e-05, "loss": 1.2614, "step": 16484 }, { "epoch": 41.2125, "grad_norm": 10.5, "learning_rate": 1.3270563700109067e-05, "loss": 1.3664, "step": 16485 }, { "epoch": 41.215, "grad_norm": 8.875, "learning_rate": 1.3269798529534665e-05, "loss": 1.3158, "step": 16486 }, { "epoch": 41.2175, "grad_norm": 5.53125, "learning_rate": 1.3269033337523598e-05, "loss": 1.3388, "step": 16487 }, { "epoch": 41.22, "grad_norm": 7.125, "learning_rate": 1.3268268124080881e-05, "loss": 1.3956, "step": 16488 }, { "epoch": 41.2225, "grad_norm": 6.5625, "learning_rate": 1.3267502889211526e-05, "loss": 1.2705, "step": 16489 }, { "epoch": 41.225, "grad_norm": 11.0, "learning_rate": 1.3266737632920553e-05, "loss": 1.3834, "step": 16490 }, { "epoch": 41.2275, "grad_norm": 5.4375, "learning_rate": 1.3265972355212977e-05, "loss": 1.3192, "step": 16491 }, { "epoch": 41.23, "grad_norm": 7.34375, "learning_rate": 1.326520705609382e-05, "loss": 1.3553, "step": 16492 }, { "epoch": 41.2325, "grad_norm": 9.375, "learning_rate": 1.3264441735568094e-05, "loss": 1.4548, "step": 16493 }, { "epoch": 41.235, "grad_norm": 8.375, "learning_rate": 1.3263676393640821e-05, "loss": 1.4245, "step": 16494 }, { "epoch": 41.2375, "grad_norm": 8.375, "learning_rate": 1.3262911030317013e-05, "loss": 1.5374, "step": 16495 }, { "epoch": 41.24, "grad_norm": 7.8125, "learning_rate": 1.3262145645601693e-05, "loss": 1.3113, "step": 16496 }, { "epoch": 41.2425, "grad_norm": 7.375, "learning_rate": 1.3261380239499877e-05, "loss": 1.4238, "step": 16497 }, { "epoch": 41.245, "grad_norm": 5.9375, "learning_rate": 1.3260614812016586e-05, "loss": 1.3546, "step": 16498 }, { "epoch": 41.2475, "grad_norm": 8.0625, "learning_rate": 1.3259849363156827e-05, "loss": 1.4945, "step": 16499 }, { "epoch": 41.25, "grad_norm": 7.3125, "learning_rate": 1.3259083892925633e-05, "loss": 1.1694, "step": 16500 }, { "epoch": 41.2525, "grad_norm": 12.625, "learning_rate": 1.325831840132801e-05, "loss": 1.3688, "step": 16501 }, { "epoch": 41.255, "grad_norm": 9.625, "learning_rate": 1.3257552888368985e-05, "loss": 1.3113, "step": 16502 }, { "epoch": 41.2575, "grad_norm": 5.84375, "learning_rate": 1.3256787354053576e-05, "loss": 1.2815, "step": 16503 }, { "epoch": 41.26, "grad_norm": 4.96875, "learning_rate": 1.3256021798386795e-05, "loss": 1.5004, "step": 16504 }, { "epoch": 41.2625, "grad_norm": 10.0, "learning_rate": 1.3255256221373666e-05, "loss": 1.4149, "step": 16505 }, { "epoch": 41.265, "grad_norm": 6.125, "learning_rate": 1.325449062301921e-05, "loss": 1.2805, "step": 16506 }, { "epoch": 41.2675, "grad_norm": 6.625, "learning_rate": 1.325372500332844e-05, "loss": 1.2452, "step": 16507 }, { "epoch": 41.27, "grad_norm": 9.1875, "learning_rate": 1.3252959362306381e-05, "loss": 1.299, "step": 16508 }, { "epoch": 41.2725, "grad_norm": 6.84375, "learning_rate": 1.3252193699958052e-05, "loss": 1.2118, "step": 16509 }, { "epoch": 41.275, "grad_norm": 9.375, "learning_rate": 1.3251428016288467e-05, "loss": 1.3004, "step": 16510 }, { "epoch": 41.2775, "grad_norm": 5.9375, "learning_rate": 1.3250662311302653e-05, "loss": 1.4222, "step": 16511 }, { "epoch": 41.28, "grad_norm": 9.0625, "learning_rate": 1.3249896585005628e-05, "loss": 1.4309, "step": 16512 }, { "epoch": 41.2825, "grad_norm": 4.375, "learning_rate": 1.3249130837402413e-05, "loss": 1.3909, "step": 16513 }, { "epoch": 41.285, "grad_norm": 6.96875, "learning_rate": 1.324836506849802e-05, "loss": 1.3464, "step": 16514 }, { "epoch": 41.2875, "grad_norm": 7.15625, "learning_rate": 1.324759927829748e-05, "loss": 1.4852, "step": 16515 }, { "epoch": 41.29, "grad_norm": 7.46875, "learning_rate": 1.3246833466805807e-05, "loss": 1.4077, "step": 16516 }, { "epoch": 41.2925, "grad_norm": 11.375, "learning_rate": 1.3246067634028025e-05, "loss": 1.367, "step": 16517 }, { "epoch": 41.295, "grad_norm": 10.0625, "learning_rate": 1.3245301779969155e-05, "loss": 1.4435, "step": 16518 }, { "epoch": 41.2975, "grad_norm": 9.625, "learning_rate": 1.3244535904634214e-05, "loss": 1.2931, "step": 16519 }, { "epoch": 41.3, "grad_norm": 7.09375, "learning_rate": 1.3243770008028225e-05, "loss": 1.4058, "step": 16520 }, { "epoch": 41.3025, "grad_norm": 6.40625, "learning_rate": 1.3243004090156211e-05, "loss": 1.3161, "step": 16521 }, { "epoch": 41.305, "grad_norm": 6.4375, "learning_rate": 1.3242238151023192e-05, "loss": 1.2792, "step": 16522 }, { "epoch": 41.3075, "grad_norm": 7.125, "learning_rate": 1.324147219063419e-05, "loss": 1.4346, "step": 16523 }, { "epoch": 41.31, "grad_norm": 8.5625, "learning_rate": 1.3240706208994226e-05, "loss": 1.253, "step": 16524 }, { "epoch": 41.3125, "grad_norm": 8.4375, "learning_rate": 1.3239940206108322e-05, "loss": 1.4015, "step": 16525 }, { "epoch": 41.315, "grad_norm": 6.5, "learning_rate": 1.3239174181981496e-05, "loss": 1.4213, "step": 16526 }, { "epoch": 41.3175, "grad_norm": 7.34375, "learning_rate": 1.3238408136618776e-05, "loss": 1.3662, "step": 16527 }, { "epoch": 41.32, "grad_norm": 9.5, "learning_rate": 1.3237642070025183e-05, "loss": 1.3894, "step": 16528 }, { "epoch": 41.3225, "grad_norm": 12.375, "learning_rate": 1.3236875982205736e-05, "loss": 1.2758, "step": 16529 }, { "epoch": 41.325, "grad_norm": 6.40625, "learning_rate": 1.3236109873165462e-05, "loss": 1.2666, "step": 16530 }, { "epoch": 41.3275, "grad_norm": 10.875, "learning_rate": 1.3235343742909379e-05, "loss": 1.319, "step": 16531 }, { "epoch": 41.33, "grad_norm": 5.34375, "learning_rate": 1.3234577591442515e-05, "loss": 1.3591, "step": 16532 }, { "epoch": 41.3325, "grad_norm": 5.34375, "learning_rate": 1.3233811418769886e-05, "loss": 1.2377, "step": 16533 }, { "epoch": 41.335, "grad_norm": 7.375, "learning_rate": 1.3233045224896519e-05, "loss": 1.4558, "step": 16534 }, { "epoch": 41.3375, "grad_norm": 7.34375, "learning_rate": 1.3232279009827438e-05, "loss": 1.3602, "step": 16535 }, { "epoch": 41.34, "grad_norm": 5.625, "learning_rate": 1.3231512773567667e-05, "loss": 1.273, "step": 16536 }, { "epoch": 41.3425, "grad_norm": 7.75, "learning_rate": 1.3230746516122226e-05, "loss": 1.3933, "step": 16537 }, { "epoch": 41.345, "grad_norm": 9.125, "learning_rate": 1.322998023749614e-05, "loss": 1.3132, "step": 16538 }, { "epoch": 41.3475, "grad_norm": 6.3125, "learning_rate": 1.3229213937694432e-05, "loss": 1.3057, "step": 16539 }, { "epoch": 41.35, "grad_norm": 6.9375, "learning_rate": 1.3228447616722128e-05, "loss": 1.4253, "step": 16540 }, { "epoch": 41.3525, "grad_norm": 8.9375, "learning_rate": 1.322768127458425e-05, "loss": 1.3071, "step": 16541 }, { "epoch": 41.355, "grad_norm": 9.6875, "learning_rate": 1.3226914911285824e-05, "loss": 1.3321, "step": 16542 }, { "epoch": 41.3575, "grad_norm": 8.25, "learning_rate": 1.322614852683187e-05, "loss": 1.2757, "step": 16543 }, { "epoch": 41.36, "grad_norm": 5.46875, "learning_rate": 1.322538212122742e-05, "loss": 1.3723, "step": 16544 }, { "epoch": 41.3625, "grad_norm": 6.125, "learning_rate": 1.3224615694477492e-05, "loss": 1.2655, "step": 16545 }, { "epoch": 41.365, "grad_norm": 6.71875, "learning_rate": 1.3223849246587111e-05, "loss": 1.2885, "step": 16546 }, { "epoch": 41.3675, "grad_norm": 6.96875, "learning_rate": 1.3223082777561302e-05, "loss": 1.3563, "step": 16547 }, { "epoch": 41.37, "grad_norm": 8.75, "learning_rate": 1.3222316287405095e-05, "loss": 1.3093, "step": 16548 }, { "epoch": 41.3725, "grad_norm": 10.375, "learning_rate": 1.322154977612351e-05, "loss": 1.4484, "step": 16549 }, { "epoch": 41.375, "grad_norm": 7.9375, "learning_rate": 1.3220783243721571e-05, "loss": 1.2615, "step": 16550 }, { "epoch": 41.3775, "grad_norm": 5.09375, "learning_rate": 1.3220016690204308e-05, "loss": 1.3919, "step": 16551 }, { "epoch": 41.38, "grad_norm": 6.125, "learning_rate": 1.3219250115576745e-05, "loss": 1.4687, "step": 16552 }, { "epoch": 41.3825, "grad_norm": 4.65625, "learning_rate": 1.3218483519843907e-05, "loss": 1.3863, "step": 16553 }, { "epoch": 41.385, "grad_norm": 7.6875, "learning_rate": 1.321771690301082e-05, "loss": 1.3616, "step": 16554 }, { "epoch": 41.3875, "grad_norm": 7.75, "learning_rate": 1.321695026508251e-05, "loss": 1.4241, "step": 16555 }, { "epoch": 41.39, "grad_norm": 6.71875, "learning_rate": 1.3216183606064e-05, "loss": 1.3203, "step": 16556 }, { "epoch": 41.3925, "grad_norm": 10.4375, "learning_rate": 1.3215416925960325e-05, "loss": 1.153, "step": 16557 }, { "epoch": 41.395, "grad_norm": 5.625, "learning_rate": 1.32146502247765e-05, "loss": 1.4955, "step": 16558 }, { "epoch": 41.3975, "grad_norm": 6.5625, "learning_rate": 1.3213883502517559e-05, "loss": 1.2885, "step": 16559 }, { "epoch": 41.4, "grad_norm": 7.75, "learning_rate": 1.3213116759188525e-05, "loss": 1.5155, "step": 16560 }, { "epoch": 41.4025, "grad_norm": 5.28125, "learning_rate": 1.3212349994794428e-05, "loss": 1.1675, "step": 16561 }, { "epoch": 41.405, "grad_norm": 7.65625, "learning_rate": 1.321158320934029e-05, "loss": 1.3701, "step": 16562 }, { "epoch": 41.4075, "grad_norm": 7.4375, "learning_rate": 1.321081640283114e-05, "loss": 1.3003, "step": 16563 }, { "epoch": 41.41, "grad_norm": 9.0625, "learning_rate": 1.321004957527201e-05, "loss": 1.293, "step": 16564 }, { "epoch": 41.4125, "grad_norm": 5.5625, "learning_rate": 1.3209282726667922e-05, "loss": 1.3194, "step": 16565 }, { "epoch": 41.415, "grad_norm": 8.3125, "learning_rate": 1.3208515857023908e-05, "loss": 1.223, "step": 16566 }, { "epoch": 41.4175, "grad_norm": 8.6875, "learning_rate": 1.3207748966344989e-05, "loss": 1.4445, "step": 16567 }, { "epoch": 41.42, "grad_norm": 5.90625, "learning_rate": 1.3206982054636195e-05, "loss": 1.3123, "step": 16568 }, { "epoch": 41.4225, "grad_norm": 8.5625, "learning_rate": 1.3206215121902558e-05, "loss": 1.1969, "step": 16569 }, { "epoch": 41.425, "grad_norm": 6.25, "learning_rate": 1.3205448168149102e-05, "loss": 1.3309, "step": 16570 }, { "epoch": 41.4275, "grad_norm": 7.0, "learning_rate": 1.3204681193380855e-05, "loss": 1.2999, "step": 16571 }, { "epoch": 41.43, "grad_norm": 10.1875, "learning_rate": 1.3203914197602847e-05, "loss": 1.3505, "step": 16572 }, { "epoch": 41.4325, "grad_norm": 7.5625, "learning_rate": 1.3203147180820107e-05, "loss": 1.1682, "step": 16573 }, { "epoch": 41.435, "grad_norm": 5.4375, "learning_rate": 1.320238014303766e-05, "loss": 1.2164, "step": 16574 }, { "epoch": 41.4375, "grad_norm": 5.71875, "learning_rate": 1.3201613084260538e-05, "loss": 1.2762, "step": 16575 }, { "epoch": 41.44, "grad_norm": 8.6875, "learning_rate": 1.320084600449377e-05, "loss": 1.2712, "step": 16576 }, { "epoch": 41.4425, "grad_norm": 5.9375, "learning_rate": 1.3200078903742381e-05, "loss": 1.3022, "step": 16577 }, { "epoch": 41.445, "grad_norm": 21.0, "learning_rate": 1.3199311782011406e-05, "loss": 1.3644, "step": 16578 }, { "epoch": 41.4475, "grad_norm": 8.625, "learning_rate": 1.3198544639305868e-05, "loss": 1.2947, "step": 16579 }, { "epoch": 41.45, "grad_norm": 8.1875, "learning_rate": 1.31977774756308e-05, "loss": 1.2385, "step": 16580 }, { "epoch": 41.4525, "grad_norm": 5.625, "learning_rate": 1.3197010290991232e-05, "loss": 1.3117, "step": 16581 }, { "epoch": 41.455, "grad_norm": 8.125, "learning_rate": 1.319624308539219e-05, "loss": 1.4359, "step": 16582 }, { "epoch": 41.4575, "grad_norm": 8.0625, "learning_rate": 1.3195475858838709e-05, "loss": 1.2857, "step": 16583 }, { "epoch": 41.46, "grad_norm": 7.15625, "learning_rate": 1.3194708611335816e-05, "loss": 1.2637, "step": 16584 }, { "epoch": 41.4625, "grad_norm": 7.875, "learning_rate": 1.319394134288854e-05, "loss": 1.3304, "step": 16585 }, { "epoch": 41.465, "grad_norm": 7.96875, "learning_rate": 1.3193174053501911e-05, "loss": 1.3512, "step": 16586 }, { "epoch": 41.4675, "grad_norm": 6.25, "learning_rate": 1.3192406743180966e-05, "loss": 1.295, "step": 16587 }, { "epoch": 41.47, "grad_norm": 8.0625, "learning_rate": 1.3191639411930726e-05, "loss": 1.4789, "step": 16588 }, { "epoch": 41.4725, "grad_norm": 6.0, "learning_rate": 1.3190872059756226e-05, "loss": 1.3153, "step": 16589 }, { "epoch": 41.475, "grad_norm": 6.0, "learning_rate": 1.3190104686662497e-05, "loss": 1.3208, "step": 16590 }, { "epoch": 41.4775, "grad_norm": 6.375, "learning_rate": 1.318933729265457e-05, "loss": 1.2885, "step": 16591 }, { "epoch": 41.48, "grad_norm": 8.6875, "learning_rate": 1.3188569877737474e-05, "loss": 1.3384, "step": 16592 }, { "epoch": 41.4825, "grad_norm": 10.375, "learning_rate": 1.3187802441916242e-05, "loss": 1.4037, "step": 16593 }, { "epoch": 41.485, "grad_norm": 11.625, "learning_rate": 1.3187034985195904e-05, "loss": 1.1483, "step": 16594 }, { "epoch": 41.4875, "grad_norm": 8.6875, "learning_rate": 1.3186267507581496e-05, "loss": 1.538, "step": 16595 }, { "epoch": 41.49, "grad_norm": 6.78125, "learning_rate": 1.3185500009078038e-05, "loss": 1.6165, "step": 16596 }, { "epoch": 41.4925, "grad_norm": 10.1875, "learning_rate": 1.3184732489690577e-05, "loss": 1.2511, "step": 16597 }, { "epoch": 41.495, "grad_norm": 8.3125, "learning_rate": 1.3183964949424132e-05, "loss": 1.5086, "step": 16598 }, { "epoch": 41.4975, "grad_norm": 14.9375, "learning_rate": 1.3183197388283743e-05, "loss": 1.3842, "step": 16599 }, { "epoch": 41.5, "grad_norm": 7.40625, "learning_rate": 1.3182429806274442e-05, "loss": 1.5535, "step": 16600 }, { "epoch": 41.5025, "grad_norm": 10.5, "learning_rate": 1.3181662203401256e-05, "loss": 1.2514, "step": 16601 }, { "epoch": 41.505, "grad_norm": 6.53125, "learning_rate": 1.3180894579669216e-05, "loss": 1.3081, "step": 16602 }, { "epoch": 41.5075, "grad_norm": 6.5, "learning_rate": 1.3180126935083361e-05, "loss": 1.3425, "step": 16603 }, { "epoch": 41.51, "grad_norm": 6.8125, "learning_rate": 1.3179359269648723e-05, "loss": 1.2621, "step": 16604 }, { "epoch": 41.5125, "grad_norm": 5.625, "learning_rate": 1.3178591583370332e-05, "loss": 1.2466, "step": 16605 }, { "epoch": 41.515, "grad_norm": 7.375, "learning_rate": 1.3177823876253221e-05, "loss": 1.2927, "step": 16606 }, { "epoch": 41.5175, "grad_norm": 8.375, "learning_rate": 1.3177056148302426e-05, "loss": 1.4124, "step": 16607 }, { "epoch": 41.52, "grad_norm": 8.25, "learning_rate": 1.3176288399522975e-05, "loss": 1.2663, "step": 16608 }, { "epoch": 41.5225, "grad_norm": 6.03125, "learning_rate": 1.3175520629919905e-05, "loss": 1.4857, "step": 16609 }, { "epoch": 41.525, "grad_norm": 7.71875, "learning_rate": 1.317475283949825e-05, "loss": 1.2697, "step": 16610 }, { "epoch": 41.5275, "grad_norm": 7.125, "learning_rate": 1.3173985028263042e-05, "loss": 1.3684, "step": 16611 }, { "epoch": 41.53, "grad_norm": 6.875, "learning_rate": 1.3173217196219317e-05, "loss": 1.33, "step": 16612 }, { "epoch": 41.5325, "grad_norm": 9.875, "learning_rate": 1.3172449343372102e-05, "loss": 1.3188, "step": 16613 }, { "epoch": 41.535, "grad_norm": 9.375, "learning_rate": 1.317168146972644e-05, "loss": 1.4759, "step": 16614 }, { "epoch": 41.5375, "grad_norm": 7.53125, "learning_rate": 1.3170913575287361e-05, "loss": 1.3501, "step": 16615 }, { "epoch": 41.54, "grad_norm": 10.625, "learning_rate": 1.3170145660059898e-05, "loss": 1.314, "step": 16616 }, { "epoch": 41.5425, "grad_norm": 8.5, "learning_rate": 1.3169377724049088e-05, "loss": 1.2963, "step": 16617 }, { "epoch": 41.545, "grad_norm": 10.375, "learning_rate": 1.3168609767259966e-05, "loss": 1.35, "step": 16618 }, { "epoch": 41.5475, "grad_norm": 8.4375, "learning_rate": 1.3167841789697562e-05, "loss": 1.4853, "step": 16619 }, { "epoch": 41.55, "grad_norm": 6.65625, "learning_rate": 1.3167073791366915e-05, "loss": 1.3857, "step": 16620 }, { "epoch": 41.5525, "grad_norm": 7.5625, "learning_rate": 1.316630577227306e-05, "loss": 1.3382, "step": 16621 }, { "epoch": 41.555, "grad_norm": 7.1875, "learning_rate": 1.316553773242103e-05, "loss": 1.2223, "step": 16622 }, { "epoch": 41.5575, "grad_norm": 9.125, "learning_rate": 1.3164769671815862e-05, "loss": 1.3963, "step": 16623 }, { "epoch": 41.56, "grad_norm": 7.53125, "learning_rate": 1.3164001590462592e-05, "loss": 1.2972, "step": 16624 }, { "epoch": 41.5625, "grad_norm": 6.875, "learning_rate": 1.3163233488366254e-05, "loss": 1.3286, "step": 16625 }, { "epoch": 41.565, "grad_norm": 4.9375, "learning_rate": 1.316246536553188e-05, "loss": 1.3034, "step": 16626 }, { "epoch": 41.5675, "grad_norm": 7.5625, "learning_rate": 1.3161697221964513e-05, "loss": 1.306, "step": 16627 }, { "epoch": 41.57, "grad_norm": 6.4375, "learning_rate": 1.3160929057669183e-05, "loss": 1.3035, "step": 16628 }, { "epoch": 41.5725, "grad_norm": 6.90625, "learning_rate": 1.3160160872650934e-05, "loss": 1.3914, "step": 16629 }, { "epoch": 41.575, "grad_norm": 8.6875, "learning_rate": 1.315939266691479e-05, "loss": 1.2976, "step": 16630 }, { "epoch": 41.5775, "grad_norm": 6.71875, "learning_rate": 1.3158624440465796e-05, "loss": 1.3061, "step": 16631 }, { "epoch": 41.58, "grad_norm": 8.75, "learning_rate": 1.3157856193308988e-05, "loss": 1.2554, "step": 16632 }, { "epoch": 41.5825, "grad_norm": 7.8125, "learning_rate": 1.3157087925449403e-05, "loss": 1.5087, "step": 16633 }, { "epoch": 41.585, "grad_norm": 6.15625, "learning_rate": 1.3156319636892072e-05, "loss": 1.4182, "step": 16634 }, { "epoch": 41.5875, "grad_norm": 10.5625, "learning_rate": 1.3155551327642038e-05, "loss": 1.4685, "step": 16635 }, { "epoch": 41.59, "grad_norm": 7.34375, "learning_rate": 1.3154782997704336e-05, "loss": 1.3338, "step": 16636 }, { "epoch": 41.5925, "grad_norm": 6.65625, "learning_rate": 1.3154014647084001e-05, "loss": 1.4318, "step": 16637 }, { "epoch": 41.595, "grad_norm": 7.1875, "learning_rate": 1.3153246275786073e-05, "loss": 1.3958, "step": 16638 }, { "epoch": 41.5975, "grad_norm": 6.71875, "learning_rate": 1.315247788381559e-05, "loss": 1.2253, "step": 16639 }, { "epoch": 41.6, "grad_norm": 8.625, "learning_rate": 1.3151709471177589e-05, "loss": 1.383, "step": 16640 }, { "epoch": 41.6025, "grad_norm": 6.375, "learning_rate": 1.3150941037877103e-05, "loss": 1.3912, "step": 16641 }, { "epoch": 41.605, "grad_norm": 7.3125, "learning_rate": 1.3150172583919176e-05, "loss": 1.4331, "step": 16642 }, { "epoch": 41.6075, "grad_norm": 7.34375, "learning_rate": 1.314940410930884e-05, "loss": 1.2622, "step": 16643 }, { "epoch": 41.61, "grad_norm": 7.15625, "learning_rate": 1.3148635614051143e-05, "loss": 1.3291, "step": 16644 }, { "epoch": 41.6125, "grad_norm": 8.625, "learning_rate": 1.3147867098151115e-05, "loss": 1.2386, "step": 16645 }, { "epoch": 41.615, "grad_norm": 11.5625, "learning_rate": 1.3147098561613794e-05, "loss": 1.2751, "step": 16646 }, { "epoch": 41.6175, "grad_norm": 7.5, "learning_rate": 1.3146330004444223e-05, "loss": 1.4998, "step": 16647 }, { "epoch": 41.62, "grad_norm": 8.75, "learning_rate": 1.3145561426647438e-05, "loss": 1.3791, "step": 16648 }, { "epoch": 41.6225, "grad_norm": 7.71875, "learning_rate": 1.3144792828228477e-05, "loss": 1.211, "step": 16649 }, { "epoch": 41.625, "grad_norm": 8.375, "learning_rate": 1.3144024209192378e-05, "loss": 1.2488, "step": 16650 }, { "epoch": 41.6275, "grad_norm": 7.53125, "learning_rate": 1.3143255569544187e-05, "loss": 1.282, "step": 16651 }, { "epoch": 41.63, "grad_norm": 6.9375, "learning_rate": 1.3142486909288934e-05, "loss": 1.2989, "step": 16652 }, { "epoch": 41.6325, "grad_norm": 10.0625, "learning_rate": 1.3141718228431666e-05, "loss": 1.4041, "step": 16653 }, { "epoch": 41.635, "grad_norm": 6.5, "learning_rate": 1.3140949526977415e-05, "loss": 1.323, "step": 16654 }, { "epoch": 41.6375, "grad_norm": 7.40625, "learning_rate": 1.314018080493123e-05, "loss": 1.2797, "step": 16655 }, { "epoch": 41.64, "grad_norm": 9.0, "learning_rate": 1.3139412062298141e-05, "loss": 1.4548, "step": 16656 }, { "epoch": 41.6425, "grad_norm": 7.9375, "learning_rate": 1.3138643299083196e-05, "loss": 1.3975, "step": 16657 }, { "epoch": 41.645, "grad_norm": 6.625, "learning_rate": 1.3137874515291427e-05, "loss": 1.2883, "step": 16658 }, { "epoch": 41.6475, "grad_norm": 10.375, "learning_rate": 1.313710571092788e-05, "loss": 1.4747, "step": 16659 }, { "epoch": 41.65, "grad_norm": 5.28125, "learning_rate": 1.3136336885997591e-05, "loss": 1.2445, "step": 16660 }, { "epoch": 41.6525, "grad_norm": 7.0625, "learning_rate": 1.3135568040505608e-05, "loss": 1.3207, "step": 16661 }, { "epoch": 41.655, "grad_norm": 8.1875, "learning_rate": 1.3134799174456963e-05, "loss": 1.2104, "step": 16662 }, { "epoch": 41.6575, "grad_norm": 8.8125, "learning_rate": 1.31340302878567e-05, "loss": 1.3913, "step": 16663 }, { "epoch": 41.66, "grad_norm": 8.125, "learning_rate": 1.3133261380709861e-05, "loss": 1.3435, "step": 16664 }, { "epoch": 41.6625, "grad_norm": 4.65625, "learning_rate": 1.3132492453021486e-05, "loss": 1.2449, "step": 16665 }, { "epoch": 41.665, "grad_norm": 9.375, "learning_rate": 1.3131723504796615e-05, "loss": 1.39, "step": 16666 }, { "epoch": 41.6675, "grad_norm": 7.0625, "learning_rate": 1.3130954536040292e-05, "loss": 1.3603, "step": 16667 }, { "epoch": 41.67, "grad_norm": 6.78125, "learning_rate": 1.3130185546757552e-05, "loss": 1.4512, "step": 16668 }, { "epoch": 41.6725, "grad_norm": 8.4375, "learning_rate": 1.3129416536953446e-05, "loss": 1.1528, "step": 16669 }, { "epoch": 41.675, "grad_norm": 8.375, "learning_rate": 1.3128647506633005e-05, "loss": 1.2873, "step": 16670 }, { "epoch": 41.6775, "grad_norm": 6.875, "learning_rate": 1.3127878455801282e-05, "loss": 1.4116, "step": 16671 }, { "epoch": 41.68, "grad_norm": 6.875, "learning_rate": 1.312710938446331e-05, "loss": 1.325, "step": 16672 }, { "epoch": 41.6825, "grad_norm": 7.59375, "learning_rate": 1.3126340292624131e-05, "loss": 1.3477, "step": 16673 }, { "epoch": 41.685, "grad_norm": 7.65625, "learning_rate": 1.3125571180288797e-05, "loss": 1.2436, "step": 16674 }, { "epoch": 41.6875, "grad_norm": 6.1875, "learning_rate": 1.3124802047462335e-05, "loss": 1.4478, "step": 16675 }, { "epoch": 41.69, "grad_norm": 6.875, "learning_rate": 1.3124032894149803e-05, "loss": 1.3202, "step": 16676 }, { "epoch": 41.6925, "grad_norm": 10.75, "learning_rate": 1.3123263720356235e-05, "loss": 1.3729, "step": 16677 }, { "epoch": 41.695, "grad_norm": 9.0, "learning_rate": 1.3122494526086672e-05, "loss": 1.6478, "step": 16678 }, { "epoch": 41.6975, "grad_norm": 8.5625, "learning_rate": 1.3121725311346162e-05, "loss": 1.4144, "step": 16679 }, { "epoch": 41.7, "grad_norm": 7.53125, "learning_rate": 1.3120956076139746e-05, "loss": 1.321, "step": 16680 }, { "epoch": 41.7025, "grad_norm": 8.25, "learning_rate": 1.3120186820472464e-05, "loss": 1.3039, "step": 16681 }, { "epoch": 41.705, "grad_norm": 7.28125, "learning_rate": 1.3119417544349369e-05, "loss": 1.2463, "step": 16682 }, { "epoch": 41.7075, "grad_norm": 7.375, "learning_rate": 1.311864824777549e-05, "loss": 1.3925, "step": 16683 }, { "epoch": 41.71, "grad_norm": 5.75, "learning_rate": 1.3117878930755881e-05, "loss": 1.2267, "step": 16684 }, { "epoch": 41.7125, "grad_norm": 8.0625, "learning_rate": 1.3117109593295581e-05, "loss": 1.358, "step": 16685 }, { "epoch": 41.715, "grad_norm": 8.25, "learning_rate": 1.3116340235399639e-05, "loss": 1.3006, "step": 16686 }, { "epoch": 41.7175, "grad_norm": 7.21875, "learning_rate": 1.3115570857073093e-05, "loss": 1.2668, "step": 16687 }, { "epoch": 41.72, "grad_norm": 5.84375, "learning_rate": 1.3114801458320988e-05, "loss": 1.3841, "step": 16688 }, { "epoch": 41.7225, "grad_norm": 7.15625, "learning_rate": 1.311403203914837e-05, "loss": 1.2824, "step": 16689 }, { "epoch": 41.725, "grad_norm": 8.5625, "learning_rate": 1.311326259956028e-05, "loss": 1.2633, "step": 16690 }, { "epoch": 41.7275, "grad_norm": 7.34375, "learning_rate": 1.3112493139561768e-05, "loss": 1.4131, "step": 16691 }, { "epoch": 41.73, "grad_norm": 6.875, "learning_rate": 1.3111723659157875e-05, "loss": 1.305, "step": 16692 }, { "epoch": 41.7325, "grad_norm": 5.8125, "learning_rate": 1.3110954158353646e-05, "loss": 1.3595, "step": 16693 }, { "epoch": 41.735, "grad_norm": 6.84375, "learning_rate": 1.3110184637154126e-05, "loss": 1.3465, "step": 16694 }, { "epoch": 41.7375, "grad_norm": 7.5625, "learning_rate": 1.3109415095564362e-05, "loss": 1.4393, "step": 16695 }, { "epoch": 41.74, "grad_norm": 5.46875, "learning_rate": 1.3108645533589394e-05, "loss": 1.5683, "step": 16696 }, { "epoch": 41.7425, "grad_norm": 5.65625, "learning_rate": 1.3107875951234275e-05, "loss": 1.2982, "step": 16697 }, { "epoch": 41.745, "grad_norm": 11.3125, "learning_rate": 1.3107106348504039e-05, "loss": 1.515, "step": 16698 }, { "epoch": 41.7475, "grad_norm": 6.9375, "learning_rate": 1.3106336725403743e-05, "loss": 1.313, "step": 16699 }, { "epoch": 41.75, "grad_norm": 8.0625, "learning_rate": 1.3105567081938423e-05, "loss": 1.4875, "step": 16700 }, { "epoch": 41.7525, "grad_norm": 7.5625, "learning_rate": 1.3104797418113136e-05, "loss": 1.3301, "step": 16701 }, { "epoch": 41.755, "grad_norm": 8.5, "learning_rate": 1.3104027733932918e-05, "loss": 1.3499, "step": 16702 }, { "epoch": 41.7575, "grad_norm": 6.625, "learning_rate": 1.3103258029402816e-05, "loss": 1.5106, "step": 16703 }, { "epoch": 41.76, "grad_norm": 6.40625, "learning_rate": 1.3102488304527883e-05, "loss": 1.3523, "step": 16704 }, { "epoch": 41.7625, "grad_norm": 7.125, "learning_rate": 1.3101718559313156e-05, "loss": 1.2665, "step": 16705 }, { "epoch": 41.765, "grad_norm": 7.6875, "learning_rate": 1.310094879376369e-05, "loss": 1.4446, "step": 16706 }, { "epoch": 41.7675, "grad_norm": 6.21875, "learning_rate": 1.3100179007884528e-05, "loss": 1.3325, "step": 16707 }, { "epoch": 41.77, "grad_norm": 7.0625, "learning_rate": 1.3099409201680715e-05, "loss": 1.3503, "step": 16708 }, { "epoch": 41.7725, "grad_norm": 5.65625, "learning_rate": 1.3098639375157301e-05, "loss": 1.3099, "step": 16709 }, { "epoch": 41.775, "grad_norm": 9.5, "learning_rate": 1.3097869528319329e-05, "loss": 1.2909, "step": 16710 }, { "epoch": 41.7775, "grad_norm": 6.875, "learning_rate": 1.3097099661171848e-05, "loss": 1.3531, "step": 16711 }, { "epoch": 41.78, "grad_norm": 8.875, "learning_rate": 1.309632977371991e-05, "loss": 1.2515, "step": 16712 }, { "epoch": 41.7825, "grad_norm": 9.625, "learning_rate": 1.309555986596855e-05, "loss": 1.3602, "step": 16713 }, { "epoch": 41.785, "grad_norm": 9.3125, "learning_rate": 1.3094789937922832e-05, "loss": 1.23, "step": 16714 }, { "epoch": 41.7875, "grad_norm": 9.1875, "learning_rate": 1.3094019989587787e-05, "loss": 1.3907, "step": 16715 }, { "epoch": 41.79, "grad_norm": 7.3125, "learning_rate": 1.3093250020968477e-05, "loss": 1.3876, "step": 16716 }, { "epoch": 41.7925, "grad_norm": 8.1875, "learning_rate": 1.309248003206994e-05, "loss": 1.4015, "step": 16717 }, { "epoch": 41.795, "grad_norm": 5.59375, "learning_rate": 1.3091710022897231e-05, "loss": 1.2489, "step": 16718 }, { "epoch": 41.7975, "grad_norm": 7.34375, "learning_rate": 1.3090939993455391e-05, "loss": 1.358, "step": 16719 }, { "epoch": 41.8, "grad_norm": 7.1875, "learning_rate": 1.3090169943749475e-05, "loss": 1.3928, "step": 16720 }, { "epoch": 41.8025, "grad_norm": 8.1875, "learning_rate": 1.3089399873784529e-05, "loss": 1.3084, "step": 16721 }, { "epoch": 41.805, "grad_norm": 8.625, "learning_rate": 1.3088629783565598e-05, "loss": 1.4575, "step": 16722 }, { "epoch": 41.8075, "grad_norm": 8.8125, "learning_rate": 1.3087859673097735e-05, "loss": 1.3595, "step": 16723 }, { "epoch": 41.81, "grad_norm": 7.25, "learning_rate": 1.308708954238599e-05, "loss": 1.3144, "step": 16724 }, { "epoch": 41.8125, "grad_norm": 7.375, "learning_rate": 1.3086319391435405e-05, "loss": 1.3284, "step": 16725 }, { "epoch": 41.815, "grad_norm": 7.9375, "learning_rate": 1.3085549220251037e-05, "loss": 1.3437, "step": 16726 }, { "epoch": 41.8175, "grad_norm": 8.3125, "learning_rate": 1.3084779028837933e-05, "loss": 1.4166, "step": 16727 }, { "epoch": 41.82, "grad_norm": 9.5625, "learning_rate": 1.3084008817201137e-05, "loss": 1.3689, "step": 16728 }, { "epoch": 41.8225, "grad_norm": 5.8125, "learning_rate": 1.3083238585345706e-05, "loss": 1.1881, "step": 16729 }, { "epoch": 41.825, "grad_norm": 7.34375, "learning_rate": 1.3082468333276683e-05, "loss": 1.3396, "step": 16730 }, { "epoch": 41.8275, "grad_norm": 8.75, "learning_rate": 1.3081698060999123e-05, "loss": 1.3439, "step": 16731 }, { "epoch": 41.83, "grad_norm": 7.875, "learning_rate": 1.3080927768518074e-05, "loss": 1.2292, "step": 16732 }, { "epoch": 41.8325, "grad_norm": 5.96875, "learning_rate": 1.3080157455838585e-05, "loss": 1.345, "step": 16733 }, { "epoch": 41.835, "grad_norm": 9.25, "learning_rate": 1.3079387122965708e-05, "loss": 1.3949, "step": 16734 }, { "epoch": 41.8375, "grad_norm": 7.34375, "learning_rate": 1.3078616769904491e-05, "loss": 1.3051, "step": 16735 }, { "epoch": 41.84, "grad_norm": 7.40625, "learning_rate": 1.3077846396659986e-05, "loss": 1.3242, "step": 16736 }, { "epoch": 41.8425, "grad_norm": 9.3125, "learning_rate": 1.3077076003237246e-05, "loss": 1.3164, "step": 16737 }, { "epoch": 41.845, "grad_norm": 10.5, "learning_rate": 1.3076305589641314e-05, "loss": 1.2994, "step": 16738 }, { "epoch": 41.8475, "grad_norm": 7.6875, "learning_rate": 1.307553515587725e-05, "loss": 1.2902, "step": 16739 }, { "epoch": 41.85, "grad_norm": 8.125, "learning_rate": 1.3074764701950095e-05, "loss": 1.4281, "step": 16740 }, { "epoch": 41.8525, "grad_norm": 5.84375, "learning_rate": 1.307399422786491e-05, "loss": 1.2604, "step": 16741 }, { "epoch": 41.855, "grad_norm": 7.3125, "learning_rate": 1.3073223733626743e-05, "loss": 1.2239, "step": 16742 }, { "epoch": 41.8575, "grad_norm": 6.75, "learning_rate": 1.307245321924064e-05, "loss": 1.3798, "step": 16743 }, { "epoch": 41.86, "grad_norm": 9.875, "learning_rate": 1.3071682684711661e-05, "loss": 1.2655, "step": 16744 }, { "epoch": 41.8625, "grad_norm": 11.1875, "learning_rate": 1.3070912130044848e-05, "loss": 1.4188, "step": 16745 }, { "epoch": 41.865, "grad_norm": 6.25, "learning_rate": 1.3070141555245263e-05, "loss": 1.3647, "step": 16746 }, { "epoch": 41.8675, "grad_norm": 5.78125, "learning_rate": 1.3069370960317951e-05, "loss": 1.2377, "step": 16747 }, { "epoch": 41.87, "grad_norm": 10.375, "learning_rate": 1.3068600345267965e-05, "loss": 1.4318, "step": 16748 }, { "epoch": 41.8725, "grad_norm": 7.8125, "learning_rate": 1.306782971010036e-05, "loss": 1.3282, "step": 16749 }, { "epoch": 41.875, "grad_norm": 8.3125, "learning_rate": 1.3067059054820184e-05, "loss": 1.2147, "step": 16750 }, { "epoch": 41.8775, "grad_norm": 7.9375, "learning_rate": 1.3066288379432493e-05, "loss": 1.3211, "step": 16751 }, { "epoch": 41.88, "grad_norm": 8.375, "learning_rate": 1.3065517683942339e-05, "loss": 1.3228, "step": 16752 }, { "epoch": 41.8825, "grad_norm": 5.375, "learning_rate": 1.3064746968354772e-05, "loss": 1.3015, "step": 16753 }, { "epoch": 41.885, "grad_norm": 8.0625, "learning_rate": 1.3063976232674848e-05, "loss": 1.4372, "step": 16754 }, { "epoch": 41.8875, "grad_norm": 9.5625, "learning_rate": 1.3063205476907614e-05, "loss": 1.4469, "step": 16755 }, { "epoch": 41.89, "grad_norm": 5.34375, "learning_rate": 1.3062434701058134e-05, "loss": 1.323, "step": 16756 }, { "epoch": 41.8925, "grad_norm": 9.0625, "learning_rate": 1.306166390513145e-05, "loss": 1.4314, "step": 16757 }, { "epoch": 41.895, "grad_norm": 8.0625, "learning_rate": 1.3060893089132623e-05, "loss": 1.4282, "step": 16758 }, { "epoch": 41.8975, "grad_norm": 9.875, "learning_rate": 1.3060122253066701e-05, "loss": 1.4331, "step": 16759 }, { "epoch": 41.9, "grad_norm": 10.0625, "learning_rate": 1.305935139693874e-05, "loss": 1.1993, "step": 16760 }, { "epoch": 41.9025, "grad_norm": 7.125, "learning_rate": 1.3058580520753794e-05, "loss": 1.4151, "step": 16761 }, { "epoch": 41.905, "grad_norm": 6.5, "learning_rate": 1.3057809624516918e-05, "loss": 1.2375, "step": 16762 }, { "epoch": 41.9075, "grad_norm": 9.9375, "learning_rate": 1.3057038708233161e-05, "loss": 1.4063, "step": 16763 }, { "epoch": 41.91, "grad_norm": 6.75, "learning_rate": 1.3056267771907583e-05, "loss": 1.3223, "step": 16764 }, { "epoch": 41.9125, "grad_norm": 7.65625, "learning_rate": 1.3055496815545237e-05, "loss": 1.3248, "step": 16765 }, { "epoch": 41.915, "grad_norm": 5.96875, "learning_rate": 1.3054725839151174e-05, "loss": 1.4208, "step": 16766 }, { "epoch": 41.9175, "grad_norm": 9.25, "learning_rate": 1.3053954842730453e-05, "loss": 1.2848, "step": 16767 }, { "epoch": 41.92, "grad_norm": 8.5625, "learning_rate": 1.3053183826288124e-05, "loss": 1.3926, "step": 16768 }, { "epoch": 41.9225, "grad_norm": 6.1875, "learning_rate": 1.3052412789829245e-05, "loss": 1.3597, "step": 16769 }, { "epoch": 41.925, "grad_norm": 9.3125, "learning_rate": 1.3051641733358869e-05, "loss": 1.2789, "step": 16770 }, { "epoch": 41.9275, "grad_norm": 6.53125, "learning_rate": 1.3050870656882054e-05, "loss": 1.2195, "step": 16771 }, { "epoch": 41.93, "grad_norm": 9.0, "learning_rate": 1.305009956040385e-05, "loss": 1.5777, "step": 16772 }, { "epoch": 41.9325, "grad_norm": 9.3125, "learning_rate": 1.3049328443929316e-05, "loss": 1.4472, "step": 16773 }, { "epoch": 41.935, "grad_norm": 10.0, "learning_rate": 1.304855730746351e-05, "loss": 1.4664, "step": 16774 }, { "epoch": 41.9375, "grad_norm": 9.125, "learning_rate": 1.304778615101148e-05, "loss": 1.3735, "step": 16775 }, { "epoch": 41.94, "grad_norm": 5.8125, "learning_rate": 1.304701497457829e-05, "loss": 1.4341, "step": 16776 }, { "epoch": 41.9425, "grad_norm": 7.1875, "learning_rate": 1.3046243778168987e-05, "loss": 1.26, "step": 16777 }, { "epoch": 41.945, "grad_norm": 9.6875, "learning_rate": 1.3045472561788637e-05, "loss": 1.3536, "step": 16778 }, { "epoch": 41.9475, "grad_norm": 7.4375, "learning_rate": 1.3044701325442288e-05, "loss": 1.3281, "step": 16779 }, { "epoch": 41.95, "grad_norm": 7.71875, "learning_rate": 1.3043930069134998e-05, "loss": 1.2191, "step": 16780 }, { "epoch": 41.9525, "grad_norm": 126.0, "learning_rate": 1.3043158792871827e-05, "loss": 1.3445, "step": 16781 }, { "epoch": 41.955, "grad_norm": 9.25, "learning_rate": 1.3042387496657826e-05, "loss": 1.2348, "step": 16782 }, { "epoch": 41.9575, "grad_norm": 10.5625, "learning_rate": 1.3041616180498055e-05, "loss": 1.3813, "step": 16783 }, { "epoch": 41.96, "grad_norm": 9.0, "learning_rate": 1.3040844844397573e-05, "loss": 1.2789, "step": 16784 }, { "epoch": 41.9625, "grad_norm": 8.4375, "learning_rate": 1.3040073488361428e-05, "loss": 1.3408, "step": 16785 }, { "epoch": 41.965, "grad_norm": 6.46875, "learning_rate": 1.303930211239469e-05, "loss": 1.2513, "step": 16786 }, { "epoch": 41.9675, "grad_norm": 7.25, "learning_rate": 1.3038530716502403e-05, "loss": 1.3379, "step": 16787 }, { "epoch": 41.97, "grad_norm": 6.78125, "learning_rate": 1.3037759300689632e-05, "loss": 1.1918, "step": 16788 }, { "epoch": 41.9725, "grad_norm": 9.25, "learning_rate": 1.3036987864961435e-05, "loss": 1.423, "step": 16789 }, { "epoch": 41.975, "grad_norm": 9.8125, "learning_rate": 1.3036216409322864e-05, "loss": 1.3325, "step": 16790 }, { "epoch": 41.9775, "grad_norm": 8.5625, "learning_rate": 1.303544493377898e-05, "loss": 1.2963, "step": 16791 }, { "epoch": 41.98, "grad_norm": 5.71875, "learning_rate": 1.3034673438334841e-05, "loss": 1.5088, "step": 16792 }, { "epoch": 41.9825, "grad_norm": 8.3125, "learning_rate": 1.3033901922995504e-05, "loss": 1.5119, "step": 16793 }, { "epoch": 41.985, "grad_norm": 7.03125, "learning_rate": 1.3033130387766031e-05, "loss": 1.3745, "step": 16794 }, { "epoch": 41.9875, "grad_norm": 7.28125, "learning_rate": 1.3032358832651472e-05, "loss": 1.517, "step": 16795 }, { "epoch": 41.99, "grad_norm": 7.0, "learning_rate": 1.3031587257656892e-05, "loss": 1.4134, "step": 16796 }, { "epoch": 41.9925, "grad_norm": 7.53125, "learning_rate": 1.3030815662787347e-05, "loss": 1.3027, "step": 16797 }, { "epoch": 41.995, "grad_norm": 7.46875, "learning_rate": 1.3030044048047894e-05, "loss": 1.3272, "step": 16798 }, { "epoch": 41.9975, "grad_norm": 8.5, "learning_rate": 1.3029272413443596e-05, "loss": 1.3719, "step": 16799 }, { "epoch": 42.0, "grad_norm": 5.78125, "learning_rate": 1.3028500758979507e-05, "loss": 1.3177, "step": 16800 }, { "epoch": 42.0025, "grad_norm": 9.625, "learning_rate": 1.3027729084660692e-05, "loss": 1.2984, "step": 16801 }, { "epoch": 42.005, "grad_norm": 6.96875, "learning_rate": 1.30269573904922e-05, "loss": 1.3353, "step": 16802 }, { "epoch": 42.0075, "grad_norm": 7.09375, "learning_rate": 1.3026185676479104e-05, "loss": 1.4472, "step": 16803 }, { "epoch": 42.01, "grad_norm": 7.71875, "learning_rate": 1.3025413942626453e-05, "loss": 1.3812, "step": 16804 }, { "epoch": 42.0125, "grad_norm": 6.8125, "learning_rate": 1.3024642188939307e-05, "loss": 1.3141, "step": 16805 }, { "epoch": 42.015, "grad_norm": 7.65625, "learning_rate": 1.302387041542273e-05, "loss": 1.2599, "step": 16806 }, { "epoch": 42.0175, "grad_norm": 8.0, "learning_rate": 1.3023098622081778e-05, "loss": 1.2772, "step": 16807 }, { "epoch": 42.02, "grad_norm": 5.71875, "learning_rate": 1.3022326808921513e-05, "loss": 1.1927, "step": 16808 }, { "epoch": 42.0225, "grad_norm": 7.53125, "learning_rate": 1.3021554975946998e-05, "loss": 1.424, "step": 16809 }, { "epoch": 42.025, "grad_norm": 8.0625, "learning_rate": 1.3020783123163282e-05, "loss": 1.3383, "step": 16810 }, { "epoch": 42.0275, "grad_norm": 6.9375, "learning_rate": 1.3020011250575437e-05, "loss": 1.2785, "step": 16811 }, { "epoch": 42.03, "grad_norm": 8.5625, "learning_rate": 1.301923935818852e-05, "loss": 1.3924, "step": 16812 }, { "epoch": 42.0325, "grad_norm": 6.90625, "learning_rate": 1.3018467446007588e-05, "loss": 1.483, "step": 16813 }, { "epoch": 42.035, "grad_norm": 6.4375, "learning_rate": 1.301769551403771e-05, "loss": 1.3084, "step": 16814 }, { "epoch": 42.0375, "grad_norm": 5.6875, "learning_rate": 1.3016923562283932e-05, "loss": 1.3579, "step": 16815 }, { "epoch": 42.04, "grad_norm": 8.625, "learning_rate": 1.3016151590751332e-05, "loss": 1.2593, "step": 16816 }, { "epoch": 42.0425, "grad_norm": 9.125, "learning_rate": 1.3015379599444958e-05, "loss": 1.5041, "step": 16817 }, { "epoch": 42.045, "grad_norm": 6.90625, "learning_rate": 1.3014607588369878e-05, "loss": 1.2496, "step": 16818 }, { "epoch": 42.0475, "grad_norm": 6.53125, "learning_rate": 1.3013835557531151e-05, "loss": 1.2967, "step": 16819 }, { "epoch": 42.05, "grad_norm": 6.375, "learning_rate": 1.3013063506933838e-05, "loss": 1.3679, "step": 16820 }, { "epoch": 42.0525, "grad_norm": 8.5, "learning_rate": 1.3012291436583002e-05, "loss": 1.1162, "step": 16821 }, { "epoch": 42.055, "grad_norm": 7.3125, "learning_rate": 1.3011519346483704e-05, "loss": 1.4246, "step": 16822 }, { "epoch": 42.0575, "grad_norm": 5.65625, "learning_rate": 1.3010747236641005e-05, "loss": 1.473, "step": 16823 }, { "epoch": 42.06, "grad_norm": 9.9375, "learning_rate": 1.300997510705997e-05, "loss": 1.359, "step": 16824 }, { "epoch": 42.0625, "grad_norm": 7.25, "learning_rate": 1.3009202957745652e-05, "loss": 1.2166, "step": 16825 }, { "epoch": 42.065, "grad_norm": 8.75, "learning_rate": 1.3008430788703127e-05, "loss": 1.3468, "step": 16826 }, { "epoch": 42.0675, "grad_norm": 8.8125, "learning_rate": 1.3007658599937446e-05, "loss": 1.246, "step": 16827 }, { "epoch": 42.07, "grad_norm": 6.34375, "learning_rate": 1.3006886391453679e-05, "loss": 1.3702, "step": 16828 }, { "epoch": 42.0725, "grad_norm": 7.34375, "learning_rate": 1.3006114163256882e-05, "loss": 1.3893, "step": 16829 }, { "epoch": 42.075, "grad_norm": 8.8125, "learning_rate": 1.3005341915352124e-05, "loss": 1.3097, "step": 16830 }, { "epoch": 42.0775, "grad_norm": 7.46875, "learning_rate": 1.3004569647744461e-05, "loss": 1.4005, "step": 16831 }, { "epoch": 42.08, "grad_norm": 8.875, "learning_rate": 1.3003797360438961e-05, "loss": 1.2757, "step": 16832 }, { "epoch": 42.0825, "grad_norm": 6.5625, "learning_rate": 1.3003025053440685e-05, "loss": 1.3814, "step": 16833 }, { "epoch": 42.085, "grad_norm": 7.75, "learning_rate": 1.3002252726754699e-05, "loss": 1.413, "step": 16834 }, { "epoch": 42.0875, "grad_norm": 7.46875, "learning_rate": 1.3001480380386062e-05, "loss": 1.3699, "step": 16835 }, { "epoch": 42.09, "grad_norm": 6.15625, "learning_rate": 1.3000708014339842e-05, "loss": 1.4196, "step": 16836 }, { "epoch": 42.0925, "grad_norm": 5.09375, "learning_rate": 1.2999935628621099e-05, "loss": 1.4495, "step": 16837 }, { "epoch": 42.095, "grad_norm": 7.3125, "learning_rate": 1.29991632232349e-05, "loss": 1.4174, "step": 16838 }, { "epoch": 42.0975, "grad_norm": 4.84375, "learning_rate": 1.2998390798186304e-05, "loss": 1.3162, "step": 16839 }, { "epoch": 42.1, "grad_norm": 9.5, "learning_rate": 1.299761835348038e-05, "loss": 1.3058, "step": 16840 }, { "epoch": 42.1025, "grad_norm": 6.25, "learning_rate": 1.299684588912219e-05, "loss": 1.2671, "step": 16841 }, { "epoch": 42.105, "grad_norm": 5.6875, "learning_rate": 1.2996073405116794e-05, "loss": 1.2197, "step": 16842 }, { "epoch": 42.1075, "grad_norm": 10.625, "learning_rate": 1.2995300901469267e-05, "loss": 1.1965, "step": 16843 }, { "epoch": 42.11, "grad_norm": 7.0, "learning_rate": 1.2994528378184663e-05, "loss": 1.2927, "step": 16844 }, { "epoch": 42.1125, "grad_norm": 5.90625, "learning_rate": 1.2993755835268054e-05, "loss": 1.4445, "step": 16845 }, { "epoch": 42.115, "grad_norm": 5.9375, "learning_rate": 1.29929832727245e-05, "loss": 1.3854, "step": 16846 }, { "epoch": 42.1175, "grad_norm": 9.125, "learning_rate": 1.2992210690559067e-05, "loss": 1.4975, "step": 16847 }, { "epoch": 42.12, "grad_norm": 8.4375, "learning_rate": 1.2991438088776818e-05, "loss": 1.4082, "step": 16848 }, { "epoch": 42.1225, "grad_norm": 7.8125, "learning_rate": 1.2990665467382824e-05, "loss": 1.5432, "step": 16849 }, { "epoch": 42.125, "grad_norm": 6.1875, "learning_rate": 1.2989892826382144e-05, "loss": 1.1669, "step": 16850 }, { "epoch": 42.1275, "grad_norm": 7.625, "learning_rate": 1.2989120165779849e-05, "loss": 1.3098, "step": 16851 }, { "epoch": 42.13, "grad_norm": 7.96875, "learning_rate": 1.2988347485581003e-05, "loss": 1.1826, "step": 16852 }, { "epoch": 42.1325, "grad_norm": 6.5, "learning_rate": 1.2987574785790669e-05, "loss": 1.3073, "step": 16853 }, { "epoch": 42.135, "grad_norm": 7.90625, "learning_rate": 1.2986802066413915e-05, "loss": 1.2815, "step": 16854 }, { "epoch": 42.1375, "grad_norm": 6.53125, "learning_rate": 1.2986029327455802e-05, "loss": 1.3342, "step": 16855 }, { "epoch": 42.14, "grad_norm": 6.03125, "learning_rate": 1.2985256568921404e-05, "loss": 1.3424, "step": 16856 }, { "epoch": 42.1425, "grad_norm": 6.1875, "learning_rate": 1.2984483790815783e-05, "loss": 1.1967, "step": 16857 }, { "epoch": 42.145, "grad_norm": 6.71875, "learning_rate": 1.2983710993144006e-05, "loss": 1.3187, "step": 16858 }, { "epoch": 42.1475, "grad_norm": 8.625, "learning_rate": 1.2982938175911138e-05, "loss": 1.4957, "step": 16859 }, { "epoch": 42.15, "grad_norm": 8.5, "learning_rate": 1.2982165339122248e-05, "loss": 1.3685, "step": 16860 }, { "epoch": 42.1525, "grad_norm": 8.0625, "learning_rate": 1.29813924827824e-05, "loss": 1.2896, "step": 16861 }, { "epoch": 42.155, "grad_norm": 9.25, "learning_rate": 1.2980619606896663e-05, "loss": 1.4478, "step": 16862 }, { "epoch": 42.1575, "grad_norm": 6.96875, "learning_rate": 1.2979846711470101e-05, "loss": 1.3125, "step": 16863 }, { "epoch": 42.16, "grad_norm": 7.0, "learning_rate": 1.2979073796507786e-05, "loss": 1.2331, "step": 16864 }, { "epoch": 42.1625, "grad_norm": 5.96875, "learning_rate": 1.297830086201478e-05, "loss": 1.3129, "step": 16865 }, { "epoch": 42.165, "grad_norm": 7.71875, "learning_rate": 1.2977527907996154e-05, "loss": 1.1292, "step": 16866 }, { "epoch": 42.1675, "grad_norm": 10.0625, "learning_rate": 1.2976754934456976e-05, "loss": 1.3006, "step": 16867 }, { "epoch": 42.17, "grad_norm": 7.1875, "learning_rate": 1.297598194140231e-05, "loss": 1.3681, "step": 16868 }, { "epoch": 42.1725, "grad_norm": 7.15625, "learning_rate": 1.2975208928837227e-05, "loss": 1.3648, "step": 16869 }, { "epoch": 42.175, "grad_norm": 12.625, "learning_rate": 1.297443589676679e-05, "loss": 1.3803, "step": 16870 }, { "epoch": 42.1775, "grad_norm": 7.21875, "learning_rate": 1.2973662845196073e-05, "loss": 1.5066, "step": 16871 }, { "epoch": 42.18, "grad_norm": 9.875, "learning_rate": 1.297288977413014e-05, "loss": 1.456, "step": 16872 }, { "epoch": 42.1825, "grad_norm": 8.375, "learning_rate": 1.2972116683574063e-05, "loss": 1.4424, "step": 16873 }, { "epoch": 42.185, "grad_norm": 8.375, "learning_rate": 1.2971343573532907e-05, "loss": 1.1735, "step": 16874 }, { "epoch": 42.1875, "grad_norm": 8.6875, "learning_rate": 1.2970570444011739e-05, "loss": 1.3064, "step": 16875 }, { "epoch": 42.19, "grad_norm": 7.25, "learning_rate": 1.2969797295015632e-05, "loss": 1.3636, "step": 16876 }, { "epoch": 42.1925, "grad_norm": 8.625, "learning_rate": 1.2969024126549653e-05, "loss": 1.2861, "step": 16877 }, { "epoch": 42.195, "grad_norm": 7.34375, "learning_rate": 1.296825093861887e-05, "loss": 1.4324, "step": 16878 }, { "epoch": 42.1975, "grad_norm": 5.65625, "learning_rate": 1.2967477731228356e-05, "loss": 1.2011, "step": 16879 }, { "epoch": 42.2, "grad_norm": 9.0625, "learning_rate": 1.296670450438317e-05, "loss": 1.3133, "step": 16880 }, { "epoch": 42.2025, "grad_norm": 7.875, "learning_rate": 1.2965931258088392e-05, "loss": 1.2777, "step": 16881 }, { "epoch": 42.205, "grad_norm": 8.1875, "learning_rate": 1.2965157992349084e-05, "loss": 1.3344, "step": 16882 }, { "epoch": 42.2075, "grad_norm": 7.0, "learning_rate": 1.2964384707170324e-05, "loss": 1.3185, "step": 16883 }, { "epoch": 42.21, "grad_norm": 10.0625, "learning_rate": 1.2963611402557174e-05, "loss": 1.3054, "step": 16884 }, { "epoch": 42.2125, "grad_norm": 11.0625, "learning_rate": 1.2962838078514706e-05, "loss": 1.3863, "step": 16885 }, { "epoch": 42.215, "grad_norm": 9.625, "learning_rate": 1.2962064735047993e-05, "loss": 1.3856, "step": 16886 }, { "epoch": 42.2175, "grad_norm": 6.5625, "learning_rate": 1.2961291372162097e-05, "loss": 1.2022, "step": 16887 }, { "epoch": 42.22, "grad_norm": 5.03125, "learning_rate": 1.2960517989862096e-05, "loss": 1.2824, "step": 16888 }, { "epoch": 42.2225, "grad_norm": 7.5, "learning_rate": 1.2959744588153057e-05, "loss": 1.4771, "step": 16889 }, { "epoch": 42.225, "grad_norm": 6.90625, "learning_rate": 1.2958971167040054e-05, "loss": 1.4101, "step": 16890 }, { "epoch": 42.2275, "grad_norm": 6.125, "learning_rate": 1.295819772652815e-05, "loss": 1.1639, "step": 16891 }, { "epoch": 42.23, "grad_norm": 8.1875, "learning_rate": 1.2957424266622422e-05, "loss": 1.3176, "step": 16892 }, { "epoch": 42.2325, "grad_norm": 6.3125, "learning_rate": 1.2956650787327938e-05, "loss": 1.2065, "step": 16893 }, { "epoch": 42.235, "grad_norm": 6.15625, "learning_rate": 1.2955877288649775e-05, "loss": 1.307, "step": 16894 }, { "epoch": 42.2375, "grad_norm": 13.875, "learning_rate": 1.2955103770592991e-05, "loss": 1.297, "step": 16895 }, { "epoch": 42.24, "grad_norm": 7.40625, "learning_rate": 1.2954330233162669e-05, "loss": 1.3593, "step": 16896 }, { "epoch": 42.2425, "grad_norm": 4.75, "learning_rate": 1.2953556676363874e-05, "loss": 1.4385, "step": 16897 }, { "epoch": 42.245, "grad_norm": 6.53125, "learning_rate": 1.2952783100201683e-05, "loss": 1.328, "step": 16898 }, { "epoch": 42.2475, "grad_norm": 8.8125, "learning_rate": 1.2952009504681161e-05, "loss": 1.5209, "step": 16899 }, { "epoch": 42.25, "grad_norm": 9.0, "learning_rate": 1.2951235889807386e-05, "loss": 1.3268, "step": 16900 }, { "epoch": 42.2525, "grad_norm": 7.0625, "learning_rate": 1.2950462255585422e-05, "loss": 1.3444, "step": 16901 }, { "epoch": 42.255, "grad_norm": 14.0625, "learning_rate": 1.2949688602020349e-05, "loss": 1.3614, "step": 16902 }, { "epoch": 42.2575, "grad_norm": 8.5625, "learning_rate": 1.2948914929117235e-05, "loss": 1.28, "step": 16903 }, { "epoch": 42.26, "grad_norm": 6.96875, "learning_rate": 1.2948141236881154e-05, "loss": 1.3869, "step": 16904 }, { "epoch": 42.2625, "grad_norm": 9.625, "learning_rate": 1.2947367525317175e-05, "loss": 1.3945, "step": 16905 }, { "epoch": 42.265, "grad_norm": 6.96875, "learning_rate": 1.2946593794430373e-05, "loss": 1.2484, "step": 16906 }, { "epoch": 42.2675, "grad_norm": 6.4375, "learning_rate": 1.2945820044225818e-05, "loss": 1.3177, "step": 16907 }, { "epoch": 42.27, "grad_norm": 8.4375, "learning_rate": 1.2945046274708588e-05, "loss": 1.5039, "step": 16908 }, { "epoch": 42.2725, "grad_norm": 7.375, "learning_rate": 1.2944272485883751e-05, "loss": 1.4641, "step": 16909 }, { "epoch": 42.275, "grad_norm": 5.09375, "learning_rate": 1.2943498677756381e-05, "loss": 1.3233, "step": 16910 }, { "epoch": 42.2775, "grad_norm": 5.8125, "learning_rate": 1.2942724850331554e-05, "loss": 1.2482, "step": 16911 }, { "epoch": 42.28, "grad_norm": 6.9375, "learning_rate": 1.2941951003614337e-05, "loss": 1.3083, "step": 16912 }, { "epoch": 42.2825, "grad_norm": 6.0625, "learning_rate": 1.294117713760981e-05, "loss": 1.3738, "step": 16913 }, { "epoch": 42.285, "grad_norm": 6.125, "learning_rate": 1.294040325232304e-05, "loss": 1.3393, "step": 16914 }, { "epoch": 42.2875, "grad_norm": 6.78125, "learning_rate": 1.2939629347759109e-05, "loss": 1.3032, "step": 16915 }, { "epoch": 42.29, "grad_norm": 9.5, "learning_rate": 1.2938855423923081e-05, "loss": 1.2521, "step": 16916 }, { "epoch": 42.2925, "grad_norm": 8.4375, "learning_rate": 1.2938081480820035e-05, "loss": 1.4884, "step": 16917 }, { "epoch": 42.295, "grad_norm": 7.46875, "learning_rate": 1.2937307518455047e-05, "loss": 1.2472, "step": 16918 }, { "epoch": 42.2975, "grad_norm": 6.96875, "learning_rate": 1.2936533536833188e-05, "loss": 1.308, "step": 16919 }, { "epoch": 42.3, "grad_norm": 5.9375, "learning_rate": 1.2935759535959528e-05, "loss": 1.235, "step": 16920 }, { "epoch": 42.3025, "grad_norm": 8.375, "learning_rate": 1.2934985515839151e-05, "loss": 1.3515, "step": 16921 }, { "epoch": 42.305, "grad_norm": 8.5625, "learning_rate": 1.2934211476477125e-05, "loss": 1.2709, "step": 16922 }, { "epoch": 42.3075, "grad_norm": 6.09375, "learning_rate": 1.2933437417878524e-05, "loss": 1.4363, "step": 16923 }, { "epoch": 42.31, "grad_norm": 7.625, "learning_rate": 1.2932663340048429e-05, "loss": 1.342, "step": 16924 }, { "epoch": 42.3125, "grad_norm": 7.15625, "learning_rate": 1.2931889242991906e-05, "loss": 1.3364, "step": 16925 }, { "epoch": 42.315, "grad_norm": 6.84375, "learning_rate": 1.2931115126714038e-05, "loss": 1.4846, "step": 16926 }, { "epoch": 42.3175, "grad_norm": 10.6875, "learning_rate": 1.2930340991219895e-05, "loss": 1.5895, "step": 16927 }, { "epoch": 42.32, "grad_norm": 8.9375, "learning_rate": 1.2929566836514556e-05, "loss": 1.2642, "step": 16928 }, { "epoch": 42.3225, "grad_norm": 6.375, "learning_rate": 1.2928792662603087e-05, "loss": 1.3094, "step": 16929 }, { "epoch": 42.325, "grad_norm": 7.875, "learning_rate": 1.292801846949058e-05, "loss": 1.3711, "step": 16930 }, { "epoch": 42.3275, "grad_norm": 6.875, "learning_rate": 1.2927244257182094e-05, "loss": 1.4306, "step": 16931 }, { "epoch": 42.33, "grad_norm": 11.0, "learning_rate": 1.2926470025682715e-05, "loss": 1.2976, "step": 16932 }, { "epoch": 42.3325, "grad_norm": 11.25, "learning_rate": 1.2925695774997515e-05, "loss": 1.3965, "step": 16933 }, { "epoch": 42.335, "grad_norm": 5.5625, "learning_rate": 1.2924921505131569e-05, "loss": 1.3055, "step": 16934 }, { "epoch": 42.3375, "grad_norm": 7.34375, "learning_rate": 1.2924147216089957e-05, "loss": 1.5517, "step": 16935 }, { "epoch": 42.34, "grad_norm": 7.78125, "learning_rate": 1.2923372907877752e-05, "loss": 1.2891, "step": 16936 }, { "epoch": 42.3425, "grad_norm": 5.75, "learning_rate": 1.2922598580500029e-05, "loss": 1.1637, "step": 16937 }, { "epoch": 42.345, "grad_norm": 7.8125, "learning_rate": 1.292182423396187e-05, "loss": 1.2856, "step": 16938 }, { "epoch": 42.3475, "grad_norm": 6.8125, "learning_rate": 1.2921049868268347e-05, "loss": 1.2337, "step": 16939 }, { "epoch": 42.35, "grad_norm": 12.0, "learning_rate": 1.2920275483424538e-05, "loss": 1.3469, "step": 16940 }, { "epoch": 42.3525, "grad_norm": 6.8125, "learning_rate": 1.291950107943552e-05, "loss": 1.3728, "step": 16941 }, { "epoch": 42.355, "grad_norm": 7.25, "learning_rate": 1.2918726656306369e-05, "loss": 1.3551, "step": 16942 }, { "epoch": 42.3575, "grad_norm": 12.8125, "learning_rate": 1.2917952214042164e-05, "loss": 1.4241, "step": 16943 }, { "epoch": 42.36, "grad_norm": 8.0625, "learning_rate": 1.291717775264798e-05, "loss": 1.3998, "step": 16944 }, { "epoch": 42.3625, "grad_norm": 9.375, "learning_rate": 1.2916403272128896e-05, "loss": 1.378, "step": 16945 }, { "epoch": 42.365, "grad_norm": 7.3125, "learning_rate": 1.291562877248999e-05, "loss": 1.3596, "step": 16946 }, { "epoch": 42.3675, "grad_norm": 8.5625, "learning_rate": 1.2914854253736337e-05, "loss": 1.354, "step": 16947 }, { "epoch": 42.37, "grad_norm": 9.6875, "learning_rate": 1.2914079715873016e-05, "loss": 1.4098, "step": 16948 }, { "epoch": 42.3725, "grad_norm": 8.3125, "learning_rate": 1.2913305158905105e-05, "loss": 1.2327, "step": 16949 }, { "epoch": 42.375, "grad_norm": 6.28125, "learning_rate": 1.2912530582837683e-05, "loss": 1.2658, "step": 16950 }, { "epoch": 42.3775, "grad_norm": 5.59375, "learning_rate": 1.2911755987675825e-05, "loss": 1.1919, "step": 16951 }, { "epoch": 42.38, "grad_norm": 7.5625, "learning_rate": 1.2910981373424614e-05, "loss": 1.2154, "step": 16952 }, { "epoch": 42.3825, "grad_norm": 5.875, "learning_rate": 1.2910206740089124e-05, "loss": 1.2955, "step": 16953 }, { "epoch": 42.385, "grad_norm": 7.59375, "learning_rate": 1.2909432087674434e-05, "loss": 1.4528, "step": 16954 }, { "epoch": 42.3875, "grad_norm": 7.96875, "learning_rate": 1.2908657416185623e-05, "loss": 1.2604, "step": 16955 }, { "epoch": 42.39, "grad_norm": 10.0, "learning_rate": 1.2907882725627776e-05, "loss": 1.3325, "step": 16956 }, { "epoch": 42.3925, "grad_norm": 8.6875, "learning_rate": 1.290710801600596e-05, "loss": 1.4227, "step": 16957 }, { "epoch": 42.395, "grad_norm": 6.0625, "learning_rate": 1.2906333287325265e-05, "loss": 1.4026, "step": 16958 }, { "epoch": 42.3975, "grad_norm": 7.65625, "learning_rate": 1.290555853959076e-05, "loss": 1.2609, "step": 16959 }, { "epoch": 42.4, "grad_norm": 6.75, "learning_rate": 1.2904783772807534e-05, "loss": 1.3149, "step": 16960 }, { "epoch": 42.4025, "grad_norm": 7.59375, "learning_rate": 1.290400898698066e-05, "loss": 1.2409, "step": 16961 }, { "epoch": 42.405, "grad_norm": 8.6875, "learning_rate": 1.2903234182115221e-05, "loss": 1.355, "step": 16962 }, { "epoch": 42.4075, "grad_norm": 8.125, "learning_rate": 1.2902459358216293e-05, "loss": 1.3532, "step": 16963 }, { "epoch": 42.41, "grad_norm": 11.375, "learning_rate": 1.2901684515288956e-05, "loss": 1.3606, "step": 16964 }, { "epoch": 42.4125, "grad_norm": 6.875, "learning_rate": 1.2900909653338296e-05, "loss": 1.3705, "step": 16965 }, { "epoch": 42.415, "grad_norm": 9.3125, "learning_rate": 1.2900134772369387e-05, "loss": 1.4449, "step": 16966 }, { "epoch": 42.4175, "grad_norm": 6.0625, "learning_rate": 1.2899359872387306e-05, "loss": 1.2646, "step": 16967 }, { "epoch": 42.42, "grad_norm": 6.09375, "learning_rate": 1.2898584953397145e-05, "loss": 1.2578, "step": 16968 }, { "epoch": 42.4225, "grad_norm": 8.375, "learning_rate": 1.2897810015403971e-05, "loss": 1.2892, "step": 16969 }, { "epoch": 42.425, "grad_norm": 8.5625, "learning_rate": 1.2897035058412874e-05, "loss": 1.4401, "step": 16970 }, { "epoch": 42.4275, "grad_norm": 12.4375, "learning_rate": 1.289626008242893e-05, "loss": 1.4126, "step": 16971 }, { "epoch": 42.43, "grad_norm": 6.5, "learning_rate": 1.289548508745722e-05, "loss": 1.3735, "step": 16972 }, { "epoch": 42.4325, "grad_norm": 7.46875, "learning_rate": 1.2894710073502825e-05, "loss": 1.3706, "step": 16973 }, { "epoch": 42.435, "grad_norm": 6.875, "learning_rate": 1.2893935040570828e-05, "loss": 1.4069, "step": 16974 }, { "epoch": 42.4375, "grad_norm": 6.4375, "learning_rate": 1.2893159988666309e-05, "loss": 1.2489, "step": 16975 }, { "epoch": 42.44, "grad_norm": 11.4375, "learning_rate": 1.2892384917794347e-05, "loss": 1.3407, "step": 16976 }, { "epoch": 42.4425, "grad_norm": 7.90625, "learning_rate": 1.2891609827960027e-05, "loss": 1.3259, "step": 16977 }, { "epoch": 42.445, "grad_norm": 15.125, "learning_rate": 1.2890834719168428e-05, "loss": 1.2832, "step": 16978 }, { "epoch": 42.4475, "grad_norm": 7.15625, "learning_rate": 1.2890059591424633e-05, "loss": 1.2174, "step": 16979 }, { "epoch": 42.45, "grad_norm": 6.9375, "learning_rate": 1.2889284444733722e-05, "loss": 1.2776, "step": 16980 }, { "epoch": 42.4525, "grad_norm": 7.8125, "learning_rate": 1.288850927910078e-05, "loss": 1.4761, "step": 16981 }, { "epoch": 42.455, "grad_norm": 6.84375, "learning_rate": 1.2887734094530884e-05, "loss": 1.2993, "step": 16982 }, { "epoch": 42.4575, "grad_norm": 10.5625, "learning_rate": 1.288695889102912e-05, "loss": 1.3306, "step": 16983 }, { "epoch": 42.46, "grad_norm": 6.125, "learning_rate": 1.2886183668600567e-05, "loss": 1.3719, "step": 16984 }, { "epoch": 42.4625, "grad_norm": 6.9375, "learning_rate": 1.2885408427250314e-05, "loss": 1.5394, "step": 16985 }, { "epoch": 42.465, "grad_norm": 6.4375, "learning_rate": 1.2884633166983435e-05, "loss": 1.2732, "step": 16986 }, { "epoch": 42.4675, "grad_norm": 7.1875, "learning_rate": 1.2883857887805018e-05, "loss": 1.3978, "step": 16987 }, { "epoch": 42.47, "grad_norm": 9.25, "learning_rate": 1.2883082589720145e-05, "loss": 1.304, "step": 16988 }, { "epoch": 42.4725, "grad_norm": 8.0, "learning_rate": 1.2882307272733895e-05, "loss": 1.3847, "step": 16989 }, { "epoch": 42.475, "grad_norm": 8.5, "learning_rate": 1.2881531936851355e-05, "loss": 1.2838, "step": 16990 }, { "epoch": 42.4775, "grad_norm": 5.75, "learning_rate": 1.2880756582077608e-05, "loss": 1.3334, "step": 16991 }, { "epoch": 42.48, "grad_norm": 9.75, "learning_rate": 1.2879981208417735e-05, "loss": 1.2645, "step": 16992 }, { "epoch": 42.4825, "grad_norm": 6.125, "learning_rate": 1.2879205815876821e-05, "loss": 1.2113, "step": 16993 }, { "epoch": 42.485, "grad_norm": 9.0625, "learning_rate": 1.2878430404459947e-05, "loss": 1.2402, "step": 16994 }, { "epoch": 42.4875, "grad_norm": 7.46875, "learning_rate": 1.2877654974172203e-05, "loss": 1.3006, "step": 16995 }, { "epoch": 42.49, "grad_norm": 8.0625, "learning_rate": 1.2876879525018664e-05, "loss": 1.3612, "step": 16996 }, { "epoch": 42.4925, "grad_norm": 6.53125, "learning_rate": 1.2876104057004417e-05, "loss": 1.3734, "step": 16997 }, { "epoch": 42.495, "grad_norm": 6.71875, "learning_rate": 1.2875328570134553e-05, "loss": 1.1981, "step": 16998 }, { "epoch": 42.4975, "grad_norm": 7.0625, "learning_rate": 1.2874553064414144e-05, "loss": 1.3191, "step": 16999 }, { "epoch": 42.5, "grad_norm": 7.59375, "learning_rate": 1.2873777539848284e-05, "loss": 1.2923, "step": 17000 }, { "epoch": 42.5025, "grad_norm": 8.125, "learning_rate": 1.2873001996442052e-05, "loss": 1.2833, "step": 17001 }, { "epoch": 42.505, "grad_norm": 7.0, "learning_rate": 1.2872226434200532e-05, "loss": 1.2377, "step": 17002 }, { "epoch": 42.5075, "grad_norm": 6.34375, "learning_rate": 1.2871450853128813e-05, "loss": 1.2967, "step": 17003 }, { "epoch": 42.51, "grad_norm": 6.65625, "learning_rate": 1.2870675253231977e-05, "loss": 1.3436, "step": 17004 }, { "epoch": 42.5125, "grad_norm": 7.625, "learning_rate": 1.2869899634515109e-05, "loss": 1.4216, "step": 17005 }, { "epoch": 42.515, "grad_norm": 7.09375, "learning_rate": 1.286912399698329e-05, "loss": 1.2947, "step": 17006 }, { "epoch": 42.5175, "grad_norm": 6.71875, "learning_rate": 1.2868348340641613e-05, "loss": 1.4916, "step": 17007 }, { "epoch": 42.52, "grad_norm": 8.5625, "learning_rate": 1.2867572665495156e-05, "loss": 1.5343, "step": 17008 }, { "epoch": 42.5225, "grad_norm": 8.125, "learning_rate": 1.2866796971549009e-05, "loss": 1.3085, "step": 17009 }, { "epoch": 42.525, "grad_norm": 6.75, "learning_rate": 1.2866021258808257e-05, "loss": 1.3558, "step": 17010 }, { "epoch": 42.5275, "grad_norm": 11.125, "learning_rate": 1.2865245527277986e-05, "loss": 1.2803, "step": 17011 }, { "epoch": 42.53, "grad_norm": 9.4375, "learning_rate": 1.2864469776963275e-05, "loss": 1.209, "step": 17012 }, { "epoch": 42.5325, "grad_norm": 8.1875, "learning_rate": 1.2863694007869218e-05, "loss": 1.292, "step": 17013 }, { "epoch": 42.535, "grad_norm": 9.9375, "learning_rate": 1.2862918220000895e-05, "loss": 1.3286, "step": 17014 }, { "epoch": 42.5375, "grad_norm": 5.875, "learning_rate": 1.2862142413363398e-05, "loss": 1.2327, "step": 17015 }, { "epoch": 42.54, "grad_norm": 7.25, "learning_rate": 1.286136658796181e-05, "loss": 1.388, "step": 17016 }, { "epoch": 42.5425, "grad_norm": 6.71875, "learning_rate": 1.2860590743801216e-05, "loss": 1.2328, "step": 17017 }, { "epoch": 42.545, "grad_norm": 7.875, "learning_rate": 1.2859814880886701e-05, "loss": 1.404, "step": 17018 }, { "epoch": 42.5475, "grad_norm": 9.5625, "learning_rate": 1.2859038999223357e-05, "loss": 1.3652, "step": 17019 }, { "epoch": 42.55, "grad_norm": 7.8125, "learning_rate": 1.2858263098816265e-05, "loss": 1.3377, "step": 17020 }, { "epoch": 42.5525, "grad_norm": 7.9375, "learning_rate": 1.2857487179670518e-05, "loss": 1.5471, "step": 17021 }, { "epoch": 42.555, "grad_norm": 9.75, "learning_rate": 1.2856711241791198e-05, "loss": 1.2536, "step": 17022 }, { "epoch": 42.5575, "grad_norm": 6.25, "learning_rate": 1.2855935285183394e-05, "loss": 1.35, "step": 17023 }, { "epoch": 42.56, "grad_norm": 8.5, "learning_rate": 1.285515930985219e-05, "loss": 1.3101, "step": 17024 }, { "epoch": 42.5625, "grad_norm": 7.09375, "learning_rate": 1.2854383315802679e-05, "loss": 1.3344, "step": 17025 }, { "epoch": 42.565, "grad_norm": 7.0625, "learning_rate": 1.2853607303039943e-05, "loss": 1.2459, "step": 17026 }, { "epoch": 42.5675, "grad_norm": 9.375, "learning_rate": 1.2852831271569073e-05, "loss": 1.3423, "step": 17027 }, { "epoch": 42.57, "grad_norm": 8.0, "learning_rate": 1.2852055221395156e-05, "loss": 1.4506, "step": 17028 }, { "epoch": 42.5725, "grad_norm": 6.28125, "learning_rate": 1.2851279152523276e-05, "loss": 1.2479, "step": 17029 }, { "epoch": 42.575, "grad_norm": 10.375, "learning_rate": 1.2850503064958528e-05, "loss": 1.4218, "step": 17030 }, { "epoch": 42.5775, "grad_norm": 8.375, "learning_rate": 1.2849726958705992e-05, "loss": 1.3551, "step": 17031 }, { "epoch": 42.58, "grad_norm": 9.0, "learning_rate": 1.2848950833770764e-05, "loss": 1.4592, "step": 17032 }, { "epoch": 42.5825, "grad_norm": 6.5625, "learning_rate": 1.2848174690157926e-05, "loss": 1.4022, "step": 17033 }, { "epoch": 42.585, "grad_norm": 11.0625, "learning_rate": 1.284739852787257e-05, "loss": 1.426, "step": 17034 }, { "epoch": 42.5875, "grad_norm": 8.25, "learning_rate": 1.284662234691978e-05, "loss": 1.4012, "step": 17035 }, { "epoch": 42.59, "grad_norm": 7.75, "learning_rate": 1.284584614730465e-05, "loss": 1.3304, "step": 17036 }, { "epoch": 42.5925, "grad_norm": 6.3125, "learning_rate": 1.2845069929032267e-05, "loss": 1.3264, "step": 17037 }, { "epoch": 42.595, "grad_norm": 5.5625, "learning_rate": 1.2844293692107722e-05, "loss": 1.2384, "step": 17038 }, { "epoch": 42.5975, "grad_norm": 6.78125, "learning_rate": 1.2843517436536095e-05, "loss": 1.3939, "step": 17039 }, { "epoch": 42.6, "grad_norm": 6.875, "learning_rate": 1.2842741162322487e-05, "loss": 1.4119, "step": 17040 }, { "epoch": 42.6025, "grad_norm": 7.09375, "learning_rate": 1.2841964869471978e-05, "loss": 1.1979, "step": 17041 }, { "epoch": 42.605, "grad_norm": 11.75, "learning_rate": 1.2841188557989666e-05, "loss": 1.4109, "step": 17042 }, { "epoch": 42.6075, "grad_norm": 8.8125, "learning_rate": 1.2840412227880629e-05, "loss": 1.3935, "step": 17043 }, { "epoch": 42.61, "grad_norm": 6.65625, "learning_rate": 1.2839635879149967e-05, "loss": 1.2702, "step": 17044 }, { "epoch": 42.6125, "grad_norm": 7.5, "learning_rate": 1.2838859511802766e-05, "loss": 1.3773, "step": 17045 }, { "epoch": 42.615, "grad_norm": 8.4375, "learning_rate": 1.2838083125844112e-05, "loss": 1.1139, "step": 17046 }, { "epoch": 42.6175, "grad_norm": 7.96875, "learning_rate": 1.2837306721279103e-05, "loss": 1.3538, "step": 17047 }, { "epoch": 42.62, "grad_norm": 5.71875, "learning_rate": 1.2836530298112824e-05, "loss": 1.3703, "step": 17048 }, { "epoch": 42.6225, "grad_norm": 5.875, "learning_rate": 1.2835753856350364e-05, "loss": 1.4294, "step": 17049 }, { "epoch": 42.625, "grad_norm": 5.53125, "learning_rate": 1.2834977395996817e-05, "loss": 1.284, "step": 17050 }, { "epoch": 42.6275, "grad_norm": 10.4375, "learning_rate": 1.2834200917057271e-05, "loss": 1.4177, "step": 17051 }, { "epoch": 42.63, "grad_norm": 10.1875, "learning_rate": 1.2833424419536815e-05, "loss": 1.2916, "step": 17052 }, { "epoch": 42.6325, "grad_norm": 13.6875, "learning_rate": 1.2832647903440547e-05, "loss": 1.4643, "step": 17053 }, { "epoch": 42.635, "grad_norm": 6.4375, "learning_rate": 1.2831871368773548e-05, "loss": 1.298, "step": 17054 }, { "epoch": 42.6375, "grad_norm": 7.09375, "learning_rate": 1.2831094815540916e-05, "loss": 1.2851, "step": 17055 }, { "epoch": 42.64, "grad_norm": 7.40625, "learning_rate": 1.2830318243747736e-05, "loss": 1.4819, "step": 17056 }, { "epoch": 42.6425, "grad_norm": 6.6875, "learning_rate": 1.2829541653399108e-05, "loss": 1.3141, "step": 17057 }, { "epoch": 42.645, "grad_norm": 6.4375, "learning_rate": 1.2828765044500115e-05, "loss": 1.341, "step": 17058 }, { "epoch": 42.6475, "grad_norm": 8.25, "learning_rate": 1.2827988417055854e-05, "loss": 1.3428, "step": 17059 }, { "epoch": 42.65, "grad_norm": 6.9375, "learning_rate": 1.282721177107141e-05, "loss": 1.3021, "step": 17060 }, { "epoch": 42.6525, "grad_norm": 6.40625, "learning_rate": 1.2826435106551882e-05, "loss": 1.3162, "step": 17061 }, { "epoch": 42.655, "grad_norm": 7.46875, "learning_rate": 1.2825658423502359e-05, "loss": 1.383, "step": 17062 }, { "epoch": 42.6575, "grad_norm": 8.25, "learning_rate": 1.282488172192793e-05, "loss": 1.2759, "step": 17063 }, { "epoch": 42.66, "grad_norm": 10.5625, "learning_rate": 1.2824105001833691e-05, "loss": 1.3248, "step": 17064 }, { "epoch": 42.6625, "grad_norm": 12.5625, "learning_rate": 1.2823328263224732e-05, "loss": 1.2841, "step": 17065 }, { "epoch": 42.665, "grad_norm": 5.65625, "learning_rate": 1.2822551506106148e-05, "loss": 1.392, "step": 17066 }, { "epoch": 42.6675, "grad_norm": 7.5625, "learning_rate": 1.2821774730483028e-05, "loss": 1.4805, "step": 17067 }, { "epoch": 42.67, "grad_norm": 7.90625, "learning_rate": 1.2820997936360464e-05, "loss": 1.4056, "step": 17068 }, { "epoch": 42.6725, "grad_norm": 6.375, "learning_rate": 1.282022112374355e-05, "loss": 1.2467, "step": 17069 }, { "epoch": 42.675, "grad_norm": 8.6875, "learning_rate": 1.2819444292637384e-05, "loss": 1.2855, "step": 17070 }, { "epoch": 42.6775, "grad_norm": 8.8125, "learning_rate": 1.2818667443047049e-05, "loss": 1.3081, "step": 17071 }, { "epoch": 42.68, "grad_norm": 5.65625, "learning_rate": 1.2817890574977648e-05, "loss": 1.389, "step": 17072 }, { "epoch": 42.6825, "grad_norm": 10.875, "learning_rate": 1.2817113688434266e-05, "loss": 1.2772, "step": 17073 }, { "epoch": 42.685, "grad_norm": 5.40625, "learning_rate": 1.2816336783421999e-05, "loss": 1.4158, "step": 17074 }, { "epoch": 42.6875, "grad_norm": 8.5, "learning_rate": 1.281555985994594e-05, "loss": 1.1115, "step": 17075 }, { "epoch": 42.69, "grad_norm": 11.375, "learning_rate": 1.2814782918011183e-05, "loss": 1.3374, "step": 17076 }, { "epoch": 42.6925, "grad_norm": 10.0, "learning_rate": 1.2814005957622826e-05, "loss": 1.381, "step": 17077 }, { "epoch": 42.695, "grad_norm": 8.8125, "learning_rate": 1.2813228978785954e-05, "loss": 1.2083, "step": 17078 }, { "epoch": 42.6975, "grad_norm": 8.0625, "learning_rate": 1.281245198150567e-05, "loss": 1.4683, "step": 17079 }, { "epoch": 42.7, "grad_norm": 8.75, "learning_rate": 1.2811674965787058e-05, "loss": 1.4082, "step": 17080 }, { "epoch": 42.7025, "grad_norm": 8.375, "learning_rate": 1.281089793163522e-05, "loss": 1.3644, "step": 17081 }, { "epoch": 42.705, "grad_norm": 6.34375, "learning_rate": 1.2810120879055247e-05, "loss": 1.1996, "step": 17082 }, { "epoch": 42.7075, "grad_norm": 9.0, "learning_rate": 1.2809343808052237e-05, "loss": 1.2117, "step": 17083 }, { "epoch": 42.71, "grad_norm": 8.5, "learning_rate": 1.2808566718631279e-05, "loss": 1.1793, "step": 17084 }, { "epoch": 42.7125, "grad_norm": 5.65625, "learning_rate": 1.2807789610797472e-05, "loss": 1.3405, "step": 17085 }, { "epoch": 42.715, "grad_norm": 8.3125, "learning_rate": 1.2807012484555903e-05, "loss": 1.2809, "step": 17086 }, { "epoch": 42.7175, "grad_norm": 7.4375, "learning_rate": 1.280623533991168e-05, "loss": 1.3123, "step": 17087 }, { "epoch": 42.72, "grad_norm": 9.5625, "learning_rate": 1.2805458176869885e-05, "loss": 1.2909, "step": 17088 }, { "epoch": 42.7225, "grad_norm": 9.75, "learning_rate": 1.2804680995435622e-05, "loss": 1.3505, "step": 17089 }, { "epoch": 42.725, "grad_norm": 5.90625, "learning_rate": 1.280390379561398e-05, "loss": 1.2553, "step": 17090 }, { "epoch": 42.7275, "grad_norm": 15.1875, "learning_rate": 1.2803126577410057e-05, "loss": 1.4126, "step": 17091 }, { "epoch": 42.73, "grad_norm": 6.78125, "learning_rate": 1.2802349340828948e-05, "loss": 1.3929, "step": 17092 }, { "epoch": 42.7325, "grad_norm": 6.71875, "learning_rate": 1.280157208587575e-05, "loss": 1.1209, "step": 17093 }, { "epoch": 42.735, "grad_norm": 7.90625, "learning_rate": 1.2800794812555557e-05, "loss": 1.3978, "step": 17094 }, { "epoch": 42.7375, "grad_norm": 5.75, "learning_rate": 1.2800017520873467e-05, "loss": 1.1538, "step": 17095 }, { "epoch": 42.74, "grad_norm": 9.0625, "learning_rate": 1.2799240210834572e-05, "loss": 1.2279, "step": 17096 }, { "epoch": 42.7425, "grad_norm": 8.5625, "learning_rate": 1.2798462882443972e-05, "loss": 1.2629, "step": 17097 }, { "epoch": 42.745, "grad_norm": 7.4375, "learning_rate": 1.2797685535706763e-05, "loss": 1.3274, "step": 17098 }, { "epoch": 42.7475, "grad_norm": 10.4375, "learning_rate": 1.2796908170628033e-05, "loss": 1.4163, "step": 17099 }, { "epoch": 42.75, "grad_norm": 8.8125, "learning_rate": 1.279613078721289e-05, "loss": 1.3653, "step": 17100 }, { "epoch": 42.7525, "grad_norm": 7.5, "learning_rate": 1.2795353385466425e-05, "loss": 1.297, "step": 17101 }, { "epoch": 42.755, "grad_norm": 7.46875, "learning_rate": 1.2794575965393734e-05, "loss": 1.1941, "step": 17102 }, { "epoch": 42.7575, "grad_norm": 7.1875, "learning_rate": 1.2793798526999916e-05, "loss": 1.3141, "step": 17103 }, { "epoch": 42.76, "grad_norm": 6.9375, "learning_rate": 1.2793021070290065e-05, "loss": 1.4298, "step": 17104 }, { "epoch": 42.7625, "grad_norm": 7.59375, "learning_rate": 1.2792243595269282e-05, "loss": 1.224, "step": 17105 }, { "epoch": 42.765, "grad_norm": 9.3125, "learning_rate": 1.279146610194266e-05, "loss": 1.3673, "step": 17106 }, { "epoch": 42.7675, "grad_norm": 6.53125, "learning_rate": 1.27906885903153e-05, "loss": 1.3824, "step": 17107 }, { "epoch": 42.77, "grad_norm": 7.375, "learning_rate": 1.2789911060392295e-05, "loss": 1.2236, "step": 17108 }, { "epoch": 42.7725, "grad_norm": 7.25, "learning_rate": 1.2789133512178744e-05, "loss": 1.2484, "step": 17109 }, { "epoch": 42.775, "grad_norm": 9.125, "learning_rate": 1.2788355945679749e-05, "loss": 1.4122, "step": 17110 }, { "epoch": 42.7775, "grad_norm": 7.25, "learning_rate": 1.2787578360900399e-05, "loss": 1.3427, "step": 17111 }, { "epoch": 42.78, "grad_norm": 7.5625, "learning_rate": 1.2786800757845802e-05, "loss": 1.2262, "step": 17112 }, { "epoch": 42.7825, "grad_norm": 6.9375, "learning_rate": 1.2786023136521046e-05, "loss": 1.314, "step": 17113 }, { "epoch": 42.785, "grad_norm": 8.4375, "learning_rate": 1.2785245496931237e-05, "loss": 1.4018, "step": 17114 }, { "epoch": 42.7875, "grad_norm": 11.375, "learning_rate": 1.2784467839081469e-05, "loss": 1.4155, "step": 17115 }, { "epoch": 42.79, "grad_norm": 5.875, "learning_rate": 1.2783690162976839e-05, "loss": 1.3392, "step": 17116 }, { "epoch": 42.7925, "grad_norm": 6.65625, "learning_rate": 1.2782912468622452e-05, "loss": 1.3666, "step": 17117 }, { "epoch": 42.795, "grad_norm": 7.0625, "learning_rate": 1.27821347560234e-05, "loss": 1.3019, "step": 17118 }, { "epoch": 42.7975, "grad_norm": 8.625, "learning_rate": 1.2781357025184785e-05, "loss": 1.3714, "step": 17119 }, { "epoch": 42.8, "grad_norm": 9.1875, "learning_rate": 1.2780579276111702e-05, "loss": 1.3363, "step": 17120 }, { "epoch": 42.8025, "grad_norm": 7.65625, "learning_rate": 1.2779801508809256e-05, "loss": 1.3226, "step": 17121 }, { "epoch": 42.805, "grad_norm": 10.625, "learning_rate": 1.277902372328254e-05, "loss": 1.3665, "step": 17122 }, { "epoch": 42.8075, "grad_norm": 6.46875, "learning_rate": 1.2778245919536657e-05, "loss": 1.369, "step": 17123 }, { "epoch": 42.81, "grad_norm": 7.28125, "learning_rate": 1.2777468097576701e-05, "loss": 1.3653, "step": 17124 }, { "epoch": 42.8125, "grad_norm": 8.0625, "learning_rate": 1.2776690257407782e-05, "loss": 1.338, "step": 17125 }, { "epoch": 42.815, "grad_norm": 6.53125, "learning_rate": 1.2775912399034987e-05, "loss": 1.3851, "step": 17126 }, { "epoch": 42.8175, "grad_norm": 9.0625, "learning_rate": 1.2775134522463426e-05, "loss": 1.4377, "step": 17127 }, { "epoch": 42.82, "grad_norm": 9.125, "learning_rate": 1.2774356627698193e-05, "loss": 1.3751, "step": 17128 }, { "epoch": 42.8225, "grad_norm": 5.6875, "learning_rate": 1.2773578714744387e-05, "loss": 1.3464, "step": 17129 }, { "epoch": 42.825, "grad_norm": 6.15625, "learning_rate": 1.277280078360711e-05, "loss": 1.278, "step": 17130 }, { "epoch": 42.8275, "grad_norm": 8.6875, "learning_rate": 1.2772022834291464e-05, "loss": 1.381, "step": 17131 }, { "epoch": 42.83, "grad_norm": 6.125, "learning_rate": 1.2771244866802546e-05, "loss": 1.2419, "step": 17132 }, { "epoch": 42.8325, "grad_norm": 5.28125, "learning_rate": 1.2770466881145457e-05, "loss": 1.4785, "step": 17133 }, { "epoch": 42.835, "grad_norm": 6.5, "learning_rate": 1.27696888773253e-05, "loss": 1.1416, "step": 17134 }, { "epoch": 42.8375, "grad_norm": 7.125, "learning_rate": 1.2768910855347172e-05, "loss": 1.4111, "step": 17135 }, { "epoch": 42.84, "grad_norm": 9.3125, "learning_rate": 1.2768132815216174e-05, "loss": 1.5005, "step": 17136 }, { "epoch": 42.8425, "grad_norm": 9.0625, "learning_rate": 1.2767354756937409e-05, "loss": 1.3221, "step": 17137 }, { "epoch": 42.845, "grad_norm": 7.75, "learning_rate": 1.276657668051598e-05, "loss": 1.3953, "step": 17138 }, { "epoch": 42.8475, "grad_norm": 10.625, "learning_rate": 1.2765798585956981e-05, "loss": 1.4234, "step": 17139 }, { "epoch": 42.85, "grad_norm": 7.8125, "learning_rate": 1.276502047326552e-05, "loss": 1.2678, "step": 17140 }, { "epoch": 42.8525, "grad_norm": 6.09375, "learning_rate": 1.2764242342446691e-05, "loss": 1.3117, "step": 17141 }, { "epoch": 42.855, "grad_norm": 7.1875, "learning_rate": 1.2763464193505606e-05, "loss": 1.3641, "step": 17142 }, { "epoch": 42.8575, "grad_norm": 7.65625, "learning_rate": 1.2762686026447358e-05, "loss": 1.3362, "step": 17143 }, { "epoch": 42.86, "grad_norm": 6.78125, "learning_rate": 1.2761907841277048e-05, "loss": 1.2579, "step": 17144 }, { "epoch": 42.8625, "grad_norm": 9.625, "learning_rate": 1.2761129637999783e-05, "loss": 1.4854, "step": 17145 }, { "epoch": 42.865, "grad_norm": 6.875, "learning_rate": 1.2760351416620661e-05, "loss": 1.452, "step": 17146 }, { "epoch": 42.8675, "grad_norm": 5.9375, "learning_rate": 1.2759573177144788e-05, "loss": 1.3566, "step": 17147 }, { "epoch": 42.87, "grad_norm": 8.25, "learning_rate": 1.2758794919577264e-05, "loss": 1.334, "step": 17148 }, { "epoch": 42.8725, "grad_norm": 8.375, "learning_rate": 1.275801664392319e-05, "loss": 1.3655, "step": 17149 }, { "epoch": 42.875, "grad_norm": 9.3125, "learning_rate": 1.2757238350187669e-05, "loss": 1.3398, "step": 17150 }, { "epoch": 42.8775, "grad_norm": 8.0625, "learning_rate": 1.2756460038375801e-05, "loss": 1.2945, "step": 17151 }, { "epoch": 42.88, "grad_norm": 5.4375, "learning_rate": 1.2755681708492696e-05, "loss": 1.3979, "step": 17152 }, { "epoch": 42.8825, "grad_norm": 6.84375, "learning_rate": 1.2754903360543449e-05, "loss": 1.2884, "step": 17153 }, { "epoch": 42.885, "grad_norm": 6.25, "learning_rate": 1.2754124994533165e-05, "loss": 1.4031, "step": 17154 }, { "epoch": 42.8875, "grad_norm": 7.1875, "learning_rate": 1.2753346610466953e-05, "loss": 1.2989, "step": 17155 }, { "epoch": 42.89, "grad_norm": 8.5625, "learning_rate": 1.2752568208349905e-05, "loss": 1.4162, "step": 17156 }, { "epoch": 42.8925, "grad_norm": 6.90625, "learning_rate": 1.2751789788187134e-05, "loss": 1.1258, "step": 17157 }, { "epoch": 42.895, "grad_norm": 5.65625, "learning_rate": 1.2751011349983735e-05, "loss": 1.4084, "step": 17158 }, { "epoch": 42.8975, "grad_norm": 6.28125, "learning_rate": 1.275023289374482e-05, "loss": 1.2651, "step": 17159 }, { "epoch": 42.9, "grad_norm": 4.8125, "learning_rate": 1.2749454419475486e-05, "loss": 1.2475, "step": 17160 }, { "epoch": 42.9025, "grad_norm": 6.125, "learning_rate": 1.274867592718084e-05, "loss": 1.4561, "step": 17161 }, { "epoch": 42.905, "grad_norm": 7.5625, "learning_rate": 1.2747897416865985e-05, "loss": 1.4, "step": 17162 }, { "epoch": 42.9075, "grad_norm": 9.6875, "learning_rate": 1.2747118888536023e-05, "loss": 1.3947, "step": 17163 }, { "epoch": 42.91, "grad_norm": 6.78125, "learning_rate": 1.274634034219606e-05, "loss": 1.4285, "step": 17164 }, { "epoch": 42.9125, "grad_norm": 6.8125, "learning_rate": 1.2745561777851198e-05, "loss": 1.2707, "step": 17165 }, { "epoch": 42.915, "grad_norm": 8.125, "learning_rate": 1.2744783195506545e-05, "loss": 1.4925, "step": 17166 }, { "epoch": 42.9175, "grad_norm": 5.34375, "learning_rate": 1.2744004595167204e-05, "loss": 1.3391, "step": 17167 }, { "epoch": 42.92, "grad_norm": 5.09375, "learning_rate": 1.2743225976838277e-05, "loss": 1.1746, "step": 17168 }, { "epoch": 42.9225, "grad_norm": 7.96875, "learning_rate": 1.2742447340524869e-05, "loss": 1.3569, "step": 17169 }, { "epoch": 42.925, "grad_norm": 8.3125, "learning_rate": 1.274166868623209e-05, "loss": 1.3048, "step": 17170 }, { "epoch": 42.9275, "grad_norm": 7.65625, "learning_rate": 1.2740890013965035e-05, "loss": 1.2869, "step": 17171 }, { "epoch": 42.93, "grad_norm": 7.875, "learning_rate": 1.2740111323728818e-05, "loss": 1.3622, "step": 17172 }, { "epoch": 42.9325, "grad_norm": 6.5625, "learning_rate": 1.273933261552854e-05, "loss": 1.363, "step": 17173 }, { "epoch": 42.935, "grad_norm": 6.1875, "learning_rate": 1.2738553889369309e-05, "loss": 1.2534, "step": 17174 }, { "epoch": 42.9375, "grad_norm": 8.25, "learning_rate": 1.2737775145256229e-05, "loss": 1.346, "step": 17175 }, { "epoch": 42.94, "grad_norm": 8.3125, "learning_rate": 1.2736996383194403e-05, "loss": 1.2284, "step": 17176 }, { "epoch": 42.9425, "grad_norm": 6.625, "learning_rate": 1.2736217603188936e-05, "loss": 1.2481, "step": 17177 }, { "epoch": 42.945, "grad_norm": 8.6875, "learning_rate": 1.273543880524494e-05, "loss": 1.4419, "step": 17178 }, { "epoch": 42.9475, "grad_norm": 7.40625, "learning_rate": 1.2734659989367515e-05, "loss": 1.3003, "step": 17179 }, { "epoch": 42.95, "grad_norm": 7.5, "learning_rate": 1.273388115556177e-05, "loss": 1.4746, "step": 17180 }, { "epoch": 42.9525, "grad_norm": 5.0, "learning_rate": 1.2733102303832804e-05, "loss": 1.3043, "step": 17181 }, { "epoch": 42.955, "grad_norm": 7.90625, "learning_rate": 1.2732323434185734e-05, "loss": 1.4613, "step": 17182 }, { "epoch": 42.9575, "grad_norm": 9.0, "learning_rate": 1.2731544546625659e-05, "loss": 1.3306, "step": 17183 }, { "epoch": 42.96, "grad_norm": 6.3125, "learning_rate": 1.2730765641157689e-05, "loss": 1.33, "step": 17184 }, { "epoch": 42.9625, "grad_norm": 7.0625, "learning_rate": 1.2729986717786926e-05, "loss": 1.3111, "step": 17185 }, { "epoch": 42.965, "grad_norm": 8.6875, "learning_rate": 1.272920777651848e-05, "loss": 1.4518, "step": 17186 }, { "epoch": 42.9675, "grad_norm": 9.125, "learning_rate": 1.272842881735746e-05, "loss": 1.3206, "step": 17187 }, { "epoch": 42.97, "grad_norm": 8.1875, "learning_rate": 1.2727649840308965e-05, "loss": 1.349, "step": 17188 }, { "epoch": 42.9725, "grad_norm": 6.9375, "learning_rate": 1.272687084537811e-05, "loss": 1.3578, "step": 17189 }, { "epoch": 42.975, "grad_norm": 9.375, "learning_rate": 1.2726091832569997e-05, "loss": 1.3098, "step": 17190 }, { "epoch": 42.9775, "grad_norm": 5.9375, "learning_rate": 1.2725312801889738e-05, "loss": 1.3808, "step": 17191 }, { "epoch": 42.98, "grad_norm": 10.25, "learning_rate": 1.2724533753342433e-05, "loss": 1.2058, "step": 17192 }, { "epoch": 42.9825, "grad_norm": 6.71875, "learning_rate": 1.27237546869332e-05, "loss": 1.451, "step": 17193 }, { "epoch": 42.985, "grad_norm": 7.375, "learning_rate": 1.2722975602667135e-05, "loss": 1.2309, "step": 17194 }, { "epoch": 42.9875, "grad_norm": 6.375, "learning_rate": 1.2722196500549355e-05, "loss": 1.3237, "step": 17195 }, { "epoch": 42.99, "grad_norm": 7.0, "learning_rate": 1.2721417380584957e-05, "loss": 1.4029, "step": 17196 }, { "epoch": 42.9925, "grad_norm": 7.5625, "learning_rate": 1.2720638242779062e-05, "loss": 1.3885, "step": 17197 }, { "epoch": 42.995, "grad_norm": 9.5625, "learning_rate": 1.2719859087136767e-05, "loss": 1.2895, "step": 17198 }, { "epoch": 42.9975, "grad_norm": 8.8125, "learning_rate": 1.271907991366319e-05, "loss": 1.2638, "step": 17199 }, { "epoch": 43.0, "grad_norm": 5.6875, "learning_rate": 1.2718300722363431e-05, "loss": 1.353, "step": 17200 }, { "epoch": 43.0025, "grad_norm": 9.6875, "learning_rate": 1.2717521513242601e-05, "loss": 1.1888, "step": 17201 }, { "epoch": 43.005, "grad_norm": 9.8125, "learning_rate": 1.2716742286305808e-05, "loss": 1.4126, "step": 17202 }, { "epoch": 43.0075, "grad_norm": 7.3125, "learning_rate": 1.2715963041558164e-05, "loss": 1.425, "step": 17203 }, { "epoch": 43.01, "grad_norm": 6.75, "learning_rate": 1.2715183779004772e-05, "loss": 1.2573, "step": 17204 }, { "epoch": 43.0125, "grad_norm": 6.84375, "learning_rate": 1.2714404498650743e-05, "loss": 1.1977, "step": 17205 }, { "epoch": 43.015, "grad_norm": 7.78125, "learning_rate": 1.2713625200501189e-05, "loss": 1.2489, "step": 17206 }, { "epoch": 43.0175, "grad_norm": 6.9375, "learning_rate": 1.2712845884561217e-05, "loss": 1.3347, "step": 17207 }, { "epoch": 43.02, "grad_norm": 6.1875, "learning_rate": 1.2712066550835936e-05, "loss": 1.3209, "step": 17208 }, { "epoch": 43.0225, "grad_norm": 6.90625, "learning_rate": 1.2711287199330453e-05, "loss": 1.2849, "step": 17209 }, { "epoch": 43.025, "grad_norm": 7.0625, "learning_rate": 1.2710507830049883e-05, "loss": 1.3308, "step": 17210 }, { "epoch": 43.0275, "grad_norm": 6.5, "learning_rate": 1.2709728442999329e-05, "loss": 1.2719, "step": 17211 }, { "epoch": 43.03, "grad_norm": 4.71875, "learning_rate": 1.2708949038183907e-05, "loss": 1.3669, "step": 17212 }, { "epoch": 43.0325, "grad_norm": 8.0625, "learning_rate": 1.2708169615608718e-05, "loss": 1.3921, "step": 17213 }, { "epoch": 43.035, "grad_norm": 4.5625, "learning_rate": 1.2707390175278885e-05, "loss": 1.4489, "step": 17214 }, { "epoch": 43.0375, "grad_norm": 6.9375, "learning_rate": 1.2706610717199505e-05, "loss": 1.3439, "step": 17215 }, { "epoch": 43.04, "grad_norm": 7.25, "learning_rate": 1.2705831241375695e-05, "loss": 1.399, "step": 17216 }, { "epoch": 43.0425, "grad_norm": 6.65625, "learning_rate": 1.2705051747812562e-05, "loss": 1.4062, "step": 17217 }, { "epoch": 43.045, "grad_norm": 9.125, "learning_rate": 1.270427223651522e-05, "loss": 1.2098, "step": 17218 }, { "epoch": 43.0475, "grad_norm": 10.625, "learning_rate": 1.2703492707488778e-05, "loss": 1.1754, "step": 17219 }, { "epoch": 43.05, "grad_norm": 6.5625, "learning_rate": 1.2702713160738344e-05, "loss": 1.3347, "step": 17220 }, { "epoch": 43.0525, "grad_norm": 6.71875, "learning_rate": 1.2701933596269031e-05, "loss": 1.3716, "step": 17221 }, { "epoch": 43.055, "grad_norm": 7.21875, "learning_rate": 1.270115401408595e-05, "loss": 1.3728, "step": 17222 }, { "epoch": 43.0575, "grad_norm": 8.375, "learning_rate": 1.2700374414194212e-05, "loss": 1.2583, "step": 17223 }, { "epoch": 43.06, "grad_norm": 9.0, "learning_rate": 1.2699594796598927e-05, "loss": 1.2858, "step": 17224 }, { "epoch": 43.0625, "grad_norm": 7.03125, "learning_rate": 1.269881516130521e-05, "loss": 1.2377, "step": 17225 }, { "epoch": 43.065, "grad_norm": 6.4375, "learning_rate": 1.269803550831816e-05, "loss": 1.2716, "step": 17226 }, { "epoch": 43.0675, "grad_norm": 6.5625, "learning_rate": 1.2697255837642906e-05, "loss": 1.331, "step": 17227 }, { "epoch": 43.07, "grad_norm": 11.3125, "learning_rate": 1.2696476149284545e-05, "loss": 1.3003, "step": 17228 }, { "epoch": 43.0725, "grad_norm": 6.5625, "learning_rate": 1.2695696443248196e-05, "loss": 1.4899, "step": 17229 }, { "epoch": 43.075, "grad_norm": 6.53125, "learning_rate": 1.269491671953897e-05, "loss": 1.1882, "step": 17230 }, { "epoch": 43.0775, "grad_norm": 7.96875, "learning_rate": 1.2694136978161977e-05, "loss": 1.294, "step": 17231 }, { "epoch": 43.08, "grad_norm": 6.46875, "learning_rate": 1.2693357219122331e-05, "loss": 1.2457, "step": 17232 }, { "epoch": 43.0825, "grad_norm": 7.34375, "learning_rate": 1.269257744242514e-05, "loss": 1.2703, "step": 17233 }, { "epoch": 43.085, "grad_norm": 6.78125, "learning_rate": 1.269179764807552e-05, "loss": 1.4795, "step": 17234 }, { "epoch": 43.0875, "grad_norm": 5.65625, "learning_rate": 1.2691017836078583e-05, "loss": 1.4333, "step": 17235 }, { "epoch": 43.09, "grad_norm": 8.625, "learning_rate": 1.269023800643944e-05, "loss": 1.2791, "step": 17236 }, { "epoch": 43.0925, "grad_norm": 6.53125, "learning_rate": 1.2689458159163206e-05, "loss": 1.3495, "step": 17237 }, { "epoch": 43.095, "grad_norm": 7.4375, "learning_rate": 1.2688678294254993e-05, "loss": 1.1748, "step": 17238 }, { "epoch": 43.0975, "grad_norm": 7.96875, "learning_rate": 1.268789841171991e-05, "loss": 1.129, "step": 17239 }, { "epoch": 43.1, "grad_norm": 10.125, "learning_rate": 1.2687118511563075e-05, "loss": 1.4127, "step": 17240 }, { "epoch": 43.1025, "grad_norm": 7.4375, "learning_rate": 1.2686338593789596e-05, "loss": 1.4067, "step": 17241 }, { "epoch": 43.105, "grad_norm": 7.90625, "learning_rate": 1.2685558658404591e-05, "loss": 1.3358, "step": 17242 }, { "epoch": 43.1075, "grad_norm": 8.375, "learning_rate": 1.2684778705413166e-05, "loss": 1.241, "step": 17243 }, { "epoch": 43.11, "grad_norm": 6.71875, "learning_rate": 1.2683998734820444e-05, "loss": 1.3521, "step": 17244 }, { "epoch": 43.1125, "grad_norm": 6.59375, "learning_rate": 1.2683218746631533e-05, "loss": 1.3951, "step": 17245 }, { "epoch": 43.115, "grad_norm": 5.40625, "learning_rate": 1.2682438740851546e-05, "loss": 1.3487, "step": 17246 }, { "epoch": 43.1175, "grad_norm": 6.25, "learning_rate": 1.2681658717485598e-05, "loss": 1.3179, "step": 17247 }, { "epoch": 43.12, "grad_norm": 6.6875, "learning_rate": 1.2680878676538804e-05, "loss": 1.28, "step": 17248 }, { "epoch": 43.1225, "grad_norm": 7.90625, "learning_rate": 1.2680098618016277e-05, "loss": 1.2989, "step": 17249 }, { "epoch": 43.125, "grad_norm": 8.75, "learning_rate": 1.2679318541923131e-05, "loss": 1.3557, "step": 17250 }, { "epoch": 43.1275, "grad_norm": 8.9375, "learning_rate": 1.2678538448264477e-05, "loss": 1.2509, "step": 17251 }, { "epoch": 43.13, "grad_norm": 7.21875, "learning_rate": 1.2677758337045436e-05, "loss": 1.2571, "step": 17252 }, { "epoch": 43.1325, "grad_norm": 8.75, "learning_rate": 1.2676978208271113e-05, "loss": 1.3561, "step": 17253 }, { "epoch": 43.135, "grad_norm": 8.3125, "learning_rate": 1.2676198061946632e-05, "loss": 1.3886, "step": 17254 }, { "epoch": 43.1375, "grad_norm": 8.8125, "learning_rate": 1.2675417898077105e-05, "loss": 1.411, "step": 17255 }, { "epoch": 43.14, "grad_norm": 8.75, "learning_rate": 1.2674637716667641e-05, "loss": 1.1704, "step": 17256 }, { "epoch": 43.1425, "grad_norm": 6.875, "learning_rate": 1.2673857517723363e-05, "loss": 1.2776, "step": 17257 }, { "epoch": 43.145, "grad_norm": 10.6875, "learning_rate": 1.267307730124938e-05, "loss": 1.3955, "step": 17258 }, { "epoch": 43.1475, "grad_norm": 8.4375, "learning_rate": 1.267229706725081e-05, "loss": 1.3408, "step": 17259 }, { "epoch": 43.15, "grad_norm": 8.0625, "learning_rate": 1.2671516815732767e-05, "loss": 1.3157, "step": 17260 }, { "epoch": 43.1525, "grad_norm": 5.0625, "learning_rate": 1.2670736546700366e-05, "loss": 1.3905, "step": 17261 }, { "epoch": 43.155, "grad_norm": 7.78125, "learning_rate": 1.2669956260158725e-05, "loss": 1.4002, "step": 17262 }, { "epoch": 43.1575, "grad_norm": 8.3125, "learning_rate": 1.2669175956112956e-05, "loss": 1.4569, "step": 17263 }, { "epoch": 43.16, "grad_norm": 7.15625, "learning_rate": 1.2668395634568175e-05, "loss": 1.1971, "step": 17264 }, { "epoch": 43.1625, "grad_norm": 9.3125, "learning_rate": 1.2667615295529502e-05, "loss": 1.39, "step": 17265 }, { "epoch": 43.165, "grad_norm": 9.0625, "learning_rate": 1.2666834939002048e-05, "loss": 1.4119, "step": 17266 }, { "epoch": 43.1675, "grad_norm": 9.125, "learning_rate": 1.2666054564990933e-05, "loss": 1.3651, "step": 17267 }, { "epoch": 43.17, "grad_norm": 5.875, "learning_rate": 1.2665274173501268e-05, "loss": 1.2596, "step": 17268 }, { "epoch": 43.1725, "grad_norm": 7.84375, "learning_rate": 1.2664493764538175e-05, "loss": 1.4149, "step": 17269 }, { "epoch": 43.175, "grad_norm": 6.71875, "learning_rate": 1.2663713338106765e-05, "loss": 1.2455, "step": 17270 }, { "epoch": 43.1775, "grad_norm": 8.875, "learning_rate": 1.2662932894212158e-05, "loss": 1.3089, "step": 17271 }, { "epoch": 43.18, "grad_norm": 6.65625, "learning_rate": 1.266215243285947e-05, "loss": 1.4495, "step": 17272 }, { "epoch": 43.1825, "grad_norm": 8.3125, "learning_rate": 1.2661371954053818e-05, "loss": 1.2751, "step": 17273 }, { "epoch": 43.185, "grad_norm": 6.65625, "learning_rate": 1.2660591457800316e-05, "loss": 1.312, "step": 17274 }, { "epoch": 43.1875, "grad_norm": 7.28125, "learning_rate": 1.2659810944104083e-05, "loss": 1.2299, "step": 17275 }, { "epoch": 43.19, "grad_norm": 12.9375, "learning_rate": 1.2659030412970236e-05, "loss": 1.3436, "step": 17276 }, { "epoch": 43.1925, "grad_norm": 10.25, "learning_rate": 1.2658249864403893e-05, "loss": 1.357, "step": 17277 }, { "epoch": 43.195, "grad_norm": 7.90625, "learning_rate": 1.2657469298410167e-05, "loss": 1.392, "step": 17278 }, { "epoch": 43.1975, "grad_norm": 7.0625, "learning_rate": 1.2656688714994183e-05, "loss": 1.4698, "step": 17279 }, { "epoch": 43.2, "grad_norm": 9.5625, "learning_rate": 1.2655908114161053e-05, "loss": 1.4036, "step": 17280 }, { "epoch": 43.2025, "grad_norm": 6.40625, "learning_rate": 1.2655127495915896e-05, "loss": 1.2615, "step": 17281 }, { "epoch": 43.205, "grad_norm": 9.125, "learning_rate": 1.2654346860263828e-05, "loss": 1.3639, "step": 17282 }, { "epoch": 43.2075, "grad_norm": 6.4375, "learning_rate": 1.2653566207209967e-05, "loss": 1.1787, "step": 17283 }, { "epoch": 43.21, "grad_norm": 7.59375, "learning_rate": 1.2652785536759438e-05, "loss": 1.4404, "step": 17284 }, { "epoch": 43.2125, "grad_norm": 5.84375, "learning_rate": 1.2652004848917346e-05, "loss": 1.353, "step": 17285 }, { "epoch": 43.215, "grad_norm": 6.71875, "learning_rate": 1.265122414368882e-05, "loss": 1.3119, "step": 17286 }, { "epoch": 43.2175, "grad_norm": 7.65625, "learning_rate": 1.2650443421078975e-05, "loss": 1.3815, "step": 17287 }, { "epoch": 43.22, "grad_norm": 5.21875, "learning_rate": 1.2649662681092926e-05, "loss": 1.4009, "step": 17288 }, { "epoch": 43.2225, "grad_norm": 9.25, "learning_rate": 1.2648881923735799e-05, "loss": 1.3286, "step": 17289 }, { "epoch": 43.225, "grad_norm": 6.3125, "learning_rate": 1.2648101149012704e-05, "loss": 1.3521, "step": 17290 }, { "epoch": 43.2275, "grad_norm": 6.53125, "learning_rate": 1.2647320356928765e-05, "loss": 1.4517, "step": 17291 }, { "epoch": 43.23, "grad_norm": 8.625, "learning_rate": 1.26465395474891e-05, "loss": 1.3507, "step": 17292 }, { "epoch": 43.2325, "grad_norm": 8.8125, "learning_rate": 1.2645758720698829e-05, "loss": 1.4239, "step": 17293 }, { "epoch": 43.235, "grad_norm": 7.21875, "learning_rate": 1.2644977876563068e-05, "loss": 1.3714, "step": 17294 }, { "epoch": 43.2375, "grad_norm": 6.96875, "learning_rate": 1.2644197015086936e-05, "loss": 1.1539, "step": 17295 }, { "epoch": 43.24, "grad_norm": 10.625, "learning_rate": 1.2643416136275557e-05, "loss": 1.1683, "step": 17296 }, { "epoch": 43.2425, "grad_norm": 7.75, "learning_rate": 1.264263524013405e-05, "loss": 1.5539, "step": 17297 }, { "epoch": 43.245, "grad_norm": 5.59375, "learning_rate": 1.2641854326667527e-05, "loss": 1.348, "step": 17298 }, { "epoch": 43.2475, "grad_norm": 7.75, "learning_rate": 1.2641073395881117e-05, "loss": 1.4604, "step": 17299 }, { "epoch": 43.25, "grad_norm": 8.1875, "learning_rate": 1.2640292447779932e-05, "loss": 1.4794, "step": 17300 }, { "epoch": 43.2525, "grad_norm": 5.0625, "learning_rate": 1.2639511482369099e-05, "loss": 1.316, "step": 17301 }, { "epoch": 43.255, "grad_norm": 7.34375, "learning_rate": 1.2638730499653731e-05, "loss": 1.3827, "step": 17302 }, { "epoch": 43.2575, "grad_norm": 6.40625, "learning_rate": 1.2637949499638952e-05, "loss": 1.2961, "step": 17303 }, { "epoch": 43.26, "grad_norm": 7.46875, "learning_rate": 1.2637168482329882e-05, "loss": 1.5518, "step": 17304 }, { "epoch": 43.2625, "grad_norm": 6.1875, "learning_rate": 1.2636387447731642e-05, "loss": 1.5072, "step": 17305 }, { "epoch": 43.265, "grad_norm": 6.3125, "learning_rate": 1.263560639584935e-05, "loss": 1.2121, "step": 17306 }, { "epoch": 43.2675, "grad_norm": 6.875, "learning_rate": 1.2634825326688129e-05, "loss": 1.3881, "step": 17307 }, { "epoch": 43.27, "grad_norm": 5.34375, "learning_rate": 1.2634044240253098e-05, "loss": 1.2797, "step": 17308 }, { "epoch": 43.2725, "grad_norm": 8.25, "learning_rate": 1.2633263136549379e-05, "loss": 1.5519, "step": 17309 }, { "epoch": 43.275, "grad_norm": 7.09375, "learning_rate": 1.263248201558209e-05, "loss": 1.3998, "step": 17310 }, { "epoch": 43.2775, "grad_norm": 8.375, "learning_rate": 1.2631700877356357e-05, "loss": 1.4595, "step": 17311 }, { "epoch": 43.28, "grad_norm": 7.4375, "learning_rate": 1.2630919721877299e-05, "loss": 1.2838, "step": 17312 }, { "epoch": 43.2825, "grad_norm": 6.53125, "learning_rate": 1.2630138549150032e-05, "loss": 1.202, "step": 17313 }, { "epoch": 43.285, "grad_norm": 7.6875, "learning_rate": 1.2629357359179688e-05, "loss": 1.2788, "step": 17314 }, { "epoch": 43.2875, "grad_norm": 8.3125, "learning_rate": 1.2628576151971377e-05, "loss": 1.353, "step": 17315 }, { "epoch": 43.29, "grad_norm": 9.1875, "learning_rate": 1.262779492753023e-05, "loss": 1.2528, "step": 17316 }, { "epoch": 43.2925, "grad_norm": 5.90625, "learning_rate": 1.2627013685861363e-05, "loss": 1.3645, "step": 17317 }, { "epoch": 43.295, "grad_norm": 7.21875, "learning_rate": 1.2626232426969898e-05, "loss": 1.3941, "step": 17318 }, { "epoch": 43.2975, "grad_norm": 5.25, "learning_rate": 1.2625451150860961e-05, "loss": 1.3753, "step": 17319 }, { "epoch": 43.3, "grad_norm": 6.625, "learning_rate": 1.2624669857539669e-05, "loss": 1.3697, "step": 17320 }, { "epoch": 43.3025, "grad_norm": 8.125, "learning_rate": 1.2623888547011147e-05, "loss": 1.3348, "step": 17321 }, { "epoch": 43.305, "grad_norm": 7.28125, "learning_rate": 1.262310721928052e-05, "loss": 1.3598, "step": 17322 }, { "epoch": 43.3075, "grad_norm": 8.0625, "learning_rate": 1.26223258743529e-05, "loss": 1.3469, "step": 17323 }, { "epoch": 43.31, "grad_norm": 8.0625, "learning_rate": 1.2621544512233422e-05, "loss": 1.3201, "step": 17324 }, { "epoch": 43.3125, "grad_norm": 10.25, "learning_rate": 1.2620763132927201e-05, "loss": 1.4456, "step": 17325 }, { "epoch": 43.315, "grad_norm": 10.125, "learning_rate": 1.2619981736439361e-05, "loss": 1.3649, "step": 17326 }, { "epoch": 43.3175, "grad_norm": 8.9375, "learning_rate": 1.261920032277503e-05, "loss": 1.0758, "step": 17327 }, { "epoch": 43.32, "grad_norm": 9.1875, "learning_rate": 1.261841889193932e-05, "loss": 1.2312, "step": 17328 }, { "epoch": 43.3225, "grad_norm": 7.21875, "learning_rate": 1.2617637443937368e-05, "loss": 1.1946, "step": 17329 }, { "epoch": 43.325, "grad_norm": 7.71875, "learning_rate": 1.2616855978774283e-05, "loss": 1.4475, "step": 17330 }, { "epoch": 43.3275, "grad_norm": 8.9375, "learning_rate": 1.2616074496455199e-05, "loss": 1.2351, "step": 17331 }, { "epoch": 43.33, "grad_norm": 6.9375, "learning_rate": 1.2615292996985234e-05, "loss": 1.3138, "step": 17332 }, { "epoch": 43.3325, "grad_norm": 6.0625, "learning_rate": 1.2614511480369514e-05, "loss": 1.3131, "step": 17333 }, { "epoch": 43.335, "grad_norm": 11.5, "learning_rate": 1.2613729946613158e-05, "loss": 1.4252, "step": 17334 }, { "epoch": 43.3375, "grad_norm": 8.875, "learning_rate": 1.2612948395721296e-05, "loss": 1.2721, "step": 17335 }, { "epoch": 43.34, "grad_norm": 8.5, "learning_rate": 1.2612166827699049e-05, "loss": 1.3974, "step": 17336 }, { "epoch": 43.3425, "grad_norm": 12.3125, "learning_rate": 1.2611385242551542e-05, "loss": 1.4017, "step": 17337 }, { "epoch": 43.345, "grad_norm": 4.625, "learning_rate": 1.2610603640283891e-05, "loss": 1.2531, "step": 17338 }, { "epoch": 43.3475, "grad_norm": 6.90625, "learning_rate": 1.2609822020901235e-05, "loss": 1.3166, "step": 17339 }, { "epoch": 43.35, "grad_norm": 5.1875, "learning_rate": 1.2609040384408685e-05, "loss": 1.339, "step": 17340 }, { "epoch": 43.3525, "grad_norm": 6.90625, "learning_rate": 1.2608258730811373e-05, "loss": 1.3468, "step": 17341 }, { "epoch": 43.355, "grad_norm": 6.71875, "learning_rate": 1.260747706011442e-05, "loss": 1.2858, "step": 17342 }, { "epoch": 43.3575, "grad_norm": 8.875, "learning_rate": 1.2606695372322955e-05, "loss": 1.2267, "step": 17343 }, { "epoch": 43.36, "grad_norm": 8.0, "learning_rate": 1.2605913667442096e-05, "loss": 1.3308, "step": 17344 }, { "epoch": 43.3625, "grad_norm": 8.3125, "learning_rate": 1.2605131945476972e-05, "loss": 1.2848, "step": 17345 }, { "epoch": 43.365, "grad_norm": 11.5625, "learning_rate": 1.2604350206432708e-05, "loss": 1.3582, "step": 17346 }, { "epoch": 43.3675, "grad_norm": 11.8125, "learning_rate": 1.2603568450314428e-05, "loss": 1.2623, "step": 17347 }, { "epoch": 43.37, "grad_norm": 5.15625, "learning_rate": 1.2602786677127256e-05, "loss": 1.2184, "step": 17348 }, { "epoch": 43.3725, "grad_norm": 7.25, "learning_rate": 1.260200488687632e-05, "loss": 1.3107, "step": 17349 }, { "epoch": 43.375, "grad_norm": 9.3125, "learning_rate": 1.2601223079566745e-05, "loss": 1.5385, "step": 17350 }, { "epoch": 43.3775, "grad_norm": 7.75, "learning_rate": 1.2600441255203652e-05, "loss": 1.2415, "step": 17351 }, { "epoch": 43.38, "grad_norm": 8.8125, "learning_rate": 1.2599659413792176e-05, "loss": 1.3249, "step": 17352 }, { "epoch": 43.3825, "grad_norm": 10.8125, "learning_rate": 1.259887755533743e-05, "loss": 1.4375, "step": 17353 }, { "epoch": 43.385, "grad_norm": 7.59375, "learning_rate": 1.2598095679844553e-05, "loss": 1.2012, "step": 17354 }, { "epoch": 43.3875, "grad_norm": 7.28125, "learning_rate": 1.2597313787318659e-05, "loss": 1.3696, "step": 17355 }, { "epoch": 43.39, "grad_norm": 7.25, "learning_rate": 1.2596531877764887e-05, "loss": 1.3038, "step": 17356 }, { "epoch": 43.3925, "grad_norm": 8.25, "learning_rate": 1.259574995118835e-05, "loss": 1.2756, "step": 17357 }, { "epoch": 43.395, "grad_norm": 6.5, "learning_rate": 1.2594968007594182e-05, "loss": 1.4075, "step": 17358 }, { "epoch": 43.3975, "grad_norm": 8.75, "learning_rate": 1.2594186046987507e-05, "loss": 1.3032, "step": 17359 }, { "epoch": 43.4, "grad_norm": 7.21875, "learning_rate": 1.2593404069373452e-05, "loss": 1.3306, "step": 17360 }, { "epoch": 43.4025, "grad_norm": 8.0625, "learning_rate": 1.2592622074757146e-05, "loss": 1.28, "step": 17361 }, { "epoch": 43.405, "grad_norm": 7.71875, "learning_rate": 1.259184006314371e-05, "loss": 1.2121, "step": 17362 }, { "epoch": 43.4075, "grad_norm": 6.96875, "learning_rate": 1.2591058034538276e-05, "loss": 1.2439, "step": 17363 }, { "epoch": 43.41, "grad_norm": 8.25, "learning_rate": 1.2590275988945967e-05, "loss": 1.3986, "step": 17364 }, { "epoch": 43.4125, "grad_norm": 9.625, "learning_rate": 1.2589493926371916e-05, "loss": 1.3063, "step": 17365 }, { "epoch": 43.415, "grad_norm": 6.375, "learning_rate": 1.2588711846821244e-05, "loss": 1.2816, "step": 17366 }, { "epoch": 43.4175, "grad_norm": 8.125, "learning_rate": 1.2587929750299086e-05, "loss": 1.446, "step": 17367 }, { "epoch": 43.42, "grad_norm": 6.78125, "learning_rate": 1.2587147636810558e-05, "loss": 1.2109, "step": 17368 }, { "epoch": 43.4225, "grad_norm": 8.6875, "learning_rate": 1.2586365506360797e-05, "loss": 1.3637, "step": 17369 }, { "epoch": 43.425, "grad_norm": 7.84375, "learning_rate": 1.2585583358954923e-05, "loss": 1.3274, "step": 17370 }, { "epoch": 43.4275, "grad_norm": 5.21875, "learning_rate": 1.258480119459807e-05, "loss": 1.1677, "step": 17371 }, { "epoch": 43.43, "grad_norm": 5.8125, "learning_rate": 1.2584019013295365e-05, "loss": 1.4118, "step": 17372 }, { "epoch": 43.4325, "grad_norm": 6.78125, "learning_rate": 1.2583236815051936e-05, "loss": 1.2414, "step": 17373 }, { "epoch": 43.435, "grad_norm": 10.125, "learning_rate": 1.2582454599872907e-05, "loss": 1.3898, "step": 17374 }, { "epoch": 43.4375, "grad_norm": 10.0, "learning_rate": 1.2581672367763408e-05, "loss": 1.3357, "step": 17375 }, { "epoch": 43.44, "grad_norm": 8.9375, "learning_rate": 1.2580890118728572e-05, "loss": 1.3798, "step": 17376 }, { "epoch": 43.4425, "grad_norm": 7.40625, "learning_rate": 1.2580107852773523e-05, "loss": 1.2418, "step": 17377 }, { "epoch": 43.445, "grad_norm": 5.28125, "learning_rate": 1.2579325569903386e-05, "loss": 1.3648, "step": 17378 }, { "epoch": 43.4475, "grad_norm": 7.34375, "learning_rate": 1.2578543270123297e-05, "loss": 1.3047, "step": 17379 }, { "epoch": 43.45, "grad_norm": 6.40625, "learning_rate": 1.2577760953438382e-05, "loss": 1.4785, "step": 17380 }, { "epoch": 43.4525, "grad_norm": 6.15625, "learning_rate": 1.2576978619853767e-05, "loss": 1.3928, "step": 17381 }, { "epoch": 43.455, "grad_norm": 6.625, "learning_rate": 1.257619626937459e-05, "loss": 1.2824, "step": 17382 }, { "epoch": 43.4575, "grad_norm": 9.9375, "learning_rate": 1.2575413902005967e-05, "loss": 1.3609, "step": 17383 }, { "epoch": 43.46, "grad_norm": 6.15625, "learning_rate": 1.2574631517753036e-05, "loss": 1.2758, "step": 17384 }, { "epoch": 43.4625, "grad_norm": 7.6875, "learning_rate": 1.257384911662092e-05, "loss": 1.3794, "step": 17385 }, { "epoch": 43.465, "grad_norm": 6.21875, "learning_rate": 1.2573066698614755e-05, "loss": 1.3814, "step": 17386 }, { "epoch": 43.4675, "grad_norm": 8.5, "learning_rate": 1.2572284263739667e-05, "loss": 1.3839, "step": 17387 }, { "epoch": 43.47, "grad_norm": 7.78125, "learning_rate": 1.2571501812000789e-05, "loss": 1.4163, "step": 17388 }, { "epoch": 43.4725, "grad_norm": 6.9375, "learning_rate": 1.2570719343403245e-05, "loss": 1.3944, "step": 17389 }, { "epoch": 43.475, "grad_norm": 7.09375, "learning_rate": 1.2569936857952167e-05, "loss": 1.4712, "step": 17390 }, { "epoch": 43.4775, "grad_norm": 8.875, "learning_rate": 1.256915435565269e-05, "loss": 1.3466, "step": 17391 }, { "epoch": 43.48, "grad_norm": 8.25, "learning_rate": 1.2568371836509936e-05, "loss": 1.2821, "step": 17392 }, { "epoch": 43.4825, "grad_norm": 11.0, "learning_rate": 1.2567589300529041e-05, "loss": 1.5584, "step": 17393 }, { "epoch": 43.485, "grad_norm": 7.5, "learning_rate": 1.2566806747715132e-05, "loss": 1.3067, "step": 17394 }, { "epoch": 43.4875, "grad_norm": 9.375, "learning_rate": 1.2566024178073341e-05, "loss": 1.4237, "step": 17395 }, { "epoch": 43.49, "grad_norm": 5.75, "learning_rate": 1.2565241591608798e-05, "loss": 1.4423, "step": 17396 }, { "epoch": 43.4925, "grad_norm": 6.25, "learning_rate": 1.2564458988326635e-05, "loss": 1.4526, "step": 17397 }, { "epoch": 43.495, "grad_norm": 7.78125, "learning_rate": 1.256367636823198e-05, "loss": 1.3439, "step": 17398 }, { "epoch": 43.4975, "grad_norm": 5.6875, "learning_rate": 1.2562893731329968e-05, "loss": 1.1957, "step": 17399 }, { "epoch": 43.5, "grad_norm": 8.5625, "learning_rate": 1.2562111077625723e-05, "loss": 1.4225, "step": 17400 }, { "epoch": 43.5025, "grad_norm": 9.4375, "learning_rate": 1.2561328407124385e-05, "loss": 1.4903, "step": 17401 }, { "epoch": 43.505, "grad_norm": 7.28125, "learning_rate": 1.2560545719831076e-05, "loss": 1.4894, "step": 17402 }, { "epoch": 43.5075, "grad_norm": 8.5, "learning_rate": 1.2559763015750937e-05, "loss": 1.2688, "step": 17403 }, { "epoch": 43.51, "grad_norm": 5.9375, "learning_rate": 1.255898029488909e-05, "loss": 1.2359, "step": 17404 }, { "epoch": 43.5125, "grad_norm": 7.90625, "learning_rate": 1.2558197557250672e-05, "loss": 1.3616, "step": 17405 }, { "epoch": 43.515, "grad_norm": 10.0625, "learning_rate": 1.2557414802840814e-05, "loss": 1.2501, "step": 17406 }, { "epoch": 43.5175, "grad_norm": 10.5, "learning_rate": 1.2556632031664645e-05, "loss": 1.4672, "step": 17407 }, { "epoch": 43.52, "grad_norm": 7.71875, "learning_rate": 1.2555849243727298e-05, "loss": 1.1782, "step": 17408 }, { "epoch": 43.5225, "grad_norm": 6.90625, "learning_rate": 1.2555066439033909e-05, "loss": 1.3135, "step": 17409 }, { "epoch": 43.525, "grad_norm": 9.1875, "learning_rate": 1.2554283617589602e-05, "loss": 1.4269, "step": 17410 }, { "epoch": 43.5275, "grad_norm": 11.8125, "learning_rate": 1.2553500779399519e-05, "loss": 1.2412, "step": 17411 }, { "epoch": 43.53, "grad_norm": 5.875, "learning_rate": 1.2552717924468783e-05, "loss": 1.313, "step": 17412 }, { "epoch": 43.5325, "grad_norm": 8.0625, "learning_rate": 1.2551935052802533e-05, "loss": 1.2932, "step": 17413 }, { "epoch": 43.535, "grad_norm": 6.65625, "learning_rate": 1.25511521644059e-05, "loss": 1.4634, "step": 17414 }, { "epoch": 43.5375, "grad_norm": 9.0625, "learning_rate": 1.2550369259284012e-05, "loss": 1.477, "step": 17415 }, { "epoch": 43.54, "grad_norm": 5.625, "learning_rate": 1.2549586337442006e-05, "loss": 1.4328, "step": 17416 }, { "epoch": 43.5425, "grad_norm": 6.0625, "learning_rate": 1.2548803398885014e-05, "loss": 1.3894, "step": 17417 }, { "epoch": 43.545, "grad_norm": 5.46875, "learning_rate": 1.254802044361817e-05, "loss": 1.5098, "step": 17418 }, { "epoch": 43.5475, "grad_norm": 6.5625, "learning_rate": 1.2547237471646605e-05, "loss": 1.3555, "step": 17419 }, { "epoch": 43.55, "grad_norm": 7.5625, "learning_rate": 1.2546454482975454e-05, "loss": 1.3551, "step": 17420 }, { "epoch": 43.5525, "grad_norm": 5.875, "learning_rate": 1.2545671477609849e-05, "loss": 1.4084, "step": 17421 }, { "epoch": 43.555, "grad_norm": 9.5625, "learning_rate": 1.2544888455554922e-05, "loss": 1.5839, "step": 17422 }, { "epoch": 43.5575, "grad_norm": 6.09375, "learning_rate": 1.254410541681581e-05, "loss": 1.3587, "step": 17423 }, { "epoch": 43.56, "grad_norm": 9.6875, "learning_rate": 1.2543322361397648e-05, "loss": 1.4126, "step": 17424 }, { "epoch": 43.5625, "grad_norm": 8.1875, "learning_rate": 1.2542539289305559e-05, "loss": 1.4601, "step": 17425 }, { "epoch": 43.565, "grad_norm": 10.8125, "learning_rate": 1.254175620054469e-05, "loss": 1.1271, "step": 17426 }, { "epoch": 43.5675, "grad_norm": 7.375, "learning_rate": 1.2540973095120165e-05, "loss": 1.3202, "step": 17427 }, { "epoch": 43.57, "grad_norm": 6.53125, "learning_rate": 1.2540189973037127e-05, "loss": 1.1977, "step": 17428 }, { "epoch": 43.5725, "grad_norm": 5.65625, "learning_rate": 1.25394068343007e-05, "loss": 1.3385, "step": 17429 }, { "epoch": 43.575, "grad_norm": 6.6875, "learning_rate": 1.2538623678916027e-05, "loss": 1.2957, "step": 17430 }, { "epoch": 43.5775, "grad_norm": 11.0625, "learning_rate": 1.2537840506888238e-05, "loss": 1.2416, "step": 17431 }, { "epoch": 43.58, "grad_norm": 7.40625, "learning_rate": 1.2537057318222468e-05, "loss": 1.3128, "step": 17432 }, { "epoch": 43.5825, "grad_norm": 9.9375, "learning_rate": 1.253627411292385e-05, "loss": 1.3282, "step": 17433 }, { "epoch": 43.585, "grad_norm": 8.4375, "learning_rate": 1.2535490890997521e-05, "loss": 1.4472, "step": 17434 }, { "epoch": 43.5875, "grad_norm": 8.9375, "learning_rate": 1.2534707652448616e-05, "loss": 1.4746, "step": 17435 }, { "epoch": 43.59, "grad_norm": 11.5, "learning_rate": 1.2533924397282268e-05, "loss": 1.3615, "step": 17436 }, { "epoch": 43.5925, "grad_norm": 6.84375, "learning_rate": 1.2533141125503616e-05, "loss": 1.2756, "step": 17437 }, { "epoch": 43.595, "grad_norm": 5.25, "learning_rate": 1.2532357837117791e-05, "loss": 1.2307, "step": 17438 }, { "epoch": 43.5975, "grad_norm": 9.0625, "learning_rate": 1.2531574532129929e-05, "loss": 1.2189, "step": 17439 }, { "epoch": 43.6, "grad_norm": 8.875, "learning_rate": 1.2530791210545163e-05, "loss": 1.3106, "step": 17440 }, { "epoch": 43.6025, "grad_norm": 7.875, "learning_rate": 1.2530007872368634e-05, "loss": 1.29, "step": 17441 }, { "epoch": 43.605, "grad_norm": 9.25, "learning_rate": 1.2529224517605474e-05, "loss": 1.2688, "step": 17442 }, { "epoch": 43.6075, "grad_norm": 6.375, "learning_rate": 1.2528441146260822e-05, "loss": 1.4809, "step": 17443 }, { "epoch": 43.61, "grad_norm": 8.125, "learning_rate": 1.2527657758339807e-05, "loss": 1.4732, "step": 17444 }, { "epoch": 43.6125, "grad_norm": 5.8125, "learning_rate": 1.2526874353847569e-05, "loss": 1.3335, "step": 17445 }, { "epoch": 43.615, "grad_norm": 10.25, "learning_rate": 1.2526090932789246e-05, "loss": 1.391, "step": 17446 }, { "epoch": 43.6175, "grad_norm": 7.28125, "learning_rate": 1.2525307495169971e-05, "loss": 1.2766, "step": 17447 }, { "epoch": 43.62, "grad_norm": 9.8125, "learning_rate": 1.2524524040994881e-05, "loss": 1.35, "step": 17448 }, { "epoch": 43.6225, "grad_norm": 13.75, "learning_rate": 1.2523740570269112e-05, "loss": 1.1852, "step": 17449 }, { "epoch": 43.625, "grad_norm": 6.21875, "learning_rate": 1.25229570829978e-05, "loss": 1.2931, "step": 17450 }, { "epoch": 43.6275, "grad_norm": 7.34375, "learning_rate": 1.2522173579186086e-05, "loss": 1.2113, "step": 17451 }, { "epoch": 43.63, "grad_norm": 7.21875, "learning_rate": 1.2521390058839097e-05, "loss": 1.1736, "step": 17452 }, { "epoch": 43.6325, "grad_norm": 9.375, "learning_rate": 1.2520606521961982e-05, "loss": 1.4412, "step": 17453 }, { "epoch": 43.635, "grad_norm": 8.4375, "learning_rate": 1.251982296855987e-05, "loss": 1.3507, "step": 17454 }, { "epoch": 43.6375, "grad_norm": 11.5625, "learning_rate": 1.2519039398637896e-05, "loss": 1.2954, "step": 17455 }, { "epoch": 43.64, "grad_norm": 5.28125, "learning_rate": 1.2518255812201203e-05, "loss": 1.4115, "step": 17456 }, { "epoch": 43.6425, "grad_norm": 8.625, "learning_rate": 1.2517472209254924e-05, "loss": 1.2491, "step": 17457 }, { "epoch": 43.645, "grad_norm": 9.625, "learning_rate": 1.2516688589804202e-05, "loss": 1.4027, "step": 17458 }, { "epoch": 43.6475, "grad_norm": 7.4375, "learning_rate": 1.2515904953854167e-05, "loss": 1.3264, "step": 17459 }, { "epoch": 43.65, "grad_norm": 6.40625, "learning_rate": 1.251512130140996e-05, "loss": 1.3744, "step": 17460 }, { "epoch": 43.6525, "grad_norm": 7.96875, "learning_rate": 1.2514337632476718e-05, "loss": 1.2641, "step": 17461 }, { "epoch": 43.655, "grad_norm": 10.3125, "learning_rate": 1.2513553947059581e-05, "loss": 1.4845, "step": 17462 }, { "epoch": 43.6575, "grad_norm": 7.3125, "learning_rate": 1.2512770245163684e-05, "loss": 1.4317, "step": 17463 }, { "epoch": 43.66, "grad_norm": 8.3125, "learning_rate": 1.2511986526794166e-05, "loss": 1.3516, "step": 17464 }, { "epoch": 43.6625, "grad_norm": 8.25, "learning_rate": 1.2511202791956163e-05, "loss": 1.0651, "step": 17465 }, { "epoch": 43.665, "grad_norm": 8.125, "learning_rate": 1.2510419040654818e-05, "loss": 1.3094, "step": 17466 }, { "epoch": 43.6675, "grad_norm": 8.375, "learning_rate": 1.2509635272895264e-05, "loss": 1.3544, "step": 17467 }, { "epoch": 43.67, "grad_norm": 8.6875, "learning_rate": 1.2508851488682642e-05, "loss": 1.3971, "step": 17468 }, { "epoch": 43.6725, "grad_norm": 4.59375, "learning_rate": 1.2508067688022094e-05, "loss": 1.4413, "step": 17469 }, { "epoch": 43.675, "grad_norm": 8.75, "learning_rate": 1.2507283870918748e-05, "loss": 1.2733, "step": 17470 }, { "epoch": 43.6775, "grad_norm": 6.5, "learning_rate": 1.2506500037377753e-05, "loss": 1.3704, "step": 17471 }, { "epoch": 43.68, "grad_norm": 10.875, "learning_rate": 1.2505716187404242e-05, "loss": 1.4142, "step": 17472 }, { "epoch": 43.6825, "grad_norm": 9.1875, "learning_rate": 1.250493232100336e-05, "loss": 1.4096, "step": 17473 }, { "epoch": 43.685, "grad_norm": 7.03125, "learning_rate": 1.250414843818024e-05, "loss": 1.3902, "step": 17474 }, { "epoch": 43.6875, "grad_norm": 13.1875, "learning_rate": 1.250336453894002e-05, "loss": 1.4429, "step": 17475 }, { "epoch": 43.69, "grad_norm": 7.0625, "learning_rate": 1.2502580623287846e-05, "loss": 1.4315, "step": 17476 }, { "epoch": 43.6925, "grad_norm": 9.3125, "learning_rate": 1.2501796691228852e-05, "loss": 1.2504, "step": 17477 }, { "epoch": 43.695, "grad_norm": 7.125, "learning_rate": 1.2501012742768178e-05, "loss": 1.3771, "step": 17478 }, { "epoch": 43.6975, "grad_norm": 8.0, "learning_rate": 1.2500228777910967e-05, "loss": 1.2792, "step": 17479 }, { "epoch": 43.7, "grad_norm": 7.03125, "learning_rate": 1.2499444796662354e-05, "loss": 1.3704, "step": 17480 }, { "epoch": 43.7025, "grad_norm": 7.25, "learning_rate": 1.2498660799027482e-05, "loss": 1.4206, "step": 17481 }, { "epoch": 43.705, "grad_norm": 6.9375, "learning_rate": 1.249787678501149e-05, "loss": 1.2381, "step": 17482 }, { "epoch": 43.7075, "grad_norm": 6.65625, "learning_rate": 1.2497092754619517e-05, "loss": 1.28, "step": 17483 }, { "epoch": 43.71, "grad_norm": 6.4375, "learning_rate": 1.2496308707856706e-05, "loss": 1.2411, "step": 17484 }, { "epoch": 43.7125, "grad_norm": 7.875, "learning_rate": 1.2495524644728193e-05, "loss": 1.4694, "step": 17485 }, { "epoch": 43.715, "grad_norm": 10.625, "learning_rate": 1.2494740565239118e-05, "loss": 1.4471, "step": 17486 }, { "epoch": 43.7175, "grad_norm": 9.1875, "learning_rate": 1.2493956469394628e-05, "loss": 1.4136, "step": 17487 }, { "epoch": 43.72, "grad_norm": 8.5, "learning_rate": 1.2493172357199856e-05, "loss": 1.3251, "step": 17488 }, { "epoch": 43.7225, "grad_norm": 5.03125, "learning_rate": 1.2492388228659945e-05, "loss": 1.2557, "step": 17489 }, { "epoch": 43.725, "grad_norm": 6.875, "learning_rate": 1.249160408378004e-05, "loss": 1.4261, "step": 17490 }, { "epoch": 43.7275, "grad_norm": 7.53125, "learning_rate": 1.2490819922565278e-05, "loss": 1.3764, "step": 17491 }, { "epoch": 43.73, "grad_norm": 8.25, "learning_rate": 1.2490035745020798e-05, "loss": 1.2782, "step": 17492 }, { "epoch": 43.7325, "grad_norm": 6.0625, "learning_rate": 1.2489251551151744e-05, "loss": 1.3243, "step": 17493 }, { "epoch": 43.735, "grad_norm": 5.84375, "learning_rate": 1.2488467340963259e-05, "loss": 1.169, "step": 17494 }, { "epoch": 43.7375, "grad_norm": 6.21875, "learning_rate": 1.2487683114460478e-05, "loss": 1.2899, "step": 17495 }, { "epoch": 43.74, "grad_norm": 7.84375, "learning_rate": 1.2486898871648552e-05, "loss": 1.2896, "step": 17496 }, { "epoch": 43.7425, "grad_norm": 6.5, "learning_rate": 1.248611461253261e-05, "loss": 1.3759, "step": 17497 }, { "epoch": 43.745, "grad_norm": 11.0, "learning_rate": 1.2485330337117803e-05, "loss": 1.3501, "step": 17498 }, { "epoch": 43.7475, "grad_norm": 7.625, "learning_rate": 1.248454604540927e-05, "loss": 1.2899, "step": 17499 }, { "epoch": 43.75, "grad_norm": 6.625, "learning_rate": 1.248376173741215e-05, "loss": 1.3945, "step": 17500 }, { "epoch": 43.7525, "grad_norm": 8.625, "learning_rate": 1.248297741313159e-05, "loss": 1.4783, "step": 17501 }, { "epoch": 43.755, "grad_norm": 8.25, "learning_rate": 1.2482193072572727e-05, "loss": 1.3442, "step": 17502 }, { "epoch": 43.7575, "grad_norm": 7.8125, "learning_rate": 1.2481408715740708e-05, "loss": 1.368, "step": 17503 }, { "epoch": 43.76, "grad_norm": 8.3125, "learning_rate": 1.2480624342640673e-05, "loss": 1.4436, "step": 17504 }, { "epoch": 43.7625, "grad_norm": 4.96875, "learning_rate": 1.247983995327776e-05, "loss": 1.4961, "step": 17505 }, { "epoch": 43.765, "grad_norm": 9.1875, "learning_rate": 1.247905554765712e-05, "loss": 1.3775, "step": 17506 }, { "epoch": 43.7675, "grad_norm": 8.3125, "learning_rate": 1.2478271125783889e-05, "loss": 1.3139, "step": 17507 }, { "epoch": 43.77, "grad_norm": 8.75, "learning_rate": 1.2477486687663212e-05, "loss": 1.4933, "step": 17508 }, { "epoch": 43.7725, "grad_norm": 10.125, "learning_rate": 1.2476702233300234e-05, "loss": 1.2747, "step": 17509 }, { "epoch": 43.775, "grad_norm": 7.59375, "learning_rate": 1.2475917762700092e-05, "loss": 1.3764, "step": 17510 }, { "epoch": 43.7775, "grad_norm": 6.53125, "learning_rate": 1.2475133275867935e-05, "loss": 1.325, "step": 17511 }, { "epoch": 43.78, "grad_norm": 9.3125, "learning_rate": 1.2474348772808897e-05, "loss": 1.262, "step": 17512 }, { "epoch": 43.7825, "grad_norm": 6.46875, "learning_rate": 1.2473564253528134e-05, "loss": 1.2139, "step": 17513 }, { "epoch": 43.785, "grad_norm": 7.78125, "learning_rate": 1.247277971803078e-05, "loss": 1.3346, "step": 17514 }, { "epoch": 43.7875, "grad_norm": 5.875, "learning_rate": 1.2471995166321982e-05, "loss": 1.372, "step": 17515 }, { "epoch": 43.79, "grad_norm": 4.53125, "learning_rate": 1.2471210598406882e-05, "loss": 1.3342, "step": 17516 }, { "epoch": 43.7925, "grad_norm": 5.4375, "learning_rate": 1.2470426014290626e-05, "loss": 1.3496, "step": 17517 }, { "epoch": 43.795, "grad_norm": 5.65625, "learning_rate": 1.2469641413978353e-05, "loss": 1.2715, "step": 17518 }, { "epoch": 43.7975, "grad_norm": 5.78125, "learning_rate": 1.2468856797475213e-05, "loss": 1.5512, "step": 17519 }, { "epoch": 43.8, "grad_norm": 7.375, "learning_rate": 1.2468072164786342e-05, "loss": 1.3324, "step": 17520 }, { "epoch": 43.8025, "grad_norm": 7.96875, "learning_rate": 1.2467287515916893e-05, "loss": 1.2672, "step": 17521 }, { "epoch": 43.805, "grad_norm": 5.3125, "learning_rate": 1.2466502850872004e-05, "loss": 1.1821, "step": 17522 }, { "epoch": 43.8075, "grad_norm": 9.0, "learning_rate": 1.2465718169656822e-05, "loss": 1.4299, "step": 17523 }, { "epoch": 43.81, "grad_norm": 9.1875, "learning_rate": 1.2464933472276488e-05, "loss": 1.3297, "step": 17524 }, { "epoch": 43.8125, "grad_norm": 6.125, "learning_rate": 1.246414875873615e-05, "loss": 1.2591, "step": 17525 }, { "epoch": 43.815, "grad_norm": 9.0625, "learning_rate": 1.2463364029040955e-05, "loss": 1.2646, "step": 17526 }, { "epoch": 43.8175, "grad_norm": 9.1875, "learning_rate": 1.246257928319604e-05, "loss": 1.3684, "step": 17527 }, { "epoch": 43.82, "grad_norm": 7.875, "learning_rate": 1.2461794521206554e-05, "loss": 1.3893, "step": 17528 }, { "epoch": 43.8225, "grad_norm": 11.4375, "learning_rate": 1.2461009743077642e-05, "loss": 1.2851, "step": 17529 }, { "epoch": 43.825, "grad_norm": 8.125, "learning_rate": 1.246022494881445e-05, "loss": 1.3985, "step": 17530 }, { "epoch": 43.8275, "grad_norm": 10.8125, "learning_rate": 1.2459440138422118e-05, "loss": 1.3742, "step": 17531 }, { "epoch": 43.83, "grad_norm": 6.0625, "learning_rate": 1.2458655311905798e-05, "loss": 1.2419, "step": 17532 }, { "epoch": 43.8325, "grad_norm": 7.875, "learning_rate": 1.2457870469270631e-05, "loss": 1.1985, "step": 17533 }, { "epoch": 43.835, "grad_norm": 6.375, "learning_rate": 1.2457085610521764e-05, "loss": 1.2731, "step": 17534 }, { "epoch": 43.8375, "grad_norm": 7.5625, "learning_rate": 1.2456300735664342e-05, "loss": 1.3764, "step": 17535 }, { "epoch": 43.84, "grad_norm": 7.21875, "learning_rate": 1.2455515844703512e-05, "loss": 1.4973, "step": 17536 }, { "epoch": 43.8425, "grad_norm": 7.71875, "learning_rate": 1.2454730937644417e-05, "loss": 1.3229, "step": 17537 }, { "epoch": 43.845, "grad_norm": 5.1875, "learning_rate": 1.2453946014492204e-05, "loss": 1.3573, "step": 17538 }, { "epoch": 43.8475, "grad_norm": 7.21875, "learning_rate": 1.2453161075252018e-05, "loss": 1.4272, "step": 17539 }, { "epoch": 43.85, "grad_norm": 4.96875, "learning_rate": 1.2452376119929009e-05, "loss": 1.4859, "step": 17540 }, { "epoch": 43.8525, "grad_norm": 8.0, "learning_rate": 1.2451591148528322e-05, "loss": 1.3723, "step": 17541 }, { "epoch": 43.855, "grad_norm": 5.8125, "learning_rate": 1.2450806161055096e-05, "loss": 1.291, "step": 17542 }, { "epoch": 43.8575, "grad_norm": 6.28125, "learning_rate": 1.2450021157514488e-05, "loss": 1.3159, "step": 17543 }, { "epoch": 43.86, "grad_norm": 6.84375, "learning_rate": 1.2449236137911634e-05, "loss": 1.3334, "step": 17544 }, { "epoch": 43.8625, "grad_norm": 5.3125, "learning_rate": 1.2448451102251688e-05, "loss": 1.2134, "step": 17545 }, { "epoch": 43.865, "grad_norm": 7.3125, "learning_rate": 1.2447666050539798e-05, "loss": 1.3786, "step": 17546 }, { "epoch": 43.8675, "grad_norm": 9.5625, "learning_rate": 1.2446880982781102e-05, "loss": 1.3154, "step": 17547 }, { "epoch": 43.87, "grad_norm": 7.375, "learning_rate": 1.2446095898980755e-05, "loss": 1.5333, "step": 17548 }, { "epoch": 43.8725, "grad_norm": 6.65625, "learning_rate": 1.2445310799143902e-05, "loss": 1.3687, "step": 17549 }, { "epoch": 43.875, "grad_norm": 11.0625, "learning_rate": 1.2444525683275687e-05, "loss": 1.3871, "step": 17550 }, { "epoch": 43.8775, "grad_norm": 5.84375, "learning_rate": 1.2443740551381262e-05, "loss": 1.2205, "step": 17551 }, { "epoch": 43.88, "grad_norm": 6.625, "learning_rate": 1.2442955403465768e-05, "loss": 1.2221, "step": 17552 }, { "epoch": 43.8825, "grad_norm": 7.59375, "learning_rate": 1.2442170239534362e-05, "loss": 1.3548, "step": 17553 }, { "epoch": 43.885, "grad_norm": 6.96875, "learning_rate": 1.244138505959218e-05, "loss": 1.3096, "step": 17554 }, { "epoch": 43.8875, "grad_norm": 9.1875, "learning_rate": 1.2440599863644378e-05, "loss": 1.249, "step": 17555 }, { "epoch": 43.89, "grad_norm": 6.875, "learning_rate": 1.24398146516961e-05, "loss": 1.3467, "step": 17556 }, { "epoch": 43.8925, "grad_norm": 11.0625, "learning_rate": 1.2439029423752496e-05, "loss": 1.3567, "step": 17557 }, { "epoch": 43.895, "grad_norm": 8.875, "learning_rate": 1.243824417981871e-05, "loss": 1.2148, "step": 17558 }, { "epoch": 43.8975, "grad_norm": 8.0, "learning_rate": 1.2437458919899894e-05, "loss": 1.3941, "step": 17559 }, { "epoch": 43.9, "grad_norm": 7.71875, "learning_rate": 1.2436673644001196e-05, "loss": 1.2998, "step": 17560 }, { "epoch": 43.9025, "grad_norm": 10.125, "learning_rate": 1.2435888352127762e-05, "loss": 1.283, "step": 17561 }, { "epoch": 43.905, "grad_norm": 6.4375, "learning_rate": 1.2435103044284741e-05, "loss": 1.36, "step": 17562 }, { "epoch": 43.9075, "grad_norm": 10.25, "learning_rate": 1.2434317720477284e-05, "loss": 1.4459, "step": 17563 }, { "epoch": 43.91, "grad_norm": 5.0, "learning_rate": 1.2433532380710535e-05, "loss": 1.3221, "step": 17564 }, { "epoch": 43.9125, "grad_norm": 6.25, "learning_rate": 1.243274702498965e-05, "loss": 1.3312, "step": 17565 }, { "epoch": 43.915, "grad_norm": 7.34375, "learning_rate": 1.2431961653319769e-05, "loss": 1.4391, "step": 17566 }, { "epoch": 43.9175, "grad_norm": 6.5625, "learning_rate": 1.2431176265706041e-05, "loss": 1.1836, "step": 17567 }, { "epoch": 43.92, "grad_norm": 6.65625, "learning_rate": 1.2430390862153625e-05, "loss": 1.403, "step": 17568 }, { "epoch": 43.9225, "grad_norm": 9.0, "learning_rate": 1.242960544266766e-05, "loss": 1.2243, "step": 17569 }, { "epoch": 43.925, "grad_norm": 6.59375, "learning_rate": 1.2428820007253302e-05, "loss": 1.3894, "step": 17570 }, { "epoch": 43.9275, "grad_norm": 7.78125, "learning_rate": 1.2428034555915695e-05, "loss": 1.2375, "step": 17571 }, { "epoch": 43.93, "grad_norm": 9.5, "learning_rate": 1.2427249088659992e-05, "loss": 1.3611, "step": 17572 }, { "epoch": 43.9325, "grad_norm": 8.625, "learning_rate": 1.2426463605491339e-05, "loss": 1.4257, "step": 17573 }, { "epoch": 43.935, "grad_norm": 8.375, "learning_rate": 1.2425678106414889e-05, "loss": 1.3675, "step": 17574 }, { "epoch": 43.9375, "grad_norm": 9.3125, "learning_rate": 1.242489259143579e-05, "loss": 1.3397, "step": 17575 }, { "epoch": 43.94, "grad_norm": 9.3125, "learning_rate": 1.2424107060559194e-05, "loss": 1.3824, "step": 17576 }, { "epoch": 43.9425, "grad_norm": 5.5625, "learning_rate": 1.2423321513790248e-05, "loss": 1.1662, "step": 17577 }, { "epoch": 43.945, "grad_norm": 7.5625, "learning_rate": 1.2422535951134102e-05, "loss": 1.2044, "step": 17578 }, { "epoch": 43.9475, "grad_norm": 10.9375, "learning_rate": 1.2421750372595907e-05, "loss": 1.4295, "step": 17579 }, { "epoch": 43.95, "grad_norm": 7.5625, "learning_rate": 1.2420964778180815e-05, "loss": 1.241, "step": 17580 }, { "epoch": 43.9525, "grad_norm": 6.40625, "learning_rate": 1.2420179167893976e-05, "loss": 1.1643, "step": 17581 }, { "epoch": 43.955, "grad_norm": 8.125, "learning_rate": 1.2419393541740536e-05, "loss": 1.3862, "step": 17582 }, { "epoch": 43.9575, "grad_norm": 7.28125, "learning_rate": 1.2418607899725651e-05, "loss": 1.2806, "step": 17583 }, { "epoch": 43.96, "grad_norm": 4.53125, "learning_rate": 1.2417822241854466e-05, "loss": 1.299, "step": 17584 }, { "epoch": 43.9625, "grad_norm": 6.0, "learning_rate": 1.2417036568132142e-05, "loss": 1.3004, "step": 17585 }, { "epoch": 43.965, "grad_norm": 9.625, "learning_rate": 1.2416250878563818e-05, "loss": 1.4231, "step": 17586 }, { "epoch": 43.9675, "grad_norm": 6.65625, "learning_rate": 1.241546517315465e-05, "loss": 1.2574, "step": 17587 }, { "epoch": 43.97, "grad_norm": 7.65625, "learning_rate": 1.241467945190979e-05, "loss": 1.3004, "step": 17588 }, { "epoch": 43.9725, "grad_norm": 8.375, "learning_rate": 1.2413893714834388e-05, "loss": 1.2936, "step": 17589 }, { "epoch": 43.975, "grad_norm": 7.8125, "learning_rate": 1.2413107961933594e-05, "loss": 1.4302, "step": 17590 }, { "epoch": 43.9775, "grad_norm": 7.34375, "learning_rate": 1.2412322193212562e-05, "loss": 1.4436, "step": 17591 }, { "epoch": 43.98, "grad_norm": 6.40625, "learning_rate": 1.2411536408676443e-05, "loss": 1.3026, "step": 17592 }, { "epoch": 43.9825, "grad_norm": 9.1875, "learning_rate": 1.2410750608330389e-05, "loss": 1.36, "step": 17593 }, { "epoch": 43.985, "grad_norm": 7.46875, "learning_rate": 1.2409964792179547e-05, "loss": 1.3253, "step": 17594 }, { "epoch": 43.9875, "grad_norm": 10.625, "learning_rate": 1.2409178960229073e-05, "loss": 1.2544, "step": 17595 }, { "epoch": 43.99, "grad_norm": 12.125, "learning_rate": 1.2408393112484123e-05, "loss": 1.4124, "step": 17596 }, { "epoch": 43.9925, "grad_norm": 7.90625, "learning_rate": 1.2407607248949836e-05, "loss": 1.2609, "step": 17597 }, { "epoch": 43.995, "grad_norm": 8.25, "learning_rate": 1.2406821369631379e-05, "loss": 1.3957, "step": 17598 }, { "epoch": 43.9975, "grad_norm": 7.375, "learning_rate": 1.2406035474533893e-05, "loss": 1.4453, "step": 17599 }, { "epoch": 44.0, "grad_norm": 8.5, "learning_rate": 1.2405249563662539e-05, "loss": 1.3083, "step": 17600 }, { "epoch": 44.0025, "grad_norm": 7.15625, "learning_rate": 1.2404463637022461e-05, "loss": 1.4017, "step": 17601 }, { "epoch": 44.005, "grad_norm": 8.25, "learning_rate": 1.2403677694618818e-05, "loss": 1.2365, "step": 17602 }, { "epoch": 44.0075, "grad_norm": 6.28125, "learning_rate": 1.240289173645676e-05, "loss": 1.2766, "step": 17603 }, { "epoch": 44.01, "grad_norm": 9.25, "learning_rate": 1.240210576254144e-05, "loss": 1.2194, "step": 17604 }, { "epoch": 44.0125, "grad_norm": 8.375, "learning_rate": 1.240131977287801e-05, "loss": 1.1556, "step": 17605 }, { "epoch": 44.015, "grad_norm": 7.96875, "learning_rate": 1.2400533767471623e-05, "loss": 1.4389, "step": 17606 }, { "epoch": 44.0175, "grad_norm": 9.9375, "learning_rate": 1.2399747746327433e-05, "loss": 1.3991, "step": 17607 }, { "epoch": 44.02, "grad_norm": 8.75, "learning_rate": 1.2398961709450593e-05, "loss": 1.2286, "step": 17608 }, { "epoch": 44.0225, "grad_norm": 8.875, "learning_rate": 1.2398175656846257e-05, "loss": 1.2953, "step": 17609 }, { "epoch": 44.025, "grad_norm": 8.0625, "learning_rate": 1.2397389588519575e-05, "loss": 1.0898, "step": 17610 }, { "epoch": 44.0275, "grad_norm": 9.0625, "learning_rate": 1.2396603504475706e-05, "loss": 1.2213, "step": 17611 }, { "epoch": 44.03, "grad_norm": 7.25, "learning_rate": 1.2395817404719797e-05, "loss": 1.2816, "step": 17612 }, { "epoch": 44.0325, "grad_norm": 6.28125, "learning_rate": 1.2395031289257009e-05, "loss": 1.2729, "step": 17613 }, { "epoch": 44.035, "grad_norm": 16.75, "learning_rate": 1.2394245158092488e-05, "loss": 1.3474, "step": 17614 }, { "epoch": 44.0375, "grad_norm": 6.6875, "learning_rate": 1.2393459011231395e-05, "loss": 1.3493, "step": 17615 }, { "epoch": 44.04, "grad_norm": 6.4375, "learning_rate": 1.2392672848678877e-05, "loss": 1.2961, "step": 17616 }, { "epoch": 44.0425, "grad_norm": 9.875, "learning_rate": 1.2391886670440094e-05, "loss": 1.3778, "step": 17617 }, { "epoch": 44.045, "grad_norm": 8.9375, "learning_rate": 1.2391100476520196e-05, "loss": 1.2912, "step": 17618 }, { "epoch": 44.0475, "grad_norm": 7.0, "learning_rate": 1.2390314266924341e-05, "loss": 1.1803, "step": 17619 }, { "epoch": 44.05, "grad_norm": 7.125, "learning_rate": 1.2389528041657679e-05, "loss": 1.2416, "step": 17620 }, { "epoch": 44.0525, "grad_norm": 6.59375, "learning_rate": 1.238874180072537e-05, "loss": 1.2889, "step": 17621 }, { "epoch": 44.055, "grad_norm": 6.9375, "learning_rate": 1.2387955544132562e-05, "loss": 1.2822, "step": 17622 }, { "epoch": 44.0575, "grad_norm": 8.25, "learning_rate": 1.2387169271884416e-05, "loss": 1.2996, "step": 17623 }, { "epoch": 44.06, "grad_norm": 10.875, "learning_rate": 1.2386382983986079e-05, "loss": 1.4232, "step": 17624 }, { "epoch": 44.0625, "grad_norm": 6.125, "learning_rate": 1.2385596680442715e-05, "loss": 1.3466, "step": 17625 }, { "epoch": 44.065, "grad_norm": 7.25, "learning_rate": 1.2384810361259473e-05, "loss": 1.2697, "step": 17626 }, { "epoch": 44.0675, "grad_norm": 6.03125, "learning_rate": 1.238402402644151e-05, "loss": 1.3396, "step": 17627 }, { "epoch": 44.07, "grad_norm": 6.75, "learning_rate": 1.2383237675993983e-05, "loss": 1.4159, "step": 17628 }, { "epoch": 44.0725, "grad_norm": 8.1875, "learning_rate": 1.2382451309922038e-05, "loss": 1.3519, "step": 17629 }, { "epoch": 44.075, "grad_norm": 8.6875, "learning_rate": 1.2381664928230846e-05, "loss": 1.3377, "step": 17630 }, { "epoch": 44.0775, "grad_norm": 6.28125, "learning_rate": 1.2380878530925549e-05, "loss": 1.3025, "step": 17631 }, { "epoch": 44.08, "grad_norm": 7.28125, "learning_rate": 1.238009211801131e-05, "loss": 1.5212, "step": 17632 }, { "epoch": 44.0825, "grad_norm": 5.75, "learning_rate": 1.2379305689493281e-05, "loss": 1.3391, "step": 17633 }, { "epoch": 44.085, "grad_norm": 4.6875, "learning_rate": 1.2378519245376621e-05, "loss": 1.4026, "step": 17634 }, { "epoch": 44.0875, "grad_norm": 7.65625, "learning_rate": 1.2377732785666484e-05, "loss": 1.267, "step": 17635 }, { "epoch": 44.09, "grad_norm": 5.65625, "learning_rate": 1.2376946310368023e-05, "loss": 1.341, "step": 17636 }, { "epoch": 44.0925, "grad_norm": 5.375, "learning_rate": 1.23761598194864e-05, "loss": 1.2372, "step": 17637 }, { "epoch": 44.095, "grad_norm": 6.65625, "learning_rate": 1.2375373313026768e-05, "loss": 1.2872, "step": 17638 }, { "epoch": 44.0975, "grad_norm": 6.53125, "learning_rate": 1.2374586790994281e-05, "loss": 1.4083, "step": 17639 }, { "epoch": 44.1, "grad_norm": 7.21875, "learning_rate": 1.23738002533941e-05, "loss": 1.1589, "step": 17640 }, { "epoch": 44.1025, "grad_norm": 11.3125, "learning_rate": 1.2373013700231377e-05, "loss": 1.3624, "step": 17641 }, { "epoch": 44.105, "grad_norm": 6.625, "learning_rate": 1.2372227131511276e-05, "loss": 1.4613, "step": 17642 }, { "epoch": 44.1075, "grad_norm": 8.4375, "learning_rate": 1.2371440547238947e-05, "loss": 1.1938, "step": 17643 }, { "epoch": 44.11, "grad_norm": 11.3125, "learning_rate": 1.2370653947419548e-05, "loss": 1.3425, "step": 17644 }, { "epoch": 44.1125, "grad_norm": 6.28125, "learning_rate": 1.2369867332058236e-05, "loss": 1.3141, "step": 17645 }, { "epoch": 44.115, "grad_norm": 6.28125, "learning_rate": 1.236908070116017e-05, "loss": 1.2711, "step": 17646 }, { "epoch": 44.1175, "grad_norm": 6.0, "learning_rate": 1.2368294054730505e-05, "loss": 1.4407, "step": 17647 }, { "epoch": 44.12, "grad_norm": 6.96875, "learning_rate": 1.2367507392774398e-05, "loss": 1.3375, "step": 17648 }, { "epoch": 44.1225, "grad_norm": 5.96875, "learning_rate": 1.2366720715297009e-05, "loss": 1.1966, "step": 17649 }, { "epoch": 44.125, "grad_norm": 5.375, "learning_rate": 1.2365934022303491e-05, "loss": 1.1772, "step": 17650 }, { "epoch": 44.1275, "grad_norm": 13.8125, "learning_rate": 1.2365147313799007e-05, "loss": 1.3396, "step": 17651 }, { "epoch": 44.13, "grad_norm": 7.6875, "learning_rate": 1.2364360589788711e-05, "loss": 1.2609, "step": 17652 }, { "epoch": 44.1325, "grad_norm": 5.21875, "learning_rate": 1.2363573850277764e-05, "loss": 1.4651, "step": 17653 }, { "epoch": 44.135, "grad_norm": 8.625, "learning_rate": 1.2362787095271316e-05, "loss": 1.3323, "step": 17654 }, { "epoch": 44.1375, "grad_norm": 9.0, "learning_rate": 1.2362000324774534e-05, "loss": 1.4777, "step": 17655 }, { "epoch": 44.14, "grad_norm": 6.4375, "learning_rate": 1.236121353879257e-05, "loss": 1.4009, "step": 17656 }, { "epoch": 44.1425, "grad_norm": 9.625, "learning_rate": 1.2360426737330589e-05, "loss": 1.3994, "step": 17657 }, { "epoch": 44.145, "grad_norm": 7.125, "learning_rate": 1.2359639920393742e-05, "loss": 1.3972, "step": 17658 }, { "epoch": 44.1475, "grad_norm": 6.4375, "learning_rate": 1.235885308798719e-05, "loss": 1.3889, "step": 17659 }, { "epoch": 44.15, "grad_norm": 5.625, "learning_rate": 1.2358066240116092e-05, "loss": 1.418, "step": 17660 }, { "epoch": 44.1525, "grad_norm": 8.4375, "learning_rate": 1.2357279376785605e-05, "loss": 1.541, "step": 17661 }, { "epoch": 44.155, "grad_norm": 11.9375, "learning_rate": 1.235649249800089e-05, "loss": 1.3175, "step": 17662 }, { "epoch": 44.1575, "grad_norm": 6.34375, "learning_rate": 1.2355705603767104e-05, "loss": 1.273, "step": 17663 }, { "epoch": 44.16, "grad_norm": 5.78125, "learning_rate": 1.2354918694089406e-05, "loss": 1.2307, "step": 17664 }, { "epoch": 44.1625, "grad_norm": 14.1875, "learning_rate": 1.2354131768972955e-05, "loss": 1.4113, "step": 17665 }, { "epoch": 44.165, "grad_norm": 6.09375, "learning_rate": 1.2353344828422911e-05, "loss": 1.3326, "step": 17666 }, { "epoch": 44.1675, "grad_norm": 6.75, "learning_rate": 1.2352557872444432e-05, "loss": 1.421, "step": 17667 }, { "epoch": 44.17, "grad_norm": 7.90625, "learning_rate": 1.235177090104268e-05, "loss": 1.3162, "step": 17668 }, { "epoch": 44.1725, "grad_norm": 4.4375, "learning_rate": 1.235098391422281e-05, "loss": 1.21, "step": 17669 }, { "epoch": 44.175, "grad_norm": 6.6875, "learning_rate": 1.2350196911989984e-05, "loss": 1.326, "step": 17670 }, { "epoch": 44.1775, "grad_norm": 8.875, "learning_rate": 1.2349409894349358e-05, "loss": 1.4652, "step": 17671 }, { "epoch": 44.18, "grad_norm": 10.0, "learning_rate": 1.23486228613061e-05, "loss": 1.5892, "step": 17672 }, { "epoch": 44.1825, "grad_norm": 6.59375, "learning_rate": 1.2347835812865362e-05, "loss": 1.3864, "step": 17673 }, { "epoch": 44.185, "grad_norm": 9.6875, "learning_rate": 1.2347048749032306e-05, "loss": 1.2771, "step": 17674 }, { "epoch": 44.1875, "grad_norm": 8.5625, "learning_rate": 1.2346261669812093e-05, "loss": 1.4217, "step": 17675 }, { "epoch": 44.19, "grad_norm": 5.09375, "learning_rate": 1.2345474575209884e-05, "loss": 1.3513, "step": 17676 }, { "epoch": 44.1925, "grad_norm": 7.84375, "learning_rate": 1.2344687465230834e-05, "loss": 1.3875, "step": 17677 }, { "epoch": 44.195, "grad_norm": 5.875, "learning_rate": 1.2343900339880109e-05, "loss": 1.4178, "step": 17678 }, { "epoch": 44.1975, "grad_norm": 8.1875, "learning_rate": 1.2343113199162866e-05, "loss": 1.2345, "step": 17679 }, { "epoch": 44.2, "grad_norm": 7.3125, "learning_rate": 1.2342326043084268e-05, "loss": 1.3015, "step": 17680 }, { "epoch": 44.2025, "grad_norm": 8.0, "learning_rate": 1.2341538871649472e-05, "loss": 1.4207, "step": 17681 }, { "epoch": 44.205, "grad_norm": 5.71875, "learning_rate": 1.2340751684863643e-05, "loss": 1.1479, "step": 17682 }, { "epoch": 44.2075, "grad_norm": 5.25, "learning_rate": 1.233996448273194e-05, "loss": 1.3055, "step": 17683 }, { "epoch": 44.21, "grad_norm": 4.84375, "learning_rate": 1.233917726525952e-05, "loss": 1.2468, "step": 17684 }, { "epoch": 44.2125, "grad_norm": 10.8125, "learning_rate": 1.2338390032451553e-05, "loss": 1.2744, "step": 17685 }, { "epoch": 44.215, "grad_norm": 6.53125, "learning_rate": 1.2337602784313189e-05, "loss": 1.3006, "step": 17686 }, { "epoch": 44.2175, "grad_norm": 11.3125, "learning_rate": 1.2336815520849602e-05, "loss": 1.4762, "step": 17687 }, { "epoch": 44.22, "grad_norm": 10.625, "learning_rate": 1.2336028242065938e-05, "loss": 1.5109, "step": 17688 }, { "epoch": 44.2225, "grad_norm": 8.0, "learning_rate": 1.2335240947967371e-05, "loss": 1.3718, "step": 17689 }, { "epoch": 44.225, "grad_norm": 7.75, "learning_rate": 1.2334453638559057e-05, "loss": 1.4406, "step": 17690 }, { "epoch": 44.2275, "grad_norm": 10.125, "learning_rate": 1.2333666313846155e-05, "loss": 1.2829, "step": 17691 }, { "epoch": 44.23, "grad_norm": 10.3125, "learning_rate": 1.2332878973833834e-05, "loss": 1.3792, "step": 17692 }, { "epoch": 44.2325, "grad_norm": 7.96875, "learning_rate": 1.2332091618527252e-05, "loss": 1.3465, "step": 17693 }, { "epoch": 44.235, "grad_norm": 9.4375, "learning_rate": 1.2331304247931568e-05, "loss": 1.4714, "step": 17694 }, { "epoch": 44.2375, "grad_norm": 8.0625, "learning_rate": 1.233051686205195e-05, "loss": 1.3563, "step": 17695 }, { "epoch": 44.24, "grad_norm": 7.3125, "learning_rate": 1.2329729460893552e-05, "loss": 1.2866, "step": 17696 }, { "epoch": 44.2425, "grad_norm": 9.0625, "learning_rate": 1.2328942044461543e-05, "loss": 1.3227, "step": 17697 }, { "epoch": 44.245, "grad_norm": 9.3125, "learning_rate": 1.2328154612761085e-05, "loss": 1.3797, "step": 17698 }, { "epoch": 44.2475, "grad_norm": 7.59375, "learning_rate": 1.2327367165797334e-05, "loss": 1.3043, "step": 17699 }, { "epoch": 44.25, "grad_norm": 6.34375, "learning_rate": 1.2326579703575464e-05, "loss": 1.1653, "step": 17700 }, { "epoch": 44.2525, "grad_norm": 7.125, "learning_rate": 1.2325792226100623e-05, "loss": 1.3177, "step": 17701 }, { "epoch": 44.255, "grad_norm": 10.75, "learning_rate": 1.2325004733377987e-05, "loss": 1.2982, "step": 17702 }, { "epoch": 44.2575, "grad_norm": 7.125, "learning_rate": 1.232421722541271e-05, "loss": 1.2327, "step": 17703 }, { "epoch": 44.26, "grad_norm": 7.0, "learning_rate": 1.232342970220996e-05, "loss": 1.4051, "step": 17704 }, { "epoch": 44.2625, "grad_norm": 6.3125, "learning_rate": 1.2322642163774896e-05, "loss": 1.3076, "step": 17705 }, { "epoch": 44.265, "grad_norm": 6.09375, "learning_rate": 1.2321854610112684e-05, "loss": 1.2883, "step": 17706 }, { "epoch": 44.2675, "grad_norm": 6.4375, "learning_rate": 1.2321067041228484e-05, "loss": 1.1672, "step": 17707 }, { "epoch": 44.27, "grad_norm": 7.75, "learning_rate": 1.2320279457127467e-05, "loss": 1.3453, "step": 17708 }, { "epoch": 44.2725, "grad_norm": 5.09375, "learning_rate": 1.2319491857814786e-05, "loss": 1.3122, "step": 17709 }, { "epoch": 44.275, "grad_norm": 6.90625, "learning_rate": 1.231870424329561e-05, "loss": 1.3182, "step": 17710 }, { "epoch": 44.2775, "grad_norm": 9.0, "learning_rate": 1.2317916613575098e-05, "loss": 1.4459, "step": 17711 }, { "epoch": 44.28, "grad_norm": 8.8125, "learning_rate": 1.2317128968658424e-05, "loss": 1.2405, "step": 17712 }, { "epoch": 44.2825, "grad_norm": 6.53125, "learning_rate": 1.2316341308550743e-05, "loss": 1.3498, "step": 17713 }, { "epoch": 44.285, "grad_norm": 5.5625, "learning_rate": 1.231555363325722e-05, "loss": 1.2076, "step": 17714 }, { "epoch": 44.2875, "grad_norm": 8.125, "learning_rate": 1.231476594278302e-05, "loss": 1.3874, "step": 17715 }, { "epoch": 44.29, "grad_norm": 6.65625, "learning_rate": 1.2313978237133308e-05, "loss": 1.3243, "step": 17716 }, { "epoch": 44.2925, "grad_norm": 8.25, "learning_rate": 1.2313190516313246e-05, "loss": 1.5726, "step": 17717 }, { "epoch": 44.295, "grad_norm": 7.5625, "learning_rate": 1.2312402780328001e-05, "loss": 1.4251, "step": 17718 }, { "epoch": 44.2975, "grad_norm": 11.9375, "learning_rate": 1.2311615029182735e-05, "loss": 1.27, "step": 17719 }, { "epoch": 44.3, "grad_norm": 5.625, "learning_rate": 1.2310827262882614e-05, "loss": 1.125, "step": 17720 }, { "epoch": 44.3025, "grad_norm": 6.84375, "learning_rate": 1.2310039481432801e-05, "loss": 1.3305, "step": 17721 }, { "epoch": 44.305, "grad_norm": 6.78125, "learning_rate": 1.230925168483846e-05, "loss": 1.2567, "step": 17722 }, { "epoch": 44.3075, "grad_norm": 4.96875, "learning_rate": 1.2308463873104763e-05, "loss": 1.3064, "step": 17723 }, { "epoch": 44.31, "grad_norm": 6.90625, "learning_rate": 1.2307676046236866e-05, "loss": 1.4824, "step": 17724 }, { "epoch": 44.3125, "grad_norm": 7.09375, "learning_rate": 1.2306888204239938e-05, "loss": 1.3461, "step": 17725 }, { "epoch": 44.315, "grad_norm": 6.71875, "learning_rate": 1.2306100347119139e-05, "loss": 1.3254, "step": 17726 }, { "epoch": 44.3175, "grad_norm": 5.34375, "learning_rate": 1.2305312474879645e-05, "loss": 1.2977, "step": 17727 }, { "epoch": 44.32, "grad_norm": 5.59375, "learning_rate": 1.2304524587526609e-05, "loss": 1.2752, "step": 17728 }, { "epoch": 44.3225, "grad_norm": 7.40625, "learning_rate": 1.2303736685065207e-05, "loss": 1.4256, "step": 17729 }, { "epoch": 44.325, "grad_norm": 8.375, "learning_rate": 1.2302948767500596e-05, "loss": 1.4108, "step": 17730 }, { "epoch": 44.3275, "grad_norm": 6.375, "learning_rate": 1.2302160834837944e-05, "loss": 1.3533, "step": 17731 }, { "epoch": 44.33, "grad_norm": 6.625, "learning_rate": 1.230137288708242e-05, "loss": 1.3916, "step": 17732 }, { "epoch": 44.3325, "grad_norm": 7.4375, "learning_rate": 1.2300584924239187e-05, "loss": 1.2791, "step": 17733 }, { "epoch": 44.335, "grad_norm": 7.40625, "learning_rate": 1.229979694631341e-05, "loss": 1.1467, "step": 17734 }, { "epoch": 44.3375, "grad_norm": 8.875, "learning_rate": 1.2299008953310258e-05, "loss": 1.3338, "step": 17735 }, { "epoch": 44.34, "grad_norm": 7.125, "learning_rate": 1.2298220945234896e-05, "loss": 1.3526, "step": 17736 }, { "epoch": 44.3425, "grad_norm": 6.21875, "learning_rate": 1.2297432922092487e-05, "loss": 1.3194, "step": 17737 }, { "epoch": 44.345, "grad_norm": 5.46875, "learning_rate": 1.2296644883888205e-05, "loss": 1.279, "step": 17738 }, { "epoch": 44.3475, "grad_norm": 7.6875, "learning_rate": 1.2295856830627205e-05, "loss": 1.2125, "step": 17739 }, { "epoch": 44.35, "grad_norm": 6.1875, "learning_rate": 1.2295068762314661e-05, "loss": 1.2687, "step": 17740 }, { "epoch": 44.3525, "grad_norm": 6.8125, "learning_rate": 1.2294280678955738e-05, "loss": 1.327, "step": 17741 }, { "epoch": 44.355, "grad_norm": 16.375, "learning_rate": 1.2293492580555606e-05, "loss": 1.2568, "step": 17742 }, { "epoch": 44.3575, "grad_norm": 6.65625, "learning_rate": 1.2292704467119425e-05, "loss": 1.1811, "step": 17743 }, { "epoch": 44.36, "grad_norm": 5.90625, "learning_rate": 1.2291916338652365e-05, "loss": 1.2707, "step": 17744 }, { "epoch": 44.3625, "grad_norm": 6.65625, "learning_rate": 1.2291128195159594e-05, "loss": 1.5357, "step": 17745 }, { "epoch": 44.365, "grad_norm": 6.59375, "learning_rate": 1.2290340036646278e-05, "loss": 1.2314, "step": 17746 }, { "epoch": 44.3675, "grad_norm": 5.75, "learning_rate": 1.2289551863117586e-05, "loss": 1.3569, "step": 17747 }, { "epoch": 44.37, "grad_norm": 8.3125, "learning_rate": 1.2288763674578681e-05, "loss": 1.4366, "step": 17748 }, { "epoch": 44.3725, "grad_norm": 8.375, "learning_rate": 1.2287975471034734e-05, "loss": 1.3803, "step": 17749 }, { "epoch": 44.375, "grad_norm": 6.78125, "learning_rate": 1.2287187252490914e-05, "loss": 1.4922, "step": 17750 }, { "epoch": 44.3775, "grad_norm": 8.375, "learning_rate": 1.2286399018952381e-05, "loss": 1.4219, "step": 17751 }, { "epoch": 44.38, "grad_norm": 7.25, "learning_rate": 1.2285610770424311e-05, "loss": 1.3108, "step": 17752 }, { "epoch": 44.3825, "grad_norm": 7.28125, "learning_rate": 1.2284822506911866e-05, "loss": 1.2529, "step": 17753 }, { "epoch": 44.385, "grad_norm": 6.4375, "learning_rate": 1.2284034228420218e-05, "loss": 1.4405, "step": 17754 }, { "epoch": 44.3875, "grad_norm": 9.625, "learning_rate": 1.2283245934954535e-05, "loss": 1.3948, "step": 17755 }, { "epoch": 44.39, "grad_norm": 10.6875, "learning_rate": 1.228245762651998e-05, "loss": 1.2386, "step": 17756 }, { "epoch": 44.3925, "grad_norm": 9.125, "learning_rate": 1.2281669303121726e-05, "loss": 1.4273, "step": 17757 }, { "epoch": 44.395, "grad_norm": 11.0, "learning_rate": 1.2280880964764935e-05, "loss": 1.2437, "step": 17758 }, { "epoch": 44.3975, "grad_norm": 5.875, "learning_rate": 1.2280092611454783e-05, "loss": 1.3142, "step": 17759 }, { "epoch": 44.4, "grad_norm": 7.03125, "learning_rate": 1.2279304243196438e-05, "loss": 1.2833, "step": 17760 }, { "epoch": 44.4025, "grad_norm": 7.0, "learning_rate": 1.2278515859995059e-05, "loss": 1.3114, "step": 17761 }, { "epoch": 44.405, "grad_norm": 10.4375, "learning_rate": 1.2277727461855826e-05, "loss": 1.3832, "step": 17762 }, { "epoch": 44.4075, "grad_norm": 17.5, "learning_rate": 1.22769390487839e-05, "loss": 1.4382, "step": 17763 }, { "epoch": 44.41, "grad_norm": 6.1875, "learning_rate": 1.2276150620784454e-05, "loss": 1.1753, "step": 17764 }, { "epoch": 44.4125, "grad_norm": 6.84375, "learning_rate": 1.227536217786266e-05, "loss": 1.4151, "step": 17765 }, { "epoch": 44.415, "grad_norm": 8.9375, "learning_rate": 1.2274573720023674e-05, "loss": 1.2678, "step": 17766 }, { "epoch": 44.4175, "grad_norm": 5.9375, "learning_rate": 1.2273785247272678e-05, "loss": 1.2648, "step": 17767 }, { "epoch": 44.42, "grad_norm": 5.5625, "learning_rate": 1.2272996759614837e-05, "loss": 1.3887, "step": 17768 }, { "epoch": 44.4225, "grad_norm": 7.96875, "learning_rate": 1.227220825705532e-05, "loss": 1.4572, "step": 17769 }, { "epoch": 44.425, "grad_norm": 6.21875, "learning_rate": 1.22714197395993e-05, "loss": 1.2397, "step": 17770 }, { "epoch": 44.4275, "grad_norm": 7.875, "learning_rate": 1.2270631207251937e-05, "loss": 1.4397, "step": 17771 }, { "epoch": 44.43, "grad_norm": 6.25, "learning_rate": 1.2269842660018412e-05, "loss": 1.4455, "step": 17772 }, { "epoch": 44.4325, "grad_norm": 7.40625, "learning_rate": 1.2269054097903885e-05, "loss": 1.2055, "step": 17773 }, { "epoch": 44.435, "grad_norm": 9.375, "learning_rate": 1.2268265520913535e-05, "loss": 1.4051, "step": 17774 }, { "epoch": 44.4375, "grad_norm": 8.75, "learning_rate": 1.2267476929052521e-05, "loss": 1.3866, "step": 17775 }, { "epoch": 44.44, "grad_norm": 7.59375, "learning_rate": 1.2266688322326024e-05, "loss": 1.4772, "step": 17776 }, { "epoch": 44.4425, "grad_norm": 5.96875, "learning_rate": 1.2265899700739208e-05, "loss": 1.352, "step": 17777 }, { "epoch": 44.445, "grad_norm": 8.1875, "learning_rate": 1.2265111064297242e-05, "loss": 1.2909, "step": 17778 }, { "epoch": 44.4475, "grad_norm": 8.0, "learning_rate": 1.22643224130053e-05, "loss": 1.3002, "step": 17779 }, { "epoch": 44.45, "grad_norm": 12.625, "learning_rate": 1.2263533746868552e-05, "loss": 1.3051, "step": 17780 }, { "epoch": 44.4525, "grad_norm": 10.8125, "learning_rate": 1.2262745065892165e-05, "loss": 1.2884, "step": 17781 }, { "epoch": 44.455, "grad_norm": 7.21875, "learning_rate": 1.2261956370081314e-05, "loss": 1.2784, "step": 17782 }, { "epoch": 44.4575, "grad_norm": 7.8125, "learning_rate": 1.2261167659441164e-05, "loss": 1.4634, "step": 17783 }, { "epoch": 44.46, "grad_norm": 10.6875, "learning_rate": 1.2260378933976895e-05, "loss": 1.3404, "step": 17784 }, { "epoch": 44.4625, "grad_norm": 8.875, "learning_rate": 1.225959019369367e-05, "loss": 1.3914, "step": 17785 }, { "epoch": 44.465, "grad_norm": 5.84375, "learning_rate": 1.225880143859666e-05, "loss": 1.3337, "step": 17786 }, { "epoch": 44.4675, "grad_norm": 8.0625, "learning_rate": 1.2258012668691039e-05, "loss": 1.2333, "step": 17787 }, { "epoch": 44.47, "grad_norm": 9.25, "learning_rate": 1.2257223883981978e-05, "loss": 1.2691, "step": 17788 }, { "epoch": 44.4725, "grad_norm": 9.875, "learning_rate": 1.2256435084474647e-05, "loss": 1.3911, "step": 17789 }, { "epoch": 44.475, "grad_norm": 7.90625, "learning_rate": 1.2255646270174217e-05, "loss": 1.3375, "step": 17790 }, { "epoch": 44.4775, "grad_norm": 8.625, "learning_rate": 1.225485744108586e-05, "loss": 1.2696, "step": 17791 }, { "epoch": 44.48, "grad_norm": 5.5625, "learning_rate": 1.225406859721475e-05, "loss": 1.3959, "step": 17792 }, { "epoch": 44.4825, "grad_norm": 10.75, "learning_rate": 1.2253279738566058e-05, "loss": 1.2206, "step": 17793 }, { "epoch": 44.485, "grad_norm": 9.625, "learning_rate": 1.225249086514495e-05, "loss": 1.446, "step": 17794 }, { "epoch": 44.4875, "grad_norm": 8.0625, "learning_rate": 1.2251701976956606e-05, "loss": 1.3765, "step": 17795 }, { "epoch": 44.49, "grad_norm": 9.375, "learning_rate": 1.225091307400619e-05, "loss": 1.4026, "step": 17796 }, { "epoch": 44.4925, "grad_norm": 6.78125, "learning_rate": 1.2250124156298882e-05, "loss": 1.3725, "step": 17797 }, { "epoch": 44.495, "grad_norm": 8.1875, "learning_rate": 1.2249335223839844e-05, "loss": 1.4241, "step": 17798 }, { "epoch": 44.4975, "grad_norm": 7.90625, "learning_rate": 1.2248546276634261e-05, "loss": 1.4387, "step": 17799 }, { "epoch": 44.5, "grad_norm": 7.46875, "learning_rate": 1.2247757314687296e-05, "loss": 1.2664, "step": 17800 }, { "epoch": 44.5025, "grad_norm": 7.40625, "learning_rate": 1.2246968338004124e-05, "loss": 1.2566, "step": 17801 }, { "epoch": 44.505, "grad_norm": 5.40625, "learning_rate": 1.2246179346589918e-05, "loss": 1.3896, "step": 17802 }, { "epoch": 44.5075, "grad_norm": 6.15625, "learning_rate": 1.2245390340449847e-05, "loss": 1.3851, "step": 17803 }, { "epoch": 44.51, "grad_norm": 8.1875, "learning_rate": 1.224460131958909e-05, "loss": 1.358, "step": 17804 }, { "epoch": 44.5125, "grad_norm": 6.9375, "learning_rate": 1.2243812284012814e-05, "loss": 1.3039, "step": 17805 }, { "epoch": 44.515, "grad_norm": 8.125, "learning_rate": 1.2243023233726196e-05, "loss": 1.2437, "step": 17806 }, { "epoch": 44.5175, "grad_norm": 9.5625, "learning_rate": 1.2242234168734409e-05, "loss": 1.3387, "step": 17807 }, { "epoch": 44.52, "grad_norm": 9.3125, "learning_rate": 1.2241445089042623e-05, "loss": 1.2199, "step": 17808 }, { "epoch": 44.5225, "grad_norm": 9.125, "learning_rate": 1.2240655994656012e-05, "loss": 1.4625, "step": 17809 }, { "epoch": 44.525, "grad_norm": 7.625, "learning_rate": 1.2239866885579755e-05, "loss": 1.3129, "step": 17810 }, { "epoch": 44.5275, "grad_norm": 6.4375, "learning_rate": 1.2239077761819012e-05, "loss": 1.4307, "step": 17811 }, { "epoch": 44.53, "grad_norm": 6.0, "learning_rate": 1.2238288623378971e-05, "loss": 1.3884, "step": 17812 }, { "epoch": 44.5325, "grad_norm": 8.625, "learning_rate": 1.2237499470264796e-05, "loss": 1.2846, "step": 17813 }, { "epoch": 44.535, "grad_norm": 8.25, "learning_rate": 1.2236710302481667e-05, "loss": 1.2989, "step": 17814 }, { "epoch": 44.5375, "grad_norm": 7.4375, "learning_rate": 1.2235921120034753e-05, "loss": 1.362, "step": 17815 }, { "epoch": 44.54, "grad_norm": 6.4375, "learning_rate": 1.223513192292923e-05, "loss": 1.2465, "step": 17816 }, { "epoch": 44.5425, "grad_norm": 5.75, "learning_rate": 1.2234342711170271e-05, "loss": 1.5746, "step": 17817 }, { "epoch": 44.545, "grad_norm": 6.9375, "learning_rate": 1.223355348476305e-05, "loss": 1.3237, "step": 17818 }, { "epoch": 44.5475, "grad_norm": 10.8125, "learning_rate": 1.2232764243712742e-05, "loss": 1.3859, "step": 17819 }, { "epoch": 44.55, "grad_norm": 9.125, "learning_rate": 1.2231974988024522e-05, "loss": 1.2605, "step": 17820 }, { "epoch": 44.5525, "grad_norm": 8.375, "learning_rate": 1.2231185717703562e-05, "loss": 1.2392, "step": 17821 }, { "epoch": 44.555, "grad_norm": 8.8125, "learning_rate": 1.2230396432755038e-05, "loss": 1.353, "step": 17822 }, { "epoch": 44.5575, "grad_norm": 10.9375, "learning_rate": 1.2229607133184125e-05, "loss": 1.3791, "step": 17823 }, { "epoch": 44.56, "grad_norm": 5.90625, "learning_rate": 1.2228817818995998e-05, "loss": 1.3813, "step": 17824 }, { "epoch": 44.5625, "grad_norm": 8.5, "learning_rate": 1.2228028490195831e-05, "loss": 1.3436, "step": 17825 }, { "epoch": 44.565, "grad_norm": 8.6875, "learning_rate": 1.2227239146788794e-05, "loss": 1.3758, "step": 17826 }, { "epoch": 44.5675, "grad_norm": 6.65625, "learning_rate": 1.222644978878007e-05, "loss": 1.2232, "step": 17827 }, { "epoch": 44.57, "grad_norm": 6.5625, "learning_rate": 1.2225660416174827e-05, "loss": 1.5053, "step": 17828 }, { "epoch": 44.5725, "grad_norm": 10.3125, "learning_rate": 1.2224871028978248e-05, "loss": 1.3399, "step": 17829 }, { "epoch": 44.575, "grad_norm": 6.59375, "learning_rate": 1.22240816271955e-05, "loss": 1.389, "step": 17830 }, { "epoch": 44.5775, "grad_norm": 5.625, "learning_rate": 1.2223292210831766e-05, "loss": 1.3352, "step": 17831 }, { "epoch": 44.58, "grad_norm": 9.1875, "learning_rate": 1.222250277989221e-05, "loss": 1.4108, "step": 17832 }, { "epoch": 44.5825, "grad_norm": 6.3125, "learning_rate": 1.2221713334382022e-05, "loss": 1.3057, "step": 17833 }, { "epoch": 44.585, "grad_norm": 9.75, "learning_rate": 1.2220923874306364e-05, "loss": 1.3923, "step": 17834 }, { "epoch": 44.5875, "grad_norm": 7.625, "learning_rate": 1.2220134399670424e-05, "loss": 1.3624, "step": 17835 }, { "epoch": 44.59, "grad_norm": 5.625, "learning_rate": 1.2219344910479368e-05, "loss": 1.5089, "step": 17836 }, { "epoch": 44.5925, "grad_norm": 6.40625, "learning_rate": 1.2218555406738378e-05, "loss": 1.2302, "step": 17837 }, { "epoch": 44.595, "grad_norm": 7.96875, "learning_rate": 1.2217765888452624e-05, "loss": 1.1913, "step": 17838 }, { "epoch": 44.5975, "grad_norm": 9.4375, "learning_rate": 1.2216976355627287e-05, "loss": 1.4347, "step": 17839 }, { "epoch": 44.6, "grad_norm": 7.21875, "learning_rate": 1.2216186808267544e-05, "loss": 1.253, "step": 17840 }, { "epoch": 44.6025, "grad_norm": 10.4375, "learning_rate": 1.2215397246378566e-05, "loss": 1.2986, "step": 17841 }, { "epoch": 44.605, "grad_norm": 15.625, "learning_rate": 1.2214607669965535e-05, "loss": 1.438, "step": 17842 }, { "epoch": 44.6075, "grad_norm": 10.0, "learning_rate": 1.2213818079033621e-05, "loss": 1.2764, "step": 17843 }, { "epoch": 44.61, "grad_norm": 6.96875, "learning_rate": 1.221302847358801e-05, "loss": 1.3697, "step": 17844 }, { "epoch": 44.6125, "grad_norm": 6.5, "learning_rate": 1.2212238853633869e-05, "loss": 1.3163, "step": 17845 }, { "epoch": 44.615, "grad_norm": 9.375, "learning_rate": 1.2211449219176379e-05, "loss": 1.3244, "step": 17846 }, { "epoch": 44.6175, "grad_norm": 9.875, "learning_rate": 1.2210659570220716e-05, "loss": 1.3691, "step": 17847 }, { "epoch": 44.62, "grad_norm": 7.4375, "learning_rate": 1.2209869906772058e-05, "loss": 1.4408, "step": 17848 }, { "epoch": 44.6225, "grad_norm": 7.59375, "learning_rate": 1.2209080228835581e-05, "loss": 1.3216, "step": 17849 }, { "epoch": 44.625, "grad_norm": 6.34375, "learning_rate": 1.2208290536416466e-05, "loss": 1.2584, "step": 17850 }, { "epoch": 44.6275, "grad_norm": 9.1875, "learning_rate": 1.2207500829519878e-05, "loss": 1.2326, "step": 17851 }, { "epoch": 44.63, "grad_norm": 6.71875, "learning_rate": 1.2206711108151011e-05, "loss": 1.214, "step": 17852 }, { "epoch": 44.6325, "grad_norm": 6.875, "learning_rate": 1.220592137231503e-05, "loss": 1.1905, "step": 17853 }, { "epoch": 44.635, "grad_norm": 7.3125, "learning_rate": 1.220513162201712e-05, "loss": 1.1768, "step": 17854 }, { "epoch": 44.6375, "grad_norm": 7.03125, "learning_rate": 1.2204341857262454e-05, "loss": 1.3831, "step": 17855 }, { "epoch": 44.64, "grad_norm": 8.5, "learning_rate": 1.2203552078056209e-05, "loss": 1.3227, "step": 17856 }, { "epoch": 44.6425, "grad_norm": 6.25, "learning_rate": 1.2202762284403566e-05, "loss": 1.2834, "step": 17857 }, { "epoch": 44.645, "grad_norm": 9.0, "learning_rate": 1.2201972476309702e-05, "loss": 1.3652, "step": 17858 }, { "epoch": 44.6475, "grad_norm": 8.8125, "learning_rate": 1.2201182653779795e-05, "loss": 1.2871, "step": 17859 }, { "epoch": 44.65, "grad_norm": 6.3125, "learning_rate": 1.2200392816819022e-05, "loss": 1.3075, "step": 17860 }, { "epoch": 44.6525, "grad_norm": 8.9375, "learning_rate": 1.2199602965432562e-05, "loss": 1.2104, "step": 17861 }, { "epoch": 44.655, "grad_norm": 5.71875, "learning_rate": 1.2198813099625592e-05, "loss": 1.3169, "step": 17862 }, { "epoch": 44.6575, "grad_norm": 7.65625, "learning_rate": 1.2198023219403292e-05, "loss": 1.2129, "step": 17863 }, { "epoch": 44.66, "grad_norm": 7.34375, "learning_rate": 1.2197233324770839e-05, "loss": 1.4166, "step": 17864 }, { "epoch": 44.6625, "grad_norm": 5.3125, "learning_rate": 1.2196443415733415e-05, "loss": 1.2186, "step": 17865 }, { "epoch": 44.665, "grad_norm": 6.3125, "learning_rate": 1.2195653492296192e-05, "loss": 1.3323, "step": 17866 }, { "epoch": 44.6675, "grad_norm": 9.125, "learning_rate": 1.2194863554464355e-05, "loss": 1.1944, "step": 17867 }, { "epoch": 44.67, "grad_norm": 7.375, "learning_rate": 1.219407360224308e-05, "loss": 1.4536, "step": 17868 }, { "epoch": 44.6725, "grad_norm": 7.5, "learning_rate": 1.2193283635637548e-05, "loss": 1.2981, "step": 17869 }, { "epoch": 44.675, "grad_norm": 6.65625, "learning_rate": 1.2192493654652933e-05, "loss": 1.4692, "step": 17870 }, { "epoch": 44.6775, "grad_norm": 7.53125, "learning_rate": 1.2191703659294419e-05, "loss": 1.2751, "step": 17871 }, { "epoch": 44.68, "grad_norm": 5.84375, "learning_rate": 1.2190913649567185e-05, "loss": 1.4739, "step": 17872 }, { "epoch": 44.6825, "grad_norm": 9.75, "learning_rate": 1.2190123625476404e-05, "loss": 1.3599, "step": 17873 }, { "epoch": 44.685, "grad_norm": 9.1875, "learning_rate": 1.2189333587027264e-05, "loss": 1.2416, "step": 17874 }, { "epoch": 44.6875, "grad_norm": 7.75, "learning_rate": 1.218854353422494e-05, "loss": 1.2818, "step": 17875 }, { "epoch": 44.69, "grad_norm": 7.0625, "learning_rate": 1.2187753467074613e-05, "loss": 1.321, "step": 17876 }, { "epoch": 44.6925, "grad_norm": 6.65625, "learning_rate": 1.2186963385581458e-05, "loss": 1.3543, "step": 17877 }, { "epoch": 44.695, "grad_norm": 14.0, "learning_rate": 1.2186173289750662e-05, "loss": 1.204, "step": 17878 }, { "epoch": 44.6975, "grad_norm": 9.25, "learning_rate": 1.21853831795874e-05, "loss": 1.3802, "step": 17879 }, { "epoch": 44.7, "grad_norm": 6.75, "learning_rate": 1.2184593055096853e-05, "loss": 1.239, "step": 17880 }, { "epoch": 44.7025, "grad_norm": 6.53125, "learning_rate": 1.2183802916284203e-05, "loss": 1.3386, "step": 17881 }, { "epoch": 44.705, "grad_norm": 11.75, "learning_rate": 1.2183012763154632e-05, "loss": 1.2916, "step": 17882 }, { "epoch": 44.7075, "grad_norm": 6.5625, "learning_rate": 1.2182222595713308e-05, "loss": 1.3239, "step": 17883 }, { "epoch": 44.71, "grad_norm": 6.90625, "learning_rate": 1.2181432413965428e-05, "loss": 1.3225, "step": 17884 }, { "epoch": 44.7125, "grad_norm": 10.25, "learning_rate": 1.218064221791616e-05, "loss": 1.3873, "step": 17885 }, { "epoch": 44.715, "grad_norm": 7.375, "learning_rate": 1.217985200757069e-05, "loss": 1.3175, "step": 17886 }, { "epoch": 44.7175, "grad_norm": 6.3125, "learning_rate": 1.2179061782934198e-05, "loss": 1.3359, "step": 17887 }, { "epoch": 44.72, "grad_norm": 13.9375, "learning_rate": 1.2178271544011864e-05, "loss": 1.361, "step": 17888 }, { "epoch": 44.7225, "grad_norm": 8.25, "learning_rate": 1.2177481290808869e-05, "loss": 1.379, "step": 17889 }, { "epoch": 44.725, "grad_norm": 9.0, "learning_rate": 1.2176691023330391e-05, "loss": 1.4944, "step": 17890 }, { "epoch": 44.7275, "grad_norm": 6.84375, "learning_rate": 1.2175900741581616e-05, "loss": 1.3213, "step": 17891 }, { "epoch": 44.73, "grad_norm": 6.9375, "learning_rate": 1.2175110445567722e-05, "loss": 1.1978, "step": 17892 }, { "epoch": 44.7325, "grad_norm": 8.1875, "learning_rate": 1.2174320135293894e-05, "loss": 1.2753, "step": 17893 }, { "epoch": 44.735, "grad_norm": 7.09375, "learning_rate": 1.2173529810765306e-05, "loss": 1.3887, "step": 17894 }, { "epoch": 44.7375, "grad_norm": 5.53125, "learning_rate": 1.2172739471987146e-05, "loss": 1.3427, "step": 17895 }, { "epoch": 44.74, "grad_norm": 7.9375, "learning_rate": 1.2171949118964592e-05, "loss": 1.2632, "step": 17896 }, { "epoch": 44.7425, "grad_norm": 5.21875, "learning_rate": 1.2171158751702828e-05, "loss": 1.2329, "step": 17897 }, { "epoch": 44.745, "grad_norm": 6.5625, "learning_rate": 1.217036837020703e-05, "loss": 1.3329, "step": 17898 }, { "epoch": 44.7475, "grad_norm": 7.53125, "learning_rate": 1.216957797448239e-05, "loss": 1.2556, "step": 17899 }, { "epoch": 44.75, "grad_norm": 7.40625, "learning_rate": 1.2168787564534078e-05, "loss": 1.2334, "step": 17900 }, { "epoch": 44.7525, "grad_norm": 8.125, "learning_rate": 1.2167997140367287e-05, "loss": 1.4015, "step": 17901 }, { "epoch": 44.755, "grad_norm": 6.46875, "learning_rate": 1.2167206701987189e-05, "loss": 1.3714, "step": 17902 }, { "epoch": 44.7575, "grad_norm": 7.75, "learning_rate": 1.2166416249398973e-05, "loss": 1.2989, "step": 17903 }, { "epoch": 44.76, "grad_norm": 8.625, "learning_rate": 1.2165625782607817e-05, "loss": 1.3866, "step": 17904 }, { "epoch": 44.7625, "grad_norm": 6.09375, "learning_rate": 1.2164835301618907e-05, "loss": 1.3417, "step": 17905 }, { "epoch": 44.765, "grad_norm": 7.6875, "learning_rate": 1.216404480643742e-05, "loss": 1.2747, "step": 17906 }, { "epoch": 44.7675, "grad_norm": 5.59375, "learning_rate": 1.2163254297068546e-05, "loss": 1.3314, "step": 17907 }, { "epoch": 44.77, "grad_norm": 5.90625, "learning_rate": 1.2162463773517462e-05, "loss": 1.1871, "step": 17908 }, { "epoch": 44.7725, "grad_norm": 6.90625, "learning_rate": 1.2161673235789351e-05, "loss": 1.3261, "step": 17909 }, { "epoch": 44.775, "grad_norm": 9.75, "learning_rate": 1.2160882683889397e-05, "loss": 1.2598, "step": 17910 }, { "epoch": 44.7775, "grad_norm": 7.90625, "learning_rate": 1.2160092117822781e-05, "loss": 1.2327, "step": 17911 }, { "epoch": 44.78, "grad_norm": 6.4375, "learning_rate": 1.2159301537594691e-05, "loss": 1.3183, "step": 17912 }, { "epoch": 44.7825, "grad_norm": 8.0625, "learning_rate": 1.2158510943210304e-05, "loss": 1.3109, "step": 17913 }, { "epoch": 44.785, "grad_norm": 6.6875, "learning_rate": 1.215772033467481e-05, "loss": 1.2927, "step": 17914 }, { "epoch": 44.7875, "grad_norm": 10.625, "learning_rate": 1.2156929711993381e-05, "loss": 1.1932, "step": 17915 }, { "epoch": 44.79, "grad_norm": 8.5625, "learning_rate": 1.2156139075171212e-05, "loss": 1.369, "step": 17916 }, { "epoch": 44.7925, "grad_norm": 6.09375, "learning_rate": 1.215534842421348e-05, "loss": 1.3472, "step": 17917 }, { "epoch": 44.795, "grad_norm": 5.78125, "learning_rate": 1.2154557759125369e-05, "loss": 1.3749, "step": 17918 }, { "epoch": 44.7975, "grad_norm": 7.9375, "learning_rate": 1.2153767079912067e-05, "loss": 1.5528, "step": 17919 }, { "epoch": 44.8, "grad_norm": 7.1875, "learning_rate": 1.215297638657875e-05, "loss": 1.3197, "step": 17920 }, { "epoch": 44.8025, "grad_norm": 8.25, "learning_rate": 1.2152185679130608e-05, "loss": 1.2889, "step": 17921 }, { "epoch": 44.805, "grad_norm": 5.21875, "learning_rate": 1.2151394957572824e-05, "loss": 1.1105, "step": 17922 }, { "epoch": 44.8075, "grad_norm": 11.5, "learning_rate": 1.2150604221910578e-05, "loss": 1.2379, "step": 17923 }, { "epoch": 44.81, "grad_norm": 8.75, "learning_rate": 1.2149813472149059e-05, "loss": 1.3439, "step": 17924 }, { "epoch": 44.8125, "grad_norm": 8.375, "learning_rate": 1.2149022708293447e-05, "loss": 1.355, "step": 17925 }, { "epoch": 44.815, "grad_norm": 5.3125, "learning_rate": 1.214823193034893e-05, "loss": 1.2511, "step": 17926 }, { "epoch": 44.8175, "grad_norm": 6.15625, "learning_rate": 1.214744113832069e-05, "loss": 1.4056, "step": 17927 }, { "epoch": 44.82, "grad_norm": 7.03125, "learning_rate": 1.2146650332213911e-05, "loss": 1.3268, "step": 17928 }, { "epoch": 44.8225, "grad_norm": 10.5625, "learning_rate": 1.214585951203378e-05, "loss": 1.6305, "step": 17929 }, { "epoch": 44.825, "grad_norm": 9.125, "learning_rate": 1.2145068677785479e-05, "loss": 1.3382, "step": 17930 }, { "epoch": 44.8275, "grad_norm": 6.03125, "learning_rate": 1.2144277829474192e-05, "loss": 1.1581, "step": 17931 }, { "epoch": 44.83, "grad_norm": 9.625, "learning_rate": 1.2143486967105107e-05, "loss": 1.3748, "step": 17932 }, { "epoch": 44.8325, "grad_norm": 5.5625, "learning_rate": 1.2142696090683405e-05, "loss": 1.2519, "step": 17933 }, { "epoch": 44.835, "grad_norm": 7.125, "learning_rate": 1.2141905200214277e-05, "loss": 1.3709, "step": 17934 }, { "epoch": 44.8375, "grad_norm": 8.875, "learning_rate": 1.21411142957029e-05, "loss": 1.4416, "step": 17935 }, { "epoch": 44.84, "grad_norm": 15.4375, "learning_rate": 1.2140323377154467e-05, "loss": 1.335, "step": 17936 }, { "epoch": 44.8425, "grad_norm": 7.9375, "learning_rate": 1.213953244457416e-05, "loss": 1.2347, "step": 17937 }, { "epoch": 44.845, "grad_norm": 7.75, "learning_rate": 1.213874149796716e-05, "loss": 1.421, "step": 17938 }, { "epoch": 44.8475, "grad_norm": 7.90625, "learning_rate": 1.213795053733866e-05, "loss": 1.2882, "step": 17939 }, { "epoch": 44.85, "grad_norm": 8.0625, "learning_rate": 1.2137159562693839e-05, "loss": 1.3794, "step": 17940 }, { "epoch": 44.8525, "grad_norm": 9.5, "learning_rate": 1.2136368574037886e-05, "loss": 1.2478, "step": 17941 }, { "epoch": 44.855, "grad_norm": 6.15625, "learning_rate": 1.2135577571375986e-05, "loss": 1.3253, "step": 17942 }, { "epoch": 44.8575, "grad_norm": 7.5625, "learning_rate": 1.2134786554713327e-05, "loss": 1.4522, "step": 17943 }, { "epoch": 44.86, "grad_norm": 5.1875, "learning_rate": 1.2133995524055094e-05, "loss": 1.326, "step": 17944 }, { "epoch": 44.8625, "grad_norm": 7.21875, "learning_rate": 1.2133204479406469e-05, "loss": 1.2761, "step": 17945 }, { "epoch": 44.865, "grad_norm": 6.125, "learning_rate": 1.2132413420772639e-05, "loss": 1.2435, "step": 17946 }, { "epoch": 44.8675, "grad_norm": 7.21875, "learning_rate": 1.2131622348158794e-05, "loss": 1.3733, "step": 17947 }, { "epoch": 44.87, "grad_norm": 5.96875, "learning_rate": 1.2130831261570117e-05, "loss": 1.4099, "step": 17948 }, { "epoch": 44.8725, "grad_norm": 6.15625, "learning_rate": 1.2130040161011799e-05, "loss": 1.3037, "step": 17949 }, { "epoch": 44.875, "grad_norm": 14.4375, "learning_rate": 1.212924904648902e-05, "loss": 1.279, "step": 17950 }, { "epoch": 44.8775, "grad_norm": 5.1875, "learning_rate": 1.2128457918006968e-05, "loss": 1.2411, "step": 17951 }, { "epoch": 44.88, "grad_norm": 10.0, "learning_rate": 1.2127666775570837e-05, "loss": 1.333, "step": 17952 }, { "epoch": 44.8825, "grad_norm": 7.40625, "learning_rate": 1.2126875619185801e-05, "loss": 1.367, "step": 17953 }, { "epoch": 44.885, "grad_norm": 5.96875, "learning_rate": 1.2126084448857059e-05, "loss": 1.2655, "step": 17954 }, { "epoch": 44.8875, "grad_norm": 7.1875, "learning_rate": 1.2125293264589786e-05, "loss": 1.3511, "step": 17955 }, { "epoch": 44.89, "grad_norm": 5.34375, "learning_rate": 1.2124502066389181e-05, "loss": 1.3792, "step": 17956 }, { "epoch": 44.8925, "grad_norm": 6.78125, "learning_rate": 1.2123710854260425e-05, "loss": 1.3497, "step": 17957 }, { "epoch": 44.895, "grad_norm": 7.96875, "learning_rate": 1.2122919628208705e-05, "loss": 1.402, "step": 17958 }, { "epoch": 44.8975, "grad_norm": 7.0625, "learning_rate": 1.2122128388239207e-05, "loss": 1.2765, "step": 17959 }, { "epoch": 44.9, "grad_norm": 7.53125, "learning_rate": 1.2121337134357121e-05, "loss": 1.325, "step": 17960 }, { "epoch": 44.9025, "grad_norm": 7.6875, "learning_rate": 1.2120545866567634e-05, "loss": 1.2314, "step": 17961 }, { "epoch": 44.905, "grad_norm": 8.375, "learning_rate": 1.2119754584875933e-05, "loss": 1.2647, "step": 17962 }, { "epoch": 44.9075, "grad_norm": 6.3125, "learning_rate": 1.2118963289287207e-05, "loss": 1.327, "step": 17963 }, { "epoch": 44.91, "grad_norm": 9.1875, "learning_rate": 1.2118171979806642e-05, "loss": 1.3823, "step": 17964 }, { "epoch": 44.9125, "grad_norm": 9.125, "learning_rate": 1.2117380656439423e-05, "loss": 1.3085, "step": 17965 }, { "epoch": 44.915, "grad_norm": 8.875, "learning_rate": 1.2116589319190746e-05, "loss": 1.3053, "step": 17966 }, { "epoch": 44.9175, "grad_norm": 9.375, "learning_rate": 1.2115797968065794e-05, "loss": 1.3905, "step": 17967 }, { "epoch": 44.92, "grad_norm": 15.9375, "learning_rate": 1.211500660306975e-05, "loss": 1.413, "step": 17968 }, { "epoch": 44.9225, "grad_norm": 6.71875, "learning_rate": 1.2114215224207813e-05, "loss": 1.3631, "step": 17969 }, { "epoch": 44.925, "grad_norm": 10.25, "learning_rate": 1.211342383148516e-05, "loss": 1.4692, "step": 17970 }, { "epoch": 44.9275, "grad_norm": 7.25, "learning_rate": 1.211263242490699e-05, "loss": 1.289, "step": 17971 }, { "epoch": 44.93, "grad_norm": 5.78125, "learning_rate": 1.2111841004478485e-05, "loss": 1.2948, "step": 17972 }, { "epoch": 44.9325, "grad_norm": 6.78125, "learning_rate": 1.2111049570204835e-05, "loss": 1.3051, "step": 17973 }, { "epoch": 44.935, "grad_norm": 8.375, "learning_rate": 1.2110258122091227e-05, "loss": 1.4632, "step": 17974 }, { "epoch": 44.9375, "grad_norm": 5.21875, "learning_rate": 1.2109466660142853e-05, "loss": 1.2341, "step": 17975 }, { "epoch": 44.94, "grad_norm": 8.625, "learning_rate": 1.2108675184364898e-05, "loss": 1.3985, "step": 17976 }, { "epoch": 44.9425, "grad_norm": 7.6875, "learning_rate": 1.2107883694762554e-05, "loss": 1.475, "step": 17977 }, { "epoch": 44.945, "grad_norm": 5.96875, "learning_rate": 1.210709219134101e-05, "loss": 1.2914, "step": 17978 }, { "epoch": 44.9475, "grad_norm": 7.6875, "learning_rate": 1.2106300674105452e-05, "loss": 1.4113, "step": 17979 }, { "epoch": 44.95, "grad_norm": 9.5, "learning_rate": 1.2105509143061072e-05, "loss": 1.2567, "step": 17980 }, { "epoch": 44.9525, "grad_norm": 7.40625, "learning_rate": 1.2104717598213056e-05, "loss": 1.2376, "step": 17981 }, { "epoch": 44.955, "grad_norm": 8.0, "learning_rate": 1.21039260395666e-05, "loss": 1.2829, "step": 17982 }, { "epoch": 44.9575, "grad_norm": 6.4375, "learning_rate": 1.2103134467126884e-05, "loss": 1.3382, "step": 17983 }, { "epoch": 44.96, "grad_norm": 7.15625, "learning_rate": 1.210234288089911e-05, "loss": 1.3666, "step": 17984 }, { "epoch": 44.9625, "grad_norm": 5.25, "learning_rate": 1.2101551280888453e-05, "loss": 1.2358, "step": 17985 }, { "epoch": 44.965, "grad_norm": 9.0, "learning_rate": 1.2100759667100116e-05, "loss": 1.2823, "step": 17986 }, { "epoch": 44.9675, "grad_norm": 7.125, "learning_rate": 1.2099968039539278e-05, "loss": 1.2627, "step": 17987 }, { "epoch": 44.97, "grad_norm": 6.75, "learning_rate": 1.2099176398211134e-05, "loss": 1.4075, "step": 17988 }, { "epoch": 44.9725, "grad_norm": 9.5625, "learning_rate": 1.2098384743120875e-05, "loss": 1.3032, "step": 17989 }, { "epoch": 44.975, "grad_norm": 8.4375, "learning_rate": 1.2097593074273688e-05, "loss": 1.4595, "step": 17990 }, { "epoch": 44.9775, "grad_norm": 7.8125, "learning_rate": 1.2096801391674764e-05, "loss": 1.2779, "step": 17991 }, { "epoch": 44.98, "grad_norm": 29.0, "learning_rate": 1.2096009695329298e-05, "loss": 1.4186, "step": 17992 }, { "epoch": 44.9825, "grad_norm": 6.4375, "learning_rate": 1.2095217985242474e-05, "loss": 1.252, "step": 17993 }, { "epoch": 44.985, "grad_norm": 9.6875, "learning_rate": 1.2094426261419486e-05, "loss": 1.3659, "step": 17994 }, { "epoch": 44.9875, "grad_norm": 5.90625, "learning_rate": 1.2093634523865519e-05, "loss": 1.5614, "step": 17995 }, { "epoch": 44.99, "grad_norm": 7.90625, "learning_rate": 1.2092842772585773e-05, "loss": 1.3558, "step": 17996 }, { "epoch": 44.9925, "grad_norm": 7.25, "learning_rate": 1.2092051007585427e-05, "loss": 1.2721, "step": 17997 }, { "epoch": 44.995, "grad_norm": 6.78125, "learning_rate": 1.2091259228869687e-05, "loss": 1.2207, "step": 17998 }, { "epoch": 44.9975, "grad_norm": 7.5625, "learning_rate": 1.2090467436443731e-05, "loss": 1.3116, "step": 17999 }, { "epoch": 45.0, "grad_norm": 7.84375, "learning_rate": 1.2089675630312755e-05, "loss": 1.233, "step": 18000 }, { "epoch": 45.0025, "grad_norm": 6.59375, "learning_rate": 1.2088883810481946e-05, "loss": 1.395, "step": 18001 }, { "epoch": 45.005, "grad_norm": 6.75, "learning_rate": 1.2088091976956498e-05, "loss": 1.3063, "step": 18002 }, { "epoch": 45.0075, "grad_norm": 7.375, "learning_rate": 1.2087300129741609e-05, "loss": 1.1984, "step": 18003 }, { "epoch": 45.01, "grad_norm": 10.625, "learning_rate": 1.2086508268842459e-05, "loss": 1.293, "step": 18004 }, { "epoch": 45.0125, "grad_norm": 8.4375, "learning_rate": 1.2085716394264244e-05, "loss": 1.2007, "step": 18005 }, { "epoch": 45.015, "grad_norm": 6.5625, "learning_rate": 1.2084924506012157e-05, "loss": 1.3482, "step": 18006 }, { "epoch": 45.0175, "grad_norm": 8.1875, "learning_rate": 1.2084132604091388e-05, "loss": 1.2113, "step": 18007 }, { "epoch": 45.02, "grad_norm": 9.4375, "learning_rate": 1.208334068850713e-05, "loss": 1.3288, "step": 18008 }, { "epoch": 45.0225, "grad_norm": 9.75, "learning_rate": 1.2082548759264574e-05, "loss": 1.3778, "step": 18009 }, { "epoch": 45.025, "grad_norm": 12.0, "learning_rate": 1.2081756816368911e-05, "loss": 1.4338, "step": 18010 }, { "epoch": 45.0275, "grad_norm": 8.1875, "learning_rate": 1.2080964859825335e-05, "loss": 1.381, "step": 18011 }, { "epoch": 45.03, "grad_norm": 7.75, "learning_rate": 1.2080172889639033e-05, "loss": 1.2681, "step": 18012 }, { "epoch": 45.0325, "grad_norm": 18.125, "learning_rate": 1.2079380905815203e-05, "loss": 1.5345, "step": 18013 }, { "epoch": 45.035, "grad_norm": 7.8125, "learning_rate": 1.2078588908359038e-05, "loss": 1.4487, "step": 18014 }, { "epoch": 45.0375, "grad_norm": 7.78125, "learning_rate": 1.2077796897275722e-05, "loss": 1.4835, "step": 18015 }, { "epoch": 45.04, "grad_norm": 9.625, "learning_rate": 1.2077004872570454e-05, "loss": 1.3819, "step": 18016 }, { "epoch": 45.0425, "grad_norm": 6.6875, "learning_rate": 1.2076212834248426e-05, "loss": 1.3005, "step": 18017 }, { "epoch": 45.045, "grad_norm": 9.125, "learning_rate": 1.207542078231483e-05, "loss": 1.3397, "step": 18018 }, { "epoch": 45.0475, "grad_norm": 8.375, "learning_rate": 1.2074628716774858e-05, "loss": 1.4765, "step": 18019 }, { "epoch": 45.05, "grad_norm": 6.96875, "learning_rate": 1.2073836637633705e-05, "loss": 1.2908, "step": 18020 }, { "epoch": 45.0525, "grad_norm": 6.03125, "learning_rate": 1.2073044544896559e-05, "loss": 1.3585, "step": 18021 }, { "epoch": 45.055, "grad_norm": 6.03125, "learning_rate": 1.2072252438568617e-05, "loss": 1.216, "step": 18022 }, { "epoch": 45.0575, "grad_norm": 7.65625, "learning_rate": 1.207146031865507e-05, "loss": 1.4468, "step": 18023 }, { "epoch": 45.06, "grad_norm": 7.0, "learning_rate": 1.2070668185161116e-05, "loss": 1.2386, "step": 18024 }, { "epoch": 45.0625, "grad_norm": 7.28125, "learning_rate": 1.2069876038091941e-05, "loss": 1.3029, "step": 18025 }, { "epoch": 45.065, "grad_norm": 7.40625, "learning_rate": 1.2069083877452743e-05, "loss": 1.324, "step": 18026 }, { "epoch": 45.0675, "grad_norm": 9.0625, "learning_rate": 1.2068291703248711e-05, "loss": 1.2579, "step": 18027 }, { "epoch": 45.07, "grad_norm": 11.6875, "learning_rate": 1.2067499515485046e-05, "loss": 1.3652, "step": 18028 }, { "epoch": 45.0725, "grad_norm": 8.75, "learning_rate": 1.2066707314166934e-05, "loss": 1.3331, "step": 18029 }, { "epoch": 45.075, "grad_norm": 6.3125, "learning_rate": 1.206591509929957e-05, "loss": 1.2095, "step": 18030 }, { "epoch": 45.0775, "grad_norm": 7.5625, "learning_rate": 1.206512287088815e-05, "loss": 1.2415, "step": 18031 }, { "epoch": 45.08, "grad_norm": 9.0625, "learning_rate": 1.206433062893787e-05, "loss": 1.4531, "step": 18032 }, { "epoch": 45.0825, "grad_norm": 13.0625, "learning_rate": 1.2063538373453917e-05, "loss": 1.4602, "step": 18033 }, { "epoch": 45.085, "grad_norm": 8.375, "learning_rate": 1.2062746104441493e-05, "loss": 1.2906, "step": 18034 }, { "epoch": 45.0875, "grad_norm": 6.59375, "learning_rate": 1.2061953821905785e-05, "loss": 1.3195, "step": 18035 }, { "epoch": 45.09, "grad_norm": 6.53125, "learning_rate": 1.2061161525851991e-05, "loss": 1.3445, "step": 18036 }, { "epoch": 45.0925, "grad_norm": 7.84375, "learning_rate": 1.2060369216285305e-05, "loss": 1.3555, "step": 18037 }, { "epoch": 45.095, "grad_norm": 10.5625, "learning_rate": 1.2059576893210922e-05, "loss": 1.3177, "step": 18038 }, { "epoch": 45.0975, "grad_norm": 9.3125, "learning_rate": 1.2058784556634037e-05, "loss": 1.3324, "step": 18039 }, { "epoch": 45.1, "grad_norm": 7.03125, "learning_rate": 1.2057992206559837e-05, "loss": 1.2637, "step": 18040 }, { "epoch": 45.1025, "grad_norm": 7.96875, "learning_rate": 1.2057199842993528e-05, "loss": 1.4857, "step": 18041 }, { "epoch": 45.105, "grad_norm": 7.84375, "learning_rate": 1.2056407465940294e-05, "loss": 1.3572, "step": 18042 }, { "epoch": 45.1075, "grad_norm": 7.125, "learning_rate": 1.205561507540534e-05, "loss": 1.38, "step": 18043 }, { "epoch": 45.11, "grad_norm": 8.9375, "learning_rate": 1.2054822671393852e-05, "loss": 1.3548, "step": 18044 }, { "epoch": 45.1125, "grad_norm": 10.9375, "learning_rate": 1.205403025391103e-05, "loss": 1.4064, "step": 18045 }, { "epoch": 45.115, "grad_norm": 7.21875, "learning_rate": 1.205323782296207e-05, "loss": 1.1867, "step": 18046 }, { "epoch": 45.1175, "grad_norm": 6.625, "learning_rate": 1.2052445378552163e-05, "loss": 1.3758, "step": 18047 }, { "epoch": 45.12, "grad_norm": 8.125, "learning_rate": 1.2051652920686505e-05, "loss": 1.3157, "step": 18048 }, { "epoch": 45.1225, "grad_norm": 5.53125, "learning_rate": 1.2050860449370293e-05, "loss": 1.2223, "step": 18049 }, { "epoch": 45.125, "grad_norm": 8.625, "learning_rate": 1.2050067964608725e-05, "loss": 1.3224, "step": 18050 }, { "epoch": 45.1275, "grad_norm": 9.4375, "learning_rate": 1.2049275466406991e-05, "loss": 1.352, "step": 18051 }, { "epoch": 45.13, "grad_norm": 12.625, "learning_rate": 1.2048482954770289e-05, "loss": 1.3416, "step": 18052 }, { "epoch": 45.1325, "grad_norm": 7.25, "learning_rate": 1.2047690429703815e-05, "loss": 1.1796, "step": 18053 }, { "epoch": 45.135, "grad_norm": 8.9375, "learning_rate": 1.2046897891212766e-05, "loss": 1.2747, "step": 18054 }, { "epoch": 45.1375, "grad_norm": 9.125, "learning_rate": 1.2046105339302332e-05, "loss": 1.5373, "step": 18055 }, { "epoch": 45.14, "grad_norm": 8.6875, "learning_rate": 1.204531277397772e-05, "loss": 1.3986, "step": 18056 }, { "epoch": 45.1425, "grad_norm": 7.28125, "learning_rate": 1.204452019524411e-05, "loss": 1.2819, "step": 18057 }, { "epoch": 45.145, "grad_norm": 6.5, "learning_rate": 1.2043727603106717e-05, "loss": 1.3199, "step": 18058 }, { "epoch": 45.1475, "grad_norm": 9.625, "learning_rate": 1.2042934997570723e-05, "loss": 1.3513, "step": 18059 }, { "epoch": 45.15, "grad_norm": 5.75, "learning_rate": 1.204214237864133e-05, "loss": 1.2353, "step": 18060 }, { "epoch": 45.1525, "grad_norm": 6.375, "learning_rate": 1.2041349746323732e-05, "loss": 1.1944, "step": 18061 }, { "epoch": 45.155, "grad_norm": 6.125, "learning_rate": 1.2040557100623127e-05, "loss": 1.3657, "step": 18062 }, { "epoch": 45.1575, "grad_norm": 10.5625, "learning_rate": 1.2039764441544712e-05, "loss": 1.4494, "step": 18063 }, { "epoch": 45.16, "grad_norm": 7.3125, "learning_rate": 1.2038971769093685e-05, "loss": 1.235, "step": 18064 }, { "epoch": 45.1625, "grad_norm": 7.0, "learning_rate": 1.2038179083275236e-05, "loss": 1.4014, "step": 18065 }, { "epoch": 45.165, "grad_norm": 6.96875, "learning_rate": 1.2037386384094572e-05, "loss": 1.2662, "step": 18066 }, { "epoch": 45.1675, "grad_norm": 9.125, "learning_rate": 1.2036593671556877e-05, "loss": 1.2885, "step": 18067 }, { "epoch": 45.17, "grad_norm": 6.78125, "learning_rate": 1.2035800945667365e-05, "loss": 1.2458, "step": 18068 }, { "epoch": 45.1725, "grad_norm": 7.3125, "learning_rate": 1.2035008206431215e-05, "loss": 1.2815, "step": 18069 }, { "epoch": 45.175, "grad_norm": 7.0, "learning_rate": 1.2034215453853636e-05, "loss": 1.456, "step": 18070 }, { "epoch": 45.1775, "grad_norm": 8.0625, "learning_rate": 1.2033422687939822e-05, "loss": 1.3457, "step": 18071 }, { "epoch": 45.18, "grad_norm": 4.9375, "learning_rate": 1.2032629908694969e-05, "loss": 1.2557, "step": 18072 }, { "epoch": 45.1825, "grad_norm": 9.1875, "learning_rate": 1.2031837116124278e-05, "loss": 1.402, "step": 18073 }, { "epoch": 45.185, "grad_norm": 8.9375, "learning_rate": 1.2031044310232941e-05, "loss": 1.2536, "step": 18074 }, { "epoch": 45.1875, "grad_norm": 12.6875, "learning_rate": 1.2030251491026162e-05, "loss": 1.3299, "step": 18075 }, { "epoch": 45.19, "grad_norm": 6.3125, "learning_rate": 1.2029458658509135e-05, "loss": 1.1271, "step": 18076 }, { "epoch": 45.1925, "grad_norm": 26.125, "learning_rate": 1.2028665812687055e-05, "loss": 1.4032, "step": 18077 }, { "epoch": 45.195, "grad_norm": 7.5625, "learning_rate": 1.2027872953565125e-05, "loss": 1.2926, "step": 18078 }, { "epoch": 45.1975, "grad_norm": 9.4375, "learning_rate": 1.2027080081148546e-05, "loss": 1.2898, "step": 18079 }, { "epoch": 45.2, "grad_norm": 9.0625, "learning_rate": 1.2026287195442503e-05, "loss": 1.3957, "step": 18080 }, { "epoch": 45.2025, "grad_norm": 7.53125, "learning_rate": 1.2025494296452206e-05, "loss": 1.3573, "step": 18081 }, { "epoch": 45.205, "grad_norm": 7.0625, "learning_rate": 1.2024701384182848e-05, "loss": 1.1783, "step": 18082 }, { "epoch": 45.2075, "grad_norm": 6.375, "learning_rate": 1.2023908458639632e-05, "loss": 1.3048, "step": 18083 }, { "epoch": 45.21, "grad_norm": 6.71875, "learning_rate": 1.2023115519827751e-05, "loss": 1.3517, "step": 18084 }, { "epoch": 45.2125, "grad_norm": 15.625, "learning_rate": 1.2022322567752407e-05, "loss": 1.519, "step": 18085 }, { "epoch": 45.215, "grad_norm": 9.1875, "learning_rate": 1.2021529602418795e-05, "loss": 1.5316, "step": 18086 }, { "epoch": 45.2175, "grad_norm": 7.5625, "learning_rate": 1.2020736623832117e-05, "loss": 1.3703, "step": 18087 }, { "epoch": 45.22, "grad_norm": 5.6875, "learning_rate": 1.201994363199757e-05, "loss": 1.363, "step": 18088 }, { "epoch": 45.2225, "grad_norm": 7.0, "learning_rate": 1.2019150626920353e-05, "loss": 1.2283, "step": 18089 }, { "epoch": 45.225, "grad_norm": 8.875, "learning_rate": 1.2018357608605667e-05, "loss": 1.4595, "step": 18090 }, { "epoch": 45.2275, "grad_norm": 6.75, "learning_rate": 1.2017564577058707e-05, "loss": 1.4037, "step": 18091 }, { "epoch": 45.23, "grad_norm": 7.625, "learning_rate": 1.2016771532284675e-05, "loss": 1.3581, "step": 18092 }, { "epoch": 45.2325, "grad_norm": 8.25, "learning_rate": 1.201597847428877e-05, "loss": 1.2669, "step": 18093 }, { "epoch": 45.235, "grad_norm": 7.78125, "learning_rate": 1.2015185403076193e-05, "loss": 1.3228, "step": 18094 }, { "epoch": 45.2375, "grad_norm": 7.65625, "learning_rate": 1.2014392318652137e-05, "loss": 1.3055, "step": 18095 }, { "epoch": 45.24, "grad_norm": 7.28125, "learning_rate": 1.201359922102181e-05, "loss": 1.2955, "step": 18096 }, { "epoch": 45.2425, "grad_norm": 6.0625, "learning_rate": 1.2012806110190402e-05, "loss": 1.4548, "step": 18097 }, { "epoch": 45.245, "grad_norm": 8.0625, "learning_rate": 1.2012012986163124e-05, "loss": 1.2011, "step": 18098 }, { "epoch": 45.2475, "grad_norm": 7.125, "learning_rate": 1.2011219848945163e-05, "loss": 1.287, "step": 18099 }, { "epoch": 45.25, "grad_norm": 5.71875, "learning_rate": 1.2010426698541728e-05, "loss": 1.2649, "step": 18100 }, { "epoch": 45.2525, "grad_norm": 6.375, "learning_rate": 1.2009633534958018e-05, "loss": 1.2073, "step": 18101 }, { "epoch": 45.255, "grad_norm": 6.8125, "learning_rate": 1.2008840358199227e-05, "loss": 1.3575, "step": 18102 }, { "epoch": 45.2575, "grad_norm": 6.375, "learning_rate": 1.2008047168270562e-05, "loss": 1.2941, "step": 18103 }, { "epoch": 45.26, "grad_norm": 8.0, "learning_rate": 1.2007253965177219e-05, "loss": 1.3361, "step": 18104 }, { "epoch": 45.2625, "grad_norm": 5.59375, "learning_rate": 1.20064607489244e-05, "loss": 1.4218, "step": 18105 }, { "epoch": 45.265, "grad_norm": 11.25, "learning_rate": 1.2005667519517303e-05, "loss": 1.4488, "step": 18106 }, { "epoch": 45.2675, "grad_norm": 10.4375, "learning_rate": 1.2004874276961132e-05, "loss": 1.3831, "step": 18107 }, { "epoch": 45.27, "grad_norm": 9.0, "learning_rate": 1.2004081021261083e-05, "loss": 1.4426, "step": 18108 }, { "epoch": 45.2725, "grad_norm": 8.6875, "learning_rate": 1.2003287752422364e-05, "loss": 1.3643, "step": 18109 }, { "epoch": 45.275, "grad_norm": 8.625, "learning_rate": 1.2002494470450164e-05, "loss": 1.3566, "step": 18110 }, { "epoch": 45.2775, "grad_norm": 7.875, "learning_rate": 1.2001701175349696e-05, "loss": 1.2946, "step": 18111 }, { "epoch": 45.28, "grad_norm": 6.5625, "learning_rate": 1.200090786712615e-05, "loss": 1.3143, "step": 18112 }, { "epoch": 45.2825, "grad_norm": 5.75, "learning_rate": 1.2000114545784736e-05, "loss": 1.1976, "step": 18113 }, { "epoch": 45.285, "grad_norm": 8.375, "learning_rate": 1.199932121133065e-05, "loss": 1.4825, "step": 18114 }, { "epoch": 45.2875, "grad_norm": 5.6875, "learning_rate": 1.1998527863769094e-05, "loss": 1.3934, "step": 18115 }, { "epoch": 45.29, "grad_norm": 7.46875, "learning_rate": 1.199773450310527e-05, "loss": 1.2425, "step": 18116 }, { "epoch": 45.2925, "grad_norm": 7.53125, "learning_rate": 1.1996941129344376e-05, "loss": 1.3706, "step": 18117 }, { "epoch": 45.295, "grad_norm": 6.0625, "learning_rate": 1.199614774249162e-05, "loss": 1.4164, "step": 18118 }, { "epoch": 45.2975, "grad_norm": 7.53125, "learning_rate": 1.1995354342552197e-05, "loss": 1.3566, "step": 18119 }, { "epoch": 45.3, "grad_norm": 7.4375, "learning_rate": 1.199456092953131e-05, "loss": 1.301, "step": 18120 }, { "epoch": 45.3025, "grad_norm": 8.25, "learning_rate": 1.1993767503434161e-05, "loss": 1.3546, "step": 18121 }, { "epoch": 45.305, "grad_norm": 7.6875, "learning_rate": 1.1992974064265954e-05, "loss": 1.2765, "step": 18122 }, { "epoch": 45.3075, "grad_norm": 7.0625, "learning_rate": 1.1992180612031888e-05, "loss": 1.3059, "step": 18123 }, { "epoch": 45.31, "grad_norm": 7.625, "learning_rate": 1.1991387146737165e-05, "loss": 1.3987, "step": 18124 }, { "epoch": 45.3125, "grad_norm": 9.0625, "learning_rate": 1.199059366838699e-05, "loss": 1.4292, "step": 18125 }, { "epoch": 45.315, "grad_norm": 7.09375, "learning_rate": 1.1989800176986562e-05, "loss": 1.3753, "step": 18126 }, { "epoch": 45.3175, "grad_norm": 8.0, "learning_rate": 1.198900667254108e-05, "loss": 1.2266, "step": 18127 }, { "epoch": 45.32, "grad_norm": 11.5, "learning_rate": 1.1988213155055754e-05, "loss": 1.3698, "step": 18128 }, { "epoch": 45.3225, "grad_norm": 6.28125, "learning_rate": 1.198741962453578e-05, "loss": 1.2775, "step": 18129 }, { "epoch": 45.325, "grad_norm": 8.0, "learning_rate": 1.1986626080986363e-05, "loss": 1.4877, "step": 18130 }, { "epoch": 45.3275, "grad_norm": 10.5625, "learning_rate": 1.1985832524412706e-05, "loss": 1.2943, "step": 18131 }, { "epoch": 45.33, "grad_norm": 6.5, "learning_rate": 1.1985038954820008e-05, "loss": 1.2771, "step": 18132 }, { "epoch": 45.3325, "grad_norm": 9.0625, "learning_rate": 1.1984245372213476e-05, "loss": 1.3982, "step": 18133 }, { "epoch": 45.335, "grad_norm": 6.75, "learning_rate": 1.198345177659831e-05, "loss": 1.3457, "step": 18134 }, { "epoch": 45.3375, "grad_norm": 7.625, "learning_rate": 1.1982658167979713e-05, "loss": 1.2048, "step": 18135 }, { "epoch": 45.34, "grad_norm": 7.34375, "learning_rate": 1.1981864546362891e-05, "loss": 1.2025, "step": 18136 }, { "epoch": 45.3425, "grad_norm": 13.75, "learning_rate": 1.1981070911753041e-05, "loss": 1.4572, "step": 18137 }, { "epoch": 45.345, "grad_norm": 7.34375, "learning_rate": 1.1980277264155371e-05, "loss": 1.2422, "step": 18138 }, { "epoch": 45.3475, "grad_norm": 9.1875, "learning_rate": 1.1979483603575081e-05, "loss": 1.3798, "step": 18139 }, { "epoch": 45.35, "grad_norm": 6.78125, "learning_rate": 1.197868993001738e-05, "loss": 1.4332, "step": 18140 }, { "epoch": 45.3525, "grad_norm": 5.71875, "learning_rate": 1.1977896243487464e-05, "loss": 1.2877, "step": 18141 }, { "epoch": 45.355, "grad_norm": 12.6875, "learning_rate": 1.1977102543990535e-05, "loss": 1.266, "step": 18142 }, { "epoch": 45.3575, "grad_norm": 6.875, "learning_rate": 1.1976308831531807e-05, "loss": 1.4247, "step": 18143 }, { "epoch": 45.36, "grad_norm": 6.5625, "learning_rate": 1.1975515106116472e-05, "loss": 1.267, "step": 18144 }, { "epoch": 45.3625, "grad_norm": 5.125, "learning_rate": 1.1974721367749747e-05, "loss": 1.5444, "step": 18145 }, { "epoch": 45.365, "grad_norm": 5.90625, "learning_rate": 1.1973927616436823e-05, "loss": 1.4217, "step": 18146 }, { "epoch": 45.3675, "grad_norm": 6.21875, "learning_rate": 1.1973133852182908e-05, "loss": 1.4327, "step": 18147 }, { "epoch": 45.37, "grad_norm": 9.6875, "learning_rate": 1.1972340074993207e-05, "loss": 1.3826, "step": 18148 }, { "epoch": 45.3725, "grad_norm": 8.0625, "learning_rate": 1.1971546284872923e-05, "loss": 1.3099, "step": 18149 }, { "epoch": 45.375, "grad_norm": 7.5625, "learning_rate": 1.1970752481827261e-05, "loss": 1.3928, "step": 18150 }, { "epoch": 45.3775, "grad_norm": 6.90625, "learning_rate": 1.1969958665861428e-05, "loss": 1.4086, "step": 18151 }, { "epoch": 45.38, "grad_norm": 9.4375, "learning_rate": 1.1969164836980618e-05, "loss": 1.4481, "step": 18152 }, { "epoch": 45.3825, "grad_norm": 5.1875, "learning_rate": 1.1968370995190048e-05, "loss": 1.3753, "step": 18153 }, { "epoch": 45.385, "grad_norm": 6.8125, "learning_rate": 1.1967577140494913e-05, "loss": 1.077, "step": 18154 }, { "epoch": 45.3875, "grad_norm": 7.875, "learning_rate": 1.1966783272900425e-05, "loss": 1.2717, "step": 18155 }, { "epoch": 45.39, "grad_norm": 9.6875, "learning_rate": 1.196598939241178e-05, "loss": 1.2479, "step": 18156 }, { "epoch": 45.3925, "grad_norm": 6.65625, "learning_rate": 1.196519549903419e-05, "loss": 1.331, "step": 18157 }, { "epoch": 45.395, "grad_norm": 7.03125, "learning_rate": 1.1964401592772855e-05, "loss": 1.3432, "step": 18158 }, { "epoch": 45.3975, "grad_norm": 7.40625, "learning_rate": 1.1963607673632981e-05, "loss": 1.2618, "step": 18159 }, { "epoch": 45.4, "grad_norm": 13.0625, "learning_rate": 1.1962813741619777e-05, "loss": 1.3575, "step": 18160 }, { "epoch": 45.4025, "grad_norm": 10.8125, "learning_rate": 1.1962019796738442e-05, "loss": 1.2576, "step": 18161 }, { "epoch": 45.405, "grad_norm": 6.5, "learning_rate": 1.1961225838994186e-05, "loss": 1.5405, "step": 18162 }, { "epoch": 45.4075, "grad_norm": 5.90625, "learning_rate": 1.196043186839221e-05, "loss": 1.2666, "step": 18163 }, { "epoch": 45.41, "grad_norm": 7.65625, "learning_rate": 1.195963788493772e-05, "loss": 1.4384, "step": 18164 }, { "epoch": 45.4125, "grad_norm": 5.625, "learning_rate": 1.1958843888635925e-05, "loss": 1.371, "step": 18165 }, { "epoch": 45.415, "grad_norm": 8.125, "learning_rate": 1.1958049879492028e-05, "loss": 1.3085, "step": 18166 }, { "epoch": 45.4175, "grad_norm": 9.0625, "learning_rate": 1.1957255857511231e-05, "loss": 1.2673, "step": 18167 }, { "epoch": 45.42, "grad_norm": 5.4375, "learning_rate": 1.1956461822698748e-05, "loss": 1.3219, "step": 18168 }, { "epoch": 45.4225, "grad_norm": 4.9375, "learning_rate": 1.1955667775059776e-05, "loss": 1.309, "step": 18169 }, { "epoch": 45.425, "grad_norm": 6.65625, "learning_rate": 1.1954873714599526e-05, "loss": 1.298, "step": 18170 }, { "epoch": 45.4275, "grad_norm": 8.4375, "learning_rate": 1.1954079641323202e-05, "loss": 1.2849, "step": 18171 }, { "epoch": 45.43, "grad_norm": 7.28125, "learning_rate": 1.195328555523601e-05, "loss": 1.4318, "step": 18172 }, { "epoch": 45.4325, "grad_norm": 6.96875, "learning_rate": 1.1952491456343154e-05, "loss": 1.3163, "step": 18173 }, { "epoch": 45.435, "grad_norm": 6.15625, "learning_rate": 1.1951697344649844e-05, "loss": 1.3448, "step": 18174 }, { "epoch": 45.4375, "grad_norm": 11.25, "learning_rate": 1.1950903220161286e-05, "loss": 1.2741, "step": 18175 }, { "epoch": 45.44, "grad_norm": 6.6875, "learning_rate": 1.1950109082882681e-05, "loss": 1.287, "step": 18176 }, { "epoch": 45.4425, "grad_norm": 6.25, "learning_rate": 1.194931493281924e-05, "loss": 1.3252, "step": 18177 }, { "epoch": 45.445, "grad_norm": 7.125, "learning_rate": 1.1948520769976169e-05, "loss": 1.3296, "step": 18178 }, { "epoch": 45.4475, "grad_norm": 7.0625, "learning_rate": 1.1947726594358672e-05, "loss": 1.2876, "step": 18179 }, { "epoch": 45.45, "grad_norm": 6.96875, "learning_rate": 1.194693240597196e-05, "loss": 1.1792, "step": 18180 }, { "epoch": 45.4525, "grad_norm": 8.0625, "learning_rate": 1.1946138204821238e-05, "loss": 1.352, "step": 18181 }, { "epoch": 45.455, "grad_norm": 6.5625, "learning_rate": 1.1945343990911707e-05, "loss": 1.308, "step": 18182 }, { "epoch": 45.4575, "grad_norm": 8.625, "learning_rate": 1.1944549764248583e-05, "loss": 1.1277, "step": 18183 }, { "epoch": 45.46, "grad_norm": 5.15625, "learning_rate": 1.1943755524837067e-05, "loss": 1.3769, "step": 18184 }, { "epoch": 45.4625, "grad_norm": 18.0, "learning_rate": 1.1942961272682368e-05, "loss": 1.4086, "step": 18185 }, { "epoch": 45.465, "grad_norm": 5.75, "learning_rate": 1.1942167007789691e-05, "loss": 1.3162, "step": 18186 }, { "epoch": 45.4675, "grad_norm": 9.25, "learning_rate": 1.1941372730164246e-05, "loss": 1.3288, "step": 18187 }, { "epoch": 45.47, "grad_norm": 6.34375, "learning_rate": 1.1940578439811241e-05, "loss": 1.4534, "step": 18188 }, { "epoch": 45.4725, "grad_norm": 7.5625, "learning_rate": 1.193978413673588e-05, "loss": 1.331, "step": 18189 }, { "epoch": 45.475, "grad_norm": 9.25, "learning_rate": 1.1938989820943369e-05, "loss": 1.4078, "step": 18190 }, { "epoch": 45.4775, "grad_norm": 6.15625, "learning_rate": 1.1938195492438923e-05, "loss": 1.4001, "step": 18191 }, { "epoch": 45.48, "grad_norm": 11.6875, "learning_rate": 1.193740115122774e-05, "loss": 1.3979, "step": 18192 }, { "epoch": 45.4825, "grad_norm": 10.6875, "learning_rate": 1.1936606797315035e-05, "loss": 1.3492, "step": 18193 }, { "epoch": 45.485, "grad_norm": 5.625, "learning_rate": 1.1935812430706016e-05, "loss": 1.2366, "step": 18194 }, { "epoch": 45.4875, "grad_norm": 6.3125, "learning_rate": 1.1935018051405885e-05, "loss": 1.4098, "step": 18195 }, { "epoch": 45.49, "grad_norm": 5.1875, "learning_rate": 1.1934223659419855e-05, "loss": 1.3974, "step": 18196 }, { "epoch": 45.4925, "grad_norm": 11.5, "learning_rate": 1.1933429254753129e-05, "loss": 1.3378, "step": 18197 }, { "epoch": 45.495, "grad_norm": 10.0625, "learning_rate": 1.1932634837410922e-05, "loss": 1.3075, "step": 18198 }, { "epoch": 45.4975, "grad_norm": 6.21875, "learning_rate": 1.1931840407398434e-05, "loss": 1.1588, "step": 18199 }, { "epoch": 45.5, "grad_norm": 10.5, "learning_rate": 1.1931045964720882e-05, "loss": 1.5717, "step": 18200 }, { "epoch": 45.5025, "grad_norm": 13.0, "learning_rate": 1.1930251509383469e-05, "loss": 1.3486, "step": 18201 }, { "epoch": 45.505, "grad_norm": 6.1875, "learning_rate": 1.19294570413914e-05, "loss": 1.3879, "step": 18202 }, { "epoch": 45.5075, "grad_norm": 7.3125, "learning_rate": 1.1928662560749893e-05, "loss": 1.3067, "step": 18203 }, { "epoch": 45.51, "grad_norm": 7.15625, "learning_rate": 1.1927868067464147e-05, "loss": 1.3256, "step": 18204 }, { "epoch": 45.5125, "grad_norm": 11.875, "learning_rate": 1.192707356153938e-05, "loss": 1.3021, "step": 18205 }, { "epoch": 45.515, "grad_norm": 5.78125, "learning_rate": 1.1926279042980795e-05, "loss": 1.2977, "step": 18206 }, { "epoch": 45.5175, "grad_norm": 10.5, "learning_rate": 1.1925484511793598e-05, "loss": 1.2929, "step": 18207 }, { "epoch": 45.52, "grad_norm": 6.75, "learning_rate": 1.1924689967983006e-05, "loss": 1.4091, "step": 18208 }, { "epoch": 45.5225, "grad_norm": 9.6875, "learning_rate": 1.1923895411554218e-05, "loss": 1.2844, "step": 18209 }, { "epoch": 45.525, "grad_norm": 6.21875, "learning_rate": 1.1923100842512452e-05, "loss": 1.3168, "step": 18210 }, { "epoch": 45.5275, "grad_norm": 7.1875, "learning_rate": 1.1922306260862918e-05, "loss": 1.4899, "step": 18211 }, { "epoch": 45.53, "grad_norm": 9.0625, "learning_rate": 1.1921511666610812e-05, "loss": 1.205, "step": 18212 }, { "epoch": 45.5325, "grad_norm": 8.9375, "learning_rate": 1.192071705976136e-05, "loss": 1.2285, "step": 18213 }, { "epoch": 45.535, "grad_norm": 7.53125, "learning_rate": 1.191992244031976e-05, "loss": 1.2734, "step": 18214 }, { "epoch": 45.5375, "grad_norm": 7.8125, "learning_rate": 1.1919127808291227e-05, "loss": 1.3348, "step": 18215 }, { "epoch": 45.54, "grad_norm": 6.1875, "learning_rate": 1.1918333163680968e-05, "loss": 1.331, "step": 18216 }, { "epoch": 45.5425, "grad_norm": 7.25, "learning_rate": 1.1917538506494196e-05, "loss": 1.4361, "step": 18217 }, { "epoch": 45.545, "grad_norm": 11.0, "learning_rate": 1.1916743836736115e-05, "loss": 1.3049, "step": 18218 }, { "epoch": 45.5475, "grad_norm": 8.5625, "learning_rate": 1.1915949154411939e-05, "loss": 1.2388, "step": 18219 }, { "epoch": 45.55, "grad_norm": 6.90625, "learning_rate": 1.1915154459526876e-05, "loss": 1.3044, "step": 18220 }, { "epoch": 45.5525, "grad_norm": 5.375, "learning_rate": 1.1914359752086143e-05, "loss": 1.4464, "step": 18221 }, { "epoch": 45.555, "grad_norm": 9.125, "learning_rate": 1.1913565032094935e-05, "loss": 1.3994, "step": 18222 }, { "epoch": 45.5575, "grad_norm": 9.4375, "learning_rate": 1.1912770299558479e-05, "loss": 1.3877, "step": 18223 }, { "epoch": 45.56, "grad_norm": 5.125, "learning_rate": 1.191197555448197e-05, "loss": 1.3555, "step": 18224 }, { "epoch": 45.5625, "grad_norm": 7.4375, "learning_rate": 1.1911180796870632e-05, "loss": 1.2316, "step": 18225 }, { "epoch": 45.565, "grad_norm": 8.0625, "learning_rate": 1.1910386026729668e-05, "loss": 1.3322, "step": 18226 }, { "epoch": 45.5675, "grad_norm": 8.25, "learning_rate": 1.190959124406429e-05, "loss": 1.2843, "step": 18227 }, { "epoch": 45.57, "grad_norm": 6.90625, "learning_rate": 1.1908796448879706e-05, "loss": 1.2148, "step": 18228 }, { "epoch": 45.5725, "grad_norm": 8.625, "learning_rate": 1.1908001641181128e-05, "loss": 1.3373, "step": 18229 }, { "epoch": 45.575, "grad_norm": 7.125, "learning_rate": 1.1907206820973768e-05, "loss": 1.1594, "step": 18230 }, { "epoch": 45.5775, "grad_norm": 6.1875, "learning_rate": 1.190641198826284e-05, "loss": 1.3044, "step": 18231 }, { "epoch": 45.58, "grad_norm": 9.875, "learning_rate": 1.190561714305355e-05, "loss": 1.3921, "step": 18232 }, { "epoch": 45.5825, "grad_norm": 7.4375, "learning_rate": 1.1904822285351107e-05, "loss": 1.5, "step": 18233 }, { "epoch": 45.585, "grad_norm": 6.34375, "learning_rate": 1.1904027415160728e-05, "loss": 1.2934, "step": 18234 }, { "epoch": 45.5875, "grad_norm": 8.3125, "learning_rate": 1.190323253248762e-05, "loss": 1.4222, "step": 18235 }, { "epoch": 45.59, "grad_norm": 6.84375, "learning_rate": 1.1902437637337e-05, "loss": 1.285, "step": 18236 }, { "epoch": 45.5925, "grad_norm": 5.03125, "learning_rate": 1.1901642729714069e-05, "loss": 1.4158, "step": 18237 }, { "epoch": 45.595, "grad_norm": 7.375, "learning_rate": 1.190084780962405e-05, "loss": 1.3716, "step": 18238 }, { "epoch": 45.5975, "grad_norm": 7.75, "learning_rate": 1.1900052877072144e-05, "loss": 1.3126, "step": 18239 }, { "epoch": 45.6, "grad_norm": 9.625, "learning_rate": 1.189925793206357e-05, "loss": 1.3025, "step": 18240 }, { "epoch": 45.6025, "grad_norm": 6.59375, "learning_rate": 1.1898462974603535e-05, "loss": 1.2708, "step": 18241 }, { "epoch": 45.605, "grad_norm": 7.34375, "learning_rate": 1.1897668004697256e-05, "loss": 1.3259, "step": 18242 }, { "epoch": 45.6075, "grad_norm": 7.53125, "learning_rate": 1.1896873022349939e-05, "loss": 1.2843, "step": 18243 }, { "epoch": 45.61, "grad_norm": 7.5, "learning_rate": 1.18960780275668e-05, "loss": 1.4102, "step": 18244 }, { "epoch": 45.6125, "grad_norm": 8.125, "learning_rate": 1.1895283020353048e-05, "loss": 1.3017, "step": 18245 }, { "epoch": 45.615, "grad_norm": 10.5625, "learning_rate": 1.1894488000713894e-05, "loss": 1.3053, "step": 18246 }, { "epoch": 45.6175, "grad_norm": 8.0, "learning_rate": 1.1893692968654558e-05, "loss": 1.3512, "step": 18247 }, { "epoch": 45.62, "grad_norm": 6.21875, "learning_rate": 1.1892897924180242e-05, "loss": 1.3445, "step": 18248 }, { "epoch": 45.6225, "grad_norm": 7.125, "learning_rate": 1.1892102867296165e-05, "loss": 1.3218, "step": 18249 }, { "epoch": 45.625, "grad_norm": 6.71875, "learning_rate": 1.1891307798007536e-05, "loss": 1.2367, "step": 18250 }, { "epoch": 45.6275, "grad_norm": 6.4375, "learning_rate": 1.1890512716319572e-05, "loss": 1.1721, "step": 18251 }, { "epoch": 45.63, "grad_norm": 7.46875, "learning_rate": 1.188971762223748e-05, "loss": 1.4402, "step": 18252 }, { "epoch": 45.6325, "grad_norm": 12.625, "learning_rate": 1.1888922515766477e-05, "loss": 1.2992, "step": 18253 }, { "epoch": 45.635, "grad_norm": 9.3125, "learning_rate": 1.1888127396911768e-05, "loss": 1.2058, "step": 18254 }, { "epoch": 45.6375, "grad_norm": 5.84375, "learning_rate": 1.1887332265678577e-05, "loss": 1.3523, "step": 18255 }, { "epoch": 45.64, "grad_norm": 7.1875, "learning_rate": 1.1886537122072106e-05, "loss": 1.332, "step": 18256 }, { "epoch": 45.6425, "grad_norm": 8.75, "learning_rate": 1.188574196609758e-05, "loss": 1.3445, "step": 18257 }, { "epoch": 45.645, "grad_norm": 10.8125, "learning_rate": 1.18849467977602e-05, "loss": 1.3512, "step": 18258 }, { "epoch": 45.6475, "grad_norm": 7.03125, "learning_rate": 1.1884151617065187e-05, "loss": 1.2327, "step": 18259 }, { "epoch": 45.65, "grad_norm": 8.3125, "learning_rate": 1.188335642401775e-05, "loss": 1.6434, "step": 18260 }, { "epoch": 45.6525, "grad_norm": 10.3125, "learning_rate": 1.1882561218623104e-05, "loss": 1.4942, "step": 18261 }, { "epoch": 45.655, "grad_norm": 7.09375, "learning_rate": 1.1881766000886461e-05, "loss": 1.2709, "step": 18262 }, { "epoch": 45.6575, "grad_norm": 14.625, "learning_rate": 1.1880970770813036e-05, "loss": 1.3764, "step": 18263 }, { "epoch": 45.66, "grad_norm": 7.25, "learning_rate": 1.188017552840804e-05, "loss": 1.3912, "step": 18264 }, { "epoch": 45.6625, "grad_norm": 7.5625, "learning_rate": 1.1879380273676691e-05, "loss": 1.4351, "step": 18265 }, { "epoch": 45.665, "grad_norm": 9.6875, "learning_rate": 1.18785850066242e-05, "loss": 1.3193, "step": 18266 }, { "epoch": 45.6675, "grad_norm": 9.1875, "learning_rate": 1.187778972725578e-05, "loss": 1.3247, "step": 18267 }, { "epoch": 45.67, "grad_norm": 6.1875, "learning_rate": 1.1876994435576648e-05, "loss": 1.1239, "step": 18268 }, { "epoch": 45.6725, "grad_norm": 6.78125, "learning_rate": 1.1876199131592012e-05, "loss": 1.2874, "step": 18269 }, { "epoch": 45.675, "grad_norm": 7.46875, "learning_rate": 1.1875403815307092e-05, "loss": 1.386, "step": 18270 }, { "epoch": 45.6775, "grad_norm": 9.0, "learning_rate": 1.1874608486727097e-05, "loss": 1.395, "step": 18271 }, { "epoch": 45.68, "grad_norm": 4.5, "learning_rate": 1.187381314585725e-05, "loss": 1.1894, "step": 18272 }, { "epoch": 45.6825, "grad_norm": 5.90625, "learning_rate": 1.1873017792702753e-05, "loss": 1.1832, "step": 18273 }, { "epoch": 45.685, "grad_norm": 9.1875, "learning_rate": 1.1872222427268828e-05, "loss": 1.4229, "step": 18274 }, { "epoch": 45.6875, "grad_norm": 8.75, "learning_rate": 1.1871427049560687e-05, "loss": 1.4238, "step": 18275 }, { "epoch": 45.69, "grad_norm": 8.625, "learning_rate": 1.1870631659583547e-05, "loss": 1.3282, "step": 18276 }, { "epoch": 45.6925, "grad_norm": 6.875, "learning_rate": 1.1869836257342619e-05, "loss": 1.2963, "step": 18277 }, { "epoch": 45.695, "grad_norm": 6.0625, "learning_rate": 1.1869040842843123e-05, "loss": 1.3235, "step": 18278 }, { "epoch": 45.6975, "grad_norm": 10.0, "learning_rate": 1.1868245416090262e-05, "loss": 1.3718, "step": 18279 }, { "epoch": 45.7, "grad_norm": 7.78125, "learning_rate": 1.1867449977089264e-05, "loss": 1.5118, "step": 18280 }, { "epoch": 45.7025, "grad_norm": 6.4375, "learning_rate": 1.1866654525845338e-05, "loss": 1.2931, "step": 18281 }, { "epoch": 45.705, "grad_norm": 6.84375, "learning_rate": 1.18658590623637e-05, "loss": 1.4433, "step": 18282 }, { "epoch": 45.7075, "grad_norm": 8.625, "learning_rate": 1.1865063586649567e-05, "loss": 1.4546, "step": 18283 }, { "epoch": 45.71, "grad_norm": 7.3125, "learning_rate": 1.1864268098708148e-05, "loss": 1.2651, "step": 18284 }, { "epoch": 45.7125, "grad_norm": 8.4375, "learning_rate": 1.1863472598544666e-05, "loss": 1.2556, "step": 18285 }, { "epoch": 45.715, "grad_norm": 7.0625, "learning_rate": 1.1862677086164328e-05, "loss": 1.2375, "step": 18286 }, { "epoch": 45.7175, "grad_norm": 7.8125, "learning_rate": 1.1861881561572355e-05, "loss": 1.482, "step": 18287 }, { "epoch": 45.72, "grad_norm": 11.375, "learning_rate": 1.1861086024773963e-05, "loss": 1.3248, "step": 18288 }, { "epoch": 45.7225, "grad_norm": 8.125, "learning_rate": 1.1860290475774363e-05, "loss": 1.2682, "step": 18289 }, { "epoch": 45.725, "grad_norm": 7.40625, "learning_rate": 1.1859494914578772e-05, "loss": 1.3457, "step": 18290 }, { "epoch": 45.7275, "grad_norm": 7.375, "learning_rate": 1.185869934119241e-05, "loss": 1.2264, "step": 18291 }, { "epoch": 45.73, "grad_norm": 8.3125, "learning_rate": 1.1857903755620488e-05, "loss": 1.3881, "step": 18292 }, { "epoch": 45.7325, "grad_norm": 7.6875, "learning_rate": 1.1857108157868224e-05, "loss": 1.4013, "step": 18293 }, { "epoch": 45.735, "grad_norm": 7.40625, "learning_rate": 1.1856312547940831e-05, "loss": 1.2809, "step": 18294 }, { "epoch": 45.7375, "grad_norm": 7.75, "learning_rate": 1.1855516925843531e-05, "loss": 1.4182, "step": 18295 }, { "epoch": 45.74, "grad_norm": 7.65625, "learning_rate": 1.185472129158153e-05, "loss": 1.3822, "step": 18296 }, { "epoch": 45.7425, "grad_norm": 8.5, "learning_rate": 1.1853925645160058e-05, "loss": 1.3642, "step": 18297 }, { "epoch": 45.745, "grad_norm": 6.40625, "learning_rate": 1.185312998658432e-05, "loss": 1.3085, "step": 18298 }, { "epoch": 45.7475, "grad_norm": 8.4375, "learning_rate": 1.1852334315859534e-05, "loss": 1.2949, "step": 18299 }, { "epoch": 45.75, "grad_norm": 11.25, "learning_rate": 1.1851538632990922e-05, "loss": 1.4745, "step": 18300 }, { "epoch": 45.7525, "grad_norm": 6.125, "learning_rate": 1.1850742937983694e-05, "loss": 1.3776, "step": 18301 }, { "epoch": 45.755, "grad_norm": 8.375, "learning_rate": 1.1849947230843073e-05, "loss": 1.3848, "step": 18302 }, { "epoch": 45.7575, "grad_norm": 10.625, "learning_rate": 1.1849151511574269e-05, "loss": 1.4052, "step": 18303 }, { "epoch": 45.76, "grad_norm": 5.5, "learning_rate": 1.1848355780182502e-05, "loss": 1.1247, "step": 18304 }, { "epoch": 45.7625, "grad_norm": 6.3125, "learning_rate": 1.184756003667299e-05, "loss": 1.3155, "step": 18305 }, { "epoch": 45.765, "grad_norm": 11.4375, "learning_rate": 1.1846764281050946e-05, "loss": 1.3507, "step": 18306 }, { "epoch": 45.7675, "grad_norm": 7.21875, "learning_rate": 1.184596851332159e-05, "loss": 1.3326, "step": 18307 }, { "epoch": 45.77, "grad_norm": 7.53125, "learning_rate": 1.184517273349014e-05, "loss": 1.2337, "step": 18308 }, { "epoch": 45.7725, "grad_norm": 9.125, "learning_rate": 1.1844376941561808e-05, "loss": 1.3974, "step": 18309 }, { "epoch": 45.775, "grad_norm": 8.25, "learning_rate": 1.1843581137541819e-05, "loss": 1.3807, "step": 18310 }, { "epoch": 45.7775, "grad_norm": 9.4375, "learning_rate": 1.1842785321435382e-05, "loss": 1.2692, "step": 18311 }, { "epoch": 45.78, "grad_norm": 5.15625, "learning_rate": 1.184198949324772e-05, "loss": 1.3177, "step": 18312 }, { "epoch": 45.7825, "grad_norm": 8.8125, "learning_rate": 1.1841193652984049e-05, "loss": 1.2455, "step": 18313 }, { "epoch": 45.785, "grad_norm": 7.65625, "learning_rate": 1.1840397800649585e-05, "loss": 1.3798, "step": 18314 }, { "epoch": 45.7875, "grad_norm": 7.125, "learning_rate": 1.1839601936249547e-05, "loss": 1.3389, "step": 18315 }, { "epoch": 45.79, "grad_norm": 6.375, "learning_rate": 1.1838806059789151e-05, "loss": 1.5147, "step": 18316 }, { "epoch": 45.7925, "grad_norm": 7.96875, "learning_rate": 1.183801017127362e-05, "loss": 1.3039, "step": 18317 }, { "epoch": 45.795, "grad_norm": 5.3125, "learning_rate": 1.1837214270708165e-05, "loss": 1.4432, "step": 18318 }, { "epoch": 45.7975, "grad_norm": 10.6875, "learning_rate": 1.1836418358098005e-05, "loss": 1.3114, "step": 18319 }, { "epoch": 45.8, "grad_norm": 6.65625, "learning_rate": 1.1835622433448361e-05, "loss": 1.3478, "step": 18320 }, { "epoch": 45.8025, "grad_norm": 6.625, "learning_rate": 1.1834826496764452e-05, "loss": 1.2315, "step": 18321 }, { "epoch": 45.805, "grad_norm": 8.0625, "learning_rate": 1.1834030548051492e-05, "loss": 1.3533, "step": 18322 }, { "epoch": 45.8075, "grad_norm": 5.46875, "learning_rate": 1.1833234587314704e-05, "loss": 1.2863, "step": 18323 }, { "epoch": 45.81, "grad_norm": 7.96875, "learning_rate": 1.1832438614559298e-05, "loss": 1.2825, "step": 18324 }, { "epoch": 45.8125, "grad_norm": 4.6875, "learning_rate": 1.1831642629790503e-05, "loss": 1.2536, "step": 18325 }, { "epoch": 45.815, "grad_norm": 7.625, "learning_rate": 1.1830846633013528e-05, "loss": 1.2863, "step": 18326 }, { "epoch": 45.8175, "grad_norm": 9.375, "learning_rate": 1.18300506242336e-05, "loss": 1.319, "step": 18327 }, { "epoch": 45.82, "grad_norm": 7.8125, "learning_rate": 1.1829254603455931e-05, "loss": 1.4536, "step": 18328 }, { "epoch": 45.8225, "grad_norm": 6.96875, "learning_rate": 1.1828458570685743e-05, "loss": 1.2762, "step": 18329 }, { "epoch": 45.825, "grad_norm": 6.40625, "learning_rate": 1.1827662525928252e-05, "loss": 1.3307, "step": 18330 }, { "epoch": 45.8275, "grad_norm": 5.21875, "learning_rate": 1.182686646918868e-05, "loss": 1.2247, "step": 18331 }, { "epoch": 45.83, "grad_norm": 7.28125, "learning_rate": 1.1826070400472242e-05, "loss": 1.2228, "step": 18332 }, { "epoch": 45.8325, "grad_norm": 8.875, "learning_rate": 1.1825274319784164e-05, "loss": 1.369, "step": 18333 }, { "epoch": 45.835, "grad_norm": 10.8125, "learning_rate": 1.1824478227129657e-05, "loss": 1.2993, "step": 18334 }, { "epoch": 45.8375, "grad_norm": 11.0625, "learning_rate": 1.1823682122513945e-05, "loss": 1.28, "step": 18335 }, { "epoch": 45.84, "grad_norm": 6.40625, "learning_rate": 1.1822886005942244e-05, "loss": 1.5197, "step": 18336 }, { "epoch": 45.8425, "grad_norm": 7.71875, "learning_rate": 1.1822089877419778e-05, "loss": 1.3032, "step": 18337 }, { "epoch": 45.845, "grad_norm": 7.5625, "learning_rate": 1.1821293736951767e-05, "loss": 1.283, "step": 18338 }, { "epoch": 45.8475, "grad_norm": 7.65625, "learning_rate": 1.182049758454342e-05, "loss": 1.4852, "step": 18339 }, { "epoch": 45.85, "grad_norm": 7.71875, "learning_rate": 1.181970142019997e-05, "loss": 1.2694, "step": 18340 }, { "epoch": 45.8525, "grad_norm": 7.5, "learning_rate": 1.1818905243926624e-05, "loss": 1.3165, "step": 18341 }, { "epoch": 45.855, "grad_norm": 9.0, "learning_rate": 1.1818109055728613e-05, "loss": 1.4978, "step": 18342 }, { "epoch": 45.8575, "grad_norm": 7.9375, "learning_rate": 1.1817312855611149e-05, "loss": 1.4019, "step": 18343 }, { "epoch": 45.86, "grad_norm": 11.0, "learning_rate": 1.1816516643579457e-05, "loss": 1.4037, "step": 18344 }, { "epoch": 45.8625, "grad_norm": 5.78125, "learning_rate": 1.1815720419638752e-05, "loss": 1.2339, "step": 18345 }, { "epoch": 45.865, "grad_norm": 9.0, "learning_rate": 1.1814924183794255e-05, "loss": 1.2395, "step": 18346 }, { "epoch": 45.8675, "grad_norm": 6.15625, "learning_rate": 1.1814127936051192e-05, "loss": 1.3118, "step": 18347 }, { "epoch": 45.87, "grad_norm": 8.0625, "learning_rate": 1.181333167641478e-05, "loss": 1.3986, "step": 18348 }, { "epoch": 45.8725, "grad_norm": 6.21875, "learning_rate": 1.1812535404890233e-05, "loss": 1.2457, "step": 18349 }, { "epoch": 45.875, "grad_norm": 6.1875, "learning_rate": 1.1811739121482777e-05, "loss": 1.4439, "step": 18350 }, { "epoch": 45.8775, "grad_norm": 6.53125, "learning_rate": 1.1810942826197633e-05, "loss": 1.3357, "step": 18351 }, { "epoch": 45.88, "grad_norm": 6.21875, "learning_rate": 1.1810146519040023e-05, "loss": 1.4312, "step": 18352 }, { "epoch": 45.8825, "grad_norm": 7.75, "learning_rate": 1.1809350200015164e-05, "loss": 1.2772, "step": 18353 }, { "epoch": 45.885, "grad_norm": 7.8125, "learning_rate": 1.1808553869128275e-05, "loss": 1.2241, "step": 18354 }, { "epoch": 45.8875, "grad_norm": 6.34375, "learning_rate": 1.180775752638458e-05, "loss": 1.2722, "step": 18355 }, { "epoch": 45.89, "grad_norm": 7.28125, "learning_rate": 1.1806961171789297e-05, "loss": 1.172, "step": 18356 }, { "epoch": 45.8925, "grad_norm": 7.0625, "learning_rate": 1.1806164805347654e-05, "loss": 1.1286, "step": 18357 }, { "epoch": 45.895, "grad_norm": 7.875, "learning_rate": 1.1805368427064862e-05, "loss": 1.3112, "step": 18358 }, { "epoch": 45.8975, "grad_norm": 8.0625, "learning_rate": 1.1804572036946151e-05, "loss": 1.2784, "step": 18359 }, { "epoch": 45.9, "grad_norm": 8.6875, "learning_rate": 1.1803775634996735e-05, "loss": 1.5422, "step": 18360 }, { "epoch": 45.9025, "grad_norm": 6.53125, "learning_rate": 1.1802979221221839e-05, "loss": 1.263, "step": 18361 }, { "epoch": 45.905, "grad_norm": 8.75, "learning_rate": 1.1802182795626685e-05, "loss": 1.4066, "step": 18362 }, { "epoch": 45.9075, "grad_norm": 8.5625, "learning_rate": 1.1801386358216488e-05, "loss": 1.2316, "step": 18363 }, { "epoch": 45.91, "grad_norm": 7.75, "learning_rate": 1.1800589908996479e-05, "loss": 1.2897, "step": 18364 }, { "epoch": 45.9125, "grad_norm": 10.875, "learning_rate": 1.1799793447971874e-05, "loss": 1.4226, "step": 18365 }, { "epoch": 45.915, "grad_norm": 11.25, "learning_rate": 1.1798996975147891e-05, "loss": 1.4661, "step": 18366 }, { "epoch": 45.9175, "grad_norm": 8.625, "learning_rate": 1.1798200490529763e-05, "loss": 1.4873, "step": 18367 }, { "epoch": 45.92, "grad_norm": 7.75, "learning_rate": 1.1797403994122698e-05, "loss": 1.3136, "step": 18368 }, { "epoch": 45.9225, "grad_norm": 8.75, "learning_rate": 1.1796607485931928e-05, "loss": 1.3289, "step": 18369 }, { "epoch": 45.925, "grad_norm": 5.15625, "learning_rate": 1.1795810965962669e-05, "loss": 1.4106, "step": 18370 }, { "epoch": 45.9275, "grad_norm": 9.0, "learning_rate": 1.1795014434220147e-05, "loss": 1.2828, "step": 18371 }, { "epoch": 45.93, "grad_norm": 6.6875, "learning_rate": 1.179421789070958e-05, "loss": 1.3829, "step": 18372 }, { "epoch": 45.9325, "grad_norm": 8.0625, "learning_rate": 1.1793421335436194e-05, "loss": 1.3927, "step": 18373 }, { "epoch": 45.935, "grad_norm": 7.4375, "learning_rate": 1.179262476840521e-05, "loss": 1.2871, "step": 18374 }, { "epoch": 45.9375, "grad_norm": 7.59375, "learning_rate": 1.1791828189621848e-05, "loss": 1.3392, "step": 18375 }, { "epoch": 45.94, "grad_norm": 10.0, "learning_rate": 1.179103159909133e-05, "loss": 1.3776, "step": 18376 }, { "epoch": 45.9425, "grad_norm": 5.53125, "learning_rate": 1.1790234996818885e-05, "loss": 1.1971, "step": 18377 }, { "epoch": 45.945, "grad_norm": 6.09375, "learning_rate": 1.178943838280973e-05, "loss": 1.2805, "step": 18378 }, { "epoch": 45.9475, "grad_norm": 6.25, "learning_rate": 1.1788641757069088e-05, "loss": 1.3287, "step": 18379 }, { "epoch": 45.95, "grad_norm": 7.78125, "learning_rate": 1.1787845119602184e-05, "loss": 1.2299, "step": 18380 }, { "epoch": 45.9525, "grad_norm": 7.4375, "learning_rate": 1.1787048470414235e-05, "loss": 1.2481, "step": 18381 }, { "epoch": 45.955, "grad_norm": 7.90625, "learning_rate": 1.1786251809510472e-05, "loss": 1.463, "step": 18382 }, { "epoch": 45.9575, "grad_norm": 9.625, "learning_rate": 1.1785455136896109e-05, "loss": 1.4252, "step": 18383 }, { "epoch": 45.96, "grad_norm": 6.9375, "learning_rate": 1.178465845257638e-05, "loss": 1.2901, "step": 18384 }, { "epoch": 45.9625, "grad_norm": 7.0, "learning_rate": 1.1783861756556495e-05, "loss": 1.4121, "step": 18385 }, { "epoch": 45.965, "grad_norm": 6.375, "learning_rate": 1.1783065048841687e-05, "loss": 1.3974, "step": 18386 }, { "epoch": 45.9675, "grad_norm": 8.4375, "learning_rate": 1.1782268329437176e-05, "loss": 1.3541, "step": 18387 }, { "epoch": 45.97, "grad_norm": 9.3125, "learning_rate": 1.1781471598348185e-05, "loss": 1.2948, "step": 18388 }, { "epoch": 45.9725, "grad_norm": 8.625, "learning_rate": 1.1780674855579936e-05, "loss": 1.3014, "step": 18389 }, { "epoch": 45.975, "grad_norm": 6.53125, "learning_rate": 1.1779878101137654e-05, "loss": 1.2622, "step": 18390 }, { "epoch": 45.9775, "grad_norm": 11.375, "learning_rate": 1.1779081335026564e-05, "loss": 1.2339, "step": 18391 }, { "epoch": 45.98, "grad_norm": 8.25, "learning_rate": 1.1778284557251887e-05, "loss": 1.3596, "step": 18392 }, { "epoch": 45.9825, "grad_norm": 8.1875, "learning_rate": 1.1777487767818848e-05, "loss": 1.1948, "step": 18393 }, { "epoch": 45.985, "grad_norm": 6.6875, "learning_rate": 1.1776690966732668e-05, "loss": 1.475, "step": 18394 }, { "epoch": 45.9875, "grad_norm": 6.3125, "learning_rate": 1.177589415399858e-05, "loss": 1.516, "step": 18395 }, { "epoch": 45.99, "grad_norm": 6.71875, "learning_rate": 1.1775097329621793e-05, "loss": 1.2865, "step": 18396 }, { "epoch": 45.9925, "grad_norm": 9.625, "learning_rate": 1.1774300493607544e-05, "loss": 1.5589, "step": 18397 }, { "epoch": 45.995, "grad_norm": 6.46875, "learning_rate": 1.1773503645961049e-05, "loss": 1.196, "step": 18398 }, { "epoch": 45.9975, "grad_norm": 8.125, "learning_rate": 1.1772706786687538e-05, "loss": 1.5073, "step": 18399 }, { "epoch": 46.0, "grad_norm": 6.40625, "learning_rate": 1.177190991579223e-05, "loss": 1.357, "step": 18400 }, { "epoch": 46.0025, "grad_norm": 9.6875, "learning_rate": 1.1771113033280352e-05, "loss": 1.2092, "step": 18401 }, { "epoch": 46.005, "grad_norm": 9.3125, "learning_rate": 1.1770316139157127e-05, "loss": 1.3461, "step": 18402 }, { "epoch": 46.0075, "grad_norm": 7.1875, "learning_rate": 1.1769519233427782e-05, "loss": 1.3002, "step": 18403 }, { "epoch": 46.01, "grad_norm": 7.875, "learning_rate": 1.1768722316097539e-05, "loss": 1.2147, "step": 18404 }, { "epoch": 46.0125, "grad_norm": 8.5625, "learning_rate": 1.1767925387171621e-05, "loss": 1.3666, "step": 18405 }, { "epoch": 46.015, "grad_norm": 17.625, "learning_rate": 1.1767128446655258e-05, "loss": 1.2933, "step": 18406 }, { "epoch": 46.0175, "grad_norm": 6.28125, "learning_rate": 1.1766331494553669e-05, "loss": 1.4681, "step": 18407 }, { "epoch": 46.02, "grad_norm": 6.4375, "learning_rate": 1.1765534530872085e-05, "loss": 1.3044, "step": 18408 }, { "epoch": 46.0225, "grad_norm": 6.0, "learning_rate": 1.1764737555615725e-05, "loss": 1.2153, "step": 18409 }, { "epoch": 46.025, "grad_norm": 8.1875, "learning_rate": 1.1763940568789817e-05, "loss": 1.3305, "step": 18410 }, { "epoch": 46.0275, "grad_norm": 8.5625, "learning_rate": 1.1763143570399581e-05, "loss": 1.5147, "step": 18411 }, { "epoch": 46.03, "grad_norm": 7.46875, "learning_rate": 1.1762346560450253e-05, "loss": 1.2293, "step": 18412 }, { "epoch": 46.0325, "grad_norm": 5.09375, "learning_rate": 1.1761549538947045e-05, "loss": 1.2927, "step": 18413 }, { "epoch": 46.035, "grad_norm": 6.625, "learning_rate": 1.1760752505895196e-05, "loss": 1.2853, "step": 18414 }, { "epoch": 46.0375, "grad_norm": 7.09375, "learning_rate": 1.175995546129992e-05, "loss": 1.295, "step": 18415 }, { "epoch": 46.04, "grad_norm": 7.53125, "learning_rate": 1.1759158405166446e-05, "loss": 1.3388, "step": 18416 }, { "epoch": 46.0425, "grad_norm": 6.71875, "learning_rate": 1.1758361337500002e-05, "loss": 1.3045, "step": 18417 }, { "epoch": 46.045, "grad_norm": 6.21875, "learning_rate": 1.175756425830581e-05, "loss": 1.2685, "step": 18418 }, { "epoch": 46.0475, "grad_norm": 5.15625, "learning_rate": 1.1756767167589096e-05, "loss": 1.313, "step": 18419 }, { "epoch": 46.05, "grad_norm": 9.25, "learning_rate": 1.1755970065355087e-05, "loss": 1.4077, "step": 18420 }, { "epoch": 46.0525, "grad_norm": 5.6875, "learning_rate": 1.1755172951609008e-05, "loss": 1.1905, "step": 18421 }, { "epoch": 46.055, "grad_norm": 8.1875, "learning_rate": 1.1754375826356087e-05, "loss": 1.3398, "step": 18422 }, { "epoch": 46.0575, "grad_norm": 7.3125, "learning_rate": 1.1753578689601547e-05, "loss": 1.5143, "step": 18423 }, { "epoch": 46.06, "grad_norm": 8.375, "learning_rate": 1.1752781541350617e-05, "loss": 1.3537, "step": 18424 }, { "epoch": 46.0625, "grad_norm": 7.59375, "learning_rate": 1.1751984381608524e-05, "loss": 1.3028, "step": 18425 }, { "epoch": 46.065, "grad_norm": 4.96875, "learning_rate": 1.1751187210380487e-05, "loss": 1.2559, "step": 18426 }, { "epoch": 46.0675, "grad_norm": 9.25, "learning_rate": 1.175039002767174e-05, "loss": 1.2779, "step": 18427 }, { "epoch": 46.07, "grad_norm": 6.40625, "learning_rate": 1.17495928334875e-05, "loss": 1.2062, "step": 18428 }, { "epoch": 46.0725, "grad_norm": 7.53125, "learning_rate": 1.1748795627833006e-05, "loss": 1.2496, "step": 18429 }, { "epoch": 46.075, "grad_norm": 8.0625, "learning_rate": 1.1747998410713475e-05, "loss": 1.3037, "step": 18430 }, { "epoch": 46.0775, "grad_norm": 8.4375, "learning_rate": 1.1747201182134136e-05, "loss": 1.4186, "step": 18431 }, { "epoch": 46.08, "grad_norm": 9.1875, "learning_rate": 1.1746403942100215e-05, "loss": 1.4583, "step": 18432 }, { "epoch": 46.0825, "grad_norm": 6.09375, "learning_rate": 1.1745606690616943e-05, "loss": 1.2944, "step": 18433 }, { "epoch": 46.085, "grad_norm": 7.5625, "learning_rate": 1.174480942768954e-05, "loss": 1.4153, "step": 18434 }, { "epoch": 46.0875, "grad_norm": 6.3125, "learning_rate": 1.1744012153323239e-05, "loss": 1.3303, "step": 18435 }, { "epoch": 46.09, "grad_norm": 8.1875, "learning_rate": 1.174321486752326e-05, "loss": 1.2505, "step": 18436 }, { "epoch": 46.0925, "grad_norm": 9.4375, "learning_rate": 1.174241757029484e-05, "loss": 1.4261, "step": 18437 }, { "epoch": 46.095, "grad_norm": 7.625, "learning_rate": 1.1741620261643195e-05, "loss": 1.4055, "step": 18438 }, { "epoch": 46.0975, "grad_norm": 7.0, "learning_rate": 1.1740822941573562e-05, "loss": 1.3875, "step": 18439 }, { "epoch": 46.1, "grad_norm": 9.8125, "learning_rate": 1.174002561009116e-05, "loss": 1.2294, "step": 18440 }, { "epoch": 46.1025, "grad_norm": 6.59375, "learning_rate": 1.173922826720122e-05, "loss": 1.138, "step": 18441 }, { "epoch": 46.105, "grad_norm": 7.40625, "learning_rate": 1.173843091290897e-05, "loss": 1.2542, "step": 18442 }, { "epoch": 46.1075, "grad_norm": 8.5625, "learning_rate": 1.1737633547219636e-05, "loss": 1.1846, "step": 18443 }, { "epoch": 46.11, "grad_norm": 7.125, "learning_rate": 1.1736836170138446e-05, "loss": 1.2929, "step": 18444 }, { "epoch": 46.1125, "grad_norm": 7.40625, "learning_rate": 1.1736038781670626e-05, "loss": 1.2502, "step": 18445 }, { "epoch": 46.115, "grad_norm": 9.6875, "learning_rate": 1.1735241381821407e-05, "loss": 1.3343, "step": 18446 }, { "epoch": 46.1175, "grad_norm": 7.9375, "learning_rate": 1.1734443970596015e-05, "loss": 1.3645, "step": 18447 }, { "epoch": 46.12, "grad_norm": 8.1875, "learning_rate": 1.1733646547999678e-05, "loss": 1.2821, "step": 18448 }, { "epoch": 46.1225, "grad_norm": 10.125, "learning_rate": 1.1732849114037622e-05, "loss": 1.3034, "step": 18449 }, { "epoch": 46.125, "grad_norm": 11.0, "learning_rate": 1.1732051668715082e-05, "loss": 1.3314, "step": 18450 }, { "epoch": 46.1275, "grad_norm": 9.3125, "learning_rate": 1.1731254212037273e-05, "loss": 1.4407, "step": 18451 }, { "epoch": 46.13, "grad_norm": 7.40625, "learning_rate": 1.1730456744009435e-05, "loss": 1.1298, "step": 18452 }, { "epoch": 46.1325, "grad_norm": 8.9375, "learning_rate": 1.1729659264636788e-05, "loss": 1.1671, "step": 18453 }, { "epoch": 46.135, "grad_norm": 7.6875, "learning_rate": 1.1728861773924568e-05, "loss": 1.282, "step": 18454 }, { "epoch": 46.1375, "grad_norm": 6.5, "learning_rate": 1.1728064271878e-05, "loss": 1.4612, "step": 18455 }, { "epoch": 46.14, "grad_norm": 11.5, "learning_rate": 1.172726675850231e-05, "loss": 1.3518, "step": 18456 }, { "epoch": 46.1425, "grad_norm": 9.625, "learning_rate": 1.1726469233802729e-05, "loss": 1.2498, "step": 18457 }, { "epoch": 46.145, "grad_norm": 9.5, "learning_rate": 1.1725671697784482e-05, "loss": 1.2126, "step": 18458 }, { "epoch": 46.1475, "grad_norm": 5.46875, "learning_rate": 1.1724874150452803e-05, "loss": 1.2045, "step": 18459 }, { "epoch": 46.15, "grad_norm": 5.5625, "learning_rate": 1.1724076591812919e-05, "loss": 1.4155, "step": 18460 }, { "epoch": 46.1525, "grad_norm": 6.09375, "learning_rate": 1.1723279021870056e-05, "loss": 1.195, "step": 18461 }, { "epoch": 46.155, "grad_norm": 6.25, "learning_rate": 1.1722481440629442e-05, "loss": 1.3406, "step": 18462 }, { "epoch": 46.1575, "grad_norm": 5.125, "learning_rate": 1.1721683848096314e-05, "loss": 1.4224, "step": 18463 }, { "epoch": 46.16, "grad_norm": 8.1875, "learning_rate": 1.1720886244275893e-05, "loss": 1.366, "step": 18464 }, { "epoch": 46.1625, "grad_norm": 8.75, "learning_rate": 1.1720088629173412e-05, "loss": 1.1142, "step": 18465 }, { "epoch": 46.165, "grad_norm": 7.78125, "learning_rate": 1.1719291002794096e-05, "loss": 1.4555, "step": 18466 }, { "epoch": 46.1675, "grad_norm": 6.75, "learning_rate": 1.1718493365143182e-05, "loss": 1.2949, "step": 18467 }, { "epoch": 46.17, "grad_norm": 6.0625, "learning_rate": 1.1717695716225887e-05, "loss": 1.3826, "step": 18468 }, { "epoch": 46.1725, "grad_norm": 7.1875, "learning_rate": 1.1716898056047454e-05, "loss": 1.3574, "step": 18469 }, { "epoch": 46.175, "grad_norm": 7.65625, "learning_rate": 1.1716100384613105e-05, "loss": 1.2604, "step": 18470 }, { "epoch": 46.1775, "grad_norm": 4.6875, "learning_rate": 1.1715302701928068e-05, "loss": 1.1898, "step": 18471 }, { "epoch": 46.18, "grad_norm": 12.25, "learning_rate": 1.1714505007997576e-05, "loss": 1.2865, "step": 18472 }, { "epoch": 46.1825, "grad_norm": 7.0625, "learning_rate": 1.1713707302826857e-05, "loss": 1.3256, "step": 18473 }, { "epoch": 46.185, "grad_norm": 6.8125, "learning_rate": 1.1712909586421144e-05, "loss": 1.2535, "step": 18474 }, { "epoch": 46.1875, "grad_norm": 6.15625, "learning_rate": 1.1712111858785663e-05, "loss": 1.297, "step": 18475 }, { "epoch": 46.19, "grad_norm": 8.3125, "learning_rate": 1.1711314119925644e-05, "loss": 1.1974, "step": 18476 }, { "epoch": 46.1925, "grad_norm": 9.75, "learning_rate": 1.171051636984632e-05, "loss": 1.3512, "step": 18477 }, { "epoch": 46.195, "grad_norm": 8.1875, "learning_rate": 1.1709718608552918e-05, "loss": 1.3789, "step": 18478 }, { "epoch": 46.1975, "grad_norm": 8.1875, "learning_rate": 1.170892083605067e-05, "loss": 1.4809, "step": 18479 }, { "epoch": 46.2, "grad_norm": 11.625, "learning_rate": 1.1708123052344803e-05, "loss": 1.4066, "step": 18480 }, { "epoch": 46.2025, "grad_norm": 10.25, "learning_rate": 1.1707325257440552e-05, "loss": 1.5049, "step": 18481 }, { "epoch": 46.205, "grad_norm": 8.75, "learning_rate": 1.1706527451343149e-05, "loss": 1.42, "step": 18482 }, { "epoch": 46.2075, "grad_norm": 8.5, "learning_rate": 1.1705729634057812e-05, "loss": 1.2779, "step": 18483 }, { "epoch": 46.21, "grad_norm": 7.53125, "learning_rate": 1.1704931805589785e-05, "loss": 1.2515, "step": 18484 }, { "epoch": 46.2125, "grad_norm": 8.0625, "learning_rate": 1.1704133965944288e-05, "loss": 1.2825, "step": 18485 }, { "epoch": 46.215, "grad_norm": 6.6875, "learning_rate": 1.1703336115126564e-05, "loss": 1.3032, "step": 18486 }, { "epoch": 46.2175, "grad_norm": 7.28125, "learning_rate": 1.1702538253141835e-05, "loss": 1.3784, "step": 18487 }, { "epoch": 46.22, "grad_norm": 7.84375, "learning_rate": 1.170174037999533e-05, "loss": 1.414, "step": 18488 }, { "epoch": 46.2225, "grad_norm": 7.40625, "learning_rate": 1.1700942495692285e-05, "loss": 1.1917, "step": 18489 }, { "epoch": 46.225, "grad_norm": 8.5, "learning_rate": 1.170014460023793e-05, "loss": 1.4526, "step": 18490 }, { "epoch": 46.2275, "grad_norm": 6.75, "learning_rate": 1.1699346693637494e-05, "loss": 1.4053, "step": 18491 }, { "epoch": 46.23, "grad_norm": 7.5, "learning_rate": 1.1698548775896208e-05, "loss": 1.4712, "step": 18492 }, { "epoch": 46.2325, "grad_norm": 8.75, "learning_rate": 1.1697750847019305e-05, "loss": 1.4005, "step": 18493 }, { "epoch": 46.235, "grad_norm": 11.4375, "learning_rate": 1.1696952907012016e-05, "loss": 1.5657, "step": 18494 }, { "epoch": 46.2375, "grad_norm": 6.90625, "learning_rate": 1.169615495587957e-05, "loss": 1.4743, "step": 18495 }, { "epoch": 46.24, "grad_norm": 6.5625, "learning_rate": 1.1695356993627203e-05, "loss": 1.3168, "step": 18496 }, { "epoch": 46.2425, "grad_norm": 6.78125, "learning_rate": 1.1694559020260143e-05, "loss": 1.4108, "step": 18497 }, { "epoch": 46.245, "grad_norm": 7.65625, "learning_rate": 1.169376103578362e-05, "loss": 1.396, "step": 18498 }, { "epoch": 46.2475, "grad_norm": 5.28125, "learning_rate": 1.1692963040202869e-05, "loss": 1.439, "step": 18499 }, { "epoch": 46.25, "grad_norm": 13.5, "learning_rate": 1.1692165033523117e-05, "loss": 1.4498, "step": 18500 }, { "epoch": 46.2525, "grad_norm": 5.78125, "learning_rate": 1.1691367015749601e-05, "loss": 1.5492, "step": 18501 }, { "epoch": 46.255, "grad_norm": 8.125, "learning_rate": 1.1690568986887552e-05, "loss": 1.3517, "step": 18502 }, { "epoch": 46.2575, "grad_norm": 6.875, "learning_rate": 1.1689770946942199e-05, "loss": 1.3554, "step": 18503 }, { "epoch": 46.26, "grad_norm": 7.0625, "learning_rate": 1.1688972895918776e-05, "loss": 1.551, "step": 18504 }, { "epoch": 46.2625, "grad_norm": 11.0625, "learning_rate": 1.1688174833822513e-05, "loss": 1.2869, "step": 18505 }, { "epoch": 46.265, "grad_norm": 7.59375, "learning_rate": 1.1687376760658641e-05, "loss": 1.161, "step": 18506 }, { "epoch": 46.2675, "grad_norm": 7.34375, "learning_rate": 1.1686578676432401e-05, "loss": 1.319, "step": 18507 }, { "epoch": 46.27, "grad_norm": 6.375, "learning_rate": 1.1685780581149011e-05, "loss": 1.3052, "step": 18508 }, { "epoch": 46.2725, "grad_norm": 9.3125, "learning_rate": 1.1684982474813717e-05, "loss": 1.4439, "step": 18509 }, { "epoch": 46.275, "grad_norm": 13.0, "learning_rate": 1.1684184357431741e-05, "loss": 1.3437, "step": 18510 }, { "epoch": 46.2775, "grad_norm": 14.0, "learning_rate": 1.1683386229008324e-05, "loss": 1.3052, "step": 18511 }, { "epoch": 46.28, "grad_norm": 8.8125, "learning_rate": 1.1682588089548692e-05, "loss": 1.3929, "step": 18512 }, { "epoch": 46.2825, "grad_norm": 7.28125, "learning_rate": 1.1681789939058082e-05, "loss": 1.1195, "step": 18513 }, { "epoch": 46.285, "grad_norm": 8.375, "learning_rate": 1.1680991777541721e-05, "loss": 1.2833, "step": 18514 }, { "epoch": 46.2875, "grad_norm": 6.9375, "learning_rate": 1.1680193605004844e-05, "loss": 1.4608, "step": 18515 }, { "epoch": 46.29, "grad_norm": 7.53125, "learning_rate": 1.1679395421452688e-05, "loss": 1.2558, "step": 18516 }, { "epoch": 46.2925, "grad_norm": 7.1875, "learning_rate": 1.1678597226890484e-05, "loss": 1.4041, "step": 18517 }, { "epoch": 46.295, "grad_norm": 7.8125, "learning_rate": 1.1677799021323462e-05, "loss": 1.3939, "step": 18518 }, { "epoch": 46.2975, "grad_norm": 11.5625, "learning_rate": 1.1677000804756857e-05, "loss": 1.4551, "step": 18519 }, { "epoch": 46.3, "grad_norm": 7.5, "learning_rate": 1.1676202577195901e-05, "loss": 1.3393, "step": 18520 }, { "epoch": 46.3025, "grad_norm": 7.0625, "learning_rate": 1.1675404338645828e-05, "loss": 1.3576, "step": 18521 }, { "epoch": 46.305, "grad_norm": 7.25, "learning_rate": 1.1674606089111876e-05, "loss": 1.3669, "step": 18522 }, { "epoch": 46.3075, "grad_norm": 7.0, "learning_rate": 1.1673807828599266e-05, "loss": 1.4011, "step": 18523 }, { "epoch": 46.31, "grad_norm": 6.6875, "learning_rate": 1.1673009557113244e-05, "loss": 1.2601, "step": 18524 }, { "epoch": 46.3125, "grad_norm": 6.65625, "learning_rate": 1.1672211274659034e-05, "loss": 1.3227, "step": 18525 }, { "epoch": 46.315, "grad_norm": 9.625, "learning_rate": 1.1671412981241878e-05, "loss": 1.3649, "step": 18526 }, { "epoch": 46.3175, "grad_norm": 6.875, "learning_rate": 1.1670614676867004e-05, "loss": 1.3521, "step": 18527 }, { "epoch": 46.32, "grad_norm": 6.46875, "learning_rate": 1.1669816361539647e-05, "loss": 1.3102, "step": 18528 }, { "epoch": 46.3225, "grad_norm": 9.25, "learning_rate": 1.166901803526504e-05, "loss": 1.2787, "step": 18529 }, { "epoch": 46.325, "grad_norm": 7.78125, "learning_rate": 1.166821969804842e-05, "loss": 1.4318, "step": 18530 }, { "epoch": 46.3275, "grad_norm": 6.59375, "learning_rate": 1.1667421349895016e-05, "loss": 1.324, "step": 18531 }, { "epoch": 46.33, "grad_norm": 8.375, "learning_rate": 1.1666622990810064e-05, "loss": 1.4655, "step": 18532 }, { "epoch": 46.3325, "grad_norm": 11.125, "learning_rate": 1.1665824620798799e-05, "loss": 1.4847, "step": 18533 }, { "epoch": 46.335, "grad_norm": 6.0625, "learning_rate": 1.1665026239866455e-05, "loss": 1.437, "step": 18534 }, { "epoch": 46.3375, "grad_norm": 8.4375, "learning_rate": 1.1664227848018263e-05, "loss": 1.3978, "step": 18535 }, { "epoch": 46.34, "grad_norm": 7.875, "learning_rate": 1.1663429445259464e-05, "loss": 1.409, "step": 18536 }, { "epoch": 46.3425, "grad_norm": 5.6875, "learning_rate": 1.1662631031595288e-05, "loss": 1.3122, "step": 18537 }, { "epoch": 46.345, "grad_norm": 8.875, "learning_rate": 1.1661832607030964e-05, "loss": 1.2411, "step": 18538 }, { "epoch": 46.3475, "grad_norm": 6.71875, "learning_rate": 1.166103417157174e-05, "loss": 1.4154, "step": 18539 }, { "epoch": 46.35, "grad_norm": 12.25, "learning_rate": 1.1660235725222835e-05, "loss": 1.3578, "step": 18540 }, { "epoch": 46.3525, "grad_norm": 9.625, "learning_rate": 1.1659437267989495e-05, "loss": 1.4211, "step": 18541 }, { "epoch": 46.355, "grad_norm": 6.34375, "learning_rate": 1.165863879987695e-05, "loss": 1.3409, "step": 18542 }, { "epoch": 46.3575, "grad_norm": 8.0625, "learning_rate": 1.1657840320890434e-05, "loss": 1.3238, "step": 18543 }, { "epoch": 46.36, "grad_norm": 5.46875, "learning_rate": 1.1657041831035186e-05, "loss": 1.3326, "step": 18544 }, { "epoch": 46.3625, "grad_norm": 5.40625, "learning_rate": 1.1656243330316437e-05, "loss": 1.3724, "step": 18545 }, { "epoch": 46.365, "grad_norm": 5.75, "learning_rate": 1.1655444818739422e-05, "loss": 1.4281, "step": 18546 }, { "epoch": 46.3675, "grad_norm": 8.625, "learning_rate": 1.1654646296309378e-05, "loss": 1.2251, "step": 18547 }, { "epoch": 46.37, "grad_norm": 6.71875, "learning_rate": 1.1653847763031538e-05, "loss": 1.3295, "step": 18548 }, { "epoch": 46.3725, "grad_norm": 9.75, "learning_rate": 1.165304921891114e-05, "loss": 1.4259, "step": 18549 }, { "epoch": 46.375, "grad_norm": 7.28125, "learning_rate": 1.1652250663953415e-05, "loss": 1.2794, "step": 18550 }, { "epoch": 46.3775, "grad_norm": 7.78125, "learning_rate": 1.1651452098163601e-05, "loss": 1.2602, "step": 18551 }, { "epoch": 46.38, "grad_norm": 8.25, "learning_rate": 1.1650653521546937e-05, "loss": 1.236, "step": 18552 }, { "epoch": 46.3825, "grad_norm": 7.75, "learning_rate": 1.1649854934108648e-05, "loss": 1.5031, "step": 18553 }, { "epoch": 46.385, "grad_norm": 10.6875, "learning_rate": 1.1649056335853983e-05, "loss": 1.4347, "step": 18554 }, { "epoch": 46.3875, "grad_norm": 9.0, "learning_rate": 1.1648257726788166e-05, "loss": 1.3944, "step": 18555 }, { "epoch": 46.39, "grad_norm": 6.40625, "learning_rate": 1.1647459106916438e-05, "loss": 1.4106, "step": 18556 }, { "epoch": 46.3925, "grad_norm": 7.84375, "learning_rate": 1.1646660476244036e-05, "loss": 1.3939, "step": 18557 }, { "epoch": 46.395, "grad_norm": 6.5625, "learning_rate": 1.1645861834776194e-05, "loss": 1.4196, "step": 18558 }, { "epoch": 46.3975, "grad_norm": 6.84375, "learning_rate": 1.1645063182518146e-05, "loss": 1.5087, "step": 18559 }, { "epoch": 46.4, "grad_norm": 6.96875, "learning_rate": 1.164426451947513e-05, "loss": 1.2221, "step": 18560 }, { "epoch": 46.4025, "grad_norm": 6.4375, "learning_rate": 1.1643465845652379e-05, "loss": 1.38, "step": 18561 }, { "epoch": 46.405, "grad_norm": 6.28125, "learning_rate": 1.1642667161055137e-05, "loss": 1.2338, "step": 18562 }, { "epoch": 46.4075, "grad_norm": 6.6875, "learning_rate": 1.164186846568863e-05, "loss": 1.376, "step": 18563 }, { "epoch": 46.41, "grad_norm": 6.59375, "learning_rate": 1.1641069759558101e-05, "loss": 1.202, "step": 18564 }, { "epoch": 46.4125, "grad_norm": 6.96875, "learning_rate": 1.1640271042668785e-05, "loss": 1.3389, "step": 18565 }, { "epoch": 46.415, "grad_norm": 6.96875, "learning_rate": 1.1639472315025916e-05, "loss": 1.2224, "step": 18566 }, { "epoch": 46.4175, "grad_norm": 7.65625, "learning_rate": 1.1638673576634735e-05, "loss": 1.4056, "step": 18567 }, { "epoch": 46.42, "grad_norm": 9.375, "learning_rate": 1.1637874827500472e-05, "loss": 1.4133, "step": 18568 }, { "epoch": 46.4225, "grad_norm": 6.6875, "learning_rate": 1.1637076067628373e-05, "loss": 1.3495, "step": 18569 }, { "epoch": 46.425, "grad_norm": 9.375, "learning_rate": 1.1636277297023661e-05, "loss": 1.2043, "step": 18570 }, { "epoch": 46.4275, "grad_norm": 7.75, "learning_rate": 1.1635478515691587e-05, "loss": 1.3362, "step": 18571 }, { "epoch": 46.43, "grad_norm": 8.0625, "learning_rate": 1.1634679723637379e-05, "loss": 1.4619, "step": 18572 }, { "epoch": 46.4325, "grad_norm": 7.0, "learning_rate": 1.1633880920866274e-05, "loss": 1.3015, "step": 18573 }, { "epoch": 46.435, "grad_norm": 9.5, "learning_rate": 1.1633082107383513e-05, "loss": 1.318, "step": 18574 }, { "epoch": 46.4375, "grad_norm": 6.0625, "learning_rate": 1.1632283283194331e-05, "loss": 1.2349, "step": 18575 }, { "epoch": 46.44, "grad_norm": 6.90625, "learning_rate": 1.1631484448303964e-05, "loss": 1.3138, "step": 18576 }, { "epoch": 46.4425, "grad_norm": 8.375, "learning_rate": 1.1630685602717655e-05, "loss": 1.2029, "step": 18577 }, { "epoch": 46.445, "grad_norm": 7.78125, "learning_rate": 1.1629886746440628e-05, "loss": 1.5119, "step": 18578 }, { "epoch": 46.4475, "grad_norm": 9.625, "learning_rate": 1.1629087879478136e-05, "loss": 1.2366, "step": 18579 }, { "epoch": 46.45, "grad_norm": 9.375, "learning_rate": 1.1628289001835405e-05, "loss": 1.469, "step": 18580 }, { "epoch": 46.4525, "grad_norm": 7.09375, "learning_rate": 1.1627490113517681e-05, "loss": 1.3797, "step": 18581 }, { "epoch": 46.455, "grad_norm": 7.96875, "learning_rate": 1.1626691214530193e-05, "loss": 1.2898, "step": 18582 }, { "epoch": 46.4575, "grad_norm": 6.125, "learning_rate": 1.1625892304878184e-05, "loss": 1.2256, "step": 18583 }, { "epoch": 46.46, "grad_norm": 9.0, "learning_rate": 1.1625093384566887e-05, "loss": 1.4382, "step": 18584 }, { "epoch": 46.4625, "grad_norm": 7.71875, "learning_rate": 1.1624294453601546e-05, "loss": 1.5258, "step": 18585 }, { "epoch": 46.465, "grad_norm": 8.0625, "learning_rate": 1.1623495511987396e-05, "loss": 1.3826, "step": 18586 }, { "epoch": 46.4675, "grad_norm": 18.25, "learning_rate": 1.1622696559729674e-05, "loss": 1.3224, "step": 18587 }, { "epoch": 46.47, "grad_norm": 6.25, "learning_rate": 1.1621897596833616e-05, "loss": 1.2151, "step": 18588 }, { "epoch": 46.4725, "grad_norm": 7.3125, "learning_rate": 1.1621098623304464e-05, "loss": 1.4098, "step": 18589 }, { "epoch": 46.475, "grad_norm": 7.59375, "learning_rate": 1.1620299639147454e-05, "loss": 1.1975, "step": 18590 }, { "epoch": 46.4775, "grad_norm": 7.0, "learning_rate": 1.1619500644367825e-05, "loss": 1.3836, "step": 18591 }, { "epoch": 46.48, "grad_norm": 6.96875, "learning_rate": 1.1618701638970815e-05, "loss": 1.3752, "step": 18592 }, { "epoch": 46.4825, "grad_norm": 8.3125, "learning_rate": 1.1617902622961658e-05, "loss": 1.3871, "step": 18593 }, { "epoch": 46.485, "grad_norm": 7.4375, "learning_rate": 1.1617103596345602e-05, "loss": 1.3693, "step": 18594 }, { "epoch": 46.4875, "grad_norm": 6.4375, "learning_rate": 1.1616304559127873e-05, "loss": 1.2177, "step": 18595 }, { "epoch": 46.49, "grad_norm": 8.25, "learning_rate": 1.1615505511313723e-05, "loss": 1.255, "step": 18596 }, { "epoch": 46.4925, "grad_norm": 7.65625, "learning_rate": 1.161470645290838e-05, "loss": 1.2683, "step": 18597 }, { "epoch": 46.495, "grad_norm": 6.90625, "learning_rate": 1.1613907383917087e-05, "loss": 1.1471, "step": 18598 }, { "epoch": 46.4975, "grad_norm": 8.625, "learning_rate": 1.161310830434508e-05, "loss": 1.5051, "step": 18599 }, { "epoch": 46.5, "grad_norm": 8.8125, "learning_rate": 1.1612309214197599e-05, "loss": 1.3684, "step": 18600 }, { "epoch": 46.5025, "grad_norm": 7.84375, "learning_rate": 1.1611510113479887e-05, "loss": 1.3251, "step": 18601 }, { "epoch": 46.505, "grad_norm": 9.6875, "learning_rate": 1.1610711002197177e-05, "loss": 1.2199, "step": 18602 }, { "epoch": 46.5075, "grad_norm": 6.34375, "learning_rate": 1.1609911880354709e-05, "loss": 1.3208, "step": 18603 }, { "epoch": 46.51, "grad_norm": 8.0, "learning_rate": 1.1609112747957724e-05, "loss": 1.496, "step": 18604 }, { "epoch": 46.5125, "grad_norm": 5.78125, "learning_rate": 1.160831360501146e-05, "loss": 1.3092, "step": 18605 }, { "epoch": 46.515, "grad_norm": 7.75, "learning_rate": 1.1607514451521155e-05, "loss": 1.433, "step": 18606 }, { "epoch": 46.5175, "grad_norm": 5.8125, "learning_rate": 1.1606715287492054e-05, "loss": 1.1866, "step": 18607 }, { "epoch": 46.52, "grad_norm": 9.625, "learning_rate": 1.1605916112929388e-05, "loss": 1.4337, "step": 18608 }, { "epoch": 46.5225, "grad_norm": 6.53125, "learning_rate": 1.1605116927838405e-05, "loss": 1.3541, "step": 18609 }, { "epoch": 46.525, "grad_norm": 6.03125, "learning_rate": 1.1604317732224333e-05, "loss": 1.1918, "step": 18610 }, { "epoch": 46.5275, "grad_norm": 9.3125, "learning_rate": 1.1603518526092424e-05, "loss": 1.3871, "step": 18611 }, { "epoch": 46.53, "grad_norm": 8.6875, "learning_rate": 1.1602719309447905e-05, "loss": 1.1914, "step": 18612 }, { "epoch": 46.5325, "grad_norm": 7.8125, "learning_rate": 1.160192008229603e-05, "loss": 1.3351, "step": 18613 }, { "epoch": 46.535, "grad_norm": 7.375, "learning_rate": 1.1601120844642027e-05, "loss": 1.4659, "step": 18614 }, { "epoch": 46.5375, "grad_norm": 5.90625, "learning_rate": 1.1600321596491139e-05, "loss": 1.4091, "step": 18615 }, { "epoch": 46.54, "grad_norm": 7.15625, "learning_rate": 1.159952233784861e-05, "loss": 1.3318, "step": 18616 }, { "epoch": 46.5425, "grad_norm": 8.3125, "learning_rate": 1.1598723068719672e-05, "loss": 1.3505, "step": 18617 }, { "epoch": 46.545, "grad_norm": 7.6875, "learning_rate": 1.1597923789109572e-05, "loss": 1.2309, "step": 18618 }, { "epoch": 46.5475, "grad_norm": 13.4375, "learning_rate": 1.1597124499023549e-05, "loss": 1.322, "step": 18619 }, { "epoch": 46.55, "grad_norm": 6.65625, "learning_rate": 1.1596325198466841e-05, "loss": 1.2869, "step": 18620 }, { "epoch": 46.5525, "grad_norm": 5.21875, "learning_rate": 1.1595525887444687e-05, "loss": 1.3325, "step": 18621 }, { "epoch": 46.555, "grad_norm": 8.5, "learning_rate": 1.1594726565962332e-05, "loss": 1.4413, "step": 18622 }, { "epoch": 46.5575, "grad_norm": 6.71875, "learning_rate": 1.1593927234025008e-05, "loss": 1.1841, "step": 18623 }, { "epoch": 46.56, "grad_norm": 6.40625, "learning_rate": 1.1593127891637968e-05, "loss": 1.4099, "step": 18624 }, { "epoch": 46.5625, "grad_norm": 6.65625, "learning_rate": 1.1592328538806439e-05, "loss": 1.271, "step": 18625 }, { "epoch": 46.565, "grad_norm": 11.125, "learning_rate": 1.1591529175535674e-05, "loss": 1.3595, "step": 18626 }, { "epoch": 46.5675, "grad_norm": 5.90625, "learning_rate": 1.1590729801830902e-05, "loss": 1.4803, "step": 18627 }, { "epoch": 46.57, "grad_norm": 10.0, "learning_rate": 1.1589930417697372e-05, "loss": 1.3282, "step": 18628 }, { "epoch": 46.5725, "grad_norm": 9.8125, "learning_rate": 1.158913102314032e-05, "loss": 1.3149, "step": 18629 }, { "epoch": 46.575, "grad_norm": 8.625, "learning_rate": 1.1588331618164992e-05, "loss": 1.4679, "step": 18630 }, { "epoch": 46.5775, "grad_norm": 7.53125, "learning_rate": 1.1587532202776621e-05, "loss": 1.3726, "step": 18631 }, { "epoch": 46.58, "grad_norm": 6.96875, "learning_rate": 1.1586732776980456e-05, "loss": 1.3436, "step": 18632 }, { "epoch": 46.5825, "grad_norm": 6.75, "learning_rate": 1.1585933340781732e-05, "loss": 1.4174, "step": 18633 }, { "epoch": 46.585, "grad_norm": 11.0625, "learning_rate": 1.1585133894185693e-05, "loss": 1.4997, "step": 18634 }, { "epoch": 46.5875, "grad_norm": 8.125, "learning_rate": 1.158433443719758e-05, "loss": 1.2495, "step": 18635 }, { "epoch": 46.59, "grad_norm": 6.9375, "learning_rate": 1.1583534969822634e-05, "loss": 1.232, "step": 18636 }, { "epoch": 46.5925, "grad_norm": 5.53125, "learning_rate": 1.1582735492066095e-05, "loss": 1.2493, "step": 18637 }, { "epoch": 46.595, "grad_norm": 7.9375, "learning_rate": 1.1581936003933208e-05, "loss": 1.4275, "step": 18638 }, { "epoch": 46.5975, "grad_norm": 5.25, "learning_rate": 1.1581136505429212e-05, "loss": 1.3667, "step": 18639 }, { "epoch": 46.6, "grad_norm": 6.625, "learning_rate": 1.1580336996559343e-05, "loss": 1.4259, "step": 18640 }, { "epoch": 46.6025, "grad_norm": 7.4375, "learning_rate": 1.1579537477328856e-05, "loss": 1.3139, "step": 18641 }, { "epoch": 46.605, "grad_norm": 4.8125, "learning_rate": 1.1578737947742975e-05, "loss": 1.442, "step": 18642 }, { "epoch": 46.6075, "grad_norm": 8.3125, "learning_rate": 1.1577938407806959e-05, "loss": 1.3971, "step": 18643 }, { "epoch": 46.61, "grad_norm": 8.125, "learning_rate": 1.157713885752604e-05, "loss": 1.3507, "step": 18644 }, { "epoch": 46.6125, "grad_norm": 8.8125, "learning_rate": 1.1576339296905461e-05, "loss": 1.2818, "step": 18645 }, { "epoch": 46.615, "grad_norm": 9.1875, "learning_rate": 1.1575539725950465e-05, "loss": 1.4038, "step": 18646 }, { "epoch": 46.6175, "grad_norm": 6.9375, "learning_rate": 1.1574740144666292e-05, "loss": 1.3157, "step": 18647 }, { "epoch": 46.62, "grad_norm": 9.0625, "learning_rate": 1.1573940553058186e-05, "loss": 1.295, "step": 18648 }, { "epoch": 46.6225, "grad_norm": 5.1875, "learning_rate": 1.1573140951131391e-05, "loss": 1.2787, "step": 18649 }, { "epoch": 46.625, "grad_norm": 5.8125, "learning_rate": 1.1572341338891145e-05, "loss": 1.3704, "step": 18650 }, { "epoch": 46.6275, "grad_norm": 6.5, "learning_rate": 1.1571541716342694e-05, "loss": 1.3556, "step": 18651 }, { "epoch": 46.63, "grad_norm": 11.75, "learning_rate": 1.1570742083491273e-05, "loss": 1.2929, "step": 18652 }, { "epoch": 46.6325, "grad_norm": 5.71875, "learning_rate": 1.1569942440342137e-05, "loss": 1.2547, "step": 18653 }, { "epoch": 46.635, "grad_norm": 5.9375, "learning_rate": 1.1569142786900519e-05, "loss": 1.2088, "step": 18654 }, { "epoch": 46.6375, "grad_norm": 6.90625, "learning_rate": 1.1568343123171659e-05, "loss": 1.3593, "step": 18655 }, { "epoch": 46.64, "grad_norm": 6.59375, "learning_rate": 1.156754344916081e-05, "loss": 1.289, "step": 18656 }, { "epoch": 46.6425, "grad_norm": 7.90625, "learning_rate": 1.1566743764873203e-05, "loss": 1.3975, "step": 18657 }, { "epoch": 46.645, "grad_norm": 6.03125, "learning_rate": 1.1565944070314093e-05, "loss": 1.3354, "step": 18658 }, { "epoch": 46.6475, "grad_norm": 8.375, "learning_rate": 1.1565144365488714e-05, "loss": 1.2046, "step": 18659 }, { "epoch": 46.65, "grad_norm": 9.0, "learning_rate": 1.156434465040231e-05, "loss": 1.2366, "step": 18660 }, { "epoch": 46.6525, "grad_norm": 8.875, "learning_rate": 1.1563544925060126e-05, "loss": 1.2313, "step": 18661 }, { "epoch": 46.655, "grad_norm": 7.6875, "learning_rate": 1.1562745189467404e-05, "loss": 1.3326, "step": 18662 }, { "epoch": 46.6575, "grad_norm": 12.9375, "learning_rate": 1.1561945443629386e-05, "loss": 1.1818, "step": 18663 }, { "epoch": 46.66, "grad_norm": 7.84375, "learning_rate": 1.156114568755132e-05, "loss": 1.3246, "step": 18664 }, { "epoch": 46.6625, "grad_norm": 8.1875, "learning_rate": 1.156034592123844e-05, "loss": 1.2713, "step": 18665 }, { "epoch": 46.665, "grad_norm": 9.0, "learning_rate": 1.1559546144695998e-05, "loss": 1.3025, "step": 18666 }, { "epoch": 46.6675, "grad_norm": 6.15625, "learning_rate": 1.1558746357929231e-05, "loss": 1.2826, "step": 18667 }, { "epoch": 46.67, "grad_norm": 7.3125, "learning_rate": 1.1557946560943389e-05, "loss": 1.16, "step": 18668 }, { "epoch": 46.6725, "grad_norm": 6.75, "learning_rate": 1.155714675374371e-05, "loss": 1.5208, "step": 18669 }, { "epoch": 46.675, "grad_norm": 7.125, "learning_rate": 1.1556346936335438e-05, "loss": 1.3207, "step": 18670 }, { "epoch": 46.6775, "grad_norm": 5.96875, "learning_rate": 1.155554710872382e-05, "loss": 1.3143, "step": 18671 }, { "epoch": 46.68, "grad_norm": 6.6875, "learning_rate": 1.1554747270914098e-05, "loss": 1.2699, "step": 18672 }, { "epoch": 46.6825, "grad_norm": 7.625, "learning_rate": 1.155394742291151e-05, "loss": 1.2806, "step": 18673 }, { "epoch": 46.685, "grad_norm": 6.8125, "learning_rate": 1.155314756472131e-05, "loss": 1.3923, "step": 18674 }, { "epoch": 46.6875, "grad_norm": 7.21875, "learning_rate": 1.1552347696348733e-05, "loss": 1.4762, "step": 18675 }, { "epoch": 46.69, "grad_norm": 8.125, "learning_rate": 1.155154781779903e-05, "loss": 1.2272, "step": 18676 }, { "epoch": 46.6925, "grad_norm": 6.1875, "learning_rate": 1.1550747929077439e-05, "loss": 1.4199, "step": 18677 }, { "epoch": 46.695, "grad_norm": 5.78125, "learning_rate": 1.1549948030189209e-05, "loss": 1.2561, "step": 18678 }, { "epoch": 46.6975, "grad_norm": 8.375, "learning_rate": 1.1549148121139581e-05, "loss": 1.2382, "step": 18679 }, { "epoch": 46.7, "grad_norm": 8.25, "learning_rate": 1.1548348201933799e-05, "loss": 1.3175, "step": 18680 }, { "epoch": 46.7025, "grad_norm": 6.71875, "learning_rate": 1.1547548272577109e-05, "loss": 1.2997, "step": 18681 }, { "epoch": 46.705, "grad_norm": 5.1875, "learning_rate": 1.1546748333074753e-05, "loss": 1.4149, "step": 18682 }, { "epoch": 46.7075, "grad_norm": 7.8125, "learning_rate": 1.154594838343198e-05, "loss": 1.5046, "step": 18683 }, { "epoch": 46.71, "grad_norm": 9.0625, "learning_rate": 1.1545148423654029e-05, "loss": 1.305, "step": 18684 }, { "epoch": 46.7125, "grad_norm": 7.40625, "learning_rate": 1.1544348453746146e-05, "loss": 1.3155, "step": 18685 }, { "epoch": 46.715, "grad_norm": 6.96875, "learning_rate": 1.1543548473713577e-05, "loss": 1.266, "step": 18686 }, { "epoch": 46.7175, "grad_norm": 7.5, "learning_rate": 1.1542748483561565e-05, "loss": 1.3632, "step": 18687 }, { "epoch": 46.72, "grad_norm": 7.28125, "learning_rate": 1.1541948483295358e-05, "loss": 1.278, "step": 18688 }, { "epoch": 46.7225, "grad_norm": 7.65625, "learning_rate": 1.1541148472920199e-05, "loss": 1.3277, "step": 18689 }, { "epoch": 46.725, "grad_norm": 9.4375, "learning_rate": 1.1540348452441327e-05, "loss": 1.3124, "step": 18690 }, { "epoch": 46.7275, "grad_norm": 8.0625, "learning_rate": 1.1539548421863997e-05, "loss": 1.3049, "step": 18691 }, { "epoch": 46.73, "grad_norm": 9.875, "learning_rate": 1.1538748381193447e-05, "loss": 1.3395, "step": 18692 }, { "epoch": 46.7325, "grad_norm": 6.9375, "learning_rate": 1.1537948330434927e-05, "loss": 1.2588, "step": 18693 }, { "epoch": 46.735, "grad_norm": 7.09375, "learning_rate": 1.1537148269593678e-05, "loss": 1.255, "step": 18694 }, { "epoch": 46.7375, "grad_norm": 6.1875, "learning_rate": 1.1536348198674945e-05, "loss": 1.2266, "step": 18695 }, { "epoch": 46.74, "grad_norm": 8.75, "learning_rate": 1.1535548117683977e-05, "loss": 1.2342, "step": 18696 }, { "epoch": 46.7425, "grad_norm": 9.5, "learning_rate": 1.1534748026626014e-05, "loss": 1.4438, "step": 18697 }, { "epoch": 46.745, "grad_norm": 10.0625, "learning_rate": 1.1533947925506308e-05, "loss": 1.3566, "step": 18698 }, { "epoch": 46.7475, "grad_norm": 9.6875, "learning_rate": 1.15331478143301e-05, "loss": 1.4887, "step": 18699 }, { "epoch": 46.75, "grad_norm": 9.0, "learning_rate": 1.1532347693102632e-05, "loss": 1.2554, "step": 18700 }, { "epoch": 46.7525, "grad_norm": 6.84375, "learning_rate": 1.1531547561829158e-05, "loss": 1.5099, "step": 18701 }, { "epoch": 46.755, "grad_norm": 10.0, "learning_rate": 1.1530747420514919e-05, "loss": 1.343, "step": 18702 }, { "epoch": 46.7575, "grad_norm": 7.625, "learning_rate": 1.1529947269165161e-05, "loss": 1.3291, "step": 18703 }, { "epoch": 46.76, "grad_norm": 9.875, "learning_rate": 1.1529147107785129e-05, "loss": 1.337, "step": 18704 }, { "epoch": 46.7625, "grad_norm": 8.3125, "learning_rate": 1.1528346936380068e-05, "loss": 1.3776, "step": 18705 }, { "epoch": 46.765, "grad_norm": 7.8125, "learning_rate": 1.1527546754955228e-05, "loss": 1.3079, "step": 18706 }, { "epoch": 46.7675, "grad_norm": 9.3125, "learning_rate": 1.1526746563515853e-05, "loss": 1.2713, "step": 18707 }, { "epoch": 46.77, "grad_norm": 7.65625, "learning_rate": 1.1525946362067188e-05, "loss": 1.3623, "step": 18708 }, { "epoch": 46.7725, "grad_norm": 11.0, "learning_rate": 1.152514615061448e-05, "loss": 1.394, "step": 18709 }, { "epoch": 46.775, "grad_norm": 10.375, "learning_rate": 1.1524345929162975e-05, "loss": 1.4949, "step": 18710 }, { "epoch": 46.7775, "grad_norm": 9.875, "learning_rate": 1.1523545697717919e-05, "loss": 1.2643, "step": 18711 }, { "epoch": 46.78, "grad_norm": 4.78125, "learning_rate": 1.1522745456284557e-05, "loss": 1.2641, "step": 18712 }, { "epoch": 46.7825, "grad_norm": 8.1875, "learning_rate": 1.1521945204868139e-05, "loss": 1.3669, "step": 18713 }, { "epoch": 46.785, "grad_norm": 6.84375, "learning_rate": 1.1521144943473907e-05, "loss": 1.3835, "step": 18714 }, { "epoch": 46.7875, "grad_norm": 7.1875, "learning_rate": 1.152034467210711e-05, "loss": 1.2946, "step": 18715 }, { "epoch": 46.79, "grad_norm": 6.71875, "learning_rate": 1.1519544390772996e-05, "loss": 1.5014, "step": 18716 }, { "epoch": 46.7925, "grad_norm": 9.375, "learning_rate": 1.1518744099476809e-05, "loss": 1.2465, "step": 18717 }, { "epoch": 46.795, "grad_norm": 6.53125, "learning_rate": 1.1517943798223796e-05, "loss": 1.421, "step": 18718 }, { "epoch": 46.7975, "grad_norm": 12.625, "learning_rate": 1.1517143487019207e-05, "loss": 1.3492, "step": 18719 }, { "epoch": 46.8, "grad_norm": 7.34375, "learning_rate": 1.151634316586828e-05, "loss": 1.3998, "step": 18720 }, { "epoch": 46.8025, "grad_norm": 7.09375, "learning_rate": 1.1515542834776272e-05, "loss": 1.4883, "step": 18721 }, { "epoch": 46.805, "grad_norm": 8.25, "learning_rate": 1.1514742493748424e-05, "loss": 1.3248, "step": 18722 }, { "epoch": 46.8075, "grad_norm": 6.4375, "learning_rate": 1.1513942142789985e-05, "loss": 1.3044, "step": 18723 }, { "epoch": 46.81, "grad_norm": 8.0, "learning_rate": 1.1513141781906202e-05, "loss": 1.4701, "step": 18724 }, { "epoch": 46.8125, "grad_norm": 7.25, "learning_rate": 1.1512341411102324e-05, "loss": 1.3049, "step": 18725 }, { "epoch": 46.815, "grad_norm": 7.53125, "learning_rate": 1.1511541030383597e-05, "loss": 1.2349, "step": 18726 }, { "epoch": 46.8175, "grad_norm": 4.71875, "learning_rate": 1.1510740639755263e-05, "loss": 1.1991, "step": 18727 }, { "epoch": 46.82, "grad_norm": 5.875, "learning_rate": 1.150994023922258e-05, "loss": 1.1548, "step": 18728 }, { "epoch": 46.8225, "grad_norm": 8.125, "learning_rate": 1.1509139828790781e-05, "loss": 1.3294, "step": 18729 }, { "epoch": 46.825, "grad_norm": 6.15625, "learning_rate": 1.1508339408465128e-05, "loss": 1.2247, "step": 18730 }, { "epoch": 46.8275, "grad_norm": 7.3125, "learning_rate": 1.150753897825086e-05, "loss": 1.3479, "step": 18731 }, { "epoch": 46.83, "grad_norm": 6.15625, "learning_rate": 1.1506738538153228e-05, "loss": 1.2973, "step": 18732 }, { "epoch": 46.8325, "grad_norm": 9.5, "learning_rate": 1.1505938088177477e-05, "loss": 1.402, "step": 18733 }, { "epoch": 46.835, "grad_norm": 6.34375, "learning_rate": 1.1505137628328858e-05, "loss": 1.1783, "step": 18734 }, { "epoch": 46.8375, "grad_norm": 7.5, "learning_rate": 1.1504337158612615e-05, "loss": 1.4441, "step": 18735 }, { "epoch": 46.84, "grad_norm": 5.53125, "learning_rate": 1.1503536679034e-05, "loss": 1.318, "step": 18736 }, { "epoch": 46.8425, "grad_norm": 6.46875, "learning_rate": 1.1502736189598253e-05, "loss": 1.2515, "step": 18737 }, { "epoch": 46.845, "grad_norm": 7.40625, "learning_rate": 1.1501935690310635e-05, "loss": 1.3185, "step": 18738 }, { "epoch": 46.8475, "grad_norm": 8.0, "learning_rate": 1.1501135181176381e-05, "loss": 1.3387, "step": 18739 }, { "epoch": 46.85, "grad_norm": 8.25, "learning_rate": 1.150033466220075e-05, "loss": 1.332, "step": 18740 }, { "epoch": 46.8525, "grad_norm": 9.6875, "learning_rate": 1.1499534133388984e-05, "loss": 1.3493, "step": 18741 }, { "epoch": 46.855, "grad_norm": 4.90625, "learning_rate": 1.1498733594746328e-05, "loss": 1.2217, "step": 18742 }, { "epoch": 46.8575, "grad_norm": 8.1875, "learning_rate": 1.1497933046278039e-05, "loss": 1.3957, "step": 18743 }, { "epoch": 46.86, "grad_norm": 7.75, "learning_rate": 1.1497132487989357e-05, "loss": 1.4153, "step": 18744 }, { "epoch": 46.8625, "grad_norm": 8.6875, "learning_rate": 1.1496331919885538e-05, "loss": 1.3508, "step": 18745 }, { "epoch": 46.865, "grad_norm": 9.625, "learning_rate": 1.1495531341971824e-05, "loss": 1.3107, "step": 18746 }, { "epoch": 46.8675, "grad_norm": 9.625, "learning_rate": 1.1494730754253467e-05, "loss": 1.4498, "step": 18747 }, { "epoch": 46.87, "grad_norm": 8.4375, "learning_rate": 1.1493930156735716e-05, "loss": 1.2691, "step": 18748 }, { "epoch": 46.8725, "grad_norm": 10.125, "learning_rate": 1.1493129549423818e-05, "loss": 1.4132, "step": 18749 }, { "epoch": 46.875, "grad_norm": 7.53125, "learning_rate": 1.1492328932323022e-05, "loss": 1.3762, "step": 18750 }, { "epoch": 46.8775, "grad_norm": 7.5625, "learning_rate": 1.149152830543858e-05, "loss": 1.4809, "step": 18751 }, { "epoch": 46.88, "grad_norm": 8.5, "learning_rate": 1.1490727668775735e-05, "loss": 1.2575, "step": 18752 }, { "epoch": 46.8825, "grad_norm": 8.375, "learning_rate": 1.1489927022339739e-05, "loss": 1.3691, "step": 18753 }, { "epoch": 46.885, "grad_norm": 10.375, "learning_rate": 1.148912636613584e-05, "loss": 1.4621, "step": 18754 }, { "epoch": 46.8875, "grad_norm": 7.03125, "learning_rate": 1.1488325700169294e-05, "loss": 1.231, "step": 18755 }, { "epoch": 46.89, "grad_norm": 8.8125, "learning_rate": 1.148752502444534e-05, "loss": 1.3996, "step": 18756 }, { "epoch": 46.8925, "grad_norm": 8.9375, "learning_rate": 1.1486724338969233e-05, "loss": 1.3813, "step": 18757 }, { "epoch": 46.895, "grad_norm": 9.75, "learning_rate": 1.148592364374622e-05, "loss": 1.5412, "step": 18758 }, { "epoch": 46.8975, "grad_norm": 6.375, "learning_rate": 1.148512293878155e-05, "loss": 1.4447, "step": 18759 }, { "epoch": 46.9, "grad_norm": 5.9375, "learning_rate": 1.1484322224080474e-05, "loss": 1.4778, "step": 18760 }, { "epoch": 46.9025, "grad_norm": 5.59375, "learning_rate": 1.148352149964824e-05, "loss": 1.4196, "step": 18761 }, { "epoch": 46.905, "grad_norm": 6.34375, "learning_rate": 1.14827207654901e-05, "loss": 1.1583, "step": 18762 }, { "epoch": 46.9075, "grad_norm": 5.875, "learning_rate": 1.1481920021611301e-05, "loss": 1.4116, "step": 18763 }, { "epoch": 46.91, "grad_norm": 7.40625, "learning_rate": 1.1481119268017094e-05, "loss": 1.2408, "step": 18764 }, { "epoch": 46.9125, "grad_norm": 5.90625, "learning_rate": 1.1480318504712728e-05, "loss": 1.278, "step": 18765 }, { "epoch": 46.915, "grad_norm": 9.25, "learning_rate": 1.1479517731703456e-05, "loss": 1.4164, "step": 18766 }, { "epoch": 46.9175, "grad_norm": 8.375, "learning_rate": 1.1478716948994519e-05, "loss": 1.4491, "step": 18767 }, { "epoch": 46.92, "grad_norm": 7.34375, "learning_rate": 1.147791615659118e-05, "loss": 1.284, "step": 18768 }, { "epoch": 46.9225, "grad_norm": 10.5, "learning_rate": 1.1477115354498676e-05, "loss": 1.4381, "step": 18769 }, { "epoch": 46.925, "grad_norm": 7.6875, "learning_rate": 1.1476314542722267e-05, "loss": 1.3742, "step": 18770 }, { "epoch": 46.9275, "grad_norm": 9.125, "learning_rate": 1.1475513721267197e-05, "loss": 1.2305, "step": 18771 }, { "epoch": 46.93, "grad_norm": 6.65625, "learning_rate": 1.1474712890138716e-05, "loss": 1.2769, "step": 18772 }, { "epoch": 46.9325, "grad_norm": 6.5, "learning_rate": 1.147391204934208e-05, "loss": 1.2967, "step": 18773 }, { "epoch": 46.935, "grad_norm": 7.1875, "learning_rate": 1.1473111198882533e-05, "loss": 1.2772, "step": 18774 }, { "epoch": 46.9375, "grad_norm": 10.25, "learning_rate": 1.147231033876533e-05, "loss": 1.3998, "step": 18775 }, { "epoch": 46.94, "grad_norm": 7.6875, "learning_rate": 1.1471509468995719e-05, "loss": 1.3875, "step": 18776 }, { "epoch": 46.9425, "grad_norm": 6.5625, "learning_rate": 1.1470708589578949e-05, "loss": 1.5214, "step": 18777 }, { "epoch": 46.945, "grad_norm": 7.6875, "learning_rate": 1.1469907700520274e-05, "loss": 1.4401, "step": 18778 }, { "epoch": 46.9475, "grad_norm": 9.1875, "learning_rate": 1.146910680182494e-05, "loss": 1.333, "step": 18779 }, { "epoch": 46.95, "grad_norm": 7.1875, "learning_rate": 1.1468305893498204e-05, "loss": 1.3218, "step": 18780 }, { "epoch": 46.9525, "grad_norm": 5.9375, "learning_rate": 1.1467504975545314e-05, "loss": 1.2786, "step": 18781 }, { "epoch": 46.955, "grad_norm": 6.09375, "learning_rate": 1.1466704047971517e-05, "loss": 1.3208, "step": 18782 }, { "epoch": 46.9575, "grad_norm": 7.5, "learning_rate": 1.146590311078207e-05, "loss": 1.3491, "step": 18783 }, { "epoch": 46.96, "grad_norm": 6.625, "learning_rate": 1.1465102163982218e-05, "loss": 1.4104, "step": 18784 }, { "epoch": 46.9625, "grad_norm": 7.84375, "learning_rate": 1.1464301207577217e-05, "loss": 1.2607, "step": 18785 }, { "epoch": 46.965, "grad_norm": 11.0, "learning_rate": 1.1463500241572314e-05, "loss": 1.3437, "step": 18786 }, { "epoch": 46.9675, "grad_norm": 5.71875, "learning_rate": 1.1462699265972764e-05, "loss": 1.2806, "step": 18787 }, { "epoch": 46.97, "grad_norm": 12.125, "learning_rate": 1.1461898280783815e-05, "loss": 1.4727, "step": 18788 }, { "epoch": 46.9725, "grad_norm": 8.3125, "learning_rate": 1.1461097286010718e-05, "loss": 1.2958, "step": 18789 }, { "epoch": 46.975, "grad_norm": 7.5625, "learning_rate": 1.1460296281658728e-05, "loss": 1.2712, "step": 18790 }, { "epoch": 46.9775, "grad_norm": 7.53125, "learning_rate": 1.1459495267733093e-05, "loss": 1.303, "step": 18791 }, { "epoch": 46.98, "grad_norm": 7.875, "learning_rate": 1.1458694244239067e-05, "loss": 1.2225, "step": 18792 }, { "epoch": 46.9825, "grad_norm": 9.4375, "learning_rate": 1.1457893211181897e-05, "loss": 1.2905, "step": 18793 }, { "epoch": 46.985, "grad_norm": 9.0, "learning_rate": 1.145709216856684e-05, "loss": 1.3218, "step": 18794 }, { "epoch": 46.9875, "grad_norm": 21.375, "learning_rate": 1.1456291116399143e-05, "loss": 1.399, "step": 18795 }, { "epoch": 46.99, "grad_norm": 8.25, "learning_rate": 1.1455490054684063e-05, "loss": 1.3976, "step": 18796 }, { "epoch": 46.9925, "grad_norm": 7.6875, "learning_rate": 1.1454688983426842e-05, "loss": 1.3197, "step": 18797 }, { "epoch": 46.995, "grad_norm": 5.03125, "learning_rate": 1.1453887902632746e-05, "loss": 1.4049, "step": 18798 }, { "epoch": 46.9975, "grad_norm": 8.3125, "learning_rate": 1.1453086812307013e-05, "loss": 1.4331, "step": 18799 }, { "epoch": 47.0, "grad_norm": 9.0, "learning_rate": 1.1452285712454905e-05, "loss": 1.4718, "step": 18800 }, { "epoch": 47.0025, "grad_norm": 7.25, "learning_rate": 1.1451484603081668e-05, "loss": 1.3877, "step": 18801 }, { "epoch": 47.005, "grad_norm": 9.125, "learning_rate": 1.1450683484192558e-05, "loss": 1.2456, "step": 18802 }, { "epoch": 47.0075, "grad_norm": 8.25, "learning_rate": 1.1449882355792821e-05, "loss": 1.5345, "step": 18803 }, { "epoch": 47.01, "grad_norm": 7.0, "learning_rate": 1.1449081217887716e-05, "loss": 1.3509, "step": 18804 }, { "epoch": 47.0125, "grad_norm": 6.59375, "learning_rate": 1.1448280070482493e-05, "loss": 1.3788, "step": 18805 }, { "epoch": 47.015, "grad_norm": 8.0625, "learning_rate": 1.1447478913582406e-05, "loss": 1.3468, "step": 18806 }, { "epoch": 47.0175, "grad_norm": 8.375, "learning_rate": 1.1446677747192698e-05, "loss": 1.444, "step": 18807 }, { "epoch": 47.02, "grad_norm": 10.5, "learning_rate": 1.1445876571318637e-05, "loss": 1.3111, "step": 18808 }, { "epoch": 47.0225, "grad_norm": 10.4375, "learning_rate": 1.144507538596546e-05, "loss": 1.3281, "step": 18809 }, { "epoch": 47.025, "grad_norm": 6.78125, "learning_rate": 1.1444274191138431e-05, "loss": 1.299, "step": 18810 }, { "epoch": 47.0275, "grad_norm": 8.1875, "learning_rate": 1.14434729868428e-05, "loss": 1.2724, "step": 18811 }, { "epoch": 47.03, "grad_norm": 6.65625, "learning_rate": 1.144267177308381e-05, "loss": 1.3971, "step": 18812 }, { "epoch": 47.0325, "grad_norm": 7.59375, "learning_rate": 1.144187054986673e-05, "loss": 1.5595, "step": 18813 }, { "epoch": 47.035, "grad_norm": 7.21875, "learning_rate": 1.14410693171968e-05, "loss": 1.2863, "step": 18814 }, { "epoch": 47.0375, "grad_norm": 8.125, "learning_rate": 1.1440268075079278e-05, "loss": 1.2248, "step": 18815 }, { "epoch": 47.04, "grad_norm": 4.84375, "learning_rate": 1.1439466823519414e-05, "loss": 1.3373, "step": 18816 }, { "epoch": 47.0425, "grad_norm": 8.9375, "learning_rate": 1.1438665562522467e-05, "loss": 1.373, "step": 18817 }, { "epoch": 47.045, "grad_norm": 7.53125, "learning_rate": 1.1437864292093683e-05, "loss": 1.3994, "step": 18818 }, { "epoch": 47.0475, "grad_norm": 9.1875, "learning_rate": 1.1437063012238318e-05, "loss": 1.4739, "step": 18819 }, { "epoch": 47.05, "grad_norm": 8.9375, "learning_rate": 1.1436261722961627e-05, "loss": 1.3951, "step": 18820 }, { "epoch": 47.0525, "grad_norm": 9.0, "learning_rate": 1.1435460424268866e-05, "loss": 1.2555, "step": 18821 }, { "epoch": 47.055, "grad_norm": 13.125, "learning_rate": 1.1434659116165277e-05, "loss": 1.3235, "step": 18822 }, { "epoch": 47.0575, "grad_norm": 7.0, "learning_rate": 1.1433857798656122e-05, "loss": 1.3492, "step": 18823 }, { "epoch": 47.06, "grad_norm": 7.28125, "learning_rate": 1.1433056471746654e-05, "loss": 1.3251, "step": 18824 }, { "epoch": 47.0625, "grad_norm": 7.4375, "learning_rate": 1.1432255135442126e-05, "loss": 1.2673, "step": 18825 }, { "epoch": 47.065, "grad_norm": 9.0625, "learning_rate": 1.143145378974779e-05, "loss": 1.397, "step": 18826 }, { "epoch": 47.0675, "grad_norm": 8.3125, "learning_rate": 1.1430652434668898e-05, "loss": 1.2285, "step": 18827 }, { "epoch": 47.07, "grad_norm": 7.96875, "learning_rate": 1.1429851070210706e-05, "loss": 1.4133, "step": 18828 }, { "epoch": 47.0725, "grad_norm": 6.5625, "learning_rate": 1.1429049696378469e-05, "loss": 1.267, "step": 18829 }, { "epoch": 47.075, "grad_norm": 9.0, "learning_rate": 1.1428248313177439e-05, "loss": 1.476, "step": 18830 }, { "epoch": 47.0775, "grad_norm": 6.90625, "learning_rate": 1.142744692061287e-05, "loss": 1.4971, "step": 18831 }, { "epoch": 47.08, "grad_norm": 6.25, "learning_rate": 1.1426645518690015e-05, "loss": 1.4602, "step": 18832 }, { "epoch": 47.0825, "grad_norm": 6.9375, "learning_rate": 1.1425844107414132e-05, "loss": 1.2732, "step": 18833 }, { "epoch": 47.085, "grad_norm": 10.625, "learning_rate": 1.1425042686790471e-05, "loss": 1.1967, "step": 18834 }, { "epoch": 47.0875, "grad_norm": 7.03125, "learning_rate": 1.1424241256824286e-05, "loss": 1.1235, "step": 18835 }, { "epoch": 47.09, "grad_norm": 5.9375, "learning_rate": 1.1423439817520837e-05, "loss": 1.1855, "step": 18836 }, { "epoch": 47.0925, "grad_norm": 7.40625, "learning_rate": 1.1422638368885366e-05, "loss": 1.3039, "step": 18837 }, { "epoch": 47.095, "grad_norm": 9.125, "learning_rate": 1.1421836910923141e-05, "loss": 1.4206, "step": 18838 }, { "epoch": 47.0975, "grad_norm": 6.25, "learning_rate": 1.1421035443639405e-05, "loss": 1.2039, "step": 18839 }, { "epoch": 47.1, "grad_norm": 6.28125, "learning_rate": 1.1420233967039423e-05, "loss": 1.3946, "step": 18840 }, { "epoch": 47.1025, "grad_norm": 7.46875, "learning_rate": 1.1419432481128441e-05, "loss": 1.4356, "step": 18841 }, { "epoch": 47.105, "grad_norm": 6.34375, "learning_rate": 1.1418630985911717e-05, "loss": 1.1293, "step": 18842 }, { "epoch": 47.1075, "grad_norm": 6.375, "learning_rate": 1.1417829481394503e-05, "loss": 1.2325, "step": 18843 }, { "epoch": 47.11, "grad_norm": 9.0625, "learning_rate": 1.1417027967582058e-05, "loss": 1.3313, "step": 18844 }, { "epoch": 47.1125, "grad_norm": 6.78125, "learning_rate": 1.1416226444479631e-05, "loss": 1.2954, "step": 18845 }, { "epoch": 47.115, "grad_norm": 8.75, "learning_rate": 1.1415424912092484e-05, "loss": 1.1624, "step": 18846 }, { "epoch": 47.1175, "grad_norm": 7.0, "learning_rate": 1.1414623370425865e-05, "loss": 1.226, "step": 18847 }, { "epoch": 47.12, "grad_norm": 7.4375, "learning_rate": 1.1413821819485035e-05, "loss": 1.5026, "step": 18848 }, { "epoch": 47.1225, "grad_norm": 5.96875, "learning_rate": 1.1413020259275246e-05, "loss": 1.2954, "step": 18849 }, { "epoch": 47.125, "grad_norm": 6.0625, "learning_rate": 1.1412218689801748e-05, "loss": 1.3394, "step": 18850 }, { "epoch": 47.1275, "grad_norm": 6.46875, "learning_rate": 1.1411417111069804e-05, "loss": 1.3123, "step": 18851 }, { "epoch": 47.13, "grad_norm": 6.90625, "learning_rate": 1.1410615523084664e-05, "loss": 1.4599, "step": 18852 }, { "epoch": 47.1325, "grad_norm": 9.5625, "learning_rate": 1.140981392585159e-05, "loss": 1.4829, "step": 18853 }, { "epoch": 47.135, "grad_norm": 6.4375, "learning_rate": 1.1409012319375828e-05, "loss": 1.4915, "step": 18854 }, { "epoch": 47.1375, "grad_norm": 8.625, "learning_rate": 1.1408210703662639e-05, "loss": 1.3691, "step": 18855 }, { "epoch": 47.14, "grad_norm": 8.5, "learning_rate": 1.1407409078717275e-05, "loss": 1.3069, "step": 18856 }, { "epoch": 47.1425, "grad_norm": 6.28125, "learning_rate": 1.1406607444544998e-05, "loss": 1.1995, "step": 18857 }, { "epoch": 47.145, "grad_norm": 6.09375, "learning_rate": 1.1405805801151053e-05, "loss": 1.2876, "step": 18858 }, { "epoch": 47.1475, "grad_norm": 11.8125, "learning_rate": 1.1405004148540706e-05, "loss": 1.331, "step": 18859 }, { "epoch": 47.15, "grad_norm": 10.0625, "learning_rate": 1.1404202486719205e-05, "loss": 1.4288, "step": 18860 }, { "epoch": 47.1525, "grad_norm": 7.5, "learning_rate": 1.140340081569181e-05, "loss": 1.355, "step": 18861 }, { "epoch": 47.155, "grad_norm": 21.0, "learning_rate": 1.1402599135463776e-05, "loss": 1.235, "step": 18862 }, { "epoch": 47.1575, "grad_norm": 7.6875, "learning_rate": 1.1401797446040357e-05, "loss": 1.3537, "step": 18863 }, { "epoch": 47.16, "grad_norm": 10.1875, "learning_rate": 1.140099574742681e-05, "loss": 1.5175, "step": 18864 }, { "epoch": 47.1625, "grad_norm": 12.6875, "learning_rate": 1.1400194039628392e-05, "loss": 1.2088, "step": 18865 }, { "epoch": 47.165, "grad_norm": 6.25, "learning_rate": 1.1399392322650358e-05, "loss": 1.2101, "step": 18866 }, { "epoch": 47.1675, "grad_norm": 5.71875, "learning_rate": 1.1398590596497963e-05, "loss": 1.3926, "step": 18867 }, { "epoch": 47.17, "grad_norm": 9.0625, "learning_rate": 1.1397788861176466e-05, "loss": 1.4635, "step": 18868 }, { "epoch": 47.1725, "grad_norm": 6.09375, "learning_rate": 1.1396987116691117e-05, "loss": 1.3243, "step": 18869 }, { "epoch": 47.175, "grad_norm": 6.46875, "learning_rate": 1.1396185363047181e-05, "loss": 1.2194, "step": 18870 }, { "epoch": 47.1775, "grad_norm": 10.5, "learning_rate": 1.1395383600249904e-05, "loss": 1.3939, "step": 18871 }, { "epoch": 47.18, "grad_norm": 6.4375, "learning_rate": 1.1394581828304555e-05, "loss": 1.4117, "step": 18872 }, { "epoch": 47.1825, "grad_norm": 8.125, "learning_rate": 1.139378004721638e-05, "loss": 1.2006, "step": 18873 }, { "epoch": 47.185, "grad_norm": 7.0, "learning_rate": 1.1392978256990639e-05, "loss": 1.2664, "step": 18874 }, { "epoch": 47.1875, "grad_norm": 8.6875, "learning_rate": 1.1392176457632586e-05, "loss": 1.3054, "step": 18875 }, { "epoch": 47.19, "grad_norm": 7.1875, "learning_rate": 1.1391374649147482e-05, "loss": 1.2352, "step": 18876 }, { "epoch": 47.1925, "grad_norm": 6.90625, "learning_rate": 1.1390572831540583e-05, "loss": 1.2421, "step": 18877 }, { "epoch": 47.195, "grad_norm": 8.5625, "learning_rate": 1.1389771004817142e-05, "loss": 1.4121, "step": 18878 }, { "epoch": 47.1975, "grad_norm": 5.8125, "learning_rate": 1.1388969168982418e-05, "loss": 1.4161, "step": 18879 }, { "epoch": 47.2, "grad_norm": 8.6875, "learning_rate": 1.138816732404167e-05, "loss": 1.4517, "step": 18880 }, { "epoch": 47.2025, "grad_norm": 10.0, "learning_rate": 1.1387365470000147e-05, "loss": 1.2972, "step": 18881 }, { "epoch": 47.205, "grad_norm": 8.875, "learning_rate": 1.1386563606863117e-05, "loss": 1.2488, "step": 18882 }, { "epoch": 47.2075, "grad_norm": 8.125, "learning_rate": 1.138576173463583e-05, "loss": 1.2155, "step": 18883 }, { "epoch": 47.21, "grad_norm": 9.375, "learning_rate": 1.1384959853323546e-05, "loss": 1.1487, "step": 18884 }, { "epoch": 47.2125, "grad_norm": 7.59375, "learning_rate": 1.1384157962931518e-05, "loss": 1.2883, "step": 18885 }, { "epoch": 47.215, "grad_norm": 6.40625, "learning_rate": 1.1383356063465006e-05, "loss": 1.2994, "step": 18886 }, { "epoch": 47.2175, "grad_norm": 7.21875, "learning_rate": 1.1382554154929266e-05, "loss": 1.3273, "step": 18887 }, { "epoch": 47.22, "grad_norm": 6.5, "learning_rate": 1.1381752237329557e-05, "loss": 1.483, "step": 18888 }, { "epoch": 47.2225, "grad_norm": 7.8125, "learning_rate": 1.1380950310671138e-05, "loss": 1.2273, "step": 18889 }, { "epoch": 47.225, "grad_norm": 8.875, "learning_rate": 1.138014837495926e-05, "loss": 1.2478, "step": 18890 }, { "epoch": 47.2275, "grad_norm": 4.375, "learning_rate": 1.137934643019919e-05, "loss": 1.3506, "step": 18891 }, { "epoch": 47.23, "grad_norm": 9.25, "learning_rate": 1.1378544476396176e-05, "loss": 1.3519, "step": 18892 }, { "epoch": 47.2325, "grad_norm": 8.9375, "learning_rate": 1.1377742513555482e-05, "loss": 1.3671, "step": 18893 }, { "epoch": 47.235, "grad_norm": 4.9375, "learning_rate": 1.137694054168236e-05, "loss": 1.4494, "step": 18894 }, { "epoch": 47.2375, "grad_norm": 6.40625, "learning_rate": 1.1376138560782074e-05, "loss": 1.2376, "step": 18895 }, { "epoch": 47.24, "grad_norm": 6.8125, "learning_rate": 1.1375336570859877e-05, "loss": 1.3324, "step": 18896 }, { "epoch": 47.2425, "grad_norm": 8.1875, "learning_rate": 1.137453457192103e-05, "loss": 1.253, "step": 18897 }, { "epoch": 47.245, "grad_norm": 8.25, "learning_rate": 1.1373732563970788e-05, "loss": 1.2274, "step": 18898 }, { "epoch": 47.2475, "grad_norm": 6.5, "learning_rate": 1.1372930547014409e-05, "loss": 1.2804, "step": 18899 }, { "epoch": 47.25, "grad_norm": 8.375, "learning_rate": 1.1372128521057155e-05, "loss": 1.3792, "step": 18900 }, { "epoch": 47.2525, "grad_norm": 8.1875, "learning_rate": 1.1371326486104279e-05, "loss": 1.2536, "step": 18901 }, { "epoch": 47.255, "grad_norm": 6.5, "learning_rate": 1.1370524442161042e-05, "loss": 1.3365, "step": 18902 }, { "epoch": 47.2575, "grad_norm": 7.25, "learning_rate": 1.1369722389232702e-05, "loss": 1.3402, "step": 18903 }, { "epoch": 47.26, "grad_norm": 8.3125, "learning_rate": 1.1368920327324517e-05, "loss": 1.2715, "step": 18904 }, { "epoch": 47.2625, "grad_norm": 7.9375, "learning_rate": 1.1368118256441745e-05, "loss": 1.3895, "step": 18905 }, { "epoch": 47.265, "grad_norm": 7.375, "learning_rate": 1.1367316176589645e-05, "loss": 1.2838, "step": 18906 }, { "epoch": 47.2675, "grad_norm": 5.65625, "learning_rate": 1.1366514087773474e-05, "loss": 1.237, "step": 18907 }, { "epoch": 47.27, "grad_norm": 7.46875, "learning_rate": 1.1365711989998494e-05, "loss": 1.3149, "step": 18908 }, { "epoch": 47.2725, "grad_norm": 6.8125, "learning_rate": 1.1364909883269957e-05, "loss": 1.3131, "step": 18909 }, { "epoch": 47.275, "grad_norm": 7.625, "learning_rate": 1.1364107767593126e-05, "loss": 1.2668, "step": 18910 }, { "epoch": 47.2775, "grad_norm": 7.46875, "learning_rate": 1.136330564297326e-05, "loss": 1.2991, "step": 18911 }, { "epoch": 47.28, "grad_norm": 6.59375, "learning_rate": 1.136250350941562e-05, "loss": 1.3053, "step": 18912 }, { "epoch": 47.2825, "grad_norm": 6.96875, "learning_rate": 1.1361701366925456e-05, "loss": 1.1695, "step": 18913 }, { "epoch": 47.285, "grad_norm": 10.0, "learning_rate": 1.1360899215508035e-05, "loss": 1.3224, "step": 18914 }, { "epoch": 47.2875, "grad_norm": 7.65625, "learning_rate": 1.1360097055168613e-05, "loss": 1.2558, "step": 18915 }, { "epoch": 47.29, "grad_norm": 7.09375, "learning_rate": 1.1359294885912449e-05, "loss": 1.4212, "step": 18916 }, { "epoch": 47.2925, "grad_norm": 8.25, "learning_rate": 1.13584927077448e-05, "loss": 1.4411, "step": 18917 }, { "epoch": 47.295, "grad_norm": 5.75, "learning_rate": 1.135769052067093e-05, "loss": 1.3455, "step": 18918 }, { "epoch": 47.2975, "grad_norm": 8.4375, "learning_rate": 1.1356888324696096e-05, "loss": 1.4448, "step": 18919 }, { "epoch": 47.3, "grad_norm": 7.40625, "learning_rate": 1.1356086119825553e-05, "loss": 1.373, "step": 18920 }, { "epoch": 47.3025, "grad_norm": 8.0625, "learning_rate": 1.1355283906064567e-05, "loss": 1.3224, "step": 18921 }, { "epoch": 47.305, "grad_norm": 9.25, "learning_rate": 1.1354481683418392e-05, "loss": 1.2072, "step": 18922 }, { "epoch": 47.3075, "grad_norm": 11.375, "learning_rate": 1.1353679451892293e-05, "loss": 1.266, "step": 18923 }, { "epoch": 47.31, "grad_norm": 7.6875, "learning_rate": 1.135287721149152e-05, "loss": 1.3378, "step": 18924 }, { "epoch": 47.3125, "grad_norm": 7.5625, "learning_rate": 1.1352074962221342e-05, "loss": 1.3049, "step": 18925 }, { "epoch": 47.315, "grad_norm": 6.65625, "learning_rate": 1.135127270408701e-05, "loss": 1.4113, "step": 18926 }, { "epoch": 47.3175, "grad_norm": 5.8125, "learning_rate": 1.1350470437093794e-05, "loss": 1.5743, "step": 18927 }, { "epoch": 47.32, "grad_norm": 8.0625, "learning_rate": 1.1349668161246945e-05, "loss": 1.2079, "step": 18928 }, { "epoch": 47.3225, "grad_norm": 9.4375, "learning_rate": 1.1348865876551727e-05, "loss": 1.4543, "step": 18929 }, { "epoch": 47.325, "grad_norm": 5.875, "learning_rate": 1.1348063583013397e-05, "loss": 1.4209, "step": 18930 }, { "epoch": 47.3275, "grad_norm": 11.625, "learning_rate": 1.1347261280637217e-05, "loss": 1.3809, "step": 18931 }, { "epoch": 47.33, "grad_norm": 9.1875, "learning_rate": 1.1346458969428445e-05, "loss": 1.298, "step": 18932 }, { "epoch": 47.3325, "grad_norm": 6.375, "learning_rate": 1.1345656649392343e-05, "loss": 1.3404, "step": 18933 }, { "epoch": 47.335, "grad_norm": 6.875, "learning_rate": 1.1344854320534167e-05, "loss": 1.3389, "step": 18934 }, { "epoch": 47.3375, "grad_norm": 7.40625, "learning_rate": 1.1344051982859181e-05, "loss": 1.3223, "step": 18935 }, { "epoch": 47.34, "grad_norm": 7.8125, "learning_rate": 1.1343249636372646e-05, "loss": 1.287, "step": 18936 }, { "epoch": 47.3425, "grad_norm": 6.59375, "learning_rate": 1.1342447281079818e-05, "loss": 1.2606, "step": 18937 }, { "epoch": 47.345, "grad_norm": 8.4375, "learning_rate": 1.1341644916985963e-05, "loss": 1.3225, "step": 18938 }, { "epoch": 47.3475, "grad_norm": 6.625, "learning_rate": 1.1340842544096333e-05, "loss": 1.3417, "step": 18939 }, { "epoch": 47.35, "grad_norm": 5.6875, "learning_rate": 1.1340040162416197e-05, "loss": 1.2398, "step": 18940 }, { "epoch": 47.3525, "grad_norm": 6.59375, "learning_rate": 1.1339237771950806e-05, "loss": 1.3783, "step": 18941 }, { "epoch": 47.355, "grad_norm": 6.1875, "learning_rate": 1.1338435372705432e-05, "loss": 1.3444, "step": 18942 }, { "epoch": 47.3575, "grad_norm": 8.125, "learning_rate": 1.1337632964685326e-05, "loss": 1.3294, "step": 18943 }, { "epoch": 47.36, "grad_norm": 6.78125, "learning_rate": 1.1336830547895752e-05, "loss": 1.2513, "step": 18944 }, { "epoch": 47.3625, "grad_norm": 9.75, "learning_rate": 1.133602812234197e-05, "loss": 1.3929, "step": 18945 }, { "epoch": 47.365, "grad_norm": 6.65625, "learning_rate": 1.1335225688029243e-05, "loss": 1.2856, "step": 18946 }, { "epoch": 47.3675, "grad_norm": 5.3125, "learning_rate": 1.1334423244962829e-05, "loss": 1.3995, "step": 18947 }, { "epoch": 47.37, "grad_norm": 8.125, "learning_rate": 1.133362079314799e-05, "loss": 1.3436, "step": 18948 }, { "epoch": 47.3725, "grad_norm": 10.6875, "learning_rate": 1.1332818332589984e-05, "loss": 1.3624, "step": 18949 }, { "epoch": 47.375, "grad_norm": 7.46875, "learning_rate": 1.1332015863294078e-05, "loss": 1.3157, "step": 18950 }, { "epoch": 47.3775, "grad_norm": 7.0, "learning_rate": 1.1331213385265524e-05, "loss": 1.332, "step": 18951 }, { "epoch": 47.38, "grad_norm": 7.65625, "learning_rate": 1.1330410898509594e-05, "loss": 1.3025, "step": 18952 }, { "epoch": 47.3825, "grad_norm": 9.25, "learning_rate": 1.132960840303154e-05, "loss": 1.4303, "step": 18953 }, { "epoch": 47.385, "grad_norm": 7.1875, "learning_rate": 1.1328805898836628e-05, "loss": 1.2601, "step": 18954 }, { "epoch": 47.3875, "grad_norm": 5.375, "learning_rate": 1.1328003385930117e-05, "loss": 1.2313, "step": 18955 }, { "epoch": 47.39, "grad_norm": 6.4375, "learning_rate": 1.132720086431727e-05, "loss": 1.2888, "step": 18956 }, { "epoch": 47.3925, "grad_norm": 6.25, "learning_rate": 1.1326398334003346e-05, "loss": 1.1214, "step": 18957 }, { "epoch": 47.395, "grad_norm": 12.0625, "learning_rate": 1.1325595794993606e-05, "loss": 1.3448, "step": 18958 }, { "epoch": 47.3975, "grad_norm": 7.0625, "learning_rate": 1.1324793247293315e-05, "loss": 1.3746, "step": 18959 }, { "epoch": 47.4, "grad_norm": 6.09375, "learning_rate": 1.1323990690907734e-05, "loss": 1.2582, "step": 18960 }, { "epoch": 47.4025, "grad_norm": 8.375, "learning_rate": 1.1323188125842119e-05, "loss": 1.4152, "step": 18961 }, { "epoch": 47.405, "grad_norm": 8.75, "learning_rate": 1.132238555210174e-05, "loss": 1.3411, "step": 18962 }, { "epoch": 47.4075, "grad_norm": 7.15625, "learning_rate": 1.1321582969691854e-05, "loss": 1.3991, "step": 18963 }, { "epoch": 47.41, "grad_norm": 8.3125, "learning_rate": 1.1320780378617717e-05, "loss": 1.4877, "step": 18964 }, { "epoch": 47.4125, "grad_norm": 7.90625, "learning_rate": 1.1319977778884604e-05, "loss": 1.389, "step": 18965 }, { "epoch": 47.415, "grad_norm": 6.46875, "learning_rate": 1.1319175170497764e-05, "loss": 1.4072, "step": 18966 }, { "epoch": 47.4175, "grad_norm": 6.5625, "learning_rate": 1.131837255346247e-05, "loss": 1.2094, "step": 18967 }, { "epoch": 47.42, "grad_norm": 5.3125, "learning_rate": 1.1317569927783971e-05, "loss": 1.2335, "step": 18968 }, { "epoch": 47.4225, "grad_norm": 7.59375, "learning_rate": 1.1316767293467544e-05, "loss": 1.3447, "step": 18969 }, { "epoch": 47.425, "grad_norm": 6.90625, "learning_rate": 1.1315964650518438e-05, "loss": 1.2858, "step": 18970 }, { "epoch": 47.4275, "grad_norm": 9.3125, "learning_rate": 1.1315161998941924e-05, "loss": 1.321, "step": 18971 }, { "epoch": 47.43, "grad_norm": 9.125, "learning_rate": 1.1314359338743257e-05, "loss": 1.3813, "step": 18972 }, { "epoch": 47.4325, "grad_norm": 10.0625, "learning_rate": 1.1313556669927705e-05, "loss": 1.333, "step": 18973 }, { "epoch": 47.435, "grad_norm": 10.9375, "learning_rate": 1.1312753992500526e-05, "loss": 1.413, "step": 18974 }, { "epoch": 47.4375, "grad_norm": 6.9375, "learning_rate": 1.1311951306466987e-05, "loss": 1.4461, "step": 18975 }, { "epoch": 47.44, "grad_norm": 6.96875, "learning_rate": 1.1311148611832346e-05, "loss": 1.4147, "step": 18976 }, { "epoch": 47.4425, "grad_norm": 7.25, "learning_rate": 1.1310345908601867e-05, "loss": 1.2185, "step": 18977 }, { "epoch": 47.445, "grad_norm": 6.25, "learning_rate": 1.1309543196780814e-05, "loss": 1.2781, "step": 18978 }, { "epoch": 47.4475, "grad_norm": 10.6875, "learning_rate": 1.1308740476374447e-05, "loss": 1.2954, "step": 18979 }, { "epoch": 47.45, "grad_norm": 9.25, "learning_rate": 1.1307937747388034e-05, "loss": 1.3549, "step": 18980 }, { "epoch": 47.4525, "grad_norm": 5.90625, "learning_rate": 1.1307135009826827e-05, "loss": 1.1587, "step": 18981 }, { "epoch": 47.455, "grad_norm": 5.65625, "learning_rate": 1.1306332263696098e-05, "loss": 1.3548, "step": 18982 }, { "epoch": 47.4575, "grad_norm": 10.0, "learning_rate": 1.1305529509001105e-05, "loss": 1.1735, "step": 18983 }, { "epoch": 47.46, "grad_norm": 6.96875, "learning_rate": 1.1304726745747119e-05, "loss": 1.3462, "step": 18984 }, { "epoch": 47.4625, "grad_norm": 6.3125, "learning_rate": 1.1303923973939392e-05, "loss": 1.2711, "step": 18985 }, { "epoch": 47.465, "grad_norm": 6.65625, "learning_rate": 1.1303121193583191e-05, "loss": 1.318, "step": 18986 }, { "epoch": 47.4675, "grad_norm": 8.3125, "learning_rate": 1.1302318404683781e-05, "loss": 1.3809, "step": 18987 }, { "epoch": 47.47, "grad_norm": 6.96875, "learning_rate": 1.1301515607246422e-05, "loss": 1.3382, "step": 18988 }, { "epoch": 47.4725, "grad_norm": 6.5, "learning_rate": 1.1300712801276381e-05, "loss": 1.3494, "step": 18989 }, { "epoch": 47.475, "grad_norm": 10.4375, "learning_rate": 1.129990998677892e-05, "loss": 1.3481, "step": 18990 }, { "epoch": 47.4775, "grad_norm": 8.1875, "learning_rate": 1.1299107163759298e-05, "loss": 1.4051, "step": 18991 }, { "epoch": 47.48, "grad_norm": 7.25, "learning_rate": 1.129830433222278e-05, "loss": 1.3322, "step": 18992 }, { "epoch": 47.4825, "grad_norm": 8.1875, "learning_rate": 1.1297501492174634e-05, "loss": 1.2295, "step": 18993 }, { "epoch": 47.485, "grad_norm": 8.0, "learning_rate": 1.129669864362012e-05, "loss": 1.2926, "step": 18994 }, { "epoch": 47.4875, "grad_norm": 7.09375, "learning_rate": 1.1295895786564503e-05, "loss": 1.3698, "step": 18995 }, { "epoch": 47.49, "grad_norm": 11.0, "learning_rate": 1.129509292101304e-05, "loss": 1.3362, "step": 18996 }, { "epoch": 47.4925, "grad_norm": 8.1875, "learning_rate": 1.1294290046971006e-05, "loss": 1.3137, "step": 18997 }, { "epoch": 47.495, "grad_norm": 9.125, "learning_rate": 1.1293487164443653e-05, "loss": 1.3065, "step": 18998 }, { "epoch": 47.4975, "grad_norm": 7.21875, "learning_rate": 1.1292684273436254e-05, "loss": 1.3518, "step": 18999 }, { "epoch": 47.5, "grad_norm": 6.59375, "learning_rate": 1.1291881373954066e-05, "loss": 1.3211, "step": 19000 }, { "epoch": 47.5025, "grad_norm": 10.875, "learning_rate": 1.1291078466002356e-05, "loss": 1.2468, "step": 19001 }, { "epoch": 47.505, "grad_norm": 6.34375, "learning_rate": 1.1290275549586387e-05, "loss": 1.2746, "step": 19002 }, { "epoch": 47.5075, "grad_norm": 9.1875, "learning_rate": 1.1289472624711425e-05, "loss": 1.3552, "step": 19003 }, { "epoch": 47.51, "grad_norm": 10.5, "learning_rate": 1.1288669691382728e-05, "loss": 1.2596, "step": 19004 }, { "epoch": 47.5125, "grad_norm": 5.8125, "learning_rate": 1.1287866749605567e-05, "loss": 1.1837, "step": 19005 }, { "epoch": 47.515, "grad_norm": 4.8125, "learning_rate": 1.1287063799385202e-05, "loss": 1.2038, "step": 19006 }, { "epoch": 47.5175, "grad_norm": 6.46875, "learning_rate": 1.12862608407269e-05, "loss": 1.3214, "step": 19007 }, { "epoch": 47.52, "grad_norm": 9.25, "learning_rate": 1.128545787363592e-05, "loss": 1.5386, "step": 19008 }, { "epoch": 47.5225, "grad_norm": 5.375, "learning_rate": 1.1284654898117533e-05, "loss": 1.3359, "step": 19009 }, { "epoch": 47.525, "grad_norm": 8.9375, "learning_rate": 1.1283851914177e-05, "loss": 1.2793, "step": 19010 }, { "epoch": 47.5275, "grad_norm": 7.03125, "learning_rate": 1.1283048921819582e-05, "loss": 1.2021, "step": 19011 }, { "epoch": 47.53, "grad_norm": 6.28125, "learning_rate": 1.1282245921050552e-05, "loss": 1.2741, "step": 19012 }, { "epoch": 47.5325, "grad_norm": 7.03125, "learning_rate": 1.1281442911875163e-05, "loss": 1.2891, "step": 19013 }, { "epoch": 47.535, "grad_norm": 6.15625, "learning_rate": 1.1280639894298689e-05, "loss": 1.1997, "step": 19014 }, { "epoch": 47.5375, "grad_norm": 6.6875, "learning_rate": 1.127983686832639e-05, "loss": 1.2716, "step": 19015 }, { "epoch": 47.54, "grad_norm": 7.15625, "learning_rate": 1.1279033833963532e-05, "loss": 1.3247, "step": 19016 }, { "epoch": 47.5425, "grad_norm": 5.625, "learning_rate": 1.127823079121538e-05, "loss": 1.4045, "step": 19017 }, { "epoch": 47.545, "grad_norm": 10.875, "learning_rate": 1.1277427740087196e-05, "loss": 1.3719, "step": 19018 }, { "epoch": 47.5475, "grad_norm": 7.6875, "learning_rate": 1.127662468058425e-05, "loss": 1.2556, "step": 19019 }, { "epoch": 47.55, "grad_norm": 6.34375, "learning_rate": 1.1275821612711803e-05, "loss": 1.4866, "step": 19020 }, { "epoch": 47.5525, "grad_norm": 8.0, "learning_rate": 1.1275018536475118e-05, "loss": 1.2604, "step": 19021 }, { "epoch": 47.555, "grad_norm": 6.78125, "learning_rate": 1.1274215451879467e-05, "loss": 1.379, "step": 19022 }, { "epoch": 47.5575, "grad_norm": 8.4375, "learning_rate": 1.1273412358930105e-05, "loss": 1.2079, "step": 19023 }, { "epoch": 47.56, "grad_norm": 5.78125, "learning_rate": 1.1272609257632305e-05, "loss": 1.2784, "step": 19024 }, { "epoch": 47.5625, "grad_norm": 7.375, "learning_rate": 1.127180614799133e-05, "loss": 1.3411, "step": 19025 }, { "epoch": 47.565, "grad_norm": 6.5, "learning_rate": 1.1271003030012447e-05, "loss": 1.2174, "step": 19026 }, { "epoch": 47.5675, "grad_norm": 7.90625, "learning_rate": 1.1270199903700915e-05, "loss": 1.3536, "step": 19027 }, { "epoch": 47.57, "grad_norm": 6.875, "learning_rate": 1.1269396769062002e-05, "loss": 1.3755, "step": 19028 }, { "epoch": 47.5725, "grad_norm": 7.625, "learning_rate": 1.126859362610098e-05, "loss": 1.3235, "step": 19029 }, { "epoch": 47.575, "grad_norm": 7.53125, "learning_rate": 1.1267790474823106e-05, "loss": 1.4953, "step": 19030 }, { "epoch": 47.5775, "grad_norm": 10.0, "learning_rate": 1.1266987315233649e-05, "loss": 1.5486, "step": 19031 }, { "epoch": 47.58, "grad_norm": 7.1875, "learning_rate": 1.1266184147337873e-05, "loss": 1.3213, "step": 19032 }, { "epoch": 47.5825, "grad_norm": 7.34375, "learning_rate": 1.1265380971141044e-05, "loss": 1.3896, "step": 19033 }, { "epoch": 47.585, "grad_norm": 7.09375, "learning_rate": 1.126457778664843e-05, "loss": 1.4915, "step": 19034 }, { "epoch": 47.5875, "grad_norm": 7.09375, "learning_rate": 1.1263774593865295e-05, "loss": 1.4826, "step": 19035 }, { "epoch": 47.59, "grad_norm": 8.5625, "learning_rate": 1.12629713927969e-05, "loss": 1.2528, "step": 19036 }, { "epoch": 47.5925, "grad_norm": 11.8125, "learning_rate": 1.1262168183448522e-05, "loss": 1.47, "step": 19037 }, { "epoch": 47.595, "grad_norm": 6.90625, "learning_rate": 1.1261364965825412e-05, "loss": 1.3246, "step": 19038 }, { "epoch": 47.5975, "grad_norm": 8.9375, "learning_rate": 1.126056173993285e-05, "loss": 1.3288, "step": 19039 }, { "epoch": 47.6, "grad_norm": 7.0625, "learning_rate": 1.1259758505776092e-05, "loss": 1.2442, "step": 19040 }, { "epoch": 47.6025, "grad_norm": 8.0625, "learning_rate": 1.125895526336041e-05, "loss": 1.3697, "step": 19041 }, { "epoch": 47.605, "grad_norm": 6.46875, "learning_rate": 1.1258152012691066e-05, "loss": 1.4096, "step": 19042 }, { "epoch": 47.6075, "grad_norm": 6.875, "learning_rate": 1.125734875377333e-05, "loss": 1.3176, "step": 19043 }, { "epoch": 47.61, "grad_norm": 6.21875, "learning_rate": 1.1256545486612462e-05, "loss": 1.444, "step": 19044 }, { "epoch": 47.6125, "grad_norm": 7.78125, "learning_rate": 1.1255742211213736e-05, "loss": 1.2389, "step": 19045 }, { "epoch": 47.615, "grad_norm": 12.875, "learning_rate": 1.1254938927582413e-05, "loss": 1.1951, "step": 19046 }, { "epoch": 47.6175, "grad_norm": 10.75, "learning_rate": 1.125413563572376e-05, "loss": 1.4985, "step": 19047 }, { "epoch": 47.62, "grad_norm": 8.125, "learning_rate": 1.1253332335643043e-05, "loss": 1.2565, "step": 19048 }, { "epoch": 47.6225, "grad_norm": 8.5, "learning_rate": 1.125252902734553e-05, "loss": 1.345, "step": 19049 }, { "epoch": 47.625, "grad_norm": 7.15625, "learning_rate": 1.125172571083649e-05, "loss": 1.2785, "step": 19050 }, { "epoch": 47.6275, "grad_norm": 7.71875, "learning_rate": 1.1250922386121181e-05, "loss": 1.3054, "step": 19051 }, { "epoch": 47.63, "grad_norm": 8.5625, "learning_rate": 1.125011905320488e-05, "loss": 1.2766, "step": 19052 }, { "epoch": 47.6325, "grad_norm": 7.46875, "learning_rate": 1.1249315712092845e-05, "loss": 1.3563, "step": 19053 }, { "epoch": 47.635, "grad_norm": 6.78125, "learning_rate": 1.124851236279035e-05, "loss": 1.3516, "step": 19054 }, { "epoch": 47.6375, "grad_norm": 6.90625, "learning_rate": 1.1247709005302654e-05, "loss": 1.2515, "step": 19055 }, { "epoch": 47.64, "grad_norm": 4.4375, "learning_rate": 1.1246905639635029e-05, "loss": 1.3421, "step": 19056 }, { "epoch": 47.6425, "grad_norm": 7.09375, "learning_rate": 1.124610226579274e-05, "loss": 1.2963, "step": 19057 }, { "epoch": 47.645, "grad_norm": 8.9375, "learning_rate": 1.1245298883781055e-05, "loss": 1.3148, "step": 19058 }, { "epoch": 47.6475, "grad_norm": 7.9375, "learning_rate": 1.1244495493605242e-05, "loss": 1.4208, "step": 19059 }, { "epoch": 47.65, "grad_norm": 8.5625, "learning_rate": 1.1243692095270565e-05, "loss": 1.3697, "step": 19060 }, { "epoch": 47.6525, "grad_norm": 10.1875, "learning_rate": 1.1242888688782289e-05, "loss": 1.2359, "step": 19061 }, { "epoch": 47.655, "grad_norm": 10.75, "learning_rate": 1.1242085274145689e-05, "loss": 1.5352, "step": 19062 }, { "epoch": 47.6575, "grad_norm": 8.625, "learning_rate": 1.1241281851366026e-05, "loss": 1.3894, "step": 19063 }, { "epoch": 47.66, "grad_norm": 5.8125, "learning_rate": 1.1240478420448568e-05, "loss": 1.2482, "step": 19064 }, { "epoch": 47.6625, "grad_norm": 10.5625, "learning_rate": 1.1239674981398587e-05, "loss": 1.3628, "step": 19065 }, { "epoch": 47.665, "grad_norm": 6.15625, "learning_rate": 1.1238871534221341e-05, "loss": 1.3487, "step": 19066 }, { "epoch": 47.6675, "grad_norm": 5.6875, "learning_rate": 1.1238068078922106e-05, "loss": 1.3273, "step": 19067 }, { "epoch": 47.67, "grad_norm": 8.875, "learning_rate": 1.1237264615506145e-05, "loss": 1.2387, "step": 19068 }, { "epoch": 47.6725, "grad_norm": 12.375, "learning_rate": 1.123646114397873e-05, "loss": 1.5169, "step": 19069 }, { "epoch": 47.675, "grad_norm": 79.0, "learning_rate": 1.123565766434512e-05, "loss": 1.4127, "step": 19070 }, { "epoch": 47.6775, "grad_norm": 8.3125, "learning_rate": 1.1234854176610592e-05, "loss": 1.2409, "step": 19071 }, { "epoch": 47.68, "grad_norm": 7.59375, "learning_rate": 1.1234050680780407e-05, "loss": 1.3414, "step": 19072 }, { "epoch": 47.6825, "grad_norm": 6.6875, "learning_rate": 1.1233247176859835e-05, "loss": 1.1733, "step": 19073 }, { "epoch": 47.685, "grad_norm": 5.8125, "learning_rate": 1.1232443664854145e-05, "loss": 1.2304, "step": 19074 }, { "epoch": 47.6875, "grad_norm": 6.375, "learning_rate": 1.1231640144768604e-05, "loss": 1.4137, "step": 19075 }, { "epoch": 47.69, "grad_norm": 6.0, "learning_rate": 1.1230836616608479e-05, "loss": 1.3297, "step": 19076 }, { "epoch": 47.6925, "grad_norm": 9.3125, "learning_rate": 1.1230033080379037e-05, "loss": 1.432, "step": 19077 }, { "epoch": 47.695, "grad_norm": 7.21875, "learning_rate": 1.1229229536085548e-05, "loss": 1.257, "step": 19078 }, { "epoch": 47.6975, "grad_norm": 8.875, "learning_rate": 1.122842598373328e-05, "loss": 1.2012, "step": 19079 }, { "epoch": 47.7, "grad_norm": 4.90625, "learning_rate": 1.1227622423327501e-05, "loss": 1.3394, "step": 19080 }, { "epoch": 47.7025, "grad_norm": 12.0625, "learning_rate": 1.1226818854873479e-05, "loss": 1.3166, "step": 19081 }, { "epoch": 47.705, "grad_norm": 9.125, "learning_rate": 1.122601527837648e-05, "loss": 1.2366, "step": 19082 }, { "epoch": 47.7075, "grad_norm": 8.9375, "learning_rate": 1.1225211693841773e-05, "loss": 1.332, "step": 19083 }, { "epoch": 47.71, "grad_norm": 8.0, "learning_rate": 1.1224408101274631e-05, "loss": 1.2878, "step": 19084 }, { "epoch": 47.7125, "grad_norm": 6.46875, "learning_rate": 1.1223604500680314e-05, "loss": 1.0656, "step": 19085 }, { "epoch": 47.715, "grad_norm": 6.75, "learning_rate": 1.1222800892064099e-05, "loss": 1.2891, "step": 19086 }, { "epoch": 47.7175, "grad_norm": 6.65625, "learning_rate": 1.1221997275431249e-05, "loss": 1.4946, "step": 19087 }, { "epoch": 47.72, "grad_norm": 7.59375, "learning_rate": 1.1221193650787032e-05, "loss": 1.3072, "step": 19088 }, { "epoch": 47.7225, "grad_norm": 6.53125, "learning_rate": 1.122039001813672e-05, "loss": 1.3225, "step": 19089 }, { "epoch": 47.725, "grad_norm": 9.5625, "learning_rate": 1.1219586377485578e-05, "loss": 1.3316, "step": 19090 }, { "epoch": 47.7275, "grad_norm": 6.78125, "learning_rate": 1.1218782728838877e-05, "loss": 1.2021, "step": 19091 }, { "epoch": 47.73, "grad_norm": 8.1875, "learning_rate": 1.1217979072201887e-05, "loss": 1.2664, "step": 19092 }, { "epoch": 47.7325, "grad_norm": 8.3125, "learning_rate": 1.1217175407579872e-05, "loss": 1.289, "step": 19093 }, { "epoch": 47.735, "grad_norm": 9.625, "learning_rate": 1.1216371734978109e-05, "loss": 1.3349, "step": 19094 }, { "epoch": 47.7375, "grad_norm": 8.625, "learning_rate": 1.1215568054401855e-05, "loss": 1.2385, "step": 19095 }, { "epoch": 47.74, "grad_norm": 9.75, "learning_rate": 1.1214764365856393e-05, "loss": 1.3458, "step": 19096 }, { "epoch": 47.7425, "grad_norm": 5.28125, "learning_rate": 1.1213960669346979e-05, "loss": 1.2631, "step": 19097 }, { "epoch": 47.745, "grad_norm": 6.84375, "learning_rate": 1.1213156964878887e-05, "loss": 1.3327, "step": 19098 }, { "epoch": 47.7475, "grad_norm": 7.59375, "learning_rate": 1.1212353252457388e-05, "loss": 1.3895, "step": 19099 }, { "epoch": 47.75, "grad_norm": 8.3125, "learning_rate": 1.1211549532087749e-05, "loss": 1.3983, "step": 19100 }, { "epoch": 47.7525, "grad_norm": 8.1875, "learning_rate": 1.1210745803775243e-05, "loss": 1.4345, "step": 19101 }, { "epoch": 47.755, "grad_norm": 6.90625, "learning_rate": 1.1209942067525132e-05, "loss": 1.3892, "step": 19102 }, { "epoch": 47.7575, "grad_norm": 7.15625, "learning_rate": 1.120913832334269e-05, "loss": 1.4734, "step": 19103 }, { "epoch": 47.76, "grad_norm": 9.4375, "learning_rate": 1.1208334571233186e-05, "loss": 1.398, "step": 19104 }, { "epoch": 47.7625, "grad_norm": 6.59375, "learning_rate": 1.120753081120189e-05, "loss": 1.3892, "step": 19105 }, { "epoch": 47.765, "grad_norm": 7.59375, "learning_rate": 1.1206727043254069e-05, "loss": 1.4516, "step": 19106 }, { "epoch": 47.7675, "grad_norm": 7.625, "learning_rate": 1.1205923267394996e-05, "loss": 1.3191, "step": 19107 }, { "epoch": 47.77, "grad_norm": 6.59375, "learning_rate": 1.1205119483629933e-05, "loss": 1.2353, "step": 19108 }, { "epoch": 47.7725, "grad_norm": 7.90625, "learning_rate": 1.120431569196416e-05, "loss": 1.3052, "step": 19109 }, { "epoch": 47.775, "grad_norm": 6.46875, "learning_rate": 1.1203511892402938e-05, "loss": 1.3036, "step": 19110 }, { "epoch": 47.7775, "grad_norm": 5.90625, "learning_rate": 1.1202708084951544e-05, "loss": 1.4168, "step": 19111 }, { "epoch": 47.78, "grad_norm": 9.1875, "learning_rate": 1.1201904269615242e-05, "loss": 1.3859, "step": 19112 }, { "epoch": 47.7825, "grad_norm": 11.125, "learning_rate": 1.1201100446399304e-05, "loss": 1.5189, "step": 19113 }, { "epoch": 47.785, "grad_norm": 7.03125, "learning_rate": 1.1200296615308998e-05, "loss": 1.3497, "step": 19114 }, { "epoch": 47.7875, "grad_norm": 23.25, "learning_rate": 1.1199492776349596e-05, "loss": 1.3299, "step": 19115 }, { "epoch": 47.79, "grad_norm": 9.9375, "learning_rate": 1.1198688929526369e-05, "loss": 1.3525, "step": 19116 }, { "epoch": 47.7925, "grad_norm": 5.84375, "learning_rate": 1.1197885074844584e-05, "loss": 1.2484, "step": 19117 }, { "epoch": 47.795, "grad_norm": 7.25, "learning_rate": 1.1197081212309513e-05, "loss": 1.488, "step": 19118 }, { "epoch": 47.7975, "grad_norm": 6.9375, "learning_rate": 1.1196277341926425e-05, "loss": 1.2241, "step": 19119 }, { "epoch": 47.8, "grad_norm": 10.6875, "learning_rate": 1.119547346370059e-05, "loss": 1.3725, "step": 19120 }, { "epoch": 47.8025, "grad_norm": 6.15625, "learning_rate": 1.1194669577637281e-05, "loss": 1.3832, "step": 19121 }, { "epoch": 47.805, "grad_norm": 7.65625, "learning_rate": 1.1193865683741767e-05, "loss": 1.454, "step": 19122 }, { "epoch": 47.8075, "grad_norm": 10.625, "learning_rate": 1.1193061782019312e-05, "loss": 1.3268, "step": 19123 }, { "epoch": 47.81, "grad_norm": 6.78125, "learning_rate": 1.1192257872475197e-05, "loss": 1.2263, "step": 19124 }, { "epoch": 47.8125, "grad_norm": 9.3125, "learning_rate": 1.1191453955114681e-05, "loss": 1.1469, "step": 19125 }, { "epoch": 47.815, "grad_norm": 6.0, "learning_rate": 1.1190650029943046e-05, "loss": 1.4942, "step": 19126 }, { "epoch": 47.8175, "grad_norm": 6.90625, "learning_rate": 1.1189846096965557e-05, "loss": 1.3333, "step": 19127 }, { "epoch": 47.82, "grad_norm": 6.5, "learning_rate": 1.1189042156187482e-05, "loss": 1.4642, "step": 19128 }, { "epoch": 47.8225, "grad_norm": 7.46875, "learning_rate": 1.1188238207614096e-05, "loss": 1.1799, "step": 19129 }, { "epoch": 47.825, "grad_norm": 7.6875, "learning_rate": 1.1187434251250666e-05, "loss": 1.5767, "step": 19130 }, { "epoch": 47.8275, "grad_norm": 7.4375, "learning_rate": 1.1186630287102467e-05, "loss": 1.3469, "step": 19131 }, { "epoch": 47.83, "grad_norm": 5.1875, "learning_rate": 1.1185826315174766e-05, "loss": 1.3461, "step": 19132 }, { "epoch": 47.8325, "grad_norm": 5.9375, "learning_rate": 1.1185022335472836e-05, "loss": 1.2598, "step": 19133 }, { "epoch": 47.835, "grad_norm": 9.0, "learning_rate": 1.1184218348001944e-05, "loss": 1.3375, "step": 19134 }, { "epoch": 47.8375, "grad_norm": 11.625, "learning_rate": 1.1183414352767366e-05, "loss": 1.3876, "step": 19135 }, { "epoch": 47.84, "grad_norm": 6.21875, "learning_rate": 1.118261034977437e-05, "loss": 1.268, "step": 19136 }, { "epoch": 47.8425, "grad_norm": 9.625, "learning_rate": 1.1181806339028232e-05, "loss": 1.2535, "step": 19137 }, { "epoch": 47.845, "grad_norm": 6.25, "learning_rate": 1.1181002320534214e-05, "loss": 1.1888, "step": 19138 }, { "epoch": 47.8475, "grad_norm": 8.25, "learning_rate": 1.1180198294297596e-05, "loss": 1.3245, "step": 19139 }, { "epoch": 47.85, "grad_norm": 11.9375, "learning_rate": 1.1179394260323639e-05, "loss": 1.4728, "step": 19140 }, { "epoch": 47.8525, "grad_norm": 6.90625, "learning_rate": 1.1178590218617626e-05, "loss": 1.4077, "step": 19141 }, { "epoch": 47.855, "grad_norm": 6.59375, "learning_rate": 1.1177786169184821e-05, "loss": 1.2598, "step": 19142 }, { "epoch": 47.8575, "grad_norm": 6.875, "learning_rate": 1.1176982112030496e-05, "loss": 1.3521, "step": 19143 }, { "epoch": 47.86, "grad_norm": 12.75, "learning_rate": 1.1176178047159926e-05, "loss": 1.5795, "step": 19144 }, { "epoch": 47.8625, "grad_norm": 7.4375, "learning_rate": 1.1175373974578378e-05, "loss": 1.1653, "step": 19145 }, { "epoch": 47.865, "grad_norm": 9.6875, "learning_rate": 1.1174569894291124e-05, "loss": 1.2861, "step": 19146 }, { "epoch": 47.8675, "grad_norm": 7.84375, "learning_rate": 1.117376580630344e-05, "loss": 1.3934, "step": 19147 }, { "epoch": 47.87, "grad_norm": 10.375, "learning_rate": 1.1172961710620591e-05, "loss": 1.279, "step": 19148 }, { "epoch": 47.8725, "grad_norm": 8.1875, "learning_rate": 1.1172157607247852e-05, "loss": 1.3088, "step": 19149 }, { "epoch": 47.875, "grad_norm": 8.3125, "learning_rate": 1.1171353496190499e-05, "loss": 1.338, "step": 19150 }, { "epoch": 47.8775, "grad_norm": 9.0, "learning_rate": 1.1170549377453797e-05, "loss": 1.3833, "step": 19151 }, { "epoch": 47.88, "grad_norm": 6.5, "learning_rate": 1.116974525104302e-05, "loss": 1.2263, "step": 19152 }, { "epoch": 47.8825, "grad_norm": 7.78125, "learning_rate": 1.1168941116963439e-05, "loss": 1.2851, "step": 19153 }, { "epoch": 47.885, "grad_norm": 6.96875, "learning_rate": 1.116813697522033e-05, "loss": 1.3047, "step": 19154 }, { "epoch": 47.8875, "grad_norm": 6.0, "learning_rate": 1.1167332825818954e-05, "loss": 1.4919, "step": 19155 }, { "epoch": 47.89, "grad_norm": 9.1875, "learning_rate": 1.11665286687646e-05, "loss": 1.2782, "step": 19156 }, { "epoch": 47.8925, "grad_norm": 12.9375, "learning_rate": 1.1165724504062525e-05, "loss": 1.2488, "step": 19157 }, { "epoch": 47.895, "grad_norm": 6.875, "learning_rate": 1.116492033171801e-05, "loss": 1.3713, "step": 19158 }, { "epoch": 47.8975, "grad_norm": 7.09375, "learning_rate": 1.1164116151736321e-05, "loss": 1.2854, "step": 19159 }, { "epoch": 47.9, "grad_norm": 9.0, "learning_rate": 1.1163311964122733e-05, "loss": 1.3282, "step": 19160 }, { "epoch": 47.9025, "grad_norm": 9.0, "learning_rate": 1.116250776888252e-05, "loss": 1.3915, "step": 19161 }, { "epoch": 47.905, "grad_norm": 6.90625, "learning_rate": 1.1161703566020956e-05, "loss": 1.4427, "step": 19162 }, { "epoch": 47.9075, "grad_norm": 8.125, "learning_rate": 1.1160899355543303e-05, "loss": 1.3647, "step": 19163 }, { "epoch": 47.91, "grad_norm": 8.5625, "learning_rate": 1.1160095137454845e-05, "loss": 1.3703, "step": 19164 }, { "epoch": 47.9125, "grad_norm": 13.0, "learning_rate": 1.1159290911760848e-05, "loss": 1.3592, "step": 19165 }, { "epoch": 47.915, "grad_norm": 6.59375, "learning_rate": 1.1158486678466584e-05, "loss": 1.3394, "step": 19166 }, { "epoch": 47.9175, "grad_norm": 7.0, "learning_rate": 1.1157682437577333e-05, "loss": 1.3463, "step": 19167 }, { "epoch": 47.92, "grad_norm": 7.53125, "learning_rate": 1.1156878189098357e-05, "loss": 1.3848, "step": 19168 }, { "epoch": 47.9225, "grad_norm": 9.75, "learning_rate": 1.115607393303494e-05, "loss": 1.3764, "step": 19169 }, { "epoch": 47.925, "grad_norm": 7.1875, "learning_rate": 1.1155269669392342e-05, "loss": 1.3477, "step": 19170 }, { "epoch": 47.9275, "grad_norm": 4.625, "learning_rate": 1.1154465398175847e-05, "loss": 1.3695, "step": 19171 }, { "epoch": 47.93, "grad_norm": 6.0, "learning_rate": 1.115366111939072e-05, "loss": 1.1516, "step": 19172 }, { "epoch": 47.9325, "grad_norm": 8.6875, "learning_rate": 1.1152856833042239e-05, "loss": 1.3478, "step": 19173 }, { "epoch": 47.935, "grad_norm": 5.4375, "learning_rate": 1.1152052539135675e-05, "loss": 1.2087, "step": 19174 }, { "epoch": 47.9375, "grad_norm": 8.4375, "learning_rate": 1.1151248237676297e-05, "loss": 1.3925, "step": 19175 }, { "epoch": 47.94, "grad_norm": 4.53125, "learning_rate": 1.1150443928669383e-05, "loss": 1.3813, "step": 19176 }, { "epoch": 47.9425, "grad_norm": 6.0, "learning_rate": 1.1149639612120208e-05, "loss": 1.3452, "step": 19177 }, { "epoch": 47.945, "grad_norm": 8.25, "learning_rate": 1.114883528803404e-05, "loss": 1.2763, "step": 19178 }, { "epoch": 47.9475, "grad_norm": 6.40625, "learning_rate": 1.1148030956416154e-05, "loss": 1.3509, "step": 19179 }, { "epoch": 47.95, "grad_norm": 6.75, "learning_rate": 1.114722661727182e-05, "loss": 1.3202, "step": 19180 }, { "epoch": 47.9525, "grad_norm": 10.75, "learning_rate": 1.1146422270606319e-05, "loss": 1.294, "step": 19181 }, { "epoch": 47.955, "grad_norm": 7.28125, "learning_rate": 1.1145617916424916e-05, "loss": 1.2318, "step": 19182 }, { "epoch": 47.9575, "grad_norm": 7.46875, "learning_rate": 1.1144813554732888e-05, "loss": 1.209, "step": 19183 }, { "epoch": 47.96, "grad_norm": 8.125, "learning_rate": 1.114400918553551e-05, "loss": 1.2873, "step": 19184 }, { "epoch": 47.9625, "grad_norm": 9.5625, "learning_rate": 1.1143204808838051e-05, "loss": 1.3415, "step": 19185 }, { "epoch": 47.965, "grad_norm": 5.90625, "learning_rate": 1.1142400424645787e-05, "loss": 1.3815, "step": 19186 }, { "epoch": 47.9675, "grad_norm": 13.125, "learning_rate": 1.1141596032963996e-05, "loss": 1.3522, "step": 19187 }, { "epoch": 47.97, "grad_norm": 10.6875, "learning_rate": 1.1140791633797943e-05, "loss": 1.3797, "step": 19188 }, { "epoch": 47.9725, "grad_norm": 10.625, "learning_rate": 1.1139987227152906e-05, "loss": 1.2233, "step": 19189 }, { "epoch": 47.975, "grad_norm": 7.78125, "learning_rate": 1.1139182813034157e-05, "loss": 1.3928, "step": 19190 }, { "epoch": 47.9775, "grad_norm": 9.4375, "learning_rate": 1.1138378391446974e-05, "loss": 1.2349, "step": 19191 }, { "epoch": 47.98, "grad_norm": 5.5625, "learning_rate": 1.113757396239663e-05, "loss": 1.3185, "step": 19192 }, { "epoch": 47.9825, "grad_norm": 8.125, "learning_rate": 1.1136769525888391e-05, "loss": 1.4249, "step": 19193 }, { "epoch": 47.985, "grad_norm": 9.6875, "learning_rate": 1.113596508192754e-05, "loss": 1.3982, "step": 19194 }, { "epoch": 47.9875, "grad_norm": 8.0625, "learning_rate": 1.1135160630519346e-05, "loss": 1.3611, "step": 19195 }, { "epoch": 47.99, "grad_norm": 7.3125, "learning_rate": 1.1134356171669085e-05, "loss": 1.3816, "step": 19196 }, { "epoch": 47.9925, "grad_norm": 7.25, "learning_rate": 1.113355170538203e-05, "loss": 1.5239, "step": 19197 }, { "epoch": 47.995, "grad_norm": 5.96875, "learning_rate": 1.1132747231663456e-05, "loss": 1.2596, "step": 19198 }, { "epoch": 47.9975, "grad_norm": 6.71875, "learning_rate": 1.1131942750518636e-05, "loss": 1.4055, "step": 19199 }, { "epoch": 48.0, "grad_norm": 5.90625, "learning_rate": 1.1131138261952845e-05, "loss": 1.2729, "step": 19200 }, { "epoch": 48.0025, "grad_norm": 8.125, "learning_rate": 1.1130333765971355e-05, "loss": 1.4269, "step": 19201 }, { "epoch": 48.005, "grad_norm": 6.0, "learning_rate": 1.1129529262579443e-05, "loss": 1.4283, "step": 19202 }, { "epoch": 48.0075, "grad_norm": 5.28125, "learning_rate": 1.1128724751782384e-05, "loss": 1.324, "step": 19203 }, { "epoch": 48.01, "grad_norm": 11.25, "learning_rate": 1.112792023358545e-05, "loss": 1.3801, "step": 19204 }, { "epoch": 48.0125, "grad_norm": 6.90625, "learning_rate": 1.1127115707993914e-05, "loss": 1.2723, "step": 19205 }, { "epoch": 48.015, "grad_norm": 8.5, "learning_rate": 1.1126311175013055e-05, "loss": 1.3129, "step": 19206 }, { "epoch": 48.0175, "grad_norm": 10.875, "learning_rate": 1.1125506634648145e-05, "loss": 1.377, "step": 19207 }, { "epoch": 48.02, "grad_norm": 6.375, "learning_rate": 1.1124702086904455e-05, "loss": 1.3255, "step": 19208 }, { "epoch": 48.0225, "grad_norm": 7.625, "learning_rate": 1.112389753178727e-05, "loss": 1.3552, "step": 19209 }, { "epoch": 48.025, "grad_norm": 5.75, "learning_rate": 1.112309296930185e-05, "loss": 1.2349, "step": 19210 }, { "epoch": 48.0275, "grad_norm": 11.0, "learning_rate": 1.1122288399453485e-05, "loss": 1.3151, "step": 19211 }, { "epoch": 48.03, "grad_norm": 7.8125, "learning_rate": 1.1121483822247434e-05, "loss": 1.397, "step": 19212 }, { "epoch": 48.0325, "grad_norm": 7.15625, "learning_rate": 1.1120679237688988e-05, "loss": 1.3946, "step": 19213 }, { "epoch": 48.035, "grad_norm": 9.9375, "learning_rate": 1.1119874645783408e-05, "loss": 1.3749, "step": 19214 }, { "epoch": 48.0375, "grad_norm": 10.0625, "learning_rate": 1.1119070046535976e-05, "loss": 1.3696, "step": 19215 }, { "epoch": 48.04, "grad_norm": 6.8125, "learning_rate": 1.1118265439951968e-05, "loss": 1.228, "step": 19216 }, { "epoch": 48.0425, "grad_norm": 9.4375, "learning_rate": 1.1117460826036653e-05, "loss": 1.341, "step": 19217 }, { "epoch": 48.045, "grad_norm": 7.375, "learning_rate": 1.111665620479531e-05, "loss": 1.2986, "step": 19218 }, { "epoch": 48.0475, "grad_norm": 6.84375, "learning_rate": 1.1115851576233216e-05, "loss": 1.2346, "step": 19219 }, { "epoch": 48.05, "grad_norm": 7.84375, "learning_rate": 1.1115046940355643e-05, "loss": 1.3635, "step": 19220 }, { "epoch": 48.0525, "grad_norm": 5.0, "learning_rate": 1.1114242297167865e-05, "loss": 1.1457, "step": 19221 }, { "epoch": 48.055, "grad_norm": 9.1875, "learning_rate": 1.111343764667516e-05, "loss": 1.3521, "step": 19222 }, { "epoch": 48.0575, "grad_norm": 6.28125, "learning_rate": 1.1112632988882804e-05, "loss": 1.3378, "step": 19223 }, { "epoch": 48.06, "grad_norm": 13.9375, "learning_rate": 1.111182832379607e-05, "loss": 1.3618, "step": 19224 }, { "epoch": 48.0625, "grad_norm": 9.625, "learning_rate": 1.1111023651420231e-05, "loss": 1.1873, "step": 19225 }, { "epoch": 48.065, "grad_norm": 8.875, "learning_rate": 1.1110218971760571e-05, "loss": 1.456, "step": 19226 }, { "epoch": 48.0675, "grad_norm": 7.125, "learning_rate": 1.1109414284822356e-05, "loss": 1.1485, "step": 19227 }, { "epoch": 48.07, "grad_norm": 6.0625, "learning_rate": 1.1108609590610867e-05, "loss": 1.3843, "step": 19228 }, { "epoch": 48.0725, "grad_norm": 8.625, "learning_rate": 1.1107804889131377e-05, "loss": 1.4033, "step": 19229 }, { "epoch": 48.075, "grad_norm": 10.1875, "learning_rate": 1.1107000180389163e-05, "loss": 1.3677, "step": 19230 }, { "epoch": 48.0775, "grad_norm": 9.1875, "learning_rate": 1.1106195464389498e-05, "loss": 1.4272, "step": 19231 }, { "epoch": 48.08, "grad_norm": 9.75, "learning_rate": 1.110539074113766e-05, "loss": 1.4532, "step": 19232 }, { "epoch": 48.0825, "grad_norm": 10.875, "learning_rate": 1.1104586010638927e-05, "loss": 1.2213, "step": 19233 }, { "epoch": 48.085, "grad_norm": 5.5625, "learning_rate": 1.1103781272898573e-05, "loss": 1.2431, "step": 19234 }, { "epoch": 48.0875, "grad_norm": 7.84375, "learning_rate": 1.110297652792187e-05, "loss": 1.3272, "step": 19235 }, { "epoch": 48.09, "grad_norm": 4.5625, "learning_rate": 1.1102171775714097e-05, "loss": 1.3407, "step": 19236 }, { "epoch": 48.0925, "grad_norm": 9.1875, "learning_rate": 1.110136701628053e-05, "loss": 1.2421, "step": 19237 }, { "epoch": 48.095, "grad_norm": 6.875, "learning_rate": 1.1100562249626447e-05, "loss": 1.2845, "step": 19238 }, { "epoch": 48.0975, "grad_norm": 18.125, "learning_rate": 1.1099757475757125e-05, "loss": 1.3194, "step": 19239 }, { "epoch": 48.1, "grad_norm": 8.0625, "learning_rate": 1.109895269467783e-05, "loss": 1.3051, "step": 19240 }, { "epoch": 48.1025, "grad_norm": 6.09375, "learning_rate": 1.109814790639385e-05, "loss": 1.109, "step": 19241 }, { "epoch": 48.105, "grad_norm": 6.84375, "learning_rate": 1.1097343110910452e-05, "loss": 1.3096, "step": 19242 }, { "epoch": 48.1075, "grad_norm": 6.3125, "learning_rate": 1.1096538308232923e-05, "loss": 1.2418, "step": 19243 }, { "epoch": 48.11, "grad_norm": 5.4375, "learning_rate": 1.1095733498366527e-05, "loss": 1.3027, "step": 19244 }, { "epoch": 48.1125, "grad_norm": 7.125, "learning_rate": 1.1094928681316548e-05, "loss": 1.2373, "step": 19245 }, { "epoch": 48.115, "grad_norm": 15.625, "learning_rate": 1.1094123857088259e-05, "loss": 1.2268, "step": 19246 }, { "epoch": 48.1175, "grad_norm": 8.125, "learning_rate": 1.109331902568694e-05, "loss": 1.2129, "step": 19247 }, { "epoch": 48.12, "grad_norm": 8.125, "learning_rate": 1.1092514187117865e-05, "loss": 1.2265, "step": 19248 }, { "epoch": 48.1225, "grad_norm": 8.625, "learning_rate": 1.109170934138631e-05, "loss": 1.3945, "step": 19249 }, { "epoch": 48.125, "grad_norm": 4.96875, "learning_rate": 1.109090448849755e-05, "loss": 1.25, "step": 19250 }, { "epoch": 48.1275, "grad_norm": 8.1875, "learning_rate": 1.1090099628456867e-05, "loss": 1.2848, "step": 19251 }, { "epoch": 48.13, "grad_norm": 7.5, "learning_rate": 1.1089294761269534e-05, "loss": 1.4171, "step": 19252 }, { "epoch": 48.1325, "grad_norm": 6.53125, "learning_rate": 1.1088489886940827e-05, "loss": 1.3651, "step": 19253 }, { "epoch": 48.135, "grad_norm": 5.9375, "learning_rate": 1.1087685005476025e-05, "loss": 1.4625, "step": 19254 }, { "epoch": 48.1375, "grad_norm": 7.0625, "learning_rate": 1.1086880116880403e-05, "loss": 1.2685, "step": 19255 }, { "epoch": 48.14, "grad_norm": 8.8125, "learning_rate": 1.1086075221159237e-05, "loss": 1.244, "step": 19256 }, { "epoch": 48.1425, "grad_norm": 6.0, "learning_rate": 1.1085270318317809e-05, "loss": 1.172, "step": 19257 }, { "epoch": 48.145, "grad_norm": 7.84375, "learning_rate": 1.1084465408361389e-05, "loss": 1.416, "step": 19258 }, { "epoch": 48.1475, "grad_norm": 10.875, "learning_rate": 1.1083660491295258e-05, "loss": 1.2352, "step": 19259 }, { "epoch": 48.15, "grad_norm": 10.0625, "learning_rate": 1.1082855567124693e-05, "loss": 1.3078, "step": 19260 }, { "epoch": 48.1525, "grad_norm": 6.96875, "learning_rate": 1.1082050635854969e-05, "loss": 1.3464, "step": 19261 }, { "epoch": 48.155, "grad_norm": 8.8125, "learning_rate": 1.1081245697491365e-05, "loss": 1.2312, "step": 19262 }, { "epoch": 48.1575, "grad_norm": 6.78125, "learning_rate": 1.1080440752039157e-05, "loss": 1.1636, "step": 19263 }, { "epoch": 48.16, "grad_norm": 6.8125, "learning_rate": 1.1079635799503625e-05, "loss": 1.286, "step": 19264 }, { "epoch": 48.1625, "grad_norm": 5.46875, "learning_rate": 1.107883083989004e-05, "loss": 1.3448, "step": 19265 }, { "epoch": 48.165, "grad_norm": 9.1875, "learning_rate": 1.1078025873203687e-05, "loss": 1.4206, "step": 19266 }, { "epoch": 48.1675, "grad_norm": 6.375, "learning_rate": 1.1077220899449834e-05, "loss": 1.3543, "step": 19267 }, { "epoch": 48.17, "grad_norm": 5.78125, "learning_rate": 1.1076415918633771e-05, "loss": 1.3606, "step": 19268 }, { "epoch": 48.1725, "grad_norm": 8.0625, "learning_rate": 1.1075610930760765e-05, "loss": 1.3044, "step": 19269 }, { "epoch": 48.175, "grad_norm": 5.25, "learning_rate": 1.1074805935836096e-05, "loss": 1.3036, "step": 19270 }, { "epoch": 48.1775, "grad_norm": 7.96875, "learning_rate": 1.1074000933865043e-05, "loss": 1.3412, "step": 19271 }, { "epoch": 48.18, "grad_norm": 8.0625, "learning_rate": 1.1073195924852882e-05, "loss": 1.3092, "step": 19272 }, { "epoch": 48.1825, "grad_norm": 6.625, "learning_rate": 1.1072390908804894e-05, "loss": 1.3224, "step": 19273 }, { "epoch": 48.185, "grad_norm": 8.0625, "learning_rate": 1.107158588572635e-05, "loss": 1.3293, "step": 19274 }, { "epoch": 48.1875, "grad_norm": 6.375, "learning_rate": 1.1070780855622537e-05, "loss": 1.2971, "step": 19275 }, { "epoch": 48.19, "grad_norm": 6.90625, "learning_rate": 1.1069975818498724e-05, "loss": 1.4287, "step": 19276 }, { "epoch": 48.1925, "grad_norm": 7.1875, "learning_rate": 1.1069170774360192e-05, "loss": 1.4073, "step": 19277 }, { "epoch": 48.195, "grad_norm": 7.625, "learning_rate": 1.1068365723212219e-05, "loss": 1.3718, "step": 19278 }, { "epoch": 48.1975, "grad_norm": 6.71875, "learning_rate": 1.1067560665060086e-05, "loss": 1.4052, "step": 19279 }, { "epoch": 48.2, "grad_norm": 6.09375, "learning_rate": 1.1066755599909065e-05, "loss": 1.3432, "step": 19280 }, { "epoch": 48.2025, "grad_norm": 9.375, "learning_rate": 1.1065950527764442e-05, "loss": 1.2255, "step": 19281 }, { "epoch": 48.205, "grad_norm": 6.8125, "learning_rate": 1.1065145448631483e-05, "loss": 1.273, "step": 19282 }, { "epoch": 48.2075, "grad_norm": 6.9375, "learning_rate": 1.1064340362515478e-05, "loss": 1.3629, "step": 19283 }, { "epoch": 48.21, "grad_norm": 9.0, "learning_rate": 1.10635352694217e-05, "loss": 1.2234, "step": 19284 }, { "epoch": 48.2125, "grad_norm": 8.75, "learning_rate": 1.1062730169355424e-05, "loss": 1.2647, "step": 19285 }, { "epoch": 48.215, "grad_norm": 7.71875, "learning_rate": 1.1061925062321935e-05, "loss": 1.4219, "step": 19286 }, { "epoch": 48.2175, "grad_norm": 8.125, "learning_rate": 1.1061119948326506e-05, "loss": 1.2652, "step": 19287 }, { "epoch": 48.22, "grad_norm": 7.46875, "learning_rate": 1.1060314827374417e-05, "loss": 1.3595, "step": 19288 }, { "epoch": 48.2225, "grad_norm": 9.75, "learning_rate": 1.1059509699470947e-05, "loss": 1.2649, "step": 19289 }, { "epoch": 48.225, "grad_norm": 7.6875, "learning_rate": 1.1058704564621372e-05, "loss": 1.3755, "step": 19290 }, { "epoch": 48.2275, "grad_norm": 7.8125, "learning_rate": 1.1057899422830975e-05, "loss": 1.3534, "step": 19291 }, { "epoch": 48.23, "grad_norm": 7.28125, "learning_rate": 1.1057094274105028e-05, "loss": 1.3281, "step": 19292 }, { "epoch": 48.2325, "grad_norm": 10.0625, "learning_rate": 1.1056289118448817e-05, "loss": 1.3163, "step": 19293 }, { "epoch": 48.235, "grad_norm": 9.5625, "learning_rate": 1.1055483955867619e-05, "loss": 1.5629, "step": 19294 }, { "epoch": 48.2375, "grad_norm": 8.4375, "learning_rate": 1.1054678786366704e-05, "loss": 1.273, "step": 19295 }, { "epoch": 48.24, "grad_norm": 21.375, "learning_rate": 1.1053873609951362e-05, "loss": 1.4022, "step": 19296 }, { "epoch": 48.2425, "grad_norm": 5.59375, "learning_rate": 1.1053068426626863e-05, "loss": 1.3601, "step": 19297 }, { "epoch": 48.245, "grad_norm": 9.875, "learning_rate": 1.1052263236398493e-05, "loss": 1.4347, "step": 19298 }, { "epoch": 48.2475, "grad_norm": 7.0625, "learning_rate": 1.1051458039271525e-05, "loss": 1.4537, "step": 19299 }, { "epoch": 48.25, "grad_norm": 11.6875, "learning_rate": 1.105065283525124e-05, "loss": 1.6263, "step": 19300 }, { "epoch": 48.2525, "grad_norm": 7.75, "learning_rate": 1.104984762434292e-05, "loss": 1.5417, "step": 19301 }, { "epoch": 48.255, "grad_norm": 9.6875, "learning_rate": 1.1049042406551837e-05, "loss": 1.5426, "step": 19302 }, { "epoch": 48.2575, "grad_norm": 10.25, "learning_rate": 1.1048237181883276e-05, "loss": 1.3601, "step": 19303 }, { "epoch": 48.26, "grad_norm": 6.9375, "learning_rate": 1.1047431950342517e-05, "loss": 1.31, "step": 19304 }, { "epoch": 48.2625, "grad_norm": 6.0, "learning_rate": 1.104662671193483e-05, "loss": 1.288, "step": 19305 }, { "epoch": 48.265, "grad_norm": 8.5, "learning_rate": 1.1045821466665504e-05, "loss": 1.3785, "step": 19306 }, { "epoch": 48.2675, "grad_norm": 8.75, "learning_rate": 1.1045016214539812e-05, "loss": 1.3221, "step": 19307 }, { "epoch": 48.27, "grad_norm": 9.6875, "learning_rate": 1.1044210955563037e-05, "loss": 1.2914, "step": 19308 }, { "epoch": 48.2725, "grad_norm": 13.25, "learning_rate": 1.1043405689740459e-05, "loss": 1.2393, "step": 19309 }, { "epoch": 48.275, "grad_norm": 9.875, "learning_rate": 1.1042600417077351e-05, "loss": 1.4409, "step": 19310 }, { "epoch": 48.2775, "grad_norm": 6.25, "learning_rate": 1.1041795137579e-05, "loss": 1.3528, "step": 19311 }, { "epoch": 48.28, "grad_norm": 10.1875, "learning_rate": 1.1040989851250678e-05, "loss": 1.2945, "step": 19312 }, { "epoch": 48.2825, "grad_norm": 6.34375, "learning_rate": 1.1040184558097674e-05, "loss": 1.3419, "step": 19313 }, { "epoch": 48.285, "grad_norm": 10.8125, "learning_rate": 1.1039379258125256e-05, "loss": 1.3747, "step": 19314 }, { "epoch": 48.2875, "grad_norm": 10.875, "learning_rate": 1.1038573951338711e-05, "loss": 1.5376, "step": 19315 }, { "epoch": 48.29, "grad_norm": 8.3125, "learning_rate": 1.1037768637743316e-05, "loss": 1.298, "step": 19316 }, { "epoch": 48.2925, "grad_norm": 6.5625, "learning_rate": 1.1036963317344351e-05, "loss": 1.4211, "step": 19317 }, { "epoch": 48.295, "grad_norm": 6.25, "learning_rate": 1.1036157990147099e-05, "loss": 1.2432, "step": 19318 }, { "epoch": 48.2975, "grad_norm": 12.0, "learning_rate": 1.1035352656156835e-05, "loss": 1.3106, "step": 19319 }, { "epoch": 48.3, "grad_norm": 9.875, "learning_rate": 1.1034547315378838e-05, "loss": 1.3551, "step": 19320 }, { "epoch": 48.3025, "grad_norm": 6.5625, "learning_rate": 1.1033741967818395e-05, "loss": 1.3437, "step": 19321 }, { "epoch": 48.305, "grad_norm": 7.5, "learning_rate": 1.1032936613480775e-05, "loss": 1.2515, "step": 19322 }, { "epoch": 48.3075, "grad_norm": 6.78125, "learning_rate": 1.103213125237127e-05, "loss": 1.3624, "step": 19323 }, { "epoch": 48.31, "grad_norm": 9.75, "learning_rate": 1.1031325884495152e-05, "loss": 1.4066, "step": 19324 }, { "epoch": 48.3125, "grad_norm": 5.5625, "learning_rate": 1.10305205098577e-05, "loss": 1.4983, "step": 19325 }, { "epoch": 48.315, "grad_norm": 4.5625, "learning_rate": 1.10297151284642e-05, "loss": 1.4334, "step": 19326 }, { "epoch": 48.3175, "grad_norm": 7.875, "learning_rate": 1.1028909740319927e-05, "loss": 1.3903, "step": 19327 }, { "epoch": 48.32, "grad_norm": 7.5, "learning_rate": 1.1028104345430161e-05, "loss": 1.3451, "step": 19328 }, { "epoch": 48.3225, "grad_norm": 8.0625, "learning_rate": 1.1027298943800188e-05, "loss": 1.3966, "step": 19329 }, { "epoch": 48.325, "grad_norm": 8.9375, "learning_rate": 1.1026493535435284e-05, "loss": 1.3777, "step": 19330 }, { "epoch": 48.3275, "grad_norm": 9.25, "learning_rate": 1.1025688120340728e-05, "loss": 1.3214, "step": 19331 }, { "epoch": 48.33, "grad_norm": 6.21875, "learning_rate": 1.10248826985218e-05, "loss": 1.2835, "step": 19332 }, { "epoch": 48.3325, "grad_norm": 7.40625, "learning_rate": 1.1024077269983784e-05, "loss": 1.4452, "step": 19333 }, { "epoch": 48.335, "grad_norm": 8.0625, "learning_rate": 1.102327183473196e-05, "loss": 1.2497, "step": 19334 }, { "epoch": 48.3375, "grad_norm": 8.5625, "learning_rate": 1.1022466392771602e-05, "loss": 1.3001, "step": 19335 }, { "epoch": 48.34, "grad_norm": 8.875, "learning_rate": 1.1021660944108e-05, "loss": 1.3989, "step": 19336 }, { "epoch": 48.3425, "grad_norm": 7.75, "learning_rate": 1.1020855488746427e-05, "loss": 1.2656, "step": 19337 }, { "epoch": 48.345, "grad_norm": 7.53125, "learning_rate": 1.1020050026692171e-05, "loss": 1.2329, "step": 19338 }, { "epoch": 48.3475, "grad_norm": 7.90625, "learning_rate": 1.10192445579505e-05, "loss": 1.1699, "step": 19339 }, { "epoch": 48.35, "grad_norm": 6.59375, "learning_rate": 1.101843908252671e-05, "loss": 1.286, "step": 19340 }, { "epoch": 48.3525, "grad_norm": 5.4375, "learning_rate": 1.101763360042607e-05, "loss": 1.3634, "step": 19341 }, { "epoch": 48.355, "grad_norm": 10.0, "learning_rate": 1.1016828111653866e-05, "loss": 1.3032, "step": 19342 }, { "epoch": 48.3575, "grad_norm": 6.78125, "learning_rate": 1.1016022616215376e-05, "loss": 1.4903, "step": 19343 }, { "epoch": 48.36, "grad_norm": 8.0, "learning_rate": 1.1015217114115884e-05, "loss": 1.4091, "step": 19344 }, { "epoch": 48.3625, "grad_norm": 6.84375, "learning_rate": 1.1014411605360669e-05, "loss": 1.2276, "step": 19345 }, { "epoch": 48.365, "grad_norm": 11.3125, "learning_rate": 1.1013606089955011e-05, "loss": 1.2598, "step": 19346 }, { "epoch": 48.3675, "grad_norm": 6.875, "learning_rate": 1.1012800567904194e-05, "loss": 1.1332, "step": 19347 }, { "epoch": 48.37, "grad_norm": 6.25, "learning_rate": 1.1011995039213494e-05, "loss": 1.3935, "step": 19348 }, { "epoch": 48.3725, "grad_norm": 6.78125, "learning_rate": 1.1011189503888198e-05, "loss": 1.2642, "step": 19349 }, { "epoch": 48.375, "grad_norm": 9.6875, "learning_rate": 1.1010383961933582e-05, "loss": 1.2454, "step": 19350 }, { "epoch": 48.3775, "grad_norm": 8.6875, "learning_rate": 1.1009578413354933e-05, "loss": 1.2547, "step": 19351 }, { "epoch": 48.38, "grad_norm": 7.28125, "learning_rate": 1.1008772858157524e-05, "loss": 1.2498, "step": 19352 }, { "epoch": 48.3825, "grad_norm": 8.875, "learning_rate": 1.1007967296346645e-05, "loss": 1.4268, "step": 19353 }, { "epoch": 48.385, "grad_norm": 9.8125, "learning_rate": 1.1007161727927567e-05, "loss": 1.3436, "step": 19354 }, { "epoch": 48.3875, "grad_norm": 6.84375, "learning_rate": 1.1006356152905581e-05, "loss": 1.36, "step": 19355 }, { "epoch": 48.39, "grad_norm": 7.6875, "learning_rate": 1.1005550571285964e-05, "loss": 1.2801, "step": 19356 }, { "epoch": 48.3925, "grad_norm": 5.71875, "learning_rate": 1.1004744983073996e-05, "loss": 1.4846, "step": 19357 }, { "epoch": 48.395, "grad_norm": 5.75, "learning_rate": 1.100393938827496e-05, "loss": 1.3462, "step": 19358 }, { "epoch": 48.3975, "grad_norm": 8.375, "learning_rate": 1.1003133786894138e-05, "loss": 1.4296, "step": 19359 }, { "epoch": 48.4, "grad_norm": 7.8125, "learning_rate": 1.1002328178936813e-05, "loss": 1.3585, "step": 19360 }, { "epoch": 48.4025, "grad_norm": 8.125, "learning_rate": 1.1001522564408263e-05, "loss": 1.3715, "step": 19361 }, { "epoch": 48.405, "grad_norm": 8.25, "learning_rate": 1.100071694331377e-05, "loss": 1.229, "step": 19362 }, { "epoch": 48.4075, "grad_norm": 9.6875, "learning_rate": 1.0999911315658618e-05, "loss": 1.4715, "step": 19363 }, { "epoch": 48.41, "grad_norm": 7.28125, "learning_rate": 1.0999105681448087e-05, "loss": 1.3055, "step": 19364 }, { "epoch": 48.4125, "grad_norm": 8.1875, "learning_rate": 1.0998300040687459e-05, "loss": 1.3837, "step": 19365 }, { "epoch": 48.415, "grad_norm": 6.0, "learning_rate": 1.099749439338202e-05, "loss": 1.1143, "step": 19366 }, { "epoch": 48.4175, "grad_norm": 6.4375, "learning_rate": 1.0996688739537041e-05, "loss": 1.3791, "step": 19367 }, { "epoch": 48.42, "grad_norm": 10.5, "learning_rate": 1.0995883079157817e-05, "loss": 1.2448, "step": 19368 }, { "epoch": 48.4225, "grad_norm": 6.59375, "learning_rate": 1.0995077412249617e-05, "loss": 1.4182, "step": 19369 }, { "epoch": 48.425, "grad_norm": 6.125, "learning_rate": 1.0994271738817734e-05, "loss": 1.3363, "step": 19370 }, { "epoch": 48.4275, "grad_norm": 10.75, "learning_rate": 1.0993466058867444e-05, "loss": 1.4122, "step": 19371 }, { "epoch": 48.43, "grad_norm": 7.25, "learning_rate": 1.099266037240403e-05, "loss": 1.2714, "step": 19372 }, { "epoch": 48.4325, "grad_norm": 5.75, "learning_rate": 1.0991854679432774e-05, "loss": 1.3191, "step": 19373 }, { "epoch": 48.435, "grad_norm": 6.4375, "learning_rate": 1.0991048979958958e-05, "loss": 1.3404, "step": 19374 }, { "epoch": 48.4375, "grad_norm": 7.75, "learning_rate": 1.0990243273987863e-05, "loss": 1.3002, "step": 19375 }, { "epoch": 48.44, "grad_norm": 12.1875, "learning_rate": 1.0989437561524776e-05, "loss": 1.2862, "step": 19376 }, { "epoch": 48.4425, "grad_norm": 7.46875, "learning_rate": 1.0988631842574976e-05, "loss": 1.2738, "step": 19377 }, { "epoch": 48.445, "grad_norm": 9.625, "learning_rate": 1.0987826117143743e-05, "loss": 1.4416, "step": 19378 }, { "epoch": 48.4475, "grad_norm": 7.5, "learning_rate": 1.0987020385236363e-05, "loss": 1.3715, "step": 19379 }, { "epoch": 48.45, "grad_norm": 9.5625, "learning_rate": 1.0986214646858115e-05, "loss": 1.3546, "step": 19380 }, { "epoch": 48.4525, "grad_norm": 8.4375, "learning_rate": 1.0985408902014288e-05, "loss": 1.3044, "step": 19381 }, { "epoch": 48.455, "grad_norm": 6.8125, "learning_rate": 1.0984603150710153e-05, "loss": 1.3611, "step": 19382 }, { "epoch": 48.4575, "grad_norm": 7.65625, "learning_rate": 1.0983797392951005e-05, "loss": 1.475, "step": 19383 }, { "epoch": 48.46, "grad_norm": 5.3125, "learning_rate": 1.0982991628742116e-05, "loss": 1.2654, "step": 19384 }, { "epoch": 48.4625, "grad_norm": 7.84375, "learning_rate": 1.0982185858088779e-05, "loss": 1.3716, "step": 19385 }, { "epoch": 48.465, "grad_norm": 9.5, "learning_rate": 1.0981380080996265e-05, "loss": 1.3332, "step": 19386 }, { "epoch": 48.4675, "grad_norm": 7.75, "learning_rate": 1.0980574297469867e-05, "loss": 1.411, "step": 19387 }, { "epoch": 48.47, "grad_norm": 7.25, "learning_rate": 1.097976850751486e-05, "loss": 1.2761, "step": 19388 }, { "epoch": 48.4725, "grad_norm": 7.84375, "learning_rate": 1.097896271113653e-05, "loss": 1.3857, "step": 19389 }, { "epoch": 48.475, "grad_norm": 9.625, "learning_rate": 1.097815690834016e-05, "loss": 1.3874, "step": 19390 }, { "epoch": 48.4775, "grad_norm": 8.125, "learning_rate": 1.0977351099131034e-05, "loss": 1.2588, "step": 19391 }, { "epoch": 48.48, "grad_norm": 8.6875, "learning_rate": 1.097654528351443e-05, "loss": 1.3707, "step": 19392 }, { "epoch": 48.4825, "grad_norm": 5.5625, "learning_rate": 1.0975739461495638e-05, "loss": 1.2337, "step": 19393 }, { "epoch": 48.485, "grad_norm": 7.78125, "learning_rate": 1.0974933633079934e-05, "loss": 1.3793, "step": 19394 }, { "epoch": 48.4875, "grad_norm": 5.125, "learning_rate": 1.0974127798272607e-05, "loss": 1.4158, "step": 19395 }, { "epoch": 48.49, "grad_norm": 6.9375, "learning_rate": 1.0973321957078935e-05, "loss": 1.1799, "step": 19396 }, { "epoch": 48.4925, "grad_norm": 6.375, "learning_rate": 1.0972516109504203e-05, "loss": 1.4699, "step": 19397 }, { "epoch": 48.495, "grad_norm": 7.0, "learning_rate": 1.0971710255553695e-05, "loss": 1.4983, "step": 19398 }, { "epoch": 48.4975, "grad_norm": 9.625, "learning_rate": 1.0970904395232696e-05, "loss": 1.4075, "step": 19399 }, { "epoch": 48.5, "grad_norm": 6.375, "learning_rate": 1.0970098528546482e-05, "loss": 1.3877, "step": 19400 }, { "epoch": 48.5025, "grad_norm": 7.4375, "learning_rate": 1.0969292655500343e-05, "loss": 1.4308, "step": 19401 }, { "epoch": 48.505, "grad_norm": 7.34375, "learning_rate": 1.096848677609956e-05, "loss": 1.3174, "step": 19402 }, { "epoch": 48.5075, "grad_norm": 7.625, "learning_rate": 1.0967680890349416e-05, "loss": 1.4792, "step": 19403 }, { "epoch": 48.51, "grad_norm": 8.1875, "learning_rate": 1.0966874998255197e-05, "loss": 1.4497, "step": 19404 }, { "epoch": 48.5125, "grad_norm": 10.6875, "learning_rate": 1.0966069099822182e-05, "loss": 1.2982, "step": 19405 }, { "epoch": 48.515, "grad_norm": 11.375, "learning_rate": 1.096526319505566e-05, "loss": 1.3831, "step": 19406 }, { "epoch": 48.5175, "grad_norm": 6.5, "learning_rate": 1.0964457283960905e-05, "loss": 1.4148, "step": 19407 }, { "epoch": 48.52, "grad_norm": 6.5, "learning_rate": 1.0963651366543214e-05, "loss": 1.4673, "step": 19408 }, { "epoch": 48.5225, "grad_norm": 7.03125, "learning_rate": 1.0962845442807857e-05, "loss": 1.3114, "step": 19409 }, { "epoch": 48.525, "grad_norm": 7.90625, "learning_rate": 1.0962039512760127e-05, "loss": 1.3585, "step": 19410 }, { "epoch": 48.5275, "grad_norm": 9.3125, "learning_rate": 1.0961233576405304e-05, "loss": 1.3661, "step": 19411 }, { "epoch": 48.53, "grad_norm": 6.84375, "learning_rate": 1.0960427633748672e-05, "loss": 1.3758, "step": 19412 }, { "epoch": 48.5325, "grad_norm": 8.4375, "learning_rate": 1.0959621684795514e-05, "loss": 1.3082, "step": 19413 }, { "epoch": 48.535, "grad_norm": 6.875, "learning_rate": 1.0958815729551112e-05, "loss": 1.2502, "step": 19414 }, { "epoch": 48.5375, "grad_norm": 5.625, "learning_rate": 1.0958009768020759e-05, "loss": 1.3727, "step": 19415 }, { "epoch": 48.54, "grad_norm": 7.21875, "learning_rate": 1.0957203800209729e-05, "loss": 1.4445, "step": 19416 }, { "epoch": 48.5425, "grad_norm": 11.5, "learning_rate": 1.0956397826123308e-05, "loss": 1.3072, "step": 19417 }, { "epoch": 48.545, "grad_norm": 6.84375, "learning_rate": 1.0955591845766781e-05, "loss": 1.3623, "step": 19418 }, { "epoch": 48.5475, "grad_norm": 7.21875, "learning_rate": 1.0954785859145433e-05, "loss": 1.3787, "step": 19419 }, { "epoch": 48.55, "grad_norm": 6.53125, "learning_rate": 1.0953979866264549e-05, "loss": 1.3273, "step": 19420 }, { "epoch": 48.5525, "grad_norm": 6.9375, "learning_rate": 1.0953173867129412e-05, "loss": 1.3036, "step": 19421 }, { "epoch": 48.555, "grad_norm": 6.28125, "learning_rate": 1.0952367861745298e-05, "loss": 1.3609, "step": 19422 }, { "epoch": 48.5575, "grad_norm": 10.125, "learning_rate": 1.0951561850117506e-05, "loss": 1.3851, "step": 19423 }, { "epoch": 48.56, "grad_norm": 6.4375, "learning_rate": 1.095075583225131e-05, "loss": 1.2184, "step": 19424 }, { "epoch": 48.5625, "grad_norm": 9.75, "learning_rate": 1.0949949808151998e-05, "loss": 1.4739, "step": 19425 }, { "epoch": 48.565, "grad_norm": 8.6875, "learning_rate": 1.094914377782485e-05, "loss": 1.3934, "step": 19426 }, { "epoch": 48.5675, "grad_norm": 8.0625, "learning_rate": 1.0948337741275157e-05, "loss": 1.4442, "step": 19427 }, { "epoch": 48.57, "grad_norm": 5.5, "learning_rate": 1.0947531698508198e-05, "loss": 1.1709, "step": 19428 }, { "epoch": 48.5725, "grad_norm": 10.375, "learning_rate": 1.0946725649529257e-05, "loss": 1.3897, "step": 19429 }, { "epoch": 48.575, "grad_norm": 7.65625, "learning_rate": 1.0945919594343624e-05, "loss": 1.3305, "step": 19430 }, { "epoch": 48.5775, "grad_norm": 7.40625, "learning_rate": 1.0945113532956578e-05, "loss": 1.2478, "step": 19431 }, { "epoch": 48.58, "grad_norm": 12.0, "learning_rate": 1.0944307465373405e-05, "loss": 1.3587, "step": 19432 }, { "epoch": 48.5825, "grad_norm": 9.8125, "learning_rate": 1.0943501391599392e-05, "loss": 1.4165, "step": 19433 }, { "epoch": 48.585, "grad_norm": 5.65625, "learning_rate": 1.0942695311639819e-05, "loss": 1.2783, "step": 19434 }, { "epoch": 48.5875, "grad_norm": 11.5625, "learning_rate": 1.0941889225499975e-05, "loss": 1.3216, "step": 19435 }, { "epoch": 48.59, "grad_norm": 6.71875, "learning_rate": 1.0941083133185146e-05, "loss": 1.1953, "step": 19436 }, { "epoch": 48.5925, "grad_norm": 7.78125, "learning_rate": 1.0940277034700608e-05, "loss": 1.4095, "step": 19437 }, { "epoch": 48.595, "grad_norm": 8.5625, "learning_rate": 1.0939470930051657e-05, "loss": 1.4078, "step": 19438 }, { "epoch": 48.5975, "grad_norm": 6.5625, "learning_rate": 1.0938664819243566e-05, "loss": 1.233, "step": 19439 }, { "epoch": 48.6, "grad_norm": 6.53125, "learning_rate": 1.0937858702281631e-05, "loss": 1.2555, "step": 19440 }, { "epoch": 48.6025, "grad_norm": 7.25, "learning_rate": 1.093705257917113e-05, "loss": 1.3489, "step": 19441 }, { "epoch": 48.605, "grad_norm": 6.0625, "learning_rate": 1.093624644991735e-05, "loss": 1.308, "step": 19442 }, { "epoch": 48.6075, "grad_norm": 6.75, "learning_rate": 1.0935440314525576e-05, "loss": 1.3716, "step": 19443 }, { "epoch": 48.61, "grad_norm": 5.3125, "learning_rate": 1.093463417300109e-05, "loss": 1.3311, "step": 19444 }, { "epoch": 48.6125, "grad_norm": 6.90625, "learning_rate": 1.0933828025349182e-05, "loss": 1.2469, "step": 19445 }, { "epoch": 48.615, "grad_norm": 7.9375, "learning_rate": 1.0933021871575135e-05, "loss": 1.2887, "step": 19446 }, { "epoch": 48.6175, "grad_norm": 8.75, "learning_rate": 1.0932215711684236e-05, "loss": 1.3451, "step": 19447 }, { "epoch": 48.62, "grad_norm": 8.75, "learning_rate": 1.0931409545681764e-05, "loss": 1.3793, "step": 19448 }, { "epoch": 48.6225, "grad_norm": 5.71875, "learning_rate": 1.093060337357301e-05, "loss": 1.4092, "step": 19449 }, { "epoch": 48.625, "grad_norm": 8.125, "learning_rate": 1.0929797195363259e-05, "loss": 1.4464, "step": 19450 }, { "epoch": 48.6275, "grad_norm": 6.40625, "learning_rate": 1.0928991011057792e-05, "loss": 1.1339, "step": 19451 }, { "epoch": 48.63, "grad_norm": 8.5625, "learning_rate": 1.0928184820661897e-05, "loss": 1.459, "step": 19452 }, { "epoch": 48.6325, "grad_norm": 8.1875, "learning_rate": 1.0927378624180864e-05, "loss": 1.1955, "step": 19453 }, { "epoch": 48.635, "grad_norm": 7.3125, "learning_rate": 1.092657242161997e-05, "loss": 1.2407, "step": 19454 }, { "epoch": 48.6375, "grad_norm": 6.03125, "learning_rate": 1.0925766212984508e-05, "loss": 1.3361, "step": 19455 }, { "epoch": 48.64, "grad_norm": 7.96875, "learning_rate": 1.0924959998279754e-05, "loss": 1.2703, "step": 19456 }, { "epoch": 48.6425, "grad_norm": 8.4375, "learning_rate": 1.0924153777511005e-05, "loss": 1.3896, "step": 19457 }, { "epoch": 48.645, "grad_norm": 6.84375, "learning_rate": 1.0923347550683537e-05, "loss": 1.3069, "step": 19458 }, { "epoch": 48.6475, "grad_norm": 7.75, "learning_rate": 1.092254131780264e-05, "loss": 1.2333, "step": 19459 }, { "epoch": 48.65, "grad_norm": 6.0625, "learning_rate": 1.0921735078873599e-05, "loss": 1.1745, "step": 19460 }, { "epoch": 48.6525, "grad_norm": 6.15625, "learning_rate": 1.0920928833901701e-05, "loss": 1.2402, "step": 19461 }, { "epoch": 48.655, "grad_norm": 7.1875, "learning_rate": 1.092012258289223e-05, "loss": 1.4738, "step": 19462 }, { "epoch": 48.6575, "grad_norm": 9.5625, "learning_rate": 1.0919316325850472e-05, "loss": 1.5302, "step": 19463 }, { "epoch": 48.66, "grad_norm": 9.5625, "learning_rate": 1.091851006278171e-05, "loss": 1.2812, "step": 19464 }, { "epoch": 48.6625, "grad_norm": 7.625, "learning_rate": 1.091770379369124e-05, "loss": 1.1232, "step": 19465 }, { "epoch": 48.665, "grad_norm": 7.46875, "learning_rate": 1.0916897518584333e-05, "loss": 1.2808, "step": 19466 }, { "epoch": 48.6675, "grad_norm": 8.8125, "learning_rate": 1.0916091237466287e-05, "loss": 1.2988, "step": 19467 }, { "epoch": 48.67, "grad_norm": 8.875, "learning_rate": 1.0915284950342381e-05, "loss": 1.5969, "step": 19468 }, { "epoch": 48.6725, "grad_norm": 8.0, "learning_rate": 1.0914478657217905e-05, "loss": 1.2431, "step": 19469 }, { "epoch": 48.675, "grad_norm": 7.125, "learning_rate": 1.0913672358098144e-05, "loss": 1.4077, "step": 19470 }, { "epoch": 48.6775, "grad_norm": 10.9375, "learning_rate": 1.091286605298838e-05, "loss": 1.2761, "step": 19471 }, { "epoch": 48.68, "grad_norm": 6.375, "learning_rate": 1.0912059741893908e-05, "loss": 1.22, "step": 19472 }, { "epoch": 48.6825, "grad_norm": 7.1875, "learning_rate": 1.0911253424820008e-05, "loss": 1.3234, "step": 19473 }, { "epoch": 48.685, "grad_norm": 5.875, "learning_rate": 1.0910447101771964e-05, "loss": 1.3114, "step": 19474 }, { "epoch": 48.6875, "grad_norm": 8.0, "learning_rate": 1.0909640772755065e-05, "loss": 1.2225, "step": 19475 }, { "epoch": 48.69, "grad_norm": 8.0, "learning_rate": 1.09088344377746e-05, "loss": 1.3387, "step": 19476 }, { "epoch": 48.6925, "grad_norm": 7.875, "learning_rate": 1.0908028096835852e-05, "loss": 1.5168, "step": 19477 }, { "epoch": 48.695, "grad_norm": 9.25, "learning_rate": 1.090722174994411e-05, "loss": 1.365, "step": 19478 }, { "epoch": 48.6975, "grad_norm": 7.03125, "learning_rate": 1.0906415397104655e-05, "loss": 1.2985, "step": 19479 }, { "epoch": 48.7, "grad_norm": 6.125, "learning_rate": 1.090560903832278e-05, "loss": 1.4804, "step": 19480 }, { "epoch": 48.7025, "grad_norm": 4.90625, "learning_rate": 1.0904802673603764e-05, "loss": 1.2968, "step": 19481 }, { "epoch": 48.705, "grad_norm": 12.9375, "learning_rate": 1.0903996302952904e-05, "loss": 1.3136, "step": 19482 }, { "epoch": 48.7075, "grad_norm": 9.4375, "learning_rate": 1.0903189926375476e-05, "loss": 1.371, "step": 19483 }, { "epoch": 48.71, "grad_norm": 7.15625, "learning_rate": 1.0902383543876773e-05, "loss": 1.3055, "step": 19484 }, { "epoch": 48.7125, "grad_norm": 7.21875, "learning_rate": 1.0901577155462077e-05, "loss": 1.3225, "step": 19485 }, { "epoch": 48.715, "grad_norm": 7.5625, "learning_rate": 1.0900770761136678e-05, "loss": 1.2787, "step": 19486 }, { "epoch": 48.7175, "grad_norm": 9.625, "learning_rate": 1.0899964360905864e-05, "loss": 1.3762, "step": 19487 }, { "epoch": 48.72, "grad_norm": 7.90625, "learning_rate": 1.089915795477492e-05, "loss": 1.287, "step": 19488 }, { "epoch": 48.7225, "grad_norm": 6.71875, "learning_rate": 1.089835154274913e-05, "loss": 1.4207, "step": 19489 }, { "epoch": 48.725, "grad_norm": 7.40625, "learning_rate": 1.0897545124833785e-05, "loss": 1.474, "step": 19490 }, { "epoch": 48.7275, "grad_norm": 7.03125, "learning_rate": 1.0896738701034168e-05, "loss": 1.3796, "step": 19491 }, { "epoch": 48.73, "grad_norm": 7.28125, "learning_rate": 1.089593227135557e-05, "loss": 1.3184, "step": 19492 }, { "epoch": 48.7325, "grad_norm": 6.90625, "learning_rate": 1.0895125835803278e-05, "loss": 1.3381, "step": 19493 }, { "epoch": 48.735, "grad_norm": 6.03125, "learning_rate": 1.089431939438257e-05, "loss": 1.205, "step": 19494 }, { "epoch": 48.7375, "grad_norm": 6.375, "learning_rate": 1.0893512947098746e-05, "loss": 1.2757, "step": 19495 }, { "epoch": 48.74, "grad_norm": 7.34375, "learning_rate": 1.0892706493957083e-05, "loss": 1.3199, "step": 19496 }, { "epoch": 48.7425, "grad_norm": 6.9375, "learning_rate": 1.0891900034962874e-05, "loss": 1.4389, "step": 19497 }, { "epoch": 48.745, "grad_norm": 5.75, "learning_rate": 1.0891093570121404e-05, "loss": 1.1685, "step": 19498 }, { "epoch": 48.7475, "grad_norm": 5.40625, "learning_rate": 1.0890287099437962e-05, "loss": 1.3533, "step": 19499 }, { "epoch": 48.75, "grad_norm": 5.9375, "learning_rate": 1.088948062291783e-05, "loss": 1.2477, "step": 19500 }, { "epoch": 48.7525, "grad_norm": 7.53125, "learning_rate": 1.08886741405663e-05, "loss": 1.2861, "step": 19501 }, { "epoch": 48.755, "grad_norm": 6.96875, "learning_rate": 1.0887867652388658e-05, "loss": 1.3333, "step": 19502 }, { "epoch": 48.7575, "grad_norm": 5.75, "learning_rate": 1.088706115839019e-05, "loss": 1.4811, "step": 19503 }, { "epoch": 48.76, "grad_norm": 7.15625, "learning_rate": 1.0886254658576186e-05, "loss": 1.2063, "step": 19504 }, { "epoch": 48.7625, "grad_norm": 9.8125, "learning_rate": 1.0885448152951931e-05, "loss": 1.2723, "step": 19505 }, { "epoch": 48.765, "grad_norm": 5.78125, "learning_rate": 1.0884641641522715e-05, "loss": 1.281, "step": 19506 }, { "epoch": 48.7675, "grad_norm": 7.8125, "learning_rate": 1.0883835124293823e-05, "loss": 1.2503, "step": 19507 }, { "epoch": 48.77, "grad_norm": 5.71875, "learning_rate": 1.0883028601270545e-05, "loss": 1.3171, "step": 19508 }, { "epoch": 48.7725, "grad_norm": 11.1875, "learning_rate": 1.0882222072458162e-05, "loss": 1.4308, "step": 19509 }, { "epoch": 48.775, "grad_norm": 6.96875, "learning_rate": 1.088141553786197e-05, "loss": 1.338, "step": 19510 }, { "epoch": 48.7775, "grad_norm": 9.125, "learning_rate": 1.088060899748725e-05, "loss": 1.2872, "step": 19511 }, { "epoch": 48.78, "grad_norm": 5.75, "learning_rate": 1.0879802451339298e-05, "loss": 1.2929, "step": 19512 }, { "epoch": 48.7825, "grad_norm": 7.28125, "learning_rate": 1.0878995899423392e-05, "loss": 1.322, "step": 19513 }, { "epoch": 48.785, "grad_norm": 11.125, "learning_rate": 1.0878189341744826e-05, "loss": 1.3608, "step": 19514 }, { "epoch": 48.7875, "grad_norm": 8.875, "learning_rate": 1.0877382778308884e-05, "loss": 1.3839, "step": 19515 }, { "epoch": 48.79, "grad_norm": 9.3125, "learning_rate": 1.0876576209120857e-05, "loss": 1.3802, "step": 19516 }, { "epoch": 48.7925, "grad_norm": 6.625, "learning_rate": 1.0875769634186031e-05, "loss": 1.377, "step": 19517 }, { "epoch": 48.795, "grad_norm": 8.3125, "learning_rate": 1.0874963053509694e-05, "loss": 1.3562, "step": 19518 }, { "epoch": 48.7975, "grad_norm": 10.6875, "learning_rate": 1.0874156467097135e-05, "loss": 1.3168, "step": 19519 }, { "epoch": 48.8, "grad_norm": 10.6875, "learning_rate": 1.087334987495364e-05, "loss": 1.2672, "step": 19520 }, { "epoch": 48.8025, "grad_norm": 5.59375, "learning_rate": 1.08725432770845e-05, "loss": 1.4461, "step": 19521 }, { "epoch": 48.805, "grad_norm": 5.96875, "learning_rate": 1.0871736673494998e-05, "loss": 1.2837, "step": 19522 }, { "epoch": 48.8075, "grad_norm": 7.625, "learning_rate": 1.0870930064190432e-05, "loss": 1.2046, "step": 19523 }, { "epoch": 48.81, "grad_norm": 7.375, "learning_rate": 1.0870123449176075e-05, "loss": 1.3426, "step": 19524 }, { "epoch": 48.8125, "grad_norm": 8.4375, "learning_rate": 1.086931682845723e-05, "loss": 1.3338, "step": 19525 }, { "epoch": 48.815, "grad_norm": 7.21875, "learning_rate": 1.0868510202039176e-05, "loss": 1.2592, "step": 19526 }, { "epoch": 48.8175, "grad_norm": 6.6875, "learning_rate": 1.0867703569927204e-05, "loss": 1.2864, "step": 19527 }, { "epoch": 48.82, "grad_norm": 8.125, "learning_rate": 1.0866896932126602e-05, "loss": 1.3772, "step": 19528 }, { "epoch": 48.8225, "grad_norm": 6.3125, "learning_rate": 1.0866090288642659e-05, "loss": 1.3774, "step": 19529 }, { "epoch": 48.825, "grad_norm": 8.1875, "learning_rate": 1.0865283639480662e-05, "loss": 1.5053, "step": 19530 }, { "epoch": 48.8275, "grad_norm": 5.9375, "learning_rate": 1.08644769846459e-05, "loss": 1.2664, "step": 19531 }, { "epoch": 48.83, "grad_norm": 29.0, "learning_rate": 1.086367032414366e-05, "loss": 1.4005, "step": 19532 }, { "epoch": 48.8325, "grad_norm": 6.0, "learning_rate": 1.0862863657979237e-05, "loss": 1.2633, "step": 19533 }, { "epoch": 48.835, "grad_norm": 5.8125, "learning_rate": 1.0862056986157909e-05, "loss": 1.1827, "step": 19534 }, { "epoch": 48.8375, "grad_norm": 8.8125, "learning_rate": 1.0861250308684975e-05, "loss": 1.3587, "step": 19535 }, { "epoch": 48.84, "grad_norm": 8.125, "learning_rate": 1.0860443625565712e-05, "loss": 1.5035, "step": 19536 }, { "epoch": 48.8425, "grad_norm": 6.59375, "learning_rate": 1.085963693680542e-05, "loss": 1.3925, "step": 19537 }, { "epoch": 48.845, "grad_norm": 8.6875, "learning_rate": 1.0858830242409382e-05, "loss": 1.2909, "step": 19538 }, { "epoch": 48.8475, "grad_norm": 5.9375, "learning_rate": 1.0858023542382885e-05, "loss": 1.3029, "step": 19539 }, { "epoch": 48.85, "grad_norm": 7.6875, "learning_rate": 1.0857216836731221e-05, "loss": 1.3016, "step": 19540 }, { "epoch": 48.8525, "grad_norm": 6.03125, "learning_rate": 1.0856410125459675e-05, "loss": 1.517, "step": 19541 }, { "epoch": 48.855, "grad_norm": 6.53125, "learning_rate": 1.0855603408573545e-05, "loss": 1.3217, "step": 19542 }, { "epoch": 48.8575, "grad_norm": 6.125, "learning_rate": 1.0854796686078109e-05, "loss": 1.2503, "step": 19543 }, { "epoch": 48.86, "grad_norm": 7.78125, "learning_rate": 1.085398995797866e-05, "loss": 1.3395, "step": 19544 }, { "epoch": 48.8625, "grad_norm": 8.0, "learning_rate": 1.0853183224280487e-05, "loss": 1.3056, "step": 19545 }, { "epoch": 48.865, "grad_norm": 7.34375, "learning_rate": 1.0852376484988882e-05, "loss": 1.2568, "step": 19546 }, { "epoch": 48.8675, "grad_norm": 8.9375, "learning_rate": 1.0851569740109129e-05, "loss": 1.4709, "step": 19547 }, { "epoch": 48.87, "grad_norm": 8.1875, "learning_rate": 1.085076298964652e-05, "loss": 1.3528, "step": 19548 }, { "epoch": 48.8725, "grad_norm": 11.25, "learning_rate": 1.084995623360634e-05, "loss": 1.3632, "step": 19549 }, { "epoch": 48.875, "grad_norm": 6.78125, "learning_rate": 1.0849149471993883e-05, "loss": 1.2314, "step": 19550 }, { "epoch": 48.8775, "grad_norm": 5.03125, "learning_rate": 1.0848342704814434e-05, "loss": 1.2785, "step": 19551 }, { "epoch": 48.88, "grad_norm": 8.125, "learning_rate": 1.0847535932073288e-05, "loss": 1.3983, "step": 19552 }, { "epoch": 48.8825, "grad_norm": 26.875, "learning_rate": 1.0846729153775728e-05, "loss": 1.2506, "step": 19553 }, { "epoch": 48.885, "grad_norm": 8.1875, "learning_rate": 1.0845922369927044e-05, "loss": 1.2783, "step": 19554 }, { "epoch": 48.8875, "grad_norm": 9.3125, "learning_rate": 1.084511558053253e-05, "loss": 1.3837, "step": 19555 }, { "epoch": 48.89, "grad_norm": 7.6875, "learning_rate": 1.084430878559747e-05, "loss": 1.14, "step": 19556 }, { "epoch": 48.8925, "grad_norm": 6.0, "learning_rate": 1.0843501985127156e-05, "loss": 1.264, "step": 19557 }, { "epoch": 48.895, "grad_norm": 10.375, "learning_rate": 1.0842695179126877e-05, "loss": 1.4259, "step": 19558 }, { "epoch": 48.8975, "grad_norm": 7.5, "learning_rate": 1.084188836760192e-05, "loss": 1.3926, "step": 19559 }, { "epoch": 48.9, "grad_norm": 6.4375, "learning_rate": 1.0841081550557577e-05, "loss": 1.3417, "step": 19560 }, { "epoch": 48.9025, "grad_norm": 10.0625, "learning_rate": 1.084027472799914e-05, "loss": 1.3166, "step": 19561 }, { "epoch": 48.905, "grad_norm": 7.1875, "learning_rate": 1.0839467899931893e-05, "loss": 1.4184, "step": 19562 }, { "epoch": 48.9075, "grad_norm": 6.90625, "learning_rate": 1.0838661066361132e-05, "loss": 1.3022, "step": 19563 }, { "epoch": 48.91, "grad_norm": 8.375, "learning_rate": 1.0837854227292138e-05, "loss": 1.2395, "step": 19564 }, { "epoch": 48.9125, "grad_norm": 6.40625, "learning_rate": 1.0837047382730206e-05, "loss": 1.1747, "step": 19565 }, { "epoch": 48.915, "grad_norm": 4.75, "learning_rate": 1.0836240532680624e-05, "loss": 1.2206, "step": 19566 }, { "epoch": 48.9175, "grad_norm": 8.1875, "learning_rate": 1.0835433677148685e-05, "loss": 1.2955, "step": 19567 }, { "epoch": 48.92, "grad_norm": 6.5, "learning_rate": 1.0834626816139678e-05, "loss": 1.2816, "step": 19568 }, { "epoch": 48.9225, "grad_norm": 7.15625, "learning_rate": 1.0833819949658886e-05, "loss": 1.2459, "step": 19569 }, { "epoch": 48.925, "grad_norm": 6.53125, "learning_rate": 1.0833013077711607e-05, "loss": 1.3858, "step": 19570 }, { "epoch": 48.9275, "grad_norm": 8.25, "learning_rate": 1.0832206200303127e-05, "loss": 1.4381, "step": 19571 }, { "epoch": 48.93, "grad_norm": 10.25, "learning_rate": 1.0831399317438736e-05, "loss": 1.2687, "step": 19572 }, { "epoch": 48.9325, "grad_norm": 6.9375, "learning_rate": 1.0830592429123722e-05, "loss": 1.3307, "step": 19573 }, { "epoch": 48.935, "grad_norm": 7.125, "learning_rate": 1.082978553536338e-05, "loss": 1.3158, "step": 19574 }, { "epoch": 48.9375, "grad_norm": 7.96875, "learning_rate": 1.0828978636162996e-05, "loss": 1.3806, "step": 19575 }, { "epoch": 48.94, "grad_norm": 7.8125, "learning_rate": 1.0828171731527863e-05, "loss": 1.1668, "step": 19576 }, { "epoch": 48.9425, "grad_norm": 9.5, "learning_rate": 1.082736482146327e-05, "loss": 1.3007, "step": 19577 }, { "epoch": 48.945, "grad_norm": 7.84375, "learning_rate": 1.0826557905974504e-05, "loss": 1.3571, "step": 19578 }, { "epoch": 48.9475, "grad_norm": 7.75, "learning_rate": 1.0825750985066858e-05, "loss": 1.3489, "step": 19579 }, { "epoch": 48.95, "grad_norm": 6.1875, "learning_rate": 1.0824944058745623e-05, "loss": 1.5051, "step": 19580 }, { "epoch": 48.9525, "grad_norm": 8.9375, "learning_rate": 1.0824137127016085e-05, "loss": 1.2204, "step": 19581 }, { "epoch": 48.955, "grad_norm": 5.75, "learning_rate": 1.082333018988354e-05, "loss": 1.2667, "step": 19582 }, { "epoch": 48.9575, "grad_norm": 6.46875, "learning_rate": 1.0822523247353271e-05, "loss": 1.262, "step": 19583 }, { "epoch": 48.96, "grad_norm": 10.6875, "learning_rate": 1.0821716299430577e-05, "loss": 1.3519, "step": 19584 }, { "epoch": 48.9625, "grad_norm": 9.375, "learning_rate": 1.0820909346120742e-05, "loss": 1.3745, "step": 19585 }, { "epoch": 48.965, "grad_norm": 6.3125, "learning_rate": 1.0820102387429059e-05, "loss": 1.3941, "step": 19586 }, { "epoch": 48.9675, "grad_norm": 6.09375, "learning_rate": 1.0819295423360816e-05, "loss": 1.2948, "step": 19587 }, { "epoch": 48.97, "grad_norm": 8.3125, "learning_rate": 1.0818488453921308e-05, "loss": 1.2445, "step": 19588 }, { "epoch": 48.9725, "grad_norm": 7.9375, "learning_rate": 1.0817681479115817e-05, "loss": 1.4045, "step": 19589 }, { "epoch": 48.975, "grad_norm": 8.0, "learning_rate": 1.0816874498949644e-05, "loss": 1.3681, "step": 19590 }, { "epoch": 48.9775, "grad_norm": 8.375, "learning_rate": 1.0816067513428072e-05, "loss": 1.3266, "step": 19591 }, { "epoch": 48.98, "grad_norm": 8.0, "learning_rate": 1.0815260522556394e-05, "loss": 1.3344, "step": 19592 }, { "epoch": 48.9825, "grad_norm": 8.1875, "learning_rate": 1.0814453526339903e-05, "loss": 1.3389, "step": 19593 }, { "epoch": 48.985, "grad_norm": 7.46875, "learning_rate": 1.0813646524783886e-05, "loss": 1.1331, "step": 19594 }, { "epoch": 48.9875, "grad_norm": 7.25, "learning_rate": 1.0812839517893636e-05, "loss": 1.1155, "step": 19595 }, { "epoch": 48.99, "grad_norm": 7.40625, "learning_rate": 1.081203250567444e-05, "loss": 1.1971, "step": 19596 }, { "epoch": 48.9925, "grad_norm": 7.8125, "learning_rate": 1.0811225488131593e-05, "loss": 1.2477, "step": 19597 }, { "epoch": 48.995, "grad_norm": 8.4375, "learning_rate": 1.0810418465270383e-05, "loss": 1.3982, "step": 19598 }, { "epoch": 48.9975, "grad_norm": 8.4375, "learning_rate": 1.0809611437096103e-05, "loss": 1.2917, "step": 19599 }, { "epoch": 49.0, "grad_norm": 7.0625, "learning_rate": 1.0808804403614044e-05, "loss": 1.3649, "step": 19600 }, { "epoch": 49.0025, "grad_norm": 8.0625, "learning_rate": 1.0807997364829495e-05, "loss": 1.3742, "step": 19601 }, { "epoch": 49.005, "grad_norm": 5.65625, "learning_rate": 1.0807190320747745e-05, "loss": 1.4939, "step": 19602 }, { "epoch": 49.0075, "grad_norm": 6.96875, "learning_rate": 1.080638327137409e-05, "loss": 1.2025, "step": 19603 }, { "epoch": 49.01, "grad_norm": 8.5625, "learning_rate": 1.0805576216713818e-05, "loss": 1.3614, "step": 19604 }, { "epoch": 49.0125, "grad_norm": 8.125, "learning_rate": 1.0804769156772221e-05, "loss": 1.3424, "step": 19605 }, { "epoch": 49.015, "grad_norm": 8.0625, "learning_rate": 1.0803962091554588e-05, "loss": 1.2846, "step": 19606 }, { "epoch": 49.0175, "grad_norm": 8.0625, "learning_rate": 1.0803155021066215e-05, "loss": 1.4799, "step": 19607 }, { "epoch": 49.02, "grad_norm": 6.3125, "learning_rate": 1.0802347945312386e-05, "loss": 1.3333, "step": 19608 }, { "epoch": 49.0225, "grad_norm": 9.4375, "learning_rate": 1.0801540864298399e-05, "loss": 1.4607, "step": 19609 }, { "epoch": 49.025, "grad_norm": 7.25, "learning_rate": 1.0800733778029542e-05, "loss": 1.3285, "step": 19610 }, { "epoch": 49.0275, "grad_norm": 7.90625, "learning_rate": 1.0799926686511102e-05, "loss": 1.2908, "step": 19611 }, { "epoch": 49.03, "grad_norm": 5.625, "learning_rate": 1.0799119589748378e-05, "loss": 1.3786, "step": 19612 }, { "epoch": 49.0325, "grad_norm": 10.3125, "learning_rate": 1.0798312487746658e-05, "loss": 1.2267, "step": 19613 }, { "epoch": 49.035, "grad_norm": 8.9375, "learning_rate": 1.0797505380511235e-05, "loss": 1.3784, "step": 19614 }, { "epoch": 49.0375, "grad_norm": 9.9375, "learning_rate": 1.0796698268047397e-05, "loss": 1.3326, "step": 19615 }, { "epoch": 49.04, "grad_norm": 6.53125, "learning_rate": 1.0795891150360435e-05, "loss": 1.2159, "step": 19616 }, { "epoch": 49.0425, "grad_norm": 7.4375, "learning_rate": 1.0795084027455645e-05, "loss": 1.3637, "step": 19617 }, { "epoch": 49.045, "grad_norm": 7.03125, "learning_rate": 1.0794276899338315e-05, "loss": 1.2667, "step": 19618 }, { "epoch": 49.0475, "grad_norm": 7.375, "learning_rate": 1.0793469766013737e-05, "loss": 1.3541, "step": 19619 }, { "epoch": 49.05, "grad_norm": 5.9375, "learning_rate": 1.0792662627487207e-05, "loss": 1.2798, "step": 19620 }, { "epoch": 49.0525, "grad_norm": 6.53125, "learning_rate": 1.0791855483764007e-05, "loss": 1.4056, "step": 19621 }, { "epoch": 49.055, "grad_norm": 7.21875, "learning_rate": 1.0791048334849436e-05, "loss": 1.3161, "step": 19622 }, { "epoch": 49.0575, "grad_norm": 7.84375, "learning_rate": 1.0790241180748783e-05, "loss": 1.3804, "step": 19623 }, { "epoch": 49.06, "grad_norm": 6.96875, "learning_rate": 1.0789434021467344e-05, "loss": 1.3657, "step": 19624 }, { "epoch": 49.0625, "grad_norm": 6.65625, "learning_rate": 1.0788626857010404e-05, "loss": 1.2339, "step": 19625 }, { "epoch": 49.065, "grad_norm": 4.3125, "learning_rate": 1.0787819687383258e-05, "loss": 1.2968, "step": 19626 }, { "epoch": 49.0675, "grad_norm": 9.4375, "learning_rate": 1.07870125125912e-05, "loss": 1.2913, "step": 19627 }, { "epoch": 49.07, "grad_norm": 4.5, "learning_rate": 1.0786205332639517e-05, "loss": 1.3808, "step": 19628 }, { "epoch": 49.0725, "grad_norm": 12.5625, "learning_rate": 1.0785398147533503e-05, "loss": 1.3849, "step": 19629 }, { "epoch": 49.075, "grad_norm": 7.03125, "learning_rate": 1.0784590957278452e-05, "loss": 1.5138, "step": 19630 }, { "epoch": 49.0775, "grad_norm": 6.34375, "learning_rate": 1.0783783761879651e-05, "loss": 1.4788, "step": 19631 }, { "epoch": 49.08, "grad_norm": 9.0, "learning_rate": 1.0782976561342398e-05, "loss": 1.4342, "step": 19632 }, { "epoch": 49.0825, "grad_norm": 9.8125, "learning_rate": 1.078216935567198e-05, "loss": 1.3526, "step": 19633 }, { "epoch": 49.085, "grad_norm": 7.28125, "learning_rate": 1.0781362144873692e-05, "loss": 1.2811, "step": 19634 }, { "epoch": 49.0875, "grad_norm": 7.125, "learning_rate": 1.078055492895283e-05, "loss": 1.3589, "step": 19635 }, { "epoch": 49.09, "grad_norm": 6.96875, "learning_rate": 1.0779747707914672e-05, "loss": 1.3735, "step": 19636 }, { "epoch": 49.0925, "grad_norm": 6.4375, "learning_rate": 1.0778940481764526e-05, "loss": 1.3208, "step": 19637 }, { "epoch": 49.095, "grad_norm": 6.3125, "learning_rate": 1.0778133250507671e-05, "loss": 1.1908, "step": 19638 }, { "epoch": 49.0975, "grad_norm": 6.5625, "learning_rate": 1.0777326014149415e-05, "loss": 1.3025, "step": 19639 }, { "epoch": 49.1, "grad_norm": 14.25, "learning_rate": 1.0776518772695035e-05, "loss": 1.2041, "step": 19640 }, { "epoch": 49.1025, "grad_norm": 8.5, "learning_rate": 1.0775711526149829e-05, "loss": 1.3939, "step": 19641 }, { "epoch": 49.105, "grad_norm": 5.625, "learning_rate": 1.0774904274519089e-05, "loss": 1.1483, "step": 19642 }, { "epoch": 49.1075, "grad_norm": 8.5625, "learning_rate": 1.0774097017808108e-05, "loss": 1.3896, "step": 19643 }, { "epoch": 49.11, "grad_norm": 6.9375, "learning_rate": 1.0773289756022178e-05, "loss": 1.3989, "step": 19644 }, { "epoch": 49.1125, "grad_norm": 8.6875, "learning_rate": 1.0772482489166592e-05, "loss": 1.1962, "step": 19645 }, { "epoch": 49.115, "grad_norm": 10.625, "learning_rate": 1.077167521724664e-05, "loss": 1.4087, "step": 19646 }, { "epoch": 49.1175, "grad_norm": 7.09375, "learning_rate": 1.0770867940267617e-05, "loss": 1.3933, "step": 19647 }, { "epoch": 49.12, "grad_norm": 13.625, "learning_rate": 1.0770060658234815e-05, "loss": 1.3627, "step": 19648 }, { "epoch": 49.1225, "grad_norm": 8.25, "learning_rate": 1.0769253371153525e-05, "loss": 1.2557, "step": 19649 }, { "epoch": 49.125, "grad_norm": 10.625, "learning_rate": 1.0768446079029044e-05, "loss": 1.3111, "step": 19650 }, { "epoch": 49.1275, "grad_norm": 6.75, "learning_rate": 1.0767638781866656e-05, "loss": 1.3933, "step": 19651 }, { "epoch": 49.13, "grad_norm": 9.1875, "learning_rate": 1.0766831479671664e-05, "loss": 1.5438, "step": 19652 }, { "epoch": 49.1325, "grad_norm": 7.96875, "learning_rate": 1.0766024172449353e-05, "loss": 1.3504, "step": 19653 }, { "epoch": 49.135, "grad_norm": 10.5, "learning_rate": 1.076521686020502e-05, "loss": 1.352, "step": 19654 }, { "epoch": 49.1375, "grad_norm": 6.59375, "learning_rate": 1.0764409542943955e-05, "loss": 1.2194, "step": 19655 }, { "epoch": 49.14, "grad_norm": 6.90625, "learning_rate": 1.076360222067145e-05, "loss": 1.3426, "step": 19656 }, { "epoch": 49.1425, "grad_norm": 6.25, "learning_rate": 1.07627948933928e-05, "loss": 1.329, "step": 19657 }, { "epoch": 49.145, "grad_norm": 8.3125, "learning_rate": 1.0761987561113299e-05, "loss": 1.3936, "step": 19658 }, { "epoch": 49.1475, "grad_norm": 6.375, "learning_rate": 1.0761180223838234e-05, "loss": 1.2926, "step": 19659 }, { "epoch": 49.15, "grad_norm": 5.8125, "learning_rate": 1.0760372881572904e-05, "loss": 1.2163, "step": 19660 }, { "epoch": 49.1525, "grad_norm": 6.53125, "learning_rate": 1.0759565534322602e-05, "loss": 1.3074, "step": 19661 }, { "epoch": 49.155, "grad_norm": 6.8125, "learning_rate": 1.0758758182092617e-05, "loss": 1.2704, "step": 19662 }, { "epoch": 49.1575, "grad_norm": 5.75, "learning_rate": 1.0757950824888244e-05, "loss": 1.3821, "step": 19663 }, { "epoch": 49.16, "grad_norm": 6.8125, "learning_rate": 1.0757143462714777e-05, "loss": 1.2356, "step": 19664 }, { "epoch": 49.1625, "grad_norm": 8.9375, "learning_rate": 1.0756336095577506e-05, "loss": 1.4224, "step": 19665 }, { "epoch": 49.165, "grad_norm": 7.46875, "learning_rate": 1.0755528723481725e-05, "loss": 1.2949, "step": 19666 }, { "epoch": 49.1675, "grad_norm": 6.5, "learning_rate": 1.0754721346432731e-05, "loss": 1.2948, "step": 19667 }, { "epoch": 49.17, "grad_norm": 7.03125, "learning_rate": 1.075391396443581e-05, "loss": 1.4702, "step": 19668 }, { "epoch": 49.1725, "grad_norm": 7.75, "learning_rate": 1.0753106577496264e-05, "loss": 1.2222, "step": 19669 }, { "epoch": 49.175, "grad_norm": 7.6875, "learning_rate": 1.0752299185619377e-05, "loss": 1.3158, "step": 19670 }, { "epoch": 49.1775, "grad_norm": 9.5625, "learning_rate": 1.075149178881045e-05, "loss": 1.5117, "step": 19671 }, { "epoch": 49.18, "grad_norm": 7.375, "learning_rate": 1.075068438707477e-05, "loss": 1.2826, "step": 19672 }, { "epoch": 49.1825, "grad_norm": 7.3125, "learning_rate": 1.0749876980417636e-05, "loss": 1.3317, "step": 19673 }, { "epoch": 49.185, "grad_norm": 6.5, "learning_rate": 1.0749069568844338e-05, "loss": 1.3493, "step": 19674 }, { "epoch": 49.1875, "grad_norm": 6.625, "learning_rate": 1.074826215236017e-05, "loss": 1.4146, "step": 19675 }, { "epoch": 49.19, "grad_norm": 7.46875, "learning_rate": 1.0747454730970422e-05, "loss": 1.3328, "step": 19676 }, { "epoch": 49.1925, "grad_norm": 6.5625, "learning_rate": 1.0746647304680394e-05, "loss": 1.292, "step": 19677 }, { "epoch": 49.195, "grad_norm": 6.5625, "learning_rate": 1.0745839873495371e-05, "loss": 1.2686, "step": 19678 }, { "epoch": 49.1975, "grad_norm": 5.78125, "learning_rate": 1.0745032437420657e-05, "loss": 1.4338, "step": 19679 }, { "epoch": 49.2, "grad_norm": 7.625, "learning_rate": 1.0744224996461541e-05, "loss": 1.2914, "step": 19680 }, { "epoch": 49.2025, "grad_norm": 9.25, "learning_rate": 1.0743417550623311e-05, "loss": 1.1894, "step": 19681 }, { "epoch": 49.205, "grad_norm": 7.84375, "learning_rate": 1.074261009991127e-05, "loss": 1.4067, "step": 19682 }, { "epoch": 49.2075, "grad_norm": 5.59375, "learning_rate": 1.0741802644330704e-05, "loss": 1.4241, "step": 19683 }, { "epoch": 49.21, "grad_norm": 8.375, "learning_rate": 1.0740995183886911e-05, "loss": 1.1814, "step": 19684 }, { "epoch": 49.2125, "grad_norm": 8.8125, "learning_rate": 1.0740187718585182e-05, "loss": 1.2372, "step": 19685 }, { "epoch": 49.215, "grad_norm": 7.21875, "learning_rate": 1.0739380248430812e-05, "loss": 1.2938, "step": 19686 }, { "epoch": 49.2175, "grad_norm": 7.28125, "learning_rate": 1.0738572773429097e-05, "loss": 1.4897, "step": 19687 }, { "epoch": 49.22, "grad_norm": 9.75, "learning_rate": 1.0737765293585324e-05, "loss": 1.4136, "step": 19688 }, { "epoch": 49.2225, "grad_norm": 9.375, "learning_rate": 1.0736957808904796e-05, "loss": 1.2317, "step": 19689 }, { "epoch": 49.225, "grad_norm": 8.375, "learning_rate": 1.0736150319392801e-05, "loss": 1.3083, "step": 19690 }, { "epoch": 49.2275, "grad_norm": 8.375, "learning_rate": 1.0735342825054631e-05, "loss": 1.4984, "step": 19691 }, { "epoch": 49.23, "grad_norm": 7.5625, "learning_rate": 1.0734535325895587e-05, "loss": 1.4373, "step": 19692 }, { "epoch": 49.2325, "grad_norm": 10.875, "learning_rate": 1.0733727821920954e-05, "loss": 1.2079, "step": 19693 }, { "epoch": 49.235, "grad_norm": 9.9375, "learning_rate": 1.0732920313136036e-05, "loss": 1.339, "step": 19694 }, { "epoch": 49.2375, "grad_norm": 9.9375, "learning_rate": 1.0732112799546117e-05, "loss": 1.2402, "step": 19695 }, { "epoch": 49.24, "grad_norm": 7.59375, "learning_rate": 1.0731305281156499e-05, "loss": 1.5365, "step": 19696 }, { "epoch": 49.2425, "grad_norm": 7.4375, "learning_rate": 1.0730497757972471e-05, "loss": 1.367, "step": 19697 }, { "epoch": 49.245, "grad_norm": 7.46875, "learning_rate": 1.072969022999933e-05, "loss": 1.3457, "step": 19698 }, { "epoch": 49.2475, "grad_norm": 7.3125, "learning_rate": 1.072888269724237e-05, "loss": 1.345, "step": 19699 }, { "epoch": 49.25, "grad_norm": 7.75, "learning_rate": 1.0728075159706881e-05, "loss": 1.3681, "step": 19700 }, { "epoch": 49.2525, "grad_norm": 9.1875, "learning_rate": 1.0727267617398163e-05, "loss": 1.3951, "step": 19701 }, { "epoch": 49.255, "grad_norm": 7.5625, "learning_rate": 1.0726460070321507e-05, "loss": 1.3676, "step": 19702 }, { "epoch": 49.2575, "grad_norm": 12.3125, "learning_rate": 1.0725652518482208e-05, "loss": 1.2975, "step": 19703 }, { "epoch": 49.26, "grad_norm": 8.25, "learning_rate": 1.072484496188556e-05, "loss": 1.3077, "step": 19704 }, { "epoch": 49.2625, "grad_norm": 6.71875, "learning_rate": 1.0724037400536858e-05, "loss": 1.2146, "step": 19705 }, { "epoch": 49.265, "grad_norm": 6.65625, "learning_rate": 1.0723229834441393e-05, "loss": 1.2098, "step": 19706 }, { "epoch": 49.2675, "grad_norm": 5.75, "learning_rate": 1.0722422263604466e-05, "loss": 1.3446, "step": 19707 }, { "epoch": 49.27, "grad_norm": 6.78125, "learning_rate": 1.0721614688031366e-05, "loss": 1.4172, "step": 19708 }, { "epoch": 49.2725, "grad_norm": 5.71875, "learning_rate": 1.0720807107727388e-05, "loss": 1.1747, "step": 19709 }, { "epoch": 49.275, "grad_norm": 8.8125, "learning_rate": 1.0719999522697828e-05, "loss": 1.295, "step": 19710 }, { "epoch": 49.2775, "grad_norm": 8.375, "learning_rate": 1.0719191932947981e-05, "loss": 1.2745, "step": 19711 }, { "epoch": 49.28, "grad_norm": 9.375, "learning_rate": 1.0718384338483141e-05, "loss": 1.4094, "step": 19712 }, { "epoch": 49.2825, "grad_norm": 6.3125, "learning_rate": 1.07175767393086e-05, "loss": 1.4452, "step": 19713 }, { "epoch": 49.285, "grad_norm": 6.09375, "learning_rate": 1.0716769135429654e-05, "loss": 1.2254, "step": 19714 }, { "epoch": 49.2875, "grad_norm": 7.0625, "learning_rate": 1.0715961526851602e-05, "loss": 1.4711, "step": 19715 }, { "epoch": 49.29, "grad_norm": 5.46875, "learning_rate": 1.071515391357973e-05, "loss": 1.2848, "step": 19716 }, { "epoch": 49.2925, "grad_norm": 6.46875, "learning_rate": 1.071434629561934e-05, "loss": 1.4052, "step": 19717 }, { "epoch": 49.295, "grad_norm": 20.75, "learning_rate": 1.0713538672975726e-05, "loss": 1.3587, "step": 19718 }, { "epoch": 49.2975, "grad_norm": 8.125, "learning_rate": 1.071273104565418e-05, "loss": 1.2238, "step": 19719 }, { "epoch": 49.3, "grad_norm": 8.75, "learning_rate": 1.0711923413659995e-05, "loss": 1.1137, "step": 19720 }, { "epoch": 49.3025, "grad_norm": 7.9375, "learning_rate": 1.071111577699847e-05, "loss": 1.2738, "step": 19721 }, { "epoch": 49.305, "grad_norm": 8.5, "learning_rate": 1.07103081356749e-05, "loss": 1.2315, "step": 19722 }, { "epoch": 49.3075, "grad_norm": 10.0, "learning_rate": 1.0709500489694578e-05, "loss": 1.3885, "step": 19723 }, { "epoch": 49.31, "grad_norm": 7.28125, "learning_rate": 1.07086928390628e-05, "loss": 1.3334, "step": 19724 }, { "epoch": 49.3125, "grad_norm": 6.84375, "learning_rate": 1.0707885183784857e-05, "loss": 1.2712, "step": 19725 }, { "epoch": 49.315, "grad_norm": 9.1875, "learning_rate": 1.0707077523866049e-05, "loss": 1.3502, "step": 19726 }, { "epoch": 49.3175, "grad_norm": 4.90625, "learning_rate": 1.0706269859311669e-05, "loss": 1.3485, "step": 19727 }, { "epoch": 49.32, "grad_norm": 6.4375, "learning_rate": 1.0705462190127011e-05, "loss": 1.402, "step": 19728 }, { "epoch": 49.3225, "grad_norm": 9.75, "learning_rate": 1.0704654516317371e-05, "loss": 1.3682, "step": 19729 }, { "epoch": 49.325, "grad_norm": 10.75, "learning_rate": 1.0703846837888046e-05, "loss": 1.3053, "step": 19730 }, { "epoch": 49.3275, "grad_norm": 5.65625, "learning_rate": 1.0703039154844329e-05, "loss": 1.2504, "step": 19731 }, { "epoch": 49.33, "grad_norm": 5.0625, "learning_rate": 1.0702231467191516e-05, "loss": 1.4227, "step": 19732 }, { "epoch": 49.3325, "grad_norm": 7.65625, "learning_rate": 1.0701423774934899e-05, "loss": 1.2717, "step": 19733 }, { "epoch": 49.335, "grad_norm": 5.6875, "learning_rate": 1.0700616078079778e-05, "loss": 1.2212, "step": 19734 }, { "epoch": 49.3375, "grad_norm": 8.0, "learning_rate": 1.0699808376631444e-05, "loss": 1.3155, "step": 19735 }, { "epoch": 49.34, "grad_norm": 9.0625, "learning_rate": 1.0699000670595194e-05, "loss": 1.3007, "step": 19736 }, { "epoch": 49.3425, "grad_norm": 7.3125, "learning_rate": 1.0698192959976327e-05, "loss": 1.1763, "step": 19737 }, { "epoch": 49.345, "grad_norm": 7.40625, "learning_rate": 1.0697385244780132e-05, "loss": 1.1941, "step": 19738 }, { "epoch": 49.3475, "grad_norm": 15.25, "learning_rate": 1.0696577525011911e-05, "loss": 1.277, "step": 19739 }, { "epoch": 49.35, "grad_norm": 8.5, "learning_rate": 1.069576980067695e-05, "loss": 1.2426, "step": 19740 }, { "epoch": 49.3525, "grad_norm": 8.8125, "learning_rate": 1.0694962071780555e-05, "loss": 1.3354, "step": 19741 }, { "epoch": 49.355, "grad_norm": 9.1875, "learning_rate": 1.0694154338328016e-05, "loss": 1.3491, "step": 19742 }, { "epoch": 49.3575, "grad_norm": 7.09375, "learning_rate": 1.0693346600324626e-05, "loss": 1.3982, "step": 19743 }, { "epoch": 49.36, "grad_norm": 7.78125, "learning_rate": 1.0692538857775685e-05, "loss": 1.3086, "step": 19744 }, { "epoch": 49.3625, "grad_norm": 10.0, "learning_rate": 1.0691731110686486e-05, "loss": 1.2361, "step": 19745 }, { "epoch": 49.365, "grad_norm": 8.6875, "learning_rate": 1.0690923359062326e-05, "loss": 1.2468, "step": 19746 }, { "epoch": 49.3675, "grad_norm": 7.875, "learning_rate": 1.0690115602908501e-05, "loss": 1.3777, "step": 19747 }, { "epoch": 49.37, "grad_norm": 8.1875, "learning_rate": 1.0689307842230305e-05, "loss": 1.2971, "step": 19748 }, { "epoch": 49.3725, "grad_norm": 6.8125, "learning_rate": 1.0688500077033033e-05, "loss": 1.3509, "step": 19749 }, { "epoch": 49.375, "grad_norm": 10.0625, "learning_rate": 1.0687692307321984e-05, "loss": 1.3489, "step": 19750 }, { "epoch": 49.3775, "grad_norm": 9.75, "learning_rate": 1.0686884533102451e-05, "loss": 1.385, "step": 19751 }, { "epoch": 49.38, "grad_norm": 10.125, "learning_rate": 1.0686076754379734e-05, "loss": 1.3313, "step": 19752 }, { "epoch": 49.3825, "grad_norm": 9.625, "learning_rate": 1.0685268971159119e-05, "loss": 1.2406, "step": 19753 }, { "epoch": 49.385, "grad_norm": 4.71875, "learning_rate": 1.0684461183445914e-05, "loss": 1.4714, "step": 19754 }, { "epoch": 49.3875, "grad_norm": 5.59375, "learning_rate": 1.0683653391245403e-05, "loss": 1.3626, "step": 19755 }, { "epoch": 49.39, "grad_norm": 13.125, "learning_rate": 1.0682845594562892e-05, "loss": 1.3915, "step": 19756 }, { "epoch": 49.3925, "grad_norm": 7.4375, "learning_rate": 1.068203779340367e-05, "loss": 1.3901, "step": 19757 }, { "epoch": 49.395, "grad_norm": 9.0, "learning_rate": 1.0681229987773038e-05, "loss": 1.378, "step": 19758 }, { "epoch": 49.3975, "grad_norm": 13.625, "learning_rate": 1.0680422177676288e-05, "loss": 1.091, "step": 19759 }, { "epoch": 49.4, "grad_norm": 11.1875, "learning_rate": 1.0679614363118718e-05, "loss": 1.4575, "step": 19760 }, { "epoch": 49.4025, "grad_norm": 9.0, "learning_rate": 1.0678806544105622e-05, "loss": 1.3578, "step": 19761 }, { "epoch": 49.405, "grad_norm": 6.46875, "learning_rate": 1.0677998720642302e-05, "loss": 1.293, "step": 19762 }, { "epoch": 49.4075, "grad_norm": 6.625, "learning_rate": 1.0677190892734042e-05, "loss": 1.232, "step": 19763 }, { "epoch": 49.41, "grad_norm": 7.46875, "learning_rate": 1.0676383060386153e-05, "loss": 1.4142, "step": 19764 }, { "epoch": 49.4125, "grad_norm": 6.75, "learning_rate": 1.0675575223603918e-05, "loss": 1.3557, "step": 19765 }, { "epoch": 49.415, "grad_norm": 6.34375, "learning_rate": 1.0674767382392644e-05, "loss": 1.2511, "step": 19766 }, { "epoch": 49.4175, "grad_norm": 8.0625, "learning_rate": 1.0673959536757619e-05, "loss": 1.339, "step": 19767 }, { "epoch": 49.42, "grad_norm": 5.875, "learning_rate": 1.0673151686704141e-05, "loss": 1.3847, "step": 19768 }, { "epoch": 49.4225, "grad_norm": 7.6875, "learning_rate": 1.067234383223751e-05, "loss": 1.26, "step": 19769 }, { "epoch": 49.425, "grad_norm": 8.75, "learning_rate": 1.0671535973363019e-05, "loss": 1.29, "step": 19770 }, { "epoch": 49.4275, "grad_norm": 7.5625, "learning_rate": 1.0670728110085964e-05, "loss": 1.5169, "step": 19771 }, { "epoch": 49.43, "grad_norm": 8.4375, "learning_rate": 1.0669920242411642e-05, "loss": 1.4221, "step": 19772 }, { "epoch": 49.4325, "grad_norm": 8.9375, "learning_rate": 1.0669112370345351e-05, "loss": 1.2656, "step": 19773 }, { "epoch": 49.435, "grad_norm": 6.5625, "learning_rate": 1.0668304493892387e-05, "loss": 1.1141, "step": 19774 }, { "epoch": 49.4375, "grad_norm": 7.65625, "learning_rate": 1.0667496613058044e-05, "loss": 1.2894, "step": 19775 }, { "epoch": 49.44, "grad_norm": 7.125, "learning_rate": 1.066668872784762e-05, "loss": 1.3194, "step": 19776 }, { "epoch": 49.4425, "grad_norm": 7.25, "learning_rate": 1.0665880838266413e-05, "loss": 1.3041, "step": 19777 }, { "epoch": 49.445, "grad_norm": 6.875, "learning_rate": 1.0665072944319715e-05, "loss": 1.2146, "step": 19778 }, { "epoch": 49.4475, "grad_norm": 8.875, "learning_rate": 1.066426504601283e-05, "loss": 1.1979, "step": 19779 }, { "epoch": 49.45, "grad_norm": 12.625, "learning_rate": 1.0663457143351044e-05, "loss": 1.1756, "step": 19780 }, { "epoch": 49.4525, "grad_norm": 7.34375, "learning_rate": 1.0662649236339664e-05, "loss": 1.4424, "step": 19781 }, { "epoch": 49.455, "grad_norm": 8.25, "learning_rate": 1.0661841324983981e-05, "loss": 1.3025, "step": 19782 }, { "epoch": 49.4575, "grad_norm": 6.34375, "learning_rate": 1.0661033409289292e-05, "loss": 1.5325, "step": 19783 }, { "epoch": 49.46, "grad_norm": 7.21875, "learning_rate": 1.0660225489260894e-05, "loss": 1.3135, "step": 19784 }, { "epoch": 49.4625, "grad_norm": 6.34375, "learning_rate": 1.0659417564904087e-05, "loss": 1.1885, "step": 19785 }, { "epoch": 49.465, "grad_norm": 7.1875, "learning_rate": 1.0658609636224163e-05, "loss": 1.3268, "step": 19786 }, { "epoch": 49.4675, "grad_norm": 7.5625, "learning_rate": 1.0657801703226418e-05, "loss": 1.2643, "step": 19787 }, { "epoch": 49.47, "grad_norm": 5.40625, "learning_rate": 1.0656993765916153e-05, "loss": 1.2774, "step": 19788 }, { "epoch": 49.4725, "grad_norm": 8.3125, "learning_rate": 1.0656185824298663e-05, "loss": 1.307, "step": 19789 }, { "epoch": 49.475, "grad_norm": 9.3125, "learning_rate": 1.0655377878379246e-05, "loss": 1.4023, "step": 19790 }, { "epoch": 49.4775, "grad_norm": 8.75, "learning_rate": 1.06545699281632e-05, "loss": 1.3832, "step": 19791 }, { "epoch": 49.48, "grad_norm": 8.0625, "learning_rate": 1.0653761973655819e-05, "loss": 1.3328, "step": 19792 }, { "epoch": 49.4825, "grad_norm": 8.5625, "learning_rate": 1.0652954014862396e-05, "loss": 1.3896, "step": 19793 }, { "epoch": 49.485, "grad_norm": 9.875, "learning_rate": 1.0652146051788236e-05, "loss": 1.3129, "step": 19794 }, { "epoch": 49.4875, "grad_norm": 8.5, "learning_rate": 1.0651338084438632e-05, "loss": 1.2452, "step": 19795 }, { "epoch": 49.49, "grad_norm": 9.8125, "learning_rate": 1.0650530112818884e-05, "loss": 1.327, "step": 19796 }, { "epoch": 49.4925, "grad_norm": 6.71875, "learning_rate": 1.0649722136934285e-05, "loss": 1.3905, "step": 19797 }, { "epoch": 49.495, "grad_norm": 8.6875, "learning_rate": 1.0648914156790131e-05, "loss": 1.1447, "step": 19798 }, { "epoch": 49.4975, "grad_norm": 5.6875, "learning_rate": 1.0648106172391724e-05, "loss": 1.3594, "step": 19799 }, { "epoch": 49.5, "grad_norm": 8.1875, "learning_rate": 1.0647298183744359e-05, "loss": 1.2601, "step": 19800 }, { "epoch": 49.5025, "grad_norm": 9.8125, "learning_rate": 1.0646490190853333e-05, "loss": 1.422, "step": 19801 }, { "epoch": 49.505, "grad_norm": 8.125, "learning_rate": 1.0645682193723943e-05, "loss": 1.2129, "step": 19802 }, { "epoch": 49.5075, "grad_norm": 7.78125, "learning_rate": 1.0644874192361485e-05, "loss": 1.3959, "step": 19803 }, { "epoch": 49.51, "grad_norm": 6.5, "learning_rate": 1.064406618677126e-05, "loss": 1.3568, "step": 19804 }, { "epoch": 49.5125, "grad_norm": 8.375, "learning_rate": 1.0643258176958561e-05, "loss": 1.3653, "step": 19805 }, { "epoch": 49.515, "grad_norm": 12.125, "learning_rate": 1.0642450162928688e-05, "loss": 1.4593, "step": 19806 }, { "epoch": 49.5175, "grad_norm": 7.59375, "learning_rate": 1.0641642144686939e-05, "loss": 1.392, "step": 19807 }, { "epoch": 49.52, "grad_norm": 6.8125, "learning_rate": 1.0640834122238606e-05, "loss": 1.278, "step": 19808 }, { "epoch": 49.5225, "grad_norm": 6.03125, "learning_rate": 1.0640026095588994e-05, "loss": 1.1509, "step": 19809 }, { "epoch": 49.525, "grad_norm": 6.375, "learning_rate": 1.0639218064743393e-05, "loss": 1.4611, "step": 19810 }, { "epoch": 49.5275, "grad_norm": 5.25, "learning_rate": 1.0638410029707108e-05, "loss": 1.3076, "step": 19811 }, { "epoch": 49.53, "grad_norm": 7.96875, "learning_rate": 1.0637601990485426e-05, "loss": 1.2875, "step": 19812 }, { "epoch": 49.5325, "grad_norm": 5.78125, "learning_rate": 1.063679394708366e-05, "loss": 1.3264, "step": 19813 }, { "epoch": 49.535, "grad_norm": 9.5, "learning_rate": 1.063598589950709e-05, "loss": 1.3688, "step": 19814 }, { "epoch": 49.5375, "grad_norm": 6.96875, "learning_rate": 1.0635177847761026e-05, "loss": 1.3031, "step": 19815 }, { "epoch": 49.54, "grad_norm": 6.125, "learning_rate": 1.0634369791850761e-05, "loss": 1.53, "step": 19816 }, { "epoch": 49.5425, "grad_norm": 11.6875, "learning_rate": 1.0633561731781592e-05, "loss": 1.2119, "step": 19817 }, { "epoch": 49.545, "grad_norm": 6.8125, "learning_rate": 1.063275366755882e-05, "loss": 1.3329, "step": 19818 }, { "epoch": 49.5475, "grad_norm": 5.625, "learning_rate": 1.0631945599187738e-05, "loss": 1.2244, "step": 19819 }, { "epoch": 49.55, "grad_norm": 6.125, "learning_rate": 1.0631137526673647e-05, "loss": 1.1844, "step": 19820 }, { "epoch": 49.5525, "grad_norm": 6.78125, "learning_rate": 1.0630329450021844e-05, "loss": 1.4078, "step": 19821 }, { "epoch": 49.555, "grad_norm": 6.6875, "learning_rate": 1.0629521369237623e-05, "loss": 1.1568, "step": 19822 }, { "epoch": 49.5575, "grad_norm": 11.75, "learning_rate": 1.0628713284326288e-05, "loss": 1.4455, "step": 19823 }, { "epoch": 49.56, "grad_norm": 9.1875, "learning_rate": 1.0627905195293135e-05, "loss": 1.3356, "step": 19824 }, { "epoch": 49.5625, "grad_norm": 8.1875, "learning_rate": 1.0627097102143458e-05, "loss": 1.2493, "step": 19825 }, { "epoch": 49.565, "grad_norm": 6.53125, "learning_rate": 1.0626289004882562e-05, "loss": 1.3784, "step": 19826 }, { "epoch": 49.5675, "grad_norm": 7.4375, "learning_rate": 1.0625480903515732e-05, "loss": 1.2539, "step": 19827 }, { "epoch": 49.57, "grad_norm": 7.09375, "learning_rate": 1.0624672798048283e-05, "loss": 1.3466, "step": 19828 }, { "epoch": 49.5725, "grad_norm": 8.25, "learning_rate": 1.06238646884855e-05, "loss": 1.4255, "step": 19829 }, { "epoch": 49.575, "grad_norm": 7.40625, "learning_rate": 1.0623056574832687e-05, "loss": 1.3504, "step": 19830 }, { "epoch": 49.5775, "grad_norm": 7.75, "learning_rate": 1.0622248457095138e-05, "loss": 1.2522, "step": 19831 }, { "epoch": 49.58, "grad_norm": 8.3125, "learning_rate": 1.0621440335278152e-05, "loss": 1.5123, "step": 19832 }, { "epoch": 49.5825, "grad_norm": 5.5, "learning_rate": 1.062063220938703e-05, "loss": 1.2617, "step": 19833 }, { "epoch": 49.585, "grad_norm": 8.6875, "learning_rate": 1.061982407942707e-05, "loss": 1.2852, "step": 19834 }, { "epoch": 49.5875, "grad_norm": 10.25, "learning_rate": 1.0619015945403563e-05, "loss": 1.3849, "step": 19835 }, { "epoch": 49.59, "grad_norm": 8.3125, "learning_rate": 1.0618207807321817e-05, "loss": 1.2768, "step": 19836 }, { "epoch": 49.5925, "grad_norm": 6.09375, "learning_rate": 1.061739966518712e-05, "loss": 1.2212, "step": 19837 }, { "epoch": 49.595, "grad_norm": 7.3125, "learning_rate": 1.0616591519004783e-05, "loss": 1.4168, "step": 19838 }, { "epoch": 49.5975, "grad_norm": 5.90625, "learning_rate": 1.0615783368780091e-05, "loss": 1.3639, "step": 19839 }, { "epoch": 49.6, "grad_norm": 10.5625, "learning_rate": 1.061497521451835e-05, "loss": 1.3508, "step": 19840 }, { "epoch": 49.6025, "grad_norm": 7.9375, "learning_rate": 1.0614167056224855e-05, "loss": 1.2751, "step": 19841 }, { "epoch": 49.605, "grad_norm": 7.71875, "learning_rate": 1.0613358893904905e-05, "loss": 1.4407, "step": 19842 }, { "epoch": 49.6075, "grad_norm": 5.71875, "learning_rate": 1.0612550727563799e-05, "loss": 1.2877, "step": 19843 }, { "epoch": 49.61, "grad_norm": 11.125, "learning_rate": 1.0611742557206837e-05, "loss": 1.1792, "step": 19844 }, { "epoch": 49.6125, "grad_norm": 7.71875, "learning_rate": 1.0610934382839312e-05, "loss": 1.3008, "step": 19845 }, { "epoch": 49.615, "grad_norm": 5.0, "learning_rate": 1.0610126204466528e-05, "loss": 1.3341, "step": 19846 }, { "epoch": 49.6175, "grad_norm": 8.75, "learning_rate": 1.060931802209378e-05, "loss": 1.3345, "step": 19847 }, { "epoch": 49.62, "grad_norm": 8.9375, "learning_rate": 1.0608509835726368e-05, "loss": 1.4015, "step": 19848 }, { "epoch": 49.6225, "grad_norm": 7.0625, "learning_rate": 1.0607701645369593e-05, "loss": 1.3636, "step": 19849 }, { "epoch": 49.625, "grad_norm": 10.9375, "learning_rate": 1.0606893451028743e-05, "loss": 1.3951, "step": 19850 }, { "epoch": 49.6275, "grad_norm": 9.75, "learning_rate": 1.0606085252709131e-05, "loss": 1.5001, "step": 19851 }, { "epoch": 49.63, "grad_norm": 7.84375, "learning_rate": 1.0605277050416043e-05, "loss": 1.4, "step": 19852 }, { "epoch": 49.6325, "grad_norm": 5.25, "learning_rate": 1.0604468844154788e-05, "loss": 1.4032, "step": 19853 }, { "epoch": 49.635, "grad_norm": 5.53125, "learning_rate": 1.0603660633930654e-05, "loss": 1.3153, "step": 19854 }, { "epoch": 49.6375, "grad_norm": 7.5, "learning_rate": 1.0602852419748947e-05, "loss": 1.2957, "step": 19855 }, { "epoch": 49.64, "grad_norm": 12.125, "learning_rate": 1.0602044201614965e-05, "loss": 1.3295, "step": 19856 }, { "epoch": 49.6425, "grad_norm": 6.625, "learning_rate": 1.0601235979534003e-05, "loss": 1.2525, "step": 19857 }, { "epoch": 49.645, "grad_norm": 8.125, "learning_rate": 1.0600427753511363e-05, "loss": 1.2901, "step": 19858 }, { "epoch": 49.6475, "grad_norm": 8.125, "learning_rate": 1.0599619523552345e-05, "loss": 1.345, "step": 19859 }, { "epoch": 49.65, "grad_norm": 7.71875, "learning_rate": 1.0598811289662243e-05, "loss": 1.3861, "step": 19860 }, { "epoch": 49.6525, "grad_norm": 9.875, "learning_rate": 1.0598003051846359e-05, "loss": 1.4537, "step": 19861 }, { "epoch": 49.655, "grad_norm": 6.53125, "learning_rate": 1.059719481010999e-05, "loss": 1.2209, "step": 19862 }, { "epoch": 49.6575, "grad_norm": 9.1875, "learning_rate": 1.0596386564458437e-05, "loss": 1.3895, "step": 19863 }, { "epoch": 49.66, "grad_norm": 8.6875, "learning_rate": 1.0595578314896999e-05, "loss": 1.4261, "step": 19864 }, { "epoch": 49.6625, "grad_norm": 9.1875, "learning_rate": 1.0594770061430968e-05, "loss": 1.2744, "step": 19865 }, { "epoch": 49.665, "grad_norm": 6.5625, "learning_rate": 1.0593961804065653e-05, "loss": 1.2992, "step": 19866 }, { "epoch": 49.6675, "grad_norm": 5.53125, "learning_rate": 1.0593153542806345e-05, "loss": 1.3742, "step": 19867 }, { "epoch": 49.67, "grad_norm": 7.59375, "learning_rate": 1.059234527765835e-05, "loss": 1.2844, "step": 19868 }, { "epoch": 49.6725, "grad_norm": 7.59375, "learning_rate": 1.0591537008626962e-05, "loss": 1.3235, "step": 19869 }, { "epoch": 49.675, "grad_norm": 6.1875, "learning_rate": 1.059072873571748e-05, "loss": 1.2787, "step": 19870 }, { "epoch": 49.6775, "grad_norm": 7.125, "learning_rate": 1.0589920458935203e-05, "loss": 1.3876, "step": 19871 }, { "epoch": 49.68, "grad_norm": 7.65625, "learning_rate": 1.0589112178285432e-05, "loss": 1.3731, "step": 19872 }, { "epoch": 49.6825, "grad_norm": 5.84375, "learning_rate": 1.0588303893773467e-05, "loss": 1.1922, "step": 19873 }, { "epoch": 49.685, "grad_norm": 6.59375, "learning_rate": 1.0587495605404601e-05, "loss": 1.4007, "step": 19874 }, { "epoch": 49.6875, "grad_norm": 6.8125, "learning_rate": 1.0586687313184141e-05, "loss": 1.3192, "step": 19875 }, { "epoch": 49.69, "grad_norm": 7.5625, "learning_rate": 1.058587901711738e-05, "loss": 1.3604, "step": 19876 }, { "epoch": 49.6925, "grad_norm": 7.9375, "learning_rate": 1.058507071720962e-05, "loss": 1.3509, "step": 19877 }, { "epoch": 49.695, "grad_norm": 6.34375, "learning_rate": 1.0584262413466162e-05, "loss": 1.3807, "step": 19878 }, { "epoch": 49.6975, "grad_norm": 9.9375, "learning_rate": 1.0583454105892304e-05, "loss": 1.267, "step": 19879 }, { "epoch": 49.7, "grad_norm": 8.875, "learning_rate": 1.0582645794493337e-05, "loss": 1.331, "step": 19880 }, { "epoch": 49.7025, "grad_norm": 6.15625, "learning_rate": 1.0581837479274574e-05, "loss": 1.2258, "step": 19881 }, { "epoch": 49.705, "grad_norm": 7.0625, "learning_rate": 1.0581029160241305e-05, "loss": 1.3155, "step": 19882 }, { "epoch": 49.7075, "grad_norm": 10.0, "learning_rate": 1.0580220837398836e-05, "loss": 1.2573, "step": 19883 }, { "epoch": 49.71, "grad_norm": 9.4375, "learning_rate": 1.057941251075246e-05, "loss": 1.1542, "step": 19884 }, { "epoch": 49.7125, "grad_norm": 7.28125, "learning_rate": 1.0578604180307477e-05, "loss": 1.2515, "step": 19885 }, { "epoch": 49.715, "grad_norm": 9.625, "learning_rate": 1.057779584606919e-05, "loss": 1.3102, "step": 19886 }, { "epoch": 49.7175, "grad_norm": 11.0, "learning_rate": 1.0576987508042896e-05, "loss": 1.4476, "step": 19887 }, { "epoch": 49.72, "grad_norm": 6.53125, "learning_rate": 1.0576179166233895e-05, "loss": 1.3538, "step": 19888 }, { "epoch": 49.7225, "grad_norm": 6.15625, "learning_rate": 1.0575370820647486e-05, "loss": 1.3263, "step": 19889 }, { "epoch": 49.725, "grad_norm": 8.9375, "learning_rate": 1.0574562471288969e-05, "loss": 1.1804, "step": 19890 }, { "epoch": 49.7275, "grad_norm": 6.875, "learning_rate": 1.0573754118163643e-05, "loss": 1.3999, "step": 19891 }, { "epoch": 49.73, "grad_norm": 10.875, "learning_rate": 1.0572945761276809e-05, "loss": 1.2477, "step": 19892 }, { "epoch": 49.7325, "grad_norm": 9.0, "learning_rate": 1.0572137400633766e-05, "loss": 1.4045, "step": 19893 }, { "epoch": 49.735, "grad_norm": 9.5, "learning_rate": 1.0571329036239814e-05, "loss": 1.3181, "step": 19894 }, { "epoch": 49.7375, "grad_norm": 6.96875, "learning_rate": 1.0570520668100249e-05, "loss": 1.4369, "step": 19895 }, { "epoch": 49.74, "grad_norm": 6.46875, "learning_rate": 1.0569712296220375e-05, "loss": 1.3507, "step": 19896 }, { "epoch": 49.7425, "grad_norm": 5.6875, "learning_rate": 1.0568903920605488e-05, "loss": 1.3472, "step": 19897 }, { "epoch": 49.745, "grad_norm": 5.25, "learning_rate": 1.0568095541260895e-05, "loss": 1.3199, "step": 19898 }, { "epoch": 49.7475, "grad_norm": 6.375, "learning_rate": 1.0567287158191886e-05, "loss": 1.25, "step": 19899 }, { "epoch": 49.75, "grad_norm": 8.4375, "learning_rate": 1.0566478771403763e-05, "loss": 1.3449, "step": 19900 }, { "epoch": 49.7525, "grad_norm": 7.84375, "learning_rate": 1.0565670380901832e-05, "loss": 1.3179, "step": 19901 }, { "epoch": 49.755, "grad_norm": 6.71875, "learning_rate": 1.0564861986691387e-05, "loss": 1.1889, "step": 19902 }, { "epoch": 49.7575, "grad_norm": 5.96875, "learning_rate": 1.056405358877773e-05, "loss": 1.2343, "step": 19903 }, { "epoch": 49.76, "grad_norm": 14.625, "learning_rate": 1.056324518716616e-05, "loss": 1.2502, "step": 19904 }, { "epoch": 49.7625, "grad_norm": 8.25, "learning_rate": 1.0562436781861976e-05, "loss": 1.3013, "step": 19905 }, { "epoch": 49.765, "grad_norm": 6.96875, "learning_rate": 1.056162837287048e-05, "loss": 1.5164, "step": 19906 }, { "epoch": 49.7675, "grad_norm": 7.875, "learning_rate": 1.0560819960196968e-05, "loss": 1.3365, "step": 19907 }, { "epoch": 49.77, "grad_norm": 8.5, "learning_rate": 1.0560011543846749e-05, "loss": 1.2852, "step": 19908 }, { "epoch": 49.7725, "grad_norm": 8.125, "learning_rate": 1.0559203123825113e-05, "loss": 1.4713, "step": 19909 }, { "epoch": 49.775, "grad_norm": 10.375, "learning_rate": 1.0558394700137362e-05, "loss": 1.3524, "step": 19910 }, { "epoch": 49.7775, "grad_norm": 6.5625, "learning_rate": 1.05575862727888e-05, "loss": 1.3479, "step": 19911 }, { "epoch": 49.78, "grad_norm": 7.59375, "learning_rate": 1.0556777841784725e-05, "loss": 1.3224, "step": 19912 }, { "epoch": 49.7825, "grad_norm": 8.75, "learning_rate": 1.0555969407130435e-05, "loss": 1.3298, "step": 19913 }, { "epoch": 49.785, "grad_norm": 6.15625, "learning_rate": 1.0555160968831232e-05, "loss": 1.2978, "step": 19914 }, { "epoch": 49.7875, "grad_norm": 7.71875, "learning_rate": 1.0554352526892417e-05, "loss": 1.1875, "step": 19915 }, { "epoch": 49.79, "grad_norm": 7.875, "learning_rate": 1.055354408131929e-05, "loss": 1.3669, "step": 19916 }, { "epoch": 49.7925, "grad_norm": 13.0, "learning_rate": 1.0552735632117148e-05, "loss": 1.2195, "step": 19917 }, { "epoch": 49.795, "grad_norm": 7.0, "learning_rate": 1.0551927179291295e-05, "loss": 1.4382, "step": 19918 }, { "epoch": 49.7975, "grad_norm": 6.75, "learning_rate": 1.0551118722847028e-05, "loss": 1.4978, "step": 19919 }, { "epoch": 49.8, "grad_norm": 10.0625, "learning_rate": 1.055031026278965e-05, "loss": 1.3805, "step": 19920 }, { "epoch": 49.8025, "grad_norm": 8.25, "learning_rate": 1.054950179912446e-05, "loss": 1.3343, "step": 19921 }, { "epoch": 49.805, "grad_norm": 6.84375, "learning_rate": 1.0548693331856757e-05, "loss": 1.4152, "step": 19922 }, { "epoch": 49.8075, "grad_norm": 8.5, "learning_rate": 1.0547884860991844e-05, "loss": 1.2322, "step": 19923 }, { "epoch": 49.81, "grad_norm": 10.0, "learning_rate": 1.0547076386535017e-05, "loss": 1.4922, "step": 19924 }, { "epoch": 49.8125, "grad_norm": 6.40625, "learning_rate": 1.0546267908491582e-05, "loss": 1.4122, "step": 19925 }, { "epoch": 49.815, "grad_norm": 7.28125, "learning_rate": 1.0545459426866835e-05, "loss": 1.3855, "step": 19926 }, { "epoch": 49.8175, "grad_norm": 6.28125, "learning_rate": 1.054465094166608e-05, "loss": 1.3121, "step": 19927 }, { "epoch": 49.82, "grad_norm": 6.5625, "learning_rate": 1.054384245289461e-05, "loss": 1.34, "step": 19928 }, { "epoch": 49.8225, "grad_norm": 7.65625, "learning_rate": 1.0543033960557737e-05, "loss": 1.2684, "step": 19929 }, { "epoch": 49.825, "grad_norm": 10.625, "learning_rate": 1.0542225464660749e-05, "loss": 1.4517, "step": 19930 }, { "epoch": 49.8275, "grad_norm": 9.0, "learning_rate": 1.0541416965208957e-05, "loss": 1.1722, "step": 19931 }, { "epoch": 49.83, "grad_norm": 7.875, "learning_rate": 1.0540608462207655e-05, "loss": 1.2808, "step": 19932 }, { "epoch": 49.8325, "grad_norm": 7.53125, "learning_rate": 1.0539799955662147e-05, "loss": 1.1487, "step": 19933 }, { "epoch": 49.835, "grad_norm": 10.6875, "learning_rate": 1.053899144557773e-05, "loss": 1.3279, "step": 19934 }, { "epoch": 49.8375, "grad_norm": 7.9375, "learning_rate": 1.0538182931959709e-05, "loss": 1.2646, "step": 19935 }, { "epoch": 49.84, "grad_norm": 10.6875, "learning_rate": 1.0537374414813384e-05, "loss": 1.4212, "step": 19936 }, { "epoch": 49.8425, "grad_norm": 13.875, "learning_rate": 1.0536565894144048e-05, "loss": 1.2658, "step": 19937 }, { "epoch": 49.845, "grad_norm": 5.21875, "learning_rate": 1.053575736995701e-05, "loss": 1.2657, "step": 19938 }, { "epoch": 49.8475, "grad_norm": 10.6875, "learning_rate": 1.0534948842257568e-05, "loss": 1.4233, "step": 19939 }, { "epoch": 49.85, "grad_norm": 6.96875, "learning_rate": 1.0534140311051026e-05, "loss": 1.2807, "step": 19940 }, { "epoch": 49.8525, "grad_norm": 7.34375, "learning_rate": 1.0533331776342679e-05, "loss": 1.3529, "step": 19941 }, { "epoch": 49.855, "grad_norm": 8.4375, "learning_rate": 1.053252323813783e-05, "loss": 1.3956, "step": 19942 }, { "epoch": 49.8575, "grad_norm": 9.0, "learning_rate": 1.053171469644178e-05, "loss": 1.4205, "step": 19943 }, { "epoch": 49.86, "grad_norm": 5.1875, "learning_rate": 1.053090615125983e-05, "loss": 1.3228, "step": 19944 }, { "epoch": 49.8625, "grad_norm": 9.875, "learning_rate": 1.053009760259728e-05, "loss": 1.3423, "step": 19945 }, { "epoch": 49.865, "grad_norm": 7.15625, "learning_rate": 1.0529289050459432e-05, "loss": 1.3254, "step": 19946 }, { "epoch": 49.8675, "grad_norm": 8.25, "learning_rate": 1.0528480494851586e-05, "loss": 1.4521, "step": 19947 }, { "epoch": 49.87, "grad_norm": 6.875, "learning_rate": 1.0527671935779042e-05, "loss": 1.3235, "step": 19948 }, { "epoch": 49.8725, "grad_norm": 6.9375, "learning_rate": 1.0526863373247102e-05, "loss": 1.4018, "step": 19949 }, { "epoch": 49.875, "grad_norm": 7.25, "learning_rate": 1.0526054807261067e-05, "loss": 1.3014, "step": 19950 }, { "epoch": 49.8775, "grad_norm": 11.1875, "learning_rate": 1.0525246237826241e-05, "loss": 1.3559, "step": 19951 }, { "epoch": 49.88, "grad_norm": 6.1875, "learning_rate": 1.0524437664947918e-05, "loss": 1.4426, "step": 19952 }, { "epoch": 49.8825, "grad_norm": 9.375, "learning_rate": 1.0523629088631405e-05, "loss": 1.34, "step": 19953 }, { "epoch": 49.885, "grad_norm": 5.90625, "learning_rate": 1.0522820508881997e-05, "loss": 1.3843, "step": 19954 }, { "epoch": 49.8875, "grad_norm": 7.5625, "learning_rate": 1.0522011925705002e-05, "loss": 1.3724, "step": 19955 }, { "epoch": 49.89, "grad_norm": 6.90625, "learning_rate": 1.0521203339105719e-05, "loss": 1.3249, "step": 19956 }, { "epoch": 49.8925, "grad_norm": 10.9375, "learning_rate": 1.0520394749089444e-05, "loss": 1.3594, "step": 19957 }, { "epoch": 49.895, "grad_norm": 4.9375, "learning_rate": 1.0519586155661483e-05, "loss": 1.4192, "step": 19958 }, { "epoch": 49.8975, "grad_norm": 6.34375, "learning_rate": 1.0518777558827136e-05, "loss": 1.3448, "step": 19959 }, { "epoch": 49.9, "grad_norm": 9.5, "learning_rate": 1.0517968958591705e-05, "loss": 1.4647, "step": 19960 }, { "epoch": 49.9025, "grad_norm": 6.34375, "learning_rate": 1.051716035496049e-05, "loss": 1.1917, "step": 19961 }, { "epoch": 49.905, "grad_norm": 7.0, "learning_rate": 1.051635174793879e-05, "loss": 1.341, "step": 19962 }, { "epoch": 49.9075, "grad_norm": 13.75, "learning_rate": 1.0515543137531912e-05, "loss": 1.3674, "step": 19963 }, { "epoch": 49.91, "grad_norm": 9.3125, "learning_rate": 1.0514734523745151e-05, "loss": 1.3196, "step": 19964 }, { "epoch": 49.9125, "grad_norm": 8.0625, "learning_rate": 1.0513925906583813e-05, "loss": 1.2677, "step": 19965 }, { "epoch": 49.915, "grad_norm": 8.5625, "learning_rate": 1.0513117286053198e-05, "loss": 1.2934, "step": 19966 }, { "epoch": 49.9175, "grad_norm": 10.6875, "learning_rate": 1.0512308662158604e-05, "loss": 1.474, "step": 19967 }, { "epoch": 49.92, "grad_norm": 6.4375, "learning_rate": 1.051150003490534e-05, "loss": 1.2371, "step": 19968 }, { "epoch": 49.9225, "grad_norm": 8.6875, "learning_rate": 1.0510691404298695e-05, "loss": 1.3778, "step": 19969 }, { "epoch": 49.925, "grad_norm": 6.40625, "learning_rate": 1.0509882770343985e-05, "loss": 1.5735, "step": 19970 }, { "epoch": 49.9275, "grad_norm": 7.1875, "learning_rate": 1.0509074133046497e-05, "loss": 1.4106, "step": 19971 }, { "epoch": 49.93, "grad_norm": 8.625, "learning_rate": 1.0508265492411545e-05, "loss": 1.3454, "step": 19972 }, { "epoch": 49.9325, "grad_norm": 6.5625, "learning_rate": 1.0507456848444421e-05, "loss": 1.3559, "step": 19973 }, { "epoch": 49.935, "grad_norm": 7.5625, "learning_rate": 1.0506648201150432e-05, "loss": 1.2616, "step": 19974 }, { "epoch": 49.9375, "grad_norm": 9.5625, "learning_rate": 1.0505839550534875e-05, "loss": 1.3319, "step": 19975 }, { "epoch": 49.94, "grad_norm": 6.875, "learning_rate": 1.0505030896603059e-05, "loss": 1.292, "step": 19976 }, { "epoch": 49.9425, "grad_norm": 8.4375, "learning_rate": 1.0504222239360276e-05, "loss": 1.4479, "step": 19977 }, { "epoch": 49.945, "grad_norm": 6.875, "learning_rate": 1.0503413578811837e-05, "loss": 1.1274, "step": 19978 }, { "epoch": 49.9475, "grad_norm": 7.71875, "learning_rate": 1.0502604914963032e-05, "loss": 1.2977, "step": 19979 }, { "epoch": 49.95, "grad_norm": 5.46875, "learning_rate": 1.0501796247819176e-05, "loss": 1.2891, "step": 19980 }, { "epoch": 49.9525, "grad_norm": 6.65625, "learning_rate": 1.0500987577385563e-05, "loss": 1.2768, "step": 19981 }, { "epoch": 49.955, "grad_norm": 6.5, "learning_rate": 1.0500178903667491e-05, "loss": 1.4442, "step": 19982 }, { "epoch": 49.9575, "grad_norm": 8.8125, "learning_rate": 1.049937022667027e-05, "loss": 1.3566, "step": 19983 }, { "epoch": 49.96, "grad_norm": 10.375, "learning_rate": 1.0498561546399194e-05, "loss": 1.3643, "step": 19984 }, { "epoch": 49.9625, "grad_norm": 5.34375, "learning_rate": 1.0497752862859572e-05, "loss": 1.3361, "step": 19985 }, { "epoch": 49.965, "grad_norm": 7.3125, "learning_rate": 1.04969441760567e-05, "loss": 1.3375, "step": 19986 }, { "epoch": 49.9675, "grad_norm": 5.6875, "learning_rate": 1.0496135485995884e-05, "loss": 1.308, "step": 19987 }, { "epoch": 49.97, "grad_norm": 13.0, "learning_rate": 1.0495326792682421e-05, "loss": 1.1985, "step": 19988 }, { "epoch": 49.9725, "grad_norm": 9.4375, "learning_rate": 1.0494518096121617e-05, "loss": 1.3908, "step": 19989 }, { "epoch": 49.975, "grad_norm": 10.0, "learning_rate": 1.0493709396318771e-05, "loss": 1.3364, "step": 19990 }, { "epoch": 49.9775, "grad_norm": 7.21875, "learning_rate": 1.049290069327919e-05, "loss": 1.2594, "step": 19991 }, { "epoch": 49.98, "grad_norm": 8.25, "learning_rate": 1.0492091987008167e-05, "loss": 1.3552, "step": 19992 }, { "epoch": 49.9825, "grad_norm": 5.34375, "learning_rate": 1.0491283277511013e-05, "loss": 1.2175, "step": 19993 }, { "epoch": 49.985, "grad_norm": 6.28125, "learning_rate": 1.049047456479302e-05, "loss": 1.328, "step": 19994 }, { "epoch": 49.9875, "grad_norm": 9.0, "learning_rate": 1.04896658488595e-05, "loss": 1.4028, "step": 19995 }, { "epoch": 49.99, "grad_norm": 7.875, "learning_rate": 1.0488857129715748e-05, "loss": 1.1689, "step": 19996 }, { "epoch": 49.9925, "grad_norm": 7.5, "learning_rate": 1.0488048407367069e-05, "loss": 1.4679, "step": 19997 }, { "epoch": 49.995, "grad_norm": 8.4375, "learning_rate": 1.0487239681818764e-05, "loss": 1.3712, "step": 19998 }, { "epoch": 49.9975, "grad_norm": 7.21875, "learning_rate": 1.0486430953076135e-05, "loss": 1.4691, "step": 19999 }, { "epoch": 50.0, "grad_norm": 7.46875, "learning_rate": 1.0485622221144485e-05, "loss": 1.2941, "step": 20000 }, { "epoch": 50.0025, "grad_norm": 8.1875, "learning_rate": 1.0484813486029113e-05, "loss": 1.3223, "step": 20001 }, { "epoch": 50.005, "grad_norm": 6.625, "learning_rate": 1.0484004747735323e-05, "loss": 1.3643, "step": 20002 }, { "epoch": 50.0075, "grad_norm": 9.5, "learning_rate": 1.0483196006268419e-05, "loss": 1.3473, "step": 20003 }, { "epoch": 50.01, "grad_norm": 10.625, "learning_rate": 1.0482387261633703e-05, "loss": 1.3319, "step": 20004 }, { "epoch": 50.0125, "grad_norm": 8.5, "learning_rate": 1.0481578513836472e-05, "loss": 1.5133, "step": 20005 }, { "epoch": 50.015, "grad_norm": 6.09375, "learning_rate": 1.0480769762882035e-05, "loss": 1.4064, "step": 20006 }, { "epoch": 50.0175, "grad_norm": 7.6875, "learning_rate": 1.0479961008775688e-05, "loss": 1.4256, "step": 20007 }, { "epoch": 50.02, "grad_norm": 8.5, "learning_rate": 1.0479152251522737e-05, "loss": 1.4198, "step": 20008 }, { "epoch": 50.0225, "grad_norm": 6.34375, "learning_rate": 1.047834349112848e-05, "loss": 1.2003, "step": 20009 }, { "epoch": 50.025, "grad_norm": 7.53125, "learning_rate": 1.047753472759823e-05, "loss": 1.4023, "step": 20010 }, { "epoch": 50.0275, "grad_norm": 6.46875, "learning_rate": 1.0476725960937273e-05, "loss": 1.3354, "step": 20011 }, { "epoch": 50.03, "grad_norm": 9.6875, "learning_rate": 1.0475917191150924e-05, "loss": 1.2011, "step": 20012 }, { "epoch": 50.0325, "grad_norm": 7.84375, "learning_rate": 1.0475108418244479e-05, "loss": 1.3678, "step": 20013 }, { "epoch": 50.035, "grad_norm": 6.65625, "learning_rate": 1.0474299642223243e-05, "loss": 1.3245, "step": 20014 }, { "epoch": 50.0375, "grad_norm": 5.65625, "learning_rate": 1.0473490863092517e-05, "loss": 1.3523, "step": 20015 }, { "epoch": 50.04, "grad_norm": 6.84375, "learning_rate": 1.0472682080857606e-05, "loss": 1.2898, "step": 20016 }, { "epoch": 50.0425, "grad_norm": 6.96875, "learning_rate": 1.0471873295523807e-05, "loss": 1.4192, "step": 20017 }, { "epoch": 50.045, "grad_norm": 7.46875, "learning_rate": 1.0471064507096427e-05, "loss": 1.366, "step": 20018 }, { "epoch": 50.0475, "grad_norm": 12.0, "learning_rate": 1.0470255715580768e-05, "loss": 1.2726, "step": 20019 }, { "epoch": 50.05, "grad_norm": 8.25, "learning_rate": 1.046944692098213e-05, "loss": 1.3511, "step": 20020 }, { "epoch": 50.0525, "grad_norm": 5.15625, "learning_rate": 1.0468638123305819e-05, "loss": 1.3031, "step": 20021 }, { "epoch": 50.055, "grad_norm": 10.0625, "learning_rate": 1.046782932255713e-05, "loss": 1.2697, "step": 20022 }, { "epoch": 50.0575, "grad_norm": 7.0625, "learning_rate": 1.0467020518741377e-05, "loss": 1.3849, "step": 20023 }, { "epoch": 50.06, "grad_norm": 6.9375, "learning_rate": 1.0466211711863851e-05, "loss": 1.3894, "step": 20024 }, { "epoch": 50.0625, "grad_norm": 7.5625, "learning_rate": 1.0465402901929864e-05, "loss": 1.4707, "step": 20025 }, { "epoch": 50.065, "grad_norm": 7.375, "learning_rate": 1.0464594088944712e-05, "loss": 1.398, "step": 20026 }, { "epoch": 50.0675, "grad_norm": 7.5, "learning_rate": 1.04637852729137e-05, "loss": 1.1691, "step": 20027 }, { "epoch": 50.07, "grad_norm": 6.96875, "learning_rate": 1.046297645384213e-05, "loss": 1.2904, "step": 20028 }, { "epoch": 50.0725, "grad_norm": 6.96875, "learning_rate": 1.0462167631735305e-05, "loss": 1.2758, "step": 20029 }, { "epoch": 50.075, "grad_norm": 8.8125, "learning_rate": 1.0461358806598526e-05, "loss": 1.2327, "step": 20030 }, { "epoch": 50.0775, "grad_norm": 9.0625, "learning_rate": 1.04605499784371e-05, "loss": 1.3032, "step": 20031 }, { "epoch": 50.08, "grad_norm": 6.875, "learning_rate": 1.0459741147256325e-05, "loss": 1.3236, "step": 20032 }, { "epoch": 50.0825, "grad_norm": 6.03125, "learning_rate": 1.0458932313061507e-05, "loss": 1.4504, "step": 20033 }, { "epoch": 50.085, "grad_norm": 8.25, "learning_rate": 1.0458123475857947e-05, "loss": 1.5021, "step": 20034 }, { "epoch": 50.0875, "grad_norm": 9.125, "learning_rate": 1.0457314635650946e-05, "loss": 1.3445, "step": 20035 }, { "epoch": 50.09, "grad_norm": 6.75, "learning_rate": 1.0456505792445812e-05, "loss": 1.3462, "step": 20036 }, { "epoch": 50.0925, "grad_norm": 7.28125, "learning_rate": 1.045569694624784e-05, "loss": 1.123, "step": 20037 }, { "epoch": 50.095, "grad_norm": 6.71875, "learning_rate": 1.0454888097062338e-05, "loss": 1.3334, "step": 20038 }, { "epoch": 50.0975, "grad_norm": 10.6875, "learning_rate": 1.0454079244894607e-05, "loss": 1.3625, "step": 20039 }, { "epoch": 50.1, "grad_norm": 5.71875, "learning_rate": 1.0453270389749956e-05, "loss": 1.3513, "step": 20040 }, { "epoch": 50.1025, "grad_norm": 8.625, "learning_rate": 1.0452461531633677e-05, "loss": 1.4061, "step": 20041 }, { "epoch": 50.105, "grad_norm": 4.875, "learning_rate": 1.045165267055108e-05, "loss": 1.3947, "step": 20042 }, { "epoch": 50.1075, "grad_norm": 8.0, "learning_rate": 1.0450843806507467e-05, "loss": 1.3789, "step": 20043 }, { "epoch": 50.11, "grad_norm": 4.84375, "learning_rate": 1.045003493950814e-05, "loss": 1.399, "step": 20044 }, { "epoch": 50.1125, "grad_norm": 7.78125, "learning_rate": 1.04492260695584e-05, "loss": 1.4051, "step": 20045 }, { "epoch": 50.115, "grad_norm": 10.5, "learning_rate": 1.0448417196663555e-05, "loss": 1.295, "step": 20046 }, { "epoch": 50.1175, "grad_norm": 6.3125, "learning_rate": 1.04476083208289e-05, "loss": 1.2959, "step": 20047 }, { "epoch": 50.12, "grad_norm": 7.59375, "learning_rate": 1.044679944205975e-05, "loss": 1.3091, "step": 20048 }, { "epoch": 50.1225, "grad_norm": 7.34375, "learning_rate": 1.0445990560361393e-05, "loss": 1.2774, "step": 20049 }, { "epoch": 50.125, "grad_norm": 8.3125, "learning_rate": 1.0445181675739144e-05, "loss": 1.2697, "step": 20050 }, { "epoch": 50.1275, "grad_norm": 7.09375, "learning_rate": 1.0444372788198303e-05, "loss": 1.3258, "step": 20051 }, { "epoch": 50.13, "grad_norm": 7.09375, "learning_rate": 1.044356389774417e-05, "loss": 1.1748, "step": 20052 }, { "epoch": 50.1325, "grad_norm": 6.3125, "learning_rate": 1.0442755004382048e-05, "loss": 1.3026, "step": 20053 }, { "epoch": 50.135, "grad_norm": 8.5, "learning_rate": 1.044194610811724e-05, "loss": 1.3373, "step": 20054 }, { "epoch": 50.1375, "grad_norm": 8.5625, "learning_rate": 1.0441137208955056e-05, "loss": 1.2012, "step": 20055 }, { "epoch": 50.14, "grad_norm": 10.25, "learning_rate": 1.0440328306900791e-05, "loss": 1.1911, "step": 20056 }, { "epoch": 50.1425, "grad_norm": 9.625, "learning_rate": 1.0439519401959755e-05, "loss": 1.0843, "step": 20057 }, { "epoch": 50.145, "grad_norm": 7.28125, "learning_rate": 1.0438710494137245e-05, "loss": 1.3649, "step": 20058 }, { "epoch": 50.1475, "grad_norm": 10.625, "learning_rate": 1.0437901583438567e-05, "loss": 1.2257, "step": 20059 }, { "epoch": 50.15, "grad_norm": 9.75, "learning_rate": 1.0437092669869025e-05, "loss": 1.3541, "step": 20060 }, { "epoch": 50.1525, "grad_norm": 9.5625, "learning_rate": 1.0436283753433918e-05, "loss": 1.1701, "step": 20061 }, { "epoch": 50.155, "grad_norm": 10.3125, "learning_rate": 1.0435474834138552e-05, "loss": 1.3213, "step": 20062 }, { "epoch": 50.1575, "grad_norm": 9.6875, "learning_rate": 1.0434665911988236e-05, "loss": 1.3277, "step": 20063 }, { "epoch": 50.16, "grad_norm": 7.8125, "learning_rate": 1.043385698698826e-05, "loss": 1.1905, "step": 20064 }, { "epoch": 50.1625, "grad_norm": 8.5625, "learning_rate": 1.0433048059143941e-05, "loss": 1.4011, "step": 20065 }, { "epoch": 50.165, "grad_norm": 6.40625, "learning_rate": 1.0432239128460573e-05, "loss": 1.341, "step": 20066 }, { "epoch": 50.1675, "grad_norm": 8.9375, "learning_rate": 1.0431430194943464e-05, "loss": 1.3002, "step": 20067 }, { "epoch": 50.17, "grad_norm": 6.5625, "learning_rate": 1.0430621258597914e-05, "loss": 1.2842, "step": 20068 }, { "epoch": 50.1725, "grad_norm": 7.875, "learning_rate": 1.042981231942923e-05, "loss": 1.3592, "step": 20069 }, { "epoch": 50.175, "grad_norm": 8.125, "learning_rate": 1.0429003377442714e-05, "loss": 1.1933, "step": 20070 }, { "epoch": 50.1775, "grad_norm": 6.3125, "learning_rate": 1.0428194432643668e-05, "loss": 1.3029, "step": 20071 }, { "epoch": 50.18, "grad_norm": 5.25, "learning_rate": 1.0427385485037398e-05, "loss": 1.3777, "step": 20072 }, { "epoch": 50.1825, "grad_norm": 11.875, "learning_rate": 1.0426576534629202e-05, "loss": 1.2672, "step": 20073 }, { "epoch": 50.185, "grad_norm": 5.46875, "learning_rate": 1.042576758142439e-05, "loss": 1.303, "step": 20074 }, { "epoch": 50.1875, "grad_norm": 8.875, "learning_rate": 1.0424958625428264e-05, "loss": 1.3981, "step": 20075 }, { "epoch": 50.19, "grad_norm": 5.34375, "learning_rate": 1.0424149666646124e-05, "loss": 1.3472, "step": 20076 }, { "epoch": 50.1925, "grad_norm": 8.125, "learning_rate": 1.0423340705083278e-05, "loss": 1.2963, "step": 20077 }, { "epoch": 50.195, "grad_norm": 6.34375, "learning_rate": 1.0422531740745027e-05, "loss": 1.3923, "step": 20078 }, { "epoch": 50.1975, "grad_norm": 6.84375, "learning_rate": 1.042172277363667e-05, "loss": 1.4278, "step": 20079 }, { "epoch": 50.2, "grad_norm": 8.0625, "learning_rate": 1.0420913803763522e-05, "loss": 1.2513, "step": 20080 }, { "epoch": 50.2025, "grad_norm": 8.8125, "learning_rate": 1.0420104831130876e-05, "loss": 1.4007, "step": 20081 }, { "epoch": 50.205, "grad_norm": 8.9375, "learning_rate": 1.0419295855744042e-05, "loss": 1.3999, "step": 20082 }, { "epoch": 50.2075, "grad_norm": 5.4375, "learning_rate": 1.041848687760832e-05, "loss": 1.218, "step": 20083 }, { "epoch": 50.21, "grad_norm": 6.53125, "learning_rate": 1.0417677896729015e-05, "loss": 1.2658, "step": 20084 }, { "epoch": 50.2125, "grad_norm": 7.78125, "learning_rate": 1.0416868913111428e-05, "loss": 1.342, "step": 20085 }, { "epoch": 50.215, "grad_norm": 5.75, "learning_rate": 1.0416059926760868e-05, "loss": 1.3116, "step": 20086 }, { "epoch": 50.2175, "grad_norm": 6.5, "learning_rate": 1.0415250937682635e-05, "loss": 1.3247, "step": 20087 }, { "epoch": 50.22, "grad_norm": 6.40625, "learning_rate": 1.0414441945882033e-05, "loss": 1.4346, "step": 20088 }, { "epoch": 50.2225, "grad_norm": 8.4375, "learning_rate": 1.0413632951364366e-05, "loss": 1.278, "step": 20089 }, { "epoch": 50.225, "grad_norm": 5.84375, "learning_rate": 1.0412823954134938e-05, "loss": 1.3206, "step": 20090 }, { "epoch": 50.2275, "grad_norm": 10.5625, "learning_rate": 1.0412014954199052e-05, "loss": 1.3151, "step": 20091 }, { "epoch": 50.23, "grad_norm": 8.875, "learning_rate": 1.0411205951562015e-05, "loss": 1.2577, "step": 20092 }, { "epoch": 50.2325, "grad_norm": 7.78125, "learning_rate": 1.0410396946229128e-05, "loss": 1.3817, "step": 20093 }, { "epoch": 50.235, "grad_norm": 8.125, "learning_rate": 1.0409587938205692e-05, "loss": 1.3718, "step": 20094 }, { "epoch": 50.2375, "grad_norm": 7.3125, "learning_rate": 1.0408778927497019e-05, "loss": 1.2903, "step": 20095 }, { "epoch": 50.24, "grad_norm": 9.3125, "learning_rate": 1.04079699141084e-05, "loss": 1.2837, "step": 20096 }, { "epoch": 50.2425, "grad_norm": 14.4375, "learning_rate": 1.0407160898045154e-05, "loss": 1.4234, "step": 20097 }, { "epoch": 50.245, "grad_norm": 7.625, "learning_rate": 1.0406351879312575e-05, "loss": 1.2511, "step": 20098 }, { "epoch": 50.2475, "grad_norm": 7.53125, "learning_rate": 1.040554285791597e-05, "loss": 1.1807, "step": 20099 }, { "epoch": 50.25, "grad_norm": 9.125, "learning_rate": 1.0404733833860639e-05, "loss": 1.294, "step": 20100 }, { "epoch": 50.2525, "grad_norm": 6.75, "learning_rate": 1.0403924807151891e-05, "loss": 1.2643, "step": 20101 }, { "epoch": 50.255, "grad_norm": 7.0625, "learning_rate": 1.0403115777795027e-05, "loss": 1.3653, "step": 20102 }, { "epoch": 50.2575, "grad_norm": 7.8125, "learning_rate": 1.0402306745795354e-05, "loss": 1.4711, "step": 20103 }, { "epoch": 50.26, "grad_norm": 8.125, "learning_rate": 1.0401497711158173e-05, "loss": 1.3384, "step": 20104 }, { "epoch": 50.2625, "grad_norm": 8.0625, "learning_rate": 1.040068867388879e-05, "loss": 1.3079, "step": 20105 }, { "epoch": 50.265, "grad_norm": 6.75, "learning_rate": 1.0399879633992507e-05, "loss": 1.2518, "step": 20106 }, { "epoch": 50.2675, "grad_norm": 6.375, "learning_rate": 1.039907059147463e-05, "loss": 1.374, "step": 20107 }, { "epoch": 50.27, "grad_norm": 7.28125, "learning_rate": 1.0398261546340464e-05, "loss": 1.2063, "step": 20108 }, { "epoch": 50.2725, "grad_norm": 6.25, "learning_rate": 1.0397452498595306e-05, "loss": 1.1567, "step": 20109 }, { "epoch": 50.275, "grad_norm": 6.59375, "learning_rate": 1.0396643448244473e-05, "loss": 1.2468, "step": 20110 }, { "epoch": 50.2775, "grad_norm": 8.1875, "learning_rate": 1.0395834395293254e-05, "loss": 1.3089, "step": 20111 }, { "epoch": 50.28, "grad_norm": 5.65625, "learning_rate": 1.0395025339746965e-05, "loss": 1.1347, "step": 20112 }, { "epoch": 50.2825, "grad_norm": 5.75, "learning_rate": 1.0394216281610903e-05, "loss": 1.428, "step": 20113 }, { "epoch": 50.285, "grad_norm": 8.4375, "learning_rate": 1.0393407220890377e-05, "loss": 1.2554, "step": 20114 }, { "epoch": 50.2875, "grad_norm": 7.53125, "learning_rate": 1.0392598157590687e-05, "loss": 1.2752, "step": 20115 }, { "epoch": 50.29, "grad_norm": 9.25, "learning_rate": 1.0391789091717142e-05, "loss": 1.2637, "step": 20116 }, { "epoch": 50.2925, "grad_norm": 10.125, "learning_rate": 1.0390980023275039e-05, "loss": 1.3843, "step": 20117 }, { "epoch": 50.295, "grad_norm": 6.96875, "learning_rate": 1.0390170952269692e-05, "loss": 1.4115, "step": 20118 }, { "epoch": 50.2975, "grad_norm": 5.53125, "learning_rate": 1.0389361878706395e-05, "loss": 1.2987, "step": 20119 }, { "epoch": 50.3, "grad_norm": 6.0, "learning_rate": 1.0388552802590461e-05, "loss": 1.2525, "step": 20120 }, { "epoch": 50.3025, "grad_norm": 7.625, "learning_rate": 1.0387743723927186e-05, "loss": 1.2008, "step": 20121 }, { "epoch": 50.305, "grad_norm": 8.0625, "learning_rate": 1.0386934642721882e-05, "loss": 1.3947, "step": 20122 }, { "epoch": 50.3075, "grad_norm": 7.5, "learning_rate": 1.0386125558979851e-05, "loss": 1.4432, "step": 20123 }, { "epoch": 50.31, "grad_norm": 6.25, "learning_rate": 1.0385316472706393e-05, "loss": 1.3228, "step": 20124 }, { "epoch": 50.3125, "grad_norm": 5.78125, "learning_rate": 1.0384507383906819e-05, "loss": 1.2489, "step": 20125 }, { "epoch": 50.315, "grad_norm": 6.75, "learning_rate": 1.0383698292586423e-05, "loss": 1.2093, "step": 20126 }, { "epoch": 50.3175, "grad_norm": 5.5625, "learning_rate": 1.0382889198750525e-05, "loss": 1.2206, "step": 20127 }, { "epoch": 50.32, "grad_norm": 5.84375, "learning_rate": 1.0382080102404417e-05, "loss": 1.3228, "step": 20128 }, { "epoch": 50.3225, "grad_norm": 6.6875, "learning_rate": 1.0381271003553409e-05, "loss": 1.4122, "step": 20129 }, { "epoch": 50.325, "grad_norm": 6.75, "learning_rate": 1.03804619022028e-05, "loss": 1.3529, "step": 20130 }, { "epoch": 50.3275, "grad_norm": 9.4375, "learning_rate": 1.03796527983579e-05, "loss": 1.3268, "step": 20131 }, { "epoch": 50.33, "grad_norm": 8.0, "learning_rate": 1.037884369202401e-05, "loss": 1.3569, "step": 20132 }, { "epoch": 50.3325, "grad_norm": 6.8125, "learning_rate": 1.037803458320644e-05, "loss": 1.4236, "step": 20133 }, { "epoch": 50.335, "grad_norm": 5.0, "learning_rate": 1.0377225471910486e-05, "loss": 1.1963, "step": 20134 }, { "epoch": 50.3375, "grad_norm": 5.875, "learning_rate": 1.037641635814146e-05, "loss": 1.4301, "step": 20135 }, { "epoch": 50.34, "grad_norm": 6.90625, "learning_rate": 1.037560724190466e-05, "loss": 1.2458, "step": 20136 }, { "epoch": 50.3425, "grad_norm": 6.40625, "learning_rate": 1.0374798123205396e-05, "loss": 1.3179, "step": 20137 }, { "epoch": 50.345, "grad_norm": 4.75, "learning_rate": 1.0373989002048972e-05, "loss": 1.3289, "step": 20138 }, { "epoch": 50.3475, "grad_norm": 7.5625, "learning_rate": 1.0373179878440688e-05, "loss": 1.3319, "step": 20139 }, { "epoch": 50.35, "grad_norm": 6.53125, "learning_rate": 1.0372370752385854e-05, "loss": 1.5673, "step": 20140 }, { "epoch": 50.3525, "grad_norm": 8.5625, "learning_rate": 1.037156162388977e-05, "loss": 1.207, "step": 20141 }, { "epoch": 50.355, "grad_norm": 11.0, "learning_rate": 1.0370752492957745e-05, "loss": 1.4484, "step": 20142 }, { "epoch": 50.3575, "grad_norm": 7.1875, "learning_rate": 1.0369943359595081e-05, "loss": 1.2174, "step": 20143 }, { "epoch": 50.36, "grad_norm": 9.625, "learning_rate": 1.0369134223807082e-05, "loss": 1.3514, "step": 20144 }, { "epoch": 50.3625, "grad_norm": 8.375, "learning_rate": 1.0368325085599053e-05, "loss": 1.149, "step": 20145 }, { "epoch": 50.365, "grad_norm": 10.0625, "learning_rate": 1.0367515944976303e-05, "loss": 1.3441, "step": 20146 }, { "epoch": 50.3675, "grad_norm": 8.5625, "learning_rate": 1.0366706801944132e-05, "loss": 1.2908, "step": 20147 }, { "epoch": 50.37, "grad_norm": 9.0625, "learning_rate": 1.0365897656507846e-05, "loss": 1.2819, "step": 20148 }, { "epoch": 50.3725, "grad_norm": 9.6875, "learning_rate": 1.0365088508672747e-05, "loss": 1.18, "step": 20149 }, { "epoch": 50.375, "grad_norm": 7.6875, "learning_rate": 1.0364279358444144e-05, "loss": 1.4459, "step": 20150 }, { "epoch": 50.3775, "grad_norm": 7.75, "learning_rate": 1.0363470205827339e-05, "loss": 1.2822, "step": 20151 }, { "epoch": 50.38, "grad_norm": 6.875, "learning_rate": 1.0362661050827643e-05, "loss": 1.1972, "step": 20152 }, { "epoch": 50.3825, "grad_norm": 7.09375, "learning_rate": 1.036185189345035e-05, "loss": 1.3009, "step": 20153 }, { "epoch": 50.385, "grad_norm": 5.0625, "learning_rate": 1.0361042733700772e-05, "loss": 1.3291, "step": 20154 }, { "epoch": 50.3875, "grad_norm": 7.96875, "learning_rate": 1.0360233571584212e-05, "loss": 1.3166, "step": 20155 }, { "epoch": 50.39, "grad_norm": 6.96875, "learning_rate": 1.0359424407105976e-05, "loss": 1.1972, "step": 20156 }, { "epoch": 50.3925, "grad_norm": 7.75, "learning_rate": 1.0358615240271369e-05, "loss": 1.4267, "step": 20157 }, { "epoch": 50.395, "grad_norm": 5.71875, "learning_rate": 1.0357806071085692e-05, "loss": 1.3441, "step": 20158 }, { "epoch": 50.3975, "grad_norm": 9.625, "learning_rate": 1.0356996899554255e-05, "loss": 1.3643, "step": 20159 }, { "epoch": 50.4, "grad_norm": 6.8125, "learning_rate": 1.0356187725682359e-05, "loss": 1.5065, "step": 20160 }, { "epoch": 50.4025, "grad_norm": 6.0625, "learning_rate": 1.035537854947531e-05, "loss": 1.2286, "step": 20161 }, { "epoch": 50.405, "grad_norm": 6.5, "learning_rate": 1.0354569370938414e-05, "loss": 1.2759, "step": 20162 }, { "epoch": 50.4075, "grad_norm": 7.53125, "learning_rate": 1.035376019007698e-05, "loss": 1.5102, "step": 20163 }, { "epoch": 50.41, "grad_norm": 6.3125, "learning_rate": 1.0352951006896302e-05, "loss": 1.4158, "step": 20164 }, { "epoch": 50.4125, "grad_norm": 6.96875, "learning_rate": 1.0352141821401696e-05, "loss": 1.255, "step": 20165 }, { "epoch": 50.415, "grad_norm": 8.625, "learning_rate": 1.0351332633598457e-05, "loss": 1.2742, "step": 20166 }, { "epoch": 50.4175, "grad_norm": 5.28125, "learning_rate": 1.03505234434919e-05, "loss": 1.3479, "step": 20167 }, { "epoch": 50.42, "grad_norm": 8.8125, "learning_rate": 1.0349714251087323e-05, "loss": 1.2739, "step": 20168 }, { "epoch": 50.4225, "grad_norm": 9.0, "learning_rate": 1.0348905056390035e-05, "loss": 1.2638, "step": 20169 }, { "epoch": 50.425, "grad_norm": 11.9375, "learning_rate": 1.0348095859405339e-05, "loss": 1.3641, "step": 20170 }, { "epoch": 50.4275, "grad_norm": 10.3125, "learning_rate": 1.0347286660138541e-05, "loss": 1.2801, "step": 20171 }, { "epoch": 50.43, "grad_norm": 8.1875, "learning_rate": 1.0346477458594944e-05, "loss": 1.4179, "step": 20172 }, { "epoch": 50.4325, "grad_norm": 7.3125, "learning_rate": 1.0345668254779854e-05, "loss": 1.3463, "step": 20173 }, { "epoch": 50.435, "grad_norm": 7.21875, "learning_rate": 1.034485904869858e-05, "loss": 1.3944, "step": 20174 }, { "epoch": 50.4375, "grad_norm": 8.9375, "learning_rate": 1.034404984035642e-05, "loss": 1.5427, "step": 20175 }, { "epoch": 50.44, "grad_norm": 6.96875, "learning_rate": 1.0343240629758683e-05, "loss": 1.3145, "step": 20176 }, { "epoch": 50.4425, "grad_norm": 7.28125, "learning_rate": 1.0342431416910676e-05, "loss": 1.2852, "step": 20177 }, { "epoch": 50.445, "grad_norm": 5.90625, "learning_rate": 1.0341622201817703e-05, "loss": 1.3813, "step": 20178 }, { "epoch": 50.4475, "grad_norm": 8.6875, "learning_rate": 1.0340812984485068e-05, "loss": 1.2142, "step": 20179 }, { "epoch": 50.45, "grad_norm": 6.09375, "learning_rate": 1.0340003764918078e-05, "loss": 1.2932, "step": 20180 }, { "epoch": 50.4525, "grad_norm": 8.375, "learning_rate": 1.0339194543122033e-05, "loss": 1.4323, "step": 20181 }, { "epoch": 50.455, "grad_norm": 8.25, "learning_rate": 1.0338385319102246e-05, "loss": 1.4797, "step": 20182 }, { "epoch": 50.4575, "grad_norm": 8.9375, "learning_rate": 1.0337576092864017e-05, "loss": 1.3787, "step": 20183 }, { "epoch": 50.46, "grad_norm": 9.1875, "learning_rate": 1.0336766864412655e-05, "loss": 1.2957, "step": 20184 }, { "epoch": 50.4625, "grad_norm": 7.0625, "learning_rate": 1.033595763375346e-05, "loss": 1.3137, "step": 20185 }, { "epoch": 50.465, "grad_norm": 7.625, "learning_rate": 1.033514840089174e-05, "loss": 1.2706, "step": 20186 }, { "epoch": 50.4675, "grad_norm": 7.1875, "learning_rate": 1.03343391658328e-05, "loss": 1.4093, "step": 20187 }, { "epoch": 50.47, "grad_norm": 11.25, "learning_rate": 1.0333529928581948e-05, "loss": 1.4187, "step": 20188 }, { "epoch": 50.4725, "grad_norm": 7.25, "learning_rate": 1.0332720689144486e-05, "loss": 1.2476, "step": 20189 }, { "epoch": 50.475, "grad_norm": 8.125, "learning_rate": 1.0331911447525724e-05, "loss": 1.3937, "step": 20190 }, { "epoch": 50.4775, "grad_norm": 7.0, "learning_rate": 1.0331102203730959e-05, "loss": 1.1959, "step": 20191 }, { "epoch": 50.48, "grad_norm": 6.9375, "learning_rate": 1.0330292957765502e-05, "loss": 1.4349, "step": 20192 }, { "epoch": 50.4825, "grad_norm": 10.0625, "learning_rate": 1.0329483709634657e-05, "loss": 1.1849, "step": 20193 }, { "epoch": 50.485, "grad_norm": 6.46875, "learning_rate": 1.032867445934373e-05, "loss": 1.2907, "step": 20194 }, { "epoch": 50.4875, "grad_norm": 8.0625, "learning_rate": 1.0327865206898032e-05, "loss": 1.2772, "step": 20195 }, { "epoch": 50.49, "grad_norm": 5.21875, "learning_rate": 1.0327055952302856e-05, "loss": 1.1694, "step": 20196 }, { "epoch": 50.4925, "grad_norm": 8.1875, "learning_rate": 1.0326246695563517e-05, "loss": 1.1118, "step": 20197 }, { "epoch": 50.495, "grad_norm": 11.25, "learning_rate": 1.0325437436685318e-05, "loss": 1.4664, "step": 20198 }, { "epoch": 50.4975, "grad_norm": 5.4375, "learning_rate": 1.0324628175673564e-05, "loss": 1.2195, "step": 20199 }, { "epoch": 50.5, "grad_norm": 8.9375, "learning_rate": 1.0323818912533561e-05, "loss": 1.3368, "step": 20200 }, { "epoch": 50.5025, "grad_norm": 6.25, "learning_rate": 1.0323009647270612e-05, "loss": 1.457, "step": 20201 }, { "epoch": 50.505, "grad_norm": 9.1875, "learning_rate": 1.0322200379890026e-05, "loss": 1.3876, "step": 20202 }, { "epoch": 50.5075, "grad_norm": 6.59375, "learning_rate": 1.0321391110397107e-05, "loss": 1.3062, "step": 20203 }, { "epoch": 50.51, "grad_norm": 7.71875, "learning_rate": 1.0320581838797163e-05, "loss": 1.2585, "step": 20204 }, { "epoch": 50.5125, "grad_norm": 9.125, "learning_rate": 1.0319772565095496e-05, "loss": 1.3423, "step": 20205 }, { "epoch": 50.515, "grad_norm": 7.1875, "learning_rate": 1.031896328929741e-05, "loss": 1.2928, "step": 20206 }, { "epoch": 50.5175, "grad_norm": 5.375, "learning_rate": 1.0318154011408219e-05, "loss": 1.3391, "step": 20207 }, { "epoch": 50.52, "grad_norm": 6.875, "learning_rate": 1.0317344731433217e-05, "loss": 1.3979, "step": 20208 }, { "epoch": 50.5225, "grad_norm": 6.03125, "learning_rate": 1.031653544937772e-05, "loss": 1.2902, "step": 20209 }, { "epoch": 50.525, "grad_norm": 6.71875, "learning_rate": 1.0315726165247029e-05, "loss": 1.4356, "step": 20210 }, { "epoch": 50.5275, "grad_norm": 6.6875, "learning_rate": 1.0314916879046448e-05, "loss": 1.357, "step": 20211 }, { "epoch": 50.53, "grad_norm": 8.5, "learning_rate": 1.0314107590781284e-05, "loss": 1.3683, "step": 20212 }, { "epoch": 50.5325, "grad_norm": 7.9375, "learning_rate": 1.0313298300456845e-05, "loss": 1.2505, "step": 20213 }, { "epoch": 50.535, "grad_norm": 6.375, "learning_rate": 1.0312489008078434e-05, "loss": 1.3012, "step": 20214 }, { "epoch": 50.5375, "grad_norm": 9.3125, "learning_rate": 1.0311679713651358e-05, "loss": 1.3282, "step": 20215 }, { "epoch": 50.54, "grad_norm": 7.6875, "learning_rate": 1.0310870417180922e-05, "loss": 1.2665, "step": 20216 }, { "epoch": 50.5425, "grad_norm": 9.375, "learning_rate": 1.0310061118672433e-05, "loss": 1.4484, "step": 20217 }, { "epoch": 50.545, "grad_norm": 5.53125, "learning_rate": 1.0309251818131194e-05, "loss": 1.1733, "step": 20218 }, { "epoch": 50.5475, "grad_norm": 13.25, "learning_rate": 1.0308442515562513e-05, "loss": 1.2996, "step": 20219 }, { "epoch": 50.55, "grad_norm": 7.34375, "learning_rate": 1.0307633210971697e-05, "loss": 1.3898, "step": 20220 }, { "epoch": 50.5525, "grad_norm": 8.75, "learning_rate": 1.0306823904364044e-05, "loss": 1.2835, "step": 20221 }, { "epoch": 50.555, "grad_norm": 6.78125, "learning_rate": 1.0306014595744873e-05, "loss": 1.3625, "step": 20222 }, { "epoch": 50.5575, "grad_norm": 6.75, "learning_rate": 1.0305205285119479e-05, "loss": 1.4141, "step": 20223 }, { "epoch": 50.56, "grad_norm": 6.21875, "learning_rate": 1.0304395972493172e-05, "loss": 1.3, "step": 20224 }, { "epoch": 50.5625, "grad_norm": 9.0, "learning_rate": 1.0303586657871258e-05, "loss": 1.3334, "step": 20225 }, { "epoch": 50.565, "grad_norm": 5.25, "learning_rate": 1.0302777341259042e-05, "loss": 1.4167, "step": 20226 }, { "epoch": 50.5675, "grad_norm": 6.40625, "learning_rate": 1.0301968022661831e-05, "loss": 1.2466, "step": 20227 }, { "epoch": 50.57, "grad_norm": 7.90625, "learning_rate": 1.0301158702084926e-05, "loss": 1.4313, "step": 20228 }, { "epoch": 50.5725, "grad_norm": 7.15625, "learning_rate": 1.0300349379533638e-05, "loss": 1.3318, "step": 20229 }, { "epoch": 50.575, "grad_norm": 9.1875, "learning_rate": 1.0299540055013273e-05, "loss": 1.4526, "step": 20230 }, { "epoch": 50.5775, "grad_norm": 10.375, "learning_rate": 1.0298730728529134e-05, "loss": 1.3077, "step": 20231 }, { "epoch": 50.58, "grad_norm": 7.15625, "learning_rate": 1.0297921400086528e-05, "loss": 1.3414, "step": 20232 }, { "epoch": 50.5825, "grad_norm": 8.1875, "learning_rate": 1.0297112069690762e-05, "loss": 1.1698, "step": 20233 }, { "epoch": 50.585, "grad_norm": 9.4375, "learning_rate": 1.0296302737347141e-05, "loss": 1.3737, "step": 20234 }, { "epoch": 50.5875, "grad_norm": 6.875, "learning_rate": 1.0295493403060973e-05, "loss": 1.478, "step": 20235 }, { "epoch": 50.59, "grad_norm": 7.875, "learning_rate": 1.0294684066837557e-05, "loss": 1.3039, "step": 20236 }, { "epoch": 50.5925, "grad_norm": 9.6875, "learning_rate": 1.029387472868221e-05, "loss": 1.3245, "step": 20237 }, { "epoch": 50.595, "grad_norm": 8.25, "learning_rate": 1.0293065388600228e-05, "loss": 1.4341, "step": 20238 }, { "epoch": 50.5975, "grad_norm": 7.90625, "learning_rate": 1.0292256046596922e-05, "loss": 1.3447, "step": 20239 }, { "epoch": 50.6, "grad_norm": 7.90625, "learning_rate": 1.0291446702677598e-05, "loss": 1.2842, "step": 20240 }, { "epoch": 50.6025, "grad_norm": 6.0625, "learning_rate": 1.0290637356847561e-05, "loss": 1.375, "step": 20241 }, { "epoch": 50.605, "grad_norm": 6.96875, "learning_rate": 1.0289828009112117e-05, "loss": 1.2228, "step": 20242 }, { "epoch": 50.6075, "grad_norm": 16.0, "learning_rate": 1.0289018659476571e-05, "loss": 1.1846, "step": 20243 }, { "epoch": 50.61, "grad_norm": 9.9375, "learning_rate": 1.0288209307946232e-05, "loss": 1.3939, "step": 20244 }, { "epoch": 50.6125, "grad_norm": 5.4375, "learning_rate": 1.0287399954526402e-05, "loss": 1.3457, "step": 20245 }, { "epoch": 50.615, "grad_norm": 11.4375, "learning_rate": 1.028659059922239e-05, "loss": 1.4191, "step": 20246 }, { "epoch": 50.6175, "grad_norm": 11.5, "learning_rate": 1.0285781242039502e-05, "loss": 1.3998, "step": 20247 }, { "epoch": 50.62, "grad_norm": 8.75, "learning_rate": 1.0284971882983043e-05, "loss": 1.4027, "step": 20248 }, { "epoch": 50.6225, "grad_norm": 8.5, "learning_rate": 1.0284162522058323e-05, "loss": 1.4787, "step": 20249 }, { "epoch": 50.625, "grad_norm": 7.59375, "learning_rate": 1.0283353159270644e-05, "loss": 1.3031, "step": 20250 }, { "epoch": 50.6275, "grad_norm": 5.125, "learning_rate": 1.0282543794625309e-05, "loss": 1.2737, "step": 20251 }, { "epoch": 50.63, "grad_norm": 9.1875, "learning_rate": 1.0281734428127633e-05, "loss": 1.5393, "step": 20252 }, { "epoch": 50.6325, "grad_norm": 8.25, "learning_rate": 1.0280925059782911e-05, "loss": 1.3327, "step": 20253 }, { "epoch": 50.635, "grad_norm": 4.375, "learning_rate": 1.0280115689596464e-05, "loss": 1.5603, "step": 20254 }, { "epoch": 50.6375, "grad_norm": 9.8125, "learning_rate": 1.0279306317573585e-05, "loss": 1.3315, "step": 20255 }, { "epoch": 50.64, "grad_norm": 9.125, "learning_rate": 1.0278496943719585e-05, "loss": 1.4018, "step": 20256 }, { "epoch": 50.6425, "grad_norm": 8.0, "learning_rate": 1.0277687568039769e-05, "loss": 1.3251, "step": 20257 }, { "epoch": 50.645, "grad_norm": 9.0625, "learning_rate": 1.0276878190539446e-05, "loss": 1.2408, "step": 20258 }, { "epoch": 50.6475, "grad_norm": 9.3125, "learning_rate": 1.027606881122392e-05, "loss": 1.4855, "step": 20259 }, { "epoch": 50.65, "grad_norm": 5.625, "learning_rate": 1.0275259430098502e-05, "loss": 1.2735, "step": 20260 }, { "epoch": 50.6525, "grad_norm": 7.03125, "learning_rate": 1.0274450047168488e-05, "loss": 1.278, "step": 20261 }, { "epoch": 50.655, "grad_norm": 8.0625, "learning_rate": 1.0273640662439193e-05, "loss": 1.3531, "step": 20262 }, { "epoch": 50.6575, "grad_norm": 7.1875, "learning_rate": 1.027283127591592e-05, "loss": 1.4234, "step": 20263 }, { "epoch": 50.66, "grad_norm": 8.5625, "learning_rate": 1.0272021887603977e-05, "loss": 1.3447, "step": 20264 }, { "epoch": 50.6625, "grad_norm": 5.78125, "learning_rate": 1.0271212497508671e-05, "loss": 1.4174, "step": 20265 }, { "epoch": 50.665, "grad_norm": 9.125, "learning_rate": 1.0270403105635302e-05, "loss": 1.2977, "step": 20266 }, { "epoch": 50.6675, "grad_norm": 7.0, "learning_rate": 1.0269593711989187e-05, "loss": 1.4536, "step": 20267 }, { "epoch": 50.67, "grad_norm": 7.71875, "learning_rate": 1.0268784316575622e-05, "loss": 1.327, "step": 20268 }, { "epoch": 50.6725, "grad_norm": 7.65625, "learning_rate": 1.0267974919399921e-05, "loss": 1.2678, "step": 20269 }, { "epoch": 50.675, "grad_norm": 7.90625, "learning_rate": 1.0267165520467385e-05, "loss": 1.324, "step": 20270 }, { "epoch": 50.6775, "grad_norm": 6.25, "learning_rate": 1.0266356119783324e-05, "loss": 1.4396, "step": 20271 }, { "epoch": 50.68, "grad_norm": 7.1875, "learning_rate": 1.0265546717353041e-05, "loss": 1.1985, "step": 20272 }, { "epoch": 50.6825, "grad_norm": 9.0625, "learning_rate": 1.0264737313181845e-05, "loss": 1.4103, "step": 20273 }, { "epoch": 50.685, "grad_norm": 6.375, "learning_rate": 1.0263927907275042e-05, "loss": 1.3711, "step": 20274 }, { "epoch": 50.6875, "grad_norm": 6.84375, "learning_rate": 1.0263118499637942e-05, "loss": 1.3498, "step": 20275 }, { "epoch": 50.69, "grad_norm": 10.9375, "learning_rate": 1.0262309090275841e-05, "loss": 1.3696, "step": 20276 }, { "epoch": 50.6925, "grad_norm": 6.90625, "learning_rate": 1.0261499679194059e-05, "loss": 1.4021, "step": 20277 }, { "epoch": 50.695, "grad_norm": 6.25, "learning_rate": 1.0260690266397889e-05, "loss": 1.2454, "step": 20278 }, { "epoch": 50.6975, "grad_norm": 7.65625, "learning_rate": 1.0259880851892648e-05, "loss": 1.3667, "step": 20279 }, { "epoch": 50.7, "grad_norm": 5.34375, "learning_rate": 1.0259071435683636e-05, "loss": 1.291, "step": 20280 }, { "epoch": 50.7025, "grad_norm": 7.46875, "learning_rate": 1.0258262017776165e-05, "loss": 1.4472, "step": 20281 }, { "epoch": 50.705, "grad_norm": 9.0625, "learning_rate": 1.0257452598175535e-05, "loss": 1.4446, "step": 20282 }, { "epoch": 50.7075, "grad_norm": 6.78125, "learning_rate": 1.0256643176887058e-05, "loss": 1.3727, "step": 20283 }, { "epoch": 50.71, "grad_norm": 7.6875, "learning_rate": 1.0255833753916038e-05, "loss": 1.2476, "step": 20284 }, { "epoch": 50.7125, "grad_norm": 6.4375, "learning_rate": 1.0255024329267782e-05, "loss": 1.2898, "step": 20285 }, { "epoch": 50.715, "grad_norm": 7.03125, "learning_rate": 1.0254214902947597e-05, "loss": 1.2932, "step": 20286 }, { "epoch": 50.7175, "grad_norm": 8.5, "learning_rate": 1.0253405474960789e-05, "loss": 1.2618, "step": 20287 }, { "epoch": 50.72, "grad_norm": 8.625, "learning_rate": 1.0252596045312666e-05, "loss": 1.3549, "step": 20288 }, { "epoch": 50.7225, "grad_norm": 6.09375, "learning_rate": 1.0251786614008533e-05, "loss": 1.586, "step": 20289 }, { "epoch": 50.725, "grad_norm": 7.125, "learning_rate": 1.0250977181053698e-05, "loss": 1.248, "step": 20290 }, { "epoch": 50.7275, "grad_norm": 7.1875, "learning_rate": 1.0250167746453464e-05, "loss": 1.3819, "step": 20291 }, { "epoch": 50.73, "grad_norm": 5.8125, "learning_rate": 1.0249358310213144e-05, "loss": 1.3564, "step": 20292 }, { "epoch": 50.7325, "grad_norm": 9.875, "learning_rate": 1.0248548872338036e-05, "loss": 1.3313, "step": 20293 }, { "epoch": 50.735, "grad_norm": 5.625, "learning_rate": 1.0247739432833457e-05, "loss": 1.4572, "step": 20294 }, { "epoch": 50.7375, "grad_norm": 6.1875, "learning_rate": 1.0246929991704704e-05, "loss": 1.3084, "step": 20295 }, { "epoch": 50.74, "grad_norm": 6.1875, "learning_rate": 1.0246120548957091e-05, "loss": 1.3234, "step": 20296 }, { "epoch": 50.7425, "grad_norm": 6.34375, "learning_rate": 1.0245311104595918e-05, "loss": 1.1722, "step": 20297 }, { "epoch": 50.745, "grad_norm": 6.90625, "learning_rate": 1.0244501658626498e-05, "loss": 1.2949, "step": 20298 }, { "epoch": 50.7475, "grad_norm": 8.25, "learning_rate": 1.0243692211054132e-05, "loss": 1.3761, "step": 20299 }, { "epoch": 50.75, "grad_norm": 7.0, "learning_rate": 1.0242882761884132e-05, "loss": 1.3219, "step": 20300 }, { "epoch": 50.7525, "grad_norm": 7.375, "learning_rate": 1.0242073311121803e-05, "loss": 1.3852, "step": 20301 }, { "epoch": 50.755, "grad_norm": 5.875, "learning_rate": 1.024126385877245e-05, "loss": 1.3938, "step": 20302 }, { "epoch": 50.7575, "grad_norm": 6.28125, "learning_rate": 1.024045440484138e-05, "loss": 1.185, "step": 20303 }, { "epoch": 50.76, "grad_norm": 5.28125, "learning_rate": 1.02396449493339e-05, "loss": 1.2296, "step": 20304 }, { "epoch": 50.7625, "grad_norm": 6.28125, "learning_rate": 1.0238835492255318e-05, "loss": 1.2858, "step": 20305 }, { "epoch": 50.765, "grad_norm": 7.5, "learning_rate": 1.0238026033610942e-05, "loss": 1.331, "step": 20306 }, { "epoch": 50.7675, "grad_norm": 6.625, "learning_rate": 1.0237216573406077e-05, "loss": 1.2905, "step": 20307 }, { "epoch": 50.77, "grad_norm": 12.375, "learning_rate": 1.0236407111646027e-05, "loss": 1.324, "step": 20308 }, { "epoch": 50.7725, "grad_norm": 7.25, "learning_rate": 1.0235597648336103e-05, "loss": 1.3071, "step": 20309 }, { "epoch": 50.775, "grad_norm": 9.0, "learning_rate": 1.0234788183481608e-05, "loss": 1.4728, "step": 20310 }, { "epoch": 50.7775, "grad_norm": 7.3125, "learning_rate": 1.0233978717087856e-05, "loss": 1.3698, "step": 20311 }, { "epoch": 50.78, "grad_norm": 7.53125, "learning_rate": 1.0233169249160145e-05, "loss": 1.3168, "step": 20312 }, { "epoch": 50.7825, "grad_norm": 5.28125, "learning_rate": 1.0232359779703787e-05, "loss": 1.4507, "step": 20313 }, { "epoch": 50.785, "grad_norm": 8.0, "learning_rate": 1.0231550308724085e-05, "loss": 1.3564, "step": 20314 }, { "epoch": 50.7875, "grad_norm": 9.375, "learning_rate": 1.0230740836226351e-05, "loss": 1.2479, "step": 20315 }, { "epoch": 50.79, "grad_norm": 9.125, "learning_rate": 1.0229931362215891e-05, "loss": 1.405, "step": 20316 }, { "epoch": 50.7925, "grad_norm": 7.8125, "learning_rate": 1.0229121886698005e-05, "loss": 1.3825, "step": 20317 }, { "epoch": 50.795, "grad_norm": 9.4375, "learning_rate": 1.022831240967801e-05, "loss": 1.1817, "step": 20318 }, { "epoch": 50.7975, "grad_norm": 9.5, "learning_rate": 1.0227502931161207e-05, "loss": 1.3081, "step": 20319 }, { "epoch": 50.8, "grad_norm": 8.625, "learning_rate": 1.02266934511529e-05, "loss": 1.4585, "step": 20320 }, { "epoch": 50.8025, "grad_norm": 6.46875, "learning_rate": 1.0225883969658403e-05, "loss": 1.328, "step": 20321 }, { "epoch": 50.805, "grad_norm": 6.9375, "learning_rate": 1.0225074486683023e-05, "loss": 1.2717, "step": 20322 }, { "epoch": 50.8075, "grad_norm": 6.875, "learning_rate": 1.0224265002232056e-05, "loss": 1.3132, "step": 20323 }, { "epoch": 50.81, "grad_norm": 7.90625, "learning_rate": 1.0223455516310825e-05, "loss": 1.4272, "step": 20324 }, { "epoch": 50.8125, "grad_norm": 8.25, "learning_rate": 1.022264602892462e-05, "loss": 1.3174, "step": 20325 }, { "epoch": 50.815, "grad_norm": 6.75, "learning_rate": 1.0221836540078764e-05, "loss": 1.4619, "step": 20326 }, { "epoch": 50.8175, "grad_norm": 17.875, "learning_rate": 1.0221027049778554e-05, "loss": 1.3571, "step": 20327 }, { "epoch": 50.82, "grad_norm": 5.28125, "learning_rate": 1.0220217558029298e-05, "loss": 1.3518, "step": 20328 }, { "epoch": 50.8225, "grad_norm": 8.625, "learning_rate": 1.0219408064836304e-05, "loss": 1.4142, "step": 20329 }, { "epoch": 50.825, "grad_norm": 10.1875, "learning_rate": 1.0218598570204882e-05, "loss": 1.3245, "step": 20330 }, { "epoch": 50.8275, "grad_norm": 7.1875, "learning_rate": 1.0217789074140337e-05, "loss": 1.3591, "step": 20331 }, { "epoch": 50.83, "grad_norm": 4.96875, "learning_rate": 1.0216979576647972e-05, "loss": 1.2543, "step": 20332 }, { "epoch": 50.8325, "grad_norm": 7.09375, "learning_rate": 1.0216170077733101e-05, "loss": 1.3257, "step": 20333 }, { "epoch": 50.835, "grad_norm": 6.75, "learning_rate": 1.0215360577401026e-05, "loss": 1.2298, "step": 20334 }, { "epoch": 50.8375, "grad_norm": 5.5625, "learning_rate": 1.0214551075657055e-05, "loss": 1.1487, "step": 20335 }, { "epoch": 50.84, "grad_norm": 7.4375, "learning_rate": 1.0213741572506497e-05, "loss": 1.2326, "step": 20336 }, { "epoch": 50.8425, "grad_norm": 8.875, "learning_rate": 1.0212932067954659e-05, "loss": 1.2605, "step": 20337 }, { "epoch": 50.845, "grad_norm": 7.28125, "learning_rate": 1.0212122562006844e-05, "loss": 1.5279, "step": 20338 }, { "epoch": 50.8475, "grad_norm": 7.59375, "learning_rate": 1.0211313054668364e-05, "loss": 1.3717, "step": 20339 }, { "epoch": 50.85, "grad_norm": 7.65625, "learning_rate": 1.0210503545944522e-05, "loss": 1.3004, "step": 20340 }, { "epoch": 50.8525, "grad_norm": 8.5, "learning_rate": 1.0209694035840629e-05, "loss": 1.3791, "step": 20341 }, { "epoch": 50.855, "grad_norm": 6.78125, "learning_rate": 1.020888452436199e-05, "loss": 1.2607, "step": 20342 }, { "epoch": 50.8575, "grad_norm": 9.0, "learning_rate": 1.0208075011513913e-05, "loss": 1.3471, "step": 20343 }, { "epoch": 50.86, "grad_norm": 8.8125, "learning_rate": 1.0207265497301704e-05, "loss": 1.3162, "step": 20344 }, { "epoch": 50.8625, "grad_norm": 9.1875, "learning_rate": 1.020645598173067e-05, "loss": 1.3488, "step": 20345 }, { "epoch": 50.865, "grad_norm": 5.40625, "learning_rate": 1.020564646480612e-05, "loss": 1.2444, "step": 20346 }, { "epoch": 50.8675, "grad_norm": 6.8125, "learning_rate": 1.0204836946533361e-05, "loss": 1.2298, "step": 20347 }, { "epoch": 50.87, "grad_norm": 8.375, "learning_rate": 1.0204027426917695e-05, "loss": 1.53, "step": 20348 }, { "epoch": 50.8725, "grad_norm": 7.5, "learning_rate": 1.0203217905964437e-05, "loss": 1.2469, "step": 20349 }, { "epoch": 50.875, "grad_norm": 6.90625, "learning_rate": 1.0202408383678887e-05, "loss": 1.2445, "step": 20350 }, { "epoch": 50.8775, "grad_norm": 10.1875, "learning_rate": 1.0201598860066362e-05, "loss": 1.3673, "step": 20351 }, { "epoch": 50.88, "grad_norm": 8.0, "learning_rate": 1.0200789335132157e-05, "loss": 1.2761, "step": 20352 }, { "epoch": 50.8825, "grad_norm": 5.53125, "learning_rate": 1.0199979808881589e-05, "loss": 1.364, "step": 20353 }, { "epoch": 50.885, "grad_norm": 7.0625, "learning_rate": 1.0199170281319959e-05, "loss": 1.3834, "step": 20354 }, { "epoch": 50.8875, "grad_norm": 8.1875, "learning_rate": 1.0198360752452577e-05, "loss": 1.3389, "step": 20355 }, { "epoch": 50.89, "grad_norm": 8.4375, "learning_rate": 1.0197551222284749e-05, "loss": 1.25, "step": 20356 }, { "epoch": 50.8925, "grad_norm": 10.25, "learning_rate": 1.0196741690821785e-05, "loss": 1.3348, "step": 20357 }, { "epoch": 50.895, "grad_norm": 6.84375, "learning_rate": 1.0195932158068988e-05, "loss": 1.2945, "step": 20358 }, { "epoch": 50.8975, "grad_norm": 5.71875, "learning_rate": 1.019512262403167e-05, "loss": 1.3012, "step": 20359 }, { "epoch": 50.9, "grad_norm": 7.625, "learning_rate": 1.0194313088715135e-05, "loss": 1.3941, "step": 20360 }, { "epoch": 50.9025, "grad_norm": 6.90625, "learning_rate": 1.0193503552124691e-05, "loss": 1.1858, "step": 20361 }, { "epoch": 50.905, "grad_norm": 10.875, "learning_rate": 1.0192694014265648e-05, "loss": 1.22, "step": 20362 }, { "epoch": 50.9075, "grad_norm": 6.53125, "learning_rate": 1.0191884475143306e-05, "loss": 1.2009, "step": 20363 }, { "epoch": 50.91, "grad_norm": 7.125, "learning_rate": 1.019107493476298e-05, "loss": 1.2274, "step": 20364 }, { "epoch": 50.9125, "grad_norm": 7.5625, "learning_rate": 1.0190265393129972e-05, "loss": 1.3007, "step": 20365 }, { "epoch": 50.915, "grad_norm": 8.75, "learning_rate": 1.0189455850249597e-05, "loss": 1.3119, "step": 20366 }, { "epoch": 50.9175, "grad_norm": 9.125, "learning_rate": 1.0188646306127152e-05, "loss": 1.197, "step": 20367 }, { "epoch": 50.92, "grad_norm": 7.0625, "learning_rate": 1.0187836760767954e-05, "loss": 1.3434, "step": 20368 }, { "epoch": 50.9225, "grad_norm": 8.125, "learning_rate": 1.01870272141773e-05, "loss": 1.3117, "step": 20369 }, { "epoch": 50.925, "grad_norm": 7.90625, "learning_rate": 1.0186217666360507e-05, "loss": 1.354, "step": 20370 }, { "epoch": 50.9275, "grad_norm": 5.59375, "learning_rate": 1.0185408117322877e-05, "loss": 1.2645, "step": 20371 }, { "epoch": 50.93, "grad_norm": 8.3125, "learning_rate": 1.0184598567069719e-05, "loss": 1.3016, "step": 20372 }, { "epoch": 50.9325, "grad_norm": 113.0, "learning_rate": 1.018378901560634e-05, "loss": 1.3561, "step": 20373 }, { "epoch": 50.935, "grad_norm": 6.90625, "learning_rate": 1.0182979462938047e-05, "loss": 1.2211, "step": 20374 }, { "epoch": 50.9375, "grad_norm": 7.0625, "learning_rate": 1.0182169909070148e-05, "loss": 1.2713, "step": 20375 }, { "epoch": 50.94, "grad_norm": 6.78125, "learning_rate": 1.0181360354007952e-05, "loss": 1.3769, "step": 20376 }, { "epoch": 50.9425, "grad_norm": 7.3125, "learning_rate": 1.0180550797756765e-05, "loss": 1.3207, "step": 20377 }, { "epoch": 50.945, "grad_norm": 10.8125, "learning_rate": 1.017974124032189e-05, "loss": 1.3205, "step": 20378 }, { "epoch": 50.9475, "grad_norm": 5.90625, "learning_rate": 1.0178931681708643e-05, "loss": 1.1766, "step": 20379 }, { "epoch": 50.95, "grad_norm": 9.25, "learning_rate": 1.0178122121922324e-05, "loss": 1.2739, "step": 20380 }, { "epoch": 50.9525, "grad_norm": 4.875, "learning_rate": 1.0177312560968248e-05, "loss": 1.3319, "step": 20381 }, { "epoch": 50.955, "grad_norm": 6.34375, "learning_rate": 1.0176502998851714e-05, "loss": 1.3212, "step": 20382 }, { "epoch": 50.9575, "grad_norm": 10.0625, "learning_rate": 1.0175693435578035e-05, "loss": 1.4515, "step": 20383 }, { "epoch": 50.96, "grad_norm": 6.34375, "learning_rate": 1.0174883871152517e-05, "loss": 1.2709, "step": 20384 }, { "epoch": 50.9625, "grad_norm": 7.1875, "learning_rate": 1.0174074305580467e-05, "loss": 1.3447, "step": 20385 }, { "epoch": 50.965, "grad_norm": 7.25, "learning_rate": 1.0173264738867192e-05, "loss": 1.1578, "step": 20386 }, { "epoch": 50.9675, "grad_norm": 7.25, "learning_rate": 1.0172455171018e-05, "loss": 1.3358, "step": 20387 }, { "epoch": 50.97, "grad_norm": 6.1875, "learning_rate": 1.01716456020382e-05, "loss": 1.2362, "step": 20388 }, { "epoch": 50.9725, "grad_norm": 9.0, "learning_rate": 1.0170836031933097e-05, "loss": 1.3934, "step": 20389 }, { "epoch": 50.975, "grad_norm": 15.3125, "learning_rate": 1.0170026460708e-05, "loss": 1.3887, "step": 20390 }, { "epoch": 50.9775, "grad_norm": 7.1875, "learning_rate": 1.0169216888368219e-05, "loss": 1.3551, "step": 20391 }, { "epoch": 50.98, "grad_norm": 8.375, "learning_rate": 1.0168407314919057e-05, "loss": 1.339, "step": 20392 }, { "epoch": 50.9825, "grad_norm": 7.15625, "learning_rate": 1.0167597740365821e-05, "loss": 1.2828, "step": 20393 }, { "epoch": 50.985, "grad_norm": 10.375, "learning_rate": 1.0166788164713825e-05, "loss": 1.2327, "step": 20394 }, { "epoch": 50.9875, "grad_norm": 7.84375, "learning_rate": 1.016597858796837e-05, "loss": 1.2762, "step": 20395 }, { "epoch": 50.99, "grad_norm": 5.625, "learning_rate": 1.0165169010134769e-05, "loss": 1.4231, "step": 20396 }, { "epoch": 50.9925, "grad_norm": 7.59375, "learning_rate": 1.0164359431218324e-05, "loss": 1.2832, "step": 20397 }, { "epoch": 50.995, "grad_norm": 10.25, "learning_rate": 1.0163549851224345e-05, "loss": 1.2754, "step": 20398 }, { "epoch": 50.9975, "grad_norm": 7.15625, "learning_rate": 1.0162740270158141e-05, "loss": 1.3742, "step": 20399 }, { "epoch": 51.0, "grad_norm": 5.53125, "learning_rate": 1.0161930688025018e-05, "loss": 1.3092, "step": 20400 }, { "epoch": 51.0025, "grad_norm": 5.40625, "learning_rate": 1.0161121104830282e-05, "loss": 1.2362, "step": 20401 }, { "epoch": 51.005, "grad_norm": 6.34375, "learning_rate": 1.0160311520579243e-05, "loss": 1.3343, "step": 20402 }, { "epoch": 51.0075, "grad_norm": 7.625, "learning_rate": 1.015950193527721e-05, "loss": 1.5074, "step": 20403 }, { "epoch": 51.01, "grad_norm": 8.8125, "learning_rate": 1.0158692348929486e-05, "loss": 1.3532, "step": 20404 }, { "epoch": 51.0125, "grad_norm": 9.0, "learning_rate": 1.0157882761541384e-05, "loss": 1.3954, "step": 20405 }, { "epoch": 51.015, "grad_norm": 9.25, "learning_rate": 1.0157073173118207e-05, "loss": 1.4466, "step": 20406 }, { "epoch": 51.0175, "grad_norm": 7.375, "learning_rate": 1.0156263583665268e-05, "loss": 1.2955, "step": 20407 }, { "epoch": 51.02, "grad_norm": 9.6875, "learning_rate": 1.0155453993187867e-05, "loss": 1.2352, "step": 20408 }, { "epoch": 51.0225, "grad_norm": 6.9375, "learning_rate": 1.0154644401691318e-05, "loss": 1.1988, "step": 20409 }, { "epoch": 51.025, "grad_norm": 6.75, "learning_rate": 1.0153834809180924e-05, "loss": 1.3403, "step": 20410 }, { "epoch": 51.0275, "grad_norm": 6.3125, "learning_rate": 1.0153025215662e-05, "loss": 1.3966, "step": 20411 }, { "epoch": 51.03, "grad_norm": 9.1875, "learning_rate": 1.0152215621139845e-05, "loss": 1.4231, "step": 20412 }, { "epoch": 51.0325, "grad_norm": 9.0, "learning_rate": 1.0151406025619772e-05, "loss": 1.1659, "step": 20413 }, { "epoch": 51.035, "grad_norm": 7.3125, "learning_rate": 1.0150596429107087e-05, "loss": 1.3158, "step": 20414 }, { "epoch": 51.0375, "grad_norm": 9.3125, "learning_rate": 1.0149786831607096e-05, "loss": 1.2872, "step": 20415 }, { "epoch": 51.04, "grad_norm": 7.59375, "learning_rate": 1.014897723312511e-05, "loss": 1.167, "step": 20416 }, { "epoch": 51.0425, "grad_norm": 7.09375, "learning_rate": 1.0148167633666437e-05, "loss": 1.4266, "step": 20417 }, { "epoch": 51.045, "grad_norm": 7.5625, "learning_rate": 1.0147358033236379e-05, "loss": 1.253, "step": 20418 }, { "epoch": 51.0475, "grad_norm": 10.3125, "learning_rate": 1.0146548431840251e-05, "loss": 1.3816, "step": 20419 }, { "epoch": 51.05, "grad_norm": 10.875, "learning_rate": 1.0145738829483354e-05, "loss": 1.3166, "step": 20420 }, { "epoch": 51.0525, "grad_norm": 10.125, "learning_rate": 1.0144929226171003e-05, "loss": 1.2151, "step": 20421 }, { "epoch": 51.055, "grad_norm": 10.1875, "learning_rate": 1.0144119621908499e-05, "loss": 1.4295, "step": 20422 }, { "epoch": 51.0575, "grad_norm": 7.9375, "learning_rate": 1.0143310016701156e-05, "loss": 1.4782, "step": 20423 }, { "epoch": 51.06, "grad_norm": 7.28125, "learning_rate": 1.0142500410554274e-05, "loss": 1.309, "step": 20424 }, { "epoch": 51.0625, "grad_norm": 7.875, "learning_rate": 1.0141690803473167e-05, "loss": 1.3415, "step": 20425 }, { "epoch": 51.065, "grad_norm": 8.875, "learning_rate": 1.0140881195463138e-05, "loss": 1.2574, "step": 20426 }, { "epoch": 51.0675, "grad_norm": 8.75, "learning_rate": 1.0140071586529501e-05, "loss": 1.4124, "step": 20427 }, { "epoch": 51.07, "grad_norm": 6.5, "learning_rate": 1.0139261976677562e-05, "loss": 1.2813, "step": 20428 }, { "epoch": 51.0725, "grad_norm": 6.75, "learning_rate": 1.0138452365912624e-05, "loss": 1.2711, "step": 20429 }, { "epoch": 51.075, "grad_norm": 8.5625, "learning_rate": 1.0137642754239996e-05, "loss": 1.1096, "step": 20430 }, { "epoch": 51.0775, "grad_norm": 8.6875, "learning_rate": 1.013683314166499e-05, "loss": 1.3513, "step": 20431 }, { "epoch": 51.08, "grad_norm": 17.875, "learning_rate": 1.013602352819291e-05, "loss": 1.3095, "step": 20432 }, { "epoch": 51.0825, "grad_norm": 8.125, "learning_rate": 1.0135213913829067e-05, "loss": 1.361, "step": 20433 }, { "epoch": 51.085, "grad_norm": 4.8125, "learning_rate": 1.0134404298578768e-05, "loss": 1.2254, "step": 20434 }, { "epoch": 51.0875, "grad_norm": 8.5, "learning_rate": 1.0133594682447316e-05, "loss": 1.2116, "step": 20435 }, { "epoch": 51.09, "grad_norm": 6.40625, "learning_rate": 1.0132785065440027e-05, "loss": 1.3849, "step": 20436 }, { "epoch": 51.0925, "grad_norm": 8.8125, "learning_rate": 1.0131975447562201e-05, "loss": 1.3328, "step": 20437 }, { "epoch": 51.095, "grad_norm": 8.75, "learning_rate": 1.0131165828819152e-05, "loss": 1.349, "step": 20438 }, { "epoch": 51.0975, "grad_norm": 8.125, "learning_rate": 1.0130356209216183e-05, "loss": 1.3514, "step": 20439 }, { "epoch": 51.1, "grad_norm": 8.375, "learning_rate": 1.0129546588758605e-05, "loss": 1.3044, "step": 20440 }, { "epoch": 51.1025, "grad_norm": 8.125, "learning_rate": 1.0128736967451724e-05, "loss": 1.289, "step": 20441 }, { "epoch": 51.105, "grad_norm": 7.75, "learning_rate": 1.0127927345300847e-05, "loss": 1.1787, "step": 20442 }, { "epoch": 51.1075, "grad_norm": 8.0625, "learning_rate": 1.0127117722311286e-05, "loss": 1.336, "step": 20443 }, { "epoch": 51.11, "grad_norm": 8.0625, "learning_rate": 1.0126308098488346e-05, "loss": 1.3104, "step": 20444 }, { "epoch": 51.1125, "grad_norm": 11.1875, "learning_rate": 1.0125498473837333e-05, "loss": 1.3598, "step": 20445 }, { "epoch": 51.115, "grad_norm": 7.4375, "learning_rate": 1.012468884836356e-05, "loss": 1.327, "step": 20446 }, { "epoch": 51.1175, "grad_norm": 6.25, "learning_rate": 1.012387922207233e-05, "loss": 1.246, "step": 20447 }, { "epoch": 51.12, "grad_norm": 8.5625, "learning_rate": 1.0123069594968952e-05, "loss": 1.3241, "step": 20448 }, { "epoch": 51.1225, "grad_norm": 10.875, "learning_rate": 1.0122259967058738e-05, "loss": 1.2985, "step": 20449 }, { "epoch": 51.125, "grad_norm": 8.0, "learning_rate": 1.012145033834699e-05, "loss": 1.2496, "step": 20450 }, { "epoch": 51.1275, "grad_norm": 11.6875, "learning_rate": 1.0120640708839021e-05, "loss": 1.3403, "step": 20451 }, { "epoch": 51.13, "grad_norm": 5.84375, "learning_rate": 1.0119831078540133e-05, "loss": 1.1712, "step": 20452 }, { "epoch": 51.1325, "grad_norm": 9.625, "learning_rate": 1.0119021447455639e-05, "loss": 1.3607, "step": 20453 }, { "epoch": 51.135, "grad_norm": 6.75, "learning_rate": 1.0118211815590846e-05, "loss": 1.3269, "step": 20454 }, { "epoch": 51.1375, "grad_norm": 6.96875, "learning_rate": 1.011740218295106e-05, "loss": 1.2305, "step": 20455 }, { "epoch": 51.14, "grad_norm": 6.34375, "learning_rate": 1.0116592549541588e-05, "loss": 1.2673, "step": 20456 }, { "epoch": 51.1425, "grad_norm": 8.6875, "learning_rate": 1.0115782915367742e-05, "loss": 1.3996, "step": 20457 }, { "epoch": 51.145, "grad_norm": 10.125, "learning_rate": 1.0114973280434828e-05, "loss": 1.3333, "step": 20458 }, { "epoch": 51.1475, "grad_norm": 7.3125, "learning_rate": 1.0114163644748152e-05, "loss": 1.4171, "step": 20459 }, { "epoch": 51.15, "grad_norm": 5.46875, "learning_rate": 1.0113354008313025e-05, "loss": 1.3509, "step": 20460 }, { "epoch": 51.1525, "grad_norm": 5.71875, "learning_rate": 1.0112544371134754e-05, "loss": 1.3773, "step": 20461 }, { "epoch": 51.155, "grad_norm": 9.375, "learning_rate": 1.0111734733218646e-05, "loss": 1.3269, "step": 20462 }, { "epoch": 51.1575, "grad_norm": 5.9375, "learning_rate": 1.011092509457001e-05, "loss": 1.3386, "step": 20463 }, { "epoch": 51.16, "grad_norm": 9.0, "learning_rate": 1.0110115455194157e-05, "loss": 1.306, "step": 20464 }, { "epoch": 51.1625, "grad_norm": 5.5625, "learning_rate": 1.0109305815096383e-05, "loss": 1.4612, "step": 20465 }, { "epoch": 51.165, "grad_norm": 5.5, "learning_rate": 1.0108496174282012e-05, "loss": 1.3015, "step": 20466 }, { "epoch": 51.1675, "grad_norm": 7.875, "learning_rate": 1.010768653275634e-05, "loss": 1.2823, "step": 20467 }, { "epoch": 51.17, "grad_norm": 6.5, "learning_rate": 1.0106876890524683e-05, "loss": 1.2727, "step": 20468 }, { "epoch": 51.1725, "grad_norm": 17.25, "learning_rate": 1.0106067247592345e-05, "loss": 1.2994, "step": 20469 }, { "epoch": 51.175, "grad_norm": 7.4375, "learning_rate": 1.0105257603964631e-05, "loss": 1.3274, "step": 20470 }, { "epoch": 51.1775, "grad_norm": 9.9375, "learning_rate": 1.0104447959646855e-05, "loss": 1.35, "step": 20471 }, { "epoch": 51.18, "grad_norm": 8.0625, "learning_rate": 1.0103638314644322e-05, "loss": 1.2805, "step": 20472 }, { "epoch": 51.1825, "grad_norm": 9.6875, "learning_rate": 1.0102828668962338e-05, "loss": 1.3925, "step": 20473 }, { "epoch": 51.185, "grad_norm": 6.5, "learning_rate": 1.0102019022606216e-05, "loss": 1.3704, "step": 20474 }, { "epoch": 51.1875, "grad_norm": 7.75, "learning_rate": 1.010120937558126e-05, "loss": 1.4171, "step": 20475 }, { "epoch": 51.19, "grad_norm": 7.75, "learning_rate": 1.010039972789278e-05, "loss": 1.3803, "step": 20476 }, { "epoch": 51.1925, "grad_norm": 10.3125, "learning_rate": 1.0099590079546084e-05, "loss": 1.4418, "step": 20477 }, { "epoch": 51.195, "grad_norm": 5.0625, "learning_rate": 1.0098780430546477e-05, "loss": 1.263, "step": 20478 }, { "epoch": 51.1975, "grad_norm": 6.21875, "learning_rate": 1.0097970780899276e-05, "loss": 1.3322, "step": 20479 }, { "epoch": 51.2, "grad_norm": 9.5, "learning_rate": 1.0097161130609774e-05, "loss": 1.4088, "step": 20480 }, { "epoch": 51.2025, "grad_norm": 7.5, "learning_rate": 1.0096351479683292e-05, "loss": 1.429, "step": 20481 }, { "epoch": 51.205, "grad_norm": 6.90625, "learning_rate": 1.0095541828125133e-05, "loss": 1.4313, "step": 20482 }, { "epoch": 51.2075, "grad_norm": 6.59375, "learning_rate": 1.0094732175940608e-05, "loss": 1.2866, "step": 20483 }, { "epoch": 51.21, "grad_norm": 6.15625, "learning_rate": 1.0093922523135021e-05, "loss": 1.3691, "step": 20484 }, { "epoch": 51.2125, "grad_norm": 7.65625, "learning_rate": 1.0093112869713679e-05, "loss": 1.414, "step": 20485 }, { "epoch": 51.215, "grad_norm": 9.3125, "learning_rate": 1.0092303215681893e-05, "loss": 1.2668, "step": 20486 }, { "epoch": 51.2175, "grad_norm": 7.90625, "learning_rate": 1.0091493561044974e-05, "loss": 1.4118, "step": 20487 }, { "epoch": 51.22, "grad_norm": 7.40625, "learning_rate": 1.0090683905808225e-05, "loss": 1.1675, "step": 20488 }, { "epoch": 51.2225, "grad_norm": 6.65625, "learning_rate": 1.0089874249976958e-05, "loss": 1.3733, "step": 20489 }, { "epoch": 51.225, "grad_norm": 9.375, "learning_rate": 1.0089064593556474e-05, "loss": 1.3616, "step": 20490 }, { "epoch": 51.2275, "grad_norm": 6.28125, "learning_rate": 1.0088254936552094e-05, "loss": 1.2094, "step": 20491 }, { "epoch": 51.23, "grad_norm": 12.375, "learning_rate": 1.008744527896911e-05, "loss": 1.3056, "step": 20492 }, { "epoch": 51.2325, "grad_norm": 6.71875, "learning_rate": 1.0086635620812844e-05, "loss": 1.2931, "step": 20493 }, { "epoch": 51.235, "grad_norm": 7.3125, "learning_rate": 1.0085825962088596e-05, "loss": 1.269, "step": 20494 }, { "epoch": 51.2375, "grad_norm": 6.59375, "learning_rate": 1.0085016302801676e-05, "loss": 1.2686, "step": 20495 }, { "epoch": 51.24, "grad_norm": 9.1875, "learning_rate": 1.0084206642957393e-05, "loss": 1.4007, "step": 20496 }, { "epoch": 51.2425, "grad_norm": 7.40625, "learning_rate": 1.0083396982561051e-05, "loss": 1.3809, "step": 20497 }, { "epoch": 51.245, "grad_norm": 6.8125, "learning_rate": 1.008258732161797e-05, "loss": 1.4456, "step": 20498 }, { "epoch": 51.2475, "grad_norm": 8.5, "learning_rate": 1.0081777660133443e-05, "loss": 1.3619, "step": 20499 }, { "epoch": 51.25, "grad_norm": 7.3125, "learning_rate": 1.0080967998112787e-05, "loss": 1.1483, "step": 20500 }, { "epoch": 51.2525, "grad_norm": 7.75, "learning_rate": 1.0080158335561308e-05, "loss": 1.3538, "step": 20501 }, { "epoch": 51.255, "grad_norm": 8.4375, "learning_rate": 1.0079348672484315e-05, "loss": 1.3881, "step": 20502 }, { "epoch": 51.2575, "grad_norm": 13.1875, "learning_rate": 1.0078539008887114e-05, "loss": 1.4099, "step": 20503 }, { "epoch": 51.26, "grad_norm": 6.90625, "learning_rate": 1.0077729344775018e-05, "loss": 1.3235, "step": 20504 }, { "epoch": 51.2625, "grad_norm": 8.5, "learning_rate": 1.0076919680153325e-05, "loss": 1.505, "step": 20505 }, { "epoch": 51.265, "grad_norm": 6.09375, "learning_rate": 1.0076110015027356e-05, "loss": 1.3339, "step": 20506 }, { "epoch": 51.2675, "grad_norm": 7.0, "learning_rate": 1.0075300349402409e-05, "loss": 1.3362, "step": 20507 }, { "epoch": 51.27, "grad_norm": 6.6875, "learning_rate": 1.0074490683283797e-05, "loss": 1.2756, "step": 20508 }, { "epoch": 51.2725, "grad_norm": 5.90625, "learning_rate": 1.0073681016676828e-05, "loss": 1.2603, "step": 20509 }, { "epoch": 51.275, "grad_norm": 6.71875, "learning_rate": 1.0072871349586808e-05, "loss": 1.3364, "step": 20510 }, { "epoch": 51.2775, "grad_norm": 8.25, "learning_rate": 1.0072061682019045e-05, "loss": 1.393, "step": 20511 }, { "epoch": 51.28, "grad_norm": 6.5, "learning_rate": 1.0071252013978852e-05, "loss": 1.3359, "step": 20512 }, { "epoch": 51.2825, "grad_norm": 5.59375, "learning_rate": 1.007044234547153e-05, "loss": 1.4247, "step": 20513 }, { "epoch": 51.285, "grad_norm": 11.8125, "learning_rate": 1.0069632676502393e-05, "loss": 1.2047, "step": 20514 }, { "epoch": 51.2875, "grad_norm": 5.40625, "learning_rate": 1.0068823007076747e-05, "loss": 1.3173, "step": 20515 }, { "epoch": 51.29, "grad_norm": 8.875, "learning_rate": 1.0068013337199898e-05, "loss": 1.3335, "step": 20516 }, { "epoch": 51.2925, "grad_norm": 6.15625, "learning_rate": 1.006720366687716e-05, "loss": 1.1378, "step": 20517 }, { "epoch": 51.295, "grad_norm": 6.28125, "learning_rate": 1.0066393996113832e-05, "loss": 1.2921, "step": 20518 }, { "epoch": 51.2975, "grad_norm": 7.28125, "learning_rate": 1.0065584324915234e-05, "loss": 1.4037, "step": 20519 }, { "epoch": 51.3, "grad_norm": 8.125, "learning_rate": 1.0064774653286662e-05, "loss": 1.4741, "step": 20520 }, { "epoch": 51.3025, "grad_norm": 9.5625, "learning_rate": 1.0063964981233436e-05, "loss": 1.1589, "step": 20521 }, { "epoch": 51.305, "grad_norm": 8.3125, "learning_rate": 1.0063155308760852e-05, "loss": 1.3186, "step": 20522 }, { "epoch": 51.3075, "grad_norm": 8.0625, "learning_rate": 1.0062345635874228e-05, "loss": 1.3545, "step": 20523 }, { "epoch": 51.31, "grad_norm": 6.40625, "learning_rate": 1.0061535962578867e-05, "loss": 1.297, "step": 20524 }, { "epoch": 51.3125, "grad_norm": 10.1875, "learning_rate": 1.0060726288880081e-05, "loss": 1.4927, "step": 20525 }, { "epoch": 51.315, "grad_norm": 10.5625, "learning_rate": 1.0059916614783175e-05, "loss": 1.4006, "step": 20526 }, { "epoch": 51.3175, "grad_norm": 7.4375, "learning_rate": 1.0059106940293455e-05, "loss": 1.4093, "step": 20527 }, { "epoch": 51.32, "grad_norm": 7.5625, "learning_rate": 1.0058297265416234e-05, "loss": 1.4406, "step": 20528 }, { "epoch": 51.3225, "grad_norm": 6.0, "learning_rate": 1.005748759015682e-05, "loss": 1.4057, "step": 20529 }, { "epoch": 51.325, "grad_norm": 8.875, "learning_rate": 1.0056677914520518e-05, "loss": 1.418, "step": 20530 }, { "epoch": 51.3275, "grad_norm": 7.9375, "learning_rate": 1.0055868238512639e-05, "loss": 1.3485, "step": 20531 }, { "epoch": 51.33, "grad_norm": 6.21875, "learning_rate": 1.0055058562138489e-05, "loss": 1.2995, "step": 20532 }, { "epoch": 51.3325, "grad_norm": 7.75, "learning_rate": 1.005424888540338e-05, "loss": 1.2782, "step": 20533 }, { "epoch": 51.335, "grad_norm": 5.875, "learning_rate": 1.0053439208312616e-05, "loss": 1.4105, "step": 20534 }, { "epoch": 51.3375, "grad_norm": 7.46875, "learning_rate": 1.0052629530871503e-05, "loss": 1.4101, "step": 20535 }, { "epoch": 51.34, "grad_norm": 9.3125, "learning_rate": 1.0051819853085358e-05, "loss": 1.2792, "step": 20536 }, { "epoch": 51.3425, "grad_norm": 6.5, "learning_rate": 1.005101017495948e-05, "loss": 1.2421, "step": 20537 }, { "epoch": 51.345, "grad_norm": 6.6875, "learning_rate": 1.0050200496499185e-05, "loss": 1.2989, "step": 20538 }, { "epoch": 51.3475, "grad_norm": 7.71875, "learning_rate": 1.0049390817709776e-05, "loss": 1.1788, "step": 20539 }, { "epoch": 51.35, "grad_norm": 9.1875, "learning_rate": 1.0048581138596563e-05, "loss": 1.4068, "step": 20540 }, { "epoch": 51.3525, "grad_norm": 7.15625, "learning_rate": 1.0047771459164853e-05, "loss": 1.3058, "step": 20541 }, { "epoch": 51.355, "grad_norm": 8.125, "learning_rate": 1.0046961779419957e-05, "loss": 1.2438, "step": 20542 }, { "epoch": 51.3575, "grad_norm": 8.5, "learning_rate": 1.0046152099367179e-05, "loss": 1.464, "step": 20543 }, { "epoch": 51.36, "grad_norm": 6.75, "learning_rate": 1.0045342419011832e-05, "loss": 1.3974, "step": 20544 }, { "epoch": 51.3625, "grad_norm": 9.25, "learning_rate": 1.004453273835922e-05, "loss": 1.408, "step": 20545 }, { "epoch": 51.365, "grad_norm": 5.40625, "learning_rate": 1.0043723057414654e-05, "loss": 1.4542, "step": 20546 }, { "epoch": 51.3675, "grad_norm": 10.1875, "learning_rate": 1.0042913376183441e-05, "loss": 1.4504, "step": 20547 }, { "epoch": 51.37, "grad_norm": 9.0625, "learning_rate": 1.004210369467089e-05, "loss": 1.2542, "step": 20548 }, { "epoch": 51.3725, "grad_norm": 5.96875, "learning_rate": 1.004129401288231e-05, "loss": 1.4286, "step": 20549 }, { "epoch": 51.375, "grad_norm": 9.4375, "learning_rate": 1.0040484330823006e-05, "loss": 1.4289, "step": 20550 }, { "epoch": 51.3775, "grad_norm": 7.53125, "learning_rate": 1.0039674648498293e-05, "loss": 1.1816, "step": 20551 }, { "epoch": 51.38, "grad_norm": 6.40625, "learning_rate": 1.0038864965913469e-05, "loss": 1.2672, "step": 20552 }, { "epoch": 51.3825, "grad_norm": 5.375, "learning_rate": 1.003805528307385e-05, "loss": 1.2784, "step": 20553 }, { "epoch": 51.385, "grad_norm": 6.40625, "learning_rate": 1.003724559998474e-05, "loss": 1.2718, "step": 20554 }, { "epoch": 51.3875, "grad_norm": 7.59375, "learning_rate": 1.0036435916651454e-05, "loss": 1.3382, "step": 20555 }, { "epoch": 51.39, "grad_norm": 7.96875, "learning_rate": 1.0035626233079291e-05, "loss": 1.4737, "step": 20556 }, { "epoch": 51.3925, "grad_norm": 14.4375, "learning_rate": 1.0034816549273568e-05, "loss": 1.4627, "step": 20557 }, { "epoch": 51.395, "grad_norm": 7.40625, "learning_rate": 1.0034006865239586e-05, "loss": 1.4266, "step": 20558 }, { "epoch": 51.3975, "grad_norm": 10.3125, "learning_rate": 1.0033197180982656e-05, "loss": 1.4002, "step": 20559 }, { "epoch": 51.4, "grad_norm": 6.09375, "learning_rate": 1.003238749650809e-05, "loss": 1.2172, "step": 20560 }, { "epoch": 51.4025, "grad_norm": 6.75, "learning_rate": 1.003157781182119e-05, "loss": 1.3809, "step": 20561 }, { "epoch": 51.405, "grad_norm": 7.28125, "learning_rate": 1.0030768126927267e-05, "loss": 1.3609, "step": 20562 }, { "epoch": 51.4075, "grad_norm": 5.78125, "learning_rate": 1.002995844183163e-05, "loss": 1.3191, "step": 20563 }, { "epoch": 51.41, "grad_norm": 7.875, "learning_rate": 1.0029148756539586e-05, "loss": 1.3075, "step": 20564 }, { "epoch": 51.4125, "grad_norm": 8.6875, "learning_rate": 1.0028339071056447e-05, "loss": 1.3227, "step": 20565 }, { "epoch": 51.415, "grad_norm": 10.8125, "learning_rate": 1.0027529385387515e-05, "loss": 1.4113, "step": 20566 }, { "epoch": 51.4175, "grad_norm": 8.1875, "learning_rate": 1.0026719699538101e-05, "loss": 1.334, "step": 20567 }, { "epoch": 51.42, "grad_norm": 7.8125, "learning_rate": 1.0025910013513519e-05, "loss": 1.2046, "step": 20568 }, { "epoch": 51.4225, "grad_norm": 7.0625, "learning_rate": 1.0025100327319064e-05, "loss": 1.2426, "step": 20569 }, { "epoch": 51.425, "grad_norm": 9.5, "learning_rate": 1.0024290640960059e-05, "loss": 1.413, "step": 20570 }, { "epoch": 51.4275, "grad_norm": 6.875, "learning_rate": 1.0023480954441802e-05, "loss": 1.3206, "step": 20571 }, { "epoch": 51.43, "grad_norm": 5.21875, "learning_rate": 1.0022671267769605e-05, "loss": 1.2517, "step": 20572 }, { "epoch": 51.4325, "grad_norm": 8.0625, "learning_rate": 1.0021861580948777e-05, "loss": 1.3908, "step": 20573 }, { "epoch": 51.435, "grad_norm": 6.4375, "learning_rate": 1.0021051893984626e-05, "loss": 1.2698, "step": 20574 }, { "epoch": 51.4375, "grad_norm": 4.75, "learning_rate": 1.0020242206882459e-05, "loss": 1.2144, "step": 20575 }, { "epoch": 51.44, "grad_norm": 7.0, "learning_rate": 1.0019432519647585e-05, "loss": 1.3257, "step": 20576 }, { "epoch": 51.4425, "grad_norm": 5.1875, "learning_rate": 1.001862283228531e-05, "loss": 1.2372, "step": 20577 }, { "epoch": 51.445, "grad_norm": 6.3125, "learning_rate": 1.0017813144800948e-05, "loss": 1.2499, "step": 20578 }, { "epoch": 51.4475, "grad_norm": 7.90625, "learning_rate": 1.0017003457199802e-05, "loss": 1.289, "step": 20579 }, { "epoch": 51.45, "grad_norm": 5.625, "learning_rate": 1.001619376948718e-05, "loss": 1.3172, "step": 20580 }, { "epoch": 51.4525, "grad_norm": 10.3125, "learning_rate": 1.0015384081668396e-05, "loss": 1.2501, "step": 20581 }, { "epoch": 51.455, "grad_norm": 11.75, "learning_rate": 1.0014574393748752e-05, "loss": 1.3492, "step": 20582 }, { "epoch": 51.4575, "grad_norm": 6.3125, "learning_rate": 1.0013764705733563e-05, "loss": 1.3396, "step": 20583 }, { "epoch": 51.46, "grad_norm": 8.375, "learning_rate": 1.0012955017628127e-05, "loss": 1.259, "step": 20584 }, { "epoch": 51.4625, "grad_norm": 4.90625, "learning_rate": 1.0012145329437761e-05, "loss": 1.2469, "step": 20585 }, { "epoch": 51.465, "grad_norm": 7.40625, "learning_rate": 1.0011335641167772e-05, "loss": 1.3911, "step": 20586 }, { "epoch": 51.4675, "grad_norm": 8.9375, "learning_rate": 1.0010525952823466e-05, "loss": 1.2847, "step": 20587 }, { "epoch": 51.47, "grad_norm": 7.1875, "learning_rate": 1.0009716264410153e-05, "loss": 1.3398, "step": 20588 }, { "epoch": 51.4725, "grad_norm": 6.09375, "learning_rate": 1.0008906575933139e-05, "loss": 1.391, "step": 20589 }, { "epoch": 51.475, "grad_norm": 6.3125, "learning_rate": 1.0008096887397734e-05, "loss": 1.2459, "step": 20590 }, { "epoch": 51.4775, "grad_norm": 7.5, "learning_rate": 1.000728719880925e-05, "loss": 1.376, "step": 20591 }, { "epoch": 51.48, "grad_norm": 9.625, "learning_rate": 1.0006477510172984e-05, "loss": 1.4132, "step": 20592 }, { "epoch": 51.4825, "grad_norm": 8.5625, "learning_rate": 1.0005667821494258e-05, "loss": 1.3503, "step": 20593 }, { "epoch": 51.485, "grad_norm": 7.9375, "learning_rate": 1.000485813277837e-05, "loss": 1.4743, "step": 20594 }, { "epoch": 51.4875, "grad_norm": 7.8125, "learning_rate": 1.0004048444030637e-05, "loss": 1.3673, "step": 20595 }, { "epoch": 51.49, "grad_norm": 6.5625, "learning_rate": 1.000323875525636e-05, "loss": 1.2756, "step": 20596 }, { "epoch": 51.4925, "grad_norm": 8.5, "learning_rate": 1.0002429066460848e-05, "loss": 1.5188, "step": 20597 }, { "epoch": 51.495, "grad_norm": 7.5625, "learning_rate": 1.0001619377649414e-05, "loss": 1.258, "step": 20598 }, { "epoch": 51.4975, "grad_norm": 8.625, "learning_rate": 1.0000809688827363e-05, "loss": 1.3195, "step": 20599 }, { "epoch": 51.5, "grad_norm": 9.1875, "learning_rate": 1e-05, "loss": 1.5174, "step": 20600 }, { "epoch": 51.5025, "grad_norm": 7.625, "learning_rate": 9.99919031117264e-06, "loss": 1.321, "step": 20601 }, { "epoch": 51.505, "grad_norm": 8.1875, "learning_rate": 9.998380622350591e-06, "loss": 1.1813, "step": 20602 }, { "epoch": 51.5075, "grad_norm": 7.375, "learning_rate": 9.997570933539154e-06, "loss": 1.4063, "step": 20603 }, { "epoch": 51.51, "grad_norm": 7.03125, "learning_rate": 9.996761244743643e-06, "loss": 1.5068, "step": 20604 }, { "epoch": 51.5125, "grad_norm": 6.375, "learning_rate": 9.995951555969367e-06, "loss": 1.2528, "step": 20605 }, { "epoch": 51.515, "grad_norm": 6.6875, "learning_rate": 9.995141867221633e-06, "loss": 1.3745, "step": 20606 }, { "epoch": 51.5175, "grad_norm": 7.90625, "learning_rate": 9.994332178505745e-06, "loss": 1.4346, "step": 20607 }, { "epoch": 51.52, "grad_norm": 5.46875, "learning_rate": 9.993522489827016e-06, "loss": 1.3165, "step": 20608 }, { "epoch": 51.5225, "grad_norm": 5.46875, "learning_rate": 9.992712801190753e-06, "loss": 1.1882, "step": 20609 }, { "epoch": 51.525, "grad_norm": 6.40625, "learning_rate": 9.99190311260227e-06, "loss": 1.3807, "step": 20610 }, { "epoch": 51.5275, "grad_norm": 7.0625, "learning_rate": 9.991093424066865e-06, "loss": 1.2143, "step": 20611 }, { "epoch": 51.53, "grad_norm": 7.03125, "learning_rate": 9.99028373558985e-06, "loss": 1.3697, "step": 20612 }, { "epoch": 51.5325, "grad_norm": 10.25, "learning_rate": 9.989474047176535e-06, "loss": 1.2217, "step": 20613 }, { "epoch": 51.535, "grad_norm": 9.6875, "learning_rate": 9.988664358832231e-06, "loss": 1.2273, "step": 20614 }, { "epoch": 51.5375, "grad_norm": 7.65625, "learning_rate": 9.98785467056224e-06, "loss": 1.3278, "step": 20615 }, { "epoch": 51.54, "grad_norm": 8.0, "learning_rate": 9.987044982371875e-06, "loss": 1.4645, "step": 20616 }, { "epoch": 51.5425, "grad_norm": 8.6875, "learning_rate": 9.986235294266444e-06, "loss": 1.2862, "step": 20617 }, { "epoch": 51.545, "grad_norm": 8.0, "learning_rate": 9.98542560625125e-06, "loss": 1.3283, "step": 20618 }, { "epoch": 51.5475, "grad_norm": 6.9375, "learning_rate": 9.984615918331605e-06, "loss": 1.3244, "step": 20619 }, { "epoch": 51.55, "grad_norm": 5.78125, "learning_rate": 9.98380623051282e-06, "loss": 1.35, "step": 20620 }, { "epoch": 51.5525, "grad_norm": 7.84375, "learning_rate": 9.982996542800203e-06, "loss": 1.2702, "step": 20621 }, { "epoch": 51.555, "grad_norm": 9.25, "learning_rate": 9.982186855199057e-06, "loss": 1.3802, "step": 20622 }, { "epoch": 51.5575, "grad_norm": 7.15625, "learning_rate": 9.981377167714691e-06, "loss": 1.2591, "step": 20623 }, { "epoch": 51.56, "grad_norm": 8.0625, "learning_rate": 9.980567480352417e-06, "loss": 1.4436, "step": 20624 }, { "epoch": 51.5625, "grad_norm": 10.25, "learning_rate": 9.979757793117545e-06, "loss": 1.2893, "step": 20625 }, { "epoch": 51.565, "grad_norm": 7.125, "learning_rate": 9.978948106015377e-06, "loss": 1.3765, "step": 20626 }, { "epoch": 51.5675, "grad_norm": 9.5625, "learning_rate": 9.978138419051227e-06, "loss": 1.5571, "step": 20627 }, { "epoch": 51.57, "grad_norm": 6.0625, "learning_rate": 9.977328732230394e-06, "loss": 1.3458, "step": 20628 }, { "epoch": 51.5725, "grad_norm": 9.1875, "learning_rate": 9.976519045558201e-06, "loss": 1.2762, "step": 20629 }, { "epoch": 51.575, "grad_norm": 7.375, "learning_rate": 9.975709359039944e-06, "loss": 1.3, "step": 20630 }, { "epoch": 51.5775, "grad_norm": 8.8125, "learning_rate": 9.97489967268094e-06, "loss": 1.3713, "step": 20631 }, { "epoch": 51.58, "grad_norm": 10.25, "learning_rate": 9.974089986486488e-06, "loss": 1.3552, "step": 20632 }, { "epoch": 51.5825, "grad_norm": 9.375, "learning_rate": 9.973280300461899e-06, "loss": 1.3347, "step": 20633 }, { "epoch": 51.585, "grad_norm": 6.25, "learning_rate": 9.972470614612488e-06, "loss": 1.3107, "step": 20634 }, { "epoch": 51.5875, "grad_norm": 8.375, "learning_rate": 9.971660928943556e-06, "loss": 1.345, "step": 20635 }, { "epoch": 51.59, "grad_norm": 5.375, "learning_rate": 9.970851243460416e-06, "loss": 1.3064, "step": 20636 }, { "epoch": 51.5925, "grad_norm": 6.34375, "learning_rate": 9.970041558168373e-06, "loss": 1.4224, "step": 20637 }, { "epoch": 51.595, "grad_norm": 8.0, "learning_rate": 9.969231873072735e-06, "loss": 1.4094, "step": 20638 }, { "epoch": 51.5975, "grad_norm": 6.375, "learning_rate": 9.968422188178811e-06, "loss": 1.2278, "step": 20639 }, { "epoch": 51.6, "grad_norm": 6.875, "learning_rate": 9.967612503491915e-06, "loss": 1.3245, "step": 20640 }, { "epoch": 51.6025, "grad_norm": 6.5, "learning_rate": 9.966802819017346e-06, "loss": 1.303, "step": 20641 }, { "epoch": 51.605, "grad_norm": 5.59375, "learning_rate": 9.96599313476042e-06, "loss": 1.1723, "step": 20642 }, { "epoch": 51.6075, "grad_norm": 6.625, "learning_rate": 9.965183450726434e-06, "loss": 1.1955, "step": 20643 }, { "epoch": 51.61, "grad_norm": 9.5625, "learning_rate": 9.96437376692071e-06, "loss": 1.3083, "step": 20644 }, { "epoch": 51.6125, "grad_norm": 6.28125, "learning_rate": 9.963564083348549e-06, "loss": 1.3352, "step": 20645 }, { "epoch": 51.615, "grad_norm": 8.25, "learning_rate": 9.962754400015263e-06, "loss": 1.1688, "step": 20646 }, { "epoch": 51.6175, "grad_norm": 6.40625, "learning_rate": 9.961944716926155e-06, "loss": 1.3253, "step": 20647 }, { "epoch": 51.62, "grad_norm": 8.75, "learning_rate": 9.961135034086533e-06, "loss": 1.3519, "step": 20648 }, { "epoch": 51.6225, "grad_norm": 7.625, "learning_rate": 9.96032535150171e-06, "loss": 1.3553, "step": 20649 }, { "epoch": 51.625, "grad_norm": 5.78125, "learning_rate": 9.959515669176997e-06, "loss": 1.3371, "step": 20650 }, { "epoch": 51.6275, "grad_norm": 7.40625, "learning_rate": 9.958705987117694e-06, "loss": 1.3318, "step": 20651 }, { "epoch": 51.63, "grad_norm": 6.21875, "learning_rate": 9.957896305329112e-06, "loss": 1.2978, "step": 20652 }, { "epoch": 51.6325, "grad_norm": 11.0625, "learning_rate": 9.957086623816559e-06, "loss": 1.4451, "step": 20653 }, { "epoch": 51.635, "grad_norm": 9.625, "learning_rate": 9.95627694258535e-06, "loss": 1.4708, "step": 20654 }, { "epoch": 51.6375, "grad_norm": 7.8125, "learning_rate": 9.955467261640782e-06, "loss": 1.4013, "step": 20655 }, { "epoch": 51.64, "grad_norm": 7.53125, "learning_rate": 9.954657580988171e-06, "loss": 1.3146, "step": 20656 }, { "epoch": 51.6425, "grad_norm": 8.5625, "learning_rate": 9.953847900632826e-06, "loss": 1.2086, "step": 20657 }, { "epoch": 51.645, "grad_norm": 6.59375, "learning_rate": 9.953038220580046e-06, "loss": 1.2041, "step": 20658 }, { "epoch": 51.6475, "grad_norm": 9.25, "learning_rate": 9.952228540835149e-06, "loss": 1.6, "step": 20659 }, { "epoch": 51.65, "grad_norm": 9.125, "learning_rate": 9.95141886140344e-06, "loss": 1.311, "step": 20660 }, { "epoch": 51.6525, "grad_norm": 8.375, "learning_rate": 9.950609182290229e-06, "loss": 1.303, "step": 20661 }, { "epoch": 51.655, "grad_norm": 9.625, "learning_rate": 9.94979950350082e-06, "loss": 1.3527, "step": 20662 }, { "epoch": 51.6575, "grad_norm": 9.5, "learning_rate": 9.94898982504052e-06, "loss": 1.3489, "step": 20663 }, { "epoch": 51.66, "grad_norm": 9.875, "learning_rate": 9.948180146914644e-06, "loss": 1.1537, "step": 20664 }, { "epoch": 51.6625, "grad_norm": 6.75, "learning_rate": 9.9473704691285e-06, "loss": 1.4491, "step": 20665 }, { "epoch": 51.665, "grad_norm": 7.09375, "learning_rate": 9.94656079168739e-06, "loss": 1.5184, "step": 20666 }, { "epoch": 51.6675, "grad_norm": 8.6875, "learning_rate": 9.945751114596624e-06, "loss": 1.1979, "step": 20667 }, { "epoch": 51.67, "grad_norm": 8.75, "learning_rate": 9.944941437861511e-06, "loss": 1.2793, "step": 20668 }, { "epoch": 51.6725, "grad_norm": 7.3125, "learning_rate": 9.944131761487363e-06, "loss": 1.48, "step": 20669 }, { "epoch": 51.675, "grad_norm": 6.3125, "learning_rate": 9.943322085479484e-06, "loss": 1.48, "step": 20670 }, { "epoch": 51.6775, "grad_norm": 19.375, "learning_rate": 9.942512409843183e-06, "loss": 1.2865, "step": 20671 }, { "epoch": 51.68, "grad_norm": 9.0625, "learning_rate": 9.941702734583771e-06, "loss": 1.298, "step": 20672 }, { "epoch": 51.6825, "grad_norm": 8.625, "learning_rate": 9.940893059706547e-06, "loss": 1.2847, "step": 20673 }, { "epoch": 51.685, "grad_norm": 7.96875, "learning_rate": 9.940083385216829e-06, "loss": 1.3289, "step": 20674 }, { "epoch": 51.6875, "grad_norm": 9.6875, "learning_rate": 9.939273711119922e-06, "loss": 1.3586, "step": 20675 }, { "epoch": 51.69, "grad_norm": 5.78125, "learning_rate": 9.938464037421138e-06, "loss": 1.2799, "step": 20676 }, { "epoch": 51.6925, "grad_norm": 10.375, "learning_rate": 9.937654364125777e-06, "loss": 1.3257, "step": 20677 }, { "epoch": 51.695, "grad_norm": 6.59375, "learning_rate": 9.93684469123915e-06, "loss": 1.3526, "step": 20678 }, { "epoch": 51.6975, "grad_norm": 7.625, "learning_rate": 9.936035018766567e-06, "loss": 1.3799, "step": 20679 }, { "epoch": 51.7, "grad_norm": 7.4375, "learning_rate": 9.935225346713341e-06, "loss": 1.1668, "step": 20680 }, { "epoch": 51.7025, "grad_norm": 10.25, "learning_rate": 9.934415675084771e-06, "loss": 1.3437, "step": 20681 }, { "epoch": 51.705, "grad_norm": 6.1875, "learning_rate": 9.93360600388617e-06, "loss": 1.4201, "step": 20682 }, { "epoch": 51.7075, "grad_norm": 9.0625, "learning_rate": 9.932796333122842e-06, "loss": 1.379, "step": 20683 }, { "epoch": 51.71, "grad_norm": 7.25, "learning_rate": 9.931986662800106e-06, "loss": 1.3406, "step": 20684 }, { "epoch": 51.7125, "grad_norm": 8.5625, "learning_rate": 9.931176992923258e-06, "loss": 1.4585, "step": 20685 }, { "epoch": 51.715, "grad_norm": 6.5625, "learning_rate": 9.93036732349761e-06, "loss": 1.3456, "step": 20686 }, { "epoch": 51.7175, "grad_norm": 7.09375, "learning_rate": 9.929557654528474e-06, "loss": 1.3139, "step": 20687 }, { "epoch": 51.72, "grad_norm": 5.625, "learning_rate": 9.928747986021153e-06, "loss": 1.1833, "step": 20688 }, { "epoch": 51.7225, "grad_norm": 8.75, "learning_rate": 9.927938317980957e-06, "loss": 1.2546, "step": 20689 }, { "epoch": 51.725, "grad_norm": 7.5, "learning_rate": 9.927128650413195e-06, "loss": 1.4005, "step": 20690 }, { "epoch": 51.7275, "grad_norm": 8.0, "learning_rate": 9.926318983323177e-06, "loss": 1.3025, "step": 20691 }, { "epoch": 51.73, "grad_norm": 8.625, "learning_rate": 9.925509316716208e-06, "loss": 1.5152, "step": 20692 }, { "epoch": 51.7325, "grad_norm": 8.625, "learning_rate": 9.924699650597593e-06, "loss": 1.2522, "step": 20693 }, { "epoch": 51.735, "grad_norm": 7.75, "learning_rate": 9.923889984972647e-06, "loss": 1.1885, "step": 20694 }, { "epoch": 51.7375, "grad_norm": 12.3125, "learning_rate": 9.923080319846678e-06, "loss": 1.4547, "step": 20695 }, { "epoch": 51.74, "grad_norm": 7.59375, "learning_rate": 9.922270655224989e-06, "loss": 1.2779, "step": 20696 }, { "epoch": 51.7425, "grad_norm": 8.125, "learning_rate": 9.921460991112891e-06, "loss": 1.2431, "step": 20697 }, { "epoch": 51.745, "grad_norm": 6.34375, "learning_rate": 9.920651327515686e-06, "loss": 1.3126, "step": 20698 }, { "epoch": 51.7475, "grad_norm": 5.875, "learning_rate": 9.919841664438693e-06, "loss": 1.128, "step": 20699 }, { "epoch": 51.75, "grad_norm": 7.9375, "learning_rate": 9.919032001887215e-06, "loss": 1.1342, "step": 20700 }, { "epoch": 51.7525, "grad_norm": 5.8125, "learning_rate": 9.91822233986656e-06, "loss": 1.2267, "step": 20701 }, { "epoch": 51.755, "grad_norm": 6.59375, "learning_rate": 9.917412678382037e-06, "loss": 1.236, "step": 20702 }, { "epoch": 51.7575, "grad_norm": 7.9375, "learning_rate": 9.916603017438949e-06, "loss": 1.4164, "step": 20703 }, { "epoch": 51.76, "grad_norm": 9.5625, "learning_rate": 9.91579335704261e-06, "loss": 1.3366, "step": 20704 }, { "epoch": 51.7625, "grad_norm": 8.375, "learning_rate": 9.914983697198327e-06, "loss": 1.437, "step": 20705 }, { "epoch": 51.765, "grad_norm": 6.59375, "learning_rate": 9.914174037911409e-06, "loss": 1.3922, "step": 20706 }, { "epoch": 51.7675, "grad_norm": 7.125, "learning_rate": 9.91336437918716e-06, "loss": 1.3886, "step": 20707 }, { "epoch": 51.77, "grad_norm": 6.71875, "learning_rate": 9.912554721030892e-06, "loss": 1.4417, "step": 20708 }, { "epoch": 51.7725, "grad_norm": 5.5625, "learning_rate": 9.911745063447911e-06, "loss": 1.4495, "step": 20709 }, { "epoch": 51.775, "grad_norm": 11.75, "learning_rate": 9.910935406443528e-06, "loss": 1.3264, "step": 20710 }, { "epoch": 51.7775, "grad_norm": 9.1875, "learning_rate": 9.910125750023047e-06, "loss": 1.4593, "step": 20711 }, { "epoch": 51.78, "grad_norm": 10.5, "learning_rate": 9.90931609419178e-06, "loss": 1.3892, "step": 20712 }, { "epoch": 51.7825, "grad_norm": 7.03125, "learning_rate": 9.908506438955026e-06, "loss": 1.1919, "step": 20713 }, { "epoch": 51.785, "grad_norm": 6.65625, "learning_rate": 9.907696784318109e-06, "loss": 1.3043, "step": 20714 }, { "epoch": 51.7875, "grad_norm": 9.375, "learning_rate": 9.906887130286325e-06, "loss": 1.2605, "step": 20715 }, { "epoch": 51.79, "grad_norm": 7.90625, "learning_rate": 9.906077476864986e-06, "loss": 1.3495, "step": 20716 }, { "epoch": 51.7925, "grad_norm": 5.15625, "learning_rate": 9.905267824059397e-06, "loss": 1.3382, "step": 20717 }, { "epoch": 51.795, "grad_norm": 9.4375, "learning_rate": 9.904458171874867e-06, "loss": 1.3813, "step": 20718 }, { "epoch": 51.7975, "grad_norm": 8.0625, "learning_rate": 9.90364852031671e-06, "loss": 1.4627, "step": 20719 }, { "epoch": 51.8, "grad_norm": 8.375, "learning_rate": 9.90283886939023e-06, "loss": 1.283, "step": 20720 }, { "epoch": 51.8025, "grad_norm": 7.71875, "learning_rate": 9.90202921910073e-06, "loss": 1.332, "step": 20721 }, { "epoch": 51.805, "grad_norm": 9.0, "learning_rate": 9.901219569453521e-06, "loss": 1.4247, "step": 20722 }, { "epoch": 51.8075, "grad_norm": 7.53125, "learning_rate": 9.900409920453918e-06, "loss": 1.23, "step": 20723 }, { "epoch": 51.81, "grad_norm": 6.8125, "learning_rate": 9.899600272107223e-06, "loss": 1.2807, "step": 20724 }, { "epoch": 51.8125, "grad_norm": 4.71875, "learning_rate": 9.898790624418743e-06, "loss": 1.3491, "step": 20725 }, { "epoch": 51.815, "grad_norm": 6.71875, "learning_rate": 9.897980977393787e-06, "loss": 1.3537, "step": 20726 }, { "epoch": 51.8175, "grad_norm": 11.4375, "learning_rate": 9.897171331037662e-06, "loss": 1.3865, "step": 20727 }, { "epoch": 51.82, "grad_norm": 8.125, "learning_rate": 9.896361685355681e-06, "loss": 1.299, "step": 20728 }, { "epoch": 51.8225, "grad_norm": 8.625, "learning_rate": 9.895552040353149e-06, "loss": 1.3132, "step": 20729 }, { "epoch": 51.825, "grad_norm": 10.75, "learning_rate": 9.894742396035372e-06, "loss": 1.3906, "step": 20730 }, { "epoch": 51.8275, "grad_norm": 7.375, "learning_rate": 9.89393275240766e-06, "loss": 1.3858, "step": 20731 }, { "epoch": 51.83, "grad_norm": 7.625, "learning_rate": 9.893123109475319e-06, "loss": 1.4017, "step": 20732 }, { "epoch": 51.8325, "grad_norm": 7.40625, "learning_rate": 9.892313467243661e-06, "loss": 1.3104, "step": 20733 }, { "epoch": 51.835, "grad_norm": 5.75, "learning_rate": 9.89150382571799e-06, "loss": 1.3063, "step": 20734 }, { "epoch": 51.8375, "grad_norm": 10.8125, "learning_rate": 9.890694184903618e-06, "loss": 1.4181, "step": 20735 }, { "epoch": 51.84, "grad_norm": 9.9375, "learning_rate": 9.88988454480585e-06, "loss": 1.3157, "step": 20736 }, { "epoch": 51.8425, "grad_norm": 9.8125, "learning_rate": 9.889074905429991e-06, "loss": 1.298, "step": 20737 }, { "epoch": 51.845, "grad_norm": 4.40625, "learning_rate": 9.888265266781354e-06, "loss": 1.4406, "step": 20738 }, { "epoch": 51.8475, "grad_norm": 7.25, "learning_rate": 9.88745562886525e-06, "loss": 1.2942, "step": 20739 }, { "epoch": 51.85, "grad_norm": 5.84375, "learning_rate": 9.886645991686977e-06, "loss": 1.3771, "step": 20740 }, { "epoch": 51.8525, "grad_norm": 7.78125, "learning_rate": 9.88583635525185e-06, "loss": 1.2104, "step": 20741 }, { "epoch": 51.855, "grad_norm": 7.28125, "learning_rate": 9.885026719565172e-06, "loss": 1.3143, "step": 20742 }, { "epoch": 51.8575, "grad_norm": 9.125, "learning_rate": 9.88421708463226e-06, "loss": 1.3021, "step": 20743 }, { "epoch": 51.86, "grad_norm": 10.375, "learning_rate": 9.883407450458415e-06, "loss": 1.2038, "step": 20744 }, { "epoch": 51.8625, "grad_norm": 10.6875, "learning_rate": 9.882597817048944e-06, "loss": 1.329, "step": 20745 }, { "epoch": 51.865, "grad_norm": 6.28125, "learning_rate": 9.88178818440916e-06, "loss": 1.2251, "step": 20746 }, { "epoch": 51.8675, "grad_norm": 8.25, "learning_rate": 9.880978552544363e-06, "loss": 1.2237, "step": 20747 }, { "epoch": 51.87, "grad_norm": 8.9375, "learning_rate": 9.880168921459869e-06, "loss": 1.2539, "step": 20748 }, { "epoch": 51.8725, "grad_norm": 7.78125, "learning_rate": 9.879359291160982e-06, "loss": 1.3482, "step": 20749 }, { "epoch": 51.875, "grad_norm": 14.3125, "learning_rate": 9.878549661653013e-06, "loss": 1.3542, "step": 20750 }, { "epoch": 51.8775, "grad_norm": 6.84375, "learning_rate": 9.877740032941265e-06, "loss": 1.2937, "step": 20751 }, { "epoch": 51.88, "grad_norm": 9.8125, "learning_rate": 9.876930405031047e-06, "loss": 1.3413, "step": 20752 }, { "epoch": 51.8825, "grad_norm": 9.3125, "learning_rate": 9.876120777927671e-06, "loss": 1.1747, "step": 20753 }, { "epoch": 51.885, "grad_norm": 8.375, "learning_rate": 9.875311151636443e-06, "loss": 1.3536, "step": 20754 }, { "epoch": 51.8875, "grad_norm": 8.5625, "learning_rate": 9.874501526162668e-06, "loss": 1.3634, "step": 20755 }, { "epoch": 51.89, "grad_norm": 4.8125, "learning_rate": 9.873691901511657e-06, "loss": 1.3273, "step": 20756 }, { "epoch": 51.8925, "grad_norm": 10.0625, "learning_rate": 9.872882277688714e-06, "loss": 1.329, "step": 20757 }, { "epoch": 51.895, "grad_norm": 10.75, "learning_rate": 9.872072654699155e-06, "loss": 1.2573, "step": 20758 }, { "epoch": 51.8975, "grad_norm": 11.75, "learning_rate": 9.87126303254828e-06, "loss": 1.3842, "step": 20759 }, { "epoch": 51.9, "grad_norm": 8.3125, "learning_rate": 9.870453411241399e-06, "loss": 1.3196, "step": 20760 }, { "epoch": 51.9025, "grad_norm": 8.5625, "learning_rate": 9.869643790783822e-06, "loss": 1.683, "step": 20761 }, { "epoch": 51.905, "grad_norm": 6.21875, "learning_rate": 9.868834171180852e-06, "loss": 1.5057, "step": 20762 }, { "epoch": 51.9075, "grad_norm": 7.375, "learning_rate": 9.8680245524378e-06, "loss": 1.2244, "step": 20763 }, { "epoch": 51.91, "grad_norm": 8.625, "learning_rate": 9.867214934559975e-06, "loss": 1.4996, "step": 20764 }, { "epoch": 51.9125, "grad_norm": 6.0, "learning_rate": 9.866405317552687e-06, "loss": 1.2496, "step": 20765 }, { "epoch": 51.915, "grad_norm": 7.125, "learning_rate": 9.865595701421237e-06, "loss": 1.304, "step": 20766 }, { "epoch": 51.9175, "grad_norm": 7.0, "learning_rate": 9.864786086170933e-06, "loss": 1.2687, "step": 20767 }, { "epoch": 51.92, "grad_norm": 12.1875, "learning_rate": 9.86397647180709e-06, "loss": 1.304, "step": 20768 }, { "epoch": 51.9225, "grad_norm": 5.8125, "learning_rate": 9.863166858335012e-06, "loss": 1.2717, "step": 20769 }, { "epoch": 51.925, "grad_norm": 8.25, "learning_rate": 9.862357245760006e-06, "loss": 1.1483, "step": 20770 }, { "epoch": 51.9275, "grad_norm": 6.03125, "learning_rate": 9.861547634087383e-06, "loss": 1.2765, "step": 20771 }, { "epoch": 51.93, "grad_norm": 4.96875, "learning_rate": 9.86073802332244e-06, "loss": 1.3061, "step": 20772 }, { "epoch": 51.9325, "grad_norm": 11.75, "learning_rate": 9.8599284134705e-06, "loss": 1.3891, "step": 20773 }, { "epoch": 51.935, "grad_norm": 7.84375, "learning_rate": 9.859118804536863e-06, "loss": 1.3778, "step": 20774 }, { "epoch": 51.9375, "grad_norm": 8.6875, "learning_rate": 9.858309196526836e-06, "loss": 1.2891, "step": 20775 }, { "epoch": 51.94, "grad_norm": 6.4375, "learning_rate": 9.857499589445731e-06, "loss": 1.3261, "step": 20776 }, { "epoch": 51.9425, "grad_norm": 5.9375, "learning_rate": 9.856689983298849e-06, "loss": 1.2747, "step": 20777 }, { "epoch": 51.945, "grad_norm": 11.0625, "learning_rate": 9.855880378091503e-06, "loss": 1.4383, "step": 20778 }, { "epoch": 51.9475, "grad_norm": 7.1875, "learning_rate": 9.855070773829e-06, "loss": 1.4687, "step": 20779 }, { "epoch": 51.95, "grad_norm": 6.125, "learning_rate": 9.854261170516648e-06, "loss": 1.3155, "step": 20780 }, { "epoch": 51.9525, "grad_norm": 9.0, "learning_rate": 9.853451568159752e-06, "loss": 1.4477, "step": 20781 }, { "epoch": 51.955, "grad_norm": 7.3125, "learning_rate": 9.852641966763623e-06, "loss": 1.2354, "step": 20782 }, { "epoch": 51.9575, "grad_norm": 8.9375, "learning_rate": 9.851832366333566e-06, "loss": 1.3854, "step": 20783 }, { "epoch": 51.96, "grad_norm": 8.125, "learning_rate": 9.851022766874892e-06, "loss": 1.294, "step": 20784 }, { "epoch": 51.9625, "grad_norm": 7.15625, "learning_rate": 9.850213168392908e-06, "loss": 1.3403, "step": 20785 }, { "epoch": 51.965, "grad_norm": 5.0625, "learning_rate": 9.84940357089292e-06, "loss": 1.3422, "step": 20786 }, { "epoch": 51.9675, "grad_norm": 5.90625, "learning_rate": 9.84859397438023e-06, "loss": 1.285, "step": 20787 }, { "epoch": 51.97, "grad_norm": 8.125, "learning_rate": 9.847784378860158e-06, "loss": 1.4094, "step": 20788 }, { "epoch": 51.9725, "grad_norm": 5.6875, "learning_rate": 9.846974784338004e-06, "loss": 1.462, "step": 20789 }, { "epoch": 51.975, "grad_norm": 7.21875, "learning_rate": 9.846165190819077e-06, "loss": 1.2312, "step": 20790 }, { "epoch": 51.9775, "grad_norm": 6.59375, "learning_rate": 9.845355598308687e-06, "loss": 1.1745, "step": 20791 }, { "epoch": 51.98, "grad_norm": 5.6875, "learning_rate": 9.844546006812135e-06, "loss": 1.3369, "step": 20792 }, { "epoch": 51.9825, "grad_norm": 7.53125, "learning_rate": 9.843736416334734e-06, "loss": 1.2923, "step": 20793 }, { "epoch": 51.985, "grad_norm": 7.03125, "learning_rate": 9.842926826881796e-06, "loss": 1.2703, "step": 20794 }, { "epoch": 51.9875, "grad_norm": 7.0, "learning_rate": 9.84211723845862e-06, "loss": 1.3729, "step": 20795 }, { "epoch": 51.99, "grad_norm": 10.1875, "learning_rate": 9.841307651070516e-06, "loss": 1.2409, "step": 20796 }, { "epoch": 51.9925, "grad_norm": 7.625, "learning_rate": 9.840498064722792e-06, "loss": 1.4079, "step": 20797 }, { "epoch": 51.995, "grad_norm": 8.1875, "learning_rate": 9.839688479420759e-06, "loss": 1.3846, "step": 20798 }, { "epoch": 51.9975, "grad_norm": 7.5625, "learning_rate": 9.838878895169721e-06, "loss": 1.3618, "step": 20799 }, { "epoch": 52.0, "grad_norm": 5.5625, "learning_rate": 9.838069311974986e-06, "loss": 1.3029, "step": 20800 }, { "epoch": 52.0025, "grad_norm": 7.59375, "learning_rate": 9.837259729841864e-06, "loss": 1.2491, "step": 20801 }, { "epoch": 52.005, "grad_norm": 8.5625, "learning_rate": 9.836450148775657e-06, "loss": 1.4108, "step": 20802 }, { "epoch": 52.0075, "grad_norm": 7.71875, "learning_rate": 9.83564056878168e-06, "loss": 1.3154, "step": 20803 }, { "epoch": 52.01, "grad_norm": 10.4375, "learning_rate": 9.834830989865233e-06, "loss": 1.2584, "step": 20804 }, { "epoch": 52.0125, "grad_norm": 7.125, "learning_rate": 9.834021412031634e-06, "loss": 1.3527, "step": 20805 }, { "epoch": 52.015, "grad_norm": 7.25, "learning_rate": 9.833211835286179e-06, "loss": 1.3261, "step": 20806 }, { "epoch": 52.0175, "grad_norm": 9.25, "learning_rate": 9.832402259634179e-06, "loss": 1.3393, "step": 20807 }, { "epoch": 52.02, "grad_norm": 7.28125, "learning_rate": 9.831592685080945e-06, "loss": 1.1924, "step": 20808 }, { "epoch": 52.0225, "grad_norm": 6.78125, "learning_rate": 9.830783111631784e-06, "loss": 1.2567, "step": 20809 }, { "epoch": 52.025, "grad_norm": 8.8125, "learning_rate": 9.829973539292002e-06, "loss": 1.4061, "step": 20810 }, { "epoch": 52.0275, "grad_norm": 8.625, "learning_rate": 9.829163968066904e-06, "loss": 1.3381, "step": 20811 }, { "epoch": 52.03, "grad_norm": 8.75, "learning_rate": 9.828354397961801e-06, "loss": 1.3734, "step": 20812 }, { "epoch": 52.0325, "grad_norm": 8.75, "learning_rate": 9.827544828982003e-06, "loss": 1.3673, "step": 20813 }, { "epoch": 52.035, "grad_norm": 7.03125, "learning_rate": 9.826735261132812e-06, "loss": 1.3335, "step": 20814 }, { "epoch": 52.0375, "grad_norm": 7.65625, "learning_rate": 9.825925694419537e-06, "loss": 1.3817, "step": 20815 }, { "epoch": 52.04, "grad_norm": 6.9375, "learning_rate": 9.825116128847488e-06, "loss": 1.2403, "step": 20816 }, { "epoch": 52.0425, "grad_norm": 7.875, "learning_rate": 9.824306564421968e-06, "loss": 1.4634, "step": 20817 }, { "epoch": 52.045, "grad_norm": 6.9375, "learning_rate": 9.823497001148287e-06, "loss": 1.2597, "step": 20818 }, { "epoch": 52.0475, "grad_norm": 6.375, "learning_rate": 9.822687439031755e-06, "loss": 1.3436, "step": 20819 }, { "epoch": 52.05, "grad_norm": 8.0625, "learning_rate": 9.821877878077678e-06, "loss": 1.3063, "step": 20820 }, { "epoch": 52.0525, "grad_norm": 5.71875, "learning_rate": 9.821068318291362e-06, "loss": 1.3015, "step": 20821 }, { "epoch": 52.055, "grad_norm": 10.75, "learning_rate": 9.820258759678109e-06, "loss": 1.3843, "step": 20822 }, { "epoch": 52.0575, "grad_norm": 4.90625, "learning_rate": 9.819449202243238e-06, "loss": 1.3281, "step": 20823 }, { "epoch": 52.06, "grad_norm": 7.3125, "learning_rate": 9.818639645992052e-06, "loss": 1.4851, "step": 20824 }, { "epoch": 52.0625, "grad_norm": 11.1875, "learning_rate": 9.817830090929856e-06, "loss": 1.3244, "step": 20825 }, { "epoch": 52.065, "grad_norm": 6.1875, "learning_rate": 9.817020537061956e-06, "loss": 1.2352, "step": 20826 }, { "epoch": 52.0675, "grad_norm": 6.59375, "learning_rate": 9.816210984393661e-06, "loss": 1.3288, "step": 20827 }, { "epoch": 52.07, "grad_norm": 7.96875, "learning_rate": 9.815401432930285e-06, "loss": 1.1847, "step": 20828 }, { "epoch": 52.0725, "grad_norm": 7.0, "learning_rate": 9.814591882677127e-06, "loss": 1.212, "step": 20829 }, { "epoch": 52.075, "grad_norm": 8.6875, "learning_rate": 9.813782333639496e-06, "loss": 1.3498, "step": 20830 }, { "epoch": 52.0775, "grad_norm": 7.0625, "learning_rate": 9.812972785822703e-06, "loss": 1.2312, "step": 20831 }, { "epoch": 52.08, "grad_norm": 7.125, "learning_rate": 9.812163239232051e-06, "loss": 1.4895, "step": 20832 }, { "epoch": 52.0825, "grad_norm": 9.5, "learning_rate": 9.81135369387285e-06, "loss": 1.3487, "step": 20833 }, { "epoch": 52.085, "grad_norm": 8.9375, "learning_rate": 9.810544149750406e-06, "loss": 1.2336, "step": 20834 }, { "epoch": 52.0875, "grad_norm": 5.75, "learning_rate": 9.80973460687003e-06, "loss": 1.2937, "step": 20835 }, { "epoch": 52.09, "grad_norm": 7.875, "learning_rate": 9.808925065237023e-06, "loss": 1.4579, "step": 20836 }, { "epoch": 52.0925, "grad_norm": 9.3125, "learning_rate": 9.808115524856694e-06, "loss": 1.2385, "step": 20837 }, { "epoch": 52.095, "grad_norm": 8.5625, "learning_rate": 9.807305985734354e-06, "loss": 1.3498, "step": 20838 }, { "epoch": 52.0975, "grad_norm": 6.21875, "learning_rate": 9.80649644787531e-06, "loss": 1.4204, "step": 20839 }, { "epoch": 52.1, "grad_norm": 11.5, "learning_rate": 9.805686911284867e-06, "loss": 1.5064, "step": 20840 }, { "epoch": 52.1025, "grad_norm": 8.5625, "learning_rate": 9.804877375968335e-06, "loss": 1.4036, "step": 20841 }, { "epoch": 52.105, "grad_norm": 8.0, "learning_rate": 9.804067841931012e-06, "loss": 1.3602, "step": 20842 }, { "epoch": 52.1075, "grad_norm": 22.375, "learning_rate": 9.803258309178217e-06, "loss": 1.3179, "step": 20843 }, { "epoch": 52.11, "grad_norm": 5.5625, "learning_rate": 9.802448777715253e-06, "loss": 1.3696, "step": 20844 }, { "epoch": 52.1125, "grad_norm": 8.1875, "learning_rate": 9.801639247547425e-06, "loss": 1.3735, "step": 20845 }, { "epoch": 52.115, "grad_norm": 8.5, "learning_rate": 9.800829718680046e-06, "loss": 1.4772, "step": 20846 }, { "epoch": 52.1175, "grad_norm": 11.3125, "learning_rate": 9.800020191118415e-06, "loss": 1.335, "step": 20847 }, { "epoch": 52.12, "grad_norm": 6.96875, "learning_rate": 9.799210664867844e-06, "loss": 1.3448, "step": 20848 }, { "epoch": 52.1225, "grad_norm": 8.125, "learning_rate": 9.798401139933641e-06, "loss": 1.1757, "step": 20849 }, { "epoch": 52.125, "grad_norm": 7.90625, "learning_rate": 9.797591616321115e-06, "loss": 1.3523, "step": 20850 }, { "epoch": 52.1275, "grad_norm": 8.4375, "learning_rate": 9.796782094035568e-06, "loss": 1.3304, "step": 20851 }, { "epoch": 52.13, "grad_norm": 9.375, "learning_rate": 9.795972573082305e-06, "loss": 1.2923, "step": 20852 }, { "epoch": 52.1325, "grad_norm": 17.25, "learning_rate": 9.795163053466642e-06, "loss": 1.3292, "step": 20853 }, { "epoch": 52.135, "grad_norm": 11.875, "learning_rate": 9.794353535193884e-06, "loss": 1.33, "step": 20854 }, { "epoch": 52.1375, "grad_norm": 12.1875, "learning_rate": 9.793544018269332e-06, "loss": 1.2022, "step": 20855 }, { "epoch": 52.14, "grad_norm": 5.8125, "learning_rate": 9.792734502698301e-06, "loss": 1.3096, "step": 20856 }, { "epoch": 52.1425, "grad_norm": 9.0, "learning_rate": 9.791924988486088e-06, "loss": 1.4242, "step": 20857 }, { "epoch": 52.145, "grad_norm": 6.46875, "learning_rate": 9.791115475638012e-06, "loss": 1.2646, "step": 20858 }, { "epoch": 52.1475, "grad_norm": 8.5625, "learning_rate": 9.790305964159373e-06, "loss": 1.4648, "step": 20859 }, { "epoch": 52.15, "grad_norm": 10.4375, "learning_rate": 9.789496454055482e-06, "loss": 1.4094, "step": 20860 }, { "epoch": 52.1525, "grad_norm": 6.46875, "learning_rate": 9.788686945331641e-06, "loss": 1.3339, "step": 20861 }, { "epoch": 52.155, "grad_norm": 7.8125, "learning_rate": 9.787877437993158e-06, "loss": 1.2435, "step": 20862 }, { "epoch": 52.1575, "grad_norm": 7.25, "learning_rate": 9.787067932045343e-06, "loss": 1.3711, "step": 20863 }, { "epoch": 52.16, "grad_norm": 5.59375, "learning_rate": 9.786258427493505e-06, "loss": 1.3176, "step": 20864 }, { "epoch": 52.1625, "grad_norm": 7.5625, "learning_rate": 9.785448924342948e-06, "loss": 1.2608, "step": 20865 }, { "epoch": 52.165, "grad_norm": 7.875, "learning_rate": 9.784639422598977e-06, "loss": 1.3402, "step": 20866 }, { "epoch": 52.1675, "grad_norm": 7.65625, "learning_rate": 9.7838299222669e-06, "loss": 1.4383, "step": 20867 }, { "epoch": 52.17, "grad_norm": 8.0625, "learning_rate": 9.78302042335203e-06, "loss": 1.2676, "step": 20868 }, { "epoch": 52.1725, "grad_norm": 11.8125, "learning_rate": 9.782210925859666e-06, "loss": 1.3745, "step": 20869 }, { "epoch": 52.175, "grad_norm": 5.90625, "learning_rate": 9.78140142979512e-06, "loss": 1.305, "step": 20870 }, { "epoch": 52.1775, "grad_norm": 8.375, "learning_rate": 9.7805919351637e-06, "loss": 1.2001, "step": 20871 }, { "epoch": 52.18, "grad_norm": 7.65625, "learning_rate": 9.779782441970702e-06, "loss": 1.373, "step": 20872 }, { "epoch": 52.1825, "grad_norm": 6.125, "learning_rate": 9.77897295022145e-06, "loss": 1.407, "step": 20873 }, { "epoch": 52.185, "grad_norm": 7.375, "learning_rate": 9.778163459921238e-06, "loss": 1.2414, "step": 20874 }, { "epoch": 52.1875, "grad_norm": 11.1875, "learning_rate": 9.777353971075381e-06, "loss": 1.4495, "step": 20875 }, { "epoch": 52.19, "grad_norm": 9.375, "learning_rate": 9.776544483689182e-06, "loss": 1.2812, "step": 20876 }, { "epoch": 52.1925, "grad_norm": 9.3125, "learning_rate": 9.775734997767944e-06, "loss": 1.3008, "step": 20877 }, { "epoch": 52.195, "grad_norm": 6.40625, "learning_rate": 9.77492551331698e-06, "loss": 1.204, "step": 20878 }, { "epoch": 52.1975, "grad_norm": 6.90625, "learning_rate": 9.7741160303416e-06, "loss": 1.3239, "step": 20879 }, { "epoch": 52.2, "grad_norm": 6.84375, "learning_rate": 9.773306548847102e-06, "loss": 1.3946, "step": 20880 }, { "epoch": 52.2025, "grad_norm": 5.96875, "learning_rate": 9.772497068838798e-06, "loss": 1.4616, "step": 20881 }, { "epoch": 52.205, "grad_norm": 7.34375, "learning_rate": 9.77168759032199e-06, "loss": 1.1947, "step": 20882 }, { "epoch": 52.2075, "grad_norm": 10.0625, "learning_rate": 9.770878113301996e-06, "loss": 1.3216, "step": 20883 }, { "epoch": 52.21, "grad_norm": 7.09375, "learning_rate": 9.770068637784114e-06, "loss": 1.5561, "step": 20884 }, { "epoch": 52.2125, "grad_norm": 8.0625, "learning_rate": 9.769259163773652e-06, "loss": 1.2479, "step": 20885 }, { "epoch": 52.215, "grad_norm": 6.21875, "learning_rate": 9.768449691275918e-06, "loss": 1.2888, "step": 20886 }, { "epoch": 52.2175, "grad_norm": 6.625, "learning_rate": 9.767640220296216e-06, "loss": 1.3603, "step": 20887 }, { "epoch": 52.22, "grad_norm": 10.375, "learning_rate": 9.766830750839857e-06, "loss": 1.6008, "step": 20888 }, { "epoch": 52.2225, "grad_norm": 10.6875, "learning_rate": 9.766021282912147e-06, "loss": 1.2351, "step": 20889 }, { "epoch": 52.225, "grad_norm": 8.3125, "learning_rate": 9.765211816518393e-06, "loss": 1.2135, "step": 20890 }, { "epoch": 52.2275, "grad_norm": 7.84375, "learning_rate": 9.7644023516639e-06, "loss": 1.3403, "step": 20891 }, { "epoch": 52.23, "grad_norm": 6.71875, "learning_rate": 9.763592888353975e-06, "loss": 1.2206, "step": 20892 }, { "epoch": 52.2325, "grad_norm": 7.96875, "learning_rate": 9.762783426593927e-06, "loss": 1.4077, "step": 20893 }, { "epoch": 52.235, "grad_norm": 6.375, "learning_rate": 9.761973966389061e-06, "loss": 1.1542, "step": 20894 }, { "epoch": 52.2375, "grad_norm": 7.5625, "learning_rate": 9.761164507744683e-06, "loss": 1.311, "step": 20895 }, { "epoch": 52.24, "grad_norm": 5.46875, "learning_rate": 9.760355050666102e-06, "loss": 1.3675, "step": 20896 }, { "epoch": 52.2425, "grad_norm": 8.75, "learning_rate": 9.75954559515862e-06, "loss": 1.2746, "step": 20897 }, { "epoch": 52.245, "grad_norm": 6.96875, "learning_rate": 9.758736141227554e-06, "loss": 1.2823, "step": 20898 }, { "epoch": 52.2475, "grad_norm": 10.6875, "learning_rate": 9.7579266888782e-06, "loss": 1.3031, "step": 20899 }, { "epoch": 52.25, "grad_norm": 8.0625, "learning_rate": 9.757117238115871e-06, "loss": 1.2726, "step": 20900 }, { "epoch": 52.2525, "grad_norm": 7.625, "learning_rate": 9.756307788945873e-06, "loss": 1.2426, "step": 20901 }, { "epoch": 52.255, "grad_norm": 6.84375, "learning_rate": 9.755498341373507e-06, "loss": 1.2428, "step": 20902 }, { "epoch": 52.2575, "grad_norm": 8.625, "learning_rate": 9.754688895404084e-06, "loss": 1.4321, "step": 20903 }, { "epoch": 52.26, "grad_norm": 10.4375, "learning_rate": 9.753879451042914e-06, "loss": 1.448, "step": 20904 }, { "epoch": 52.2625, "grad_norm": 8.875, "learning_rate": 9.7530700082953e-06, "loss": 1.176, "step": 20905 }, { "epoch": 52.265, "grad_norm": 8.0625, "learning_rate": 9.752260567166548e-06, "loss": 1.5272, "step": 20906 }, { "epoch": 52.2675, "grad_norm": 8.5625, "learning_rate": 9.751451127661966e-06, "loss": 1.2584, "step": 20907 }, { "epoch": 52.27, "grad_norm": 8.4375, "learning_rate": 9.750641689786859e-06, "loss": 1.3448, "step": 20908 }, { "epoch": 52.2725, "grad_norm": 7.125, "learning_rate": 9.74983225354654e-06, "loss": 1.2627, "step": 20909 }, { "epoch": 52.275, "grad_norm": 7.6875, "learning_rate": 9.749022818946307e-06, "loss": 1.2954, "step": 20910 }, { "epoch": 52.2775, "grad_norm": 7.09375, "learning_rate": 9.748213385991472e-06, "loss": 1.3329, "step": 20911 }, { "epoch": 52.28, "grad_norm": 9.5, "learning_rate": 9.747403954687334e-06, "loss": 1.2738, "step": 20912 }, { "epoch": 52.2825, "grad_norm": 11.4375, "learning_rate": 9.746594525039213e-06, "loss": 1.2349, "step": 20913 }, { "epoch": 52.285, "grad_norm": 6.0, "learning_rate": 9.745785097052406e-06, "loss": 1.3694, "step": 20914 }, { "epoch": 52.2875, "grad_norm": 7.125, "learning_rate": 9.74497567073222e-06, "loss": 1.3979, "step": 20915 }, { "epoch": 52.29, "grad_norm": 9.3125, "learning_rate": 9.744166246083967e-06, "loss": 1.4565, "step": 20916 }, { "epoch": 52.2925, "grad_norm": 5.5625, "learning_rate": 9.743356823112945e-06, "loss": 1.2924, "step": 20917 }, { "epoch": 52.295, "grad_norm": 9.0, "learning_rate": 9.742547401824469e-06, "loss": 1.2742, "step": 20918 }, { "epoch": 52.2975, "grad_norm": 7.09375, "learning_rate": 9.74173798222384e-06, "loss": 1.1688, "step": 20919 }, { "epoch": 52.3, "grad_norm": 5.8125, "learning_rate": 9.740928564316369e-06, "loss": 1.4098, "step": 20920 }, { "epoch": 52.3025, "grad_norm": 8.5, "learning_rate": 9.740119148107359e-06, "loss": 1.2588, "step": 20921 }, { "epoch": 52.305, "grad_norm": 6.9375, "learning_rate": 9.739309733602113e-06, "loss": 1.1393, "step": 20922 }, { "epoch": 52.3075, "grad_norm": 8.5, "learning_rate": 9.738500320805946e-06, "loss": 1.3653, "step": 20923 }, { "epoch": 52.31, "grad_norm": 7.21875, "learning_rate": 9.737690909724162e-06, "loss": 1.2075, "step": 20924 }, { "epoch": 52.3125, "grad_norm": 8.6875, "learning_rate": 9.736881500362064e-06, "loss": 1.333, "step": 20925 }, { "epoch": 52.315, "grad_norm": 5.15625, "learning_rate": 9.736072092724963e-06, "loss": 1.3209, "step": 20926 }, { "epoch": 52.3175, "grad_norm": 5.28125, "learning_rate": 9.735262686818154e-06, "loss": 1.2833, "step": 20927 }, { "epoch": 52.32, "grad_norm": 7.1875, "learning_rate": 9.734453282646962e-06, "loss": 1.1798, "step": 20928 }, { "epoch": 52.3225, "grad_norm": 5.4375, "learning_rate": 9.733643880216678e-06, "loss": 1.507, "step": 20929 }, { "epoch": 52.325, "grad_norm": 7.1875, "learning_rate": 9.73283447953262e-06, "loss": 1.4355, "step": 20930 }, { "epoch": 52.3275, "grad_norm": 7.5, "learning_rate": 9.732025080600084e-06, "loss": 1.3055, "step": 20931 }, { "epoch": 52.33, "grad_norm": 5.78125, "learning_rate": 9.731215683424378e-06, "loss": 1.2222, "step": 20932 }, { "epoch": 52.3325, "grad_norm": 5.96875, "learning_rate": 9.730406288010815e-06, "loss": 1.2421, "step": 20933 }, { "epoch": 52.335, "grad_norm": 12.5, "learning_rate": 9.7295968943647e-06, "loss": 1.2666, "step": 20934 }, { "epoch": 52.3375, "grad_norm": 6.5625, "learning_rate": 9.728787502491334e-06, "loss": 1.425, "step": 20935 }, { "epoch": 52.34, "grad_norm": 12.5625, "learning_rate": 9.727978112396026e-06, "loss": 1.3457, "step": 20936 }, { "epoch": 52.3425, "grad_norm": 7.09375, "learning_rate": 9.72716872408408e-06, "loss": 1.2528, "step": 20937 }, { "epoch": 52.345, "grad_norm": 7.03125, "learning_rate": 9.72635933756081e-06, "loss": 1.364, "step": 20938 }, { "epoch": 52.3475, "grad_norm": 7.40625, "learning_rate": 9.725549952831516e-06, "loss": 1.5041, "step": 20939 }, { "epoch": 52.35, "grad_norm": 8.8125, "learning_rate": 9.724740569901503e-06, "loss": 1.4032, "step": 20940 }, { "epoch": 52.3525, "grad_norm": 10.8125, "learning_rate": 9.723931188776085e-06, "loss": 1.5156, "step": 20941 }, { "epoch": 52.355, "grad_norm": 10.25, "learning_rate": 9.723121809460554e-06, "loss": 1.2391, "step": 20942 }, { "epoch": 52.3575, "grad_norm": 7.1875, "learning_rate": 9.722312431960233e-06, "loss": 1.3418, "step": 20943 }, { "epoch": 52.36, "grad_norm": 7.125, "learning_rate": 9.721503056280418e-06, "loss": 1.4495, "step": 20944 }, { "epoch": 52.3625, "grad_norm": 7.125, "learning_rate": 9.720693682426421e-06, "loss": 1.286, "step": 20945 }, { "epoch": 52.365, "grad_norm": 5.40625, "learning_rate": 9.719884310403543e-06, "loss": 1.2248, "step": 20946 }, { "epoch": 52.3675, "grad_norm": 5.90625, "learning_rate": 9.719074940217089e-06, "loss": 1.2784, "step": 20947 }, { "epoch": 52.37, "grad_norm": 49.0, "learning_rate": 9.71826557187237e-06, "loss": 1.3533, "step": 20948 }, { "epoch": 52.3725, "grad_norm": 6.375, "learning_rate": 9.717456205374693e-06, "loss": 1.2739, "step": 20949 }, { "epoch": 52.375, "grad_norm": 6.125, "learning_rate": 9.71664684072936e-06, "loss": 1.3112, "step": 20950 }, { "epoch": 52.3775, "grad_norm": 5.65625, "learning_rate": 9.71583747794168e-06, "loss": 1.3213, "step": 20951 }, { "epoch": 52.38, "grad_norm": 7.625, "learning_rate": 9.715028117016955e-06, "loss": 1.2263, "step": 20952 }, { "epoch": 52.3825, "grad_norm": 6.4375, "learning_rate": 9.7142187579605e-06, "loss": 1.307, "step": 20953 }, { "epoch": 52.385, "grad_norm": 8.875, "learning_rate": 9.713409400777612e-06, "loss": 1.2471, "step": 20954 }, { "epoch": 52.3875, "grad_norm": 11.1875, "learning_rate": 9.7126000454736e-06, "loss": 1.4091, "step": 20955 }, { "epoch": 52.39, "grad_norm": 6.0625, "learning_rate": 9.711790692053773e-06, "loss": 1.2399, "step": 20956 }, { "epoch": 52.3925, "grad_norm": 6.78125, "learning_rate": 9.710981340523432e-06, "loss": 1.4195, "step": 20957 }, { "epoch": 52.395, "grad_norm": 6.25, "learning_rate": 9.710171990887886e-06, "loss": 1.3463, "step": 20958 }, { "epoch": 52.3975, "grad_norm": 6.65625, "learning_rate": 9.709362643152442e-06, "loss": 1.2864, "step": 20959 }, { "epoch": 52.4, "grad_norm": 8.5, "learning_rate": 9.708553297322407e-06, "loss": 1.3818, "step": 20960 }, { "epoch": 52.4025, "grad_norm": 5.78125, "learning_rate": 9.707743953403082e-06, "loss": 1.2654, "step": 20961 }, { "epoch": 52.405, "grad_norm": 6.46875, "learning_rate": 9.706934611399774e-06, "loss": 1.3295, "step": 20962 }, { "epoch": 52.4075, "grad_norm": 7.375, "learning_rate": 9.706125271317794e-06, "loss": 1.3104, "step": 20963 }, { "epoch": 52.41, "grad_norm": 11.25, "learning_rate": 9.705315933162445e-06, "loss": 1.401, "step": 20964 }, { "epoch": 52.4125, "grad_norm": 8.1875, "learning_rate": 9.704506596939032e-06, "loss": 1.4003, "step": 20965 }, { "epoch": 52.415, "grad_norm": 5.9375, "learning_rate": 9.70369726265286e-06, "loss": 1.2582, "step": 20966 }, { "epoch": 52.4175, "grad_norm": 8.25, "learning_rate": 9.70288793030924e-06, "loss": 1.3763, "step": 20967 }, { "epoch": 52.42, "grad_norm": 6.9375, "learning_rate": 9.702078599913475e-06, "loss": 1.3277, "step": 20968 }, { "epoch": 52.4225, "grad_norm": 8.4375, "learning_rate": 9.70126927147087e-06, "loss": 1.287, "step": 20969 }, { "epoch": 52.425, "grad_norm": 6.71875, "learning_rate": 9.700459944986732e-06, "loss": 1.2644, "step": 20970 }, { "epoch": 52.4275, "grad_norm": 9.75, "learning_rate": 9.699650620466362e-06, "loss": 1.448, "step": 20971 }, { "epoch": 52.43, "grad_norm": 7.65625, "learning_rate": 9.698841297915077e-06, "loss": 1.3335, "step": 20972 }, { "epoch": 52.4325, "grad_norm": 4.25, "learning_rate": 9.698031977338174e-06, "loss": 1.4196, "step": 20973 }, { "epoch": 52.435, "grad_norm": 7.90625, "learning_rate": 9.697222658740961e-06, "loss": 1.1423, "step": 20974 }, { "epoch": 52.4375, "grad_norm": 7.96875, "learning_rate": 9.696413342128747e-06, "loss": 1.4657, "step": 20975 }, { "epoch": 52.44, "grad_norm": 8.6875, "learning_rate": 9.69560402750683e-06, "loss": 1.407, "step": 20976 }, { "epoch": 52.4425, "grad_norm": 7.8125, "learning_rate": 9.694794714880523e-06, "loss": 1.3302, "step": 20977 }, { "epoch": 52.445, "grad_norm": 9.25, "learning_rate": 9.693985404255129e-06, "loss": 1.2252, "step": 20978 }, { "epoch": 52.4475, "grad_norm": 5.75, "learning_rate": 9.693176095635957e-06, "loss": 1.3381, "step": 20979 }, { "epoch": 52.45, "grad_norm": 8.5625, "learning_rate": 9.692366789028308e-06, "loss": 1.3573, "step": 20980 }, { "epoch": 52.4525, "grad_norm": 5.65625, "learning_rate": 9.691557484437488e-06, "loss": 1.2823, "step": 20981 }, { "epoch": 52.455, "grad_norm": 7.5, "learning_rate": 9.690748181868807e-06, "loss": 1.2519, "step": 20982 }, { "epoch": 52.4575, "grad_norm": 10.3125, "learning_rate": 9.68993888132757e-06, "loss": 1.2563, "step": 20983 }, { "epoch": 52.46, "grad_norm": 6.53125, "learning_rate": 9.689129582819081e-06, "loss": 1.2698, "step": 20984 }, { "epoch": 52.4625, "grad_norm": 8.125, "learning_rate": 9.688320286348644e-06, "loss": 1.4518, "step": 20985 }, { "epoch": 52.465, "grad_norm": 7.15625, "learning_rate": 9.687510991921567e-06, "loss": 1.2962, "step": 20986 }, { "epoch": 52.4675, "grad_norm": 5.75, "learning_rate": 9.686701699543157e-06, "loss": 1.3126, "step": 20987 }, { "epoch": 52.47, "grad_norm": 6.1875, "learning_rate": 9.685892409218718e-06, "loss": 1.413, "step": 20988 }, { "epoch": 52.4725, "grad_norm": 9.625, "learning_rate": 9.685083120953555e-06, "loss": 1.4287, "step": 20989 }, { "epoch": 52.475, "grad_norm": 9.625, "learning_rate": 9.684273834752976e-06, "loss": 1.2167, "step": 20990 }, { "epoch": 52.4775, "grad_norm": 8.25, "learning_rate": 9.683464550622283e-06, "loss": 1.3473, "step": 20991 }, { "epoch": 52.48, "grad_norm": 6.71875, "learning_rate": 9.682655268566783e-06, "loss": 1.468, "step": 20992 }, { "epoch": 52.4825, "grad_norm": 6.40625, "learning_rate": 9.681845988591784e-06, "loss": 1.3383, "step": 20993 }, { "epoch": 52.485, "grad_norm": 6.28125, "learning_rate": 9.681036710702592e-06, "loss": 1.326, "step": 20994 }, { "epoch": 52.4875, "grad_norm": 8.0625, "learning_rate": 9.680227434904508e-06, "loss": 1.3705, "step": 20995 }, { "epoch": 52.49, "grad_norm": 6.59375, "learning_rate": 9.679418161202838e-06, "loss": 1.2581, "step": 20996 }, { "epoch": 52.4925, "grad_norm": 8.75, "learning_rate": 9.678608889602891e-06, "loss": 1.2271, "step": 20997 }, { "epoch": 52.495, "grad_norm": 7.25, "learning_rate": 9.677799620109976e-06, "loss": 1.1973, "step": 20998 }, { "epoch": 52.4975, "grad_norm": 6.6875, "learning_rate": 9.67699035272939e-06, "loss": 1.5268, "step": 20999 }, { "epoch": 52.5, "grad_norm": 7.9375, "learning_rate": 9.676181087466444e-06, "loss": 1.3625, "step": 21000 }, { "epoch": 52.5025, "grad_norm": 8.5, "learning_rate": 9.675371824326436e-06, "loss": 1.3546, "step": 21001 }, { "epoch": 52.505, "grad_norm": 10.4375, "learning_rate": 9.674562563314684e-06, "loss": 1.3215, "step": 21002 }, { "epoch": 52.5075, "grad_norm": 9.5, "learning_rate": 9.673753304436484e-06, "loss": 1.3159, "step": 21003 }, { "epoch": 52.51, "grad_norm": 7.40625, "learning_rate": 9.672944047697147e-06, "loss": 1.3247, "step": 21004 }, { "epoch": 52.5125, "grad_norm": 6.96875, "learning_rate": 9.672134793101973e-06, "loss": 1.3324, "step": 21005 }, { "epoch": 52.515, "grad_norm": 10.9375, "learning_rate": 9.671325540656268e-06, "loss": 1.407, "step": 21006 }, { "epoch": 52.5175, "grad_norm": 7.96875, "learning_rate": 9.670516290365344e-06, "loss": 1.3532, "step": 21007 }, { "epoch": 52.52, "grad_norm": 5.6875, "learning_rate": 9.669707042234502e-06, "loss": 1.3272, "step": 21008 }, { "epoch": 52.5225, "grad_norm": 7.0625, "learning_rate": 9.668897796269046e-06, "loss": 1.4848, "step": 21009 }, { "epoch": 52.525, "grad_norm": 6.25, "learning_rate": 9.668088552474282e-06, "loss": 1.2293, "step": 21010 }, { "epoch": 52.5275, "grad_norm": 8.875, "learning_rate": 9.667279310855514e-06, "loss": 1.2832, "step": 21011 }, { "epoch": 52.53, "grad_norm": 5.3125, "learning_rate": 9.666470071418053e-06, "loss": 1.3502, "step": 21012 }, { "epoch": 52.5325, "grad_norm": 11.3125, "learning_rate": 9.665660834167201e-06, "loss": 1.4218, "step": 21013 }, { "epoch": 52.535, "grad_norm": 7.0, "learning_rate": 9.664851599108263e-06, "loss": 1.3362, "step": 21014 }, { "epoch": 52.5375, "grad_norm": 6.46875, "learning_rate": 9.664042366246546e-06, "loss": 1.4099, "step": 21015 }, { "epoch": 52.54, "grad_norm": 7.78125, "learning_rate": 9.663233135587347e-06, "loss": 1.4342, "step": 21016 }, { "epoch": 52.5425, "grad_norm": 7.6875, "learning_rate": 9.662423907135985e-06, "loss": 1.3852, "step": 21017 }, { "epoch": 52.545, "grad_norm": 7.40625, "learning_rate": 9.661614680897756e-06, "loss": 1.1574, "step": 21018 }, { "epoch": 52.5475, "grad_norm": 10.875, "learning_rate": 9.660805456877968e-06, "loss": 1.2842, "step": 21019 }, { "epoch": 52.55, "grad_norm": 6.5625, "learning_rate": 9.659996235081926e-06, "loss": 1.268, "step": 21020 }, { "epoch": 52.5525, "grad_norm": 8.0, "learning_rate": 9.659187015514933e-06, "loss": 1.371, "step": 21021 }, { "epoch": 52.555, "grad_norm": 10.5625, "learning_rate": 9.658377798182297e-06, "loss": 1.4389, "step": 21022 }, { "epoch": 52.5575, "grad_norm": 6.3125, "learning_rate": 9.657568583089327e-06, "loss": 1.3273, "step": 21023 }, { "epoch": 52.56, "grad_norm": 5.46875, "learning_rate": 9.656759370241318e-06, "loss": 1.2383, "step": 21024 }, { "epoch": 52.5625, "grad_norm": 9.0625, "learning_rate": 9.655950159643583e-06, "loss": 1.1179, "step": 21025 }, { "epoch": 52.565, "grad_norm": 9.6875, "learning_rate": 9.655140951301422e-06, "loss": 1.3493, "step": 21026 }, { "epoch": 52.5675, "grad_norm": 8.125, "learning_rate": 9.65433174522015e-06, "loss": 1.2944, "step": 21027 }, { "epoch": 52.57, "grad_norm": 7.09375, "learning_rate": 9.65352254140506e-06, "loss": 1.2354, "step": 21028 }, { "epoch": 52.5725, "grad_norm": 7.21875, "learning_rate": 9.652713339861464e-06, "loss": 1.4428, "step": 21029 }, { "epoch": 52.575, "grad_norm": 8.75, "learning_rate": 9.651904140594666e-06, "loss": 1.5019, "step": 21030 }, { "epoch": 52.5775, "grad_norm": 6.90625, "learning_rate": 9.651094943609967e-06, "loss": 1.4946, "step": 21031 }, { "epoch": 52.58, "grad_norm": 7.9375, "learning_rate": 9.650285748912678e-06, "loss": 1.3154, "step": 21032 }, { "epoch": 52.5825, "grad_norm": 8.0625, "learning_rate": 9.649476556508101e-06, "loss": 1.268, "step": 21033 }, { "epoch": 52.585, "grad_norm": 7.96875, "learning_rate": 9.648667366401546e-06, "loss": 1.4025, "step": 21034 }, { "epoch": 52.5875, "grad_norm": 5.53125, "learning_rate": 9.64785817859831e-06, "loss": 1.18, "step": 21035 }, { "epoch": 52.59, "grad_norm": 6.78125, "learning_rate": 9.6470489931037e-06, "loss": 1.2484, "step": 21036 }, { "epoch": 52.5925, "grad_norm": 7.25, "learning_rate": 9.646239809923024e-06, "loss": 1.4578, "step": 21037 }, { "epoch": 52.595, "grad_norm": 8.25, "learning_rate": 9.645430629061588e-06, "loss": 1.2811, "step": 21038 }, { "epoch": 52.5975, "grad_norm": 6.5, "learning_rate": 9.644621450524693e-06, "loss": 1.235, "step": 21039 }, { "epoch": 52.6, "grad_norm": 9.125, "learning_rate": 9.643812274317644e-06, "loss": 1.2363, "step": 21040 }, { "epoch": 52.6025, "grad_norm": 7.21875, "learning_rate": 9.643003100445746e-06, "loss": 1.2594, "step": 21041 }, { "epoch": 52.605, "grad_norm": 5.75, "learning_rate": 9.642193928914311e-06, "loss": 1.1921, "step": 21042 }, { "epoch": 52.6075, "grad_norm": 6.6875, "learning_rate": 9.641384759728636e-06, "loss": 1.3282, "step": 21043 }, { "epoch": 52.61, "grad_norm": 9.8125, "learning_rate": 9.640575592894027e-06, "loss": 1.2886, "step": 21044 }, { "epoch": 52.6125, "grad_norm": 5.46875, "learning_rate": 9.639766428415793e-06, "loss": 1.2726, "step": 21045 }, { "epoch": 52.615, "grad_norm": 6.375, "learning_rate": 9.63895726629923e-06, "loss": 1.3269, "step": 21046 }, { "epoch": 52.6175, "grad_norm": 6.375, "learning_rate": 9.638148106549653e-06, "loss": 1.21, "step": 21047 }, { "epoch": 52.62, "grad_norm": 8.4375, "learning_rate": 9.637338949172362e-06, "loss": 1.4098, "step": 21048 }, { "epoch": 52.6225, "grad_norm": 9.625, "learning_rate": 9.636529794172663e-06, "loss": 1.3376, "step": 21049 }, { "epoch": 52.625, "grad_norm": 10.6875, "learning_rate": 9.63572064155586e-06, "loss": 1.3754, "step": 21050 }, { "epoch": 52.6275, "grad_norm": 6.8125, "learning_rate": 9.634911491327255e-06, "loss": 1.2286, "step": 21051 }, { "epoch": 52.63, "grad_norm": 7.875, "learning_rate": 9.634102343492157e-06, "loss": 1.2488, "step": 21052 }, { "epoch": 52.6325, "grad_norm": 8.75, "learning_rate": 9.633293198055872e-06, "loss": 1.4204, "step": 21053 }, { "epoch": 52.635, "grad_norm": 8.1875, "learning_rate": 9.632484055023702e-06, "loss": 1.3528, "step": 21054 }, { "epoch": 52.6375, "grad_norm": 10.375, "learning_rate": 9.631674914400949e-06, "loss": 1.2593, "step": 21055 }, { "epoch": 52.64, "grad_norm": 6.84375, "learning_rate": 9.630865776192918e-06, "loss": 1.4665, "step": 21056 }, { "epoch": 52.6425, "grad_norm": 8.0625, "learning_rate": 9.630056640404922e-06, "loss": 1.4335, "step": 21057 }, { "epoch": 52.645, "grad_norm": 7.21875, "learning_rate": 9.629247507042258e-06, "loss": 1.2497, "step": 21058 }, { "epoch": 52.6475, "grad_norm": 7.78125, "learning_rate": 9.628438376110231e-06, "loss": 1.3621, "step": 21059 }, { "epoch": 52.65, "grad_norm": 5.25, "learning_rate": 9.627629247614151e-06, "loss": 1.1587, "step": 21060 }, { "epoch": 52.6525, "grad_norm": 6.875, "learning_rate": 9.626820121559313e-06, "loss": 1.3356, "step": 21061 }, { "epoch": 52.655, "grad_norm": 7.84375, "learning_rate": 9.626010997951031e-06, "loss": 1.4438, "step": 21062 }, { "epoch": 52.6575, "grad_norm": 8.0625, "learning_rate": 9.625201876794605e-06, "loss": 1.3823, "step": 21063 }, { "epoch": 52.66, "grad_norm": 6.625, "learning_rate": 9.624392758095344e-06, "loss": 1.3105, "step": 21064 }, { "epoch": 52.6625, "grad_norm": 4.96875, "learning_rate": 9.623583641858544e-06, "loss": 1.1946, "step": 21065 }, { "epoch": 52.665, "grad_norm": 7.8125, "learning_rate": 9.622774528089516e-06, "loss": 1.2778, "step": 21066 }, { "epoch": 52.6675, "grad_norm": 6.21875, "learning_rate": 9.621965416793563e-06, "loss": 1.3853, "step": 21067 }, { "epoch": 52.67, "grad_norm": 8.1875, "learning_rate": 9.621156307975992e-06, "loss": 1.2514, "step": 21068 }, { "epoch": 52.6725, "grad_norm": 8.875, "learning_rate": 9.620347201642102e-06, "loss": 1.3264, "step": 21069 }, { "epoch": 52.675, "grad_norm": 7.84375, "learning_rate": 9.619538097797205e-06, "loss": 1.2216, "step": 21070 }, { "epoch": 52.6775, "grad_norm": 6.1875, "learning_rate": 9.618728996446593e-06, "loss": 1.4291, "step": 21071 }, { "epoch": 52.68, "grad_norm": 8.875, "learning_rate": 9.617919897595586e-06, "loss": 1.2961, "step": 21072 }, { "epoch": 52.6825, "grad_norm": 5.71875, "learning_rate": 9.617110801249477e-06, "loss": 1.37, "step": 21073 }, { "epoch": 52.685, "grad_norm": 7.9375, "learning_rate": 9.616301707413578e-06, "loss": 1.4868, "step": 21074 }, { "epoch": 52.6875, "grad_norm": 6.59375, "learning_rate": 9.615492616093188e-06, "loss": 1.3227, "step": 21075 }, { "epoch": 52.69, "grad_norm": 7.9375, "learning_rate": 9.614683527293608e-06, "loss": 1.2421, "step": 21076 }, { "epoch": 52.6925, "grad_norm": 7.125, "learning_rate": 9.613874441020152e-06, "loss": 1.1853, "step": 21077 }, { "epoch": 52.695, "grad_norm": 6.25, "learning_rate": 9.613065357278121e-06, "loss": 1.2718, "step": 21078 }, { "epoch": 52.6975, "grad_norm": 7.40625, "learning_rate": 9.612256276072817e-06, "loss": 1.4228, "step": 21079 }, { "epoch": 52.7, "grad_norm": 6.625, "learning_rate": 9.611447197409544e-06, "loss": 1.5234, "step": 21080 }, { "epoch": 52.7025, "grad_norm": 9.4375, "learning_rate": 9.610638121293605e-06, "loss": 1.4319, "step": 21081 }, { "epoch": 52.705, "grad_norm": 8.3125, "learning_rate": 9.60982904773031e-06, "loss": 1.3623, "step": 21082 }, { "epoch": 52.7075, "grad_norm": 11.6875, "learning_rate": 9.609019976724963e-06, "loss": 1.2715, "step": 21083 }, { "epoch": 52.71, "grad_norm": 5.59375, "learning_rate": 9.608210908282862e-06, "loss": 1.4163, "step": 21084 }, { "epoch": 52.7125, "grad_norm": 6.84375, "learning_rate": 9.607401842409318e-06, "loss": 1.3045, "step": 21085 }, { "epoch": 52.715, "grad_norm": 7.0625, "learning_rate": 9.606592779109623e-06, "loss": 1.4094, "step": 21086 }, { "epoch": 52.7175, "grad_norm": 6.375, "learning_rate": 9.605783718389099e-06, "loss": 1.3511, "step": 21087 }, { "epoch": 52.72, "grad_norm": 5.90625, "learning_rate": 9.604974660253039e-06, "loss": 1.4439, "step": 21088 }, { "epoch": 52.7225, "grad_norm": 7.59375, "learning_rate": 9.60416560470675e-06, "loss": 1.3103, "step": 21089 }, { "epoch": 52.725, "grad_norm": 10.0625, "learning_rate": 9.603356551755534e-06, "loss": 1.2795, "step": 21090 }, { "epoch": 52.7275, "grad_norm": 8.75, "learning_rate": 9.602547501404694e-06, "loss": 1.2798, "step": 21091 }, { "epoch": 52.73, "grad_norm": 5.84375, "learning_rate": 9.601738453659537e-06, "loss": 1.2844, "step": 21092 }, { "epoch": 52.7325, "grad_norm": 7.46875, "learning_rate": 9.600929408525372e-06, "loss": 1.3876, "step": 21093 }, { "epoch": 52.735, "grad_norm": 8.0625, "learning_rate": 9.600120366007495e-06, "loss": 1.3203, "step": 21094 }, { "epoch": 52.7375, "grad_norm": 7.15625, "learning_rate": 9.599311326111213e-06, "loss": 1.4484, "step": 21095 }, { "epoch": 52.74, "grad_norm": 7.71875, "learning_rate": 9.598502288841827e-06, "loss": 1.3389, "step": 21096 }, { "epoch": 52.7425, "grad_norm": 6.71875, "learning_rate": 9.597693254204649e-06, "loss": 1.3227, "step": 21097 }, { "epoch": 52.745, "grad_norm": 8.0, "learning_rate": 9.596884222204974e-06, "loss": 1.2001, "step": 21098 }, { "epoch": 52.7475, "grad_norm": 9.25, "learning_rate": 9.596075192848112e-06, "loss": 1.4888, "step": 21099 }, { "epoch": 52.75, "grad_norm": 5.75, "learning_rate": 9.595266166139366e-06, "loss": 1.2483, "step": 21100 }, { "epoch": 52.7525, "grad_norm": 10.125, "learning_rate": 9.594457142084034e-06, "loss": 1.3528, "step": 21101 }, { "epoch": 52.755, "grad_norm": 8.375, "learning_rate": 9.593648120687428e-06, "loss": 1.2167, "step": 21102 }, { "epoch": 52.7575, "grad_norm": 6.78125, "learning_rate": 9.592839101954848e-06, "loss": 1.2862, "step": 21103 }, { "epoch": 52.76, "grad_norm": 8.375, "learning_rate": 9.592030085891602e-06, "loss": 1.5891, "step": 21104 }, { "epoch": 52.7625, "grad_norm": 7.40625, "learning_rate": 9.591221072502988e-06, "loss": 1.3615, "step": 21105 }, { "epoch": 52.765, "grad_norm": 11.75, "learning_rate": 9.590412061794308e-06, "loss": 1.2492, "step": 21106 }, { "epoch": 52.7675, "grad_norm": 5.40625, "learning_rate": 9.589603053770873e-06, "loss": 1.296, "step": 21107 }, { "epoch": 52.77, "grad_norm": 6.875, "learning_rate": 9.588794048437988e-06, "loss": 1.3851, "step": 21108 }, { "epoch": 52.7725, "grad_norm": 7.28125, "learning_rate": 9.58798504580095e-06, "loss": 1.363, "step": 21109 }, { "epoch": 52.775, "grad_norm": 5.25, "learning_rate": 9.587176045865063e-06, "loss": 1.2635, "step": 21110 }, { "epoch": 52.7775, "grad_norm": 7.125, "learning_rate": 9.586367048635634e-06, "loss": 1.3197, "step": 21111 }, { "epoch": 52.78, "grad_norm": 9.0625, "learning_rate": 9.58555805411797e-06, "loss": 1.257, "step": 21112 }, { "epoch": 52.7825, "grad_norm": 7.46875, "learning_rate": 9.584749062317368e-06, "loss": 1.2565, "step": 21113 }, { "epoch": 52.785, "grad_norm": 8.25, "learning_rate": 9.583940073239135e-06, "loss": 1.478, "step": 21114 }, { "epoch": 52.7875, "grad_norm": 6.6875, "learning_rate": 9.583131086888577e-06, "loss": 1.2292, "step": 21115 }, { "epoch": 52.79, "grad_norm": 9.4375, "learning_rate": 9.582322103270989e-06, "loss": 1.2903, "step": 21116 }, { "epoch": 52.7925, "grad_norm": 7.40625, "learning_rate": 9.581513122391684e-06, "loss": 1.3167, "step": 21117 }, { "epoch": 52.795, "grad_norm": 6.5625, "learning_rate": 9.580704144255961e-06, "loss": 1.2693, "step": 21118 }, { "epoch": 52.7975, "grad_norm": 7.0625, "learning_rate": 9.579895168869128e-06, "loss": 1.3673, "step": 21119 }, { "epoch": 52.8, "grad_norm": 5.34375, "learning_rate": 9.579086196236483e-06, "loss": 1.2225, "step": 21120 }, { "epoch": 52.8025, "grad_norm": 9.5, "learning_rate": 9.57827722636333e-06, "loss": 1.3588, "step": 21121 }, { "epoch": 52.805, "grad_norm": 10.0, "learning_rate": 9.577468259254976e-06, "loss": 1.2788, "step": 21122 }, { "epoch": 52.8075, "grad_norm": 11.1875, "learning_rate": 9.576659294916725e-06, "loss": 1.3815, "step": 21123 }, { "epoch": 52.81, "grad_norm": 6.625, "learning_rate": 9.57585033335388e-06, "loss": 1.2566, "step": 21124 }, { "epoch": 52.8125, "grad_norm": 6.96875, "learning_rate": 9.57504137457174e-06, "loss": 1.4585, "step": 21125 }, { "epoch": 52.815, "grad_norm": 12.8125, "learning_rate": 9.57423241857561e-06, "loss": 1.4385, "step": 21126 }, { "epoch": 52.8175, "grad_norm": 7.28125, "learning_rate": 9.5734234653708e-06, "loss": 1.2567, "step": 21127 }, { "epoch": 52.82, "grad_norm": 9.8125, "learning_rate": 9.572614514962607e-06, "loss": 1.3811, "step": 21128 }, { "epoch": 52.8225, "grad_norm": 7.25, "learning_rate": 9.571805567356335e-06, "loss": 1.5728, "step": 21129 }, { "epoch": 52.825, "grad_norm": 8.75, "learning_rate": 9.570996622557291e-06, "loss": 1.499, "step": 21130 }, { "epoch": 52.8275, "grad_norm": 5.96875, "learning_rate": 9.570187680570771e-06, "loss": 1.2774, "step": 21131 }, { "epoch": 52.83, "grad_norm": 7.71875, "learning_rate": 9.569378741402087e-06, "loss": 1.2908, "step": 21132 }, { "epoch": 52.8325, "grad_norm": 9.875, "learning_rate": 9.56856980505654e-06, "loss": 1.3584, "step": 21133 }, { "epoch": 52.835, "grad_norm": 6.75, "learning_rate": 9.56776087153943e-06, "loss": 1.4137, "step": 21134 }, { "epoch": 52.8375, "grad_norm": 7.1875, "learning_rate": 9.566951940856064e-06, "loss": 1.3038, "step": 21135 }, { "epoch": 52.84, "grad_norm": 7.125, "learning_rate": 9.56614301301174e-06, "loss": 1.5351, "step": 21136 }, { "epoch": 52.8425, "grad_norm": 7.875, "learning_rate": 9.565334088011768e-06, "loss": 1.4892, "step": 21137 }, { "epoch": 52.845, "grad_norm": 9.375, "learning_rate": 9.56452516586145e-06, "loss": 1.1968, "step": 21138 }, { "epoch": 52.8475, "grad_norm": 6.96875, "learning_rate": 9.563716246566085e-06, "loss": 1.3566, "step": 21139 }, { "epoch": 52.85, "grad_norm": 6.90625, "learning_rate": 9.562907330130981e-06, "loss": 1.4172, "step": 21140 }, { "epoch": 52.8525, "grad_norm": 7.21875, "learning_rate": 9.562098416561434e-06, "loss": 1.344, "step": 21141 }, { "epoch": 52.855, "grad_norm": 7.96875, "learning_rate": 9.561289505862758e-06, "loss": 1.2635, "step": 21142 }, { "epoch": 52.8575, "grad_norm": 7.84375, "learning_rate": 9.560480598040246e-06, "loss": 1.3589, "step": 21143 }, { "epoch": 52.86, "grad_norm": 8.625, "learning_rate": 9.55967169309921e-06, "loss": 1.3751, "step": 21144 }, { "epoch": 52.8625, "grad_norm": 10.3125, "learning_rate": 9.558862791044947e-06, "loss": 1.4755, "step": 21145 }, { "epoch": 52.865, "grad_norm": 6.53125, "learning_rate": 9.55805389188276e-06, "loss": 1.3327, "step": 21146 }, { "epoch": 52.8675, "grad_norm": 6.1875, "learning_rate": 9.557244995617955e-06, "loss": 1.344, "step": 21147 }, { "epoch": 52.87, "grad_norm": 8.375, "learning_rate": 9.556436102255834e-06, "loss": 1.3899, "step": 21148 }, { "epoch": 52.8725, "grad_norm": 7.125, "learning_rate": 9.555627211801704e-06, "loss": 1.5345, "step": 21149 }, { "epoch": 52.875, "grad_norm": 7.375, "learning_rate": 9.55481832426086e-06, "loss": 1.5124, "step": 21150 }, { "epoch": 52.8775, "grad_norm": 5.3125, "learning_rate": 9.554009439638607e-06, "loss": 1.2693, "step": 21151 }, { "epoch": 52.88, "grad_norm": 7.5, "learning_rate": 9.553200557940254e-06, "loss": 1.3705, "step": 21152 }, { "epoch": 52.8825, "grad_norm": 7.65625, "learning_rate": 9.552391679171101e-06, "loss": 1.4667, "step": 21153 }, { "epoch": 52.885, "grad_norm": 7.34375, "learning_rate": 9.55158280333645e-06, "loss": 1.4073, "step": 21154 }, { "epoch": 52.8875, "grad_norm": 8.5, "learning_rate": 9.550773930441604e-06, "loss": 1.3445, "step": 21155 }, { "epoch": 52.89, "grad_norm": 7.5, "learning_rate": 9.549965060491862e-06, "loss": 1.2971, "step": 21156 }, { "epoch": 52.8925, "grad_norm": 9.1875, "learning_rate": 9.549156193492535e-06, "loss": 1.1945, "step": 21157 }, { "epoch": 52.895, "grad_norm": 10.6875, "learning_rate": 9.548347329448921e-06, "loss": 1.3831, "step": 21158 }, { "epoch": 52.8975, "grad_norm": 9.875, "learning_rate": 9.547538468366326e-06, "loss": 1.4381, "step": 21159 }, { "epoch": 52.9, "grad_norm": 7.09375, "learning_rate": 9.54672961025005e-06, "loss": 1.3464, "step": 21160 }, { "epoch": 52.9025, "grad_norm": 7.1875, "learning_rate": 9.545920755105392e-06, "loss": 1.1425, "step": 21161 }, { "epoch": 52.905, "grad_norm": 7.3125, "learning_rate": 9.545111902937663e-06, "loss": 1.323, "step": 21162 }, { "epoch": 52.9075, "grad_norm": 4.90625, "learning_rate": 9.544303053752165e-06, "loss": 1.4664, "step": 21163 }, { "epoch": 52.91, "grad_norm": 8.125, "learning_rate": 9.543494207554193e-06, "loss": 1.3339, "step": 21164 }, { "epoch": 52.9125, "grad_norm": 5.28125, "learning_rate": 9.542685364349057e-06, "loss": 1.4211, "step": 21165 }, { "epoch": 52.915, "grad_norm": 5.5625, "learning_rate": 9.541876524142053e-06, "loss": 1.38, "step": 21166 }, { "epoch": 52.9175, "grad_norm": 6.59375, "learning_rate": 9.541067686938496e-06, "loss": 1.3568, "step": 21167 }, { "epoch": 52.92, "grad_norm": 7.84375, "learning_rate": 9.540258852743676e-06, "loss": 1.4454, "step": 21168 }, { "epoch": 52.9225, "grad_norm": 8.375, "learning_rate": 9.539450021562901e-06, "loss": 1.1592, "step": 21169 }, { "epoch": 52.925, "grad_norm": 7.9375, "learning_rate": 9.538641193401477e-06, "loss": 1.3241, "step": 21170 }, { "epoch": 52.9275, "grad_norm": 8.75, "learning_rate": 9.537832368264697e-06, "loss": 1.2038, "step": 21171 }, { "epoch": 52.93, "grad_norm": 7.65625, "learning_rate": 9.537023546157873e-06, "loss": 1.3065, "step": 21172 }, { "epoch": 52.9325, "grad_norm": 17.125, "learning_rate": 9.536214727086303e-06, "loss": 1.2543, "step": 21173 }, { "epoch": 52.935, "grad_norm": 9.125, "learning_rate": 9.535405911055293e-06, "loss": 1.365, "step": 21174 }, { "epoch": 52.9375, "grad_norm": 6.875, "learning_rate": 9.53459709807014e-06, "loss": 1.2683, "step": 21175 }, { "epoch": 52.94, "grad_norm": 8.625, "learning_rate": 9.533788288136149e-06, "loss": 1.2935, "step": 21176 }, { "epoch": 52.9425, "grad_norm": 8.1875, "learning_rate": 9.532979481258626e-06, "loss": 1.3206, "step": 21177 }, { "epoch": 52.945, "grad_norm": 6.0, "learning_rate": 9.532170677442871e-06, "loss": 1.1239, "step": 21178 }, { "epoch": 52.9475, "grad_norm": 7.28125, "learning_rate": 9.531361876694186e-06, "loss": 1.2261, "step": 21179 }, { "epoch": 52.95, "grad_norm": 9.4375, "learning_rate": 9.530553079017872e-06, "loss": 1.3923, "step": 21180 }, { "epoch": 52.9525, "grad_norm": 7.71875, "learning_rate": 9.529744284419234e-06, "loss": 1.3374, "step": 21181 }, { "epoch": 52.955, "grad_norm": 13.9375, "learning_rate": 9.528935492903575e-06, "loss": 1.2123, "step": 21182 }, { "epoch": 52.9575, "grad_norm": 8.375, "learning_rate": 9.528126704476195e-06, "loss": 1.3717, "step": 21183 }, { "epoch": 52.96, "grad_norm": 14.5625, "learning_rate": 9.527317919142398e-06, "loss": 1.2381, "step": 21184 }, { "epoch": 52.9625, "grad_norm": 7.25, "learning_rate": 9.526509136907488e-06, "loss": 1.2567, "step": 21185 }, { "epoch": 52.965, "grad_norm": 7.25, "learning_rate": 9.525700357776758e-06, "loss": 1.5213, "step": 21186 }, { "epoch": 52.9675, "grad_norm": 7.5625, "learning_rate": 9.524891581755523e-06, "loss": 1.3918, "step": 21187 }, { "epoch": 52.97, "grad_norm": 6.40625, "learning_rate": 9.52408280884908e-06, "loss": 1.2781, "step": 21188 }, { "epoch": 52.9725, "grad_norm": 8.625, "learning_rate": 9.523274039062732e-06, "loss": 1.214, "step": 21189 }, { "epoch": 52.975, "grad_norm": 11.6875, "learning_rate": 9.522465272401777e-06, "loss": 1.3266, "step": 21190 }, { "epoch": 52.9775, "grad_norm": 7.625, "learning_rate": 9.52165650887152e-06, "loss": 1.514, "step": 21191 }, { "epoch": 52.98, "grad_norm": 6.5, "learning_rate": 9.520847748477266e-06, "loss": 1.2611, "step": 21192 }, { "epoch": 52.9825, "grad_norm": 7.625, "learning_rate": 9.520038991224315e-06, "loss": 1.3239, "step": 21193 }, { "epoch": 52.985, "grad_norm": 5.78125, "learning_rate": 9.519230237117969e-06, "loss": 1.3436, "step": 21194 }, { "epoch": 52.9875, "grad_norm": 6.84375, "learning_rate": 9.518421486163531e-06, "loss": 1.3553, "step": 21195 }, { "epoch": 52.99, "grad_norm": 7.96875, "learning_rate": 9.517612738366299e-06, "loss": 1.334, "step": 21196 }, { "epoch": 52.9925, "grad_norm": 14.25, "learning_rate": 9.516803993731583e-06, "loss": 1.2532, "step": 21197 }, { "epoch": 52.995, "grad_norm": 6.6875, "learning_rate": 9.515995252264678e-06, "loss": 1.2286, "step": 21198 }, { "epoch": 52.9975, "grad_norm": 11.0, "learning_rate": 9.51518651397089e-06, "loss": 1.4134, "step": 21199 }, { "epoch": 53.0, "grad_norm": 5.875, "learning_rate": 9.514377778855521e-06, "loss": 1.1952, "step": 21200 }, { "epoch": 53.0025, "grad_norm": 7.75, "learning_rate": 9.513569046923869e-06, "loss": 1.1524, "step": 21201 }, { "epoch": 53.005, "grad_norm": 6.6875, "learning_rate": 9.512760318181239e-06, "loss": 1.3383, "step": 21202 }, { "epoch": 53.0075, "grad_norm": 10.625, "learning_rate": 9.511951592632935e-06, "loss": 1.3553, "step": 21203 }, { "epoch": 53.01, "grad_norm": 7.28125, "learning_rate": 9.511142870284257e-06, "loss": 1.2779, "step": 21204 }, { "epoch": 53.0125, "grad_norm": 6.625, "learning_rate": 9.510334151140504e-06, "loss": 1.3054, "step": 21205 }, { "epoch": 53.015, "grad_norm": 7.125, "learning_rate": 9.509525435206981e-06, "loss": 1.3018, "step": 21206 }, { "epoch": 53.0175, "grad_norm": 6.09375, "learning_rate": 9.508716722488992e-06, "loss": 1.1776, "step": 21207 }, { "epoch": 53.02, "grad_norm": 8.125, "learning_rate": 9.507908012991835e-06, "loss": 1.4093, "step": 21208 }, { "epoch": 53.0225, "grad_norm": 10.0, "learning_rate": 9.507099306720815e-06, "loss": 1.323, "step": 21209 }, { "epoch": 53.025, "grad_norm": 8.75, "learning_rate": 9.506290603681229e-06, "loss": 1.3509, "step": 21210 }, { "epoch": 53.0275, "grad_norm": 8.5, "learning_rate": 9.505481903878382e-06, "loss": 1.339, "step": 21211 }, { "epoch": 53.03, "grad_norm": 7.3125, "learning_rate": 9.50467320731758e-06, "loss": 1.2816, "step": 21212 }, { "epoch": 53.0325, "grad_norm": 8.125, "learning_rate": 9.50386451400412e-06, "loss": 1.3189, "step": 21213 }, { "epoch": 53.035, "grad_norm": 10.1875, "learning_rate": 9.503055823943304e-06, "loss": 1.3421, "step": 21214 }, { "epoch": 53.0375, "grad_norm": 7.5625, "learning_rate": 9.502247137140428e-06, "loss": 1.2863, "step": 21215 }, { "epoch": 53.04, "grad_norm": 7.84375, "learning_rate": 9.501438453600808e-06, "loss": 1.3341, "step": 21216 }, { "epoch": 53.0425, "grad_norm": 8.3125, "learning_rate": 9.500629773329733e-06, "loss": 1.3347, "step": 21217 }, { "epoch": 53.045, "grad_norm": 6.75, "learning_rate": 9.49982109633251e-06, "loss": 1.3242, "step": 21218 }, { "epoch": 53.0475, "grad_norm": 6.8125, "learning_rate": 9.499012422614444e-06, "loss": 1.3568, "step": 21219 }, { "epoch": 53.05, "grad_norm": 6.3125, "learning_rate": 9.498203752180827e-06, "loss": 1.2304, "step": 21220 }, { "epoch": 53.0525, "grad_norm": 8.1875, "learning_rate": 9.497395085036968e-06, "loss": 1.3714, "step": 21221 }, { "epoch": 53.055, "grad_norm": 7.84375, "learning_rate": 9.496586421188166e-06, "loss": 1.3613, "step": 21222 }, { "epoch": 53.0575, "grad_norm": 7.09375, "learning_rate": 9.495777760639727e-06, "loss": 1.3314, "step": 21223 }, { "epoch": 53.06, "grad_norm": 5.625, "learning_rate": 9.494969103396946e-06, "loss": 1.1494, "step": 21224 }, { "epoch": 53.0625, "grad_norm": 9.75, "learning_rate": 9.494160449465123e-06, "loss": 1.2561, "step": 21225 }, { "epoch": 53.065, "grad_norm": 9.25, "learning_rate": 9.49335179884957e-06, "loss": 1.2877, "step": 21226 }, { "epoch": 53.0675, "grad_norm": 9.5625, "learning_rate": 9.492543151555582e-06, "loss": 1.3953, "step": 21227 }, { "epoch": 53.07, "grad_norm": 5.78125, "learning_rate": 9.49173450758846e-06, "loss": 1.2027, "step": 21228 }, { "epoch": 53.0725, "grad_norm": 6.0625, "learning_rate": 9.490925866953506e-06, "loss": 1.3486, "step": 21229 }, { "epoch": 53.075, "grad_norm": 5.75, "learning_rate": 9.490117229656017e-06, "loss": 1.4029, "step": 21230 }, { "epoch": 53.0775, "grad_norm": 5.5, "learning_rate": 9.489308595701305e-06, "loss": 1.2827, "step": 21231 }, { "epoch": 53.08, "grad_norm": 9.8125, "learning_rate": 9.488499965094664e-06, "loss": 1.3944, "step": 21232 }, { "epoch": 53.0825, "grad_norm": 6.96875, "learning_rate": 9.487691337841398e-06, "loss": 1.3685, "step": 21233 }, { "epoch": 53.085, "grad_norm": 7.28125, "learning_rate": 9.486882713946805e-06, "loss": 1.1988, "step": 21234 }, { "epoch": 53.0875, "grad_norm": 7.0, "learning_rate": 9.486074093416188e-06, "loss": 1.311, "step": 21235 }, { "epoch": 53.09, "grad_norm": 10.125, "learning_rate": 9.485265476254849e-06, "loss": 1.506, "step": 21236 }, { "epoch": 53.0925, "grad_norm": 9.3125, "learning_rate": 9.484456862468092e-06, "loss": 1.252, "step": 21237 }, { "epoch": 53.095, "grad_norm": 6.96875, "learning_rate": 9.483648252061211e-06, "loss": 1.4133, "step": 21238 }, { "epoch": 53.0975, "grad_norm": 7.25, "learning_rate": 9.482839645039513e-06, "loss": 1.2469, "step": 21239 }, { "epoch": 53.1, "grad_norm": 8.375, "learning_rate": 9.482031041408296e-06, "loss": 1.3459, "step": 21240 }, { "epoch": 53.1025, "grad_norm": 8.5625, "learning_rate": 9.481222441172865e-06, "loss": 1.2835, "step": 21241 }, { "epoch": 53.105, "grad_norm": 7.0625, "learning_rate": 9.480413844338519e-06, "loss": 1.333, "step": 21242 }, { "epoch": 53.1075, "grad_norm": 7.4375, "learning_rate": 9.479605250910558e-06, "loss": 1.2533, "step": 21243 }, { "epoch": 53.11, "grad_norm": 7.0, "learning_rate": 9.478796660894288e-06, "loss": 1.4988, "step": 21244 }, { "epoch": 53.1125, "grad_norm": 8.25, "learning_rate": 9.477988074294996e-06, "loss": 1.4086, "step": 21245 }, { "epoch": 53.115, "grad_norm": 7.5, "learning_rate": 9.477179491118004e-06, "loss": 1.2265, "step": 21246 }, { "epoch": 53.1175, "grad_norm": 8.8125, "learning_rate": 9.476370911368597e-06, "loss": 1.4425, "step": 21247 }, { "epoch": 53.12, "grad_norm": 8.75, "learning_rate": 9.475562335052086e-06, "loss": 1.4063, "step": 21248 }, { "epoch": 53.1225, "grad_norm": 7.75, "learning_rate": 9.474753762173762e-06, "loss": 1.3035, "step": 21249 }, { "epoch": 53.125, "grad_norm": 6.75, "learning_rate": 9.473945192738933e-06, "loss": 1.3217, "step": 21250 }, { "epoch": 53.1275, "grad_norm": 9.8125, "learning_rate": 9.473136626752898e-06, "loss": 1.2392, "step": 21251 }, { "epoch": 53.13, "grad_norm": 11.625, "learning_rate": 9.47232806422096e-06, "loss": 1.3103, "step": 21252 }, { "epoch": 53.1325, "grad_norm": 8.625, "learning_rate": 9.471519505148418e-06, "loss": 1.3433, "step": 21253 }, { "epoch": 53.135, "grad_norm": 8.875, "learning_rate": 9.470710949540571e-06, "loss": 1.1983, "step": 21254 }, { "epoch": 53.1375, "grad_norm": 9.4375, "learning_rate": 9.46990239740272e-06, "loss": 1.3188, "step": 21255 }, { "epoch": 53.14, "grad_norm": 6.125, "learning_rate": 9.469093848740172e-06, "loss": 1.2558, "step": 21256 }, { "epoch": 53.1425, "grad_norm": 15.4375, "learning_rate": 9.468285303558222e-06, "loss": 1.4808, "step": 21257 }, { "epoch": 53.145, "grad_norm": 6.6875, "learning_rate": 9.467476761862173e-06, "loss": 1.3694, "step": 21258 }, { "epoch": 53.1475, "grad_norm": 8.375, "learning_rate": 9.466668223657326e-06, "loss": 1.4088, "step": 21259 }, { "epoch": 53.15, "grad_norm": 7.6875, "learning_rate": 9.465859688948977e-06, "loss": 1.288, "step": 21260 }, { "epoch": 53.1525, "grad_norm": 6.875, "learning_rate": 9.465051157742432e-06, "loss": 1.4071, "step": 21261 }, { "epoch": 53.155, "grad_norm": 6.96875, "learning_rate": 9.464242630042991e-06, "loss": 1.2352, "step": 21262 }, { "epoch": 53.1575, "grad_norm": 7.71875, "learning_rate": 9.463434105855955e-06, "loss": 1.1631, "step": 21263 }, { "epoch": 53.16, "grad_norm": 7.21875, "learning_rate": 9.462625585186621e-06, "loss": 1.279, "step": 21264 }, { "epoch": 53.1625, "grad_norm": 7.96875, "learning_rate": 9.461817068040293e-06, "loss": 1.3392, "step": 21265 }, { "epoch": 53.165, "grad_norm": 6.15625, "learning_rate": 9.46100855442227e-06, "loss": 1.3578, "step": 21266 }, { "epoch": 53.1675, "grad_norm": 8.4375, "learning_rate": 9.460200044337856e-06, "loss": 1.145, "step": 21267 }, { "epoch": 53.17, "grad_norm": 7.21875, "learning_rate": 9.459391537792347e-06, "loss": 1.3014, "step": 21268 }, { "epoch": 53.1725, "grad_norm": 10.625, "learning_rate": 9.458583034791046e-06, "loss": 1.2938, "step": 21269 }, { "epoch": 53.175, "grad_norm": 9.5625, "learning_rate": 9.457774535339251e-06, "loss": 1.4407, "step": 21270 }, { "epoch": 53.1775, "grad_norm": 13.0625, "learning_rate": 9.456966039442269e-06, "loss": 1.3708, "step": 21271 }, { "epoch": 53.18, "grad_norm": 5.71875, "learning_rate": 9.45615754710539e-06, "loss": 1.3083, "step": 21272 }, { "epoch": 53.1825, "grad_norm": 12.0, "learning_rate": 9.455349058333924e-06, "loss": 1.4445, "step": 21273 }, { "epoch": 53.185, "grad_norm": 10.5, "learning_rate": 9.45454057313317e-06, "loss": 1.2531, "step": 21274 }, { "epoch": 53.1875, "grad_norm": 6.09375, "learning_rate": 9.45373209150842e-06, "loss": 1.5335, "step": 21275 }, { "epoch": 53.19, "grad_norm": 7.53125, "learning_rate": 9.452923613464984e-06, "loss": 1.2722, "step": 21276 }, { "epoch": 53.1925, "grad_norm": 8.25, "learning_rate": 9.452115139008159e-06, "loss": 1.3416, "step": 21277 }, { "epoch": 53.195, "grad_norm": 8.625, "learning_rate": 9.451306668143248e-06, "loss": 1.3454, "step": 21278 }, { "epoch": 53.1975, "grad_norm": 7.0625, "learning_rate": 9.450498200875546e-06, "loss": 1.3148, "step": 21279 }, { "epoch": 53.2, "grad_norm": 7.96875, "learning_rate": 9.449689737210352e-06, "loss": 1.2993, "step": 21280 }, { "epoch": 53.2025, "grad_norm": 5.71875, "learning_rate": 9.448881277152973e-06, "loss": 1.382, "step": 21281 }, { "epoch": 53.205, "grad_norm": 6.8125, "learning_rate": 9.448072820708709e-06, "loss": 1.3862, "step": 21282 }, { "epoch": 53.2075, "grad_norm": 7.1875, "learning_rate": 9.447264367882855e-06, "loss": 1.1408, "step": 21283 }, { "epoch": 53.21, "grad_norm": 8.4375, "learning_rate": 9.446455918680715e-06, "loss": 1.4004, "step": 21284 }, { "epoch": 53.2125, "grad_norm": 6.0, "learning_rate": 9.445647473107583e-06, "loss": 1.277, "step": 21285 }, { "epoch": 53.215, "grad_norm": 5.53125, "learning_rate": 9.44483903116877e-06, "loss": 1.2984, "step": 21286 }, { "epoch": 53.2175, "grad_norm": 6.6875, "learning_rate": 9.444030592869568e-06, "loss": 1.3776, "step": 21287 }, { "epoch": 53.22, "grad_norm": 14.3125, "learning_rate": 9.443222158215278e-06, "loss": 1.2292, "step": 21288 }, { "epoch": 53.2225, "grad_norm": 8.3125, "learning_rate": 9.442413727211204e-06, "loss": 1.2069, "step": 21289 }, { "epoch": 53.225, "grad_norm": 8.25, "learning_rate": 9.44160529986264e-06, "loss": 1.4062, "step": 21290 }, { "epoch": 53.2275, "grad_norm": 6.90625, "learning_rate": 9.44079687617489e-06, "loss": 1.3838, "step": 21291 }, { "epoch": 53.23, "grad_norm": 6.34375, "learning_rate": 9.439988456153254e-06, "loss": 1.2626, "step": 21292 }, { "epoch": 53.2325, "grad_norm": 8.125, "learning_rate": 9.439180039803035e-06, "loss": 1.3676, "step": 21293 }, { "epoch": 53.235, "grad_norm": 8.3125, "learning_rate": 9.438371627129525e-06, "loss": 1.3784, "step": 21294 }, { "epoch": 53.2375, "grad_norm": 10.375, "learning_rate": 9.437563218138026e-06, "loss": 1.3273, "step": 21295 }, { "epoch": 53.24, "grad_norm": 6.90625, "learning_rate": 9.436754812833843e-06, "loss": 1.3187, "step": 21296 }, { "epoch": 53.2425, "grad_norm": 9.25, "learning_rate": 9.435946411222275e-06, "loss": 1.4562, "step": 21297 }, { "epoch": 53.245, "grad_norm": 6.09375, "learning_rate": 9.435138013308617e-06, "loss": 1.3625, "step": 21298 }, { "epoch": 53.2475, "grad_norm": 7.5, "learning_rate": 9.434329619098173e-06, "loss": 1.3847, "step": 21299 }, { "epoch": 53.25, "grad_norm": 8.125, "learning_rate": 9.433521228596237e-06, "loss": 1.2314, "step": 21300 }, { "epoch": 53.2525, "grad_norm": 6.875, "learning_rate": 9.432712841808118e-06, "loss": 1.2278, "step": 21301 }, { "epoch": 53.255, "grad_norm": 5.84375, "learning_rate": 9.43190445873911e-06, "loss": 1.3188, "step": 21302 }, { "epoch": 53.2575, "grad_norm": 7.3125, "learning_rate": 9.431096079394515e-06, "loss": 1.2606, "step": 21303 }, { "epoch": 53.26, "grad_norm": 9.8125, "learning_rate": 9.43028770377963e-06, "loss": 1.4108, "step": 21304 }, { "epoch": 53.2625, "grad_norm": 8.4375, "learning_rate": 9.429479331899753e-06, "loss": 1.2737, "step": 21305 }, { "epoch": 53.265, "grad_norm": 8.0625, "learning_rate": 9.428670963760187e-06, "loss": 1.2006, "step": 21306 }, { "epoch": 53.2675, "grad_norm": 9.75, "learning_rate": 9.427862599366237e-06, "loss": 1.2656, "step": 21307 }, { "epoch": 53.27, "grad_norm": 4.6875, "learning_rate": 9.427054238723193e-06, "loss": 1.2839, "step": 21308 }, { "epoch": 53.2725, "grad_norm": 8.6875, "learning_rate": 9.426245881836358e-06, "loss": 1.1761, "step": 21309 }, { "epoch": 53.275, "grad_norm": 6.6875, "learning_rate": 9.425437528711031e-06, "loss": 1.384, "step": 21310 }, { "epoch": 53.2775, "grad_norm": 7.34375, "learning_rate": 9.424629179352517e-06, "loss": 1.3982, "step": 21311 }, { "epoch": 53.28, "grad_norm": 11.3125, "learning_rate": 9.423820833766108e-06, "loss": 1.4013, "step": 21312 }, { "epoch": 53.2825, "grad_norm": 8.75, "learning_rate": 9.423012491957106e-06, "loss": 1.2828, "step": 21313 }, { "epoch": 53.285, "grad_norm": 9.6875, "learning_rate": 9.422204153930815e-06, "loss": 1.2218, "step": 21314 }, { "epoch": 53.2875, "grad_norm": 6.8125, "learning_rate": 9.421395819692523e-06, "loss": 1.3222, "step": 21315 }, { "epoch": 53.29, "grad_norm": 9.4375, "learning_rate": 9.420587489247544e-06, "loss": 1.3216, "step": 21316 }, { "epoch": 53.2925, "grad_norm": 7.0, "learning_rate": 9.419779162601168e-06, "loss": 1.387, "step": 21317 }, { "epoch": 53.295, "grad_norm": 10.5, "learning_rate": 9.418970839758698e-06, "loss": 1.2933, "step": 21318 }, { "epoch": 53.2975, "grad_norm": 8.875, "learning_rate": 9.418162520725429e-06, "loss": 1.3137, "step": 21319 }, { "epoch": 53.3, "grad_norm": 8.125, "learning_rate": 9.417354205506663e-06, "loss": 1.3563, "step": 21320 }, { "epoch": 53.3025, "grad_norm": 9.25, "learning_rate": 9.4165458941077e-06, "loss": 1.3679, "step": 21321 }, { "epoch": 53.305, "grad_norm": 7.4375, "learning_rate": 9.415737586533843e-06, "loss": 1.2715, "step": 21322 }, { "epoch": 53.3075, "grad_norm": 6.3125, "learning_rate": 9.414929282790382e-06, "loss": 1.3902, "step": 21323 }, { "epoch": 53.31, "grad_norm": 9.125, "learning_rate": 9.414120982882622e-06, "loss": 1.2501, "step": 21324 }, { "epoch": 53.3125, "grad_norm": 7.53125, "learning_rate": 9.41331268681586e-06, "loss": 1.3658, "step": 21325 }, { "epoch": 53.315, "grad_norm": 9.0, "learning_rate": 9.412504394595402e-06, "loss": 1.3711, "step": 21326 }, { "epoch": 53.3175, "grad_norm": 6.53125, "learning_rate": 9.411696106226538e-06, "loss": 1.3976, "step": 21327 }, { "epoch": 53.32, "grad_norm": 8.9375, "learning_rate": 9.410887821714571e-06, "loss": 1.3191, "step": 21328 }, { "epoch": 53.3225, "grad_norm": 9.0625, "learning_rate": 9.410079541064802e-06, "loss": 1.1514, "step": 21329 }, { "epoch": 53.325, "grad_norm": 7.78125, "learning_rate": 9.409271264282523e-06, "loss": 1.419, "step": 21330 }, { "epoch": 53.3275, "grad_norm": 8.25, "learning_rate": 9.408462991373043e-06, "loss": 1.0727, "step": 21331 }, { "epoch": 53.33, "grad_norm": 8.6875, "learning_rate": 9.407654722341653e-06, "loss": 1.2856, "step": 21332 }, { "epoch": 53.3325, "grad_norm": 6.84375, "learning_rate": 9.406846457193657e-06, "loss": 1.1963, "step": 21333 }, { "epoch": 53.335, "grad_norm": 7.59375, "learning_rate": 9.40603819593435e-06, "loss": 1.394, "step": 21334 }, { "epoch": 53.3375, "grad_norm": 10.1875, "learning_rate": 9.405229938569033e-06, "loss": 1.4808, "step": 21335 }, { "epoch": 53.34, "grad_norm": 6.3125, "learning_rate": 9.404421685103004e-06, "loss": 1.4209, "step": 21336 }, { "epoch": 53.3425, "grad_norm": 8.25, "learning_rate": 9.403613435541566e-06, "loss": 1.2996, "step": 21337 }, { "epoch": 53.345, "grad_norm": 9.3125, "learning_rate": 9.402805189890012e-06, "loss": 1.3306, "step": 21338 }, { "epoch": 53.3475, "grad_norm": 7.34375, "learning_rate": 9.401996948153644e-06, "loss": 1.5094, "step": 21339 }, { "epoch": 53.35, "grad_norm": 8.0625, "learning_rate": 9.401188710337757e-06, "loss": 1.4652, "step": 21340 }, { "epoch": 53.3525, "grad_norm": 9.4375, "learning_rate": 9.400380476447658e-06, "loss": 1.2522, "step": 21341 }, { "epoch": 53.355, "grad_norm": 5.8125, "learning_rate": 9.399572246488638e-06, "loss": 1.0695, "step": 21342 }, { "epoch": 53.3575, "grad_norm": 9.3125, "learning_rate": 9.398764020465999e-06, "loss": 1.2814, "step": 21343 }, { "epoch": 53.36, "grad_norm": 8.75, "learning_rate": 9.39795579838504e-06, "loss": 1.2487, "step": 21344 }, { "epoch": 53.3625, "grad_norm": 8.4375, "learning_rate": 9.397147580251054e-06, "loss": 1.3654, "step": 21345 }, { "epoch": 53.365, "grad_norm": 9.4375, "learning_rate": 9.396339366069347e-06, "loss": 1.3749, "step": 21346 }, { "epoch": 53.3675, "grad_norm": 6.59375, "learning_rate": 9.395531155845217e-06, "loss": 1.2402, "step": 21347 }, { "epoch": 53.37, "grad_norm": 6.4375, "learning_rate": 9.394722949583962e-06, "loss": 1.2391, "step": 21348 }, { "epoch": 53.3725, "grad_norm": 5.03125, "learning_rate": 9.393914747290874e-06, "loss": 1.2422, "step": 21349 }, { "epoch": 53.375, "grad_norm": 14.25, "learning_rate": 9.393106548971257e-06, "loss": 1.3512, "step": 21350 }, { "epoch": 53.3775, "grad_norm": 20.75, "learning_rate": 9.39229835463041e-06, "loss": 1.4164, "step": 21351 }, { "epoch": 53.38, "grad_norm": 6.71875, "learning_rate": 9.391490164273635e-06, "loss": 1.3453, "step": 21352 }, { "epoch": 53.3825, "grad_norm": 9.5, "learning_rate": 9.390681977906223e-06, "loss": 1.5417, "step": 21353 }, { "epoch": 53.385, "grad_norm": 7.65625, "learning_rate": 9.389873795533477e-06, "loss": 1.3453, "step": 21354 }, { "epoch": 53.3875, "grad_norm": 15.25, "learning_rate": 9.389065617160688e-06, "loss": 1.3041, "step": 21355 }, { "epoch": 53.39, "grad_norm": 10.3125, "learning_rate": 9.388257442793167e-06, "loss": 1.3186, "step": 21356 }, { "epoch": 53.3925, "grad_norm": 10.3125, "learning_rate": 9.387449272436203e-06, "loss": 1.3851, "step": 21357 }, { "epoch": 53.395, "grad_norm": 6.28125, "learning_rate": 9.386641106095096e-06, "loss": 1.4605, "step": 21358 }, { "epoch": 53.3975, "grad_norm": 7.78125, "learning_rate": 9.38583294377515e-06, "loss": 1.4519, "step": 21359 }, { "epoch": 53.4, "grad_norm": 9.6875, "learning_rate": 9.385024785481653e-06, "loss": 1.2077, "step": 21360 }, { "epoch": 53.4025, "grad_norm": 6.21875, "learning_rate": 9.38421663121991e-06, "loss": 1.2787, "step": 21361 }, { "epoch": 53.405, "grad_norm": 6.21875, "learning_rate": 9.38340848099522e-06, "loss": 1.3448, "step": 21362 }, { "epoch": 53.4075, "grad_norm": 6.0625, "learning_rate": 9.382600334812882e-06, "loss": 1.3397, "step": 21363 }, { "epoch": 53.41, "grad_norm": 5.53125, "learning_rate": 9.381792192678188e-06, "loss": 1.3114, "step": 21364 }, { "epoch": 53.4125, "grad_norm": 7.96875, "learning_rate": 9.380984054596439e-06, "loss": 1.3423, "step": 21365 }, { "epoch": 53.415, "grad_norm": 6.8125, "learning_rate": 9.380175920572934e-06, "loss": 1.2838, "step": 21366 }, { "epoch": 53.4175, "grad_norm": 6.78125, "learning_rate": 9.379367790612974e-06, "loss": 1.2007, "step": 21367 }, { "epoch": 53.42, "grad_norm": 5.5, "learning_rate": 9.37855966472185e-06, "loss": 1.4732, "step": 21368 }, { "epoch": 53.4225, "grad_norm": 9.4375, "learning_rate": 9.377751542904868e-06, "loss": 1.4705, "step": 21369 }, { "epoch": 53.425, "grad_norm": 7.65625, "learning_rate": 9.376943425167315e-06, "loss": 1.2224, "step": 21370 }, { "epoch": 53.4275, "grad_norm": 7.09375, "learning_rate": 9.376135311514502e-06, "loss": 1.3003, "step": 21371 }, { "epoch": 53.43, "grad_norm": 6.46875, "learning_rate": 9.37532720195172e-06, "loss": 1.2686, "step": 21372 }, { "epoch": 53.4325, "grad_norm": 7.53125, "learning_rate": 9.37451909648427e-06, "loss": 1.225, "step": 21373 }, { "epoch": 53.435, "grad_norm": 15.125, "learning_rate": 9.373710995117444e-06, "loss": 1.2947, "step": 21374 }, { "epoch": 53.4375, "grad_norm": 7.5625, "learning_rate": 9.372902897856542e-06, "loss": 1.4191, "step": 21375 }, { "epoch": 53.44, "grad_norm": 11.3125, "learning_rate": 9.372094804706867e-06, "loss": 1.2959, "step": 21376 }, { "epoch": 53.4425, "grad_norm": 5.5625, "learning_rate": 9.371286715673714e-06, "loss": 1.4495, "step": 21377 }, { "epoch": 53.445, "grad_norm": 7.9375, "learning_rate": 9.370478630762379e-06, "loss": 1.2797, "step": 21378 }, { "epoch": 53.4475, "grad_norm": 7.625, "learning_rate": 9.36967054997816e-06, "loss": 1.4444, "step": 21379 }, { "epoch": 53.45, "grad_norm": 7.5, "learning_rate": 9.368862473326355e-06, "loss": 1.1827, "step": 21380 }, { "epoch": 53.4525, "grad_norm": 8.75, "learning_rate": 9.368054400812266e-06, "loss": 1.3347, "step": 21381 }, { "epoch": 53.455, "grad_norm": 8.3125, "learning_rate": 9.367246332441184e-06, "loss": 1.3255, "step": 21382 }, { "epoch": 53.4575, "grad_norm": 7.09375, "learning_rate": 9.36643826821841e-06, "loss": 1.3586, "step": 21383 }, { "epoch": 53.46, "grad_norm": 8.125, "learning_rate": 9.365630208149244e-06, "loss": 1.3883, "step": 21384 }, { "epoch": 53.4625, "grad_norm": 8.3125, "learning_rate": 9.364822152238974e-06, "loss": 1.4159, "step": 21385 }, { "epoch": 53.465, "grad_norm": 10.125, "learning_rate": 9.364014100492911e-06, "loss": 1.4749, "step": 21386 }, { "epoch": 53.4675, "grad_norm": 9.9375, "learning_rate": 9.363206052916344e-06, "loss": 1.3121, "step": 21387 }, { "epoch": 53.47, "grad_norm": 7.875, "learning_rate": 9.362398009514575e-06, "loss": 1.401, "step": 21388 }, { "epoch": 53.4725, "grad_norm": 6.21875, "learning_rate": 9.361589970292897e-06, "loss": 1.1901, "step": 21389 }, { "epoch": 53.475, "grad_norm": 8.375, "learning_rate": 9.360781935256607e-06, "loss": 1.3282, "step": 21390 }, { "epoch": 53.4775, "grad_norm": 5.75, "learning_rate": 9.359973904411008e-06, "loss": 1.2992, "step": 21391 }, { "epoch": 53.48, "grad_norm": 9.1875, "learning_rate": 9.359165877761396e-06, "loss": 1.3457, "step": 21392 }, { "epoch": 53.4825, "grad_norm": 10.25, "learning_rate": 9.358357855313066e-06, "loss": 1.5993, "step": 21393 }, { "epoch": 53.485, "grad_norm": 7.625, "learning_rate": 9.357549837071315e-06, "loss": 1.2907, "step": 21394 }, { "epoch": 53.4875, "grad_norm": 9.375, "learning_rate": 9.35674182304144e-06, "loss": 1.2571, "step": 21395 }, { "epoch": 53.49, "grad_norm": 7.25, "learning_rate": 9.355933813228743e-06, "loss": 1.3668, "step": 21396 }, { "epoch": 53.4925, "grad_norm": 8.6875, "learning_rate": 9.355125807638517e-06, "loss": 1.2653, "step": 21397 }, { "epoch": 53.495, "grad_norm": 6.09375, "learning_rate": 9.35431780627606e-06, "loss": 1.4062, "step": 21398 }, { "epoch": 53.4975, "grad_norm": 5.15625, "learning_rate": 9.353509809146672e-06, "loss": 1.4608, "step": 21399 }, { "epoch": 53.5, "grad_norm": 7.4375, "learning_rate": 9.352701816255643e-06, "loss": 1.3526, "step": 21400 }, { "epoch": 53.5025, "grad_norm": 6.90625, "learning_rate": 9.351893827608278e-06, "loss": 1.3078, "step": 21401 }, { "epoch": 53.505, "grad_norm": 9.4375, "learning_rate": 9.35108584320987e-06, "loss": 1.3955, "step": 21402 }, { "epoch": 53.5075, "grad_norm": 10.0, "learning_rate": 9.35027786306572e-06, "loss": 1.6041, "step": 21403 }, { "epoch": 53.51, "grad_norm": 5.65625, "learning_rate": 9.349469887181122e-06, "loss": 1.4451, "step": 21404 }, { "epoch": 53.5125, "grad_norm": 8.4375, "learning_rate": 9.34866191556137e-06, "loss": 1.2914, "step": 21405 }, { "epoch": 53.515, "grad_norm": 9.0, "learning_rate": 9.347853948211765e-06, "loss": 1.2155, "step": 21406 }, { "epoch": 53.5175, "grad_norm": 8.9375, "learning_rate": 9.347045985137607e-06, "loss": 1.3756, "step": 21407 }, { "epoch": 53.52, "grad_norm": 6.9375, "learning_rate": 9.346238026344186e-06, "loss": 1.452, "step": 21408 }, { "epoch": 53.5225, "grad_norm": 7.5, "learning_rate": 9.345430071836804e-06, "loss": 1.4761, "step": 21409 }, { "epoch": 53.525, "grad_norm": 7.5, "learning_rate": 9.344622121620754e-06, "loss": 1.2853, "step": 21410 }, { "epoch": 53.5275, "grad_norm": 5.59375, "learning_rate": 9.343814175701339e-06, "loss": 1.3291, "step": 21411 }, { "epoch": 53.53, "grad_norm": 6.9375, "learning_rate": 9.343006234083848e-06, "loss": 1.3425, "step": 21412 }, { "epoch": 53.5325, "grad_norm": 6.8125, "learning_rate": 9.342198296773585e-06, "loss": 1.2487, "step": 21413 }, { "epoch": 53.535, "grad_norm": 8.75, "learning_rate": 9.341390363775844e-06, "loss": 1.4701, "step": 21414 }, { "epoch": 53.5375, "grad_norm": 7.375, "learning_rate": 9.340582435095918e-06, "loss": 1.2336, "step": 21415 }, { "epoch": 53.54, "grad_norm": 7.71875, "learning_rate": 9.339774510739107e-06, "loss": 1.3794, "step": 21416 }, { "epoch": 53.5425, "grad_norm": 7.9375, "learning_rate": 9.338966590710711e-06, "loss": 1.1429, "step": 21417 }, { "epoch": 53.545, "grad_norm": 6.96875, "learning_rate": 9.338158675016024e-06, "loss": 1.3676, "step": 21418 }, { "epoch": 53.5475, "grad_norm": 8.9375, "learning_rate": 9.337350763660341e-06, "loss": 1.447, "step": 21419 }, { "epoch": 53.55, "grad_norm": 5.59375, "learning_rate": 9.336542856648958e-06, "loss": 1.1209, "step": 21420 }, { "epoch": 53.5525, "grad_norm": 7.5625, "learning_rate": 9.335734953987173e-06, "loss": 1.4087, "step": 21421 }, { "epoch": 53.555, "grad_norm": 12.3125, "learning_rate": 9.334927055680287e-06, "loss": 1.4048, "step": 21422 }, { "epoch": 53.5575, "grad_norm": 5.875, "learning_rate": 9.334119161733592e-06, "loss": 1.221, "step": 21423 }, { "epoch": 53.56, "grad_norm": 5.90625, "learning_rate": 9.333311272152385e-06, "loss": 1.1991, "step": 21424 }, { "epoch": 53.5625, "grad_norm": 9.0625, "learning_rate": 9.332503386941958e-06, "loss": 1.383, "step": 21425 }, { "epoch": 53.565, "grad_norm": 9.0, "learning_rate": 9.331695506107616e-06, "loss": 1.3073, "step": 21426 }, { "epoch": 53.5675, "grad_norm": 7.125, "learning_rate": 9.330887629654652e-06, "loss": 1.2611, "step": 21427 }, { "epoch": 53.57, "grad_norm": 7.03125, "learning_rate": 9.33007975758836e-06, "loss": 1.4512, "step": 21428 }, { "epoch": 53.5725, "grad_norm": 8.125, "learning_rate": 9.32927188991404e-06, "loss": 1.3732, "step": 21429 }, { "epoch": 53.575, "grad_norm": 7.625, "learning_rate": 9.328464026636984e-06, "loss": 1.201, "step": 21430 }, { "epoch": 53.5775, "grad_norm": 8.6875, "learning_rate": 9.327656167762494e-06, "loss": 1.4115, "step": 21431 }, { "epoch": 53.58, "grad_norm": 7.5, "learning_rate": 9.32684831329586e-06, "loss": 1.2375, "step": 21432 }, { "epoch": 53.5825, "grad_norm": 6.5, "learning_rate": 9.326040463242388e-06, "loss": 1.3251, "step": 21433 }, { "epoch": 53.585, "grad_norm": 6.09375, "learning_rate": 9.325232617607363e-06, "loss": 1.2709, "step": 21434 }, { "epoch": 53.5875, "grad_norm": 6.09375, "learning_rate": 9.324424776396082e-06, "loss": 1.3665, "step": 21435 }, { "epoch": 53.59, "grad_norm": 6.125, "learning_rate": 9.32361693961385e-06, "loss": 1.3386, "step": 21436 }, { "epoch": 53.5925, "grad_norm": 6.5, "learning_rate": 9.32280910726596e-06, "loss": 1.4837, "step": 21437 }, { "epoch": 53.595, "grad_norm": 6.71875, "learning_rate": 9.322001279357704e-06, "loss": 1.2964, "step": 21438 }, { "epoch": 53.5975, "grad_norm": 4.15625, "learning_rate": 9.321193455894383e-06, "loss": 1.3816, "step": 21439 }, { "epoch": 53.6, "grad_norm": 6.25, "learning_rate": 9.320385636881283e-06, "loss": 1.4887, "step": 21440 }, { "epoch": 53.6025, "grad_norm": 7.21875, "learning_rate": 9.319577822323716e-06, "loss": 1.1905, "step": 21441 }, { "epoch": 53.605, "grad_norm": 8.0625, "learning_rate": 9.318770012226966e-06, "loss": 1.3328, "step": 21442 }, { "epoch": 53.6075, "grad_norm": 7.625, "learning_rate": 9.317962206596334e-06, "loss": 1.3865, "step": 21443 }, { "epoch": 53.61, "grad_norm": 9.0, "learning_rate": 9.317154405437113e-06, "loss": 1.4317, "step": 21444 }, { "epoch": 53.6125, "grad_norm": 6.78125, "learning_rate": 9.316346608754599e-06, "loss": 1.3529, "step": 21445 }, { "epoch": 53.615, "grad_norm": 6.21875, "learning_rate": 9.315538816554091e-06, "loss": 1.4682, "step": 21446 }, { "epoch": 53.6175, "grad_norm": 7.625, "learning_rate": 9.314731028840885e-06, "loss": 1.2599, "step": 21447 }, { "epoch": 53.62, "grad_norm": 10.4375, "learning_rate": 9.313923245620271e-06, "loss": 1.426, "step": 21448 }, { "epoch": 53.6225, "grad_norm": 8.4375, "learning_rate": 9.31311546689755e-06, "loss": 1.1987, "step": 21449 }, { "epoch": 53.625, "grad_norm": 9.75, "learning_rate": 9.312307692678016e-06, "loss": 1.5597, "step": 21450 }, { "epoch": 53.6275, "grad_norm": 7.71875, "learning_rate": 9.311499922966969e-06, "loss": 1.4409, "step": 21451 }, { "epoch": 53.63, "grad_norm": 6.1875, "learning_rate": 9.310692157769699e-06, "loss": 1.2922, "step": 21452 }, { "epoch": 53.6325, "grad_norm": 8.4375, "learning_rate": 9.309884397091502e-06, "loss": 1.1989, "step": 21453 }, { "epoch": 53.635, "grad_norm": 7.03125, "learning_rate": 9.309076640937674e-06, "loss": 1.3809, "step": 21454 }, { "epoch": 53.6375, "grad_norm": 7.09375, "learning_rate": 9.308268889313514e-06, "loss": 1.3059, "step": 21455 }, { "epoch": 53.64, "grad_norm": 5.59375, "learning_rate": 9.307461142224318e-06, "loss": 1.2746, "step": 21456 }, { "epoch": 53.6425, "grad_norm": 6.78125, "learning_rate": 9.306653399675378e-06, "loss": 1.3126, "step": 21457 }, { "epoch": 53.645, "grad_norm": 5.3125, "learning_rate": 9.30584566167199e-06, "loss": 1.3338, "step": 21458 }, { "epoch": 53.6475, "grad_norm": 7.59375, "learning_rate": 9.305037928219445e-06, "loss": 1.3989, "step": 21459 }, { "epoch": 53.65, "grad_norm": 5.8125, "learning_rate": 9.30423019932305e-06, "loss": 1.1861, "step": 21460 }, { "epoch": 53.6525, "grad_norm": 8.9375, "learning_rate": 9.303422474988092e-06, "loss": 1.3664, "step": 21461 }, { "epoch": 53.655, "grad_norm": 5.75, "learning_rate": 9.302614755219871e-06, "loss": 1.2525, "step": 21462 }, { "epoch": 53.6575, "grad_norm": 9.0625, "learning_rate": 9.301807040023676e-06, "loss": 1.3538, "step": 21463 }, { "epoch": 53.66, "grad_norm": 6.9375, "learning_rate": 9.300999329404805e-06, "loss": 1.4154, "step": 21464 }, { "epoch": 53.6625, "grad_norm": 7.65625, "learning_rate": 9.300191623368558e-06, "loss": 1.4308, "step": 21465 }, { "epoch": 53.665, "grad_norm": 8.5625, "learning_rate": 9.299383921920226e-06, "loss": 1.321, "step": 21466 }, { "epoch": 53.6675, "grad_norm": 8.375, "learning_rate": 9.298576225065104e-06, "loss": 1.223, "step": 21467 }, { "epoch": 53.67, "grad_norm": 7.375, "learning_rate": 9.29776853280849e-06, "loss": 1.2907, "step": 21468 }, { "epoch": 53.6725, "grad_norm": 5.625, "learning_rate": 9.296960845155673e-06, "loss": 1.304, "step": 21469 }, { "epoch": 53.675, "grad_norm": 8.625, "learning_rate": 9.296153162111956e-06, "loss": 1.3083, "step": 21470 }, { "epoch": 53.6775, "grad_norm": 8.4375, "learning_rate": 9.29534548368263e-06, "loss": 1.194, "step": 21471 }, { "epoch": 53.68, "grad_norm": 11.25, "learning_rate": 9.29453780987299e-06, "loss": 1.1667, "step": 21472 }, { "epoch": 53.6825, "grad_norm": 6.125, "learning_rate": 9.293730140688336e-06, "loss": 1.1332, "step": 21473 }, { "epoch": 53.685, "grad_norm": 9.1875, "learning_rate": 9.292922476133954e-06, "loss": 1.3293, "step": 21474 }, { "epoch": 53.6875, "grad_norm": 7.625, "learning_rate": 9.292114816215145e-06, "loss": 1.3469, "step": 21475 }, { "epoch": 53.69, "grad_norm": 10.0625, "learning_rate": 9.291307160937205e-06, "loss": 1.3946, "step": 21476 }, { "epoch": 53.6925, "grad_norm": 5.65625, "learning_rate": 9.290499510305427e-06, "loss": 1.4316, "step": 21477 }, { "epoch": 53.695, "grad_norm": 6.8125, "learning_rate": 9.289691864325103e-06, "loss": 1.3999, "step": 21478 }, { "epoch": 53.6975, "grad_norm": 5.4375, "learning_rate": 9.288884223001532e-06, "loss": 1.3505, "step": 21479 }, { "epoch": 53.7, "grad_norm": 8.0, "learning_rate": 9.288076586340005e-06, "loss": 1.2709, "step": 21480 }, { "epoch": 53.7025, "grad_norm": 8.625, "learning_rate": 9.287268954345826e-06, "loss": 1.2773, "step": 21481 }, { "epoch": 53.705, "grad_norm": 7.1875, "learning_rate": 9.28646132702428e-06, "loss": 1.286, "step": 21482 }, { "epoch": 53.7075, "grad_norm": 5.53125, "learning_rate": 9.285653704380662e-06, "loss": 1.1358, "step": 21483 }, { "epoch": 53.71, "grad_norm": 9.875, "learning_rate": 9.28484608642027e-06, "loss": 1.3428, "step": 21484 }, { "epoch": 53.7125, "grad_norm": 6.3125, "learning_rate": 9.284038473148402e-06, "loss": 1.3939, "step": 21485 }, { "epoch": 53.715, "grad_norm": 7.25, "learning_rate": 9.283230864570348e-06, "loss": 1.4327, "step": 21486 }, { "epoch": 53.7175, "grad_norm": 8.5625, "learning_rate": 9.282423260691404e-06, "loss": 1.1876, "step": 21487 }, { "epoch": 53.72, "grad_norm": 7.28125, "learning_rate": 9.281615661516866e-06, "loss": 1.4013, "step": 21488 }, { "epoch": 53.7225, "grad_norm": 11.0, "learning_rate": 9.28080806705202e-06, "loss": 1.458, "step": 21489 }, { "epoch": 53.725, "grad_norm": 7.96875, "learning_rate": 9.280000477302174e-06, "loss": 1.4104, "step": 21490 }, { "epoch": 53.7275, "grad_norm": 7.40625, "learning_rate": 9.279192892272613e-06, "loss": 1.2929, "step": 21491 }, { "epoch": 53.73, "grad_norm": 7.125, "learning_rate": 9.278385311968638e-06, "loss": 1.2484, "step": 21492 }, { "epoch": 53.7325, "grad_norm": 14.0625, "learning_rate": 9.277577736395538e-06, "loss": 1.3959, "step": 21493 }, { "epoch": 53.735, "grad_norm": 7.0, "learning_rate": 9.276770165558607e-06, "loss": 1.3549, "step": 21494 }, { "epoch": 53.7375, "grad_norm": 9.3125, "learning_rate": 9.275962599463143e-06, "loss": 1.4743, "step": 21495 }, { "epoch": 53.74, "grad_norm": 4.75, "learning_rate": 9.275155038114442e-06, "loss": 1.2634, "step": 21496 }, { "epoch": 53.7425, "grad_norm": 9.0, "learning_rate": 9.274347481517796e-06, "loss": 1.4804, "step": 21497 }, { "epoch": 53.745, "grad_norm": 5.65625, "learning_rate": 9.273539929678496e-06, "loss": 1.309, "step": 21498 }, { "epoch": 53.7475, "grad_norm": 9.5625, "learning_rate": 9.272732382601837e-06, "loss": 1.2084, "step": 21499 }, { "epoch": 53.75, "grad_norm": 8.125, "learning_rate": 9.27192484029312e-06, "loss": 1.4648, "step": 21500 }, { "epoch": 53.7525, "grad_norm": 6.25, "learning_rate": 9.271117302757634e-06, "loss": 1.2692, "step": 21501 }, { "epoch": 53.755, "grad_norm": 6.875, "learning_rate": 9.270309770000672e-06, "loss": 1.3478, "step": 21502 }, { "epoch": 53.7575, "grad_norm": 6.78125, "learning_rate": 9.269502242027532e-06, "loss": 1.2484, "step": 21503 }, { "epoch": 53.76, "grad_norm": 7.75, "learning_rate": 9.268694718843503e-06, "loss": 1.3405, "step": 21504 }, { "epoch": 53.7625, "grad_norm": 9.875, "learning_rate": 9.267887200453885e-06, "loss": 1.3932, "step": 21505 }, { "epoch": 53.765, "grad_norm": 5.9375, "learning_rate": 9.267079686863966e-06, "loss": 1.3753, "step": 21506 }, { "epoch": 53.7675, "grad_norm": 7.21875, "learning_rate": 9.26627217807905e-06, "loss": 1.3175, "step": 21507 }, { "epoch": 53.77, "grad_norm": 8.0625, "learning_rate": 9.265464674104418e-06, "loss": 1.3004, "step": 21508 }, { "epoch": 53.7725, "grad_norm": 9.125, "learning_rate": 9.26465717494537e-06, "loss": 1.3499, "step": 21509 }, { "epoch": 53.775, "grad_norm": 5.5625, "learning_rate": 9.2638496806072e-06, "loss": 1.253, "step": 21510 }, { "epoch": 53.7775, "grad_norm": 7.0625, "learning_rate": 9.263042191095208e-06, "loss": 1.3754, "step": 21511 }, { "epoch": 53.78, "grad_norm": 8.4375, "learning_rate": 9.262234706414677e-06, "loss": 1.251, "step": 21512 }, { "epoch": 53.7825, "grad_norm": 6.0, "learning_rate": 9.26142722657091e-06, "loss": 1.4036, "step": 21513 }, { "epoch": 53.785, "grad_norm": 8.875, "learning_rate": 9.260619751569188e-06, "loss": 1.2733, "step": 21514 }, { "epoch": 53.7875, "grad_norm": 10.0625, "learning_rate": 9.25981228141482e-06, "loss": 1.3032, "step": 21515 }, { "epoch": 53.79, "grad_norm": 6.65625, "learning_rate": 9.259004816113092e-06, "loss": 1.2431, "step": 21516 }, { "epoch": 53.7925, "grad_norm": 9.125, "learning_rate": 9.258197355669301e-06, "loss": 1.2637, "step": 21517 }, { "epoch": 53.795, "grad_norm": 8.25, "learning_rate": 9.257389900088735e-06, "loss": 1.1832, "step": 21518 }, { "epoch": 53.7975, "grad_norm": 6.0625, "learning_rate": 9.256582449376689e-06, "loss": 1.2801, "step": 21519 }, { "epoch": 53.8, "grad_norm": 7.59375, "learning_rate": 9.255775003538462e-06, "loss": 1.1393, "step": 21520 }, { "epoch": 53.8025, "grad_norm": 8.1875, "learning_rate": 9.254967562579346e-06, "loss": 1.3779, "step": 21521 }, { "epoch": 53.805, "grad_norm": 8.0, "learning_rate": 9.25416012650463e-06, "loss": 1.2754, "step": 21522 }, { "epoch": 53.8075, "grad_norm": 12.9375, "learning_rate": 9.253352695319611e-06, "loss": 1.3678, "step": 21523 }, { "epoch": 53.81, "grad_norm": 15.0, "learning_rate": 9.25254526902958e-06, "loss": 1.3698, "step": 21524 }, { "epoch": 53.8125, "grad_norm": 7.25, "learning_rate": 9.251737847639834e-06, "loss": 1.2728, "step": 21525 }, { "epoch": 53.815, "grad_norm": 7.65625, "learning_rate": 9.250930431155667e-06, "loss": 1.3142, "step": 21526 }, { "epoch": 53.8175, "grad_norm": 6.90625, "learning_rate": 9.250123019582368e-06, "loss": 1.1901, "step": 21527 }, { "epoch": 53.82, "grad_norm": 7.875, "learning_rate": 9.249315612925235e-06, "loss": 1.2493, "step": 21528 }, { "epoch": 53.8225, "grad_norm": 5.5625, "learning_rate": 9.24850821118955e-06, "loss": 1.3782, "step": 21529 }, { "epoch": 53.825, "grad_norm": 8.8125, "learning_rate": 9.247700814380625e-06, "loss": 1.4133, "step": 21530 }, { "epoch": 53.8275, "grad_norm": 6.375, "learning_rate": 9.24689342250374e-06, "loss": 1.3567, "step": 21531 }, { "epoch": 53.83, "grad_norm": 7.125, "learning_rate": 9.246086035564194e-06, "loss": 1.2969, "step": 21532 }, { "epoch": 53.8325, "grad_norm": 7.0, "learning_rate": 9.245278653567274e-06, "loss": 1.4423, "step": 21533 }, { "epoch": 53.835, "grad_norm": 9.3125, "learning_rate": 9.244471276518276e-06, "loss": 1.3086, "step": 21534 }, { "epoch": 53.8375, "grad_norm": 8.5, "learning_rate": 9.243663904422497e-06, "loss": 1.2405, "step": 21535 }, { "epoch": 53.84, "grad_norm": 7.59375, "learning_rate": 9.242856537285227e-06, "loss": 1.3542, "step": 21536 }, { "epoch": 53.8425, "grad_norm": 6.5625, "learning_rate": 9.24204917511176e-06, "loss": 1.3385, "step": 21537 }, { "epoch": 53.845, "grad_norm": 15.5625, "learning_rate": 9.241241817907387e-06, "loss": 1.3808, "step": 21538 }, { "epoch": 53.8475, "grad_norm": 9.625, "learning_rate": 9.2404344656774e-06, "loss": 1.3085, "step": 21539 }, { "epoch": 53.85, "grad_norm": 11.6875, "learning_rate": 9.239627118427098e-06, "loss": 1.4835, "step": 21540 }, { "epoch": 53.8525, "grad_norm": 6.375, "learning_rate": 9.238819776161767e-06, "loss": 1.2878, "step": 21541 }, { "epoch": 53.855, "grad_norm": 11.0, "learning_rate": 9.238012438886706e-06, "loss": 1.3333, "step": 21542 }, { "epoch": 53.8575, "grad_norm": 7.875, "learning_rate": 9.237205106607205e-06, "loss": 1.2515, "step": 21543 }, { "epoch": 53.86, "grad_norm": 7.15625, "learning_rate": 9.236397779328554e-06, "loss": 1.3203, "step": 21544 }, { "epoch": 53.8625, "grad_norm": 6.65625, "learning_rate": 9.235590457056049e-06, "loss": 1.4269, "step": 21545 }, { "epoch": 53.865, "grad_norm": 7.5625, "learning_rate": 9.234783139794982e-06, "loss": 1.2694, "step": 21546 }, { "epoch": 53.8675, "grad_norm": 6.09375, "learning_rate": 9.23397582755065e-06, "loss": 1.3451, "step": 21547 }, { "epoch": 53.87, "grad_norm": 6.6875, "learning_rate": 9.23316852032834e-06, "loss": 1.3353, "step": 21548 }, { "epoch": 53.8725, "grad_norm": 5.5, "learning_rate": 9.232361218133344e-06, "loss": 1.2724, "step": 21549 }, { "epoch": 53.875, "grad_norm": 7.59375, "learning_rate": 9.231553920970958e-06, "loss": 1.3554, "step": 21550 }, { "epoch": 53.8775, "grad_norm": 5.09375, "learning_rate": 9.230746628846478e-06, "loss": 1.2811, "step": 21551 }, { "epoch": 53.88, "grad_norm": 8.4375, "learning_rate": 9.229939341765188e-06, "loss": 1.2649, "step": 21552 }, { "epoch": 53.8825, "grad_norm": 6.9375, "learning_rate": 9.229132059732386e-06, "loss": 1.4087, "step": 21553 }, { "epoch": 53.885, "grad_norm": 5.4375, "learning_rate": 9.22832478275336e-06, "loss": 1.2944, "step": 21554 }, { "epoch": 53.8875, "grad_norm": 6.25, "learning_rate": 9.227517510833413e-06, "loss": 1.4239, "step": 21555 }, { "epoch": 53.89, "grad_norm": 9.5, "learning_rate": 9.226710243977825e-06, "loss": 1.3293, "step": 21556 }, { "epoch": 53.8925, "grad_norm": 8.75, "learning_rate": 9.225902982191895e-06, "loss": 1.3406, "step": 21557 }, { "epoch": 53.895, "grad_norm": 7.59375, "learning_rate": 9.225095725480918e-06, "loss": 1.2838, "step": 21558 }, { "epoch": 53.8975, "grad_norm": 7.71875, "learning_rate": 9.224288473850175e-06, "loss": 1.1728, "step": 21559 }, { "epoch": 53.9, "grad_norm": 6.6875, "learning_rate": 9.22348122730497e-06, "loss": 1.3857, "step": 21560 }, { "epoch": 53.9025, "grad_norm": 8.6875, "learning_rate": 9.222673985850589e-06, "loss": 1.2555, "step": 21561 }, { "epoch": 53.905, "grad_norm": 8.375, "learning_rate": 9.22186674949233e-06, "loss": 1.426, "step": 21562 }, { "epoch": 53.9075, "grad_norm": 10.0625, "learning_rate": 9.221059518235479e-06, "loss": 1.3573, "step": 21563 }, { "epoch": 53.91, "grad_norm": 9.5, "learning_rate": 9.220252292085328e-06, "loss": 1.3676, "step": 21564 }, { "epoch": 53.9125, "grad_norm": 6.5625, "learning_rate": 9.219445071047174e-06, "loss": 1.2567, "step": 21565 }, { "epoch": 53.915, "grad_norm": 6.9375, "learning_rate": 9.21863785512631e-06, "loss": 1.2966, "step": 21566 }, { "epoch": 53.9175, "grad_norm": 9.5625, "learning_rate": 9.217830644328023e-06, "loss": 1.4346, "step": 21567 }, { "epoch": 53.92, "grad_norm": 8.25, "learning_rate": 9.217023438657606e-06, "loss": 1.3749, "step": 21568 }, { "epoch": 53.9225, "grad_norm": 6.875, "learning_rate": 9.216216238120349e-06, "loss": 1.1916, "step": 21569 }, { "epoch": 53.925, "grad_norm": 11.5, "learning_rate": 9.215409042721553e-06, "loss": 1.301, "step": 21570 }, { "epoch": 53.9275, "grad_norm": 7.75, "learning_rate": 9.2146018524665e-06, "loss": 1.352, "step": 21571 }, { "epoch": 53.93, "grad_norm": 7.71875, "learning_rate": 9.213794667360488e-06, "loss": 1.3477, "step": 21572 }, { "epoch": 53.9325, "grad_norm": 11.0625, "learning_rate": 9.212987487408807e-06, "loss": 1.47, "step": 21573 }, { "epoch": 53.935, "grad_norm": 9.5625, "learning_rate": 9.212180312616744e-06, "loss": 1.4502, "step": 21574 }, { "epoch": 53.9375, "grad_norm": 8.6875, "learning_rate": 9.2113731429896e-06, "loss": 1.2885, "step": 21575 }, { "epoch": 53.94, "grad_norm": 5.9375, "learning_rate": 9.21056597853266e-06, "loss": 1.2754, "step": 21576 }, { "epoch": 53.9425, "grad_norm": 6.875, "learning_rate": 9.20975881925122e-06, "loss": 1.1652, "step": 21577 }, { "epoch": 53.945, "grad_norm": 8.6875, "learning_rate": 9.208951665150567e-06, "loss": 1.3154, "step": 21578 }, { "epoch": 53.9475, "grad_norm": 4.875, "learning_rate": 9.208144516235994e-06, "loss": 1.281, "step": 21579 }, { "epoch": 53.95, "grad_norm": 5.1875, "learning_rate": 9.207337372512797e-06, "loss": 1.3478, "step": 21580 }, { "epoch": 53.9525, "grad_norm": 6.28125, "learning_rate": 9.206530233986265e-06, "loss": 1.3088, "step": 21581 }, { "epoch": 53.955, "grad_norm": 11.0625, "learning_rate": 9.205723100661687e-06, "loss": 1.3414, "step": 21582 }, { "epoch": 53.9575, "grad_norm": 11.0, "learning_rate": 9.20491597254436e-06, "loss": 1.3043, "step": 21583 }, { "epoch": 53.96, "grad_norm": 6.6875, "learning_rate": 9.204108849639565e-06, "loss": 1.478, "step": 21584 }, { "epoch": 53.9625, "grad_norm": 7.53125, "learning_rate": 9.203301731952606e-06, "loss": 1.2819, "step": 21585 }, { "epoch": 53.965, "grad_norm": 5.875, "learning_rate": 9.202494619488768e-06, "loss": 1.2451, "step": 21586 }, { "epoch": 53.9675, "grad_norm": 6.65625, "learning_rate": 9.201687512253346e-06, "loss": 1.4668, "step": 21587 }, { "epoch": 53.97, "grad_norm": 7.90625, "learning_rate": 9.200880410251627e-06, "loss": 1.3179, "step": 21588 }, { "epoch": 53.9725, "grad_norm": 10.9375, "learning_rate": 9.2000733134889e-06, "loss": 1.3074, "step": 21589 }, { "epoch": 53.975, "grad_norm": 8.5625, "learning_rate": 9.199266221970461e-06, "loss": 1.3607, "step": 21590 }, { "epoch": 53.9775, "grad_norm": 6.0625, "learning_rate": 9.198459135701605e-06, "loss": 1.4396, "step": 21591 }, { "epoch": 53.98, "grad_norm": 8.125, "learning_rate": 9.197652054687619e-06, "loss": 1.2883, "step": 21592 }, { "epoch": 53.9825, "grad_norm": 7.1875, "learning_rate": 9.19684497893379e-06, "loss": 1.2714, "step": 21593 }, { "epoch": 53.985, "grad_norm": 12.625, "learning_rate": 9.196037908445412e-06, "loss": 1.3826, "step": 21594 }, { "epoch": 53.9875, "grad_norm": 8.875, "learning_rate": 9.19523084322778e-06, "loss": 1.445, "step": 21595 }, { "epoch": 53.99, "grad_norm": 7.8125, "learning_rate": 9.194423783286185e-06, "loss": 1.2896, "step": 21596 }, { "epoch": 53.9925, "grad_norm": 5.8125, "learning_rate": 9.193616728625913e-06, "loss": 1.3576, "step": 21597 }, { "epoch": 53.995, "grad_norm": 7.21875, "learning_rate": 9.19280967925226e-06, "loss": 1.5077, "step": 21598 }, { "epoch": 53.9975, "grad_norm": 8.625, "learning_rate": 9.192002635170507e-06, "loss": 1.2272, "step": 21599 }, { "epoch": 54.0, "grad_norm": 9.375, "learning_rate": 9.19119559638596e-06, "loss": 1.313, "step": 21600 }, { "epoch": 54.0025, "grad_norm": 7.375, "learning_rate": 9.190388562903898e-06, "loss": 1.5198, "step": 21601 }, { "epoch": 54.005, "grad_norm": 10.125, "learning_rate": 9.189581534729622e-06, "loss": 1.3906, "step": 21602 }, { "epoch": 54.0075, "grad_norm": 10.625, "learning_rate": 9.188774511868412e-06, "loss": 1.4111, "step": 21603 }, { "epoch": 54.01, "grad_norm": 7.96875, "learning_rate": 9.187967494325562e-06, "loss": 1.4002, "step": 21604 }, { "epoch": 54.0125, "grad_norm": 7.8125, "learning_rate": 9.187160482106367e-06, "loss": 1.3165, "step": 21605 }, { "epoch": 54.015, "grad_norm": 6.40625, "learning_rate": 9.186353475216118e-06, "loss": 1.2854, "step": 21606 }, { "epoch": 54.0175, "grad_norm": 6.8125, "learning_rate": 9.1855464736601e-06, "loss": 1.331, "step": 21607 }, { "epoch": 54.02, "grad_norm": 8.1875, "learning_rate": 9.184739477443608e-06, "loss": 1.2926, "step": 21608 }, { "epoch": 54.0225, "grad_norm": 7.125, "learning_rate": 9.183932486571928e-06, "loss": 1.3602, "step": 21609 }, { "epoch": 54.025, "grad_norm": 8.9375, "learning_rate": 9.183125501050358e-06, "loss": 1.2935, "step": 21610 }, { "epoch": 54.0275, "grad_norm": 7.1875, "learning_rate": 9.182318520884184e-06, "loss": 1.2914, "step": 21611 }, { "epoch": 54.03, "grad_norm": 6.84375, "learning_rate": 9.181511546078696e-06, "loss": 1.4082, "step": 21612 }, { "epoch": 54.0325, "grad_norm": 6.25, "learning_rate": 9.180704576639188e-06, "loss": 1.6583, "step": 21613 }, { "epoch": 54.035, "grad_norm": 6.6875, "learning_rate": 9.179897612570944e-06, "loss": 1.3885, "step": 21614 }, { "epoch": 54.0375, "grad_norm": 11.0625, "learning_rate": 9.17909065387926e-06, "loss": 1.2175, "step": 21615 }, { "epoch": 54.04, "grad_norm": 6.53125, "learning_rate": 9.178283700569424e-06, "loss": 1.3279, "step": 21616 }, { "epoch": 54.0425, "grad_norm": 6.78125, "learning_rate": 9.177476752646732e-06, "loss": 1.2331, "step": 21617 }, { "epoch": 54.045, "grad_norm": 6.78125, "learning_rate": 9.176669810116465e-06, "loss": 1.1736, "step": 21618 }, { "epoch": 54.0475, "grad_norm": 6.75, "learning_rate": 9.175862872983917e-06, "loss": 1.301, "step": 21619 }, { "epoch": 54.05, "grad_norm": 6.3125, "learning_rate": 9.17505594125438e-06, "loss": 1.42, "step": 21620 }, { "epoch": 54.0525, "grad_norm": 9.375, "learning_rate": 9.174249014933145e-06, "loss": 1.1769, "step": 21621 }, { "epoch": 54.055, "grad_norm": 7.6875, "learning_rate": 9.1734420940255e-06, "loss": 1.2861, "step": 21622 }, { "epoch": 54.0575, "grad_norm": 9.0, "learning_rate": 9.172635178536734e-06, "loss": 1.4322, "step": 21623 }, { "epoch": 54.06, "grad_norm": 7.75, "learning_rate": 9.171828268472137e-06, "loss": 1.3597, "step": 21624 }, { "epoch": 54.0625, "grad_norm": 5.03125, "learning_rate": 9.171021363837005e-06, "loss": 1.4922, "step": 21625 }, { "epoch": 54.065, "grad_norm": 9.375, "learning_rate": 9.170214464636622e-06, "loss": 1.3472, "step": 21626 }, { "epoch": 54.0675, "grad_norm": 8.0625, "learning_rate": 9.16940757087628e-06, "loss": 1.3514, "step": 21627 }, { "epoch": 54.07, "grad_norm": 5.90625, "learning_rate": 9.16860068256127e-06, "loss": 1.2866, "step": 21628 }, { "epoch": 54.0725, "grad_norm": 6.71875, "learning_rate": 9.167793799696877e-06, "loss": 1.4293, "step": 21629 }, { "epoch": 54.075, "grad_norm": 7.65625, "learning_rate": 9.166986922288396e-06, "loss": 1.454, "step": 21630 }, { "epoch": 54.0775, "grad_norm": 8.125, "learning_rate": 9.166180050341117e-06, "loss": 1.3844, "step": 21631 }, { "epoch": 54.08, "grad_norm": 10.0, "learning_rate": 9.165373183860329e-06, "loss": 1.3242, "step": 21632 }, { "epoch": 54.0825, "grad_norm": 6.28125, "learning_rate": 9.16456632285132e-06, "loss": 1.3739, "step": 21633 }, { "epoch": 54.085, "grad_norm": 7.09375, "learning_rate": 9.163759467319376e-06, "loss": 1.1987, "step": 21634 }, { "epoch": 54.0875, "grad_norm": 9.3125, "learning_rate": 9.162952617269796e-06, "loss": 1.4383, "step": 21635 }, { "epoch": 54.09, "grad_norm": 5.84375, "learning_rate": 9.162145772707867e-06, "loss": 1.2429, "step": 21636 }, { "epoch": 54.0925, "grad_norm": 10.25, "learning_rate": 9.161338933638873e-06, "loss": 1.331, "step": 21637 }, { "epoch": 54.095, "grad_norm": 7.34375, "learning_rate": 9.160532100068109e-06, "loss": 1.3334, "step": 21638 }, { "epoch": 54.0975, "grad_norm": 6.8125, "learning_rate": 9.159725272000861e-06, "loss": 1.2809, "step": 21639 }, { "epoch": 54.1, "grad_norm": 7.84375, "learning_rate": 9.158918449442425e-06, "loss": 1.3305, "step": 21640 }, { "epoch": 54.1025, "grad_norm": 7.09375, "learning_rate": 9.158111632398083e-06, "loss": 1.3821, "step": 21641 }, { "epoch": 54.105, "grad_norm": 8.6875, "learning_rate": 9.157304820873128e-06, "loss": 1.3292, "step": 21642 }, { "epoch": 54.1075, "grad_norm": 8.125, "learning_rate": 9.15649801487285e-06, "loss": 1.4478, "step": 21643 }, { "epoch": 54.11, "grad_norm": 7.28125, "learning_rate": 9.155691214402533e-06, "loss": 1.2686, "step": 21644 }, { "epoch": 54.1125, "grad_norm": 8.4375, "learning_rate": 9.154884419467474e-06, "loss": 1.2011, "step": 21645 }, { "epoch": 54.115, "grad_norm": 6.46875, "learning_rate": 9.154077630072957e-06, "loss": 1.2835, "step": 21646 }, { "epoch": 54.1175, "grad_norm": 6.40625, "learning_rate": 9.153270846224277e-06, "loss": 1.2303, "step": 21647 }, { "epoch": 54.12, "grad_norm": 7.5625, "learning_rate": 9.152464067926717e-06, "loss": 1.2827, "step": 21648 }, { "epoch": 54.1225, "grad_norm": 7.09375, "learning_rate": 9.151657295185566e-06, "loss": 1.3799, "step": 21649 }, { "epoch": 54.125, "grad_norm": 6.125, "learning_rate": 9.150850528006118e-06, "loss": 1.2912, "step": 21650 }, { "epoch": 54.1275, "grad_norm": 6.625, "learning_rate": 9.150043766393664e-06, "loss": 1.2592, "step": 21651 }, { "epoch": 54.13, "grad_norm": 8.25, "learning_rate": 9.149237010353485e-06, "loss": 1.484, "step": 21652 }, { "epoch": 54.1325, "grad_norm": 7.3125, "learning_rate": 9.148430259890876e-06, "loss": 1.1639, "step": 21653 }, { "epoch": 54.135, "grad_norm": 7.375, "learning_rate": 9.14762351501112e-06, "loss": 1.2839, "step": 21654 }, { "epoch": 54.1375, "grad_norm": 5.375, "learning_rate": 9.146816775719514e-06, "loss": 1.4073, "step": 21655 }, { "epoch": 54.14, "grad_norm": 10.6875, "learning_rate": 9.146010042021341e-06, "loss": 1.3326, "step": 21656 }, { "epoch": 54.1425, "grad_norm": 6.09375, "learning_rate": 9.145203313921896e-06, "loss": 1.2621, "step": 21657 }, { "epoch": 54.145, "grad_norm": 5.375, "learning_rate": 9.144396591426462e-06, "loss": 1.3059, "step": 21658 }, { "epoch": 54.1475, "grad_norm": 7.09375, "learning_rate": 9.143589874540323e-06, "loss": 1.2149, "step": 21659 }, { "epoch": 54.15, "grad_norm": 6.0, "learning_rate": 9.142783163268782e-06, "loss": 1.2678, "step": 21660 }, { "epoch": 54.1525, "grad_norm": 7.5, "learning_rate": 9.141976457617118e-06, "loss": 1.5471, "step": 21661 }, { "epoch": 54.155, "grad_norm": 11.0625, "learning_rate": 9.141169757590625e-06, "loss": 1.332, "step": 21662 }, { "epoch": 54.1575, "grad_norm": 7.71875, "learning_rate": 9.140363063194586e-06, "loss": 1.1501, "step": 21663 }, { "epoch": 54.16, "grad_norm": 6.0, "learning_rate": 9.139556374434288e-06, "loss": 1.1391, "step": 21664 }, { "epoch": 54.1625, "grad_norm": 6.8125, "learning_rate": 9.13874969131503e-06, "loss": 1.2562, "step": 21665 }, { "epoch": 54.165, "grad_norm": 6.5, "learning_rate": 9.137943013842094e-06, "loss": 1.3444, "step": 21666 }, { "epoch": 54.1675, "grad_norm": 6.875, "learning_rate": 9.137136342020768e-06, "loss": 1.3019, "step": 21667 }, { "epoch": 54.17, "grad_norm": 6.34375, "learning_rate": 9.136329675856344e-06, "loss": 1.284, "step": 21668 }, { "epoch": 54.1725, "grad_norm": 9.125, "learning_rate": 9.1355230153541e-06, "loss": 1.234, "step": 21669 }, { "epoch": 54.175, "grad_norm": 7.875, "learning_rate": 9.134716360519341e-06, "loss": 1.2695, "step": 21670 }, { "epoch": 54.1775, "grad_norm": 6.0625, "learning_rate": 9.133909711357345e-06, "loss": 1.2444, "step": 21671 }, { "epoch": 54.18, "grad_norm": 7.25, "learning_rate": 9.133103067873403e-06, "loss": 1.3839, "step": 21672 }, { "epoch": 54.1825, "grad_norm": 7.125, "learning_rate": 9.1322964300728e-06, "loss": 1.2461, "step": 21673 }, { "epoch": 54.185, "grad_norm": 7.53125, "learning_rate": 9.131489797960826e-06, "loss": 1.1832, "step": 21674 }, { "epoch": 54.1875, "grad_norm": 10.8125, "learning_rate": 9.130683171542772e-06, "loss": 1.2527, "step": 21675 }, { "epoch": 54.19, "grad_norm": 5.5, "learning_rate": 9.129876550823926e-06, "loss": 1.2053, "step": 21676 }, { "epoch": 54.1925, "grad_norm": 7.28125, "learning_rate": 9.129069935809573e-06, "loss": 1.322, "step": 21677 }, { "epoch": 54.195, "grad_norm": 6.0, "learning_rate": 9.128263326505003e-06, "loss": 1.4185, "step": 21678 }, { "epoch": 54.1975, "grad_norm": 9.1875, "learning_rate": 9.1274567229155e-06, "loss": 1.2212, "step": 21679 }, { "epoch": 54.2, "grad_norm": 8.375, "learning_rate": 9.126650125046361e-06, "loss": 1.2392, "step": 21680 }, { "epoch": 54.2025, "grad_norm": 8.5625, "learning_rate": 9.125843532902867e-06, "loss": 1.3689, "step": 21681 }, { "epoch": 54.205, "grad_norm": 5.84375, "learning_rate": 9.125036946490307e-06, "loss": 1.2621, "step": 21682 }, { "epoch": 54.2075, "grad_norm": 6.6875, "learning_rate": 9.124230365813974e-06, "loss": 1.3422, "step": 21683 }, { "epoch": 54.21, "grad_norm": 9.1875, "learning_rate": 9.123423790879146e-06, "loss": 1.3508, "step": 21684 }, { "epoch": 54.2125, "grad_norm": 9.75, "learning_rate": 9.122617221691117e-06, "loss": 1.3177, "step": 21685 }, { "epoch": 54.215, "grad_norm": 6.8125, "learning_rate": 9.121810658255177e-06, "loss": 1.297, "step": 21686 }, { "epoch": 54.2175, "grad_norm": 7.875, "learning_rate": 9.121004100576613e-06, "loss": 1.1469, "step": 21687 }, { "epoch": 54.22, "grad_norm": 7.9375, "learning_rate": 9.120197548660707e-06, "loss": 1.0942, "step": 21688 }, { "epoch": 54.2225, "grad_norm": 8.625, "learning_rate": 9.11939100251275e-06, "loss": 1.3731, "step": 21689 }, { "epoch": 54.225, "grad_norm": 7.59375, "learning_rate": 9.11858446213803e-06, "loss": 1.2866, "step": 21690 }, { "epoch": 54.2275, "grad_norm": 5.96875, "learning_rate": 9.117777927541841e-06, "loss": 1.3918, "step": 21691 }, { "epoch": 54.23, "grad_norm": 6.25, "learning_rate": 9.116971398729462e-06, "loss": 1.2046, "step": 21692 }, { "epoch": 54.2325, "grad_norm": 9.0, "learning_rate": 9.116164875706179e-06, "loss": 1.3319, "step": 21693 }, { "epoch": 54.235, "grad_norm": 5.34375, "learning_rate": 9.115358358477287e-06, "loss": 1.3599, "step": 21694 }, { "epoch": 54.2375, "grad_norm": 6.53125, "learning_rate": 9.11455184704807e-06, "loss": 1.2485, "step": 21695 }, { "epoch": 54.24, "grad_norm": 8.4375, "learning_rate": 9.113745341423816e-06, "loss": 1.2077, "step": 21696 }, { "epoch": 54.2425, "grad_norm": 8.4375, "learning_rate": 9.112938841609811e-06, "loss": 1.2304, "step": 21697 }, { "epoch": 54.245, "grad_norm": 6.59375, "learning_rate": 9.112132347611344e-06, "loss": 1.3324, "step": 21698 }, { "epoch": 54.2475, "grad_norm": 8.0625, "learning_rate": 9.111325859433703e-06, "loss": 1.4036, "step": 21699 }, { "epoch": 54.25, "grad_norm": 6.21875, "learning_rate": 9.110519377082174e-06, "loss": 1.3139, "step": 21700 }, { "epoch": 54.2525, "grad_norm": 12.0, "learning_rate": 9.109712900562043e-06, "loss": 1.3114, "step": 21701 }, { "epoch": 54.255, "grad_norm": 7.6875, "learning_rate": 9.1089064298786e-06, "loss": 1.3042, "step": 21702 }, { "epoch": 54.2575, "grad_norm": 9.6875, "learning_rate": 9.108099965037128e-06, "loss": 1.2178, "step": 21703 }, { "epoch": 54.26, "grad_norm": 11.8125, "learning_rate": 9.107293506042918e-06, "loss": 1.3009, "step": 21704 }, { "epoch": 54.2625, "grad_norm": 7.5625, "learning_rate": 9.106487052901256e-06, "loss": 1.3517, "step": 21705 }, { "epoch": 54.265, "grad_norm": 5.46875, "learning_rate": 9.105680605617433e-06, "loss": 1.3399, "step": 21706 }, { "epoch": 54.2675, "grad_norm": 5.90625, "learning_rate": 9.104874164196729e-06, "loss": 1.1519, "step": 21707 }, { "epoch": 54.27, "grad_norm": 8.75, "learning_rate": 9.10406772864443e-06, "loss": 1.253, "step": 21708 }, { "epoch": 54.2725, "grad_norm": 8.3125, "learning_rate": 9.103261298965833e-06, "loss": 1.2972, "step": 21709 }, { "epoch": 54.275, "grad_norm": 7.03125, "learning_rate": 9.102454875166219e-06, "loss": 1.523, "step": 21710 }, { "epoch": 54.2775, "grad_norm": 9.3125, "learning_rate": 9.101648457250873e-06, "loss": 1.4057, "step": 21711 }, { "epoch": 54.28, "grad_norm": 10.375, "learning_rate": 9.100842045225084e-06, "loss": 1.2688, "step": 21712 }, { "epoch": 54.2825, "grad_norm": 6.40625, "learning_rate": 9.100035639094136e-06, "loss": 1.3548, "step": 21713 }, { "epoch": 54.285, "grad_norm": 7.0, "learning_rate": 9.099229238863324e-06, "loss": 1.3122, "step": 21714 }, { "epoch": 54.2875, "grad_norm": 4.9375, "learning_rate": 9.098422844537925e-06, "loss": 1.434, "step": 21715 }, { "epoch": 54.29, "grad_norm": 6.375, "learning_rate": 9.09761645612323e-06, "loss": 1.1692, "step": 21716 }, { "epoch": 54.2925, "grad_norm": 10.5625, "learning_rate": 9.096810073624529e-06, "loss": 1.4267, "step": 21717 }, { "epoch": 54.295, "grad_norm": 8.9375, "learning_rate": 9.0960036970471e-06, "loss": 1.2473, "step": 21718 }, { "epoch": 54.2975, "grad_norm": 6.53125, "learning_rate": 9.095197326396237e-06, "loss": 1.2083, "step": 21719 }, { "epoch": 54.3, "grad_norm": 10.75, "learning_rate": 9.094390961677223e-06, "loss": 1.3288, "step": 21720 }, { "epoch": 54.3025, "grad_norm": 7.78125, "learning_rate": 9.09358460289535e-06, "loss": 1.4033, "step": 21721 }, { "epoch": 54.305, "grad_norm": 6.3125, "learning_rate": 9.092778250055895e-06, "loss": 1.1411, "step": 21722 }, { "epoch": 54.3075, "grad_norm": 8.375, "learning_rate": 9.091971903164148e-06, "loss": 1.2434, "step": 21723 }, { "epoch": 54.31, "grad_norm": 5.5625, "learning_rate": 9.091165562225401e-06, "loss": 1.1863, "step": 21724 }, { "epoch": 54.3125, "grad_norm": 8.125, "learning_rate": 9.090359227244936e-06, "loss": 1.23, "step": 21725 }, { "epoch": 54.315, "grad_norm": 8.4375, "learning_rate": 9.08955289822804e-06, "loss": 1.3808, "step": 21726 }, { "epoch": 54.3175, "grad_norm": 18.5, "learning_rate": 9.088746575179999e-06, "loss": 1.3967, "step": 21727 }, { "epoch": 54.32, "grad_norm": 7.65625, "learning_rate": 9.087940258106093e-06, "loss": 1.3541, "step": 21728 }, { "epoch": 54.3225, "grad_norm": 8.5625, "learning_rate": 9.08713394701162e-06, "loss": 1.359, "step": 21729 }, { "epoch": 54.325, "grad_norm": 8.1875, "learning_rate": 9.08632764190186e-06, "loss": 1.3946, "step": 21730 }, { "epoch": 54.3275, "grad_norm": 9.6875, "learning_rate": 9.085521342782096e-06, "loss": 1.3915, "step": 21731 }, { "epoch": 54.33, "grad_norm": 7.71875, "learning_rate": 9.084715049657622e-06, "loss": 1.3883, "step": 21732 }, { "epoch": 54.3325, "grad_norm": 5.78125, "learning_rate": 9.083908762533714e-06, "loss": 1.3074, "step": 21733 }, { "epoch": 54.335, "grad_norm": 7.875, "learning_rate": 9.083102481415668e-06, "loss": 1.4673, "step": 21734 }, { "epoch": 54.3375, "grad_norm": 6.96875, "learning_rate": 9.082296206308764e-06, "loss": 1.0971, "step": 21735 }, { "epoch": 54.34, "grad_norm": 11.375, "learning_rate": 9.081489937218293e-06, "loss": 1.4511, "step": 21736 }, { "epoch": 54.3425, "grad_norm": 6.875, "learning_rate": 9.080683674149533e-06, "loss": 1.3683, "step": 21737 }, { "epoch": 54.345, "grad_norm": 8.125, "learning_rate": 9.079877417107771e-06, "loss": 1.221, "step": 21738 }, { "epoch": 54.3475, "grad_norm": 8.6875, "learning_rate": 9.0790711660983e-06, "loss": 1.3569, "step": 21739 }, { "epoch": 54.35, "grad_norm": 8.3125, "learning_rate": 9.078264921126405e-06, "loss": 1.35, "step": 21740 }, { "epoch": 54.3525, "grad_norm": 8.875, "learning_rate": 9.077458682197364e-06, "loss": 1.5313, "step": 21741 }, { "epoch": 54.355, "grad_norm": 8.6875, "learning_rate": 9.076652449316468e-06, "loss": 1.4187, "step": 21742 }, { "epoch": 54.3575, "grad_norm": 11.875, "learning_rate": 9.075846222488997e-06, "loss": 1.3686, "step": 21743 }, { "epoch": 54.36, "grad_norm": 12.125, "learning_rate": 9.075040001720247e-06, "loss": 1.309, "step": 21744 }, { "epoch": 54.3625, "grad_norm": 6.9375, "learning_rate": 9.074233787015497e-06, "loss": 1.1134, "step": 21745 }, { "epoch": 54.365, "grad_norm": 6.46875, "learning_rate": 9.073427578380033e-06, "loss": 1.4314, "step": 21746 }, { "epoch": 54.3675, "grad_norm": 8.3125, "learning_rate": 9.072621375819141e-06, "loss": 1.2754, "step": 21747 }, { "epoch": 54.37, "grad_norm": 7.90625, "learning_rate": 9.071815179338101e-06, "loss": 1.472, "step": 21748 }, { "epoch": 54.3725, "grad_norm": 6.75, "learning_rate": 9.071008988942208e-06, "loss": 1.0816, "step": 21749 }, { "epoch": 54.375, "grad_norm": 6.5, "learning_rate": 9.070202804636745e-06, "loss": 1.2521, "step": 21750 }, { "epoch": 54.3775, "grad_norm": 5.625, "learning_rate": 9.069396626426993e-06, "loss": 1.2438, "step": 21751 }, { "epoch": 54.38, "grad_norm": 5.53125, "learning_rate": 9.06859045431824e-06, "loss": 1.3697, "step": 21752 }, { "epoch": 54.3825, "grad_norm": 9.125, "learning_rate": 9.067784288315766e-06, "loss": 1.4025, "step": 21753 }, { "epoch": 54.385, "grad_norm": 7.6875, "learning_rate": 9.066978128424867e-06, "loss": 1.3422, "step": 21754 }, { "epoch": 54.3875, "grad_norm": 9.125, "learning_rate": 9.066171974650821e-06, "loss": 1.4802, "step": 21755 }, { "epoch": 54.39, "grad_norm": 7.53125, "learning_rate": 9.065365826998913e-06, "loss": 1.4634, "step": 21756 }, { "epoch": 54.3925, "grad_norm": 7.34375, "learning_rate": 9.06455968547443e-06, "loss": 1.4276, "step": 21757 }, { "epoch": 54.395, "grad_norm": 7.5, "learning_rate": 9.063753550082652e-06, "loss": 1.3203, "step": 21758 }, { "epoch": 54.3975, "grad_norm": 12.5625, "learning_rate": 9.062947420828873e-06, "loss": 1.358, "step": 21759 }, { "epoch": 54.4, "grad_norm": 8.75, "learning_rate": 9.062141297718372e-06, "loss": 1.343, "step": 21760 }, { "epoch": 54.4025, "grad_norm": 8.375, "learning_rate": 9.061335180756436e-06, "loss": 1.4257, "step": 21761 }, { "epoch": 54.405, "grad_norm": 8.8125, "learning_rate": 9.06052906994835e-06, "loss": 1.4372, "step": 21762 }, { "epoch": 54.4075, "grad_norm": 8.4375, "learning_rate": 9.059722965299393e-06, "loss": 1.295, "step": 21763 }, { "epoch": 54.41, "grad_norm": 7.34375, "learning_rate": 9.058916866814857e-06, "loss": 1.3232, "step": 21764 }, { "epoch": 54.4125, "grad_norm": 8.75, "learning_rate": 9.058110774500026e-06, "loss": 1.5238, "step": 21765 }, { "epoch": 54.415, "grad_norm": 8.25, "learning_rate": 9.057304688360183e-06, "loss": 1.2712, "step": 21766 }, { "epoch": 54.4175, "grad_norm": 6.34375, "learning_rate": 9.056498608400611e-06, "loss": 1.0807, "step": 21767 }, { "epoch": 54.42, "grad_norm": 6.1875, "learning_rate": 9.055692534626595e-06, "loss": 1.4189, "step": 21768 }, { "epoch": 54.4225, "grad_norm": 7.15625, "learning_rate": 9.054886467043425e-06, "loss": 1.6536, "step": 21769 }, { "epoch": 54.425, "grad_norm": 7.65625, "learning_rate": 9.05408040565638e-06, "loss": 1.2885, "step": 21770 }, { "epoch": 54.4275, "grad_norm": 9.25, "learning_rate": 9.053274350470745e-06, "loss": 1.3285, "step": 21771 }, { "epoch": 54.43, "grad_norm": 11.25, "learning_rate": 9.052468301491809e-06, "loss": 1.2288, "step": 21772 }, { "epoch": 54.4325, "grad_norm": 9.5, "learning_rate": 9.051662258724848e-06, "loss": 1.3456, "step": 21773 }, { "epoch": 54.435, "grad_norm": 6.5, "learning_rate": 9.050856222175151e-06, "loss": 1.4505, "step": 21774 }, { "epoch": 54.4375, "grad_norm": 8.375, "learning_rate": 9.050050191848006e-06, "loss": 1.3115, "step": 21775 }, { "epoch": 54.44, "grad_norm": 7.75, "learning_rate": 9.049244167748694e-06, "loss": 1.2938, "step": 21776 }, { "epoch": 54.4425, "grad_norm": 13.125, "learning_rate": 9.048438149882497e-06, "loss": 1.4285, "step": 21777 }, { "epoch": 54.445, "grad_norm": 6.4375, "learning_rate": 9.0476321382547e-06, "loss": 1.2603, "step": 21778 }, { "epoch": 54.4475, "grad_norm": 5.96875, "learning_rate": 9.046826132870591e-06, "loss": 1.2439, "step": 21779 }, { "epoch": 54.45, "grad_norm": 6.375, "learning_rate": 9.046020133735455e-06, "loss": 1.4251, "step": 21780 }, { "epoch": 54.4525, "grad_norm": 9.0, "learning_rate": 9.045214140854568e-06, "loss": 1.2702, "step": 21781 }, { "epoch": 54.455, "grad_norm": 8.125, "learning_rate": 9.04440815423322e-06, "loss": 1.3684, "step": 21782 }, { "epoch": 54.4575, "grad_norm": 6.875, "learning_rate": 9.043602173876694e-06, "loss": 1.1047, "step": 21783 }, { "epoch": 54.46, "grad_norm": 11.8125, "learning_rate": 9.042796199790274e-06, "loss": 1.286, "step": 21784 }, { "epoch": 54.4625, "grad_norm": 8.875, "learning_rate": 9.041990231979244e-06, "loss": 1.6203, "step": 21785 }, { "epoch": 54.465, "grad_norm": 6.46875, "learning_rate": 9.04118427044889e-06, "loss": 1.458, "step": 21786 }, { "epoch": 54.4675, "grad_norm": 8.875, "learning_rate": 9.040378315204492e-06, "loss": 1.3554, "step": 21787 }, { "epoch": 54.47, "grad_norm": 9.8125, "learning_rate": 9.039572366251333e-06, "loss": 1.3021, "step": 21788 }, { "epoch": 54.4725, "grad_norm": 10.875, "learning_rate": 9.0387664235947e-06, "loss": 1.2351, "step": 21789 }, { "epoch": 54.475, "grad_norm": 11.4375, "learning_rate": 9.037960487239876e-06, "loss": 1.2669, "step": 21790 }, { "epoch": 54.4775, "grad_norm": 8.5625, "learning_rate": 9.037154557192147e-06, "loss": 1.3075, "step": 21791 }, { "epoch": 54.48, "grad_norm": 7.59375, "learning_rate": 9.036348633456791e-06, "loss": 1.2629, "step": 21792 }, { "epoch": 54.4825, "grad_norm": 4.96875, "learning_rate": 9.035542716039095e-06, "loss": 1.3965, "step": 21793 }, { "epoch": 54.485, "grad_norm": 8.0625, "learning_rate": 9.034736804944343e-06, "loss": 1.4788, "step": 21794 }, { "epoch": 54.4875, "grad_norm": 7.90625, "learning_rate": 9.033930900177821e-06, "loss": 1.3275, "step": 21795 }, { "epoch": 54.49, "grad_norm": 9.0, "learning_rate": 9.033125001744807e-06, "loss": 1.4012, "step": 21796 }, { "epoch": 54.4925, "grad_norm": 8.0625, "learning_rate": 9.032319109650589e-06, "loss": 1.327, "step": 21797 }, { "epoch": 54.495, "grad_norm": 8.6875, "learning_rate": 9.03151322390044e-06, "loss": 1.3609, "step": 21798 }, { "epoch": 54.4975, "grad_norm": 6.375, "learning_rate": 9.030707344499659e-06, "loss": 1.3579, "step": 21799 }, { "epoch": 54.5, "grad_norm": 9.0625, "learning_rate": 9.02990147145352e-06, "loss": 1.465, "step": 21800 }, { "epoch": 54.5025, "grad_norm": 7.03125, "learning_rate": 9.029095604767309e-06, "loss": 1.3058, "step": 21801 }, { "epoch": 54.505, "grad_norm": 6.90625, "learning_rate": 9.02828974444631e-06, "loss": 1.2696, "step": 21802 }, { "epoch": 54.5075, "grad_norm": 13.5625, "learning_rate": 9.027483890495798e-06, "loss": 1.2507, "step": 21803 }, { "epoch": 54.51, "grad_norm": 8.0625, "learning_rate": 9.026678042921068e-06, "loss": 1.3018, "step": 21804 }, { "epoch": 54.5125, "grad_norm": 9.0, "learning_rate": 9.025872201727396e-06, "loss": 1.2897, "step": 21805 }, { "epoch": 54.515, "grad_norm": 21.125, "learning_rate": 9.02506636692007e-06, "loss": 1.2199, "step": 21806 }, { "epoch": 54.5175, "grad_norm": 7.6875, "learning_rate": 9.024260538504367e-06, "loss": 1.3848, "step": 21807 }, { "epoch": 54.52, "grad_norm": 6.46875, "learning_rate": 9.023454716485572e-06, "loss": 1.3608, "step": 21808 }, { "epoch": 54.5225, "grad_norm": 5.4375, "learning_rate": 9.022648900868969e-06, "loss": 1.2815, "step": 21809 }, { "epoch": 54.525, "grad_norm": 7.90625, "learning_rate": 9.021843091659842e-06, "loss": 1.4628, "step": 21810 }, { "epoch": 54.5275, "grad_norm": 9.3125, "learning_rate": 9.021037288863473e-06, "loss": 1.3419, "step": 21811 }, { "epoch": 54.53, "grad_norm": 11.75, "learning_rate": 9.020231492485146e-06, "loss": 1.3104, "step": 21812 }, { "epoch": 54.5325, "grad_norm": 8.4375, "learning_rate": 9.019425702530135e-06, "loss": 1.2401, "step": 21813 }, { "epoch": 54.535, "grad_norm": 5.59375, "learning_rate": 9.018619919003737e-06, "loss": 1.5678, "step": 21814 }, { "epoch": 54.5375, "grad_norm": 8.1875, "learning_rate": 9.017814141911226e-06, "loss": 1.2458, "step": 21815 }, { "epoch": 54.54, "grad_norm": 6.0, "learning_rate": 9.017008371257887e-06, "loss": 1.3644, "step": 21816 }, { "epoch": 54.5425, "grad_norm": 11.8125, "learning_rate": 9.016202607049e-06, "loss": 1.2766, "step": 21817 }, { "epoch": 54.545, "grad_norm": 7.21875, "learning_rate": 9.015396849289847e-06, "loss": 1.1453, "step": 21818 }, { "epoch": 54.5475, "grad_norm": 7.625, "learning_rate": 9.014591097985715e-06, "loss": 1.2413, "step": 21819 }, { "epoch": 54.55, "grad_norm": 5.6875, "learning_rate": 9.013785353141887e-06, "loss": 1.1901, "step": 21820 }, { "epoch": 54.5525, "grad_norm": 7.46875, "learning_rate": 9.012979614763639e-06, "loss": 1.374, "step": 21821 }, { "epoch": 54.555, "grad_norm": 6.3125, "learning_rate": 9.012173882856259e-06, "loss": 1.4331, "step": 21822 }, { "epoch": 54.5575, "grad_norm": 5.6875, "learning_rate": 9.011368157425024e-06, "loss": 1.3322, "step": 21823 }, { "epoch": 54.56, "grad_norm": 9.25, "learning_rate": 9.010562438475225e-06, "loss": 1.3523, "step": 21824 }, { "epoch": 54.5625, "grad_norm": 7.0625, "learning_rate": 9.009756726012138e-06, "loss": 1.3532, "step": 21825 }, { "epoch": 54.565, "grad_norm": 7.9375, "learning_rate": 9.008951020041044e-06, "loss": 1.4193, "step": 21826 }, { "epoch": 54.5675, "grad_norm": 6.8125, "learning_rate": 9.008145320567233e-06, "loss": 1.2668, "step": 21827 }, { "epoch": 54.57, "grad_norm": 14.9375, "learning_rate": 9.007339627595972e-06, "loss": 1.3791, "step": 21828 }, { "epoch": 54.5725, "grad_norm": 9.0, "learning_rate": 9.00653394113256e-06, "loss": 1.5375, "step": 21829 }, { "epoch": 54.575, "grad_norm": 10.75, "learning_rate": 9.005728261182268e-06, "loss": 1.3695, "step": 21830 }, { "epoch": 54.5775, "grad_norm": 7.3125, "learning_rate": 9.004922587750386e-06, "loss": 1.3183, "step": 21831 }, { "epoch": 54.58, "grad_norm": 8.75, "learning_rate": 9.004116920842188e-06, "loss": 1.2762, "step": 21832 }, { "epoch": 54.5825, "grad_norm": 7.4375, "learning_rate": 9.003311260462959e-06, "loss": 1.271, "step": 21833 }, { "epoch": 54.585, "grad_norm": 10.3125, "learning_rate": 9.002505606617983e-06, "loss": 1.4721, "step": 21834 }, { "epoch": 54.5875, "grad_norm": 8.375, "learning_rate": 9.001699959312543e-06, "loss": 1.265, "step": 21835 }, { "epoch": 54.59, "grad_norm": 8.1875, "learning_rate": 9.000894318551914e-06, "loss": 1.2669, "step": 21836 }, { "epoch": 54.5925, "grad_norm": 7.15625, "learning_rate": 9.000088684341385e-06, "loss": 1.2235, "step": 21837 }, { "epoch": 54.595, "grad_norm": 7.15625, "learning_rate": 8.99928305668623e-06, "loss": 1.181, "step": 21838 }, { "epoch": 54.5975, "grad_norm": 7.375, "learning_rate": 8.99847743559174e-06, "loss": 1.1787, "step": 21839 }, { "epoch": 54.6, "grad_norm": 10.1875, "learning_rate": 8.99767182106319e-06, "loss": 1.3248, "step": 21840 }, { "epoch": 54.6025, "grad_norm": 6.59375, "learning_rate": 8.996866213105863e-06, "loss": 1.1815, "step": 21841 }, { "epoch": 54.605, "grad_norm": 5.875, "learning_rate": 8.996060611725045e-06, "loss": 1.3904, "step": 21842 }, { "epoch": 54.6075, "grad_norm": 10.3125, "learning_rate": 8.995255016926007e-06, "loss": 1.3323, "step": 21843 }, { "epoch": 54.61, "grad_norm": 7.40625, "learning_rate": 8.994449428714041e-06, "loss": 1.3443, "step": 21844 }, { "epoch": 54.6125, "grad_norm": 5.96875, "learning_rate": 8.993643847094422e-06, "loss": 1.3549, "step": 21845 }, { "epoch": 54.615, "grad_norm": 10.8125, "learning_rate": 8.992838272072438e-06, "loss": 1.3974, "step": 21846 }, { "epoch": 54.6175, "grad_norm": 6.46875, "learning_rate": 8.992032703653362e-06, "loss": 1.4329, "step": 21847 }, { "epoch": 54.62, "grad_norm": 6.96875, "learning_rate": 8.991227141842477e-06, "loss": 1.1531, "step": 21848 }, { "epoch": 54.6225, "grad_norm": 8.0, "learning_rate": 8.99042158664507e-06, "loss": 1.377, "step": 21849 }, { "epoch": 54.625, "grad_norm": 6.625, "learning_rate": 8.98961603806642e-06, "loss": 1.4887, "step": 21850 }, { "epoch": 54.6275, "grad_norm": 6.09375, "learning_rate": 8.988810496111805e-06, "loss": 1.3384, "step": 21851 }, { "epoch": 54.63, "grad_norm": 6.125, "learning_rate": 8.988004960786508e-06, "loss": 1.2479, "step": 21852 }, { "epoch": 54.6325, "grad_norm": 9.375, "learning_rate": 8.987199432095808e-06, "loss": 1.4459, "step": 21853 }, { "epoch": 54.635, "grad_norm": 11.5625, "learning_rate": 8.986393910044992e-06, "loss": 1.4078, "step": 21854 }, { "epoch": 54.6375, "grad_norm": 8.5, "learning_rate": 8.985588394639335e-06, "loss": 1.1635, "step": 21855 }, { "epoch": 54.64, "grad_norm": 9.8125, "learning_rate": 8.984782885884119e-06, "loss": 1.4759, "step": 21856 }, { "epoch": 54.6425, "grad_norm": 7.9375, "learning_rate": 8.98397738378463e-06, "loss": 1.2555, "step": 21857 }, { "epoch": 54.645, "grad_norm": 7.15625, "learning_rate": 8.983171888346137e-06, "loss": 1.2732, "step": 21858 }, { "epoch": 54.6475, "grad_norm": 6.375, "learning_rate": 8.982366399573933e-06, "loss": 1.2384, "step": 21859 }, { "epoch": 54.65, "grad_norm": 8.375, "learning_rate": 8.981560917473292e-06, "loss": 1.3216, "step": 21860 }, { "epoch": 54.6525, "grad_norm": 9.875, "learning_rate": 8.980755442049503e-06, "loss": 1.4143, "step": 21861 }, { "epoch": 54.655, "grad_norm": 7.34375, "learning_rate": 8.979949973307836e-06, "loss": 1.2843, "step": 21862 }, { "epoch": 54.6575, "grad_norm": 9.0, "learning_rate": 8.979144511253573e-06, "loss": 1.614, "step": 21863 }, { "epoch": 54.66, "grad_norm": 8.5625, "learning_rate": 8.978339055892001e-06, "loss": 1.3997, "step": 21864 }, { "epoch": 54.6625, "grad_norm": 6.4375, "learning_rate": 8.9775336072284e-06, "loss": 1.2924, "step": 21865 }, { "epoch": 54.665, "grad_norm": 6.78125, "learning_rate": 8.976728165268046e-06, "loss": 1.2716, "step": 21866 }, { "epoch": 54.6675, "grad_norm": 7.09375, "learning_rate": 8.975922730016222e-06, "loss": 1.343, "step": 21867 }, { "epoch": 54.67, "grad_norm": 6.71875, "learning_rate": 8.9751173014782e-06, "loss": 1.3334, "step": 21868 }, { "epoch": 54.6725, "grad_norm": 8.875, "learning_rate": 8.974311879659277e-06, "loss": 1.3492, "step": 21869 }, { "epoch": 54.675, "grad_norm": 8.5625, "learning_rate": 8.973506464564721e-06, "loss": 1.3232, "step": 21870 }, { "epoch": 54.6775, "grad_norm": 6.03125, "learning_rate": 8.972701056199815e-06, "loss": 1.3515, "step": 21871 }, { "epoch": 54.68, "grad_norm": 9.4375, "learning_rate": 8.971895654569842e-06, "loss": 1.3921, "step": 21872 }, { "epoch": 54.6825, "grad_norm": 8.0, "learning_rate": 8.971090259680077e-06, "loss": 1.3591, "step": 21873 }, { "epoch": 54.685, "grad_norm": 10.5, "learning_rate": 8.970284871535804e-06, "loss": 1.3212, "step": 21874 }, { "epoch": 54.6875, "grad_norm": 5.0625, "learning_rate": 8.969479490142302e-06, "loss": 1.3035, "step": 21875 }, { "epoch": 54.69, "grad_norm": 9.0, "learning_rate": 8.968674115504853e-06, "loss": 1.2789, "step": 21876 }, { "epoch": 54.6925, "grad_norm": 8.6875, "learning_rate": 8.967868747628735e-06, "loss": 1.3499, "step": 21877 }, { "epoch": 54.695, "grad_norm": 5.375, "learning_rate": 8.967063386519224e-06, "loss": 1.3243, "step": 21878 }, { "epoch": 54.6975, "grad_norm": 7.09375, "learning_rate": 8.966258032181608e-06, "loss": 1.2439, "step": 21879 }, { "epoch": 54.7, "grad_norm": 8.8125, "learning_rate": 8.965452684621164e-06, "loss": 1.2304, "step": 21880 }, { "epoch": 54.7025, "grad_norm": 6.8125, "learning_rate": 8.96464734384317e-06, "loss": 1.3059, "step": 21881 }, { "epoch": 54.705, "grad_norm": 11.5, "learning_rate": 8.963842009852908e-06, "loss": 1.2543, "step": 21882 }, { "epoch": 54.7075, "grad_norm": 5.65625, "learning_rate": 8.963036682655649e-06, "loss": 1.3824, "step": 21883 }, { "epoch": 54.71, "grad_norm": 6.40625, "learning_rate": 8.962231362256688e-06, "loss": 1.2626, "step": 21884 }, { "epoch": 54.7125, "grad_norm": 6.875, "learning_rate": 8.961426048661292e-06, "loss": 1.3734, "step": 21885 }, { "epoch": 54.715, "grad_norm": 5.40625, "learning_rate": 8.960620741874749e-06, "loss": 1.2328, "step": 21886 }, { "epoch": 54.7175, "grad_norm": 14.875, "learning_rate": 8.959815441902333e-06, "loss": 1.2844, "step": 21887 }, { "epoch": 54.72, "grad_norm": 11.875, "learning_rate": 8.959010148749324e-06, "loss": 1.2116, "step": 21888 }, { "epoch": 54.7225, "grad_norm": 8.75, "learning_rate": 8.958204862421003e-06, "loss": 1.2913, "step": 21889 }, { "epoch": 54.725, "grad_norm": 8.0625, "learning_rate": 8.957399582922652e-06, "loss": 1.3479, "step": 21890 }, { "epoch": 54.7275, "grad_norm": 5.25, "learning_rate": 8.956594310259546e-06, "loss": 1.3382, "step": 21891 }, { "epoch": 54.73, "grad_norm": 7.40625, "learning_rate": 8.955789044436966e-06, "loss": 1.4318, "step": 21892 }, { "epoch": 54.7325, "grad_norm": 5.5, "learning_rate": 8.954983785460188e-06, "loss": 1.3432, "step": 21893 }, { "epoch": 54.735, "grad_norm": 7.28125, "learning_rate": 8.954178533334498e-06, "loss": 1.3322, "step": 21894 }, { "epoch": 54.7375, "grad_norm": 7.78125, "learning_rate": 8.953373288065173e-06, "loss": 1.1827, "step": 21895 }, { "epoch": 54.74, "grad_norm": 8.0625, "learning_rate": 8.952568049657488e-06, "loss": 1.3108, "step": 21896 }, { "epoch": 54.7425, "grad_norm": 7.03125, "learning_rate": 8.951762818116729e-06, "loss": 1.4041, "step": 21897 }, { "epoch": 54.745, "grad_norm": 6.5625, "learning_rate": 8.950957593448163e-06, "loss": 1.2618, "step": 21898 }, { "epoch": 54.7475, "grad_norm": 5.53125, "learning_rate": 8.950152375657084e-06, "loss": 1.3293, "step": 21899 }, { "epoch": 54.75, "grad_norm": 4.90625, "learning_rate": 8.949347164748761e-06, "loss": 1.4548, "step": 21900 }, { "epoch": 54.7525, "grad_norm": 5.90625, "learning_rate": 8.94854196072848e-06, "loss": 1.248, "step": 21901 }, { "epoch": 54.755, "grad_norm": 8.625, "learning_rate": 8.947736763601512e-06, "loss": 1.5157, "step": 21902 }, { "epoch": 54.7575, "grad_norm": 8.625, "learning_rate": 8.946931573373138e-06, "loss": 1.3648, "step": 21903 }, { "epoch": 54.76, "grad_norm": 9.3125, "learning_rate": 8.94612639004864e-06, "loss": 1.4063, "step": 21904 }, { "epoch": 54.7625, "grad_norm": 8.375, "learning_rate": 8.9453212136333e-06, "loss": 1.3135, "step": 21905 }, { "epoch": 54.765, "grad_norm": 7.65625, "learning_rate": 8.944516044132388e-06, "loss": 1.2379, "step": 21906 }, { "epoch": 54.7675, "grad_norm": 9.0625, "learning_rate": 8.943710881551186e-06, "loss": 1.3146, "step": 21907 }, { "epoch": 54.77, "grad_norm": 6.5625, "learning_rate": 8.94290572589497e-06, "loss": 1.435, "step": 21908 }, { "epoch": 54.7725, "grad_norm": 7.6875, "learning_rate": 8.942100577169028e-06, "loss": 1.2335, "step": 21909 }, { "epoch": 54.775, "grad_norm": 11.9375, "learning_rate": 8.94129543537863e-06, "loss": 1.2534, "step": 21910 }, { "epoch": 54.7775, "grad_norm": 7.34375, "learning_rate": 8.940490300529056e-06, "loss": 1.3822, "step": 21911 }, { "epoch": 54.78, "grad_norm": 5.625, "learning_rate": 8.939685172625588e-06, "loss": 1.258, "step": 21912 }, { "epoch": 54.7825, "grad_norm": 5.4375, "learning_rate": 8.938880051673498e-06, "loss": 1.3766, "step": 21913 }, { "epoch": 54.785, "grad_norm": 9.625, "learning_rate": 8.938074937678069e-06, "loss": 1.349, "step": 21914 }, { "epoch": 54.7875, "grad_norm": 7.75, "learning_rate": 8.937269830644578e-06, "loss": 1.2353, "step": 21915 }, { "epoch": 54.79, "grad_norm": 6.40625, "learning_rate": 8.936464730578305e-06, "loss": 1.396, "step": 21916 }, { "epoch": 54.7925, "grad_norm": 10.3125, "learning_rate": 8.935659637484526e-06, "loss": 1.2676, "step": 21917 }, { "epoch": 54.795, "grad_norm": 8.0625, "learning_rate": 8.934854551368517e-06, "loss": 1.4587, "step": 21918 }, { "epoch": 54.7975, "grad_norm": 7.5, "learning_rate": 8.934049472235563e-06, "loss": 1.1729, "step": 21919 }, { "epoch": 54.8, "grad_norm": 8.625, "learning_rate": 8.933244400090937e-06, "loss": 1.4559, "step": 21920 }, { "epoch": 54.8025, "grad_norm": 7.9375, "learning_rate": 8.932439334939917e-06, "loss": 1.4784, "step": 21921 }, { "epoch": 54.805, "grad_norm": 7.28125, "learning_rate": 8.931634276787783e-06, "loss": 1.4622, "step": 21922 }, { "epoch": 54.8075, "grad_norm": 9.8125, "learning_rate": 8.93082922563981e-06, "loss": 1.3562, "step": 21923 }, { "epoch": 54.81, "grad_norm": 6.0, "learning_rate": 8.93002418150128e-06, "loss": 1.3701, "step": 21924 }, { "epoch": 54.8125, "grad_norm": 8.5, "learning_rate": 8.929219144377468e-06, "loss": 1.3441, "step": 21925 }, { "epoch": 54.815, "grad_norm": 6.03125, "learning_rate": 8.928414114273651e-06, "loss": 1.263, "step": 21926 }, { "epoch": 54.8175, "grad_norm": 8.0625, "learning_rate": 8.927609091195111e-06, "loss": 1.3599, "step": 21927 }, { "epoch": 54.82, "grad_norm": 11.75, "learning_rate": 8.92680407514712e-06, "loss": 1.4052, "step": 21928 }, { "epoch": 54.8225, "grad_norm": 5.21875, "learning_rate": 8.92599906613496e-06, "loss": 1.2346, "step": 21929 }, { "epoch": 54.825, "grad_norm": 6.75, "learning_rate": 8.925194064163907e-06, "loss": 1.4359, "step": 21930 }, { "epoch": 54.8275, "grad_norm": 9.9375, "learning_rate": 8.924389069239242e-06, "loss": 1.2743, "step": 21931 }, { "epoch": 54.83, "grad_norm": 9.25, "learning_rate": 8.923584081366234e-06, "loss": 1.2329, "step": 21932 }, { "epoch": 54.8325, "grad_norm": 7.5625, "learning_rate": 8.922779100550164e-06, "loss": 1.3373, "step": 21933 }, { "epoch": 54.835, "grad_norm": 9.3125, "learning_rate": 8.921974126796315e-06, "loss": 1.2904, "step": 21934 }, { "epoch": 54.8375, "grad_norm": 8.125, "learning_rate": 8.921169160109965e-06, "loss": 1.3876, "step": 21935 }, { "epoch": 54.84, "grad_norm": 9.1875, "learning_rate": 8.92036420049638e-06, "loss": 1.1643, "step": 21936 }, { "epoch": 54.8425, "grad_norm": 7.3125, "learning_rate": 8.919559247960845e-06, "loss": 1.3543, "step": 21937 }, { "epoch": 54.845, "grad_norm": 6.4375, "learning_rate": 8.918754302508636e-06, "loss": 1.1987, "step": 21938 }, { "epoch": 54.8475, "grad_norm": 10.5625, "learning_rate": 8.917949364145035e-06, "loss": 1.3034, "step": 21939 }, { "epoch": 54.85, "grad_norm": 8.5625, "learning_rate": 8.91714443287531e-06, "loss": 1.2983, "step": 21940 }, { "epoch": 54.8525, "grad_norm": 9.4375, "learning_rate": 8.916339508704745e-06, "loss": 1.3937, "step": 21941 }, { "epoch": 54.855, "grad_norm": 6.375, "learning_rate": 8.915534591638611e-06, "loss": 1.4304, "step": 21942 }, { "epoch": 54.8575, "grad_norm": 10.875, "learning_rate": 8.914729681682194e-06, "loss": 1.4326, "step": 21943 }, { "epoch": 54.86, "grad_norm": 21.625, "learning_rate": 8.913924778840764e-06, "loss": 1.2488, "step": 21944 }, { "epoch": 54.8625, "grad_norm": 8.0, "learning_rate": 8.9131198831196e-06, "loss": 1.3718, "step": 21945 }, { "epoch": 54.865, "grad_norm": 6.3125, "learning_rate": 8.91231499452398e-06, "loss": 1.3066, "step": 21946 }, { "epoch": 54.8675, "grad_norm": 5.5625, "learning_rate": 8.911510113059174e-06, "loss": 1.3976, "step": 21947 }, { "epoch": 54.87, "grad_norm": 8.3125, "learning_rate": 8.910705238730467e-06, "loss": 1.3147, "step": 21948 }, { "epoch": 54.8725, "grad_norm": 7.3125, "learning_rate": 8.909900371543134e-06, "loss": 1.2762, "step": 21949 }, { "epoch": 54.875, "grad_norm": 8.75, "learning_rate": 8.909095511502452e-06, "loss": 1.4605, "step": 21950 }, { "epoch": 54.8775, "grad_norm": 5.3125, "learning_rate": 8.908290658613694e-06, "loss": 1.3037, "step": 21951 }, { "epoch": 54.88, "grad_norm": 7.6875, "learning_rate": 8.907485812882137e-06, "loss": 1.4794, "step": 21952 }, { "epoch": 54.8825, "grad_norm": 11.5625, "learning_rate": 8.90668097431306e-06, "loss": 1.2674, "step": 21953 }, { "epoch": 54.885, "grad_norm": 7.5625, "learning_rate": 8.905876142911744e-06, "loss": 1.0888, "step": 21954 }, { "epoch": 54.8875, "grad_norm": 8.0625, "learning_rate": 8.905071318683454e-06, "loss": 1.3152, "step": 21955 }, { "epoch": 54.89, "grad_norm": 7.8125, "learning_rate": 8.904266501633478e-06, "loss": 1.4773, "step": 21956 }, { "epoch": 54.8925, "grad_norm": 9.25, "learning_rate": 8.903461691767079e-06, "loss": 1.3819, "step": 21957 }, { "epoch": 54.895, "grad_norm": 7.53125, "learning_rate": 8.902656889089548e-06, "loss": 1.4514, "step": 21958 }, { "epoch": 54.8975, "grad_norm": 7.46875, "learning_rate": 8.901852093606152e-06, "loss": 1.2011, "step": 21959 }, { "epoch": 54.9, "grad_norm": 5.9375, "learning_rate": 8.901047305322172e-06, "loss": 1.3653, "step": 21960 }, { "epoch": 54.9025, "grad_norm": 8.8125, "learning_rate": 8.900242524242882e-06, "loss": 1.3208, "step": 21961 }, { "epoch": 54.905, "grad_norm": 8.625, "learning_rate": 8.899437750373553e-06, "loss": 1.3247, "step": 21962 }, { "epoch": 54.9075, "grad_norm": 8.125, "learning_rate": 8.898632983719468e-06, "loss": 1.2332, "step": 21963 }, { "epoch": 54.91, "grad_norm": 7.65625, "learning_rate": 8.897828224285905e-06, "loss": 1.3544, "step": 21964 }, { "epoch": 54.9125, "grad_norm": 6.9375, "learning_rate": 8.897023472078134e-06, "loss": 1.2939, "step": 21965 }, { "epoch": 54.915, "grad_norm": 7.6875, "learning_rate": 8.896218727101433e-06, "loss": 1.2369, "step": 21966 }, { "epoch": 54.9175, "grad_norm": 7.21875, "learning_rate": 8.895413989361074e-06, "loss": 1.2153, "step": 21967 }, { "epoch": 54.92, "grad_norm": 13.0625, "learning_rate": 8.89460925886234e-06, "loss": 1.151, "step": 21968 }, { "epoch": 54.9225, "grad_norm": 7.875, "learning_rate": 8.893804535610505e-06, "loss": 1.2179, "step": 21969 }, { "epoch": 54.925, "grad_norm": 8.0, "learning_rate": 8.892999819610842e-06, "loss": 1.2811, "step": 21970 }, { "epoch": 54.9275, "grad_norm": 9.25, "learning_rate": 8.89219511086863e-06, "loss": 1.4793, "step": 21971 }, { "epoch": 54.93, "grad_norm": 6.5, "learning_rate": 8.891390409389133e-06, "loss": 1.3171, "step": 21972 }, { "epoch": 54.9325, "grad_norm": 6.84375, "learning_rate": 8.890585715177646e-06, "loss": 1.3488, "step": 21973 }, { "epoch": 54.935, "grad_norm": 7.125, "learning_rate": 8.889781028239432e-06, "loss": 1.4812, "step": 21974 }, { "epoch": 54.9375, "grad_norm": 8.875, "learning_rate": 8.88897634857977e-06, "loss": 1.3623, "step": 21975 }, { "epoch": 54.94, "grad_norm": 8.25, "learning_rate": 8.888171676203933e-06, "loss": 1.3857, "step": 21976 }, { "epoch": 54.9425, "grad_norm": 8.375, "learning_rate": 8.887367011117198e-06, "loss": 1.352, "step": 21977 }, { "epoch": 54.945, "grad_norm": 6.21875, "learning_rate": 8.88656235332484e-06, "loss": 1.4401, "step": 21978 }, { "epoch": 54.9475, "grad_norm": 7.96875, "learning_rate": 8.885757702832137e-06, "loss": 1.201, "step": 21979 }, { "epoch": 54.95, "grad_norm": 7.625, "learning_rate": 8.88495305964436e-06, "loss": 1.1535, "step": 21980 }, { "epoch": 54.9525, "grad_norm": 7.84375, "learning_rate": 8.884148423766787e-06, "loss": 1.4238, "step": 21981 }, { "epoch": 54.955, "grad_norm": 7.6875, "learning_rate": 8.88334379520469e-06, "loss": 1.3023, "step": 21982 }, { "epoch": 54.9575, "grad_norm": 9.1875, "learning_rate": 8.882539173963349e-06, "loss": 1.5152, "step": 21983 }, { "epoch": 54.96, "grad_norm": 6.25, "learning_rate": 8.881734560048037e-06, "loss": 1.3967, "step": 21984 }, { "epoch": 54.9625, "grad_norm": 6.8125, "learning_rate": 8.880929953464025e-06, "loss": 1.3434, "step": 21985 }, { "epoch": 54.965, "grad_norm": 7.25, "learning_rate": 8.880125354216597e-06, "loss": 1.4322, "step": 21986 }, { "epoch": 54.9675, "grad_norm": 8.8125, "learning_rate": 8.879320762311017e-06, "loss": 1.2988, "step": 21987 }, { "epoch": 54.97, "grad_norm": 8.125, "learning_rate": 8.878516177752566e-06, "loss": 1.2694, "step": 21988 }, { "epoch": 54.9725, "grad_norm": 7.03125, "learning_rate": 8.877711600546519e-06, "loss": 1.2665, "step": 21989 }, { "epoch": 54.975, "grad_norm": 7.46875, "learning_rate": 8.876907030698152e-06, "loss": 1.3974, "step": 21990 }, { "epoch": 54.9775, "grad_norm": 7.09375, "learning_rate": 8.876102468212735e-06, "loss": 1.2332, "step": 21991 }, { "epoch": 54.98, "grad_norm": 8.125, "learning_rate": 8.875297913095544e-06, "loss": 1.324, "step": 21992 }, { "epoch": 54.9825, "grad_norm": 6.3125, "learning_rate": 8.874493365351857e-06, "loss": 1.2175, "step": 21993 }, { "epoch": 54.985, "grad_norm": 9.625, "learning_rate": 8.873688824986948e-06, "loss": 1.3426, "step": 21994 }, { "epoch": 54.9875, "grad_norm": 8.0625, "learning_rate": 8.872884292006088e-06, "loss": 1.3608, "step": 21995 }, { "epoch": 54.99, "grad_norm": 5.09375, "learning_rate": 8.872079766414554e-06, "loss": 1.2671, "step": 21996 }, { "epoch": 54.9925, "grad_norm": 8.4375, "learning_rate": 8.871275248217617e-06, "loss": 1.2463, "step": 21997 }, { "epoch": 54.995, "grad_norm": 8.875, "learning_rate": 8.870470737420559e-06, "loss": 1.3807, "step": 21998 }, { "epoch": 54.9975, "grad_norm": 8.1875, "learning_rate": 8.869666234028647e-06, "loss": 1.4282, "step": 21999 }, { "epoch": 55.0, "grad_norm": 5.0, "learning_rate": 8.868861738047158e-06, "loss": 1.2946, "step": 22000 }, { "epoch": 55.0025, "grad_norm": 10.75, "learning_rate": 8.868057249481369e-06, "loss": 1.4796, "step": 22001 }, { "epoch": 55.005, "grad_norm": 6.96875, "learning_rate": 8.867252768336547e-06, "loss": 1.1368, "step": 22002 }, { "epoch": 55.0075, "grad_norm": 8.3125, "learning_rate": 8.866448294617973e-06, "loss": 1.4077, "step": 22003 }, { "epoch": 55.01, "grad_norm": 9.0625, "learning_rate": 8.865643828330917e-06, "loss": 1.361, "step": 22004 }, { "epoch": 55.0125, "grad_norm": 6.75, "learning_rate": 8.864839369480658e-06, "loss": 1.2476, "step": 22005 }, { "epoch": 55.015, "grad_norm": 8.0625, "learning_rate": 8.864034918072464e-06, "loss": 1.2271, "step": 22006 }, { "epoch": 55.0175, "grad_norm": 9.375, "learning_rate": 8.86323047411161e-06, "loss": 1.2702, "step": 22007 }, { "epoch": 55.02, "grad_norm": 9.75, "learning_rate": 8.862426037603373e-06, "loss": 1.3221, "step": 22008 }, { "epoch": 55.0225, "grad_norm": 7.46875, "learning_rate": 8.861621608553027e-06, "loss": 1.5315, "step": 22009 }, { "epoch": 55.025, "grad_norm": 10.4375, "learning_rate": 8.860817186965845e-06, "loss": 1.3715, "step": 22010 }, { "epoch": 55.0275, "grad_norm": 6.09375, "learning_rate": 8.860012772847097e-06, "loss": 1.2923, "step": 22011 }, { "epoch": 55.03, "grad_norm": 7.375, "learning_rate": 8.859208366202059e-06, "loss": 1.2604, "step": 22012 }, { "epoch": 55.0325, "grad_norm": 6.96875, "learning_rate": 8.85840396703601e-06, "loss": 1.1981, "step": 22013 }, { "epoch": 55.035, "grad_norm": 6.46875, "learning_rate": 8.857599575354214e-06, "loss": 1.5332, "step": 22014 }, { "epoch": 55.0375, "grad_norm": 7.90625, "learning_rate": 8.856795191161952e-06, "loss": 1.3677, "step": 22015 }, { "epoch": 55.04, "grad_norm": 7.96875, "learning_rate": 8.855990814464497e-06, "loss": 1.2278, "step": 22016 }, { "epoch": 55.0425, "grad_norm": 6.59375, "learning_rate": 8.855186445267114e-06, "loss": 1.351, "step": 22017 }, { "epoch": 55.045, "grad_norm": 11.375, "learning_rate": 8.854382083575088e-06, "loss": 1.2775, "step": 22018 }, { "epoch": 55.0475, "grad_norm": 6.53125, "learning_rate": 8.853577729393686e-06, "loss": 1.3087, "step": 22019 }, { "epoch": 55.05, "grad_norm": 7.4375, "learning_rate": 8.852773382728184e-06, "loss": 1.3366, "step": 22020 }, { "epoch": 55.0525, "grad_norm": 6.59375, "learning_rate": 8.851969043583851e-06, "loss": 1.3309, "step": 22021 }, { "epoch": 55.055, "grad_norm": 6.1875, "learning_rate": 8.851164711965962e-06, "loss": 1.2392, "step": 22022 }, { "epoch": 55.0575, "grad_norm": 6.6875, "learning_rate": 8.850360387879793e-06, "loss": 1.2442, "step": 22023 }, { "epoch": 55.06, "grad_norm": 9.8125, "learning_rate": 8.849556071330618e-06, "loss": 1.4025, "step": 22024 }, { "epoch": 55.0625, "grad_norm": 6.9375, "learning_rate": 8.848751762323705e-06, "loss": 1.2144, "step": 22025 }, { "epoch": 55.065, "grad_norm": 13.0, "learning_rate": 8.847947460864332e-06, "loss": 1.2826, "step": 22026 }, { "epoch": 55.0675, "grad_norm": 6.65625, "learning_rate": 8.847143166957763e-06, "loss": 1.4156, "step": 22027 }, { "epoch": 55.07, "grad_norm": 8.1875, "learning_rate": 8.846338880609282e-06, "loss": 1.3104, "step": 22028 }, { "epoch": 55.0725, "grad_norm": 7.90625, "learning_rate": 8.845534601824156e-06, "loss": 1.5047, "step": 22029 }, { "epoch": 55.075, "grad_norm": 10.25, "learning_rate": 8.844730330607661e-06, "loss": 1.2862, "step": 22030 }, { "epoch": 55.0775, "grad_norm": 5.875, "learning_rate": 8.843926066965066e-06, "loss": 1.4319, "step": 22031 }, { "epoch": 55.08, "grad_norm": 7.46875, "learning_rate": 8.843121810901643e-06, "loss": 1.3054, "step": 22032 }, { "epoch": 55.0825, "grad_norm": 6.5, "learning_rate": 8.84231756242267e-06, "loss": 1.0713, "step": 22033 }, { "epoch": 55.085, "grad_norm": 9.9375, "learning_rate": 8.841513321533418e-06, "loss": 1.4983, "step": 22034 }, { "epoch": 55.0875, "grad_norm": 8.1875, "learning_rate": 8.840709088239155e-06, "loss": 1.3846, "step": 22035 }, { "epoch": 55.09, "grad_norm": 5.40625, "learning_rate": 8.839904862545159e-06, "loss": 1.3142, "step": 22036 }, { "epoch": 55.0925, "grad_norm": 6.5, "learning_rate": 8.839100644456698e-06, "loss": 1.2142, "step": 22037 }, { "epoch": 55.095, "grad_norm": 8.25, "learning_rate": 8.838296433979047e-06, "loss": 1.2766, "step": 22038 }, { "epoch": 55.0975, "grad_norm": 5.875, "learning_rate": 8.837492231117481e-06, "loss": 1.244, "step": 22039 }, { "epoch": 55.1, "grad_norm": 6.46875, "learning_rate": 8.836688035877268e-06, "loss": 1.3112, "step": 22040 }, { "epoch": 55.1025, "grad_norm": 10.3125, "learning_rate": 8.835883848263684e-06, "loss": 1.3466, "step": 22041 }, { "epoch": 55.105, "grad_norm": 7.21875, "learning_rate": 8.835079668281992e-06, "loss": 1.1795, "step": 22042 }, { "epoch": 55.1075, "grad_norm": 8.5625, "learning_rate": 8.834275495937477e-06, "loss": 1.3503, "step": 22043 }, { "epoch": 55.11, "grad_norm": 7.4375, "learning_rate": 8.833471331235404e-06, "loss": 1.2966, "step": 22044 }, { "epoch": 55.1125, "grad_norm": 11.25, "learning_rate": 8.832667174181047e-06, "loss": 1.282, "step": 22045 }, { "epoch": 55.115, "grad_norm": 8.5, "learning_rate": 8.831863024779676e-06, "loss": 1.3808, "step": 22046 }, { "epoch": 55.1175, "grad_norm": 8.0625, "learning_rate": 8.831058883036563e-06, "loss": 1.277, "step": 22047 }, { "epoch": 55.12, "grad_norm": 7.1875, "learning_rate": 8.830254748956983e-06, "loss": 1.2855, "step": 22048 }, { "epoch": 55.1225, "grad_norm": 8.25, "learning_rate": 8.829450622546208e-06, "loss": 1.2798, "step": 22049 }, { "epoch": 55.125, "grad_norm": 6.6875, "learning_rate": 8.828646503809505e-06, "loss": 1.1716, "step": 22050 }, { "epoch": 55.1275, "grad_norm": 20.125, "learning_rate": 8.82784239275215e-06, "loss": 1.3314, "step": 22051 }, { "epoch": 55.13, "grad_norm": 6.40625, "learning_rate": 8.827038289379409e-06, "loss": 1.2803, "step": 22052 }, { "epoch": 55.1325, "grad_norm": 5.875, "learning_rate": 8.826234193696564e-06, "loss": 1.273, "step": 22053 }, { "epoch": 55.135, "grad_norm": 7.25, "learning_rate": 8.825430105708877e-06, "loss": 1.3581, "step": 22054 }, { "epoch": 55.1375, "grad_norm": 5.53125, "learning_rate": 8.824626025421625e-06, "loss": 1.3118, "step": 22055 }, { "epoch": 55.14, "grad_norm": 8.5, "learning_rate": 8.823821952840079e-06, "loss": 1.2331, "step": 22056 }, { "epoch": 55.1425, "grad_norm": 6.46875, "learning_rate": 8.823017887969506e-06, "loss": 1.3191, "step": 22057 }, { "epoch": 55.145, "grad_norm": 7.59375, "learning_rate": 8.822213830815182e-06, "loss": 1.3775, "step": 22058 }, { "epoch": 55.1475, "grad_norm": 5.8125, "learning_rate": 8.821409781382376e-06, "loss": 1.374, "step": 22059 }, { "epoch": 55.15, "grad_norm": 10.25, "learning_rate": 8.820605739676363e-06, "loss": 1.3744, "step": 22060 }, { "epoch": 55.1525, "grad_norm": 6.34375, "learning_rate": 8.819801705702411e-06, "loss": 1.4315, "step": 22061 }, { "epoch": 55.155, "grad_norm": 8.1875, "learning_rate": 8.818997679465788e-06, "loss": 1.3186, "step": 22062 }, { "epoch": 55.1575, "grad_norm": 12.4375, "learning_rate": 8.818193660971771e-06, "loss": 1.3427, "step": 22063 }, { "epoch": 55.16, "grad_norm": 8.125, "learning_rate": 8.817389650225631e-06, "loss": 1.4745, "step": 22064 }, { "epoch": 55.1625, "grad_norm": 9.1875, "learning_rate": 8.816585647232637e-06, "loss": 1.1609, "step": 22065 }, { "epoch": 55.165, "grad_norm": 8.1875, "learning_rate": 8.815781651998059e-06, "loss": 1.2875, "step": 22066 }, { "epoch": 55.1675, "grad_norm": 7.78125, "learning_rate": 8.814977664527166e-06, "loss": 1.3494, "step": 22067 }, { "epoch": 55.17, "grad_norm": 10.1875, "learning_rate": 8.814173684825237e-06, "loss": 1.1626, "step": 22068 }, { "epoch": 55.1725, "grad_norm": 7.59375, "learning_rate": 8.813369712897538e-06, "loss": 1.452, "step": 22069 }, { "epoch": 55.175, "grad_norm": 7.75, "learning_rate": 8.812565748749336e-06, "loss": 1.3967, "step": 22070 }, { "epoch": 55.1775, "grad_norm": 6.5, "learning_rate": 8.81176179238591e-06, "loss": 1.2616, "step": 22071 }, { "epoch": 55.18, "grad_norm": 9.5, "learning_rate": 8.81095784381252e-06, "loss": 1.2539, "step": 22072 }, { "epoch": 55.1825, "grad_norm": 6.03125, "learning_rate": 8.810153903034446e-06, "loss": 1.4651, "step": 22073 }, { "epoch": 55.185, "grad_norm": 7.3125, "learning_rate": 8.809349970056956e-06, "loss": 1.3183, "step": 22074 }, { "epoch": 55.1875, "grad_norm": 14.875, "learning_rate": 8.80854604488532e-06, "loss": 1.2383, "step": 22075 }, { "epoch": 55.19, "grad_norm": 7.875, "learning_rate": 8.807742127524808e-06, "loss": 1.3777, "step": 22076 }, { "epoch": 55.1925, "grad_norm": 8.1875, "learning_rate": 8.80693821798069e-06, "loss": 1.3221, "step": 22077 }, { "epoch": 55.195, "grad_norm": 8.1875, "learning_rate": 8.806134316258236e-06, "loss": 1.3704, "step": 22078 }, { "epoch": 55.1975, "grad_norm": 7.8125, "learning_rate": 8.805330422362722e-06, "loss": 1.3267, "step": 22079 }, { "epoch": 55.2, "grad_norm": 8.0, "learning_rate": 8.804526536299413e-06, "loss": 1.4119, "step": 22080 }, { "epoch": 55.2025, "grad_norm": 6.65625, "learning_rate": 8.803722658073578e-06, "loss": 1.3166, "step": 22081 }, { "epoch": 55.205, "grad_norm": 6.78125, "learning_rate": 8.802918787690487e-06, "loss": 1.4371, "step": 22082 }, { "epoch": 55.2075, "grad_norm": 11.875, "learning_rate": 8.802114925155418e-06, "loss": 1.2654, "step": 22083 }, { "epoch": 55.21, "grad_norm": 6.65625, "learning_rate": 8.801311070473634e-06, "loss": 1.2612, "step": 22084 }, { "epoch": 55.2125, "grad_norm": 5.15625, "learning_rate": 8.800507223650406e-06, "loss": 1.187, "step": 22085 }, { "epoch": 55.215, "grad_norm": 7.28125, "learning_rate": 8.799703384691005e-06, "loss": 1.279, "step": 22086 }, { "epoch": 55.2175, "grad_norm": 7.125, "learning_rate": 8.798899553600699e-06, "loss": 1.3754, "step": 22087 }, { "epoch": 55.22, "grad_norm": 6.5, "learning_rate": 8.79809573038476e-06, "loss": 1.3165, "step": 22088 }, { "epoch": 55.2225, "grad_norm": 6.9375, "learning_rate": 8.797291915048458e-06, "loss": 1.3823, "step": 22089 }, { "epoch": 55.225, "grad_norm": 7.0625, "learning_rate": 8.796488107597064e-06, "loss": 1.5274, "step": 22090 }, { "epoch": 55.2275, "grad_norm": 5.8125, "learning_rate": 8.795684308035844e-06, "loss": 1.2966, "step": 22091 }, { "epoch": 55.23, "grad_norm": 6.53125, "learning_rate": 8.794880516370066e-06, "loss": 1.3459, "step": 22092 }, { "epoch": 55.2325, "grad_norm": 8.125, "learning_rate": 8.794076732605009e-06, "loss": 1.1752, "step": 22093 }, { "epoch": 55.235, "grad_norm": 9.4375, "learning_rate": 8.793272956745934e-06, "loss": 1.4739, "step": 22094 }, { "epoch": 55.2375, "grad_norm": 7.0625, "learning_rate": 8.792469188798114e-06, "loss": 1.1987, "step": 22095 }, { "epoch": 55.24, "grad_norm": 12.125, "learning_rate": 8.79166542876682e-06, "loss": 1.4644, "step": 22096 }, { "epoch": 55.2425, "grad_norm": 10.25, "learning_rate": 8.79086167665731e-06, "loss": 1.4663, "step": 22097 }, { "epoch": 55.245, "grad_norm": 8.0, "learning_rate": 8.790057932474871e-06, "loss": 1.3686, "step": 22098 }, { "epoch": 55.2475, "grad_norm": 12.25, "learning_rate": 8.789254196224762e-06, "loss": 1.2916, "step": 22099 }, { "epoch": 55.25, "grad_norm": 7.3125, "learning_rate": 8.788450467912254e-06, "loss": 1.2994, "step": 22100 }, { "epoch": 55.2525, "grad_norm": 6.375, "learning_rate": 8.787646747542617e-06, "loss": 1.5303, "step": 22101 }, { "epoch": 55.255, "grad_norm": 7.6875, "learning_rate": 8.786843035121115e-06, "loss": 1.395, "step": 22102 }, { "epoch": 55.2575, "grad_norm": 4.90625, "learning_rate": 8.786039330653025e-06, "loss": 1.3322, "step": 22103 }, { "epoch": 55.26, "grad_norm": 8.5, "learning_rate": 8.785235634143612e-06, "loss": 1.3861, "step": 22104 }, { "epoch": 55.2625, "grad_norm": 6.8125, "learning_rate": 8.784431945598146e-06, "loss": 1.1427, "step": 22105 }, { "epoch": 55.265, "grad_norm": 6.1875, "learning_rate": 8.783628265021896e-06, "loss": 1.2622, "step": 22106 }, { "epoch": 55.2675, "grad_norm": 10.25, "learning_rate": 8.782824592420128e-06, "loss": 1.3419, "step": 22107 }, { "epoch": 55.27, "grad_norm": 8.8125, "learning_rate": 8.782020927798114e-06, "loss": 1.396, "step": 22108 }, { "epoch": 55.2725, "grad_norm": 6.34375, "learning_rate": 8.781217271161125e-06, "loss": 1.2124, "step": 22109 }, { "epoch": 55.275, "grad_norm": 5.9375, "learning_rate": 8.780413622514424e-06, "loss": 1.2484, "step": 22110 }, { "epoch": 55.2775, "grad_norm": 8.375, "learning_rate": 8.779609981863287e-06, "loss": 1.3784, "step": 22111 }, { "epoch": 55.28, "grad_norm": 6.84375, "learning_rate": 8.778806349212968e-06, "loss": 1.2051, "step": 22112 }, { "epoch": 55.2825, "grad_norm": 7.46875, "learning_rate": 8.778002724568754e-06, "loss": 1.2189, "step": 22113 }, { "epoch": 55.285, "grad_norm": 7.75, "learning_rate": 8.777199107935904e-06, "loss": 1.5361, "step": 22114 }, { "epoch": 55.2875, "grad_norm": 7.03125, "learning_rate": 8.776395499319689e-06, "loss": 1.4324, "step": 22115 }, { "epoch": 55.29, "grad_norm": 7.3125, "learning_rate": 8.775591898725374e-06, "loss": 1.3341, "step": 22116 }, { "epoch": 55.2925, "grad_norm": 6.96875, "learning_rate": 8.774788306158227e-06, "loss": 1.3998, "step": 22117 }, { "epoch": 55.295, "grad_norm": 9.6875, "learning_rate": 8.773984721623521e-06, "loss": 1.3062, "step": 22118 }, { "epoch": 55.2975, "grad_norm": 9.75, "learning_rate": 8.773181145126526e-06, "loss": 1.3941, "step": 22119 }, { "epoch": 55.3, "grad_norm": 6.6875, "learning_rate": 8.772377576672502e-06, "loss": 1.3023, "step": 22120 }, { "epoch": 55.3025, "grad_norm": 7.28125, "learning_rate": 8.771574016266722e-06, "loss": 1.2116, "step": 22121 }, { "epoch": 55.305, "grad_norm": 8.8125, "learning_rate": 8.770770463914452e-06, "loss": 1.3969, "step": 22122 }, { "epoch": 55.3075, "grad_norm": 7.59375, "learning_rate": 8.769966919620966e-06, "loss": 1.3755, "step": 22123 }, { "epoch": 55.31, "grad_norm": 8.625, "learning_rate": 8.769163383391525e-06, "loss": 1.2858, "step": 22124 }, { "epoch": 55.3125, "grad_norm": 5.25, "learning_rate": 8.7683598552314e-06, "loss": 1.2485, "step": 22125 }, { "epoch": 55.315, "grad_norm": 7.1875, "learning_rate": 8.76755633514586e-06, "loss": 1.4445, "step": 22126 }, { "epoch": 55.3175, "grad_norm": 8.75, "learning_rate": 8.766752823140168e-06, "loss": 1.3335, "step": 22127 }, { "epoch": 55.32, "grad_norm": 7.15625, "learning_rate": 8.765949319219595e-06, "loss": 1.3234, "step": 22128 }, { "epoch": 55.3225, "grad_norm": 7.09375, "learning_rate": 8.765145823389412e-06, "loss": 1.2744, "step": 22129 }, { "epoch": 55.325, "grad_norm": 14.25, "learning_rate": 8.764342335654884e-06, "loss": 1.3702, "step": 22130 }, { "epoch": 55.3275, "grad_norm": 6.3125, "learning_rate": 8.763538856021277e-06, "loss": 1.3296, "step": 22131 }, { "epoch": 55.33, "grad_norm": 8.1875, "learning_rate": 8.762735384493856e-06, "loss": 1.3582, "step": 22132 }, { "epoch": 55.3325, "grad_norm": 7.65625, "learning_rate": 8.761931921077896e-06, "loss": 1.3445, "step": 22133 }, { "epoch": 55.335, "grad_norm": 25.375, "learning_rate": 8.761128465778662e-06, "loss": 1.3957, "step": 22134 }, { "epoch": 55.3375, "grad_norm": 6.65625, "learning_rate": 8.76032501860142e-06, "loss": 1.4429, "step": 22135 }, { "epoch": 55.34, "grad_norm": 7.625, "learning_rate": 8.759521579551434e-06, "loss": 1.4173, "step": 22136 }, { "epoch": 55.3425, "grad_norm": 10.625, "learning_rate": 8.758718148633976e-06, "loss": 1.2649, "step": 22137 }, { "epoch": 55.345, "grad_norm": 7.03125, "learning_rate": 8.757914725854314e-06, "loss": 1.3112, "step": 22138 }, { "epoch": 55.3475, "grad_norm": 6.90625, "learning_rate": 8.757111311217712e-06, "loss": 1.4554, "step": 22139 }, { "epoch": 55.35, "grad_norm": 8.5625, "learning_rate": 8.75630790472944e-06, "loss": 1.3106, "step": 22140 }, { "epoch": 55.3525, "grad_norm": 11.1875, "learning_rate": 8.755504506394763e-06, "loss": 1.3202, "step": 22141 }, { "epoch": 55.355, "grad_norm": 9.5, "learning_rate": 8.754701116218948e-06, "loss": 1.3507, "step": 22142 }, { "epoch": 55.3575, "grad_norm": 8.375, "learning_rate": 8.753897734207262e-06, "loss": 1.4171, "step": 22143 }, { "epoch": 55.36, "grad_norm": 7.46875, "learning_rate": 8.753094360364973e-06, "loss": 1.3802, "step": 22144 }, { "epoch": 55.3625, "grad_norm": 6.5625, "learning_rate": 8.752290994697351e-06, "loss": 1.3905, "step": 22145 }, { "epoch": 55.365, "grad_norm": 6.90625, "learning_rate": 8.751487637209656e-06, "loss": 1.1979, "step": 22146 }, { "epoch": 55.3675, "grad_norm": 7.0, "learning_rate": 8.750684287907156e-06, "loss": 1.3661, "step": 22147 }, { "epoch": 55.37, "grad_norm": 11.25, "learning_rate": 8.749880946795122e-06, "loss": 1.4379, "step": 22148 }, { "epoch": 55.3725, "grad_norm": 7.71875, "learning_rate": 8.74907761387882e-06, "loss": 1.3807, "step": 22149 }, { "epoch": 55.375, "grad_norm": 5.8125, "learning_rate": 8.748274289163514e-06, "loss": 1.3862, "step": 22150 }, { "epoch": 55.3775, "grad_norm": 8.5625, "learning_rate": 8.747470972654471e-06, "loss": 1.3505, "step": 22151 }, { "epoch": 55.38, "grad_norm": 9.25, "learning_rate": 8.746667664356957e-06, "loss": 1.2966, "step": 22152 }, { "epoch": 55.3825, "grad_norm": 9.5, "learning_rate": 8.745864364276245e-06, "loss": 1.3019, "step": 22153 }, { "epoch": 55.385, "grad_norm": 8.0, "learning_rate": 8.74506107241759e-06, "loss": 1.4229, "step": 22154 }, { "epoch": 55.3875, "grad_norm": 8.1875, "learning_rate": 8.744257788786269e-06, "loss": 1.254, "step": 22155 }, { "epoch": 55.39, "grad_norm": 10.0625, "learning_rate": 8.743454513387541e-06, "loss": 1.3331, "step": 22156 }, { "epoch": 55.3925, "grad_norm": 10.75, "learning_rate": 8.742651246226675e-06, "loss": 1.2822, "step": 22157 }, { "epoch": 55.395, "grad_norm": 6.1875, "learning_rate": 8.741847987308935e-06, "loss": 1.3486, "step": 22158 }, { "epoch": 55.3975, "grad_norm": 7.375, "learning_rate": 8.741044736639592e-06, "loss": 1.3093, "step": 22159 }, { "epoch": 55.4, "grad_norm": 6.96875, "learning_rate": 8.740241494223911e-06, "loss": 1.3096, "step": 22160 }, { "epoch": 55.4025, "grad_norm": 9.1875, "learning_rate": 8.739438260067155e-06, "loss": 1.2634, "step": 22161 }, { "epoch": 55.405, "grad_norm": 9.8125, "learning_rate": 8.738635034174588e-06, "loss": 1.2642, "step": 22162 }, { "epoch": 55.4075, "grad_norm": 8.5625, "learning_rate": 8.737831816551482e-06, "loss": 1.328, "step": 22163 }, { "epoch": 55.41, "grad_norm": 8.375, "learning_rate": 8.737028607203101e-06, "loss": 1.2336, "step": 22164 }, { "epoch": 55.4125, "grad_norm": 7.28125, "learning_rate": 8.73622540613471e-06, "loss": 1.3265, "step": 22165 }, { "epoch": 55.415, "grad_norm": 10.875, "learning_rate": 8.735422213351575e-06, "loss": 1.3329, "step": 22166 }, { "epoch": 55.4175, "grad_norm": 8.5625, "learning_rate": 8.734619028858956e-06, "loss": 1.1111, "step": 22167 }, { "epoch": 55.42, "grad_norm": 9.375, "learning_rate": 8.73381585266213e-06, "loss": 1.368, "step": 22168 }, { "epoch": 55.4225, "grad_norm": 6.6875, "learning_rate": 8.733012684766355e-06, "loss": 1.3229, "step": 22169 }, { "epoch": 55.425, "grad_norm": 7.5, "learning_rate": 8.7322095251769e-06, "loss": 1.3332, "step": 22170 }, { "epoch": 55.4275, "grad_norm": 9.1875, "learning_rate": 8.731406373899026e-06, "loss": 1.4161, "step": 22171 }, { "epoch": 55.43, "grad_norm": 8.0, "learning_rate": 8.730603230938e-06, "loss": 1.3007, "step": 22172 }, { "epoch": 55.4325, "grad_norm": 8.3125, "learning_rate": 8.729800096299088e-06, "loss": 1.3213, "step": 22173 }, { "epoch": 55.435, "grad_norm": 6.28125, "learning_rate": 8.728996969987558e-06, "loss": 1.3641, "step": 22174 }, { "epoch": 55.4375, "grad_norm": 8.3125, "learning_rate": 8.728193852008674e-06, "loss": 1.2632, "step": 22175 }, { "epoch": 55.44, "grad_norm": 8.0625, "learning_rate": 8.727390742367698e-06, "loss": 1.2173, "step": 22176 }, { "epoch": 55.4425, "grad_norm": 6.34375, "learning_rate": 8.726587641069896e-06, "loss": 1.2636, "step": 22177 }, { "epoch": 55.445, "grad_norm": 7.46875, "learning_rate": 8.725784548120538e-06, "loss": 1.3941, "step": 22178 }, { "epoch": 55.4475, "grad_norm": 8.75, "learning_rate": 8.724981463524886e-06, "loss": 1.324, "step": 22179 }, { "epoch": 55.45, "grad_norm": 8.0, "learning_rate": 8.724178387288202e-06, "loss": 1.4049, "step": 22180 }, { "epoch": 55.4525, "grad_norm": 9.9375, "learning_rate": 8.723375319415752e-06, "loss": 1.3624, "step": 22181 }, { "epoch": 55.455, "grad_norm": 8.0625, "learning_rate": 8.722572259912804e-06, "loss": 1.3649, "step": 22182 }, { "epoch": 55.4575, "grad_norm": 6.65625, "learning_rate": 8.721769208784623e-06, "loss": 1.3108, "step": 22183 }, { "epoch": 55.46, "grad_norm": 9.375, "learning_rate": 8.720966166036471e-06, "loss": 1.2849, "step": 22184 }, { "epoch": 55.4625, "grad_norm": 5.8125, "learning_rate": 8.720163131673615e-06, "loss": 1.2378, "step": 22185 }, { "epoch": 55.465, "grad_norm": 7.96875, "learning_rate": 8.719360105701311e-06, "loss": 1.3252, "step": 22186 }, { "epoch": 55.4675, "grad_norm": 7.3125, "learning_rate": 8.718557088124837e-06, "loss": 1.2499, "step": 22187 }, { "epoch": 55.47, "grad_norm": 8.875, "learning_rate": 8.717754078949451e-06, "loss": 1.1939, "step": 22188 }, { "epoch": 55.4725, "grad_norm": 9.5, "learning_rate": 8.716951078180422e-06, "loss": 1.2207, "step": 22189 }, { "epoch": 55.475, "grad_norm": 8.5, "learning_rate": 8.716148085823004e-06, "loss": 1.2232, "step": 22190 }, { "epoch": 55.4775, "grad_norm": 5.9375, "learning_rate": 8.715345101882469e-06, "loss": 1.395, "step": 22191 }, { "epoch": 55.48, "grad_norm": 9.6875, "learning_rate": 8.71454212636408e-06, "loss": 1.3303, "step": 22192 }, { "epoch": 55.4825, "grad_norm": 6.96875, "learning_rate": 8.713739159273105e-06, "loss": 1.3224, "step": 22193 }, { "epoch": 55.485, "grad_norm": 8.4375, "learning_rate": 8.712936200614801e-06, "loss": 1.3143, "step": 22194 }, { "epoch": 55.4875, "grad_norm": 6.96875, "learning_rate": 8.712133250394436e-06, "loss": 1.275, "step": 22195 }, { "epoch": 55.49, "grad_norm": 6.9375, "learning_rate": 8.711330308617274e-06, "loss": 1.2602, "step": 22196 }, { "epoch": 55.4925, "grad_norm": 7.40625, "learning_rate": 8.71052737528858e-06, "loss": 1.3542, "step": 22197 }, { "epoch": 55.495, "grad_norm": 9.375, "learning_rate": 8.709724450413617e-06, "loss": 1.2094, "step": 22198 }, { "epoch": 55.4975, "grad_norm": 8.0625, "learning_rate": 8.708921533997647e-06, "loss": 1.2707, "step": 22199 }, { "epoch": 55.5, "grad_norm": 7.625, "learning_rate": 8.708118626045939e-06, "loss": 1.2395, "step": 22200 }, { "epoch": 55.5025, "grad_norm": 10.625, "learning_rate": 8.707315726563747e-06, "loss": 1.4108, "step": 22201 }, { "epoch": 55.505, "grad_norm": 7.46875, "learning_rate": 8.706512835556349e-06, "loss": 1.2072, "step": 22202 }, { "epoch": 55.5075, "grad_norm": 6.46875, "learning_rate": 8.705709953028999e-06, "loss": 1.5572, "step": 22203 }, { "epoch": 55.51, "grad_norm": 8.8125, "learning_rate": 8.704907078986962e-06, "loss": 1.3164, "step": 22204 }, { "epoch": 55.5125, "grad_norm": 7.875, "learning_rate": 8.704104213435502e-06, "loss": 1.2079, "step": 22205 }, { "epoch": 55.515, "grad_norm": 9.0625, "learning_rate": 8.703301356379881e-06, "loss": 1.1543, "step": 22206 }, { "epoch": 55.5175, "grad_norm": 8.875, "learning_rate": 8.702498507825366e-06, "loss": 1.3952, "step": 22207 }, { "epoch": 55.52, "grad_norm": 6.46875, "learning_rate": 8.701695667777221e-06, "loss": 1.1487, "step": 22208 }, { "epoch": 55.5225, "grad_norm": 9.875, "learning_rate": 8.700892836240707e-06, "loss": 1.2535, "step": 22209 }, { "epoch": 55.525, "grad_norm": 8.0, "learning_rate": 8.700090013221086e-06, "loss": 1.3522, "step": 22210 }, { "epoch": 55.5275, "grad_norm": 7.15625, "learning_rate": 8.69928719872362e-06, "loss": 1.4033, "step": 22211 }, { "epoch": 55.53, "grad_norm": 8.625, "learning_rate": 8.69848439275358e-06, "loss": 1.3546, "step": 22212 }, { "epoch": 55.5325, "grad_norm": 7.5, "learning_rate": 8.697681595316222e-06, "loss": 1.3585, "step": 22213 }, { "epoch": 55.535, "grad_norm": 10.125, "learning_rate": 8.696878806416812e-06, "loss": 1.3171, "step": 22214 }, { "epoch": 55.5375, "grad_norm": 10.1875, "learning_rate": 8.696076026060614e-06, "loss": 1.453, "step": 22215 }, { "epoch": 55.54, "grad_norm": 7.6875, "learning_rate": 8.695273254252884e-06, "loss": 1.3604, "step": 22216 }, { "epoch": 55.5425, "grad_norm": 9.3125, "learning_rate": 8.694470490998895e-06, "loss": 1.3679, "step": 22217 }, { "epoch": 55.545, "grad_norm": 8.6875, "learning_rate": 8.693667736303904e-06, "loss": 1.2701, "step": 22218 }, { "epoch": 55.5475, "grad_norm": 6.71875, "learning_rate": 8.692864990173176e-06, "loss": 1.256, "step": 22219 }, { "epoch": 55.55, "grad_norm": 6.6875, "learning_rate": 8.692062252611973e-06, "loss": 1.2628, "step": 22220 }, { "epoch": 55.5525, "grad_norm": 7.53125, "learning_rate": 8.691259523625553e-06, "loss": 1.3072, "step": 22221 }, { "epoch": 55.555, "grad_norm": 6.25, "learning_rate": 8.690456803219188e-06, "loss": 1.3198, "step": 22222 }, { "epoch": 55.5575, "grad_norm": 8.0, "learning_rate": 8.689654091398135e-06, "loss": 1.3193, "step": 22223 }, { "epoch": 55.56, "grad_norm": 7.625, "learning_rate": 8.688851388167658e-06, "loss": 1.2179, "step": 22224 }, { "epoch": 55.5625, "grad_norm": 8.1875, "learning_rate": 8.688048693533017e-06, "loss": 1.2892, "step": 22225 }, { "epoch": 55.565, "grad_norm": 8.875, "learning_rate": 8.687246007499473e-06, "loss": 1.3029, "step": 22226 }, { "epoch": 55.5675, "grad_norm": 10.3125, "learning_rate": 8.686443330072298e-06, "loss": 1.3293, "step": 22227 }, { "epoch": 55.57, "grad_norm": 10.0625, "learning_rate": 8.685640661256746e-06, "loss": 1.4864, "step": 22228 }, { "epoch": 55.5725, "grad_norm": 11.125, "learning_rate": 8.68483800105808e-06, "loss": 1.302, "step": 22229 }, { "epoch": 55.575, "grad_norm": 7.1875, "learning_rate": 8.684035349481567e-06, "loss": 1.3313, "step": 22230 }, { "epoch": 55.5775, "grad_norm": 7.9375, "learning_rate": 8.68323270653246e-06, "loss": 1.3535, "step": 22231 }, { "epoch": 55.58, "grad_norm": 6.0625, "learning_rate": 8.682430072216029e-06, "loss": 1.2809, "step": 22232 }, { "epoch": 55.5825, "grad_norm": 7.25, "learning_rate": 8.681627446537534e-06, "loss": 1.3499, "step": 22233 }, { "epoch": 55.585, "grad_norm": 11.375, "learning_rate": 8.680824829502239e-06, "loss": 1.3057, "step": 22234 }, { "epoch": 55.5875, "grad_norm": 10.4375, "learning_rate": 8.680022221115401e-06, "loss": 1.3897, "step": 22235 }, { "epoch": 55.59, "grad_norm": 10.0625, "learning_rate": 8.679219621382283e-06, "loss": 1.4028, "step": 22236 }, { "epoch": 55.5925, "grad_norm": 9.1875, "learning_rate": 8.67841703030815e-06, "loss": 1.427, "step": 22237 }, { "epoch": 55.595, "grad_norm": 8.5, "learning_rate": 8.677614447898264e-06, "loss": 1.2016, "step": 22238 }, { "epoch": 55.5975, "grad_norm": 7.40625, "learning_rate": 8.676811874157883e-06, "loss": 1.2926, "step": 22239 }, { "epoch": 55.6, "grad_norm": 9.0, "learning_rate": 8.676009309092273e-06, "loss": 1.4569, "step": 22240 }, { "epoch": 55.6025, "grad_norm": 7.375, "learning_rate": 8.675206752706685e-06, "loss": 1.373, "step": 22241 }, { "epoch": 55.605, "grad_norm": 8.0, "learning_rate": 8.674404205006395e-06, "loss": 1.2826, "step": 22242 }, { "epoch": 55.6075, "grad_norm": 8.0625, "learning_rate": 8.673601665996659e-06, "loss": 1.1236, "step": 22243 }, { "epoch": 55.61, "grad_norm": 9.125, "learning_rate": 8.672799135682735e-06, "loss": 1.4529, "step": 22244 }, { "epoch": 55.6125, "grad_norm": 9.5625, "learning_rate": 8.671996614069888e-06, "loss": 1.3502, "step": 22245 }, { "epoch": 55.615, "grad_norm": 9.875, "learning_rate": 8.671194101163375e-06, "loss": 1.6011, "step": 22246 }, { "epoch": 55.6175, "grad_norm": 14.0625, "learning_rate": 8.670391596968463e-06, "loss": 1.3027, "step": 22247 }, { "epoch": 55.62, "grad_norm": 6.96875, "learning_rate": 8.66958910149041e-06, "loss": 1.3679, "step": 22248 }, { "epoch": 55.6225, "grad_norm": 6.75, "learning_rate": 8.668786614734479e-06, "loss": 1.4161, "step": 22249 }, { "epoch": 55.625, "grad_norm": 5.53125, "learning_rate": 8.667984136705927e-06, "loss": 1.2536, "step": 22250 }, { "epoch": 55.6275, "grad_norm": 6.59375, "learning_rate": 8.667181667410018e-06, "loss": 1.3898, "step": 22251 }, { "epoch": 55.63, "grad_norm": 6.96875, "learning_rate": 8.666379206852012e-06, "loss": 1.3665, "step": 22252 }, { "epoch": 55.6325, "grad_norm": 7.6875, "learning_rate": 8.665576755037175e-06, "loss": 1.3296, "step": 22253 }, { "epoch": 55.635, "grad_norm": 8.375, "learning_rate": 8.66477431197076e-06, "loss": 1.2726, "step": 22254 }, { "epoch": 55.6375, "grad_norm": 5.53125, "learning_rate": 8.663971877658034e-06, "loss": 1.5131, "step": 22255 }, { "epoch": 55.64, "grad_norm": 5.6875, "learning_rate": 8.663169452104248e-06, "loss": 1.3371, "step": 22256 }, { "epoch": 55.6425, "grad_norm": 6.5625, "learning_rate": 8.662367035314676e-06, "loss": 1.2849, "step": 22257 }, { "epoch": 55.645, "grad_norm": 5.6875, "learning_rate": 8.66156462729457e-06, "loss": 1.3129, "step": 22258 }, { "epoch": 55.6475, "grad_norm": 8.0, "learning_rate": 8.660762228049196e-06, "loss": 1.3449, "step": 22259 }, { "epoch": 55.65, "grad_norm": 6.1875, "learning_rate": 8.659959837583808e-06, "loss": 1.3151, "step": 22260 }, { "epoch": 55.6525, "grad_norm": 11.5, "learning_rate": 8.659157455903667e-06, "loss": 1.4345, "step": 22261 }, { "epoch": 55.655, "grad_norm": 6.25, "learning_rate": 8.65835508301404e-06, "loss": 1.3639, "step": 22262 }, { "epoch": 55.6575, "grad_norm": 8.0625, "learning_rate": 8.657552718920183e-06, "loss": 1.4784, "step": 22263 }, { "epoch": 55.66, "grad_norm": 8.375, "learning_rate": 8.656750363627356e-06, "loss": 1.4459, "step": 22264 }, { "epoch": 55.6625, "grad_norm": 7.375, "learning_rate": 8.65594801714082e-06, "loss": 1.3958, "step": 22265 }, { "epoch": 55.665, "grad_norm": 8.375, "learning_rate": 8.655145679465833e-06, "loss": 1.3709, "step": 22266 }, { "epoch": 55.6675, "grad_norm": 8.6875, "learning_rate": 8.65434335060766e-06, "loss": 1.3533, "step": 22267 }, { "epoch": 55.67, "grad_norm": 6.625, "learning_rate": 8.653541030571559e-06, "loss": 1.2522, "step": 22268 }, { "epoch": 55.6725, "grad_norm": 6.09375, "learning_rate": 8.652738719362786e-06, "loss": 1.2945, "step": 22269 }, { "epoch": 55.675, "grad_norm": 5.625, "learning_rate": 8.651936416986608e-06, "loss": 1.3361, "step": 22270 }, { "epoch": 55.6775, "grad_norm": 7.125, "learning_rate": 8.651134123448273e-06, "loss": 1.4338, "step": 22271 }, { "epoch": 55.68, "grad_norm": 9.9375, "learning_rate": 8.650331838753057e-06, "loss": 1.4165, "step": 22272 }, { "epoch": 55.6825, "grad_norm": 7.84375, "learning_rate": 8.649529562906208e-06, "loss": 1.3354, "step": 22273 }, { "epoch": 55.685, "grad_norm": 12.5, "learning_rate": 8.648727295912993e-06, "loss": 1.5023, "step": 22274 }, { "epoch": 55.6875, "grad_norm": 9.0, "learning_rate": 8.647925037778663e-06, "loss": 1.346, "step": 22275 }, { "epoch": 55.69, "grad_norm": 6.21875, "learning_rate": 8.647122788508482e-06, "loss": 1.4474, "step": 22276 }, { "epoch": 55.6925, "grad_norm": 6.0, "learning_rate": 8.64632054810771e-06, "loss": 1.2886, "step": 22277 }, { "epoch": 55.695, "grad_norm": 8.4375, "learning_rate": 8.645518316581611e-06, "loss": 1.5408, "step": 22278 }, { "epoch": 55.6975, "grad_norm": 8.875, "learning_rate": 8.644716093935436e-06, "loss": 1.3236, "step": 22279 }, { "epoch": 55.7, "grad_norm": 10.0, "learning_rate": 8.643913880174449e-06, "loss": 1.2696, "step": 22280 }, { "epoch": 55.7025, "grad_norm": 13.5, "learning_rate": 8.643111675303906e-06, "loss": 1.3995, "step": 22281 }, { "epoch": 55.705, "grad_norm": 9.75, "learning_rate": 8.642309479329072e-06, "loss": 1.4134, "step": 22282 }, { "epoch": 55.7075, "grad_norm": 6.59375, "learning_rate": 8.6415072922552e-06, "loss": 1.2782, "step": 22283 }, { "epoch": 55.71, "grad_norm": 7.0625, "learning_rate": 8.640705114087554e-06, "loss": 1.3903, "step": 22284 }, { "epoch": 55.7125, "grad_norm": 6.71875, "learning_rate": 8.639902944831392e-06, "loss": 1.2351, "step": 22285 }, { "epoch": 55.715, "grad_norm": 7.0, "learning_rate": 8.639100784491966e-06, "loss": 1.459, "step": 22286 }, { "epoch": 55.7175, "grad_norm": 9.8125, "learning_rate": 8.638298633074547e-06, "loss": 1.3361, "step": 22287 }, { "epoch": 55.72, "grad_norm": 8.75, "learning_rate": 8.637496490584385e-06, "loss": 1.3089, "step": 22288 }, { "epoch": 55.7225, "grad_norm": 6.53125, "learning_rate": 8.636694357026744e-06, "loss": 1.4222, "step": 22289 }, { "epoch": 55.725, "grad_norm": 5.25, "learning_rate": 8.635892232406877e-06, "loss": 1.1474, "step": 22290 }, { "epoch": 55.7275, "grad_norm": 7.625, "learning_rate": 8.635090116730045e-06, "loss": 1.4322, "step": 22291 }, { "epoch": 55.73, "grad_norm": 16.75, "learning_rate": 8.63428801000151e-06, "loss": 1.4526, "step": 22292 }, { "epoch": 55.7325, "grad_norm": 9.4375, "learning_rate": 8.633485912226529e-06, "loss": 1.4269, "step": 22293 }, { "epoch": 55.735, "grad_norm": 5.625, "learning_rate": 8.632683823410358e-06, "loss": 1.4051, "step": 22294 }, { "epoch": 55.7375, "grad_norm": 7.125, "learning_rate": 8.631881743558256e-06, "loss": 1.4017, "step": 22295 }, { "epoch": 55.74, "grad_norm": 9.5, "learning_rate": 8.631079672675483e-06, "loss": 1.3839, "step": 22296 }, { "epoch": 55.7425, "grad_norm": 7.4375, "learning_rate": 8.630277610767301e-06, "loss": 1.4115, "step": 22297 }, { "epoch": 55.745, "grad_norm": 7.78125, "learning_rate": 8.62947555783896e-06, "loss": 1.4905, "step": 22298 }, { "epoch": 55.7475, "grad_norm": 5.90625, "learning_rate": 8.628673513895723e-06, "loss": 1.304, "step": 22299 }, { "epoch": 55.75, "grad_norm": 7.90625, "learning_rate": 8.62787147894285e-06, "loss": 1.325, "step": 22300 }, { "epoch": 55.7525, "grad_norm": 9.6875, "learning_rate": 8.627069452985593e-06, "loss": 1.2558, "step": 22301 }, { "epoch": 55.755, "grad_norm": 6.71875, "learning_rate": 8.626267436029216e-06, "loss": 1.2522, "step": 22302 }, { "epoch": 55.7575, "grad_norm": 6.59375, "learning_rate": 8.625465428078974e-06, "loss": 1.3079, "step": 22303 }, { "epoch": 55.76, "grad_norm": 11.4375, "learning_rate": 8.624663429140128e-06, "loss": 1.4161, "step": 22304 }, { "epoch": 55.7625, "grad_norm": 6.65625, "learning_rate": 8.62386143921793e-06, "loss": 1.3996, "step": 22305 }, { "epoch": 55.765, "grad_norm": 8.1875, "learning_rate": 8.62305945831764e-06, "loss": 1.1981, "step": 22306 }, { "epoch": 55.7675, "grad_norm": 5.03125, "learning_rate": 8.622257486444521e-06, "loss": 1.4271, "step": 22307 }, { "epoch": 55.77, "grad_norm": 6.28125, "learning_rate": 8.621455523603827e-06, "loss": 1.1391, "step": 22308 }, { "epoch": 55.7725, "grad_norm": 6.0625, "learning_rate": 8.620653569800814e-06, "loss": 1.346, "step": 22309 }, { "epoch": 55.775, "grad_norm": 9.75, "learning_rate": 8.619851625040743e-06, "loss": 1.3895, "step": 22310 }, { "epoch": 55.7775, "grad_norm": 6.1875, "learning_rate": 8.619049689328864e-06, "loss": 1.3831, "step": 22311 }, { "epoch": 55.78, "grad_norm": 8.375, "learning_rate": 8.618247762670445e-06, "loss": 1.2243, "step": 22312 }, { "epoch": 55.7825, "grad_norm": 4.96875, "learning_rate": 8.617445845070736e-06, "loss": 1.2092, "step": 22313 }, { "epoch": 55.785, "grad_norm": 7.3125, "learning_rate": 8.616643936534998e-06, "loss": 1.3068, "step": 22314 }, { "epoch": 55.7875, "grad_norm": 5.625, "learning_rate": 8.615842037068487e-06, "loss": 1.3176, "step": 22315 }, { "epoch": 55.79, "grad_norm": 9.125, "learning_rate": 8.61504014667646e-06, "loss": 1.3733, "step": 22316 }, { "epoch": 55.7925, "grad_norm": 6.5, "learning_rate": 8.614238265364172e-06, "loss": 1.2607, "step": 22317 }, { "epoch": 55.795, "grad_norm": 6.09375, "learning_rate": 8.613436393136885e-06, "loss": 1.4552, "step": 22318 }, { "epoch": 55.7975, "grad_norm": 8.1875, "learning_rate": 8.612634529999854e-06, "loss": 1.3379, "step": 22319 }, { "epoch": 55.8, "grad_norm": 13.3125, "learning_rate": 8.611832675958335e-06, "loss": 1.4239, "step": 22320 }, { "epoch": 55.8025, "grad_norm": 5.28125, "learning_rate": 8.611030831017583e-06, "loss": 1.3549, "step": 22321 }, { "epoch": 55.805, "grad_norm": 7.90625, "learning_rate": 8.61022899518286e-06, "loss": 1.3529, "step": 22322 }, { "epoch": 55.8075, "grad_norm": 10.0625, "learning_rate": 8.609427168459422e-06, "loss": 1.3689, "step": 22323 }, { "epoch": 55.81, "grad_norm": 7.03125, "learning_rate": 8.60862535085252e-06, "loss": 1.2763, "step": 22324 }, { "epoch": 55.8125, "grad_norm": 4.625, "learning_rate": 8.607823542367418e-06, "loss": 1.2396, "step": 22325 }, { "epoch": 55.815, "grad_norm": 9.1875, "learning_rate": 8.607021743009363e-06, "loss": 1.2495, "step": 22326 }, { "epoch": 55.8175, "grad_norm": 9.5625, "learning_rate": 8.606219952783623e-06, "loss": 1.294, "step": 22327 }, { "epoch": 55.82, "grad_norm": 7.1875, "learning_rate": 8.605418171695447e-06, "loss": 1.246, "step": 22328 }, { "epoch": 55.8225, "grad_norm": 6.53125, "learning_rate": 8.604616399750097e-06, "loss": 1.3872, "step": 22329 }, { "epoch": 55.825, "grad_norm": 7.09375, "learning_rate": 8.603814636952824e-06, "loss": 1.2146, "step": 22330 }, { "epoch": 55.8275, "grad_norm": 6.4375, "learning_rate": 8.603012883308883e-06, "loss": 1.4293, "step": 22331 }, { "epoch": 55.83, "grad_norm": 6.96875, "learning_rate": 8.602211138823537e-06, "loss": 1.3202, "step": 22332 }, { "epoch": 55.8325, "grad_norm": 6.28125, "learning_rate": 8.60140940350204e-06, "loss": 1.3985, "step": 22333 }, { "epoch": 55.835, "grad_norm": 10.25, "learning_rate": 8.600607677349647e-06, "loss": 1.353, "step": 22334 }, { "epoch": 55.8375, "grad_norm": 7.21875, "learning_rate": 8.599805960371611e-06, "loss": 1.2846, "step": 22335 }, { "epoch": 55.84, "grad_norm": 7.625, "learning_rate": 8.599004252573191e-06, "loss": 1.2182, "step": 22336 }, { "epoch": 55.8425, "grad_norm": 6.46875, "learning_rate": 8.598202553959646e-06, "loss": 1.2362, "step": 22337 }, { "epoch": 55.845, "grad_norm": 9.3125, "learning_rate": 8.597400864536228e-06, "loss": 1.1582, "step": 22338 }, { "epoch": 55.8475, "grad_norm": 9.75, "learning_rate": 8.596599184308192e-06, "loss": 1.3281, "step": 22339 }, { "epoch": 55.85, "grad_norm": 7.90625, "learning_rate": 8.595797513280799e-06, "loss": 1.2705, "step": 22340 }, { "epoch": 55.8525, "grad_norm": 7.21875, "learning_rate": 8.594995851459295e-06, "loss": 1.2401, "step": 22341 }, { "epoch": 55.855, "grad_norm": 7.875, "learning_rate": 8.594194198848948e-06, "loss": 1.3533, "step": 22342 }, { "epoch": 55.8575, "grad_norm": 10.6875, "learning_rate": 8.593392555455007e-06, "loss": 1.3554, "step": 22343 }, { "epoch": 55.86, "grad_norm": 8.6875, "learning_rate": 8.592590921282728e-06, "loss": 1.2691, "step": 22344 }, { "epoch": 55.8625, "grad_norm": 5.09375, "learning_rate": 8.591789296337365e-06, "loss": 1.3722, "step": 22345 }, { "epoch": 55.865, "grad_norm": 7.4375, "learning_rate": 8.590987680624174e-06, "loss": 1.4607, "step": 22346 }, { "epoch": 55.8675, "grad_norm": 6.6875, "learning_rate": 8.590186074148412e-06, "loss": 1.3361, "step": 22347 }, { "epoch": 55.87, "grad_norm": 7.84375, "learning_rate": 8.589384476915337e-06, "loss": 1.2943, "step": 22348 }, { "epoch": 55.8725, "grad_norm": 8.8125, "learning_rate": 8.588582888930199e-06, "loss": 1.3775, "step": 22349 }, { "epoch": 55.875, "grad_norm": 5.78125, "learning_rate": 8.587781310198253e-06, "loss": 1.3981, "step": 22350 }, { "epoch": 55.8775, "grad_norm": 4.90625, "learning_rate": 8.586979740724756e-06, "loss": 1.2566, "step": 22351 }, { "epoch": 55.88, "grad_norm": 8.3125, "learning_rate": 8.586178180514968e-06, "loss": 1.3384, "step": 22352 }, { "epoch": 55.8825, "grad_norm": 6.875, "learning_rate": 8.585376629574137e-06, "loss": 1.3654, "step": 22353 }, { "epoch": 55.885, "grad_norm": 7.25, "learning_rate": 8.584575087907518e-06, "loss": 1.2733, "step": 22354 }, { "epoch": 55.8875, "grad_norm": 7.8125, "learning_rate": 8.583773555520372e-06, "loss": 1.4012, "step": 22355 }, { "epoch": 55.89, "grad_norm": 6.40625, "learning_rate": 8.582972032417946e-06, "loss": 1.2447, "step": 22356 }, { "epoch": 55.8925, "grad_norm": 7.15625, "learning_rate": 8.582170518605499e-06, "loss": 1.5046, "step": 22357 }, { "epoch": 55.895, "grad_norm": 7.03125, "learning_rate": 8.581369014088286e-06, "loss": 1.563, "step": 22358 }, { "epoch": 55.8975, "grad_norm": 8.25, "learning_rate": 8.580567518871564e-06, "loss": 1.4937, "step": 22359 }, { "epoch": 55.9, "grad_norm": 9.25, "learning_rate": 8.579766032960582e-06, "loss": 1.4846, "step": 22360 }, { "epoch": 55.9025, "grad_norm": 7.84375, "learning_rate": 8.578964556360595e-06, "loss": 1.2347, "step": 22361 }, { "epoch": 55.905, "grad_norm": 6.625, "learning_rate": 8.578163089076862e-06, "loss": 1.2788, "step": 22362 }, { "epoch": 55.9075, "grad_norm": 5.25, "learning_rate": 8.577361631114635e-06, "loss": 1.1345, "step": 22363 }, { "epoch": 55.91, "grad_norm": 7.0, "learning_rate": 8.576560182479168e-06, "loss": 1.2866, "step": 22364 }, { "epoch": 55.9125, "grad_norm": 9.0, "learning_rate": 8.575758743175715e-06, "loss": 1.1909, "step": 22365 }, { "epoch": 55.915, "grad_norm": 7.09375, "learning_rate": 8.57495731320953e-06, "loss": 1.26, "step": 22366 }, { "epoch": 55.9175, "grad_norm": 7.3125, "learning_rate": 8.574155892585871e-06, "loss": 1.1881, "step": 22367 }, { "epoch": 55.92, "grad_norm": 10.6875, "learning_rate": 8.573354481309986e-06, "loss": 1.2511, "step": 22368 }, { "epoch": 55.9225, "grad_norm": 8.5, "learning_rate": 8.572553079387133e-06, "loss": 1.3058, "step": 22369 }, { "epoch": 55.925, "grad_norm": 6.9375, "learning_rate": 8.571751686822567e-06, "loss": 1.3531, "step": 22370 }, { "epoch": 55.9275, "grad_norm": 5.59375, "learning_rate": 8.570950303621533e-06, "loss": 1.1906, "step": 22371 }, { "epoch": 55.93, "grad_norm": 8.5625, "learning_rate": 8.570148929789297e-06, "loss": 1.2539, "step": 22372 }, { "epoch": 55.9325, "grad_norm": 7.1875, "learning_rate": 8.569347565331105e-06, "loss": 1.356, "step": 22373 }, { "epoch": 55.935, "grad_norm": 5.96875, "learning_rate": 8.568546210252216e-06, "loss": 1.4139, "step": 22374 }, { "epoch": 55.9375, "grad_norm": 8.625, "learning_rate": 8.567744864557879e-06, "loss": 1.1867, "step": 22375 }, { "epoch": 55.94, "grad_norm": 8.4375, "learning_rate": 8.566943528253348e-06, "loss": 1.3643, "step": 22376 }, { "epoch": 55.9425, "grad_norm": 7.03125, "learning_rate": 8.56614220134388e-06, "loss": 1.2742, "step": 22377 }, { "epoch": 55.945, "grad_norm": 5.84375, "learning_rate": 8.565340883834727e-06, "loss": 1.256, "step": 22378 }, { "epoch": 55.9475, "grad_norm": 8.3125, "learning_rate": 8.56453957573114e-06, "loss": 1.2996, "step": 22379 }, { "epoch": 55.95, "grad_norm": 6.90625, "learning_rate": 8.563738277038376e-06, "loss": 1.3625, "step": 22380 }, { "epoch": 55.9525, "grad_norm": 8.8125, "learning_rate": 8.56293698776168e-06, "loss": 1.1503, "step": 22381 }, { "epoch": 55.955, "grad_norm": 7.9375, "learning_rate": 8.56213570790632e-06, "loss": 1.2763, "step": 22382 }, { "epoch": 55.9575, "grad_norm": 8.5625, "learning_rate": 8.561334437477537e-06, "loss": 1.5041, "step": 22383 }, { "epoch": 55.96, "grad_norm": 6.8125, "learning_rate": 8.560533176480588e-06, "loss": 1.3931, "step": 22384 }, { "epoch": 55.9625, "grad_norm": 6.125, "learning_rate": 8.559731924920729e-06, "loss": 1.2805, "step": 22385 }, { "epoch": 55.965, "grad_norm": 5.96875, "learning_rate": 8.558930682803204e-06, "loss": 1.1759, "step": 22386 }, { "epoch": 55.9675, "grad_norm": 6.875, "learning_rate": 8.558129450133274e-06, "loss": 1.2797, "step": 22387 }, { "epoch": 55.97, "grad_norm": 7.21875, "learning_rate": 8.557328226916191e-06, "loss": 1.2344, "step": 22388 }, { "epoch": 55.9725, "grad_norm": 7.0, "learning_rate": 8.556527013157207e-06, "loss": 1.2607, "step": 22389 }, { "epoch": 55.975, "grad_norm": 7.5, "learning_rate": 8.555725808861574e-06, "loss": 1.2355, "step": 22390 }, { "epoch": 55.9775, "grad_norm": 7.65625, "learning_rate": 8.554924614034542e-06, "loss": 1.5504, "step": 22391 }, { "epoch": 55.98, "grad_norm": 6.40625, "learning_rate": 8.554123428681367e-06, "loss": 1.3043, "step": 22392 }, { "epoch": 55.9825, "grad_norm": 6.96875, "learning_rate": 8.553322252807303e-06, "loss": 1.4173, "step": 22393 }, { "epoch": 55.985, "grad_norm": 6.96875, "learning_rate": 8.5525210864176e-06, "loss": 1.3262, "step": 22394 }, { "epoch": 55.9875, "grad_norm": 8.5625, "learning_rate": 8.551719929517512e-06, "loss": 1.3921, "step": 22395 }, { "epoch": 55.99, "grad_norm": 9.0625, "learning_rate": 8.550918782112284e-06, "loss": 1.2795, "step": 22396 }, { "epoch": 55.9925, "grad_norm": 8.875, "learning_rate": 8.55011764420718e-06, "loss": 1.1871, "step": 22397 }, { "epoch": 55.995, "grad_norm": 6.5, "learning_rate": 8.549316515807447e-06, "loss": 1.2452, "step": 22398 }, { "epoch": 55.9975, "grad_norm": 10.8125, "learning_rate": 8.548515396918337e-06, "loss": 1.2668, "step": 22399 }, { "epoch": 56.0, "grad_norm": 6.6875, "learning_rate": 8.5477142875451e-06, "loss": 1.627, "step": 22400 }, { "epoch": 56.0025, "grad_norm": 6.0, "learning_rate": 8.546913187692989e-06, "loss": 1.2627, "step": 22401 }, { "epoch": 56.005, "grad_norm": 9.125, "learning_rate": 8.546112097367258e-06, "loss": 1.3867, "step": 22402 }, { "epoch": 56.0075, "grad_norm": 6.8125, "learning_rate": 8.54531101657316e-06, "loss": 1.2963, "step": 22403 }, { "epoch": 56.01, "grad_norm": 6.09375, "learning_rate": 8.544509945315943e-06, "loss": 1.3909, "step": 22404 }, { "epoch": 56.0125, "grad_norm": 6.34375, "learning_rate": 8.543708883600862e-06, "loss": 1.3026, "step": 22405 }, { "epoch": 56.015, "grad_norm": 6.65625, "learning_rate": 8.542907831433162e-06, "loss": 1.2898, "step": 22406 }, { "epoch": 56.0175, "grad_norm": 6.15625, "learning_rate": 8.542106788818106e-06, "loss": 1.2629, "step": 22407 }, { "epoch": 56.02, "grad_norm": 7.28125, "learning_rate": 8.541305755760938e-06, "loss": 1.2148, "step": 22408 }, { "epoch": 56.0225, "grad_norm": 8.1875, "learning_rate": 8.54050473226691e-06, "loss": 1.2889, "step": 22409 }, { "epoch": 56.025, "grad_norm": 9.5, "learning_rate": 8.539703718341277e-06, "loss": 1.318, "step": 22410 }, { "epoch": 56.0275, "grad_norm": 8.125, "learning_rate": 8.538902713989282e-06, "loss": 1.2925, "step": 22411 }, { "epoch": 56.03, "grad_norm": 9.625, "learning_rate": 8.538101719216189e-06, "loss": 1.4677, "step": 22412 }, { "epoch": 56.0325, "grad_norm": 11.3125, "learning_rate": 8.53730073402724e-06, "loss": 1.3681, "step": 22413 }, { "epoch": 56.035, "grad_norm": 6.15625, "learning_rate": 8.53649975842769e-06, "loss": 1.3126, "step": 22414 }, { "epoch": 56.0375, "grad_norm": 11.375, "learning_rate": 8.535698792422788e-06, "loss": 1.3907, "step": 22415 }, { "epoch": 56.04, "grad_norm": 6.96875, "learning_rate": 8.534897836017784e-06, "loss": 1.3966, "step": 22416 }, { "epoch": 56.0425, "grad_norm": 6.59375, "learning_rate": 8.534096889217933e-06, "loss": 1.2822, "step": 22417 }, { "epoch": 56.045, "grad_norm": 7.03125, "learning_rate": 8.533295952028486e-06, "loss": 1.2758, "step": 22418 }, { "epoch": 56.0475, "grad_norm": 9.3125, "learning_rate": 8.53249502445469e-06, "loss": 1.2219, "step": 22419 }, { "epoch": 56.05, "grad_norm": 5.90625, "learning_rate": 8.531694106501796e-06, "loss": 1.2682, "step": 22420 }, { "epoch": 56.0525, "grad_norm": 8.0, "learning_rate": 8.53089319817506e-06, "loss": 1.3519, "step": 22421 }, { "epoch": 56.055, "grad_norm": 7.125, "learning_rate": 8.53009229947973e-06, "loss": 1.359, "step": 22422 }, { "epoch": 56.0575, "grad_norm": 4.5, "learning_rate": 8.529291410421055e-06, "loss": 1.3509, "step": 22423 }, { "epoch": 56.06, "grad_norm": 6.9375, "learning_rate": 8.528490531004285e-06, "loss": 1.2177, "step": 22424 }, { "epoch": 56.0625, "grad_norm": 6.03125, "learning_rate": 8.52768966123467e-06, "loss": 1.3458, "step": 22425 }, { "epoch": 56.065, "grad_norm": 6.1875, "learning_rate": 8.526888801117468e-06, "loss": 1.358, "step": 22426 }, { "epoch": 56.0675, "grad_norm": 8.6875, "learning_rate": 8.526087950657924e-06, "loss": 1.1907, "step": 22427 }, { "epoch": 56.07, "grad_norm": 5.84375, "learning_rate": 8.525287109861285e-06, "loss": 1.27, "step": 22428 }, { "epoch": 56.0725, "grad_norm": 7.25, "learning_rate": 8.524486278732808e-06, "loss": 1.3695, "step": 22429 }, { "epoch": 56.075, "grad_norm": 8.0625, "learning_rate": 8.523685457277736e-06, "loss": 1.3312, "step": 22430 }, { "epoch": 56.0775, "grad_norm": 6.21875, "learning_rate": 8.522884645501324e-06, "loss": 1.1889, "step": 22431 }, { "epoch": 56.08, "grad_norm": 6.75, "learning_rate": 8.522083843408823e-06, "loss": 1.2409, "step": 22432 }, { "epoch": 56.0825, "grad_norm": 7.09375, "learning_rate": 8.521283051005483e-06, "loss": 1.2246, "step": 22433 }, { "epoch": 56.085, "grad_norm": 7.6875, "learning_rate": 8.52048226829655e-06, "loss": 1.461, "step": 22434 }, { "epoch": 56.0875, "grad_norm": 9.75, "learning_rate": 8.519681495287273e-06, "loss": 1.3338, "step": 22435 }, { "epoch": 56.09, "grad_norm": 7.40625, "learning_rate": 8.518880731982908e-06, "loss": 1.2262, "step": 22436 }, { "epoch": 56.0925, "grad_norm": 6.71875, "learning_rate": 8.518079978388702e-06, "loss": 1.2403, "step": 22437 }, { "epoch": 56.095, "grad_norm": 8.375, "learning_rate": 8.517279234509902e-06, "loss": 1.3422, "step": 22438 }, { "epoch": 56.0975, "grad_norm": 8.0, "learning_rate": 8.516478500351763e-06, "loss": 1.3053, "step": 22439 }, { "epoch": 56.1, "grad_norm": 8.125, "learning_rate": 8.515677775919528e-06, "loss": 1.3437, "step": 22440 }, { "epoch": 56.1025, "grad_norm": 6.375, "learning_rate": 8.514877061218453e-06, "loss": 1.272, "step": 22441 }, { "epoch": 56.105, "grad_norm": 11.1875, "learning_rate": 8.514076356253783e-06, "loss": 1.3224, "step": 22442 }, { "epoch": 56.1075, "grad_norm": 8.125, "learning_rate": 8.51327566103077e-06, "loss": 1.3191, "step": 22443 }, { "epoch": 56.11, "grad_norm": 6.9375, "learning_rate": 8.512474975554665e-06, "loss": 1.2388, "step": 22444 }, { "epoch": 56.1125, "grad_norm": 6.8125, "learning_rate": 8.51167429983071e-06, "loss": 1.4466, "step": 22445 }, { "epoch": 56.115, "grad_norm": 8.75, "learning_rate": 8.51087363386416e-06, "loss": 1.3622, "step": 22446 }, { "epoch": 56.1175, "grad_norm": 11.375, "learning_rate": 8.510072977660263e-06, "loss": 1.461, "step": 22447 }, { "epoch": 56.12, "grad_norm": 7.75, "learning_rate": 8.50927233122427e-06, "loss": 1.3585, "step": 22448 }, { "epoch": 56.1225, "grad_norm": 8.9375, "learning_rate": 8.508471694561425e-06, "loss": 1.3827, "step": 22449 }, { "epoch": 56.125, "grad_norm": 8.125, "learning_rate": 8.50767106767698e-06, "loss": 1.3658, "step": 22450 }, { "epoch": 56.1275, "grad_norm": 9.3125, "learning_rate": 8.506870450576184e-06, "loss": 1.3343, "step": 22451 }, { "epoch": 56.13, "grad_norm": 8.9375, "learning_rate": 8.506069843264288e-06, "loss": 1.3765, "step": 22452 }, { "epoch": 56.1325, "grad_norm": 6.78125, "learning_rate": 8.505269245746537e-06, "loss": 1.4719, "step": 22453 }, { "epoch": 56.135, "grad_norm": 7.875, "learning_rate": 8.50446865802818e-06, "loss": 1.3911, "step": 22454 }, { "epoch": 56.1375, "grad_norm": 7.8125, "learning_rate": 8.503668080114464e-06, "loss": 1.3313, "step": 22455 }, { "epoch": 56.14, "grad_norm": 6.28125, "learning_rate": 8.502867512010645e-06, "loss": 1.2994, "step": 22456 }, { "epoch": 56.1425, "grad_norm": 7.4375, "learning_rate": 8.502066953721966e-06, "loss": 1.2904, "step": 22457 }, { "epoch": 56.145, "grad_norm": 5.84375, "learning_rate": 8.501266405253674e-06, "loss": 1.3187, "step": 22458 }, { "epoch": 56.1475, "grad_norm": 5.71875, "learning_rate": 8.500465866611023e-06, "loss": 1.2779, "step": 22459 }, { "epoch": 56.15, "grad_norm": 9.875, "learning_rate": 8.499665337799254e-06, "loss": 1.4046, "step": 22460 }, { "epoch": 56.1525, "grad_norm": 7.4375, "learning_rate": 8.498864818823619e-06, "loss": 1.297, "step": 22461 }, { "epoch": 56.155, "grad_norm": 8.875, "learning_rate": 8.498064309689368e-06, "loss": 1.4577, "step": 22462 }, { "epoch": 56.1575, "grad_norm": 9.125, "learning_rate": 8.497263810401748e-06, "loss": 1.3156, "step": 22463 }, { "epoch": 56.16, "grad_norm": 7.15625, "learning_rate": 8.496463320966004e-06, "loss": 1.2971, "step": 22464 }, { "epoch": 56.1625, "grad_norm": 8.75, "learning_rate": 8.495662841387386e-06, "loss": 1.4011, "step": 22465 }, { "epoch": 56.165, "grad_norm": 7.8125, "learning_rate": 8.494862371671144e-06, "loss": 1.2533, "step": 22466 }, { "epoch": 56.1675, "grad_norm": 7.90625, "learning_rate": 8.494061911822527e-06, "loss": 1.4805, "step": 22467 }, { "epoch": 56.17, "grad_norm": 14.1875, "learning_rate": 8.493261461846775e-06, "loss": 1.2771, "step": 22468 }, { "epoch": 56.1725, "grad_norm": 8.0, "learning_rate": 8.492461021749145e-06, "loss": 1.2261, "step": 22469 }, { "epoch": 56.175, "grad_norm": 6.21875, "learning_rate": 8.491660591534872e-06, "loss": 1.2689, "step": 22470 }, { "epoch": 56.1775, "grad_norm": 8.5, "learning_rate": 8.49086017120922e-06, "loss": 1.4007, "step": 22471 }, { "epoch": 56.18, "grad_norm": 9.25, "learning_rate": 8.490059760777425e-06, "loss": 1.3195, "step": 22472 }, { "epoch": 56.1825, "grad_norm": 5.375, "learning_rate": 8.48925936024474e-06, "loss": 1.3124, "step": 22473 }, { "epoch": 56.185, "grad_norm": 8.6875, "learning_rate": 8.48845896961641e-06, "loss": 1.3526, "step": 22474 }, { "epoch": 56.1875, "grad_norm": 7.84375, "learning_rate": 8.487658588897676e-06, "loss": 1.2832, "step": 22475 }, { "epoch": 56.19, "grad_norm": 6.09375, "learning_rate": 8.486858218093798e-06, "loss": 1.3616, "step": 22476 }, { "epoch": 56.1925, "grad_norm": 9.5625, "learning_rate": 8.486057857210017e-06, "loss": 1.2762, "step": 22477 }, { "epoch": 56.195, "grad_norm": 7.90625, "learning_rate": 8.485257506251578e-06, "loss": 1.4621, "step": 22478 }, { "epoch": 56.1975, "grad_norm": 9.125, "learning_rate": 8.484457165223731e-06, "loss": 1.4562, "step": 22479 }, { "epoch": 56.2, "grad_norm": 8.6875, "learning_rate": 8.48365683413172e-06, "loss": 1.3703, "step": 22480 }, { "epoch": 56.2025, "grad_norm": 9.0, "learning_rate": 8.482856512980798e-06, "loss": 1.3424, "step": 22481 }, { "epoch": 56.205, "grad_norm": 6.8125, "learning_rate": 8.482056201776207e-06, "loss": 1.1573, "step": 22482 }, { "epoch": 56.2075, "grad_norm": 7.15625, "learning_rate": 8.481255900523195e-06, "loss": 1.3119, "step": 22483 }, { "epoch": 56.21, "grad_norm": 5.5, "learning_rate": 8.480455609227009e-06, "loss": 1.4128, "step": 22484 }, { "epoch": 56.2125, "grad_norm": 7.34375, "learning_rate": 8.479655327892888e-06, "loss": 1.3261, "step": 22485 }, { "epoch": 56.215, "grad_norm": 10.125, "learning_rate": 8.478855056526095e-06, "loss": 1.2795, "step": 22486 }, { "epoch": 56.2175, "grad_norm": 10.375, "learning_rate": 8.478054795131864e-06, "loss": 1.3269, "step": 22487 }, { "epoch": 56.22, "grad_norm": 5.96875, "learning_rate": 8.477254543715446e-06, "loss": 1.2316, "step": 22488 }, { "epoch": 56.2225, "grad_norm": 7.34375, "learning_rate": 8.476454302282086e-06, "loss": 1.1974, "step": 22489 }, { "epoch": 56.225, "grad_norm": 9.0, "learning_rate": 8.475654070837027e-06, "loss": 1.1806, "step": 22490 }, { "epoch": 56.2275, "grad_norm": 13.1875, "learning_rate": 8.47485384938552e-06, "loss": 1.2223, "step": 22491 }, { "epoch": 56.23, "grad_norm": 6.84375, "learning_rate": 8.474053637932816e-06, "loss": 1.397, "step": 22492 }, { "epoch": 56.2325, "grad_norm": 6.09375, "learning_rate": 8.47325343648415e-06, "loss": 1.4013, "step": 22493 }, { "epoch": 56.235, "grad_norm": 6.90625, "learning_rate": 8.472453245044774e-06, "loss": 1.2842, "step": 22494 }, { "epoch": 56.2375, "grad_norm": 5.75, "learning_rate": 8.471653063619932e-06, "loss": 1.2536, "step": 22495 }, { "epoch": 56.24, "grad_norm": 6.1875, "learning_rate": 8.470852892214875e-06, "loss": 1.3916, "step": 22496 }, { "epoch": 56.2425, "grad_norm": 7.4375, "learning_rate": 8.470052730834844e-06, "loss": 1.3078, "step": 22497 }, { "epoch": 56.245, "grad_norm": 7.875, "learning_rate": 8.469252579485084e-06, "loss": 1.3076, "step": 22498 }, { "epoch": 56.2475, "grad_norm": 6.5625, "learning_rate": 8.468452438170847e-06, "loss": 1.3316, "step": 22499 }, { "epoch": 56.25, "grad_norm": 10.625, "learning_rate": 8.46765230689737e-06, "loss": 1.3927, "step": 22500 }, { "epoch": 56.2525, "grad_norm": 7.65625, "learning_rate": 8.466852185669906e-06, "loss": 1.3185, "step": 22501 }, { "epoch": 56.255, "grad_norm": 6.625, "learning_rate": 8.466052074493695e-06, "loss": 1.4193, "step": 22502 }, { "epoch": 56.2575, "grad_norm": 8.75, "learning_rate": 8.465251973373989e-06, "loss": 1.2766, "step": 22503 }, { "epoch": 56.26, "grad_norm": 13.0, "learning_rate": 8.464451882316028e-06, "loss": 1.384, "step": 22504 }, { "epoch": 56.2625, "grad_norm": 9.1875, "learning_rate": 8.463651801325056e-06, "loss": 1.3202, "step": 22505 }, { "epoch": 56.265, "grad_norm": 8.4375, "learning_rate": 8.462851730406323e-06, "loss": 1.3089, "step": 22506 }, { "epoch": 56.2675, "grad_norm": 7.9375, "learning_rate": 8.462051669565077e-06, "loss": 1.3511, "step": 22507 }, { "epoch": 56.27, "grad_norm": 8.5, "learning_rate": 8.461251618806554e-06, "loss": 1.2762, "step": 22508 }, { "epoch": 56.2725, "grad_norm": 7.875, "learning_rate": 8.460451578136005e-06, "loss": 1.3615, "step": 22509 }, { "epoch": 56.275, "grad_norm": 7.40625, "learning_rate": 8.459651547558671e-06, "loss": 1.2801, "step": 22510 }, { "epoch": 56.2775, "grad_norm": 5.53125, "learning_rate": 8.458851527079806e-06, "loss": 1.3159, "step": 22511 }, { "epoch": 56.28, "grad_norm": 7.3125, "learning_rate": 8.458051516704644e-06, "loss": 1.3695, "step": 22512 }, { "epoch": 56.2825, "grad_norm": 7.5, "learning_rate": 8.457251516438436e-06, "loss": 1.3209, "step": 22513 }, { "epoch": 56.285, "grad_norm": 9.625, "learning_rate": 8.456451526286428e-06, "loss": 1.2753, "step": 22514 }, { "epoch": 56.2875, "grad_norm": 10.1875, "learning_rate": 8.455651546253857e-06, "loss": 1.26, "step": 22515 }, { "epoch": 56.29, "grad_norm": 8.375, "learning_rate": 8.454851576345975e-06, "loss": 1.2881, "step": 22516 }, { "epoch": 56.2925, "grad_norm": 8.3125, "learning_rate": 8.454051616568024e-06, "loss": 1.4226, "step": 22517 }, { "epoch": 56.295, "grad_norm": 7.1875, "learning_rate": 8.45325166692525e-06, "loss": 1.3235, "step": 22518 }, { "epoch": 56.2975, "grad_norm": 6.46875, "learning_rate": 8.452451727422896e-06, "loss": 1.3787, "step": 22519 }, { "epoch": 56.3, "grad_norm": 7.9375, "learning_rate": 8.451651798066203e-06, "loss": 1.4379, "step": 22520 }, { "epoch": 56.3025, "grad_norm": 6.5, "learning_rate": 8.45085187886042e-06, "loss": 1.3778, "step": 22521 }, { "epoch": 56.305, "grad_norm": 6.4375, "learning_rate": 8.450051969810795e-06, "loss": 1.3511, "step": 22522 }, { "epoch": 56.3075, "grad_norm": 5.15625, "learning_rate": 8.449252070922563e-06, "loss": 1.2732, "step": 22523 }, { "epoch": 56.31, "grad_norm": 6.4375, "learning_rate": 8.448452182200974e-06, "loss": 1.245, "step": 22524 }, { "epoch": 56.3125, "grad_norm": 7.40625, "learning_rate": 8.447652303651267e-06, "loss": 1.4073, "step": 22525 }, { "epoch": 56.315, "grad_norm": 5.375, "learning_rate": 8.446852435278694e-06, "loss": 1.3279, "step": 22526 }, { "epoch": 56.3175, "grad_norm": 8.3125, "learning_rate": 8.446052577088491e-06, "loss": 1.2462, "step": 22527 }, { "epoch": 56.32, "grad_norm": 6.6875, "learning_rate": 8.445252729085907e-06, "loss": 1.1692, "step": 22528 }, { "epoch": 56.3225, "grad_norm": 8.375, "learning_rate": 8.444452891276184e-06, "loss": 1.456, "step": 22529 }, { "epoch": 56.325, "grad_norm": 5.90625, "learning_rate": 8.443653063664563e-06, "loss": 1.343, "step": 22530 }, { "epoch": 56.3275, "grad_norm": 7.03125, "learning_rate": 8.442853246256293e-06, "loss": 1.3335, "step": 22531 }, { "epoch": 56.33, "grad_norm": 6.53125, "learning_rate": 8.442053439056613e-06, "loss": 1.3007, "step": 22532 }, { "epoch": 56.3325, "grad_norm": 5.71875, "learning_rate": 8.44125364207077e-06, "loss": 1.2722, "step": 22533 }, { "epoch": 56.335, "grad_norm": 5.28125, "learning_rate": 8.440453855304007e-06, "loss": 1.2881, "step": 22534 }, { "epoch": 56.3375, "grad_norm": 9.0, "learning_rate": 8.439654078761561e-06, "loss": 1.3644, "step": 22535 }, { "epoch": 56.34, "grad_norm": 5.625, "learning_rate": 8.438854312448683e-06, "loss": 1.2068, "step": 22536 }, { "epoch": 56.3425, "grad_norm": 8.0, "learning_rate": 8.438054556370616e-06, "loss": 1.3712, "step": 22537 }, { "epoch": 56.345, "grad_norm": 6.21875, "learning_rate": 8.437254810532599e-06, "loss": 1.4712, "step": 22538 }, { "epoch": 56.3475, "grad_norm": 10.6875, "learning_rate": 8.43645507493988e-06, "loss": 1.4389, "step": 22539 }, { "epoch": 56.35, "grad_norm": 5.34375, "learning_rate": 8.43565534959769e-06, "loss": 1.286, "step": 22540 }, { "epoch": 56.3525, "grad_norm": 8.8125, "learning_rate": 8.43485563451129e-06, "loss": 1.4476, "step": 22541 }, { "epoch": 56.355, "grad_norm": 6.75, "learning_rate": 8.434055929685909e-06, "loss": 1.4915, "step": 22542 }, { "epoch": 56.3575, "grad_norm": 10.9375, "learning_rate": 8.433256235126799e-06, "loss": 1.2313, "step": 22543 }, { "epoch": 56.36, "grad_norm": 5.75, "learning_rate": 8.432456550839196e-06, "loss": 1.2604, "step": 22544 }, { "epoch": 56.3625, "grad_norm": 6.78125, "learning_rate": 8.431656876828343e-06, "loss": 1.3009, "step": 22545 }, { "epoch": 56.365, "grad_norm": 6.90625, "learning_rate": 8.430857213099486e-06, "loss": 1.316, "step": 22546 }, { "epoch": 56.3675, "grad_norm": 8.1875, "learning_rate": 8.430057559657867e-06, "loss": 1.3654, "step": 22547 }, { "epoch": 56.37, "grad_norm": 7.78125, "learning_rate": 8.429257916508729e-06, "loss": 1.3456, "step": 22548 }, { "epoch": 56.3725, "grad_norm": 8.125, "learning_rate": 8.428458283657311e-06, "loss": 1.2551, "step": 22549 }, { "epoch": 56.375, "grad_norm": 6.0, "learning_rate": 8.427658661108857e-06, "loss": 1.2402, "step": 22550 }, { "epoch": 56.3775, "grad_norm": 6.09375, "learning_rate": 8.42685904886861e-06, "loss": 1.3176, "step": 22551 }, { "epoch": 56.38, "grad_norm": 7.9375, "learning_rate": 8.426059446941817e-06, "loss": 1.631, "step": 22552 }, { "epoch": 56.3825, "grad_norm": 7.6875, "learning_rate": 8.425259855333713e-06, "loss": 1.1541, "step": 22553 }, { "epoch": 56.385, "grad_norm": 5.65625, "learning_rate": 8.424460274049542e-06, "loss": 1.2705, "step": 22554 }, { "epoch": 56.3875, "grad_norm": 6.78125, "learning_rate": 8.42366070309454e-06, "loss": 1.1898, "step": 22555 }, { "epoch": 56.39, "grad_norm": 7.25, "learning_rate": 8.422861142473964e-06, "loss": 1.2601, "step": 22556 }, { "epoch": 56.3925, "grad_norm": 4.875, "learning_rate": 8.422061592193045e-06, "loss": 1.3573, "step": 22557 }, { "epoch": 56.395, "grad_norm": 6.03125, "learning_rate": 8.421262052257026e-06, "loss": 1.3635, "step": 22558 }, { "epoch": 56.3975, "grad_norm": 8.0, "learning_rate": 8.420462522671151e-06, "loss": 1.2055, "step": 22559 }, { "epoch": 56.4, "grad_norm": 8.9375, "learning_rate": 8.419663003440657e-06, "loss": 1.145, "step": 22560 }, { "epoch": 56.4025, "grad_norm": 8.75, "learning_rate": 8.418863494570792e-06, "loss": 1.4004, "step": 22561 }, { "epoch": 56.405, "grad_norm": 5.5625, "learning_rate": 8.418063996066796e-06, "loss": 1.2215, "step": 22562 }, { "epoch": 56.4075, "grad_norm": 6.78125, "learning_rate": 8.417264507933906e-06, "loss": 1.2624, "step": 22563 }, { "epoch": 56.41, "grad_norm": 5.59375, "learning_rate": 8.416465030177369e-06, "loss": 1.3707, "step": 22564 }, { "epoch": 56.4125, "grad_norm": 8.5, "learning_rate": 8.41566556280242e-06, "loss": 1.3527, "step": 22565 }, { "epoch": 56.415, "grad_norm": 14.375, "learning_rate": 8.41486610581431e-06, "loss": 1.3199, "step": 22566 }, { "epoch": 56.4175, "grad_norm": 10.125, "learning_rate": 8.41406665921827e-06, "loss": 1.3008, "step": 22567 }, { "epoch": 56.42, "grad_norm": 9.625, "learning_rate": 8.413267223019548e-06, "loss": 1.3156, "step": 22568 }, { "epoch": 56.4225, "grad_norm": 8.3125, "learning_rate": 8.412467797223382e-06, "loss": 1.3077, "step": 22569 }, { "epoch": 56.425, "grad_norm": 12.375, "learning_rate": 8.411668381835013e-06, "loss": 1.4065, "step": 22570 }, { "epoch": 56.4275, "grad_norm": 7.15625, "learning_rate": 8.410868976859681e-06, "loss": 1.235, "step": 22571 }, { "epoch": 56.43, "grad_norm": 8.375, "learning_rate": 8.41006958230263e-06, "loss": 1.3948, "step": 22572 }, { "epoch": 56.4325, "grad_norm": 7.0, "learning_rate": 8.409270198169101e-06, "loss": 1.3663, "step": 22573 }, { "epoch": 56.435, "grad_norm": 5.78125, "learning_rate": 8.408470824464333e-06, "loss": 1.2626, "step": 22574 }, { "epoch": 56.4375, "grad_norm": 8.0, "learning_rate": 8.407671461193561e-06, "loss": 1.3483, "step": 22575 }, { "epoch": 56.44, "grad_norm": 6.625, "learning_rate": 8.406872108362034e-06, "loss": 1.2679, "step": 22576 }, { "epoch": 56.4425, "grad_norm": 7.90625, "learning_rate": 8.406072765974993e-06, "loss": 1.4946, "step": 22577 }, { "epoch": 56.445, "grad_norm": 6.5625, "learning_rate": 8.405273434037673e-06, "loss": 1.32, "step": 22578 }, { "epoch": 56.4475, "grad_norm": 6.65625, "learning_rate": 8.404474112555316e-06, "loss": 1.3878, "step": 22579 }, { "epoch": 56.45, "grad_norm": 13.6875, "learning_rate": 8.40367480153316e-06, "loss": 1.5638, "step": 22580 }, { "epoch": 56.4525, "grad_norm": 7.03125, "learning_rate": 8.402875500976455e-06, "loss": 1.3649, "step": 22581 }, { "epoch": 56.455, "grad_norm": 8.375, "learning_rate": 8.40207621089043e-06, "loss": 1.2836, "step": 22582 }, { "epoch": 56.4575, "grad_norm": 9.125, "learning_rate": 8.40127693128033e-06, "loss": 1.5052, "step": 22583 }, { "epoch": 56.46, "grad_norm": 6.90625, "learning_rate": 8.400477662151397e-06, "loss": 1.5015, "step": 22584 }, { "epoch": 56.4625, "grad_norm": 6.25, "learning_rate": 8.399678403508863e-06, "loss": 1.4195, "step": 22585 }, { "epoch": 56.465, "grad_norm": 10.0625, "learning_rate": 8.398879155357977e-06, "loss": 1.4498, "step": 22586 }, { "epoch": 56.4675, "grad_norm": 8.875, "learning_rate": 8.398079917703973e-06, "loss": 1.3926, "step": 22587 }, { "epoch": 56.47, "grad_norm": 7.21875, "learning_rate": 8.397280690552096e-06, "loss": 1.383, "step": 22588 }, { "epoch": 56.4725, "grad_norm": 8.375, "learning_rate": 8.396481473907581e-06, "loss": 1.4137, "step": 22589 }, { "epoch": 56.475, "grad_norm": 6.28125, "learning_rate": 8.395682267775667e-06, "loss": 1.2522, "step": 22590 }, { "epoch": 56.4775, "grad_norm": 7.625, "learning_rate": 8.3948830721616e-06, "loss": 1.4078, "step": 22591 }, { "epoch": 56.48, "grad_norm": 5.5, "learning_rate": 8.394083887070614e-06, "loss": 1.3137, "step": 22592 }, { "epoch": 56.4825, "grad_norm": 7.59375, "learning_rate": 8.39328471250795e-06, "loss": 1.2402, "step": 22593 }, { "epoch": 56.485, "grad_norm": 7.3125, "learning_rate": 8.392485548478847e-06, "loss": 1.171, "step": 22594 }, { "epoch": 56.4875, "grad_norm": 8.8125, "learning_rate": 8.391686394988541e-06, "loss": 1.2948, "step": 22595 }, { "epoch": 56.49, "grad_norm": 4.71875, "learning_rate": 8.390887252042279e-06, "loss": 1.2183, "step": 22596 }, { "epoch": 56.4925, "grad_norm": 6.09375, "learning_rate": 8.390088119645294e-06, "loss": 1.2913, "step": 22597 }, { "epoch": 56.495, "grad_norm": 7.0, "learning_rate": 8.389288997802826e-06, "loss": 1.3326, "step": 22598 }, { "epoch": 56.4975, "grad_norm": 6.5625, "learning_rate": 8.38848988652012e-06, "loss": 1.2159, "step": 22599 }, { "epoch": 56.5, "grad_norm": 8.375, "learning_rate": 8.387690785802403e-06, "loss": 1.3764, "step": 22600 }, { "epoch": 56.5025, "grad_norm": 6.21875, "learning_rate": 8.386891695654924e-06, "loss": 1.2928, "step": 22601 }, { "epoch": 56.505, "grad_norm": 5.53125, "learning_rate": 8.386092616082916e-06, "loss": 1.1478, "step": 22602 }, { "epoch": 56.5075, "grad_norm": 6.78125, "learning_rate": 8.385293547091625e-06, "loss": 1.3931, "step": 22603 }, { "epoch": 56.51, "grad_norm": 5.15625, "learning_rate": 8.384494488686282e-06, "loss": 1.2988, "step": 22604 }, { "epoch": 56.5125, "grad_norm": 6.6875, "learning_rate": 8.383695440872127e-06, "loss": 1.287, "step": 22605 }, { "epoch": 56.515, "grad_norm": 7.0, "learning_rate": 8.382896403654401e-06, "loss": 1.4524, "step": 22606 }, { "epoch": 56.5175, "grad_norm": 7.6875, "learning_rate": 8.382097377038343e-06, "loss": 1.2861, "step": 22607 }, { "epoch": 56.52, "grad_norm": 10.0625, "learning_rate": 8.38129836102919e-06, "loss": 1.3619, "step": 22608 }, { "epoch": 56.5225, "grad_norm": 9.0625, "learning_rate": 8.380499355632181e-06, "loss": 1.4268, "step": 22609 }, { "epoch": 56.525, "grad_norm": 9.25, "learning_rate": 8.379700360852548e-06, "loss": 1.2117, "step": 22610 }, { "epoch": 56.5275, "grad_norm": 6.6875, "learning_rate": 8.378901376695538e-06, "loss": 1.2863, "step": 22611 }, { "epoch": 56.53, "grad_norm": 7.6875, "learning_rate": 8.378102403166388e-06, "loss": 1.2864, "step": 22612 }, { "epoch": 56.5325, "grad_norm": 7.09375, "learning_rate": 8.377303440270333e-06, "loss": 1.4084, "step": 22613 }, { "epoch": 56.535, "grad_norm": 7.25, "learning_rate": 8.37650448801261e-06, "loss": 1.1947, "step": 22614 }, { "epoch": 56.5375, "grad_norm": 5.1875, "learning_rate": 8.375705546398456e-06, "loss": 1.2412, "step": 22615 }, { "epoch": 56.54, "grad_norm": 13.875, "learning_rate": 8.374906615433114e-06, "loss": 1.3763, "step": 22616 }, { "epoch": 56.5425, "grad_norm": 6.5625, "learning_rate": 8.374107695121819e-06, "loss": 1.387, "step": 22617 }, { "epoch": 56.545, "grad_norm": 6.375, "learning_rate": 8.373308785469812e-06, "loss": 1.2133, "step": 22618 }, { "epoch": 56.5475, "grad_norm": 7.15625, "learning_rate": 8.372509886482326e-06, "loss": 1.2826, "step": 22619 }, { "epoch": 56.55, "grad_norm": 8.0625, "learning_rate": 8.371710998164595e-06, "loss": 1.3088, "step": 22620 }, { "epoch": 56.5525, "grad_norm": 7.53125, "learning_rate": 8.370912120521866e-06, "loss": 1.3534, "step": 22621 }, { "epoch": 56.555, "grad_norm": 8.9375, "learning_rate": 8.370113253559374e-06, "loss": 1.4517, "step": 22622 }, { "epoch": 56.5575, "grad_norm": 6.46875, "learning_rate": 8.369314397282352e-06, "loss": 1.4075, "step": 22623 }, { "epoch": 56.56, "grad_norm": 5.3125, "learning_rate": 8.36851555169604e-06, "loss": 1.268, "step": 22624 }, { "epoch": 56.5625, "grad_norm": 8.4375, "learning_rate": 8.36771671680567e-06, "loss": 1.3114, "step": 22625 }, { "epoch": 56.565, "grad_norm": 6.59375, "learning_rate": 8.366917892616489e-06, "loss": 1.2227, "step": 22626 }, { "epoch": 56.5675, "grad_norm": 5.8125, "learning_rate": 8.366119079133728e-06, "loss": 1.4177, "step": 22627 }, { "epoch": 56.57, "grad_norm": 8.8125, "learning_rate": 8.365320276362628e-06, "loss": 1.3189, "step": 22628 }, { "epoch": 56.5725, "grad_norm": 5.71875, "learning_rate": 8.36452148430842e-06, "loss": 1.2278, "step": 22629 }, { "epoch": 56.575, "grad_norm": 6.3125, "learning_rate": 8.36372270297634e-06, "loss": 1.2818, "step": 22630 }, { "epoch": 56.5775, "grad_norm": 6.40625, "learning_rate": 8.362923932371632e-06, "loss": 1.2585, "step": 22631 }, { "epoch": 56.58, "grad_norm": 6.8125, "learning_rate": 8.36212517249953e-06, "loss": 1.3391, "step": 22632 }, { "epoch": 56.5825, "grad_norm": 7.78125, "learning_rate": 8.36132642336527e-06, "loss": 1.2545, "step": 22633 }, { "epoch": 56.585, "grad_norm": 7.875, "learning_rate": 8.360527684974087e-06, "loss": 1.4114, "step": 22634 }, { "epoch": 56.5875, "grad_norm": 7.4375, "learning_rate": 8.359728957331215e-06, "loss": 1.4274, "step": 22635 }, { "epoch": 56.59, "grad_norm": 10.9375, "learning_rate": 8.358930240441902e-06, "loss": 1.2513, "step": 22636 }, { "epoch": 56.5925, "grad_norm": 9.375, "learning_rate": 8.358131534311372e-06, "loss": 1.2483, "step": 22637 }, { "epoch": 56.595, "grad_norm": 9.875, "learning_rate": 8.357332838944866e-06, "loss": 1.3115, "step": 22638 }, { "epoch": 56.5975, "grad_norm": 11.5, "learning_rate": 8.356534154347624e-06, "loss": 1.4292, "step": 22639 }, { "epoch": 56.6, "grad_norm": 6.125, "learning_rate": 8.355735480524874e-06, "loss": 1.4682, "step": 22640 }, { "epoch": 56.6025, "grad_norm": 5.71875, "learning_rate": 8.354936817481859e-06, "loss": 1.1407, "step": 22641 }, { "epoch": 56.605, "grad_norm": 7.5, "learning_rate": 8.35413816522381e-06, "loss": 1.3368, "step": 22642 }, { "epoch": 56.6075, "grad_norm": 12.3125, "learning_rate": 8.353339523755967e-06, "loss": 1.4417, "step": 22643 }, { "epoch": 56.61, "grad_norm": 8.6875, "learning_rate": 8.352540893083565e-06, "loss": 1.2281, "step": 22644 }, { "epoch": 56.6125, "grad_norm": 7.40625, "learning_rate": 8.351742273211836e-06, "loss": 1.4554, "step": 22645 }, { "epoch": 56.615, "grad_norm": 6.625, "learning_rate": 8.35094366414602e-06, "loss": 1.369, "step": 22646 }, { "epoch": 56.6175, "grad_norm": 7.0, "learning_rate": 8.350145065891353e-06, "loss": 1.4695, "step": 22647 }, { "epoch": 56.62, "grad_norm": 9.0625, "learning_rate": 8.349346478453068e-06, "loss": 1.4387, "step": 22648 }, { "epoch": 56.6225, "grad_norm": 8.375, "learning_rate": 8.3485479018364e-06, "loss": 1.4725, "step": 22649 }, { "epoch": 56.625, "grad_norm": 9.3125, "learning_rate": 8.347749336046587e-06, "loss": 1.3722, "step": 22650 }, { "epoch": 56.6275, "grad_norm": 6.1875, "learning_rate": 8.346950781088866e-06, "loss": 1.3205, "step": 22651 }, { "epoch": 56.63, "grad_norm": 9.1875, "learning_rate": 8.346152236968465e-06, "loss": 1.3313, "step": 22652 }, { "epoch": 56.6325, "grad_norm": 8.125, "learning_rate": 8.345353703690627e-06, "loss": 1.381, "step": 22653 }, { "epoch": 56.635, "grad_norm": 8.1875, "learning_rate": 8.344555181260583e-06, "loss": 1.2906, "step": 22654 }, { "epoch": 56.6375, "grad_norm": 8.4375, "learning_rate": 8.343756669683566e-06, "loss": 1.3383, "step": 22655 }, { "epoch": 56.64, "grad_norm": 8.1875, "learning_rate": 8.342958168964816e-06, "loss": 1.2679, "step": 22656 }, { "epoch": 56.6425, "grad_norm": 7.375, "learning_rate": 8.342159679109567e-06, "loss": 1.3902, "step": 22657 }, { "epoch": 56.645, "grad_norm": 5.9375, "learning_rate": 8.341361200123055e-06, "loss": 1.3044, "step": 22658 }, { "epoch": 56.6475, "grad_norm": 5.5, "learning_rate": 8.340562732010508e-06, "loss": 1.3552, "step": 22659 }, { "epoch": 56.65, "grad_norm": 6.34375, "learning_rate": 8.339764274777165e-06, "loss": 1.2696, "step": 22660 }, { "epoch": 56.6525, "grad_norm": 7.75, "learning_rate": 8.338965828428263e-06, "loss": 1.4629, "step": 22661 }, { "epoch": 56.655, "grad_norm": 8.1875, "learning_rate": 8.338167392969037e-06, "loss": 1.3927, "step": 22662 }, { "epoch": 56.6575, "grad_norm": 6.4375, "learning_rate": 8.337368968404718e-06, "loss": 1.2086, "step": 22663 }, { "epoch": 56.66, "grad_norm": 6.78125, "learning_rate": 8.336570554740538e-06, "loss": 1.3335, "step": 22664 }, { "epoch": 56.6625, "grad_norm": 7.65625, "learning_rate": 8.335772151981735e-06, "loss": 1.3508, "step": 22665 }, { "epoch": 56.665, "grad_norm": 8.0, "learning_rate": 8.334973760133549e-06, "loss": 1.4279, "step": 22666 }, { "epoch": 56.6675, "grad_norm": 14.0625, "learning_rate": 8.334175379201203e-06, "loss": 1.4934, "step": 22667 }, { "epoch": 56.67, "grad_norm": 9.75, "learning_rate": 8.333377009189939e-06, "loss": 1.326, "step": 22668 }, { "epoch": 56.6725, "grad_norm": 7.125, "learning_rate": 8.332578650104985e-06, "loss": 1.2707, "step": 22669 }, { "epoch": 56.675, "grad_norm": 6.4375, "learning_rate": 8.331780301951585e-06, "loss": 1.2398, "step": 22670 }, { "epoch": 56.6775, "grad_norm": 9.5, "learning_rate": 8.330981964734962e-06, "loss": 1.3882, "step": 22671 }, { "epoch": 56.68, "grad_norm": 7.09375, "learning_rate": 8.330183638460356e-06, "loss": 1.2906, "step": 22672 }, { "epoch": 56.6825, "grad_norm": 8.4375, "learning_rate": 8.329385323133e-06, "loss": 1.2576, "step": 22673 }, { "epoch": 56.685, "grad_norm": 6.28125, "learning_rate": 8.328587018758124e-06, "loss": 1.3278, "step": 22674 }, { "epoch": 56.6875, "grad_norm": 7.78125, "learning_rate": 8.327788725340966e-06, "loss": 1.2992, "step": 22675 }, { "epoch": 56.69, "grad_norm": 6.46875, "learning_rate": 8.32699044288676e-06, "loss": 1.1321, "step": 22676 }, { "epoch": 56.6925, "grad_norm": 7.46875, "learning_rate": 8.326192171400737e-06, "loss": 1.2796, "step": 22677 }, { "epoch": 56.695, "grad_norm": 5.4375, "learning_rate": 8.32539391088813e-06, "loss": 1.3113, "step": 22678 }, { "epoch": 56.6975, "grad_norm": 10.3125, "learning_rate": 8.324595661354172e-06, "loss": 1.3933, "step": 22679 }, { "epoch": 56.7, "grad_norm": 4.78125, "learning_rate": 8.3237974228041e-06, "loss": 1.2924, "step": 22680 }, { "epoch": 56.7025, "grad_norm": 6.9375, "learning_rate": 8.322999195243146e-06, "loss": 1.2073, "step": 22681 }, { "epoch": 56.705, "grad_norm": 8.125, "learning_rate": 8.322200978676541e-06, "loss": 1.5413, "step": 22682 }, { "epoch": 56.7075, "grad_norm": 21.5, "learning_rate": 8.32140277310952e-06, "loss": 1.402, "step": 22683 }, { "epoch": 56.71, "grad_norm": 6.3125, "learning_rate": 8.320604578547312e-06, "loss": 1.1744, "step": 22684 }, { "epoch": 56.7125, "grad_norm": 8.25, "learning_rate": 8.319806394995157e-06, "loss": 1.2558, "step": 22685 }, { "epoch": 56.715, "grad_norm": 9.9375, "learning_rate": 8.319008222458282e-06, "loss": 1.474, "step": 22686 }, { "epoch": 56.7175, "grad_norm": 6.5625, "learning_rate": 8.318210060941922e-06, "loss": 1.3733, "step": 22687 }, { "epoch": 56.72, "grad_norm": 6.875, "learning_rate": 8.317411910451313e-06, "loss": 1.3482, "step": 22688 }, { "epoch": 56.7225, "grad_norm": 5.34375, "learning_rate": 8.316613770991679e-06, "loss": 1.4617, "step": 22689 }, { "epoch": 56.725, "grad_norm": 9.5625, "learning_rate": 8.315815642568259e-06, "loss": 1.301, "step": 22690 }, { "epoch": 56.7275, "grad_norm": 6.59375, "learning_rate": 8.315017525186285e-06, "loss": 1.4005, "step": 22691 }, { "epoch": 56.73, "grad_norm": 6.25, "learning_rate": 8.31421941885099e-06, "loss": 1.1892, "step": 22692 }, { "epoch": 56.7325, "grad_norm": 9.25, "learning_rate": 8.313421323567606e-06, "loss": 1.4207, "step": 22693 }, { "epoch": 56.735, "grad_norm": 7.09375, "learning_rate": 8.312623239341358e-06, "loss": 1.2995, "step": 22694 }, { "epoch": 56.7375, "grad_norm": 7.5625, "learning_rate": 8.311825166177489e-06, "loss": 1.3276, "step": 22695 }, { "epoch": 56.74, "grad_norm": 8.1875, "learning_rate": 8.311027104081228e-06, "loss": 1.3005, "step": 22696 }, { "epoch": 56.7425, "grad_norm": 8.4375, "learning_rate": 8.310229053057805e-06, "loss": 1.2223, "step": 22697 }, { "epoch": 56.745, "grad_norm": 7.90625, "learning_rate": 8.309431013112453e-06, "loss": 1.4197, "step": 22698 }, { "epoch": 56.7475, "grad_norm": 8.4375, "learning_rate": 8.308632984250397e-06, "loss": 1.4437, "step": 22699 }, { "epoch": 56.75, "grad_norm": 7.78125, "learning_rate": 8.307834966476885e-06, "loss": 1.3749, "step": 22700 }, { "epoch": 56.7525, "grad_norm": 8.3125, "learning_rate": 8.307036959797134e-06, "loss": 1.3856, "step": 22701 }, { "epoch": 56.755, "grad_norm": 8.5625, "learning_rate": 8.306238964216384e-06, "loss": 1.3942, "step": 22702 }, { "epoch": 56.7575, "grad_norm": 7.0625, "learning_rate": 8.305440979739862e-06, "loss": 1.267, "step": 22703 }, { "epoch": 56.76, "grad_norm": 5.59375, "learning_rate": 8.304643006372797e-06, "loss": 1.2861, "step": 22704 }, { "epoch": 56.7625, "grad_norm": 7.03125, "learning_rate": 8.303845044120428e-06, "loss": 1.2113, "step": 22705 }, { "epoch": 56.765, "grad_norm": 6.90625, "learning_rate": 8.303047092987987e-06, "loss": 1.3792, "step": 22706 }, { "epoch": 56.7675, "grad_norm": 8.4375, "learning_rate": 8.302249152980697e-06, "loss": 1.336, "step": 22707 }, { "epoch": 56.77, "grad_norm": 7.6875, "learning_rate": 8.301451224103796e-06, "loss": 1.3395, "step": 22708 }, { "epoch": 56.7725, "grad_norm": 10.25, "learning_rate": 8.300653306362508e-06, "loss": 1.1967, "step": 22709 }, { "epoch": 56.775, "grad_norm": 6.5625, "learning_rate": 8.299855399762073e-06, "loss": 1.2221, "step": 22710 }, { "epoch": 56.7775, "grad_norm": 8.875, "learning_rate": 8.299057504307717e-06, "loss": 1.2935, "step": 22711 }, { "epoch": 56.78, "grad_norm": 9.9375, "learning_rate": 8.29825962000467e-06, "loss": 1.4814, "step": 22712 }, { "epoch": 56.7825, "grad_norm": 7.09375, "learning_rate": 8.29746174685817e-06, "loss": 1.3792, "step": 22713 }, { "epoch": 56.785, "grad_norm": 7.65625, "learning_rate": 8.296663884873436e-06, "loss": 1.3304, "step": 22714 }, { "epoch": 56.7875, "grad_norm": 9.0, "learning_rate": 8.29586603405571e-06, "loss": 1.4317, "step": 22715 }, { "epoch": 56.79, "grad_norm": 8.625, "learning_rate": 8.295068194410218e-06, "loss": 1.2804, "step": 22716 }, { "epoch": 56.7925, "grad_norm": 8.8125, "learning_rate": 8.294270365942191e-06, "loss": 1.3021, "step": 22717 }, { "epoch": 56.795, "grad_norm": 7.875, "learning_rate": 8.293472548656858e-06, "loss": 1.4874, "step": 22718 }, { "epoch": 56.7975, "grad_norm": 11.625, "learning_rate": 8.292674742559448e-06, "loss": 1.3774, "step": 22719 }, { "epoch": 56.8, "grad_norm": 7.46875, "learning_rate": 8.291876947655197e-06, "loss": 1.2427, "step": 22720 }, { "epoch": 56.8025, "grad_norm": 6.4375, "learning_rate": 8.291079163949334e-06, "loss": 1.1855, "step": 22721 }, { "epoch": 56.805, "grad_norm": 4.46875, "learning_rate": 8.290281391447085e-06, "loss": 1.2793, "step": 22722 }, { "epoch": 56.8075, "grad_norm": 9.9375, "learning_rate": 8.289483630153682e-06, "loss": 1.4116, "step": 22723 }, { "epoch": 56.81, "grad_norm": 9.25, "learning_rate": 8.288685880074357e-06, "loss": 1.2676, "step": 22724 }, { "epoch": 56.8125, "grad_norm": 7.4375, "learning_rate": 8.287888141214339e-06, "loss": 1.3945, "step": 22725 }, { "epoch": 56.815, "grad_norm": 5.0625, "learning_rate": 8.28709041357886e-06, "loss": 1.3185, "step": 22726 }, { "epoch": 56.8175, "grad_norm": 9.0, "learning_rate": 8.286292697173144e-06, "loss": 1.2412, "step": 22727 }, { "epoch": 56.82, "grad_norm": 5.3125, "learning_rate": 8.285494992002429e-06, "loss": 1.3722, "step": 22728 }, { "epoch": 56.8225, "grad_norm": 7.34375, "learning_rate": 8.284697298071935e-06, "loss": 1.4468, "step": 22729 }, { "epoch": 56.825, "grad_norm": 5.5, "learning_rate": 8.2838996153869e-06, "loss": 1.3875, "step": 22730 }, { "epoch": 56.8275, "grad_norm": 9.6875, "learning_rate": 8.283101943952549e-06, "loss": 1.4569, "step": 22731 }, { "epoch": 56.83, "grad_norm": 5.21875, "learning_rate": 8.282304283774116e-06, "loss": 1.3367, "step": 22732 }, { "epoch": 56.8325, "grad_norm": 8.6875, "learning_rate": 8.281506634856823e-06, "loss": 1.2844, "step": 22733 }, { "epoch": 56.835, "grad_norm": 7.90625, "learning_rate": 8.280708997205904e-06, "loss": 1.2738, "step": 22734 }, { "epoch": 56.8375, "grad_norm": 6.625, "learning_rate": 8.27991137082659e-06, "loss": 1.2319, "step": 22735 }, { "epoch": 56.84, "grad_norm": 8.3125, "learning_rate": 8.27911375572411e-06, "loss": 1.251, "step": 22736 }, { "epoch": 56.8425, "grad_norm": 7.46875, "learning_rate": 8.278316151903689e-06, "loss": 1.4254, "step": 22737 }, { "epoch": 56.845, "grad_norm": 10.4375, "learning_rate": 8.27751855937056e-06, "loss": 1.2959, "step": 22738 }, { "epoch": 56.8475, "grad_norm": 5.875, "learning_rate": 8.276720978129945e-06, "loss": 1.2172, "step": 22739 }, { "epoch": 56.85, "grad_norm": 5.90625, "learning_rate": 8.275923408187086e-06, "loss": 1.22, "step": 22740 }, { "epoch": 56.8525, "grad_norm": 6.625, "learning_rate": 8.2751258495472e-06, "loss": 1.2571, "step": 22741 }, { "epoch": 56.855, "grad_norm": 7.34375, "learning_rate": 8.27432830221552e-06, "loss": 1.2508, "step": 22742 }, { "epoch": 56.8575, "grad_norm": 6.5, "learning_rate": 8.273530766197278e-06, "loss": 1.2602, "step": 22743 }, { "epoch": 56.86, "grad_norm": 9.25, "learning_rate": 8.272733241497694e-06, "loss": 1.2718, "step": 22744 }, { "epoch": 56.8625, "grad_norm": 7.40625, "learning_rate": 8.271935728122002e-06, "loss": 1.3609, "step": 22745 }, { "epoch": 56.865, "grad_norm": 11.25, "learning_rate": 8.271138226075433e-06, "loss": 1.3774, "step": 22746 }, { "epoch": 56.8675, "grad_norm": 10.75, "learning_rate": 8.270340735363214e-06, "loss": 1.2919, "step": 22747 }, { "epoch": 56.87, "grad_norm": 8.625, "learning_rate": 8.26954325599057e-06, "loss": 1.3402, "step": 22748 }, { "epoch": 56.8725, "grad_norm": 9.3125, "learning_rate": 8.268745787962729e-06, "loss": 1.3195, "step": 22749 }, { "epoch": 56.875, "grad_norm": 15.625, "learning_rate": 8.267948331284923e-06, "loss": 1.2537, "step": 22750 }, { "epoch": 56.8775, "grad_norm": 9.125, "learning_rate": 8.267150885962381e-06, "loss": 1.2459, "step": 22751 }, { "epoch": 56.88, "grad_norm": 5.5, "learning_rate": 8.266353452000326e-06, "loss": 1.2675, "step": 22752 }, { "epoch": 56.8825, "grad_norm": 6.0, "learning_rate": 8.265556029403989e-06, "loss": 1.2988, "step": 22753 }, { "epoch": 56.885, "grad_norm": 6.09375, "learning_rate": 8.264758618178593e-06, "loss": 1.2475, "step": 22754 }, { "epoch": 56.8875, "grad_norm": 8.375, "learning_rate": 8.263961218329375e-06, "loss": 1.4466, "step": 22755 }, { "epoch": 56.89, "grad_norm": 6.96875, "learning_rate": 8.263163829861558e-06, "loss": 1.2917, "step": 22756 }, { "epoch": 56.8925, "grad_norm": 7.65625, "learning_rate": 8.262366452780367e-06, "loss": 1.3671, "step": 22757 }, { "epoch": 56.895, "grad_norm": 8.5, "learning_rate": 8.261569087091036e-06, "loss": 1.3144, "step": 22758 }, { "epoch": 56.8975, "grad_norm": 5.59375, "learning_rate": 8.260771732798782e-06, "loss": 1.2416, "step": 22759 }, { "epoch": 56.9, "grad_norm": 9.0625, "learning_rate": 8.259974389908842e-06, "loss": 1.2514, "step": 22760 }, { "epoch": 56.9025, "grad_norm": 8.0625, "learning_rate": 8.259177058426441e-06, "loss": 1.3395, "step": 22761 }, { "epoch": 56.905, "grad_norm": 7.125, "learning_rate": 8.258379738356807e-06, "loss": 1.4626, "step": 22762 }, { "epoch": 56.9075, "grad_norm": 8.0625, "learning_rate": 8.257582429705165e-06, "loss": 1.3308, "step": 22763 }, { "epoch": 56.91, "grad_norm": 7.21875, "learning_rate": 8.256785132476738e-06, "loss": 1.3717, "step": 22764 }, { "epoch": 56.9125, "grad_norm": 8.8125, "learning_rate": 8.255987846676763e-06, "loss": 1.3921, "step": 22765 }, { "epoch": 56.915, "grad_norm": 6.96875, "learning_rate": 8.255190572310463e-06, "loss": 1.4265, "step": 22766 }, { "epoch": 56.9175, "grad_norm": 9.6875, "learning_rate": 8.254393309383062e-06, "loss": 1.3492, "step": 22767 }, { "epoch": 56.92, "grad_norm": 7.875, "learning_rate": 8.25359605789979e-06, "loss": 1.1197, "step": 22768 }, { "epoch": 56.9225, "grad_norm": 9.125, "learning_rate": 8.252798817865865e-06, "loss": 1.2872, "step": 22769 }, { "epoch": 56.925, "grad_norm": 6.84375, "learning_rate": 8.25200158928653e-06, "loss": 1.3569, "step": 22770 }, { "epoch": 56.9275, "grad_norm": 9.375, "learning_rate": 8.251204372166998e-06, "loss": 1.2514, "step": 22771 }, { "epoch": 56.93, "grad_norm": 13.3125, "learning_rate": 8.250407166512502e-06, "loss": 1.2634, "step": 22772 }, { "epoch": 56.9325, "grad_norm": 5.71875, "learning_rate": 8.249609972328268e-06, "loss": 1.4902, "step": 22773 }, { "epoch": 56.935, "grad_norm": 9.25, "learning_rate": 8.248812789619515e-06, "loss": 1.2212, "step": 22774 }, { "epoch": 56.9375, "grad_norm": 8.0, "learning_rate": 8.24801561839148e-06, "loss": 1.1762, "step": 22775 }, { "epoch": 56.94, "grad_norm": 8.375, "learning_rate": 8.247218458649385e-06, "loss": 1.3256, "step": 22776 }, { "epoch": 56.9425, "grad_norm": 9.625, "learning_rate": 8.246421310398454e-06, "loss": 1.487, "step": 22777 }, { "epoch": 56.945, "grad_norm": 10.5, "learning_rate": 8.245624173643914e-06, "loss": 1.4811, "step": 22778 }, { "epoch": 56.9475, "grad_norm": 7.4375, "learning_rate": 8.24482704839099e-06, "loss": 1.3056, "step": 22779 }, { "epoch": 56.95, "grad_norm": 6.40625, "learning_rate": 8.244029934644916e-06, "loss": 1.2522, "step": 22780 }, { "epoch": 56.9525, "grad_norm": 8.0625, "learning_rate": 8.243232832410907e-06, "loss": 1.4965, "step": 22781 }, { "epoch": 56.955, "grad_norm": 6.59375, "learning_rate": 8.242435741694195e-06, "loss": 1.4733, "step": 22782 }, { "epoch": 56.9575, "grad_norm": 7.65625, "learning_rate": 8.241638662500004e-06, "loss": 1.1834, "step": 22783 }, { "epoch": 56.96, "grad_norm": 7.21875, "learning_rate": 8.240841594833554e-06, "loss": 1.2941, "step": 22784 }, { "epoch": 56.9625, "grad_norm": 10.0, "learning_rate": 8.240044538700082e-06, "loss": 1.3032, "step": 22785 }, { "epoch": 56.965, "grad_norm": 6.875, "learning_rate": 8.239247494104807e-06, "loss": 1.3177, "step": 22786 }, { "epoch": 56.9675, "grad_norm": 6.25, "learning_rate": 8.238450461052956e-06, "loss": 1.2154, "step": 22787 }, { "epoch": 56.97, "grad_norm": 6.40625, "learning_rate": 8.23765343954975e-06, "loss": 1.2357, "step": 22788 }, { "epoch": 56.9725, "grad_norm": 6.0625, "learning_rate": 8.236856429600419e-06, "loss": 1.3281, "step": 22789 }, { "epoch": 56.975, "grad_norm": 7.96875, "learning_rate": 8.236059431210186e-06, "loss": 1.2741, "step": 22790 }, { "epoch": 56.9775, "grad_norm": 8.5625, "learning_rate": 8.235262444384279e-06, "loss": 1.4182, "step": 22791 }, { "epoch": 56.98, "grad_norm": 6.4375, "learning_rate": 8.234465469127919e-06, "loss": 1.3926, "step": 22792 }, { "epoch": 56.9825, "grad_norm": 11.9375, "learning_rate": 8.233668505446333e-06, "loss": 1.3267, "step": 22793 }, { "epoch": 56.985, "grad_norm": 6.71875, "learning_rate": 8.232871553344743e-06, "loss": 1.218, "step": 22794 }, { "epoch": 56.9875, "grad_norm": 5.1875, "learning_rate": 8.23207461282838e-06, "loss": 1.1772, "step": 22795 }, { "epoch": 56.99, "grad_norm": 7.5, "learning_rate": 8.231277683902466e-06, "loss": 1.28, "step": 22796 }, { "epoch": 56.9925, "grad_norm": 9.9375, "learning_rate": 8.230480766572223e-06, "loss": 1.4343, "step": 22797 }, { "epoch": 56.995, "grad_norm": 6.40625, "learning_rate": 8.229683860842878e-06, "loss": 1.1811, "step": 22798 }, { "epoch": 56.9975, "grad_norm": 5.71875, "learning_rate": 8.228886966719652e-06, "loss": 1.3729, "step": 22799 }, { "epoch": 57.0, "grad_norm": 6.71875, "learning_rate": 8.228090084207773e-06, "loss": 1.3966, "step": 22800 }, { "epoch": 57.0025, "grad_norm": 7.84375, "learning_rate": 8.227293213312467e-06, "loss": 1.3159, "step": 22801 }, { "epoch": 57.005, "grad_norm": 31.625, "learning_rate": 8.226496354038956e-06, "loss": 1.4085, "step": 22802 }, { "epoch": 57.0075, "grad_norm": 7.34375, "learning_rate": 8.225699506392461e-06, "loss": 1.3418, "step": 22803 }, { "epoch": 57.01, "grad_norm": 5.5625, "learning_rate": 8.224902670378207e-06, "loss": 1.1738, "step": 22804 }, { "epoch": 57.0125, "grad_norm": 8.6875, "learning_rate": 8.224105846001424e-06, "loss": 1.3291, "step": 22805 }, { "epoch": 57.015, "grad_norm": 8.25, "learning_rate": 8.223309033267333e-06, "loss": 1.2602, "step": 22806 }, { "epoch": 57.0175, "grad_norm": 7.625, "learning_rate": 8.222512232181156e-06, "loss": 1.3997, "step": 22807 }, { "epoch": 57.02, "grad_norm": 8.875, "learning_rate": 8.221715442748117e-06, "loss": 1.3682, "step": 22808 }, { "epoch": 57.0225, "grad_norm": 7.09375, "learning_rate": 8.220918664973438e-06, "loss": 1.4863, "step": 22809 }, { "epoch": 57.025, "grad_norm": 6.0625, "learning_rate": 8.220121898862348e-06, "loss": 1.3917, "step": 22810 }, { "epoch": 57.0275, "grad_norm": 5.1875, "learning_rate": 8.219325144420067e-06, "loss": 1.3393, "step": 22811 }, { "epoch": 57.03, "grad_norm": 7.125, "learning_rate": 8.21852840165182e-06, "loss": 1.4138, "step": 22812 }, { "epoch": 57.0325, "grad_norm": 9.5625, "learning_rate": 8.217731670562829e-06, "loss": 1.3576, "step": 22813 }, { "epoch": 57.035, "grad_norm": 7.875, "learning_rate": 8.216934951158315e-06, "loss": 1.3954, "step": 22814 }, { "epoch": 57.0375, "grad_norm": 8.0, "learning_rate": 8.216138243443506e-06, "loss": 1.315, "step": 22815 }, { "epoch": 57.04, "grad_norm": 8.3125, "learning_rate": 8.215341547423624e-06, "loss": 1.3517, "step": 22816 }, { "epoch": 57.0425, "grad_norm": 6.40625, "learning_rate": 8.214544863103894e-06, "loss": 1.4527, "step": 22817 }, { "epoch": 57.045, "grad_norm": 7.03125, "learning_rate": 8.213748190489533e-06, "loss": 1.1951, "step": 22818 }, { "epoch": 57.0475, "grad_norm": 6.625, "learning_rate": 8.212951529585765e-06, "loss": 1.2868, "step": 22819 }, { "epoch": 57.05, "grad_norm": 8.9375, "learning_rate": 8.212154880397817e-06, "loss": 1.4435, "step": 22820 }, { "epoch": 57.0525, "grad_norm": 8.625, "learning_rate": 8.211358242930915e-06, "loss": 1.4151, "step": 22821 }, { "epoch": 57.055, "grad_norm": 6.6875, "learning_rate": 8.210561617190273e-06, "loss": 1.2358, "step": 22822 }, { "epoch": 57.0575, "grad_norm": 9.3125, "learning_rate": 8.20976500318112e-06, "loss": 1.2459, "step": 22823 }, { "epoch": 57.06, "grad_norm": 6.25, "learning_rate": 8.208968400908668e-06, "loss": 1.1685, "step": 22824 }, { "epoch": 57.0625, "grad_norm": 9.125, "learning_rate": 8.208171810378155e-06, "loss": 1.3615, "step": 22825 }, { "epoch": 57.065, "grad_norm": 9.0, "learning_rate": 8.207375231594793e-06, "loss": 1.4606, "step": 22826 }, { "epoch": 57.0675, "grad_norm": 4.75, "learning_rate": 8.206578664563809e-06, "loss": 1.2962, "step": 22827 }, { "epoch": 57.07, "grad_norm": 8.25, "learning_rate": 8.205782109290424e-06, "loss": 1.2799, "step": 22828 }, { "epoch": 57.0725, "grad_norm": 8.1875, "learning_rate": 8.204985565779856e-06, "loss": 1.3328, "step": 22829 }, { "epoch": 57.075, "grad_norm": 8.4375, "learning_rate": 8.204189034037333e-06, "loss": 1.3747, "step": 22830 }, { "epoch": 57.0775, "grad_norm": 6.65625, "learning_rate": 8.203392514068075e-06, "loss": 1.4177, "step": 22831 }, { "epoch": 57.08, "grad_norm": 9.0, "learning_rate": 8.202596005877307e-06, "loss": 1.1159, "step": 22832 }, { "epoch": 57.0825, "grad_norm": 7.6875, "learning_rate": 8.201799509470244e-06, "loss": 1.3727, "step": 22833 }, { "epoch": 57.085, "grad_norm": 7.21875, "learning_rate": 8.201003024852107e-06, "loss": 1.2652, "step": 22834 }, { "epoch": 57.0875, "grad_norm": 8.5, "learning_rate": 8.200206552028129e-06, "loss": 1.3732, "step": 22835 }, { "epoch": 57.09, "grad_norm": 6.625, "learning_rate": 8.199410091003523e-06, "loss": 1.4201, "step": 22836 }, { "epoch": 57.0925, "grad_norm": 5.78125, "learning_rate": 8.198613641783514e-06, "loss": 1.4157, "step": 22837 }, { "epoch": 57.095, "grad_norm": 6.03125, "learning_rate": 8.197817204373322e-06, "loss": 1.3549, "step": 22838 }, { "epoch": 57.0975, "grad_norm": 7.5625, "learning_rate": 8.197020778778161e-06, "loss": 1.3274, "step": 22839 }, { "epoch": 57.1, "grad_norm": 7.0, "learning_rate": 8.196224365003267e-06, "loss": 1.2889, "step": 22840 }, { "epoch": 57.1025, "grad_norm": 7.78125, "learning_rate": 8.195427963053852e-06, "loss": 1.2837, "step": 22841 }, { "epoch": 57.105, "grad_norm": 10.0, "learning_rate": 8.194631572935141e-06, "loss": 1.2692, "step": 22842 }, { "epoch": 57.1075, "grad_norm": 7.625, "learning_rate": 8.19383519465235e-06, "loss": 1.3519, "step": 22843 }, { "epoch": 57.11, "grad_norm": 5.71875, "learning_rate": 8.193038828210703e-06, "loss": 1.3238, "step": 22844 }, { "epoch": 57.1125, "grad_norm": 7.21875, "learning_rate": 8.192242473615422e-06, "loss": 1.2751, "step": 22845 }, { "epoch": 57.115, "grad_norm": 6.5625, "learning_rate": 8.19144613087173e-06, "loss": 1.4735, "step": 22846 }, { "epoch": 57.1175, "grad_norm": 10.125, "learning_rate": 8.190649799984841e-06, "loss": 1.2803, "step": 22847 }, { "epoch": 57.12, "grad_norm": 6.625, "learning_rate": 8.189853480959982e-06, "loss": 1.3241, "step": 22848 }, { "epoch": 57.1225, "grad_norm": 15.3125, "learning_rate": 8.189057173802367e-06, "loss": 1.3815, "step": 22849 }, { "epoch": 57.125, "grad_norm": 6.0, "learning_rate": 8.188260878517224e-06, "loss": 1.2564, "step": 22850 }, { "epoch": 57.1275, "grad_norm": 4.84375, "learning_rate": 8.18746459510977e-06, "loss": 1.3468, "step": 22851 }, { "epoch": 57.13, "grad_norm": 8.8125, "learning_rate": 8.186668323585226e-06, "loss": 1.362, "step": 22852 }, { "epoch": 57.1325, "grad_norm": 10.375, "learning_rate": 8.185872063948813e-06, "loss": 1.3062, "step": 22853 }, { "epoch": 57.135, "grad_norm": 10.125, "learning_rate": 8.185075816205744e-06, "loss": 1.1694, "step": 22854 }, { "epoch": 57.1375, "grad_norm": 10.125, "learning_rate": 8.184279580361253e-06, "loss": 1.4124, "step": 22855 }, { "epoch": 57.14, "grad_norm": 7.625, "learning_rate": 8.183483356420547e-06, "loss": 1.38, "step": 22856 }, { "epoch": 57.1425, "grad_norm": 8.75, "learning_rate": 8.182687144388856e-06, "loss": 1.2824, "step": 22857 }, { "epoch": 57.145, "grad_norm": 9.9375, "learning_rate": 8.181890944271394e-06, "loss": 1.2661, "step": 22858 }, { "epoch": 57.1475, "grad_norm": 7.0, "learning_rate": 8.181094756073378e-06, "loss": 1.3372, "step": 22859 }, { "epoch": 57.15, "grad_norm": 6.4375, "learning_rate": 8.180298579800034e-06, "loss": 1.3322, "step": 22860 }, { "epoch": 57.1525, "grad_norm": 7.0625, "learning_rate": 8.179502415456583e-06, "loss": 1.1963, "step": 22861 }, { "epoch": 57.155, "grad_norm": 8.125, "learning_rate": 8.17870626304824e-06, "loss": 1.3554, "step": 22862 }, { "epoch": 57.1575, "grad_norm": 7.5, "learning_rate": 8.177910122580224e-06, "loss": 1.5496, "step": 22863 }, { "epoch": 57.16, "grad_norm": 6.5, "learning_rate": 8.177113994057756e-06, "loss": 1.3967, "step": 22864 }, { "epoch": 57.1625, "grad_norm": 7.15625, "learning_rate": 8.176317877486057e-06, "loss": 1.3924, "step": 22865 }, { "epoch": 57.165, "grad_norm": 10.75, "learning_rate": 8.175521772870345e-06, "loss": 1.3773, "step": 22866 }, { "epoch": 57.1675, "grad_norm": 6.0625, "learning_rate": 8.17472568021584e-06, "loss": 1.2299, "step": 22867 }, { "epoch": 57.17, "grad_norm": 7.0625, "learning_rate": 8.173929599527761e-06, "loss": 1.2539, "step": 22868 }, { "epoch": 57.1725, "grad_norm": 7.375, "learning_rate": 8.173133530811323e-06, "loss": 1.2537, "step": 22869 }, { "epoch": 57.175, "grad_norm": 6.3125, "learning_rate": 8.172337474071751e-06, "loss": 1.4688, "step": 22870 }, { "epoch": 57.1775, "grad_norm": 8.75, "learning_rate": 8.17154142931426e-06, "loss": 1.2638, "step": 22871 }, { "epoch": 57.18, "grad_norm": 7.34375, "learning_rate": 8.170745396544072e-06, "loss": 1.2611, "step": 22872 }, { "epoch": 57.1825, "grad_norm": 11.3125, "learning_rate": 8.169949375766404e-06, "loss": 1.3567, "step": 22873 }, { "epoch": 57.185, "grad_norm": 8.4375, "learning_rate": 8.169153366986472e-06, "loss": 1.3445, "step": 22874 }, { "epoch": 57.1875, "grad_norm": 7.28125, "learning_rate": 8.168357370209499e-06, "loss": 1.2782, "step": 22875 }, { "epoch": 57.19, "grad_norm": 6.53125, "learning_rate": 8.167561385440704e-06, "loss": 1.3639, "step": 22876 }, { "epoch": 57.1925, "grad_norm": 9.8125, "learning_rate": 8.1667654126853e-06, "loss": 1.3222, "step": 22877 }, { "epoch": 57.195, "grad_norm": 5.8125, "learning_rate": 8.16596945194851e-06, "loss": 1.1923, "step": 22878 }, { "epoch": 57.1975, "grad_norm": 6.59375, "learning_rate": 8.16517350323555e-06, "loss": 1.2555, "step": 22879 }, { "epoch": 57.2, "grad_norm": 12.0, "learning_rate": 8.16437756655164e-06, "loss": 1.3374, "step": 22880 }, { "epoch": 57.2025, "grad_norm": 9.125, "learning_rate": 8.163581641901997e-06, "loss": 1.2129, "step": 22881 }, { "epoch": 57.205, "grad_norm": 8.375, "learning_rate": 8.162785729291839e-06, "loss": 1.3734, "step": 22882 }, { "epoch": 57.2075, "grad_norm": 14.8125, "learning_rate": 8.161989828726387e-06, "loss": 1.4558, "step": 22883 }, { "epoch": 57.21, "grad_norm": 8.3125, "learning_rate": 8.16119394021085e-06, "loss": 1.4242, "step": 22884 }, { "epoch": 57.2125, "grad_norm": 7.90625, "learning_rate": 8.160398063750456e-06, "loss": 1.1943, "step": 22885 }, { "epoch": 57.215, "grad_norm": 8.0, "learning_rate": 8.159602199350416e-06, "loss": 1.3523, "step": 22886 }, { "epoch": 57.2175, "grad_norm": 5.75, "learning_rate": 8.158806347015956e-06, "loss": 1.3351, "step": 22887 }, { "epoch": 57.22, "grad_norm": 9.125, "learning_rate": 8.158010506752283e-06, "loss": 1.2137, "step": 22888 }, { "epoch": 57.2225, "grad_norm": 6.625, "learning_rate": 8.15721467856462e-06, "loss": 1.4805, "step": 22889 }, { "epoch": 57.225, "grad_norm": 7.34375, "learning_rate": 8.156418862458184e-06, "loss": 1.3919, "step": 22890 }, { "epoch": 57.2275, "grad_norm": 7.0, "learning_rate": 8.155623058438194e-06, "loss": 1.2517, "step": 22891 }, { "epoch": 57.23, "grad_norm": 6.75, "learning_rate": 8.154827266509865e-06, "loss": 1.2127, "step": 22892 }, { "epoch": 57.2325, "grad_norm": 6.03125, "learning_rate": 8.154031486678416e-06, "loss": 1.4072, "step": 22893 }, { "epoch": 57.235, "grad_norm": 7.9375, "learning_rate": 8.153235718949056e-06, "loss": 1.3762, "step": 22894 }, { "epoch": 57.2375, "grad_norm": 9.125, "learning_rate": 8.152439963327015e-06, "loss": 1.3788, "step": 22895 }, { "epoch": 57.24, "grad_norm": 5.375, "learning_rate": 8.1516442198175e-06, "loss": 1.2666, "step": 22896 }, { "epoch": 57.2425, "grad_norm": 9.875, "learning_rate": 8.150848488425735e-06, "loss": 1.4232, "step": 22897 }, { "epoch": 57.245, "grad_norm": 6.78125, "learning_rate": 8.150052769156933e-06, "loss": 1.3209, "step": 22898 }, { "epoch": 57.2475, "grad_norm": 6.65625, "learning_rate": 8.149257062016307e-06, "loss": 1.4082, "step": 22899 }, { "epoch": 57.25, "grad_norm": 9.4375, "learning_rate": 8.148461367009081e-06, "loss": 1.4126, "step": 22900 }, { "epoch": 57.2525, "grad_norm": 9.75, "learning_rate": 8.147665684140467e-06, "loss": 1.1808, "step": 22901 }, { "epoch": 57.255, "grad_norm": 7.5625, "learning_rate": 8.146870013415685e-06, "loss": 1.2555, "step": 22902 }, { "epoch": 57.2575, "grad_norm": 8.0625, "learning_rate": 8.146074354839947e-06, "loss": 1.353, "step": 22903 }, { "epoch": 57.26, "grad_norm": 8.25, "learning_rate": 8.14527870841847e-06, "loss": 1.265, "step": 22904 }, { "epoch": 57.2625, "grad_norm": 7.8125, "learning_rate": 8.144483074156472e-06, "loss": 1.2354, "step": 22905 }, { "epoch": 57.265, "grad_norm": 6.8125, "learning_rate": 8.143687452059172e-06, "loss": 1.2619, "step": 22906 }, { "epoch": 57.2675, "grad_norm": 9.625, "learning_rate": 8.14289184213178e-06, "loss": 1.3134, "step": 22907 }, { "epoch": 57.27, "grad_norm": 7.4375, "learning_rate": 8.142096244379514e-06, "loss": 1.341, "step": 22908 }, { "epoch": 57.2725, "grad_norm": 7.28125, "learning_rate": 8.14130065880759e-06, "loss": 1.1823, "step": 22909 }, { "epoch": 57.275, "grad_norm": 6.28125, "learning_rate": 8.14050508542123e-06, "loss": 1.3418, "step": 22910 }, { "epoch": 57.2775, "grad_norm": 5.34375, "learning_rate": 8.13970952422564e-06, "loss": 1.2418, "step": 22911 }, { "epoch": 57.28, "grad_norm": 9.375, "learning_rate": 8.138913975226044e-06, "loss": 1.2582, "step": 22912 }, { "epoch": 57.2825, "grad_norm": 11.6875, "learning_rate": 8.138118438427645e-06, "loss": 1.3789, "step": 22913 }, { "epoch": 57.285, "grad_norm": 8.0625, "learning_rate": 8.137322913835674e-06, "loss": 1.3688, "step": 22914 }, { "epoch": 57.2875, "grad_norm": 6.4375, "learning_rate": 8.136527401455338e-06, "loss": 1.219, "step": 22915 }, { "epoch": 57.29, "grad_norm": 10.5625, "learning_rate": 8.135731901291854e-06, "loss": 1.3829, "step": 22916 }, { "epoch": 57.2925, "grad_norm": 11.5625, "learning_rate": 8.134936413350438e-06, "loss": 1.305, "step": 22917 }, { "epoch": 57.295, "grad_norm": 7.5625, "learning_rate": 8.1341409376363e-06, "loss": 1.3354, "step": 22918 }, { "epoch": 57.2975, "grad_norm": 6.25, "learning_rate": 8.133345474154662e-06, "loss": 1.5333, "step": 22919 }, { "epoch": 57.3, "grad_norm": 12.0, "learning_rate": 8.132550022910737e-06, "loss": 1.536, "step": 22920 }, { "epoch": 57.3025, "grad_norm": 6.375, "learning_rate": 8.13175458390974e-06, "loss": 1.3412, "step": 22921 }, { "epoch": 57.305, "grad_norm": 6.03125, "learning_rate": 8.130959157156884e-06, "loss": 1.3413, "step": 22922 }, { "epoch": 57.3075, "grad_norm": 6.6875, "learning_rate": 8.130163742657383e-06, "loss": 1.3633, "step": 22923 }, { "epoch": 57.31, "grad_norm": 7.9375, "learning_rate": 8.129368340416454e-06, "loss": 1.1745, "step": 22924 }, { "epoch": 57.3125, "grad_norm": 8.5, "learning_rate": 8.128572950439314e-06, "loss": 1.2645, "step": 22925 }, { "epoch": 57.315, "grad_norm": 8.0625, "learning_rate": 8.127777572731174e-06, "loss": 1.3604, "step": 22926 }, { "epoch": 57.3175, "grad_norm": 15.8125, "learning_rate": 8.12698220729725e-06, "loss": 1.3695, "step": 22927 }, { "epoch": 57.32, "grad_norm": 5.09375, "learning_rate": 8.126186854142752e-06, "loss": 1.3133, "step": 22928 }, { "epoch": 57.3225, "grad_norm": 7.125, "learning_rate": 8.125391513272903e-06, "loss": 1.3384, "step": 22929 }, { "epoch": 57.325, "grad_norm": 22.625, "learning_rate": 8.12459618469291e-06, "loss": 1.3404, "step": 22930 }, { "epoch": 57.3275, "grad_norm": 12.625, "learning_rate": 8.123800868407991e-06, "loss": 1.3458, "step": 22931 }, { "epoch": 57.33, "grad_norm": 8.375, "learning_rate": 8.123005564423356e-06, "loss": 1.3208, "step": 22932 }, { "epoch": 57.3325, "grad_norm": 5.5, "learning_rate": 8.122210272744222e-06, "loss": 1.3141, "step": 22933 }, { "epoch": 57.335, "grad_norm": 8.5625, "learning_rate": 8.121414993375802e-06, "loss": 1.4147, "step": 22934 }, { "epoch": 57.3375, "grad_norm": 6.46875, "learning_rate": 8.12061972632331e-06, "loss": 1.2527, "step": 22935 }, { "epoch": 57.34, "grad_norm": 6.5625, "learning_rate": 8.119824471591962e-06, "loss": 1.3522, "step": 22936 }, { "epoch": 57.3425, "grad_norm": 6.4375, "learning_rate": 8.119029229186967e-06, "loss": 1.3839, "step": 22937 }, { "epoch": 57.345, "grad_norm": 6.75, "learning_rate": 8.11823399911354e-06, "loss": 1.3777, "step": 22938 }, { "epoch": 57.3475, "grad_norm": 8.375, "learning_rate": 8.117438781376901e-06, "loss": 1.5028, "step": 22939 }, { "epoch": 57.35, "grad_norm": 8.0625, "learning_rate": 8.116643575982254e-06, "loss": 1.2932, "step": 22940 }, { "epoch": 57.3525, "grad_norm": 9.1875, "learning_rate": 8.115848382934816e-06, "loss": 1.3921, "step": 22941 }, { "epoch": 57.355, "grad_norm": 9.8125, "learning_rate": 8.115053202239805e-06, "loss": 1.4415, "step": 22942 }, { "epoch": 57.3575, "grad_norm": 8.375, "learning_rate": 8.114258033902424e-06, "loss": 1.2796, "step": 22943 }, { "epoch": 57.36, "grad_norm": 8.8125, "learning_rate": 8.113462877927893e-06, "loss": 1.4338, "step": 22944 }, { "epoch": 57.3625, "grad_norm": 9.8125, "learning_rate": 8.112667734321425e-06, "loss": 1.4074, "step": 22945 }, { "epoch": 57.365, "grad_norm": 7.5625, "learning_rate": 8.111872603088234e-06, "loss": 1.4052, "step": 22946 }, { "epoch": 57.3675, "grad_norm": 5.8125, "learning_rate": 8.11107748423353e-06, "loss": 1.3408, "step": 22947 }, { "epoch": 57.37, "grad_norm": 8.5625, "learning_rate": 8.110282377762522e-06, "loss": 1.2743, "step": 22948 }, { "epoch": 57.3725, "grad_norm": 6.125, "learning_rate": 8.10948728368043e-06, "loss": 1.2441, "step": 22949 }, { "epoch": 57.375, "grad_norm": 7.65625, "learning_rate": 8.108692201992466e-06, "loss": 1.3552, "step": 22950 }, { "epoch": 57.3775, "grad_norm": 9.8125, "learning_rate": 8.107897132703836e-06, "loss": 1.3533, "step": 22951 }, { "epoch": 57.38, "grad_norm": 9.125, "learning_rate": 8.10710207581976e-06, "loss": 1.2437, "step": 22952 }, { "epoch": 57.3825, "grad_norm": 5.9375, "learning_rate": 8.106307031345443e-06, "loss": 1.3347, "step": 22953 }, { "epoch": 57.385, "grad_norm": 7.03125, "learning_rate": 8.105511999286107e-06, "loss": 1.4048, "step": 22954 }, { "epoch": 57.3875, "grad_norm": 10.875, "learning_rate": 8.104716979646956e-06, "loss": 1.1593, "step": 22955 }, { "epoch": 57.39, "grad_norm": 9.1875, "learning_rate": 8.103921972433205e-06, "loss": 1.3663, "step": 22956 }, { "epoch": 57.3925, "grad_norm": 7.625, "learning_rate": 8.103126977650066e-06, "loss": 1.3543, "step": 22957 }, { "epoch": 57.395, "grad_norm": 9.0, "learning_rate": 8.102331995302746e-06, "loss": 1.3534, "step": 22958 }, { "epoch": 57.3975, "grad_norm": 8.9375, "learning_rate": 8.101537025396465e-06, "loss": 1.2632, "step": 22959 }, { "epoch": 57.4, "grad_norm": 7.59375, "learning_rate": 8.100742067936432e-06, "loss": 1.3158, "step": 22960 }, { "epoch": 57.4025, "grad_norm": 7.1875, "learning_rate": 8.099947122927859e-06, "loss": 1.3006, "step": 22961 }, { "epoch": 57.405, "grad_norm": 8.5, "learning_rate": 8.099152190375955e-06, "loss": 1.3719, "step": 22962 }, { "epoch": 57.4075, "grad_norm": 6.625, "learning_rate": 8.09835727028593e-06, "loss": 1.3012, "step": 22963 }, { "epoch": 57.41, "grad_norm": 8.0625, "learning_rate": 8.097562362663002e-06, "loss": 1.4025, "step": 22964 }, { "epoch": 57.4125, "grad_norm": 9.625, "learning_rate": 8.096767467512381e-06, "loss": 1.3687, "step": 22965 }, { "epoch": 57.415, "grad_norm": 7.6875, "learning_rate": 8.095972584839274e-06, "loss": 1.3483, "step": 22966 }, { "epoch": 57.4175, "grad_norm": 6.15625, "learning_rate": 8.095177714648894e-06, "loss": 1.3215, "step": 22967 }, { "epoch": 57.42, "grad_norm": 7.0625, "learning_rate": 8.094382856946452e-06, "loss": 1.3394, "step": 22968 }, { "epoch": 57.4225, "grad_norm": 5.75, "learning_rate": 8.093588011737163e-06, "loss": 1.3722, "step": 22969 }, { "epoch": 57.425, "grad_norm": 6.71875, "learning_rate": 8.092793179026233e-06, "loss": 1.2552, "step": 22970 }, { "epoch": 57.4275, "grad_norm": 9.1875, "learning_rate": 8.091998358818875e-06, "loss": 1.2834, "step": 22971 }, { "epoch": 57.43, "grad_norm": 7.6875, "learning_rate": 8.0912035511203e-06, "loss": 1.1623, "step": 22972 }, { "epoch": 57.4325, "grad_norm": 14.4375, "learning_rate": 8.090408755935714e-06, "loss": 1.4088, "step": 22973 }, { "epoch": 57.435, "grad_norm": 6.6875, "learning_rate": 8.089613973270335e-06, "loss": 1.2925, "step": 22974 }, { "epoch": 57.4375, "grad_norm": 8.5625, "learning_rate": 8.08881920312937e-06, "loss": 1.3208, "step": 22975 }, { "epoch": 57.44, "grad_norm": 8.5, "learning_rate": 8.088024445518033e-06, "loss": 1.4413, "step": 22976 }, { "epoch": 57.4425, "grad_norm": 10.1875, "learning_rate": 8.087229700441526e-06, "loss": 1.3072, "step": 22977 }, { "epoch": 57.445, "grad_norm": 8.0, "learning_rate": 8.086434967905066e-06, "loss": 1.4499, "step": 22978 }, { "epoch": 57.4475, "grad_norm": 11.375, "learning_rate": 8.085640247913862e-06, "loss": 1.2602, "step": 22979 }, { "epoch": 57.45, "grad_norm": 6.75, "learning_rate": 8.084845540473127e-06, "loss": 1.347, "step": 22980 }, { "epoch": 57.4525, "grad_norm": 7.1875, "learning_rate": 8.084050845588064e-06, "loss": 1.2544, "step": 22981 }, { "epoch": 57.455, "grad_norm": 6.28125, "learning_rate": 8.08325616326389e-06, "loss": 1.3834, "step": 22982 }, { "epoch": 57.4575, "grad_norm": 10.625, "learning_rate": 8.082461493505806e-06, "loss": 1.433, "step": 22983 }, { "epoch": 57.46, "grad_norm": 6.5, "learning_rate": 8.081666836319034e-06, "loss": 1.384, "step": 22984 }, { "epoch": 57.4625, "grad_norm": 7.3125, "learning_rate": 8.080872191708775e-06, "loss": 1.3162, "step": 22985 }, { "epoch": 57.465, "grad_norm": 5.65625, "learning_rate": 8.080077559680243e-06, "loss": 1.3279, "step": 22986 }, { "epoch": 57.4675, "grad_norm": 6.25, "learning_rate": 8.079282940238645e-06, "loss": 1.2503, "step": 22987 }, { "epoch": 57.47, "grad_norm": 9.6875, "learning_rate": 8.078488333389188e-06, "loss": 1.4068, "step": 22988 }, { "epoch": 57.4725, "grad_norm": 10.4375, "learning_rate": 8.077693739137087e-06, "loss": 1.3253, "step": 22989 }, { "epoch": 57.475, "grad_norm": 8.4375, "learning_rate": 8.076899157487551e-06, "loss": 1.221, "step": 22990 }, { "epoch": 57.4775, "grad_norm": 8.3125, "learning_rate": 8.076104588445784e-06, "loss": 1.3211, "step": 22991 }, { "epoch": 57.48, "grad_norm": 7.625, "learning_rate": 8.075310032017e-06, "loss": 1.2899, "step": 22992 }, { "epoch": 57.4825, "grad_norm": 7.375, "learning_rate": 8.074515488206404e-06, "loss": 1.3013, "step": 22993 }, { "epoch": 57.485, "grad_norm": 7.15625, "learning_rate": 8.073720957019208e-06, "loss": 1.2411, "step": 22994 }, { "epoch": 57.4875, "grad_norm": 7.3125, "learning_rate": 8.072926438460625e-06, "loss": 1.3316, "step": 22995 }, { "epoch": 57.49, "grad_norm": 7.875, "learning_rate": 8.072131932535855e-06, "loss": 1.2706, "step": 22996 }, { "epoch": 57.4925, "grad_norm": 7.875, "learning_rate": 8.071337439250114e-06, "loss": 1.1835, "step": 22997 }, { "epoch": 57.495, "grad_norm": 5.8125, "learning_rate": 8.0705429586086e-06, "loss": 1.2104, "step": 22998 }, { "epoch": 57.4975, "grad_norm": 9.5625, "learning_rate": 8.069748490616536e-06, "loss": 1.1816, "step": 22999 }, { "epoch": 57.5, "grad_norm": 5.15625, "learning_rate": 8.068954035279121e-06, "loss": 1.189, "step": 23000 }, { "epoch": 57.5025, "grad_norm": 9.3125, "learning_rate": 8.06815959260157e-06, "loss": 1.4573, "step": 23001 }, { "epoch": 57.505, "grad_norm": 9.125, "learning_rate": 8.067365162589083e-06, "loss": 1.3821, "step": 23002 }, { "epoch": 57.5075, "grad_norm": 7.6875, "learning_rate": 8.066570745246873e-06, "loss": 1.2933, "step": 23003 }, { "epoch": 57.51, "grad_norm": 8.125, "learning_rate": 8.065776340580148e-06, "loss": 1.3096, "step": 23004 }, { "epoch": 57.5125, "grad_norm": 7.75, "learning_rate": 8.06498194859412e-06, "loss": 1.2832, "step": 23005 }, { "epoch": 57.515, "grad_norm": 5.5625, "learning_rate": 8.064187569293989e-06, "loss": 1.207, "step": 23006 }, { "epoch": 57.5175, "grad_norm": 9.6875, "learning_rate": 8.063393202684968e-06, "loss": 1.3524, "step": 23007 }, { "epoch": 57.52, "grad_norm": 9.25, "learning_rate": 8.062598848772261e-06, "loss": 1.4192, "step": 23008 }, { "epoch": 57.5225, "grad_norm": 6.78125, "learning_rate": 8.061804507561082e-06, "loss": 1.2797, "step": 23009 }, { "epoch": 57.525, "grad_norm": 8.3125, "learning_rate": 8.061010179056633e-06, "loss": 1.3361, "step": 23010 }, { "epoch": 57.5275, "grad_norm": 6.71875, "learning_rate": 8.060215863264125e-06, "loss": 1.3385, "step": 23011 }, { "epoch": 57.53, "grad_norm": 9.75, "learning_rate": 8.059421560188765e-06, "loss": 1.3951, "step": 23012 }, { "epoch": 57.5325, "grad_norm": 7.46875, "learning_rate": 8.058627269835755e-06, "loss": 1.2217, "step": 23013 }, { "epoch": 57.535, "grad_norm": 11.0, "learning_rate": 8.05783299221031e-06, "loss": 1.3422, "step": 23014 }, { "epoch": 57.5375, "grad_norm": 8.5625, "learning_rate": 8.057038727317634e-06, "loss": 1.3284, "step": 23015 }, { "epoch": 57.54, "grad_norm": 5.8125, "learning_rate": 8.056244475162938e-06, "loss": 1.2423, "step": 23016 }, { "epoch": 57.5425, "grad_norm": 4.6875, "learning_rate": 8.05545023575142e-06, "loss": 1.3424, "step": 23017 }, { "epoch": 57.545, "grad_norm": 8.0, "learning_rate": 8.054656009088293e-06, "loss": 1.3561, "step": 23018 }, { "epoch": 57.5475, "grad_norm": 9.625, "learning_rate": 8.053861795178765e-06, "loss": 1.3558, "step": 23019 }, { "epoch": 57.55, "grad_norm": 7.9375, "learning_rate": 8.053067594028044e-06, "loss": 1.3686, "step": 23020 }, { "epoch": 57.5525, "grad_norm": 6.21875, "learning_rate": 8.05227340564133e-06, "loss": 1.3016, "step": 23021 }, { "epoch": 57.555, "grad_norm": 6.5625, "learning_rate": 8.051479230023835e-06, "loss": 1.2728, "step": 23022 }, { "epoch": 57.5575, "grad_norm": 7.875, "learning_rate": 8.050685067180761e-06, "loss": 1.4846, "step": 23023 }, { "epoch": 57.56, "grad_norm": 6.75, "learning_rate": 8.049890917117322e-06, "loss": 1.4124, "step": 23024 }, { "epoch": 57.5625, "grad_norm": 8.3125, "learning_rate": 8.04909677983872e-06, "loss": 1.3195, "step": 23025 }, { "epoch": 57.565, "grad_norm": 9.5, "learning_rate": 8.048302655350158e-06, "loss": 1.327, "step": 23026 }, { "epoch": 57.5675, "grad_norm": 6.40625, "learning_rate": 8.04750854365685e-06, "loss": 1.1916, "step": 23027 }, { "epoch": 57.57, "grad_norm": 8.4375, "learning_rate": 8.046714444763994e-06, "loss": 1.3449, "step": 23028 }, { "epoch": 57.5725, "grad_norm": 7.25, "learning_rate": 8.045920358676801e-06, "loss": 1.3007, "step": 23029 }, { "epoch": 57.575, "grad_norm": 31.625, "learning_rate": 8.045126285400477e-06, "loss": 1.3162, "step": 23030 }, { "epoch": 57.5775, "grad_norm": 7.5625, "learning_rate": 8.044332224940227e-06, "loss": 1.3929, "step": 23031 }, { "epoch": 57.58, "grad_norm": 8.125, "learning_rate": 8.043538177301256e-06, "loss": 1.278, "step": 23032 }, { "epoch": 57.5825, "grad_norm": 7.96875, "learning_rate": 8.042744142488767e-06, "loss": 1.4261, "step": 23033 }, { "epoch": 57.585, "grad_norm": 7.0, "learning_rate": 8.041950120507974e-06, "loss": 1.181, "step": 23034 }, { "epoch": 57.5875, "grad_norm": 7.5, "learning_rate": 8.041156111364077e-06, "loss": 1.3957, "step": 23035 }, { "epoch": 57.59, "grad_norm": 7.09375, "learning_rate": 8.040362115062282e-06, "loss": 1.3652, "step": 23036 }, { "epoch": 57.5925, "grad_norm": 6.28125, "learning_rate": 8.039568131607794e-06, "loss": 1.1849, "step": 23037 }, { "epoch": 57.595, "grad_norm": 8.0, "learning_rate": 8.038774161005816e-06, "loss": 1.235, "step": 23038 }, { "epoch": 57.5975, "grad_norm": 10.6875, "learning_rate": 8.03798020326156e-06, "loss": 1.4304, "step": 23039 }, { "epoch": 57.6, "grad_norm": 7.84375, "learning_rate": 8.037186258380226e-06, "loss": 1.1951, "step": 23040 }, { "epoch": 57.6025, "grad_norm": 10.4375, "learning_rate": 8.03639232636702e-06, "loss": 1.2875, "step": 23041 }, { "epoch": 57.605, "grad_norm": 6.125, "learning_rate": 8.03559840722715e-06, "loss": 1.3295, "step": 23042 }, { "epoch": 57.6075, "grad_norm": 6.40625, "learning_rate": 8.034804500965814e-06, "loss": 1.3937, "step": 23043 }, { "epoch": 57.61, "grad_norm": 6.5625, "learning_rate": 8.034010607588223e-06, "loss": 1.3839, "step": 23044 }, { "epoch": 57.6125, "grad_norm": 6.59375, "learning_rate": 8.03321672709958e-06, "loss": 1.1794, "step": 23045 }, { "epoch": 57.615, "grad_norm": 9.5625, "learning_rate": 8.03242285950509e-06, "loss": 1.3692, "step": 23046 }, { "epoch": 57.6175, "grad_norm": 7.125, "learning_rate": 8.031629004809957e-06, "loss": 1.3383, "step": 23047 }, { "epoch": 57.62, "grad_norm": 23.375, "learning_rate": 8.030835163019382e-06, "loss": 1.366, "step": 23048 }, { "epoch": 57.6225, "grad_norm": 6.28125, "learning_rate": 8.030041334138575e-06, "loss": 1.3861, "step": 23049 }, { "epoch": 57.625, "grad_norm": 5.65625, "learning_rate": 8.02924751817274e-06, "loss": 1.3473, "step": 23050 }, { "epoch": 57.6275, "grad_norm": 9.4375, "learning_rate": 8.028453715127079e-06, "loss": 1.3284, "step": 23051 }, { "epoch": 57.63, "grad_norm": 9.1875, "learning_rate": 8.027659925006798e-06, "loss": 1.3201, "step": 23052 }, { "epoch": 57.6325, "grad_norm": 9.875, "learning_rate": 8.026866147817091e-06, "loss": 1.4333, "step": 23053 }, { "epoch": 57.635, "grad_norm": 8.625, "learning_rate": 8.02607238356318e-06, "loss": 1.4765, "step": 23054 }, { "epoch": 57.6375, "grad_norm": 7.125, "learning_rate": 8.025278632250256e-06, "loss": 1.2502, "step": 23055 }, { "epoch": 57.64, "grad_norm": 9.25, "learning_rate": 8.02448489388353e-06, "loss": 1.3108, "step": 23056 }, { "epoch": 57.6425, "grad_norm": 9.375, "learning_rate": 8.023691168468197e-06, "loss": 1.2329, "step": 23057 }, { "epoch": 57.645, "grad_norm": 7.3125, "learning_rate": 8.022897456009465e-06, "loss": 1.335, "step": 23058 }, { "epoch": 57.6475, "grad_norm": 7.96875, "learning_rate": 8.02210375651254e-06, "loss": 1.3236, "step": 23059 }, { "epoch": 57.65, "grad_norm": 8.1875, "learning_rate": 8.021310069982624e-06, "loss": 1.2525, "step": 23060 }, { "epoch": 57.6525, "grad_norm": 9.25, "learning_rate": 8.020516396424922e-06, "loss": 1.4079, "step": 23061 }, { "epoch": 57.655, "grad_norm": 8.375, "learning_rate": 8.019722735844632e-06, "loss": 1.2952, "step": 23062 }, { "epoch": 57.6575, "grad_norm": 8.625, "learning_rate": 8.01892908824696e-06, "loss": 1.6968, "step": 23063 }, { "epoch": 57.66, "grad_norm": 5.46875, "learning_rate": 8.018135453637112e-06, "loss": 1.2275, "step": 23064 }, { "epoch": 57.6625, "grad_norm": 8.5, "learning_rate": 8.01734183202029e-06, "loss": 1.2576, "step": 23065 }, { "epoch": 57.665, "grad_norm": 8.8125, "learning_rate": 8.016548223401694e-06, "loss": 1.2969, "step": 23066 }, { "epoch": 57.6675, "grad_norm": 6.5, "learning_rate": 8.015754627786529e-06, "loss": 1.3166, "step": 23067 }, { "epoch": 57.67, "grad_norm": 7.34375, "learning_rate": 8.014961045179992e-06, "loss": 1.2756, "step": 23068 }, { "epoch": 57.6725, "grad_norm": 7.75, "learning_rate": 8.014167475587297e-06, "loss": 1.3894, "step": 23069 }, { "epoch": 57.675, "grad_norm": 7.8125, "learning_rate": 8.013373919013639e-06, "loss": 1.4444, "step": 23070 }, { "epoch": 57.6775, "grad_norm": 6.3125, "learning_rate": 8.012580375464224e-06, "loss": 1.2449, "step": 23071 }, { "epoch": 57.68, "grad_norm": 7.4375, "learning_rate": 8.01178684494425e-06, "loss": 1.3306, "step": 23072 }, { "epoch": 57.6825, "grad_norm": 6.21875, "learning_rate": 8.01099332745892e-06, "loss": 1.3249, "step": 23073 }, { "epoch": 57.685, "grad_norm": 6.1875, "learning_rate": 8.010199823013441e-06, "loss": 1.2275, "step": 23074 }, { "epoch": 57.6875, "grad_norm": 7.125, "learning_rate": 8.009406331613014e-06, "loss": 1.3753, "step": 23075 }, { "epoch": 57.69, "grad_norm": 6.3125, "learning_rate": 8.008612853262837e-06, "loss": 1.3441, "step": 23076 }, { "epoch": 57.6925, "grad_norm": 7.3125, "learning_rate": 8.007819387968114e-06, "loss": 1.3221, "step": 23077 }, { "epoch": 57.695, "grad_norm": 4.6875, "learning_rate": 8.007025935734046e-06, "loss": 1.1855, "step": 23078 }, { "epoch": 57.6975, "grad_norm": 9.0, "learning_rate": 8.00623249656584e-06, "loss": 1.3069, "step": 23079 }, { "epoch": 57.7, "grad_norm": 8.125, "learning_rate": 8.005439070468692e-06, "loss": 1.3856, "step": 23080 }, { "epoch": 57.7025, "grad_norm": 6.8125, "learning_rate": 8.004645657447806e-06, "loss": 1.3893, "step": 23081 }, { "epoch": 57.705, "grad_norm": 8.3125, "learning_rate": 8.003852257508385e-06, "loss": 1.3482, "step": 23082 }, { "epoch": 57.7075, "grad_norm": 6.96875, "learning_rate": 8.003058870655625e-06, "loss": 1.264, "step": 23083 }, { "epoch": 57.71, "grad_norm": 7.90625, "learning_rate": 8.002265496894733e-06, "loss": 1.1911, "step": 23084 }, { "epoch": 57.7125, "grad_norm": 6.15625, "learning_rate": 8.001472136230907e-06, "loss": 1.2003, "step": 23085 }, { "epoch": 57.715, "grad_norm": 11.5, "learning_rate": 8.000678788669354e-06, "loss": 1.2721, "step": 23086 }, { "epoch": 57.7175, "grad_norm": 8.9375, "learning_rate": 7.999885454215269e-06, "loss": 1.3835, "step": 23087 }, { "epoch": 57.72, "grad_norm": 5.40625, "learning_rate": 7.999092132873851e-06, "loss": 1.313, "step": 23088 }, { "epoch": 57.7225, "grad_norm": 6.53125, "learning_rate": 7.998298824650307e-06, "loss": 1.3961, "step": 23089 }, { "epoch": 57.725, "grad_norm": 8.8125, "learning_rate": 7.99750552954984e-06, "loss": 1.1138, "step": 23090 }, { "epoch": 57.7275, "grad_norm": 11.8125, "learning_rate": 7.996712247577643e-06, "loss": 1.4906, "step": 23091 }, { "epoch": 57.73, "grad_norm": 9.5625, "learning_rate": 7.995918978738918e-06, "loss": 1.484, "step": 23092 }, { "epoch": 57.7325, "grad_norm": 8.5625, "learning_rate": 7.99512572303887e-06, "loss": 1.2552, "step": 23093 }, { "epoch": 57.735, "grad_norm": 7.09375, "learning_rate": 7.994332480482699e-06, "loss": 1.3684, "step": 23094 }, { "epoch": 57.7375, "grad_norm": 7.40625, "learning_rate": 7.993539251075604e-06, "loss": 1.326, "step": 23095 }, { "epoch": 57.74, "grad_norm": 6.3125, "learning_rate": 7.992746034822783e-06, "loss": 1.3196, "step": 23096 }, { "epoch": 57.7425, "grad_norm": 4.96875, "learning_rate": 7.991952831729443e-06, "loss": 1.3455, "step": 23097 }, { "epoch": 57.745, "grad_norm": 8.1875, "learning_rate": 7.991159641800774e-06, "loss": 1.294, "step": 23098 }, { "epoch": 57.7475, "grad_norm": 5.4375, "learning_rate": 7.990366465041985e-06, "loss": 1.331, "step": 23099 }, { "epoch": 57.75, "grad_norm": 6.5, "learning_rate": 7.989573301458274e-06, "loss": 1.2356, "step": 23100 }, { "epoch": 57.7525, "grad_norm": 5.0625, "learning_rate": 7.98878015105484e-06, "loss": 1.4646, "step": 23101 }, { "epoch": 57.755, "grad_norm": 8.625, "learning_rate": 7.987987013836883e-06, "loss": 1.3465, "step": 23102 }, { "epoch": 57.7575, "grad_norm": 9.25, "learning_rate": 7.987193889809598e-06, "loss": 1.2047, "step": 23103 }, { "epoch": 57.76, "grad_norm": 8.4375, "learning_rate": 7.986400778978192e-06, "loss": 1.1709, "step": 23104 }, { "epoch": 57.7625, "grad_norm": 5.90625, "learning_rate": 7.985607681347866e-06, "loss": 1.3609, "step": 23105 }, { "epoch": 57.765, "grad_norm": 8.8125, "learning_rate": 7.984814596923812e-06, "loss": 1.3109, "step": 23106 }, { "epoch": 57.7675, "grad_norm": 9.8125, "learning_rate": 7.984021525711232e-06, "loss": 1.2449, "step": 23107 }, { "epoch": 57.77, "grad_norm": 8.5625, "learning_rate": 7.983228467715325e-06, "loss": 1.3098, "step": 23108 }, { "epoch": 57.7725, "grad_norm": 11.8125, "learning_rate": 7.982435422941297e-06, "loss": 1.3829, "step": 23109 }, { "epoch": 57.775, "grad_norm": 6.59375, "learning_rate": 7.981642391394338e-06, "loss": 1.3517, "step": 23110 }, { "epoch": 57.7775, "grad_norm": 5.53125, "learning_rate": 7.98084937307965e-06, "loss": 1.4137, "step": 23111 }, { "epoch": 57.78, "grad_norm": 7.34375, "learning_rate": 7.980056368002435e-06, "loss": 1.2518, "step": 23112 }, { "epoch": 57.7825, "grad_norm": 5.71875, "learning_rate": 7.979263376167886e-06, "loss": 1.299, "step": 23113 }, { "epoch": 57.785, "grad_norm": 7.5625, "learning_rate": 7.978470397581209e-06, "loss": 1.431, "step": 23114 }, { "epoch": 57.7875, "grad_norm": 5.25, "learning_rate": 7.977677432247596e-06, "loss": 1.2, "step": 23115 }, { "epoch": 57.79, "grad_norm": 6.28125, "learning_rate": 7.976884480172254e-06, "loss": 1.2895, "step": 23116 }, { "epoch": 57.7925, "grad_norm": 10.4375, "learning_rate": 7.976091541360373e-06, "loss": 1.1777, "step": 23117 }, { "epoch": 57.795, "grad_norm": 7.34375, "learning_rate": 7.975298615817152e-06, "loss": 1.3551, "step": 23118 }, { "epoch": 57.7975, "grad_norm": 9.625, "learning_rate": 7.974505703547795e-06, "loss": 1.4894, "step": 23119 }, { "epoch": 57.8, "grad_norm": 7.8125, "learning_rate": 7.9737128045575e-06, "loss": 1.4849, "step": 23120 }, { "epoch": 57.8025, "grad_norm": 6.96875, "learning_rate": 7.972919918851461e-06, "loss": 1.1981, "step": 23121 }, { "epoch": 57.805, "grad_norm": 10.5, "learning_rate": 7.972127046434878e-06, "loss": 1.2699, "step": 23122 }, { "epoch": 57.8075, "grad_norm": 7.65625, "learning_rate": 7.971334187312945e-06, "loss": 1.1688, "step": 23123 }, { "epoch": 57.81, "grad_norm": 7.40625, "learning_rate": 7.97054134149087e-06, "loss": 1.4503, "step": 23124 }, { "epoch": 57.8125, "grad_norm": 6.28125, "learning_rate": 7.969748508973842e-06, "loss": 1.323, "step": 23125 }, { "epoch": 57.815, "grad_norm": 7.3125, "learning_rate": 7.968955689767062e-06, "loss": 1.2166, "step": 23126 }, { "epoch": 57.8175, "grad_norm": 11.0625, "learning_rate": 7.968162883875728e-06, "loss": 1.3356, "step": 23127 }, { "epoch": 57.82, "grad_norm": 8.75, "learning_rate": 7.967370091305033e-06, "loss": 1.2837, "step": 23128 }, { "epoch": 57.8225, "grad_norm": 8.0, "learning_rate": 7.966577312060182e-06, "loss": 1.3267, "step": 23129 }, { "epoch": 57.825, "grad_norm": 9.0625, "learning_rate": 7.965784546146365e-06, "loss": 1.6649, "step": 23130 }, { "epoch": 57.8275, "grad_norm": 6.8125, "learning_rate": 7.964991793568788e-06, "loss": 1.3393, "step": 23131 }, { "epoch": 57.83, "grad_norm": 6.3125, "learning_rate": 7.964199054332642e-06, "loss": 1.2758, "step": 23132 }, { "epoch": 57.8325, "grad_norm": 7.5625, "learning_rate": 7.963406328443121e-06, "loss": 1.1765, "step": 23133 }, { "epoch": 57.835, "grad_norm": 8.9375, "learning_rate": 7.962613615905431e-06, "loss": 1.3858, "step": 23134 }, { "epoch": 57.8375, "grad_norm": 5.40625, "learning_rate": 7.961820916724767e-06, "loss": 1.2132, "step": 23135 }, { "epoch": 57.84, "grad_norm": 9.875, "learning_rate": 7.96102823090632e-06, "loss": 1.2747, "step": 23136 }, { "epoch": 57.8425, "grad_norm": 6.21875, "learning_rate": 7.960235558455293e-06, "loss": 1.3341, "step": 23137 }, { "epoch": 57.845, "grad_norm": 18.375, "learning_rate": 7.959442899376873e-06, "loss": 1.2342, "step": 23138 }, { "epoch": 57.8475, "grad_norm": 8.5625, "learning_rate": 7.95865025367627e-06, "loss": 1.3377, "step": 23139 }, { "epoch": 57.85, "grad_norm": 8.8125, "learning_rate": 7.957857621358674e-06, "loss": 1.2565, "step": 23140 }, { "epoch": 57.8525, "grad_norm": 7.71875, "learning_rate": 7.957065002429282e-06, "loss": 1.3397, "step": 23141 }, { "epoch": 57.855, "grad_norm": 10.1875, "learning_rate": 7.956272396893288e-06, "loss": 1.2871, "step": 23142 }, { "epoch": 57.8575, "grad_norm": 8.375, "learning_rate": 7.955479804755887e-06, "loss": 1.4849, "step": 23143 }, { "epoch": 57.86, "grad_norm": 6.21875, "learning_rate": 7.954687226022285e-06, "loss": 1.266, "step": 23144 }, { "epoch": 57.8625, "grad_norm": 5.6875, "learning_rate": 7.95389466069767e-06, "loss": 1.0934, "step": 23145 }, { "epoch": 57.865, "grad_norm": 5.15625, "learning_rate": 7.953102108787239e-06, "loss": 1.4427, "step": 23146 }, { "epoch": 57.8675, "grad_norm": 9.4375, "learning_rate": 7.952309570296187e-06, "loss": 1.3625, "step": 23147 }, { "epoch": 57.87, "grad_norm": 8.5625, "learning_rate": 7.951517045229713e-06, "loss": 1.302, "step": 23148 }, { "epoch": 57.8725, "grad_norm": 8.0625, "learning_rate": 7.950724533593014e-06, "loss": 1.2889, "step": 23149 }, { "epoch": 57.875, "grad_norm": 8.6875, "learning_rate": 7.949932035391279e-06, "loss": 1.3036, "step": 23150 }, { "epoch": 57.8775, "grad_norm": 5.28125, "learning_rate": 7.949139550629708e-06, "loss": 1.2829, "step": 23151 }, { "epoch": 57.88, "grad_norm": 5.90625, "learning_rate": 7.948347079313494e-06, "loss": 1.27, "step": 23152 }, { "epoch": 57.8825, "grad_norm": 10.5625, "learning_rate": 7.94755462144784e-06, "loss": 1.462, "step": 23153 }, { "epoch": 57.885, "grad_norm": 6.0625, "learning_rate": 7.946762177037934e-06, "loss": 1.2374, "step": 23154 }, { "epoch": 57.8875, "grad_norm": 12.25, "learning_rate": 7.945969746088971e-06, "loss": 1.4361, "step": 23155 }, { "epoch": 57.89, "grad_norm": 5.9375, "learning_rate": 7.945177328606153e-06, "loss": 1.261, "step": 23156 }, { "epoch": 57.8925, "grad_norm": 9.3125, "learning_rate": 7.94438492459466e-06, "loss": 1.2129, "step": 23157 }, { "epoch": 57.895, "grad_norm": 8.0625, "learning_rate": 7.943592534059707e-06, "loss": 1.2868, "step": 23158 }, { "epoch": 57.8975, "grad_norm": 6.09375, "learning_rate": 7.942800157006476e-06, "loss": 1.2718, "step": 23159 }, { "epoch": 57.9, "grad_norm": 6.59375, "learning_rate": 7.942007793440165e-06, "loss": 1.2746, "step": 23160 }, { "epoch": 57.9025, "grad_norm": 6.71875, "learning_rate": 7.941215443365968e-06, "loss": 1.3018, "step": 23161 }, { "epoch": 57.905, "grad_norm": 7.5625, "learning_rate": 7.940423106789078e-06, "loss": 1.4219, "step": 23162 }, { "epoch": 57.9075, "grad_norm": 8.1875, "learning_rate": 7.939630783714693e-06, "loss": 1.3779, "step": 23163 }, { "epoch": 57.91, "grad_norm": 6.5625, "learning_rate": 7.93883847414801e-06, "loss": 1.3874, "step": 23164 }, { "epoch": 57.9125, "grad_norm": 5.375, "learning_rate": 7.938046178094217e-06, "loss": 1.3696, "step": 23165 }, { "epoch": 57.915, "grad_norm": 6.0, "learning_rate": 7.93725389555851e-06, "loss": 1.3326, "step": 23166 }, { "epoch": 57.9175, "grad_norm": 5.34375, "learning_rate": 7.936461626546081e-06, "loss": 1.2396, "step": 23167 }, { "epoch": 57.92, "grad_norm": 8.5625, "learning_rate": 7.935669371062132e-06, "loss": 1.1867, "step": 23168 }, { "epoch": 57.9225, "grad_norm": 5.9375, "learning_rate": 7.934877129111851e-06, "loss": 1.2903, "step": 23169 }, { "epoch": 57.925, "grad_norm": 10.9375, "learning_rate": 7.934084900700432e-06, "loss": 1.388, "step": 23170 }, { "epoch": 57.9275, "grad_norm": 7.9375, "learning_rate": 7.93329268583307e-06, "loss": 1.346, "step": 23171 }, { "epoch": 57.93, "grad_norm": 7.6875, "learning_rate": 7.932500484514957e-06, "loss": 1.4461, "step": 23172 }, { "epoch": 57.9325, "grad_norm": 5.65625, "learning_rate": 7.93170829675129e-06, "loss": 1.364, "step": 23173 }, { "epoch": 57.935, "grad_norm": 6.59375, "learning_rate": 7.930916122547259e-06, "loss": 1.4007, "step": 23174 }, { "epoch": 57.9375, "grad_norm": 6.6875, "learning_rate": 7.930123961908062e-06, "loss": 1.3103, "step": 23175 }, { "epoch": 57.94, "grad_norm": 9.5, "learning_rate": 7.929331814838889e-06, "loss": 1.1886, "step": 23176 }, { "epoch": 57.9425, "grad_norm": 6.28125, "learning_rate": 7.92853968134493e-06, "loss": 1.3015, "step": 23177 }, { "epoch": 57.945, "grad_norm": 6.34375, "learning_rate": 7.927747561431383e-06, "loss": 1.5633, "step": 23178 }, { "epoch": 57.9475, "grad_norm": 6.96875, "learning_rate": 7.926955455103444e-06, "loss": 1.4048, "step": 23179 }, { "epoch": 57.95, "grad_norm": 11.25, "learning_rate": 7.9261633623663e-06, "loss": 1.319, "step": 23180 }, { "epoch": 57.9525, "grad_norm": 8.0625, "learning_rate": 7.925371283225145e-06, "loss": 1.367, "step": 23181 }, { "epoch": 57.955, "grad_norm": 7.59375, "learning_rate": 7.924579217685172e-06, "loss": 1.3934, "step": 23182 }, { "epoch": 57.9575, "grad_norm": 6.78125, "learning_rate": 7.923787165751576e-06, "loss": 1.4289, "step": 23183 }, { "epoch": 57.96, "grad_norm": 7.40625, "learning_rate": 7.922995127429547e-06, "loss": 1.4669, "step": 23184 }, { "epoch": 57.9625, "grad_norm": 7.3125, "learning_rate": 7.922203102724282e-06, "loss": 1.2549, "step": 23185 }, { "epoch": 57.965, "grad_norm": 9.9375, "learning_rate": 7.921411091640969e-06, "loss": 1.3455, "step": 23186 }, { "epoch": 57.9675, "grad_norm": 7.0, "learning_rate": 7.920619094184799e-06, "loss": 1.3143, "step": 23187 }, { "epoch": 57.97, "grad_norm": 7.75, "learning_rate": 7.919827110360968e-06, "loss": 1.4659, "step": 23188 }, { "epoch": 57.9725, "grad_norm": 55.0, "learning_rate": 7.919035140174668e-06, "loss": 1.4425, "step": 23189 }, { "epoch": 57.975, "grad_norm": 9.75, "learning_rate": 7.918243183631094e-06, "loss": 1.2764, "step": 23190 }, { "epoch": 57.9775, "grad_norm": 8.1875, "learning_rate": 7.91745124073543e-06, "loss": 1.1726, "step": 23191 }, { "epoch": 57.98, "grad_norm": 8.125, "learning_rate": 7.916659311492871e-06, "loss": 1.3046, "step": 23192 }, { "epoch": 57.9825, "grad_norm": 6.78125, "learning_rate": 7.915867395908612e-06, "loss": 1.2157, "step": 23193 }, { "epoch": 57.985, "grad_norm": 6.34375, "learning_rate": 7.915075493987846e-06, "loss": 1.4831, "step": 23194 }, { "epoch": 57.9875, "grad_norm": 6.65625, "learning_rate": 7.91428360573576e-06, "loss": 1.3808, "step": 23195 }, { "epoch": 57.99, "grad_norm": 6.40625, "learning_rate": 7.913491731157546e-06, "loss": 1.2447, "step": 23196 }, { "epoch": 57.9925, "grad_norm": 11.3125, "learning_rate": 7.912699870258393e-06, "loss": 1.2861, "step": 23197 }, { "epoch": 57.995, "grad_norm": 7.40625, "learning_rate": 7.911908023043502e-06, "loss": 1.2132, "step": 23198 }, { "epoch": 57.9975, "grad_norm": 9.25, "learning_rate": 7.911116189518057e-06, "loss": 1.3508, "step": 23199 }, { "epoch": 58.0, "grad_norm": 9.375, "learning_rate": 7.91032436968725e-06, "loss": 1.278, "step": 23200 }, { "epoch": 58.0025, "grad_norm": 10.0, "learning_rate": 7.909532563556275e-06, "loss": 1.29, "step": 23201 }, { "epoch": 58.005, "grad_norm": 8.625, "learning_rate": 7.908740771130317e-06, "loss": 1.3194, "step": 23202 }, { "epoch": 58.0075, "grad_norm": 7.40625, "learning_rate": 7.907948992414571e-06, "loss": 1.2155, "step": 23203 }, { "epoch": 58.01, "grad_norm": 6.75, "learning_rate": 7.90715722741423e-06, "loss": 1.2677, "step": 23204 }, { "epoch": 58.0125, "grad_norm": 4.96875, "learning_rate": 7.906365476134483e-06, "loss": 1.217, "step": 23205 }, { "epoch": 58.015, "grad_norm": 7.8125, "learning_rate": 7.90557373858052e-06, "loss": 1.5161, "step": 23206 }, { "epoch": 58.0175, "grad_norm": 8.875, "learning_rate": 7.904782014757528e-06, "loss": 1.2056, "step": 23207 }, { "epoch": 58.02, "grad_norm": 7.96875, "learning_rate": 7.903990304670704e-06, "loss": 1.2165, "step": 23208 }, { "epoch": 58.0225, "grad_norm": 10.0, "learning_rate": 7.903198608325238e-06, "loss": 1.2886, "step": 23209 }, { "epoch": 58.025, "grad_norm": 7.4375, "learning_rate": 7.902406925726315e-06, "loss": 1.1467, "step": 23210 }, { "epoch": 58.0275, "grad_norm": 7.65625, "learning_rate": 7.901615256879131e-06, "loss": 1.2912, "step": 23211 }, { "epoch": 58.03, "grad_norm": 7.75, "learning_rate": 7.900823601788867e-06, "loss": 1.5041, "step": 23212 }, { "epoch": 58.0325, "grad_norm": 9.0, "learning_rate": 7.900031960460725e-06, "loss": 1.3383, "step": 23213 }, { "epoch": 58.035, "grad_norm": 5.625, "learning_rate": 7.89924033289989e-06, "loss": 1.3765, "step": 23214 }, { "epoch": 58.0375, "grad_norm": 7.875, "learning_rate": 7.898448719111552e-06, "loss": 1.2969, "step": 23215 }, { "epoch": 58.04, "grad_norm": 6.8125, "learning_rate": 7.897657119100896e-06, "loss": 1.3933, "step": 23216 }, { "epoch": 58.0425, "grad_norm": 7.84375, "learning_rate": 7.896865532873114e-06, "loss": 1.3181, "step": 23217 }, { "epoch": 58.045, "grad_norm": 7.96875, "learning_rate": 7.896073960433401e-06, "loss": 1.3144, "step": 23218 }, { "epoch": 58.0475, "grad_norm": 8.5625, "learning_rate": 7.895282401786946e-06, "loss": 1.2529, "step": 23219 }, { "epoch": 58.05, "grad_norm": 5.9375, "learning_rate": 7.894490856938931e-06, "loss": 1.1962, "step": 23220 }, { "epoch": 58.0525, "grad_norm": 11.0625, "learning_rate": 7.893699325894552e-06, "loss": 1.412, "step": 23221 }, { "epoch": 58.055, "grad_norm": 7.34375, "learning_rate": 7.892907808658991e-06, "loss": 1.3106, "step": 23222 }, { "epoch": 58.0575, "grad_norm": 9.4375, "learning_rate": 7.892116305237449e-06, "loss": 1.3004, "step": 23223 }, { "epoch": 58.06, "grad_norm": 6.9375, "learning_rate": 7.891324815635104e-06, "loss": 1.3468, "step": 23224 }, { "epoch": 58.0625, "grad_norm": 11.0625, "learning_rate": 7.89053333985715e-06, "loss": 1.3149, "step": 23225 }, { "epoch": 58.065, "grad_norm": 7.84375, "learning_rate": 7.889741877908778e-06, "loss": 1.5334, "step": 23226 }, { "epoch": 58.0675, "grad_norm": 6.1875, "learning_rate": 7.888950429795166e-06, "loss": 1.2853, "step": 23227 }, { "epoch": 58.07, "grad_norm": 7.96875, "learning_rate": 7.888158995521518e-06, "loss": 1.3182, "step": 23228 }, { "epoch": 58.0725, "grad_norm": 7.21875, "learning_rate": 7.887367575093011e-06, "loss": 1.192, "step": 23229 }, { "epoch": 58.075, "grad_norm": 7.3125, "learning_rate": 7.886576168514841e-06, "loss": 1.2731, "step": 23230 }, { "epoch": 58.0775, "grad_norm": 8.125, "learning_rate": 7.88578477579219e-06, "loss": 1.3183, "step": 23231 }, { "epoch": 58.08, "grad_norm": 9.3125, "learning_rate": 7.88499339693025e-06, "loss": 1.4564, "step": 23232 }, { "epoch": 58.0825, "grad_norm": 5.78125, "learning_rate": 7.88420203193421e-06, "loss": 1.3818, "step": 23233 }, { "epoch": 58.085, "grad_norm": 10.1875, "learning_rate": 7.883410680809257e-06, "loss": 1.275, "step": 23234 }, { "epoch": 58.0875, "grad_norm": 9.6875, "learning_rate": 7.882619343560578e-06, "loss": 1.5128, "step": 23235 }, { "epoch": 58.09, "grad_norm": 5.84375, "learning_rate": 7.881828020193362e-06, "loss": 1.2852, "step": 23236 }, { "epoch": 58.0925, "grad_norm": 7.3125, "learning_rate": 7.881036710712794e-06, "loss": 1.409, "step": 23237 }, { "epoch": 58.095, "grad_norm": 6.9375, "learning_rate": 7.880245415124069e-06, "loss": 1.2975, "step": 23238 }, { "epoch": 58.0975, "grad_norm": 6.75, "learning_rate": 7.879454133432369e-06, "loss": 1.1728, "step": 23239 }, { "epoch": 58.1, "grad_norm": 6.53125, "learning_rate": 7.87866286564288e-06, "loss": 1.3669, "step": 23240 }, { "epoch": 58.1025, "grad_norm": 4.15625, "learning_rate": 7.877871611760798e-06, "loss": 1.1809, "step": 23241 }, { "epoch": 58.105, "grad_norm": 5.8125, "learning_rate": 7.877080371791299e-06, "loss": 1.2947, "step": 23242 }, { "epoch": 58.1075, "grad_norm": 8.9375, "learning_rate": 7.876289145739577e-06, "loss": 1.2002, "step": 23243 }, { "epoch": 58.11, "grad_norm": 8.3125, "learning_rate": 7.87549793361082e-06, "loss": 1.3233, "step": 23244 }, { "epoch": 58.1125, "grad_norm": 9.3125, "learning_rate": 7.874706735410216e-06, "loss": 1.2611, "step": 23245 }, { "epoch": 58.115, "grad_norm": 8.0625, "learning_rate": 7.873915551142946e-06, "loss": 1.3829, "step": 23246 }, { "epoch": 58.1175, "grad_norm": 7.375, "learning_rate": 7.8731243808142e-06, "loss": 1.477, "step": 23247 }, { "epoch": 58.12, "grad_norm": 8.0, "learning_rate": 7.872333224429166e-06, "loss": 1.1319, "step": 23248 }, { "epoch": 58.1225, "grad_norm": 8.1875, "learning_rate": 7.871542081993034e-06, "loss": 1.3751, "step": 23249 }, { "epoch": 58.125, "grad_norm": 10.0, "learning_rate": 7.870750953510983e-06, "loss": 1.3211, "step": 23250 }, { "epoch": 58.1275, "grad_norm": 7.0, "learning_rate": 7.869959838988206e-06, "loss": 1.2823, "step": 23251 }, { "epoch": 58.13, "grad_norm": 10.3125, "learning_rate": 7.869168738429883e-06, "loss": 1.4153, "step": 23252 }, { "epoch": 58.1325, "grad_norm": 4.90625, "learning_rate": 7.868377651841209e-06, "loss": 1.2821, "step": 23253 }, { "epoch": 58.135, "grad_norm": 7.25, "learning_rate": 7.867586579227364e-06, "loss": 1.2926, "step": 23254 }, { "epoch": 58.1375, "grad_norm": 6.65625, "learning_rate": 7.866795520593536e-06, "loss": 1.4473, "step": 23255 }, { "epoch": 58.14, "grad_norm": 8.5625, "learning_rate": 7.866004475944913e-06, "loss": 1.3293, "step": 23256 }, { "epoch": 58.1425, "grad_norm": 7.03125, "learning_rate": 7.865213445286675e-06, "loss": 1.3974, "step": 23257 }, { "epoch": 58.145, "grad_norm": 10.375, "learning_rate": 7.864422428624015e-06, "loss": 1.2684, "step": 23258 }, { "epoch": 58.1475, "grad_norm": 6.375, "learning_rate": 7.863631425962115e-06, "loss": 1.2254, "step": 23259 }, { "epoch": 58.15, "grad_norm": 10.0625, "learning_rate": 7.862840437306165e-06, "loss": 1.3269, "step": 23260 }, { "epoch": 58.1525, "grad_norm": 9.75, "learning_rate": 7.862049462661345e-06, "loss": 1.4355, "step": 23261 }, { "epoch": 58.155, "grad_norm": 8.5625, "learning_rate": 7.861258502032842e-06, "loss": 1.4001, "step": 23262 }, { "epoch": 58.1575, "grad_norm": 15.5, "learning_rate": 7.860467555425843e-06, "loss": 1.1932, "step": 23263 }, { "epoch": 58.16, "grad_norm": 7.40625, "learning_rate": 7.859676622845535e-06, "loss": 1.3208, "step": 23264 }, { "epoch": 58.1625, "grad_norm": 5.40625, "learning_rate": 7.858885704297102e-06, "loss": 1.4737, "step": 23265 }, { "epoch": 58.165, "grad_norm": 8.0625, "learning_rate": 7.85809479978573e-06, "loss": 1.2762, "step": 23266 }, { "epoch": 58.1675, "grad_norm": 6.875, "learning_rate": 7.857303909316594e-06, "loss": 1.4035, "step": 23267 }, { "epoch": 58.17, "grad_norm": 10.0, "learning_rate": 7.856513032894896e-06, "loss": 1.3613, "step": 23268 }, { "epoch": 58.1725, "grad_norm": 6.8125, "learning_rate": 7.855722170525812e-06, "loss": 1.2352, "step": 23269 }, { "epoch": 58.175, "grad_norm": 8.0, "learning_rate": 7.854931322214525e-06, "loss": 1.2885, "step": 23270 }, { "epoch": 58.1775, "grad_norm": 11.875, "learning_rate": 7.854140487966226e-06, "loss": 1.4152, "step": 23271 }, { "epoch": 58.18, "grad_norm": 6.5, "learning_rate": 7.85334966778609e-06, "loss": 1.3911, "step": 23272 }, { "epoch": 58.1825, "grad_norm": 35.25, "learning_rate": 7.852558861679313e-06, "loss": 1.3778, "step": 23273 }, { "epoch": 58.185, "grad_norm": 6.90625, "learning_rate": 7.851768069651073e-06, "loss": 1.2538, "step": 23274 }, { "epoch": 58.1875, "grad_norm": 9.75, "learning_rate": 7.850977291706557e-06, "loss": 1.3333, "step": 23275 }, { "epoch": 58.19, "grad_norm": 12.75, "learning_rate": 7.850186527850944e-06, "loss": 1.3591, "step": 23276 }, { "epoch": 58.1925, "grad_norm": 6.875, "learning_rate": 7.849395778089423e-06, "loss": 1.3291, "step": 23277 }, { "epoch": 58.195, "grad_norm": 9.875, "learning_rate": 7.848605042427177e-06, "loss": 1.4106, "step": 23278 }, { "epoch": 58.1975, "grad_norm": 6.65625, "learning_rate": 7.847814320869395e-06, "loss": 1.2206, "step": 23279 }, { "epoch": 58.2, "grad_norm": 11.125, "learning_rate": 7.847023613421251e-06, "loss": 1.3402, "step": 23280 }, { "epoch": 58.2025, "grad_norm": 8.25, "learning_rate": 7.84623292008794e-06, "loss": 1.4081, "step": 23281 }, { "epoch": 58.205, "grad_norm": 8.4375, "learning_rate": 7.84544224087463e-06, "loss": 1.3548, "step": 23282 }, { "epoch": 58.2075, "grad_norm": 8.4375, "learning_rate": 7.844651575786523e-06, "loss": 1.306, "step": 23283 }, { "epoch": 58.21, "grad_norm": 7.46875, "learning_rate": 7.84386092482879e-06, "loss": 1.2386, "step": 23284 }, { "epoch": 58.2125, "grad_norm": 7.6875, "learning_rate": 7.843070288006622e-06, "loss": 1.2295, "step": 23285 }, { "epoch": 58.215, "grad_norm": 7.625, "learning_rate": 7.842279665325197e-06, "loss": 1.4694, "step": 23286 }, { "epoch": 58.2175, "grad_norm": 7.96875, "learning_rate": 7.841489056789696e-06, "loss": 1.4317, "step": 23287 }, { "epoch": 58.22, "grad_norm": 6.65625, "learning_rate": 7.84069846240531e-06, "loss": 1.3624, "step": 23288 }, { "epoch": 58.2225, "grad_norm": 7.4375, "learning_rate": 7.839907882177222e-06, "loss": 1.3326, "step": 23289 }, { "epoch": 58.225, "grad_norm": 9.8125, "learning_rate": 7.839117316110608e-06, "loss": 1.279, "step": 23290 }, { "epoch": 58.2275, "grad_norm": 6.9375, "learning_rate": 7.838326764210654e-06, "loss": 1.2721, "step": 23291 }, { "epoch": 58.23, "grad_norm": 5.9375, "learning_rate": 7.83753622648254e-06, "loss": 1.268, "step": 23292 }, { "epoch": 58.2325, "grad_norm": 9.0625, "learning_rate": 7.836745702931457e-06, "loss": 1.4152, "step": 23293 }, { "epoch": 58.235, "grad_norm": 8.5625, "learning_rate": 7.835955193562581e-06, "loss": 1.2966, "step": 23294 }, { "epoch": 58.2375, "grad_norm": 6.40625, "learning_rate": 7.835164698381097e-06, "loss": 1.5695, "step": 23295 }, { "epoch": 58.24, "grad_norm": 7.78125, "learning_rate": 7.834374217392188e-06, "loss": 1.3214, "step": 23296 }, { "epoch": 58.2425, "grad_norm": 8.0, "learning_rate": 7.833583750601029e-06, "loss": 1.1161, "step": 23297 }, { "epoch": 58.245, "grad_norm": 6.125, "learning_rate": 7.832793298012813e-06, "loss": 1.4133, "step": 23298 }, { "epoch": 58.2475, "grad_norm": 7.875, "learning_rate": 7.832002859632718e-06, "loss": 1.1687, "step": 23299 }, { "epoch": 58.25, "grad_norm": 7.0625, "learning_rate": 7.831212435465925e-06, "loss": 1.3987, "step": 23300 }, { "epoch": 58.2525, "grad_norm": 7.25, "learning_rate": 7.830422025517615e-06, "loss": 1.375, "step": 23301 }, { "epoch": 58.255, "grad_norm": 8.6875, "learning_rate": 7.82963162979297e-06, "loss": 1.434, "step": 23302 }, { "epoch": 58.2575, "grad_norm": 11.0, "learning_rate": 7.828841248297174e-06, "loss": 1.1336, "step": 23303 }, { "epoch": 58.26, "grad_norm": 6.40625, "learning_rate": 7.828050881035412e-06, "loss": 1.3456, "step": 23304 }, { "epoch": 58.2625, "grad_norm": 5.3125, "learning_rate": 7.827260528012858e-06, "loss": 1.3472, "step": 23305 }, { "epoch": 58.265, "grad_norm": 6.15625, "learning_rate": 7.826470189234697e-06, "loss": 1.271, "step": 23306 }, { "epoch": 58.2675, "grad_norm": 21.125, "learning_rate": 7.825679864706108e-06, "loss": 1.4245, "step": 23307 }, { "epoch": 58.27, "grad_norm": 9.8125, "learning_rate": 7.82488955443228e-06, "loss": 1.3086, "step": 23308 }, { "epoch": 58.2725, "grad_norm": 4.3125, "learning_rate": 7.824099258418388e-06, "loss": 1.3979, "step": 23309 }, { "epoch": 58.275, "grad_norm": 6.75, "learning_rate": 7.82330897666961e-06, "loss": 1.2901, "step": 23310 }, { "epoch": 58.2775, "grad_norm": 10.0625, "learning_rate": 7.822518709191138e-06, "loss": 1.4764, "step": 23311 }, { "epoch": 58.28, "grad_norm": 9.25, "learning_rate": 7.82172845598814e-06, "loss": 1.2336, "step": 23312 }, { "epoch": 58.2825, "grad_norm": 10.0, "learning_rate": 7.820938217065805e-06, "loss": 1.3372, "step": 23313 }, { "epoch": 58.285, "grad_norm": 7.875, "learning_rate": 7.820147992429312e-06, "loss": 1.2634, "step": 23314 }, { "epoch": 58.2875, "grad_norm": 10.5, "learning_rate": 7.819357782083844e-06, "loss": 1.3748, "step": 23315 }, { "epoch": 58.29, "grad_norm": 5.9375, "learning_rate": 7.818567586034578e-06, "loss": 1.2617, "step": 23316 }, { "epoch": 58.2925, "grad_norm": 6.84375, "learning_rate": 7.817777404286692e-06, "loss": 1.4036, "step": 23317 }, { "epoch": 58.295, "grad_norm": 8.5, "learning_rate": 7.816987236845372e-06, "loss": 1.3562, "step": 23318 }, { "epoch": 58.2975, "grad_norm": 14.0625, "learning_rate": 7.816197083715798e-06, "loss": 1.2667, "step": 23319 }, { "epoch": 58.3, "grad_norm": 8.0625, "learning_rate": 7.815406944903148e-06, "loss": 1.3128, "step": 23320 }, { "epoch": 58.3025, "grad_norm": 6.40625, "learning_rate": 7.814616820412601e-06, "loss": 1.2152, "step": 23321 }, { "epoch": 58.305, "grad_norm": 9.0625, "learning_rate": 7.813826710249338e-06, "loss": 1.392, "step": 23322 }, { "epoch": 58.3075, "grad_norm": 8.6875, "learning_rate": 7.813036614418543e-06, "loss": 1.3004, "step": 23323 }, { "epoch": 58.31, "grad_norm": 8.3125, "learning_rate": 7.812246532925392e-06, "loss": 1.399, "step": 23324 }, { "epoch": 58.3125, "grad_norm": 7.09375, "learning_rate": 7.811456465775064e-06, "loss": 1.3731, "step": 23325 }, { "epoch": 58.315, "grad_norm": 6.75, "learning_rate": 7.810666412972741e-06, "loss": 1.3909, "step": 23326 }, { "epoch": 58.3175, "grad_norm": 5.78125, "learning_rate": 7.809876374523597e-06, "loss": 1.3543, "step": 23327 }, { "epoch": 58.32, "grad_norm": 9.5625, "learning_rate": 7.80908635043282e-06, "loss": 1.5036, "step": 23328 }, { "epoch": 58.3225, "grad_norm": 7.4375, "learning_rate": 7.808296340705583e-06, "loss": 1.4378, "step": 23329 }, { "epoch": 58.325, "grad_norm": 7.5625, "learning_rate": 7.807506345347072e-06, "loss": 1.3607, "step": 23330 }, { "epoch": 58.3275, "grad_norm": 7.75, "learning_rate": 7.806716364362457e-06, "loss": 1.3572, "step": 23331 }, { "epoch": 58.33, "grad_norm": 10.0625, "learning_rate": 7.805926397756921e-06, "loss": 1.2606, "step": 23332 }, { "epoch": 58.3325, "grad_norm": 6.21875, "learning_rate": 7.805136445535647e-06, "loss": 1.2626, "step": 23333 }, { "epoch": 58.335, "grad_norm": 6.5625, "learning_rate": 7.80434650770381e-06, "loss": 1.3435, "step": 23334 }, { "epoch": 58.3375, "grad_norm": 6.40625, "learning_rate": 7.80355658426659e-06, "loss": 1.4984, "step": 23335 }, { "epoch": 58.34, "grad_norm": 7.71875, "learning_rate": 7.802766675229166e-06, "loss": 1.2723, "step": 23336 }, { "epoch": 58.3425, "grad_norm": 8.0625, "learning_rate": 7.80197678059671e-06, "loss": 1.2115, "step": 23337 }, { "epoch": 58.345, "grad_norm": 11.8125, "learning_rate": 7.801186900374411e-06, "loss": 1.2882, "step": 23338 }, { "epoch": 58.3475, "grad_norm": 8.25, "learning_rate": 7.800397034567443e-06, "loss": 1.3927, "step": 23339 }, { "epoch": 58.35, "grad_norm": 6.46875, "learning_rate": 7.799607183180981e-06, "loss": 1.337, "step": 23340 }, { "epoch": 58.3525, "grad_norm": 6.15625, "learning_rate": 7.798817346220211e-06, "loss": 1.1969, "step": 23341 }, { "epoch": 58.355, "grad_norm": 9.75, "learning_rate": 7.798027523690301e-06, "loss": 1.192, "step": 23342 }, { "epoch": 58.3575, "grad_norm": 6.46875, "learning_rate": 7.797237715596436e-06, "loss": 1.4459, "step": 23343 }, { "epoch": 58.36, "grad_norm": 9.4375, "learning_rate": 7.796447921943793e-06, "loss": 1.4211, "step": 23344 }, { "epoch": 58.3625, "grad_norm": 9.6875, "learning_rate": 7.795658142737553e-06, "loss": 1.4135, "step": 23345 }, { "epoch": 58.365, "grad_norm": 7.15625, "learning_rate": 7.794868377982886e-06, "loss": 1.255, "step": 23346 }, { "epoch": 58.3675, "grad_norm": 6.75, "learning_rate": 7.794078627684972e-06, "loss": 1.4528, "step": 23347 }, { "epoch": 58.37, "grad_norm": 6.90625, "learning_rate": 7.79328889184899e-06, "loss": 1.3658, "step": 23348 }, { "epoch": 58.3725, "grad_norm": 6.90625, "learning_rate": 7.792499170480123e-06, "loss": 1.3611, "step": 23349 }, { "epoch": 58.375, "grad_norm": 6.3125, "learning_rate": 7.791709463583541e-06, "loss": 1.3883, "step": 23350 }, { "epoch": 58.3775, "grad_norm": 5.0625, "learning_rate": 7.790919771164425e-06, "loss": 1.2963, "step": 23351 }, { "epoch": 58.38, "grad_norm": 8.5625, "learning_rate": 7.790130093227943e-06, "loss": 1.3991, "step": 23352 }, { "epoch": 58.3825, "grad_norm": 5.78125, "learning_rate": 7.789340429779286e-06, "loss": 1.2588, "step": 23353 }, { "epoch": 58.385, "grad_norm": 6.6875, "learning_rate": 7.788550780823624e-06, "loss": 1.4024, "step": 23354 }, { "epoch": 58.3875, "grad_norm": 9.4375, "learning_rate": 7.787761146366136e-06, "loss": 1.43, "step": 23355 }, { "epoch": 58.39, "grad_norm": 6.0, "learning_rate": 7.786971526411996e-06, "loss": 1.3887, "step": 23356 }, { "epoch": 58.3925, "grad_norm": 5.0, "learning_rate": 7.786181920966379e-06, "loss": 1.2826, "step": 23357 }, { "epoch": 58.395, "grad_norm": 8.375, "learning_rate": 7.785392330034467e-06, "loss": 1.3796, "step": 23358 }, { "epoch": 58.3975, "grad_norm": 6.40625, "learning_rate": 7.784602753621436e-06, "loss": 1.4096, "step": 23359 }, { "epoch": 58.4, "grad_norm": 7.6875, "learning_rate": 7.78381319173246e-06, "loss": 1.1917, "step": 23360 }, { "epoch": 58.4025, "grad_norm": 5.6875, "learning_rate": 7.783023644372714e-06, "loss": 1.2139, "step": 23361 }, { "epoch": 58.405, "grad_norm": 9.5625, "learning_rate": 7.782234111547376e-06, "loss": 1.4502, "step": 23362 }, { "epoch": 58.4075, "grad_norm": 6.21875, "learning_rate": 7.781444593261627e-06, "loss": 1.3374, "step": 23363 }, { "epoch": 58.41, "grad_norm": 5.84375, "learning_rate": 7.780655089520634e-06, "loss": 1.2785, "step": 23364 }, { "epoch": 58.4125, "grad_norm": 8.125, "learning_rate": 7.77986560032958e-06, "loss": 1.3377, "step": 23365 }, { "epoch": 58.415, "grad_norm": 15.25, "learning_rate": 7.779076125693639e-06, "loss": 1.4532, "step": 23366 }, { "epoch": 58.4175, "grad_norm": 7.9375, "learning_rate": 7.77828666561798e-06, "loss": 1.3679, "step": 23367 }, { "epoch": 58.42, "grad_norm": 9.25, "learning_rate": 7.777497220107791e-06, "loss": 1.2354, "step": 23368 }, { "epoch": 58.4225, "grad_norm": 9.0, "learning_rate": 7.77670778916824e-06, "loss": 1.2983, "step": 23369 }, { "epoch": 58.425, "grad_norm": 8.0, "learning_rate": 7.775918372804505e-06, "loss": 1.2289, "step": 23370 }, { "epoch": 58.4275, "grad_norm": 9.8125, "learning_rate": 7.775128971021757e-06, "loss": 1.3845, "step": 23371 }, { "epoch": 58.43, "grad_norm": 7.25, "learning_rate": 7.774339583825173e-06, "loss": 1.2742, "step": 23372 }, { "epoch": 58.4325, "grad_norm": 6.90625, "learning_rate": 7.773550211219932e-06, "loss": 1.4653, "step": 23373 }, { "epoch": 58.435, "grad_norm": 8.8125, "learning_rate": 7.772760853211209e-06, "loss": 1.3543, "step": 23374 }, { "epoch": 58.4375, "grad_norm": 6.8125, "learning_rate": 7.771971509804174e-06, "loss": 1.3537, "step": 23375 }, { "epoch": 58.44, "grad_norm": 7.75, "learning_rate": 7.771182181004005e-06, "loss": 1.2373, "step": 23376 }, { "epoch": 58.4425, "grad_norm": 10.375, "learning_rate": 7.770392866815875e-06, "loss": 1.3629, "step": 23377 }, { "epoch": 58.445, "grad_norm": 7.96875, "learning_rate": 7.769603567244964e-06, "loss": 1.3643, "step": 23378 }, { "epoch": 58.4475, "grad_norm": 6.875, "learning_rate": 7.76881428229644e-06, "loss": 1.376, "step": 23379 }, { "epoch": 58.45, "grad_norm": 8.375, "learning_rate": 7.768025011975481e-06, "loss": 1.3493, "step": 23380 }, { "epoch": 58.4525, "grad_norm": 6.59375, "learning_rate": 7.767235756287263e-06, "loss": 1.3735, "step": 23381 }, { "epoch": 58.455, "grad_norm": 6.78125, "learning_rate": 7.766446515236953e-06, "loss": 1.3245, "step": 23382 }, { "epoch": 58.4575, "grad_norm": 7.78125, "learning_rate": 7.765657288829732e-06, "loss": 1.3524, "step": 23383 }, { "epoch": 58.46, "grad_norm": 8.125, "learning_rate": 7.764868077070772e-06, "loss": 1.4047, "step": 23384 }, { "epoch": 58.4625, "grad_norm": 8.75, "learning_rate": 7.764078879965252e-06, "loss": 1.396, "step": 23385 }, { "epoch": 58.465, "grad_norm": 8.75, "learning_rate": 7.763289697518338e-06, "loss": 1.3133, "step": 23386 }, { "epoch": 58.4675, "grad_norm": 6.28125, "learning_rate": 7.762500529735205e-06, "loss": 1.3629, "step": 23387 }, { "epoch": 58.47, "grad_norm": 7.6875, "learning_rate": 7.761711376621032e-06, "loss": 1.3503, "step": 23388 }, { "epoch": 58.4725, "grad_norm": 6.78125, "learning_rate": 7.76092223818099e-06, "loss": 1.3872, "step": 23389 }, { "epoch": 58.475, "grad_norm": 10.5, "learning_rate": 7.760133114420252e-06, "loss": 1.4195, "step": 23390 }, { "epoch": 58.4775, "grad_norm": 7.40625, "learning_rate": 7.759344005343988e-06, "loss": 1.2427, "step": 23391 }, { "epoch": 58.48, "grad_norm": 8.3125, "learning_rate": 7.758554910957378e-06, "loss": 1.347, "step": 23392 }, { "epoch": 58.4825, "grad_norm": 26.0, "learning_rate": 7.757765831265593e-06, "loss": 1.5163, "step": 23393 }, { "epoch": 58.485, "grad_norm": 9.0625, "learning_rate": 7.756976766273805e-06, "loss": 1.5179, "step": 23394 }, { "epoch": 58.4875, "grad_norm": 5.96875, "learning_rate": 7.756187715987187e-06, "loss": 1.4316, "step": 23395 }, { "epoch": 58.49, "grad_norm": 8.25, "learning_rate": 7.75539868041091e-06, "loss": 1.3777, "step": 23396 }, { "epoch": 58.4925, "grad_norm": 7.125, "learning_rate": 7.754609659550155e-06, "loss": 1.3747, "step": 23397 }, { "epoch": 58.495, "grad_norm": 7.15625, "learning_rate": 7.753820653410086e-06, "loss": 1.342, "step": 23398 }, { "epoch": 58.4975, "grad_norm": 10.5, "learning_rate": 7.75303166199588e-06, "loss": 1.2308, "step": 23399 }, { "epoch": 58.5, "grad_norm": 9.5, "learning_rate": 7.752242685312709e-06, "loss": 1.3985, "step": 23400 }, { "epoch": 58.5025, "grad_norm": 11.6875, "learning_rate": 7.751453723365742e-06, "loss": 1.1566, "step": 23401 }, { "epoch": 58.505, "grad_norm": 8.625, "learning_rate": 7.750664776160156e-06, "loss": 1.3315, "step": 23402 }, { "epoch": 58.5075, "grad_norm": 6.65625, "learning_rate": 7.749875843701121e-06, "loss": 1.3999, "step": 23403 }, { "epoch": 58.51, "grad_norm": 14.5625, "learning_rate": 7.749086925993814e-06, "loss": 1.1434, "step": 23404 }, { "epoch": 58.5125, "grad_norm": 8.3125, "learning_rate": 7.748298023043399e-06, "loss": 1.3465, "step": 23405 }, { "epoch": 58.515, "grad_norm": 8.625, "learning_rate": 7.747509134855051e-06, "loss": 1.4989, "step": 23406 }, { "epoch": 58.5175, "grad_norm": 7.875, "learning_rate": 7.746720261433944e-06, "loss": 1.4295, "step": 23407 }, { "epoch": 58.52, "grad_norm": 9.875, "learning_rate": 7.745931402785252e-06, "loss": 1.3535, "step": 23408 }, { "epoch": 58.5225, "grad_norm": 7.625, "learning_rate": 7.745142558914141e-06, "loss": 1.2908, "step": 23409 }, { "epoch": 58.525, "grad_norm": 8.4375, "learning_rate": 7.744353729825786e-06, "loss": 1.4322, "step": 23410 }, { "epoch": 58.5275, "grad_norm": 8.75, "learning_rate": 7.743564915525354e-06, "loss": 1.1725, "step": 23411 }, { "epoch": 58.53, "grad_norm": 9.375, "learning_rate": 7.742776116018024e-06, "loss": 1.3071, "step": 23412 }, { "epoch": 58.5325, "grad_norm": 6.71875, "learning_rate": 7.741987331308964e-06, "loss": 1.2104, "step": 23413 }, { "epoch": 58.535, "grad_norm": 7.875, "learning_rate": 7.741198561403342e-06, "loss": 1.3572, "step": 23414 }, { "epoch": 58.5375, "grad_norm": 7.1875, "learning_rate": 7.740409806306335e-06, "loss": 1.208, "step": 23415 }, { "epoch": 58.54, "grad_norm": 6.78125, "learning_rate": 7.739621066023108e-06, "loss": 1.2742, "step": 23416 }, { "epoch": 58.5425, "grad_norm": 7.75, "learning_rate": 7.738832340558836e-06, "loss": 1.211, "step": 23417 }, { "epoch": 58.545, "grad_norm": 8.6875, "learning_rate": 7.738043629918688e-06, "loss": 1.3398, "step": 23418 }, { "epoch": 58.5475, "grad_norm": 6.0625, "learning_rate": 7.737254934107838e-06, "loss": 1.1379, "step": 23419 }, { "epoch": 58.55, "grad_norm": 7.34375, "learning_rate": 7.736466253131451e-06, "loss": 1.1982, "step": 23420 }, { "epoch": 58.5525, "grad_norm": 7.3125, "learning_rate": 7.735677586994701e-06, "loss": 1.1592, "step": 23421 }, { "epoch": 58.555, "grad_norm": 7.5, "learning_rate": 7.734888935702758e-06, "loss": 1.219, "step": 23422 }, { "epoch": 58.5575, "grad_norm": 7.40625, "learning_rate": 7.734100299260797e-06, "loss": 1.4594, "step": 23423 }, { "epoch": 58.56, "grad_norm": 10.0, "learning_rate": 7.733311677673979e-06, "loss": 1.3124, "step": 23424 }, { "epoch": 58.5625, "grad_norm": 7.375, "learning_rate": 7.732523070947482e-06, "loss": 1.3937, "step": 23425 }, { "epoch": 58.565, "grad_norm": 10.0, "learning_rate": 7.731734479086467e-06, "loss": 1.3649, "step": 23426 }, { "epoch": 58.5675, "grad_norm": 8.375, "learning_rate": 7.730945902096117e-06, "loss": 1.2524, "step": 23427 }, { "epoch": 58.57, "grad_norm": 6.40625, "learning_rate": 7.730157339981591e-06, "loss": 1.3115, "step": 23428 }, { "epoch": 58.5725, "grad_norm": 8.0625, "learning_rate": 7.729368792748066e-06, "loss": 1.3827, "step": 23429 }, { "epoch": 58.575, "grad_norm": 9.9375, "learning_rate": 7.728580260400706e-06, "loss": 1.3666, "step": 23430 }, { "epoch": 58.5775, "grad_norm": 7.8125, "learning_rate": 7.72779174294468e-06, "loss": 1.3179, "step": 23431 }, { "epoch": 58.58, "grad_norm": 6.0625, "learning_rate": 7.727003240385163e-06, "loss": 1.3894, "step": 23432 }, { "epoch": 58.5825, "grad_norm": 9.6875, "learning_rate": 7.726214752727323e-06, "loss": 1.2476, "step": 23433 }, { "epoch": 58.585, "grad_norm": 7.96875, "learning_rate": 7.725426279976327e-06, "loss": 1.2452, "step": 23434 }, { "epoch": 58.5875, "grad_norm": 7.46875, "learning_rate": 7.724637822137346e-06, "loss": 1.407, "step": 23435 }, { "epoch": 58.59, "grad_norm": 7.6875, "learning_rate": 7.723849379215545e-06, "loss": 1.246, "step": 23436 }, { "epoch": 58.5925, "grad_norm": 7.9375, "learning_rate": 7.723060951216099e-06, "loss": 1.2423, "step": 23437 }, { "epoch": 58.595, "grad_norm": 8.75, "learning_rate": 7.722272538144178e-06, "loss": 1.2788, "step": 23438 }, { "epoch": 58.5975, "grad_norm": 9.3125, "learning_rate": 7.721484140004943e-06, "loss": 1.4099, "step": 23439 }, { "epoch": 58.6, "grad_norm": 10.5, "learning_rate": 7.720695756803569e-06, "loss": 1.4146, "step": 23440 }, { "epoch": 58.6025, "grad_norm": 8.4375, "learning_rate": 7.719907388545215e-06, "loss": 1.2486, "step": 23441 }, { "epoch": 58.605, "grad_norm": 7.0, "learning_rate": 7.719119035235065e-06, "loss": 1.4266, "step": 23442 }, { "epoch": 58.6075, "grad_norm": 7.9375, "learning_rate": 7.718330696878277e-06, "loss": 1.4294, "step": 23443 }, { "epoch": 58.61, "grad_norm": 7.90625, "learning_rate": 7.717542373480025e-06, "loss": 1.3605, "step": 23444 }, { "epoch": 58.6125, "grad_norm": 9.4375, "learning_rate": 7.71675406504547e-06, "loss": 1.221, "step": 23445 }, { "epoch": 58.615, "grad_norm": 10.625, "learning_rate": 7.715965771579782e-06, "loss": 1.3393, "step": 23446 }, { "epoch": 58.6175, "grad_norm": 5.71875, "learning_rate": 7.715177493088134e-06, "loss": 1.3505, "step": 23447 }, { "epoch": 58.62, "grad_norm": 8.9375, "learning_rate": 7.714389229575692e-06, "loss": 1.3944, "step": 23448 }, { "epoch": 58.6225, "grad_norm": 7.8125, "learning_rate": 7.71360098104762e-06, "loss": 1.2883, "step": 23449 }, { "epoch": 58.625, "grad_norm": 8.0, "learning_rate": 7.712812747509091e-06, "loss": 1.2084, "step": 23450 }, { "epoch": 58.6275, "grad_norm": 8.125, "learning_rate": 7.712024528965264e-06, "loss": 1.3899, "step": 23451 }, { "epoch": 58.63, "grad_norm": 5.6875, "learning_rate": 7.71123632542132e-06, "loss": 1.4023, "step": 23452 }, { "epoch": 58.6325, "grad_norm": 6.65625, "learning_rate": 7.710448136882418e-06, "loss": 1.2246, "step": 23453 }, { "epoch": 58.635, "grad_norm": 11.6875, "learning_rate": 7.709659963353723e-06, "loss": 1.3799, "step": 23454 }, { "epoch": 58.6375, "grad_norm": 5.03125, "learning_rate": 7.70887180484041e-06, "loss": 1.3335, "step": 23455 }, { "epoch": 58.64, "grad_norm": 9.9375, "learning_rate": 7.708083661347637e-06, "loss": 1.2989, "step": 23456 }, { "epoch": 58.6425, "grad_norm": 8.875, "learning_rate": 7.707295532880578e-06, "loss": 1.2308, "step": 23457 }, { "epoch": 58.645, "grad_norm": 8.6875, "learning_rate": 7.706507419444397e-06, "loss": 1.2362, "step": 23458 }, { "epoch": 58.6475, "grad_norm": 7.09375, "learning_rate": 7.705719321044264e-06, "loss": 1.3336, "step": 23459 }, { "epoch": 58.65, "grad_norm": 10.9375, "learning_rate": 7.704931237685342e-06, "loss": 1.3545, "step": 23460 }, { "epoch": 58.6525, "grad_norm": 6.1875, "learning_rate": 7.704143169372797e-06, "loss": 1.2866, "step": 23461 }, { "epoch": 58.655, "grad_norm": 8.1875, "learning_rate": 7.7033551161118e-06, "loss": 1.3095, "step": 23462 }, { "epoch": 58.6575, "grad_norm": 6.875, "learning_rate": 7.702567077907515e-06, "loss": 1.346, "step": 23463 }, { "epoch": 58.66, "grad_norm": 6.9375, "learning_rate": 7.701779054765107e-06, "loss": 1.2869, "step": 23464 }, { "epoch": 58.6625, "grad_norm": 6.28125, "learning_rate": 7.700991046689744e-06, "loss": 1.272, "step": 23465 }, { "epoch": 58.665, "grad_norm": 10.875, "learning_rate": 7.70020305368659e-06, "loss": 1.4044, "step": 23466 }, { "epoch": 58.6675, "grad_norm": 6.8125, "learning_rate": 7.699415075760816e-06, "loss": 1.2323, "step": 23467 }, { "epoch": 58.67, "grad_norm": 12.0, "learning_rate": 7.698627112917583e-06, "loss": 1.3836, "step": 23468 }, { "epoch": 58.6725, "grad_norm": 7.5625, "learning_rate": 7.697839165162058e-06, "loss": 1.4145, "step": 23469 }, { "epoch": 58.675, "grad_norm": 5.75, "learning_rate": 7.69705123249941e-06, "loss": 1.2569, "step": 23470 }, { "epoch": 58.6775, "grad_norm": 7.21875, "learning_rate": 7.696263314934798e-06, "loss": 1.2483, "step": 23471 }, { "epoch": 58.68, "grad_norm": 9.9375, "learning_rate": 7.695475412473393e-06, "loss": 1.3614, "step": 23472 }, { "epoch": 58.6825, "grad_norm": 12.75, "learning_rate": 7.69468752512036e-06, "loss": 1.4161, "step": 23473 }, { "epoch": 58.685, "grad_norm": 5.3125, "learning_rate": 7.693899652880864e-06, "loss": 1.4107, "step": 23474 }, { "epoch": 58.6875, "grad_norm": 7.1875, "learning_rate": 7.693111795760067e-06, "loss": 1.5214, "step": 23475 }, { "epoch": 58.69, "grad_norm": 10.8125, "learning_rate": 7.692323953763136e-06, "loss": 1.4248, "step": 23476 }, { "epoch": 58.6925, "grad_norm": 9.0, "learning_rate": 7.69153612689524e-06, "loss": 1.2154, "step": 23477 }, { "epoch": 58.695, "grad_norm": 7.0, "learning_rate": 7.690748315161541e-06, "loss": 1.3057, "step": 23478 }, { "epoch": 58.6975, "grad_norm": 5.34375, "learning_rate": 7.689960518567202e-06, "loss": 1.3345, "step": 23479 }, { "epoch": 58.7, "grad_norm": 11.3125, "learning_rate": 7.689172737117389e-06, "loss": 1.5956, "step": 23480 }, { "epoch": 58.7025, "grad_norm": 7.8125, "learning_rate": 7.688384970817267e-06, "loss": 1.3262, "step": 23481 }, { "epoch": 58.705, "grad_norm": 10.75, "learning_rate": 7.687597219672002e-06, "loss": 1.2845, "step": 23482 }, { "epoch": 58.7075, "grad_norm": 10.375, "learning_rate": 7.686809483686757e-06, "loss": 1.3329, "step": 23483 }, { "epoch": 58.71, "grad_norm": 6.78125, "learning_rate": 7.686021762866695e-06, "loss": 1.2968, "step": 23484 }, { "epoch": 58.7125, "grad_norm": 13.25, "learning_rate": 7.685234057216984e-06, "loss": 1.4454, "step": 23485 }, { "epoch": 58.715, "grad_norm": 5.8125, "learning_rate": 7.684446366742782e-06, "loss": 1.328, "step": 23486 }, { "epoch": 58.7175, "grad_norm": 9.0, "learning_rate": 7.68365869144926e-06, "loss": 1.2455, "step": 23487 }, { "epoch": 58.72, "grad_norm": 6.59375, "learning_rate": 7.682871031341579e-06, "loss": 1.196, "step": 23488 }, { "epoch": 58.7225, "grad_norm": 6.125, "learning_rate": 7.682083386424903e-06, "loss": 1.1785, "step": 23489 }, { "epoch": 58.725, "grad_norm": 7.125, "learning_rate": 7.681295756704395e-06, "loss": 1.401, "step": 23490 }, { "epoch": 58.7275, "grad_norm": 8.0625, "learning_rate": 7.680508142185216e-06, "loss": 1.2397, "step": 23491 }, { "epoch": 58.73, "grad_norm": 8.125, "learning_rate": 7.679720542872537e-06, "loss": 1.2459, "step": 23492 }, { "epoch": 58.7325, "grad_norm": 6.78125, "learning_rate": 7.678932958771517e-06, "loss": 1.4176, "step": 23493 }, { "epoch": 58.735, "grad_norm": 8.0625, "learning_rate": 7.678145389887319e-06, "loss": 1.2424, "step": 23494 }, { "epoch": 58.7375, "grad_norm": 6.75, "learning_rate": 7.67735783622511e-06, "loss": 1.2934, "step": 23495 }, { "epoch": 58.74, "grad_norm": 15.3125, "learning_rate": 7.676570297790042e-06, "loss": 1.4222, "step": 23496 }, { "epoch": 58.7425, "grad_norm": 10.1875, "learning_rate": 7.675782774587291e-06, "loss": 1.3588, "step": 23497 }, { "epoch": 58.745, "grad_norm": 5.59375, "learning_rate": 7.674995266622016e-06, "loss": 1.0731, "step": 23498 }, { "epoch": 58.7475, "grad_norm": 8.1875, "learning_rate": 7.67420777389938e-06, "loss": 1.4167, "step": 23499 }, { "epoch": 58.75, "grad_norm": 10.625, "learning_rate": 7.673420296424541e-06, "loss": 1.4211, "step": 23500 }, { "epoch": 58.7525, "grad_norm": 7.40625, "learning_rate": 7.672632834202666e-06, "loss": 1.2045, "step": 23501 }, { "epoch": 58.755, "grad_norm": 7.5625, "learning_rate": 7.671845387238918e-06, "loss": 1.2413, "step": 23502 }, { "epoch": 58.7575, "grad_norm": 7.28125, "learning_rate": 7.67105795553846e-06, "loss": 1.4199, "step": 23503 }, { "epoch": 58.76, "grad_norm": 5.65625, "learning_rate": 7.670270539106452e-06, "loss": 1.2933, "step": 23504 }, { "epoch": 58.7625, "grad_norm": 9.75, "learning_rate": 7.669483137948055e-06, "loss": 1.3407, "step": 23505 }, { "epoch": 58.765, "grad_norm": 5.5, "learning_rate": 7.668695752068434e-06, "loss": 1.4434, "step": 23506 }, { "epoch": 58.7675, "grad_norm": 8.5625, "learning_rate": 7.66790838147275e-06, "loss": 1.4756, "step": 23507 }, { "epoch": 58.77, "grad_norm": 6.8125, "learning_rate": 7.667121026166168e-06, "loss": 1.3477, "step": 23508 }, { "epoch": 58.7725, "grad_norm": 7.6875, "learning_rate": 7.666333686153847e-06, "loss": 1.3478, "step": 23509 }, { "epoch": 58.775, "grad_norm": 9.1875, "learning_rate": 7.66554636144095e-06, "loss": 1.3105, "step": 23510 }, { "epoch": 58.7775, "grad_norm": 7.90625, "learning_rate": 7.66475905203263e-06, "loss": 1.2937, "step": 23511 }, { "epoch": 58.78, "grad_norm": 6.59375, "learning_rate": 7.663971757934064e-06, "loss": 1.218, "step": 23512 }, { "epoch": 58.7825, "grad_norm": 16.5, "learning_rate": 7.663184479150403e-06, "loss": 1.3941, "step": 23513 }, { "epoch": 58.785, "grad_norm": 8.0625, "learning_rate": 7.662397215686813e-06, "loss": 1.3241, "step": 23514 }, { "epoch": 58.7875, "grad_norm": 8.25, "learning_rate": 7.661609967548452e-06, "loss": 1.2903, "step": 23515 }, { "epoch": 58.79, "grad_norm": 6.0, "learning_rate": 7.660822734740478e-06, "loss": 1.2855, "step": 23516 }, { "epoch": 58.7925, "grad_norm": 7.375, "learning_rate": 7.660035517268063e-06, "loss": 1.3642, "step": 23517 }, { "epoch": 58.795, "grad_norm": 6.09375, "learning_rate": 7.659248315136359e-06, "loss": 1.4294, "step": 23518 }, { "epoch": 58.7975, "grad_norm": 7.0625, "learning_rate": 7.65846112835053e-06, "loss": 1.2545, "step": 23519 }, { "epoch": 58.8, "grad_norm": 7.0, "learning_rate": 7.657673956915735e-06, "loss": 1.3149, "step": 23520 }, { "epoch": 58.8025, "grad_norm": 6.75, "learning_rate": 7.656886800837135e-06, "loss": 1.3204, "step": 23521 }, { "epoch": 58.805, "grad_norm": 10.4375, "learning_rate": 7.656099660119894e-06, "loss": 1.3633, "step": 23522 }, { "epoch": 58.8075, "grad_norm": 6.4375, "learning_rate": 7.655312534769168e-06, "loss": 1.4197, "step": 23523 }, { "epoch": 58.81, "grad_norm": 9.125, "learning_rate": 7.654525424790121e-06, "loss": 1.4809, "step": 23524 }, { "epoch": 58.8125, "grad_norm": 6.625, "learning_rate": 7.653738330187912e-06, "loss": 1.364, "step": 23525 }, { "epoch": 58.815, "grad_norm": 6.5, "learning_rate": 7.652951250967696e-06, "loss": 1.2811, "step": 23526 }, { "epoch": 58.8175, "grad_norm": 8.9375, "learning_rate": 7.652164187134641e-06, "loss": 1.3537, "step": 23527 }, { "epoch": 58.82, "grad_norm": 8.0, "learning_rate": 7.651377138693903e-06, "loss": 1.3436, "step": 23528 }, { "epoch": 58.8225, "grad_norm": 5.46875, "learning_rate": 7.650590105650643e-06, "loss": 1.1985, "step": 23529 }, { "epoch": 58.825, "grad_norm": 10.4375, "learning_rate": 7.64980308801002e-06, "loss": 1.1846, "step": 23530 }, { "epoch": 58.8275, "grad_norm": 6.0, "learning_rate": 7.649016085777191e-06, "loss": 1.3191, "step": 23531 }, { "epoch": 58.83, "grad_norm": 6.3125, "learning_rate": 7.648229098957323e-06, "loss": 1.2817, "step": 23532 }, { "epoch": 58.8325, "grad_norm": 5.96875, "learning_rate": 7.647442127555571e-06, "loss": 1.4704, "step": 23533 }, { "epoch": 58.835, "grad_norm": 6.4375, "learning_rate": 7.646655171577092e-06, "loss": 1.2323, "step": 23534 }, { "epoch": 58.8375, "grad_norm": 8.125, "learning_rate": 7.645868231027047e-06, "loss": 1.3487, "step": 23535 }, { "epoch": 58.84, "grad_norm": 8.75, "learning_rate": 7.645081305910596e-06, "loss": 1.4929, "step": 23536 }, { "epoch": 58.8425, "grad_norm": 7.9375, "learning_rate": 7.644294396232898e-06, "loss": 1.1675, "step": 23537 }, { "epoch": 58.845, "grad_norm": 5.96875, "learning_rate": 7.643507501999114e-06, "loss": 1.3441, "step": 23538 }, { "epoch": 58.8475, "grad_norm": 7.9375, "learning_rate": 7.642720623214396e-06, "loss": 1.3631, "step": 23539 }, { "epoch": 58.85, "grad_norm": 10.25, "learning_rate": 7.641933759883913e-06, "loss": 1.318, "step": 23540 }, { "epoch": 58.8525, "grad_norm": 7.34375, "learning_rate": 7.641146912012812e-06, "loss": 1.2402, "step": 23541 }, { "epoch": 58.855, "grad_norm": 7.3125, "learning_rate": 7.64036007960626e-06, "loss": 1.329, "step": 23542 }, { "epoch": 58.8575, "grad_norm": 6.84375, "learning_rate": 7.639573262669413e-06, "loss": 1.2971, "step": 23543 }, { "epoch": 58.86, "grad_norm": 6.125, "learning_rate": 7.638786461207433e-06, "loss": 1.2987, "step": 23544 }, { "epoch": 58.8625, "grad_norm": 7.40625, "learning_rate": 7.63799967522547e-06, "loss": 1.2884, "step": 23545 }, { "epoch": 58.865, "grad_norm": 8.1875, "learning_rate": 7.637212904728684e-06, "loss": 1.207, "step": 23546 }, { "epoch": 58.8675, "grad_norm": 9.9375, "learning_rate": 7.63642614972224e-06, "loss": 1.1989, "step": 23547 }, { "epoch": 58.87, "grad_norm": 7.78125, "learning_rate": 7.635639410211292e-06, "loss": 1.2414, "step": 23548 }, { "epoch": 58.8725, "grad_norm": 6.75, "learning_rate": 7.634852686200995e-06, "loss": 1.2953, "step": 23549 }, { "epoch": 58.875, "grad_norm": 9.25, "learning_rate": 7.63406597769651e-06, "loss": 1.397, "step": 23550 }, { "epoch": 58.8775, "grad_norm": 7.21875, "learning_rate": 7.633279284702993e-06, "loss": 1.3699, "step": 23551 }, { "epoch": 58.88, "grad_norm": 5.625, "learning_rate": 7.632492607225604e-06, "loss": 1.2738, "step": 23552 }, { "epoch": 58.8825, "grad_norm": 6.34375, "learning_rate": 7.631705945269498e-06, "loss": 1.2891, "step": 23553 }, { "epoch": 58.885, "grad_norm": 6.5, "learning_rate": 7.630919298839833e-06, "loss": 1.3827, "step": 23554 }, { "epoch": 58.8875, "grad_norm": 10.5625, "learning_rate": 7.630132667941769e-06, "loss": 1.3689, "step": 23555 }, { "epoch": 58.89, "grad_norm": 6.875, "learning_rate": 7.629346052580455e-06, "loss": 1.2803, "step": 23556 }, { "epoch": 58.8925, "grad_norm": 7.3125, "learning_rate": 7.628559452761055e-06, "loss": 1.3258, "step": 23557 }, { "epoch": 58.895, "grad_norm": 8.0625, "learning_rate": 7.627772868488726e-06, "loss": 1.4227, "step": 23558 }, { "epoch": 58.8975, "grad_norm": 7.15625, "learning_rate": 7.626986299768624e-06, "loss": 1.3574, "step": 23559 }, { "epoch": 58.9, "grad_norm": 5.3125, "learning_rate": 7.6261997466059035e-06, "loss": 1.477, "step": 23560 }, { "epoch": 58.9025, "grad_norm": 8.1875, "learning_rate": 7.62541320900572e-06, "loss": 1.4278, "step": 23561 }, { "epoch": 58.905, "grad_norm": 13.4375, "learning_rate": 7.624626686973234e-06, "loss": 1.3121, "step": 23562 }, { "epoch": 58.9075, "grad_norm": 5.625, "learning_rate": 7.623840180513605e-06, "loss": 1.2645, "step": 23563 }, { "epoch": 58.91, "grad_norm": 6.28125, "learning_rate": 7.62305368963198e-06, "loss": 1.1513, "step": 23564 }, { "epoch": 58.9125, "grad_norm": 10.25, "learning_rate": 7.622267214333522e-06, "loss": 1.3113, "step": 23565 }, { "epoch": 58.915, "grad_norm": 10.5, "learning_rate": 7.621480754623379e-06, "loss": 1.3369, "step": 23566 }, { "epoch": 58.9175, "grad_norm": 8.5, "learning_rate": 7.62069431050672e-06, "loss": 1.3647, "step": 23567 }, { "epoch": 58.92, "grad_norm": 10.0625, "learning_rate": 7.619907881988692e-06, "loss": 1.3702, "step": 23568 }, { "epoch": 58.9225, "grad_norm": 6.59375, "learning_rate": 7.6191214690744545e-06, "loss": 1.3361, "step": 23569 }, { "epoch": 58.925, "grad_norm": 8.1875, "learning_rate": 7.618335071769159e-06, "loss": 1.3111, "step": 23570 }, { "epoch": 58.9275, "grad_norm": 6.375, "learning_rate": 7.617548690077961e-06, "loss": 1.3489, "step": 23571 }, { "epoch": 58.93, "grad_norm": 7.6875, "learning_rate": 7.6167623240060205e-06, "loss": 1.1872, "step": 23572 }, { "epoch": 58.9325, "grad_norm": 7.1875, "learning_rate": 7.615975973558492e-06, "loss": 1.3602, "step": 23573 }, { "epoch": 58.935, "grad_norm": 8.6875, "learning_rate": 7.615189638740531e-06, "loss": 1.3052, "step": 23574 }, { "epoch": 58.9375, "grad_norm": 6.65625, "learning_rate": 7.6144033195572886e-06, "loss": 1.2454, "step": 23575 }, { "epoch": 58.94, "grad_norm": 10.625, "learning_rate": 7.613617016013921e-06, "loss": 1.2459, "step": 23576 }, { "epoch": 58.9425, "grad_norm": 12.1875, "learning_rate": 7.612830728115588e-06, "loss": 1.3869, "step": 23577 }, { "epoch": 58.945, "grad_norm": 10.0625, "learning_rate": 7.612044455867442e-06, "loss": 1.3908, "step": 23578 }, { "epoch": 58.9475, "grad_norm": 7.3125, "learning_rate": 7.611258199274635e-06, "loss": 1.1676, "step": 23579 }, { "epoch": 58.95, "grad_norm": 8.625, "learning_rate": 7.610471958342326e-06, "loss": 1.2309, "step": 23580 }, { "epoch": 58.9525, "grad_norm": 9.4375, "learning_rate": 7.60968573307566e-06, "loss": 1.3095, "step": 23581 }, { "epoch": 58.955, "grad_norm": 9.4375, "learning_rate": 7.608899523479805e-06, "loss": 1.41, "step": 23582 }, { "epoch": 58.9575, "grad_norm": 10.0625, "learning_rate": 7.608113329559909e-06, "loss": 1.2973, "step": 23583 }, { "epoch": 58.96, "grad_norm": 5.1875, "learning_rate": 7.607327151321127e-06, "loss": 1.3503, "step": 23584 }, { "epoch": 58.9625, "grad_norm": 6.875, "learning_rate": 7.60654098876861e-06, "loss": 1.2011, "step": 23585 }, { "epoch": 58.965, "grad_norm": 10.5625, "learning_rate": 7.605754841907514e-06, "loss": 1.3735, "step": 23586 }, { "epoch": 58.9675, "grad_norm": 6.1875, "learning_rate": 7.604968710742994e-06, "loss": 1.4078, "step": 23587 }, { "epoch": 58.97, "grad_norm": 5.875, "learning_rate": 7.604182595280205e-06, "loss": 1.2633, "step": 23588 }, { "epoch": 58.9725, "grad_norm": 6.125, "learning_rate": 7.603396495524298e-06, "loss": 1.2221, "step": 23589 }, { "epoch": 58.975, "grad_norm": 8.875, "learning_rate": 7.602610411480427e-06, "loss": 1.097, "step": 23590 }, { "epoch": 58.9775, "grad_norm": 8.375, "learning_rate": 7.6018243431537455e-06, "loss": 1.2864, "step": 23591 }, { "epoch": 58.98, "grad_norm": 10.4375, "learning_rate": 7.60103829054941e-06, "loss": 1.3141, "step": 23592 }, { "epoch": 58.9825, "grad_norm": 8.1875, "learning_rate": 7.600252253672571e-06, "loss": 1.3357, "step": 23593 }, { "epoch": 58.985, "grad_norm": 6.90625, "learning_rate": 7.599466232528381e-06, "loss": 1.3419, "step": 23594 }, { "epoch": 58.9875, "grad_norm": 7.46875, "learning_rate": 7.598680227121996e-06, "loss": 1.3611, "step": 23595 }, { "epoch": 58.99, "grad_norm": 6.65625, "learning_rate": 7.597894237458564e-06, "loss": 1.2815, "step": 23596 }, { "epoch": 58.9925, "grad_norm": 7.59375, "learning_rate": 7.597108263543243e-06, "loss": 1.238, "step": 23597 }, { "epoch": 58.995, "grad_norm": 5.8125, "learning_rate": 7.5963223053811855e-06, "loss": 1.2523, "step": 23598 }, { "epoch": 58.9975, "grad_norm": 6.25, "learning_rate": 7.595536362977544e-06, "loss": 1.2569, "step": 23599 }, { "epoch": 59.0, "grad_norm": 8.0, "learning_rate": 7.594750436337467e-06, "loss": 1.2559, "step": 23600 }, { "epoch": 59.0025, "grad_norm": 8.125, "learning_rate": 7.593964525466107e-06, "loss": 1.2818, "step": 23601 }, { "epoch": 59.005, "grad_norm": 7.5625, "learning_rate": 7.593178630368624e-06, "loss": 1.2316, "step": 23602 }, { "epoch": 59.0075, "grad_norm": 8.5625, "learning_rate": 7.592392751050166e-06, "loss": 1.4853, "step": 23603 }, { "epoch": 59.01, "grad_norm": 8.3125, "learning_rate": 7.591606887515883e-06, "loss": 1.3236, "step": 23604 }, { "epoch": 59.0125, "grad_norm": 5.8125, "learning_rate": 7.59082103977093e-06, "loss": 1.2038, "step": 23605 }, { "epoch": 59.015, "grad_norm": 13.75, "learning_rate": 7.590035207820454e-06, "loss": 1.3866, "step": 23606 }, { "epoch": 59.0175, "grad_norm": 7.65625, "learning_rate": 7.589249391669615e-06, "loss": 1.3006, "step": 23607 }, { "epoch": 59.02, "grad_norm": 7.3125, "learning_rate": 7.588463591323559e-06, "loss": 1.2863, "step": 23608 }, { "epoch": 59.0225, "grad_norm": 6.75, "learning_rate": 7.587677806787439e-06, "loss": 1.2403, "step": 23609 }, { "epoch": 59.025, "grad_norm": 10.0, "learning_rate": 7.586892038066409e-06, "loss": 1.2899, "step": 23610 }, { "epoch": 59.0275, "grad_norm": 5.3125, "learning_rate": 7.586106285165614e-06, "loss": 1.127, "step": 23611 }, { "epoch": 59.03, "grad_norm": 8.9375, "learning_rate": 7.585320548090213e-06, "loss": 1.2997, "step": 23612 }, { "epoch": 59.0325, "grad_norm": 8.5625, "learning_rate": 7.584534826845352e-06, "loss": 1.2111, "step": 23613 }, { "epoch": 59.035, "grad_norm": 9.375, "learning_rate": 7.583749121436187e-06, "loss": 1.4657, "step": 23614 }, { "epoch": 59.0375, "grad_norm": 6.71875, "learning_rate": 7.582963431867863e-06, "loss": 1.4484, "step": 23615 }, { "epoch": 59.04, "grad_norm": 13.3125, "learning_rate": 7.582177758145532e-06, "loss": 1.2878, "step": 23616 }, { "epoch": 59.0425, "grad_norm": 6.6875, "learning_rate": 7.581392100274351e-06, "loss": 1.2343, "step": 23617 }, { "epoch": 59.045, "grad_norm": 8.0625, "learning_rate": 7.580606458259468e-06, "loss": 1.2088, "step": 23618 }, { "epoch": 59.0475, "grad_norm": 7.9375, "learning_rate": 7.579820832106028e-06, "loss": 1.2968, "step": 23619 }, { "epoch": 59.05, "grad_norm": 4.21875, "learning_rate": 7.579035221819188e-06, "loss": 1.2904, "step": 23620 }, { "epoch": 59.0525, "grad_norm": 8.5625, "learning_rate": 7.578249627404093e-06, "loss": 1.3711, "step": 23621 }, { "epoch": 59.055, "grad_norm": 8.0625, "learning_rate": 7.577464048865901e-06, "loss": 1.2922, "step": 23622 }, { "epoch": 59.0575, "grad_norm": 10.4375, "learning_rate": 7.576678486209755e-06, "loss": 1.5035, "step": 23623 }, { "epoch": 59.06, "grad_norm": 10.1875, "learning_rate": 7.57589293944081e-06, "loss": 1.271, "step": 23624 }, { "epoch": 59.0625, "grad_norm": 7.1875, "learning_rate": 7.575107408564214e-06, "loss": 1.2063, "step": 23625 }, { "epoch": 59.065, "grad_norm": 6.8125, "learning_rate": 7.574321893585113e-06, "loss": 1.2387, "step": 23626 }, { "epoch": 59.0675, "grad_norm": 8.4375, "learning_rate": 7.573536394508663e-06, "loss": 1.3649, "step": 23627 }, { "epoch": 59.07, "grad_norm": 8.5, "learning_rate": 7.572750911340011e-06, "loss": 1.2005, "step": 23628 }, { "epoch": 59.0725, "grad_norm": 7.3125, "learning_rate": 7.5719654440843095e-06, "loss": 1.3144, "step": 23629 }, { "epoch": 59.075, "grad_norm": 7.90625, "learning_rate": 7.571179992746703e-06, "loss": 1.2638, "step": 23630 }, { "epoch": 59.0775, "grad_norm": 6.1875, "learning_rate": 7.570394557332341e-06, "loss": 1.4336, "step": 23631 }, { "epoch": 59.08, "grad_norm": 8.3125, "learning_rate": 7.569609137846376e-06, "loss": 1.3903, "step": 23632 }, { "epoch": 59.0825, "grad_norm": 7.1875, "learning_rate": 7.5688237342939604e-06, "loss": 1.346, "step": 23633 }, { "epoch": 59.085, "grad_norm": 8.8125, "learning_rate": 7.568038346680236e-06, "loss": 1.2539, "step": 23634 }, { "epoch": 59.0875, "grad_norm": 10.125, "learning_rate": 7.567252975010353e-06, "loss": 1.39, "step": 23635 }, { "epoch": 59.09, "grad_norm": 10.9375, "learning_rate": 7.566467619289464e-06, "loss": 1.2938, "step": 23636 }, { "epoch": 59.0925, "grad_norm": 8.0625, "learning_rate": 7.565682279522719e-06, "loss": 1.3868, "step": 23637 }, { "epoch": 59.095, "grad_norm": 9.1875, "learning_rate": 7.5648969557152595e-06, "loss": 1.188, "step": 23638 }, { "epoch": 59.0975, "grad_norm": 7.84375, "learning_rate": 7.564111647872242e-06, "loss": 1.2854, "step": 23639 }, { "epoch": 59.1, "grad_norm": 12.1875, "learning_rate": 7.5633263559988035e-06, "loss": 1.2975, "step": 23640 }, { "epoch": 59.1025, "grad_norm": 6.78125, "learning_rate": 7.562541080100107e-06, "loss": 1.1903, "step": 23641 }, { "epoch": 59.105, "grad_norm": 10.9375, "learning_rate": 7.561755820181291e-06, "loss": 1.1669, "step": 23642 }, { "epoch": 59.1075, "grad_norm": 9.25, "learning_rate": 7.560970576247507e-06, "loss": 1.1658, "step": 23643 }, { "epoch": 59.11, "grad_norm": 5.78125, "learning_rate": 7.560185348303904e-06, "loss": 1.3647, "step": 23644 }, { "epoch": 59.1125, "grad_norm": 6.9375, "learning_rate": 7.559400136355625e-06, "loss": 1.2719, "step": 23645 }, { "epoch": 59.115, "grad_norm": 6.3125, "learning_rate": 7.558614940407822e-06, "loss": 1.3846, "step": 23646 }, { "epoch": 59.1175, "grad_norm": 6.0625, "learning_rate": 7.557829760465641e-06, "loss": 1.4393, "step": 23647 }, { "epoch": 59.12, "grad_norm": 6.3125, "learning_rate": 7.557044596534234e-06, "loss": 1.4112, "step": 23648 }, { "epoch": 59.1225, "grad_norm": 6.96875, "learning_rate": 7.5562594486187426e-06, "loss": 1.351, "step": 23649 }, { "epoch": 59.125, "grad_norm": 6.84375, "learning_rate": 7.5554743167243135e-06, "loss": 1.255, "step": 23650 }, { "epoch": 59.1275, "grad_norm": 7.75, "learning_rate": 7.5546892008561e-06, "loss": 1.1761, "step": 23651 }, { "epoch": 59.13, "grad_norm": 9.875, "learning_rate": 7.553904101019247e-06, "loss": 1.2506, "step": 23652 }, { "epoch": 59.1325, "grad_norm": 6.59375, "learning_rate": 7.553119017218899e-06, "loss": 1.3242, "step": 23653 }, { "epoch": 59.135, "grad_norm": 7.5625, "learning_rate": 7.552333949460209e-06, "loss": 1.4043, "step": 23654 }, { "epoch": 59.1375, "grad_norm": 18.75, "learning_rate": 7.551548897748311e-06, "loss": 1.4107, "step": 23655 }, { "epoch": 59.14, "grad_norm": 6.0625, "learning_rate": 7.550763862088369e-06, "loss": 1.2267, "step": 23656 }, { "epoch": 59.1425, "grad_norm": 8.25, "learning_rate": 7.5499788424855166e-06, "loss": 1.2545, "step": 23657 }, { "epoch": 59.145, "grad_norm": 6.8125, "learning_rate": 7.549193838944908e-06, "loss": 1.4801, "step": 23658 }, { "epoch": 59.1475, "grad_norm": 8.25, "learning_rate": 7.548408851471684e-06, "loss": 1.4282, "step": 23659 }, { "epoch": 59.15, "grad_norm": 6.8125, "learning_rate": 7.547623880070992e-06, "loss": 1.2896, "step": 23660 }, { "epoch": 59.1525, "grad_norm": 7.34375, "learning_rate": 7.546838924747981e-06, "loss": 1.3982, "step": 23661 }, { "epoch": 59.155, "grad_norm": 9.0625, "learning_rate": 7.5460539855077995e-06, "loss": 1.3973, "step": 23662 }, { "epoch": 59.1575, "grad_norm": 8.75, "learning_rate": 7.545269062355587e-06, "loss": 1.3392, "step": 23663 }, { "epoch": 59.16, "grad_norm": 9.375, "learning_rate": 7.544484155296492e-06, "loss": 1.3415, "step": 23664 }, { "epoch": 59.1625, "grad_norm": 8.3125, "learning_rate": 7.543699264335657e-06, "loss": 1.2794, "step": 23665 }, { "epoch": 59.165, "grad_norm": 6.8125, "learning_rate": 7.542914389478237e-06, "loss": 1.123, "step": 23666 }, { "epoch": 59.1675, "grad_norm": 8.3125, "learning_rate": 7.542129530729371e-06, "loss": 1.3633, "step": 23667 }, { "epoch": 59.17, "grad_norm": 7.09375, "learning_rate": 7.541344688094204e-06, "loss": 1.4259, "step": 23668 }, { "epoch": 59.1725, "grad_norm": 11.6875, "learning_rate": 7.540559861577885e-06, "loss": 1.2633, "step": 23669 }, { "epoch": 59.175, "grad_norm": 10.5, "learning_rate": 7.5397750511855515e-06, "loss": 1.3536, "step": 23670 }, { "epoch": 59.1775, "grad_norm": 6.375, "learning_rate": 7.538990256922361e-06, "loss": 1.2224, "step": 23671 }, { "epoch": 59.18, "grad_norm": 6.34375, "learning_rate": 7.538205478793448e-06, "loss": 1.4538, "step": 23672 }, { "epoch": 59.1825, "grad_norm": 7.0625, "learning_rate": 7.537420716803965e-06, "loss": 1.3333, "step": 23673 }, { "epoch": 59.185, "grad_norm": 6.625, "learning_rate": 7.536635970959051e-06, "loss": 1.4186, "step": 23674 }, { "epoch": 59.1875, "grad_norm": 7.90625, "learning_rate": 7.53585124126385e-06, "loss": 1.3995, "step": 23675 }, { "epoch": 59.19, "grad_norm": 8.0625, "learning_rate": 7.535066527723512e-06, "loss": 1.2975, "step": 23676 }, { "epoch": 59.1925, "grad_norm": 6.75, "learning_rate": 7.534281830343182e-06, "loss": 1.3117, "step": 23677 }, { "epoch": 59.195, "grad_norm": 7.28125, "learning_rate": 7.533497149128e-06, "loss": 1.1725, "step": 23678 }, { "epoch": 59.1975, "grad_norm": 5.96875, "learning_rate": 7.53271248408311e-06, "loss": 1.3348, "step": 23679 }, { "epoch": 59.2, "grad_norm": 10.4375, "learning_rate": 7.531927835213657e-06, "loss": 1.2346, "step": 23680 }, { "epoch": 59.2025, "grad_norm": 7.375, "learning_rate": 7.531143202524791e-06, "loss": 1.2903, "step": 23681 }, { "epoch": 59.205, "grad_norm": 9.75, "learning_rate": 7.530358586021649e-06, "loss": 1.3122, "step": 23682 }, { "epoch": 59.2075, "grad_norm": 7.3125, "learning_rate": 7.529573985709377e-06, "loss": 1.2999, "step": 23683 }, { "epoch": 59.21, "grad_norm": 7.5625, "learning_rate": 7.528789401593122e-06, "loss": 1.5274, "step": 23684 }, { "epoch": 59.2125, "grad_norm": 6.875, "learning_rate": 7.52800483367802e-06, "loss": 1.2655, "step": 23685 }, { "epoch": 59.215, "grad_norm": 7.90625, "learning_rate": 7.5272202819692216e-06, "loss": 1.367, "step": 23686 }, { "epoch": 59.2175, "grad_norm": 9.0, "learning_rate": 7.526435746471869e-06, "loss": 1.3286, "step": 23687 }, { "epoch": 59.22, "grad_norm": 9.625, "learning_rate": 7.525651227191105e-06, "loss": 1.3288, "step": 23688 }, { "epoch": 59.2225, "grad_norm": 8.375, "learning_rate": 7.524866724132072e-06, "loss": 1.3069, "step": 23689 }, { "epoch": 59.225, "grad_norm": 7.0, "learning_rate": 7.52408223729991e-06, "loss": 1.3267, "step": 23690 }, { "epoch": 59.2275, "grad_norm": 15.25, "learning_rate": 7.523297766699769e-06, "loss": 1.4443, "step": 23691 }, { "epoch": 59.23, "grad_norm": 8.1875, "learning_rate": 7.52251331233679e-06, "loss": 1.3773, "step": 23692 }, { "epoch": 59.2325, "grad_norm": 5.59375, "learning_rate": 7.5217288742161135e-06, "loss": 1.254, "step": 23693 }, { "epoch": 59.235, "grad_norm": 8.5, "learning_rate": 7.520944452342883e-06, "loss": 1.4321, "step": 23694 }, { "epoch": 59.2375, "grad_norm": 7.78125, "learning_rate": 7.520160046722239e-06, "loss": 1.2996, "step": 23695 }, { "epoch": 59.24, "grad_norm": 6.75, "learning_rate": 7.519375657359331e-06, "loss": 1.1385, "step": 23696 }, { "epoch": 59.2425, "grad_norm": 10.0, "learning_rate": 7.518591284259294e-06, "loss": 1.4629, "step": 23697 }, { "epoch": 59.245, "grad_norm": 8.0625, "learning_rate": 7.517806927427274e-06, "loss": 1.3634, "step": 23698 }, { "epoch": 59.2475, "grad_norm": 8.1875, "learning_rate": 7.517022586868415e-06, "loss": 1.4908, "step": 23699 }, { "epoch": 59.25, "grad_norm": 6.0625, "learning_rate": 7.516238262587851e-06, "loss": 1.4681, "step": 23700 }, { "epoch": 59.2525, "grad_norm": 7.9375, "learning_rate": 7.515453954590733e-06, "loss": 1.2646, "step": 23701 }, { "epoch": 59.255, "grad_norm": 7.9375, "learning_rate": 7.514669662882199e-06, "loss": 1.2519, "step": 23702 }, { "epoch": 59.2575, "grad_norm": 8.375, "learning_rate": 7.513885387467394e-06, "loss": 1.373, "step": 23703 }, { "epoch": 59.26, "grad_norm": 5.59375, "learning_rate": 7.513101128351454e-06, "loss": 1.2686, "step": 23704 }, { "epoch": 59.2625, "grad_norm": 10.9375, "learning_rate": 7.512316885539521e-06, "loss": 1.3298, "step": 23705 }, { "epoch": 59.265, "grad_norm": 10.5625, "learning_rate": 7.511532659036742e-06, "loss": 1.3589, "step": 23706 }, { "epoch": 59.2675, "grad_norm": 6.84375, "learning_rate": 7.510748448848257e-06, "loss": 1.1976, "step": 23707 }, { "epoch": 59.27, "grad_norm": 6.40625, "learning_rate": 7.509964254979204e-06, "loss": 1.299, "step": 23708 }, { "epoch": 59.2725, "grad_norm": 6.125, "learning_rate": 7.509180077434727e-06, "loss": 1.4271, "step": 23709 }, { "epoch": 59.275, "grad_norm": 8.4375, "learning_rate": 7.508395916219959e-06, "loss": 1.207, "step": 23710 }, { "epoch": 59.2775, "grad_norm": 5.34375, "learning_rate": 7.507611771340055e-06, "loss": 1.1115, "step": 23711 }, { "epoch": 59.28, "grad_norm": 7.59375, "learning_rate": 7.506827642800146e-06, "loss": 1.4156, "step": 23712 }, { "epoch": 59.2825, "grad_norm": 8.125, "learning_rate": 7.506043530605376e-06, "loss": 1.2598, "step": 23713 }, { "epoch": 59.285, "grad_norm": 14.1875, "learning_rate": 7.5052594347608854e-06, "loss": 1.3483, "step": 23714 }, { "epoch": 59.2875, "grad_norm": 8.4375, "learning_rate": 7.504475355271811e-06, "loss": 1.4261, "step": 23715 }, { "epoch": 59.29, "grad_norm": 6.84375, "learning_rate": 7.503691292143298e-06, "loss": 1.3617, "step": 23716 }, { "epoch": 59.2925, "grad_norm": 8.8125, "learning_rate": 7.502907245380485e-06, "loss": 1.3455, "step": 23717 }, { "epoch": 59.295, "grad_norm": 9.1875, "learning_rate": 7.502123214988515e-06, "loss": 1.2681, "step": 23718 }, { "epoch": 59.2975, "grad_norm": 7.84375, "learning_rate": 7.501339200972521e-06, "loss": 1.5672, "step": 23719 }, { "epoch": 59.3, "grad_norm": 7.5625, "learning_rate": 7.500555203337647e-06, "loss": 1.44, "step": 23720 }, { "epoch": 59.3025, "grad_norm": 7.65625, "learning_rate": 7.499771222089035e-06, "loss": 1.3229, "step": 23721 }, { "epoch": 59.305, "grad_norm": 6.65625, "learning_rate": 7.498987257231825e-06, "loss": 1.2689, "step": 23722 }, { "epoch": 59.3075, "grad_norm": 5.90625, "learning_rate": 7.4982033087711525e-06, "loss": 1.2363, "step": 23723 }, { "epoch": 59.31, "grad_norm": 6.46875, "learning_rate": 7.49741937671216e-06, "loss": 1.3805, "step": 23724 }, { "epoch": 59.3125, "grad_norm": 12.125, "learning_rate": 7.49663546105998e-06, "loss": 1.4311, "step": 23725 }, { "epoch": 59.315, "grad_norm": 7.59375, "learning_rate": 7.495851561819765e-06, "loss": 1.3642, "step": 23726 }, { "epoch": 59.3175, "grad_norm": 7.1875, "learning_rate": 7.495067678996644e-06, "loss": 1.1902, "step": 23727 }, { "epoch": 59.32, "grad_norm": 8.6875, "learning_rate": 7.49428381259576e-06, "loss": 1.5036, "step": 23728 }, { "epoch": 59.3225, "grad_norm": 7.53125, "learning_rate": 7.493499962622251e-06, "loss": 1.3102, "step": 23729 }, { "epoch": 59.325, "grad_norm": 9.5, "learning_rate": 7.492716129081253e-06, "loss": 1.2599, "step": 23730 }, { "epoch": 59.3275, "grad_norm": 8.1875, "learning_rate": 7.49193231197791e-06, "loss": 1.3451, "step": 23731 }, { "epoch": 59.33, "grad_norm": 11.25, "learning_rate": 7.491148511317361e-06, "loss": 1.2562, "step": 23732 }, { "epoch": 59.3325, "grad_norm": 7.6875, "learning_rate": 7.490364727104739e-06, "loss": 1.3953, "step": 23733 }, { "epoch": 59.335, "grad_norm": 7.0, "learning_rate": 7.489580959345185e-06, "loss": 1.3714, "step": 23734 }, { "epoch": 59.3375, "grad_norm": 6.1875, "learning_rate": 7.488797208043837e-06, "loss": 1.3554, "step": 23735 }, { "epoch": 59.34, "grad_norm": 7.21875, "learning_rate": 7.488013473205838e-06, "loss": 1.2438, "step": 23736 }, { "epoch": 59.3425, "grad_norm": 7.5, "learning_rate": 7.487229754836318e-06, "loss": 1.3101, "step": 23737 }, { "epoch": 59.345, "grad_norm": 6.40625, "learning_rate": 7.4864460529404215e-06, "loss": 1.2954, "step": 23738 }, { "epoch": 59.3475, "grad_norm": 10.4375, "learning_rate": 7.485662367523286e-06, "loss": 1.338, "step": 23739 }, { "epoch": 59.35, "grad_norm": 6.96875, "learning_rate": 7.48487869859004e-06, "loss": 1.3414, "step": 23740 }, { "epoch": 59.3525, "grad_norm": 6.8125, "learning_rate": 7.484095046145836e-06, "loss": 1.3222, "step": 23741 }, { "epoch": 59.355, "grad_norm": 7.65625, "learning_rate": 7.4833114101958015e-06, "loss": 1.3425, "step": 23742 }, { "epoch": 59.3575, "grad_norm": 7.90625, "learning_rate": 7.4825277907450774e-06, "loss": 1.4148, "step": 23743 }, { "epoch": 59.36, "grad_norm": 5.6875, "learning_rate": 7.4817441877988005e-06, "loss": 1.3318, "step": 23744 }, { "epoch": 59.3625, "grad_norm": 6.96875, "learning_rate": 7.480960601362105e-06, "loss": 1.1681, "step": 23745 }, { "epoch": 59.365, "grad_norm": 5.96875, "learning_rate": 7.480177031440132e-06, "loss": 1.3525, "step": 23746 }, { "epoch": 59.3675, "grad_norm": 5.96875, "learning_rate": 7.479393478038022e-06, "loss": 1.3205, "step": 23747 }, { "epoch": 59.37, "grad_norm": 9.375, "learning_rate": 7.4786099411609035e-06, "loss": 1.4404, "step": 23748 }, { "epoch": 59.3725, "grad_norm": 7.5625, "learning_rate": 7.4778264208139185e-06, "loss": 1.3057, "step": 23749 }, { "epoch": 59.375, "grad_norm": 4.8125, "learning_rate": 7.4770429170022e-06, "loss": 1.2979, "step": 23750 }, { "epoch": 59.3775, "grad_norm": 6.1875, "learning_rate": 7.476259429730891e-06, "loss": 1.4304, "step": 23751 }, { "epoch": 59.38, "grad_norm": 6.65625, "learning_rate": 7.475475959005123e-06, "loss": 1.2106, "step": 23752 }, { "epoch": 59.3825, "grad_norm": 8.125, "learning_rate": 7.474692504830033e-06, "loss": 1.3964, "step": 23753 }, { "epoch": 59.385, "grad_norm": 9.375, "learning_rate": 7.473909067210759e-06, "loss": 1.2931, "step": 23754 }, { "epoch": 59.3875, "grad_norm": 8.5625, "learning_rate": 7.473125646152433e-06, "loss": 1.283, "step": 23755 }, { "epoch": 59.39, "grad_norm": 8.0625, "learning_rate": 7.472342241660197e-06, "loss": 1.3453, "step": 23756 }, { "epoch": 59.3925, "grad_norm": 7.34375, "learning_rate": 7.471558853739182e-06, "loss": 1.1366, "step": 23757 }, { "epoch": 59.395, "grad_norm": 9.9375, "learning_rate": 7.47077548239453e-06, "loss": 1.4011, "step": 23758 }, { "epoch": 59.3975, "grad_norm": 6.84375, "learning_rate": 7.4699921276313694e-06, "loss": 1.3385, "step": 23759 }, { "epoch": 59.4, "grad_norm": 9.875, "learning_rate": 7.469208789454838e-06, "loss": 1.3993, "step": 23760 }, { "epoch": 59.4025, "grad_norm": 8.625, "learning_rate": 7.4684254678700745e-06, "loss": 1.2203, "step": 23761 }, { "epoch": 59.405, "grad_norm": 8.6875, "learning_rate": 7.467642162882214e-06, "loss": 1.3254, "step": 23762 }, { "epoch": 59.4075, "grad_norm": 10.4375, "learning_rate": 7.466858874496388e-06, "loss": 1.1556, "step": 23763 }, { "epoch": 59.41, "grad_norm": 22.0, "learning_rate": 7.466075602717733e-06, "loss": 1.22, "step": 23764 }, { "epoch": 59.4125, "grad_norm": 8.875, "learning_rate": 7.465292347551385e-06, "loss": 1.3936, "step": 23765 }, { "epoch": 59.415, "grad_norm": 5.125, "learning_rate": 7.464509109002481e-06, "loss": 1.289, "step": 23766 }, { "epoch": 59.4175, "grad_norm": 7.25, "learning_rate": 7.463725887076153e-06, "loss": 1.4702, "step": 23767 }, { "epoch": 59.42, "grad_norm": 8.25, "learning_rate": 7.462942681777537e-06, "loss": 1.2846, "step": 23768 }, { "epoch": 59.4225, "grad_norm": 5.84375, "learning_rate": 7.462159493111767e-06, "loss": 1.3281, "step": 23769 }, { "epoch": 59.425, "grad_norm": 6.21875, "learning_rate": 7.461376321083975e-06, "loss": 1.1735, "step": 23770 }, { "epoch": 59.4275, "grad_norm": 11.5625, "learning_rate": 7.460593165699301e-06, "loss": 1.1377, "step": 23771 }, { "epoch": 59.43, "grad_norm": 11.5, "learning_rate": 7.459810026962877e-06, "loss": 1.5374, "step": 23772 }, { "epoch": 59.4325, "grad_norm": 9.1875, "learning_rate": 7.459026904879838e-06, "loss": 1.3217, "step": 23773 }, { "epoch": 59.435, "grad_norm": 7.21875, "learning_rate": 7.4582437994553134e-06, "loss": 1.3061, "step": 23774 }, { "epoch": 59.4375, "grad_norm": 6.03125, "learning_rate": 7.457460710694441e-06, "loss": 1.2901, "step": 23775 }, { "epoch": 59.44, "grad_norm": 6.71875, "learning_rate": 7.456677638602355e-06, "loss": 1.302, "step": 23776 }, { "epoch": 59.4425, "grad_norm": 7.625, "learning_rate": 7.455894583184192e-06, "loss": 1.4568, "step": 23777 }, { "epoch": 59.445, "grad_norm": 7.84375, "learning_rate": 7.45511154444508e-06, "loss": 1.2962, "step": 23778 }, { "epoch": 59.4475, "grad_norm": 8.5, "learning_rate": 7.454328522390156e-06, "loss": 1.3363, "step": 23779 }, { "epoch": 59.45, "grad_norm": 8.0625, "learning_rate": 7.4535455170245476e-06, "loss": 1.4521, "step": 23780 }, { "epoch": 59.4525, "grad_norm": 8.0, "learning_rate": 7.452762528353397e-06, "loss": 1.4186, "step": 23781 }, { "epoch": 59.455, "grad_norm": 7.46875, "learning_rate": 7.451979556381832e-06, "loss": 1.3329, "step": 23782 }, { "epoch": 59.4575, "grad_norm": 29.625, "learning_rate": 7.451196601114987e-06, "loss": 1.2237, "step": 23783 }, { "epoch": 59.46, "grad_norm": 8.0, "learning_rate": 7.450413662557998e-06, "loss": 1.5021, "step": 23784 }, { "epoch": 59.4625, "grad_norm": 7.0, "learning_rate": 7.449630740715992e-06, "loss": 1.3721, "step": 23785 }, { "epoch": 59.465, "grad_norm": 7.0625, "learning_rate": 7.448847835594104e-06, "loss": 1.1405, "step": 23786 }, { "epoch": 59.4675, "grad_norm": 15.5625, "learning_rate": 7.448064947197469e-06, "loss": 1.3219, "step": 23787 }, { "epoch": 59.47, "grad_norm": 6.8125, "learning_rate": 7.44728207553122e-06, "loss": 1.3061, "step": 23788 }, { "epoch": 59.4725, "grad_norm": 5.71875, "learning_rate": 7.446499220600486e-06, "loss": 1.2746, "step": 23789 }, { "epoch": 59.475, "grad_norm": 5.875, "learning_rate": 7.445716382410398e-06, "loss": 1.392, "step": 23790 }, { "epoch": 59.4775, "grad_norm": 8.0625, "learning_rate": 7.444933560966093e-06, "loss": 1.4125, "step": 23791 }, { "epoch": 59.48, "grad_norm": 8.3125, "learning_rate": 7.444150756272704e-06, "loss": 1.3333, "step": 23792 }, { "epoch": 59.4825, "grad_norm": 9.125, "learning_rate": 7.4433679683353586e-06, "loss": 1.4431, "step": 23793 }, { "epoch": 59.485, "grad_norm": 5.0, "learning_rate": 7.442585197159193e-06, "loss": 1.3434, "step": 23794 }, { "epoch": 59.4875, "grad_norm": 7.25, "learning_rate": 7.441802442749328e-06, "loss": 1.4214, "step": 23795 }, { "epoch": 59.49, "grad_norm": 8.125, "learning_rate": 7.441019705110912e-06, "loss": 1.4172, "step": 23796 }, { "epoch": 59.4925, "grad_norm": 5.71875, "learning_rate": 7.4402369842490665e-06, "loss": 1.3372, "step": 23797 }, { "epoch": 59.495, "grad_norm": 5.90625, "learning_rate": 7.439454280168926e-06, "loss": 1.377, "step": 23798 }, { "epoch": 59.4975, "grad_norm": 7.375, "learning_rate": 7.438671592875619e-06, "loss": 1.4335, "step": 23799 }, { "epoch": 59.5, "grad_norm": 9.1875, "learning_rate": 7.4378889223742766e-06, "loss": 1.3586, "step": 23800 }, { "epoch": 59.5025, "grad_norm": 7.25, "learning_rate": 7.437106268670033e-06, "loss": 1.321, "step": 23801 }, { "epoch": 59.505, "grad_norm": 6.9375, "learning_rate": 7.436323631768022e-06, "loss": 1.3998, "step": 23802 }, { "epoch": 59.5075, "grad_norm": 6.96875, "learning_rate": 7.435541011673368e-06, "loss": 1.4417, "step": 23803 }, { "epoch": 59.51, "grad_norm": 5.59375, "learning_rate": 7.434758408391205e-06, "loss": 1.1838, "step": 23804 }, { "epoch": 59.5125, "grad_norm": 9.0625, "learning_rate": 7.43397582192666e-06, "loss": 1.4181, "step": 23805 }, { "epoch": 59.515, "grad_norm": 8.5, "learning_rate": 7.4331932522848715e-06, "loss": 1.2829, "step": 23806 }, { "epoch": 59.5175, "grad_norm": 8.125, "learning_rate": 7.432410699470962e-06, "loss": 1.4254, "step": 23807 }, { "epoch": 59.52, "grad_norm": 8.875, "learning_rate": 7.431628163490067e-06, "loss": 1.1621, "step": 23808 }, { "epoch": 59.5225, "grad_norm": 9.625, "learning_rate": 7.430845644347316e-06, "loss": 1.4083, "step": 23809 }, { "epoch": 59.525, "grad_norm": 6.25, "learning_rate": 7.430063142047832e-06, "loss": 1.2986, "step": 23810 }, { "epoch": 59.5275, "grad_norm": 11.1875, "learning_rate": 7.429280656596758e-06, "loss": 1.2814, "step": 23811 }, { "epoch": 59.53, "grad_norm": 8.0625, "learning_rate": 7.428498187999215e-06, "loss": 1.1716, "step": 23812 }, { "epoch": 59.5325, "grad_norm": 7.28125, "learning_rate": 7.427715736260337e-06, "loss": 1.3793, "step": 23813 }, { "epoch": 59.535, "grad_norm": 5.1875, "learning_rate": 7.426933301385249e-06, "loss": 1.2652, "step": 23814 }, { "epoch": 59.5375, "grad_norm": 4.71875, "learning_rate": 7.426150883379082e-06, "loss": 1.3189, "step": 23815 }, { "epoch": 59.54, "grad_norm": 7.1875, "learning_rate": 7.4253684822469684e-06, "loss": 1.3222, "step": 23816 }, { "epoch": 59.5425, "grad_norm": 6.0625, "learning_rate": 7.424586097994038e-06, "loss": 1.1866, "step": 23817 }, { "epoch": 59.545, "grad_norm": 11.1875, "learning_rate": 7.423803730625417e-06, "loss": 1.4412, "step": 23818 }, { "epoch": 59.5475, "grad_norm": 6.25, "learning_rate": 7.4230213801462334e-06, "loss": 1.3217, "step": 23819 }, { "epoch": 59.55, "grad_norm": 6.0625, "learning_rate": 7.422239046561619e-06, "loss": 1.2777, "step": 23820 }, { "epoch": 59.5525, "grad_norm": 6.625, "learning_rate": 7.4214567298767046e-06, "loss": 1.3256, "step": 23821 }, { "epoch": 59.555, "grad_norm": 5.15625, "learning_rate": 7.420674430096615e-06, "loss": 1.3573, "step": 23822 }, { "epoch": 59.5575, "grad_norm": 9.9375, "learning_rate": 7.419892147226482e-06, "loss": 1.2474, "step": 23823 }, { "epoch": 59.56, "grad_norm": 5.40625, "learning_rate": 7.419109881271434e-06, "loss": 1.1733, "step": 23824 }, { "epoch": 59.5625, "grad_norm": 6.96875, "learning_rate": 7.4183276322365925e-06, "loss": 1.3271, "step": 23825 }, { "epoch": 59.565, "grad_norm": 7.59375, "learning_rate": 7.417545400127095e-06, "loss": 1.2572, "step": 23826 }, { "epoch": 59.5675, "grad_norm": 8.8125, "learning_rate": 7.416763184948068e-06, "loss": 1.4003, "step": 23827 }, { "epoch": 59.57, "grad_norm": 8.6875, "learning_rate": 7.415980986704638e-06, "loss": 1.3392, "step": 23828 }, { "epoch": 59.5725, "grad_norm": 6.65625, "learning_rate": 7.415198805401933e-06, "loss": 1.1603, "step": 23829 }, { "epoch": 59.575, "grad_norm": 9.25, "learning_rate": 7.414416641045079e-06, "loss": 1.4933, "step": 23830 }, { "epoch": 59.5775, "grad_norm": 6.875, "learning_rate": 7.413634493639207e-06, "loss": 1.3736, "step": 23831 }, { "epoch": 59.58, "grad_norm": 7.125, "learning_rate": 7.4128523631894464e-06, "loss": 1.2218, "step": 23832 }, { "epoch": 59.5825, "grad_norm": 7.5625, "learning_rate": 7.41207024970092e-06, "loss": 1.2464, "step": 23833 }, { "epoch": 59.585, "grad_norm": 9.4375, "learning_rate": 7.411288153178757e-06, "loss": 1.1533, "step": 23834 }, { "epoch": 59.5875, "grad_norm": 9.125, "learning_rate": 7.410506073628084e-06, "loss": 1.163, "step": 23835 }, { "epoch": 59.59, "grad_norm": 7.21875, "learning_rate": 7.4097240110540334e-06, "loss": 1.188, "step": 23836 }, { "epoch": 59.5925, "grad_norm": 6.75, "learning_rate": 7.408941965461727e-06, "loss": 1.3064, "step": 23837 }, { "epoch": 59.595, "grad_norm": 7.46875, "learning_rate": 7.408159936856292e-06, "loss": 1.2143, "step": 23838 }, { "epoch": 59.5975, "grad_norm": 8.875, "learning_rate": 7.40737792524286e-06, "loss": 1.3983, "step": 23839 }, { "epoch": 59.6, "grad_norm": 9.3125, "learning_rate": 7.40659593062655e-06, "loss": 1.2585, "step": 23840 }, { "epoch": 59.6025, "grad_norm": 6.46875, "learning_rate": 7.405813953012495e-06, "loss": 1.3432, "step": 23841 }, { "epoch": 59.605, "grad_norm": 5.75, "learning_rate": 7.405031992405821e-06, "loss": 1.2833, "step": 23842 }, { "epoch": 59.6075, "grad_norm": 9.25, "learning_rate": 7.4042500488116546e-06, "loss": 1.3592, "step": 23843 }, { "epoch": 59.61, "grad_norm": 5.5, "learning_rate": 7.40346812223512e-06, "loss": 1.264, "step": 23844 }, { "epoch": 59.6125, "grad_norm": 17.625, "learning_rate": 7.4026862126813404e-06, "loss": 1.392, "step": 23845 }, { "epoch": 59.615, "grad_norm": 7.1875, "learning_rate": 7.40190432015545e-06, "loss": 1.4495, "step": 23846 }, { "epoch": 59.6175, "grad_norm": 10.9375, "learning_rate": 7.401122444662572e-06, "loss": 1.2843, "step": 23847 }, { "epoch": 59.62, "grad_norm": 5.9375, "learning_rate": 7.40034058620783e-06, "loss": 1.2544, "step": 23848 }, { "epoch": 59.6225, "grad_norm": 19.75, "learning_rate": 7.399558744796352e-06, "loss": 1.2914, "step": 23849 }, { "epoch": 59.625, "grad_norm": 6.78125, "learning_rate": 7.398776920433257e-06, "loss": 1.446, "step": 23850 }, { "epoch": 59.6275, "grad_norm": 9.375, "learning_rate": 7.3979951131236834e-06, "loss": 1.2038, "step": 23851 }, { "epoch": 59.63, "grad_norm": 5.96875, "learning_rate": 7.397213322872747e-06, "loss": 1.267, "step": 23852 }, { "epoch": 59.6325, "grad_norm": 6.59375, "learning_rate": 7.396431549685577e-06, "loss": 1.2356, "step": 23853 }, { "epoch": 59.635, "grad_norm": 8.1875, "learning_rate": 7.395649793567297e-06, "loss": 1.3312, "step": 23854 }, { "epoch": 59.6375, "grad_norm": 5.15625, "learning_rate": 7.39486805452303e-06, "loss": 1.2479, "step": 23855 }, { "epoch": 59.64, "grad_norm": 9.125, "learning_rate": 7.394086332557907e-06, "loss": 1.4189, "step": 23856 }, { "epoch": 59.6425, "grad_norm": 9.4375, "learning_rate": 7.3933046276770495e-06, "loss": 1.3339, "step": 23857 }, { "epoch": 59.645, "grad_norm": 8.0, "learning_rate": 7.3925229398855825e-06, "loss": 1.2241, "step": 23858 }, { "epoch": 59.6475, "grad_norm": 6.0625, "learning_rate": 7.39174126918863e-06, "loss": 1.404, "step": 23859 }, { "epoch": 59.65, "grad_norm": 6.6875, "learning_rate": 7.390959615591315e-06, "loss": 1.3964, "step": 23860 }, { "epoch": 59.6525, "grad_norm": 6.03125, "learning_rate": 7.390177979098767e-06, "loss": 1.2477, "step": 23861 }, { "epoch": 59.655, "grad_norm": 8.25, "learning_rate": 7.38939635971611e-06, "loss": 1.3746, "step": 23862 }, { "epoch": 59.6575, "grad_norm": 8.0625, "learning_rate": 7.388614757448465e-06, "loss": 1.3927, "step": 23863 }, { "epoch": 59.66, "grad_norm": 10.6875, "learning_rate": 7.387833172300958e-06, "loss": 1.4357, "step": 23864 }, { "epoch": 59.6625, "grad_norm": 6.5, "learning_rate": 7.3870516042787045e-06, "loss": 1.4507, "step": 23865 }, { "epoch": 59.665, "grad_norm": 6.625, "learning_rate": 7.386270053386844e-06, "loss": 1.2738, "step": 23866 }, { "epoch": 59.6675, "grad_norm": 7.46875, "learning_rate": 7.38548851963049e-06, "loss": 1.3915, "step": 23867 }, { "epoch": 59.67, "grad_norm": 7.6875, "learning_rate": 7.3847070030147704e-06, "loss": 1.3412, "step": 23868 }, { "epoch": 59.6725, "grad_norm": 8.3125, "learning_rate": 7.383925503544804e-06, "loss": 1.2465, "step": 23869 }, { "epoch": 59.675, "grad_norm": 4.78125, "learning_rate": 7.3831440212257165e-06, "loss": 1.4335, "step": 23870 }, { "epoch": 59.6775, "grad_norm": 7.125, "learning_rate": 7.382362556062635e-06, "loss": 1.1784, "step": 23871 }, { "epoch": 59.68, "grad_norm": 10.0, "learning_rate": 7.38158110806068e-06, "loss": 1.3193, "step": 23872 }, { "epoch": 59.6825, "grad_norm": 6.125, "learning_rate": 7.380799677224975e-06, "loss": 1.3072, "step": 23873 }, { "epoch": 59.685, "grad_norm": 6.03125, "learning_rate": 7.380018263560638e-06, "loss": 1.3098, "step": 23874 }, { "epoch": 59.6875, "grad_norm": 7.65625, "learning_rate": 7.379236867072799e-06, "loss": 1.344, "step": 23875 }, { "epoch": 59.69, "grad_norm": 7.0, "learning_rate": 7.378455487766581e-06, "loss": 1.2814, "step": 23876 }, { "epoch": 59.6925, "grad_norm": 8.8125, "learning_rate": 7.3776741256471016e-06, "loss": 1.355, "step": 23877 }, { "epoch": 59.695, "grad_norm": 5.96875, "learning_rate": 7.376892780719486e-06, "loss": 1.4097, "step": 23878 }, { "epoch": 59.6975, "grad_norm": 6.9375, "learning_rate": 7.376111452988853e-06, "loss": 1.237, "step": 23879 }, { "epoch": 59.7, "grad_norm": 8.8125, "learning_rate": 7.375330142460331e-06, "loss": 1.3825, "step": 23880 }, { "epoch": 59.7025, "grad_norm": 8.5625, "learning_rate": 7.374548849139043e-06, "loss": 1.4809, "step": 23881 }, { "epoch": 59.705, "grad_norm": 8.6875, "learning_rate": 7.373767573030104e-06, "loss": 1.2851, "step": 23882 }, { "epoch": 59.7075, "grad_norm": 6.375, "learning_rate": 7.372986314138641e-06, "loss": 1.4511, "step": 23883 }, { "epoch": 59.71, "grad_norm": 7.3125, "learning_rate": 7.3722050724697705e-06, "loss": 1.2662, "step": 23884 }, { "epoch": 59.7125, "grad_norm": 12.0625, "learning_rate": 7.371423848028624e-06, "loss": 1.3829, "step": 23885 }, { "epoch": 59.715, "grad_norm": 10.25, "learning_rate": 7.370642640820315e-06, "loss": 1.304, "step": 23886 }, { "epoch": 59.7175, "grad_norm": 7.125, "learning_rate": 7.36986145084997e-06, "loss": 1.3019, "step": 23887 }, { "epoch": 59.72, "grad_norm": 7.53125, "learning_rate": 7.3690802781227056e-06, "loss": 1.1288, "step": 23888 }, { "epoch": 59.7225, "grad_norm": 8.6875, "learning_rate": 7.368299122643644e-06, "loss": 1.3946, "step": 23889 }, { "epoch": 59.725, "grad_norm": 9.6875, "learning_rate": 7.36751798441791e-06, "loss": 1.2515, "step": 23890 }, { "epoch": 59.7275, "grad_norm": 4.03125, "learning_rate": 7.3667368634506255e-06, "loss": 1.3523, "step": 23891 }, { "epoch": 59.73, "grad_norm": 6.9375, "learning_rate": 7.365955759746906e-06, "loss": 1.2765, "step": 23892 }, { "epoch": 59.7325, "grad_norm": 5.5625, "learning_rate": 7.365174673311874e-06, "loss": 1.1443, "step": 23893 }, { "epoch": 59.735, "grad_norm": 7.71875, "learning_rate": 7.36439360415065e-06, "loss": 1.3308, "step": 23894 }, { "epoch": 59.7375, "grad_norm": 16.25, "learning_rate": 7.363612552268361e-06, "loss": 1.3138, "step": 23895 }, { "epoch": 59.74, "grad_norm": 8.375, "learning_rate": 7.362831517670121e-06, "loss": 1.2881, "step": 23896 }, { "epoch": 59.7425, "grad_norm": 7.84375, "learning_rate": 7.362050500361051e-06, "loss": 1.3004, "step": 23897 }, { "epoch": 59.745, "grad_norm": 7.0, "learning_rate": 7.361269500346274e-06, "loss": 1.3526, "step": 23898 }, { "epoch": 59.7475, "grad_norm": 6.9375, "learning_rate": 7.360488517630905e-06, "loss": 1.3365, "step": 23899 }, { "epoch": 59.75, "grad_norm": 6.875, "learning_rate": 7.35970755222007e-06, "loss": 1.2868, "step": 23900 }, { "epoch": 59.7525, "grad_norm": 8.1875, "learning_rate": 7.358926604118886e-06, "loss": 1.3288, "step": 23901 }, { "epoch": 59.755, "grad_norm": 5.125, "learning_rate": 7.3581456733324756e-06, "loss": 1.3568, "step": 23902 }, { "epoch": 59.7575, "grad_norm": 8.5, "learning_rate": 7.357364759865955e-06, "loss": 1.3259, "step": 23903 }, { "epoch": 59.76, "grad_norm": 10.25, "learning_rate": 7.356583863724442e-06, "loss": 1.4824, "step": 23904 }, { "epoch": 59.7625, "grad_norm": 8.25, "learning_rate": 7.355802984913063e-06, "loss": 1.2961, "step": 23905 }, { "epoch": 59.765, "grad_norm": 4.9375, "learning_rate": 7.355022123436936e-06, "loss": 1.2878, "step": 23906 }, { "epoch": 59.7675, "grad_norm": 7.78125, "learning_rate": 7.3542412793011754e-06, "loss": 1.3862, "step": 23907 }, { "epoch": 59.77, "grad_norm": 6.34375, "learning_rate": 7.353460452510903e-06, "loss": 1.3437, "step": 23908 }, { "epoch": 59.7725, "grad_norm": 5.9375, "learning_rate": 7.352679643071235e-06, "loss": 1.5289, "step": 23909 }, { "epoch": 59.775, "grad_norm": 9.6875, "learning_rate": 7.351898850987299e-06, "loss": 1.2724, "step": 23910 }, { "epoch": 59.7775, "grad_norm": 7.6875, "learning_rate": 7.351118076264205e-06, "loss": 1.4851, "step": 23911 }, { "epoch": 59.78, "grad_norm": 8.625, "learning_rate": 7.350337318907075e-06, "loss": 1.4225, "step": 23912 }, { "epoch": 59.7825, "grad_norm": 7.6875, "learning_rate": 7.349556578921031e-06, "loss": 1.5524, "step": 23913 }, { "epoch": 59.785, "grad_norm": 9.0, "learning_rate": 7.3487758563111825e-06, "loss": 1.443, "step": 23914 }, { "epoch": 59.7875, "grad_norm": 9.625, "learning_rate": 7.347995151082656e-06, "loss": 1.4939, "step": 23915 }, { "epoch": 59.79, "grad_norm": 7.1875, "learning_rate": 7.347214463240566e-06, "loss": 1.2326, "step": 23916 }, { "epoch": 59.7925, "grad_norm": 8.0625, "learning_rate": 7.3464337927900355e-06, "loss": 1.2774, "step": 23917 }, { "epoch": 59.795, "grad_norm": 7.625, "learning_rate": 7.345653139736175e-06, "loss": 1.1921, "step": 23918 }, { "epoch": 59.7975, "grad_norm": 8.1875, "learning_rate": 7.344872504084106e-06, "loss": 1.2995, "step": 23919 }, { "epoch": 59.8, "grad_norm": 9.9375, "learning_rate": 7.344091885838949e-06, "loss": 1.3109, "step": 23920 }, { "epoch": 59.8025, "grad_norm": 7.25, "learning_rate": 7.34331128500582e-06, "loss": 1.3005, "step": 23921 }, { "epoch": 59.805, "grad_norm": 10.0625, "learning_rate": 7.342530701589834e-06, "loss": 1.2712, "step": 23922 }, { "epoch": 59.8075, "grad_norm": 10.0625, "learning_rate": 7.341750135596111e-06, "loss": 1.318, "step": 23923 }, { "epoch": 59.81, "grad_norm": 8.6875, "learning_rate": 7.340969587029765e-06, "loss": 1.392, "step": 23924 }, { "epoch": 59.8125, "grad_norm": 7.40625, "learning_rate": 7.34018905589592e-06, "loss": 1.3689, "step": 23925 }, { "epoch": 59.815, "grad_norm": 6.40625, "learning_rate": 7.3394085421996874e-06, "loss": 1.5429, "step": 23926 }, { "epoch": 59.8175, "grad_norm": 7.3125, "learning_rate": 7.338628045946185e-06, "loss": 1.4113, "step": 23927 }, { "epoch": 59.82, "grad_norm": 6.90625, "learning_rate": 7.337847567140534e-06, "loss": 1.3759, "step": 23928 }, { "epoch": 59.8225, "grad_norm": 6.71875, "learning_rate": 7.337067105787844e-06, "loss": 1.3166, "step": 23929 }, { "epoch": 59.825, "grad_norm": 8.375, "learning_rate": 7.336286661893237e-06, "loss": 1.2977, "step": 23930 }, { "epoch": 59.8275, "grad_norm": 6.21875, "learning_rate": 7.335506235461828e-06, "loss": 1.4607, "step": 23931 }, { "epoch": 59.83, "grad_norm": 6.28125, "learning_rate": 7.334725826498735e-06, "loss": 1.2705, "step": 23932 }, { "epoch": 59.8325, "grad_norm": 8.25, "learning_rate": 7.333945435009071e-06, "loss": 1.3746, "step": 23933 }, { "epoch": 59.835, "grad_norm": 9.1875, "learning_rate": 7.333165060997953e-06, "loss": 1.2832, "step": 23934 }, { "epoch": 59.8375, "grad_norm": 11.875, "learning_rate": 7.332384704470499e-06, "loss": 1.3502, "step": 23935 }, { "epoch": 59.84, "grad_norm": 8.375, "learning_rate": 7.331604365431826e-06, "loss": 1.3613, "step": 23936 }, { "epoch": 59.8425, "grad_norm": 7.0, "learning_rate": 7.330824043887047e-06, "loss": 1.3519, "step": 23937 }, { "epoch": 59.845, "grad_norm": 6.40625, "learning_rate": 7.33004373984128e-06, "loss": 1.4029, "step": 23938 }, { "epoch": 59.8475, "grad_norm": 7.59375, "learning_rate": 7.329263453299633e-06, "loss": 1.2463, "step": 23939 }, { "epoch": 59.85, "grad_norm": 7.6875, "learning_rate": 7.328483184267236e-06, "loss": 1.268, "step": 23940 }, { "epoch": 59.8525, "grad_norm": 5.4375, "learning_rate": 7.327702932749192e-06, "loss": 1.3342, "step": 23941 }, { "epoch": 59.855, "grad_norm": 9.375, "learning_rate": 7.3269226987506246e-06, "loss": 1.3308, "step": 23942 }, { "epoch": 59.8575, "grad_norm": 8.4375, "learning_rate": 7.326142482276642e-06, "loss": 1.424, "step": 23943 }, { "epoch": 59.86, "grad_norm": 6.53125, "learning_rate": 7.32536228333236e-06, "loss": 1.2416, "step": 23944 }, { "epoch": 59.8625, "grad_norm": 6.625, "learning_rate": 7.324582101922898e-06, "loss": 1.3398, "step": 23945 }, { "epoch": 59.865, "grad_norm": 7.3125, "learning_rate": 7.323801938053371e-06, "loss": 1.3939, "step": 23946 }, { "epoch": 59.8675, "grad_norm": 9.1875, "learning_rate": 7.32302179172889e-06, "loss": 1.388, "step": 23947 }, { "epoch": 59.87, "grad_norm": 6.28125, "learning_rate": 7.32224166295457e-06, "loss": 1.2695, "step": 23948 }, { "epoch": 59.8725, "grad_norm": 8.875, "learning_rate": 7.321461551735524e-06, "loss": 1.34, "step": 23949 }, { "epoch": 59.875, "grad_norm": 6.03125, "learning_rate": 7.320681458076871e-06, "loss": 1.2942, "step": 23950 }, { "epoch": 59.8775, "grad_norm": 7.875, "learning_rate": 7.319901381983726e-06, "loss": 1.3909, "step": 23951 }, { "epoch": 59.88, "grad_norm": 8.0, "learning_rate": 7.319121323461198e-06, "loss": 1.4228, "step": 23952 }, { "epoch": 59.8825, "grad_norm": 6.3125, "learning_rate": 7.318341282514405e-06, "loss": 1.2306, "step": 23953 }, { "epoch": 59.885, "grad_norm": 8.5625, "learning_rate": 7.317561259148454e-06, "loss": 1.3293, "step": 23954 }, { "epoch": 59.8875, "grad_norm": 8.5, "learning_rate": 7.31678125336847e-06, "loss": 1.3218, "step": 23955 }, { "epoch": 59.89, "grad_norm": 4.9375, "learning_rate": 7.316001265179557e-06, "loss": 1.3708, "step": 23956 }, { "epoch": 59.8925, "grad_norm": 5.1875, "learning_rate": 7.3152212945868365e-06, "loss": 1.3179, "step": 23957 }, { "epoch": 59.895, "grad_norm": 6.0625, "learning_rate": 7.314441341595416e-06, "loss": 1.3524, "step": 23958 }, { "epoch": 59.8975, "grad_norm": 6.15625, "learning_rate": 7.313661406210406e-06, "loss": 1.2684, "step": 23959 }, { "epoch": 59.9, "grad_norm": 7.9375, "learning_rate": 7.312881488436928e-06, "loss": 1.4164, "step": 23960 }, { "epoch": 59.9025, "grad_norm": 5.8125, "learning_rate": 7.3121015882800936e-06, "loss": 1.2107, "step": 23961 }, { "epoch": 59.905, "grad_norm": 6.03125, "learning_rate": 7.311321705745011e-06, "loss": 1.333, "step": 23962 }, { "epoch": 59.9075, "grad_norm": 8.4375, "learning_rate": 7.310541840836796e-06, "loss": 1.5086, "step": 23963 }, { "epoch": 59.91, "grad_norm": 6.6875, "learning_rate": 7.309761993560559e-06, "loss": 1.4019, "step": 23964 }, { "epoch": 59.9125, "grad_norm": 11.625, "learning_rate": 7.308982163921419e-06, "loss": 1.2734, "step": 23965 }, { "epoch": 59.915, "grad_norm": 8.3125, "learning_rate": 7.308202351924483e-06, "loss": 1.3956, "step": 23966 }, { "epoch": 59.9175, "grad_norm": 8.6875, "learning_rate": 7.307422557574863e-06, "loss": 1.218, "step": 23967 }, { "epoch": 59.92, "grad_norm": 7.40625, "learning_rate": 7.3066427808776754e-06, "loss": 1.222, "step": 23968 }, { "epoch": 59.9225, "grad_norm": 7.9375, "learning_rate": 7.305863021838027e-06, "loss": 1.4693, "step": 23969 }, { "epoch": 59.925, "grad_norm": 4.625, "learning_rate": 7.3050832804610325e-06, "loss": 1.2238, "step": 23970 }, { "epoch": 59.9275, "grad_norm": 8.25, "learning_rate": 7.304303556751806e-06, "loss": 1.2704, "step": 23971 }, { "epoch": 59.93, "grad_norm": 10.8125, "learning_rate": 7.3035238507154595e-06, "loss": 1.321, "step": 23972 }, { "epoch": 59.9325, "grad_norm": 11.6875, "learning_rate": 7.3027441623571e-06, "loss": 1.3814, "step": 23973 }, { "epoch": 59.935, "grad_norm": 11.0, "learning_rate": 7.30196449168184e-06, "loss": 1.3184, "step": 23974 }, { "epoch": 59.9375, "grad_norm": 6.0, "learning_rate": 7.301184838694795e-06, "loss": 1.2734, "step": 23975 }, { "epoch": 59.94, "grad_norm": 6.03125, "learning_rate": 7.300405203401076e-06, "loss": 1.2054, "step": 23976 }, { "epoch": 59.9425, "grad_norm": 6.78125, "learning_rate": 7.299625585805792e-06, "loss": 1.3944, "step": 23977 }, { "epoch": 59.945, "grad_norm": 8.0, "learning_rate": 7.2988459859140525e-06, "loss": 1.3466, "step": 23978 }, { "epoch": 59.9475, "grad_norm": 5.75, "learning_rate": 7.298066403730969e-06, "loss": 1.4649, "step": 23979 }, { "epoch": 59.95, "grad_norm": 7.4375, "learning_rate": 7.297286839261659e-06, "loss": 1.3117, "step": 23980 }, { "epoch": 59.9525, "grad_norm": 9.875, "learning_rate": 7.296507292511226e-06, "loss": 1.3589, "step": 23981 }, { "epoch": 59.955, "grad_norm": 6.75, "learning_rate": 7.295727763484783e-06, "loss": 1.3425, "step": 23982 }, { "epoch": 59.9575, "grad_norm": 7.71875, "learning_rate": 7.294948252187441e-06, "loss": 1.2611, "step": 23983 }, { "epoch": 59.96, "grad_norm": 8.375, "learning_rate": 7.294168758624307e-06, "loss": 1.1513, "step": 23984 }, { "epoch": 59.9625, "grad_norm": 7.40625, "learning_rate": 7.293389282800498e-06, "loss": 1.2281, "step": 23985 }, { "epoch": 59.965, "grad_norm": 7.40625, "learning_rate": 7.2926098247211185e-06, "loss": 1.3787, "step": 23986 }, { "epoch": 59.9675, "grad_norm": 9.3125, "learning_rate": 7.291830384391283e-06, "loss": 1.3306, "step": 23987 }, { "epoch": 59.97, "grad_norm": 6.1875, "learning_rate": 7.2910509618160974e-06, "loss": 1.397, "step": 23988 }, { "epoch": 59.9725, "grad_norm": 5.8125, "learning_rate": 7.290271557000672e-06, "loss": 1.2639, "step": 23989 }, { "epoch": 59.975, "grad_norm": 8.75, "learning_rate": 7.289492169950119e-06, "loss": 1.468, "step": 23990 }, { "epoch": 59.9775, "grad_norm": 9.75, "learning_rate": 7.28871280066955e-06, "loss": 1.5465, "step": 23991 }, { "epoch": 59.98, "grad_norm": 7.46875, "learning_rate": 7.287933449164068e-06, "loss": 1.3171, "step": 23992 }, { "epoch": 59.9825, "grad_norm": 8.5, "learning_rate": 7.287154115438786e-06, "loss": 1.2915, "step": 23993 }, { "epoch": 59.985, "grad_norm": 6.0625, "learning_rate": 7.286374799498811e-06, "loss": 1.3692, "step": 23994 }, { "epoch": 59.9875, "grad_norm": 9.5, "learning_rate": 7.285595501349259e-06, "loss": 1.3867, "step": 23995 }, { "epoch": 59.99, "grad_norm": 8.375, "learning_rate": 7.2848162209952325e-06, "loss": 1.3264, "step": 23996 }, { "epoch": 59.9925, "grad_norm": 6.84375, "learning_rate": 7.28403695844184e-06, "loss": 1.3931, "step": 23997 }, { "epoch": 59.995, "grad_norm": 5.46875, "learning_rate": 7.2832577136941965e-06, "loss": 1.2379, "step": 23998 }, { "epoch": 59.9975, "grad_norm": 6.09375, "learning_rate": 7.282478486757403e-06, "loss": 1.1373, "step": 23999 }, { "epoch": 60.0, "grad_norm": 8.0625, "learning_rate": 7.2816992776365714e-06, "loss": 1.4392, "step": 24000 }, { "epoch": 60.0025, "grad_norm": 8.125, "learning_rate": 7.280920086336814e-06, "loss": 1.2368, "step": 24001 }, { "epoch": 60.005, "grad_norm": 6.21875, "learning_rate": 7.280140912863235e-06, "loss": 1.2229, "step": 24002 }, { "epoch": 60.0075, "grad_norm": 8.5625, "learning_rate": 7.279361757220942e-06, "loss": 1.4185, "step": 24003 }, { "epoch": 60.01, "grad_norm": 6.09375, "learning_rate": 7.2785826194150426e-06, "loss": 1.3283, "step": 24004 }, { "epoch": 60.0125, "grad_norm": 7.84375, "learning_rate": 7.277803499450649e-06, "loss": 1.2793, "step": 24005 }, { "epoch": 60.015, "grad_norm": 7.9375, "learning_rate": 7.277024397332868e-06, "loss": 1.4036, "step": 24006 }, { "epoch": 60.0175, "grad_norm": 8.625, "learning_rate": 7.276245313066806e-06, "loss": 1.1933, "step": 24007 }, { "epoch": 60.02, "grad_norm": 7.03125, "learning_rate": 7.27546624665757e-06, "loss": 1.3527, "step": 24008 }, { "epoch": 60.0225, "grad_norm": 9.6875, "learning_rate": 7.274687198110263e-06, "loss": 1.1772, "step": 24009 }, { "epoch": 60.025, "grad_norm": 8.875, "learning_rate": 7.2739081674300035e-06, "loss": 1.2191, "step": 24010 }, { "epoch": 60.0275, "grad_norm": 7.4375, "learning_rate": 7.273129154621892e-06, "loss": 1.1343, "step": 24011 }, { "epoch": 60.03, "grad_norm": 9.125, "learning_rate": 7.272350159691039e-06, "loss": 1.3903, "step": 24012 }, { "epoch": 60.0325, "grad_norm": 8.1875, "learning_rate": 7.271571182642546e-06, "loss": 1.3032, "step": 24013 }, { "epoch": 60.035, "grad_norm": 9.125, "learning_rate": 7.270792223481522e-06, "loss": 1.3218, "step": 24014 }, { "epoch": 60.0375, "grad_norm": 8.75, "learning_rate": 7.270013282213076e-06, "loss": 1.3785, "step": 24015 }, { "epoch": 60.04, "grad_norm": 7.25, "learning_rate": 7.269234358842314e-06, "loss": 1.388, "step": 24016 }, { "epoch": 60.0425, "grad_norm": 7.09375, "learning_rate": 7.268455453374345e-06, "loss": 1.4225, "step": 24017 }, { "epoch": 60.045, "grad_norm": 11.875, "learning_rate": 7.26767656581427e-06, "loss": 1.3824, "step": 24018 }, { "epoch": 60.0475, "grad_norm": 9.6875, "learning_rate": 7.266897696167197e-06, "loss": 1.1381, "step": 24019 }, { "epoch": 60.05, "grad_norm": 5.25, "learning_rate": 7.2661188444382345e-06, "loss": 1.3516, "step": 24020 }, { "epoch": 60.0525, "grad_norm": 7.46875, "learning_rate": 7.265340010632489e-06, "loss": 1.3982, "step": 24021 }, { "epoch": 60.055, "grad_norm": 6.875, "learning_rate": 7.264561194755064e-06, "loss": 1.1401, "step": 24022 }, { "epoch": 60.0575, "grad_norm": 9.4375, "learning_rate": 7.263782396811068e-06, "loss": 1.3028, "step": 24023 }, { "epoch": 60.06, "grad_norm": 6.125, "learning_rate": 7.263003616805599e-06, "loss": 1.3629, "step": 24024 }, { "epoch": 60.0625, "grad_norm": 6.21875, "learning_rate": 7.262224854743773e-06, "loss": 1.2777, "step": 24025 }, { "epoch": 60.065, "grad_norm": 11.3125, "learning_rate": 7.2614461106306925e-06, "loss": 1.5691, "step": 24026 }, { "epoch": 60.0675, "grad_norm": 5.96875, "learning_rate": 7.260667384471463e-06, "loss": 1.2757, "step": 24027 }, { "epoch": 60.07, "grad_norm": 7.625, "learning_rate": 7.259888676271185e-06, "loss": 1.4034, "step": 24028 }, { "epoch": 60.0725, "grad_norm": 6.40625, "learning_rate": 7.259109986034966e-06, "loss": 1.4468, "step": 24029 }, { "epoch": 60.075, "grad_norm": 7.28125, "learning_rate": 7.258331313767914e-06, "loss": 1.4008, "step": 24030 }, { "epoch": 60.0775, "grad_norm": 7.125, "learning_rate": 7.257552659475135e-06, "loss": 1.3867, "step": 24031 }, { "epoch": 60.08, "grad_norm": 5.3125, "learning_rate": 7.256774023161728e-06, "loss": 1.3383, "step": 24032 }, { "epoch": 60.0825, "grad_norm": 7.59375, "learning_rate": 7.255995404832801e-06, "loss": 1.358, "step": 24033 }, { "epoch": 60.085, "grad_norm": 6.65625, "learning_rate": 7.255216804493457e-06, "loss": 1.3443, "step": 24034 }, { "epoch": 60.0875, "grad_norm": 5.03125, "learning_rate": 7.254438222148805e-06, "loss": 1.1599, "step": 24035 }, { "epoch": 60.09, "grad_norm": 10.375, "learning_rate": 7.253659657803945e-06, "loss": 1.4838, "step": 24036 }, { "epoch": 60.0925, "grad_norm": 9.375, "learning_rate": 7.252881111463981e-06, "loss": 1.2859, "step": 24037 }, { "epoch": 60.095, "grad_norm": 7.0, "learning_rate": 7.252102583134022e-06, "loss": 1.1988, "step": 24038 }, { "epoch": 60.0975, "grad_norm": 8.0, "learning_rate": 7.251324072819163e-06, "loss": 1.2873, "step": 24039 }, { "epoch": 60.1, "grad_norm": 8.4375, "learning_rate": 7.250545580524515e-06, "loss": 1.4812, "step": 24040 }, { "epoch": 60.1025, "grad_norm": 6.4375, "learning_rate": 7.2497671062551826e-06, "loss": 1.2388, "step": 24041 }, { "epoch": 60.105, "grad_norm": 8.25, "learning_rate": 7.248988650016268e-06, "loss": 1.3001, "step": 24042 }, { "epoch": 60.1075, "grad_norm": 8.4375, "learning_rate": 7.248210211812871e-06, "loss": 1.1936, "step": 24043 }, { "epoch": 60.11, "grad_norm": 7.46875, "learning_rate": 7.247431791650096e-06, "loss": 1.3929, "step": 24044 }, { "epoch": 60.1125, "grad_norm": 9.5, "learning_rate": 7.2466533895330495e-06, "loss": 1.3426, "step": 24045 }, { "epoch": 60.115, "grad_norm": 6.96875, "learning_rate": 7.245875005466836e-06, "loss": 1.2796, "step": 24046 }, { "epoch": 60.1175, "grad_norm": 5.46875, "learning_rate": 7.245096639456555e-06, "loss": 1.3842, "step": 24047 }, { "epoch": 60.12, "grad_norm": 10.5625, "learning_rate": 7.244318291507308e-06, "loss": 1.3939, "step": 24048 }, { "epoch": 60.1225, "grad_norm": 8.875, "learning_rate": 7.243539961624199e-06, "loss": 1.2053, "step": 24049 }, { "epoch": 60.125, "grad_norm": 11.125, "learning_rate": 7.2427616498123356e-06, "loss": 1.4206, "step": 24050 }, { "epoch": 60.1275, "grad_norm": 7.28125, "learning_rate": 7.241983356076814e-06, "loss": 1.3626, "step": 24051 }, { "epoch": 60.13, "grad_norm": 8.0, "learning_rate": 7.24120508042274e-06, "loss": 1.434, "step": 24052 }, { "epoch": 60.1325, "grad_norm": 10.4375, "learning_rate": 7.2404268228552156e-06, "loss": 1.4573, "step": 24053 }, { "epoch": 60.135, "grad_norm": 6.375, "learning_rate": 7.239648583379339e-06, "loss": 1.3688, "step": 24054 }, { "epoch": 60.1375, "grad_norm": 5.6875, "learning_rate": 7.238870362000219e-06, "loss": 1.1467, "step": 24055 }, { "epoch": 60.14, "grad_norm": 5.59375, "learning_rate": 7.2380921587229536e-06, "loss": 1.364, "step": 24056 }, { "epoch": 60.1425, "grad_norm": 7.4375, "learning_rate": 7.237313973552649e-06, "loss": 1.2465, "step": 24057 }, { "epoch": 60.145, "grad_norm": 5.4375, "learning_rate": 7.236535806494399e-06, "loss": 1.2776, "step": 24058 }, { "epoch": 60.1475, "grad_norm": 6.75, "learning_rate": 7.2357576575533085e-06, "loss": 1.4309, "step": 24059 }, { "epoch": 60.15, "grad_norm": 5.5, "learning_rate": 7.234979526734482e-06, "loss": 1.4029, "step": 24060 }, { "epoch": 60.1525, "grad_norm": 9.3125, "learning_rate": 7.234201414043023e-06, "loss": 1.471, "step": 24061 }, { "epoch": 60.155, "grad_norm": 6.6875, "learning_rate": 7.233423319484026e-06, "loss": 1.1865, "step": 24062 }, { "epoch": 60.1575, "grad_norm": 7.53125, "learning_rate": 7.232645243062592e-06, "loss": 1.2167, "step": 24063 }, { "epoch": 60.16, "grad_norm": 7.46875, "learning_rate": 7.231867184783826e-06, "loss": 1.2641, "step": 24064 }, { "epoch": 60.1625, "grad_norm": 8.375, "learning_rate": 7.231089144652832e-06, "loss": 1.4284, "step": 24065 }, { "epoch": 60.165, "grad_norm": 14.125, "learning_rate": 7.2303111226747045e-06, "loss": 1.4304, "step": 24066 }, { "epoch": 60.1675, "grad_norm": 8.0625, "learning_rate": 7.229533118854546e-06, "loss": 1.2471, "step": 24067 }, { "epoch": 60.17, "grad_norm": 9.4375, "learning_rate": 7.228755133197459e-06, "loss": 1.22, "step": 24068 }, { "epoch": 60.1725, "grad_norm": 5.78125, "learning_rate": 7.22797716570854e-06, "loss": 1.3828, "step": 24069 }, { "epoch": 60.175, "grad_norm": 5.71875, "learning_rate": 7.2271992163928926e-06, "loss": 1.3088, "step": 24070 }, { "epoch": 60.1775, "grad_norm": 9.0625, "learning_rate": 7.226421285255616e-06, "loss": 1.3398, "step": 24071 }, { "epoch": 60.18, "grad_norm": 7.5, "learning_rate": 7.225643372301812e-06, "loss": 1.3006, "step": 24072 }, { "epoch": 60.1825, "grad_norm": 9.125, "learning_rate": 7.224865477536578e-06, "loss": 1.2155, "step": 24073 }, { "epoch": 60.185, "grad_norm": 8.6875, "learning_rate": 7.224087600965013e-06, "loss": 1.3104, "step": 24074 }, { "epoch": 60.1875, "grad_norm": 8.9375, "learning_rate": 7.223309742592221e-06, "loss": 1.4144, "step": 24075 }, { "epoch": 60.19, "grad_norm": 8.25, "learning_rate": 7.222531902423299e-06, "loss": 1.2609, "step": 24076 }, { "epoch": 60.1925, "grad_norm": 6.65625, "learning_rate": 7.221754080463348e-06, "loss": 1.1966, "step": 24077 }, { "epoch": 60.195, "grad_norm": 9.8125, "learning_rate": 7.2209762767174664e-06, "loss": 1.3403, "step": 24078 }, { "epoch": 60.1975, "grad_norm": 5.625, "learning_rate": 7.220198491190746e-06, "loss": 1.4139, "step": 24079 }, { "epoch": 60.2, "grad_norm": 6.03125, "learning_rate": 7.219420723888301e-06, "loss": 1.347, "step": 24080 }, { "epoch": 60.2025, "grad_norm": 8.25, "learning_rate": 7.21864297481522e-06, "loss": 1.4521, "step": 24081 }, { "epoch": 60.205, "grad_norm": 7.96875, "learning_rate": 7.217865243976605e-06, "loss": 1.3779, "step": 24082 }, { "epoch": 60.2075, "grad_norm": 6.375, "learning_rate": 7.217087531377554e-06, "loss": 1.3486, "step": 24083 }, { "epoch": 60.21, "grad_norm": 8.125, "learning_rate": 7.216309837023161e-06, "loss": 1.3331, "step": 24084 }, { "epoch": 60.2125, "grad_norm": 7.15625, "learning_rate": 7.215532160918535e-06, "loss": 1.3514, "step": 24085 }, { "epoch": 60.215, "grad_norm": 9.25, "learning_rate": 7.214754503068766e-06, "loss": 1.3325, "step": 24086 }, { "epoch": 60.2175, "grad_norm": 4.71875, "learning_rate": 7.213976863478958e-06, "loss": 1.3026, "step": 24087 }, { "epoch": 60.22, "grad_norm": 8.4375, "learning_rate": 7.213199242154203e-06, "loss": 1.4123, "step": 24088 }, { "epoch": 60.2225, "grad_norm": 7.8125, "learning_rate": 7.212421639099603e-06, "loss": 1.2972, "step": 24089 }, { "epoch": 60.225, "grad_norm": 10.5625, "learning_rate": 7.211644054320255e-06, "loss": 1.2591, "step": 24090 }, { "epoch": 60.2275, "grad_norm": 7.5, "learning_rate": 7.210866487821259e-06, "loss": 1.3033, "step": 24091 }, { "epoch": 60.23, "grad_norm": 6.3125, "learning_rate": 7.210088939607709e-06, "loss": 1.305, "step": 24092 }, { "epoch": 60.2325, "grad_norm": 8.6875, "learning_rate": 7.2093114096847075e-06, "loss": 1.4927, "step": 24093 }, { "epoch": 60.235, "grad_norm": 6.15625, "learning_rate": 7.20853389805734e-06, "loss": 1.1938, "step": 24094 }, { "epoch": 60.2375, "grad_norm": 7.5625, "learning_rate": 7.2077564047307214e-06, "loss": 1.1904, "step": 24095 }, { "epoch": 60.24, "grad_norm": 6.71875, "learning_rate": 7.2069789297099355e-06, "loss": 1.3577, "step": 24096 }, { "epoch": 60.2425, "grad_norm": 7.40625, "learning_rate": 7.206201473000088e-06, "loss": 1.3866, "step": 24097 }, { "epoch": 60.245, "grad_norm": 7.625, "learning_rate": 7.20542403460627e-06, "loss": 1.2291, "step": 24098 }, { "epoch": 60.2475, "grad_norm": 9.3125, "learning_rate": 7.204646614533577e-06, "loss": 1.2162, "step": 24099 }, { "epoch": 60.25, "grad_norm": 7.84375, "learning_rate": 7.203869212787112e-06, "loss": 1.3502, "step": 24100 }, { "epoch": 60.2525, "grad_norm": 9.875, "learning_rate": 7.203091829371969e-06, "loss": 1.5062, "step": 24101 }, { "epoch": 60.255, "grad_norm": 9.0, "learning_rate": 7.202314464293243e-06, "loss": 1.3675, "step": 24102 }, { "epoch": 60.2575, "grad_norm": 8.0, "learning_rate": 7.2015371175560325e-06, "loss": 1.2192, "step": 24103 }, { "epoch": 60.26, "grad_norm": 6.375, "learning_rate": 7.2007597891654275e-06, "loss": 1.2516, "step": 24104 }, { "epoch": 60.2625, "grad_norm": 10.5625, "learning_rate": 7.199982479126536e-06, "loss": 1.3942, "step": 24105 }, { "epoch": 60.265, "grad_norm": 8.25, "learning_rate": 7.199205187444445e-06, "loss": 1.3123, "step": 24106 }, { "epoch": 60.2675, "grad_norm": 5.125, "learning_rate": 7.198427914124252e-06, "loss": 1.258, "step": 24107 }, { "epoch": 60.27, "grad_norm": 8.375, "learning_rate": 7.197650659171056e-06, "loss": 1.3976, "step": 24108 }, { "epoch": 60.2725, "grad_norm": 7.09375, "learning_rate": 7.196873422589945e-06, "loss": 1.2654, "step": 24109 }, { "epoch": 60.275, "grad_norm": 9.1875, "learning_rate": 7.196096204386022e-06, "loss": 1.3355, "step": 24110 }, { "epoch": 60.2775, "grad_norm": 6.65625, "learning_rate": 7.1953190045643826e-06, "loss": 1.4273, "step": 24111 }, { "epoch": 60.28, "grad_norm": 6.28125, "learning_rate": 7.194541823130119e-06, "loss": 1.3397, "step": 24112 }, { "epoch": 60.2825, "grad_norm": 9.4375, "learning_rate": 7.193764660088326e-06, "loss": 1.3655, "step": 24113 }, { "epoch": 60.285, "grad_norm": 8.5, "learning_rate": 7.192987515444096e-06, "loss": 1.5493, "step": 24114 }, { "epoch": 60.2875, "grad_norm": 19.375, "learning_rate": 7.192210389202532e-06, "loss": 1.345, "step": 24115 }, { "epoch": 60.29, "grad_norm": 9.5, "learning_rate": 7.191433281368725e-06, "loss": 1.3441, "step": 24116 }, { "epoch": 60.2925, "grad_norm": 7.28125, "learning_rate": 7.190656191947767e-06, "loss": 1.3257, "step": 24117 }, { "epoch": 60.295, "grad_norm": 8.8125, "learning_rate": 7.189879120944752e-06, "loss": 1.3749, "step": 24118 }, { "epoch": 60.2975, "grad_norm": 5.96875, "learning_rate": 7.18910206836478e-06, "loss": 1.2435, "step": 24119 }, { "epoch": 60.3, "grad_norm": 5.8125, "learning_rate": 7.188325034212944e-06, "loss": 1.2673, "step": 24120 }, { "epoch": 60.3025, "grad_norm": 8.125, "learning_rate": 7.187548018494335e-06, "loss": 1.4184, "step": 24121 }, { "epoch": 60.305, "grad_norm": 6.34375, "learning_rate": 7.186771021214047e-06, "loss": 1.2424, "step": 24122 }, { "epoch": 60.3075, "grad_norm": 8.0625, "learning_rate": 7.185994042377176e-06, "loss": 1.3057, "step": 24123 }, { "epoch": 60.31, "grad_norm": 7.875, "learning_rate": 7.185217081988818e-06, "loss": 1.2675, "step": 24124 }, { "epoch": 60.3125, "grad_norm": 8.5, "learning_rate": 7.184440140054063e-06, "loss": 1.1823, "step": 24125 }, { "epoch": 60.315, "grad_norm": 6.75, "learning_rate": 7.183663216578005e-06, "loss": 1.28, "step": 24126 }, { "epoch": 60.3175, "grad_norm": 7.53125, "learning_rate": 7.1828863115657395e-06, "loss": 1.2915, "step": 24127 }, { "epoch": 60.32, "grad_norm": 8.75, "learning_rate": 7.182109425022357e-06, "loss": 1.5103, "step": 24128 }, { "epoch": 60.3225, "grad_norm": 6.125, "learning_rate": 7.181332556952951e-06, "loss": 1.2739, "step": 24129 }, { "epoch": 60.325, "grad_norm": 7.90625, "learning_rate": 7.180555707362619e-06, "loss": 1.2651, "step": 24130 }, { "epoch": 60.3275, "grad_norm": 9.4375, "learning_rate": 7.179778876256451e-06, "loss": 1.3325, "step": 24131 }, { "epoch": 60.33, "grad_norm": 5.3125, "learning_rate": 7.179002063639539e-06, "loss": 1.2052, "step": 24132 }, { "epoch": 60.3325, "grad_norm": 6.3125, "learning_rate": 7.178225269516974e-06, "loss": 1.2461, "step": 24133 }, { "epoch": 60.335, "grad_norm": 11.625, "learning_rate": 7.177448493893854e-06, "loss": 1.4412, "step": 24134 }, { "epoch": 60.3375, "grad_norm": 7.0625, "learning_rate": 7.17667173677527e-06, "loss": 1.3574, "step": 24135 }, { "epoch": 60.34, "grad_norm": 6.40625, "learning_rate": 7.175894998166311e-06, "loss": 1.3461, "step": 24136 }, { "epoch": 60.3425, "grad_norm": 8.125, "learning_rate": 7.175118278072072e-06, "loss": 1.2949, "step": 24137 }, { "epoch": 60.345, "grad_norm": 8.6875, "learning_rate": 7.174341576497642e-06, "loss": 1.2803, "step": 24138 }, { "epoch": 60.3475, "grad_norm": 8.3125, "learning_rate": 7.173564893448119e-06, "loss": 1.3953, "step": 24139 }, { "epoch": 60.35, "grad_norm": 7.15625, "learning_rate": 7.1727882289285915e-06, "loss": 1.269, "step": 24140 }, { "epoch": 60.3525, "grad_norm": 6.875, "learning_rate": 7.1720115829441495e-06, "loss": 1.3358, "step": 24141 }, { "epoch": 60.355, "grad_norm": 8.5625, "learning_rate": 7.171234955499888e-06, "loss": 1.1864, "step": 24142 }, { "epoch": 60.3575, "grad_norm": 7.625, "learning_rate": 7.170458346600895e-06, "loss": 1.2416, "step": 24143 }, { "epoch": 60.36, "grad_norm": 5.25, "learning_rate": 7.169681756252265e-06, "loss": 1.1807, "step": 24144 }, { "epoch": 60.3625, "grad_norm": 8.1875, "learning_rate": 7.168905184459087e-06, "loss": 1.5166, "step": 24145 }, { "epoch": 60.365, "grad_norm": 7.65625, "learning_rate": 7.168128631226456e-06, "loss": 1.2415, "step": 24146 }, { "epoch": 60.3675, "grad_norm": 9.1875, "learning_rate": 7.1673520965594585e-06, "loss": 1.463, "step": 24147 }, { "epoch": 60.37, "grad_norm": 7.53125, "learning_rate": 7.166575580463184e-06, "loss": 1.2247, "step": 24148 }, { "epoch": 60.3725, "grad_norm": 5.78125, "learning_rate": 7.165799082942731e-06, "loss": 1.364, "step": 24149 }, { "epoch": 60.375, "grad_norm": 7.65625, "learning_rate": 7.165022604003187e-06, "loss": 1.4863, "step": 24150 }, { "epoch": 60.3775, "grad_norm": 6.28125, "learning_rate": 7.164246143649638e-06, "loss": 1.3914, "step": 24151 }, { "epoch": 60.38, "grad_norm": 5.65625, "learning_rate": 7.163469701887182e-06, "loss": 1.2109, "step": 24152 }, { "epoch": 60.3825, "grad_norm": 5.84375, "learning_rate": 7.162693278720897e-06, "loss": 1.359, "step": 24153 }, { "epoch": 60.385, "grad_norm": 5.6875, "learning_rate": 7.161916874155888e-06, "loss": 1.217, "step": 24154 }, { "epoch": 60.3875, "grad_norm": 7.09375, "learning_rate": 7.161140488197237e-06, "loss": 1.3603, "step": 24155 }, { "epoch": 60.39, "grad_norm": 7.71875, "learning_rate": 7.160364120850034e-06, "loss": 1.3854, "step": 24156 }, { "epoch": 60.3925, "grad_norm": 6.25, "learning_rate": 7.159587772119374e-06, "loss": 1.3266, "step": 24157 }, { "epoch": 60.395, "grad_norm": 7.71875, "learning_rate": 7.158811442010339e-06, "loss": 1.23, "step": 24158 }, { "epoch": 60.3975, "grad_norm": 5.8125, "learning_rate": 7.158035130528023e-06, "loss": 1.3215, "step": 24159 }, { "epoch": 60.4, "grad_norm": 7.4375, "learning_rate": 7.157258837677514e-06, "loss": 1.3855, "step": 24160 }, { "epoch": 60.4025, "grad_norm": 7.53125, "learning_rate": 7.156482563463907e-06, "loss": 1.211, "step": 24161 }, { "epoch": 60.405, "grad_norm": 7.71875, "learning_rate": 7.155706307892283e-06, "loss": 1.399, "step": 24162 }, { "epoch": 60.4075, "grad_norm": 10.6875, "learning_rate": 7.154930070967734e-06, "loss": 1.2987, "step": 24163 }, { "epoch": 60.41, "grad_norm": 7.1875, "learning_rate": 7.154153852695349e-06, "loss": 1.3993, "step": 24164 }, { "epoch": 60.4125, "grad_norm": 6.1875, "learning_rate": 7.153377653080223e-06, "loss": 1.2642, "step": 24165 }, { "epoch": 60.415, "grad_norm": 11.3125, "learning_rate": 7.152601472127434e-06, "loss": 1.4477, "step": 24166 }, { "epoch": 60.4175, "grad_norm": 6.84375, "learning_rate": 7.15182530984208e-06, "loss": 1.3349, "step": 24167 }, { "epoch": 60.42, "grad_norm": 7.71875, "learning_rate": 7.151049166229237e-06, "loss": 1.3533, "step": 24168 }, { "epoch": 60.4225, "grad_norm": 9.875, "learning_rate": 7.150273041294009e-06, "loss": 1.3585, "step": 24169 }, { "epoch": 60.425, "grad_norm": 10.125, "learning_rate": 7.149496935041475e-06, "loss": 1.5314, "step": 24170 }, { "epoch": 60.4275, "grad_norm": 12.6875, "learning_rate": 7.148720847476727e-06, "loss": 1.3434, "step": 24171 }, { "epoch": 60.43, "grad_norm": 6.03125, "learning_rate": 7.1479447786048495e-06, "loss": 1.395, "step": 24172 }, { "epoch": 60.4325, "grad_norm": 6.46875, "learning_rate": 7.147168728430929e-06, "loss": 1.1787, "step": 24173 }, { "epoch": 60.435, "grad_norm": 7.375, "learning_rate": 7.146392696960058e-06, "loss": 1.1684, "step": 24174 }, { "epoch": 60.4375, "grad_norm": 8.875, "learning_rate": 7.145616684197325e-06, "loss": 1.5679, "step": 24175 }, { "epoch": 60.44, "grad_norm": 9.6875, "learning_rate": 7.144840690147812e-06, "loss": 1.3531, "step": 24176 }, { "epoch": 60.4425, "grad_norm": 11.4375, "learning_rate": 7.14406471481661e-06, "loss": 1.3677, "step": 24177 }, { "epoch": 60.445, "grad_norm": 11.1875, "learning_rate": 7.143288758208803e-06, "loss": 1.2106, "step": 24178 }, { "epoch": 60.4475, "grad_norm": 7.40625, "learning_rate": 7.142512820329485e-06, "loss": 1.384, "step": 24179 }, { "epoch": 60.45, "grad_norm": 4.28125, "learning_rate": 7.1417369011837355e-06, "loss": 1.198, "step": 24180 }, { "epoch": 60.4525, "grad_norm": 14.0625, "learning_rate": 7.140961000776646e-06, "loss": 1.3275, "step": 24181 }, { "epoch": 60.455, "grad_norm": 9.5625, "learning_rate": 7.140185119113303e-06, "loss": 1.3062, "step": 24182 }, { "epoch": 60.4575, "grad_norm": 9.25, "learning_rate": 7.139409256198786e-06, "loss": 1.218, "step": 24183 }, { "epoch": 60.46, "grad_norm": 6.03125, "learning_rate": 7.138633412038194e-06, "loss": 1.2835, "step": 24184 }, { "epoch": 60.4625, "grad_norm": 6.03125, "learning_rate": 7.137857586636603e-06, "loss": 1.2839, "step": 24185 }, { "epoch": 60.465, "grad_norm": 8.3125, "learning_rate": 7.137081779999107e-06, "loss": 1.2203, "step": 24186 }, { "epoch": 60.4675, "grad_norm": 8.5625, "learning_rate": 7.136305992130786e-06, "loss": 1.2708, "step": 24187 }, { "epoch": 60.47, "grad_norm": 8.3125, "learning_rate": 7.135530223036726e-06, "loss": 1.3051, "step": 24188 }, { "epoch": 60.4725, "grad_norm": 11.4375, "learning_rate": 7.134754472722017e-06, "loss": 1.2344, "step": 24189 }, { "epoch": 60.475, "grad_norm": 6.5625, "learning_rate": 7.133978741191745e-06, "loss": 1.229, "step": 24190 }, { "epoch": 60.4775, "grad_norm": 11.0625, "learning_rate": 7.133203028450993e-06, "loss": 1.4682, "step": 24191 }, { "epoch": 60.48, "grad_norm": 5.40625, "learning_rate": 7.132427334504846e-06, "loss": 1.2746, "step": 24192 }, { "epoch": 60.4825, "grad_norm": 8.3125, "learning_rate": 7.131651659358387e-06, "loss": 1.2579, "step": 24193 }, { "epoch": 60.485, "grad_norm": 7.84375, "learning_rate": 7.130876003016711e-06, "loss": 1.3381, "step": 24194 }, { "epoch": 60.4875, "grad_norm": 8.125, "learning_rate": 7.130100365484895e-06, "loss": 1.2903, "step": 24195 }, { "epoch": 60.49, "grad_norm": 6.03125, "learning_rate": 7.129324746768027e-06, "loss": 1.3762, "step": 24196 }, { "epoch": 60.4925, "grad_norm": 9.875, "learning_rate": 7.128549146871192e-06, "loss": 1.2749, "step": 24197 }, { "epoch": 60.495, "grad_norm": 11.5, "learning_rate": 7.12777356579947e-06, "loss": 1.2856, "step": 24198 }, { "epoch": 60.4975, "grad_norm": 9.1875, "learning_rate": 7.126998003557951e-06, "loss": 1.3972, "step": 24199 }, { "epoch": 60.5, "grad_norm": 8.75, "learning_rate": 7.126222460151719e-06, "loss": 1.2367, "step": 24200 }, { "epoch": 60.5025, "grad_norm": 7.96875, "learning_rate": 7.125446935585859e-06, "loss": 1.3786, "step": 24201 }, { "epoch": 60.505, "grad_norm": 8.0, "learning_rate": 7.124671429865453e-06, "loss": 1.2299, "step": 24202 }, { "epoch": 60.5075, "grad_norm": 6.34375, "learning_rate": 7.123895942995583e-06, "loss": 1.2688, "step": 24203 }, { "epoch": 60.51, "grad_norm": 6.75, "learning_rate": 7.123120474981338e-06, "loss": 1.4254, "step": 24204 }, { "epoch": 60.5125, "grad_norm": 6.59375, "learning_rate": 7.122345025827802e-06, "loss": 1.3609, "step": 24205 }, { "epoch": 60.515, "grad_norm": 10.375, "learning_rate": 7.121569595540054e-06, "loss": 1.2636, "step": 24206 }, { "epoch": 60.5175, "grad_norm": 6.9375, "learning_rate": 7.120794184123183e-06, "loss": 1.2812, "step": 24207 }, { "epoch": 60.52, "grad_norm": 6.5, "learning_rate": 7.120018791582266e-06, "loss": 1.3668, "step": 24208 }, { "epoch": 60.5225, "grad_norm": 7.59375, "learning_rate": 7.119243417922396e-06, "loss": 1.2831, "step": 24209 }, { "epoch": 60.525, "grad_norm": 6.84375, "learning_rate": 7.118468063148647e-06, "loss": 1.4191, "step": 24210 }, { "epoch": 60.5275, "grad_norm": 6.46875, "learning_rate": 7.117692727266108e-06, "loss": 1.3654, "step": 24211 }, { "epoch": 60.53, "grad_norm": 7.3125, "learning_rate": 7.116917410279862e-06, "loss": 1.2913, "step": 24212 }, { "epoch": 60.5325, "grad_norm": 13.4375, "learning_rate": 7.116142112194985e-06, "loss": 1.3422, "step": 24213 }, { "epoch": 60.535, "grad_norm": 7.75, "learning_rate": 7.115366833016567e-06, "loss": 1.2667, "step": 24214 }, { "epoch": 60.5375, "grad_norm": 8.5625, "learning_rate": 7.114591572749689e-06, "loss": 1.3626, "step": 24215 }, { "epoch": 60.54, "grad_norm": 6.34375, "learning_rate": 7.113816331399435e-06, "loss": 1.2774, "step": 24216 }, { "epoch": 60.5425, "grad_norm": 7.875, "learning_rate": 7.113041108970884e-06, "loss": 1.4062, "step": 24217 }, { "epoch": 60.545, "grad_norm": 8.6875, "learning_rate": 7.112265905469118e-06, "loss": 1.3163, "step": 24218 }, { "epoch": 60.5475, "grad_norm": 7.125, "learning_rate": 7.111490720899223e-06, "loss": 1.3079, "step": 24219 }, { "epoch": 60.55, "grad_norm": 8.9375, "learning_rate": 7.110715555266281e-06, "loss": 1.3412, "step": 24220 }, { "epoch": 60.5525, "grad_norm": 5.71875, "learning_rate": 7.109940408575371e-06, "loss": 1.405, "step": 24221 }, { "epoch": 60.555, "grad_norm": 6.46875, "learning_rate": 7.109165280831576e-06, "loss": 1.2562, "step": 24222 }, { "epoch": 60.5575, "grad_norm": 8.375, "learning_rate": 7.108390172039974e-06, "loss": 1.3534, "step": 24223 }, { "epoch": 60.56, "grad_norm": 10.375, "learning_rate": 7.107615082205654e-06, "loss": 1.2505, "step": 24224 }, { "epoch": 60.5625, "grad_norm": 9.625, "learning_rate": 7.106840011333694e-06, "loss": 1.508, "step": 24225 }, { "epoch": 60.565, "grad_norm": 5.5, "learning_rate": 7.106064959429176e-06, "loss": 1.3825, "step": 24226 }, { "epoch": 60.5675, "grad_norm": 10.125, "learning_rate": 7.105289926497178e-06, "loss": 1.3433, "step": 24227 }, { "epoch": 60.57, "grad_norm": 6.25, "learning_rate": 7.104514912542784e-06, "loss": 1.415, "step": 24228 }, { "epoch": 60.5725, "grad_norm": 11.0625, "learning_rate": 7.103739917571073e-06, "loss": 1.3148, "step": 24229 }, { "epoch": 60.575, "grad_norm": 7.3125, "learning_rate": 7.1029649415871296e-06, "loss": 1.4385, "step": 24230 }, { "epoch": 60.5775, "grad_norm": 9.3125, "learning_rate": 7.102189984596033e-06, "loss": 1.2937, "step": 24231 }, { "epoch": 60.58, "grad_norm": 8.75, "learning_rate": 7.1014150466028605e-06, "loss": 1.3841, "step": 24232 }, { "epoch": 60.5825, "grad_norm": 6.375, "learning_rate": 7.1006401276126935e-06, "loss": 1.222, "step": 24233 }, { "epoch": 60.585, "grad_norm": 6.5625, "learning_rate": 7.099865227630616e-06, "loss": 1.2458, "step": 24234 }, { "epoch": 60.5875, "grad_norm": 6.3125, "learning_rate": 7.099090346661708e-06, "loss": 1.4424, "step": 24235 }, { "epoch": 60.59, "grad_norm": 7.625, "learning_rate": 7.0983154847110445e-06, "loss": 1.3845, "step": 24236 }, { "epoch": 60.5925, "grad_norm": 7.5625, "learning_rate": 7.097540641783711e-06, "loss": 1.243, "step": 24237 }, { "epoch": 60.595, "grad_norm": 8.1875, "learning_rate": 7.09676581788478e-06, "loss": 1.3526, "step": 24238 }, { "epoch": 60.5975, "grad_norm": 10.375, "learning_rate": 7.095991013019341e-06, "loss": 1.4444, "step": 24239 }, { "epoch": 60.6, "grad_norm": 14.0625, "learning_rate": 7.095216227192467e-06, "loss": 1.1864, "step": 24240 }, { "epoch": 60.6025, "grad_norm": 10.6875, "learning_rate": 7.0944414604092425e-06, "loss": 1.3922, "step": 24241 }, { "epoch": 60.605, "grad_norm": 5.65625, "learning_rate": 7.09366671267474e-06, "loss": 1.2516, "step": 24242 }, { "epoch": 60.6075, "grad_norm": 6.875, "learning_rate": 7.09289198399404e-06, "loss": 1.1304, "step": 24243 }, { "epoch": 60.61, "grad_norm": 6.15625, "learning_rate": 7.092117274372227e-06, "loss": 1.268, "step": 24244 }, { "epoch": 60.6125, "grad_norm": 9.875, "learning_rate": 7.091342583814378e-06, "loss": 1.3683, "step": 24245 }, { "epoch": 60.615, "grad_norm": 9.1875, "learning_rate": 7.090567912325569e-06, "loss": 1.3707, "step": 24246 }, { "epoch": 60.6175, "grad_norm": 5.25, "learning_rate": 7.08979325991088e-06, "loss": 1.3285, "step": 24247 }, { "epoch": 60.62, "grad_norm": 8.25, "learning_rate": 7.089018626575388e-06, "loss": 1.3197, "step": 24248 }, { "epoch": 60.6225, "grad_norm": 7.71875, "learning_rate": 7.088244012324177e-06, "loss": 1.3432, "step": 24249 }, { "epoch": 60.625, "grad_norm": 9.375, "learning_rate": 7.08746941716232e-06, "loss": 1.3111, "step": 24250 }, { "epoch": 60.6275, "grad_norm": 5.71875, "learning_rate": 7.086694841094898e-06, "loss": 1.2101, "step": 24251 }, { "epoch": 60.63, "grad_norm": 6.34375, "learning_rate": 7.085920284126989e-06, "loss": 1.3775, "step": 24252 }, { "epoch": 60.6325, "grad_norm": 5.6875, "learning_rate": 7.085145746263664e-06, "loss": 1.3297, "step": 24253 }, { "epoch": 60.635, "grad_norm": 7.5625, "learning_rate": 7.084371227510013e-06, "loss": 1.2741, "step": 24254 }, { "epoch": 60.6375, "grad_norm": 7.34375, "learning_rate": 7.083596727871105e-06, "loss": 1.3539, "step": 24255 }, { "epoch": 60.64, "grad_norm": 6.71875, "learning_rate": 7.082822247352024e-06, "loss": 1.3534, "step": 24256 }, { "epoch": 60.6425, "grad_norm": 5.65625, "learning_rate": 7.082047785957839e-06, "loss": 1.27, "step": 24257 }, { "epoch": 60.645, "grad_norm": 8.5625, "learning_rate": 7.081273343693631e-06, "loss": 1.3561, "step": 24258 }, { "epoch": 60.6475, "grad_norm": 6.5625, "learning_rate": 7.080498920564481e-06, "loss": 1.3665, "step": 24259 }, { "epoch": 60.65, "grad_norm": 6.875, "learning_rate": 7.0797245165754654e-06, "loss": 1.2998, "step": 24260 }, { "epoch": 60.6525, "grad_norm": 7.21875, "learning_rate": 7.078950131731655e-06, "loss": 1.3416, "step": 24261 }, { "epoch": 60.655, "grad_norm": 12.6875, "learning_rate": 7.078175766038133e-06, "loss": 1.4435, "step": 24262 }, { "epoch": 60.6575, "grad_norm": 8.6875, "learning_rate": 7.077401419499969e-06, "loss": 1.3683, "step": 24263 }, { "epoch": 60.66, "grad_norm": 6.15625, "learning_rate": 7.076627092122251e-06, "loss": 1.268, "step": 24264 }, { "epoch": 60.6625, "grad_norm": 10.1875, "learning_rate": 7.075852783910047e-06, "loss": 1.1689, "step": 24265 }, { "epoch": 60.665, "grad_norm": 5.65625, "learning_rate": 7.075078494868433e-06, "loss": 1.3226, "step": 24266 }, { "epoch": 60.6675, "grad_norm": 9.25, "learning_rate": 7.074304225002491e-06, "loss": 1.3323, "step": 24267 }, { "epoch": 60.67, "grad_norm": 11.0, "learning_rate": 7.073529974317288e-06, "loss": 1.5608, "step": 24268 }, { "epoch": 60.6725, "grad_norm": 6.46875, "learning_rate": 7.072755742817909e-06, "loss": 1.2949, "step": 24269 }, { "epoch": 60.675, "grad_norm": 5.90625, "learning_rate": 7.071981530509424e-06, "loss": 1.3447, "step": 24270 }, { "epoch": 60.6775, "grad_norm": 7.90625, "learning_rate": 7.0712073373969145e-06, "loss": 1.2069, "step": 24271 }, { "epoch": 60.68, "grad_norm": 9.3125, "learning_rate": 7.070433163485451e-06, "loss": 1.2822, "step": 24272 }, { "epoch": 60.6825, "grad_norm": 9.1875, "learning_rate": 7.069659008780108e-06, "loss": 1.4704, "step": 24273 }, { "epoch": 60.685, "grad_norm": 7.0625, "learning_rate": 7.068884873285964e-06, "loss": 1.2378, "step": 24274 }, { "epoch": 60.6875, "grad_norm": 7.46875, "learning_rate": 7.068110757008095e-06, "loss": 1.2896, "step": 24275 }, { "epoch": 60.69, "grad_norm": 7.8125, "learning_rate": 7.067336659951576e-06, "loss": 1.3156, "step": 24276 }, { "epoch": 60.6925, "grad_norm": 10.9375, "learning_rate": 7.066562582121477e-06, "loss": 1.2866, "step": 24277 }, { "epoch": 60.695, "grad_norm": 13.6875, "learning_rate": 7.065788523522876e-06, "loss": 1.3316, "step": 24278 }, { "epoch": 60.6975, "grad_norm": 6.71875, "learning_rate": 7.065014484160853e-06, "loss": 1.2294, "step": 24279 }, { "epoch": 60.7, "grad_norm": 6.625, "learning_rate": 7.064240464040472e-06, "loss": 1.2641, "step": 24280 }, { "epoch": 60.7025, "grad_norm": 8.875, "learning_rate": 7.0634664631668174e-06, "loss": 1.2836, "step": 24281 }, { "epoch": 60.705, "grad_norm": 6.59375, "learning_rate": 7.062692481544959e-06, "loss": 1.276, "step": 24282 }, { "epoch": 60.7075, "grad_norm": 6.8125, "learning_rate": 7.061918519179967e-06, "loss": 1.2287, "step": 24283 }, { "epoch": 60.71, "grad_norm": 6.71875, "learning_rate": 7.0611445760769216e-06, "loss": 1.3001, "step": 24284 }, { "epoch": 60.7125, "grad_norm": 6.78125, "learning_rate": 7.060370652240895e-06, "loss": 1.2357, "step": 24285 }, { "epoch": 60.715, "grad_norm": 8.375, "learning_rate": 7.059596747676963e-06, "loss": 1.4196, "step": 24286 }, { "epoch": 60.7175, "grad_norm": 9.3125, "learning_rate": 7.058822862390195e-06, "loss": 1.2696, "step": 24287 }, { "epoch": 60.72, "grad_norm": 7.78125, "learning_rate": 7.0580489963856646e-06, "loss": 1.2744, "step": 24288 }, { "epoch": 60.7225, "grad_norm": 6.875, "learning_rate": 7.0572751496684485e-06, "loss": 1.4014, "step": 24289 }, { "epoch": 60.725, "grad_norm": 7.1875, "learning_rate": 7.0565013222436215e-06, "loss": 1.3662, "step": 24290 }, { "epoch": 60.7275, "grad_norm": 10.375, "learning_rate": 7.055727514116253e-06, "loss": 1.502, "step": 24291 }, { "epoch": 60.73, "grad_norm": 9.25, "learning_rate": 7.054953725291418e-06, "loss": 1.1843, "step": 24292 }, { "epoch": 60.7325, "grad_norm": 8.375, "learning_rate": 7.054179955774181e-06, "loss": 1.2385, "step": 24293 }, { "epoch": 60.735, "grad_norm": 7.625, "learning_rate": 7.053406205569631e-06, "loss": 1.1585, "step": 24294 }, { "epoch": 60.7375, "grad_norm": 6.46875, "learning_rate": 7.052632474682828e-06, "loss": 1.4577, "step": 24295 }, { "epoch": 60.74, "grad_norm": 6.40625, "learning_rate": 7.05185876311885e-06, "loss": 1.2045, "step": 24296 }, { "epoch": 60.7425, "grad_norm": 8.875, "learning_rate": 7.05108507088277e-06, "loss": 1.2954, "step": 24297 }, { "epoch": 60.745, "grad_norm": 6.9375, "learning_rate": 7.050311397979652e-06, "loss": 1.3106, "step": 24298 }, { "epoch": 60.7475, "grad_norm": 6.28125, "learning_rate": 7.049537744414578e-06, "loss": 1.25, "step": 24299 }, { "epoch": 60.75, "grad_norm": 5.5, "learning_rate": 7.048764110192618e-06, "loss": 1.2419, "step": 24300 }, { "epoch": 60.7525, "grad_norm": 6.90625, "learning_rate": 7.047990495318843e-06, "loss": 1.3013, "step": 24301 }, { "epoch": 60.755, "grad_norm": 6.625, "learning_rate": 7.047216899798321e-06, "loss": 1.2796, "step": 24302 }, { "epoch": 60.7575, "grad_norm": 8.75, "learning_rate": 7.046443323636127e-06, "loss": 1.3809, "step": 24303 }, { "epoch": 60.76, "grad_norm": 7.1875, "learning_rate": 7.045669766837333e-06, "loss": 1.2563, "step": 24304 }, { "epoch": 60.7625, "grad_norm": 7.84375, "learning_rate": 7.044896229407012e-06, "loss": 1.46, "step": 24305 }, { "epoch": 60.765, "grad_norm": 8.5625, "learning_rate": 7.044122711350232e-06, "loss": 1.3886, "step": 24306 }, { "epoch": 60.7675, "grad_norm": 5.84375, "learning_rate": 7.043349212672066e-06, "loss": 1.2476, "step": 24307 }, { "epoch": 60.77, "grad_norm": 5.71875, "learning_rate": 7.0425757333775794e-06, "loss": 1.4409, "step": 24308 }, { "epoch": 60.7725, "grad_norm": 9.125, "learning_rate": 7.041802273471852e-06, "loss": 1.2259, "step": 24309 }, { "epoch": 60.775, "grad_norm": 8.625, "learning_rate": 7.0410288329599505e-06, "loss": 1.3426, "step": 24310 }, { "epoch": 60.7775, "grad_norm": 7.0, "learning_rate": 7.040255411846946e-06, "loss": 1.2488, "step": 24311 }, { "epoch": 60.78, "grad_norm": 10.125, "learning_rate": 7.039482010137908e-06, "loss": 1.5175, "step": 24312 }, { "epoch": 60.7825, "grad_norm": 8.4375, "learning_rate": 7.038708627837905e-06, "loss": 1.3027, "step": 24313 }, { "epoch": 60.785, "grad_norm": 9.875, "learning_rate": 7.037935264952012e-06, "loss": 1.2941, "step": 24314 }, { "epoch": 60.7875, "grad_norm": 6.90625, "learning_rate": 7.037161921485299e-06, "loss": 1.467, "step": 24315 }, { "epoch": 60.79, "grad_norm": 7.3125, "learning_rate": 7.036388597442829e-06, "loss": 1.2176, "step": 24316 }, { "epoch": 60.7925, "grad_norm": 7.3125, "learning_rate": 7.035615292829681e-06, "loss": 1.1666, "step": 24317 }, { "epoch": 60.795, "grad_norm": 8.875, "learning_rate": 7.034842007650915e-06, "loss": 1.253, "step": 24318 }, { "epoch": 60.7975, "grad_norm": 7.1875, "learning_rate": 7.0340687419116126e-06, "loss": 1.3157, "step": 24319 }, { "epoch": 60.8, "grad_norm": 10.3125, "learning_rate": 7.033295495616834e-06, "loss": 1.226, "step": 24320 }, { "epoch": 60.8025, "grad_norm": 7.09375, "learning_rate": 7.032522268771651e-06, "loss": 1.3512, "step": 24321 }, { "epoch": 60.805, "grad_norm": 8.125, "learning_rate": 7.031749061381136e-06, "loss": 1.3332, "step": 24322 }, { "epoch": 60.8075, "grad_norm": 6.21875, "learning_rate": 7.030975873450349e-06, "loss": 1.2898, "step": 24323 }, { "epoch": 60.81, "grad_norm": 5.9375, "learning_rate": 7.0302027049843706e-06, "loss": 1.5311, "step": 24324 }, { "epoch": 60.8125, "grad_norm": 7.53125, "learning_rate": 7.029429555988263e-06, "loss": 1.3133, "step": 24325 }, { "epoch": 60.815, "grad_norm": 9.0, "learning_rate": 7.0286564264671e-06, "loss": 1.3677, "step": 24326 }, { "epoch": 60.8175, "grad_norm": 5.78125, "learning_rate": 7.027883316425943e-06, "loss": 1.2398, "step": 24327 }, { "epoch": 60.82, "grad_norm": 7.78125, "learning_rate": 7.027110225869862e-06, "loss": 1.2483, "step": 24328 }, { "epoch": 60.8225, "grad_norm": 8.1875, "learning_rate": 7.026337154803928e-06, "loss": 1.3994, "step": 24329 }, { "epoch": 60.825, "grad_norm": 9.1875, "learning_rate": 7.025564103233214e-06, "loss": 1.2286, "step": 24330 }, { "epoch": 60.8275, "grad_norm": 8.6875, "learning_rate": 7.024791071162777e-06, "loss": 1.225, "step": 24331 }, { "epoch": 60.83, "grad_norm": 8.0, "learning_rate": 7.024018058597694e-06, "loss": 1.2703, "step": 24332 }, { "epoch": 60.8325, "grad_norm": 6.3125, "learning_rate": 7.0232450655430255e-06, "loss": 1.3305, "step": 24333 }, { "epoch": 60.835, "grad_norm": 8.5, "learning_rate": 7.022472092003847e-06, "loss": 1.2781, "step": 24334 }, { "epoch": 60.8375, "grad_norm": 5.9375, "learning_rate": 7.021699137985222e-06, "loss": 1.4149, "step": 24335 }, { "epoch": 60.84, "grad_norm": 9.0625, "learning_rate": 7.020926203492218e-06, "loss": 1.1622, "step": 24336 }, { "epoch": 60.8425, "grad_norm": 9.375, "learning_rate": 7.020153288529903e-06, "loss": 1.325, "step": 24337 }, { "epoch": 60.845, "grad_norm": 6.875, "learning_rate": 7.019380393103341e-06, "loss": 1.384, "step": 24338 }, { "epoch": 60.8475, "grad_norm": 7.125, "learning_rate": 7.018607517217604e-06, "loss": 1.2924, "step": 24339 }, { "epoch": 60.85, "grad_norm": 7.53125, "learning_rate": 7.017834660877756e-06, "loss": 1.3883, "step": 24340 }, { "epoch": 60.8525, "grad_norm": 7.59375, "learning_rate": 7.017061824088867e-06, "loss": 1.4274, "step": 24341 }, { "epoch": 60.855, "grad_norm": 11.3125, "learning_rate": 7.016289006855998e-06, "loss": 1.2304, "step": 24342 }, { "epoch": 60.8575, "grad_norm": 8.375, "learning_rate": 7.0155162091842185e-06, "loss": 1.313, "step": 24343 }, { "epoch": 60.86, "grad_norm": 5.71875, "learning_rate": 7.0147434310785975e-06, "loss": 1.3286, "step": 24344 }, { "epoch": 60.8625, "grad_norm": 8.9375, "learning_rate": 7.013970672544201e-06, "loss": 1.368, "step": 24345 }, { "epoch": 60.865, "grad_norm": 7.65625, "learning_rate": 7.013197933586091e-06, "loss": 1.1935, "step": 24346 }, { "epoch": 60.8675, "grad_norm": 9.125, "learning_rate": 7.012425214209335e-06, "loss": 1.3687, "step": 24347 }, { "epoch": 60.87, "grad_norm": 9.6875, "learning_rate": 7.011652514418998e-06, "loss": 1.4451, "step": 24348 }, { "epoch": 60.8725, "grad_norm": 8.3125, "learning_rate": 7.010879834220152e-06, "loss": 1.2647, "step": 24349 }, { "epoch": 60.875, "grad_norm": 6.53125, "learning_rate": 7.010107173617857e-06, "loss": 1.4329, "step": 24350 }, { "epoch": 60.8775, "grad_norm": 6.375, "learning_rate": 7.009334532617179e-06, "loss": 1.1786, "step": 24351 }, { "epoch": 60.88, "grad_norm": 7.25, "learning_rate": 7.008561911223186e-06, "loss": 1.3091, "step": 24352 }, { "epoch": 60.8825, "grad_norm": 8.75, "learning_rate": 7.0077893094409375e-06, "loss": 1.448, "step": 24353 }, { "epoch": 60.885, "grad_norm": 14.25, "learning_rate": 7.0070167272755045e-06, "loss": 1.391, "step": 24354 }, { "epoch": 60.8875, "grad_norm": 6.46875, "learning_rate": 7.0062441647319504e-06, "loss": 1.4616, "step": 24355 }, { "epoch": 60.89, "grad_norm": 5.8125, "learning_rate": 7.005471621815341e-06, "loss": 1.3434, "step": 24356 }, { "epoch": 60.8925, "grad_norm": 6.9375, "learning_rate": 7.004699098530738e-06, "loss": 1.4693, "step": 24357 }, { "epoch": 60.895, "grad_norm": 7.78125, "learning_rate": 7.003926594883206e-06, "loss": 1.3403, "step": 24358 }, { "epoch": 60.8975, "grad_norm": 5.0, "learning_rate": 7.003154110877813e-06, "loss": 1.1416, "step": 24359 }, { "epoch": 60.9, "grad_norm": 7.09375, "learning_rate": 7.002381646519625e-06, "loss": 1.2734, "step": 24360 }, { "epoch": 60.9025, "grad_norm": 6.9375, "learning_rate": 7.0016092018136995e-06, "loss": 1.5763, "step": 24361 }, { "epoch": 60.905, "grad_norm": 8.3125, "learning_rate": 7.000836776765103e-06, "loss": 1.3914, "step": 24362 }, { "epoch": 60.9075, "grad_norm": 9.0, "learning_rate": 7.000064371378902e-06, "loss": 1.3159, "step": 24363 }, { "epoch": 60.91, "grad_norm": 8.5, "learning_rate": 6.9992919856601615e-06, "loss": 1.2271, "step": 24364 }, { "epoch": 60.9125, "grad_norm": 6.53125, "learning_rate": 6.998519619613939e-06, "loss": 1.2165, "step": 24365 }, { "epoch": 60.915, "grad_norm": 8.375, "learning_rate": 6.997747273245303e-06, "loss": 1.3041, "step": 24366 }, { "epoch": 60.9175, "grad_norm": 5.40625, "learning_rate": 6.996974946559314e-06, "loss": 1.4131, "step": 24367 }, { "epoch": 60.92, "grad_norm": 5.40625, "learning_rate": 6.9962026395610416e-06, "loss": 1.2733, "step": 24368 }, { "epoch": 60.9225, "grad_norm": 7.875, "learning_rate": 6.995430352255541e-06, "loss": 1.2825, "step": 24369 }, { "epoch": 60.925, "grad_norm": 6.40625, "learning_rate": 6.99465808464788e-06, "loss": 1.3465, "step": 24370 }, { "epoch": 60.9275, "grad_norm": 9.875, "learning_rate": 6.9938858367431214e-06, "loss": 1.464, "step": 24371 }, { "epoch": 60.93, "grad_norm": 7.90625, "learning_rate": 6.993113608546325e-06, "loss": 1.1537, "step": 24372 }, { "epoch": 60.9325, "grad_norm": 9.75, "learning_rate": 6.9923414000625544e-06, "loss": 1.2983, "step": 24373 }, { "epoch": 60.935, "grad_norm": 7.78125, "learning_rate": 6.991569211296876e-06, "loss": 1.5176, "step": 24374 }, { "epoch": 60.9375, "grad_norm": 10.5, "learning_rate": 6.990797042254349e-06, "loss": 1.4192, "step": 24375 }, { "epoch": 60.94, "grad_norm": 9.125, "learning_rate": 6.990024892940036e-06, "loss": 1.2767, "step": 24376 }, { "epoch": 60.9425, "grad_norm": 6.8125, "learning_rate": 6.989252763358997e-06, "loss": 1.2826, "step": 24377 }, { "epoch": 60.945, "grad_norm": 6.78125, "learning_rate": 6.988480653516299e-06, "loss": 1.3455, "step": 24378 }, { "epoch": 60.9475, "grad_norm": 6.1875, "learning_rate": 6.987708563417002e-06, "loss": 1.3107, "step": 24379 }, { "epoch": 60.95, "grad_norm": 9.1875, "learning_rate": 6.986936493066165e-06, "loss": 1.3595, "step": 24380 }, { "epoch": 60.9525, "grad_norm": 9.75, "learning_rate": 6.9861644424688544e-06, "loss": 1.3459, "step": 24381 }, { "epoch": 60.955, "grad_norm": 6.4375, "learning_rate": 6.985392411630123e-06, "loss": 1.3709, "step": 24382 }, { "epoch": 60.9575, "grad_norm": 7.3125, "learning_rate": 6.984620400555044e-06, "loss": 1.3697, "step": 24383 }, { "epoch": 60.96, "grad_norm": 6.25, "learning_rate": 6.983848409248672e-06, "loss": 1.2858, "step": 24384 }, { "epoch": 60.9625, "grad_norm": 7.71875, "learning_rate": 6.98307643771607e-06, "loss": 1.0913, "step": 24385 }, { "epoch": 60.965, "grad_norm": 7.0625, "learning_rate": 6.9823044859622965e-06, "loss": 1.3654, "step": 24386 }, { "epoch": 60.9675, "grad_norm": 8.375, "learning_rate": 6.9815325539924116e-06, "loss": 1.3282, "step": 24387 }, { "epoch": 60.97, "grad_norm": 11.0, "learning_rate": 6.980760641811481e-06, "loss": 1.3245, "step": 24388 }, { "epoch": 60.9725, "grad_norm": 6.90625, "learning_rate": 6.979988749424564e-06, "loss": 1.2866, "step": 24389 }, { "epoch": 60.975, "grad_norm": 8.6875, "learning_rate": 6.979216876836719e-06, "loss": 1.3225, "step": 24390 }, { "epoch": 60.9775, "grad_norm": 6.84375, "learning_rate": 6.978445024053008e-06, "loss": 1.2562, "step": 24391 }, { "epoch": 60.98, "grad_norm": 10.5, "learning_rate": 6.977673191078487e-06, "loss": 1.2247, "step": 24392 }, { "epoch": 60.9825, "grad_norm": 10.375, "learning_rate": 6.976901377918221e-06, "loss": 1.2173, "step": 24393 }, { "epoch": 60.985, "grad_norm": 7.5, "learning_rate": 6.976129584577274e-06, "loss": 1.6164, "step": 24394 }, { "epoch": 60.9875, "grad_norm": 6.15625, "learning_rate": 6.975357811060695e-06, "loss": 1.2135, "step": 24395 }, { "epoch": 60.99, "grad_norm": 6.09375, "learning_rate": 6.974586057373552e-06, "loss": 1.4042, "step": 24396 }, { "epoch": 60.9925, "grad_norm": 7.5, "learning_rate": 6.973814323520897e-06, "loss": 1.3858, "step": 24397 }, { "epoch": 60.995, "grad_norm": 8.625, "learning_rate": 6.973042609507799e-06, "loss": 1.39, "step": 24398 }, { "epoch": 60.9975, "grad_norm": 7.59375, "learning_rate": 6.97227091533931e-06, "loss": 1.3484, "step": 24399 }, { "epoch": 61.0, "grad_norm": 7.4375, "learning_rate": 6.971499241020495e-06, "loss": 1.3098, "step": 24400 }, { "epoch": 61.0025, "grad_norm": 6.15625, "learning_rate": 6.970727586556407e-06, "loss": 1.5175, "step": 24401 }, { "epoch": 61.005, "grad_norm": 9.6875, "learning_rate": 6.969955951952105e-06, "loss": 1.235, "step": 24402 }, { "epoch": 61.0075, "grad_norm": 9.5625, "learning_rate": 6.969184337212655e-06, "loss": 1.3934, "step": 24403 }, { "epoch": 61.01, "grad_norm": 6.34375, "learning_rate": 6.968412742343112e-06, "loss": 1.3971, "step": 24404 }, { "epoch": 61.0125, "grad_norm": 6.9375, "learning_rate": 6.967641167348531e-06, "loss": 1.2646, "step": 24405 }, { "epoch": 61.015, "grad_norm": 8.75, "learning_rate": 6.966869612233973e-06, "loss": 1.3564, "step": 24406 }, { "epoch": 61.0175, "grad_norm": 7.4375, "learning_rate": 6.966098077004495e-06, "loss": 1.2263, "step": 24407 }, { "epoch": 61.02, "grad_norm": 8.0625, "learning_rate": 6.96532656166516e-06, "loss": 1.3157, "step": 24408 }, { "epoch": 61.0225, "grad_norm": 8.125, "learning_rate": 6.964555066221022e-06, "loss": 1.1178, "step": 24409 }, { "epoch": 61.025, "grad_norm": 6.84375, "learning_rate": 6.963783590677139e-06, "loss": 1.2206, "step": 24410 }, { "epoch": 61.0275, "grad_norm": 7.9375, "learning_rate": 6.963012135038572e-06, "loss": 1.3914, "step": 24411 }, { "epoch": 61.03, "grad_norm": 11.75, "learning_rate": 6.962240699310369e-06, "loss": 1.4258, "step": 24412 }, { "epoch": 61.0325, "grad_norm": 6.34375, "learning_rate": 6.9614692834975995e-06, "loss": 1.5034, "step": 24413 }, { "epoch": 61.035, "grad_norm": 6.9375, "learning_rate": 6.960697887605314e-06, "loss": 1.388, "step": 24414 }, { "epoch": 61.0375, "grad_norm": 7.6875, "learning_rate": 6.959926511638575e-06, "loss": 1.3308, "step": 24415 }, { "epoch": 61.04, "grad_norm": 5.65625, "learning_rate": 6.959155155602433e-06, "loss": 1.3532, "step": 24416 }, { "epoch": 61.0425, "grad_norm": 8.6875, "learning_rate": 6.958383819501946e-06, "loss": 1.5006, "step": 24417 }, { "epoch": 61.045, "grad_norm": 8.625, "learning_rate": 6.9576125033421744e-06, "loss": 1.2625, "step": 24418 }, { "epoch": 61.0475, "grad_norm": 10.25, "learning_rate": 6.956841207128177e-06, "loss": 1.2612, "step": 24419 }, { "epoch": 61.05, "grad_norm": 8.4375, "learning_rate": 6.956069930865005e-06, "loss": 1.2312, "step": 24420 }, { "epoch": 61.0525, "grad_norm": 9.6875, "learning_rate": 6.9552986745577155e-06, "loss": 1.3756, "step": 24421 }, { "epoch": 61.055, "grad_norm": 7.03125, "learning_rate": 6.954527438211365e-06, "loss": 1.2542, "step": 24422 }, { "epoch": 61.0575, "grad_norm": 5.4375, "learning_rate": 6.953756221831014e-06, "loss": 1.4538, "step": 24423 }, { "epoch": 61.06, "grad_norm": 8.5625, "learning_rate": 6.9529850254217135e-06, "loss": 1.3331, "step": 24424 }, { "epoch": 61.0625, "grad_norm": 8.3125, "learning_rate": 6.952213848988522e-06, "loss": 1.2133, "step": 24425 }, { "epoch": 61.065, "grad_norm": 7.875, "learning_rate": 6.951442692536495e-06, "loss": 1.3059, "step": 24426 }, { "epoch": 61.0675, "grad_norm": 7.71875, "learning_rate": 6.950671556070685e-06, "loss": 1.2124, "step": 24427 }, { "epoch": 61.07, "grad_norm": 7.3125, "learning_rate": 6.949900439596151e-06, "loss": 1.2376, "step": 24428 }, { "epoch": 61.0725, "grad_norm": 11.125, "learning_rate": 6.949129343117949e-06, "loss": 1.2991, "step": 24429 }, { "epoch": 61.075, "grad_norm": 10.0, "learning_rate": 6.948358266641136e-06, "loss": 1.3179, "step": 24430 }, { "epoch": 61.0775, "grad_norm": 7.21875, "learning_rate": 6.947587210170759e-06, "loss": 1.2987, "step": 24431 }, { "epoch": 61.08, "grad_norm": 7.28125, "learning_rate": 6.946816173711878e-06, "loss": 1.3306, "step": 24432 }, { "epoch": 61.0825, "grad_norm": 7.875, "learning_rate": 6.94604515726955e-06, "loss": 1.3219, "step": 24433 }, { "epoch": 61.085, "grad_norm": 7.5625, "learning_rate": 6.945274160848829e-06, "loss": 1.3029, "step": 24434 }, { "epoch": 61.0875, "grad_norm": 8.625, "learning_rate": 6.944503184454768e-06, "loss": 1.1844, "step": 24435 }, { "epoch": 61.09, "grad_norm": 6.03125, "learning_rate": 6.9437322280924195e-06, "loss": 1.1468, "step": 24436 }, { "epoch": 61.0925, "grad_norm": 7.0625, "learning_rate": 6.942961291766839e-06, "loss": 1.3429, "step": 24437 }, { "epoch": 61.095, "grad_norm": 12.625, "learning_rate": 6.942190375483086e-06, "loss": 1.3651, "step": 24438 }, { "epoch": 61.0975, "grad_norm": 8.6875, "learning_rate": 6.941419479246209e-06, "loss": 1.2742, "step": 24439 }, { "epoch": 61.1, "grad_norm": 6.84375, "learning_rate": 6.940648603061263e-06, "loss": 1.2947, "step": 24440 }, { "epoch": 61.1025, "grad_norm": 8.125, "learning_rate": 6.939877746933305e-06, "loss": 1.3483, "step": 24441 }, { "epoch": 61.105, "grad_norm": 8.375, "learning_rate": 6.939106910867381e-06, "loss": 1.3949, "step": 24442 }, { "epoch": 61.1075, "grad_norm": 7.59375, "learning_rate": 6.938336094868552e-06, "loss": 1.3439, "step": 24443 }, { "epoch": 61.11, "grad_norm": 6.78125, "learning_rate": 6.93756529894187e-06, "loss": 1.3171, "step": 24444 }, { "epoch": 61.1125, "grad_norm": 6.9375, "learning_rate": 6.936794523092388e-06, "loss": 1.3753, "step": 24445 }, { "epoch": 61.115, "grad_norm": 7.84375, "learning_rate": 6.936023767325157e-06, "loss": 1.3053, "step": 24446 }, { "epoch": 61.1175, "grad_norm": 13.25, "learning_rate": 6.935253031645231e-06, "loss": 1.286, "step": 24447 }, { "epoch": 61.12, "grad_norm": 9.25, "learning_rate": 6.934482316057663e-06, "loss": 1.3175, "step": 24448 }, { "epoch": 61.1225, "grad_norm": 7.875, "learning_rate": 6.933711620567509e-06, "loss": 1.345, "step": 24449 }, { "epoch": 61.125, "grad_norm": 8.5625, "learning_rate": 6.932940945179818e-06, "loss": 1.2859, "step": 24450 }, { "epoch": 61.1275, "grad_norm": 5.875, "learning_rate": 6.932170289899644e-06, "loss": 1.1677, "step": 24451 }, { "epoch": 61.13, "grad_norm": 5.125, "learning_rate": 6.931399654732034e-06, "loss": 1.3876, "step": 24452 }, { "epoch": 61.1325, "grad_norm": 5.53125, "learning_rate": 6.930629039682051e-06, "loss": 1.3991, "step": 24453 }, { "epoch": 61.135, "grad_norm": 7.0, "learning_rate": 6.929858444754738e-06, "loss": 1.3264, "step": 24454 }, { "epoch": 61.1375, "grad_norm": 7.125, "learning_rate": 6.9290878699551535e-06, "loss": 1.1962, "step": 24455 }, { "epoch": 61.14, "grad_norm": 11.125, "learning_rate": 6.928317315288344e-06, "loss": 1.2862, "step": 24456 }, { "epoch": 61.1425, "grad_norm": 8.3125, "learning_rate": 6.92754678075936e-06, "loss": 1.3107, "step": 24457 }, { "epoch": 61.145, "grad_norm": 5.71875, "learning_rate": 6.92677626637326e-06, "loss": 1.2323, "step": 24458 }, { "epoch": 61.1475, "grad_norm": 8.6875, "learning_rate": 6.926005772135093e-06, "loss": 1.3277, "step": 24459 }, { "epoch": 61.15, "grad_norm": 8.875, "learning_rate": 6.925235298049906e-06, "loss": 1.4565, "step": 24460 }, { "epoch": 61.1525, "grad_norm": 6.3125, "learning_rate": 6.924464844122756e-06, "loss": 1.458, "step": 24461 }, { "epoch": 61.155, "grad_norm": 7.46875, "learning_rate": 6.923694410358686e-06, "loss": 1.2595, "step": 24462 }, { "epoch": 61.1575, "grad_norm": 7.59375, "learning_rate": 6.922923996762757e-06, "loss": 1.189, "step": 24463 }, { "epoch": 61.16, "grad_norm": 7.65625, "learning_rate": 6.922153603340016e-06, "loss": 1.3964, "step": 24464 }, { "epoch": 61.1625, "grad_norm": 10.6875, "learning_rate": 6.9213832300955104e-06, "loss": 1.3281, "step": 24465 }, { "epoch": 61.165, "grad_norm": 6.875, "learning_rate": 6.9206128770342975e-06, "loss": 1.255, "step": 24466 }, { "epoch": 61.1675, "grad_norm": 5.71875, "learning_rate": 6.919842544161414e-06, "loss": 1.2535, "step": 24467 }, { "epoch": 61.17, "grad_norm": 14.375, "learning_rate": 6.919072231481929e-06, "loss": 1.4938, "step": 24468 }, { "epoch": 61.1725, "grad_norm": 6.625, "learning_rate": 6.918301939000879e-06, "loss": 1.3561, "step": 24469 }, { "epoch": 61.175, "grad_norm": 7.375, "learning_rate": 6.917531666723321e-06, "loss": 1.493, "step": 24470 }, { "epoch": 61.1775, "grad_norm": 5.40625, "learning_rate": 6.9167614146542995e-06, "loss": 1.4517, "step": 24471 }, { "epoch": 61.18, "grad_norm": 7.625, "learning_rate": 6.915991182798865e-06, "loss": 1.2442, "step": 24472 }, { "epoch": 61.1825, "grad_norm": 11.8125, "learning_rate": 6.915220971162071e-06, "loss": 1.451, "step": 24473 }, { "epoch": 61.185, "grad_norm": 8.625, "learning_rate": 6.914450779748967e-06, "loss": 1.2886, "step": 24474 }, { "epoch": 61.1875, "grad_norm": 10.4375, "learning_rate": 6.913680608564597e-06, "loss": 1.3931, "step": 24475 }, { "epoch": 61.19, "grad_norm": 6.40625, "learning_rate": 6.912910457614015e-06, "loss": 1.3393, "step": 24476 }, { "epoch": 61.1925, "grad_norm": 11.0, "learning_rate": 6.912140326902265e-06, "loss": 1.2685, "step": 24477 }, { "epoch": 61.195, "grad_norm": 9.625, "learning_rate": 6.911370216434404e-06, "loss": 1.3385, "step": 24478 }, { "epoch": 61.1975, "grad_norm": 8.6875, "learning_rate": 6.910600126215474e-06, "loss": 1.2757, "step": 24479 }, { "epoch": 61.2, "grad_norm": 5.75, "learning_rate": 6.909830056250527e-06, "loss": 1.4492, "step": 24480 }, { "epoch": 61.2025, "grad_norm": 10.0625, "learning_rate": 6.909060006544612e-06, "loss": 1.3574, "step": 24481 }, { "epoch": 61.205, "grad_norm": 8.25, "learning_rate": 6.908289977102771e-06, "loss": 1.2968, "step": 24482 }, { "epoch": 61.2075, "grad_norm": 9.0625, "learning_rate": 6.9075199679300604e-06, "loss": 1.1811, "step": 24483 }, { "epoch": 61.21, "grad_norm": 7.375, "learning_rate": 6.906749979031525e-06, "loss": 1.4334, "step": 24484 }, { "epoch": 61.2125, "grad_norm": 5.5625, "learning_rate": 6.905980010412214e-06, "loss": 1.1966, "step": 24485 }, { "epoch": 61.215, "grad_norm": 10.4375, "learning_rate": 6.905210062077175e-06, "loss": 1.478, "step": 24486 }, { "epoch": 61.2175, "grad_norm": 9.9375, "learning_rate": 6.904440134031449e-06, "loss": 1.4403, "step": 24487 }, { "epoch": 61.22, "grad_norm": 7.875, "learning_rate": 6.903670226280095e-06, "loss": 1.3274, "step": 24488 }, { "epoch": 61.2225, "grad_norm": 7.4375, "learning_rate": 6.902900338828156e-06, "loss": 1.4307, "step": 24489 }, { "epoch": 61.225, "grad_norm": 8.1875, "learning_rate": 6.9021304716806745e-06, "loss": 1.2933, "step": 24490 }, { "epoch": 61.2275, "grad_norm": 7.125, "learning_rate": 6.901360624842704e-06, "loss": 1.431, "step": 24491 }, { "epoch": 61.23, "grad_norm": 5.6875, "learning_rate": 6.900590798319287e-06, "loss": 1.2736, "step": 24492 }, { "epoch": 61.2325, "grad_norm": 10.9375, "learning_rate": 6.8998209921154755e-06, "loss": 1.1172, "step": 24493 }, { "epoch": 61.235, "grad_norm": 8.25, "learning_rate": 6.899051206236311e-06, "loss": 1.3216, "step": 24494 }, { "epoch": 61.2375, "grad_norm": 8.0, "learning_rate": 6.898281440686845e-06, "loss": 1.24, "step": 24495 }, { "epoch": 61.24, "grad_norm": 7.71875, "learning_rate": 6.897511695472124e-06, "loss": 1.3757, "step": 24496 }, { "epoch": 61.2425, "grad_norm": 7.65625, "learning_rate": 6.896741970597185e-06, "loss": 1.1264, "step": 24497 }, { "epoch": 61.245, "grad_norm": 6.1875, "learning_rate": 6.8959722660670855e-06, "loss": 1.1936, "step": 24498 }, { "epoch": 61.2475, "grad_norm": 9.1875, "learning_rate": 6.895202581886868e-06, "loss": 1.2231, "step": 24499 }, { "epoch": 61.25, "grad_norm": 12.3125, "learning_rate": 6.894432918061579e-06, "loss": 1.2986, "step": 24500 }, { "epoch": 61.2525, "grad_norm": 7.9375, "learning_rate": 6.893663274596261e-06, "loss": 1.3528, "step": 24501 }, { "epoch": 61.255, "grad_norm": 8.0625, "learning_rate": 6.892893651495962e-06, "loss": 1.2981, "step": 24502 }, { "epoch": 61.2575, "grad_norm": 6.84375, "learning_rate": 6.892124048765729e-06, "loss": 1.1384, "step": 24503 }, { "epoch": 61.26, "grad_norm": 6.15625, "learning_rate": 6.891354466410609e-06, "loss": 1.2642, "step": 24504 }, { "epoch": 61.2625, "grad_norm": 8.9375, "learning_rate": 6.890584904435642e-06, "loss": 1.2218, "step": 24505 }, { "epoch": 61.265, "grad_norm": 8.5625, "learning_rate": 6.889815362845875e-06, "loss": 1.3109, "step": 24506 }, { "epoch": 61.2675, "grad_norm": 5.8125, "learning_rate": 6.889045841646353e-06, "loss": 1.4592, "step": 24507 }, { "epoch": 61.27, "grad_norm": 8.75, "learning_rate": 6.888276340842127e-06, "loss": 1.2066, "step": 24508 }, { "epoch": 61.2725, "grad_norm": 8.375, "learning_rate": 6.887506860438233e-06, "loss": 1.2748, "step": 24509 }, { "epoch": 61.275, "grad_norm": 7.625, "learning_rate": 6.886737400439722e-06, "loss": 1.1575, "step": 24510 }, { "epoch": 61.2775, "grad_norm": 5.75, "learning_rate": 6.885967960851636e-06, "loss": 1.4701, "step": 24511 }, { "epoch": 61.28, "grad_norm": 6.90625, "learning_rate": 6.885198541679016e-06, "loss": 1.24, "step": 24512 }, { "epoch": 61.2825, "grad_norm": 6.125, "learning_rate": 6.884429142926911e-06, "loss": 1.3246, "step": 24513 }, { "epoch": 61.285, "grad_norm": 5.46875, "learning_rate": 6.883659764600364e-06, "loss": 1.3379, "step": 24514 }, { "epoch": 61.2875, "grad_norm": 7.15625, "learning_rate": 6.882890406704422e-06, "loss": 1.3343, "step": 24515 }, { "epoch": 61.29, "grad_norm": 9.875, "learning_rate": 6.882121069244123e-06, "loss": 1.2314, "step": 24516 }, { "epoch": 61.2925, "grad_norm": 9.125, "learning_rate": 6.881351752224512e-06, "loss": 1.3841, "step": 24517 }, { "epoch": 61.295, "grad_norm": 7.375, "learning_rate": 6.880582455650635e-06, "loss": 1.2097, "step": 24518 }, { "epoch": 61.2975, "grad_norm": 11.0, "learning_rate": 6.879813179527537e-06, "loss": 1.3056, "step": 24519 }, { "epoch": 61.3, "grad_norm": 8.1875, "learning_rate": 6.8790439238602576e-06, "loss": 1.3713, "step": 24520 }, { "epoch": 61.3025, "grad_norm": 24.5, "learning_rate": 6.878274688653843e-06, "loss": 1.3892, "step": 24521 }, { "epoch": 61.305, "grad_norm": 10.375, "learning_rate": 6.877505473913328e-06, "loss": 1.2602, "step": 24522 }, { "epoch": 61.3075, "grad_norm": 5.34375, "learning_rate": 6.87673627964377e-06, "loss": 1.3968, "step": 24523 }, { "epoch": 61.31, "grad_norm": 8.125, "learning_rate": 6.8759671058501995e-06, "loss": 1.2554, "step": 24524 }, { "epoch": 61.3125, "grad_norm": 7.09375, "learning_rate": 6.875197952537667e-06, "loss": 1.3062, "step": 24525 }, { "epoch": 61.315, "grad_norm": 19.5, "learning_rate": 6.87442881971121e-06, "loss": 1.2628, "step": 24526 }, { "epoch": 61.3175, "grad_norm": 8.8125, "learning_rate": 6.8736597073758704e-06, "loss": 1.2034, "step": 24527 }, { "epoch": 61.32, "grad_norm": 5.46875, "learning_rate": 6.872890615536694e-06, "loss": 1.1727, "step": 24528 }, { "epoch": 61.3225, "grad_norm": 8.5, "learning_rate": 6.8721215441987224e-06, "loss": 1.3172, "step": 24529 }, { "epoch": 61.325, "grad_norm": 5.6875, "learning_rate": 6.871352493366998e-06, "loss": 1.2415, "step": 24530 }, { "epoch": 61.3275, "grad_norm": 7.53125, "learning_rate": 6.870583463046559e-06, "loss": 1.459, "step": 24531 }, { "epoch": 61.33, "grad_norm": 7.78125, "learning_rate": 6.8698144532424484e-06, "loss": 1.2418, "step": 24532 }, { "epoch": 61.3325, "grad_norm": 6.34375, "learning_rate": 6.869045463959711e-06, "loss": 1.3227, "step": 24533 }, { "epoch": 61.335, "grad_norm": 10.3125, "learning_rate": 6.868276495203388e-06, "loss": 1.2771, "step": 24534 }, { "epoch": 61.3375, "grad_norm": 11.25, "learning_rate": 6.867507546978517e-06, "loss": 1.2344, "step": 24535 }, { "epoch": 61.34, "grad_norm": 6.5, "learning_rate": 6.866738619290143e-06, "loss": 1.1692, "step": 24536 }, { "epoch": 61.3425, "grad_norm": 9.4375, "learning_rate": 6.8659697121433e-06, "loss": 1.2207, "step": 24537 }, { "epoch": 61.345, "grad_norm": 14.875, "learning_rate": 6.8652008255430395e-06, "loss": 1.4828, "step": 24538 }, { "epoch": 61.3475, "grad_norm": 6.3125, "learning_rate": 6.864431959494396e-06, "loss": 1.3129, "step": 24539 }, { "epoch": 61.35, "grad_norm": 9.4375, "learning_rate": 6.863663114002411e-06, "loss": 1.3762, "step": 24540 }, { "epoch": 61.3525, "grad_norm": 6.90625, "learning_rate": 6.862894289072124e-06, "loss": 1.326, "step": 24541 }, { "epoch": 61.355, "grad_norm": 7.34375, "learning_rate": 6.862125484708575e-06, "loss": 1.1923, "step": 24542 }, { "epoch": 61.3575, "grad_norm": 7.3125, "learning_rate": 6.861356700916808e-06, "loss": 1.3081, "step": 24543 }, { "epoch": 61.36, "grad_norm": 7.65625, "learning_rate": 6.860587937701862e-06, "loss": 1.2877, "step": 24544 }, { "epoch": 61.3625, "grad_norm": 5.875, "learning_rate": 6.859819195068775e-06, "loss": 1.4155, "step": 24545 }, { "epoch": 61.365, "grad_norm": 8.0625, "learning_rate": 6.859050473022586e-06, "loss": 1.4851, "step": 24546 }, { "epoch": 61.3675, "grad_norm": 7.15625, "learning_rate": 6.8582817715683355e-06, "loss": 1.3467, "step": 24547 }, { "epoch": 61.37, "grad_norm": 6.75, "learning_rate": 6.857513090711068e-06, "loss": 1.3565, "step": 24548 }, { "epoch": 61.3725, "grad_norm": 7.9375, "learning_rate": 6.856744430455818e-06, "loss": 1.2705, "step": 24549 }, { "epoch": 61.375, "grad_norm": 10.9375, "learning_rate": 6.855975790807623e-06, "loss": 1.3591, "step": 24550 }, { "epoch": 61.3775, "grad_norm": 6.5, "learning_rate": 6.85520717177153e-06, "loss": 1.4224, "step": 24551 }, { "epoch": 61.38, "grad_norm": 8.3125, "learning_rate": 6.8544385733525665e-06, "loss": 1.3474, "step": 24552 }, { "epoch": 61.3825, "grad_norm": 7.0, "learning_rate": 6.8536699955557805e-06, "loss": 1.3611, "step": 24553 }, { "epoch": 61.385, "grad_norm": 6.25, "learning_rate": 6.852901438386208e-06, "loss": 1.4922, "step": 24554 }, { "epoch": 61.3875, "grad_norm": 7.5625, "learning_rate": 6.85213290184889e-06, "loss": 1.3616, "step": 24555 }, { "epoch": 61.39, "grad_norm": 10.75, "learning_rate": 6.851364385948861e-06, "loss": 1.2851, "step": 24556 }, { "epoch": 61.3925, "grad_norm": 9.6875, "learning_rate": 6.850595890691159e-06, "loss": 1.3112, "step": 24557 }, { "epoch": 61.395, "grad_norm": 6.59375, "learning_rate": 6.849827416080825e-06, "loss": 1.4362, "step": 24558 }, { "epoch": 61.3975, "grad_norm": 6.3125, "learning_rate": 6.849058962122901e-06, "loss": 1.3221, "step": 24559 }, { "epoch": 61.4, "grad_norm": 7.8125, "learning_rate": 6.848290528822417e-06, "loss": 1.3849, "step": 24560 }, { "epoch": 61.4025, "grad_norm": 8.5625, "learning_rate": 6.847522116184413e-06, "loss": 1.3918, "step": 24561 }, { "epoch": 61.405, "grad_norm": 7.90625, "learning_rate": 6.846753724213927e-06, "loss": 1.3477, "step": 24562 }, { "epoch": 61.4075, "grad_norm": 7.125, "learning_rate": 6.845985352916001e-06, "loss": 1.3345, "step": 24563 }, { "epoch": 61.41, "grad_norm": 8.8125, "learning_rate": 6.845217002295668e-06, "loss": 1.3171, "step": 24564 }, { "epoch": 61.4125, "grad_norm": 7.75, "learning_rate": 6.844448672357964e-06, "loss": 1.2502, "step": 24565 }, { "epoch": 61.415, "grad_norm": 7.1875, "learning_rate": 6.843680363107932e-06, "loss": 1.2603, "step": 24566 }, { "epoch": 61.4175, "grad_norm": 9.0, "learning_rate": 6.842912074550601e-06, "loss": 1.2927, "step": 24567 }, { "epoch": 61.42, "grad_norm": 5.625, "learning_rate": 6.842143806691013e-06, "loss": 1.1869, "step": 24568 }, { "epoch": 61.4225, "grad_norm": 9.1875, "learning_rate": 6.841375559534204e-06, "loss": 1.3206, "step": 24569 }, { "epoch": 61.425, "grad_norm": 5.53125, "learning_rate": 6.8406073330852145e-06, "loss": 1.2922, "step": 24570 }, { "epoch": 61.4275, "grad_norm": 11.3125, "learning_rate": 6.839839127349073e-06, "loss": 1.3438, "step": 24571 }, { "epoch": 61.43, "grad_norm": 6.34375, "learning_rate": 6.839070942330817e-06, "loss": 1.4763, "step": 24572 }, { "epoch": 61.4325, "grad_norm": 7.1875, "learning_rate": 6.83830277803549e-06, "loss": 1.3045, "step": 24573 }, { "epoch": 61.435, "grad_norm": 7.21875, "learning_rate": 6.837534634468124e-06, "loss": 1.2634, "step": 24574 }, { "epoch": 61.4375, "grad_norm": 8.375, "learning_rate": 6.836766511633752e-06, "loss": 1.33, "step": 24575 }, { "epoch": 61.44, "grad_norm": 6.84375, "learning_rate": 6.835998409537412e-06, "loss": 1.2638, "step": 24576 }, { "epoch": 61.4425, "grad_norm": 6.90625, "learning_rate": 6.835230328184139e-06, "loss": 1.2685, "step": 24577 }, { "epoch": 61.445, "grad_norm": 8.3125, "learning_rate": 6.8344622675789716e-06, "loss": 1.2638, "step": 24578 }, { "epoch": 61.4475, "grad_norm": 6.90625, "learning_rate": 6.833694227726942e-06, "loss": 1.4281, "step": 24579 }, { "epoch": 61.45, "grad_norm": 5.03125, "learning_rate": 6.8329262086330864e-06, "loss": 1.396, "step": 24580 }, { "epoch": 61.4525, "grad_norm": 7.71875, "learning_rate": 6.832158210302442e-06, "loss": 1.4029, "step": 24581 }, { "epoch": 61.455, "grad_norm": 9.625, "learning_rate": 6.831390232740038e-06, "loss": 1.5065, "step": 24582 }, { "epoch": 61.4575, "grad_norm": 6.0, "learning_rate": 6.830622275950913e-06, "loss": 1.5154, "step": 24583 }, { "epoch": 61.46, "grad_norm": 9.5625, "learning_rate": 6.829854339940103e-06, "loss": 1.2861, "step": 24584 }, { "epoch": 61.4625, "grad_norm": 9.625, "learning_rate": 6.829086424712643e-06, "loss": 1.2028, "step": 24585 }, { "epoch": 61.465, "grad_norm": 5.9375, "learning_rate": 6.828318530273563e-06, "loss": 1.2217, "step": 24586 }, { "epoch": 61.4675, "grad_norm": 6.90625, "learning_rate": 6.827550656627898e-06, "loss": 1.3328, "step": 24587 }, { "epoch": 61.47, "grad_norm": 6.09375, "learning_rate": 6.826782803780687e-06, "loss": 1.3384, "step": 24588 }, { "epoch": 61.4725, "grad_norm": 7.0, "learning_rate": 6.826014971736962e-06, "loss": 1.2354, "step": 24589 }, { "epoch": 61.475, "grad_norm": 6.78125, "learning_rate": 6.825247160501754e-06, "loss": 1.2505, "step": 24590 }, { "epoch": 61.4775, "grad_norm": 5.46875, "learning_rate": 6.8244793700801e-06, "loss": 1.192, "step": 24591 }, { "epoch": 61.48, "grad_norm": 7.65625, "learning_rate": 6.823711600477025e-06, "loss": 1.3386, "step": 24592 }, { "epoch": 61.4825, "grad_norm": 8.25, "learning_rate": 6.8229438516975785e-06, "loss": 1.2431, "step": 24593 }, { "epoch": 61.485, "grad_norm": 7.875, "learning_rate": 6.822176123746782e-06, "loss": 1.2828, "step": 24594 }, { "epoch": 61.4875, "grad_norm": 6.71875, "learning_rate": 6.821408416629674e-06, "loss": 1.3461, "step": 24595 }, { "epoch": 61.49, "grad_norm": 7.25, "learning_rate": 6.820640730351281e-06, "loss": 1.3204, "step": 24596 }, { "epoch": 61.4925, "grad_norm": 8.0625, "learning_rate": 6.81987306491664e-06, "loss": 1.2673, "step": 24597 }, { "epoch": 61.495, "grad_norm": 8.25, "learning_rate": 6.819105420330785e-06, "loss": 1.2709, "step": 24598 }, { "epoch": 61.4975, "grad_norm": 6.84375, "learning_rate": 6.818337796598749e-06, "loss": 1.3287, "step": 24599 }, { "epoch": 61.5, "grad_norm": 4.8125, "learning_rate": 6.8175701937255645e-06, "loss": 1.1802, "step": 24600 }, { "epoch": 61.5025, "grad_norm": 5.0, "learning_rate": 6.81680261171626e-06, "loss": 1.2632, "step": 24601 }, { "epoch": 61.505, "grad_norm": 8.9375, "learning_rate": 6.816035050575868e-06, "loss": 1.4181, "step": 24602 }, { "epoch": 61.5075, "grad_norm": 6.4375, "learning_rate": 6.815267510309424e-06, "loss": 1.3987, "step": 24603 }, { "epoch": 61.51, "grad_norm": 7.15625, "learning_rate": 6.814499990921963e-06, "loss": 1.3918, "step": 24604 }, { "epoch": 61.5125, "grad_norm": 6.375, "learning_rate": 6.81373249241851e-06, "loss": 1.3178, "step": 24605 }, { "epoch": 61.515, "grad_norm": 8.375, "learning_rate": 6.812965014804096e-06, "loss": 1.4566, "step": 24606 }, { "epoch": 61.5175, "grad_norm": 8.3125, "learning_rate": 6.812197558083759e-06, "loss": 1.277, "step": 24607 }, { "epoch": 61.52, "grad_norm": 7.125, "learning_rate": 6.811430122262529e-06, "loss": 1.3343, "step": 24608 }, { "epoch": 61.5225, "grad_norm": 9.0625, "learning_rate": 6.810662707345434e-06, "loss": 1.5114, "step": 24609 }, { "epoch": 61.525, "grad_norm": 11.25, "learning_rate": 6.809895313337508e-06, "loss": 1.494, "step": 24610 }, { "epoch": 61.5275, "grad_norm": 8.8125, "learning_rate": 6.809127940243773e-06, "loss": 1.4072, "step": 24611 }, { "epoch": 61.53, "grad_norm": 7.1875, "learning_rate": 6.808360588069276e-06, "loss": 1.2028, "step": 24612 }, { "epoch": 61.5325, "grad_norm": 7.84375, "learning_rate": 6.807593256819036e-06, "loss": 1.3249, "step": 24613 }, { "epoch": 61.535, "grad_norm": 9.125, "learning_rate": 6.80682594649809e-06, "loss": 1.3082, "step": 24614 }, { "epoch": 61.5375, "grad_norm": 12.125, "learning_rate": 6.806058657111463e-06, "loss": 1.3381, "step": 24615 }, { "epoch": 61.54, "grad_norm": 10.625, "learning_rate": 6.805291388664186e-06, "loss": 1.3551, "step": 24616 }, { "epoch": 61.5425, "grad_norm": 9.875, "learning_rate": 6.804524141161292e-06, "loss": 1.3045, "step": 24617 }, { "epoch": 61.545, "grad_norm": 8.625, "learning_rate": 6.803756914607811e-06, "loss": 1.5145, "step": 24618 }, { "epoch": 61.5475, "grad_norm": 9.8125, "learning_rate": 6.802989709008772e-06, "loss": 1.4442, "step": 24619 }, { "epoch": 61.55, "grad_norm": 10.375, "learning_rate": 6.802222524369202e-06, "loss": 1.2941, "step": 24620 }, { "epoch": 61.5525, "grad_norm": 8.625, "learning_rate": 6.801455360694133e-06, "loss": 1.3122, "step": 24621 }, { "epoch": 61.555, "grad_norm": 6.9375, "learning_rate": 6.800688217988598e-06, "loss": 1.3058, "step": 24622 }, { "epoch": 61.5575, "grad_norm": 9.1875, "learning_rate": 6.799921096257621e-06, "loss": 1.36, "step": 24623 }, { "epoch": 61.56, "grad_norm": 7.3125, "learning_rate": 6.799153995506234e-06, "loss": 1.4381, "step": 24624 }, { "epoch": 61.5625, "grad_norm": 6.03125, "learning_rate": 6.798386915739466e-06, "loss": 1.3616, "step": 24625 }, { "epoch": 61.565, "grad_norm": 6.1875, "learning_rate": 6.7976198569623395e-06, "loss": 1.1936, "step": 24626 }, { "epoch": 61.5675, "grad_norm": 6.5625, "learning_rate": 6.7968528191798956e-06, "loss": 1.3207, "step": 24627 }, { "epoch": 61.57, "grad_norm": 4.8125, "learning_rate": 6.796085802397154e-06, "loss": 1.2166, "step": 24628 }, { "epoch": 61.5725, "grad_norm": 10.375, "learning_rate": 6.795318806619149e-06, "loss": 1.5043, "step": 24629 }, { "epoch": 61.575, "grad_norm": 10.25, "learning_rate": 6.794551831850902e-06, "loss": 1.3756, "step": 24630 }, { "epoch": 61.5775, "grad_norm": 7.40625, "learning_rate": 6.793784878097443e-06, "loss": 1.2722, "step": 24631 }, { "epoch": 61.58, "grad_norm": 11.125, "learning_rate": 6.793017945363804e-06, "loss": 1.246, "step": 24632 }, { "epoch": 61.5825, "grad_norm": 9.1875, "learning_rate": 6.7922510336550156e-06, "loss": 1.2732, "step": 24633 }, { "epoch": 61.585, "grad_norm": 6.0625, "learning_rate": 6.791484142976097e-06, "loss": 1.2674, "step": 24634 }, { "epoch": 61.5875, "grad_norm": 6.6875, "learning_rate": 6.79071727333208e-06, "loss": 1.1159, "step": 24635 }, { "epoch": 61.59, "grad_norm": 7.0, "learning_rate": 6.789950424727989e-06, "loss": 1.1803, "step": 24636 }, { "epoch": 61.5925, "grad_norm": 6.03125, "learning_rate": 6.78918359716886e-06, "loss": 1.3503, "step": 24637 }, { "epoch": 61.595, "grad_norm": 8.625, "learning_rate": 6.7884167906597135e-06, "loss": 1.2887, "step": 24638 }, { "epoch": 61.5975, "grad_norm": 8.5, "learning_rate": 6.787650005205576e-06, "loss": 1.332, "step": 24639 }, { "epoch": 61.6, "grad_norm": 6.46875, "learning_rate": 6.786883240811479e-06, "loss": 1.4498, "step": 24640 }, { "epoch": 61.6025, "grad_norm": 8.875, "learning_rate": 6.786116497482445e-06, "loss": 1.3865, "step": 24641 }, { "epoch": 61.605, "grad_norm": 7.40625, "learning_rate": 6.785349775223502e-06, "loss": 1.2106, "step": 24642 }, { "epoch": 61.6075, "grad_norm": 6.5625, "learning_rate": 6.784583074039679e-06, "loss": 1.2348, "step": 24643 }, { "epoch": 61.61, "grad_norm": 9.1875, "learning_rate": 6.7838163939360026e-06, "loss": 1.3512, "step": 24644 }, { "epoch": 61.6125, "grad_norm": 7.3125, "learning_rate": 6.783049734917494e-06, "loss": 1.3272, "step": 24645 }, { "epoch": 61.615, "grad_norm": 5.8125, "learning_rate": 6.78228309698918e-06, "loss": 1.2621, "step": 24646 }, { "epoch": 61.6175, "grad_norm": 5.46875, "learning_rate": 6.781516480156094e-06, "loss": 1.357, "step": 24647 }, { "epoch": 61.62, "grad_norm": 7.25, "learning_rate": 6.780749884423257e-06, "loss": 1.4505, "step": 24648 }, { "epoch": 61.6225, "grad_norm": 8.75, "learning_rate": 6.779983309795693e-06, "loss": 1.3816, "step": 24649 }, { "epoch": 61.625, "grad_norm": 8.4375, "learning_rate": 6.77921675627843e-06, "loss": 1.5155, "step": 24650 }, { "epoch": 61.6275, "grad_norm": 8.4375, "learning_rate": 6.778450223876492e-06, "loss": 1.2485, "step": 24651 }, { "epoch": 61.63, "grad_norm": 5.90625, "learning_rate": 6.777683712594909e-06, "loss": 1.3295, "step": 24652 }, { "epoch": 61.6325, "grad_norm": 12.4375, "learning_rate": 6.776917222438699e-06, "loss": 1.2863, "step": 24653 }, { "epoch": 61.635, "grad_norm": 8.875, "learning_rate": 6.776150753412894e-06, "loss": 1.3237, "step": 24654 }, { "epoch": 61.6375, "grad_norm": 5.6875, "learning_rate": 6.775384305522514e-06, "loss": 1.3179, "step": 24655 }, { "epoch": 61.64, "grad_norm": 8.125, "learning_rate": 6.774617878772582e-06, "loss": 1.3359, "step": 24656 }, { "epoch": 61.6425, "grad_norm": 8.4375, "learning_rate": 6.77385147316813e-06, "loss": 1.3224, "step": 24657 }, { "epoch": 61.645, "grad_norm": 5.0625, "learning_rate": 6.773085088714177e-06, "loss": 1.255, "step": 24658 }, { "epoch": 61.6475, "grad_norm": 8.125, "learning_rate": 6.772318725415753e-06, "loss": 1.2448, "step": 24659 }, { "epoch": 61.65, "grad_norm": 6.1875, "learning_rate": 6.771552383277875e-06, "loss": 1.3212, "step": 24660 }, { "epoch": 61.6525, "grad_norm": 8.75, "learning_rate": 6.770786062305569e-06, "loss": 1.4559, "step": 24661 }, { "epoch": 61.655, "grad_norm": 7.78125, "learning_rate": 6.770019762503861e-06, "loss": 1.3587, "step": 24662 }, { "epoch": 61.6575, "grad_norm": 8.5625, "learning_rate": 6.769253483877778e-06, "loss": 1.42, "step": 24663 }, { "epoch": 61.66, "grad_norm": 8.875, "learning_rate": 6.768487226432336e-06, "loss": 1.3412, "step": 24664 }, { "epoch": 61.6625, "grad_norm": 8.375, "learning_rate": 6.767720990172566e-06, "loss": 1.3954, "step": 24665 }, { "epoch": 61.665, "grad_norm": 9.0, "learning_rate": 6.76695477510348e-06, "loss": 1.4174, "step": 24666 }, { "epoch": 61.6675, "grad_norm": 11.25, "learning_rate": 6.766188581230117e-06, "loss": 1.5185, "step": 24667 }, { "epoch": 61.67, "grad_norm": 10.9375, "learning_rate": 6.76542240855749e-06, "loss": 1.2757, "step": 24668 }, { "epoch": 61.6725, "grad_norm": 7.0, "learning_rate": 6.764656257090623e-06, "loss": 1.2478, "step": 24669 }, { "epoch": 61.675, "grad_norm": 6.53125, "learning_rate": 6.763890126834542e-06, "loss": 1.2854, "step": 24670 }, { "epoch": 61.6775, "grad_norm": 9.4375, "learning_rate": 6.763124017794266e-06, "loss": 1.2377, "step": 24671 }, { "epoch": 61.68, "grad_norm": 5.15625, "learning_rate": 6.76235792997482e-06, "loss": 1.2934, "step": 24672 }, { "epoch": 61.6825, "grad_norm": 7.0, "learning_rate": 6.761591863381225e-06, "loss": 1.407, "step": 24673 }, { "epoch": 61.685, "grad_norm": 7.40625, "learning_rate": 6.7608258180185085e-06, "loss": 1.3469, "step": 24674 }, { "epoch": 61.6875, "grad_norm": 9.5, "learning_rate": 6.760059793891684e-06, "loss": 1.3236, "step": 24675 }, { "epoch": 61.69, "grad_norm": 5.78125, "learning_rate": 6.759293791005777e-06, "loss": 1.3838, "step": 24676 }, { "epoch": 61.6925, "grad_norm": 6.5625, "learning_rate": 6.758527809365812e-06, "loss": 1.3779, "step": 24677 }, { "epoch": 61.695, "grad_norm": 6.3125, "learning_rate": 6.7577618489768115e-06, "loss": 1.2518, "step": 24678 }, { "epoch": 61.6975, "grad_norm": 6.90625, "learning_rate": 6.756995909843791e-06, "loss": 1.3535, "step": 24679 }, { "epoch": 61.7, "grad_norm": 6.0625, "learning_rate": 6.756229991971779e-06, "loss": 1.3533, "step": 24680 }, { "epoch": 61.7025, "grad_norm": 6.40625, "learning_rate": 6.755464095365787e-06, "loss": 1.3287, "step": 24681 }, { "epoch": 61.705, "grad_norm": 6.4375, "learning_rate": 6.754698220030849e-06, "loss": 1.4344, "step": 24682 }, { "epoch": 61.7075, "grad_norm": 8.25, "learning_rate": 6.753932365971976e-06, "loss": 1.4608, "step": 24683 }, { "epoch": 61.71, "grad_norm": 5.09375, "learning_rate": 6.7531665331941955e-06, "loss": 1.2066, "step": 24684 }, { "epoch": 61.7125, "grad_norm": 11.375, "learning_rate": 6.7524007217025235e-06, "loss": 1.4638, "step": 24685 }, { "epoch": 61.715, "grad_norm": 8.3125, "learning_rate": 6.75163493150198e-06, "loss": 1.2649, "step": 24686 }, { "epoch": 61.7175, "grad_norm": 5.09375, "learning_rate": 6.750869162597591e-06, "loss": 1.2245, "step": 24687 }, { "epoch": 61.72, "grad_norm": 5.5, "learning_rate": 6.750103414994374e-06, "loss": 1.559, "step": 24688 }, { "epoch": 61.7225, "grad_norm": 7.09375, "learning_rate": 6.749337688697348e-06, "loss": 1.2479, "step": 24689 }, { "epoch": 61.725, "grad_norm": 5.9375, "learning_rate": 6.748571983711534e-06, "loss": 1.2746, "step": 24690 }, { "epoch": 61.7275, "grad_norm": 7.0, "learning_rate": 6.7478063000419506e-06, "loss": 1.3971, "step": 24691 }, { "epoch": 61.73, "grad_norm": 7.96875, "learning_rate": 6.747040637693622e-06, "loss": 1.2409, "step": 24692 }, { "epoch": 61.7325, "grad_norm": 7.4375, "learning_rate": 6.746274996671562e-06, "loss": 1.4425, "step": 24693 }, { "epoch": 61.735, "grad_norm": 13.25, "learning_rate": 6.745509376980794e-06, "loss": 1.3658, "step": 24694 }, { "epoch": 61.7375, "grad_norm": 6.03125, "learning_rate": 6.7447437786263395e-06, "loss": 1.3402, "step": 24695 }, { "epoch": 61.74, "grad_norm": 8.5, "learning_rate": 6.743978201613206e-06, "loss": 1.4592, "step": 24696 }, { "epoch": 61.7425, "grad_norm": 8.8125, "learning_rate": 6.743212645946429e-06, "loss": 1.3392, "step": 24697 }, { "epoch": 61.745, "grad_norm": 8.4375, "learning_rate": 6.742447111631016e-06, "loss": 1.3008, "step": 24698 }, { "epoch": 61.7475, "grad_norm": 8.0625, "learning_rate": 6.741681598671993e-06, "loss": 1.4012, "step": 24699 }, { "epoch": 61.75, "grad_norm": 7.21875, "learning_rate": 6.740916107074372e-06, "loss": 1.2924, "step": 24700 }, { "epoch": 61.7525, "grad_norm": 6.375, "learning_rate": 6.7401506368431725e-06, "loss": 1.348, "step": 24701 }, { "epoch": 61.755, "grad_norm": 11.1875, "learning_rate": 6.739385187983418e-06, "loss": 1.4254, "step": 24702 }, { "epoch": 61.7575, "grad_norm": 6.5625, "learning_rate": 6.7386197605001245e-06, "loss": 1.4243, "step": 24703 }, { "epoch": 61.76, "grad_norm": 6.90625, "learning_rate": 6.737854354398308e-06, "loss": 1.2732, "step": 24704 }, { "epoch": 61.7625, "grad_norm": 5.96875, "learning_rate": 6.737088969682988e-06, "loss": 1.3264, "step": 24705 }, { "epoch": 61.765, "grad_norm": 8.125, "learning_rate": 6.736323606359181e-06, "loss": 1.2385, "step": 24706 }, { "epoch": 61.7675, "grad_norm": 6.96875, "learning_rate": 6.735558264431907e-06, "loss": 1.3337, "step": 24707 }, { "epoch": 61.77, "grad_norm": 6.25, "learning_rate": 6.734792943906184e-06, "loss": 1.3278, "step": 24708 }, { "epoch": 61.7725, "grad_norm": 7.25, "learning_rate": 6.734027644787024e-06, "loss": 1.3724, "step": 24709 }, { "epoch": 61.775, "grad_norm": 6.53125, "learning_rate": 6.7332623670794525e-06, "loss": 1.3503, "step": 24710 }, { "epoch": 61.7775, "grad_norm": 6.59375, "learning_rate": 6.732497110788479e-06, "loss": 1.2914, "step": 24711 }, { "epoch": 61.78, "grad_norm": 5.90625, "learning_rate": 6.731731875919123e-06, "loss": 1.2104, "step": 24712 }, { "epoch": 61.7825, "grad_norm": 7.5625, "learning_rate": 6.7309666624764035e-06, "loss": 1.1394, "step": 24713 }, { "epoch": 61.785, "grad_norm": 7.75, "learning_rate": 6.730201470465339e-06, "loss": 1.3062, "step": 24714 }, { "epoch": 61.7875, "grad_norm": 7.53125, "learning_rate": 6.729436299890938e-06, "loss": 1.3791, "step": 24715 }, { "epoch": 61.79, "grad_norm": 5.75, "learning_rate": 6.728671150758222e-06, "loss": 1.3236, "step": 24716 }, { "epoch": 61.7925, "grad_norm": 9.5625, "learning_rate": 6.727906023072207e-06, "loss": 1.4915, "step": 24717 }, { "epoch": 61.795, "grad_norm": 10.1875, "learning_rate": 6.727140916837913e-06, "loss": 1.5449, "step": 24718 }, { "epoch": 61.7975, "grad_norm": 7.21875, "learning_rate": 6.72637583206035e-06, "loss": 1.4177, "step": 24719 }, { "epoch": 61.8, "grad_norm": 8.875, "learning_rate": 6.725610768744535e-06, "loss": 1.3324, "step": 24720 }, { "epoch": 61.8025, "grad_norm": 5.71875, "learning_rate": 6.724845726895484e-06, "loss": 1.3818, "step": 24721 }, { "epoch": 61.805, "grad_norm": 7.53125, "learning_rate": 6.724080706518215e-06, "loss": 1.4065, "step": 24722 }, { "epoch": 61.8075, "grad_norm": 6.84375, "learning_rate": 6.723315707617743e-06, "loss": 1.2807, "step": 24723 }, { "epoch": 61.81, "grad_norm": 7.59375, "learning_rate": 6.72255073019908e-06, "loss": 1.1122, "step": 24724 }, { "epoch": 61.8125, "grad_norm": 8.4375, "learning_rate": 6.7217857742672465e-06, "loss": 1.3105, "step": 24725 }, { "epoch": 61.815, "grad_norm": 8.0, "learning_rate": 6.721020839827248e-06, "loss": 1.4199, "step": 24726 }, { "epoch": 61.8175, "grad_norm": 6.71875, "learning_rate": 6.7202559268841095e-06, "loss": 1.2779, "step": 24727 }, { "epoch": 61.82, "grad_norm": 10.3125, "learning_rate": 6.719491035442842e-06, "loss": 1.2095, "step": 24728 }, { "epoch": 61.8225, "grad_norm": 6.0625, "learning_rate": 6.7187261655084625e-06, "loss": 1.2847, "step": 24729 }, { "epoch": 61.825, "grad_norm": 7.8125, "learning_rate": 6.717961317085979e-06, "loss": 1.3385, "step": 24730 }, { "epoch": 61.8275, "grad_norm": 9.1875, "learning_rate": 6.717196490180409e-06, "loss": 1.273, "step": 24731 }, { "epoch": 61.83, "grad_norm": 6.125, "learning_rate": 6.716431684796769e-06, "loss": 1.1836, "step": 24732 }, { "epoch": 61.8325, "grad_norm": 8.5625, "learning_rate": 6.715666900940074e-06, "loss": 1.3717, "step": 24733 }, { "epoch": 61.835, "grad_norm": 5.90625, "learning_rate": 6.714902138615334e-06, "loss": 1.3509, "step": 24734 }, { "epoch": 61.8375, "grad_norm": 6.125, "learning_rate": 6.7141373978275646e-06, "loss": 1.4095, "step": 24735 }, { "epoch": 61.84, "grad_norm": 7.625, "learning_rate": 6.713372678581773e-06, "loss": 1.4413, "step": 24736 }, { "epoch": 61.8425, "grad_norm": 7.0625, "learning_rate": 6.712607980882986e-06, "loss": 1.3326, "step": 24737 }, { "epoch": 61.845, "grad_norm": 6.75, "learning_rate": 6.7118433047362065e-06, "loss": 1.3597, "step": 24738 }, { "epoch": 61.8475, "grad_norm": 8.75, "learning_rate": 6.7110786501464506e-06, "loss": 1.2171, "step": 24739 }, { "epoch": 61.85, "grad_norm": 8.1875, "learning_rate": 6.710314017118734e-06, "loss": 1.3919, "step": 24740 }, { "epoch": 61.8525, "grad_norm": 8.5625, "learning_rate": 6.709549405658063e-06, "loss": 1.1613, "step": 24741 }, { "epoch": 61.855, "grad_norm": 7.375, "learning_rate": 6.7087848157694554e-06, "loss": 1.2455, "step": 24742 }, { "epoch": 61.8575, "grad_norm": 5.25, "learning_rate": 6.708020247457925e-06, "loss": 1.3186, "step": 24743 }, { "epoch": 61.86, "grad_norm": 5.78125, "learning_rate": 6.707255700728482e-06, "loss": 1.4055, "step": 24744 }, { "epoch": 61.8625, "grad_norm": 6.6875, "learning_rate": 6.706491175586138e-06, "loss": 1.4596, "step": 24745 }, { "epoch": 61.865, "grad_norm": 7.03125, "learning_rate": 6.705726672035903e-06, "loss": 1.352, "step": 24746 }, { "epoch": 61.8675, "grad_norm": 7.53125, "learning_rate": 6.704962190082795e-06, "loss": 1.3028, "step": 24747 }, { "epoch": 61.87, "grad_norm": 9.0, "learning_rate": 6.704197729731825e-06, "loss": 1.2273, "step": 24748 }, { "epoch": 61.8725, "grad_norm": 10.8125, "learning_rate": 6.703433290988002e-06, "loss": 1.3041, "step": 24749 }, { "epoch": 61.875, "grad_norm": 10.125, "learning_rate": 6.702668873856339e-06, "loss": 1.2419, "step": 24750 }, { "epoch": 61.8775, "grad_norm": 10.1875, "learning_rate": 6.701904478341841e-06, "loss": 1.466, "step": 24751 }, { "epoch": 61.88, "grad_norm": 7.03125, "learning_rate": 6.7011401044495304e-06, "loss": 1.2631, "step": 24752 }, { "epoch": 61.8825, "grad_norm": 8.3125, "learning_rate": 6.700375752184411e-06, "loss": 1.4005, "step": 24753 }, { "epoch": 61.885, "grad_norm": 8.0625, "learning_rate": 6.6996114215515e-06, "loss": 1.2873, "step": 24754 }, { "epoch": 61.8875, "grad_norm": 6.21875, "learning_rate": 6.698847112555802e-06, "loss": 1.3347, "step": 24755 }, { "epoch": 61.89, "grad_norm": 9.4375, "learning_rate": 6.6980828252023275e-06, "loss": 1.2413, "step": 24756 }, { "epoch": 61.8925, "grad_norm": 8.0625, "learning_rate": 6.697318559496093e-06, "loss": 1.3405, "step": 24757 }, { "epoch": 61.895, "grad_norm": 7.46875, "learning_rate": 6.696554315442106e-06, "loss": 1.3198, "step": 24758 }, { "epoch": 61.8975, "grad_norm": 8.0625, "learning_rate": 6.695790093045377e-06, "loss": 1.4201, "step": 24759 }, { "epoch": 61.9, "grad_norm": 7.125, "learning_rate": 6.695025892310913e-06, "loss": 1.2962, "step": 24760 }, { "epoch": 61.9025, "grad_norm": 7.53125, "learning_rate": 6.694261713243729e-06, "loss": 1.3077, "step": 24761 }, { "epoch": 61.905, "grad_norm": 9.375, "learning_rate": 6.693497555848834e-06, "loss": 1.4028, "step": 24762 }, { "epoch": 61.9075, "grad_norm": 7.28125, "learning_rate": 6.692733420131234e-06, "loss": 1.2489, "step": 24763 }, { "epoch": 61.91, "grad_norm": 7.4375, "learning_rate": 6.691969306095943e-06, "loss": 1.5085, "step": 24764 }, { "epoch": 61.9125, "grad_norm": 6.34375, "learning_rate": 6.691205213747971e-06, "loss": 1.392, "step": 24765 }, { "epoch": 61.915, "grad_norm": 12.5625, "learning_rate": 6.690441143092319e-06, "loss": 1.4378, "step": 24766 }, { "epoch": 61.9175, "grad_norm": 7.65625, "learning_rate": 6.689677094134008e-06, "loss": 1.1912, "step": 24767 }, { "epoch": 61.92, "grad_norm": 7.96875, "learning_rate": 6.68891306687804e-06, "loss": 1.2794, "step": 24768 }, { "epoch": 61.9225, "grad_norm": 6.9375, "learning_rate": 6.688149061329428e-06, "loss": 1.3949, "step": 24769 }, { "epoch": 61.925, "grad_norm": 9.3125, "learning_rate": 6.687385077493174e-06, "loss": 1.4231, "step": 24770 }, { "epoch": 61.9275, "grad_norm": 7.34375, "learning_rate": 6.6866211153742896e-06, "loss": 1.3031, "step": 24771 }, { "epoch": 61.93, "grad_norm": 8.625, "learning_rate": 6.685857174977788e-06, "loss": 1.3751, "step": 24772 }, { "epoch": 61.9325, "grad_norm": 10.875, "learning_rate": 6.685093256308676e-06, "loss": 1.2801, "step": 24773 }, { "epoch": 61.935, "grad_norm": 24.875, "learning_rate": 6.6843293593719576e-06, "loss": 1.2233, "step": 24774 }, { "epoch": 61.9375, "grad_norm": 7.5, "learning_rate": 6.683565484172643e-06, "loss": 1.3917, "step": 24775 }, { "epoch": 61.94, "grad_norm": 6.78125, "learning_rate": 6.682801630715737e-06, "loss": 1.4176, "step": 24776 }, { "epoch": 61.9425, "grad_norm": 10.0625, "learning_rate": 6.6820377990062555e-06, "loss": 1.1835, "step": 24777 }, { "epoch": 61.945, "grad_norm": 8.125, "learning_rate": 6.681273989049201e-06, "loss": 1.4455, "step": 24778 }, { "epoch": 61.9475, "grad_norm": 9.5625, "learning_rate": 6.68051020084958e-06, "loss": 1.2983, "step": 24779 }, { "epoch": 61.95, "grad_norm": 6.40625, "learning_rate": 6.6797464344124045e-06, "loss": 1.2702, "step": 24780 }, { "epoch": 61.9525, "grad_norm": 5.71875, "learning_rate": 6.678982689742673e-06, "loss": 1.2331, "step": 24781 }, { "epoch": 61.955, "grad_norm": 10.3125, "learning_rate": 6.6782189668454e-06, "loss": 1.303, "step": 24782 }, { "epoch": 61.9575, "grad_norm": 7.1875, "learning_rate": 6.677455265725592e-06, "loss": 1.2873, "step": 24783 }, { "epoch": 61.96, "grad_norm": 7.375, "learning_rate": 6.676691586388255e-06, "loss": 1.2508, "step": 24784 }, { "epoch": 61.9625, "grad_norm": 7.15625, "learning_rate": 6.675927928838395e-06, "loss": 1.3711, "step": 24785 }, { "epoch": 61.965, "grad_norm": 7.75, "learning_rate": 6.6751642930810145e-06, "loss": 1.334, "step": 24786 }, { "epoch": 61.9675, "grad_norm": 8.25, "learning_rate": 6.674400679121127e-06, "loss": 1.4116, "step": 24787 }, { "epoch": 61.97, "grad_norm": 6.0625, "learning_rate": 6.6736370869637355e-06, "loss": 1.2065, "step": 24788 }, { "epoch": 61.9725, "grad_norm": 8.8125, "learning_rate": 6.672873516613847e-06, "loss": 1.3071, "step": 24789 }, { "epoch": 61.975, "grad_norm": 7.875, "learning_rate": 6.672109968076465e-06, "loss": 1.3268, "step": 24790 }, { "epoch": 61.9775, "grad_norm": 6.625, "learning_rate": 6.671346441356594e-06, "loss": 1.2608, "step": 24791 }, { "epoch": 61.98, "grad_norm": 8.375, "learning_rate": 6.670582936459249e-06, "loss": 1.141, "step": 24792 }, { "epoch": 61.9825, "grad_norm": 18.5, "learning_rate": 6.6698194533894245e-06, "loss": 1.4041, "step": 24793 }, { "epoch": 61.985, "grad_norm": 6.96875, "learning_rate": 6.669055992152131e-06, "loss": 1.296, "step": 24794 }, { "epoch": 61.9875, "grad_norm": 5.53125, "learning_rate": 6.668292552752376e-06, "loss": 1.2947, "step": 24795 }, { "epoch": 61.99, "grad_norm": 10.8125, "learning_rate": 6.667529135195158e-06, "loss": 1.532, "step": 24796 }, { "epoch": 61.9925, "grad_norm": 6.75, "learning_rate": 6.666765739485487e-06, "loss": 1.2174, "step": 24797 }, { "epoch": 61.995, "grad_norm": 7.28125, "learning_rate": 6.666002365628366e-06, "loss": 1.5566, "step": 24798 }, { "epoch": 61.9975, "grad_norm": 6.40625, "learning_rate": 6.665239013628803e-06, "loss": 1.261, "step": 24799 }, { "epoch": 62.0, "grad_norm": 7.65625, "learning_rate": 6.664475683491797e-06, "loss": 1.3776, "step": 24800 }, { "epoch": 62.0025, "grad_norm": 6.71875, "learning_rate": 6.663712375222353e-06, "loss": 1.3222, "step": 24801 }, { "epoch": 62.005, "grad_norm": 6.28125, "learning_rate": 6.662949088825479e-06, "loss": 1.3691, "step": 24802 }, { "epoch": 62.0075, "grad_norm": 7.84375, "learning_rate": 6.6621858243061784e-06, "loss": 1.3441, "step": 24803 }, { "epoch": 62.01, "grad_norm": 6.3125, "learning_rate": 6.661422581669453e-06, "loss": 1.3429, "step": 24804 }, { "epoch": 62.0125, "grad_norm": 6.53125, "learning_rate": 6.660659360920309e-06, "loss": 1.5881, "step": 24805 }, { "epoch": 62.015, "grad_norm": 9.625, "learning_rate": 6.659896162063743e-06, "loss": 1.3675, "step": 24806 }, { "epoch": 62.0175, "grad_norm": 8.0, "learning_rate": 6.65913298510477e-06, "loss": 1.2684, "step": 24807 }, { "epoch": 62.02, "grad_norm": 7.96875, "learning_rate": 6.6583698300483864e-06, "loss": 1.1816, "step": 24808 }, { "epoch": 62.0225, "grad_norm": 9.625, "learning_rate": 6.657606696899597e-06, "loss": 1.2811, "step": 24809 }, { "epoch": 62.025, "grad_norm": 7.8125, "learning_rate": 6.656843585663404e-06, "loss": 1.4372, "step": 24810 }, { "epoch": 62.0275, "grad_norm": 6.6875, "learning_rate": 6.656080496344808e-06, "loss": 1.2587, "step": 24811 }, { "epoch": 62.03, "grad_norm": 9.75, "learning_rate": 6.6553174289488156e-06, "loss": 1.5201, "step": 24812 }, { "epoch": 62.0325, "grad_norm": 4.53125, "learning_rate": 6.6545543834804295e-06, "loss": 1.3013, "step": 24813 }, { "epoch": 62.035, "grad_norm": 8.5625, "learning_rate": 6.653791359944652e-06, "loss": 1.3069, "step": 24814 }, { "epoch": 62.0375, "grad_norm": 9.0, "learning_rate": 6.6530283583464825e-06, "loss": 1.41, "step": 24815 }, { "epoch": 62.04, "grad_norm": 6.4375, "learning_rate": 6.652265378690923e-06, "loss": 1.3817, "step": 24816 }, { "epoch": 62.0425, "grad_norm": 7.71875, "learning_rate": 6.6515024209829805e-06, "loss": 1.3765, "step": 24817 }, { "epoch": 62.045, "grad_norm": 7.28125, "learning_rate": 6.650739485227655e-06, "loss": 1.325, "step": 24818 }, { "epoch": 62.0475, "grad_norm": 7.75, "learning_rate": 6.649976571429947e-06, "loss": 1.4142, "step": 24819 }, { "epoch": 62.05, "grad_norm": 6.3125, "learning_rate": 6.649213679594859e-06, "loss": 1.2283, "step": 24820 }, { "epoch": 62.0525, "grad_norm": 9.875, "learning_rate": 6.648450809727387e-06, "loss": 1.3073, "step": 24821 }, { "epoch": 62.055, "grad_norm": 6.46875, "learning_rate": 6.647687961832543e-06, "loss": 1.3712, "step": 24822 }, { "epoch": 62.0575, "grad_norm": 7.78125, "learning_rate": 6.646925135915321e-06, "loss": 1.3307, "step": 24823 }, { "epoch": 62.06, "grad_norm": 7.75, "learning_rate": 6.646162331980726e-06, "loss": 1.3002, "step": 24824 }, { "epoch": 62.0625, "grad_norm": 10.25, "learning_rate": 6.645399550033753e-06, "loss": 1.3297, "step": 24825 }, { "epoch": 62.065, "grad_norm": 7.75, "learning_rate": 6.6446367900794065e-06, "loss": 1.3803, "step": 24826 }, { "epoch": 62.0675, "grad_norm": 9.875, "learning_rate": 6.643874052122686e-06, "loss": 1.3077, "step": 24827 }, { "epoch": 62.07, "grad_norm": 7.625, "learning_rate": 6.643111336168597e-06, "loss": 1.4029, "step": 24828 }, { "epoch": 62.0725, "grad_norm": 7.375, "learning_rate": 6.642348642222134e-06, "loss": 1.3153, "step": 24829 }, { "epoch": 62.075, "grad_norm": 9.3125, "learning_rate": 6.641585970288298e-06, "loss": 1.2934, "step": 24830 }, { "epoch": 62.0775, "grad_norm": 5.625, "learning_rate": 6.640823320372087e-06, "loss": 1.1712, "step": 24831 }, { "epoch": 62.08, "grad_norm": 6.90625, "learning_rate": 6.6400606924785095e-06, "loss": 1.352, "step": 24832 }, { "epoch": 62.0825, "grad_norm": 7.375, "learning_rate": 6.6392980866125574e-06, "loss": 1.3236, "step": 24833 }, { "epoch": 62.085, "grad_norm": 6.84375, "learning_rate": 6.638535502779232e-06, "loss": 1.2258, "step": 24834 }, { "epoch": 62.0875, "grad_norm": 7.03125, "learning_rate": 6.637772940983536e-06, "loss": 1.231, "step": 24835 }, { "epoch": 62.09, "grad_norm": 8.6875, "learning_rate": 6.637010401230459e-06, "loss": 1.3074, "step": 24836 }, { "epoch": 62.0925, "grad_norm": 8.4375, "learning_rate": 6.636247883525014e-06, "loss": 1.1098, "step": 24837 }, { "epoch": 62.095, "grad_norm": 13.8125, "learning_rate": 6.635485387872191e-06, "loss": 1.3245, "step": 24838 }, { "epoch": 62.0975, "grad_norm": 5.8125, "learning_rate": 6.634722914276994e-06, "loss": 1.3454, "step": 24839 }, { "epoch": 62.1, "grad_norm": 6.09375, "learning_rate": 6.633960462744415e-06, "loss": 1.2135, "step": 24840 }, { "epoch": 62.1025, "grad_norm": 13.8125, "learning_rate": 6.633198033279457e-06, "loss": 1.291, "step": 24841 }, { "epoch": 62.105, "grad_norm": 7.625, "learning_rate": 6.632435625887119e-06, "loss": 1.1573, "step": 24842 }, { "epoch": 62.1075, "grad_norm": 9.75, "learning_rate": 6.631673240572399e-06, "loss": 1.35, "step": 24843 }, { "epoch": 62.11, "grad_norm": 8.0, "learning_rate": 6.6309108773402945e-06, "loss": 1.2994, "step": 24844 }, { "epoch": 62.1125, "grad_norm": 8.5, "learning_rate": 6.6301485361957994e-06, "loss": 1.3835, "step": 24845 }, { "epoch": 62.115, "grad_norm": 7.53125, "learning_rate": 6.629386217143918e-06, "loss": 1.3814, "step": 24846 }, { "epoch": 62.1175, "grad_norm": 12.0, "learning_rate": 6.62862392018965e-06, "loss": 1.3201, "step": 24847 }, { "epoch": 62.12, "grad_norm": 7.21875, "learning_rate": 6.627861645337984e-06, "loss": 1.3186, "step": 24848 }, { "epoch": 62.1225, "grad_norm": 7.78125, "learning_rate": 6.627099392593923e-06, "loss": 1.1863, "step": 24849 }, { "epoch": 62.125, "grad_norm": 8.8125, "learning_rate": 6.6263371619624615e-06, "loss": 1.2984, "step": 24850 }, { "epoch": 62.1275, "grad_norm": 6.90625, "learning_rate": 6.625574953448604e-06, "loss": 1.2676, "step": 24851 }, { "epoch": 62.13, "grad_norm": 6.1875, "learning_rate": 6.624812767057338e-06, "loss": 1.3473, "step": 24852 }, { "epoch": 62.1325, "grad_norm": 8.375, "learning_rate": 6.624050602793665e-06, "loss": 1.2005, "step": 24853 }, { "epoch": 62.135, "grad_norm": 5.21875, "learning_rate": 6.623288460662583e-06, "loss": 1.1522, "step": 24854 }, { "epoch": 62.1375, "grad_norm": 8.6875, "learning_rate": 6.622526340669082e-06, "loss": 1.4451, "step": 24855 }, { "epoch": 62.14, "grad_norm": 8.0625, "learning_rate": 6.621764242818167e-06, "loss": 1.2769, "step": 24856 }, { "epoch": 62.1425, "grad_norm": 8.75, "learning_rate": 6.6210021671148294e-06, "loss": 1.41, "step": 24857 }, { "epoch": 62.145, "grad_norm": 7.15625, "learning_rate": 6.620240113564068e-06, "loss": 1.4698, "step": 24858 }, { "epoch": 62.1475, "grad_norm": 7.96875, "learning_rate": 6.619478082170876e-06, "loss": 1.2561, "step": 24859 }, { "epoch": 62.15, "grad_norm": 8.8125, "learning_rate": 6.618716072940248e-06, "loss": 1.4214, "step": 24860 }, { "epoch": 62.1525, "grad_norm": 6.46875, "learning_rate": 6.617954085877184e-06, "loss": 1.46, "step": 24861 }, { "epoch": 62.155, "grad_norm": 10.125, "learning_rate": 6.61719212098668e-06, "loss": 1.2573, "step": 24862 }, { "epoch": 62.1575, "grad_norm": 4.9375, "learning_rate": 6.616430178273727e-06, "loss": 1.228, "step": 24863 }, { "epoch": 62.16, "grad_norm": 6.65625, "learning_rate": 6.615668257743322e-06, "loss": 1.3544, "step": 24864 }, { "epoch": 62.1625, "grad_norm": 8.625, "learning_rate": 6.614906359400459e-06, "loss": 1.2007, "step": 24865 }, { "epoch": 62.165, "grad_norm": 10.0, "learning_rate": 6.614144483250138e-06, "loss": 1.3067, "step": 24866 }, { "epoch": 62.1675, "grad_norm": 14.875, "learning_rate": 6.613382629297349e-06, "loss": 1.4971, "step": 24867 }, { "epoch": 62.17, "grad_norm": 8.8125, "learning_rate": 6.612620797547087e-06, "loss": 1.4865, "step": 24868 }, { "epoch": 62.1725, "grad_norm": 7.34375, "learning_rate": 6.6118589880043496e-06, "loss": 1.2798, "step": 24869 }, { "epoch": 62.175, "grad_norm": 6.71875, "learning_rate": 6.611097200674126e-06, "loss": 1.2348, "step": 24870 }, { "epoch": 62.1775, "grad_norm": 6.78125, "learning_rate": 6.610335435561415e-06, "loss": 1.2157, "step": 24871 }, { "epoch": 62.18, "grad_norm": 9.6875, "learning_rate": 6.609573692671209e-06, "loss": 1.3611, "step": 24872 }, { "epoch": 62.1825, "grad_norm": 6.03125, "learning_rate": 6.608811972008504e-06, "loss": 1.3644, "step": 24873 }, { "epoch": 62.185, "grad_norm": 8.3125, "learning_rate": 6.60805027357829e-06, "loss": 1.3202, "step": 24874 }, { "epoch": 62.1875, "grad_norm": 6.875, "learning_rate": 6.607288597385561e-06, "loss": 1.261, "step": 24875 }, { "epoch": 62.19, "grad_norm": 7.8125, "learning_rate": 6.606526943435314e-06, "loss": 1.3644, "step": 24876 }, { "epoch": 62.1925, "grad_norm": 16.25, "learning_rate": 6.605765311732544e-06, "loss": 1.4305, "step": 24877 }, { "epoch": 62.195, "grad_norm": 6.375, "learning_rate": 6.605003702282238e-06, "loss": 1.1743, "step": 24878 }, { "epoch": 62.1975, "grad_norm": 8.4375, "learning_rate": 6.6042421150893906e-06, "loss": 1.2987, "step": 24879 }, { "epoch": 62.2, "grad_norm": 6.8125, "learning_rate": 6.603480550158995e-06, "loss": 1.301, "step": 24880 }, { "epoch": 62.2025, "grad_norm": 7.53125, "learning_rate": 6.602719007496049e-06, "loss": 1.3913, "step": 24881 }, { "epoch": 62.205, "grad_norm": 4.65625, "learning_rate": 6.6019574871055396e-06, "loss": 1.2709, "step": 24882 }, { "epoch": 62.2075, "grad_norm": 7.1875, "learning_rate": 6.601195988992459e-06, "loss": 1.5818, "step": 24883 }, { "epoch": 62.21, "grad_norm": 9.6875, "learning_rate": 6.600434513161807e-06, "loss": 1.3444, "step": 24884 }, { "epoch": 62.2125, "grad_norm": 5.625, "learning_rate": 6.599673059618563e-06, "loss": 1.1214, "step": 24885 }, { "epoch": 62.215, "grad_norm": 7.21875, "learning_rate": 6.598911628367729e-06, "loss": 1.3254, "step": 24886 }, { "epoch": 62.2175, "grad_norm": 5.9375, "learning_rate": 6.5981502194142945e-06, "loss": 1.3503, "step": 24887 }, { "epoch": 62.22, "grad_norm": 7.375, "learning_rate": 6.597388832763254e-06, "loss": 1.2303, "step": 24888 }, { "epoch": 62.2225, "grad_norm": 6.125, "learning_rate": 6.5966274684195945e-06, "loss": 1.3703, "step": 24889 }, { "epoch": 62.225, "grad_norm": 6.75, "learning_rate": 6.595866126388305e-06, "loss": 1.2795, "step": 24890 }, { "epoch": 62.2275, "grad_norm": 10.625, "learning_rate": 6.5951048066743835e-06, "loss": 1.5218, "step": 24891 }, { "epoch": 62.23, "grad_norm": 9.5625, "learning_rate": 6.594343509282821e-06, "loss": 1.4285, "step": 24892 }, { "epoch": 62.2325, "grad_norm": 8.625, "learning_rate": 6.593582234218605e-06, "loss": 1.3749, "step": 24893 }, { "epoch": 62.235, "grad_norm": 8.125, "learning_rate": 6.592820981486729e-06, "loss": 1.2404, "step": 24894 }, { "epoch": 62.2375, "grad_norm": 9.4375, "learning_rate": 6.592059751092176e-06, "loss": 1.2893, "step": 24895 }, { "epoch": 62.24, "grad_norm": 7.8125, "learning_rate": 6.591298543039949e-06, "loss": 1.4113, "step": 24896 }, { "epoch": 62.2425, "grad_norm": 8.625, "learning_rate": 6.590537357335031e-06, "loss": 1.3124, "step": 24897 }, { "epoch": 62.245, "grad_norm": 6.5625, "learning_rate": 6.5897761939824146e-06, "loss": 1.326, "step": 24898 }, { "epoch": 62.2475, "grad_norm": 10.875, "learning_rate": 6.589015052987086e-06, "loss": 1.3877, "step": 24899 }, { "epoch": 62.25, "grad_norm": 7.78125, "learning_rate": 6.588253934354039e-06, "loss": 1.1678, "step": 24900 }, { "epoch": 62.2525, "grad_norm": 6.4375, "learning_rate": 6.587492838088263e-06, "loss": 1.3463, "step": 24901 }, { "epoch": 62.255, "grad_norm": 8.9375, "learning_rate": 6.58673176419475e-06, "loss": 1.3247, "step": 24902 }, { "epoch": 62.2575, "grad_norm": 6.5625, "learning_rate": 6.585970712678484e-06, "loss": 1.3893, "step": 24903 }, { "epoch": 62.26, "grad_norm": 11.0, "learning_rate": 6.585209683544459e-06, "loss": 1.3407, "step": 24904 }, { "epoch": 62.2625, "grad_norm": 10.1875, "learning_rate": 6.584448676797659e-06, "loss": 1.3735, "step": 24905 }, { "epoch": 62.265, "grad_norm": 5.90625, "learning_rate": 6.58368769244308e-06, "loss": 1.254, "step": 24906 }, { "epoch": 62.2675, "grad_norm": 6.84375, "learning_rate": 6.58292673048571e-06, "loss": 1.4172, "step": 24907 }, { "epoch": 62.27, "grad_norm": 7.5625, "learning_rate": 6.582165790930533e-06, "loss": 1.297, "step": 24908 }, { "epoch": 62.2725, "grad_norm": 9.8125, "learning_rate": 6.581404873782542e-06, "loss": 1.2368, "step": 24909 }, { "epoch": 62.275, "grad_norm": 8.3125, "learning_rate": 6.580643979046718e-06, "loss": 1.2982, "step": 24910 }, { "epoch": 62.2775, "grad_norm": 4.78125, "learning_rate": 6.579883106728061e-06, "loss": 1.3527, "step": 24911 }, { "epoch": 62.28, "grad_norm": 8.875, "learning_rate": 6.579122256831551e-06, "loss": 1.3854, "step": 24912 }, { "epoch": 62.2825, "grad_norm": 7.875, "learning_rate": 6.578361429362182e-06, "loss": 1.4146, "step": 24913 }, { "epoch": 62.285, "grad_norm": 7.96875, "learning_rate": 6.5776006243249355e-06, "loss": 1.3758, "step": 24914 }, { "epoch": 62.2875, "grad_norm": 6.28125, "learning_rate": 6.576839841724801e-06, "loss": 1.319, "step": 24915 }, { "epoch": 62.29, "grad_norm": 7.875, "learning_rate": 6.576079081566771e-06, "loss": 1.4298, "step": 24916 }, { "epoch": 62.2925, "grad_norm": 9.375, "learning_rate": 6.575318343855829e-06, "loss": 1.3198, "step": 24917 }, { "epoch": 62.295, "grad_norm": 7.84375, "learning_rate": 6.5745576285969614e-06, "loss": 1.4427, "step": 24918 }, { "epoch": 62.2975, "grad_norm": 7.9375, "learning_rate": 6.573796935795158e-06, "loss": 1.1078, "step": 24919 }, { "epoch": 62.3, "grad_norm": 6.53125, "learning_rate": 6.5730362654554015e-06, "loss": 1.3194, "step": 24920 }, { "epoch": 62.3025, "grad_norm": 7.15625, "learning_rate": 6.572275617582686e-06, "loss": 1.491, "step": 24921 }, { "epoch": 62.305, "grad_norm": 7.78125, "learning_rate": 6.571514992181993e-06, "loss": 1.4985, "step": 24922 }, { "epoch": 62.3075, "grad_norm": 8.0, "learning_rate": 6.5707543892583115e-06, "loss": 1.3962, "step": 24923 }, { "epoch": 62.31, "grad_norm": 6.28125, "learning_rate": 6.569993808816628e-06, "loss": 1.2631, "step": 24924 }, { "epoch": 62.3125, "grad_norm": 5.46875, "learning_rate": 6.569233250861924e-06, "loss": 1.3559, "step": 24925 }, { "epoch": 62.315, "grad_norm": 7.1875, "learning_rate": 6.56847271539919e-06, "loss": 1.2441, "step": 24926 }, { "epoch": 62.3175, "grad_norm": 9.875, "learning_rate": 6.567712202433414e-06, "loss": 1.2883, "step": 24927 }, { "epoch": 62.32, "grad_norm": 8.8125, "learning_rate": 6.566951711969581e-06, "loss": 1.266, "step": 24928 }, { "epoch": 62.3225, "grad_norm": 5.78125, "learning_rate": 6.5661912440126715e-06, "loss": 1.3604, "step": 24929 }, { "epoch": 62.325, "grad_norm": 7.90625, "learning_rate": 6.565430798567674e-06, "loss": 1.3731, "step": 24930 }, { "epoch": 62.3275, "grad_norm": 5.375, "learning_rate": 6.564670375639577e-06, "loss": 1.2615, "step": 24931 }, { "epoch": 62.33, "grad_norm": 9.1875, "learning_rate": 6.563909975233365e-06, "loss": 1.3935, "step": 24932 }, { "epoch": 62.3325, "grad_norm": 9.5625, "learning_rate": 6.563149597354019e-06, "loss": 1.5722, "step": 24933 }, { "epoch": 62.335, "grad_norm": 6.84375, "learning_rate": 6.562389242006528e-06, "loss": 1.3209, "step": 24934 }, { "epoch": 62.3375, "grad_norm": 9.4375, "learning_rate": 6.561628909195873e-06, "loss": 1.3316, "step": 24935 }, { "epoch": 62.34, "grad_norm": 7.34375, "learning_rate": 6.560868598927046e-06, "loss": 1.2245, "step": 24936 }, { "epoch": 62.3425, "grad_norm": 8.1875, "learning_rate": 6.560108311205024e-06, "loss": 1.3011, "step": 24937 }, { "epoch": 62.345, "grad_norm": 9.375, "learning_rate": 6.559348046034794e-06, "loss": 1.2468, "step": 24938 }, { "epoch": 62.3475, "grad_norm": 6.625, "learning_rate": 6.558587803421343e-06, "loss": 1.3099, "step": 24939 }, { "epoch": 62.35, "grad_norm": 7.375, "learning_rate": 6.5578275833696485e-06, "loss": 1.4509, "step": 24940 }, { "epoch": 62.3525, "grad_norm": 6.96875, "learning_rate": 6.557067385884701e-06, "loss": 1.3163, "step": 24941 }, { "epoch": 62.355, "grad_norm": 6.03125, "learning_rate": 6.556307210971482e-06, "loss": 1.4243, "step": 24942 }, { "epoch": 62.3575, "grad_norm": 7.90625, "learning_rate": 6.555547058634977e-06, "loss": 1.3726, "step": 24943 }, { "epoch": 62.36, "grad_norm": 7.125, "learning_rate": 6.554786928880165e-06, "loss": 1.3619, "step": 24944 }, { "epoch": 62.3625, "grad_norm": 6.40625, "learning_rate": 6.5540268217120295e-06, "loss": 1.3346, "step": 24945 }, { "epoch": 62.365, "grad_norm": 7.28125, "learning_rate": 6.5532667371355595e-06, "loss": 1.4586, "step": 24946 }, { "epoch": 62.3675, "grad_norm": 9.0625, "learning_rate": 6.5525066751557364e-06, "loss": 1.3656, "step": 24947 }, { "epoch": 62.37, "grad_norm": 7.28125, "learning_rate": 6.551746635777541e-06, "loss": 1.1602, "step": 24948 }, { "epoch": 62.3725, "grad_norm": 9.125, "learning_rate": 6.550986619005955e-06, "loss": 1.2397, "step": 24949 }, { "epoch": 62.375, "grad_norm": 7.34375, "learning_rate": 6.550226624845961e-06, "loss": 1.1903, "step": 24950 }, { "epoch": 62.3775, "grad_norm": 6.6875, "learning_rate": 6.549466653302547e-06, "loss": 1.1961, "step": 24951 }, { "epoch": 62.38, "grad_norm": 7.53125, "learning_rate": 6.54870670438069e-06, "loss": 1.3293, "step": 24952 }, { "epoch": 62.3825, "grad_norm": 6.65625, "learning_rate": 6.547946778085373e-06, "loss": 1.2676, "step": 24953 }, { "epoch": 62.385, "grad_norm": 10.0, "learning_rate": 6.547186874421581e-06, "loss": 1.4035, "step": 24954 }, { "epoch": 62.3875, "grad_norm": 8.625, "learning_rate": 6.54642699339429e-06, "loss": 1.3141, "step": 24955 }, { "epoch": 62.39, "grad_norm": 8.875, "learning_rate": 6.545667135008489e-06, "loss": 1.3784, "step": 24956 }, { "epoch": 62.3925, "grad_norm": 6.84375, "learning_rate": 6.544907299269154e-06, "loss": 1.2274, "step": 24957 }, { "epoch": 62.395, "grad_norm": 7.625, "learning_rate": 6.54414748618127e-06, "loss": 1.496, "step": 24958 }, { "epoch": 62.3975, "grad_norm": 5.8125, "learning_rate": 6.543387695749816e-06, "loss": 1.1445, "step": 24959 }, { "epoch": 62.4, "grad_norm": 6.59375, "learning_rate": 6.542627927979772e-06, "loss": 1.3533, "step": 24960 }, { "epoch": 62.4025, "grad_norm": 4.71875, "learning_rate": 6.5418681828761234e-06, "loss": 1.2631, "step": 24961 }, { "epoch": 62.405, "grad_norm": 10.4375, "learning_rate": 6.541108460443849e-06, "loss": 1.4322, "step": 24962 }, { "epoch": 62.4075, "grad_norm": 7.375, "learning_rate": 6.540348760687928e-06, "loss": 1.2665, "step": 24963 }, { "epoch": 62.41, "grad_norm": 11.6875, "learning_rate": 6.5395890836133445e-06, "loss": 1.338, "step": 24964 }, { "epoch": 62.4125, "grad_norm": 7.5625, "learning_rate": 6.538829429225068e-06, "loss": 1.2392, "step": 24965 }, { "epoch": 62.415, "grad_norm": 8.4375, "learning_rate": 6.538069797528097e-06, "loss": 1.2398, "step": 24966 }, { "epoch": 62.4175, "grad_norm": 6.875, "learning_rate": 6.537310188527397e-06, "loss": 1.2536, "step": 24967 }, { "epoch": 62.42, "grad_norm": 6.4375, "learning_rate": 6.5365506022279564e-06, "loss": 1.4467, "step": 24968 }, { "epoch": 62.4225, "grad_norm": 6.1875, "learning_rate": 6.535791038634749e-06, "loss": 1.2658, "step": 24969 }, { "epoch": 62.425, "grad_norm": 10.25, "learning_rate": 6.535031497752755e-06, "loss": 1.384, "step": 24970 }, { "epoch": 62.4275, "grad_norm": 6.625, "learning_rate": 6.534271979586958e-06, "loss": 1.2471, "step": 24971 }, { "epoch": 62.43, "grad_norm": 8.625, "learning_rate": 6.533512484142335e-06, "loss": 1.3227, "step": 24972 }, { "epoch": 62.4325, "grad_norm": 7.40625, "learning_rate": 6.532753011423868e-06, "loss": 1.261, "step": 24973 }, { "epoch": 62.435, "grad_norm": 5.25, "learning_rate": 6.5319935614365304e-06, "loss": 1.3366, "step": 24974 }, { "epoch": 62.4375, "grad_norm": 7.0625, "learning_rate": 6.531234134185303e-06, "loss": 1.3116, "step": 24975 }, { "epoch": 62.44, "grad_norm": 5.21875, "learning_rate": 6.530474729675167e-06, "loss": 1.4044, "step": 24976 }, { "epoch": 62.4425, "grad_norm": 11.0625, "learning_rate": 6.529715347911103e-06, "loss": 1.2498, "step": 24977 }, { "epoch": 62.445, "grad_norm": 5.09375, "learning_rate": 6.528955988898083e-06, "loss": 1.3433, "step": 24978 }, { "epoch": 62.4475, "grad_norm": 7.375, "learning_rate": 6.528196652641092e-06, "loss": 1.3281, "step": 24979 }, { "epoch": 62.45, "grad_norm": 10.0, "learning_rate": 6.527437339145097e-06, "loss": 1.3512, "step": 24980 }, { "epoch": 62.4525, "grad_norm": 6.46875, "learning_rate": 6.5266780484150916e-06, "loss": 1.2839, "step": 24981 }, { "epoch": 62.455, "grad_norm": 5.75, "learning_rate": 6.525918780456044e-06, "loss": 1.4581, "step": 24982 }, { "epoch": 62.4575, "grad_norm": 6.46875, "learning_rate": 6.525159535272936e-06, "loss": 1.2453, "step": 24983 }, { "epoch": 62.46, "grad_norm": 8.3125, "learning_rate": 6.52440031287074e-06, "loss": 1.3714, "step": 24984 }, { "epoch": 62.4625, "grad_norm": 8.5, "learning_rate": 6.523641113254434e-06, "loss": 1.2031, "step": 24985 }, { "epoch": 62.465, "grad_norm": 9.25, "learning_rate": 6.522881936429e-06, "loss": 1.3014, "step": 24986 }, { "epoch": 62.4675, "grad_norm": 9.6875, "learning_rate": 6.522122782399416e-06, "loss": 1.3599, "step": 24987 }, { "epoch": 62.47, "grad_norm": 8.4375, "learning_rate": 6.521363651170654e-06, "loss": 1.2364, "step": 24988 }, { "epoch": 62.4725, "grad_norm": 6.28125, "learning_rate": 6.520604542747691e-06, "loss": 1.2848, "step": 24989 }, { "epoch": 62.475, "grad_norm": 9.8125, "learning_rate": 6.519845457135503e-06, "loss": 1.2433, "step": 24990 }, { "epoch": 62.4775, "grad_norm": 6.03125, "learning_rate": 6.519086394339076e-06, "loss": 1.3143, "step": 24991 }, { "epoch": 62.48, "grad_norm": 5.78125, "learning_rate": 6.518327354363374e-06, "loss": 1.4663, "step": 24992 }, { "epoch": 62.4825, "grad_norm": 7.28125, "learning_rate": 6.51756833721338e-06, "loss": 1.3781, "step": 24993 }, { "epoch": 62.485, "grad_norm": 12.25, "learning_rate": 6.516809342894071e-06, "loss": 1.2743, "step": 24994 }, { "epoch": 62.4875, "grad_norm": 6.9375, "learning_rate": 6.516050371410416e-06, "loss": 1.3807, "step": 24995 }, { "epoch": 62.49, "grad_norm": 6.34375, "learning_rate": 6.515291422767398e-06, "loss": 1.3101, "step": 24996 }, { "epoch": 62.4925, "grad_norm": 5.71875, "learning_rate": 6.514532496969989e-06, "loss": 1.3818, "step": 24997 }, { "epoch": 62.495, "grad_norm": 7.6875, "learning_rate": 6.513773594023168e-06, "loss": 1.2422, "step": 24998 }, { "epoch": 62.4975, "grad_norm": 8.125, "learning_rate": 6.513014713931907e-06, "loss": 1.2505, "step": 24999 }, { "epoch": 62.5, "grad_norm": 5.9375, "learning_rate": 6.5122558567011775e-06, "loss": 1.3463, "step": 25000 }, { "epoch": 62.5025, "grad_norm": 7.8125, "learning_rate": 6.511497022335962e-06, "loss": 1.2618, "step": 25001 }, { "epoch": 62.505, "grad_norm": 10.625, "learning_rate": 6.5107382108412355e-06, "loss": 1.2478, "step": 25002 }, { "epoch": 62.5075, "grad_norm": 6.90625, "learning_rate": 6.509979422221966e-06, "loss": 1.311, "step": 25003 }, { "epoch": 62.51, "grad_norm": 8.0, "learning_rate": 6.509220656483133e-06, "loss": 1.3276, "step": 25004 }, { "epoch": 62.5125, "grad_norm": 6.59375, "learning_rate": 6.508461913629707e-06, "loss": 1.4261, "step": 25005 }, { "epoch": 62.515, "grad_norm": 9.6875, "learning_rate": 6.50770319366667e-06, "loss": 1.3611, "step": 25006 }, { "epoch": 62.5175, "grad_norm": 8.3125, "learning_rate": 6.506944496598988e-06, "loss": 1.2731, "step": 25007 }, { "epoch": 62.52, "grad_norm": 10.5625, "learning_rate": 6.506185822431639e-06, "loss": 1.3764, "step": 25008 }, { "epoch": 62.5225, "grad_norm": 8.125, "learning_rate": 6.505427171169597e-06, "loss": 1.372, "step": 25009 }, { "epoch": 62.525, "grad_norm": 6.03125, "learning_rate": 6.504668542817832e-06, "loss": 1.3334, "step": 25010 }, { "epoch": 62.5275, "grad_norm": 8.125, "learning_rate": 6.503909937381321e-06, "loss": 1.215, "step": 25011 }, { "epoch": 62.53, "grad_norm": 5.4375, "learning_rate": 6.503151354865036e-06, "loss": 1.2938, "step": 25012 }, { "epoch": 62.5325, "grad_norm": 7.0, "learning_rate": 6.502392795273953e-06, "loss": 1.2206, "step": 25013 }, { "epoch": 62.535, "grad_norm": 6.0625, "learning_rate": 6.501634258613041e-06, "loss": 1.2801, "step": 25014 }, { "epoch": 62.5375, "grad_norm": 8.3125, "learning_rate": 6.500875744887274e-06, "loss": 1.4385, "step": 25015 }, { "epoch": 62.54, "grad_norm": 4.34375, "learning_rate": 6.500117254101627e-06, "loss": 1.421, "step": 25016 }, { "epoch": 62.5425, "grad_norm": 6.84375, "learning_rate": 6.499358786261072e-06, "loss": 1.2697, "step": 25017 }, { "epoch": 62.545, "grad_norm": 9.5, "learning_rate": 6.49860034137058e-06, "loss": 1.4842, "step": 25018 }, { "epoch": 62.5475, "grad_norm": 7.125, "learning_rate": 6.497841919435124e-06, "loss": 1.2119, "step": 25019 }, { "epoch": 62.55, "grad_norm": 7.125, "learning_rate": 6.497083520459674e-06, "loss": 1.4185, "step": 25020 }, { "epoch": 62.5525, "grad_norm": 7.25, "learning_rate": 6.496325144449208e-06, "loss": 1.2531, "step": 25021 }, { "epoch": 62.555, "grad_norm": 8.6875, "learning_rate": 6.495566791408693e-06, "loss": 1.2487, "step": 25022 }, { "epoch": 62.5575, "grad_norm": 7.0625, "learning_rate": 6.494808461343101e-06, "loss": 1.3378, "step": 25023 }, { "epoch": 62.56, "grad_norm": 6.21875, "learning_rate": 6.494050154257408e-06, "loss": 1.2569, "step": 25024 }, { "epoch": 62.5625, "grad_norm": 7.125, "learning_rate": 6.493291870156575e-06, "loss": 1.2932, "step": 25025 }, { "epoch": 62.565, "grad_norm": 6.3125, "learning_rate": 6.4925336090455834e-06, "loss": 1.3577, "step": 25026 }, { "epoch": 62.5675, "grad_norm": 7.0, "learning_rate": 6.491775370929403e-06, "loss": 1.3294, "step": 25027 }, { "epoch": 62.57, "grad_norm": 6.96875, "learning_rate": 6.4910171558130016e-06, "loss": 1.1629, "step": 25028 }, { "epoch": 62.5725, "grad_norm": 8.5625, "learning_rate": 6.4902589637013524e-06, "loss": 1.4913, "step": 25029 }, { "epoch": 62.575, "grad_norm": 7.53125, "learning_rate": 6.489500794599422e-06, "loss": 1.406, "step": 25030 }, { "epoch": 62.5775, "grad_norm": 7.0625, "learning_rate": 6.4887426485121855e-06, "loss": 1.377, "step": 25031 }, { "epoch": 62.58, "grad_norm": 5.71875, "learning_rate": 6.487984525444613e-06, "loss": 1.2263, "step": 25032 }, { "epoch": 62.5825, "grad_norm": 8.8125, "learning_rate": 6.4872264254016714e-06, "loss": 1.3161, "step": 25033 }, { "epoch": 62.585, "grad_norm": 8.5, "learning_rate": 6.486468348388336e-06, "loss": 1.2631, "step": 25034 }, { "epoch": 62.5875, "grad_norm": 7.78125, "learning_rate": 6.485710294409567e-06, "loss": 1.1986, "step": 25035 }, { "epoch": 62.59, "grad_norm": 6.34375, "learning_rate": 6.484952263470348e-06, "loss": 1.1096, "step": 25036 }, { "epoch": 62.5925, "grad_norm": 9.5, "learning_rate": 6.484194255575637e-06, "loss": 1.3765, "step": 25037 }, { "epoch": 62.595, "grad_norm": 7.28125, "learning_rate": 6.48343627073041e-06, "loss": 1.5369, "step": 25038 }, { "epoch": 62.5975, "grad_norm": 8.3125, "learning_rate": 6.482678308939634e-06, "loss": 1.1537, "step": 25039 }, { "epoch": 62.6, "grad_norm": 8.4375, "learning_rate": 6.481920370208274e-06, "loss": 1.4623, "step": 25040 }, { "epoch": 62.6025, "grad_norm": 5.875, "learning_rate": 6.481162454541308e-06, "loss": 1.3534, "step": 25041 }, { "epoch": 62.605, "grad_norm": 7.8125, "learning_rate": 6.480404561943697e-06, "loss": 1.4037, "step": 25042 }, { "epoch": 62.6075, "grad_norm": 6.71875, "learning_rate": 6.479646692420416e-06, "loss": 1.4224, "step": 25043 }, { "epoch": 62.61, "grad_norm": 9.375, "learning_rate": 6.47888884597643e-06, "loss": 1.3514, "step": 25044 }, { "epoch": 62.6125, "grad_norm": 6.9375, "learning_rate": 6.4781310226167035e-06, "loss": 1.3322, "step": 25045 }, { "epoch": 62.615, "grad_norm": 9.1875, "learning_rate": 6.4773732223462125e-06, "loss": 1.4356, "step": 25046 }, { "epoch": 62.6175, "grad_norm": 7.75, "learning_rate": 6.476615445169924e-06, "loss": 1.369, "step": 25047 }, { "epoch": 62.62, "grad_norm": 8.25, "learning_rate": 6.475857691092801e-06, "loss": 1.4667, "step": 25048 }, { "epoch": 62.6225, "grad_norm": 10.4375, "learning_rate": 6.475099960119815e-06, "loss": 1.3727, "step": 25049 }, { "epoch": 62.625, "grad_norm": 7.125, "learning_rate": 6.474342252255927e-06, "loss": 1.4622, "step": 25050 }, { "epoch": 62.6275, "grad_norm": 9.0, "learning_rate": 6.473584567506117e-06, "loss": 1.3968, "step": 25051 }, { "epoch": 62.63, "grad_norm": 8.125, "learning_rate": 6.472826905875343e-06, "loss": 1.3206, "step": 25052 }, { "epoch": 62.6325, "grad_norm": 8.5625, "learning_rate": 6.472069267368577e-06, "loss": 1.3536, "step": 25053 }, { "epoch": 62.635, "grad_norm": 7.8125, "learning_rate": 6.4713116519907814e-06, "loss": 1.2907, "step": 25054 }, { "epoch": 62.6375, "grad_norm": 8.0625, "learning_rate": 6.4705540597469245e-06, "loss": 1.3256, "step": 25055 }, { "epoch": 62.64, "grad_norm": 7.15625, "learning_rate": 6.469796490641974e-06, "loss": 1.4116, "step": 25056 }, { "epoch": 62.6425, "grad_norm": 9.8125, "learning_rate": 6.4690389446809e-06, "loss": 1.2338, "step": 25057 }, { "epoch": 62.645, "grad_norm": 6.34375, "learning_rate": 6.468281421868663e-06, "loss": 1.3511, "step": 25058 }, { "epoch": 62.6475, "grad_norm": 8.0625, "learning_rate": 6.467523922210231e-06, "loss": 1.3685, "step": 25059 }, { "epoch": 62.65, "grad_norm": 6.65625, "learning_rate": 6.466766445710568e-06, "loss": 1.2572, "step": 25060 }, { "epoch": 62.6525, "grad_norm": 7.875, "learning_rate": 6.466008992374649e-06, "loss": 1.2291, "step": 25061 }, { "epoch": 62.655, "grad_norm": 6.1875, "learning_rate": 6.465251562207431e-06, "loss": 1.4018, "step": 25062 }, { "epoch": 62.6575, "grad_norm": 7.875, "learning_rate": 6.464494155213881e-06, "loss": 1.1774, "step": 25063 }, { "epoch": 62.66, "grad_norm": 7.78125, "learning_rate": 6.46373677139897e-06, "loss": 1.3596, "step": 25064 }, { "epoch": 62.6625, "grad_norm": 5.34375, "learning_rate": 6.4629794107676535e-06, "loss": 1.1633, "step": 25065 }, { "epoch": 62.665, "grad_norm": 10.5625, "learning_rate": 6.4622220733249045e-06, "loss": 1.435, "step": 25066 }, { "epoch": 62.6675, "grad_norm": 6.5, "learning_rate": 6.461464759075686e-06, "loss": 1.2709, "step": 25067 }, { "epoch": 62.67, "grad_norm": 6.84375, "learning_rate": 6.460707468024965e-06, "loss": 1.3887, "step": 25068 }, { "epoch": 62.6725, "grad_norm": 6.46875, "learning_rate": 6.459950200177702e-06, "loss": 1.1849, "step": 25069 }, { "epoch": 62.675, "grad_norm": 6.625, "learning_rate": 6.459192955538864e-06, "loss": 1.4267, "step": 25070 }, { "epoch": 62.6775, "grad_norm": 6.53125, "learning_rate": 6.458435734113415e-06, "loss": 1.374, "step": 25071 }, { "epoch": 62.68, "grad_norm": 7.40625, "learning_rate": 6.4576785359063225e-06, "loss": 1.4642, "step": 25072 }, { "epoch": 62.6825, "grad_norm": 7.78125, "learning_rate": 6.4569213609225455e-06, "loss": 1.308, "step": 25073 }, { "epoch": 62.685, "grad_norm": 8.1875, "learning_rate": 6.45616420916705e-06, "loss": 1.364, "step": 25074 }, { "epoch": 62.6875, "grad_norm": 8.25, "learning_rate": 6.455407080644797e-06, "loss": 1.4834, "step": 25075 }, { "epoch": 62.69, "grad_norm": 7.8125, "learning_rate": 6.45464997536076e-06, "loss": 1.2591, "step": 25076 }, { "epoch": 62.6925, "grad_norm": 7.375, "learning_rate": 6.453892893319893e-06, "loss": 1.3163, "step": 25077 }, { "epoch": 62.695, "grad_norm": 9.6875, "learning_rate": 6.453135834527162e-06, "loss": 1.4112, "step": 25078 }, { "epoch": 62.6975, "grad_norm": 8.0, "learning_rate": 6.452378798987533e-06, "loss": 1.3059, "step": 25079 }, { "epoch": 62.7, "grad_norm": 9.8125, "learning_rate": 6.4516217867059615e-06, "loss": 1.1881, "step": 25080 }, { "epoch": 62.7025, "grad_norm": 9.0, "learning_rate": 6.450864797687418e-06, "loss": 1.3137, "step": 25081 }, { "epoch": 62.705, "grad_norm": 7.625, "learning_rate": 6.450107831936863e-06, "loss": 1.3821, "step": 25082 }, { "epoch": 62.7075, "grad_norm": 7.9375, "learning_rate": 6.44935088945926e-06, "loss": 1.2849, "step": 25083 }, { "epoch": 62.71, "grad_norm": 6.96875, "learning_rate": 6.448593970259571e-06, "loss": 1.1581, "step": 25084 }, { "epoch": 62.7125, "grad_norm": 6.625, "learning_rate": 6.4478370743427535e-06, "loss": 1.1725, "step": 25085 }, { "epoch": 62.715, "grad_norm": 6.1875, "learning_rate": 6.447080201713777e-06, "loss": 1.2007, "step": 25086 }, { "epoch": 62.7175, "grad_norm": 8.9375, "learning_rate": 6.446323352377603e-06, "loss": 1.394, "step": 25087 }, { "epoch": 62.72, "grad_norm": 5.5625, "learning_rate": 6.445566526339187e-06, "loss": 1.3791, "step": 25088 }, { "epoch": 62.7225, "grad_norm": 7.4375, "learning_rate": 6.4448097236034935e-06, "loss": 1.3984, "step": 25089 }, { "epoch": 62.725, "grad_norm": 9.5625, "learning_rate": 6.4440529441754885e-06, "loss": 1.3351, "step": 25090 }, { "epoch": 62.7275, "grad_norm": 6.125, "learning_rate": 6.443296188060131e-06, "loss": 1.1411, "step": 25091 }, { "epoch": 62.73, "grad_norm": 6.9375, "learning_rate": 6.44253945526238e-06, "loss": 1.1043, "step": 25092 }, { "epoch": 62.7325, "grad_norm": 9.125, "learning_rate": 6.441782745787197e-06, "loss": 1.5843, "step": 25093 }, { "epoch": 62.735, "grad_norm": 6.03125, "learning_rate": 6.441026059639542e-06, "loss": 1.3579, "step": 25094 }, { "epoch": 62.7375, "grad_norm": 6.5, "learning_rate": 6.440269396824381e-06, "loss": 1.4403, "step": 25095 }, { "epoch": 62.74, "grad_norm": 7.0, "learning_rate": 6.439512757346671e-06, "loss": 1.2959, "step": 25096 }, { "epoch": 62.7425, "grad_norm": 7.09375, "learning_rate": 6.438756141211373e-06, "loss": 1.3842, "step": 25097 }, { "epoch": 62.745, "grad_norm": 6.84375, "learning_rate": 6.437999548423448e-06, "loss": 1.2109, "step": 25098 }, { "epoch": 62.7475, "grad_norm": 6.96875, "learning_rate": 6.437242978987852e-06, "loss": 1.4011, "step": 25099 }, { "epoch": 62.75, "grad_norm": 8.0625, "learning_rate": 6.43648643290955e-06, "loss": 1.384, "step": 25100 }, { "epoch": 62.7525, "grad_norm": 5.8125, "learning_rate": 6.4357299101935025e-06, "loss": 1.2729, "step": 25101 }, { "epoch": 62.755, "grad_norm": 6.75, "learning_rate": 6.434973410844667e-06, "loss": 1.4311, "step": 25102 }, { "epoch": 62.7575, "grad_norm": 7.625, "learning_rate": 6.434216934868e-06, "loss": 1.4185, "step": 25103 }, { "epoch": 62.76, "grad_norm": 4.9375, "learning_rate": 6.4334604822684645e-06, "loss": 1.2747, "step": 25104 }, { "epoch": 62.7625, "grad_norm": 9.25, "learning_rate": 6.43270405305102e-06, "loss": 1.3445, "step": 25105 }, { "epoch": 62.765, "grad_norm": 10.3125, "learning_rate": 6.4319476472206274e-06, "loss": 1.3674, "step": 25106 }, { "epoch": 62.7675, "grad_norm": 8.9375, "learning_rate": 6.431191264782242e-06, "loss": 1.4909, "step": 25107 }, { "epoch": 62.77, "grad_norm": 8.1875, "learning_rate": 6.430434905740825e-06, "loss": 1.2052, "step": 25108 }, { "epoch": 62.7725, "grad_norm": 5.65625, "learning_rate": 6.429678570101328e-06, "loss": 1.2199, "step": 25109 }, { "epoch": 62.775, "grad_norm": 6.90625, "learning_rate": 6.428922257868721e-06, "loss": 1.2007, "step": 25110 }, { "epoch": 62.7775, "grad_norm": 8.375, "learning_rate": 6.428165969047955e-06, "loss": 1.5365, "step": 25111 }, { "epoch": 62.78, "grad_norm": 8.3125, "learning_rate": 6.42740970364399e-06, "loss": 1.3559, "step": 25112 }, { "epoch": 62.7825, "grad_norm": 5.28125, "learning_rate": 6.426653461661787e-06, "loss": 1.3424, "step": 25113 }, { "epoch": 62.785, "grad_norm": 7.0625, "learning_rate": 6.425897243106296e-06, "loss": 1.4254, "step": 25114 }, { "epoch": 62.7875, "grad_norm": 9.0, "learning_rate": 6.425141047982482e-06, "loss": 1.3501, "step": 25115 }, { "epoch": 62.79, "grad_norm": 7.5, "learning_rate": 6.4243848762953e-06, "loss": 1.3737, "step": 25116 }, { "epoch": 62.7925, "grad_norm": 5.53125, "learning_rate": 6.423628728049709e-06, "loss": 1.3553, "step": 25117 }, { "epoch": 62.795, "grad_norm": 5.09375, "learning_rate": 6.422872603250664e-06, "loss": 1.2905, "step": 25118 }, { "epoch": 62.7975, "grad_norm": 8.375, "learning_rate": 6.42211650190312e-06, "loss": 1.2134, "step": 25119 }, { "epoch": 62.8, "grad_norm": 7.1875, "learning_rate": 6.421360424012039e-06, "loss": 1.3216, "step": 25120 }, { "epoch": 62.8025, "grad_norm": 13.125, "learning_rate": 6.420604369582379e-06, "loss": 1.3087, "step": 25121 }, { "epoch": 62.805, "grad_norm": 10.125, "learning_rate": 6.419848338619091e-06, "loss": 1.2579, "step": 25122 }, { "epoch": 62.8075, "grad_norm": 5.9375, "learning_rate": 6.419092331127137e-06, "loss": 1.4131, "step": 25123 }, { "epoch": 62.81, "grad_norm": 8.625, "learning_rate": 6.418336347111463e-06, "loss": 1.3172, "step": 25124 }, { "epoch": 62.8125, "grad_norm": 9.4375, "learning_rate": 6.4175803865770395e-06, "loss": 1.2955, "step": 25125 }, { "epoch": 62.815, "grad_norm": 8.3125, "learning_rate": 6.416824449528813e-06, "loss": 1.4138, "step": 25126 }, { "epoch": 62.8175, "grad_norm": 6.03125, "learning_rate": 6.416068535971744e-06, "loss": 1.1318, "step": 25127 }, { "epoch": 62.82, "grad_norm": 4.9375, "learning_rate": 6.415312645910785e-06, "loss": 1.1868, "step": 25128 }, { "epoch": 62.8225, "grad_norm": 14.125, "learning_rate": 6.414556779350889e-06, "loss": 1.2555, "step": 25129 }, { "epoch": 62.825, "grad_norm": 7.28125, "learning_rate": 6.413800936297019e-06, "loss": 1.317, "step": 25130 }, { "epoch": 62.8275, "grad_norm": 8.875, "learning_rate": 6.413045116754129e-06, "loss": 1.6023, "step": 25131 }, { "epoch": 62.83, "grad_norm": 11.875, "learning_rate": 6.412289320727167e-06, "loss": 1.3532, "step": 25132 }, { "epoch": 62.8325, "grad_norm": 5.5, "learning_rate": 6.4115335482210964e-06, "loss": 1.2299, "step": 25133 }, { "epoch": 62.835, "grad_norm": 10.0, "learning_rate": 6.410777799240864e-06, "loss": 1.302, "step": 25134 }, { "epoch": 62.8375, "grad_norm": 9.3125, "learning_rate": 6.410022073791432e-06, "loss": 1.1807, "step": 25135 }, { "epoch": 62.84, "grad_norm": 6.09375, "learning_rate": 6.409266371877751e-06, "loss": 1.2284, "step": 25136 }, { "epoch": 62.8425, "grad_norm": 9.9375, "learning_rate": 6.408510693504774e-06, "loss": 1.3726, "step": 25137 }, { "epoch": 62.845, "grad_norm": 5.0, "learning_rate": 6.4077550386774625e-06, "loss": 1.1811, "step": 25138 }, { "epoch": 62.8475, "grad_norm": 7.8125, "learning_rate": 6.406999407400757e-06, "loss": 1.3931, "step": 25139 }, { "epoch": 62.85, "grad_norm": 4.4375, "learning_rate": 6.406243799679625e-06, "loss": 1.2874, "step": 25140 }, { "epoch": 62.8525, "grad_norm": 7.875, "learning_rate": 6.405488215519014e-06, "loss": 1.3684, "step": 25141 }, { "epoch": 62.855, "grad_norm": 8.1875, "learning_rate": 6.404732654923882e-06, "loss": 1.3382, "step": 25142 }, { "epoch": 62.8575, "grad_norm": 6.0625, "learning_rate": 6.403977117899174e-06, "loss": 1.2865, "step": 25143 }, { "epoch": 62.86, "grad_norm": 8.3125, "learning_rate": 6.403221604449847e-06, "loss": 1.2476, "step": 25144 }, { "epoch": 62.8625, "grad_norm": 7.1875, "learning_rate": 6.4024661145808586e-06, "loss": 1.1554, "step": 25145 }, { "epoch": 62.865, "grad_norm": 8.625, "learning_rate": 6.401710648297159e-06, "loss": 1.4726, "step": 25146 }, { "epoch": 62.8675, "grad_norm": 6.90625, "learning_rate": 6.400955205603698e-06, "loss": 1.5035, "step": 25147 }, { "epoch": 62.87, "grad_norm": 9.5, "learning_rate": 6.400199786505432e-06, "loss": 1.3441, "step": 25148 }, { "epoch": 62.8725, "grad_norm": 9.9375, "learning_rate": 6.399444391007309e-06, "loss": 1.3101, "step": 25149 }, { "epoch": 62.875, "grad_norm": 6.0, "learning_rate": 6.398689019114289e-06, "loss": 1.1852, "step": 25150 }, { "epoch": 62.8775, "grad_norm": 8.6875, "learning_rate": 6.397933670831318e-06, "loss": 1.2941, "step": 25151 }, { "epoch": 62.88, "grad_norm": 6.90625, "learning_rate": 6.397178346163348e-06, "loss": 1.2951, "step": 25152 }, { "epoch": 62.8825, "grad_norm": 6.125, "learning_rate": 6.396423045115336e-06, "loss": 1.2452, "step": 25153 }, { "epoch": 62.885, "grad_norm": 4.75, "learning_rate": 6.395667767692226e-06, "loss": 1.2933, "step": 25154 }, { "epoch": 62.8875, "grad_norm": 7.3125, "learning_rate": 6.394912513898977e-06, "loss": 1.3799, "step": 25155 }, { "epoch": 62.89, "grad_norm": 9.3125, "learning_rate": 6.394157283740536e-06, "loss": 1.2669, "step": 25156 }, { "epoch": 62.8925, "grad_norm": 6.9375, "learning_rate": 6.3934020772218585e-06, "loss": 1.0974, "step": 25157 }, { "epoch": 62.895, "grad_norm": 7.625, "learning_rate": 6.39264689434789e-06, "loss": 1.3029, "step": 25158 }, { "epoch": 62.8975, "grad_norm": 8.5625, "learning_rate": 6.391891735123582e-06, "loss": 1.278, "step": 25159 }, { "epoch": 62.9, "grad_norm": 7.84375, "learning_rate": 6.39113659955389e-06, "loss": 1.2005, "step": 25160 }, { "epoch": 62.9025, "grad_norm": 8.125, "learning_rate": 6.390381487643763e-06, "loss": 1.2454, "step": 25161 }, { "epoch": 62.905, "grad_norm": 9.5625, "learning_rate": 6.38962639939815e-06, "loss": 1.2307, "step": 25162 }, { "epoch": 62.9075, "grad_norm": 6.25, "learning_rate": 6.388871334822e-06, "loss": 1.345, "step": 25163 }, { "epoch": 62.91, "grad_norm": 9.5, "learning_rate": 6.3881162939202645e-06, "loss": 1.268, "step": 25164 }, { "epoch": 62.9125, "grad_norm": 7.46875, "learning_rate": 6.387361276697897e-06, "loss": 1.3045, "step": 25165 }, { "epoch": 62.915, "grad_norm": 7.5625, "learning_rate": 6.3866062831598444e-06, "loss": 1.404, "step": 25166 }, { "epoch": 62.9175, "grad_norm": 10.4375, "learning_rate": 6.385851313311053e-06, "loss": 1.4067, "step": 25167 }, { "epoch": 62.92, "grad_norm": 7.25, "learning_rate": 6.38509636715648e-06, "loss": 1.4585, "step": 25168 }, { "epoch": 62.9225, "grad_norm": 6.125, "learning_rate": 6.384341444701068e-06, "loss": 1.4089, "step": 25169 }, { "epoch": 62.925, "grad_norm": 8.9375, "learning_rate": 6.383586545949768e-06, "loss": 1.347, "step": 25170 }, { "epoch": 62.9275, "grad_norm": 7.375, "learning_rate": 6.3828316709075314e-06, "loss": 1.3356, "step": 25171 }, { "epoch": 62.93, "grad_norm": 11.0, "learning_rate": 6.382076819579306e-06, "loss": 1.5319, "step": 25172 }, { "epoch": 62.9325, "grad_norm": 6.1875, "learning_rate": 6.38132199197004e-06, "loss": 1.2235, "step": 25173 }, { "epoch": 62.935, "grad_norm": 8.1875, "learning_rate": 6.380567188084679e-06, "loss": 1.32, "step": 25174 }, { "epoch": 62.9375, "grad_norm": 7.4375, "learning_rate": 6.379812407928178e-06, "loss": 1.2407, "step": 25175 }, { "epoch": 62.94, "grad_norm": 7.1875, "learning_rate": 6.3790576515054824e-06, "loss": 1.267, "step": 25176 }, { "epoch": 62.9425, "grad_norm": 12.5625, "learning_rate": 6.378302918821539e-06, "loss": 1.3813, "step": 25177 }, { "epoch": 62.945, "grad_norm": 6.8125, "learning_rate": 6.377548209881299e-06, "loss": 1.2215, "step": 25178 }, { "epoch": 62.9475, "grad_norm": 5.71875, "learning_rate": 6.376793524689702e-06, "loss": 1.4509, "step": 25179 }, { "epoch": 62.95, "grad_norm": 8.8125, "learning_rate": 6.376038863251706e-06, "loss": 1.3136, "step": 25180 }, { "epoch": 62.9525, "grad_norm": 7.5, "learning_rate": 6.375284225572253e-06, "loss": 1.2798, "step": 25181 }, { "epoch": 62.955, "grad_norm": 6.09375, "learning_rate": 6.374529611656293e-06, "loss": 1.2945, "step": 25182 }, { "epoch": 62.9575, "grad_norm": 8.6875, "learning_rate": 6.373775021508773e-06, "loss": 1.2683, "step": 25183 }, { "epoch": 62.96, "grad_norm": 7.0625, "learning_rate": 6.373020455134633e-06, "loss": 1.2823, "step": 25184 }, { "epoch": 62.9625, "grad_norm": 7.84375, "learning_rate": 6.37226591253883e-06, "loss": 1.401, "step": 25185 }, { "epoch": 62.965, "grad_norm": 7.40625, "learning_rate": 6.371511393726306e-06, "loss": 1.2633, "step": 25186 }, { "epoch": 62.9675, "grad_norm": 8.9375, "learning_rate": 6.370756898702009e-06, "loss": 1.3537, "step": 25187 }, { "epoch": 62.97, "grad_norm": 6.75, "learning_rate": 6.370002427470884e-06, "loss": 1.3281, "step": 25188 }, { "epoch": 62.9725, "grad_norm": 6.59375, "learning_rate": 6.369247980037875e-06, "loss": 1.2426, "step": 25189 }, { "epoch": 62.975, "grad_norm": 12.875, "learning_rate": 6.368493556407933e-06, "loss": 1.4031, "step": 25190 }, { "epoch": 62.9775, "grad_norm": 9.125, "learning_rate": 6.367739156586004e-06, "loss": 1.319, "step": 25191 }, { "epoch": 62.98, "grad_norm": 8.25, "learning_rate": 6.36698478057703e-06, "loss": 1.3301, "step": 25192 }, { "epoch": 62.9825, "grad_norm": 8.1875, "learning_rate": 6.36623042838596e-06, "loss": 1.2351, "step": 25193 }, { "epoch": 62.985, "grad_norm": 7.0, "learning_rate": 6.365476100017732e-06, "loss": 1.1789, "step": 25194 }, { "epoch": 62.9875, "grad_norm": 5.59375, "learning_rate": 6.3647217954773045e-06, "loss": 1.1144, "step": 25195 }, { "epoch": 62.99, "grad_norm": 7.3125, "learning_rate": 6.363967514769611e-06, "loss": 1.2678, "step": 25196 }, { "epoch": 62.9925, "grad_norm": 8.8125, "learning_rate": 6.363213257899605e-06, "loss": 1.4704, "step": 25197 }, { "epoch": 62.995, "grad_norm": 7.78125, "learning_rate": 6.3624590248722255e-06, "loss": 1.3212, "step": 25198 }, { "epoch": 62.9975, "grad_norm": 4.65625, "learning_rate": 6.361704815692415e-06, "loss": 1.2973, "step": 25199 }, { "epoch": 63.0, "grad_norm": 7.1875, "learning_rate": 6.360950630365126e-06, "loss": 1.3612, "step": 25200 }, { "epoch": 63.0025, "grad_norm": 7.625, "learning_rate": 6.3601964688953e-06, "loss": 1.3077, "step": 25201 }, { "epoch": 63.005, "grad_norm": 5.28125, "learning_rate": 6.3594423312878785e-06, "loss": 1.1081, "step": 25202 }, { "epoch": 63.0075, "grad_norm": 7.125, "learning_rate": 6.358688217547807e-06, "loss": 1.3001, "step": 25203 }, { "epoch": 63.01, "grad_norm": 7.125, "learning_rate": 6.357934127680029e-06, "loss": 1.2385, "step": 25204 }, { "epoch": 63.0125, "grad_norm": 9.5, "learning_rate": 6.357180061689492e-06, "loss": 1.3477, "step": 25205 }, { "epoch": 63.015, "grad_norm": 6.125, "learning_rate": 6.356426019581134e-06, "loss": 1.1585, "step": 25206 }, { "epoch": 63.0175, "grad_norm": 7.34375, "learning_rate": 6.355672001359904e-06, "loss": 1.3226, "step": 25207 }, { "epoch": 63.02, "grad_norm": 8.125, "learning_rate": 6.354918007030743e-06, "loss": 1.2589, "step": 25208 }, { "epoch": 63.0225, "grad_norm": 7.53125, "learning_rate": 6.354164036598587e-06, "loss": 1.3151, "step": 25209 }, { "epoch": 63.025, "grad_norm": 7.28125, "learning_rate": 6.353410090068393e-06, "loss": 1.4207, "step": 25210 }, { "epoch": 63.0275, "grad_norm": 4.96875, "learning_rate": 6.352656167445092e-06, "loss": 1.356, "step": 25211 }, { "epoch": 63.03, "grad_norm": 8.25, "learning_rate": 6.351902268733636e-06, "loss": 1.279, "step": 25212 }, { "epoch": 63.0325, "grad_norm": 8.5625, "learning_rate": 6.3511483939389595e-06, "loss": 1.388, "step": 25213 }, { "epoch": 63.035, "grad_norm": 6.09375, "learning_rate": 6.350394543066007e-06, "loss": 1.3295, "step": 25214 }, { "epoch": 63.0375, "grad_norm": 8.3125, "learning_rate": 6.349640716119723e-06, "loss": 1.2115, "step": 25215 }, { "epoch": 63.04, "grad_norm": 7.375, "learning_rate": 6.3488869131050505e-06, "loss": 1.3348, "step": 25216 }, { "epoch": 63.0425, "grad_norm": 5.59375, "learning_rate": 6.348133134026928e-06, "loss": 1.237, "step": 25217 }, { "epoch": 63.045, "grad_norm": 7.90625, "learning_rate": 6.347379378890297e-06, "loss": 1.3422, "step": 25218 }, { "epoch": 63.0475, "grad_norm": 7.34375, "learning_rate": 6.3466256477001e-06, "loss": 1.2668, "step": 25219 }, { "epoch": 63.05, "grad_norm": 7.3125, "learning_rate": 6.345871940461282e-06, "loss": 1.1319, "step": 25220 }, { "epoch": 63.0525, "grad_norm": 7.625, "learning_rate": 6.345118257178779e-06, "loss": 1.4466, "step": 25221 }, { "epoch": 63.055, "grad_norm": 5.5, "learning_rate": 6.344364597857538e-06, "loss": 1.2184, "step": 25222 }, { "epoch": 63.0575, "grad_norm": 9.75, "learning_rate": 6.343610962502494e-06, "loss": 1.2334, "step": 25223 }, { "epoch": 63.06, "grad_norm": 5.28125, "learning_rate": 6.342857351118588e-06, "loss": 1.337, "step": 25224 }, { "epoch": 63.0625, "grad_norm": 5.625, "learning_rate": 6.342103763710765e-06, "loss": 1.2854, "step": 25225 }, { "epoch": 63.065, "grad_norm": 7.3125, "learning_rate": 6.341350200283963e-06, "loss": 1.4964, "step": 25226 }, { "epoch": 63.0675, "grad_norm": 7.09375, "learning_rate": 6.340596660843126e-06, "loss": 1.188, "step": 25227 }, { "epoch": 63.07, "grad_norm": 6.53125, "learning_rate": 6.339843145393188e-06, "loss": 1.2876, "step": 25228 }, { "epoch": 63.0725, "grad_norm": 4.5, "learning_rate": 6.339089653939088e-06, "loss": 1.3848, "step": 25229 }, { "epoch": 63.075, "grad_norm": 8.0, "learning_rate": 6.338336186485774e-06, "loss": 1.3614, "step": 25230 }, { "epoch": 63.0775, "grad_norm": 7.65625, "learning_rate": 6.3375827430381816e-06, "loss": 1.3389, "step": 25231 }, { "epoch": 63.08, "grad_norm": 6.40625, "learning_rate": 6.33682932360125e-06, "loss": 1.3732, "step": 25232 }, { "epoch": 63.0825, "grad_norm": 5.125, "learning_rate": 6.336075928179919e-06, "loss": 1.386, "step": 25233 }, { "epoch": 63.085, "grad_norm": 5.59375, "learning_rate": 6.335322556779124e-06, "loss": 1.2868, "step": 25234 }, { "epoch": 63.0875, "grad_norm": 9.625, "learning_rate": 6.334569209403811e-06, "loss": 1.5036, "step": 25235 }, { "epoch": 63.09, "grad_norm": 7.65625, "learning_rate": 6.333815886058916e-06, "loss": 1.3662, "step": 25236 }, { "epoch": 63.0925, "grad_norm": 15.4375, "learning_rate": 6.333062586749377e-06, "loss": 1.2662, "step": 25237 }, { "epoch": 63.095, "grad_norm": 6.09375, "learning_rate": 6.332309311480134e-06, "loss": 1.5082, "step": 25238 }, { "epoch": 63.0975, "grad_norm": 9.3125, "learning_rate": 6.33155606025612e-06, "loss": 1.3363, "step": 25239 }, { "epoch": 63.1, "grad_norm": 7.25, "learning_rate": 6.33080283308228e-06, "loss": 1.287, "step": 25240 }, { "epoch": 63.1025, "grad_norm": 6.625, "learning_rate": 6.330049629963551e-06, "loss": 1.4281, "step": 25241 }, { "epoch": 63.105, "grad_norm": 6.9375, "learning_rate": 6.329296450904869e-06, "loss": 1.4633, "step": 25242 }, { "epoch": 63.1075, "grad_norm": 7.5625, "learning_rate": 6.328543295911173e-06, "loss": 1.3685, "step": 25243 }, { "epoch": 63.11, "grad_norm": 7.4375, "learning_rate": 6.3277901649873974e-06, "loss": 1.3126, "step": 25244 }, { "epoch": 63.1125, "grad_norm": 5.90625, "learning_rate": 6.327037058138486e-06, "loss": 1.2489, "step": 25245 }, { "epoch": 63.115, "grad_norm": 7.5625, "learning_rate": 6.3262839753693715e-06, "loss": 1.3737, "step": 25246 }, { "epoch": 63.1175, "grad_norm": 10.0625, "learning_rate": 6.325530916684992e-06, "loss": 1.3476, "step": 25247 }, { "epoch": 63.12, "grad_norm": 5.125, "learning_rate": 6.324777882090287e-06, "loss": 1.1144, "step": 25248 }, { "epoch": 63.1225, "grad_norm": 4.5625, "learning_rate": 6.324024871590186e-06, "loss": 1.2352, "step": 25249 }, { "epoch": 63.125, "grad_norm": 6.0625, "learning_rate": 6.323271885189636e-06, "loss": 1.5374, "step": 25250 }, { "epoch": 63.1275, "grad_norm": 7.96875, "learning_rate": 6.322518922893565e-06, "loss": 1.3877, "step": 25251 }, { "epoch": 63.13, "grad_norm": 8.5625, "learning_rate": 6.321765984706915e-06, "loss": 1.1973, "step": 25252 }, { "epoch": 63.1325, "grad_norm": 6.1875, "learning_rate": 6.32101307063462e-06, "loss": 1.2594, "step": 25253 }, { "epoch": 63.135, "grad_norm": 7.46875, "learning_rate": 6.3202601806816144e-06, "loss": 1.3265, "step": 25254 }, { "epoch": 63.1375, "grad_norm": 7.4375, "learning_rate": 6.319507314852835e-06, "loss": 1.3883, "step": 25255 }, { "epoch": 63.14, "grad_norm": 9.125, "learning_rate": 6.318754473153221e-06, "loss": 1.3483, "step": 25256 }, { "epoch": 63.1425, "grad_norm": 6.28125, "learning_rate": 6.318001655587706e-06, "loss": 1.4124, "step": 25257 }, { "epoch": 63.145, "grad_norm": 6.9375, "learning_rate": 6.317248862161222e-06, "loss": 1.4409, "step": 25258 }, { "epoch": 63.1475, "grad_norm": 6.53125, "learning_rate": 6.316496092878708e-06, "loss": 1.2366, "step": 25259 }, { "epoch": 63.15, "grad_norm": 8.25, "learning_rate": 6.315743347745098e-06, "loss": 1.308, "step": 25260 }, { "epoch": 63.1525, "grad_norm": 9.25, "learning_rate": 6.314990626765329e-06, "loss": 1.3228, "step": 25261 }, { "epoch": 63.155, "grad_norm": 8.1875, "learning_rate": 6.314237929944333e-06, "loss": 1.3924, "step": 25262 }, { "epoch": 63.1575, "grad_norm": 9.125, "learning_rate": 6.313485257287047e-06, "loss": 1.3846, "step": 25263 }, { "epoch": 63.16, "grad_norm": 10.0, "learning_rate": 6.3127326087983974e-06, "loss": 1.3779, "step": 25264 }, { "epoch": 63.1625, "grad_norm": 6.65625, "learning_rate": 6.3119799844833315e-06, "loss": 1.2278, "step": 25265 }, { "epoch": 63.165, "grad_norm": 5.9375, "learning_rate": 6.311227384346776e-06, "loss": 1.3587, "step": 25266 }, { "epoch": 63.1675, "grad_norm": 5.90625, "learning_rate": 6.31047480839367e-06, "loss": 1.2151, "step": 25267 }, { "epoch": 63.17, "grad_norm": 6.8125, "learning_rate": 6.309722256628938e-06, "loss": 1.3578, "step": 25268 }, { "epoch": 63.1725, "grad_norm": 10.9375, "learning_rate": 6.308969729057519e-06, "loss": 1.2488, "step": 25269 }, { "epoch": 63.175, "grad_norm": 7.125, "learning_rate": 6.308217225684348e-06, "loss": 1.3785, "step": 25270 }, { "epoch": 63.1775, "grad_norm": 6.6875, "learning_rate": 6.30746474651436e-06, "loss": 1.4397, "step": 25271 }, { "epoch": 63.18, "grad_norm": 8.5, "learning_rate": 6.306712291552484e-06, "loss": 1.4475, "step": 25272 }, { "epoch": 63.1825, "grad_norm": 7.46875, "learning_rate": 6.305959860803652e-06, "loss": 1.2435, "step": 25273 }, { "epoch": 63.185, "grad_norm": 7.625, "learning_rate": 6.305207454272799e-06, "loss": 1.3617, "step": 25274 }, { "epoch": 63.1875, "grad_norm": 5.71875, "learning_rate": 6.304455071964861e-06, "loss": 1.2997, "step": 25275 }, { "epoch": 63.19, "grad_norm": 5.34375, "learning_rate": 6.303702713884767e-06, "loss": 1.3745, "step": 25276 }, { "epoch": 63.1925, "grad_norm": 7.3125, "learning_rate": 6.3029503800374495e-06, "loss": 1.3486, "step": 25277 }, { "epoch": 63.195, "grad_norm": 10.375, "learning_rate": 6.302198070427843e-06, "loss": 1.3884, "step": 25278 }, { "epoch": 63.1975, "grad_norm": 7.09375, "learning_rate": 6.301445785060873e-06, "loss": 1.3281, "step": 25279 }, { "epoch": 63.2, "grad_norm": 5.96875, "learning_rate": 6.300693523941481e-06, "loss": 1.308, "step": 25280 }, { "epoch": 63.2025, "grad_norm": 6.75, "learning_rate": 6.299941287074594e-06, "loss": 1.3229, "step": 25281 }, { "epoch": 63.205, "grad_norm": 7.3125, "learning_rate": 6.299189074465144e-06, "loss": 1.2995, "step": 25282 }, { "epoch": 63.2075, "grad_norm": 7.5625, "learning_rate": 6.298436886118062e-06, "loss": 1.199, "step": 25283 }, { "epoch": 63.21, "grad_norm": 19.75, "learning_rate": 6.297684722038277e-06, "loss": 1.5258, "step": 25284 }, { "epoch": 63.2125, "grad_norm": 5.78125, "learning_rate": 6.296932582230724e-06, "loss": 1.3279, "step": 25285 }, { "epoch": 63.215, "grad_norm": 8.25, "learning_rate": 6.2961804667003345e-06, "loss": 1.4095, "step": 25286 }, { "epoch": 63.2175, "grad_norm": 6.96875, "learning_rate": 6.295428375452037e-06, "loss": 1.3129, "step": 25287 }, { "epoch": 63.22, "grad_norm": 6.0, "learning_rate": 6.2946763084907625e-06, "loss": 1.3398, "step": 25288 }, { "epoch": 63.2225, "grad_norm": 6.25, "learning_rate": 6.293924265821441e-06, "loss": 1.3143, "step": 25289 }, { "epoch": 63.225, "grad_norm": 5.6875, "learning_rate": 6.293172247449005e-06, "loss": 1.3007, "step": 25290 }, { "epoch": 63.2275, "grad_norm": 8.1875, "learning_rate": 6.292420253378383e-06, "loss": 1.3153, "step": 25291 }, { "epoch": 63.23, "grad_norm": 8.125, "learning_rate": 6.291668283614505e-06, "loss": 1.334, "step": 25292 }, { "epoch": 63.2325, "grad_norm": 9.125, "learning_rate": 6.290916338162304e-06, "loss": 1.1299, "step": 25293 }, { "epoch": 63.235, "grad_norm": 6.96875, "learning_rate": 6.290164417026703e-06, "loss": 1.3547, "step": 25294 }, { "epoch": 63.2375, "grad_norm": 6.46875, "learning_rate": 6.289412520212637e-06, "loss": 1.4057, "step": 25295 }, { "epoch": 63.24, "grad_norm": 9.125, "learning_rate": 6.2886606477250345e-06, "loss": 1.1214, "step": 25296 }, { "epoch": 63.2425, "grad_norm": 7.78125, "learning_rate": 6.287908799568826e-06, "loss": 1.3709, "step": 25297 }, { "epoch": 63.245, "grad_norm": 10.625, "learning_rate": 6.287156975748937e-06, "loss": 1.2795, "step": 25298 }, { "epoch": 63.2475, "grad_norm": 8.625, "learning_rate": 6.286405176270296e-06, "loss": 1.2858, "step": 25299 }, { "epoch": 63.25, "grad_norm": 6.84375, "learning_rate": 6.2856534011378365e-06, "loss": 1.2641, "step": 25300 }, { "epoch": 63.2525, "grad_norm": 9.875, "learning_rate": 6.284901650356486e-06, "loss": 1.2678, "step": 25301 }, { "epoch": 63.255, "grad_norm": 8.4375, "learning_rate": 6.284149923931171e-06, "loss": 1.3022, "step": 25302 }, { "epoch": 63.2575, "grad_norm": 7.25, "learning_rate": 6.283398221866819e-06, "loss": 1.3711, "step": 25303 }, { "epoch": 63.26, "grad_norm": 7.34375, "learning_rate": 6.282646544168358e-06, "loss": 1.2651, "step": 25304 }, { "epoch": 63.2625, "grad_norm": 6.4375, "learning_rate": 6.281894890840721e-06, "loss": 1.3208, "step": 25305 }, { "epoch": 63.265, "grad_norm": 5.53125, "learning_rate": 6.2811432618888315e-06, "loss": 1.3161, "step": 25306 }, { "epoch": 63.2675, "grad_norm": 6.71875, "learning_rate": 6.280391657317617e-06, "loss": 1.336, "step": 25307 }, { "epoch": 63.27, "grad_norm": 6.0625, "learning_rate": 6.279640077132008e-06, "loss": 1.5039, "step": 25308 }, { "epoch": 63.2725, "grad_norm": 7.125, "learning_rate": 6.2788885213369255e-06, "loss": 1.2578, "step": 25309 }, { "epoch": 63.275, "grad_norm": 7.65625, "learning_rate": 6.278136989937305e-06, "loss": 1.2156, "step": 25310 }, { "epoch": 63.2775, "grad_norm": 13.5, "learning_rate": 6.277385482938068e-06, "loss": 1.2896, "step": 25311 }, { "epoch": 63.28, "grad_norm": 9.0625, "learning_rate": 6.276634000344144e-06, "loss": 1.2823, "step": 25312 }, { "epoch": 63.2825, "grad_norm": 6.75, "learning_rate": 6.275882542160457e-06, "loss": 1.2929, "step": 25313 }, { "epoch": 63.285, "grad_norm": 10.4375, "learning_rate": 6.275131108391933e-06, "loss": 1.427, "step": 25314 }, { "epoch": 63.2875, "grad_norm": 10.25, "learning_rate": 6.2743796990435025e-06, "loss": 1.438, "step": 25315 }, { "epoch": 63.29, "grad_norm": 8.6875, "learning_rate": 6.273628314120092e-06, "loss": 1.4406, "step": 25316 }, { "epoch": 63.2925, "grad_norm": 9.0625, "learning_rate": 6.272876953626623e-06, "loss": 1.363, "step": 25317 }, { "epoch": 63.295, "grad_norm": 7.875, "learning_rate": 6.272125617568026e-06, "loss": 1.2934, "step": 25318 }, { "epoch": 63.2975, "grad_norm": 7.09375, "learning_rate": 6.271374305949216e-06, "loss": 1.2469, "step": 25319 }, { "epoch": 63.3, "grad_norm": 9.4375, "learning_rate": 6.270623018775135e-06, "loss": 1.328, "step": 25320 }, { "epoch": 63.3025, "grad_norm": 9.125, "learning_rate": 6.269871756050699e-06, "loss": 1.2906, "step": 25321 }, { "epoch": 63.305, "grad_norm": 5.96875, "learning_rate": 6.269120517780833e-06, "loss": 1.211, "step": 25322 }, { "epoch": 63.3075, "grad_norm": 8.125, "learning_rate": 6.268369303970465e-06, "loss": 1.312, "step": 25323 }, { "epoch": 63.31, "grad_norm": 6.21875, "learning_rate": 6.267618114624515e-06, "loss": 1.3406, "step": 25324 }, { "epoch": 63.3125, "grad_norm": 10.25, "learning_rate": 6.266866949747914e-06, "loss": 1.3863, "step": 25325 }, { "epoch": 63.315, "grad_norm": 8.25, "learning_rate": 6.266115809345583e-06, "loss": 1.4272, "step": 25326 }, { "epoch": 63.3175, "grad_norm": 8.5625, "learning_rate": 6.26536469342245e-06, "loss": 1.422, "step": 25327 }, { "epoch": 63.32, "grad_norm": 8.875, "learning_rate": 6.264613601983435e-06, "loss": 1.2612, "step": 25328 }, { "epoch": 63.3225, "grad_norm": 9.75, "learning_rate": 6.26386253503346e-06, "loss": 1.3565, "step": 25329 }, { "epoch": 63.325, "grad_norm": 5.0, "learning_rate": 6.263111492577456e-06, "loss": 1.5846, "step": 25330 }, { "epoch": 63.3275, "grad_norm": 4.875, "learning_rate": 6.262360474620346e-06, "loss": 1.2586, "step": 25331 }, { "epoch": 63.33, "grad_norm": 7.0, "learning_rate": 6.261609481167048e-06, "loss": 1.2881, "step": 25332 }, { "epoch": 63.3325, "grad_norm": 5.53125, "learning_rate": 6.2608585122224875e-06, "loss": 1.3259, "step": 25333 }, { "epoch": 63.335, "grad_norm": 7.625, "learning_rate": 6.260107567791591e-06, "loss": 1.271, "step": 25334 }, { "epoch": 63.3375, "grad_norm": 6.21875, "learning_rate": 6.259356647879281e-06, "loss": 1.2633, "step": 25335 }, { "epoch": 63.34, "grad_norm": 7.375, "learning_rate": 6.258605752490477e-06, "loss": 1.2123, "step": 25336 }, { "epoch": 63.3425, "grad_norm": 6.625, "learning_rate": 6.257854881630108e-06, "loss": 1.2526, "step": 25337 }, { "epoch": 63.345, "grad_norm": 6.625, "learning_rate": 6.257104035303083e-06, "loss": 1.3276, "step": 25338 }, { "epoch": 63.3475, "grad_norm": 10.125, "learning_rate": 6.256353213514343e-06, "loss": 1.573, "step": 25339 }, { "epoch": 63.35, "grad_norm": 5.46875, "learning_rate": 6.255602416268799e-06, "loss": 1.1488, "step": 25340 }, { "epoch": 63.3525, "grad_norm": 16.5, "learning_rate": 6.254851643571378e-06, "loss": 1.3095, "step": 25341 }, { "epoch": 63.355, "grad_norm": 8.1875, "learning_rate": 6.2541008954269964e-06, "loss": 1.1908, "step": 25342 }, { "epoch": 63.3575, "grad_norm": 9.5625, "learning_rate": 6.253350171840578e-06, "loss": 1.1825, "step": 25343 }, { "epoch": 63.36, "grad_norm": 7.90625, "learning_rate": 6.2525994728170495e-06, "loss": 1.4531, "step": 25344 }, { "epoch": 63.3625, "grad_norm": 7.3125, "learning_rate": 6.251848798361328e-06, "loss": 1.3725, "step": 25345 }, { "epoch": 63.365, "grad_norm": 6.0, "learning_rate": 6.251098148478336e-06, "loss": 1.2645, "step": 25346 }, { "epoch": 63.3675, "grad_norm": 7.25, "learning_rate": 6.250347523172994e-06, "loss": 1.2024, "step": 25347 }, { "epoch": 63.37, "grad_norm": 7.75, "learning_rate": 6.249596922450221e-06, "loss": 1.4424, "step": 25348 }, { "epoch": 63.3725, "grad_norm": 4.875, "learning_rate": 6.2488463463149425e-06, "loss": 1.2889, "step": 25349 }, { "epoch": 63.375, "grad_norm": 7.15625, "learning_rate": 6.24809579477208e-06, "loss": 1.3864, "step": 25350 }, { "epoch": 63.3775, "grad_norm": 8.875, "learning_rate": 6.247345267826549e-06, "loss": 1.2859, "step": 25351 }, { "epoch": 63.38, "grad_norm": 9.6875, "learning_rate": 6.246594765483274e-06, "loss": 1.418, "step": 25352 }, { "epoch": 63.3825, "grad_norm": 10.125, "learning_rate": 6.245844287747168e-06, "loss": 1.5831, "step": 25353 }, { "epoch": 63.385, "grad_norm": 8.1875, "learning_rate": 6.245093834623163e-06, "loss": 1.2386, "step": 25354 }, { "epoch": 63.3875, "grad_norm": 6.03125, "learning_rate": 6.244343406116169e-06, "loss": 1.4778, "step": 25355 }, { "epoch": 63.39, "grad_norm": 6.0625, "learning_rate": 6.243593002231112e-06, "loss": 1.1793, "step": 25356 }, { "epoch": 63.3925, "grad_norm": 7.03125, "learning_rate": 6.2428426229729075e-06, "loss": 1.3991, "step": 25357 }, { "epoch": 63.395, "grad_norm": 6.1875, "learning_rate": 6.242092268346475e-06, "loss": 1.2395, "step": 25358 }, { "epoch": 63.3975, "grad_norm": 5.1875, "learning_rate": 6.241341938356736e-06, "loss": 1.3331, "step": 25359 }, { "epoch": 63.4, "grad_norm": 8.5, "learning_rate": 6.2405916330086106e-06, "loss": 1.2537, "step": 25360 }, { "epoch": 63.4025, "grad_norm": 8.4375, "learning_rate": 6.239841352307015e-06, "loss": 1.4258, "step": 25361 }, { "epoch": 63.405, "grad_norm": 7.5, "learning_rate": 6.239091096256869e-06, "loss": 1.386, "step": 25362 }, { "epoch": 63.4075, "grad_norm": 5.09375, "learning_rate": 6.2383408648630885e-06, "loss": 1.2157, "step": 25363 }, { "epoch": 63.41, "grad_norm": 7.96875, "learning_rate": 6.237590658130599e-06, "loss": 1.1913, "step": 25364 }, { "epoch": 63.4125, "grad_norm": 10.25, "learning_rate": 6.236840476064312e-06, "loss": 1.3184, "step": 25365 }, { "epoch": 63.415, "grad_norm": 5.4375, "learning_rate": 6.23609031866915e-06, "loss": 1.2828, "step": 25366 }, { "epoch": 63.4175, "grad_norm": 7.53125, "learning_rate": 6.235340185950029e-06, "loss": 1.4712, "step": 25367 }, { "epoch": 63.42, "grad_norm": 7.90625, "learning_rate": 6.234590077911865e-06, "loss": 1.1945, "step": 25368 }, { "epoch": 63.4225, "grad_norm": 6.8125, "learning_rate": 6.233839994559579e-06, "loss": 1.1967, "step": 25369 }, { "epoch": 63.425, "grad_norm": 9.9375, "learning_rate": 6.233089935898088e-06, "loss": 1.2661, "step": 25370 }, { "epoch": 63.4275, "grad_norm": 7.65625, "learning_rate": 6.232339901932309e-06, "loss": 1.1596, "step": 25371 }, { "epoch": 63.43, "grad_norm": 8.1875, "learning_rate": 6.231589892667158e-06, "loss": 1.1957, "step": 25372 }, { "epoch": 63.4325, "grad_norm": 10.25, "learning_rate": 6.230839908107551e-06, "loss": 1.3502, "step": 25373 }, { "epoch": 63.435, "grad_norm": 9.5625, "learning_rate": 6.230089948258409e-06, "loss": 1.3371, "step": 25374 }, { "epoch": 63.4375, "grad_norm": 7.3125, "learning_rate": 6.229340013124648e-06, "loss": 1.3714, "step": 25375 }, { "epoch": 63.44, "grad_norm": 5.90625, "learning_rate": 6.2285901027111806e-06, "loss": 1.3628, "step": 25376 }, { "epoch": 63.4425, "grad_norm": 7.375, "learning_rate": 6.2278402170229245e-06, "loss": 1.2999, "step": 25377 }, { "epoch": 63.445, "grad_norm": 9.0, "learning_rate": 6.227090356064797e-06, "loss": 1.2772, "step": 25378 }, { "epoch": 63.4475, "grad_norm": 8.9375, "learning_rate": 6.226340519841718e-06, "loss": 1.3184, "step": 25379 }, { "epoch": 63.45, "grad_norm": 7.90625, "learning_rate": 6.225590708358596e-06, "loss": 1.3511, "step": 25380 }, { "epoch": 63.4525, "grad_norm": 6.5, "learning_rate": 6.224840921620352e-06, "loss": 1.3616, "step": 25381 }, { "epoch": 63.455, "grad_norm": 8.1875, "learning_rate": 6.224091159631901e-06, "loss": 1.242, "step": 25382 }, { "epoch": 63.4575, "grad_norm": 11.5625, "learning_rate": 6.223341422398153e-06, "loss": 1.3757, "step": 25383 }, { "epoch": 63.46, "grad_norm": 5.28125, "learning_rate": 6.22259170992403e-06, "loss": 1.1544, "step": 25384 }, { "epoch": 63.4625, "grad_norm": 6.65625, "learning_rate": 6.221842022214444e-06, "loss": 1.3598, "step": 25385 }, { "epoch": 63.465, "grad_norm": 7.0625, "learning_rate": 6.221092359274314e-06, "loss": 1.3153, "step": 25386 }, { "epoch": 63.4675, "grad_norm": 5.28125, "learning_rate": 6.220342721108548e-06, "loss": 1.3015, "step": 25387 }, { "epoch": 63.47, "grad_norm": 8.875, "learning_rate": 6.219593107722062e-06, "loss": 1.3573, "step": 25388 }, { "epoch": 63.4725, "grad_norm": 7.125, "learning_rate": 6.2188435191197735e-06, "loss": 1.503, "step": 25389 }, { "epoch": 63.475, "grad_norm": 5.65625, "learning_rate": 6.218093955306599e-06, "loss": 1.3527, "step": 25390 }, { "epoch": 63.4775, "grad_norm": 7.15625, "learning_rate": 6.2173444162874465e-06, "loss": 1.4422, "step": 25391 }, { "epoch": 63.48, "grad_norm": 9.5, "learning_rate": 6.216594902067233e-06, "loss": 1.3756, "step": 25392 }, { "epoch": 63.4825, "grad_norm": 6.71875, "learning_rate": 6.215845412650869e-06, "loss": 1.2683, "step": 25393 }, { "epoch": 63.485, "grad_norm": 5.65625, "learning_rate": 6.215095948043276e-06, "loss": 1.33, "step": 25394 }, { "epoch": 63.4875, "grad_norm": 10.875, "learning_rate": 6.21434650824936e-06, "loss": 1.4187, "step": 25395 }, { "epoch": 63.49, "grad_norm": 6.96875, "learning_rate": 6.213597093274038e-06, "loss": 1.2805, "step": 25396 }, { "epoch": 63.4925, "grad_norm": 7.25, "learning_rate": 6.212847703122223e-06, "loss": 1.2226, "step": 25397 }, { "epoch": 63.495, "grad_norm": 7.59375, "learning_rate": 6.212098337798822e-06, "loss": 1.267, "step": 25398 }, { "epoch": 63.4975, "grad_norm": 9.375, "learning_rate": 6.211348997308755e-06, "loss": 1.2547, "step": 25399 }, { "epoch": 63.5, "grad_norm": 6.40625, "learning_rate": 6.210599681656933e-06, "loss": 1.3414, "step": 25400 }, { "epoch": 63.5025, "grad_norm": 8.875, "learning_rate": 6.20985039084827e-06, "loss": 1.3509, "step": 25401 }, { "epoch": 63.505, "grad_norm": 8.75, "learning_rate": 6.209101124887674e-06, "loss": 1.2208, "step": 25402 }, { "epoch": 63.5075, "grad_norm": 5.4375, "learning_rate": 6.208351883780057e-06, "loss": 1.3895, "step": 25403 }, { "epoch": 63.51, "grad_norm": 8.6875, "learning_rate": 6.207602667530336e-06, "loss": 1.3586, "step": 25404 }, { "epoch": 63.5125, "grad_norm": 9.375, "learning_rate": 6.206853476143422e-06, "loss": 1.2933, "step": 25405 }, { "epoch": 63.515, "grad_norm": 7.125, "learning_rate": 6.206104309624221e-06, "loss": 1.3537, "step": 25406 }, { "epoch": 63.5175, "grad_norm": 8.3125, "learning_rate": 6.2053551679776535e-06, "loss": 1.2724, "step": 25407 }, { "epoch": 63.52, "grad_norm": 7.21875, "learning_rate": 6.204606051208617e-06, "loss": 1.3984, "step": 25408 }, { "epoch": 63.5225, "grad_norm": 7.875, "learning_rate": 6.203856959322039e-06, "loss": 1.3739, "step": 25409 }, { "epoch": 63.525, "grad_norm": 5.9375, "learning_rate": 6.20310789232282e-06, "loss": 1.4186, "step": 25410 }, { "epoch": 63.5275, "grad_norm": 7.5625, "learning_rate": 6.202358850215876e-06, "loss": 1.2317, "step": 25411 }, { "epoch": 63.53, "grad_norm": 7.96875, "learning_rate": 6.201609833006113e-06, "loss": 1.3743, "step": 25412 }, { "epoch": 63.5325, "grad_norm": 7.53125, "learning_rate": 6.200860840698443e-06, "loss": 1.3422, "step": 25413 }, { "epoch": 63.535, "grad_norm": 8.4375, "learning_rate": 6.200111873297778e-06, "loss": 1.2285, "step": 25414 }, { "epoch": 63.5375, "grad_norm": 6.96875, "learning_rate": 6.1993629308090294e-06, "loss": 1.4202, "step": 25415 }, { "epoch": 63.54, "grad_norm": 8.625, "learning_rate": 6.198614013237105e-06, "loss": 1.3552, "step": 25416 }, { "epoch": 63.5425, "grad_norm": 6.4375, "learning_rate": 6.197865120586914e-06, "loss": 1.4103, "step": 25417 }, { "epoch": 63.545, "grad_norm": 10.625, "learning_rate": 6.197116252863364e-06, "loss": 1.3565, "step": 25418 }, { "epoch": 63.5475, "grad_norm": 7.84375, "learning_rate": 6.196367410071371e-06, "loss": 1.343, "step": 25419 }, { "epoch": 63.55, "grad_norm": 7.21875, "learning_rate": 6.1956185922158445e-06, "loss": 1.316, "step": 25420 }, { "epoch": 63.5525, "grad_norm": 10.125, "learning_rate": 6.194869799301686e-06, "loss": 1.4721, "step": 25421 }, { "epoch": 63.555, "grad_norm": 7.625, "learning_rate": 6.194121031333811e-06, "loss": 1.3844, "step": 25422 }, { "epoch": 63.5575, "grad_norm": 7.1875, "learning_rate": 6.193372288317121e-06, "loss": 1.2641, "step": 25423 }, { "epoch": 63.56, "grad_norm": 8.5, "learning_rate": 6.192623570256535e-06, "loss": 1.3168, "step": 25424 }, { "epoch": 63.5625, "grad_norm": 7.28125, "learning_rate": 6.191874877156956e-06, "loss": 1.1376, "step": 25425 }, { "epoch": 63.565, "grad_norm": 10.875, "learning_rate": 6.191126209023293e-06, "loss": 1.3225, "step": 25426 }, { "epoch": 63.5675, "grad_norm": 8.75, "learning_rate": 6.190377565860454e-06, "loss": 1.4319, "step": 25427 }, { "epoch": 63.57, "grad_norm": 5.59375, "learning_rate": 6.189628947673344e-06, "loss": 1.1474, "step": 25428 }, { "epoch": 63.5725, "grad_norm": 7.03125, "learning_rate": 6.188880354466876e-06, "loss": 1.3823, "step": 25429 }, { "epoch": 63.575, "grad_norm": 7.09375, "learning_rate": 6.188131786245959e-06, "loss": 1.2946, "step": 25430 }, { "epoch": 63.5775, "grad_norm": 7.0625, "learning_rate": 6.1873832430154945e-06, "loss": 1.2921, "step": 25431 }, { "epoch": 63.58, "grad_norm": 8.5625, "learning_rate": 6.186634724780394e-06, "loss": 1.3271, "step": 25432 }, { "epoch": 63.5825, "grad_norm": 7.4375, "learning_rate": 6.1858862315455604e-06, "loss": 1.393, "step": 25433 }, { "epoch": 63.585, "grad_norm": 7.25, "learning_rate": 6.1851377633159096e-06, "loss": 1.2798, "step": 25434 }, { "epoch": 63.5875, "grad_norm": 6.125, "learning_rate": 6.184389320096339e-06, "loss": 1.3719, "step": 25435 }, { "epoch": 63.59, "grad_norm": 6.71875, "learning_rate": 6.18364090189176e-06, "loss": 1.4098, "step": 25436 }, { "epoch": 63.5925, "grad_norm": 7.75, "learning_rate": 6.182892508707081e-06, "loss": 1.3, "step": 25437 }, { "epoch": 63.595, "grad_norm": 7.625, "learning_rate": 6.182144140547201e-06, "loss": 1.4473, "step": 25438 }, { "epoch": 63.5975, "grad_norm": 7.625, "learning_rate": 6.181395797417035e-06, "loss": 1.3586, "step": 25439 }, { "epoch": 63.6, "grad_norm": 10.5625, "learning_rate": 6.180647479321484e-06, "loss": 1.2538, "step": 25440 }, { "epoch": 63.6025, "grad_norm": 8.4375, "learning_rate": 6.179899186265457e-06, "loss": 1.2728, "step": 25441 }, { "epoch": 63.605, "grad_norm": 10.875, "learning_rate": 6.179150918253857e-06, "loss": 1.3376, "step": 25442 }, { "epoch": 63.6075, "grad_norm": 7.5625, "learning_rate": 6.178402675291588e-06, "loss": 1.1964, "step": 25443 }, { "epoch": 63.61, "grad_norm": 5.21875, "learning_rate": 6.177654457383561e-06, "loss": 1.2699, "step": 25444 }, { "epoch": 63.6125, "grad_norm": 7.5, "learning_rate": 6.1769062645346785e-06, "loss": 1.1452, "step": 25445 }, { "epoch": 63.615, "grad_norm": 47.75, "learning_rate": 6.176158096749846e-06, "loss": 1.4388, "step": 25446 }, { "epoch": 63.6175, "grad_norm": 6.625, "learning_rate": 6.1754099540339665e-06, "loss": 1.3047, "step": 25447 }, { "epoch": 63.62, "grad_norm": 6.28125, "learning_rate": 6.174661836391945e-06, "loss": 1.1175, "step": 25448 }, { "epoch": 63.6225, "grad_norm": 6.65625, "learning_rate": 6.173913743828691e-06, "loss": 1.418, "step": 25449 }, { "epoch": 63.625, "grad_norm": 7.25, "learning_rate": 6.173165676349103e-06, "loss": 1.2813, "step": 25450 }, { "epoch": 63.6275, "grad_norm": 13.375, "learning_rate": 6.172417633958089e-06, "loss": 1.1512, "step": 25451 }, { "epoch": 63.63, "grad_norm": 8.5625, "learning_rate": 6.171669616660553e-06, "loss": 1.2952, "step": 25452 }, { "epoch": 63.6325, "grad_norm": 8.0625, "learning_rate": 6.170921624461393e-06, "loss": 1.1905, "step": 25453 }, { "epoch": 63.635, "grad_norm": 7.96875, "learning_rate": 6.170173657365521e-06, "loss": 1.3831, "step": 25454 }, { "epoch": 63.6375, "grad_norm": 7.03125, "learning_rate": 6.169425715377837e-06, "loss": 1.4985, "step": 25455 }, { "epoch": 63.64, "grad_norm": 8.5625, "learning_rate": 6.168677798503246e-06, "loss": 1.256, "step": 25456 }, { "epoch": 63.6425, "grad_norm": 11.9375, "learning_rate": 6.167929906746648e-06, "loss": 1.3062, "step": 25457 }, { "epoch": 63.645, "grad_norm": 6.5625, "learning_rate": 6.167182040112947e-06, "loss": 1.2568, "step": 25458 }, { "epoch": 63.6475, "grad_norm": 8.4375, "learning_rate": 6.166434198607048e-06, "loss": 1.3591, "step": 25459 }, { "epoch": 63.65, "grad_norm": 9.25, "learning_rate": 6.165686382233856e-06, "loss": 1.431, "step": 25460 }, { "epoch": 63.6525, "grad_norm": 8.5625, "learning_rate": 6.164938590998269e-06, "loss": 1.2868, "step": 25461 }, { "epoch": 63.655, "grad_norm": 5.03125, "learning_rate": 6.164190824905189e-06, "loss": 1.243, "step": 25462 }, { "epoch": 63.6575, "grad_norm": 8.3125, "learning_rate": 6.1634430839595205e-06, "loss": 1.2981, "step": 25463 }, { "epoch": 63.66, "grad_norm": 5.5, "learning_rate": 6.16269536816617e-06, "loss": 1.436, "step": 25464 }, { "epoch": 63.6625, "grad_norm": 8.4375, "learning_rate": 6.161947677530031e-06, "loss": 1.4022, "step": 25465 }, { "epoch": 63.665, "grad_norm": 8.3125, "learning_rate": 6.161200012056011e-06, "loss": 1.3473, "step": 25466 }, { "epoch": 63.6675, "grad_norm": 9.375, "learning_rate": 6.160452371749012e-06, "loss": 1.4149, "step": 25467 }, { "epoch": 63.67, "grad_norm": 5.125, "learning_rate": 6.159704756613929e-06, "loss": 1.2531, "step": 25468 }, { "epoch": 63.6725, "grad_norm": 8.375, "learning_rate": 6.15895716665567e-06, "loss": 1.4085, "step": 25469 }, { "epoch": 63.675, "grad_norm": 8.0625, "learning_rate": 6.158209601879135e-06, "loss": 1.1527, "step": 25470 }, { "epoch": 63.6775, "grad_norm": 7.96875, "learning_rate": 6.157462062289226e-06, "loss": 1.4068, "step": 25471 }, { "epoch": 63.68, "grad_norm": 10.5625, "learning_rate": 6.156714547890838e-06, "loss": 1.2546, "step": 25472 }, { "epoch": 63.6825, "grad_norm": 5.84375, "learning_rate": 6.1559670586888755e-06, "loss": 1.4938, "step": 25473 }, { "epoch": 63.685, "grad_norm": 4.53125, "learning_rate": 6.15521959468824e-06, "loss": 1.3935, "step": 25474 }, { "epoch": 63.6875, "grad_norm": 7.09375, "learning_rate": 6.154472155893833e-06, "loss": 1.2628, "step": 25475 }, { "epoch": 63.69, "grad_norm": 7.25, "learning_rate": 6.153724742310551e-06, "loss": 1.5112, "step": 25476 }, { "epoch": 63.6925, "grad_norm": 14.0, "learning_rate": 6.152977353943299e-06, "loss": 1.309, "step": 25477 }, { "epoch": 63.695, "grad_norm": 9.75, "learning_rate": 6.152229990796966e-06, "loss": 1.3364, "step": 25478 }, { "epoch": 63.6975, "grad_norm": 7.4375, "learning_rate": 6.151482652876466e-06, "loss": 1.4606, "step": 25479 }, { "epoch": 63.7, "grad_norm": 5.78125, "learning_rate": 6.150735340186689e-06, "loss": 1.3476, "step": 25480 }, { "epoch": 63.7025, "grad_norm": 7.78125, "learning_rate": 6.149988052732541e-06, "loss": 1.3466, "step": 25481 }, { "epoch": 63.705, "grad_norm": 7.90625, "learning_rate": 6.149240790518913e-06, "loss": 1.4161, "step": 25482 }, { "epoch": 63.7075, "grad_norm": 8.4375, "learning_rate": 6.148493553550707e-06, "loss": 1.5241, "step": 25483 }, { "epoch": 63.71, "grad_norm": 6.03125, "learning_rate": 6.1477463418328255e-06, "loss": 1.3123, "step": 25484 }, { "epoch": 63.7125, "grad_norm": 8.5, "learning_rate": 6.146999155370164e-06, "loss": 1.2933, "step": 25485 }, { "epoch": 63.715, "grad_norm": 7.5, "learning_rate": 6.1462519941676245e-06, "loss": 1.3775, "step": 25486 }, { "epoch": 63.7175, "grad_norm": 6.84375, "learning_rate": 6.1455048582301e-06, "loss": 1.3156, "step": 25487 }, { "epoch": 63.72, "grad_norm": 7.4375, "learning_rate": 6.14475774756249e-06, "loss": 1.2117, "step": 25488 }, { "epoch": 63.7225, "grad_norm": 10.625, "learning_rate": 6.1440106621696945e-06, "loss": 1.3782, "step": 25489 }, { "epoch": 63.725, "grad_norm": 9.4375, "learning_rate": 6.143263602056615e-06, "loss": 1.2392, "step": 25490 }, { "epoch": 63.7275, "grad_norm": 14.5, "learning_rate": 6.142516567228143e-06, "loss": 1.2458, "step": 25491 }, { "epoch": 63.73, "grad_norm": 7.5, "learning_rate": 6.141769557689179e-06, "loss": 1.3449, "step": 25492 }, { "epoch": 63.7325, "grad_norm": 7.28125, "learning_rate": 6.1410225734446124e-06, "loss": 1.2943, "step": 25493 }, { "epoch": 63.735, "grad_norm": 8.6875, "learning_rate": 6.140275614499356e-06, "loss": 1.4018, "step": 25494 }, { "epoch": 63.7375, "grad_norm": 7.40625, "learning_rate": 6.1395286808582935e-06, "loss": 1.3879, "step": 25495 }, { "epoch": 63.74, "grad_norm": 6.96875, "learning_rate": 6.13878177252633e-06, "loss": 1.1986, "step": 25496 }, { "epoch": 63.7425, "grad_norm": 6.75, "learning_rate": 6.138034889508356e-06, "loss": 1.2446, "step": 25497 }, { "epoch": 63.745, "grad_norm": 6.90625, "learning_rate": 6.13728803180927e-06, "loss": 1.392, "step": 25498 }, { "epoch": 63.7475, "grad_norm": 7.09375, "learning_rate": 6.136541199433969e-06, "loss": 1.4564, "step": 25499 }, { "epoch": 63.75, "grad_norm": 4.84375, "learning_rate": 6.135794392387353e-06, "loss": 1.3051, "step": 25500 }, { "epoch": 63.7525, "grad_norm": 7.78125, "learning_rate": 6.135047610674312e-06, "loss": 1.2365, "step": 25501 }, { "epoch": 63.755, "grad_norm": 9.5, "learning_rate": 6.1343008542997436e-06, "loss": 1.2664, "step": 25502 }, { "epoch": 63.7575, "grad_norm": 8.4375, "learning_rate": 6.133554123268542e-06, "loss": 1.4141, "step": 25503 }, { "epoch": 63.76, "grad_norm": 9.25, "learning_rate": 6.13280741758561e-06, "loss": 1.362, "step": 25504 }, { "epoch": 63.7625, "grad_norm": 6.6875, "learning_rate": 6.132060737255835e-06, "loss": 1.1704, "step": 25505 }, { "epoch": 63.765, "grad_norm": 9.375, "learning_rate": 6.131314082284115e-06, "loss": 1.3903, "step": 25506 }, { "epoch": 63.7675, "grad_norm": 8.125, "learning_rate": 6.1305674526753465e-06, "loss": 1.3234, "step": 25507 }, { "epoch": 63.77, "grad_norm": 5.5, "learning_rate": 6.12982084843442e-06, "loss": 1.3833, "step": 25508 }, { "epoch": 63.7725, "grad_norm": 8.8125, "learning_rate": 6.129074269566236e-06, "loss": 1.172, "step": 25509 }, { "epoch": 63.775, "grad_norm": 5.875, "learning_rate": 6.128327716075684e-06, "loss": 1.5441, "step": 25510 }, { "epoch": 63.7775, "grad_norm": 6.71875, "learning_rate": 6.127581187967664e-06, "loss": 1.3242, "step": 25511 }, { "epoch": 63.78, "grad_norm": 4.75, "learning_rate": 6.126834685247065e-06, "loss": 1.336, "step": 25512 }, { "epoch": 63.7825, "grad_norm": 10.125, "learning_rate": 6.126088207918779e-06, "loss": 1.2705, "step": 25513 }, { "epoch": 63.785, "grad_norm": 7.5625, "learning_rate": 6.125341755987707e-06, "loss": 1.2272, "step": 25514 }, { "epoch": 63.7875, "grad_norm": 7.96875, "learning_rate": 6.124595329458741e-06, "loss": 1.4629, "step": 25515 }, { "epoch": 63.79, "grad_norm": 9.25, "learning_rate": 6.123848928336771e-06, "loss": 1.4808, "step": 25516 }, { "epoch": 63.7925, "grad_norm": 8.1875, "learning_rate": 6.123102552626693e-06, "loss": 1.3318, "step": 25517 }, { "epoch": 63.795, "grad_norm": 10.0, "learning_rate": 6.1223562023333974e-06, "loss": 1.3113, "step": 25518 }, { "epoch": 63.7975, "grad_norm": 8.8125, "learning_rate": 6.121609877461783e-06, "loss": 1.3872, "step": 25519 }, { "epoch": 63.8, "grad_norm": 6.96875, "learning_rate": 6.120863578016736e-06, "loss": 1.3, "step": 25520 }, { "epoch": 63.8025, "grad_norm": 9.4375, "learning_rate": 6.120117304003153e-06, "loss": 1.4644, "step": 25521 }, { "epoch": 63.805, "grad_norm": 5.5, "learning_rate": 6.119371055425929e-06, "loss": 1.3871, "step": 25522 }, { "epoch": 63.8075, "grad_norm": 7.40625, "learning_rate": 6.1186248322899475e-06, "loss": 1.3904, "step": 25523 }, { "epoch": 63.81, "grad_norm": 11.75, "learning_rate": 6.117878634600107e-06, "loss": 1.3956, "step": 25524 }, { "epoch": 63.8125, "grad_norm": 8.875, "learning_rate": 6.1171324623613016e-06, "loss": 1.451, "step": 25525 }, { "epoch": 63.815, "grad_norm": 6.0, "learning_rate": 6.116386315578422e-06, "loss": 1.3469, "step": 25526 }, { "epoch": 63.8175, "grad_norm": 6.5625, "learning_rate": 6.115640194256355e-06, "loss": 1.1725, "step": 25527 }, { "epoch": 63.82, "grad_norm": 6.4375, "learning_rate": 6.114894098399994e-06, "loss": 1.2918, "step": 25528 }, { "epoch": 63.8225, "grad_norm": 6.21875, "learning_rate": 6.114148028014233e-06, "loss": 1.322, "step": 25529 }, { "epoch": 63.825, "grad_norm": 7.125, "learning_rate": 6.1134019831039655e-06, "loss": 1.2993, "step": 25530 }, { "epoch": 63.8275, "grad_norm": 8.625, "learning_rate": 6.112655963674077e-06, "loss": 1.2902, "step": 25531 }, { "epoch": 63.83, "grad_norm": 7.53125, "learning_rate": 6.11190996972946e-06, "loss": 1.3653, "step": 25532 }, { "epoch": 63.8325, "grad_norm": 9.5625, "learning_rate": 6.111164001275003e-06, "loss": 1.3567, "step": 25533 }, { "epoch": 63.835, "grad_norm": 8.75, "learning_rate": 6.110418058315604e-06, "loss": 1.3957, "step": 25534 }, { "epoch": 63.8375, "grad_norm": 9.4375, "learning_rate": 6.109672140856146e-06, "loss": 1.4971, "step": 25535 }, { "epoch": 63.84, "grad_norm": 7.71875, "learning_rate": 6.108926248901521e-06, "loss": 1.2827, "step": 25536 }, { "epoch": 63.8425, "grad_norm": 6.90625, "learning_rate": 6.108180382456623e-06, "loss": 1.2449, "step": 25537 }, { "epoch": 63.845, "grad_norm": 7.53125, "learning_rate": 6.107434541526334e-06, "loss": 1.116, "step": 25538 }, { "epoch": 63.8475, "grad_norm": 10.0, "learning_rate": 6.106688726115552e-06, "loss": 1.1793, "step": 25539 }, { "epoch": 63.85, "grad_norm": 8.875, "learning_rate": 6.1059429362291615e-06, "loss": 1.2535, "step": 25540 }, { "epoch": 63.8525, "grad_norm": 7.1875, "learning_rate": 6.105197171872055e-06, "loss": 1.2164, "step": 25541 }, { "epoch": 63.855, "grad_norm": 10.75, "learning_rate": 6.104451433049119e-06, "loss": 1.2758, "step": 25542 }, { "epoch": 63.8575, "grad_norm": 6.5625, "learning_rate": 6.103705719765241e-06, "loss": 1.2568, "step": 25543 }, { "epoch": 63.86, "grad_norm": 6.84375, "learning_rate": 6.102960032025313e-06, "loss": 1.3983, "step": 25544 }, { "epoch": 63.8625, "grad_norm": 9.5625, "learning_rate": 6.102214369834227e-06, "loss": 1.3826, "step": 25545 }, { "epoch": 63.865, "grad_norm": 9.0, "learning_rate": 6.101468733196866e-06, "loss": 1.3212, "step": 25546 }, { "epoch": 63.8675, "grad_norm": 8.3125, "learning_rate": 6.100723122118121e-06, "loss": 1.3482, "step": 25547 }, { "epoch": 63.87, "grad_norm": 8.5625, "learning_rate": 6.099977536602872e-06, "loss": 1.1813, "step": 25548 }, { "epoch": 63.8725, "grad_norm": 8.4375, "learning_rate": 6.099231976656023e-06, "loss": 1.4923, "step": 25549 }, { "epoch": 63.875, "grad_norm": 7.75, "learning_rate": 6.09848644228245e-06, "loss": 1.4099, "step": 25550 }, { "epoch": 63.8775, "grad_norm": 7.09375, "learning_rate": 6.097740933487044e-06, "loss": 1.2476, "step": 25551 }, { "epoch": 63.88, "grad_norm": 9.9375, "learning_rate": 6.0969954502746916e-06, "loss": 1.3181, "step": 25552 }, { "epoch": 63.8825, "grad_norm": 8.375, "learning_rate": 6.0962499926502785e-06, "loss": 1.1127, "step": 25553 }, { "epoch": 63.885, "grad_norm": 8.25, "learning_rate": 6.0955045606186965e-06, "loss": 1.2739, "step": 25554 }, { "epoch": 63.8875, "grad_norm": 6.3125, "learning_rate": 6.0947591541848306e-06, "loss": 1.2583, "step": 25555 }, { "epoch": 63.89, "grad_norm": 6.5, "learning_rate": 6.094013773353569e-06, "loss": 1.3104, "step": 25556 }, { "epoch": 63.8925, "grad_norm": 6.1875, "learning_rate": 6.093268418129795e-06, "loss": 1.4069, "step": 25557 }, { "epoch": 63.895, "grad_norm": 7.53125, "learning_rate": 6.092523088518395e-06, "loss": 1.4552, "step": 25558 }, { "epoch": 63.8975, "grad_norm": 10.0625, "learning_rate": 6.09177778452426e-06, "loss": 1.3678, "step": 25559 }, { "epoch": 63.9, "grad_norm": 7.25, "learning_rate": 6.091032506152274e-06, "loss": 1.2226, "step": 25560 }, { "epoch": 63.9025, "grad_norm": 8.6875, "learning_rate": 6.0902872534073205e-06, "loss": 1.4769, "step": 25561 }, { "epoch": 63.905, "grad_norm": 11.625, "learning_rate": 6.08954202629429e-06, "loss": 1.307, "step": 25562 }, { "epoch": 63.9075, "grad_norm": 7.65625, "learning_rate": 6.08879682481806e-06, "loss": 1.2183, "step": 25563 }, { "epoch": 63.91, "grad_norm": 9.0, "learning_rate": 6.0880516489835265e-06, "loss": 1.3607, "step": 25564 }, { "epoch": 63.9125, "grad_norm": 6.125, "learning_rate": 6.087306498795568e-06, "loss": 1.3634, "step": 25565 }, { "epoch": 63.915, "grad_norm": 5.34375, "learning_rate": 6.086561374259074e-06, "loss": 1.3062, "step": 25566 }, { "epoch": 63.9175, "grad_norm": 9.0625, "learning_rate": 6.085816275378924e-06, "loss": 1.1947, "step": 25567 }, { "epoch": 63.92, "grad_norm": 15.9375, "learning_rate": 6.0850712021600044e-06, "loss": 1.2124, "step": 25568 }, { "epoch": 63.9225, "grad_norm": 8.125, "learning_rate": 6.084326154607204e-06, "loss": 1.2276, "step": 25569 }, { "epoch": 63.925, "grad_norm": 6.59375, "learning_rate": 6.083581132725407e-06, "loss": 1.308, "step": 25570 }, { "epoch": 63.9275, "grad_norm": 6.25, "learning_rate": 6.082836136519492e-06, "loss": 1.3043, "step": 25571 }, { "epoch": 63.93, "grad_norm": 7.09375, "learning_rate": 6.082091165994345e-06, "loss": 1.2327, "step": 25572 }, { "epoch": 63.9325, "grad_norm": 7.375, "learning_rate": 6.081346221154853e-06, "loss": 1.5523, "step": 25573 }, { "epoch": 63.935, "grad_norm": 9.4375, "learning_rate": 6.080601302005902e-06, "loss": 1.3604, "step": 25574 }, { "epoch": 63.9375, "grad_norm": 6.875, "learning_rate": 6.079856408552368e-06, "loss": 1.4388, "step": 25575 }, { "epoch": 63.94, "grad_norm": 12.5625, "learning_rate": 6.07911154079914e-06, "loss": 1.3338, "step": 25576 }, { "epoch": 63.9425, "grad_norm": 6.28125, "learning_rate": 6.078366698751097e-06, "loss": 1.3427, "step": 25577 }, { "epoch": 63.945, "grad_norm": 7.3125, "learning_rate": 6.07762188241313e-06, "loss": 1.355, "step": 25578 }, { "epoch": 63.9475, "grad_norm": 6.40625, "learning_rate": 6.076877091790113e-06, "loss": 1.1842, "step": 25579 }, { "epoch": 63.95, "grad_norm": 7.90625, "learning_rate": 6.076132326886934e-06, "loss": 1.3402, "step": 25580 }, { "epoch": 63.9525, "grad_norm": 7.96875, "learning_rate": 6.0753875877084764e-06, "loss": 1.1794, "step": 25581 }, { "epoch": 63.955, "grad_norm": 6.5625, "learning_rate": 6.074642874259615e-06, "loss": 1.1531, "step": 25582 }, { "epoch": 63.9575, "grad_norm": 9.1875, "learning_rate": 6.073898186545244e-06, "loss": 1.4138, "step": 25583 }, { "epoch": 63.96, "grad_norm": 9.5625, "learning_rate": 6.073153524570236e-06, "loss": 1.2766, "step": 25584 }, { "epoch": 63.9625, "grad_norm": 7.53125, "learning_rate": 6.07240888833948e-06, "loss": 1.3057, "step": 25585 }, { "epoch": 63.965, "grad_norm": 7.125, "learning_rate": 6.071664277857852e-06, "loss": 1.2392, "step": 25586 }, { "epoch": 63.9675, "grad_norm": 7.5625, "learning_rate": 6.070919693130233e-06, "loss": 1.2379, "step": 25587 }, { "epoch": 63.97, "grad_norm": 6.78125, "learning_rate": 6.0701751341615086e-06, "loss": 1.3182, "step": 25588 }, { "epoch": 63.9725, "grad_norm": 4.53125, "learning_rate": 6.069430600956563e-06, "loss": 1.2326, "step": 25589 }, { "epoch": 63.975, "grad_norm": 6.375, "learning_rate": 6.06868609352027e-06, "loss": 1.3383, "step": 25590 }, { "epoch": 63.9775, "grad_norm": 6.375, "learning_rate": 6.067941611857513e-06, "loss": 1.2731, "step": 25591 }, { "epoch": 63.98, "grad_norm": 8.375, "learning_rate": 6.067197155973172e-06, "loss": 1.2942, "step": 25592 }, { "epoch": 63.9825, "grad_norm": 5.75, "learning_rate": 6.066452725872133e-06, "loss": 1.3186, "step": 25593 }, { "epoch": 63.985, "grad_norm": 6.34375, "learning_rate": 6.065708321559271e-06, "loss": 1.3241, "step": 25594 }, { "epoch": 63.9875, "grad_norm": 11.1875, "learning_rate": 6.064963943039467e-06, "loss": 1.3841, "step": 25595 }, { "epoch": 63.99, "grad_norm": 8.0625, "learning_rate": 6.064219590317604e-06, "loss": 1.2515, "step": 25596 }, { "epoch": 63.9925, "grad_norm": 8.3125, "learning_rate": 6.063475263398556e-06, "loss": 1.4207, "step": 25597 }, { "epoch": 63.995, "grad_norm": 7.34375, "learning_rate": 6.062730962287208e-06, "loss": 1.255, "step": 25598 }, { "epoch": 63.9975, "grad_norm": 9.75, "learning_rate": 6.061986686988439e-06, "loss": 1.3277, "step": 25599 }, { "epoch": 64.0, "grad_norm": 7.46875, "learning_rate": 6.061242437507131e-06, "loss": 1.2917, "step": 25600 }, { "epoch": 64.0025, "grad_norm": 8.9375, "learning_rate": 6.060498213848156e-06, "loss": 1.4916, "step": 25601 }, { "epoch": 64.005, "grad_norm": 6.03125, "learning_rate": 6.059754016016395e-06, "loss": 1.199, "step": 25602 }, { "epoch": 64.0075, "grad_norm": 4.5, "learning_rate": 6.059009844016732e-06, "loss": 1.5325, "step": 25603 }, { "epoch": 64.01, "grad_norm": 10.0625, "learning_rate": 6.058265697854044e-06, "loss": 1.2912, "step": 25604 }, { "epoch": 64.0125, "grad_norm": 8.625, "learning_rate": 6.057521577533207e-06, "loss": 1.1326, "step": 25605 }, { "epoch": 64.015, "grad_norm": 6.625, "learning_rate": 6.056777483059101e-06, "loss": 1.3533, "step": 25606 }, { "epoch": 64.0175, "grad_norm": 10.6875, "learning_rate": 6.056033414436602e-06, "loss": 1.457, "step": 25607 }, { "epoch": 64.02, "grad_norm": 6.125, "learning_rate": 6.0552893716705925e-06, "loss": 1.2181, "step": 25608 }, { "epoch": 64.0225, "grad_norm": 8.5625, "learning_rate": 6.054545354765948e-06, "loss": 1.5348, "step": 25609 }, { "epoch": 64.025, "grad_norm": 5.96875, "learning_rate": 6.053801363727545e-06, "loss": 1.5078, "step": 25610 }, { "epoch": 64.0275, "grad_norm": 7.28125, "learning_rate": 6.053057398560264e-06, "loss": 1.339, "step": 25611 }, { "epoch": 64.03, "grad_norm": 6.78125, "learning_rate": 6.052313459268979e-06, "loss": 1.2733, "step": 25612 }, { "epoch": 64.0325, "grad_norm": 7.46875, "learning_rate": 6.0515695458585684e-06, "loss": 1.36, "step": 25613 }, { "epoch": 64.035, "grad_norm": 8.375, "learning_rate": 6.050825658333912e-06, "loss": 1.3917, "step": 25614 }, { "epoch": 64.0375, "grad_norm": 7.84375, "learning_rate": 6.050081796699885e-06, "loss": 1.3825, "step": 25615 }, { "epoch": 64.04, "grad_norm": 7.3125, "learning_rate": 6.049337960961362e-06, "loss": 1.2747, "step": 25616 }, { "epoch": 64.0425, "grad_norm": 7.8125, "learning_rate": 6.048594151123219e-06, "loss": 1.1961, "step": 25617 }, { "epoch": 64.045, "grad_norm": 8.0, "learning_rate": 6.047850367190338e-06, "loss": 1.3508, "step": 25618 }, { "epoch": 64.0475, "grad_norm": 8.5625, "learning_rate": 6.047106609167592e-06, "loss": 1.4536, "step": 25619 }, { "epoch": 64.05, "grad_norm": 5.78125, "learning_rate": 6.0463628770598574e-06, "loss": 1.4172, "step": 25620 }, { "epoch": 64.0525, "grad_norm": 7.375, "learning_rate": 6.0456191708720095e-06, "loss": 1.2457, "step": 25621 }, { "epoch": 64.055, "grad_norm": 9.25, "learning_rate": 6.044875490608918e-06, "loss": 1.374, "step": 25622 }, { "epoch": 64.0575, "grad_norm": 7.75, "learning_rate": 6.0441318362754705e-06, "loss": 1.3362, "step": 25623 }, { "epoch": 64.06, "grad_norm": 7.34375, "learning_rate": 6.043388207876535e-06, "loss": 1.4502, "step": 25624 }, { "epoch": 64.0625, "grad_norm": 11.3125, "learning_rate": 6.042644605416987e-06, "loss": 1.2857, "step": 25625 }, { "epoch": 64.065, "grad_norm": 9.75, "learning_rate": 6.041901028901707e-06, "loss": 1.3753, "step": 25626 }, { "epoch": 64.0675, "grad_norm": 5.3125, "learning_rate": 6.04115747833556e-06, "loss": 1.2739, "step": 25627 }, { "epoch": 64.07, "grad_norm": 6.53125, "learning_rate": 6.040413953723427e-06, "loss": 1.4485, "step": 25628 }, { "epoch": 64.0725, "grad_norm": 8.1875, "learning_rate": 6.0396704550701835e-06, "loss": 1.3678, "step": 25629 }, { "epoch": 64.075, "grad_norm": 9.75, "learning_rate": 6.038926982380704e-06, "loss": 1.2796, "step": 25630 }, { "epoch": 64.0775, "grad_norm": 8.25, "learning_rate": 6.038183535659858e-06, "loss": 1.3118, "step": 25631 }, { "epoch": 64.08, "grad_norm": 6.28125, "learning_rate": 6.037440114912521e-06, "loss": 1.4515, "step": 25632 }, { "epoch": 64.0825, "grad_norm": 6.375, "learning_rate": 6.036696720143568e-06, "loss": 1.1601, "step": 25633 }, { "epoch": 64.085, "grad_norm": 5.9375, "learning_rate": 6.035953351357876e-06, "loss": 1.3933, "step": 25634 }, { "epoch": 64.0875, "grad_norm": 8.6875, "learning_rate": 6.0352100085603124e-06, "loss": 1.1839, "step": 25635 }, { "epoch": 64.09, "grad_norm": 10.1875, "learning_rate": 6.034466691755757e-06, "loss": 1.3155, "step": 25636 }, { "epoch": 64.0925, "grad_norm": 7.1875, "learning_rate": 6.033723400949072e-06, "loss": 1.2919, "step": 25637 }, { "epoch": 64.095, "grad_norm": 7.8125, "learning_rate": 6.032980136145144e-06, "loss": 1.3006, "step": 25638 }, { "epoch": 64.0975, "grad_norm": 8.0625, "learning_rate": 6.032236897348838e-06, "loss": 1.4186, "step": 25639 }, { "epoch": 64.1, "grad_norm": 7.40625, "learning_rate": 6.0314936845650296e-06, "loss": 1.2986, "step": 25640 }, { "epoch": 64.1025, "grad_norm": 6.5, "learning_rate": 6.030750497798588e-06, "loss": 1.3421, "step": 25641 }, { "epoch": 64.105, "grad_norm": 5.46875, "learning_rate": 6.030007337054384e-06, "loss": 1.3138, "step": 25642 }, { "epoch": 64.1075, "grad_norm": 7.75, "learning_rate": 6.029264202337297e-06, "loss": 1.2977, "step": 25643 }, { "epoch": 64.11, "grad_norm": 7.6875, "learning_rate": 6.028521093652195e-06, "loss": 1.3289, "step": 25644 }, { "epoch": 64.1125, "grad_norm": 10.25, "learning_rate": 6.02777801100395e-06, "loss": 1.3643, "step": 25645 }, { "epoch": 64.115, "grad_norm": 7.6875, "learning_rate": 6.0270349543974325e-06, "loss": 1.3618, "step": 25646 }, { "epoch": 64.1175, "grad_norm": 11.1875, "learning_rate": 6.026291923837513e-06, "loss": 1.2905, "step": 25647 }, { "epoch": 64.12, "grad_norm": 7.40625, "learning_rate": 6.0255489193290675e-06, "loss": 1.1988, "step": 25648 }, { "epoch": 64.1225, "grad_norm": 8.75, "learning_rate": 6.024805940876963e-06, "loss": 1.3254, "step": 25649 }, { "epoch": 64.125, "grad_norm": 4.71875, "learning_rate": 6.024062988486072e-06, "loss": 1.2837, "step": 25650 }, { "epoch": 64.1275, "grad_norm": 9.6875, "learning_rate": 6.023320062161266e-06, "loss": 1.136, "step": 25651 }, { "epoch": 64.13, "grad_norm": 8.0625, "learning_rate": 6.0225771619074105e-06, "loss": 1.3445, "step": 25652 }, { "epoch": 64.1325, "grad_norm": 6.3125, "learning_rate": 6.0218342877293835e-06, "loss": 1.4562, "step": 25653 }, { "epoch": 64.135, "grad_norm": 5.3125, "learning_rate": 6.021091439632051e-06, "loss": 1.4436, "step": 25654 }, { "epoch": 64.1375, "grad_norm": 6.5625, "learning_rate": 6.020348617620287e-06, "loss": 1.3768, "step": 25655 }, { "epoch": 64.14, "grad_norm": 6.3125, "learning_rate": 6.019605821698953e-06, "loss": 1.2916, "step": 25656 }, { "epoch": 64.1425, "grad_norm": 6.78125, "learning_rate": 6.018863051872924e-06, "loss": 1.3793, "step": 25657 }, { "epoch": 64.145, "grad_norm": 8.25, "learning_rate": 6.018120308147071e-06, "loss": 1.2893, "step": 25658 }, { "epoch": 64.1475, "grad_norm": 6.625, "learning_rate": 6.017377590526265e-06, "loss": 1.2149, "step": 25659 }, { "epoch": 64.15, "grad_norm": 6.53125, "learning_rate": 6.016634899015369e-06, "loss": 1.3568, "step": 25660 }, { "epoch": 64.1525, "grad_norm": 8.75, "learning_rate": 6.015892233619256e-06, "loss": 1.3803, "step": 25661 }, { "epoch": 64.155, "grad_norm": 10.1875, "learning_rate": 6.01514959434279e-06, "loss": 1.4158, "step": 25662 }, { "epoch": 64.1575, "grad_norm": 7.28125, "learning_rate": 6.01440698119085e-06, "loss": 1.2095, "step": 25663 }, { "epoch": 64.16, "grad_norm": 7.28125, "learning_rate": 6.013664394168297e-06, "loss": 1.2977, "step": 25664 }, { "epoch": 64.1625, "grad_norm": 8.625, "learning_rate": 6.012921833279999e-06, "loss": 1.3033, "step": 25665 }, { "epoch": 64.165, "grad_norm": 7.96875, "learning_rate": 6.012179298530828e-06, "loss": 1.3307, "step": 25666 }, { "epoch": 64.1675, "grad_norm": 7.84375, "learning_rate": 6.011436789925646e-06, "loss": 1.3296, "step": 25667 }, { "epoch": 64.17, "grad_norm": 7.125, "learning_rate": 6.010694307469327e-06, "loss": 1.3779, "step": 25668 }, { "epoch": 64.1725, "grad_norm": 6.1875, "learning_rate": 6.009951851166738e-06, "loss": 1.367, "step": 25669 }, { "epoch": 64.175, "grad_norm": 6.65625, "learning_rate": 6.009209421022745e-06, "loss": 1.3588, "step": 25670 }, { "epoch": 64.1775, "grad_norm": 5.03125, "learning_rate": 6.008467017042215e-06, "loss": 1.4004, "step": 25671 }, { "epoch": 64.18, "grad_norm": 8.0, "learning_rate": 6.00772463923001e-06, "loss": 1.3151, "step": 25672 }, { "epoch": 64.1825, "grad_norm": 10.5625, "learning_rate": 6.006982287591008e-06, "loss": 1.3894, "step": 25673 }, { "epoch": 64.185, "grad_norm": 6.0625, "learning_rate": 6.006239962130071e-06, "loss": 1.3995, "step": 25674 }, { "epoch": 64.1875, "grad_norm": 7.1875, "learning_rate": 6.005497662852063e-06, "loss": 1.1678, "step": 25675 }, { "epoch": 64.19, "grad_norm": 7.125, "learning_rate": 6.004755389761854e-06, "loss": 1.2501, "step": 25676 }, { "epoch": 64.1925, "grad_norm": 7.8125, "learning_rate": 6.004013142864305e-06, "loss": 1.2415, "step": 25677 }, { "epoch": 64.195, "grad_norm": 6.71875, "learning_rate": 6.0032709221642905e-06, "loss": 1.3637, "step": 25678 }, { "epoch": 64.1975, "grad_norm": 8.5, "learning_rate": 6.00252872766667e-06, "loss": 1.2201, "step": 25679 }, { "epoch": 64.2, "grad_norm": 4.78125, "learning_rate": 6.00178655937631e-06, "loss": 1.4713, "step": 25680 }, { "epoch": 64.2025, "grad_norm": 8.5625, "learning_rate": 6.0010444172980805e-06, "loss": 1.3354, "step": 25681 }, { "epoch": 64.205, "grad_norm": 9.9375, "learning_rate": 6.00030230143684e-06, "loss": 1.4786, "step": 25682 }, { "epoch": 64.2075, "grad_norm": 6.96875, "learning_rate": 5.999560211797459e-06, "loss": 1.2236, "step": 25683 }, { "epoch": 64.21, "grad_norm": 8.0625, "learning_rate": 5.998818148384801e-06, "loss": 1.3495, "step": 25684 }, { "epoch": 64.2125, "grad_norm": 7.0625, "learning_rate": 5.998076111203735e-06, "loss": 1.3636, "step": 25685 }, { "epoch": 64.215, "grad_norm": 8.8125, "learning_rate": 5.9973341002591165e-06, "loss": 1.2786, "step": 25686 }, { "epoch": 64.2175, "grad_norm": 8.0625, "learning_rate": 5.996592115555815e-06, "loss": 1.2454, "step": 25687 }, { "epoch": 64.22, "grad_norm": 9.8125, "learning_rate": 5.995850157098698e-06, "loss": 1.4329, "step": 25688 }, { "epoch": 64.2225, "grad_norm": 7.21875, "learning_rate": 5.995108224892629e-06, "loss": 1.1789, "step": 25689 }, { "epoch": 64.225, "grad_norm": 7.09375, "learning_rate": 5.994366318942469e-06, "loss": 1.2907, "step": 25690 }, { "epoch": 64.2275, "grad_norm": 6.65625, "learning_rate": 5.993624439253083e-06, "loss": 1.313, "step": 25691 }, { "epoch": 64.23, "grad_norm": 7.0625, "learning_rate": 5.99288258582933e-06, "loss": 1.3761, "step": 25692 }, { "epoch": 64.2325, "grad_norm": 5.21875, "learning_rate": 5.9921407586760835e-06, "loss": 1.1351, "step": 25693 }, { "epoch": 64.235, "grad_norm": 6.25, "learning_rate": 5.991398957798203e-06, "loss": 1.4363, "step": 25694 }, { "epoch": 64.2375, "grad_norm": 10.1875, "learning_rate": 5.990657183200546e-06, "loss": 1.3439, "step": 25695 }, { "epoch": 64.24, "grad_norm": 7.5, "learning_rate": 5.989915434887985e-06, "loss": 1.2449, "step": 25696 }, { "epoch": 64.2425, "grad_norm": 11.875, "learning_rate": 5.989173712865373e-06, "loss": 1.211, "step": 25697 }, { "epoch": 64.245, "grad_norm": 5.21875, "learning_rate": 5.98843201713758e-06, "loss": 1.3153, "step": 25698 }, { "epoch": 64.2475, "grad_norm": 8.875, "learning_rate": 5.987690347709465e-06, "loss": 1.1514, "step": 25699 }, { "epoch": 64.25, "grad_norm": 5.875, "learning_rate": 5.986948704585895e-06, "loss": 1.3393, "step": 25700 }, { "epoch": 64.2525, "grad_norm": 4.5, "learning_rate": 5.986207087771727e-06, "loss": 1.2895, "step": 25701 }, { "epoch": 64.255, "grad_norm": 12.75, "learning_rate": 5.985465497271821e-06, "loss": 1.3937, "step": 25702 }, { "epoch": 64.2575, "grad_norm": 8.875, "learning_rate": 5.984723933091046e-06, "loss": 1.3596, "step": 25703 }, { "epoch": 64.26, "grad_norm": 6.84375, "learning_rate": 5.983982395234261e-06, "loss": 1.367, "step": 25704 }, { "epoch": 64.2625, "grad_norm": 5.75, "learning_rate": 5.983240883706325e-06, "loss": 1.2397, "step": 25705 }, { "epoch": 64.265, "grad_norm": 7.625, "learning_rate": 5.982499398512104e-06, "loss": 1.3693, "step": 25706 }, { "epoch": 64.2675, "grad_norm": 10.625, "learning_rate": 5.98175793965645e-06, "loss": 1.3067, "step": 25707 }, { "epoch": 64.27, "grad_norm": 8.6875, "learning_rate": 5.981016507144235e-06, "loss": 1.3941, "step": 25708 }, { "epoch": 64.2725, "grad_norm": 6.75, "learning_rate": 5.980275100980314e-06, "loss": 1.2321, "step": 25709 }, { "epoch": 64.275, "grad_norm": 7.4375, "learning_rate": 5.9795337211695505e-06, "loss": 1.3488, "step": 25710 }, { "epoch": 64.2775, "grad_norm": 7.9375, "learning_rate": 5.9787923677168e-06, "loss": 1.2955, "step": 25711 }, { "epoch": 64.28, "grad_norm": 7.21875, "learning_rate": 5.9780510406269245e-06, "loss": 1.0843, "step": 25712 }, { "epoch": 64.2825, "grad_norm": 8.125, "learning_rate": 5.977309739904789e-06, "loss": 1.2286, "step": 25713 }, { "epoch": 64.285, "grad_norm": 7.3125, "learning_rate": 5.9765684655552495e-06, "loss": 1.3339, "step": 25714 }, { "epoch": 64.2875, "grad_norm": 7.5625, "learning_rate": 5.975827217583166e-06, "loss": 1.2453, "step": 25715 }, { "epoch": 64.29, "grad_norm": 7.46875, "learning_rate": 5.975085995993396e-06, "loss": 1.3879, "step": 25716 }, { "epoch": 64.2925, "grad_norm": 5.6875, "learning_rate": 5.9743448007908e-06, "loss": 1.1498, "step": 25717 }, { "epoch": 64.295, "grad_norm": 8.125, "learning_rate": 5.9736036319802424e-06, "loss": 1.3476, "step": 25718 }, { "epoch": 64.2975, "grad_norm": 6.25, "learning_rate": 5.972862489566578e-06, "loss": 1.2562, "step": 25719 }, { "epoch": 64.3, "grad_norm": 7.75, "learning_rate": 5.972121373554665e-06, "loss": 1.3696, "step": 25720 }, { "epoch": 64.3025, "grad_norm": 10.0625, "learning_rate": 5.971380283949364e-06, "loss": 1.3615, "step": 25721 }, { "epoch": 64.305, "grad_norm": 6.59375, "learning_rate": 5.970639220755528e-06, "loss": 1.4162, "step": 25722 }, { "epoch": 64.3075, "grad_norm": 8.75, "learning_rate": 5.969898183978025e-06, "loss": 1.3513, "step": 25723 }, { "epoch": 64.31, "grad_norm": 6.5, "learning_rate": 5.969157173621706e-06, "loss": 1.4762, "step": 25724 }, { "epoch": 64.3125, "grad_norm": 9.6875, "learning_rate": 5.968416189691433e-06, "loss": 1.4462, "step": 25725 }, { "epoch": 64.315, "grad_norm": 9.1875, "learning_rate": 5.967675232192061e-06, "loss": 1.2642, "step": 25726 }, { "epoch": 64.3175, "grad_norm": 8.125, "learning_rate": 5.966934301128446e-06, "loss": 1.3177, "step": 25727 }, { "epoch": 64.32, "grad_norm": 6.875, "learning_rate": 5.966193396505452e-06, "loss": 1.2211, "step": 25728 }, { "epoch": 64.3225, "grad_norm": 7.25, "learning_rate": 5.9654525183279345e-06, "loss": 1.2428, "step": 25729 }, { "epoch": 64.325, "grad_norm": 6.8125, "learning_rate": 5.964711666600745e-06, "loss": 1.3276, "step": 25730 }, { "epoch": 64.3275, "grad_norm": 8.375, "learning_rate": 5.963970841328746e-06, "loss": 1.3457, "step": 25731 }, { "epoch": 64.33, "grad_norm": 8.125, "learning_rate": 5.96323004251679e-06, "loss": 1.3773, "step": 25732 }, { "epoch": 64.3325, "grad_norm": 7.75, "learning_rate": 5.96248927016974e-06, "loss": 1.3521, "step": 25733 }, { "epoch": 64.335, "grad_norm": 9.0, "learning_rate": 5.961748524292448e-06, "loss": 1.2271, "step": 25734 }, { "epoch": 64.3375, "grad_norm": 5.78125, "learning_rate": 5.961007804889769e-06, "loss": 1.3868, "step": 25735 }, { "epoch": 64.34, "grad_norm": 6.125, "learning_rate": 5.960267111966565e-06, "loss": 1.3822, "step": 25736 }, { "epoch": 64.3425, "grad_norm": 11.0625, "learning_rate": 5.959526445527683e-06, "loss": 1.4683, "step": 25737 }, { "epoch": 64.345, "grad_norm": 4.71875, "learning_rate": 5.958785805577987e-06, "loss": 1.2735, "step": 25738 }, { "epoch": 64.3475, "grad_norm": 6.9375, "learning_rate": 5.958045192122329e-06, "loss": 1.3599, "step": 25739 }, { "epoch": 64.35, "grad_norm": 7.8125, "learning_rate": 5.957304605165567e-06, "loss": 1.2812, "step": 25740 }, { "epoch": 64.3525, "grad_norm": 9.0, "learning_rate": 5.9565640447125515e-06, "loss": 1.2917, "step": 25741 }, { "epoch": 64.355, "grad_norm": 9.625, "learning_rate": 5.955823510768139e-06, "loss": 1.2096, "step": 25742 }, { "epoch": 64.3575, "grad_norm": 7.5625, "learning_rate": 5.955083003337189e-06, "loss": 1.2852, "step": 25743 }, { "epoch": 64.36, "grad_norm": 10.0, "learning_rate": 5.954342522424553e-06, "loss": 1.3507, "step": 25744 }, { "epoch": 64.3625, "grad_norm": 7.8125, "learning_rate": 5.9536020680350845e-06, "loss": 1.2592, "step": 25745 }, { "epoch": 64.365, "grad_norm": 11.0625, "learning_rate": 5.952861640173639e-06, "loss": 1.3428, "step": 25746 }, { "epoch": 64.3675, "grad_norm": 8.8125, "learning_rate": 5.952121238845068e-06, "loss": 1.3567, "step": 25747 }, { "epoch": 64.37, "grad_norm": 8.8125, "learning_rate": 5.9513808640542325e-06, "loss": 1.2852, "step": 25748 }, { "epoch": 64.3725, "grad_norm": 9.4375, "learning_rate": 5.950640515805981e-06, "loss": 1.3112, "step": 25749 }, { "epoch": 64.375, "grad_norm": 6.46875, "learning_rate": 5.949900194105167e-06, "loss": 1.2193, "step": 25750 }, { "epoch": 64.3775, "grad_norm": 9.3125, "learning_rate": 5.949159898956647e-06, "loss": 1.2995, "step": 25751 }, { "epoch": 64.38, "grad_norm": 7.46875, "learning_rate": 5.948419630365269e-06, "loss": 1.2801, "step": 25752 }, { "epoch": 64.3825, "grad_norm": 8.375, "learning_rate": 5.947679388335892e-06, "loss": 1.2913, "step": 25753 }, { "epoch": 64.385, "grad_norm": 7.90625, "learning_rate": 5.946939172873367e-06, "loss": 1.3447, "step": 25754 }, { "epoch": 64.3875, "grad_norm": 8.0, "learning_rate": 5.946198983982547e-06, "loss": 1.3313, "step": 25755 }, { "epoch": 64.39, "grad_norm": 5.65625, "learning_rate": 5.945458821668284e-06, "loss": 1.1049, "step": 25756 }, { "epoch": 64.3925, "grad_norm": 7.46875, "learning_rate": 5.944718685935429e-06, "loss": 1.2442, "step": 25757 }, { "epoch": 64.395, "grad_norm": 7.375, "learning_rate": 5.943978576788836e-06, "loss": 1.1941, "step": 25758 }, { "epoch": 64.3975, "grad_norm": 6.71875, "learning_rate": 5.943238494233361e-06, "loss": 1.28, "step": 25759 }, { "epoch": 64.4, "grad_norm": 6.875, "learning_rate": 5.942498438273849e-06, "loss": 1.2033, "step": 25760 }, { "epoch": 64.4025, "grad_norm": 7.6875, "learning_rate": 5.941758408915157e-06, "loss": 1.3582, "step": 25761 }, { "epoch": 64.405, "grad_norm": 6.1875, "learning_rate": 5.941018406162129e-06, "loss": 1.2758, "step": 25762 }, { "epoch": 64.4075, "grad_norm": 7.09375, "learning_rate": 5.940278430019628e-06, "loss": 1.1594, "step": 25763 }, { "epoch": 64.41, "grad_norm": 11.875, "learning_rate": 5.9395384804924974e-06, "loss": 1.3619, "step": 25764 }, { "epoch": 64.4125, "grad_norm": 7.34375, "learning_rate": 5.938798557585589e-06, "loss": 1.2779, "step": 25765 }, { "epoch": 64.415, "grad_norm": 5.8125, "learning_rate": 5.938058661303756e-06, "loss": 1.3537, "step": 25766 }, { "epoch": 64.4175, "grad_norm": 8.1875, "learning_rate": 5.937318791651846e-06, "loss": 1.2881, "step": 25767 }, { "epoch": 64.42, "grad_norm": 7.21875, "learning_rate": 5.936578948634713e-06, "loss": 1.3716, "step": 25768 }, { "epoch": 64.4225, "grad_norm": 7.46875, "learning_rate": 5.935839132257204e-06, "loss": 1.4743, "step": 25769 }, { "epoch": 64.425, "grad_norm": 6.84375, "learning_rate": 5.935099342524175e-06, "loss": 1.26, "step": 25770 }, { "epoch": 64.4275, "grad_norm": 13.25, "learning_rate": 5.934359579440469e-06, "loss": 1.309, "step": 25771 }, { "epoch": 64.43, "grad_norm": 8.5625, "learning_rate": 5.933619843010937e-06, "loss": 1.2372, "step": 25772 }, { "epoch": 64.4325, "grad_norm": 11.5625, "learning_rate": 5.932880133240433e-06, "loss": 1.3756, "step": 25773 }, { "epoch": 64.435, "grad_norm": 6.15625, "learning_rate": 5.932140450133805e-06, "loss": 1.3779, "step": 25774 }, { "epoch": 64.4375, "grad_norm": 6.25, "learning_rate": 5.9314007936959006e-06, "loss": 1.3095, "step": 25775 }, { "epoch": 64.44, "grad_norm": 8.25, "learning_rate": 5.930661163931572e-06, "loss": 1.4256, "step": 25776 }, { "epoch": 64.4425, "grad_norm": 9.6875, "learning_rate": 5.929921560845659e-06, "loss": 1.3523, "step": 25777 }, { "epoch": 64.445, "grad_norm": 8.1875, "learning_rate": 5.929181984443024e-06, "loss": 1.2547, "step": 25778 }, { "epoch": 64.4475, "grad_norm": 9.125, "learning_rate": 5.928442434728507e-06, "loss": 1.2601, "step": 25779 }, { "epoch": 64.45, "grad_norm": 7.15625, "learning_rate": 5.927702911706961e-06, "loss": 1.361, "step": 25780 }, { "epoch": 64.4525, "grad_norm": 8.0, "learning_rate": 5.926963415383229e-06, "loss": 1.3334, "step": 25781 }, { "epoch": 64.455, "grad_norm": 8.6875, "learning_rate": 5.9262239457621594e-06, "loss": 1.4556, "step": 25782 }, { "epoch": 64.4575, "grad_norm": 6.75, "learning_rate": 5.925484502848605e-06, "loss": 1.3541, "step": 25783 }, { "epoch": 64.46, "grad_norm": 9.375, "learning_rate": 5.924745086647413e-06, "loss": 1.2783, "step": 25784 }, { "epoch": 64.4625, "grad_norm": 7.5625, "learning_rate": 5.9240056971634276e-06, "loss": 1.5332, "step": 25785 }, { "epoch": 64.465, "grad_norm": 11.375, "learning_rate": 5.923266334401498e-06, "loss": 1.2771, "step": 25786 }, { "epoch": 64.4675, "grad_norm": 7.53125, "learning_rate": 5.922526998366468e-06, "loss": 1.3628, "step": 25787 }, { "epoch": 64.47, "grad_norm": 6.96875, "learning_rate": 5.921787689063192e-06, "loss": 1.5131, "step": 25788 }, { "epoch": 64.4725, "grad_norm": 5.875, "learning_rate": 5.921048406496512e-06, "loss": 1.103, "step": 25789 }, { "epoch": 64.475, "grad_norm": 5.46875, "learning_rate": 5.920309150671275e-06, "loss": 1.38, "step": 25790 }, { "epoch": 64.4775, "grad_norm": 9.75, "learning_rate": 5.91956992159233e-06, "loss": 1.2227, "step": 25791 }, { "epoch": 64.48, "grad_norm": 7.59375, "learning_rate": 5.918830719264514e-06, "loss": 1.3354, "step": 25792 }, { "epoch": 64.4825, "grad_norm": 8.5625, "learning_rate": 5.918091543692688e-06, "loss": 1.4292, "step": 25793 }, { "epoch": 64.485, "grad_norm": 9.0625, "learning_rate": 5.917352394881687e-06, "loss": 1.2212, "step": 25794 }, { "epoch": 64.4875, "grad_norm": 6.9375, "learning_rate": 5.916613272836364e-06, "loss": 1.2163, "step": 25795 }, { "epoch": 64.49, "grad_norm": 6.03125, "learning_rate": 5.915874177561558e-06, "loss": 1.3628, "step": 25796 }, { "epoch": 64.4925, "grad_norm": 8.8125, "learning_rate": 5.915135109062116e-06, "loss": 1.3138, "step": 25797 }, { "epoch": 64.495, "grad_norm": 9.375, "learning_rate": 5.914396067342885e-06, "loss": 1.4204, "step": 25798 }, { "epoch": 64.4975, "grad_norm": 6.46875, "learning_rate": 5.913657052408713e-06, "loss": 1.2782, "step": 25799 }, { "epoch": 64.5, "grad_norm": 7.25, "learning_rate": 5.912918064264441e-06, "loss": 1.3628, "step": 25800 }, { "epoch": 64.5025, "grad_norm": 8.4375, "learning_rate": 5.912179102914913e-06, "loss": 1.3246, "step": 25801 }, { "epoch": 64.505, "grad_norm": 6.28125, "learning_rate": 5.9114401683649745e-06, "loss": 1.339, "step": 25802 }, { "epoch": 64.5075, "grad_norm": 6.59375, "learning_rate": 5.910701260619473e-06, "loss": 1.3598, "step": 25803 }, { "epoch": 64.51, "grad_norm": 8.875, "learning_rate": 5.909962379683249e-06, "loss": 1.3939, "step": 25804 }, { "epoch": 64.5125, "grad_norm": 7.40625, "learning_rate": 5.90922352556115e-06, "loss": 1.2146, "step": 25805 }, { "epoch": 64.515, "grad_norm": 8.875, "learning_rate": 5.908484698258018e-06, "loss": 1.4647, "step": 25806 }, { "epoch": 64.5175, "grad_norm": 5.46875, "learning_rate": 5.907745897778693e-06, "loss": 1.1442, "step": 25807 }, { "epoch": 64.52, "grad_norm": 6.96875, "learning_rate": 5.9070071241280235e-06, "loss": 1.4539, "step": 25808 }, { "epoch": 64.5225, "grad_norm": 7.71875, "learning_rate": 5.906268377310852e-06, "loss": 1.4266, "step": 25809 }, { "epoch": 64.525, "grad_norm": 8.0625, "learning_rate": 5.905529657332023e-06, "loss": 1.4732, "step": 25810 }, { "epoch": 64.5275, "grad_norm": 6.625, "learning_rate": 5.9047909641963766e-06, "loss": 1.3231, "step": 25811 }, { "epoch": 64.53, "grad_norm": 11.25, "learning_rate": 5.904052297908754e-06, "loss": 1.285, "step": 25812 }, { "epoch": 64.5325, "grad_norm": 5.53125, "learning_rate": 5.903313658474001e-06, "loss": 1.3958, "step": 25813 }, { "epoch": 64.535, "grad_norm": 5.40625, "learning_rate": 5.902575045896965e-06, "loss": 1.2904, "step": 25814 }, { "epoch": 64.5375, "grad_norm": 7.5625, "learning_rate": 5.901836460182479e-06, "loss": 1.4085, "step": 25815 }, { "epoch": 64.54, "grad_norm": 8.8125, "learning_rate": 5.901097901335388e-06, "loss": 1.4097, "step": 25816 }, { "epoch": 64.5425, "grad_norm": 8.4375, "learning_rate": 5.900359369360535e-06, "loss": 1.3864, "step": 25817 }, { "epoch": 64.545, "grad_norm": 6.375, "learning_rate": 5.899620864262767e-06, "loss": 1.3669, "step": 25818 }, { "epoch": 64.5475, "grad_norm": 7.5, "learning_rate": 5.898882386046917e-06, "loss": 1.1707, "step": 25819 }, { "epoch": 64.55, "grad_norm": 7.8125, "learning_rate": 5.898143934717831e-06, "loss": 1.3136, "step": 25820 }, { "epoch": 64.5525, "grad_norm": 9.0625, "learning_rate": 5.897405510280345e-06, "loss": 1.4692, "step": 25821 }, { "epoch": 64.555, "grad_norm": 6.8125, "learning_rate": 5.89666711273931e-06, "loss": 1.2752, "step": 25822 }, { "epoch": 64.5575, "grad_norm": 5.78125, "learning_rate": 5.8959287420995594e-06, "loss": 1.4239, "step": 25823 }, { "epoch": 64.56, "grad_norm": 6.3125, "learning_rate": 5.895190398365935e-06, "loss": 1.3833, "step": 25824 }, { "epoch": 64.5625, "grad_norm": 8.5, "learning_rate": 5.89445208154328e-06, "loss": 1.4367, "step": 25825 }, { "epoch": 64.565, "grad_norm": 7.78125, "learning_rate": 5.893713791636428e-06, "loss": 1.1554, "step": 25826 }, { "epoch": 64.5675, "grad_norm": 8.25, "learning_rate": 5.8929755286502286e-06, "loss": 1.221, "step": 25827 }, { "epoch": 64.57, "grad_norm": 9.25, "learning_rate": 5.892237292589516e-06, "loss": 1.0728, "step": 25828 }, { "epoch": 64.5725, "grad_norm": 5.78125, "learning_rate": 5.891499083459134e-06, "loss": 1.3058, "step": 25829 }, { "epoch": 64.575, "grad_norm": 9.375, "learning_rate": 5.890760901263916e-06, "loss": 1.5075, "step": 25830 }, { "epoch": 64.5775, "grad_norm": 9.1875, "learning_rate": 5.890022746008704e-06, "loss": 1.2576, "step": 25831 }, { "epoch": 64.58, "grad_norm": 7.28125, "learning_rate": 5.889284617698339e-06, "loss": 1.3613, "step": 25832 }, { "epoch": 64.5825, "grad_norm": 6.625, "learning_rate": 5.888546516337663e-06, "loss": 1.0787, "step": 25833 }, { "epoch": 64.585, "grad_norm": 7.59375, "learning_rate": 5.8878084419315095e-06, "loss": 1.3311, "step": 25834 }, { "epoch": 64.5875, "grad_norm": 9.5, "learning_rate": 5.88707039448472e-06, "loss": 1.1792, "step": 25835 }, { "epoch": 64.59, "grad_norm": 8.8125, "learning_rate": 5.886332374002128e-06, "loss": 1.5329, "step": 25836 }, { "epoch": 64.5925, "grad_norm": 7.8125, "learning_rate": 5.885594380488582e-06, "loss": 1.4166, "step": 25837 }, { "epoch": 64.595, "grad_norm": 8.125, "learning_rate": 5.884856413948913e-06, "loss": 1.2292, "step": 25838 }, { "epoch": 64.5975, "grad_norm": 13.875, "learning_rate": 5.884118474387961e-06, "loss": 1.5775, "step": 25839 }, { "epoch": 64.6, "grad_norm": 6.78125, "learning_rate": 5.8833805618105635e-06, "loss": 1.2779, "step": 25840 }, { "epoch": 64.6025, "grad_norm": 6.625, "learning_rate": 5.882642676221556e-06, "loss": 1.2657, "step": 25841 }, { "epoch": 64.605, "grad_norm": 8.5625, "learning_rate": 5.88190481762578e-06, "loss": 1.3712, "step": 25842 }, { "epoch": 64.6075, "grad_norm": 6.53125, "learning_rate": 5.881166986028071e-06, "loss": 1.3224, "step": 25843 }, { "epoch": 64.61, "grad_norm": 5.46875, "learning_rate": 5.880429181433268e-06, "loss": 1.4141, "step": 25844 }, { "epoch": 64.6125, "grad_norm": 5.875, "learning_rate": 5.879691403846205e-06, "loss": 1.3317, "step": 25845 }, { "epoch": 64.615, "grad_norm": 5.59375, "learning_rate": 5.878953653271718e-06, "loss": 1.3543, "step": 25846 }, { "epoch": 64.6175, "grad_norm": 9.125, "learning_rate": 5.878215929714648e-06, "loss": 1.3931, "step": 25847 }, { "epoch": 64.62, "grad_norm": 7.15625, "learning_rate": 5.877478233179832e-06, "loss": 1.3028, "step": 25848 }, { "epoch": 64.6225, "grad_norm": 6.4375, "learning_rate": 5.876740563672101e-06, "loss": 1.4083, "step": 25849 }, { "epoch": 64.625, "grad_norm": 12.125, "learning_rate": 5.876002921196296e-06, "loss": 1.3533, "step": 25850 }, { "epoch": 64.6275, "grad_norm": 6.96875, "learning_rate": 5.875265305757245e-06, "loss": 1.3187, "step": 25851 }, { "epoch": 64.63, "grad_norm": 10.3125, "learning_rate": 5.874527717359796e-06, "loss": 1.3152, "step": 25852 }, { "epoch": 64.6325, "grad_norm": 7.9375, "learning_rate": 5.873790156008775e-06, "loss": 1.3959, "step": 25853 }, { "epoch": 64.635, "grad_norm": 9.3125, "learning_rate": 5.873052621709023e-06, "loss": 1.3451, "step": 25854 }, { "epoch": 64.6375, "grad_norm": 8.4375, "learning_rate": 5.872315114465373e-06, "loss": 1.3974, "step": 25855 }, { "epoch": 64.64, "grad_norm": 5.125, "learning_rate": 5.871577634282655e-06, "loss": 1.4241, "step": 25856 }, { "epoch": 64.6425, "grad_norm": 5.8125, "learning_rate": 5.8708401811657116e-06, "loss": 1.5054, "step": 25857 }, { "epoch": 64.645, "grad_norm": 8.9375, "learning_rate": 5.870102755119378e-06, "loss": 1.3762, "step": 25858 }, { "epoch": 64.6475, "grad_norm": 7.375, "learning_rate": 5.869365356148482e-06, "loss": 1.438, "step": 25859 }, { "epoch": 64.65, "grad_norm": 5.9375, "learning_rate": 5.868627984257862e-06, "loss": 1.2479, "step": 25860 }, { "epoch": 64.6525, "grad_norm": 7.875, "learning_rate": 5.867890639452349e-06, "loss": 1.1872, "step": 25861 }, { "epoch": 64.655, "grad_norm": 8.4375, "learning_rate": 5.867153321736783e-06, "loss": 1.2364, "step": 25862 }, { "epoch": 64.6575, "grad_norm": 10.25, "learning_rate": 5.866416031115996e-06, "loss": 1.3263, "step": 25863 }, { "epoch": 64.66, "grad_norm": 9.1875, "learning_rate": 5.865678767594817e-06, "loss": 1.3321, "step": 25864 }, { "epoch": 64.6625, "grad_norm": 6.4375, "learning_rate": 5.864941531178086e-06, "loss": 1.2431, "step": 25865 }, { "epoch": 64.665, "grad_norm": 6.90625, "learning_rate": 5.864204321870626e-06, "loss": 1.2661, "step": 25866 }, { "epoch": 64.6675, "grad_norm": 6.75, "learning_rate": 5.863467139677283e-06, "loss": 1.2866, "step": 25867 }, { "epoch": 64.67, "grad_norm": 6.65625, "learning_rate": 5.862729984602882e-06, "loss": 1.1691, "step": 25868 }, { "epoch": 64.6725, "grad_norm": 5.9375, "learning_rate": 5.861992856652259e-06, "loss": 1.19, "step": 25869 }, { "epoch": 64.675, "grad_norm": 7.5, "learning_rate": 5.861255755830244e-06, "loss": 1.2429, "step": 25870 }, { "epoch": 64.6775, "grad_norm": 7.71875, "learning_rate": 5.860518682141668e-06, "loss": 1.3847, "step": 25871 }, { "epoch": 64.68, "grad_norm": 7.5625, "learning_rate": 5.8597816355913685e-06, "loss": 1.3111, "step": 25872 }, { "epoch": 64.6825, "grad_norm": 8.125, "learning_rate": 5.859044616184177e-06, "loss": 1.3653, "step": 25873 }, { "epoch": 64.685, "grad_norm": 7.3125, "learning_rate": 5.8583076239249216e-06, "loss": 1.1836, "step": 25874 }, { "epoch": 64.6875, "grad_norm": 5.15625, "learning_rate": 5.857570658818434e-06, "loss": 1.2259, "step": 25875 }, { "epoch": 64.69, "grad_norm": 6.96875, "learning_rate": 5.856833720869547e-06, "loss": 1.3242, "step": 25876 }, { "epoch": 64.6925, "grad_norm": 6.8125, "learning_rate": 5.856096810083096e-06, "loss": 1.3805, "step": 25877 }, { "epoch": 64.695, "grad_norm": 5.3125, "learning_rate": 5.855359926463906e-06, "loss": 1.3706, "step": 25878 }, { "epoch": 64.6975, "grad_norm": 7.5625, "learning_rate": 5.854623070016812e-06, "loss": 1.3716, "step": 25879 }, { "epoch": 64.7, "grad_norm": 7.25, "learning_rate": 5.853886240746643e-06, "loss": 1.4534, "step": 25880 }, { "epoch": 64.7025, "grad_norm": 7.0, "learning_rate": 5.8531494386582275e-06, "loss": 1.1809, "step": 25881 }, { "epoch": 64.705, "grad_norm": 10.0, "learning_rate": 5.8524126637563995e-06, "loss": 1.2169, "step": 25882 }, { "epoch": 64.7075, "grad_norm": 10.5, "learning_rate": 5.851675916045988e-06, "loss": 1.1367, "step": 25883 }, { "epoch": 64.71, "grad_norm": 7.3125, "learning_rate": 5.850939195531826e-06, "loss": 1.334, "step": 25884 }, { "epoch": 64.7125, "grad_norm": 9.125, "learning_rate": 5.850202502218738e-06, "loss": 1.3605, "step": 25885 }, { "epoch": 64.715, "grad_norm": 6.71875, "learning_rate": 5.849465836111554e-06, "loss": 1.3076, "step": 25886 }, { "epoch": 64.7175, "grad_norm": 9.875, "learning_rate": 5.848729197215109e-06, "loss": 1.4312, "step": 25887 }, { "epoch": 64.72, "grad_norm": 8.0625, "learning_rate": 5.84799258553423e-06, "loss": 1.267, "step": 25888 }, { "epoch": 64.7225, "grad_norm": 6.84375, "learning_rate": 5.847256001073744e-06, "loss": 1.3535, "step": 25889 }, { "epoch": 64.725, "grad_norm": 11.9375, "learning_rate": 5.846519443838482e-06, "loss": 1.3064, "step": 25890 }, { "epoch": 64.7275, "grad_norm": 5.9375, "learning_rate": 5.84578291383327e-06, "loss": 1.3171, "step": 25891 }, { "epoch": 64.73, "grad_norm": 7.46875, "learning_rate": 5.845046411062942e-06, "loss": 1.3501, "step": 25892 }, { "epoch": 64.7325, "grad_norm": 8.0, "learning_rate": 5.844309935532323e-06, "loss": 1.3756, "step": 25893 }, { "epoch": 64.735, "grad_norm": 8.0625, "learning_rate": 5.843573487246241e-06, "loss": 1.3729, "step": 25894 }, { "epoch": 64.7375, "grad_norm": 5.90625, "learning_rate": 5.842837066209528e-06, "loss": 1.3363, "step": 25895 }, { "epoch": 64.74, "grad_norm": 12.0, "learning_rate": 5.842100672427002e-06, "loss": 1.3127, "step": 25896 }, { "epoch": 64.7425, "grad_norm": 8.3125, "learning_rate": 5.841364305903503e-06, "loss": 1.1255, "step": 25897 }, { "epoch": 64.745, "grad_norm": 8.75, "learning_rate": 5.8406279666438505e-06, "loss": 1.3171, "step": 25898 }, { "epoch": 64.7475, "grad_norm": 6.8125, "learning_rate": 5.839891654652878e-06, "loss": 1.2047, "step": 25899 }, { "epoch": 64.75, "grad_norm": 9.75, "learning_rate": 5.839155369935407e-06, "loss": 1.4165, "step": 25900 }, { "epoch": 64.7525, "grad_norm": 8.875, "learning_rate": 5.838419112496265e-06, "loss": 1.1049, "step": 25901 }, { "epoch": 64.755, "grad_norm": 6.90625, "learning_rate": 5.837682882340286e-06, "loss": 1.2388, "step": 25902 }, { "epoch": 64.7575, "grad_norm": 8.625, "learning_rate": 5.836946679472285e-06, "loss": 1.2868, "step": 25903 }, { "epoch": 64.76, "grad_norm": 8.0, "learning_rate": 5.836210503897099e-06, "loss": 1.2066, "step": 25904 }, { "epoch": 64.7625, "grad_norm": 9.5, "learning_rate": 5.835474355619553e-06, "loss": 1.2887, "step": 25905 }, { "epoch": 64.765, "grad_norm": 6.03125, "learning_rate": 5.834738234644464e-06, "loss": 1.3187, "step": 25906 }, { "epoch": 64.7675, "grad_norm": 8.1875, "learning_rate": 5.8340021409766655e-06, "loss": 1.3562, "step": 25907 }, { "epoch": 64.77, "grad_norm": 8.9375, "learning_rate": 5.833266074620987e-06, "loss": 1.4088, "step": 25908 }, { "epoch": 64.7725, "grad_norm": 8.125, "learning_rate": 5.832530035582248e-06, "loss": 1.3091, "step": 25909 }, { "epoch": 64.775, "grad_norm": 6.78125, "learning_rate": 5.831794023865271e-06, "loss": 1.4669, "step": 25910 }, { "epoch": 64.7775, "grad_norm": 8.0625, "learning_rate": 5.831058039474886e-06, "loss": 1.2875, "step": 25911 }, { "epoch": 64.78, "grad_norm": 9.25, "learning_rate": 5.830322082415922e-06, "loss": 1.4217, "step": 25912 }, { "epoch": 64.7825, "grad_norm": 7.0625, "learning_rate": 5.8295861526932005e-06, "loss": 1.3378, "step": 25913 }, { "epoch": 64.785, "grad_norm": 9.625, "learning_rate": 5.8288502503115395e-06, "loss": 1.3707, "step": 25914 }, { "epoch": 64.7875, "grad_norm": 12.0625, "learning_rate": 5.828114375275775e-06, "loss": 1.4106, "step": 25915 }, { "epoch": 64.79, "grad_norm": 7.5, "learning_rate": 5.827378527590722e-06, "loss": 1.4195, "step": 25916 }, { "epoch": 64.7925, "grad_norm": 11.5625, "learning_rate": 5.82664270726121e-06, "loss": 1.2459, "step": 25917 }, { "epoch": 64.795, "grad_norm": 7.46875, "learning_rate": 5.8259069142920585e-06, "loss": 1.3445, "step": 25918 }, { "epoch": 64.7975, "grad_norm": 6.03125, "learning_rate": 5.825171148688098e-06, "loss": 1.2091, "step": 25919 }, { "epoch": 64.8, "grad_norm": 9.25, "learning_rate": 5.82443541045415e-06, "loss": 1.4175, "step": 25920 }, { "epoch": 64.8025, "grad_norm": 6.40625, "learning_rate": 5.823699699595031e-06, "loss": 1.2063, "step": 25921 }, { "epoch": 64.805, "grad_norm": 7.0, "learning_rate": 5.82296401611557e-06, "loss": 1.3807, "step": 25922 }, { "epoch": 64.8075, "grad_norm": 6.625, "learning_rate": 5.822228360020593e-06, "loss": 1.3491, "step": 25923 }, { "epoch": 64.81, "grad_norm": 6.96875, "learning_rate": 5.82149273131492e-06, "loss": 1.4701, "step": 25924 }, { "epoch": 64.8125, "grad_norm": 10.6875, "learning_rate": 5.820757130003369e-06, "loss": 1.2932, "step": 25925 }, { "epoch": 64.815, "grad_norm": 10.0, "learning_rate": 5.820021556090769e-06, "loss": 1.3522, "step": 25926 }, { "epoch": 64.8175, "grad_norm": 12.125, "learning_rate": 5.819286009581944e-06, "loss": 1.3766, "step": 25927 }, { "epoch": 64.82, "grad_norm": 6.53125, "learning_rate": 5.818550490481711e-06, "loss": 1.175, "step": 25928 }, { "epoch": 64.8225, "grad_norm": 6.625, "learning_rate": 5.8178149987948904e-06, "loss": 1.2294, "step": 25929 }, { "epoch": 64.825, "grad_norm": 7.90625, "learning_rate": 5.817079534526311e-06, "loss": 1.3607, "step": 25930 }, { "epoch": 64.8275, "grad_norm": 7.03125, "learning_rate": 5.816344097680787e-06, "loss": 1.2706, "step": 25931 }, { "epoch": 64.83, "grad_norm": 13.875, "learning_rate": 5.815608688263148e-06, "loss": 1.3049, "step": 25932 }, { "epoch": 64.8325, "grad_norm": 7.0, "learning_rate": 5.814873306278206e-06, "loss": 1.2325, "step": 25933 }, { "epoch": 64.835, "grad_norm": 5.34375, "learning_rate": 5.814137951730791e-06, "loss": 1.2471, "step": 25934 }, { "epoch": 64.8375, "grad_norm": 6.21875, "learning_rate": 5.813402624625722e-06, "loss": 1.2765, "step": 25935 }, { "epoch": 64.84, "grad_norm": 6.8125, "learning_rate": 5.812667324967813e-06, "loss": 1.3177, "step": 25936 }, { "epoch": 64.8425, "grad_norm": 6.09375, "learning_rate": 5.811932052761889e-06, "loss": 1.2604, "step": 25937 }, { "epoch": 64.845, "grad_norm": 8.1875, "learning_rate": 5.811196808012776e-06, "loss": 1.2717, "step": 25938 }, { "epoch": 64.8475, "grad_norm": 10.875, "learning_rate": 5.810461590725287e-06, "loss": 1.6, "step": 25939 }, { "epoch": 64.85, "grad_norm": 6.0625, "learning_rate": 5.809726400904242e-06, "loss": 1.314, "step": 25940 }, { "epoch": 64.8525, "grad_norm": 7.03125, "learning_rate": 5.808991238554462e-06, "loss": 1.2614, "step": 25941 }, { "epoch": 64.855, "grad_norm": 6.40625, "learning_rate": 5.808256103680773e-06, "loss": 1.1963, "step": 25942 }, { "epoch": 64.8575, "grad_norm": 8.125, "learning_rate": 5.807520996287987e-06, "loss": 1.2893, "step": 25943 }, { "epoch": 64.86, "grad_norm": 6.5, "learning_rate": 5.806785916380924e-06, "loss": 1.1901, "step": 25944 }, { "epoch": 64.8625, "grad_norm": 7.90625, "learning_rate": 5.8060508639644075e-06, "loss": 1.4817, "step": 25945 }, { "epoch": 64.865, "grad_norm": 6.5, "learning_rate": 5.805315839043251e-06, "loss": 1.3582, "step": 25946 }, { "epoch": 64.8675, "grad_norm": 6.40625, "learning_rate": 5.804580841622278e-06, "loss": 1.389, "step": 25947 }, { "epoch": 64.87, "grad_norm": 7.03125, "learning_rate": 5.803845871706302e-06, "loss": 1.2537, "step": 25948 }, { "epoch": 64.8725, "grad_norm": 7.625, "learning_rate": 5.803110929300148e-06, "loss": 1.2333, "step": 25949 }, { "epoch": 64.875, "grad_norm": 7.375, "learning_rate": 5.802376014408632e-06, "loss": 1.3261, "step": 25950 }, { "epoch": 64.8775, "grad_norm": 7.65625, "learning_rate": 5.801641127036566e-06, "loss": 1.4826, "step": 25951 }, { "epoch": 64.88, "grad_norm": 5.625, "learning_rate": 5.800906267188773e-06, "loss": 1.3046, "step": 25952 }, { "epoch": 64.8825, "grad_norm": 7.21875, "learning_rate": 5.800171434870074e-06, "loss": 1.3848, "step": 25953 }, { "epoch": 64.885, "grad_norm": 7.34375, "learning_rate": 5.799436630085284e-06, "loss": 1.4972, "step": 25954 }, { "epoch": 64.8875, "grad_norm": 11.625, "learning_rate": 5.7987018528392195e-06, "loss": 1.2802, "step": 25955 }, { "epoch": 64.89, "grad_norm": 8.375, "learning_rate": 5.797967103136688e-06, "loss": 1.3633, "step": 25956 }, { "epoch": 64.8925, "grad_norm": 6.1875, "learning_rate": 5.797232380982525e-06, "loss": 1.2572, "step": 25957 }, { "epoch": 64.895, "grad_norm": 9.5, "learning_rate": 5.796497686381537e-06, "loss": 1.2003, "step": 25958 }, { "epoch": 64.8975, "grad_norm": 9.0625, "learning_rate": 5.795763019338544e-06, "loss": 1.3938, "step": 25959 }, { "epoch": 64.9, "grad_norm": 8.8125, "learning_rate": 5.795028379858355e-06, "loss": 1.2372, "step": 25960 }, { "epoch": 64.9025, "grad_norm": 6.375, "learning_rate": 5.794293767945793e-06, "loss": 1.2554, "step": 25961 }, { "epoch": 64.905, "grad_norm": 13.3125, "learning_rate": 5.793559183605675e-06, "loss": 1.4308, "step": 25962 }, { "epoch": 64.9075, "grad_norm": 6.71875, "learning_rate": 5.792824626842812e-06, "loss": 1.2648, "step": 25963 }, { "epoch": 64.91, "grad_norm": 5.96875, "learning_rate": 5.7920900976620265e-06, "loss": 1.4861, "step": 25964 }, { "epoch": 64.9125, "grad_norm": 6.9375, "learning_rate": 5.791355596068129e-06, "loss": 1.3421, "step": 25965 }, { "epoch": 64.915, "grad_norm": 7.6875, "learning_rate": 5.790621122065932e-06, "loss": 1.4487, "step": 25966 }, { "epoch": 64.9175, "grad_norm": 6.125, "learning_rate": 5.789886675660255e-06, "loss": 1.2565, "step": 25967 }, { "epoch": 64.92, "grad_norm": 7.96875, "learning_rate": 5.789152256855917e-06, "loss": 1.2645, "step": 25968 }, { "epoch": 64.9225, "grad_norm": 6.03125, "learning_rate": 5.788417865657727e-06, "loss": 1.3182, "step": 25969 }, { "epoch": 64.925, "grad_norm": 10.625, "learning_rate": 5.7876835020705e-06, "loss": 1.3333, "step": 25970 }, { "epoch": 64.9275, "grad_norm": 7.5, "learning_rate": 5.786949166099045e-06, "loss": 1.3466, "step": 25971 }, { "epoch": 64.93, "grad_norm": 6.21875, "learning_rate": 5.786214857748192e-06, "loss": 1.3276, "step": 25972 }, { "epoch": 64.9325, "grad_norm": 8.0625, "learning_rate": 5.785480577022744e-06, "loss": 1.4379, "step": 25973 }, { "epoch": 64.935, "grad_norm": 8.625, "learning_rate": 5.784746323927516e-06, "loss": 1.2624, "step": 25974 }, { "epoch": 64.9375, "grad_norm": 8.9375, "learning_rate": 5.78401209846732e-06, "loss": 1.3478, "step": 25975 }, { "epoch": 64.94, "grad_norm": 7.5, "learning_rate": 5.783277900646971e-06, "loss": 1.3706, "step": 25976 }, { "epoch": 64.9425, "grad_norm": 8.5, "learning_rate": 5.782543730471288e-06, "loss": 1.3002, "step": 25977 }, { "epoch": 64.945, "grad_norm": 6.5625, "learning_rate": 5.781809587945078e-06, "loss": 1.2206, "step": 25978 }, { "epoch": 64.9475, "grad_norm": 19.0, "learning_rate": 5.781075473073153e-06, "loss": 1.3305, "step": 25979 }, { "epoch": 64.95, "grad_norm": 5.96875, "learning_rate": 5.780341385860333e-06, "loss": 1.4283, "step": 25980 }, { "epoch": 64.9525, "grad_norm": 7.625, "learning_rate": 5.77960732631142e-06, "loss": 1.2956, "step": 25981 }, { "epoch": 64.955, "grad_norm": 7.78125, "learning_rate": 5.778873294431238e-06, "loss": 1.4721, "step": 25982 }, { "epoch": 64.9575, "grad_norm": 7.9375, "learning_rate": 5.778139290224589e-06, "loss": 1.422, "step": 25983 }, { "epoch": 64.96, "grad_norm": 11.0625, "learning_rate": 5.777405313696294e-06, "loss": 1.3265, "step": 25984 }, { "epoch": 64.9625, "grad_norm": 6.84375, "learning_rate": 5.7766713648511605e-06, "loss": 1.3236, "step": 25985 }, { "epoch": 64.965, "grad_norm": 14.125, "learning_rate": 5.775937443693997e-06, "loss": 1.4612, "step": 25986 }, { "epoch": 64.9675, "grad_norm": 5.5, "learning_rate": 5.77520355022962e-06, "loss": 1.4079, "step": 25987 }, { "epoch": 64.97, "grad_norm": 6.5, "learning_rate": 5.774469684462841e-06, "loss": 1.208, "step": 25988 }, { "epoch": 64.9725, "grad_norm": 8.0, "learning_rate": 5.7737358463984715e-06, "loss": 1.3862, "step": 25989 }, { "epoch": 64.975, "grad_norm": 8.5, "learning_rate": 5.7730020360413155e-06, "loss": 1.433, "step": 25990 }, { "epoch": 64.9775, "grad_norm": 6.625, "learning_rate": 5.772268253396189e-06, "loss": 1.2895, "step": 25991 }, { "epoch": 64.98, "grad_norm": 10.4375, "learning_rate": 5.771534498467908e-06, "loss": 1.3378, "step": 25992 }, { "epoch": 64.9825, "grad_norm": 7.125, "learning_rate": 5.770800771261276e-06, "loss": 1.119, "step": 25993 }, { "epoch": 64.985, "grad_norm": 9.0625, "learning_rate": 5.770067071781101e-06, "loss": 1.3233, "step": 25994 }, { "epoch": 64.9875, "grad_norm": 6.75, "learning_rate": 5.7693334000322025e-06, "loss": 1.4932, "step": 25995 }, { "epoch": 64.99, "grad_norm": 8.375, "learning_rate": 5.76859975601938e-06, "loss": 1.3188, "step": 25996 }, { "epoch": 64.9925, "grad_norm": 6.125, "learning_rate": 5.767866139747451e-06, "loss": 1.2281, "step": 25997 }, { "epoch": 64.995, "grad_norm": 6.59375, "learning_rate": 5.76713255122122e-06, "loss": 1.3852, "step": 25998 }, { "epoch": 64.9975, "grad_norm": 6.375, "learning_rate": 5.766398990445501e-06, "loss": 1.4633, "step": 25999 }, { "epoch": 65.0, "grad_norm": 7.5625, "learning_rate": 5.765665457425102e-06, "loss": 1.3582, "step": 26000 }, { "epoch": 65.0025, "grad_norm": 5.59375, "learning_rate": 5.764931952164827e-06, "loss": 1.3321, "step": 26001 }, { "epoch": 65.005, "grad_norm": 5.1875, "learning_rate": 5.764198474669488e-06, "loss": 1.4015, "step": 26002 }, { "epoch": 65.0075, "grad_norm": 6.6875, "learning_rate": 5.763465024943898e-06, "loss": 1.1471, "step": 26003 }, { "epoch": 65.01, "grad_norm": 10.5, "learning_rate": 5.762731602992862e-06, "loss": 1.3022, "step": 26004 }, { "epoch": 65.0125, "grad_norm": 7.84375, "learning_rate": 5.761998208821184e-06, "loss": 1.254, "step": 26005 }, { "epoch": 65.015, "grad_norm": 10.375, "learning_rate": 5.761264842433677e-06, "loss": 1.3071, "step": 26006 }, { "epoch": 65.0175, "grad_norm": 9.375, "learning_rate": 5.760531503835151e-06, "loss": 1.4042, "step": 26007 }, { "epoch": 65.02, "grad_norm": 9.3125, "learning_rate": 5.759798193030413e-06, "loss": 1.3642, "step": 26008 }, { "epoch": 65.0225, "grad_norm": 8.125, "learning_rate": 5.759064910024261e-06, "loss": 1.2434, "step": 26009 }, { "epoch": 65.025, "grad_norm": 5.5625, "learning_rate": 5.758331654821516e-06, "loss": 1.4098, "step": 26010 }, { "epoch": 65.0275, "grad_norm": 9.1875, "learning_rate": 5.757598427426975e-06, "loss": 1.2929, "step": 26011 }, { "epoch": 65.03, "grad_norm": 6.65625, "learning_rate": 5.756865227845452e-06, "loss": 1.3042, "step": 26012 }, { "epoch": 65.0325, "grad_norm": 4.46875, "learning_rate": 5.756132056081747e-06, "loss": 1.3342, "step": 26013 }, { "epoch": 65.035, "grad_norm": 8.6875, "learning_rate": 5.755398912140675e-06, "loss": 1.4123, "step": 26014 }, { "epoch": 65.0375, "grad_norm": 7.34375, "learning_rate": 5.754665796027037e-06, "loss": 1.3237, "step": 26015 }, { "epoch": 65.04, "grad_norm": 15.9375, "learning_rate": 5.753932707745635e-06, "loss": 1.3851, "step": 26016 }, { "epoch": 65.0425, "grad_norm": 8.875, "learning_rate": 5.753199647301283e-06, "loss": 1.2445, "step": 26017 }, { "epoch": 65.045, "grad_norm": 6.25, "learning_rate": 5.752466614698786e-06, "loss": 1.2602, "step": 26018 }, { "epoch": 65.0475, "grad_norm": 7.28125, "learning_rate": 5.751733609942949e-06, "loss": 1.2808, "step": 26019 }, { "epoch": 65.05, "grad_norm": 9.25, "learning_rate": 5.751000633038573e-06, "loss": 1.3789, "step": 26020 }, { "epoch": 65.0525, "grad_norm": 9.25, "learning_rate": 5.750267683990466e-06, "loss": 1.3861, "step": 26021 }, { "epoch": 65.055, "grad_norm": 8.1875, "learning_rate": 5.749534762803437e-06, "loss": 1.2957, "step": 26022 }, { "epoch": 65.0575, "grad_norm": 5.71875, "learning_rate": 5.74880186948229e-06, "loss": 1.4709, "step": 26023 }, { "epoch": 65.06, "grad_norm": 7.15625, "learning_rate": 5.748069004031822e-06, "loss": 1.508, "step": 26024 }, { "epoch": 65.0625, "grad_norm": 9.3125, "learning_rate": 5.747336166456849e-06, "loss": 1.3015, "step": 26025 }, { "epoch": 65.065, "grad_norm": 8.5625, "learning_rate": 5.746603356762166e-06, "loss": 1.3401, "step": 26026 }, { "epoch": 65.0675, "grad_norm": 6.5625, "learning_rate": 5.745870574952583e-06, "loss": 1.3467, "step": 26027 }, { "epoch": 65.07, "grad_norm": 6.625, "learning_rate": 5.7451378210329e-06, "loss": 1.2539, "step": 26028 }, { "epoch": 65.0725, "grad_norm": 8.3125, "learning_rate": 5.744405095007927e-06, "loss": 1.4678, "step": 26029 }, { "epoch": 65.075, "grad_norm": 8.0625, "learning_rate": 5.7436723968824635e-06, "loss": 1.3635, "step": 26030 }, { "epoch": 65.0775, "grad_norm": 7.625, "learning_rate": 5.74293972666131e-06, "loss": 1.2783, "step": 26031 }, { "epoch": 65.08, "grad_norm": 9.5625, "learning_rate": 5.742207084349274e-06, "loss": 1.381, "step": 26032 }, { "epoch": 65.0825, "grad_norm": 7.34375, "learning_rate": 5.74147446995116e-06, "loss": 1.2885, "step": 26033 }, { "epoch": 65.085, "grad_norm": 6.65625, "learning_rate": 5.74074188347177e-06, "loss": 1.4397, "step": 26034 }, { "epoch": 65.0875, "grad_norm": 8.375, "learning_rate": 5.7400093249159014e-06, "loss": 1.245, "step": 26035 }, { "epoch": 65.09, "grad_norm": 7.15625, "learning_rate": 5.739276794288362e-06, "loss": 1.2781, "step": 26036 }, { "epoch": 65.0925, "grad_norm": 8.3125, "learning_rate": 5.738544291593957e-06, "loss": 1.399, "step": 26037 }, { "epoch": 65.095, "grad_norm": 11.3125, "learning_rate": 5.737811816837484e-06, "loss": 1.3289, "step": 26038 }, { "epoch": 65.0975, "grad_norm": 6.34375, "learning_rate": 5.737079370023744e-06, "loss": 1.3342, "step": 26039 }, { "epoch": 65.1, "grad_norm": 6.9375, "learning_rate": 5.736346951157544e-06, "loss": 1.256, "step": 26040 }, { "epoch": 65.1025, "grad_norm": 9.0, "learning_rate": 5.735614560243679e-06, "loss": 1.225, "step": 26041 }, { "epoch": 65.105, "grad_norm": 7.0, "learning_rate": 5.734882197286958e-06, "loss": 1.2827, "step": 26042 }, { "epoch": 65.1075, "grad_norm": 7.6875, "learning_rate": 5.734149862292175e-06, "loss": 1.3996, "step": 26043 }, { "epoch": 65.11, "grad_norm": 14.9375, "learning_rate": 5.733417555264139e-06, "loss": 1.2115, "step": 26044 }, { "epoch": 65.1125, "grad_norm": 6.375, "learning_rate": 5.732685276207646e-06, "loss": 1.335, "step": 26045 }, { "epoch": 65.115, "grad_norm": 7.6875, "learning_rate": 5.731953025127494e-06, "loss": 1.3502, "step": 26046 }, { "epoch": 65.1175, "grad_norm": 5.8125, "learning_rate": 5.7312208020284876e-06, "loss": 1.3632, "step": 26047 }, { "epoch": 65.12, "grad_norm": 5.8125, "learning_rate": 5.73048860691543e-06, "loss": 1.5482, "step": 26048 }, { "epoch": 65.1225, "grad_norm": 9.5, "learning_rate": 5.729756439793119e-06, "loss": 1.4116, "step": 26049 }, { "epoch": 65.125, "grad_norm": 9.375, "learning_rate": 5.729024300666349e-06, "loss": 1.2082, "step": 26050 }, { "epoch": 65.1275, "grad_norm": 7.65625, "learning_rate": 5.728292189539926e-06, "loss": 1.3882, "step": 26051 }, { "epoch": 65.13, "grad_norm": 10.25, "learning_rate": 5.727560106418651e-06, "loss": 1.3305, "step": 26052 }, { "epoch": 65.1325, "grad_norm": 6.0625, "learning_rate": 5.726828051307322e-06, "loss": 1.4362, "step": 26053 }, { "epoch": 65.135, "grad_norm": 6.03125, "learning_rate": 5.726096024210733e-06, "loss": 1.2142, "step": 26054 }, { "epoch": 65.1375, "grad_norm": 8.875, "learning_rate": 5.725364025133692e-06, "loss": 1.2689, "step": 26055 }, { "epoch": 65.14, "grad_norm": 7.71875, "learning_rate": 5.724632054080989e-06, "loss": 1.2899, "step": 26056 }, { "epoch": 65.1425, "grad_norm": 10.3125, "learning_rate": 5.72390011105743e-06, "loss": 1.2897, "step": 26057 }, { "epoch": 65.145, "grad_norm": 6.78125, "learning_rate": 5.7231681960678075e-06, "loss": 1.2613, "step": 26058 }, { "epoch": 65.1475, "grad_norm": 6.1875, "learning_rate": 5.722436309116928e-06, "loss": 1.1662, "step": 26059 }, { "epoch": 65.15, "grad_norm": 7.03125, "learning_rate": 5.721704450209581e-06, "loss": 1.3516, "step": 26060 }, { "epoch": 65.1525, "grad_norm": 6.21875, "learning_rate": 5.720972619350571e-06, "loss": 1.2881, "step": 26061 }, { "epoch": 65.155, "grad_norm": 9.4375, "learning_rate": 5.720240816544691e-06, "loss": 1.337, "step": 26062 }, { "epoch": 65.1575, "grad_norm": 5.53125, "learning_rate": 5.719509041796742e-06, "loss": 1.2968, "step": 26063 }, { "epoch": 65.16, "grad_norm": 12.25, "learning_rate": 5.718777295111524e-06, "loss": 1.4292, "step": 26064 }, { "epoch": 65.1625, "grad_norm": 11.875, "learning_rate": 5.718045576493825e-06, "loss": 1.4994, "step": 26065 }, { "epoch": 65.165, "grad_norm": 9.0, "learning_rate": 5.717313885948448e-06, "loss": 1.3468, "step": 26066 }, { "epoch": 65.1675, "grad_norm": 10.0625, "learning_rate": 5.716582223480194e-06, "loss": 1.3942, "step": 26067 }, { "epoch": 65.17, "grad_norm": 8.3125, "learning_rate": 5.715850589093855e-06, "loss": 1.3995, "step": 26068 }, { "epoch": 65.1725, "grad_norm": 9.4375, "learning_rate": 5.715118982794223e-06, "loss": 1.3592, "step": 26069 }, { "epoch": 65.175, "grad_norm": 9.375, "learning_rate": 5.714387404586101e-06, "loss": 1.3098, "step": 26070 }, { "epoch": 65.1775, "grad_norm": 5.65625, "learning_rate": 5.713655854474288e-06, "loss": 1.3295, "step": 26071 }, { "epoch": 65.18, "grad_norm": 7.65625, "learning_rate": 5.712924332463575e-06, "loss": 1.237, "step": 26072 }, { "epoch": 65.1825, "grad_norm": 7.375, "learning_rate": 5.712192838558754e-06, "loss": 1.1773, "step": 26073 }, { "epoch": 65.185, "grad_norm": 7.5, "learning_rate": 5.7114613727646305e-06, "loss": 1.2271, "step": 26074 }, { "epoch": 65.1875, "grad_norm": 6.65625, "learning_rate": 5.71072993508599e-06, "loss": 1.1528, "step": 26075 }, { "epoch": 65.19, "grad_norm": 6.0, "learning_rate": 5.709998525527637e-06, "loss": 1.3031, "step": 26076 }, { "epoch": 65.1925, "grad_norm": 6.75, "learning_rate": 5.709267144094357e-06, "loss": 1.3549, "step": 26077 }, { "epoch": 65.195, "grad_norm": 8.5, "learning_rate": 5.708535790790954e-06, "loss": 1.3804, "step": 26078 }, { "epoch": 65.1975, "grad_norm": 6.5, "learning_rate": 5.70780446562222e-06, "loss": 1.3148, "step": 26079 }, { "epoch": 65.2, "grad_norm": 7.09375, "learning_rate": 5.707073168592943e-06, "loss": 1.4165, "step": 26080 }, { "epoch": 65.2025, "grad_norm": 5.78125, "learning_rate": 5.706341899707921e-06, "loss": 1.3694, "step": 26081 }, { "epoch": 65.205, "grad_norm": 9.0, "learning_rate": 5.705610658971956e-06, "loss": 1.3457, "step": 26082 }, { "epoch": 65.2075, "grad_norm": 8.875, "learning_rate": 5.704879446389835e-06, "loss": 1.3439, "step": 26083 }, { "epoch": 65.21, "grad_norm": 8.625, "learning_rate": 5.704148261966349e-06, "loss": 1.47, "step": 26084 }, { "epoch": 65.2125, "grad_norm": 7.75, "learning_rate": 5.703417105706293e-06, "loss": 1.3749, "step": 26085 }, { "epoch": 65.215, "grad_norm": 7.65625, "learning_rate": 5.702685977614468e-06, "loss": 1.4758, "step": 26086 }, { "epoch": 65.2175, "grad_norm": 12.0, "learning_rate": 5.7019548776956626e-06, "loss": 1.3712, "step": 26087 }, { "epoch": 65.22, "grad_norm": 6.625, "learning_rate": 5.701223805954664e-06, "loss": 1.3659, "step": 26088 }, { "epoch": 65.2225, "grad_norm": 13.0, "learning_rate": 5.700492762396275e-06, "loss": 1.4258, "step": 26089 }, { "epoch": 65.225, "grad_norm": 5.375, "learning_rate": 5.699761747025278e-06, "loss": 1.3392, "step": 26090 }, { "epoch": 65.2275, "grad_norm": 6.15625, "learning_rate": 5.699030759846476e-06, "loss": 1.3218, "step": 26091 }, { "epoch": 65.23, "grad_norm": 8.375, "learning_rate": 5.698299800864651e-06, "loss": 1.2643, "step": 26092 }, { "epoch": 65.2325, "grad_norm": 7.8125, "learning_rate": 5.697568870084605e-06, "loss": 1.3776, "step": 26093 }, { "epoch": 65.235, "grad_norm": 5.0625, "learning_rate": 5.696837967511126e-06, "loss": 1.2796, "step": 26094 }, { "epoch": 65.2375, "grad_norm": 6.125, "learning_rate": 5.6961070931490005e-06, "loss": 1.2256, "step": 26095 }, { "epoch": 65.24, "grad_norm": 7.96875, "learning_rate": 5.695376247003025e-06, "loss": 1.3034, "step": 26096 }, { "epoch": 65.2425, "grad_norm": 8.25, "learning_rate": 5.694645429077994e-06, "loss": 1.1725, "step": 26097 }, { "epoch": 65.245, "grad_norm": 9.9375, "learning_rate": 5.693914639378695e-06, "loss": 1.3449, "step": 26098 }, { "epoch": 65.2475, "grad_norm": 6.3125, "learning_rate": 5.69318387790992e-06, "loss": 1.4521, "step": 26099 }, { "epoch": 65.25, "grad_norm": 7.03125, "learning_rate": 5.692453144676451e-06, "loss": 1.3334, "step": 26100 }, { "epoch": 65.2525, "grad_norm": 12.875, "learning_rate": 5.691722439683095e-06, "loss": 1.4188, "step": 26101 }, { "epoch": 65.255, "grad_norm": 8.375, "learning_rate": 5.6909917629346346e-06, "loss": 1.3533, "step": 26102 }, { "epoch": 65.2575, "grad_norm": 5.5625, "learning_rate": 5.690261114435856e-06, "loss": 1.2058, "step": 26103 }, { "epoch": 65.26, "grad_norm": 5.375, "learning_rate": 5.689530494191555e-06, "loss": 1.2659, "step": 26104 }, { "epoch": 65.2625, "grad_norm": 7.25, "learning_rate": 5.688799902206516e-06, "loss": 1.387, "step": 26105 }, { "epoch": 65.265, "grad_norm": 6.03125, "learning_rate": 5.688069338485538e-06, "loss": 1.27, "step": 26106 }, { "epoch": 65.2675, "grad_norm": 6.125, "learning_rate": 5.6873388030334e-06, "loss": 1.3647, "step": 26107 }, { "epoch": 65.27, "grad_norm": 6.3125, "learning_rate": 5.686608295854901e-06, "loss": 1.1789, "step": 26108 }, { "epoch": 65.2725, "grad_norm": 7.0625, "learning_rate": 5.685877816954825e-06, "loss": 1.2536, "step": 26109 }, { "epoch": 65.275, "grad_norm": 7.25, "learning_rate": 5.685147366337956e-06, "loss": 1.2867, "step": 26110 }, { "epoch": 65.2775, "grad_norm": 7.90625, "learning_rate": 5.68441694400909e-06, "loss": 1.2921, "step": 26111 }, { "epoch": 65.28, "grad_norm": 6.53125, "learning_rate": 5.683686549973018e-06, "loss": 1.2393, "step": 26112 }, { "epoch": 65.2825, "grad_norm": 6.375, "learning_rate": 5.682956184234523e-06, "loss": 1.2358, "step": 26113 }, { "epoch": 65.285, "grad_norm": 9.125, "learning_rate": 5.682225846798395e-06, "loss": 1.4198, "step": 26114 }, { "epoch": 65.2875, "grad_norm": 7.25, "learning_rate": 5.681495537669413e-06, "loss": 1.2512, "step": 26115 }, { "epoch": 65.29, "grad_norm": 7.125, "learning_rate": 5.680765256852381e-06, "loss": 1.2945, "step": 26116 }, { "epoch": 65.2925, "grad_norm": 6.5, "learning_rate": 5.6800350043520804e-06, "loss": 1.2621, "step": 26117 }, { "epoch": 65.295, "grad_norm": 6.53125, "learning_rate": 5.679304780173297e-06, "loss": 1.2469, "step": 26118 }, { "epoch": 65.2975, "grad_norm": 7.6875, "learning_rate": 5.678574584320815e-06, "loss": 1.1769, "step": 26119 }, { "epoch": 65.3, "grad_norm": 6.25, "learning_rate": 5.677844416799424e-06, "loss": 1.2759, "step": 26120 }, { "epoch": 65.3025, "grad_norm": 18.375, "learning_rate": 5.677114277613918e-06, "loss": 1.4076, "step": 26121 }, { "epoch": 65.305, "grad_norm": 8.375, "learning_rate": 5.676384166769075e-06, "loss": 1.2586, "step": 26122 }, { "epoch": 65.3075, "grad_norm": 6.875, "learning_rate": 5.675654084269681e-06, "loss": 1.3921, "step": 26123 }, { "epoch": 65.31, "grad_norm": 10.0625, "learning_rate": 5.674924030120533e-06, "loss": 1.376, "step": 26124 }, { "epoch": 65.3125, "grad_norm": 10.625, "learning_rate": 5.674194004326403e-06, "loss": 1.3625, "step": 26125 }, { "epoch": 65.315, "grad_norm": 7.375, "learning_rate": 5.673464006892088e-06, "loss": 1.2447, "step": 26126 }, { "epoch": 65.3175, "grad_norm": 6.25, "learning_rate": 5.672734037822367e-06, "loss": 1.4488, "step": 26127 }, { "epoch": 65.32, "grad_norm": 12.1875, "learning_rate": 5.672004097122033e-06, "loss": 1.2676, "step": 26128 }, { "epoch": 65.3225, "grad_norm": 9.4375, "learning_rate": 5.671274184795865e-06, "loss": 1.38, "step": 26129 }, { "epoch": 65.325, "grad_norm": 7.75, "learning_rate": 5.670544300848643e-06, "loss": 1.4517, "step": 26130 }, { "epoch": 65.3275, "grad_norm": 6.84375, "learning_rate": 5.6698144452851665e-06, "loss": 1.2311, "step": 26131 }, { "epoch": 65.33, "grad_norm": 7.5625, "learning_rate": 5.669084618110215e-06, "loss": 1.1779, "step": 26132 }, { "epoch": 65.3325, "grad_norm": 7.09375, "learning_rate": 5.6683548193285695e-06, "loss": 1.4127, "step": 26133 }, { "epoch": 65.335, "grad_norm": 9.25, "learning_rate": 5.6676250489450135e-06, "loss": 1.3375, "step": 26134 }, { "epoch": 65.3375, "grad_norm": 8.625, "learning_rate": 5.666895306964335e-06, "loss": 1.2853, "step": 26135 }, { "epoch": 65.34, "grad_norm": 8.25, "learning_rate": 5.666165593391322e-06, "loss": 1.3948, "step": 26136 }, { "epoch": 65.3425, "grad_norm": 6.9375, "learning_rate": 5.665435908230752e-06, "loss": 1.2456, "step": 26137 }, { "epoch": 65.345, "grad_norm": 11.25, "learning_rate": 5.664706251487407e-06, "loss": 1.2402, "step": 26138 }, { "epoch": 65.3475, "grad_norm": 7.84375, "learning_rate": 5.6639766231660785e-06, "loss": 1.2633, "step": 26139 }, { "epoch": 65.35, "grad_norm": 8.125, "learning_rate": 5.663247023271543e-06, "loss": 1.4635, "step": 26140 }, { "epoch": 65.3525, "grad_norm": 7.875, "learning_rate": 5.662517451808589e-06, "loss": 1.2064, "step": 26141 }, { "epoch": 65.355, "grad_norm": 6.9375, "learning_rate": 5.661787908781993e-06, "loss": 1.2488, "step": 26142 }, { "epoch": 65.3575, "grad_norm": 6.15625, "learning_rate": 5.661058394196547e-06, "loss": 1.4105, "step": 26143 }, { "epoch": 65.36, "grad_norm": 7.625, "learning_rate": 5.6603289080570274e-06, "loss": 1.6245, "step": 26144 }, { "epoch": 65.3625, "grad_norm": 7.71875, "learning_rate": 5.659599450368214e-06, "loss": 1.3277, "step": 26145 }, { "epoch": 65.365, "grad_norm": 6.96875, "learning_rate": 5.658870021134895e-06, "loss": 1.3358, "step": 26146 }, { "epoch": 65.3675, "grad_norm": 9.5, "learning_rate": 5.658140620361851e-06, "loss": 1.2946, "step": 26147 }, { "epoch": 65.37, "grad_norm": 8.6875, "learning_rate": 5.657411248053866e-06, "loss": 1.3206, "step": 26148 }, { "epoch": 65.3725, "grad_norm": 10.0, "learning_rate": 5.656681904215713e-06, "loss": 1.6196, "step": 26149 }, { "epoch": 65.375, "grad_norm": 8.3125, "learning_rate": 5.655952588852181e-06, "loss": 1.2124, "step": 26150 }, { "epoch": 65.3775, "grad_norm": 8.3125, "learning_rate": 5.655223301968054e-06, "loss": 1.1786, "step": 26151 }, { "epoch": 65.38, "grad_norm": 6.75, "learning_rate": 5.654494043568109e-06, "loss": 1.1952, "step": 26152 }, { "epoch": 65.3825, "grad_norm": 9.375, "learning_rate": 5.653764813657123e-06, "loss": 1.2416, "step": 26153 }, { "epoch": 65.385, "grad_norm": 7.65625, "learning_rate": 5.653035612239886e-06, "loss": 1.2582, "step": 26154 }, { "epoch": 65.3875, "grad_norm": 8.875, "learning_rate": 5.652306439321167e-06, "loss": 1.4397, "step": 26155 }, { "epoch": 65.39, "grad_norm": 6.625, "learning_rate": 5.651577294905759e-06, "loss": 1.2243, "step": 26156 }, { "epoch": 65.3925, "grad_norm": 5.65625, "learning_rate": 5.650848178998432e-06, "loss": 1.2506, "step": 26157 }, { "epoch": 65.395, "grad_norm": 8.9375, "learning_rate": 5.650119091603974e-06, "loss": 1.3087, "step": 26158 }, { "epoch": 65.3975, "grad_norm": 6.34375, "learning_rate": 5.649390032727161e-06, "loss": 1.2772, "step": 26159 }, { "epoch": 65.4, "grad_norm": 5.09375, "learning_rate": 5.648661002372769e-06, "loss": 1.2206, "step": 26160 }, { "epoch": 65.4025, "grad_norm": 6.1875, "learning_rate": 5.647932000545582e-06, "loss": 1.2618, "step": 26161 }, { "epoch": 65.405, "grad_norm": 7.75, "learning_rate": 5.647203027250381e-06, "loss": 1.3184, "step": 26162 }, { "epoch": 65.4075, "grad_norm": 7.75, "learning_rate": 5.646474082491944e-06, "loss": 1.4629, "step": 26163 }, { "epoch": 65.41, "grad_norm": 7.4375, "learning_rate": 5.645745166275045e-06, "loss": 1.3277, "step": 26164 }, { "epoch": 65.4125, "grad_norm": 7.34375, "learning_rate": 5.645016278604466e-06, "loss": 1.1854, "step": 26165 }, { "epoch": 65.415, "grad_norm": 8.625, "learning_rate": 5.64428741948499e-06, "loss": 1.4191, "step": 26166 }, { "epoch": 65.4175, "grad_norm": 8.1875, "learning_rate": 5.643558588921392e-06, "loss": 1.4087, "step": 26167 }, { "epoch": 65.42, "grad_norm": 8.875, "learning_rate": 5.6428297869184444e-06, "loss": 1.3697, "step": 26168 }, { "epoch": 65.4225, "grad_norm": 7.3125, "learning_rate": 5.642101013480934e-06, "loss": 1.416, "step": 26169 }, { "epoch": 65.425, "grad_norm": 9.125, "learning_rate": 5.641372268613632e-06, "loss": 1.2284, "step": 26170 }, { "epoch": 65.4275, "grad_norm": 7.90625, "learning_rate": 5.640643552321323e-06, "loss": 1.2184, "step": 26171 }, { "epoch": 65.43, "grad_norm": 7.1875, "learning_rate": 5.639914864608777e-06, "loss": 1.4044, "step": 26172 }, { "epoch": 65.4325, "grad_norm": 7.3125, "learning_rate": 5.6391862054807776e-06, "loss": 1.4449, "step": 26173 }, { "epoch": 65.435, "grad_norm": 8.3125, "learning_rate": 5.638457574942099e-06, "loss": 1.2489, "step": 26174 }, { "epoch": 65.4375, "grad_norm": 6.59375, "learning_rate": 5.637728972997514e-06, "loss": 1.3675, "step": 26175 }, { "epoch": 65.44, "grad_norm": 7.40625, "learning_rate": 5.637000399651804e-06, "loss": 1.3167, "step": 26176 }, { "epoch": 65.4425, "grad_norm": 9.0, "learning_rate": 5.636271854909748e-06, "loss": 1.3743, "step": 26177 }, { "epoch": 65.445, "grad_norm": 4.96875, "learning_rate": 5.635543338776119e-06, "loss": 1.4103, "step": 26178 }, { "epoch": 65.4475, "grad_norm": 6.40625, "learning_rate": 5.63481485125569e-06, "loss": 1.3463, "step": 26179 }, { "epoch": 65.45, "grad_norm": 10.125, "learning_rate": 5.63408639235324e-06, "loss": 1.3592, "step": 26180 }, { "epoch": 65.4525, "grad_norm": 8.75, "learning_rate": 5.633357962073548e-06, "loss": 1.2432, "step": 26181 }, { "epoch": 65.455, "grad_norm": 9.125, "learning_rate": 5.632629560421388e-06, "loss": 1.3079, "step": 26182 }, { "epoch": 65.4575, "grad_norm": 5.03125, "learning_rate": 5.631901187401527e-06, "loss": 1.3073, "step": 26183 }, { "epoch": 65.46, "grad_norm": 7.5625, "learning_rate": 5.631172843018754e-06, "loss": 1.4094, "step": 26184 }, { "epoch": 65.4625, "grad_norm": 7.625, "learning_rate": 5.630444527277832e-06, "loss": 1.1542, "step": 26185 }, { "epoch": 65.465, "grad_norm": 6.84375, "learning_rate": 5.6297162401835445e-06, "loss": 1.1685, "step": 26186 }, { "epoch": 65.4675, "grad_norm": 7.125, "learning_rate": 5.628987981740659e-06, "loss": 1.1782, "step": 26187 }, { "epoch": 65.47, "grad_norm": 10.0625, "learning_rate": 5.6282597519539575e-06, "loss": 1.211, "step": 26188 }, { "epoch": 65.4725, "grad_norm": 7.75, "learning_rate": 5.627531550828209e-06, "loss": 1.285, "step": 26189 }, { "epoch": 65.475, "grad_norm": 10.0, "learning_rate": 5.626803378368185e-06, "loss": 1.1721, "step": 26190 }, { "epoch": 65.4775, "grad_norm": 7.46875, "learning_rate": 5.626075234578664e-06, "loss": 1.2764, "step": 26191 }, { "epoch": 65.48, "grad_norm": 8.6875, "learning_rate": 5.625347119464422e-06, "loss": 1.3136, "step": 26192 }, { "epoch": 65.4825, "grad_norm": 6.84375, "learning_rate": 5.62461903303023e-06, "loss": 1.3691, "step": 26193 }, { "epoch": 65.485, "grad_norm": 6.46875, "learning_rate": 5.6238909752808546e-06, "loss": 1.2137, "step": 26194 }, { "epoch": 65.4875, "grad_norm": 7.40625, "learning_rate": 5.623162946221077e-06, "loss": 1.2488, "step": 26195 }, { "epoch": 65.49, "grad_norm": 6.03125, "learning_rate": 5.622434945855671e-06, "loss": 1.4647, "step": 26196 }, { "epoch": 65.4925, "grad_norm": 8.0, "learning_rate": 5.621706974189406e-06, "loss": 1.3217, "step": 26197 }, { "epoch": 65.495, "grad_norm": 6.5625, "learning_rate": 5.620979031227052e-06, "loss": 1.3586, "step": 26198 }, { "epoch": 65.4975, "grad_norm": 6.25, "learning_rate": 5.620251116973388e-06, "loss": 1.288, "step": 26199 }, { "epoch": 65.5, "grad_norm": 8.1875, "learning_rate": 5.619523231433177e-06, "loss": 1.2488, "step": 26200 }, { "epoch": 65.5025, "grad_norm": 6.71875, "learning_rate": 5.618795374611201e-06, "loss": 1.1511, "step": 26201 }, { "epoch": 65.505, "grad_norm": 6.15625, "learning_rate": 5.618067546512224e-06, "loss": 1.3159, "step": 26202 }, { "epoch": 65.5075, "grad_norm": 10.625, "learning_rate": 5.617339747141024e-06, "loss": 1.1849, "step": 26203 }, { "epoch": 65.51, "grad_norm": 7.15625, "learning_rate": 5.616611976502369e-06, "loss": 1.3381, "step": 26204 }, { "epoch": 65.5125, "grad_norm": 8.8125, "learning_rate": 5.6158842346010276e-06, "loss": 1.4573, "step": 26205 }, { "epoch": 65.515, "grad_norm": 5.84375, "learning_rate": 5.615156521441774e-06, "loss": 1.3109, "step": 26206 }, { "epoch": 65.5175, "grad_norm": 7.75, "learning_rate": 5.614428837029381e-06, "loss": 1.2785, "step": 26207 }, { "epoch": 65.52, "grad_norm": 8.0625, "learning_rate": 5.613701181368618e-06, "loss": 1.5041, "step": 26208 }, { "epoch": 65.5225, "grad_norm": 8.5625, "learning_rate": 5.61297355446425e-06, "loss": 1.3229, "step": 26209 }, { "epoch": 65.525, "grad_norm": 8.625, "learning_rate": 5.612245956321053e-06, "loss": 1.3314, "step": 26210 }, { "epoch": 65.5275, "grad_norm": 8.75, "learning_rate": 5.6115183869438e-06, "loss": 1.3465, "step": 26211 }, { "epoch": 65.53, "grad_norm": 7.59375, "learning_rate": 5.610790846337255e-06, "loss": 1.3242, "step": 26212 }, { "epoch": 65.5325, "grad_norm": 7.75, "learning_rate": 5.610063334506186e-06, "loss": 1.3801, "step": 26213 }, { "epoch": 65.535, "grad_norm": 7.375, "learning_rate": 5.609335851455371e-06, "loss": 1.289, "step": 26214 }, { "epoch": 65.5375, "grad_norm": 8.625, "learning_rate": 5.608608397189571e-06, "loss": 1.1994, "step": 26215 }, { "epoch": 65.54, "grad_norm": 8.5625, "learning_rate": 5.6078809717135615e-06, "loss": 1.2748, "step": 26216 }, { "epoch": 65.5425, "grad_norm": 7.09375, "learning_rate": 5.607153575032105e-06, "loss": 1.2407, "step": 26217 }, { "epoch": 65.545, "grad_norm": 7.71875, "learning_rate": 5.6064262071499775e-06, "loss": 1.304, "step": 26218 }, { "epoch": 65.5475, "grad_norm": 6.25, "learning_rate": 5.6056988680719445e-06, "loss": 1.1776, "step": 26219 }, { "epoch": 65.55, "grad_norm": 10.25, "learning_rate": 5.604971557802769e-06, "loss": 1.3308, "step": 26220 }, { "epoch": 65.5525, "grad_norm": 7.15625, "learning_rate": 5.604244276347226e-06, "loss": 1.234, "step": 26221 }, { "epoch": 65.555, "grad_norm": 7.96875, "learning_rate": 5.603517023710083e-06, "loss": 1.2381, "step": 26222 }, { "epoch": 65.5575, "grad_norm": 8.125, "learning_rate": 5.602789799896109e-06, "loss": 1.2989, "step": 26223 }, { "epoch": 65.56, "grad_norm": 7.40625, "learning_rate": 5.602062604910064e-06, "loss": 1.2937, "step": 26224 }, { "epoch": 65.5625, "grad_norm": 7.9375, "learning_rate": 5.60133543875672e-06, "loss": 1.2614, "step": 26225 }, { "epoch": 65.565, "grad_norm": 8.75, "learning_rate": 5.600608301440848e-06, "loss": 1.3115, "step": 26226 }, { "epoch": 65.5675, "grad_norm": 7.3125, "learning_rate": 5.599881192967213e-06, "loss": 1.404, "step": 26227 }, { "epoch": 65.57, "grad_norm": 7.375, "learning_rate": 5.599154113340578e-06, "loss": 1.4415, "step": 26228 }, { "epoch": 65.5725, "grad_norm": 6.25, "learning_rate": 5.598427062565715e-06, "loss": 1.4134, "step": 26229 }, { "epoch": 65.575, "grad_norm": 6.125, "learning_rate": 5.597700040647385e-06, "loss": 1.307, "step": 26230 }, { "epoch": 65.5775, "grad_norm": 5.5625, "learning_rate": 5.596973047590362e-06, "loss": 1.369, "step": 26231 }, { "epoch": 65.58, "grad_norm": 8.375, "learning_rate": 5.596246083399402e-06, "loss": 1.2628, "step": 26232 }, { "epoch": 65.5825, "grad_norm": 7.96875, "learning_rate": 5.59551914807928e-06, "loss": 1.3884, "step": 26233 }, { "epoch": 65.585, "grad_norm": 9.625, "learning_rate": 5.594792241634759e-06, "loss": 1.3304, "step": 26234 }, { "epoch": 65.5875, "grad_norm": 9.9375, "learning_rate": 5.5940653640706e-06, "loss": 1.4435, "step": 26235 }, { "epoch": 65.59, "grad_norm": 9.25, "learning_rate": 5.5933385153915735e-06, "loss": 1.4062, "step": 26236 }, { "epoch": 65.5925, "grad_norm": 8.0, "learning_rate": 5.592611695602446e-06, "loss": 1.2903, "step": 26237 }, { "epoch": 65.595, "grad_norm": 7.1875, "learning_rate": 5.591884904707978e-06, "loss": 1.3304, "step": 26238 }, { "epoch": 65.5975, "grad_norm": 6.53125, "learning_rate": 5.591158142712938e-06, "loss": 1.31, "step": 26239 }, { "epoch": 65.6, "grad_norm": 8.75, "learning_rate": 5.590431409622081e-06, "loss": 1.1979, "step": 26240 }, { "epoch": 65.6025, "grad_norm": 6.125, "learning_rate": 5.5897047054401866e-06, "loss": 1.382, "step": 26241 }, { "epoch": 65.605, "grad_norm": 7.15625, "learning_rate": 5.588978030172012e-06, "loss": 1.3321, "step": 26242 }, { "epoch": 65.6075, "grad_norm": 6.78125, "learning_rate": 5.588251383822317e-06, "loss": 1.4222, "step": 26243 }, { "epoch": 65.61, "grad_norm": 7.625, "learning_rate": 5.587524766395873e-06, "loss": 1.1657, "step": 26244 }, { "epoch": 65.6125, "grad_norm": 7.40625, "learning_rate": 5.5867981778974366e-06, "loss": 1.3164, "step": 26245 }, { "epoch": 65.615, "grad_norm": 6.125, "learning_rate": 5.5860716183317785e-06, "loss": 1.3549, "step": 26246 }, { "epoch": 65.6175, "grad_norm": 6.28125, "learning_rate": 5.585345087703654e-06, "loss": 1.3791, "step": 26247 }, { "epoch": 65.62, "grad_norm": 11.125, "learning_rate": 5.5846185860178344e-06, "loss": 1.2903, "step": 26248 }, { "epoch": 65.6225, "grad_norm": 7.75, "learning_rate": 5.583892113279079e-06, "loss": 1.2431, "step": 26249 }, { "epoch": 65.625, "grad_norm": 9.625, "learning_rate": 5.5831656694921465e-06, "loss": 1.153, "step": 26250 }, { "epoch": 65.6275, "grad_norm": 6.75, "learning_rate": 5.5824392546618025e-06, "loss": 1.3831, "step": 26251 }, { "epoch": 65.63, "grad_norm": 8.5625, "learning_rate": 5.581712868792814e-06, "loss": 1.3301, "step": 26252 }, { "epoch": 65.6325, "grad_norm": 8.5, "learning_rate": 5.580986511889939e-06, "loss": 1.3292, "step": 26253 }, { "epoch": 65.635, "grad_norm": 7.15625, "learning_rate": 5.58026018395794e-06, "loss": 1.2522, "step": 26254 }, { "epoch": 65.6375, "grad_norm": 10.75, "learning_rate": 5.579533885001571e-06, "loss": 1.3105, "step": 26255 }, { "epoch": 65.64, "grad_norm": 8.5, "learning_rate": 5.5788076150256075e-06, "loss": 1.2749, "step": 26256 }, { "epoch": 65.6425, "grad_norm": 7.53125, "learning_rate": 5.578081374034805e-06, "loss": 1.2012, "step": 26257 }, { "epoch": 65.645, "grad_norm": 7.53125, "learning_rate": 5.577355162033924e-06, "loss": 1.3501, "step": 26258 }, { "epoch": 65.6475, "grad_norm": 7.9375, "learning_rate": 5.576628979027722e-06, "loss": 1.3871, "step": 26259 }, { "epoch": 65.65, "grad_norm": 7.3125, "learning_rate": 5.575902825020962e-06, "loss": 1.3559, "step": 26260 }, { "epoch": 65.6525, "grad_norm": 7.28125, "learning_rate": 5.575176700018412e-06, "loss": 1.2141, "step": 26261 }, { "epoch": 65.655, "grad_norm": 9.625, "learning_rate": 5.574450604024826e-06, "loss": 1.3089, "step": 26262 }, { "epoch": 65.6575, "grad_norm": 7.3125, "learning_rate": 5.57372453704496e-06, "loss": 1.2315, "step": 26263 }, { "epoch": 65.66, "grad_norm": 7.375, "learning_rate": 5.572998499083583e-06, "loss": 1.3114, "step": 26264 }, { "epoch": 65.6625, "grad_norm": 9.0, "learning_rate": 5.572272490145447e-06, "loss": 1.292, "step": 26265 }, { "epoch": 65.665, "grad_norm": 5.0625, "learning_rate": 5.57154651023532e-06, "loss": 1.3196, "step": 26266 }, { "epoch": 65.6675, "grad_norm": 13.875, "learning_rate": 5.570820559357951e-06, "loss": 1.3973, "step": 26267 }, { "epoch": 65.67, "grad_norm": 5.8125, "learning_rate": 5.570094637518111e-06, "loss": 1.2931, "step": 26268 }, { "epoch": 65.6725, "grad_norm": 7.53125, "learning_rate": 5.569368744720551e-06, "loss": 1.3553, "step": 26269 }, { "epoch": 65.675, "grad_norm": 10.5, "learning_rate": 5.568642880970026e-06, "loss": 1.3383, "step": 26270 }, { "epoch": 65.6775, "grad_norm": 11.4375, "learning_rate": 5.56791704627131e-06, "loss": 1.5756, "step": 26271 }, { "epoch": 65.68, "grad_norm": 10.375, "learning_rate": 5.567191240629151e-06, "loss": 1.4146, "step": 26272 }, { "epoch": 65.6825, "grad_norm": 5.78125, "learning_rate": 5.566465464048307e-06, "loss": 1.1502, "step": 26273 }, { "epoch": 65.685, "grad_norm": 6.40625, "learning_rate": 5.565739716533537e-06, "loss": 1.348, "step": 26274 }, { "epoch": 65.6875, "grad_norm": 8.0, "learning_rate": 5.5650139980895985e-06, "loss": 1.2164, "step": 26275 }, { "epoch": 65.69, "grad_norm": 7.1875, "learning_rate": 5.564288308721255e-06, "loss": 1.3972, "step": 26276 }, { "epoch": 65.6925, "grad_norm": 5.53125, "learning_rate": 5.56356264843326e-06, "loss": 1.3025, "step": 26277 }, { "epoch": 65.695, "grad_norm": 8.9375, "learning_rate": 5.562837017230367e-06, "loss": 1.2725, "step": 26278 }, { "epoch": 65.6975, "grad_norm": 8.125, "learning_rate": 5.562111415117341e-06, "loss": 1.401, "step": 26279 }, { "epoch": 65.7, "grad_norm": 4.0625, "learning_rate": 5.56138584209893e-06, "loss": 1.3492, "step": 26280 }, { "epoch": 65.7025, "grad_norm": 6.6875, "learning_rate": 5.560660298179901e-06, "loss": 1.4432, "step": 26281 }, { "epoch": 65.705, "grad_norm": 7.9375, "learning_rate": 5.5599347833650005e-06, "loss": 1.2601, "step": 26282 }, { "epoch": 65.7075, "grad_norm": 6.6875, "learning_rate": 5.5592092976589936e-06, "loss": 1.3563, "step": 26283 }, { "epoch": 65.71, "grad_norm": 6.78125, "learning_rate": 5.558483841066636e-06, "loss": 1.542, "step": 26284 }, { "epoch": 65.7125, "grad_norm": 11.1875, "learning_rate": 5.557758413592673e-06, "loss": 1.3852, "step": 26285 }, { "epoch": 65.715, "grad_norm": 5.78125, "learning_rate": 5.557033015241868e-06, "loss": 1.4156, "step": 26286 }, { "epoch": 65.7175, "grad_norm": 7.875, "learning_rate": 5.556307646018983e-06, "loss": 1.3036, "step": 26287 }, { "epoch": 65.72, "grad_norm": 5.375, "learning_rate": 5.555582305928766e-06, "loss": 1.3631, "step": 26288 }, { "epoch": 65.7225, "grad_norm": 7.34375, "learning_rate": 5.5548569949759704e-06, "loss": 1.3223, "step": 26289 }, { "epoch": 65.725, "grad_norm": 7.46875, "learning_rate": 5.554131713165353e-06, "loss": 1.2578, "step": 26290 }, { "epoch": 65.7275, "grad_norm": 8.375, "learning_rate": 5.553406460501675e-06, "loss": 1.3977, "step": 26291 }, { "epoch": 65.73, "grad_norm": 11.75, "learning_rate": 5.5526812369896855e-06, "loss": 1.157, "step": 26292 }, { "epoch": 65.7325, "grad_norm": 9.625, "learning_rate": 5.551956042634137e-06, "loss": 1.4484, "step": 26293 }, { "epoch": 65.735, "grad_norm": 7.09375, "learning_rate": 5.55123087743979e-06, "loss": 1.159, "step": 26294 }, { "epoch": 65.7375, "grad_norm": 6.5, "learning_rate": 5.550505741411392e-06, "loss": 1.3481, "step": 26295 }, { "epoch": 65.74, "grad_norm": 4.9375, "learning_rate": 5.549780634553704e-06, "loss": 1.1862, "step": 26296 }, { "epoch": 65.7425, "grad_norm": 5.71875, "learning_rate": 5.5490555568714725e-06, "loss": 1.3892, "step": 26297 }, { "epoch": 65.745, "grad_norm": 4.59375, "learning_rate": 5.548330508369457e-06, "loss": 1.3948, "step": 26298 }, { "epoch": 65.7475, "grad_norm": 6.84375, "learning_rate": 5.547605489052406e-06, "loss": 1.3231, "step": 26299 }, { "epoch": 65.75, "grad_norm": 6.28125, "learning_rate": 5.546880498925079e-06, "loss": 1.3689, "step": 26300 }, { "epoch": 65.7525, "grad_norm": 8.75, "learning_rate": 5.5461555379922215e-06, "loss": 1.3174, "step": 26301 }, { "epoch": 65.755, "grad_norm": 8.6875, "learning_rate": 5.545430606258593e-06, "loss": 1.217, "step": 26302 }, { "epoch": 65.7575, "grad_norm": 17.0, "learning_rate": 5.544705703728944e-06, "loss": 1.3007, "step": 26303 }, { "epoch": 65.76, "grad_norm": 8.0625, "learning_rate": 5.5439808304080225e-06, "loss": 1.3038, "step": 26304 }, { "epoch": 65.7625, "grad_norm": 9.5, "learning_rate": 5.543255986300584e-06, "loss": 1.383, "step": 26305 }, { "epoch": 65.765, "grad_norm": 9.125, "learning_rate": 5.542531171411385e-06, "loss": 1.272, "step": 26306 }, { "epoch": 65.7675, "grad_norm": 9.375, "learning_rate": 5.541806385745172e-06, "loss": 1.4687, "step": 26307 }, { "epoch": 65.77, "grad_norm": 8.875, "learning_rate": 5.541081629306695e-06, "loss": 1.2242, "step": 26308 }, { "epoch": 65.7725, "grad_norm": 6.125, "learning_rate": 5.54035690210071e-06, "loss": 1.293, "step": 26309 }, { "epoch": 65.775, "grad_norm": 9.8125, "learning_rate": 5.539632204131968e-06, "loss": 1.2543, "step": 26310 }, { "epoch": 65.7775, "grad_norm": 6.28125, "learning_rate": 5.53890753540522e-06, "loss": 1.2644, "step": 26311 }, { "epoch": 65.78, "grad_norm": 6.59375, "learning_rate": 5.538182895925212e-06, "loss": 1.2255, "step": 26312 }, { "epoch": 65.7825, "grad_norm": 8.125, "learning_rate": 5.537458285696702e-06, "loss": 1.4619, "step": 26313 }, { "epoch": 65.785, "grad_norm": 6.0, "learning_rate": 5.536733704724432e-06, "loss": 1.3558, "step": 26314 }, { "epoch": 65.7875, "grad_norm": 6.6875, "learning_rate": 5.536009153013162e-06, "loss": 1.2737, "step": 26315 }, { "epoch": 65.79, "grad_norm": 6.8125, "learning_rate": 5.535284630567634e-06, "loss": 1.1666, "step": 26316 }, { "epoch": 65.7925, "grad_norm": 8.8125, "learning_rate": 5.534560137392604e-06, "loss": 1.4306, "step": 26317 }, { "epoch": 65.795, "grad_norm": 8.5625, "learning_rate": 5.53383567349282e-06, "loss": 1.2595, "step": 26318 }, { "epoch": 65.7975, "grad_norm": 7.375, "learning_rate": 5.533111238873026e-06, "loss": 1.3411, "step": 26319 }, { "epoch": 65.8, "grad_norm": 7.8125, "learning_rate": 5.5323868335379775e-06, "loss": 1.443, "step": 26320 }, { "epoch": 65.8025, "grad_norm": 6.5625, "learning_rate": 5.531662457492424e-06, "loss": 1.2593, "step": 26321 }, { "epoch": 65.805, "grad_norm": 7.6875, "learning_rate": 5.530938110741113e-06, "loss": 1.3817, "step": 26322 }, { "epoch": 65.8075, "grad_norm": 6.34375, "learning_rate": 5.53021379328879e-06, "loss": 1.408, "step": 26323 }, { "epoch": 65.81, "grad_norm": 10.0625, "learning_rate": 5.529489505140206e-06, "loss": 1.3019, "step": 26324 }, { "epoch": 65.8125, "grad_norm": 7.1875, "learning_rate": 5.528765246300114e-06, "loss": 1.2346, "step": 26325 }, { "epoch": 65.815, "grad_norm": 6.71875, "learning_rate": 5.528041016773258e-06, "loss": 1.4026, "step": 26326 }, { "epoch": 65.8175, "grad_norm": 8.5, "learning_rate": 5.527316816564381e-06, "loss": 1.2062, "step": 26327 }, { "epoch": 65.82, "grad_norm": 8.4375, "learning_rate": 5.526592645678241e-06, "loss": 1.394, "step": 26328 }, { "epoch": 65.8225, "grad_norm": 8.5, "learning_rate": 5.525868504119577e-06, "loss": 1.4022, "step": 26329 }, { "epoch": 65.825, "grad_norm": 5.65625, "learning_rate": 5.525144391893143e-06, "loss": 1.3661, "step": 26330 }, { "epoch": 65.8275, "grad_norm": 9.5625, "learning_rate": 5.524420309003681e-06, "loss": 1.3882, "step": 26331 }, { "epoch": 65.83, "grad_norm": 7.78125, "learning_rate": 5.523696255455944e-06, "loss": 1.3478, "step": 26332 }, { "epoch": 65.8325, "grad_norm": 7.5, "learning_rate": 5.522972231254673e-06, "loss": 1.344, "step": 26333 }, { "epoch": 65.835, "grad_norm": 7.15625, "learning_rate": 5.522248236404614e-06, "loss": 1.3499, "step": 26334 }, { "epoch": 65.8375, "grad_norm": 7.6875, "learning_rate": 5.521524270910518e-06, "loss": 1.4638, "step": 26335 }, { "epoch": 65.84, "grad_norm": 6.8125, "learning_rate": 5.520800334777132e-06, "loss": 1.2258, "step": 26336 }, { "epoch": 65.8425, "grad_norm": 8.125, "learning_rate": 5.5200764280092e-06, "loss": 1.2968, "step": 26337 }, { "epoch": 65.845, "grad_norm": 7.40625, "learning_rate": 5.519352550611464e-06, "loss": 1.2615, "step": 26338 }, { "epoch": 65.8475, "grad_norm": 6.53125, "learning_rate": 5.518628702588673e-06, "loss": 1.3044, "step": 26339 }, { "epoch": 65.85, "grad_norm": 7.375, "learning_rate": 5.517904883945577e-06, "loss": 1.4143, "step": 26340 }, { "epoch": 65.8525, "grad_norm": 8.625, "learning_rate": 5.517181094686919e-06, "loss": 1.2892, "step": 26341 }, { "epoch": 65.855, "grad_norm": 6.96875, "learning_rate": 5.516457334817435e-06, "loss": 1.2366, "step": 26342 }, { "epoch": 65.8575, "grad_norm": 9.875, "learning_rate": 5.5157336043418844e-06, "loss": 1.3716, "step": 26343 }, { "epoch": 65.86, "grad_norm": 7.1875, "learning_rate": 5.515009903264998e-06, "loss": 1.1942, "step": 26344 }, { "epoch": 65.8625, "grad_norm": 8.375, "learning_rate": 5.5142862315915335e-06, "loss": 1.1963, "step": 26345 }, { "epoch": 65.865, "grad_norm": 6.75, "learning_rate": 5.513562589326224e-06, "loss": 1.3609, "step": 26346 }, { "epoch": 65.8675, "grad_norm": 4.875, "learning_rate": 5.5128389764738225e-06, "loss": 1.3023, "step": 26347 }, { "epoch": 65.87, "grad_norm": 7.71875, "learning_rate": 5.512115393039071e-06, "loss": 1.3059, "step": 26348 }, { "epoch": 65.8725, "grad_norm": 6.8125, "learning_rate": 5.511391839026705e-06, "loss": 1.3069, "step": 26349 }, { "epoch": 65.875, "grad_norm": 9.6875, "learning_rate": 5.510668314441474e-06, "loss": 1.233, "step": 26350 }, { "epoch": 65.8775, "grad_norm": 5.71875, "learning_rate": 5.509944819288128e-06, "loss": 1.2264, "step": 26351 }, { "epoch": 65.88, "grad_norm": 8.6875, "learning_rate": 5.509221353571404e-06, "loss": 1.1856, "step": 26352 }, { "epoch": 65.8825, "grad_norm": 6.46875, "learning_rate": 5.508497917296039e-06, "loss": 1.4612, "step": 26353 }, { "epoch": 65.885, "grad_norm": 6.40625, "learning_rate": 5.507774510466784e-06, "loss": 1.2056, "step": 26354 }, { "epoch": 65.8875, "grad_norm": 6.1875, "learning_rate": 5.507051133088381e-06, "loss": 1.4432, "step": 26355 }, { "epoch": 65.89, "grad_norm": 7.65625, "learning_rate": 5.506327785165574e-06, "loss": 1.3119, "step": 26356 }, { "epoch": 65.8925, "grad_norm": 6.5625, "learning_rate": 5.505604466703096e-06, "loss": 1.3106, "step": 26357 }, { "epoch": 65.895, "grad_norm": 5.25, "learning_rate": 5.504881177705701e-06, "loss": 1.1322, "step": 26358 }, { "epoch": 65.8975, "grad_norm": 6.8125, "learning_rate": 5.50415791817812e-06, "loss": 1.2771, "step": 26359 }, { "epoch": 65.9, "grad_norm": 5.71875, "learning_rate": 5.503434688125104e-06, "loss": 1.3526, "step": 26360 }, { "epoch": 65.9025, "grad_norm": 8.625, "learning_rate": 5.502711487551387e-06, "loss": 1.3647, "step": 26361 }, { "epoch": 65.905, "grad_norm": 6.15625, "learning_rate": 5.501988316461716e-06, "loss": 1.2567, "step": 26362 }, { "epoch": 65.9075, "grad_norm": 8.5625, "learning_rate": 5.5012651748608305e-06, "loss": 1.3816, "step": 26363 }, { "epoch": 65.91, "grad_norm": 8.25, "learning_rate": 5.500542062753467e-06, "loss": 1.2289, "step": 26364 }, { "epoch": 65.9125, "grad_norm": 6.0, "learning_rate": 5.499818980144369e-06, "loss": 1.3985, "step": 26365 }, { "epoch": 65.915, "grad_norm": 6.09375, "learning_rate": 5.499095927038281e-06, "loss": 1.2035, "step": 26366 }, { "epoch": 65.9175, "grad_norm": 7.65625, "learning_rate": 5.4983729034399415e-06, "loss": 1.2841, "step": 26367 }, { "epoch": 65.92, "grad_norm": 7.375, "learning_rate": 5.497649909354084e-06, "loss": 1.2295, "step": 26368 }, { "epoch": 65.9225, "grad_norm": 8.25, "learning_rate": 5.496926944785455e-06, "loss": 1.3249, "step": 26369 }, { "epoch": 65.925, "grad_norm": 7.25, "learning_rate": 5.4962040097387945e-06, "loss": 1.3372, "step": 26370 }, { "epoch": 65.9275, "grad_norm": 7.84375, "learning_rate": 5.4954811042188405e-06, "loss": 1.2374, "step": 26371 }, { "epoch": 65.93, "grad_norm": 8.5625, "learning_rate": 5.49475822823033e-06, "loss": 1.1565, "step": 26372 }, { "epoch": 65.9325, "grad_norm": 5.25, "learning_rate": 5.494035381778007e-06, "loss": 1.3178, "step": 26373 }, { "epoch": 65.935, "grad_norm": 8.125, "learning_rate": 5.493312564866604e-06, "loss": 1.2703, "step": 26374 }, { "epoch": 65.9375, "grad_norm": 6.125, "learning_rate": 5.492589777500868e-06, "loss": 1.2342, "step": 26375 }, { "epoch": 65.94, "grad_norm": 7.625, "learning_rate": 5.491867019685528e-06, "loss": 1.3508, "step": 26376 }, { "epoch": 65.9425, "grad_norm": 6.625, "learning_rate": 5.491144291425332e-06, "loss": 1.1978, "step": 26377 }, { "epoch": 65.945, "grad_norm": 5.9375, "learning_rate": 5.490421592725014e-06, "loss": 1.2366, "step": 26378 }, { "epoch": 65.9475, "grad_norm": 6.5, "learning_rate": 5.489698923589306e-06, "loss": 1.5544, "step": 26379 }, { "epoch": 65.95, "grad_norm": 8.1875, "learning_rate": 5.488976284022953e-06, "loss": 1.3714, "step": 26380 }, { "epoch": 65.9525, "grad_norm": 8.0, "learning_rate": 5.488253674030695e-06, "loss": 1.1812, "step": 26381 }, { "epoch": 65.955, "grad_norm": 8.0, "learning_rate": 5.487531093617265e-06, "loss": 1.328, "step": 26382 }, { "epoch": 65.9575, "grad_norm": 7.59375, "learning_rate": 5.486808542787396e-06, "loss": 1.3874, "step": 26383 }, { "epoch": 65.96, "grad_norm": 6.90625, "learning_rate": 5.486086021545829e-06, "loss": 1.5169, "step": 26384 }, { "epoch": 65.9625, "grad_norm": 9.6875, "learning_rate": 5.485363529897307e-06, "loss": 1.3525, "step": 26385 }, { "epoch": 65.965, "grad_norm": 9.0, "learning_rate": 5.48464106784656e-06, "loss": 1.4294, "step": 26386 }, { "epoch": 65.9675, "grad_norm": 8.0625, "learning_rate": 5.483918635398322e-06, "loss": 1.2791, "step": 26387 }, { "epoch": 65.97, "grad_norm": 10.3125, "learning_rate": 5.483196232557338e-06, "loss": 1.3161, "step": 26388 }, { "epoch": 65.9725, "grad_norm": 8.625, "learning_rate": 5.482473859328333e-06, "loss": 1.4056, "step": 26389 }, { "epoch": 65.975, "grad_norm": 10.25, "learning_rate": 5.481751515716051e-06, "loss": 1.2834, "step": 26390 }, { "epoch": 65.9775, "grad_norm": 8.875, "learning_rate": 5.481029201725223e-06, "loss": 1.2764, "step": 26391 }, { "epoch": 65.98, "grad_norm": 7.6875, "learning_rate": 5.4803069173605915e-06, "loss": 1.2831, "step": 26392 }, { "epoch": 65.9825, "grad_norm": 9.125, "learning_rate": 5.479584662626887e-06, "loss": 1.3994, "step": 26393 }, { "epoch": 65.985, "grad_norm": 6.03125, "learning_rate": 5.47886243752884e-06, "loss": 1.352, "step": 26394 }, { "epoch": 65.9875, "grad_norm": 7.78125, "learning_rate": 5.478140242071189e-06, "loss": 1.3534, "step": 26395 }, { "epoch": 65.99, "grad_norm": 6.96875, "learning_rate": 5.477418076258675e-06, "loss": 1.2075, "step": 26396 }, { "epoch": 65.9925, "grad_norm": 7.75, "learning_rate": 5.476695940096026e-06, "loss": 1.404, "step": 26397 }, { "epoch": 65.995, "grad_norm": 7.03125, "learning_rate": 5.475973833587977e-06, "loss": 1.3375, "step": 26398 }, { "epoch": 65.9975, "grad_norm": 7.59375, "learning_rate": 5.475251756739257e-06, "loss": 1.3064, "step": 26399 }, { "epoch": 66.0, "grad_norm": 8.5625, "learning_rate": 5.4745297095546125e-06, "loss": 1.208, "step": 26400 }, { "epoch": 66.0025, "grad_norm": 5.59375, "learning_rate": 5.473807692038768e-06, "loss": 1.352, "step": 26401 }, { "epoch": 66.005, "grad_norm": 8.1875, "learning_rate": 5.473085704196462e-06, "loss": 1.4092, "step": 26402 }, { "epoch": 66.0075, "grad_norm": 6.90625, "learning_rate": 5.472363746032418e-06, "loss": 1.4218, "step": 26403 }, { "epoch": 66.01, "grad_norm": 8.3125, "learning_rate": 5.4716418175513784e-06, "loss": 1.2665, "step": 26404 }, { "epoch": 66.0125, "grad_norm": 6.125, "learning_rate": 5.470919918758078e-06, "loss": 1.2458, "step": 26405 }, { "epoch": 66.015, "grad_norm": 9.4375, "learning_rate": 5.470198049657244e-06, "loss": 1.3696, "step": 26406 }, { "epoch": 66.0175, "grad_norm": 12.4375, "learning_rate": 5.469476210253606e-06, "loss": 1.4843, "step": 26407 }, { "epoch": 66.02, "grad_norm": 7.03125, "learning_rate": 5.468754400551905e-06, "loss": 1.3931, "step": 26408 }, { "epoch": 66.0225, "grad_norm": 8.625, "learning_rate": 5.468032620556867e-06, "loss": 1.2224, "step": 26409 }, { "epoch": 66.025, "grad_norm": 8.5625, "learning_rate": 5.467310870273224e-06, "loss": 1.3358, "step": 26410 }, { "epoch": 66.0275, "grad_norm": 6.96875, "learning_rate": 5.466589149705713e-06, "loss": 1.2456, "step": 26411 }, { "epoch": 66.03, "grad_norm": 6.6875, "learning_rate": 5.4658674588590625e-06, "loss": 1.3434, "step": 26412 }, { "epoch": 66.0325, "grad_norm": 8.3125, "learning_rate": 5.465145797738004e-06, "loss": 1.3545, "step": 26413 }, { "epoch": 66.035, "grad_norm": 9.4375, "learning_rate": 5.464424166347259e-06, "loss": 1.2783, "step": 26414 }, { "epoch": 66.0375, "grad_norm": 7.46875, "learning_rate": 5.463702564691577e-06, "loss": 1.2902, "step": 26415 }, { "epoch": 66.04, "grad_norm": 7.5, "learning_rate": 5.4629809927756794e-06, "loss": 1.1309, "step": 26416 }, { "epoch": 66.0425, "grad_norm": 7.0, "learning_rate": 5.462259450604297e-06, "loss": 1.4555, "step": 26417 }, { "epoch": 66.045, "grad_norm": 9.625, "learning_rate": 5.461537938182156e-06, "loss": 1.4007, "step": 26418 }, { "epoch": 66.0475, "grad_norm": 7.53125, "learning_rate": 5.4608164555139884e-06, "loss": 1.4428, "step": 26419 }, { "epoch": 66.05, "grad_norm": 6.875, "learning_rate": 5.460095002604533e-06, "loss": 1.2871, "step": 26420 }, { "epoch": 66.0525, "grad_norm": 6.96875, "learning_rate": 5.459373579458512e-06, "loss": 1.486, "step": 26421 }, { "epoch": 66.055, "grad_norm": 5.9375, "learning_rate": 5.458652186080653e-06, "loss": 1.4576, "step": 26422 }, { "epoch": 66.0575, "grad_norm": 6.21875, "learning_rate": 5.457930822475692e-06, "loss": 1.321, "step": 26423 }, { "epoch": 66.06, "grad_norm": 8.625, "learning_rate": 5.457209488648351e-06, "loss": 1.5104, "step": 26424 }, { "epoch": 66.0625, "grad_norm": 5.40625, "learning_rate": 5.456488184603366e-06, "loss": 1.2, "step": 26425 }, { "epoch": 66.065, "grad_norm": 10.8125, "learning_rate": 5.455766910345459e-06, "loss": 1.4312, "step": 26426 }, { "epoch": 66.0675, "grad_norm": 6.84375, "learning_rate": 5.455045665879367e-06, "loss": 1.4162, "step": 26427 }, { "epoch": 66.07, "grad_norm": 8.5625, "learning_rate": 5.454324451209814e-06, "loss": 1.2914, "step": 26428 }, { "epoch": 66.0725, "grad_norm": 6.875, "learning_rate": 5.453603266341523e-06, "loss": 1.235, "step": 26429 }, { "epoch": 66.075, "grad_norm": 13.8125, "learning_rate": 5.452882111279228e-06, "loss": 1.3425, "step": 26430 }, { "epoch": 66.0775, "grad_norm": 5.34375, "learning_rate": 5.452160986027658e-06, "loss": 1.3619, "step": 26431 }, { "epoch": 66.08, "grad_norm": 8.4375, "learning_rate": 5.451439890591539e-06, "loss": 1.4581, "step": 26432 }, { "epoch": 66.0825, "grad_norm": 4.90625, "learning_rate": 5.450718824975596e-06, "loss": 1.3513, "step": 26433 }, { "epoch": 66.085, "grad_norm": 6.15625, "learning_rate": 5.4499977891845565e-06, "loss": 1.3281, "step": 26434 }, { "epoch": 66.0875, "grad_norm": 7.375, "learning_rate": 5.449276783223154e-06, "loss": 1.1139, "step": 26435 }, { "epoch": 66.09, "grad_norm": 5.0625, "learning_rate": 5.448555807096111e-06, "loss": 1.1492, "step": 26436 }, { "epoch": 66.0925, "grad_norm": 5.03125, "learning_rate": 5.44783486080815e-06, "loss": 1.327, "step": 26437 }, { "epoch": 66.095, "grad_norm": 6.4375, "learning_rate": 5.447113944364005e-06, "loss": 1.3651, "step": 26438 }, { "epoch": 66.0975, "grad_norm": 10.25, "learning_rate": 5.446393057768396e-06, "loss": 1.4113, "step": 26439 }, { "epoch": 66.1, "grad_norm": 6.59375, "learning_rate": 5.445672201026054e-06, "loss": 1.2407, "step": 26440 }, { "epoch": 66.1025, "grad_norm": 6.5625, "learning_rate": 5.4449513741417005e-06, "loss": 1.2826, "step": 26441 }, { "epoch": 66.105, "grad_norm": 8.6875, "learning_rate": 5.444230577120067e-06, "loss": 1.431, "step": 26442 }, { "epoch": 66.1075, "grad_norm": 4.96875, "learning_rate": 5.443509809965874e-06, "loss": 1.2795, "step": 26443 }, { "epoch": 66.11, "grad_norm": 5.3125, "learning_rate": 5.442789072683846e-06, "loss": 1.2944, "step": 26444 }, { "epoch": 66.1125, "grad_norm": 6.5, "learning_rate": 5.44206836527871e-06, "loss": 1.22, "step": 26445 }, { "epoch": 66.115, "grad_norm": 4.78125, "learning_rate": 5.4413476877551966e-06, "loss": 1.2451, "step": 26446 }, { "epoch": 66.1175, "grad_norm": 5.96875, "learning_rate": 5.440627040118023e-06, "loss": 1.3561, "step": 26447 }, { "epoch": 66.12, "grad_norm": 8.0625, "learning_rate": 5.439906422371914e-06, "loss": 1.3804, "step": 26448 }, { "epoch": 66.1225, "grad_norm": 17.75, "learning_rate": 5.439185834521596e-06, "loss": 1.5419, "step": 26449 }, { "epoch": 66.125, "grad_norm": 5.75, "learning_rate": 5.438465276571796e-06, "loss": 1.336, "step": 26450 }, { "epoch": 66.1275, "grad_norm": 6.25, "learning_rate": 5.437744748527236e-06, "loss": 1.5013, "step": 26451 }, { "epoch": 66.13, "grad_norm": 10.0, "learning_rate": 5.437024250392634e-06, "loss": 1.3652, "step": 26452 }, { "epoch": 66.1325, "grad_norm": 6.96875, "learning_rate": 5.436303782172723e-06, "loss": 1.2952, "step": 26453 }, { "epoch": 66.135, "grad_norm": 8.125, "learning_rate": 5.435583343872217e-06, "loss": 1.4805, "step": 26454 }, { "epoch": 66.1375, "grad_norm": 7.96875, "learning_rate": 5.434862935495848e-06, "loss": 1.1889, "step": 26455 }, { "epoch": 66.14, "grad_norm": 7.625, "learning_rate": 5.434142557048332e-06, "loss": 1.2414, "step": 26456 }, { "epoch": 66.1425, "grad_norm": 7.71875, "learning_rate": 5.433422208534398e-06, "loss": 1.3379, "step": 26457 }, { "epoch": 66.145, "grad_norm": 7.15625, "learning_rate": 5.432701889958767e-06, "loss": 1.2728, "step": 26458 }, { "epoch": 66.1475, "grad_norm": 8.875, "learning_rate": 5.431981601326153e-06, "loss": 1.4165, "step": 26459 }, { "epoch": 66.15, "grad_norm": 5.6875, "learning_rate": 5.431261342641287e-06, "loss": 1.0294, "step": 26460 }, { "epoch": 66.1525, "grad_norm": 8.25, "learning_rate": 5.430541113908891e-06, "loss": 1.4186, "step": 26461 }, { "epoch": 66.155, "grad_norm": 6.46875, "learning_rate": 5.429820915133685e-06, "loss": 1.3319, "step": 26462 }, { "epoch": 66.1575, "grad_norm": 10.1875, "learning_rate": 5.429100746320386e-06, "loss": 1.3748, "step": 26463 }, { "epoch": 66.16, "grad_norm": 8.5, "learning_rate": 5.42838060747372e-06, "loss": 1.3219, "step": 26464 }, { "epoch": 66.1625, "grad_norm": 9.6875, "learning_rate": 5.427660498598413e-06, "loss": 1.6093, "step": 26465 }, { "epoch": 66.165, "grad_norm": 8.25, "learning_rate": 5.426940419699179e-06, "loss": 1.2528, "step": 26466 }, { "epoch": 66.1675, "grad_norm": 7.5, "learning_rate": 5.426220370780737e-06, "loss": 1.2068, "step": 26467 }, { "epoch": 66.17, "grad_norm": 9.875, "learning_rate": 5.4255003518478165e-06, "loss": 1.3431, "step": 26468 }, { "epoch": 66.1725, "grad_norm": 6.46875, "learning_rate": 5.4247803629051265e-06, "loss": 1.1549, "step": 26469 }, { "epoch": 66.175, "grad_norm": 10.375, "learning_rate": 5.424060403957399e-06, "loss": 1.3204, "step": 26470 }, { "epoch": 66.1775, "grad_norm": 7.96875, "learning_rate": 5.423340475009344e-06, "loss": 1.3874, "step": 26471 }, { "epoch": 66.18, "grad_norm": 8.3125, "learning_rate": 5.422620576065689e-06, "loss": 1.3209, "step": 26472 }, { "epoch": 66.1825, "grad_norm": 6.53125, "learning_rate": 5.421900707131151e-06, "loss": 1.2621, "step": 26473 }, { "epoch": 66.185, "grad_norm": 9.8125, "learning_rate": 5.421180868210445e-06, "loss": 1.2936, "step": 26474 }, { "epoch": 66.1875, "grad_norm": 7.625, "learning_rate": 5.420461059308293e-06, "loss": 1.1847, "step": 26475 }, { "epoch": 66.19, "grad_norm": 7.4375, "learning_rate": 5.41974128042942e-06, "loss": 1.3824, "step": 26476 }, { "epoch": 66.1925, "grad_norm": 6.21875, "learning_rate": 5.41902153157854e-06, "loss": 1.2442, "step": 26477 }, { "epoch": 66.195, "grad_norm": 9.75, "learning_rate": 5.418301812760367e-06, "loss": 1.3576, "step": 26478 }, { "epoch": 66.1975, "grad_norm": 7.375, "learning_rate": 5.417582123979623e-06, "loss": 1.3063, "step": 26479 }, { "epoch": 66.2, "grad_norm": 7.5, "learning_rate": 5.416862465241033e-06, "loss": 1.4874, "step": 26480 }, { "epoch": 66.2025, "grad_norm": 6.65625, "learning_rate": 5.416142836549308e-06, "loss": 1.1795, "step": 26481 }, { "epoch": 66.205, "grad_norm": 6.34375, "learning_rate": 5.415423237909164e-06, "loss": 1.2615, "step": 26482 }, { "epoch": 66.2075, "grad_norm": 8.8125, "learning_rate": 5.4147036693253255e-06, "loss": 1.2612, "step": 26483 }, { "epoch": 66.21, "grad_norm": 5.90625, "learning_rate": 5.4139841308025036e-06, "loss": 1.2713, "step": 26484 }, { "epoch": 66.2125, "grad_norm": 7.75, "learning_rate": 5.41326462234542e-06, "loss": 1.3231, "step": 26485 }, { "epoch": 66.215, "grad_norm": 7.625, "learning_rate": 5.4125451439587865e-06, "loss": 1.342, "step": 26486 }, { "epoch": 66.2175, "grad_norm": 6.375, "learning_rate": 5.411825695647328e-06, "loss": 1.3335, "step": 26487 }, { "epoch": 66.22, "grad_norm": 7.625, "learning_rate": 5.411106277415756e-06, "loss": 1.2603, "step": 26488 }, { "epoch": 66.2225, "grad_norm": 8.625, "learning_rate": 5.410386889268784e-06, "loss": 1.2953, "step": 26489 }, { "epoch": 66.225, "grad_norm": 8.9375, "learning_rate": 5.4096675312111316e-06, "loss": 1.3579, "step": 26490 }, { "epoch": 66.2275, "grad_norm": 6.5625, "learning_rate": 5.40894820324752e-06, "loss": 1.4417, "step": 26491 }, { "epoch": 66.23, "grad_norm": 10.4375, "learning_rate": 5.40822890538266e-06, "loss": 1.3495, "step": 26492 }, { "epoch": 66.2325, "grad_norm": 7.3125, "learning_rate": 5.407509637621262e-06, "loss": 1.3946, "step": 26493 }, { "epoch": 66.235, "grad_norm": 6.09375, "learning_rate": 5.4067903999680495e-06, "loss": 1.3727, "step": 26494 }, { "epoch": 66.2375, "grad_norm": 6.21875, "learning_rate": 5.4060711924277376e-06, "loss": 1.2083, "step": 26495 }, { "epoch": 66.24, "grad_norm": 8.1875, "learning_rate": 5.405352015005039e-06, "loss": 1.3056, "step": 26496 }, { "epoch": 66.2425, "grad_norm": 9.8125, "learning_rate": 5.404632867704665e-06, "loss": 1.3573, "step": 26497 }, { "epoch": 66.245, "grad_norm": 7.96875, "learning_rate": 5.40391375053134e-06, "loss": 1.2941, "step": 26498 }, { "epoch": 66.2475, "grad_norm": 9.375, "learning_rate": 5.4031946634897655e-06, "loss": 1.323, "step": 26499 }, { "epoch": 66.25, "grad_norm": 8.25, "learning_rate": 5.40247560658467e-06, "loss": 1.3204, "step": 26500 }, { "epoch": 66.2525, "grad_norm": 5.84375, "learning_rate": 5.4017565798207536e-06, "loss": 1.328, "step": 26501 }, { "epoch": 66.255, "grad_norm": 7.25, "learning_rate": 5.401037583202742e-06, "loss": 1.2813, "step": 26502 }, { "epoch": 66.2575, "grad_norm": 9.9375, "learning_rate": 5.4003186167353436e-06, "loss": 1.3841, "step": 26503 }, { "epoch": 66.26, "grad_norm": 8.25, "learning_rate": 5.399599680423269e-06, "loss": 1.444, "step": 26504 }, { "epoch": 66.2625, "grad_norm": 7.40625, "learning_rate": 5.398880774271234e-06, "loss": 1.2997, "step": 26505 }, { "epoch": 66.265, "grad_norm": 6.84375, "learning_rate": 5.3981618982839575e-06, "loss": 1.3767, "step": 26506 }, { "epoch": 66.2675, "grad_norm": 10.9375, "learning_rate": 5.397443052466147e-06, "loss": 1.3148, "step": 26507 }, { "epoch": 66.27, "grad_norm": 11.5, "learning_rate": 5.396724236822513e-06, "loss": 1.2753, "step": 26508 }, { "epoch": 66.2725, "grad_norm": 5.15625, "learning_rate": 5.396005451357769e-06, "loss": 1.307, "step": 26509 }, { "epoch": 66.275, "grad_norm": 6.65625, "learning_rate": 5.395286696076633e-06, "loss": 1.4052, "step": 26510 }, { "epoch": 66.2775, "grad_norm": 8.1875, "learning_rate": 5.394567970983813e-06, "loss": 1.3523, "step": 26511 }, { "epoch": 66.28, "grad_norm": 8.9375, "learning_rate": 5.3938492760840176e-06, "loss": 1.2592, "step": 26512 }, { "epoch": 66.2825, "grad_norm": 9.5625, "learning_rate": 5.393130611381967e-06, "loss": 1.4075, "step": 26513 }, { "epoch": 66.285, "grad_norm": 11.375, "learning_rate": 5.392411976882362e-06, "loss": 1.3441, "step": 26514 }, { "epoch": 66.2875, "grad_norm": 5.40625, "learning_rate": 5.391693372589924e-06, "loss": 1.3497, "step": 26515 }, { "epoch": 66.29, "grad_norm": 9.625, "learning_rate": 5.3909747985093565e-06, "loss": 1.4107, "step": 26516 }, { "epoch": 66.2925, "grad_norm": 7.78125, "learning_rate": 5.390256254645379e-06, "loss": 1.2169, "step": 26517 }, { "epoch": 66.295, "grad_norm": 10.0625, "learning_rate": 5.389537741002695e-06, "loss": 1.3192, "step": 26518 }, { "epoch": 66.2975, "grad_norm": 7.53125, "learning_rate": 5.388819257586013e-06, "loss": 1.3849, "step": 26519 }, { "epoch": 66.3, "grad_norm": 5.9375, "learning_rate": 5.3881008044000495e-06, "loss": 1.3282, "step": 26520 }, { "epoch": 66.3025, "grad_norm": 7.1875, "learning_rate": 5.387382381449514e-06, "loss": 1.477, "step": 26521 }, { "epoch": 66.305, "grad_norm": 8.875, "learning_rate": 5.386663988739116e-06, "loss": 1.3702, "step": 26522 }, { "epoch": 66.3075, "grad_norm": 9.5625, "learning_rate": 5.385945626273561e-06, "loss": 1.2958, "step": 26523 }, { "epoch": 66.31, "grad_norm": 6.875, "learning_rate": 5.385227294057561e-06, "loss": 1.2993, "step": 26524 }, { "epoch": 66.3125, "grad_norm": 7.40625, "learning_rate": 5.38450899209583e-06, "loss": 1.2427, "step": 26525 }, { "epoch": 66.315, "grad_norm": 7.375, "learning_rate": 5.383790720393074e-06, "loss": 1.3219, "step": 26526 }, { "epoch": 66.3175, "grad_norm": 7.71875, "learning_rate": 5.383072478953998e-06, "loss": 1.2728, "step": 26527 }, { "epoch": 66.32, "grad_norm": 5.90625, "learning_rate": 5.382354267783316e-06, "loss": 1.2146, "step": 26528 }, { "epoch": 66.3225, "grad_norm": 7.75, "learning_rate": 5.381636086885733e-06, "loss": 1.4082, "step": 26529 }, { "epoch": 66.325, "grad_norm": 7.25, "learning_rate": 5.380917936265961e-06, "loss": 1.2764, "step": 26530 }, { "epoch": 66.3275, "grad_norm": 9.5625, "learning_rate": 5.3801998159287035e-06, "loss": 1.6057, "step": 26531 }, { "epoch": 66.33, "grad_norm": 9.6875, "learning_rate": 5.379481725878674e-06, "loss": 1.3956, "step": 26532 }, { "epoch": 66.3325, "grad_norm": 7.71875, "learning_rate": 5.378763666120579e-06, "loss": 1.2334, "step": 26533 }, { "epoch": 66.335, "grad_norm": 24.0, "learning_rate": 5.3780456366591195e-06, "loss": 1.3326, "step": 26534 }, { "epoch": 66.3375, "grad_norm": 4.6875, "learning_rate": 5.377327637499008e-06, "loss": 1.4539, "step": 26535 }, { "epoch": 66.34, "grad_norm": 9.125, "learning_rate": 5.376609668644954e-06, "loss": 1.2333, "step": 26536 }, { "epoch": 66.3425, "grad_norm": 9.9375, "learning_rate": 5.375891730101664e-06, "loss": 1.332, "step": 26537 }, { "epoch": 66.345, "grad_norm": 8.375, "learning_rate": 5.375173821873843e-06, "loss": 1.3549, "step": 26538 }, { "epoch": 66.3475, "grad_norm": 5.90625, "learning_rate": 5.374455943966188e-06, "loss": 1.3097, "step": 26539 }, { "epoch": 66.35, "grad_norm": 7.71875, "learning_rate": 5.373738096383423e-06, "loss": 1.3028, "step": 26540 }, { "epoch": 66.3525, "grad_norm": 7.09375, "learning_rate": 5.373020279130247e-06, "loss": 1.2139, "step": 26541 }, { "epoch": 66.355, "grad_norm": 5.71875, "learning_rate": 5.372302492211363e-06, "loss": 1.2939, "step": 26542 }, { "epoch": 66.3575, "grad_norm": 8.625, "learning_rate": 5.371584735631472e-06, "loss": 1.2518, "step": 26543 }, { "epoch": 66.36, "grad_norm": 9.1875, "learning_rate": 5.370867009395294e-06, "loss": 1.4363, "step": 26544 }, { "epoch": 66.3625, "grad_norm": 4.9375, "learning_rate": 5.370149313507527e-06, "loss": 1.1335, "step": 26545 }, { "epoch": 66.365, "grad_norm": 7.5625, "learning_rate": 5.3694316479728715e-06, "loss": 1.4363, "step": 26546 }, { "epoch": 66.3675, "grad_norm": 6.28125, "learning_rate": 5.36871401279604e-06, "loss": 1.3874, "step": 26547 }, { "epoch": 66.37, "grad_norm": 4.4375, "learning_rate": 5.367996407981732e-06, "loss": 1.5092, "step": 26548 }, { "epoch": 66.3725, "grad_norm": 6.71875, "learning_rate": 5.367278833534656e-06, "loss": 1.2841, "step": 26549 }, { "epoch": 66.375, "grad_norm": 6.5625, "learning_rate": 5.366561289459512e-06, "loss": 1.2159, "step": 26550 }, { "epoch": 66.3775, "grad_norm": 6.28125, "learning_rate": 5.36584377576101e-06, "loss": 1.2745, "step": 26551 }, { "epoch": 66.38, "grad_norm": 9.625, "learning_rate": 5.365126292443852e-06, "loss": 1.3419, "step": 26552 }, { "epoch": 66.3825, "grad_norm": 9.125, "learning_rate": 5.364408839512735e-06, "loss": 1.4323, "step": 26553 }, { "epoch": 66.385, "grad_norm": 7.6875, "learning_rate": 5.363691416972369e-06, "loss": 1.2251, "step": 26554 }, { "epoch": 66.3875, "grad_norm": 6.03125, "learning_rate": 5.3629740248274605e-06, "loss": 1.2547, "step": 26555 }, { "epoch": 66.39, "grad_norm": 9.0, "learning_rate": 5.362256663082708e-06, "loss": 1.3564, "step": 26556 }, { "epoch": 66.3925, "grad_norm": 9.0, "learning_rate": 5.361539331742816e-06, "loss": 1.2799, "step": 26557 }, { "epoch": 66.395, "grad_norm": 8.0625, "learning_rate": 5.360822030812478e-06, "loss": 1.2875, "step": 26558 }, { "epoch": 66.3975, "grad_norm": 7.0, "learning_rate": 5.360104760296414e-06, "loss": 1.2456, "step": 26559 }, { "epoch": 66.4, "grad_norm": 7.46875, "learning_rate": 5.359387520199317e-06, "loss": 1.3265, "step": 26560 }, { "epoch": 66.4025, "grad_norm": 7.3125, "learning_rate": 5.358670310525889e-06, "loss": 1.4534, "step": 26561 }, { "epoch": 66.405, "grad_norm": 7.21875, "learning_rate": 5.35795313128083e-06, "loss": 1.2475, "step": 26562 }, { "epoch": 66.4075, "grad_norm": 6.875, "learning_rate": 5.357235982468843e-06, "loss": 1.2004, "step": 26563 }, { "epoch": 66.41, "grad_norm": 9.1875, "learning_rate": 5.3565188640946366e-06, "loss": 1.363, "step": 26564 }, { "epoch": 66.4125, "grad_norm": 7.59375, "learning_rate": 5.355801776162906e-06, "loss": 1.17, "step": 26565 }, { "epoch": 66.415, "grad_norm": 10.625, "learning_rate": 5.355084718678349e-06, "loss": 1.2672, "step": 26566 }, { "epoch": 66.4175, "grad_norm": 8.625, "learning_rate": 5.354367691645674e-06, "loss": 1.4477, "step": 26567 }, { "epoch": 66.42, "grad_norm": 8.8125, "learning_rate": 5.353650695069577e-06, "loss": 1.2253, "step": 26568 }, { "epoch": 66.4225, "grad_norm": 8.3125, "learning_rate": 5.3529337289547615e-06, "loss": 1.2732, "step": 26569 }, { "epoch": 66.425, "grad_norm": 8.25, "learning_rate": 5.352216793305924e-06, "loss": 1.3576, "step": 26570 }, { "epoch": 66.4275, "grad_norm": 9.75, "learning_rate": 5.3514998881277715e-06, "loss": 1.2985, "step": 26571 }, { "epoch": 66.43, "grad_norm": 6.84375, "learning_rate": 5.3507830134249985e-06, "loss": 1.3108, "step": 26572 }, { "epoch": 66.4325, "grad_norm": 13.1875, "learning_rate": 5.350066169202298e-06, "loss": 1.2462, "step": 26573 }, { "epoch": 66.435, "grad_norm": 9.3125, "learning_rate": 5.349349355464386e-06, "loss": 1.3475, "step": 26574 }, { "epoch": 66.4375, "grad_norm": 6.3125, "learning_rate": 5.3486325722159535e-06, "loss": 1.4347, "step": 26575 }, { "epoch": 66.44, "grad_norm": 8.75, "learning_rate": 5.3479158194617e-06, "loss": 1.3647, "step": 26576 }, { "epoch": 66.4425, "grad_norm": 9.8125, "learning_rate": 5.347199097206319e-06, "loss": 1.353, "step": 26577 }, { "epoch": 66.445, "grad_norm": 5.96875, "learning_rate": 5.346482405454516e-06, "loss": 1.2697, "step": 26578 }, { "epoch": 66.4475, "grad_norm": 8.4375, "learning_rate": 5.3457657442109925e-06, "loss": 1.2466, "step": 26579 }, { "epoch": 66.45, "grad_norm": 7.75, "learning_rate": 5.3450491134804416e-06, "loss": 1.328, "step": 26580 }, { "epoch": 66.4525, "grad_norm": 9.375, "learning_rate": 5.3443325132675586e-06, "loss": 1.364, "step": 26581 }, { "epoch": 66.455, "grad_norm": 9.0625, "learning_rate": 5.3436159435770495e-06, "loss": 1.2244, "step": 26582 }, { "epoch": 66.4575, "grad_norm": 7.875, "learning_rate": 5.342899404413604e-06, "loss": 1.4624, "step": 26583 }, { "epoch": 66.46, "grad_norm": 5.6875, "learning_rate": 5.342182895781926e-06, "loss": 1.4742, "step": 26584 }, { "epoch": 66.4625, "grad_norm": 7.75, "learning_rate": 5.341466417686709e-06, "loss": 1.2468, "step": 26585 }, { "epoch": 66.465, "grad_norm": 6.40625, "learning_rate": 5.340749970132655e-06, "loss": 1.339, "step": 26586 }, { "epoch": 66.4675, "grad_norm": 6.71875, "learning_rate": 5.340033553124456e-06, "loss": 1.2826, "step": 26587 }, { "epoch": 66.47, "grad_norm": 8.1875, "learning_rate": 5.3393171666668086e-06, "loss": 1.2105, "step": 26588 }, { "epoch": 66.4725, "grad_norm": 13.9375, "learning_rate": 5.33860081076441e-06, "loss": 1.2606, "step": 26589 }, { "epoch": 66.475, "grad_norm": 10.5, "learning_rate": 5.337884485421963e-06, "loss": 1.2908, "step": 26590 }, { "epoch": 66.4775, "grad_norm": 6.5, "learning_rate": 5.337168190644158e-06, "loss": 1.2552, "step": 26591 }, { "epoch": 66.48, "grad_norm": 8.6875, "learning_rate": 5.336451926435688e-06, "loss": 1.3129, "step": 26592 }, { "epoch": 66.4825, "grad_norm": 8.5625, "learning_rate": 5.335735692801254e-06, "loss": 1.4724, "step": 26593 }, { "epoch": 66.485, "grad_norm": 7.5, "learning_rate": 5.335019489745552e-06, "loss": 1.1523, "step": 26594 }, { "epoch": 66.4875, "grad_norm": 10.3125, "learning_rate": 5.3343033172732765e-06, "loss": 1.246, "step": 26595 }, { "epoch": 66.49, "grad_norm": 8.75, "learning_rate": 5.333587175389117e-06, "loss": 1.2228, "step": 26596 }, { "epoch": 66.4925, "grad_norm": 8.125, "learning_rate": 5.3328710640977775e-06, "loss": 1.2559, "step": 26597 }, { "epoch": 66.495, "grad_norm": 7.375, "learning_rate": 5.332154983403945e-06, "loss": 1.1555, "step": 26598 }, { "epoch": 66.4975, "grad_norm": 7.40625, "learning_rate": 5.331438933312321e-06, "loss": 1.4355, "step": 26599 }, { "epoch": 66.5, "grad_norm": 12.75, "learning_rate": 5.330722913827594e-06, "loss": 1.3256, "step": 26600 }, { "epoch": 66.5025, "grad_norm": 8.375, "learning_rate": 5.330006924954463e-06, "loss": 1.402, "step": 26601 }, { "epoch": 66.505, "grad_norm": 7.375, "learning_rate": 5.329290966697621e-06, "loss": 1.3157, "step": 26602 }, { "epoch": 66.5075, "grad_norm": 7.875, "learning_rate": 5.328575039061756e-06, "loss": 1.2138, "step": 26603 }, { "epoch": 66.51, "grad_norm": 9.25, "learning_rate": 5.327859142051565e-06, "loss": 1.4295, "step": 26604 }, { "epoch": 66.5125, "grad_norm": 9.0625, "learning_rate": 5.3271432756717485e-06, "loss": 1.4221, "step": 26605 }, { "epoch": 66.515, "grad_norm": 6.78125, "learning_rate": 5.326427439926993e-06, "loss": 1.3944, "step": 26606 }, { "epoch": 66.5175, "grad_norm": 6.03125, "learning_rate": 5.325711634821988e-06, "loss": 1.273, "step": 26607 }, { "epoch": 66.52, "grad_norm": 7.96875, "learning_rate": 5.32499586036143e-06, "loss": 1.3881, "step": 26608 }, { "epoch": 66.5225, "grad_norm": 7.75, "learning_rate": 5.324280116550016e-06, "loss": 1.3365, "step": 26609 }, { "epoch": 66.525, "grad_norm": 6.71875, "learning_rate": 5.323564403392436e-06, "loss": 1.2996, "step": 26610 }, { "epoch": 66.5275, "grad_norm": 7.71875, "learning_rate": 5.322848720893375e-06, "loss": 1.4615, "step": 26611 }, { "epoch": 66.53, "grad_norm": 8.6875, "learning_rate": 5.322133069057535e-06, "loss": 1.3175, "step": 26612 }, { "epoch": 66.5325, "grad_norm": 6.90625, "learning_rate": 5.3214174478896e-06, "loss": 1.1248, "step": 26613 }, { "epoch": 66.535, "grad_norm": 7.0625, "learning_rate": 5.3207018573942684e-06, "loss": 1.4757, "step": 26614 }, { "epoch": 66.5375, "grad_norm": 5.96875, "learning_rate": 5.319986297576224e-06, "loss": 1.3278, "step": 26615 }, { "epoch": 66.54, "grad_norm": 6.53125, "learning_rate": 5.3192707684401665e-06, "loss": 1.2497, "step": 26616 }, { "epoch": 66.5425, "grad_norm": 7.96875, "learning_rate": 5.318555269990782e-06, "loss": 1.2573, "step": 26617 }, { "epoch": 66.545, "grad_norm": 5.84375, "learning_rate": 5.317839802232758e-06, "loss": 1.1602, "step": 26618 }, { "epoch": 66.5475, "grad_norm": 8.25, "learning_rate": 5.3171243651707894e-06, "loss": 1.4228, "step": 26619 }, { "epoch": 66.55, "grad_norm": 7.65625, "learning_rate": 5.3164089588095705e-06, "loss": 1.2034, "step": 26620 }, { "epoch": 66.5525, "grad_norm": 7.9375, "learning_rate": 5.3156935831537855e-06, "loss": 1.3222, "step": 26621 }, { "epoch": 66.555, "grad_norm": 10.3125, "learning_rate": 5.314978238208124e-06, "loss": 1.4255, "step": 26622 }, { "epoch": 66.5575, "grad_norm": 6.875, "learning_rate": 5.314262923977276e-06, "loss": 1.3525, "step": 26623 }, { "epoch": 66.56, "grad_norm": 7.0625, "learning_rate": 5.313547640465937e-06, "loss": 1.3843, "step": 26624 }, { "epoch": 66.5625, "grad_norm": 9.6875, "learning_rate": 5.312832387678793e-06, "loss": 1.287, "step": 26625 }, { "epoch": 66.565, "grad_norm": 7.125, "learning_rate": 5.312117165620527e-06, "loss": 1.1947, "step": 26626 }, { "epoch": 66.5675, "grad_norm": 6.5625, "learning_rate": 5.3114019742958375e-06, "loss": 1.4594, "step": 26627 }, { "epoch": 66.57, "grad_norm": 7.40625, "learning_rate": 5.3106868137094055e-06, "loss": 1.307, "step": 26628 }, { "epoch": 66.5725, "grad_norm": 7.375, "learning_rate": 5.3099716838659276e-06, "loss": 1.3111, "step": 26629 }, { "epoch": 66.575, "grad_norm": 7.0625, "learning_rate": 5.309256584770084e-06, "loss": 1.3239, "step": 26630 }, { "epoch": 66.5775, "grad_norm": 8.8125, "learning_rate": 5.308541516426569e-06, "loss": 1.4222, "step": 26631 }, { "epoch": 66.58, "grad_norm": 7.84375, "learning_rate": 5.307826478840068e-06, "loss": 1.3492, "step": 26632 }, { "epoch": 66.5825, "grad_norm": 7.625, "learning_rate": 5.307111472015266e-06, "loss": 1.4174, "step": 26633 }, { "epoch": 66.585, "grad_norm": 8.125, "learning_rate": 5.306396495956853e-06, "loss": 1.2405, "step": 26634 }, { "epoch": 66.5875, "grad_norm": 11.0625, "learning_rate": 5.30568155066952e-06, "loss": 1.2704, "step": 26635 }, { "epoch": 66.59, "grad_norm": 9.1875, "learning_rate": 5.304966636157952e-06, "loss": 1.2277, "step": 26636 }, { "epoch": 66.5925, "grad_norm": 28.75, "learning_rate": 5.30425175242683e-06, "loss": 1.2874, "step": 26637 }, { "epoch": 66.595, "grad_norm": 9.8125, "learning_rate": 5.3035368994808465e-06, "loss": 1.4743, "step": 26638 }, { "epoch": 66.5975, "grad_norm": 8.5, "learning_rate": 5.30282207732469e-06, "loss": 1.2855, "step": 26639 }, { "epoch": 66.6, "grad_norm": 12.3125, "learning_rate": 5.302107285963045e-06, "loss": 1.3246, "step": 26640 }, { "epoch": 66.6025, "grad_norm": 7.40625, "learning_rate": 5.301392525400593e-06, "loss": 1.2579, "step": 26641 }, { "epoch": 66.605, "grad_norm": 9.875, "learning_rate": 5.300677795642026e-06, "loss": 1.3787, "step": 26642 }, { "epoch": 66.6075, "grad_norm": 7.90625, "learning_rate": 5.299963096692024e-06, "loss": 1.3876, "step": 26643 }, { "epoch": 66.61, "grad_norm": 6.3125, "learning_rate": 5.29924842855528e-06, "loss": 1.2876, "step": 26644 }, { "epoch": 66.6125, "grad_norm": 6.3125, "learning_rate": 5.298533791236471e-06, "loss": 1.3755, "step": 26645 }, { "epoch": 66.615, "grad_norm": 8.1875, "learning_rate": 5.297819184740291e-06, "loss": 1.2386, "step": 26646 }, { "epoch": 66.6175, "grad_norm": 8.75, "learning_rate": 5.29710460907142e-06, "loss": 1.407, "step": 26647 }, { "epoch": 66.62, "grad_norm": 5.90625, "learning_rate": 5.2963900642345385e-06, "loss": 1.2882, "step": 26648 }, { "epoch": 66.6225, "grad_norm": 7.25, "learning_rate": 5.295675550234336e-06, "loss": 1.3123, "step": 26649 }, { "epoch": 66.625, "grad_norm": 13.125, "learning_rate": 5.2949610670755e-06, "loss": 1.252, "step": 26650 }, { "epoch": 66.6275, "grad_norm": 7.53125, "learning_rate": 5.294246614762712e-06, "loss": 1.3243, "step": 26651 }, { "epoch": 66.63, "grad_norm": 8.125, "learning_rate": 5.29353219330065e-06, "loss": 1.3111, "step": 26652 }, { "epoch": 66.6325, "grad_norm": 10.875, "learning_rate": 5.2928178026940016e-06, "loss": 1.3674, "step": 26653 }, { "epoch": 66.635, "grad_norm": 5.75, "learning_rate": 5.2921034429474564e-06, "loss": 1.3002, "step": 26654 }, { "epoch": 66.6375, "grad_norm": 7.6875, "learning_rate": 5.2913891140656925e-06, "loss": 1.3719, "step": 26655 }, { "epoch": 66.64, "grad_norm": 10.4375, "learning_rate": 5.2906748160533895e-06, "loss": 1.2637, "step": 26656 }, { "epoch": 66.6425, "grad_norm": 6.625, "learning_rate": 5.2899605489152375e-06, "loss": 1.2689, "step": 26657 }, { "epoch": 66.645, "grad_norm": 5.8125, "learning_rate": 5.289246312655911e-06, "loss": 1.1705, "step": 26658 }, { "epoch": 66.6475, "grad_norm": 7.5625, "learning_rate": 5.288532107280102e-06, "loss": 1.3296, "step": 26659 }, { "epoch": 66.65, "grad_norm": 8.0625, "learning_rate": 5.287817932792485e-06, "loss": 1.2553, "step": 26660 }, { "epoch": 66.6525, "grad_norm": 9.5625, "learning_rate": 5.287103789197747e-06, "loss": 1.3048, "step": 26661 }, { "epoch": 66.655, "grad_norm": 8.3125, "learning_rate": 5.2863896765005695e-06, "loss": 1.359, "step": 26662 }, { "epoch": 66.6575, "grad_norm": 6.59375, "learning_rate": 5.285675594705628e-06, "loss": 1.3028, "step": 26663 }, { "epoch": 66.66, "grad_norm": 7.0, "learning_rate": 5.284961543817608e-06, "loss": 1.3528, "step": 26664 }, { "epoch": 66.6625, "grad_norm": 8.3125, "learning_rate": 5.284247523841198e-06, "loss": 1.3584, "step": 26665 }, { "epoch": 66.665, "grad_norm": 7.5, "learning_rate": 5.283533534781069e-06, "loss": 1.2475, "step": 26666 }, { "epoch": 66.6675, "grad_norm": 6.5, "learning_rate": 5.282819576641903e-06, "loss": 1.2267, "step": 26667 }, { "epoch": 66.67, "grad_norm": 8.6875, "learning_rate": 5.282105649428384e-06, "loss": 1.3782, "step": 26668 }, { "epoch": 66.6725, "grad_norm": 6.09375, "learning_rate": 5.281391753145194e-06, "loss": 1.2385, "step": 26669 }, { "epoch": 66.675, "grad_norm": 7.53125, "learning_rate": 5.280677887797012e-06, "loss": 1.3343, "step": 26670 }, { "epoch": 66.6775, "grad_norm": 8.5, "learning_rate": 5.279964053388511e-06, "loss": 1.2406, "step": 26671 }, { "epoch": 66.68, "grad_norm": 11.625, "learning_rate": 5.279250249924384e-06, "loss": 1.3026, "step": 26672 }, { "epoch": 66.6825, "grad_norm": 7.71875, "learning_rate": 5.278536477409297e-06, "loss": 1.3416, "step": 26673 }, { "epoch": 66.685, "grad_norm": 8.4375, "learning_rate": 5.27782273584794e-06, "loss": 1.3494, "step": 26674 }, { "epoch": 66.6875, "grad_norm": 6.625, "learning_rate": 5.2771090252449855e-06, "loss": 1.2004, "step": 26675 }, { "epoch": 66.69, "grad_norm": 11.1875, "learning_rate": 5.276395345605119e-06, "loss": 1.3394, "step": 26676 }, { "epoch": 66.6925, "grad_norm": 7.03125, "learning_rate": 5.275681696933013e-06, "loss": 1.4331, "step": 26677 }, { "epoch": 66.695, "grad_norm": 6.5, "learning_rate": 5.274968079233348e-06, "loss": 1.4074, "step": 26678 }, { "epoch": 66.6975, "grad_norm": 7.59375, "learning_rate": 5.274254492510802e-06, "loss": 1.4074, "step": 26679 }, { "epoch": 66.7, "grad_norm": 7.125, "learning_rate": 5.273540936770059e-06, "loss": 1.196, "step": 26680 }, { "epoch": 66.7025, "grad_norm": 7.96875, "learning_rate": 5.272827412015791e-06, "loss": 1.4682, "step": 26681 }, { "epoch": 66.705, "grad_norm": 9.3125, "learning_rate": 5.272113918252678e-06, "loss": 1.3527, "step": 26682 }, { "epoch": 66.7075, "grad_norm": 9.3125, "learning_rate": 5.271400455485389e-06, "loss": 1.4252, "step": 26683 }, { "epoch": 66.71, "grad_norm": 8.375, "learning_rate": 5.270687023718618e-06, "loss": 1.2629, "step": 26684 }, { "epoch": 66.7125, "grad_norm": 9.5625, "learning_rate": 5.2699736229570335e-06, "loss": 1.2815, "step": 26685 }, { "epoch": 66.715, "grad_norm": 10.125, "learning_rate": 5.269260253205308e-06, "loss": 1.409, "step": 26686 }, { "epoch": 66.7175, "grad_norm": 5.84375, "learning_rate": 5.2685469144681265e-06, "loss": 1.3177, "step": 26687 }, { "epoch": 66.72, "grad_norm": 7.59375, "learning_rate": 5.26783360675016e-06, "loss": 1.3031, "step": 26688 }, { "epoch": 66.7225, "grad_norm": 7.6875, "learning_rate": 5.26712033005609e-06, "loss": 1.5176, "step": 26689 }, { "epoch": 66.725, "grad_norm": 7.1875, "learning_rate": 5.2664070843905855e-06, "loss": 1.3347, "step": 26690 }, { "epoch": 66.7275, "grad_norm": 7.65625, "learning_rate": 5.265693869758331e-06, "loss": 1.416, "step": 26691 }, { "epoch": 66.73, "grad_norm": 5.84375, "learning_rate": 5.2649806861639975e-06, "loss": 1.288, "step": 26692 }, { "epoch": 66.7325, "grad_norm": 6.96875, "learning_rate": 5.264267533612257e-06, "loss": 1.39, "step": 26693 }, { "epoch": 66.735, "grad_norm": 12.25, "learning_rate": 5.26355441210779e-06, "loss": 1.4161, "step": 26694 }, { "epoch": 66.7375, "grad_norm": 5.875, "learning_rate": 5.262841321655274e-06, "loss": 1.3994, "step": 26695 }, { "epoch": 66.74, "grad_norm": 8.4375, "learning_rate": 5.26212826225938e-06, "loss": 1.2287, "step": 26696 }, { "epoch": 66.7425, "grad_norm": 9.25, "learning_rate": 5.2614152339247845e-06, "loss": 1.3221, "step": 26697 }, { "epoch": 66.745, "grad_norm": 6.71875, "learning_rate": 5.2607022366561544e-06, "loss": 1.3181, "step": 26698 }, { "epoch": 66.7475, "grad_norm": 7.28125, "learning_rate": 5.259989270458177e-06, "loss": 1.3085, "step": 26699 }, { "epoch": 66.75, "grad_norm": 7.5, "learning_rate": 5.259276335335522e-06, "loss": 1.294, "step": 26700 }, { "epoch": 66.7525, "grad_norm": 6.84375, "learning_rate": 5.25856343129286e-06, "loss": 1.3272, "step": 26701 }, { "epoch": 66.755, "grad_norm": 10.0, "learning_rate": 5.257850558334862e-06, "loss": 1.3273, "step": 26702 }, { "epoch": 66.7575, "grad_norm": 8.75, "learning_rate": 5.257137716466208e-06, "loss": 1.3896, "step": 26703 }, { "epoch": 66.76, "grad_norm": 7.625, "learning_rate": 5.2564249056915704e-06, "loss": 1.2108, "step": 26704 }, { "epoch": 66.7625, "grad_norm": 6.28125, "learning_rate": 5.255712126015624e-06, "loss": 1.3313, "step": 26705 }, { "epoch": 66.765, "grad_norm": 5.78125, "learning_rate": 5.254999377443033e-06, "loss": 1.2387, "step": 26706 }, { "epoch": 66.7675, "grad_norm": 6.6875, "learning_rate": 5.254286659978482e-06, "loss": 1.2405, "step": 26707 }, { "epoch": 66.77, "grad_norm": 6.3125, "learning_rate": 5.253573973626634e-06, "loss": 1.1942, "step": 26708 }, { "epoch": 66.7725, "grad_norm": 8.625, "learning_rate": 5.252861318392166e-06, "loss": 1.3597, "step": 26709 }, { "epoch": 66.775, "grad_norm": 9.75, "learning_rate": 5.252148694279749e-06, "loss": 1.2769, "step": 26710 }, { "epoch": 66.7775, "grad_norm": 8.5, "learning_rate": 5.251436101294057e-06, "loss": 1.3643, "step": 26711 }, { "epoch": 66.78, "grad_norm": 7.15625, "learning_rate": 5.2507235394397595e-06, "loss": 1.4082, "step": 26712 }, { "epoch": 66.7825, "grad_norm": 11.6875, "learning_rate": 5.250011008721522e-06, "loss": 1.308, "step": 26713 }, { "epoch": 66.785, "grad_norm": 6.46875, "learning_rate": 5.24929850914403e-06, "loss": 1.2762, "step": 26714 }, { "epoch": 66.7875, "grad_norm": 9.375, "learning_rate": 5.248586040711945e-06, "loss": 1.362, "step": 26715 }, { "epoch": 66.79, "grad_norm": 8.1875, "learning_rate": 5.2478736034299415e-06, "loss": 1.3765, "step": 26716 }, { "epoch": 66.7925, "grad_norm": 7.09375, "learning_rate": 5.247161197302684e-06, "loss": 1.4312, "step": 26717 }, { "epoch": 66.795, "grad_norm": 6.1875, "learning_rate": 5.246448822334848e-06, "loss": 1.3399, "step": 26718 }, { "epoch": 66.7975, "grad_norm": 7.65625, "learning_rate": 5.2457364785311074e-06, "loss": 1.4244, "step": 26719 }, { "epoch": 66.8, "grad_norm": 8.75, "learning_rate": 5.245024165896126e-06, "loss": 1.2889, "step": 26720 }, { "epoch": 66.8025, "grad_norm": 7.4375, "learning_rate": 5.244311884434574e-06, "loss": 1.2534, "step": 26721 }, { "epoch": 66.805, "grad_norm": 9.0, "learning_rate": 5.243599634151126e-06, "loss": 1.2382, "step": 26722 }, { "epoch": 66.8075, "grad_norm": 7.0625, "learning_rate": 5.242887415050446e-06, "loss": 1.3331, "step": 26723 }, { "epoch": 66.81, "grad_norm": 6.5625, "learning_rate": 5.242175227137209e-06, "loss": 1.3688, "step": 26724 }, { "epoch": 66.8125, "grad_norm": 6.0, "learning_rate": 5.241463070416076e-06, "loss": 1.3301, "step": 26725 }, { "epoch": 66.815, "grad_norm": 5.75, "learning_rate": 5.240750944891726e-06, "loss": 1.2699, "step": 26726 }, { "epoch": 66.8175, "grad_norm": 6.71875, "learning_rate": 5.240038850568821e-06, "loss": 1.3477, "step": 26727 }, { "epoch": 66.82, "grad_norm": 8.375, "learning_rate": 5.239326787452028e-06, "loss": 1.2813, "step": 26728 }, { "epoch": 66.8225, "grad_norm": 10.3125, "learning_rate": 5.2386147555460165e-06, "loss": 1.2545, "step": 26729 }, { "epoch": 66.825, "grad_norm": 5.59375, "learning_rate": 5.237902754855462e-06, "loss": 1.3826, "step": 26730 }, { "epoch": 66.8275, "grad_norm": 6.84375, "learning_rate": 5.237190785385024e-06, "loss": 1.3169, "step": 26731 }, { "epoch": 66.83, "grad_norm": 8.6875, "learning_rate": 5.2364788471393714e-06, "loss": 1.2575, "step": 26732 }, { "epoch": 66.8325, "grad_norm": 7.53125, "learning_rate": 5.235766940123171e-06, "loss": 1.357, "step": 26733 }, { "epoch": 66.835, "grad_norm": 6.5625, "learning_rate": 5.2350550643410965e-06, "loss": 1.3439, "step": 26734 }, { "epoch": 66.8375, "grad_norm": 7.375, "learning_rate": 5.23434321979781e-06, "loss": 1.2798, "step": 26735 }, { "epoch": 66.84, "grad_norm": 10.8125, "learning_rate": 5.2336314064979766e-06, "loss": 1.3685, "step": 26736 }, { "epoch": 66.8425, "grad_norm": 6.90625, "learning_rate": 5.2329196244462665e-06, "loss": 1.2273, "step": 26737 }, { "epoch": 66.845, "grad_norm": 6.84375, "learning_rate": 5.232207873647341e-06, "loss": 1.3491, "step": 26738 }, { "epoch": 66.8475, "grad_norm": 5.0, "learning_rate": 5.231496154105875e-06, "loss": 1.3936, "step": 26739 }, { "epoch": 66.85, "grad_norm": 6.84375, "learning_rate": 5.2307844658265236e-06, "loss": 1.2353, "step": 26740 }, { "epoch": 66.8525, "grad_norm": 8.0625, "learning_rate": 5.230072808813964e-06, "loss": 1.3353, "step": 26741 }, { "epoch": 66.855, "grad_norm": 6.84375, "learning_rate": 5.2293611830728565e-06, "loss": 1.514, "step": 26742 }, { "epoch": 66.8575, "grad_norm": 6.4375, "learning_rate": 5.2286495886078614e-06, "loss": 1.2481, "step": 26743 }, { "epoch": 66.86, "grad_norm": 7.65625, "learning_rate": 5.2279380254236475e-06, "loss": 1.5044, "step": 26744 }, { "epoch": 66.8625, "grad_norm": 7.59375, "learning_rate": 5.227226493524886e-06, "loss": 1.3558, "step": 26745 }, { "epoch": 66.865, "grad_norm": 6.0625, "learning_rate": 5.226514992916237e-06, "loss": 1.1858, "step": 26746 }, { "epoch": 66.8675, "grad_norm": 10.3125, "learning_rate": 5.22580352360236e-06, "loss": 1.3712, "step": 26747 }, { "epoch": 66.87, "grad_norm": 9.875, "learning_rate": 5.225092085587925e-06, "loss": 1.364, "step": 26748 }, { "epoch": 66.8725, "grad_norm": 7.59375, "learning_rate": 5.224380678877599e-06, "loss": 1.3658, "step": 26749 }, { "epoch": 66.875, "grad_norm": 6.625, "learning_rate": 5.223669303476041e-06, "loss": 1.3285, "step": 26750 }, { "epoch": 66.8775, "grad_norm": 6.28125, "learning_rate": 5.2229579593879134e-06, "loss": 1.3233, "step": 26751 }, { "epoch": 66.88, "grad_norm": 9.625, "learning_rate": 5.222246646617886e-06, "loss": 1.2728, "step": 26752 }, { "epoch": 66.8825, "grad_norm": 6.71875, "learning_rate": 5.221535365170614e-06, "loss": 1.2924, "step": 26753 }, { "epoch": 66.885, "grad_norm": 14.5, "learning_rate": 5.220824115050767e-06, "loss": 1.4558, "step": 26754 }, { "epoch": 66.8875, "grad_norm": 8.3125, "learning_rate": 5.220112896263004e-06, "loss": 1.3629, "step": 26755 }, { "epoch": 66.89, "grad_norm": 6.96875, "learning_rate": 5.219401708811993e-06, "loss": 1.305, "step": 26756 }, { "epoch": 66.8925, "grad_norm": 7.09375, "learning_rate": 5.218690552702394e-06, "loss": 1.3778, "step": 26757 }, { "epoch": 66.895, "grad_norm": 5.5625, "learning_rate": 5.217979427938864e-06, "loss": 1.2502, "step": 26758 }, { "epoch": 66.8975, "grad_norm": 7.78125, "learning_rate": 5.217268334526069e-06, "loss": 1.2522, "step": 26759 }, { "epoch": 66.9, "grad_norm": 8.75, "learning_rate": 5.216557272468675e-06, "loss": 1.2989, "step": 26760 }, { "epoch": 66.9025, "grad_norm": 5.3125, "learning_rate": 5.215846241771341e-06, "loss": 1.4213, "step": 26761 }, { "epoch": 66.905, "grad_norm": 6.9375, "learning_rate": 5.215135242438722e-06, "loss": 1.455, "step": 26762 }, { "epoch": 66.9075, "grad_norm": 8.1875, "learning_rate": 5.214424274475487e-06, "loss": 1.2641, "step": 26763 }, { "epoch": 66.91, "grad_norm": 6.375, "learning_rate": 5.213713337886299e-06, "loss": 1.3723, "step": 26764 }, { "epoch": 66.9125, "grad_norm": 6.6875, "learning_rate": 5.213002432675813e-06, "loss": 1.3389, "step": 26765 }, { "epoch": 66.915, "grad_norm": 8.3125, "learning_rate": 5.212291558848689e-06, "loss": 1.3685, "step": 26766 }, { "epoch": 66.9175, "grad_norm": 7.9375, "learning_rate": 5.211580716409593e-06, "loss": 1.2877, "step": 26767 }, { "epoch": 66.92, "grad_norm": 7.25, "learning_rate": 5.210869905363178e-06, "loss": 1.3438, "step": 26768 }, { "epoch": 66.9225, "grad_norm": 6.15625, "learning_rate": 5.2101591257141135e-06, "loss": 1.3336, "step": 26769 }, { "epoch": 66.925, "grad_norm": 6.125, "learning_rate": 5.20944837746705e-06, "loss": 1.3064, "step": 26770 }, { "epoch": 66.9275, "grad_norm": 7.28125, "learning_rate": 5.208737660626655e-06, "loss": 1.3456, "step": 26771 }, { "epoch": 66.93, "grad_norm": 8.0, "learning_rate": 5.208026975197583e-06, "loss": 1.2155, "step": 26772 }, { "epoch": 66.9325, "grad_norm": 5.96875, "learning_rate": 5.207316321184491e-06, "loss": 1.2314, "step": 26773 }, { "epoch": 66.935, "grad_norm": 7.6875, "learning_rate": 5.2066056985920425e-06, "loss": 1.3948, "step": 26774 }, { "epoch": 66.9375, "grad_norm": 7.78125, "learning_rate": 5.2058951074248985e-06, "loss": 1.3385, "step": 26775 }, { "epoch": 66.94, "grad_norm": 8.0, "learning_rate": 5.205184547687714e-06, "loss": 1.3021, "step": 26776 }, { "epoch": 66.9425, "grad_norm": 7.40625, "learning_rate": 5.204474019385144e-06, "loss": 1.3302, "step": 26777 }, { "epoch": 66.945, "grad_norm": 9.3125, "learning_rate": 5.20376352252185e-06, "loss": 1.3896, "step": 26778 }, { "epoch": 66.9475, "grad_norm": 11.0, "learning_rate": 5.203053057102494e-06, "loss": 1.4351, "step": 26779 }, { "epoch": 66.95, "grad_norm": 7.0, "learning_rate": 5.202342623131731e-06, "loss": 1.3054, "step": 26780 }, { "epoch": 66.9525, "grad_norm": 8.0625, "learning_rate": 5.201632220614215e-06, "loss": 1.3827, "step": 26781 }, { "epoch": 66.955, "grad_norm": 8.5625, "learning_rate": 5.200921849554608e-06, "loss": 1.4451, "step": 26782 }, { "epoch": 66.9575, "grad_norm": 6.1875, "learning_rate": 5.200211509957562e-06, "loss": 1.2777, "step": 26783 }, { "epoch": 66.96, "grad_norm": 6.71875, "learning_rate": 5.199501201827741e-06, "loss": 1.1344, "step": 26784 }, { "epoch": 66.9625, "grad_norm": 7.28125, "learning_rate": 5.198790925169795e-06, "loss": 1.4311, "step": 26785 }, { "epoch": 66.965, "grad_norm": 8.5625, "learning_rate": 5.198080679988387e-06, "loss": 1.4143, "step": 26786 }, { "epoch": 66.9675, "grad_norm": 7.375, "learning_rate": 5.1973704662881655e-06, "loss": 1.2469, "step": 26787 }, { "epoch": 66.97, "grad_norm": 8.1875, "learning_rate": 5.196660284073795e-06, "loss": 1.3679, "step": 26788 }, { "epoch": 66.9725, "grad_norm": 9.1875, "learning_rate": 5.1959501333499256e-06, "loss": 1.2013, "step": 26789 }, { "epoch": 66.975, "grad_norm": 8.0625, "learning_rate": 5.195240014121217e-06, "loss": 1.3956, "step": 26790 }, { "epoch": 66.9775, "grad_norm": 7.46875, "learning_rate": 5.194529926392322e-06, "loss": 1.392, "step": 26791 }, { "epoch": 66.98, "grad_norm": 8.0625, "learning_rate": 5.193819870167893e-06, "loss": 1.3279, "step": 26792 }, { "epoch": 66.9825, "grad_norm": 9.25, "learning_rate": 5.193109845452589e-06, "loss": 1.44, "step": 26793 }, { "epoch": 66.985, "grad_norm": 8.0, "learning_rate": 5.1923998522510685e-06, "loss": 1.3341, "step": 26794 }, { "epoch": 66.9875, "grad_norm": 20.25, "learning_rate": 5.1916898905679815e-06, "loss": 1.2123, "step": 26795 }, { "epoch": 66.99, "grad_norm": 8.625, "learning_rate": 5.19097996040798e-06, "loss": 1.2653, "step": 26796 }, { "epoch": 66.9925, "grad_norm": 11.5625, "learning_rate": 5.190270061775722e-06, "loss": 1.4904, "step": 26797 }, { "epoch": 66.995, "grad_norm": 10.875, "learning_rate": 5.1895601946758625e-06, "loss": 1.495, "step": 26798 }, { "epoch": 66.9975, "grad_norm": 7.6875, "learning_rate": 5.188850359113056e-06, "loss": 1.4488, "step": 26799 }, { "epoch": 67.0, "grad_norm": 8.25, "learning_rate": 5.18814055509195e-06, "loss": 1.3865, "step": 26800 }, { "epoch": 67.0025, "grad_norm": 8.9375, "learning_rate": 5.187430782617205e-06, "loss": 1.3396, "step": 26801 }, { "epoch": 67.005, "grad_norm": 6.8125, "learning_rate": 5.186721041693468e-06, "loss": 1.2638, "step": 26802 }, { "epoch": 67.0075, "grad_norm": 5.75, "learning_rate": 5.186011332325398e-06, "loss": 1.3659, "step": 26803 }, { "epoch": 67.01, "grad_norm": 16.75, "learning_rate": 5.185301654517643e-06, "loss": 1.2938, "step": 26804 }, { "epoch": 67.0125, "grad_norm": 10.3125, "learning_rate": 5.184592008274861e-06, "loss": 1.4149, "step": 26805 }, { "epoch": 67.015, "grad_norm": 8.4375, "learning_rate": 5.1838823936017005e-06, "loss": 1.4514, "step": 26806 }, { "epoch": 67.0175, "grad_norm": 8.0625, "learning_rate": 5.18317281050281e-06, "loss": 1.3598, "step": 26807 }, { "epoch": 67.02, "grad_norm": 8.0625, "learning_rate": 5.1824632589828465e-06, "loss": 1.3716, "step": 26808 }, { "epoch": 67.0225, "grad_norm": 8.625, "learning_rate": 5.181753739046464e-06, "loss": 1.3275, "step": 26809 }, { "epoch": 67.025, "grad_norm": 10.25, "learning_rate": 5.181044250698313e-06, "loss": 1.4004, "step": 26810 }, { "epoch": 67.0275, "grad_norm": 6.375, "learning_rate": 5.180334793943038e-06, "loss": 1.3003, "step": 26811 }, { "epoch": 67.03, "grad_norm": 6.25, "learning_rate": 5.1796253687852945e-06, "loss": 1.3777, "step": 26812 }, { "epoch": 67.0325, "grad_norm": 7.46875, "learning_rate": 5.178915975229739e-06, "loss": 1.2475, "step": 26813 }, { "epoch": 67.035, "grad_norm": 6.25, "learning_rate": 5.178206613281018e-06, "loss": 1.2669, "step": 26814 }, { "epoch": 67.0375, "grad_norm": 7.25, "learning_rate": 5.177497282943778e-06, "loss": 1.0979, "step": 26815 }, { "epoch": 67.04, "grad_norm": 6.53125, "learning_rate": 5.1767879842226745e-06, "loss": 1.2923, "step": 26816 }, { "epoch": 67.0425, "grad_norm": 6.375, "learning_rate": 5.176078717122352e-06, "loss": 1.3005, "step": 26817 }, { "epoch": 67.045, "grad_norm": 7.59375, "learning_rate": 5.1753694816474705e-06, "loss": 1.4507, "step": 26818 }, { "epoch": 67.0475, "grad_norm": 9.75, "learning_rate": 5.174660277802669e-06, "loss": 1.4195, "step": 26819 }, { "epoch": 67.05, "grad_norm": 7.59375, "learning_rate": 5.173951105592605e-06, "loss": 1.2219, "step": 26820 }, { "epoch": 67.0525, "grad_norm": 8.5625, "learning_rate": 5.173241965021925e-06, "loss": 1.4161, "step": 26821 }, { "epoch": 67.055, "grad_norm": 8.625, "learning_rate": 5.172532856095274e-06, "loss": 1.3555, "step": 26822 }, { "epoch": 67.0575, "grad_norm": 5.28125, "learning_rate": 5.171823778817303e-06, "loss": 1.4176, "step": 26823 }, { "epoch": 67.06, "grad_norm": 11.375, "learning_rate": 5.171114733192667e-06, "loss": 1.2864, "step": 26824 }, { "epoch": 67.0625, "grad_norm": 7.53125, "learning_rate": 5.170405719226009e-06, "loss": 1.4607, "step": 26825 }, { "epoch": 67.065, "grad_norm": 9.375, "learning_rate": 5.169696736921974e-06, "loss": 1.4492, "step": 26826 }, { "epoch": 67.0675, "grad_norm": 7.625, "learning_rate": 5.168987786285214e-06, "loss": 1.2263, "step": 26827 }, { "epoch": 67.07, "grad_norm": 8.125, "learning_rate": 5.168278867320382e-06, "loss": 1.3071, "step": 26828 }, { "epoch": 67.0725, "grad_norm": 7.96875, "learning_rate": 5.167569980032119e-06, "loss": 1.4754, "step": 26829 }, { "epoch": 67.075, "grad_norm": 6.78125, "learning_rate": 5.166861124425069e-06, "loss": 1.3042, "step": 26830 }, { "epoch": 67.0775, "grad_norm": 7.46875, "learning_rate": 5.1661523005038905e-06, "loss": 1.3684, "step": 26831 }, { "epoch": 67.08, "grad_norm": 6.0, "learning_rate": 5.165443508273218e-06, "loss": 1.2631, "step": 26832 }, { "epoch": 67.0825, "grad_norm": 9.25, "learning_rate": 5.16473474773771e-06, "loss": 1.3498, "step": 26833 }, { "epoch": 67.085, "grad_norm": 13.1875, "learning_rate": 5.164026018902002e-06, "loss": 1.3952, "step": 26834 }, { "epoch": 67.0875, "grad_norm": 8.625, "learning_rate": 5.16331732177075e-06, "loss": 1.2312, "step": 26835 }, { "epoch": 67.09, "grad_norm": 7.71875, "learning_rate": 5.162608656348597e-06, "loss": 1.4702, "step": 26836 }, { "epoch": 67.0925, "grad_norm": 7.3125, "learning_rate": 5.161900022640182e-06, "loss": 1.314, "step": 26837 }, { "epoch": 67.095, "grad_norm": 5.75, "learning_rate": 5.16119142065016e-06, "loss": 1.2313, "step": 26838 }, { "epoch": 67.0975, "grad_norm": 9.625, "learning_rate": 5.160482850383175e-06, "loss": 1.2942, "step": 26839 }, { "epoch": 67.1, "grad_norm": 9.0, "learning_rate": 5.1597743118438725e-06, "loss": 1.307, "step": 26840 }, { "epoch": 67.1025, "grad_norm": 6.53125, "learning_rate": 5.159065805036895e-06, "loss": 1.2519, "step": 26841 }, { "epoch": 67.105, "grad_norm": 9.9375, "learning_rate": 5.158357329966881e-06, "loss": 1.3313, "step": 26842 }, { "epoch": 67.1075, "grad_norm": 7.15625, "learning_rate": 5.157648886638489e-06, "loss": 1.3185, "step": 26843 }, { "epoch": 67.11, "grad_norm": 6.0625, "learning_rate": 5.156940475056358e-06, "loss": 1.2004, "step": 26844 }, { "epoch": 67.1125, "grad_norm": 10.25, "learning_rate": 5.156232095225131e-06, "loss": 1.377, "step": 26845 }, { "epoch": 67.115, "grad_norm": 7.875, "learning_rate": 5.15552374714945e-06, "loss": 1.3824, "step": 26846 }, { "epoch": 67.1175, "grad_norm": 5.46875, "learning_rate": 5.15481543083396e-06, "loss": 1.2604, "step": 26847 }, { "epoch": 67.12, "grad_norm": 5.75, "learning_rate": 5.154107146283311e-06, "loss": 1.2999, "step": 26848 }, { "epoch": 67.1225, "grad_norm": 6.125, "learning_rate": 5.153398893502142e-06, "loss": 1.3025, "step": 26849 }, { "epoch": 67.125, "grad_norm": 6.03125, "learning_rate": 5.152690672495091e-06, "loss": 1.3358, "step": 26850 }, { "epoch": 67.1275, "grad_norm": 7.03125, "learning_rate": 5.1519824832668095e-06, "loss": 1.2111, "step": 26851 }, { "epoch": 67.13, "grad_norm": 4.6875, "learning_rate": 5.151274325821934e-06, "loss": 1.2556, "step": 26852 }, { "epoch": 67.1325, "grad_norm": 6.15625, "learning_rate": 5.15056620016511e-06, "loss": 1.385, "step": 26853 }, { "epoch": 67.135, "grad_norm": 8.25, "learning_rate": 5.149858106300982e-06, "loss": 1.2758, "step": 26854 }, { "epoch": 67.1375, "grad_norm": 6.375, "learning_rate": 5.149150044234192e-06, "loss": 1.3568, "step": 26855 }, { "epoch": 67.14, "grad_norm": 5.53125, "learning_rate": 5.1484420139693805e-06, "loss": 1.3962, "step": 26856 }, { "epoch": 67.1425, "grad_norm": 7.03125, "learning_rate": 5.147734015511181e-06, "loss": 1.2483, "step": 26857 }, { "epoch": 67.145, "grad_norm": 7.46875, "learning_rate": 5.14702604886425e-06, "loss": 1.3364, "step": 26858 }, { "epoch": 67.1475, "grad_norm": 9.8125, "learning_rate": 5.146318114033222e-06, "loss": 1.2665, "step": 26859 }, { "epoch": 67.15, "grad_norm": 5.90625, "learning_rate": 5.145610211022738e-06, "loss": 1.1887, "step": 26860 }, { "epoch": 67.1525, "grad_norm": 7.4375, "learning_rate": 5.144902339837437e-06, "loss": 1.3166, "step": 26861 }, { "epoch": 67.155, "grad_norm": 6.71875, "learning_rate": 5.144194500481961e-06, "loss": 1.3556, "step": 26862 }, { "epoch": 67.1575, "grad_norm": 8.5625, "learning_rate": 5.143486692960955e-06, "loss": 1.3412, "step": 26863 }, { "epoch": 67.16, "grad_norm": 8.5625, "learning_rate": 5.1427789172790565e-06, "loss": 1.1665, "step": 26864 }, { "epoch": 67.1625, "grad_norm": 7.0625, "learning_rate": 5.1420711734409e-06, "loss": 1.347, "step": 26865 }, { "epoch": 67.165, "grad_norm": 6.25, "learning_rate": 5.141363461451136e-06, "loss": 1.4785, "step": 26866 }, { "epoch": 67.1675, "grad_norm": 8.625, "learning_rate": 5.140655781314394e-06, "loss": 1.2455, "step": 26867 }, { "epoch": 67.17, "grad_norm": 7.125, "learning_rate": 5.139948133035322e-06, "loss": 1.2995, "step": 26868 }, { "epoch": 67.1725, "grad_norm": 8.75, "learning_rate": 5.139240516618551e-06, "loss": 1.481, "step": 26869 }, { "epoch": 67.175, "grad_norm": 9.5625, "learning_rate": 5.1385329320687296e-06, "loss": 1.3546, "step": 26870 }, { "epoch": 67.1775, "grad_norm": 15.3125, "learning_rate": 5.137825379390492e-06, "loss": 1.3743, "step": 26871 }, { "epoch": 67.18, "grad_norm": 7.96875, "learning_rate": 5.137117858588472e-06, "loss": 1.2909, "step": 26872 }, { "epoch": 67.1825, "grad_norm": 8.0625, "learning_rate": 5.136410369667313e-06, "loss": 1.3131, "step": 26873 }, { "epoch": 67.185, "grad_norm": 8.6875, "learning_rate": 5.135702912631657e-06, "loss": 1.4508, "step": 26874 }, { "epoch": 67.1875, "grad_norm": 15.9375, "learning_rate": 5.134995487486139e-06, "loss": 1.4049, "step": 26875 }, { "epoch": 67.19, "grad_norm": 5.9375, "learning_rate": 5.13428809423539e-06, "loss": 1.4624, "step": 26876 }, { "epoch": 67.1925, "grad_norm": 7.25, "learning_rate": 5.133580732884054e-06, "loss": 1.4812, "step": 26877 }, { "epoch": 67.195, "grad_norm": 7.25, "learning_rate": 5.132873403436773e-06, "loss": 1.3986, "step": 26878 }, { "epoch": 67.1975, "grad_norm": 8.5, "learning_rate": 5.132166105898178e-06, "loss": 1.2269, "step": 26879 }, { "epoch": 67.2, "grad_norm": 7.6875, "learning_rate": 5.131458840272905e-06, "loss": 1.1815, "step": 26880 }, { "epoch": 67.2025, "grad_norm": 8.0, "learning_rate": 5.130751606565595e-06, "loss": 1.2453, "step": 26881 }, { "epoch": 67.205, "grad_norm": 7.03125, "learning_rate": 5.130044404780879e-06, "loss": 1.3202, "step": 26882 }, { "epoch": 67.2075, "grad_norm": 8.1875, "learning_rate": 5.129337234923402e-06, "loss": 1.3299, "step": 26883 }, { "epoch": 67.21, "grad_norm": 7.0, "learning_rate": 5.128630096997791e-06, "loss": 1.2881, "step": 26884 }, { "epoch": 67.2125, "grad_norm": 8.3125, "learning_rate": 5.1279229910086894e-06, "loss": 1.4258, "step": 26885 }, { "epoch": 67.215, "grad_norm": 6.46875, "learning_rate": 5.1272159169607305e-06, "loss": 1.3418, "step": 26886 }, { "epoch": 67.2175, "grad_norm": 6.75, "learning_rate": 5.126508874858545e-06, "loss": 1.2794, "step": 26887 }, { "epoch": 67.22, "grad_norm": 5.9375, "learning_rate": 5.125801864706771e-06, "loss": 1.2341, "step": 26888 }, { "epoch": 67.2225, "grad_norm": 6.28125, "learning_rate": 5.125094886510049e-06, "loss": 1.4224, "step": 26889 }, { "epoch": 67.225, "grad_norm": 9.4375, "learning_rate": 5.124387940273011e-06, "loss": 1.4516, "step": 26890 }, { "epoch": 67.2275, "grad_norm": 10.375, "learning_rate": 5.1236810260002865e-06, "loss": 1.2855, "step": 26891 }, { "epoch": 67.23, "grad_norm": 6.125, "learning_rate": 5.122974143696513e-06, "loss": 1.303, "step": 26892 }, { "epoch": 67.2325, "grad_norm": 7.4375, "learning_rate": 5.122267293366332e-06, "loss": 1.4123, "step": 26893 }, { "epoch": 67.235, "grad_norm": 9.0625, "learning_rate": 5.121560475014369e-06, "loss": 1.3224, "step": 26894 }, { "epoch": 67.2375, "grad_norm": 11.6875, "learning_rate": 5.120853688645258e-06, "loss": 1.4635, "step": 26895 }, { "epoch": 67.24, "grad_norm": 8.5625, "learning_rate": 5.120146934263638e-06, "loss": 1.3209, "step": 26896 }, { "epoch": 67.2425, "grad_norm": 6.90625, "learning_rate": 5.119440211874136e-06, "loss": 1.2278, "step": 26897 }, { "epoch": 67.245, "grad_norm": 6.625, "learning_rate": 5.1187335214813914e-06, "loss": 1.3553, "step": 26898 }, { "epoch": 67.2475, "grad_norm": 8.375, "learning_rate": 5.118026863090032e-06, "loss": 1.1973, "step": 26899 }, { "epoch": 67.25, "grad_norm": 8.8125, "learning_rate": 5.117320236704697e-06, "loss": 1.3524, "step": 26900 }, { "epoch": 67.2525, "grad_norm": 8.0625, "learning_rate": 5.116613642330015e-06, "loss": 1.3691, "step": 26901 }, { "epoch": 67.255, "grad_norm": 5.9375, "learning_rate": 5.115907079970615e-06, "loss": 1.268, "step": 26902 }, { "epoch": 67.2575, "grad_norm": 6.21875, "learning_rate": 5.115200549631131e-06, "loss": 1.2322, "step": 26903 }, { "epoch": 67.26, "grad_norm": 7.53125, "learning_rate": 5.114494051316201e-06, "loss": 1.3408, "step": 26904 }, { "epoch": 67.2625, "grad_norm": 8.4375, "learning_rate": 5.1137875850304545e-06, "loss": 1.239, "step": 26905 }, { "epoch": 67.265, "grad_norm": 8.8125, "learning_rate": 5.113081150778515e-06, "loss": 1.3613, "step": 26906 }, { "epoch": 67.2675, "grad_norm": 7.78125, "learning_rate": 5.112374748565021e-06, "loss": 1.3032, "step": 26907 }, { "epoch": 67.27, "grad_norm": 8.125, "learning_rate": 5.1116683783946055e-06, "loss": 1.1495, "step": 26908 }, { "epoch": 67.2725, "grad_norm": 8.0, "learning_rate": 5.110962040271897e-06, "loss": 1.293, "step": 26909 }, { "epoch": 67.275, "grad_norm": 8.0, "learning_rate": 5.1102557342015224e-06, "loss": 1.3841, "step": 26910 }, { "epoch": 67.2775, "grad_norm": 9.25, "learning_rate": 5.109549460188119e-06, "loss": 1.2215, "step": 26911 }, { "epoch": 67.28, "grad_norm": 8.75, "learning_rate": 5.10884321823631e-06, "loss": 1.4627, "step": 26912 }, { "epoch": 67.2825, "grad_norm": 5.96875, "learning_rate": 5.108137008350733e-06, "loss": 1.1874, "step": 26913 }, { "epoch": 67.285, "grad_norm": 6.9375, "learning_rate": 5.10743083053601e-06, "loss": 1.3766, "step": 26914 }, { "epoch": 67.2875, "grad_norm": 7.3125, "learning_rate": 5.1067246847967776e-06, "loss": 1.3864, "step": 26915 }, { "epoch": 67.29, "grad_norm": 7.75, "learning_rate": 5.106018571137663e-06, "loss": 1.334, "step": 26916 }, { "epoch": 67.2925, "grad_norm": 6.90625, "learning_rate": 5.105312489563292e-06, "loss": 1.3573, "step": 26917 }, { "epoch": 67.295, "grad_norm": 6.96875, "learning_rate": 5.1046064400782945e-06, "loss": 1.195, "step": 26918 }, { "epoch": 67.2975, "grad_norm": 5.5625, "learning_rate": 5.103900422687307e-06, "loss": 1.2378, "step": 26919 }, { "epoch": 67.3, "grad_norm": 6.0625, "learning_rate": 5.103194437394952e-06, "loss": 1.4103, "step": 26920 }, { "epoch": 67.3025, "grad_norm": 6.78125, "learning_rate": 5.102488484205854e-06, "loss": 1.3095, "step": 26921 }, { "epoch": 67.305, "grad_norm": 5.1875, "learning_rate": 5.1017825631246455e-06, "loss": 1.2669, "step": 26922 }, { "epoch": 67.3075, "grad_norm": 9.0, "learning_rate": 5.101076674155958e-06, "loss": 1.2602, "step": 26923 }, { "epoch": 67.31, "grad_norm": 7.90625, "learning_rate": 5.100370817304417e-06, "loss": 1.3246, "step": 26924 }, { "epoch": 67.3125, "grad_norm": 6.78125, "learning_rate": 5.099664992574645e-06, "loss": 1.3727, "step": 26925 }, { "epoch": 67.315, "grad_norm": 6.6875, "learning_rate": 5.0989591999712754e-06, "loss": 1.2475, "step": 26926 }, { "epoch": 67.3175, "grad_norm": 7.03125, "learning_rate": 5.0982534394989305e-06, "loss": 1.2263, "step": 26927 }, { "epoch": 67.32, "grad_norm": 9.25, "learning_rate": 5.097547711162243e-06, "loss": 1.5376, "step": 26928 }, { "epoch": 67.3225, "grad_norm": 6.8125, "learning_rate": 5.096842014965833e-06, "loss": 1.2933, "step": 26929 }, { "epoch": 67.325, "grad_norm": 6.53125, "learning_rate": 5.096136350914335e-06, "loss": 1.3936, "step": 26930 }, { "epoch": 67.3275, "grad_norm": 7.375, "learning_rate": 5.09543071901237e-06, "loss": 1.4571, "step": 26931 }, { "epoch": 67.33, "grad_norm": 7.65625, "learning_rate": 5.0947251192645604e-06, "loss": 1.2435, "step": 26932 }, { "epoch": 67.3325, "grad_norm": 8.375, "learning_rate": 5.0940195516755386e-06, "loss": 1.1598, "step": 26933 }, { "epoch": 67.335, "grad_norm": 6.96875, "learning_rate": 5.093314016249931e-06, "loss": 1.3867, "step": 26934 }, { "epoch": 67.3375, "grad_norm": 6.6875, "learning_rate": 5.092608512992361e-06, "loss": 1.3693, "step": 26935 }, { "epoch": 67.34, "grad_norm": 7.65625, "learning_rate": 5.091903041907449e-06, "loss": 1.2645, "step": 26936 }, { "epoch": 67.3425, "grad_norm": 6.625, "learning_rate": 5.091197602999824e-06, "loss": 1.388, "step": 26937 }, { "epoch": 67.345, "grad_norm": 8.375, "learning_rate": 5.090492196274115e-06, "loss": 1.234, "step": 26938 }, { "epoch": 67.3475, "grad_norm": 6.875, "learning_rate": 5.089786821734944e-06, "loss": 1.1827, "step": 26939 }, { "epoch": 67.35, "grad_norm": 8.0625, "learning_rate": 5.089081479386928e-06, "loss": 1.2707, "step": 26940 }, { "epoch": 67.3525, "grad_norm": 6.125, "learning_rate": 5.0883761692347015e-06, "loss": 1.4586, "step": 26941 }, { "epoch": 67.355, "grad_norm": 6.5625, "learning_rate": 5.087670891282881e-06, "loss": 1.3557, "step": 26942 }, { "epoch": 67.3575, "grad_norm": 6.65625, "learning_rate": 5.086965645536097e-06, "loss": 1.3515, "step": 26943 }, { "epoch": 67.36, "grad_norm": 9.125, "learning_rate": 5.086260431998967e-06, "loss": 1.3515, "step": 26944 }, { "epoch": 67.3625, "grad_norm": 7.96875, "learning_rate": 5.08555525067612e-06, "loss": 1.323, "step": 26945 }, { "epoch": 67.365, "grad_norm": 8.0, "learning_rate": 5.084850101572176e-06, "loss": 1.3376, "step": 26946 }, { "epoch": 67.3675, "grad_norm": 9.5, "learning_rate": 5.084144984691755e-06, "loss": 1.243, "step": 26947 }, { "epoch": 67.37, "grad_norm": 8.1875, "learning_rate": 5.083439900039482e-06, "loss": 1.373, "step": 26948 }, { "epoch": 67.3725, "grad_norm": 6.625, "learning_rate": 5.082734847619984e-06, "loss": 1.2451, "step": 26949 }, { "epoch": 67.375, "grad_norm": 5.5, "learning_rate": 5.08202982743788e-06, "loss": 1.2733, "step": 26950 }, { "epoch": 67.3775, "grad_norm": 7.6875, "learning_rate": 5.081324839497788e-06, "loss": 1.2173, "step": 26951 }, { "epoch": 67.38, "grad_norm": 5.625, "learning_rate": 5.080619883804333e-06, "loss": 1.2848, "step": 26952 }, { "epoch": 67.3825, "grad_norm": 6.625, "learning_rate": 5.079914960362141e-06, "loss": 1.3485, "step": 26953 }, { "epoch": 67.385, "grad_norm": 7.6875, "learning_rate": 5.079210069175831e-06, "loss": 1.2847, "step": 26954 }, { "epoch": 67.3875, "grad_norm": 8.5625, "learning_rate": 5.078505210250017e-06, "loss": 1.4494, "step": 26955 }, { "epoch": 67.39, "grad_norm": 7.8125, "learning_rate": 5.077800383589331e-06, "loss": 1.3234, "step": 26956 }, { "epoch": 67.3925, "grad_norm": 11.125, "learning_rate": 5.0770955891983845e-06, "loss": 1.3735, "step": 26957 }, { "epoch": 67.395, "grad_norm": 8.375, "learning_rate": 5.076390827081805e-06, "loss": 1.245, "step": 26958 }, { "epoch": 67.3975, "grad_norm": 8.4375, "learning_rate": 5.075686097244208e-06, "loss": 1.296, "step": 26959 }, { "epoch": 67.4, "grad_norm": 7.0, "learning_rate": 5.074981399690219e-06, "loss": 1.443, "step": 26960 }, { "epoch": 67.4025, "grad_norm": 9.625, "learning_rate": 5.0742767344244545e-06, "loss": 1.3188, "step": 26961 }, { "epoch": 67.405, "grad_norm": 9.25, "learning_rate": 5.073572101451532e-06, "loss": 1.3567, "step": 26962 }, { "epoch": 67.4075, "grad_norm": 6.5, "learning_rate": 5.072867500776073e-06, "loss": 1.3731, "step": 26963 }, { "epoch": 67.41, "grad_norm": 7.8125, "learning_rate": 5.072162932402701e-06, "loss": 1.2708, "step": 26964 }, { "epoch": 67.4125, "grad_norm": 7.71875, "learning_rate": 5.071458396336032e-06, "loss": 1.3034, "step": 26965 }, { "epoch": 67.415, "grad_norm": 8.5, "learning_rate": 5.07075389258068e-06, "loss": 1.3052, "step": 26966 }, { "epoch": 67.4175, "grad_norm": 10.5, "learning_rate": 5.07004942114127e-06, "loss": 1.2833, "step": 26967 }, { "epoch": 67.42, "grad_norm": 6.09375, "learning_rate": 5.069344982022421e-06, "loss": 1.3249, "step": 26968 }, { "epoch": 67.4225, "grad_norm": 7.1875, "learning_rate": 5.068640575228749e-06, "loss": 1.2692, "step": 26969 }, { "epoch": 67.425, "grad_norm": 6.6875, "learning_rate": 5.067936200764869e-06, "loss": 1.3334, "step": 26970 }, { "epoch": 67.4275, "grad_norm": 11.125, "learning_rate": 5.067231858635404e-06, "loss": 1.3946, "step": 26971 }, { "epoch": 67.43, "grad_norm": 6.71875, "learning_rate": 5.066527548844968e-06, "loss": 1.3101, "step": 26972 }, { "epoch": 67.4325, "grad_norm": 7.59375, "learning_rate": 5.065823271398183e-06, "loss": 1.2485, "step": 26973 }, { "epoch": 67.435, "grad_norm": 10.0625, "learning_rate": 5.0651190262996584e-06, "loss": 1.37, "step": 26974 }, { "epoch": 67.4375, "grad_norm": 9.0, "learning_rate": 5.064414813554022e-06, "loss": 1.374, "step": 26975 }, { "epoch": 67.44, "grad_norm": 9.8125, "learning_rate": 5.0637106331658815e-06, "loss": 1.354, "step": 26976 }, { "epoch": 67.4425, "grad_norm": 6.09375, "learning_rate": 5.063006485139854e-06, "loss": 1.4658, "step": 26977 }, { "epoch": 67.445, "grad_norm": 6.71875, "learning_rate": 5.062302369480559e-06, "loss": 1.2479, "step": 26978 }, { "epoch": 67.4475, "grad_norm": 8.0625, "learning_rate": 5.061598286192616e-06, "loss": 1.1977, "step": 26979 }, { "epoch": 67.45, "grad_norm": 5.6875, "learning_rate": 5.060894235280637e-06, "loss": 1.2367, "step": 26980 }, { "epoch": 67.4525, "grad_norm": 7.15625, "learning_rate": 5.0601902167492366e-06, "loss": 1.3487, "step": 26981 }, { "epoch": 67.455, "grad_norm": 6.65625, "learning_rate": 5.0594862306030254e-06, "loss": 1.5379, "step": 26982 }, { "epoch": 67.4575, "grad_norm": 7.5, "learning_rate": 5.0587822768466325e-06, "loss": 1.2602, "step": 26983 }, { "epoch": 67.46, "grad_norm": 8.1875, "learning_rate": 5.058078355484663e-06, "loss": 1.291, "step": 26984 }, { "epoch": 67.4625, "grad_norm": 7.5, "learning_rate": 5.057374466521736e-06, "loss": 1.2954, "step": 26985 }, { "epoch": 67.465, "grad_norm": 11.125, "learning_rate": 5.05667060996246e-06, "loss": 1.3433, "step": 26986 }, { "epoch": 67.4675, "grad_norm": 10.3125, "learning_rate": 5.055966785811452e-06, "loss": 1.5258, "step": 26987 }, { "epoch": 67.47, "grad_norm": 12.0625, "learning_rate": 5.055262994073334e-06, "loss": 1.4542, "step": 26988 }, { "epoch": 67.4725, "grad_norm": 8.6875, "learning_rate": 5.05455923475271e-06, "loss": 1.4704, "step": 26989 }, { "epoch": 67.475, "grad_norm": 8.75, "learning_rate": 5.053855507854201e-06, "loss": 1.4077, "step": 26990 }, { "epoch": 67.4775, "grad_norm": 9.875, "learning_rate": 5.053151813382417e-06, "loss": 1.4113, "step": 26991 }, { "epoch": 67.48, "grad_norm": 9.9375, "learning_rate": 5.0524481513419675e-06, "loss": 1.4152, "step": 26992 }, { "epoch": 67.4825, "grad_norm": 6.78125, "learning_rate": 5.051744521737471e-06, "loss": 1.2181, "step": 26993 }, { "epoch": 67.485, "grad_norm": 6.875, "learning_rate": 5.051040924573542e-06, "loss": 1.2522, "step": 26994 }, { "epoch": 67.4875, "grad_norm": 7.4375, "learning_rate": 5.050337359854792e-06, "loss": 1.2491, "step": 26995 }, { "epoch": 67.49, "grad_norm": 7.125, "learning_rate": 5.049633827585832e-06, "loss": 1.2531, "step": 26996 }, { "epoch": 67.4925, "grad_norm": 7.125, "learning_rate": 5.048930327771266e-06, "loss": 1.3294, "step": 26997 }, { "epoch": 67.495, "grad_norm": 6.71875, "learning_rate": 5.048226860415724e-06, "loss": 1.2583, "step": 26998 }, { "epoch": 67.4975, "grad_norm": 6.625, "learning_rate": 5.047523425523806e-06, "loss": 1.2418, "step": 26999 }, { "epoch": 67.5, "grad_norm": 7.625, "learning_rate": 5.046820023100129e-06, "loss": 1.3348, "step": 27000 }, { "epoch": 67.5025, "grad_norm": 7.0625, "learning_rate": 5.0461166531492955e-06, "loss": 1.1439, "step": 27001 }, { "epoch": 67.505, "grad_norm": 7.25, "learning_rate": 5.045413315675925e-06, "loss": 1.4208, "step": 27002 }, { "epoch": 67.5075, "grad_norm": 8.25, "learning_rate": 5.044710010684628e-06, "loss": 1.2997, "step": 27003 }, { "epoch": 67.51, "grad_norm": 7.625, "learning_rate": 5.0440067381800165e-06, "loss": 1.3061, "step": 27004 }, { "epoch": 67.5125, "grad_norm": 8.625, "learning_rate": 5.043303498166693e-06, "loss": 1.2631, "step": 27005 }, { "epoch": 67.515, "grad_norm": 8.75, "learning_rate": 5.042600290649278e-06, "loss": 1.3853, "step": 27006 }, { "epoch": 67.5175, "grad_norm": 9.5625, "learning_rate": 5.0418971156323725e-06, "loss": 1.3727, "step": 27007 }, { "epoch": 67.52, "grad_norm": 6.71875, "learning_rate": 5.041193973120595e-06, "loss": 1.3024, "step": 27008 }, { "epoch": 67.5225, "grad_norm": 6.5, "learning_rate": 5.040490863118548e-06, "loss": 1.3545, "step": 27009 }, { "epoch": 67.525, "grad_norm": 6.71875, "learning_rate": 5.039787785630848e-06, "loss": 1.3072, "step": 27010 }, { "epoch": 67.5275, "grad_norm": 7.78125, "learning_rate": 5.039084740662101e-06, "loss": 1.3529, "step": 27011 }, { "epoch": 67.53, "grad_norm": 8.375, "learning_rate": 5.038381728216912e-06, "loss": 1.2652, "step": 27012 }, { "epoch": 67.5325, "grad_norm": 12.3125, "learning_rate": 5.037678748299893e-06, "loss": 1.3724, "step": 27013 }, { "epoch": 67.535, "grad_norm": 5.78125, "learning_rate": 5.036975800915659e-06, "loss": 1.1596, "step": 27014 }, { "epoch": 67.5375, "grad_norm": 5.84375, "learning_rate": 5.036272886068812e-06, "loss": 1.3, "step": 27015 }, { "epoch": 67.54, "grad_norm": 8.4375, "learning_rate": 5.035570003763958e-06, "loss": 1.1695, "step": 27016 }, { "epoch": 67.5425, "grad_norm": 6.78125, "learning_rate": 5.034867154005708e-06, "loss": 1.4111, "step": 27017 }, { "epoch": 67.545, "grad_norm": 5.53125, "learning_rate": 5.034164336798674e-06, "loss": 1.2285, "step": 27018 }, { "epoch": 67.5475, "grad_norm": 8.0, "learning_rate": 5.0334615521474604e-06, "loss": 1.2943, "step": 27019 }, { "epoch": 67.55, "grad_norm": 6.0625, "learning_rate": 5.03275880005667e-06, "loss": 1.3615, "step": 27020 }, { "epoch": 67.5525, "grad_norm": 7.8125, "learning_rate": 5.032056080530918e-06, "loss": 1.337, "step": 27021 }, { "epoch": 67.555, "grad_norm": 5.84375, "learning_rate": 5.031353393574803e-06, "loss": 1.371, "step": 27022 }, { "epoch": 67.5575, "grad_norm": 10.9375, "learning_rate": 5.0306507391929415e-06, "loss": 1.4589, "step": 27023 }, { "epoch": 67.56, "grad_norm": 7.6875, "learning_rate": 5.02994811738993e-06, "loss": 1.333, "step": 27024 }, { "epoch": 67.5625, "grad_norm": 6.125, "learning_rate": 5.029245528170383e-06, "loss": 1.3794, "step": 27025 }, { "epoch": 67.565, "grad_norm": 9.4375, "learning_rate": 5.028542971538906e-06, "loss": 1.2299, "step": 27026 }, { "epoch": 67.5675, "grad_norm": 10.6875, "learning_rate": 5.027840447500096e-06, "loss": 1.3237, "step": 27027 }, { "epoch": 67.57, "grad_norm": 10.375, "learning_rate": 5.027137956058565e-06, "loss": 1.3852, "step": 27028 }, { "epoch": 67.5725, "grad_norm": 6.5, "learning_rate": 5.026435497218924e-06, "loss": 1.1828, "step": 27029 }, { "epoch": 67.575, "grad_norm": 6.3125, "learning_rate": 5.025733070985772e-06, "loss": 1.3391, "step": 27030 }, { "epoch": 67.5775, "grad_norm": 8.25, "learning_rate": 5.025030677363711e-06, "loss": 1.4069, "step": 27031 }, { "epoch": 67.58, "grad_norm": 9.75, "learning_rate": 5.02432831635735e-06, "loss": 1.146, "step": 27032 }, { "epoch": 67.5825, "grad_norm": 7.15625, "learning_rate": 5.023625987971298e-06, "loss": 1.1885, "step": 27033 }, { "epoch": 67.585, "grad_norm": 9.8125, "learning_rate": 5.022923692210153e-06, "loss": 1.4674, "step": 27034 }, { "epoch": 67.5875, "grad_norm": 8.875, "learning_rate": 5.02222142907852e-06, "loss": 1.2565, "step": 27035 }, { "epoch": 67.59, "grad_norm": 6.21875, "learning_rate": 5.021519198581002e-06, "loss": 1.2743, "step": 27036 }, { "epoch": 67.5925, "grad_norm": 10.125, "learning_rate": 5.020817000722209e-06, "loss": 1.2661, "step": 27037 }, { "epoch": 67.595, "grad_norm": 8.4375, "learning_rate": 5.020114835506741e-06, "loss": 1.4034, "step": 27038 }, { "epoch": 67.5975, "grad_norm": 7.90625, "learning_rate": 5.019412702939197e-06, "loss": 1.3708, "step": 27039 }, { "epoch": 67.6, "grad_norm": 7.625, "learning_rate": 5.018710603024187e-06, "loss": 1.2775, "step": 27040 }, { "epoch": 67.6025, "grad_norm": 5.78125, "learning_rate": 5.0180085357663075e-06, "loss": 1.3184, "step": 27041 }, { "epoch": 67.605, "grad_norm": 10.25, "learning_rate": 5.01730650117017e-06, "loss": 1.3452, "step": 27042 }, { "epoch": 67.6075, "grad_norm": 5.71875, "learning_rate": 5.016604499240365e-06, "loss": 1.2115, "step": 27043 }, { "epoch": 67.61, "grad_norm": 4.9375, "learning_rate": 5.015902529981508e-06, "loss": 1.2693, "step": 27044 }, { "epoch": 67.6125, "grad_norm": 5.875, "learning_rate": 5.0152005933981925e-06, "loss": 1.4089, "step": 27045 }, { "epoch": 67.615, "grad_norm": 6.40625, "learning_rate": 5.014498689495019e-06, "loss": 1.5037, "step": 27046 }, { "epoch": 67.6175, "grad_norm": 5.875, "learning_rate": 5.013796818276595e-06, "loss": 1.2759, "step": 27047 }, { "epoch": 67.62, "grad_norm": 9.75, "learning_rate": 5.013094979747522e-06, "loss": 1.5353, "step": 27048 }, { "epoch": 67.6225, "grad_norm": 5.8125, "learning_rate": 5.012393173912398e-06, "loss": 1.4132, "step": 27049 }, { "epoch": 67.625, "grad_norm": 6.625, "learning_rate": 5.01169140077582e-06, "loss": 1.4833, "step": 27050 }, { "epoch": 67.6275, "grad_norm": 8.5, "learning_rate": 5.0109896603423954e-06, "loss": 1.4924, "step": 27051 }, { "epoch": 67.63, "grad_norm": 7.625, "learning_rate": 5.010287952616726e-06, "loss": 1.3251, "step": 27052 }, { "epoch": 67.6325, "grad_norm": 9.875, "learning_rate": 5.009586277603409e-06, "loss": 1.2364, "step": 27053 }, { "epoch": 67.635, "grad_norm": 5.4375, "learning_rate": 5.008884635307041e-06, "loss": 1.336, "step": 27054 }, { "epoch": 67.6375, "grad_norm": 7.625, "learning_rate": 5.00818302573223e-06, "loss": 1.2587, "step": 27055 }, { "epoch": 67.64, "grad_norm": 8.3125, "learning_rate": 5.007481448883567e-06, "loss": 1.1598, "step": 27056 }, { "epoch": 67.6425, "grad_norm": 10.375, "learning_rate": 5.006779904765659e-06, "loss": 1.3573, "step": 27057 }, { "epoch": 67.645, "grad_norm": 8.1875, "learning_rate": 5.006078393383098e-06, "loss": 1.2817, "step": 27058 }, { "epoch": 67.6475, "grad_norm": 7.1875, "learning_rate": 5.005376914740493e-06, "loss": 1.3773, "step": 27059 }, { "epoch": 67.65, "grad_norm": 7.84375, "learning_rate": 5.004675468842436e-06, "loss": 1.2305, "step": 27060 }, { "epoch": 67.6525, "grad_norm": 4.4375, "learning_rate": 5.003974055693524e-06, "loss": 1.418, "step": 27061 }, { "epoch": 67.655, "grad_norm": 7.71875, "learning_rate": 5.0032726752983565e-06, "loss": 1.2538, "step": 27062 }, { "epoch": 67.6575, "grad_norm": 6.90625, "learning_rate": 5.002571327661537e-06, "loss": 1.2624, "step": 27063 }, { "epoch": 67.66, "grad_norm": 7.78125, "learning_rate": 5.0018700127876606e-06, "loss": 1.3592, "step": 27064 }, { "epoch": 67.6625, "grad_norm": 9.1875, "learning_rate": 5.001168730681321e-06, "loss": 1.31, "step": 27065 }, { "epoch": 67.665, "grad_norm": 8.1875, "learning_rate": 5.000467481347119e-06, "loss": 1.2691, "step": 27066 }, { "epoch": 67.6675, "grad_norm": 5.5625, "learning_rate": 4.999766264789655e-06, "loss": 1.2938, "step": 27067 }, { "epoch": 67.67, "grad_norm": 9.375, "learning_rate": 4.9990650810135246e-06, "loss": 1.3872, "step": 27068 }, { "epoch": 67.6725, "grad_norm": 6.3125, "learning_rate": 4.9983639300233185e-06, "loss": 1.3356, "step": 27069 }, { "epoch": 67.675, "grad_norm": 6.84375, "learning_rate": 4.9976628118236425e-06, "loss": 1.4333, "step": 27070 }, { "epoch": 67.6775, "grad_norm": 12.25, "learning_rate": 4.996961726419085e-06, "loss": 1.3484, "step": 27071 }, { "epoch": 67.68, "grad_norm": 5.90625, "learning_rate": 4.99626067381425e-06, "loss": 1.1971, "step": 27072 }, { "epoch": 67.6825, "grad_norm": 7.09375, "learning_rate": 4.9955596540137275e-06, "loss": 1.381, "step": 27073 }, { "epoch": 67.685, "grad_norm": 7.09375, "learning_rate": 4.994858667022118e-06, "loss": 1.3087, "step": 27074 }, { "epoch": 67.6875, "grad_norm": 7.34375, "learning_rate": 4.9941577128440144e-06, "loss": 1.277, "step": 27075 }, { "epoch": 67.69, "grad_norm": 7.125, "learning_rate": 4.993456791484009e-06, "loss": 1.3113, "step": 27076 }, { "epoch": 67.6925, "grad_norm": 8.3125, "learning_rate": 4.992755902946701e-06, "loss": 1.2418, "step": 27077 }, { "epoch": 67.695, "grad_norm": 8.0, "learning_rate": 4.992055047236689e-06, "loss": 1.4099, "step": 27078 }, { "epoch": 67.6975, "grad_norm": 6.65625, "learning_rate": 4.991354224358564e-06, "loss": 1.286, "step": 27079 }, { "epoch": 67.7, "grad_norm": 7.40625, "learning_rate": 4.990653434316915e-06, "loss": 1.4256, "step": 27080 }, { "epoch": 67.7025, "grad_norm": 6.875, "learning_rate": 4.989952677116342e-06, "loss": 1.369, "step": 27081 }, { "epoch": 67.705, "grad_norm": 6.0, "learning_rate": 4.989251952761444e-06, "loss": 1.2943, "step": 27082 }, { "epoch": 67.7075, "grad_norm": 9.875, "learning_rate": 4.988551261256808e-06, "loss": 1.3925, "step": 27083 }, { "epoch": 67.71, "grad_norm": 9.0625, "learning_rate": 4.9878506026070265e-06, "loss": 1.2953, "step": 27084 }, { "epoch": 67.7125, "grad_norm": 5.90625, "learning_rate": 4.987149976816698e-06, "loss": 1.252, "step": 27085 }, { "epoch": 67.715, "grad_norm": 10.375, "learning_rate": 4.9864493838904115e-06, "loss": 1.3501, "step": 27086 }, { "epoch": 67.7175, "grad_norm": 8.0625, "learning_rate": 4.985748823832766e-06, "loss": 1.3912, "step": 27087 }, { "epoch": 67.72, "grad_norm": 9.75, "learning_rate": 4.985048296648346e-06, "loss": 1.3563, "step": 27088 }, { "epoch": 67.7225, "grad_norm": 7.125, "learning_rate": 4.9843478023417535e-06, "loss": 1.2521, "step": 27089 }, { "epoch": 67.725, "grad_norm": 7.5625, "learning_rate": 4.983647340917576e-06, "loss": 1.3767, "step": 27090 }, { "epoch": 67.7275, "grad_norm": 5.21875, "learning_rate": 4.982946912380401e-06, "loss": 1.2614, "step": 27091 }, { "epoch": 67.73, "grad_norm": 10.5625, "learning_rate": 4.982246516734826e-06, "loss": 1.386, "step": 27092 }, { "epoch": 67.7325, "grad_norm": 7.59375, "learning_rate": 4.981546153985445e-06, "loss": 1.2151, "step": 27093 }, { "epoch": 67.735, "grad_norm": 8.0625, "learning_rate": 4.9808458241368475e-06, "loss": 1.2775, "step": 27094 }, { "epoch": 67.7375, "grad_norm": 7.625, "learning_rate": 4.9801455271936195e-06, "loss": 1.2276, "step": 27095 }, { "epoch": 67.74, "grad_norm": 5.875, "learning_rate": 4.979445263160357e-06, "loss": 1.2725, "step": 27096 }, { "epoch": 67.7425, "grad_norm": 7.0625, "learning_rate": 4.978745032041654e-06, "loss": 1.4056, "step": 27097 }, { "epoch": 67.745, "grad_norm": 8.25, "learning_rate": 4.978044833842098e-06, "loss": 1.3047, "step": 27098 }, { "epoch": 67.7475, "grad_norm": 7.5, "learning_rate": 4.977344668566275e-06, "loss": 1.2378, "step": 27099 }, { "epoch": 67.75, "grad_norm": 8.125, "learning_rate": 4.976644536218783e-06, "loss": 1.2472, "step": 27100 }, { "epoch": 67.7525, "grad_norm": 6.84375, "learning_rate": 4.975944436804205e-06, "loss": 1.1048, "step": 27101 }, { "epoch": 67.755, "grad_norm": 9.125, "learning_rate": 4.975244370327138e-06, "loss": 1.4428, "step": 27102 }, { "epoch": 67.7575, "grad_norm": 6.25, "learning_rate": 4.974544336792163e-06, "loss": 1.2306, "step": 27103 }, { "epoch": 67.76, "grad_norm": 8.4375, "learning_rate": 4.973844336203879e-06, "loss": 1.3258, "step": 27104 }, { "epoch": 67.7625, "grad_norm": 6.28125, "learning_rate": 4.97314436856687e-06, "loss": 1.2711, "step": 27105 }, { "epoch": 67.765, "grad_norm": 7.3125, "learning_rate": 4.972444433885723e-06, "loss": 1.3507, "step": 27106 }, { "epoch": 67.7675, "grad_norm": 8.1875, "learning_rate": 4.971744532165027e-06, "loss": 1.2744, "step": 27107 }, { "epoch": 67.77, "grad_norm": 6.96875, "learning_rate": 4.971044663409378e-06, "loss": 1.416, "step": 27108 }, { "epoch": 67.7725, "grad_norm": 6.40625, "learning_rate": 4.9703448276233575e-06, "loss": 1.4514, "step": 27109 }, { "epoch": 67.775, "grad_norm": 7.21875, "learning_rate": 4.969645024811552e-06, "loss": 1.3281, "step": 27110 }, { "epoch": 67.7775, "grad_norm": 6.90625, "learning_rate": 4.9689452549785535e-06, "loss": 1.2039, "step": 27111 }, { "epoch": 67.78, "grad_norm": 8.0, "learning_rate": 4.96824551812895e-06, "loss": 1.1954, "step": 27112 }, { "epoch": 67.7825, "grad_norm": 6.0625, "learning_rate": 4.967545814267329e-06, "loss": 1.2942, "step": 27113 }, { "epoch": 67.785, "grad_norm": 14.875, "learning_rate": 4.966846143398273e-06, "loss": 1.3094, "step": 27114 }, { "epoch": 67.7875, "grad_norm": 7.84375, "learning_rate": 4.9661465055263744e-06, "loss": 1.2995, "step": 27115 }, { "epoch": 67.79, "grad_norm": 10.25, "learning_rate": 4.965446900656215e-06, "loss": 1.3389, "step": 27116 }, { "epoch": 67.7925, "grad_norm": 5.21875, "learning_rate": 4.964747328792389e-06, "loss": 1.5391, "step": 27117 }, { "epoch": 67.795, "grad_norm": 7.71875, "learning_rate": 4.964047789939474e-06, "loss": 1.3079, "step": 27118 }, { "epoch": 67.7975, "grad_norm": 7.46875, "learning_rate": 4.963348284102062e-06, "loss": 1.2861, "step": 27119 }, { "epoch": 67.8, "grad_norm": 9.125, "learning_rate": 4.9626488112847384e-06, "loss": 1.3282, "step": 27120 }, { "epoch": 67.8025, "grad_norm": 9.5625, "learning_rate": 4.961949371492084e-06, "loss": 1.3219, "step": 27121 }, { "epoch": 67.805, "grad_norm": 7.78125, "learning_rate": 4.961249964728687e-06, "loss": 1.2611, "step": 27122 }, { "epoch": 67.8075, "grad_norm": 6.53125, "learning_rate": 4.960550590999138e-06, "loss": 1.3518, "step": 27123 }, { "epoch": 67.81, "grad_norm": 9.125, "learning_rate": 4.959851250308019e-06, "loss": 1.3222, "step": 27124 }, { "epoch": 67.8125, "grad_norm": 4.8125, "learning_rate": 4.959151942659911e-06, "loss": 1.3331, "step": 27125 }, { "epoch": 67.815, "grad_norm": 5.15625, "learning_rate": 4.958452668059395e-06, "loss": 1.1641, "step": 27126 }, { "epoch": 67.8175, "grad_norm": 6.4375, "learning_rate": 4.95775342651107e-06, "loss": 1.4313, "step": 27127 }, { "epoch": 67.82, "grad_norm": 9.125, "learning_rate": 4.957054218019511e-06, "loss": 1.3954, "step": 27128 }, { "epoch": 67.8225, "grad_norm": 6.84375, "learning_rate": 4.956355042589297e-06, "loss": 1.3273, "step": 27129 }, { "epoch": 67.825, "grad_norm": 6.3125, "learning_rate": 4.955655900225024e-06, "loss": 1.3722, "step": 27130 }, { "epoch": 67.8275, "grad_norm": 8.1875, "learning_rate": 4.954956790931264e-06, "loss": 1.2562, "step": 27131 }, { "epoch": 67.83, "grad_norm": 6.6875, "learning_rate": 4.9542577147126095e-06, "loss": 1.3163, "step": 27132 }, { "epoch": 67.8325, "grad_norm": 8.4375, "learning_rate": 4.953558671573635e-06, "loss": 1.3088, "step": 27133 }, { "epoch": 67.835, "grad_norm": 8.375, "learning_rate": 4.952859661518932e-06, "loss": 1.3716, "step": 27134 }, { "epoch": 67.8375, "grad_norm": 8.4375, "learning_rate": 4.952160684553079e-06, "loss": 1.2981, "step": 27135 }, { "epoch": 67.84, "grad_norm": 6.75, "learning_rate": 4.951461740680655e-06, "loss": 1.2677, "step": 27136 }, { "epoch": 67.8425, "grad_norm": 9.875, "learning_rate": 4.950762829906246e-06, "loss": 1.3002, "step": 27137 }, { "epoch": 67.845, "grad_norm": 5.3125, "learning_rate": 4.9500639522344375e-06, "loss": 1.3124, "step": 27138 }, { "epoch": 67.8475, "grad_norm": 8.125, "learning_rate": 4.9493651076698076e-06, "loss": 1.2609, "step": 27139 }, { "epoch": 67.85, "grad_norm": 7.84375, "learning_rate": 4.948666296216938e-06, "loss": 1.3629, "step": 27140 }, { "epoch": 67.8525, "grad_norm": 6.3125, "learning_rate": 4.947967517880403e-06, "loss": 1.3779, "step": 27141 }, { "epoch": 67.855, "grad_norm": 7.875, "learning_rate": 4.947268772664797e-06, "loss": 1.4945, "step": 27142 }, { "epoch": 67.8575, "grad_norm": 7.03125, "learning_rate": 4.946570060574696e-06, "loss": 1.2901, "step": 27143 }, { "epoch": 67.86, "grad_norm": 6.03125, "learning_rate": 4.945871381614679e-06, "loss": 1.3737, "step": 27144 }, { "epoch": 67.8625, "grad_norm": 7.84375, "learning_rate": 4.945172735789323e-06, "loss": 1.1112, "step": 27145 }, { "epoch": 67.865, "grad_norm": 5.25, "learning_rate": 4.944474123103212e-06, "loss": 1.1661, "step": 27146 }, { "epoch": 67.8675, "grad_norm": 8.5, "learning_rate": 4.943775543560931e-06, "loss": 1.2801, "step": 27147 }, { "epoch": 67.87, "grad_norm": 8.5, "learning_rate": 4.943076997167055e-06, "loss": 1.3012, "step": 27148 }, { "epoch": 67.8725, "grad_norm": 6.15625, "learning_rate": 4.9423784839261605e-06, "loss": 1.2736, "step": 27149 }, { "epoch": 67.875, "grad_norm": 8.375, "learning_rate": 4.9416800038428326e-06, "loss": 1.4119, "step": 27150 }, { "epoch": 67.8775, "grad_norm": 5.28125, "learning_rate": 4.940981556921645e-06, "loss": 1.2824, "step": 27151 }, { "epoch": 67.88, "grad_norm": 6.53125, "learning_rate": 4.9402831431671834e-06, "loss": 1.2859, "step": 27152 }, { "epoch": 67.8825, "grad_norm": 9.3125, "learning_rate": 4.9395847625840196e-06, "loss": 1.2714, "step": 27153 }, { "epoch": 67.885, "grad_norm": 8.125, "learning_rate": 4.93888641517674e-06, "loss": 1.3633, "step": 27154 }, { "epoch": 67.8875, "grad_norm": 10.0625, "learning_rate": 4.938188100949918e-06, "loss": 1.4208, "step": 27155 }, { "epoch": 67.89, "grad_norm": 7.40625, "learning_rate": 4.9374898199081245e-06, "loss": 1.3485, "step": 27156 }, { "epoch": 67.8925, "grad_norm": 9.375, "learning_rate": 4.936791572055953e-06, "loss": 1.172, "step": 27157 }, { "epoch": 67.895, "grad_norm": 6.6875, "learning_rate": 4.936093357397973e-06, "loss": 1.3058, "step": 27158 }, { "epoch": 67.8975, "grad_norm": 6.9375, "learning_rate": 4.935395175938762e-06, "loss": 1.3904, "step": 27159 }, { "epoch": 67.9, "grad_norm": 6.96875, "learning_rate": 4.934697027682894e-06, "loss": 1.3127, "step": 27160 }, { "epoch": 67.9025, "grad_norm": 6.625, "learning_rate": 4.9339989126349494e-06, "loss": 1.2389, "step": 27161 }, { "epoch": 67.905, "grad_norm": 10.375, "learning_rate": 4.93330083079951e-06, "loss": 1.4804, "step": 27162 }, { "epoch": 67.9075, "grad_norm": 6.21875, "learning_rate": 4.9326027821811475e-06, "loss": 1.2909, "step": 27163 }, { "epoch": 67.91, "grad_norm": 10.0, "learning_rate": 4.931904766784434e-06, "loss": 1.2887, "step": 27164 }, { "epoch": 67.9125, "grad_norm": 8.9375, "learning_rate": 4.9312067846139536e-06, "loss": 1.3901, "step": 27165 }, { "epoch": 67.915, "grad_norm": 8.0625, "learning_rate": 4.930508835674276e-06, "loss": 1.3519, "step": 27166 }, { "epoch": 67.9175, "grad_norm": 7.46875, "learning_rate": 4.929810919969982e-06, "loss": 1.1187, "step": 27167 }, { "epoch": 67.92, "grad_norm": 5.9375, "learning_rate": 4.929113037505642e-06, "loss": 1.3336, "step": 27168 }, { "epoch": 67.9225, "grad_norm": 6.75, "learning_rate": 4.928415188285837e-06, "loss": 1.418, "step": 27169 }, { "epoch": 67.925, "grad_norm": 7.84375, "learning_rate": 4.92771737231514e-06, "loss": 1.2737, "step": 27170 }, { "epoch": 67.9275, "grad_norm": 6.625, "learning_rate": 4.92701958959812e-06, "loss": 1.2321, "step": 27171 }, { "epoch": 67.93, "grad_norm": 9.0, "learning_rate": 4.926321840139357e-06, "loss": 1.4227, "step": 27172 }, { "epoch": 67.9325, "grad_norm": 9.625, "learning_rate": 4.925624123943429e-06, "loss": 1.4381, "step": 27173 }, { "epoch": 67.935, "grad_norm": 6.75, "learning_rate": 4.924926441014906e-06, "loss": 1.2526, "step": 27174 }, { "epoch": 67.9375, "grad_norm": 9.125, "learning_rate": 4.924228791358358e-06, "loss": 1.2693, "step": 27175 }, { "epoch": 67.94, "grad_norm": 5.28125, "learning_rate": 4.923531174978363e-06, "loss": 1.5426, "step": 27176 }, { "epoch": 67.9425, "grad_norm": 7.84375, "learning_rate": 4.922833591879499e-06, "loss": 1.3114, "step": 27177 }, { "epoch": 67.945, "grad_norm": 6.75, "learning_rate": 4.9221360420663345e-06, "loss": 1.2637, "step": 27178 }, { "epoch": 67.9475, "grad_norm": 11.125, "learning_rate": 4.9214385255434384e-06, "loss": 1.4205, "step": 27179 }, { "epoch": 67.95, "grad_norm": 8.0, "learning_rate": 4.9207410423153925e-06, "loss": 1.4629, "step": 27180 }, { "epoch": 67.9525, "grad_norm": 9.0625, "learning_rate": 4.920043592386762e-06, "loss": 1.2968, "step": 27181 }, { "epoch": 67.955, "grad_norm": 8.9375, "learning_rate": 4.9193461757621255e-06, "loss": 1.3763, "step": 27182 }, { "epoch": 67.9575, "grad_norm": 5.875, "learning_rate": 4.918648792446048e-06, "loss": 1.2204, "step": 27183 }, { "epoch": 67.96, "grad_norm": 8.5625, "learning_rate": 4.91795144244311e-06, "loss": 1.3497, "step": 27184 }, { "epoch": 67.9625, "grad_norm": 8.8125, "learning_rate": 4.9172541257578775e-06, "loss": 1.48, "step": 27185 }, { "epoch": 67.965, "grad_norm": 6.71875, "learning_rate": 4.916556842394922e-06, "loss": 1.3238, "step": 27186 }, { "epoch": 67.9675, "grad_norm": 6.8125, "learning_rate": 4.915859592358815e-06, "loss": 1.2295, "step": 27187 }, { "epoch": 67.97, "grad_norm": 7.0, "learning_rate": 4.9151623756541334e-06, "loss": 1.3566, "step": 27188 }, { "epoch": 67.9725, "grad_norm": 5.625, "learning_rate": 4.914465192285443e-06, "loss": 1.2855, "step": 27189 }, { "epoch": 67.975, "grad_norm": 7.03125, "learning_rate": 4.913768042257313e-06, "loss": 1.3529, "step": 27190 }, { "epoch": 67.9775, "grad_norm": 7.625, "learning_rate": 4.9130709255743145e-06, "loss": 1.336, "step": 27191 }, { "epoch": 67.98, "grad_norm": 7.0625, "learning_rate": 4.912373842241025e-06, "loss": 1.302, "step": 27192 }, { "epoch": 67.9825, "grad_norm": 5.0625, "learning_rate": 4.911676792262008e-06, "loss": 1.2496, "step": 27193 }, { "epoch": 67.985, "grad_norm": 6.46875, "learning_rate": 4.91097977564183e-06, "loss": 1.3559, "step": 27194 }, { "epoch": 67.9875, "grad_norm": 5.28125, "learning_rate": 4.910282792385069e-06, "loss": 1.4425, "step": 27195 }, { "epoch": 67.99, "grad_norm": 6.75, "learning_rate": 4.909585842496287e-06, "loss": 1.2939, "step": 27196 }, { "epoch": 67.9925, "grad_norm": 11.625, "learning_rate": 4.908888925980061e-06, "loss": 1.3132, "step": 27197 }, { "epoch": 67.995, "grad_norm": 9.8125, "learning_rate": 4.90819204284095e-06, "loss": 1.244, "step": 27198 }, { "epoch": 67.9975, "grad_norm": 8.25, "learning_rate": 4.907495193083533e-06, "loss": 1.2571, "step": 27199 }, { "epoch": 68.0, "grad_norm": 7.90625, "learning_rate": 4.9067983767123736e-06, "loss": 1.1322, "step": 27200 }, { "epoch": 68.0025, "grad_norm": 9.5, "learning_rate": 4.906101593732037e-06, "loss": 1.3471, "step": 27201 }, { "epoch": 68.005, "grad_norm": 9.375, "learning_rate": 4.905404844147094e-06, "loss": 1.1263, "step": 27202 }, { "epoch": 68.0075, "grad_norm": 5.75, "learning_rate": 4.904708127962115e-06, "loss": 1.408, "step": 27203 }, { "epoch": 68.01, "grad_norm": 5.53125, "learning_rate": 4.904011445181668e-06, "loss": 1.3222, "step": 27204 }, { "epoch": 68.0125, "grad_norm": 7.5625, "learning_rate": 4.903314795810313e-06, "loss": 1.4423, "step": 27205 }, { "epoch": 68.015, "grad_norm": 6.6875, "learning_rate": 4.902618179852624e-06, "loss": 1.4004, "step": 27206 }, { "epoch": 68.0175, "grad_norm": 7.65625, "learning_rate": 4.901921597313168e-06, "loss": 1.1749, "step": 27207 }, { "epoch": 68.02, "grad_norm": 11.3125, "learning_rate": 4.901225048196511e-06, "loss": 1.401, "step": 27208 }, { "epoch": 68.0225, "grad_norm": 7.03125, "learning_rate": 4.900528532507215e-06, "loss": 1.3089, "step": 27209 }, { "epoch": 68.025, "grad_norm": 7.65625, "learning_rate": 4.899832050249853e-06, "loss": 1.3763, "step": 27210 }, { "epoch": 68.0275, "grad_norm": 6.6875, "learning_rate": 4.899135601428983e-06, "loss": 1.2844, "step": 27211 }, { "epoch": 68.03, "grad_norm": 5.84375, "learning_rate": 4.898439186049182e-06, "loss": 1.3123, "step": 27212 }, { "epoch": 68.0325, "grad_norm": 8.25, "learning_rate": 4.8977428041150055e-06, "loss": 1.2658, "step": 27213 }, { "epoch": 68.035, "grad_norm": 8.625, "learning_rate": 4.897046455631026e-06, "loss": 1.4017, "step": 27214 }, { "epoch": 68.0375, "grad_norm": 6.15625, "learning_rate": 4.896350140601805e-06, "loss": 1.2471, "step": 27215 }, { "epoch": 68.04, "grad_norm": 7.8125, "learning_rate": 4.895653859031906e-06, "loss": 1.2319, "step": 27216 }, { "epoch": 68.0425, "grad_norm": 9.0, "learning_rate": 4.894957610925894e-06, "loss": 1.2962, "step": 27217 }, { "epoch": 68.045, "grad_norm": 6.375, "learning_rate": 4.894261396288341e-06, "loss": 1.4277, "step": 27218 }, { "epoch": 68.0475, "grad_norm": 8.125, "learning_rate": 4.893565215123806e-06, "loss": 1.4439, "step": 27219 }, { "epoch": 68.05, "grad_norm": 8.0625, "learning_rate": 4.8928690674368495e-06, "loss": 1.3242, "step": 27220 }, { "epoch": 68.0525, "grad_norm": 6.625, "learning_rate": 4.892172953232039e-06, "loss": 1.2752, "step": 27221 }, { "epoch": 68.055, "grad_norm": 7.09375, "learning_rate": 4.89147687251394e-06, "loss": 1.3837, "step": 27222 }, { "epoch": 68.0575, "grad_norm": 8.6875, "learning_rate": 4.890780825287116e-06, "loss": 1.3317, "step": 27223 }, { "epoch": 68.06, "grad_norm": 7.96875, "learning_rate": 4.890084811556125e-06, "loss": 1.2736, "step": 27224 }, { "epoch": 68.0625, "grad_norm": 8.875, "learning_rate": 4.889388831325537e-06, "loss": 1.3207, "step": 27225 }, { "epoch": 68.065, "grad_norm": 7.3125, "learning_rate": 4.888692884599906e-06, "loss": 1.3844, "step": 27226 }, { "epoch": 68.0675, "grad_norm": 5.8125, "learning_rate": 4.887996971383805e-06, "loss": 1.2236, "step": 27227 }, { "epoch": 68.07, "grad_norm": 8.8125, "learning_rate": 4.8873010916817874e-06, "loss": 1.2275, "step": 27228 }, { "epoch": 68.0725, "grad_norm": 5.0625, "learning_rate": 4.886605245498422e-06, "loss": 1.213, "step": 27229 }, { "epoch": 68.075, "grad_norm": 6.90625, "learning_rate": 4.885909432838269e-06, "loss": 1.3193, "step": 27230 }, { "epoch": 68.0775, "grad_norm": 7.5625, "learning_rate": 4.885213653705886e-06, "loss": 1.3121, "step": 27231 }, { "epoch": 68.08, "grad_norm": 9.0, "learning_rate": 4.884517908105837e-06, "loss": 1.3453, "step": 27232 }, { "epoch": 68.0825, "grad_norm": 6.40625, "learning_rate": 4.883822196042688e-06, "loss": 1.3498, "step": 27233 }, { "epoch": 68.085, "grad_norm": 8.25, "learning_rate": 4.883126517520995e-06, "loss": 1.2231, "step": 27234 }, { "epoch": 68.0875, "grad_norm": 8.75, "learning_rate": 4.882430872545317e-06, "loss": 1.3956, "step": 27235 }, { "epoch": 68.09, "grad_norm": 7.46875, "learning_rate": 4.881735261120216e-06, "loss": 1.5222, "step": 27236 }, { "epoch": 68.0925, "grad_norm": 9.4375, "learning_rate": 4.88103968325026e-06, "loss": 1.4233, "step": 27237 }, { "epoch": 68.095, "grad_norm": 8.125, "learning_rate": 4.880344138940001e-06, "loss": 1.1533, "step": 27238 }, { "epoch": 68.0975, "grad_norm": 6.15625, "learning_rate": 4.879648628193997e-06, "loss": 1.3575, "step": 27239 }, { "epoch": 68.1, "grad_norm": 7.875, "learning_rate": 4.878953151016816e-06, "loss": 1.2946, "step": 27240 }, { "epoch": 68.1025, "grad_norm": 8.3125, "learning_rate": 4.87825770741301e-06, "loss": 1.458, "step": 27241 }, { "epoch": 68.105, "grad_norm": 9.4375, "learning_rate": 4.877562297387144e-06, "loss": 1.3101, "step": 27242 }, { "epoch": 68.1075, "grad_norm": 4.25, "learning_rate": 4.876866920943773e-06, "loss": 1.1664, "step": 27243 }, { "epoch": 68.11, "grad_norm": 6.4375, "learning_rate": 4.876171578087459e-06, "loss": 1.4392, "step": 27244 }, { "epoch": 68.1125, "grad_norm": 7.875, "learning_rate": 4.8754762688227596e-06, "loss": 1.2036, "step": 27245 }, { "epoch": 68.115, "grad_norm": 8.375, "learning_rate": 4.874780993154229e-06, "loss": 1.3343, "step": 27246 }, { "epoch": 68.1175, "grad_norm": 6.625, "learning_rate": 4.8740857510864305e-06, "loss": 1.3451, "step": 27247 }, { "epoch": 68.12, "grad_norm": 8.9375, "learning_rate": 4.873390542623922e-06, "loss": 1.5123, "step": 27248 }, { "epoch": 68.1225, "grad_norm": 9.25, "learning_rate": 4.872695367771262e-06, "loss": 1.4666, "step": 27249 }, { "epoch": 68.125, "grad_norm": 5.46875, "learning_rate": 4.872000226533001e-06, "loss": 1.3893, "step": 27250 }, { "epoch": 68.1275, "grad_norm": 6.25, "learning_rate": 4.871305118913702e-06, "loss": 1.2381, "step": 27251 }, { "epoch": 68.13, "grad_norm": 6.0625, "learning_rate": 4.870610044917925e-06, "loss": 1.2937, "step": 27252 }, { "epoch": 68.1325, "grad_norm": 7.5625, "learning_rate": 4.869915004550223e-06, "loss": 1.2963, "step": 27253 }, { "epoch": 68.135, "grad_norm": 5.875, "learning_rate": 4.86921999781515e-06, "loss": 1.187, "step": 27254 }, { "epoch": 68.1375, "grad_norm": 6.28125, "learning_rate": 4.8685250247172685e-06, "loss": 1.2962, "step": 27255 }, { "epoch": 68.14, "grad_norm": 10.9375, "learning_rate": 4.867830085261128e-06, "loss": 1.3712, "step": 27256 }, { "epoch": 68.1425, "grad_norm": 5.84375, "learning_rate": 4.8671351794512926e-06, "loss": 1.2137, "step": 27257 }, { "epoch": 68.145, "grad_norm": 6.8125, "learning_rate": 4.866440307292308e-06, "loss": 1.3213, "step": 27258 }, { "epoch": 68.1475, "grad_norm": 7.40625, "learning_rate": 4.865745468788739e-06, "loss": 1.2372, "step": 27259 }, { "epoch": 68.15, "grad_norm": 7.59375, "learning_rate": 4.8650506639451385e-06, "loss": 1.3176, "step": 27260 }, { "epoch": 68.1525, "grad_norm": 8.25, "learning_rate": 4.8643558927660575e-06, "loss": 1.42, "step": 27261 }, { "epoch": 68.155, "grad_norm": 8.375, "learning_rate": 4.863661155256052e-06, "loss": 1.4687, "step": 27262 }, { "epoch": 68.1575, "grad_norm": 7.5625, "learning_rate": 4.862966451419683e-06, "loss": 1.3051, "step": 27263 }, { "epoch": 68.16, "grad_norm": 6.28125, "learning_rate": 4.8622717812615e-06, "loss": 1.359, "step": 27264 }, { "epoch": 68.1625, "grad_norm": 10.125, "learning_rate": 4.861577144786058e-06, "loss": 1.3578, "step": 27265 }, { "epoch": 68.165, "grad_norm": 10.9375, "learning_rate": 4.860882541997902e-06, "loss": 1.24, "step": 27266 }, { "epoch": 68.1675, "grad_norm": 8.3125, "learning_rate": 4.8601879729016035e-06, "loss": 1.4359, "step": 27267 }, { "epoch": 68.17, "grad_norm": 8.25, "learning_rate": 4.8594934375017065e-06, "loss": 1.4592, "step": 27268 }, { "epoch": 68.1725, "grad_norm": 15.9375, "learning_rate": 4.858798935802761e-06, "loss": 1.2752, "step": 27269 }, { "epoch": 68.175, "grad_norm": 8.1875, "learning_rate": 4.858104467809323e-06, "loss": 1.2229, "step": 27270 }, { "epoch": 68.1775, "grad_norm": 9.25, "learning_rate": 4.857410033525951e-06, "loss": 1.1928, "step": 27271 }, { "epoch": 68.18, "grad_norm": 5.90625, "learning_rate": 4.856715632957193e-06, "loss": 1.4563, "step": 27272 }, { "epoch": 68.1825, "grad_norm": 8.75, "learning_rate": 4.856021266107598e-06, "loss": 1.2702, "step": 27273 }, { "epoch": 68.185, "grad_norm": 6.75, "learning_rate": 4.8553269329817245e-06, "loss": 1.3848, "step": 27274 }, { "epoch": 68.1875, "grad_norm": 6.5, "learning_rate": 4.854632633584118e-06, "loss": 1.2297, "step": 27275 }, { "epoch": 68.19, "grad_norm": 8.875, "learning_rate": 4.853938367919338e-06, "loss": 1.3759, "step": 27276 }, { "epoch": 68.1925, "grad_norm": 8.25, "learning_rate": 4.853244135991929e-06, "loss": 1.2568, "step": 27277 }, { "epoch": 68.195, "grad_norm": 9.0, "learning_rate": 4.85254993780645e-06, "loss": 1.3198, "step": 27278 }, { "epoch": 68.1975, "grad_norm": 4.9375, "learning_rate": 4.851855773367447e-06, "loss": 1.3602, "step": 27279 }, { "epoch": 68.2, "grad_norm": 8.5625, "learning_rate": 4.851161642679466e-06, "loss": 1.2831, "step": 27280 }, { "epoch": 68.2025, "grad_norm": 6.8125, "learning_rate": 4.850467545747065e-06, "loss": 1.4861, "step": 27281 }, { "epoch": 68.205, "grad_norm": 6.125, "learning_rate": 4.8497734825747964e-06, "loss": 1.336, "step": 27282 }, { "epoch": 68.2075, "grad_norm": 6.5, "learning_rate": 4.849079453167206e-06, "loss": 1.2319, "step": 27283 }, { "epoch": 68.21, "grad_norm": 6.5625, "learning_rate": 4.848385457528845e-06, "loss": 1.5052, "step": 27284 }, { "epoch": 68.2125, "grad_norm": 6.6875, "learning_rate": 4.8476914956642555e-06, "loss": 1.3131, "step": 27285 }, { "epoch": 68.215, "grad_norm": 9.1875, "learning_rate": 4.846997567578001e-06, "loss": 1.4027, "step": 27286 }, { "epoch": 68.2175, "grad_norm": 8.3125, "learning_rate": 4.846303673274625e-06, "loss": 1.1386, "step": 27287 }, { "epoch": 68.22, "grad_norm": 5.90625, "learning_rate": 4.845609812758676e-06, "loss": 1.2193, "step": 27288 }, { "epoch": 68.2225, "grad_norm": 9.25, "learning_rate": 4.8449159860347e-06, "loss": 1.3851, "step": 27289 }, { "epoch": 68.225, "grad_norm": 8.625, "learning_rate": 4.844222193107247e-06, "loss": 1.3237, "step": 27290 }, { "epoch": 68.2275, "grad_norm": 6.75, "learning_rate": 4.84352843398087e-06, "loss": 1.2338, "step": 27291 }, { "epoch": 68.23, "grad_norm": 8.0, "learning_rate": 4.842834708660115e-06, "loss": 1.4828, "step": 27292 }, { "epoch": 68.2325, "grad_norm": 6.5625, "learning_rate": 4.8421410171495265e-06, "loss": 1.3366, "step": 27293 }, { "epoch": 68.235, "grad_norm": 6.625, "learning_rate": 4.8414473594536584e-06, "loss": 1.1988, "step": 27294 }, { "epoch": 68.2375, "grad_norm": 7.46875, "learning_rate": 4.84075373557705e-06, "loss": 1.4298, "step": 27295 }, { "epoch": 68.24, "grad_norm": 7.03125, "learning_rate": 4.840060145524254e-06, "loss": 1.1953, "step": 27296 }, { "epoch": 68.2425, "grad_norm": 10.1875, "learning_rate": 4.839366589299822e-06, "loss": 1.2361, "step": 27297 }, { "epoch": 68.245, "grad_norm": 7.78125, "learning_rate": 4.838673066908296e-06, "loss": 1.4811, "step": 27298 }, { "epoch": 68.2475, "grad_norm": 9.3125, "learning_rate": 4.837979578354223e-06, "loss": 1.3558, "step": 27299 }, { "epoch": 68.25, "grad_norm": 7.5625, "learning_rate": 4.837286123642141e-06, "loss": 1.322, "step": 27300 }, { "epoch": 68.2525, "grad_norm": 9.0625, "learning_rate": 4.836592702776613e-06, "loss": 1.3069, "step": 27301 }, { "epoch": 68.255, "grad_norm": 6.21875, "learning_rate": 4.835899315762176e-06, "loss": 1.285, "step": 27302 }, { "epoch": 68.2575, "grad_norm": 10.0, "learning_rate": 4.8352059626033756e-06, "loss": 1.3079, "step": 27303 }, { "epoch": 68.26, "grad_norm": 7.59375, "learning_rate": 4.834512643304755e-06, "loss": 1.2435, "step": 27304 }, { "epoch": 68.2625, "grad_norm": 6.25, "learning_rate": 4.833819357870863e-06, "loss": 1.3614, "step": 27305 }, { "epoch": 68.265, "grad_norm": 5.75, "learning_rate": 4.833126106306248e-06, "loss": 1.1439, "step": 27306 }, { "epoch": 68.2675, "grad_norm": 11.6875, "learning_rate": 4.832432888615453e-06, "loss": 1.3293, "step": 27307 }, { "epoch": 68.27, "grad_norm": 7.34375, "learning_rate": 4.831739704803016e-06, "loss": 1.2807, "step": 27308 }, { "epoch": 68.2725, "grad_norm": 9.3125, "learning_rate": 4.831046554873491e-06, "loss": 1.3495, "step": 27309 }, { "epoch": 68.275, "grad_norm": 4.8125, "learning_rate": 4.830353438831413e-06, "loss": 1.3486, "step": 27310 }, { "epoch": 68.2775, "grad_norm": 7.6875, "learning_rate": 4.829660356681336e-06, "loss": 1.2412, "step": 27311 }, { "epoch": 68.28, "grad_norm": 5.59375, "learning_rate": 4.828967308427795e-06, "loss": 1.1567, "step": 27312 }, { "epoch": 68.2825, "grad_norm": 7.875, "learning_rate": 4.828274294075341e-06, "loss": 1.3344, "step": 27313 }, { "epoch": 68.285, "grad_norm": 8.3125, "learning_rate": 4.827581313628513e-06, "loss": 1.261, "step": 27314 }, { "epoch": 68.2875, "grad_norm": 6.25, "learning_rate": 4.826888367091852e-06, "loss": 1.4154, "step": 27315 }, { "epoch": 68.29, "grad_norm": 5.59375, "learning_rate": 4.8261954544699045e-06, "loss": 1.3477, "step": 27316 }, { "epoch": 68.2925, "grad_norm": 8.375, "learning_rate": 4.825502575767215e-06, "loss": 1.3847, "step": 27317 }, { "epoch": 68.295, "grad_norm": 6.75, "learning_rate": 4.8248097309883234e-06, "loss": 1.2384, "step": 27318 }, { "epoch": 68.2975, "grad_norm": 7.40625, "learning_rate": 4.824116920137769e-06, "loss": 1.3072, "step": 27319 }, { "epoch": 68.3, "grad_norm": 7.625, "learning_rate": 4.823424143220097e-06, "loss": 1.2648, "step": 27320 }, { "epoch": 68.3025, "grad_norm": 5.90625, "learning_rate": 4.822731400239853e-06, "loss": 1.1899, "step": 27321 }, { "epoch": 68.305, "grad_norm": 6.84375, "learning_rate": 4.822038691201575e-06, "loss": 1.4117, "step": 27322 }, { "epoch": 68.3075, "grad_norm": 6.53125, "learning_rate": 4.821346016109799e-06, "loss": 1.3026, "step": 27323 }, { "epoch": 68.31, "grad_norm": 9.4375, "learning_rate": 4.820653374969077e-06, "loss": 1.3039, "step": 27324 }, { "epoch": 68.3125, "grad_norm": 6.0, "learning_rate": 4.819960767783939e-06, "loss": 1.3189, "step": 27325 }, { "epoch": 68.315, "grad_norm": 9.375, "learning_rate": 4.819268194558934e-06, "loss": 1.3926, "step": 27326 }, { "epoch": 68.3175, "grad_norm": 7.59375, "learning_rate": 4.818575655298597e-06, "loss": 1.2499, "step": 27327 }, { "epoch": 68.32, "grad_norm": 6.1875, "learning_rate": 4.817883150007474e-06, "loss": 1.2317, "step": 27328 }, { "epoch": 68.3225, "grad_norm": 8.375, "learning_rate": 4.817190678690102e-06, "loss": 1.2205, "step": 27329 }, { "epoch": 68.325, "grad_norm": 4.59375, "learning_rate": 4.8164982413510175e-06, "loss": 1.174, "step": 27330 }, { "epoch": 68.3275, "grad_norm": 12.5625, "learning_rate": 4.815805837994761e-06, "loss": 1.4226, "step": 27331 }, { "epoch": 68.33, "grad_norm": 8.375, "learning_rate": 4.8151134686258794e-06, "loss": 1.4248, "step": 27332 }, { "epoch": 68.3325, "grad_norm": 8.4375, "learning_rate": 4.814421133248907e-06, "loss": 1.4536, "step": 27333 }, { "epoch": 68.335, "grad_norm": 9.875, "learning_rate": 4.813728831868376e-06, "loss": 1.1824, "step": 27334 }, { "epoch": 68.3375, "grad_norm": 10.9375, "learning_rate": 4.813036564488833e-06, "loss": 1.1974, "step": 27335 }, { "epoch": 68.34, "grad_norm": 7.09375, "learning_rate": 4.8123443311148175e-06, "loss": 1.3448, "step": 27336 }, { "epoch": 68.3425, "grad_norm": 6.5, "learning_rate": 4.811652131750866e-06, "loss": 1.3329, "step": 27337 }, { "epoch": 68.345, "grad_norm": 7.4375, "learning_rate": 4.8109599664015106e-06, "loss": 1.2811, "step": 27338 }, { "epoch": 68.3475, "grad_norm": 6.34375, "learning_rate": 4.810267835071298e-06, "loss": 1.2982, "step": 27339 }, { "epoch": 68.35, "grad_norm": 7.9375, "learning_rate": 4.809575737764759e-06, "loss": 1.3047, "step": 27340 }, { "epoch": 68.3525, "grad_norm": 8.75, "learning_rate": 4.808883674486436e-06, "loss": 1.3431, "step": 27341 }, { "epoch": 68.355, "grad_norm": 7.46875, "learning_rate": 4.8081916452408615e-06, "loss": 1.3705, "step": 27342 }, { "epoch": 68.3575, "grad_norm": 6.125, "learning_rate": 4.807499650032579e-06, "loss": 1.2933, "step": 27343 }, { "epoch": 68.36, "grad_norm": 9.1875, "learning_rate": 4.806807688866119e-06, "loss": 1.3898, "step": 27344 }, { "epoch": 68.3625, "grad_norm": 6.625, "learning_rate": 4.806115761746018e-06, "loss": 1.1833, "step": 27345 }, { "epoch": 68.365, "grad_norm": 13.5, "learning_rate": 4.805423868676814e-06, "loss": 1.4045, "step": 27346 }, { "epoch": 68.3675, "grad_norm": 6.9375, "learning_rate": 4.804732009663046e-06, "loss": 1.3135, "step": 27347 }, { "epoch": 68.37, "grad_norm": 9.1875, "learning_rate": 4.804040184709248e-06, "loss": 1.3736, "step": 27348 }, { "epoch": 68.3725, "grad_norm": 8.0625, "learning_rate": 4.80334839381995e-06, "loss": 1.2224, "step": 27349 }, { "epoch": 68.375, "grad_norm": 7.5625, "learning_rate": 4.802656636999693e-06, "loss": 1.201, "step": 27350 }, { "epoch": 68.3775, "grad_norm": 7.84375, "learning_rate": 4.801964914253013e-06, "loss": 1.358, "step": 27351 }, { "epoch": 68.38, "grad_norm": 6.375, "learning_rate": 4.801273225584445e-06, "loss": 1.2956, "step": 27352 }, { "epoch": 68.3825, "grad_norm": 6.90625, "learning_rate": 4.800581570998517e-06, "loss": 1.367, "step": 27353 }, { "epoch": 68.385, "grad_norm": 5.65625, "learning_rate": 4.799889950499772e-06, "loss": 1.3364, "step": 27354 }, { "epoch": 68.3875, "grad_norm": 9.0, "learning_rate": 4.7991983640927375e-06, "loss": 1.3773, "step": 27355 }, { "epoch": 68.39, "grad_norm": 8.5, "learning_rate": 4.798506811781953e-06, "loss": 1.3423, "step": 27356 }, { "epoch": 68.3925, "grad_norm": 5.53125, "learning_rate": 4.797815293571947e-06, "loss": 1.2998, "step": 27357 }, { "epoch": 68.395, "grad_norm": 7.21875, "learning_rate": 4.797123809467258e-06, "loss": 1.2805, "step": 27358 }, { "epoch": 68.3975, "grad_norm": 6.65625, "learning_rate": 4.796432359472418e-06, "loss": 1.3854, "step": 27359 }, { "epoch": 68.4, "grad_norm": 6.25, "learning_rate": 4.795740943591955e-06, "loss": 1.4565, "step": 27360 }, { "epoch": 68.4025, "grad_norm": 7.75, "learning_rate": 4.795049561830407e-06, "loss": 1.2816, "step": 27361 }, { "epoch": 68.405, "grad_norm": 9.4375, "learning_rate": 4.794358214192309e-06, "loss": 1.4346, "step": 27362 }, { "epoch": 68.4075, "grad_norm": 6.90625, "learning_rate": 4.79366690068219e-06, "loss": 1.3612, "step": 27363 }, { "epoch": 68.41, "grad_norm": 8.375, "learning_rate": 4.7929756213045796e-06, "loss": 1.4011, "step": 27364 }, { "epoch": 68.4125, "grad_norm": 5.8125, "learning_rate": 4.792284376064011e-06, "loss": 1.4457, "step": 27365 }, { "epoch": 68.415, "grad_norm": 9.3125, "learning_rate": 4.791593164965024e-06, "loss": 1.2547, "step": 27366 }, { "epoch": 68.4175, "grad_norm": 7.84375, "learning_rate": 4.790901988012141e-06, "loss": 1.3692, "step": 27367 }, { "epoch": 68.42, "grad_norm": 7.59375, "learning_rate": 4.790210845209894e-06, "loss": 1.2569, "step": 27368 }, { "epoch": 68.4225, "grad_norm": 12.0625, "learning_rate": 4.78951973656282e-06, "loss": 1.384, "step": 27369 }, { "epoch": 68.425, "grad_norm": 9.9375, "learning_rate": 4.788828662075441e-06, "loss": 1.4166, "step": 27370 }, { "epoch": 68.4275, "grad_norm": 7.75, "learning_rate": 4.788137621752298e-06, "loss": 1.2401, "step": 27371 }, { "epoch": 68.43, "grad_norm": 8.125, "learning_rate": 4.7874466155979116e-06, "loss": 1.3944, "step": 27372 }, { "epoch": 68.4325, "grad_norm": 6.40625, "learning_rate": 4.78675564361682e-06, "loss": 1.3822, "step": 27373 }, { "epoch": 68.435, "grad_norm": 7.0625, "learning_rate": 4.786064705813551e-06, "loss": 1.2947, "step": 27374 }, { "epoch": 68.4375, "grad_norm": 8.375, "learning_rate": 4.7853738021926284e-06, "loss": 1.3891, "step": 27375 }, { "epoch": 68.44, "grad_norm": 6.0625, "learning_rate": 4.784682932758588e-06, "loss": 1.3753, "step": 27376 }, { "epoch": 68.4425, "grad_norm": 7.4375, "learning_rate": 4.783992097515959e-06, "loss": 1.3575, "step": 27377 }, { "epoch": 68.445, "grad_norm": 8.6875, "learning_rate": 4.783301296469271e-06, "loss": 1.4136, "step": 27378 }, { "epoch": 68.4475, "grad_norm": 6.75, "learning_rate": 4.782610529623046e-06, "loss": 1.42, "step": 27379 }, { "epoch": 68.45, "grad_norm": 12.0, "learning_rate": 4.781919796981818e-06, "loss": 1.4536, "step": 27380 }, { "epoch": 68.4525, "grad_norm": 5.96875, "learning_rate": 4.781229098550119e-06, "loss": 1.375, "step": 27381 }, { "epoch": 68.455, "grad_norm": 7.40625, "learning_rate": 4.780538434332472e-06, "loss": 1.2491, "step": 27382 }, { "epoch": 68.4575, "grad_norm": 20.25, "learning_rate": 4.779847804333404e-06, "loss": 1.2329, "step": 27383 }, { "epoch": 68.46, "grad_norm": 7.84375, "learning_rate": 4.779157208557448e-06, "loss": 1.2159, "step": 27384 }, { "epoch": 68.4625, "grad_norm": 12.75, "learning_rate": 4.778466647009124e-06, "loss": 1.1929, "step": 27385 }, { "epoch": 68.465, "grad_norm": 12.125, "learning_rate": 4.777776119692968e-06, "loss": 1.3327, "step": 27386 }, { "epoch": 68.4675, "grad_norm": 6.8125, "learning_rate": 4.777085626613499e-06, "loss": 1.2574, "step": 27387 }, { "epoch": 68.47, "grad_norm": 9.5625, "learning_rate": 4.776395167775252e-06, "loss": 1.4056, "step": 27388 }, { "epoch": 68.4725, "grad_norm": 6.71875, "learning_rate": 4.775704743182749e-06, "loss": 1.2092, "step": 27389 }, { "epoch": 68.475, "grad_norm": 7.59375, "learning_rate": 4.775014352840512e-06, "loss": 1.2618, "step": 27390 }, { "epoch": 68.4775, "grad_norm": 5.59375, "learning_rate": 4.774323996753073e-06, "loss": 1.3278, "step": 27391 }, { "epoch": 68.48, "grad_norm": 7.0, "learning_rate": 4.77363367492496e-06, "loss": 1.3635, "step": 27392 }, { "epoch": 68.4825, "grad_norm": 8.625, "learning_rate": 4.772943387360696e-06, "loss": 1.4068, "step": 27393 }, { "epoch": 68.485, "grad_norm": 7.125, "learning_rate": 4.7722531340648e-06, "loss": 1.3605, "step": 27394 }, { "epoch": 68.4875, "grad_norm": 7.46875, "learning_rate": 4.771562915041805e-06, "loss": 1.4245, "step": 27395 }, { "epoch": 68.49, "grad_norm": 8.875, "learning_rate": 4.770872730296239e-06, "loss": 1.2915, "step": 27396 }, { "epoch": 68.4925, "grad_norm": 5.75, "learning_rate": 4.770182579832621e-06, "loss": 1.3514, "step": 27397 }, { "epoch": 68.495, "grad_norm": 10.875, "learning_rate": 4.769492463655473e-06, "loss": 1.3024, "step": 27398 }, { "epoch": 68.4975, "grad_norm": 7.6875, "learning_rate": 4.768802381769326e-06, "loss": 1.4122, "step": 27399 }, { "epoch": 68.5, "grad_norm": 10.5, "learning_rate": 4.7681123341787e-06, "loss": 1.3745, "step": 27400 }, { "epoch": 68.5025, "grad_norm": 9.8125, "learning_rate": 4.767422320888121e-06, "loss": 1.2957, "step": 27401 }, { "epoch": 68.505, "grad_norm": 9.6875, "learning_rate": 4.7667323419021095e-06, "loss": 1.2486, "step": 27402 }, { "epoch": 68.5075, "grad_norm": 8.625, "learning_rate": 4.766042397225196e-06, "loss": 1.2688, "step": 27403 }, { "epoch": 68.51, "grad_norm": 8.4375, "learning_rate": 4.7653524868618975e-06, "loss": 1.3596, "step": 27404 }, { "epoch": 68.5125, "grad_norm": 8.6875, "learning_rate": 4.7646626108167346e-06, "loss": 1.2478, "step": 27405 }, { "epoch": 68.515, "grad_norm": 7.0625, "learning_rate": 4.763972769094236e-06, "loss": 1.3896, "step": 27406 }, { "epoch": 68.5175, "grad_norm": 8.75, "learning_rate": 4.763282961698925e-06, "loss": 1.2635, "step": 27407 }, { "epoch": 68.52, "grad_norm": 8.8125, "learning_rate": 4.7625931886353215e-06, "loss": 1.4089, "step": 27408 }, { "epoch": 68.5225, "grad_norm": 7.1875, "learning_rate": 4.761903449907943e-06, "loss": 1.3964, "step": 27409 }, { "epoch": 68.525, "grad_norm": 6.1875, "learning_rate": 4.761213745521317e-06, "loss": 1.4273, "step": 27410 }, { "epoch": 68.5275, "grad_norm": 7.65625, "learning_rate": 4.760524075479968e-06, "loss": 1.2121, "step": 27411 }, { "epoch": 68.53, "grad_norm": 5.09375, "learning_rate": 4.759834439788412e-06, "loss": 1.4095, "step": 27412 }, { "epoch": 68.5325, "grad_norm": 6.28125, "learning_rate": 4.759144838451169e-06, "loss": 1.3539, "step": 27413 }, { "epoch": 68.535, "grad_norm": 6.65625, "learning_rate": 4.758455271472766e-06, "loss": 1.2457, "step": 27414 }, { "epoch": 68.5375, "grad_norm": 8.25, "learning_rate": 4.757765738857717e-06, "loss": 1.3272, "step": 27415 }, { "epoch": 68.54, "grad_norm": 8.9375, "learning_rate": 4.75707624061055e-06, "loss": 1.3596, "step": 27416 }, { "epoch": 68.5425, "grad_norm": 7.25, "learning_rate": 4.756386776735777e-06, "loss": 1.3181, "step": 27417 }, { "epoch": 68.545, "grad_norm": 8.3125, "learning_rate": 4.755697347237927e-06, "loss": 1.2308, "step": 27418 }, { "epoch": 68.5475, "grad_norm": 5.375, "learning_rate": 4.755007952121516e-06, "loss": 1.3174, "step": 27419 }, { "epoch": 68.55, "grad_norm": 8.5625, "learning_rate": 4.754318591391057e-06, "loss": 1.3944, "step": 27420 }, { "epoch": 68.5525, "grad_norm": 6.6875, "learning_rate": 4.753629265051076e-06, "loss": 1.2875, "step": 27421 }, { "epoch": 68.555, "grad_norm": 8.625, "learning_rate": 4.752939973106097e-06, "loss": 1.4146, "step": 27422 }, { "epoch": 68.5575, "grad_norm": 4.875, "learning_rate": 4.752250715560633e-06, "loss": 1.1648, "step": 27423 }, { "epoch": 68.56, "grad_norm": 8.5, "learning_rate": 4.7515614924192026e-06, "loss": 1.2818, "step": 27424 }, { "epoch": 68.5625, "grad_norm": 8.6875, "learning_rate": 4.750872303686317e-06, "loss": 1.2378, "step": 27425 }, { "epoch": 68.565, "grad_norm": 11.5, "learning_rate": 4.750183149366511e-06, "loss": 1.4364, "step": 27426 }, { "epoch": 68.5675, "grad_norm": 9.375, "learning_rate": 4.749494029464293e-06, "loss": 1.3209, "step": 27427 }, { "epoch": 68.57, "grad_norm": 9.3125, "learning_rate": 4.748804943984183e-06, "loss": 1.3747, "step": 27428 }, { "epoch": 68.5725, "grad_norm": 6.8125, "learning_rate": 4.7481158929306935e-06, "loss": 1.2834, "step": 27429 }, { "epoch": 68.575, "grad_norm": 9.5625, "learning_rate": 4.747426876308346e-06, "loss": 1.3797, "step": 27430 }, { "epoch": 68.5775, "grad_norm": 7.875, "learning_rate": 4.746737894121662e-06, "loss": 1.3416, "step": 27431 }, { "epoch": 68.58, "grad_norm": 8.625, "learning_rate": 4.74604894637515e-06, "loss": 1.2874, "step": 27432 }, { "epoch": 68.5825, "grad_norm": 7.71875, "learning_rate": 4.745360033073335e-06, "loss": 1.3345, "step": 27433 }, { "epoch": 68.585, "grad_norm": 6.125, "learning_rate": 4.744671154220729e-06, "loss": 1.3579, "step": 27434 }, { "epoch": 68.5875, "grad_norm": 7.59375, "learning_rate": 4.7439823098218455e-06, "loss": 1.3082, "step": 27435 }, { "epoch": 68.59, "grad_norm": 8.875, "learning_rate": 4.743293499881203e-06, "loss": 1.3201, "step": 27436 }, { "epoch": 68.5925, "grad_norm": 11.875, "learning_rate": 4.742604724403322e-06, "loss": 1.2253, "step": 27437 }, { "epoch": 68.595, "grad_norm": 7.0, "learning_rate": 4.7419159833927145e-06, "loss": 1.3724, "step": 27438 }, { "epoch": 68.5975, "grad_norm": 8.375, "learning_rate": 4.741227276853895e-06, "loss": 1.3857, "step": 27439 }, { "epoch": 68.6, "grad_norm": 7.03125, "learning_rate": 4.740538604791371e-06, "loss": 1.3201, "step": 27440 }, { "epoch": 68.6025, "grad_norm": 6.8125, "learning_rate": 4.739849967209673e-06, "loss": 1.2747, "step": 27441 }, { "epoch": 68.605, "grad_norm": 6.90625, "learning_rate": 4.739161364113308e-06, "loss": 1.2584, "step": 27442 }, { "epoch": 68.6075, "grad_norm": 6.21875, "learning_rate": 4.738472795506791e-06, "loss": 1.2526, "step": 27443 }, { "epoch": 68.61, "grad_norm": 7.53125, "learning_rate": 4.737784261394632e-06, "loss": 1.2975, "step": 27444 }, { "epoch": 68.6125, "grad_norm": 13.5625, "learning_rate": 4.737095761781348e-06, "loss": 1.2504, "step": 27445 }, { "epoch": 68.615, "grad_norm": 8.1875, "learning_rate": 4.736407296671459e-06, "loss": 1.3205, "step": 27446 }, { "epoch": 68.6175, "grad_norm": 11.5, "learning_rate": 4.735718866069472e-06, "loss": 1.3635, "step": 27447 }, { "epoch": 68.62, "grad_norm": 6.6875, "learning_rate": 4.7350304699798976e-06, "loss": 1.1866, "step": 27448 }, { "epoch": 68.6225, "grad_norm": 5.59375, "learning_rate": 4.734342108407257e-06, "loss": 1.3439, "step": 27449 }, { "epoch": 68.625, "grad_norm": 7.15625, "learning_rate": 4.733653781356055e-06, "loss": 1.3137, "step": 27450 }, { "epoch": 68.6275, "grad_norm": 7.03125, "learning_rate": 4.732965488830812e-06, "loss": 1.2851, "step": 27451 }, { "epoch": 68.63, "grad_norm": 4.875, "learning_rate": 4.732277230836032e-06, "loss": 1.3116, "step": 27452 }, { "epoch": 68.6325, "grad_norm": 6.71875, "learning_rate": 4.731589007376236e-06, "loss": 1.2974, "step": 27453 }, { "epoch": 68.635, "grad_norm": 7.21875, "learning_rate": 4.730900818455932e-06, "loss": 1.3679, "step": 27454 }, { "epoch": 68.6375, "grad_norm": 8.0625, "learning_rate": 4.730212664079626e-06, "loss": 1.3963, "step": 27455 }, { "epoch": 68.64, "grad_norm": 7.34375, "learning_rate": 4.729524544251837e-06, "loss": 1.4727, "step": 27456 }, { "epoch": 68.6425, "grad_norm": 7.84375, "learning_rate": 4.7288364589770765e-06, "loss": 1.2122, "step": 27457 }, { "epoch": 68.645, "grad_norm": 6.28125, "learning_rate": 4.7281484082598536e-06, "loss": 1.2709, "step": 27458 }, { "epoch": 68.6475, "grad_norm": 8.5625, "learning_rate": 4.727460392104675e-06, "loss": 1.2273, "step": 27459 }, { "epoch": 68.65, "grad_norm": 8.125, "learning_rate": 4.726772410516055e-06, "loss": 1.3296, "step": 27460 }, { "epoch": 68.6525, "grad_norm": 7.59375, "learning_rate": 4.726084463498508e-06, "loss": 1.274, "step": 27461 }, { "epoch": 68.655, "grad_norm": 8.625, "learning_rate": 4.72539655105654e-06, "loss": 1.3784, "step": 27462 }, { "epoch": 68.6575, "grad_norm": 5.71875, "learning_rate": 4.724708673194658e-06, "loss": 1.436, "step": 27463 }, { "epoch": 68.66, "grad_norm": 7.15625, "learning_rate": 4.724020829917378e-06, "loss": 1.3646, "step": 27464 }, { "epoch": 68.6625, "grad_norm": 6.53125, "learning_rate": 4.7233330212292024e-06, "loss": 1.2759, "step": 27465 }, { "epoch": 68.665, "grad_norm": 8.4375, "learning_rate": 4.722645247134648e-06, "loss": 1.2385, "step": 27466 }, { "epoch": 68.6675, "grad_norm": 8.875, "learning_rate": 4.721957507638217e-06, "loss": 1.1134, "step": 27467 }, { "epoch": 68.67, "grad_norm": 4.25, "learning_rate": 4.721269802744425e-06, "loss": 1.1541, "step": 27468 }, { "epoch": 68.6725, "grad_norm": 7.71875, "learning_rate": 4.720582132457777e-06, "loss": 1.2248, "step": 27469 }, { "epoch": 68.675, "grad_norm": 7.59375, "learning_rate": 4.719894496782778e-06, "loss": 1.2003, "step": 27470 }, { "epoch": 68.6775, "grad_norm": 9.625, "learning_rate": 4.719206895723938e-06, "loss": 1.4017, "step": 27471 }, { "epoch": 68.68, "grad_norm": 6.9375, "learning_rate": 4.718519329285771e-06, "loss": 1.2583, "step": 27472 }, { "epoch": 68.6825, "grad_norm": 8.5625, "learning_rate": 4.717831797472779e-06, "loss": 1.3503, "step": 27473 }, { "epoch": 68.685, "grad_norm": 7.5, "learning_rate": 4.717144300289468e-06, "loss": 1.4201, "step": 27474 }, { "epoch": 68.6875, "grad_norm": 7.90625, "learning_rate": 4.716456837740347e-06, "loss": 1.131, "step": 27475 }, { "epoch": 68.69, "grad_norm": 7.125, "learning_rate": 4.715769409829926e-06, "loss": 1.3009, "step": 27476 }, { "epoch": 68.6925, "grad_norm": 8.25, "learning_rate": 4.715082016562712e-06, "loss": 1.3605, "step": 27477 }, { "epoch": 68.695, "grad_norm": 7.15625, "learning_rate": 4.714394657943203e-06, "loss": 1.2401, "step": 27478 }, { "epoch": 68.6975, "grad_norm": 10.1875, "learning_rate": 4.7137073339759155e-06, "loss": 1.3444, "step": 27479 }, { "epoch": 68.7, "grad_norm": 6.90625, "learning_rate": 4.713020044665348e-06, "loss": 1.4789, "step": 27480 }, { "epoch": 68.7025, "grad_norm": 5.75, "learning_rate": 4.712332790016012e-06, "loss": 1.3371, "step": 27481 }, { "epoch": 68.705, "grad_norm": 6.28125, "learning_rate": 4.711645570032407e-06, "loss": 1.2485, "step": 27482 }, { "epoch": 68.7075, "grad_norm": 10.9375, "learning_rate": 4.710958384719046e-06, "loss": 1.3371, "step": 27483 }, { "epoch": 68.71, "grad_norm": 7.28125, "learning_rate": 4.71027123408043e-06, "loss": 1.3886, "step": 27484 }, { "epoch": 68.7125, "grad_norm": 9.75, "learning_rate": 4.709584118121061e-06, "loss": 1.3346, "step": 27485 }, { "epoch": 68.715, "grad_norm": 6.375, "learning_rate": 4.708897036845447e-06, "loss": 1.1988, "step": 27486 }, { "epoch": 68.7175, "grad_norm": 7.03125, "learning_rate": 4.708209990258096e-06, "loss": 1.4467, "step": 27487 }, { "epoch": 68.72, "grad_norm": 8.5, "learning_rate": 4.707522978363508e-06, "loss": 1.3244, "step": 27488 }, { "epoch": 68.7225, "grad_norm": 6.96875, "learning_rate": 4.706836001166183e-06, "loss": 1.6149, "step": 27489 }, { "epoch": 68.725, "grad_norm": 8.0625, "learning_rate": 4.70614905867063e-06, "loss": 1.4175, "step": 27490 }, { "epoch": 68.7275, "grad_norm": 6.46875, "learning_rate": 4.705462150881357e-06, "loss": 1.2928, "step": 27491 }, { "epoch": 68.73, "grad_norm": 6.53125, "learning_rate": 4.704775277802861e-06, "loss": 1.2402, "step": 27492 }, { "epoch": 68.7325, "grad_norm": 10.8125, "learning_rate": 4.704088439439643e-06, "loss": 1.4485, "step": 27493 }, { "epoch": 68.735, "grad_norm": 6.5625, "learning_rate": 4.703401635796212e-06, "loss": 1.3757, "step": 27494 }, { "epoch": 68.7375, "grad_norm": 9.0, "learning_rate": 4.702714866877065e-06, "loss": 1.4212, "step": 27495 }, { "epoch": 68.74, "grad_norm": 7.34375, "learning_rate": 4.702028132686711e-06, "loss": 1.4496, "step": 27496 }, { "epoch": 68.7425, "grad_norm": 7.6875, "learning_rate": 4.701341433229646e-06, "loss": 1.3243, "step": 27497 }, { "epoch": 68.745, "grad_norm": 9.625, "learning_rate": 4.700654768510376e-06, "loss": 1.309, "step": 27498 }, { "epoch": 68.7475, "grad_norm": 12.75, "learning_rate": 4.699968138533403e-06, "loss": 1.3377, "step": 27499 }, { "epoch": 68.75, "grad_norm": 5.59375, "learning_rate": 4.699281543303222e-06, "loss": 1.2275, "step": 27500 }, { "epoch": 68.7525, "grad_norm": 7.25, "learning_rate": 4.69859498282434e-06, "loss": 1.3292, "step": 27501 }, { "epoch": 68.755, "grad_norm": 8.0, "learning_rate": 4.69790845710126e-06, "loss": 1.4498, "step": 27502 }, { "epoch": 68.7575, "grad_norm": 8.125, "learning_rate": 4.697221966138481e-06, "loss": 1.2851, "step": 27503 }, { "epoch": 68.76, "grad_norm": 7.59375, "learning_rate": 4.696535509940499e-06, "loss": 1.2495, "step": 27504 }, { "epoch": 68.7625, "grad_norm": 7.375, "learning_rate": 4.6958490885118165e-06, "loss": 1.3459, "step": 27505 }, { "epoch": 68.765, "grad_norm": 7.0, "learning_rate": 4.695162701856939e-06, "loss": 1.4808, "step": 27506 }, { "epoch": 68.7675, "grad_norm": 9.625, "learning_rate": 4.694476349980364e-06, "loss": 1.2666, "step": 27507 }, { "epoch": 68.77, "grad_norm": 7.1875, "learning_rate": 4.693790032886585e-06, "loss": 1.2431, "step": 27508 }, { "epoch": 68.7725, "grad_norm": 5.84375, "learning_rate": 4.69310375058011e-06, "loss": 1.4942, "step": 27509 }, { "epoch": 68.775, "grad_norm": 5.78125, "learning_rate": 4.6924175030654306e-06, "loss": 1.3635, "step": 27510 }, { "epoch": 68.7775, "grad_norm": 7.71875, "learning_rate": 4.691731290347053e-06, "loss": 1.2783, "step": 27511 }, { "epoch": 68.78, "grad_norm": 11.0, "learning_rate": 4.69104511242947e-06, "loss": 1.3693, "step": 27512 }, { "epoch": 68.7825, "grad_norm": 10.125, "learning_rate": 4.690358969317186e-06, "loss": 1.2709, "step": 27513 }, { "epoch": 68.785, "grad_norm": 7.84375, "learning_rate": 4.6896728610146925e-06, "loss": 1.2189, "step": 27514 }, { "epoch": 68.7875, "grad_norm": 8.1875, "learning_rate": 4.688986787526495e-06, "loss": 1.3525, "step": 27515 }, { "epoch": 68.79, "grad_norm": 5.59375, "learning_rate": 4.688300748857083e-06, "loss": 1.3968, "step": 27516 }, { "epoch": 68.7925, "grad_norm": 5.9375, "learning_rate": 4.687614745010963e-06, "loss": 1.2385, "step": 27517 }, { "epoch": 68.795, "grad_norm": 15.75, "learning_rate": 4.6869287759926284e-06, "loss": 1.3893, "step": 27518 }, { "epoch": 68.7975, "grad_norm": 6.71875, "learning_rate": 4.686242841806573e-06, "loss": 1.3467, "step": 27519 }, { "epoch": 68.8, "grad_norm": 6.5, "learning_rate": 4.685556942457296e-06, "loss": 1.3487, "step": 27520 }, { "epoch": 68.8025, "grad_norm": 6.3125, "learning_rate": 4.684871077949299e-06, "loss": 1.4277, "step": 27521 }, { "epoch": 68.805, "grad_norm": 5.1875, "learning_rate": 4.6841852482870745e-06, "loss": 1.3095, "step": 27522 }, { "epoch": 68.8075, "grad_norm": 6.125, "learning_rate": 4.683499453475114e-06, "loss": 1.3348, "step": 27523 }, { "epoch": 68.81, "grad_norm": 9.8125, "learning_rate": 4.68281369351792e-06, "loss": 1.2417, "step": 27524 }, { "epoch": 68.8125, "grad_norm": 7.5625, "learning_rate": 4.68212796841999e-06, "loss": 1.3868, "step": 27525 }, { "epoch": 68.815, "grad_norm": 8.8125, "learning_rate": 4.6814422781858145e-06, "loss": 1.3826, "step": 27526 }, { "epoch": 68.8175, "grad_norm": 7.4375, "learning_rate": 4.680756622819889e-06, "loss": 1.1941, "step": 27527 }, { "epoch": 68.82, "grad_norm": 7.15625, "learning_rate": 4.680071002326714e-06, "loss": 1.2992, "step": 27528 }, { "epoch": 68.8225, "grad_norm": 6.3125, "learning_rate": 4.679385416710776e-06, "loss": 1.192, "step": 27529 }, { "epoch": 68.825, "grad_norm": 7.90625, "learning_rate": 4.678699865976577e-06, "loss": 1.2274, "step": 27530 }, { "epoch": 68.8275, "grad_norm": 5.53125, "learning_rate": 4.678014350128607e-06, "loss": 1.2912, "step": 27531 }, { "epoch": 68.83, "grad_norm": 8.1875, "learning_rate": 4.677328869171365e-06, "loss": 1.4248, "step": 27532 }, { "epoch": 68.8325, "grad_norm": 7.6875, "learning_rate": 4.676643423109341e-06, "loss": 1.2914, "step": 27533 }, { "epoch": 68.835, "grad_norm": 8.4375, "learning_rate": 4.675958011947027e-06, "loss": 1.3423, "step": 27534 }, { "epoch": 68.8375, "grad_norm": 7.1875, "learning_rate": 4.675272635688919e-06, "loss": 1.2814, "step": 27535 }, { "epoch": 68.84, "grad_norm": 10.6875, "learning_rate": 4.674587294339513e-06, "loss": 1.4046, "step": 27536 }, { "epoch": 68.8425, "grad_norm": 6.9375, "learning_rate": 4.673901987903301e-06, "loss": 1.3105, "step": 27537 }, { "epoch": 68.845, "grad_norm": 7.0625, "learning_rate": 4.673216716384771e-06, "loss": 1.3909, "step": 27538 }, { "epoch": 68.8475, "grad_norm": 6.65625, "learning_rate": 4.672531479788418e-06, "loss": 1.2887, "step": 27539 }, { "epoch": 68.85, "grad_norm": 10.3125, "learning_rate": 4.67184627811874e-06, "loss": 1.4187, "step": 27540 }, { "epoch": 68.8525, "grad_norm": 6.59375, "learning_rate": 4.671161111380225e-06, "loss": 1.3676, "step": 27541 }, { "epoch": 68.855, "grad_norm": 6.53125, "learning_rate": 4.6704759795773595e-06, "loss": 1.2589, "step": 27542 }, { "epoch": 68.8575, "grad_norm": 6.1875, "learning_rate": 4.6697908827146446e-06, "loss": 1.3649, "step": 27543 }, { "epoch": 68.86, "grad_norm": 5.625, "learning_rate": 4.669105820796564e-06, "loss": 1.3074, "step": 27544 }, { "epoch": 68.8625, "grad_norm": 6.8125, "learning_rate": 4.668420793827615e-06, "loss": 1.2104, "step": 27545 }, { "epoch": 68.865, "grad_norm": 6.6875, "learning_rate": 4.667735801812282e-06, "loss": 1.376, "step": 27546 }, { "epoch": 68.8675, "grad_norm": 11.8125, "learning_rate": 4.667050844755064e-06, "loss": 1.3619, "step": 27547 }, { "epoch": 68.87, "grad_norm": 8.9375, "learning_rate": 4.6663659226604474e-06, "loss": 1.4476, "step": 27548 }, { "epoch": 68.8725, "grad_norm": 6.4375, "learning_rate": 4.665681035532918e-06, "loss": 1.3985, "step": 27549 }, { "epoch": 68.875, "grad_norm": 6.625, "learning_rate": 4.664996183376972e-06, "loss": 1.3457, "step": 27550 }, { "epoch": 68.8775, "grad_norm": 7.90625, "learning_rate": 4.6643113661971e-06, "loss": 1.3217, "step": 27551 }, { "epoch": 68.88, "grad_norm": 9.0, "learning_rate": 4.663626583997789e-06, "loss": 1.4296, "step": 27552 }, { "epoch": 68.8825, "grad_norm": 9.3125, "learning_rate": 4.662941836783525e-06, "loss": 1.2484, "step": 27553 }, { "epoch": 68.885, "grad_norm": 9.3125, "learning_rate": 4.662257124558801e-06, "loss": 1.2034, "step": 27554 }, { "epoch": 68.8875, "grad_norm": 9.0, "learning_rate": 4.6615724473281085e-06, "loss": 1.4035, "step": 27555 }, { "epoch": 68.89, "grad_norm": 10.5625, "learning_rate": 4.660887805095935e-06, "loss": 1.3041, "step": 27556 }, { "epoch": 68.8925, "grad_norm": 8.125, "learning_rate": 4.660203197866762e-06, "loss": 1.157, "step": 27557 }, { "epoch": 68.895, "grad_norm": 8.375, "learning_rate": 4.659518625645089e-06, "loss": 1.2973, "step": 27558 }, { "epoch": 68.8975, "grad_norm": 13.5, "learning_rate": 4.658834088435392e-06, "loss": 1.262, "step": 27559 }, { "epoch": 68.9, "grad_norm": 7.59375, "learning_rate": 4.65814958624217e-06, "loss": 1.3591, "step": 27560 }, { "epoch": 68.9025, "grad_norm": 7.125, "learning_rate": 4.6574651190699015e-06, "loss": 1.3446, "step": 27561 }, { "epoch": 68.905, "grad_norm": 7.90625, "learning_rate": 4.656780686923083e-06, "loss": 1.2326, "step": 27562 }, { "epoch": 68.9075, "grad_norm": 8.4375, "learning_rate": 4.656096289806195e-06, "loss": 1.374, "step": 27563 }, { "epoch": 68.91, "grad_norm": 8.0, "learning_rate": 4.655411927723722e-06, "loss": 1.4784, "step": 27564 }, { "epoch": 68.9125, "grad_norm": 5.9375, "learning_rate": 4.6547276006801554e-06, "loss": 1.3348, "step": 27565 }, { "epoch": 68.915, "grad_norm": 8.5, "learning_rate": 4.654043308679984e-06, "loss": 1.2878, "step": 27566 }, { "epoch": 68.9175, "grad_norm": 6.625, "learning_rate": 4.65335905172769e-06, "loss": 1.4851, "step": 27567 }, { "epoch": 68.92, "grad_norm": 7.03125, "learning_rate": 4.652674829827762e-06, "loss": 1.4104, "step": 27568 }, { "epoch": 68.9225, "grad_norm": 7.53125, "learning_rate": 4.651990642984675e-06, "loss": 1.2862, "step": 27569 }, { "epoch": 68.925, "grad_norm": 4.78125, "learning_rate": 4.65130649120293e-06, "loss": 1.5281, "step": 27570 }, { "epoch": 68.9275, "grad_norm": 9.1875, "learning_rate": 4.6506223744870076e-06, "loss": 1.1337, "step": 27571 }, { "epoch": 68.93, "grad_norm": 9.25, "learning_rate": 4.649938292841386e-06, "loss": 1.2713, "step": 27572 }, { "epoch": 68.9325, "grad_norm": 8.6875, "learning_rate": 4.649254246270558e-06, "loss": 1.414, "step": 27573 }, { "epoch": 68.935, "grad_norm": 7.875, "learning_rate": 4.648570234779003e-06, "loss": 1.4376, "step": 27574 }, { "epoch": 68.9375, "grad_norm": 8.75, "learning_rate": 4.6478862583712096e-06, "loss": 1.2603, "step": 27575 }, { "epoch": 68.94, "grad_norm": 8.1875, "learning_rate": 4.647202317051656e-06, "loss": 1.3735, "step": 27576 }, { "epoch": 68.9425, "grad_norm": 6.0, "learning_rate": 4.646518410824835e-06, "loss": 1.293, "step": 27577 }, { "epoch": 68.945, "grad_norm": 8.3125, "learning_rate": 4.645834539695224e-06, "loss": 1.4365, "step": 27578 }, { "epoch": 68.9475, "grad_norm": 7.6875, "learning_rate": 4.645150703667304e-06, "loss": 1.248, "step": 27579 }, { "epoch": 68.95, "grad_norm": 7.9375, "learning_rate": 4.6444669027455615e-06, "loss": 1.1656, "step": 27580 }, { "epoch": 68.9525, "grad_norm": 4.625, "learning_rate": 4.643783136934483e-06, "loss": 1.2802, "step": 27581 }, { "epoch": 68.955, "grad_norm": 5.90625, "learning_rate": 4.6430994062385484e-06, "loss": 1.3346, "step": 27582 }, { "epoch": 68.9575, "grad_norm": 7.40625, "learning_rate": 4.64241571066224e-06, "loss": 1.3229, "step": 27583 }, { "epoch": 68.96, "grad_norm": 7.3125, "learning_rate": 4.641732050210032e-06, "loss": 1.3208, "step": 27584 }, { "epoch": 68.9625, "grad_norm": 9.6875, "learning_rate": 4.641048424886422e-06, "loss": 1.3919, "step": 27585 }, { "epoch": 68.965, "grad_norm": 7.53125, "learning_rate": 4.640364834695884e-06, "loss": 1.2694, "step": 27586 }, { "epoch": 68.9675, "grad_norm": 6.96875, "learning_rate": 4.639681279642899e-06, "loss": 1.1217, "step": 27587 }, { "epoch": 68.97, "grad_norm": 6.46875, "learning_rate": 4.638997759731946e-06, "loss": 1.2714, "step": 27588 }, { "epoch": 68.9725, "grad_norm": 5.90625, "learning_rate": 4.638314274967509e-06, "loss": 1.2588, "step": 27589 }, { "epoch": 68.975, "grad_norm": 6.375, "learning_rate": 4.637630825354072e-06, "loss": 1.1481, "step": 27590 }, { "epoch": 68.9775, "grad_norm": 6.34375, "learning_rate": 4.6369474108961145e-06, "loss": 1.2861, "step": 27591 }, { "epoch": 68.98, "grad_norm": 7.34375, "learning_rate": 4.63626403159811e-06, "loss": 1.2194, "step": 27592 }, { "epoch": 68.9825, "grad_norm": 8.75, "learning_rate": 4.6355806874645485e-06, "loss": 1.2919, "step": 27593 }, { "epoch": 68.985, "grad_norm": 6.40625, "learning_rate": 4.634897378499901e-06, "loss": 1.3908, "step": 27594 }, { "epoch": 68.9875, "grad_norm": 6.96875, "learning_rate": 4.634214104708656e-06, "loss": 1.2198, "step": 27595 }, { "epoch": 68.99, "grad_norm": 5.34375, "learning_rate": 4.6335308660952824e-06, "loss": 1.4524, "step": 27596 }, { "epoch": 68.9925, "grad_norm": 7.375, "learning_rate": 4.6328476626642714e-06, "loss": 1.2323, "step": 27597 }, { "epoch": 68.995, "grad_norm": 7.96875, "learning_rate": 4.6321644944200975e-06, "loss": 1.3132, "step": 27598 }, { "epoch": 68.9975, "grad_norm": 8.375, "learning_rate": 4.6314813613672295e-06, "loss": 1.373, "step": 27599 }, { "epoch": 69.0, "grad_norm": 10.8125, "learning_rate": 4.630798263510162e-06, "loss": 1.2198, "step": 27600 }, { "epoch": 69.0025, "grad_norm": 8.25, "learning_rate": 4.630115200853367e-06, "loss": 1.3397, "step": 27601 }, { "epoch": 69.005, "grad_norm": 6.21875, "learning_rate": 4.6294321734013215e-06, "loss": 1.1982, "step": 27602 }, { "epoch": 69.0075, "grad_norm": 10.6875, "learning_rate": 4.628749181158499e-06, "loss": 1.331, "step": 27603 }, { "epoch": 69.01, "grad_norm": 7.5625, "learning_rate": 4.6280662241293835e-06, "loss": 1.4183, "step": 27604 }, { "epoch": 69.0125, "grad_norm": 6.09375, "learning_rate": 4.6273833023184545e-06, "loss": 1.3488, "step": 27605 }, { "epoch": 69.015, "grad_norm": 7.3125, "learning_rate": 4.626700415730185e-06, "loss": 1.4305, "step": 27606 }, { "epoch": 69.0175, "grad_norm": 5.1875, "learning_rate": 4.626017564369049e-06, "loss": 1.3272, "step": 27607 }, { "epoch": 69.02, "grad_norm": 16.75, "learning_rate": 4.625334748239531e-06, "loss": 1.1743, "step": 27608 }, { "epoch": 69.0225, "grad_norm": 9.25, "learning_rate": 4.624651967346098e-06, "loss": 1.2888, "step": 27609 }, { "epoch": 69.025, "grad_norm": 6.65625, "learning_rate": 4.6239692216932365e-06, "loss": 1.2966, "step": 27610 }, { "epoch": 69.0275, "grad_norm": 6.3125, "learning_rate": 4.623286511285414e-06, "loss": 1.4425, "step": 27611 }, { "epoch": 69.03, "grad_norm": 10.375, "learning_rate": 4.622603836127113e-06, "loss": 1.2875, "step": 27612 }, { "epoch": 69.0325, "grad_norm": 6.125, "learning_rate": 4.621921196222807e-06, "loss": 1.4064, "step": 27613 }, { "epoch": 69.035, "grad_norm": 10.4375, "learning_rate": 4.6212385915769665e-06, "loss": 1.5576, "step": 27614 }, { "epoch": 69.0375, "grad_norm": 9.5625, "learning_rate": 4.620556022194071e-06, "loss": 1.3454, "step": 27615 }, { "epoch": 69.04, "grad_norm": 7.8125, "learning_rate": 4.619873488078597e-06, "loss": 1.2626, "step": 27616 }, { "epoch": 69.0425, "grad_norm": 7.0625, "learning_rate": 4.619190989235019e-06, "loss": 1.3541, "step": 27617 }, { "epoch": 69.045, "grad_norm": 5.8125, "learning_rate": 4.618508525667804e-06, "loss": 1.3373, "step": 27618 }, { "epoch": 69.0475, "grad_norm": 6.59375, "learning_rate": 4.617826097381433e-06, "loss": 1.3518, "step": 27619 }, { "epoch": 69.05, "grad_norm": 6.8125, "learning_rate": 4.617143704380382e-06, "loss": 1.3757, "step": 27620 }, { "epoch": 69.0525, "grad_norm": 7.25, "learning_rate": 4.616461346669121e-06, "loss": 1.2304, "step": 27621 }, { "epoch": 69.055, "grad_norm": 8.1875, "learning_rate": 4.61577902425212e-06, "loss": 1.3228, "step": 27622 }, { "epoch": 69.0575, "grad_norm": 6.59375, "learning_rate": 4.615096737133862e-06, "loss": 1.2509, "step": 27623 }, { "epoch": 69.06, "grad_norm": 5.84375, "learning_rate": 4.614414485318809e-06, "loss": 1.3122, "step": 27624 }, { "epoch": 69.0625, "grad_norm": 6.3125, "learning_rate": 4.613732268811444e-06, "loss": 1.3822, "step": 27625 }, { "epoch": 69.065, "grad_norm": 9.9375, "learning_rate": 4.61305008761623e-06, "loss": 1.2302, "step": 27626 }, { "epoch": 69.0675, "grad_norm": 8.875, "learning_rate": 4.612367941737648e-06, "loss": 1.1654, "step": 27627 }, { "epoch": 69.07, "grad_norm": 8.125, "learning_rate": 4.611685831180166e-06, "loss": 1.4205, "step": 27628 }, { "epoch": 69.0725, "grad_norm": 6.78125, "learning_rate": 4.611003755948252e-06, "loss": 1.2004, "step": 27629 }, { "epoch": 69.075, "grad_norm": 6.5625, "learning_rate": 4.610321716046382e-06, "loss": 1.2891, "step": 27630 }, { "epoch": 69.0775, "grad_norm": 11.0, "learning_rate": 4.6096397114790315e-06, "loss": 1.4216, "step": 27631 }, { "epoch": 69.08, "grad_norm": 6.0, "learning_rate": 4.608957742250667e-06, "loss": 1.3307, "step": 27632 }, { "epoch": 69.0825, "grad_norm": 10.5625, "learning_rate": 4.6082758083657555e-06, "loss": 1.2233, "step": 27633 }, { "epoch": 69.085, "grad_norm": 6.125, "learning_rate": 4.607593909828772e-06, "loss": 1.3416, "step": 27634 }, { "epoch": 69.0875, "grad_norm": 10.6875, "learning_rate": 4.606912046644192e-06, "loss": 1.1124, "step": 27635 }, { "epoch": 69.09, "grad_norm": 7.5, "learning_rate": 4.60623021881648e-06, "loss": 1.3121, "step": 27636 }, { "epoch": 69.0925, "grad_norm": 8.9375, "learning_rate": 4.605548426350104e-06, "loss": 1.3552, "step": 27637 }, { "epoch": 69.095, "grad_norm": 6.96875, "learning_rate": 4.604866669249539e-06, "loss": 1.3666, "step": 27638 }, { "epoch": 69.0975, "grad_norm": 10.0625, "learning_rate": 4.6041849475192494e-06, "loss": 1.3728, "step": 27639 }, { "epoch": 69.1, "grad_norm": 5.71875, "learning_rate": 4.60350326116371e-06, "loss": 1.1683, "step": 27640 }, { "epoch": 69.1025, "grad_norm": 8.5625, "learning_rate": 4.6028216101873845e-06, "loss": 1.388, "step": 27641 }, { "epoch": 69.105, "grad_norm": 6.125, "learning_rate": 4.602139994594748e-06, "loss": 1.2964, "step": 27642 }, { "epoch": 69.1075, "grad_norm": 8.9375, "learning_rate": 4.601458414390266e-06, "loss": 1.2642, "step": 27643 }, { "epoch": 69.11, "grad_norm": 7.75, "learning_rate": 4.600776869578403e-06, "loss": 1.2929, "step": 27644 }, { "epoch": 69.1125, "grad_norm": 6.8125, "learning_rate": 4.60009536016363e-06, "loss": 1.3507, "step": 27645 }, { "epoch": 69.115, "grad_norm": 7.8125, "learning_rate": 4.59941388615042e-06, "loss": 1.3145, "step": 27646 }, { "epoch": 69.1175, "grad_norm": 7.6875, "learning_rate": 4.598732447543236e-06, "loss": 1.2846, "step": 27647 }, { "epoch": 69.12, "grad_norm": 6.15625, "learning_rate": 4.598051044346542e-06, "loss": 1.1559, "step": 27648 }, { "epoch": 69.1225, "grad_norm": 7.75, "learning_rate": 4.59736967656481e-06, "loss": 1.2443, "step": 27649 }, { "epoch": 69.125, "grad_norm": 8.875, "learning_rate": 4.596688344202509e-06, "loss": 1.2978, "step": 27650 }, { "epoch": 69.1275, "grad_norm": 5.375, "learning_rate": 4.5960070472641036e-06, "loss": 1.3178, "step": 27651 }, { "epoch": 69.13, "grad_norm": 7.875, "learning_rate": 4.595325785754056e-06, "loss": 1.4151, "step": 27652 }, { "epoch": 69.1325, "grad_norm": 10.0, "learning_rate": 4.594644559676841e-06, "loss": 1.283, "step": 27653 }, { "epoch": 69.135, "grad_norm": 6.78125, "learning_rate": 4.5939633690369145e-06, "loss": 1.4648, "step": 27654 }, { "epoch": 69.1375, "grad_norm": 8.25, "learning_rate": 4.593282213838753e-06, "loss": 1.3185, "step": 27655 }, { "epoch": 69.14, "grad_norm": 8.0625, "learning_rate": 4.592601094086811e-06, "loss": 1.3013, "step": 27656 }, { "epoch": 69.1425, "grad_norm": 7.53125, "learning_rate": 4.5919200097855655e-06, "loss": 1.2635, "step": 27657 }, { "epoch": 69.145, "grad_norm": 6.15625, "learning_rate": 4.591238960939476e-06, "loss": 1.3232, "step": 27658 }, { "epoch": 69.1475, "grad_norm": 6.71875, "learning_rate": 4.590557947553003e-06, "loss": 1.4664, "step": 27659 }, { "epoch": 69.15, "grad_norm": 9.125, "learning_rate": 4.589876969630616e-06, "loss": 1.2514, "step": 27660 }, { "epoch": 69.1525, "grad_norm": 7.21875, "learning_rate": 4.589196027176782e-06, "loss": 1.3404, "step": 27661 }, { "epoch": 69.155, "grad_norm": 6.46875, "learning_rate": 4.588515120195964e-06, "loss": 1.2806, "step": 27662 }, { "epoch": 69.1575, "grad_norm": 7.84375, "learning_rate": 4.587834248692619e-06, "loss": 1.3727, "step": 27663 }, { "epoch": 69.16, "grad_norm": 5.84375, "learning_rate": 4.587153412671217e-06, "loss": 1.2613, "step": 27664 }, { "epoch": 69.1625, "grad_norm": 9.125, "learning_rate": 4.5864726121362235e-06, "loss": 1.4298, "step": 27665 }, { "epoch": 69.165, "grad_norm": 8.875, "learning_rate": 4.5857918470920995e-06, "loss": 1.3987, "step": 27666 }, { "epoch": 69.1675, "grad_norm": 6.53125, "learning_rate": 4.585111117543304e-06, "loss": 1.3043, "step": 27667 }, { "epoch": 69.17, "grad_norm": 7.28125, "learning_rate": 4.584430423494306e-06, "loss": 1.271, "step": 27668 }, { "epoch": 69.1725, "grad_norm": 7.75, "learning_rate": 4.583749764949563e-06, "loss": 1.3362, "step": 27669 }, { "epoch": 69.175, "grad_norm": 8.625, "learning_rate": 4.583069141913543e-06, "loss": 1.2346, "step": 27670 }, { "epoch": 69.1775, "grad_norm": 6.96875, "learning_rate": 4.582388554390701e-06, "loss": 1.2468, "step": 27671 }, { "epoch": 69.18, "grad_norm": 7.90625, "learning_rate": 4.581708002385506e-06, "loss": 1.2557, "step": 27672 }, { "epoch": 69.1825, "grad_norm": 7.75, "learning_rate": 4.581027485902417e-06, "loss": 1.4126, "step": 27673 }, { "epoch": 69.185, "grad_norm": 7.34375, "learning_rate": 4.580347004945892e-06, "loss": 1.2265, "step": 27674 }, { "epoch": 69.1875, "grad_norm": 7.9375, "learning_rate": 4.579666559520395e-06, "loss": 1.3245, "step": 27675 }, { "epoch": 69.19, "grad_norm": 12.125, "learning_rate": 4.57898614963039e-06, "loss": 1.2631, "step": 27676 }, { "epoch": 69.1925, "grad_norm": 7.40625, "learning_rate": 4.578305775280335e-06, "loss": 1.5127, "step": 27677 }, { "epoch": 69.195, "grad_norm": 6.875, "learning_rate": 4.577625436474688e-06, "loss": 1.3737, "step": 27678 }, { "epoch": 69.1975, "grad_norm": 7.9375, "learning_rate": 4.5769451332179105e-06, "loss": 1.3971, "step": 27679 }, { "epoch": 69.2, "grad_norm": 7.8125, "learning_rate": 4.576264865514467e-06, "loss": 1.2687, "step": 27680 }, { "epoch": 69.2025, "grad_norm": 6.0625, "learning_rate": 4.575584633368815e-06, "loss": 1.4084, "step": 27681 }, { "epoch": 69.205, "grad_norm": 8.25, "learning_rate": 4.574904436785409e-06, "loss": 1.303, "step": 27682 }, { "epoch": 69.2075, "grad_norm": 8.0625, "learning_rate": 4.574224275768715e-06, "loss": 1.5407, "step": 27683 }, { "epoch": 69.21, "grad_norm": 6.46875, "learning_rate": 4.573544150323187e-06, "loss": 1.3627, "step": 27684 }, { "epoch": 69.2125, "grad_norm": 9.4375, "learning_rate": 4.57286406045329e-06, "loss": 1.2788, "step": 27685 }, { "epoch": 69.215, "grad_norm": 6.8125, "learning_rate": 4.572184006163475e-06, "loss": 1.2983, "step": 27686 }, { "epoch": 69.2175, "grad_norm": 5.78125, "learning_rate": 4.571503987458209e-06, "loss": 1.3568, "step": 27687 }, { "epoch": 69.22, "grad_norm": 11.25, "learning_rate": 4.570824004341945e-06, "loss": 1.3791, "step": 27688 }, { "epoch": 69.2225, "grad_norm": 6.96875, "learning_rate": 4.570144056819138e-06, "loss": 1.2178, "step": 27689 }, { "epoch": 69.225, "grad_norm": 6.75, "learning_rate": 4.569464144894249e-06, "loss": 1.2123, "step": 27690 }, { "epoch": 69.2275, "grad_norm": 6.9375, "learning_rate": 4.56878426857174e-06, "loss": 1.1701, "step": 27691 }, { "epoch": 69.23, "grad_norm": 9.375, "learning_rate": 4.568104427856063e-06, "loss": 1.4235, "step": 27692 }, { "epoch": 69.2325, "grad_norm": 11.1875, "learning_rate": 4.5674246227516725e-06, "loss": 1.4034, "step": 27693 }, { "epoch": 69.235, "grad_norm": 9.8125, "learning_rate": 4.56674485326303e-06, "loss": 1.3836, "step": 27694 }, { "epoch": 69.2375, "grad_norm": 9.6875, "learning_rate": 4.566065119394592e-06, "loss": 1.4989, "step": 27695 }, { "epoch": 69.24, "grad_norm": 7.09375, "learning_rate": 4.565385421150817e-06, "loss": 1.4037, "step": 27696 }, { "epoch": 69.2425, "grad_norm": 6.875, "learning_rate": 4.564705758536152e-06, "loss": 1.2692, "step": 27697 }, { "epoch": 69.245, "grad_norm": 8.9375, "learning_rate": 4.564026131555063e-06, "loss": 1.2429, "step": 27698 }, { "epoch": 69.2475, "grad_norm": 7.125, "learning_rate": 4.563346540211996e-06, "loss": 1.3659, "step": 27699 }, { "epoch": 69.25, "grad_norm": 9.1875, "learning_rate": 4.562666984511416e-06, "loss": 1.4227, "step": 27700 }, { "epoch": 69.2525, "grad_norm": 10.125, "learning_rate": 4.56198746445777e-06, "loss": 1.3271, "step": 27701 }, { "epoch": 69.255, "grad_norm": 8.6875, "learning_rate": 4.56130798005552e-06, "loss": 1.438, "step": 27702 }, { "epoch": 69.2575, "grad_norm": 4.96875, "learning_rate": 4.560628531309118e-06, "loss": 1.334, "step": 27703 }, { "epoch": 69.26, "grad_norm": 7.78125, "learning_rate": 4.5599491182230134e-06, "loss": 1.2912, "step": 27704 }, { "epoch": 69.2625, "grad_norm": 11.8125, "learning_rate": 4.559269740801666e-06, "loss": 1.2896, "step": 27705 }, { "epoch": 69.265, "grad_norm": 5.46875, "learning_rate": 4.5585903990495305e-06, "loss": 1.374, "step": 27706 }, { "epoch": 69.2675, "grad_norm": 8.8125, "learning_rate": 4.557911092971059e-06, "loss": 1.302, "step": 27707 }, { "epoch": 69.27, "grad_norm": 8.3125, "learning_rate": 4.557231822570704e-06, "loss": 1.376, "step": 27708 }, { "epoch": 69.2725, "grad_norm": 6.125, "learning_rate": 4.556552587852914e-06, "loss": 1.2574, "step": 27709 }, { "epoch": 69.275, "grad_norm": 8.5, "learning_rate": 4.5558733888221525e-06, "loss": 1.4481, "step": 27710 }, { "epoch": 69.2775, "grad_norm": 8.375, "learning_rate": 4.5551942254828695e-06, "loss": 1.5774, "step": 27711 }, { "epoch": 69.28, "grad_norm": 8.875, "learning_rate": 4.554515097839511e-06, "loss": 1.4012, "step": 27712 }, { "epoch": 69.2825, "grad_norm": 5.9375, "learning_rate": 4.553836005896536e-06, "loss": 1.4192, "step": 27713 }, { "epoch": 69.285, "grad_norm": 5.6875, "learning_rate": 4.5531569496583905e-06, "loss": 1.4475, "step": 27714 }, { "epoch": 69.2875, "grad_norm": 6.78125, "learning_rate": 4.552477929129534e-06, "loss": 1.1859, "step": 27715 }, { "epoch": 69.29, "grad_norm": 5.6875, "learning_rate": 4.551798944314412e-06, "loss": 1.1803, "step": 27716 }, { "epoch": 69.2925, "grad_norm": 6.375, "learning_rate": 4.55111999521748e-06, "loss": 1.3464, "step": 27717 }, { "epoch": 69.295, "grad_norm": 3.921875, "learning_rate": 4.550441081843187e-06, "loss": 1.2132, "step": 27718 }, { "epoch": 69.2975, "grad_norm": 6.9375, "learning_rate": 4.549762204195982e-06, "loss": 1.3552, "step": 27719 }, { "epoch": 69.3, "grad_norm": 8.6875, "learning_rate": 4.549083362280318e-06, "loss": 1.3451, "step": 27720 }, { "epoch": 69.3025, "grad_norm": 11.9375, "learning_rate": 4.548404556100647e-06, "loss": 1.3398, "step": 27721 }, { "epoch": 69.305, "grad_norm": 9.8125, "learning_rate": 4.54772578566142e-06, "loss": 1.2326, "step": 27722 }, { "epoch": 69.3075, "grad_norm": 7.15625, "learning_rate": 4.547047050967083e-06, "loss": 1.2803, "step": 27723 }, { "epoch": 69.31, "grad_norm": 8.5, "learning_rate": 4.54636835202208e-06, "loss": 1.2585, "step": 27724 }, { "epoch": 69.3125, "grad_norm": 7.0, "learning_rate": 4.545689688830877e-06, "loss": 1.3284, "step": 27725 }, { "epoch": 69.315, "grad_norm": 8.0625, "learning_rate": 4.545011061397911e-06, "loss": 1.4249, "step": 27726 }, { "epoch": 69.3175, "grad_norm": 9.0625, "learning_rate": 4.544332469727637e-06, "loss": 1.2755, "step": 27727 }, { "epoch": 69.32, "grad_norm": 8.25, "learning_rate": 4.543653913824496e-06, "loss": 1.2827, "step": 27728 }, { "epoch": 69.3225, "grad_norm": 7.40625, "learning_rate": 4.542975393692942e-06, "loss": 1.4472, "step": 27729 }, { "epoch": 69.325, "grad_norm": 8.9375, "learning_rate": 4.542296909337426e-06, "loss": 1.3282, "step": 27730 }, { "epoch": 69.3275, "grad_norm": 9.25, "learning_rate": 4.5416184607623935e-06, "loss": 1.3078, "step": 27731 }, { "epoch": 69.33, "grad_norm": 6.25, "learning_rate": 4.540940047972289e-06, "loss": 1.3912, "step": 27732 }, { "epoch": 69.3325, "grad_norm": 4.8125, "learning_rate": 4.540261670971565e-06, "loss": 1.3554, "step": 27733 }, { "epoch": 69.335, "grad_norm": 8.5, "learning_rate": 4.5395833297646655e-06, "loss": 1.3522, "step": 27734 }, { "epoch": 69.3375, "grad_norm": 8.4375, "learning_rate": 4.538905024356042e-06, "loss": 1.3107, "step": 27735 }, { "epoch": 69.34, "grad_norm": 8.8125, "learning_rate": 4.5382267547501335e-06, "loss": 1.4659, "step": 27736 }, { "epoch": 69.3425, "grad_norm": 5.5625, "learning_rate": 4.537548520951397e-06, "loss": 1.2938, "step": 27737 }, { "epoch": 69.345, "grad_norm": 8.5625, "learning_rate": 4.5368703229642744e-06, "loss": 1.3223, "step": 27738 }, { "epoch": 69.3475, "grad_norm": 6.15625, "learning_rate": 4.5361921607932025e-06, "loss": 1.3586, "step": 27739 }, { "epoch": 69.35, "grad_norm": 10.0, "learning_rate": 4.535514034442644e-06, "loss": 1.3147, "step": 27740 }, { "epoch": 69.3525, "grad_norm": 6.1875, "learning_rate": 4.534835943917037e-06, "loss": 1.2784, "step": 27741 }, { "epoch": 69.355, "grad_norm": 8.3125, "learning_rate": 4.534157889220826e-06, "loss": 1.4153, "step": 27742 }, { "epoch": 69.3575, "grad_norm": 8.375, "learning_rate": 4.533479870358455e-06, "loss": 1.4263, "step": 27743 }, { "epoch": 69.36, "grad_norm": 6.875, "learning_rate": 4.53280188733437e-06, "loss": 1.2754, "step": 27744 }, { "epoch": 69.3625, "grad_norm": 5.71875, "learning_rate": 4.532123940153022e-06, "loss": 1.4493, "step": 27745 }, { "epoch": 69.365, "grad_norm": 6.4375, "learning_rate": 4.531446028818849e-06, "loss": 1.3615, "step": 27746 }, { "epoch": 69.3675, "grad_norm": 10.0625, "learning_rate": 4.530768153336293e-06, "loss": 1.3374, "step": 27747 }, { "epoch": 69.37, "grad_norm": 6.8125, "learning_rate": 4.530090313709808e-06, "loss": 1.4921, "step": 27748 }, { "epoch": 69.3725, "grad_norm": 7.3125, "learning_rate": 4.5294125099438265e-06, "loss": 1.6121, "step": 27749 }, { "epoch": 69.375, "grad_norm": 8.375, "learning_rate": 4.528734742042803e-06, "loss": 1.2249, "step": 27750 }, { "epoch": 69.3775, "grad_norm": 16.625, "learning_rate": 4.528057010011171e-06, "loss": 1.3329, "step": 27751 }, { "epoch": 69.38, "grad_norm": 6.21875, "learning_rate": 4.527379313853381e-06, "loss": 1.1792, "step": 27752 }, { "epoch": 69.3825, "grad_norm": 8.1875, "learning_rate": 4.526701653573875e-06, "loss": 1.4011, "step": 27753 }, { "epoch": 69.385, "grad_norm": 5.9375, "learning_rate": 4.5260240291770885e-06, "loss": 1.2424, "step": 27754 }, { "epoch": 69.3875, "grad_norm": 12.75, "learning_rate": 4.525346440667471e-06, "loss": 1.5543, "step": 27755 }, { "epoch": 69.39, "grad_norm": 8.9375, "learning_rate": 4.524668888049466e-06, "loss": 1.3068, "step": 27756 }, { "epoch": 69.3925, "grad_norm": 7.46875, "learning_rate": 4.523991371327513e-06, "loss": 1.3544, "step": 27757 }, { "epoch": 69.395, "grad_norm": 7.09375, "learning_rate": 4.523313890506049e-06, "loss": 1.3306, "step": 27758 }, { "epoch": 69.3975, "grad_norm": 7.25, "learning_rate": 4.52263644558952e-06, "loss": 1.2954, "step": 27759 }, { "epoch": 69.4, "grad_norm": 6.40625, "learning_rate": 4.521959036582372e-06, "loss": 1.2462, "step": 27760 }, { "epoch": 69.4025, "grad_norm": 8.5, "learning_rate": 4.521281663489041e-06, "loss": 1.3532, "step": 27761 }, { "epoch": 69.405, "grad_norm": 8.9375, "learning_rate": 4.520604326313964e-06, "loss": 1.2545, "step": 27762 }, { "epoch": 69.4075, "grad_norm": 9.1875, "learning_rate": 4.519927025061586e-06, "loss": 1.3292, "step": 27763 }, { "epoch": 69.41, "grad_norm": 8.1875, "learning_rate": 4.519249759736351e-06, "loss": 1.4004, "step": 27764 }, { "epoch": 69.4125, "grad_norm": 7.78125, "learning_rate": 4.5185725303426955e-06, "loss": 1.2914, "step": 27765 }, { "epoch": 69.415, "grad_norm": 7.4375, "learning_rate": 4.517895336885054e-06, "loss": 1.4732, "step": 27766 }, { "epoch": 69.4175, "grad_norm": 5.6875, "learning_rate": 4.517218179367876e-06, "loss": 1.2721, "step": 27767 }, { "epoch": 69.42, "grad_norm": 7.0, "learning_rate": 4.516541057795593e-06, "loss": 1.2762, "step": 27768 }, { "epoch": 69.4225, "grad_norm": 8.875, "learning_rate": 4.51586397217265e-06, "loss": 1.2391, "step": 27769 }, { "epoch": 69.425, "grad_norm": 13.375, "learning_rate": 4.515186922503481e-06, "loss": 1.3481, "step": 27770 }, { "epoch": 69.4275, "grad_norm": 9.6875, "learning_rate": 4.5145099087925294e-06, "loss": 1.2298, "step": 27771 }, { "epoch": 69.43, "grad_norm": 10.0, "learning_rate": 4.513832931044232e-06, "loss": 1.2905, "step": 27772 }, { "epoch": 69.4325, "grad_norm": 6.59375, "learning_rate": 4.513155989263023e-06, "loss": 1.3441, "step": 27773 }, { "epoch": 69.435, "grad_norm": 10.1875, "learning_rate": 4.512479083453343e-06, "loss": 1.2583, "step": 27774 }, { "epoch": 69.4375, "grad_norm": 8.25, "learning_rate": 4.511802213619635e-06, "loss": 1.2718, "step": 27775 }, { "epoch": 69.44, "grad_norm": 9.9375, "learning_rate": 4.511125379766332e-06, "loss": 1.3165, "step": 27776 }, { "epoch": 69.4425, "grad_norm": 9.5625, "learning_rate": 4.510448581897867e-06, "loss": 1.3236, "step": 27777 }, { "epoch": 69.445, "grad_norm": 7.40625, "learning_rate": 4.509771820018682e-06, "loss": 1.2618, "step": 27778 }, { "epoch": 69.4475, "grad_norm": 7.25, "learning_rate": 4.509095094133215e-06, "loss": 1.3273, "step": 27779 }, { "epoch": 69.45, "grad_norm": 8.8125, "learning_rate": 4.508418404245903e-06, "loss": 1.3927, "step": 27780 }, { "epoch": 69.4525, "grad_norm": 9.625, "learning_rate": 4.507741750361177e-06, "loss": 1.436, "step": 27781 }, { "epoch": 69.455, "grad_norm": 9.25, "learning_rate": 4.507065132483478e-06, "loss": 1.2441, "step": 27782 }, { "epoch": 69.4575, "grad_norm": 7.3125, "learning_rate": 4.506388550617238e-06, "loss": 1.2665, "step": 27783 }, { "epoch": 69.46, "grad_norm": 10.6875, "learning_rate": 4.5057120047668975e-06, "loss": 1.2503, "step": 27784 }, { "epoch": 69.4625, "grad_norm": 6.0, "learning_rate": 4.505035494936886e-06, "loss": 1.2716, "step": 27785 }, { "epoch": 69.465, "grad_norm": 9.8125, "learning_rate": 4.504359021131646e-06, "loss": 1.3649, "step": 27786 }, { "epoch": 69.4675, "grad_norm": 10.6875, "learning_rate": 4.503682583355607e-06, "loss": 1.2625, "step": 27787 }, { "epoch": 69.47, "grad_norm": 9.25, "learning_rate": 4.503006181613203e-06, "loss": 1.3027, "step": 27788 }, { "epoch": 69.4725, "grad_norm": 7.5625, "learning_rate": 4.50232981590887e-06, "loss": 1.3154, "step": 27789 }, { "epoch": 69.475, "grad_norm": 5.46875, "learning_rate": 4.501653486247046e-06, "loss": 1.2634, "step": 27790 }, { "epoch": 69.4775, "grad_norm": 6.8125, "learning_rate": 4.500977192632162e-06, "loss": 1.3232, "step": 27791 }, { "epoch": 69.48, "grad_norm": 6.875, "learning_rate": 4.500300935068647e-06, "loss": 1.1834, "step": 27792 }, { "epoch": 69.4825, "grad_norm": 7.375, "learning_rate": 4.49962471356094e-06, "loss": 1.1486, "step": 27793 }, { "epoch": 69.485, "grad_norm": 8.5625, "learning_rate": 4.498948528113477e-06, "loss": 1.4961, "step": 27794 }, { "epoch": 69.4875, "grad_norm": 8.3125, "learning_rate": 4.498272378730687e-06, "loss": 1.2094, "step": 27795 }, { "epoch": 69.49, "grad_norm": 5.96875, "learning_rate": 4.497596265416999e-06, "loss": 1.1629, "step": 27796 }, { "epoch": 69.4925, "grad_norm": 5.4375, "learning_rate": 4.496920188176853e-06, "loss": 1.4065, "step": 27797 }, { "epoch": 69.495, "grad_norm": 8.1875, "learning_rate": 4.496244147014675e-06, "loss": 1.2178, "step": 27798 }, { "epoch": 69.4975, "grad_norm": 8.625, "learning_rate": 4.495568141934904e-06, "loss": 1.2491, "step": 27799 }, { "epoch": 69.5, "grad_norm": 7.53125, "learning_rate": 4.494892172941965e-06, "loss": 1.5451, "step": 27800 }, { "epoch": 69.5025, "grad_norm": 7.5625, "learning_rate": 4.494216240040296e-06, "loss": 1.2518, "step": 27801 }, { "epoch": 69.505, "grad_norm": 10.5625, "learning_rate": 4.493540343234324e-06, "loss": 1.2472, "step": 27802 }, { "epoch": 69.5075, "grad_norm": 8.0625, "learning_rate": 4.492864482528477e-06, "loss": 1.4095, "step": 27803 }, { "epoch": 69.51, "grad_norm": 7.3125, "learning_rate": 4.492188657927191e-06, "loss": 1.4307, "step": 27804 }, { "epoch": 69.5125, "grad_norm": 10.625, "learning_rate": 4.4915128694349e-06, "loss": 1.3505, "step": 27805 }, { "epoch": 69.515, "grad_norm": 6.3125, "learning_rate": 4.49083711705603e-06, "loss": 1.3928, "step": 27806 }, { "epoch": 69.5175, "grad_norm": 10.3125, "learning_rate": 4.490161400795007e-06, "loss": 1.3893, "step": 27807 }, { "epoch": 69.52, "grad_norm": 9.75, "learning_rate": 4.489485720656266e-06, "loss": 1.2139, "step": 27808 }, { "epoch": 69.5225, "grad_norm": 11.125, "learning_rate": 4.488810076644239e-06, "loss": 1.3682, "step": 27809 }, { "epoch": 69.525, "grad_norm": 7.0, "learning_rate": 4.488134468763353e-06, "loss": 1.2721, "step": 27810 }, { "epoch": 69.5275, "grad_norm": 10.5, "learning_rate": 4.4874588970180334e-06, "loss": 1.3571, "step": 27811 }, { "epoch": 69.53, "grad_norm": 11.125, "learning_rate": 4.486783361412717e-06, "loss": 1.3196, "step": 27812 }, { "epoch": 69.5325, "grad_norm": 10.6875, "learning_rate": 4.4861078619518225e-06, "loss": 1.3151, "step": 27813 }, { "epoch": 69.535, "grad_norm": 7.125, "learning_rate": 4.48543239863979e-06, "loss": 1.4725, "step": 27814 }, { "epoch": 69.5375, "grad_norm": 6.4375, "learning_rate": 4.4847569714810365e-06, "loss": 1.3236, "step": 27815 }, { "epoch": 69.54, "grad_norm": 6.21875, "learning_rate": 4.4840815804800005e-06, "loss": 1.2388, "step": 27816 }, { "epoch": 69.5425, "grad_norm": 6.5, "learning_rate": 4.483406225641105e-06, "loss": 1.2947, "step": 27817 }, { "epoch": 69.545, "grad_norm": 8.4375, "learning_rate": 4.482730906968773e-06, "loss": 1.2062, "step": 27818 }, { "epoch": 69.5475, "grad_norm": 9.0, "learning_rate": 4.4820556244674375e-06, "loss": 1.3086, "step": 27819 }, { "epoch": 69.55, "grad_norm": 8.125, "learning_rate": 4.481380378141528e-06, "loss": 1.4268, "step": 27820 }, { "epoch": 69.5525, "grad_norm": 6.09375, "learning_rate": 4.480705167995467e-06, "loss": 1.1171, "step": 27821 }, { "epoch": 69.555, "grad_norm": 5.46875, "learning_rate": 4.4800299940336785e-06, "loss": 1.2279, "step": 27822 }, { "epoch": 69.5575, "grad_norm": 6.5, "learning_rate": 4.479354856260593e-06, "loss": 1.3553, "step": 27823 }, { "epoch": 69.56, "grad_norm": 9.0, "learning_rate": 4.478679754680639e-06, "loss": 1.2257, "step": 27824 }, { "epoch": 69.5625, "grad_norm": 9.25, "learning_rate": 4.478004689298241e-06, "loss": 1.2757, "step": 27825 }, { "epoch": 69.565, "grad_norm": 7.96875, "learning_rate": 4.477329660117819e-06, "loss": 1.319, "step": 27826 }, { "epoch": 69.5675, "grad_norm": 6.3125, "learning_rate": 4.476654667143806e-06, "loss": 1.2832, "step": 27827 }, { "epoch": 69.57, "grad_norm": 4.90625, "learning_rate": 4.4759797103806205e-06, "loss": 1.2929, "step": 27828 }, { "epoch": 69.5725, "grad_norm": 7.5, "learning_rate": 4.475304789832695e-06, "loss": 1.3629, "step": 27829 }, { "epoch": 69.575, "grad_norm": 5.21875, "learning_rate": 4.474629905504446e-06, "loss": 1.2868, "step": 27830 }, { "epoch": 69.5775, "grad_norm": 7.34375, "learning_rate": 4.473955057400306e-06, "loss": 1.5215, "step": 27831 }, { "epoch": 69.58, "grad_norm": 7.03125, "learning_rate": 4.473280245524696e-06, "loss": 1.3316, "step": 27832 }, { "epoch": 69.5825, "grad_norm": 7.0, "learning_rate": 4.472605469882035e-06, "loss": 1.4536, "step": 27833 }, { "epoch": 69.585, "grad_norm": 7.0, "learning_rate": 4.471930730476752e-06, "loss": 1.261, "step": 27834 }, { "epoch": 69.5875, "grad_norm": 7.0, "learning_rate": 4.471256027313273e-06, "loss": 1.3778, "step": 27835 }, { "epoch": 69.59, "grad_norm": 8.125, "learning_rate": 4.470581360396019e-06, "loss": 1.2283, "step": 27836 }, { "epoch": 69.5925, "grad_norm": 7.5625, "learning_rate": 4.469906729729408e-06, "loss": 1.2966, "step": 27837 }, { "epoch": 69.595, "grad_norm": 7.1875, "learning_rate": 4.469232135317868e-06, "loss": 1.3213, "step": 27838 }, { "epoch": 69.5975, "grad_norm": 7.875, "learning_rate": 4.468557577165824e-06, "loss": 1.4119, "step": 27839 }, { "epoch": 69.6, "grad_norm": 6.15625, "learning_rate": 4.467883055277696e-06, "loss": 1.1995, "step": 27840 }, { "epoch": 69.6025, "grad_norm": 9.875, "learning_rate": 4.467208569657899e-06, "loss": 1.4579, "step": 27841 }, { "epoch": 69.605, "grad_norm": 7.78125, "learning_rate": 4.466534120310867e-06, "loss": 1.2998, "step": 27842 }, { "epoch": 69.6075, "grad_norm": 6.03125, "learning_rate": 4.465859707241012e-06, "loss": 1.2665, "step": 27843 }, { "epoch": 69.61, "grad_norm": 9.125, "learning_rate": 4.465185330452763e-06, "loss": 1.3064, "step": 27844 }, { "epoch": 69.6125, "grad_norm": 13.625, "learning_rate": 4.464510989950534e-06, "loss": 1.5021, "step": 27845 }, { "epoch": 69.615, "grad_norm": 10.75, "learning_rate": 4.463836685738752e-06, "loss": 1.3619, "step": 27846 }, { "epoch": 69.6175, "grad_norm": 9.5, "learning_rate": 4.463162417821836e-06, "loss": 1.3283, "step": 27847 }, { "epoch": 69.62, "grad_norm": 5.15625, "learning_rate": 4.462488186204201e-06, "loss": 1.2121, "step": 27848 }, { "epoch": 69.6225, "grad_norm": 7.6875, "learning_rate": 4.4618139908902735e-06, "loss": 1.2946, "step": 27849 }, { "epoch": 69.625, "grad_norm": 10.125, "learning_rate": 4.461139831884475e-06, "loss": 1.2471, "step": 27850 }, { "epoch": 69.6275, "grad_norm": 6.65625, "learning_rate": 4.460465709191222e-06, "loss": 1.2938, "step": 27851 }, { "epoch": 69.63, "grad_norm": 6.6875, "learning_rate": 4.45979162281493e-06, "loss": 1.2952, "step": 27852 }, { "epoch": 69.6325, "grad_norm": 6.40625, "learning_rate": 4.459117572760022e-06, "loss": 1.4174, "step": 27853 }, { "epoch": 69.635, "grad_norm": 8.5625, "learning_rate": 4.4584435590309215e-06, "loss": 1.1788, "step": 27854 }, { "epoch": 69.6375, "grad_norm": 6.96875, "learning_rate": 4.4577695816320445e-06, "loss": 1.2994, "step": 27855 }, { "epoch": 69.64, "grad_norm": 7.375, "learning_rate": 4.457095640567804e-06, "loss": 1.3257, "step": 27856 }, { "epoch": 69.6425, "grad_norm": 6.0, "learning_rate": 4.456421735842625e-06, "loss": 1.2798, "step": 27857 }, { "epoch": 69.645, "grad_norm": 5.75, "learning_rate": 4.455747867460921e-06, "loss": 1.4452, "step": 27858 }, { "epoch": 69.6475, "grad_norm": 7.375, "learning_rate": 4.4550740354271145e-06, "loss": 1.2999, "step": 27859 }, { "epoch": 69.65, "grad_norm": 6.9375, "learning_rate": 4.454400239745619e-06, "loss": 1.3051, "step": 27860 }, { "epoch": 69.6525, "grad_norm": 10.625, "learning_rate": 4.453726480420856e-06, "loss": 1.2223, "step": 27861 }, { "epoch": 69.655, "grad_norm": 4.90625, "learning_rate": 4.4530527574572405e-06, "loss": 1.3199, "step": 27862 }, { "epoch": 69.6575, "grad_norm": 7.6875, "learning_rate": 4.452379070859185e-06, "loss": 1.3445, "step": 27863 }, { "epoch": 69.66, "grad_norm": 7.15625, "learning_rate": 4.451705420631111e-06, "loss": 1.3259, "step": 27864 }, { "epoch": 69.6625, "grad_norm": 7.71875, "learning_rate": 4.451031806777438e-06, "loss": 1.4303, "step": 27865 }, { "epoch": 69.665, "grad_norm": 10.25, "learning_rate": 4.4503582293025786e-06, "loss": 1.3667, "step": 27866 }, { "epoch": 69.6675, "grad_norm": 10.125, "learning_rate": 4.449684688210948e-06, "loss": 1.3547, "step": 27867 }, { "epoch": 69.67, "grad_norm": 7.15625, "learning_rate": 4.449011183506955e-06, "loss": 1.3861, "step": 27868 }, { "epoch": 69.6725, "grad_norm": 6.90625, "learning_rate": 4.44833771519503e-06, "loss": 1.2809, "step": 27869 }, { "epoch": 69.675, "grad_norm": 8.25, "learning_rate": 4.44766428327958e-06, "loss": 1.4067, "step": 27870 }, { "epoch": 69.6775, "grad_norm": 5.90625, "learning_rate": 4.446990887765023e-06, "loss": 1.2219, "step": 27871 }, { "epoch": 69.68, "grad_norm": 6.0625, "learning_rate": 4.4463175286557654e-06, "loss": 1.2774, "step": 27872 }, { "epoch": 69.6825, "grad_norm": 6.53125, "learning_rate": 4.445644205956229e-06, "loss": 1.3873, "step": 27873 }, { "epoch": 69.685, "grad_norm": 8.5, "learning_rate": 4.444970919670831e-06, "loss": 1.3302, "step": 27874 }, { "epoch": 69.6875, "grad_norm": 8.6875, "learning_rate": 4.444297669803981e-06, "loss": 1.355, "step": 27875 }, { "epoch": 69.69, "grad_norm": 5.9375, "learning_rate": 4.44362445636009e-06, "loss": 1.227, "step": 27876 }, { "epoch": 69.6925, "grad_norm": 6.90625, "learning_rate": 4.442951279343577e-06, "loss": 1.3256, "step": 27877 }, { "epoch": 69.695, "grad_norm": 6.65625, "learning_rate": 4.44227813875885e-06, "loss": 1.2429, "step": 27878 }, { "epoch": 69.6975, "grad_norm": 8.8125, "learning_rate": 4.441605034610325e-06, "loss": 1.2621, "step": 27879 }, { "epoch": 69.7, "grad_norm": 6.3125, "learning_rate": 4.440931966902419e-06, "loss": 1.2372, "step": 27880 }, { "epoch": 69.7025, "grad_norm": 8.0625, "learning_rate": 4.44025893563954e-06, "loss": 1.4251, "step": 27881 }, { "epoch": 69.705, "grad_norm": 8.8125, "learning_rate": 4.439585940826101e-06, "loss": 1.2552, "step": 27882 }, { "epoch": 69.7075, "grad_norm": 10.5, "learning_rate": 4.4389129824665075e-06, "loss": 1.3275, "step": 27883 }, { "epoch": 69.71, "grad_norm": 5.9375, "learning_rate": 4.438240060565185e-06, "loss": 1.4479, "step": 27884 }, { "epoch": 69.7125, "grad_norm": 7.0625, "learning_rate": 4.437567175126538e-06, "loss": 1.2578, "step": 27885 }, { "epoch": 69.715, "grad_norm": 8.4375, "learning_rate": 4.436894326154978e-06, "loss": 1.1783, "step": 27886 }, { "epoch": 69.7175, "grad_norm": 5.84375, "learning_rate": 4.436221513654913e-06, "loss": 1.3385, "step": 27887 }, { "epoch": 69.72, "grad_norm": 6.34375, "learning_rate": 4.435548737630756e-06, "loss": 1.429, "step": 27888 }, { "epoch": 69.7225, "grad_norm": 7.59375, "learning_rate": 4.434875998086923e-06, "loss": 1.3231, "step": 27889 }, { "epoch": 69.725, "grad_norm": 8.375, "learning_rate": 4.43420329502782e-06, "loss": 1.3906, "step": 27890 }, { "epoch": 69.7275, "grad_norm": 13.3125, "learning_rate": 4.433530628457854e-06, "loss": 1.4201, "step": 27891 }, { "epoch": 69.73, "grad_norm": 7.21875, "learning_rate": 4.4328579983814425e-06, "loss": 1.2571, "step": 27892 }, { "epoch": 69.7325, "grad_norm": 6.6875, "learning_rate": 4.432185404802987e-06, "loss": 1.3073, "step": 27893 }, { "epoch": 69.735, "grad_norm": 7.03125, "learning_rate": 4.431512847726905e-06, "loss": 1.3288, "step": 27894 }, { "epoch": 69.7375, "grad_norm": 7.34375, "learning_rate": 4.4308403271575975e-06, "loss": 1.5027, "step": 27895 }, { "epoch": 69.74, "grad_norm": 6.25, "learning_rate": 4.430167843099483e-06, "loss": 1.3604, "step": 27896 }, { "epoch": 69.7425, "grad_norm": 7.90625, "learning_rate": 4.429495395556964e-06, "loss": 1.2952, "step": 27897 }, { "epoch": 69.745, "grad_norm": 6.65625, "learning_rate": 4.4288229845344475e-06, "loss": 1.173, "step": 27898 }, { "epoch": 69.7475, "grad_norm": 8.1875, "learning_rate": 4.428150610036345e-06, "loss": 1.3319, "step": 27899 }, { "epoch": 69.75, "grad_norm": 8.0625, "learning_rate": 4.427478272067066e-06, "loss": 1.1296, "step": 27900 }, { "epoch": 69.7525, "grad_norm": 8.375, "learning_rate": 4.4268059706310175e-06, "loss": 1.3658, "step": 27901 }, { "epoch": 69.755, "grad_norm": 7.0, "learning_rate": 4.426133705732603e-06, "loss": 1.3176, "step": 27902 }, { "epoch": 69.7575, "grad_norm": 7.875, "learning_rate": 4.425461477376232e-06, "loss": 1.23, "step": 27903 }, { "epoch": 69.76, "grad_norm": 7.125, "learning_rate": 4.4247892855663164e-06, "loss": 1.3665, "step": 27904 }, { "epoch": 69.7625, "grad_norm": 6.3125, "learning_rate": 4.424117130307259e-06, "loss": 1.395, "step": 27905 }, { "epoch": 69.765, "grad_norm": 5.875, "learning_rate": 4.423445011603463e-06, "loss": 1.3179, "step": 27906 }, { "epoch": 69.7675, "grad_norm": 7.375, "learning_rate": 4.422772929459342e-06, "loss": 1.2356, "step": 27907 }, { "epoch": 69.77, "grad_norm": 5.6875, "learning_rate": 4.422100883879296e-06, "loss": 1.3172, "step": 27908 }, { "epoch": 69.7725, "grad_norm": 8.875, "learning_rate": 4.4214288748677345e-06, "loss": 1.3134, "step": 27909 }, { "epoch": 69.775, "grad_norm": 11.0, "learning_rate": 4.42075690242906e-06, "loss": 1.3819, "step": 27910 }, { "epoch": 69.7775, "grad_norm": 7.40625, "learning_rate": 4.420084966567683e-06, "loss": 1.3271, "step": 27911 }, { "epoch": 69.78, "grad_norm": 6.71875, "learning_rate": 4.419413067288006e-06, "loss": 1.3902, "step": 27912 }, { "epoch": 69.7825, "grad_norm": 5.6875, "learning_rate": 4.418741204594431e-06, "loss": 1.3323, "step": 27913 }, { "epoch": 69.785, "grad_norm": 7.5625, "learning_rate": 4.4180693784913644e-06, "loss": 1.4093, "step": 27914 }, { "epoch": 69.7875, "grad_norm": 7.84375, "learning_rate": 4.417397588983215e-06, "loss": 1.4159, "step": 27915 }, { "epoch": 69.79, "grad_norm": 6.4375, "learning_rate": 4.416725836074384e-06, "loss": 1.3154, "step": 27916 }, { "epoch": 69.7925, "grad_norm": 12.5625, "learning_rate": 4.416054119769271e-06, "loss": 1.2237, "step": 27917 }, { "epoch": 69.795, "grad_norm": 7.625, "learning_rate": 4.415382440072283e-06, "loss": 1.3289, "step": 27918 }, { "epoch": 69.7975, "grad_norm": 7.28125, "learning_rate": 4.41471079698783e-06, "loss": 1.3034, "step": 27919 }, { "epoch": 69.8, "grad_norm": 4.34375, "learning_rate": 4.414039190520308e-06, "loss": 1.2408, "step": 27920 }, { "epoch": 69.8025, "grad_norm": 8.625, "learning_rate": 4.413367620674119e-06, "loss": 1.3243, "step": 27921 }, { "epoch": 69.805, "grad_norm": 5.25, "learning_rate": 4.412696087453671e-06, "loss": 1.2013, "step": 27922 }, { "epoch": 69.8075, "grad_norm": 12.6875, "learning_rate": 4.412024590863359e-06, "loss": 1.2971, "step": 27923 }, { "epoch": 69.81, "grad_norm": 6.53125, "learning_rate": 4.411353130907595e-06, "loss": 1.4134, "step": 27924 }, { "epoch": 69.8125, "grad_norm": 6.9375, "learning_rate": 4.410681707590774e-06, "loss": 1.2558, "step": 27925 }, { "epoch": 69.815, "grad_norm": 7.28125, "learning_rate": 4.410010320917301e-06, "loss": 1.3608, "step": 27926 }, { "epoch": 69.8175, "grad_norm": 8.0625, "learning_rate": 4.409338970891579e-06, "loss": 1.2592, "step": 27927 }, { "epoch": 69.82, "grad_norm": 15.5625, "learning_rate": 4.408667657518002e-06, "loss": 1.4121, "step": 27928 }, { "epoch": 69.8225, "grad_norm": 6.59375, "learning_rate": 4.407996380800976e-06, "loss": 1.2663, "step": 27929 }, { "epoch": 69.825, "grad_norm": 6.34375, "learning_rate": 4.407325140744907e-06, "loss": 1.1623, "step": 27930 }, { "epoch": 69.8275, "grad_norm": 7.4375, "learning_rate": 4.406653937354189e-06, "loss": 1.3962, "step": 27931 }, { "epoch": 69.83, "grad_norm": 6.90625, "learning_rate": 4.405982770633221e-06, "loss": 1.1515, "step": 27932 }, { "epoch": 69.8325, "grad_norm": 5.46875, "learning_rate": 4.405311640586406e-06, "loss": 1.4149, "step": 27933 }, { "epoch": 69.835, "grad_norm": 6.84375, "learning_rate": 4.404640547218148e-06, "loss": 1.4365, "step": 27934 }, { "epoch": 69.8375, "grad_norm": 7.40625, "learning_rate": 4.403969490532842e-06, "loss": 1.2927, "step": 27935 }, { "epoch": 69.84, "grad_norm": 7.03125, "learning_rate": 4.403298470534885e-06, "loss": 1.4156, "step": 27936 }, { "epoch": 69.8425, "grad_norm": 6.0, "learning_rate": 4.402627487228683e-06, "loss": 1.3693, "step": 27937 }, { "epoch": 69.845, "grad_norm": 8.4375, "learning_rate": 4.401956540618627e-06, "loss": 1.3304, "step": 27938 }, { "epoch": 69.8475, "grad_norm": 9.5625, "learning_rate": 4.4012856307091236e-06, "loss": 1.3429, "step": 27939 }, { "epoch": 69.85, "grad_norm": 11.125, "learning_rate": 4.400614757504565e-06, "loss": 1.4261, "step": 27940 }, { "epoch": 69.8525, "grad_norm": 6.53125, "learning_rate": 4.399943921009354e-06, "loss": 1.4561, "step": 27941 }, { "epoch": 69.855, "grad_norm": 10.4375, "learning_rate": 4.399273121227887e-06, "loss": 1.4539, "step": 27942 }, { "epoch": 69.8575, "grad_norm": 7.6875, "learning_rate": 4.398602358164559e-06, "loss": 1.3456, "step": 27943 }, { "epoch": 69.86, "grad_norm": 7.3125, "learning_rate": 4.397931631823767e-06, "loss": 1.347, "step": 27944 }, { "epoch": 69.8625, "grad_norm": 6.1875, "learning_rate": 4.397260942209918e-06, "loss": 1.2307, "step": 27945 }, { "epoch": 69.865, "grad_norm": 8.8125, "learning_rate": 4.3965902893274005e-06, "loss": 1.3759, "step": 27946 }, { "epoch": 69.8675, "grad_norm": 7.15625, "learning_rate": 4.395919673180609e-06, "loss": 1.324, "step": 27947 }, { "epoch": 69.87, "grad_norm": 9.4375, "learning_rate": 4.395249093773943e-06, "loss": 1.3717, "step": 27948 }, { "epoch": 69.8725, "grad_norm": 6.5625, "learning_rate": 4.394578551111804e-06, "loss": 1.4117, "step": 27949 }, { "epoch": 69.875, "grad_norm": 7.53125, "learning_rate": 4.393908045198585e-06, "loss": 1.339, "step": 27950 }, { "epoch": 69.8775, "grad_norm": 10.3125, "learning_rate": 4.393237576038676e-06, "loss": 1.3547, "step": 27951 }, { "epoch": 69.88, "grad_norm": 5.71875, "learning_rate": 4.39256714363648e-06, "loss": 1.3694, "step": 27952 }, { "epoch": 69.8825, "grad_norm": 6.875, "learning_rate": 4.391896747996387e-06, "loss": 1.2155, "step": 27953 }, { "epoch": 69.885, "grad_norm": 8.5, "learning_rate": 4.391226389122797e-06, "loss": 1.2235, "step": 27954 }, { "epoch": 69.8875, "grad_norm": 8.1875, "learning_rate": 4.390556067020098e-06, "loss": 1.2794, "step": 27955 }, { "epoch": 69.89, "grad_norm": 9.5625, "learning_rate": 4.389885781692695e-06, "loss": 1.3403, "step": 27956 }, { "epoch": 69.8925, "grad_norm": 10.5625, "learning_rate": 4.3892155331449745e-06, "loss": 1.3252, "step": 27957 }, { "epoch": 69.895, "grad_norm": 11.4375, "learning_rate": 4.38854532138133e-06, "loss": 1.1861, "step": 27958 }, { "epoch": 69.8975, "grad_norm": 6.21875, "learning_rate": 4.387875146406158e-06, "loss": 1.0722, "step": 27959 }, { "epoch": 69.9, "grad_norm": 7.0, "learning_rate": 4.3872050082238535e-06, "loss": 1.2087, "step": 27960 }, { "epoch": 69.9025, "grad_norm": 9.75, "learning_rate": 4.3865349068388096e-06, "loss": 1.2646, "step": 27961 }, { "epoch": 69.905, "grad_norm": 7.375, "learning_rate": 4.385864842255415e-06, "loss": 1.4462, "step": 27962 }, { "epoch": 69.9075, "grad_norm": 7.75, "learning_rate": 4.385194814478065e-06, "loss": 1.2941, "step": 27963 }, { "epoch": 69.91, "grad_norm": 12.875, "learning_rate": 4.384524823511157e-06, "loss": 1.3852, "step": 27964 }, { "epoch": 69.9125, "grad_norm": 6.0, "learning_rate": 4.38385486935908e-06, "loss": 1.3788, "step": 27965 }, { "epoch": 69.915, "grad_norm": 8.9375, "learning_rate": 4.383184952026223e-06, "loss": 1.3103, "step": 27966 }, { "epoch": 69.9175, "grad_norm": 6.71875, "learning_rate": 4.382515071516982e-06, "loss": 1.4436, "step": 27967 }, { "epoch": 69.92, "grad_norm": 9.5625, "learning_rate": 4.3818452278357445e-06, "loss": 1.3384, "step": 27968 }, { "epoch": 69.9225, "grad_norm": 5.8125, "learning_rate": 4.381175420986909e-06, "loss": 1.1886, "step": 27969 }, { "epoch": 69.925, "grad_norm": 11.5, "learning_rate": 4.380505650974859e-06, "loss": 1.369, "step": 27970 }, { "epoch": 69.9275, "grad_norm": 8.5625, "learning_rate": 4.379835917803993e-06, "loss": 1.2443, "step": 27971 }, { "epoch": 69.93, "grad_norm": 9.4375, "learning_rate": 4.379166221478697e-06, "loss": 1.2592, "step": 27972 }, { "epoch": 69.9325, "grad_norm": 7.0, "learning_rate": 4.378496562003359e-06, "loss": 1.2272, "step": 27973 }, { "epoch": 69.935, "grad_norm": 7.84375, "learning_rate": 4.377826939382372e-06, "loss": 1.4101, "step": 27974 }, { "epoch": 69.9375, "grad_norm": 11.125, "learning_rate": 4.3771573536201314e-06, "loss": 1.3984, "step": 27975 }, { "epoch": 69.94, "grad_norm": 6.96875, "learning_rate": 4.376487804721022e-06, "loss": 1.3097, "step": 27976 }, { "epoch": 69.9425, "grad_norm": 6.0625, "learning_rate": 4.37581829268943e-06, "loss": 1.2864, "step": 27977 }, { "epoch": 69.945, "grad_norm": 10.1875, "learning_rate": 4.375148817529747e-06, "loss": 1.2807, "step": 27978 }, { "epoch": 69.9475, "grad_norm": 7.40625, "learning_rate": 4.374479379246368e-06, "loss": 1.189, "step": 27979 }, { "epoch": 69.95, "grad_norm": 6.34375, "learning_rate": 4.373809977843676e-06, "loss": 1.3867, "step": 27980 }, { "epoch": 69.9525, "grad_norm": 8.5, "learning_rate": 4.373140613326058e-06, "loss": 1.3656, "step": 27981 }, { "epoch": 69.955, "grad_norm": 9.5625, "learning_rate": 4.372471285697909e-06, "loss": 1.298, "step": 27982 }, { "epoch": 69.9575, "grad_norm": 20.875, "learning_rate": 4.371801994963608e-06, "loss": 1.5235, "step": 27983 }, { "epoch": 69.96, "grad_norm": 7.0, "learning_rate": 4.371132741127553e-06, "loss": 1.3074, "step": 27984 }, { "epoch": 69.9625, "grad_norm": 9.9375, "learning_rate": 4.3704635241941215e-06, "loss": 1.2544, "step": 27985 }, { "epoch": 69.965, "grad_norm": 6.125, "learning_rate": 4.369794344167711e-06, "loss": 1.4261, "step": 27986 }, { "epoch": 69.9675, "grad_norm": 8.0, "learning_rate": 4.369125201052702e-06, "loss": 1.2581, "step": 27987 }, { "epoch": 69.97, "grad_norm": 7.71875, "learning_rate": 4.36845609485348e-06, "loss": 1.3045, "step": 27988 }, { "epoch": 69.9725, "grad_norm": 8.375, "learning_rate": 4.367787025574434e-06, "loss": 1.2987, "step": 27989 }, { "epoch": 69.975, "grad_norm": 7.5625, "learning_rate": 4.3671179932199546e-06, "loss": 1.3558, "step": 27990 }, { "epoch": 69.9775, "grad_norm": 6.3125, "learning_rate": 4.3664489977944246e-06, "loss": 1.1714, "step": 27991 }, { "epoch": 69.98, "grad_norm": 9.0625, "learning_rate": 4.3657800393022255e-06, "loss": 1.3268, "step": 27992 }, { "epoch": 69.9825, "grad_norm": 9.625, "learning_rate": 4.365111117747747e-06, "loss": 1.3394, "step": 27993 }, { "epoch": 69.985, "grad_norm": 9.5, "learning_rate": 4.364442233135379e-06, "loss": 1.2828, "step": 27994 }, { "epoch": 69.9875, "grad_norm": 6.3125, "learning_rate": 4.363773385469501e-06, "loss": 1.1732, "step": 27995 }, { "epoch": 69.99, "grad_norm": 6.8125, "learning_rate": 4.363104574754495e-06, "loss": 1.2217, "step": 27996 }, { "epoch": 69.9925, "grad_norm": 5.375, "learning_rate": 4.362435800994751e-06, "loss": 1.2862, "step": 27997 }, { "epoch": 69.995, "grad_norm": 6.78125, "learning_rate": 4.361767064194656e-06, "loss": 1.3145, "step": 27998 }, { "epoch": 69.9975, "grad_norm": 10.625, "learning_rate": 4.361098364358591e-06, "loss": 1.3472, "step": 27999 }, { "epoch": 70.0, "grad_norm": 6.4375, "learning_rate": 4.360429701490935e-06, "loss": 1.2733, "step": 28000 }, { "epoch": 70.0025, "grad_norm": 5.125, "learning_rate": 4.35976107559608e-06, "loss": 1.2535, "step": 28001 }, { "epoch": 70.005, "grad_norm": 7.0, "learning_rate": 4.359092486678401e-06, "loss": 1.3484, "step": 28002 }, { "epoch": 70.0075, "grad_norm": 6.375, "learning_rate": 4.358423934742291e-06, "loss": 1.2366, "step": 28003 }, { "epoch": 70.01, "grad_norm": 10.0625, "learning_rate": 4.357755419792123e-06, "loss": 1.2581, "step": 28004 }, { "epoch": 70.0125, "grad_norm": 6.28125, "learning_rate": 4.3570869418322905e-06, "loss": 1.3806, "step": 28005 }, { "epoch": 70.015, "grad_norm": 6.875, "learning_rate": 4.356418500867168e-06, "loss": 1.2733, "step": 28006 }, { "epoch": 70.0175, "grad_norm": 8.0, "learning_rate": 4.3557500969011364e-06, "loss": 1.2868, "step": 28007 }, { "epoch": 70.02, "grad_norm": 7.0625, "learning_rate": 4.355081729938583e-06, "loss": 1.3032, "step": 28008 }, { "epoch": 70.0225, "grad_norm": 8.3125, "learning_rate": 4.35441339998389e-06, "loss": 1.2946, "step": 28009 }, { "epoch": 70.025, "grad_norm": 8.4375, "learning_rate": 4.353745107041438e-06, "loss": 1.3899, "step": 28010 }, { "epoch": 70.0275, "grad_norm": 8.875, "learning_rate": 4.353076851115606e-06, "loss": 1.1298, "step": 28011 }, { "epoch": 70.03, "grad_norm": 5.75, "learning_rate": 4.35240863221077e-06, "loss": 1.3695, "step": 28012 }, { "epoch": 70.0325, "grad_norm": 7.4375, "learning_rate": 4.351740450331323e-06, "loss": 1.2964, "step": 28013 }, { "epoch": 70.035, "grad_norm": 5.59375, "learning_rate": 4.35107230548164e-06, "loss": 1.3121, "step": 28014 }, { "epoch": 70.0375, "grad_norm": 8.1875, "learning_rate": 4.350404197666098e-06, "loss": 1.2722, "step": 28015 }, { "epoch": 70.04, "grad_norm": 5.9375, "learning_rate": 4.349736126889084e-06, "loss": 1.2735, "step": 28016 }, { "epoch": 70.0425, "grad_norm": 6.375, "learning_rate": 4.349068093154969e-06, "loss": 1.3263, "step": 28017 }, { "epoch": 70.045, "grad_norm": 7.03125, "learning_rate": 4.348400096468142e-06, "loss": 1.1765, "step": 28018 }, { "epoch": 70.0475, "grad_norm": 8.375, "learning_rate": 4.347732136832973e-06, "loss": 1.3228, "step": 28019 }, { "epoch": 70.05, "grad_norm": 6.875, "learning_rate": 4.34706421425385e-06, "loss": 1.3537, "step": 28020 }, { "epoch": 70.0525, "grad_norm": 8.375, "learning_rate": 4.346396328735148e-06, "loss": 1.3212, "step": 28021 }, { "epoch": 70.055, "grad_norm": 6.625, "learning_rate": 4.345728480281242e-06, "loss": 1.305, "step": 28022 }, { "epoch": 70.0575, "grad_norm": 6.375, "learning_rate": 4.345060668896513e-06, "loss": 1.2622, "step": 28023 }, { "epoch": 70.06, "grad_norm": 9.6875, "learning_rate": 4.344392894585344e-06, "loss": 1.3313, "step": 28024 }, { "epoch": 70.0625, "grad_norm": 5.0, "learning_rate": 4.34372515735211e-06, "loss": 1.2662, "step": 28025 }, { "epoch": 70.065, "grad_norm": 7.75, "learning_rate": 4.3430574572011865e-06, "loss": 1.3511, "step": 28026 }, { "epoch": 70.0675, "grad_norm": 7.53125, "learning_rate": 4.342389794136944e-06, "loss": 1.2624, "step": 28027 }, { "epoch": 70.07, "grad_norm": 6.75, "learning_rate": 4.341722168163776e-06, "loss": 1.2168, "step": 28028 }, { "epoch": 70.0725, "grad_norm": 7.40625, "learning_rate": 4.34105457928605e-06, "loss": 1.3487, "step": 28029 }, { "epoch": 70.075, "grad_norm": 5.125, "learning_rate": 4.340387027508145e-06, "loss": 1.3596, "step": 28030 }, { "epoch": 70.0775, "grad_norm": 7.21875, "learning_rate": 4.339719512834431e-06, "loss": 1.3527, "step": 28031 }, { "epoch": 70.08, "grad_norm": 8.8125, "learning_rate": 4.339052035269291e-06, "loss": 1.2099, "step": 28032 }, { "epoch": 70.0825, "grad_norm": 7.3125, "learning_rate": 4.338384594817102e-06, "loss": 1.3248, "step": 28033 }, { "epoch": 70.085, "grad_norm": 7.28125, "learning_rate": 4.337717191482237e-06, "loss": 1.291, "step": 28034 }, { "epoch": 70.0875, "grad_norm": 8.1875, "learning_rate": 4.3370498252690684e-06, "loss": 1.3544, "step": 28035 }, { "epoch": 70.09, "grad_norm": 13.0, "learning_rate": 4.336382496181978e-06, "loss": 1.358, "step": 28036 }, { "epoch": 70.0925, "grad_norm": 5.71875, "learning_rate": 4.335715204225334e-06, "loss": 1.328, "step": 28037 }, { "epoch": 70.095, "grad_norm": 5.84375, "learning_rate": 4.3350479494035165e-06, "loss": 1.4668, "step": 28038 }, { "epoch": 70.0975, "grad_norm": 7.34375, "learning_rate": 4.334380731720896e-06, "loss": 1.1946, "step": 28039 }, { "epoch": 70.1, "grad_norm": 6.15625, "learning_rate": 4.3337135511818514e-06, "loss": 1.2685, "step": 28040 }, { "epoch": 70.1025, "grad_norm": 7.6875, "learning_rate": 4.3330464077907545e-06, "loss": 1.4175, "step": 28041 }, { "epoch": 70.105, "grad_norm": 11.9375, "learning_rate": 4.332379301551971e-06, "loss": 1.3263, "step": 28042 }, { "epoch": 70.1075, "grad_norm": 6.5, "learning_rate": 4.3317122324698905e-06, "loss": 1.229, "step": 28043 }, { "epoch": 70.11, "grad_norm": 8.0625, "learning_rate": 4.331045200548876e-06, "loss": 1.4231, "step": 28044 }, { "epoch": 70.1125, "grad_norm": 6.25, "learning_rate": 4.330378205793303e-06, "loss": 1.1661, "step": 28045 }, { "epoch": 70.115, "grad_norm": 6.625, "learning_rate": 4.3297112482075395e-06, "loss": 1.28, "step": 28046 }, { "epoch": 70.1175, "grad_norm": 7.34375, "learning_rate": 4.329044327795963e-06, "loss": 1.233, "step": 28047 }, { "epoch": 70.12, "grad_norm": 7.8125, "learning_rate": 4.328377444562948e-06, "loss": 1.4453, "step": 28048 }, { "epoch": 70.1225, "grad_norm": 7.1875, "learning_rate": 4.327710598512864e-06, "loss": 1.2154, "step": 28049 }, { "epoch": 70.125, "grad_norm": 4.6875, "learning_rate": 4.327043789650078e-06, "loss": 1.2455, "step": 28050 }, { "epoch": 70.1275, "grad_norm": 7.46875, "learning_rate": 4.32637701797897e-06, "loss": 1.2195, "step": 28051 }, { "epoch": 70.13, "grad_norm": 7.4375, "learning_rate": 4.325710283503903e-06, "loss": 1.4656, "step": 28052 }, { "epoch": 70.1325, "grad_norm": 7.125, "learning_rate": 4.3250435862292575e-06, "loss": 1.3095, "step": 28053 }, { "epoch": 70.135, "grad_norm": 5.90625, "learning_rate": 4.324376926159394e-06, "loss": 1.3028, "step": 28054 }, { "epoch": 70.1375, "grad_norm": 6.09375, "learning_rate": 4.323710303298693e-06, "loss": 1.4367, "step": 28055 }, { "epoch": 70.14, "grad_norm": 9.125, "learning_rate": 4.323043717651518e-06, "loss": 1.2056, "step": 28056 }, { "epoch": 70.1425, "grad_norm": 6.90625, "learning_rate": 4.322377169222239e-06, "loss": 1.456, "step": 28057 }, { "epoch": 70.145, "grad_norm": 8.1875, "learning_rate": 4.321710658015229e-06, "loss": 1.4252, "step": 28058 }, { "epoch": 70.1475, "grad_norm": 5.25, "learning_rate": 4.32104418403486e-06, "loss": 1.4132, "step": 28059 }, { "epoch": 70.15, "grad_norm": 6.28125, "learning_rate": 4.320377747285497e-06, "loss": 1.2276, "step": 28060 }, { "epoch": 70.1525, "grad_norm": 7.59375, "learning_rate": 4.3197113477715065e-06, "loss": 1.1934, "step": 28061 }, { "epoch": 70.155, "grad_norm": 9.6875, "learning_rate": 4.319044985497261e-06, "loss": 1.4013, "step": 28062 }, { "epoch": 70.1575, "grad_norm": 6.28125, "learning_rate": 4.318378660467134e-06, "loss": 1.2518, "step": 28063 }, { "epoch": 70.16, "grad_norm": 7.59375, "learning_rate": 4.31771237268549e-06, "loss": 1.4466, "step": 28064 }, { "epoch": 70.1625, "grad_norm": 8.3125, "learning_rate": 4.317046122156691e-06, "loss": 1.2287, "step": 28065 }, { "epoch": 70.165, "grad_norm": 6.21875, "learning_rate": 4.316379908885113e-06, "loss": 1.2609, "step": 28066 }, { "epoch": 70.1675, "grad_norm": 8.3125, "learning_rate": 4.315713732875118e-06, "loss": 1.2779, "step": 28067 }, { "epoch": 70.17, "grad_norm": 8.0, "learning_rate": 4.315047594131081e-06, "loss": 1.3758, "step": 28068 }, { "epoch": 70.1725, "grad_norm": 7.375, "learning_rate": 4.31438149265736e-06, "loss": 1.3659, "step": 28069 }, { "epoch": 70.175, "grad_norm": 12.5625, "learning_rate": 4.313715428458331e-06, "loss": 1.3224, "step": 28070 }, { "epoch": 70.1775, "grad_norm": 7.84375, "learning_rate": 4.313049401538355e-06, "loss": 1.3539, "step": 28071 }, { "epoch": 70.18, "grad_norm": 6.4375, "learning_rate": 4.312383411901796e-06, "loss": 1.3829, "step": 28072 }, { "epoch": 70.1825, "grad_norm": 11.4375, "learning_rate": 4.311717459553024e-06, "loss": 1.3722, "step": 28073 }, { "epoch": 70.185, "grad_norm": 8.9375, "learning_rate": 4.311051544496408e-06, "loss": 1.1985, "step": 28074 }, { "epoch": 70.1875, "grad_norm": 6.6875, "learning_rate": 4.310385666736311e-06, "loss": 1.3941, "step": 28075 }, { "epoch": 70.19, "grad_norm": 7.46875, "learning_rate": 4.309719826277094e-06, "loss": 1.303, "step": 28076 }, { "epoch": 70.1925, "grad_norm": 14.625, "learning_rate": 4.309054023123126e-06, "loss": 1.3414, "step": 28077 }, { "epoch": 70.195, "grad_norm": 5.8125, "learning_rate": 4.308388257278776e-06, "loss": 1.2842, "step": 28078 }, { "epoch": 70.1975, "grad_norm": 7.28125, "learning_rate": 4.307722528748405e-06, "loss": 1.3395, "step": 28079 }, { "epoch": 70.2, "grad_norm": 8.125, "learning_rate": 4.307056837536373e-06, "loss": 1.4228, "step": 28080 }, { "epoch": 70.2025, "grad_norm": 7.46875, "learning_rate": 4.306391183647052e-06, "loss": 1.4223, "step": 28081 }, { "epoch": 70.205, "grad_norm": 9.0625, "learning_rate": 4.3057255670848e-06, "loss": 1.2977, "step": 28082 }, { "epoch": 70.2075, "grad_norm": 6.9375, "learning_rate": 4.305059987853986e-06, "loss": 1.4511, "step": 28083 }, { "epoch": 70.21, "grad_norm": 7.09375, "learning_rate": 4.304394445958967e-06, "loss": 1.4231, "step": 28084 }, { "epoch": 70.2125, "grad_norm": 8.6875, "learning_rate": 4.303728941404115e-06, "loss": 1.3984, "step": 28085 }, { "epoch": 70.215, "grad_norm": 7.59375, "learning_rate": 4.303063474193787e-06, "loss": 1.3771, "step": 28086 }, { "epoch": 70.2175, "grad_norm": 7.90625, "learning_rate": 4.302398044332343e-06, "loss": 1.353, "step": 28087 }, { "epoch": 70.22, "grad_norm": 7.03125, "learning_rate": 4.301732651824151e-06, "loss": 1.4927, "step": 28088 }, { "epoch": 70.2225, "grad_norm": 11.9375, "learning_rate": 4.301067296673574e-06, "loss": 1.3716, "step": 28089 }, { "epoch": 70.225, "grad_norm": 10.0625, "learning_rate": 4.300401978884973e-06, "loss": 1.4436, "step": 28090 }, { "epoch": 70.2275, "grad_norm": 5.96875, "learning_rate": 4.299736698462703e-06, "loss": 1.3853, "step": 28091 }, { "epoch": 70.23, "grad_norm": 6.53125, "learning_rate": 4.299071455411132e-06, "loss": 1.3735, "step": 28092 }, { "epoch": 70.2325, "grad_norm": 8.5625, "learning_rate": 4.298406249734624e-06, "loss": 1.3108, "step": 28093 }, { "epoch": 70.235, "grad_norm": 7.40625, "learning_rate": 4.297741081437536e-06, "loss": 1.2643, "step": 28094 }, { "epoch": 70.2375, "grad_norm": 7.4375, "learning_rate": 4.297075950524228e-06, "loss": 1.3313, "step": 28095 }, { "epoch": 70.24, "grad_norm": 5.46875, "learning_rate": 4.296410856999062e-06, "loss": 1.2877, "step": 28096 }, { "epoch": 70.2425, "grad_norm": 9.125, "learning_rate": 4.295745800866397e-06, "loss": 1.2609, "step": 28097 }, { "epoch": 70.245, "grad_norm": 9.3125, "learning_rate": 4.295080782130597e-06, "loss": 1.4713, "step": 28098 }, { "epoch": 70.2475, "grad_norm": 8.1875, "learning_rate": 4.294415800796016e-06, "loss": 1.2552, "step": 28099 }, { "epoch": 70.25, "grad_norm": 9.4375, "learning_rate": 4.2937508568670194e-06, "loss": 1.2282, "step": 28100 }, { "epoch": 70.2525, "grad_norm": 9.75, "learning_rate": 4.293085950347965e-06, "loss": 1.0979, "step": 28101 }, { "epoch": 70.255, "grad_norm": 8.375, "learning_rate": 4.292421081243206e-06, "loss": 1.2051, "step": 28102 }, { "epoch": 70.2575, "grad_norm": 6.78125, "learning_rate": 4.291756249557106e-06, "loss": 1.3674, "step": 28103 }, { "epoch": 70.26, "grad_norm": 7.46875, "learning_rate": 4.291091455294028e-06, "loss": 1.4546, "step": 28104 }, { "epoch": 70.2625, "grad_norm": 8.9375, "learning_rate": 4.290426698458327e-06, "loss": 1.3869, "step": 28105 }, { "epoch": 70.265, "grad_norm": 7.875, "learning_rate": 4.289761979054354e-06, "loss": 1.2775, "step": 28106 }, { "epoch": 70.2675, "grad_norm": 14.25, "learning_rate": 4.289097297086474e-06, "loss": 1.296, "step": 28107 }, { "epoch": 70.27, "grad_norm": 8.0625, "learning_rate": 4.2884326525590466e-06, "loss": 1.2447, "step": 28108 }, { "epoch": 70.2725, "grad_norm": 7.0625, "learning_rate": 4.287768045476428e-06, "loss": 1.3185, "step": 28109 }, { "epoch": 70.275, "grad_norm": 5.46875, "learning_rate": 4.287103475842969e-06, "loss": 1.2907, "step": 28110 }, { "epoch": 70.2775, "grad_norm": 9.875, "learning_rate": 4.286438943663034e-06, "loss": 1.1792, "step": 28111 }, { "epoch": 70.28, "grad_norm": 8.3125, "learning_rate": 4.2857744489409725e-06, "loss": 1.3882, "step": 28112 }, { "epoch": 70.2825, "grad_norm": 8.875, "learning_rate": 4.28510999168115e-06, "loss": 1.4389, "step": 28113 }, { "epoch": 70.285, "grad_norm": 9.0, "learning_rate": 4.284445571887914e-06, "loss": 1.358, "step": 28114 }, { "epoch": 70.2875, "grad_norm": 10.375, "learning_rate": 4.283781189565626e-06, "loss": 1.5442, "step": 28115 }, { "epoch": 70.29, "grad_norm": 7.09375, "learning_rate": 4.2831168447186425e-06, "loss": 1.3466, "step": 28116 }, { "epoch": 70.2925, "grad_norm": 7.0625, "learning_rate": 4.2824525373513105e-06, "loss": 1.2403, "step": 28117 }, { "epoch": 70.295, "grad_norm": 10.0625, "learning_rate": 4.281788267467991e-06, "loss": 1.2335, "step": 28118 }, { "epoch": 70.2975, "grad_norm": 8.625, "learning_rate": 4.2811240350730435e-06, "loss": 1.3298, "step": 28119 }, { "epoch": 70.3, "grad_norm": 7.65625, "learning_rate": 4.280459840170818e-06, "loss": 1.2066, "step": 28120 }, { "epoch": 70.3025, "grad_norm": 6.84375, "learning_rate": 4.279795682765664e-06, "loss": 1.2109, "step": 28121 }, { "epoch": 70.305, "grad_norm": 10.125, "learning_rate": 4.279131562861941e-06, "loss": 1.2434, "step": 28122 }, { "epoch": 70.3075, "grad_norm": 9.0, "learning_rate": 4.278467480464007e-06, "loss": 1.3594, "step": 28123 }, { "epoch": 70.31, "grad_norm": 12.75, "learning_rate": 4.277803435576211e-06, "loss": 1.4573, "step": 28124 }, { "epoch": 70.3125, "grad_norm": 8.6875, "learning_rate": 4.277139428202902e-06, "loss": 1.2836, "step": 28125 }, { "epoch": 70.315, "grad_norm": 9.0, "learning_rate": 4.2764754583484424e-06, "loss": 1.2351, "step": 28126 }, { "epoch": 70.3175, "grad_norm": 4.90625, "learning_rate": 4.275811526017177e-06, "loss": 1.2207, "step": 28127 }, { "epoch": 70.32, "grad_norm": 5.1875, "learning_rate": 4.2751476312134655e-06, "loss": 1.4112, "step": 28128 }, { "epoch": 70.3225, "grad_norm": 7.90625, "learning_rate": 4.274483773941653e-06, "loss": 1.2447, "step": 28129 }, { "epoch": 70.325, "grad_norm": 8.875, "learning_rate": 4.273819954206101e-06, "loss": 1.3173, "step": 28130 }, { "epoch": 70.3275, "grad_norm": 13.125, "learning_rate": 4.2731561720111544e-06, "loss": 1.2161, "step": 28131 }, { "epoch": 70.33, "grad_norm": 10.3125, "learning_rate": 4.272492427361165e-06, "loss": 1.3081, "step": 28132 }, { "epoch": 70.3325, "grad_norm": 5.875, "learning_rate": 4.271828720260485e-06, "loss": 1.1715, "step": 28133 }, { "epoch": 70.335, "grad_norm": 5.15625, "learning_rate": 4.271165050713469e-06, "loss": 1.2132, "step": 28134 }, { "epoch": 70.3375, "grad_norm": 6.5, "learning_rate": 4.270501418724467e-06, "loss": 1.1882, "step": 28135 }, { "epoch": 70.34, "grad_norm": 13.4375, "learning_rate": 4.2698378242978254e-06, "loss": 1.5123, "step": 28136 }, { "epoch": 70.3425, "grad_norm": 6.46875, "learning_rate": 4.269174267437897e-06, "loss": 1.1861, "step": 28137 }, { "epoch": 70.345, "grad_norm": 7.15625, "learning_rate": 4.268510748149037e-06, "loss": 1.2442, "step": 28138 }, { "epoch": 70.3475, "grad_norm": 5.4375, "learning_rate": 4.267847266435589e-06, "loss": 1.2516, "step": 28139 }, { "epoch": 70.35, "grad_norm": 5.53125, "learning_rate": 4.267183822301903e-06, "loss": 1.298, "step": 28140 }, { "epoch": 70.3525, "grad_norm": 7.46875, "learning_rate": 4.2665204157523344e-06, "loss": 1.3244, "step": 28141 }, { "epoch": 70.355, "grad_norm": 9.125, "learning_rate": 4.265857046791225e-06, "loss": 1.4547, "step": 28142 }, { "epoch": 70.3575, "grad_norm": 6.0, "learning_rate": 4.265193715422931e-06, "loss": 1.4846, "step": 28143 }, { "epoch": 70.36, "grad_norm": 7.78125, "learning_rate": 4.264530421651792e-06, "loss": 1.38, "step": 28144 }, { "epoch": 70.3625, "grad_norm": 7.34375, "learning_rate": 4.263867165482167e-06, "loss": 1.3178, "step": 28145 }, { "epoch": 70.365, "grad_norm": 8.5, "learning_rate": 4.2632039469183995e-06, "loss": 1.2923, "step": 28146 }, { "epoch": 70.3675, "grad_norm": 8.0625, "learning_rate": 4.262540765964835e-06, "loss": 1.5007, "step": 28147 }, { "epoch": 70.37, "grad_norm": 8.5625, "learning_rate": 4.261877622625821e-06, "loss": 1.2059, "step": 28148 }, { "epoch": 70.3725, "grad_norm": 10.125, "learning_rate": 4.2612145169057125e-06, "loss": 1.3943, "step": 28149 }, { "epoch": 70.375, "grad_norm": 8.0, "learning_rate": 4.260551448808852e-06, "loss": 1.3841, "step": 28150 }, { "epoch": 70.3775, "grad_norm": 7.21875, "learning_rate": 4.259888418339586e-06, "loss": 1.3361, "step": 28151 }, { "epoch": 70.38, "grad_norm": 6.75, "learning_rate": 4.259225425502256e-06, "loss": 1.3872, "step": 28152 }, { "epoch": 70.3825, "grad_norm": 6.53125, "learning_rate": 4.258562470301221e-06, "loss": 1.3387, "step": 28153 }, { "epoch": 70.385, "grad_norm": 8.6875, "learning_rate": 4.25789955274082e-06, "loss": 1.3, "step": 28154 }, { "epoch": 70.3875, "grad_norm": 8.6875, "learning_rate": 4.257236672825395e-06, "loss": 1.1865, "step": 28155 }, { "epoch": 70.39, "grad_norm": 9.125, "learning_rate": 4.256573830559303e-06, "loss": 1.2932, "step": 28156 }, { "epoch": 70.3925, "grad_norm": 7.0, "learning_rate": 4.2559110259468785e-06, "loss": 1.3227, "step": 28157 }, { "epoch": 70.395, "grad_norm": 7.34375, "learning_rate": 4.2552482589924735e-06, "loss": 1.3746, "step": 28158 }, { "epoch": 70.3975, "grad_norm": 8.5625, "learning_rate": 4.254585529700429e-06, "loss": 1.3617, "step": 28159 }, { "epoch": 70.4, "grad_norm": 11.6875, "learning_rate": 4.2539228380750955e-06, "loss": 1.34, "step": 28160 }, { "epoch": 70.4025, "grad_norm": 8.0625, "learning_rate": 4.253260184120813e-06, "loss": 1.2801, "step": 28161 }, { "epoch": 70.405, "grad_norm": 5.1875, "learning_rate": 4.252597567841924e-06, "loss": 1.2619, "step": 28162 }, { "epoch": 70.4075, "grad_norm": 6.5625, "learning_rate": 4.251934989242775e-06, "loss": 1.2181, "step": 28163 }, { "epoch": 70.41, "grad_norm": 6.03125, "learning_rate": 4.251272448327713e-06, "loss": 1.248, "step": 28164 }, { "epoch": 70.4125, "grad_norm": 9.1875, "learning_rate": 4.250609945101081e-06, "loss": 1.2911, "step": 28165 }, { "epoch": 70.415, "grad_norm": 8.1875, "learning_rate": 4.249947479567218e-06, "loss": 1.1219, "step": 28166 }, { "epoch": 70.4175, "grad_norm": 6.9375, "learning_rate": 4.249285051730463e-06, "loss": 1.4184, "step": 28167 }, { "epoch": 70.42, "grad_norm": 7.15625, "learning_rate": 4.248622661595172e-06, "loss": 1.3355, "step": 28168 }, { "epoch": 70.4225, "grad_norm": 6.75, "learning_rate": 4.247960309165681e-06, "loss": 1.3616, "step": 28169 }, { "epoch": 70.425, "grad_norm": 5.9375, "learning_rate": 4.247297994446331e-06, "loss": 1.2569, "step": 28170 }, { "epoch": 70.4275, "grad_norm": 6.53125, "learning_rate": 4.2466357174414626e-06, "loss": 1.1882, "step": 28171 }, { "epoch": 70.43, "grad_norm": 9.1875, "learning_rate": 4.24597347815542e-06, "loss": 1.249, "step": 28172 }, { "epoch": 70.4325, "grad_norm": 7.21875, "learning_rate": 4.245311276592548e-06, "loss": 1.2009, "step": 28173 }, { "epoch": 70.435, "grad_norm": 11.75, "learning_rate": 4.244649112757185e-06, "loss": 1.2763, "step": 28174 }, { "epoch": 70.4375, "grad_norm": 10.0, "learning_rate": 4.24398698665367e-06, "loss": 1.2534, "step": 28175 }, { "epoch": 70.44, "grad_norm": 8.75, "learning_rate": 4.243324898286349e-06, "loss": 1.4065, "step": 28176 }, { "epoch": 70.4425, "grad_norm": 9.25, "learning_rate": 4.2426628476595546e-06, "loss": 1.2982, "step": 28177 }, { "epoch": 70.445, "grad_norm": 9.0, "learning_rate": 4.242000834777637e-06, "loss": 1.3287, "step": 28178 }, { "epoch": 70.4475, "grad_norm": 8.75, "learning_rate": 4.2413388596449275e-06, "loss": 1.2414, "step": 28179 }, { "epoch": 70.45, "grad_norm": 6.8125, "learning_rate": 4.240676922265774e-06, "loss": 1.1876, "step": 28180 }, { "epoch": 70.4525, "grad_norm": 7.46875, "learning_rate": 4.240015022644511e-06, "loss": 1.2663, "step": 28181 }, { "epoch": 70.455, "grad_norm": 7.71875, "learning_rate": 4.239353160785473e-06, "loss": 1.3838, "step": 28182 }, { "epoch": 70.4575, "grad_norm": 9.125, "learning_rate": 4.2386913366930115e-06, "loss": 1.3982, "step": 28183 }, { "epoch": 70.46, "grad_norm": 10.8125, "learning_rate": 4.238029550371459e-06, "loss": 1.3217, "step": 28184 }, { "epoch": 70.4625, "grad_norm": 10.1875, "learning_rate": 4.2373678018251565e-06, "loss": 1.2506, "step": 28185 }, { "epoch": 70.465, "grad_norm": 9.375, "learning_rate": 4.236706091058433e-06, "loss": 1.2462, "step": 28186 }, { "epoch": 70.4675, "grad_norm": 8.0, "learning_rate": 4.236044418075636e-06, "loss": 1.363, "step": 28187 }, { "epoch": 70.47, "grad_norm": 11.875, "learning_rate": 4.2353827828811046e-06, "loss": 1.1625, "step": 28188 }, { "epoch": 70.4725, "grad_norm": 5.25, "learning_rate": 4.234721185479173e-06, "loss": 1.3885, "step": 28189 }, { "epoch": 70.475, "grad_norm": 9.0625, "learning_rate": 4.234059625874174e-06, "loss": 1.3759, "step": 28190 }, { "epoch": 70.4775, "grad_norm": 9.125, "learning_rate": 4.233398104070453e-06, "loss": 1.5092, "step": 28191 }, { "epoch": 70.48, "grad_norm": 9.125, "learning_rate": 4.2327366200723404e-06, "loss": 1.3018, "step": 28192 }, { "epoch": 70.4825, "grad_norm": 6.71875, "learning_rate": 4.232075173884179e-06, "loss": 1.2348, "step": 28193 }, { "epoch": 70.485, "grad_norm": 7.53125, "learning_rate": 4.231413765510298e-06, "loss": 1.2283, "step": 28194 }, { "epoch": 70.4875, "grad_norm": 9.4375, "learning_rate": 4.2307523949550425e-06, "loss": 1.5174, "step": 28195 }, { "epoch": 70.49, "grad_norm": 11.1875, "learning_rate": 4.230091062222742e-06, "loss": 1.2823, "step": 28196 }, { "epoch": 70.4925, "grad_norm": 6.8125, "learning_rate": 4.229429767317731e-06, "loss": 1.3213, "step": 28197 }, { "epoch": 70.495, "grad_norm": 6.65625, "learning_rate": 4.2287685102443465e-06, "loss": 1.2871, "step": 28198 }, { "epoch": 70.4975, "grad_norm": 6.8125, "learning_rate": 4.228107291006928e-06, "loss": 1.4198, "step": 28199 }, { "epoch": 70.5, "grad_norm": 5.96875, "learning_rate": 4.2274461096098085e-06, "loss": 1.2019, "step": 28200 }, { "epoch": 70.5025, "grad_norm": 6.1875, "learning_rate": 4.226784966057317e-06, "loss": 1.3066, "step": 28201 }, { "epoch": 70.505, "grad_norm": 6.78125, "learning_rate": 4.226123860353793e-06, "loss": 1.3556, "step": 28202 }, { "epoch": 70.5075, "grad_norm": 7.25, "learning_rate": 4.225462792503573e-06, "loss": 1.2518, "step": 28203 }, { "epoch": 70.51, "grad_norm": 6.53125, "learning_rate": 4.224801762510987e-06, "loss": 1.4094, "step": 28204 }, { "epoch": 70.5125, "grad_norm": 6.1875, "learning_rate": 4.224140770380367e-06, "loss": 1.2049, "step": 28205 }, { "epoch": 70.515, "grad_norm": 7.6875, "learning_rate": 4.223479816116052e-06, "loss": 1.3349, "step": 28206 }, { "epoch": 70.5175, "grad_norm": 6.375, "learning_rate": 4.222818899722369e-06, "loss": 1.2895, "step": 28207 }, { "epoch": 70.52, "grad_norm": 7.4375, "learning_rate": 4.222158021203657e-06, "loss": 1.2578, "step": 28208 }, { "epoch": 70.5225, "grad_norm": 7.125, "learning_rate": 4.221497180564242e-06, "loss": 1.1966, "step": 28209 }, { "epoch": 70.525, "grad_norm": 8.8125, "learning_rate": 4.220836377808465e-06, "loss": 1.4221, "step": 28210 }, { "epoch": 70.5275, "grad_norm": 5.5625, "learning_rate": 4.220175612940654e-06, "loss": 1.4398, "step": 28211 }, { "epoch": 70.53, "grad_norm": 7.25, "learning_rate": 4.2195148859651344e-06, "loss": 1.3254, "step": 28212 }, { "epoch": 70.5325, "grad_norm": 8.8125, "learning_rate": 4.218854196886245e-06, "loss": 1.3192, "step": 28213 }, { "epoch": 70.535, "grad_norm": 9.6875, "learning_rate": 4.21819354570832e-06, "loss": 1.4557, "step": 28214 }, { "epoch": 70.5375, "grad_norm": 8.6875, "learning_rate": 4.217532932435687e-06, "loss": 1.4177, "step": 28215 }, { "epoch": 70.54, "grad_norm": 6.3125, "learning_rate": 4.2168723570726735e-06, "loss": 1.1981, "step": 28216 }, { "epoch": 70.5425, "grad_norm": 8.5, "learning_rate": 4.216211819623612e-06, "loss": 1.3067, "step": 28217 }, { "epoch": 70.545, "grad_norm": 8.6875, "learning_rate": 4.2155513200928386e-06, "loss": 1.4408, "step": 28218 }, { "epoch": 70.5475, "grad_norm": 10.375, "learning_rate": 4.21489085848468e-06, "loss": 1.25, "step": 28219 }, { "epoch": 70.55, "grad_norm": 9.0625, "learning_rate": 4.21423043480346e-06, "loss": 1.3222, "step": 28220 }, { "epoch": 70.5525, "grad_norm": 8.375, "learning_rate": 4.213570049053519e-06, "loss": 1.3307, "step": 28221 }, { "epoch": 70.555, "grad_norm": 6.4375, "learning_rate": 4.212909701239177e-06, "loss": 1.2523, "step": 28222 }, { "epoch": 70.5575, "grad_norm": 6.53125, "learning_rate": 4.212249391364771e-06, "loss": 1.4222, "step": 28223 }, { "epoch": 70.56, "grad_norm": 6.25, "learning_rate": 4.211589119434622e-06, "loss": 1.3193, "step": 28224 }, { "epoch": 70.5625, "grad_norm": 7.0625, "learning_rate": 4.210928885453068e-06, "loss": 1.2339, "step": 28225 }, { "epoch": 70.565, "grad_norm": 11.375, "learning_rate": 4.2102686894244325e-06, "loss": 1.2521, "step": 28226 }, { "epoch": 70.5675, "grad_norm": 6.0625, "learning_rate": 4.20960853135304e-06, "loss": 1.5317, "step": 28227 }, { "epoch": 70.57, "grad_norm": 6.96875, "learning_rate": 4.208948411243222e-06, "loss": 1.2758, "step": 28228 }, { "epoch": 70.5725, "grad_norm": 5.59375, "learning_rate": 4.208288329099311e-06, "loss": 1.189, "step": 28229 }, { "epoch": 70.575, "grad_norm": 5.75, "learning_rate": 4.207628284925628e-06, "loss": 1.3025, "step": 28230 }, { "epoch": 70.5775, "grad_norm": 8.1875, "learning_rate": 4.2069682787265e-06, "loss": 1.3554, "step": 28231 }, { "epoch": 70.58, "grad_norm": 6.0625, "learning_rate": 4.206308310506255e-06, "loss": 1.339, "step": 28232 }, { "epoch": 70.5825, "grad_norm": 11.0, "learning_rate": 4.205648380269226e-06, "loss": 1.275, "step": 28233 }, { "epoch": 70.585, "grad_norm": 8.0625, "learning_rate": 4.2049884880197335e-06, "loss": 1.2027, "step": 28234 }, { "epoch": 70.5875, "grad_norm": 8.25, "learning_rate": 4.204328633762101e-06, "loss": 1.2867, "step": 28235 }, { "epoch": 70.59, "grad_norm": 6.84375, "learning_rate": 4.203668817500662e-06, "loss": 1.2249, "step": 28236 }, { "epoch": 70.5925, "grad_norm": 5.375, "learning_rate": 4.203009039239734e-06, "loss": 1.0539, "step": 28237 }, { "epoch": 70.595, "grad_norm": 6.78125, "learning_rate": 4.20234929898365e-06, "loss": 1.3849, "step": 28238 }, { "epoch": 70.5975, "grad_norm": 9.3125, "learning_rate": 4.20168959673673e-06, "loss": 1.2859, "step": 28239 }, { "epoch": 70.6, "grad_norm": 8.1875, "learning_rate": 4.201029932503303e-06, "loss": 1.3052, "step": 28240 }, { "epoch": 70.6025, "grad_norm": 9.5, "learning_rate": 4.200370306287689e-06, "loss": 1.2973, "step": 28241 }, { "epoch": 70.605, "grad_norm": 9.75, "learning_rate": 4.199710718094219e-06, "loss": 1.3498, "step": 28242 }, { "epoch": 70.6075, "grad_norm": 6.78125, "learning_rate": 4.199051167927209e-06, "loss": 1.2022, "step": 28243 }, { "epoch": 70.61, "grad_norm": 7.3125, "learning_rate": 4.198391655790991e-06, "loss": 1.3113, "step": 28244 }, { "epoch": 70.6125, "grad_norm": 9.25, "learning_rate": 4.197732181689886e-06, "loss": 1.4388, "step": 28245 }, { "epoch": 70.615, "grad_norm": 7.1875, "learning_rate": 4.197072745628213e-06, "loss": 1.3202, "step": 28246 }, { "epoch": 70.6175, "grad_norm": 7.59375, "learning_rate": 4.196413347610298e-06, "loss": 1.456, "step": 28247 }, { "epoch": 70.62, "grad_norm": 8.6875, "learning_rate": 4.1957539876404685e-06, "loss": 1.3136, "step": 28248 }, { "epoch": 70.6225, "grad_norm": 13.8125, "learning_rate": 4.195094665723044e-06, "loss": 1.3907, "step": 28249 }, { "epoch": 70.625, "grad_norm": 7.5, "learning_rate": 4.194435381862343e-06, "loss": 1.2591, "step": 28250 }, { "epoch": 70.6275, "grad_norm": 9.0625, "learning_rate": 4.193776136062691e-06, "loss": 1.221, "step": 28251 }, { "epoch": 70.63, "grad_norm": 5.84375, "learning_rate": 4.1931169283284134e-06, "loss": 1.3304, "step": 28252 }, { "epoch": 70.6325, "grad_norm": 7.78125, "learning_rate": 4.19245775866383e-06, "loss": 1.3525, "step": 28253 }, { "epoch": 70.635, "grad_norm": 6.40625, "learning_rate": 4.191798627073257e-06, "loss": 1.3059, "step": 28254 }, { "epoch": 70.6375, "grad_norm": 7.125, "learning_rate": 4.1911395335610235e-06, "loss": 1.2929, "step": 28255 }, { "epoch": 70.64, "grad_norm": 6.78125, "learning_rate": 4.190480478131443e-06, "loss": 1.4071, "step": 28256 }, { "epoch": 70.6425, "grad_norm": 5.375, "learning_rate": 4.189821460788844e-06, "loss": 1.4093, "step": 28257 }, { "epoch": 70.645, "grad_norm": 6.96875, "learning_rate": 4.189162481537538e-06, "loss": 1.4949, "step": 28258 }, { "epoch": 70.6475, "grad_norm": 4.6875, "learning_rate": 4.188503540381856e-06, "loss": 1.375, "step": 28259 }, { "epoch": 70.65, "grad_norm": 9.75, "learning_rate": 4.18784463732611e-06, "loss": 1.271, "step": 28260 }, { "epoch": 70.6525, "grad_norm": 8.1875, "learning_rate": 4.18718577237462e-06, "loss": 1.2555, "step": 28261 }, { "epoch": 70.655, "grad_norm": 10.125, "learning_rate": 4.1865269455317074e-06, "loss": 1.3681, "step": 28262 }, { "epoch": 70.6575, "grad_norm": 6.84375, "learning_rate": 4.185868156801695e-06, "loss": 1.3547, "step": 28263 }, { "epoch": 70.66, "grad_norm": 10.0625, "learning_rate": 4.185209406188898e-06, "loss": 1.3664, "step": 28264 }, { "epoch": 70.6625, "grad_norm": 5.03125, "learning_rate": 4.184550693697632e-06, "loss": 1.364, "step": 28265 }, { "epoch": 70.665, "grad_norm": 5.625, "learning_rate": 4.183892019332219e-06, "loss": 1.2822, "step": 28266 }, { "epoch": 70.6675, "grad_norm": 6.0625, "learning_rate": 4.18323338309698e-06, "loss": 1.2024, "step": 28267 }, { "epoch": 70.67, "grad_norm": 8.875, "learning_rate": 4.182574784996231e-06, "loss": 1.2089, "step": 28268 }, { "epoch": 70.6725, "grad_norm": 7.8125, "learning_rate": 4.181916225034285e-06, "loss": 1.3673, "step": 28269 }, { "epoch": 70.675, "grad_norm": 8.125, "learning_rate": 4.181257703215467e-06, "loss": 1.2729, "step": 28270 }, { "epoch": 70.6775, "grad_norm": 6.84375, "learning_rate": 4.180599219544087e-06, "loss": 1.4752, "step": 28271 }, { "epoch": 70.68, "grad_norm": 6.5, "learning_rate": 4.179940774024469e-06, "loss": 1.1063, "step": 28272 }, { "epoch": 70.6825, "grad_norm": 8.125, "learning_rate": 4.179282366660923e-06, "loss": 1.2198, "step": 28273 }, { "epoch": 70.685, "grad_norm": 5.65625, "learning_rate": 4.178623997457773e-06, "loss": 1.1723, "step": 28274 }, { "epoch": 70.6875, "grad_norm": 9.375, "learning_rate": 4.17796566641933e-06, "loss": 1.3352, "step": 28275 }, { "epoch": 70.69, "grad_norm": 5.6875, "learning_rate": 4.177307373549908e-06, "loss": 1.35, "step": 28276 }, { "epoch": 70.6925, "grad_norm": 6.09375, "learning_rate": 4.176649118853826e-06, "loss": 1.2846, "step": 28277 }, { "epoch": 70.695, "grad_norm": 6.375, "learning_rate": 4.175990902335403e-06, "loss": 1.3603, "step": 28278 }, { "epoch": 70.6975, "grad_norm": 7.8125, "learning_rate": 4.175332723998951e-06, "loss": 1.2208, "step": 28279 }, { "epoch": 70.7, "grad_norm": 6.21875, "learning_rate": 4.17467458384878e-06, "loss": 1.3655, "step": 28280 }, { "epoch": 70.7025, "grad_norm": 10.125, "learning_rate": 4.17401648188921e-06, "loss": 1.2898, "step": 28281 }, { "epoch": 70.705, "grad_norm": 5.53125, "learning_rate": 4.173358418124558e-06, "loss": 1.1971, "step": 28282 }, { "epoch": 70.7075, "grad_norm": 7.78125, "learning_rate": 4.172700392559136e-06, "loss": 1.3807, "step": 28283 }, { "epoch": 70.71, "grad_norm": 7.28125, "learning_rate": 4.1720424051972515e-06, "loss": 1.234, "step": 28284 }, { "epoch": 70.7125, "grad_norm": 6.6875, "learning_rate": 4.171384456043229e-06, "loss": 1.2578, "step": 28285 }, { "epoch": 70.715, "grad_norm": 6.59375, "learning_rate": 4.1707265451013735e-06, "loss": 1.3149, "step": 28286 }, { "epoch": 70.7175, "grad_norm": 7.625, "learning_rate": 4.170068672376004e-06, "loss": 1.3457, "step": 28287 }, { "epoch": 70.72, "grad_norm": 8.375, "learning_rate": 4.169410837871427e-06, "loss": 1.2789, "step": 28288 }, { "epoch": 70.7225, "grad_norm": 9.5, "learning_rate": 4.168753041591964e-06, "loss": 1.4046, "step": 28289 }, { "epoch": 70.725, "grad_norm": 8.5, "learning_rate": 4.168095283541922e-06, "loss": 1.4346, "step": 28290 }, { "epoch": 70.7275, "grad_norm": 9.875, "learning_rate": 4.16743756372561e-06, "loss": 1.2679, "step": 28291 }, { "epoch": 70.73, "grad_norm": 7.625, "learning_rate": 4.166779882147345e-06, "loss": 1.2857, "step": 28292 }, { "epoch": 70.7325, "grad_norm": 8.3125, "learning_rate": 4.166122238811441e-06, "loss": 1.3353, "step": 28293 }, { "epoch": 70.735, "grad_norm": 11.25, "learning_rate": 4.165464633722206e-06, "loss": 1.3018, "step": 28294 }, { "epoch": 70.7375, "grad_norm": 5.46875, "learning_rate": 4.164807066883948e-06, "loss": 1.2768, "step": 28295 }, { "epoch": 70.74, "grad_norm": 5.78125, "learning_rate": 4.16414953830098e-06, "loss": 1.4622, "step": 28296 }, { "epoch": 70.7425, "grad_norm": 6.3125, "learning_rate": 4.16349204797762e-06, "loss": 1.3577, "step": 28297 }, { "epoch": 70.745, "grad_norm": 9.3125, "learning_rate": 4.162834595918171e-06, "loss": 1.4521, "step": 28298 }, { "epoch": 70.7475, "grad_norm": 9.0, "learning_rate": 4.162177182126941e-06, "loss": 1.322, "step": 28299 }, { "epoch": 70.75, "grad_norm": 7.90625, "learning_rate": 4.1615198066082475e-06, "loss": 1.3468, "step": 28300 }, { "epoch": 70.7525, "grad_norm": 11.125, "learning_rate": 4.160862469366394e-06, "loss": 1.5619, "step": 28301 }, { "epoch": 70.755, "grad_norm": 5.75, "learning_rate": 4.160205170405694e-06, "loss": 1.3056, "step": 28302 }, { "epoch": 70.7575, "grad_norm": 10.5, "learning_rate": 4.159547909730454e-06, "loss": 1.2473, "step": 28303 }, { "epoch": 70.76, "grad_norm": 6.75, "learning_rate": 4.158890687344986e-06, "loss": 1.3257, "step": 28304 }, { "epoch": 70.7625, "grad_norm": 7.15625, "learning_rate": 4.158233503253598e-06, "loss": 1.4966, "step": 28305 }, { "epoch": 70.765, "grad_norm": 8.1875, "learning_rate": 4.157576357460593e-06, "loss": 1.3086, "step": 28306 }, { "epoch": 70.7675, "grad_norm": 7.65625, "learning_rate": 4.1569192499702845e-06, "loss": 1.3218, "step": 28307 }, { "epoch": 70.77, "grad_norm": 6.96875, "learning_rate": 4.156262180786983e-06, "loss": 1.4844, "step": 28308 }, { "epoch": 70.7725, "grad_norm": 7.53125, "learning_rate": 4.155605149914992e-06, "loss": 1.3202, "step": 28309 }, { "epoch": 70.775, "grad_norm": 7.34375, "learning_rate": 4.15494815735862e-06, "loss": 1.2805, "step": 28310 }, { "epoch": 70.7775, "grad_norm": 10.625, "learning_rate": 4.154291203122167e-06, "loss": 1.446, "step": 28311 }, { "epoch": 70.78, "grad_norm": 8.3125, "learning_rate": 4.153634287209955e-06, "loss": 1.5009, "step": 28312 }, { "epoch": 70.7825, "grad_norm": 7.03125, "learning_rate": 4.152977409626282e-06, "loss": 1.2213, "step": 28313 }, { "epoch": 70.785, "grad_norm": 8.5625, "learning_rate": 4.152320570375454e-06, "loss": 1.389, "step": 28314 }, { "epoch": 70.7875, "grad_norm": 6.3125, "learning_rate": 4.151663769461776e-06, "loss": 1.1406, "step": 28315 }, { "epoch": 70.79, "grad_norm": 11.3125, "learning_rate": 4.151007006889556e-06, "loss": 1.2968, "step": 28316 }, { "epoch": 70.7925, "grad_norm": 4.46875, "learning_rate": 4.150350282663104e-06, "loss": 1.3243, "step": 28317 }, { "epoch": 70.795, "grad_norm": 5.9375, "learning_rate": 4.14969359678672e-06, "loss": 1.4563, "step": 28318 }, { "epoch": 70.7975, "grad_norm": 9.4375, "learning_rate": 4.1490369492647085e-06, "loss": 1.3354, "step": 28319 }, { "epoch": 70.8, "grad_norm": 6.46875, "learning_rate": 4.14838034010138e-06, "loss": 1.5011, "step": 28320 }, { "epoch": 70.8025, "grad_norm": 8.625, "learning_rate": 4.1477237693010315e-06, "loss": 1.2975, "step": 28321 }, { "epoch": 70.805, "grad_norm": 5.4375, "learning_rate": 4.147067236867973e-06, "loss": 1.2581, "step": 28322 }, { "epoch": 70.8075, "grad_norm": 7.84375, "learning_rate": 4.14641074280651e-06, "loss": 1.2353, "step": 28323 }, { "epoch": 70.81, "grad_norm": 6.46875, "learning_rate": 4.1457542871209445e-06, "loss": 1.306, "step": 28324 }, { "epoch": 70.8125, "grad_norm": 7.96875, "learning_rate": 4.145097869815579e-06, "loss": 1.2834, "step": 28325 }, { "epoch": 70.815, "grad_norm": 8.3125, "learning_rate": 4.144441490894711e-06, "loss": 1.4074, "step": 28326 }, { "epoch": 70.8175, "grad_norm": 6.59375, "learning_rate": 4.143785150362657e-06, "loss": 1.0525, "step": 28327 }, { "epoch": 70.82, "grad_norm": 7.125, "learning_rate": 4.143128848223714e-06, "loss": 1.3245, "step": 28328 }, { "epoch": 70.8225, "grad_norm": 10.9375, "learning_rate": 4.142472584482183e-06, "loss": 1.4232, "step": 28329 }, { "epoch": 70.825, "grad_norm": 4.4375, "learning_rate": 4.141816359142365e-06, "loss": 1.3435, "step": 28330 }, { "epoch": 70.8275, "grad_norm": 6.1875, "learning_rate": 4.141160172208563e-06, "loss": 1.2068, "step": 28331 }, { "epoch": 70.83, "grad_norm": 7.3125, "learning_rate": 4.140504023685085e-06, "loss": 1.3402, "step": 28332 }, { "epoch": 70.8325, "grad_norm": 8.875, "learning_rate": 4.139847913576227e-06, "loss": 1.2304, "step": 28333 }, { "epoch": 70.835, "grad_norm": 9.3125, "learning_rate": 4.139191841886288e-06, "loss": 1.4895, "step": 28334 }, { "epoch": 70.8375, "grad_norm": 6.46875, "learning_rate": 4.138535808619578e-06, "loss": 1.2502, "step": 28335 }, { "epoch": 70.84, "grad_norm": 7.0625, "learning_rate": 4.137879813780388e-06, "loss": 1.3429, "step": 28336 }, { "epoch": 70.8425, "grad_norm": 8.75, "learning_rate": 4.137223857373027e-06, "loss": 1.3225, "step": 28337 }, { "epoch": 70.845, "grad_norm": 7.84375, "learning_rate": 4.1365679394017886e-06, "loss": 1.1963, "step": 28338 }, { "epoch": 70.8475, "grad_norm": 6.9375, "learning_rate": 4.135912059870979e-06, "loss": 1.2211, "step": 28339 }, { "epoch": 70.85, "grad_norm": 9.875, "learning_rate": 4.135256218784896e-06, "loss": 1.3751, "step": 28340 }, { "epoch": 70.8525, "grad_norm": 6.03125, "learning_rate": 4.134600416147835e-06, "loss": 1.2281, "step": 28341 }, { "epoch": 70.855, "grad_norm": 6.125, "learning_rate": 4.133944651964098e-06, "loss": 1.3724, "step": 28342 }, { "epoch": 70.8575, "grad_norm": 9.8125, "learning_rate": 4.13328892623799e-06, "loss": 1.4188, "step": 28343 }, { "epoch": 70.86, "grad_norm": 6.84375, "learning_rate": 4.132633238973803e-06, "loss": 1.4396, "step": 28344 }, { "epoch": 70.8625, "grad_norm": 13.5625, "learning_rate": 4.131977590175836e-06, "loss": 1.3981, "step": 28345 }, { "epoch": 70.865, "grad_norm": 7.3125, "learning_rate": 4.131321979848388e-06, "loss": 1.5113, "step": 28346 }, { "epoch": 70.8675, "grad_norm": 7.34375, "learning_rate": 4.1306664079957625e-06, "loss": 1.3506, "step": 28347 }, { "epoch": 70.87, "grad_norm": 8.9375, "learning_rate": 4.130010874622253e-06, "loss": 1.3437, "step": 28348 }, { "epoch": 70.8725, "grad_norm": 9.8125, "learning_rate": 4.129355379732154e-06, "loss": 1.4223, "step": 28349 }, { "epoch": 70.875, "grad_norm": 8.6875, "learning_rate": 4.12869992332977e-06, "loss": 1.3018, "step": 28350 }, { "epoch": 70.8775, "grad_norm": 5.59375, "learning_rate": 4.12804450541939e-06, "loss": 1.3695, "step": 28351 }, { "epoch": 70.88, "grad_norm": 8.125, "learning_rate": 4.127389126005319e-06, "loss": 1.2135, "step": 28352 }, { "epoch": 70.8825, "grad_norm": 8.0, "learning_rate": 4.126733785091846e-06, "loss": 1.2692, "step": 28353 }, { "epoch": 70.885, "grad_norm": 6.84375, "learning_rate": 4.126078482683276e-06, "loss": 1.3456, "step": 28354 }, { "epoch": 70.8875, "grad_norm": 7.03125, "learning_rate": 4.125423218783899e-06, "loss": 1.4237, "step": 28355 }, { "epoch": 70.89, "grad_norm": 8.5, "learning_rate": 4.124767993398009e-06, "loss": 1.3525, "step": 28356 }, { "epoch": 70.8925, "grad_norm": 5.75, "learning_rate": 4.124112806529906e-06, "loss": 1.2405, "step": 28357 }, { "epoch": 70.895, "grad_norm": 6.15625, "learning_rate": 4.123457658183887e-06, "loss": 1.3004, "step": 28358 }, { "epoch": 70.8975, "grad_norm": 8.75, "learning_rate": 4.1228025483642446e-06, "loss": 1.3712, "step": 28359 }, { "epoch": 70.9, "grad_norm": 7.4375, "learning_rate": 4.12214747707527e-06, "loss": 1.3802, "step": 28360 }, { "epoch": 70.9025, "grad_norm": 6.0625, "learning_rate": 4.121492444321261e-06, "loss": 1.3878, "step": 28361 }, { "epoch": 70.905, "grad_norm": 7.8125, "learning_rate": 4.120837450106517e-06, "loss": 1.516, "step": 28362 }, { "epoch": 70.9075, "grad_norm": 8.6875, "learning_rate": 4.120182494435325e-06, "loss": 1.4693, "step": 28363 }, { "epoch": 70.91, "grad_norm": 6.0, "learning_rate": 4.1195275773119805e-06, "loss": 1.2573, "step": 28364 }, { "epoch": 70.9125, "grad_norm": 7.40625, "learning_rate": 4.118872698740779e-06, "loss": 1.4115, "step": 28365 }, { "epoch": 70.915, "grad_norm": 9.1875, "learning_rate": 4.1182178587260106e-06, "loss": 1.3458, "step": 28366 }, { "epoch": 70.9175, "grad_norm": 10.25, "learning_rate": 4.117563057271974e-06, "loss": 1.3145, "step": 28367 }, { "epoch": 70.92, "grad_norm": 7.375, "learning_rate": 4.116908294382956e-06, "loss": 1.3322, "step": 28368 }, { "epoch": 70.9225, "grad_norm": 10.9375, "learning_rate": 4.116253570063254e-06, "loss": 1.3641, "step": 28369 }, { "epoch": 70.925, "grad_norm": 8.6875, "learning_rate": 4.115598884317157e-06, "loss": 1.3365, "step": 28370 }, { "epoch": 70.9275, "grad_norm": 6.3125, "learning_rate": 4.114944237148957e-06, "loss": 1.2113, "step": 28371 }, { "epoch": 70.93, "grad_norm": 8.0, "learning_rate": 4.1142896285629455e-06, "loss": 1.3844, "step": 28372 }, { "epoch": 70.9325, "grad_norm": 5.59375, "learning_rate": 4.11363505856342e-06, "loss": 1.4174, "step": 28373 }, { "epoch": 70.935, "grad_norm": 6.5, "learning_rate": 4.112980527154668e-06, "loss": 1.3646, "step": 28374 }, { "epoch": 70.9375, "grad_norm": 7.1875, "learning_rate": 4.1123260343409754e-06, "loss": 1.181, "step": 28375 }, { "epoch": 70.94, "grad_norm": 6.84375, "learning_rate": 4.111671580126637e-06, "loss": 1.2631, "step": 28376 }, { "epoch": 70.9425, "grad_norm": 10.3125, "learning_rate": 4.111017164515949e-06, "loss": 1.2078, "step": 28377 }, { "epoch": 70.945, "grad_norm": 9.5, "learning_rate": 4.110362787513197e-06, "loss": 1.5101, "step": 28378 }, { "epoch": 70.9475, "grad_norm": 6.5, "learning_rate": 4.109708449122667e-06, "loss": 1.3534, "step": 28379 }, { "epoch": 70.95, "grad_norm": 8.4375, "learning_rate": 4.1090541493486555e-06, "loss": 1.3871, "step": 28380 }, { "epoch": 70.9525, "grad_norm": 6.40625, "learning_rate": 4.108399888195447e-06, "loss": 1.2956, "step": 28381 }, { "epoch": 70.955, "grad_norm": 7.21875, "learning_rate": 4.107745665667335e-06, "loss": 1.2647, "step": 28382 }, { "epoch": 70.9575, "grad_norm": 6.34375, "learning_rate": 4.107091481768604e-06, "loss": 1.3267, "step": 28383 }, { "epoch": 70.96, "grad_norm": 9.125, "learning_rate": 4.10643733650355e-06, "loss": 1.2372, "step": 28384 }, { "epoch": 70.9625, "grad_norm": 9.125, "learning_rate": 4.105783229876455e-06, "loss": 1.0848, "step": 28385 }, { "epoch": 70.965, "grad_norm": 6.75, "learning_rate": 4.105129161891608e-06, "loss": 1.2918, "step": 28386 }, { "epoch": 70.9675, "grad_norm": 5.8125, "learning_rate": 4.104475132553296e-06, "loss": 1.4414, "step": 28387 }, { "epoch": 70.97, "grad_norm": 6.28125, "learning_rate": 4.103821141865815e-06, "loss": 1.2939, "step": 28388 }, { "epoch": 70.9725, "grad_norm": 6.5, "learning_rate": 4.103167189833444e-06, "loss": 1.3725, "step": 28389 }, { "epoch": 70.975, "grad_norm": 10.6875, "learning_rate": 4.102513276460471e-06, "loss": 1.3655, "step": 28390 }, { "epoch": 70.9775, "grad_norm": 8.3125, "learning_rate": 4.1018594017511845e-06, "loss": 1.4373, "step": 28391 }, { "epoch": 70.98, "grad_norm": 7.65625, "learning_rate": 4.101205565709876e-06, "loss": 1.2322, "step": 28392 }, { "epoch": 70.9825, "grad_norm": 12.4375, "learning_rate": 4.100551768340827e-06, "loss": 1.3153, "step": 28393 }, { "epoch": 70.985, "grad_norm": 8.5, "learning_rate": 4.099898009648321e-06, "loss": 1.3224, "step": 28394 }, { "epoch": 70.9875, "grad_norm": 9.8125, "learning_rate": 4.099244289636652e-06, "loss": 1.4023, "step": 28395 }, { "epoch": 70.99, "grad_norm": 10.5, "learning_rate": 4.098590608310097e-06, "loss": 1.3949, "step": 28396 }, { "epoch": 70.9925, "grad_norm": 7.0625, "learning_rate": 4.097936965672949e-06, "loss": 1.2806, "step": 28397 }, { "epoch": 70.995, "grad_norm": 7.34375, "learning_rate": 4.0972833617294875e-06, "loss": 1.3322, "step": 28398 }, { "epoch": 70.9975, "grad_norm": 6.5625, "learning_rate": 4.096629796484002e-06, "loss": 1.4921, "step": 28399 }, { "epoch": 71.0, "grad_norm": 7.1875, "learning_rate": 4.095976269940777e-06, "loss": 1.3478, "step": 28400 }, { "epoch": 71.0025, "grad_norm": 5.4375, "learning_rate": 4.0953227821040904e-06, "loss": 1.3746, "step": 28401 }, { "epoch": 71.005, "grad_norm": 7.40625, "learning_rate": 4.0946693329782325e-06, "loss": 1.2594, "step": 28402 }, { "epoch": 71.0075, "grad_norm": 6.6875, "learning_rate": 4.094015922567489e-06, "loss": 1.5132, "step": 28403 }, { "epoch": 71.01, "grad_norm": 9.625, "learning_rate": 4.093362550876141e-06, "loss": 1.3815, "step": 28404 }, { "epoch": 71.0125, "grad_norm": 5.09375, "learning_rate": 4.092709217908468e-06, "loss": 1.1706, "step": 28405 }, { "epoch": 71.015, "grad_norm": 7.15625, "learning_rate": 4.0920559236687576e-06, "loss": 1.3352, "step": 28406 }, { "epoch": 71.0175, "grad_norm": 8.5625, "learning_rate": 4.091402668161295e-06, "loss": 1.3134, "step": 28407 }, { "epoch": 71.02, "grad_norm": 6.5625, "learning_rate": 4.09074945139036e-06, "loss": 1.2639, "step": 28408 }, { "epoch": 71.0225, "grad_norm": 10.375, "learning_rate": 4.0900962733602325e-06, "loss": 1.2531, "step": 28409 }, { "epoch": 71.025, "grad_norm": 6.21875, "learning_rate": 4.0894431340752e-06, "loss": 1.4006, "step": 28410 }, { "epoch": 71.0275, "grad_norm": 6.5, "learning_rate": 4.088790033539538e-06, "loss": 1.3595, "step": 28411 }, { "epoch": 71.03, "grad_norm": 6.28125, "learning_rate": 4.088136971757537e-06, "loss": 1.3586, "step": 28412 }, { "epoch": 71.0325, "grad_norm": 7.5625, "learning_rate": 4.087483948733469e-06, "loss": 1.207, "step": 28413 }, { "epoch": 71.035, "grad_norm": 7.0, "learning_rate": 4.086830964471622e-06, "loss": 1.2574, "step": 28414 }, { "epoch": 71.0375, "grad_norm": 7.625, "learning_rate": 4.086178018976275e-06, "loss": 1.2641, "step": 28415 }, { "epoch": 71.04, "grad_norm": 7.40625, "learning_rate": 4.085525112251706e-06, "loss": 1.436, "step": 28416 }, { "epoch": 71.0425, "grad_norm": 9.4375, "learning_rate": 4.084872244302197e-06, "loss": 1.2931, "step": 28417 }, { "epoch": 71.045, "grad_norm": 6.84375, "learning_rate": 4.0842194151320326e-06, "loss": 1.2258, "step": 28418 }, { "epoch": 71.0475, "grad_norm": 7.46875, "learning_rate": 4.083566624745488e-06, "loss": 1.6106, "step": 28419 }, { "epoch": 71.05, "grad_norm": 7.6875, "learning_rate": 4.082913873146842e-06, "loss": 1.3306, "step": 28420 }, { "epoch": 71.0525, "grad_norm": 8.625, "learning_rate": 4.082261160340375e-06, "loss": 1.3928, "step": 28421 }, { "epoch": 71.055, "grad_norm": 8.6875, "learning_rate": 4.0816084863303705e-06, "loss": 1.4379, "step": 28422 }, { "epoch": 71.0575, "grad_norm": 6.5, "learning_rate": 4.080955851121104e-06, "loss": 1.154, "step": 28423 }, { "epoch": 71.06, "grad_norm": 8.375, "learning_rate": 4.080303254716851e-06, "loss": 1.333, "step": 28424 }, { "epoch": 71.0625, "grad_norm": 13.5, "learning_rate": 4.079650697121895e-06, "loss": 1.4641, "step": 28425 }, { "epoch": 71.065, "grad_norm": 6.40625, "learning_rate": 4.07899817834051e-06, "loss": 1.3585, "step": 28426 }, { "epoch": 71.0675, "grad_norm": 10.375, "learning_rate": 4.07834569837698e-06, "loss": 1.4352, "step": 28427 }, { "epoch": 71.07, "grad_norm": 12.4375, "learning_rate": 4.077693257235573e-06, "loss": 1.3148, "step": 28428 }, { "epoch": 71.0725, "grad_norm": 7.9375, "learning_rate": 4.077040854920578e-06, "loss": 1.3752, "step": 28429 }, { "epoch": 71.075, "grad_norm": 7.46875, "learning_rate": 4.076388491436264e-06, "loss": 1.3389, "step": 28430 }, { "epoch": 71.0775, "grad_norm": 5.46875, "learning_rate": 4.075736166786908e-06, "loss": 1.1137, "step": 28431 }, { "epoch": 71.08, "grad_norm": 9.9375, "learning_rate": 4.0750838809767875e-06, "loss": 1.3696, "step": 28432 }, { "epoch": 71.0825, "grad_norm": 7.9375, "learning_rate": 4.074431634010184e-06, "loss": 1.3296, "step": 28433 }, { "epoch": 71.085, "grad_norm": 5.84375, "learning_rate": 4.07377942589137e-06, "loss": 1.3139, "step": 28434 }, { "epoch": 71.0875, "grad_norm": 7.4375, "learning_rate": 4.073127256624616e-06, "loss": 1.3336, "step": 28435 }, { "epoch": 71.09, "grad_norm": 5.625, "learning_rate": 4.072475126214204e-06, "loss": 1.1905, "step": 28436 }, { "epoch": 71.0925, "grad_norm": 7.90625, "learning_rate": 4.07182303466441e-06, "loss": 1.2273, "step": 28437 }, { "epoch": 71.095, "grad_norm": 6.6875, "learning_rate": 4.071170981979509e-06, "loss": 1.3688, "step": 28438 }, { "epoch": 71.0975, "grad_norm": 7.65625, "learning_rate": 4.070518968163768e-06, "loss": 1.3592, "step": 28439 }, { "epoch": 71.1, "grad_norm": 7.375, "learning_rate": 4.069866993221473e-06, "loss": 1.3183, "step": 28440 }, { "epoch": 71.1025, "grad_norm": 7.875, "learning_rate": 4.069215057156888e-06, "loss": 1.3448, "step": 28441 }, { "epoch": 71.105, "grad_norm": 6.59375, "learning_rate": 4.068563159974295e-06, "loss": 1.2713, "step": 28442 }, { "epoch": 71.1075, "grad_norm": 8.8125, "learning_rate": 4.067911301677962e-06, "loss": 1.2353, "step": 28443 }, { "epoch": 71.11, "grad_norm": 8.1875, "learning_rate": 4.067259482272168e-06, "loss": 1.3777, "step": 28444 }, { "epoch": 71.1125, "grad_norm": 7.84375, "learning_rate": 4.066607701761183e-06, "loss": 1.379, "step": 28445 }, { "epoch": 71.115, "grad_norm": 7.09375, "learning_rate": 4.065955960149277e-06, "loss": 1.3394, "step": 28446 }, { "epoch": 71.1175, "grad_norm": 11.0, "learning_rate": 4.065304257440728e-06, "loss": 1.3379, "step": 28447 }, { "epoch": 71.12, "grad_norm": 6.375, "learning_rate": 4.0646525936398086e-06, "loss": 1.4032, "step": 28448 }, { "epoch": 71.1225, "grad_norm": 9.5, "learning_rate": 4.0640009687507885e-06, "loss": 1.2657, "step": 28449 }, { "epoch": 71.125, "grad_norm": 7.09375, "learning_rate": 4.0633493827779425e-06, "loss": 1.2262, "step": 28450 }, { "epoch": 71.1275, "grad_norm": 11.6875, "learning_rate": 4.062697835725532e-06, "loss": 1.3348, "step": 28451 }, { "epoch": 71.13, "grad_norm": 7.6875, "learning_rate": 4.062046327597845e-06, "loss": 1.3255, "step": 28452 }, { "epoch": 71.1325, "grad_norm": 5.65625, "learning_rate": 4.0613948583991435e-06, "loss": 1.3223, "step": 28453 }, { "epoch": 71.135, "grad_norm": 7.0625, "learning_rate": 4.060743428133701e-06, "loss": 1.3051, "step": 28454 }, { "epoch": 71.1375, "grad_norm": 6.21875, "learning_rate": 4.060092036805783e-06, "loss": 1.2448, "step": 28455 }, { "epoch": 71.14, "grad_norm": 11.875, "learning_rate": 4.0594406844196636e-06, "loss": 1.3425, "step": 28456 }, { "epoch": 71.1425, "grad_norm": 5.96875, "learning_rate": 4.058789370979616e-06, "loss": 1.3663, "step": 28457 }, { "epoch": 71.145, "grad_norm": 7.28125, "learning_rate": 4.0581380964899055e-06, "loss": 1.3625, "step": 28458 }, { "epoch": 71.1475, "grad_norm": 10.0, "learning_rate": 4.057486860954806e-06, "loss": 1.123, "step": 28459 }, { "epoch": 71.15, "grad_norm": 7.90625, "learning_rate": 4.056835664378585e-06, "loss": 1.4587, "step": 28460 }, { "epoch": 71.1525, "grad_norm": 9.6875, "learning_rate": 4.056184506765509e-06, "loss": 1.3334, "step": 28461 }, { "epoch": 71.155, "grad_norm": 5.375, "learning_rate": 4.055533388119849e-06, "loss": 1.4223, "step": 28462 }, { "epoch": 71.1575, "grad_norm": 8.875, "learning_rate": 4.054882308445878e-06, "loss": 1.3118, "step": 28463 }, { "epoch": 71.16, "grad_norm": 9.625, "learning_rate": 4.054231267747862e-06, "loss": 1.3255, "step": 28464 }, { "epoch": 71.1625, "grad_norm": 8.25, "learning_rate": 4.053580266030065e-06, "loss": 1.4044, "step": 28465 }, { "epoch": 71.165, "grad_norm": 10.5625, "learning_rate": 4.052929303296752e-06, "loss": 1.5738, "step": 28466 }, { "epoch": 71.1675, "grad_norm": 7.0625, "learning_rate": 4.052278379552203e-06, "loss": 1.3858, "step": 28467 }, { "epoch": 71.17, "grad_norm": 7.625, "learning_rate": 4.05162749480068e-06, "loss": 1.172, "step": 28468 }, { "epoch": 71.1725, "grad_norm": 7.8125, "learning_rate": 4.05097664904645e-06, "loss": 1.2591, "step": 28469 }, { "epoch": 71.175, "grad_norm": 9.1875, "learning_rate": 4.050325842293773e-06, "loss": 1.2724, "step": 28470 }, { "epoch": 71.1775, "grad_norm": 7.0, "learning_rate": 4.049675074546924e-06, "loss": 1.5337, "step": 28471 }, { "epoch": 71.18, "grad_norm": 10.6875, "learning_rate": 4.049024345810169e-06, "loss": 1.242, "step": 28472 }, { "epoch": 71.1825, "grad_norm": 6.84375, "learning_rate": 4.048373656087773e-06, "loss": 1.2814, "step": 28473 }, { "epoch": 71.185, "grad_norm": 7.5625, "learning_rate": 4.047723005383997e-06, "loss": 1.3224, "step": 28474 }, { "epoch": 71.1875, "grad_norm": 7.125, "learning_rate": 4.047072393703115e-06, "loss": 1.2646, "step": 28475 }, { "epoch": 71.19, "grad_norm": 6.46875, "learning_rate": 4.046421821049385e-06, "loss": 1.2883, "step": 28476 }, { "epoch": 71.1925, "grad_norm": 8.125, "learning_rate": 4.045771287427077e-06, "loss": 1.4211, "step": 28477 }, { "epoch": 71.195, "grad_norm": 8.5625, "learning_rate": 4.045120792840452e-06, "loss": 1.3519, "step": 28478 }, { "epoch": 71.1975, "grad_norm": 6.46875, "learning_rate": 4.04447033729378e-06, "loss": 1.4204, "step": 28479 }, { "epoch": 71.2, "grad_norm": 6.125, "learning_rate": 4.043819920791322e-06, "loss": 1.3674, "step": 28480 }, { "epoch": 71.2025, "grad_norm": 7.625, "learning_rate": 4.04316954333734e-06, "loss": 1.2906, "step": 28481 }, { "epoch": 71.205, "grad_norm": 8.5625, "learning_rate": 4.042519204936098e-06, "loss": 1.2997, "step": 28482 }, { "epoch": 71.2075, "grad_norm": 6.53125, "learning_rate": 4.041868905591867e-06, "loss": 1.1538, "step": 28483 }, { "epoch": 71.21, "grad_norm": 7.40625, "learning_rate": 4.041218645308905e-06, "loss": 1.2583, "step": 28484 }, { "epoch": 71.2125, "grad_norm": 9.8125, "learning_rate": 4.0405684240914666e-06, "loss": 1.5071, "step": 28485 }, { "epoch": 71.215, "grad_norm": 7.78125, "learning_rate": 4.039918241943831e-06, "loss": 1.2838, "step": 28486 }, { "epoch": 71.2175, "grad_norm": 10.375, "learning_rate": 4.039268098870252e-06, "loss": 1.2649, "step": 28487 }, { "epoch": 71.22, "grad_norm": 7.40625, "learning_rate": 4.038617994874994e-06, "loss": 1.5061, "step": 28488 }, { "epoch": 71.2225, "grad_norm": 6.875, "learning_rate": 4.037967929962314e-06, "loss": 1.4021, "step": 28489 }, { "epoch": 71.225, "grad_norm": 7.875, "learning_rate": 4.037317904136477e-06, "loss": 1.2567, "step": 28490 }, { "epoch": 71.2275, "grad_norm": 8.6875, "learning_rate": 4.03666791740175e-06, "loss": 1.415, "step": 28491 }, { "epoch": 71.23, "grad_norm": 8.5, "learning_rate": 4.036017969762389e-06, "loss": 1.2959, "step": 28492 }, { "epoch": 71.2325, "grad_norm": 6.90625, "learning_rate": 4.035368061222652e-06, "loss": 1.3209, "step": 28493 }, { "epoch": 71.235, "grad_norm": 8.9375, "learning_rate": 4.0347181917868065e-06, "loss": 1.1477, "step": 28494 }, { "epoch": 71.2375, "grad_norm": 6.71875, "learning_rate": 4.034068361459107e-06, "loss": 1.3699, "step": 28495 }, { "epoch": 71.24, "grad_norm": 10.0625, "learning_rate": 4.033418570243819e-06, "loss": 1.2851, "step": 28496 }, { "epoch": 71.2425, "grad_norm": 7.625, "learning_rate": 4.0327688181451975e-06, "loss": 1.2609, "step": 28497 }, { "epoch": 71.245, "grad_norm": 6.125, "learning_rate": 4.032119105167508e-06, "loss": 1.3596, "step": 28498 }, { "epoch": 71.2475, "grad_norm": 6.5625, "learning_rate": 4.031469431315007e-06, "loss": 1.4252, "step": 28499 }, { "epoch": 71.25, "grad_norm": 6.78125, "learning_rate": 4.03081979659195e-06, "loss": 1.3223, "step": 28500 }, { "epoch": 71.2525, "grad_norm": 6.25, "learning_rate": 4.030170201002599e-06, "loss": 1.3715, "step": 28501 }, { "epoch": 71.255, "grad_norm": 7.375, "learning_rate": 4.029520644551217e-06, "loss": 1.2223, "step": 28502 }, { "epoch": 71.2575, "grad_norm": 9.3125, "learning_rate": 4.028871127242059e-06, "loss": 1.3288, "step": 28503 }, { "epoch": 71.26, "grad_norm": 7.84375, "learning_rate": 4.028221649079379e-06, "loss": 1.2536, "step": 28504 }, { "epoch": 71.2625, "grad_norm": 7.71875, "learning_rate": 4.02757221006744e-06, "loss": 1.3058, "step": 28505 }, { "epoch": 71.265, "grad_norm": 5.875, "learning_rate": 4.0269228102105e-06, "loss": 1.2781, "step": 28506 }, { "epoch": 71.2675, "grad_norm": 7.9375, "learning_rate": 4.026273449512818e-06, "loss": 1.4101, "step": 28507 }, { "epoch": 71.27, "grad_norm": 10.625, "learning_rate": 4.025624127978642e-06, "loss": 1.3512, "step": 28508 }, { "epoch": 71.2725, "grad_norm": 5.75, "learning_rate": 4.024974845612239e-06, "loss": 1.3286, "step": 28509 }, { "epoch": 71.275, "grad_norm": 6.90625, "learning_rate": 4.02432560241786e-06, "loss": 1.177, "step": 28510 }, { "epoch": 71.2775, "grad_norm": 6.5625, "learning_rate": 4.023676398399766e-06, "loss": 1.4253, "step": 28511 }, { "epoch": 71.28, "grad_norm": 8.0625, "learning_rate": 4.0230272335622065e-06, "loss": 1.352, "step": 28512 }, { "epoch": 71.2825, "grad_norm": 8.1875, "learning_rate": 4.022378107909445e-06, "loss": 1.2491, "step": 28513 }, { "epoch": 71.285, "grad_norm": 6.28125, "learning_rate": 4.0217290214457326e-06, "loss": 1.3533, "step": 28514 }, { "epoch": 71.2875, "grad_norm": 9.0625, "learning_rate": 4.021079974175322e-06, "loss": 1.3174, "step": 28515 }, { "epoch": 71.29, "grad_norm": 5.4375, "learning_rate": 4.020430966102474e-06, "loss": 1.1689, "step": 28516 }, { "epoch": 71.2925, "grad_norm": 6.6875, "learning_rate": 4.019781997231443e-06, "loss": 1.4346, "step": 28517 }, { "epoch": 71.295, "grad_norm": 9.4375, "learning_rate": 4.019133067566482e-06, "loss": 1.3855, "step": 28518 }, { "epoch": 71.2975, "grad_norm": 7.625, "learning_rate": 4.0184841771118425e-06, "loss": 1.3549, "step": 28519 }, { "epoch": 71.3, "grad_norm": 7.03125, "learning_rate": 4.017835325871781e-06, "loss": 1.2429, "step": 28520 }, { "epoch": 71.3025, "grad_norm": 6.46875, "learning_rate": 4.017186513850554e-06, "loss": 1.1489, "step": 28521 }, { "epoch": 71.305, "grad_norm": 6.5, "learning_rate": 4.016537741052414e-06, "loss": 1.2785, "step": 28522 }, { "epoch": 71.3075, "grad_norm": 14.75, "learning_rate": 4.015889007481609e-06, "loss": 1.3678, "step": 28523 }, { "epoch": 71.31, "grad_norm": 6.65625, "learning_rate": 4.015240313142401e-06, "loss": 1.3338, "step": 28524 }, { "epoch": 71.3125, "grad_norm": 7.28125, "learning_rate": 4.0145916580390335e-06, "loss": 1.5029, "step": 28525 }, { "epoch": 71.315, "grad_norm": 7.46875, "learning_rate": 4.013943042175768e-06, "loss": 1.3561, "step": 28526 }, { "epoch": 71.3175, "grad_norm": 6.09375, "learning_rate": 4.013294465556847e-06, "loss": 1.4681, "step": 28527 }, { "epoch": 71.32, "grad_norm": 8.0625, "learning_rate": 4.012645928186533e-06, "loss": 1.2461, "step": 28528 }, { "epoch": 71.3225, "grad_norm": 7.15625, "learning_rate": 4.011997430069072e-06, "loss": 1.314, "step": 28529 }, { "epoch": 71.325, "grad_norm": 8.0625, "learning_rate": 4.011348971208713e-06, "loss": 1.2183, "step": 28530 }, { "epoch": 71.3275, "grad_norm": 7.75, "learning_rate": 4.010700551609711e-06, "loss": 1.2263, "step": 28531 }, { "epoch": 71.33, "grad_norm": 5.90625, "learning_rate": 4.010052171276319e-06, "loss": 1.1723, "step": 28532 }, { "epoch": 71.3325, "grad_norm": 8.6875, "learning_rate": 4.0094038302127855e-06, "loss": 1.3494, "step": 28533 }, { "epoch": 71.335, "grad_norm": 9.375, "learning_rate": 4.008755528423358e-06, "loss": 1.3179, "step": 28534 }, { "epoch": 71.3375, "grad_norm": 9.5625, "learning_rate": 4.0081072659122886e-06, "loss": 1.2137, "step": 28535 }, { "epoch": 71.34, "grad_norm": 7.6875, "learning_rate": 4.007459042683832e-06, "loss": 1.334, "step": 28536 }, { "epoch": 71.3425, "grad_norm": 6.65625, "learning_rate": 4.006810858742235e-06, "loss": 1.411, "step": 28537 }, { "epoch": 71.345, "grad_norm": 7.34375, "learning_rate": 4.006162714091741e-06, "loss": 1.3684, "step": 28538 }, { "epoch": 71.3475, "grad_norm": 9.3125, "learning_rate": 4.005514608736609e-06, "loss": 1.2013, "step": 28539 }, { "epoch": 71.35, "grad_norm": 7.09375, "learning_rate": 4.004866542681079e-06, "loss": 1.2474, "step": 28540 }, { "epoch": 71.3525, "grad_norm": 6.8125, "learning_rate": 4.004218515929409e-06, "loss": 1.2902, "step": 28541 }, { "epoch": 71.355, "grad_norm": 8.8125, "learning_rate": 4.003570528485838e-06, "loss": 1.4628, "step": 28542 }, { "epoch": 71.3575, "grad_norm": 7.5625, "learning_rate": 4.002922580354623e-06, "loss": 1.3277, "step": 28543 }, { "epoch": 71.36, "grad_norm": 8.375, "learning_rate": 4.002274671540006e-06, "loss": 1.2254, "step": 28544 }, { "epoch": 71.3625, "grad_norm": 5.75, "learning_rate": 4.001626802046233e-06, "loss": 1.126, "step": 28545 }, { "epoch": 71.365, "grad_norm": 10.5, "learning_rate": 4.000978971877555e-06, "loss": 1.3007, "step": 28546 }, { "epoch": 71.3675, "grad_norm": 5.5, "learning_rate": 4.000331181038223e-06, "loss": 1.1948, "step": 28547 }, { "epoch": 71.37, "grad_norm": 7.21875, "learning_rate": 3.9996834295324796e-06, "loss": 1.2694, "step": 28548 }, { "epoch": 71.3725, "grad_norm": 6.9375, "learning_rate": 3.999035717364566e-06, "loss": 1.3698, "step": 28549 }, { "epoch": 71.375, "grad_norm": 12.125, "learning_rate": 3.998388044538737e-06, "loss": 1.235, "step": 28550 }, { "epoch": 71.3775, "grad_norm": 12.9375, "learning_rate": 3.997740411059238e-06, "loss": 1.3944, "step": 28551 }, { "epoch": 71.38, "grad_norm": 5.90625, "learning_rate": 3.997092816930313e-06, "loss": 1.4425, "step": 28552 }, { "epoch": 71.3825, "grad_norm": 7.6875, "learning_rate": 3.996445262156202e-06, "loss": 1.2752, "step": 28553 }, { "epoch": 71.385, "grad_norm": 10.25, "learning_rate": 3.9957977467411615e-06, "loss": 1.3149, "step": 28554 }, { "epoch": 71.3875, "grad_norm": 7.9375, "learning_rate": 3.995150270689426e-06, "loss": 1.4679, "step": 28555 }, { "epoch": 71.39, "grad_norm": 6.59375, "learning_rate": 3.994502834005249e-06, "loss": 1.4267, "step": 28556 }, { "epoch": 71.3925, "grad_norm": 7.09375, "learning_rate": 3.9938554366928676e-06, "loss": 1.1856, "step": 28557 }, { "epoch": 71.395, "grad_norm": 4.5625, "learning_rate": 3.993208078756532e-06, "loss": 1.3811, "step": 28558 }, { "epoch": 71.3975, "grad_norm": 5.96875, "learning_rate": 3.992560760200485e-06, "loss": 1.1604, "step": 28559 }, { "epoch": 71.4, "grad_norm": 7.5, "learning_rate": 3.991913481028965e-06, "loss": 1.4207, "step": 28560 }, { "epoch": 71.4025, "grad_norm": 8.375, "learning_rate": 3.991266241246221e-06, "loss": 1.3703, "step": 28561 }, { "epoch": 71.405, "grad_norm": 6.5, "learning_rate": 3.990619040856497e-06, "loss": 1.2466, "step": 28562 }, { "epoch": 71.4075, "grad_norm": 7.125, "learning_rate": 3.9899718798640355e-06, "loss": 1.2068, "step": 28563 }, { "epoch": 71.41, "grad_norm": 5.3125, "learning_rate": 3.9893247582730736e-06, "loss": 1.1231, "step": 28564 }, { "epoch": 71.4125, "grad_norm": 10.8125, "learning_rate": 3.988677676087858e-06, "loss": 1.2649, "step": 28565 }, { "epoch": 71.415, "grad_norm": 7.5, "learning_rate": 3.988030633312635e-06, "loss": 1.3569, "step": 28566 }, { "epoch": 71.4175, "grad_norm": 8.0625, "learning_rate": 3.987383629951642e-06, "loss": 1.3388, "step": 28567 }, { "epoch": 71.42, "grad_norm": 12.0625, "learning_rate": 3.9867366660091176e-06, "loss": 1.4652, "step": 28568 }, { "epoch": 71.4225, "grad_norm": 6.03125, "learning_rate": 3.986089741489312e-06, "loss": 1.1811, "step": 28569 }, { "epoch": 71.425, "grad_norm": 6.40625, "learning_rate": 3.985442856396457e-06, "loss": 1.2311, "step": 28570 }, { "epoch": 71.4275, "grad_norm": 6.375, "learning_rate": 3.9847960107348e-06, "loss": 1.2597, "step": 28571 }, { "epoch": 71.43, "grad_norm": 6.59375, "learning_rate": 3.984149204508578e-06, "loss": 1.3322, "step": 28572 }, { "epoch": 71.4325, "grad_norm": 8.75, "learning_rate": 3.983502437722037e-06, "loss": 1.2784, "step": 28573 }, { "epoch": 71.435, "grad_norm": 8.5, "learning_rate": 3.9828557103794115e-06, "loss": 1.3343, "step": 28574 }, { "epoch": 71.4375, "grad_norm": 8.6875, "learning_rate": 3.98220902248494e-06, "loss": 1.5666, "step": 28575 }, { "epoch": 71.44, "grad_norm": 7.0625, "learning_rate": 3.981562374042867e-06, "loss": 1.3502, "step": 28576 }, { "epoch": 71.4425, "grad_norm": 6.1875, "learning_rate": 3.980915765057431e-06, "loss": 1.2876, "step": 28577 }, { "epoch": 71.445, "grad_norm": 5.0625, "learning_rate": 3.980269195532872e-06, "loss": 1.1806, "step": 28578 }, { "epoch": 71.4475, "grad_norm": 8.9375, "learning_rate": 3.979622665473424e-06, "loss": 1.4076, "step": 28579 }, { "epoch": 71.45, "grad_norm": 8.125, "learning_rate": 3.978976174883329e-06, "loss": 1.1943, "step": 28580 }, { "epoch": 71.4525, "grad_norm": 5.28125, "learning_rate": 3.978329723766828e-06, "loss": 1.4816, "step": 28581 }, { "epoch": 71.455, "grad_norm": 8.0625, "learning_rate": 3.977683312128157e-06, "loss": 1.339, "step": 28582 }, { "epoch": 71.4575, "grad_norm": 7.5, "learning_rate": 3.977036939971549e-06, "loss": 1.4166, "step": 28583 }, { "epoch": 71.46, "grad_norm": 5.46875, "learning_rate": 3.976390607301249e-06, "loss": 1.076, "step": 28584 }, { "epoch": 71.4625, "grad_norm": 6.71875, "learning_rate": 3.9757443141214894e-06, "loss": 1.3398, "step": 28585 }, { "epoch": 71.465, "grad_norm": 10.6875, "learning_rate": 3.97509806043651e-06, "loss": 1.4067, "step": 28586 }, { "epoch": 71.4675, "grad_norm": 8.6875, "learning_rate": 3.974451846250545e-06, "loss": 1.3184, "step": 28587 }, { "epoch": 71.47, "grad_norm": 8.0625, "learning_rate": 3.9738056715678355e-06, "loss": 1.5335, "step": 28588 }, { "epoch": 71.4725, "grad_norm": 6.5625, "learning_rate": 3.973159536392614e-06, "loss": 1.4876, "step": 28589 }, { "epoch": 71.475, "grad_norm": 7.3125, "learning_rate": 3.972513440729115e-06, "loss": 1.2174, "step": 28590 }, { "epoch": 71.4775, "grad_norm": 6.9375, "learning_rate": 3.971867384581577e-06, "loss": 1.3381, "step": 28591 }, { "epoch": 71.48, "grad_norm": 7.15625, "learning_rate": 3.971221367954239e-06, "loss": 1.417, "step": 28592 }, { "epoch": 71.4825, "grad_norm": 7.5625, "learning_rate": 3.970575390851331e-06, "loss": 1.2605, "step": 28593 }, { "epoch": 71.485, "grad_norm": 6.8125, "learning_rate": 3.969929453277091e-06, "loss": 1.4443, "step": 28594 }, { "epoch": 71.4875, "grad_norm": 7.53125, "learning_rate": 3.969283555235744e-06, "loss": 1.1606, "step": 28595 }, { "epoch": 71.49, "grad_norm": 7.0625, "learning_rate": 3.96863769673154e-06, "loss": 1.2774, "step": 28596 }, { "epoch": 71.4925, "grad_norm": 8.0, "learning_rate": 3.967991877768705e-06, "loss": 1.3457, "step": 28597 }, { "epoch": 71.495, "grad_norm": 8.4375, "learning_rate": 3.9673460983514715e-06, "loss": 1.4154, "step": 28598 }, { "epoch": 71.4975, "grad_norm": 6.125, "learning_rate": 3.966700358484079e-06, "loss": 1.2199, "step": 28599 }, { "epoch": 71.5, "grad_norm": 9.4375, "learning_rate": 3.966054658170754e-06, "loss": 1.3107, "step": 28600 }, { "epoch": 71.5025, "grad_norm": 5.75, "learning_rate": 3.965408997415736e-06, "loss": 1.2264, "step": 28601 }, { "epoch": 71.505, "grad_norm": 7.1875, "learning_rate": 3.9647633762232515e-06, "loss": 1.2554, "step": 28602 }, { "epoch": 71.5075, "grad_norm": 6.0, "learning_rate": 3.964117794597542e-06, "loss": 1.3272, "step": 28603 }, { "epoch": 71.51, "grad_norm": 6.25, "learning_rate": 3.963472252542833e-06, "loss": 1.3392, "step": 28604 }, { "epoch": 71.5125, "grad_norm": 6.40625, "learning_rate": 3.962826750063356e-06, "loss": 1.2059, "step": 28605 }, { "epoch": 71.515, "grad_norm": 7.40625, "learning_rate": 3.962181287163344e-06, "loss": 1.3753, "step": 28606 }, { "epoch": 71.5175, "grad_norm": 9.5625, "learning_rate": 3.961535863847033e-06, "loss": 1.2286, "step": 28607 }, { "epoch": 71.52, "grad_norm": 7.21875, "learning_rate": 3.960890480118653e-06, "loss": 1.301, "step": 28608 }, { "epoch": 71.5225, "grad_norm": 6.90625, "learning_rate": 3.9602451359824334e-06, "loss": 1.3805, "step": 28609 }, { "epoch": 71.525, "grad_norm": 6.71875, "learning_rate": 3.959599831442597e-06, "loss": 1.267, "step": 28610 }, { "epoch": 71.5275, "grad_norm": 11.625, "learning_rate": 3.958954566503389e-06, "loss": 1.1885, "step": 28611 }, { "epoch": 71.53, "grad_norm": 7.0625, "learning_rate": 3.9583093411690345e-06, "loss": 1.323, "step": 28612 }, { "epoch": 71.5325, "grad_norm": 11.625, "learning_rate": 3.957664155443761e-06, "loss": 1.4346, "step": 28613 }, { "epoch": 71.535, "grad_norm": 9.25, "learning_rate": 3.9570190093317966e-06, "loss": 1.3337, "step": 28614 }, { "epoch": 71.5375, "grad_norm": 5.34375, "learning_rate": 3.956373902837373e-06, "loss": 1.2535, "step": 28615 }, { "epoch": 71.54, "grad_norm": 6.375, "learning_rate": 3.955728835964724e-06, "loss": 1.2224, "step": 28616 }, { "epoch": 71.5425, "grad_norm": 7.40625, "learning_rate": 3.955083808718076e-06, "loss": 1.2598, "step": 28617 }, { "epoch": 71.545, "grad_norm": 5.4375, "learning_rate": 3.9544388211016515e-06, "loss": 1.2556, "step": 28618 }, { "epoch": 71.5475, "grad_norm": 7.71875, "learning_rate": 3.9537938731196886e-06, "loss": 1.284, "step": 28619 }, { "epoch": 71.55, "grad_norm": 7.78125, "learning_rate": 3.953148964776408e-06, "loss": 1.3534, "step": 28620 }, { "epoch": 71.5525, "grad_norm": 7.875, "learning_rate": 3.952504096076043e-06, "loss": 1.4386, "step": 28621 }, { "epoch": 71.555, "grad_norm": 7.8125, "learning_rate": 3.9518592670228165e-06, "loss": 1.342, "step": 28622 }, { "epoch": 71.5575, "grad_norm": 6.40625, "learning_rate": 3.951214477620962e-06, "loss": 1.4639, "step": 28623 }, { "epoch": 71.56, "grad_norm": 7.875, "learning_rate": 3.950569727874704e-06, "loss": 1.4078, "step": 28624 }, { "epoch": 71.5625, "grad_norm": 5.40625, "learning_rate": 3.949925017788261e-06, "loss": 1.3296, "step": 28625 }, { "epoch": 71.565, "grad_norm": 5.0, "learning_rate": 3.949280347365874e-06, "loss": 1.1867, "step": 28626 }, { "epoch": 71.5675, "grad_norm": 8.3125, "learning_rate": 3.948635716611764e-06, "loss": 1.2875, "step": 28627 }, { "epoch": 71.57, "grad_norm": 9.4375, "learning_rate": 3.947991125530155e-06, "loss": 1.2948, "step": 28628 }, { "epoch": 71.5725, "grad_norm": 11.875, "learning_rate": 3.947346574125271e-06, "loss": 1.3335, "step": 28629 }, { "epoch": 71.575, "grad_norm": 5.8125, "learning_rate": 3.946702062401341e-06, "loss": 1.1808, "step": 28630 }, { "epoch": 71.5775, "grad_norm": 5.4375, "learning_rate": 3.946057590362593e-06, "loss": 1.1771, "step": 28631 }, { "epoch": 71.58, "grad_norm": 9.9375, "learning_rate": 3.945413158013249e-06, "loss": 1.4946, "step": 28632 }, { "epoch": 71.5825, "grad_norm": 11.9375, "learning_rate": 3.944768765357532e-06, "loss": 1.347, "step": 28633 }, { "epoch": 71.585, "grad_norm": 5.5625, "learning_rate": 3.94412441239967e-06, "loss": 1.2407, "step": 28634 }, { "epoch": 71.5875, "grad_norm": 6.1875, "learning_rate": 3.943480099143884e-06, "loss": 1.1691, "step": 28635 }, { "epoch": 71.59, "grad_norm": 6.8125, "learning_rate": 3.942835825594403e-06, "loss": 1.308, "step": 28636 }, { "epoch": 71.5925, "grad_norm": 8.0625, "learning_rate": 3.942191591755446e-06, "loss": 1.2798, "step": 28637 }, { "epoch": 71.595, "grad_norm": 10.6875, "learning_rate": 3.9415473976312405e-06, "loss": 1.347, "step": 28638 }, { "epoch": 71.5975, "grad_norm": 6.90625, "learning_rate": 3.940903243226008e-06, "loss": 1.2737, "step": 28639 }, { "epoch": 71.6, "grad_norm": 8.0625, "learning_rate": 3.940259128543967e-06, "loss": 1.1766, "step": 28640 }, { "epoch": 71.6025, "grad_norm": 7.375, "learning_rate": 3.939615053589345e-06, "loss": 1.4598, "step": 28641 }, { "epoch": 71.605, "grad_norm": 11.4375, "learning_rate": 3.9389710183663684e-06, "loss": 1.3319, "step": 28642 }, { "epoch": 71.6075, "grad_norm": 10.0625, "learning_rate": 3.938327022879256e-06, "loss": 1.3246, "step": 28643 }, { "epoch": 71.61, "grad_norm": 6.0, "learning_rate": 3.9376830671322245e-06, "loss": 1.2274, "step": 28644 }, { "epoch": 71.6125, "grad_norm": 9.8125, "learning_rate": 3.9370391511295e-06, "loss": 1.2929, "step": 28645 }, { "epoch": 71.615, "grad_norm": 8.375, "learning_rate": 3.9363952748753075e-06, "loss": 1.3317, "step": 28646 }, { "epoch": 71.6175, "grad_norm": 8.4375, "learning_rate": 3.935751438373866e-06, "loss": 1.2817, "step": 28647 }, { "epoch": 71.62, "grad_norm": 6.0625, "learning_rate": 3.935107641629391e-06, "loss": 1.2181, "step": 28648 }, { "epoch": 71.6225, "grad_norm": 9.0625, "learning_rate": 3.934463884646112e-06, "loss": 1.248, "step": 28649 }, { "epoch": 71.625, "grad_norm": 8.8125, "learning_rate": 3.933820167428241e-06, "loss": 1.3488, "step": 28650 }, { "epoch": 71.6275, "grad_norm": 8.25, "learning_rate": 3.933176489980005e-06, "loss": 1.2431, "step": 28651 }, { "epoch": 71.63, "grad_norm": 6.96875, "learning_rate": 3.932532852305618e-06, "loss": 1.2966, "step": 28652 }, { "epoch": 71.6325, "grad_norm": 5.71875, "learning_rate": 3.931889254409306e-06, "loss": 1.286, "step": 28653 }, { "epoch": 71.635, "grad_norm": 8.5625, "learning_rate": 3.931245696295287e-06, "loss": 1.4149, "step": 28654 }, { "epoch": 71.6375, "grad_norm": 7.34375, "learning_rate": 3.930602177967774e-06, "loss": 1.465, "step": 28655 }, { "epoch": 71.64, "grad_norm": 6.125, "learning_rate": 3.9299586994309905e-06, "loss": 1.3789, "step": 28656 }, { "epoch": 71.6425, "grad_norm": 5.5, "learning_rate": 3.9293152606891575e-06, "loss": 1.2009, "step": 28657 }, { "epoch": 71.645, "grad_norm": 10.8125, "learning_rate": 3.928671861746491e-06, "loss": 1.3798, "step": 28658 }, { "epoch": 71.6475, "grad_norm": 8.4375, "learning_rate": 3.9280285026072055e-06, "loss": 1.2158, "step": 28659 }, { "epoch": 71.65, "grad_norm": 7.0625, "learning_rate": 3.927385183275522e-06, "loss": 1.3984, "step": 28660 }, { "epoch": 71.6525, "grad_norm": 9.375, "learning_rate": 3.926741903755661e-06, "loss": 1.4088, "step": 28661 }, { "epoch": 71.655, "grad_norm": 6.5625, "learning_rate": 3.926098664051838e-06, "loss": 1.1722, "step": 28662 }, { "epoch": 71.6575, "grad_norm": 8.625, "learning_rate": 3.925455464168265e-06, "loss": 1.4627, "step": 28663 }, { "epoch": 71.66, "grad_norm": 6.3125, "learning_rate": 3.924812304109167e-06, "loss": 1.3336, "step": 28664 }, { "epoch": 71.6625, "grad_norm": 6.5625, "learning_rate": 3.924169183878753e-06, "loss": 1.2161, "step": 28665 }, { "epoch": 71.665, "grad_norm": 6.875, "learning_rate": 3.923526103481245e-06, "loss": 1.3929, "step": 28666 }, { "epoch": 71.6675, "grad_norm": 6.34375, "learning_rate": 3.922883062920854e-06, "loss": 1.2615, "step": 28667 }, { "epoch": 71.67, "grad_norm": 7.09375, "learning_rate": 3.922240062201803e-06, "loss": 1.2634, "step": 28668 }, { "epoch": 71.6725, "grad_norm": 6.4375, "learning_rate": 3.921597101328302e-06, "loss": 1.2607, "step": 28669 }, { "epoch": 71.675, "grad_norm": 7.75, "learning_rate": 3.920954180304564e-06, "loss": 1.4253, "step": 28670 }, { "epoch": 71.6775, "grad_norm": 8.125, "learning_rate": 3.920311299134806e-06, "loss": 1.2462, "step": 28671 }, { "epoch": 71.68, "grad_norm": 5.0625, "learning_rate": 3.919668457823248e-06, "loss": 1.3669, "step": 28672 }, { "epoch": 71.6825, "grad_norm": 6.875, "learning_rate": 3.9190256563741006e-06, "loss": 1.3428, "step": 28673 }, { "epoch": 71.685, "grad_norm": 8.25, "learning_rate": 3.918382894791574e-06, "loss": 1.273, "step": 28674 }, { "epoch": 71.6875, "grad_norm": 5.5625, "learning_rate": 3.917740173079886e-06, "loss": 1.3347, "step": 28675 }, { "epoch": 71.69, "grad_norm": 10.6875, "learning_rate": 3.917097491243252e-06, "loss": 1.3298, "step": 28676 }, { "epoch": 71.6925, "grad_norm": 12.6875, "learning_rate": 3.916454849285885e-06, "loss": 1.3759, "step": 28677 }, { "epoch": 71.695, "grad_norm": 7.8125, "learning_rate": 3.915812247211993e-06, "loss": 1.2916, "step": 28678 }, { "epoch": 71.6975, "grad_norm": 9.5625, "learning_rate": 3.915169685025796e-06, "loss": 1.3739, "step": 28679 }, { "epoch": 71.7, "grad_norm": 11.75, "learning_rate": 3.914527162731498e-06, "loss": 1.4237, "step": 28680 }, { "epoch": 71.7025, "grad_norm": 11.4375, "learning_rate": 3.913884680333322e-06, "loss": 1.5058, "step": 28681 }, { "epoch": 71.705, "grad_norm": 7.0625, "learning_rate": 3.913242237835469e-06, "loss": 1.2278, "step": 28682 }, { "epoch": 71.7075, "grad_norm": 11.4375, "learning_rate": 3.912599835242162e-06, "loss": 1.2917, "step": 28683 }, { "epoch": 71.71, "grad_norm": 4.875, "learning_rate": 3.911957472557605e-06, "loss": 1.2843, "step": 28684 }, { "epoch": 71.7125, "grad_norm": 18.875, "learning_rate": 3.911315149786009e-06, "loss": 1.3392, "step": 28685 }, { "epoch": 71.715, "grad_norm": 7.90625, "learning_rate": 3.9106728669315865e-06, "loss": 1.1873, "step": 28686 }, { "epoch": 71.7175, "grad_norm": 7.3125, "learning_rate": 3.9100306239985515e-06, "loss": 1.1584, "step": 28687 }, { "epoch": 71.72, "grad_norm": 8.0625, "learning_rate": 3.909388420991113e-06, "loss": 1.4989, "step": 28688 }, { "epoch": 71.7225, "grad_norm": 6.3125, "learning_rate": 3.908746257913478e-06, "loss": 1.2428, "step": 28689 }, { "epoch": 71.725, "grad_norm": 8.875, "learning_rate": 3.908104134769858e-06, "loss": 1.3426, "step": 28690 }, { "epoch": 71.7275, "grad_norm": 7.625, "learning_rate": 3.907462051564467e-06, "loss": 1.2971, "step": 28691 }, { "epoch": 71.73, "grad_norm": 7.96875, "learning_rate": 3.906820008301509e-06, "loss": 1.3776, "step": 28692 }, { "epoch": 71.7325, "grad_norm": 11.3125, "learning_rate": 3.9061780049851936e-06, "loss": 1.2239, "step": 28693 }, { "epoch": 71.735, "grad_norm": 5.09375, "learning_rate": 3.905536041619735e-06, "loss": 1.4238, "step": 28694 }, { "epoch": 71.7375, "grad_norm": 9.9375, "learning_rate": 3.904894118209335e-06, "loss": 1.5709, "step": 28695 }, { "epoch": 71.74, "grad_norm": 9.0, "learning_rate": 3.904252234758208e-06, "loss": 1.4125, "step": 28696 }, { "epoch": 71.7425, "grad_norm": 6.78125, "learning_rate": 3.903610391270556e-06, "loss": 1.2657, "step": 28697 }, { "epoch": 71.745, "grad_norm": 7.3125, "learning_rate": 3.902968587750594e-06, "loss": 1.4009, "step": 28698 }, { "epoch": 71.7475, "grad_norm": 6.5, "learning_rate": 3.902326824202526e-06, "loss": 1.3654, "step": 28699 }, { "epoch": 71.75, "grad_norm": 6.6875, "learning_rate": 3.901685100630554e-06, "loss": 1.3231, "step": 28700 }, { "epoch": 71.7525, "grad_norm": 8.625, "learning_rate": 3.901043417038893e-06, "loss": 1.4097, "step": 28701 }, { "epoch": 71.755, "grad_norm": 7.6875, "learning_rate": 3.90040177343175e-06, "loss": 1.2854, "step": 28702 }, { "epoch": 71.7575, "grad_norm": 7.375, "learning_rate": 3.899760169813328e-06, "loss": 1.3702, "step": 28703 }, { "epoch": 71.76, "grad_norm": 10.3125, "learning_rate": 3.899118606187832e-06, "loss": 1.3343, "step": 28704 }, { "epoch": 71.7625, "grad_norm": 8.0, "learning_rate": 3.89847708255947e-06, "loss": 1.4209, "step": 28705 }, { "epoch": 71.765, "grad_norm": 6.625, "learning_rate": 3.897835598932451e-06, "loss": 1.2892, "step": 28706 }, { "epoch": 71.7675, "grad_norm": 7.59375, "learning_rate": 3.897194155310977e-06, "loss": 1.3953, "step": 28707 }, { "epoch": 71.77, "grad_norm": 7.3125, "learning_rate": 3.8965527516992514e-06, "loss": 1.2966, "step": 28708 }, { "epoch": 71.7725, "grad_norm": 7.5625, "learning_rate": 3.895911388101485e-06, "loss": 1.4646, "step": 28709 }, { "epoch": 71.775, "grad_norm": 5.875, "learning_rate": 3.895270064521877e-06, "loss": 1.3333, "step": 28710 }, { "epoch": 71.7775, "grad_norm": 5.9375, "learning_rate": 3.894628780964636e-06, "loss": 1.2989, "step": 28711 }, { "epoch": 71.78, "grad_norm": 7.25, "learning_rate": 3.893987537433961e-06, "loss": 1.4844, "step": 28712 }, { "epoch": 71.7825, "grad_norm": 6.21875, "learning_rate": 3.893346333934064e-06, "loss": 1.2443, "step": 28713 }, { "epoch": 71.785, "grad_norm": 5.78125, "learning_rate": 3.8927051704691435e-06, "loss": 1.4017, "step": 28714 }, { "epoch": 71.7875, "grad_norm": 8.0, "learning_rate": 3.892064047043399e-06, "loss": 1.1436, "step": 28715 }, { "epoch": 71.79, "grad_norm": 7.84375, "learning_rate": 3.891422963661039e-06, "loss": 1.313, "step": 28716 }, { "epoch": 71.7925, "grad_norm": 7.21875, "learning_rate": 3.890781920326268e-06, "loss": 1.2866, "step": 28717 }, { "epoch": 71.795, "grad_norm": 6.875, "learning_rate": 3.890140917043288e-06, "loss": 1.4257, "step": 28718 }, { "epoch": 71.7975, "grad_norm": 10.0, "learning_rate": 3.889499953816296e-06, "loss": 1.3402, "step": 28719 }, { "epoch": 71.8, "grad_norm": 8.8125, "learning_rate": 3.888859030649498e-06, "loss": 1.3093, "step": 28720 }, { "epoch": 71.8025, "grad_norm": 9.75, "learning_rate": 3.888218147547098e-06, "loss": 1.2959, "step": 28721 }, { "epoch": 71.805, "grad_norm": 7.46875, "learning_rate": 3.887577304513296e-06, "loss": 1.3116, "step": 28722 }, { "epoch": 71.8075, "grad_norm": 9.5625, "learning_rate": 3.886936501552288e-06, "loss": 1.2638, "step": 28723 }, { "epoch": 71.81, "grad_norm": 7.71875, "learning_rate": 3.886295738668281e-06, "loss": 1.3013, "step": 28724 }, { "epoch": 71.8125, "grad_norm": 8.0, "learning_rate": 3.885655015865477e-06, "loss": 1.3754, "step": 28725 }, { "epoch": 71.815, "grad_norm": 9.0625, "learning_rate": 3.8850143331480735e-06, "loss": 1.6128, "step": 28726 }, { "epoch": 71.8175, "grad_norm": 7.34375, "learning_rate": 3.88437369052027e-06, "loss": 1.292, "step": 28727 }, { "epoch": 71.82, "grad_norm": 7.6875, "learning_rate": 3.88373308798627e-06, "loss": 1.2042, "step": 28728 }, { "epoch": 71.8225, "grad_norm": 6.5625, "learning_rate": 3.883092525550267e-06, "loss": 1.2819, "step": 28729 }, { "epoch": 71.825, "grad_norm": 7.65625, "learning_rate": 3.88245200321647e-06, "loss": 1.3496, "step": 28730 }, { "epoch": 71.8275, "grad_norm": 7.125, "learning_rate": 3.881811520989067e-06, "loss": 1.3696, "step": 28731 }, { "epoch": 71.83, "grad_norm": 10.0, "learning_rate": 3.8811710788722674e-06, "loss": 1.3308, "step": 28732 }, { "epoch": 71.8325, "grad_norm": 5.0625, "learning_rate": 3.880530676870266e-06, "loss": 1.255, "step": 28733 }, { "epoch": 71.835, "grad_norm": 8.625, "learning_rate": 3.879890314987257e-06, "loss": 1.4516, "step": 28734 }, { "epoch": 71.8375, "grad_norm": 6.75, "learning_rate": 3.879249993227441e-06, "loss": 1.3763, "step": 28735 }, { "epoch": 71.84, "grad_norm": 5.1875, "learning_rate": 3.878609711595022e-06, "loss": 1.2453, "step": 28736 }, { "epoch": 71.8425, "grad_norm": 4.625, "learning_rate": 3.877969470094191e-06, "loss": 1.326, "step": 28737 }, { "epoch": 71.845, "grad_norm": 8.125, "learning_rate": 3.8773292687291454e-06, "loss": 1.3564, "step": 28738 }, { "epoch": 71.8475, "grad_norm": 6.5, "learning_rate": 3.876689107504084e-06, "loss": 1.3379, "step": 28739 }, { "epoch": 71.85, "grad_norm": 8.0, "learning_rate": 3.876048986423207e-06, "loss": 1.2784, "step": 28740 }, { "epoch": 71.8525, "grad_norm": 8.3125, "learning_rate": 3.875408905490706e-06, "loss": 1.2523, "step": 28741 }, { "epoch": 71.855, "grad_norm": 8.375, "learning_rate": 3.874768864710779e-06, "loss": 1.2354, "step": 28742 }, { "epoch": 71.8575, "grad_norm": 12.5, "learning_rate": 3.874128864087623e-06, "loss": 1.3125, "step": 28743 }, { "epoch": 71.86, "grad_norm": 6.40625, "learning_rate": 3.87348890362543e-06, "loss": 1.3923, "step": 28744 }, { "epoch": 71.8625, "grad_norm": 7.3125, "learning_rate": 3.872848983328402e-06, "loss": 1.3548, "step": 28745 }, { "epoch": 71.865, "grad_norm": 8.6875, "learning_rate": 3.872209103200728e-06, "loss": 1.3081, "step": 28746 }, { "epoch": 71.8675, "grad_norm": 8.8125, "learning_rate": 3.87156926324661e-06, "loss": 1.3824, "step": 28747 }, { "epoch": 71.87, "grad_norm": 7.53125, "learning_rate": 3.8709294634702374e-06, "loss": 1.3033, "step": 28748 }, { "epoch": 71.8725, "grad_norm": 7.9375, "learning_rate": 3.870289703875803e-06, "loss": 1.2816, "step": 28749 }, { "epoch": 71.875, "grad_norm": 7.28125, "learning_rate": 3.869649984467504e-06, "loss": 1.5299, "step": 28750 }, { "epoch": 71.8775, "grad_norm": 5.8125, "learning_rate": 3.869010305249538e-06, "loss": 1.1442, "step": 28751 }, { "epoch": 71.88, "grad_norm": 7.65625, "learning_rate": 3.8683706662260945e-06, "loss": 1.2948, "step": 28752 }, { "epoch": 71.8825, "grad_norm": 8.3125, "learning_rate": 3.867731067401368e-06, "loss": 1.6241, "step": 28753 }, { "epoch": 71.885, "grad_norm": 12.4375, "learning_rate": 3.867091508779545e-06, "loss": 1.2991, "step": 28754 }, { "epoch": 71.8875, "grad_norm": 7.40625, "learning_rate": 3.8664519903648305e-06, "loss": 1.336, "step": 28755 }, { "epoch": 71.89, "grad_norm": 7.15625, "learning_rate": 3.8658125121614106e-06, "loss": 1.249, "step": 28756 }, { "epoch": 71.8925, "grad_norm": 6.4375, "learning_rate": 3.86517307417348e-06, "loss": 1.4469, "step": 28757 }, { "epoch": 71.895, "grad_norm": 8.0, "learning_rate": 3.864533676405225e-06, "loss": 1.3238, "step": 28758 }, { "epoch": 71.8975, "grad_norm": 16.375, "learning_rate": 3.8638943188608424e-06, "loss": 1.3028, "step": 28759 }, { "epoch": 71.9, "grad_norm": 8.125, "learning_rate": 3.863255001544526e-06, "loss": 1.3633, "step": 28760 }, { "epoch": 71.9025, "grad_norm": 9.625, "learning_rate": 3.862615724460465e-06, "loss": 1.1898, "step": 28761 }, { "epoch": 71.905, "grad_norm": 6.21875, "learning_rate": 3.861976487612844e-06, "loss": 1.32, "step": 28762 }, { "epoch": 71.9075, "grad_norm": 6.59375, "learning_rate": 3.861337291005865e-06, "loss": 1.1691, "step": 28763 }, { "epoch": 71.91, "grad_norm": 7.9375, "learning_rate": 3.860698134643709e-06, "loss": 1.2391, "step": 28764 }, { "epoch": 71.9125, "grad_norm": 8.4375, "learning_rate": 3.860059018530571e-06, "loss": 1.466, "step": 28765 }, { "epoch": 71.915, "grad_norm": 11.6875, "learning_rate": 3.859419942670642e-06, "loss": 1.3409, "step": 28766 }, { "epoch": 71.9175, "grad_norm": 8.125, "learning_rate": 3.858780907068113e-06, "loss": 1.3349, "step": 28767 }, { "epoch": 71.92, "grad_norm": 7.15625, "learning_rate": 3.858141911727168e-06, "loss": 1.1935, "step": 28768 }, { "epoch": 71.9225, "grad_norm": 10.8125, "learning_rate": 3.857502956651993e-06, "loss": 1.2792, "step": 28769 }, { "epoch": 71.925, "grad_norm": 6.59375, "learning_rate": 3.85686404184679e-06, "loss": 1.2442, "step": 28770 }, { "epoch": 71.9275, "grad_norm": 7.21875, "learning_rate": 3.85622516731574e-06, "loss": 1.2767, "step": 28771 }, { "epoch": 71.93, "grad_norm": 7.9375, "learning_rate": 3.855586333063031e-06, "loss": 1.3074, "step": 28772 }, { "epoch": 71.9325, "grad_norm": 7.71875, "learning_rate": 3.85494753909285e-06, "loss": 1.2264, "step": 28773 }, { "epoch": 71.935, "grad_norm": 5.9375, "learning_rate": 3.854308785409388e-06, "loss": 1.224, "step": 28774 }, { "epoch": 71.9375, "grad_norm": 7.53125, "learning_rate": 3.853670072016833e-06, "loss": 1.5677, "step": 28775 }, { "epoch": 71.94, "grad_norm": 9.3125, "learning_rate": 3.853031398919372e-06, "loss": 1.2151, "step": 28776 }, { "epoch": 71.9425, "grad_norm": 11.0625, "learning_rate": 3.852392766121188e-06, "loss": 1.2025, "step": 28777 }, { "epoch": 71.945, "grad_norm": 7.1875, "learning_rate": 3.851754173626474e-06, "loss": 1.1036, "step": 28778 }, { "epoch": 71.9475, "grad_norm": 7.9375, "learning_rate": 3.8511156214394094e-06, "loss": 1.134, "step": 28779 }, { "epoch": 71.95, "grad_norm": 5.53125, "learning_rate": 3.8504771095641905e-06, "loss": 1.2316, "step": 28780 }, { "epoch": 71.9525, "grad_norm": 7.03125, "learning_rate": 3.849838638004992e-06, "loss": 1.2469, "step": 28781 }, { "epoch": 71.955, "grad_norm": 7.3125, "learning_rate": 3.84920020676601e-06, "loss": 1.4211, "step": 28782 }, { "epoch": 71.9575, "grad_norm": 8.6875, "learning_rate": 3.848561815851424e-06, "loss": 1.2527, "step": 28783 }, { "epoch": 71.96, "grad_norm": 7.6875, "learning_rate": 3.847923465265418e-06, "loss": 1.3278, "step": 28784 }, { "epoch": 71.9625, "grad_norm": 6.59375, "learning_rate": 3.8472851550121795e-06, "loss": 1.2127, "step": 28785 }, { "epoch": 71.965, "grad_norm": 8.5, "learning_rate": 3.846646885095896e-06, "loss": 1.3426, "step": 28786 }, { "epoch": 71.9675, "grad_norm": 8.1875, "learning_rate": 3.84600865552075e-06, "loss": 1.3742, "step": 28787 }, { "epoch": 71.97, "grad_norm": 9.625, "learning_rate": 3.845370466290922e-06, "loss": 1.3188, "step": 28788 }, { "epoch": 71.9725, "grad_norm": 6.78125, "learning_rate": 3.8447323174105976e-06, "loss": 1.1773, "step": 28789 }, { "epoch": 71.975, "grad_norm": 8.625, "learning_rate": 3.8440942088839665e-06, "loss": 1.2957, "step": 28790 }, { "epoch": 71.9775, "grad_norm": 6.125, "learning_rate": 3.843456140715207e-06, "loss": 1.4992, "step": 28791 }, { "epoch": 71.98, "grad_norm": 6.75, "learning_rate": 3.842818112908498e-06, "loss": 1.4188, "step": 28792 }, { "epoch": 71.9825, "grad_norm": 8.25, "learning_rate": 3.842180125468032e-06, "loss": 1.3555, "step": 28793 }, { "epoch": 71.985, "grad_norm": 10.5, "learning_rate": 3.841542178397983e-06, "loss": 1.3525, "step": 28794 }, { "epoch": 71.9875, "grad_norm": 7.84375, "learning_rate": 3.840904271702542e-06, "loss": 1.5017, "step": 28795 }, { "epoch": 71.99, "grad_norm": 8.75, "learning_rate": 3.84026640538588e-06, "loss": 1.355, "step": 28796 }, { "epoch": 71.9925, "grad_norm": 9.5, "learning_rate": 3.8396285794521895e-06, "loss": 1.4224, "step": 28797 }, { "epoch": 71.995, "grad_norm": 8.3125, "learning_rate": 3.838990793905647e-06, "loss": 1.2235, "step": 28798 }, { "epoch": 71.9975, "grad_norm": 9.0, "learning_rate": 3.838353048750432e-06, "loss": 1.5578, "step": 28799 }, { "epoch": 72.0, "grad_norm": 9.625, "learning_rate": 3.837715343990727e-06, "loss": 1.3128, "step": 28800 }, { "epoch": 72.0025, "grad_norm": 7.1875, "learning_rate": 3.837077679630718e-06, "loss": 1.2949, "step": 28801 }, { "epoch": 72.005, "grad_norm": 7.90625, "learning_rate": 3.83644005567458e-06, "loss": 1.4254, "step": 28802 }, { "epoch": 72.0075, "grad_norm": 5.15625, "learning_rate": 3.835802472126491e-06, "loss": 1.5209, "step": 28803 }, { "epoch": 72.01, "grad_norm": 9.9375, "learning_rate": 3.835164928990635e-06, "loss": 1.415, "step": 28804 }, { "epoch": 72.0125, "grad_norm": 8.3125, "learning_rate": 3.834527426271193e-06, "loss": 1.4203, "step": 28805 }, { "epoch": 72.015, "grad_norm": 6.40625, "learning_rate": 3.833889963972344e-06, "loss": 1.2188, "step": 28806 }, { "epoch": 72.0175, "grad_norm": 7.53125, "learning_rate": 3.833252542098261e-06, "loss": 1.275, "step": 28807 }, { "epoch": 72.02, "grad_norm": 5.6875, "learning_rate": 3.832615160653131e-06, "loss": 1.2093, "step": 28808 }, { "epoch": 72.0225, "grad_norm": 6.78125, "learning_rate": 3.831977819641127e-06, "loss": 1.2326, "step": 28809 }, { "epoch": 72.025, "grad_norm": 6.53125, "learning_rate": 3.831340519066433e-06, "loss": 1.2387, "step": 28810 }, { "epoch": 72.0275, "grad_norm": 7.0625, "learning_rate": 3.83070325893322e-06, "loss": 1.3014, "step": 28811 }, { "epoch": 72.03, "grad_norm": 9.8125, "learning_rate": 3.830066039245673e-06, "loss": 1.4155, "step": 28812 }, { "epoch": 72.0325, "grad_norm": 12.25, "learning_rate": 3.829428860007966e-06, "loss": 1.401, "step": 28813 }, { "epoch": 72.035, "grad_norm": 7.90625, "learning_rate": 3.828791721224272e-06, "loss": 1.2132, "step": 28814 }, { "epoch": 72.0375, "grad_norm": 8.375, "learning_rate": 3.828154622898773e-06, "loss": 1.2599, "step": 28815 }, { "epoch": 72.04, "grad_norm": 7.84375, "learning_rate": 3.8275175650356485e-06, "loss": 1.2822, "step": 28816 }, { "epoch": 72.0425, "grad_norm": 9.375, "learning_rate": 3.826880547639073e-06, "loss": 1.3789, "step": 28817 }, { "epoch": 72.045, "grad_norm": 7.78125, "learning_rate": 3.826243570713217e-06, "loss": 1.2506, "step": 28818 }, { "epoch": 72.0475, "grad_norm": 6.71875, "learning_rate": 3.825606634262262e-06, "loss": 1.3583, "step": 28819 }, { "epoch": 72.05, "grad_norm": 7.28125, "learning_rate": 3.824969738290386e-06, "loss": 1.3873, "step": 28820 }, { "epoch": 72.0525, "grad_norm": 9.3125, "learning_rate": 3.824332882801761e-06, "loss": 1.3963, "step": 28821 }, { "epoch": 72.055, "grad_norm": 6.65625, "learning_rate": 3.823696067800559e-06, "loss": 1.2464, "step": 28822 }, { "epoch": 72.0575, "grad_norm": 10.0625, "learning_rate": 3.823059293290963e-06, "loss": 1.3388, "step": 28823 }, { "epoch": 72.06, "grad_norm": 5.15625, "learning_rate": 3.822422559277139e-06, "loss": 1.2363, "step": 28824 }, { "epoch": 72.0625, "grad_norm": 10.5625, "learning_rate": 3.821785865763269e-06, "loss": 1.3919, "step": 28825 }, { "epoch": 72.065, "grad_norm": 8.0625, "learning_rate": 3.82114921275352e-06, "loss": 1.2381, "step": 28826 }, { "epoch": 72.0675, "grad_norm": 9.3125, "learning_rate": 3.820512600252072e-06, "loss": 1.4626, "step": 28827 }, { "epoch": 72.07, "grad_norm": 7.75, "learning_rate": 3.819876028263097e-06, "loss": 1.357, "step": 28828 }, { "epoch": 72.0725, "grad_norm": 15.25, "learning_rate": 3.8192394967907645e-06, "loss": 1.4059, "step": 28829 }, { "epoch": 72.075, "grad_norm": 7.46875, "learning_rate": 3.8186030058392495e-06, "loss": 1.3207, "step": 28830 }, { "epoch": 72.0775, "grad_norm": 9.0, "learning_rate": 3.81796655541273e-06, "loss": 1.4761, "step": 28831 }, { "epoch": 72.08, "grad_norm": 6.96875, "learning_rate": 3.817330145515374e-06, "loss": 1.3621, "step": 28832 }, { "epoch": 72.0825, "grad_norm": 5.84375, "learning_rate": 3.816693776151351e-06, "loss": 1.3701, "step": 28833 }, { "epoch": 72.085, "grad_norm": 7.5625, "learning_rate": 3.8160574473248355e-06, "loss": 1.319, "step": 28834 }, { "epoch": 72.0875, "grad_norm": 8.1875, "learning_rate": 3.815421159040005e-06, "loss": 1.3481, "step": 28835 }, { "epoch": 72.09, "grad_norm": 7.84375, "learning_rate": 3.814784911301024e-06, "loss": 1.19, "step": 28836 }, { "epoch": 72.0925, "grad_norm": 5.15625, "learning_rate": 3.814148704112063e-06, "loss": 1.3598, "step": 28837 }, { "epoch": 72.095, "grad_norm": 9.3125, "learning_rate": 3.8135125374772985e-06, "loss": 1.291, "step": 28838 }, { "epoch": 72.0975, "grad_norm": 10.6875, "learning_rate": 3.8128764114008952e-06, "loss": 1.3661, "step": 28839 }, { "epoch": 72.1, "grad_norm": 5.9375, "learning_rate": 3.81224032588703e-06, "loss": 1.2878, "step": 28840 }, { "epoch": 72.1025, "grad_norm": 8.0, "learning_rate": 3.811604280939867e-06, "loss": 1.2977, "step": 28841 }, { "epoch": 72.105, "grad_norm": 7.15625, "learning_rate": 3.8109682765635803e-06, "loss": 1.2689, "step": 28842 }, { "epoch": 72.1075, "grad_norm": 6.46875, "learning_rate": 3.8103323127623395e-06, "loss": 1.3928, "step": 28843 }, { "epoch": 72.11, "grad_norm": 7.9375, "learning_rate": 3.8096963895403084e-06, "loss": 1.2397, "step": 28844 }, { "epoch": 72.1125, "grad_norm": 6.5, "learning_rate": 3.8090605069016596e-06, "loss": 1.215, "step": 28845 }, { "epoch": 72.115, "grad_norm": 7.21875, "learning_rate": 3.8084246648505663e-06, "loss": 1.4393, "step": 28846 }, { "epoch": 72.1175, "grad_norm": 9.9375, "learning_rate": 3.807788863391193e-06, "loss": 1.2719, "step": 28847 }, { "epoch": 72.12, "grad_norm": 7.4375, "learning_rate": 3.807153102527704e-06, "loss": 1.3354, "step": 28848 }, { "epoch": 72.1225, "grad_norm": 7.0, "learning_rate": 3.8065173822642722e-06, "loss": 1.174, "step": 28849 }, { "epoch": 72.125, "grad_norm": 6.4375, "learning_rate": 3.8058817026050676e-06, "loss": 1.4344, "step": 28850 }, { "epoch": 72.1275, "grad_norm": 6.53125, "learning_rate": 3.8052460635542555e-06, "loss": 1.2558, "step": 28851 }, { "epoch": 72.13, "grad_norm": 5.25, "learning_rate": 3.8046104651159987e-06, "loss": 1.217, "step": 28852 }, { "epoch": 72.1325, "grad_norm": 8.3125, "learning_rate": 3.803974907294471e-06, "loss": 1.3375, "step": 28853 }, { "epoch": 72.135, "grad_norm": 9.375, "learning_rate": 3.8033393900938322e-06, "loss": 1.3334, "step": 28854 }, { "epoch": 72.1375, "grad_norm": 5.78125, "learning_rate": 3.802703913518255e-06, "loss": 1.3008, "step": 28855 }, { "epoch": 72.14, "grad_norm": 9.3125, "learning_rate": 3.802068477571901e-06, "loss": 1.254, "step": 28856 }, { "epoch": 72.1425, "grad_norm": 7.5, "learning_rate": 3.8014330822589406e-06, "loss": 1.3547, "step": 28857 }, { "epoch": 72.145, "grad_norm": 7.96875, "learning_rate": 3.800797727583537e-06, "loss": 1.2167, "step": 28858 }, { "epoch": 72.1475, "grad_norm": 5.25, "learning_rate": 3.8001624135498527e-06, "loss": 1.457, "step": 28859 }, { "epoch": 72.15, "grad_norm": 5.71875, "learning_rate": 3.7995271401620548e-06, "loss": 1.3158, "step": 28860 }, { "epoch": 72.1525, "grad_norm": 7.125, "learning_rate": 3.7988919074243124e-06, "loss": 1.331, "step": 28861 }, { "epoch": 72.155, "grad_norm": 7.1875, "learning_rate": 3.7982567153407867e-06, "loss": 1.3245, "step": 28862 }, { "epoch": 72.1575, "grad_norm": 8.0625, "learning_rate": 3.797621563915638e-06, "loss": 1.2919, "step": 28863 }, { "epoch": 72.16, "grad_norm": 10.1875, "learning_rate": 3.7969864531530344e-06, "loss": 1.5267, "step": 28864 }, { "epoch": 72.1625, "grad_norm": 7.34375, "learning_rate": 3.796351383057142e-06, "loss": 1.2913, "step": 28865 }, { "epoch": 72.165, "grad_norm": 9.0625, "learning_rate": 3.7957163536321228e-06, "loss": 1.2864, "step": 28866 }, { "epoch": 72.1675, "grad_norm": 7.84375, "learning_rate": 3.795081364882135e-06, "loss": 1.2969, "step": 28867 }, { "epoch": 72.17, "grad_norm": 7.59375, "learning_rate": 3.7944464168113494e-06, "loss": 1.2329, "step": 28868 }, { "epoch": 72.1725, "grad_norm": 6.875, "learning_rate": 3.7938115094239202e-06, "loss": 1.2413, "step": 28869 }, { "epoch": 72.175, "grad_norm": 8.625, "learning_rate": 3.793176642724019e-06, "loss": 1.3708, "step": 28870 }, { "epoch": 72.1775, "grad_norm": 6.90625, "learning_rate": 3.792541816715799e-06, "loss": 1.3396, "step": 28871 }, { "epoch": 72.18, "grad_norm": 10.25, "learning_rate": 3.79190703140343e-06, "loss": 1.3295, "step": 28872 }, { "epoch": 72.1825, "grad_norm": 7.3125, "learning_rate": 3.7912722867910712e-06, "loss": 1.3989, "step": 28873 }, { "epoch": 72.185, "grad_norm": 6.46875, "learning_rate": 3.790637582882878e-06, "loss": 1.4964, "step": 28874 }, { "epoch": 72.1875, "grad_norm": 5.96875, "learning_rate": 3.7900029196830167e-06, "loss": 1.3418, "step": 28875 }, { "epoch": 72.19, "grad_norm": 8.9375, "learning_rate": 3.7893682971956512e-06, "loss": 1.3178, "step": 28876 }, { "epoch": 72.1925, "grad_norm": 7.4375, "learning_rate": 3.7887337154249393e-06, "loss": 1.3942, "step": 28877 }, { "epoch": 72.195, "grad_norm": 4.53125, "learning_rate": 3.788099174375036e-06, "loss": 1.3436, "step": 28878 }, { "epoch": 72.1975, "grad_norm": 5.6875, "learning_rate": 3.7874646740501076e-06, "loss": 1.2809, "step": 28879 }, { "epoch": 72.2, "grad_norm": 5.28125, "learning_rate": 3.7868302144543146e-06, "loss": 1.2099, "step": 28880 }, { "epoch": 72.2025, "grad_norm": 9.5, "learning_rate": 3.786195795591814e-06, "loss": 1.3816, "step": 28881 }, { "epoch": 72.205, "grad_norm": 7.125, "learning_rate": 3.785561417466761e-06, "loss": 1.1683, "step": 28882 }, { "epoch": 72.2075, "grad_norm": 4.6875, "learning_rate": 3.7849270800833226e-06, "loss": 1.3263, "step": 28883 }, { "epoch": 72.21, "grad_norm": 5.625, "learning_rate": 3.7842927834456512e-06, "loss": 1.2697, "step": 28884 }, { "epoch": 72.2125, "grad_norm": 5.0625, "learning_rate": 3.7836585275579096e-06, "loss": 1.3264, "step": 28885 }, { "epoch": 72.215, "grad_norm": 5.5625, "learning_rate": 3.7830243124242514e-06, "loss": 1.2709, "step": 28886 }, { "epoch": 72.2175, "grad_norm": 8.125, "learning_rate": 3.782390138048839e-06, "loss": 1.2435, "step": 28887 }, { "epoch": 72.22, "grad_norm": 7.25, "learning_rate": 3.7817560044358302e-06, "loss": 1.2681, "step": 28888 }, { "epoch": 72.2225, "grad_norm": 11.125, "learning_rate": 3.781121911589375e-06, "loss": 1.2316, "step": 28889 }, { "epoch": 72.225, "grad_norm": 7.8125, "learning_rate": 3.7804878595136363e-06, "loss": 1.2647, "step": 28890 }, { "epoch": 72.2275, "grad_norm": 9.1875, "learning_rate": 3.7798538482127746e-06, "loss": 1.5266, "step": 28891 }, { "epoch": 72.23, "grad_norm": 6.09375, "learning_rate": 3.7792198776909406e-06, "loss": 1.2909, "step": 28892 }, { "epoch": 72.2325, "grad_norm": 9.875, "learning_rate": 3.7785859479522934e-06, "loss": 1.2491, "step": 28893 }, { "epoch": 72.235, "grad_norm": 7.625, "learning_rate": 3.77795205900098e-06, "loss": 1.5251, "step": 28894 }, { "epoch": 72.2375, "grad_norm": 8.0625, "learning_rate": 3.777318210841171e-06, "loss": 1.3278, "step": 28895 }, { "epoch": 72.24, "grad_norm": 7.0, "learning_rate": 3.7766844034770155e-06, "loss": 1.3518, "step": 28896 }, { "epoch": 72.2425, "grad_norm": 7.34375, "learning_rate": 3.7760506369126672e-06, "loss": 1.285, "step": 28897 }, { "epoch": 72.245, "grad_norm": 8.125, "learning_rate": 3.7754169111522787e-06, "loss": 1.3381, "step": 28898 }, { "epoch": 72.2475, "grad_norm": 6.3125, "learning_rate": 3.7747832262000084e-06, "loss": 1.256, "step": 28899 }, { "epoch": 72.25, "grad_norm": 7.46875, "learning_rate": 3.7741495820600128e-06, "loss": 1.3666, "step": 28900 }, { "epoch": 72.2525, "grad_norm": 6.90625, "learning_rate": 3.7735159787364393e-06, "loss": 1.3098, "step": 28901 }, { "epoch": 72.255, "grad_norm": 9.1875, "learning_rate": 3.772882416233451e-06, "loss": 1.5396, "step": 28902 }, { "epoch": 72.2575, "grad_norm": 10.75, "learning_rate": 3.772248894555195e-06, "loss": 1.2825, "step": 28903 }, { "epoch": 72.26, "grad_norm": 7.125, "learning_rate": 3.7716154137058236e-06, "loss": 1.3405, "step": 28904 }, { "epoch": 72.2625, "grad_norm": 6.40625, "learning_rate": 3.7709819736894913e-06, "loss": 1.3163, "step": 28905 }, { "epoch": 72.265, "grad_norm": 9.75, "learning_rate": 3.7703485745103564e-06, "loss": 1.3432, "step": 28906 }, { "epoch": 72.2675, "grad_norm": 5.21875, "learning_rate": 3.769715216172567e-06, "loss": 1.4273, "step": 28907 }, { "epoch": 72.27, "grad_norm": 9.0, "learning_rate": 3.7690818986802744e-06, "loss": 1.2236, "step": 28908 }, { "epoch": 72.2725, "grad_norm": 7.28125, "learning_rate": 3.768448622037625e-06, "loss": 1.1867, "step": 28909 }, { "epoch": 72.275, "grad_norm": 6.28125, "learning_rate": 3.7678153862487844e-06, "loss": 1.4788, "step": 28910 }, { "epoch": 72.2775, "grad_norm": 11.625, "learning_rate": 3.7671821913178953e-06, "loss": 1.4583, "step": 28911 }, { "epoch": 72.28, "grad_norm": 7.15625, "learning_rate": 3.766549037249112e-06, "loss": 1.1729, "step": 28912 }, { "epoch": 72.2825, "grad_norm": 8.0, "learning_rate": 3.765915924046579e-06, "loss": 1.2466, "step": 28913 }, { "epoch": 72.285, "grad_norm": 7.5, "learning_rate": 3.7652828517144523e-06, "loss": 1.4373, "step": 28914 }, { "epoch": 72.2875, "grad_norm": 7.1875, "learning_rate": 3.764649820256886e-06, "loss": 1.3555, "step": 28915 }, { "epoch": 72.29, "grad_norm": 8.1875, "learning_rate": 3.7640168296780243e-06, "loss": 1.3739, "step": 28916 }, { "epoch": 72.2925, "grad_norm": 6.6875, "learning_rate": 3.7633838799820164e-06, "loss": 1.3981, "step": 28917 }, { "epoch": 72.295, "grad_norm": 11.8125, "learning_rate": 3.7627509711730183e-06, "loss": 1.3595, "step": 28918 }, { "epoch": 72.2975, "grad_norm": 6.75, "learning_rate": 3.76211810325517e-06, "loss": 1.3031, "step": 28919 }, { "epoch": 72.3, "grad_norm": 8.5625, "learning_rate": 3.7614852762326303e-06, "loss": 1.4314, "step": 28920 }, { "epoch": 72.3025, "grad_norm": 5.96875, "learning_rate": 3.76085249010954e-06, "loss": 1.533, "step": 28921 }, { "epoch": 72.305, "grad_norm": 6.46875, "learning_rate": 3.7602197448900547e-06, "loss": 1.3831, "step": 28922 }, { "epoch": 72.3075, "grad_norm": 8.5625, "learning_rate": 3.7595870405783185e-06, "loss": 1.3703, "step": 28923 }, { "epoch": 72.31, "grad_norm": 9.75, "learning_rate": 3.7589543771784765e-06, "loss": 1.3236, "step": 28924 }, { "epoch": 72.3125, "grad_norm": 8.5, "learning_rate": 3.7583217546946805e-06, "loss": 1.3445, "step": 28925 }, { "epoch": 72.315, "grad_norm": 6.9375, "learning_rate": 3.7576891731310805e-06, "loss": 1.2899, "step": 28926 }, { "epoch": 72.3175, "grad_norm": 7.21875, "learning_rate": 3.75705663249182e-06, "loss": 1.4756, "step": 28927 }, { "epoch": 72.32, "grad_norm": 5.28125, "learning_rate": 3.7564241327810436e-06, "loss": 1.2318, "step": 28928 }, { "epoch": 72.3225, "grad_norm": 5.78125, "learning_rate": 3.7557916740029e-06, "loss": 1.3638, "step": 28929 }, { "epoch": 72.325, "grad_norm": 7.84375, "learning_rate": 3.7551592561615403e-06, "loss": 1.3773, "step": 28930 }, { "epoch": 72.3275, "grad_norm": 10.6875, "learning_rate": 3.754526879261107e-06, "loss": 1.3777, "step": 28931 }, { "epoch": 72.33, "grad_norm": 9.0625, "learning_rate": 3.7538945433057407e-06, "loss": 1.3738, "step": 28932 }, { "epoch": 72.3325, "grad_norm": 7.65625, "learning_rate": 3.7532622482995964e-06, "loss": 1.3371, "step": 28933 }, { "epoch": 72.335, "grad_norm": 6.03125, "learning_rate": 3.7526299942468115e-06, "loss": 1.2461, "step": 28934 }, { "epoch": 72.3375, "grad_norm": 9.3125, "learning_rate": 3.7519977811515373e-06, "loss": 1.4077, "step": 28935 }, { "epoch": 72.34, "grad_norm": 9.0625, "learning_rate": 3.7513656090179116e-06, "loss": 1.2456, "step": 28936 }, { "epoch": 72.3425, "grad_norm": 13.3125, "learning_rate": 3.750733477850087e-06, "loss": 1.2941, "step": 28937 }, { "epoch": 72.345, "grad_norm": 11.8125, "learning_rate": 3.7501013876522043e-06, "loss": 1.355, "step": 28938 }, { "epoch": 72.3475, "grad_norm": 10.0, "learning_rate": 3.749469338428402e-06, "loss": 1.3282, "step": 28939 }, { "epoch": 72.35, "grad_norm": 7.375, "learning_rate": 3.7488373301828296e-06, "loss": 1.3838, "step": 28940 }, { "epoch": 72.3525, "grad_norm": 8.125, "learning_rate": 3.7482053629196323e-06, "loss": 1.3471, "step": 28941 }, { "epoch": 72.355, "grad_norm": 5.28125, "learning_rate": 3.747573436642952e-06, "loss": 1.3312, "step": 28942 }, { "epoch": 72.3575, "grad_norm": 7.0, "learning_rate": 3.746941551356925e-06, "loss": 1.2836, "step": 28943 }, { "epoch": 72.36, "grad_norm": 8.3125, "learning_rate": 3.7463097070656995e-06, "loss": 1.2065, "step": 28944 }, { "epoch": 72.3625, "grad_norm": 8.9375, "learning_rate": 3.7456779037734203e-06, "loss": 1.3715, "step": 28945 }, { "epoch": 72.365, "grad_norm": 6.875, "learning_rate": 3.7450461414842264e-06, "loss": 1.2888, "step": 28946 }, { "epoch": 72.3675, "grad_norm": 7.125, "learning_rate": 3.744414420202257e-06, "loss": 1.3037, "step": 28947 }, { "epoch": 72.37, "grad_norm": 9.1875, "learning_rate": 3.7437827399316594e-06, "loss": 1.3909, "step": 28948 }, { "epoch": 72.3725, "grad_norm": 5.875, "learning_rate": 3.7431511006765685e-06, "loss": 1.2352, "step": 28949 }, { "epoch": 72.375, "grad_norm": 7.90625, "learning_rate": 3.742519502441132e-06, "loss": 1.1742, "step": 28950 }, { "epoch": 72.3775, "grad_norm": 5.71875, "learning_rate": 3.741887945229483e-06, "loss": 1.4165, "step": 28951 }, { "epoch": 72.38, "grad_norm": 8.4375, "learning_rate": 3.741256429045771e-06, "loss": 1.2518, "step": 28952 }, { "epoch": 72.3825, "grad_norm": 12.125, "learning_rate": 3.740624953894131e-06, "loss": 1.4023, "step": 28953 }, { "epoch": 72.385, "grad_norm": 9.625, "learning_rate": 3.7399935197786986e-06, "loss": 1.365, "step": 28954 }, { "epoch": 72.3875, "grad_norm": 8.4375, "learning_rate": 3.7393621267036194e-06, "loss": 1.355, "step": 28955 }, { "epoch": 72.39, "grad_norm": 7.0, "learning_rate": 3.738730774673035e-06, "loss": 1.1768, "step": 28956 }, { "epoch": 72.3925, "grad_norm": 6.875, "learning_rate": 3.7380994636910807e-06, "loss": 1.218, "step": 28957 }, { "epoch": 72.395, "grad_norm": 7.09375, "learning_rate": 3.737468193761893e-06, "loss": 1.2622, "step": 28958 }, { "epoch": 72.3975, "grad_norm": 4.84375, "learning_rate": 3.736836964889612e-06, "loss": 1.3288, "step": 28959 }, { "epoch": 72.4, "grad_norm": 8.75, "learning_rate": 3.736205777078381e-06, "loss": 1.2821, "step": 28960 }, { "epoch": 72.4025, "grad_norm": 7.0, "learning_rate": 3.735574630332335e-06, "loss": 1.3459, "step": 28961 }, { "epoch": 72.405, "grad_norm": 8.5625, "learning_rate": 3.7349435246556077e-06, "loss": 1.3779, "step": 28962 }, { "epoch": 72.4075, "grad_norm": 9.9375, "learning_rate": 3.7343124600523427e-06, "loss": 1.2593, "step": 28963 }, { "epoch": 72.41, "grad_norm": 7.4375, "learning_rate": 3.7336814365266706e-06, "loss": 1.232, "step": 28964 }, { "epoch": 72.4125, "grad_norm": 6.4375, "learning_rate": 3.733050454082736e-06, "loss": 1.3354, "step": 28965 }, { "epoch": 72.415, "grad_norm": 6.75, "learning_rate": 3.7324195127246685e-06, "loss": 1.1217, "step": 28966 }, { "epoch": 72.4175, "grad_norm": 9.6875, "learning_rate": 3.7317886124566115e-06, "loss": 1.4326, "step": 28967 }, { "epoch": 72.42, "grad_norm": 10.75, "learning_rate": 3.7311577532826946e-06, "loss": 1.2134, "step": 28968 }, { "epoch": 72.4225, "grad_norm": 6.1875, "learning_rate": 3.7305269352070594e-06, "loss": 1.3897, "step": 28969 }, { "epoch": 72.425, "grad_norm": 7.6875, "learning_rate": 3.7298961582338356e-06, "loss": 1.3609, "step": 28970 }, { "epoch": 72.4275, "grad_norm": 7.375, "learning_rate": 3.7292654223671644e-06, "loss": 1.3176, "step": 28971 }, { "epoch": 72.43, "grad_norm": 7.21875, "learning_rate": 3.728634727611179e-06, "loss": 1.4567, "step": 28972 }, { "epoch": 72.4325, "grad_norm": 9.0625, "learning_rate": 3.72800407397001e-06, "loss": 1.2671, "step": 28973 }, { "epoch": 72.435, "grad_norm": 6.84375, "learning_rate": 3.727373461447794e-06, "loss": 1.3566, "step": 28974 }, { "epoch": 72.4375, "grad_norm": 5.40625, "learning_rate": 3.726742890048671e-06, "loss": 1.4712, "step": 28975 }, { "epoch": 72.44, "grad_norm": 6.65625, "learning_rate": 3.72611235977677e-06, "loss": 1.4087, "step": 28976 }, { "epoch": 72.4425, "grad_norm": 11.5, "learning_rate": 3.725481870636223e-06, "loss": 1.2954, "step": 28977 }, { "epoch": 72.445, "grad_norm": 8.8125, "learning_rate": 3.7248514226311648e-06, "loss": 1.3092, "step": 28978 }, { "epoch": 72.4475, "grad_norm": 5.375, "learning_rate": 3.7242210157657334e-06, "loss": 1.1914, "step": 28979 }, { "epoch": 72.45, "grad_norm": 6.34375, "learning_rate": 3.7235906500440576e-06, "loss": 1.4086, "step": 28980 }, { "epoch": 72.4525, "grad_norm": 8.25, "learning_rate": 3.7229603254702673e-06, "loss": 1.2304, "step": 28981 }, { "epoch": 72.455, "grad_norm": 6.6875, "learning_rate": 3.722330042048501e-06, "loss": 1.3112, "step": 28982 }, { "epoch": 72.4575, "grad_norm": 6.625, "learning_rate": 3.7216997997828843e-06, "loss": 1.3692, "step": 28983 }, { "epoch": 72.46, "grad_norm": 6.84375, "learning_rate": 3.721069598677556e-06, "loss": 1.3032, "step": 28984 }, { "epoch": 72.4625, "grad_norm": 12.5625, "learning_rate": 3.7204394387366405e-06, "loss": 1.347, "step": 28985 }, { "epoch": 72.465, "grad_norm": 9.5, "learning_rate": 3.719809319964276e-06, "loss": 1.2259, "step": 28986 }, { "epoch": 72.4675, "grad_norm": 5.8125, "learning_rate": 3.719179242364591e-06, "loss": 1.3292, "step": 28987 }, { "epoch": 72.47, "grad_norm": 6.78125, "learning_rate": 3.7185492059417115e-06, "loss": 1.2788, "step": 28988 }, { "epoch": 72.4725, "grad_norm": 6.8125, "learning_rate": 3.717919210699772e-06, "loss": 1.2811, "step": 28989 }, { "epoch": 72.475, "grad_norm": 11.5, "learning_rate": 3.7172892566429064e-06, "loss": 1.1593, "step": 28990 }, { "epoch": 72.4775, "grad_norm": 6.4375, "learning_rate": 3.7166593437752418e-06, "loss": 1.2686, "step": 28991 }, { "epoch": 72.48, "grad_norm": 8.6875, "learning_rate": 3.7160294721009026e-06, "loss": 1.3226, "step": 28992 }, { "epoch": 72.4825, "grad_norm": 6.25, "learning_rate": 3.7153996416240235e-06, "loss": 1.224, "step": 28993 }, { "epoch": 72.485, "grad_norm": 7.0, "learning_rate": 3.7147698523487352e-06, "loss": 1.4556, "step": 28994 }, { "epoch": 72.4875, "grad_norm": 11.4375, "learning_rate": 3.7141401042791646e-06, "loss": 1.241, "step": 28995 }, { "epoch": 72.49, "grad_norm": 7.875, "learning_rate": 3.7135103974194376e-06, "loss": 1.3579, "step": 28996 }, { "epoch": 72.4925, "grad_norm": 6.375, "learning_rate": 3.712880731773687e-06, "loss": 1.2954, "step": 28997 }, { "epoch": 72.495, "grad_norm": 9.125, "learning_rate": 3.7122511073460356e-06, "loss": 1.4856, "step": 28998 }, { "epoch": 72.4975, "grad_norm": 7.0625, "learning_rate": 3.7116215241406184e-06, "loss": 1.3715, "step": 28999 }, { "epoch": 72.5, "grad_norm": 9.9375, "learning_rate": 3.7109919821615546e-06, "loss": 1.3617, "step": 29000 }, { "epoch": 72.5025, "grad_norm": 6.40625, "learning_rate": 3.710362481412979e-06, "loss": 1.2033, "step": 29001 }, { "epoch": 72.505, "grad_norm": 6.46875, "learning_rate": 3.709733021899016e-06, "loss": 1.2447, "step": 29002 }, { "epoch": 72.5075, "grad_norm": 8.9375, "learning_rate": 3.7091036036237883e-06, "loss": 1.3642, "step": 29003 }, { "epoch": 72.51, "grad_norm": 14.375, "learning_rate": 3.708474226591425e-06, "loss": 1.3119, "step": 29004 }, { "epoch": 72.5125, "grad_norm": 7.53125, "learning_rate": 3.707844890806056e-06, "loss": 1.2773, "step": 29005 }, { "epoch": 72.515, "grad_norm": 9.25, "learning_rate": 3.707215596271805e-06, "loss": 1.2497, "step": 29006 }, { "epoch": 72.5175, "grad_norm": 7.25, "learning_rate": 3.706586342992793e-06, "loss": 1.2404, "step": 29007 }, { "epoch": 72.52, "grad_norm": 35.0, "learning_rate": 3.705957130973149e-06, "loss": 1.239, "step": 29008 }, { "epoch": 72.5225, "grad_norm": 5.625, "learning_rate": 3.705327960217001e-06, "loss": 1.2859, "step": 29009 }, { "epoch": 72.525, "grad_norm": 10.0625, "learning_rate": 3.7046988307284725e-06, "loss": 1.2798, "step": 29010 }, { "epoch": 72.5275, "grad_norm": 10.9375, "learning_rate": 3.7040697425116824e-06, "loss": 1.1548, "step": 29011 }, { "epoch": 72.53, "grad_norm": 7.03125, "learning_rate": 3.703440695570761e-06, "loss": 1.2884, "step": 29012 }, { "epoch": 72.5325, "grad_norm": 5.9375, "learning_rate": 3.702811689909829e-06, "loss": 1.4112, "step": 29013 }, { "epoch": 72.535, "grad_norm": 6.46875, "learning_rate": 3.7021827255330143e-06, "loss": 1.3097, "step": 29014 }, { "epoch": 72.5375, "grad_norm": 8.25, "learning_rate": 3.701553802444434e-06, "loss": 1.1533, "step": 29015 }, { "epoch": 72.54, "grad_norm": 6.75, "learning_rate": 3.700924920648219e-06, "loss": 1.5098, "step": 29016 }, { "epoch": 72.5425, "grad_norm": 6.5, "learning_rate": 3.7002960801484876e-06, "loss": 1.2377, "step": 29017 }, { "epoch": 72.545, "grad_norm": 6.96875, "learning_rate": 3.6996672809493594e-06, "loss": 1.368, "step": 29018 }, { "epoch": 72.5475, "grad_norm": 7.375, "learning_rate": 3.6990385230549607e-06, "loss": 1.3872, "step": 29019 }, { "epoch": 72.55, "grad_norm": 7.96875, "learning_rate": 3.6984098064694174e-06, "loss": 1.3357, "step": 29020 }, { "epoch": 72.5525, "grad_norm": 8.75, "learning_rate": 3.697781131196846e-06, "loss": 1.2893, "step": 29021 }, { "epoch": 72.555, "grad_norm": 5.96875, "learning_rate": 3.6971524972413666e-06, "loss": 1.3891, "step": 29022 }, { "epoch": 72.5575, "grad_norm": 6.53125, "learning_rate": 3.6965239046071032e-06, "loss": 1.3152, "step": 29023 }, { "epoch": 72.56, "grad_norm": 6.96875, "learning_rate": 3.69589535329818e-06, "loss": 1.2806, "step": 29024 }, { "epoch": 72.5625, "grad_norm": 8.4375, "learning_rate": 3.6952668433187145e-06, "loss": 1.4716, "step": 29025 }, { "epoch": 72.565, "grad_norm": 5.6875, "learning_rate": 3.6946383746728243e-06, "loss": 1.351, "step": 29026 }, { "epoch": 72.5675, "grad_norm": 9.75, "learning_rate": 3.694009947364636e-06, "loss": 1.2547, "step": 29027 }, { "epoch": 72.57, "grad_norm": 10.4375, "learning_rate": 3.6933815613982626e-06, "loss": 1.3956, "step": 29028 }, { "epoch": 72.5725, "grad_norm": 5.46875, "learning_rate": 3.69275321677783e-06, "loss": 1.2323, "step": 29029 }, { "epoch": 72.575, "grad_norm": 7.5625, "learning_rate": 3.692124913507452e-06, "loss": 1.3095, "step": 29030 }, { "epoch": 72.5775, "grad_norm": 6.46875, "learning_rate": 3.6914966515912544e-06, "loss": 1.5347, "step": 29031 }, { "epoch": 72.58, "grad_norm": 10.625, "learning_rate": 3.690868431033352e-06, "loss": 1.313, "step": 29032 }, { "epoch": 72.5825, "grad_norm": 6.125, "learning_rate": 3.69024025183786e-06, "loss": 1.3558, "step": 29033 }, { "epoch": 72.585, "grad_norm": 6.25, "learning_rate": 3.6896121140089014e-06, "loss": 1.4688, "step": 29034 }, { "epoch": 72.5875, "grad_norm": 5.6875, "learning_rate": 3.6889840175505952e-06, "loss": 1.4115, "step": 29035 }, { "epoch": 72.59, "grad_norm": 9.875, "learning_rate": 3.6883559624670585e-06, "loss": 1.4015, "step": 29036 }, { "epoch": 72.5925, "grad_norm": 7.09375, "learning_rate": 3.687727948762407e-06, "loss": 1.344, "step": 29037 }, { "epoch": 72.595, "grad_norm": 7.59375, "learning_rate": 3.687099976440751e-06, "loss": 1.3812, "step": 29038 }, { "epoch": 72.5975, "grad_norm": 6.8125, "learning_rate": 3.6864720455062232e-06, "loss": 1.3335, "step": 29039 }, { "epoch": 72.6, "grad_norm": 8.25, "learning_rate": 3.685844155962931e-06, "loss": 1.2811, "step": 29040 }, { "epoch": 72.6025, "grad_norm": 7.53125, "learning_rate": 3.6852163078149883e-06, "loss": 1.3583, "step": 29041 }, { "epoch": 72.605, "grad_norm": 6.21875, "learning_rate": 3.6845885010665196e-06, "loss": 1.4428, "step": 29042 }, { "epoch": 72.6075, "grad_norm": 7.875, "learning_rate": 3.683960735721631e-06, "loss": 1.2981, "step": 29043 }, { "epoch": 72.61, "grad_norm": 6.125, "learning_rate": 3.683333011784447e-06, "loss": 1.2019, "step": 29044 }, { "epoch": 72.6125, "grad_norm": 7.0625, "learning_rate": 3.6827053292590753e-06, "loss": 1.3943, "step": 29045 }, { "epoch": 72.615, "grad_norm": 6.5625, "learning_rate": 3.682077688149638e-06, "loss": 1.313, "step": 29046 }, { "epoch": 72.6175, "grad_norm": 7.0, "learning_rate": 3.6814500884602477e-06, "loss": 1.3355, "step": 29047 }, { "epoch": 72.62, "grad_norm": 7.25, "learning_rate": 3.680822530195013e-06, "loss": 1.1684, "step": 29048 }, { "epoch": 72.6225, "grad_norm": 6.90625, "learning_rate": 3.6801950133580535e-06, "loss": 1.3058, "step": 29049 }, { "epoch": 72.625, "grad_norm": 8.75, "learning_rate": 3.6795675379534857e-06, "loss": 1.3968, "step": 29050 }, { "epoch": 72.6275, "grad_norm": 7.03125, "learning_rate": 3.678940103985421e-06, "loss": 1.348, "step": 29051 }, { "epoch": 72.63, "grad_norm": 8.875, "learning_rate": 3.678312711457971e-06, "loss": 1.1417, "step": 29052 }, { "epoch": 72.6325, "grad_norm": 7.84375, "learning_rate": 3.677685360375244e-06, "loss": 1.3577, "step": 29053 }, { "epoch": 72.635, "grad_norm": 8.4375, "learning_rate": 3.6770580507413643e-06, "loss": 1.2079, "step": 29054 }, { "epoch": 72.6375, "grad_norm": 7.3125, "learning_rate": 3.6764307825604408e-06, "loss": 1.3727, "step": 29055 }, { "epoch": 72.64, "grad_norm": 10.5, "learning_rate": 3.675803555836582e-06, "loss": 1.4516, "step": 29056 }, { "epoch": 72.6425, "grad_norm": 9.4375, "learning_rate": 3.6751763705738996e-06, "loss": 1.4402, "step": 29057 }, { "epoch": 72.645, "grad_norm": 5.5, "learning_rate": 3.6745492267765083e-06, "loss": 1.3526, "step": 29058 }, { "epoch": 72.6475, "grad_norm": 9.625, "learning_rate": 3.6739221244485225e-06, "loss": 1.3638, "step": 29059 }, { "epoch": 72.65, "grad_norm": 6.1875, "learning_rate": 3.673295063594049e-06, "loss": 1.2549, "step": 29060 }, { "epoch": 72.6525, "grad_norm": 7.75, "learning_rate": 3.672668044217198e-06, "loss": 1.3073, "step": 29061 }, { "epoch": 72.655, "grad_norm": 5.625, "learning_rate": 3.672041066322084e-06, "loss": 1.2034, "step": 29062 }, { "epoch": 72.6575, "grad_norm": 7.71875, "learning_rate": 3.6714141299128127e-06, "loss": 1.39, "step": 29063 }, { "epoch": 72.66, "grad_norm": 7.0, "learning_rate": 3.6707872349935016e-06, "loss": 1.5618, "step": 29064 }, { "epoch": 72.6625, "grad_norm": 6.125, "learning_rate": 3.670160381568251e-06, "loss": 1.3643, "step": 29065 }, { "epoch": 72.665, "grad_norm": 9.3125, "learning_rate": 3.6695335696411793e-06, "loss": 1.2636, "step": 29066 }, { "epoch": 72.6675, "grad_norm": 5.875, "learning_rate": 3.668906799216393e-06, "loss": 1.1393, "step": 29067 }, { "epoch": 72.67, "grad_norm": 6.6875, "learning_rate": 3.668280070297994e-06, "loss": 1.2243, "step": 29068 }, { "epoch": 72.6725, "grad_norm": 7.28125, "learning_rate": 3.667653382890103e-06, "loss": 1.3238, "step": 29069 }, { "epoch": 72.675, "grad_norm": 6.0, "learning_rate": 3.6670267369968237e-06, "loss": 1.1919, "step": 29070 }, { "epoch": 72.6775, "grad_norm": 7.03125, "learning_rate": 3.6664001326222632e-06, "loss": 1.2811, "step": 29071 }, { "epoch": 72.68, "grad_norm": 8.8125, "learning_rate": 3.6657735697705267e-06, "loss": 1.2639, "step": 29072 }, { "epoch": 72.6825, "grad_norm": 10.0625, "learning_rate": 3.665147048445725e-06, "loss": 1.3714, "step": 29073 }, { "epoch": 72.685, "grad_norm": 9.8125, "learning_rate": 3.6645205686519705e-06, "loss": 1.4425, "step": 29074 }, { "epoch": 72.6875, "grad_norm": 8.75, "learning_rate": 3.663894130393364e-06, "loss": 1.3203, "step": 29075 }, { "epoch": 72.69, "grad_norm": 7.53125, "learning_rate": 3.6632677336740107e-06, "loss": 1.4151, "step": 29076 }, { "epoch": 72.6925, "grad_norm": 6.875, "learning_rate": 3.662641378498025e-06, "loss": 1.2673, "step": 29077 }, { "epoch": 72.695, "grad_norm": 5.53125, "learning_rate": 3.662015064869505e-06, "loss": 1.2484, "step": 29078 }, { "epoch": 72.6975, "grad_norm": 7.71875, "learning_rate": 3.661388792792564e-06, "loss": 1.4484, "step": 29079 }, { "epoch": 72.7, "grad_norm": 7.5, "learning_rate": 3.6607625622713005e-06, "loss": 1.1613, "step": 29080 }, { "epoch": 72.7025, "grad_norm": 8.125, "learning_rate": 3.660136373309828e-06, "loss": 1.348, "step": 29081 }, { "epoch": 72.705, "grad_norm": 6.8125, "learning_rate": 3.6595102259122473e-06, "loss": 1.3026, "step": 29082 }, { "epoch": 72.7075, "grad_norm": 6.84375, "learning_rate": 3.658884120082661e-06, "loss": 1.3136, "step": 29083 }, { "epoch": 72.71, "grad_norm": 6.34375, "learning_rate": 3.6582580558251767e-06, "loss": 1.3553, "step": 29084 }, { "epoch": 72.7125, "grad_norm": 5.875, "learning_rate": 3.657632033143902e-06, "loss": 1.1707, "step": 29085 }, { "epoch": 72.715, "grad_norm": 5.96875, "learning_rate": 3.657006052042937e-06, "loss": 1.2497, "step": 29086 }, { "epoch": 72.7175, "grad_norm": 6.84375, "learning_rate": 3.656380112526384e-06, "loss": 1.2843, "step": 29087 }, { "epoch": 72.72, "grad_norm": 4.90625, "learning_rate": 3.6557542145983495e-06, "loss": 1.3329, "step": 29088 }, { "epoch": 72.7225, "grad_norm": 9.4375, "learning_rate": 3.655128358262938e-06, "loss": 1.3369, "step": 29089 }, { "epoch": 72.725, "grad_norm": 7.90625, "learning_rate": 3.654502543524253e-06, "loss": 1.2244, "step": 29090 }, { "epoch": 72.7275, "grad_norm": 6.34375, "learning_rate": 3.6538767703863907e-06, "loss": 1.2621, "step": 29091 }, { "epoch": 72.73, "grad_norm": 9.375, "learning_rate": 3.6532510388534626e-06, "loss": 1.3749, "step": 29092 }, { "epoch": 72.7325, "grad_norm": 7.1875, "learning_rate": 3.6526253489295636e-06, "loss": 1.3681, "step": 29093 }, { "epoch": 72.735, "grad_norm": 7.46875, "learning_rate": 3.6519997006188013e-06, "loss": 1.2697, "step": 29094 }, { "epoch": 72.7375, "grad_norm": 7.625, "learning_rate": 3.651374093925272e-06, "loss": 1.3466, "step": 29095 }, { "epoch": 72.74, "grad_norm": 8.375, "learning_rate": 3.650748528853084e-06, "loss": 1.2833, "step": 29096 }, { "epoch": 72.7425, "grad_norm": 8.8125, "learning_rate": 3.6501230054063343e-06, "loss": 1.172, "step": 29097 }, { "epoch": 72.745, "grad_norm": 8.3125, "learning_rate": 3.6494975235891194e-06, "loss": 1.4498, "step": 29098 }, { "epoch": 72.7475, "grad_norm": 7.84375, "learning_rate": 3.648872083405546e-06, "loss": 1.3274, "step": 29099 }, { "epoch": 72.75, "grad_norm": 7.5625, "learning_rate": 3.6482466848597164e-06, "loss": 1.3532, "step": 29100 }, { "epoch": 72.7525, "grad_norm": 4.8125, "learning_rate": 3.647621327955726e-06, "loss": 1.2498, "step": 29101 }, { "epoch": 72.755, "grad_norm": 6.34375, "learning_rate": 3.6469960126976743e-06, "loss": 1.3976, "step": 29102 }, { "epoch": 72.7575, "grad_norm": 6.40625, "learning_rate": 3.646370739089662e-06, "loss": 1.3781, "step": 29103 }, { "epoch": 72.76, "grad_norm": 7.34375, "learning_rate": 3.6457455071357918e-06, "loss": 1.4076, "step": 29104 }, { "epoch": 72.7625, "grad_norm": 8.875, "learning_rate": 3.6451203168401593e-06, "loss": 1.2122, "step": 29105 }, { "epoch": 72.765, "grad_norm": 7.1875, "learning_rate": 3.644495168206862e-06, "loss": 1.2433, "step": 29106 }, { "epoch": 72.7675, "grad_norm": 5.28125, "learning_rate": 3.6438700612400014e-06, "loss": 1.2853, "step": 29107 }, { "epoch": 72.77, "grad_norm": 5.25, "learning_rate": 3.643244995943672e-06, "loss": 1.1148, "step": 29108 }, { "epoch": 72.7725, "grad_norm": 10.375, "learning_rate": 3.642619972321978e-06, "loss": 1.1924, "step": 29109 }, { "epoch": 72.775, "grad_norm": 8.6875, "learning_rate": 3.6419949903790086e-06, "loss": 1.2179, "step": 29110 }, { "epoch": 72.7775, "grad_norm": 5.90625, "learning_rate": 3.6413700501188696e-06, "loss": 1.3559, "step": 29111 }, { "epoch": 72.78, "grad_norm": 5.6875, "learning_rate": 3.6407451515456537e-06, "loss": 1.2387, "step": 29112 }, { "epoch": 72.7825, "grad_norm": 4.5625, "learning_rate": 3.640120294663455e-06, "loss": 1.2966, "step": 29113 }, { "epoch": 72.785, "grad_norm": 8.5, "learning_rate": 3.6394954794763737e-06, "loss": 1.1673, "step": 29114 }, { "epoch": 72.7875, "grad_norm": 7.65625, "learning_rate": 3.6388707059885077e-06, "loss": 1.3103, "step": 29115 }, { "epoch": 72.79, "grad_norm": 6.4375, "learning_rate": 3.638245974203951e-06, "loss": 1.2759, "step": 29116 }, { "epoch": 72.7925, "grad_norm": 12.125, "learning_rate": 3.6376212841267964e-06, "loss": 1.3382, "step": 29117 }, { "epoch": 72.795, "grad_norm": 7.65625, "learning_rate": 3.636996635761142e-06, "loss": 1.3684, "step": 29118 }, { "epoch": 72.7975, "grad_norm": 8.875, "learning_rate": 3.636372029111086e-06, "loss": 1.4223, "step": 29119 }, { "epoch": 72.8, "grad_norm": 9.1875, "learning_rate": 3.63574746418072e-06, "loss": 1.4002, "step": 29120 }, { "epoch": 72.8025, "grad_norm": 8.5625, "learning_rate": 3.635122940974136e-06, "loss": 1.3638, "step": 29121 }, { "epoch": 72.805, "grad_norm": 11.625, "learning_rate": 3.6344984594954356e-06, "loss": 1.3471, "step": 29122 }, { "epoch": 72.8075, "grad_norm": 8.25, "learning_rate": 3.6338740197487042e-06, "loss": 1.4088, "step": 29123 }, { "epoch": 72.81, "grad_norm": 5.5625, "learning_rate": 3.6332496217380432e-06, "loss": 1.2038, "step": 29124 }, { "epoch": 72.8125, "grad_norm": 11.0625, "learning_rate": 3.63262526546754e-06, "loss": 1.3257, "step": 29125 }, { "epoch": 72.815, "grad_norm": 8.3125, "learning_rate": 3.6320009509412933e-06, "loss": 1.433, "step": 29126 }, { "epoch": 72.8175, "grad_norm": 10.8125, "learning_rate": 3.6313766781633943e-06, "loss": 1.364, "step": 29127 }, { "epoch": 72.82, "grad_norm": 8.0625, "learning_rate": 3.6307524471379317e-06, "loss": 1.2884, "step": 29128 }, { "epoch": 72.8225, "grad_norm": 12.25, "learning_rate": 3.6301282578690013e-06, "loss": 1.3166, "step": 29129 }, { "epoch": 72.825, "grad_norm": 8.8125, "learning_rate": 3.6295041103606976e-06, "loss": 1.2636, "step": 29130 }, { "epoch": 72.8275, "grad_norm": 10.9375, "learning_rate": 3.62888000461711e-06, "loss": 1.2823, "step": 29131 }, { "epoch": 72.83, "grad_norm": 6.6875, "learning_rate": 3.628255940642328e-06, "loss": 1.2211, "step": 29132 }, { "epoch": 72.8325, "grad_norm": 7.6875, "learning_rate": 3.627631918440444e-06, "loss": 1.233, "step": 29133 }, { "epoch": 72.835, "grad_norm": 7.65625, "learning_rate": 3.627007938015553e-06, "loss": 1.3624, "step": 29134 }, { "epoch": 72.8375, "grad_norm": 7.84375, "learning_rate": 3.6263839993717443e-06, "loss": 1.3682, "step": 29135 }, { "epoch": 72.84, "grad_norm": 6.40625, "learning_rate": 3.625760102513103e-06, "loss": 1.1948, "step": 29136 }, { "epoch": 72.8425, "grad_norm": 9.1875, "learning_rate": 3.625136247443727e-06, "loss": 1.2953, "step": 29137 }, { "epoch": 72.845, "grad_norm": 10.125, "learning_rate": 3.6245124341676996e-06, "loss": 1.3636, "step": 29138 }, { "epoch": 72.8475, "grad_norm": 6.6875, "learning_rate": 3.623888662689116e-06, "loss": 1.3653, "step": 29139 }, { "epoch": 72.85, "grad_norm": 10.0, "learning_rate": 3.6232649330120608e-06, "loss": 1.2834, "step": 29140 }, { "epoch": 72.8525, "grad_norm": 6.09375, "learning_rate": 3.622641245140628e-06, "loss": 1.3013, "step": 29141 }, { "epoch": 72.855, "grad_norm": 6.5625, "learning_rate": 3.6220175990789043e-06, "loss": 1.3419, "step": 29142 }, { "epoch": 72.8575, "grad_norm": 6.34375, "learning_rate": 3.621393994830975e-06, "loss": 1.2864, "step": 29143 }, { "epoch": 72.86, "grad_norm": 14.5625, "learning_rate": 3.6207704324009308e-06, "loss": 1.2957, "step": 29144 }, { "epoch": 72.8625, "grad_norm": 6.5625, "learning_rate": 3.6201469117928647e-06, "loss": 1.3147, "step": 29145 }, { "epoch": 72.865, "grad_norm": 13.0625, "learning_rate": 3.6195234330108586e-06, "loss": 1.2158, "step": 29146 }, { "epoch": 72.8675, "grad_norm": 7.5, "learning_rate": 3.6188999960589986e-06, "loss": 1.2714, "step": 29147 }, { "epoch": 72.87, "grad_norm": 8.25, "learning_rate": 3.6182766009413763e-06, "loss": 1.2923, "step": 29148 }, { "epoch": 72.8725, "grad_norm": 8.9375, "learning_rate": 3.6176532476620795e-06, "loss": 1.3278, "step": 29149 }, { "epoch": 72.875, "grad_norm": 8.375, "learning_rate": 3.6170299362251926e-06, "loss": 1.3226, "step": 29150 }, { "epoch": 72.8775, "grad_norm": 9.0625, "learning_rate": 3.6164066666347987e-06, "loss": 1.2488, "step": 29151 }, { "epoch": 72.88, "grad_norm": 7.28125, "learning_rate": 3.6157834388949907e-06, "loss": 1.182, "step": 29152 }, { "epoch": 72.8825, "grad_norm": 15.375, "learning_rate": 3.6151602530098474e-06, "loss": 1.3085, "step": 29153 }, { "epoch": 72.885, "grad_norm": 8.125, "learning_rate": 3.6145371089834614e-06, "loss": 1.2433, "step": 29154 }, { "epoch": 72.8875, "grad_norm": 10.625, "learning_rate": 3.613914006819911e-06, "loss": 1.3833, "step": 29155 }, { "epoch": 72.89, "grad_norm": 9.875, "learning_rate": 3.613290946523289e-06, "loss": 1.3052, "step": 29156 }, { "epoch": 72.8925, "grad_norm": 6.71875, "learning_rate": 3.612667928097675e-06, "loss": 1.2335, "step": 29157 }, { "epoch": 72.895, "grad_norm": 8.4375, "learning_rate": 3.612044951547151e-06, "loss": 1.349, "step": 29158 }, { "epoch": 72.8975, "grad_norm": 6.15625, "learning_rate": 3.6114220168758063e-06, "loss": 1.465, "step": 29159 }, { "epoch": 72.9, "grad_norm": 13.875, "learning_rate": 3.610799124087725e-06, "loss": 1.341, "step": 29160 }, { "epoch": 72.9025, "grad_norm": 11.1875, "learning_rate": 3.6101762731869893e-06, "loss": 1.333, "step": 29161 }, { "epoch": 72.905, "grad_norm": 7.03125, "learning_rate": 3.60955346417768e-06, "loss": 1.2714, "step": 29162 }, { "epoch": 72.9075, "grad_norm": 9.625, "learning_rate": 3.6089306970638804e-06, "loss": 1.3465, "step": 29163 }, { "epoch": 72.91, "grad_norm": 6.15625, "learning_rate": 3.6083079718496804e-06, "loss": 1.2784, "step": 29164 }, { "epoch": 72.9125, "grad_norm": 7.71875, "learning_rate": 3.6076852885391566e-06, "loss": 1.4149, "step": 29165 }, { "epoch": 72.915, "grad_norm": 11.1875, "learning_rate": 3.607062647136389e-06, "loss": 1.4446, "step": 29166 }, { "epoch": 72.9175, "grad_norm": 5.15625, "learning_rate": 3.6064400476454675e-06, "loss": 1.142, "step": 29167 }, { "epoch": 72.92, "grad_norm": 9.3125, "learning_rate": 3.6058174900704646e-06, "loss": 1.5847, "step": 29168 }, { "epoch": 72.9225, "grad_norm": 8.75, "learning_rate": 3.60519497441547e-06, "loss": 1.3774, "step": 29169 }, { "epoch": 72.925, "grad_norm": 9.625, "learning_rate": 3.6045725006845576e-06, "loss": 1.2744, "step": 29170 }, { "epoch": 72.9275, "grad_norm": 9.0, "learning_rate": 3.6039500688818154e-06, "loss": 1.2638, "step": 29171 }, { "epoch": 72.93, "grad_norm": 9.8125, "learning_rate": 3.60332767901132e-06, "loss": 1.3318, "step": 29172 }, { "epoch": 72.9325, "grad_norm": 8.9375, "learning_rate": 3.602705331077151e-06, "loss": 1.4599, "step": 29173 }, { "epoch": 72.935, "grad_norm": 9.9375, "learning_rate": 3.602083025083388e-06, "loss": 1.3392, "step": 29174 }, { "epoch": 72.9375, "grad_norm": 6.5625, "learning_rate": 3.601460761034117e-06, "loss": 1.4051, "step": 29175 }, { "epoch": 72.94, "grad_norm": 5.4375, "learning_rate": 3.6008385389334122e-06, "loss": 1.2996, "step": 29176 }, { "epoch": 72.9425, "grad_norm": 5.4375, "learning_rate": 3.6002163587853547e-06, "loss": 1.4848, "step": 29177 }, { "epoch": 72.945, "grad_norm": 11.4375, "learning_rate": 3.599594220594016e-06, "loss": 1.2845, "step": 29178 }, { "epoch": 72.9475, "grad_norm": 8.625, "learning_rate": 3.598972124363488e-06, "loss": 1.4798, "step": 29179 }, { "epoch": 72.95, "grad_norm": 5.375, "learning_rate": 3.5983500700978425e-06, "loss": 1.3506, "step": 29180 }, { "epoch": 72.9525, "grad_norm": 9.5, "learning_rate": 3.597728057801154e-06, "loss": 1.3215, "step": 29181 }, { "epoch": 72.955, "grad_norm": 8.5, "learning_rate": 3.5971060874775076e-06, "loss": 1.3226, "step": 29182 }, { "epoch": 72.9575, "grad_norm": 7.9375, "learning_rate": 3.596484159130974e-06, "loss": 1.4323, "step": 29183 }, { "epoch": 72.96, "grad_norm": 8.25, "learning_rate": 3.595862272765638e-06, "loss": 1.4303, "step": 29184 }, { "epoch": 72.9625, "grad_norm": 7.0625, "learning_rate": 3.595240428385569e-06, "loss": 1.3103, "step": 29185 }, { "epoch": 72.965, "grad_norm": 8.5, "learning_rate": 3.594618625994851e-06, "loss": 1.3622, "step": 29186 }, { "epoch": 72.9675, "grad_norm": 8.625, "learning_rate": 3.5939968655975565e-06, "loss": 1.2782, "step": 29187 }, { "epoch": 72.97, "grad_norm": 5.15625, "learning_rate": 3.5933751471977597e-06, "loss": 1.2487, "step": 29188 }, { "epoch": 72.9725, "grad_norm": 7.9375, "learning_rate": 3.5927534707995382e-06, "loss": 1.3977, "step": 29189 }, { "epoch": 72.975, "grad_norm": 8.3125, "learning_rate": 3.5921318364069724e-06, "loss": 1.3534, "step": 29190 }, { "epoch": 72.9775, "grad_norm": 11.0625, "learning_rate": 3.5915102440241334e-06, "loss": 1.3438, "step": 29191 }, { "epoch": 72.98, "grad_norm": 8.3125, "learning_rate": 3.5908886936550967e-06, "loss": 1.2194, "step": 29192 }, { "epoch": 72.9825, "grad_norm": 7.96875, "learning_rate": 3.5902671853039307e-06, "loss": 1.4126, "step": 29193 }, { "epoch": 72.985, "grad_norm": 6.03125, "learning_rate": 3.5896457189747235e-06, "loss": 1.3211, "step": 29194 }, { "epoch": 72.9875, "grad_norm": 7.8125, "learning_rate": 3.5890242946715426e-06, "loss": 1.3095, "step": 29195 }, { "epoch": 72.99, "grad_norm": 8.625, "learning_rate": 3.588402912398462e-06, "loss": 1.3034, "step": 29196 }, { "epoch": 72.9925, "grad_norm": 5.34375, "learning_rate": 3.587781572159551e-06, "loss": 1.2208, "step": 29197 }, { "epoch": 72.995, "grad_norm": 6.84375, "learning_rate": 3.5871602739588882e-06, "loss": 1.3379, "step": 29198 }, { "epoch": 72.9975, "grad_norm": 5.90625, "learning_rate": 3.586539017800549e-06, "loss": 1.2444, "step": 29199 }, { "epoch": 73.0, "grad_norm": 4.09375, "learning_rate": 3.585917803688603e-06, "loss": 1.269, "step": 29200 }, { "epoch": 73.0025, "grad_norm": 7.25, "learning_rate": 3.585296631627121e-06, "loss": 1.3682, "step": 29201 }, { "epoch": 73.005, "grad_norm": 10.875, "learning_rate": 3.5846755016201807e-06, "loss": 1.2178, "step": 29202 }, { "epoch": 73.0075, "grad_norm": 6.28125, "learning_rate": 3.584054413671847e-06, "loss": 1.2299, "step": 29203 }, { "epoch": 73.01, "grad_norm": 8.0625, "learning_rate": 3.5834333677861998e-06, "loss": 1.443, "step": 29204 }, { "epoch": 73.0125, "grad_norm": 9.4375, "learning_rate": 3.582812363967303e-06, "loss": 1.3178, "step": 29205 }, { "epoch": 73.015, "grad_norm": 8.25, "learning_rate": 3.582191402219234e-06, "loss": 1.3498, "step": 29206 }, { "epoch": 73.0175, "grad_norm": 8.5, "learning_rate": 3.5815704825460627e-06, "loss": 1.323, "step": 29207 }, { "epoch": 73.02, "grad_norm": 9.5625, "learning_rate": 3.580949604951851e-06, "loss": 1.3225, "step": 29208 }, { "epoch": 73.0225, "grad_norm": 5.65625, "learning_rate": 3.580328769440684e-06, "loss": 1.2443, "step": 29209 }, { "epoch": 73.025, "grad_norm": 8.75, "learning_rate": 3.5797079760166242e-06, "loss": 1.2517, "step": 29210 }, { "epoch": 73.0275, "grad_norm": 8.3125, "learning_rate": 3.5790872246837428e-06, "loss": 1.3341, "step": 29211 }, { "epoch": 73.03, "grad_norm": 8.6875, "learning_rate": 3.578466515446101e-06, "loss": 1.3791, "step": 29212 }, { "epoch": 73.0325, "grad_norm": 7.59375, "learning_rate": 3.5778458483077827e-06, "loss": 1.3727, "step": 29213 }, { "epoch": 73.035, "grad_norm": 7.46875, "learning_rate": 3.57722522327285e-06, "loss": 1.3735, "step": 29214 }, { "epoch": 73.0375, "grad_norm": 9.375, "learning_rate": 3.5766046403453723e-06, "loss": 1.3385, "step": 29215 }, { "epoch": 73.04, "grad_norm": 8.8125, "learning_rate": 3.5759840995294136e-06, "loss": 1.3764, "step": 29216 }, { "epoch": 73.0425, "grad_norm": 9.9375, "learning_rate": 3.575363600829046e-06, "loss": 1.3747, "step": 29217 }, { "epoch": 73.045, "grad_norm": 6.625, "learning_rate": 3.5747431442483417e-06, "loss": 1.1883, "step": 29218 }, { "epoch": 73.0475, "grad_norm": 4.53125, "learning_rate": 3.574122729791364e-06, "loss": 1.1841, "step": 29219 }, { "epoch": 73.05, "grad_norm": 5.53125, "learning_rate": 3.5735023574621765e-06, "loss": 1.2211, "step": 29220 }, { "epoch": 73.0525, "grad_norm": 6.34375, "learning_rate": 3.5728820272648535e-06, "loss": 1.2825, "step": 29221 }, { "epoch": 73.055, "grad_norm": 8.5, "learning_rate": 3.5722617392034564e-06, "loss": 1.4631, "step": 29222 }, { "epoch": 73.0575, "grad_norm": 6.59375, "learning_rate": 3.5716414932820563e-06, "loss": 1.3262, "step": 29223 }, { "epoch": 73.06, "grad_norm": 7.78125, "learning_rate": 3.5710212895047146e-06, "loss": 1.2692, "step": 29224 }, { "epoch": 73.0625, "grad_norm": 9.6875, "learning_rate": 3.5704011278755035e-06, "loss": 1.3326, "step": 29225 }, { "epoch": 73.065, "grad_norm": 5.5, "learning_rate": 3.569781008398485e-06, "loss": 1.2251, "step": 29226 }, { "epoch": 73.0675, "grad_norm": 6.03125, "learning_rate": 3.569160931077722e-06, "loss": 1.2181, "step": 29227 }, { "epoch": 73.07, "grad_norm": 6.3125, "learning_rate": 3.568540895917283e-06, "loss": 1.2892, "step": 29228 }, { "epoch": 73.0725, "grad_norm": 7.375, "learning_rate": 3.5679209029212347e-06, "loss": 1.2571, "step": 29229 }, { "epoch": 73.075, "grad_norm": 7.625, "learning_rate": 3.567300952093641e-06, "loss": 1.2554, "step": 29230 }, { "epoch": 73.0775, "grad_norm": 8.875, "learning_rate": 3.5666810434385602e-06, "loss": 1.3569, "step": 29231 }, { "epoch": 73.08, "grad_norm": 9.25, "learning_rate": 3.5660611769600604e-06, "loss": 1.3146, "step": 29232 }, { "epoch": 73.0825, "grad_norm": 11.5, "learning_rate": 3.565441352662211e-06, "loss": 1.2305, "step": 29233 }, { "epoch": 73.085, "grad_norm": 7.9375, "learning_rate": 3.5648215705490707e-06, "loss": 1.387, "step": 29234 }, { "epoch": 73.0875, "grad_norm": 7.71875, "learning_rate": 3.5642018306246984e-06, "loss": 1.3455, "step": 29235 }, { "epoch": 73.09, "grad_norm": 9.0, "learning_rate": 3.563582132893164e-06, "loss": 1.3227, "step": 29236 }, { "epoch": 73.0925, "grad_norm": 12.125, "learning_rate": 3.562962477358526e-06, "loss": 1.3108, "step": 29237 }, { "epoch": 73.095, "grad_norm": 7.75, "learning_rate": 3.5623428640248505e-06, "loss": 1.3508, "step": 29238 }, { "epoch": 73.0975, "grad_norm": 9.5, "learning_rate": 3.5617232928961942e-06, "loss": 1.2774, "step": 29239 }, { "epoch": 73.1, "grad_norm": 8.375, "learning_rate": 3.5611037639766267e-06, "loss": 1.3438, "step": 29240 }, { "epoch": 73.1025, "grad_norm": 6.71875, "learning_rate": 3.560484277270204e-06, "loss": 1.2684, "step": 29241 }, { "epoch": 73.105, "grad_norm": 6.5625, "learning_rate": 3.559864832780987e-06, "loss": 1.1942, "step": 29242 }, { "epoch": 73.1075, "grad_norm": 6.28125, "learning_rate": 3.5592454305130377e-06, "loss": 1.3052, "step": 29243 }, { "epoch": 73.11, "grad_norm": 7.28125, "learning_rate": 3.55862607047042e-06, "loss": 1.3409, "step": 29244 }, { "epoch": 73.1125, "grad_norm": 7.90625, "learning_rate": 3.5580067526571925e-06, "loss": 1.2627, "step": 29245 }, { "epoch": 73.115, "grad_norm": 10.25, "learning_rate": 3.557387477077412e-06, "loss": 1.515, "step": 29246 }, { "epoch": 73.1175, "grad_norm": 36.0, "learning_rate": 3.5567682437351412e-06, "loss": 1.3817, "step": 29247 }, { "epoch": 73.12, "grad_norm": 8.5, "learning_rate": 3.556149052634443e-06, "loss": 1.2324, "step": 29248 }, { "epoch": 73.1225, "grad_norm": 6.34375, "learning_rate": 3.5555299037793743e-06, "loss": 1.3078, "step": 29249 }, { "epoch": 73.125, "grad_norm": 10.25, "learning_rate": 3.5549107971739905e-06, "loss": 1.3167, "step": 29250 }, { "epoch": 73.1275, "grad_norm": 8.375, "learning_rate": 3.554291732822356e-06, "loss": 1.249, "step": 29251 }, { "epoch": 73.13, "grad_norm": 8.125, "learning_rate": 3.553672710728524e-06, "loss": 1.3101, "step": 29252 }, { "epoch": 73.1325, "grad_norm": 7.15625, "learning_rate": 3.5530537308965594e-06, "loss": 1.2603, "step": 29253 }, { "epoch": 73.135, "grad_norm": 8.6875, "learning_rate": 3.5524347933305126e-06, "loss": 1.5085, "step": 29254 }, { "epoch": 73.1375, "grad_norm": 7.1875, "learning_rate": 3.5518158980344496e-06, "loss": 1.4934, "step": 29255 }, { "epoch": 73.14, "grad_norm": 6.90625, "learning_rate": 3.5511970450124223e-06, "loss": 1.3358, "step": 29256 }, { "epoch": 73.1425, "grad_norm": 9.0, "learning_rate": 3.550578234268487e-06, "loss": 1.2844, "step": 29257 }, { "epoch": 73.145, "grad_norm": 6.0625, "learning_rate": 3.5499594658067016e-06, "loss": 1.317, "step": 29258 }, { "epoch": 73.1475, "grad_norm": 8.25, "learning_rate": 3.5493407396311273e-06, "loss": 1.3393, "step": 29259 }, { "epoch": 73.15, "grad_norm": 7.28125, "learning_rate": 3.548722055745818e-06, "loss": 1.3897, "step": 29260 }, { "epoch": 73.1525, "grad_norm": 9.75, "learning_rate": 3.5481034141548243e-06, "loss": 1.3222, "step": 29261 }, { "epoch": 73.155, "grad_norm": 6.3125, "learning_rate": 3.5474848148622064e-06, "loss": 1.2429, "step": 29262 }, { "epoch": 73.1575, "grad_norm": 6.9375, "learning_rate": 3.546866257872025e-06, "loss": 1.3373, "step": 29263 }, { "epoch": 73.16, "grad_norm": 6.5, "learning_rate": 3.546247743188328e-06, "loss": 1.2412, "step": 29264 }, { "epoch": 73.1625, "grad_norm": 6.625, "learning_rate": 3.5456292708151698e-06, "loss": 1.3245, "step": 29265 }, { "epoch": 73.165, "grad_norm": 6.90625, "learning_rate": 3.545010840756612e-06, "loss": 1.2686, "step": 29266 }, { "epoch": 73.1675, "grad_norm": 7.4375, "learning_rate": 3.5443924530167005e-06, "loss": 1.2502, "step": 29267 }, { "epoch": 73.17, "grad_norm": 6.25, "learning_rate": 3.543774107599497e-06, "loss": 1.3486, "step": 29268 }, { "epoch": 73.1725, "grad_norm": 6.9375, "learning_rate": 3.54315580450905e-06, "loss": 1.2674, "step": 29269 }, { "epoch": 73.175, "grad_norm": 6.875, "learning_rate": 3.542537543749417e-06, "loss": 1.2554, "step": 29270 }, { "epoch": 73.1775, "grad_norm": 6.96875, "learning_rate": 3.54191932532465e-06, "loss": 1.3954, "step": 29271 }, { "epoch": 73.18, "grad_norm": 6.375, "learning_rate": 3.541301149238798e-06, "loss": 1.4858, "step": 29272 }, { "epoch": 73.1825, "grad_norm": 7.46875, "learning_rate": 3.540683015495918e-06, "loss": 1.3516, "step": 29273 }, { "epoch": 73.185, "grad_norm": 7.65625, "learning_rate": 3.540064924100064e-06, "loss": 1.3543, "step": 29274 }, { "epoch": 73.1875, "grad_norm": 6.125, "learning_rate": 3.539446875055287e-06, "loss": 1.3597, "step": 29275 }, { "epoch": 73.19, "grad_norm": 5.96875, "learning_rate": 3.5388288683656336e-06, "loss": 1.3409, "step": 29276 }, { "epoch": 73.1925, "grad_norm": 8.5, "learning_rate": 3.53821090403516e-06, "loss": 1.2849, "step": 29277 }, { "epoch": 73.195, "grad_norm": 11.0, "learning_rate": 3.537592982067919e-06, "loss": 1.3718, "step": 29278 }, { "epoch": 73.1975, "grad_norm": 7.875, "learning_rate": 3.536975102467961e-06, "loss": 1.2987, "step": 29279 }, { "epoch": 73.2, "grad_norm": 5.5, "learning_rate": 3.536357265239333e-06, "loss": 1.3606, "step": 29280 }, { "epoch": 73.2025, "grad_norm": 6.375, "learning_rate": 3.5357394703860904e-06, "loss": 1.3364, "step": 29281 }, { "epoch": 73.205, "grad_norm": 7.9375, "learning_rate": 3.5351217179122775e-06, "loss": 1.2458, "step": 29282 }, { "epoch": 73.2075, "grad_norm": 7.46875, "learning_rate": 3.5345040078219515e-06, "loss": 1.4454, "step": 29283 }, { "epoch": 73.21, "grad_norm": 8.5, "learning_rate": 3.5338863401191557e-06, "loss": 1.444, "step": 29284 }, { "epoch": 73.2125, "grad_norm": 10.0, "learning_rate": 3.5332687148079457e-06, "loss": 1.295, "step": 29285 }, { "epoch": 73.215, "grad_norm": 7.34375, "learning_rate": 3.5326511318923674e-06, "loss": 1.2476, "step": 29286 }, { "epoch": 73.2175, "grad_norm": 8.0, "learning_rate": 3.532033591376465e-06, "loss": 1.2605, "step": 29287 }, { "epoch": 73.22, "grad_norm": 7.625, "learning_rate": 3.5314160932642925e-06, "loss": 1.2353, "step": 29288 }, { "epoch": 73.2225, "grad_norm": 7.625, "learning_rate": 3.5307986375599e-06, "loss": 1.3368, "step": 29289 }, { "epoch": 73.225, "grad_norm": 9.3125, "learning_rate": 3.5301812242673327e-06, "loss": 1.3092, "step": 29290 }, { "epoch": 73.2275, "grad_norm": 8.0625, "learning_rate": 3.529563853390635e-06, "loss": 1.367, "step": 29291 }, { "epoch": 73.23, "grad_norm": 7.9375, "learning_rate": 3.5289465249338585e-06, "loss": 1.3045, "step": 29292 }, { "epoch": 73.2325, "grad_norm": 8.5625, "learning_rate": 3.5283292389010526e-06, "loss": 1.3965, "step": 29293 }, { "epoch": 73.235, "grad_norm": 5.34375, "learning_rate": 3.5277119952962613e-06, "loss": 1.2836, "step": 29294 }, { "epoch": 73.2375, "grad_norm": 8.125, "learning_rate": 3.527094794123528e-06, "loss": 1.5668, "step": 29295 }, { "epoch": 73.24, "grad_norm": 5.0625, "learning_rate": 3.5264776353869046e-06, "loss": 1.3966, "step": 29296 }, { "epoch": 73.2425, "grad_norm": 9.1875, "learning_rate": 3.5258605190904327e-06, "loss": 1.2016, "step": 29297 }, { "epoch": 73.245, "grad_norm": 6.5, "learning_rate": 3.5252434452381625e-06, "loss": 1.2262, "step": 29298 }, { "epoch": 73.2475, "grad_norm": 8.875, "learning_rate": 3.5246264138341336e-06, "loss": 1.3673, "step": 29299 }, { "epoch": 73.25, "grad_norm": 5.5625, "learning_rate": 3.5240094248824e-06, "loss": 1.3014, "step": 29300 }, { "epoch": 73.2525, "grad_norm": 8.125, "learning_rate": 3.523392478386999e-06, "loss": 1.3936, "step": 29301 }, { "epoch": 73.255, "grad_norm": 5.125, "learning_rate": 3.5227755743519753e-06, "loss": 1.2906, "step": 29302 }, { "epoch": 73.2575, "grad_norm": 6.96875, "learning_rate": 3.522158712781376e-06, "loss": 1.351, "step": 29303 }, { "epoch": 73.26, "grad_norm": 5.1875, "learning_rate": 3.5215418936792487e-06, "loss": 1.3421, "step": 29304 }, { "epoch": 73.2625, "grad_norm": 7.8125, "learning_rate": 3.5209251170496327e-06, "loss": 1.2669, "step": 29305 }, { "epoch": 73.265, "grad_norm": 8.1875, "learning_rate": 3.52030838289657e-06, "loss": 1.2948, "step": 29306 }, { "epoch": 73.2675, "grad_norm": 10.75, "learning_rate": 3.5196916912241054e-06, "loss": 1.389, "step": 29307 }, { "epoch": 73.27, "grad_norm": 6.21875, "learning_rate": 3.5190750420362863e-06, "loss": 1.2459, "step": 29308 }, { "epoch": 73.2725, "grad_norm": 4.875, "learning_rate": 3.518458435337152e-06, "loss": 1.2812, "step": 29309 }, { "epoch": 73.275, "grad_norm": 10.5625, "learning_rate": 3.5178418711307416e-06, "loss": 1.3143, "step": 29310 }, { "epoch": 73.2775, "grad_norm": 7.71875, "learning_rate": 3.517225349421104e-06, "loss": 1.2251, "step": 29311 }, { "epoch": 73.28, "grad_norm": 8.5, "learning_rate": 3.5166088702122738e-06, "loss": 1.1626, "step": 29312 }, { "epoch": 73.2825, "grad_norm": 7.75, "learning_rate": 3.5159924335082997e-06, "loss": 1.1876, "step": 29313 }, { "epoch": 73.285, "grad_norm": 8.4375, "learning_rate": 3.5153760393132165e-06, "loss": 1.5002, "step": 29314 }, { "epoch": 73.2875, "grad_norm": 7.34375, "learning_rate": 3.5147596876310707e-06, "loss": 1.2934, "step": 29315 }, { "epoch": 73.29, "grad_norm": 7.75, "learning_rate": 3.514143378465903e-06, "loss": 1.3801, "step": 29316 }, { "epoch": 73.2925, "grad_norm": 7.15625, "learning_rate": 3.5135271118217474e-06, "loss": 1.4042, "step": 29317 }, { "epoch": 73.295, "grad_norm": 6.1875, "learning_rate": 3.5129108877026474e-06, "loss": 1.2361, "step": 29318 }, { "epoch": 73.2975, "grad_norm": 9.4375, "learning_rate": 3.512294706112649e-06, "loss": 1.2726, "step": 29319 }, { "epoch": 73.3, "grad_norm": 7.25, "learning_rate": 3.511678567055786e-06, "loss": 1.2882, "step": 29320 }, { "epoch": 73.3025, "grad_norm": 5.6875, "learning_rate": 3.5110624705360962e-06, "loss": 1.1266, "step": 29321 }, { "epoch": 73.305, "grad_norm": 11.1875, "learning_rate": 3.51044641655762e-06, "loss": 1.3245, "step": 29322 }, { "epoch": 73.3075, "grad_norm": 8.125, "learning_rate": 3.5098304051244015e-06, "loss": 1.3, "step": 29323 }, { "epoch": 73.31, "grad_norm": 5.78125, "learning_rate": 3.5092144362404757e-06, "loss": 1.1085, "step": 29324 }, { "epoch": 73.3125, "grad_norm": 7.5625, "learning_rate": 3.5085985099098753e-06, "loss": 1.272, "step": 29325 }, { "epoch": 73.315, "grad_norm": 9.4375, "learning_rate": 3.507982626136648e-06, "loss": 1.2384, "step": 29326 }, { "epoch": 73.3175, "grad_norm": 9.25, "learning_rate": 3.5073667849248226e-06, "loss": 1.2223, "step": 29327 }, { "epoch": 73.32, "grad_norm": 10.125, "learning_rate": 3.5067509862784455e-06, "loss": 1.2645, "step": 29328 }, { "epoch": 73.3225, "grad_norm": 10.8125, "learning_rate": 3.5061352302015437e-06, "loss": 1.4837, "step": 29329 }, { "epoch": 73.325, "grad_norm": 7.75, "learning_rate": 3.505519516698165e-06, "loss": 1.4467, "step": 29330 }, { "epoch": 73.3275, "grad_norm": 6.5625, "learning_rate": 3.504903845772338e-06, "loss": 1.3425, "step": 29331 }, { "epoch": 73.33, "grad_norm": 12.875, "learning_rate": 3.5042882174281e-06, "loss": 1.3621, "step": 29332 }, { "epoch": 73.3325, "grad_norm": 11.4375, "learning_rate": 3.503672631669488e-06, "loss": 1.19, "step": 29333 }, { "epoch": 73.335, "grad_norm": 7.03125, "learning_rate": 3.503057088500542e-06, "loss": 1.4229, "step": 29334 }, { "epoch": 73.3375, "grad_norm": 6.75, "learning_rate": 3.5024415879252928e-06, "loss": 1.2229, "step": 29335 }, { "epoch": 73.34, "grad_norm": 6.625, "learning_rate": 3.5018261299477772e-06, "loss": 1.2338, "step": 29336 }, { "epoch": 73.3425, "grad_norm": 7.03125, "learning_rate": 3.501210714572023e-06, "loss": 1.2074, "step": 29337 }, { "epoch": 73.345, "grad_norm": 6.78125, "learning_rate": 3.500595341802078e-06, "loss": 1.2307, "step": 29338 }, { "epoch": 73.3475, "grad_norm": 8.0625, "learning_rate": 3.499980011641969e-06, "loss": 1.4184, "step": 29339 }, { "epoch": 73.35, "grad_norm": 8.0625, "learning_rate": 3.4993647240957307e-06, "loss": 1.3564, "step": 29340 }, { "epoch": 73.3525, "grad_norm": 7.96875, "learning_rate": 3.498749479167395e-06, "loss": 1.1555, "step": 29341 }, { "epoch": 73.355, "grad_norm": 7.53125, "learning_rate": 3.498134276860997e-06, "loss": 1.4301, "step": 29342 }, { "epoch": 73.3575, "grad_norm": 6.28125, "learning_rate": 3.497519117180573e-06, "loss": 1.2716, "step": 29343 }, { "epoch": 73.36, "grad_norm": 9.3125, "learning_rate": 3.4969040001301513e-06, "loss": 1.2945, "step": 29344 }, { "epoch": 73.3625, "grad_norm": 5.5, "learning_rate": 3.49628892571377e-06, "loss": 1.2224, "step": 29345 }, { "epoch": 73.365, "grad_norm": 6.625, "learning_rate": 3.495673893935457e-06, "loss": 1.3313, "step": 29346 }, { "epoch": 73.3675, "grad_norm": 7.59375, "learning_rate": 3.4950589047992444e-06, "loss": 1.3885, "step": 29347 }, { "epoch": 73.37, "grad_norm": 8.4375, "learning_rate": 3.494443958309164e-06, "loss": 1.3008, "step": 29348 }, { "epoch": 73.3725, "grad_norm": 6.3125, "learning_rate": 3.493829054469252e-06, "loss": 1.262, "step": 29349 }, { "epoch": 73.375, "grad_norm": 5.46875, "learning_rate": 3.4932141932835362e-06, "loss": 1.3897, "step": 29350 }, { "epoch": 73.3775, "grad_norm": 7.34375, "learning_rate": 3.4925993747560484e-06, "loss": 1.4909, "step": 29351 }, { "epoch": 73.38, "grad_norm": 6.96875, "learning_rate": 3.491984598890812e-06, "loss": 1.3356, "step": 29352 }, { "epoch": 73.3825, "grad_norm": 9.0625, "learning_rate": 3.4913698656918705e-06, "loss": 1.2881, "step": 29353 }, { "epoch": 73.385, "grad_norm": 8.25, "learning_rate": 3.490755175163247e-06, "loss": 1.4024, "step": 29354 }, { "epoch": 73.3875, "grad_norm": 9.4375, "learning_rate": 3.4901405273089726e-06, "loss": 1.2303, "step": 29355 }, { "epoch": 73.39, "grad_norm": 6.21875, "learning_rate": 3.4895259221330736e-06, "loss": 1.1516, "step": 29356 }, { "epoch": 73.3925, "grad_norm": 8.375, "learning_rate": 3.488911359639582e-06, "loss": 1.2471, "step": 29357 }, { "epoch": 73.395, "grad_norm": 7.5625, "learning_rate": 3.4882968398325303e-06, "loss": 1.3824, "step": 29358 }, { "epoch": 73.3975, "grad_norm": 7.5625, "learning_rate": 3.487682362715944e-06, "loss": 1.3834, "step": 29359 }, { "epoch": 73.4, "grad_norm": 7.25, "learning_rate": 3.487067928293848e-06, "loss": 1.4078, "step": 29360 }, { "epoch": 73.4025, "grad_norm": 9.875, "learning_rate": 3.486453536570277e-06, "loss": 1.2377, "step": 29361 }, { "epoch": 73.405, "grad_norm": 7.875, "learning_rate": 3.4858391875492527e-06, "loss": 1.2569, "step": 29362 }, { "epoch": 73.4075, "grad_norm": 7.84375, "learning_rate": 3.48522488123481e-06, "loss": 1.3075, "step": 29363 }, { "epoch": 73.41, "grad_norm": 10.8125, "learning_rate": 3.484610617630968e-06, "loss": 1.3236, "step": 29364 }, { "epoch": 73.4125, "grad_norm": 8.6875, "learning_rate": 3.483996396741762e-06, "loss": 1.3087, "step": 29365 }, { "epoch": 73.415, "grad_norm": 8.0625, "learning_rate": 3.4833822185712154e-06, "loss": 1.3587, "step": 29366 }, { "epoch": 73.4175, "grad_norm": 7.875, "learning_rate": 3.482768083123349e-06, "loss": 1.277, "step": 29367 }, { "epoch": 73.42, "grad_norm": 6.40625, "learning_rate": 3.4821539904021963e-06, "loss": 1.4558, "step": 29368 }, { "epoch": 73.4225, "grad_norm": 6.5625, "learning_rate": 3.4815399404117835e-06, "loss": 1.2866, "step": 29369 }, { "epoch": 73.425, "grad_norm": 7.9375, "learning_rate": 3.480925933156134e-06, "loss": 1.2915, "step": 29370 }, { "epoch": 73.4275, "grad_norm": 7.9375, "learning_rate": 3.4803119686392695e-06, "loss": 1.4065, "step": 29371 }, { "epoch": 73.43, "grad_norm": 5.96875, "learning_rate": 3.47969804686522e-06, "loss": 1.3552, "step": 29372 }, { "epoch": 73.4325, "grad_norm": 6.0625, "learning_rate": 3.4790841678380118e-06, "loss": 1.3956, "step": 29373 }, { "epoch": 73.435, "grad_norm": 7.71875, "learning_rate": 3.4784703315616666e-06, "loss": 1.3123, "step": 29374 }, { "epoch": 73.4375, "grad_norm": 5.4375, "learning_rate": 3.4778565380402064e-06, "loss": 1.2391, "step": 29375 }, { "epoch": 73.44, "grad_norm": 8.1875, "learning_rate": 3.4772427872776606e-06, "loss": 1.3148, "step": 29376 }, { "epoch": 73.4425, "grad_norm": 6.34375, "learning_rate": 3.4766290792780466e-06, "loss": 1.386, "step": 29377 }, { "epoch": 73.445, "grad_norm": 5.65625, "learning_rate": 3.476015414045395e-06, "loss": 1.299, "step": 29378 }, { "epoch": 73.4475, "grad_norm": 6.5, "learning_rate": 3.4754017915837214e-06, "loss": 1.4881, "step": 29379 }, { "epoch": 73.45, "grad_norm": 8.5625, "learning_rate": 3.4747882118970565e-06, "loss": 1.4127, "step": 29380 }, { "epoch": 73.4525, "grad_norm": 7.84375, "learning_rate": 3.474174674989418e-06, "loss": 1.4895, "step": 29381 }, { "epoch": 73.455, "grad_norm": 9.0625, "learning_rate": 3.473561180864826e-06, "loss": 1.2737, "step": 29382 }, { "epoch": 73.4575, "grad_norm": 7.5, "learning_rate": 3.4729477295273073e-06, "loss": 1.1709, "step": 29383 }, { "epoch": 73.46, "grad_norm": 6.28125, "learning_rate": 3.4723343209808834e-06, "loss": 1.3417, "step": 29384 }, { "epoch": 73.4625, "grad_norm": 6.21875, "learning_rate": 3.471720955229576e-06, "loss": 1.2995, "step": 29385 }, { "epoch": 73.465, "grad_norm": 7.0, "learning_rate": 3.4711076322774006e-06, "loss": 1.4178, "step": 29386 }, { "epoch": 73.4675, "grad_norm": 7.5625, "learning_rate": 3.470494352128383e-06, "loss": 1.2856, "step": 29387 }, { "epoch": 73.47, "grad_norm": 6.28125, "learning_rate": 3.4698811147865454e-06, "loss": 1.3218, "step": 29388 }, { "epoch": 73.4725, "grad_norm": 6.46875, "learning_rate": 3.4692679202559065e-06, "loss": 1.318, "step": 29389 }, { "epoch": 73.475, "grad_norm": 10.125, "learning_rate": 3.4686547685404826e-06, "loss": 1.4357, "step": 29390 }, { "epoch": 73.4775, "grad_norm": 6.5, "learning_rate": 3.4680416596443e-06, "loss": 1.3662, "step": 29391 }, { "epoch": 73.48, "grad_norm": 8.125, "learning_rate": 3.4674285935713715e-06, "loss": 1.2008, "step": 29392 }, { "epoch": 73.4825, "grad_norm": 8.0625, "learning_rate": 3.4668155703257235e-06, "loss": 1.4141, "step": 29393 }, { "epoch": 73.485, "grad_norm": 10.6875, "learning_rate": 3.466202589911366e-06, "loss": 1.3965, "step": 29394 }, { "epoch": 73.4875, "grad_norm": 9.6875, "learning_rate": 3.4655896523323284e-06, "loss": 1.3924, "step": 29395 }, { "epoch": 73.49, "grad_norm": 6.875, "learning_rate": 3.4649767575926217e-06, "loss": 1.469, "step": 29396 }, { "epoch": 73.4925, "grad_norm": 8.75, "learning_rate": 3.464363905696264e-06, "loss": 1.2663, "step": 29397 }, { "epoch": 73.495, "grad_norm": 8.4375, "learning_rate": 3.4637510966472744e-06, "loss": 1.4266, "step": 29398 }, { "epoch": 73.4975, "grad_norm": 8.125, "learning_rate": 3.463138330449675e-06, "loss": 1.3428, "step": 29399 }, { "epoch": 73.5, "grad_norm": 7.03125, "learning_rate": 3.4625256071074776e-06, "loss": 1.5264, "step": 29400 }, { "epoch": 73.5025, "grad_norm": 5.5, "learning_rate": 3.4619129266246986e-06, "loss": 1.2483, "step": 29401 }, { "epoch": 73.505, "grad_norm": 5.46875, "learning_rate": 3.461300289005357e-06, "loss": 1.2647, "step": 29402 }, { "epoch": 73.5075, "grad_norm": 7.03125, "learning_rate": 3.460687694253472e-06, "loss": 1.5978, "step": 29403 }, { "epoch": 73.51, "grad_norm": 7.15625, "learning_rate": 3.460075142373056e-06, "loss": 1.3352, "step": 29404 }, { "epoch": 73.5125, "grad_norm": 8.4375, "learning_rate": 3.4594626333681234e-06, "loss": 1.3606, "step": 29405 }, { "epoch": 73.515, "grad_norm": 6.84375, "learning_rate": 3.4588501672426942e-06, "loss": 1.2697, "step": 29406 }, { "epoch": 73.5175, "grad_norm": 8.875, "learning_rate": 3.4582377440007796e-06, "loss": 1.4181, "step": 29407 }, { "epoch": 73.52, "grad_norm": 6.65625, "learning_rate": 3.4576253636463996e-06, "loss": 1.3241, "step": 29408 }, { "epoch": 73.5225, "grad_norm": 9.0625, "learning_rate": 3.4570130261835623e-06, "loss": 1.409, "step": 29409 }, { "epoch": 73.525, "grad_norm": 9.8125, "learning_rate": 3.4564007316162883e-06, "loss": 1.2844, "step": 29410 }, { "epoch": 73.5275, "grad_norm": 7.875, "learning_rate": 3.4557884799485895e-06, "loss": 1.3333, "step": 29411 }, { "epoch": 73.53, "grad_norm": 6.21875, "learning_rate": 3.455176271184477e-06, "loss": 1.3351, "step": 29412 }, { "epoch": 73.5325, "grad_norm": 11.125, "learning_rate": 3.4545641053279666e-06, "loss": 1.3887, "step": 29413 }, { "epoch": 73.535, "grad_norm": 6.46875, "learning_rate": 3.4539519823830757e-06, "loss": 1.1855, "step": 29414 }, { "epoch": 73.5375, "grad_norm": 7.40625, "learning_rate": 3.453339902353814e-06, "loss": 1.4178, "step": 29415 }, { "epoch": 73.54, "grad_norm": 7.78125, "learning_rate": 3.4527278652441896e-06, "loss": 1.2819, "step": 29416 }, { "epoch": 73.5425, "grad_norm": 6.3125, "learning_rate": 3.452115871058219e-06, "loss": 1.3385, "step": 29417 }, { "epoch": 73.545, "grad_norm": 8.5625, "learning_rate": 3.451503919799919e-06, "loss": 1.3433, "step": 29418 }, { "epoch": 73.5475, "grad_norm": 9.3125, "learning_rate": 3.450892011473297e-06, "loss": 1.3138, "step": 29419 }, { "epoch": 73.55, "grad_norm": 7.03125, "learning_rate": 3.450280146082361e-06, "loss": 1.3762, "step": 29420 }, { "epoch": 73.5525, "grad_norm": 5.90625, "learning_rate": 3.4496683236311303e-06, "loss": 1.2178, "step": 29421 }, { "epoch": 73.555, "grad_norm": 9.625, "learning_rate": 3.449056544123609e-06, "loss": 1.2414, "step": 29422 }, { "epoch": 73.5575, "grad_norm": 6.40625, "learning_rate": 3.4484448075638134e-06, "loss": 1.3447, "step": 29423 }, { "epoch": 73.56, "grad_norm": 6.5, "learning_rate": 3.4478331139557475e-06, "loss": 1.3244, "step": 29424 }, { "epoch": 73.5625, "grad_norm": 5.96875, "learning_rate": 3.4472214633034295e-06, "loss": 1.3541, "step": 29425 }, { "epoch": 73.565, "grad_norm": 7.375, "learning_rate": 3.4466098556108662e-06, "loss": 1.427, "step": 29426 }, { "epoch": 73.5675, "grad_norm": 9.875, "learning_rate": 3.445998290882062e-06, "loss": 1.3375, "step": 29427 }, { "epoch": 73.57, "grad_norm": 5.96875, "learning_rate": 3.4453867691210307e-06, "loss": 1.4177, "step": 29428 }, { "epoch": 73.5725, "grad_norm": 5.03125, "learning_rate": 3.4447752903317853e-06, "loss": 1.2995, "step": 29429 }, { "epoch": 73.575, "grad_norm": 7.53125, "learning_rate": 3.4441638545183308e-06, "loss": 1.3725, "step": 29430 }, { "epoch": 73.5775, "grad_norm": 13.6875, "learning_rate": 3.4435524616846726e-06, "loss": 1.4845, "step": 29431 }, { "epoch": 73.58, "grad_norm": 8.375, "learning_rate": 3.442941111834822e-06, "loss": 1.3527, "step": 29432 }, { "epoch": 73.5825, "grad_norm": 6.0625, "learning_rate": 3.442329804972789e-06, "loss": 1.3478, "step": 29433 }, { "epoch": 73.585, "grad_norm": 7.34375, "learning_rate": 3.441718541102581e-06, "loss": 1.1501, "step": 29434 }, { "epoch": 73.5875, "grad_norm": 9.5625, "learning_rate": 3.4411073202282e-06, "loss": 1.3007, "step": 29435 }, { "epoch": 73.59, "grad_norm": 7.96875, "learning_rate": 3.440496142353661e-06, "loss": 1.4191, "step": 29436 }, { "epoch": 73.5925, "grad_norm": 6.1875, "learning_rate": 3.4398850074829627e-06, "loss": 1.4131, "step": 29437 }, { "epoch": 73.595, "grad_norm": 6.875, "learning_rate": 3.439273915620118e-06, "loss": 1.3587, "step": 29438 }, { "epoch": 73.5975, "grad_norm": 8.625, "learning_rate": 3.438662866769129e-06, "loss": 1.2875, "step": 29439 }, { "epoch": 73.6, "grad_norm": 7.84375, "learning_rate": 3.4380518609340076e-06, "loss": 1.2825, "step": 29440 }, { "epoch": 73.6025, "grad_norm": 7.75, "learning_rate": 3.4374408981187547e-06, "loss": 1.3317, "step": 29441 }, { "epoch": 73.605, "grad_norm": 7.4375, "learning_rate": 3.4368299783273727e-06, "loss": 1.2119, "step": 29442 }, { "epoch": 73.6075, "grad_norm": 7.875, "learning_rate": 3.4362191015638715e-06, "loss": 1.4596, "step": 29443 }, { "epoch": 73.61, "grad_norm": 5.53125, "learning_rate": 3.435608267832259e-06, "loss": 1.2341, "step": 29444 }, { "epoch": 73.6125, "grad_norm": 7.5625, "learning_rate": 3.4349974771365347e-06, "loss": 1.3184, "step": 29445 }, { "epoch": 73.615, "grad_norm": 6.8125, "learning_rate": 3.434386729480702e-06, "loss": 1.3104, "step": 29446 }, { "epoch": 73.6175, "grad_norm": 5.15625, "learning_rate": 3.433776024868767e-06, "loss": 1.1725, "step": 29447 }, { "epoch": 73.62, "grad_norm": 6.53125, "learning_rate": 3.4331653633047356e-06, "loss": 1.313, "step": 29448 }, { "epoch": 73.6225, "grad_norm": 7.21875, "learning_rate": 3.432554744792611e-06, "loss": 1.2127, "step": 29449 }, { "epoch": 73.625, "grad_norm": 7.3125, "learning_rate": 3.431944169336391e-06, "loss": 1.1984, "step": 29450 }, { "epoch": 73.6275, "grad_norm": 12.125, "learning_rate": 3.431333636940085e-06, "loss": 1.3, "step": 29451 }, { "epoch": 73.63, "grad_norm": 8.1875, "learning_rate": 3.4307231476076896e-06, "loss": 1.4832, "step": 29452 }, { "epoch": 73.6325, "grad_norm": 5.40625, "learning_rate": 3.4301127013432146e-06, "loss": 1.272, "step": 29453 }, { "epoch": 73.635, "grad_norm": 8.75, "learning_rate": 3.4295022981506533e-06, "loss": 1.2877, "step": 29454 }, { "epoch": 73.6375, "grad_norm": 9.4375, "learning_rate": 3.4288919380340157e-06, "loss": 1.3666, "step": 29455 }, { "epoch": 73.64, "grad_norm": 7.84375, "learning_rate": 3.428281620997296e-06, "loss": 1.2531, "step": 29456 }, { "epoch": 73.6425, "grad_norm": 8.0625, "learning_rate": 3.4276713470445033e-06, "loss": 1.372, "step": 29457 }, { "epoch": 73.645, "grad_norm": 9.8125, "learning_rate": 3.4270611161796296e-06, "loss": 1.1986, "step": 29458 }, { "epoch": 73.6475, "grad_norm": 6.6875, "learning_rate": 3.426450928406685e-06, "loss": 1.3537, "step": 29459 }, { "epoch": 73.65, "grad_norm": 10.5625, "learning_rate": 3.4258407837296635e-06, "loss": 1.3031, "step": 29460 }, { "epoch": 73.6525, "grad_norm": 5.5625, "learning_rate": 3.4252306821525637e-06, "loss": 1.2725, "step": 29461 }, { "epoch": 73.655, "grad_norm": 7.6875, "learning_rate": 3.42462062367939e-06, "loss": 1.2772, "step": 29462 }, { "epoch": 73.6575, "grad_norm": 6.34375, "learning_rate": 3.424010608314142e-06, "loss": 1.1965, "step": 29463 }, { "epoch": 73.66, "grad_norm": 6.90625, "learning_rate": 3.4234006360608173e-06, "loss": 1.3964, "step": 29464 }, { "epoch": 73.6625, "grad_norm": 6.0, "learning_rate": 3.4227907069234125e-06, "loss": 1.3775, "step": 29465 }, { "epoch": 73.665, "grad_norm": 10.6875, "learning_rate": 3.4221808209059283e-06, "loss": 1.2725, "step": 29466 }, { "epoch": 73.6675, "grad_norm": 5.875, "learning_rate": 3.421570978012366e-06, "loss": 1.4096, "step": 29467 }, { "epoch": 73.67, "grad_norm": 9.8125, "learning_rate": 3.4209611782467224e-06, "loss": 1.4435, "step": 29468 }, { "epoch": 73.6725, "grad_norm": 7.1875, "learning_rate": 3.4203514216129907e-06, "loss": 1.3151, "step": 29469 }, { "epoch": 73.675, "grad_norm": 6.34375, "learning_rate": 3.419741708115174e-06, "loss": 1.2919, "step": 29470 }, { "epoch": 73.6775, "grad_norm": 7.6875, "learning_rate": 3.4191320377572645e-06, "loss": 1.2484, "step": 29471 }, { "epoch": 73.68, "grad_norm": 7.5, "learning_rate": 3.418522410543266e-06, "loss": 1.322, "step": 29472 }, { "epoch": 73.6825, "grad_norm": 7.9375, "learning_rate": 3.4179128264771668e-06, "loss": 1.3351, "step": 29473 }, { "epoch": 73.685, "grad_norm": 9.9375, "learning_rate": 3.4173032855629705e-06, "loss": 1.4172, "step": 29474 }, { "epoch": 73.6875, "grad_norm": 13.75, "learning_rate": 3.4166937878046723e-06, "loss": 1.293, "step": 29475 }, { "epoch": 73.69, "grad_norm": 6.375, "learning_rate": 3.4160843332062623e-06, "loss": 1.3463, "step": 29476 }, { "epoch": 73.6925, "grad_norm": 6.1875, "learning_rate": 3.4154749217717386e-06, "loss": 1.3738, "step": 29477 }, { "epoch": 73.695, "grad_norm": 5.59375, "learning_rate": 3.4148655535051023e-06, "loss": 1.4147, "step": 29478 }, { "epoch": 73.6975, "grad_norm": 9.5, "learning_rate": 3.4142562284103444e-06, "loss": 1.3286, "step": 29479 }, { "epoch": 73.7, "grad_norm": 4.96875, "learning_rate": 3.413646946491458e-06, "loss": 1.2362, "step": 29480 }, { "epoch": 73.7025, "grad_norm": 7.1875, "learning_rate": 3.4130377077524324e-06, "loss": 1.413, "step": 29481 }, { "epoch": 73.705, "grad_norm": 7.6875, "learning_rate": 3.4124285121972746e-06, "loss": 1.2863, "step": 29482 }, { "epoch": 73.7075, "grad_norm": 7.4375, "learning_rate": 3.411819359829972e-06, "loss": 1.3281, "step": 29483 }, { "epoch": 73.71, "grad_norm": 7.4375, "learning_rate": 3.411210250654514e-06, "loss": 1.4503, "step": 29484 }, { "epoch": 73.7125, "grad_norm": 6.53125, "learning_rate": 3.410601184674902e-06, "loss": 1.5185, "step": 29485 }, { "epoch": 73.715, "grad_norm": 7.34375, "learning_rate": 3.4099921618951213e-06, "loss": 1.2812, "step": 29486 }, { "epoch": 73.7175, "grad_norm": 6.4375, "learning_rate": 3.4093831823191727e-06, "loss": 1.3536, "step": 29487 }, { "epoch": 73.72, "grad_norm": 5.40625, "learning_rate": 3.4087742459510396e-06, "loss": 1.4159, "step": 29488 }, { "epoch": 73.7225, "grad_norm": 6.71875, "learning_rate": 3.4081653527947222e-06, "loss": 1.4171, "step": 29489 }, { "epoch": 73.725, "grad_norm": 7.1875, "learning_rate": 3.407556502854209e-06, "loss": 1.3299, "step": 29490 }, { "epoch": 73.7275, "grad_norm": 8.6875, "learning_rate": 3.406947696133489e-06, "loss": 1.3053, "step": 29491 }, { "epoch": 73.73, "grad_norm": 7.96875, "learning_rate": 3.406338932636556e-06, "loss": 1.3559, "step": 29492 }, { "epoch": 73.7325, "grad_norm": 6.8125, "learning_rate": 3.405730212367403e-06, "loss": 1.3493, "step": 29493 }, { "epoch": 73.735, "grad_norm": 8.0625, "learning_rate": 3.4051215353300202e-06, "loss": 1.2606, "step": 29494 }, { "epoch": 73.7375, "grad_norm": 6.90625, "learning_rate": 3.4045129015283964e-06, "loss": 1.2106, "step": 29495 }, { "epoch": 73.74, "grad_norm": 4.96875, "learning_rate": 3.403904310966515e-06, "loss": 1.2827, "step": 29496 }, { "epoch": 73.7425, "grad_norm": 9.125, "learning_rate": 3.40329576364838e-06, "loss": 1.2168, "step": 29497 }, { "epoch": 73.745, "grad_norm": 10.125, "learning_rate": 3.4026872595779735e-06, "loss": 1.3528, "step": 29498 }, { "epoch": 73.7475, "grad_norm": 8.3125, "learning_rate": 3.4020787987592862e-06, "loss": 1.1082, "step": 29499 }, { "epoch": 73.75, "grad_norm": 6.71875, "learning_rate": 3.4014703811963024e-06, "loss": 1.3634, "step": 29500 }, { "epoch": 73.7525, "grad_norm": 7.40625, "learning_rate": 3.4008620068930153e-06, "loss": 1.3174, "step": 29501 }, { "epoch": 73.755, "grad_norm": 6.78125, "learning_rate": 3.4002536758534154e-06, "loss": 1.3564, "step": 29502 }, { "epoch": 73.7575, "grad_norm": 6.8125, "learning_rate": 3.3996453880814883e-06, "loss": 1.2703, "step": 29503 }, { "epoch": 73.76, "grad_norm": 7.6875, "learning_rate": 3.3990371435812185e-06, "loss": 1.3567, "step": 29504 }, { "epoch": 73.7625, "grad_norm": 6.96875, "learning_rate": 3.398428942356601e-06, "loss": 1.3547, "step": 29505 }, { "epoch": 73.765, "grad_norm": 5.53125, "learning_rate": 3.3978207844116164e-06, "loss": 1.4767, "step": 29506 }, { "epoch": 73.7675, "grad_norm": 10.8125, "learning_rate": 3.397212669750256e-06, "loss": 1.3872, "step": 29507 }, { "epoch": 73.77, "grad_norm": 5.6875, "learning_rate": 3.396604598376503e-06, "loss": 1.3174, "step": 29508 }, { "epoch": 73.7725, "grad_norm": 6.9375, "learning_rate": 3.395996570294349e-06, "loss": 1.3819, "step": 29509 }, { "epoch": 73.775, "grad_norm": 6.84375, "learning_rate": 3.3953885855077773e-06, "loss": 1.3437, "step": 29510 }, { "epoch": 73.7775, "grad_norm": 10.125, "learning_rate": 3.3947806440207677e-06, "loss": 1.1988, "step": 29511 }, { "epoch": 73.78, "grad_norm": 8.75, "learning_rate": 3.3941727458373177e-06, "loss": 1.1646, "step": 29512 }, { "epoch": 73.7825, "grad_norm": 8.875, "learning_rate": 3.3935648909614073e-06, "loss": 1.2866, "step": 29513 }, { "epoch": 73.785, "grad_norm": 7.5, "learning_rate": 3.3929570793970202e-06, "loss": 1.2468, "step": 29514 }, { "epoch": 73.7875, "grad_norm": 6.21875, "learning_rate": 3.3923493111481397e-06, "loss": 1.2926, "step": 29515 }, { "epoch": 73.79, "grad_norm": 7.53125, "learning_rate": 3.391741586218752e-06, "loss": 1.2507, "step": 29516 }, { "epoch": 73.7925, "grad_norm": 9.4375, "learning_rate": 3.3911339046128455e-06, "loss": 1.3846, "step": 29517 }, { "epoch": 73.795, "grad_norm": 6.0625, "learning_rate": 3.390526266334402e-06, "loss": 1.308, "step": 29518 }, { "epoch": 73.7975, "grad_norm": 19.25, "learning_rate": 3.389918671387399e-06, "loss": 1.2651, "step": 29519 }, { "epoch": 73.8, "grad_norm": 5.3125, "learning_rate": 3.3893111197758276e-06, "loss": 1.1895, "step": 29520 }, { "epoch": 73.8025, "grad_norm": 7.78125, "learning_rate": 3.388703611503665e-06, "loss": 1.4139, "step": 29521 }, { "epoch": 73.805, "grad_norm": 5.1875, "learning_rate": 3.388096146574902e-06, "loss": 1.3141, "step": 29522 }, { "epoch": 73.8075, "grad_norm": 8.3125, "learning_rate": 3.3874887249935107e-06, "loss": 1.3761, "step": 29523 }, { "epoch": 73.81, "grad_norm": 9.5625, "learning_rate": 3.3868813467634833e-06, "loss": 1.317, "step": 29524 }, { "epoch": 73.8125, "grad_norm": 7.71875, "learning_rate": 3.386274011888796e-06, "loss": 1.2714, "step": 29525 }, { "epoch": 73.815, "grad_norm": 7.65625, "learning_rate": 3.3856667203734293e-06, "loss": 1.2409, "step": 29526 }, { "epoch": 73.8175, "grad_norm": 8.9375, "learning_rate": 3.385059472221366e-06, "loss": 1.2866, "step": 29527 }, { "epoch": 73.82, "grad_norm": 8.5625, "learning_rate": 3.3844522674365922e-06, "loss": 1.2879, "step": 29528 }, { "epoch": 73.8225, "grad_norm": 8.6875, "learning_rate": 3.383845106023084e-06, "loss": 1.2421, "step": 29529 }, { "epoch": 73.825, "grad_norm": 10.625, "learning_rate": 3.3832379879848197e-06, "loss": 1.2018, "step": 29530 }, { "epoch": 73.8275, "grad_norm": 6.5625, "learning_rate": 3.382630913325782e-06, "loss": 1.3048, "step": 29531 }, { "epoch": 73.83, "grad_norm": 8.9375, "learning_rate": 3.382023882049954e-06, "loss": 1.3907, "step": 29532 }, { "epoch": 73.8325, "grad_norm": 7.46875, "learning_rate": 3.3814168941613136e-06, "loss": 1.3997, "step": 29533 }, { "epoch": 73.835, "grad_norm": 6.6875, "learning_rate": 3.3808099496638357e-06, "loss": 1.4396, "step": 29534 }, { "epoch": 73.8375, "grad_norm": 8.9375, "learning_rate": 3.3802030485615057e-06, "loss": 1.4203, "step": 29535 }, { "epoch": 73.84, "grad_norm": 9.1875, "learning_rate": 3.3795961908582965e-06, "loss": 1.2519, "step": 29536 }, { "epoch": 73.8425, "grad_norm": 8.25, "learning_rate": 3.3789893765581936e-06, "loss": 1.2332, "step": 29537 }, { "epoch": 73.845, "grad_norm": 8.25, "learning_rate": 3.3783826056651682e-06, "loss": 1.3269, "step": 29538 }, { "epoch": 73.8475, "grad_norm": 8.625, "learning_rate": 3.377775878183204e-06, "loss": 1.2355, "step": 29539 }, { "epoch": 73.85, "grad_norm": 6.65625, "learning_rate": 3.3771691941162755e-06, "loss": 1.3447, "step": 29540 }, { "epoch": 73.8525, "grad_norm": 6.9375, "learning_rate": 3.376562553468359e-06, "loss": 1.4089, "step": 29541 }, { "epoch": 73.855, "grad_norm": 10.25, "learning_rate": 3.375955956243432e-06, "loss": 1.2752, "step": 29542 }, { "epoch": 73.8575, "grad_norm": 9.0625, "learning_rate": 3.375349402445477e-06, "loss": 1.4672, "step": 29543 }, { "epoch": 73.86, "grad_norm": 6.125, "learning_rate": 3.374742892078465e-06, "loss": 1.2216, "step": 29544 }, { "epoch": 73.8625, "grad_norm": 7.3125, "learning_rate": 3.3741364251463714e-06, "loss": 1.1047, "step": 29545 }, { "epoch": 73.865, "grad_norm": 6.71875, "learning_rate": 3.3735300016531736e-06, "loss": 1.1824, "step": 29546 }, { "epoch": 73.8675, "grad_norm": 8.125, "learning_rate": 3.3729236216028515e-06, "loss": 1.3092, "step": 29547 }, { "epoch": 73.87, "grad_norm": 9.625, "learning_rate": 3.372317284999377e-06, "loss": 1.3807, "step": 29548 }, { "epoch": 73.8725, "grad_norm": 7.65625, "learning_rate": 3.3717109918467217e-06, "loss": 1.2941, "step": 29549 }, { "epoch": 73.875, "grad_norm": 10.1875, "learning_rate": 3.3711047421488676e-06, "loss": 1.3358, "step": 29550 }, { "epoch": 73.8775, "grad_norm": 6.15625, "learning_rate": 3.370498535909782e-06, "loss": 1.358, "step": 29551 }, { "epoch": 73.88, "grad_norm": 7.75, "learning_rate": 3.3698923731334453e-06, "loss": 1.3666, "step": 29552 }, { "epoch": 73.8825, "grad_norm": 8.6875, "learning_rate": 3.369286253823827e-06, "loss": 1.5151, "step": 29553 }, { "epoch": 73.885, "grad_norm": 5.75, "learning_rate": 3.3686801779849045e-06, "loss": 1.2231, "step": 29554 }, { "epoch": 73.8875, "grad_norm": 8.75, "learning_rate": 3.36807414562065e-06, "loss": 1.4345, "step": 29555 }, { "epoch": 73.89, "grad_norm": 7.03125, "learning_rate": 3.3674681567350332e-06, "loss": 1.2179, "step": 29556 }, { "epoch": 73.8925, "grad_norm": 6.75, "learning_rate": 3.3668622113320294e-06, "loss": 1.2915, "step": 29557 }, { "epoch": 73.895, "grad_norm": 7.75, "learning_rate": 3.366256309415614e-06, "loss": 1.2515, "step": 29558 }, { "epoch": 73.8975, "grad_norm": 5.0, "learning_rate": 3.365650450989757e-06, "loss": 1.3975, "step": 29559 }, { "epoch": 73.9, "grad_norm": 7.34375, "learning_rate": 3.3650446360584276e-06, "loss": 1.3416, "step": 29560 }, { "epoch": 73.9025, "grad_norm": 7.59375, "learning_rate": 3.3644388646255998e-06, "loss": 1.2232, "step": 29561 }, { "epoch": 73.905, "grad_norm": 6.84375, "learning_rate": 3.3638331366952483e-06, "loss": 1.3846, "step": 29562 }, { "epoch": 73.9075, "grad_norm": 5.90625, "learning_rate": 3.3632274522713414e-06, "loss": 1.3327, "step": 29563 }, { "epoch": 73.91, "grad_norm": 6.53125, "learning_rate": 3.3626218113578467e-06, "loss": 1.3416, "step": 29564 }, { "epoch": 73.9125, "grad_norm": 6.875, "learning_rate": 3.362016213958741e-06, "loss": 1.2846, "step": 29565 }, { "epoch": 73.915, "grad_norm": 8.9375, "learning_rate": 3.3614106600779874e-06, "loss": 1.1542, "step": 29566 }, { "epoch": 73.9175, "grad_norm": 7.375, "learning_rate": 3.360805149719564e-06, "loss": 1.3056, "step": 29567 }, { "epoch": 73.92, "grad_norm": 5.75, "learning_rate": 3.360199682887433e-06, "loss": 1.4395, "step": 29568 }, { "epoch": 73.9225, "grad_norm": 7.4375, "learning_rate": 3.359594259585569e-06, "loss": 1.4041, "step": 29569 }, { "epoch": 73.925, "grad_norm": 6.84375, "learning_rate": 3.3589888798179415e-06, "loss": 1.3992, "step": 29570 }, { "epoch": 73.9275, "grad_norm": 8.0, "learning_rate": 3.3583835435885126e-06, "loss": 1.3236, "step": 29571 }, { "epoch": 73.93, "grad_norm": 8.4375, "learning_rate": 3.3577782509012557e-06, "loss": 1.22, "step": 29572 }, { "epoch": 73.9325, "grad_norm": 7.40625, "learning_rate": 3.3571730017601414e-06, "loss": 1.3324, "step": 29573 }, { "epoch": 73.935, "grad_norm": 10.4375, "learning_rate": 3.356567796169136e-06, "loss": 1.4381, "step": 29574 }, { "epoch": 73.9375, "grad_norm": 14.625, "learning_rate": 3.3559626341322027e-06, "loss": 1.317, "step": 29575 }, { "epoch": 73.94, "grad_norm": 9.9375, "learning_rate": 3.3553575156533125e-06, "loss": 1.2787, "step": 29576 }, { "epoch": 73.9425, "grad_norm": 8.9375, "learning_rate": 3.3547524407364353e-06, "loss": 1.1792, "step": 29577 }, { "epoch": 73.945, "grad_norm": 10.8125, "learning_rate": 3.3541474093855352e-06, "loss": 1.3118, "step": 29578 }, { "epoch": 73.9475, "grad_norm": 6.9375, "learning_rate": 3.3535424216045763e-06, "loss": 1.293, "step": 29579 }, { "epoch": 73.95, "grad_norm": 8.3125, "learning_rate": 3.35293747739753e-06, "loss": 1.2974, "step": 29580 }, { "epoch": 73.9525, "grad_norm": 7.75, "learning_rate": 3.352332576768357e-06, "loss": 1.4124, "step": 29581 }, { "epoch": 73.955, "grad_norm": 8.3125, "learning_rate": 3.3517277197210274e-06, "loss": 1.1805, "step": 29582 }, { "epoch": 73.9575, "grad_norm": 6.71875, "learning_rate": 3.351122906259503e-06, "loss": 1.389, "step": 29583 }, { "epoch": 73.96, "grad_norm": 8.5625, "learning_rate": 3.3505181363877536e-06, "loss": 1.3439, "step": 29584 }, { "epoch": 73.9625, "grad_norm": 6.625, "learning_rate": 3.3499134101097408e-06, "loss": 1.3882, "step": 29585 }, { "epoch": 73.965, "grad_norm": 8.5625, "learning_rate": 3.3493087274294265e-06, "loss": 1.4906, "step": 29586 }, { "epoch": 73.9675, "grad_norm": 9.6875, "learning_rate": 3.348704088350778e-06, "loss": 1.4789, "step": 29587 }, { "epoch": 73.97, "grad_norm": 6.6875, "learning_rate": 3.3480994928777633e-06, "loss": 1.312, "step": 29588 }, { "epoch": 73.9725, "grad_norm": 12.3125, "learning_rate": 3.347494941014341e-06, "loss": 1.3243, "step": 29589 }, { "epoch": 73.975, "grad_norm": 7.375, "learning_rate": 3.3468904327644736e-06, "loss": 1.3234, "step": 29590 }, { "epoch": 73.9775, "grad_norm": 7.03125, "learning_rate": 3.346285968132125e-06, "loss": 1.2497, "step": 29591 }, { "epoch": 73.98, "grad_norm": 6.71875, "learning_rate": 3.3456815471212634e-06, "loss": 1.4107, "step": 29592 }, { "epoch": 73.9825, "grad_norm": 8.625, "learning_rate": 3.3450771697358476e-06, "loss": 1.2945, "step": 29593 }, { "epoch": 73.985, "grad_norm": 6.53125, "learning_rate": 3.344472835979835e-06, "loss": 1.162, "step": 29594 }, { "epoch": 73.9875, "grad_norm": 6.5625, "learning_rate": 3.343868545857196e-06, "loss": 1.4343, "step": 29595 }, { "epoch": 73.99, "grad_norm": 6.8125, "learning_rate": 3.3432642993718856e-06, "loss": 1.2679, "step": 29596 }, { "epoch": 73.9925, "grad_norm": 5.90625, "learning_rate": 3.3426600965278712e-06, "loss": 1.3152, "step": 29597 }, { "epoch": 73.995, "grad_norm": 8.75, "learning_rate": 3.3420559373291073e-06, "loss": 1.0794, "step": 29598 }, { "epoch": 73.9975, "grad_norm": 9.9375, "learning_rate": 3.3414518217795623e-06, "loss": 1.3056, "step": 29599 }, { "epoch": 74.0, "grad_norm": 11.0, "learning_rate": 3.3408477498831917e-06, "loss": 1.3746, "step": 29600 }, { "epoch": 74.0025, "grad_norm": 7.15625, "learning_rate": 3.340243721643954e-06, "loss": 1.2646, "step": 29601 }, { "epoch": 74.005, "grad_norm": 7.53125, "learning_rate": 3.3396397370658117e-06, "loss": 1.2407, "step": 29602 }, { "epoch": 74.0075, "grad_norm": 8.875, "learning_rate": 3.3390357961527285e-06, "loss": 1.3423, "step": 29603 }, { "epoch": 74.01, "grad_norm": 5.8125, "learning_rate": 3.3384318989086596e-06, "loss": 1.3296, "step": 29604 }, { "epoch": 74.0125, "grad_norm": 6.96875, "learning_rate": 3.3378280453375613e-06, "loss": 1.3433, "step": 29605 }, { "epoch": 74.015, "grad_norm": 9.1875, "learning_rate": 3.337224235443396e-06, "loss": 1.3243, "step": 29606 }, { "epoch": 74.0175, "grad_norm": 6.03125, "learning_rate": 3.3366204692301253e-06, "loss": 1.1608, "step": 29607 }, { "epoch": 74.02, "grad_norm": 13.0, "learning_rate": 3.3360167467017036e-06, "loss": 1.4189, "step": 29608 }, { "epoch": 74.0225, "grad_norm": 7.25, "learning_rate": 3.3354130678620868e-06, "loss": 1.5046, "step": 29609 }, { "epoch": 74.025, "grad_norm": 7.3125, "learning_rate": 3.334809432715238e-06, "loss": 1.2953, "step": 29610 }, { "epoch": 74.0275, "grad_norm": 8.25, "learning_rate": 3.334205841265108e-06, "loss": 1.3577, "step": 29611 }, { "epoch": 74.03, "grad_norm": 15.1875, "learning_rate": 3.333602293515662e-06, "loss": 1.4057, "step": 29612 }, { "epoch": 74.0325, "grad_norm": 8.375, "learning_rate": 3.3329987894708494e-06, "loss": 1.2345, "step": 29613 }, { "epoch": 74.035, "grad_norm": 6.40625, "learning_rate": 3.3323953291346325e-06, "loss": 1.3393, "step": 29614 }, { "epoch": 74.0375, "grad_norm": 7.0625, "learning_rate": 3.3317919125109653e-06, "loss": 1.3375, "step": 29615 }, { "epoch": 74.04, "grad_norm": 9.625, "learning_rate": 3.3311885396038002e-06, "loss": 1.2546, "step": 29616 }, { "epoch": 74.0425, "grad_norm": 8.0625, "learning_rate": 3.3305852104170967e-06, "loss": 1.4297, "step": 29617 }, { "epoch": 74.045, "grad_norm": 8.0625, "learning_rate": 3.3299819249548128e-06, "loss": 1.3642, "step": 29618 }, { "epoch": 74.0475, "grad_norm": 9.4375, "learning_rate": 3.3293786832208995e-06, "loss": 1.5212, "step": 29619 }, { "epoch": 74.05, "grad_norm": 7.3125, "learning_rate": 3.3287754852193143e-06, "loss": 1.3674, "step": 29620 }, { "epoch": 74.0525, "grad_norm": 7.9375, "learning_rate": 3.3281723309540013e-06, "loss": 1.2768, "step": 29621 }, { "epoch": 74.055, "grad_norm": 8.0625, "learning_rate": 3.3275692204289323e-06, "loss": 1.3192, "step": 29622 }, { "epoch": 74.0575, "grad_norm": 10.4375, "learning_rate": 3.3269661536480514e-06, "loss": 1.4022, "step": 29623 }, { "epoch": 74.06, "grad_norm": 8.125, "learning_rate": 3.32636313061531e-06, "loss": 1.2345, "step": 29624 }, { "epoch": 74.0625, "grad_norm": 4.71875, "learning_rate": 3.325760151334668e-06, "loss": 1.2978, "step": 29625 }, { "epoch": 74.065, "grad_norm": 7.84375, "learning_rate": 3.325157215810072e-06, "loss": 1.2748, "step": 29626 }, { "epoch": 74.0675, "grad_norm": 7.1875, "learning_rate": 3.3245543240454825e-06, "loss": 1.3211, "step": 29627 }, { "epoch": 74.07, "grad_norm": 6.6875, "learning_rate": 3.323951476044843e-06, "loss": 1.3977, "step": 29628 }, { "epoch": 74.0725, "grad_norm": 7.96875, "learning_rate": 3.3233486718121145e-06, "loss": 1.2736, "step": 29629 }, { "epoch": 74.075, "grad_norm": 5.71875, "learning_rate": 3.322745911351245e-06, "loss": 1.347, "step": 29630 }, { "epoch": 74.0775, "grad_norm": 8.4375, "learning_rate": 3.322143194666183e-06, "loss": 1.3651, "step": 29631 }, { "epoch": 74.08, "grad_norm": 13.375, "learning_rate": 3.321540521760883e-06, "loss": 1.4789, "step": 29632 }, { "epoch": 74.0825, "grad_norm": 9.625, "learning_rate": 3.3209378926392986e-06, "loss": 1.2905, "step": 29633 }, { "epoch": 74.085, "grad_norm": 8.875, "learning_rate": 3.320335307305379e-06, "loss": 1.4075, "step": 29634 }, { "epoch": 74.0875, "grad_norm": 7.21875, "learning_rate": 3.319732765763073e-06, "loss": 1.3071, "step": 29635 }, { "epoch": 74.09, "grad_norm": 6.0625, "learning_rate": 3.319130268016325e-06, "loss": 1.3951, "step": 29636 }, { "epoch": 74.0925, "grad_norm": 9.5, "learning_rate": 3.3185278140690992e-06, "loss": 1.3996, "step": 29637 }, { "epoch": 74.095, "grad_norm": 7.71875, "learning_rate": 3.3179254039253373e-06, "loss": 1.3799, "step": 29638 }, { "epoch": 74.0975, "grad_norm": 8.5, "learning_rate": 3.3173230375889886e-06, "loss": 1.2248, "step": 29639 }, { "epoch": 74.1, "grad_norm": 6.46875, "learning_rate": 3.3167207150640003e-06, "loss": 1.4666, "step": 29640 }, { "epoch": 74.1025, "grad_norm": 8.4375, "learning_rate": 3.3161184363543218e-06, "loss": 1.322, "step": 29641 }, { "epoch": 74.105, "grad_norm": 8.1875, "learning_rate": 3.3155162014639086e-06, "loss": 1.3538, "step": 29642 }, { "epoch": 74.1075, "grad_norm": 7.9375, "learning_rate": 3.3149140103967024e-06, "loss": 1.2193, "step": 29643 }, { "epoch": 74.11, "grad_norm": 10.0625, "learning_rate": 3.3143118631566496e-06, "loss": 1.2691, "step": 29644 }, { "epoch": 74.1125, "grad_norm": 7.09375, "learning_rate": 3.313709759747704e-06, "loss": 1.3168, "step": 29645 }, { "epoch": 74.115, "grad_norm": 6.09375, "learning_rate": 3.3131077001738066e-06, "loss": 1.2867, "step": 29646 }, { "epoch": 74.1175, "grad_norm": 8.75, "learning_rate": 3.3125056844389104e-06, "loss": 1.2849, "step": 29647 }, { "epoch": 74.12, "grad_norm": 6.875, "learning_rate": 3.3119037125469553e-06, "loss": 1.3189, "step": 29648 }, { "epoch": 74.1225, "grad_norm": 8.5, "learning_rate": 3.311301784501896e-06, "loss": 1.2574, "step": 29649 }, { "epoch": 74.125, "grad_norm": 5.78125, "learning_rate": 3.3106999003076745e-06, "loss": 1.4176, "step": 29650 }, { "epoch": 74.1275, "grad_norm": 6.71875, "learning_rate": 3.31009805996823e-06, "loss": 1.224, "step": 29651 }, { "epoch": 74.13, "grad_norm": 9.4375, "learning_rate": 3.3094962634875216e-06, "loss": 1.2826, "step": 29652 }, { "epoch": 74.1325, "grad_norm": 10.125, "learning_rate": 3.308894510869488e-06, "loss": 1.2375, "step": 29653 }, { "epoch": 74.135, "grad_norm": 6.90625, "learning_rate": 3.308292802118075e-06, "loss": 1.2357, "step": 29654 }, { "epoch": 74.1375, "grad_norm": 7.21875, "learning_rate": 3.307691137237222e-06, "loss": 1.2962, "step": 29655 }, { "epoch": 74.14, "grad_norm": 9.875, "learning_rate": 3.307089516230878e-06, "loss": 1.358, "step": 29656 }, { "epoch": 74.1425, "grad_norm": 9.1875, "learning_rate": 3.3064879391029916e-06, "loss": 1.3102, "step": 29657 }, { "epoch": 74.145, "grad_norm": 9.875, "learning_rate": 3.3058864058575024e-06, "loss": 1.146, "step": 29658 }, { "epoch": 74.1475, "grad_norm": 7.96875, "learning_rate": 3.3052849164983504e-06, "loss": 1.5169, "step": 29659 }, { "epoch": 74.15, "grad_norm": 8.4375, "learning_rate": 3.304683471029485e-06, "loss": 1.3435, "step": 29660 }, { "epoch": 74.1525, "grad_norm": 5.1875, "learning_rate": 3.3040820694548447e-06, "loss": 1.2302, "step": 29661 }, { "epoch": 74.155, "grad_norm": 7.46875, "learning_rate": 3.3034807117783774e-06, "loss": 1.3435, "step": 29662 }, { "epoch": 74.1575, "grad_norm": 9.5625, "learning_rate": 3.3028793980040198e-06, "loss": 1.3124, "step": 29663 }, { "epoch": 74.16, "grad_norm": 12.25, "learning_rate": 3.3022781281357184e-06, "loss": 1.5085, "step": 29664 }, { "epoch": 74.1625, "grad_norm": 8.1875, "learning_rate": 3.3016769021774152e-06, "loss": 1.3395, "step": 29665 }, { "epoch": 74.165, "grad_norm": 7.90625, "learning_rate": 3.301075720133047e-06, "loss": 1.5288, "step": 29666 }, { "epoch": 74.1675, "grad_norm": 14.6875, "learning_rate": 3.3004745820065576e-06, "loss": 1.4084, "step": 29667 }, { "epoch": 74.17, "grad_norm": 6.71875, "learning_rate": 3.299873487801891e-06, "loss": 1.297, "step": 29668 }, { "epoch": 74.1725, "grad_norm": 5.46875, "learning_rate": 3.299272437522988e-06, "loss": 1.1678, "step": 29669 }, { "epoch": 74.175, "grad_norm": 10.4375, "learning_rate": 3.298671431173782e-06, "loss": 1.333, "step": 29670 }, { "epoch": 74.1775, "grad_norm": 6.5, "learning_rate": 3.2980704687582176e-06, "loss": 1.3941, "step": 29671 }, { "epoch": 74.18, "grad_norm": 8.9375, "learning_rate": 3.297469550280239e-06, "loss": 1.4048, "step": 29672 }, { "epoch": 74.1825, "grad_norm": 7.5, "learning_rate": 3.296868675743782e-06, "loss": 1.3518, "step": 29673 }, { "epoch": 74.185, "grad_norm": 8.4375, "learning_rate": 3.2962678451527817e-06, "loss": 1.4109, "step": 29674 }, { "epoch": 74.1875, "grad_norm": 7.8125, "learning_rate": 3.295667058511186e-06, "loss": 1.4653, "step": 29675 }, { "epoch": 74.19, "grad_norm": 7.0, "learning_rate": 3.2950663158229233e-06, "loss": 1.3381, "step": 29676 }, { "epoch": 74.1925, "grad_norm": 11.3125, "learning_rate": 3.294465617091942e-06, "loss": 1.2961, "step": 29677 }, { "epoch": 74.195, "grad_norm": 6.0, "learning_rate": 3.293864962322173e-06, "loss": 1.2287, "step": 29678 }, { "epoch": 74.1975, "grad_norm": 6.28125, "learning_rate": 3.29326435151756e-06, "loss": 1.3632, "step": 29679 }, { "epoch": 74.2, "grad_norm": 7.625, "learning_rate": 3.2926637846820366e-06, "loss": 1.4194, "step": 29680 }, { "epoch": 74.2025, "grad_norm": 10.125, "learning_rate": 3.292063261819538e-06, "loss": 1.4128, "step": 29681 }, { "epoch": 74.205, "grad_norm": 5.9375, "learning_rate": 3.291462782934005e-06, "loss": 1.139, "step": 29682 }, { "epoch": 74.2075, "grad_norm": 8.375, "learning_rate": 3.2908623480293754e-06, "loss": 1.2844, "step": 29683 }, { "epoch": 74.21, "grad_norm": 6.21875, "learning_rate": 3.290261957109585e-06, "loss": 1.2344, "step": 29684 }, { "epoch": 74.2125, "grad_norm": 8.5, "learning_rate": 3.289661610178565e-06, "loss": 1.3816, "step": 29685 }, { "epoch": 74.215, "grad_norm": 8.625, "learning_rate": 3.2890613072402557e-06, "loss": 1.3677, "step": 29686 }, { "epoch": 74.2175, "grad_norm": 7.09375, "learning_rate": 3.2884610482985945e-06, "loss": 1.2623, "step": 29687 }, { "epoch": 74.22, "grad_norm": 7.5, "learning_rate": 3.2878608333575136e-06, "loss": 1.3415, "step": 29688 }, { "epoch": 74.2225, "grad_norm": 7.28125, "learning_rate": 3.287260662420946e-06, "loss": 1.334, "step": 29689 }, { "epoch": 74.225, "grad_norm": 6.15625, "learning_rate": 3.2866605354928316e-06, "loss": 1.3913, "step": 29690 }, { "epoch": 74.2275, "grad_norm": 6.5, "learning_rate": 3.286060452577099e-06, "loss": 1.393, "step": 29691 }, { "epoch": 74.23, "grad_norm": 6.75, "learning_rate": 3.285460413677689e-06, "loss": 1.2591, "step": 29692 }, { "epoch": 74.2325, "grad_norm": 6.9375, "learning_rate": 3.284860418798528e-06, "loss": 1.4805, "step": 29693 }, { "epoch": 74.235, "grad_norm": 6.8125, "learning_rate": 3.2842604679435573e-06, "loss": 1.507, "step": 29694 }, { "epoch": 74.2375, "grad_norm": 6.03125, "learning_rate": 3.283660561116703e-06, "loss": 1.37, "step": 29695 }, { "epoch": 74.24, "grad_norm": 6.75, "learning_rate": 3.2830606983219038e-06, "loss": 1.4363, "step": 29696 }, { "epoch": 74.2425, "grad_norm": 6.40625, "learning_rate": 3.2824608795630863e-06, "loss": 1.2479, "step": 29697 }, { "epoch": 74.245, "grad_norm": 7.4375, "learning_rate": 3.2818611048441905e-06, "loss": 1.4215, "step": 29698 }, { "epoch": 74.2475, "grad_norm": 6.34375, "learning_rate": 3.2812613741691445e-06, "loss": 1.4931, "step": 29699 }, { "epoch": 74.25, "grad_norm": 7.53125, "learning_rate": 3.280661687541876e-06, "loss": 1.1999, "step": 29700 }, { "epoch": 74.2525, "grad_norm": 7.9375, "learning_rate": 3.280062044966321e-06, "loss": 1.3923, "step": 29701 }, { "epoch": 74.255, "grad_norm": 6.84375, "learning_rate": 3.2794624464464144e-06, "loss": 1.4837, "step": 29702 }, { "epoch": 74.2575, "grad_norm": 9.25, "learning_rate": 3.278862891986082e-06, "loss": 1.2422, "step": 29703 }, { "epoch": 74.26, "grad_norm": 8.9375, "learning_rate": 3.278263381589253e-06, "loss": 1.3203, "step": 29704 }, { "epoch": 74.2625, "grad_norm": 6.21875, "learning_rate": 3.27766391525986e-06, "loss": 1.4158, "step": 29705 }, { "epoch": 74.265, "grad_norm": 6.75, "learning_rate": 3.277064493001837e-06, "loss": 1.4181, "step": 29706 }, { "epoch": 74.2675, "grad_norm": 9.0625, "learning_rate": 3.2764651148191094e-06, "loss": 1.3088, "step": 29707 }, { "epoch": 74.27, "grad_norm": 11.875, "learning_rate": 3.2758657807156046e-06, "loss": 1.2843, "step": 29708 }, { "epoch": 74.2725, "grad_norm": 8.9375, "learning_rate": 3.275266490695258e-06, "loss": 1.2723, "step": 29709 }, { "epoch": 74.275, "grad_norm": 9.875, "learning_rate": 3.2746672447619908e-06, "loss": 1.2321, "step": 29710 }, { "epoch": 74.2775, "grad_norm": 8.0625, "learning_rate": 3.2740680429197414e-06, "loss": 1.1805, "step": 29711 }, { "epoch": 74.28, "grad_norm": 8.25, "learning_rate": 3.2734688851724273e-06, "loss": 1.4414, "step": 29712 }, { "epoch": 74.2825, "grad_norm": 7.5625, "learning_rate": 3.2728697715239866e-06, "loss": 1.3101, "step": 29713 }, { "epoch": 74.285, "grad_norm": 7.15625, "learning_rate": 3.272270701978343e-06, "loss": 1.3422, "step": 29714 }, { "epoch": 74.2875, "grad_norm": 7.28125, "learning_rate": 3.2716716765394187e-06, "loss": 1.3006, "step": 29715 }, { "epoch": 74.29, "grad_norm": 6.5, "learning_rate": 3.271072695211146e-06, "loss": 1.3295, "step": 29716 }, { "epoch": 74.2925, "grad_norm": 4.84375, "learning_rate": 3.2704737579974557e-06, "loss": 1.3787, "step": 29717 }, { "epoch": 74.295, "grad_norm": 10.25, "learning_rate": 3.2698748649022693e-06, "loss": 1.2767, "step": 29718 }, { "epoch": 74.2975, "grad_norm": 6.71875, "learning_rate": 3.26927601592951e-06, "loss": 1.2358, "step": 29719 }, { "epoch": 74.3, "grad_norm": 6.34375, "learning_rate": 3.268677211083109e-06, "loss": 1.4174, "step": 29720 }, { "epoch": 74.3025, "grad_norm": 7.46875, "learning_rate": 3.2680784503669936e-06, "loss": 1.1299, "step": 29721 }, { "epoch": 74.305, "grad_norm": 7.59375, "learning_rate": 3.267479733785086e-06, "loss": 1.3849, "step": 29722 }, { "epoch": 74.3075, "grad_norm": 6.3125, "learning_rate": 3.266881061341309e-06, "loss": 1.2637, "step": 29723 }, { "epoch": 74.31, "grad_norm": 8.5625, "learning_rate": 3.2662824330395946e-06, "loss": 1.3545, "step": 29724 }, { "epoch": 74.3125, "grad_norm": 7.46875, "learning_rate": 3.265683848883859e-06, "loss": 1.3555, "step": 29725 }, { "epoch": 74.315, "grad_norm": 9.6875, "learning_rate": 3.265085308878033e-06, "loss": 1.3383, "step": 29726 }, { "epoch": 74.3175, "grad_norm": 7.25, "learning_rate": 3.2644868130260367e-06, "loss": 1.3377, "step": 29727 }, { "epoch": 74.32, "grad_norm": 5.21875, "learning_rate": 3.2638883613317974e-06, "loss": 1.2969, "step": 29728 }, { "epoch": 74.3225, "grad_norm": 6.9375, "learning_rate": 3.263289953799237e-06, "loss": 1.3415, "step": 29729 }, { "epoch": 74.325, "grad_norm": 7.875, "learning_rate": 3.2626915904322743e-06, "loss": 1.4282, "step": 29730 }, { "epoch": 74.3275, "grad_norm": 7.46875, "learning_rate": 3.2620932712348364e-06, "loss": 1.3395, "step": 29731 }, { "epoch": 74.33, "grad_norm": 6.375, "learning_rate": 3.2614949962108478e-06, "loss": 1.2822, "step": 29732 }, { "epoch": 74.3325, "grad_norm": 7.3125, "learning_rate": 3.260896765364229e-06, "loss": 1.303, "step": 29733 }, { "epoch": 74.335, "grad_norm": 9.0, "learning_rate": 3.2602985786988984e-06, "loss": 1.2841, "step": 29734 }, { "epoch": 74.3375, "grad_norm": 8.375, "learning_rate": 3.25970043621878e-06, "loss": 1.3779, "step": 29735 }, { "epoch": 74.34, "grad_norm": 6.75, "learning_rate": 3.2591023379277986e-06, "loss": 1.1364, "step": 29736 }, { "epoch": 74.3425, "grad_norm": 8.3125, "learning_rate": 3.258504283829874e-06, "loss": 1.3656, "step": 29737 }, { "epoch": 74.345, "grad_norm": 6.1875, "learning_rate": 3.2579062739289215e-06, "loss": 1.3853, "step": 29738 }, { "epoch": 74.3475, "grad_norm": 6.03125, "learning_rate": 3.257308308228868e-06, "loss": 1.4219, "step": 29739 }, { "epoch": 74.35, "grad_norm": 7.25, "learning_rate": 3.256710386733629e-06, "loss": 1.3997, "step": 29740 }, { "epoch": 74.3525, "grad_norm": 6.34375, "learning_rate": 3.25611250944713e-06, "loss": 1.2827, "step": 29741 }, { "epoch": 74.355, "grad_norm": 5.8125, "learning_rate": 3.2555146763732848e-06, "loss": 1.2844, "step": 29742 }, { "epoch": 74.3575, "grad_norm": 11.5625, "learning_rate": 3.2549168875160187e-06, "loss": 1.5099, "step": 29743 }, { "epoch": 74.36, "grad_norm": 6.75, "learning_rate": 3.2543191428792466e-06, "loss": 1.4211, "step": 29744 }, { "epoch": 74.3625, "grad_norm": 7.0625, "learning_rate": 3.253721442466886e-06, "loss": 1.3983, "step": 29745 }, { "epoch": 74.365, "grad_norm": 6.09375, "learning_rate": 3.253123786282857e-06, "loss": 1.3761, "step": 29746 }, { "epoch": 74.3675, "grad_norm": 7.59375, "learning_rate": 3.252526174331082e-06, "loss": 1.3313, "step": 29747 }, { "epoch": 74.37, "grad_norm": 6.15625, "learning_rate": 3.2519286066154753e-06, "loss": 1.0814, "step": 29748 }, { "epoch": 74.3725, "grad_norm": 11.375, "learning_rate": 3.2513310831399513e-06, "loss": 1.4532, "step": 29749 }, { "epoch": 74.375, "grad_norm": 7.4375, "learning_rate": 3.2507336039084315e-06, "loss": 1.238, "step": 29750 }, { "epoch": 74.3775, "grad_norm": 6.46875, "learning_rate": 3.2501361689248333e-06, "loss": 1.4068, "step": 29751 }, { "epoch": 74.38, "grad_norm": 7.09375, "learning_rate": 3.249538778193074e-06, "loss": 1.4209, "step": 29752 }, { "epoch": 74.3825, "grad_norm": 9.0625, "learning_rate": 3.2489414317170655e-06, "loss": 1.3412, "step": 29753 }, { "epoch": 74.385, "grad_norm": 8.3125, "learning_rate": 3.248344129500729e-06, "loss": 1.3904, "step": 29754 }, { "epoch": 74.3875, "grad_norm": 8.0, "learning_rate": 3.247746871547974e-06, "loss": 1.3337, "step": 29755 }, { "epoch": 74.39, "grad_norm": 6.375, "learning_rate": 3.247149657862725e-06, "loss": 1.2522, "step": 29756 }, { "epoch": 74.3925, "grad_norm": 7.21875, "learning_rate": 3.2465524884488898e-06, "loss": 1.2566, "step": 29757 }, { "epoch": 74.395, "grad_norm": 6.5, "learning_rate": 3.245955363310388e-06, "loss": 1.4092, "step": 29758 }, { "epoch": 74.3975, "grad_norm": 8.875, "learning_rate": 3.245358282451133e-06, "loss": 1.3331, "step": 29759 }, { "epoch": 74.4, "grad_norm": 6.84375, "learning_rate": 3.2447612458750365e-06, "loss": 1.3153, "step": 29760 }, { "epoch": 74.4025, "grad_norm": 6.875, "learning_rate": 3.244164253586014e-06, "loss": 1.3656, "step": 29761 }, { "epoch": 74.405, "grad_norm": 9.5, "learning_rate": 3.2435673055879845e-06, "loss": 1.424, "step": 29762 }, { "epoch": 74.4075, "grad_norm": 6.0, "learning_rate": 3.242970401884856e-06, "loss": 1.2613, "step": 29763 }, { "epoch": 74.41, "grad_norm": 6.53125, "learning_rate": 3.2423735424805413e-06, "loss": 1.4681, "step": 29764 }, { "epoch": 74.4125, "grad_norm": 8.375, "learning_rate": 3.2417767273789536e-06, "loss": 1.1792, "step": 29765 }, { "epoch": 74.415, "grad_norm": 6.3125, "learning_rate": 3.241179956584012e-06, "loss": 1.3415, "step": 29766 }, { "epoch": 74.4175, "grad_norm": 7.96875, "learning_rate": 3.2405832300996232e-06, "loss": 1.4028, "step": 29767 }, { "epoch": 74.42, "grad_norm": 5.625, "learning_rate": 3.2399865479296978e-06, "loss": 1.3044, "step": 29768 }, { "epoch": 74.4225, "grad_norm": 11.5, "learning_rate": 3.2393899100781522e-06, "loss": 1.2553, "step": 29769 }, { "epoch": 74.425, "grad_norm": 7.40625, "learning_rate": 3.238793316548894e-06, "loss": 1.2709, "step": 29770 }, { "epoch": 74.4275, "grad_norm": 10.25, "learning_rate": 3.238196767345838e-06, "loss": 1.3989, "step": 29771 }, { "epoch": 74.43, "grad_norm": 10.6875, "learning_rate": 3.237600262472891e-06, "loss": 1.2953, "step": 29772 }, { "epoch": 74.4325, "grad_norm": 8.0, "learning_rate": 3.237003801933969e-06, "loss": 1.3217, "step": 29773 }, { "epoch": 74.435, "grad_norm": 9.625, "learning_rate": 3.2364073857329804e-06, "loss": 1.3782, "step": 29774 }, { "epoch": 74.4375, "grad_norm": 7.3125, "learning_rate": 3.2358110138738297e-06, "loss": 1.3785, "step": 29775 }, { "epoch": 74.44, "grad_norm": 7.78125, "learning_rate": 3.2352146863604317e-06, "loss": 1.3327, "step": 29776 }, { "epoch": 74.4425, "grad_norm": 8.8125, "learning_rate": 3.2346184031966987e-06, "loss": 1.3492, "step": 29777 }, { "epoch": 74.445, "grad_norm": 5.59375, "learning_rate": 3.2340221643865378e-06, "loss": 1.2941, "step": 29778 }, { "epoch": 74.4475, "grad_norm": 7.53125, "learning_rate": 3.2334259699338554e-06, "loss": 1.3001, "step": 29779 }, { "epoch": 74.45, "grad_norm": 8.4375, "learning_rate": 3.2328298198425556e-06, "loss": 1.3384, "step": 29780 }, { "epoch": 74.4525, "grad_norm": 4.34375, "learning_rate": 3.232233714116559e-06, "loss": 1.3372, "step": 29781 }, { "epoch": 74.455, "grad_norm": 8.5, "learning_rate": 3.231637652759768e-06, "loss": 1.2438, "step": 29782 }, { "epoch": 74.4575, "grad_norm": 6.9375, "learning_rate": 3.231041635776089e-06, "loss": 1.3922, "step": 29783 }, { "epoch": 74.46, "grad_norm": 7.8125, "learning_rate": 3.230445663169427e-06, "loss": 1.3932, "step": 29784 }, { "epoch": 74.4625, "grad_norm": 7.0625, "learning_rate": 3.229849734943693e-06, "loss": 1.323, "step": 29785 }, { "epoch": 74.465, "grad_norm": 8.3125, "learning_rate": 3.229253851102796e-06, "loss": 1.2619, "step": 29786 }, { "epoch": 74.4675, "grad_norm": 5.8125, "learning_rate": 3.2286580116506394e-06, "loss": 1.4487, "step": 29787 }, { "epoch": 74.47, "grad_norm": 10.75, "learning_rate": 3.2280622165911267e-06, "loss": 1.4849, "step": 29788 }, { "epoch": 74.4725, "grad_norm": 7.90625, "learning_rate": 3.227466465928171e-06, "loss": 1.3985, "step": 29789 }, { "epoch": 74.475, "grad_norm": 10.125, "learning_rate": 3.2268707596656712e-06, "loss": 1.3989, "step": 29790 }, { "epoch": 74.4775, "grad_norm": 7.78125, "learning_rate": 3.2262750978075342e-06, "loss": 1.1546, "step": 29791 }, { "epoch": 74.48, "grad_norm": 7.46875, "learning_rate": 3.2256794803576707e-06, "loss": 1.2295, "step": 29792 }, { "epoch": 74.4825, "grad_norm": 6.09375, "learning_rate": 3.2250839073199815e-06, "loss": 1.1349, "step": 29793 }, { "epoch": 74.485, "grad_norm": 7.6875, "learning_rate": 3.2244883786983717e-06, "loss": 1.3015, "step": 29794 }, { "epoch": 74.4875, "grad_norm": 6.0625, "learning_rate": 3.2238928944967374e-06, "loss": 1.2986, "step": 29795 }, { "epoch": 74.49, "grad_norm": 9.1875, "learning_rate": 3.2232974547189967e-06, "loss": 1.2863, "step": 29796 }, { "epoch": 74.4925, "grad_norm": 8.25, "learning_rate": 3.222702059369047e-06, "loss": 1.2629, "step": 29797 }, { "epoch": 74.495, "grad_norm": 7.40625, "learning_rate": 3.2221067084507907e-06, "loss": 1.1648, "step": 29798 }, { "epoch": 74.4975, "grad_norm": 7.34375, "learning_rate": 3.221511401968129e-06, "loss": 1.422, "step": 29799 }, { "epoch": 74.5, "grad_norm": 5.9375, "learning_rate": 3.2209161399249677e-06, "loss": 1.2335, "step": 29800 }, { "epoch": 74.5025, "grad_norm": 7.21875, "learning_rate": 3.2203209223252117e-06, "loss": 1.2326, "step": 29801 }, { "epoch": 74.505, "grad_norm": 8.5625, "learning_rate": 3.219725749172761e-06, "loss": 1.3845, "step": 29802 }, { "epoch": 74.5075, "grad_norm": 5.90625, "learning_rate": 3.2191306204715122e-06, "loss": 1.2838, "step": 29803 }, { "epoch": 74.51, "grad_norm": 8.5625, "learning_rate": 3.2185355362253755e-06, "loss": 1.2733, "step": 29804 }, { "epoch": 74.5125, "grad_norm": 6.34375, "learning_rate": 3.2179404964382465e-06, "loss": 1.2849, "step": 29805 }, { "epoch": 74.515, "grad_norm": 9.125, "learning_rate": 3.2173455011140296e-06, "loss": 1.2384, "step": 29806 }, { "epoch": 74.5175, "grad_norm": 8.875, "learning_rate": 3.216750550256622e-06, "loss": 1.1864, "step": 29807 }, { "epoch": 74.52, "grad_norm": 7.15625, "learning_rate": 3.2161556438699303e-06, "loss": 1.4389, "step": 29808 }, { "epoch": 74.5225, "grad_norm": 9.25, "learning_rate": 3.2155607819578495e-06, "loss": 1.2464, "step": 29809 }, { "epoch": 74.525, "grad_norm": 6.75, "learning_rate": 3.214965964524278e-06, "loss": 1.2377, "step": 29810 }, { "epoch": 74.5275, "grad_norm": 9.625, "learning_rate": 3.214371191573118e-06, "loss": 1.397, "step": 29811 }, { "epoch": 74.53, "grad_norm": 7.375, "learning_rate": 3.2137764631082715e-06, "loss": 1.3822, "step": 29812 }, { "epoch": 74.5325, "grad_norm": 9.0625, "learning_rate": 3.213181779133636e-06, "loss": 1.236, "step": 29813 }, { "epoch": 74.535, "grad_norm": 8.3125, "learning_rate": 3.212587139653105e-06, "loss": 1.2888, "step": 29814 }, { "epoch": 74.5375, "grad_norm": 8.6875, "learning_rate": 3.2119925446705824e-06, "loss": 1.3986, "step": 29815 }, { "epoch": 74.54, "grad_norm": 11.875, "learning_rate": 3.2113979941899666e-06, "loss": 1.46, "step": 29816 }, { "epoch": 74.5425, "grad_norm": 10.8125, "learning_rate": 3.210803488215155e-06, "loss": 1.37, "step": 29817 }, { "epoch": 74.545, "grad_norm": 5.90625, "learning_rate": 3.2102090267500406e-06, "loss": 1.305, "step": 29818 }, { "epoch": 74.5475, "grad_norm": 5.71875, "learning_rate": 3.209614609798527e-06, "loss": 1.3077, "step": 29819 }, { "epoch": 74.55, "grad_norm": 11.25, "learning_rate": 3.209020237364505e-06, "loss": 1.2938, "step": 29820 }, { "epoch": 74.5525, "grad_norm": 5.03125, "learning_rate": 3.208425909451879e-06, "loss": 1.3789, "step": 29821 }, { "epoch": 74.555, "grad_norm": 7.5, "learning_rate": 3.207831626064537e-06, "loss": 1.4389, "step": 29822 }, { "epoch": 74.5575, "grad_norm": 10.8125, "learning_rate": 3.207237387206381e-06, "loss": 1.4671, "step": 29823 }, { "epoch": 74.56, "grad_norm": 7.25, "learning_rate": 3.2066431928813068e-06, "loss": 1.3427, "step": 29824 }, { "epoch": 74.5625, "grad_norm": 8.1875, "learning_rate": 3.2060490430932033e-06, "loss": 1.1978, "step": 29825 }, { "epoch": 74.565, "grad_norm": 8.1875, "learning_rate": 3.205454937845971e-06, "loss": 1.2695, "step": 29826 }, { "epoch": 74.5675, "grad_norm": 9.5625, "learning_rate": 3.2048608771435076e-06, "loss": 1.2477, "step": 29827 }, { "epoch": 74.57, "grad_norm": 9.0625, "learning_rate": 3.2042668609897042e-06, "loss": 1.3352, "step": 29828 }, { "epoch": 74.5725, "grad_norm": 6.90625, "learning_rate": 3.203672889388453e-06, "loss": 1.2977, "step": 29829 }, { "epoch": 74.575, "grad_norm": 7.125, "learning_rate": 3.2030789623436485e-06, "loss": 1.2128, "step": 29830 }, { "epoch": 74.5775, "grad_norm": 6.6875, "learning_rate": 3.2024850798591913e-06, "loss": 1.3395, "step": 29831 }, { "epoch": 74.58, "grad_norm": 7.59375, "learning_rate": 3.201891241938969e-06, "loss": 1.3193, "step": 29832 }, { "epoch": 74.5825, "grad_norm": 7.71875, "learning_rate": 3.201297448586872e-06, "loss": 1.2377, "step": 29833 }, { "epoch": 74.585, "grad_norm": 7.71875, "learning_rate": 3.2007036998068007e-06, "loss": 1.3692, "step": 29834 }, { "epoch": 74.5875, "grad_norm": 7.75, "learning_rate": 3.2001099956026407e-06, "loss": 1.3363, "step": 29835 }, { "epoch": 74.59, "grad_norm": 6.09375, "learning_rate": 3.199516335978291e-06, "loss": 1.3345, "step": 29836 }, { "epoch": 74.5925, "grad_norm": 7.46875, "learning_rate": 3.198922720937636e-06, "loss": 1.3151, "step": 29837 }, { "epoch": 74.595, "grad_norm": 7.78125, "learning_rate": 3.198329150484575e-06, "loss": 1.2043, "step": 29838 }, { "epoch": 74.5975, "grad_norm": 12.0625, "learning_rate": 3.197735624622996e-06, "loss": 1.4046, "step": 29839 }, { "epoch": 74.6, "grad_norm": 9.0, "learning_rate": 3.197142143356787e-06, "loss": 1.2326, "step": 29840 }, { "epoch": 74.6025, "grad_norm": 10.5, "learning_rate": 3.1965487066898415e-06, "loss": 1.2779, "step": 29841 }, { "epoch": 74.605, "grad_norm": 7.625, "learning_rate": 3.1959553146260538e-06, "loss": 1.2235, "step": 29842 }, { "epoch": 74.6075, "grad_norm": 9.4375, "learning_rate": 3.1953619671693114e-06, "loss": 1.392, "step": 29843 }, { "epoch": 74.61, "grad_norm": 8.5625, "learning_rate": 3.1947686643235e-06, "loss": 1.1587, "step": 29844 }, { "epoch": 74.6125, "grad_norm": 6.5625, "learning_rate": 3.1941754060925123e-06, "loss": 1.4586, "step": 29845 }, { "epoch": 74.615, "grad_norm": 6.53125, "learning_rate": 3.1935821924802423e-06, "loss": 1.2133, "step": 29846 }, { "epoch": 74.6175, "grad_norm": 5.0625, "learning_rate": 3.1929890234905747e-06, "loss": 1.369, "step": 29847 }, { "epoch": 74.62, "grad_norm": 7.46875, "learning_rate": 3.192395899127395e-06, "loss": 1.367, "step": 29848 }, { "epoch": 74.6225, "grad_norm": 5.75, "learning_rate": 3.1918028193945994e-06, "loss": 1.1916, "step": 29849 }, { "epoch": 74.625, "grad_norm": 6.6875, "learning_rate": 3.1912097842960676e-06, "loss": 1.4047, "step": 29850 }, { "epoch": 74.6275, "grad_norm": 5.6875, "learning_rate": 3.190616793835697e-06, "loss": 1.3457, "step": 29851 }, { "epoch": 74.63, "grad_norm": 8.625, "learning_rate": 3.190023848017365e-06, "loss": 1.2266, "step": 29852 }, { "epoch": 74.6325, "grad_norm": 10.1875, "learning_rate": 3.1894309468449693e-06, "loss": 1.4904, "step": 29853 }, { "epoch": 74.635, "grad_norm": 8.5625, "learning_rate": 3.1888380903223902e-06, "loss": 1.4037, "step": 29854 }, { "epoch": 74.6375, "grad_norm": 9.6875, "learning_rate": 3.1882452784535132e-06, "loss": 1.4554, "step": 29855 }, { "epoch": 74.64, "grad_norm": 9.625, "learning_rate": 3.1876525112422283e-06, "loss": 1.3398, "step": 29856 }, { "epoch": 74.6425, "grad_norm": 8.4375, "learning_rate": 3.187059788692424e-06, "loss": 1.3215, "step": 29857 }, { "epoch": 74.645, "grad_norm": 10.5625, "learning_rate": 3.1864671108079824e-06, "loss": 1.3193, "step": 29858 }, { "epoch": 74.6475, "grad_norm": 6.625, "learning_rate": 3.1858744775927875e-06, "loss": 1.2252, "step": 29859 }, { "epoch": 74.65, "grad_norm": 9.75, "learning_rate": 3.1852818890507255e-06, "loss": 1.4326, "step": 29860 }, { "epoch": 74.6525, "grad_norm": 8.6875, "learning_rate": 3.1846893451856874e-06, "loss": 1.4414, "step": 29861 }, { "epoch": 74.655, "grad_norm": 8.1875, "learning_rate": 3.184096846001552e-06, "loss": 1.2641, "step": 29862 }, { "epoch": 74.6575, "grad_norm": 7.75, "learning_rate": 3.1835043915022025e-06, "loss": 1.3723, "step": 29863 }, { "epoch": 74.66, "grad_norm": 8.1875, "learning_rate": 3.1829119816915276e-06, "loss": 1.4265, "step": 29864 }, { "epoch": 74.6625, "grad_norm": 7.65625, "learning_rate": 3.182319616573407e-06, "loss": 1.2473, "step": 29865 }, { "epoch": 74.665, "grad_norm": 7.53125, "learning_rate": 3.181727296151729e-06, "loss": 1.2977, "step": 29866 }, { "epoch": 74.6675, "grad_norm": 6.78125, "learning_rate": 3.1811350204303706e-06, "loss": 1.3862, "step": 29867 }, { "epoch": 74.67, "grad_norm": 6.8125, "learning_rate": 3.180542789413221e-06, "loss": 1.4529, "step": 29868 }, { "epoch": 74.6725, "grad_norm": 9.6875, "learning_rate": 3.1799506031041606e-06, "loss": 1.333, "step": 29869 }, { "epoch": 74.675, "grad_norm": 8.5625, "learning_rate": 3.1793584615070682e-06, "loss": 1.396, "step": 29870 }, { "epoch": 74.6775, "grad_norm": 8.6875, "learning_rate": 3.178766364625828e-06, "loss": 1.1653, "step": 29871 }, { "epoch": 74.68, "grad_norm": 5.75, "learning_rate": 3.178174312464326e-06, "loss": 1.1716, "step": 29872 }, { "epoch": 74.6825, "grad_norm": 6.90625, "learning_rate": 3.1775823050264397e-06, "loss": 1.3031, "step": 29873 }, { "epoch": 74.685, "grad_norm": 7.40625, "learning_rate": 3.1769903423160485e-06, "loss": 1.3095, "step": 29874 }, { "epoch": 74.6875, "grad_norm": 7.875, "learning_rate": 3.176398424337035e-06, "loss": 1.2767, "step": 29875 }, { "epoch": 74.69, "grad_norm": 11.125, "learning_rate": 3.175806551093283e-06, "loss": 1.3074, "step": 29876 }, { "epoch": 74.6925, "grad_norm": 7.0, "learning_rate": 3.175214722588672e-06, "loss": 1.2056, "step": 29877 }, { "epoch": 74.695, "grad_norm": 6.625, "learning_rate": 3.1746229388270756e-06, "loss": 1.3761, "step": 29878 }, { "epoch": 74.6975, "grad_norm": 8.125, "learning_rate": 3.174031199812382e-06, "loss": 1.3717, "step": 29879 }, { "epoch": 74.7, "grad_norm": 7.09375, "learning_rate": 3.1734395055484623e-06, "loss": 1.2907, "step": 29880 }, { "epoch": 74.7025, "grad_norm": 8.875, "learning_rate": 3.172847856039204e-06, "loss": 1.18, "step": 29881 }, { "epoch": 74.705, "grad_norm": 5.59375, "learning_rate": 3.172256251288479e-06, "loss": 1.2542, "step": 29882 }, { "epoch": 74.7075, "grad_norm": 12.3125, "learning_rate": 3.1716646913001715e-06, "loss": 1.2446, "step": 29883 }, { "epoch": 74.71, "grad_norm": 12.3125, "learning_rate": 3.1710731760781575e-06, "loss": 1.4352, "step": 29884 }, { "epoch": 74.7125, "grad_norm": 4.46875, "learning_rate": 3.1704817056263106e-06, "loss": 1.181, "step": 29885 }, { "epoch": 74.715, "grad_norm": 10.5, "learning_rate": 3.1698902799485133e-06, "loss": 1.295, "step": 29886 }, { "epoch": 74.7175, "grad_norm": 11.5, "learning_rate": 3.169298899048644e-06, "loss": 1.3112, "step": 29887 }, { "epoch": 74.72, "grad_norm": 6.84375, "learning_rate": 3.1687075629305787e-06, "loss": 1.2115, "step": 29888 }, { "epoch": 74.7225, "grad_norm": 5.375, "learning_rate": 3.1681162715981895e-06, "loss": 1.3923, "step": 29889 }, { "epoch": 74.725, "grad_norm": 6.0625, "learning_rate": 3.1675250250553568e-06, "loss": 1.1871, "step": 29890 }, { "epoch": 74.7275, "grad_norm": 6.5625, "learning_rate": 3.16693382330596e-06, "loss": 1.307, "step": 29891 }, { "epoch": 74.73, "grad_norm": 6.46875, "learning_rate": 3.166342666353872e-06, "loss": 1.2301, "step": 29892 }, { "epoch": 74.7325, "grad_norm": 8.3125, "learning_rate": 3.1657515542029636e-06, "loss": 1.4416, "step": 29893 }, { "epoch": 74.735, "grad_norm": 7.5, "learning_rate": 3.1651604868571196e-06, "loss": 1.1969, "step": 29894 }, { "epoch": 74.7375, "grad_norm": 9.1875, "learning_rate": 3.164569464320205e-06, "loss": 1.2136, "step": 29895 }, { "epoch": 74.74, "grad_norm": 5.3125, "learning_rate": 3.163978486596103e-06, "loss": 1.2608, "step": 29896 }, { "epoch": 74.7425, "grad_norm": 7.0625, "learning_rate": 3.1633875536886815e-06, "loss": 1.3075, "step": 29897 }, { "epoch": 74.745, "grad_norm": 9.125, "learning_rate": 3.1627966656018204e-06, "loss": 1.2511, "step": 29898 }, { "epoch": 74.7475, "grad_norm": 6.78125, "learning_rate": 3.1622058223393914e-06, "loss": 1.2849, "step": 29899 }, { "epoch": 74.75, "grad_norm": 7.53125, "learning_rate": 3.1616150239052647e-06, "loss": 1.2779, "step": 29900 }, { "epoch": 74.7525, "grad_norm": 10.5, "learning_rate": 3.1610242703033158e-06, "loss": 1.2386, "step": 29901 }, { "epoch": 74.755, "grad_norm": 10.5625, "learning_rate": 3.1604335615374203e-06, "loss": 1.2541, "step": 29902 }, { "epoch": 74.7575, "grad_norm": 5.84375, "learning_rate": 3.1598428976114492e-06, "loss": 1.415, "step": 29903 }, { "epoch": 74.76, "grad_norm": 6.78125, "learning_rate": 3.1592522785292714e-06, "loss": 1.4571, "step": 29904 }, { "epoch": 74.7625, "grad_norm": 5.84375, "learning_rate": 3.158661704294762e-06, "loss": 1.4874, "step": 29905 }, { "epoch": 74.765, "grad_norm": 7.375, "learning_rate": 3.1580711749117955e-06, "loss": 1.2938, "step": 29906 }, { "epoch": 74.7675, "grad_norm": 7.3125, "learning_rate": 3.1574806903842416e-06, "loss": 1.2595, "step": 29907 }, { "epoch": 74.77, "grad_norm": 9.5, "learning_rate": 3.156890250715966e-06, "loss": 1.5634, "step": 29908 }, { "epoch": 74.7725, "grad_norm": 9.625, "learning_rate": 3.156299855910847e-06, "loss": 1.2556, "step": 29909 }, { "epoch": 74.775, "grad_norm": 7.46875, "learning_rate": 3.1557095059727506e-06, "loss": 1.2291, "step": 29910 }, { "epoch": 74.7775, "grad_norm": 8.6875, "learning_rate": 3.155119200905551e-06, "loss": 1.2952, "step": 29911 }, { "epoch": 74.78, "grad_norm": 9.4375, "learning_rate": 3.1545289407131128e-06, "loss": 1.5104, "step": 29912 }, { "epoch": 74.7825, "grad_norm": 6.5, "learning_rate": 3.1539387253993127e-06, "loss": 1.3942, "step": 29913 }, { "epoch": 74.785, "grad_norm": 11.125, "learning_rate": 3.153348554968015e-06, "loss": 1.3677, "step": 29914 }, { "epoch": 74.7875, "grad_norm": 6.78125, "learning_rate": 3.1527584294230883e-06, "loss": 1.2874, "step": 29915 }, { "epoch": 74.79, "grad_norm": 10.4375, "learning_rate": 3.152168348768403e-06, "loss": 1.3127, "step": 29916 }, { "epoch": 74.7925, "grad_norm": 6.15625, "learning_rate": 3.1515783130078313e-06, "loss": 1.0975, "step": 29917 }, { "epoch": 74.795, "grad_norm": 7.625, "learning_rate": 3.150988322145239e-06, "loss": 1.259, "step": 29918 }, { "epoch": 74.7975, "grad_norm": 8.0625, "learning_rate": 3.1503983761844914e-06, "loss": 1.3205, "step": 29919 }, { "epoch": 74.8, "grad_norm": 6.875, "learning_rate": 3.1498084751294523e-06, "loss": 1.3424, "step": 29920 }, { "epoch": 74.8025, "grad_norm": 7.875, "learning_rate": 3.1492186189840014e-06, "loss": 1.2654, "step": 29921 }, { "epoch": 74.805, "grad_norm": 10.375, "learning_rate": 3.148628807751999e-06, "loss": 1.4203, "step": 29922 }, { "epoch": 74.8075, "grad_norm": 14.3125, "learning_rate": 3.1480390414373106e-06, "loss": 1.2842, "step": 29923 }, { "epoch": 74.81, "grad_norm": 10.375, "learning_rate": 3.147449320043803e-06, "loss": 1.3977, "step": 29924 }, { "epoch": 74.8125, "grad_norm": 8.6875, "learning_rate": 3.1468596435753418e-06, "loss": 1.1809, "step": 29925 }, { "epoch": 74.815, "grad_norm": 8.625, "learning_rate": 3.146270012035798e-06, "loss": 1.2752, "step": 29926 }, { "epoch": 74.8175, "grad_norm": 7.96875, "learning_rate": 3.1456804254290318e-06, "loss": 1.337, "step": 29927 }, { "epoch": 74.82, "grad_norm": 7.3125, "learning_rate": 3.145090883758912e-06, "loss": 1.5135, "step": 29928 }, { "epoch": 74.8225, "grad_norm": 7.78125, "learning_rate": 3.144501387029303e-06, "loss": 1.2792, "step": 29929 }, { "epoch": 74.825, "grad_norm": 8.5625, "learning_rate": 3.143911935244065e-06, "loss": 1.2923, "step": 29930 }, { "epoch": 74.8275, "grad_norm": 9.3125, "learning_rate": 3.1433225284070667e-06, "loss": 1.3691, "step": 29931 }, { "epoch": 74.83, "grad_norm": 9.6875, "learning_rate": 3.142733166522174e-06, "loss": 1.2739, "step": 29932 }, { "epoch": 74.8325, "grad_norm": 8.75, "learning_rate": 3.1421438495932476e-06, "loss": 1.3267, "step": 29933 }, { "epoch": 74.835, "grad_norm": 9.125, "learning_rate": 3.1415545776241527e-06, "loss": 1.2411, "step": 29934 }, { "epoch": 74.8375, "grad_norm": 10.0, "learning_rate": 3.1409653506187452e-06, "loss": 1.4478, "step": 29935 }, { "epoch": 74.84, "grad_norm": 10.125, "learning_rate": 3.1403761685809007e-06, "loss": 1.1927, "step": 29936 }, { "epoch": 74.8425, "grad_norm": 7.53125, "learning_rate": 3.139787031514475e-06, "loss": 1.4412, "step": 29937 }, { "epoch": 74.845, "grad_norm": 7.9375, "learning_rate": 3.139197939423332e-06, "loss": 1.2574, "step": 29938 }, { "epoch": 74.8475, "grad_norm": 7.59375, "learning_rate": 3.1386088923113257e-06, "loss": 1.3315, "step": 29939 }, { "epoch": 74.85, "grad_norm": 7.875, "learning_rate": 3.1380198901823313e-06, "loss": 1.3301, "step": 29940 }, { "epoch": 74.8525, "grad_norm": 5.4375, "learning_rate": 3.1374309330402033e-06, "loss": 1.1964, "step": 29941 }, { "epoch": 74.855, "grad_norm": 6.0, "learning_rate": 3.136842020888804e-06, "loss": 1.2588, "step": 29942 }, { "epoch": 74.8575, "grad_norm": 7.875, "learning_rate": 3.1362531537319908e-06, "loss": 1.2358, "step": 29943 }, { "epoch": 74.86, "grad_norm": 5.9375, "learning_rate": 3.135664331573627e-06, "loss": 1.3754, "step": 29944 }, { "epoch": 74.8625, "grad_norm": 4.4375, "learning_rate": 3.135075554417576e-06, "loss": 1.19, "step": 29945 }, { "epoch": 74.865, "grad_norm": 6.75, "learning_rate": 3.134486822267696e-06, "loss": 1.1745, "step": 29946 }, { "epoch": 74.8675, "grad_norm": 4.03125, "learning_rate": 3.1338981351278417e-06, "loss": 1.2805, "step": 29947 }, { "epoch": 74.87, "grad_norm": 7.71875, "learning_rate": 3.1333094930018793e-06, "loss": 1.228, "step": 29948 }, { "epoch": 74.8725, "grad_norm": 8.25, "learning_rate": 3.1327208958936627e-06, "loss": 1.2457, "step": 29949 }, { "epoch": 74.875, "grad_norm": 7.0, "learning_rate": 3.132132343807056e-06, "loss": 1.3778, "step": 29950 }, { "epoch": 74.8775, "grad_norm": 7.71875, "learning_rate": 3.131543836745912e-06, "loss": 1.3178, "step": 29951 }, { "epoch": 74.88, "grad_norm": 7.3125, "learning_rate": 3.130955374714094e-06, "loss": 1.2675, "step": 29952 }, { "epoch": 74.8825, "grad_norm": 7.71875, "learning_rate": 3.1303669577154583e-06, "loss": 1.4026, "step": 29953 }, { "epoch": 74.885, "grad_norm": 7.8125, "learning_rate": 3.1297785857538553e-06, "loss": 1.5336, "step": 29954 }, { "epoch": 74.8875, "grad_norm": 7.5625, "learning_rate": 3.1291902588331556e-06, "loss": 1.3722, "step": 29955 }, { "epoch": 74.89, "grad_norm": 8.625, "learning_rate": 3.1286019769572095e-06, "loss": 1.4209, "step": 29956 }, { "epoch": 74.8925, "grad_norm": 7.59375, "learning_rate": 3.1280137401298737e-06, "loss": 1.2667, "step": 29957 }, { "epoch": 74.895, "grad_norm": 6.53125, "learning_rate": 3.127425548355002e-06, "loss": 1.4969, "step": 29958 }, { "epoch": 74.8975, "grad_norm": 6.96875, "learning_rate": 3.1268374016364533e-06, "loss": 1.268, "step": 29959 }, { "epoch": 74.9, "grad_norm": 6.84375, "learning_rate": 3.126249299978086e-06, "loss": 1.2191, "step": 29960 }, { "epoch": 74.9025, "grad_norm": 11.8125, "learning_rate": 3.125661243383754e-06, "loss": 1.2998, "step": 29961 }, { "epoch": 74.905, "grad_norm": 8.9375, "learning_rate": 3.1250732318573087e-06, "loss": 1.3914, "step": 29962 }, { "epoch": 74.9075, "grad_norm": 8.0625, "learning_rate": 3.12448526540261e-06, "loss": 1.2826, "step": 29963 }, { "epoch": 74.91, "grad_norm": 7.34375, "learning_rate": 3.1238973440235098e-06, "loss": 1.4915, "step": 29964 }, { "epoch": 74.9125, "grad_norm": 6.78125, "learning_rate": 3.123309467723865e-06, "loss": 1.304, "step": 29965 }, { "epoch": 74.915, "grad_norm": 7.28125, "learning_rate": 3.122721636507525e-06, "loss": 1.3437, "step": 29966 }, { "epoch": 74.9175, "grad_norm": 9.125, "learning_rate": 3.1221338503783505e-06, "loss": 1.4254, "step": 29967 }, { "epoch": 74.92, "grad_norm": 8.25, "learning_rate": 3.1215461093401913e-06, "loss": 1.4311, "step": 29968 }, { "epoch": 74.9225, "grad_norm": 8.3125, "learning_rate": 3.1209584133968973e-06, "loss": 1.2855, "step": 29969 }, { "epoch": 74.925, "grad_norm": 5.5, "learning_rate": 3.120370762552325e-06, "loss": 1.4767, "step": 29970 }, { "epoch": 74.9275, "grad_norm": 5.875, "learning_rate": 3.1197831568103285e-06, "loss": 1.2675, "step": 29971 }, { "epoch": 74.93, "grad_norm": 7.3125, "learning_rate": 3.11919559617476e-06, "loss": 1.2927, "step": 29972 }, { "epoch": 74.9325, "grad_norm": 11.375, "learning_rate": 3.1186080806494657e-06, "loss": 1.2993, "step": 29973 }, { "epoch": 74.935, "grad_norm": 6.90625, "learning_rate": 3.1180206102383013e-06, "loss": 1.2971, "step": 29974 }, { "epoch": 74.9375, "grad_norm": 8.0625, "learning_rate": 3.117433184945121e-06, "loss": 1.4397, "step": 29975 }, { "epoch": 74.94, "grad_norm": 9.375, "learning_rate": 3.116845804773775e-06, "loss": 1.2783, "step": 29976 }, { "epoch": 74.9425, "grad_norm": 6.65625, "learning_rate": 3.1162584697281084e-06, "loss": 1.291, "step": 29977 }, { "epoch": 74.945, "grad_norm": 8.4375, "learning_rate": 3.1156711798119787e-06, "loss": 1.3585, "step": 29978 }, { "epoch": 74.9475, "grad_norm": 10.1875, "learning_rate": 3.1150839350292306e-06, "loss": 1.1969, "step": 29979 }, { "epoch": 74.95, "grad_norm": 5.59375, "learning_rate": 3.1144967353837196e-06, "loss": 1.256, "step": 29980 }, { "epoch": 74.9525, "grad_norm": 6.28125, "learning_rate": 3.1139095808792884e-06, "loss": 1.4107, "step": 29981 }, { "epoch": 74.955, "grad_norm": 8.0, "learning_rate": 3.113322471519795e-06, "loss": 1.3567, "step": 29982 }, { "epoch": 74.9575, "grad_norm": 6.28125, "learning_rate": 3.1127354073090832e-06, "loss": 1.3885, "step": 29983 }, { "epoch": 74.96, "grad_norm": 12.125, "learning_rate": 3.1121483882509996e-06, "loss": 1.2417, "step": 29984 }, { "epoch": 74.9625, "grad_norm": 5.34375, "learning_rate": 3.1115614143493956e-06, "loss": 1.347, "step": 29985 }, { "epoch": 74.965, "grad_norm": 6.65625, "learning_rate": 3.1109744856081227e-06, "loss": 1.3484, "step": 29986 }, { "epoch": 74.9675, "grad_norm": 6.46875, "learning_rate": 3.1103876020310243e-06, "loss": 1.2052, "step": 29987 }, { "epoch": 74.97, "grad_norm": 7.96875, "learning_rate": 3.1098007636219464e-06, "loss": 1.4264, "step": 29988 }, { "epoch": 74.9725, "grad_norm": 7.125, "learning_rate": 3.109213970384739e-06, "loss": 1.2052, "step": 29989 }, { "epoch": 74.975, "grad_norm": 6.6875, "learning_rate": 3.108627222323253e-06, "loss": 1.3823, "step": 29990 }, { "epoch": 74.9775, "grad_norm": 8.3125, "learning_rate": 3.10804051944133e-06, "loss": 1.2752, "step": 29991 }, { "epoch": 74.98, "grad_norm": 7.9375, "learning_rate": 3.107453861742815e-06, "loss": 1.2095, "step": 29992 }, { "epoch": 74.9825, "grad_norm": 6.125, "learning_rate": 3.10686724923156e-06, "loss": 1.3134, "step": 29993 }, { "epoch": 74.985, "grad_norm": 6.625, "learning_rate": 3.1062806819114044e-06, "loss": 1.2851, "step": 29994 }, { "epoch": 74.9875, "grad_norm": 6.84375, "learning_rate": 3.105694159786199e-06, "loss": 1.3169, "step": 29995 }, { "epoch": 74.99, "grad_norm": 6.8125, "learning_rate": 3.105107682859785e-06, "loss": 1.2552, "step": 29996 }, { "epoch": 74.9925, "grad_norm": 7.1875, "learning_rate": 3.1045212511360113e-06, "loss": 1.541, "step": 29997 }, { "epoch": 74.995, "grad_norm": 9.8125, "learning_rate": 3.1039348646187206e-06, "loss": 1.3861, "step": 29998 }, { "epoch": 74.9975, "grad_norm": 8.25, "learning_rate": 3.1033485233117534e-06, "loss": 1.2272, "step": 29999 }, { "epoch": 75.0, "grad_norm": 9.5625, "learning_rate": 3.1027622272189572e-06, "loss": 1.2089, "step": 30000 }, { "epoch": 75.0025, "grad_norm": 8.8125, "learning_rate": 3.1021759763441794e-06, "loss": 1.2825, "step": 30001 }, { "epoch": 75.005, "grad_norm": 8.75, "learning_rate": 3.1015897706912602e-06, "loss": 1.3942, "step": 30002 }, { "epoch": 75.0075, "grad_norm": 6.15625, "learning_rate": 3.1010036102640386e-06, "loss": 1.5096, "step": 30003 }, { "epoch": 75.01, "grad_norm": 9.0, "learning_rate": 3.1004174950663613e-06, "loss": 1.3775, "step": 30004 }, { "epoch": 75.0125, "grad_norm": 9.5, "learning_rate": 3.0998314251020746e-06, "loss": 1.3363, "step": 30005 }, { "epoch": 75.015, "grad_norm": 8.0625, "learning_rate": 3.099245400375016e-06, "loss": 1.4156, "step": 30006 }, { "epoch": 75.0175, "grad_norm": 9.0, "learning_rate": 3.0986594208890266e-06, "loss": 1.2521, "step": 30007 }, { "epoch": 75.02, "grad_norm": 9.3125, "learning_rate": 3.0980734866479525e-06, "loss": 1.3509, "step": 30008 }, { "epoch": 75.0225, "grad_norm": 6.4375, "learning_rate": 3.0974875976556283e-06, "loss": 1.2408, "step": 30009 }, { "epoch": 75.025, "grad_norm": 7.59375, "learning_rate": 3.096901753915903e-06, "loss": 1.2649, "step": 30010 }, { "epoch": 75.0275, "grad_norm": 8.25, "learning_rate": 3.0963159554326105e-06, "loss": 1.3531, "step": 30011 }, { "epoch": 75.03, "grad_norm": 8.1875, "learning_rate": 3.095730202209597e-06, "loss": 1.4133, "step": 30012 }, { "epoch": 75.0325, "grad_norm": 10.1875, "learning_rate": 3.0951444942507015e-06, "loss": 1.3472, "step": 30013 }, { "epoch": 75.035, "grad_norm": 7.9375, "learning_rate": 3.0945588315597573e-06, "loss": 1.414, "step": 30014 }, { "epoch": 75.0375, "grad_norm": 7.1875, "learning_rate": 3.09397321414061e-06, "loss": 1.2594, "step": 30015 }, { "epoch": 75.04, "grad_norm": 8.375, "learning_rate": 3.0933876419971008e-06, "loss": 1.3497, "step": 30016 }, { "epoch": 75.0425, "grad_norm": 6.125, "learning_rate": 3.0928021151330668e-06, "loss": 1.3699, "step": 30017 }, { "epoch": 75.045, "grad_norm": 7.4375, "learning_rate": 3.0922166335523416e-06, "loss": 1.4812, "step": 30018 }, { "epoch": 75.0475, "grad_norm": 8.4375, "learning_rate": 3.0916311972587675e-06, "loss": 1.2297, "step": 30019 }, { "epoch": 75.05, "grad_norm": 6.4375, "learning_rate": 3.0910458062561865e-06, "loss": 1.6288, "step": 30020 }, { "epoch": 75.0525, "grad_norm": 8.625, "learning_rate": 3.0904604605484323e-06, "loss": 1.3048, "step": 30021 }, { "epoch": 75.055, "grad_norm": 8.5, "learning_rate": 3.0898751601393407e-06, "loss": 1.3403, "step": 30022 }, { "epoch": 75.0575, "grad_norm": 9.875, "learning_rate": 3.089289905032753e-06, "loss": 1.253, "step": 30023 }, { "epoch": 75.06, "grad_norm": 6.21875, "learning_rate": 3.0887046952325016e-06, "loss": 1.2699, "step": 30024 }, { "epoch": 75.0625, "grad_norm": 10.4375, "learning_rate": 3.0881195307424282e-06, "loss": 1.2772, "step": 30025 }, { "epoch": 75.065, "grad_norm": 11.125, "learning_rate": 3.087534411566364e-06, "loss": 1.3463, "step": 30026 }, { "epoch": 75.0675, "grad_norm": 7.90625, "learning_rate": 3.0869493377081506e-06, "loss": 1.4027, "step": 30027 }, { "epoch": 75.07, "grad_norm": 6.375, "learning_rate": 3.086364309171621e-06, "loss": 1.4029, "step": 30028 }, { "epoch": 75.0725, "grad_norm": 8.75, "learning_rate": 3.085779325960606e-06, "loss": 1.1686, "step": 30029 }, { "epoch": 75.075, "grad_norm": 8.1875, "learning_rate": 3.0851943880789457e-06, "loss": 1.4467, "step": 30030 }, { "epoch": 75.0775, "grad_norm": 6.5625, "learning_rate": 3.0846094955304784e-06, "loss": 1.2939, "step": 30031 }, { "epoch": 75.08, "grad_norm": 7.125, "learning_rate": 3.0840246483190338e-06, "loss": 1.2882, "step": 30032 }, { "epoch": 75.0825, "grad_norm": 7.0625, "learning_rate": 3.083439846448443e-06, "loss": 1.317, "step": 30033 }, { "epoch": 75.085, "grad_norm": 4.875, "learning_rate": 3.082855089922544e-06, "loss": 1.3896, "step": 30034 }, { "epoch": 75.0875, "grad_norm": 6.125, "learning_rate": 3.0822703787451747e-06, "loss": 1.2502, "step": 30035 }, { "epoch": 75.09, "grad_norm": 9.625, "learning_rate": 3.0816857129201627e-06, "loss": 1.3126, "step": 30036 }, { "epoch": 75.0925, "grad_norm": 5.25, "learning_rate": 3.0811010924513396e-06, "loss": 1.1912, "step": 30037 }, { "epoch": 75.095, "grad_norm": 5.84375, "learning_rate": 3.080516517342543e-06, "loss": 1.4365, "step": 30038 }, { "epoch": 75.0975, "grad_norm": 8.0, "learning_rate": 3.079931987597602e-06, "loss": 1.1928, "step": 30039 }, { "epoch": 75.1, "grad_norm": 6.625, "learning_rate": 3.0793475032203513e-06, "loss": 1.1828, "step": 30040 }, { "epoch": 75.1025, "grad_norm": 8.5625, "learning_rate": 3.0787630642146194e-06, "loss": 1.5563, "step": 30041 }, { "epoch": 75.105, "grad_norm": 9.0, "learning_rate": 3.078178670584242e-06, "loss": 1.3525, "step": 30042 }, { "epoch": 75.1075, "grad_norm": 7.5625, "learning_rate": 3.0775943223330495e-06, "loss": 1.4732, "step": 30043 }, { "epoch": 75.11, "grad_norm": 9.125, "learning_rate": 3.0770100194648677e-06, "loss": 1.3157, "step": 30044 }, { "epoch": 75.1125, "grad_norm": 7.71875, "learning_rate": 3.0764257619835315e-06, "loss": 1.3238, "step": 30045 }, { "epoch": 75.115, "grad_norm": 7.40625, "learning_rate": 3.075841549892874e-06, "loss": 1.2665, "step": 30046 }, { "epoch": 75.1175, "grad_norm": 6.4375, "learning_rate": 3.0752573831967214e-06, "loss": 1.3459, "step": 30047 }, { "epoch": 75.12, "grad_norm": 6.3125, "learning_rate": 3.074673261898903e-06, "loss": 1.3568, "step": 30048 }, { "epoch": 75.1225, "grad_norm": 7.84375, "learning_rate": 3.0740891860032474e-06, "loss": 1.3705, "step": 30049 }, { "epoch": 75.125, "grad_norm": 7.25, "learning_rate": 3.073505155513591e-06, "loss": 1.3048, "step": 30050 }, { "epoch": 75.1275, "grad_norm": 7.59375, "learning_rate": 3.072921170433756e-06, "loss": 1.2957, "step": 30051 }, { "epoch": 75.13, "grad_norm": 9.5625, "learning_rate": 3.07233723076757e-06, "loss": 1.3638, "step": 30052 }, { "epoch": 75.1325, "grad_norm": 8.5, "learning_rate": 3.071753336518868e-06, "loss": 1.2078, "step": 30053 }, { "epoch": 75.135, "grad_norm": 8.375, "learning_rate": 3.071169487691469e-06, "loss": 1.386, "step": 30054 }, { "epoch": 75.1375, "grad_norm": 7.34375, "learning_rate": 3.07058568428921e-06, "loss": 1.3181, "step": 30055 }, { "epoch": 75.14, "grad_norm": 7.53125, "learning_rate": 3.070001926315911e-06, "loss": 1.3758, "step": 30056 }, { "epoch": 75.1425, "grad_norm": 9.75, "learning_rate": 3.0694182137754037e-06, "loss": 1.2404, "step": 30057 }, { "epoch": 75.145, "grad_norm": 10.3125, "learning_rate": 3.068834546671514e-06, "loss": 1.2269, "step": 30058 }, { "epoch": 75.1475, "grad_norm": 7.53125, "learning_rate": 3.068250925008065e-06, "loss": 1.266, "step": 30059 }, { "epoch": 75.15, "grad_norm": 8.5, "learning_rate": 3.0676673487888854e-06, "loss": 1.1771, "step": 30060 }, { "epoch": 75.1525, "grad_norm": 6.0, "learning_rate": 3.067083818017804e-06, "loss": 1.2992, "step": 30061 }, { "epoch": 75.155, "grad_norm": 7.3125, "learning_rate": 3.0665003326986443e-06, "loss": 1.3629, "step": 30062 }, { "epoch": 75.1575, "grad_norm": 7.1875, "learning_rate": 3.06591689283523e-06, "loss": 1.2581, "step": 30063 }, { "epoch": 75.16, "grad_norm": 10.0, "learning_rate": 3.065333498431381e-06, "loss": 1.3862, "step": 30064 }, { "epoch": 75.1625, "grad_norm": 8.0625, "learning_rate": 3.064750149490935e-06, "loss": 1.4004, "step": 30065 }, { "epoch": 75.165, "grad_norm": 5.90625, "learning_rate": 3.064166846017709e-06, "loss": 1.3212, "step": 30066 }, { "epoch": 75.1675, "grad_norm": 8.4375, "learning_rate": 3.0635835880155252e-06, "loss": 1.3093, "step": 30067 }, { "epoch": 75.17, "grad_norm": 7.875, "learning_rate": 3.063000375488212e-06, "loss": 1.2691, "step": 30068 }, { "epoch": 75.1725, "grad_norm": 11.5, "learning_rate": 3.0624172084395875e-06, "loss": 1.3691, "step": 30069 }, { "epoch": 75.175, "grad_norm": 9.75, "learning_rate": 3.061834086873482e-06, "loss": 1.3091, "step": 30070 }, { "epoch": 75.1775, "grad_norm": 9.0625, "learning_rate": 3.061251010793711e-06, "loss": 1.3321, "step": 30071 }, { "epoch": 75.18, "grad_norm": 14.6875, "learning_rate": 3.060667980204104e-06, "loss": 1.2912, "step": 30072 }, { "epoch": 75.1825, "grad_norm": 9.8125, "learning_rate": 3.060084995108481e-06, "loss": 1.3317, "step": 30073 }, { "epoch": 75.185, "grad_norm": 8.9375, "learning_rate": 3.059502055510658e-06, "loss": 1.2912, "step": 30074 }, { "epoch": 75.1875, "grad_norm": 6.375, "learning_rate": 3.058919161414463e-06, "loss": 1.411, "step": 30075 }, { "epoch": 75.19, "grad_norm": 7.96875, "learning_rate": 3.058336312823719e-06, "loss": 1.3341, "step": 30076 }, { "epoch": 75.1925, "grad_norm": 7.40625, "learning_rate": 3.0577535097422452e-06, "loss": 1.2576, "step": 30077 }, { "epoch": 75.195, "grad_norm": 4.96875, "learning_rate": 3.05717075217386e-06, "loss": 1.3176, "step": 30078 }, { "epoch": 75.1975, "grad_norm": 9.25, "learning_rate": 3.05658804012238e-06, "loss": 1.4562, "step": 30079 }, { "epoch": 75.2, "grad_norm": 8.0625, "learning_rate": 3.0560053735916372e-06, "loss": 1.2466, "step": 30080 }, { "epoch": 75.2025, "grad_norm": 6.78125, "learning_rate": 3.055422752585445e-06, "loss": 1.2056, "step": 30081 }, { "epoch": 75.205, "grad_norm": 9.25, "learning_rate": 3.0548401771076243e-06, "loss": 1.2468, "step": 30082 }, { "epoch": 75.2075, "grad_norm": 7.3125, "learning_rate": 3.05425764716199e-06, "loss": 1.392, "step": 30083 }, { "epoch": 75.21, "grad_norm": 8.375, "learning_rate": 3.0536751627523643e-06, "loss": 1.3805, "step": 30084 }, { "epoch": 75.2125, "grad_norm": 8.5, "learning_rate": 3.0530927238825694e-06, "loss": 1.3008, "step": 30085 }, { "epoch": 75.215, "grad_norm": 8.0625, "learning_rate": 3.0525103305564206e-06, "loss": 1.3167, "step": 30086 }, { "epoch": 75.2175, "grad_norm": 6.09375, "learning_rate": 3.0519279827777325e-06, "loss": 1.2926, "step": 30087 }, { "epoch": 75.22, "grad_norm": 7.21875, "learning_rate": 3.0513456805503305e-06, "loss": 1.3456, "step": 30088 }, { "epoch": 75.2225, "grad_norm": 11.25, "learning_rate": 3.0507634238780247e-06, "loss": 1.256, "step": 30089 }, { "epoch": 75.225, "grad_norm": 9.1875, "learning_rate": 3.050181212764639e-06, "loss": 1.2953, "step": 30090 }, { "epoch": 75.2275, "grad_norm": 7.96875, "learning_rate": 3.0495990472139834e-06, "loss": 1.3623, "step": 30091 }, { "epoch": 75.23, "grad_norm": 7.21875, "learning_rate": 3.049016927229882e-06, "loss": 1.2914, "step": 30092 }, { "epoch": 75.2325, "grad_norm": 8.5625, "learning_rate": 3.048434852816147e-06, "loss": 1.4267, "step": 30093 }, { "epoch": 75.235, "grad_norm": 8.8125, "learning_rate": 3.0478528239765893e-06, "loss": 1.2123, "step": 30094 }, { "epoch": 75.2375, "grad_norm": 5.40625, "learning_rate": 3.0472708407150365e-06, "loss": 1.2242, "step": 30095 }, { "epoch": 75.24, "grad_norm": 5.96875, "learning_rate": 3.0466889030352976e-06, "loss": 1.2082, "step": 30096 }, { "epoch": 75.2425, "grad_norm": 6.9375, "learning_rate": 3.046107010941187e-06, "loss": 1.2638, "step": 30097 }, { "epoch": 75.245, "grad_norm": 12.625, "learning_rate": 3.045525164436519e-06, "loss": 1.329, "step": 30098 }, { "epoch": 75.2475, "grad_norm": 10.4375, "learning_rate": 3.0449433635251093e-06, "loss": 1.3473, "step": 30099 }, { "epoch": 75.25, "grad_norm": 12.875, "learning_rate": 3.0443616082107753e-06, "loss": 1.3322, "step": 30100 }, { "epoch": 75.2525, "grad_norm": 6.03125, "learning_rate": 3.0437798984973287e-06, "loss": 1.3104, "step": 30101 }, { "epoch": 75.255, "grad_norm": 9.1875, "learning_rate": 3.043198234388579e-06, "loss": 1.3375, "step": 30102 }, { "epoch": 75.2575, "grad_norm": 6.5, "learning_rate": 3.0426166158883475e-06, "loss": 1.3898, "step": 30103 }, { "epoch": 75.26, "grad_norm": 6.46875, "learning_rate": 3.0420350430004388e-06, "loss": 1.6091, "step": 30104 }, { "epoch": 75.2625, "grad_norm": 8.9375, "learning_rate": 3.041453515728673e-06, "loss": 1.2665, "step": 30105 }, { "epoch": 75.265, "grad_norm": 6.96875, "learning_rate": 3.040872034076857e-06, "loss": 1.2873, "step": 30106 }, { "epoch": 75.2675, "grad_norm": 6.59375, "learning_rate": 3.0402905980488085e-06, "loss": 1.2722, "step": 30107 }, { "epoch": 75.27, "grad_norm": 6.90625, "learning_rate": 3.039709207648336e-06, "loss": 1.3171, "step": 30108 }, { "epoch": 75.2725, "grad_norm": 7.9375, "learning_rate": 3.0391278628792485e-06, "loss": 1.4587, "step": 30109 }, { "epoch": 75.275, "grad_norm": 9.25, "learning_rate": 3.0385465637453604e-06, "loss": 1.313, "step": 30110 }, { "epoch": 75.2775, "grad_norm": 6.375, "learning_rate": 3.0379653102504847e-06, "loss": 1.3916, "step": 30111 }, { "epoch": 75.28, "grad_norm": 8.1875, "learning_rate": 3.037384102398431e-06, "loss": 1.4148, "step": 30112 }, { "epoch": 75.2825, "grad_norm": 7.8125, "learning_rate": 3.0368029401930045e-06, "loss": 1.3896, "step": 30113 }, { "epoch": 75.285, "grad_norm": 6.15625, "learning_rate": 3.036221823638019e-06, "loss": 1.2935, "step": 30114 }, { "epoch": 75.2875, "grad_norm": 8.25, "learning_rate": 3.035640752737289e-06, "loss": 1.3803, "step": 30115 }, { "epoch": 75.29, "grad_norm": 10.875, "learning_rate": 3.035059727494619e-06, "loss": 1.3263, "step": 30116 }, { "epoch": 75.2925, "grad_norm": 7.15625, "learning_rate": 3.0344787479138147e-06, "loss": 1.3266, "step": 30117 }, { "epoch": 75.295, "grad_norm": 10.5, "learning_rate": 3.033897813998693e-06, "loss": 1.4306, "step": 30118 }, { "epoch": 75.2975, "grad_norm": 6.46875, "learning_rate": 3.0333169257530548e-06, "loss": 1.1599, "step": 30119 }, { "epoch": 75.3, "grad_norm": 10.0, "learning_rate": 3.032736083180716e-06, "loss": 1.2514, "step": 30120 }, { "epoch": 75.3025, "grad_norm": 7.21875, "learning_rate": 3.0321552862854774e-06, "loss": 1.3349, "step": 30121 }, { "epoch": 75.305, "grad_norm": 9.1875, "learning_rate": 3.0315745350711524e-06, "loss": 1.2065, "step": 30122 }, { "epoch": 75.3075, "grad_norm": 8.375, "learning_rate": 3.0309938295415465e-06, "loss": 1.4298, "step": 30123 }, { "epoch": 75.31, "grad_norm": 7.59375, "learning_rate": 3.0304131697004624e-06, "loss": 1.4437, "step": 30124 }, { "epoch": 75.3125, "grad_norm": 6.28125, "learning_rate": 3.0298325555517105e-06, "loss": 1.3936, "step": 30125 }, { "epoch": 75.315, "grad_norm": 7.96875, "learning_rate": 3.029251987099101e-06, "loss": 1.2574, "step": 30126 }, { "epoch": 75.3175, "grad_norm": 9.875, "learning_rate": 3.0286714643464366e-06, "loss": 1.2799, "step": 30127 }, { "epoch": 75.32, "grad_norm": 6.3125, "learning_rate": 3.0280909872975194e-06, "loss": 1.2184, "step": 30128 }, { "epoch": 75.3225, "grad_norm": 7.34375, "learning_rate": 3.0275105559561593e-06, "loss": 1.3844, "step": 30129 }, { "epoch": 75.325, "grad_norm": 6.375, "learning_rate": 3.0269301703261646e-06, "loss": 1.3497, "step": 30130 }, { "epoch": 75.3275, "grad_norm": 6.65625, "learning_rate": 3.026349830411336e-06, "loss": 1.199, "step": 30131 }, { "epoch": 75.33, "grad_norm": 6.09375, "learning_rate": 3.025769536215475e-06, "loss": 1.4403, "step": 30132 }, { "epoch": 75.3325, "grad_norm": 7.6875, "learning_rate": 3.025189287742394e-06, "loss": 1.2799, "step": 30133 }, { "epoch": 75.335, "grad_norm": 5.75, "learning_rate": 3.02460908499589e-06, "loss": 1.3522, "step": 30134 }, { "epoch": 75.3375, "grad_norm": 8.6875, "learning_rate": 3.0240289279797717e-06, "loss": 1.4095, "step": 30135 }, { "epoch": 75.34, "grad_norm": 9.9375, "learning_rate": 3.023448816697838e-06, "loss": 1.2241, "step": 30136 }, { "epoch": 75.3425, "grad_norm": 9.375, "learning_rate": 3.022868751153898e-06, "loss": 1.2572, "step": 30137 }, { "epoch": 75.345, "grad_norm": 8.3125, "learning_rate": 3.0222887313517513e-06, "loss": 1.2856, "step": 30138 }, { "epoch": 75.3475, "grad_norm": 7.21875, "learning_rate": 3.0217087572951966e-06, "loss": 1.4827, "step": 30139 }, { "epoch": 75.35, "grad_norm": 10.1875, "learning_rate": 3.0211288289880404e-06, "loss": 1.4045, "step": 30140 }, { "epoch": 75.3525, "grad_norm": 7.375, "learning_rate": 3.0205489464340875e-06, "loss": 1.3227, "step": 30141 }, { "epoch": 75.355, "grad_norm": 9.8125, "learning_rate": 3.0199691096371364e-06, "loss": 1.4211, "step": 30142 }, { "epoch": 75.3575, "grad_norm": 8.0625, "learning_rate": 3.019389318600986e-06, "loss": 1.477, "step": 30143 }, { "epoch": 75.36, "grad_norm": 9.875, "learning_rate": 3.0188095733294388e-06, "loss": 1.3689, "step": 30144 }, { "epoch": 75.3625, "grad_norm": 7.09375, "learning_rate": 3.0182298738263006e-06, "loss": 1.3738, "step": 30145 }, { "epoch": 75.365, "grad_norm": 7.25, "learning_rate": 3.0176502200953685e-06, "loss": 1.3717, "step": 30146 }, { "epoch": 75.3675, "grad_norm": 5.375, "learning_rate": 3.017070612140438e-06, "loss": 1.463, "step": 30147 }, { "epoch": 75.37, "grad_norm": 7.0625, "learning_rate": 3.0164910499653155e-06, "loss": 1.2958, "step": 30148 }, { "epoch": 75.3725, "grad_norm": 8.5625, "learning_rate": 3.0159115335737966e-06, "loss": 1.3251, "step": 30149 }, { "epoch": 75.375, "grad_norm": 9.375, "learning_rate": 3.015332062969685e-06, "loss": 1.4345, "step": 30150 }, { "epoch": 75.3775, "grad_norm": 9.6875, "learning_rate": 3.0147526381567725e-06, "loss": 1.3451, "step": 30151 }, { "epoch": 75.38, "grad_norm": 9.1875, "learning_rate": 3.014173259138867e-06, "loss": 1.2054, "step": 30152 }, { "epoch": 75.3825, "grad_norm": 6.46875, "learning_rate": 3.013593925919761e-06, "loss": 1.3687, "step": 30153 }, { "epoch": 75.385, "grad_norm": 7.34375, "learning_rate": 3.0130146385032512e-06, "loss": 1.4679, "step": 30154 }, { "epoch": 75.3875, "grad_norm": 9.375, "learning_rate": 3.0124353968931376e-06, "loss": 1.4237, "step": 30155 }, { "epoch": 75.39, "grad_norm": 6.6875, "learning_rate": 3.011856201093221e-06, "loss": 1.3685, "step": 30156 }, { "epoch": 75.3925, "grad_norm": 9.25, "learning_rate": 3.011277051107295e-06, "loss": 1.3584, "step": 30157 }, { "epoch": 75.395, "grad_norm": 6.21875, "learning_rate": 3.010697946939155e-06, "loss": 1.302, "step": 30158 }, { "epoch": 75.3975, "grad_norm": 6.375, "learning_rate": 3.010118888592598e-06, "loss": 1.2628, "step": 30159 }, { "epoch": 75.4, "grad_norm": 6.875, "learning_rate": 3.009539876071427e-06, "loss": 1.17, "step": 30160 }, { "epoch": 75.4025, "grad_norm": 10.1875, "learning_rate": 3.0089609093794312e-06, "loss": 1.4183, "step": 30161 }, { "epoch": 75.405, "grad_norm": 11.8125, "learning_rate": 3.0083819885204045e-06, "loss": 1.401, "step": 30162 }, { "epoch": 75.4075, "grad_norm": 9.875, "learning_rate": 3.00780311349815e-06, "loss": 1.2598, "step": 30163 }, { "epoch": 75.41, "grad_norm": 7.21875, "learning_rate": 3.0072242843164555e-06, "loss": 1.3521, "step": 30164 }, { "epoch": 75.4125, "grad_norm": 8.1875, "learning_rate": 3.0066455009791206e-06, "loss": 1.2821, "step": 30165 }, { "epoch": 75.415, "grad_norm": 14.375, "learning_rate": 3.0060667634899354e-06, "loss": 1.253, "step": 30166 }, { "epoch": 75.4175, "grad_norm": 6.78125, "learning_rate": 3.0054880718527003e-06, "loss": 1.3708, "step": 30167 }, { "epoch": 75.42, "grad_norm": 11.25, "learning_rate": 3.0049094260712053e-06, "loss": 1.1725, "step": 30168 }, { "epoch": 75.4225, "grad_norm": 7.0, "learning_rate": 3.004330826149241e-06, "loss": 1.3804, "step": 30169 }, { "epoch": 75.425, "grad_norm": 6.75, "learning_rate": 3.0037522720906033e-06, "loss": 1.4098, "step": 30170 }, { "epoch": 75.4275, "grad_norm": 7.1875, "learning_rate": 3.0031737638990887e-06, "loss": 1.2111, "step": 30171 }, { "epoch": 75.43, "grad_norm": 8.625, "learning_rate": 3.002595301578488e-06, "loss": 1.2414, "step": 30172 }, { "epoch": 75.4325, "grad_norm": 32.0, "learning_rate": 3.002016885132588e-06, "loss": 1.3288, "step": 30173 }, { "epoch": 75.435, "grad_norm": 6.5625, "learning_rate": 3.001438514565187e-06, "loss": 1.3238, "step": 30174 }, { "epoch": 75.4375, "grad_norm": 8.4375, "learning_rate": 3.0008601898800772e-06, "loss": 1.3443, "step": 30175 }, { "epoch": 75.44, "grad_norm": 9.6875, "learning_rate": 3.0002819110810475e-06, "loss": 1.4656, "step": 30176 }, { "epoch": 75.4425, "grad_norm": 8.1875, "learning_rate": 2.999703678171887e-06, "loss": 1.3109, "step": 30177 }, { "epoch": 75.445, "grad_norm": 7.96875, "learning_rate": 2.9991254911563915e-06, "loss": 1.5384, "step": 30178 }, { "epoch": 75.4475, "grad_norm": 6.0, "learning_rate": 2.9985473500383467e-06, "loss": 1.387, "step": 30179 }, { "epoch": 75.45, "grad_norm": 6.5625, "learning_rate": 2.997969254821548e-06, "loss": 1.3375, "step": 30180 }, { "epoch": 75.4525, "grad_norm": 6.59375, "learning_rate": 2.9973912055097796e-06, "loss": 1.2371, "step": 30181 }, { "epoch": 75.455, "grad_norm": 8.5625, "learning_rate": 2.9968132021068376e-06, "loss": 1.3826, "step": 30182 }, { "epoch": 75.4575, "grad_norm": 8.6875, "learning_rate": 2.996235244616504e-06, "loss": 1.3718, "step": 30183 }, { "epoch": 75.46, "grad_norm": 11.1875, "learning_rate": 2.9956573330425766e-06, "loss": 1.227, "step": 30184 }, { "epoch": 75.4625, "grad_norm": 7.21875, "learning_rate": 2.9950794673888352e-06, "loss": 1.2488, "step": 30185 }, { "epoch": 75.465, "grad_norm": 4.5625, "learning_rate": 2.994501647659076e-06, "loss": 1.4989, "step": 30186 }, { "epoch": 75.4675, "grad_norm": 7.25, "learning_rate": 2.993923873857083e-06, "loss": 1.2006, "step": 30187 }, { "epoch": 75.47, "grad_norm": 17.5, "learning_rate": 2.993346145986642e-06, "loss": 1.2653, "step": 30188 }, { "epoch": 75.4725, "grad_norm": 7.9375, "learning_rate": 2.992768464051543e-06, "loss": 1.3716, "step": 30189 }, { "epoch": 75.475, "grad_norm": 6.4375, "learning_rate": 2.9921908280555777e-06, "loss": 1.3329, "step": 30190 }, { "epoch": 75.4775, "grad_norm": 7.65625, "learning_rate": 2.9916132380025274e-06, "loss": 1.2836, "step": 30191 }, { "epoch": 75.48, "grad_norm": 7.6875, "learning_rate": 2.9910356938961782e-06, "loss": 1.1754, "step": 30192 }, { "epoch": 75.4825, "grad_norm": 8.5625, "learning_rate": 2.9904581957403187e-06, "loss": 1.4488, "step": 30193 }, { "epoch": 75.485, "grad_norm": 8.125, "learning_rate": 2.9898807435387376e-06, "loss": 1.5441, "step": 30194 }, { "epoch": 75.4875, "grad_norm": 6.5625, "learning_rate": 2.9893033372952173e-06, "loss": 1.2316, "step": 30195 }, { "epoch": 75.49, "grad_norm": 6.03125, "learning_rate": 2.9887259770135415e-06, "loss": 1.3003, "step": 30196 }, { "epoch": 75.4925, "grad_norm": 6.5625, "learning_rate": 2.9881486626975e-06, "loss": 1.3447, "step": 30197 }, { "epoch": 75.495, "grad_norm": 10.75, "learning_rate": 2.987571394350872e-06, "loss": 1.2387, "step": 30198 }, { "epoch": 75.4975, "grad_norm": 8.625, "learning_rate": 2.9869941719774486e-06, "loss": 1.3872, "step": 30199 }, { "epoch": 75.5, "grad_norm": 7.84375, "learning_rate": 2.9864169955810085e-06, "loss": 1.3262, "step": 30200 }, { "epoch": 75.5025, "grad_norm": 6.09375, "learning_rate": 2.98583986516534e-06, "loss": 1.4315, "step": 30201 }, { "epoch": 75.505, "grad_norm": 5.96875, "learning_rate": 2.985262780734225e-06, "loss": 1.2988, "step": 30202 }, { "epoch": 75.5075, "grad_norm": 11.0625, "learning_rate": 2.9846857422914434e-06, "loss": 1.2507, "step": 30203 }, { "epoch": 75.51, "grad_norm": 6.75, "learning_rate": 2.9841087498407817e-06, "loss": 1.3293, "step": 30204 }, { "epoch": 75.5125, "grad_norm": 6.125, "learning_rate": 2.983531803386025e-06, "loss": 1.1866, "step": 30205 }, { "epoch": 75.515, "grad_norm": 8.875, "learning_rate": 2.9829549029309525e-06, "loss": 1.4212, "step": 30206 }, { "epoch": 75.5175, "grad_norm": 6.96875, "learning_rate": 2.982378048479344e-06, "loss": 1.2962, "step": 30207 }, { "epoch": 75.52, "grad_norm": 9.6875, "learning_rate": 2.981801240034985e-06, "loss": 1.2509, "step": 30208 }, { "epoch": 75.5225, "grad_norm": 8.4375, "learning_rate": 2.9812244776016586e-06, "loss": 1.345, "step": 30209 }, { "epoch": 75.525, "grad_norm": 5.15625, "learning_rate": 2.980647761183144e-06, "loss": 1.2155, "step": 30210 }, { "epoch": 75.5275, "grad_norm": 5.78125, "learning_rate": 2.9800710907832188e-06, "loss": 1.3349, "step": 30211 }, { "epoch": 75.53, "grad_norm": 7.4375, "learning_rate": 2.979494466405669e-06, "loss": 1.27, "step": 30212 }, { "epoch": 75.5325, "grad_norm": 8.75, "learning_rate": 2.9789178880542703e-06, "loss": 1.2589, "step": 30213 }, { "epoch": 75.535, "grad_norm": 7.0625, "learning_rate": 2.978341355732807e-06, "loss": 1.181, "step": 30214 }, { "epoch": 75.5375, "grad_norm": 5.09375, "learning_rate": 2.9777648694450544e-06, "loss": 1.217, "step": 30215 }, { "epoch": 75.54, "grad_norm": 6.46875, "learning_rate": 2.977188429194797e-06, "loss": 1.283, "step": 30216 }, { "epoch": 75.5425, "grad_norm": 11.25, "learning_rate": 2.976612034985812e-06, "loss": 1.3501, "step": 30217 }, { "epoch": 75.545, "grad_norm": 6.375, "learning_rate": 2.976035686821873e-06, "loss": 1.3995, "step": 30218 }, { "epoch": 75.5475, "grad_norm": 8.5, "learning_rate": 2.975459384706765e-06, "loss": 1.2128, "step": 30219 }, { "epoch": 75.55, "grad_norm": 5.9375, "learning_rate": 2.974883128644266e-06, "loss": 1.3022, "step": 30220 }, { "epoch": 75.5525, "grad_norm": 7.8125, "learning_rate": 2.974306918638151e-06, "loss": 1.291, "step": 30221 }, { "epoch": 75.555, "grad_norm": 7.09375, "learning_rate": 2.9737307546922e-06, "loss": 1.2205, "step": 30222 }, { "epoch": 75.5575, "grad_norm": 8.125, "learning_rate": 2.9731546368101827e-06, "loss": 1.4206, "step": 30223 }, { "epoch": 75.56, "grad_norm": 11.25, "learning_rate": 2.9725785649958895e-06, "loss": 1.4522, "step": 30224 }, { "epoch": 75.5625, "grad_norm": 6.40625, "learning_rate": 2.972002539253088e-06, "loss": 1.3404, "step": 30225 }, { "epoch": 75.565, "grad_norm": 7.9375, "learning_rate": 2.9714265595855585e-06, "loss": 1.3827, "step": 30226 }, { "epoch": 75.5675, "grad_norm": 7.4375, "learning_rate": 2.9708506259970706e-06, "loss": 1.2067, "step": 30227 }, { "epoch": 75.57, "grad_norm": 5.90625, "learning_rate": 2.9702747384914055e-06, "loss": 1.2805, "step": 30228 }, { "epoch": 75.5725, "grad_norm": 6.09375, "learning_rate": 2.9696988970723407e-06, "loss": 1.4587, "step": 30229 }, { "epoch": 75.575, "grad_norm": 9.9375, "learning_rate": 2.9691231017436495e-06, "loss": 1.3671, "step": 30230 }, { "epoch": 75.5775, "grad_norm": 8.6875, "learning_rate": 2.9685473525091014e-06, "loss": 1.3396, "step": 30231 }, { "epoch": 75.58, "grad_norm": 8.6875, "learning_rate": 2.9679716493724795e-06, "loss": 1.3182, "step": 30232 }, { "epoch": 75.5825, "grad_norm": 9.25, "learning_rate": 2.9673959923375507e-06, "loss": 1.2628, "step": 30233 }, { "epoch": 75.585, "grad_norm": 5.6875, "learning_rate": 2.966820381408092e-06, "loss": 1.2164, "step": 30234 }, { "epoch": 75.5875, "grad_norm": 7.4375, "learning_rate": 2.966244816587881e-06, "loss": 1.2955, "step": 30235 }, { "epoch": 75.59, "grad_norm": 7.59375, "learning_rate": 2.965669297880688e-06, "loss": 1.3918, "step": 30236 }, { "epoch": 75.5925, "grad_norm": 6.59375, "learning_rate": 2.965093825290285e-06, "loss": 1.3088, "step": 30237 }, { "epoch": 75.595, "grad_norm": 6.46875, "learning_rate": 2.9645183988204384e-06, "loss": 1.3535, "step": 30238 }, { "epoch": 75.5975, "grad_norm": 8.5, "learning_rate": 2.9639430184749353e-06, "loss": 1.3977, "step": 30239 }, { "epoch": 75.6, "grad_norm": 7.375, "learning_rate": 2.9633676842575386e-06, "loss": 1.3009, "step": 30240 }, { "epoch": 75.6025, "grad_norm": 5.71875, "learning_rate": 2.962792396172023e-06, "loss": 1.3723, "step": 30241 }, { "epoch": 75.605, "grad_norm": 10.375, "learning_rate": 2.9622171542221555e-06, "loss": 1.3781, "step": 30242 }, { "epoch": 75.6075, "grad_norm": 6.53125, "learning_rate": 2.9616419584117106e-06, "loss": 1.3502, "step": 30243 }, { "epoch": 75.61, "grad_norm": 9.0625, "learning_rate": 2.9610668087444617e-06, "loss": 1.2891, "step": 30244 }, { "epoch": 75.6125, "grad_norm": 8.25, "learning_rate": 2.9604917052241776e-06, "loss": 1.4257, "step": 30245 }, { "epoch": 75.615, "grad_norm": 12.5, "learning_rate": 2.959916647854626e-06, "loss": 1.1557, "step": 30246 }, { "epoch": 75.6175, "grad_norm": 7.5625, "learning_rate": 2.959341636639581e-06, "loss": 1.381, "step": 30247 }, { "epoch": 75.62, "grad_norm": 6.40625, "learning_rate": 2.958766671582809e-06, "loss": 1.2785, "step": 30248 }, { "epoch": 75.6225, "grad_norm": 8.8125, "learning_rate": 2.958191752688082e-06, "loss": 1.347, "step": 30249 }, { "epoch": 75.625, "grad_norm": 7.0625, "learning_rate": 2.9576168799591663e-06, "loss": 1.2049, "step": 30250 }, { "epoch": 75.6275, "grad_norm": 6.3125, "learning_rate": 2.9570420533998358e-06, "loss": 1.2287, "step": 30251 }, { "epoch": 75.63, "grad_norm": 8.1875, "learning_rate": 2.9564672730138542e-06, "loss": 1.3835, "step": 30252 }, { "epoch": 75.6325, "grad_norm": 9.5, "learning_rate": 2.95589253880499e-06, "loss": 1.3973, "step": 30253 }, { "epoch": 75.635, "grad_norm": 7.4375, "learning_rate": 2.955317850777011e-06, "loss": 1.2237, "step": 30254 }, { "epoch": 75.6375, "grad_norm": 7.40625, "learning_rate": 2.9547432089336902e-06, "loss": 1.2736, "step": 30255 }, { "epoch": 75.64, "grad_norm": 7.125, "learning_rate": 2.9541686132787907e-06, "loss": 1.2176, "step": 30256 }, { "epoch": 75.6425, "grad_norm": 6.71875, "learning_rate": 2.9535940638160766e-06, "loss": 1.3515, "step": 30257 }, { "epoch": 75.645, "grad_norm": 9.9375, "learning_rate": 2.9530195605493184e-06, "loss": 1.2125, "step": 30258 }, { "epoch": 75.6475, "grad_norm": 5.21875, "learning_rate": 2.952445103482284e-06, "loss": 1.2445, "step": 30259 }, { "epoch": 75.65, "grad_norm": 10.1875, "learning_rate": 2.951870692618739e-06, "loss": 1.3742, "step": 30260 }, { "epoch": 75.6525, "grad_norm": 6.34375, "learning_rate": 2.951296327962443e-06, "loss": 1.3112, "step": 30261 }, { "epoch": 75.655, "grad_norm": 8.125, "learning_rate": 2.950722009517171e-06, "loss": 1.3997, "step": 30262 }, { "epoch": 75.6575, "grad_norm": 10.375, "learning_rate": 2.950147737286678e-06, "loss": 1.2689, "step": 30263 }, { "epoch": 75.66, "grad_norm": 7.46875, "learning_rate": 2.949573511274739e-06, "loss": 1.3259, "step": 30264 }, { "epoch": 75.6625, "grad_norm": 7.8125, "learning_rate": 2.9489993314851104e-06, "loss": 1.4064, "step": 30265 }, { "epoch": 75.665, "grad_norm": 5.875, "learning_rate": 2.948425197921564e-06, "loss": 1.2104, "step": 30266 }, { "epoch": 75.6675, "grad_norm": 7.71875, "learning_rate": 2.947851110587859e-06, "loss": 1.1897, "step": 30267 }, { "epoch": 75.67, "grad_norm": 9.0, "learning_rate": 2.9472770694877574e-06, "loss": 1.3235, "step": 30268 }, { "epoch": 75.6725, "grad_norm": 6.59375, "learning_rate": 2.9467030746250235e-06, "loss": 1.2125, "step": 30269 }, { "epoch": 75.675, "grad_norm": 6.65625, "learning_rate": 2.9461291260034263e-06, "loss": 1.3733, "step": 30270 }, { "epoch": 75.6775, "grad_norm": 9.25, "learning_rate": 2.945555223626725e-06, "loss": 1.3367, "step": 30271 }, { "epoch": 75.68, "grad_norm": 7.4375, "learning_rate": 2.944981367498677e-06, "loss": 1.3341, "step": 30272 }, { "epoch": 75.6825, "grad_norm": 7.3125, "learning_rate": 2.944407557623049e-06, "loss": 1.3496, "step": 30273 }, { "epoch": 75.685, "grad_norm": 7.28125, "learning_rate": 2.9438337940036067e-06, "loss": 1.3673, "step": 30274 }, { "epoch": 75.6875, "grad_norm": 8.5, "learning_rate": 2.9432600766441066e-06, "loss": 1.2984, "step": 30275 }, { "epoch": 75.69, "grad_norm": 5.40625, "learning_rate": 2.9426864055483085e-06, "loss": 1.201, "step": 30276 }, { "epoch": 75.6925, "grad_norm": 7.59375, "learning_rate": 2.942112780719979e-06, "loss": 1.2858, "step": 30277 }, { "epoch": 75.695, "grad_norm": 7.03125, "learning_rate": 2.941539202162872e-06, "loss": 1.3726, "step": 30278 }, { "epoch": 75.6975, "grad_norm": 5.84375, "learning_rate": 2.940965669880755e-06, "loss": 1.2751, "step": 30279 }, { "epoch": 75.7, "grad_norm": 7.9375, "learning_rate": 2.940392183877382e-06, "loss": 1.2854, "step": 30280 }, { "epoch": 75.7025, "grad_norm": 5.4375, "learning_rate": 2.9398187441565173e-06, "loss": 1.42, "step": 30281 }, { "epoch": 75.705, "grad_norm": 7.5625, "learning_rate": 2.93924535072192e-06, "loss": 1.2667, "step": 30282 }, { "epoch": 75.7075, "grad_norm": 7.0, "learning_rate": 2.938672003577343e-06, "loss": 1.2856, "step": 30283 }, { "epoch": 75.71, "grad_norm": 7.875, "learning_rate": 2.9380987027265506e-06, "loss": 1.3369, "step": 30284 }, { "epoch": 75.7125, "grad_norm": 9.5, "learning_rate": 2.9375254481733028e-06, "loss": 1.3272, "step": 30285 }, { "epoch": 75.715, "grad_norm": 8.5625, "learning_rate": 2.9369522399213568e-06, "loss": 1.3545, "step": 30286 }, { "epoch": 75.7175, "grad_norm": 7.71875, "learning_rate": 2.936379077974465e-06, "loss": 1.2884, "step": 30287 }, { "epoch": 75.72, "grad_norm": 9.9375, "learning_rate": 2.93580596233639e-06, "loss": 1.3685, "step": 30288 }, { "epoch": 75.7225, "grad_norm": 8.5625, "learning_rate": 2.935232893010891e-06, "loss": 1.2468, "step": 30289 }, { "epoch": 75.725, "grad_norm": 7.78125, "learning_rate": 2.9346598700017224e-06, "loss": 1.4227, "step": 30290 }, { "epoch": 75.7275, "grad_norm": 7.96875, "learning_rate": 2.934086893312639e-06, "loss": 1.5794, "step": 30291 }, { "epoch": 75.73, "grad_norm": 9.5625, "learning_rate": 2.933513962947401e-06, "loss": 1.3079, "step": 30292 }, { "epoch": 75.7325, "grad_norm": 7.40625, "learning_rate": 2.93294107890976e-06, "loss": 1.3179, "step": 30293 }, { "epoch": 75.735, "grad_norm": 15.0, "learning_rate": 2.932368241203477e-06, "loss": 1.2229, "step": 30294 }, { "epoch": 75.7375, "grad_norm": 11.1875, "learning_rate": 2.931795449832303e-06, "loss": 1.2433, "step": 30295 }, { "epoch": 75.74, "grad_norm": 8.125, "learning_rate": 2.9312227047999974e-06, "loss": 1.2694, "step": 30296 }, { "epoch": 75.7425, "grad_norm": 6.9375, "learning_rate": 2.9306500061103136e-06, "loss": 1.4903, "step": 30297 }, { "epoch": 75.745, "grad_norm": 7.375, "learning_rate": 2.9300773537670023e-06, "loss": 1.3303, "step": 30298 }, { "epoch": 75.7475, "grad_norm": 6.59375, "learning_rate": 2.92950474777382e-06, "loss": 1.3583, "step": 30299 }, { "epoch": 75.75, "grad_norm": 9.8125, "learning_rate": 2.9289321881345257e-06, "loss": 1.1805, "step": 30300 }, { "epoch": 75.7525, "grad_norm": 6.78125, "learning_rate": 2.928359674852869e-06, "loss": 1.1731, "step": 30301 }, { "epoch": 75.755, "grad_norm": 7.84375, "learning_rate": 2.9277872079326e-06, "loss": 1.264, "step": 30302 }, { "epoch": 75.7575, "grad_norm": 48.25, "learning_rate": 2.927214787377475e-06, "loss": 1.3183, "step": 30303 }, { "epoch": 75.76, "grad_norm": 7.5625, "learning_rate": 2.9266424131912495e-06, "loss": 1.3202, "step": 30304 }, { "epoch": 75.7625, "grad_norm": 6.9375, "learning_rate": 2.9260700853776734e-06, "loss": 1.3559, "step": 30305 }, { "epoch": 75.765, "grad_norm": 8.25, "learning_rate": 2.925497803940496e-06, "loss": 1.2242, "step": 30306 }, { "epoch": 75.7675, "grad_norm": 6.375, "learning_rate": 2.924925568883474e-06, "loss": 1.2458, "step": 30307 }, { "epoch": 75.77, "grad_norm": 4.59375, "learning_rate": 2.9243533802103553e-06, "loss": 1.3905, "step": 30308 }, { "epoch": 75.7725, "grad_norm": 7.0, "learning_rate": 2.923781237924894e-06, "loss": 1.3911, "step": 30309 }, { "epoch": 75.775, "grad_norm": 5.5, "learning_rate": 2.9232091420308385e-06, "loss": 1.4552, "step": 30310 }, { "epoch": 75.7775, "grad_norm": 9.0625, "learning_rate": 2.9226370925319427e-06, "loss": 1.3044, "step": 30311 }, { "epoch": 75.78, "grad_norm": 6.5625, "learning_rate": 2.9220650894319557e-06, "loss": 1.2701, "step": 30312 }, { "epoch": 75.7825, "grad_norm": 8.75, "learning_rate": 2.921493132734623e-06, "loss": 1.2882, "step": 30313 }, { "epoch": 75.785, "grad_norm": 9.0, "learning_rate": 2.920921222443698e-06, "loss": 1.421, "step": 30314 }, { "epoch": 75.7875, "grad_norm": 9.8125, "learning_rate": 2.920349358562933e-06, "loss": 1.5469, "step": 30315 }, { "epoch": 75.79, "grad_norm": 12.0, "learning_rate": 2.919777541096075e-06, "loss": 1.3254, "step": 30316 }, { "epoch": 75.7925, "grad_norm": 7.46875, "learning_rate": 2.919205770046869e-06, "loss": 1.3631, "step": 30317 }, { "epoch": 75.795, "grad_norm": 8.125, "learning_rate": 2.9186340454190654e-06, "loss": 1.2862, "step": 30318 }, { "epoch": 75.7975, "grad_norm": 7.28125, "learning_rate": 2.918062367216418e-06, "loss": 1.2441, "step": 30319 }, { "epoch": 75.8, "grad_norm": 8.25, "learning_rate": 2.9174907354426696e-06, "loss": 1.1503, "step": 30320 }, { "epoch": 75.8025, "grad_norm": 6.75, "learning_rate": 2.916919150101566e-06, "loss": 1.3004, "step": 30321 }, { "epoch": 75.805, "grad_norm": 7.0, "learning_rate": 2.91634761119686e-06, "loss": 1.2609, "step": 30322 }, { "epoch": 75.8075, "grad_norm": 7.6875, "learning_rate": 2.915776118732292e-06, "loss": 1.3614, "step": 30323 }, { "epoch": 75.81, "grad_norm": 6.875, "learning_rate": 2.9152046727116157e-06, "loss": 1.2551, "step": 30324 }, { "epoch": 75.8125, "grad_norm": 10.875, "learning_rate": 2.914633273138572e-06, "loss": 1.2351, "step": 30325 }, { "epoch": 75.815, "grad_norm": 13.3125, "learning_rate": 2.9140619200169107e-06, "loss": 1.2446, "step": 30326 }, { "epoch": 75.8175, "grad_norm": 6.59375, "learning_rate": 2.913490613350376e-06, "loss": 1.1876, "step": 30327 }, { "epoch": 75.82, "grad_norm": 6.21875, "learning_rate": 2.91291935314271e-06, "loss": 1.3798, "step": 30328 }, { "epoch": 75.8225, "grad_norm": 8.625, "learning_rate": 2.9123481393976615e-06, "loss": 1.32, "step": 30329 }, { "epoch": 75.825, "grad_norm": 7.84375, "learning_rate": 2.911776972118978e-06, "loss": 1.3546, "step": 30330 }, { "epoch": 75.8275, "grad_norm": 5.3125, "learning_rate": 2.9112058513104015e-06, "loss": 1.2655, "step": 30331 }, { "epoch": 75.83, "grad_norm": 7.25, "learning_rate": 2.9106347769756728e-06, "loss": 1.3169, "step": 30332 }, { "epoch": 75.8325, "grad_norm": 7.90625, "learning_rate": 2.910063749118538e-06, "loss": 1.3134, "step": 30333 }, { "epoch": 75.835, "grad_norm": 8.0625, "learning_rate": 2.9094927677427454e-06, "loss": 1.3279, "step": 30334 }, { "epoch": 75.8375, "grad_norm": 8.1875, "learning_rate": 2.908921832852034e-06, "loss": 1.364, "step": 30335 }, { "epoch": 75.84, "grad_norm": 7.15625, "learning_rate": 2.9083509444501433e-06, "loss": 1.3559, "step": 30336 }, { "epoch": 75.8425, "grad_norm": 7.53125, "learning_rate": 2.9077801025408246e-06, "loss": 1.375, "step": 30337 }, { "epoch": 75.845, "grad_norm": 9.8125, "learning_rate": 2.9072093071278116e-06, "loss": 1.2106, "step": 30338 }, { "epoch": 75.8475, "grad_norm": 9.75, "learning_rate": 2.9066385582148547e-06, "loss": 1.586, "step": 30339 }, { "epoch": 75.85, "grad_norm": 9.6875, "learning_rate": 2.9060678558056876e-06, "loss": 1.1796, "step": 30340 }, { "epoch": 75.8525, "grad_norm": 8.3125, "learning_rate": 2.9054971999040595e-06, "loss": 1.3009, "step": 30341 }, { "epoch": 75.855, "grad_norm": 9.375, "learning_rate": 2.904926590513707e-06, "loss": 1.3074, "step": 30342 }, { "epoch": 75.8575, "grad_norm": 8.9375, "learning_rate": 2.9043560276383697e-06, "loss": 1.355, "step": 30343 }, { "epoch": 75.86, "grad_norm": 7.78125, "learning_rate": 2.9037855112817893e-06, "loss": 1.2238, "step": 30344 }, { "epoch": 75.8625, "grad_norm": 8.375, "learning_rate": 2.903215041447711e-06, "loss": 1.3245, "step": 30345 }, { "epoch": 75.865, "grad_norm": 7.0, "learning_rate": 2.9026446181398704e-06, "loss": 1.4399, "step": 30346 }, { "epoch": 75.8675, "grad_norm": 10.3125, "learning_rate": 2.902074241362005e-06, "loss": 1.5759, "step": 30347 }, { "epoch": 75.87, "grad_norm": 6.0, "learning_rate": 2.9015039111178566e-06, "loss": 1.3312, "step": 30348 }, { "epoch": 75.8725, "grad_norm": 8.9375, "learning_rate": 2.9009336274111677e-06, "loss": 1.4753, "step": 30349 }, { "epoch": 75.875, "grad_norm": 7.03125, "learning_rate": 2.900363390245674e-06, "loss": 1.2339, "step": 30350 }, { "epoch": 75.8775, "grad_norm": 5.875, "learning_rate": 2.8997931996251092e-06, "loss": 1.2726, "step": 30351 }, { "epoch": 75.88, "grad_norm": 5.625, "learning_rate": 2.899223055553221e-06, "loss": 1.198, "step": 30352 }, { "epoch": 75.8825, "grad_norm": 8.25, "learning_rate": 2.8986529580337387e-06, "loss": 1.1585, "step": 30353 }, { "epoch": 75.885, "grad_norm": 4.59375, "learning_rate": 2.8980829070704064e-06, "loss": 1.4173, "step": 30354 }, { "epoch": 75.8875, "grad_norm": 7.25, "learning_rate": 2.8975129026669557e-06, "loss": 1.4502, "step": 30355 }, { "epoch": 75.89, "grad_norm": 6.125, "learning_rate": 2.896942944827129e-06, "loss": 1.4008, "step": 30356 }, { "epoch": 75.8925, "grad_norm": 8.5, "learning_rate": 2.896373033554659e-06, "loss": 1.3088, "step": 30357 }, { "epoch": 75.895, "grad_norm": 5.96875, "learning_rate": 2.895803168853282e-06, "loss": 1.18, "step": 30358 }, { "epoch": 75.8975, "grad_norm": 7.40625, "learning_rate": 2.895233350726734e-06, "loss": 1.3671, "step": 30359 }, { "epoch": 75.9, "grad_norm": 7.09375, "learning_rate": 2.8946635791787546e-06, "loss": 1.3958, "step": 30360 }, { "epoch": 75.9025, "grad_norm": 7.75, "learning_rate": 2.894093854213077e-06, "loss": 1.3121, "step": 30361 }, { "epoch": 75.905, "grad_norm": 13.6875, "learning_rate": 2.8935241758334356e-06, "loss": 1.1613, "step": 30362 }, { "epoch": 75.9075, "grad_norm": 6.875, "learning_rate": 2.892954544043559e-06, "loss": 1.245, "step": 30363 }, { "epoch": 75.91, "grad_norm": 7.09375, "learning_rate": 2.892384958847194e-06, "loss": 1.3537, "step": 30364 }, { "epoch": 75.9125, "grad_norm": 8.25, "learning_rate": 2.8918154202480697e-06, "loss": 1.4297, "step": 30365 }, { "epoch": 75.915, "grad_norm": 5.84375, "learning_rate": 2.8912459282499185e-06, "loss": 1.3156, "step": 30366 }, { "epoch": 75.9175, "grad_norm": 9.0625, "learning_rate": 2.8906764828564716e-06, "loss": 1.4253, "step": 30367 }, { "epoch": 75.92, "grad_norm": 5.8125, "learning_rate": 2.890107084071465e-06, "loss": 1.5083, "step": 30368 }, { "epoch": 75.9225, "grad_norm": 6.875, "learning_rate": 2.8895377318986353e-06, "loss": 1.3944, "step": 30369 }, { "epoch": 75.925, "grad_norm": 11.4375, "learning_rate": 2.8889684263417073e-06, "loss": 1.5078, "step": 30370 }, { "epoch": 75.9275, "grad_norm": 7.28125, "learning_rate": 2.888399167404422e-06, "loss": 1.3014, "step": 30371 }, { "epoch": 75.93, "grad_norm": 6.78125, "learning_rate": 2.8878299550905064e-06, "loss": 1.402, "step": 30372 }, { "epoch": 75.9325, "grad_norm": 7.15625, "learning_rate": 2.8872607894036906e-06, "loss": 1.3443, "step": 30373 }, { "epoch": 75.935, "grad_norm": 6.625, "learning_rate": 2.8866916703477086e-06, "loss": 1.386, "step": 30374 }, { "epoch": 75.9375, "grad_norm": 6.0, "learning_rate": 2.886122597926294e-06, "loss": 1.2461, "step": 30375 }, { "epoch": 75.94, "grad_norm": 9.1875, "learning_rate": 2.885553572143175e-06, "loss": 1.4177, "step": 30376 }, { "epoch": 75.9425, "grad_norm": 10.0625, "learning_rate": 2.8849845930020827e-06, "loss": 1.3246, "step": 30377 }, { "epoch": 75.945, "grad_norm": 5.5625, "learning_rate": 2.88441566050674e-06, "loss": 1.3138, "step": 30378 }, { "epoch": 75.9475, "grad_norm": 17.0, "learning_rate": 2.883846774660889e-06, "loss": 1.4145, "step": 30379 }, { "epoch": 75.95, "grad_norm": 10.4375, "learning_rate": 2.883277935468254e-06, "loss": 1.2763, "step": 30380 }, { "epoch": 75.9525, "grad_norm": 5.71875, "learning_rate": 2.882709142932565e-06, "loss": 1.3511, "step": 30381 }, { "epoch": 75.955, "grad_norm": 7.625, "learning_rate": 2.8821403970575455e-06, "loss": 1.1968, "step": 30382 }, { "epoch": 75.9575, "grad_norm": 9.375, "learning_rate": 2.88157169784693e-06, "loss": 1.2599, "step": 30383 }, { "epoch": 75.96, "grad_norm": 6.25, "learning_rate": 2.881003045304448e-06, "loss": 1.3577, "step": 30384 }, { "epoch": 75.9625, "grad_norm": 6.96875, "learning_rate": 2.880434439433826e-06, "loss": 1.3129, "step": 30385 }, { "epoch": 75.965, "grad_norm": 10.1875, "learning_rate": 2.8798658802387867e-06, "loss": 1.3829, "step": 30386 }, { "epoch": 75.9675, "grad_norm": 9.4375, "learning_rate": 2.8792973677230663e-06, "loss": 1.2977, "step": 30387 }, { "epoch": 75.97, "grad_norm": 6.09375, "learning_rate": 2.878728901890383e-06, "loss": 1.378, "step": 30388 }, { "epoch": 75.9725, "grad_norm": 6.65625, "learning_rate": 2.878160482744471e-06, "loss": 1.4679, "step": 30389 }, { "epoch": 75.975, "grad_norm": 8.75, "learning_rate": 2.8775921102890514e-06, "loss": 1.3933, "step": 30390 }, { "epoch": 75.9775, "grad_norm": 6.90625, "learning_rate": 2.8770237845278557e-06, "loss": 1.3087, "step": 30391 }, { "epoch": 75.98, "grad_norm": 8.1875, "learning_rate": 2.8764555054646083e-06, "loss": 1.6041, "step": 30392 }, { "epoch": 75.9825, "grad_norm": 11.375, "learning_rate": 2.8758872731030286e-06, "loss": 1.2836, "step": 30393 }, { "epoch": 75.985, "grad_norm": 8.5, "learning_rate": 2.8753190874468473e-06, "loss": 1.4567, "step": 30394 }, { "epoch": 75.9875, "grad_norm": 11.625, "learning_rate": 2.8747509484997926e-06, "loss": 1.3791, "step": 30395 }, { "epoch": 75.99, "grad_norm": 6.8125, "learning_rate": 2.874182856265586e-06, "loss": 1.3503, "step": 30396 }, { "epoch": 75.9925, "grad_norm": 8.1875, "learning_rate": 2.8736148107479467e-06, "loss": 1.2734, "step": 30397 }, { "epoch": 75.995, "grad_norm": 8.5, "learning_rate": 2.873046811950605e-06, "loss": 1.2062, "step": 30398 }, { "epoch": 75.9975, "grad_norm": 7.65625, "learning_rate": 2.8724788598772857e-06, "loss": 1.4513, "step": 30399 }, { "epoch": 76.0, "grad_norm": 7.03125, "learning_rate": 2.8719109545317102e-06, "loss": 1.3283, "step": 30400 }, { "epoch": 76.0025, "grad_norm": 5.8125, "learning_rate": 2.871343095917597e-06, "loss": 1.3042, "step": 30401 }, { "epoch": 76.005, "grad_norm": 7.6875, "learning_rate": 2.870775284038677e-06, "loss": 1.2494, "step": 30402 }, { "epoch": 76.0075, "grad_norm": 8.0625, "learning_rate": 2.870207518898667e-06, "loss": 1.2745, "step": 30403 }, { "epoch": 76.01, "grad_norm": 6.78125, "learning_rate": 2.8696398005012927e-06, "loss": 1.4313, "step": 30404 }, { "epoch": 76.0125, "grad_norm": 6.09375, "learning_rate": 2.869072128850272e-06, "loss": 1.195, "step": 30405 }, { "epoch": 76.015, "grad_norm": 10.875, "learning_rate": 2.868504503949332e-06, "loss": 1.4263, "step": 30406 }, { "epoch": 76.0175, "grad_norm": 10.5, "learning_rate": 2.8679369258021907e-06, "loss": 1.4244, "step": 30407 }, { "epoch": 76.02, "grad_norm": 12.9375, "learning_rate": 2.867369394412567e-06, "loss": 1.2932, "step": 30408 }, { "epoch": 76.0225, "grad_norm": 9.0, "learning_rate": 2.866801909784184e-06, "loss": 1.4772, "step": 30409 }, { "epoch": 76.025, "grad_norm": 7.71875, "learning_rate": 2.866234471920766e-06, "loss": 1.2225, "step": 30410 }, { "epoch": 76.0275, "grad_norm": 8.6875, "learning_rate": 2.8656670808260278e-06, "loss": 1.4545, "step": 30411 }, { "epoch": 76.03, "grad_norm": 6.90625, "learning_rate": 2.865099736503688e-06, "loss": 1.1826, "step": 30412 }, { "epoch": 76.0325, "grad_norm": 6.4375, "learning_rate": 2.8645324389574704e-06, "loss": 1.3735, "step": 30413 }, { "epoch": 76.035, "grad_norm": 6.6875, "learning_rate": 2.8639651881910935e-06, "loss": 1.3584, "step": 30414 }, { "epoch": 76.0375, "grad_norm": 11.25, "learning_rate": 2.8633979842082772e-06, "loss": 1.2255, "step": 30415 }, { "epoch": 76.04, "grad_norm": 7.625, "learning_rate": 2.8628308270127335e-06, "loss": 1.325, "step": 30416 }, { "epoch": 76.0425, "grad_norm": 9.6875, "learning_rate": 2.862263716608189e-06, "loss": 1.5106, "step": 30417 }, { "epoch": 76.045, "grad_norm": 7.8125, "learning_rate": 2.8616966529983547e-06, "loss": 1.249, "step": 30418 }, { "epoch": 76.0475, "grad_norm": 6.46875, "learning_rate": 2.8611296361869545e-06, "loss": 1.4596, "step": 30419 }, { "epoch": 76.05, "grad_norm": 7.78125, "learning_rate": 2.8605626661776995e-06, "loss": 1.2449, "step": 30420 }, { "epoch": 76.0525, "grad_norm": 4.8125, "learning_rate": 2.859995742974314e-06, "loss": 1.1554, "step": 30421 }, { "epoch": 76.055, "grad_norm": 5.875, "learning_rate": 2.859428866580507e-06, "loss": 1.3092, "step": 30422 }, { "epoch": 76.0575, "grad_norm": 8.375, "learning_rate": 2.858862037000002e-06, "loss": 1.3688, "step": 30423 }, { "epoch": 76.06, "grad_norm": 7.84375, "learning_rate": 2.8582952542365083e-06, "loss": 1.3116, "step": 30424 }, { "epoch": 76.0625, "grad_norm": 10.0625, "learning_rate": 2.8577285182937477e-06, "loss": 1.3733, "step": 30425 }, { "epoch": 76.065, "grad_norm": 5.78125, "learning_rate": 2.8571618291754344e-06, "loss": 1.3543, "step": 30426 }, { "epoch": 76.0675, "grad_norm": 4.875, "learning_rate": 2.856595186885278e-06, "loss": 1.3308, "step": 30427 }, { "epoch": 76.07, "grad_norm": 7.25, "learning_rate": 2.856028591426998e-06, "loss": 1.4372, "step": 30428 }, { "epoch": 76.0725, "grad_norm": 8.875, "learning_rate": 2.855462042804312e-06, "loss": 1.5071, "step": 30429 }, { "epoch": 76.075, "grad_norm": 8.375, "learning_rate": 2.8548955410209312e-06, "loss": 1.3915, "step": 30430 }, { "epoch": 76.0775, "grad_norm": 6.71875, "learning_rate": 2.854329086080565e-06, "loss": 1.2233, "step": 30431 }, { "epoch": 76.08, "grad_norm": 7.84375, "learning_rate": 2.853762677986932e-06, "loss": 1.2918, "step": 30432 }, { "epoch": 76.0825, "grad_norm": 10.8125, "learning_rate": 2.8531963167437494e-06, "loss": 1.3378, "step": 30433 }, { "epoch": 76.085, "grad_norm": 8.375, "learning_rate": 2.8526300023547236e-06, "loss": 1.4498, "step": 30434 }, { "epoch": 76.0875, "grad_norm": 6.46875, "learning_rate": 2.852063734823568e-06, "loss": 1.4484, "step": 30435 }, { "epoch": 76.09, "grad_norm": 6.6875, "learning_rate": 2.8514975141539993e-06, "loss": 1.4149, "step": 30436 }, { "epoch": 76.0925, "grad_norm": 7.5625, "learning_rate": 2.850931340349723e-06, "loss": 1.3105, "step": 30437 }, { "epoch": 76.095, "grad_norm": 7.84375, "learning_rate": 2.8503652134144576e-06, "loss": 1.2643, "step": 30438 }, { "epoch": 76.0975, "grad_norm": 8.3125, "learning_rate": 2.849799133351909e-06, "loss": 1.2351, "step": 30439 }, { "epoch": 76.1, "grad_norm": 6.46875, "learning_rate": 2.849233100165795e-06, "loss": 1.3689, "step": 30440 }, { "epoch": 76.1025, "grad_norm": 8.25, "learning_rate": 2.8486671138598223e-06, "loss": 1.3621, "step": 30441 }, { "epoch": 76.105, "grad_norm": 8.5625, "learning_rate": 2.8481011744376985e-06, "loss": 1.3835, "step": 30442 }, { "epoch": 76.1075, "grad_norm": 6.9375, "learning_rate": 2.8475352819031377e-06, "loss": 1.3495, "step": 30443 }, { "epoch": 76.11, "grad_norm": 7.1875, "learning_rate": 2.846969436259852e-06, "loss": 1.225, "step": 30444 }, { "epoch": 76.1125, "grad_norm": 4.8125, "learning_rate": 2.8464036375115477e-06, "loss": 1.1395, "step": 30445 }, { "epoch": 76.115, "grad_norm": 10.5, "learning_rate": 2.845837885661933e-06, "loss": 1.2588, "step": 30446 }, { "epoch": 76.1175, "grad_norm": 6.5625, "learning_rate": 2.8452721807147188e-06, "loss": 1.3746, "step": 30447 }, { "epoch": 76.12, "grad_norm": 8.375, "learning_rate": 2.844706522673616e-06, "loss": 1.3509, "step": 30448 }, { "epoch": 76.1225, "grad_norm": 7.90625, "learning_rate": 2.8441409115423313e-06, "loss": 1.4527, "step": 30449 }, { "epoch": 76.125, "grad_norm": 6.1875, "learning_rate": 2.8435753473245697e-06, "loss": 1.404, "step": 30450 }, { "epoch": 76.1275, "grad_norm": 7.8125, "learning_rate": 2.843009830024045e-06, "loss": 1.3881, "step": 30451 }, { "epoch": 76.13, "grad_norm": 8.9375, "learning_rate": 2.842444359644457e-06, "loss": 1.3316, "step": 30452 }, { "epoch": 76.1325, "grad_norm": 6.09375, "learning_rate": 2.8418789361895216e-06, "loss": 1.2924, "step": 30453 }, { "epoch": 76.135, "grad_norm": 6.9375, "learning_rate": 2.8413135596629383e-06, "loss": 1.1389, "step": 30454 }, { "epoch": 76.1375, "grad_norm": 7.71875, "learning_rate": 2.8407482300684195e-06, "loss": 1.3766, "step": 30455 }, { "epoch": 76.14, "grad_norm": 8.5625, "learning_rate": 2.8401829474096696e-06, "loss": 1.4293, "step": 30456 }, { "epoch": 76.1425, "grad_norm": 5.875, "learning_rate": 2.83961771169039e-06, "loss": 1.2952, "step": 30457 }, { "epoch": 76.145, "grad_norm": 7.0, "learning_rate": 2.8390525229142907e-06, "loss": 1.3667, "step": 30458 }, { "epoch": 76.1475, "grad_norm": 6.1875, "learning_rate": 2.838487381085079e-06, "loss": 1.2604, "step": 30459 }, { "epoch": 76.15, "grad_norm": 11.3125, "learning_rate": 2.837922286206457e-06, "loss": 1.4302, "step": 30460 }, { "epoch": 76.1525, "grad_norm": 5.75, "learning_rate": 2.837357238282128e-06, "loss": 1.2939, "step": 30461 }, { "epoch": 76.155, "grad_norm": 8.375, "learning_rate": 2.8367922373157974e-06, "loss": 1.4087, "step": 30462 }, { "epoch": 76.1575, "grad_norm": 6.46875, "learning_rate": 2.8362272833111737e-06, "loss": 1.2896, "step": 30463 }, { "epoch": 76.16, "grad_norm": 7.125, "learning_rate": 2.835662376271957e-06, "loss": 1.298, "step": 30464 }, { "epoch": 76.1625, "grad_norm": 6.125, "learning_rate": 2.835097516201849e-06, "loss": 1.2707, "step": 30465 }, { "epoch": 76.165, "grad_norm": 9.0625, "learning_rate": 2.8345327031045568e-06, "loss": 1.4029, "step": 30466 }, { "epoch": 76.1675, "grad_norm": 7.3125, "learning_rate": 2.8339679369837793e-06, "loss": 1.2981, "step": 30467 }, { "epoch": 76.17, "grad_norm": 8.6875, "learning_rate": 2.8334032178432234e-06, "loss": 1.223, "step": 30468 }, { "epoch": 76.1725, "grad_norm": 6.40625, "learning_rate": 2.832838545686586e-06, "loss": 1.3628, "step": 30469 }, { "epoch": 76.175, "grad_norm": 6.40625, "learning_rate": 2.8322739205175763e-06, "loss": 1.3085, "step": 30470 }, { "epoch": 76.1775, "grad_norm": 6.96875, "learning_rate": 2.831709342339891e-06, "loss": 1.3791, "step": 30471 }, { "epoch": 76.18, "grad_norm": 5.9375, "learning_rate": 2.8311448111572304e-06, "loss": 1.2608, "step": 30472 }, { "epoch": 76.1825, "grad_norm": 7.9375, "learning_rate": 2.830580326973297e-06, "loss": 1.3232, "step": 30473 }, { "epoch": 76.185, "grad_norm": 8.875, "learning_rate": 2.830015889791795e-06, "loss": 1.3074, "step": 30474 }, { "epoch": 76.1875, "grad_norm": 9.0, "learning_rate": 2.8294514996164224e-06, "loss": 1.3838, "step": 30475 }, { "epoch": 76.19, "grad_norm": 9.0625, "learning_rate": 2.8288871564508745e-06, "loss": 1.2473, "step": 30476 }, { "epoch": 76.1925, "grad_norm": 5.65625, "learning_rate": 2.8283228602988565e-06, "loss": 1.4191, "step": 30477 }, { "epoch": 76.195, "grad_norm": 9.9375, "learning_rate": 2.8277586111640696e-06, "loss": 1.3931, "step": 30478 }, { "epoch": 76.1975, "grad_norm": 5.75, "learning_rate": 2.8271944090502114e-06, "loss": 1.2521, "step": 30479 }, { "epoch": 76.2, "grad_norm": 10.6875, "learning_rate": 2.8266302539609747e-06, "loss": 1.3248, "step": 30480 }, { "epoch": 76.2025, "grad_norm": 9.0, "learning_rate": 2.826066145900067e-06, "loss": 1.2376, "step": 30481 }, { "epoch": 76.205, "grad_norm": 7.53125, "learning_rate": 2.82550208487118e-06, "loss": 1.4679, "step": 30482 }, { "epoch": 76.2075, "grad_norm": 9.1875, "learning_rate": 2.8249380708780162e-06, "loss": 1.3951, "step": 30483 }, { "epoch": 76.21, "grad_norm": 9.125, "learning_rate": 2.824374103924269e-06, "loss": 1.4001, "step": 30484 }, { "epoch": 76.2125, "grad_norm": 5.9375, "learning_rate": 2.82381018401364e-06, "loss": 1.2626, "step": 30485 }, { "epoch": 76.215, "grad_norm": 8.0625, "learning_rate": 2.823246311149825e-06, "loss": 1.2277, "step": 30486 }, { "epoch": 76.2175, "grad_norm": 8.5, "learning_rate": 2.8226824853365177e-06, "loss": 1.2457, "step": 30487 }, { "epoch": 76.22, "grad_norm": 9.8125, "learning_rate": 2.822118706577416e-06, "loss": 1.2333, "step": 30488 }, { "epoch": 76.2225, "grad_norm": 5.21875, "learning_rate": 2.8215549748762193e-06, "loss": 1.2984, "step": 30489 }, { "epoch": 76.225, "grad_norm": 8.0625, "learning_rate": 2.820991290236622e-06, "loss": 1.3342, "step": 30490 }, { "epoch": 76.2275, "grad_norm": 6.28125, "learning_rate": 2.8204276526623155e-06, "loss": 1.3869, "step": 30491 }, { "epoch": 76.23, "grad_norm": 6.65625, "learning_rate": 2.8198640621569973e-06, "loss": 1.1855, "step": 30492 }, { "epoch": 76.2325, "grad_norm": 8.8125, "learning_rate": 2.819300518724366e-06, "loss": 1.5069, "step": 30493 }, { "epoch": 76.235, "grad_norm": 9.8125, "learning_rate": 2.8187370223681134e-06, "loss": 1.2315, "step": 30494 }, { "epoch": 76.2375, "grad_norm": 7.0, "learning_rate": 2.8181735730919303e-06, "loss": 1.1769, "step": 30495 }, { "epoch": 76.24, "grad_norm": 7.34375, "learning_rate": 2.8176101708995174e-06, "loss": 1.2216, "step": 30496 }, { "epoch": 76.2425, "grad_norm": 7.6875, "learning_rate": 2.817046815794561e-06, "loss": 1.4088, "step": 30497 }, { "epoch": 76.245, "grad_norm": 6.90625, "learning_rate": 2.8164835077807616e-06, "loss": 1.3798, "step": 30498 }, { "epoch": 76.2475, "grad_norm": 6.25, "learning_rate": 2.8159202468618064e-06, "loss": 1.2914, "step": 30499 }, { "epoch": 76.25, "grad_norm": 6.25, "learning_rate": 2.8153570330413925e-06, "loss": 1.3095, "step": 30500 }, { "epoch": 76.2525, "grad_norm": 5.84375, "learning_rate": 2.814793866323211e-06, "loss": 1.3039, "step": 30501 }, { "epoch": 76.255, "grad_norm": 6.78125, "learning_rate": 2.81423074671095e-06, "loss": 1.3086, "step": 30502 }, { "epoch": 76.2575, "grad_norm": 9.5625, "learning_rate": 2.8136676742083046e-06, "loss": 1.2521, "step": 30503 }, { "epoch": 76.26, "grad_norm": 12.875, "learning_rate": 2.8131046488189705e-06, "loss": 1.4107, "step": 30504 }, { "epoch": 76.2625, "grad_norm": 7.34375, "learning_rate": 2.8125416705466335e-06, "loss": 1.3672, "step": 30505 }, { "epoch": 76.265, "grad_norm": 8.3125, "learning_rate": 2.811978739394986e-06, "loss": 1.4002, "step": 30506 }, { "epoch": 76.2675, "grad_norm": 8.8125, "learning_rate": 2.811415855367713e-06, "loss": 1.4303, "step": 30507 }, { "epoch": 76.27, "grad_norm": 8.125, "learning_rate": 2.8108530184685143e-06, "loss": 1.3638, "step": 30508 }, { "epoch": 76.2725, "grad_norm": 6.34375, "learning_rate": 2.810290228701077e-06, "loss": 1.3073, "step": 30509 }, { "epoch": 76.275, "grad_norm": 5.53125, "learning_rate": 2.809727486069086e-06, "loss": 1.1695, "step": 30510 }, { "epoch": 76.2775, "grad_norm": 8.25, "learning_rate": 2.8091647905762366e-06, "loss": 1.3292, "step": 30511 }, { "epoch": 76.28, "grad_norm": 9.4375, "learning_rate": 2.808602142226212e-06, "loss": 1.2906, "step": 30512 }, { "epoch": 76.2825, "grad_norm": 11.4375, "learning_rate": 2.808039541022708e-06, "loss": 1.3402, "step": 30513 }, { "epoch": 76.285, "grad_norm": 10.25, "learning_rate": 2.807476986969404e-06, "loss": 1.2886, "step": 30514 }, { "epoch": 76.2875, "grad_norm": 7.28125, "learning_rate": 2.8069144800699977e-06, "loss": 1.2008, "step": 30515 }, { "epoch": 76.29, "grad_norm": 11.1875, "learning_rate": 2.8063520203281714e-06, "loss": 1.3258, "step": 30516 }, { "epoch": 76.2925, "grad_norm": 7.5625, "learning_rate": 2.805789607747611e-06, "loss": 1.2883, "step": 30517 }, { "epoch": 76.295, "grad_norm": 7.78125, "learning_rate": 2.805227242332005e-06, "loss": 1.3594, "step": 30518 }, { "epoch": 76.2975, "grad_norm": 5.96875, "learning_rate": 2.8046649240850453e-06, "loss": 1.3461, "step": 30519 }, { "epoch": 76.3, "grad_norm": 6.84375, "learning_rate": 2.8041026530104144e-06, "loss": 1.2694, "step": 30520 }, { "epoch": 76.3025, "grad_norm": 6.5625, "learning_rate": 2.8035404291117973e-06, "loss": 1.1631, "step": 30521 }, { "epoch": 76.305, "grad_norm": 11.0625, "learning_rate": 2.8029782523928763e-06, "loss": 1.2531, "step": 30522 }, { "epoch": 76.3075, "grad_norm": 4.8125, "learning_rate": 2.8024161228573476e-06, "loss": 1.3663, "step": 30523 }, { "epoch": 76.31, "grad_norm": 8.1875, "learning_rate": 2.8018540405088914e-06, "loss": 1.1588, "step": 30524 }, { "epoch": 76.3125, "grad_norm": 9.375, "learning_rate": 2.8012920053511916e-06, "loss": 1.3659, "step": 30525 }, { "epoch": 76.315, "grad_norm": 6.84375, "learning_rate": 2.8007300173879293e-06, "loss": 1.1969, "step": 30526 }, { "epoch": 76.3175, "grad_norm": 6.8125, "learning_rate": 2.8001680766227947e-06, "loss": 1.2218, "step": 30527 }, { "epoch": 76.32, "grad_norm": 5.46875, "learning_rate": 2.7996061830594714e-06, "loss": 1.1547, "step": 30528 }, { "epoch": 76.3225, "grad_norm": 5.0, "learning_rate": 2.799044336701643e-06, "loss": 1.3, "step": 30529 }, { "epoch": 76.325, "grad_norm": 6.46875, "learning_rate": 2.7984825375529876e-06, "loss": 1.2655, "step": 30530 }, { "epoch": 76.3275, "grad_norm": 6.71875, "learning_rate": 2.7979207856171963e-06, "loss": 1.2194, "step": 30531 }, { "epoch": 76.33, "grad_norm": 6.53125, "learning_rate": 2.7973590808979455e-06, "loss": 1.1739, "step": 30532 }, { "epoch": 76.3325, "grad_norm": 7.375, "learning_rate": 2.7967974233989226e-06, "loss": 1.4771, "step": 30533 }, { "epoch": 76.335, "grad_norm": 6.875, "learning_rate": 2.7962358131238055e-06, "loss": 1.3107, "step": 30534 }, { "epoch": 76.3375, "grad_norm": 7.8125, "learning_rate": 2.79567425007628e-06, "loss": 1.2184, "step": 30535 }, { "epoch": 76.34, "grad_norm": 8.1875, "learning_rate": 2.795112734260026e-06, "loss": 1.2714, "step": 30536 }, { "epoch": 76.3425, "grad_norm": 6.78125, "learning_rate": 2.7945512656787186e-06, "loss": 1.3608, "step": 30537 }, { "epoch": 76.345, "grad_norm": 11.3125, "learning_rate": 2.7939898443360513e-06, "loss": 1.1961, "step": 30538 }, { "epoch": 76.3475, "grad_norm": 5.625, "learning_rate": 2.7934284702356983e-06, "loss": 1.3485, "step": 30539 }, { "epoch": 76.35, "grad_norm": 7.46875, "learning_rate": 2.7928671433813392e-06, "loss": 1.3954, "step": 30540 }, { "epoch": 76.3525, "grad_norm": 6.875, "learning_rate": 2.792305863776652e-06, "loss": 1.3244, "step": 30541 }, { "epoch": 76.355, "grad_norm": 8.375, "learning_rate": 2.7917446314253195e-06, "loss": 1.1676, "step": 30542 }, { "epoch": 76.3575, "grad_norm": 9.3125, "learning_rate": 2.791183446331023e-06, "loss": 1.2848, "step": 30543 }, { "epoch": 76.36, "grad_norm": 6.59375, "learning_rate": 2.7906223084974405e-06, "loss": 1.1317, "step": 30544 }, { "epoch": 76.3625, "grad_norm": 8.9375, "learning_rate": 2.7900612179282448e-06, "loss": 1.3885, "step": 30545 }, { "epoch": 76.365, "grad_norm": 7.03125, "learning_rate": 2.7895001746271234e-06, "loss": 1.2554, "step": 30546 }, { "epoch": 76.3675, "grad_norm": 5.71875, "learning_rate": 2.788939178597747e-06, "loss": 1.3202, "step": 30547 }, { "epoch": 76.37, "grad_norm": 10.25, "learning_rate": 2.7883782298438002e-06, "loss": 1.3312, "step": 30548 }, { "epoch": 76.3725, "grad_norm": 7.6875, "learning_rate": 2.7878173283689536e-06, "loss": 1.2663, "step": 30549 }, { "epoch": 76.375, "grad_norm": 8.6875, "learning_rate": 2.7872564741768917e-06, "loss": 1.3016, "step": 30550 }, { "epoch": 76.3775, "grad_norm": 8.6875, "learning_rate": 2.7866956672712873e-06, "loss": 1.2907, "step": 30551 }, { "epoch": 76.38, "grad_norm": 10.0, "learning_rate": 2.786134907655814e-06, "loss": 1.2843, "step": 30552 }, { "epoch": 76.3825, "grad_norm": 17.625, "learning_rate": 2.785574195334152e-06, "loss": 1.3663, "step": 30553 }, { "epoch": 76.385, "grad_norm": 6.75, "learning_rate": 2.78501353030998e-06, "loss": 1.3184, "step": 30554 }, { "epoch": 76.3875, "grad_norm": 7.09375, "learning_rate": 2.7844529125869714e-06, "loss": 1.2426, "step": 30555 }, { "epoch": 76.39, "grad_norm": 7.78125, "learning_rate": 2.783892342168796e-06, "loss": 1.2183, "step": 30556 }, { "epoch": 76.3925, "grad_norm": 6.5, "learning_rate": 2.783331819059134e-06, "loss": 1.1222, "step": 30557 }, { "epoch": 76.395, "grad_norm": 6.59375, "learning_rate": 2.782771343261663e-06, "loss": 1.3383, "step": 30558 }, { "epoch": 76.3975, "grad_norm": 9.3125, "learning_rate": 2.7822109147800545e-06, "loss": 1.367, "step": 30559 }, { "epoch": 76.4, "grad_norm": 6.84375, "learning_rate": 2.78165053361798e-06, "loss": 1.3904, "step": 30560 }, { "epoch": 76.4025, "grad_norm": 8.875, "learning_rate": 2.7810901997791183e-06, "loss": 1.398, "step": 30561 }, { "epoch": 76.405, "grad_norm": 6.3125, "learning_rate": 2.780529913267137e-06, "loss": 1.393, "step": 30562 }, { "epoch": 76.4075, "grad_norm": 8.9375, "learning_rate": 2.7799696740857163e-06, "loss": 1.2377, "step": 30563 }, { "epoch": 76.41, "grad_norm": 20.25, "learning_rate": 2.779409482238522e-06, "loss": 1.3986, "step": 30564 }, { "epoch": 76.4125, "grad_norm": 8.5, "learning_rate": 2.7788493377292324e-06, "loss": 1.3045, "step": 30565 }, { "epoch": 76.415, "grad_norm": 11.1875, "learning_rate": 2.778289240561519e-06, "loss": 1.3914, "step": 30566 }, { "epoch": 76.4175, "grad_norm": 6.75, "learning_rate": 2.777729190739048e-06, "loss": 1.3221, "step": 30567 }, { "epoch": 76.42, "grad_norm": 5.6875, "learning_rate": 2.777169188265496e-06, "loss": 1.335, "step": 30568 }, { "epoch": 76.4225, "grad_norm": 12.25, "learning_rate": 2.776609233144537e-06, "loss": 1.3811, "step": 30569 }, { "epoch": 76.425, "grad_norm": 8.1875, "learning_rate": 2.776049325379839e-06, "loss": 1.423, "step": 30570 }, { "epoch": 76.4275, "grad_norm": 6.9375, "learning_rate": 2.77548946497507e-06, "loss": 1.2414, "step": 30571 }, { "epoch": 76.43, "grad_norm": 5.65625, "learning_rate": 2.774929651933902e-06, "loss": 1.3206, "step": 30572 }, { "epoch": 76.4325, "grad_norm": 7.1875, "learning_rate": 2.774369886260009e-06, "loss": 1.3076, "step": 30573 }, { "epoch": 76.435, "grad_norm": 7.375, "learning_rate": 2.7738101679570585e-06, "loss": 1.2934, "step": 30574 }, { "epoch": 76.4375, "grad_norm": 6.59375, "learning_rate": 2.7732504970287154e-06, "loss": 1.3142, "step": 30575 }, { "epoch": 76.44, "grad_norm": 7.65625, "learning_rate": 2.772690873478656e-06, "loss": 1.226, "step": 30576 }, { "epoch": 76.4425, "grad_norm": 8.1875, "learning_rate": 2.772131297310544e-06, "loss": 1.2699, "step": 30577 }, { "epoch": 76.445, "grad_norm": 5.4375, "learning_rate": 2.771571768528053e-06, "loss": 1.3575, "step": 30578 }, { "epoch": 76.4475, "grad_norm": 5.9375, "learning_rate": 2.7710122871348443e-06, "loss": 1.336, "step": 30579 }, { "epoch": 76.45, "grad_norm": 8.0625, "learning_rate": 2.770452853134593e-06, "loss": 1.2697, "step": 30580 }, { "epoch": 76.4525, "grad_norm": 7.6875, "learning_rate": 2.769893466530964e-06, "loss": 1.2272, "step": 30581 }, { "epoch": 76.455, "grad_norm": 8.6875, "learning_rate": 2.7693341273276198e-06, "loss": 1.3948, "step": 30582 }, { "epoch": 76.4575, "grad_norm": 6.84375, "learning_rate": 2.768774835528233e-06, "loss": 1.3047, "step": 30583 }, { "epoch": 76.46, "grad_norm": 6.25, "learning_rate": 2.7682155911364718e-06, "loss": 1.2643, "step": 30584 }, { "epoch": 76.4625, "grad_norm": 7.75, "learning_rate": 2.7676563941559995e-06, "loss": 1.278, "step": 30585 }, { "epoch": 76.465, "grad_norm": 8.75, "learning_rate": 2.7670972445904797e-06, "loss": 1.33, "step": 30586 }, { "epoch": 76.4675, "grad_norm": 7.65625, "learning_rate": 2.76653814244358e-06, "loss": 1.1453, "step": 30587 }, { "epoch": 76.47, "grad_norm": 8.625, "learning_rate": 2.765979087718971e-06, "loss": 1.2909, "step": 30588 }, { "epoch": 76.4725, "grad_norm": 6.21875, "learning_rate": 2.7654200804203137e-06, "loss": 1.3066, "step": 30589 }, { "epoch": 76.475, "grad_norm": 6.625, "learning_rate": 2.764861120551269e-06, "loss": 1.2635, "step": 30590 }, { "epoch": 76.4775, "grad_norm": 7.78125, "learning_rate": 2.764302208115508e-06, "loss": 1.3925, "step": 30591 }, { "epoch": 76.48, "grad_norm": 8.625, "learning_rate": 2.7637433431166903e-06, "loss": 1.4417, "step": 30592 }, { "epoch": 76.4825, "grad_norm": 7.34375, "learning_rate": 2.7631845255584846e-06, "loss": 1.1401, "step": 30593 }, { "epoch": 76.485, "grad_norm": 5.03125, "learning_rate": 2.762625755444547e-06, "loss": 1.2574, "step": 30594 }, { "epoch": 76.4875, "grad_norm": 6.5625, "learning_rate": 2.762067032778549e-06, "loss": 1.3499, "step": 30595 }, { "epoch": 76.49, "grad_norm": 11.3125, "learning_rate": 2.761508357564151e-06, "loss": 1.4482, "step": 30596 }, { "epoch": 76.4925, "grad_norm": 6.0, "learning_rate": 2.7609497298050104e-06, "loss": 1.3257, "step": 30597 }, { "epoch": 76.495, "grad_norm": 5.65625, "learning_rate": 2.760391149504794e-06, "loss": 1.3114, "step": 30598 }, { "epoch": 76.4975, "grad_norm": 9.0, "learning_rate": 2.7598326166671676e-06, "loss": 1.169, "step": 30599 }, { "epoch": 76.5, "grad_norm": 8.5, "learning_rate": 2.759274131295787e-06, "loss": 1.3563, "step": 30600 }, { "epoch": 76.5025, "grad_norm": 8.4375, "learning_rate": 2.7587156933943138e-06, "loss": 1.3442, "step": 30601 }, { "epoch": 76.505, "grad_norm": 6.09375, "learning_rate": 2.75815730296641e-06, "loss": 1.0501, "step": 30602 }, { "epoch": 76.5075, "grad_norm": 14.125, "learning_rate": 2.7575989600157405e-06, "loss": 1.2421, "step": 30603 }, { "epoch": 76.51, "grad_norm": 9.4375, "learning_rate": 2.757040664545961e-06, "loss": 1.3019, "step": 30604 }, { "epoch": 76.5125, "grad_norm": 5.6875, "learning_rate": 2.756482416560732e-06, "loss": 1.2941, "step": 30605 }, { "epoch": 76.515, "grad_norm": 9.25, "learning_rate": 2.7559242160637167e-06, "loss": 1.2429, "step": 30606 }, { "epoch": 76.5175, "grad_norm": 8.5, "learning_rate": 2.7553660630585687e-06, "loss": 1.3439, "step": 30607 }, { "epoch": 76.52, "grad_norm": 7.15625, "learning_rate": 2.754807957548955e-06, "loss": 1.2257, "step": 30608 }, { "epoch": 76.5225, "grad_norm": 7.875, "learning_rate": 2.754249899538527e-06, "loss": 1.4409, "step": 30609 }, { "epoch": 76.525, "grad_norm": 9.0, "learning_rate": 2.7536918890309494e-06, "loss": 1.4859, "step": 30610 }, { "epoch": 76.5275, "grad_norm": 10.4375, "learning_rate": 2.7531339260298782e-06, "loss": 1.2543, "step": 30611 }, { "epoch": 76.53, "grad_norm": 7.46875, "learning_rate": 2.7525760105389676e-06, "loss": 1.2299, "step": 30612 }, { "epoch": 76.5325, "grad_norm": 5.875, "learning_rate": 2.7520181425618785e-06, "loss": 1.3616, "step": 30613 }, { "epoch": 76.535, "grad_norm": 6.0625, "learning_rate": 2.7514603221022717e-06, "loss": 1.3778, "step": 30614 }, { "epoch": 76.5375, "grad_norm": 7.0, "learning_rate": 2.750902549163801e-06, "loss": 1.3071, "step": 30615 }, { "epoch": 76.54, "grad_norm": 8.125, "learning_rate": 2.7503448237501208e-06, "loss": 1.4496, "step": 30616 }, { "epoch": 76.5425, "grad_norm": 6.3125, "learning_rate": 2.749787145864888e-06, "loss": 1.3015, "step": 30617 }, { "epoch": 76.545, "grad_norm": 6.09375, "learning_rate": 2.7492295155117653e-06, "loss": 1.3557, "step": 30618 }, { "epoch": 76.5475, "grad_norm": 7.3125, "learning_rate": 2.7486719326944034e-06, "loss": 1.235, "step": 30619 }, { "epoch": 76.55, "grad_norm": 6.5625, "learning_rate": 2.7481143974164548e-06, "loss": 1.3216, "step": 30620 }, { "epoch": 76.5525, "grad_norm": 6.375, "learning_rate": 2.7475569096815803e-06, "loss": 1.4192, "step": 30621 }, { "epoch": 76.555, "grad_norm": 7.15625, "learning_rate": 2.74699946949343e-06, "loss": 1.2867, "step": 30622 }, { "epoch": 76.5575, "grad_norm": 10.5625, "learning_rate": 2.7464420768556645e-06, "loss": 1.1887, "step": 30623 }, { "epoch": 76.56, "grad_norm": 8.3125, "learning_rate": 2.745884731771931e-06, "loss": 1.32, "step": 30624 }, { "epoch": 76.5625, "grad_norm": 6.625, "learning_rate": 2.7453274342458903e-06, "loss": 1.3594, "step": 30625 }, { "epoch": 76.565, "grad_norm": 8.0, "learning_rate": 2.7447701842811923e-06, "loss": 1.3322, "step": 30626 }, { "epoch": 76.5675, "grad_norm": 8.75, "learning_rate": 2.7442129818814876e-06, "loss": 1.3468, "step": 30627 }, { "epoch": 76.57, "grad_norm": 4.6875, "learning_rate": 2.743655827050432e-06, "loss": 1.2566, "step": 30628 }, { "epoch": 76.5725, "grad_norm": 8.125, "learning_rate": 2.743098719791681e-06, "loss": 1.2955, "step": 30629 }, { "epoch": 76.575, "grad_norm": 7.6875, "learning_rate": 2.742541660108886e-06, "loss": 1.457, "step": 30630 }, { "epoch": 76.5775, "grad_norm": 6.3125, "learning_rate": 2.7419846480056933e-06, "loss": 1.3634, "step": 30631 }, { "epoch": 76.58, "grad_norm": 9.0625, "learning_rate": 2.741427683485759e-06, "loss": 1.341, "step": 30632 }, { "epoch": 76.5825, "grad_norm": 6.28125, "learning_rate": 2.7408707665527368e-06, "loss": 1.24, "step": 30633 }, { "epoch": 76.585, "grad_norm": 7.0625, "learning_rate": 2.740313897210276e-06, "loss": 1.4752, "step": 30634 }, { "epoch": 76.5875, "grad_norm": 4.84375, "learning_rate": 2.739757075462024e-06, "loss": 1.3858, "step": 30635 }, { "epoch": 76.59, "grad_norm": 6.5, "learning_rate": 2.7392003013116362e-06, "loss": 1.3084, "step": 30636 }, { "epoch": 76.5925, "grad_norm": 7.25, "learning_rate": 2.7386435747627593e-06, "loss": 1.1692, "step": 30637 }, { "epoch": 76.595, "grad_norm": 7.34375, "learning_rate": 2.7380868958190454e-06, "loss": 1.4121, "step": 30638 }, { "epoch": 76.5975, "grad_norm": 11.5, "learning_rate": 2.7375302644841417e-06, "loss": 1.4337, "step": 30639 }, { "epoch": 76.6, "grad_norm": 6.59375, "learning_rate": 2.736973680761702e-06, "loss": 1.1917, "step": 30640 }, { "epoch": 76.6025, "grad_norm": 8.125, "learning_rate": 2.736417144655371e-06, "loss": 1.4616, "step": 30641 }, { "epoch": 76.605, "grad_norm": 8.0, "learning_rate": 2.7358606561687962e-06, "loss": 1.2981, "step": 30642 }, { "epoch": 76.6075, "grad_norm": 8.25, "learning_rate": 2.7353042153056286e-06, "loss": 1.4642, "step": 30643 }, { "epoch": 76.61, "grad_norm": 7.21875, "learning_rate": 2.734747822069519e-06, "loss": 1.3279, "step": 30644 }, { "epoch": 76.6125, "grad_norm": 6.75, "learning_rate": 2.734191476464111e-06, "loss": 1.2235, "step": 30645 }, { "epoch": 76.615, "grad_norm": 12.0625, "learning_rate": 2.7336351784930535e-06, "loss": 1.3062, "step": 30646 }, { "epoch": 76.6175, "grad_norm": 8.125, "learning_rate": 2.7330789281599867e-06, "loss": 1.2261, "step": 30647 }, { "epoch": 76.62, "grad_norm": 8.5625, "learning_rate": 2.7325227254685704e-06, "loss": 1.3519, "step": 30648 }, { "epoch": 76.6225, "grad_norm": 8.5, "learning_rate": 2.7319665704224428e-06, "loss": 1.3862, "step": 30649 }, { "epoch": 76.625, "grad_norm": 7.90625, "learning_rate": 2.7314104630252502e-06, "loss": 1.1909, "step": 30650 }, { "epoch": 76.6275, "grad_norm": 9.375, "learning_rate": 2.7308544032806418e-06, "loss": 1.5091, "step": 30651 }, { "epoch": 76.63, "grad_norm": 6.90625, "learning_rate": 2.7302983911922576e-06, "loss": 1.432, "step": 30652 }, { "epoch": 76.6325, "grad_norm": 8.625, "learning_rate": 2.729742426763751e-06, "loss": 1.2943, "step": 30653 }, { "epoch": 76.635, "grad_norm": 6.0625, "learning_rate": 2.729186509998757e-06, "loss": 1.355, "step": 30654 }, { "epoch": 76.6375, "grad_norm": 7.53125, "learning_rate": 2.72863064090093e-06, "loss": 1.2883, "step": 30655 }, { "epoch": 76.64, "grad_norm": 7.28125, "learning_rate": 2.728074819473908e-06, "loss": 1.4233, "step": 30656 }, { "epoch": 76.6425, "grad_norm": 7.375, "learning_rate": 2.727519045721334e-06, "loss": 1.4078, "step": 30657 }, { "epoch": 76.645, "grad_norm": 8.1875, "learning_rate": 2.726963319646855e-06, "loss": 1.2957, "step": 30658 }, { "epoch": 76.6475, "grad_norm": 10.8125, "learning_rate": 2.7264076412541153e-06, "loss": 1.403, "step": 30659 }, { "epoch": 76.65, "grad_norm": 7.96875, "learning_rate": 2.7258520105467566e-06, "loss": 1.3141, "step": 30660 }, { "epoch": 76.6525, "grad_norm": 8.125, "learning_rate": 2.725296427528421e-06, "loss": 1.4053, "step": 30661 }, { "epoch": 76.655, "grad_norm": 5.90625, "learning_rate": 2.724740892202745e-06, "loss": 1.3863, "step": 30662 }, { "epoch": 76.6575, "grad_norm": 7.65625, "learning_rate": 2.7241854045733817e-06, "loss": 1.2596, "step": 30663 }, { "epoch": 76.66, "grad_norm": 6.5, "learning_rate": 2.723629964643969e-06, "loss": 1.1857, "step": 30664 }, { "epoch": 76.6625, "grad_norm": 6.1875, "learning_rate": 2.723074572418146e-06, "loss": 1.3045, "step": 30665 }, { "epoch": 76.665, "grad_norm": 6.0625, "learning_rate": 2.7225192278995496e-06, "loss": 1.2529, "step": 30666 }, { "epoch": 76.6675, "grad_norm": 5.78125, "learning_rate": 2.7219639310918324e-06, "loss": 1.2205, "step": 30667 }, { "epoch": 76.67, "grad_norm": 6.75, "learning_rate": 2.7214086819986272e-06, "loss": 1.2618, "step": 30668 }, { "epoch": 76.6725, "grad_norm": 5.65625, "learning_rate": 2.7208534806235766e-06, "loss": 1.3511, "step": 30669 }, { "epoch": 76.675, "grad_norm": 8.4375, "learning_rate": 2.7202983269703153e-06, "loss": 1.1743, "step": 30670 }, { "epoch": 76.6775, "grad_norm": 8.0625, "learning_rate": 2.7197432210424867e-06, "loss": 1.1661, "step": 30671 }, { "epoch": 76.68, "grad_norm": 8.3125, "learning_rate": 2.7191881628437335e-06, "loss": 1.4332, "step": 30672 }, { "epoch": 76.6825, "grad_norm": 9.125, "learning_rate": 2.718633152377692e-06, "loss": 1.4695, "step": 30673 }, { "epoch": 76.685, "grad_norm": 9.5625, "learning_rate": 2.718078189647997e-06, "loss": 1.5505, "step": 30674 }, { "epoch": 76.6875, "grad_norm": 7.84375, "learning_rate": 2.7175232746582926e-06, "loss": 1.3805, "step": 30675 }, { "epoch": 76.69, "grad_norm": 8.5, "learning_rate": 2.7169684074122105e-06, "loss": 1.2792, "step": 30676 }, { "epoch": 76.6925, "grad_norm": 8.875, "learning_rate": 2.7164135879133935e-06, "loss": 1.2693, "step": 30677 }, { "epoch": 76.695, "grad_norm": 9.5625, "learning_rate": 2.7158588161654795e-06, "loss": 1.2346, "step": 30678 }, { "epoch": 76.6975, "grad_norm": 5.0625, "learning_rate": 2.715304092172104e-06, "loss": 1.3718, "step": 30679 }, { "epoch": 76.7, "grad_norm": 7.5625, "learning_rate": 2.714749415936904e-06, "loss": 1.183, "step": 30680 }, { "epoch": 76.7025, "grad_norm": 7.71875, "learning_rate": 2.714194787463508e-06, "loss": 1.3914, "step": 30681 }, { "epoch": 76.705, "grad_norm": 6.46875, "learning_rate": 2.7136402067555658e-06, "loss": 1.4652, "step": 30682 }, { "epoch": 76.7075, "grad_norm": 9.5, "learning_rate": 2.7130856738167066e-06, "loss": 1.472, "step": 30683 }, { "epoch": 76.71, "grad_norm": 7.40625, "learning_rate": 2.7125311886505665e-06, "loss": 1.3208, "step": 30684 }, { "epoch": 76.7125, "grad_norm": 10.6875, "learning_rate": 2.7119767512607774e-06, "loss": 1.2968, "step": 30685 }, { "epoch": 76.715, "grad_norm": 4.46875, "learning_rate": 2.711422361650977e-06, "loss": 1.348, "step": 30686 }, { "epoch": 76.7175, "grad_norm": 6.90625, "learning_rate": 2.7108680198248037e-06, "loss": 1.3687, "step": 30687 }, { "epoch": 76.72, "grad_norm": 7.1875, "learning_rate": 2.7103137257858867e-06, "loss": 1.3154, "step": 30688 }, { "epoch": 76.7225, "grad_norm": 10.625, "learning_rate": 2.709759479537859e-06, "loss": 1.3995, "step": 30689 }, { "epoch": 76.725, "grad_norm": 6.3125, "learning_rate": 2.7092052810843592e-06, "loss": 1.3316, "step": 30690 }, { "epoch": 76.7275, "grad_norm": 8.125, "learning_rate": 2.708651130429014e-06, "loss": 1.4042, "step": 30691 }, { "epoch": 76.73, "grad_norm": 6.09375, "learning_rate": 2.7080970275754647e-06, "loss": 1.3225, "step": 30692 }, { "epoch": 76.7325, "grad_norm": 16.75, "learning_rate": 2.707542972527335e-06, "loss": 1.2289, "step": 30693 }, { "epoch": 76.735, "grad_norm": 9.6875, "learning_rate": 2.7069889652882662e-06, "loss": 1.2625, "step": 30694 }, { "epoch": 76.7375, "grad_norm": 7.59375, "learning_rate": 2.706435005861885e-06, "loss": 1.3471, "step": 30695 }, { "epoch": 76.74, "grad_norm": 7.5, "learning_rate": 2.705881094251821e-06, "loss": 1.3457, "step": 30696 }, { "epoch": 76.7425, "grad_norm": 6.28125, "learning_rate": 2.7053272304617084e-06, "loss": 1.3902, "step": 30697 }, { "epoch": 76.745, "grad_norm": 9.9375, "learning_rate": 2.704773414495182e-06, "loss": 1.3273, "step": 30698 }, { "epoch": 76.7475, "grad_norm": 9.1875, "learning_rate": 2.70421964635587e-06, "loss": 1.4032, "step": 30699 }, { "epoch": 76.75, "grad_norm": 5.5625, "learning_rate": 2.7036659260473973e-06, "loss": 1.3477, "step": 30700 }, { "epoch": 76.7525, "grad_norm": 8.5, "learning_rate": 2.7031122535734e-06, "loss": 1.3069, "step": 30701 }, { "epoch": 76.755, "grad_norm": 6.78125, "learning_rate": 2.7025586289375092e-06, "loss": 1.3269, "step": 30702 }, { "epoch": 76.7575, "grad_norm": 7.71875, "learning_rate": 2.7020050521433525e-06, "loss": 1.2995, "step": 30703 }, { "epoch": 76.76, "grad_norm": 6.625, "learning_rate": 2.7014515231945557e-06, "loss": 1.2769, "step": 30704 }, { "epoch": 76.7625, "grad_norm": 8.625, "learning_rate": 2.700898042094754e-06, "loss": 1.2299, "step": 30705 }, { "epoch": 76.765, "grad_norm": 5.375, "learning_rate": 2.700344608847568e-06, "loss": 1.1383, "step": 30706 }, { "epoch": 76.7675, "grad_norm": 9.125, "learning_rate": 2.6997912234566346e-06, "loss": 1.2426, "step": 30707 }, { "epoch": 76.77, "grad_norm": 7.0, "learning_rate": 2.6992378859255754e-06, "loss": 1.2225, "step": 30708 }, { "epoch": 76.7725, "grad_norm": 8.75, "learning_rate": 2.698684596258022e-06, "loss": 1.2187, "step": 30709 }, { "epoch": 76.775, "grad_norm": 6.96875, "learning_rate": 2.6981313544576016e-06, "loss": 1.3635, "step": 30710 }, { "epoch": 76.7775, "grad_norm": 4.8125, "learning_rate": 2.697578160527936e-06, "loss": 1.3007, "step": 30711 }, { "epoch": 76.78, "grad_norm": 6.5, "learning_rate": 2.6970250144726563e-06, "loss": 1.5431, "step": 30712 }, { "epoch": 76.7825, "grad_norm": 10.25, "learning_rate": 2.696471916295391e-06, "loss": 1.3414, "step": 30713 }, { "epoch": 76.785, "grad_norm": 9.9375, "learning_rate": 2.6959188659997636e-06, "loss": 1.2047, "step": 30714 }, { "epoch": 76.7875, "grad_norm": 6.9375, "learning_rate": 2.6953658635893977e-06, "loss": 1.3902, "step": 30715 }, { "epoch": 76.79, "grad_norm": 5.75, "learning_rate": 2.69481290906792e-06, "loss": 1.2578, "step": 30716 }, { "epoch": 76.7925, "grad_norm": 8.6875, "learning_rate": 2.6942600024389597e-06, "loss": 1.4842, "step": 30717 }, { "epoch": 76.795, "grad_norm": 7.0, "learning_rate": 2.693707143706138e-06, "loss": 1.275, "step": 30718 }, { "epoch": 76.7975, "grad_norm": 13.25, "learning_rate": 2.693154332873077e-06, "loss": 1.4486, "step": 30719 }, { "epoch": 76.8, "grad_norm": 9.0625, "learning_rate": 2.692601569943407e-06, "loss": 1.3194, "step": 30720 }, { "epoch": 76.8025, "grad_norm": 10.1875, "learning_rate": 2.6920488549207458e-06, "loss": 1.1383, "step": 30721 }, { "epoch": 76.805, "grad_norm": 6.0625, "learning_rate": 2.691496187808723e-06, "loss": 1.1971, "step": 30722 }, { "epoch": 76.8075, "grad_norm": 5.65625, "learning_rate": 2.690943568610955e-06, "loss": 1.254, "step": 30723 }, { "epoch": 76.81, "grad_norm": 7.84375, "learning_rate": 2.6903909973310725e-06, "loss": 1.2981, "step": 30724 }, { "epoch": 76.8125, "grad_norm": 8.25, "learning_rate": 2.6898384739726934e-06, "loss": 1.3713, "step": 30725 }, { "epoch": 76.815, "grad_norm": 6.875, "learning_rate": 2.689285998539438e-06, "loss": 1.4193, "step": 30726 }, { "epoch": 76.8175, "grad_norm": 6.40625, "learning_rate": 2.6887335710349305e-06, "loss": 1.2035, "step": 30727 }, { "epoch": 76.82, "grad_norm": 9.875, "learning_rate": 2.6881811914627976e-06, "loss": 1.3878, "step": 30728 }, { "epoch": 76.8225, "grad_norm": 8.3125, "learning_rate": 2.6876288598266543e-06, "loss": 1.39, "step": 30729 }, { "epoch": 76.825, "grad_norm": 8.125, "learning_rate": 2.6870765761301223e-06, "loss": 1.2205, "step": 30730 }, { "epoch": 76.8275, "grad_norm": 12.25, "learning_rate": 2.6865243403768227e-06, "loss": 1.463, "step": 30731 }, { "epoch": 76.83, "grad_norm": 6.5, "learning_rate": 2.6859721525703797e-06, "loss": 1.2365, "step": 30732 }, { "epoch": 76.8325, "grad_norm": 4.875, "learning_rate": 2.6854200127144104e-06, "loss": 1.3776, "step": 30733 }, { "epoch": 76.835, "grad_norm": 6.09375, "learning_rate": 2.684867920812533e-06, "loss": 1.2395, "step": 30734 }, { "epoch": 76.8375, "grad_norm": 5.90625, "learning_rate": 2.684315876868371e-06, "loss": 1.1704, "step": 30735 }, { "epoch": 76.84, "grad_norm": 7.4375, "learning_rate": 2.683763880885538e-06, "loss": 1.357, "step": 30736 }, { "epoch": 76.8425, "grad_norm": 7.09375, "learning_rate": 2.68321193286766e-06, "loss": 1.2276, "step": 30737 }, { "epoch": 76.845, "grad_norm": 7.875, "learning_rate": 2.6826600328183474e-06, "loss": 1.3442, "step": 30738 }, { "epoch": 76.8475, "grad_norm": 8.125, "learning_rate": 2.6821081807412264e-06, "loss": 1.2658, "step": 30739 }, { "epoch": 76.85, "grad_norm": 6.71875, "learning_rate": 2.6815563766399122e-06, "loss": 1.1553, "step": 30740 }, { "epoch": 76.8525, "grad_norm": 9.3125, "learning_rate": 2.6810046205180175e-06, "loss": 1.4417, "step": 30741 }, { "epoch": 76.855, "grad_norm": 12.0625, "learning_rate": 2.6804529123791644e-06, "loss": 1.2755, "step": 30742 }, { "epoch": 76.8575, "grad_norm": 7.71875, "learning_rate": 2.6799012522269716e-06, "loss": 1.2064, "step": 30743 }, { "epoch": 76.86, "grad_norm": 8.8125, "learning_rate": 2.679349640065053e-06, "loss": 1.4635, "step": 30744 }, { "epoch": 76.8625, "grad_norm": 6.96875, "learning_rate": 2.6787980758970235e-06, "loss": 1.1646, "step": 30745 }, { "epoch": 76.865, "grad_norm": 9.4375, "learning_rate": 2.6782465597264994e-06, "loss": 1.5165, "step": 30746 }, { "epoch": 76.8675, "grad_norm": 7.6875, "learning_rate": 2.677695091557102e-06, "loss": 1.4065, "step": 30747 }, { "epoch": 76.87, "grad_norm": 7.0625, "learning_rate": 2.6771436713924417e-06, "loss": 1.3416, "step": 30748 }, { "epoch": 76.8725, "grad_norm": 7.28125, "learning_rate": 2.6765922992361317e-06, "loss": 1.1604, "step": 30749 }, { "epoch": 76.875, "grad_norm": 8.5625, "learning_rate": 2.6760409750917925e-06, "loss": 1.4586, "step": 30750 }, { "epoch": 76.8775, "grad_norm": 11.125, "learning_rate": 2.675489698963033e-06, "loss": 1.3444, "step": 30751 }, { "epoch": 76.88, "grad_norm": 8.0625, "learning_rate": 2.674938470853472e-06, "loss": 1.2944, "step": 30752 }, { "epoch": 76.8825, "grad_norm": 7.3125, "learning_rate": 2.674387290766719e-06, "loss": 1.376, "step": 30753 }, { "epoch": 76.885, "grad_norm": 8.375, "learning_rate": 2.6738361587063922e-06, "loss": 1.2555, "step": 30754 }, { "epoch": 76.8875, "grad_norm": 6.53125, "learning_rate": 2.673285074676103e-06, "loss": 1.2338, "step": 30755 }, { "epoch": 76.89, "grad_norm": 6.53125, "learning_rate": 2.672734038679461e-06, "loss": 1.419, "step": 30756 }, { "epoch": 76.8925, "grad_norm": 6.5625, "learning_rate": 2.6721830507200806e-06, "loss": 1.219, "step": 30757 }, { "epoch": 76.895, "grad_norm": 7.6875, "learning_rate": 2.6716321108015775e-06, "loss": 1.3278, "step": 30758 }, { "epoch": 76.8975, "grad_norm": 6.96875, "learning_rate": 2.6710812189275625e-06, "loss": 1.4249, "step": 30759 }, { "epoch": 76.9, "grad_norm": 7.625, "learning_rate": 2.670530375101641e-06, "loss": 1.3447, "step": 30760 }, { "epoch": 76.9025, "grad_norm": 9.0625, "learning_rate": 2.6699795793274295e-06, "loss": 1.2646, "step": 30761 }, { "epoch": 76.905, "grad_norm": 6.09375, "learning_rate": 2.6694288316085427e-06, "loss": 1.372, "step": 30762 }, { "epoch": 76.9075, "grad_norm": 6.6875, "learning_rate": 2.6688781319485866e-06, "loss": 1.4324, "step": 30763 }, { "epoch": 76.91, "grad_norm": 5.90625, "learning_rate": 2.668327480351168e-06, "loss": 1.339, "step": 30764 }, { "epoch": 76.9125, "grad_norm": 9.75, "learning_rate": 2.667776876819905e-06, "loss": 1.2623, "step": 30765 }, { "epoch": 76.915, "grad_norm": 7.59375, "learning_rate": 2.6672263213584005e-06, "loss": 1.2431, "step": 30766 }, { "epoch": 76.9175, "grad_norm": 5.3125, "learning_rate": 2.66667581397027e-06, "loss": 1.2986, "step": 30767 }, { "epoch": 76.92, "grad_norm": 8.0625, "learning_rate": 2.6661253546591158e-06, "loss": 1.368, "step": 30768 }, { "epoch": 76.9225, "grad_norm": 5.125, "learning_rate": 2.665574943428554e-06, "loss": 1.2149, "step": 30769 }, { "epoch": 76.925, "grad_norm": 8.5625, "learning_rate": 2.6650245802821883e-06, "loss": 1.2752, "step": 30770 }, { "epoch": 76.9275, "grad_norm": 10.0, "learning_rate": 2.6644742652236254e-06, "loss": 1.2456, "step": 30771 }, { "epoch": 76.93, "grad_norm": 7.8125, "learning_rate": 2.6639239982564757e-06, "loss": 1.3506, "step": 30772 }, { "epoch": 76.9325, "grad_norm": 6.46875, "learning_rate": 2.6633737793843493e-06, "loss": 1.3815, "step": 30773 }, { "epoch": 76.935, "grad_norm": 7.3125, "learning_rate": 2.6628236086108516e-06, "loss": 1.3591, "step": 30774 }, { "epoch": 76.9375, "grad_norm": 6.5, "learning_rate": 2.662273485939586e-06, "loss": 1.3667, "step": 30775 }, { "epoch": 76.94, "grad_norm": 7.5, "learning_rate": 2.661723411374161e-06, "loss": 1.3438, "step": 30776 }, { "epoch": 76.9425, "grad_norm": 6.0, "learning_rate": 2.6611733849181863e-06, "loss": 1.2994, "step": 30777 }, { "epoch": 76.945, "grad_norm": 8.9375, "learning_rate": 2.660623406575266e-06, "loss": 1.2956, "step": 30778 }, { "epoch": 76.9475, "grad_norm": 6.53125, "learning_rate": 2.660073476349001e-06, "loss": 1.2294, "step": 30779 }, { "epoch": 76.95, "grad_norm": 9.75, "learning_rate": 2.6595235942430044e-06, "loss": 1.3346, "step": 30780 }, { "epoch": 76.9525, "grad_norm": 6.5, "learning_rate": 2.6589737602608747e-06, "loss": 1.3319, "step": 30781 }, { "epoch": 76.955, "grad_norm": 9.5625, "learning_rate": 2.658423974406221e-06, "loss": 1.4066, "step": 30782 }, { "epoch": 76.9575, "grad_norm": 6.875, "learning_rate": 2.657874236682644e-06, "loss": 1.1918, "step": 30783 }, { "epoch": 76.96, "grad_norm": 6.3125, "learning_rate": 2.6573245470937527e-06, "loss": 1.3111, "step": 30784 }, { "epoch": 76.9625, "grad_norm": 4.03125, "learning_rate": 2.656774905643147e-06, "loss": 1.3734, "step": 30785 }, { "epoch": 76.965, "grad_norm": 10.0, "learning_rate": 2.6562253123344285e-06, "loss": 1.301, "step": 30786 }, { "epoch": 76.9675, "grad_norm": 8.3125, "learning_rate": 2.655675767171203e-06, "loss": 1.4203, "step": 30787 }, { "epoch": 76.97, "grad_norm": 7.5, "learning_rate": 2.6551262701570768e-06, "loss": 1.27, "step": 30788 }, { "epoch": 76.9725, "grad_norm": 8.5, "learning_rate": 2.654576821295648e-06, "loss": 1.3848, "step": 30789 }, { "epoch": 76.975, "grad_norm": 8.625, "learning_rate": 2.654027420590517e-06, "loss": 1.3335, "step": 30790 }, { "epoch": 76.9775, "grad_norm": 8.5625, "learning_rate": 2.653478068045289e-06, "loss": 1.2504, "step": 30791 }, { "epoch": 76.98, "grad_norm": 6.40625, "learning_rate": 2.652928763663567e-06, "loss": 1.3495, "step": 30792 }, { "epoch": 76.9825, "grad_norm": 8.875, "learning_rate": 2.65237950744895e-06, "loss": 1.1278, "step": 30793 }, { "epoch": 76.985, "grad_norm": 7.6875, "learning_rate": 2.651830299405036e-06, "loss": 1.3579, "step": 30794 }, { "epoch": 76.9875, "grad_norm": 9.4375, "learning_rate": 2.651281139535431e-06, "loss": 1.3443, "step": 30795 }, { "epoch": 76.99, "grad_norm": 7.71875, "learning_rate": 2.6507320278437296e-06, "loss": 1.4178, "step": 30796 }, { "epoch": 76.9925, "grad_norm": 6.125, "learning_rate": 2.6501829643335396e-06, "loss": 1.1735, "step": 30797 }, { "epoch": 76.995, "grad_norm": 6.5625, "learning_rate": 2.6496339490084524e-06, "loss": 1.4066, "step": 30798 }, { "epoch": 76.9975, "grad_norm": 7.53125, "learning_rate": 2.649084981872073e-06, "loss": 1.4261, "step": 30799 }, { "epoch": 77.0, "grad_norm": 5.5, "learning_rate": 2.648536062927999e-06, "loss": 1.2637, "step": 30800 }, { "epoch": 77.0025, "grad_norm": 6.375, "learning_rate": 2.647987192179825e-06, "loss": 1.367, "step": 30801 }, { "epoch": 77.005, "grad_norm": 8.6875, "learning_rate": 2.647438369631152e-06, "loss": 1.225, "step": 30802 }, { "epoch": 77.0075, "grad_norm": 7.21875, "learning_rate": 2.646889595285583e-06, "loss": 1.316, "step": 30803 }, { "epoch": 77.01, "grad_norm": 7.03125, "learning_rate": 2.6463408691467107e-06, "loss": 1.2743, "step": 30804 }, { "epoch": 77.0125, "grad_norm": 8.125, "learning_rate": 2.645792191218135e-06, "loss": 1.3064, "step": 30805 }, { "epoch": 77.015, "grad_norm": 6.125, "learning_rate": 2.6452435615034434e-06, "loss": 1.2513, "step": 30806 }, { "epoch": 77.0175, "grad_norm": 9.5, "learning_rate": 2.6446949800062484e-06, "loss": 1.39, "step": 30807 }, { "epoch": 77.02, "grad_norm": 6.75, "learning_rate": 2.644146446730138e-06, "loss": 1.4017, "step": 30808 }, { "epoch": 77.0225, "grad_norm": 8.3125, "learning_rate": 2.643597961678709e-06, "loss": 1.3549, "step": 30809 }, { "epoch": 77.025, "grad_norm": 8.75, "learning_rate": 2.6430495248555545e-06, "loss": 1.3727, "step": 30810 }, { "epoch": 77.0275, "grad_norm": 5.96875, "learning_rate": 2.6425011362642726e-06, "loss": 1.2667, "step": 30811 }, { "epoch": 77.03, "grad_norm": 7.09375, "learning_rate": 2.641952795908462e-06, "loss": 1.492, "step": 30812 }, { "epoch": 77.0325, "grad_norm": 6.9375, "learning_rate": 2.6414045037917113e-06, "loss": 1.3115, "step": 30813 }, { "epoch": 77.035, "grad_norm": 6.40625, "learning_rate": 2.640856259917621e-06, "loss": 1.3543, "step": 30814 }, { "epoch": 77.0375, "grad_norm": 8.5, "learning_rate": 2.640308064289784e-06, "loss": 1.412, "step": 30815 }, { "epoch": 77.04, "grad_norm": 6.625, "learning_rate": 2.639759916911788e-06, "loss": 1.4392, "step": 30816 }, { "epoch": 77.0425, "grad_norm": 6.71875, "learning_rate": 2.639211817787233e-06, "loss": 1.2873, "step": 30817 }, { "epoch": 77.045, "grad_norm": 8.5625, "learning_rate": 2.6386637669197125e-06, "loss": 1.3759, "step": 30818 }, { "epoch": 77.0475, "grad_norm": 6.53125, "learning_rate": 2.6381157643128184e-06, "loss": 1.3205, "step": 30819 }, { "epoch": 77.05, "grad_norm": 6.8125, "learning_rate": 2.637567809970143e-06, "loss": 1.2669, "step": 30820 }, { "epoch": 77.0525, "grad_norm": 6.96875, "learning_rate": 2.6370199038952717e-06, "loss": 1.3982, "step": 30821 }, { "epoch": 77.055, "grad_norm": 6.4375, "learning_rate": 2.636472046091809e-06, "loss": 1.3134, "step": 30822 }, { "epoch": 77.0575, "grad_norm": 7.8125, "learning_rate": 2.635924236563342e-06, "loss": 1.2795, "step": 30823 }, { "epoch": 77.06, "grad_norm": 12.8125, "learning_rate": 2.6353764753134603e-06, "loss": 1.2315, "step": 30824 }, { "epoch": 77.0625, "grad_norm": 10.25, "learning_rate": 2.6348287623457534e-06, "loss": 1.3191, "step": 30825 }, { "epoch": 77.065, "grad_norm": 7.75, "learning_rate": 2.6342810976638132e-06, "loss": 1.2947, "step": 30826 }, { "epoch": 77.0675, "grad_norm": 7.1875, "learning_rate": 2.6337334812712357e-06, "loss": 1.1425, "step": 30827 }, { "epoch": 77.07, "grad_norm": 10.375, "learning_rate": 2.6331859131716064e-06, "loss": 1.3126, "step": 30828 }, { "epoch": 77.0725, "grad_norm": 5.8125, "learning_rate": 2.6326383933685117e-06, "loss": 1.5099, "step": 30829 }, { "epoch": 77.075, "grad_norm": 7.3125, "learning_rate": 2.632090921865549e-06, "loss": 1.2111, "step": 30830 }, { "epoch": 77.0775, "grad_norm": 7.1875, "learning_rate": 2.631543498666299e-06, "loss": 1.3939, "step": 30831 }, { "epoch": 77.08, "grad_norm": 7.875, "learning_rate": 2.6309961237743587e-06, "loss": 1.4751, "step": 30832 }, { "epoch": 77.0825, "grad_norm": 10.3125, "learning_rate": 2.63044879719331e-06, "loss": 1.377, "step": 30833 }, { "epoch": 77.085, "grad_norm": 6.78125, "learning_rate": 2.629901518926746e-06, "loss": 1.3267, "step": 30834 }, { "epoch": 77.0875, "grad_norm": 6.96875, "learning_rate": 2.629354288978254e-06, "loss": 1.2887, "step": 30835 }, { "epoch": 77.09, "grad_norm": 6.96875, "learning_rate": 2.6288071073514176e-06, "loss": 1.3741, "step": 30836 }, { "epoch": 77.0925, "grad_norm": 6.65625, "learning_rate": 2.628259974049826e-06, "loss": 1.2314, "step": 30837 }, { "epoch": 77.095, "grad_norm": 5.84375, "learning_rate": 2.6277128890770697e-06, "loss": 1.3083, "step": 30838 }, { "epoch": 77.0975, "grad_norm": 5.46875, "learning_rate": 2.627165852436734e-06, "loss": 1.1531, "step": 30839 }, { "epoch": 77.1, "grad_norm": 5.25, "learning_rate": 2.6266188641324e-06, "loss": 1.322, "step": 30840 }, { "epoch": 77.1025, "grad_norm": 5.9375, "learning_rate": 2.626071924167657e-06, "loss": 1.1792, "step": 30841 }, { "epoch": 77.105, "grad_norm": 6.8125, "learning_rate": 2.6255250325460955e-06, "loss": 1.3332, "step": 30842 }, { "epoch": 77.1075, "grad_norm": 7.25, "learning_rate": 2.624978189271297e-06, "loss": 1.4633, "step": 30843 }, { "epoch": 77.11, "grad_norm": 9.25, "learning_rate": 2.6244313943468426e-06, "loss": 1.3381, "step": 30844 }, { "epoch": 77.1125, "grad_norm": 8.75, "learning_rate": 2.6238846477763234e-06, "loss": 1.3313, "step": 30845 }, { "epoch": 77.115, "grad_norm": 8.0625, "learning_rate": 2.6233379495633195e-06, "loss": 1.3196, "step": 30846 }, { "epoch": 77.1175, "grad_norm": 10.0625, "learning_rate": 2.622791299711419e-06, "loss": 1.3766, "step": 30847 }, { "epoch": 77.12, "grad_norm": 9.6875, "learning_rate": 2.6222446982242e-06, "loss": 1.5043, "step": 30848 }, { "epoch": 77.1225, "grad_norm": 6.96875, "learning_rate": 2.6216981451052538e-06, "loss": 1.3059, "step": 30849 }, { "epoch": 77.125, "grad_norm": 6.75, "learning_rate": 2.6211516403581585e-06, "loss": 1.2674, "step": 30850 }, { "epoch": 77.1275, "grad_norm": 6.21875, "learning_rate": 2.620605183986494e-06, "loss": 1.2306, "step": 30851 }, { "epoch": 77.13, "grad_norm": 7.28125, "learning_rate": 2.620058775993849e-06, "loss": 1.335, "step": 30852 }, { "epoch": 77.1325, "grad_norm": 5.96875, "learning_rate": 2.619512416383805e-06, "loss": 1.1322, "step": 30853 }, { "epoch": 77.135, "grad_norm": 8.125, "learning_rate": 2.6189661051599414e-06, "loss": 1.2191, "step": 30854 }, { "epoch": 77.1375, "grad_norm": 7.8125, "learning_rate": 2.6184198423258387e-06, "loss": 1.2362, "step": 30855 }, { "epoch": 77.14, "grad_norm": 8.1875, "learning_rate": 2.6178736278850792e-06, "loss": 1.2897, "step": 30856 }, { "epoch": 77.1425, "grad_norm": 7.8125, "learning_rate": 2.6173274618412493e-06, "loss": 1.3392, "step": 30857 }, { "epoch": 77.145, "grad_norm": 6.46875, "learning_rate": 2.616781344197924e-06, "loss": 1.3571, "step": 30858 }, { "epoch": 77.1475, "grad_norm": 10.25, "learning_rate": 2.6162352749586815e-06, "loss": 1.2531, "step": 30859 }, { "epoch": 77.15, "grad_norm": 9.625, "learning_rate": 2.6156892541271083e-06, "loss": 1.5373, "step": 30860 }, { "epoch": 77.1525, "grad_norm": 11.8125, "learning_rate": 2.6151432817067778e-06, "loss": 1.3191, "step": 30861 }, { "epoch": 77.155, "grad_norm": 6.4375, "learning_rate": 2.6145973577012763e-06, "loss": 1.1657, "step": 30862 }, { "epoch": 77.1575, "grad_norm": 9.8125, "learning_rate": 2.614051482114175e-06, "loss": 1.209, "step": 30863 }, { "epoch": 77.16, "grad_norm": 6.03125, "learning_rate": 2.61350565494906e-06, "loss": 1.211, "step": 30864 }, { "epoch": 77.1625, "grad_norm": 7.21875, "learning_rate": 2.6129598762095063e-06, "loss": 1.3719, "step": 30865 }, { "epoch": 77.165, "grad_norm": 5.90625, "learning_rate": 2.612414145899089e-06, "loss": 1.3347, "step": 30866 }, { "epoch": 77.1675, "grad_norm": 13.5, "learning_rate": 2.61186846402139e-06, "loss": 1.4083, "step": 30867 }, { "epoch": 77.17, "grad_norm": 7.84375, "learning_rate": 2.6113228305799867e-06, "loss": 1.2275, "step": 30868 }, { "epoch": 77.1725, "grad_norm": 7.59375, "learning_rate": 2.610777245578456e-06, "loss": 1.3961, "step": 30869 }, { "epoch": 77.175, "grad_norm": 6.75, "learning_rate": 2.610231709020372e-06, "loss": 1.5188, "step": 30870 }, { "epoch": 77.1775, "grad_norm": 8.75, "learning_rate": 2.609686220909312e-06, "loss": 1.2728, "step": 30871 }, { "epoch": 77.18, "grad_norm": 5.3125, "learning_rate": 2.6091407812488567e-06, "loss": 1.305, "step": 30872 }, { "epoch": 77.1825, "grad_norm": 8.375, "learning_rate": 2.6085953900425786e-06, "loss": 1.2914, "step": 30873 }, { "epoch": 77.185, "grad_norm": 9.0, "learning_rate": 2.6080500472940507e-06, "loss": 1.309, "step": 30874 }, { "epoch": 77.1875, "grad_norm": 6.84375, "learning_rate": 2.6075047530068544e-06, "loss": 1.1639, "step": 30875 }, { "epoch": 77.19, "grad_norm": 9.375, "learning_rate": 2.6069595071845566e-06, "loss": 1.3232, "step": 30876 }, { "epoch": 77.1925, "grad_norm": 6.5, "learning_rate": 2.6064143098307406e-06, "loss": 1.4037, "step": 30877 }, { "epoch": 77.195, "grad_norm": 5.65625, "learning_rate": 2.6058691609489728e-06, "loss": 1.2029, "step": 30878 }, { "epoch": 77.1975, "grad_norm": 9.3125, "learning_rate": 2.605324060542834e-06, "loss": 1.3544, "step": 30879 }, { "epoch": 77.2, "grad_norm": 11.5625, "learning_rate": 2.604779008615895e-06, "loss": 1.4184, "step": 30880 }, { "epoch": 77.2025, "grad_norm": 6.4375, "learning_rate": 2.604234005171726e-06, "loss": 1.3573, "step": 30881 }, { "epoch": 77.205, "grad_norm": 9.25, "learning_rate": 2.603689050213902e-06, "loss": 1.3135, "step": 30882 }, { "epoch": 77.2075, "grad_norm": 5.9375, "learning_rate": 2.6031441437460012e-06, "loss": 1.4054, "step": 30883 }, { "epoch": 77.21, "grad_norm": 6.5625, "learning_rate": 2.60259928577159e-06, "loss": 1.3421, "step": 30884 }, { "epoch": 77.2125, "grad_norm": 7.75, "learning_rate": 2.6020544762942403e-06, "loss": 1.4251, "step": 30885 }, { "epoch": 77.215, "grad_norm": 7.90625, "learning_rate": 2.601509715317524e-06, "loss": 1.3279, "step": 30886 }, { "epoch": 77.2175, "grad_norm": 8.8125, "learning_rate": 2.6009650028450175e-06, "loss": 1.3253, "step": 30887 }, { "epoch": 77.22, "grad_norm": 8.5, "learning_rate": 2.6004203388802883e-06, "loss": 1.3364, "step": 30888 }, { "epoch": 77.2225, "grad_norm": 6.59375, "learning_rate": 2.5998757234269033e-06, "loss": 1.4204, "step": 30889 }, { "epoch": 77.225, "grad_norm": 10.0, "learning_rate": 2.5993311564884417e-06, "loss": 1.4747, "step": 30890 }, { "epoch": 77.2275, "grad_norm": 7.1875, "learning_rate": 2.5987866380684644e-06, "loss": 1.4243, "step": 30891 }, { "epoch": 77.23, "grad_norm": 7.9375, "learning_rate": 2.5982421681705493e-06, "loss": 1.15, "step": 30892 }, { "epoch": 77.2325, "grad_norm": 7.21875, "learning_rate": 2.59769774679826e-06, "loss": 1.391, "step": 30893 }, { "epoch": 77.235, "grad_norm": 6.9375, "learning_rate": 2.5971533739551703e-06, "loss": 1.2162, "step": 30894 }, { "epoch": 77.2375, "grad_norm": 5.75, "learning_rate": 2.596609049644847e-06, "loss": 1.3078, "step": 30895 }, { "epoch": 77.24, "grad_norm": 6.46875, "learning_rate": 2.5960647738708553e-06, "loss": 1.1661, "step": 30896 }, { "epoch": 77.2425, "grad_norm": 5.6875, "learning_rate": 2.595520546636766e-06, "loss": 1.313, "step": 30897 }, { "epoch": 77.245, "grad_norm": 5.46875, "learning_rate": 2.5949763679461515e-06, "loss": 1.2734, "step": 30898 }, { "epoch": 77.2475, "grad_norm": 4.78125, "learning_rate": 2.5944322378025765e-06, "loss": 1.5434, "step": 30899 }, { "epoch": 77.25, "grad_norm": 6.0625, "learning_rate": 2.593888156209603e-06, "loss": 1.1764, "step": 30900 }, { "epoch": 77.2525, "grad_norm": 6.9375, "learning_rate": 2.5933441231708032e-06, "loss": 1.4262, "step": 30901 }, { "epoch": 77.255, "grad_norm": 7.6875, "learning_rate": 2.592800138689745e-06, "loss": 1.4512, "step": 30902 }, { "epoch": 77.2575, "grad_norm": 8.5625, "learning_rate": 2.592256202769994e-06, "loss": 1.2552, "step": 30903 }, { "epoch": 77.26, "grad_norm": 9.5625, "learning_rate": 2.59171231541511e-06, "loss": 1.3313, "step": 30904 }, { "epoch": 77.2625, "grad_norm": 7.09375, "learning_rate": 2.591168476628668e-06, "loss": 1.3105, "step": 30905 }, { "epoch": 77.265, "grad_norm": 7.6875, "learning_rate": 2.590624686414226e-06, "loss": 1.2156, "step": 30906 }, { "epoch": 77.2675, "grad_norm": 7.59375, "learning_rate": 2.590080944775355e-06, "loss": 1.5234, "step": 30907 }, { "epoch": 77.27, "grad_norm": 8.0, "learning_rate": 2.5895372517156137e-06, "loss": 1.2801, "step": 30908 }, { "epoch": 77.2725, "grad_norm": 5.1875, "learning_rate": 2.5889936072385722e-06, "loss": 1.1946, "step": 30909 }, { "epoch": 77.275, "grad_norm": 6.8125, "learning_rate": 2.5884500113477895e-06, "loss": 1.4812, "step": 30910 }, { "epoch": 77.2775, "grad_norm": 8.0, "learning_rate": 2.587906464046834e-06, "loss": 1.3077, "step": 30911 }, { "epoch": 77.28, "grad_norm": 7.1875, "learning_rate": 2.5873629653392653e-06, "loss": 1.3378, "step": 30912 }, { "epoch": 77.2825, "grad_norm": 7.25, "learning_rate": 2.586819515228651e-06, "loss": 1.448, "step": 30913 }, { "epoch": 77.285, "grad_norm": 6.6875, "learning_rate": 2.586276113718551e-06, "loss": 1.3628, "step": 30914 }, { "epoch": 77.2875, "grad_norm": 8.1875, "learning_rate": 2.585732760812526e-06, "loss": 1.256, "step": 30915 }, { "epoch": 77.29, "grad_norm": 7.96875, "learning_rate": 2.585189456514139e-06, "loss": 1.2421, "step": 30916 }, { "epoch": 77.2925, "grad_norm": 8.0625, "learning_rate": 2.5846462008269567e-06, "loss": 1.3232, "step": 30917 }, { "epoch": 77.295, "grad_norm": 6.0625, "learning_rate": 2.584102993754538e-06, "loss": 1.2862, "step": 30918 }, { "epoch": 77.2975, "grad_norm": 9.8125, "learning_rate": 2.58355983530044e-06, "loss": 1.2193, "step": 30919 }, { "epoch": 77.3, "grad_norm": 10.5, "learning_rate": 2.583016725468226e-06, "loss": 1.4311, "step": 30920 }, { "epoch": 77.3025, "grad_norm": 7.375, "learning_rate": 2.5824736642614613e-06, "loss": 1.2704, "step": 30921 }, { "epoch": 77.305, "grad_norm": 7.6875, "learning_rate": 2.581930651683702e-06, "loss": 1.4515, "step": 30922 }, { "epoch": 77.3075, "grad_norm": 7.3125, "learning_rate": 2.581387687738506e-06, "loss": 1.2283, "step": 30923 }, { "epoch": 77.31, "grad_norm": 5.65625, "learning_rate": 2.580844772429438e-06, "loss": 1.3983, "step": 30924 }, { "epoch": 77.3125, "grad_norm": 10.125, "learning_rate": 2.580301905760052e-06, "loss": 1.2071, "step": 30925 }, { "epoch": 77.315, "grad_norm": 5.84375, "learning_rate": 2.5797590877339127e-06, "loss": 1.1713, "step": 30926 }, { "epoch": 77.3175, "grad_norm": 7.59375, "learning_rate": 2.5792163183545726e-06, "loss": 1.2029, "step": 30927 }, { "epoch": 77.32, "grad_norm": 9.1875, "learning_rate": 2.578673597625597e-06, "loss": 1.356, "step": 30928 }, { "epoch": 77.3225, "grad_norm": 7.03125, "learning_rate": 2.5781309255505403e-06, "loss": 1.2291, "step": 30929 }, { "epoch": 77.325, "grad_norm": 7.375, "learning_rate": 2.577588302132956e-06, "loss": 1.2737, "step": 30930 }, { "epoch": 77.3275, "grad_norm": 10.0625, "learning_rate": 2.577045727376407e-06, "loss": 1.4942, "step": 30931 }, { "epoch": 77.33, "grad_norm": 7.71875, "learning_rate": 2.5765032012844526e-06, "loss": 1.3265, "step": 30932 }, { "epoch": 77.3325, "grad_norm": 7.28125, "learning_rate": 2.5759607238606443e-06, "loss": 1.3024, "step": 30933 }, { "epoch": 77.335, "grad_norm": 9.5, "learning_rate": 2.5754182951085396e-06, "loss": 1.3403, "step": 30934 }, { "epoch": 77.3375, "grad_norm": 9.0625, "learning_rate": 2.574875915031694e-06, "loss": 1.2532, "step": 30935 }, { "epoch": 77.34, "grad_norm": 11.625, "learning_rate": 2.574333583633668e-06, "loss": 1.318, "step": 30936 }, { "epoch": 77.3425, "grad_norm": 6.53125, "learning_rate": 2.573791300918014e-06, "loss": 1.335, "step": 30937 }, { "epoch": 77.345, "grad_norm": 6.71875, "learning_rate": 2.5732490668882835e-06, "loss": 1.2708, "step": 30938 }, { "epoch": 77.3475, "grad_norm": 10.5625, "learning_rate": 2.572706881548038e-06, "loss": 1.2967, "step": 30939 }, { "epoch": 77.35, "grad_norm": 9.5, "learning_rate": 2.572164744900827e-06, "loss": 1.4314, "step": 30940 }, { "epoch": 77.3525, "grad_norm": 10.0, "learning_rate": 2.571622656950209e-06, "loss": 1.4042, "step": 30941 }, { "epoch": 77.355, "grad_norm": 5.90625, "learning_rate": 2.5710806176997328e-06, "loss": 1.3423, "step": 30942 }, { "epoch": 77.3575, "grad_norm": 14.3125, "learning_rate": 2.5705386271529575e-06, "loss": 1.4326, "step": 30943 }, { "epoch": 77.36, "grad_norm": 8.0, "learning_rate": 2.569996685313434e-06, "loss": 1.3037, "step": 30944 }, { "epoch": 77.3625, "grad_norm": 6.0, "learning_rate": 2.5694547921847112e-06, "loss": 1.2981, "step": 30945 }, { "epoch": 77.365, "grad_norm": 12.0, "learning_rate": 2.5689129477703465e-06, "loss": 1.3578, "step": 30946 }, { "epoch": 77.3675, "grad_norm": 7.71875, "learning_rate": 2.568371152073893e-06, "loss": 1.362, "step": 30947 }, { "epoch": 77.37, "grad_norm": 6.1875, "learning_rate": 2.5678294050989017e-06, "loss": 1.1933, "step": 30948 }, { "epoch": 77.3725, "grad_norm": 6.21875, "learning_rate": 2.567287706848923e-06, "loss": 1.4772, "step": 30949 }, { "epoch": 77.375, "grad_norm": 8.5, "learning_rate": 2.5667460573275028e-06, "loss": 1.3575, "step": 30950 }, { "epoch": 77.3775, "grad_norm": 7.4375, "learning_rate": 2.5662044565382036e-06, "loss": 1.2511, "step": 30951 }, { "epoch": 77.38, "grad_norm": 7.84375, "learning_rate": 2.5656629044845714e-06, "loss": 1.3912, "step": 30952 }, { "epoch": 77.3825, "grad_norm": 7.53125, "learning_rate": 2.565121401170152e-06, "loss": 1.3544, "step": 30953 }, { "epoch": 77.385, "grad_norm": 6.21875, "learning_rate": 2.5645799465985023e-06, "loss": 1.1959, "step": 30954 }, { "epoch": 77.3875, "grad_norm": 8.375, "learning_rate": 2.564038540773166e-06, "loss": 1.5431, "step": 30955 }, { "epoch": 77.39, "grad_norm": 7.0625, "learning_rate": 2.5634971836976987e-06, "loss": 1.3753, "step": 30956 }, { "epoch": 77.3925, "grad_norm": 9.0, "learning_rate": 2.562955875375642e-06, "loss": 1.3683, "step": 30957 }, { "epoch": 77.395, "grad_norm": 15.125, "learning_rate": 2.5624146158105533e-06, "loss": 1.4164, "step": 30958 }, { "epoch": 77.3975, "grad_norm": 10.5625, "learning_rate": 2.5618734050059766e-06, "loss": 1.2147, "step": 30959 }, { "epoch": 77.4, "grad_norm": 5.875, "learning_rate": 2.5613322429654573e-06, "loss": 1.3167, "step": 30960 }, { "epoch": 77.4025, "grad_norm": 7.65625, "learning_rate": 2.560791129692546e-06, "loss": 1.3743, "step": 30961 }, { "epoch": 77.405, "grad_norm": 5.96875, "learning_rate": 2.5602500651907924e-06, "loss": 1.1955, "step": 30962 }, { "epoch": 77.4075, "grad_norm": 8.125, "learning_rate": 2.5597090494637424e-06, "loss": 1.4445, "step": 30963 }, { "epoch": 77.41, "grad_norm": 6.5625, "learning_rate": 2.5591680825149423e-06, "loss": 1.5216, "step": 30964 }, { "epoch": 77.4125, "grad_norm": 9.9375, "learning_rate": 2.558627164347932e-06, "loss": 1.3518, "step": 30965 }, { "epoch": 77.415, "grad_norm": 5.8125, "learning_rate": 2.5580862949662712e-06, "loss": 1.3255, "step": 30966 }, { "epoch": 77.4175, "grad_norm": 11.875, "learning_rate": 2.557545474373497e-06, "loss": 1.5222, "step": 30967 }, { "epoch": 77.42, "grad_norm": 6.0, "learning_rate": 2.5570047025731594e-06, "loss": 1.2355, "step": 30968 }, { "epoch": 77.4225, "grad_norm": 6.4375, "learning_rate": 2.556463979568796e-06, "loss": 1.2838, "step": 30969 }, { "epoch": 77.425, "grad_norm": 8.4375, "learning_rate": 2.5559233053639575e-06, "loss": 1.3335, "step": 30970 }, { "epoch": 77.4275, "grad_norm": 7.8125, "learning_rate": 2.555382679962192e-06, "loss": 1.3371, "step": 30971 }, { "epoch": 77.43, "grad_norm": 6.84375, "learning_rate": 2.554842103367038e-06, "loss": 1.2382, "step": 30972 }, { "epoch": 77.4325, "grad_norm": 8.0625, "learning_rate": 2.554301575582039e-06, "loss": 1.4625, "step": 30973 }, { "epoch": 77.435, "grad_norm": 7.6875, "learning_rate": 2.553761096610744e-06, "loss": 1.4396, "step": 30974 }, { "epoch": 77.4375, "grad_norm": 7.78125, "learning_rate": 2.5532206664566907e-06, "loss": 1.2993, "step": 30975 }, { "epoch": 77.44, "grad_norm": 6.875, "learning_rate": 2.5526802851234268e-06, "loss": 1.3024, "step": 30976 }, { "epoch": 77.4425, "grad_norm": 7.4375, "learning_rate": 2.5521399526144895e-06, "loss": 1.4119, "step": 30977 }, { "epoch": 77.445, "grad_norm": 8.1875, "learning_rate": 2.551599668933429e-06, "loss": 1.1678, "step": 30978 }, { "epoch": 77.4475, "grad_norm": 11.5, "learning_rate": 2.5510594340837826e-06, "loss": 1.2118, "step": 30979 }, { "epoch": 77.45, "grad_norm": 9.1875, "learning_rate": 2.5505192480690865e-06, "loss": 1.3266, "step": 30980 }, { "epoch": 77.4525, "grad_norm": 6.0625, "learning_rate": 2.549979110892894e-06, "loss": 1.2839, "step": 30981 }, { "epoch": 77.455, "grad_norm": 4.5625, "learning_rate": 2.5494390225587393e-06, "loss": 1.2481, "step": 30982 }, { "epoch": 77.4575, "grad_norm": 10.0, "learning_rate": 2.548898983070166e-06, "loss": 1.3372, "step": 30983 }, { "epoch": 77.46, "grad_norm": 10.3125, "learning_rate": 2.5483589924307086e-06, "loss": 1.4754, "step": 30984 }, { "epoch": 77.4625, "grad_norm": 5.90625, "learning_rate": 2.5478190506439117e-06, "loss": 1.1278, "step": 30985 }, { "epoch": 77.465, "grad_norm": 5.3125, "learning_rate": 2.547279157713318e-06, "loss": 1.2666, "step": 30986 }, { "epoch": 77.4675, "grad_norm": 7.625, "learning_rate": 2.5467393136424634e-06, "loss": 1.4166, "step": 30987 }, { "epoch": 77.47, "grad_norm": 6.1875, "learning_rate": 2.5461995184348854e-06, "loss": 1.3235, "step": 30988 }, { "epoch": 77.4725, "grad_norm": 6.40625, "learning_rate": 2.545659772094128e-06, "loss": 1.0017, "step": 30989 }, { "epoch": 77.475, "grad_norm": 7.84375, "learning_rate": 2.5451200746237235e-06, "loss": 1.4491, "step": 30990 }, { "epoch": 77.4775, "grad_norm": 5.8125, "learning_rate": 2.5445804260272165e-06, "loss": 1.2598, "step": 30991 }, { "epoch": 77.48, "grad_norm": 7.375, "learning_rate": 2.5440408263081385e-06, "loss": 1.2329, "step": 30992 }, { "epoch": 77.4825, "grad_norm": 8.5, "learning_rate": 2.5435012754700338e-06, "loss": 1.4599, "step": 30993 }, { "epoch": 77.485, "grad_norm": 8.125, "learning_rate": 2.542961773516437e-06, "loss": 1.2921, "step": 30994 }, { "epoch": 77.4875, "grad_norm": 5.84375, "learning_rate": 2.5424223204508804e-06, "loss": 1.3054, "step": 30995 }, { "epoch": 77.49, "grad_norm": 6.6875, "learning_rate": 2.5418829162769053e-06, "loss": 1.3885, "step": 30996 }, { "epoch": 77.4925, "grad_norm": 6.65625, "learning_rate": 2.5413435609980498e-06, "loss": 1.3045, "step": 30997 }, { "epoch": 77.495, "grad_norm": 8.25, "learning_rate": 2.540804254617848e-06, "loss": 1.4113, "step": 30998 }, { "epoch": 77.4975, "grad_norm": 10.125, "learning_rate": 2.5402649971398308e-06, "loss": 1.3338, "step": 30999 }, { "epoch": 77.5, "grad_norm": 6.75, "learning_rate": 2.5397257885675396e-06, "loss": 1.3899, "step": 31000 }, { "epoch": 77.5025, "grad_norm": 6.46875, "learning_rate": 2.5391866289045096e-06, "loss": 1.3021, "step": 31001 }, { "epoch": 77.505, "grad_norm": 7.65625, "learning_rate": 2.5386475181542734e-06, "loss": 1.2035, "step": 31002 }, { "epoch": 77.5075, "grad_norm": 8.375, "learning_rate": 2.5381084563203628e-06, "loss": 1.2658, "step": 31003 }, { "epoch": 77.51, "grad_norm": 9.9375, "learning_rate": 2.537569443406317e-06, "loss": 1.351, "step": 31004 }, { "epoch": 77.5125, "grad_norm": 9.6875, "learning_rate": 2.537030479415664e-06, "loss": 1.3862, "step": 31005 }, { "epoch": 77.515, "grad_norm": 7.78125, "learning_rate": 2.536491564351945e-06, "loss": 1.1597, "step": 31006 }, { "epoch": 77.5175, "grad_norm": 4.8125, "learning_rate": 2.5359526982186845e-06, "loss": 1.211, "step": 31007 }, { "epoch": 77.52, "grad_norm": 5.46875, "learning_rate": 2.535413881019423e-06, "loss": 1.2602, "step": 31008 }, { "epoch": 77.5225, "grad_norm": 8.625, "learning_rate": 2.5348751127576888e-06, "loss": 1.3776, "step": 31009 }, { "epoch": 77.525, "grad_norm": 9.4375, "learning_rate": 2.5343363934370125e-06, "loss": 1.4562, "step": 31010 }, { "epoch": 77.5275, "grad_norm": 6.90625, "learning_rate": 2.533797723060928e-06, "loss": 1.3622, "step": 31011 }, { "epoch": 77.53, "grad_norm": 5.46875, "learning_rate": 2.533259101632969e-06, "loss": 1.2673, "step": 31012 }, { "epoch": 77.5325, "grad_norm": 7.375, "learning_rate": 2.532720529156665e-06, "loss": 1.2647, "step": 31013 }, { "epoch": 77.535, "grad_norm": 5.53125, "learning_rate": 2.5321820056355427e-06, "loss": 1.2703, "step": 31014 }, { "epoch": 77.5375, "grad_norm": 6.6875, "learning_rate": 2.5316435310731367e-06, "loss": 1.4563, "step": 31015 }, { "epoch": 77.54, "grad_norm": 7.4375, "learning_rate": 2.5311051054729806e-06, "loss": 1.2206, "step": 31016 }, { "epoch": 77.5425, "grad_norm": 6.40625, "learning_rate": 2.5305667288385994e-06, "loss": 1.2157, "step": 31017 }, { "epoch": 77.545, "grad_norm": 8.75, "learning_rate": 2.5300284011735222e-06, "loss": 1.2805, "step": 31018 }, { "epoch": 77.5475, "grad_norm": 7.40625, "learning_rate": 2.5294901224812816e-06, "loss": 1.2353, "step": 31019 }, { "epoch": 77.55, "grad_norm": 6.96875, "learning_rate": 2.528951892765402e-06, "loss": 1.3708, "step": 31020 }, { "epoch": 77.5525, "grad_norm": 7.0, "learning_rate": 2.5284137120294183e-06, "loss": 1.2369, "step": 31021 }, { "epoch": 77.555, "grad_norm": 7.84375, "learning_rate": 2.5278755802768518e-06, "loss": 1.402, "step": 31022 }, { "epoch": 77.5575, "grad_norm": 10.1875, "learning_rate": 2.5273374975112376e-06, "loss": 1.145, "step": 31023 }, { "epoch": 77.56, "grad_norm": 6.8125, "learning_rate": 2.526799463736099e-06, "loss": 1.1974, "step": 31024 }, { "epoch": 77.5625, "grad_norm": 6.6875, "learning_rate": 2.5262614789549624e-06, "loss": 1.2106, "step": 31025 }, { "epoch": 77.565, "grad_norm": 8.875, "learning_rate": 2.525723543171356e-06, "loss": 1.3036, "step": 31026 }, { "epoch": 77.5675, "grad_norm": 6.6875, "learning_rate": 2.525185656388809e-06, "loss": 1.3993, "step": 31027 }, { "epoch": 77.57, "grad_norm": 5.5625, "learning_rate": 2.524647818610846e-06, "loss": 1.3034, "step": 31028 }, { "epoch": 77.5725, "grad_norm": 7.6875, "learning_rate": 2.5241100298409893e-06, "loss": 1.2921, "step": 31029 }, { "epoch": 77.575, "grad_norm": 7.875, "learning_rate": 2.5235722900827686e-06, "loss": 1.2994, "step": 31030 }, { "epoch": 77.5775, "grad_norm": 6.1875, "learning_rate": 2.523034599339712e-06, "loss": 1.328, "step": 31031 }, { "epoch": 77.58, "grad_norm": 11.0, "learning_rate": 2.5224969576153413e-06, "loss": 1.37, "step": 31032 }, { "epoch": 77.5825, "grad_norm": 7.3125, "learning_rate": 2.521959364913179e-06, "loss": 1.2278, "step": 31033 }, { "epoch": 77.585, "grad_norm": 6.6875, "learning_rate": 2.5214218212367537e-06, "loss": 1.3185, "step": 31034 }, { "epoch": 77.5875, "grad_norm": 9.625, "learning_rate": 2.5208843265895853e-06, "loss": 1.3056, "step": 31035 }, { "epoch": 77.59, "grad_norm": 7.875, "learning_rate": 2.520346880975203e-06, "loss": 1.225, "step": 31036 }, { "epoch": 77.5925, "grad_norm": 9.0, "learning_rate": 2.5198094843971234e-06, "loss": 1.2868, "step": 31037 }, { "epoch": 77.595, "grad_norm": 10.125, "learning_rate": 2.519272136858877e-06, "loss": 1.326, "step": 31038 }, { "epoch": 77.5975, "grad_norm": 7.53125, "learning_rate": 2.5187348383639833e-06, "loss": 1.3431, "step": 31039 }, { "epoch": 77.6, "grad_norm": 9.375, "learning_rate": 2.5181975889159615e-06, "loss": 1.3377, "step": 31040 }, { "epoch": 77.6025, "grad_norm": 7.1875, "learning_rate": 2.517660388518337e-06, "loss": 1.3112, "step": 31041 }, { "epoch": 77.605, "grad_norm": 7.5, "learning_rate": 2.5171232371746344e-06, "loss": 1.4807, "step": 31042 }, { "epoch": 77.6075, "grad_norm": 6.65625, "learning_rate": 2.5165861348883725e-06, "loss": 1.5846, "step": 31043 }, { "epoch": 77.61, "grad_norm": 5.6875, "learning_rate": 2.516049081663069e-06, "loss": 1.3774, "step": 31044 }, { "epoch": 77.6125, "grad_norm": 8.875, "learning_rate": 2.515512077502249e-06, "loss": 1.4042, "step": 31045 }, { "epoch": 77.615, "grad_norm": 6.59375, "learning_rate": 2.5149751224094344e-06, "loss": 1.4574, "step": 31046 }, { "epoch": 77.6175, "grad_norm": 7.03125, "learning_rate": 2.514438216388144e-06, "loss": 1.2716, "step": 31047 }, { "epoch": 77.62, "grad_norm": 8.9375, "learning_rate": 2.513901359441894e-06, "loss": 1.6553, "step": 31048 }, { "epoch": 77.6225, "grad_norm": 9.5625, "learning_rate": 2.51336455157421e-06, "loss": 1.3323, "step": 31049 }, { "epoch": 77.625, "grad_norm": 9.625, "learning_rate": 2.512827792788606e-06, "loss": 1.3445, "step": 31050 }, { "epoch": 77.6275, "grad_norm": 9.4375, "learning_rate": 2.5122910830886047e-06, "loss": 1.4044, "step": 31051 }, { "epoch": 77.63, "grad_norm": 6.25, "learning_rate": 2.5117544224777222e-06, "loss": 1.2402, "step": 31052 }, { "epoch": 77.6325, "grad_norm": 7.0625, "learning_rate": 2.511217810959481e-06, "loss": 1.3471, "step": 31053 }, { "epoch": 77.635, "grad_norm": 7.875, "learning_rate": 2.5106812485373955e-06, "loss": 1.2239, "step": 31054 }, { "epoch": 77.6375, "grad_norm": 7.0625, "learning_rate": 2.510144735214981e-06, "loss": 1.3804, "step": 31055 }, { "epoch": 77.64, "grad_norm": 11.8125, "learning_rate": 2.509608270995758e-06, "loss": 1.4328, "step": 31056 }, { "epoch": 77.6425, "grad_norm": 6.28125, "learning_rate": 2.509071855883246e-06, "loss": 1.4357, "step": 31057 }, { "epoch": 77.645, "grad_norm": 14.1875, "learning_rate": 2.50853548988096e-06, "loss": 1.2391, "step": 31058 }, { "epoch": 77.6475, "grad_norm": 15.4375, "learning_rate": 2.5079991729924112e-06, "loss": 1.3783, "step": 31059 }, { "epoch": 77.65, "grad_norm": 8.0, "learning_rate": 2.507462905221122e-06, "loss": 1.3496, "step": 31060 }, { "epoch": 77.6525, "grad_norm": 7.40625, "learning_rate": 2.5069266865706077e-06, "loss": 1.3348, "step": 31061 }, { "epoch": 77.655, "grad_norm": 8.375, "learning_rate": 2.5063905170443825e-06, "loss": 1.2146, "step": 31062 }, { "epoch": 77.6575, "grad_norm": 8.5, "learning_rate": 2.505854396645958e-06, "loss": 1.3608, "step": 31063 }, { "epoch": 77.66, "grad_norm": 9.4375, "learning_rate": 2.505318325378856e-06, "loss": 1.2817, "step": 31064 }, { "epoch": 77.6625, "grad_norm": 7.625, "learning_rate": 2.5047823032465836e-06, "loss": 1.4744, "step": 31065 }, { "epoch": 77.665, "grad_norm": 5.75, "learning_rate": 2.504246330252661e-06, "loss": 1.1772, "step": 31066 }, { "epoch": 77.6675, "grad_norm": 5.9375, "learning_rate": 2.5037104064005967e-06, "loss": 1.3472, "step": 31067 }, { "epoch": 77.67, "grad_norm": 8.1875, "learning_rate": 2.5031745316939106e-06, "loss": 1.2419, "step": 31068 }, { "epoch": 77.6725, "grad_norm": 7.65625, "learning_rate": 2.5026387061361114e-06, "loss": 1.2835, "step": 31069 }, { "epoch": 77.675, "grad_norm": 7.9375, "learning_rate": 2.502102929730711e-06, "loss": 1.3286, "step": 31070 }, { "epoch": 77.6775, "grad_norm": 8.1875, "learning_rate": 2.501567202481222e-06, "loss": 1.1511, "step": 31071 }, { "epoch": 77.68, "grad_norm": 6.75, "learning_rate": 2.501031524391163e-06, "loss": 1.3882, "step": 31072 }, { "epoch": 77.6825, "grad_norm": 9.1875, "learning_rate": 2.5004958954640403e-06, "loss": 1.259, "step": 31073 }, { "epoch": 77.685, "grad_norm": 6.25, "learning_rate": 2.499960315703364e-06, "loss": 1.2799, "step": 31074 }, { "epoch": 77.6875, "grad_norm": 7.46875, "learning_rate": 2.4994247851126475e-06, "loss": 1.3686, "step": 31075 }, { "epoch": 77.69, "grad_norm": 10.6875, "learning_rate": 2.4988893036954045e-06, "loss": 1.401, "step": 31076 }, { "epoch": 77.6925, "grad_norm": 5.375, "learning_rate": 2.4983538714551436e-06, "loss": 1.3085, "step": 31077 }, { "epoch": 77.695, "grad_norm": 10.9375, "learning_rate": 2.4978184883953715e-06, "loss": 1.1825, "step": 31078 }, { "epoch": 77.6975, "grad_norm": 7.3125, "learning_rate": 2.497283154519604e-06, "loss": 1.2869, "step": 31079 }, { "epoch": 77.7, "grad_norm": 15.1875, "learning_rate": 2.496747869831345e-06, "loss": 1.3446, "step": 31080 }, { "epoch": 77.7025, "grad_norm": 6.875, "learning_rate": 2.49621263433411e-06, "loss": 1.3438, "step": 31081 }, { "epoch": 77.705, "grad_norm": 9.625, "learning_rate": 2.4956774480314004e-06, "loss": 1.3836, "step": 31082 }, { "epoch": 77.7075, "grad_norm": 9.5625, "learning_rate": 2.4951423109267337e-06, "loss": 1.3082, "step": 31083 }, { "epoch": 77.71, "grad_norm": 7.53125, "learning_rate": 2.494607223023613e-06, "loss": 1.3373, "step": 31084 }, { "epoch": 77.7125, "grad_norm": 10.375, "learning_rate": 2.4940721843255443e-06, "loss": 1.4474, "step": 31085 }, { "epoch": 77.715, "grad_norm": 5.96875, "learning_rate": 2.4935371948360376e-06, "loss": 1.3458, "step": 31086 }, { "epoch": 77.7175, "grad_norm": 8.6875, "learning_rate": 2.4930022545586043e-06, "loss": 1.342, "step": 31087 }, { "epoch": 77.72, "grad_norm": 7.4375, "learning_rate": 2.492467363496747e-06, "loss": 1.2134, "step": 31088 }, { "epoch": 77.7225, "grad_norm": 5.9375, "learning_rate": 2.4919325216539737e-06, "loss": 1.3584, "step": 31089 }, { "epoch": 77.725, "grad_norm": 8.6875, "learning_rate": 2.4913977290337844e-06, "loss": 1.2438, "step": 31090 }, { "epoch": 77.7275, "grad_norm": 6.6875, "learning_rate": 2.490862985639696e-06, "loss": 1.3332, "step": 31091 }, { "epoch": 77.73, "grad_norm": 7.21875, "learning_rate": 2.4903282914752104e-06, "loss": 1.4271, "step": 31092 }, { "epoch": 77.7325, "grad_norm": 6.40625, "learning_rate": 2.489793646543829e-06, "loss": 1.2529, "step": 31093 }, { "epoch": 77.735, "grad_norm": 7.21875, "learning_rate": 2.4892590508490624e-06, "loss": 1.4114, "step": 31094 }, { "epoch": 77.7375, "grad_norm": 10.1875, "learning_rate": 2.4887245043944107e-06, "loss": 1.3548, "step": 31095 }, { "epoch": 77.74, "grad_norm": 8.25, "learning_rate": 2.488190007183383e-06, "loss": 1.2673, "step": 31096 }, { "epoch": 77.7425, "grad_norm": 10.5, "learning_rate": 2.4876555592194783e-06, "loss": 1.3348, "step": 31097 }, { "epoch": 77.745, "grad_norm": 7.46875, "learning_rate": 2.4871211605062063e-06, "loss": 1.3076, "step": 31098 }, { "epoch": 77.7475, "grad_norm": 7.3125, "learning_rate": 2.486586811047066e-06, "loss": 1.4343, "step": 31099 }, { "epoch": 77.75, "grad_norm": 5.59375, "learning_rate": 2.48605251084556e-06, "loss": 1.3245, "step": 31100 }, { "epoch": 77.7525, "grad_norm": 6.8125, "learning_rate": 2.4855182599051932e-06, "loss": 1.2934, "step": 31101 }, { "epoch": 77.755, "grad_norm": 7.84375, "learning_rate": 2.4849840582294704e-06, "loss": 1.2675, "step": 31102 }, { "epoch": 77.7575, "grad_norm": 5.4375, "learning_rate": 2.4844499058218917e-06, "loss": 1.2507, "step": 31103 }, { "epoch": 77.76, "grad_norm": 5.53125, "learning_rate": 2.483915802685959e-06, "loss": 1.269, "step": 31104 }, { "epoch": 77.7625, "grad_norm": 13.0, "learning_rate": 2.4833817488251677e-06, "loss": 1.4611, "step": 31105 }, { "epoch": 77.765, "grad_norm": 6.25, "learning_rate": 2.48284774424303e-06, "loss": 1.3501, "step": 31106 }, { "epoch": 77.7675, "grad_norm": 7.71875, "learning_rate": 2.4823137889430416e-06, "loss": 1.4945, "step": 31107 }, { "epoch": 77.77, "grad_norm": 7.75, "learning_rate": 2.481779882928703e-06, "loss": 1.384, "step": 31108 }, { "epoch": 77.7725, "grad_norm": 7.875, "learning_rate": 2.481246026203513e-06, "loss": 1.4181, "step": 31109 }, { "epoch": 77.775, "grad_norm": 5.6875, "learning_rate": 2.480712218770972e-06, "loss": 1.1524, "step": 31110 }, { "epoch": 77.7775, "grad_norm": 7.0, "learning_rate": 2.4801784606345846e-06, "loss": 1.1635, "step": 31111 }, { "epoch": 77.78, "grad_norm": 7.375, "learning_rate": 2.479644751797845e-06, "loss": 1.1928, "step": 31112 }, { "epoch": 77.7825, "grad_norm": 6.3125, "learning_rate": 2.4791110922642515e-06, "loss": 1.3383, "step": 31113 }, { "epoch": 77.785, "grad_norm": 7.90625, "learning_rate": 2.4785774820373077e-06, "loss": 1.2919, "step": 31114 }, { "epoch": 77.7875, "grad_norm": 7.375, "learning_rate": 2.4780439211205055e-06, "loss": 1.3258, "step": 31115 }, { "epoch": 77.79, "grad_norm": 7.40625, "learning_rate": 2.4775104095173495e-06, "loss": 1.2266, "step": 31116 }, { "epoch": 77.7925, "grad_norm": 11.25, "learning_rate": 2.476976947231332e-06, "loss": 1.2265, "step": 31117 }, { "epoch": 77.795, "grad_norm": 7.84375, "learning_rate": 2.4764435342659542e-06, "loss": 1.2728, "step": 31118 }, { "epoch": 77.7975, "grad_norm": 8.6875, "learning_rate": 2.4759101706247123e-06, "loss": 1.2474, "step": 31119 }, { "epoch": 77.8, "grad_norm": 8.4375, "learning_rate": 2.475376856311097e-06, "loss": 1.335, "step": 31120 }, { "epoch": 77.8025, "grad_norm": 5.3125, "learning_rate": 2.474843591328614e-06, "loss": 1.332, "step": 31121 }, { "epoch": 77.805, "grad_norm": 8.9375, "learning_rate": 2.4743103756807564e-06, "loss": 1.2219, "step": 31122 }, { "epoch": 77.8075, "grad_norm": 9.25, "learning_rate": 2.473777209371019e-06, "loss": 1.3638, "step": 31123 }, { "epoch": 77.81, "grad_norm": 6.375, "learning_rate": 2.4732440924028935e-06, "loss": 1.4505, "step": 31124 }, { "epoch": 77.8125, "grad_norm": 6.53125, "learning_rate": 2.472711024779879e-06, "loss": 1.3498, "step": 31125 }, { "epoch": 77.815, "grad_norm": 6.78125, "learning_rate": 2.4721780065054723e-06, "loss": 1.3547, "step": 31126 }, { "epoch": 77.8175, "grad_norm": 8.0625, "learning_rate": 2.4716450375831656e-06, "loss": 1.2704, "step": 31127 }, { "epoch": 77.82, "grad_norm": 8.4375, "learning_rate": 2.47111211801645e-06, "loss": 1.2903, "step": 31128 }, { "epoch": 77.8225, "grad_norm": 10.8125, "learning_rate": 2.4705792478088253e-06, "loss": 1.1832, "step": 31129 }, { "epoch": 77.825, "grad_norm": 9.375, "learning_rate": 2.470046426963778e-06, "loss": 1.4268, "step": 31130 }, { "epoch": 77.8275, "grad_norm": 5.875, "learning_rate": 2.4695136554848085e-06, "loss": 1.3203, "step": 31131 }, { "epoch": 77.83, "grad_norm": 11.5625, "learning_rate": 2.4689809333754022e-06, "loss": 1.3445, "step": 31132 }, { "epoch": 77.8325, "grad_norm": 8.125, "learning_rate": 2.4684482606390594e-06, "loss": 1.3266, "step": 31133 }, { "epoch": 77.835, "grad_norm": 11.5625, "learning_rate": 2.467915637279269e-06, "loss": 1.2254, "step": 31134 }, { "epoch": 77.8375, "grad_norm": 6.90625, "learning_rate": 2.4673830632995176e-06, "loss": 1.1463, "step": 31135 }, { "epoch": 77.84, "grad_norm": 9.125, "learning_rate": 2.4668505387033025e-06, "loss": 1.4178, "step": 31136 }, { "epoch": 77.8425, "grad_norm": 10.375, "learning_rate": 2.4663180634941162e-06, "loss": 1.3474, "step": 31137 }, { "epoch": 77.845, "grad_norm": 9.875, "learning_rate": 2.4657856376754487e-06, "loss": 1.412, "step": 31138 }, { "epoch": 77.8475, "grad_norm": 10.3125, "learning_rate": 2.465253261250784e-06, "loss": 1.4995, "step": 31139 }, { "epoch": 77.85, "grad_norm": 8.3125, "learning_rate": 2.464720934223619e-06, "loss": 1.2599, "step": 31140 }, { "epoch": 77.8525, "grad_norm": 6.46875, "learning_rate": 2.4641886565974437e-06, "loss": 1.1939, "step": 31141 }, { "epoch": 77.855, "grad_norm": 9.4375, "learning_rate": 2.463656428375747e-06, "loss": 1.323, "step": 31142 }, { "epoch": 77.8575, "grad_norm": 10.25, "learning_rate": 2.4631242495620144e-06, "loss": 1.3621, "step": 31143 }, { "epoch": 77.86, "grad_norm": 5.3125, "learning_rate": 2.462592120159739e-06, "loss": 1.2739, "step": 31144 }, { "epoch": 77.8625, "grad_norm": 10.9375, "learning_rate": 2.462060040172407e-06, "loss": 1.2306, "step": 31145 }, { "epoch": 77.865, "grad_norm": 6.09375, "learning_rate": 2.46152800960351e-06, "loss": 1.2925, "step": 31146 }, { "epoch": 77.8675, "grad_norm": 8.1875, "learning_rate": 2.4609960284565302e-06, "loss": 1.2902, "step": 31147 }, { "epoch": 77.87, "grad_norm": 10.0, "learning_rate": 2.4604640967349614e-06, "loss": 1.3781, "step": 31148 }, { "epoch": 77.8725, "grad_norm": 7.125, "learning_rate": 2.4599322144422864e-06, "loss": 1.3529, "step": 31149 }, { "epoch": 77.875, "grad_norm": 5.21875, "learning_rate": 2.459400381581997e-06, "loss": 1.3499, "step": 31150 }, { "epoch": 77.8775, "grad_norm": 7.6875, "learning_rate": 2.4588685981575732e-06, "loss": 1.4533, "step": 31151 }, { "epoch": 77.88, "grad_norm": 9.75, "learning_rate": 2.458336864172508e-06, "loss": 1.489, "step": 31152 }, { "epoch": 77.8825, "grad_norm": 11.5625, "learning_rate": 2.4578051796302847e-06, "loss": 1.3896, "step": 31153 }, { "epoch": 77.885, "grad_norm": 9.5625, "learning_rate": 2.4572735445343867e-06, "loss": 1.3825, "step": 31154 }, { "epoch": 77.8875, "grad_norm": 9.0625, "learning_rate": 2.4567419588883e-06, "loss": 1.3292, "step": 31155 }, { "epoch": 77.89, "grad_norm": 7.0625, "learning_rate": 2.456210422695515e-06, "loss": 1.3617, "step": 31156 }, { "epoch": 77.8925, "grad_norm": 8.8125, "learning_rate": 2.4556789359595134e-06, "loss": 1.3016, "step": 31157 }, { "epoch": 77.895, "grad_norm": 8.1875, "learning_rate": 2.4551474986837743e-06, "loss": 1.3423, "step": 31158 }, { "epoch": 77.8975, "grad_norm": 8.1875, "learning_rate": 2.454616110871787e-06, "loss": 1.4724, "step": 31159 }, { "epoch": 77.9, "grad_norm": 7.0625, "learning_rate": 2.4540847725270376e-06, "loss": 1.2955, "step": 31160 }, { "epoch": 77.9025, "grad_norm": 8.8125, "learning_rate": 2.453553483653006e-06, "loss": 1.3291, "step": 31161 }, { "epoch": 77.905, "grad_norm": 7.9375, "learning_rate": 2.4530222442531738e-06, "loss": 1.3514, "step": 31162 }, { "epoch": 77.9075, "grad_norm": 8.875, "learning_rate": 2.4524910543310274e-06, "loss": 1.3712, "step": 31163 }, { "epoch": 77.91, "grad_norm": 9.8125, "learning_rate": 2.4519599138900463e-06, "loss": 1.3636, "step": 31164 }, { "epoch": 77.9125, "grad_norm": 7.28125, "learning_rate": 2.451428822933716e-06, "loss": 1.265, "step": 31165 }, { "epoch": 77.915, "grad_norm": 5.9375, "learning_rate": 2.450897781465512e-06, "loss": 1.16, "step": 31166 }, { "epoch": 77.9175, "grad_norm": 6.4375, "learning_rate": 2.4503667894889248e-06, "loss": 1.3168, "step": 31167 }, { "epoch": 77.92, "grad_norm": 5.28125, "learning_rate": 2.44983584700743e-06, "loss": 1.2266, "step": 31168 }, { "epoch": 77.9225, "grad_norm": 6.59375, "learning_rate": 2.4493049540245063e-06, "loss": 1.1739, "step": 31169 }, { "epoch": 77.925, "grad_norm": 9.75, "learning_rate": 2.448774110543638e-06, "loss": 1.1894, "step": 31170 }, { "epoch": 77.9275, "grad_norm": 8.0625, "learning_rate": 2.4482433165683063e-06, "loss": 1.2436, "step": 31171 }, { "epoch": 77.93, "grad_norm": 8.4375, "learning_rate": 2.44771257210199e-06, "loss": 1.2736, "step": 31172 }, { "epoch": 77.9325, "grad_norm": 7.875, "learning_rate": 2.447181877148165e-06, "loss": 1.4929, "step": 31173 }, { "epoch": 77.935, "grad_norm": 6.875, "learning_rate": 2.446651231710312e-06, "loss": 1.2667, "step": 31174 }, { "epoch": 77.9375, "grad_norm": 6.96875, "learning_rate": 2.4461206357919154e-06, "loss": 1.3827, "step": 31175 }, { "epoch": 77.94, "grad_norm": 8.3125, "learning_rate": 2.445590089396449e-06, "loss": 1.2931, "step": 31176 }, { "epoch": 77.9425, "grad_norm": 5.90625, "learning_rate": 2.4450595925273903e-06, "loss": 1.352, "step": 31177 }, { "epoch": 77.945, "grad_norm": 6.75, "learning_rate": 2.444529145188219e-06, "loss": 1.2795, "step": 31178 }, { "epoch": 77.9475, "grad_norm": 6.75, "learning_rate": 2.4439987473824113e-06, "loss": 1.249, "step": 31179 }, { "epoch": 77.95, "grad_norm": 7.8125, "learning_rate": 2.4434683991134476e-06, "loss": 1.562, "step": 31180 }, { "epoch": 77.9525, "grad_norm": 6.6875, "learning_rate": 2.4429381003847995e-06, "loss": 1.2777, "step": 31181 }, { "epoch": 77.955, "grad_norm": 8.8125, "learning_rate": 2.4424078511999505e-06, "loss": 1.3503, "step": 31182 }, { "epoch": 77.9575, "grad_norm": 7.90625, "learning_rate": 2.4418776515623733e-06, "loss": 1.2331, "step": 31183 }, { "epoch": 77.96, "grad_norm": 5.75, "learning_rate": 2.4413475014755396e-06, "loss": 1.238, "step": 31184 }, { "epoch": 77.9625, "grad_norm": 5.25, "learning_rate": 2.44081740094293e-06, "loss": 1.3258, "step": 31185 }, { "epoch": 77.965, "grad_norm": 5.3125, "learning_rate": 2.4402873499680226e-06, "loss": 1.4072, "step": 31186 }, { "epoch": 77.9675, "grad_norm": 5.65625, "learning_rate": 2.439757348554287e-06, "loss": 1.1983, "step": 31187 }, { "epoch": 77.97, "grad_norm": 8.3125, "learning_rate": 2.4392273967051983e-06, "loss": 1.3212, "step": 31188 }, { "epoch": 77.9725, "grad_norm": 5.8125, "learning_rate": 2.4386974944242314e-06, "loss": 1.4756, "step": 31189 }, { "epoch": 77.975, "grad_norm": 8.0, "learning_rate": 2.438167641714865e-06, "loss": 1.2592, "step": 31190 }, { "epoch": 77.9775, "grad_norm": 5.84375, "learning_rate": 2.437637838580568e-06, "loss": 1.3394, "step": 31191 }, { "epoch": 77.98, "grad_norm": 9.25, "learning_rate": 2.437108085024812e-06, "loss": 1.1803, "step": 31192 }, { "epoch": 77.9825, "grad_norm": 9.4375, "learning_rate": 2.4365783810510756e-06, "loss": 1.2818, "step": 31193 }, { "epoch": 77.985, "grad_norm": 8.4375, "learning_rate": 2.436048726662826e-06, "loss": 1.3765, "step": 31194 }, { "epoch": 77.9875, "grad_norm": 7.03125, "learning_rate": 2.435519121863541e-06, "loss": 1.3166, "step": 31195 }, { "epoch": 77.99, "grad_norm": 9.0625, "learning_rate": 2.434989566656687e-06, "loss": 1.2761, "step": 31196 }, { "epoch": 77.9925, "grad_norm": 11.5625, "learning_rate": 2.4344600610457423e-06, "loss": 1.3185, "step": 31197 }, { "epoch": 77.995, "grad_norm": 6.96875, "learning_rate": 2.4339306050341736e-06, "loss": 1.2192, "step": 31198 }, { "epoch": 77.9975, "grad_norm": 6.28125, "learning_rate": 2.4334011986254514e-06, "loss": 1.346, "step": 31199 }, { "epoch": 78.0, "grad_norm": 9.4375, "learning_rate": 2.432871841823047e-06, "loss": 1.3302, "step": 31200 }, { "epoch": 78.0025, "grad_norm": 6.75, "learning_rate": 2.432342534630435e-06, "loss": 1.2403, "step": 31201 }, { "epoch": 78.005, "grad_norm": 10.6875, "learning_rate": 2.431813277051083e-06, "loss": 1.3785, "step": 31202 }, { "epoch": 78.0075, "grad_norm": 10.625, "learning_rate": 2.431284069088458e-06, "loss": 1.3489, "step": 31203 }, { "epoch": 78.01, "grad_norm": 6.84375, "learning_rate": 2.4307549107460304e-06, "loss": 1.3307, "step": 31204 }, { "epoch": 78.0125, "grad_norm": 8.125, "learning_rate": 2.430225802027274e-06, "loss": 1.3251, "step": 31205 }, { "epoch": 78.015, "grad_norm": 6.78125, "learning_rate": 2.4296967429356542e-06, "loss": 1.4773, "step": 31206 }, { "epoch": 78.0175, "grad_norm": 6.875, "learning_rate": 2.4291677334746367e-06, "loss": 1.2982, "step": 31207 }, { "epoch": 78.02, "grad_norm": 11.25, "learning_rate": 2.428638773647696e-06, "loss": 1.3031, "step": 31208 }, { "epoch": 78.0225, "grad_norm": 7.1875, "learning_rate": 2.4281098634582935e-06, "loss": 1.2033, "step": 31209 }, { "epoch": 78.025, "grad_norm": 8.1875, "learning_rate": 2.4275810029099025e-06, "loss": 1.3492, "step": 31210 }, { "epoch": 78.0275, "grad_norm": 9.25, "learning_rate": 2.427052192005984e-06, "loss": 1.3251, "step": 31211 }, { "epoch": 78.03, "grad_norm": 8.0, "learning_rate": 2.4265234307500117e-06, "loss": 1.2507, "step": 31212 }, { "epoch": 78.0325, "grad_norm": 13.375, "learning_rate": 2.4259947191454493e-06, "loss": 1.4379, "step": 31213 }, { "epoch": 78.035, "grad_norm": 9.0625, "learning_rate": 2.425466057195759e-06, "loss": 1.3296, "step": 31214 }, { "epoch": 78.0375, "grad_norm": 6.15625, "learning_rate": 2.4249374449044094e-06, "loss": 1.3231, "step": 31215 }, { "epoch": 78.04, "grad_norm": 4.3125, "learning_rate": 2.42440888227487e-06, "loss": 1.1196, "step": 31216 }, { "epoch": 78.0425, "grad_norm": 9.875, "learning_rate": 2.423880369310604e-06, "loss": 1.273, "step": 31217 }, { "epoch": 78.045, "grad_norm": 7.375, "learning_rate": 2.423351906015071e-06, "loss": 1.3601, "step": 31218 }, { "epoch": 78.0475, "grad_norm": 11.875, "learning_rate": 2.422823492391739e-06, "loss": 1.555, "step": 31219 }, { "epoch": 78.05, "grad_norm": 8.0625, "learning_rate": 2.4222951284440776e-06, "loss": 1.2716, "step": 31220 }, { "epoch": 78.0525, "grad_norm": 6.96875, "learning_rate": 2.421766814175546e-06, "loss": 1.3467, "step": 31221 }, { "epoch": 78.055, "grad_norm": 6.96875, "learning_rate": 2.421238549589604e-06, "loss": 1.4213, "step": 31222 }, { "epoch": 78.0575, "grad_norm": 8.5, "learning_rate": 2.420710334689722e-06, "loss": 1.489, "step": 31223 }, { "epoch": 78.06, "grad_norm": 5.125, "learning_rate": 2.420182169479357e-06, "loss": 1.2747, "step": 31224 }, { "epoch": 78.0625, "grad_norm": 9.0, "learning_rate": 2.4196540539619774e-06, "loss": 1.3449, "step": 31225 }, { "epoch": 78.065, "grad_norm": 8.0, "learning_rate": 2.419125988141039e-06, "loss": 1.4179, "step": 31226 }, { "epoch": 78.0675, "grad_norm": 6.125, "learning_rate": 2.41859797202001e-06, "loss": 1.408, "step": 31227 }, { "epoch": 78.07, "grad_norm": 6.53125, "learning_rate": 2.41807000560235e-06, "loss": 1.323, "step": 31228 }, { "epoch": 78.0725, "grad_norm": 6.6875, "learning_rate": 2.417542088891517e-06, "loss": 1.2194, "step": 31229 }, { "epoch": 78.075, "grad_norm": 6.84375, "learning_rate": 2.417014221890974e-06, "loss": 1.1787, "step": 31230 }, { "epoch": 78.0775, "grad_norm": 7.40625, "learning_rate": 2.416486404604186e-06, "loss": 1.2939, "step": 31231 }, { "epoch": 78.08, "grad_norm": 8.8125, "learning_rate": 2.415958637034609e-06, "loss": 1.4489, "step": 31232 }, { "epoch": 78.0825, "grad_norm": 7.1875, "learning_rate": 2.4154309191857008e-06, "loss": 1.2645, "step": 31233 }, { "epoch": 78.085, "grad_norm": 7.40625, "learning_rate": 2.4149032510609237e-06, "loss": 1.3488, "step": 31234 }, { "epoch": 78.0875, "grad_norm": 9.3125, "learning_rate": 2.4143756326637403e-06, "loss": 1.2412, "step": 31235 }, { "epoch": 78.09, "grad_norm": 7.40625, "learning_rate": 2.413848063997607e-06, "loss": 1.1018, "step": 31236 }, { "epoch": 78.0925, "grad_norm": 6.25, "learning_rate": 2.413320545065979e-06, "loss": 1.4016, "step": 31237 }, { "epoch": 78.095, "grad_norm": 6.125, "learning_rate": 2.4127930758723207e-06, "loss": 1.331, "step": 31238 }, { "epoch": 78.0975, "grad_norm": 6.34375, "learning_rate": 2.4122656564200853e-06, "loss": 1.2568, "step": 31239 }, { "epoch": 78.1, "grad_norm": 6.875, "learning_rate": 2.411738286712735e-06, "loss": 1.3164, "step": 31240 }, { "epoch": 78.1025, "grad_norm": 5.96875, "learning_rate": 2.411210966753722e-06, "loss": 1.2382, "step": 31241 }, { "epoch": 78.105, "grad_norm": 6.53125, "learning_rate": 2.4106836965465084e-06, "loss": 1.304, "step": 31242 }, { "epoch": 78.1075, "grad_norm": 6.71875, "learning_rate": 2.4101564760945495e-06, "loss": 1.3233, "step": 31243 }, { "epoch": 78.11, "grad_norm": 6.75, "learning_rate": 2.409629305401299e-06, "loss": 1.265, "step": 31244 }, { "epoch": 78.1125, "grad_norm": 7.53125, "learning_rate": 2.4091021844702132e-06, "loss": 1.4012, "step": 31245 }, { "epoch": 78.115, "grad_norm": 9.875, "learning_rate": 2.408575113304754e-06, "loss": 1.2977, "step": 31246 }, { "epoch": 78.1175, "grad_norm": 6.78125, "learning_rate": 2.4080480919083714e-06, "loss": 1.3736, "step": 31247 }, { "epoch": 78.12, "grad_norm": 7.59375, "learning_rate": 2.407521120284523e-06, "loss": 1.1883, "step": 31248 }, { "epoch": 78.1225, "grad_norm": 5.75, "learning_rate": 2.4069941984366564e-06, "loss": 1.1093, "step": 31249 }, { "epoch": 78.125, "grad_norm": 8.125, "learning_rate": 2.406467326368237e-06, "loss": 1.3479, "step": 31250 }, { "epoch": 78.1275, "grad_norm": 5.46875, "learning_rate": 2.4059405040827133e-06, "loss": 1.4351, "step": 31251 }, { "epoch": 78.13, "grad_norm": 7.90625, "learning_rate": 2.405413731583541e-06, "loss": 1.5103, "step": 31252 }, { "epoch": 78.1325, "grad_norm": 6.40625, "learning_rate": 2.4048870088741693e-06, "loss": 1.3433, "step": 31253 }, { "epoch": 78.135, "grad_norm": 9.875, "learning_rate": 2.4043603359580536e-06, "loss": 1.4185, "step": 31254 }, { "epoch": 78.1375, "grad_norm": 6.625, "learning_rate": 2.403833712838651e-06, "loss": 1.439, "step": 31255 }, { "epoch": 78.14, "grad_norm": 4.5625, "learning_rate": 2.403307139519411e-06, "loss": 1.2716, "step": 31256 }, { "epoch": 78.1425, "grad_norm": 6.5, "learning_rate": 2.4027806160037825e-06, "loss": 1.3916, "step": 31257 }, { "epoch": 78.145, "grad_norm": 9.125, "learning_rate": 2.402254142295223e-06, "loss": 1.3617, "step": 31258 }, { "epoch": 78.1475, "grad_norm": 8.25, "learning_rate": 2.4017277183971778e-06, "loss": 1.3556, "step": 31259 }, { "epoch": 78.15, "grad_norm": 7.875, "learning_rate": 2.401201344313102e-06, "loss": 1.3427, "step": 31260 }, { "epoch": 78.1525, "grad_norm": 7.25, "learning_rate": 2.400675020046449e-06, "loss": 1.1909, "step": 31261 }, { "epoch": 78.155, "grad_norm": 8.75, "learning_rate": 2.400148745600667e-06, "loss": 1.3348, "step": 31262 }, { "epoch": 78.1575, "grad_norm": 10.1875, "learning_rate": 2.3996225209792046e-06, "loss": 1.3595, "step": 31263 }, { "epoch": 78.16, "grad_norm": 6.65625, "learning_rate": 2.3990963461855075e-06, "loss": 1.2976, "step": 31264 }, { "epoch": 78.1625, "grad_norm": 7.21875, "learning_rate": 2.3985702212230366e-06, "loss": 1.2644, "step": 31265 }, { "epoch": 78.165, "grad_norm": 7.125, "learning_rate": 2.3980441460952353e-06, "loss": 1.2012, "step": 31266 }, { "epoch": 78.1675, "grad_norm": 5.8125, "learning_rate": 2.397518120805552e-06, "loss": 1.3376, "step": 31267 }, { "epoch": 78.17, "grad_norm": 6.96875, "learning_rate": 2.396992145357433e-06, "loss": 1.2966, "step": 31268 }, { "epoch": 78.1725, "grad_norm": 8.4375, "learning_rate": 2.3964662197543297e-06, "loss": 1.3196, "step": 31269 }, { "epoch": 78.175, "grad_norm": 10.0, "learning_rate": 2.395940343999691e-06, "loss": 1.3127, "step": 31270 }, { "epoch": 78.1775, "grad_norm": 4.59375, "learning_rate": 2.3954145180969644e-06, "loss": 1.2386, "step": 31271 }, { "epoch": 78.18, "grad_norm": 8.625, "learning_rate": 2.3948887420495927e-06, "loss": 1.4207, "step": 31272 }, { "epoch": 78.1825, "grad_norm": 7.25, "learning_rate": 2.3943630158610286e-06, "loss": 1.3202, "step": 31273 }, { "epoch": 78.185, "grad_norm": 6.84375, "learning_rate": 2.3938373395347136e-06, "loss": 1.391, "step": 31274 }, { "epoch": 78.1875, "grad_norm": 8.4375, "learning_rate": 2.3933117130741e-06, "loss": 1.4458, "step": 31275 }, { "epoch": 78.19, "grad_norm": 9.25, "learning_rate": 2.3927861364826265e-06, "loss": 1.2869, "step": 31276 }, { "epoch": 78.1925, "grad_norm": 7.75, "learning_rate": 2.3922606097637456e-06, "loss": 1.2912, "step": 31277 }, { "epoch": 78.195, "grad_norm": 6.375, "learning_rate": 2.3917351329209004e-06, "loss": 1.3228, "step": 31278 }, { "epoch": 78.1975, "grad_norm": 6.4375, "learning_rate": 2.3912097059575322e-06, "loss": 1.1264, "step": 31279 }, { "epoch": 78.2, "grad_norm": 6.5625, "learning_rate": 2.390684328877089e-06, "loss": 1.3847, "step": 31280 }, { "epoch": 78.2025, "grad_norm": 9.9375, "learning_rate": 2.3901590016830167e-06, "loss": 1.2237, "step": 31281 }, { "epoch": 78.205, "grad_norm": 9.6875, "learning_rate": 2.3896337243787583e-06, "loss": 1.3006, "step": 31282 }, { "epoch": 78.2075, "grad_norm": 8.9375, "learning_rate": 2.3891084969677534e-06, "loss": 1.3835, "step": 31283 }, { "epoch": 78.21, "grad_norm": 8.5625, "learning_rate": 2.388583319453449e-06, "loss": 1.5005, "step": 31284 }, { "epoch": 78.2125, "grad_norm": 7.09375, "learning_rate": 2.38805819183929e-06, "loss": 1.4267, "step": 31285 }, { "epoch": 78.215, "grad_norm": 7.0625, "learning_rate": 2.3875331141287173e-06, "loss": 1.3606, "step": 31286 }, { "epoch": 78.2175, "grad_norm": 6.84375, "learning_rate": 2.38700808632517e-06, "loss": 1.3535, "step": 31287 }, { "epoch": 78.22, "grad_norm": 5.96875, "learning_rate": 2.386483108432096e-06, "loss": 1.2847, "step": 31288 }, { "epoch": 78.2225, "grad_norm": 5.4375, "learning_rate": 2.385958180452931e-06, "loss": 1.2325, "step": 31289 }, { "epoch": 78.225, "grad_norm": 7.03125, "learning_rate": 2.3854333023911225e-06, "loss": 1.5219, "step": 31290 }, { "epoch": 78.2275, "grad_norm": 7.1875, "learning_rate": 2.384908474250105e-06, "loss": 1.3034, "step": 31291 }, { "epoch": 78.23, "grad_norm": 12.75, "learning_rate": 2.3843836960333267e-06, "loss": 1.4151, "step": 31292 }, { "epoch": 78.2325, "grad_norm": 7.71875, "learning_rate": 2.3838589677442236e-06, "loss": 1.3147, "step": 31293 }, { "epoch": 78.235, "grad_norm": 7.75, "learning_rate": 2.3833342893862334e-06, "loss": 1.3879, "step": 31294 }, { "epoch": 78.2375, "grad_norm": 5.8125, "learning_rate": 2.3828096609627992e-06, "loss": 1.4339, "step": 31295 }, { "epoch": 78.24, "grad_norm": 5.53125, "learning_rate": 2.3822850824773623e-06, "loss": 1.3796, "step": 31296 }, { "epoch": 78.2425, "grad_norm": 7.21875, "learning_rate": 2.3817605539333597e-06, "loss": 1.2773, "step": 31297 }, { "epoch": 78.245, "grad_norm": 8.5625, "learning_rate": 2.381236075334228e-06, "loss": 1.2337, "step": 31298 }, { "epoch": 78.2475, "grad_norm": 10.8125, "learning_rate": 2.380711646683407e-06, "loss": 1.3169, "step": 31299 }, { "epoch": 78.25, "grad_norm": 8.25, "learning_rate": 2.3801872679843384e-06, "loss": 1.3443, "step": 31300 }, { "epoch": 78.2525, "grad_norm": 10.625, "learning_rate": 2.3796629392404568e-06, "loss": 1.3487, "step": 31301 }, { "epoch": 78.255, "grad_norm": 6.84375, "learning_rate": 2.379138660455198e-06, "loss": 1.3221, "step": 31302 }, { "epoch": 78.2575, "grad_norm": 8.4375, "learning_rate": 2.378614431632004e-06, "loss": 1.2896, "step": 31303 }, { "epoch": 78.26, "grad_norm": 6.6875, "learning_rate": 2.3780902527743054e-06, "loss": 1.3013, "step": 31304 }, { "epoch": 78.2625, "grad_norm": 6.875, "learning_rate": 2.3775661238855454e-06, "loss": 1.2678, "step": 31305 }, { "epoch": 78.265, "grad_norm": 8.5, "learning_rate": 2.3770420449691534e-06, "loss": 1.2246, "step": 31306 }, { "epoch": 78.2675, "grad_norm": 7.625, "learning_rate": 2.376518016028572e-06, "loss": 1.3442, "step": 31307 }, { "epoch": 78.27, "grad_norm": 6.46875, "learning_rate": 2.375994037067233e-06, "loss": 1.1416, "step": 31308 }, { "epoch": 78.2725, "grad_norm": 6.96875, "learning_rate": 2.375470108088569e-06, "loss": 1.2652, "step": 31309 }, { "epoch": 78.275, "grad_norm": 7.96875, "learning_rate": 2.374946229096019e-06, "loss": 1.3011, "step": 31310 }, { "epoch": 78.2775, "grad_norm": 6.25, "learning_rate": 2.3744224000930182e-06, "loss": 1.2452, "step": 31311 }, { "epoch": 78.28, "grad_norm": 10.25, "learning_rate": 2.3738986210829997e-06, "loss": 1.4009, "step": 31312 }, { "epoch": 78.2825, "grad_norm": 6.9375, "learning_rate": 2.373374892069393e-06, "loss": 1.2727, "step": 31313 }, { "epoch": 78.285, "grad_norm": 5.71875, "learning_rate": 2.372851213055636e-06, "loss": 1.3079, "step": 31314 }, { "epoch": 78.2875, "grad_norm": 8.125, "learning_rate": 2.372327584045163e-06, "loss": 1.3133, "step": 31315 }, { "epoch": 78.29, "grad_norm": 8.0625, "learning_rate": 2.371804005041406e-06, "loss": 1.275, "step": 31316 }, { "epoch": 78.2925, "grad_norm": 8.4375, "learning_rate": 2.371280476047794e-06, "loss": 1.3595, "step": 31317 }, { "epoch": 78.295, "grad_norm": 7.71875, "learning_rate": 2.3707569970677647e-06, "loss": 1.445, "step": 31318 }, { "epoch": 78.2975, "grad_norm": 8.6875, "learning_rate": 2.370233568104744e-06, "loss": 1.3375, "step": 31319 }, { "epoch": 78.3, "grad_norm": 7.65625, "learning_rate": 2.36971018916217e-06, "loss": 1.3141, "step": 31320 }, { "epoch": 78.3025, "grad_norm": 9.625, "learning_rate": 2.3691868602434676e-06, "loss": 1.3514, "step": 31321 }, { "epoch": 78.305, "grad_norm": 7.03125, "learning_rate": 2.3686635813520733e-06, "loss": 1.2476, "step": 31322 }, { "epoch": 78.3075, "grad_norm": 6.90625, "learning_rate": 2.3681403524914147e-06, "loss": 1.1616, "step": 31323 }, { "epoch": 78.31, "grad_norm": 7.5, "learning_rate": 2.3676171736649212e-06, "loss": 1.4814, "step": 31324 }, { "epoch": 78.3125, "grad_norm": 8.125, "learning_rate": 2.367094044876023e-06, "loss": 1.2768, "step": 31325 }, { "epoch": 78.315, "grad_norm": 12.8125, "learning_rate": 2.3665709661281534e-06, "loss": 1.3224, "step": 31326 }, { "epoch": 78.3175, "grad_norm": 8.0625, "learning_rate": 2.3660479374247405e-06, "loss": 1.4281, "step": 31327 }, { "epoch": 78.32, "grad_norm": 10.4375, "learning_rate": 2.3655249587692073e-06, "loss": 1.4032, "step": 31328 }, { "epoch": 78.3225, "grad_norm": 8.1875, "learning_rate": 2.365002030164988e-06, "loss": 1.297, "step": 31329 }, { "epoch": 78.325, "grad_norm": 10.0, "learning_rate": 2.364479151615513e-06, "loss": 1.2406, "step": 31330 }, { "epoch": 78.3275, "grad_norm": 7.3125, "learning_rate": 2.3639563231242057e-06, "loss": 1.2337, "step": 31331 }, { "epoch": 78.33, "grad_norm": 7.4375, "learning_rate": 2.3634335446944936e-06, "loss": 1.4409, "step": 31332 }, { "epoch": 78.3325, "grad_norm": 9.8125, "learning_rate": 2.3629108163298077e-06, "loss": 1.3575, "step": 31333 }, { "epoch": 78.335, "grad_norm": 8.75, "learning_rate": 2.36238813803357e-06, "loss": 1.3625, "step": 31334 }, { "epoch": 78.3375, "grad_norm": 4.875, "learning_rate": 2.3618655098092137e-06, "loss": 1.216, "step": 31335 }, { "epoch": 78.34, "grad_norm": 7.5, "learning_rate": 2.3613429316601577e-06, "loss": 1.2716, "step": 31336 }, { "epoch": 78.3425, "grad_norm": 5.6875, "learning_rate": 2.360820403589835e-06, "loss": 1.2632, "step": 31337 }, { "epoch": 78.345, "grad_norm": 6.9375, "learning_rate": 2.3602979256016688e-06, "loss": 1.3454, "step": 31338 }, { "epoch": 78.3475, "grad_norm": 5.125, "learning_rate": 2.3597754976990794e-06, "loss": 1.2453, "step": 31339 }, { "epoch": 78.35, "grad_norm": 7.03125, "learning_rate": 2.3592531198854974e-06, "loss": 1.2586, "step": 31340 }, { "epoch": 78.3525, "grad_norm": 6.09375, "learning_rate": 2.358730792164349e-06, "loss": 1.1824, "step": 31341 }, { "epoch": 78.355, "grad_norm": 6.75, "learning_rate": 2.358208514539054e-06, "loss": 1.3452, "step": 31342 }, { "epoch": 78.3575, "grad_norm": 10.0625, "learning_rate": 2.357686287013037e-06, "loss": 1.3221, "step": 31343 }, { "epoch": 78.36, "grad_norm": 10.5, "learning_rate": 2.3571641095897223e-06, "loss": 1.359, "step": 31344 }, { "epoch": 78.3625, "grad_norm": 6.65625, "learning_rate": 2.3566419822725363e-06, "loss": 1.347, "step": 31345 }, { "epoch": 78.365, "grad_norm": 7.15625, "learning_rate": 2.356119905064901e-06, "loss": 1.268, "step": 31346 }, { "epoch": 78.3675, "grad_norm": 6.625, "learning_rate": 2.3555978779702336e-06, "loss": 1.1792, "step": 31347 }, { "epoch": 78.37, "grad_norm": 7.3125, "learning_rate": 2.355075900991963e-06, "loss": 1.2495, "step": 31348 }, { "epoch": 78.3725, "grad_norm": 7.96875, "learning_rate": 2.3545539741335068e-06, "loss": 1.1797, "step": 31349 }, { "epoch": 78.375, "grad_norm": 6.875, "learning_rate": 2.3540320973982924e-06, "loss": 1.2878, "step": 31350 }, { "epoch": 78.3775, "grad_norm": 7.78125, "learning_rate": 2.353510270789734e-06, "loss": 1.3603, "step": 31351 }, { "epoch": 78.38, "grad_norm": 6.875, "learning_rate": 2.352988494311259e-06, "loss": 1.3845, "step": 31352 }, { "epoch": 78.3825, "grad_norm": 8.1875, "learning_rate": 2.3524667679662848e-06, "loss": 1.3128, "step": 31353 }, { "epoch": 78.385, "grad_norm": 9.0625, "learning_rate": 2.351945091758231e-06, "loss": 1.2942, "step": 31354 }, { "epoch": 78.3875, "grad_norm": 7.71875, "learning_rate": 2.3514234656905177e-06, "loss": 1.2005, "step": 31355 }, { "epoch": 78.39, "grad_norm": 6.78125, "learning_rate": 2.35090188976657e-06, "loss": 1.348, "step": 31356 }, { "epoch": 78.3925, "grad_norm": 9.5625, "learning_rate": 2.3503803639898025e-06, "loss": 1.3683, "step": 31357 }, { "epoch": 78.395, "grad_norm": 10.375, "learning_rate": 2.349858888363633e-06, "loss": 1.438, "step": 31358 }, { "epoch": 78.3975, "grad_norm": 6.625, "learning_rate": 2.349337462891482e-06, "loss": 1.3344, "step": 31359 }, { "epoch": 78.4, "grad_norm": 10.1875, "learning_rate": 2.3488160875767717e-06, "loss": 1.4886, "step": 31360 }, { "epoch": 78.4025, "grad_norm": 8.3125, "learning_rate": 2.348294762422917e-06, "loss": 1.2576, "step": 31361 }, { "epoch": 78.405, "grad_norm": 7.84375, "learning_rate": 2.3477734874333334e-06, "loss": 1.2507, "step": 31362 }, { "epoch": 78.4075, "grad_norm": 9.0, "learning_rate": 2.3472522626114424e-06, "loss": 1.2164, "step": 31363 }, { "epoch": 78.41, "grad_norm": 6.09375, "learning_rate": 2.3467310879606566e-06, "loss": 1.4326, "step": 31364 }, { "epoch": 78.4125, "grad_norm": 9.3125, "learning_rate": 2.3462099634843994e-06, "loss": 1.495, "step": 31365 }, { "epoch": 78.415, "grad_norm": 6.6875, "learning_rate": 2.3456888891860797e-06, "loss": 1.3706, "step": 31366 }, { "epoch": 78.4175, "grad_norm": 4.59375, "learning_rate": 2.3451678650691202e-06, "loss": 1.2713, "step": 31367 }, { "epoch": 78.42, "grad_norm": 7.625, "learning_rate": 2.3446468911369334e-06, "loss": 1.1288, "step": 31368 }, { "epoch": 78.4225, "grad_norm": 8.1875, "learning_rate": 2.3441259673929327e-06, "loss": 1.2515, "step": 31369 }, { "epoch": 78.425, "grad_norm": 11.25, "learning_rate": 2.3436050938405363e-06, "loss": 1.377, "step": 31370 }, { "epoch": 78.4275, "grad_norm": 6.875, "learning_rate": 2.3430842704831603e-06, "loss": 1.3067, "step": 31371 }, { "epoch": 78.43, "grad_norm": 7.09375, "learning_rate": 2.3425634973242182e-06, "loss": 1.3485, "step": 31372 }, { "epoch": 78.4325, "grad_norm": 7.1875, "learning_rate": 2.3420427743671203e-06, "loss": 1.189, "step": 31373 }, { "epoch": 78.435, "grad_norm": 7.65625, "learning_rate": 2.3415221016152834e-06, "loss": 1.1468, "step": 31374 }, { "epoch": 78.4375, "grad_norm": 9.6875, "learning_rate": 2.3410014790721245e-06, "loss": 1.3017, "step": 31375 }, { "epoch": 78.44, "grad_norm": 4.9375, "learning_rate": 2.340480906741053e-06, "loss": 1.3427, "step": 31376 }, { "epoch": 78.4425, "grad_norm": 7.5, "learning_rate": 2.339960384625479e-06, "loss": 1.2886, "step": 31377 }, { "epoch": 78.445, "grad_norm": 9.375, "learning_rate": 2.3394399127288227e-06, "loss": 1.4721, "step": 31378 }, { "epoch": 78.4475, "grad_norm": 5.5, "learning_rate": 2.3389194910544865e-06, "loss": 1.2627, "step": 31379 }, { "epoch": 78.45, "grad_norm": 7.25, "learning_rate": 2.3383991196058918e-06, "loss": 1.2858, "step": 31380 }, { "epoch": 78.4525, "grad_norm": 7.4375, "learning_rate": 2.3378787983864436e-06, "loss": 1.4369, "step": 31381 }, { "epoch": 78.455, "grad_norm": 7.65625, "learning_rate": 2.3373585273995567e-06, "loss": 1.3825, "step": 31382 }, { "epoch": 78.4575, "grad_norm": 7.53125, "learning_rate": 2.3368383066486422e-06, "loss": 1.3932, "step": 31383 }, { "epoch": 78.46, "grad_norm": 11.625, "learning_rate": 2.3363181361371047e-06, "loss": 1.2055, "step": 31384 }, { "epoch": 78.4625, "grad_norm": 7.9375, "learning_rate": 2.3357980158683602e-06, "loss": 1.2704, "step": 31385 }, { "epoch": 78.465, "grad_norm": 7.78125, "learning_rate": 2.3352779458458186e-06, "loss": 1.2911, "step": 31386 }, { "epoch": 78.4675, "grad_norm": 5.90625, "learning_rate": 2.334757926072889e-06, "loss": 1.2171, "step": 31387 }, { "epoch": 78.47, "grad_norm": 9.5, "learning_rate": 2.334237956552979e-06, "loss": 1.3424, "step": 31388 }, { "epoch": 78.4725, "grad_norm": 7.46875, "learning_rate": 2.333718037289494e-06, "loss": 1.3438, "step": 31389 }, { "epoch": 78.475, "grad_norm": 6.1875, "learning_rate": 2.333198168285851e-06, "loss": 1.2693, "step": 31390 }, { "epoch": 78.4775, "grad_norm": 7.96875, "learning_rate": 2.3326783495454543e-06, "loss": 1.4844, "step": 31391 }, { "epoch": 78.48, "grad_norm": 6.8125, "learning_rate": 2.332158581071712e-06, "loss": 1.428, "step": 31392 }, { "epoch": 78.4825, "grad_norm": 8.0, "learning_rate": 2.3316388628680254e-06, "loss": 1.0696, "step": 31393 }, { "epoch": 78.485, "grad_norm": 6.21875, "learning_rate": 2.3311191949378132e-06, "loss": 1.3421, "step": 31394 }, { "epoch": 78.4875, "grad_norm": 7.375, "learning_rate": 2.330599577284476e-06, "loss": 1.1671, "step": 31395 }, { "epoch": 78.49, "grad_norm": 6.625, "learning_rate": 2.3300800099114186e-06, "loss": 1.3485, "step": 31396 }, { "epoch": 78.4925, "grad_norm": 8.9375, "learning_rate": 2.3295604928220526e-06, "loss": 1.2366, "step": 31397 }, { "epoch": 78.495, "grad_norm": 8.125, "learning_rate": 2.3290410260197782e-06, "loss": 1.3386, "step": 31398 }, { "epoch": 78.4975, "grad_norm": 7.8125, "learning_rate": 2.3285216095080078e-06, "loss": 1.3535, "step": 31399 }, { "epoch": 78.5, "grad_norm": 8.6875, "learning_rate": 2.328002243290138e-06, "loss": 1.1742, "step": 31400 }, { "epoch": 78.5025, "grad_norm": 6.875, "learning_rate": 2.327482927369583e-06, "loss": 1.3177, "step": 31401 }, { "epoch": 78.505, "grad_norm": 7.03125, "learning_rate": 2.3269636617497416e-06, "loss": 1.3137, "step": 31402 }, { "epoch": 78.5075, "grad_norm": 8.125, "learning_rate": 2.3264444464340175e-06, "loss": 1.3791, "step": 31403 }, { "epoch": 78.51, "grad_norm": 8.0625, "learning_rate": 2.325925281425816e-06, "loss": 1.3771, "step": 31404 }, { "epoch": 78.5125, "grad_norm": 7.28125, "learning_rate": 2.3254061667285444e-06, "loss": 1.3874, "step": 31405 }, { "epoch": 78.515, "grad_norm": 8.875, "learning_rate": 2.324887102345602e-06, "loss": 1.2838, "step": 31406 }, { "epoch": 78.5175, "grad_norm": 5.90625, "learning_rate": 2.3243680882803933e-06, "loss": 1.5188, "step": 31407 }, { "epoch": 78.52, "grad_norm": 7.96875, "learning_rate": 2.323849124536315e-06, "loss": 1.2845, "step": 31408 }, { "epoch": 78.5225, "grad_norm": 6.65625, "learning_rate": 2.323330211116779e-06, "loss": 1.1166, "step": 31409 }, { "epoch": 78.525, "grad_norm": 7.84375, "learning_rate": 2.3228113480251845e-06, "loss": 1.4516, "step": 31410 }, { "epoch": 78.5275, "grad_norm": 12.6875, "learning_rate": 2.3222925352649306e-06, "loss": 1.2534, "step": 31411 }, { "epoch": 78.53, "grad_norm": 7.4375, "learning_rate": 2.321773772839416e-06, "loss": 1.1268, "step": 31412 }, { "epoch": 78.5325, "grad_norm": 9.1875, "learning_rate": 2.321255060752047e-06, "loss": 1.3542, "step": 31413 }, { "epoch": 78.535, "grad_norm": 5.71875, "learning_rate": 2.320736399006225e-06, "loss": 1.3957, "step": 31414 }, { "epoch": 78.5375, "grad_norm": 9.25, "learning_rate": 2.320217787605348e-06, "loss": 1.3914, "step": 31415 }, { "epoch": 78.54, "grad_norm": 7.90625, "learning_rate": 2.3196992265528118e-06, "loss": 1.5579, "step": 31416 }, { "epoch": 78.5425, "grad_norm": 6.90625, "learning_rate": 2.3191807158520243e-06, "loss": 1.361, "step": 31417 }, { "epoch": 78.545, "grad_norm": 7.8125, "learning_rate": 2.3186622555063765e-06, "loss": 1.3462, "step": 31418 }, { "epoch": 78.5475, "grad_norm": 9.375, "learning_rate": 2.3181438455192752e-06, "loss": 1.3231, "step": 31419 }, { "epoch": 78.55, "grad_norm": 7.8125, "learning_rate": 2.317625485894113e-06, "loss": 1.3312, "step": 31420 }, { "epoch": 78.5525, "grad_norm": 6.75, "learning_rate": 2.317107176634292e-06, "loss": 1.2868, "step": 31421 }, { "epoch": 78.555, "grad_norm": 6.21875, "learning_rate": 2.316588917743211e-06, "loss": 1.2865, "step": 31422 }, { "epoch": 78.5575, "grad_norm": 9.8125, "learning_rate": 2.3160707092242584e-06, "loss": 1.2627, "step": 31423 }, { "epoch": 78.56, "grad_norm": 6.90625, "learning_rate": 2.3155525510808453e-06, "loss": 1.192, "step": 31424 }, { "epoch": 78.5625, "grad_norm": 8.8125, "learning_rate": 2.3150344433163617e-06, "loss": 1.3067, "step": 31425 }, { "epoch": 78.565, "grad_norm": 6.90625, "learning_rate": 2.3145163859342046e-06, "loss": 1.4055, "step": 31426 }, { "epoch": 78.5675, "grad_norm": 6.28125, "learning_rate": 2.313998378937767e-06, "loss": 1.4217, "step": 31427 }, { "epoch": 78.57, "grad_norm": 6.15625, "learning_rate": 2.31348042233045e-06, "loss": 1.1584, "step": 31428 }, { "epoch": 78.5725, "grad_norm": 6.5, "learning_rate": 2.3129625161156498e-06, "loss": 1.3157, "step": 31429 }, { "epoch": 78.575, "grad_norm": 7.65625, "learning_rate": 2.31244466029676e-06, "loss": 1.3268, "step": 31430 }, { "epoch": 78.5775, "grad_norm": 5.3125, "learning_rate": 2.311926854877172e-06, "loss": 1.3499, "step": 31431 }, { "epoch": 78.58, "grad_norm": 6.6875, "learning_rate": 2.311409099860288e-06, "loss": 1.4245, "step": 31432 }, { "epoch": 78.5825, "grad_norm": 10.5625, "learning_rate": 2.310891395249494e-06, "loss": 1.3646, "step": 31433 }, { "epoch": 78.585, "grad_norm": 9.3125, "learning_rate": 2.310373741048192e-06, "loss": 1.3831, "step": 31434 }, { "epoch": 78.5875, "grad_norm": 6.375, "learning_rate": 2.309856137259768e-06, "loss": 1.2285, "step": 31435 }, { "epoch": 78.59, "grad_norm": 7.90625, "learning_rate": 2.3093385838876236e-06, "loss": 1.4939, "step": 31436 }, { "epoch": 78.5925, "grad_norm": 8.375, "learning_rate": 2.3088210809351473e-06, "loss": 1.3396, "step": 31437 }, { "epoch": 78.595, "grad_norm": 6.4375, "learning_rate": 2.3083036284057292e-06, "loss": 1.333, "step": 31438 }, { "epoch": 78.5975, "grad_norm": 8.375, "learning_rate": 2.307786226302764e-06, "loss": 1.4075, "step": 31439 }, { "epoch": 78.6, "grad_norm": 9.9375, "learning_rate": 2.307268874629649e-06, "loss": 1.4804, "step": 31440 }, { "epoch": 78.6025, "grad_norm": 5.78125, "learning_rate": 2.3067515733897703e-06, "loss": 1.3498, "step": 31441 }, { "epoch": 78.605, "grad_norm": 5.5625, "learning_rate": 2.306234322586518e-06, "loss": 1.2187, "step": 31442 }, { "epoch": 78.6075, "grad_norm": 6.78125, "learning_rate": 2.3057171222232845e-06, "loss": 1.2193, "step": 31443 }, { "epoch": 78.61, "grad_norm": 7.9375, "learning_rate": 2.3051999723034655e-06, "loss": 1.3484, "step": 31444 }, { "epoch": 78.6125, "grad_norm": 9.75, "learning_rate": 2.304682872830447e-06, "loss": 1.3132, "step": 31445 }, { "epoch": 78.615, "grad_norm": 11.6875, "learning_rate": 2.304165823807618e-06, "loss": 1.2486, "step": 31446 }, { "epoch": 78.6175, "grad_norm": 6.625, "learning_rate": 2.3036488252383714e-06, "loss": 1.2643, "step": 31447 }, { "epoch": 78.62, "grad_norm": 6.125, "learning_rate": 2.3031318771260935e-06, "loss": 1.2098, "step": 31448 }, { "epoch": 78.6225, "grad_norm": 5.875, "learning_rate": 2.302614979474177e-06, "loss": 1.3143, "step": 31449 }, { "epoch": 78.625, "grad_norm": 7.75, "learning_rate": 2.3020981322860057e-06, "loss": 1.258, "step": 31450 }, { "epoch": 78.6275, "grad_norm": 9.8125, "learning_rate": 2.3015813355649742e-06, "loss": 1.2964, "step": 31451 }, { "epoch": 78.63, "grad_norm": 5.4375, "learning_rate": 2.3010645893144666e-06, "loss": 1.3065, "step": 31452 }, { "epoch": 78.6325, "grad_norm": 6.9375, "learning_rate": 2.30054789353787e-06, "loss": 1.3391, "step": 31453 }, { "epoch": 78.635, "grad_norm": 8.8125, "learning_rate": 2.300031248238572e-06, "loss": 1.378, "step": 31454 }, { "epoch": 78.6375, "grad_norm": 7.9375, "learning_rate": 2.299514653419965e-06, "loss": 1.3689, "step": 31455 }, { "epoch": 78.64, "grad_norm": 5.84375, "learning_rate": 2.2989981090854306e-06, "loss": 1.2634, "step": 31456 }, { "epoch": 78.6425, "grad_norm": 6.84375, "learning_rate": 2.298481615238355e-06, "loss": 1.3236, "step": 31457 }, { "epoch": 78.645, "grad_norm": 7.21875, "learning_rate": 2.297965171882125e-06, "loss": 1.3339, "step": 31458 }, { "epoch": 78.6475, "grad_norm": 6.5625, "learning_rate": 2.2974487790201296e-06, "loss": 1.3035, "step": 31459 }, { "epoch": 78.65, "grad_norm": 7.84375, "learning_rate": 2.296932436655752e-06, "loss": 1.347, "step": 31460 }, { "epoch": 78.6525, "grad_norm": 6.75, "learning_rate": 2.2964161447923738e-06, "loss": 1.373, "step": 31461 }, { "epoch": 78.655, "grad_norm": 3.765625, "learning_rate": 2.2958999034333875e-06, "loss": 1.2892, "step": 31462 }, { "epoch": 78.6575, "grad_norm": 6.125, "learning_rate": 2.2953837125821686e-06, "loss": 1.3212, "step": 31463 }, { "epoch": 78.66, "grad_norm": 7.09375, "learning_rate": 2.2948675722421086e-06, "loss": 1.4539, "step": 31464 }, { "epoch": 78.6625, "grad_norm": 8.0625, "learning_rate": 2.294351482416587e-06, "loss": 1.4286, "step": 31465 }, { "epoch": 78.665, "grad_norm": 8.6875, "learning_rate": 2.2938354431089906e-06, "loss": 1.3942, "step": 31466 }, { "epoch": 78.6675, "grad_norm": 7.46875, "learning_rate": 2.293319454322701e-06, "loss": 1.2114, "step": 31467 }, { "epoch": 78.67, "grad_norm": 8.5625, "learning_rate": 2.2928035160610973e-06, "loss": 1.5268, "step": 31468 }, { "epoch": 78.6725, "grad_norm": 7.25, "learning_rate": 2.292287628327565e-06, "loss": 1.2637, "step": 31469 }, { "epoch": 78.675, "grad_norm": 8.75, "learning_rate": 2.291771791125491e-06, "loss": 1.4198, "step": 31470 }, { "epoch": 78.6775, "grad_norm": 7.21875, "learning_rate": 2.2912560044582512e-06, "loss": 1.2818, "step": 31471 }, { "epoch": 78.68, "grad_norm": 7.5625, "learning_rate": 2.2907402683292268e-06, "loss": 1.3414, "step": 31472 }, { "epoch": 78.6825, "grad_norm": 8.0, "learning_rate": 2.2902245827418002e-06, "loss": 1.4197, "step": 31473 }, { "epoch": 78.685, "grad_norm": 6.15625, "learning_rate": 2.289708947699355e-06, "loss": 1.2372, "step": 31474 }, { "epoch": 78.6875, "grad_norm": 9.875, "learning_rate": 2.2891933632052697e-06, "loss": 1.3768, "step": 31475 }, { "epoch": 78.69, "grad_norm": 7.75, "learning_rate": 2.2886778292629217e-06, "loss": 1.309, "step": 31476 }, { "epoch": 78.6925, "grad_norm": 5.78125, "learning_rate": 2.288162345875696e-06, "loss": 1.418, "step": 31477 }, { "epoch": 78.695, "grad_norm": 8.25, "learning_rate": 2.287646913046966e-06, "loss": 1.3991, "step": 31478 }, { "epoch": 78.6975, "grad_norm": 12.0625, "learning_rate": 2.2871315307801177e-06, "loss": 1.407, "step": 31479 }, { "epoch": 78.7, "grad_norm": 6.09375, "learning_rate": 2.2866161990785228e-06, "loss": 1.332, "step": 31480 }, { "epoch": 78.7025, "grad_norm": 7.4375, "learning_rate": 2.2861009179455673e-06, "loss": 1.2601, "step": 31481 }, { "epoch": 78.705, "grad_norm": 7.71875, "learning_rate": 2.285585687384625e-06, "loss": 1.4118, "step": 31482 }, { "epoch": 78.7075, "grad_norm": 7.90625, "learning_rate": 2.285070507399071e-06, "loss": 1.3218, "step": 31483 }, { "epoch": 78.71, "grad_norm": 9.375, "learning_rate": 2.2845553779922867e-06, "loss": 1.3673, "step": 31484 }, { "epoch": 78.7125, "grad_norm": 8.6875, "learning_rate": 2.2840402991676514e-06, "loss": 1.4073, "step": 31485 }, { "epoch": 78.715, "grad_norm": 7.46875, "learning_rate": 2.283525270928538e-06, "loss": 1.2682, "step": 31486 }, { "epoch": 78.7175, "grad_norm": 9.0, "learning_rate": 2.2830102932783227e-06, "loss": 1.4579, "step": 31487 }, { "epoch": 78.72, "grad_norm": 8.0, "learning_rate": 2.2824953662203832e-06, "loss": 1.2839, "step": 31488 }, { "epoch": 78.7225, "grad_norm": 6.09375, "learning_rate": 2.281980489758097e-06, "loss": 1.2573, "step": 31489 }, { "epoch": 78.725, "grad_norm": 5.59375, "learning_rate": 2.2814656638948384e-06, "loss": 1.1719, "step": 31490 }, { "epoch": 78.7275, "grad_norm": 6.5625, "learning_rate": 2.2809508886339783e-06, "loss": 1.3826, "step": 31491 }, { "epoch": 78.73, "grad_norm": 8.875, "learning_rate": 2.2804361639788997e-06, "loss": 1.196, "step": 31492 }, { "epoch": 78.7325, "grad_norm": 6.75, "learning_rate": 2.279921489932968e-06, "loss": 1.3714, "step": 31493 }, { "epoch": 78.735, "grad_norm": 8.1875, "learning_rate": 2.279406866499566e-06, "loss": 1.405, "step": 31494 }, { "epoch": 78.7375, "grad_norm": 8.0625, "learning_rate": 2.278892293682061e-06, "loss": 1.2233, "step": 31495 }, { "epoch": 78.74, "grad_norm": 9.625, "learning_rate": 2.2783777714838316e-06, "loss": 1.3229, "step": 31496 }, { "epoch": 78.7425, "grad_norm": 6.75, "learning_rate": 2.2778632999082482e-06, "loss": 1.3271, "step": 31497 }, { "epoch": 78.745, "grad_norm": 5.90625, "learning_rate": 2.277348878958682e-06, "loss": 1.1839, "step": 31498 }, { "epoch": 78.7475, "grad_norm": 7.59375, "learning_rate": 2.2768345086385068e-06, "loss": 1.1557, "step": 31499 }, { "epoch": 78.75, "grad_norm": 7.90625, "learning_rate": 2.2763201889510987e-06, "loss": 1.3177, "step": 31500 }, { "epoch": 78.7525, "grad_norm": 7.4375, "learning_rate": 2.2758059198998262e-06, "loss": 1.3433, "step": 31501 }, { "epoch": 78.755, "grad_norm": 6.0625, "learning_rate": 2.275291701488058e-06, "loss": 1.2793, "step": 31502 }, { "epoch": 78.7575, "grad_norm": 8.125, "learning_rate": 2.274777533719169e-06, "loss": 1.3686, "step": 31503 }, { "epoch": 78.76, "grad_norm": 6.4375, "learning_rate": 2.2742634165965317e-06, "loss": 1.2875, "step": 31504 }, { "epoch": 78.7625, "grad_norm": 11.0625, "learning_rate": 2.2737493501235144e-06, "loss": 1.2804, "step": 31505 }, { "epoch": 78.765, "grad_norm": 8.375, "learning_rate": 2.273235334303484e-06, "loss": 1.2303, "step": 31506 }, { "epoch": 78.7675, "grad_norm": 7.0, "learning_rate": 2.2727213691398165e-06, "loss": 1.2673, "step": 31507 }, { "epoch": 78.77, "grad_norm": 6.375, "learning_rate": 2.2722074546358763e-06, "loss": 1.2209, "step": 31508 }, { "epoch": 78.7725, "grad_norm": 8.9375, "learning_rate": 2.2716935907950366e-06, "loss": 1.244, "step": 31509 }, { "epoch": 78.775, "grad_norm": 4.9375, "learning_rate": 2.2711797776206623e-06, "loss": 1.3555, "step": 31510 }, { "epoch": 78.7775, "grad_norm": 10.9375, "learning_rate": 2.2706660151161263e-06, "loss": 1.3612, "step": 31511 }, { "epoch": 78.78, "grad_norm": 10.375, "learning_rate": 2.270152303284795e-06, "loss": 1.3321, "step": 31512 }, { "epoch": 78.7825, "grad_norm": 8.875, "learning_rate": 2.269638642130033e-06, "loss": 1.3558, "step": 31513 }, { "epoch": 78.785, "grad_norm": 9.125, "learning_rate": 2.2691250316552103e-06, "loss": 1.1304, "step": 31514 }, { "epoch": 78.7875, "grad_norm": 6.84375, "learning_rate": 2.2686114718636974e-06, "loss": 1.2656, "step": 31515 }, { "epoch": 78.79, "grad_norm": 6.4375, "learning_rate": 2.2680979627588593e-06, "loss": 1.1994, "step": 31516 }, { "epoch": 78.7925, "grad_norm": 9.375, "learning_rate": 2.2675845043440582e-06, "loss": 1.5383, "step": 31517 }, { "epoch": 78.795, "grad_norm": 11.0625, "learning_rate": 2.267071096622663e-06, "loss": 1.3664, "step": 31518 }, { "epoch": 78.7975, "grad_norm": 8.875, "learning_rate": 2.2665577395980444e-06, "loss": 1.3341, "step": 31519 }, { "epoch": 78.8, "grad_norm": 7.125, "learning_rate": 2.266044433273562e-06, "loss": 1.4395, "step": 31520 }, { "epoch": 78.8025, "grad_norm": 5.9375, "learning_rate": 2.2655311776525814e-06, "loss": 1.3433, "step": 31521 }, { "epoch": 78.805, "grad_norm": 5.4375, "learning_rate": 2.265017972738471e-06, "loss": 1.4713, "step": 31522 }, { "epoch": 78.8075, "grad_norm": 6.71875, "learning_rate": 2.264504818534591e-06, "loss": 1.4933, "step": 31523 }, { "epoch": 78.81, "grad_norm": 6.90625, "learning_rate": 2.2639917150443104e-06, "loss": 1.4048, "step": 31524 }, { "epoch": 78.8125, "grad_norm": 5.625, "learning_rate": 2.263478662270987e-06, "loss": 1.3106, "step": 31525 }, { "epoch": 78.815, "grad_norm": 7.375, "learning_rate": 2.2629656602179917e-06, "loss": 1.3119, "step": 31526 }, { "epoch": 78.8175, "grad_norm": 10.1875, "learning_rate": 2.262452708888684e-06, "loss": 1.3209, "step": 31527 }, { "epoch": 78.82, "grad_norm": 6.625, "learning_rate": 2.2619398082864232e-06, "loss": 1.3665, "step": 31528 }, { "epoch": 78.8225, "grad_norm": 9.5625, "learning_rate": 2.2614269584145754e-06, "loss": 1.4233, "step": 31529 }, { "epoch": 78.825, "grad_norm": 10.4375, "learning_rate": 2.260914159276506e-06, "loss": 1.3127, "step": 31530 }, { "epoch": 78.8275, "grad_norm": 7.4375, "learning_rate": 2.2604014108755735e-06, "loss": 1.2574, "step": 31531 }, { "epoch": 78.83, "grad_norm": 6.0625, "learning_rate": 2.2598887132151394e-06, "loss": 1.2913, "step": 31532 }, { "epoch": 78.8325, "grad_norm": 10.125, "learning_rate": 2.2593760662985598e-06, "loss": 1.2594, "step": 31533 }, { "epoch": 78.835, "grad_norm": 10.6875, "learning_rate": 2.2588634701292068e-06, "loss": 1.1552, "step": 31534 }, { "epoch": 78.8375, "grad_norm": 5.78125, "learning_rate": 2.258350924710434e-06, "loss": 1.3481, "step": 31535 }, { "epoch": 78.84, "grad_norm": 6.03125, "learning_rate": 2.2578384300456014e-06, "loss": 1.2577, "step": 31536 }, { "epoch": 78.8425, "grad_norm": 7.125, "learning_rate": 2.2573259861380727e-06, "loss": 1.3239, "step": 31537 }, { "epoch": 78.845, "grad_norm": 7.28125, "learning_rate": 2.2568135929912017e-06, "loss": 1.3633, "step": 31538 }, { "epoch": 78.8475, "grad_norm": 8.25, "learning_rate": 2.256301250608354e-06, "loss": 1.2545, "step": 31539 }, { "epoch": 78.85, "grad_norm": 9.0, "learning_rate": 2.2557889589928815e-06, "loss": 1.3939, "step": 31540 }, { "epoch": 78.8525, "grad_norm": 6.34375, "learning_rate": 2.2552767181481506e-06, "loss": 1.369, "step": 31541 }, { "epoch": 78.855, "grad_norm": 9.5, "learning_rate": 2.2547645280775154e-06, "loss": 1.2081, "step": 31542 }, { "epoch": 78.8575, "grad_norm": 7.0, "learning_rate": 2.2542523887843316e-06, "loss": 1.3346, "step": 31543 }, { "epoch": 78.86, "grad_norm": 5.5, "learning_rate": 2.253740300271959e-06, "loss": 1.3075, "step": 31544 }, { "epoch": 78.8625, "grad_norm": 7.84375, "learning_rate": 2.2532282625437576e-06, "loss": 1.3108, "step": 31545 }, { "epoch": 78.865, "grad_norm": 7.9375, "learning_rate": 2.252716275603082e-06, "loss": 1.3215, "step": 31546 }, { "epoch": 78.8675, "grad_norm": 7.125, "learning_rate": 2.252204339453288e-06, "loss": 1.3342, "step": 31547 }, { "epoch": 78.87, "grad_norm": 11.0625, "learning_rate": 2.2516924540977268e-06, "loss": 1.2933, "step": 31548 }, { "epoch": 78.8725, "grad_norm": 6.21875, "learning_rate": 2.2511806195397656e-06, "loss": 1.1778, "step": 31549 }, { "epoch": 78.875, "grad_norm": 6.46875, "learning_rate": 2.2506688357827546e-06, "loss": 1.2625, "step": 31550 }, { "epoch": 78.8775, "grad_norm": 7.21875, "learning_rate": 2.2501571028300475e-06, "loss": 1.2504, "step": 31551 }, { "epoch": 78.88, "grad_norm": 7.65625, "learning_rate": 2.249645420684998e-06, "loss": 1.482, "step": 31552 }, { "epoch": 78.8825, "grad_norm": 7.3125, "learning_rate": 2.2491337893509633e-06, "loss": 1.3246, "step": 31553 }, { "epoch": 78.885, "grad_norm": 7.75, "learning_rate": 2.2486222088313e-06, "loss": 1.3784, "step": 31554 }, { "epoch": 78.8875, "grad_norm": 8.75, "learning_rate": 2.2481106791293593e-06, "loss": 1.2227, "step": 31555 }, { "epoch": 78.89, "grad_norm": 9.3125, "learning_rate": 2.2475992002484915e-06, "loss": 1.294, "step": 31556 }, { "epoch": 78.8925, "grad_norm": 8.25, "learning_rate": 2.247087772192056e-06, "loss": 1.4366, "step": 31557 }, { "epoch": 78.895, "grad_norm": 10.5, "learning_rate": 2.2465763949634e-06, "loss": 1.4777, "step": 31558 }, { "epoch": 78.8975, "grad_norm": 7.34375, "learning_rate": 2.2460650685658823e-06, "loss": 1.2619, "step": 31559 }, { "epoch": 78.9, "grad_norm": 6.0, "learning_rate": 2.245553793002849e-06, "loss": 1.4368, "step": 31560 }, { "epoch": 78.9025, "grad_norm": 6.75, "learning_rate": 2.2450425682776565e-06, "loss": 1.3167, "step": 31561 }, { "epoch": 78.905, "grad_norm": 7.5625, "learning_rate": 2.244531394393655e-06, "loss": 1.1933, "step": 31562 }, { "epoch": 78.9075, "grad_norm": 6.46875, "learning_rate": 2.24402027135419e-06, "loss": 1.2655, "step": 31563 }, { "epoch": 78.91, "grad_norm": 7.75, "learning_rate": 2.2435091991626233e-06, "loss": 1.2994, "step": 31564 }, { "epoch": 78.9125, "grad_norm": 9.375, "learning_rate": 2.2429981778222996e-06, "loss": 1.3387, "step": 31565 }, { "epoch": 78.915, "grad_norm": 11.0625, "learning_rate": 2.24248720733657e-06, "loss": 1.2643, "step": 31566 }, { "epoch": 78.9175, "grad_norm": 4.875, "learning_rate": 2.2419762877087814e-06, "loss": 1.2279, "step": 31567 }, { "epoch": 78.92, "grad_norm": 6.71875, "learning_rate": 2.2414654189422845e-06, "loss": 1.4346, "step": 31568 }, { "epoch": 78.9225, "grad_norm": 6.8125, "learning_rate": 2.2409546010404337e-06, "loss": 1.2775, "step": 31569 }, { "epoch": 78.925, "grad_norm": 9.0625, "learning_rate": 2.2404438340065736e-06, "loss": 1.3354, "step": 31570 }, { "epoch": 78.9275, "grad_norm": 5.46875, "learning_rate": 2.23993311784405e-06, "loss": 1.3395, "step": 31571 }, { "epoch": 78.93, "grad_norm": 7.65625, "learning_rate": 2.239422452556218e-06, "loss": 1.4394, "step": 31572 }, { "epoch": 78.9325, "grad_norm": 6.90625, "learning_rate": 2.238911838146418e-06, "loss": 1.3725, "step": 31573 }, { "epoch": 78.935, "grad_norm": 7.6875, "learning_rate": 2.238401274618005e-06, "loss": 1.4016, "step": 31574 }, { "epoch": 78.9375, "grad_norm": 9.8125, "learning_rate": 2.2378907619743196e-06, "loss": 1.4625, "step": 31575 }, { "epoch": 78.94, "grad_norm": 9.8125, "learning_rate": 2.2373803002187146e-06, "loss": 1.2352, "step": 31576 }, { "epoch": 78.9425, "grad_norm": 8.3125, "learning_rate": 2.2368698893545338e-06, "loss": 1.4382, "step": 31577 }, { "epoch": 78.945, "grad_norm": 7.84375, "learning_rate": 2.2363595293851204e-06, "loss": 1.4365, "step": 31578 }, { "epoch": 78.9475, "grad_norm": 6.21875, "learning_rate": 2.235849220313823e-06, "loss": 1.3548, "step": 31579 }, { "epoch": 78.95, "grad_norm": 8.9375, "learning_rate": 2.23533896214399e-06, "loss": 1.3914, "step": 31580 }, { "epoch": 78.9525, "grad_norm": 9.75, "learning_rate": 2.234828754878965e-06, "loss": 1.2195, "step": 31581 }, { "epoch": 78.955, "grad_norm": 8.375, "learning_rate": 2.2343185985220885e-06, "loss": 1.2744, "step": 31582 }, { "epoch": 78.9575, "grad_norm": 8.375, "learning_rate": 2.233808493076708e-06, "loss": 1.3882, "step": 31583 }, { "epoch": 78.96, "grad_norm": 10.5625, "learning_rate": 2.233298438546172e-06, "loss": 1.4354, "step": 31584 }, { "epoch": 78.9625, "grad_norm": 8.0625, "learning_rate": 2.23278843493382e-06, "loss": 1.2004, "step": 31585 }, { "epoch": 78.965, "grad_norm": 7.6875, "learning_rate": 2.2322784822429946e-06, "loss": 1.3471, "step": 31586 }, { "epoch": 78.9675, "grad_norm": 7.9375, "learning_rate": 2.2317685804770417e-06, "loss": 1.2076, "step": 31587 }, { "epoch": 78.97, "grad_norm": 8.6875, "learning_rate": 2.2312587296393005e-06, "loss": 1.4696, "step": 31588 }, { "epoch": 78.9725, "grad_norm": 7.0, "learning_rate": 2.230748929733121e-06, "loss": 1.2748, "step": 31589 }, { "epoch": 78.975, "grad_norm": 5.78125, "learning_rate": 2.2302391807618363e-06, "loss": 1.2106, "step": 31590 }, { "epoch": 78.9775, "grad_norm": 9.375, "learning_rate": 2.2297294827287952e-06, "loss": 1.3359, "step": 31591 }, { "epoch": 78.98, "grad_norm": 7.3125, "learning_rate": 2.2292198356373362e-06, "loss": 1.267, "step": 31592 }, { "epoch": 78.9825, "grad_norm": 8.0625, "learning_rate": 2.2287102394907988e-06, "loss": 1.4947, "step": 31593 }, { "epoch": 78.985, "grad_norm": 9.125, "learning_rate": 2.2282006942925274e-06, "loss": 1.4107, "step": 31594 }, { "epoch": 78.9875, "grad_norm": 10.0625, "learning_rate": 2.2276912000458627e-06, "loss": 1.227, "step": 31595 }, { "epoch": 78.99, "grad_norm": 10.375, "learning_rate": 2.227181756754143e-06, "loss": 1.3557, "step": 31596 }, { "epoch": 78.9925, "grad_norm": 9.5625, "learning_rate": 2.2266723644207068e-06, "loss": 1.3949, "step": 31597 }, { "epoch": 78.995, "grad_norm": 10.0, "learning_rate": 2.226163023048895e-06, "loss": 1.2734, "step": 31598 }, { "epoch": 78.9975, "grad_norm": 7.375, "learning_rate": 2.22565373264205e-06, "loss": 1.2876, "step": 31599 }, { "epoch": 79.0, "grad_norm": 7.90625, "learning_rate": 2.2251444932035094e-06, "loss": 1.2953, "step": 31600 }, { "epoch": 79.0025, "grad_norm": 6.875, "learning_rate": 2.2246353047366065e-06, "loss": 1.2553, "step": 31601 }, { "epoch": 79.005, "grad_norm": 6.875, "learning_rate": 2.2241261672446855e-06, "loss": 1.2395, "step": 31602 }, { "epoch": 79.0075, "grad_norm": 6.3125, "learning_rate": 2.2236170807310795e-06, "loss": 1.2631, "step": 31603 }, { "epoch": 79.01, "grad_norm": 6.15625, "learning_rate": 2.2231080451991326e-06, "loss": 1.321, "step": 31604 }, { "epoch": 79.0125, "grad_norm": 8.25, "learning_rate": 2.2225990606521752e-06, "loss": 1.4322, "step": 31605 }, { "epoch": 79.015, "grad_norm": 6.9375, "learning_rate": 2.222090127093549e-06, "loss": 1.4208, "step": 31606 }, { "epoch": 79.0175, "grad_norm": 10.3125, "learning_rate": 2.2215812445265896e-06, "loss": 1.4402, "step": 31607 }, { "epoch": 79.02, "grad_norm": 9.1875, "learning_rate": 2.221072412954629e-06, "loss": 1.3542, "step": 31608 }, { "epoch": 79.0225, "grad_norm": 7.625, "learning_rate": 2.2205636323810066e-06, "loss": 1.244, "step": 31609 }, { "epoch": 79.025, "grad_norm": 7.90625, "learning_rate": 2.2200549028090603e-06, "loss": 1.178, "step": 31610 }, { "epoch": 79.0275, "grad_norm": 6.9375, "learning_rate": 2.2195462242421227e-06, "loss": 1.1532, "step": 31611 }, { "epoch": 79.03, "grad_norm": 8.8125, "learning_rate": 2.2190375966835266e-06, "loss": 1.1969, "step": 31612 }, { "epoch": 79.0325, "grad_norm": 9.25, "learning_rate": 2.2185290201366085e-06, "loss": 1.1633, "step": 31613 }, { "epoch": 79.035, "grad_norm": 8.125, "learning_rate": 2.2180204946047047e-06, "loss": 1.3769, "step": 31614 }, { "epoch": 79.0375, "grad_norm": 6.96875, "learning_rate": 2.217512020091147e-06, "loss": 1.3736, "step": 31615 }, { "epoch": 79.04, "grad_norm": 8.9375, "learning_rate": 2.2170035965992674e-06, "loss": 1.4109, "step": 31616 }, { "epoch": 79.0425, "grad_norm": 7.875, "learning_rate": 2.216495224132402e-06, "loss": 1.2784, "step": 31617 }, { "epoch": 79.045, "grad_norm": 5.21875, "learning_rate": 2.2159869026938807e-06, "loss": 1.3227, "step": 31618 }, { "epoch": 79.0475, "grad_norm": 7.46875, "learning_rate": 2.2154786322870415e-06, "loss": 1.2577, "step": 31619 }, { "epoch": 79.05, "grad_norm": 6.6875, "learning_rate": 2.2149704129152083e-06, "loss": 1.2802, "step": 31620 }, { "epoch": 79.0525, "grad_norm": 9.9375, "learning_rate": 2.214462244581721e-06, "loss": 1.1749, "step": 31621 }, { "epoch": 79.055, "grad_norm": 8.25, "learning_rate": 2.213954127289908e-06, "loss": 1.3617, "step": 31622 }, { "epoch": 79.0575, "grad_norm": 5.625, "learning_rate": 2.2134460610430974e-06, "loss": 1.2438, "step": 31623 }, { "epoch": 79.06, "grad_norm": 8.8125, "learning_rate": 2.2129380458446226e-06, "loss": 1.4941, "step": 31624 }, { "epoch": 79.0625, "grad_norm": 7.71875, "learning_rate": 2.212430081697816e-06, "loss": 1.2004, "step": 31625 }, { "epoch": 79.065, "grad_norm": 8.0, "learning_rate": 2.211922168606008e-06, "loss": 1.5045, "step": 31626 }, { "epoch": 79.0675, "grad_norm": 7.21875, "learning_rate": 2.2114143065725224e-06, "loss": 1.344, "step": 31627 }, { "epoch": 79.07, "grad_norm": 7.96875, "learning_rate": 2.2109064956006944e-06, "loss": 1.3115, "step": 31628 }, { "epoch": 79.0725, "grad_norm": 7.46875, "learning_rate": 2.2103987356938528e-06, "loss": 1.3155, "step": 31629 }, { "epoch": 79.075, "grad_norm": 10.5, "learning_rate": 2.2098910268553263e-06, "loss": 1.3678, "step": 31630 }, { "epoch": 79.0775, "grad_norm": 8.25, "learning_rate": 2.2093833690884405e-06, "loss": 1.271, "step": 31631 }, { "epoch": 79.08, "grad_norm": 10.3125, "learning_rate": 2.2088757623965263e-06, "loss": 1.3324, "step": 31632 }, { "epoch": 79.0825, "grad_norm": 6.65625, "learning_rate": 2.208368206782909e-06, "loss": 1.4703, "step": 31633 }, { "epoch": 79.085, "grad_norm": 7.03125, "learning_rate": 2.2078607022509213e-06, "loss": 1.3001, "step": 31634 }, { "epoch": 79.0875, "grad_norm": 8.375, "learning_rate": 2.207353248803884e-06, "loss": 1.3595, "step": 31635 }, { "epoch": 79.09, "grad_norm": 7.0625, "learning_rate": 2.2068458464451292e-06, "loss": 1.3202, "step": 31636 }, { "epoch": 79.0925, "grad_norm": 8.375, "learning_rate": 2.206338495177979e-06, "loss": 1.382, "step": 31637 }, { "epoch": 79.095, "grad_norm": 5.96875, "learning_rate": 2.2058311950057644e-06, "loss": 1.307, "step": 31638 }, { "epoch": 79.0975, "grad_norm": 7.90625, "learning_rate": 2.2053239459318055e-06, "loss": 1.377, "step": 31639 }, { "epoch": 79.1, "grad_norm": 8.375, "learning_rate": 2.204816747959434e-06, "loss": 1.3993, "step": 31640 }, { "epoch": 79.1025, "grad_norm": 8.5, "learning_rate": 2.204309601091972e-06, "loss": 1.4068, "step": 31641 }, { "epoch": 79.105, "grad_norm": 7.4375, "learning_rate": 2.2038025053327417e-06, "loss": 1.3364, "step": 31642 }, { "epoch": 79.1075, "grad_norm": 9.625, "learning_rate": 2.20329546068507e-06, "loss": 1.3289, "step": 31643 }, { "epoch": 79.11, "grad_norm": 6.5, "learning_rate": 2.2027884671522835e-06, "loss": 1.3642, "step": 31644 }, { "epoch": 79.1125, "grad_norm": 5.875, "learning_rate": 2.202281524737704e-06, "loss": 1.2492, "step": 31645 }, { "epoch": 79.115, "grad_norm": 7.65625, "learning_rate": 2.201774633444652e-06, "loss": 1.3124, "step": 31646 }, { "epoch": 79.1175, "grad_norm": 7.15625, "learning_rate": 2.2012677932764537e-06, "loss": 1.3437, "step": 31647 }, { "epoch": 79.12, "grad_norm": 8.5, "learning_rate": 2.2007610042364337e-06, "loss": 1.5457, "step": 31648 }, { "epoch": 79.1225, "grad_norm": 8.3125, "learning_rate": 2.200254266327914e-06, "loss": 1.1802, "step": 31649 }, { "epoch": 79.125, "grad_norm": 5.34375, "learning_rate": 2.1997475795542113e-06, "loss": 1.3551, "step": 31650 }, { "epoch": 79.1275, "grad_norm": 5.5625, "learning_rate": 2.1992409439186534e-06, "loss": 1.3363, "step": 31651 }, { "epoch": 79.13, "grad_norm": 6.75, "learning_rate": 2.198734359424558e-06, "loss": 1.2473, "step": 31652 }, { "epoch": 79.1325, "grad_norm": 7.53125, "learning_rate": 2.198227826075251e-06, "loss": 1.2563, "step": 31653 }, { "epoch": 79.135, "grad_norm": 5.875, "learning_rate": 2.1977213438740463e-06, "loss": 1.3724, "step": 31654 }, { "epoch": 79.1375, "grad_norm": 8.375, "learning_rate": 2.1972149128242713e-06, "loss": 1.1509, "step": 31655 }, { "epoch": 79.14, "grad_norm": 6.6875, "learning_rate": 2.1967085329292435e-06, "loss": 1.1707, "step": 31656 }, { "epoch": 79.1425, "grad_norm": 8.3125, "learning_rate": 2.1962022041922792e-06, "loss": 1.2999, "step": 31657 }, { "epoch": 79.145, "grad_norm": 6.625, "learning_rate": 2.195695926616702e-06, "loss": 1.2458, "step": 31658 }, { "epoch": 79.1475, "grad_norm": 10.375, "learning_rate": 2.195189700205832e-06, "loss": 1.3917, "step": 31659 }, { "epoch": 79.15, "grad_norm": 7.5, "learning_rate": 2.194683524962986e-06, "loss": 1.4882, "step": 31660 }, { "epoch": 79.1525, "grad_norm": 12.125, "learning_rate": 2.1941774008914796e-06, "loss": 1.229, "step": 31661 }, { "epoch": 79.155, "grad_norm": 6.90625, "learning_rate": 2.1936713279946332e-06, "loss": 1.2633, "step": 31662 }, { "epoch": 79.1575, "grad_norm": 13.0625, "learning_rate": 2.1931653062757687e-06, "loss": 1.4464, "step": 31663 }, { "epoch": 79.16, "grad_norm": 10.9375, "learning_rate": 2.1926593357382e-06, "loss": 1.3171, "step": 31664 }, { "epoch": 79.1625, "grad_norm": 7.4375, "learning_rate": 2.192153416385241e-06, "loss": 1.4623, "step": 31665 }, { "epoch": 79.165, "grad_norm": 10.6875, "learning_rate": 2.191647548220215e-06, "loss": 1.316, "step": 31666 }, { "epoch": 79.1675, "grad_norm": 6.90625, "learning_rate": 2.1911417312464323e-06, "loss": 1.4068, "step": 31667 }, { "epoch": 79.17, "grad_norm": 7.5, "learning_rate": 2.190635965467215e-06, "loss": 1.369, "step": 31668 }, { "epoch": 79.1725, "grad_norm": 7.9375, "learning_rate": 2.190130250885872e-06, "loss": 1.2332, "step": 31669 }, { "epoch": 79.175, "grad_norm": 5.34375, "learning_rate": 2.1896245875057266e-06, "loss": 1.3031, "step": 31670 }, { "epoch": 79.1775, "grad_norm": 8.25, "learning_rate": 2.1891189753300903e-06, "loss": 1.2811, "step": 31671 }, { "epoch": 79.18, "grad_norm": 5.625, "learning_rate": 2.188613414362273e-06, "loss": 1.1821, "step": 31672 }, { "epoch": 79.1825, "grad_norm": 6.6875, "learning_rate": 2.1881079046055944e-06, "loss": 1.2821, "step": 31673 }, { "epoch": 79.185, "grad_norm": 7.4375, "learning_rate": 2.1876024460633706e-06, "loss": 1.4811, "step": 31674 }, { "epoch": 79.1875, "grad_norm": 9.5, "learning_rate": 2.1870970387389124e-06, "loss": 1.1959, "step": 31675 }, { "epoch": 79.19, "grad_norm": 9.0, "learning_rate": 2.1865916826355293e-06, "loss": 1.3961, "step": 31676 }, { "epoch": 79.1925, "grad_norm": 6.65625, "learning_rate": 2.18608637775654e-06, "loss": 1.3658, "step": 31677 }, { "epoch": 79.195, "grad_norm": 4.21875, "learning_rate": 2.1855811241052583e-06, "loss": 1.2664, "step": 31678 }, { "epoch": 79.1975, "grad_norm": 8.125, "learning_rate": 2.1850759216849937e-06, "loss": 1.1883, "step": 31679 }, { "epoch": 79.2, "grad_norm": 16.0, "learning_rate": 2.184570770499056e-06, "loss": 1.2379, "step": 31680 }, { "epoch": 79.2025, "grad_norm": 11.375, "learning_rate": 2.184065670550762e-06, "loss": 1.4121, "step": 31681 }, { "epoch": 79.205, "grad_norm": 8.75, "learning_rate": 2.183560621843419e-06, "loss": 1.2052, "step": 31682 }, { "epoch": 79.2075, "grad_norm": 5.875, "learning_rate": 2.1830556243803414e-06, "loss": 1.4501, "step": 31683 }, { "epoch": 79.21, "grad_norm": 15.3125, "learning_rate": 2.1825506781648365e-06, "loss": 1.4289, "step": 31684 }, { "epoch": 79.2125, "grad_norm": 8.3125, "learning_rate": 2.1820457832002185e-06, "loss": 1.0801, "step": 31685 }, { "epoch": 79.215, "grad_norm": 9.75, "learning_rate": 2.1815409394897957e-06, "loss": 1.3418, "step": 31686 }, { "epoch": 79.2175, "grad_norm": 5.53125, "learning_rate": 2.1810361470368746e-06, "loss": 1.6374, "step": 31687 }, { "epoch": 79.22, "grad_norm": 8.4375, "learning_rate": 2.180531405844768e-06, "loss": 1.2977, "step": 31688 }, { "epoch": 79.2225, "grad_norm": 8.625, "learning_rate": 2.1800267159167877e-06, "loss": 1.274, "step": 31689 }, { "epoch": 79.225, "grad_norm": 7.21875, "learning_rate": 2.1795220772562385e-06, "loss": 1.3925, "step": 31690 }, { "epoch": 79.2275, "grad_norm": 5.78125, "learning_rate": 2.1790174898664296e-06, "loss": 1.2741, "step": 31691 }, { "epoch": 79.23, "grad_norm": 6.1875, "learning_rate": 2.1785129537506646e-06, "loss": 1.34, "step": 31692 }, { "epoch": 79.2325, "grad_norm": 6.875, "learning_rate": 2.1780084689122604e-06, "loss": 1.345, "step": 31693 }, { "epoch": 79.235, "grad_norm": 8.0, "learning_rate": 2.1775040353545195e-06, "loss": 1.3499, "step": 31694 }, { "epoch": 79.2375, "grad_norm": 7.53125, "learning_rate": 2.1769996530807493e-06, "loss": 1.3437, "step": 31695 }, { "epoch": 79.24, "grad_norm": 7.0625, "learning_rate": 2.176495322094254e-06, "loss": 1.4193, "step": 31696 }, { "epoch": 79.2425, "grad_norm": 7.1875, "learning_rate": 2.175991042398341e-06, "loss": 1.3146, "step": 31697 }, { "epoch": 79.245, "grad_norm": 6.4375, "learning_rate": 2.1754868139963213e-06, "loss": 1.4076, "step": 31698 }, { "epoch": 79.2475, "grad_norm": 7.5, "learning_rate": 2.174982636891497e-06, "loss": 1.4345, "step": 31699 }, { "epoch": 79.25, "grad_norm": 6.40625, "learning_rate": 2.1744785110871713e-06, "loss": 1.2197, "step": 31700 }, { "epoch": 79.2525, "grad_norm": 5.125, "learning_rate": 2.173974436586653e-06, "loss": 1.1534, "step": 31701 }, { "epoch": 79.255, "grad_norm": 9.0, "learning_rate": 2.173470413393243e-06, "loss": 1.3154, "step": 31702 }, { "epoch": 79.2575, "grad_norm": 10.3125, "learning_rate": 2.172966441510247e-06, "loss": 1.234, "step": 31703 }, { "epoch": 79.26, "grad_norm": 6.4375, "learning_rate": 2.1724625209409735e-06, "loss": 1.2792, "step": 31704 }, { "epoch": 79.2625, "grad_norm": 8.0625, "learning_rate": 2.1719586516887214e-06, "loss": 1.2785, "step": 31705 }, { "epoch": 79.265, "grad_norm": 7.46875, "learning_rate": 2.1714548337567962e-06, "loss": 1.371, "step": 31706 }, { "epoch": 79.2675, "grad_norm": 6.9375, "learning_rate": 2.1709510671484935e-06, "loss": 1.1615, "step": 31707 }, { "epoch": 79.27, "grad_norm": 6.53125, "learning_rate": 2.1704473518671287e-06, "loss": 1.2421, "step": 31708 }, { "epoch": 79.2725, "grad_norm": 6.0, "learning_rate": 2.169943687915996e-06, "loss": 1.2494, "step": 31709 }, { "epoch": 79.275, "grad_norm": 7.03125, "learning_rate": 2.1694400752984014e-06, "loss": 1.1904, "step": 31710 }, { "epoch": 79.2775, "grad_norm": 7.5625, "learning_rate": 2.16893651401764e-06, "loss": 1.445, "step": 31711 }, { "epoch": 79.28, "grad_norm": 5.90625, "learning_rate": 2.1684330040770183e-06, "loss": 1.3022, "step": 31712 }, { "epoch": 79.2825, "grad_norm": 8.75, "learning_rate": 2.1679295454798386e-06, "loss": 1.3297, "step": 31713 }, { "epoch": 79.285, "grad_norm": 6.5625, "learning_rate": 2.1674261382293993e-06, "loss": 1.4392, "step": 31714 }, { "epoch": 79.2875, "grad_norm": 4.90625, "learning_rate": 2.1669227823289984e-06, "loss": 1.279, "step": 31715 }, { "epoch": 79.29, "grad_norm": 13.75, "learning_rate": 2.1664194777819414e-06, "loss": 1.3162, "step": 31716 }, { "epoch": 79.2925, "grad_norm": 9.5, "learning_rate": 2.165916224591521e-06, "loss": 1.3373, "step": 31717 }, { "epoch": 79.295, "grad_norm": 7.0, "learning_rate": 2.1654130227610435e-06, "loss": 1.2684, "step": 31718 }, { "epoch": 79.2975, "grad_norm": 7.4375, "learning_rate": 2.1649098722938023e-06, "loss": 1.2227, "step": 31719 }, { "epoch": 79.3, "grad_norm": 6.90625, "learning_rate": 2.1644067731931005e-06, "loss": 1.295, "step": 31720 }, { "epoch": 79.3025, "grad_norm": 9.25, "learning_rate": 2.163903725462235e-06, "loss": 1.2211, "step": 31721 }, { "epoch": 79.305, "grad_norm": 8.5625, "learning_rate": 2.1634007291045013e-06, "loss": 1.3686, "step": 31722 }, { "epoch": 79.3075, "grad_norm": 10.3125, "learning_rate": 2.162897784123197e-06, "loss": 1.3058, "step": 31723 }, { "epoch": 79.31, "grad_norm": 7.78125, "learning_rate": 2.162394890521625e-06, "loss": 1.1678, "step": 31724 }, { "epoch": 79.3125, "grad_norm": 7.9375, "learning_rate": 2.161892048303078e-06, "loss": 1.2562, "step": 31725 }, { "epoch": 79.315, "grad_norm": 5.21875, "learning_rate": 2.1613892574708504e-06, "loss": 1.352, "step": 31726 }, { "epoch": 79.3175, "grad_norm": 8.625, "learning_rate": 2.1608865180282413e-06, "loss": 1.3991, "step": 31727 }, { "epoch": 79.32, "grad_norm": 5.46875, "learning_rate": 2.1603838299785486e-06, "loss": 1.2528, "step": 31728 }, { "epoch": 79.3225, "grad_norm": 9.5625, "learning_rate": 2.1598811933250664e-06, "loss": 1.2453, "step": 31729 }, { "epoch": 79.325, "grad_norm": 8.3125, "learning_rate": 2.1593786080710867e-06, "loss": 1.3515, "step": 31730 }, { "epoch": 79.3275, "grad_norm": 8.25, "learning_rate": 2.15887607421991e-06, "loss": 1.2986, "step": 31731 }, { "epoch": 79.33, "grad_norm": 6.125, "learning_rate": 2.1583735917748238e-06, "loss": 1.3434, "step": 31732 }, { "epoch": 79.3325, "grad_norm": 8.75, "learning_rate": 2.1578711607391312e-06, "loss": 1.3631, "step": 31733 }, { "epoch": 79.335, "grad_norm": 11.1875, "learning_rate": 2.1573687811161172e-06, "loss": 1.3587, "step": 31734 }, { "epoch": 79.3375, "grad_norm": 6.875, "learning_rate": 2.156866452909083e-06, "loss": 1.5404, "step": 31735 }, { "epoch": 79.34, "grad_norm": 7.6875, "learning_rate": 2.1563641761213183e-06, "loss": 1.3724, "step": 31736 }, { "epoch": 79.3425, "grad_norm": 7.78125, "learning_rate": 2.155861950756114e-06, "loss": 1.3521, "step": 31737 }, { "epoch": 79.345, "grad_norm": 6.4375, "learning_rate": 2.1553597768167646e-06, "loss": 1.2993, "step": 31738 }, { "epoch": 79.3475, "grad_norm": 7.5625, "learning_rate": 2.1548576543065648e-06, "loss": 1.2101, "step": 31739 }, { "epoch": 79.35, "grad_norm": 7.71875, "learning_rate": 2.1543555832288056e-06, "loss": 1.5274, "step": 31740 }, { "epoch": 79.3525, "grad_norm": 7.375, "learning_rate": 2.153853563586773e-06, "loss": 1.3901, "step": 31741 }, { "epoch": 79.355, "grad_norm": 7.09375, "learning_rate": 2.1533515953837625e-06, "loss": 1.3576, "step": 31742 }, { "epoch": 79.3575, "grad_norm": 6.90625, "learning_rate": 2.152849678623069e-06, "loss": 1.2062, "step": 31743 }, { "epoch": 79.36, "grad_norm": 6.1875, "learning_rate": 2.1523478133079776e-06, "loss": 1.2268, "step": 31744 }, { "epoch": 79.3625, "grad_norm": 6.0, "learning_rate": 2.1518459994417772e-06, "loss": 1.276, "step": 31745 }, { "epoch": 79.365, "grad_norm": 5.59375, "learning_rate": 2.1513442370277635e-06, "loss": 1.3735, "step": 31746 }, { "epoch": 79.3675, "grad_norm": 8.1875, "learning_rate": 2.1508425260692213e-06, "loss": 1.3187, "step": 31747 }, { "epoch": 79.37, "grad_norm": 6.28125, "learning_rate": 2.1503408665694427e-06, "loss": 1.3482, "step": 31748 }, { "epoch": 79.3725, "grad_norm": 9.1875, "learning_rate": 2.149839258531713e-06, "loss": 1.2625, "step": 31749 }, { "epoch": 79.375, "grad_norm": 6.40625, "learning_rate": 2.149337701959325e-06, "loss": 1.3031, "step": 31750 }, { "epoch": 79.3775, "grad_norm": 5.71875, "learning_rate": 2.148836196855566e-06, "loss": 1.1672, "step": 31751 }, { "epoch": 79.38, "grad_norm": 9.1875, "learning_rate": 2.148334743223719e-06, "loss": 1.308, "step": 31752 }, { "epoch": 79.3825, "grad_norm": 7.125, "learning_rate": 2.147833341067076e-06, "loss": 1.3027, "step": 31753 }, { "epoch": 79.385, "grad_norm": 7.71875, "learning_rate": 2.1473319903889257e-06, "loss": 1.3495, "step": 31754 }, { "epoch": 79.3875, "grad_norm": 7.15625, "learning_rate": 2.146830691192553e-06, "loss": 1.2032, "step": 31755 }, { "epoch": 79.39, "grad_norm": 12.4375, "learning_rate": 2.1463294434812408e-06, "loss": 1.3907, "step": 31756 }, { "epoch": 79.3925, "grad_norm": 5.40625, "learning_rate": 2.1458282472582793e-06, "loss": 1.4139, "step": 31757 }, { "epoch": 79.395, "grad_norm": 8.5, "learning_rate": 2.145327102526955e-06, "loss": 1.3969, "step": 31758 }, { "epoch": 79.3975, "grad_norm": 6.46875, "learning_rate": 2.1448260092905526e-06, "loss": 1.2215, "step": 31759 }, { "epoch": 79.4, "grad_norm": 10.25, "learning_rate": 2.1443249675523536e-06, "loss": 1.3177, "step": 31760 }, { "epoch": 79.4025, "grad_norm": 8.5625, "learning_rate": 2.1438239773156487e-06, "loss": 1.3192, "step": 31761 }, { "epoch": 79.405, "grad_norm": 8.625, "learning_rate": 2.143323038583717e-06, "loss": 1.4672, "step": 31762 }, { "epoch": 79.4075, "grad_norm": 9.5, "learning_rate": 2.1428221513598467e-06, "loss": 1.3263, "step": 31763 }, { "epoch": 79.41, "grad_norm": 9.3125, "learning_rate": 2.1423213156473175e-06, "loss": 1.4146, "step": 31764 }, { "epoch": 79.4125, "grad_norm": 6.75, "learning_rate": 2.1418205314494188e-06, "loss": 1.363, "step": 31765 }, { "epoch": 79.415, "grad_norm": 6.75, "learning_rate": 2.14131979876943e-06, "loss": 1.3499, "step": 31766 }, { "epoch": 79.4175, "grad_norm": 6.96875, "learning_rate": 2.140819117610631e-06, "loss": 1.3439, "step": 31767 }, { "epoch": 79.42, "grad_norm": 5.5625, "learning_rate": 2.1403184879763083e-06, "loss": 1.3072, "step": 31768 }, { "epoch": 79.4225, "grad_norm": 6.53125, "learning_rate": 2.1398179098697448e-06, "loss": 1.2866, "step": 31769 }, { "epoch": 79.425, "grad_norm": 9.3125, "learning_rate": 2.1393173832942203e-06, "loss": 1.1973, "step": 31770 }, { "epoch": 79.4275, "grad_norm": 8.1875, "learning_rate": 2.1388169082530143e-06, "loss": 1.3455, "step": 31771 }, { "epoch": 79.43, "grad_norm": 7.25, "learning_rate": 2.1383164847494107e-06, "loss": 1.3435, "step": 31772 }, { "epoch": 79.4325, "grad_norm": 9.6875, "learning_rate": 2.1378161127866913e-06, "loss": 1.3798, "step": 31773 }, { "epoch": 79.435, "grad_norm": 7.125, "learning_rate": 2.1373157923681353e-06, "loss": 1.257, "step": 31774 }, { "epoch": 79.4375, "grad_norm": 8.25, "learning_rate": 2.136815523497019e-06, "loss": 1.1295, "step": 31775 }, { "epoch": 79.44, "grad_norm": 8.0625, "learning_rate": 2.1363153061766297e-06, "loss": 1.3256, "step": 31776 }, { "epoch": 79.4425, "grad_norm": 7.8125, "learning_rate": 2.1358151404102378e-06, "loss": 1.2294, "step": 31777 }, { "epoch": 79.445, "grad_norm": 7.125, "learning_rate": 2.135315026201131e-06, "loss": 1.2012, "step": 31778 }, { "epoch": 79.4475, "grad_norm": 5.875, "learning_rate": 2.1348149635525815e-06, "loss": 1.4024, "step": 31779 }, { "epoch": 79.45, "grad_norm": 5.65625, "learning_rate": 2.134314952467873e-06, "loss": 1.2333, "step": 31780 }, { "epoch": 79.4525, "grad_norm": 6.34375, "learning_rate": 2.1338149929502803e-06, "loss": 1.2232, "step": 31781 }, { "epoch": 79.455, "grad_norm": 11.75, "learning_rate": 2.13331508500308e-06, "loss": 1.2478, "step": 31782 }, { "epoch": 79.4575, "grad_norm": 8.6875, "learning_rate": 2.1328152286295512e-06, "loss": 1.3359, "step": 31783 }, { "epoch": 79.46, "grad_norm": 7.75, "learning_rate": 2.1323154238329722e-06, "loss": 1.3247, "step": 31784 }, { "epoch": 79.4625, "grad_norm": 7.875, "learning_rate": 2.1318156706166205e-06, "loss": 1.3715, "step": 31785 }, { "epoch": 79.465, "grad_norm": 39.5, "learning_rate": 2.131315968983766e-06, "loss": 1.2281, "step": 31786 }, { "epoch": 79.4675, "grad_norm": 6.46875, "learning_rate": 2.1308163189376897e-06, "loss": 1.3192, "step": 31787 }, { "epoch": 79.47, "grad_norm": 8.75, "learning_rate": 2.1303167204816698e-06, "loss": 1.2633, "step": 31788 }, { "epoch": 79.4725, "grad_norm": 5.75, "learning_rate": 2.1298171736189776e-06, "loss": 1.4018, "step": 31789 }, { "epoch": 79.475, "grad_norm": 7.25, "learning_rate": 2.1293176783528867e-06, "loss": 1.4168, "step": 31790 }, { "epoch": 79.4775, "grad_norm": 8.4375, "learning_rate": 2.128818234686677e-06, "loss": 1.3755, "step": 31791 }, { "epoch": 79.48, "grad_norm": 8.125, "learning_rate": 2.128318842623618e-06, "loss": 1.4042, "step": 31792 }, { "epoch": 79.4825, "grad_norm": 6.28125, "learning_rate": 2.127819502166988e-06, "loss": 1.208, "step": 31793 }, { "epoch": 79.485, "grad_norm": 8.5625, "learning_rate": 2.1273202133200543e-06, "loss": 1.3769, "step": 31794 }, { "epoch": 79.4875, "grad_norm": 8.4375, "learning_rate": 2.126820976086098e-06, "loss": 1.3204, "step": 31795 }, { "epoch": 79.49, "grad_norm": 6.3125, "learning_rate": 2.1263217904683884e-06, "loss": 1.3686, "step": 31796 }, { "epoch": 79.4925, "grad_norm": 4.625, "learning_rate": 2.1258226564701957e-06, "loss": 1.1284, "step": 31797 }, { "epoch": 79.495, "grad_norm": 10.625, "learning_rate": 2.125323574094794e-06, "loss": 1.2473, "step": 31798 }, { "epoch": 79.4975, "grad_norm": 10.3125, "learning_rate": 2.1248245433454583e-06, "loss": 1.3766, "step": 31799 }, { "epoch": 79.5, "grad_norm": 10.625, "learning_rate": 2.124325564225458e-06, "loss": 1.3126, "step": 31800 }, { "epoch": 79.5025, "grad_norm": 6.125, "learning_rate": 2.1238266367380613e-06, "loss": 1.4694, "step": 31801 }, { "epoch": 79.505, "grad_norm": 7.625, "learning_rate": 2.123327760886542e-06, "loss": 1.193, "step": 31802 }, { "epoch": 79.5075, "grad_norm": 8.375, "learning_rate": 2.122828936674174e-06, "loss": 1.2461, "step": 31803 }, { "epoch": 79.51, "grad_norm": 7.09375, "learning_rate": 2.1223301641042237e-06, "loss": 1.1282, "step": 31804 }, { "epoch": 79.5125, "grad_norm": 9.875, "learning_rate": 2.1218314431799593e-06, "loss": 1.3805, "step": 31805 }, { "epoch": 79.515, "grad_norm": 8.125, "learning_rate": 2.121332773904655e-06, "loss": 1.3638, "step": 31806 }, { "epoch": 79.5175, "grad_norm": 9.5625, "learning_rate": 2.120834156281575e-06, "loss": 1.3968, "step": 31807 }, { "epoch": 79.52, "grad_norm": 9.125, "learning_rate": 2.1203355903139934e-06, "loss": 1.3695, "step": 31808 }, { "epoch": 79.5225, "grad_norm": 6.625, "learning_rate": 2.1198370760051744e-06, "loss": 1.4138, "step": 31809 }, { "epoch": 79.525, "grad_norm": 10.25, "learning_rate": 2.1193386133583906e-06, "loss": 1.2408, "step": 31810 }, { "epoch": 79.5275, "grad_norm": 6.59375, "learning_rate": 2.118840202376907e-06, "loss": 1.4276, "step": 31811 }, { "epoch": 79.53, "grad_norm": 5.53125, "learning_rate": 2.1183418430639903e-06, "loss": 1.4514, "step": 31812 }, { "epoch": 79.5325, "grad_norm": 7.96875, "learning_rate": 2.117843535422908e-06, "loss": 1.2195, "step": 31813 }, { "epoch": 79.535, "grad_norm": 6.9375, "learning_rate": 2.1173452794569303e-06, "loss": 1.3867, "step": 31814 }, { "epoch": 79.5375, "grad_norm": 8.6875, "learning_rate": 2.1168470751693227e-06, "loss": 1.3734, "step": 31815 }, { "epoch": 79.54, "grad_norm": 6.84375, "learning_rate": 2.116348922563346e-06, "loss": 1.3411, "step": 31816 }, { "epoch": 79.5425, "grad_norm": 14.6875, "learning_rate": 2.1158508216422714e-06, "loss": 1.4148, "step": 31817 }, { "epoch": 79.545, "grad_norm": 9.5625, "learning_rate": 2.115352772409366e-06, "loss": 1.538, "step": 31818 }, { "epoch": 79.5475, "grad_norm": 8.5, "learning_rate": 2.1148547748678915e-06, "loss": 1.4547, "step": 31819 }, { "epoch": 79.55, "grad_norm": 8.8125, "learning_rate": 2.1143568290211115e-06, "loss": 1.2454, "step": 31820 }, { "epoch": 79.5525, "grad_norm": 7.84375, "learning_rate": 2.113858934872296e-06, "loss": 1.3308, "step": 31821 }, { "epoch": 79.555, "grad_norm": 5.65625, "learning_rate": 2.1133610924247017e-06, "loss": 1.2899, "step": 31822 }, { "epoch": 79.5575, "grad_norm": 10.9375, "learning_rate": 2.1128633016816e-06, "loss": 1.4153, "step": 31823 }, { "epoch": 79.56, "grad_norm": 5.9375, "learning_rate": 2.112365562646248e-06, "loss": 1.1979, "step": 31824 }, { "epoch": 79.5625, "grad_norm": 8.9375, "learning_rate": 2.1118678753219137e-06, "loss": 1.3332, "step": 31825 }, { "epoch": 79.565, "grad_norm": 8.0, "learning_rate": 2.1113702397118583e-06, "loss": 1.3361, "step": 31826 }, { "epoch": 79.5675, "grad_norm": 9.3125, "learning_rate": 2.1108726558193413e-06, "loss": 1.337, "step": 31827 }, { "epoch": 79.57, "grad_norm": 8.1875, "learning_rate": 2.110375123647629e-06, "loss": 1.2069, "step": 31828 }, { "epoch": 79.5725, "grad_norm": 8.5625, "learning_rate": 2.109877643199982e-06, "loss": 1.2805, "step": 31829 }, { "epoch": 79.575, "grad_norm": 8.1875, "learning_rate": 2.1093802144796626e-06, "loss": 1.2844, "step": 31830 }, { "epoch": 79.5775, "grad_norm": 8.375, "learning_rate": 2.1088828374899308e-06, "loss": 1.2815, "step": 31831 }, { "epoch": 79.58, "grad_norm": 8.8125, "learning_rate": 2.108385512234041e-06, "loss": 1.2918, "step": 31832 }, { "epoch": 79.5825, "grad_norm": 4.8125, "learning_rate": 2.1078882387152665e-06, "loss": 1.3851, "step": 31833 }, { "epoch": 79.585, "grad_norm": 6.3125, "learning_rate": 2.1073910169368607e-06, "loss": 1.1976, "step": 31834 }, { "epoch": 79.5875, "grad_norm": 6.90625, "learning_rate": 2.106893846902083e-06, "loss": 1.3819, "step": 31835 }, { "epoch": 79.59, "grad_norm": 6.4375, "learning_rate": 2.106396728614192e-06, "loss": 1.3431, "step": 31836 }, { "epoch": 79.5925, "grad_norm": 8.375, "learning_rate": 2.105899662076446e-06, "loss": 1.4735, "step": 31837 }, { "epoch": 79.595, "grad_norm": 11.8125, "learning_rate": 2.1054026472921096e-06, "loss": 1.434, "step": 31838 }, { "epoch": 79.5975, "grad_norm": 6.1875, "learning_rate": 2.1049056842644355e-06, "loss": 1.1764, "step": 31839 }, { "epoch": 79.6, "grad_norm": 8.5, "learning_rate": 2.1044087729966856e-06, "loss": 1.2822, "step": 31840 }, { "epoch": 79.6025, "grad_norm": 8.8125, "learning_rate": 2.103911913492117e-06, "loss": 1.4389, "step": 31841 }, { "epoch": 79.605, "grad_norm": 7.03125, "learning_rate": 2.103415105753982e-06, "loss": 1.33, "step": 31842 }, { "epoch": 79.6075, "grad_norm": 8.5, "learning_rate": 2.102918349785542e-06, "loss": 1.3156, "step": 31843 }, { "epoch": 79.61, "grad_norm": 9.0, "learning_rate": 2.102421645590056e-06, "loss": 1.3969, "step": 31844 }, { "epoch": 79.6125, "grad_norm": 7.5, "learning_rate": 2.1019249931707776e-06, "loss": 1.2195, "step": 31845 }, { "epoch": 79.615, "grad_norm": 6.5625, "learning_rate": 2.101428392530963e-06, "loss": 1.332, "step": 31846 }, { "epoch": 79.6175, "grad_norm": 8.125, "learning_rate": 2.1009318436738622e-06, "loss": 1.5282, "step": 31847 }, { "epoch": 79.62, "grad_norm": 8.4375, "learning_rate": 2.100435346602742e-06, "loss": 1.2913, "step": 31848 }, { "epoch": 79.6225, "grad_norm": 4.9375, "learning_rate": 2.0999389013208506e-06, "loss": 1.1379, "step": 31849 }, { "epoch": 79.625, "grad_norm": 10.375, "learning_rate": 2.099442507831444e-06, "loss": 1.1953, "step": 31850 }, { "epoch": 79.6275, "grad_norm": 7.78125, "learning_rate": 2.098946166137774e-06, "loss": 1.212, "step": 31851 }, { "epoch": 79.63, "grad_norm": 7.5, "learning_rate": 2.098449876243096e-06, "loss": 1.3835, "step": 31852 }, { "epoch": 79.6325, "grad_norm": 5.1875, "learning_rate": 2.097953638150667e-06, "loss": 1.2388, "step": 31853 }, { "epoch": 79.635, "grad_norm": 9.6875, "learning_rate": 2.097457451863739e-06, "loss": 1.337, "step": 31854 }, { "epoch": 79.6375, "grad_norm": 10.125, "learning_rate": 2.09696131738556e-06, "loss": 1.5802, "step": 31855 }, { "epoch": 79.64, "grad_norm": 9.1875, "learning_rate": 2.0964652347193894e-06, "loss": 1.3854, "step": 31856 }, { "epoch": 79.6425, "grad_norm": 6.125, "learning_rate": 2.095969203868473e-06, "loss": 1.34, "step": 31857 }, { "epoch": 79.645, "grad_norm": 7.21875, "learning_rate": 2.0954732248360698e-06, "loss": 1.2166, "step": 31858 }, { "epoch": 79.6475, "grad_norm": 7.03125, "learning_rate": 2.0949772976254245e-06, "loss": 1.2986, "step": 31859 }, { "epoch": 79.65, "grad_norm": 9.0, "learning_rate": 2.0944814222397948e-06, "loss": 1.3026, "step": 31860 }, { "epoch": 79.6525, "grad_norm": 7.59375, "learning_rate": 2.093985598682429e-06, "loss": 1.3434, "step": 31861 }, { "epoch": 79.655, "grad_norm": 9.375, "learning_rate": 2.093489826956574e-06, "loss": 1.3737, "step": 31862 }, { "epoch": 79.6575, "grad_norm": 5.625, "learning_rate": 2.0929941070654824e-06, "loss": 1.3826, "step": 31863 }, { "epoch": 79.66, "grad_norm": 6.03125, "learning_rate": 2.0924984390124093e-06, "loss": 1.2986, "step": 31864 }, { "epoch": 79.6625, "grad_norm": 7.53125, "learning_rate": 2.0920028228005985e-06, "loss": 1.2314, "step": 31865 }, { "epoch": 79.665, "grad_norm": 7.21875, "learning_rate": 2.0915072584332998e-06, "loss": 1.3323, "step": 31866 }, { "epoch": 79.6675, "grad_norm": 7.53125, "learning_rate": 2.0910117459137615e-06, "loss": 1.4221, "step": 31867 }, { "epoch": 79.67, "grad_norm": 7.21875, "learning_rate": 2.0905162852452366e-06, "loss": 1.3801, "step": 31868 }, { "epoch": 79.6725, "grad_norm": 5.6875, "learning_rate": 2.0900208764309706e-06, "loss": 1.2889, "step": 31869 }, { "epoch": 79.675, "grad_norm": 7.125, "learning_rate": 2.0895255194742093e-06, "loss": 1.3151, "step": 31870 }, { "epoch": 79.6775, "grad_norm": 6.53125, "learning_rate": 2.0890302143782035e-06, "loss": 1.4741, "step": 31871 }, { "epoch": 79.68, "grad_norm": 6.40625, "learning_rate": 2.088534961146197e-06, "loss": 1.3313, "step": 31872 }, { "epoch": 79.6825, "grad_norm": 7.40625, "learning_rate": 2.0880397597814418e-06, "loss": 1.2687, "step": 31873 }, { "epoch": 79.685, "grad_norm": 5.96875, "learning_rate": 2.087544610287178e-06, "loss": 1.2481, "step": 31874 }, { "epoch": 79.6875, "grad_norm": 6.625, "learning_rate": 2.087049512666658e-06, "loss": 1.2199, "step": 31875 }, { "epoch": 79.69, "grad_norm": 8.6875, "learning_rate": 2.086554466923125e-06, "loss": 1.2415, "step": 31876 }, { "epoch": 79.6925, "grad_norm": 7.90625, "learning_rate": 2.0860594730598218e-06, "loss": 1.3431, "step": 31877 }, { "epoch": 79.695, "grad_norm": 6.34375, "learning_rate": 2.0855645310799944e-06, "loss": 1.2958, "step": 31878 }, { "epoch": 79.6975, "grad_norm": 7.34375, "learning_rate": 2.085069640986893e-06, "loss": 1.251, "step": 31879 }, { "epoch": 79.7, "grad_norm": 6.9375, "learning_rate": 2.0845748027837585e-06, "loss": 1.1341, "step": 31880 }, { "epoch": 79.7025, "grad_norm": 7.90625, "learning_rate": 2.084080016473832e-06, "loss": 1.3671, "step": 31881 }, { "epoch": 79.705, "grad_norm": 8.9375, "learning_rate": 2.0835852820603596e-06, "loss": 1.4165, "step": 31882 }, { "epoch": 79.7075, "grad_norm": 7.9375, "learning_rate": 2.083090599546589e-06, "loss": 1.3486, "step": 31883 }, { "epoch": 79.71, "grad_norm": 7.5625, "learning_rate": 2.0825959689357588e-06, "loss": 1.3355, "step": 31884 }, { "epoch": 79.7125, "grad_norm": 6.5625, "learning_rate": 2.0821013902311105e-06, "loss": 1.33, "step": 31885 }, { "epoch": 79.715, "grad_norm": 9.625, "learning_rate": 2.0816068634358875e-06, "loss": 1.3981, "step": 31886 }, { "epoch": 79.7175, "grad_norm": 7.53125, "learning_rate": 2.081112388553337e-06, "loss": 1.2624, "step": 31887 }, { "epoch": 79.72, "grad_norm": 8.0625, "learning_rate": 2.0806179655866964e-06, "loss": 1.295, "step": 31888 }, { "epoch": 79.7225, "grad_norm": 8.5625, "learning_rate": 2.080123594539205e-06, "loss": 1.3454, "step": 31889 }, { "epoch": 79.725, "grad_norm": 6.75, "learning_rate": 2.079629275414109e-06, "loss": 1.3426, "step": 31890 }, { "epoch": 79.7275, "grad_norm": 14.375, "learning_rate": 2.0791350082146433e-06, "loss": 1.2196, "step": 31891 }, { "epoch": 79.73, "grad_norm": 8.4375, "learning_rate": 2.0786407929440543e-06, "loss": 1.3146, "step": 31892 }, { "epoch": 79.7325, "grad_norm": 4.90625, "learning_rate": 2.078146629605576e-06, "loss": 1.1834, "step": 31893 }, { "epoch": 79.735, "grad_norm": 4.75, "learning_rate": 2.0776525182024552e-06, "loss": 1.1799, "step": 31894 }, { "epoch": 79.7375, "grad_norm": 7.1875, "learning_rate": 2.077158458737928e-06, "loss": 1.2315, "step": 31895 }, { "epoch": 79.74, "grad_norm": 9.0, "learning_rate": 2.076664451215229e-06, "loss": 1.3778, "step": 31896 }, { "epoch": 79.7425, "grad_norm": 8.125, "learning_rate": 2.0761704956376006e-06, "loss": 1.4268, "step": 31897 }, { "epoch": 79.745, "grad_norm": 8.625, "learning_rate": 2.0756765920082845e-06, "loss": 1.3241, "step": 31898 }, { "epoch": 79.7475, "grad_norm": 5.5, "learning_rate": 2.0751827403305158e-06, "loss": 1.1498, "step": 31899 }, { "epoch": 79.75, "grad_norm": 8.8125, "learning_rate": 2.074688940607529e-06, "loss": 1.2794, "step": 31900 }, { "epoch": 79.7525, "grad_norm": 10.375, "learning_rate": 2.074195192842564e-06, "loss": 1.2674, "step": 31901 }, { "epoch": 79.755, "grad_norm": 7.375, "learning_rate": 2.0737014970388614e-06, "loss": 1.4388, "step": 31902 }, { "epoch": 79.7575, "grad_norm": 6.78125, "learning_rate": 2.0732078531996545e-06, "loss": 1.3677, "step": 31903 }, { "epoch": 79.76, "grad_norm": 8.875, "learning_rate": 2.072714261328177e-06, "loss": 1.4499, "step": 31904 }, { "epoch": 79.7625, "grad_norm": 8.25, "learning_rate": 2.0722207214276702e-06, "loss": 1.3828, "step": 31905 }, { "epoch": 79.765, "grad_norm": 7.34375, "learning_rate": 2.071727233501364e-06, "loss": 1.4591, "step": 31906 }, { "epoch": 79.7675, "grad_norm": 8.25, "learning_rate": 2.0712337975525e-06, "loss": 1.339, "step": 31907 }, { "epoch": 79.77, "grad_norm": 7.71875, "learning_rate": 2.070740413584307e-06, "loss": 1.1821, "step": 31908 }, { "epoch": 79.7725, "grad_norm": 7.875, "learning_rate": 2.070247081600025e-06, "loss": 1.3613, "step": 31909 }, { "epoch": 79.775, "grad_norm": 8.0625, "learning_rate": 2.0697538016028864e-06, "loss": 1.1721, "step": 31910 }, { "epoch": 79.7775, "grad_norm": 5.75, "learning_rate": 2.0692605735961213e-06, "loss": 1.4265, "step": 31911 }, { "epoch": 79.78, "grad_norm": 8.4375, "learning_rate": 2.068767397582967e-06, "loss": 1.2937, "step": 31912 }, { "epoch": 79.7825, "grad_norm": 10.125, "learning_rate": 2.068274273566658e-06, "loss": 1.4065, "step": 31913 }, { "epoch": 79.785, "grad_norm": 7.5, "learning_rate": 2.0677812015504272e-06, "loss": 1.3287, "step": 31914 }, { "epoch": 79.7875, "grad_norm": 6.375, "learning_rate": 2.067288181537501e-06, "loss": 1.2779, "step": 31915 }, { "epoch": 79.79, "grad_norm": 7.40625, "learning_rate": 2.0667952135311178e-06, "loss": 1.2418, "step": 31916 }, { "epoch": 79.7925, "grad_norm": 6.1875, "learning_rate": 2.066302297534508e-06, "loss": 1.2904, "step": 31917 }, { "epoch": 79.795, "grad_norm": 6.1875, "learning_rate": 2.0658094335509047e-06, "loss": 1.4012, "step": 31918 }, { "epoch": 79.7975, "grad_norm": 6.78125, "learning_rate": 2.065316621583534e-06, "loss": 1.1846, "step": 31919 }, { "epoch": 79.8, "grad_norm": 6.5, "learning_rate": 2.064823861635633e-06, "loss": 1.3002, "step": 31920 }, { "epoch": 79.8025, "grad_norm": 10.3125, "learning_rate": 2.064331153710426e-06, "loss": 1.3097, "step": 31921 }, { "epoch": 79.805, "grad_norm": 7.5625, "learning_rate": 2.063838497811149e-06, "loss": 1.3851, "step": 31922 }, { "epoch": 79.8075, "grad_norm": 4.65625, "learning_rate": 2.0633458939410266e-06, "loss": 1.2141, "step": 31923 }, { "epoch": 79.81, "grad_norm": 11.75, "learning_rate": 2.0628533421032937e-06, "loss": 1.4024, "step": 31924 }, { "epoch": 79.8125, "grad_norm": 6.75, "learning_rate": 2.062360842301178e-06, "loss": 1.1211, "step": 31925 }, { "epoch": 79.815, "grad_norm": 6.21875, "learning_rate": 2.0618683945379024e-06, "loss": 1.3614, "step": 31926 }, { "epoch": 79.8175, "grad_norm": 8.0625, "learning_rate": 2.0613759988167005e-06, "loss": 1.4713, "step": 31927 }, { "epoch": 79.82, "grad_norm": 10.0, "learning_rate": 2.060883655140803e-06, "loss": 1.4463, "step": 31928 }, { "epoch": 79.8225, "grad_norm": 6.15625, "learning_rate": 2.0603913635134344e-06, "loss": 1.2626, "step": 31929 }, { "epoch": 79.825, "grad_norm": 7.71875, "learning_rate": 2.0598991239378186e-06, "loss": 1.241, "step": 31930 }, { "epoch": 79.8275, "grad_norm": 9.0625, "learning_rate": 2.0594069364171878e-06, "loss": 1.5462, "step": 31931 }, { "epoch": 79.83, "grad_norm": 7.25, "learning_rate": 2.058914800954769e-06, "loss": 1.2893, "step": 31932 }, { "epoch": 79.8325, "grad_norm": 7.96875, "learning_rate": 2.0584227175537874e-06, "loss": 1.4349, "step": 31933 }, { "epoch": 79.835, "grad_norm": 7.28125, "learning_rate": 2.0579306862174663e-06, "loss": 1.2951, "step": 31934 }, { "epoch": 79.8375, "grad_norm": 8.625, "learning_rate": 2.0574387069490365e-06, "loss": 1.38, "step": 31935 }, { "epoch": 79.84, "grad_norm": 7.53125, "learning_rate": 2.0569467797517173e-06, "loss": 1.1506, "step": 31936 }, { "epoch": 79.8425, "grad_norm": 6.71875, "learning_rate": 2.0564549046287407e-06, "loss": 1.3419, "step": 31937 }, { "epoch": 79.845, "grad_norm": 6.34375, "learning_rate": 2.055963081583324e-06, "loss": 1.2597, "step": 31938 }, { "epoch": 79.8475, "grad_norm": 8.5, "learning_rate": 2.0554713106186995e-06, "loss": 1.3656, "step": 31939 }, { "epoch": 79.85, "grad_norm": 9.5, "learning_rate": 2.0549795917380867e-06, "loss": 1.365, "step": 31940 }, { "epoch": 79.8525, "grad_norm": 5.84375, "learning_rate": 2.054487924944707e-06, "loss": 1.368, "step": 31941 }, { "epoch": 79.855, "grad_norm": 8.9375, "learning_rate": 2.053996310241786e-06, "loss": 1.2936, "step": 31942 }, { "epoch": 79.8575, "grad_norm": 6.15625, "learning_rate": 2.0535047476325497e-06, "loss": 1.2781, "step": 31943 }, { "epoch": 79.86, "grad_norm": 7.0625, "learning_rate": 2.0530132371202195e-06, "loss": 1.3006, "step": 31944 }, { "epoch": 79.8625, "grad_norm": 6.75, "learning_rate": 2.052521778708013e-06, "loss": 1.3998, "step": 31945 }, { "epoch": 79.865, "grad_norm": 10.125, "learning_rate": 2.0520303723991563e-06, "loss": 1.2566, "step": 31946 }, { "epoch": 79.8675, "grad_norm": 8.0, "learning_rate": 2.051539018196872e-06, "loss": 1.3367, "step": 31947 }, { "epoch": 79.87, "grad_norm": 9.5625, "learning_rate": 2.0510477161043805e-06, "loss": 1.3544, "step": 31948 }, { "epoch": 79.8725, "grad_norm": 7.5625, "learning_rate": 2.0505564661249e-06, "loss": 1.3111, "step": 31949 }, { "epoch": 79.875, "grad_norm": 6.4375, "learning_rate": 2.050065268261655e-06, "loss": 1.306, "step": 31950 }, { "epoch": 79.8775, "grad_norm": 8.0625, "learning_rate": 2.0495741225178613e-06, "loss": 1.2783, "step": 31951 }, { "epoch": 79.88, "grad_norm": 9.0625, "learning_rate": 2.0490830288967443e-06, "loss": 1.3315, "step": 31952 }, { "epoch": 79.8825, "grad_norm": 6.46875, "learning_rate": 2.048591987401518e-06, "loss": 1.2302, "step": 31953 }, { "epoch": 79.885, "grad_norm": 7.375, "learning_rate": 2.048100998035407e-06, "loss": 1.4431, "step": 31954 }, { "epoch": 79.8875, "grad_norm": 5.25, "learning_rate": 2.047610060801628e-06, "loss": 1.3536, "step": 31955 }, { "epoch": 79.89, "grad_norm": 7.09375, "learning_rate": 2.047119175703397e-06, "loss": 1.3444, "step": 31956 }, { "epoch": 79.8925, "grad_norm": 8.125, "learning_rate": 2.046628342743934e-06, "loss": 1.4142, "step": 31957 }, { "epoch": 79.895, "grad_norm": 5.75, "learning_rate": 2.04613756192646e-06, "loss": 1.3324, "step": 31958 }, { "epoch": 79.8975, "grad_norm": 6.4375, "learning_rate": 2.045646833254189e-06, "loss": 1.3355, "step": 31959 }, { "epoch": 79.9, "grad_norm": 7.5625, "learning_rate": 2.0451561567303378e-06, "loss": 1.3933, "step": 31960 }, { "epoch": 79.9025, "grad_norm": 11.625, "learning_rate": 2.0446655323581234e-06, "loss": 1.355, "step": 31961 }, { "epoch": 79.905, "grad_norm": 11.0625, "learning_rate": 2.0441749601407668e-06, "loss": 1.3089, "step": 31962 }, { "epoch": 79.9075, "grad_norm": 8.625, "learning_rate": 2.0436844400814802e-06, "loss": 1.4489, "step": 31963 }, { "epoch": 79.91, "grad_norm": 6.6875, "learning_rate": 2.0431939721834783e-06, "loss": 1.3067, "step": 31964 }, { "epoch": 79.9125, "grad_norm": 6.78125, "learning_rate": 2.0427035564499796e-06, "loss": 1.3789, "step": 31965 }, { "epoch": 79.915, "grad_norm": 5.875, "learning_rate": 2.0422131928841958e-06, "loss": 1.4115, "step": 31966 }, { "epoch": 79.9175, "grad_norm": 5.46875, "learning_rate": 2.0417228814893476e-06, "loss": 1.2386, "step": 31967 }, { "epoch": 79.92, "grad_norm": 8.125, "learning_rate": 2.041232622268642e-06, "loss": 1.2637, "step": 31968 }, { "epoch": 79.9225, "grad_norm": 7.90625, "learning_rate": 2.0407424152252998e-06, "loss": 1.3344, "step": 31969 }, { "epoch": 79.925, "grad_norm": 9.8125, "learning_rate": 2.0402522603625308e-06, "loss": 1.3303, "step": 31970 }, { "epoch": 79.9275, "grad_norm": 7.4375, "learning_rate": 2.039762157683548e-06, "loss": 1.3416, "step": 31971 }, { "epoch": 79.93, "grad_norm": 5.75, "learning_rate": 2.0392721071915656e-06, "loss": 1.5162, "step": 31972 }, { "epoch": 79.9325, "grad_norm": 8.25, "learning_rate": 2.0387821088898e-06, "loss": 1.3743, "step": 31973 }, { "epoch": 79.935, "grad_norm": 5.625, "learning_rate": 2.038292162781459e-06, "loss": 1.0987, "step": 31974 }, { "epoch": 79.9375, "grad_norm": 5.78125, "learning_rate": 2.0378022688697563e-06, "loss": 1.2087, "step": 31975 }, { "epoch": 79.94, "grad_norm": 7.9375, "learning_rate": 2.037312427157898e-06, "loss": 1.2414, "step": 31976 }, { "epoch": 79.9425, "grad_norm": 6.125, "learning_rate": 2.036822637649107e-06, "loss": 1.4918, "step": 31977 }, { "epoch": 79.945, "grad_norm": 5.90625, "learning_rate": 2.036332900346587e-06, "loss": 1.2497, "step": 31978 }, { "epoch": 79.9475, "grad_norm": 8.6875, "learning_rate": 2.0358432152535477e-06, "loss": 1.2111, "step": 31979 }, { "epoch": 79.95, "grad_norm": 7.28125, "learning_rate": 2.0353535823732053e-06, "loss": 1.3465, "step": 31980 }, { "epoch": 79.9525, "grad_norm": 10.0625, "learning_rate": 2.0348640017087615e-06, "loss": 1.4267, "step": 31981 }, { "epoch": 79.955, "grad_norm": 5.90625, "learning_rate": 2.034374473263434e-06, "loss": 1.3624, "step": 31982 }, { "epoch": 79.9575, "grad_norm": 7.09375, "learning_rate": 2.033884997040426e-06, "loss": 1.3542, "step": 31983 }, { "epoch": 79.96, "grad_norm": 6.71875, "learning_rate": 2.033395573042952e-06, "loss": 1.246, "step": 31984 }, { "epoch": 79.9625, "grad_norm": 7.375, "learning_rate": 2.0329062012742184e-06, "loss": 1.5027, "step": 31985 }, { "epoch": 79.965, "grad_norm": 14.75, "learning_rate": 2.032416881737429e-06, "loss": 1.4032, "step": 31986 }, { "epoch": 79.9675, "grad_norm": 6.375, "learning_rate": 2.0319276144357967e-06, "loss": 1.329, "step": 31987 }, { "epoch": 79.97, "grad_norm": 8.3125, "learning_rate": 2.031438399372531e-06, "loss": 1.4368, "step": 31988 }, { "epoch": 79.9725, "grad_norm": 8.4375, "learning_rate": 2.030949236550834e-06, "loss": 1.1539, "step": 31989 }, { "epoch": 79.975, "grad_norm": 9.8125, "learning_rate": 2.0304601259739176e-06, "loss": 1.3849, "step": 31990 }, { "epoch": 79.9775, "grad_norm": 6.4375, "learning_rate": 2.029971067644979e-06, "loss": 1.2495, "step": 31991 }, { "epoch": 79.98, "grad_norm": 5.96875, "learning_rate": 2.029482061567237e-06, "loss": 1.1795, "step": 31992 }, { "epoch": 79.9825, "grad_norm": 9.375, "learning_rate": 2.0289931077438907e-06, "loss": 1.3975, "step": 31993 }, { "epoch": 79.985, "grad_norm": 8.5, "learning_rate": 2.028504206178148e-06, "loss": 1.3709, "step": 31994 }, { "epoch": 79.9875, "grad_norm": 7.78125, "learning_rate": 2.0280153568732085e-06, "loss": 1.3523, "step": 31995 }, { "epoch": 79.99, "grad_norm": 6.21875, "learning_rate": 2.0275265598322822e-06, "loss": 1.3131, "step": 31996 }, { "epoch": 79.9925, "grad_norm": 9.3125, "learning_rate": 2.0270378150585756e-06, "loss": 1.3377, "step": 31997 }, { "epoch": 79.995, "grad_norm": 5.875, "learning_rate": 2.0265491225552903e-06, "loss": 1.352, "step": 31998 }, { "epoch": 79.9975, "grad_norm": 7.6875, "learning_rate": 2.026060482325627e-06, "loss": 1.4511, "step": 31999 }, { "epoch": 80.0, "grad_norm": 4.875, "learning_rate": 2.025571894372794e-06, "loss": 1.3682, "step": 32000 }, { "epoch": 80.0025, "grad_norm": 8.125, "learning_rate": 2.0250833586999896e-06, "loss": 1.3661, "step": 32001 }, { "epoch": 80.005, "grad_norm": 9.75, "learning_rate": 2.024594875310424e-06, "loss": 1.3682, "step": 32002 }, { "epoch": 80.0075, "grad_norm": 6.75, "learning_rate": 2.024106444207291e-06, "loss": 1.3276, "step": 32003 }, { "epoch": 80.01, "grad_norm": 6.9375, "learning_rate": 2.0236180653937998e-06, "loss": 1.1808, "step": 32004 }, { "epoch": 80.0125, "grad_norm": 7.375, "learning_rate": 2.0231297388731498e-06, "loss": 1.5061, "step": 32005 }, { "epoch": 80.015, "grad_norm": 6.59375, "learning_rate": 2.022641464648537e-06, "loss": 1.2211, "step": 32006 }, { "epoch": 80.0175, "grad_norm": 6.90625, "learning_rate": 2.022153242723173e-06, "loss": 1.4318, "step": 32007 }, { "epoch": 80.02, "grad_norm": 5.03125, "learning_rate": 2.0216650731002517e-06, "loss": 1.4017, "step": 32008 }, { "epoch": 80.0225, "grad_norm": 8.625, "learning_rate": 2.0211769557829754e-06, "loss": 1.2475, "step": 32009 }, { "epoch": 80.025, "grad_norm": 9.75, "learning_rate": 2.0206888907745415e-06, "loss": 1.4134, "step": 32010 }, { "epoch": 80.0275, "grad_norm": 8.375, "learning_rate": 2.0202008780781525e-06, "loss": 1.1686, "step": 32011 }, { "epoch": 80.03, "grad_norm": 10.125, "learning_rate": 2.0197129176970087e-06, "loss": 1.2144, "step": 32012 }, { "epoch": 80.0325, "grad_norm": 7.78125, "learning_rate": 2.019225009634308e-06, "loss": 1.4633, "step": 32013 }, { "epoch": 80.035, "grad_norm": 7.65625, "learning_rate": 2.0187371538932465e-06, "loss": 1.3309, "step": 32014 }, { "epoch": 80.0375, "grad_norm": 9.4375, "learning_rate": 2.0182493504770274e-06, "loss": 1.3871, "step": 32015 }, { "epoch": 80.04, "grad_norm": 6.53125, "learning_rate": 2.017761599388842e-06, "loss": 1.3411, "step": 32016 }, { "epoch": 80.0425, "grad_norm": 7.6875, "learning_rate": 2.0172739006318962e-06, "loss": 1.3239, "step": 32017 }, { "epoch": 80.045, "grad_norm": 5.25, "learning_rate": 2.016786254209381e-06, "loss": 1.378, "step": 32018 }, { "epoch": 80.0475, "grad_norm": 7.3125, "learning_rate": 2.0162986601244985e-06, "loss": 1.266, "step": 32019 }, { "epoch": 80.05, "grad_norm": 8.8125, "learning_rate": 2.0158111183804407e-06, "loss": 1.4643, "step": 32020 }, { "epoch": 80.0525, "grad_norm": 6.96875, "learning_rate": 2.0153236289804044e-06, "loss": 1.291, "step": 32021 }, { "epoch": 80.055, "grad_norm": 8.0, "learning_rate": 2.014836191927586e-06, "loss": 1.2765, "step": 32022 }, { "epoch": 80.0575, "grad_norm": 6.4375, "learning_rate": 2.014348807225186e-06, "loss": 1.1012, "step": 32023 }, { "epoch": 80.06, "grad_norm": 8.25, "learning_rate": 2.013861474876395e-06, "loss": 1.1663, "step": 32024 }, { "epoch": 80.0625, "grad_norm": 6.0625, "learning_rate": 2.0133741948844056e-06, "loss": 1.3714, "step": 32025 }, { "epoch": 80.065, "grad_norm": 7.34375, "learning_rate": 2.0128869672524155e-06, "loss": 1.2226, "step": 32026 }, { "epoch": 80.0675, "grad_norm": 5.8125, "learning_rate": 2.0123997919836214e-06, "loss": 1.2115, "step": 32027 }, { "epoch": 80.07, "grad_norm": 7.1875, "learning_rate": 2.011912669081215e-06, "loss": 1.3578, "step": 32028 }, { "epoch": 80.0725, "grad_norm": 4.84375, "learning_rate": 2.0114255985483867e-06, "loss": 1.4278, "step": 32029 }, { "epoch": 80.075, "grad_norm": 9.125, "learning_rate": 2.010938580388334e-06, "loss": 1.3936, "step": 32030 }, { "epoch": 80.0775, "grad_norm": 6.625, "learning_rate": 2.0104516146042476e-06, "loss": 1.2463, "step": 32031 }, { "epoch": 80.08, "grad_norm": 8.8125, "learning_rate": 2.0099647011993217e-06, "loss": 1.382, "step": 32032 }, { "epoch": 80.0825, "grad_norm": 8.6875, "learning_rate": 2.0094778401767458e-06, "loss": 1.3063, "step": 32033 }, { "epoch": 80.085, "grad_norm": 6.96875, "learning_rate": 2.0089910315397153e-06, "loss": 1.2929, "step": 32034 }, { "epoch": 80.0875, "grad_norm": 6.46875, "learning_rate": 2.0085042752914207e-06, "loss": 1.2331, "step": 32035 }, { "epoch": 80.09, "grad_norm": 6.875, "learning_rate": 2.0080175714350492e-06, "loss": 1.213, "step": 32036 }, { "epoch": 80.0925, "grad_norm": 9.75, "learning_rate": 2.0075309199737947e-06, "loss": 1.4224, "step": 32037 }, { "epoch": 80.095, "grad_norm": 6.875, "learning_rate": 2.0070443209108504e-06, "loss": 1.1365, "step": 32038 }, { "epoch": 80.0975, "grad_norm": 8.0, "learning_rate": 2.0065577742494037e-06, "loss": 1.3574, "step": 32039 }, { "epoch": 80.1, "grad_norm": 8.3125, "learning_rate": 2.0060712799926407e-06, "loss": 1.3695, "step": 32040 }, { "epoch": 80.1025, "grad_norm": 5.90625, "learning_rate": 2.005584838143756e-06, "loss": 1.2982, "step": 32041 }, { "epoch": 80.105, "grad_norm": 5.9375, "learning_rate": 2.005098448705939e-06, "loss": 1.2878, "step": 32042 }, { "epoch": 80.1075, "grad_norm": 9.625, "learning_rate": 2.0046121116823768e-06, "loss": 1.4445, "step": 32043 }, { "epoch": 80.11, "grad_norm": 11.875, "learning_rate": 2.004125827076255e-06, "loss": 1.2216, "step": 32044 }, { "epoch": 80.1125, "grad_norm": 8.5, "learning_rate": 2.0036395948907674e-06, "loss": 1.3504, "step": 32045 }, { "epoch": 80.115, "grad_norm": 7.9375, "learning_rate": 2.0031534151290944e-06, "loss": 1.3746, "step": 32046 }, { "epoch": 80.1175, "grad_norm": 5.65625, "learning_rate": 2.002667287794432e-06, "loss": 1.374, "step": 32047 }, { "epoch": 80.12, "grad_norm": 6.46875, "learning_rate": 2.00218121288996e-06, "loss": 1.2645, "step": 32048 }, { "epoch": 80.1225, "grad_norm": 9.1875, "learning_rate": 2.0016951904188696e-06, "loss": 1.3439, "step": 32049 }, { "epoch": 80.125, "grad_norm": 7.0, "learning_rate": 2.001209220384346e-06, "loss": 1.306, "step": 32050 }, { "epoch": 80.1275, "grad_norm": 9.0, "learning_rate": 2.0007233027895713e-06, "loss": 1.3719, "step": 32051 }, { "epoch": 80.13, "grad_norm": 7.21875, "learning_rate": 2.0002374376377354e-06, "loss": 1.1956, "step": 32052 }, { "epoch": 80.1325, "grad_norm": 7.34375, "learning_rate": 1.999751624932025e-06, "loss": 1.3233, "step": 32053 }, { "epoch": 80.135, "grad_norm": 8.4375, "learning_rate": 1.9992658646756235e-06, "loss": 1.3244, "step": 32054 }, { "epoch": 80.1375, "grad_norm": 8.375, "learning_rate": 1.9987801568717115e-06, "loss": 1.3448, "step": 32055 }, { "epoch": 80.14, "grad_norm": 8.0, "learning_rate": 1.998294501523477e-06, "loss": 1.3111, "step": 32056 }, { "epoch": 80.1425, "grad_norm": 7.125, "learning_rate": 1.997808898634105e-06, "loss": 1.4146, "step": 32057 }, { "epoch": 80.145, "grad_norm": 6.40625, "learning_rate": 1.997323348206779e-06, "loss": 1.3372, "step": 32058 }, { "epoch": 80.1475, "grad_norm": 5.1875, "learning_rate": 1.9968378502446773e-06, "loss": 1.2805, "step": 32059 }, { "epoch": 80.15, "grad_norm": 7.28125, "learning_rate": 1.9963524047509898e-06, "loss": 1.459, "step": 32060 }, { "epoch": 80.1525, "grad_norm": 8.5, "learning_rate": 1.9958670117288927e-06, "loss": 1.3514, "step": 32061 }, { "epoch": 80.155, "grad_norm": 5.625, "learning_rate": 1.9953816711815743e-06, "loss": 1.4416, "step": 32062 }, { "epoch": 80.1575, "grad_norm": 7.34375, "learning_rate": 1.9948963831122093e-06, "loss": 1.368, "step": 32063 }, { "epoch": 80.16, "grad_norm": 7.96875, "learning_rate": 1.994411147523987e-06, "loss": 1.4072, "step": 32064 }, { "epoch": 80.1625, "grad_norm": 6.65625, "learning_rate": 1.993925964420085e-06, "loss": 1.2564, "step": 32065 }, { "epoch": 80.165, "grad_norm": 7.90625, "learning_rate": 1.99344083380368e-06, "loss": 1.4394, "step": 32066 }, { "epoch": 80.1675, "grad_norm": 9.25, "learning_rate": 1.9929557556779587e-06, "loss": 1.4632, "step": 32067 }, { "epoch": 80.17, "grad_norm": 6.5625, "learning_rate": 1.9924707300461e-06, "loss": 1.4263, "step": 32068 }, { "epoch": 80.1725, "grad_norm": 7.65625, "learning_rate": 1.991985756911283e-06, "loss": 1.4355, "step": 32069 }, { "epoch": 80.175, "grad_norm": 6.9375, "learning_rate": 1.9915008362766854e-06, "loss": 1.3534, "step": 32070 }, { "epoch": 80.1775, "grad_norm": 5.875, "learning_rate": 1.9910159681454868e-06, "loss": 1.2143, "step": 32071 }, { "epoch": 80.18, "grad_norm": 12.1875, "learning_rate": 1.990531152520869e-06, "loss": 1.1962, "step": 32072 }, { "epoch": 80.1825, "grad_norm": 7.59375, "learning_rate": 1.9900463894060096e-06, "loss": 1.4033, "step": 32073 }, { "epoch": 80.185, "grad_norm": 6.90625, "learning_rate": 1.989561678804083e-06, "loss": 1.3321, "step": 32074 }, { "epoch": 80.1875, "grad_norm": 8.0, "learning_rate": 1.9890770207182706e-06, "loss": 1.2236, "step": 32075 }, { "epoch": 80.19, "grad_norm": 6.4375, "learning_rate": 1.9885924151517477e-06, "loss": 1.5354, "step": 32076 }, { "epoch": 80.1925, "grad_norm": 10.6875, "learning_rate": 1.9881078621076943e-06, "loss": 1.3748, "step": 32077 }, { "epoch": 80.195, "grad_norm": 8.0625, "learning_rate": 1.9876233615892827e-06, "loss": 1.3722, "step": 32078 }, { "epoch": 80.1975, "grad_norm": 7.34375, "learning_rate": 1.9871389135996943e-06, "loss": 1.3264, "step": 32079 }, { "epoch": 80.2, "grad_norm": 6.9375, "learning_rate": 1.9866545181421016e-06, "loss": 1.1308, "step": 32080 }, { "epoch": 80.2025, "grad_norm": 11.8125, "learning_rate": 1.9861701752196793e-06, "loss": 1.3718, "step": 32081 }, { "epoch": 80.205, "grad_norm": 8.8125, "learning_rate": 1.985685884835605e-06, "loss": 1.2692, "step": 32082 }, { "epoch": 80.2075, "grad_norm": 6.03125, "learning_rate": 1.985201646993056e-06, "loss": 1.3771, "step": 32083 }, { "epoch": 80.21, "grad_norm": 8.8125, "learning_rate": 1.9847174616952035e-06, "loss": 1.4735, "step": 32084 }, { "epoch": 80.2125, "grad_norm": 7.96875, "learning_rate": 1.9842333289452208e-06, "loss": 1.313, "step": 32085 }, { "epoch": 80.215, "grad_norm": 7.65625, "learning_rate": 1.9837492487462827e-06, "loss": 1.4209, "step": 32086 }, { "epoch": 80.2175, "grad_norm": 6.125, "learning_rate": 1.983265221101567e-06, "loss": 1.3081, "step": 32087 }, { "epoch": 80.22, "grad_norm": 10.0625, "learning_rate": 1.9827812460142435e-06, "loss": 1.2611, "step": 32088 }, { "epoch": 80.2225, "grad_norm": 8.4375, "learning_rate": 1.9822973234874833e-06, "loss": 1.2211, "step": 32089 }, { "epoch": 80.225, "grad_norm": 10.4375, "learning_rate": 1.9818134535244627e-06, "loss": 1.4185, "step": 32090 }, { "epoch": 80.2275, "grad_norm": 6.34375, "learning_rate": 1.9813296361283496e-06, "loss": 1.3588, "step": 32091 }, { "epoch": 80.23, "grad_norm": 8.75, "learning_rate": 1.9808458713023215e-06, "loss": 1.5046, "step": 32092 }, { "epoch": 80.2325, "grad_norm": 7.0625, "learning_rate": 1.980362159049545e-06, "loss": 1.2443, "step": 32093 }, { "epoch": 80.235, "grad_norm": 10.25, "learning_rate": 1.9798784993731946e-06, "loss": 1.302, "step": 32094 }, { "epoch": 80.2375, "grad_norm": 7.59375, "learning_rate": 1.9793948922764404e-06, "loss": 1.2679, "step": 32095 }, { "epoch": 80.24, "grad_norm": 8.3125, "learning_rate": 1.97891133776245e-06, "loss": 1.2526, "step": 32096 }, { "epoch": 80.2425, "grad_norm": 12.5, "learning_rate": 1.9784278358343955e-06, "loss": 1.3789, "step": 32097 }, { "epoch": 80.245, "grad_norm": 10.5625, "learning_rate": 1.9779443864954508e-06, "loss": 1.4187, "step": 32098 }, { "epoch": 80.2475, "grad_norm": 12.3125, "learning_rate": 1.9774609897487796e-06, "loss": 1.3346, "step": 32099 }, { "epoch": 80.25, "grad_norm": 6.875, "learning_rate": 1.976977645597552e-06, "loss": 1.2227, "step": 32100 }, { "epoch": 80.2525, "grad_norm": 6.625, "learning_rate": 1.9764943540449377e-06, "loss": 1.2956, "step": 32101 }, { "epoch": 80.255, "grad_norm": 5.9375, "learning_rate": 1.9760111150941075e-06, "loss": 1.3169, "step": 32102 }, { "epoch": 80.2575, "grad_norm": 7.46875, "learning_rate": 1.9755279287482275e-06, "loss": 1.3481, "step": 32103 }, { "epoch": 80.26, "grad_norm": 8.6875, "learning_rate": 1.9750447950104635e-06, "loss": 1.2938, "step": 32104 }, { "epoch": 80.2625, "grad_norm": 9.5, "learning_rate": 1.9745617138839856e-06, "loss": 1.3997, "step": 32105 }, { "epoch": 80.265, "grad_norm": 7.71875, "learning_rate": 1.9740786853719584e-06, "loss": 1.2748, "step": 32106 }, { "epoch": 80.2675, "grad_norm": 7.9375, "learning_rate": 1.9735957094775526e-06, "loss": 1.2632, "step": 32107 }, { "epoch": 80.27, "grad_norm": 7.3125, "learning_rate": 1.97311278620393e-06, "loss": 1.4087, "step": 32108 }, { "epoch": 80.2725, "grad_norm": 9.75, "learning_rate": 1.9726299155542604e-06, "loss": 1.2572, "step": 32109 }, { "epoch": 80.275, "grad_norm": 4.59375, "learning_rate": 1.972147097531709e-06, "loss": 1.3309, "step": 32110 }, { "epoch": 80.2775, "grad_norm": 7.65625, "learning_rate": 1.9716643321394368e-06, "loss": 1.2702, "step": 32111 }, { "epoch": 80.28, "grad_norm": 6.875, "learning_rate": 1.971181619380611e-06, "loss": 1.4127, "step": 32112 }, { "epoch": 80.2825, "grad_norm": 10.875, "learning_rate": 1.9706989592584016e-06, "loss": 1.363, "step": 32113 }, { "epoch": 80.285, "grad_norm": 12.1875, "learning_rate": 1.9702163517759676e-06, "loss": 1.3103, "step": 32114 }, { "epoch": 80.2875, "grad_norm": 6.9375, "learning_rate": 1.969733796936474e-06, "loss": 1.4734, "step": 32115 }, { "epoch": 80.29, "grad_norm": 8.0625, "learning_rate": 1.9692512947430788e-06, "loss": 1.3072, "step": 32116 }, { "epoch": 80.2925, "grad_norm": 7.96875, "learning_rate": 1.9687688451989562e-06, "loss": 1.4773, "step": 32117 }, { "epoch": 80.295, "grad_norm": 6.8125, "learning_rate": 1.9682864483072627e-06, "loss": 1.2548, "step": 32118 }, { "epoch": 80.2975, "grad_norm": 8.375, "learning_rate": 1.9678041040711606e-06, "loss": 1.1856, "step": 32119 }, { "epoch": 80.3, "grad_norm": 6.53125, "learning_rate": 1.967321812493813e-06, "loss": 1.2316, "step": 32120 }, { "epoch": 80.3025, "grad_norm": 8.6875, "learning_rate": 1.966839573578384e-06, "loss": 1.3941, "step": 32121 }, { "epoch": 80.305, "grad_norm": 13.8125, "learning_rate": 1.966357387328034e-06, "loss": 1.3138, "step": 32122 }, { "epoch": 80.3075, "grad_norm": 6.9375, "learning_rate": 1.9658752537459214e-06, "loss": 1.3458, "step": 32123 }, { "epoch": 80.31, "grad_norm": 7.09375, "learning_rate": 1.9653931728352106e-06, "loss": 1.3492, "step": 32124 }, { "epoch": 80.3125, "grad_norm": 6.84375, "learning_rate": 1.9649111445990588e-06, "loss": 1.2481, "step": 32125 }, { "epoch": 80.315, "grad_norm": 7.125, "learning_rate": 1.9644291690406305e-06, "loss": 1.2233, "step": 32126 }, { "epoch": 80.3175, "grad_norm": 7.25, "learning_rate": 1.9639472461630793e-06, "loss": 1.2794, "step": 32127 }, { "epoch": 80.32, "grad_norm": 9.25, "learning_rate": 1.963465375969572e-06, "loss": 1.3159, "step": 32128 }, { "epoch": 80.3225, "grad_norm": 8.6875, "learning_rate": 1.962983558463264e-06, "loss": 1.3546, "step": 32129 }, { "epoch": 80.325, "grad_norm": 9.5625, "learning_rate": 1.962501793647312e-06, "loss": 1.3504, "step": 32130 }, { "epoch": 80.3275, "grad_norm": 6.46875, "learning_rate": 1.9620200815248757e-06, "loss": 1.2934, "step": 32131 }, { "epoch": 80.33, "grad_norm": 6.0625, "learning_rate": 1.9615384220991173e-06, "loss": 1.3393, "step": 32132 }, { "epoch": 80.3325, "grad_norm": 6.6875, "learning_rate": 1.961056815373191e-06, "loss": 1.2396, "step": 32133 }, { "epoch": 80.335, "grad_norm": 8.3125, "learning_rate": 1.9605752613502548e-06, "loss": 1.3571, "step": 32134 }, { "epoch": 80.3375, "grad_norm": 8.0625, "learning_rate": 1.9600937600334602e-06, "loss": 1.2781, "step": 32135 }, { "epoch": 80.34, "grad_norm": 10.1875, "learning_rate": 1.9596123114259747e-06, "loss": 1.3299, "step": 32136 }, { "epoch": 80.3425, "grad_norm": 7.1875, "learning_rate": 1.959130915530948e-06, "loss": 1.2807, "step": 32137 }, { "epoch": 80.345, "grad_norm": 9.125, "learning_rate": 1.9586495723515386e-06, "loss": 1.4032, "step": 32138 }, { "epoch": 80.3475, "grad_norm": 9.0625, "learning_rate": 1.9581682818908976e-06, "loss": 1.2721, "step": 32139 }, { "epoch": 80.35, "grad_norm": 6.9375, "learning_rate": 1.9576870441521834e-06, "loss": 1.5047, "step": 32140 }, { "epoch": 80.3525, "grad_norm": 7.46875, "learning_rate": 1.9572058591385533e-06, "loss": 1.4198, "step": 32141 }, { "epoch": 80.355, "grad_norm": 8.1875, "learning_rate": 1.9567247268531608e-06, "loss": 1.434, "step": 32142 }, { "epoch": 80.3575, "grad_norm": 6.375, "learning_rate": 1.956243647299155e-06, "loss": 1.2934, "step": 32143 }, { "epoch": 80.36, "grad_norm": 8.625, "learning_rate": 1.955762620479699e-06, "loss": 1.5311, "step": 32144 }, { "epoch": 80.3625, "grad_norm": 6.65625, "learning_rate": 1.9552816463979364e-06, "loss": 1.2341, "step": 32145 }, { "epoch": 80.365, "grad_norm": 8.5, "learning_rate": 1.954800725057027e-06, "loss": 1.2289, "step": 32146 }, { "epoch": 80.3675, "grad_norm": 6.15625, "learning_rate": 1.954319856460123e-06, "loss": 1.3324, "step": 32147 }, { "epoch": 80.37, "grad_norm": 8.0, "learning_rate": 1.953839040610377e-06, "loss": 1.4258, "step": 32148 }, { "epoch": 80.3725, "grad_norm": 7.65625, "learning_rate": 1.9533582775109404e-06, "loss": 1.3818, "step": 32149 }, { "epoch": 80.375, "grad_norm": 11.1875, "learning_rate": 1.9528775671649593e-06, "loss": 1.2033, "step": 32150 }, { "epoch": 80.3775, "grad_norm": 5.625, "learning_rate": 1.952396909575597e-06, "loss": 1.3135, "step": 32151 }, { "epoch": 80.38, "grad_norm": 8.8125, "learning_rate": 1.9519163047459978e-06, "loss": 1.1967, "step": 32152 }, { "epoch": 80.3825, "grad_norm": 8.125, "learning_rate": 1.9514357526793125e-06, "loss": 1.3221, "step": 32153 }, { "epoch": 80.385, "grad_norm": 8.6875, "learning_rate": 1.9509552533786913e-06, "loss": 1.3174, "step": 32154 }, { "epoch": 80.3875, "grad_norm": 10.4375, "learning_rate": 1.9504748068472846e-06, "loss": 1.2051, "step": 32155 }, { "epoch": 80.39, "grad_norm": 5.6875, "learning_rate": 1.9499944130882455e-06, "loss": 1.2367, "step": 32156 }, { "epoch": 80.3925, "grad_norm": 7.9375, "learning_rate": 1.94951407210472e-06, "loss": 1.2631, "step": 32157 }, { "epoch": 80.395, "grad_norm": 12.375, "learning_rate": 1.9490337838998575e-06, "loss": 1.4006, "step": 32158 }, { "epoch": 80.3975, "grad_norm": 10.875, "learning_rate": 1.9485535484768083e-06, "loss": 1.2274, "step": 32159 }, { "epoch": 80.4, "grad_norm": 7.5, "learning_rate": 1.9480733658387175e-06, "loss": 1.2617, "step": 32160 }, { "epoch": 80.4025, "grad_norm": 8.375, "learning_rate": 1.9475932359887385e-06, "loss": 1.2268, "step": 32161 }, { "epoch": 80.405, "grad_norm": 8.0625, "learning_rate": 1.947113158930013e-06, "loss": 1.3792, "step": 32162 }, { "epoch": 80.4075, "grad_norm": 9.1875, "learning_rate": 1.946633134665694e-06, "loss": 1.3961, "step": 32163 }, { "epoch": 80.41, "grad_norm": 9.875, "learning_rate": 1.9461531631989262e-06, "loss": 1.4394, "step": 32164 }, { "epoch": 80.4125, "grad_norm": 10.3125, "learning_rate": 1.9456732445328542e-06, "loss": 1.3464, "step": 32165 }, { "epoch": 80.415, "grad_norm": 11.25, "learning_rate": 1.9451933786706247e-06, "loss": 1.2794, "step": 32166 }, { "epoch": 80.4175, "grad_norm": 6.46875, "learning_rate": 1.9447135656153882e-06, "loss": 1.3821, "step": 32167 }, { "epoch": 80.42, "grad_norm": 8.0, "learning_rate": 1.944233805370287e-06, "loss": 1.276, "step": 32168 }, { "epoch": 80.4225, "grad_norm": 6.9375, "learning_rate": 1.943754097938464e-06, "loss": 1.1978, "step": 32169 }, { "epoch": 80.425, "grad_norm": 9.1875, "learning_rate": 1.9432744433230667e-06, "loss": 1.3668, "step": 32170 }, { "epoch": 80.4275, "grad_norm": 6.53125, "learning_rate": 1.9427948415272425e-06, "loss": 1.3324, "step": 32171 }, { "epoch": 80.43, "grad_norm": 8.5, "learning_rate": 1.942315292554133e-06, "loss": 1.4433, "step": 32172 }, { "epoch": 80.4325, "grad_norm": 6.75, "learning_rate": 1.941835796406879e-06, "loss": 1.2157, "step": 32173 }, { "epoch": 80.435, "grad_norm": 6.03125, "learning_rate": 1.9413563530886303e-06, "loss": 1.3563, "step": 32174 }, { "epoch": 80.4375, "grad_norm": 6.375, "learning_rate": 1.9408769626025237e-06, "loss": 1.2165, "step": 32175 }, { "epoch": 80.44, "grad_norm": 11.3125, "learning_rate": 1.940397624951709e-06, "loss": 1.4605, "step": 32176 }, { "epoch": 80.4425, "grad_norm": 8.375, "learning_rate": 1.9399183401393207e-06, "loss": 1.2927, "step": 32177 }, { "epoch": 80.445, "grad_norm": 9.3125, "learning_rate": 1.9394391081685083e-06, "loss": 1.2117, "step": 32178 }, { "epoch": 80.4475, "grad_norm": 7.65625, "learning_rate": 1.938959929042411e-06, "loss": 1.264, "step": 32179 }, { "epoch": 80.45, "grad_norm": 7.375, "learning_rate": 1.9384808027641666e-06, "loss": 1.3999, "step": 32180 }, { "epoch": 80.4525, "grad_norm": 8.875, "learning_rate": 1.93800172933692e-06, "loss": 1.3293, "step": 32181 }, { "epoch": 80.455, "grad_norm": 7.25, "learning_rate": 1.937522708763814e-06, "loss": 1.4718, "step": 32182 }, { "epoch": 80.4575, "grad_norm": 8.25, "learning_rate": 1.9370437410479857e-06, "loss": 1.418, "step": 32183 }, { "epoch": 80.46, "grad_norm": 4.1875, "learning_rate": 1.936564826192574e-06, "loss": 1.2032, "step": 32184 }, { "epoch": 80.4625, "grad_norm": 9.875, "learning_rate": 1.9360859642007213e-06, "loss": 1.4288, "step": 32185 }, { "epoch": 80.465, "grad_norm": 9.0, "learning_rate": 1.9356071550755684e-06, "loss": 1.2913, "step": 32186 }, { "epoch": 80.4675, "grad_norm": 7.90625, "learning_rate": 1.9351283988202517e-06, "loss": 1.2498, "step": 32187 }, { "epoch": 80.47, "grad_norm": 9.125, "learning_rate": 1.934649695437908e-06, "loss": 1.2782, "step": 32188 }, { "epoch": 80.4725, "grad_norm": 8.5625, "learning_rate": 1.934171044931681e-06, "loss": 1.3808, "step": 32189 }, { "epoch": 80.475, "grad_norm": 8.5, "learning_rate": 1.9336924473047036e-06, "loss": 1.5299, "step": 32190 }, { "epoch": 80.4775, "grad_norm": 7.3125, "learning_rate": 1.933213902560118e-06, "loss": 1.2327, "step": 32191 }, { "epoch": 80.48, "grad_norm": 6.90625, "learning_rate": 1.9327354107010566e-06, "loss": 1.4438, "step": 32192 }, { "epoch": 80.4825, "grad_norm": 6.9375, "learning_rate": 1.9322569717306616e-06, "loss": 1.3577, "step": 32193 }, { "epoch": 80.485, "grad_norm": 9.9375, "learning_rate": 1.9317785856520676e-06, "loss": 1.114, "step": 32194 }, { "epoch": 80.4875, "grad_norm": 8.1875, "learning_rate": 1.9313002524684068e-06, "loss": 1.3588, "step": 32195 }, { "epoch": 80.49, "grad_norm": 10.3125, "learning_rate": 1.9308219721828192e-06, "loss": 1.423, "step": 32196 }, { "epoch": 80.4925, "grad_norm": 6.5625, "learning_rate": 1.930343744798442e-06, "loss": 1.3148, "step": 32197 }, { "epoch": 80.495, "grad_norm": 7.21875, "learning_rate": 1.9298655703184088e-06, "loss": 1.3279, "step": 32198 }, { "epoch": 80.4975, "grad_norm": 7.46875, "learning_rate": 1.929387448745851e-06, "loss": 1.4001, "step": 32199 }, { "epoch": 80.5, "grad_norm": 6.84375, "learning_rate": 1.9289093800839067e-06, "loss": 1.238, "step": 32200 }, { "epoch": 80.5025, "grad_norm": 7.1875, "learning_rate": 1.9284313643357112e-06, "loss": 1.3143, "step": 32201 }, { "epoch": 80.505, "grad_norm": 8.9375, "learning_rate": 1.927953401504398e-06, "loss": 1.358, "step": 32202 }, { "epoch": 80.5075, "grad_norm": 9.125, "learning_rate": 1.927475491593095e-06, "loss": 1.3497, "step": 32203 }, { "epoch": 80.51, "grad_norm": 6.0, "learning_rate": 1.9269976346049433e-06, "loss": 1.1683, "step": 32204 }, { "epoch": 80.5125, "grad_norm": 5.8125, "learning_rate": 1.9265198305430687e-06, "loss": 1.2055, "step": 32205 }, { "epoch": 80.515, "grad_norm": 6.03125, "learning_rate": 1.9260420794106104e-06, "loss": 1.4896, "step": 32206 }, { "epoch": 80.5175, "grad_norm": 4.21875, "learning_rate": 1.925564381210695e-06, "loss": 1.32, "step": 32207 }, { "epoch": 80.52, "grad_norm": 7.875, "learning_rate": 1.9250867359464575e-06, "loss": 1.2495, "step": 32208 }, { "epoch": 80.5225, "grad_norm": 6.4375, "learning_rate": 1.924609143621029e-06, "loss": 1.3352, "step": 32209 }, { "epoch": 80.525, "grad_norm": 6.96875, "learning_rate": 1.924131604237537e-06, "loss": 1.1955, "step": 32210 }, { "epoch": 80.5275, "grad_norm": 7.90625, "learning_rate": 1.9236541177991152e-06, "loss": 1.3311, "step": 32211 }, { "epoch": 80.53, "grad_norm": 9.0, "learning_rate": 1.9231766843088974e-06, "loss": 1.3924, "step": 32212 }, { "epoch": 80.5325, "grad_norm": 7.1875, "learning_rate": 1.922699303770008e-06, "loss": 1.4335, "step": 32213 }, { "epoch": 80.535, "grad_norm": 6.0, "learning_rate": 1.922221976185578e-06, "loss": 1.2589, "step": 32214 }, { "epoch": 80.5375, "grad_norm": 7.84375, "learning_rate": 1.9217447015587364e-06, "loss": 1.16, "step": 32215 }, { "epoch": 80.54, "grad_norm": 6.03125, "learning_rate": 1.9212674798926156e-06, "loss": 1.3655, "step": 32216 }, { "epoch": 80.5425, "grad_norm": 8.0625, "learning_rate": 1.9207903111903427e-06, "loss": 1.3035, "step": 32217 }, { "epoch": 80.545, "grad_norm": 9.1875, "learning_rate": 1.9203131954550424e-06, "loss": 1.3125, "step": 32218 }, { "epoch": 80.5475, "grad_norm": 6.96875, "learning_rate": 1.9198361326898474e-06, "loss": 1.5737, "step": 32219 }, { "epoch": 80.55, "grad_norm": 10.9375, "learning_rate": 1.9193591228978815e-06, "loss": 1.2146, "step": 32220 }, { "epoch": 80.5525, "grad_norm": 7.78125, "learning_rate": 1.9188821660822764e-06, "loss": 1.4897, "step": 32221 }, { "epoch": 80.555, "grad_norm": 6.0, "learning_rate": 1.9184052622461537e-06, "loss": 1.349, "step": 32222 }, { "epoch": 80.5575, "grad_norm": 5.5, "learning_rate": 1.9179284113926456e-06, "loss": 1.3017, "step": 32223 }, { "epoch": 80.56, "grad_norm": 9.3125, "learning_rate": 1.9174516135248745e-06, "loss": 1.4086, "step": 32224 }, { "epoch": 80.5625, "grad_norm": 9.8125, "learning_rate": 1.9169748686459655e-06, "loss": 1.3879, "step": 32225 }, { "epoch": 80.565, "grad_norm": 10.25, "learning_rate": 1.9164981767590452e-06, "loss": 1.3198, "step": 32226 }, { "epoch": 80.5675, "grad_norm": 9.0625, "learning_rate": 1.916021537867243e-06, "loss": 1.2567, "step": 32227 }, { "epoch": 80.57, "grad_norm": 5.96875, "learning_rate": 1.915544951973679e-06, "loss": 1.3333, "step": 32228 }, { "epoch": 80.5725, "grad_norm": 4.96875, "learning_rate": 1.915068419081476e-06, "loss": 1.317, "step": 32229 }, { "epoch": 80.575, "grad_norm": 7.78125, "learning_rate": 1.9145919391937617e-06, "loss": 1.5708, "step": 32230 }, { "epoch": 80.5775, "grad_norm": 7.34375, "learning_rate": 1.9141155123136613e-06, "loss": 1.3099, "step": 32231 }, { "epoch": 80.58, "grad_norm": 7.0, "learning_rate": 1.9136391384442964e-06, "loss": 1.2154, "step": 32232 }, { "epoch": 80.5825, "grad_norm": 8.125, "learning_rate": 1.9131628175887873e-06, "loss": 1.259, "step": 32233 }, { "epoch": 80.585, "grad_norm": 6.40625, "learning_rate": 1.9126865497502613e-06, "loss": 1.2936, "step": 32234 }, { "epoch": 80.5875, "grad_norm": 9.625, "learning_rate": 1.9122103349318354e-06, "loss": 1.333, "step": 32235 }, { "epoch": 80.59, "grad_norm": 9.0, "learning_rate": 1.911734173136638e-06, "loss": 1.1982, "step": 32236 }, { "epoch": 80.5925, "grad_norm": 6.0625, "learning_rate": 1.9112580643677857e-06, "loss": 1.1567, "step": 32237 }, { "epoch": 80.595, "grad_norm": 8.375, "learning_rate": 1.9107820086284035e-06, "loss": 1.2273, "step": 32238 }, { "epoch": 80.5975, "grad_norm": 7.8125, "learning_rate": 1.9103060059216115e-06, "loss": 1.3657, "step": 32239 }, { "epoch": 80.6, "grad_norm": 6.875, "learning_rate": 1.9098300562505266e-06, "loss": 1.3228, "step": 32240 }, { "epoch": 80.6025, "grad_norm": 7.0625, "learning_rate": 1.909354159618272e-06, "loss": 1.2941, "step": 32241 }, { "epoch": 80.605, "grad_norm": 14.125, "learning_rate": 1.9088783160279713e-06, "loss": 1.5311, "step": 32242 }, { "epoch": 80.6075, "grad_norm": 8.0, "learning_rate": 1.908402525482739e-06, "loss": 1.3904, "step": 32243 }, { "epoch": 80.61, "grad_norm": 8.1875, "learning_rate": 1.9079267879856946e-06, "loss": 1.2759, "step": 32244 }, { "epoch": 80.6125, "grad_norm": 7.875, "learning_rate": 1.907451103539958e-06, "loss": 1.276, "step": 32245 }, { "epoch": 80.615, "grad_norm": 7.15625, "learning_rate": 1.90697547214865e-06, "loss": 1.3468, "step": 32246 }, { "epoch": 80.6175, "grad_norm": 6.0625, "learning_rate": 1.9064998938148883e-06, "loss": 1.1935, "step": 32247 }, { "epoch": 80.62, "grad_norm": 8.4375, "learning_rate": 1.9060243685417857e-06, "loss": 1.4085, "step": 32248 }, { "epoch": 80.6225, "grad_norm": 10.875, "learning_rate": 1.9055488963324665e-06, "loss": 1.3393, "step": 32249 }, { "epoch": 80.625, "grad_norm": 6.96875, "learning_rate": 1.9050734771900414e-06, "loss": 1.315, "step": 32250 }, { "epoch": 80.6275, "grad_norm": 9.5625, "learning_rate": 1.9045981111176348e-06, "loss": 1.4066, "step": 32251 }, { "epoch": 80.63, "grad_norm": 7.90625, "learning_rate": 1.9041227981183553e-06, "loss": 1.3807, "step": 32252 }, { "epoch": 80.6325, "grad_norm": 8.0, "learning_rate": 1.9036475381953257e-06, "loss": 1.3864, "step": 32253 }, { "epoch": 80.635, "grad_norm": 5.5, "learning_rate": 1.9031723313516582e-06, "loss": 1.1746, "step": 32254 }, { "epoch": 80.6375, "grad_norm": 4.5625, "learning_rate": 1.9026971775904667e-06, "loss": 1.1523, "step": 32255 }, { "epoch": 80.64, "grad_norm": 9.9375, "learning_rate": 1.902222076914869e-06, "loss": 1.4744, "step": 32256 }, { "epoch": 80.6425, "grad_norm": 12.375, "learning_rate": 1.9017470293279817e-06, "loss": 1.2868, "step": 32257 }, { "epoch": 80.645, "grad_norm": 7.53125, "learning_rate": 1.9012720348329162e-06, "loss": 1.3352, "step": 32258 }, { "epoch": 80.6475, "grad_norm": 7.0625, "learning_rate": 1.9007970934327858e-06, "loss": 1.2763, "step": 32259 }, { "epoch": 80.65, "grad_norm": 5.8125, "learning_rate": 1.9003222051307046e-06, "loss": 1.4309, "step": 32260 }, { "epoch": 80.6525, "grad_norm": 11.25, "learning_rate": 1.8998473699297893e-06, "loss": 1.2456, "step": 32261 }, { "epoch": 80.655, "grad_norm": 6.53125, "learning_rate": 1.8993725878331505e-06, "loss": 1.1339, "step": 32262 }, { "epoch": 80.6575, "grad_norm": 9.0, "learning_rate": 1.898897858843899e-06, "loss": 1.2712, "step": 32263 }, { "epoch": 80.66, "grad_norm": 9.25, "learning_rate": 1.8984231829651512e-06, "loss": 1.2787, "step": 32264 }, { "epoch": 80.6625, "grad_norm": 6.71875, "learning_rate": 1.8979485602000135e-06, "loss": 1.2532, "step": 32265 }, { "epoch": 80.665, "grad_norm": 5.78125, "learning_rate": 1.897473990551605e-06, "loss": 1.0751, "step": 32266 }, { "epoch": 80.6675, "grad_norm": 6.15625, "learning_rate": 1.8969994740230291e-06, "loss": 1.2706, "step": 32267 }, { "epoch": 80.67, "grad_norm": 5.96875, "learning_rate": 1.8965250106174026e-06, "loss": 1.2376, "step": 32268 }, { "epoch": 80.6725, "grad_norm": 6.78125, "learning_rate": 1.8960506003378343e-06, "loss": 1.2215, "step": 32269 }, { "epoch": 80.675, "grad_norm": 7.53125, "learning_rate": 1.8955762431874315e-06, "loss": 1.3609, "step": 32270 }, { "epoch": 80.6775, "grad_norm": 6.0625, "learning_rate": 1.895101939169306e-06, "loss": 1.4066, "step": 32271 }, { "epoch": 80.68, "grad_norm": 6.28125, "learning_rate": 1.894627688286571e-06, "loss": 1.3483, "step": 32272 }, { "epoch": 80.6825, "grad_norm": 6.53125, "learning_rate": 1.894153490542333e-06, "loss": 1.4058, "step": 32273 }, { "epoch": 80.685, "grad_norm": 6.5, "learning_rate": 1.8936793459396995e-06, "loss": 1.3164, "step": 32274 }, { "epoch": 80.6875, "grad_norm": 6.75, "learning_rate": 1.8932052544817747e-06, "loss": 1.1803, "step": 32275 }, { "epoch": 80.69, "grad_norm": 9.0, "learning_rate": 1.8927312161716771e-06, "loss": 1.3447, "step": 32276 }, { "epoch": 80.6925, "grad_norm": 9.125, "learning_rate": 1.8922572310125099e-06, "loss": 1.2775, "step": 32277 }, { "epoch": 80.695, "grad_norm": 7.625, "learning_rate": 1.8917832990073782e-06, "loss": 1.2922, "step": 32278 }, { "epoch": 80.6975, "grad_norm": 7.125, "learning_rate": 1.8913094201593897e-06, "loss": 1.2647, "step": 32279 }, { "epoch": 80.7, "grad_norm": 7.40625, "learning_rate": 1.8908355944716516e-06, "loss": 1.3415, "step": 32280 }, { "epoch": 80.7025, "grad_norm": 10.4375, "learning_rate": 1.890361821947273e-06, "loss": 1.2699, "step": 32281 }, { "epoch": 80.705, "grad_norm": 7.78125, "learning_rate": 1.8898881025893557e-06, "loss": 1.3514, "step": 32282 }, { "epoch": 80.7075, "grad_norm": 8.4375, "learning_rate": 1.8894144364010103e-06, "loss": 1.2555, "step": 32283 }, { "epoch": 80.71, "grad_norm": 6.03125, "learning_rate": 1.888940823385339e-06, "loss": 1.2463, "step": 32284 }, { "epoch": 80.7125, "grad_norm": 6.59375, "learning_rate": 1.8884672635454438e-06, "loss": 1.3277, "step": 32285 }, { "epoch": 80.715, "grad_norm": 6.5625, "learning_rate": 1.887993756884433e-06, "loss": 1.2342, "step": 32286 }, { "epoch": 80.7175, "grad_norm": 7.5625, "learning_rate": 1.8875203034054135e-06, "loss": 1.3276, "step": 32287 }, { "epoch": 80.72, "grad_norm": 8.0, "learning_rate": 1.8870469031114868e-06, "loss": 1.4187, "step": 32288 }, { "epoch": 80.7225, "grad_norm": 7.5625, "learning_rate": 1.8865735560057552e-06, "loss": 1.3744, "step": 32289 }, { "epoch": 80.725, "grad_norm": 5.34375, "learning_rate": 1.886100262091317e-06, "loss": 1.3372, "step": 32290 }, { "epoch": 80.7275, "grad_norm": 10.6875, "learning_rate": 1.8856270213712868e-06, "loss": 1.3088, "step": 32291 }, { "epoch": 80.73, "grad_norm": 8.1875, "learning_rate": 1.8851538338487608e-06, "loss": 1.1374, "step": 32292 }, { "epoch": 80.7325, "grad_norm": 10.4375, "learning_rate": 1.884680699526842e-06, "loss": 1.2259, "step": 32293 }, { "epoch": 80.735, "grad_norm": 5.59375, "learning_rate": 1.8842076184086277e-06, "loss": 1.3489, "step": 32294 }, { "epoch": 80.7375, "grad_norm": 20.0, "learning_rate": 1.8837345904972248e-06, "loss": 1.4367, "step": 32295 }, { "epoch": 80.74, "grad_norm": 9.6875, "learning_rate": 1.8832616157957352e-06, "loss": 1.3428, "step": 32296 }, { "epoch": 80.7425, "grad_norm": 11.0, "learning_rate": 1.882788694307258e-06, "loss": 1.1896, "step": 32297 }, { "epoch": 80.745, "grad_norm": 9.5625, "learning_rate": 1.8823158260348905e-06, "loss": 1.4664, "step": 32298 }, { "epoch": 80.7475, "grad_norm": 8.4375, "learning_rate": 1.8818430109817388e-06, "loss": 1.3712, "step": 32299 }, { "epoch": 80.75, "grad_norm": 6.40625, "learning_rate": 1.8813702491508956e-06, "loss": 1.3434, "step": 32300 }, { "epoch": 80.7525, "grad_norm": 16.5, "learning_rate": 1.8808975405454677e-06, "loss": 1.2003, "step": 32301 }, { "epoch": 80.755, "grad_norm": 8.875, "learning_rate": 1.8804248851685481e-06, "loss": 1.4112, "step": 32302 }, { "epoch": 80.7575, "grad_norm": 6.25, "learning_rate": 1.8799522830232396e-06, "loss": 1.3235, "step": 32303 }, { "epoch": 80.76, "grad_norm": 8.875, "learning_rate": 1.8794797341126403e-06, "loss": 1.3755, "step": 32304 }, { "epoch": 80.7625, "grad_norm": 7.0625, "learning_rate": 1.8790072384398438e-06, "loss": 1.2238, "step": 32305 }, { "epoch": 80.765, "grad_norm": 6.59375, "learning_rate": 1.8785347960079504e-06, "loss": 1.3689, "step": 32306 }, { "epoch": 80.7675, "grad_norm": 6.5, "learning_rate": 1.8780624068200603e-06, "loss": 1.2054, "step": 32307 }, { "epoch": 80.77, "grad_norm": 7.84375, "learning_rate": 1.8775900708792693e-06, "loss": 1.139, "step": 32308 }, { "epoch": 80.7725, "grad_norm": 8.5625, "learning_rate": 1.8771177881886693e-06, "loss": 1.3173, "step": 32309 }, { "epoch": 80.775, "grad_norm": 6.40625, "learning_rate": 1.87664555875136e-06, "loss": 1.2781, "step": 32310 }, { "epoch": 80.7775, "grad_norm": 8.6875, "learning_rate": 1.8761733825704398e-06, "loss": 1.363, "step": 32311 }, { "epoch": 80.78, "grad_norm": 7.1875, "learning_rate": 1.875701259649002e-06, "loss": 1.2514, "step": 32312 }, { "epoch": 80.7825, "grad_norm": 7.28125, "learning_rate": 1.8752291899901398e-06, "loss": 1.2212, "step": 32313 }, { "epoch": 80.785, "grad_norm": 8.625, "learning_rate": 1.8747571735969516e-06, "loss": 1.5267, "step": 32314 }, { "epoch": 80.7875, "grad_norm": 8.4375, "learning_rate": 1.8742852104725274e-06, "loss": 1.3739, "step": 32315 }, { "epoch": 80.79, "grad_norm": 7.875, "learning_rate": 1.8738133006199677e-06, "loss": 1.317, "step": 32316 }, { "epoch": 80.7925, "grad_norm": 6.3125, "learning_rate": 1.8733414440423592e-06, "loss": 1.3413, "step": 32317 }, { "epoch": 80.795, "grad_norm": 7.0, "learning_rate": 1.8728696407428027e-06, "loss": 1.1488, "step": 32318 }, { "epoch": 80.7975, "grad_norm": 5.40625, "learning_rate": 1.8723978907243866e-06, "loss": 1.2946, "step": 32319 }, { "epoch": 80.8, "grad_norm": 7.8125, "learning_rate": 1.8719261939902023e-06, "loss": 1.3156, "step": 32320 }, { "epoch": 80.8025, "grad_norm": 4.25, "learning_rate": 1.8714545505433446e-06, "loss": 1.2752, "step": 32321 }, { "epoch": 80.805, "grad_norm": 5.75, "learning_rate": 1.870982960386909e-06, "loss": 1.4182, "step": 32322 }, { "epoch": 80.8075, "grad_norm": 6.34375, "learning_rate": 1.870511423523983e-06, "loss": 1.3994, "step": 32323 }, { "epoch": 80.81, "grad_norm": 7.4375, "learning_rate": 1.8700399399576563e-06, "loss": 1.3372, "step": 32324 }, { "epoch": 80.8125, "grad_norm": 5.28125, "learning_rate": 1.869568509691022e-06, "loss": 1.3044, "step": 32325 }, { "epoch": 80.815, "grad_norm": 6.375, "learning_rate": 1.8690971327271746e-06, "loss": 1.3882, "step": 32326 }, { "epoch": 80.8175, "grad_norm": 4.96875, "learning_rate": 1.868625809069201e-06, "loss": 1.3179, "step": 32327 }, { "epoch": 80.82, "grad_norm": 7.875, "learning_rate": 1.8681545387201883e-06, "loss": 1.3381, "step": 32328 }, { "epoch": 80.8225, "grad_norm": 8.3125, "learning_rate": 1.8676833216832313e-06, "loss": 1.2914, "step": 32329 }, { "epoch": 80.825, "grad_norm": 8.875, "learning_rate": 1.8672121579614144e-06, "loss": 1.2259, "step": 32330 }, { "epoch": 80.8275, "grad_norm": 6.65625, "learning_rate": 1.8667410475578317e-06, "loss": 1.2153, "step": 32331 }, { "epoch": 80.83, "grad_norm": 6.0625, "learning_rate": 1.8662699904755677e-06, "loss": 1.4452, "step": 32332 }, { "epoch": 80.8325, "grad_norm": 8.0625, "learning_rate": 1.865798986717714e-06, "loss": 1.3287, "step": 32333 }, { "epoch": 80.835, "grad_norm": 7.90625, "learning_rate": 1.865328036287356e-06, "loss": 1.2952, "step": 32334 }, { "epoch": 80.8375, "grad_norm": 6.625, "learning_rate": 1.8648571391875803e-06, "loss": 1.3778, "step": 32335 }, { "epoch": 80.84, "grad_norm": 9.1875, "learning_rate": 1.8643862954214754e-06, "loss": 1.3291, "step": 32336 }, { "epoch": 80.8425, "grad_norm": 6.5625, "learning_rate": 1.863915504992131e-06, "loss": 1.2715, "step": 32337 }, { "epoch": 80.845, "grad_norm": 8.1875, "learning_rate": 1.8634447679026302e-06, "loss": 1.2556, "step": 32338 }, { "epoch": 80.8475, "grad_norm": 7.5, "learning_rate": 1.8629740841560584e-06, "loss": 1.3351, "step": 32339 }, { "epoch": 80.85, "grad_norm": 6.0, "learning_rate": 1.862503453755502e-06, "loss": 1.2522, "step": 32340 }, { "epoch": 80.8525, "grad_norm": 11.25, "learning_rate": 1.8620328767040508e-06, "loss": 1.2584, "step": 32341 }, { "epoch": 80.855, "grad_norm": 10.5625, "learning_rate": 1.8615623530047855e-06, "loss": 1.4136, "step": 32342 }, { "epoch": 80.8575, "grad_norm": 8.375, "learning_rate": 1.8610918826607893e-06, "loss": 1.5109, "step": 32343 }, { "epoch": 80.86, "grad_norm": 7.875, "learning_rate": 1.860621465675152e-06, "loss": 1.3965, "step": 32344 }, { "epoch": 80.8625, "grad_norm": 8.375, "learning_rate": 1.8601511020509523e-06, "loss": 1.262, "step": 32345 }, { "epoch": 80.865, "grad_norm": 6.25, "learning_rate": 1.8596807917912785e-06, "loss": 1.1126, "step": 32346 }, { "epoch": 80.8675, "grad_norm": 8.4375, "learning_rate": 1.8592105348992085e-06, "loss": 1.2277, "step": 32347 }, { "epoch": 80.87, "grad_norm": 7.625, "learning_rate": 1.858740331377832e-06, "loss": 1.2594, "step": 32348 }, { "epoch": 80.8725, "grad_norm": 7.0, "learning_rate": 1.8582701812302274e-06, "loss": 1.3276, "step": 32349 }, { "epoch": 80.875, "grad_norm": 7.40625, "learning_rate": 1.8578000844594746e-06, "loss": 1.4041, "step": 32350 }, { "epoch": 80.8775, "grad_norm": 6.1875, "learning_rate": 1.85733004106866e-06, "loss": 1.2944, "step": 32351 }, { "epoch": 80.88, "grad_norm": 5.96875, "learning_rate": 1.8568600510608659e-06, "loss": 1.1967, "step": 32352 }, { "epoch": 80.8825, "grad_norm": 7.34375, "learning_rate": 1.8563901144391705e-06, "loss": 1.2195, "step": 32353 }, { "epoch": 80.885, "grad_norm": 7.78125, "learning_rate": 1.8559202312066537e-06, "loss": 1.3705, "step": 32354 }, { "epoch": 80.8875, "grad_norm": 6.21875, "learning_rate": 1.8554504013663987e-06, "loss": 1.2539, "step": 32355 }, { "epoch": 80.89, "grad_norm": 6.9375, "learning_rate": 1.8549806249214863e-06, "loss": 1.6295, "step": 32356 }, { "epoch": 80.8925, "grad_norm": 5.25, "learning_rate": 1.8545109018749963e-06, "loss": 1.2464, "step": 32357 }, { "epoch": 80.895, "grad_norm": 7.40625, "learning_rate": 1.8540412322300027e-06, "loss": 1.1881, "step": 32358 }, { "epoch": 80.8975, "grad_norm": 17.625, "learning_rate": 1.8535716159895922e-06, "loss": 1.4247, "step": 32359 }, { "epoch": 80.9, "grad_norm": 11.6875, "learning_rate": 1.8531020531568377e-06, "loss": 1.4963, "step": 32360 }, { "epoch": 80.9025, "grad_norm": 6.875, "learning_rate": 1.8526325437348225e-06, "loss": 1.3116, "step": 32361 }, { "epoch": 80.905, "grad_norm": 9.375, "learning_rate": 1.8521630877266205e-06, "loss": 1.2117, "step": 32362 }, { "epoch": 80.9075, "grad_norm": 6.40625, "learning_rate": 1.851693685135313e-06, "loss": 1.3308, "step": 32363 }, { "epoch": 80.91, "grad_norm": 9.5625, "learning_rate": 1.8512243359639727e-06, "loss": 1.3834, "step": 32364 }, { "epoch": 80.9125, "grad_norm": 6.90625, "learning_rate": 1.8507550402156826e-06, "loss": 1.247, "step": 32365 }, { "epoch": 80.915, "grad_norm": 10.4375, "learning_rate": 1.850285797893514e-06, "loss": 1.2446, "step": 32366 }, { "epoch": 80.9175, "grad_norm": 8.25, "learning_rate": 1.8498166090005477e-06, "loss": 1.2928, "step": 32367 }, { "epoch": 80.92, "grad_norm": 7.0625, "learning_rate": 1.8493474735398575e-06, "loss": 1.4698, "step": 32368 }, { "epoch": 80.9225, "grad_norm": 6.21875, "learning_rate": 1.8488783915145169e-06, "loss": 1.3964, "step": 32369 }, { "epoch": 80.925, "grad_norm": 9.25, "learning_rate": 1.8484093629276034e-06, "loss": 1.3001, "step": 32370 }, { "epoch": 80.9275, "grad_norm": 5.6875, "learning_rate": 1.8479403877821945e-06, "loss": 1.2971, "step": 32371 }, { "epoch": 80.93, "grad_norm": 8.75, "learning_rate": 1.8474714660813619e-06, "loss": 1.2398, "step": 32372 }, { "epoch": 80.9325, "grad_norm": 9.3125, "learning_rate": 1.847002597828178e-06, "loss": 1.3723, "step": 32373 }, { "epoch": 80.935, "grad_norm": 7.0625, "learning_rate": 1.8465337830257192e-06, "loss": 1.3491, "step": 32374 }, { "epoch": 80.9375, "grad_norm": 9.6875, "learning_rate": 1.8460650216770604e-06, "loss": 1.3424, "step": 32375 }, { "epoch": 80.94, "grad_norm": 8.8125, "learning_rate": 1.8455963137852728e-06, "loss": 1.4301, "step": 32376 }, { "epoch": 80.9425, "grad_norm": 8.3125, "learning_rate": 1.8451276593534272e-06, "loss": 1.2182, "step": 32377 }, { "epoch": 80.945, "grad_norm": 6.53125, "learning_rate": 1.8446590583846013e-06, "loss": 1.3489, "step": 32378 }, { "epoch": 80.9475, "grad_norm": 8.0, "learning_rate": 1.8441905108818603e-06, "loss": 1.4203, "step": 32379 }, { "epoch": 80.95, "grad_norm": 5.875, "learning_rate": 1.8437220168482839e-06, "loss": 1.3528, "step": 32380 }, { "epoch": 80.9525, "grad_norm": 10.4375, "learning_rate": 1.8432535762869364e-06, "loss": 1.3402, "step": 32381 }, { "epoch": 80.955, "grad_norm": 7.96875, "learning_rate": 1.8427851892008952e-06, "loss": 1.2998, "step": 32382 }, { "epoch": 80.9575, "grad_norm": 5.4375, "learning_rate": 1.842316855593227e-06, "loss": 1.3467, "step": 32383 }, { "epoch": 80.96, "grad_norm": 8.75, "learning_rate": 1.8418485754670013e-06, "loss": 1.4035, "step": 32384 }, { "epoch": 80.9625, "grad_norm": 6.9375, "learning_rate": 1.8413803488252891e-06, "loss": 1.2975, "step": 32385 }, { "epoch": 80.965, "grad_norm": 14.4375, "learning_rate": 1.8409121756711646e-06, "loss": 1.3501, "step": 32386 }, { "epoch": 80.9675, "grad_norm": 8.1875, "learning_rate": 1.840444056007692e-06, "loss": 1.441, "step": 32387 }, { "epoch": 80.97, "grad_norm": 6.65625, "learning_rate": 1.8399759898379398e-06, "loss": 1.2959, "step": 32388 }, { "epoch": 80.9725, "grad_norm": 12.3125, "learning_rate": 1.8395079771649783e-06, "loss": 1.2153, "step": 32389 }, { "epoch": 80.975, "grad_norm": 6.75, "learning_rate": 1.8390400179918789e-06, "loss": 1.2346, "step": 32390 }, { "epoch": 80.9775, "grad_norm": 12.25, "learning_rate": 1.838572112321706e-06, "loss": 1.4644, "step": 32391 }, { "epoch": 80.98, "grad_norm": 8.625, "learning_rate": 1.838104260157525e-06, "loss": 1.2839, "step": 32392 }, { "epoch": 80.9825, "grad_norm": 8.4375, "learning_rate": 1.837636461502409e-06, "loss": 1.487, "step": 32393 }, { "epoch": 80.985, "grad_norm": 12.0, "learning_rate": 1.8371687163594187e-06, "loss": 1.3681, "step": 32394 }, { "epoch": 80.9875, "grad_norm": 7.5, "learning_rate": 1.8367010247316264e-06, "loss": 1.3043, "step": 32395 }, { "epoch": 80.99, "grad_norm": 5.6875, "learning_rate": 1.836233386622094e-06, "loss": 1.2985, "step": 32396 }, { "epoch": 80.9925, "grad_norm": 8.1875, "learning_rate": 1.8357658020338897e-06, "loss": 1.3975, "step": 32397 }, { "epoch": 80.995, "grad_norm": 7.6875, "learning_rate": 1.8352982709700784e-06, "loss": 1.2957, "step": 32398 }, { "epoch": 80.9975, "grad_norm": 8.875, "learning_rate": 1.8348307934337228e-06, "loss": 1.0722, "step": 32399 }, { "epoch": 81.0, "grad_norm": 9.0, "learning_rate": 1.8343633694278895e-06, "loss": 1.2672, "step": 32400 }, { "epoch": 81.0025, "grad_norm": 10.5625, "learning_rate": 1.8338959989556459e-06, "loss": 1.3814, "step": 32401 }, { "epoch": 81.005, "grad_norm": 7.625, "learning_rate": 1.833428682020054e-06, "loss": 1.4465, "step": 32402 }, { "epoch": 81.0075, "grad_norm": 5.625, "learning_rate": 1.832961418624173e-06, "loss": 1.3473, "step": 32403 }, { "epoch": 81.01, "grad_norm": 7.71875, "learning_rate": 1.8324942087710706e-06, "loss": 1.28, "step": 32404 }, { "epoch": 81.0125, "grad_norm": 10.1875, "learning_rate": 1.8320270524638129e-06, "loss": 1.488, "step": 32405 }, { "epoch": 81.015, "grad_norm": 5.125, "learning_rate": 1.8315599497054582e-06, "loss": 1.2206, "step": 32406 }, { "epoch": 81.0175, "grad_norm": 7.90625, "learning_rate": 1.8310929004990663e-06, "loss": 1.3658, "step": 32407 }, { "epoch": 81.02, "grad_norm": 9.125, "learning_rate": 1.830625904847706e-06, "loss": 1.5041, "step": 32408 }, { "epoch": 81.0225, "grad_norm": 6.40625, "learning_rate": 1.8301589627544324e-06, "loss": 1.2632, "step": 32409 }, { "epoch": 81.025, "grad_norm": 8.875, "learning_rate": 1.8296920742223123e-06, "loss": 1.2697, "step": 32410 }, { "epoch": 81.0275, "grad_norm": 7.65625, "learning_rate": 1.8292252392544008e-06, "loss": 1.1746, "step": 32411 }, { "epoch": 81.03, "grad_norm": 8.25, "learning_rate": 1.8287584578537653e-06, "loss": 1.2136, "step": 32412 }, { "epoch": 81.0325, "grad_norm": 9.375, "learning_rate": 1.8282917300234615e-06, "loss": 1.405, "step": 32413 }, { "epoch": 81.035, "grad_norm": 9.125, "learning_rate": 1.8278250557665477e-06, "loss": 1.1993, "step": 32414 }, { "epoch": 81.0375, "grad_norm": 6.65625, "learning_rate": 1.8273584350860862e-06, "loss": 1.3181, "step": 32415 }, { "epoch": 81.04, "grad_norm": 8.9375, "learning_rate": 1.8268918679851388e-06, "loss": 1.2871, "step": 32416 }, { "epoch": 81.0425, "grad_norm": 10.0, "learning_rate": 1.82642535446676e-06, "loss": 1.25, "step": 32417 }, { "epoch": 81.045, "grad_norm": 6.1875, "learning_rate": 1.8259588945340102e-06, "loss": 1.2138, "step": 32418 }, { "epoch": 81.0475, "grad_norm": 6.9375, "learning_rate": 1.8254924881899416e-06, "loss": 1.2385, "step": 32419 }, { "epoch": 81.05, "grad_norm": 7.15625, "learning_rate": 1.825026135437622e-06, "loss": 1.4032, "step": 32420 }, { "epoch": 81.0525, "grad_norm": 8.25, "learning_rate": 1.8245598362801043e-06, "loss": 1.4504, "step": 32421 }, { "epoch": 81.055, "grad_norm": 9.0, "learning_rate": 1.8240935907204427e-06, "loss": 1.5095, "step": 32422 }, { "epoch": 81.0575, "grad_norm": 8.375, "learning_rate": 1.823627398761698e-06, "loss": 1.3406, "step": 32423 }, { "epoch": 81.06, "grad_norm": 9.125, "learning_rate": 1.8231612604069238e-06, "loss": 1.4506, "step": 32424 }, { "epoch": 81.0625, "grad_norm": 7.34375, "learning_rate": 1.8226951756591783e-06, "loss": 1.2413, "step": 32425 }, { "epoch": 81.065, "grad_norm": 8.25, "learning_rate": 1.8222291445215145e-06, "loss": 1.2039, "step": 32426 }, { "epoch": 81.0675, "grad_norm": 7.5625, "learning_rate": 1.8217631669969905e-06, "loss": 1.3534, "step": 32427 }, { "epoch": 81.07, "grad_norm": 7.59375, "learning_rate": 1.8212972430886612e-06, "loss": 1.3064, "step": 32428 }, { "epoch": 81.0725, "grad_norm": 10.0625, "learning_rate": 1.8208313727995763e-06, "loss": 1.4151, "step": 32429 }, { "epoch": 81.075, "grad_norm": 8.1875, "learning_rate": 1.8203655561327938e-06, "loss": 1.3917, "step": 32430 }, { "epoch": 81.0775, "grad_norm": 7.875, "learning_rate": 1.8198997930913708e-06, "loss": 1.4, "step": 32431 }, { "epoch": 81.08, "grad_norm": 10.8125, "learning_rate": 1.8194340836783565e-06, "loss": 1.3814, "step": 32432 }, { "epoch": 81.0825, "grad_norm": 8.875, "learning_rate": 1.8189684278968057e-06, "loss": 1.4098, "step": 32433 }, { "epoch": 81.085, "grad_norm": 8.75, "learning_rate": 1.818502825749764e-06, "loss": 1.3225, "step": 32434 }, { "epoch": 81.0875, "grad_norm": 7.0, "learning_rate": 1.8180372772402965e-06, "loss": 1.1184, "step": 32435 }, { "epoch": 81.09, "grad_norm": 5.46875, "learning_rate": 1.817571782371449e-06, "loss": 1.4634, "step": 32436 }, { "epoch": 81.0925, "grad_norm": 8.1875, "learning_rate": 1.817106341146274e-06, "loss": 1.2021, "step": 32437 }, { "epoch": 81.095, "grad_norm": 5.46875, "learning_rate": 1.8166409535678199e-06, "loss": 1.151, "step": 32438 }, { "epoch": 81.0975, "grad_norm": 7.96875, "learning_rate": 1.816175619639139e-06, "loss": 1.2849, "step": 32439 }, { "epoch": 81.1, "grad_norm": 9.0625, "learning_rate": 1.8157103393632869e-06, "loss": 1.2192, "step": 32440 }, { "epoch": 81.1025, "grad_norm": 5.40625, "learning_rate": 1.815245112743309e-06, "loss": 1.3254, "step": 32441 }, { "epoch": 81.105, "grad_norm": 8.5, "learning_rate": 1.8147799397822552e-06, "loss": 1.3181, "step": 32442 }, { "epoch": 81.1075, "grad_norm": 10.25, "learning_rate": 1.8143148204831795e-06, "loss": 1.2703, "step": 32443 }, { "epoch": 81.11, "grad_norm": 8.125, "learning_rate": 1.813849754849125e-06, "loss": 1.3763, "step": 32444 }, { "epoch": 81.1125, "grad_norm": 7.28125, "learning_rate": 1.8133847428831464e-06, "loss": 1.2552, "step": 32445 }, { "epoch": 81.115, "grad_norm": 8.5625, "learning_rate": 1.8129197845882873e-06, "loss": 1.4393, "step": 32446 }, { "epoch": 81.1175, "grad_norm": 8.375, "learning_rate": 1.8124548799676012e-06, "loss": 1.3216, "step": 32447 }, { "epoch": 81.12, "grad_norm": 6.4375, "learning_rate": 1.8119900290241331e-06, "loss": 1.3473, "step": 32448 }, { "epoch": 81.1225, "grad_norm": 7.0, "learning_rate": 1.8115252317609256e-06, "loss": 1.4075, "step": 32449 }, { "epoch": 81.125, "grad_norm": 7.5625, "learning_rate": 1.8110604881810357e-06, "loss": 1.4535, "step": 32450 }, { "epoch": 81.1275, "grad_norm": 7.71875, "learning_rate": 1.8105957982875066e-06, "loss": 1.3694, "step": 32451 }, { "epoch": 81.13, "grad_norm": 8.0, "learning_rate": 1.8101311620833828e-06, "loss": 1.3594, "step": 32452 }, { "epoch": 81.1325, "grad_norm": 7.71875, "learning_rate": 1.8096665795717093e-06, "loss": 1.1413, "step": 32453 }, { "epoch": 81.135, "grad_norm": 7.96875, "learning_rate": 1.8092020507555341e-06, "loss": 1.2628, "step": 32454 }, { "epoch": 81.1375, "grad_norm": 10.0, "learning_rate": 1.8087375756379045e-06, "loss": 1.2483, "step": 32455 }, { "epoch": 81.14, "grad_norm": 9.3125, "learning_rate": 1.8082731542218635e-06, "loss": 1.4419, "step": 32456 }, { "epoch": 81.1425, "grad_norm": 7.84375, "learning_rate": 1.8078087865104544e-06, "loss": 1.1893, "step": 32457 }, { "epoch": 81.145, "grad_norm": 8.625, "learning_rate": 1.8073444725067245e-06, "loss": 1.2792, "step": 32458 }, { "epoch": 81.1475, "grad_norm": 9.875, "learning_rate": 1.806880212213714e-06, "loss": 1.3731, "step": 32459 }, { "epoch": 81.15, "grad_norm": 7.6875, "learning_rate": 1.8064160056344714e-06, "loss": 1.3117, "step": 32460 }, { "epoch": 81.1525, "grad_norm": 6.875, "learning_rate": 1.8059518527720354e-06, "loss": 1.3801, "step": 32461 }, { "epoch": 81.155, "grad_norm": 5.96875, "learning_rate": 1.8054877536294535e-06, "loss": 1.319, "step": 32462 }, { "epoch": 81.1575, "grad_norm": 7.65625, "learning_rate": 1.8050237082097665e-06, "loss": 1.1686, "step": 32463 }, { "epoch": 81.16, "grad_norm": 7.21875, "learning_rate": 1.8045597165160134e-06, "loss": 1.2967, "step": 32464 }, { "epoch": 81.1625, "grad_norm": 7.875, "learning_rate": 1.804095778551238e-06, "loss": 1.3064, "step": 32465 }, { "epoch": 81.165, "grad_norm": 5.65625, "learning_rate": 1.8036318943184861e-06, "loss": 1.3566, "step": 32466 }, { "epoch": 81.1675, "grad_norm": 6.65625, "learning_rate": 1.8031680638207949e-06, "loss": 1.3131, "step": 32467 }, { "epoch": 81.17, "grad_norm": 4.6875, "learning_rate": 1.8027042870612033e-06, "loss": 1.4218, "step": 32468 }, { "epoch": 81.1725, "grad_norm": 6.5625, "learning_rate": 1.8022405640427553e-06, "loss": 1.3124, "step": 32469 }, { "epoch": 81.175, "grad_norm": 6.6875, "learning_rate": 1.8017768947684921e-06, "loss": 1.3131, "step": 32470 }, { "epoch": 81.1775, "grad_norm": 5.25, "learning_rate": 1.8013132792414512e-06, "loss": 1.2261, "step": 32471 }, { "epoch": 81.18, "grad_norm": 13.4375, "learning_rate": 1.80084971746467e-06, "loss": 1.3956, "step": 32472 }, { "epoch": 81.1825, "grad_norm": 8.0625, "learning_rate": 1.8003862094411928e-06, "loss": 1.2778, "step": 32473 }, { "epoch": 81.185, "grad_norm": 6.6875, "learning_rate": 1.7999227551740528e-06, "loss": 1.5089, "step": 32474 }, { "epoch": 81.1875, "grad_norm": 8.4375, "learning_rate": 1.799459354666293e-06, "loss": 1.4018, "step": 32475 }, { "epoch": 81.19, "grad_norm": 5.0, "learning_rate": 1.7989960079209479e-06, "loss": 1.2344, "step": 32476 }, { "epoch": 81.1925, "grad_norm": 9.1875, "learning_rate": 1.7985327149410582e-06, "loss": 1.3898, "step": 32477 }, { "epoch": 81.195, "grad_norm": 8.0, "learning_rate": 1.7980694757296601e-06, "loss": 1.3635, "step": 32478 }, { "epoch": 81.1975, "grad_norm": 7.0625, "learning_rate": 1.7976062902897883e-06, "loss": 1.28, "step": 32479 }, { "epoch": 81.2, "grad_norm": 8.1875, "learning_rate": 1.7971431586244814e-06, "loss": 1.4293, "step": 32480 }, { "epoch": 81.2025, "grad_norm": 5.9375, "learning_rate": 1.7966800807367779e-06, "loss": 1.3206, "step": 32481 }, { "epoch": 81.205, "grad_norm": 7.375, "learning_rate": 1.7962170566297121e-06, "loss": 1.3869, "step": 32482 }, { "epoch": 81.2075, "grad_norm": 7.71875, "learning_rate": 1.795754086306316e-06, "loss": 1.3707, "step": 32483 }, { "epoch": 81.21, "grad_norm": 7.15625, "learning_rate": 1.7952911697696273e-06, "loss": 1.2522, "step": 32484 }, { "epoch": 81.2125, "grad_norm": 4.90625, "learning_rate": 1.7948283070226846e-06, "loss": 1.3506, "step": 32485 }, { "epoch": 81.215, "grad_norm": 6.125, "learning_rate": 1.7943654980685198e-06, "loss": 1.3161, "step": 32486 }, { "epoch": 81.2175, "grad_norm": 7.21875, "learning_rate": 1.793902742910163e-06, "loss": 1.3171, "step": 32487 }, { "epoch": 81.22, "grad_norm": 7.625, "learning_rate": 1.7934400415506537e-06, "loss": 1.2914, "step": 32488 }, { "epoch": 81.2225, "grad_norm": 6.9375, "learning_rate": 1.7929773939930206e-06, "loss": 1.3259, "step": 32489 }, { "epoch": 81.225, "grad_norm": 7.125, "learning_rate": 1.7925148002403025e-06, "loss": 1.3057, "step": 32490 }, { "epoch": 81.2275, "grad_norm": 8.125, "learning_rate": 1.792052260295526e-06, "loss": 1.3193, "step": 32491 }, { "epoch": 81.23, "grad_norm": 8.75, "learning_rate": 1.7915897741617283e-06, "loss": 1.3669, "step": 32492 }, { "epoch": 81.2325, "grad_norm": 8.5625, "learning_rate": 1.7911273418419407e-06, "loss": 1.3905, "step": 32493 }, { "epoch": 81.235, "grad_norm": 7.96875, "learning_rate": 1.7906649633391905e-06, "loss": 1.1826, "step": 32494 }, { "epoch": 81.2375, "grad_norm": 7.1875, "learning_rate": 1.790202638656513e-06, "loss": 1.3218, "step": 32495 }, { "epoch": 81.24, "grad_norm": 5.90625, "learning_rate": 1.7897403677969405e-06, "loss": 1.3422, "step": 32496 }, { "epoch": 81.2425, "grad_norm": 8.4375, "learning_rate": 1.7892781507635016e-06, "loss": 1.3442, "step": 32497 }, { "epoch": 81.245, "grad_norm": 5.15625, "learning_rate": 1.7888159875592236e-06, "loss": 1.2983, "step": 32498 }, { "epoch": 81.2475, "grad_norm": 8.625, "learning_rate": 1.7883538781871402e-06, "loss": 1.4465, "step": 32499 }, { "epoch": 81.25, "grad_norm": 5.9375, "learning_rate": 1.7878918226502816e-06, "loss": 1.2353, "step": 32500 }, { "epoch": 81.2525, "grad_norm": 6.09375, "learning_rate": 1.787429820951677e-06, "loss": 1.459, "step": 32501 }, { "epoch": 81.255, "grad_norm": 10.5, "learning_rate": 1.7869678730943506e-06, "loss": 1.2976, "step": 32502 }, { "epoch": 81.2575, "grad_norm": 5.59375, "learning_rate": 1.7865059790813367e-06, "loss": 1.4952, "step": 32503 }, { "epoch": 81.26, "grad_norm": 8.0, "learning_rate": 1.786044138915658e-06, "loss": 1.4333, "step": 32504 }, { "epoch": 81.2625, "grad_norm": 9.25, "learning_rate": 1.785582352600348e-06, "loss": 1.2411, "step": 32505 }, { "epoch": 81.265, "grad_norm": 6.46875, "learning_rate": 1.7851206201384297e-06, "loss": 1.3067, "step": 32506 }, { "epoch": 81.2675, "grad_norm": 10.375, "learning_rate": 1.7846589415329329e-06, "loss": 1.266, "step": 32507 }, { "epoch": 81.27, "grad_norm": 6.625, "learning_rate": 1.7841973167868843e-06, "loss": 1.3162, "step": 32508 }, { "epoch": 81.2725, "grad_norm": 6.375, "learning_rate": 1.7837357459033066e-06, "loss": 1.4508, "step": 32509 }, { "epoch": 81.275, "grad_norm": 9.3125, "learning_rate": 1.7832742288852289e-06, "loss": 1.3344, "step": 32510 }, { "epoch": 81.2775, "grad_norm": 8.4375, "learning_rate": 1.7828127657356785e-06, "loss": 1.4167, "step": 32511 }, { "epoch": 81.28, "grad_norm": 8.25, "learning_rate": 1.7823513564576788e-06, "loss": 1.238, "step": 32512 }, { "epoch": 81.2825, "grad_norm": 11.125, "learning_rate": 1.7818900010542528e-06, "loss": 1.3341, "step": 32513 }, { "epoch": 81.285, "grad_norm": 6.65625, "learning_rate": 1.7814286995284268e-06, "loss": 1.3495, "step": 32514 }, { "epoch": 81.2875, "grad_norm": 9.125, "learning_rate": 1.7809674518832277e-06, "loss": 1.3154, "step": 32515 }, { "epoch": 81.29, "grad_norm": 6.96875, "learning_rate": 1.780506258121677e-06, "loss": 1.4601, "step": 32516 }, { "epoch": 81.2925, "grad_norm": 6.25, "learning_rate": 1.7800451182467958e-06, "loss": 1.4208, "step": 32517 }, { "epoch": 81.295, "grad_norm": 6.25, "learning_rate": 1.779584032261612e-06, "loss": 1.4137, "step": 32518 }, { "epoch": 81.2975, "grad_norm": 9.3125, "learning_rate": 1.7791230001691451e-06, "loss": 1.4559, "step": 32519 }, { "epoch": 81.3, "grad_norm": 10.9375, "learning_rate": 1.7786620219724205e-06, "loss": 1.517, "step": 32520 }, { "epoch": 81.3025, "grad_norm": 5.21875, "learning_rate": 1.7782010976744557e-06, "loss": 1.2994, "step": 32521 }, { "epoch": 81.305, "grad_norm": 9.3125, "learning_rate": 1.7777402272782796e-06, "loss": 1.4765, "step": 32522 }, { "epoch": 81.3075, "grad_norm": 7.1875, "learning_rate": 1.7772794107869084e-06, "loss": 1.2714, "step": 32523 }, { "epoch": 81.31, "grad_norm": 13.0, "learning_rate": 1.7768186482033611e-06, "loss": 1.185, "step": 32524 }, { "epoch": 81.3125, "grad_norm": 6.96875, "learning_rate": 1.776357939530663e-06, "loss": 1.4324, "step": 32525 }, { "epoch": 81.315, "grad_norm": 6.75, "learning_rate": 1.7758972847718347e-06, "loss": 1.3511, "step": 32526 }, { "epoch": 81.3175, "grad_norm": 6.0, "learning_rate": 1.7754366839298954e-06, "loss": 1.3764, "step": 32527 }, { "epoch": 81.32, "grad_norm": 5.84375, "learning_rate": 1.774976137007861e-06, "loss": 1.2817, "step": 32528 }, { "epoch": 81.3225, "grad_norm": 12.0625, "learning_rate": 1.774515644008754e-06, "loss": 1.417, "step": 32529 }, { "epoch": 81.325, "grad_norm": 6.6875, "learning_rate": 1.7740552049355964e-06, "loss": 1.3334, "step": 32530 }, { "epoch": 81.3275, "grad_norm": 8.8125, "learning_rate": 1.7735948197914032e-06, "loss": 1.3796, "step": 32531 }, { "epoch": 81.33, "grad_norm": 7.90625, "learning_rate": 1.7731344885791911e-06, "loss": 1.2975, "step": 32532 }, { "epoch": 81.3325, "grad_norm": 7.3125, "learning_rate": 1.7726742113019824e-06, "loss": 1.3355, "step": 32533 }, { "epoch": 81.335, "grad_norm": 6.8125, "learning_rate": 1.7722139879627897e-06, "loss": 1.4191, "step": 32534 }, { "epoch": 81.3375, "grad_norm": 9.0, "learning_rate": 1.7717538185646354e-06, "loss": 1.3869, "step": 32535 }, { "epoch": 81.34, "grad_norm": 5.375, "learning_rate": 1.7712937031105304e-06, "loss": 1.2275, "step": 32536 }, { "epoch": 81.3425, "grad_norm": 8.0, "learning_rate": 1.7708336416034976e-06, "loss": 1.2756, "step": 32537 }, { "epoch": 81.345, "grad_norm": 7.84375, "learning_rate": 1.7703736340465505e-06, "loss": 1.2362, "step": 32538 }, { "epoch": 81.3475, "grad_norm": 8.75, "learning_rate": 1.769913680442702e-06, "loss": 1.319, "step": 32539 }, { "epoch": 81.35, "grad_norm": 6.3125, "learning_rate": 1.7694537807949707e-06, "loss": 1.3618, "step": 32540 }, { "epoch": 81.3525, "grad_norm": 6.6875, "learning_rate": 1.7689939351063724e-06, "loss": 1.3225, "step": 32541 }, { "epoch": 81.355, "grad_norm": 11.6875, "learning_rate": 1.7685341433799207e-06, "loss": 1.3676, "step": 32542 }, { "epoch": 81.3575, "grad_norm": 10.625, "learning_rate": 1.7680744056186272e-06, "loss": 1.2698, "step": 32543 }, { "epoch": 81.36, "grad_norm": 6.375, "learning_rate": 1.7676147218255092e-06, "loss": 1.3344, "step": 32544 }, { "epoch": 81.3625, "grad_norm": 6.4375, "learning_rate": 1.7671550920035818e-06, "loss": 1.3459, "step": 32545 }, { "epoch": 81.365, "grad_norm": 6.3125, "learning_rate": 1.7666955161558564e-06, "loss": 1.3198, "step": 32546 }, { "epoch": 81.3675, "grad_norm": 11.3125, "learning_rate": 1.7662359942853424e-06, "loss": 1.2436, "step": 32547 }, { "epoch": 81.37, "grad_norm": 7.1875, "learning_rate": 1.7657765263950588e-06, "loss": 1.2773, "step": 32548 }, { "epoch": 81.3725, "grad_norm": 7.5625, "learning_rate": 1.765317112488012e-06, "loss": 1.2749, "step": 32549 }, { "epoch": 81.375, "grad_norm": 7.875, "learning_rate": 1.7648577525672195e-06, "loss": 1.3351, "step": 32550 }, { "epoch": 81.3775, "grad_norm": 6.3125, "learning_rate": 1.7643984466356867e-06, "loss": 1.2933, "step": 32551 }, { "epoch": 81.38, "grad_norm": 7.03125, "learning_rate": 1.7639391946964312e-06, "loss": 1.5292, "step": 32552 }, { "epoch": 81.3825, "grad_norm": 11.25, "learning_rate": 1.763479996752462e-06, "loss": 1.3309, "step": 32553 }, { "epoch": 81.385, "grad_norm": 9.0, "learning_rate": 1.7630208528067838e-06, "loss": 1.3733, "step": 32554 }, { "epoch": 81.3875, "grad_norm": 8.0625, "learning_rate": 1.7625617628624126e-06, "loss": 1.3875, "step": 32555 }, { "epoch": 81.39, "grad_norm": 10.5625, "learning_rate": 1.762102726922359e-06, "loss": 1.1762, "step": 32556 }, { "epoch": 81.3925, "grad_norm": 7.875, "learning_rate": 1.7616437449896295e-06, "loss": 1.3365, "step": 32557 }, { "epoch": 81.395, "grad_norm": 6.15625, "learning_rate": 1.7611848170672341e-06, "loss": 1.314, "step": 32558 }, { "epoch": 81.3975, "grad_norm": 11.25, "learning_rate": 1.760725943158177e-06, "loss": 1.155, "step": 32559 }, { "epoch": 81.4, "grad_norm": 7.65625, "learning_rate": 1.7602671232654755e-06, "loss": 1.4327, "step": 32560 }, { "epoch": 81.4025, "grad_norm": 10.25, "learning_rate": 1.7598083573921333e-06, "loss": 1.3613, "step": 32561 }, { "epoch": 81.405, "grad_norm": 7.59375, "learning_rate": 1.7593496455411552e-06, "loss": 1.2184, "step": 32562 }, { "epoch": 81.4075, "grad_norm": 9.5625, "learning_rate": 1.7588909877155536e-06, "loss": 1.4394, "step": 32563 }, { "epoch": 81.41, "grad_norm": 8.0, "learning_rate": 1.7584323839183305e-06, "loss": 1.4142, "step": 32564 }, { "epoch": 81.4125, "grad_norm": 10.0625, "learning_rate": 1.7579738341524965e-06, "loss": 1.445, "step": 32565 }, { "epoch": 81.415, "grad_norm": 7.28125, "learning_rate": 1.7575153384210553e-06, "loss": 1.2916, "step": 32566 }, { "epoch": 81.4175, "grad_norm": 9.9375, "learning_rate": 1.7570568967270152e-06, "loss": 1.3928, "step": 32567 }, { "epoch": 81.42, "grad_norm": 5.46875, "learning_rate": 1.7565985090733806e-06, "loss": 1.1695, "step": 32568 }, { "epoch": 81.4225, "grad_norm": 6.75, "learning_rate": 1.7561401754631525e-06, "loss": 1.2255, "step": 32569 }, { "epoch": 81.425, "grad_norm": 7.21875, "learning_rate": 1.7556818958993405e-06, "loss": 1.2674, "step": 32570 }, { "epoch": 81.4275, "grad_norm": 7.125, "learning_rate": 1.7552236703849513e-06, "loss": 1.2606, "step": 32571 }, { "epoch": 81.43, "grad_norm": 9.8125, "learning_rate": 1.7547654989229846e-06, "loss": 1.2764, "step": 32572 }, { "epoch": 81.4325, "grad_norm": 8.625, "learning_rate": 1.7543073815164458e-06, "loss": 1.271, "step": 32573 }, { "epoch": 81.435, "grad_norm": 8.875, "learning_rate": 1.7538493181683326e-06, "loss": 1.4394, "step": 32574 }, { "epoch": 81.4375, "grad_norm": 7.21875, "learning_rate": 1.753391308881659e-06, "loss": 1.4401, "step": 32575 }, { "epoch": 81.44, "grad_norm": 5.6875, "learning_rate": 1.7529333536594217e-06, "loss": 1.4098, "step": 32576 }, { "epoch": 81.4425, "grad_norm": 7.5, "learning_rate": 1.752475452504623e-06, "loss": 1.248, "step": 32577 }, { "epoch": 81.445, "grad_norm": 9.125, "learning_rate": 1.752017605420262e-06, "loss": 1.2587, "step": 32578 }, { "epoch": 81.4475, "grad_norm": 9.1875, "learning_rate": 1.7515598124093448e-06, "loss": 1.1596, "step": 32579 }, { "epoch": 81.45, "grad_norm": 8.75, "learning_rate": 1.751102073474873e-06, "loss": 1.2799, "step": 32580 }, { "epoch": 81.4525, "grad_norm": 8.375, "learning_rate": 1.7506443886198465e-06, "loss": 1.2936, "step": 32581 }, { "epoch": 81.455, "grad_norm": 19.375, "learning_rate": 1.750186757847262e-06, "loss": 1.3809, "step": 32582 }, { "epoch": 81.4575, "grad_norm": 6.65625, "learning_rate": 1.749729181160127e-06, "loss": 1.3642, "step": 32583 }, { "epoch": 81.46, "grad_norm": 8.125, "learning_rate": 1.7492716585614334e-06, "loss": 1.4938, "step": 32584 }, { "epoch": 81.4625, "grad_norm": 7.0625, "learning_rate": 1.7488141900541878e-06, "loss": 1.3899, "step": 32585 }, { "epoch": 81.465, "grad_norm": 7.875, "learning_rate": 1.7483567756413834e-06, "loss": 1.3612, "step": 32586 }, { "epoch": 81.4675, "grad_norm": 9.25, "learning_rate": 1.7478994153260242e-06, "loss": 1.4776, "step": 32587 }, { "epoch": 81.47, "grad_norm": 5.9375, "learning_rate": 1.7474421091111061e-06, "loss": 1.231, "step": 32588 }, { "epoch": 81.4725, "grad_norm": 9.9375, "learning_rate": 1.7469848569996228e-06, "loss": 1.3536, "step": 32589 }, { "epoch": 81.475, "grad_norm": 7.90625, "learning_rate": 1.7465276589945813e-06, "loss": 1.1968, "step": 32590 }, { "epoch": 81.4775, "grad_norm": 7.28125, "learning_rate": 1.7460705150989732e-06, "loss": 1.3332, "step": 32591 }, { "epoch": 81.48, "grad_norm": 11.1875, "learning_rate": 1.7456134253157976e-06, "loss": 1.3674, "step": 32592 }, { "epoch": 81.4825, "grad_norm": 12.0625, "learning_rate": 1.7451563896480472e-06, "loss": 1.2559, "step": 32593 }, { "epoch": 81.485, "grad_norm": 5.75, "learning_rate": 1.7446994080987212e-06, "loss": 1.2968, "step": 32594 }, { "epoch": 81.4875, "grad_norm": 6.59375, "learning_rate": 1.744242480670819e-06, "loss": 1.1674, "step": 32595 }, { "epoch": 81.49, "grad_norm": 5.65625, "learning_rate": 1.743785607367332e-06, "loss": 1.2452, "step": 32596 }, { "epoch": 81.4925, "grad_norm": 8.125, "learning_rate": 1.743328788191253e-06, "loss": 1.401, "step": 32597 }, { "epoch": 81.495, "grad_norm": 6.78125, "learning_rate": 1.742872023145583e-06, "loss": 1.2337, "step": 32598 }, { "epoch": 81.4975, "grad_norm": 10.25, "learning_rate": 1.7424153122333109e-06, "loss": 1.3008, "step": 32599 }, { "epoch": 81.5, "grad_norm": 7.125, "learning_rate": 1.7419586554574364e-06, "loss": 1.1843, "step": 32600 }, { "epoch": 81.5025, "grad_norm": 9.1875, "learning_rate": 1.7415020528209481e-06, "loss": 1.4317, "step": 32601 }, { "epoch": 81.505, "grad_norm": 7.5, "learning_rate": 1.7410455043268438e-06, "loss": 1.3284, "step": 32602 }, { "epoch": 81.5075, "grad_norm": 6.90625, "learning_rate": 1.7405890099781153e-06, "loss": 1.2916, "step": 32603 }, { "epoch": 81.51, "grad_norm": 14.1875, "learning_rate": 1.7401325697777515e-06, "loss": 1.2798, "step": 32604 }, { "epoch": 81.5125, "grad_norm": 9.6875, "learning_rate": 1.739676183728749e-06, "loss": 1.2045, "step": 32605 }, { "epoch": 81.515, "grad_norm": 5.28125, "learning_rate": 1.7392198518341008e-06, "loss": 1.2597, "step": 32606 }, { "epoch": 81.5175, "grad_norm": 7.4375, "learning_rate": 1.7387635740967968e-06, "loss": 1.2095, "step": 32607 }, { "epoch": 81.52, "grad_norm": 6.59375, "learning_rate": 1.7383073505198255e-06, "loss": 1.2415, "step": 32608 }, { "epoch": 81.5225, "grad_norm": 8.0625, "learning_rate": 1.7378511811061804e-06, "loss": 1.2186, "step": 32609 }, { "epoch": 81.525, "grad_norm": 8.5625, "learning_rate": 1.7373950658588557e-06, "loss": 1.4421, "step": 32610 }, { "epoch": 81.5275, "grad_norm": 7.46875, "learning_rate": 1.7369390047808376e-06, "loss": 1.3197, "step": 32611 }, { "epoch": 81.53, "grad_norm": 7.125, "learning_rate": 1.736482997875114e-06, "loss": 1.2964, "step": 32612 }, { "epoch": 81.5325, "grad_norm": 8.4375, "learning_rate": 1.736027045144678e-06, "loss": 1.2307, "step": 32613 }, { "epoch": 81.535, "grad_norm": 7.25, "learning_rate": 1.7355711465925206e-06, "loss": 1.2433, "step": 32614 }, { "epoch": 81.5375, "grad_norm": 6.84375, "learning_rate": 1.7351153022216272e-06, "loss": 1.3789, "step": 32615 }, { "epoch": 81.54, "grad_norm": 5.625, "learning_rate": 1.7346595120349852e-06, "loss": 1.3682, "step": 32616 }, { "epoch": 81.5425, "grad_norm": 5.84375, "learning_rate": 1.7342037760355868e-06, "loss": 1.2775, "step": 32617 }, { "epoch": 81.545, "grad_norm": 7.28125, "learning_rate": 1.733748094226415e-06, "loss": 1.2572, "step": 32618 }, { "epoch": 81.5475, "grad_norm": 10.1875, "learning_rate": 1.7332924666104623e-06, "loss": 1.1578, "step": 32619 }, { "epoch": 81.55, "grad_norm": 7.84375, "learning_rate": 1.7328368931907114e-06, "loss": 1.229, "step": 32620 }, { "epoch": 81.5525, "grad_norm": 6.6875, "learning_rate": 1.7323813739701534e-06, "loss": 1.1768, "step": 32621 }, { "epoch": 81.555, "grad_norm": 7.125, "learning_rate": 1.7319259089517727e-06, "loss": 1.2508, "step": 32622 }, { "epoch": 81.5575, "grad_norm": 7.5625, "learning_rate": 1.731470498138551e-06, "loss": 1.2444, "step": 32623 }, { "epoch": 81.56, "grad_norm": 9.5, "learning_rate": 1.7310151415334798e-06, "loss": 1.2634, "step": 32624 }, { "epoch": 81.5625, "grad_norm": 6.96875, "learning_rate": 1.7305598391395429e-06, "loss": 1.5903, "step": 32625 }, { "epoch": 81.565, "grad_norm": 6.46875, "learning_rate": 1.730104590959726e-06, "loss": 1.2945, "step": 32626 }, { "epoch": 81.5675, "grad_norm": 13.0, "learning_rate": 1.7296493969970096e-06, "loss": 1.4033, "step": 32627 }, { "epoch": 81.57, "grad_norm": 8.1875, "learning_rate": 1.7291942572543806e-06, "loss": 1.2748, "step": 32628 }, { "epoch": 81.5725, "grad_norm": 7.28125, "learning_rate": 1.7287391717348256e-06, "loss": 1.3674, "step": 32629 }, { "epoch": 81.575, "grad_norm": 8.625, "learning_rate": 1.7282841404413265e-06, "loss": 1.2061, "step": 32630 }, { "epoch": 81.5775, "grad_norm": 8.0, "learning_rate": 1.7278291633768617e-06, "loss": 1.3889, "step": 32631 }, { "epoch": 81.58, "grad_norm": 7.625, "learning_rate": 1.7273742405444217e-06, "loss": 1.2262, "step": 32632 }, { "epoch": 81.5825, "grad_norm": 7.75, "learning_rate": 1.7269193719469824e-06, "loss": 1.3295, "step": 32633 }, { "epoch": 81.585, "grad_norm": 6.84375, "learning_rate": 1.7264645575875304e-06, "loss": 1.4495, "step": 32634 }, { "epoch": 81.5875, "grad_norm": 9.4375, "learning_rate": 1.7260097974690438e-06, "loss": 1.4056, "step": 32635 }, { "epoch": 81.59, "grad_norm": 8.5, "learning_rate": 1.7255550915945073e-06, "loss": 1.2998, "step": 32636 }, { "epoch": 81.5925, "grad_norm": 8.0, "learning_rate": 1.7251004399669013e-06, "loss": 1.31, "step": 32637 }, { "epoch": 81.595, "grad_norm": 6.46875, "learning_rate": 1.7246458425892031e-06, "loss": 1.3264, "step": 32638 }, { "epoch": 81.5975, "grad_norm": 7.75, "learning_rate": 1.724191299464396e-06, "loss": 1.3132, "step": 32639 }, { "epoch": 81.6, "grad_norm": 6.75, "learning_rate": 1.723736810595461e-06, "loss": 1.3242, "step": 32640 }, { "epoch": 81.6025, "grad_norm": 7.03125, "learning_rate": 1.7232823759853767e-06, "loss": 1.2169, "step": 32641 }, { "epoch": 81.605, "grad_norm": 8.3125, "learning_rate": 1.7228279956371196e-06, "loss": 1.3193, "step": 32642 }, { "epoch": 81.6075, "grad_norm": 5.4375, "learning_rate": 1.7223736695536719e-06, "loss": 1.3511, "step": 32643 }, { "epoch": 81.61, "grad_norm": 6.0625, "learning_rate": 1.7219193977380121e-06, "loss": 1.2076, "step": 32644 }, { "epoch": 81.6125, "grad_norm": 6.53125, "learning_rate": 1.721465180193118e-06, "loss": 1.2929, "step": 32645 }, { "epoch": 81.615, "grad_norm": 6.46875, "learning_rate": 1.7210110169219651e-06, "loss": 1.3513, "step": 32646 }, { "epoch": 81.6175, "grad_norm": 8.0, "learning_rate": 1.720556907927534e-06, "loss": 1.4371, "step": 32647 }, { "epoch": 81.62, "grad_norm": 7.53125, "learning_rate": 1.7201028532127994e-06, "loss": 1.2312, "step": 32648 }, { "epoch": 81.6225, "grad_norm": 5.6875, "learning_rate": 1.7196488527807397e-06, "loss": 1.2391, "step": 32649 }, { "epoch": 81.625, "grad_norm": 8.4375, "learning_rate": 1.7191949066343306e-06, "loss": 1.228, "step": 32650 }, { "epoch": 81.6275, "grad_norm": 9.5625, "learning_rate": 1.7187410147765493e-06, "loss": 1.2933, "step": 32651 }, { "epoch": 81.63, "grad_norm": 8.5, "learning_rate": 1.7182871772103705e-06, "loss": 1.3485, "step": 32652 }, { "epoch": 81.6325, "grad_norm": 8.8125, "learning_rate": 1.717833393938767e-06, "loss": 1.3493, "step": 32653 }, { "epoch": 81.635, "grad_norm": 7.25, "learning_rate": 1.7173796649647167e-06, "loss": 1.2524, "step": 32654 }, { "epoch": 81.6375, "grad_norm": 5.8125, "learning_rate": 1.716925990291196e-06, "loss": 1.282, "step": 32655 }, { "epoch": 81.64, "grad_norm": 6.78125, "learning_rate": 1.7164723699211782e-06, "loss": 1.3872, "step": 32656 }, { "epoch": 81.6425, "grad_norm": 4.71875, "learning_rate": 1.716018803857632e-06, "loss": 1.2662, "step": 32657 }, { "epoch": 81.645, "grad_norm": 5.0625, "learning_rate": 1.715565292103536e-06, "loss": 1.4078, "step": 32658 }, { "epoch": 81.6475, "grad_norm": 7.5, "learning_rate": 1.7151118346618644e-06, "loss": 1.4284, "step": 32659 }, { "epoch": 81.65, "grad_norm": 6.875, "learning_rate": 1.7146584315355886e-06, "loss": 1.3496, "step": 32660 }, { "epoch": 81.6525, "grad_norm": 5.84375, "learning_rate": 1.714205082727678e-06, "loss": 1.0775, "step": 32661 }, { "epoch": 81.655, "grad_norm": 10.375, "learning_rate": 1.7137517882411092e-06, "loss": 1.2899, "step": 32662 }, { "epoch": 81.6575, "grad_norm": 7.5, "learning_rate": 1.71329854807885e-06, "loss": 1.2722, "step": 32663 }, { "epoch": 81.66, "grad_norm": 7.25, "learning_rate": 1.7128453622438768e-06, "loss": 1.353, "step": 32664 }, { "epoch": 81.6625, "grad_norm": 7.8125, "learning_rate": 1.712392230739155e-06, "loss": 1.365, "step": 32665 }, { "epoch": 81.665, "grad_norm": 5.375, "learning_rate": 1.71193915356766e-06, "loss": 1.3315, "step": 32666 }, { "epoch": 81.6675, "grad_norm": 8.1875, "learning_rate": 1.7114861307323605e-06, "loss": 1.2101, "step": 32667 }, { "epoch": 81.67, "grad_norm": 6.65625, "learning_rate": 1.7110331622362242e-06, "loss": 1.3384, "step": 32668 }, { "epoch": 81.6725, "grad_norm": 7.25, "learning_rate": 1.710580248082223e-06, "loss": 1.3104, "step": 32669 }, { "epoch": 81.675, "grad_norm": 6.4375, "learning_rate": 1.7101273882733271e-06, "loss": 1.3244, "step": 32670 }, { "epoch": 81.6775, "grad_norm": 8.125, "learning_rate": 1.709674582812506e-06, "loss": 1.3579, "step": 32671 }, { "epoch": 81.68, "grad_norm": 6.34375, "learning_rate": 1.709221831702723e-06, "loss": 1.3624, "step": 32672 }, { "epoch": 81.6825, "grad_norm": 7.03125, "learning_rate": 1.7087691349469503e-06, "loss": 1.3339, "step": 32673 }, { "epoch": 81.685, "grad_norm": 9.625, "learning_rate": 1.7083164925481577e-06, "loss": 1.3085, "step": 32674 }, { "epoch": 81.6875, "grad_norm": 8.5625, "learning_rate": 1.7078639045093105e-06, "loss": 1.3753, "step": 32675 }, { "epoch": 81.69, "grad_norm": 7.84375, "learning_rate": 1.7074113708333727e-06, "loss": 1.2481, "step": 32676 }, { "epoch": 81.6925, "grad_norm": 9.625, "learning_rate": 1.706958891523317e-06, "loss": 1.4253, "step": 32677 }, { "epoch": 81.695, "grad_norm": 5.6875, "learning_rate": 1.706506466582104e-06, "loss": 1.3157, "step": 32678 }, { "epoch": 81.6975, "grad_norm": 6.0, "learning_rate": 1.706054096012706e-06, "loss": 1.4103, "step": 32679 }, { "epoch": 81.7, "grad_norm": 6.0625, "learning_rate": 1.7056017798180824e-06, "loss": 1.3103, "step": 32680 }, { "epoch": 81.7025, "grad_norm": 5.78125, "learning_rate": 1.7051495180012045e-06, "loss": 1.2499, "step": 32681 }, { "epoch": 81.705, "grad_norm": 7.09375, "learning_rate": 1.7046973105650344e-06, "loss": 1.3978, "step": 32682 }, { "epoch": 81.7075, "grad_norm": 7.75, "learning_rate": 1.704245157512534e-06, "loss": 1.3536, "step": 32683 }, { "epoch": 81.71, "grad_norm": 8.25, "learning_rate": 1.7037930588466701e-06, "loss": 1.297, "step": 32684 }, { "epoch": 81.7125, "grad_norm": 8.0, "learning_rate": 1.70334101457041e-06, "loss": 1.3729, "step": 32685 }, { "epoch": 81.715, "grad_norm": 9.4375, "learning_rate": 1.7028890246867136e-06, "loss": 1.5636, "step": 32686 }, { "epoch": 81.7175, "grad_norm": 10.5, "learning_rate": 1.7024370891985432e-06, "loss": 1.42, "step": 32687 }, { "epoch": 81.72, "grad_norm": 7.34375, "learning_rate": 1.7019852081088616e-06, "loss": 1.3363, "step": 32688 }, { "epoch": 81.7225, "grad_norm": 8.25, "learning_rate": 1.701533381420637e-06, "loss": 1.2964, "step": 32689 }, { "epoch": 81.725, "grad_norm": 6.0625, "learning_rate": 1.7010816091368265e-06, "loss": 1.3975, "step": 32690 }, { "epoch": 81.7275, "grad_norm": 9.25, "learning_rate": 1.7006298912603915e-06, "loss": 1.4529, "step": 32691 }, { "epoch": 81.73, "grad_norm": 4.9375, "learning_rate": 1.7001782277942958e-06, "loss": 1.2951, "step": 32692 }, { "epoch": 81.7325, "grad_norm": 6.0, "learning_rate": 1.6997266187414973e-06, "loss": 1.2822, "step": 32693 }, { "epoch": 81.735, "grad_norm": 8.875, "learning_rate": 1.6992750641049628e-06, "loss": 1.4347, "step": 32694 }, { "epoch": 81.7375, "grad_norm": 6.0, "learning_rate": 1.698823563887645e-06, "loss": 1.2789, "step": 32695 }, { "epoch": 81.74, "grad_norm": 7.8125, "learning_rate": 1.6983721180925106e-06, "loss": 1.3049, "step": 32696 }, { "epoch": 81.7425, "grad_norm": 5.96875, "learning_rate": 1.6979207267225162e-06, "loss": 1.3352, "step": 32697 }, { "epoch": 81.745, "grad_norm": 7.25, "learning_rate": 1.6974693897806193e-06, "loss": 1.2385, "step": 32698 }, { "epoch": 81.7475, "grad_norm": 8.5625, "learning_rate": 1.69701810726978e-06, "loss": 1.3471, "step": 32699 }, { "epoch": 81.75, "grad_norm": 7.625, "learning_rate": 1.69656687919296e-06, "loss": 1.3228, "step": 32700 }, { "epoch": 81.7525, "grad_norm": 8.1875, "learning_rate": 1.6961157055531164e-06, "loss": 1.2979, "step": 32701 }, { "epoch": 81.755, "grad_norm": 8.8125, "learning_rate": 1.6956645863532028e-06, "loss": 1.0745, "step": 32702 }, { "epoch": 81.7575, "grad_norm": 8.25, "learning_rate": 1.6952135215961807e-06, "loss": 1.1831, "step": 32703 }, { "epoch": 81.76, "grad_norm": 7.125, "learning_rate": 1.6947625112850074e-06, "loss": 1.3749, "step": 32704 }, { "epoch": 81.7625, "grad_norm": 7.8125, "learning_rate": 1.6943115554226397e-06, "loss": 1.2809, "step": 32705 }, { "epoch": 81.765, "grad_norm": 11.125, "learning_rate": 1.6938606540120306e-06, "loss": 1.334, "step": 32706 }, { "epoch": 81.7675, "grad_norm": 7.25, "learning_rate": 1.6934098070561412e-06, "loss": 1.4922, "step": 32707 }, { "epoch": 81.77, "grad_norm": 6.8125, "learning_rate": 1.6929590145579223e-06, "loss": 1.2224, "step": 32708 }, { "epoch": 81.7725, "grad_norm": 8.8125, "learning_rate": 1.6925082765203339e-06, "loss": 1.5011, "step": 32709 }, { "epoch": 81.775, "grad_norm": 8.125, "learning_rate": 1.6920575929463257e-06, "loss": 1.2686, "step": 32710 }, { "epoch": 81.7775, "grad_norm": 6.78125, "learning_rate": 1.6916069638388589e-06, "loss": 1.3091, "step": 32711 }, { "epoch": 81.78, "grad_norm": 5.25, "learning_rate": 1.691156389200883e-06, "loss": 1.4277, "step": 32712 }, { "epoch": 81.7825, "grad_norm": 7.90625, "learning_rate": 1.6907058690353528e-06, "loss": 1.1888, "step": 32713 }, { "epoch": 81.785, "grad_norm": 9.1875, "learning_rate": 1.6902554033452212e-06, "loss": 1.3638, "step": 32714 }, { "epoch": 81.7875, "grad_norm": 7.25, "learning_rate": 1.6898049921334448e-06, "loss": 1.2741, "step": 32715 }, { "epoch": 81.79, "grad_norm": 7.21875, "learning_rate": 1.6893546354029754e-06, "loss": 1.2349, "step": 32716 }, { "epoch": 81.7925, "grad_norm": 10.75, "learning_rate": 1.6889043331567633e-06, "loss": 1.3322, "step": 32717 }, { "epoch": 81.795, "grad_norm": 7.6875, "learning_rate": 1.6884540853977571e-06, "loss": 1.4908, "step": 32718 }, { "epoch": 81.7975, "grad_norm": 7.90625, "learning_rate": 1.6880038921289187e-06, "loss": 1.4928, "step": 32719 }, { "epoch": 81.8, "grad_norm": 7.84375, "learning_rate": 1.687553753353195e-06, "loss": 1.3273, "step": 32720 }, { "epoch": 81.8025, "grad_norm": 7.25, "learning_rate": 1.6871036690735343e-06, "loss": 1.3281, "step": 32721 }, { "epoch": 81.805, "grad_norm": 6.78125, "learning_rate": 1.6866536392928878e-06, "loss": 1.2834, "step": 32722 }, { "epoch": 81.8075, "grad_norm": 9.3125, "learning_rate": 1.6862036640142077e-06, "loss": 1.3838, "step": 32723 }, { "epoch": 81.81, "grad_norm": 5.5625, "learning_rate": 1.6857537432404458e-06, "loss": 1.072, "step": 32724 }, { "epoch": 81.8125, "grad_norm": 8.25, "learning_rate": 1.6853038769745466e-06, "loss": 1.3221, "step": 32725 }, { "epoch": 81.815, "grad_norm": 10.0, "learning_rate": 1.6848540652194657e-06, "loss": 1.2898, "step": 32726 }, { "epoch": 81.8175, "grad_norm": 6.875, "learning_rate": 1.6844043079781492e-06, "loss": 1.3363, "step": 32727 }, { "epoch": 81.82, "grad_norm": 6.0625, "learning_rate": 1.6839546052535416e-06, "loss": 1.4148, "step": 32728 }, { "epoch": 81.8225, "grad_norm": 9.125, "learning_rate": 1.683504957048595e-06, "loss": 1.3597, "step": 32729 }, { "epoch": 81.825, "grad_norm": 8.6875, "learning_rate": 1.6830553633662604e-06, "loss": 1.496, "step": 32730 }, { "epoch": 81.8275, "grad_norm": 8.0625, "learning_rate": 1.6826058242094822e-06, "loss": 1.3328, "step": 32731 }, { "epoch": 81.83, "grad_norm": 5.78125, "learning_rate": 1.6821563395812057e-06, "loss": 1.3199, "step": 32732 }, { "epoch": 81.8325, "grad_norm": 5.75, "learning_rate": 1.6817069094843763e-06, "loss": 1.2182, "step": 32733 }, { "epoch": 81.835, "grad_norm": 7.6875, "learning_rate": 1.6812575339219472e-06, "loss": 1.2317, "step": 32734 }, { "epoch": 81.8375, "grad_norm": 5.4375, "learning_rate": 1.6808082128968618e-06, "loss": 1.2322, "step": 32735 }, { "epoch": 81.84, "grad_norm": 5.28125, "learning_rate": 1.680358946412064e-06, "loss": 1.4166, "step": 32736 }, { "epoch": 81.8425, "grad_norm": 6.5, "learning_rate": 1.6799097344704973e-06, "loss": 1.3695, "step": 32737 }, { "epoch": 81.845, "grad_norm": 8.0625, "learning_rate": 1.6794605770751104e-06, "loss": 1.2623, "step": 32738 }, { "epoch": 81.8475, "grad_norm": 7.34375, "learning_rate": 1.6790114742288489e-06, "loss": 1.2637, "step": 32739 }, { "epoch": 81.85, "grad_norm": 7.65625, "learning_rate": 1.6785624259346556e-06, "loss": 1.3813, "step": 32740 }, { "epoch": 81.8525, "grad_norm": 11.25, "learning_rate": 1.6781134321954705e-06, "loss": 1.4255, "step": 32741 }, { "epoch": 81.855, "grad_norm": 7.25, "learning_rate": 1.6776644930142449e-06, "loss": 1.2635, "step": 32742 }, { "epoch": 81.8575, "grad_norm": 7.9375, "learning_rate": 1.677215608393914e-06, "loss": 1.2641, "step": 32743 }, { "epoch": 81.86, "grad_norm": 6.40625, "learning_rate": 1.6767667783374276e-06, "loss": 1.1487, "step": 32744 }, { "epoch": 81.8625, "grad_norm": 8.0625, "learning_rate": 1.6763180028477222e-06, "loss": 1.2333, "step": 32745 }, { "epoch": 81.865, "grad_norm": 4.75, "learning_rate": 1.6758692819277456e-06, "loss": 1.3866, "step": 32746 }, { "epoch": 81.8675, "grad_norm": 8.1875, "learning_rate": 1.6754206155804376e-06, "loss": 1.2645, "step": 32747 }, { "epoch": 81.87, "grad_norm": 9.0, "learning_rate": 1.6749720038087359e-06, "loss": 1.3421, "step": 32748 }, { "epoch": 81.8725, "grad_norm": 11.3125, "learning_rate": 1.6745234466155835e-06, "loss": 1.2936, "step": 32749 }, { "epoch": 81.875, "grad_norm": 7.71875, "learning_rate": 1.6740749440039262e-06, "loss": 1.3677, "step": 32750 }, { "epoch": 81.8775, "grad_norm": 10.5625, "learning_rate": 1.6736264959766991e-06, "loss": 1.287, "step": 32751 }, { "epoch": 81.88, "grad_norm": 9.1875, "learning_rate": 1.6731781025368422e-06, "loss": 1.3583, "step": 32752 }, { "epoch": 81.8825, "grad_norm": 6.625, "learning_rate": 1.6727297636872963e-06, "loss": 1.3745, "step": 32753 }, { "epoch": 81.885, "grad_norm": 5.0, "learning_rate": 1.6722814794310027e-06, "loss": 1.2796, "step": 32754 }, { "epoch": 81.8875, "grad_norm": 6.25, "learning_rate": 1.671833249770899e-06, "loss": 1.3632, "step": 32755 }, { "epoch": 81.89, "grad_norm": 7.28125, "learning_rate": 1.6713850747099202e-06, "loss": 1.481, "step": 32756 }, { "epoch": 81.8925, "grad_norm": 7.4375, "learning_rate": 1.6709369542510101e-06, "loss": 1.3517, "step": 32757 }, { "epoch": 81.895, "grad_norm": 6.875, "learning_rate": 1.6704888883971016e-06, "loss": 1.095, "step": 32758 }, { "epoch": 81.8975, "grad_norm": 8.25, "learning_rate": 1.6700408771511378e-06, "loss": 1.3316, "step": 32759 }, { "epoch": 81.9, "grad_norm": 6.78125, "learning_rate": 1.669592920516049e-06, "loss": 1.2987, "step": 32760 }, { "epoch": 81.9025, "grad_norm": 8.125, "learning_rate": 1.669145018494779e-06, "loss": 1.3136, "step": 32761 }, { "epoch": 81.905, "grad_norm": 6.625, "learning_rate": 1.6686971710902611e-06, "loss": 1.3172, "step": 32762 }, { "epoch": 81.9075, "grad_norm": 6.71875, "learning_rate": 1.6682493783054277e-06, "loss": 1.4849, "step": 32763 }, { "epoch": 81.91, "grad_norm": 7.875, "learning_rate": 1.6678016401432184e-06, "loss": 1.2341, "step": 32764 }, { "epoch": 81.9125, "grad_norm": 7.4375, "learning_rate": 1.667353956606571e-06, "loss": 1.3318, "step": 32765 }, { "epoch": 81.915, "grad_norm": 7.78125, "learning_rate": 1.6669063276984177e-06, "loss": 1.1991, "step": 32766 }, { "epoch": 81.9175, "grad_norm": 9.0625, "learning_rate": 1.6664587534216892e-06, "loss": 1.3938, "step": 32767 }, { "epoch": 81.92, "grad_norm": 6.09375, "learning_rate": 1.6660112337793256e-06, "loss": 1.385, "step": 32768 }, { "epoch": 81.9225, "grad_norm": 7.21875, "learning_rate": 1.6655637687742597e-06, "loss": 1.2635, "step": 32769 }, { "epoch": 81.925, "grad_norm": 7.875, "learning_rate": 1.6651163584094244e-06, "loss": 1.2026, "step": 32770 }, { "epoch": 81.9275, "grad_norm": 8.0625, "learning_rate": 1.664669002687751e-06, "loss": 1.2511, "step": 32771 }, { "epoch": 81.93, "grad_norm": 9.4375, "learning_rate": 1.6642217016121765e-06, "loss": 1.3591, "step": 32772 }, { "epoch": 81.9325, "grad_norm": 5.15625, "learning_rate": 1.6637744551856284e-06, "loss": 1.2166, "step": 32773 }, { "epoch": 81.935, "grad_norm": 7.6875, "learning_rate": 1.6633272634110432e-06, "loss": 1.2351, "step": 32774 }, { "epoch": 81.9375, "grad_norm": 12.0625, "learning_rate": 1.6628801262913485e-06, "loss": 1.5272, "step": 32775 }, { "epoch": 81.94, "grad_norm": 10.0625, "learning_rate": 1.662433043829481e-06, "loss": 1.4651, "step": 32776 }, { "epoch": 81.9425, "grad_norm": 13.5, "learning_rate": 1.6619860160283684e-06, "loss": 1.4468, "step": 32777 }, { "epoch": 81.945, "grad_norm": 9.625, "learning_rate": 1.6615390428909394e-06, "loss": 1.3132, "step": 32778 }, { "epoch": 81.9475, "grad_norm": 7.40625, "learning_rate": 1.6610921244201272e-06, "loss": 1.2434, "step": 32779 }, { "epoch": 81.95, "grad_norm": 8.3125, "learning_rate": 1.660645260618864e-06, "loss": 1.3604, "step": 32780 }, { "epoch": 81.9525, "grad_norm": 8.3125, "learning_rate": 1.660198451490076e-06, "loss": 1.2618, "step": 32781 }, { "epoch": 81.955, "grad_norm": 5.78125, "learning_rate": 1.65975169703669e-06, "loss": 1.3005, "step": 32782 }, { "epoch": 81.9575, "grad_norm": 9.25, "learning_rate": 1.6593049972616393e-06, "loss": 1.4578, "step": 32783 }, { "epoch": 81.96, "grad_norm": 7.875, "learning_rate": 1.6588583521678536e-06, "loss": 1.4595, "step": 32784 }, { "epoch": 81.9625, "grad_norm": 6.59375, "learning_rate": 1.6584117617582585e-06, "loss": 1.3389, "step": 32785 }, { "epoch": 81.965, "grad_norm": 7.1875, "learning_rate": 1.6579652260357803e-06, "loss": 1.2601, "step": 32786 }, { "epoch": 81.9675, "grad_norm": 9.0, "learning_rate": 1.6575187450033504e-06, "loss": 1.4395, "step": 32787 }, { "epoch": 81.97, "grad_norm": 8.3125, "learning_rate": 1.6570723186638904e-06, "loss": 1.4245, "step": 32788 }, { "epoch": 81.9725, "grad_norm": 5.8125, "learning_rate": 1.6566259470203338e-06, "loss": 1.2215, "step": 32789 }, { "epoch": 81.975, "grad_norm": 8.75, "learning_rate": 1.6561796300756006e-06, "loss": 1.252, "step": 32790 }, { "epoch": 81.9775, "grad_norm": 10.4375, "learning_rate": 1.6557333678326227e-06, "loss": 1.2605, "step": 32791 }, { "epoch": 81.98, "grad_norm": 5.65625, "learning_rate": 1.6552871602943233e-06, "loss": 1.1751, "step": 32792 }, { "epoch": 81.9825, "grad_norm": 8.9375, "learning_rate": 1.6548410074636246e-06, "loss": 1.266, "step": 32793 }, { "epoch": 81.985, "grad_norm": 11.3125, "learning_rate": 1.6543949093434541e-06, "loss": 1.377, "step": 32794 }, { "epoch": 81.9875, "grad_norm": 7.53125, "learning_rate": 1.6539488659367387e-06, "loss": 1.2608, "step": 32795 }, { "epoch": 81.99, "grad_norm": 5.84375, "learning_rate": 1.6535028772464013e-06, "loss": 1.3629, "step": 32796 }, { "epoch": 81.9925, "grad_norm": 6.53125, "learning_rate": 1.6530569432753618e-06, "loss": 1.1665, "step": 32797 }, { "epoch": 81.995, "grad_norm": 6.9375, "learning_rate": 1.652611064026548e-06, "loss": 1.3297, "step": 32798 }, { "epoch": 81.9975, "grad_norm": 7.09375, "learning_rate": 1.652165239502883e-06, "loss": 1.1471, "step": 32799 }, { "epoch": 82.0, "grad_norm": 10.0, "learning_rate": 1.6517194697072903e-06, "loss": 1.3532, "step": 32800 }, { "epoch": 82.0025, "grad_norm": 6.8125, "learning_rate": 1.6512737546426871e-06, "loss": 1.0871, "step": 32801 }, { "epoch": 82.005, "grad_norm": 7.0625, "learning_rate": 1.6508280943120014e-06, "loss": 1.3904, "step": 32802 }, { "epoch": 82.0075, "grad_norm": 7.03125, "learning_rate": 1.6503824887181509e-06, "loss": 1.4014, "step": 32803 }, { "epoch": 82.01, "grad_norm": 12.25, "learning_rate": 1.6499369378640606e-06, "loss": 1.3486, "step": 32804 }, { "epoch": 82.0125, "grad_norm": 8.8125, "learning_rate": 1.6494914417526475e-06, "loss": 1.357, "step": 32805 }, { "epoch": 82.015, "grad_norm": 5.78125, "learning_rate": 1.649046000386837e-06, "loss": 1.1908, "step": 32806 }, { "epoch": 82.0175, "grad_norm": 6.53125, "learning_rate": 1.6486006137695453e-06, "loss": 1.1638, "step": 32807 }, { "epoch": 82.02, "grad_norm": 5.21875, "learning_rate": 1.6481552819036927e-06, "loss": 1.2049, "step": 32808 }, { "epoch": 82.0225, "grad_norm": 9.9375, "learning_rate": 1.6477100047922e-06, "loss": 1.4279, "step": 32809 }, { "epoch": 82.025, "grad_norm": 7.5625, "learning_rate": 1.6472647824379883e-06, "loss": 1.3382, "step": 32810 }, { "epoch": 82.0275, "grad_norm": 6.03125, "learning_rate": 1.6468196148439742e-06, "loss": 1.3238, "step": 32811 }, { "epoch": 82.03, "grad_norm": 7.15625, "learning_rate": 1.6463745020130738e-06, "loss": 1.2864, "step": 32812 }, { "epoch": 82.0325, "grad_norm": 6.71875, "learning_rate": 1.6459294439482087e-06, "loss": 1.3164, "step": 32813 }, { "epoch": 82.035, "grad_norm": 7.40625, "learning_rate": 1.6454844406522974e-06, "loss": 1.3342, "step": 32814 }, { "epoch": 82.0375, "grad_norm": 7.59375, "learning_rate": 1.6450394921282565e-06, "loss": 1.3064, "step": 32815 }, { "epoch": 82.04, "grad_norm": 8.4375, "learning_rate": 1.644594598378999e-06, "loss": 1.2019, "step": 32816 }, { "epoch": 82.0425, "grad_norm": 6.21875, "learning_rate": 1.6441497594074485e-06, "loss": 1.3003, "step": 32817 }, { "epoch": 82.045, "grad_norm": 9.125, "learning_rate": 1.6437049752165145e-06, "loss": 1.4037, "step": 32818 }, { "epoch": 82.0475, "grad_norm": 6.9375, "learning_rate": 1.6432602458091196e-06, "loss": 1.289, "step": 32819 }, { "epoch": 82.05, "grad_norm": 7.84375, "learning_rate": 1.6428155711881722e-06, "loss": 1.278, "step": 32820 }, { "epoch": 82.0525, "grad_norm": 6.625, "learning_rate": 1.6423709513565955e-06, "loss": 1.3792, "step": 32821 }, { "epoch": 82.055, "grad_norm": 6.1875, "learning_rate": 1.6419263863172997e-06, "loss": 1.5255, "step": 32822 }, { "epoch": 82.0575, "grad_norm": 6.8125, "learning_rate": 1.6414818760731976e-06, "loss": 1.3283, "step": 32823 }, { "epoch": 82.06, "grad_norm": 8.0, "learning_rate": 1.6410374206272061e-06, "loss": 1.5836, "step": 32824 }, { "epoch": 82.0625, "grad_norm": 7.59375, "learning_rate": 1.6405930199822406e-06, "loss": 1.3507, "step": 32825 }, { "epoch": 82.065, "grad_norm": 6.78125, "learning_rate": 1.6401486741412132e-06, "loss": 1.263, "step": 32826 }, { "epoch": 82.0675, "grad_norm": 7.3125, "learning_rate": 1.6397043831070337e-06, "loss": 1.1876, "step": 32827 }, { "epoch": 82.07, "grad_norm": 5.1875, "learning_rate": 1.6392601468826187e-06, "loss": 1.2118, "step": 32828 }, { "epoch": 82.0725, "grad_norm": 10.5625, "learning_rate": 1.6388159654708814e-06, "loss": 1.3093, "step": 32829 }, { "epoch": 82.075, "grad_norm": 11.125, "learning_rate": 1.6383718388747315e-06, "loss": 1.2579, "step": 32830 }, { "epoch": 82.0775, "grad_norm": 6.4375, "learning_rate": 1.6379277670970796e-06, "loss": 1.2591, "step": 32831 }, { "epoch": 82.08, "grad_norm": 7.96875, "learning_rate": 1.6374837501408403e-06, "loss": 1.3312, "step": 32832 }, { "epoch": 82.0825, "grad_norm": 8.5, "learning_rate": 1.6370397880089206e-06, "loss": 1.0845, "step": 32833 }, { "epoch": 82.085, "grad_norm": 6.375, "learning_rate": 1.636595880704236e-06, "loss": 1.4021, "step": 32834 }, { "epoch": 82.0875, "grad_norm": 7.0, "learning_rate": 1.6361520282296928e-06, "loss": 1.1603, "step": 32835 }, { "epoch": 82.09, "grad_norm": 7.84375, "learning_rate": 1.6357082305882032e-06, "loss": 1.2765, "step": 32836 }, { "epoch": 82.0925, "grad_norm": 7.21875, "learning_rate": 1.6352644877826763e-06, "loss": 1.4267, "step": 32837 }, { "epoch": 82.095, "grad_norm": 10.5625, "learning_rate": 1.6348207998160182e-06, "loss": 1.5464, "step": 32838 }, { "epoch": 82.0975, "grad_norm": 7.21875, "learning_rate": 1.63437716669114e-06, "loss": 1.39, "step": 32839 }, { "epoch": 82.1, "grad_norm": 9.0625, "learning_rate": 1.633933588410952e-06, "loss": 1.3084, "step": 32840 }, { "epoch": 82.1025, "grad_norm": 6.40625, "learning_rate": 1.6334900649783615e-06, "loss": 1.1087, "step": 32841 }, { "epoch": 82.105, "grad_norm": 8.375, "learning_rate": 1.6330465963962726e-06, "loss": 1.2398, "step": 32842 }, { "epoch": 82.1075, "grad_norm": 7.46875, "learning_rate": 1.632603182667596e-06, "loss": 1.2929, "step": 32843 }, { "epoch": 82.11, "grad_norm": 8.5625, "learning_rate": 1.6321598237952408e-06, "loss": 1.3352, "step": 32844 }, { "epoch": 82.1125, "grad_norm": 6.71875, "learning_rate": 1.6317165197821094e-06, "loss": 1.2774, "step": 32845 }, { "epoch": 82.115, "grad_norm": 8.5, "learning_rate": 1.6312732706311086e-06, "loss": 1.2477, "step": 32846 }, { "epoch": 82.1175, "grad_norm": 10.4375, "learning_rate": 1.6308300763451445e-06, "loss": 1.521, "step": 32847 }, { "epoch": 82.12, "grad_norm": 7.1875, "learning_rate": 1.6303869369271264e-06, "loss": 1.2412, "step": 32848 }, { "epoch": 82.1225, "grad_norm": 9.6875, "learning_rate": 1.629943852379956e-06, "loss": 1.3568, "step": 32849 }, { "epoch": 82.125, "grad_norm": 7.1875, "learning_rate": 1.6295008227065367e-06, "loss": 1.4178, "step": 32850 }, { "epoch": 82.1275, "grad_norm": 5.8125, "learning_rate": 1.6290578479097785e-06, "loss": 1.4338, "step": 32851 }, { "epoch": 82.13, "grad_norm": 7.71875, "learning_rate": 1.6286149279925778e-06, "loss": 1.2392, "step": 32852 }, { "epoch": 82.1325, "grad_norm": 8.875, "learning_rate": 1.6281720629578456e-06, "loss": 1.4786, "step": 32853 }, { "epoch": 82.135, "grad_norm": 6.0625, "learning_rate": 1.627729252808481e-06, "loss": 1.1625, "step": 32854 }, { "epoch": 82.1375, "grad_norm": 6.53125, "learning_rate": 1.62728649754739e-06, "loss": 1.225, "step": 32855 }, { "epoch": 82.14, "grad_norm": 16.125, "learning_rate": 1.626843797177473e-06, "loss": 1.4949, "step": 32856 }, { "epoch": 82.1425, "grad_norm": 6.40625, "learning_rate": 1.626401151701632e-06, "loss": 1.3676, "step": 32857 }, { "epoch": 82.145, "grad_norm": 7.96875, "learning_rate": 1.625958561122769e-06, "loss": 1.2682, "step": 32858 }, { "epoch": 82.1475, "grad_norm": 4.65625, "learning_rate": 1.6255160254437895e-06, "loss": 1.2108, "step": 32859 }, { "epoch": 82.15, "grad_norm": 7.8125, "learning_rate": 1.6250735446675914e-06, "loss": 1.3849, "step": 32860 }, { "epoch": 82.1525, "grad_norm": 6.84375, "learning_rate": 1.6246311187970755e-06, "loss": 1.0811, "step": 32861 }, { "epoch": 82.155, "grad_norm": 6.71875, "learning_rate": 1.6241887478351393e-06, "loss": 1.2428, "step": 32862 }, { "epoch": 82.1575, "grad_norm": 6.8125, "learning_rate": 1.6237464317846907e-06, "loss": 1.3544, "step": 32863 }, { "epoch": 82.16, "grad_norm": 6.5, "learning_rate": 1.6233041706486253e-06, "loss": 1.4078, "step": 32864 }, { "epoch": 82.1625, "grad_norm": 7.34375, "learning_rate": 1.6228619644298394e-06, "loss": 1.4335, "step": 32865 }, { "epoch": 82.165, "grad_norm": 10.4375, "learning_rate": 1.6224198131312385e-06, "loss": 1.3075, "step": 32866 }, { "epoch": 82.1675, "grad_norm": 6.40625, "learning_rate": 1.621977716755715e-06, "loss": 1.4137, "step": 32867 }, { "epoch": 82.17, "grad_norm": 6.625, "learning_rate": 1.621535675306174e-06, "loss": 1.2372, "step": 32868 }, { "epoch": 82.1725, "grad_norm": 6.84375, "learning_rate": 1.6210936887855055e-06, "loss": 1.3292, "step": 32869 }, { "epoch": 82.175, "grad_norm": 6.625, "learning_rate": 1.6206517571966152e-06, "loss": 1.4461, "step": 32870 }, { "epoch": 82.1775, "grad_norm": 7.375, "learning_rate": 1.6202098805423971e-06, "loss": 1.4765, "step": 32871 }, { "epoch": 82.18, "grad_norm": 7.15625, "learning_rate": 1.6197680588257435e-06, "loss": 1.4118, "step": 32872 }, { "epoch": 82.1825, "grad_norm": 6.28125, "learning_rate": 1.6193262920495567e-06, "loss": 1.3149, "step": 32873 }, { "epoch": 82.185, "grad_norm": 8.4375, "learning_rate": 1.618884580216733e-06, "loss": 1.3281, "step": 32874 }, { "epoch": 82.1875, "grad_norm": 7.8125, "learning_rate": 1.6184429233301669e-06, "loss": 1.3175, "step": 32875 }, { "epoch": 82.19, "grad_norm": 6.84375, "learning_rate": 1.6180013213927538e-06, "loss": 1.1986, "step": 32876 }, { "epoch": 82.1925, "grad_norm": 7.3125, "learning_rate": 1.6175597744073834e-06, "loss": 1.2207, "step": 32877 }, { "epoch": 82.195, "grad_norm": 7.0, "learning_rate": 1.6171182823769603e-06, "loss": 1.3193, "step": 32878 }, { "epoch": 82.1975, "grad_norm": 6.75, "learning_rate": 1.6166768453043757e-06, "loss": 1.2099, "step": 32879 }, { "epoch": 82.2, "grad_norm": 7.46875, "learning_rate": 1.6162354631925203e-06, "loss": 1.3406, "step": 32880 }, { "epoch": 82.2025, "grad_norm": 5.6875, "learning_rate": 1.6157941360442886e-06, "loss": 1.2761, "step": 32881 }, { "epoch": 82.205, "grad_norm": 7.5625, "learning_rate": 1.6153528638625748e-06, "loss": 1.4022, "step": 32882 }, { "epoch": 82.2075, "grad_norm": 5.84375, "learning_rate": 1.6149116466502734e-06, "loss": 1.2371, "step": 32883 }, { "epoch": 82.21, "grad_norm": 5.9375, "learning_rate": 1.6144704844102776e-06, "loss": 1.1714, "step": 32884 }, { "epoch": 82.2125, "grad_norm": 5.46875, "learning_rate": 1.614029377145474e-06, "loss": 1.1941, "step": 32885 }, { "epoch": 82.215, "grad_norm": 9.6875, "learning_rate": 1.6135883248587602e-06, "loss": 1.4994, "step": 32886 }, { "epoch": 82.2175, "grad_norm": 8.1875, "learning_rate": 1.6131473275530241e-06, "loss": 1.3685, "step": 32887 }, { "epoch": 82.22, "grad_norm": 6.9375, "learning_rate": 1.6127063852311608e-06, "loss": 1.2872, "step": 32888 }, { "epoch": 82.2225, "grad_norm": 4.8125, "learning_rate": 1.6122654978960572e-06, "loss": 1.3082, "step": 32889 }, { "epoch": 82.225, "grad_norm": 7.5, "learning_rate": 1.6118246655506065e-06, "loss": 1.2666, "step": 32890 }, { "epoch": 82.2275, "grad_norm": 8.375, "learning_rate": 1.6113838881976985e-06, "loss": 1.2992, "step": 32891 }, { "epoch": 82.23, "grad_norm": 7.59375, "learning_rate": 1.6109431658402163e-06, "loss": 1.2572, "step": 32892 }, { "epoch": 82.2325, "grad_norm": 11.0, "learning_rate": 1.6105024984810601e-06, "loss": 1.372, "step": 32893 }, { "epoch": 82.235, "grad_norm": 11.75, "learning_rate": 1.6100618861231144e-06, "loss": 1.391, "step": 32894 }, { "epoch": 82.2375, "grad_norm": 5.5, "learning_rate": 1.6096213287692663e-06, "loss": 1.3222, "step": 32895 }, { "epoch": 82.24, "grad_norm": 7.3125, "learning_rate": 1.609180826422404e-06, "loss": 1.2823, "step": 32896 }, { "epoch": 82.2425, "grad_norm": 8.0, "learning_rate": 1.608740379085415e-06, "loss": 1.2744, "step": 32897 }, { "epoch": 82.245, "grad_norm": 5.84375, "learning_rate": 1.6082999867611915e-06, "loss": 1.3451, "step": 32898 }, { "epoch": 82.2475, "grad_norm": 6.5625, "learning_rate": 1.6078596494526177e-06, "loss": 1.3529, "step": 32899 }, { "epoch": 82.25, "grad_norm": 9.875, "learning_rate": 1.607419367162577e-06, "loss": 1.1821, "step": 32900 }, { "epoch": 82.2525, "grad_norm": 11.375, "learning_rate": 1.6069791398939616e-06, "loss": 1.2541, "step": 32901 }, { "epoch": 82.255, "grad_norm": 5.6875, "learning_rate": 1.6065389676496535e-06, "loss": 1.4063, "step": 32902 }, { "epoch": 82.2575, "grad_norm": 7.625, "learning_rate": 1.6060988504325414e-06, "loss": 1.1719, "step": 32903 }, { "epoch": 82.26, "grad_norm": 7.8125, "learning_rate": 1.6056587882455078e-06, "loss": 1.3139, "step": 32904 }, { "epoch": 82.2625, "grad_norm": 7.0, "learning_rate": 1.605218781091441e-06, "loss": 1.4102, "step": 32905 }, { "epoch": 82.265, "grad_norm": 5.96875, "learning_rate": 1.6047788289732236e-06, "loss": 1.3788, "step": 32906 }, { "epoch": 82.2675, "grad_norm": 6.09375, "learning_rate": 1.6043389318937385e-06, "loss": 1.2885, "step": 32907 }, { "epoch": 82.27, "grad_norm": 6.78125, "learning_rate": 1.6038990898558715e-06, "loss": 1.3425, "step": 32908 }, { "epoch": 82.2725, "grad_norm": 8.375, "learning_rate": 1.6034593028625078e-06, "loss": 1.2524, "step": 32909 }, { "epoch": 82.275, "grad_norm": 7.9375, "learning_rate": 1.6030195709165297e-06, "loss": 1.2444, "step": 32910 }, { "epoch": 82.2775, "grad_norm": 6.84375, "learning_rate": 1.6025798940208159e-06, "loss": 1.3739, "step": 32911 }, { "epoch": 82.28, "grad_norm": 12.8125, "learning_rate": 1.602140272178253e-06, "loss": 1.4009, "step": 32912 }, { "epoch": 82.2825, "grad_norm": 7.03125, "learning_rate": 1.6017007053917256e-06, "loss": 1.2617, "step": 32913 }, { "epoch": 82.285, "grad_norm": 5.15625, "learning_rate": 1.6012611936641109e-06, "loss": 1.2167, "step": 32914 }, { "epoch": 82.2875, "grad_norm": 7.25, "learning_rate": 1.6008217369982892e-06, "loss": 1.2945, "step": 32915 }, { "epoch": 82.29, "grad_norm": 6.90625, "learning_rate": 1.600382335397147e-06, "loss": 1.3415, "step": 32916 }, { "epoch": 82.2925, "grad_norm": 9.625, "learning_rate": 1.5999429888635598e-06, "loss": 1.164, "step": 32917 }, { "epoch": 82.295, "grad_norm": 5.71875, "learning_rate": 1.599503697400413e-06, "loss": 1.4856, "step": 32918 }, { "epoch": 82.2975, "grad_norm": 7.40625, "learning_rate": 1.599064461010581e-06, "loss": 1.3446, "step": 32919 }, { "epoch": 82.3, "grad_norm": 6.96875, "learning_rate": 1.5986252796969482e-06, "loss": 1.2426, "step": 32920 }, { "epoch": 82.3025, "grad_norm": 9.25, "learning_rate": 1.5981861534623922e-06, "loss": 1.3503, "step": 32921 }, { "epoch": 82.305, "grad_norm": 7.03125, "learning_rate": 1.5977470823097897e-06, "loss": 1.3953, "step": 32922 }, { "epoch": 82.3075, "grad_norm": 8.9375, "learning_rate": 1.5973080662420204e-06, "loss": 1.2611, "step": 32923 }, { "epoch": 82.31, "grad_norm": 5.3125, "learning_rate": 1.5968691052619657e-06, "loss": 1.373, "step": 32924 }, { "epoch": 82.3125, "grad_norm": 8.3125, "learning_rate": 1.5964301993725006e-06, "loss": 1.3466, "step": 32925 }, { "epoch": 82.315, "grad_norm": 10.4375, "learning_rate": 1.595991348576501e-06, "loss": 1.3376, "step": 32926 }, { "epoch": 82.3175, "grad_norm": 6.59375, "learning_rate": 1.5955525528768456e-06, "loss": 1.3571, "step": 32927 }, { "epoch": 82.32, "grad_norm": 6.5625, "learning_rate": 1.5951138122764132e-06, "loss": 1.3348, "step": 32928 }, { "epoch": 82.3225, "grad_norm": 8.1875, "learning_rate": 1.594675126778079e-06, "loss": 1.3075, "step": 32929 }, { "epoch": 82.325, "grad_norm": 8.6875, "learning_rate": 1.5942364963847158e-06, "loss": 1.181, "step": 32930 }, { "epoch": 82.3275, "grad_norm": 9.0, "learning_rate": 1.593797921099204e-06, "loss": 1.3464, "step": 32931 }, { "epoch": 82.33, "grad_norm": 5.15625, "learning_rate": 1.5933594009244147e-06, "loss": 1.1386, "step": 32932 }, { "epoch": 82.3325, "grad_norm": 7.25, "learning_rate": 1.592920935863227e-06, "loss": 1.276, "step": 32933 }, { "epoch": 82.335, "grad_norm": 7.84375, "learning_rate": 1.5924825259185094e-06, "loss": 1.1512, "step": 32934 }, { "epoch": 82.3375, "grad_norm": 6.96875, "learning_rate": 1.5920441710931444e-06, "loss": 1.2043, "step": 32935 }, { "epoch": 82.34, "grad_norm": 8.125, "learning_rate": 1.5916058713899995e-06, "loss": 1.1458, "step": 32936 }, { "epoch": 82.3425, "grad_norm": 7.84375, "learning_rate": 1.591167626811949e-06, "loss": 1.1853, "step": 32937 }, { "epoch": 82.345, "grad_norm": 6.8125, "learning_rate": 1.5907294373618665e-06, "loss": 1.0861, "step": 32938 }, { "epoch": 82.3475, "grad_norm": 11.625, "learning_rate": 1.5902913030426281e-06, "loss": 1.4467, "step": 32939 }, { "epoch": 82.35, "grad_norm": 8.5625, "learning_rate": 1.589853223857103e-06, "loss": 1.3268, "step": 32940 }, { "epoch": 82.3525, "grad_norm": 6.09375, "learning_rate": 1.5894151998081609e-06, "loss": 1.5204, "step": 32941 }, { "epoch": 82.355, "grad_norm": 11.125, "learning_rate": 1.5889772308986763e-06, "loss": 1.5514, "step": 32942 }, { "epoch": 82.3575, "grad_norm": 5.3125, "learning_rate": 1.5885393171315234e-06, "loss": 1.2341, "step": 32943 }, { "epoch": 82.36, "grad_norm": 7.3125, "learning_rate": 1.58810145850957e-06, "loss": 1.3657, "step": 32944 }, { "epoch": 82.3625, "grad_norm": 14.0, "learning_rate": 1.5876636550356838e-06, "loss": 1.2249, "step": 32945 }, { "epoch": 82.365, "grad_norm": 6.78125, "learning_rate": 1.5872259067127405e-06, "loss": 1.3314, "step": 32946 }, { "epoch": 82.3675, "grad_norm": 6.25, "learning_rate": 1.5867882135436053e-06, "loss": 1.3413, "step": 32947 }, { "epoch": 82.37, "grad_norm": 9.3125, "learning_rate": 1.5863505755311525e-06, "loss": 1.3107, "step": 32948 }, { "epoch": 82.3725, "grad_norm": 11.25, "learning_rate": 1.5859129926782458e-06, "loss": 1.4401, "step": 32949 }, { "epoch": 82.375, "grad_norm": 9.9375, "learning_rate": 1.58547546498776e-06, "loss": 1.2654, "step": 32950 }, { "epoch": 82.3775, "grad_norm": 6.65625, "learning_rate": 1.5850379924625592e-06, "loss": 1.2488, "step": 32951 }, { "epoch": 82.38, "grad_norm": 5.3125, "learning_rate": 1.5846005751055116e-06, "loss": 1.2802, "step": 32952 }, { "epoch": 82.3825, "grad_norm": 10.5625, "learning_rate": 1.5841632129194852e-06, "loss": 1.215, "step": 32953 }, { "epoch": 82.385, "grad_norm": 5.25, "learning_rate": 1.5837259059073496e-06, "loss": 1.28, "step": 32954 }, { "epoch": 82.3875, "grad_norm": 9.0625, "learning_rate": 1.5832886540719706e-06, "loss": 1.3112, "step": 32955 }, { "epoch": 82.39, "grad_norm": 9.3125, "learning_rate": 1.5828514574162124e-06, "loss": 1.4013, "step": 32956 }, { "epoch": 82.3925, "grad_norm": 7.0625, "learning_rate": 1.5824143159429427e-06, "loss": 1.5287, "step": 32957 }, { "epoch": 82.395, "grad_norm": 7.34375, "learning_rate": 1.5819772296550306e-06, "loss": 1.3676, "step": 32958 }, { "epoch": 82.3975, "grad_norm": 7.96875, "learning_rate": 1.581540198555339e-06, "loss": 1.2755, "step": 32959 }, { "epoch": 82.4, "grad_norm": 6.875, "learning_rate": 1.5811032226467304e-06, "loss": 1.3026, "step": 32960 }, { "epoch": 82.4025, "grad_norm": 8.875, "learning_rate": 1.5806663019320744e-06, "loss": 1.3903, "step": 32961 }, { "epoch": 82.405, "grad_norm": 6.09375, "learning_rate": 1.5802294364142313e-06, "loss": 1.356, "step": 32962 }, { "epoch": 82.4075, "grad_norm": 8.4375, "learning_rate": 1.5797926260960695e-06, "loss": 1.231, "step": 32963 }, { "epoch": 82.41, "grad_norm": 9.0625, "learning_rate": 1.5793558709804469e-06, "loss": 1.3398, "step": 32964 }, { "epoch": 82.4125, "grad_norm": 8.6875, "learning_rate": 1.5789191710702334e-06, "loss": 1.2967, "step": 32965 }, { "epoch": 82.415, "grad_norm": 11.5625, "learning_rate": 1.578482526368289e-06, "loss": 1.4869, "step": 32966 }, { "epoch": 82.4175, "grad_norm": 7.75, "learning_rate": 1.5780459368774736e-06, "loss": 1.473, "step": 32967 }, { "epoch": 82.42, "grad_norm": 7.90625, "learning_rate": 1.5776094026006516e-06, "loss": 1.3651, "step": 32968 }, { "epoch": 82.4225, "grad_norm": 12.0, "learning_rate": 1.5771729235406885e-06, "loss": 1.4233, "step": 32969 }, { "epoch": 82.425, "grad_norm": 9.5625, "learning_rate": 1.5767364997004419e-06, "loss": 1.4557, "step": 32970 }, { "epoch": 82.4275, "grad_norm": 8.125, "learning_rate": 1.5763001310827718e-06, "loss": 1.3535, "step": 32971 }, { "epoch": 82.43, "grad_norm": 6.4375, "learning_rate": 1.5758638176905406e-06, "loss": 1.3024, "step": 32972 }, { "epoch": 82.4325, "grad_norm": 5.875, "learning_rate": 1.575427559526611e-06, "loss": 1.2871, "step": 32973 }, { "epoch": 82.435, "grad_norm": 7.25, "learning_rate": 1.5749913565938425e-06, "loss": 1.2012, "step": 32974 }, { "epoch": 82.4375, "grad_norm": 6.09375, "learning_rate": 1.5745552088950899e-06, "loss": 1.2742, "step": 32975 }, { "epoch": 82.44, "grad_norm": 6.375, "learning_rate": 1.5741191164332192e-06, "loss": 1.3061, "step": 32976 }, { "epoch": 82.4425, "grad_norm": 8.0, "learning_rate": 1.5736830792110846e-06, "loss": 1.3552, "step": 32977 }, { "epoch": 82.445, "grad_norm": 5.5, "learning_rate": 1.5732470972315471e-06, "loss": 1.3043, "step": 32978 }, { "epoch": 82.4475, "grad_norm": 7.5, "learning_rate": 1.5728111704974636e-06, "loss": 1.2119, "step": 32979 }, { "epoch": 82.45, "grad_norm": 11.25, "learning_rate": 1.5723752990116948e-06, "loss": 1.4718, "step": 32980 }, { "epoch": 82.4525, "grad_norm": 7.75, "learning_rate": 1.5719394827770962e-06, "loss": 1.1619, "step": 32981 }, { "epoch": 82.455, "grad_norm": 9.4375, "learning_rate": 1.5715037217965235e-06, "loss": 1.3284, "step": 32982 }, { "epoch": 82.4575, "grad_norm": 9.0625, "learning_rate": 1.5710680160728341e-06, "loss": 1.2512, "step": 32983 }, { "epoch": 82.46, "grad_norm": 8.4375, "learning_rate": 1.5706323656088873e-06, "loss": 1.2979, "step": 32984 }, { "epoch": 82.4625, "grad_norm": 7.71875, "learning_rate": 1.5701967704075382e-06, "loss": 1.4664, "step": 32985 }, { "epoch": 82.465, "grad_norm": 9.0, "learning_rate": 1.5697612304716404e-06, "loss": 1.5478, "step": 32986 }, { "epoch": 82.4675, "grad_norm": 9.125, "learning_rate": 1.569325745804049e-06, "loss": 1.3648, "step": 32987 }, { "epoch": 82.47, "grad_norm": 9.6875, "learning_rate": 1.5688903164076242e-06, "loss": 1.34, "step": 32988 }, { "epoch": 82.4725, "grad_norm": 6.0625, "learning_rate": 1.5684549422852158e-06, "loss": 1.3644, "step": 32989 }, { "epoch": 82.475, "grad_norm": 9.0625, "learning_rate": 1.5680196234396783e-06, "loss": 1.4209, "step": 32990 }, { "epoch": 82.4775, "grad_norm": 8.0, "learning_rate": 1.5675843598738682e-06, "loss": 1.29, "step": 32991 }, { "epoch": 82.48, "grad_norm": 6.65625, "learning_rate": 1.5671491515906355e-06, "loss": 1.3133, "step": 32992 }, { "epoch": 82.4825, "grad_norm": 8.3125, "learning_rate": 1.566713998592838e-06, "loss": 1.2398, "step": 32993 }, { "epoch": 82.485, "grad_norm": 7.53125, "learning_rate": 1.5662789008833235e-06, "loss": 1.3107, "step": 32994 }, { "epoch": 82.4875, "grad_norm": 8.8125, "learning_rate": 1.5658438584649493e-06, "loss": 1.4041, "step": 32995 }, { "epoch": 82.49, "grad_norm": 7.375, "learning_rate": 1.5654088713405646e-06, "loss": 1.4843, "step": 32996 }, { "epoch": 82.4925, "grad_norm": 6.3125, "learning_rate": 1.5649739395130204e-06, "loss": 1.2891, "step": 32997 }, { "epoch": 82.495, "grad_norm": 8.5625, "learning_rate": 1.5645390629851699e-06, "loss": 1.3254, "step": 32998 }, { "epoch": 82.4975, "grad_norm": 5.625, "learning_rate": 1.564104241759865e-06, "loss": 1.3163, "step": 32999 }, { "epoch": 82.5, "grad_norm": 7.78125, "learning_rate": 1.5636694758399563e-06, "loss": 1.3903, "step": 33000 }, { "epoch": 82.5025, "grad_norm": 7.25, "learning_rate": 1.563234765228292e-06, "loss": 1.1897, "step": 33001 }, { "epoch": 82.505, "grad_norm": 10.875, "learning_rate": 1.5628001099277179e-06, "loss": 1.4544, "step": 33002 }, { "epoch": 82.5075, "grad_norm": 10.875, "learning_rate": 1.5623655099410938e-06, "loss": 1.4459, "step": 33003 }, { "epoch": 82.51, "grad_norm": 6.21875, "learning_rate": 1.5619309652712644e-06, "loss": 1.3039, "step": 33004 }, { "epoch": 82.5125, "grad_norm": 8.8125, "learning_rate": 1.5614964759210749e-06, "loss": 1.4203, "step": 33005 }, { "epoch": 82.515, "grad_norm": 6.71875, "learning_rate": 1.5610620418933796e-06, "loss": 1.4351, "step": 33006 }, { "epoch": 82.5175, "grad_norm": 7.875, "learning_rate": 1.5606276631910212e-06, "loss": 1.3595, "step": 33007 }, { "epoch": 82.52, "grad_norm": 9.4375, "learning_rate": 1.5601933398168523e-06, "loss": 1.1963, "step": 33008 }, { "epoch": 82.5225, "grad_norm": 7.0, "learning_rate": 1.5597590717737165e-06, "loss": 1.3984, "step": 33009 }, { "epoch": 82.525, "grad_norm": 8.9375, "learning_rate": 1.5593248590644638e-06, "loss": 1.2457, "step": 33010 }, { "epoch": 82.5275, "grad_norm": 10.0, "learning_rate": 1.5588907016919396e-06, "loss": 1.3312, "step": 33011 }, { "epoch": 82.53, "grad_norm": 7.5, "learning_rate": 1.5584565996589884e-06, "loss": 1.4688, "step": 33012 }, { "epoch": 82.5325, "grad_norm": 6.90625, "learning_rate": 1.5580225529684579e-06, "loss": 1.2866, "step": 33013 }, { "epoch": 82.535, "grad_norm": 8.9375, "learning_rate": 1.5575885616231967e-06, "loss": 1.4063, "step": 33014 }, { "epoch": 82.5375, "grad_norm": 7.3125, "learning_rate": 1.5571546256260462e-06, "loss": 1.2938, "step": 33015 }, { "epoch": 82.54, "grad_norm": 5.84375, "learning_rate": 1.5567207449798517e-06, "loss": 1.2794, "step": 33016 }, { "epoch": 82.5425, "grad_norm": 8.4375, "learning_rate": 1.5562869196874542e-06, "loss": 1.2837, "step": 33017 }, { "epoch": 82.545, "grad_norm": 8.125, "learning_rate": 1.555853149751706e-06, "loss": 1.2756, "step": 33018 }, { "epoch": 82.5475, "grad_norm": 7.0625, "learning_rate": 1.555419435175447e-06, "loss": 1.2747, "step": 33019 }, { "epoch": 82.55, "grad_norm": 8.8125, "learning_rate": 1.5549857759615195e-06, "loss": 1.3865, "step": 33020 }, { "epoch": 82.5525, "grad_norm": 8.125, "learning_rate": 1.5545521721127653e-06, "loss": 1.3468, "step": 33021 }, { "epoch": 82.555, "grad_norm": 6.96875, "learning_rate": 1.5541186236320293e-06, "loss": 1.3454, "step": 33022 }, { "epoch": 82.5575, "grad_norm": 6.6875, "learning_rate": 1.5536851305221567e-06, "loss": 1.2367, "step": 33023 }, { "epoch": 82.56, "grad_norm": 6.25, "learning_rate": 1.5532516927859853e-06, "loss": 1.3104, "step": 33024 }, { "epoch": 82.5625, "grad_norm": 7.65625, "learning_rate": 1.552818310426356e-06, "loss": 1.2366, "step": 33025 }, { "epoch": 82.565, "grad_norm": 7.25, "learning_rate": 1.5523849834461136e-06, "loss": 1.348, "step": 33026 }, { "epoch": 82.5675, "grad_norm": 8.0, "learning_rate": 1.5519517118480953e-06, "loss": 1.5033, "step": 33027 }, { "epoch": 82.57, "grad_norm": 10.5625, "learning_rate": 1.551518495635146e-06, "loss": 1.2349, "step": 33028 }, { "epoch": 82.5725, "grad_norm": 7.3125, "learning_rate": 1.5510853348101006e-06, "loss": 1.33, "step": 33029 }, { "epoch": 82.575, "grad_norm": 8.5, "learning_rate": 1.5506522293758042e-06, "loss": 1.2311, "step": 33030 }, { "epoch": 82.5775, "grad_norm": 11.9375, "learning_rate": 1.550219179335093e-06, "loss": 1.3353, "step": 33031 }, { "epoch": 82.58, "grad_norm": 7.0625, "learning_rate": 1.5497861846908024e-06, "loss": 1.4151, "step": 33032 }, { "epoch": 82.5825, "grad_norm": 6.625, "learning_rate": 1.5493532454457804e-06, "loss": 1.2629, "step": 33033 }, { "epoch": 82.585, "grad_norm": 6.96875, "learning_rate": 1.5489203616028602e-06, "loss": 1.5323, "step": 33034 }, { "epoch": 82.5875, "grad_norm": 6.15625, "learning_rate": 1.5484875331648807e-06, "loss": 1.5103, "step": 33035 }, { "epoch": 82.59, "grad_norm": 7.28125, "learning_rate": 1.5480547601346751e-06, "loss": 1.1648, "step": 33036 }, { "epoch": 82.5925, "grad_norm": 10.8125, "learning_rate": 1.5476220425150855e-06, "loss": 1.4141, "step": 33037 }, { "epoch": 82.595, "grad_norm": 8.625, "learning_rate": 1.547189380308949e-06, "loss": 1.1861, "step": 33038 }, { "epoch": 82.5975, "grad_norm": 8.8125, "learning_rate": 1.5467567735191013e-06, "loss": 1.3335, "step": 33039 }, { "epoch": 82.6, "grad_norm": 7.15625, "learning_rate": 1.5463242221483742e-06, "loss": 1.3693, "step": 33040 }, { "epoch": 82.6025, "grad_norm": 7.6875, "learning_rate": 1.5458917261996109e-06, "loss": 1.2326, "step": 33041 }, { "epoch": 82.605, "grad_norm": 6.625, "learning_rate": 1.5454592856756401e-06, "loss": 1.3692, "step": 33042 }, { "epoch": 82.6075, "grad_norm": 9.375, "learning_rate": 1.5450269005793028e-06, "loss": 1.3913, "step": 33043 }, { "epoch": 82.61, "grad_norm": 13.0625, "learning_rate": 1.544594570913428e-06, "loss": 1.2229, "step": 33044 }, { "epoch": 82.6125, "grad_norm": 7.9375, "learning_rate": 1.5441622966808557e-06, "loss": 1.1806, "step": 33045 }, { "epoch": 82.615, "grad_norm": 6.9375, "learning_rate": 1.5437300778844166e-06, "loss": 1.3741, "step": 33046 }, { "epoch": 82.6175, "grad_norm": 7.6875, "learning_rate": 1.5432979145269421e-06, "loss": 1.2404, "step": 33047 }, { "epoch": 82.62, "grad_norm": 7.1875, "learning_rate": 1.5428658066112689e-06, "loss": 1.3267, "step": 33048 }, { "epoch": 82.6225, "grad_norm": 7.15625, "learning_rate": 1.5424337541402302e-06, "loss": 1.3418, "step": 33049 }, { "epoch": 82.625, "grad_norm": 9.375, "learning_rate": 1.542001757116658e-06, "loss": 1.3299, "step": 33050 }, { "epoch": 82.6275, "grad_norm": 7.09375, "learning_rate": 1.5415698155433812e-06, "loss": 1.2167, "step": 33051 }, { "epoch": 82.63, "grad_norm": 7.03125, "learning_rate": 1.5411379294232355e-06, "loss": 1.3524, "step": 33052 }, { "epoch": 82.6325, "grad_norm": 9.625, "learning_rate": 1.5407060987590517e-06, "loss": 1.3324, "step": 33053 }, { "epoch": 82.635, "grad_norm": 8.5, "learning_rate": 1.540274323553661e-06, "loss": 1.2735, "step": 33054 }, { "epoch": 82.6375, "grad_norm": 8.875, "learning_rate": 1.53984260380989e-06, "loss": 1.4914, "step": 33055 }, { "epoch": 82.64, "grad_norm": 8.125, "learning_rate": 1.5394109395305757e-06, "loss": 1.2952, "step": 33056 }, { "epoch": 82.6425, "grad_norm": 5.5, "learning_rate": 1.5389793307185418e-06, "loss": 1.3948, "step": 33057 }, { "epoch": 82.645, "grad_norm": 6.4375, "learning_rate": 1.5385477773766245e-06, "loss": 1.2899, "step": 33058 }, { "epoch": 82.6475, "grad_norm": 9.0625, "learning_rate": 1.5381162795076455e-06, "loss": 1.1574, "step": 33059 }, { "epoch": 82.65, "grad_norm": 7.0, "learning_rate": 1.5376848371144404e-06, "loss": 1.392, "step": 33060 }, { "epoch": 82.6525, "grad_norm": 8.375, "learning_rate": 1.5372534501998348e-06, "loss": 1.3007, "step": 33061 }, { "epoch": 82.655, "grad_norm": 9.1875, "learning_rate": 1.5368221187666544e-06, "loss": 1.1868, "step": 33062 }, { "epoch": 82.6575, "grad_norm": 8.625, "learning_rate": 1.5363908428177299e-06, "loss": 1.4481, "step": 33063 }, { "epoch": 82.66, "grad_norm": 10.3125, "learning_rate": 1.5359596223558903e-06, "loss": 1.3837, "step": 33064 }, { "epoch": 82.6625, "grad_norm": 8.0625, "learning_rate": 1.5355284573839613e-06, "loss": 1.2272, "step": 33065 }, { "epoch": 82.665, "grad_norm": 8.25, "learning_rate": 1.535097347904766e-06, "loss": 1.2229, "step": 33066 }, { "epoch": 82.6675, "grad_norm": 8.875, "learning_rate": 1.5346662939211343e-06, "loss": 1.5296, "step": 33067 }, { "epoch": 82.67, "grad_norm": 8.875, "learning_rate": 1.534235295435894e-06, "loss": 1.3984, "step": 33068 }, { "epoch": 82.6725, "grad_norm": 7.53125, "learning_rate": 1.5338043524518675e-06, "loss": 1.3102, "step": 33069 }, { "epoch": 82.675, "grad_norm": 13.25, "learning_rate": 1.533373464971879e-06, "loss": 1.3282, "step": 33070 }, { "epoch": 82.6775, "grad_norm": 11.0, "learning_rate": 1.5329426329987573e-06, "loss": 1.264, "step": 33071 }, { "epoch": 82.68, "grad_norm": 7.9375, "learning_rate": 1.5325118565353237e-06, "loss": 1.2452, "step": 33072 }, { "epoch": 82.6825, "grad_norm": 8.625, "learning_rate": 1.5320811355844044e-06, "loss": 1.3252, "step": 33073 }, { "epoch": 82.685, "grad_norm": 7.09375, "learning_rate": 1.5316504701488211e-06, "loss": 1.3239, "step": 33074 }, { "epoch": 82.6875, "grad_norm": 6.375, "learning_rate": 1.5312198602314e-06, "loss": 1.2658, "step": 33075 }, { "epoch": 82.69, "grad_norm": 6.875, "learning_rate": 1.5307893058349632e-06, "loss": 1.202, "step": 33076 }, { "epoch": 82.6925, "grad_norm": 9.25, "learning_rate": 1.5303588069623299e-06, "loss": 1.2686, "step": 33077 }, { "epoch": 82.695, "grad_norm": 7.3125, "learning_rate": 1.5299283636163253e-06, "loss": 1.3884, "step": 33078 }, { "epoch": 82.6975, "grad_norm": 6.375, "learning_rate": 1.5294979757997741e-06, "loss": 1.4156, "step": 33079 }, { "epoch": 82.7, "grad_norm": 5.46875, "learning_rate": 1.5290676435154949e-06, "loss": 1.5398, "step": 33080 }, { "epoch": 82.7025, "grad_norm": 7.15625, "learning_rate": 1.5286373667663068e-06, "loss": 1.1534, "step": 33081 }, { "epoch": 82.705, "grad_norm": 5.78125, "learning_rate": 1.528207145555034e-06, "loss": 1.3344, "step": 33082 }, { "epoch": 82.7075, "grad_norm": 9.5625, "learning_rate": 1.5277769798844977e-06, "loss": 1.2996, "step": 33083 }, { "epoch": 82.71, "grad_norm": 8.0625, "learning_rate": 1.5273468697575167e-06, "loss": 1.312, "step": 33084 }, { "epoch": 82.7125, "grad_norm": 8.3125, "learning_rate": 1.5269168151769075e-06, "loss": 1.3642, "step": 33085 }, { "epoch": 82.715, "grad_norm": 9.6875, "learning_rate": 1.526486816145496e-06, "loss": 1.3745, "step": 33086 }, { "epoch": 82.7175, "grad_norm": 5.90625, "learning_rate": 1.5260568726660952e-06, "loss": 1.1993, "step": 33087 }, { "epoch": 82.72, "grad_norm": 7.90625, "learning_rate": 1.5256269847415283e-06, "loss": 1.323, "step": 33088 }, { "epoch": 82.7225, "grad_norm": 10.375, "learning_rate": 1.5251971523746101e-06, "loss": 1.346, "step": 33089 }, { "epoch": 82.725, "grad_norm": 7.59375, "learning_rate": 1.5247673755681625e-06, "loss": 1.3591, "step": 33090 }, { "epoch": 82.7275, "grad_norm": 5.75, "learning_rate": 1.5243376543249976e-06, "loss": 1.3923, "step": 33091 }, { "epoch": 82.73, "grad_norm": 6.59375, "learning_rate": 1.523907988647939e-06, "loss": 1.24, "step": 33092 }, { "epoch": 82.7325, "grad_norm": 7.34375, "learning_rate": 1.5234783785397978e-06, "loss": 1.3014, "step": 33093 }, { "epoch": 82.735, "grad_norm": 5.28125, "learning_rate": 1.523048824003396e-06, "loss": 1.3746, "step": 33094 }, { "epoch": 82.7375, "grad_norm": 9.0, "learning_rate": 1.5226193250415467e-06, "loss": 1.1781, "step": 33095 }, { "epoch": 82.74, "grad_norm": 11.3125, "learning_rate": 1.5221898816570623e-06, "loss": 1.2706, "step": 33096 }, { "epoch": 82.7425, "grad_norm": 8.9375, "learning_rate": 1.521760493852763e-06, "loss": 1.3401, "step": 33097 }, { "epoch": 82.745, "grad_norm": 14.375, "learning_rate": 1.5213311616314652e-06, "loss": 1.2867, "step": 33098 }, { "epoch": 82.7475, "grad_norm": 9.875, "learning_rate": 1.52090188499598e-06, "loss": 1.3047, "step": 33099 }, { "epoch": 82.75, "grad_norm": 7.96875, "learning_rate": 1.520472663949122e-06, "loss": 1.4652, "step": 33100 }, { "epoch": 82.7525, "grad_norm": 8.75, "learning_rate": 1.520043498493704e-06, "loss": 1.2176, "step": 33101 }, { "epoch": 82.755, "grad_norm": 7.40625, "learning_rate": 1.5196143886325442e-06, "loss": 1.3597, "step": 33102 }, { "epoch": 82.7575, "grad_norm": 7.34375, "learning_rate": 1.5191853343684538e-06, "loss": 1.2815, "step": 33103 }, { "epoch": 82.76, "grad_norm": 6.375, "learning_rate": 1.5187563357042423e-06, "loss": 1.2565, "step": 33104 }, { "epoch": 82.7625, "grad_norm": 11.0625, "learning_rate": 1.5183273926427277e-06, "loss": 1.3145, "step": 33105 }, { "epoch": 82.765, "grad_norm": 12.625, "learning_rate": 1.5178985051867169e-06, "loss": 1.3893, "step": 33106 }, { "epoch": 82.7675, "grad_norm": 7.375, "learning_rate": 1.5174696733390259e-06, "loss": 1.524, "step": 33107 }, { "epoch": 82.77, "grad_norm": 6.59375, "learning_rate": 1.517040897102462e-06, "loss": 1.2284, "step": 33108 }, { "epoch": 82.7725, "grad_norm": 8.3125, "learning_rate": 1.5166121764798403e-06, "loss": 1.4544, "step": 33109 }, { "epoch": 82.775, "grad_norm": 8.375, "learning_rate": 1.5161835114739708e-06, "loss": 1.3849, "step": 33110 }, { "epoch": 82.7775, "grad_norm": 7.21875, "learning_rate": 1.5157549020876595e-06, "loss": 1.3034, "step": 33111 }, { "epoch": 82.78, "grad_norm": 5.84375, "learning_rate": 1.51532634832372e-06, "loss": 1.3451, "step": 33112 }, { "epoch": 82.7825, "grad_norm": 8.9375, "learning_rate": 1.5148978501849642e-06, "loss": 1.3721, "step": 33113 }, { "epoch": 82.785, "grad_norm": 6.59375, "learning_rate": 1.5144694076741973e-06, "loss": 1.2865, "step": 33114 }, { "epoch": 82.7875, "grad_norm": 4.8125, "learning_rate": 1.514041020794228e-06, "loss": 1.4686, "step": 33115 }, { "epoch": 82.79, "grad_norm": 20.75, "learning_rate": 1.5136126895478664e-06, "loss": 1.3952, "step": 33116 }, { "epoch": 82.7925, "grad_norm": 7.8125, "learning_rate": 1.5131844139379216e-06, "loss": 1.2671, "step": 33117 }, { "epoch": 82.795, "grad_norm": 7.8125, "learning_rate": 1.5127561939672008e-06, "loss": 1.4292, "step": 33118 }, { "epoch": 82.7975, "grad_norm": 6.625, "learning_rate": 1.512328029638509e-06, "loss": 1.2218, "step": 33119 }, { "epoch": 82.8, "grad_norm": 7.8125, "learning_rate": 1.511899920954656e-06, "loss": 1.2377, "step": 33120 }, { "epoch": 82.8025, "grad_norm": 5.78125, "learning_rate": 1.5114718679184458e-06, "loss": 1.377, "step": 33121 }, { "epoch": 82.805, "grad_norm": 10.9375, "learning_rate": 1.511043870532689e-06, "loss": 1.3913, "step": 33122 }, { "epoch": 82.8075, "grad_norm": 8.25, "learning_rate": 1.5106159288001865e-06, "loss": 1.2821, "step": 33123 }, { "epoch": 82.81, "grad_norm": 11.0, "learning_rate": 1.5101880427237481e-06, "loss": 1.3621, "step": 33124 }, { "epoch": 82.8125, "grad_norm": 8.8125, "learning_rate": 1.5097602123061772e-06, "loss": 1.2265, "step": 33125 }, { "epoch": 82.815, "grad_norm": 7.65625, "learning_rate": 1.5093324375502772e-06, "loss": 1.2547, "step": 33126 }, { "epoch": 82.8175, "grad_norm": 6.59375, "learning_rate": 1.5089047184588524e-06, "loss": 1.3147, "step": 33127 }, { "epoch": 82.82, "grad_norm": 7.75, "learning_rate": 1.5084770550347115e-06, "loss": 1.4532, "step": 33128 }, { "epoch": 82.8225, "grad_norm": 12.0625, "learning_rate": 1.5080494472806562e-06, "loss": 1.517, "step": 33129 }, { "epoch": 82.825, "grad_norm": 8.5625, "learning_rate": 1.5076218951994859e-06, "loss": 1.444, "step": 33130 }, { "epoch": 82.8275, "grad_norm": 6.21875, "learning_rate": 1.5071943987940063e-06, "loss": 1.459, "step": 33131 }, { "epoch": 82.83, "grad_norm": 6.84375, "learning_rate": 1.5067669580670218e-06, "loss": 1.4539, "step": 33132 }, { "epoch": 82.8325, "grad_norm": 7.46875, "learning_rate": 1.5063395730213348e-06, "loss": 1.2629, "step": 33133 }, { "epoch": 82.835, "grad_norm": 9.3125, "learning_rate": 1.5059122436597428e-06, "loss": 1.3094, "step": 33134 }, { "epoch": 82.8375, "grad_norm": 7.1875, "learning_rate": 1.5054849699850528e-06, "loss": 1.3682, "step": 33135 }, { "epoch": 82.84, "grad_norm": 10.4375, "learning_rate": 1.5050577520000608e-06, "loss": 1.2768, "step": 33136 }, { "epoch": 82.8425, "grad_norm": 7.09375, "learning_rate": 1.504630589707572e-06, "loss": 1.2607, "step": 33137 }, { "epoch": 82.845, "grad_norm": 13.0625, "learning_rate": 1.5042034831103836e-06, "loss": 1.5704, "step": 33138 }, { "epoch": 82.8475, "grad_norm": 7.15625, "learning_rate": 1.503776432211299e-06, "loss": 1.1946, "step": 33139 }, { "epoch": 82.85, "grad_norm": 26.875, "learning_rate": 1.5033494370131162e-06, "loss": 1.3249, "step": 33140 }, { "epoch": 82.8525, "grad_norm": 6.1875, "learning_rate": 1.502922497518633e-06, "loss": 1.2657, "step": 33141 }, { "epoch": 82.855, "grad_norm": 5.21875, "learning_rate": 1.502495613730649e-06, "loss": 1.2701, "step": 33142 }, { "epoch": 82.8575, "grad_norm": 6.6875, "learning_rate": 1.5020687856519656e-06, "loss": 1.3067, "step": 33143 }, { "epoch": 82.86, "grad_norm": 10.4375, "learning_rate": 1.5016420132853804e-06, "loss": 1.4111, "step": 33144 }, { "epoch": 82.8625, "grad_norm": 8.6875, "learning_rate": 1.5012152966336869e-06, "loss": 1.2561, "step": 33145 }, { "epoch": 82.865, "grad_norm": 8.9375, "learning_rate": 1.500788635699686e-06, "loss": 1.2059, "step": 33146 }, { "epoch": 82.8675, "grad_norm": 7.03125, "learning_rate": 1.5003620304861766e-06, "loss": 1.4842, "step": 33147 }, { "epoch": 82.87, "grad_norm": 6.875, "learning_rate": 1.4999354809959542e-06, "loss": 1.302, "step": 33148 }, { "epoch": 82.8725, "grad_norm": 8.0, "learning_rate": 1.4995089872318125e-06, "loss": 1.2638, "step": 33149 }, { "epoch": 82.875, "grad_norm": 8.625, "learning_rate": 1.4990825491965522e-06, "loss": 1.3421, "step": 33150 }, { "epoch": 82.8775, "grad_norm": 7.53125, "learning_rate": 1.4986561668929633e-06, "loss": 1.3036, "step": 33151 }, { "epoch": 82.88, "grad_norm": 6.90625, "learning_rate": 1.498229840323847e-06, "loss": 1.272, "step": 33152 }, { "epoch": 82.8825, "grad_norm": 8.5625, "learning_rate": 1.4978035694919934e-06, "loss": 1.2224, "step": 33153 }, { "epoch": 82.885, "grad_norm": 7.625, "learning_rate": 1.4973773544002013e-06, "loss": 1.3374, "step": 33154 }, { "epoch": 82.8875, "grad_norm": 8.125, "learning_rate": 1.4969511950512638e-06, "loss": 1.2683, "step": 33155 }, { "epoch": 82.89, "grad_norm": 8.75, "learning_rate": 1.4965250914479712e-06, "loss": 1.2775, "step": 33156 }, { "epoch": 82.8925, "grad_norm": 6.0, "learning_rate": 1.49609904359312e-06, "loss": 1.1931, "step": 33157 }, { "epoch": 82.895, "grad_norm": 5.9375, "learning_rate": 1.4956730514895047e-06, "loss": 1.3626, "step": 33158 }, { "epoch": 82.8975, "grad_norm": 7.84375, "learning_rate": 1.4952471151399174e-06, "loss": 1.3838, "step": 33159 }, { "epoch": 82.9, "grad_norm": 6.5, "learning_rate": 1.4948212345471492e-06, "loss": 1.2277, "step": 33160 }, { "epoch": 82.9025, "grad_norm": 6.8125, "learning_rate": 1.494395409713989e-06, "loss": 1.3259, "step": 33161 }, { "epoch": 82.905, "grad_norm": 7.625, "learning_rate": 1.4939696406432358e-06, "loss": 1.4824, "step": 33162 }, { "epoch": 82.9075, "grad_norm": 8.0625, "learning_rate": 1.4935439273376773e-06, "loss": 1.4442, "step": 33163 }, { "epoch": 82.91, "grad_norm": 7.0625, "learning_rate": 1.4931182698001046e-06, "loss": 1.3182, "step": 33164 }, { "epoch": 82.9125, "grad_norm": 4.65625, "learning_rate": 1.4926926680333053e-06, "loss": 1.2649, "step": 33165 }, { "epoch": 82.915, "grad_norm": 5.0, "learning_rate": 1.492267122040073e-06, "loss": 1.2174, "step": 33166 }, { "epoch": 82.9175, "grad_norm": 6.3125, "learning_rate": 1.4918416318231998e-06, "loss": 1.4046, "step": 33167 }, { "epoch": 82.92, "grad_norm": 8.375, "learning_rate": 1.4914161973854714e-06, "loss": 1.4011, "step": 33168 }, { "epoch": 82.9225, "grad_norm": 7.59375, "learning_rate": 1.4909908187296763e-06, "loss": 1.4644, "step": 33169 }, { "epoch": 82.925, "grad_norm": 8.6875, "learning_rate": 1.4905654958586069e-06, "loss": 1.4493, "step": 33170 }, { "epoch": 82.9275, "grad_norm": 9.1875, "learning_rate": 1.4901402287750465e-06, "loss": 1.0944, "step": 33171 }, { "epoch": 82.93, "grad_norm": 6.75, "learning_rate": 1.4897150174817875e-06, "loss": 1.2808, "step": 33172 }, { "epoch": 82.9325, "grad_norm": 7.65625, "learning_rate": 1.4892898619816176e-06, "loss": 1.3457, "step": 33173 }, { "epoch": 82.935, "grad_norm": 7.53125, "learning_rate": 1.4888647622773234e-06, "loss": 1.3162, "step": 33174 }, { "epoch": 82.9375, "grad_norm": 8.25, "learning_rate": 1.4884397183716902e-06, "loss": 1.2189, "step": 33175 }, { "epoch": 82.94, "grad_norm": 6.53125, "learning_rate": 1.488014730267502e-06, "loss": 1.3318, "step": 33176 }, { "epoch": 82.9425, "grad_norm": 6.25, "learning_rate": 1.4875897979675536e-06, "loss": 1.2611, "step": 33177 }, { "epoch": 82.945, "grad_norm": 9.0625, "learning_rate": 1.4871649214746252e-06, "loss": 1.316, "step": 33178 }, { "epoch": 82.9475, "grad_norm": 6.09375, "learning_rate": 1.4867401007915028e-06, "loss": 1.4453, "step": 33179 }, { "epoch": 82.95, "grad_norm": 6.875, "learning_rate": 1.4863153359209693e-06, "loss": 1.3594, "step": 33180 }, { "epoch": 82.9525, "grad_norm": 6.71875, "learning_rate": 1.4858906268658124e-06, "loss": 1.1563, "step": 33181 }, { "epoch": 82.955, "grad_norm": 8.9375, "learning_rate": 1.485465973628818e-06, "loss": 1.3945, "step": 33182 }, { "epoch": 82.9575, "grad_norm": 9.125, "learning_rate": 1.485041376212767e-06, "loss": 1.364, "step": 33183 }, { "epoch": 82.96, "grad_norm": 7.5625, "learning_rate": 1.4846168346204425e-06, "loss": 1.3427, "step": 33184 }, { "epoch": 82.9625, "grad_norm": 6.4375, "learning_rate": 1.4841923488546318e-06, "loss": 1.2987, "step": 33185 }, { "epoch": 82.965, "grad_norm": 6.9375, "learning_rate": 1.4837679189181121e-06, "loss": 1.3414, "step": 33186 }, { "epoch": 82.9675, "grad_norm": 4.3125, "learning_rate": 1.4833435448136713e-06, "loss": 1.3434, "step": 33187 }, { "epoch": 82.97, "grad_norm": 12.6875, "learning_rate": 1.4829192265440883e-06, "loss": 1.2073, "step": 33188 }, { "epoch": 82.9725, "grad_norm": 7.65625, "learning_rate": 1.4824949641121477e-06, "loss": 1.4411, "step": 33189 }, { "epoch": 82.975, "grad_norm": 9.25, "learning_rate": 1.4820707575206295e-06, "loss": 1.2528, "step": 33190 }, { "epoch": 82.9775, "grad_norm": 7.09375, "learning_rate": 1.4816466067723113e-06, "loss": 1.2242, "step": 33191 }, { "epoch": 82.98, "grad_norm": 8.0625, "learning_rate": 1.4812225118699786e-06, "loss": 1.4861, "step": 33192 }, { "epoch": 82.9825, "grad_norm": 6.90625, "learning_rate": 1.4807984728164115e-06, "loss": 1.467, "step": 33193 }, { "epoch": 82.985, "grad_norm": 8.625, "learning_rate": 1.480374489614389e-06, "loss": 1.3263, "step": 33194 }, { "epoch": 82.9875, "grad_norm": 6.40625, "learning_rate": 1.4799505622666877e-06, "loss": 1.2988, "step": 33195 }, { "epoch": 82.99, "grad_norm": 10.25, "learning_rate": 1.4795266907760897e-06, "loss": 1.3863, "step": 33196 }, { "epoch": 82.9925, "grad_norm": 8.4375, "learning_rate": 1.4791028751453762e-06, "loss": 1.2041, "step": 33197 }, { "epoch": 82.995, "grad_norm": 7.28125, "learning_rate": 1.4786791153773228e-06, "loss": 1.3022, "step": 33198 }, { "epoch": 82.9975, "grad_norm": 7.15625, "learning_rate": 1.478255411474706e-06, "loss": 1.2047, "step": 33199 }, { "epoch": 83.0, "grad_norm": 7.40625, "learning_rate": 1.4778317634403082e-06, "loss": 1.4802, "step": 33200 }, { "epoch": 83.0025, "grad_norm": 5.625, "learning_rate": 1.4774081712769028e-06, "loss": 1.355, "step": 33201 }, { "epoch": 83.005, "grad_norm": 4.96875, "learning_rate": 1.4769846349872696e-06, "loss": 1.2552, "step": 33202 }, { "epoch": 83.0075, "grad_norm": 7.75, "learning_rate": 1.4765611545741821e-06, "loss": 1.2425, "step": 33203 }, { "epoch": 83.01, "grad_norm": 7.90625, "learning_rate": 1.476137730040421e-06, "loss": 1.3486, "step": 33204 }, { "epoch": 83.0125, "grad_norm": 6.28125, "learning_rate": 1.4757143613887604e-06, "loss": 1.2586, "step": 33205 }, { "epoch": 83.015, "grad_norm": 10.0625, "learning_rate": 1.4752910486219719e-06, "loss": 1.3519, "step": 33206 }, { "epoch": 83.0175, "grad_norm": 8.5, "learning_rate": 1.4748677917428346e-06, "loss": 1.5218, "step": 33207 }, { "epoch": 83.02, "grad_norm": 8.375, "learning_rate": 1.4744445907541249e-06, "loss": 1.4432, "step": 33208 }, { "epoch": 83.0225, "grad_norm": 8.25, "learning_rate": 1.4740214456586155e-06, "loss": 1.1669, "step": 33209 }, { "epoch": 83.025, "grad_norm": 11.1875, "learning_rate": 1.4735983564590784e-06, "loss": 1.3621, "step": 33210 }, { "epoch": 83.0275, "grad_norm": 6.59375, "learning_rate": 1.473175323158289e-06, "loss": 1.2615, "step": 33211 }, { "epoch": 83.03, "grad_norm": 6.125, "learning_rate": 1.4727523457590232e-06, "loss": 1.3679, "step": 33212 }, { "epoch": 83.0325, "grad_norm": 8.0625, "learning_rate": 1.4723294242640518e-06, "loss": 1.475, "step": 33213 }, { "epoch": 83.035, "grad_norm": 6.375, "learning_rate": 1.471906558676145e-06, "loss": 1.3275, "step": 33214 }, { "epoch": 83.0375, "grad_norm": 7.875, "learning_rate": 1.4714837489980805e-06, "loss": 1.4704, "step": 33215 }, { "epoch": 83.04, "grad_norm": 6.5625, "learning_rate": 1.4710609952326239e-06, "loss": 1.4262, "step": 33216 }, { "epoch": 83.0425, "grad_norm": 8.1875, "learning_rate": 1.4706382973825528e-06, "loss": 1.5335, "step": 33217 }, { "epoch": 83.045, "grad_norm": 8.375, "learning_rate": 1.470215655450634e-06, "loss": 1.2322, "step": 33218 }, { "epoch": 83.0475, "grad_norm": 7.21875, "learning_rate": 1.4697930694396413e-06, "loss": 1.3013, "step": 33219 }, { "epoch": 83.05, "grad_norm": 8.375, "learning_rate": 1.469370539352345e-06, "loss": 1.3292, "step": 33220 }, { "epoch": 83.0525, "grad_norm": 7.875, "learning_rate": 1.4689480651915112e-06, "loss": 1.6344, "step": 33221 }, { "epoch": 83.055, "grad_norm": 7.28125, "learning_rate": 1.4685256469599118e-06, "loss": 1.3187, "step": 33222 }, { "epoch": 83.0575, "grad_norm": 11.8125, "learning_rate": 1.4681032846603204e-06, "loss": 1.2272, "step": 33223 }, { "epoch": 83.06, "grad_norm": 8.875, "learning_rate": 1.4676809782955014e-06, "loss": 1.3525, "step": 33224 }, { "epoch": 83.0625, "grad_norm": 5.3125, "learning_rate": 1.4672587278682228e-06, "loss": 1.2172, "step": 33225 }, { "epoch": 83.065, "grad_norm": 9.0, "learning_rate": 1.4668365333812552e-06, "loss": 1.2266, "step": 33226 }, { "epoch": 83.0675, "grad_norm": 6.46875, "learning_rate": 1.4664143948373666e-06, "loss": 1.1952, "step": 33227 }, { "epoch": 83.07, "grad_norm": 14.9375, "learning_rate": 1.465992312239325e-06, "loss": 1.2263, "step": 33228 }, { "epoch": 83.0725, "grad_norm": 9.625, "learning_rate": 1.4655702855898923e-06, "loss": 1.4711, "step": 33229 }, { "epoch": 83.075, "grad_norm": 7.625, "learning_rate": 1.4651483148918432e-06, "loss": 1.3061, "step": 33230 }, { "epoch": 83.0775, "grad_norm": 6.6875, "learning_rate": 1.4647264001479377e-06, "loss": 1.2753, "step": 33231 }, { "epoch": 83.08, "grad_norm": 9.0625, "learning_rate": 1.464304541360946e-06, "loss": 1.4303, "step": 33232 }, { "epoch": 83.0825, "grad_norm": 10.3125, "learning_rate": 1.4638827385336296e-06, "loss": 1.4201, "step": 33233 }, { "epoch": 83.085, "grad_norm": 10.1875, "learning_rate": 1.4634609916687593e-06, "loss": 1.3265, "step": 33234 }, { "epoch": 83.0875, "grad_norm": 6.59375, "learning_rate": 1.463039300769098e-06, "loss": 1.3012, "step": 33235 }, { "epoch": 83.09, "grad_norm": 9.0625, "learning_rate": 1.4626176658374058e-06, "loss": 1.4229, "step": 33236 }, { "epoch": 83.0925, "grad_norm": 7.09375, "learning_rate": 1.4621960868764506e-06, "loss": 1.2819, "step": 33237 }, { "epoch": 83.095, "grad_norm": 9.5625, "learning_rate": 1.461774563888999e-06, "loss": 1.2154, "step": 33238 }, { "epoch": 83.0975, "grad_norm": 6.46875, "learning_rate": 1.4613530968778112e-06, "loss": 1.2244, "step": 33239 }, { "epoch": 83.1, "grad_norm": 8.5, "learning_rate": 1.460931685845649e-06, "loss": 1.2265, "step": 33240 }, { "epoch": 83.1025, "grad_norm": 7.28125, "learning_rate": 1.4605103307952774e-06, "loss": 1.1492, "step": 33241 }, { "epoch": 83.105, "grad_norm": 9.1875, "learning_rate": 1.4600890317294592e-06, "loss": 1.4169, "step": 33242 }, { "epoch": 83.1075, "grad_norm": 8.875, "learning_rate": 1.459667788650957e-06, "loss": 1.5186, "step": 33243 }, { "epoch": 83.11, "grad_norm": 6.53125, "learning_rate": 1.4592466015625274e-06, "loss": 1.2494, "step": 33244 }, { "epoch": 83.1125, "grad_norm": 8.1875, "learning_rate": 1.4588254704669381e-06, "loss": 1.3251, "step": 33245 }, { "epoch": 83.115, "grad_norm": 10.625, "learning_rate": 1.4584043953669457e-06, "loss": 1.459, "step": 33246 }, { "epoch": 83.1175, "grad_norm": 9.5, "learning_rate": 1.457983376265314e-06, "loss": 1.2926, "step": 33247 }, { "epoch": 83.12, "grad_norm": 8.4375, "learning_rate": 1.457562413164799e-06, "loss": 1.2857, "step": 33248 }, { "epoch": 83.1225, "grad_norm": 8.25, "learning_rate": 1.4571415060681648e-06, "loss": 1.4023, "step": 33249 }, { "epoch": 83.125, "grad_norm": 8.1875, "learning_rate": 1.4567206549781699e-06, "loss": 1.2276, "step": 33250 }, { "epoch": 83.1275, "grad_norm": 6.9375, "learning_rate": 1.4562998598975696e-06, "loss": 1.3395, "step": 33251 }, { "epoch": 83.13, "grad_norm": 7.1875, "learning_rate": 1.4558791208291256e-06, "loss": 1.4217, "step": 33252 }, { "epoch": 83.1325, "grad_norm": 6.21875, "learning_rate": 1.4554584377755986e-06, "loss": 1.4096, "step": 33253 }, { "epoch": 83.135, "grad_norm": 6.9375, "learning_rate": 1.4550378107397446e-06, "loss": 1.5382, "step": 33254 }, { "epoch": 83.1375, "grad_norm": 9.9375, "learning_rate": 1.4546172397243186e-06, "loss": 1.3026, "step": 33255 }, { "epoch": 83.14, "grad_norm": 7.0625, "learning_rate": 1.4541967247320787e-06, "loss": 1.3124, "step": 33256 }, { "epoch": 83.1425, "grad_norm": 6.3125, "learning_rate": 1.4537762657657861e-06, "loss": 1.3649, "step": 33257 }, { "epoch": 83.145, "grad_norm": 7.71875, "learning_rate": 1.4533558628281953e-06, "loss": 1.33, "step": 33258 }, { "epoch": 83.1475, "grad_norm": 8.25, "learning_rate": 1.4529355159220581e-06, "loss": 1.168, "step": 33259 }, { "epoch": 83.15, "grad_norm": 7.625, "learning_rate": 1.4525152250501362e-06, "loss": 1.2743, "step": 33260 }, { "epoch": 83.1525, "grad_norm": 7.1875, "learning_rate": 1.4520949902151793e-06, "loss": 1.2019, "step": 33261 }, { "epoch": 83.155, "grad_norm": 12.3125, "learning_rate": 1.4516748114199486e-06, "loss": 1.3601, "step": 33262 }, { "epoch": 83.1575, "grad_norm": 7.875, "learning_rate": 1.451254688667194e-06, "loss": 1.4483, "step": 33263 }, { "epoch": 83.16, "grad_norm": 8.0, "learning_rate": 1.4508346219596725e-06, "loss": 1.3944, "step": 33264 }, { "epoch": 83.1625, "grad_norm": 6.1875, "learning_rate": 1.4504146113001384e-06, "loss": 1.2979, "step": 33265 }, { "epoch": 83.165, "grad_norm": 8.25, "learning_rate": 1.4499946566913415e-06, "loss": 1.3915, "step": 33266 }, { "epoch": 83.1675, "grad_norm": 9.375, "learning_rate": 1.4495747581360364e-06, "loss": 1.4219, "step": 33267 }, { "epoch": 83.17, "grad_norm": 7.0625, "learning_rate": 1.4491549156369799e-06, "loss": 1.2806, "step": 33268 }, { "epoch": 83.1725, "grad_norm": 5.875, "learning_rate": 1.448735129196922e-06, "loss": 1.2727, "step": 33269 }, { "epoch": 83.175, "grad_norm": 7.5, "learning_rate": 1.4483153988186115e-06, "loss": 1.2657, "step": 33270 }, { "epoch": 83.1775, "grad_norm": 6.3125, "learning_rate": 1.4478957245048042e-06, "loss": 1.2528, "step": 33271 }, { "epoch": 83.18, "grad_norm": 9.625, "learning_rate": 1.4474761062582509e-06, "loss": 1.3056, "step": 33272 }, { "epoch": 83.1825, "grad_norm": 9.125, "learning_rate": 1.447056544081703e-06, "loss": 1.2298, "step": 33273 }, { "epoch": 83.185, "grad_norm": 6.6875, "learning_rate": 1.4466370379779083e-06, "loss": 1.2275, "step": 33274 }, { "epoch": 83.1875, "grad_norm": 9.0, "learning_rate": 1.4462175879496198e-06, "loss": 1.3258, "step": 33275 }, { "epoch": 83.19, "grad_norm": 7.90625, "learning_rate": 1.4457981939995847e-06, "loss": 1.4175, "step": 33276 }, { "epoch": 83.1925, "grad_norm": 12.9375, "learning_rate": 1.4453788561305571e-06, "loss": 1.1986, "step": 33277 }, { "epoch": 83.195, "grad_norm": 10.6875, "learning_rate": 1.4449595743452816e-06, "loss": 1.3712, "step": 33278 }, { "epoch": 83.1975, "grad_norm": 8.125, "learning_rate": 1.4445403486465103e-06, "loss": 1.3377, "step": 33279 }, { "epoch": 83.2, "grad_norm": 12.25, "learning_rate": 1.4441211790369892e-06, "loss": 1.464, "step": 33280 }, { "epoch": 83.2025, "grad_norm": 8.1875, "learning_rate": 1.4437020655194667e-06, "loss": 1.3675, "step": 33281 }, { "epoch": 83.205, "grad_norm": 9.375, "learning_rate": 1.4432830080966897e-06, "loss": 1.3334, "step": 33282 }, { "epoch": 83.2075, "grad_norm": 9.625, "learning_rate": 1.4428640067714095e-06, "loss": 1.3319, "step": 33283 }, { "epoch": 83.21, "grad_norm": 6.65625, "learning_rate": 1.4424450615463714e-06, "loss": 1.2718, "step": 33284 }, { "epoch": 83.2125, "grad_norm": 10.3125, "learning_rate": 1.442026172424318e-06, "loss": 1.2068, "step": 33285 }, { "epoch": 83.215, "grad_norm": 6.4375, "learning_rate": 1.4416073394079977e-06, "loss": 1.2327, "step": 33286 }, { "epoch": 83.2175, "grad_norm": 8.6875, "learning_rate": 1.441188562500161e-06, "loss": 1.2499, "step": 33287 }, { "epoch": 83.22, "grad_norm": 7.71875, "learning_rate": 1.4407698417035488e-06, "loss": 1.2386, "step": 33288 }, { "epoch": 83.2225, "grad_norm": 10.0625, "learning_rate": 1.4403511770209044e-06, "loss": 1.5028, "step": 33289 }, { "epoch": 83.225, "grad_norm": 6.03125, "learning_rate": 1.4399325684549782e-06, "loss": 1.2485, "step": 33290 }, { "epoch": 83.2275, "grad_norm": 8.5, "learning_rate": 1.4395140160085086e-06, "loss": 1.3425, "step": 33291 }, { "epoch": 83.23, "grad_norm": 7.78125, "learning_rate": 1.439095519684245e-06, "loss": 1.2913, "step": 33292 }, { "epoch": 83.2325, "grad_norm": 6.15625, "learning_rate": 1.438677079484927e-06, "loss": 1.3354, "step": 33293 }, { "epoch": 83.235, "grad_norm": 7.0625, "learning_rate": 1.4382586954133005e-06, "loss": 1.481, "step": 33294 }, { "epoch": 83.2375, "grad_norm": 9.0, "learning_rate": 1.4378403674721087e-06, "loss": 1.3059, "step": 33295 }, { "epoch": 83.24, "grad_norm": 11.0, "learning_rate": 1.4374220956640895e-06, "loss": 1.3372, "step": 33296 }, { "epoch": 83.2425, "grad_norm": 7.34375, "learning_rate": 1.4370038799919884e-06, "loss": 1.3733, "step": 33297 }, { "epoch": 83.245, "grad_norm": 6.09375, "learning_rate": 1.4365857204585498e-06, "loss": 1.2321, "step": 33298 }, { "epoch": 83.2475, "grad_norm": 5.53125, "learning_rate": 1.436167617066513e-06, "loss": 1.2884, "step": 33299 }, { "epoch": 83.25, "grad_norm": 10.625, "learning_rate": 1.4357495698186186e-06, "loss": 1.1545, "step": 33300 }, { "epoch": 83.2525, "grad_norm": 6.3125, "learning_rate": 1.4353315787176015e-06, "loss": 1.2943, "step": 33301 }, { "epoch": 83.255, "grad_norm": 5.96875, "learning_rate": 1.4349136437662136e-06, "loss": 1.3637, "step": 33302 }, { "epoch": 83.2575, "grad_norm": 7.34375, "learning_rate": 1.4344957649671887e-06, "loss": 1.2361, "step": 33303 }, { "epoch": 83.26, "grad_norm": 8.0, "learning_rate": 1.4340779423232665e-06, "loss": 1.4347, "step": 33304 }, { "epoch": 83.2625, "grad_norm": 6.9375, "learning_rate": 1.433660175837185e-06, "loss": 1.3605, "step": 33305 }, { "epoch": 83.265, "grad_norm": 5.59375, "learning_rate": 1.4332424655116838e-06, "loss": 1.2987, "step": 33306 }, { "epoch": 83.2675, "grad_norm": 8.3125, "learning_rate": 1.4328248113495046e-06, "loss": 1.4204, "step": 33307 }, { "epoch": 83.27, "grad_norm": 6.875, "learning_rate": 1.4324072133533807e-06, "loss": 1.3058, "step": 33308 }, { "epoch": 83.2725, "grad_norm": 7.65625, "learning_rate": 1.431989671526055e-06, "loss": 1.3534, "step": 33309 }, { "epoch": 83.275, "grad_norm": 6.46875, "learning_rate": 1.4315721858702613e-06, "loss": 1.3669, "step": 33310 }, { "epoch": 83.2775, "grad_norm": 5.90625, "learning_rate": 1.4311547563887362e-06, "loss": 1.3537, "step": 33311 }, { "epoch": 83.28, "grad_norm": 6.71875, "learning_rate": 1.4307373830842174e-06, "loss": 1.1546, "step": 33312 }, { "epoch": 83.2825, "grad_norm": 6.84375, "learning_rate": 1.430320065959443e-06, "loss": 1.2671, "step": 33313 }, { "epoch": 83.285, "grad_norm": 6.9375, "learning_rate": 1.4299028050171482e-06, "loss": 1.2146, "step": 33314 }, { "epoch": 83.2875, "grad_norm": 7.0, "learning_rate": 1.4294856002600676e-06, "loss": 1.2532, "step": 33315 }, { "epoch": 83.29, "grad_norm": 6.71875, "learning_rate": 1.4290684516909314e-06, "loss": 1.2797, "step": 33316 }, { "epoch": 83.2925, "grad_norm": 6.1875, "learning_rate": 1.428651359312484e-06, "loss": 1.2467, "step": 33317 }, { "epoch": 83.295, "grad_norm": 10.375, "learning_rate": 1.4282343231274554e-06, "loss": 1.3277, "step": 33318 }, { "epoch": 83.2975, "grad_norm": 7.3125, "learning_rate": 1.4278173431385789e-06, "loss": 1.3088, "step": 33319 }, { "epoch": 83.3, "grad_norm": 8.25, "learning_rate": 1.427400419348588e-06, "loss": 1.3777, "step": 33320 }, { "epoch": 83.3025, "grad_norm": 7.9375, "learning_rate": 1.4269835517602159e-06, "loss": 1.2647, "step": 33321 }, { "epoch": 83.305, "grad_norm": 8.0625, "learning_rate": 1.4265667403761996e-06, "loss": 1.3525, "step": 33322 }, { "epoch": 83.3075, "grad_norm": 7.5625, "learning_rate": 1.4261499851992678e-06, "loss": 1.374, "step": 33323 }, { "epoch": 83.31, "grad_norm": 9.0625, "learning_rate": 1.4257332862321516e-06, "loss": 1.3025, "step": 33324 }, { "epoch": 83.3125, "grad_norm": 6.15625, "learning_rate": 1.4253166434775867e-06, "loss": 1.339, "step": 33325 }, { "epoch": 83.315, "grad_norm": 5.03125, "learning_rate": 1.4249000569383009e-06, "loss": 1.1928, "step": 33326 }, { "epoch": 83.3175, "grad_norm": 10.1875, "learning_rate": 1.4244835266170299e-06, "loss": 1.0936, "step": 33327 }, { "epoch": 83.32, "grad_norm": 8.125, "learning_rate": 1.424067052516499e-06, "loss": 1.287, "step": 33328 }, { "epoch": 83.3225, "grad_norm": 9.9375, "learning_rate": 1.4236506346394441e-06, "loss": 1.3822, "step": 33329 }, { "epoch": 83.325, "grad_norm": 6.71875, "learning_rate": 1.4232342729885916e-06, "loss": 1.4038, "step": 33330 }, { "epoch": 83.3275, "grad_norm": 8.875, "learning_rate": 1.4228179675666697e-06, "loss": 1.336, "step": 33331 }, { "epoch": 83.33, "grad_norm": 10.1875, "learning_rate": 1.4224017183764104e-06, "loss": 1.541, "step": 33332 }, { "epoch": 83.3325, "grad_norm": 10.75, "learning_rate": 1.421985525420545e-06, "loss": 1.4134, "step": 33333 }, { "epoch": 83.335, "grad_norm": 6.46875, "learning_rate": 1.421569388701798e-06, "loss": 1.2564, "step": 33334 }, { "epoch": 83.3375, "grad_norm": 7.21875, "learning_rate": 1.4211533082228957e-06, "loss": 1.3208, "step": 33335 }, { "epoch": 83.34, "grad_norm": 4.34375, "learning_rate": 1.420737283986572e-06, "loss": 1.3614, "step": 33336 }, { "epoch": 83.3425, "grad_norm": 8.875, "learning_rate": 1.420321315995552e-06, "loss": 1.2322, "step": 33337 }, { "epoch": 83.345, "grad_norm": 6.15625, "learning_rate": 1.4199054042525628e-06, "loss": 1.3516, "step": 33338 }, { "epoch": 83.3475, "grad_norm": 8.875, "learning_rate": 1.4194895487603265e-06, "loss": 1.362, "step": 33339 }, { "epoch": 83.35, "grad_norm": 9.0625, "learning_rate": 1.4190737495215746e-06, "loss": 1.2608, "step": 33340 }, { "epoch": 83.3525, "grad_norm": 6.21875, "learning_rate": 1.4186580065390344e-06, "loss": 1.2782, "step": 33341 }, { "epoch": 83.355, "grad_norm": 4.34375, "learning_rate": 1.4182423198154282e-06, "loss": 1.3316, "step": 33342 }, { "epoch": 83.3575, "grad_norm": 5.90625, "learning_rate": 1.4178266893534797e-06, "loss": 1.3039, "step": 33343 }, { "epoch": 83.36, "grad_norm": 12.8125, "learning_rate": 1.4174111151559188e-06, "loss": 1.2835, "step": 33344 }, { "epoch": 83.3625, "grad_norm": 6.5625, "learning_rate": 1.4169955972254645e-06, "loss": 1.1672, "step": 33345 }, { "epoch": 83.365, "grad_norm": 9.0, "learning_rate": 1.4165801355648456e-06, "loss": 1.3319, "step": 33346 }, { "epoch": 83.3675, "grad_norm": 12.625, "learning_rate": 1.4161647301767811e-06, "loss": 1.2697, "step": 33347 }, { "epoch": 83.37, "grad_norm": 7.59375, "learning_rate": 1.4157493810640011e-06, "loss": 1.2999, "step": 33348 }, { "epoch": 83.3725, "grad_norm": 6.3125, "learning_rate": 1.4153340882292232e-06, "loss": 1.4369, "step": 33349 }, { "epoch": 83.375, "grad_norm": 7.6875, "learning_rate": 1.41491885167517e-06, "loss": 1.245, "step": 33350 }, { "epoch": 83.3775, "grad_norm": 6.3125, "learning_rate": 1.4145036714045646e-06, "loss": 1.2926, "step": 33351 }, { "epoch": 83.38, "grad_norm": 6.53125, "learning_rate": 1.4140885474201315e-06, "loss": 1.411, "step": 33352 }, { "epoch": 83.3825, "grad_norm": 6.46875, "learning_rate": 1.413673479724591e-06, "loss": 1.466, "step": 33353 }, { "epoch": 83.385, "grad_norm": 5.78125, "learning_rate": 1.4132584683206619e-06, "loss": 1.5104, "step": 33354 }, { "epoch": 83.3875, "grad_norm": 7.21875, "learning_rate": 1.4128435132110652e-06, "loss": 1.237, "step": 33355 }, { "epoch": 83.39, "grad_norm": 7.78125, "learning_rate": 1.4124286143985267e-06, "loss": 1.2646, "step": 33356 }, { "epoch": 83.3925, "grad_norm": 7.09375, "learning_rate": 1.4120137718857606e-06, "loss": 1.3333, "step": 33357 }, { "epoch": 83.395, "grad_norm": 7.09375, "learning_rate": 1.4115989856754874e-06, "loss": 1.305, "step": 33358 }, { "epoch": 83.3975, "grad_norm": 9.25, "learning_rate": 1.4111842557704303e-06, "loss": 1.3922, "step": 33359 }, { "epoch": 83.4, "grad_norm": 8.375, "learning_rate": 1.4107695821733026e-06, "loss": 1.2069, "step": 33360 }, { "epoch": 83.4025, "grad_norm": 5.375, "learning_rate": 1.4103549648868287e-06, "loss": 1.1874, "step": 33361 }, { "epoch": 83.405, "grad_norm": 5.78125, "learning_rate": 1.409940403913721e-06, "loss": 1.3359, "step": 33362 }, { "epoch": 83.4075, "grad_norm": 9.625, "learning_rate": 1.409525899256703e-06, "loss": 1.3351, "step": 33363 }, { "epoch": 83.41, "grad_norm": 7.4375, "learning_rate": 1.4091114509184889e-06, "loss": 1.4193, "step": 33364 }, { "epoch": 83.4125, "grad_norm": 10.9375, "learning_rate": 1.4086970589017935e-06, "loss": 1.3428, "step": 33365 }, { "epoch": 83.415, "grad_norm": 8.5625, "learning_rate": 1.4082827232093377e-06, "loss": 1.4036, "step": 33366 }, { "epoch": 83.4175, "grad_norm": 8.25, "learning_rate": 1.4078684438438385e-06, "loss": 1.5899, "step": 33367 }, { "epoch": 83.42, "grad_norm": 9.1875, "learning_rate": 1.407454220808009e-06, "loss": 1.3648, "step": 33368 }, { "epoch": 83.4225, "grad_norm": 10.5, "learning_rate": 1.4070400541045648e-06, "loss": 1.3972, "step": 33369 }, { "epoch": 83.425, "grad_norm": 6.78125, "learning_rate": 1.4066259437362207e-06, "loss": 1.3437, "step": 33370 }, { "epoch": 83.4275, "grad_norm": 6.3125, "learning_rate": 1.4062118897056964e-06, "loss": 1.3161, "step": 33371 }, { "epoch": 83.43, "grad_norm": 7.34375, "learning_rate": 1.4057978920157033e-06, "loss": 1.327, "step": 33372 }, { "epoch": 83.4325, "grad_norm": 8.4375, "learning_rate": 1.4053839506689527e-06, "loss": 1.3447, "step": 33373 }, { "epoch": 83.435, "grad_norm": 11.125, "learning_rate": 1.4049700656681619e-06, "loss": 1.3137, "step": 33374 }, { "epoch": 83.4375, "grad_norm": 7.1875, "learning_rate": 1.4045562370160426e-06, "loss": 1.4069, "step": 33375 }, { "epoch": 83.44, "grad_norm": 6.21875, "learning_rate": 1.4041424647153112e-06, "loss": 1.2788, "step": 33376 }, { "epoch": 83.4425, "grad_norm": 10.875, "learning_rate": 1.4037287487686747e-06, "loss": 1.3394, "step": 33377 }, { "epoch": 83.445, "grad_norm": 7.34375, "learning_rate": 1.4033150891788506e-06, "loss": 1.34, "step": 33378 }, { "epoch": 83.4475, "grad_norm": 5.15625, "learning_rate": 1.402901485948549e-06, "loss": 1.1844, "step": 33379 }, { "epoch": 83.45, "grad_norm": 10.0, "learning_rate": 1.402487939080479e-06, "loss": 1.3541, "step": 33380 }, { "epoch": 83.4525, "grad_norm": 7.03125, "learning_rate": 1.4020744485773552e-06, "loss": 1.3338, "step": 33381 }, { "epoch": 83.455, "grad_norm": 11.6875, "learning_rate": 1.4016610144418885e-06, "loss": 1.1908, "step": 33382 }, { "epoch": 83.4575, "grad_norm": 6.125, "learning_rate": 1.401247636676788e-06, "loss": 1.4001, "step": 33383 }, { "epoch": 83.46, "grad_norm": 10.4375, "learning_rate": 1.4008343152847615e-06, "loss": 1.3764, "step": 33384 }, { "epoch": 83.4625, "grad_norm": 10.875, "learning_rate": 1.4004210502685212e-06, "loss": 1.2508, "step": 33385 }, { "epoch": 83.465, "grad_norm": 8.125, "learning_rate": 1.4000078416307783e-06, "loss": 1.2785, "step": 33386 }, { "epoch": 83.4675, "grad_norm": 9.0, "learning_rate": 1.3995946893742408e-06, "loss": 1.2739, "step": 33387 }, { "epoch": 83.47, "grad_norm": 5.96875, "learning_rate": 1.3991815935016128e-06, "loss": 1.3122, "step": 33388 }, { "epoch": 83.4725, "grad_norm": 9.625, "learning_rate": 1.3987685540156092e-06, "loss": 1.264, "step": 33389 }, { "epoch": 83.475, "grad_norm": 6.25, "learning_rate": 1.3983555709189333e-06, "loss": 1.2569, "step": 33390 }, { "epoch": 83.4775, "grad_norm": 6.46875, "learning_rate": 1.3979426442142952e-06, "loss": 1.3699, "step": 33391 }, { "epoch": 83.48, "grad_norm": 7.0, "learning_rate": 1.3975297739043992e-06, "loss": 1.3533, "step": 33392 }, { "epoch": 83.4825, "grad_norm": 6.1875, "learning_rate": 1.3971169599919553e-06, "loss": 1.313, "step": 33393 }, { "epoch": 83.485, "grad_norm": 6.375, "learning_rate": 1.3967042024796685e-06, "loss": 1.4395, "step": 33394 }, { "epoch": 83.4875, "grad_norm": 6.0, "learning_rate": 1.3962915013702428e-06, "loss": 1.1422, "step": 33395 }, { "epoch": 83.49, "grad_norm": 7.34375, "learning_rate": 1.395878856666385e-06, "loss": 1.3894, "step": 33396 }, { "epoch": 83.4925, "grad_norm": 6.96875, "learning_rate": 1.3954662683708042e-06, "loss": 1.2948, "step": 33397 }, { "epoch": 83.495, "grad_norm": 9.6875, "learning_rate": 1.3950537364862017e-06, "loss": 1.3421, "step": 33398 }, { "epoch": 83.4975, "grad_norm": 6.875, "learning_rate": 1.3946412610152805e-06, "loss": 1.3455, "step": 33399 }, { "epoch": 83.5, "grad_norm": 7.0625, "learning_rate": 1.3942288419607476e-06, "loss": 1.3301, "step": 33400 }, { "epoch": 83.5025, "grad_norm": 6.78125, "learning_rate": 1.3938164793253073e-06, "loss": 1.2402, "step": 33401 }, { "epoch": 83.505, "grad_norm": 8.25, "learning_rate": 1.393404173111661e-06, "loss": 1.2954, "step": 33402 }, { "epoch": 83.5075, "grad_norm": 6.375, "learning_rate": 1.3929919233225119e-06, "loss": 1.4192, "step": 33403 }, { "epoch": 83.51, "grad_norm": 8.625, "learning_rate": 1.3925797299605649e-06, "loss": 1.3092, "step": 33404 }, { "epoch": 83.5125, "grad_norm": 8.0, "learning_rate": 1.3921675930285184e-06, "loss": 1.3229, "step": 33405 }, { "epoch": 83.515, "grad_norm": 7.84375, "learning_rate": 1.3917555125290794e-06, "loss": 1.3173, "step": 33406 }, { "epoch": 83.5175, "grad_norm": 8.3125, "learning_rate": 1.3913434884649435e-06, "loss": 1.4959, "step": 33407 }, { "epoch": 83.52, "grad_norm": 8.125, "learning_rate": 1.3909315208388185e-06, "loss": 1.385, "step": 33408 }, { "epoch": 83.5225, "grad_norm": 10.0625, "learning_rate": 1.3905196096534014e-06, "loss": 1.296, "step": 33409 }, { "epoch": 83.525, "grad_norm": 6.59375, "learning_rate": 1.3901077549113905e-06, "loss": 1.2795, "step": 33410 }, { "epoch": 83.5275, "grad_norm": 10.3125, "learning_rate": 1.3896959566154899e-06, "loss": 1.3801, "step": 33411 }, { "epoch": 83.53, "grad_norm": 5.96875, "learning_rate": 1.3892842147683994e-06, "loss": 1.2656, "step": 33412 }, { "epoch": 83.5325, "grad_norm": 5.46875, "learning_rate": 1.3888725293728166e-06, "loss": 1.3582, "step": 33413 }, { "epoch": 83.535, "grad_norm": 7.1875, "learning_rate": 1.3884609004314398e-06, "loss": 1.4108, "step": 33414 }, { "epoch": 83.5375, "grad_norm": 8.5625, "learning_rate": 1.3880493279469686e-06, "loss": 1.2199, "step": 33415 }, { "epoch": 83.54, "grad_norm": 11.75, "learning_rate": 1.3876378119221024e-06, "loss": 1.4103, "step": 33416 }, { "epoch": 83.5425, "grad_norm": 5.65625, "learning_rate": 1.3872263523595398e-06, "loss": 1.3624, "step": 33417 }, { "epoch": 83.545, "grad_norm": 10.5, "learning_rate": 1.3868149492619732e-06, "loss": 1.3607, "step": 33418 }, { "epoch": 83.5475, "grad_norm": 6.6875, "learning_rate": 1.3864036026321048e-06, "loss": 1.2755, "step": 33419 }, { "epoch": 83.55, "grad_norm": 8.0, "learning_rate": 1.3859923124726283e-06, "loss": 1.3698, "step": 33420 }, { "epoch": 83.5525, "grad_norm": 7.6875, "learning_rate": 1.3855810787862434e-06, "loss": 1.412, "step": 33421 }, { "epoch": 83.555, "grad_norm": 7.5, "learning_rate": 1.3851699015756426e-06, "loss": 1.3382, "step": 33422 }, { "epoch": 83.5575, "grad_norm": 6.90625, "learning_rate": 1.3847587808435248e-06, "loss": 1.3642, "step": 33423 }, { "epoch": 83.56, "grad_norm": 7.21875, "learning_rate": 1.3843477165925846e-06, "loss": 1.3698, "step": 33424 }, { "epoch": 83.5625, "grad_norm": 6.625, "learning_rate": 1.383936708825513e-06, "loss": 1.2711, "step": 33425 }, { "epoch": 83.565, "grad_norm": 6.6875, "learning_rate": 1.3835257575450067e-06, "loss": 1.2934, "step": 33426 }, { "epoch": 83.5675, "grad_norm": 6.53125, "learning_rate": 1.3831148627537639e-06, "loss": 1.3809, "step": 33427 }, { "epoch": 83.57, "grad_norm": 6.34375, "learning_rate": 1.3827040244544754e-06, "loss": 1.3338, "step": 33428 }, { "epoch": 83.5725, "grad_norm": 8.5625, "learning_rate": 1.3822932426498326e-06, "loss": 1.3143, "step": 33429 }, { "epoch": 83.575, "grad_norm": 8.0625, "learning_rate": 1.3818825173425288e-06, "loss": 1.4566, "step": 33430 }, { "epoch": 83.5775, "grad_norm": 7.28125, "learning_rate": 1.381471848535262e-06, "loss": 1.373, "step": 33431 }, { "epoch": 83.58, "grad_norm": 9.8125, "learning_rate": 1.3810612362307208e-06, "loss": 1.1795, "step": 33432 }, { "epoch": 83.5825, "grad_norm": 6.5625, "learning_rate": 1.3806506804315944e-06, "loss": 1.4495, "step": 33433 }, { "epoch": 83.585, "grad_norm": 5.25, "learning_rate": 1.3802401811405796e-06, "loss": 1.2376, "step": 33434 }, { "epoch": 83.5875, "grad_norm": 7.6875, "learning_rate": 1.3798297383603632e-06, "loss": 1.268, "step": 33435 }, { "epoch": 83.59, "grad_norm": 8.4375, "learning_rate": 1.3794193520936405e-06, "loss": 1.3986, "step": 33436 }, { "epoch": 83.5925, "grad_norm": 8.3125, "learning_rate": 1.3790090223430964e-06, "loss": 1.4251, "step": 33437 }, { "epoch": 83.595, "grad_norm": 9.75, "learning_rate": 1.3785987491114272e-06, "loss": 1.2728, "step": 33438 }, { "epoch": 83.5975, "grad_norm": 6.84375, "learning_rate": 1.37818853240132e-06, "loss": 1.3883, "step": 33439 }, { "epoch": 83.6, "grad_norm": 6.8125, "learning_rate": 1.3777783722154603e-06, "loss": 1.2573, "step": 33440 }, { "epoch": 83.6025, "grad_norm": 6.71875, "learning_rate": 1.3773682685565415e-06, "loss": 1.2839, "step": 33441 }, { "epoch": 83.605, "grad_norm": 7.125, "learning_rate": 1.3769582214272526e-06, "loss": 1.2114, "step": 33442 }, { "epoch": 83.6075, "grad_norm": 9.4375, "learning_rate": 1.3765482308302813e-06, "loss": 1.3992, "step": 33443 }, { "epoch": 83.61, "grad_norm": 12.1875, "learning_rate": 1.3761382967683146e-06, "loss": 1.3018, "step": 33444 }, { "epoch": 83.6125, "grad_norm": 5.8125, "learning_rate": 1.3757284192440356e-06, "loss": 1.5095, "step": 33445 }, { "epoch": 83.615, "grad_norm": 5.75, "learning_rate": 1.3753185982601402e-06, "loss": 1.395, "step": 33446 }, { "epoch": 83.6175, "grad_norm": 6.15625, "learning_rate": 1.3749088338193118e-06, "loss": 1.4293, "step": 33447 }, { "epoch": 83.62, "grad_norm": 6.25, "learning_rate": 1.3744991259242324e-06, "loss": 1.2756, "step": 33448 }, { "epoch": 83.6225, "grad_norm": 7.9375, "learning_rate": 1.3740894745775946e-06, "loss": 1.2817, "step": 33449 }, { "epoch": 83.625, "grad_norm": 6.34375, "learning_rate": 1.3736798797820783e-06, "loss": 1.4088, "step": 33450 }, { "epoch": 83.6275, "grad_norm": 5.65625, "learning_rate": 1.3732703415403748e-06, "loss": 1.2214, "step": 33451 }, { "epoch": 83.63, "grad_norm": 8.375, "learning_rate": 1.372860859855163e-06, "loss": 1.4892, "step": 33452 }, { "epoch": 83.6325, "grad_norm": 7.0, "learning_rate": 1.3724514347291317e-06, "loss": 1.2058, "step": 33453 }, { "epoch": 83.635, "grad_norm": 5.65625, "learning_rate": 1.3720420661649636e-06, "loss": 1.4371, "step": 33454 }, { "epoch": 83.6375, "grad_norm": 5.4375, "learning_rate": 1.3716327541653408e-06, "loss": 1.2276, "step": 33455 }, { "epoch": 83.64, "grad_norm": 6.375, "learning_rate": 1.3712234987329486e-06, "loss": 1.2752, "step": 33456 }, { "epoch": 83.6425, "grad_norm": 7.28125, "learning_rate": 1.370814299870471e-06, "loss": 1.3126, "step": 33457 }, { "epoch": 83.645, "grad_norm": 8.4375, "learning_rate": 1.3704051575805898e-06, "loss": 1.4724, "step": 33458 }, { "epoch": 83.6475, "grad_norm": 7.96875, "learning_rate": 1.3699960718659877e-06, "loss": 1.4163, "step": 33459 }, { "epoch": 83.65, "grad_norm": 6.0, "learning_rate": 1.369587042729341e-06, "loss": 1.1993, "step": 33460 }, { "epoch": 83.6525, "grad_norm": 7.65625, "learning_rate": 1.3691780701733415e-06, "loss": 1.2355, "step": 33461 }, { "epoch": 83.655, "grad_norm": 8.5625, "learning_rate": 1.3687691542006643e-06, "loss": 1.2377, "step": 33462 }, { "epoch": 83.6575, "grad_norm": 5.65625, "learning_rate": 1.3683602948139917e-06, "loss": 1.2991, "step": 33463 }, { "epoch": 83.66, "grad_norm": 5.0625, "learning_rate": 1.367951492016001e-06, "loss": 1.3421, "step": 33464 }, { "epoch": 83.6625, "grad_norm": 6.8125, "learning_rate": 1.367542745809375e-06, "loss": 1.1211, "step": 33465 }, { "epoch": 83.665, "grad_norm": 6.0, "learning_rate": 1.3671340561967961e-06, "loss": 1.5439, "step": 33466 }, { "epoch": 83.6675, "grad_norm": 7.1875, "learning_rate": 1.3667254231809402e-06, "loss": 1.4358, "step": 33467 }, { "epoch": 83.67, "grad_norm": 6.875, "learning_rate": 1.366316846764485e-06, "loss": 1.4698, "step": 33468 }, { "epoch": 83.6725, "grad_norm": 6.78125, "learning_rate": 1.3659083269501127e-06, "loss": 1.3182, "step": 33469 }, { "epoch": 83.675, "grad_norm": 6.90625, "learning_rate": 1.3654998637404981e-06, "loss": 1.4195, "step": 33470 }, { "epoch": 83.6775, "grad_norm": 8.5625, "learning_rate": 1.3650914571383233e-06, "loss": 1.3298, "step": 33471 }, { "epoch": 83.68, "grad_norm": 7.0625, "learning_rate": 1.3646831071462606e-06, "loss": 1.3229, "step": 33472 }, { "epoch": 83.6825, "grad_norm": 7.1875, "learning_rate": 1.3642748137669926e-06, "loss": 1.2262, "step": 33473 }, { "epoch": 83.685, "grad_norm": 10.375, "learning_rate": 1.3638665770031933e-06, "loss": 1.3531, "step": 33474 }, { "epoch": 83.6875, "grad_norm": 9.4375, "learning_rate": 1.3634583968575343e-06, "loss": 1.4031, "step": 33475 }, { "epoch": 83.69, "grad_norm": 17.875, "learning_rate": 1.363050273332701e-06, "loss": 1.3973, "step": 33476 }, { "epoch": 83.6925, "grad_norm": 7.15625, "learning_rate": 1.362642206431365e-06, "loss": 1.2586, "step": 33477 }, { "epoch": 83.695, "grad_norm": 8.1875, "learning_rate": 1.3622341961562003e-06, "loss": 1.2725, "step": 33478 }, { "epoch": 83.6975, "grad_norm": 9.75, "learning_rate": 1.3618262425098805e-06, "loss": 1.3331, "step": 33479 }, { "epoch": 83.7, "grad_norm": 8.4375, "learning_rate": 1.3614183454950824e-06, "loss": 1.3792, "step": 33480 }, { "epoch": 83.7025, "grad_norm": 7.75, "learning_rate": 1.3610105051144817e-06, "loss": 1.3263, "step": 33481 }, { "epoch": 83.705, "grad_norm": 6.28125, "learning_rate": 1.3606027213707507e-06, "loss": 1.3816, "step": 33482 }, { "epoch": 83.7075, "grad_norm": 8.125, "learning_rate": 1.3601949942665604e-06, "loss": 1.472, "step": 33483 }, { "epoch": 83.71, "grad_norm": 7.25, "learning_rate": 1.359787323804589e-06, "loss": 1.3713, "step": 33484 }, { "epoch": 83.7125, "grad_norm": 8.25, "learning_rate": 1.3593797099875028e-06, "loss": 1.1785, "step": 33485 }, { "epoch": 83.715, "grad_norm": 8.625, "learning_rate": 1.35897215281798e-06, "loss": 1.3101, "step": 33486 }, { "epoch": 83.7175, "grad_norm": 7.78125, "learning_rate": 1.3585646522986872e-06, "loss": 1.3636, "step": 33487 }, { "epoch": 83.72, "grad_norm": 6.40625, "learning_rate": 1.3581572084323014e-06, "loss": 1.2925, "step": 33488 }, { "epoch": 83.7225, "grad_norm": 8.8125, "learning_rate": 1.3577498212214912e-06, "loss": 1.2871, "step": 33489 }, { "epoch": 83.725, "grad_norm": 6.75, "learning_rate": 1.357342490668926e-06, "loss": 1.2964, "step": 33490 }, { "epoch": 83.7275, "grad_norm": 7.4375, "learning_rate": 1.3569352167772765e-06, "loss": 1.2506, "step": 33491 }, { "epoch": 83.73, "grad_norm": 6.84375, "learning_rate": 1.3565279995492165e-06, "loss": 1.3111, "step": 33492 }, { "epoch": 83.7325, "grad_norm": 9.1875, "learning_rate": 1.3561208389874126e-06, "loss": 1.2859, "step": 33493 }, { "epoch": 83.735, "grad_norm": 12.75, "learning_rate": 1.3557137350945326e-06, "loss": 1.3987, "step": 33494 }, { "epoch": 83.7375, "grad_norm": 7.40625, "learning_rate": 1.355306687873247e-06, "loss": 1.1702, "step": 33495 }, { "epoch": 83.74, "grad_norm": 8.5, "learning_rate": 1.3548996973262273e-06, "loss": 1.2823, "step": 33496 }, { "epoch": 83.7425, "grad_norm": 9.0, "learning_rate": 1.354492763456139e-06, "loss": 1.2872, "step": 33497 }, { "epoch": 83.745, "grad_norm": 8.75, "learning_rate": 1.3540858862656491e-06, "loss": 1.2165, "step": 33498 }, { "epoch": 83.7475, "grad_norm": 8.8125, "learning_rate": 1.3536790657574273e-06, "loss": 1.3564, "step": 33499 }, { "epoch": 83.75, "grad_norm": 7.875, "learning_rate": 1.3532723019341376e-06, "loss": 1.1903, "step": 33500 }, { "epoch": 83.7525, "grad_norm": 11.5625, "learning_rate": 1.3528655947984504e-06, "loss": 1.329, "step": 33501 }, { "epoch": 83.755, "grad_norm": 8.4375, "learning_rate": 1.3524589443530288e-06, "loss": 1.3261, "step": 33502 }, { "epoch": 83.7575, "grad_norm": 8.875, "learning_rate": 1.3520523506005422e-06, "loss": 1.5046, "step": 33503 }, { "epoch": 83.76, "grad_norm": 7.21875, "learning_rate": 1.3516458135436539e-06, "loss": 1.4412, "step": 33504 }, { "epoch": 83.7625, "grad_norm": 8.0, "learning_rate": 1.3512393331850281e-06, "loss": 1.3771, "step": 33505 }, { "epoch": 83.765, "grad_norm": 6.5625, "learning_rate": 1.3508329095273308e-06, "loss": 1.2873, "step": 33506 }, { "epoch": 83.7675, "grad_norm": 7.71875, "learning_rate": 1.3504265425732289e-06, "loss": 1.344, "step": 33507 }, { "epoch": 83.77, "grad_norm": 13.9375, "learning_rate": 1.3500202323253853e-06, "loss": 1.3774, "step": 33508 }, { "epoch": 83.7725, "grad_norm": 6.375, "learning_rate": 1.3496139787864593e-06, "loss": 1.2544, "step": 33509 }, { "epoch": 83.775, "grad_norm": 7.46875, "learning_rate": 1.3492077819591188e-06, "loss": 1.3561, "step": 33510 }, { "epoch": 83.7775, "grad_norm": 8.625, "learning_rate": 1.3488016418460282e-06, "loss": 1.3364, "step": 33511 }, { "epoch": 83.78, "grad_norm": 5.78125, "learning_rate": 1.3483955584498476e-06, "loss": 1.2444, "step": 33512 }, { "epoch": 83.7825, "grad_norm": 7.28125, "learning_rate": 1.3479895317732372e-06, "loss": 1.3905, "step": 33513 }, { "epoch": 83.785, "grad_norm": 9.4375, "learning_rate": 1.3475835618188638e-06, "loss": 1.4273, "step": 33514 }, { "epoch": 83.7875, "grad_norm": 7.84375, "learning_rate": 1.347177648589383e-06, "loss": 1.364, "step": 33515 }, { "epoch": 83.79, "grad_norm": 6.40625, "learning_rate": 1.3467717920874624e-06, "loss": 1.4452, "step": 33516 }, { "epoch": 83.7925, "grad_norm": 8.125, "learning_rate": 1.346365992315758e-06, "loss": 1.4664, "step": 33517 }, { "epoch": 83.795, "grad_norm": 8.3125, "learning_rate": 1.345960249276933e-06, "loss": 1.2873, "step": 33518 }, { "epoch": 83.7975, "grad_norm": 7.9375, "learning_rate": 1.3455545629736466e-06, "loss": 1.4918, "step": 33519 }, { "epoch": 83.8, "grad_norm": 5.90625, "learning_rate": 1.3451489334085555e-06, "loss": 1.3955, "step": 33520 }, { "epoch": 83.8025, "grad_norm": 7.65625, "learning_rate": 1.344743360584323e-06, "loss": 1.472, "step": 33521 }, { "epoch": 83.805, "grad_norm": 7.28125, "learning_rate": 1.344337844503607e-06, "loss": 1.2406, "step": 33522 }, { "epoch": 83.8075, "grad_norm": 6.03125, "learning_rate": 1.3439323851690667e-06, "loss": 1.5354, "step": 33523 }, { "epoch": 83.81, "grad_norm": 8.5625, "learning_rate": 1.3435269825833575e-06, "loss": 1.4302, "step": 33524 }, { "epoch": 83.8125, "grad_norm": 7.5625, "learning_rate": 1.3431216367491384e-06, "loss": 1.4303, "step": 33525 }, { "epoch": 83.815, "grad_norm": 7.78125, "learning_rate": 1.3427163476690696e-06, "loss": 1.2843, "step": 33526 }, { "epoch": 83.8175, "grad_norm": 4.625, "learning_rate": 1.3423111153458067e-06, "loss": 1.5357, "step": 33527 }, { "epoch": 83.82, "grad_norm": 9.0, "learning_rate": 1.3419059397820033e-06, "loss": 1.1686, "step": 33528 }, { "epoch": 83.8225, "grad_norm": 9.25, "learning_rate": 1.3415008209803193e-06, "loss": 1.2982, "step": 33529 }, { "epoch": 83.825, "grad_norm": 8.25, "learning_rate": 1.3410957589434082e-06, "loss": 1.1815, "step": 33530 }, { "epoch": 83.8275, "grad_norm": 5.4375, "learning_rate": 1.3406907536739288e-06, "loss": 1.3107, "step": 33531 }, { "epoch": 83.83, "grad_norm": 7.15625, "learning_rate": 1.3402858051745315e-06, "loss": 1.2039, "step": 33532 }, { "epoch": 83.8325, "grad_norm": 7.5625, "learning_rate": 1.3398809134478763e-06, "loss": 1.3056, "step": 33533 }, { "epoch": 83.835, "grad_norm": 9.375, "learning_rate": 1.3394760784966155e-06, "loss": 1.2794, "step": 33534 }, { "epoch": 83.8375, "grad_norm": 6.0625, "learning_rate": 1.3390713003234013e-06, "loss": 1.2694, "step": 33535 }, { "epoch": 83.84, "grad_norm": 5.40625, "learning_rate": 1.3386665789308885e-06, "loss": 1.4266, "step": 33536 }, { "epoch": 83.8425, "grad_norm": 6.9375, "learning_rate": 1.3382619143217335e-06, "loss": 1.3657, "step": 33537 }, { "epoch": 83.845, "grad_norm": 5.21875, "learning_rate": 1.3378573064985867e-06, "loss": 1.2373, "step": 33538 }, { "epoch": 83.8475, "grad_norm": 6.4375, "learning_rate": 1.3374527554640982e-06, "loss": 1.255, "step": 33539 }, { "epoch": 83.85, "grad_norm": 7.78125, "learning_rate": 1.3370482612209224e-06, "loss": 1.2206, "step": 33540 }, { "epoch": 83.8525, "grad_norm": 6.78125, "learning_rate": 1.336643823771715e-06, "loss": 1.1854, "step": 33541 }, { "epoch": 83.855, "grad_norm": 7.5, "learning_rate": 1.3362394431191227e-06, "loss": 1.3446, "step": 33542 }, { "epoch": 83.8575, "grad_norm": 6.96875, "learning_rate": 1.335835119265796e-06, "loss": 1.2462, "step": 33543 }, { "epoch": 83.86, "grad_norm": 9.875, "learning_rate": 1.33543085221439e-06, "loss": 1.3569, "step": 33544 }, { "epoch": 83.8625, "grad_norm": 7.71875, "learning_rate": 1.3350266419675506e-06, "loss": 1.4532, "step": 33545 }, { "epoch": 83.865, "grad_norm": 8.875, "learning_rate": 1.3346224885279325e-06, "loss": 1.2715, "step": 33546 }, { "epoch": 83.8675, "grad_norm": 8.0625, "learning_rate": 1.3342183918981789e-06, "loss": 1.2043, "step": 33547 }, { "epoch": 83.87, "grad_norm": 11.9375, "learning_rate": 1.3338143520809465e-06, "loss": 1.3635, "step": 33548 }, { "epoch": 83.8725, "grad_norm": 6.09375, "learning_rate": 1.3334103690788791e-06, "loss": 1.3005, "step": 33549 }, { "epoch": 83.875, "grad_norm": 8.375, "learning_rate": 1.3330064428946255e-06, "loss": 1.3606, "step": 33550 }, { "epoch": 83.8775, "grad_norm": 6.78125, "learning_rate": 1.3326025735308335e-06, "loss": 1.2194, "step": 33551 }, { "epoch": 83.88, "grad_norm": 6.84375, "learning_rate": 1.3321987609901553e-06, "loss": 1.359, "step": 33552 }, { "epoch": 83.8825, "grad_norm": 7.25, "learning_rate": 1.3317950052752348e-06, "loss": 1.2859, "step": 33553 }, { "epoch": 83.885, "grad_norm": 6.6875, "learning_rate": 1.3313913063887175e-06, "loss": 1.2975, "step": 33554 }, { "epoch": 83.8875, "grad_norm": 6.03125, "learning_rate": 1.330987664333252e-06, "loss": 1.1628, "step": 33555 }, { "epoch": 83.89, "grad_norm": 9.8125, "learning_rate": 1.3305840791114854e-06, "loss": 1.3519, "step": 33556 }, { "epoch": 83.8925, "grad_norm": 7.25, "learning_rate": 1.3301805507260645e-06, "loss": 1.2864, "step": 33557 }, { "epoch": 83.895, "grad_norm": 8.75, "learning_rate": 1.3297770791796294e-06, "loss": 1.3131, "step": 33558 }, { "epoch": 83.8975, "grad_norm": 9.1875, "learning_rate": 1.3293736644748311e-06, "loss": 1.2922, "step": 33559 }, { "epoch": 83.9, "grad_norm": 7.46875, "learning_rate": 1.3289703066143112e-06, "loss": 1.2422, "step": 33560 }, { "epoch": 83.9025, "grad_norm": 6.875, "learning_rate": 1.3285670056007161e-06, "loss": 1.3168, "step": 33561 }, { "epoch": 83.905, "grad_norm": 7.25, "learning_rate": 1.328163761436686e-06, "loss": 1.2143, "step": 33562 }, { "epoch": 83.9075, "grad_norm": 10.4375, "learning_rate": 1.32776057412487e-06, "loss": 1.2918, "step": 33563 }, { "epoch": 83.91, "grad_norm": 7.84375, "learning_rate": 1.3273574436679093e-06, "loss": 1.3489, "step": 33564 }, { "epoch": 83.9125, "grad_norm": 7.78125, "learning_rate": 1.3269543700684428e-06, "loss": 1.182, "step": 33565 }, { "epoch": 83.915, "grad_norm": 9.375, "learning_rate": 1.3265513533291174e-06, "loss": 1.3158, "step": 33566 }, { "epoch": 83.9175, "grad_norm": 6.65625, "learning_rate": 1.3261483934525754e-06, "loss": 1.3683, "step": 33567 }, { "epoch": 83.92, "grad_norm": 8.375, "learning_rate": 1.325745490441458e-06, "loss": 1.1148, "step": 33568 }, { "epoch": 83.9225, "grad_norm": 6.4375, "learning_rate": 1.3253426442984041e-06, "loss": 1.2787, "step": 33569 }, { "epoch": 83.925, "grad_norm": 10.4375, "learning_rate": 1.324939855026056e-06, "loss": 1.5306, "step": 33570 }, { "epoch": 83.9275, "grad_norm": 7.09375, "learning_rate": 1.3245371226270576e-06, "loss": 1.373, "step": 33571 }, { "epoch": 83.93, "grad_norm": 7.875, "learning_rate": 1.3241344471040473e-06, "loss": 1.3476, "step": 33572 }, { "epoch": 83.9325, "grad_norm": 7.125, "learning_rate": 1.3237318284596612e-06, "loss": 1.3822, "step": 33573 }, { "epoch": 83.935, "grad_norm": 9.625, "learning_rate": 1.3233292666965425e-06, "loss": 1.2994, "step": 33574 }, { "epoch": 83.9375, "grad_norm": 7.3125, "learning_rate": 1.3229267618173324e-06, "loss": 1.2895, "step": 33575 }, { "epoch": 83.94, "grad_norm": 8.5, "learning_rate": 1.3225243138246669e-06, "loss": 1.265, "step": 33576 }, { "epoch": 83.9425, "grad_norm": 8.375, "learning_rate": 1.3221219227211824e-06, "loss": 1.5107, "step": 33577 }, { "epoch": 83.945, "grad_norm": 7.96875, "learning_rate": 1.3217195885095213e-06, "loss": 1.3636, "step": 33578 }, { "epoch": 83.9475, "grad_norm": 11.0, "learning_rate": 1.3213173111923173e-06, "loss": 1.2021, "step": 33579 }, { "epoch": 83.95, "grad_norm": 8.5, "learning_rate": 1.3209150907722124e-06, "loss": 1.3413, "step": 33580 }, { "epoch": 83.9525, "grad_norm": 8.0625, "learning_rate": 1.320512927251838e-06, "loss": 1.4128, "step": 33581 }, { "epoch": 83.955, "grad_norm": 6.28125, "learning_rate": 1.3201108206338353e-06, "loss": 1.1987, "step": 33582 }, { "epoch": 83.9575, "grad_norm": 7.5, "learning_rate": 1.31970877092084e-06, "loss": 1.3122, "step": 33583 }, { "epoch": 83.96, "grad_norm": 8.8125, "learning_rate": 1.3193067781154835e-06, "loss": 1.1358, "step": 33584 }, { "epoch": 83.9625, "grad_norm": 7.8125, "learning_rate": 1.3189048422204043e-06, "loss": 1.1875, "step": 33585 }, { "epoch": 83.965, "grad_norm": 16.875, "learning_rate": 1.3185029632382407e-06, "loss": 1.3761, "step": 33586 }, { "epoch": 83.9675, "grad_norm": 9.0, "learning_rate": 1.3181011411716238e-06, "loss": 1.4217, "step": 33587 }, { "epoch": 83.97, "grad_norm": 7.125, "learning_rate": 1.317699376023186e-06, "loss": 1.3357, "step": 33588 }, { "epoch": 83.9725, "grad_norm": 5.21875, "learning_rate": 1.317297667795563e-06, "loss": 1.1533, "step": 33589 }, { "epoch": 83.975, "grad_norm": 6.5, "learning_rate": 1.3168960164913924e-06, "loss": 1.3195, "step": 33590 }, { "epoch": 83.9775, "grad_norm": 6.90625, "learning_rate": 1.3164944221133037e-06, "loss": 1.2235, "step": 33591 }, { "epoch": 83.98, "grad_norm": 9.25, "learning_rate": 1.3160928846639275e-06, "loss": 1.206, "step": 33592 }, { "epoch": 83.9825, "grad_norm": 9.75, "learning_rate": 1.3156914041459013e-06, "loss": 1.275, "step": 33593 }, { "epoch": 83.985, "grad_norm": 8.4375, "learning_rate": 1.3152899805618524e-06, "loss": 1.3578, "step": 33594 }, { "epoch": 83.9875, "grad_norm": 8.125, "learning_rate": 1.3148886139144167e-06, "loss": 1.2553, "step": 33595 }, { "epoch": 83.99, "grad_norm": 7.5625, "learning_rate": 1.3144873042062222e-06, "loss": 1.2962, "step": 33596 }, { "epoch": 83.9925, "grad_norm": 8.4375, "learning_rate": 1.3140860514399033e-06, "loss": 1.249, "step": 33597 }, { "epoch": 83.995, "grad_norm": 5.0, "learning_rate": 1.3136848556180893e-06, "loss": 1.2727, "step": 33598 }, { "epoch": 83.9975, "grad_norm": 13.1875, "learning_rate": 1.3132837167434065e-06, "loss": 1.3236, "step": 33599 }, { "epoch": 84.0, "grad_norm": 7.25, "learning_rate": 1.3128826348184886e-06, "loss": 1.2701, "step": 33600 }, { "epoch": 84.0025, "grad_norm": 7.6875, "learning_rate": 1.312481609845967e-06, "loss": 1.279, "step": 33601 }, { "epoch": 84.005, "grad_norm": 7.71875, "learning_rate": 1.3120806418284671e-06, "loss": 1.2365, "step": 33602 }, { "epoch": 84.0075, "grad_norm": 10.0625, "learning_rate": 1.3116797307686202e-06, "loss": 1.367, "step": 33603 }, { "epoch": 84.01, "grad_norm": 5.75, "learning_rate": 1.3112788766690477e-06, "loss": 1.3345, "step": 33604 }, { "epoch": 84.0125, "grad_norm": 7.15625, "learning_rate": 1.3108780795323884e-06, "loss": 1.2752, "step": 33605 }, { "epoch": 84.015, "grad_norm": 7.15625, "learning_rate": 1.3104773393612647e-06, "loss": 1.1706, "step": 33606 }, { "epoch": 84.0175, "grad_norm": 10.3125, "learning_rate": 1.3100766561583035e-06, "loss": 1.4306, "step": 33607 }, { "epoch": 84.02, "grad_norm": 8.0, "learning_rate": 1.3096760299261303e-06, "loss": 1.4276, "step": 33608 }, { "epoch": 84.0225, "grad_norm": 8.25, "learning_rate": 1.309275460667373e-06, "loss": 1.2372, "step": 33609 }, { "epoch": 84.025, "grad_norm": 7.71875, "learning_rate": 1.3088749483846597e-06, "loss": 1.4071, "step": 33610 }, { "epoch": 84.0275, "grad_norm": 6.625, "learning_rate": 1.3084744930806148e-06, "loss": 1.5105, "step": 33611 }, { "epoch": 84.03, "grad_norm": 7.59375, "learning_rate": 1.3080740947578618e-06, "loss": 1.2859, "step": 33612 }, { "epoch": 84.0325, "grad_norm": 4.78125, "learning_rate": 1.3076737534190286e-06, "loss": 1.6205, "step": 33613 }, { "epoch": 84.035, "grad_norm": 7.375, "learning_rate": 1.3072734690667366e-06, "loss": 1.3917, "step": 33614 }, { "epoch": 84.0375, "grad_norm": 8.0625, "learning_rate": 1.3068732417036112e-06, "loss": 1.1628, "step": 33615 }, { "epoch": 84.04, "grad_norm": 7.78125, "learning_rate": 1.3064730713322793e-06, "loss": 1.214, "step": 33616 }, { "epoch": 84.0425, "grad_norm": 7.78125, "learning_rate": 1.3060729579553622e-06, "loss": 1.3613, "step": 33617 }, { "epoch": 84.045, "grad_norm": 9.4375, "learning_rate": 1.3056729015754832e-06, "loss": 1.3947, "step": 33618 }, { "epoch": 84.0475, "grad_norm": 5.625, "learning_rate": 1.3052729021952603e-06, "loss": 1.3616, "step": 33619 }, { "epoch": 84.05, "grad_norm": 8.375, "learning_rate": 1.3048729598173248e-06, "loss": 1.2999, "step": 33620 }, { "epoch": 84.0525, "grad_norm": 5.0, "learning_rate": 1.3044730744442935e-06, "loss": 1.4768, "step": 33621 }, { "epoch": 84.055, "grad_norm": 5.90625, "learning_rate": 1.3040732460787898e-06, "loss": 1.2268, "step": 33622 }, { "epoch": 84.0575, "grad_norm": 6.9375, "learning_rate": 1.3036734747234303e-06, "loss": 1.1725, "step": 33623 }, { "epoch": 84.06, "grad_norm": 11.8125, "learning_rate": 1.303273760380841e-06, "loss": 1.4459, "step": 33624 }, { "epoch": 84.0625, "grad_norm": 6.3125, "learning_rate": 1.3028741030536418e-06, "loss": 1.3581, "step": 33625 }, { "epoch": 84.065, "grad_norm": 7.46875, "learning_rate": 1.3024745027444529e-06, "loss": 1.2506, "step": 33626 }, { "epoch": 84.0675, "grad_norm": 5.125, "learning_rate": 1.3020749594558912e-06, "loss": 1.235, "step": 33627 }, { "epoch": 84.07, "grad_norm": 7.09375, "learning_rate": 1.301675473190579e-06, "loss": 1.2494, "step": 33628 }, { "epoch": 84.0725, "grad_norm": 5.5, "learning_rate": 1.301276043951133e-06, "loss": 1.2722, "step": 33629 }, { "epoch": 84.075, "grad_norm": 7.71875, "learning_rate": 1.3008766717401743e-06, "loss": 1.299, "step": 33630 }, { "epoch": 84.0775, "grad_norm": 8.0625, "learning_rate": 1.300477356560319e-06, "loss": 1.2865, "step": 33631 }, { "epoch": 84.08, "grad_norm": 8.5, "learning_rate": 1.3000780984141881e-06, "loss": 1.2661, "step": 33632 }, { "epoch": 84.0825, "grad_norm": 8.0625, "learning_rate": 1.2996788973043961e-06, "loss": 1.18, "step": 33633 }, { "epoch": 84.085, "grad_norm": 6.75, "learning_rate": 1.29927975323356e-06, "loss": 1.2332, "step": 33634 }, { "epoch": 84.0875, "grad_norm": 6.9375, "learning_rate": 1.2988806662042975e-06, "loss": 1.4355, "step": 33635 }, { "epoch": 84.09, "grad_norm": 8.5625, "learning_rate": 1.2984816362192265e-06, "loss": 1.2619, "step": 33636 }, { "epoch": 84.0925, "grad_norm": 7.5625, "learning_rate": 1.2980826632809617e-06, "loss": 1.2233, "step": 33637 }, { "epoch": 84.095, "grad_norm": 12.6875, "learning_rate": 1.2976837473921177e-06, "loss": 1.3167, "step": 33638 }, { "epoch": 84.0975, "grad_norm": 6.0, "learning_rate": 1.2972848885553102e-06, "loss": 1.2342, "step": 33639 }, { "epoch": 84.1, "grad_norm": 6.34375, "learning_rate": 1.296886086773157e-06, "loss": 1.3499, "step": 33640 }, { "epoch": 84.1025, "grad_norm": 5.71875, "learning_rate": 1.2964873420482693e-06, "loss": 1.1642, "step": 33641 }, { "epoch": 84.105, "grad_norm": 6.78125, "learning_rate": 1.2960886543832607e-06, "loss": 1.3999, "step": 33642 }, { "epoch": 84.1075, "grad_norm": 7.96875, "learning_rate": 1.295690023780749e-06, "loss": 1.2311, "step": 33643 }, { "epoch": 84.11, "grad_norm": 9.6875, "learning_rate": 1.2952914502433433e-06, "loss": 1.2393, "step": 33644 }, { "epoch": 84.1125, "grad_norm": 8.875, "learning_rate": 1.2948929337736604e-06, "loss": 1.3543, "step": 33645 }, { "epoch": 84.115, "grad_norm": 6.8125, "learning_rate": 1.2944944743743093e-06, "loss": 1.1284, "step": 33646 }, { "epoch": 84.1175, "grad_norm": 6.15625, "learning_rate": 1.2940960720479056e-06, "loss": 1.3784, "step": 33647 }, { "epoch": 84.12, "grad_norm": 8.875, "learning_rate": 1.2936977267970597e-06, "loss": 1.3042, "step": 33648 }, { "epoch": 84.1225, "grad_norm": 6.75, "learning_rate": 1.2932994386243813e-06, "loss": 1.3502, "step": 33649 }, { "epoch": 84.125, "grad_norm": 10.0625, "learning_rate": 1.2929012075324832e-06, "loss": 1.511, "step": 33650 }, { "epoch": 84.1275, "grad_norm": 5.125, "learning_rate": 1.2925030335239786e-06, "loss": 1.3375, "step": 33651 }, { "epoch": 84.13, "grad_norm": 9.0625, "learning_rate": 1.2921049166014755e-06, "loss": 1.32, "step": 33652 }, { "epoch": 84.1325, "grad_norm": 7.125, "learning_rate": 1.2917068567675816e-06, "loss": 1.314, "step": 33653 }, { "epoch": 84.135, "grad_norm": 4.96875, "learning_rate": 1.2913088540249097e-06, "loss": 1.4756, "step": 33654 }, { "epoch": 84.1375, "grad_norm": 5.5, "learning_rate": 1.2909109083760707e-06, "loss": 1.223, "step": 33655 }, { "epoch": 84.14, "grad_norm": 7.75, "learning_rate": 1.2905130198236704e-06, "loss": 1.5078, "step": 33656 }, { "epoch": 84.1425, "grad_norm": 11.0, "learning_rate": 1.2901151883703166e-06, "loss": 1.258, "step": 33657 }, { "epoch": 84.145, "grad_norm": 12.5625, "learning_rate": 1.2897174140186207e-06, "loss": 1.2628, "step": 33658 }, { "epoch": 84.1475, "grad_norm": 9.125, "learning_rate": 1.2893196967711875e-06, "loss": 1.3139, "step": 33659 }, { "epoch": 84.15, "grad_norm": 7.15625, "learning_rate": 1.2889220366306276e-06, "loss": 1.2514, "step": 33660 }, { "epoch": 84.1525, "grad_norm": 7.75, "learning_rate": 1.288524433599544e-06, "loss": 1.3073, "step": 33661 }, { "epoch": 84.155, "grad_norm": 11.3125, "learning_rate": 1.2881268876805486e-06, "loss": 1.3622, "step": 33662 }, { "epoch": 84.1575, "grad_norm": 8.25, "learning_rate": 1.287729398876243e-06, "loss": 1.3427, "step": 33663 }, { "epoch": 84.16, "grad_norm": 9.125, "learning_rate": 1.2873319671892337e-06, "loss": 1.4116, "step": 33664 }, { "epoch": 84.1625, "grad_norm": 7.03125, "learning_rate": 1.2869345926221277e-06, "loss": 1.3561, "step": 33665 }, { "epoch": 84.165, "grad_norm": 7.0, "learning_rate": 1.2865372751775319e-06, "loss": 1.3343, "step": 33666 }, { "epoch": 84.1675, "grad_norm": 7.15625, "learning_rate": 1.2861400148580483e-06, "loss": 1.3871, "step": 33667 }, { "epoch": 84.17, "grad_norm": 8.75, "learning_rate": 1.2857428116662808e-06, "loss": 1.4353, "step": 33668 }, { "epoch": 84.1725, "grad_norm": 7.125, "learning_rate": 1.2853456656048336e-06, "loss": 1.1787, "step": 33669 }, { "epoch": 84.175, "grad_norm": 5.21875, "learning_rate": 1.2849485766763147e-06, "loss": 1.2874, "step": 33670 }, { "epoch": 84.1775, "grad_norm": 9.0625, "learning_rate": 1.2845515448833234e-06, "loss": 1.3384, "step": 33671 }, { "epoch": 84.18, "grad_norm": 4.875, "learning_rate": 1.2841545702284618e-06, "loss": 1.3076, "step": 33672 }, { "epoch": 84.1825, "grad_norm": 9.8125, "learning_rate": 1.283757652714336e-06, "loss": 1.3543, "step": 33673 }, { "epoch": 84.185, "grad_norm": 7.28125, "learning_rate": 1.2833607923435432e-06, "loss": 1.2339, "step": 33674 }, { "epoch": 84.1875, "grad_norm": 6.875, "learning_rate": 1.2829639891186917e-06, "loss": 1.3623, "step": 33675 }, { "epoch": 84.19, "grad_norm": 9.0625, "learning_rate": 1.2825672430423763e-06, "loss": 1.4307, "step": 33676 }, { "epoch": 84.1925, "grad_norm": 6.65625, "learning_rate": 1.2821705541172047e-06, "loss": 1.3037, "step": 33677 }, { "epoch": 84.195, "grad_norm": 5.8125, "learning_rate": 1.2817739223457726e-06, "loss": 1.2996, "step": 33678 }, { "epoch": 84.1975, "grad_norm": 7.78125, "learning_rate": 1.28137734773068e-06, "loss": 1.2217, "step": 33679 }, { "epoch": 84.2, "grad_norm": 8.3125, "learning_rate": 1.2809808302745298e-06, "loss": 1.4601, "step": 33680 }, { "epoch": 84.2025, "grad_norm": 7.1875, "learning_rate": 1.2805843699799214e-06, "loss": 1.2944, "step": 33681 }, { "epoch": 84.205, "grad_norm": 7.5, "learning_rate": 1.2801879668494531e-06, "loss": 1.3253, "step": 33682 }, { "epoch": 84.2075, "grad_norm": 6.1875, "learning_rate": 1.2797916208857209e-06, "loss": 1.3326, "step": 33683 }, { "epoch": 84.21, "grad_norm": 6.5625, "learning_rate": 1.2793953320913267e-06, "loss": 1.1996, "step": 33684 }, { "epoch": 84.2125, "grad_norm": 8.125, "learning_rate": 1.2789991004688694e-06, "loss": 1.4148, "step": 33685 }, { "epoch": 84.215, "grad_norm": 9.4375, "learning_rate": 1.2786029260209454e-06, "loss": 1.3218, "step": 33686 }, { "epoch": 84.2175, "grad_norm": 6.25, "learning_rate": 1.2782068087501487e-06, "loss": 1.429, "step": 33687 }, { "epoch": 84.22, "grad_norm": 4.53125, "learning_rate": 1.277810748659082e-06, "loss": 1.358, "step": 33688 }, { "epoch": 84.2225, "grad_norm": 6.46875, "learning_rate": 1.2774147457503372e-06, "loss": 1.2907, "step": 33689 }, { "epoch": 84.225, "grad_norm": 7.96875, "learning_rate": 1.2770188000265138e-06, "loss": 1.3252, "step": 33690 }, { "epoch": 84.2275, "grad_norm": 6.0625, "learning_rate": 1.276622911490204e-06, "loss": 1.2379, "step": 33691 }, { "epoch": 84.23, "grad_norm": 7.34375, "learning_rate": 1.276227080144007e-06, "loss": 1.2174, "step": 33692 }, { "epoch": 84.2325, "grad_norm": 4.9375, "learning_rate": 1.275831305990517e-06, "loss": 1.2762, "step": 33693 }, { "epoch": 84.235, "grad_norm": 7.34375, "learning_rate": 1.2754355890323244e-06, "loss": 1.3509, "step": 33694 }, { "epoch": 84.2375, "grad_norm": 5.875, "learning_rate": 1.2750399292720284e-06, "loss": 1.3594, "step": 33695 }, { "epoch": 84.24, "grad_norm": 7.03125, "learning_rate": 1.2746443267122233e-06, "loss": 1.3298, "step": 33696 }, { "epoch": 84.2425, "grad_norm": 5.4375, "learning_rate": 1.2742487813555004e-06, "loss": 1.4108, "step": 33697 }, { "epoch": 84.245, "grad_norm": 6.28125, "learning_rate": 1.2738532932044512e-06, "loss": 1.1522, "step": 33698 }, { "epoch": 84.2475, "grad_norm": 9.6875, "learning_rate": 1.2734578622616711e-06, "loss": 1.3679, "step": 33699 }, { "epoch": 84.25, "grad_norm": 7.21875, "learning_rate": 1.2730624885297537e-06, "loss": 1.4115, "step": 33700 }, { "epoch": 84.2525, "grad_norm": 5.90625, "learning_rate": 1.2726671720112893e-06, "loss": 1.1828, "step": 33701 }, { "epoch": 84.255, "grad_norm": 5.78125, "learning_rate": 1.272271912708868e-06, "loss": 1.2796, "step": 33702 }, { "epoch": 84.2575, "grad_norm": 7.1875, "learning_rate": 1.271876710625084e-06, "loss": 1.4131, "step": 33703 }, { "epoch": 84.26, "grad_norm": 10.75, "learning_rate": 1.2714815657625257e-06, "loss": 1.2702, "step": 33704 }, { "epoch": 84.2625, "grad_norm": 10.25, "learning_rate": 1.2710864781237876e-06, "loss": 1.1544, "step": 33705 }, { "epoch": 84.265, "grad_norm": 9.8125, "learning_rate": 1.270691447711454e-06, "loss": 1.3788, "step": 33706 }, { "epoch": 84.2675, "grad_norm": 7.6875, "learning_rate": 1.2702964745281198e-06, "loss": 1.3777, "step": 33707 }, { "epoch": 84.27, "grad_norm": 9.0625, "learning_rate": 1.269901558576374e-06, "loss": 1.3785, "step": 33708 }, { "epoch": 84.2725, "grad_norm": 9.5, "learning_rate": 1.2695066998588013e-06, "loss": 1.2481, "step": 33709 }, { "epoch": 84.275, "grad_norm": 7.875, "learning_rate": 1.2691118983779938e-06, "loss": 1.3836, "step": 33710 }, { "epoch": 84.2775, "grad_norm": 6.0625, "learning_rate": 1.2687171541365406e-06, "loss": 1.405, "step": 33711 }, { "epoch": 84.28, "grad_norm": 6.125, "learning_rate": 1.2683224671370286e-06, "loss": 1.2835, "step": 33712 }, { "epoch": 84.2825, "grad_norm": 7.375, "learning_rate": 1.2679278373820437e-06, "loss": 1.1578, "step": 33713 }, { "epoch": 84.285, "grad_norm": 8.0, "learning_rate": 1.2675332648741735e-06, "loss": 1.3088, "step": 33714 }, { "epoch": 84.2875, "grad_norm": 7.875, "learning_rate": 1.267138749616008e-06, "loss": 1.4384, "step": 33715 }, { "epoch": 84.29, "grad_norm": 7.59375, "learning_rate": 1.2667442916101313e-06, "loss": 1.2281, "step": 33716 }, { "epoch": 84.2925, "grad_norm": 9.4375, "learning_rate": 1.2663498908591277e-06, "loss": 1.3324, "step": 33717 }, { "epoch": 84.295, "grad_norm": 6.875, "learning_rate": 1.265955547365587e-06, "loss": 1.2044, "step": 33718 }, { "epoch": 84.2975, "grad_norm": 6.34375, "learning_rate": 1.26556126113209e-06, "loss": 1.4274, "step": 33719 }, { "epoch": 84.3, "grad_norm": 8.625, "learning_rate": 1.2651670321612264e-06, "loss": 1.1847, "step": 33720 }, { "epoch": 84.3025, "grad_norm": 7.03125, "learning_rate": 1.2647728604555754e-06, "loss": 1.2955, "step": 33721 }, { "epoch": 84.305, "grad_norm": 11.75, "learning_rate": 1.264378746017726e-06, "loss": 1.3203, "step": 33722 }, { "epoch": 84.3075, "grad_norm": 8.5, "learning_rate": 1.2639846888502604e-06, "loss": 1.2803, "step": 33723 }, { "epoch": 84.31, "grad_norm": 6.0625, "learning_rate": 1.2635906889557591e-06, "loss": 1.2278, "step": 33724 }, { "epoch": 84.3125, "grad_norm": 7.71875, "learning_rate": 1.2631967463368077e-06, "loss": 1.3796, "step": 33725 }, { "epoch": 84.315, "grad_norm": 10.1875, "learning_rate": 1.262802860995991e-06, "loss": 1.2846, "step": 33726 }, { "epoch": 84.3175, "grad_norm": 11.4375, "learning_rate": 1.2624090329358896e-06, "loss": 1.3719, "step": 33727 }, { "epoch": 84.32, "grad_norm": 10.25, "learning_rate": 1.262015262159082e-06, "loss": 1.2644, "step": 33728 }, { "epoch": 84.3225, "grad_norm": 11.0, "learning_rate": 1.2616215486681527e-06, "loss": 1.459, "step": 33729 }, { "epoch": 84.325, "grad_norm": 6.5, "learning_rate": 1.2612278924656863e-06, "loss": 1.3659, "step": 33730 }, { "epoch": 84.3275, "grad_norm": 5.65625, "learning_rate": 1.2608342935542584e-06, "loss": 1.3959, "step": 33731 }, { "epoch": 84.33, "grad_norm": 9.625, "learning_rate": 1.2604407519364504e-06, "loss": 1.3459, "step": 33732 }, { "epoch": 84.3325, "grad_norm": 8.125, "learning_rate": 1.2600472676148445e-06, "loss": 1.5723, "step": 33733 }, { "epoch": 84.335, "grad_norm": 6.5625, "learning_rate": 1.2596538405920168e-06, "loss": 1.125, "step": 33734 }, { "epoch": 84.3375, "grad_norm": 10.4375, "learning_rate": 1.2592604708705503e-06, "loss": 1.3526, "step": 33735 }, { "epoch": 84.34, "grad_norm": 11.25, "learning_rate": 1.258867158453021e-06, "loss": 1.2697, "step": 33736 }, { "epoch": 84.3425, "grad_norm": 6.59375, "learning_rate": 1.2584739033420114e-06, "loss": 1.3563, "step": 33737 }, { "epoch": 84.345, "grad_norm": 7.34375, "learning_rate": 1.2580807055400969e-06, "loss": 1.4206, "step": 33738 }, { "epoch": 84.3475, "grad_norm": 13.375, "learning_rate": 1.257687565049852e-06, "loss": 1.3221, "step": 33739 }, { "epoch": 84.35, "grad_norm": 7.5, "learning_rate": 1.2572944818738587e-06, "loss": 1.2761, "step": 33740 }, { "epoch": 84.3525, "grad_norm": 9.125, "learning_rate": 1.2569014560146954e-06, "loss": 1.1981, "step": 33741 }, { "epoch": 84.355, "grad_norm": 6.53125, "learning_rate": 1.2565084874749355e-06, "loss": 1.1737, "step": 33742 }, { "epoch": 84.3575, "grad_norm": 8.125, "learning_rate": 1.2561155762571563e-06, "loss": 1.3861, "step": 33743 }, { "epoch": 84.36, "grad_norm": 6.71875, "learning_rate": 1.255722722363929e-06, "loss": 1.2421, "step": 33744 }, { "epoch": 84.3625, "grad_norm": 10.1875, "learning_rate": 1.255329925797838e-06, "loss": 1.3737, "step": 33745 }, { "epoch": 84.365, "grad_norm": 11.4375, "learning_rate": 1.2549371865614546e-06, "loss": 1.3154, "step": 33746 }, { "epoch": 84.3675, "grad_norm": 8.4375, "learning_rate": 1.2545445046573523e-06, "loss": 1.185, "step": 33747 }, { "epoch": 84.37, "grad_norm": 7.21875, "learning_rate": 1.2541518800881047e-06, "loss": 1.3142, "step": 33748 }, { "epoch": 84.3725, "grad_norm": 5.03125, "learning_rate": 1.2537593128562863e-06, "loss": 1.259, "step": 33749 }, { "epoch": 84.375, "grad_norm": 10.4375, "learning_rate": 1.2533668029644751e-06, "loss": 1.355, "step": 33750 }, { "epoch": 84.3775, "grad_norm": 6.65625, "learning_rate": 1.252974350415239e-06, "loss": 1.3458, "step": 33751 }, { "epoch": 84.38, "grad_norm": 6.375, "learning_rate": 1.252581955211155e-06, "loss": 1.3264, "step": 33752 }, { "epoch": 84.3825, "grad_norm": 5.65625, "learning_rate": 1.252189617354793e-06, "loss": 1.4201, "step": 33753 }, { "epoch": 84.385, "grad_norm": 8.3125, "learning_rate": 1.2517973368487246e-06, "loss": 1.2308, "step": 33754 }, { "epoch": 84.3875, "grad_norm": 6.34375, "learning_rate": 1.2514051136955218e-06, "loss": 1.2855, "step": 33755 }, { "epoch": 84.39, "grad_norm": 9.1875, "learning_rate": 1.2510129478977594e-06, "loss": 1.319, "step": 33756 }, { "epoch": 84.3925, "grad_norm": 8.6875, "learning_rate": 1.2506208394580066e-06, "loss": 1.4821, "step": 33757 }, { "epoch": 84.395, "grad_norm": 7.625, "learning_rate": 1.2502287883788333e-06, "loss": 1.2663, "step": 33758 }, { "epoch": 84.3975, "grad_norm": 9.3125, "learning_rate": 1.2498367946628053e-06, "loss": 1.2478, "step": 33759 }, { "epoch": 84.4, "grad_norm": 6.8125, "learning_rate": 1.249444858312502e-06, "loss": 1.2509, "step": 33760 }, { "epoch": 84.4025, "grad_norm": 6.65625, "learning_rate": 1.2490529793304873e-06, "loss": 1.2443, "step": 33761 }, { "epoch": 84.405, "grad_norm": 7.09375, "learning_rate": 1.248661157719331e-06, "loss": 1.3104, "step": 33762 }, { "epoch": 84.4075, "grad_norm": 8.0, "learning_rate": 1.2482693934816003e-06, "loss": 1.3827, "step": 33763 }, { "epoch": 84.41, "grad_norm": 8.0625, "learning_rate": 1.247877686619865e-06, "loss": 1.3201, "step": 33764 }, { "epoch": 84.4125, "grad_norm": 5.53125, "learning_rate": 1.2474860371366959e-06, "loss": 1.2732, "step": 33765 }, { "epoch": 84.415, "grad_norm": 5.21875, "learning_rate": 1.2470944450346567e-06, "loss": 1.2673, "step": 33766 }, { "epoch": 84.4175, "grad_norm": 7.15625, "learning_rate": 1.2467029103163154e-06, "loss": 1.2969, "step": 33767 }, { "epoch": 84.42, "grad_norm": 8.375, "learning_rate": 1.2463114329842406e-06, "loss": 1.3839, "step": 33768 }, { "epoch": 84.4225, "grad_norm": 5.53125, "learning_rate": 1.2459200130409965e-06, "loss": 1.369, "step": 33769 }, { "epoch": 84.425, "grad_norm": 6.71875, "learning_rate": 1.2455286504891528e-06, "loss": 1.3231, "step": 33770 }, { "epoch": 84.4275, "grad_norm": 8.375, "learning_rate": 1.24513734533127e-06, "loss": 1.3379, "step": 33771 }, { "epoch": 84.43, "grad_norm": 6.4375, "learning_rate": 1.2447460975699189e-06, "loss": 1.4159, "step": 33772 }, { "epoch": 84.4325, "grad_norm": 10.5625, "learning_rate": 1.244354907207662e-06, "loss": 1.2993, "step": 33773 }, { "epoch": 84.435, "grad_norm": 7.65625, "learning_rate": 1.2439637742470623e-06, "loss": 1.2094, "step": 33774 }, { "epoch": 84.4375, "grad_norm": 10.625, "learning_rate": 1.243572698690685e-06, "loss": 1.3548, "step": 33775 }, { "epoch": 84.44, "grad_norm": 8.9375, "learning_rate": 1.2431816805410968e-06, "loss": 1.2925, "step": 33776 }, { "epoch": 84.4425, "grad_norm": 11.6875, "learning_rate": 1.2427907198008594e-06, "loss": 1.4038, "step": 33777 }, { "epoch": 84.445, "grad_norm": 8.375, "learning_rate": 1.2423998164725338e-06, "loss": 1.3392, "step": 33778 }, { "epoch": 84.4475, "grad_norm": 7.75, "learning_rate": 1.2420089705586835e-06, "loss": 1.2256, "step": 33779 }, { "epoch": 84.45, "grad_norm": 10.25, "learning_rate": 1.2416181820618745e-06, "loss": 1.1605, "step": 33780 }, { "epoch": 84.4525, "grad_norm": 6.90625, "learning_rate": 1.2412274509846666e-06, "loss": 1.2555, "step": 33781 }, { "epoch": 84.455, "grad_norm": 6.96875, "learning_rate": 1.2408367773296182e-06, "loss": 1.2492, "step": 33782 }, { "epoch": 84.4575, "grad_norm": 8.9375, "learning_rate": 1.2404461610992956e-06, "loss": 1.3462, "step": 33783 }, { "epoch": 84.46, "grad_norm": 10.0625, "learning_rate": 1.2400556022962562e-06, "loss": 1.313, "step": 33784 }, { "epoch": 84.4625, "grad_norm": 7.375, "learning_rate": 1.239665100923063e-06, "loss": 1.2986, "step": 33785 }, { "epoch": 84.465, "grad_norm": 15.25, "learning_rate": 1.2392746569822733e-06, "loss": 1.244, "step": 33786 }, { "epoch": 84.4675, "grad_norm": 6.78125, "learning_rate": 1.2388842704764503e-06, "loss": 1.314, "step": 33787 }, { "epoch": 84.47, "grad_norm": 7.6875, "learning_rate": 1.238493941408151e-06, "loss": 1.3038, "step": 33788 }, { "epoch": 84.4725, "grad_norm": 5.71875, "learning_rate": 1.238103669779933e-06, "loss": 1.4779, "step": 33789 }, { "epoch": 84.475, "grad_norm": 8.875, "learning_rate": 1.2377134555943571e-06, "loss": 1.3999, "step": 33790 }, { "epoch": 84.4775, "grad_norm": 7.40625, "learning_rate": 1.2373232988539829e-06, "loss": 1.3725, "step": 33791 }, { "epoch": 84.48, "grad_norm": 7.125, "learning_rate": 1.2369331995613664e-06, "loss": 1.3145, "step": 33792 }, { "epoch": 84.4825, "grad_norm": 8.75, "learning_rate": 1.2365431577190623e-06, "loss": 1.2149, "step": 33793 }, { "epoch": 84.485, "grad_norm": 6.5625, "learning_rate": 1.2361531733296317e-06, "loss": 1.3973, "step": 33794 }, { "epoch": 84.4875, "grad_norm": 6.90625, "learning_rate": 1.2357632463956315e-06, "loss": 1.3105, "step": 33795 }, { "epoch": 84.49, "grad_norm": 8.0625, "learning_rate": 1.2353733769196174e-06, "loss": 1.4065, "step": 33796 }, { "epoch": 84.4925, "grad_norm": 5.0, "learning_rate": 1.234983564904143e-06, "loss": 1.2635, "step": 33797 }, { "epoch": 84.495, "grad_norm": 8.4375, "learning_rate": 1.2345938103517662e-06, "loss": 1.3331, "step": 33798 }, { "epoch": 84.4975, "grad_norm": 7.6875, "learning_rate": 1.2342041132650406e-06, "loss": 1.4211, "step": 33799 }, { "epoch": 84.5, "grad_norm": 7.34375, "learning_rate": 1.233814473646524e-06, "loss": 1.2086, "step": 33800 }, { "epoch": 84.5025, "grad_norm": 8.25, "learning_rate": 1.2334248914987668e-06, "loss": 1.2835, "step": 33801 }, { "epoch": 84.505, "grad_norm": 9.75, "learning_rate": 1.2330353668243279e-06, "loss": 1.2688, "step": 33802 }, { "epoch": 84.5075, "grad_norm": 11.5625, "learning_rate": 1.2326458996257573e-06, "loss": 1.3123, "step": 33803 }, { "epoch": 84.51, "grad_norm": 5.4375, "learning_rate": 1.2322564899056077e-06, "loss": 1.3984, "step": 33804 }, { "epoch": 84.5125, "grad_norm": 11.1875, "learning_rate": 1.2318671376664325e-06, "loss": 1.2317, "step": 33805 }, { "epoch": 84.515, "grad_norm": 6.34375, "learning_rate": 1.2314778429107888e-06, "loss": 1.5337, "step": 33806 }, { "epoch": 84.5175, "grad_norm": 12.4375, "learning_rate": 1.2310886056412251e-06, "loss": 1.288, "step": 33807 }, { "epoch": 84.52, "grad_norm": 7.1875, "learning_rate": 1.2306994258602922e-06, "loss": 1.244, "step": 33808 }, { "epoch": 84.5225, "grad_norm": 7.59375, "learning_rate": 1.2303103035705422e-06, "loss": 1.2571, "step": 33809 }, { "epoch": 84.525, "grad_norm": 6.3125, "learning_rate": 1.2299212387745297e-06, "loss": 1.1208, "step": 33810 }, { "epoch": 84.5275, "grad_norm": 10.6875, "learning_rate": 1.2295322314748015e-06, "loss": 1.4894, "step": 33811 }, { "epoch": 84.53, "grad_norm": 6.1875, "learning_rate": 1.2291432816739079e-06, "loss": 1.3545, "step": 33812 }, { "epoch": 84.5325, "grad_norm": 12.5, "learning_rate": 1.2287543893744013e-06, "loss": 1.3825, "step": 33813 }, { "epoch": 84.535, "grad_norm": 5.90625, "learning_rate": 1.2283655545788286e-06, "loss": 1.278, "step": 33814 }, { "epoch": 84.5375, "grad_norm": 8.125, "learning_rate": 1.227976777289742e-06, "loss": 1.522, "step": 33815 }, { "epoch": 84.54, "grad_norm": 7.6875, "learning_rate": 1.2275880575096865e-06, "loss": 1.2614, "step": 33816 }, { "epoch": 84.5425, "grad_norm": 8.375, "learning_rate": 1.2271993952412141e-06, "loss": 1.5676, "step": 33817 }, { "epoch": 84.545, "grad_norm": 7.90625, "learning_rate": 1.22681079048687e-06, "loss": 1.2168, "step": 33818 }, { "epoch": 84.5475, "grad_norm": 8.875, "learning_rate": 1.2264222432492057e-06, "loss": 1.2931, "step": 33819 }, { "epoch": 84.55, "grad_norm": 7.71875, "learning_rate": 1.226033753530763e-06, "loss": 1.1985, "step": 33820 }, { "epoch": 84.5525, "grad_norm": 7.40625, "learning_rate": 1.2256453213340947e-06, "loss": 1.2599, "step": 33821 }, { "epoch": 84.555, "grad_norm": 7.40625, "learning_rate": 1.2252569466617447e-06, "loss": 1.5701, "step": 33822 }, { "epoch": 84.5575, "grad_norm": 7.65625, "learning_rate": 1.224868629516257e-06, "loss": 1.4238, "step": 33823 }, { "epoch": 84.56, "grad_norm": 5.5625, "learning_rate": 1.2244803699001785e-06, "loss": 1.3415, "step": 33824 }, { "epoch": 84.5625, "grad_norm": 8.5, "learning_rate": 1.224092167816059e-06, "loss": 1.2781, "step": 33825 }, { "epoch": 84.565, "grad_norm": 7.9375, "learning_rate": 1.2237040232664388e-06, "loss": 1.3855, "step": 33826 }, { "epoch": 84.5675, "grad_norm": 9.75, "learning_rate": 1.2233159362538615e-06, "loss": 1.339, "step": 33827 }, { "epoch": 84.57, "grad_norm": 7.875, "learning_rate": 1.2229279067808752e-06, "loss": 1.2351, "step": 33828 }, { "epoch": 84.5725, "grad_norm": 8.3125, "learning_rate": 1.2225399348500234e-06, "loss": 1.5248, "step": 33829 }, { "epoch": 84.575, "grad_norm": 10.875, "learning_rate": 1.222152020463848e-06, "loss": 1.3053, "step": 33830 }, { "epoch": 84.5775, "grad_norm": 11.5, "learning_rate": 1.2217641636248922e-06, "loss": 1.5068, "step": 33831 }, { "epoch": 84.58, "grad_norm": 7.8125, "learning_rate": 1.2213763643357002e-06, "loss": 1.3997, "step": 33832 }, { "epoch": 84.5825, "grad_norm": 8.875, "learning_rate": 1.220988622598811e-06, "loss": 1.3802, "step": 33833 }, { "epoch": 84.585, "grad_norm": 5.84375, "learning_rate": 1.2206009384167716e-06, "loss": 1.2623, "step": 33834 }, { "epoch": 84.5875, "grad_norm": 8.625, "learning_rate": 1.2202133117921199e-06, "loss": 1.1167, "step": 33835 }, { "epoch": 84.59, "grad_norm": 7.59375, "learning_rate": 1.2198257427273997e-06, "loss": 1.2622, "step": 33836 }, { "epoch": 84.5925, "grad_norm": 8.125, "learning_rate": 1.2194382312251507e-06, "loss": 1.537, "step": 33837 }, { "epoch": 84.595, "grad_norm": 9.6875, "learning_rate": 1.2190507772879111e-06, "loss": 1.4499, "step": 33838 }, { "epoch": 84.5975, "grad_norm": 6.6875, "learning_rate": 1.2186633809182225e-06, "loss": 1.3312, "step": 33839 }, { "epoch": 84.6, "grad_norm": 4.75, "learning_rate": 1.218276042118629e-06, "loss": 1.1349, "step": 33840 }, { "epoch": 84.6025, "grad_norm": 7.875, "learning_rate": 1.2178887608916645e-06, "loss": 1.3828, "step": 33841 }, { "epoch": 84.605, "grad_norm": 6.96875, "learning_rate": 1.217501537239869e-06, "loss": 1.2884, "step": 33842 }, { "epoch": 84.6075, "grad_norm": 10.5625, "learning_rate": 1.2171143711657818e-06, "loss": 1.3817, "step": 33843 }, { "epoch": 84.61, "grad_norm": 7.8125, "learning_rate": 1.216727262671944e-06, "loss": 1.2322, "step": 33844 }, { "epoch": 84.6125, "grad_norm": 6.625, "learning_rate": 1.216340211760889e-06, "loss": 1.2888, "step": 33845 }, { "epoch": 84.615, "grad_norm": 7.0, "learning_rate": 1.2159532184351551e-06, "loss": 1.4137, "step": 33846 }, { "epoch": 84.6175, "grad_norm": 7.1875, "learning_rate": 1.2155662826972836e-06, "loss": 1.4613, "step": 33847 }, { "epoch": 84.62, "grad_norm": 5.84375, "learning_rate": 1.2151794045498045e-06, "loss": 1.1351, "step": 33848 }, { "epoch": 84.6225, "grad_norm": 10.0625, "learning_rate": 1.2147925839952602e-06, "loss": 1.3074, "step": 33849 }, { "epoch": 84.625, "grad_norm": 7.0, "learning_rate": 1.214405821036182e-06, "loss": 1.3793, "step": 33850 }, { "epoch": 84.6275, "grad_norm": 8.125, "learning_rate": 1.2140191156751091e-06, "loss": 1.2979, "step": 33851 }, { "epoch": 84.63, "grad_norm": 7.59375, "learning_rate": 1.213632467914576e-06, "loss": 1.2644, "step": 33852 }, { "epoch": 84.6325, "grad_norm": 6.40625, "learning_rate": 1.2132458777571144e-06, "loss": 1.3487, "step": 33853 }, { "epoch": 84.635, "grad_norm": 5.78125, "learning_rate": 1.2128593452052618e-06, "loss": 1.2568, "step": 33854 }, { "epoch": 84.6375, "grad_norm": 8.375, "learning_rate": 1.2124728702615529e-06, "loss": 1.2731, "step": 33855 }, { "epoch": 84.64, "grad_norm": 5.21875, "learning_rate": 1.2120864529285203e-06, "loss": 1.1843, "step": 33856 }, { "epoch": 84.6425, "grad_norm": 7.53125, "learning_rate": 1.2117000932086964e-06, "loss": 1.2538, "step": 33857 }, { "epoch": 84.645, "grad_norm": 8.9375, "learning_rate": 1.2113137911046135e-06, "loss": 1.3187, "step": 33858 }, { "epoch": 84.6475, "grad_norm": 9.0625, "learning_rate": 1.2109275466188087e-06, "loss": 1.2096, "step": 33859 }, { "epoch": 84.65, "grad_norm": 16.0, "learning_rate": 1.2105413597538107e-06, "loss": 1.2324, "step": 33860 }, { "epoch": 84.6525, "grad_norm": 6.0625, "learning_rate": 1.21015523051215e-06, "loss": 1.2184, "step": 33861 }, { "epoch": 84.655, "grad_norm": 6.375, "learning_rate": 1.2097691588963622e-06, "loss": 1.4046, "step": 33862 }, { "epoch": 84.6575, "grad_norm": 7.3125, "learning_rate": 1.2093831449089731e-06, "loss": 1.6101, "step": 33863 }, { "epoch": 84.66, "grad_norm": 7.96875, "learning_rate": 1.2089971885525187e-06, "loss": 1.3747, "step": 33864 }, { "epoch": 84.6625, "grad_norm": 8.1875, "learning_rate": 1.2086112898295256e-06, "loss": 1.2428, "step": 33865 }, { "epoch": 84.665, "grad_norm": 8.375, "learning_rate": 1.208225448742526e-06, "loss": 1.3218, "step": 33866 }, { "epoch": 84.6675, "grad_norm": 10.125, "learning_rate": 1.2078396652940493e-06, "loss": 1.3063, "step": 33867 }, { "epoch": 84.67, "grad_norm": 6.59375, "learning_rate": 1.2074539394866215e-06, "loss": 1.2799, "step": 33868 }, { "epoch": 84.6725, "grad_norm": 6.96875, "learning_rate": 1.2070682713227744e-06, "loss": 1.559, "step": 33869 }, { "epoch": 84.675, "grad_norm": 7.34375, "learning_rate": 1.2066826608050375e-06, "loss": 1.3068, "step": 33870 }, { "epoch": 84.6775, "grad_norm": 11.125, "learning_rate": 1.2062971079359375e-06, "loss": 1.5726, "step": 33871 }, { "epoch": 84.68, "grad_norm": 10.75, "learning_rate": 1.2059116127179993e-06, "loss": 1.2557, "step": 33872 }, { "epoch": 84.6825, "grad_norm": 6.9375, "learning_rate": 1.205526175153753e-06, "loss": 1.2054, "step": 33873 }, { "epoch": 84.685, "grad_norm": 5.8125, "learning_rate": 1.2051407952457261e-06, "loss": 1.1832, "step": 33874 }, { "epoch": 84.6875, "grad_norm": 6.25, "learning_rate": 1.204755472996445e-06, "loss": 1.3891, "step": 33875 }, { "epoch": 84.69, "grad_norm": 6.46875, "learning_rate": 1.204370208408433e-06, "loss": 1.3247, "step": 33876 }, { "epoch": 84.6925, "grad_norm": 9.5625, "learning_rate": 1.2039850014842203e-06, "loss": 1.3151, "step": 33877 }, { "epoch": 84.695, "grad_norm": 7.46875, "learning_rate": 1.2035998522263282e-06, "loss": 1.3097, "step": 33878 }, { "epoch": 84.6975, "grad_norm": 6.25, "learning_rate": 1.2032147606372847e-06, "loss": 1.3024, "step": 33879 }, { "epoch": 84.7, "grad_norm": 7.15625, "learning_rate": 1.202829726719611e-06, "loss": 1.1941, "step": 33880 }, { "epoch": 84.7025, "grad_norm": 7.34375, "learning_rate": 1.2024447504758363e-06, "loss": 1.2293, "step": 33881 }, { "epoch": 84.705, "grad_norm": 5.46875, "learning_rate": 1.202059831908481e-06, "loss": 1.193, "step": 33882 }, { "epoch": 84.7075, "grad_norm": 5.875, "learning_rate": 1.2016749710200681e-06, "loss": 1.3714, "step": 33883 }, { "epoch": 84.71, "grad_norm": 6.75, "learning_rate": 1.2012901678131216e-06, "loss": 1.3034, "step": 33884 }, { "epoch": 84.7125, "grad_norm": 6.375, "learning_rate": 1.2009054222901662e-06, "loss": 1.2803, "step": 33885 }, { "epoch": 84.715, "grad_norm": 6.28125, "learning_rate": 1.200520734453724e-06, "loss": 1.2776, "step": 33886 }, { "epoch": 84.7175, "grad_norm": 11.75, "learning_rate": 1.2001361043063153e-06, "loss": 1.337, "step": 33887 }, { "epoch": 84.72, "grad_norm": 6.96875, "learning_rate": 1.199751531850457e-06, "loss": 1.2512, "step": 33888 }, { "epoch": 84.7225, "grad_norm": 8.375, "learning_rate": 1.1993670170886805e-06, "loss": 1.4239, "step": 33889 }, { "epoch": 84.725, "grad_norm": 6.375, "learning_rate": 1.1989825600235017e-06, "loss": 1.3668, "step": 33890 }, { "epoch": 84.7275, "grad_norm": 5.625, "learning_rate": 1.1985981606574381e-06, "loss": 1.307, "step": 33891 }, { "epoch": 84.73, "grad_norm": 6.03125, "learning_rate": 1.1982138189930159e-06, "loss": 1.261, "step": 33892 }, { "epoch": 84.7325, "grad_norm": 6.25, "learning_rate": 1.1978295350327496e-06, "loss": 1.3781, "step": 33893 }, { "epoch": 84.735, "grad_norm": 8.125, "learning_rate": 1.1974453087791627e-06, "loss": 1.2388, "step": 33894 }, { "epoch": 84.7375, "grad_norm": 12.4375, "learning_rate": 1.1970611402347687e-06, "loss": 1.3737, "step": 33895 }, { "epoch": 84.74, "grad_norm": 8.0625, "learning_rate": 1.1966770294020934e-06, "loss": 1.4129, "step": 33896 }, { "epoch": 84.7425, "grad_norm": 7.5, "learning_rate": 1.1962929762836495e-06, "loss": 1.3196, "step": 33897 }, { "epoch": 84.745, "grad_norm": 8.875, "learning_rate": 1.1959089808819557e-06, "loss": 1.4046, "step": 33898 }, { "epoch": 84.7475, "grad_norm": 5.875, "learning_rate": 1.1955250431995302e-06, "loss": 1.2567, "step": 33899 }, { "epoch": 84.75, "grad_norm": 9.75, "learning_rate": 1.195141163238892e-06, "loss": 1.2791, "step": 33900 }, { "epoch": 84.7525, "grad_norm": 7.09375, "learning_rate": 1.1947573410025558e-06, "loss": 1.2615, "step": 33901 }, { "epoch": 84.755, "grad_norm": 7.84375, "learning_rate": 1.1943735764930386e-06, "loss": 1.3253, "step": 33902 }, { "epoch": 84.7575, "grad_norm": 6.59375, "learning_rate": 1.1939898697128516e-06, "loss": 1.3598, "step": 33903 }, { "epoch": 84.76, "grad_norm": 7.71875, "learning_rate": 1.1936062206645183e-06, "loss": 1.3298, "step": 33904 }, { "epoch": 84.7625, "grad_norm": 7.25, "learning_rate": 1.193222629350551e-06, "loss": 1.249, "step": 33905 }, { "epoch": 84.765, "grad_norm": 8.25, "learning_rate": 1.1928390957734625e-06, "loss": 1.2469, "step": 33906 }, { "epoch": 84.7675, "grad_norm": 8.125, "learning_rate": 1.1924556199357673e-06, "loss": 1.3663, "step": 33907 }, { "epoch": 84.77, "grad_norm": 8.125, "learning_rate": 1.19207220183998e-06, "loss": 1.3385, "step": 33908 }, { "epoch": 84.7725, "grad_norm": 6.21875, "learning_rate": 1.1916888414886173e-06, "loss": 1.2313, "step": 33909 }, { "epoch": 84.775, "grad_norm": 7.625, "learning_rate": 1.191305538884191e-06, "loss": 1.2357, "step": 33910 }, { "epoch": 84.7775, "grad_norm": 7.90625, "learning_rate": 1.1909222940292098e-06, "loss": 1.2674, "step": 33911 }, { "epoch": 84.78, "grad_norm": 8.5625, "learning_rate": 1.1905391069261918e-06, "loss": 1.3323, "step": 33912 }, { "epoch": 84.7825, "grad_norm": 7.71875, "learning_rate": 1.190155977577645e-06, "loss": 1.3055, "step": 33913 }, { "epoch": 84.785, "grad_norm": 9.3125, "learning_rate": 1.1897729059860852e-06, "loss": 1.3298, "step": 33914 }, { "epoch": 84.7875, "grad_norm": 7.34375, "learning_rate": 1.1893898921540191e-06, "loss": 1.3822, "step": 33915 }, { "epoch": 84.79, "grad_norm": 7.3125, "learning_rate": 1.189006936083963e-06, "loss": 1.342, "step": 33916 }, { "epoch": 84.7925, "grad_norm": 7.03125, "learning_rate": 1.1886240377784241e-06, "loss": 1.2758, "step": 33917 }, { "epoch": 84.795, "grad_norm": 5.75, "learning_rate": 1.18824119723991e-06, "loss": 1.376, "step": 33918 }, { "epoch": 84.7975, "grad_norm": 6.8125, "learning_rate": 1.1878584144709383e-06, "loss": 1.3674, "step": 33919 }, { "epoch": 84.8, "grad_norm": 7.65625, "learning_rate": 1.1874756894740137e-06, "loss": 1.2883, "step": 33920 }, { "epoch": 84.8025, "grad_norm": 6.375, "learning_rate": 1.1870930222516452e-06, "loss": 1.2724, "step": 33921 }, { "epoch": 84.805, "grad_norm": 10.1875, "learning_rate": 1.1867104128063411e-06, "loss": 1.2365, "step": 33922 }, { "epoch": 84.8075, "grad_norm": 5.6875, "learning_rate": 1.1863278611406104e-06, "loss": 1.3256, "step": 33923 }, { "epoch": 84.81, "grad_norm": 7.875, "learning_rate": 1.1859453672569632e-06, "loss": 1.437, "step": 33924 }, { "epoch": 84.8125, "grad_norm": 7.71875, "learning_rate": 1.1855629311579065e-06, "loss": 1.2212, "step": 33925 }, { "epoch": 84.815, "grad_norm": 6.40625, "learning_rate": 1.1851805528459437e-06, "loss": 1.3082, "step": 33926 }, { "epoch": 84.8175, "grad_norm": 15.0625, "learning_rate": 1.1847982323235863e-06, "loss": 1.4709, "step": 33927 }, { "epoch": 84.82, "grad_norm": 7.78125, "learning_rate": 1.1844159695933377e-06, "loss": 1.356, "step": 33928 }, { "epoch": 84.8225, "grad_norm": 8.4375, "learning_rate": 1.184033764657706e-06, "loss": 1.2925, "step": 33929 }, { "epoch": 84.825, "grad_norm": 8.25, "learning_rate": 1.1836516175191947e-06, "loss": 1.3227, "step": 33930 }, { "epoch": 84.8275, "grad_norm": 8.5, "learning_rate": 1.183269528180313e-06, "loss": 1.359, "step": 33931 }, { "epoch": 84.83, "grad_norm": 9.25, "learning_rate": 1.1828874966435632e-06, "loss": 1.2399, "step": 33932 }, { "epoch": 84.8325, "grad_norm": 8.5, "learning_rate": 1.182505522911449e-06, "loss": 1.2848, "step": 33933 }, { "epoch": 84.835, "grad_norm": 6.78125, "learning_rate": 1.182123606986475e-06, "loss": 1.3375, "step": 33934 }, { "epoch": 84.8375, "grad_norm": 9.6875, "learning_rate": 1.181741748871147e-06, "loss": 1.2753, "step": 33935 }, { "epoch": 84.84, "grad_norm": 6.625, "learning_rate": 1.1813599485679684e-06, "loss": 1.2547, "step": 33936 }, { "epoch": 84.8425, "grad_norm": 6.34375, "learning_rate": 1.1809782060794383e-06, "loss": 1.3621, "step": 33937 }, { "epoch": 84.845, "grad_norm": 8.5625, "learning_rate": 1.1805965214080627e-06, "loss": 1.3515, "step": 33938 }, { "epoch": 84.8475, "grad_norm": 6.34375, "learning_rate": 1.1802148945563452e-06, "loss": 1.2517, "step": 33939 }, { "epoch": 84.85, "grad_norm": 6.96875, "learning_rate": 1.1798333255267857e-06, "loss": 1.3683, "step": 33940 }, { "epoch": 84.8525, "grad_norm": 6.78125, "learning_rate": 1.1794518143218835e-06, "loss": 1.2151, "step": 33941 }, { "epoch": 84.855, "grad_norm": 7.5, "learning_rate": 1.1790703609441456e-06, "loss": 1.4915, "step": 33942 }, { "epoch": 84.8575, "grad_norm": 6.375, "learning_rate": 1.1786889653960664e-06, "loss": 1.4092, "step": 33943 }, { "epoch": 84.86, "grad_norm": 5.6875, "learning_rate": 1.1783076276801508e-06, "loss": 1.4033, "step": 33944 }, { "epoch": 84.8625, "grad_norm": 7.40625, "learning_rate": 1.1779263477988966e-06, "loss": 1.2885, "step": 33945 }, { "epoch": 84.865, "grad_norm": 8.875, "learning_rate": 1.1775451257548053e-06, "loss": 1.3218, "step": 33946 }, { "epoch": 84.8675, "grad_norm": 7.875, "learning_rate": 1.1771639615503761e-06, "loss": 1.3725, "step": 33947 }, { "epoch": 84.87, "grad_norm": 7.5625, "learning_rate": 1.1767828551881045e-06, "loss": 1.3836, "step": 33948 }, { "epoch": 84.8725, "grad_norm": 7.28125, "learning_rate": 1.1764018066704908e-06, "loss": 1.3229, "step": 33949 }, { "epoch": 84.875, "grad_norm": 6.59375, "learning_rate": 1.1760208160000364e-06, "loss": 1.4142, "step": 33950 }, { "epoch": 84.8775, "grad_norm": 6.59375, "learning_rate": 1.175639883179236e-06, "loss": 1.269, "step": 33951 }, { "epoch": 84.88, "grad_norm": 6.625, "learning_rate": 1.1752590082105863e-06, "loss": 1.316, "step": 33952 }, { "epoch": 84.8825, "grad_norm": 6.78125, "learning_rate": 1.1748781910965846e-06, "loss": 1.4113, "step": 33953 }, { "epoch": 84.885, "grad_norm": 6.78125, "learning_rate": 1.1744974318397294e-06, "loss": 1.4368, "step": 33954 }, { "epoch": 84.8875, "grad_norm": 7.3125, "learning_rate": 1.1741167304425171e-06, "loss": 1.2794, "step": 33955 }, { "epoch": 84.89, "grad_norm": 7.15625, "learning_rate": 1.173736086907441e-06, "loss": 1.4349, "step": 33956 }, { "epoch": 84.8925, "grad_norm": 6.78125, "learning_rate": 1.173355501236999e-06, "loss": 1.3385, "step": 33957 }, { "epoch": 84.895, "grad_norm": 7.21875, "learning_rate": 1.1729749734336826e-06, "loss": 1.3976, "step": 33958 }, { "epoch": 84.8975, "grad_norm": 9.1875, "learning_rate": 1.172594503499992e-06, "loss": 1.6862, "step": 33959 }, { "epoch": 84.9, "grad_norm": 9.9375, "learning_rate": 1.1722140914384162e-06, "loss": 1.4239, "step": 33960 }, { "epoch": 84.9025, "grad_norm": 8.125, "learning_rate": 1.1718337372514532e-06, "loss": 1.5072, "step": 33961 }, { "epoch": 84.905, "grad_norm": 7.84375, "learning_rate": 1.1714534409415957e-06, "loss": 1.3683, "step": 33962 }, { "epoch": 84.9075, "grad_norm": 6.96875, "learning_rate": 1.1710732025113336e-06, "loss": 1.2872, "step": 33963 }, { "epoch": 84.91, "grad_norm": 6.875, "learning_rate": 1.1706930219631619e-06, "loss": 1.2465, "step": 33964 }, { "epoch": 84.9125, "grad_norm": 6.59375, "learning_rate": 1.170312899299576e-06, "loss": 1.1932, "step": 33965 }, { "epoch": 84.915, "grad_norm": 8.125, "learning_rate": 1.1699328345230654e-06, "loss": 1.2796, "step": 33966 }, { "epoch": 84.9175, "grad_norm": 10.3125, "learning_rate": 1.169552827636119e-06, "loss": 1.5181, "step": 33967 }, { "epoch": 84.92, "grad_norm": 10.375, "learning_rate": 1.1691728786412315e-06, "loss": 1.359, "step": 33968 }, { "epoch": 84.9225, "grad_norm": 9.375, "learning_rate": 1.1687929875408953e-06, "loss": 1.3657, "step": 33969 }, { "epoch": 84.925, "grad_norm": 9.3125, "learning_rate": 1.1684131543375987e-06, "loss": 1.3311, "step": 33970 }, { "epoch": 84.9275, "grad_norm": 10.5, "learning_rate": 1.1680333790338304e-06, "loss": 1.4142, "step": 33971 }, { "epoch": 84.93, "grad_norm": 6.375, "learning_rate": 1.1676536616320833e-06, "loss": 1.3108, "step": 33972 }, { "epoch": 84.9325, "grad_norm": 9.375, "learning_rate": 1.1672740021348438e-06, "loss": 1.3166, "step": 33973 }, { "epoch": 84.935, "grad_norm": 9.0625, "learning_rate": 1.1668944005446047e-06, "loss": 1.4694, "step": 33974 }, { "epoch": 84.9375, "grad_norm": 7.125, "learning_rate": 1.1665148568638496e-06, "loss": 1.3123, "step": 33975 }, { "epoch": 84.94, "grad_norm": 5.28125, "learning_rate": 1.166135371095073e-06, "loss": 1.1247, "step": 33976 }, { "epoch": 84.9425, "grad_norm": 7.8125, "learning_rate": 1.1657559432407584e-06, "loss": 1.2857, "step": 33977 }, { "epoch": 84.945, "grad_norm": 5.71875, "learning_rate": 1.1653765733033917e-06, "loss": 1.1336, "step": 33978 }, { "epoch": 84.9475, "grad_norm": 7.03125, "learning_rate": 1.1649972612854632e-06, "loss": 1.2101, "step": 33979 }, { "epoch": 84.95, "grad_norm": 8.875, "learning_rate": 1.1646180071894608e-06, "loss": 1.3137, "step": 33980 }, { "epoch": 84.9525, "grad_norm": 8.9375, "learning_rate": 1.1642388110178704e-06, "loss": 1.359, "step": 33981 }, { "epoch": 84.955, "grad_norm": 6.4375, "learning_rate": 1.1638596727731732e-06, "loss": 1.3248, "step": 33982 }, { "epoch": 84.9575, "grad_norm": 7.71875, "learning_rate": 1.1634805924578597e-06, "loss": 1.237, "step": 33983 }, { "epoch": 84.96, "grad_norm": 6.09375, "learning_rate": 1.1631015700744153e-06, "loss": 1.3701, "step": 33984 }, { "epoch": 84.9625, "grad_norm": 11.875, "learning_rate": 1.1627226056253228e-06, "loss": 1.2164, "step": 33985 }, { "epoch": 84.965, "grad_norm": 7.5625, "learning_rate": 1.1623436991130654e-06, "loss": 1.466, "step": 33986 }, { "epoch": 84.9675, "grad_norm": 7.4375, "learning_rate": 1.1619648505401316e-06, "loss": 1.4385, "step": 33987 }, { "epoch": 84.97, "grad_norm": 5.59375, "learning_rate": 1.1615860599090012e-06, "loss": 1.5229, "step": 33988 }, { "epoch": 84.9725, "grad_norm": 6.28125, "learning_rate": 1.1612073272221592e-06, "loss": 1.271, "step": 33989 }, { "epoch": 84.975, "grad_norm": 5.90625, "learning_rate": 1.1608286524820877e-06, "loss": 1.3494, "step": 33990 }, { "epoch": 84.9775, "grad_norm": 6.71875, "learning_rate": 1.1604500356912706e-06, "loss": 1.4113, "step": 33991 }, { "epoch": 84.98, "grad_norm": 6.625, "learning_rate": 1.1600714768521903e-06, "loss": 1.2295, "step": 33992 }, { "epoch": 84.9825, "grad_norm": 7.46875, "learning_rate": 1.159692975967326e-06, "loss": 1.2136, "step": 33993 }, { "epoch": 84.985, "grad_norm": 7.46875, "learning_rate": 1.1593145330391598e-06, "loss": 1.3082, "step": 33994 }, { "epoch": 84.9875, "grad_norm": 12.0625, "learning_rate": 1.1589361480701755e-06, "loss": 1.3442, "step": 33995 }, { "epoch": 84.99, "grad_norm": 7.21875, "learning_rate": 1.1585578210628524e-06, "loss": 1.3531, "step": 33996 }, { "epoch": 84.9925, "grad_norm": 6.625, "learning_rate": 1.1581795520196693e-06, "loss": 1.3226, "step": 33997 }, { "epoch": 84.995, "grad_norm": 15.125, "learning_rate": 1.1578013409431055e-06, "loss": 1.2765, "step": 33998 }, { "epoch": 84.9975, "grad_norm": 10.4375, "learning_rate": 1.1574231878356457e-06, "loss": 1.2348, "step": 33999 }, { "epoch": 85.0, "grad_norm": 6.5625, "learning_rate": 1.1570450926997657e-06, "loss": 1.2349, "step": 34000 }, { "epoch": 85.0025, "grad_norm": 7.21875, "learning_rate": 1.1566670555379412e-06, "loss": 1.2196, "step": 34001 }, { "epoch": 85.005, "grad_norm": 7.9375, "learning_rate": 1.1562890763526558e-06, "loss": 1.2722, "step": 34002 }, { "epoch": 85.0075, "grad_norm": 5.71875, "learning_rate": 1.1559111551463842e-06, "loss": 1.277, "step": 34003 }, { "epoch": 85.01, "grad_norm": 11.0, "learning_rate": 1.1555332919216057e-06, "loss": 1.3145, "step": 34004 }, { "epoch": 85.0125, "grad_norm": 6.9375, "learning_rate": 1.1551554866807956e-06, "loss": 1.2124, "step": 34005 }, { "epoch": 85.015, "grad_norm": 8.375, "learning_rate": 1.1547777394264347e-06, "loss": 1.2062, "step": 34006 }, { "epoch": 85.0175, "grad_norm": 8.0, "learning_rate": 1.1544000501609965e-06, "loss": 1.187, "step": 34007 }, { "epoch": 85.02, "grad_norm": 11.0, "learning_rate": 1.1540224188869565e-06, "loss": 1.2101, "step": 34008 }, { "epoch": 85.0225, "grad_norm": 11.375, "learning_rate": 1.1536448456067905e-06, "loss": 1.5222, "step": 34009 }, { "epoch": 85.025, "grad_norm": 9.4375, "learning_rate": 1.153267330322977e-06, "loss": 1.3178, "step": 34010 }, { "epoch": 85.0275, "grad_norm": 6.15625, "learning_rate": 1.152889873037989e-06, "loss": 1.2298, "step": 34011 }, { "epoch": 85.03, "grad_norm": 8.75, "learning_rate": 1.1525124737542992e-06, "loss": 1.2964, "step": 34012 }, { "epoch": 85.0325, "grad_norm": 6.5, "learning_rate": 1.1521351324743835e-06, "loss": 1.175, "step": 34013 }, { "epoch": 85.035, "grad_norm": 8.6875, "learning_rate": 1.1517578492007176e-06, "loss": 1.2425, "step": 34014 }, { "epoch": 85.0375, "grad_norm": 5.34375, "learning_rate": 1.1513806239357728e-06, "loss": 1.3721, "step": 34015 }, { "epoch": 85.04, "grad_norm": 8.5, "learning_rate": 1.1510034566820205e-06, "loss": 1.1164, "step": 34016 }, { "epoch": 85.0425, "grad_norm": 7.40625, "learning_rate": 1.1506263474419388e-06, "loss": 1.2116, "step": 34017 }, { "epoch": 85.045, "grad_norm": 6.40625, "learning_rate": 1.1502492962179934e-06, "loss": 1.3667, "step": 34018 }, { "epoch": 85.0475, "grad_norm": 11.25, "learning_rate": 1.1498723030126612e-06, "loss": 1.3138, "step": 34019 }, { "epoch": 85.05, "grad_norm": 9.8125, "learning_rate": 1.1494953678284105e-06, "loss": 1.3761, "step": 34020 }, { "epoch": 85.0525, "grad_norm": 7.78125, "learning_rate": 1.1491184906677166e-06, "loss": 1.3294, "step": 34021 }, { "epoch": 85.055, "grad_norm": 5.78125, "learning_rate": 1.148741671533047e-06, "loss": 1.4089, "step": 34022 }, { "epoch": 85.0575, "grad_norm": 6.5, "learning_rate": 1.1483649104268712e-06, "loss": 1.5313, "step": 34023 }, { "epoch": 85.06, "grad_norm": 8.75, "learning_rate": 1.14798820735166e-06, "loss": 1.2972, "step": 34024 }, { "epoch": 85.0625, "grad_norm": 10.5, "learning_rate": 1.147611562309887e-06, "loss": 1.3819, "step": 34025 }, { "epoch": 85.065, "grad_norm": 6.875, "learning_rate": 1.147234975304018e-06, "loss": 1.2702, "step": 34026 }, { "epoch": 85.0675, "grad_norm": 6.53125, "learning_rate": 1.1468584463365218e-06, "loss": 1.3271, "step": 34027 }, { "epoch": 85.07, "grad_norm": 6.1875, "learning_rate": 1.1464819754098633e-06, "loss": 1.2839, "step": 34028 }, { "epoch": 85.0725, "grad_norm": 7.1875, "learning_rate": 1.1461055625265195e-06, "loss": 1.2598, "step": 34029 }, { "epoch": 85.075, "grad_norm": 8.1875, "learning_rate": 1.1457292076889525e-06, "loss": 1.3852, "step": 34030 }, { "epoch": 85.0775, "grad_norm": 5.9375, "learning_rate": 1.1453529108996285e-06, "loss": 1.3863, "step": 34031 }, { "epoch": 85.08, "grad_norm": 8.3125, "learning_rate": 1.144976672161019e-06, "loss": 1.2438, "step": 34032 }, { "epoch": 85.0825, "grad_norm": 8.625, "learning_rate": 1.1446004914755849e-06, "loss": 1.4766, "step": 34033 }, { "epoch": 85.085, "grad_norm": 8.375, "learning_rate": 1.1442243688457987e-06, "loss": 1.3041, "step": 34034 }, { "epoch": 85.0875, "grad_norm": 6.5, "learning_rate": 1.143848304274121e-06, "loss": 1.2461, "step": 34035 }, { "epoch": 85.09, "grad_norm": 8.75, "learning_rate": 1.143472297763022e-06, "loss": 1.2094, "step": 34036 }, { "epoch": 85.0925, "grad_norm": 9.8125, "learning_rate": 1.1430963493149627e-06, "loss": 1.3131, "step": 34037 }, { "epoch": 85.095, "grad_norm": 7.5625, "learning_rate": 1.1427204589324081e-06, "loss": 1.2207, "step": 34038 }, { "epoch": 85.0975, "grad_norm": 6.09375, "learning_rate": 1.1423446266178239e-06, "loss": 1.5171, "step": 34039 }, { "epoch": 85.1, "grad_norm": 6.40625, "learning_rate": 1.1419688523736761e-06, "loss": 1.2855, "step": 34040 }, { "epoch": 85.1025, "grad_norm": 7.40625, "learning_rate": 1.1415931362024257e-06, "loss": 1.505, "step": 34041 }, { "epoch": 85.105, "grad_norm": 5.3125, "learning_rate": 1.1412174781065367e-06, "loss": 1.2552, "step": 34042 }, { "epoch": 85.1075, "grad_norm": 9.3125, "learning_rate": 1.1408418780884656e-06, "loss": 1.4196, "step": 34043 }, { "epoch": 85.11, "grad_norm": 4.78125, "learning_rate": 1.1404663361506863e-06, "loss": 1.2179, "step": 34044 }, { "epoch": 85.1125, "grad_norm": 4.90625, "learning_rate": 1.1400908522956545e-06, "loss": 1.3599, "step": 34045 }, { "epoch": 85.115, "grad_norm": 6.09375, "learning_rate": 1.1397154265258337e-06, "loss": 1.3822, "step": 34046 }, { "epoch": 85.1175, "grad_norm": 7.5, "learning_rate": 1.139340058843681e-06, "loss": 1.4026, "step": 34047 }, { "epoch": 85.12, "grad_norm": 9.75, "learning_rate": 1.1389647492516598e-06, "loss": 1.2147, "step": 34048 }, { "epoch": 85.1225, "grad_norm": 7.5, "learning_rate": 1.1385894977522338e-06, "loss": 1.4193, "step": 34049 }, { "epoch": 85.125, "grad_norm": 6.5, "learning_rate": 1.1382143043478599e-06, "loss": 1.3593, "step": 34050 }, { "epoch": 85.1275, "grad_norm": 6.8125, "learning_rate": 1.1378391690409973e-06, "loss": 1.2578, "step": 34051 }, { "epoch": 85.13, "grad_norm": 6.875, "learning_rate": 1.1374640918341074e-06, "loss": 1.4231, "step": 34052 }, { "epoch": 85.1325, "grad_norm": 5.53125, "learning_rate": 1.137089072729647e-06, "loss": 1.1961, "step": 34053 }, { "epoch": 85.135, "grad_norm": 7.0, "learning_rate": 1.1367141117300783e-06, "loss": 1.4411, "step": 34054 }, { "epoch": 85.1375, "grad_norm": 6.625, "learning_rate": 1.1363392088378545e-06, "loss": 1.3756, "step": 34055 }, { "epoch": 85.14, "grad_norm": 9.4375, "learning_rate": 1.1359643640554384e-06, "loss": 1.2359, "step": 34056 }, { "epoch": 85.1425, "grad_norm": 9.3125, "learning_rate": 1.1355895773852854e-06, "loss": 1.3204, "step": 34057 }, { "epoch": 85.145, "grad_norm": 6.8125, "learning_rate": 1.1352148488298475e-06, "loss": 1.3571, "step": 34058 }, { "epoch": 85.1475, "grad_norm": 8.0, "learning_rate": 1.1348401783915907e-06, "loss": 1.2622, "step": 34059 }, { "epoch": 85.15, "grad_norm": 6.84375, "learning_rate": 1.1344655660729676e-06, "loss": 1.2657, "step": 34060 }, { "epoch": 85.1525, "grad_norm": 7.0625, "learning_rate": 1.1340910118764326e-06, "loss": 1.2872, "step": 34061 }, { "epoch": 85.155, "grad_norm": 6.0625, "learning_rate": 1.1337165158044394e-06, "loss": 1.2861, "step": 34062 }, { "epoch": 85.1575, "grad_norm": 8.4375, "learning_rate": 1.1333420778594484e-06, "loss": 1.3347, "step": 34063 }, { "epoch": 85.16, "grad_norm": 8.375, "learning_rate": 1.132967698043913e-06, "loss": 1.3512, "step": 34064 }, { "epoch": 85.1625, "grad_norm": 9.375, "learning_rate": 1.132593376360286e-06, "loss": 1.3699, "step": 34065 }, { "epoch": 85.165, "grad_norm": 7.96875, "learning_rate": 1.1322191128110205e-06, "loss": 1.3362, "step": 34066 }, { "epoch": 85.1675, "grad_norm": 6.03125, "learning_rate": 1.1318449073985704e-06, "loss": 1.3776, "step": 34067 }, { "epoch": 85.17, "grad_norm": 8.1875, "learning_rate": 1.1314707601253938e-06, "loss": 1.1878, "step": 34068 }, { "epoch": 85.1725, "grad_norm": 8.625, "learning_rate": 1.1310966709939386e-06, "loss": 1.3054, "step": 34069 }, { "epoch": 85.175, "grad_norm": 5.875, "learning_rate": 1.1307226400066573e-06, "loss": 1.2251, "step": 34070 }, { "epoch": 85.1775, "grad_norm": 8.25, "learning_rate": 1.1303486671660058e-06, "loss": 1.3184, "step": 34071 }, { "epoch": 85.18, "grad_norm": 6.09375, "learning_rate": 1.1299747524744309e-06, "loss": 1.3363, "step": 34072 }, { "epoch": 85.1825, "grad_norm": 9.375, "learning_rate": 1.1296008959343884e-06, "loss": 1.3974, "step": 34073 }, { "epoch": 85.185, "grad_norm": 14.3125, "learning_rate": 1.1292270975483266e-06, "loss": 1.3223, "step": 34074 }, { "epoch": 85.1875, "grad_norm": 6.8125, "learning_rate": 1.1288533573186976e-06, "loss": 1.161, "step": 34075 }, { "epoch": 85.19, "grad_norm": 9.0, "learning_rate": 1.128479675247952e-06, "loss": 1.2785, "step": 34076 }, { "epoch": 85.1925, "grad_norm": 9.125, "learning_rate": 1.1281060513385367e-06, "loss": 1.3141, "step": 34077 }, { "epoch": 85.195, "grad_norm": 6.96875, "learning_rate": 1.1277324855929027e-06, "loss": 1.3664, "step": 34078 }, { "epoch": 85.1975, "grad_norm": 6.5, "learning_rate": 1.1273589780135007e-06, "loss": 1.2264, "step": 34079 }, { "epoch": 85.2, "grad_norm": 8.9375, "learning_rate": 1.1269855286027798e-06, "loss": 1.3066, "step": 34080 }, { "epoch": 85.2025, "grad_norm": 8.5625, "learning_rate": 1.1266121373631834e-06, "loss": 1.3657, "step": 34081 }, { "epoch": 85.205, "grad_norm": 6.75, "learning_rate": 1.1262388042971627e-06, "loss": 1.2457, "step": 34082 }, { "epoch": 85.2075, "grad_norm": 7.59375, "learning_rate": 1.1258655294071686e-06, "loss": 1.1203, "step": 34083 }, { "epoch": 85.21, "grad_norm": 7.8125, "learning_rate": 1.1254923126956441e-06, "loss": 1.4198, "step": 34084 }, { "epoch": 85.2125, "grad_norm": 10.25, "learning_rate": 1.125119154165034e-06, "loss": 1.3323, "step": 34085 }, { "epoch": 85.215, "grad_norm": 9.0625, "learning_rate": 1.1247460538177911e-06, "loss": 1.3106, "step": 34086 }, { "epoch": 85.2175, "grad_norm": 6.375, "learning_rate": 1.1243730116563544e-06, "loss": 1.2547, "step": 34087 }, { "epoch": 85.22, "grad_norm": 7.6875, "learning_rate": 1.1240000276831765e-06, "loss": 1.2963, "step": 34088 }, { "epoch": 85.2225, "grad_norm": 7.0, "learning_rate": 1.1236271019006962e-06, "loss": 1.2891, "step": 34089 }, { "epoch": 85.225, "grad_norm": 6.15625, "learning_rate": 1.123254234311364e-06, "loss": 1.2532, "step": 34090 }, { "epoch": 85.2275, "grad_norm": 5.9375, "learning_rate": 1.1228814249176224e-06, "loss": 1.3061, "step": 34091 }, { "epoch": 85.23, "grad_norm": 6.6875, "learning_rate": 1.1225086737219114e-06, "loss": 1.3584, "step": 34092 }, { "epoch": 85.2325, "grad_norm": 8.75, "learning_rate": 1.1221359807266796e-06, "loss": 1.2968, "step": 34093 }, { "epoch": 85.235, "grad_norm": 5.875, "learning_rate": 1.1217633459343702e-06, "loss": 1.159, "step": 34094 }, { "epoch": 85.2375, "grad_norm": 5.65625, "learning_rate": 1.1213907693474246e-06, "loss": 1.2024, "step": 34095 }, { "epoch": 85.24, "grad_norm": 9.5, "learning_rate": 1.1210182509682854e-06, "loss": 1.2168, "step": 34096 }, { "epoch": 85.2425, "grad_norm": 6.34375, "learning_rate": 1.120645790799394e-06, "loss": 1.1127, "step": 34097 }, { "epoch": 85.245, "grad_norm": 7.0, "learning_rate": 1.120273388843196e-06, "loss": 1.2924, "step": 34098 }, { "epoch": 85.2475, "grad_norm": 6.71875, "learning_rate": 1.1199010451021297e-06, "loss": 1.1745, "step": 34099 }, { "epoch": 85.25, "grad_norm": 8.625, "learning_rate": 1.1195287595786352e-06, "loss": 1.2062, "step": 34100 }, { "epoch": 85.2525, "grad_norm": 7.46875, "learning_rate": 1.1191565322751575e-06, "loss": 1.3602, "step": 34101 }, { "epoch": 85.255, "grad_norm": 9.375, "learning_rate": 1.118784363194131e-06, "loss": 1.2838, "step": 34102 }, { "epoch": 85.2575, "grad_norm": 8.3125, "learning_rate": 1.1184122523380014e-06, "loss": 1.3668, "step": 34103 }, { "epoch": 85.26, "grad_norm": 6.125, "learning_rate": 1.118040199709204e-06, "loss": 1.2358, "step": 34104 }, { "epoch": 85.2625, "grad_norm": 9.4375, "learning_rate": 1.1176682053101806e-06, "loss": 1.2751, "step": 34105 }, { "epoch": 85.265, "grad_norm": 7.1875, "learning_rate": 1.1172962691433699e-06, "loss": 1.3847, "step": 34106 }, { "epoch": 85.2675, "grad_norm": 6.71875, "learning_rate": 1.1169243912112072e-06, "loss": 1.2975, "step": 34107 }, { "epoch": 85.27, "grad_norm": 6.0, "learning_rate": 1.1165525715161317e-06, "loss": 1.1444, "step": 34108 }, { "epoch": 85.2725, "grad_norm": 6.84375, "learning_rate": 1.1161808100605854e-06, "loss": 1.3492, "step": 34109 }, { "epoch": 85.275, "grad_norm": 7.84375, "learning_rate": 1.1158091068470012e-06, "loss": 1.2618, "step": 34110 }, { "epoch": 85.2775, "grad_norm": 5.71875, "learning_rate": 1.1154374618778152e-06, "loss": 1.2415, "step": 34111 }, { "epoch": 85.28, "grad_norm": 11.9375, "learning_rate": 1.1150658751554667e-06, "loss": 1.3843, "step": 34112 }, { "epoch": 85.2825, "grad_norm": 5.96875, "learning_rate": 1.1146943466823913e-06, "loss": 1.3177, "step": 34113 }, { "epoch": 85.285, "grad_norm": 11.9375, "learning_rate": 1.1143228764610248e-06, "loss": 1.3765, "step": 34114 }, { "epoch": 85.2875, "grad_norm": 8.5, "learning_rate": 1.113951464493801e-06, "loss": 1.4779, "step": 34115 }, { "epoch": 85.29, "grad_norm": 7.8125, "learning_rate": 1.1135801107831568e-06, "loss": 1.3264, "step": 34116 }, { "epoch": 85.2925, "grad_norm": 9.1875, "learning_rate": 1.1132088153315235e-06, "loss": 1.2651, "step": 34117 }, { "epoch": 85.295, "grad_norm": 8.875, "learning_rate": 1.1128375781413392e-06, "loss": 1.2508, "step": 34118 }, { "epoch": 85.2975, "grad_norm": 7.125, "learning_rate": 1.1124663992150353e-06, "loss": 1.2685, "step": 34119 }, { "epoch": 85.3, "grad_norm": 7.5625, "learning_rate": 1.1120952785550477e-06, "loss": 1.363, "step": 34120 }, { "epoch": 85.3025, "grad_norm": 8.25, "learning_rate": 1.1117242161638075e-06, "loss": 1.2789, "step": 34121 }, { "epoch": 85.305, "grad_norm": 7.53125, "learning_rate": 1.1113532120437454e-06, "loss": 1.3654, "step": 34122 }, { "epoch": 85.3075, "grad_norm": 8.125, "learning_rate": 1.110982266197297e-06, "loss": 1.2538, "step": 34123 }, { "epoch": 85.31, "grad_norm": 5.4375, "learning_rate": 1.1106113786268935e-06, "loss": 1.2138, "step": 34124 }, { "epoch": 85.3125, "grad_norm": 7.46875, "learning_rate": 1.1102405493349676e-06, "loss": 1.3671, "step": 34125 }, { "epoch": 85.315, "grad_norm": 6.875, "learning_rate": 1.1098697783239465e-06, "loss": 1.2698, "step": 34126 }, { "epoch": 85.3175, "grad_norm": 7.09375, "learning_rate": 1.1094990655962623e-06, "loss": 1.3904, "step": 34127 }, { "epoch": 85.32, "grad_norm": 10.5, "learning_rate": 1.1091284111543499e-06, "loss": 1.4318, "step": 34128 }, { "epoch": 85.3225, "grad_norm": 35.25, "learning_rate": 1.1087578150006351e-06, "loss": 1.2788, "step": 34129 }, { "epoch": 85.325, "grad_norm": 6.5625, "learning_rate": 1.1083872771375458e-06, "loss": 1.2515, "step": 34130 }, { "epoch": 85.3275, "grad_norm": 8.5625, "learning_rate": 1.108016797567517e-06, "loss": 1.3596, "step": 34131 }, { "epoch": 85.33, "grad_norm": 9.25, "learning_rate": 1.1076463762929712e-06, "loss": 1.3643, "step": 34132 }, { "epoch": 85.3325, "grad_norm": 9.0, "learning_rate": 1.107276013316343e-06, "loss": 1.3367, "step": 34133 }, { "epoch": 85.335, "grad_norm": 6.75, "learning_rate": 1.1069057086400536e-06, "loss": 1.2065, "step": 34134 }, { "epoch": 85.3375, "grad_norm": 7.125, "learning_rate": 1.106535462266538e-06, "loss": 1.321, "step": 34135 }, { "epoch": 85.34, "grad_norm": 10.25, "learning_rate": 1.1061652741982198e-06, "loss": 1.4039, "step": 34136 }, { "epoch": 85.3425, "grad_norm": 8.5625, "learning_rate": 1.1057951444375237e-06, "loss": 1.3832, "step": 34137 }, { "epoch": 85.345, "grad_norm": 5.9375, "learning_rate": 1.1054250729868787e-06, "loss": 1.2615, "step": 34138 }, { "epoch": 85.3475, "grad_norm": 8.25, "learning_rate": 1.1050550598487132e-06, "loss": 1.333, "step": 34139 }, { "epoch": 85.35, "grad_norm": 9.8125, "learning_rate": 1.1046851050254504e-06, "loss": 1.2861, "step": 34140 }, { "epoch": 85.3525, "grad_norm": 7.1875, "learning_rate": 1.1043152085195152e-06, "loss": 1.2003, "step": 34141 }, { "epoch": 85.355, "grad_norm": 10.625, "learning_rate": 1.1039453703333324e-06, "loss": 1.348, "step": 34142 }, { "epoch": 85.3575, "grad_norm": 6.875, "learning_rate": 1.1035755904693302e-06, "loss": 1.3888, "step": 34143 }, { "epoch": 85.36, "grad_norm": 8.0, "learning_rate": 1.1032058689299297e-06, "loss": 1.2139, "step": 34144 }, { "epoch": 85.3625, "grad_norm": 7.4375, "learning_rate": 1.1028362057175546e-06, "loss": 1.4045, "step": 34145 }, { "epoch": 85.365, "grad_norm": 8.1875, "learning_rate": 1.1024666008346307e-06, "loss": 1.2747, "step": 34146 }, { "epoch": 85.3675, "grad_norm": 8.9375, "learning_rate": 1.1020970542835774e-06, "loss": 1.4566, "step": 34147 }, { "epoch": 85.37, "grad_norm": 6.78125, "learning_rate": 1.1017275660668224e-06, "loss": 1.4118, "step": 34148 }, { "epoch": 85.3725, "grad_norm": 7.21875, "learning_rate": 1.1013581361867832e-06, "loss": 1.4929, "step": 34149 }, { "epoch": 85.375, "grad_norm": 6.09375, "learning_rate": 1.1009887646458862e-06, "loss": 1.1165, "step": 34150 }, { "epoch": 85.3775, "grad_norm": 8.375, "learning_rate": 1.1006194514465497e-06, "loss": 1.2215, "step": 34151 }, { "epoch": 85.38, "grad_norm": 6.3125, "learning_rate": 1.100250196591195e-06, "loss": 1.3589, "step": 34152 }, { "epoch": 85.3825, "grad_norm": 8.8125, "learning_rate": 1.0998810000822435e-06, "loss": 1.2902, "step": 34153 }, { "epoch": 85.385, "grad_norm": 5.15625, "learning_rate": 1.099511861922118e-06, "loss": 1.3595, "step": 34154 }, { "epoch": 85.3875, "grad_norm": 6.78125, "learning_rate": 1.0991427821132373e-06, "loss": 1.3514, "step": 34155 }, { "epoch": 85.39, "grad_norm": 8.75, "learning_rate": 1.0987737606580174e-06, "loss": 1.3476, "step": 34156 }, { "epoch": 85.3925, "grad_norm": 6.375, "learning_rate": 1.0984047975588807e-06, "loss": 1.2371, "step": 34157 }, { "epoch": 85.395, "grad_norm": 9.125, "learning_rate": 1.0980358928182477e-06, "loss": 1.2037, "step": 34158 }, { "epoch": 85.3975, "grad_norm": 7.03125, "learning_rate": 1.0976670464385352e-06, "loss": 1.3657, "step": 34159 }, { "epoch": 85.4, "grad_norm": 11.625, "learning_rate": 1.0972982584221592e-06, "loss": 1.327, "step": 34160 }, { "epoch": 85.4025, "grad_norm": 9.75, "learning_rate": 1.096929528771542e-06, "loss": 1.5442, "step": 34161 }, { "epoch": 85.405, "grad_norm": 6.46875, "learning_rate": 1.0965608574890963e-06, "loss": 1.3703, "step": 34162 }, { "epoch": 85.4075, "grad_norm": 7.09375, "learning_rate": 1.0961922445772432e-06, "loss": 1.4009, "step": 34163 }, { "epoch": 85.41, "grad_norm": 6.9375, "learning_rate": 1.0958236900383957e-06, "loss": 1.241, "step": 34164 }, { "epoch": 85.4125, "grad_norm": 8.6875, "learning_rate": 1.0954551938749735e-06, "loss": 1.1233, "step": 34165 }, { "epoch": 85.415, "grad_norm": 6.40625, "learning_rate": 1.0950867560893918e-06, "loss": 1.3697, "step": 34166 }, { "epoch": 85.4175, "grad_norm": 7.875, "learning_rate": 1.094718376684062e-06, "loss": 1.3611, "step": 34167 }, { "epoch": 85.42, "grad_norm": 10.0, "learning_rate": 1.0943500556614017e-06, "loss": 1.3244, "step": 34168 }, { "epoch": 85.4225, "grad_norm": 7.15625, "learning_rate": 1.0939817930238283e-06, "loss": 1.3852, "step": 34169 }, { "epoch": 85.425, "grad_norm": 10.5625, "learning_rate": 1.0936135887737542e-06, "loss": 1.3154, "step": 34170 }, { "epoch": 85.4275, "grad_norm": 7.6875, "learning_rate": 1.093245442913592e-06, "loss": 1.3446, "step": 34171 }, { "epoch": 85.43, "grad_norm": 9.4375, "learning_rate": 1.092877355445755e-06, "loss": 1.3948, "step": 34172 }, { "epoch": 85.4325, "grad_norm": 7.9375, "learning_rate": 1.0925093263726594e-06, "loss": 1.4554, "step": 34173 }, { "epoch": 85.435, "grad_norm": 8.25, "learning_rate": 1.0921413556967164e-06, "loss": 1.4034, "step": 34174 }, { "epoch": 85.4375, "grad_norm": 6.78125, "learning_rate": 1.0917734434203365e-06, "loss": 1.21, "step": 34175 }, { "epoch": 85.44, "grad_norm": 5.875, "learning_rate": 1.0914055895459353e-06, "loss": 1.1668, "step": 34176 }, { "epoch": 85.4425, "grad_norm": 7.25, "learning_rate": 1.091037794075921e-06, "loss": 1.3364, "step": 34177 }, { "epoch": 85.445, "grad_norm": 6.875, "learning_rate": 1.0906700570127081e-06, "loss": 1.3989, "step": 34178 }, { "epoch": 85.4475, "grad_norm": 7.96875, "learning_rate": 1.090302378358703e-06, "loss": 1.5103, "step": 34179 }, { "epoch": 85.45, "grad_norm": 5.96875, "learning_rate": 1.0899347581163222e-06, "loss": 1.3231, "step": 34180 }, { "epoch": 85.4525, "grad_norm": 8.8125, "learning_rate": 1.0895671962879716e-06, "loss": 1.3241, "step": 34181 }, { "epoch": 85.455, "grad_norm": 8.3125, "learning_rate": 1.0891996928760606e-06, "loss": 1.5034, "step": 34182 }, { "epoch": 85.4575, "grad_norm": 8.625, "learning_rate": 1.0888322478829993e-06, "loss": 1.2103, "step": 34183 }, { "epoch": 85.46, "grad_norm": 7.8125, "learning_rate": 1.0884648613111992e-06, "loss": 1.2894, "step": 34184 }, { "epoch": 85.4625, "grad_norm": 9.1875, "learning_rate": 1.0880975331630672e-06, "loss": 1.3992, "step": 34185 }, { "epoch": 85.465, "grad_norm": 7.5, "learning_rate": 1.0877302634410115e-06, "loss": 1.3309, "step": 34186 }, { "epoch": 85.4675, "grad_norm": 8.1875, "learning_rate": 1.0873630521474343e-06, "loss": 1.2347, "step": 34187 }, { "epoch": 85.47, "grad_norm": 12.625, "learning_rate": 1.086995899284753e-06, "loss": 1.4946, "step": 34188 }, { "epoch": 85.4725, "grad_norm": 9.125, "learning_rate": 1.0866288048553696e-06, "loss": 1.4299, "step": 34189 }, { "epoch": 85.475, "grad_norm": 9.9375, "learning_rate": 1.0862617688616916e-06, "loss": 1.3289, "step": 34190 }, { "epoch": 85.4775, "grad_norm": 6.53125, "learning_rate": 1.0858947913061225e-06, "loss": 1.4112, "step": 34191 }, { "epoch": 85.48, "grad_norm": 8.25, "learning_rate": 1.08552787219107e-06, "loss": 1.3246, "step": 34192 }, { "epoch": 85.4825, "grad_norm": 8.5, "learning_rate": 1.0851610115189427e-06, "loss": 1.2466, "step": 34193 }, { "epoch": 85.485, "grad_norm": 4.875, "learning_rate": 1.0847942092921415e-06, "loss": 1.4003, "step": 34194 }, { "epoch": 85.4875, "grad_norm": 7.15625, "learning_rate": 1.0844274655130738e-06, "loss": 1.3444, "step": 34195 }, { "epoch": 85.49, "grad_norm": 10.375, "learning_rate": 1.084060780184143e-06, "loss": 1.468, "step": 34196 }, { "epoch": 85.4925, "grad_norm": 9.0625, "learning_rate": 1.0836941533077505e-06, "loss": 1.3491, "step": 34197 }, { "epoch": 85.495, "grad_norm": 6.71875, "learning_rate": 1.0833275848863035e-06, "loss": 1.2585, "step": 34198 }, { "epoch": 85.4975, "grad_norm": 7.0625, "learning_rate": 1.0829610749222053e-06, "loss": 1.2242, "step": 34199 }, { "epoch": 85.5, "grad_norm": 7.9375, "learning_rate": 1.0825946234178575e-06, "loss": 1.3705, "step": 34200 }, { "epoch": 85.5025, "grad_norm": 18.125, "learning_rate": 1.0822282303756616e-06, "loss": 1.2735, "step": 34201 }, { "epoch": 85.505, "grad_norm": 4.8125, "learning_rate": 1.0818618957980176e-06, "loss": 1.3504, "step": 34202 }, { "epoch": 85.5075, "grad_norm": 5.6875, "learning_rate": 1.0814956196873338e-06, "loss": 1.1184, "step": 34203 }, { "epoch": 85.51, "grad_norm": 7.75, "learning_rate": 1.0811294020460083e-06, "loss": 1.3881, "step": 34204 }, { "epoch": 85.5125, "grad_norm": 7.65625, "learning_rate": 1.0807632428764415e-06, "loss": 1.2483, "step": 34205 }, { "epoch": 85.515, "grad_norm": 7.78125, "learning_rate": 1.0803971421810311e-06, "loss": 1.4249, "step": 34206 }, { "epoch": 85.5175, "grad_norm": 7.4375, "learning_rate": 1.080031099962181e-06, "loss": 1.2602, "step": 34207 }, { "epoch": 85.52, "grad_norm": 12.0, "learning_rate": 1.0796651162222916e-06, "loss": 1.5374, "step": 34208 }, { "epoch": 85.5225, "grad_norm": 13.25, "learning_rate": 1.0792991909637607e-06, "loss": 1.184, "step": 34209 }, { "epoch": 85.525, "grad_norm": 10.4375, "learning_rate": 1.0789333241889855e-06, "loss": 1.3221, "step": 34210 }, { "epoch": 85.5275, "grad_norm": 6.90625, "learning_rate": 1.0785675159003684e-06, "loss": 1.24, "step": 34211 }, { "epoch": 85.53, "grad_norm": 8.1875, "learning_rate": 1.0782017661003041e-06, "loss": 1.2193, "step": 34212 }, { "epoch": 85.5325, "grad_norm": 9.375, "learning_rate": 1.0778360747911932e-06, "loss": 1.3383, "step": 34213 }, { "epoch": 85.535, "grad_norm": 10.6875, "learning_rate": 1.07747044197543e-06, "loss": 1.2176, "step": 34214 }, { "epoch": 85.5375, "grad_norm": 5.875, "learning_rate": 1.0771048676554163e-06, "loss": 1.3684, "step": 34215 }, { "epoch": 85.54, "grad_norm": 8.3125, "learning_rate": 1.0767393518335445e-06, "loss": 1.3285, "step": 34216 }, { "epoch": 85.5425, "grad_norm": 6.3125, "learning_rate": 1.0763738945122116e-06, "loss": 1.3836, "step": 34217 }, { "epoch": 85.545, "grad_norm": 7.125, "learning_rate": 1.0760084956938144e-06, "loss": 1.5137, "step": 34218 }, { "epoch": 85.5475, "grad_norm": 7.71875, "learning_rate": 1.07564315538075e-06, "loss": 1.3867, "step": 34219 }, { "epoch": 85.55, "grad_norm": 6.375, "learning_rate": 1.0752778735754121e-06, "loss": 1.1741, "step": 34220 }, { "epoch": 85.5525, "grad_norm": 8.5, "learning_rate": 1.0749126502801932e-06, "loss": 1.4417, "step": 34221 }, { "epoch": 85.555, "grad_norm": 9.25, "learning_rate": 1.0745474854974902e-06, "loss": 1.3473, "step": 34222 }, { "epoch": 85.5575, "grad_norm": 6.9375, "learning_rate": 1.074182379229698e-06, "loss": 1.287, "step": 34223 }, { "epoch": 85.56, "grad_norm": 7.4375, "learning_rate": 1.07381733147921e-06, "loss": 1.3561, "step": 34224 }, { "epoch": 85.5625, "grad_norm": 9.1875, "learning_rate": 1.0734523422484156e-06, "loss": 1.3465, "step": 34225 }, { "epoch": 85.565, "grad_norm": 7.6875, "learning_rate": 1.0730874115397128e-06, "loss": 1.2538, "step": 34226 }, { "epoch": 85.5675, "grad_norm": 8.3125, "learning_rate": 1.0727225393554885e-06, "loss": 1.2612, "step": 34227 }, { "epoch": 85.57, "grad_norm": 9.25, "learning_rate": 1.072357725698141e-06, "loss": 1.3495, "step": 34228 }, { "epoch": 85.5725, "grad_norm": 6.53125, "learning_rate": 1.0719929705700572e-06, "loss": 1.4736, "step": 34229 }, { "epoch": 85.575, "grad_norm": 5.96875, "learning_rate": 1.0716282739736318e-06, "loss": 1.2984, "step": 34230 }, { "epoch": 85.5775, "grad_norm": 7.4375, "learning_rate": 1.071263635911255e-06, "loss": 1.1968, "step": 34231 }, { "epoch": 85.58, "grad_norm": 5.28125, "learning_rate": 1.0708990563853127e-06, "loss": 1.2252, "step": 34232 }, { "epoch": 85.5825, "grad_norm": 7.4375, "learning_rate": 1.0705345353982e-06, "loss": 1.3115, "step": 34233 }, { "epoch": 85.585, "grad_norm": 7.5625, "learning_rate": 1.070170072952308e-06, "loss": 1.2961, "step": 34234 }, { "epoch": 85.5875, "grad_norm": 7.375, "learning_rate": 1.0698056690500224e-06, "loss": 1.1787, "step": 34235 }, { "epoch": 85.59, "grad_norm": 7.0, "learning_rate": 1.0694413236937329e-06, "loss": 1.1753, "step": 34236 }, { "epoch": 85.5925, "grad_norm": 6.8125, "learning_rate": 1.0690770368858273e-06, "loss": 1.249, "step": 34237 }, { "epoch": 85.595, "grad_norm": 5.375, "learning_rate": 1.068712808628698e-06, "loss": 1.3875, "step": 34238 }, { "epoch": 85.5975, "grad_norm": 7.03125, "learning_rate": 1.06834863892473e-06, "loss": 1.297, "step": 34239 }, { "epoch": 85.6, "grad_norm": 5.125, "learning_rate": 1.067984527776309e-06, "loss": 1.2137, "step": 34240 }, { "epoch": 85.6025, "grad_norm": 8.9375, "learning_rate": 1.0676204751858265e-06, "loss": 1.2374, "step": 34241 }, { "epoch": 85.605, "grad_norm": 6.21875, "learning_rate": 1.067256481155664e-06, "loss": 1.4455, "step": 34242 }, { "epoch": 85.6075, "grad_norm": 7.96875, "learning_rate": 1.0668925456882129e-06, "loss": 1.3291, "step": 34243 }, { "epoch": 85.61, "grad_norm": 8.125, "learning_rate": 1.0665286687858544e-06, "loss": 1.3455, "step": 34244 }, { "epoch": 85.6125, "grad_norm": 5.5625, "learning_rate": 1.0661648504509782e-06, "loss": 1.281, "step": 34245 }, { "epoch": 85.615, "grad_norm": 6.75, "learning_rate": 1.0658010906859683e-06, "loss": 1.4448, "step": 34246 }, { "epoch": 85.6175, "grad_norm": 6.9375, "learning_rate": 1.065437389493207e-06, "loss": 1.2584, "step": 34247 }, { "epoch": 85.62, "grad_norm": 9.875, "learning_rate": 1.0650737468750806e-06, "loss": 1.4159, "step": 34248 }, { "epoch": 85.6225, "grad_norm": 6.125, "learning_rate": 1.064710162833975e-06, "loss": 1.2425, "step": 34249 }, { "epoch": 85.625, "grad_norm": 6.9375, "learning_rate": 1.064346637372271e-06, "loss": 1.2509, "step": 34250 }, { "epoch": 85.6275, "grad_norm": 12.0, "learning_rate": 1.063983170492352e-06, "loss": 1.377, "step": 34251 }, { "epoch": 85.63, "grad_norm": 7.53125, "learning_rate": 1.0636197621966016e-06, "loss": 1.36, "step": 34252 }, { "epoch": 85.6325, "grad_norm": 8.1875, "learning_rate": 1.0632564124874045e-06, "loss": 1.4606, "step": 34253 }, { "epoch": 85.635, "grad_norm": 6.375, "learning_rate": 1.062893121367141e-06, "loss": 1.4404, "step": 34254 }, { "epoch": 85.6375, "grad_norm": 7.4375, "learning_rate": 1.0625298888381897e-06, "loss": 1.3136, "step": 34255 }, { "epoch": 85.64, "grad_norm": 7.5, "learning_rate": 1.062166714902938e-06, "loss": 1.4212, "step": 34256 }, { "epoch": 85.6425, "grad_norm": 10.25, "learning_rate": 1.061803599563761e-06, "loss": 1.326, "step": 34257 }, { "epoch": 85.645, "grad_norm": 6.34375, "learning_rate": 1.0614405428230435e-06, "loss": 1.19, "step": 34258 }, { "epoch": 85.6475, "grad_norm": 11.375, "learning_rate": 1.0610775446831623e-06, "loss": 1.4475, "step": 34259 }, { "epoch": 85.65, "grad_norm": 6.75, "learning_rate": 1.0607146051465011e-06, "loss": 1.2014, "step": 34260 }, { "epoch": 85.6525, "grad_norm": 7.15625, "learning_rate": 1.0603517242154372e-06, "loss": 1.266, "step": 34261 }, { "epoch": 85.655, "grad_norm": 7.125, "learning_rate": 1.0599889018923482e-06, "loss": 1.4651, "step": 34262 }, { "epoch": 85.6575, "grad_norm": 11.25, "learning_rate": 1.0596261381796135e-06, "loss": 1.2177, "step": 34263 }, { "epoch": 85.66, "grad_norm": 5.28125, "learning_rate": 1.0592634330796137e-06, "loss": 1.2408, "step": 34264 }, { "epoch": 85.6625, "grad_norm": 6.65625, "learning_rate": 1.0589007865947254e-06, "loss": 1.2563, "step": 34265 }, { "epoch": 85.665, "grad_norm": 7.71875, "learning_rate": 1.0585381987273247e-06, "loss": 1.276, "step": 34266 }, { "epoch": 85.6675, "grad_norm": 5.5625, "learning_rate": 1.0581756694797885e-06, "loss": 1.4477, "step": 34267 }, { "epoch": 85.67, "grad_norm": 11.3125, "learning_rate": 1.057813198854497e-06, "loss": 1.1539, "step": 34268 }, { "epoch": 85.6725, "grad_norm": 5.9375, "learning_rate": 1.0574507868538253e-06, "loss": 1.247, "step": 34269 }, { "epoch": 85.675, "grad_norm": 7.25, "learning_rate": 1.0570884334801456e-06, "loss": 1.4248, "step": 34270 }, { "epoch": 85.6775, "grad_norm": 6.78125, "learning_rate": 1.0567261387358384e-06, "loss": 1.2529, "step": 34271 }, { "epoch": 85.68, "grad_norm": 6.375, "learning_rate": 1.0563639026232742e-06, "loss": 1.3261, "step": 34272 }, { "epoch": 85.6825, "grad_norm": 9.25, "learning_rate": 1.0560017251448328e-06, "loss": 1.4611, "step": 34273 }, { "epoch": 85.685, "grad_norm": 7.0, "learning_rate": 1.055639606302885e-06, "loss": 1.3144, "step": 34274 }, { "epoch": 85.6875, "grad_norm": 8.375, "learning_rate": 1.0552775460998067e-06, "loss": 1.2233, "step": 34275 }, { "epoch": 85.69, "grad_norm": 8.375, "learning_rate": 1.054915544537971e-06, "loss": 1.3564, "step": 34276 }, { "epoch": 85.6925, "grad_norm": 7.65625, "learning_rate": 1.05455360161975e-06, "loss": 1.3482, "step": 34277 }, { "epoch": 85.695, "grad_norm": 7.5, "learning_rate": 1.054191717347517e-06, "loss": 1.1905, "step": 34278 }, { "epoch": 85.6975, "grad_norm": 6.0, "learning_rate": 1.0538298917236466e-06, "loss": 1.3834, "step": 34279 }, { "epoch": 85.7, "grad_norm": 8.0, "learning_rate": 1.0534681247505107e-06, "loss": 1.2566, "step": 34280 }, { "epoch": 85.7025, "grad_norm": 7.3125, "learning_rate": 1.053106416430477e-06, "loss": 1.2564, "step": 34281 }, { "epoch": 85.705, "grad_norm": 6.625, "learning_rate": 1.0527447667659197e-06, "loss": 1.3717, "step": 34282 }, { "epoch": 85.7075, "grad_norm": 7.0, "learning_rate": 1.0523831757592106e-06, "loss": 1.4087, "step": 34283 }, { "epoch": 85.71, "grad_norm": 7.53125, "learning_rate": 1.0520216434127207e-06, "loss": 1.2934, "step": 34284 }, { "epoch": 85.7125, "grad_norm": 7.78125, "learning_rate": 1.0516601697288164e-06, "loss": 1.3257, "step": 34285 }, { "epoch": 85.715, "grad_norm": 7.125, "learning_rate": 1.0512987547098718e-06, "loss": 1.3154, "step": 34286 }, { "epoch": 85.7175, "grad_norm": 5.59375, "learning_rate": 1.0509373983582527e-06, "loss": 1.4912, "step": 34287 }, { "epoch": 85.72, "grad_norm": 7.25, "learning_rate": 1.0505761006763315e-06, "loss": 1.2653, "step": 34288 }, { "epoch": 85.7225, "grad_norm": 6.5625, "learning_rate": 1.0502148616664743e-06, "loss": 1.3642, "step": 34289 }, { "epoch": 85.725, "grad_norm": 5.875, "learning_rate": 1.049853681331051e-06, "loss": 1.3232, "step": 34290 }, { "epoch": 85.7275, "grad_norm": 7.09375, "learning_rate": 1.0494925596724293e-06, "loss": 1.2967, "step": 34291 }, { "epoch": 85.73, "grad_norm": 6.9375, "learning_rate": 1.0491314966929743e-06, "loss": 1.3762, "step": 34292 }, { "epoch": 85.7325, "grad_norm": 8.0, "learning_rate": 1.0487704923950559e-06, "loss": 1.3426, "step": 34293 }, { "epoch": 85.735, "grad_norm": 8.875, "learning_rate": 1.0484095467810407e-06, "loss": 1.4279, "step": 34294 }, { "epoch": 85.7375, "grad_norm": 6.375, "learning_rate": 1.0480486598532946e-06, "loss": 1.3039, "step": 34295 }, { "epoch": 85.74, "grad_norm": 9.375, "learning_rate": 1.0476878316141814e-06, "loss": 1.3848, "step": 34296 }, { "epoch": 85.7425, "grad_norm": 6.3125, "learning_rate": 1.047327062066068e-06, "loss": 1.4215, "step": 34297 }, { "epoch": 85.745, "grad_norm": 7.8125, "learning_rate": 1.0469663512113226e-06, "loss": 1.3071, "step": 34298 }, { "epoch": 85.7475, "grad_norm": 9.125, "learning_rate": 1.0466056990523078e-06, "loss": 1.3411, "step": 34299 }, { "epoch": 85.75, "grad_norm": 6.25, "learning_rate": 1.0462451055913847e-06, "loss": 1.205, "step": 34300 }, { "epoch": 85.7525, "grad_norm": 7.5, "learning_rate": 1.045884570830923e-06, "loss": 1.2161, "step": 34301 }, { "epoch": 85.755, "grad_norm": 7.40625, "learning_rate": 1.0455240947732826e-06, "loss": 1.259, "step": 34302 }, { "epoch": 85.7575, "grad_norm": 7.875, "learning_rate": 1.0451636774208284e-06, "loss": 1.2454, "step": 34303 }, { "epoch": 85.76, "grad_norm": 6.8125, "learning_rate": 1.044803318775922e-06, "loss": 1.1678, "step": 34304 }, { "epoch": 85.7625, "grad_norm": 6.78125, "learning_rate": 1.0444430188409283e-06, "loss": 1.3556, "step": 34305 }, { "epoch": 85.765, "grad_norm": 8.75, "learning_rate": 1.044082777618205e-06, "loss": 1.3319, "step": 34306 }, { "epoch": 85.7675, "grad_norm": 5.21875, "learning_rate": 1.0437225951101193e-06, "loss": 1.1902, "step": 34307 }, { "epoch": 85.77, "grad_norm": 8.25, "learning_rate": 1.0433624713190281e-06, "loss": 1.216, "step": 34308 }, { "epoch": 85.7725, "grad_norm": 7.4375, "learning_rate": 1.0430024062472953e-06, "loss": 1.2711, "step": 34309 }, { "epoch": 85.775, "grad_norm": 8.0625, "learning_rate": 1.0426423998972812e-06, "loss": 1.2769, "step": 34310 }, { "epoch": 85.7775, "grad_norm": 6.90625, "learning_rate": 1.0422824522713426e-06, "loss": 1.2475, "step": 34311 }, { "epoch": 85.78, "grad_norm": 5.15625, "learning_rate": 1.041922563371842e-06, "loss": 1.3991, "step": 34312 }, { "epoch": 85.7825, "grad_norm": 9.0625, "learning_rate": 1.0415627332011402e-06, "loss": 1.2938, "step": 34313 }, { "epoch": 85.785, "grad_norm": 8.5625, "learning_rate": 1.0412029617615948e-06, "loss": 1.3784, "step": 34314 }, { "epoch": 85.7875, "grad_norm": 10.75, "learning_rate": 1.040843249055562e-06, "loss": 1.2975, "step": 34315 }, { "epoch": 85.79, "grad_norm": 14.8125, "learning_rate": 1.040483595085403e-06, "loss": 1.2194, "step": 34316 }, { "epoch": 85.7925, "grad_norm": 7.5, "learning_rate": 1.0401239998534762e-06, "loss": 1.4307, "step": 34317 }, { "epoch": 85.795, "grad_norm": 9.6875, "learning_rate": 1.0397644633621385e-06, "loss": 1.6077, "step": 34318 }, { "epoch": 85.7975, "grad_norm": 8.875, "learning_rate": 1.0394049856137445e-06, "loss": 1.3278, "step": 34319 }, { "epoch": 85.8, "grad_norm": 6.09375, "learning_rate": 1.0390455666106547e-06, "loss": 1.2463, "step": 34320 }, { "epoch": 85.8025, "grad_norm": 6.25, "learning_rate": 1.0386862063552217e-06, "loss": 1.4885, "step": 34321 }, { "epoch": 85.805, "grad_norm": 6.46875, "learning_rate": 1.0383269048498045e-06, "loss": 1.2582, "step": 34322 }, { "epoch": 85.8075, "grad_norm": 7.375, "learning_rate": 1.037967662096757e-06, "loss": 1.3534, "step": 34323 }, { "epoch": 85.81, "grad_norm": 8.0, "learning_rate": 1.0376084780984353e-06, "loss": 1.2638, "step": 34324 }, { "epoch": 85.8125, "grad_norm": 7.5, "learning_rate": 1.0372493528571947e-06, "loss": 1.2875, "step": 34325 }, { "epoch": 85.815, "grad_norm": 5.21875, "learning_rate": 1.036890286375386e-06, "loss": 1.3233, "step": 34326 }, { "epoch": 85.8175, "grad_norm": 6.1875, "learning_rate": 1.036531278655367e-06, "loss": 1.3472, "step": 34327 }, { "epoch": 85.82, "grad_norm": 5.75, "learning_rate": 1.0361723296994908e-06, "loss": 1.3448, "step": 34328 }, { "epoch": 85.8225, "grad_norm": 5.6875, "learning_rate": 1.0358134395101117e-06, "loss": 1.3041, "step": 34329 }, { "epoch": 85.825, "grad_norm": 8.125, "learning_rate": 1.03545460808958e-06, "loss": 1.3718, "step": 34330 }, { "epoch": 85.8275, "grad_norm": 6.75, "learning_rate": 1.0350958354402452e-06, "loss": 1.2904, "step": 34331 }, { "epoch": 85.83, "grad_norm": 8.1875, "learning_rate": 1.0347371215644675e-06, "loss": 1.2798, "step": 34332 }, { "epoch": 85.8325, "grad_norm": 7.375, "learning_rate": 1.034378466464594e-06, "loss": 1.3596, "step": 34333 }, { "epoch": 85.835, "grad_norm": 7.5, "learning_rate": 1.034019870142975e-06, "loss": 1.2638, "step": 34334 }, { "epoch": 85.8375, "grad_norm": 8.625, "learning_rate": 1.033661332601965e-06, "loss": 1.1949, "step": 34335 }, { "epoch": 85.84, "grad_norm": 7.75, "learning_rate": 1.0333028538439093e-06, "loss": 1.1409, "step": 34336 }, { "epoch": 85.8425, "grad_norm": 7.5, "learning_rate": 1.0329444338711647e-06, "loss": 1.3497, "step": 34337 }, { "epoch": 85.845, "grad_norm": 5.96875, "learning_rate": 1.0325860726860748e-06, "loss": 1.2401, "step": 34338 }, { "epoch": 85.8475, "grad_norm": 5.8125, "learning_rate": 1.0322277702909943e-06, "loss": 1.3562, "step": 34339 }, { "epoch": 85.85, "grad_norm": 10.375, "learning_rate": 1.0318695266882696e-06, "loss": 1.345, "step": 34340 }, { "epoch": 85.8525, "grad_norm": 8.125, "learning_rate": 1.031511341880247e-06, "loss": 1.2996, "step": 34341 }, { "epoch": 85.855, "grad_norm": 7.46875, "learning_rate": 1.0311532158692773e-06, "loss": 1.4861, "step": 34342 }, { "epoch": 85.8575, "grad_norm": 6.875, "learning_rate": 1.0307951486577106e-06, "loss": 1.4119, "step": 34343 }, { "epoch": 85.86, "grad_norm": 8.8125, "learning_rate": 1.0304371402478918e-06, "loss": 1.2593, "step": 34344 }, { "epoch": 85.8625, "grad_norm": 8.5625, "learning_rate": 1.030079190642168e-06, "loss": 1.4822, "step": 34345 }, { "epoch": 85.865, "grad_norm": 9.5, "learning_rate": 1.0297212998428818e-06, "loss": 1.3619, "step": 34346 }, { "epoch": 85.8675, "grad_norm": 6.3125, "learning_rate": 1.0293634678523879e-06, "loss": 1.2743, "step": 34347 }, { "epoch": 85.87, "grad_norm": 8.375, "learning_rate": 1.0290056946730288e-06, "loss": 1.1857, "step": 34348 }, { "epoch": 85.8725, "grad_norm": 6.4375, "learning_rate": 1.028647980307148e-06, "loss": 1.3279, "step": 34349 }, { "epoch": 85.875, "grad_norm": 10.1875, "learning_rate": 1.0282903247570908e-06, "loss": 1.2901, "step": 34350 }, { "epoch": 85.8775, "grad_norm": 6.09375, "learning_rate": 1.0279327280252037e-06, "loss": 1.2832, "step": 34351 }, { "epoch": 85.88, "grad_norm": 5.9375, "learning_rate": 1.027575190113832e-06, "loss": 1.3419, "step": 34352 }, { "epoch": 85.8825, "grad_norm": 11.3125, "learning_rate": 1.027217711025319e-06, "loss": 1.3652, "step": 34353 }, { "epoch": 85.885, "grad_norm": 5.25, "learning_rate": 1.026860290762005e-06, "loss": 1.361, "step": 34354 }, { "epoch": 85.8875, "grad_norm": 7.0, "learning_rate": 1.0265029293262375e-06, "loss": 1.3228, "step": 34355 }, { "epoch": 85.89, "grad_norm": 7.34375, "learning_rate": 1.0261456267203561e-06, "loss": 1.2788, "step": 34356 }, { "epoch": 85.8925, "grad_norm": 7.6875, "learning_rate": 1.0257883829467075e-06, "loss": 1.4532, "step": 34357 }, { "epoch": 85.895, "grad_norm": 10.875, "learning_rate": 1.0254311980076282e-06, "loss": 1.2486, "step": 34358 }, { "epoch": 85.8975, "grad_norm": 8.625, "learning_rate": 1.0250740719054653e-06, "loss": 1.346, "step": 34359 }, { "epoch": 85.9, "grad_norm": 17.75, "learning_rate": 1.024717004642557e-06, "loss": 1.4898, "step": 34360 }, { "epoch": 85.9025, "grad_norm": 9.1875, "learning_rate": 1.0243599962212414e-06, "loss": 1.3216, "step": 34361 }, { "epoch": 85.905, "grad_norm": 8.875, "learning_rate": 1.0240030466438667e-06, "loss": 1.2586, "step": 34362 }, { "epoch": 85.9075, "grad_norm": 8.25, "learning_rate": 1.0236461559127687e-06, "loss": 1.1933, "step": 34363 }, { "epoch": 85.91, "grad_norm": 8.0625, "learning_rate": 1.0232893240302865e-06, "loss": 1.274, "step": 34364 }, { "epoch": 85.9125, "grad_norm": 10.3125, "learning_rate": 1.0229325509987587e-06, "loss": 1.2446, "step": 34365 }, { "epoch": 85.915, "grad_norm": 6.84375, "learning_rate": 1.0225758368205263e-06, "loss": 1.2493, "step": 34366 }, { "epoch": 85.9175, "grad_norm": 6.96875, "learning_rate": 1.0222191814979287e-06, "loss": 1.173, "step": 34367 }, { "epoch": 85.92, "grad_norm": 9.1875, "learning_rate": 1.021862585033304e-06, "loss": 1.3896, "step": 34368 }, { "epoch": 85.9225, "grad_norm": 9.8125, "learning_rate": 1.021506047428986e-06, "loss": 1.4811, "step": 34369 }, { "epoch": 85.925, "grad_norm": 8.5, "learning_rate": 1.0211495686873173e-06, "loss": 1.3213, "step": 34370 }, { "epoch": 85.9275, "grad_norm": 14.0, "learning_rate": 1.0207931488106304e-06, "loss": 1.3329, "step": 34371 }, { "epoch": 85.93, "grad_norm": 4.78125, "learning_rate": 1.0204367878012666e-06, "loss": 1.2712, "step": 34372 }, { "epoch": 85.9325, "grad_norm": 6.8125, "learning_rate": 1.0200804856615588e-06, "loss": 1.2784, "step": 34373 }, { "epoch": 85.935, "grad_norm": 6.5, "learning_rate": 1.0197242423938447e-06, "loss": 1.4523, "step": 34374 }, { "epoch": 85.9375, "grad_norm": 3.875, "learning_rate": 1.0193680580004594e-06, "loss": 1.1622, "step": 34375 }, { "epoch": 85.94, "grad_norm": 8.5625, "learning_rate": 1.0190119324837356e-06, "loss": 1.3051, "step": 34376 }, { "epoch": 85.9425, "grad_norm": 7.71875, "learning_rate": 1.018655865846011e-06, "loss": 1.3777, "step": 34377 }, { "epoch": 85.945, "grad_norm": 7.09375, "learning_rate": 1.0182998580896197e-06, "loss": 1.4157, "step": 34378 }, { "epoch": 85.9475, "grad_norm": 10.9375, "learning_rate": 1.0179439092168963e-06, "loss": 1.3284, "step": 34379 }, { "epoch": 85.95, "grad_norm": 9.1875, "learning_rate": 1.0175880192301713e-06, "loss": 1.372, "step": 34380 }, { "epoch": 85.9525, "grad_norm": 7.125, "learning_rate": 1.0172321881317793e-06, "loss": 1.3584, "step": 34381 }, { "epoch": 85.955, "grad_norm": 7.90625, "learning_rate": 1.0168764159240552e-06, "loss": 1.4171, "step": 34382 }, { "epoch": 85.9575, "grad_norm": 8.0625, "learning_rate": 1.0165207026093304e-06, "loss": 1.3059, "step": 34383 }, { "epoch": 85.96, "grad_norm": 6.5625, "learning_rate": 1.0161650481899344e-06, "loss": 1.3083, "step": 34384 }, { "epoch": 85.9625, "grad_norm": 12.5625, "learning_rate": 1.0158094526682017e-06, "loss": 1.2514, "step": 34385 }, { "epoch": 85.965, "grad_norm": 7.4375, "learning_rate": 1.0154539160464616e-06, "loss": 1.256, "step": 34386 }, { "epoch": 85.9675, "grad_norm": 6.375, "learning_rate": 1.015098438327049e-06, "loss": 1.1162, "step": 34387 }, { "epoch": 85.97, "grad_norm": 5.78125, "learning_rate": 1.0147430195122888e-06, "loss": 1.2921, "step": 34388 }, { "epoch": 85.9725, "grad_norm": 8.375, "learning_rate": 1.0143876596045156e-06, "loss": 1.2691, "step": 34389 }, { "epoch": 85.975, "grad_norm": 7.59375, "learning_rate": 1.0140323586060563e-06, "loss": 1.3799, "step": 34390 }, { "epoch": 85.9775, "grad_norm": 7.1875, "learning_rate": 1.0136771165192404e-06, "loss": 1.1085, "step": 34391 }, { "epoch": 85.98, "grad_norm": 8.5625, "learning_rate": 1.0133219333463983e-06, "loss": 1.58, "step": 34392 }, { "epoch": 85.9825, "grad_norm": 7.0625, "learning_rate": 1.0129668090898593e-06, "loss": 1.3748, "step": 34393 }, { "epoch": 85.985, "grad_norm": 5.5625, "learning_rate": 1.01261174375195e-06, "loss": 1.4117, "step": 34394 }, { "epoch": 85.9875, "grad_norm": 10.375, "learning_rate": 1.0122567373349968e-06, "loss": 1.3215, "step": 34395 }, { "epoch": 85.99, "grad_norm": 11.5, "learning_rate": 1.0119017898413286e-06, "loss": 1.3581, "step": 34396 }, { "epoch": 85.9925, "grad_norm": 9.0625, "learning_rate": 1.0115469012732748e-06, "loss": 1.2863, "step": 34397 }, { "epoch": 85.995, "grad_norm": 6.15625, "learning_rate": 1.0111920716331591e-06, "loss": 1.2902, "step": 34398 }, { "epoch": 85.9975, "grad_norm": 8.75, "learning_rate": 1.0108373009233075e-06, "loss": 1.1993, "step": 34399 }, { "epoch": 86.0, "grad_norm": 7.8125, "learning_rate": 1.010482589146048e-06, "loss": 1.4433, "step": 34400 }, { "epoch": 86.0025, "grad_norm": 10.625, "learning_rate": 1.010127936303703e-06, "loss": 1.3237, "step": 34401 }, { "epoch": 86.005, "grad_norm": 5.21875, "learning_rate": 1.0097733423986022e-06, "loss": 1.3852, "step": 34402 }, { "epoch": 86.0075, "grad_norm": 7.6875, "learning_rate": 1.0094188074330646e-06, "loss": 1.5583, "step": 34403 }, { "epoch": 86.01, "grad_norm": 7.3125, "learning_rate": 1.0090643314094207e-06, "loss": 1.3556, "step": 34404 }, { "epoch": 86.0125, "grad_norm": 6.5, "learning_rate": 1.0087099143299906e-06, "loss": 1.3176, "step": 34405 }, { "epoch": 86.015, "grad_norm": 7.0, "learning_rate": 1.008355556197096e-06, "loss": 1.3142, "step": 34406 }, { "epoch": 86.0175, "grad_norm": 6.375, "learning_rate": 1.0080012570130627e-06, "loss": 1.3572, "step": 34407 }, { "epoch": 86.02, "grad_norm": 6.28125, "learning_rate": 1.0076470167802156e-06, "loss": 1.1898, "step": 34408 }, { "epoch": 86.0225, "grad_norm": 6.5625, "learning_rate": 1.007292835500875e-06, "loss": 1.243, "step": 34409 }, { "epoch": 86.025, "grad_norm": 10.3125, "learning_rate": 1.006938713177361e-06, "loss": 1.3746, "step": 34410 }, { "epoch": 86.0275, "grad_norm": 6.96875, "learning_rate": 1.0065846498119958e-06, "loss": 1.2678, "step": 34411 }, { "epoch": 86.03, "grad_norm": 5.78125, "learning_rate": 1.0062306454071047e-06, "loss": 1.4207, "step": 34412 }, { "epoch": 86.0325, "grad_norm": 7.0625, "learning_rate": 1.005876699965006e-06, "loss": 1.3948, "step": 34413 }, { "epoch": 86.035, "grad_norm": 6.03125, "learning_rate": 1.0055228134880168e-06, "loss": 1.3555, "step": 34414 }, { "epoch": 86.0375, "grad_norm": 8.5625, "learning_rate": 1.005168985978462e-06, "loss": 1.4147, "step": 34415 }, { "epoch": 86.04, "grad_norm": 6.9375, "learning_rate": 1.0048152174386584e-06, "loss": 1.2157, "step": 34416 }, { "epoch": 86.0425, "grad_norm": 7.96875, "learning_rate": 1.0044615078709274e-06, "loss": 1.3181, "step": 34417 }, { "epoch": 86.045, "grad_norm": 7.0625, "learning_rate": 1.0041078572775842e-06, "loss": 1.3211, "step": 34418 }, { "epoch": 86.0475, "grad_norm": 11.125, "learning_rate": 1.0037542656609534e-06, "loss": 1.185, "step": 34419 }, { "epoch": 86.05, "grad_norm": 12.8125, "learning_rate": 1.0034007330233487e-06, "loss": 1.2709, "step": 34420 }, { "epoch": 86.0525, "grad_norm": 5.9375, "learning_rate": 1.0030472593670871e-06, "loss": 1.2604, "step": 34421 }, { "epoch": 86.055, "grad_norm": 8.125, "learning_rate": 1.0026938446944868e-06, "loss": 1.3233, "step": 34422 }, { "epoch": 86.0575, "grad_norm": 8.75, "learning_rate": 1.002340489007868e-06, "loss": 1.4691, "step": 34423 }, { "epoch": 86.06, "grad_norm": 11.1875, "learning_rate": 1.0019871923095459e-06, "loss": 1.2824, "step": 34424 }, { "epoch": 86.0625, "grad_norm": 6.78125, "learning_rate": 1.0016339546018328e-06, "loss": 1.334, "step": 34425 }, { "epoch": 86.065, "grad_norm": 7.71875, "learning_rate": 1.0012807758870469e-06, "loss": 1.309, "step": 34426 }, { "epoch": 86.0675, "grad_norm": 8.1875, "learning_rate": 1.0009276561675075e-06, "loss": 1.3372, "step": 34427 }, { "epoch": 86.07, "grad_norm": 7.5, "learning_rate": 1.0005745954455248e-06, "loss": 1.3851, "step": 34428 }, { "epoch": 86.0725, "grad_norm": 7.0, "learning_rate": 1.0002215937234138e-06, "loss": 1.2743, "step": 34429 }, { "epoch": 86.075, "grad_norm": 8.5625, "learning_rate": 9.998686510034915e-07, "loss": 1.2099, "step": 34430 }, { "epoch": 86.0775, "grad_norm": 11.1875, "learning_rate": 9.995157672880683e-07, "loss": 1.4012, "step": 34431 }, { "epoch": 86.08, "grad_norm": 7.40625, "learning_rate": 9.991629425794624e-07, "loss": 1.1843, "step": 34432 }, { "epoch": 86.0825, "grad_norm": 8.125, "learning_rate": 9.988101768799808e-07, "loss": 1.4081, "step": 34433 }, { "epoch": 86.085, "grad_norm": 4.875, "learning_rate": 9.984574701919425e-07, "loss": 1.2007, "step": 34434 }, { "epoch": 86.0875, "grad_norm": 8.375, "learning_rate": 9.981048225176559e-07, "loss": 1.6075, "step": 34435 }, { "epoch": 86.09, "grad_norm": 7.59375, "learning_rate": 9.977522338594325e-07, "loss": 1.3417, "step": 34436 }, { "epoch": 86.0925, "grad_norm": 8.4375, "learning_rate": 9.973997042195848e-07, "loss": 1.1851, "step": 34437 }, { "epoch": 86.095, "grad_norm": 8.3125, "learning_rate": 9.970472336004266e-07, "loss": 1.3741, "step": 34438 }, { "epoch": 86.0975, "grad_norm": 13.8125, "learning_rate": 9.96694822004266e-07, "loss": 1.3158, "step": 34439 }, { "epoch": 86.1, "grad_norm": 6.34375, "learning_rate": 9.963424694334122e-07, "loss": 1.3519, "step": 34440 }, { "epoch": 86.1025, "grad_norm": 17.5, "learning_rate": 9.95990175890177e-07, "loss": 1.4277, "step": 34441 }, { "epoch": 86.105, "grad_norm": 7.75, "learning_rate": 9.956379413768714e-07, "loss": 1.3103, "step": 34442 }, { "epoch": 86.1075, "grad_norm": 6.59375, "learning_rate": 9.952857658958038e-07, "loss": 1.4461, "step": 34443 }, { "epoch": 86.11, "grad_norm": 6.1875, "learning_rate": 9.9493364944928e-07, "loss": 1.33, "step": 34444 }, { "epoch": 86.1125, "grad_norm": 5.8125, "learning_rate": 9.945815920396129e-07, "loss": 1.25, "step": 34445 }, { "epoch": 86.115, "grad_norm": 7.0, "learning_rate": 9.94229593669106e-07, "loss": 1.342, "step": 34446 }, { "epoch": 86.1175, "grad_norm": 7.65625, "learning_rate": 9.938776543400718e-07, "loss": 1.0999, "step": 34447 }, { "epoch": 86.12, "grad_norm": 10.5625, "learning_rate": 9.935257740548143e-07, "loss": 1.3379, "step": 34448 }, { "epoch": 86.1225, "grad_norm": 4.625, "learning_rate": 9.931739528156426e-07, "loss": 1.2258, "step": 34449 }, { "epoch": 86.125, "grad_norm": 8.5625, "learning_rate": 9.928221906248614e-07, "loss": 1.2727, "step": 34450 }, { "epoch": 86.1275, "grad_norm": 5.0625, "learning_rate": 9.924704874847769e-07, "loss": 1.2381, "step": 34451 }, { "epoch": 86.13, "grad_norm": 7.3125, "learning_rate": 9.921188433976947e-07, "loss": 1.4867, "step": 34452 }, { "epoch": 86.1325, "grad_norm": 8.3125, "learning_rate": 9.91767258365922e-07, "loss": 1.3699, "step": 34453 }, { "epoch": 86.135, "grad_norm": 8.3125, "learning_rate": 9.91415732391764e-07, "loss": 1.4013, "step": 34454 }, { "epoch": 86.1375, "grad_norm": 9.1875, "learning_rate": 9.910642654775216e-07, "loss": 1.3021, "step": 34455 }, { "epoch": 86.14, "grad_norm": 7.65625, "learning_rate": 9.90712857625501e-07, "loss": 1.334, "step": 34456 }, { "epoch": 86.1425, "grad_norm": 8.1875, "learning_rate": 9.90361508838008e-07, "loss": 1.3598, "step": 34457 }, { "epoch": 86.145, "grad_norm": 7.0, "learning_rate": 9.900102191173456e-07, "loss": 1.4692, "step": 34458 }, { "epoch": 86.1475, "grad_norm": 13.8125, "learning_rate": 9.896589884658126e-07, "loss": 1.2604, "step": 34459 }, { "epoch": 86.15, "grad_norm": 4.25, "learning_rate": 9.893078168857173e-07, "loss": 1.2274, "step": 34460 }, { "epoch": 86.1525, "grad_norm": 7.59375, "learning_rate": 9.889567043793569e-07, "loss": 1.3085, "step": 34461 }, { "epoch": 86.155, "grad_norm": 6.375, "learning_rate": 9.886056509490383e-07, "loss": 1.2129, "step": 34462 }, { "epoch": 86.1575, "grad_norm": 5.8125, "learning_rate": 9.882546565970574e-07, "loss": 1.3529, "step": 34463 }, { "epoch": 86.16, "grad_norm": 8.5625, "learning_rate": 9.879037213257214e-07, "loss": 1.5072, "step": 34464 }, { "epoch": 86.1625, "grad_norm": 6.1875, "learning_rate": 9.875528451373274e-07, "loss": 1.2118, "step": 34465 }, { "epoch": 86.165, "grad_norm": 8.9375, "learning_rate": 9.872020280341744e-07, "loss": 1.1977, "step": 34466 }, { "epoch": 86.1675, "grad_norm": 7.34375, "learning_rate": 9.868512700185651e-07, "loss": 1.1957, "step": 34467 }, { "epoch": 86.17, "grad_norm": 6.09375, "learning_rate": 9.865005710927989e-07, "loss": 1.1884, "step": 34468 }, { "epoch": 86.1725, "grad_norm": 6.78125, "learning_rate": 9.861499312591749e-07, "loss": 1.2592, "step": 34469 }, { "epoch": 86.175, "grad_norm": 8.25, "learning_rate": 9.857993505199913e-07, "loss": 1.3694, "step": 34470 }, { "epoch": 86.1775, "grad_norm": 6.1875, "learning_rate": 9.85448828877542e-07, "loss": 1.2424, "step": 34471 }, { "epoch": 86.18, "grad_norm": 6.9375, "learning_rate": 9.85098366334134e-07, "loss": 1.3464, "step": 34472 }, { "epoch": 86.1825, "grad_norm": 8.625, "learning_rate": 9.847479628920597e-07, "loss": 1.3605, "step": 34473 }, { "epoch": 86.185, "grad_norm": 5.59375, "learning_rate": 9.843976185536153e-07, "loss": 1.3412, "step": 34474 }, { "epoch": 86.1875, "grad_norm": 5.75, "learning_rate": 9.84047333321102e-07, "loss": 1.2737, "step": 34475 }, { "epoch": 86.19, "grad_norm": 8.5, "learning_rate": 9.836971071968104e-07, "loss": 1.1899, "step": 34476 }, { "epoch": 86.1925, "grad_norm": 11.1875, "learning_rate": 9.833469401830432e-07, "loss": 1.46, "step": 34477 }, { "epoch": 86.195, "grad_norm": 6.8125, "learning_rate": 9.829968322820894e-07, "loss": 1.2588, "step": 34478 }, { "epoch": 86.1975, "grad_norm": 9.1875, "learning_rate": 9.826467834962494e-07, "loss": 1.3189, "step": 34479 }, { "epoch": 86.2, "grad_norm": 7.09375, "learning_rate": 9.822967938278172e-07, "loss": 1.334, "step": 34480 }, { "epoch": 86.2025, "grad_norm": 8.375, "learning_rate": 9.819468632790842e-07, "loss": 1.353, "step": 34481 }, { "epoch": 86.205, "grad_norm": 8.0625, "learning_rate": 9.81596991852346e-07, "loss": 1.4959, "step": 34482 }, { "epoch": 86.2075, "grad_norm": 8.0625, "learning_rate": 9.81247179549899e-07, "loss": 1.2276, "step": 34483 }, { "epoch": 86.21, "grad_norm": 7.90625, "learning_rate": 9.808974263740345e-07, "loss": 1.4489, "step": 34484 }, { "epoch": 86.2125, "grad_norm": 7.0, "learning_rate": 9.805477323270463e-07, "loss": 1.2612, "step": 34485 }, { "epoch": 86.215, "grad_norm": 7.0, "learning_rate": 9.801980974112224e-07, "loss": 1.2438, "step": 34486 }, { "epoch": 86.2175, "grad_norm": 8.6875, "learning_rate": 9.798485216288623e-07, "loss": 1.163, "step": 34487 }, { "epoch": 86.22, "grad_norm": 10.5625, "learning_rate": 9.79499004982254e-07, "loss": 1.393, "step": 34488 }, { "epoch": 86.2225, "grad_norm": 5.59375, "learning_rate": 9.7914954747369e-07, "loss": 1.3295, "step": 34489 }, { "epoch": 86.225, "grad_norm": 6.1875, "learning_rate": 9.788001491054578e-07, "loss": 1.2106, "step": 34490 }, { "epoch": 86.2275, "grad_norm": 6.71875, "learning_rate": 9.784508098798517e-07, "loss": 1.3827, "step": 34491 }, { "epoch": 86.23, "grad_norm": 8.5, "learning_rate": 9.781015297991626e-07, "loss": 1.3864, "step": 34492 }, { "epoch": 86.2325, "grad_norm": 7.5, "learning_rate": 9.777523088656782e-07, "loss": 1.34, "step": 34493 }, { "epoch": 86.235, "grad_norm": 7.28125, "learning_rate": 9.774031470816868e-07, "loss": 1.1909, "step": 34494 }, { "epoch": 86.2375, "grad_norm": 10.1875, "learning_rate": 9.770540444494826e-07, "loss": 1.2019, "step": 34495 }, { "epoch": 86.24, "grad_norm": 5.40625, "learning_rate": 9.767050009713476e-07, "loss": 1.4095, "step": 34496 }, { "epoch": 86.2425, "grad_norm": 8.3125, "learning_rate": 9.763560166495766e-07, "loss": 1.3683, "step": 34497 }, { "epoch": 86.245, "grad_norm": 7.71875, "learning_rate": 9.760070914864517e-07, "loss": 1.1649, "step": 34498 }, { "epoch": 86.2475, "grad_norm": 7.90625, "learning_rate": 9.75658225484265e-07, "loss": 1.3417, "step": 34499 }, { "epoch": 86.25, "grad_norm": 8.0625, "learning_rate": 9.753094186453028e-07, "loss": 1.2862, "step": 34500 }, { "epoch": 86.2525, "grad_norm": 5.90625, "learning_rate": 9.749606709718463e-07, "loss": 1.3745, "step": 34501 }, { "epoch": 86.255, "grad_norm": 8.9375, "learning_rate": 9.746119824661904e-07, "loss": 1.195, "step": 34502 }, { "epoch": 86.2575, "grad_norm": 9.875, "learning_rate": 9.742633531306168e-07, "loss": 1.3592, "step": 34503 }, { "epoch": 86.26, "grad_norm": 6.875, "learning_rate": 9.739147829674124e-07, "loss": 1.3589, "step": 34504 }, { "epoch": 86.2625, "grad_norm": 6.59375, "learning_rate": 9.735662719788586e-07, "loss": 1.2231, "step": 34505 }, { "epoch": 86.265, "grad_norm": 12.8125, "learning_rate": 9.732178201672438e-07, "loss": 1.5373, "step": 34506 }, { "epoch": 86.2675, "grad_norm": 8.4375, "learning_rate": 9.728694275348539e-07, "loss": 1.4278, "step": 34507 }, { "epoch": 86.27, "grad_norm": 8.3125, "learning_rate": 9.725210940839703e-07, "loss": 1.2537, "step": 34508 }, { "epoch": 86.2725, "grad_norm": 9.1875, "learning_rate": 9.721728198168745e-07, "loss": 1.2908, "step": 34509 }, { "epoch": 86.275, "grad_norm": 5.5, "learning_rate": 9.71824604735856e-07, "loss": 1.3072, "step": 34510 }, { "epoch": 86.2775, "grad_norm": 6.90625, "learning_rate": 9.714764488431904e-07, "loss": 1.3398, "step": 34511 }, { "epoch": 86.28, "grad_norm": 9.25, "learning_rate": 9.711283521411674e-07, "loss": 1.4106, "step": 34512 }, { "epoch": 86.2825, "grad_norm": 6.5, "learning_rate": 9.707803146320626e-07, "loss": 1.3787, "step": 34513 }, { "epoch": 86.285, "grad_norm": 8.125, "learning_rate": 9.704323363181634e-07, "loss": 1.3005, "step": 34514 }, { "epoch": 86.2875, "grad_norm": 10.1875, "learning_rate": 9.700844172017475e-07, "loss": 1.3599, "step": 34515 }, { "epoch": 86.29, "grad_norm": 7.03125, "learning_rate": 9.697365572850958e-07, "loss": 1.3356, "step": 34516 }, { "epoch": 86.2925, "grad_norm": 7.5625, "learning_rate": 9.693887565704897e-07, "loss": 1.3071, "step": 34517 }, { "epoch": 86.295, "grad_norm": 5.84375, "learning_rate": 9.690410150602115e-07, "loss": 1.2966, "step": 34518 }, { "epoch": 86.2975, "grad_norm": 6.03125, "learning_rate": 9.686933327565384e-07, "loss": 1.2316, "step": 34519 }, { "epoch": 86.3, "grad_norm": 6.25, "learning_rate": 9.683457096617487e-07, "loss": 1.3312, "step": 34520 }, { "epoch": 86.3025, "grad_norm": 8.8125, "learning_rate": 9.679981457781228e-07, "loss": 1.3263, "step": 34521 }, { "epoch": 86.305, "grad_norm": 8.875, "learning_rate": 9.676506411079423e-07, "loss": 1.3226, "step": 34522 }, { "epoch": 86.3075, "grad_norm": 6.1875, "learning_rate": 9.673031956534828e-07, "loss": 1.334, "step": 34523 }, { "epoch": 86.31, "grad_norm": 7.375, "learning_rate": 9.669558094170194e-07, "loss": 1.2323, "step": 34524 }, { "epoch": 86.3125, "grad_norm": 8.6875, "learning_rate": 9.666084824008349e-07, "loss": 1.2966, "step": 34525 }, { "epoch": 86.315, "grad_norm": 5.40625, "learning_rate": 9.662612146072014e-07, "loss": 1.1615, "step": 34526 }, { "epoch": 86.3175, "grad_norm": 6.25, "learning_rate": 9.659140060383997e-07, "loss": 1.1499, "step": 34527 }, { "epoch": 86.32, "grad_norm": 7.4375, "learning_rate": 9.655668566967026e-07, "loss": 1.4057, "step": 34528 }, { "epoch": 86.3225, "grad_norm": 8.0, "learning_rate": 9.652197665843898e-07, "loss": 1.3338, "step": 34529 }, { "epoch": 86.325, "grad_norm": 6.46875, "learning_rate": 9.648727357037336e-07, "loss": 1.5343, "step": 34530 }, { "epoch": 86.3275, "grad_norm": 4.78125, "learning_rate": 9.645257640570095e-07, "loss": 1.2308, "step": 34531 }, { "epoch": 86.33, "grad_norm": 6.28125, "learning_rate": 9.641788516464922e-07, "loss": 1.3721, "step": 34532 }, { "epoch": 86.3325, "grad_norm": 6.65625, "learning_rate": 9.638319984744581e-07, "loss": 1.3134, "step": 34533 }, { "epoch": 86.335, "grad_norm": 8.0625, "learning_rate": 9.63485204543181e-07, "loss": 1.2635, "step": 34534 }, { "epoch": 86.3375, "grad_norm": 6.90625, "learning_rate": 9.631384698549307e-07, "loss": 1.325, "step": 34535 }, { "epoch": 86.34, "grad_norm": 7.875, "learning_rate": 9.627917944119835e-07, "loss": 1.5346, "step": 34536 }, { "epoch": 86.3425, "grad_norm": 7.46875, "learning_rate": 9.624451782166133e-07, "loss": 1.1816, "step": 34537 }, { "epoch": 86.345, "grad_norm": 9.625, "learning_rate": 9.620986212710904e-07, "loss": 1.2563, "step": 34538 }, { "epoch": 86.3475, "grad_norm": 9.5625, "learning_rate": 9.617521235776861e-07, "loss": 1.5325, "step": 34539 }, { "epoch": 86.35, "grad_norm": 7.28125, "learning_rate": 9.614056851386743e-07, "loss": 1.2149, "step": 34540 }, { "epoch": 86.3525, "grad_norm": 11.125, "learning_rate": 9.610593059563234e-07, "loss": 1.3758, "step": 34541 }, { "epoch": 86.355, "grad_norm": 7.8125, "learning_rate": 9.60712986032909e-07, "loss": 1.3855, "step": 34542 }, { "epoch": 86.3575, "grad_norm": 9.9375, "learning_rate": 9.60366725370695e-07, "loss": 1.3411, "step": 34543 }, { "epoch": 86.36, "grad_norm": 4.28125, "learning_rate": 9.600205239719584e-07, "loss": 1.3281, "step": 34544 }, { "epoch": 86.3625, "grad_norm": 7.0625, "learning_rate": 9.59674381838962e-07, "loss": 1.3239, "step": 34545 }, { "epoch": 86.365, "grad_norm": 6.53125, "learning_rate": 9.593282989739816e-07, "loss": 1.3939, "step": 34546 }, { "epoch": 86.3675, "grad_norm": 8.875, "learning_rate": 9.58982275379281e-07, "loss": 1.3303, "step": 34547 }, { "epoch": 86.37, "grad_norm": 8.3125, "learning_rate": 9.58636311057133e-07, "loss": 1.2125, "step": 34548 }, { "epoch": 86.3725, "grad_norm": 8.5, "learning_rate": 9.58290406009803e-07, "loss": 1.3823, "step": 34549 }, { "epoch": 86.375, "grad_norm": 7.375, "learning_rate": 9.579445602395577e-07, "loss": 1.3957, "step": 34550 }, { "epoch": 86.3775, "grad_norm": 5.4375, "learning_rate": 9.575987737486647e-07, "loss": 1.3778, "step": 34551 }, { "epoch": 86.38, "grad_norm": 9.5, "learning_rate": 9.57253046539396e-07, "loss": 1.2801, "step": 34552 }, { "epoch": 86.3825, "grad_norm": 23.875, "learning_rate": 9.569073786140137e-07, "loss": 1.2356, "step": 34553 }, { "epoch": 86.385, "grad_norm": 8.4375, "learning_rate": 9.56561769974783e-07, "loss": 1.2536, "step": 34554 }, { "epoch": 86.3875, "grad_norm": 5.5, "learning_rate": 9.56216220623971e-07, "loss": 1.3706, "step": 34555 }, { "epoch": 86.39, "grad_norm": 9.625, "learning_rate": 9.558707305638458e-07, "loss": 1.366, "step": 34556 }, { "epoch": 86.3925, "grad_norm": 7.46875, "learning_rate": 9.555252997966702e-07, "loss": 1.2567, "step": 34557 }, { "epoch": 86.395, "grad_norm": 8.6875, "learning_rate": 9.551799283247065e-07, "loss": 1.4007, "step": 34558 }, { "epoch": 86.3975, "grad_norm": 8.625, "learning_rate": 9.548346161502242e-07, "loss": 1.4946, "step": 34559 }, { "epoch": 86.4, "grad_norm": 10.5625, "learning_rate": 9.544893632754816e-07, "loss": 1.2699, "step": 34560 }, { "epoch": 86.4025, "grad_norm": 5.4375, "learning_rate": 9.541441697027465e-07, "loss": 1.1869, "step": 34561 }, { "epoch": 86.405, "grad_norm": 21.5, "learning_rate": 9.537990354342785e-07, "loss": 1.3398, "step": 34562 }, { "epoch": 86.4075, "grad_norm": 9.75, "learning_rate": 9.534539604723436e-07, "loss": 1.2724, "step": 34563 }, { "epoch": 86.41, "grad_norm": 6.90625, "learning_rate": 9.531089448192022e-07, "loss": 1.6121, "step": 34564 }, { "epoch": 86.4125, "grad_norm": 7.09375, "learning_rate": 9.527639884771156e-07, "loss": 1.2347, "step": 34565 }, { "epoch": 86.415, "grad_norm": 12.125, "learning_rate": 9.524190914483455e-07, "loss": 1.3483, "step": 34566 }, { "epoch": 86.4175, "grad_norm": 12.3125, "learning_rate": 9.520742537351568e-07, "loss": 1.2963, "step": 34567 }, { "epoch": 86.42, "grad_norm": 7.34375, "learning_rate": 9.517294753398066e-07, "loss": 1.2422, "step": 34568 }, { "epoch": 86.4225, "grad_norm": 7.84375, "learning_rate": 9.513847562645528e-07, "loss": 1.357, "step": 34569 }, { "epoch": 86.425, "grad_norm": 6.34375, "learning_rate": 9.510400965116595e-07, "loss": 1.4099, "step": 34570 }, { "epoch": 86.4275, "grad_norm": 7.375, "learning_rate": 9.506954960833869e-07, "loss": 1.4132, "step": 34571 }, { "epoch": 86.43, "grad_norm": 6.65625, "learning_rate": 9.503509549819934e-07, "loss": 1.3544, "step": 34572 }, { "epoch": 86.4325, "grad_norm": 16.5, "learning_rate": 9.500064732097336e-07, "loss": 1.3823, "step": 34573 }, { "epoch": 86.435, "grad_norm": 7.875, "learning_rate": 9.496620507688725e-07, "loss": 1.3138, "step": 34574 }, { "epoch": 86.4375, "grad_norm": 6.96875, "learning_rate": 9.493176876616616e-07, "loss": 1.3529, "step": 34575 }, { "epoch": 86.44, "grad_norm": 7.34375, "learning_rate": 9.489733838903648e-07, "loss": 1.347, "step": 34576 }, { "epoch": 86.4425, "grad_norm": 8.25, "learning_rate": 9.486291394572333e-07, "loss": 1.3595, "step": 34577 }, { "epoch": 86.445, "grad_norm": 8.1875, "learning_rate": 9.482849543645289e-07, "loss": 1.3026, "step": 34578 }, { "epoch": 86.4475, "grad_norm": 7.40625, "learning_rate": 9.479408286145065e-07, "loss": 1.243, "step": 34579 }, { "epoch": 86.45, "grad_norm": 5.75, "learning_rate": 9.475967622094207e-07, "loss": 1.4048, "step": 34580 }, { "epoch": 86.4525, "grad_norm": 8.1875, "learning_rate": 9.472527551515276e-07, "loss": 1.2926, "step": 34581 }, { "epoch": 86.455, "grad_norm": 9.5, "learning_rate": 9.469088074430844e-07, "loss": 1.3461, "step": 34582 }, { "epoch": 86.4575, "grad_norm": 9.25, "learning_rate": 9.465649190863458e-07, "loss": 1.2593, "step": 34583 }, { "epoch": 86.46, "grad_norm": 5.28125, "learning_rate": 9.462210900835622e-07, "loss": 1.2852, "step": 34584 }, { "epoch": 86.4625, "grad_norm": 7.21875, "learning_rate": 9.458773204369919e-07, "loss": 1.1625, "step": 34585 }, { "epoch": 86.465, "grad_norm": 5.875, "learning_rate": 9.455336101488887e-07, "loss": 1.3949, "step": 34586 }, { "epoch": 86.4675, "grad_norm": 13.0, "learning_rate": 9.451899592215052e-07, "loss": 1.3596, "step": 34587 }, { "epoch": 86.47, "grad_norm": 8.625, "learning_rate": 9.448463676570929e-07, "loss": 1.2602, "step": 34588 }, { "epoch": 86.4725, "grad_norm": 6.3125, "learning_rate": 9.445028354579067e-07, "loss": 1.3425, "step": 34589 }, { "epoch": 86.475, "grad_norm": 7.15625, "learning_rate": 9.441593626261947e-07, "loss": 1.351, "step": 34590 }, { "epoch": 86.4775, "grad_norm": 6.28125, "learning_rate": 9.438159491642141e-07, "loss": 1.1315, "step": 34591 }, { "epoch": 86.48, "grad_norm": 8.0625, "learning_rate": 9.434725950742119e-07, "loss": 1.2904, "step": 34592 }, { "epoch": 86.4825, "grad_norm": 5.28125, "learning_rate": 9.43129300358443e-07, "loss": 1.415, "step": 34593 }, { "epoch": 86.485, "grad_norm": 7.6875, "learning_rate": 9.427860650191556e-07, "loss": 1.3545, "step": 34594 }, { "epoch": 86.4875, "grad_norm": 11.0, "learning_rate": 9.42442889058599e-07, "loss": 1.3359, "step": 34595 }, { "epoch": 86.49, "grad_norm": 5.75, "learning_rate": 9.420997724790238e-07, "loss": 1.2431, "step": 34596 }, { "epoch": 86.4925, "grad_norm": 6.59375, "learning_rate": 9.417567152826823e-07, "loss": 1.3032, "step": 34597 }, { "epoch": 86.495, "grad_norm": 8.875, "learning_rate": 9.414137174718219e-07, "loss": 1.2618, "step": 34598 }, { "epoch": 86.4975, "grad_norm": 8.125, "learning_rate": 9.410707790486873e-07, "loss": 1.3294, "step": 34599 }, { "epoch": 86.5, "grad_norm": 5.78125, "learning_rate": 9.407279000155311e-07, "loss": 1.1003, "step": 34600 }, { "epoch": 86.5025, "grad_norm": 4.9375, "learning_rate": 9.403850803746017e-07, "loss": 1.2218, "step": 34601 }, { "epoch": 86.505, "grad_norm": 7.875, "learning_rate": 9.40042320128145e-07, "loss": 1.4246, "step": 34602 }, { "epoch": 86.5075, "grad_norm": 8.0, "learning_rate": 9.396996192784069e-07, "loss": 1.368, "step": 34603 }, { "epoch": 86.51, "grad_norm": 6.84375, "learning_rate": 9.393569778276379e-07, "loss": 1.3707, "step": 34604 }, { "epoch": 86.5125, "grad_norm": 7.96875, "learning_rate": 9.390143957780795e-07, "loss": 1.3614, "step": 34605 }, { "epoch": 86.515, "grad_norm": 8.9375, "learning_rate": 9.38671873131981e-07, "loss": 1.3894, "step": 34606 }, { "epoch": 86.5175, "grad_norm": 8.0625, "learning_rate": 9.383294098915863e-07, "loss": 1.4228, "step": 34607 }, { "epoch": 86.52, "grad_norm": 7.28125, "learning_rate": 9.379870060591434e-07, "loss": 1.4519, "step": 34608 }, { "epoch": 86.5225, "grad_norm": 9.625, "learning_rate": 9.37644661636895e-07, "loss": 1.4393, "step": 34609 }, { "epoch": 86.525, "grad_norm": 10.9375, "learning_rate": 9.373023766270828e-07, "loss": 1.4524, "step": 34610 }, { "epoch": 86.5275, "grad_norm": 11.625, "learning_rate": 9.369601510319548e-07, "loss": 1.4459, "step": 34611 }, { "epoch": 86.53, "grad_norm": 8.1875, "learning_rate": 9.366179848537549e-07, "loss": 1.3914, "step": 34612 }, { "epoch": 86.5325, "grad_norm": 11.1875, "learning_rate": 9.362758780947234e-07, "loss": 1.3548, "step": 34613 }, { "epoch": 86.535, "grad_norm": 9.0625, "learning_rate": 9.359338307571042e-07, "loss": 1.2737, "step": 34614 }, { "epoch": 86.5375, "grad_norm": 8.6875, "learning_rate": 9.355918428431398e-07, "loss": 1.4787, "step": 34615 }, { "epoch": 86.54, "grad_norm": 7.4375, "learning_rate": 9.352499143550742e-07, "loss": 1.4726, "step": 34616 }, { "epoch": 86.5425, "grad_norm": 7.125, "learning_rate": 9.349080452951476e-07, "loss": 1.3348, "step": 34617 }, { "epoch": 86.545, "grad_norm": 7.625, "learning_rate": 9.345662356655994e-07, "loss": 1.3119, "step": 34618 }, { "epoch": 86.5475, "grad_norm": 9.0, "learning_rate": 9.342244854686733e-07, "loss": 1.1778, "step": 34619 }, { "epoch": 86.55, "grad_norm": 8.25, "learning_rate": 9.338827947066076e-07, "loss": 1.2333, "step": 34620 }, { "epoch": 86.5525, "grad_norm": 7.21875, "learning_rate": 9.33541163381645e-07, "loss": 1.2562, "step": 34621 }, { "epoch": 86.555, "grad_norm": 7.59375, "learning_rate": 9.331995914960212e-07, "loss": 1.4103, "step": 34622 }, { "epoch": 86.5575, "grad_norm": 9.0, "learning_rate": 9.328580790519804e-07, "loss": 1.3345, "step": 34623 }, { "epoch": 86.56, "grad_norm": 7.0625, "learning_rate": 9.325166260517593e-07, "loss": 1.3706, "step": 34624 }, { "epoch": 86.5625, "grad_norm": 5.90625, "learning_rate": 9.321752324975952e-07, "loss": 1.515, "step": 34625 }, { "epoch": 86.565, "grad_norm": 10.3125, "learning_rate": 9.318338983917263e-07, "loss": 1.4841, "step": 34626 }, { "epoch": 86.5675, "grad_norm": 10.875, "learning_rate": 9.314926237363942e-07, "loss": 1.2787, "step": 34627 }, { "epoch": 86.57, "grad_norm": 6.9375, "learning_rate": 9.311514085338325e-07, "loss": 1.1404, "step": 34628 }, { "epoch": 86.5725, "grad_norm": 6.4375, "learning_rate": 9.308102527862805e-07, "loss": 1.3605, "step": 34629 }, { "epoch": 86.575, "grad_norm": 9.3125, "learning_rate": 9.304691564959689e-07, "loss": 1.3103, "step": 34630 }, { "epoch": 86.5775, "grad_norm": 5.71875, "learning_rate": 9.301281196651424e-07, "loss": 1.2963, "step": 34631 }, { "epoch": 86.58, "grad_norm": 9.5, "learning_rate": 9.297871422960336e-07, "loss": 1.4965, "step": 34632 }, { "epoch": 86.5825, "grad_norm": 5.65625, "learning_rate": 9.294462243908764e-07, "loss": 1.1627, "step": 34633 }, { "epoch": 86.585, "grad_norm": 10.0, "learning_rate": 9.291053659519056e-07, "loss": 1.3345, "step": 34634 }, { "epoch": 86.5875, "grad_norm": 8.4375, "learning_rate": 9.287645669813572e-07, "loss": 1.4492, "step": 34635 }, { "epoch": 86.59, "grad_norm": 6.46875, "learning_rate": 9.284238274814672e-07, "loss": 1.2187, "step": 34636 }, { "epoch": 86.5925, "grad_norm": 7.34375, "learning_rate": 9.280831474544672e-07, "loss": 1.3124, "step": 34637 }, { "epoch": 86.595, "grad_norm": 8.0625, "learning_rate": 9.277425269025886e-07, "loss": 1.3624, "step": 34638 }, { "epoch": 86.5975, "grad_norm": 7.0, "learning_rate": 9.274019658280697e-07, "loss": 1.4137, "step": 34639 }, { "epoch": 86.6, "grad_norm": 5.5, "learning_rate": 9.270614642331377e-07, "loss": 1.1636, "step": 34640 }, { "epoch": 86.6025, "grad_norm": 5.90625, "learning_rate": 9.267210221200285e-07, "loss": 1.2821, "step": 34641 }, { "epoch": 86.605, "grad_norm": 9.875, "learning_rate": 9.263806394909747e-07, "loss": 1.2353, "step": 34642 }, { "epoch": 86.6075, "grad_norm": 7.65625, "learning_rate": 9.260403163482068e-07, "loss": 1.3039, "step": 34643 }, { "epoch": 86.61, "grad_norm": 10.6875, "learning_rate": 9.257000526939541e-07, "loss": 1.3135, "step": 34644 }, { "epoch": 86.6125, "grad_norm": 10.8125, "learning_rate": 9.25359848530446e-07, "loss": 1.3914, "step": 34645 }, { "epoch": 86.615, "grad_norm": 7.40625, "learning_rate": 9.250197038599195e-07, "loss": 1.4323, "step": 34646 }, { "epoch": 86.6175, "grad_norm": 5.21875, "learning_rate": 9.246796186846008e-07, "loss": 1.2562, "step": 34647 }, { "epoch": 86.62, "grad_norm": 10.375, "learning_rate": 9.243395930067189e-07, "loss": 1.326, "step": 34648 }, { "epoch": 86.6225, "grad_norm": 7.59375, "learning_rate": 9.239996268285012e-07, "loss": 1.2652, "step": 34649 }, { "epoch": 86.625, "grad_norm": 7.96875, "learning_rate": 9.23659720152179e-07, "loss": 1.3883, "step": 34650 }, { "epoch": 86.6275, "grad_norm": 6.4375, "learning_rate": 9.23319872979983e-07, "loss": 1.4757, "step": 34651 }, { "epoch": 86.63, "grad_norm": 9.125, "learning_rate": 9.22980085314138e-07, "loss": 1.4507, "step": 34652 }, { "epoch": 86.6325, "grad_norm": 8.75, "learning_rate": 9.22640357156871e-07, "loss": 1.4032, "step": 34653 }, { "epoch": 86.635, "grad_norm": 8.875, "learning_rate": 9.223006885104114e-07, "loss": 1.2674, "step": 34654 }, { "epoch": 86.6375, "grad_norm": 8.3125, "learning_rate": 9.21961079376984e-07, "loss": 1.4099, "step": 34655 }, { "epoch": 86.64, "grad_norm": 6.6875, "learning_rate": 9.216215297588182e-07, "loss": 1.3462, "step": 34656 }, { "epoch": 86.6425, "grad_norm": 6.4375, "learning_rate": 9.212820396581368e-07, "loss": 1.273, "step": 34657 }, { "epoch": 86.645, "grad_norm": 6.15625, "learning_rate": 9.20942609077169e-07, "loss": 1.4349, "step": 34658 }, { "epoch": 86.6475, "grad_norm": 7.09375, "learning_rate": 9.206032380181385e-07, "loss": 1.3058, "step": 34659 }, { "epoch": 86.65, "grad_norm": 8.625, "learning_rate": 9.202639264832669e-07, "loss": 1.4087, "step": 34660 }, { "epoch": 86.6525, "grad_norm": 9.75, "learning_rate": 9.199246744747825e-07, "loss": 1.2325, "step": 34661 }, { "epoch": 86.655, "grad_norm": 8.875, "learning_rate": 9.19585481994909e-07, "loss": 1.3521, "step": 34662 }, { "epoch": 86.6575, "grad_norm": 7.34375, "learning_rate": 9.192463490458714e-07, "loss": 1.4068, "step": 34663 }, { "epoch": 86.66, "grad_norm": 6.5625, "learning_rate": 9.189072756298889e-07, "loss": 1.3207, "step": 34664 }, { "epoch": 86.6625, "grad_norm": 7.6875, "learning_rate": 9.185682617491865e-07, "loss": 1.2546, "step": 34665 }, { "epoch": 86.665, "grad_norm": 9.5625, "learning_rate": 9.182293074059889e-07, "loss": 1.3518, "step": 34666 }, { "epoch": 86.6675, "grad_norm": 8.1875, "learning_rate": 9.178904126025179e-07, "loss": 1.2328, "step": 34667 }, { "epoch": 86.67, "grad_norm": 9.125, "learning_rate": 9.175515773409904e-07, "loss": 1.4619, "step": 34668 }, { "epoch": 86.6725, "grad_norm": 8.4375, "learning_rate": 9.172128016236337e-07, "loss": 1.1572, "step": 34669 }, { "epoch": 86.675, "grad_norm": 6.40625, "learning_rate": 9.168740854526648e-07, "loss": 1.3665, "step": 34670 }, { "epoch": 86.6775, "grad_norm": 10.25, "learning_rate": 9.165354288303085e-07, "loss": 1.2383, "step": 34671 }, { "epoch": 86.68, "grad_norm": 4.65625, "learning_rate": 9.161968317587788e-07, "loss": 1.4069, "step": 34672 }, { "epoch": 86.6825, "grad_norm": 6.53125, "learning_rate": 9.158582942403027e-07, "loss": 1.2717, "step": 34673 }, { "epoch": 86.685, "grad_norm": 5.71875, "learning_rate": 9.15519816277095e-07, "loss": 1.2903, "step": 34674 }, { "epoch": 86.6875, "grad_norm": 7.8125, "learning_rate": 9.151813978713741e-07, "loss": 1.4002, "step": 34675 }, { "epoch": 86.69, "grad_norm": 11.5625, "learning_rate": 9.148430390253615e-07, "loss": 1.4242, "step": 34676 }, { "epoch": 86.6925, "grad_norm": 6.0, "learning_rate": 9.145047397412754e-07, "loss": 1.3574, "step": 34677 }, { "epoch": 86.695, "grad_norm": 6.9375, "learning_rate": 9.141665000213318e-07, "loss": 1.2812, "step": 34678 }, { "epoch": 86.6975, "grad_norm": 7.0625, "learning_rate": 9.138283198677489e-07, "loss": 1.3239, "step": 34679 }, { "epoch": 86.7, "grad_norm": 9.875, "learning_rate": 9.134901992827427e-07, "loss": 1.2529, "step": 34680 }, { "epoch": 86.7025, "grad_norm": 4.875, "learning_rate": 9.131521382685327e-07, "loss": 1.3863, "step": 34681 }, { "epoch": 86.705, "grad_norm": 6.90625, "learning_rate": 9.128141368273347e-07, "loss": 1.3483, "step": 34682 }, { "epoch": 86.7075, "grad_norm": 5.1875, "learning_rate": 9.124761949613614e-07, "loss": 1.2825, "step": 34683 }, { "epoch": 86.71, "grad_norm": 7.625, "learning_rate": 9.121383126728322e-07, "loss": 1.3391, "step": 34684 }, { "epoch": 86.7125, "grad_norm": 7.96875, "learning_rate": 9.118004899639587e-07, "loss": 1.2244, "step": 34685 }, { "epoch": 86.715, "grad_norm": 5.25, "learning_rate": 9.114627268369592e-07, "loss": 1.321, "step": 34686 }, { "epoch": 86.7175, "grad_norm": 8.875, "learning_rate": 9.111250232940449e-07, "loss": 1.3639, "step": 34687 }, { "epoch": 86.72, "grad_norm": 6.96875, "learning_rate": 9.107873793374322e-07, "loss": 1.3182, "step": 34688 }, { "epoch": 86.7225, "grad_norm": 8.75, "learning_rate": 9.104497949693348e-07, "loss": 1.4467, "step": 34689 }, { "epoch": 86.725, "grad_norm": 7.78125, "learning_rate": 9.101122701919629e-07, "loss": 1.2839, "step": 34690 }, { "epoch": 86.7275, "grad_norm": 9.875, "learning_rate": 9.097748050075306e-07, "loss": 1.3805, "step": 34691 }, { "epoch": 86.73, "grad_norm": 7.59375, "learning_rate": 9.094373994182525e-07, "loss": 1.3652, "step": 34692 }, { "epoch": 86.7325, "grad_norm": 12.5625, "learning_rate": 9.091000534263394e-07, "loss": 1.5022, "step": 34693 }, { "epoch": 86.735, "grad_norm": 6.53125, "learning_rate": 9.087627670340016e-07, "loss": 1.0917, "step": 34694 }, { "epoch": 86.7375, "grad_norm": 8.25, "learning_rate": 9.084255402434506e-07, "loss": 1.3029, "step": 34695 }, { "epoch": 86.74, "grad_norm": 8.5625, "learning_rate": 9.08088373056899e-07, "loss": 1.1948, "step": 34696 }, { "epoch": 86.7425, "grad_norm": 7.09375, "learning_rate": 9.077512654765574e-07, "loss": 1.2311, "step": 34697 }, { "epoch": 86.745, "grad_norm": 5.78125, "learning_rate": 9.074142175046318e-07, "loss": 1.2249, "step": 34698 }, { "epoch": 86.7475, "grad_norm": 11.6875, "learning_rate": 9.070772291433372e-07, "loss": 1.3368, "step": 34699 }, { "epoch": 86.75, "grad_norm": 8.8125, "learning_rate": 9.067403003948783e-07, "loss": 1.4522, "step": 34700 }, { "epoch": 86.7525, "grad_norm": 7.15625, "learning_rate": 9.064034312614689e-07, "loss": 1.145, "step": 34701 }, { "epoch": 86.755, "grad_norm": 5.375, "learning_rate": 9.060666217453117e-07, "loss": 1.2638, "step": 34702 }, { "epoch": 86.7575, "grad_norm": 7.1875, "learning_rate": 9.057298718486196e-07, "loss": 1.2935, "step": 34703 }, { "epoch": 86.76, "grad_norm": 5.46875, "learning_rate": 9.053931815735995e-07, "loss": 1.2804, "step": 34704 }, { "epoch": 86.7625, "grad_norm": 10.125, "learning_rate": 9.050565509224552e-07, "loss": 1.3687, "step": 34705 }, { "epoch": 86.765, "grad_norm": 6.25, "learning_rate": 9.04719979897396e-07, "loss": 1.2946, "step": 34706 }, { "epoch": 86.7675, "grad_norm": 8.125, "learning_rate": 9.043834685006303e-07, "loss": 1.1705, "step": 34707 }, { "epoch": 86.77, "grad_norm": 8.875, "learning_rate": 9.04047016734364e-07, "loss": 1.1552, "step": 34708 }, { "epoch": 86.7725, "grad_norm": 6.84375, "learning_rate": 9.037106246007987e-07, "loss": 1.3773, "step": 34709 }, { "epoch": 86.775, "grad_norm": 7.71875, "learning_rate": 9.033742921021415e-07, "loss": 1.32, "step": 34710 }, { "epoch": 86.7775, "grad_norm": 9.3125, "learning_rate": 9.030380192406019e-07, "loss": 1.2984, "step": 34711 }, { "epoch": 86.78, "grad_norm": 8.375, "learning_rate": 9.027018060183801e-07, "loss": 1.1153, "step": 34712 }, { "epoch": 86.7825, "grad_norm": 9.5, "learning_rate": 9.023656524376801e-07, "loss": 1.2282, "step": 34713 }, { "epoch": 86.785, "grad_norm": 8.1875, "learning_rate": 9.020295585007077e-07, "loss": 1.3088, "step": 34714 }, { "epoch": 86.7875, "grad_norm": 5.9375, "learning_rate": 9.016935242096647e-07, "loss": 1.3958, "step": 34715 }, { "epoch": 86.79, "grad_norm": 4.8125, "learning_rate": 9.013575495667559e-07, "loss": 1.3637, "step": 34716 }, { "epoch": 86.7925, "grad_norm": 7.5625, "learning_rate": 9.010216345741807e-07, "loss": 1.3955, "step": 34717 }, { "epoch": 86.795, "grad_norm": 9.875, "learning_rate": 9.006857792341462e-07, "loss": 1.428, "step": 34718 }, { "epoch": 86.7975, "grad_norm": 7.125, "learning_rate": 9.003499835488517e-07, "loss": 1.3324, "step": 34719 }, { "epoch": 86.8, "grad_norm": 7.5, "learning_rate": 9.000142475204965e-07, "loss": 1.4074, "step": 34720 }, { "epoch": 86.8025, "grad_norm": 6.125, "learning_rate": 8.996785711512834e-07, "loss": 1.3005, "step": 34721 }, { "epoch": 86.805, "grad_norm": 7.46875, "learning_rate": 8.993429544434152e-07, "loss": 1.3065, "step": 34722 }, { "epoch": 86.8075, "grad_norm": 10.25, "learning_rate": 8.990073973990909e-07, "loss": 1.2003, "step": 34723 }, { "epoch": 86.81, "grad_norm": 6.46875, "learning_rate": 8.986719000205079e-07, "loss": 1.46, "step": 34724 }, { "epoch": 86.8125, "grad_norm": 9.9375, "learning_rate": 8.983364623098678e-07, "loss": 1.3251, "step": 34725 }, { "epoch": 86.815, "grad_norm": 9.125, "learning_rate": 8.980010842693709e-07, "loss": 1.2787, "step": 34726 }, { "epoch": 86.8175, "grad_norm": 8.5625, "learning_rate": 8.976657659012155e-07, "loss": 1.3056, "step": 34727 }, { "epoch": 86.82, "grad_norm": 9.5625, "learning_rate": 8.973305072075966e-07, "loss": 1.2817, "step": 34728 }, { "epoch": 86.8225, "grad_norm": 9.625, "learning_rate": 8.969953081907167e-07, "loss": 1.4786, "step": 34729 }, { "epoch": 86.825, "grad_norm": 6.78125, "learning_rate": 8.966601688527687e-07, "loss": 1.2615, "step": 34730 }, { "epoch": 86.8275, "grad_norm": 7.0625, "learning_rate": 8.963250891959552e-07, "loss": 1.2932, "step": 34731 }, { "epoch": 86.83, "grad_norm": 8.625, "learning_rate": 8.959900692224676e-07, "loss": 1.3495, "step": 34732 }, { "epoch": 86.8325, "grad_norm": 7.40625, "learning_rate": 8.956551089345078e-07, "loss": 1.1837, "step": 34733 }, { "epoch": 86.835, "grad_norm": 7.34375, "learning_rate": 8.953202083342671e-07, "loss": 1.4114, "step": 34734 }, { "epoch": 86.8375, "grad_norm": 6.8125, "learning_rate": 8.949853674239418e-07, "loss": 1.3481, "step": 34735 }, { "epoch": 86.84, "grad_norm": 8.875, "learning_rate": 8.946505862057286e-07, "loss": 1.2716, "step": 34736 }, { "epoch": 86.8425, "grad_norm": 8.4375, "learning_rate": 8.943158646818229e-07, "loss": 1.236, "step": 34737 }, { "epoch": 86.845, "grad_norm": 9.25, "learning_rate": 8.939812028544182e-07, "loss": 1.4246, "step": 34738 }, { "epoch": 86.8475, "grad_norm": 5.28125, "learning_rate": 8.936466007257061e-07, "loss": 1.331, "step": 34739 }, { "epoch": 86.85, "grad_norm": 6.34375, "learning_rate": 8.933120582978827e-07, "loss": 1.5227, "step": 34740 }, { "epoch": 86.8525, "grad_norm": 9.25, "learning_rate": 8.929775755731429e-07, "loss": 1.4096, "step": 34741 }, { "epoch": 86.855, "grad_norm": 5.53125, "learning_rate": 8.926431525536783e-07, "loss": 1.1901, "step": 34742 }, { "epoch": 86.8575, "grad_norm": 6.59375, "learning_rate": 8.923087892416782e-07, "loss": 1.284, "step": 34743 }, { "epoch": 86.86, "grad_norm": 24.625, "learning_rate": 8.919744856393386e-07, "loss": 1.2582, "step": 34744 }, { "epoch": 86.8625, "grad_norm": 8.3125, "learning_rate": 8.91640241748849e-07, "loss": 1.318, "step": 34745 }, { "epoch": 86.865, "grad_norm": 14.6875, "learning_rate": 8.91306057572403e-07, "loss": 1.1861, "step": 34746 }, { "epoch": 86.8675, "grad_norm": 6.21875, "learning_rate": 8.90971933112188e-07, "loss": 1.3754, "step": 34747 }, { "epoch": 86.87, "grad_norm": 7.9375, "learning_rate": 8.906378683703986e-07, "loss": 1.2171, "step": 34748 }, { "epoch": 86.8725, "grad_norm": 9.375, "learning_rate": 8.903038633492223e-07, "loss": 1.2108, "step": 34749 }, { "epoch": 86.875, "grad_norm": 9.0625, "learning_rate": 8.89969918050847e-07, "loss": 1.3773, "step": 34750 }, { "epoch": 86.8775, "grad_norm": 5.3125, "learning_rate": 8.896360324774656e-07, "loss": 1.2943, "step": 34751 }, { "epoch": 86.88, "grad_norm": 4.78125, "learning_rate": 8.893022066312674e-07, "loss": 1.2587, "step": 34752 }, { "epoch": 86.8825, "grad_norm": 7.65625, "learning_rate": 8.889684405144383e-07, "loss": 1.3337, "step": 34753 }, { "epoch": 86.885, "grad_norm": 6.8125, "learning_rate": 8.886347341291668e-07, "loss": 1.2338, "step": 34754 }, { "epoch": 86.8875, "grad_norm": 11.0, "learning_rate": 8.88301087477641e-07, "loss": 1.4842, "step": 34755 }, { "epoch": 86.89, "grad_norm": 7.65625, "learning_rate": 8.879675005620503e-07, "loss": 1.2265, "step": 34756 }, { "epoch": 86.8925, "grad_norm": 12.0625, "learning_rate": 8.876339733845807e-07, "loss": 1.3301, "step": 34757 }, { "epoch": 86.895, "grad_norm": 7.84375, "learning_rate": 8.873005059474149e-07, "loss": 1.2404, "step": 34758 }, { "epoch": 86.8975, "grad_norm": 6.0625, "learning_rate": 8.869670982527456e-07, "loss": 1.2852, "step": 34759 }, { "epoch": 86.9, "grad_norm": 9.375, "learning_rate": 8.866337503027523e-07, "loss": 1.226, "step": 34760 }, { "epoch": 86.9025, "grad_norm": 6.6875, "learning_rate": 8.863004620996263e-07, "loss": 1.3539, "step": 34761 }, { "epoch": 86.905, "grad_norm": 6.375, "learning_rate": 8.859672336455471e-07, "loss": 1.448, "step": 34762 }, { "epoch": 86.9075, "grad_norm": 6.21875, "learning_rate": 8.856340649427042e-07, "loss": 1.2098, "step": 34763 }, { "epoch": 86.91, "grad_norm": 6.46875, "learning_rate": 8.853009559932802e-07, "loss": 1.2692, "step": 34764 }, { "epoch": 86.9125, "grad_norm": 9.25, "learning_rate": 8.849679067994555e-07, "loss": 1.3203, "step": 34765 }, { "epoch": 86.915, "grad_norm": 8.6875, "learning_rate": 8.846349173634173e-07, "loss": 1.2206, "step": 34766 }, { "epoch": 86.9175, "grad_norm": 6.65625, "learning_rate": 8.843019876873493e-07, "loss": 1.315, "step": 34767 }, { "epoch": 86.92, "grad_norm": 6.9375, "learning_rate": 8.839691177734322e-07, "loss": 1.561, "step": 34768 }, { "epoch": 86.9225, "grad_norm": 7.09375, "learning_rate": 8.836363076238497e-07, "loss": 1.3815, "step": 34769 }, { "epoch": 86.925, "grad_norm": 7.03125, "learning_rate": 8.833035572407778e-07, "loss": 1.4108, "step": 34770 }, { "epoch": 86.9275, "grad_norm": 7.8125, "learning_rate": 8.829708666264081e-07, "loss": 1.3077, "step": 34771 }, { "epoch": 86.93, "grad_norm": 9.0, "learning_rate": 8.826382357829155e-07, "loss": 1.302, "step": 34772 }, { "epoch": 86.9325, "grad_norm": 8.25, "learning_rate": 8.823056647124817e-07, "loss": 1.4117, "step": 34773 }, { "epoch": 86.935, "grad_norm": 6.5, "learning_rate": 8.819731534172859e-07, "loss": 1.3627, "step": 34774 }, { "epoch": 86.9375, "grad_norm": 7.0, "learning_rate": 8.816407018995088e-07, "loss": 1.1225, "step": 34775 }, { "epoch": 86.94, "grad_norm": 5.6875, "learning_rate": 8.813083101613329e-07, "loss": 1.1996, "step": 34776 }, { "epoch": 86.9425, "grad_norm": 7.625, "learning_rate": 8.809759782049321e-07, "loss": 1.3396, "step": 34777 }, { "epoch": 86.945, "grad_norm": 8.4375, "learning_rate": 8.806437060324902e-07, "loss": 1.3903, "step": 34778 }, { "epoch": 86.9475, "grad_norm": 9.125, "learning_rate": 8.803114936461832e-07, "loss": 1.2324, "step": 34779 }, { "epoch": 86.95, "grad_norm": 6.5625, "learning_rate": 8.799793410481871e-07, "loss": 1.2973, "step": 34780 }, { "epoch": 86.9525, "grad_norm": 7.0625, "learning_rate": 8.796472482406814e-07, "loss": 1.3232, "step": 34781 }, { "epoch": 86.955, "grad_norm": 6.65625, "learning_rate": 8.793152152258455e-07, "loss": 1.3718, "step": 34782 }, { "epoch": 86.9575, "grad_norm": 5.90625, "learning_rate": 8.789832420058541e-07, "loss": 1.2768, "step": 34783 }, { "epoch": 86.96, "grad_norm": 5.875, "learning_rate": 8.786513285828835e-07, "loss": 1.4184, "step": 34784 }, { "epoch": 86.9625, "grad_norm": 9.375, "learning_rate": 8.783194749591061e-07, "loss": 1.4204, "step": 34785 }, { "epoch": 86.965, "grad_norm": 11.25, "learning_rate": 8.77987681136705e-07, "loss": 1.3339, "step": 34786 }, { "epoch": 86.9675, "grad_norm": 17.5, "learning_rate": 8.776559471178503e-07, "loss": 1.377, "step": 34787 }, { "epoch": 86.97, "grad_norm": 5.34375, "learning_rate": 8.773242729047193e-07, "loss": 1.2158, "step": 34788 }, { "epoch": 86.9725, "grad_norm": 5.15625, "learning_rate": 8.769926584994814e-07, "loss": 1.2024, "step": 34789 }, { "epoch": 86.975, "grad_norm": 7.0625, "learning_rate": 8.766611039043193e-07, "loss": 1.3306, "step": 34790 }, { "epoch": 86.9775, "grad_norm": 6.125, "learning_rate": 8.763296091214002e-07, "loss": 1.3199, "step": 34791 }, { "epoch": 86.98, "grad_norm": 8.625, "learning_rate": 8.759981741529e-07, "loss": 1.4323, "step": 34792 }, { "epoch": 86.9825, "grad_norm": 8.9375, "learning_rate": 8.756667990009893e-07, "loss": 1.3317, "step": 34793 }, { "epoch": 86.985, "grad_norm": 10.0625, "learning_rate": 8.753354836678407e-07, "loss": 1.4188, "step": 34794 }, { "epoch": 86.9875, "grad_norm": 5.03125, "learning_rate": 8.750042281556303e-07, "loss": 1.3665, "step": 34795 }, { "epoch": 86.99, "grad_norm": 9.6875, "learning_rate": 8.746730324665264e-07, "loss": 1.5914, "step": 34796 }, { "epoch": 86.9925, "grad_norm": 8.6875, "learning_rate": 8.743418966026995e-07, "loss": 1.3941, "step": 34797 }, { "epoch": 86.995, "grad_norm": 6.28125, "learning_rate": 8.740108205663234e-07, "loss": 1.3482, "step": 34798 }, { "epoch": 86.9975, "grad_norm": 7.46875, "learning_rate": 8.736798043595652e-07, "loss": 1.3189, "step": 34799 }, { "epoch": 87.0, "grad_norm": 5.875, "learning_rate": 8.733488479845997e-07, "loss": 1.3618, "step": 34800 }, { "epoch": 87.0025, "grad_norm": 5.0625, "learning_rate": 8.73017951443591e-07, "loss": 1.4687, "step": 34801 }, { "epoch": 87.005, "grad_norm": 10.3125, "learning_rate": 8.726871147387139e-07, "loss": 1.3216, "step": 34802 }, { "epoch": 87.0075, "grad_norm": 5.84375, "learning_rate": 8.723563378721345e-07, "loss": 1.1509, "step": 34803 }, { "epoch": 87.01, "grad_norm": 6.65625, "learning_rate": 8.720256208460187e-07, "loss": 1.2836, "step": 34804 }, { "epoch": 87.0125, "grad_norm": 9.0625, "learning_rate": 8.716949636625405e-07, "loss": 1.3754, "step": 34805 }, { "epoch": 87.015, "grad_norm": 6.59375, "learning_rate": 8.713643663238657e-07, "loss": 1.3139, "step": 34806 }, { "epoch": 87.0175, "grad_norm": 6.84375, "learning_rate": 8.710338288321596e-07, "loss": 1.2412, "step": 34807 }, { "epoch": 87.02, "grad_norm": 8.625, "learning_rate": 8.70703351189589e-07, "loss": 1.2971, "step": 34808 }, { "epoch": 87.0225, "grad_norm": 7.71875, "learning_rate": 8.703729333983224e-07, "loss": 1.1906, "step": 34809 }, { "epoch": 87.025, "grad_norm": 8.5, "learning_rate": 8.700425754605268e-07, "loss": 1.304, "step": 34810 }, { "epoch": 87.0275, "grad_norm": 6.5, "learning_rate": 8.697122773783673e-07, "loss": 1.2714, "step": 34811 }, { "epoch": 87.03, "grad_norm": 9.3125, "learning_rate": 8.693820391540064e-07, "loss": 1.2415, "step": 34812 }, { "epoch": 87.0325, "grad_norm": 7.625, "learning_rate": 8.690518607896137e-07, "loss": 1.3204, "step": 34813 }, { "epoch": 87.035, "grad_norm": 7.625, "learning_rate": 8.687217422873495e-07, "loss": 1.2161, "step": 34814 }, { "epoch": 87.0375, "grad_norm": 7.46875, "learning_rate": 8.683916836493822e-07, "loss": 1.2329, "step": 34815 }, { "epoch": 87.04, "grad_norm": 10.5625, "learning_rate": 8.680616848778711e-07, "loss": 1.3215, "step": 34816 }, { "epoch": 87.0425, "grad_norm": 6.75, "learning_rate": 8.677317459749845e-07, "loss": 1.307, "step": 34817 }, { "epoch": 87.045, "grad_norm": 5.40625, "learning_rate": 8.67401866942883e-07, "loss": 1.3078, "step": 34818 }, { "epoch": 87.0475, "grad_norm": 7.0, "learning_rate": 8.67072047783728e-07, "loss": 1.42, "step": 34819 }, { "epoch": 87.05, "grad_norm": 11.5625, "learning_rate": 8.667422884996823e-07, "loss": 1.4467, "step": 34820 }, { "epoch": 87.0525, "grad_norm": 8.3125, "learning_rate": 8.664125890929109e-07, "loss": 1.3138, "step": 34821 }, { "epoch": 87.055, "grad_norm": 6.71875, "learning_rate": 8.660829495655732e-07, "loss": 1.5251, "step": 34822 }, { "epoch": 87.0575, "grad_norm": 5.6875, "learning_rate": 8.657533699198284e-07, "loss": 1.4039, "step": 34823 }, { "epoch": 87.06, "grad_norm": 6.15625, "learning_rate": 8.654238501578393e-07, "loss": 1.2693, "step": 34824 }, { "epoch": 87.0625, "grad_norm": 6.28125, "learning_rate": 8.650943902817677e-07, "loss": 1.2576, "step": 34825 }, { "epoch": 87.065, "grad_norm": 6.4375, "learning_rate": 8.647649902937705e-07, "loss": 1.1893, "step": 34826 }, { "epoch": 87.0675, "grad_norm": 7.0625, "learning_rate": 8.644356501960072e-07, "loss": 1.3501, "step": 34827 }, { "epoch": 87.07, "grad_norm": 6.28125, "learning_rate": 8.641063699906404e-07, "loss": 1.3634, "step": 34828 }, { "epoch": 87.0725, "grad_norm": 5.875, "learning_rate": 8.637771496798253e-07, "loss": 1.4263, "step": 34829 }, { "epoch": 87.075, "grad_norm": 8.0, "learning_rate": 8.634479892657222e-07, "loss": 1.5292, "step": 34830 }, { "epoch": 87.0775, "grad_norm": 6.3125, "learning_rate": 8.631188887504871e-07, "loss": 1.4354, "step": 34831 }, { "epoch": 87.08, "grad_norm": 6.78125, "learning_rate": 8.627898481362817e-07, "loss": 1.1834, "step": 34832 }, { "epoch": 87.0825, "grad_norm": 7.0625, "learning_rate": 8.624608674252589e-07, "loss": 1.314, "step": 34833 }, { "epoch": 87.085, "grad_norm": 9.375, "learning_rate": 8.621319466195766e-07, "loss": 1.5911, "step": 34834 }, { "epoch": 87.0875, "grad_norm": 8.4375, "learning_rate": 8.618030857213899e-07, "loss": 1.2471, "step": 34835 }, { "epoch": 87.09, "grad_norm": 8.8125, "learning_rate": 8.614742847328594e-07, "loss": 1.2529, "step": 34836 }, { "epoch": 87.0925, "grad_norm": 7.84375, "learning_rate": 8.611455436561378e-07, "loss": 1.3934, "step": 34837 }, { "epoch": 87.095, "grad_norm": 8.1875, "learning_rate": 8.608168624933789e-07, "loss": 1.3561, "step": 34838 }, { "epoch": 87.0975, "grad_norm": 8.4375, "learning_rate": 8.604882412467375e-07, "loss": 1.3034, "step": 34839 }, { "epoch": 87.1, "grad_norm": 12.25, "learning_rate": 8.60159679918372e-07, "loss": 1.3902, "step": 34840 }, { "epoch": 87.1025, "grad_norm": 6.15625, "learning_rate": 8.59831178510434e-07, "loss": 1.3619, "step": 34841 }, { "epoch": 87.105, "grad_norm": 6.875, "learning_rate": 8.595027370250752e-07, "loss": 1.3409, "step": 34842 }, { "epoch": 87.1075, "grad_norm": 8.4375, "learning_rate": 8.591743554644527e-07, "loss": 1.3066, "step": 34843 }, { "epoch": 87.11, "grad_norm": 7.3125, "learning_rate": 8.588460338307159e-07, "loss": 1.3027, "step": 34844 }, { "epoch": 87.1125, "grad_norm": 7.53125, "learning_rate": 8.585177721260197e-07, "loss": 1.2803, "step": 34845 }, { "epoch": 87.115, "grad_norm": 6.84375, "learning_rate": 8.581895703525134e-07, "loss": 1.2787, "step": 34846 }, { "epoch": 87.1175, "grad_norm": 7.96875, "learning_rate": 8.578614285123521e-07, "loss": 1.4743, "step": 34847 }, { "epoch": 87.12, "grad_norm": 5.875, "learning_rate": 8.575333466076863e-07, "loss": 1.2509, "step": 34848 }, { "epoch": 87.1225, "grad_norm": 7.21875, "learning_rate": 8.572053246406631e-07, "loss": 1.315, "step": 34849 }, { "epoch": 87.125, "grad_norm": 7.5625, "learning_rate": 8.568773626134363e-07, "loss": 1.2585, "step": 34850 }, { "epoch": 87.1275, "grad_norm": 7.78125, "learning_rate": 8.565494605281577e-07, "loss": 1.2436, "step": 34851 }, { "epoch": 87.13, "grad_norm": 8.1875, "learning_rate": 8.562216183869754e-07, "loss": 1.3777, "step": 34852 }, { "epoch": 87.1325, "grad_norm": 11.0625, "learning_rate": 8.558938361920355e-07, "loss": 1.3163, "step": 34853 }, { "epoch": 87.135, "grad_norm": 9.75, "learning_rate": 8.555661139454896e-07, "loss": 1.4053, "step": 34854 }, { "epoch": 87.1375, "grad_norm": 7.21875, "learning_rate": 8.552384516494894e-07, "loss": 1.3094, "step": 34855 }, { "epoch": 87.14, "grad_norm": 8.4375, "learning_rate": 8.549108493061786e-07, "loss": 1.3023, "step": 34856 }, { "epoch": 87.1425, "grad_norm": 7.96875, "learning_rate": 8.545833069177046e-07, "loss": 1.3072, "step": 34857 }, { "epoch": 87.145, "grad_norm": 11.1875, "learning_rate": 8.542558244862187e-07, "loss": 1.3935, "step": 34858 }, { "epoch": 87.1475, "grad_norm": 7.375, "learning_rate": 8.539284020138638e-07, "loss": 1.3983, "step": 34859 }, { "epoch": 87.15, "grad_norm": 9.125, "learning_rate": 8.536010395027905e-07, "loss": 1.1927, "step": 34860 }, { "epoch": 87.1525, "grad_norm": 8.375, "learning_rate": 8.532737369551403e-07, "loss": 1.2524, "step": 34861 }, { "epoch": 87.155, "grad_norm": 9.0, "learning_rate": 8.529464943730636e-07, "loss": 1.4009, "step": 34862 }, { "epoch": 87.1575, "grad_norm": 6.875, "learning_rate": 8.526193117587034e-07, "loss": 1.3998, "step": 34863 }, { "epoch": 87.16, "grad_norm": 9.625, "learning_rate": 8.522921891142034e-07, "loss": 1.3729, "step": 34864 }, { "epoch": 87.1625, "grad_norm": 7.78125, "learning_rate": 8.519651264417094e-07, "loss": 1.4441, "step": 34865 }, { "epoch": 87.165, "grad_norm": 7.25, "learning_rate": 8.516381237433679e-07, "loss": 1.2619, "step": 34866 }, { "epoch": 87.1675, "grad_norm": 7.03125, "learning_rate": 8.513111810213215e-07, "loss": 1.3065, "step": 34867 }, { "epoch": 87.17, "grad_norm": 8.25, "learning_rate": 8.509842982777105e-07, "loss": 1.3494, "step": 34868 }, { "epoch": 87.1725, "grad_norm": 6.34375, "learning_rate": 8.506574755146802e-07, "loss": 1.2033, "step": 34869 }, { "epoch": 87.175, "grad_norm": 6.8125, "learning_rate": 8.503307127343752e-07, "loss": 1.3565, "step": 34870 }, { "epoch": 87.1775, "grad_norm": 9.25, "learning_rate": 8.500040099389362e-07, "loss": 1.5115, "step": 34871 }, { "epoch": 87.18, "grad_norm": 5.90625, "learning_rate": 8.496773671305025e-07, "loss": 1.4306, "step": 34872 }, { "epoch": 87.1825, "grad_norm": 6.78125, "learning_rate": 8.493507843112192e-07, "loss": 1.2355, "step": 34873 }, { "epoch": 87.185, "grad_norm": 6.65625, "learning_rate": 8.490242614832245e-07, "loss": 1.3301, "step": 34874 }, { "epoch": 87.1875, "grad_norm": 9.75, "learning_rate": 8.486977986486633e-07, "loss": 1.2844, "step": 34875 }, { "epoch": 87.19, "grad_norm": 7.15625, "learning_rate": 8.483713958096706e-07, "loss": 1.4125, "step": 34876 }, { "epoch": 87.1925, "grad_norm": 8.0625, "learning_rate": 8.480450529683903e-07, "loss": 1.3873, "step": 34877 }, { "epoch": 87.195, "grad_norm": 8.75, "learning_rate": 8.477187701269607e-07, "loss": 1.2827, "step": 34878 }, { "epoch": 87.1975, "grad_norm": 7.59375, "learning_rate": 8.473925472875189e-07, "loss": 1.3917, "step": 34879 }, { "epoch": 87.2, "grad_norm": 6.5625, "learning_rate": 8.470663844522053e-07, "loss": 1.4499, "step": 34880 }, { "epoch": 87.2025, "grad_norm": 5.90625, "learning_rate": 8.467402816231596e-07, "loss": 1.4077, "step": 34881 }, { "epoch": 87.205, "grad_norm": 6.90625, "learning_rate": 8.464142388025188e-07, "loss": 1.3288, "step": 34882 }, { "epoch": 87.2075, "grad_norm": 10.6875, "learning_rate": 8.460882559924177e-07, "loss": 1.313, "step": 34883 }, { "epoch": 87.21, "grad_norm": 7.1875, "learning_rate": 8.457623331949961e-07, "loss": 1.2763, "step": 34884 }, { "epoch": 87.2125, "grad_norm": 7.90625, "learning_rate": 8.454364704123919e-07, "loss": 1.3748, "step": 34885 }, { "epoch": 87.215, "grad_norm": 10.125, "learning_rate": 8.451106676467402e-07, "loss": 1.3007, "step": 34886 }, { "epoch": 87.2175, "grad_norm": 7.28125, "learning_rate": 8.447849249001739e-07, "loss": 1.3901, "step": 34887 }, { "epoch": 87.22, "grad_norm": 8.125, "learning_rate": 8.444592421748332e-07, "loss": 1.4207, "step": 34888 }, { "epoch": 87.2225, "grad_norm": 7.34375, "learning_rate": 8.4413361947285e-07, "loss": 1.1625, "step": 34889 }, { "epoch": 87.225, "grad_norm": 7.34375, "learning_rate": 8.438080567963625e-07, "loss": 1.2458, "step": 34890 }, { "epoch": 87.2275, "grad_norm": 10.75, "learning_rate": 8.434825541475011e-07, "loss": 1.4139, "step": 34891 }, { "epoch": 87.23, "grad_norm": 5.625, "learning_rate": 8.431571115284032e-07, "loss": 1.3522, "step": 34892 }, { "epoch": 87.2325, "grad_norm": 7.25, "learning_rate": 8.428317289412003e-07, "loss": 1.2651, "step": 34893 }, { "epoch": 87.235, "grad_norm": 8.9375, "learning_rate": 8.425064063880251e-07, "loss": 1.4437, "step": 34894 }, { "epoch": 87.2375, "grad_norm": 7.4375, "learning_rate": 8.421811438710114e-07, "loss": 1.3029, "step": 34895 }, { "epoch": 87.24, "grad_norm": 7.71875, "learning_rate": 8.418559413922933e-07, "loss": 1.3224, "step": 34896 }, { "epoch": 87.2425, "grad_norm": 11.3125, "learning_rate": 8.415307989540012e-07, "loss": 1.3791, "step": 34897 }, { "epoch": 87.245, "grad_norm": 6.59375, "learning_rate": 8.412057165582644e-07, "loss": 1.3252, "step": 34898 }, { "epoch": 87.2475, "grad_norm": 4.4375, "learning_rate": 8.408806942072168e-07, "loss": 1.3518, "step": 34899 }, { "epoch": 87.25, "grad_norm": 8.75, "learning_rate": 8.405557319029911e-07, "loss": 1.2967, "step": 34900 }, { "epoch": 87.2525, "grad_norm": 6.75, "learning_rate": 8.402308296477146e-07, "loss": 1.3142, "step": 34901 }, { "epoch": 87.255, "grad_norm": 8.5625, "learning_rate": 8.399059874435167e-07, "loss": 1.2462, "step": 34902 }, { "epoch": 87.2575, "grad_norm": 10.125, "learning_rate": 8.395812052925312e-07, "loss": 1.3451, "step": 34903 }, { "epoch": 87.26, "grad_norm": 10.5625, "learning_rate": 8.392564831968819e-07, "loss": 1.5361, "step": 34904 }, { "epoch": 87.2625, "grad_norm": 7.25, "learning_rate": 8.389318211587028e-07, "loss": 1.4321, "step": 34905 }, { "epoch": 87.265, "grad_norm": 6.71875, "learning_rate": 8.386072191801176e-07, "loss": 1.3295, "step": 34906 }, { "epoch": 87.2675, "grad_norm": 10.5, "learning_rate": 8.38282677263259e-07, "loss": 1.3547, "step": 34907 }, { "epoch": 87.27, "grad_norm": 5.6875, "learning_rate": 8.379581954102534e-07, "loss": 1.326, "step": 34908 }, { "epoch": 87.2725, "grad_norm": 6.8125, "learning_rate": 8.376337736232243e-07, "loss": 1.3712, "step": 34909 }, { "epoch": 87.275, "grad_norm": 10.4375, "learning_rate": 8.373094119043024e-07, "loss": 1.4732, "step": 34910 }, { "epoch": 87.2775, "grad_norm": 10.625, "learning_rate": 8.369851102556148e-07, "loss": 1.372, "step": 34911 }, { "epoch": 87.28, "grad_norm": 8.125, "learning_rate": 8.366608686792854e-07, "loss": 1.3267, "step": 34912 }, { "epoch": 87.2825, "grad_norm": 7.96875, "learning_rate": 8.363366871774415e-07, "loss": 1.5274, "step": 34913 }, { "epoch": 87.285, "grad_norm": 7.1875, "learning_rate": 8.360125657522033e-07, "loss": 1.3623, "step": 34914 }, { "epoch": 87.2875, "grad_norm": 7.65625, "learning_rate": 8.356885044057051e-07, "loss": 1.2542, "step": 34915 }, { "epoch": 87.29, "grad_norm": 7.9375, "learning_rate": 8.353645031400648e-07, "loss": 1.4427, "step": 34916 }, { "epoch": 87.2925, "grad_norm": 15.0, "learning_rate": 8.350405619574064e-07, "loss": 1.3322, "step": 34917 }, { "epoch": 87.295, "grad_norm": 6.9375, "learning_rate": 8.347166808598584e-07, "loss": 1.4334, "step": 34918 }, { "epoch": 87.2975, "grad_norm": 7.6875, "learning_rate": 8.343928598495377e-07, "loss": 1.3428, "step": 34919 }, { "epoch": 87.3, "grad_norm": 9.0, "learning_rate": 8.340690989285727e-07, "loss": 1.3335, "step": 34920 }, { "epoch": 87.3025, "grad_norm": 8.75, "learning_rate": 8.337453980990828e-07, "loss": 1.3944, "step": 34921 }, { "epoch": 87.305, "grad_norm": 7.5, "learning_rate": 8.334217573631931e-07, "loss": 1.2998, "step": 34922 }, { "epoch": 87.3075, "grad_norm": 5.15625, "learning_rate": 8.330981767230228e-07, "loss": 1.2722, "step": 34923 }, { "epoch": 87.31, "grad_norm": 7.375, "learning_rate": 8.327746561806926e-07, "loss": 1.2038, "step": 34924 }, { "epoch": 87.3125, "grad_norm": 6.59375, "learning_rate": 8.324511957383252e-07, "loss": 1.3694, "step": 34925 }, { "epoch": 87.315, "grad_norm": 7.03125, "learning_rate": 8.321277953980422e-07, "loss": 1.2526, "step": 34926 }, { "epoch": 87.3175, "grad_norm": 8.125, "learning_rate": 8.31804455161962e-07, "loss": 1.3325, "step": 34927 }, { "epoch": 87.32, "grad_norm": 6.84375, "learning_rate": 8.31481175032206e-07, "loss": 1.3332, "step": 34928 }, { "epoch": 87.3225, "grad_norm": 5.5625, "learning_rate": 8.311579550108883e-07, "loss": 1.228, "step": 34929 }, { "epoch": 87.325, "grad_norm": 6.1875, "learning_rate": 8.308347951001361e-07, "loss": 1.2364, "step": 34930 }, { "epoch": 87.3275, "grad_norm": 6.40625, "learning_rate": 8.305116953020642e-07, "loss": 1.4047, "step": 34931 }, { "epoch": 87.33, "grad_norm": 9.125, "learning_rate": 8.301886556187899e-07, "loss": 1.1541, "step": 34932 }, { "epoch": 87.3325, "grad_norm": 11.4375, "learning_rate": 8.298656760524304e-07, "loss": 1.3368, "step": 34933 }, { "epoch": 87.335, "grad_norm": 7.03125, "learning_rate": 8.29542756605105e-07, "loss": 1.2634, "step": 34934 }, { "epoch": 87.3375, "grad_norm": 8.4375, "learning_rate": 8.292198972789311e-07, "loss": 1.3913, "step": 34935 }, { "epoch": 87.34, "grad_norm": 8.0, "learning_rate": 8.288970980760258e-07, "loss": 1.439, "step": 34936 }, { "epoch": 87.3425, "grad_norm": 6.40625, "learning_rate": 8.285743589985007e-07, "loss": 1.2876, "step": 34937 }, { "epoch": 87.345, "grad_norm": 8.25, "learning_rate": 8.282516800484786e-07, "loss": 1.3627, "step": 34938 }, { "epoch": 87.3475, "grad_norm": 6.8125, "learning_rate": 8.279290612280688e-07, "loss": 1.2266, "step": 34939 }, { "epoch": 87.35, "grad_norm": 5.71875, "learning_rate": 8.276065025393909e-07, "loss": 1.3324, "step": 34940 }, { "epoch": 87.3525, "grad_norm": 6.5, "learning_rate": 8.272840039845565e-07, "loss": 1.353, "step": 34941 }, { "epoch": 87.355, "grad_norm": 9.0, "learning_rate": 8.269615655656815e-07, "loss": 1.5093, "step": 34942 }, { "epoch": 87.3575, "grad_norm": 8.5, "learning_rate": 8.266391872848801e-07, "loss": 1.2653, "step": 34943 }, { "epoch": 87.36, "grad_norm": 10.8125, "learning_rate": 8.263168691442624e-07, "loss": 1.2995, "step": 34944 }, { "epoch": 87.3625, "grad_norm": 5.34375, "learning_rate": 8.259946111459471e-07, "loss": 1.4146, "step": 34945 }, { "epoch": 87.365, "grad_norm": 8.625, "learning_rate": 8.256724132920435e-07, "loss": 1.3106, "step": 34946 }, { "epoch": 87.3675, "grad_norm": 8.75, "learning_rate": 8.253502755846655e-07, "loss": 1.4395, "step": 34947 }, { "epoch": 87.37, "grad_norm": 7.0, "learning_rate": 8.250281980259212e-07, "loss": 1.1585, "step": 34948 }, { "epoch": 87.3725, "grad_norm": 16.25, "learning_rate": 8.247061806179246e-07, "loss": 1.2442, "step": 34949 }, { "epoch": 87.375, "grad_norm": 8.625, "learning_rate": 8.243842233627897e-07, "loss": 1.4421, "step": 34950 }, { "epoch": 87.3775, "grad_norm": 7.09375, "learning_rate": 8.240623262626246e-07, "loss": 1.3093, "step": 34951 }, { "epoch": 87.38, "grad_norm": 10.375, "learning_rate": 8.237404893195377e-07, "loss": 1.4215, "step": 34952 }, { "epoch": 87.3825, "grad_norm": 9.9375, "learning_rate": 8.23418712535643e-07, "loss": 1.371, "step": 34953 }, { "epoch": 87.385, "grad_norm": 7.78125, "learning_rate": 8.230969959130453e-07, "loss": 1.2212, "step": 34954 }, { "epoch": 87.3875, "grad_norm": 5.1875, "learning_rate": 8.227753394538595e-07, "loss": 1.252, "step": 34955 }, { "epoch": 87.39, "grad_norm": 11.125, "learning_rate": 8.224537431601886e-07, "loss": 1.3505, "step": 34956 }, { "epoch": 87.3925, "grad_norm": 8.5625, "learning_rate": 8.221322070341464e-07, "loss": 1.5117, "step": 34957 }, { "epoch": 87.395, "grad_norm": 8.625, "learning_rate": 8.218107310778367e-07, "loss": 1.3242, "step": 34958 }, { "epoch": 87.3975, "grad_norm": 10.5625, "learning_rate": 8.214893152933678e-07, "loss": 1.3072, "step": 34959 }, { "epoch": 87.4, "grad_norm": 11.75, "learning_rate": 8.211679596828481e-07, "loss": 1.4685, "step": 34960 }, { "epoch": 87.4025, "grad_norm": 9.25, "learning_rate": 8.208466642483848e-07, "loss": 1.4062, "step": 34961 }, { "epoch": 87.405, "grad_norm": 7.0, "learning_rate": 8.205254289920839e-07, "loss": 1.2264, "step": 34962 }, { "epoch": 87.4075, "grad_norm": 7.3125, "learning_rate": 8.202042539160493e-07, "loss": 1.3392, "step": 34963 }, { "epoch": 87.41, "grad_norm": 5.28125, "learning_rate": 8.198831390223871e-07, "loss": 1.3048, "step": 34964 }, { "epoch": 87.4125, "grad_norm": 6.875, "learning_rate": 8.195620843132068e-07, "loss": 1.3019, "step": 34965 }, { "epoch": 87.415, "grad_norm": 4.4375, "learning_rate": 8.1924108979061e-07, "loss": 1.3397, "step": 34966 }, { "epoch": 87.4175, "grad_norm": 8.0625, "learning_rate": 8.189201554566983e-07, "loss": 1.5048, "step": 34967 }, { "epoch": 87.42, "grad_norm": 6.25, "learning_rate": 8.185992813135813e-07, "loss": 1.4246, "step": 34968 }, { "epoch": 87.4225, "grad_norm": 7.6875, "learning_rate": 8.182784673633582e-07, "loss": 1.2497, "step": 34969 }, { "epoch": 87.425, "grad_norm": 8.875, "learning_rate": 8.179577136081363e-07, "loss": 1.3742, "step": 34970 }, { "epoch": 87.4275, "grad_norm": 6.8125, "learning_rate": 8.176370200500139e-07, "loss": 1.2677, "step": 34971 }, { "epoch": 87.43, "grad_norm": 8.875, "learning_rate": 8.173163866910972e-07, "loss": 1.3637, "step": 34972 }, { "epoch": 87.4325, "grad_norm": 6.40625, "learning_rate": 8.169958135334876e-07, "loss": 1.1589, "step": 34973 }, { "epoch": 87.435, "grad_norm": 7.25, "learning_rate": 8.166753005792837e-07, "loss": 1.2973, "step": 34974 }, { "epoch": 87.4375, "grad_norm": 9.5, "learning_rate": 8.163548478305883e-07, "loss": 1.2, "step": 34975 }, { "epoch": 87.44, "grad_norm": 8.375, "learning_rate": 8.160344552895061e-07, "loss": 1.2881, "step": 34976 }, { "epoch": 87.4425, "grad_norm": 7.5, "learning_rate": 8.157141229581333e-07, "loss": 1.4035, "step": 34977 }, { "epoch": 87.445, "grad_norm": 7.15625, "learning_rate": 8.153938508385705e-07, "loss": 1.3435, "step": 34978 }, { "epoch": 87.4475, "grad_norm": 6.5, "learning_rate": 8.150736389329172e-07, "loss": 1.3092, "step": 34979 }, { "epoch": 87.45, "grad_norm": 7.0625, "learning_rate": 8.147534872432761e-07, "loss": 1.4664, "step": 34980 }, { "epoch": 87.4525, "grad_norm": 7.65625, "learning_rate": 8.144333957717432e-07, "loss": 1.2205, "step": 34981 }, { "epoch": 87.455, "grad_norm": 7.4375, "learning_rate": 8.141133645204147e-07, "loss": 1.2141, "step": 34982 }, { "epoch": 87.4575, "grad_norm": 6.9375, "learning_rate": 8.137933934913944e-07, "loss": 1.4577, "step": 34983 }, { "epoch": 87.46, "grad_norm": 6.84375, "learning_rate": 8.134734826867751e-07, "loss": 1.2444, "step": 34984 }, { "epoch": 87.4625, "grad_norm": 6.9375, "learning_rate": 8.131536321086575e-07, "loss": 1.3068, "step": 34985 }, { "epoch": 87.465, "grad_norm": 6.8125, "learning_rate": 8.128338417591353e-07, "loss": 1.3052, "step": 34986 }, { "epoch": 87.4675, "grad_norm": 6.53125, "learning_rate": 8.125141116403079e-07, "loss": 1.4204, "step": 34987 }, { "epoch": 87.47, "grad_norm": 8.5, "learning_rate": 8.121944417542715e-07, "loss": 1.3766, "step": 34988 }, { "epoch": 87.4725, "grad_norm": 7.71875, "learning_rate": 8.118748321031178e-07, "loss": 1.3557, "step": 34989 }, { "epoch": 87.475, "grad_norm": 5.9375, "learning_rate": 8.115552826889461e-07, "loss": 1.3995, "step": 34990 }, { "epoch": 87.4775, "grad_norm": 8.9375, "learning_rate": 8.112357935138515e-07, "loss": 1.3216, "step": 34991 }, { "epoch": 87.48, "grad_norm": 6.9375, "learning_rate": 8.109163645799267e-07, "loss": 1.3712, "step": 34992 }, { "epoch": 87.4825, "grad_norm": 7.375, "learning_rate": 8.105969958892656e-07, "loss": 1.3501, "step": 34993 }, { "epoch": 87.485, "grad_norm": 5.5, "learning_rate": 8.102776874439633e-07, "loss": 1.2288, "step": 34994 }, { "epoch": 87.4875, "grad_norm": 7.625, "learning_rate": 8.099584392461135e-07, "loss": 1.3908, "step": 34995 }, { "epoch": 87.49, "grad_norm": 6.46875, "learning_rate": 8.096392512978091e-07, "loss": 1.4957, "step": 34996 }, { "epoch": 87.4925, "grad_norm": 10.75, "learning_rate": 8.093201236011395e-07, "loss": 1.3111, "step": 34997 }, { "epoch": 87.495, "grad_norm": 6.5625, "learning_rate": 8.090010561582017e-07, "loss": 1.4117, "step": 34998 }, { "epoch": 87.4975, "grad_norm": 7.375, "learning_rate": 8.086820489710833e-07, "loss": 1.4048, "step": 34999 }, { "epoch": 87.5, "grad_norm": 7.78125, "learning_rate": 8.083631020418792e-07, "loss": 1.4232, "step": 35000 }, { "epoch": 87.5025, "grad_norm": 6.125, "learning_rate": 8.080442153726764e-07, "loss": 1.4473, "step": 35001 }, { "epoch": 87.505, "grad_norm": 10.1875, "learning_rate": 8.077253889655701e-07, "loss": 1.4673, "step": 35002 }, { "epoch": 87.5075, "grad_norm": 10.8125, "learning_rate": 8.074066228226484e-07, "loss": 1.2545, "step": 35003 }, { "epoch": 87.51, "grad_norm": 8.0625, "learning_rate": 8.070879169459999e-07, "loss": 1.2951, "step": 35004 }, { "epoch": 87.5125, "grad_norm": 7.28125, "learning_rate": 8.067692713377151e-07, "loss": 1.4397, "step": 35005 }, { "epoch": 87.515, "grad_norm": 8.125, "learning_rate": 8.064506859998833e-07, "loss": 1.2901, "step": 35006 }, { "epoch": 87.5175, "grad_norm": 6.0, "learning_rate": 8.06132160934594e-07, "loss": 1.3477, "step": 35007 }, { "epoch": 87.52, "grad_norm": 7.0, "learning_rate": 8.058136961439333e-07, "loss": 1.47, "step": 35008 }, { "epoch": 87.5225, "grad_norm": 6.625, "learning_rate": 8.054952916299885e-07, "loss": 1.2335, "step": 35009 }, { "epoch": 87.525, "grad_norm": 9.0, "learning_rate": 8.051769473948512e-07, "loss": 1.3799, "step": 35010 }, { "epoch": 87.5275, "grad_norm": 6.40625, "learning_rate": 8.048586634406064e-07, "loss": 1.4062, "step": 35011 }, { "epoch": 87.53, "grad_norm": 7.125, "learning_rate": 8.045404397693379e-07, "loss": 1.5256, "step": 35012 }, { "epoch": 87.5325, "grad_norm": 7.34375, "learning_rate": 8.042222763831353e-07, "loss": 1.4155, "step": 35013 }, { "epoch": 87.535, "grad_norm": 4.8125, "learning_rate": 8.039041732840824e-07, "loss": 1.2643, "step": 35014 }, { "epoch": 87.5375, "grad_norm": 9.6875, "learning_rate": 8.035861304742676e-07, "loss": 1.4265, "step": 35015 }, { "epoch": 87.54, "grad_norm": 7.875, "learning_rate": 8.032681479557725e-07, "loss": 1.2539, "step": 35016 }, { "epoch": 87.5425, "grad_norm": 9.25, "learning_rate": 8.029502257306854e-07, "loss": 1.3546, "step": 35017 }, { "epoch": 87.545, "grad_norm": 7.65625, "learning_rate": 8.02632363801088e-07, "loss": 1.3236, "step": 35018 }, { "epoch": 87.5475, "grad_norm": 7.8125, "learning_rate": 8.023145621690631e-07, "loss": 1.3815, "step": 35019 }, { "epoch": 87.55, "grad_norm": 8.0625, "learning_rate": 8.019968208366958e-07, "loss": 1.2562, "step": 35020 }, { "epoch": 87.5525, "grad_norm": 7.125, "learning_rate": 8.016791398060708e-07, "loss": 1.2105, "step": 35021 }, { "epoch": 87.555, "grad_norm": 9.9375, "learning_rate": 8.0136151907927e-07, "loss": 1.3709, "step": 35022 }, { "epoch": 87.5575, "grad_norm": 6.90625, "learning_rate": 8.010439586583729e-07, "loss": 1.3208, "step": 35023 }, { "epoch": 87.56, "grad_norm": 8.3125, "learning_rate": 8.007264585454632e-07, "loss": 1.3398, "step": 35024 }, { "epoch": 87.5625, "grad_norm": 8.4375, "learning_rate": 8.004090187426238e-07, "loss": 1.3171, "step": 35025 }, { "epoch": 87.565, "grad_norm": 5.96875, "learning_rate": 8.000916392519364e-07, "loss": 1.2444, "step": 35026 }, { "epoch": 87.5675, "grad_norm": 11.75, "learning_rate": 7.997743200754771e-07, "loss": 1.3284, "step": 35027 }, { "epoch": 87.57, "grad_norm": 9.9375, "learning_rate": 7.994570612153319e-07, "loss": 1.2625, "step": 35028 }, { "epoch": 87.5725, "grad_norm": 7.46875, "learning_rate": 7.991398626735747e-07, "loss": 1.2715, "step": 35029 }, { "epoch": 87.575, "grad_norm": 6.0625, "learning_rate": 7.988227244522917e-07, "loss": 1.3659, "step": 35030 }, { "epoch": 87.5775, "grad_norm": 10.0625, "learning_rate": 7.985056465535556e-07, "loss": 1.4911, "step": 35031 }, { "epoch": 87.58, "grad_norm": 6.15625, "learning_rate": 7.981886289794516e-07, "loss": 1.2256, "step": 35032 }, { "epoch": 87.5825, "grad_norm": 25.75, "learning_rate": 7.978716717320511e-07, "loss": 1.332, "step": 35033 }, { "epoch": 87.585, "grad_norm": 7.28125, "learning_rate": 7.975547748134394e-07, "loss": 1.4925, "step": 35034 }, { "epoch": 87.5875, "grad_norm": 5.96875, "learning_rate": 7.972379382256868e-07, "loss": 1.3084, "step": 35035 }, { "epoch": 87.59, "grad_norm": 6.46875, "learning_rate": 7.969211619708773e-07, "loss": 1.1504, "step": 35036 }, { "epoch": 87.5925, "grad_norm": 7.5625, "learning_rate": 7.966044460510836e-07, "loss": 1.4162, "step": 35037 }, { "epoch": 87.595, "grad_norm": 5.6875, "learning_rate": 7.962877904683808e-07, "loss": 1.3824, "step": 35038 }, { "epoch": 87.5975, "grad_norm": 7.75, "learning_rate": 7.959711952248472e-07, "loss": 1.3947, "step": 35039 }, { "epoch": 87.6, "grad_norm": 5.40625, "learning_rate": 7.956546603225601e-07, "loss": 1.3849, "step": 35040 }, { "epoch": 87.6025, "grad_norm": 7.46875, "learning_rate": 7.953381857635922e-07, "loss": 1.3455, "step": 35041 }, { "epoch": 87.605, "grad_norm": 8.6875, "learning_rate": 7.950217715500175e-07, "loss": 1.254, "step": 35042 }, { "epoch": 87.6075, "grad_norm": 5.4375, "learning_rate": 7.947054176839119e-07, "loss": 1.213, "step": 35043 }, { "epoch": 87.61, "grad_norm": 6.71875, "learning_rate": 7.943891241673507e-07, "loss": 1.2295, "step": 35044 }, { "epoch": 87.6125, "grad_norm": 6.46875, "learning_rate": 7.940728910024053e-07, "loss": 1.2998, "step": 35045 }, { "epoch": 87.615, "grad_norm": 7.96875, "learning_rate": 7.937567181911487e-07, "loss": 1.2462, "step": 35046 }, { "epoch": 87.6175, "grad_norm": 6.09375, "learning_rate": 7.934406057356569e-07, "loss": 1.1986, "step": 35047 }, { "epoch": 87.62, "grad_norm": 7.65625, "learning_rate": 7.931245536379973e-07, "loss": 1.3449, "step": 35048 }, { "epoch": 87.6225, "grad_norm": 6.78125, "learning_rate": 7.928085619002468e-07, "loss": 1.3344, "step": 35049 }, { "epoch": 87.625, "grad_norm": 15.6875, "learning_rate": 7.924926305244729e-07, "loss": 1.3769, "step": 35050 }, { "epoch": 87.6275, "grad_norm": 8.875, "learning_rate": 7.921767595127516e-07, "loss": 1.2023, "step": 35051 }, { "epoch": 87.63, "grad_norm": 6.59375, "learning_rate": 7.918609488671503e-07, "loss": 1.3205, "step": 35052 }, { "epoch": 87.6325, "grad_norm": 10.4375, "learning_rate": 7.915451985897382e-07, "loss": 1.3629, "step": 35053 }, { "epoch": 87.635, "grad_norm": 9.25, "learning_rate": 7.912295086825872e-07, "loss": 1.2439, "step": 35054 }, { "epoch": 87.6375, "grad_norm": 6.28125, "learning_rate": 7.909138791477689e-07, "loss": 1.3335, "step": 35055 }, { "epoch": 87.64, "grad_norm": 6.84375, "learning_rate": 7.905983099873504e-07, "loss": 1.2329, "step": 35056 }, { "epoch": 87.6425, "grad_norm": 7.90625, "learning_rate": 7.902828012034003e-07, "loss": 1.4315, "step": 35057 }, { "epoch": 87.645, "grad_norm": 10.0625, "learning_rate": 7.899673527979868e-07, "loss": 1.1693, "step": 35058 }, { "epoch": 87.6475, "grad_norm": 7.90625, "learning_rate": 7.896519647731804e-07, "loss": 1.3431, "step": 35059 }, { "epoch": 87.65, "grad_norm": 8.0625, "learning_rate": 7.893366371310463e-07, "loss": 1.3804, "step": 35060 }, { "epoch": 87.6525, "grad_norm": 8.5, "learning_rate": 7.890213698736526e-07, "loss": 1.2374, "step": 35061 }, { "epoch": 87.655, "grad_norm": 6.40625, "learning_rate": 7.887061630030668e-07, "loss": 1.2897, "step": 35062 }, { "epoch": 87.6575, "grad_norm": 6.96875, "learning_rate": 7.883910165213537e-07, "loss": 1.3673, "step": 35063 }, { "epoch": 87.66, "grad_norm": 9.0625, "learning_rate": 7.880759304305818e-07, "loss": 1.143, "step": 35064 }, { "epoch": 87.6625, "grad_norm": 12.375, "learning_rate": 7.877609047328139e-07, "loss": 1.3253, "step": 35065 }, { "epoch": 87.665, "grad_norm": 7.09375, "learning_rate": 7.874459394301182e-07, "loss": 1.614, "step": 35066 }, { "epoch": 87.6675, "grad_norm": 7.9375, "learning_rate": 7.871310345245586e-07, "loss": 1.3629, "step": 35067 }, { "epoch": 87.67, "grad_norm": 7.4375, "learning_rate": 7.86816190018197e-07, "loss": 1.3788, "step": 35068 }, { "epoch": 87.6725, "grad_norm": 8.8125, "learning_rate": 7.865014059131004e-07, "loss": 1.2293, "step": 35069 }, { "epoch": 87.675, "grad_norm": 11.25, "learning_rate": 7.861866822113339e-07, "loss": 1.2916, "step": 35070 }, { "epoch": 87.6775, "grad_norm": 10.5, "learning_rate": 7.858720189149582e-07, "loss": 1.3125, "step": 35071 }, { "epoch": 87.68, "grad_norm": 8.3125, "learning_rate": 7.855574160260371e-07, "loss": 1.3241, "step": 35072 }, { "epoch": 87.6825, "grad_norm": 6.34375, "learning_rate": 7.852428735466289e-07, "loss": 1.0924, "step": 35073 }, { "epoch": 87.685, "grad_norm": 8.625, "learning_rate": 7.84928391478803e-07, "loss": 1.3566, "step": 35074 }, { "epoch": 87.6875, "grad_norm": 9.3125, "learning_rate": 7.846139698246191e-07, "loss": 1.2691, "step": 35075 }, { "epoch": 87.69, "grad_norm": 5.96875, "learning_rate": 7.842996085861365e-07, "loss": 1.2478, "step": 35076 }, { "epoch": 87.6925, "grad_norm": 7.53125, "learning_rate": 7.839853077654147e-07, "loss": 1.355, "step": 35077 }, { "epoch": 87.695, "grad_norm": 7.5, "learning_rate": 7.836710673645165e-07, "loss": 1.3504, "step": 35078 }, { "epoch": 87.6975, "grad_norm": 9.375, "learning_rate": 7.833568873855047e-07, "loss": 1.1134, "step": 35079 }, { "epoch": 87.7, "grad_norm": 6.8125, "learning_rate": 7.830427678304353e-07, "loss": 1.2465, "step": 35080 }, { "epoch": 87.7025, "grad_norm": 8.5, "learning_rate": 7.827287087013669e-07, "loss": 1.2554, "step": 35081 }, { "epoch": 87.705, "grad_norm": 6.4375, "learning_rate": 7.82414710000362e-07, "loss": 1.3344, "step": 35082 }, { "epoch": 87.7075, "grad_norm": 7.75, "learning_rate": 7.821007717294759e-07, "loss": 1.1934, "step": 35083 }, { "epoch": 87.71, "grad_norm": 7.25, "learning_rate": 7.817868938907691e-07, "loss": 1.3004, "step": 35084 }, { "epoch": 87.7125, "grad_norm": 7.75, "learning_rate": 7.814730764862988e-07, "loss": 1.169, "step": 35085 }, { "epoch": 87.715, "grad_norm": 7.5, "learning_rate": 7.811593195181233e-07, "loss": 1.1944, "step": 35086 }, { "epoch": 87.7175, "grad_norm": 9.4375, "learning_rate": 7.808456229882977e-07, "loss": 1.3579, "step": 35087 }, { "epoch": 87.72, "grad_norm": 10.0, "learning_rate": 7.805319868988759e-07, "loss": 1.2369, "step": 35088 }, { "epoch": 87.7225, "grad_norm": 7.71875, "learning_rate": 7.802184112519207e-07, "loss": 1.4208, "step": 35089 }, { "epoch": 87.725, "grad_norm": 9.75, "learning_rate": 7.799048960494848e-07, "loss": 1.3524, "step": 35090 }, { "epoch": 87.7275, "grad_norm": 8.0625, "learning_rate": 7.795914412936234e-07, "loss": 1.3613, "step": 35091 }, { "epoch": 87.73, "grad_norm": 15.625, "learning_rate": 7.792780469863903e-07, "loss": 1.2511, "step": 35092 }, { "epoch": 87.7325, "grad_norm": 7.0625, "learning_rate": 7.789647131298417e-07, "loss": 1.2795, "step": 35093 }, { "epoch": 87.735, "grad_norm": 7.6875, "learning_rate": 7.786514397260314e-07, "loss": 1.3729, "step": 35094 }, { "epoch": 87.7375, "grad_norm": 9.25, "learning_rate": 7.783382267770145e-07, "loss": 1.2825, "step": 35095 }, { "epoch": 87.74, "grad_norm": 6.71875, "learning_rate": 7.780250742848416e-07, "loss": 1.1613, "step": 35096 }, { "epoch": 87.7425, "grad_norm": 8.5625, "learning_rate": 7.777119822515688e-07, "loss": 1.4304, "step": 35097 }, { "epoch": 87.745, "grad_norm": 7.1875, "learning_rate": 7.773989506792446e-07, "loss": 1.4319, "step": 35098 }, { "epoch": 87.7475, "grad_norm": 6.0625, "learning_rate": 7.770859795699271e-07, "loss": 1.2863, "step": 35099 }, { "epoch": 87.75, "grad_norm": 9.8125, "learning_rate": 7.767730689256614e-07, "loss": 1.2945, "step": 35100 }, { "epoch": 87.7525, "grad_norm": 8.75, "learning_rate": 7.764602187485048e-07, "loss": 1.2906, "step": 35101 }, { "epoch": 87.755, "grad_norm": 8.4375, "learning_rate": 7.761474290405069e-07, "loss": 1.4811, "step": 35102 }, { "epoch": 87.7575, "grad_norm": 6.625, "learning_rate": 7.758346998037148e-07, "loss": 1.2809, "step": 35103 }, { "epoch": 87.76, "grad_norm": 8.5625, "learning_rate": 7.755220310401812e-07, "loss": 1.6005, "step": 35104 }, { "epoch": 87.7625, "grad_norm": 8.6875, "learning_rate": 7.752094227519569e-07, "loss": 1.2078, "step": 35105 }, { "epoch": 87.765, "grad_norm": 5.53125, "learning_rate": 7.748968749410913e-07, "loss": 1.3416, "step": 35106 }, { "epoch": 87.7675, "grad_norm": 7.75, "learning_rate": 7.745843876096315e-07, "loss": 1.316, "step": 35107 }, { "epoch": 87.77, "grad_norm": 6.59375, "learning_rate": 7.742719607596261e-07, "loss": 1.5163, "step": 35108 }, { "epoch": 87.7725, "grad_norm": 8.875, "learning_rate": 7.739595943931278e-07, "loss": 1.2858, "step": 35109 }, { "epoch": 87.775, "grad_norm": 10.625, "learning_rate": 7.736472885121793e-07, "loss": 1.3383, "step": 35110 }, { "epoch": 87.7775, "grad_norm": 6.75, "learning_rate": 7.733350431188291e-07, "loss": 1.2377, "step": 35111 }, { "epoch": 87.78, "grad_norm": 9.25, "learning_rate": 7.730228582151278e-07, "loss": 1.3364, "step": 35112 }, { "epoch": 87.7825, "grad_norm": 10.6875, "learning_rate": 7.72710733803117e-07, "loss": 1.2877, "step": 35113 }, { "epoch": 87.785, "grad_norm": 8.375, "learning_rate": 7.723986698848462e-07, "loss": 1.2874, "step": 35114 }, { "epoch": 87.7875, "grad_norm": 7.25, "learning_rate": 7.720866664623594e-07, "loss": 1.3733, "step": 35115 }, { "epoch": 87.79, "grad_norm": 6.0625, "learning_rate": 7.717747235377048e-07, "loss": 1.3534, "step": 35116 }, { "epoch": 87.7925, "grad_norm": 9.875, "learning_rate": 7.714628411129266e-07, "loss": 1.2334, "step": 35117 }, { "epoch": 87.795, "grad_norm": 9.0, "learning_rate": 7.711510191900662e-07, "loss": 1.2568, "step": 35118 }, { "epoch": 87.7975, "grad_norm": 7.59375, "learning_rate": 7.708392577711698e-07, "loss": 1.2571, "step": 35119 }, { "epoch": 87.8, "grad_norm": 7.65625, "learning_rate": 7.705275568582848e-07, "loss": 1.3206, "step": 35120 }, { "epoch": 87.8025, "grad_norm": 10.125, "learning_rate": 7.702159164534518e-07, "loss": 1.2278, "step": 35121 }, { "epoch": 87.805, "grad_norm": 8.375, "learning_rate": 7.699043365587111e-07, "loss": 1.5078, "step": 35122 }, { "epoch": 87.8075, "grad_norm": 5.78125, "learning_rate": 7.695928171761092e-07, "loss": 1.3142, "step": 35123 }, { "epoch": 87.81, "grad_norm": 5.78125, "learning_rate": 7.692813583076885e-07, "loss": 1.2526, "step": 35124 }, { "epoch": 87.8125, "grad_norm": 6.4375, "learning_rate": 7.689699599554901e-07, "loss": 1.2674, "step": 35125 }, { "epoch": 87.815, "grad_norm": 6.3125, "learning_rate": 7.68658622121553e-07, "loss": 1.2959, "step": 35126 }, { "epoch": 87.8175, "grad_norm": 7.625, "learning_rate": 7.683473448079226e-07, "loss": 1.274, "step": 35127 }, { "epoch": 87.82, "grad_norm": 7.09375, "learning_rate": 7.680361280166359e-07, "loss": 1.2522, "step": 35128 }, { "epoch": 87.8225, "grad_norm": 7.0, "learning_rate": 7.677249717497371e-07, "loss": 1.234, "step": 35129 }, { "epoch": 87.825, "grad_norm": 6.4375, "learning_rate": 7.67413876009262e-07, "loss": 1.3511, "step": 35130 }, { "epoch": 87.8275, "grad_norm": 7.875, "learning_rate": 7.671028407972536e-07, "loss": 1.2622, "step": 35131 }, { "epoch": 87.83, "grad_norm": 6.28125, "learning_rate": 7.667918661157492e-07, "loss": 1.168, "step": 35132 }, { "epoch": 87.8325, "grad_norm": 8.9375, "learning_rate": 7.66480951966787e-07, "loss": 1.3186, "step": 35133 }, { "epoch": 87.835, "grad_norm": 9.9375, "learning_rate": 7.661700983524056e-07, "loss": 1.2998, "step": 35134 }, { "epoch": 87.8375, "grad_norm": 6.75, "learning_rate": 7.658593052746455e-07, "loss": 1.4881, "step": 35135 }, { "epoch": 87.84, "grad_norm": 6.8125, "learning_rate": 7.655485727355416e-07, "loss": 1.3875, "step": 35136 }, { "epoch": 87.8425, "grad_norm": 6.875, "learning_rate": 7.652379007371303e-07, "loss": 1.3377, "step": 35137 }, { "epoch": 87.845, "grad_norm": 6.21875, "learning_rate": 7.649272892814496e-07, "loss": 1.4348, "step": 35138 }, { "epoch": 87.8475, "grad_norm": 9.4375, "learning_rate": 7.646167383705383e-07, "loss": 1.2851, "step": 35139 }, { "epoch": 87.85, "grad_norm": 6.03125, "learning_rate": 7.643062480064301e-07, "loss": 1.3884, "step": 35140 }, { "epoch": 87.8525, "grad_norm": 8.875, "learning_rate": 7.639958181911577e-07, "loss": 1.3712, "step": 35141 }, { "epoch": 87.855, "grad_norm": 6.625, "learning_rate": 7.63685448926762e-07, "loss": 1.3794, "step": 35142 }, { "epoch": 87.8575, "grad_norm": 7.5625, "learning_rate": 7.633751402152734e-07, "loss": 1.2449, "step": 35143 }, { "epoch": 87.86, "grad_norm": 10.0625, "learning_rate": 7.63064892058728e-07, "loss": 1.4691, "step": 35144 }, { "epoch": 87.8625, "grad_norm": 7.53125, "learning_rate": 7.627547044591588e-07, "loss": 1.3393, "step": 35145 }, { "epoch": 87.865, "grad_norm": 8.0625, "learning_rate": 7.624445774186018e-07, "loss": 1.266, "step": 35146 }, { "epoch": 87.8675, "grad_norm": 7.6875, "learning_rate": 7.621345109390887e-07, "loss": 1.3072, "step": 35147 }, { "epoch": 87.87, "grad_norm": 8.75, "learning_rate": 7.618245050226503e-07, "loss": 1.3072, "step": 35148 }, { "epoch": 87.8725, "grad_norm": 6.40625, "learning_rate": 7.615145596713202e-07, "loss": 1.2831, "step": 35149 }, { "epoch": 87.875, "grad_norm": 6.21875, "learning_rate": 7.612046748871327e-07, "loss": 1.295, "step": 35150 }, { "epoch": 87.8775, "grad_norm": 7.84375, "learning_rate": 7.608948506721182e-07, "loss": 1.2249, "step": 35151 }, { "epoch": 87.88, "grad_norm": 6.6875, "learning_rate": 7.60585087028305e-07, "loss": 1.3358, "step": 35152 }, { "epoch": 87.8825, "grad_norm": 6.6875, "learning_rate": 7.60275383957727e-07, "loss": 1.2639, "step": 35153 }, { "epoch": 87.885, "grad_norm": 11.0, "learning_rate": 7.59965741462415e-07, "loss": 1.418, "step": 35154 }, { "epoch": 87.8875, "grad_norm": 9.625, "learning_rate": 7.596561595443974e-07, "loss": 1.2924, "step": 35155 }, { "epoch": 87.89, "grad_norm": 6.21875, "learning_rate": 7.593466382057035e-07, "loss": 1.2677, "step": 35156 }, { "epoch": 87.8925, "grad_norm": 9.125, "learning_rate": 7.59037177448364e-07, "loss": 1.3681, "step": 35157 }, { "epoch": 87.895, "grad_norm": 8.1875, "learning_rate": 7.587277772744062e-07, "loss": 1.2465, "step": 35158 }, { "epoch": 87.8975, "grad_norm": 5.46875, "learning_rate": 7.584184376858606e-07, "loss": 1.3127, "step": 35159 }, { "epoch": 87.9, "grad_norm": 8.25, "learning_rate": 7.581091586847522e-07, "loss": 1.3747, "step": 35160 }, { "epoch": 87.9025, "grad_norm": 8.6875, "learning_rate": 7.577999402731118e-07, "loss": 1.3866, "step": 35161 }, { "epoch": 87.905, "grad_norm": 6.0, "learning_rate": 7.574907824529653e-07, "loss": 1.3035, "step": 35162 }, { "epoch": 87.9075, "grad_norm": 5.8125, "learning_rate": 7.571816852263381e-07, "loss": 1.3867, "step": 35163 }, { "epoch": 87.91, "grad_norm": 5.53125, "learning_rate": 7.568726485952571e-07, "loss": 1.2827, "step": 35164 }, { "epoch": 87.9125, "grad_norm": 7.5, "learning_rate": 7.565636725617509e-07, "loss": 1.332, "step": 35165 }, { "epoch": 87.915, "grad_norm": 14.75, "learning_rate": 7.562547571278434e-07, "loss": 1.229, "step": 35166 }, { "epoch": 87.9175, "grad_norm": 7.5625, "learning_rate": 7.559459022955584e-07, "loss": 1.3031, "step": 35167 }, { "epoch": 87.92, "grad_norm": 8.0625, "learning_rate": 7.556371080669222e-07, "loss": 1.4953, "step": 35168 }, { "epoch": 87.9225, "grad_norm": 6.0, "learning_rate": 7.55328374443961e-07, "loss": 1.1593, "step": 35169 }, { "epoch": 87.925, "grad_norm": 9.4375, "learning_rate": 7.550197014286964e-07, "loss": 1.3935, "step": 35170 }, { "epoch": 87.9275, "grad_norm": 5.84375, "learning_rate": 7.547110890231513e-07, "loss": 1.2403, "step": 35171 }, { "epoch": 87.93, "grad_norm": 7.375, "learning_rate": 7.544025372293528e-07, "loss": 1.3122, "step": 35172 }, { "epoch": 87.9325, "grad_norm": 7.78125, "learning_rate": 7.540940460493185e-07, "loss": 1.223, "step": 35173 }, { "epoch": 87.935, "grad_norm": 8.875, "learning_rate": 7.537856154850753e-07, "loss": 1.3346, "step": 35174 }, { "epoch": 87.9375, "grad_norm": 7.5625, "learning_rate": 7.53477245538643e-07, "loss": 1.3034, "step": 35175 }, { "epoch": 87.94, "grad_norm": 7.6875, "learning_rate": 7.531689362120442e-07, "loss": 1.3412, "step": 35176 }, { "epoch": 87.9425, "grad_norm": 6.9375, "learning_rate": 7.528606875073008e-07, "loss": 1.268, "step": 35177 }, { "epoch": 87.945, "grad_norm": 6.75, "learning_rate": 7.525524994264311e-07, "loss": 1.2751, "step": 35178 }, { "epoch": 87.9475, "grad_norm": 9.25, "learning_rate": 7.522443719714567e-07, "loss": 1.3651, "step": 35179 }, { "epoch": 87.95, "grad_norm": 8.875, "learning_rate": 7.519363051443996e-07, "loss": 1.3794, "step": 35180 }, { "epoch": 87.9525, "grad_norm": 12.6875, "learning_rate": 7.516282989472778e-07, "loss": 1.5146, "step": 35181 }, { "epoch": 87.955, "grad_norm": 7.75, "learning_rate": 7.5132035338211e-07, "loss": 1.303, "step": 35182 }, { "epoch": 87.9575, "grad_norm": 10.4375, "learning_rate": 7.510124684509146e-07, "loss": 1.3564, "step": 35183 }, { "epoch": 87.96, "grad_norm": 6.21875, "learning_rate": 7.507046441557142e-07, "loss": 1.3423, "step": 35184 }, { "epoch": 87.9625, "grad_norm": 5.59375, "learning_rate": 7.50396880498524e-07, "loss": 1.3232, "step": 35185 }, { "epoch": 87.965, "grad_norm": 9.0625, "learning_rate": 7.50089177481359e-07, "loss": 1.3153, "step": 35186 }, { "epoch": 87.9675, "grad_norm": 6.21875, "learning_rate": 7.497815351062421e-07, "loss": 1.4494, "step": 35187 }, { "epoch": 87.97, "grad_norm": 7.125, "learning_rate": 7.494739533751849e-07, "loss": 1.1669, "step": 35188 }, { "epoch": 87.9725, "grad_norm": 8.5625, "learning_rate": 7.491664322902082e-07, "loss": 1.3926, "step": 35189 }, { "epoch": 87.975, "grad_norm": 7.375, "learning_rate": 7.488589718533234e-07, "loss": 1.2986, "step": 35190 }, { "epoch": 87.9775, "grad_norm": 9.0, "learning_rate": 7.485515720665527e-07, "loss": 1.3956, "step": 35191 }, { "epoch": 87.98, "grad_norm": 7.0625, "learning_rate": 7.482442329319062e-07, "loss": 1.2297, "step": 35192 }, { "epoch": 87.9825, "grad_norm": 50.25, "learning_rate": 7.479369544513981e-07, "loss": 1.4586, "step": 35193 }, { "epoch": 87.985, "grad_norm": 6.375, "learning_rate": 7.476297366270469e-07, "loss": 1.2213, "step": 35194 }, { "epoch": 87.9875, "grad_norm": 6.90625, "learning_rate": 7.473225794608652e-07, "loss": 1.3505, "step": 35195 }, { "epoch": 87.99, "grad_norm": 5.0625, "learning_rate": 7.470154829548671e-07, "loss": 1.1256, "step": 35196 }, { "epoch": 87.9925, "grad_norm": 6.21875, "learning_rate": 7.467084471110631e-07, "loss": 1.2869, "step": 35197 }, { "epoch": 87.995, "grad_norm": 6.0625, "learning_rate": 7.464014719314683e-07, "loss": 1.3833, "step": 35198 }, { "epoch": 87.9975, "grad_norm": 6.96875, "learning_rate": 7.460945574180966e-07, "loss": 1.2955, "step": 35199 }, { "epoch": 88.0, "grad_norm": 5.8125, "learning_rate": 7.457877035729588e-07, "loss": 1.3956, "step": 35200 }, { "epoch": 88.0025, "grad_norm": 7.125, "learning_rate": 7.454809103980654e-07, "loss": 1.111, "step": 35201 }, { "epoch": 88.005, "grad_norm": 8.625, "learning_rate": 7.451741778954302e-07, "loss": 1.4071, "step": 35202 }, { "epoch": 88.0075, "grad_norm": 10.375, "learning_rate": 7.448675060670607e-07, "loss": 1.418, "step": 35203 }, { "epoch": 88.01, "grad_norm": 6.78125, "learning_rate": 7.445608949149708e-07, "loss": 1.0908, "step": 35204 }, { "epoch": 88.0125, "grad_norm": 9.4375, "learning_rate": 7.442543444411687e-07, "loss": 1.3, "step": 35205 }, { "epoch": 88.015, "grad_norm": 6.84375, "learning_rate": 7.439478546476653e-07, "loss": 1.2359, "step": 35206 }, { "epoch": 88.0175, "grad_norm": 8.3125, "learning_rate": 7.436414255364699e-07, "loss": 1.3184, "step": 35207 }, { "epoch": 88.02, "grad_norm": 8.75, "learning_rate": 7.433350571095898e-07, "loss": 1.3664, "step": 35208 }, { "epoch": 88.0225, "grad_norm": 6.8125, "learning_rate": 7.430287493690335e-07, "loss": 1.3388, "step": 35209 }, { "epoch": 88.025, "grad_norm": 7.875, "learning_rate": 7.427225023168128e-07, "loss": 1.264, "step": 35210 }, { "epoch": 88.0275, "grad_norm": 6.6875, "learning_rate": 7.424163159549324e-07, "loss": 1.2594, "step": 35211 }, { "epoch": 88.03, "grad_norm": 6.90625, "learning_rate": 7.421101902854e-07, "loss": 1.3643, "step": 35212 }, { "epoch": 88.0325, "grad_norm": 9.625, "learning_rate": 7.418041253102204e-07, "loss": 1.442, "step": 35213 }, { "epoch": 88.035, "grad_norm": 7.28125, "learning_rate": 7.414981210314054e-07, "loss": 1.3761, "step": 35214 }, { "epoch": 88.0375, "grad_norm": 5.03125, "learning_rate": 7.411921774509578e-07, "loss": 1.2579, "step": 35215 }, { "epoch": 88.04, "grad_norm": 5.34375, "learning_rate": 7.408862945708839e-07, "loss": 1.3606, "step": 35216 }, { "epoch": 88.0425, "grad_norm": 10.375, "learning_rate": 7.405804723931876e-07, "loss": 1.408, "step": 35217 }, { "epoch": 88.045, "grad_norm": 6.09375, "learning_rate": 7.402747109198738e-07, "loss": 1.2474, "step": 35218 }, { "epoch": 88.0475, "grad_norm": 8.3125, "learning_rate": 7.399690101529522e-07, "loss": 1.1997, "step": 35219 }, { "epoch": 88.05, "grad_norm": 10.875, "learning_rate": 7.3966337009442e-07, "loss": 1.2948, "step": 35220 }, { "epoch": 88.0525, "grad_norm": 7.25, "learning_rate": 7.393577907462868e-07, "loss": 1.2534, "step": 35221 }, { "epoch": 88.055, "grad_norm": 8.3125, "learning_rate": 7.39052272110552e-07, "loss": 1.2215, "step": 35222 }, { "epoch": 88.0575, "grad_norm": 7.6875, "learning_rate": 7.387468141892195e-07, "loss": 1.2533, "step": 35223 }, { "epoch": 88.06, "grad_norm": 12.0625, "learning_rate": 7.384414169842924e-07, "loss": 1.3343, "step": 35224 }, { "epoch": 88.0625, "grad_norm": 6.59375, "learning_rate": 7.381360804977733e-07, "loss": 1.3966, "step": 35225 }, { "epoch": 88.065, "grad_norm": 5.84375, "learning_rate": 7.37830804731664e-07, "loss": 1.2268, "step": 35226 }, { "epoch": 88.0675, "grad_norm": 7.28125, "learning_rate": 7.375255896879651e-07, "loss": 1.2507, "step": 35227 }, { "epoch": 88.07, "grad_norm": 9.4375, "learning_rate": 7.372204353686751e-07, "loss": 1.5511, "step": 35228 }, { "epoch": 88.0725, "grad_norm": 7.28125, "learning_rate": 7.369153417758002e-07, "loss": 1.2623, "step": 35229 }, { "epoch": 88.075, "grad_norm": 9.0625, "learning_rate": 7.366103089113374e-07, "loss": 1.2143, "step": 35230 }, { "epoch": 88.0775, "grad_norm": 7.59375, "learning_rate": 7.363053367772866e-07, "loss": 1.3463, "step": 35231 }, { "epoch": 88.08, "grad_norm": 11.5625, "learning_rate": 7.360004253756459e-07, "loss": 1.3821, "step": 35232 }, { "epoch": 88.0825, "grad_norm": 6.09375, "learning_rate": 7.35695574708416e-07, "loss": 1.3418, "step": 35233 }, { "epoch": 88.085, "grad_norm": 7.28125, "learning_rate": 7.353907847775965e-07, "loss": 1.2366, "step": 35234 }, { "epoch": 88.0875, "grad_norm": 6.15625, "learning_rate": 7.350860555851847e-07, "loss": 1.4498, "step": 35235 }, { "epoch": 88.09, "grad_norm": 6.875, "learning_rate": 7.347813871331755e-07, "loss": 1.2518, "step": 35236 }, { "epoch": 88.0925, "grad_norm": 6.09375, "learning_rate": 7.344767794235707e-07, "loss": 1.2869, "step": 35237 }, { "epoch": 88.095, "grad_norm": 7.15625, "learning_rate": 7.341722324583644e-07, "loss": 1.1754, "step": 35238 }, { "epoch": 88.0975, "grad_norm": 6.1875, "learning_rate": 7.33867746239555e-07, "loss": 1.183, "step": 35239 }, { "epoch": 88.1, "grad_norm": 9.0625, "learning_rate": 7.335633207691362e-07, "loss": 1.3792, "step": 35240 }, { "epoch": 88.1025, "grad_norm": 7.8125, "learning_rate": 7.332589560491077e-07, "loss": 1.3585, "step": 35241 }, { "epoch": 88.105, "grad_norm": 5.84375, "learning_rate": 7.329546520814623e-07, "loss": 1.2065, "step": 35242 }, { "epoch": 88.1075, "grad_norm": 6.1875, "learning_rate": 7.32650408868193e-07, "loss": 1.4504, "step": 35243 }, { "epoch": 88.11, "grad_norm": 8.25, "learning_rate": 7.32346226411298e-07, "loss": 1.536, "step": 35244 }, { "epoch": 88.1125, "grad_norm": 15.25, "learning_rate": 7.320421047127712e-07, "loss": 1.4336, "step": 35245 }, { "epoch": 88.115, "grad_norm": 7.03125, "learning_rate": 7.317380437746058e-07, "loss": 1.2428, "step": 35246 }, { "epoch": 88.1175, "grad_norm": 8.8125, "learning_rate": 7.314340435987921e-07, "loss": 1.3014, "step": 35247 }, { "epoch": 88.12, "grad_norm": 6.09375, "learning_rate": 7.311301041873276e-07, "loss": 1.4098, "step": 35248 }, { "epoch": 88.1225, "grad_norm": 8.25, "learning_rate": 7.308262255422038e-07, "loss": 1.3565, "step": 35249 }, { "epoch": 88.125, "grad_norm": 4.59375, "learning_rate": 7.305224076654127e-07, "loss": 1.3417, "step": 35250 }, { "epoch": 88.1275, "grad_norm": 7.84375, "learning_rate": 7.302186505589437e-07, "loss": 1.2296, "step": 35251 }, { "epoch": 88.13, "grad_norm": 6.53125, "learning_rate": 7.299149542247929e-07, "loss": 1.2553, "step": 35252 }, { "epoch": 88.1325, "grad_norm": 5.875, "learning_rate": 7.296113186649456e-07, "loss": 1.354, "step": 35253 }, { "epoch": 88.135, "grad_norm": 9.0625, "learning_rate": 7.29307743881399e-07, "loss": 1.5909, "step": 35254 }, { "epoch": 88.1375, "grad_norm": 7.96875, "learning_rate": 7.290042298761368e-07, "loss": 1.1783, "step": 35255 }, { "epoch": 88.14, "grad_norm": 7.46875, "learning_rate": 7.287007766511534e-07, "loss": 1.3772, "step": 35256 }, { "epoch": 88.1425, "grad_norm": 7.625, "learning_rate": 7.28397384208438e-07, "loss": 1.331, "step": 35257 }, { "epoch": 88.145, "grad_norm": 12.0625, "learning_rate": 7.280940525499758e-07, "loss": 1.2085, "step": 35258 }, { "epoch": 88.1475, "grad_norm": 6.125, "learning_rate": 7.277907816777574e-07, "loss": 1.1807, "step": 35259 }, { "epoch": 88.15, "grad_norm": 8.25, "learning_rate": 7.274875715937746e-07, "loss": 1.4174, "step": 35260 }, { "epoch": 88.1525, "grad_norm": 7.09375, "learning_rate": 7.271844223000113e-07, "loss": 1.4159, "step": 35261 }, { "epoch": 88.155, "grad_norm": 7.8125, "learning_rate": 7.268813337984537e-07, "loss": 1.3457, "step": 35262 }, { "epoch": 88.1575, "grad_norm": 10.3125, "learning_rate": 7.265783060910913e-07, "loss": 1.3426, "step": 35263 }, { "epoch": 88.16, "grad_norm": 8.5, "learning_rate": 7.262753391799127e-07, "loss": 1.266, "step": 35264 }, { "epoch": 88.1625, "grad_norm": 6.03125, "learning_rate": 7.259724330669016e-07, "loss": 1.367, "step": 35265 }, { "epoch": 88.165, "grad_norm": 6.25, "learning_rate": 7.256695877540431e-07, "loss": 1.4997, "step": 35266 }, { "epoch": 88.1675, "grad_norm": 12.125, "learning_rate": 7.253668032433248e-07, "loss": 1.3299, "step": 35267 }, { "epoch": 88.17, "grad_norm": 7.78125, "learning_rate": 7.250640795367281e-07, "loss": 1.2581, "step": 35268 }, { "epoch": 88.1725, "grad_norm": 6.875, "learning_rate": 7.247614166362427e-07, "loss": 1.3262, "step": 35269 }, { "epoch": 88.175, "grad_norm": 5.65625, "learning_rate": 7.24458814543848e-07, "loss": 1.3414, "step": 35270 }, { "epoch": 88.1775, "grad_norm": 7.34375, "learning_rate": 7.241562732615326e-07, "loss": 1.2232, "step": 35271 }, { "epoch": 88.18, "grad_norm": 5.40625, "learning_rate": 7.238537927912747e-07, "loss": 1.3168, "step": 35272 }, { "epoch": 88.1825, "grad_norm": 5.1875, "learning_rate": 7.235513731350629e-07, "loss": 1.3575, "step": 35273 }, { "epoch": 88.185, "grad_norm": 8.0, "learning_rate": 7.232490142948744e-07, "loss": 1.2187, "step": 35274 }, { "epoch": 88.1875, "grad_norm": 5.4375, "learning_rate": 7.229467162726966e-07, "loss": 1.3569, "step": 35275 }, { "epoch": 88.19, "grad_norm": 6.4375, "learning_rate": 7.22644479070509e-07, "loss": 1.3986, "step": 35276 }, { "epoch": 88.1925, "grad_norm": 7.40625, "learning_rate": 7.22342302690291e-07, "loss": 1.2166, "step": 35277 }, { "epoch": 88.195, "grad_norm": 6.5625, "learning_rate": 7.220401871340255e-07, "loss": 1.2405, "step": 35278 }, { "epoch": 88.1975, "grad_norm": 13.375, "learning_rate": 7.217381324036943e-07, "loss": 1.3773, "step": 35279 }, { "epoch": 88.2, "grad_norm": 11.0625, "learning_rate": 7.21436138501278e-07, "loss": 1.1594, "step": 35280 }, { "epoch": 88.2025, "grad_norm": 6.46875, "learning_rate": 7.211342054287518e-07, "loss": 1.2892, "step": 35281 }, { "epoch": 88.205, "grad_norm": 7.96875, "learning_rate": 7.208323331880996e-07, "loss": 1.2967, "step": 35282 }, { "epoch": 88.2075, "grad_norm": 7.15625, "learning_rate": 7.205305217813007e-07, "loss": 1.2622, "step": 35283 }, { "epoch": 88.21, "grad_norm": 9.9375, "learning_rate": 7.202287712103329e-07, "loss": 1.2589, "step": 35284 }, { "epoch": 88.2125, "grad_norm": 7.625, "learning_rate": 7.199270814771708e-07, "loss": 1.3242, "step": 35285 }, { "epoch": 88.215, "grad_norm": 8.4375, "learning_rate": 7.196254525837987e-07, "loss": 1.3395, "step": 35286 }, { "epoch": 88.2175, "grad_norm": 9.875, "learning_rate": 7.193238845321881e-07, "loss": 1.1662, "step": 35287 }, { "epoch": 88.22, "grad_norm": 6.0, "learning_rate": 7.190223773243199e-07, "loss": 1.1777, "step": 35288 }, { "epoch": 88.2225, "grad_norm": 8.5, "learning_rate": 7.18720930962169e-07, "loss": 1.3854, "step": 35289 }, { "epoch": 88.225, "grad_norm": 8.5625, "learning_rate": 7.184195454477138e-07, "loss": 1.3459, "step": 35290 }, { "epoch": 88.2275, "grad_norm": 6.03125, "learning_rate": 7.181182207829285e-07, "loss": 1.2128, "step": 35291 }, { "epoch": 88.23, "grad_norm": 11.1875, "learning_rate": 7.178169569697868e-07, "loss": 1.306, "step": 35292 }, { "epoch": 88.2325, "grad_norm": 6.78125, "learning_rate": 7.175157540102651e-07, "loss": 1.3458, "step": 35293 }, { "epoch": 88.235, "grad_norm": 7.90625, "learning_rate": 7.172146119063417e-07, "loss": 1.3895, "step": 35294 }, { "epoch": 88.2375, "grad_norm": 9.125, "learning_rate": 7.169135306599862e-07, "loss": 1.3137, "step": 35295 }, { "epoch": 88.24, "grad_norm": 12.0, "learning_rate": 7.166125102731735e-07, "loss": 1.2817, "step": 35296 }, { "epoch": 88.2425, "grad_norm": 8.9375, "learning_rate": 7.16311550747878e-07, "loss": 1.3288, "step": 35297 }, { "epoch": 88.245, "grad_norm": 7.75, "learning_rate": 7.160106520860733e-07, "loss": 1.4022, "step": 35298 }, { "epoch": 88.2475, "grad_norm": 5.28125, "learning_rate": 7.157098142897312e-07, "loss": 1.2278, "step": 35299 }, { "epoch": 88.25, "grad_norm": 14.6875, "learning_rate": 7.154090373608236e-07, "loss": 1.1847, "step": 35300 }, { "epoch": 88.2525, "grad_norm": 7.09375, "learning_rate": 7.151083213013232e-07, "loss": 1.4302, "step": 35301 }, { "epoch": 88.255, "grad_norm": 6.65625, "learning_rate": 7.148076661131997e-07, "loss": 1.3231, "step": 35302 }, { "epoch": 88.2575, "grad_norm": 5.03125, "learning_rate": 7.14507071798427e-07, "loss": 1.3496, "step": 35303 }, { "epoch": 88.26, "grad_norm": 11.125, "learning_rate": 7.142065383589736e-07, "loss": 1.2728, "step": 35304 }, { "epoch": 88.2625, "grad_norm": 8.1875, "learning_rate": 7.139060657968122e-07, "loss": 1.3598, "step": 35305 }, { "epoch": 88.265, "grad_norm": 7.65625, "learning_rate": 7.136056541139103e-07, "loss": 1.084, "step": 35306 }, { "epoch": 88.2675, "grad_norm": 9.0, "learning_rate": 7.133053033122361e-07, "loss": 1.4779, "step": 35307 }, { "epoch": 88.27, "grad_norm": 7.28125, "learning_rate": 7.130050133937617e-07, "loss": 1.3558, "step": 35308 }, { "epoch": 88.2725, "grad_norm": 4.84375, "learning_rate": 7.127047843604562e-07, "loss": 1.1035, "step": 35309 }, { "epoch": 88.275, "grad_norm": 8.875, "learning_rate": 7.124046162142862e-07, "loss": 1.2346, "step": 35310 }, { "epoch": 88.2775, "grad_norm": 6.25, "learning_rate": 7.121045089572176e-07, "loss": 1.2074, "step": 35311 }, { "epoch": 88.28, "grad_norm": 6.90625, "learning_rate": 7.118044625912213e-07, "loss": 1.2323, "step": 35312 }, { "epoch": 88.2825, "grad_norm": 10.0625, "learning_rate": 7.115044771182645e-07, "loss": 1.3965, "step": 35313 }, { "epoch": 88.285, "grad_norm": 5.625, "learning_rate": 7.112045525403122e-07, "loss": 1.3259, "step": 35314 }, { "epoch": 88.2875, "grad_norm": 8.9375, "learning_rate": 7.109046888593296e-07, "loss": 1.257, "step": 35315 }, { "epoch": 88.29, "grad_norm": 9.75, "learning_rate": 7.106048860772863e-07, "loss": 1.1819, "step": 35316 }, { "epoch": 88.2925, "grad_norm": 8.875, "learning_rate": 7.10305144196144e-07, "loss": 1.277, "step": 35317 }, { "epoch": 88.295, "grad_norm": 9.0625, "learning_rate": 7.10005463217871e-07, "loss": 1.445, "step": 35318 }, { "epoch": 88.2975, "grad_norm": 5.75, "learning_rate": 7.097058431444281e-07, "loss": 1.2346, "step": 35319 }, { "epoch": 88.3, "grad_norm": 8.1875, "learning_rate": 7.094062839777838e-07, "loss": 1.5027, "step": 35320 }, { "epoch": 88.3025, "grad_norm": 6.46875, "learning_rate": 7.091067857199008e-07, "loss": 1.5255, "step": 35321 }, { "epoch": 88.305, "grad_norm": 7.59375, "learning_rate": 7.088073483727398e-07, "loss": 1.3603, "step": 35322 }, { "epoch": 88.3075, "grad_norm": 7.90625, "learning_rate": 7.085079719382659e-07, "loss": 1.282, "step": 35323 }, { "epoch": 88.31, "grad_norm": 7.8125, "learning_rate": 7.082086564184432e-07, "loss": 1.4202, "step": 35324 }, { "epoch": 88.3125, "grad_norm": 7.1875, "learning_rate": 7.079094018152333e-07, "loss": 1.4544, "step": 35325 }, { "epoch": 88.315, "grad_norm": 4.84375, "learning_rate": 7.076102081305957e-07, "loss": 1.2939, "step": 35326 }, { "epoch": 88.3175, "grad_norm": 6.96875, "learning_rate": 7.073110753664936e-07, "loss": 1.442, "step": 35327 }, { "epoch": 88.32, "grad_norm": 8.0625, "learning_rate": 7.070120035248906e-07, "loss": 1.3745, "step": 35328 }, { "epoch": 88.3225, "grad_norm": 9.5, "learning_rate": 7.067129926077443e-07, "loss": 1.5234, "step": 35329 }, { "epoch": 88.325, "grad_norm": 7.59375, "learning_rate": 7.064140426170141e-07, "loss": 1.2258, "step": 35330 }, { "epoch": 88.3275, "grad_norm": 6.9375, "learning_rate": 7.06115153554664e-07, "loss": 1.2832, "step": 35331 }, { "epoch": 88.33, "grad_norm": 5.65625, "learning_rate": 7.05816325422648e-07, "loss": 1.2346, "step": 35332 }, { "epoch": 88.3325, "grad_norm": 6.09375, "learning_rate": 7.055175582229313e-07, "loss": 1.2689, "step": 35333 }, { "epoch": 88.335, "grad_norm": 7.84375, "learning_rate": 7.052188519574665e-07, "loss": 1.4249, "step": 35334 }, { "epoch": 88.3375, "grad_norm": 9.625, "learning_rate": 7.049202066282179e-07, "loss": 1.2705, "step": 35335 }, { "epoch": 88.34, "grad_norm": 16.0, "learning_rate": 7.046216222371393e-07, "loss": 1.3452, "step": 35336 }, { "epoch": 88.3425, "grad_norm": 7.4375, "learning_rate": 7.043230987861883e-07, "loss": 1.3229, "step": 35337 }, { "epoch": 88.345, "grad_norm": 7.21875, "learning_rate": 7.040246362773229e-07, "loss": 1.4462, "step": 35338 }, { "epoch": 88.3475, "grad_norm": 5.75, "learning_rate": 7.037262347125018e-07, "loss": 1.2831, "step": 35339 }, { "epoch": 88.35, "grad_norm": 7.25, "learning_rate": 7.03427894093679e-07, "loss": 1.2377, "step": 35340 }, { "epoch": 88.3525, "grad_norm": 7.4375, "learning_rate": 7.031296144228106e-07, "loss": 1.3227, "step": 35341 }, { "epoch": 88.355, "grad_norm": 7.46875, "learning_rate": 7.028313957018507e-07, "loss": 1.3581, "step": 35342 }, { "epoch": 88.3575, "grad_norm": 8.5, "learning_rate": 7.025332379327587e-07, "loss": 1.2831, "step": 35343 }, { "epoch": 88.36, "grad_norm": 6.03125, "learning_rate": 7.022351411174866e-07, "loss": 1.5628, "step": 35344 }, { "epoch": 88.3625, "grad_norm": 7.0625, "learning_rate": 7.019371052579871e-07, "loss": 1.3723, "step": 35345 }, { "epoch": 88.365, "grad_norm": 8.125, "learning_rate": 7.016391303562175e-07, "loss": 1.1847, "step": 35346 }, { "epoch": 88.3675, "grad_norm": 5.78125, "learning_rate": 7.013412164141287e-07, "loss": 1.3125, "step": 35347 }, { "epoch": 88.37, "grad_norm": 9.125, "learning_rate": 7.010433634336755e-07, "loss": 1.2417, "step": 35348 }, { "epoch": 88.3725, "grad_norm": 8.3125, "learning_rate": 7.007455714168088e-07, "loss": 1.3782, "step": 35349 }, { "epoch": 88.375, "grad_norm": 11.1875, "learning_rate": 7.004478403654835e-07, "loss": 1.3714, "step": 35350 }, { "epoch": 88.3775, "grad_norm": 10.25, "learning_rate": 7.001501702816493e-07, "loss": 1.5589, "step": 35351 }, { "epoch": 88.38, "grad_norm": 6.03125, "learning_rate": 6.99852561167258e-07, "loss": 1.4405, "step": 35352 }, { "epoch": 88.3825, "grad_norm": 8.375, "learning_rate": 6.995550130242601e-07, "loss": 1.3643, "step": 35353 }, { "epoch": 88.385, "grad_norm": 9.75, "learning_rate": 6.992575258546098e-07, "loss": 1.1432, "step": 35354 }, { "epoch": 88.3875, "grad_norm": 6.625, "learning_rate": 6.989600996602541e-07, "loss": 1.2495, "step": 35355 }, { "epoch": 88.39, "grad_norm": 6.96875, "learning_rate": 6.98662734443144e-07, "loss": 1.3169, "step": 35356 }, { "epoch": 88.3925, "grad_norm": 5.0625, "learning_rate": 6.983654302052268e-07, "loss": 1.2939, "step": 35357 }, { "epoch": 88.395, "grad_norm": 7.0, "learning_rate": 6.980681869484551e-07, "loss": 1.23, "step": 35358 }, { "epoch": 88.3975, "grad_norm": 9.0625, "learning_rate": 6.977710046747766e-07, "loss": 1.2251, "step": 35359 }, { "epoch": 88.4, "grad_norm": 8.625, "learning_rate": 6.974738833861383e-07, "loss": 1.1879, "step": 35360 }, { "epoch": 88.4025, "grad_norm": 8.5, "learning_rate": 6.9717682308449e-07, "loss": 1.3677, "step": 35361 }, { "epoch": 88.405, "grad_norm": 7.5, "learning_rate": 6.968798237717766e-07, "loss": 1.379, "step": 35362 }, { "epoch": 88.4075, "grad_norm": 5.1875, "learning_rate": 6.96582885449949e-07, "loss": 1.3385, "step": 35363 }, { "epoch": 88.41, "grad_norm": 6.40625, "learning_rate": 6.962860081209489e-07, "loss": 1.317, "step": 35364 }, { "epoch": 88.4125, "grad_norm": 5.5625, "learning_rate": 6.959891917867279e-07, "loss": 1.3528, "step": 35365 }, { "epoch": 88.415, "grad_norm": 6.0, "learning_rate": 6.956924364492291e-07, "loss": 1.4127, "step": 35366 }, { "epoch": 88.4175, "grad_norm": 6.5625, "learning_rate": 6.953957421103963e-07, "loss": 1.3665, "step": 35367 }, { "epoch": 88.42, "grad_norm": 7.09375, "learning_rate": 6.950991087721781e-07, "loss": 1.1769, "step": 35368 }, { "epoch": 88.4225, "grad_norm": 10.5, "learning_rate": 6.948025364365185e-07, "loss": 1.3185, "step": 35369 }, { "epoch": 88.425, "grad_norm": 10.5625, "learning_rate": 6.945060251053603e-07, "loss": 1.3145, "step": 35370 }, { "epoch": 88.4275, "grad_norm": 7.09375, "learning_rate": 6.942095747806499e-07, "loss": 1.2712, "step": 35371 }, { "epoch": 88.43, "grad_norm": 8.0, "learning_rate": 6.939131854643244e-07, "loss": 1.3089, "step": 35372 }, { "epoch": 88.4325, "grad_norm": 7.90625, "learning_rate": 6.936168571583357e-07, "loss": 1.3178, "step": 35373 }, { "epoch": 88.435, "grad_norm": 10.0625, "learning_rate": 6.933205898646223e-07, "loss": 1.3313, "step": 35374 }, { "epoch": 88.4375, "grad_norm": 5.5, "learning_rate": 6.930243835851258e-07, "loss": 1.2746, "step": 35375 }, { "epoch": 88.44, "grad_norm": 7.71875, "learning_rate": 6.927282383217893e-07, "loss": 1.2375, "step": 35376 }, { "epoch": 88.4425, "grad_norm": 5.9375, "learning_rate": 6.924321540765522e-07, "loss": 1.2883, "step": 35377 }, { "epoch": 88.445, "grad_norm": 8.8125, "learning_rate": 6.921361308513597e-07, "loss": 1.3226, "step": 35378 }, { "epoch": 88.4475, "grad_norm": 8.75, "learning_rate": 6.918401686481501e-07, "loss": 1.3364, "step": 35379 }, { "epoch": 88.45, "grad_norm": 6.84375, "learning_rate": 6.915442674688633e-07, "loss": 1.3674, "step": 35380 }, { "epoch": 88.4525, "grad_norm": 8.5625, "learning_rate": 6.912484273154396e-07, "loss": 1.279, "step": 35381 }, { "epoch": 88.455, "grad_norm": 7.09375, "learning_rate": 6.909526481898188e-07, "loss": 1.3978, "step": 35382 }, { "epoch": 88.4575, "grad_norm": 9.125, "learning_rate": 6.906569300939403e-07, "loss": 1.403, "step": 35383 }, { "epoch": 88.46, "grad_norm": 7.09375, "learning_rate": 6.903612730297416e-07, "loss": 1.3459, "step": 35384 }, { "epoch": 88.4625, "grad_norm": 7.40625, "learning_rate": 6.900656769991632e-07, "loss": 1.3452, "step": 35385 }, { "epoch": 88.465, "grad_norm": 6.6875, "learning_rate": 6.897701420041414e-07, "loss": 1.4639, "step": 35386 }, { "epoch": 88.4675, "grad_norm": 8.1875, "learning_rate": 6.894746680466113e-07, "loss": 1.3651, "step": 35387 }, { "epoch": 88.47, "grad_norm": 9.25, "learning_rate": 6.891792551285148e-07, "loss": 1.4746, "step": 35388 }, { "epoch": 88.4725, "grad_norm": 6.46875, "learning_rate": 6.88883903251788e-07, "loss": 1.2959, "step": 35389 }, { "epoch": 88.475, "grad_norm": 9.4375, "learning_rate": 6.88588612418365e-07, "loss": 1.323, "step": 35390 }, { "epoch": 88.4775, "grad_norm": 7.4375, "learning_rate": 6.882933826301807e-07, "loss": 1.3642, "step": 35391 }, { "epoch": 88.48, "grad_norm": 5.71875, "learning_rate": 6.879982138891717e-07, "loss": 1.3782, "step": 35392 }, { "epoch": 88.4825, "grad_norm": 9.125, "learning_rate": 6.877031061972761e-07, "loss": 1.2961, "step": 35393 }, { "epoch": 88.485, "grad_norm": 9.0625, "learning_rate": 6.874080595564269e-07, "loss": 1.3221, "step": 35394 }, { "epoch": 88.4875, "grad_norm": 6.25, "learning_rate": 6.87113073968555e-07, "loss": 1.3216, "step": 35395 }, { "epoch": 88.49, "grad_norm": 5.59375, "learning_rate": 6.868181494355996e-07, "loss": 1.1196, "step": 35396 }, { "epoch": 88.4925, "grad_norm": 6.84375, "learning_rate": 6.865232859594895e-07, "loss": 1.3316, "step": 35397 }, { "epoch": 88.495, "grad_norm": 8.9375, "learning_rate": 6.862284835421629e-07, "loss": 1.2085, "step": 35398 }, { "epoch": 88.4975, "grad_norm": 5.21875, "learning_rate": 6.859337421855461e-07, "loss": 1.2045, "step": 35399 }, { "epoch": 88.5, "grad_norm": 8.4375, "learning_rate": 6.856390618915775e-07, "loss": 1.5002, "step": 35400 }, { "epoch": 88.5025, "grad_norm": 6.21875, "learning_rate": 6.853444426621869e-07, "loss": 1.4986, "step": 35401 }, { "epoch": 88.505, "grad_norm": 6.71875, "learning_rate": 6.850498844993037e-07, "loss": 1.2899, "step": 35402 }, { "epoch": 88.5075, "grad_norm": 7.4375, "learning_rate": 6.847553874048607e-07, "loss": 1.5376, "step": 35403 }, { "epoch": 88.51, "grad_norm": 6.40625, "learning_rate": 6.844609513807898e-07, "loss": 1.2156, "step": 35404 }, { "epoch": 88.5125, "grad_norm": 9.0625, "learning_rate": 6.841665764290195e-07, "loss": 1.3931, "step": 35405 }, { "epoch": 88.515, "grad_norm": 7.75, "learning_rate": 6.838722625514793e-07, "loss": 1.1838, "step": 35406 }, { "epoch": 88.5175, "grad_norm": 7.5625, "learning_rate": 6.835780097500999e-07, "loss": 1.4341, "step": 35407 }, { "epoch": 88.52, "grad_norm": 8.5625, "learning_rate": 6.83283818026812e-07, "loss": 1.5234, "step": 35408 }, { "epoch": 88.5225, "grad_norm": 9.9375, "learning_rate": 6.829896873835429e-07, "loss": 1.3162, "step": 35409 }, { "epoch": 88.525, "grad_norm": 8.5, "learning_rate": 6.826956178222178e-07, "loss": 1.2329, "step": 35410 }, { "epoch": 88.5275, "grad_norm": 4.90625, "learning_rate": 6.824016093447705e-07, "loss": 1.2413, "step": 35411 }, { "epoch": 88.53, "grad_norm": 9.5625, "learning_rate": 6.82107661953122e-07, "loss": 1.2215, "step": 35412 }, { "epoch": 88.5325, "grad_norm": 10.125, "learning_rate": 6.81813775649206e-07, "loss": 1.4739, "step": 35413 }, { "epoch": 88.535, "grad_norm": 8.5625, "learning_rate": 6.815199504349435e-07, "loss": 1.4648, "step": 35414 }, { "epoch": 88.5375, "grad_norm": 11.875, "learning_rate": 6.812261863122649e-07, "loss": 1.3538, "step": 35415 }, { "epoch": 88.54, "grad_norm": 7.96875, "learning_rate": 6.809324832830944e-07, "loss": 1.2976, "step": 35416 }, { "epoch": 88.5425, "grad_norm": 7.21875, "learning_rate": 6.80638841349357e-07, "loss": 1.3851, "step": 35417 }, { "epoch": 88.545, "grad_norm": 6.875, "learning_rate": 6.803452605129767e-07, "loss": 1.4028, "step": 35418 }, { "epoch": 88.5475, "grad_norm": 8.6875, "learning_rate": 6.800517407758834e-07, "loss": 1.2229, "step": 35419 }, { "epoch": 88.55, "grad_norm": 6.9375, "learning_rate": 6.797582821399973e-07, "loss": 1.3355, "step": 35420 }, { "epoch": 88.5525, "grad_norm": 7.9375, "learning_rate": 6.794648846072405e-07, "loss": 1.3807, "step": 35421 }, { "epoch": 88.555, "grad_norm": 7.125, "learning_rate": 6.791715481795391e-07, "loss": 1.2988, "step": 35422 }, { "epoch": 88.5575, "grad_norm": 7.5625, "learning_rate": 6.788782728588184e-07, "loss": 1.3662, "step": 35423 }, { "epoch": 88.56, "grad_norm": 8.3125, "learning_rate": 6.785850586469989e-07, "loss": 1.4572, "step": 35424 }, { "epoch": 88.5625, "grad_norm": 8.5625, "learning_rate": 6.782919055460002e-07, "loss": 1.2415, "step": 35425 }, { "epoch": 88.565, "grad_norm": 8.375, "learning_rate": 6.779988135577487e-07, "loss": 1.2941, "step": 35426 }, { "epoch": 88.5675, "grad_norm": 7.90625, "learning_rate": 6.777057826841627e-07, "loss": 1.2226, "step": 35427 }, { "epoch": 88.57, "grad_norm": 7.78125, "learning_rate": 6.774128129271651e-07, "loss": 1.3374, "step": 35428 }, { "epoch": 88.5725, "grad_norm": 23.375, "learning_rate": 6.771199042886755e-07, "loss": 1.5701, "step": 35429 }, { "epoch": 88.575, "grad_norm": 5.875, "learning_rate": 6.768270567706159e-07, "loss": 1.4563, "step": 35430 }, { "epoch": 88.5775, "grad_norm": 8.3125, "learning_rate": 6.765342703749056e-07, "loss": 1.4011, "step": 35431 }, { "epoch": 88.58, "grad_norm": 10.875, "learning_rate": 6.76241545103461e-07, "loss": 1.3019, "step": 35432 }, { "epoch": 88.5825, "grad_norm": 7.90625, "learning_rate": 6.759488809582037e-07, "loss": 1.2401, "step": 35433 }, { "epoch": 88.585, "grad_norm": 6.78125, "learning_rate": 6.756562779410547e-07, "loss": 1.2954, "step": 35434 }, { "epoch": 88.5875, "grad_norm": 8.25, "learning_rate": 6.7536373605393e-07, "loss": 1.3853, "step": 35435 }, { "epoch": 88.59, "grad_norm": 5.34375, "learning_rate": 6.75071255298746e-07, "loss": 1.3448, "step": 35436 }, { "epoch": 88.5925, "grad_norm": 10.25, "learning_rate": 6.74778835677421e-07, "loss": 1.2721, "step": 35437 }, { "epoch": 88.595, "grad_norm": 10.9375, "learning_rate": 6.744864771918747e-07, "loss": 1.4025, "step": 35438 }, { "epoch": 88.5975, "grad_norm": 8.5, "learning_rate": 6.741941798440222e-07, "loss": 1.3806, "step": 35439 }, { "epoch": 88.6, "grad_norm": 7.46875, "learning_rate": 6.739019436357774e-07, "loss": 1.3402, "step": 35440 }, { "epoch": 88.6025, "grad_norm": 6.375, "learning_rate": 6.736097685690601e-07, "loss": 1.4786, "step": 35441 }, { "epoch": 88.605, "grad_norm": 6.375, "learning_rate": 6.73317654645782e-07, "loss": 1.2368, "step": 35442 }, { "epoch": 88.6075, "grad_norm": 6.53125, "learning_rate": 6.730256018678616e-07, "loss": 1.16, "step": 35443 }, { "epoch": 88.61, "grad_norm": 9.8125, "learning_rate": 6.727336102372106e-07, "loss": 1.3125, "step": 35444 }, { "epoch": 88.6125, "grad_norm": 8.25, "learning_rate": 6.724416797557476e-07, "loss": 1.4041, "step": 35445 }, { "epoch": 88.615, "grad_norm": 6.375, "learning_rate": 6.721498104253821e-07, "loss": 1.2012, "step": 35446 }, { "epoch": 88.6175, "grad_norm": 6.40625, "learning_rate": 6.718580022480281e-07, "loss": 1.3313, "step": 35447 }, { "epoch": 88.62, "grad_norm": 7.71875, "learning_rate": 6.715662552255998e-07, "loss": 1.4361, "step": 35448 }, { "epoch": 88.6225, "grad_norm": 10.5625, "learning_rate": 6.71274569360012e-07, "loss": 1.4929, "step": 35449 }, { "epoch": 88.625, "grad_norm": 5.78125, "learning_rate": 6.709829446531734e-07, "loss": 1.2284, "step": 35450 }, { "epoch": 88.6275, "grad_norm": 7.75, "learning_rate": 6.706913811069971e-07, "loss": 1.5254, "step": 35451 }, { "epoch": 88.63, "grad_norm": 10.6875, "learning_rate": 6.703998787233946e-07, "loss": 1.3082, "step": 35452 }, { "epoch": 88.6325, "grad_norm": 11.0625, "learning_rate": 6.701084375042776e-07, "loss": 1.3617, "step": 35453 }, { "epoch": 88.635, "grad_norm": 4.875, "learning_rate": 6.698170574515573e-07, "loss": 1.3489, "step": 35454 }, { "epoch": 88.6375, "grad_norm": 7.875, "learning_rate": 6.695257385671417e-07, "loss": 1.3656, "step": 35455 }, { "epoch": 88.64, "grad_norm": 4.4375, "learning_rate": 6.692344808529427e-07, "loss": 1.3361, "step": 35456 }, { "epoch": 88.6425, "grad_norm": 5.8125, "learning_rate": 6.689432843108678e-07, "loss": 1.3393, "step": 35457 }, { "epoch": 88.645, "grad_norm": 7.15625, "learning_rate": 6.686521489428288e-07, "loss": 1.1988, "step": 35458 }, { "epoch": 88.6475, "grad_norm": 9.625, "learning_rate": 6.683610747507318e-07, "loss": 1.3729, "step": 35459 }, { "epoch": 88.65, "grad_norm": 8.1875, "learning_rate": 6.680700617364877e-07, "loss": 1.4157, "step": 35460 }, { "epoch": 88.6525, "grad_norm": 12.0, "learning_rate": 6.677791099020015e-07, "loss": 1.3405, "step": 35461 }, { "epoch": 88.655, "grad_norm": 8.6875, "learning_rate": 6.674882192491816e-07, "loss": 1.2382, "step": 35462 }, { "epoch": 88.6575, "grad_norm": 5.34375, "learning_rate": 6.671973897799355e-07, "loss": 1.2562, "step": 35463 }, { "epoch": 88.66, "grad_norm": 9.25, "learning_rate": 6.669066214961717e-07, "loss": 1.0958, "step": 35464 }, { "epoch": 88.6625, "grad_norm": 6.625, "learning_rate": 6.666159143997941e-07, "loss": 1.333, "step": 35465 }, { "epoch": 88.665, "grad_norm": 6.71875, "learning_rate": 6.663252684927068e-07, "loss": 1.2848, "step": 35466 }, { "epoch": 88.6675, "grad_norm": 8.625, "learning_rate": 6.660346837768183e-07, "loss": 1.4907, "step": 35467 }, { "epoch": 88.67, "grad_norm": 8.875, "learning_rate": 6.657441602540348e-07, "loss": 1.3319, "step": 35468 }, { "epoch": 88.6725, "grad_norm": 11.0, "learning_rate": 6.654536979262583e-07, "loss": 1.4017, "step": 35469 }, { "epoch": 88.675, "grad_norm": 10.3125, "learning_rate": 6.651632967953925e-07, "loss": 1.3163, "step": 35470 }, { "epoch": 88.6775, "grad_norm": 11.4375, "learning_rate": 6.648729568633449e-07, "loss": 1.3691, "step": 35471 }, { "epoch": 88.68, "grad_norm": 8.25, "learning_rate": 6.645826781320141e-07, "loss": 1.3625, "step": 35472 }, { "epoch": 88.6825, "grad_norm": 6.9375, "learning_rate": 6.642924606033075e-07, "loss": 1.4246, "step": 35473 }, { "epoch": 88.685, "grad_norm": 5.78125, "learning_rate": 6.640023042791244e-07, "loss": 1.3295, "step": 35474 }, { "epoch": 88.6875, "grad_norm": 7.6875, "learning_rate": 6.637122091613702e-07, "loss": 1.2245, "step": 35475 }, { "epoch": 88.69, "grad_norm": 10.375, "learning_rate": 6.634221752519454e-07, "loss": 1.4299, "step": 35476 }, { "epoch": 88.6925, "grad_norm": 8.1875, "learning_rate": 6.631322025527487e-07, "loss": 1.2439, "step": 35477 }, { "epoch": 88.695, "grad_norm": 6.84375, "learning_rate": 6.628422910656839e-07, "loss": 1.2239, "step": 35478 }, { "epoch": 88.6975, "grad_norm": 8.4375, "learning_rate": 6.625524407926542e-07, "loss": 1.3737, "step": 35479 }, { "epoch": 88.7, "grad_norm": 9.0, "learning_rate": 6.622626517355557e-07, "loss": 1.2951, "step": 35480 }, { "epoch": 88.7025, "grad_norm": 5.5625, "learning_rate": 6.61972923896288e-07, "loss": 1.2008, "step": 35481 }, { "epoch": 88.705, "grad_norm": 7.34375, "learning_rate": 6.616832572767529e-07, "loss": 1.3681, "step": 35482 }, { "epoch": 88.7075, "grad_norm": 8.125, "learning_rate": 6.613936518788499e-07, "loss": 1.3507, "step": 35483 }, { "epoch": 88.71, "grad_norm": 8.125, "learning_rate": 6.611041077044766e-07, "loss": 1.3685, "step": 35484 }, { "epoch": 88.7125, "grad_norm": 7.15625, "learning_rate": 6.608146247555291e-07, "loss": 1.3582, "step": 35485 }, { "epoch": 88.715, "grad_norm": 7.25, "learning_rate": 6.605252030339093e-07, "loss": 1.2053, "step": 35486 }, { "epoch": 88.7175, "grad_norm": 6.5, "learning_rate": 6.602358425415101e-07, "loss": 1.2291, "step": 35487 }, { "epoch": 88.72, "grad_norm": 9.75, "learning_rate": 6.599465432802332e-07, "loss": 1.2799, "step": 35488 }, { "epoch": 88.7225, "grad_norm": 7.125, "learning_rate": 6.596573052519717e-07, "loss": 1.1511, "step": 35489 }, { "epoch": 88.725, "grad_norm": 6.4375, "learning_rate": 6.59368128458624e-07, "loss": 1.287, "step": 35490 }, { "epoch": 88.7275, "grad_norm": 8.4375, "learning_rate": 6.590790129020852e-07, "loss": 1.3447, "step": 35491 }, { "epoch": 88.73, "grad_norm": 7.5625, "learning_rate": 6.587899585842494e-07, "loss": 1.2935, "step": 35492 }, { "epoch": 88.7325, "grad_norm": 5.96875, "learning_rate": 6.585009655070118e-07, "loss": 1.3323, "step": 35493 }, { "epoch": 88.735, "grad_norm": 11.125, "learning_rate": 6.582120336722708e-07, "loss": 1.2787, "step": 35494 }, { "epoch": 88.7375, "grad_norm": 11.0, "learning_rate": 6.579231630819171e-07, "loss": 1.3216, "step": 35495 }, { "epoch": 88.74, "grad_norm": 5.65625, "learning_rate": 6.57634353737846e-07, "loss": 1.2661, "step": 35496 }, { "epoch": 88.7425, "grad_norm": 6.125, "learning_rate": 6.57345605641947e-07, "loss": 1.2406, "step": 35497 }, { "epoch": 88.745, "grad_norm": 8.8125, "learning_rate": 6.570569187961185e-07, "loss": 1.3727, "step": 35498 }, { "epoch": 88.7475, "grad_norm": 7.34375, "learning_rate": 6.567682932022523e-07, "loss": 1.1933, "step": 35499 }, { "epoch": 88.75, "grad_norm": 6.84375, "learning_rate": 6.564797288622371e-07, "loss": 1.0957, "step": 35500 }, { "epoch": 88.7525, "grad_norm": 8.375, "learning_rate": 6.561912257779679e-07, "loss": 1.2928, "step": 35501 }, { "epoch": 88.755, "grad_norm": 9.5, "learning_rate": 6.559027839513343e-07, "loss": 1.3164, "step": 35502 }, { "epoch": 88.7575, "grad_norm": 8.0625, "learning_rate": 6.556144033842294e-07, "loss": 1.3089, "step": 35503 }, { "epoch": 88.76, "grad_norm": 11.9375, "learning_rate": 6.553260840785414e-07, "loss": 1.2343, "step": 35504 }, { "epoch": 88.7625, "grad_norm": 7.625, "learning_rate": 6.550378260361622e-07, "loss": 1.3712, "step": 35505 }, { "epoch": 88.765, "grad_norm": 7.0625, "learning_rate": 6.547496292589817e-07, "loss": 1.4093, "step": 35506 }, { "epoch": 88.7675, "grad_norm": 7.03125, "learning_rate": 6.544614937488869e-07, "loss": 1.3406, "step": 35507 }, { "epoch": 88.77, "grad_norm": 6.8125, "learning_rate": 6.541734195077687e-07, "loss": 1.2729, "step": 35508 }, { "epoch": 88.7725, "grad_norm": 7.6875, "learning_rate": 6.538854065375178e-07, "loss": 1.2092, "step": 35509 }, { "epoch": 88.775, "grad_norm": 9.4375, "learning_rate": 6.535974548400193e-07, "loss": 1.4291, "step": 35510 }, { "epoch": 88.7775, "grad_norm": 7.59375, "learning_rate": 6.533095644171616e-07, "loss": 1.2875, "step": 35511 }, { "epoch": 88.78, "grad_norm": 5.71875, "learning_rate": 6.530217352708301e-07, "loss": 1.4174, "step": 35512 }, { "epoch": 88.7825, "grad_norm": 6.34375, "learning_rate": 6.527339674029165e-07, "loss": 1.2394, "step": 35513 }, { "epoch": 88.785, "grad_norm": 9.625, "learning_rate": 6.524462608153049e-07, "loss": 1.2982, "step": 35514 }, { "epoch": 88.7875, "grad_norm": 8.125, "learning_rate": 6.521586155098825e-07, "loss": 1.292, "step": 35515 }, { "epoch": 88.79, "grad_norm": 6.96875, "learning_rate": 6.518710314885302e-07, "loss": 1.2745, "step": 35516 }, { "epoch": 88.7925, "grad_norm": 6.34375, "learning_rate": 6.515835087531408e-07, "loss": 1.2821, "step": 35517 }, { "epoch": 88.795, "grad_norm": 6.71875, "learning_rate": 6.512960473055952e-07, "loss": 1.2471, "step": 35518 }, { "epoch": 88.7975, "grad_norm": 6.96875, "learning_rate": 6.510086471477794e-07, "loss": 1.2941, "step": 35519 }, { "epoch": 88.8, "grad_norm": 5.5625, "learning_rate": 6.507213082815745e-07, "loss": 1.265, "step": 35520 }, { "epoch": 88.8025, "grad_norm": 8.0625, "learning_rate": 6.504340307088674e-07, "loss": 1.2731, "step": 35521 }, { "epoch": 88.805, "grad_norm": 7.25, "learning_rate": 6.501468144315415e-07, "loss": 1.4499, "step": 35522 }, { "epoch": 88.8075, "grad_norm": 7.09375, "learning_rate": 6.498596594514795e-07, "loss": 1.501, "step": 35523 }, { "epoch": 88.81, "grad_norm": 7.0625, "learning_rate": 6.495725657705609e-07, "loss": 1.2571, "step": 35524 }, { "epoch": 88.8125, "grad_norm": 7.6875, "learning_rate": 6.492855333906733e-07, "loss": 1.3114, "step": 35525 }, { "epoch": 88.815, "grad_norm": 5.625, "learning_rate": 6.48998562313693e-07, "loss": 1.2357, "step": 35526 }, { "epoch": 88.8175, "grad_norm": 6.46875, "learning_rate": 6.487116525415038e-07, "loss": 1.4084, "step": 35527 }, { "epoch": 88.82, "grad_norm": 5.78125, "learning_rate": 6.484248040759888e-07, "loss": 1.1204, "step": 35528 }, { "epoch": 88.8225, "grad_norm": 6.28125, "learning_rate": 6.481380169190265e-07, "loss": 1.3433, "step": 35529 }, { "epoch": 88.825, "grad_norm": 7.53125, "learning_rate": 6.478512910724966e-07, "loss": 1.2068, "step": 35530 }, { "epoch": 88.8275, "grad_norm": 8.5, "learning_rate": 6.475646265382773e-07, "loss": 1.4268, "step": 35531 }, { "epoch": 88.83, "grad_norm": 7.75, "learning_rate": 6.47278023318253e-07, "loss": 1.3896, "step": 35532 }, { "epoch": 88.8325, "grad_norm": 6.96875, "learning_rate": 6.469914814142986e-07, "loss": 1.2424, "step": 35533 }, { "epoch": 88.835, "grad_norm": 6.15625, "learning_rate": 6.467050008282949e-07, "loss": 1.3364, "step": 35534 }, { "epoch": 88.8375, "grad_norm": 5.84375, "learning_rate": 6.464185815621171e-07, "loss": 1.3311, "step": 35535 }, { "epoch": 88.84, "grad_norm": 6.59375, "learning_rate": 6.461322236176438e-07, "loss": 1.5582, "step": 35536 }, { "epoch": 88.8425, "grad_norm": 8.25, "learning_rate": 6.458459269967565e-07, "loss": 1.2483, "step": 35537 }, { "epoch": 88.845, "grad_norm": 6.78125, "learning_rate": 6.455596917013274e-07, "loss": 1.2536, "step": 35538 }, { "epoch": 88.8475, "grad_norm": 4.96875, "learning_rate": 6.452735177332348e-07, "loss": 1.3619, "step": 35539 }, { "epoch": 88.85, "grad_norm": 10.25, "learning_rate": 6.449874050943549e-07, "loss": 1.4114, "step": 35540 }, { "epoch": 88.8525, "grad_norm": 6.46875, "learning_rate": 6.447013537865621e-07, "loss": 1.2075, "step": 35541 }, { "epoch": 88.855, "grad_norm": 8.4375, "learning_rate": 6.444153638117334e-07, "loss": 1.2844, "step": 35542 }, { "epoch": 88.8575, "grad_norm": 8.5, "learning_rate": 6.441294351717431e-07, "loss": 1.3942, "step": 35543 }, { "epoch": 88.86, "grad_norm": 7.28125, "learning_rate": 6.438435678684662e-07, "loss": 1.4684, "step": 35544 }, { "epoch": 88.8625, "grad_norm": 5.65625, "learning_rate": 6.435577619037781e-07, "loss": 1.5233, "step": 35545 }, { "epoch": 88.865, "grad_norm": 7.3125, "learning_rate": 6.432720172795482e-07, "loss": 1.3097, "step": 35546 }, { "epoch": 88.8675, "grad_norm": 6.71875, "learning_rate": 6.429863339976516e-07, "loss": 1.2174, "step": 35547 }, { "epoch": 88.87, "grad_norm": 6.03125, "learning_rate": 6.427007120599649e-07, "loss": 1.2897, "step": 35548 }, { "epoch": 88.8725, "grad_norm": 7.375, "learning_rate": 6.424151514683585e-07, "loss": 1.2198, "step": 35549 }, { "epoch": 88.875, "grad_norm": 8.6875, "learning_rate": 6.421296522247012e-07, "loss": 1.485, "step": 35550 }, { "epoch": 88.8775, "grad_norm": 10.25, "learning_rate": 6.418442143308678e-07, "loss": 1.3559, "step": 35551 }, { "epoch": 88.88, "grad_norm": 9.0625, "learning_rate": 6.415588377887305e-07, "loss": 1.4609, "step": 35552 }, { "epoch": 88.8825, "grad_norm": 5.46875, "learning_rate": 6.412735226001587e-07, "loss": 1.2834, "step": 35553 }, { "epoch": 88.885, "grad_norm": 7.625, "learning_rate": 6.409882687670221e-07, "loss": 1.1809, "step": 35554 }, { "epoch": 88.8875, "grad_norm": 6.9375, "learning_rate": 6.407030762911936e-07, "loss": 1.212, "step": 35555 }, { "epoch": 88.89, "grad_norm": 5.84375, "learning_rate": 6.404179451745396e-07, "loss": 1.3379, "step": 35556 }, { "epoch": 88.8925, "grad_norm": 8.5, "learning_rate": 6.401328754189329e-07, "loss": 1.3966, "step": 35557 }, { "epoch": 88.895, "grad_norm": 11.5, "learning_rate": 6.398478670262386e-07, "loss": 1.5071, "step": 35558 }, { "epoch": 88.8975, "grad_norm": 8.0, "learning_rate": 6.395629199983278e-07, "loss": 1.2654, "step": 35559 }, { "epoch": 88.9, "grad_norm": 6.28125, "learning_rate": 6.392780343370686e-07, "loss": 1.1589, "step": 35560 }, { "epoch": 88.9025, "grad_norm": 9.375, "learning_rate": 6.389932100443274e-07, "loss": 1.318, "step": 35561 }, { "epoch": 88.905, "grad_norm": 7.6875, "learning_rate": 6.387084471219707e-07, "loss": 1.2843, "step": 35562 }, { "epoch": 88.9075, "grad_norm": 6.8125, "learning_rate": 6.38423745571869e-07, "loss": 1.1441, "step": 35563 }, { "epoch": 88.91, "grad_norm": 10.0625, "learning_rate": 6.381391053958863e-07, "loss": 1.425, "step": 35564 }, { "epoch": 88.9125, "grad_norm": 8.0625, "learning_rate": 6.37854526595888e-07, "loss": 1.2823, "step": 35565 }, { "epoch": 88.915, "grad_norm": 9.1875, "learning_rate": 6.375700091737402e-07, "loss": 1.2976, "step": 35566 }, { "epoch": 88.9175, "grad_norm": 12.375, "learning_rate": 6.372855531313105e-07, "loss": 1.4393, "step": 35567 }, { "epoch": 88.92, "grad_norm": 9.5625, "learning_rate": 6.370011584704617e-07, "loss": 1.3828, "step": 35568 }, { "epoch": 88.9225, "grad_norm": 8.5, "learning_rate": 6.367168251930578e-07, "loss": 1.27, "step": 35569 }, { "epoch": 88.925, "grad_norm": 7.375, "learning_rate": 6.364325533009652e-07, "loss": 1.3783, "step": 35570 }, { "epoch": 88.9275, "grad_norm": 8.75, "learning_rate": 6.361483427960435e-07, "loss": 1.327, "step": 35571 }, { "epoch": 88.93, "grad_norm": 7.78125, "learning_rate": 6.358641936801613e-07, "loss": 1.2686, "step": 35572 }, { "epoch": 88.9325, "grad_norm": 18.5, "learning_rate": 6.35580105955177e-07, "loss": 1.4715, "step": 35573 }, { "epoch": 88.935, "grad_norm": 7.5, "learning_rate": 6.352960796229557e-07, "loss": 1.3473, "step": 35574 }, { "epoch": 88.9375, "grad_norm": 6.65625, "learning_rate": 6.350121146853582e-07, "loss": 1.3636, "step": 35575 }, { "epoch": 88.94, "grad_norm": 7.0, "learning_rate": 6.347282111442455e-07, "loss": 1.36, "step": 35576 }, { "epoch": 88.9425, "grad_norm": 8.1875, "learning_rate": 6.344443690014801e-07, "loss": 1.3094, "step": 35577 }, { "epoch": 88.945, "grad_norm": 6.78125, "learning_rate": 6.341605882589241e-07, "loss": 1.5903, "step": 35578 }, { "epoch": 88.9475, "grad_norm": 5.21875, "learning_rate": 6.338768689184371e-07, "loss": 1.261, "step": 35579 }, { "epoch": 88.95, "grad_norm": 8.375, "learning_rate": 6.335932109818754e-07, "loss": 1.2811, "step": 35580 }, { "epoch": 88.9525, "grad_norm": 8.4375, "learning_rate": 6.33309614451103e-07, "loss": 1.2685, "step": 35581 }, { "epoch": 88.955, "grad_norm": 6.4375, "learning_rate": 6.330260793279797e-07, "loss": 1.1452, "step": 35582 }, { "epoch": 88.9575, "grad_norm": 8.4375, "learning_rate": 6.327426056143615e-07, "loss": 1.4874, "step": 35583 }, { "epoch": 88.96, "grad_norm": 6.8125, "learning_rate": 6.324591933121072e-07, "loss": 1.4814, "step": 35584 }, { "epoch": 88.9625, "grad_norm": 5.75, "learning_rate": 6.321758424230762e-07, "loss": 1.3416, "step": 35585 }, { "epoch": 88.965, "grad_norm": 10.4375, "learning_rate": 6.318925529491249e-07, "loss": 1.3505, "step": 35586 }, { "epoch": 88.9675, "grad_norm": 12.8125, "learning_rate": 6.31609324892113e-07, "loss": 1.5189, "step": 35587 }, { "epoch": 88.97, "grad_norm": 5.625, "learning_rate": 6.313261582538932e-07, "loss": 1.2226, "step": 35588 }, { "epoch": 88.9725, "grad_norm": 6.625, "learning_rate": 6.310430530363254e-07, "loss": 1.3692, "step": 35589 }, { "epoch": 88.975, "grad_norm": 7.125, "learning_rate": 6.307600092412647e-07, "loss": 1.2793, "step": 35590 }, { "epoch": 88.9775, "grad_norm": 8.6875, "learning_rate": 6.304770268705651e-07, "loss": 1.3445, "step": 35591 }, { "epoch": 88.98, "grad_norm": 6.40625, "learning_rate": 6.30194105926083e-07, "loss": 1.3298, "step": 35592 }, { "epoch": 88.9825, "grad_norm": 8.1875, "learning_rate": 6.299112464096757e-07, "loss": 1.3693, "step": 35593 }, { "epoch": 88.985, "grad_norm": 6.65625, "learning_rate": 6.29628448323194e-07, "loss": 1.3173, "step": 35594 }, { "epoch": 88.9875, "grad_norm": 7.09375, "learning_rate": 6.29345711668492e-07, "loss": 1.2184, "step": 35595 }, { "epoch": 88.99, "grad_norm": 5.59375, "learning_rate": 6.290630364474248e-07, "loss": 1.2054, "step": 35596 }, { "epoch": 88.9925, "grad_norm": 8.625, "learning_rate": 6.287804226618477e-07, "loss": 1.3111, "step": 35597 }, { "epoch": 88.995, "grad_norm": 5.84375, "learning_rate": 6.284978703136102e-07, "loss": 1.4154, "step": 35598 }, { "epoch": 88.9975, "grad_norm": 7.34375, "learning_rate": 6.282153794045642e-07, "loss": 1.3171, "step": 35599 }, { "epoch": 89.0, "grad_norm": 7.40625, "learning_rate": 6.279329499365649e-07, "loss": 1.2635, "step": 35600 }, { "epoch": 89.0025, "grad_norm": 8.0625, "learning_rate": 6.276505819114598e-07, "loss": 1.3727, "step": 35601 }, { "epoch": 89.005, "grad_norm": 6.875, "learning_rate": 6.27368275331105e-07, "loss": 1.2422, "step": 35602 }, { "epoch": 89.0075, "grad_norm": 8.25, "learning_rate": 6.27086030197347e-07, "loss": 1.2812, "step": 35603 }, { "epoch": 89.01, "grad_norm": 8.8125, "learning_rate": 6.268038465120386e-07, "loss": 1.2464, "step": 35604 }, { "epoch": 89.0125, "grad_norm": 8.0625, "learning_rate": 6.265217242770294e-07, "loss": 1.404, "step": 35605 }, { "epoch": 89.015, "grad_norm": 6.25, "learning_rate": 6.26239663494168e-07, "loss": 1.2802, "step": 35606 }, { "epoch": 89.0175, "grad_norm": 7.21875, "learning_rate": 6.259576641653031e-07, "loss": 1.3267, "step": 35607 }, { "epoch": 89.02, "grad_norm": 5.5625, "learning_rate": 6.256757262922863e-07, "loss": 1.2739, "step": 35608 }, { "epoch": 89.0225, "grad_norm": 7.0625, "learning_rate": 6.253938498769652e-07, "loss": 1.2502, "step": 35609 }, { "epoch": 89.025, "grad_norm": 6.84375, "learning_rate": 6.251120349211847e-07, "loss": 1.1837, "step": 35610 }, { "epoch": 89.0275, "grad_norm": 7.34375, "learning_rate": 6.248302814267948e-07, "loss": 1.203, "step": 35611 }, { "epoch": 89.03, "grad_norm": 4.625, "learning_rate": 6.245485893956438e-07, "loss": 1.3463, "step": 35612 }, { "epoch": 89.0325, "grad_norm": 8.0, "learning_rate": 6.242669588295769e-07, "loss": 1.3053, "step": 35613 }, { "epoch": 89.035, "grad_norm": 6.65625, "learning_rate": 6.239853897304393e-07, "loss": 1.317, "step": 35614 }, { "epoch": 89.0375, "grad_norm": 8.0625, "learning_rate": 6.237038821000795e-07, "loss": 1.1584, "step": 35615 }, { "epoch": 89.04, "grad_norm": 8.5625, "learning_rate": 6.234224359403407e-07, "loss": 1.3288, "step": 35616 }, { "epoch": 89.0425, "grad_norm": 6.375, "learning_rate": 6.231410512530712e-07, "loss": 1.2215, "step": 35617 }, { "epoch": 89.045, "grad_norm": 6.75, "learning_rate": 6.228597280401116e-07, "loss": 1.3183, "step": 35618 }, { "epoch": 89.0475, "grad_norm": 8.0, "learning_rate": 6.225784663033097e-07, "loss": 1.2583, "step": 35619 }, { "epoch": 89.05, "grad_norm": 9.0625, "learning_rate": 6.222972660445082e-07, "loss": 1.1408, "step": 35620 }, { "epoch": 89.0525, "grad_norm": 9.5625, "learning_rate": 6.220161272655489e-07, "loss": 1.3441, "step": 35621 }, { "epoch": 89.055, "grad_norm": 6.59375, "learning_rate": 6.217350499682772e-07, "loss": 1.3031, "step": 35622 }, { "epoch": 89.0575, "grad_norm": 7.09375, "learning_rate": 6.214540341545361e-07, "loss": 1.3442, "step": 35623 }, { "epoch": 89.06, "grad_norm": 9.0, "learning_rate": 6.211730798261673e-07, "loss": 1.3767, "step": 35624 }, { "epoch": 89.0625, "grad_norm": 6.5, "learning_rate": 6.208921869850104e-07, "loss": 1.1261, "step": 35625 }, { "epoch": 89.065, "grad_norm": 7.15625, "learning_rate": 6.206113556329096e-07, "loss": 1.3683, "step": 35626 }, { "epoch": 89.0675, "grad_norm": 9.625, "learning_rate": 6.203305857717068e-07, "loss": 1.4287, "step": 35627 }, { "epoch": 89.07, "grad_norm": 9.0, "learning_rate": 6.200498774032415e-07, "loss": 1.3649, "step": 35628 }, { "epoch": 89.0725, "grad_norm": 8.0, "learning_rate": 6.197692305293524e-07, "loss": 1.4058, "step": 35629 }, { "epoch": 89.075, "grad_norm": 5.84375, "learning_rate": 6.194886451518823e-07, "loss": 1.1555, "step": 35630 }, { "epoch": 89.0775, "grad_norm": 8.75, "learning_rate": 6.192081212726675e-07, "loss": 1.3934, "step": 35631 }, { "epoch": 89.08, "grad_norm": 6.90625, "learning_rate": 6.1892765889355e-07, "loss": 1.2459, "step": 35632 }, { "epoch": 89.0825, "grad_norm": 7.5625, "learning_rate": 6.186472580163671e-07, "loss": 1.404, "step": 35633 }, { "epoch": 89.085, "grad_norm": 19.25, "learning_rate": 6.183669186429576e-07, "loss": 1.3913, "step": 35634 }, { "epoch": 89.0875, "grad_norm": 6.78125, "learning_rate": 6.180866407751595e-07, "loss": 1.3994, "step": 35635 }, { "epoch": 89.09, "grad_norm": 7.15625, "learning_rate": 6.178064244148085e-07, "loss": 1.3303, "step": 35636 }, { "epoch": 89.0925, "grad_norm": 6.71875, "learning_rate": 6.175262695637441e-07, "loss": 1.3543, "step": 35637 }, { "epoch": 89.095, "grad_norm": 8.6875, "learning_rate": 6.172461762238024e-07, "loss": 1.3349, "step": 35638 }, { "epoch": 89.0975, "grad_norm": 8.875, "learning_rate": 6.169661443968189e-07, "loss": 1.2274, "step": 35639 }, { "epoch": 89.1, "grad_norm": 7.46875, "learning_rate": 6.166861740846297e-07, "loss": 1.3359, "step": 35640 }, { "epoch": 89.1025, "grad_norm": 6.96875, "learning_rate": 6.16406265289069e-07, "loss": 1.2025, "step": 35641 }, { "epoch": 89.105, "grad_norm": 5.78125, "learning_rate": 6.161264180119763e-07, "loss": 1.3578, "step": 35642 }, { "epoch": 89.1075, "grad_norm": 6.71875, "learning_rate": 6.158466322551826e-07, "loss": 1.3157, "step": 35643 }, { "epoch": 89.11, "grad_norm": 8.0, "learning_rate": 6.155669080205206e-07, "loss": 1.3749, "step": 35644 }, { "epoch": 89.1125, "grad_norm": 5.28125, "learning_rate": 6.15287245309829e-07, "loss": 1.3843, "step": 35645 }, { "epoch": 89.115, "grad_norm": 6.34375, "learning_rate": 6.150076441249364e-07, "loss": 1.1854, "step": 35646 }, { "epoch": 89.1175, "grad_norm": 8.3125, "learning_rate": 6.1472810446768e-07, "loss": 1.2349, "step": 35647 }, { "epoch": 89.12, "grad_norm": 6.125, "learning_rate": 6.144486263398886e-07, "loss": 1.2277, "step": 35648 }, { "epoch": 89.1225, "grad_norm": 7.25, "learning_rate": 6.141692097433982e-07, "loss": 1.2846, "step": 35649 }, { "epoch": 89.125, "grad_norm": 6.59375, "learning_rate": 6.138898546800398e-07, "loss": 1.249, "step": 35650 }, { "epoch": 89.1275, "grad_norm": 7.75, "learning_rate": 6.136105611516407e-07, "loss": 1.3123, "step": 35651 }, { "epoch": 89.13, "grad_norm": 5.25, "learning_rate": 6.133313291600362e-07, "loss": 1.3198, "step": 35652 }, { "epoch": 89.1325, "grad_norm": 7.1875, "learning_rate": 6.130521587070581e-07, "loss": 1.3367, "step": 35653 }, { "epoch": 89.135, "grad_norm": 8.75, "learning_rate": 6.127730497945327e-07, "loss": 1.2933, "step": 35654 }, { "epoch": 89.1375, "grad_norm": 5.875, "learning_rate": 6.124940024242931e-07, "loss": 1.3138, "step": 35655 }, { "epoch": 89.14, "grad_norm": 7.15625, "learning_rate": 6.122150165981633e-07, "loss": 1.3331, "step": 35656 }, { "epoch": 89.1425, "grad_norm": 7.25, "learning_rate": 6.119360923179785e-07, "loss": 1.2868, "step": 35657 }, { "epoch": 89.145, "grad_norm": 8.3125, "learning_rate": 6.11657229585566e-07, "loss": 1.186, "step": 35658 }, { "epoch": 89.1475, "grad_norm": 8.5, "learning_rate": 6.113784284027536e-07, "loss": 1.4559, "step": 35659 }, { "epoch": 89.15, "grad_norm": 10.5, "learning_rate": 6.11099688771366e-07, "loss": 1.3501, "step": 35660 }, { "epoch": 89.1525, "grad_norm": 4.25, "learning_rate": 6.108210106932333e-07, "loss": 1.2482, "step": 35661 }, { "epoch": 89.155, "grad_norm": 7.65625, "learning_rate": 6.105423941701838e-07, "loss": 1.4186, "step": 35662 }, { "epoch": 89.1575, "grad_norm": 9.4375, "learning_rate": 6.102638392040405e-07, "loss": 1.3093, "step": 35663 }, { "epoch": 89.16, "grad_norm": 15.4375, "learning_rate": 6.099853457966342e-07, "loss": 1.26, "step": 35664 }, { "epoch": 89.1625, "grad_norm": 8.8125, "learning_rate": 6.097069139497879e-07, "loss": 1.2653, "step": 35665 }, { "epoch": 89.165, "grad_norm": 8.75, "learning_rate": 6.094285436653258e-07, "loss": 1.3346, "step": 35666 }, { "epoch": 89.1675, "grad_norm": 5.90625, "learning_rate": 6.09150234945074e-07, "loss": 1.2638, "step": 35667 }, { "epoch": 89.17, "grad_norm": 5.96875, "learning_rate": 6.088719877908589e-07, "loss": 1.3449, "step": 35668 }, { "epoch": 89.1725, "grad_norm": 7.25, "learning_rate": 6.085938022045046e-07, "loss": 1.3899, "step": 35669 }, { "epoch": 89.175, "grad_norm": 8.625, "learning_rate": 6.083156781878319e-07, "loss": 1.351, "step": 35670 }, { "epoch": 89.1775, "grad_norm": 7.625, "learning_rate": 6.080376157426638e-07, "loss": 1.0974, "step": 35671 }, { "epoch": 89.18, "grad_norm": 7.0, "learning_rate": 6.077596148708275e-07, "loss": 1.1603, "step": 35672 }, { "epoch": 89.1825, "grad_norm": 6.28125, "learning_rate": 6.074816755741442e-07, "loss": 1.3495, "step": 35673 }, { "epoch": 89.185, "grad_norm": 10.5, "learning_rate": 6.072037978544343e-07, "loss": 1.2765, "step": 35674 }, { "epoch": 89.1875, "grad_norm": 20.0, "learning_rate": 6.069259817135187e-07, "loss": 1.2435, "step": 35675 }, { "epoch": 89.19, "grad_norm": 9.5625, "learning_rate": 6.066482271532215e-07, "loss": 1.2438, "step": 35676 }, { "epoch": 89.1925, "grad_norm": 10.1875, "learning_rate": 6.063705341753634e-07, "loss": 1.2634, "step": 35677 }, { "epoch": 89.195, "grad_norm": 8.1875, "learning_rate": 6.060929027817642e-07, "loss": 1.1455, "step": 35678 }, { "epoch": 89.1975, "grad_norm": 9.3125, "learning_rate": 6.058153329742433e-07, "loss": 1.2228, "step": 35679 }, { "epoch": 89.2, "grad_norm": 9.0625, "learning_rate": 6.055378247546217e-07, "loss": 1.2772, "step": 35680 }, { "epoch": 89.2025, "grad_norm": 7.71875, "learning_rate": 6.052603781247168e-07, "loss": 1.4454, "step": 35681 }, { "epoch": 89.205, "grad_norm": 9.8125, "learning_rate": 6.049829930863516e-07, "loss": 1.3361, "step": 35682 }, { "epoch": 89.2075, "grad_norm": 8.8125, "learning_rate": 6.04705669641339e-07, "loss": 1.4472, "step": 35683 }, { "epoch": 89.21, "grad_norm": 6.8125, "learning_rate": 6.044284077915019e-07, "loss": 1.3104, "step": 35684 }, { "epoch": 89.2125, "grad_norm": 8.125, "learning_rate": 6.041512075386568e-07, "loss": 1.3718, "step": 35685 }, { "epoch": 89.215, "grad_norm": 6.9375, "learning_rate": 6.038740688846178e-07, "loss": 1.3111, "step": 35686 }, { "epoch": 89.2175, "grad_norm": 5.90625, "learning_rate": 6.035969918312056e-07, "loss": 1.1775, "step": 35687 }, { "epoch": 89.22, "grad_norm": 10.0625, "learning_rate": 6.033199763802356e-07, "loss": 1.4346, "step": 35688 }, { "epoch": 89.2225, "grad_norm": 6.03125, "learning_rate": 6.030430225335249e-07, "loss": 1.3058, "step": 35689 }, { "epoch": 89.225, "grad_norm": 6.25, "learning_rate": 6.027661302928867e-07, "loss": 1.4736, "step": 35690 }, { "epoch": 89.2275, "grad_norm": 8.8125, "learning_rate": 6.024892996601372e-07, "loss": 1.1946, "step": 35691 }, { "epoch": 89.23, "grad_norm": 6.25, "learning_rate": 6.022125306370929e-07, "loss": 1.3242, "step": 35692 }, { "epoch": 89.2325, "grad_norm": 10.375, "learning_rate": 6.019358232255679e-07, "loss": 1.2836, "step": 35693 }, { "epoch": 89.235, "grad_norm": 10.0, "learning_rate": 6.016591774273739e-07, "loss": 1.3593, "step": 35694 }, { "epoch": 89.2375, "grad_norm": 7.0, "learning_rate": 6.013825932443273e-07, "loss": 1.3665, "step": 35695 }, { "epoch": 89.24, "grad_norm": 6.59375, "learning_rate": 6.01106070678239e-07, "loss": 1.2101, "step": 35696 }, { "epoch": 89.2425, "grad_norm": 5.96875, "learning_rate": 6.008296097309251e-07, "loss": 1.2932, "step": 35697 }, { "epoch": 89.245, "grad_norm": 10.9375, "learning_rate": 6.005532104041945e-07, "loss": 1.3376, "step": 35698 }, { "epoch": 89.2475, "grad_norm": 5.8125, "learning_rate": 6.00276872699862e-07, "loss": 1.3714, "step": 35699 }, { "epoch": 89.25, "grad_norm": 8.625, "learning_rate": 6.000005966197387e-07, "loss": 1.2806, "step": 35700 }, { "epoch": 89.2525, "grad_norm": 7.5625, "learning_rate": 5.997243821656329e-07, "loss": 1.3424, "step": 35701 }, { "epoch": 89.255, "grad_norm": 13.0625, "learning_rate": 5.994482293393588e-07, "loss": 1.2807, "step": 35702 }, { "epoch": 89.2575, "grad_norm": 8.0625, "learning_rate": 5.991721381427273e-07, "loss": 1.4197, "step": 35703 }, { "epoch": 89.26, "grad_norm": 7.5, "learning_rate": 5.988961085775468e-07, "loss": 1.4731, "step": 35704 }, { "epoch": 89.2625, "grad_norm": 8.25, "learning_rate": 5.986201406456261e-07, "loss": 1.459, "step": 35705 }, { "epoch": 89.265, "grad_norm": 8.6875, "learning_rate": 5.983442343487744e-07, "loss": 1.2458, "step": 35706 }, { "epoch": 89.2675, "grad_norm": 6.6875, "learning_rate": 5.98068389688804e-07, "loss": 1.3455, "step": 35707 }, { "epoch": 89.27, "grad_norm": 8.3125, "learning_rate": 5.97792606667521e-07, "loss": 1.4445, "step": 35708 }, { "epoch": 89.2725, "grad_norm": 9.75, "learning_rate": 5.975168852867319e-07, "loss": 1.4134, "step": 35709 }, { "epoch": 89.275, "grad_norm": 7.71875, "learning_rate": 5.972412255482474e-07, "loss": 1.3678, "step": 35710 }, { "epoch": 89.2775, "grad_norm": 6.5, "learning_rate": 5.969656274538716e-07, "loss": 1.2561, "step": 35711 }, { "epoch": 89.28, "grad_norm": 9.625, "learning_rate": 5.966900910054141e-07, "loss": 1.3891, "step": 35712 }, { "epoch": 89.2825, "grad_norm": 6.09375, "learning_rate": 5.964146162046792e-07, "loss": 1.3721, "step": 35713 }, { "epoch": 89.285, "grad_norm": 6.90625, "learning_rate": 5.961392030534752e-07, "loss": 1.3288, "step": 35714 }, { "epoch": 89.2875, "grad_norm": 5.34375, "learning_rate": 5.958638515536053e-07, "loss": 1.2124, "step": 35715 }, { "epoch": 89.29, "grad_norm": 12.0625, "learning_rate": 5.955885617068758e-07, "loss": 1.3025, "step": 35716 }, { "epoch": 89.2925, "grad_norm": 5.40625, "learning_rate": 5.953133335150895e-07, "loss": 1.3688, "step": 35717 }, { "epoch": 89.295, "grad_norm": 6.25, "learning_rate": 5.950381669800553e-07, "loss": 1.3473, "step": 35718 }, { "epoch": 89.2975, "grad_norm": 9.0, "learning_rate": 5.947630621035739e-07, "loss": 1.3711, "step": 35719 }, { "epoch": 89.3, "grad_norm": 8.3125, "learning_rate": 5.94488018887448e-07, "loss": 1.5222, "step": 35720 }, { "epoch": 89.3025, "grad_norm": 4.53125, "learning_rate": 5.942130373334831e-07, "loss": 1.2112, "step": 35721 }, { "epoch": 89.305, "grad_norm": 8.25, "learning_rate": 5.939381174434821e-07, "loss": 1.3657, "step": 35722 }, { "epoch": 89.3075, "grad_norm": 6.9375, "learning_rate": 5.93663259219246e-07, "loss": 1.3099, "step": 35723 }, { "epoch": 89.31, "grad_norm": 7.40625, "learning_rate": 5.933884626625764e-07, "loss": 1.3126, "step": 35724 }, { "epoch": 89.3125, "grad_norm": 6.40625, "learning_rate": 5.931137277752764e-07, "loss": 1.479, "step": 35725 }, { "epoch": 89.315, "grad_norm": 7.625, "learning_rate": 5.928390545591456e-07, "loss": 1.3681, "step": 35726 }, { "epoch": 89.3175, "grad_norm": 8.375, "learning_rate": 5.925644430159861e-07, "loss": 1.2365, "step": 35727 }, { "epoch": 89.32, "grad_norm": 5.46875, "learning_rate": 5.922898931475973e-07, "loss": 1.325, "step": 35728 }, { "epoch": 89.3225, "grad_norm": 9.25, "learning_rate": 5.920154049557802e-07, "loss": 1.4302, "step": 35729 }, { "epoch": 89.325, "grad_norm": 7.46875, "learning_rate": 5.917409784423334e-07, "loss": 1.3701, "step": 35730 }, { "epoch": 89.3275, "grad_norm": 7.21875, "learning_rate": 5.914666136090552e-07, "loss": 1.4563, "step": 35731 }, { "epoch": 89.33, "grad_norm": 7.09375, "learning_rate": 5.911923104577455e-07, "loss": 1.1817, "step": 35732 }, { "epoch": 89.3325, "grad_norm": 6.8125, "learning_rate": 5.909180689902039e-07, "loss": 1.1688, "step": 35733 }, { "epoch": 89.335, "grad_norm": 8.3125, "learning_rate": 5.906438892082267e-07, "loss": 1.3252, "step": 35734 }, { "epoch": 89.3375, "grad_norm": 7.5, "learning_rate": 5.903697711136092e-07, "loss": 1.4256, "step": 35735 }, { "epoch": 89.34, "grad_norm": 11.25, "learning_rate": 5.90095714708152e-07, "loss": 1.3583, "step": 35736 }, { "epoch": 89.3425, "grad_norm": 9.1875, "learning_rate": 5.898217199936529e-07, "loss": 1.2531, "step": 35737 }, { "epoch": 89.345, "grad_norm": 8.125, "learning_rate": 5.895477869719046e-07, "loss": 1.3049, "step": 35738 }, { "epoch": 89.3475, "grad_norm": 6.96875, "learning_rate": 5.892739156447036e-07, "loss": 1.2919, "step": 35739 }, { "epoch": 89.35, "grad_norm": 10.5, "learning_rate": 5.890001060138484e-07, "loss": 1.2008, "step": 35740 }, { "epoch": 89.3525, "grad_norm": 6.78125, "learning_rate": 5.887263580811297e-07, "loss": 1.2661, "step": 35741 }, { "epoch": 89.355, "grad_norm": 10.0625, "learning_rate": 5.884526718483452e-07, "loss": 1.2799, "step": 35742 }, { "epoch": 89.3575, "grad_norm": 5.1875, "learning_rate": 5.881790473172877e-07, "loss": 1.3097, "step": 35743 }, { "epoch": 89.36, "grad_norm": 6.9375, "learning_rate": 5.879054844897536e-07, "loss": 1.4959, "step": 35744 }, { "epoch": 89.3625, "grad_norm": 19.125, "learning_rate": 5.876319833675337e-07, "loss": 1.352, "step": 35745 }, { "epoch": 89.365, "grad_norm": 6.53125, "learning_rate": 5.873585439524199e-07, "loss": 1.3111, "step": 35746 }, { "epoch": 89.3675, "grad_norm": 6.46875, "learning_rate": 5.870851662462085e-07, "loss": 1.2727, "step": 35747 }, { "epoch": 89.37, "grad_norm": 7.78125, "learning_rate": 5.868118502506903e-07, "loss": 1.2483, "step": 35748 }, { "epoch": 89.3725, "grad_norm": 5.90625, "learning_rate": 5.865385959676573e-07, "loss": 1.3593, "step": 35749 }, { "epoch": 89.375, "grad_norm": 7.15625, "learning_rate": 5.86265403398899e-07, "loss": 1.4553, "step": 35750 }, { "epoch": 89.3775, "grad_norm": 7.59375, "learning_rate": 5.859922725462075e-07, "loss": 1.4495, "step": 35751 }, { "epoch": 89.38, "grad_norm": 5.84375, "learning_rate": 5.857192034113757e-07, "loss": 1.3597, "step": 35752 }, { "epoch": 89.3825, "grad_norm": 5.96875, "learning_rate": 5.854461959961922e-07, "loss": 1.1572, "step": 35753 }, { "epoch": 89.385, "grad_norm": 11.0, "learning_rate": 5.851732503024455e-07, "loss": 1.3366, "step": 35754 }, { "epoch": 89.3875, "grad_norm": 6.90625, "learning_rate": 5.849003663319275e-07, "loss": 1.292, "step": 35755 }, { "epoch": 89.39, "grad_norm": 5.1875, "learning_rate": 5.846275440864246e-07, "loss": 1.3724, "step": 35756 }, { "epoch": 89.3925, "grad_norm": 8.75, "learning_rate": 5.843547835677277e-07, "loss": 1.1617, "step": 35757 }, { "epoch": 89.395, "grad_norm": 7.96875, "learning_rate": 5.84082084777623e-07, "loss": 1.2767, "step": 35758 }, { "epoch": 89.3975, "grad_norm": 7.1875, "learning_rate": 5.838094477179013e-07, "loss": 1.1656, "step": 35759 }, { "epoch": 89.4, "grad_norm": 5.78125, "learning_rate": 5.835368723903456e-07, "loss": 1.2038, "step": 35760 }, { "epoch": 89.4025, "grad_norm": 8.3125, "learning_rate": 5.832643587967468e-07, "loss": 1.2859, "step": 35761 }, { "epoch": 89.405, "grad_norm": 8.3125, "learning_rate": 5.829919069388889e-07, "loss": 1.3323, "step": 35762 }, { "epoch": 89.4075, "grad_norm": 7.9375, "learning_rate": 5.827195168185607e-07, "loss": 1.4419, "step": 35763 }, { "epoch": 89.41, "grad_norm": 5.09375, "learning_rate": 5.824471884375471e-07, "loss": 1.1962, "step": 35764 }, { "epoch": 89.4125, "grad_norm": 9.375, "learning_rate": 5.821749217976303e-07, "loss": 1.2335, "step": 35765 }, { "epoch": 89.415, "grad_norm": 8.0625, "learning_rate": 5.819027169005987e-07, "loss": 1.2854, "step": 35766 }, { "epoch": 89.4175, "grad_norm": 8.5625, "learning_rate": 5.816305737482364e-07, "loss": 1.4693, "step": 35767 }, { "epoch": 89.42, "grad_norm": 6.96875, "learning_rate": 5.813584923423287e-07, "loss": 1.2674, "step": 35768 }, { "epoch": 89.4225, "grad_norm": 7.03125, "learning_rate": 5.810864726846554e-07, "loss": 1.3615, "step": 35769 }, { "epoch": 89.425, "grad_norm": 7.46875, "learning_rate": 5.808145147770017e-07, "loss": 1.3823, "step": 35770 }, { "epoch": 89.4275, "grad_norm": 5.34375, "learning_rate": 5.805426186211538e-07, "loss": 1.3857, "step": 35771 }, { "epoch": 89.43, "grad_norm": 7.21875, "learning_rate": 5.802707842188915e-07, "loss": 1.4379, "step": 35772 }, { "epoch": 89.4325, "grad_norm": 8.75, "learning_rate": 5.799990115719944e-07, "loss": 1.3155, "step": 35773 }, { "epoch": 89.435, "grad_norm": 10.1875, "learning_rate": 5.79727300682249e-07, "loss": 1.4702, "step": 35774 }, { "epoch": 89.4375, "grad_norm": 7.625, "learning_rate": 5.794556515514327e-07, "loss": 1.3319, "step": 35775 }, { "epoch": 89.44, "grad_norm": 7.78125, "learning_rate": 5.791840641813295e-07, "loss": 1.2509, "step": 35776 }, { "epoch": 89.4425, "grad_norm": 5.03125, "learning_rate": 5.789125385737182e-07, "loss": 1.2741, "step": 35777 }, { "epoch": 89.445, "grad_norm": 7.34375, "learning_rate": 5.786410747303794e-07, "loss": 1.5268, "step": 35778 }, { "epoch": 89.4475, "grad_norm": 6.875, "learning_rate": 5.783696726530929e-07, "loss": 1.374, "step": 35779 }, { "epoch": 89.45, "grad_norm": 8.25, "learning_rate": 5.780983323436374e-07, "loss": 1.3278, "step": 35780 }, { "epoch": 89.4525, "grad_norm": 8.6875, "learning_rate": 5.778270538037912e-07, "loss": 1.1491, "step": 35781 }, { "epoch": 89.455, "grad_norm": 7.90625, "learning_rate": 5.775558370353351e-07, "loss": 1.2055, "step": 35782 }, { "epoch": 89.4575, "grad_norm": 5.96875, "learning_rate": 5.772846820400469e-07, "loss": 1.397, "step": 35783 }, { "epoch": 89.46, "grad_norm": 5.15625, "learning_rate": 5.770135888197004e-07, "loss": 1.3614, "step": 35784 }, { "epoch": 89.4625, "grad_norm": 8.625, "learning_rate": 5.767425573760765e-07, "loss": 1.3027, "step": 35785 }, { "epoch": 89.465, "grad_norm": 7.90625, "learning_rate": 5.764715877109528e-07, "loss": 1.3123, "step": 35786 }, { "epoch": 89.4675, "grad_norm": 7.6875, "learning_rate": 5.762006798261044e-07, "loss": 1.4162, "step": 35787 }, { "epoch": 89.47, "grad_norm": 7.1875, "learning_rate": 5.759298337233054e-07, "loss": 1.2501, "step": 35788 }, { "epoch": 89.4725, "grad_norm": 9.3125, "learning_rate": 5.756590494043357e-07, "loss": 1.3547, "step": 35789 }, { "epoch": 89.475, "grad_norm": 7.8125, "learning_rate": 5.753883268709659e-07, "loss": 1.3958, "step": 35790 }, { "epoch": 89.4775, "grad_norm": 8.25, "learning_rate": 5.751176661249747e-07, "loss": 1.3726, "step": 35791 }, { "epoch": 89.48, "grad_norm": 7.1875, "learning_rate": 5.748470671681328e-07, "loss": 1.2406, "step": 35792 }, { "epoch": 89.4825, "grad_norm": 6.0, "learning_rate": 5.745765300022177e-07, "loss": 1.3816, "step": 35793 }, { "epoch": 89.485, "grad_norm": 9.5625, "learning_rate": 5.743060546290024e-07, "loss": 1.2009, "step": 35794 }, { "epoch": 89.4875, "grad_norm": 10.8125, "learning_rate": 5.740356410502579e-07, "loss": 1.4315, "step": 35795 }, { "epoch": 89.49, "grad_norm": 5.0625, "learning_rate": 5.737652892677592e-07, "loss": 1.3874, "step": 35796 }, { "epoch": 89.4925, "grad_norm": 6.5625, "learning_rate": 5.734949992832784e-07, "loss": 1.3711, "step": 35797 }, { "epoch": 89.495, "grad_norm": 6.59375, "learning_rate": 5.732247710985872e-07, "loss": 1.364, "step": 35798 }, { "epoch": 89.4975, "grad_norm": 7.40625, "learning_rate": 5.729546047154577e-07, "loss": 1.3741, "step": 35799 }, { "epoch": 89.5, "grad_norm": 10.75, "learning_rate": 5.726845001356573e-07, "loss": 1.2632, "step": 35800 }, { "epoch": 89.5025, "grad_norm": 8.8125, "learning_rate": 5.724144573609636e-07, "loss": 1.2284, "step": 35801 }, { "epoch": 89.505, "grad_norm": 8.0625, "learning_rate": 5.721444763931428e-07, "loss": 1.4104, "step": 35802 }, { "epoch": 89.5075, "grad_norm": 7.40625, "learning_rate": 5.718745572339634e-07, "loss": 1.3886, "step": 35803 }, { "epoch": 89.51, "grad_norm": 6.71875, "learning_rate": 5.716046998851999e-07, "loss": 1.2544, "step": 35804 }, { "epoch": 89.5125, "grad_norm": 8.25, "learning_rate": 5.713349043486172e-07, "loss": 1.2129, "step": 35805 }, { "epoch": 89.515, "grad_norm": 6.90625, "learning_rate": 5.710651706259862e-07, "loss": 1.2819, "step": 35806 }, { "epoch": 89.5175, "grad_norm": 8.0625, "learning_rate": 5.707954987190734e-07, "loss": 1.3449, "step": 35807 }, { "epoch": 89.52, "grad_norm": 7.75, "learning_rate": 5.705258886296494e-07, "loss": 1.2809, "step": 35808 }, { "epoch": 89.5225, "grad_norm": 9.375, "learning_rate": 5.702563403594808e-07, "loss": 1.2108, "step": 35809 }, { "epoch": 89.525, "grad_norm": 6.6875, "learning_rate": 5.699868539103326e-07, "loss": 1.2238, "step": 35810 }, { "epoch": 89.5275, "grad_norm": 5.625, "learning_rate": 5.697174292839725e-07, "loss": 1.3346, "step": 35811 }, { "epoch": 89.53, "grad_norm": 6.96875, "learning_rate": 5.6944806648217e-07, "loss": 1.3636, "step": 35812 }, { "epoch": 89.5325, "grad_norm": 7.75, "learning_rate": 5.691787655066883e-07, "loss": 1.2175, "step": 35813 }, { "epoch": 89.535, "grad_norm": 7.25, "learning_rate": 5.689095263592925e-07, "loss": 1.2699, "step": 35814 }, { "epoch": 89.5375, "grad_norm": 10.1875, "learning_rate": 5.68640349041747e-07, "loss": 1.191, "step": 35815 }, { "epoch": 89.54, "grad_norm": 5.8125, "learning_rate": 5.683712335558189e-07, "loss": 1.3102, "step": 35816 }, { "epoch": 89.5425, "grad_norm": 9.75, "learning_rate": 5.681021799032726e-07, "loss": 1.3545, "step": 35817 }, { "epoch": 89.545, "grad_norm": 8.125, "learning_rate": 5.678331880858711e-07, "loss": 1.527, "step": 35818 }, { "epoch": 89.5475, "grad_norm": 8.5625, "learning_rate": 5.675642581053764e-07, "loss": 1.4435, "step": 35819 }, { "epoch": 89.55, "grad_norm": 5.96875, "learning_rate": 5.672953899635524e-07, "loss": 1.4262, "step": 35820 }, { "epoch": 89.5525, "grad_norm": 7.4375, "learning_rate": 5.670265836621646e-07, "loss": 1.2801, "step": 35821 }, { "epoch": 89.555, "grad_norm": 6.46875, "learning_rate": 5.667578392029727e-07, "loss": 1.4328, "step": 35822 }, { "epoch": 89.5575, "grad_norm": 7.25, "learning_rate": 5.664891565877373e-07, "loss": 1.3375, "step": 35823 }, { "epoch": 89.56, "grad_norm": 8.3125, "learning_rate": 5.662205358182226e-07, "loss": 1.3452, "step": 35824 }, { "epoch": 89.5625, "grad_norm": 8.9375, "learning_rate": 5.659519768961885e-07, "loss": 1.2368, "step": 35825 }, { "epoch": 89.565, "grad_norm": 8.375, "learning_rate": 5.656834798233957e-07, "loss": 1.4087, "step": 35826 }, { "epoch": 89.5675, "grad_norm": 5.0, "learning_rate": 5.654150446016038e-07, "loss": 1.2772, "step": 35827 }, { "epoch": 89.57, "grad_norm": 9.75, "learning_rate": 5.651466712325737e-07, "loss": 1.3845, "step": 35828 }, { "epoch": 89.5725, "grad_norm": 7.0, "learning_rate": 5.648783597180651e-07, "loss": 1.3043, "step": 35829 }, { "epoch": 89.575, "grad_norm": 8.875, "learning_rate": 5.646101100598344e-07, "loss": 1.3636, "step": 35830 }, { "epoch": 89.5775, "grad_norm": 13.375, "learning_rate": 5.643419222596436e-07, "loss": 1.4544, "step": 35831 }, { "epoch": 89.58, "grad_norm": 8.5, "learning_rate": 5.640737963192511e-07, "loss": 1.5339, "step": 35832 }, { "epoch": 89.5825, "grad_norm": 6.6875, "learning_rate": 5.638057322404122e-07, "loss": 1.406, "step": 35833 }, { "epoch": 89.585, "grad_norm": 7.625, "learning_rate": 5.635377300248834e-07, "loss": 1.2238, "step": 35834 }, { "epoch": 89.5875, "grad_norm": 5.34375, "learning_rate": 5.632697896744243e-07, "loss": 1.344, "step": 35835 }, { "epoch": 89.59, "grad_norm": 5.9375, "learning_rate": 5.630019111907925e-07, "loss": 1.5269, "step": 35836 }, { "epoch": 89.5925, "grad_norm": 8.75, "learning_rate": 5.62734094575742e-07, "loss": 1.1608, "step": 35837 }, { "epoch": 89.595, "grad_norm": 7.28125, "learning_rate": 5.624663398310281e-07, "loss": 1.1909, "step": 35838 }, { "epoch": 89.5975, "grad_norm": 6.625, "learning_rate": 5.621986469584084e-07, "loss": 1.2792, "step": 35839 }, { "epoch": 89.6, "grad_norm": 7.15625, "learning_rate": 5.619310159596358e-07, "loss": 1.3476, "step": 35840 }, { "epoch": 89.6025, "grad_norm": 5.90625, "learning_rate": 5.616634468364668e-07, "loss": 1.3012, "step": 35841 }, { "epoch": 89.605, "grad_norm": 8.5, "learning_rate": 5.613959395906531e-07, "loss": 1.3927, "step": 35842 }, { "epoch": 89.6075, "grad_norm": 7.1875, "learning_rate": 5.611284942239514e-07, "loss": 1.2175, "step": 35843 }, { "epoch": 89.61, "grad_norm": 6.28125, "learning_rate": 5.608611107381135e-07, "loss": 1.351, "step": 35844 }, { "epoch": 89.6125, "grad_norm": 5.90625, "learning_rate": 5.6059378913489e-07, "loss": 1.316, "step": 35845 }, { "epoch": 89.615, "grad_norm": 6.46875, "learning_rate": 5.603265294160376e-07, "loss": 1.3727, "step": 35846 }, { "epoch": 89.6175, "grad_norm": 7.71875, "learning_rate": 5.600593315833069e-07, "loss": 1.2786, "step": 35847 }, { "epoch": 89.62, "grad_norm": 7.90625, "learning_rate": 5.5979219563845e-07, "loss": 1.4479, "step": 35848 }, { "epoch": 89.6225, "grad_norm": 5.71875, "learning_rate": 5.595251215832165e-07, "loss": 1.2891, "step": 35849 }, { "epoch": 89.625, "grad_norm": 10.0, "learning_rate": 5.592581094193584e-07, "loss": 1.2981, "step": 35850 }, { "epoch": 89.6275, "grad_norm": 6.46875, "learning_rate": 5.589911591486275e-07, "loss": 1.4523, "step": 35851 }, { "epoch": 89.63, "grad_norm": 8.125, "learning_rate": 5.587242707727725e-07, "loss": 1.2904, "step": 35852 }, { "epoch": 89.6325, "grad_norm": 7.5, "learning_rate": 5.584574442935419e-07, "loss": 1.4232, "step": 35853 }, { "epoch": 89.635, "grad_norm": 6.625, "learning_rate": 5.58190679712689e-07, "loss": 1.2413, "step": 35854 }, { "epoch": 89.6375, "grad_norm": 11.625, "learning_rate": 5.579239770319577e-07, "loss": 1.3213, "step": 35855 }, { "epoch": 89.64, "grad_norm": 8.3125, "learning_rate": 5.576573362531001e-07, "loss": 1.2302, "step": 35856 }, { "epoch": 89.6425, "grad_norm": 8.3125, "learning_rate": 5.573907573778625e-07, "loss": 1.2633, "step": 35857 }, { "epoch": 89.645, "grad_norm": 6.84375, "learning_rate": 5.571242404079946e-07, "loss": 1.3059, "step": 35858 }, { "epoch": 89.6475, "grad_norm": 7.875, "learning_rate": 5.568577853452417e-07, "loss": 1.2251, "step": 35859 }, { "epoch": 89.65, "grad_norm": 7.4375, "learning_rate": 5.565913921913513e-07, "loss": 1.5214, "step": 35860 }, { "epoch": 89.6525, "grad_norm": 8.0625, "learning_rate": 5.563250609480686e-07, "loss": 1.2877, "step": 35861 }, { "epoch": 89.655, "grad_norm": 6.6875, "learning_rate": 5.560587916171422e-07, "loss": 1.3332, "step": 35862 }, { "epoch": 89.6575, "grad_norm": 9.0, "learning_rate": 5.557925842003164e-07, "loss": 1.4023, "step": 35863 }, { "epoch": 89.66, "grad_norm": 5.375, "learning_rate": 5.555264386993353e-07, "loss": 1.2767, "step": 35864 }, { "epoch": 89.6625, "grad_norm": 10.75, "learning_rate": 5.552603551159452e-07, "loss": 1.335, "step": 35865 }, { "epoch": 89.665, "grad_norm": 7.9375, "learning_rate": 5.549943334518915e-07, "loss": 1.3876, "step": 35866 }, { "epoch": 89.6675, "grad_norm": 7.71875, "learning_rate": 5.547283737089171e-07, "loss": 1.2667, "step": 35867 }, { "epoch": 89.67, "grad_norm": 9.875, "learning_rate": 5.54462475888764e-07, "loss": 1.276, "step": 35868 }, { "epoch": 89.6725, "grad_norm": 11.0, "learning_rate": 5.541966399931786e-07, "loss": 1.2913, "step": 35869 }, { "epoch": 89.675, "grad_norm": 9.5625, "learning_rate": 5.539308660239007e-07, "loss": 1.2532, "step": 35870 }, { "epoch": 89.6775, "grad_norm": 10.375, "learning_rate": 5.536651539826743e-07, "loss": 1.2629, "step": 35871 }, { "epoch": 89.68, "grad_norm": 8.5, "learning_rate": 5.533995038712403e-07, "loss": 1.2334, "step": 35872 }, { "epoch": 89.6825, "grad_norm": 5.96875, "learning_rate": 5.531339156913429e-07, "loss": 1.338, "step": 35873 }, { "epoch": 89.685, "grad_norm": 9.0, "learning_rate": 5.528683894447218e-07, "loss": 1.3672, "step": 35874 }, { "epoch": 89.6875, "grad_norm": 8.5, "learning_rate": 5.526029251331155e-07, "loss": 1.3527, "step": 35875 }, { "epoch": 89.69, "grad_norm": 10.4375, "learning_rate": 5.523375227582662e-07, "loss": 1.2952, "step": 35876 }, { "epoch": 89.6925, "grad_norm": 8.125, "learning_rate": 5.520721823219155e-07, "loss": 1.3717, "step": 35877 }, { "epoch": 89.695, "grad_norm": 5.9375, "learning_rate": 5.518069038258023e-07, "loss": 1.1902, "step": 35878 }, { "epoch": 89.6975, "grad_norm": 5.8125, "learning_rate": 5.515416872716628e-07, "loss": 1.1605, "step": 35879 }, { "epoch": 89.7, "grad_norm": 8.875, "learning_rate": 5.51276532661238e-07, "loss": 1.3107, "step": 35880 }, { "epoch": 89.7025, "grad_norm": 5.8125, "learning_rate": 5.510114399962674e-07, "loss": 1.226, "step": 35881 }, { "epoch": 89.705, "grad_norm": 10.0, "learning_rate": 5.507464092784876e-07, "loss": 1.3828, "step": 35882 }, { "epoch": 89.7075, "grad_norm": 8.4375, "learning_rate": 5.504814405096348e-07, "loss": 1.325, "step": 35883 }, { "epoch": 89.71, "grad_norm": 9.625, "learning_rate": 5.5021653369145e-07, "loss": 1.3802, "step": 35884 }, { "epoch": 89.7125, "grad_norm": 11.75, "learning_rate": 5.499516888256651e-07, "loss": 1.2394, "step": 35885 }, { "epoch": 89.715, "grad_norm": 4.84375, "learning_rate": 5.496869059140197e-07, "loss": 1.4317, "step": 35886 }, { "epoch": 89.7175, "grad_norm": 9.5, "learning_rate": 5.494221849582481e-07, "loss": 1.312, "step": 35887 }, { "epoch": 89.72, "grad_norm": 7.6875, "learning_rate": 5.491575259600879e-07, "loss": 1.4772, "step": 35888 }, { "epoch": 89.7225, "grad_norm": 7.125, "learning_rate": 5.488929289212719e-07, "loss": 1.1326, "step": 35889 }, { "epoch": 89.725, "grad_norm": 11.125, "learning_rate": 5.486283938435356e-07, "loss": 1.3118, "step": 35890 }, { "epoch": 89.7275, "grad_norm": 10.0, "learning_rate": 5.483639207286118e-07, "loss": 1.412, "step": 35891 }, { "epoch": 89.73, "grad_norm": 8.1875, "learning_rate": 5.480995095782382e-07, "loss": 1.33, "step": 35892 }, { "epoch": 89.7325, "grad_norm": 7.40625, "learning_rate": 5.478351603941467e-07, "loss": 1.3611, "step": 35893 }, { "epoch": 89.735, "grad_norm": 11.1875, "learning_rate": 5.475708731780671e-07, "loss": 1.2813, "step": 35894 }, { "epoch": 89.7375, "grad_norm": 7.65625, "learning_rate": 5.473066479317346e-07, "loss": 1.3233, "step": 35895 }, { "epoch": 89.74, "grad_norm": 6.90625, "learning_rate": 5.470424846568833e-07, "loss": 1.1328, "step": 35896 }, { "epoch": 89.7425, "grad_norm": 9.8125, "learning_rate": 5.467783833552431e-07, "loss": 1.3805, "step": 35897 }, { "epoch": 89.745, "grad_norm": 7.28125, "learning_rate": 5.465143440285437e-07, "loss": 1.316, "step": 35898 }, { "epoch": 89.7475, "grad_norm": 7.03125, "learning_rate": 5.462503666785202e-07, "loss": 1.4606, "step": 35899 }, { "epoch": 89.75, "grad_norm": 7.5625, "learning_rate": 5.459864513068991e-07, "loss": 1.2605, "step": 35900 }, { "epoch": 89.7525, "grad_norm": 6.1875, "learning_rate": 5.457225979154135e-07, "loss": 1.3201, "step": 35901 }, { "epoch": 89.755, "grad_norm": 8.75, "learning_rate": 5.454588065057919e-07, "loss": 1.3419, "step": 35902 }, { "epoch": 89.7575, "grad_norm": 9.3125, "learning_rate": 5.451950770797643e-07, "loss": 1.3231, "step": 35903 }, { "epoch": 89.76, "grad_norm": 7.875, "learning_rate": 5.449314096390601e-07, "loss": 1.3527, "step": 35904 }, { "epoch": 89.7625, "grad_norm": 7.53125, "learning_rate": 5.446678041854059e-07, "loss": 1.453, "step": 35905 }, { "epoch": 89.765, "grad_norm": 8.75, "learning_rate": 5.444042607205302e-07, "loss": 1.2373, "step": 35906 }, { "epoch": 89.7675, "grad_norm": 6.15625, "learning_rate": 5.441407792461639e-07, "loss": 1.2789, "step": 35907 }, { "epoch": 89.77, "grad_norm": 5.75, "learning_rate": 5.438773597640323e-07, "loss": 1.2924, "step": 35908 }, { "epoch": 89.7725, "grad_norm": 7.90625, "learning_rate": 5.436140022758607e-07, "loss": 1.3496, "step": 35909 }, { "epoch": 89.775, "grad_norm": 7.09375, "learning_rate": 5.433507067833777e-07, "loss": 1.3602, "step": 35910 }, { "epoch": 89.7775, "grad_norm": 7.34375, "learning_rate": 5.430874732883096e-07, "loss": 1.394, "step": 35911 }, { "epoch": 89.78, "grad_norm": 7.875, "learning_rate": 5.428243017923817e-07, "loss": 1.3673, "step": 35912 }, { "epoch": 89.7825, "grad_norm": 6.0, "learning_rate": 5.425611922973185e-07, "loss": 1.2545, "step": 35913 }, { "epoch": 89.785, "grad_norm": 6.75, "learning_rate": 5.42298144804847e-07, "loss": 1.3234, "step": 35914 }, { "epoch": 89.7875, "grad_norm": 7.375, "learning_rate": 5.420351593166884e-07, "loss": 1.3451, "step": 35915 }, { "epoch": 89.79, "grad_norm": 6.6875, "learning_rate": 5.417722358345712e-07, "loss": 1.3702, "step": 35916 }, { "epoch": 89.7925, "grad_norm": 10.6875, "learning_rate": 5.41509374360214e-07, "loss": 1.5036, "step": 35917 }, { "epoch": 89.795, "grad_norm": 8.0, "learning_rate": 5.412465748953454e-07, "loss": 1.3058, "step": 35918 }, { "epoch": 89.7975, "grad_norm": 7.46875, "learning_rate": 5.409838374416853e-07, "loss": 1.2673, "step": 35919 }, { "epoch": 89.8, "grad_norm": 5.125, "learning_rate": 5.407211620009545e-07, "loss": 1.2728, "step": 35920 }, { "epoch": 89.8025, "grad_norm": 4.5, "learning_rate": 5.404585485748781e-07, "loss": 1.2908, "step": 35921 }, { "epoch": 89.805, "grad_norm": 6.40625, "learning_rate": 5.401959971651782e-07, "loss": 1.4407, "step": 35922 }, { "epoch": 89.8075, "grad_norm": 12.625, "learning_rate": 5.399335077735746e-07, "loss": 1.468, "step": 35923 }, { "epoch": 89.81, "grad_norm": 7.46875, "learning_rate": 5.396710804017868e-07, "loss": 1.357, "step": 35924 }, { "epoch": 89.8125, "grad_norm": 4.625, "learning_rate": 5.39408715051537e-07, "loss": 1.3063, "step": 35925 }, { "epoch": 89.815, "grad_norm": 7.125, "learning_rate": 5.391464117245471e-07, "loss": 1.2843, "step": 35926 }, { "epoch": 89.8175, "grad_norm": 6.125, "learning_rate": 5.388841704225344e-07, "loss": 1.4097, "step": 35927 }, { "epoch": 89.82, "grad_norm": 7.5625, "learning_rate": 5.386219911472168e-07, "loss": 1.4057, "step": 35928 }, { "epoch": 89.8225, "grad_norm": 8.0625, "learning_rate": 5.38359873900316e-07, "loss": 1.2444, "step": 35929 }, { "epoch": 89.825, "grad_norm": 8.6875, "learning_rate": 5.380978186835484e-07, "loss": 1.358, "step": 35930 }, { "epoch": 89.8275, "grad_norm": 6.375, "learning_rate": 5.378358254986338e-07, "loss": 1.2516, "step": 35931 }, { "epoch": 89.83, "grad_norm": 10.625, "learning_rate": 5.375738943472874e-07, "loss": 1.3585, "step": 35932 }, { "epoch": 89.8325, "grad_norm": 6.3125, "learning_rate": 5.373120252312291e-07, "loss": 1.3153, "step": 35933 }, { "epoch": 89.835, "grad_norm": 5.25, "learning_rate": 5.370502181521752e-07, "loss": 1.4067, "step": 35934 }, { "epoch": 89.8375, "grad_norm": 4.96875, "learning_rate": 5.367884731118389e-07, "loss": 1.4136, "step": 35935 }, { "epoch": 89.84, "grad_norm": 6.53125, "learning_rate": 5.365267901119398e-07, "loss": 1.3024, "step": 35936 }, { "epoch": 89.8425, "grad_norm": 6.96875, "learning_rate": 5.362651691541932e-07, "loss": 1.4492, "step": 35937 }, { "epoch": 89.845, "grad_norm": 8.0625, "learning_rate": 5.360036102403144e-07, "loss": 1.2208, "step": 35938 }, { "epoch": 89.8475, "grad_norm": 8.375, "learning_rate": 5.357421133720164e-07, "loss": 1.4877, "step": 35939 }, { "epoch": 89.85, "grad_norm": 6.8125, "learning_rate": 5.354806785510113e-07, "loss": 1.2856, "step": 35940 }, { "epoch": 89.8525, "grad_norm": 10.125, "learning_rate": 5.35219305779019e-07, "loss": 1.2462, "step": 35941 }, { "epoch": 89.855, "grad_norm": 7.625, "learning_rate": 5.349579950577499e-07, "loss": 1.3514, "step": 35942 }, { "epoch": 89.8575, "grad_norm": 8.875, "learning_rate": 5.346967463889174e-07, "loss": 1.472, "step": 35943 }, { "epoch": 89.86, "grad_norm": 8.6875, "learning_rate": 5.344355597742345e-07, "loss": 1.2391, "step": 35944 }, { "epoch": 89.8625, "grad_norm": 6.65625, "learning_rate": 5.341744352154121e-07, "loss": 1.2411, "step": 35945 }, { "epoch": 89.865, "grad_norm": 10.125, "learning_rate": 5.339133727141643e-07, "loss": 1.3733, "step": 35946 }, { "epoch": 89.8675, "grad_norm": 11.9375, "learning_rate": 5.336523722722009e-07, "loss": 1.1854, "step": 35947 }, { "epoch": 89.87, "grad_norm": 6.46875, "learning_rate": 5.333914338912349e-07, "loss": 1.3239, "step": 35948 }, { "epoch": 89.8725, "grad_norm": 7.40625, "learning_rate": 5.331305575729762e-07, "loss": 1.4186, "step": 35949 }, { "epoch": 89.875, "grad_norm": 6.03125, "learning_rate": 5.328697433191321e-07, "loss": 1.251, "step": 35950 }, { "epoch": 89.8775, "grad_norm": 7.84375, "learning_rate": 5.326089911314159e-07, "loss": 1.3759, "step": 35951 }, { "epoch": 89.88, "grad_norm": 9.125, "learning_rate": 5.323483010115382e-07, "loss": 1.3394, "step": 35952 }, { "epoch": 89.8825, "grad_norm": 6.59375, "learning_rate": 5.320876729612057e-07, "loss": 1.3497, "step": 35953 }, { "epoch": 89.885, "grad_norm": 10.875, "learning_rate": 5.318271069821268e-07, "loss": 1.3816, "step": 35954 }, { "epoch": 89.8875, "grad_norm": 7.5625, "learning_rate": 5.315666030760091e-07, "loss": 1.3012, "step": 35955 }, { "epoch": 89.89, "grad_norm": 8.0625, "learning_rate": 5.313061612445636e-07, "loss": 1.2721, "step": 35956 }, { "epoch": 89.8925, "grad_norm": 7.0, "learning_rate": 5.310457814894965e-07, "loss": 1.2241, "step": 35957 }, { "epoch": 89.895, "grad_norm": 10.1875, "learning_rate": 5.307854638125143e-07, "loss": 1.3135, "step": 35958 }, { "epoch": 89.8975, "grad_norm": 8.5, "learning_rate": 5.305252082153222e-07, "loss": 1.3041, "step": 35959 }, { "epoch": 89.9, "grad_norm": 7.09375, "learning_rate": 5.30265014699628e-07, "loss": 1.3465, "step": 35960 }, { "epoch": 89.9025, "grad_norm": 6.125, "learning_rate": 5.300048832671378e-07, "loss": 1.3463, "step": 35961 }, { "epoch": 89.905, "grad_norm": 7.40625, "learning_rate": 5.297448139195583e-07, "loss": 1.2202, "step": 35962 }, { "epoch": 89.9075, "grad_norm": 7.0, "learning_rate": 5.2948480665859e-07, "loss": 1.5093, "step": 35963 }, { "epoch": 89.91, "grad_norm": 10.75, "learning_rate": 5.292248614859419e-07, "loss": 1.4397, "step": 35964 }, { "epoch": 89.9125, "grad_norm": 8.6875, "learning_rate": 5.289649784033157e-07, "loss": 1.3071, "step": 35965 }, { "epoch": 89.915, "grad_norm": 6.8125, "learning_rate": 5.28705157412418e-07, "loss": 1.2456, "step": 35966 }, { "epoch": 89.9175, "grad_norm": 6.21875, "learning_rate": 5.284453985149474e-07, "loss": 1.3432, "step": 35967 }, { "epoch": 89.92, "grad_norm": 7.375, "learning_rate": 5.281857017126124e-07, "loss": 1.4625, "step": 35968 }, { "epoch": 89.9225, "grad_norm": 7.96875, "learning_rate": 5.279260670071129e-07, "loss": 1.1864, "step": 35969 }, { "epoch": 89.925, "grad_norm": 6.28125, "learning_rate": 5.276664944001475e-07, "loss": 1.3331, "step": 35970 }, { "epoch": 89.9275, "grad_norm": 7.84375, "learning_rate": 5.274069838934248e-07, "loss": 1.2704, "step": 35971 }, { "epoch": 89.93, "grad_norm": 7.78125, "learning_rate": 5.271475354886424e-07, "loss": 1.239, "step": 35972 }, { "epoch": 89.9325, "grad_norm": 8.0, "learning_rate": 5.268881491875022e-07, "loss": 1.4016, "step": 35973 }, { "epoch": 89.935, "grad_norm": 7.28125, "learning_rate": 5.266288249917029e-07, "loss": 1.3749, "step": 35974 }, { "epoch": 89.9375, "grad_norm": 10.5625, "learning_rate": 5.263695629029452e-07, "loss": 1.2693, "step": 35975 }, { "epoch": 89.94, "grad_norm": 5.90625, "learning_rate": 5.261103629229314e-07, "loss": 1.4709, "step": 35976 }, { "epoch": 89.9425, "grad_norm": 8.5625, "learning_rate": 5.258512250533587e-07, "loss": 1.347, "step": 35977 }, { "epoch": 89.945, "grad_norm": 6.0625, "learning_rate": 5.255921492959248e-07, "loss": 1.5437, "step": 35978 }, { "epoch": 89.9475, "grad_norm": 6.1875, "learning_rate": 5.253331356523317e-07, "loss": 1.3483, "step": 35979 }, { "epoch": 89.95, "grad_norm": 7.65625, "learning_rate": 5.250741841242735e-07, "loss": 1.3168, "step": 35980 }, { "epoch": 89.9525, "grad_norm": 8.0, "learning_rate": 5.24815294713451e-07, "loss": 1.315, "step": 35981 }, { "epoch": 89.955, "grad_norm": 8.75, "learning_rate": 5.245564674215586e-07, "loss": 1.4976, "step": 35982 }, { "epoch": 89.9575, "grad_norm": 11.375, "learning_rate": 5.242977022502971e-07, "loss": 1.3723, "step": 35983 }, { "epoch": 89.96, "grad_norm": 7.4375, "learning_rate": 5.240389992013606e-07, "loss": 1.2136, "step": 35984 }, { "epoch": 89.9625, "grad_norm": 7.21875, "learning_rate": 5.237803582764434e-07, "loss": 1.3718, "step": 35985 }, { "epoch": 89.965, "grad_norm": 7.6875, "learning_rate": 5.235217794772429e-07, "loss": 1.281, "step": 35986 }, { "epoch": 89.9675, "grad_norm": 7.96875, "learning_rate": 5.232632628054557e-07, "loss": 1.4638, "step": 35987 }, { "epoch": 89.97, "grad_norm": 6.875, "learning_rate": 5.230048082627748e-07, "loss": 1.3544, "step": 35988 }, { "epoch": 89.9725, "grad_norm": 5.71875, "learning_rate": 5.227464158508954e-07, "loss": 1.3505, "step": 35989 }, { "epoch": 89.975, "grad_norm": 9.0, "learning_rate": 5.224880855715098e-07, "loss": 1.1591, "step": 35990 }, { "epoch": 89.9775, "grad_norm": 9.0625, "learning_rate": 5.222298174263152e-07, "loss": 1.5472, "step": 35991 }, { "epoch": 89.98, "grad_norm": 6.3125, "learning_rate": 5.219716114170026e-07, "loss": 1.3758, "step": 35992 }, { "epoch": 89.9825, "grad_norm": 5.9375, "learning_rate": 5.217134675452629e-07, "loss": 1.2561, "step": 35993 }, { "epoch": 89.985, "grad_norm": 5.0, "learning_rate": 5.214553858127913e-07, "loss": 1.1972, "step": 35994 }, { "epoch": 89.9875, "grad_norm": 8.1875, "learning_rate": 5.211973662212788e-07, "loss": 1.263, "step": 35995 }, { "epoch": 89.99, "grad_norm": 7.46875, "learning_rate": 5.209394087724174e-07, "loss": 1.4073, "step": 35996 }, { "epoch": 89.9925, "grad_norm": 5.21875, "learning_rate": 5.206815134678977e-07, "loss": 1.3146, "step": 35997 }, { "epoch": 89.995, "grad_norm": 7.90625, "learning_rate": 5.204236803094109e-07, "loss": 1.4089, "step": 35998 }, { "epoch": 89.9975, "grad_norm": 8.75, "learning_rate": 5.201659092986467e-07, "loss": 1.2265, "step": 35999 }, { "epoch": 90.0, "grad_norm": 6.90625, "learning_rate": 5.199082004372958e-07, "loss": 1.2652, "step": 36000 }, { "epoch": 90.0025, "grad_norm": 8.75, "learning_rate": 5.196505537270469e-07, "loss": 1.2789, "step": 36001 }, { "epoch": 90.005, "grad_norm": 6.65625, "learning_rate": 5.193929691695898e-07, "loss": 1.2836, "step": 36002 }, { "epoch": 90.0075, "grad_norm": 7.0625, "learning_rate": 5.191354467666143e-07, "loss": 1.2752, "step": 36003 }, { "epoch": 90.01, "grad_norm": 7.84375, "learning_rate": 5.188779865198057e-07, "loss": 1.2292, "step": 36004 }, { "epoch": 90.0125, "grad_norm": 6.8125, "learning_rate": 5.186205884308527e-07, "loss": 1.3064, "step": 36005 }, { "epoch": 90.015, "grad_norm": 8.375, "learning_rate": 5.183632525014459e-07, "loss": 1.4594, "step": 36006 }, { "epoch": 90.0175, "grad_norm": 9.0, "learning_rate": 5.181059787332698e-07, "loss": 1.3099, "step": 36007 }, { "epoch": 90.02, "grad_norm": 8.0625, "learning_rate": 5.178487671280108e-07, "loss": 1.3329, "step": 36008 }, { "epoch": 90.0225, "grad_norm": 6.1875, "learning_rate": 5.175916176873552e-07, "loss": 1.2451, "step": 36009 }, { "epoch": 90.025, "grad_norm": 6.375, "learning_rate": 5.173345304129918e-07, "loss": 1.3898, "step": 36010 }, { "epoch": 90.0275, "grad_norm": 7.4375, "learning_rate": 5.170775053066023e-07, "loss": 1.2402, "step": 36011 }, { "epoch": 90.03, "grad_norm": 7.90625, "learning_rate": 5.168205423698724e-07, "loss": 1.2909, "step": 36012 }, { "epoch": 90.0325, "grad_norm": 7.71875, "learning_rate": 5.165636416044895e-07, "loss": 1.37, "step": 36013 }, { "epoch": 90.035, "grad_norm": 7.5, "learning_rate": 5.163068030121332e-07, "loss": 1.4174, "step": 36014 }, { "epoch": 90.0375, "grad_norm": 8.4375, "learning_rate": 5.160500265944923e-07, "loss": 1.348, "step": 36015 }, { "epoch": 90.04, "grad_norm": 7.21875, "learning_rate": 5.157933123532466e-07, "loss": 1.332, "step": 36016 }, { "epoch": 90.0425, "grad_norm": 9.875, "learning_rate": 5.155366602900813e-07, "loss": 1.3077, "step": 36017 }, { "epoch": 90.045, "grad_norm": 6.34375, "learning_rate": 5.152800704066785e-07, "loss": 1.2499, "step": 36018 }, { "epoch": 90.0475, "grad_norm": 6.0625, "learning_rate": 5.15023542704719e-07, "loss": 1.2672, "step": 36019 }, { "epoch": 90.05, "grad_norm": 6.59375, "learning_rate": 5.147670771858848e-07, "loss": 1.3202, "step": 36020 }, { "epoch": 90.0525, "grad_norm": 5.4375, "learning_rate": 5.145106738518602e-07, "loss": 1.2289, "step": 36021 }, { "epoch": 90.055, "grad_norm": 9.9375, "learning_rate": 5.142543327043225e-07, "loss": 1.3186, "step": 36022 }, { "epoch": 90.0575, "grad_norm": 9.75, "learning_rate": 5.13998053744954e-07, "loss": 1.3797, "step": 36023 }, { "epoch": 90.06, "grad_norm": 9.4375, "learning_rate": 5.137418369754343e-07, "loss": 1.5112, "step": 36024 }, { "epoch": 90.0625, "grad_norm": 7.3125, "learning_rate": 5.134856823974444e-07, "loss": 1.2773, "step": 36025 }, { "epoch": 90.065, "grad_norm": 6.3125, "learning_rate": 5.132295900126627e-07, "loss": 1.2059, "step": 36026 }, { "epoch": 90.0675, "grad_norm": 5.03125, "learning_rate": 5.12973559822767e-07, "loss": 1.0889, "step": 36027 }, { "epoch": 90.07, "grad_norm": 7.96875, "learning_rate": 5.127175918294392e-07, "loss": 1.3537, "step": 36028 }, { "epoch": 90.0725, "grad_norm": 8.0, "learning_rate": 5.124616860343523e-07, "loss": 1.284, "step": 36029 }, { "epoch": 90.075, "grad_norm": 6.1875, "learning_rate": 5.122058424391896e-07, "loss": 1.2977, "step": 36030 }, { "epoch": 90.0775, "grad_norm": 6.625, "learning_rate": 5.119500610456241e-07, "loss": 1.5717, "step": 36031 }, { "epoch": 90.08, "grad_norm": 6.0, "learning_rate": 5.116943418553355e-07, "loss": 1.2787, "step": 36032 }, { "epoch": 90.0825, "grad_norm": 5.09375, "learning_rate": 5.114386848699992e-07, "loss": 1.2707, "step": 36033 }, { "epoch": 90.085, "grad_norm": 7.15625, "learning_rate": 5.111830900912895e-07, "loss": 1.4494, "step": 36034 }, { "epoch": 90.0875, "grad_norm": 5.6875, "learning_rate": 5.109275575208838e-07, "loss": 1.4743, "step": 36035 }, { "epoch": 90.09, "grad_norm": 8.75, "learning_rate": 5.106720871604586e-07, "loss": 1.1579, "step": 36036 }, { "epoch": 90.0925, "grad_norm": 8.375, "learning_rate": 5.104166790116871e-07, "loss": 1.2953, "step": 36037 }, { "epoch": 90.095, "grad_norm": 8.625, "learning_rate": 5.101613330762434e-07, "loss": 1.3211, "step": 36038 }, { "epoch": 90.0975, "grad_norm": 8.0, "learning_rate": 5.099060493558017e-07, "loss": 1.379, "step": 36039 }, { "epoch": 90.1, "grad_norm": 9.5625, "learning_rate": 5.096508278520385e-07, "loss": 1.3864, "step": 36040 }, { "epoch": 90.1025, "grad_norm": 7.25, "learning_rate": 5.093956685666234e-07, "loss": 1.3851, "step": 36041 }, { "epoch": 90.105, "grad_norm": 7.21875, "learning_rate": 5.091405715012309e-07, "loss": 1.3426, "step": 36042 }, { "epoch": 90.1075, "grad_norm": 8.375, "learning_rate": 5.088855366575329e-07, "loss": 1.3113, "step": 36043 }, { "epoch": 90.11, "grad_norm": 7.125, "learning_rate": 5.086305640372002e-07, "loss": 1.2205, "step": 36044 }, { "epoch": 90.1125, "grad_norm": 9.4375, "learning_rate": 5.083756536419082e-07, "loss": 1.275, "step": 36045 }, { "epoch": 90.115, "grad_norm": 6.90625, "learning_rate": 5.081208054733233e-07, "loss": 1.3561, "step": 36046 }, { "epoch": 90.1175, "grad_norm": 9.5625, "learning_rate": 5.078660195331198e-07, "loss": 1.3771, "step": 36047 }, { "epoch": 90.12, "grad_norm": 6.3125, "learning_rate": 5.076112958229673e-07, "loss": 1.3343, "step": 36048 }, { "epoch": 90.1225, "grad_norm": 8.375, "learning_rate": 5.073566343445335e-07, "loss": 1.4226, "step": 36049 }, { "epoch": 90.125, "grad_norm": 5.71875, "learning_rate": 5.071020350994893e-07, "loss": 1.2277, "step": 36050 }, { "epoch": 90.1275, "grad_norm": 8.1875, "learning_rate": 5.068474980895066e-07, "loss": 1.2702, "step": 36051 }, { "epoch": 90.13, "grad_norm": 8.5625, "learning_rate": 5.06593023316252e-07, "loss": 1.2797, "step": 36052 }, { "epoch": 90.1325, "grad_norm": 6.625, "learning_rate": 5.063386107813917e-07, "loss": 1.2055, "step": 36053 }, { "epoch": 90.135, "grad_norm": 9.75, "learning_rate": 5.060842604865956e-07, "loss": 1.3437, "step": 36054 }, { "epoch": 90.1375, "grad_norm": 6.78125, "learning_rate": 5.058299724335325e-07, "loss": 1.356, "step": 36055 }, { "epoch": 90.14, "grad_norm": 7.25, "learning_rate": 5.055757466238676e-07, "loss": 1.3613, "step": 36056 }, { "epoch": 90.1425, "grad_norm": 7.0, "learning_rate": 5.053215830592672e-07, "loss": 1.4543, "step": 36057 }, { "epoch": 90.145, "grad_norm": 9.1875, "learning_rate": 5.050674817414003e-07, "loss": 1.4403, "step": 36058 }, { "epoch": 90.1475, "grad_norm": 7.53125, "learning_rate": 5.048134426719298e-07, "loss": 1.2935, "step": 36059 }, { "epoch": 90.15, "grad_norm": 5.59375, "learning_rate": 5.045594658525232e-07, "loss": 1.352, "step": 36060 }, { "epoch": 90.1525, "grad_norm": 9.8125, "learning_rate": 5.043055512848438e-07, "loss": 1.3092, "step": 36061 }, { "epoch": 90.155, "grad_norm": 6.65625, "learning_rate": 5.04051698970558e-07, "loss": 1.3163, "step": 36062 }, { "epoch": 90.1575, "grad_norm": 8.3125, "learning_rate": 5.037979089113298e-07, "loss": 1.5396, "step": 36063 }, { "epoch": 90.16, "grad_norm": 6.5625, "learning_rate": 5.035441811088204e-07, "loss": 1.2161, "step": 36064 }, { "epoch": 90.1625, "grad_norm": 7.125, "learning_rate": 5.032905155646961e-07, "loss": 1.1793, "step": 36065 }, { "epoch": 90.165, "grad_norm": 10.0, "learning_rate": 5.030369122806201e-07, "loss": 1.3251, "step": 36066 }, { "epoch": 90.1675, "grad_norm": 8.75, "learning_rate": 5.027833712582542e-07, "loss": 1.5473, "step": 36067 }, { "epoch": 90.17, "grad_norm": 5.5, "learning_rate": 5.025298924992594e-07, "loss": 1.44, "step": 36068 }, { "epoch": 90.1725, "grad_norm": 8.5625, "learning_rate": 5.02276476005299e-07, "loss": 1.2906, "step": 36069 }, { "epoch": 90.175, "grad_norm": 8.0625, "learning_rate": 5.020231217780336e-07, "loss": 1.3348, "step": 36070 }, { "epoch": 90.1775, "grad_norm": 10.6875, "learning_rate": 5.017698298191264e-07, "loss": 1.3033, "step": 36071 }, { "epoch": 90.18, "grad_norm": 8.1875, "learning_rate": 5.01516600130233e-07, "loss": 1.3888, "step": 36072 }, { "epoch": 90.1825, "grad_norm": 9.75, "learning_rate": 5.012634327130195e-07, "loss": 1.4757, "step": 36073 }, { "epoch": 90.185, "grad_norm": 8.9375, "learning_rate": 5.010103275691402e-07, "loss": 1.4039, "step": 36074 }, { "epoch": 90.1875, "grad_norm": 8.25, "learning_rate": 5.007572847002595e-07, "loss": 1.3888, "step": 36075 }, { "epoch": 90.19, "grad_norm": 6.875, "learning_rate": 5.005043041080327e-07, "loss": 1.275, "step": 36076 }, { "epoch": 90.1925, "grad_norm": 12.4375, "learning_rate": 5.002513857941194e-07, "loss": 1.3461, "step": 36077 }, { "epoch": 90.195, "grad_norm": 15.625, "learning_rate": 4.999985297601795e-07, "loss": 1.4896, "step": 36078 }, { "epoch": 90.1975, "grad_norm": 10.125, "learning_rate": 4.997457360078661e-07, "loss": 1.4171, "step": 36079 }, { "epoch": 90.2, "grad_norm": 7.90625, "learning_rate": 4.994930045388414e-07, "loss": 1.3154, "step": 36080 }, { "epoch": 90.2025, "grad_norm": 6.09375, "learning_rate": 4.992403353547604e-07, "loss": 1.2758, "step": 36081 }, { "epoch": 90.205, "grad_norm": 6.34375, "learning_rate": 4.989877284572797e-07, "loss": 1.3508, "step": 36082 }, { "epoch": 90.2075, "grad_norm": 9.0625, "learning_rate": 4.987351838480536e-07, "loss": 1.2418, "step": 36083 }, { "epoch": 90.21, "grad_norm": 9.25, "learning_rate": 4.984827015287397e-07, "loss": 1.3177, "step": 36084 }, { "epoch": 90.2125, "grad_norm": 5.84375, "learning_rate": 4.982302815009954e-07, "loss": 1.1809, "step": 36085 }, { "epoch": 90.215, "grad_norm": 8.75, "learning_rate": 4.979779237664716e-07, "loss": 1.4253, "step": 36086 }, { "epoch": 90.2175, "grad_norm": 7.0, "learning_rate": 4.977256283268239e-07, "loss": 1.2354, "step": 36087 }, { "epoch": 90.22, "grad_norm": 7.71875, "learning_rate": 4.974733951837074e-07, "loss": 1.2935, "step": 36088 }, { "epoch": 90.2225, "grad_norm": 9.875, "learning_rate": 4.972212243387742e-07, "loss": 1.3841, "step": 36089 }, { "epoch": 90.225, "grad_norm": 6.59375, "learning_rate": 4.969691157936796e-07, "loss": 1.3641, "step": 36090 }, { "epoch": 90.2275, "grad_norm": 15.1875, "learning_rate": 4.967170695500734e-07, "loss": 1.4232, "step": 36091 }, { "epoch": 90.23, "grad_norm": 6.0625, "learning_rate": 4.964650856096121e-07, "loss": 1.365, "step": 36092 }, { "epoch": 90.2325, "grad_norm": 6.0, "learning_rate": 4.962131639739442e-07, "loss": 1.3999, "step": 36093 }, { "epoch": 90.235, "grad_norm": 7.5, "learning_rate": 4.95961304644722e-07, "loss": 1.3045, "step": 36094 }, { "epoch": 90.2375, "grad_norm": 7.46875, "learning_rate": 4.957095076235974e-07, "loss": 1.3048, "step": 36095 }, { "epoch": 90.24, "grad_norm": 6.4375, "learning_rate": 4.954577729122212e-07, "loss": 1.4948, "step": 36096 }, { "epoch": 90.2425, "grad_norm": 6.1875, "learning_rate": 4.952061005122444e-07, "loss": 1.3803, "step": 36097 }, { "epoch": 90.245, "grad_norm": 27.0, "learning_rate": 4.949544904253156e-07, "loss": 1.4389, "step": 36098 }, { "epoch": 90.2475, "grad_norm": 8.0625, "learning_rate": 4.947029426530825e-07, "loss": 1.2557, "step": 36099 }, { "epoch": 90.25, "grad_norm": 6.125, "learning_rate": 4.944514571971981e-07, "loss": 1.3391, "step": 36100 }, { "epoch": 90.2525, "grad_norm": 6.25, "learning_rate": 4.942000340593111e-07, "loss": 1.1916, "step": 36101 }, { "epoch": 90.255, "grad_norm": 10.375, "learning_rate": 4.93948673241067e-07, "loss": 1.201, "step": 36102 }, { "epoch": 90.2575, "grad_norm": 10.5625, "learning_rate": 4.936973747441142e-07, "loss": 1.4771, "step": 36103 }, { "epoch": 90.26, "grad_norm": 10.1875, "learning_rate": 4.934461385701006e-07, "loss": 1.4251, "step": 36104 }, { "epoch": 90.2625, "grad_norm": 10.1875, "learning_rate": 4.931949647206757e-07, "loss": 1.5044, "step": 36105 }, { "epoch": 90.265, "grad_norm": 5.40625, "learning_rate": 4.92943853197484e-07, "loss": 1.2915, "step": 36106 }, { "epoch": 90.2675, "grad_norm": 7.21875, "learning_rate": 4.926928040021705e-07, "loss": 1.3918, "step": 36107 }, { "epoch": 90.27, "grad_norm": 6.8125, "learning_rate": 4.924418171363843e-07, "loss": 1.281, "step": 36108 }, { "epoch": 90.2725, "grad_norm": 6.90625, "learning_rate": 4.921908926017671e-07, "loss": 1.1816, "step": 36109 }, { "epoch": 90.275, "grad_norm": 8.625, "learning_rate": 4.919400303999655e-07, "loss": 1.3478, "step": 36110 }, { "epoch": 90.2775, "grad_norm": 8.9375, "learning_rate": 4.916892305326271e-07, "loss": 1.1933, "step": 36111 }, { "epoch": 90.28, "grad_norm": 7.875, "learning_rate": 4.914384930013927e-07, "loss": 1.225, "step": 36112 }, { "epoch": 90.2825, "grad_norm": 6.46875, "learning_rate": 4.911878178079077e-07, "loss": 1.2413, "step": 36113 }, { "epoch": 90.285, "grad_norm": 6.9375, "learning_rate": 4.909372049538119e-07, "loss": 1.3703, "step": 36114 }, { "epoch": 90.2875, "grad_norm": 6.21875, "learning_rate": 4.90686654440754e-07, "loss": 1.2736, "step": 36115 }, { "epoch": 90.29, "grad_norm": 8.3125, "learning_rate": 4.904361662703739e-07, "loss": 1.3431, "step": 36116 }, { "epoch": 90.2925, "grad_norm": 8.6875, "learning_rate": 4.901857404443133e-07, "loss": 1.3631, "step": 36117 }, { "epoch": 90.295, "grad_norm": 6.375, "learning_rate": 4.899353769642135e-07, "loss": 1.2389, "step": 36118 }, { "epoch": 90.2975, "grad_norm": 12.875, "learning_rate": 4.896850758317162e-07, "loss": 1.3997, "step": 36119 }, { "epoch": 90.3, "grad_norm": 7.59375, "learning_rate": 4.894348370484648e-07, "loss": 1.2854, "step": 36120 }, { "epoch": 90.3025, "grad_norm": 8.875, "learning_rate": 4.891846606160977e-07, "loss": 1.2143, "step": 36121 }, { "epoch": 90.305, "grad_norm": 6.1875, "learning_rate": 4.889345465362539e-07, "loss": 1.1236, "step": 36122 }, { "epoch": 90.3075, "grad_norm": 10.375, "learning_rate": 4.886844948105773e-07, "loss": 1.3377, "step": 36123 }, { "epoch": 90.31, "grad_norm": 5.96875, "learning_rate": 4.884345054407014e-07, "loss": 1.1324, "step": 36124 }, { "epoch": 90.3125, "grad_norm": 10.25, "learning_rate": 4.881845784282701e-07, "loss": 1.3106, "step": 36125 }, { "epoch": 90.315, "grad_norm": 7.6875, "learning_rate": 4.87934713774919e-07, "loss": 1.195, "step": 36126 }, { "epoch": 90.3175, "grad_norm": 9.3125, "learning_rate": 4.87684911482289e-07, "loss": 1.2136, "step": 36127 }, { "epoch": 90.32, "grad_norm": 9.625, "learning_rate": 4.874351715520154e-07, "loss": 1.3057, "step": 36128 }, { "epoch": 90.3225, "grad_norm": 6.21875, "learning_rate": 4.871854939857356e-07, "loss": 1.2981, "step": 36129 }, { "epoch": 90.325, "grad_norm": 6.8125, "learning_rate": 4.869358787850864e-07, "loss": 1.3409, "step": 36130 }, { "epoch": 90.3275, "grad_norm": 5.75, "learning_rate": 4.866863259517063e-07, "loss": 1.2492, "step": 36131 }, { "epoch": 90.33, "grad_norm": 6.59375, "learning_rate": 4.864368354872306e-07, "loss": 1.2416, "step": 36132 }, { "epoch": 90.3325, "grad_norm": 9.25, "learning_rate": 4.861874073932915e-07, "loss": 1.2811, "step": 36133 }, { "epoch": 90.335, "grad_norm": 7.21875, "learning_rate": 4.859380416715287e-07, "loss": 1.2431, "step": 36134 }, { "epoch": 90.3375, "grad_norm": 8.875, "learning_rate": 4.856887383235754e-07, "loss": 1.2952, "step": 36135 }, { "epoch": 90.34, "grad_norm": 6.0625, "learning_rate": 4.854394973510667e-07, "loss": 1.2952, "step": 36136 }, { "epoch": 90.3425, "grad_norm": 9.0, "learning_rate": 4.851903187556339e-07, "loss": 1.2588, "step": 36137 }, { "epoch": 90.345, "grad_norm": 11.875, "learning_rate": 4.849412025389144e-07, "loss": 1.3812, "step": 36138 }, { "epoch": 90.3475, "grad_norm": 6.9375, "learning_rate": 4.84692148702538e-07, "loss": 1.2878, "step": 36139 }, { "epoch": 90.35, "grad_norm": 8.5, "learning_rate": 4.844431572481412e-07, "loss": 1.2586, "step": 36140 }, { "epoch": 90.3525, "grad_norm": 5.46875, "learning_rate": 4.841942281773526e-07, "loss": 1.2467, "step": 36141 }, { "epoch": 90.355, "grad_norm": 6.96875, "learning_rate": 4.839453614918066e-07, "loss": 1.3259, "step": 36142 }, { "epoch": 90.3575, "grad_norm": 9.0, "learning_rate": 4.83696557193134e-07, "loss": 1.3535, "step": 36143 }, { "epoch": 90.36, "grad_norm": 33.25, "learning_rate": 4.834478152829658e-07, "loss": 1.2506, "step": 36144 }, { "epoch": 90.3625, "grad_norm": 8.25, "learning_rate": 4.831991357629318e-07, "loss": 1.3099, "step": 36145 }, { "epoch": 90.365, "grad_norm": 7.625, "learning_rate": 4.82950518634665e-07, "loss": 1.4139, "step": 36146 }, { "epoch": 90.3675, "grad_norm": 7.34375, "learning_rate": 4.827019638997943e-07, "loss": 1.2655, "step": 36147 }, { "epoch": 90.37, "grad_norm": 9.125, "learning_rate": 4.824534715599461e-07, "loss": 1.407, "step": 36148 }, { "epoch": 90.3725, "grad_norm": 7.3125, "learning_rate": 4.822050416167534e-07, "loss": 1.4571, "step": 36149 }, { "epoch": 90.375, "grad_norm": 11.5, "learning_rate": 4.81956674071844e-07, "loss": 1.3702, "step": 36150 }, { "epoch": 90.3775, "grad_norm": 6.90625, "learning_rate": 4.817083689268453e-07, "loss": 1.2433, "step": 36151 }, { "epoch": 90.38, "grad_norm": 6.90625, "learning_rate": 4.814601261833851e-07, "loss": 1.3827, "step": 36152 }, { "epoch": 90.3825, "grad_norm": 9.3125, "learning_rate": 4.81211945843093e-07, "loss": 1.2844, "step": 36153 }, { "epoch": 90.385, "grad_norm": 5.53125, "learning_rate": 4.809638279075935e-07, "loss": 1.3229, "step": 36154 }, { "epoch": 90.3875, "grad_norm": 7.40625, "learning_rate": 4.807157723785149e-07, "loss": 1.3707, "step": 36155 }, { "epoch": 90.39, "grad_norm": 6.125, "learning_rate": 4.804677792574808e-07, "loss": 1.234, "step": 36156 }, { "epoch": 90.3925, "grad_norm": 7.3125, "learning_rate": 4.802198485461207e-07, "loss": 1.3587, "step": 36157 }, { "epoch": 90.395, "grad_norm": 7.0625, "learning_rate": 4.79971980246059e-07, "loss": 1.3362, "step": 36158 }, { "epoch": 90.3975, "grad_norm": 5.625, "learning_rate": 4.797241743589176e-07, "loss": 1.1656, "step": 36159 }, { "epoch": 90.4, "grad_norm": 9.5625, "learning_rate": 4.794764308863242e-07, "loss": 1.4144, "step": 36160 }, { "epoch": 90.4025, "grad_norm": 8.0625, "learning_rate": 4.792287498299042e-07, "loss": 1.3527, "step": 36161 }, { "epoch": 90.405, "grad_norm": 10.375, "learning_rate": 4.789811311912784e-07, "loss": 1.1627, "step": 36162 }, { "epoch": 90.4075, "grad_norm": 7.1875, "learning_rate": 4.787335749720712e-07, "loss": 1.2776, "step": 36163 }, { "epoch": 90.41, "grad_norm": 8.75, "learning_rate": 4.784860811739045e-07, "loss": 1.3189, "step": 36164 }, { "epoch": 90.4125, "grad_norm": 6.71875, "learning_rate": 4.782386497984038e-07, "loss": 1.2183, "step": 36165 }, { "epoch": 90.415, "grad_norm": 6.34375, "learning_rate": 4.779912808471898e-07, "loss": 1.1539, "step": 36166 }, { "epoch": 90.4175, "grad_norm": 5.6875, "learning_rate": 4.777439743218815e-07, "loss": 1.3125, "step": 36167 }, { "epoch": 90.42, "grad_norm": 7.75, "learning_rate": 4.774967302241051e-07, "loss": 1.227, "step": 36168 }, { "epoch": 90.4225, "grad_norm": 9.6875, "learning_rate": 4.772495485554773e-07, "loss": 1.3957, "step": 36169 }, { "epoch": 90.425, "grad_norm": 8.375, "learning_rate": 4.770024293176223e-07, "loss": 1.3429, "step": 36170 }, { "epoch": 90.4275, "grad_norm": 8.375, "learning_rate": 4.7675537251215655e-07, "loss": 1.2868, "step": 36171 }, { "epoch": 90.43, "grad_norm": 8.75, "learning_rate": 4.7650837814070203e-07, "loss": 1.2657, "step": 36172 }, { "epoch": 90.4325, "grad_norm": 6.09375, "learning_rate": 4.762614462048787e-07, "loss": 1.2739, "step": 36173 }, { "epoch": 90.435, "grad_norm": 7.15625, "learning_rate": 4.760145767063018e-07, "loss": 1.4728, "step": 36174 }, { "epoch": 90.4375, "grad_norm": 6.34375, "learning_rate": 4.7576776964659344e-07, "loss": 1.331, "step": 36175 }, { "epoch": 90.44, "grad_norm": 10.3125, "learning_rate": 4.755210250273701e-07, "loss": 1.3162, "step": 36176 }, { "epoch": 90.4425, "grad_norm": 6.65625, "learning_rate": 4.7527434285025153e-07, "loss": 1.2209, "step": 36177 }, { "epoch": 90.445, "grad_norm": 9.6875, "learning_rate": 4.75027723116851e-07, "loss": 1.2837, "step": 36178 }, { "epoch": 90.4475, "grad_norm": 7.65625, "learning_rate": 4.747811658287871e-07, "loss": 1.3672, "step": 36179 }, { "epoch": 90.45, "grad_norm": 5.15625, "learning_rate": 4.745346709876786e-07, "loss": 1.3186, "step": 36180 }, { "epoch": 90.4525, "grad_norm": 7.75, "learning_rate": 4.742882385951397e-07, "loss": 1.1457, "step": 36181 }, { "epoch": 90.455, "grad_norm": 9.1875, "learning_rate": 4.7404186865278476e-07, "loss": 1.2763, "step": 36182 }, { "epoch": 90.4575, "grad_norm": 7.3125, "learning_rate": 4.7379556116223133e-07, "loss": 1.2912, "step": 36183 }, { "epoch": 90.46, "grad_norm": 6.3125, "learning_rate": 4.735493161250915e-07, "loss": 1.3866, "step": 36184 }, { "epoch": 90.4625, "grad_norm": 6.875, "learning_rate": 4.7330313354298165e-07, "loss": 1.3222, "step": 36185 }, { "epoch": 90.465, "grad_norm": 4.21875, "learning_rate": 4.7305701341751497e-07, "loss": 1.3145, "step": 36186 }, { "epoch": 90.4675, "grad_norm": 11.625, "learning_rate": 4.728109557503069e-07, "loss": 1.2209, "step": 36187 }, { "epoch": 90.47, "grad_norm": 8.5625, "learning_rate": 4.7256496054296827e-07, "loss": 1.1965, "step": 36188 }, { "epoch": 90.4725, "grad_norm": 7.53125, "learning_rate": 4.7231902779711126e-07, "loss": 1.3531, "step": 36189 }, { "epoch": 90.475, "grad_norm": 7.9375, "learning_rate": 4.720731575143489e-07, "loss": 1.3958, "step": 36190 }, { "epoch": 90.4775, "grad_norm": 5.84375, "learning_rate": 4.718273496962955e-07, "loss": 1.2895, "step": 36191 }, { "epoch": 90.48, "grad_norm": 5.75, "learning_rate": 4.715816043445609e-07, "loss": 1.178, "step": 36192 }, { "epoch": 90.4825, "grad_norm": 6.90625, "learning_rate": 4.713359214607538e-07, "loss": 1.371, "step": 36193 }, { "epoch": 90.485, "grad_norm": 7.3125, "learning_rate": 4.7109030104648845e-07, "loss": 1.2374, "step": 36194 }, { "epoch": 90.4875, "grad_norm": 6.625, "learning_rate": 4.708447431033736e-07, "loss": 1.3625, "step": 36195 }, { "epoch": 90.49, "grad_norm": 7.25, "learning_rate": 4.7059924763302013e-07, "loss": 1.3259, "step": 36196 }, { "epoch": 90.4925, "grad_norm": 7.375, "learning_rate": 4.703538146370357e-07, "loss": 1.3269, "step": 36197 }, { "epoch": 90.495, "grad_norm": 5.71875, "learning_rate": 4.7010844411703114e-07, "loss": 1.2141, "step": 36198 }, { "epoch": 90.4975, "grad_norm": 6.71875, "learning_rate": 4.698631360746131e-07, "loss": 1.3704, "step": 36199 }, { "epoch": 90.5, "grad_norm": 10.4375, "learning_rate": 4.696178905113913e-07, "loss": 1.4041, "step": 36200 }, { "epoch": 90.5025, "grad_norm": 7.3125, "learning_rate": 4.693727074289733e-07, "loss": 1.1572, "step": 36201 }, { "epoch": 90.505, "grad_norm": 7.1875, "learning_rate": 4.6912758682896684e-07, "loss": 1.2735, "step": 36202 }, { "epoch": 90.5075, "grad_norm": 7.71875, "learning_rate": 4.6888252871297946e-07, "loss": 1.5155, "step": 36203 }, { "epoch": 90.51, "grad_norm": 7.71875, "learning_rate": 4.686375330826143e-07, "loss": 1.4677, "step": 36204 }, { "epoch": 90.5125, "grad_norm": 6.75, "learning_rate": 4.6839259993948117e-07, "loss": 1.3448, "step": 36205 }, { "epoch": 90.515, "grad_norm": 7.59375, "learning_rate": 4.681477292851855e-07, "loss": 1.3019, "step": 36206 }, { "epoch": 90.5175, "grad_norm": 6.71875, "learning_rate": 4.679029211213315e-07, "loss": 1.4262, "step": 36207 }, { "epoch": 90.52, "grad_norm": 7.0625, "learning_rate": 4.676581754495235e-07, "loss": 1.3178, "step": 36208 }, { "epoch": 90.5225, "grad_norm": 7.09375, "learning_rate": 4.674134922713669e-07, "loss": 1.3888, "step": 36209 }, { "epoch": 90.525, "grad_norm": 11.625, "learning_rate": 4.6716887158846815e-07, "loss": 1.3889, "step": 36210 }, { "epoch": 90.5275, "grad_norm": 8.875, "learning_rate": 4.669243134024282e-07, "loss": 1.4644, "step": 36211 }, { "epoch": 90.53, "grad_norm": 7.1875, "learning_rate": 4.6667981771484903e-07, "loss": 1.3963, "step": 36212 }, { "epoch": 90.5325, "grad_norm": 4.9375, "learning_rate": 4.6643538452733727e-07, "loss": 1.1245, "step": 36213 }, { "epoch": 90.535, "grad_norm": 8.6875, "learning_rate": 4.6619101384149265e-07, "loss": 1.4514, "step": 36214 }, { "epoch": 90.5375, "grad_norm": 6.34375, "learning_rate": 4.6594670565891955e-07, "loss": 1.2809, "step": 36215 }, { "epoch": 90.54, "grad_norm": 5.40625, "learning_rate": 4.6570245998121654e-07, "loss": 1.1959, "step": 36216 }, { "epoch": 90.5425, "grad_norm": 5.375, "learning_rate": 4.65458276809988e-07, "loss": 1.2168, "step": 36217 }, { "epoch": 90.545, "grad_norm": 9.5625, "learning_rate": 4.6521415614683375e-07, "loss": 1.4493, "step": 36218 }, { "epoch": 90.5475, "grad_norm": 9.1875, "learning_rate": 4.649700979933525e-07, "loss": 1.4466, "step": 36219 }, { "epoch": 90.55, "grad_norm": 6.15625, "learning_rate": 4.6472610235114513e-07, "loss": 1.3604, "step": 36220 }, { "epoch": 90.5525, "grad_norm": 12.625, "learning_rate": 4.6448216922181376e-07, "loss": 1.243, "step": 36221 }, { "epoch": 90.555, "grad_norm": 5.75, "learning_rate": 4.64238298606956e-07, "loss": 1.1531, "step": 36222 }, { "epoch": 90.5575, "grad_norm": 6.5625, "learning_rate": 4.6399449050816836e-07, "loss": 1.2525, "step": 36223 }, { "epoch": 90.56, "grad_norm": 7.625, "learning_rate": 4.6375074492705173e-07, "loss": 1.3391, "step": 36224 }, { "epoch": 90.5625, "grad_norm": 10.75, "learning_rate": 4.635070618652049e-07, "loss": 1.4257, "step": 36225 }, { "epoch": 90.565, "grad_norm": 6.6875, "learning_rate": 4.632634413242243e-07, "loss": 1.2514, "step": 36226 }, { "epoch": 90.5675, "grad_norm": 7.65625, "learning_rate": 4.6301988330570533e-07, "loss": 1.282, "step": 36227 }, { "epoch": 90.57, "grad_norm": 5.6875, "learning_rate": 4.627763878112479e-07, "loss": 1.1939, "step": 36228 }, { "epoch": 90.5725, "grad_norm": 8.8125, "learning_rate": 4.6253295484244507e-07, "loss": 1.2674, "step": 36229 }, { "epoch": 90.575, "grad_norm": 7.78125, "learning_rate": 4.622895844008957e-07, "loss": 1.2332, "step": 36230 }, { "epoch": 90.5775, "grad_norm": 6.3125, "learning_rate": 4.6204627648819276e-07, "loss": 1.2063, "step": 36231 }, { "epoch": 90.58, "grad_norm": 6.96875, "learning_rate": 4.618030311059352e-07, "loss": 1.2743, "step": 36232 }, { "epoch": 90.5825, "grad_norm": 7.875, "learning_rate": 4.615598482557138e-07, "loss": 1.3499, "step": 36233 }, { "epoch": 90.585, "grad_norm": 6.75, "learning_rate": 4.6131672793912395e-07, "loss": 1.2659, "step": 36234 }, { "epoch": 90.5875, "grad_norm": 8.375, "learning_rate": 4.6107367015776006e-07, "loss": 1.2778, "step": 36235 }, { "epoch": 90.59, "grad_norm": 7.5625, "learning_rate": 4.6083067491321633e-07, "loss": 1.265, "step": 36236 }, { "epoch": 90.5925, "grad_norm": 7.71875, "learning_rate": 4.605877422070848e-07, "loss": 1.3058, "step": 36237 }, { "epoch": 90.595, "grad_norm": 9.3125, "learning_rate": 4.603448720409587e-07, "loss": 1.27, "step": 36238 }, { "epoch": 90.5975, "grad_norm": 8.4375, "learning_rate": 4.6010206441642667e-07, "loss": 1.3991, "step": 36239 }, { "epoch": 90.6, "grad_norm": 7.09375, "learning_rate": 4.5985931933508757e-07, "loss": 1.2683, "step": 36240 }, { "epoch": 90.6025, "grad_norm": 6.875, "learning_rate": 4.5961663679852777e-07, "loss": 1.2962, "step": 36241 }, { "epoch": 90.605, "grad_norm": 7.75, "learning_rate": 4.593740168083394e-07, "loss": 1.3311, "step": 36242 }, { "epoch": 90.6075, "grad_norm": 7.84375, "learning_rate": 4.5913145936611115e-07, "loss": 1.2095, "step": 36243 }, { "epoch": 90.61, "grad_norm": 9.875, "learning_rate": 4.5888896447343843e-07, "loss": 1.1783, "step": 36244 }, { "epoch": 90.6125, "grad_norm": 6.96875, "learning_rate": 4.586465321319067e-07, "loss": 1.3528, "step": 36245 }, { "epoch": 90.615, "grad_norm": 7.0625, "learning_rate": 4.584041623431057e-07, "loss": 1.194, "step": 36246 }, { "epoch": 90.6175, "grad_norm": 7.625, "learning_rate": 4.5816185510862644e-07, "loss": 1.4489, "step": 36247 }, { "epoch": 90.62, "grad_norm": 5.25, "learning_rate": 4.579196104300554e-07, "loss": 1.4961, "step": 36248 }, { "epoch": 90.6225, "grad_norm": 8.25, "learning_rate": 4.576774283089824e-07, "loss": 1.285, "step": 36249 }, { "epoch": 90.625, "grad_norm": 8.1875, "learning_rate": 4.5743530874699293e-07, "loss": 1.4339, "step": 36250 }, { "epoch": 90.6275, "grad_norm": 5.125, "learning_rate": 4.5719325174567674e-07, "loss": 1.2877, "step": 36251 }, { "epoch": 90.63, "grad_norm": 7.9375, "learning_rate": 4.569512573066204e-07, "loss": 1.3402, "step": 36252 }, { "epoch": 90.6325, "grad_norm": 6.46875, "learning_rate": 4.567093254314081e-07, "loss": 1.3702, "step": 36253 }, { "epoch": 90.635, "grad_norm": 12.0, "learning_rate": 4.5646745612162757e-07, "loss": 1.2986, "step": 36254 }, { "epoch": 90.6375, "grad_norm": 11.625, "learning_rate": 4.562256493788653e-07, "loss": 1.2233, "step": 36255 }, { "epoch": 90.64, "grad_norm": 6.96875, "learning_rate": 4.559839052047066e-07, "loss": 1.4126, "step": 36256 }, { "epoch": 90.6425, "grad_norm": 8.5625, "learning_rate": 4.557422236007347e-07, "loss": 1.3763, "step": 36257 }, { "epoch": 90.645, "grad_norm": 7.5625, "learning_rate": 4.5550060456853285e-07, "loss": 1.1708, "step": 36258 }, { "epoch": 90.6475, "grad_norm": 8.4375, "learning_rate": 4.5525904810968966e-07, "loss": 1.2406, "step": 36259 }, { "epoch": 90.65, "grad_norm": 7.09375, "learning_rate": 4.550175542257862e-07, "loss": 1.2461, "step": 36260 }, { "epoch": 90.6525, "grad_norm": 14.1875, "learning_rate": 4.5477612291840555e-07, "loss": 1.393, "step": 36261 }, { "epoch": 90.655, "grad_norm": 6.71875, "learning_rate": 4.545347541891287e-07, "loss": 1.2653, "step": 36262 }, { "epoch": 90.6575, "grad_norm": 6.3125, "learning_rate": 4.5429344803954113e-07, "loss": 1.3164, "step": 36263 }, { "epoch": 90.66, "grad_norm": 7.28125, "learning_rate": 4.540522044712248e-07, "loss": 1.1588, "step": 36264 }, { "epoch": 90.6625, "grad_norm": 9.5625, "learning_rate": 4.5381102348575954e-07, "loss": 1.2591, "step": 36265 }, { "epoch": 90.665, "grad_norm": 6.53125, "learning_rate": 4.535699050847264e-07, "loss": 1.2455, "step": 36266 }, { "epoch": 90.6675, "grad_norm": 5.28125, "learning_rate": 4.5332884926970853e-07, "loss": 1.442, "step": 36267 }, { "epoch": 90.67, "grad_norm": 7.65625, "learning_rate": 4.5308785604228245e-07, "loss": 1.2336, "step": 36268 }, { "epoch": 90.6725, "grad_norm": 7.59375, "learning_rate": 4.5284692540403244e-07, "loss": 1.3065, "step": 36269 }, { "epoch": 90.675, "grad_norm": 8.5, "learning_rate": 4.5260605735653387e-07, "loss": 1.2984, "step": 36270 }, { "epoch": 90.6775, "grad_norm": 7.6875, "learning_rate": 4.5236525190136995e-07, "loss": 1.3638, "step": 36271 }, { "epoch": 90.68, "grad_norm": 11.0625, "learning_rate": 4.521245090401172e-07, "loss": 1.3235, "step": 36272 }, { "epoch": 90.6825, "grad_norm": 7.96875, "learning_rate": 4.51883828774351e-07, "loss": 1.2515, "step": 36273 }, { "epoch": 90.685, "grad_norm": 6.0625, "learning_rate": 4.516432111056557e-07, "loss": 1.4316, "step": 36274 }, { "epoch": 90.6875, "grad_norm": 5.40625, "learning_rate": 4.514026560356044e-07, "loss": 1.2895, "step": 36275 }, { "epoch": 90.69, "grad_norm": 7.125, "learning_rate": 4.511621635657759e-07, "loss": 1.3186, "step": 36276 }, { "epoch": 90.6925, "grad_norm": 7.375, "learning_rate": 4.509217336977456e-07, "loss": 1.1736, "step": 36277 }, { "epoch": 90.695, "grad_norm": 8.375, "learning_rate": 4.506813664330889e-07, "loss": 1.2815, "step": 36278 }, { "epoch": 90.6975, "grad_norm": 11.25, "learning_rate": 4.5044106177338453e-07, "loss": 1.1254, "step": 36279 }, { "epoch": 90.7, "grad_norm": 7.15625, "learning_rate": 4.502008197202068e-07, "loss": 1.3442, "step": 36280 }, { "epoch": 90.7025, "grad_norm": 7.125, "learning_rate": 4.499606402751289e-07, "loss": 1.3259, "step": 36281 }, { "epoch": 90.705, "grad_norm": 8.25, "learning_rate": 4.497205234397284e-07, "loss": 1.3778, "step": 36282 }, { "epoch": 90.7075, "grad_norm": 9.25, "learning_rate": 4.494804692155763e-07, "loss": 1.5464, "step": 36283 }, { "epoch": 90.71, "grad_norm": 9.125, "learning_rate": 4.492404776042503e-07, "loss": 1.4005, "step": 36284 }, { "epoch": 90.7125, "grad_norm": 6.09375, "learning_rate": 4.49000548607319e-07, "loss": 1.3219, "step": 36285 }, { "epoch": 90.715, "grad_norm": 4.90625, "learning_rate": 4.487606822263602e-07, "loss": 1.333, "step": 36286 }, { "epoch": 90.7175, "grad_norm": 9.5, "learning_rate": 4.485208784629436e-07, "loss": 1.3848, "step": 36287 }, { "epoch": 90.72, "grad_norm": 6.09375, "learning_rate": 4.482811373186402e-07, "loss": 1.4892, "step": 36288 }, { "epoch": 90.7225, "grad_norm": 10.0, "learning_rate": 4.4804145879502325e-07, "loss": 1.2527, "step": 36289 }, { "epoch": 90.725, "grad_norm": 8.5625, "learning_rate": 4.47801842893667e-07, "loss": 1.3212, "step": 36290 }, { "epoch": 90.7275, "grad_norm": 11.5625, "learning_rate": 4.47562289616138e-07, "loss": 1.2585, "step": 36291 }, { "epoch": 90.73, "grad_norm": 5.5625, "learning_rate": 4.473227989640083e-07, "loss": 1.3153, "step": 36292 }, { "epoch": 90.7325, "grad_norm": 9.75, "learning_rate": 4.4708337093884777e-07, "loss": 1.3909, "step": 36293 }, { "epoch": 90.735, "grad_norm": 5.625, "learning_rate": 4.468440055422274e-07, "loss": 1.3914, "step": 36294 }, { "epoch": 90.7375, "grad_norm": 6.71875, "learning_rate": 4.466047027757148e-07, "loss": 1.4116, "step": 36295 }, { "epoch": 90.74, "grad_norm": 9.25, "learning_rate": 4.463654626408798e-07, "loss": 1.2925, "step": 36296 }, { "epoch": 90.7425, "grad_norm": 5.59375, "learning_rate": 4.4612628513929e-07, "loss": 1.3483, "step": 36297 }, { "epoch": 90.745, "grad_norm": 8.0625, "learning_rate": 4.4588717027251425e-07, "loss": 1.1885, "step": 36298 }, { "epoch": 90.7475, "grad_norm": 9.3125, "learning_rate": 4.4564811804212017e-07, "loss": 1.4358, "step": 36299 }, { "epoch": 90.75, "grad_norm": 7.25, "learning_rate": 4.454091284496731e-07, "loss": 1.324, "step": 36300 }, { "epoch": 90.7525, "grad_norm": 7.65625, "learning_rate": 4.4517020149674406e-07, "loss": 1.2192, "step": 36301 }, { "epoch": 90.755, "grad_norm": 9.1875, "learning_rate": 4.449313371848962e-07, "loss": 1.2574, "step": 36302 }, { "epoch": 90.7575, "grad_norm": 8.25, "learning_rate": 4.446925355156939e-07, "loss": 1.2199, "step": 36303 }, { "epoch": 90.76, "grad_norm": 8.3125, "learning_rate": 4.4445379649070587e-07, "loss": 1.316, "step": 36304 }, { "epoch": 90.7625, "grad_norm": 10.125, "learning_rate": 4.4421512011149746e-07, "loss": 1.2868, "step": 36305 }, { "epoch": 90.765, "grad_norm": 5.96875, "learning_rate": 4.439765063796331e-07, "loss": 1.4297, "step": 36306 }, { "epoch": 90.7675, "grad_norm": 7.78125, "learning_rate": 4.4373795529667363e-07, "loss": 1.1902, "step": 36307 }, { "epoch": 90.77, "grad_norm": 8.4375, "learning_rate": 4.434994668641868e-07, "loss": 1.277, "step": 36308 }, { "epoch": 90.7725, "grad_norm": 7.0625, "learning_rate": 4.432610410837357e-07, "loss": 1.344, "step": 36309 }, { "epoch": 90.775, "grad_norm": 9.1875, "learning_rate": 4.4302267795688357e-07, "loss": 1.3571, "step": 36310 }, { "epoch": 90.7775, "grad_norm": 9.3125, "learning_rate": 4.427843774851903e-07, "loss": 1.344, "step": 36311 }, { "epoch": 90.78, "grad_norm": 6.625, "learning_rate": 4.425461396702213e-07, "loss": 1.214, "step": 36312 }, { "epoch": 90.7825, "grad_norm": 6.71875, "learning_rate": 4.423079645135364e-07, "loss": 1.3694, "step": 36313 }, { "epoch": 90.785, "grad_norm": 8.4375, "learning_rate": 4.420698520166988e-07, "loss": 1.2557, "step": 36314 }, { "epoch": 90.7875, "grad_norm": 8.5, "learning_rate": 4.4183180218126844e-07, "loss": 1.4613, "step": 36315 }, { "epoch": 90.79, "grad_norm": 7.59375, "learning_rate": 4.415938150088073e-07, "loss": 1.3307, "step": 36316 }, { "epoch": 90.7925, "grad_norm": 5.34375, "learning_rate": 4.4135589050087414e-07, "loss": 1.2429, "step": 36317 }, { "epoch": 90.795, "grad_norm": 7.6875, "learning_rate": 4.4111802865902886e-07, "loss": 1.39, "step": 36318 }, { "epoch": 90.7975, "grad_norm": 8.125, "learning_rate": 4.408802294848302e-07, "loss": 1.1591, "step": 36319 }, { "epoch": 90.8, "grad_norm": 6.65625, "learning_rate": 4.406424929798403e-07, "loss": 1.2521, "step": 36320 }, { "epoch": 90.8025, "grad_norm": 6.8125, "learning_rate": 4.404048191456145e-07, "loss": 1.1487, "step": 36321 }, { "epoch": 90.805, "grad_norm": 7.09375, "learning_rate": 4.401672079837116e-07, "loss": 1.3801, "step": 36322 }, { "epoch": 90.8075, "grad_norm": 9.6875, "learning_rate": 4.3992965949569033e-07, "loss": 1.2425, "step": 36323 }, { "epoch": 90.81, "grad_norm": 7.46875, "learning_rate": 4.3969217368310834e-07, "loss": 1.2673, "step": 36324 }, { "epoch": 90.8125, "grad_norm": 7.84375, "learning_rate": 4.3945475054752216e-07, "loss": 1.3483, "step": 36325 }, { "epoch": 90.815, "grad_norm": 7.15625, "learning_rate": 4.3921739009048725e-07, "loss": 1.2474, "step": 36326 }, { "epoch": 90.8175, "grad_norm": 11.1875, "learning_rate": 4.389800923135612e-07, "loss": 1.2803, "step": 36327 }, { "epoch": 90.82, "grad_norm": 6.5, "learning_rate": 4.3874285721829835e-07, "loss": 1.2709, "step": 36328 }, { "epoch": 90.8225, "grad_norm": 6.4375, "learning_rate": 4.3850568480625523e-07, "loss": 1.3932, "step": 36329 }, { "epoch": 90.825, "grad_norm": 8.625, "learning_rate": 4.38268575078985e-07, "loss": 1.5823, "step": 36330 }, { "epoch": 90.8275, "grad_norm": 9.8125, "learning_rate": 4.3803152803804427e-07, "loss": 1.2508, "step": 36331 }, { "epoch": 90.83, "grad_norm": 8.125, "learning_rate": 4.377945436849873e-07, "loss": 1.4301, "step": 36332 }, { "epoch": 90.8325, "grad_norm": 5.125, "learning_rate": 4.375576220213651e-07, "loss": 1.3717, "step": 36333 }, { "epoch": 90.835, "grad_norm": 8.375, "learning_rate": 4.3732076304873194e-07, "loss": 1.229, "step": 36334 }, { "epoch": 90.8375, "grad_norm": 5.28125, "learning_rate": 4.370839667686422e-07, "loss": 1.4313, "step": 36335 }, { "epoch": 90.84, "grad_norm": 6.84375, "learning_rate": 4.368472331826479e-07, "loss": 1.3188, "step": 36336 }, { "epoch": 90.8425, "grad_norm": 14.5, "learning_rate": 4.366105622922989e-07, "loss": 1.3587, "step": 36337 }, { "epoch": 90.845, "grad_norm": 6.75, "learning_rate": 4.363739540991485e-07, "loss": 1.2738, "step": 36338 }, { "epoch": 90.8475, "grad_norm": 8.8125, "learning_rate": 4.361374086047498e-07, "loss": 1.5613, "step": 36339 }, { "epoch": 90.85, "grad_norm": 8.3125, "learning_rate": 4.3590092581065055e-07, "loss": 1.3349, "step": 36340 }, { "epoch": 90.8525, "grad_norm": 8.1875, "learning_rate": 4.3566450571840057e-07, "loss": 1.2954, "step": 36341 }, { "epoch": 90.855, "grad_norm": 7.125, "learning_rate": 4.35428148329553e-07, "loss": 1.3087, "step": 36342 }, { "epoch": 90.8575, "grad_norm": 7.15625, "learning_rate": 4.3519185364565454e-07, "loss": 1.2631, "step": 36343 }, { "epoch": 90.86, "grad_norm": 8.9375, "learning_rate": 4.3495562166825714e-07, "loss": 1.2893, "step": 36344 }, { "epoch": 90.8625, "grad_norm": 6.46875, "learning_rate": 4.3471945239890624e-07, "loss": 1.1718, "step": 36345 }, { "epoch": 90.865, "grad_norm": 10.375, "learning_rate": 4.3448334583915286e-07, "loss": 1.2376, "step": 36346 }, { "epoch": 90.8675, "grad_norm": 3.578125, "learning_rate": 4.3424730199054464e-07, "loss": 1.364, "step": 36347 }, { "epoch": 90.87, "grad_norm": 10.3125, "learning_rate": 4.3401132085462705e-07, "loss": 1.2728, "step": 36348 }, { "epoch": 90.8725, "grad_norm": 5.46875, "learning_rate": 4.337754024329477e-07, "loss": 1.3325, "step": 36349 }, { "epoch": 90.875, "grad_norm": 7.6875, "learning_rate": 4.3353954672705533e-07, "loss": 1.2664, "step": 36350 }, { "epoch": 90.8775, "grad_norm": 7.40625, "learning_rate": 4.333037537384954e-07, "loss": 1.2413, "step": 36351 }, { "epoch": 90.88, "grad_norm": 6.1875, "learning_rate": 4.3306802346881116e-07, "loss": 1.3099, "step": 36352 }, { "epoch": 90.8825, "grad_norm": 7.96875, "learning_rate": 4.328323559195502e-07, "loss": 1.1159, "step": 36353 }, { "epoch": 90.885, "grad_norm": 7.875, "learning_rate": 4.3259675109225906e-07, "loss": 1.1709, "step": 36354 }, { "epoch": 90.8875, "grad_norm": 8.125, "learning_rate": 4.32361208988481e-07, "loss": 1.3156, "step": 36355 }, { "epoch": 90.89, "grad_norm": 7.46875, "learning_rate": 4.321257296097592e-07, "loss": 1.2553, "step": 36356 }, { "epoch": 90.8925, "grad_norm": 8.25, "learning_rate": 4.3189031295763793e-07, "loss": 1.3419, "step": 36357 }, { "epoch": 90.895, "grad_norm": 9.125, "learning_rate": 4.3165495903366053e-07, "loss": 1.4013, "step": 36358 }, { "epoch": 90.8975, "grad_norm": 12.5625, "learning_rate": 4.3141966783937227e-07, "loss": 1.3881, "step": 36359 }, { "epoch": 90.9, "grad_norm": 7.875, "learning_rate": 4.3118443937631094e-07, "loss": 1.4112, "step": 36360 }, { "epoch": 90.9025, "grad_norm": 4.40625, "learning_rate": 4.3094927364602413e-07, "loss": 1.2958, "step": 36361 }, { "epoch": 90.905, "grad_norm": 6.40625, "learning_rate": 4.3071417065005063e-07, "loss": 1.26, "step": 36362 }, { "epoch": 90.9075, "grad_norm": 6.09375, "learning_rate": 4.304791303899303e-07, "loss": 1.3079, "step": 36363 }, { "epoch": 90.91, "grad_norm": 8.5, "learning_rate": 4.3024415286720635e-07, "loss": 1.3124, "step": 36364 }, { "epoch": 90.9125, "grad_norm": 6.5, "learning_rate": 4.300092380834198e-07, "loss": 1.2888, "step": 36365 }, { "epoch": 90.915, "grad_norm": 6.6875, "learning_rate": 4.2977438604010933e-07, "loss": 1.326, "step": 36366 }, { "epoch": 90.9175, "grad_norm": 5.4375, "learning_rate": 4.2953959673881383e-07, "loss": 1.2032, "step": 36367 }, { "epoch": 90.92, "grad_norm": 7.65625, "learning_rate": 4.2930487018107425e-07, "loss": 1.4478, "step": 36368 }, { "epoch": 90.9225, "grad_norm": 8.125, "learning_rate": 4.2907020636843045e-07, "loss": 1.2698, "step": 36369 }, { "epoch": 90.925, "grad_norm": 8.75, "learning_rate": 4.28835605302419e-07, "loss": 1.2922, "step": 36370 }, { "epoch": 90.9275, "grad_norm": 7.78125, "learning_rate": 4.286010669845764e-07, "loss": 1.2926, "step": 36371 }, { "epoch": 90.93, "grad_norm": 6.4375, "learning_rate": 4.283665914164437e-07, "loss": 1.3704, "step": 36372 }, { "epoch": 90.9325, "grad_norm": 6.875, "learning_rate": 4.2813217859955625e-07, "loss": 1.3714, "step": 36373 }, { "epoch": 90.935, "grad_norm": 7.84375, "learning_rate": 4.278978285354518e-07, "loss": 1.2704, "step": 36374 }, { "epoch": 90.9375, "grad_norm": 4.4375, "learning_rate": 4.2766354122566465e-07, "loss": 1.3547, "step": 36375 }, { "epoch": 90.94, "grad_norm": 6.4375, "learning_rate": 4.274293166717336e-07, "loss": 1.3241, "step": 36376 }, { "epoch": 90.9425, "grad_norm": 7.84375, "learning_rate": 4.2719515487519294e-07, "loss": 1.4115, "step": 36377 }, { "epoch": 90.945, "grad_norm": 6.40625, "learning_rate": 4.2696105583757696e-07, "loss": 1.186, "step": 36378 }, { "epoch": 90.9475, "grad_norm": 5.375, "learning_rate": 4.267270195604212e-07, "loss": 1.2529, "step": 36379 }, { "epoch": 90.95, "grad_norm": 5.6875, "learning_rate": 4.26493046045261e-07, "loss": 1.2817, "step": 36380 }, { "epoch": 90.9525, "grad_norm": 6.8125, "learning_rate": 4.262591352936285e-07, "loss": 1.2475, "step": 36381 }, { "epoch": 90.955, "grad_norm": 8.125, "learning_rate": 4.2602528730705914e-07, "loss": 1.3632, "step": 36382 }, { "epoch": 90.9575, "grad_norm": 6.40625, "learning_rate": 4.2579150208708176e-07, "loss": 1.4377, "step": 36383 }, { "epoch": 90.96, "grad_norm": 9.0625, "learning_rate": 4.2555777963523506e-07, "loss": 1.2388, "step": 36384 }, { "epoch": 90.9625, "grad_norm": 7.1875, "learning_rate": 4.2532411995304776e-07, "loss": 1.2763, "step": 36385 }, { "epoch": 90.965, "grad_norm": 7.4375, "learning_rate": 4.250905230420521e-07, "loss": 1.2789, "step": 36386 }, { "epoch": 90.9675, "grad_norm": 7.375, "learning_rate": 4.248569889037801e-07, "loss": 1.2712, "step": 36387 }, { "epoch": 90.97, "grad_norm": 7.5625, "learning_rate": 4.2462351753976174e-07, "loss": 1.3798, "step": 36388 }, { "epoch": 90.9725, "grad_norm": 6.1875, "learning_rate": 4.243901089515301e-07, "loss": 1.262, "step": 36389 }, { "epoch": 90.975, "grad_norm": 5.84375, "learning_rate": 4.2415676314061183e-07, "loss": 1.3937, "step": 36390 }, { "epoch": 90.9775, "grad_norm": 7.34375, "learning_rate": 4.239234801085401e-07, "loss": 1.4444, "step": 36391 }, { "epoch": 90.98, "grad_norm": 6.5625, "learning_rate": 4.2369025985684264e-07, "loss": 1.3271, "step": 36392 }, { "epoch": 90.9825, "grad_norm": 9.1875, "learning_rate": 4.234571023870471e-07, "loss": 1.4697, "step": 36393 }, { "epoch": 90.985, "grad_norm": 7.875, "learning_rate": 4.2322400770068326e-07, "loss": 1.3844, "step": 36394 }, { "epoch": 90.9875, "grad_norm": 5.4375, "learning_rate": 4.2299097579928225e-07, "loss": 1.3654, "step": 36395 }, { "epoch": 90.99, "grad_norm": 10.0, "learning_rate": 4.227580066843684e-07, "loss": 1.2365, "step": 36396 }, { "epoch": 90.9925, "grad_norm": 13.375, "learning_rate": 4.225251003574693e-07, "loss": 1.4373, "step": 36397 }, { "epoch": 90.995, "grad_norm": 7.90625, "learning_rate": 4.222922568201104e-07, "loss": 1.405, "step": 36398 }, { "epoch": 90.9975, "grad_norm": 8.0625, "learning_rate": 4.220594760738228e-07, "loss": 1.3144, "step": 36399 }, { "epoch": 91.0, "grad_norm": 11.1875, "learning_rate": 4.218267581201296e-07, "loss": 1.4116, "step": 36400 }, { "epoch": 91.0025, "grad_norm": 7.15625, "learning_rate": 4.2159410296055745e-07, "loss": 1.2508, "step": 36401 }, { "epoch": 91.005, "grad_norm": 5.96875, "learning_rate": 4.2136151059662955e-07, "loss": 1.324, "step": 36402 }, { "epoch": 91.0075, "grad_norm": 7.21875, "learning_rate": 4.2112898102987133e-07, "loss": 1.2727, "step": 36403 }, { "epoch": 91.01, "grad_norm": 10.875, "learning_rate": 4.2089651426181044e-07, "loss": 1.4358, "step": 36404 }, { "epoch": 91.0125, "grad_norm": 7.40625, "learning_rate": 4.20664110293969e-07, "loss": 1.1059, "step": 36405 }, { "epoch": 91.015, "grad_norm": 7.375, "learning_rate": 4.204317691278681e-07, "loss": 1.3494, "step": 36406 }, { "epoch": 91.0175, "grad_norm": 6.6875, "learning_rate": 4.201994907650353e-07, "loss": 1.413, "step": 36407 }, { "epoch": 91.02, "grad_norm": 5.4375, "learning_rate": 4.199672752069894e-07, "loss": 1.2467, "step": 36408 }, { "epoch": 91.0225, "grad_norm": 8.0625, "learning_rate": 4.197351224552559e-07, "loss": 1.3913, "step": 36409 }, { "epoch": 91.025, "grad_norm": 8.625, "learning_rate": 4.195030325113547e-07, "loss": 1.3779, "step": 36410 }, { "epoch": 91.0275, "grad_norm": 5.53125, "learning_rate": 4.19271005376809e-07, "loss": 1.3906, "step": 36411 }, { "epoch": 91.03, "grad_norm": 8.1875, "learning_rate": 4.1903904105313976e-07, "loss": 1.3094, "step": 36412 }, { "epoch": 91.0325, "grad_norm": 7.25, "learning_rate": 4.188071395418647e-07, "loss": 1.2558, "step": 36413 }, { "epoch": 91.035, "grad_norm": 7.34375, "learning_rate": 4.1857530084450813e-07, "loss": 1.2665, "step": 36414 }, { "epoch": 91.0375, "grad_norm": 9.3125, "learning_rate": 4.183435249625889e-07, "loss": 1.3932, "step": 36415 }, { "epoch": 91.04, "grad_norm": 6.0, "learning_rate": 4.1811181189762684e-07, "loss": 1.2924, "step": 36416 }, { "epoch": 91.0425, "grad_norm": 6.125, "learning_rate": 4.1788016165113856e-07, "loss": 1.41, "step": 36417 }, { "epoch": 91.045, "grad_norm": 8.875, "learning_rate": 4.176485742246439e-07, "loss": 1.1181, "step": 36418 }, { "epoch": 91.0475, "grad_norm": 10.8125, "learning_rate": 4.1741704961966387e-07, "loss": 1.3773, "step": 36419 }, { "epoch": 91.05, "grad_norm": 6.59375, "learning_rate": 4.17185587837714e-07, "loss": 1.2701, "step": 36420 }, { "epoch": 91.0525, "grad_norm": 7.1875, "learning_rate": 4.169541888803108e-07, "loss": 1.4966, "step": 36421 }, { "epoch": 91.055, "grad_norm": 8.3125, "learning_rate": 4.1672285274897307e-07, "loss": 1.3653, "step": 36422 }, { "epoch": 91.0575, "grad_norm": 11.625, "learning_rate": 4.1649157944521624e-07, "loss": 1.2943, "step": 36423 }, { "epoch": 91.06, "grad_norm": 7.5, "learning_rate": 4.16260368970558e-07, "loss": 1.2781, "step": 36424 }, { "epoch": 91.0625, "grad_norm": 5.78125, "learning_rate": 4.160292213265116e-07, "loss": 1.4824, "step": 36425 }, { "epoch": 91.065, "grad_norm": 7.78125, "learning_rate": 4.157981365145958e-07, "loss": 1.409, "step": 36426 }, { "epoch": 91.0675, "grad_norm": 6.25, "learning_rate": 4.155671145363238e-07, "loss": 1.1774, "step": 36427 }, { "epoch": 91.07, "grad_norm": 6.4375, "learning_rate": 4.1533615539320893e-07, "loss": 1.395, "step": 36428 }, { "epoch": 91.0725, "grad_norm": 5.6875, "learning_rate": 4.151052590867666e-07, "loss": 1.2531, "step": 36429 }, { "epoch": 91.075, "grad_norm": 7.15625, "learning_rate": 4.148744256185122e-07, "loss": 1.3164, "step": 36430 }, { "epoch": 91.0775, "grad_norm": 8.625, "learning_rate": 4.1464365498995793e-07, "loss": 1.4879, "step": 36431 }, { "epoch": 91.08, "grad_norm": 8.9375, "learning_rate": 4.1441294720261373e-07, "loss": 1.3634, "step": 36432 }, { "epoch": 91.0825, "grad_norm": 7.3125, "learning_rate": 4.14182302257996e-07, "loss": 1.284, "step": 36433 }, { "epoch": 91.085, "grad_norm": 7.03125, "learning_rate": 4.1395172015761596e-07, "loss": 1.2315, "step": 36434 }, { "epoch": 91.0875, "grad_norm": 13.4375, "learning_rate": 4.137212009029845e-07, "loss": 1.2144, "step": 36435 }, { "epoch": 91.09, "grad_norm": 8.5625, "learning_rate": 4.1349074449561267e-07, "loss": 1.2377, "step": 36436 }, { "epoch": 91.0925, "grad_norm": 7.1875, "learning_rate": 4.1326035093701366e-07, "loss": 1.2972, "step": 36437 }, { "epoch": 91.095, "grad_norm": 8.125, "learning_rate": 4.130300202286941e-07, "loss": 1.3417, "step": 36438 }, { "epoch": 91.0975, "grad_norm": 7.34375, "learning_rate": 4.127997523721672e-07, "loss": 1.4523, "step": 36439 }, { "epoch": 91.1, "grad_norm": 9.9375, "learning_rate": 4.125695473689406e-07, "loss": 1.3635, "step": 36440 }, { "epoch": 91.1025, "grad_norm": 6.8125, "learning_rate": 4.123394052205254e-07, "loss": 1.2885, "step": 36441 }, { "epoch": 91.105, "grad_norm": 5.5, "learning_rate": 4.121093259284292e-07, "loss": 1.3167, "step": 36442 }, { "epoch": 91.1075, "grad_norm": 6.03125, "learning_rate": 4.1187930949415975e-07, "loss": 1.2313, "step": 36443 }, { "epoch": 91.11, "grad_norm": 7.9375, "learning_rate": 4.1164935591922583e-07, "loss": 1.5433, "step": 36444 }, { "epoch": 91.1125, "grad_norm": 7.875, "learning_rate": 4.114194652051362e-07, "loss": 1.1701, "step": 36445 }, { "epoch": 91.115, "grad_norm": 11.625, "learning_rate": 4.1118963735339636e-07, "loss": 1.2198, "step": 36446 }, { "epoch": 91.1175, "grad_norm": 7.03125, "learning_rate": 4.109598723655128e-07, "loss": 1.3125, "step": 36447 }, { "epoch": 91.12, "grad_norm": 8.125, "learning_rate": 4.107301702429922e-07, "loss": 1.3865, "step": 36448 }, { "epoch": 91.1225, "grad_norm": 6.875, "learning_rate": 4.105005309873422e-07, "loss": 1.2797, "step": 36449 }, { "epoch": 91.125, "grad_norm": 5.78125, "learning_rate": 4.1027095460006715e-07, "loss": 1.39, "step": 36450 }, { "epoch": 91.1275, "grad_norm": 9.8125, "learning_rate": 4.100414410826714e-07, "loss": 1.323, "step": 36451 }, { "epoch": 91.13, "grad_norm": 7.09375, "learning_rate": 4.098119904366604e-07, "loss": 1.3186, "step": 36452 }, { "epoch": 91.1325, "grad_norm": 10.3125, "learning_rate": 4.0958260266353745e-07, "loss": 1.3459, "step": 36453 }, { "epoch": 91.135, "grad_norm": 7.25, "learning_rate": 4.0935327776480794e-07, "loss": 1.3776, "step": 36454 }, { "epoch": 91.1375, "grad_norm": 4.96875, "learning_rate": 4.0912401574197403e-07, "loss": 1.4918, "step": 36455 }, { "epoch": 91.14, "grad_norm": 7.625, "learning_rate": 4.0889481659654006e-07, "loss": 1.3199, "step": 36456 }, { "epoch": 91.1425, "grad_norm": 5.875, "learning_rate": 4.086656803300082e-07, "loss": 1.3106, "step": 36457 }, { "epoch": 91.145, "grad_norm": 6.6875, "learning_rate": 4.0843660694387835e-07, "loss": 1.3381, "step": 36458 }, { "epoch": 91.1475, "grad_norm": 7.40625, "learning_rate": 4.082075964396548e-07, "loss": 1.2904, "step": 36459 }, { "epoch": 91.15, "grad_norm": 5.4375, "learning_rate": 4.0797864881883977e-07, "loss": 1.1852, "step": 36460 }, { "epoch": 91.1525, "grad_norm": 10.6875, "learning_rate": 4.0774976408293196e-07, "loss": 1.429, "step": 36461 }, { "epoch": 91.155, "grad_norm": 4.65625, "learning_rate": 4.075209422334325e-07, "loss": 1.1906, "step": 36462 }, { "epoch": 91.1575, "grad_norm": 5.9375, "learning_rate": 4.072921832718424e-07, "loss": 1.0967, "step": 36463 }, { "epoch": 91.16, "grad_norm": 8.9375, "learning_rate": 4.070634871996615e-07, "loss": 1.3117, "step": 36464 }, { "epoch": 91.1625, "grad_norm": 5.34375, "learning_rate": 4.068348540183875e-07, "loss": 1.3273, "step": 36465 }, { "epoch": 91.165, "grad_norm": 5.375, "learning_rate": 4.066062837295204e-07, "loss": 1.1982, "step": 36466 }, { "epoch": 91.1675, "grad_norm": 5.65625, "learning_rate": 4.063777763345589e-07, "loss": 1.4769, "step": 36467 }, { "epoch": 91.17, "grad_norm": 7.0625, "learning_rate": 4.061493318350007e-07, "loss": 1.2531, "step": 36468 }, { "epoch": 91.1725, "grad_norm": 7.625, "learning_rate": 4.0592095023234355e-07, "loss": 1.2627, "step": 36469 }, { "epoch": 91.175, "grad_norm": 8.0, "learning_rate": 4.05692631528084e-07, "loss": 1.2855, "step": 36470 }, { "epoch": 91.1775, "grad_norm": 9.625, "learning_rate": 4.054643757237209e-07, "loss": 1.2709, "step": 36471 }, { "epoch": 91.18, "grad_norm": 14.75, "learning_rate": 4.0523618282074964e-07, "loss": 1.4213, "step": 36472 }, { "epoch": 91.1825, "grad_norm": 7.6875, "learning_rate": 4.0500805282066457e-07, "loss": 1.5156, "step": 36473 }, { "epoch": 91.185, "grad_norm": 7.875, "learning_rate": 4.0477998572496237e-07, "loss": 1.5021, "step": 36474 }, { "epoch": 91.1875, "grad_norm": 8.3125, "learning_rate": 4.0455198153514064e-07, "loss": 1.2345, "step": 36475 }, { "epoch": 91.19, "grad_norm": 6.0625, "learning_rate": 4.043240402526916e-07, "loss": 1.3726, "step": 36476 }, { "epoch": 91.1925, "grad_norm": 9.6875, "learning_rate": 4.0409616187910836e-07, "loss": 1.3431, "step": 36477 }, { "epoch": 91.195, "grad_norm": 7.15625, "learning_rate": 4.038683464158877e-07, "loss": 1.4153, "step": 36478 }, { "epoch": 91.1975, "grad_norm": 5.96875, "learning_rate": 4.036405938645227e-07, "loss": 1.341, "step": 36479 }, { "epoch": 91.2, "grad_norm": 5.4375, "learning_rate": 4.034129042265067e-07, "loss": 1.3128, "step": 36480 }, { "epoch": 91.2025, "grad_norm": 7.09375, "learning_rate": 4.031852775033296e-07, "loss": 1.405, "step": 36481 }, { "epoch": 91.205, "grad_norm": 7.625, "learning_rate": 4.0295771369648795e-07, "loss": 1.1939, "step": 36482 }, { "epoch": 91.2075, "grad_norm": 9.0625, "learning_rate": 4.027302128074695e-07, "loss": 1.3597, "step": 36483 }, { "epoch": 91.21, "grad_norm": 8.25, "learning_rate": 4.0250277483776854e-07, "loss": 1.4756, "step": 36484 }, { "epoch": 91.2125, "grad_norm": 7.71875, "learning_rate": 4.022753997888751e-07, "loss": 1.283, "step": 36485 }, { "epoch": 91.215, "grad_norm": 6.28125, "learning_rate": 4.020480876622812e-07, "loss": 1.2577, "step": 36486 }, { "epoch": 91.2175, "grad_norm": 6.34375, "learning_rate": 4.018208384594735e-07, "loss": 1.2997, "step": 36487 }, { "epoch": 91.22, "grad_norm": 17.625, "learning_rate": 4.0159365218194633e-07, "loss": 1.288, "step": 36488 }, { "epoch": 91.2225, "grad_norm": 10.1875, "learning_rate": 4.0136652883118634e-07, "loss": 1.3808, "step": 36489 }, { "epoch": 91.225, "grad_norm": 7.53125, "learning_rate": 4.0113946840868443e-07, "loss": 1.1189, "step": 36490 }, { "epoch": 91.2275, "grad_norm": 6.59375, "learning_rate": 4.0091247091592735e-07, "loss": 1.3799, "step": 36491 }, { "epoch": 91.23, "grad_norm": 8.9375, "learning_rate": 4.0068553635440264e-07, "loss": 1.3943, "step": 36492 }, { "epoch": 91.2325, "grad_norm": 7.625, "learning_rate": 4.0045866472560034e-07, "loss": 1.3971, "step": 36493 }, { "epoch": 91.235, "grad_norm": 7.0625, "learning_rate": 4.00231856031007e-07, "loss": 1.3262, "step": 36494 }, { "epoch": 91.2375, "grad_norm": 8.6875, "learning_rate": 4.0000511027211033e-07, "loss": 1.2133, "step": 36495 }, { "epoch": 91.24, "grad_norm": 8.25, "learning_rate": 3.9977842745039464e-07, "loss": 1.3481, "step": 36496 }, { "epoch": 91.2425, "grad_norm": 8.25, "learning_rate": 3.995518075673466e-07, "loss": 1.3511, "step": 36497 }, { "epoch": 91.245, "grad_norm": 7.28125, "learning_rate": 3.993252506244549e-07, "loss": 1.3955, "step": 36498 }, { "epoch": 91.2475, "grad_norm": 7.3125, "learning_rate": 3.9909875662320296e-07, "loss": 1.3083, "step": 36499 }, { "epoch": 91.25, "grad_norm": 9.875, "learning_rate": 3.988723255650728e-07, "loss": 1.3305, "step": 36500 }, { "epoch": 91.2525, "grad_norm": 8.5625, "learning_rate": 3.986459574515533e-07, "loss": 1.3106, "step": 36501 }, { "epoch": 91.255, "grad_norm": 7.9375, "learning_rate": 3.984196522841255e-07, "loss": 1.2692, "step": 36502 }, { "epoch": 91.2575, "grad_norm": 6.34375, "learning_rate": 3.981934100642759e-07, "loss": 1.1446, "step": 36503 }, { "epoch": 91.26, "grad_norm": 8.25, "learning_rate": 3.9796723079348455e-07, "loss": 1.2524, "step": 36504 }, { "epoch": 91.2625, "grad_norm": 5.78125, "learning_rate": 3.977411144732368e-07, "loss": 1.1296, "step": 36505 }, { "epoch": 91.265, "grad_norm": 8.0, "learning_rate": 3.9751506110501383e-07, "loss": 1.3188, "step": 36506 }, { "epoch": 91.2675, "grad_norm": 8.3125, "learning_rate": 3.9728907069029765e-07, "loss": 1.237, "step": 36507 }, { "epoch": 91.27, "grad_norm": 7.75, "learning_rate": 3.9706314323056936e-07, "loss": 1.2407, "step": 36508 }, { "epoch": 91.2725, "grad_norm": 6.21875, "learning_rate": 3.9683727872731116e-07, "loss": 1.2499, "step": 36509 }, { "epoch": 91.275, "grad_norm": 6.84375, "learning_rate": 3.966114771820051e-07, "loss": 1.2785, "step": 36510 }, { "epoch": 91.2775, "grad_norm": 6.59375, "learning_rate": 3.9638573859612783e-07, "loss": 1.4999, "step": 36511 }, { "epoch": 91.28, "grad_norm": 6.46875, "learning_rate": 3.961600629711615e-07, "loss": 1.4901, "step": 36512 }, { "epoch": 91.2825, "grad_norm": 6.53125, "learning_rate": 3.9593445030858604e-07, "loss": 1.3074, "step": 36513 }, { "epoch": 91.285, "grad_norm": 7.21875, "learning_rate": 3.957089006098802e-07, "loss": 1.3441, "step": 36514 }, { "epoch": 91.2875, "grad_norm": 6.15625, "learning_rate": 3.954834138765218e-07, "loss": 1.3206, "step": 36515 }, { "epoch": 91.29, "grad_norm": 6.5625, "learning_rate": 3.9525799010999064e-07, "loss": 1.3123, "step": 36516 }, { "epoch": 91.2925, "grad_norm": 6.875, "learning_rate": 3.9503262931176346e-07, "loss": 1.1996, "step": 36517 }, { "epoch": 91.295, "grad_norm": 9.375, "learning_rate": 3.948073314833178e-07, "loss": 1.2791, "step": 36518 }, { "epoch": 91.2975, "grad_norm": 6.71875, "learning_rate": 3.9458209662613047e-07, "loss": 1.4258, "step": 36519 }, { "epoch": 91.3, "grad_norm": 8.375, "learning_rate": 3.943569247416801e-07, "loss": 1.1846, "step": 36520 }, { "epoch": 91.3025, "grad_norm": 6.875, "learning_rate": 3.9413181583144e-07, "loss": 1.4368, "step": 36521 }, { "epoch": 91.305, "grad_norm": 6.9375, "learning_rate": 3.939067698968868e-07, "loss": 1.2279, "step": 36522 }, { "epoch": 91.3075, "grad_norm": 7.59375, "learning_rate": 3.936817869394971e-07, "loss": 1.3628, "step": 36523 }, { "epoch": 91.31, "grad_norm": 6.0625, "learning_rate": 3.9345686696074526e-07, "loss": 1.2882, "step": 36524 }, { "epoch": 91.3125, "grad_norm": 7.25, "learning_rate": 3.9323200996210673e-07, "loss": 1.3273, "step": 36525 }, { "epoch": 91.315, "grad_norm": 7.84375, "learning_rate": 3.9300721594505266e-07, "loss": 1.3451, "step": 36526 }, { "epoch": 91.3175, "grad_norm": 7.4375, "learning_rate": 3.9278248491105954e-07, "loss": 1.285, "step": 36527 }, { "epoch": 91.32, "grad_norm": 7.5625, "learning_rate": 3.925578168616007e-07, "loss": 1.2596, "step": 36528 }, { "epoch": 91.3225, "grad_norm": 6.96875, "learning_rate": 3.923332117981482e-07, "loss": 1.342, "step": 36529 }, { "epoch": 91.325, "grad_norm": 6.625, "learning_rate": 3.9210866972217433e-07, "loss": 1.4004, "step": 36530 }, { "epoch": 91.3275, "grad_norm": 10.25, "learning_rate": 3.9188419063515225e-07, "loss": 1.3812, "step": 36531 }, { "epoch": 91.33, "grad_norm": 7.5, "learning_rate": 3.91659774538552e-07, "loss": 1.2237, "step": 36532 }, { "epoch": 91.3325, "grad_norm": 5.6875, "learning_rate": 3.914354214338467e-07, "loss": 1.4618, "step": 36533 }, { "epoch": 91.335, "grad_norm": 6.0, "learning_rate": 3.912111313225042e-07, "loss": 1.4401, "step": 36534 }, { "epoch": 91.3375, "grad_norm": 8.8125, "learning_rate": 3.909869042059999e-07, "loss": 1.2901, "step": 36535 }, { "epoch": 91.34, "grad_norm": 9.375, "learning_rate": 3.907627400857994e-07, "loss": 1.3251, "step": 36536 }, { "epoch": 91.3425, "grad_norm": 7.21875, "learning_rate": 3.9053863896337365e-07, "loss": 1.5114, "step": 36537 }, { "epoch": 91.345, "grad_norm": 12.875, "learning_rate": 3.903146008401915e-07, "loss": 1.2899, "step": 36538 }, { "epoch": 91.3475, "grad_norm": 8.4375, "learning_rate": 3.9009062571772396e-07, "loss": 1.2404, "step": 36539 }, { "epoch": 91.35, "grad_norm": 9.0625, "learning_rate": 3.8986671359743767e-07, "loss": 1.2329, "step": 36540 }, { "epoch": 91.3525, "grad_norm": 8.0, "learning_rate": 3.8964286448080036e-07, "loss": 1.3614, "step": 36541 }, { "epoch": 91.355, "grad_norm": 4.90625, "learning_rate": 3.894190783692775e-07, "loss": 1.3272, "step": 36542 }, { "epoch": 91.3575, "grad_norm": 8.5625, "learning_rate": 3.8919535526434014e-07, "loss": 1.352, "step": 36543 }, { "epoch": 91.36, "grad_norm": 9.5, "learning_rate": 3.889716951674549e-07, "loss": 1.3711, "step": 36544 }, { "epoch": 91.3625, "grad_norm": 8.25, "learning_rate": 3.887480980800851e-07, "loss": 1.3172, "step": 36545 }, { "epoch": 91.365, "grad_norm": 8.625, "learning_rate": 3.8852456400369833e-07, "loss": 1.432, "step": 36546 }, { "epoch": 91.3675, "grad_norm": 6.96875, "learning_rate": 3.88301092939759e-07, "loss": 1.5555, "step": 36547 }, { "epoch": 91.37, "grad_norm": 10.125, "learning_rate": 3.88077684889735e-07, "loss": 1.394, "step": 36548 }, { "epoch": 91.3725, "grad_norm": 5.78125, "learning_rate": 3.8785433985508827e-07, "loss": 1.2864, "step": 36549 }, { "epoch": 91.375, "grad_norm": 9.6875, "learning_rate": 3.876310578372833e-07, "loss": 1.4458, "step": 36550 }, { "epoch": 91.3775, "grad_norm": 14.1875, "learning_rate": 3.8740783883778556e-07, "loss": 1.1899, "step": 36551 }, { "epoch": 91.38, "grad_norm": 13.8125, "learning_rate": 3.87184682858055e-07, "loss": 1.2889, "step": 36552 }, { "epoch": 91.3825, "grad_norm": 7.09375, "learning_rate": 3.869615898995582e-07, "loss": 1.2171, "step": 36553 }, { "epoch": 91.385, "grad_norm": 8.625, "learning_rate": 3.867385599637574e-07, "loss": 1.407, "step": 36554 }, { "epoch": 91.3875, "grad_norm": 6.40625, "learning_rate": 3.865155930521147e-07, "loss": 1.1285, "step": 36555 }, { "epoch": 91.39, "grad_norm": 13.0625, "learning_rate": 3.862926891660901e-07, "loss": 1.3967, "step": 36556 }, { "epoch": 91.3925, "grad_norm": 10.75, "learning_rate": 3.8606984830714346e-07, "loss": 1.3518, "step": 36557 }, { "epoch": 91.395, "grad_norm": 7.28125, "learning_rate": 3.858470704767403e-07, "loss": 1.2846, "step": 36558 }, { "epoch": 91.3975, "grad_norm": 7.0, "learning_rate": 3.856243556763395e-07, "loss": 1.3923, "step": 36559 }, { "epoch": 91.4, "grad_norm": 9.3125, "learning_rate": 3.8540170390740097e-07, "loss": 1.3389, "step": 36560 }, { "epoch": 91.4025, "grad_norm": 6.15625, "learning_rate": 3.851791151713824e-07, "loss": 1.4611, "step": 36561 }, { "epoch": 91.405, "grad_norm": 5.5, "learning_rate": 3.84956589469746e-07, "loss": 1.2858, "step": 36562 }, { "epoch": 91.4075, "grad_norm": 5.25, "learning_rate": 3.847341268039495e-07, "loss": 1.1927, "step": 36563 }, { "epoch": 91.41, "grad_norm": 10.5625, "learning_rate": 3.845117271754517e-07, "loss": 1.4936, "step": 36564 }, { "epoch": 91.4125, "grad_norm": 7.625, "learning_rate": 3.842893905857081e-07, "loss": 1.3016, "step": 36565 }, { "epoch": 91.415, "grad_norm": 7.0625, "learning_rate": 3.8406711703618095e-07, "loss": 1.1863, "step": 36566 }, { "epoch": 91.4175, "grad_norm": 8.4375, "learning_rate": 3.838449065283223e-07, "loss": 1.2887, "step": 36567 }, { "epoch": 91.42, "grad_norm": 9.9375, "learning_rate": 3.836227590635944e-07, "loss": 1.2386, "step": 36568 }, { "epoch": 91.4225, "grad_norm": 6.8125, "learning_rate": 3.834006746434482e-07, "loss": 1.1811, "step": 36569 }, { "epoch": 91.425, "grad_norm": 7.1875, "learning_rate": 3.831786532693449e-07, "loss": 1.4077, "step": 36570 }, { "epoch": 91.4275, "grad_norm": 4.6875, "learning_rate": 3.8295669494273656e-07, "loss": 1.2603, "step": 36571 }, { "epoch": 91.43, "grad_norm": 9.9375, "learning_rate": 3.8273479966507876e-07, "loss": 1.2323, "step": 36572 }, { "epoch": 91.4325, "grad_norm": 8.5625, "learning_rate": 3.8251296743782584e-07, "loss": 1.2466, "step": 36573 }, { "epoch": 91.435, "grad_norm": 15.4375, "learning_rate": 3.8229119826243555e-07, "loss": 1.3078, "step": 36574 }, { "epoch": 91.4375, "grad_norm": 7.09375, "learning_rate": 3.8206949214035785e-07, "loss": 1.3286, "step": 36575 }, { "epoch": 91.44, "grad_norm": 8.0625, "learning_rate": 3.8184784907304704e-07, "loss": 1.3178, "step": 36576 }, { "epoch": 91.4425, "grad_norm": 6.78125, "learning_rate": 3.8162626906195763e-07, "loss": 1.2708, "step": 36577 }, { "epoch": 91.445, "grad_norm": 11.4375, "learning_rate": 3.8140475210854176e-07, "loss": 1.4972, "step": 36578 }, { "epoch": 91.4475, "grad_norm": 6.96875, "learning_rate": 3.811832982142516e-07, "loss": 1.4542, "step": 36579 }, { "epoch": 91.45, "grad_norm": 9.6875, "learning_rate": 3.8096190738053815e-07, "loss": 1.422, "step": 36580 }, { "epoch": 91.4525, "grad_norm": 7.09375, "learning_rate": 3.807405796088537e-07, "loss": 1.5497, "step": 36581 }, { "epoch": 91.455, "grad_norm": 8.6875, "learning_rate": 3.805193149006492e-07, "loss": 1.3204, "step": 36582 }, { "epoch": 91.4575, "grad_norm": 7.4375, "learning_rate": 3.8029811325737464e-07, "loss": 1.2375, "step": 36583 }, { "epoch": 91.46, "grad_norm": 7.40625, "learning_rate": 3.800769746804811e-07, "loss": 1.3194, "step": 36584 }, { "epoch": 91.4625, "grad_norm": 9.8125, "learning_rate": 3.798558991714185e-07, "loss": 1.3223, "step": 36585 }, { "epoch": 91.465, "grad_norm": 6.5, "learning_rate": 3.796348867316346e-07, "loss": 1.3409, "step": 36586 }, { "epoch": 91.4675, "grad_norm": 6.40625, "learning_rate": 3.7941393736257936e-07, "loss": 1.3533, "step": 36587 }, { "epoch": 91.47, "grad_norm": 8.25, "learning_rate": 3.791930510657016e-07, "loss": 1.1636, "step": 36588 }, { "epoch": 91.4725, "grad_norm": 9.75, "learning_rate": 3.789722278424501e-07, "loss": 1.1962, "step": 36589 }, { "epoch": 91.475, "grad_norm": 9.5, "learning_rate": 3.787514676942716e-07, "loss": 1.3306, "step": 36590 }, { "epoch": 91.4775, "grad_norm": 6.625, "learning_rate": 3.7853077062261154e-07, "loss": 1.3077, "step": 36591 }, { "epoch": 91.48, "grad_norm": 5.9375, "learning_rate": 3.783101366289199e-07, "loss": 1.2353, "step": 36592 }, { "epoch": 91.4825, "grad_norm": 7.625, "learning_rate": 3.7808956571464215e-07, "loss": 1.287, "step": 36593 }, { "epoch": 91.485, "grad_norm": 8.3125, "learning_rate": 3.7786905788122497e-07, "loss": 1.5156, "step": 36594 }, { "epoch": 91.4875, "grad_norm": 8.1875, "learning_rate": 3.7764861313011156e-07, "loss": 1.3613, "step": 36595 }, { "epoch": 91.49, "grad_norm": 8.75, "learning_rate": 3.7742823146274977e-07, "loss": 1.2361, "step": 36596 }, { "epoch": 91.4925, "grad_norm": 7.9375, "learning_rate": 3.772079128805828e-07, "loss": 1.3171, "step": 36597 }, { "epoch": 91.495, "grad_norm": 7.15625, "learning_rate": 3.769876573850573e-07, "loss": 1.303, "step": 36598 }, { "epoch": 91.4975, "grad_norm": 9.125, "learning_rate": 3.767674649776132e-07, "loss": 1.42, "step": 36599 }, { "epoch": 91.5, "grad_norm": 6.15625, "learning_rate": 3.7654733565969826e-07, "loss": 1.3059, "step": 36600 }, { "epoch": 91.5025, "grad_norm": 12.875, "learning_rate": 3.7632726943275356e-07, "loss": 1.3545, "step": 36601 }, { "epoch": 91.505, "grad_norm": 9.375, "learning_rate": 3.761072662982213e-07, "loss": 1.3612, "step": 36602 }, { "epoch": 91.5075, "grad_norm": 7.5, "learning_rate": 3.7588732625754466e-07, "loss": 1.3451, "step": 36603 }, { "epoch": 91.51, "grad_norm": 10.625, "learning_rate": 3.7566744931216704e-07, "loss": 1.2357, "step": 36604 }, { "epoch": 91.5125, "grad_norm": 10.875, "learning_rate": 3.7544763546352834e-07, "loss": 1.2609, "step": 36605 }, { "epoch": 91.515, "grad_norm": 5.09375, "learning_rate": 3.752278847130686e-07, "loss": 1.3999, "step": 36606 }, { "epoch": 91.5175, "grad_norm": 6.875, "learning_rate": 3.750081970622288e-07, "loss": 1.3376, "step": 36607 }, { "epoch": 91.52, "grad_norm": 7.625, "learning_rate": 3.747885725124523e-07, "loss": 1.2944, "step": 36608 }, { "epoch": 91.5225, "grad_norm": 6.0625, "learning_rate": 3.7456901106517676e-07, "loss": 1.2987, "step": 36609 }, { "epoch": 91.525, "grad_norm": 7.96875, "learning_rate": 3.7434951272183993e-07, "loss": 1.2884, "step": 36610 }, { "epoch": 91.5275, "grad_norm": 6.78125, "learning_rate": 3.741300774838841e-07, "loss": 1.2705, "step": 36611 }, { "epoch": 91.53, "grad_norm": 6.3125, "learning_rate": 3.7391070535274465e-07, "loss": 1.1093, "step": 36612 }, { "epoch": 91.5325, "grad_norm": 7.15625, "learning_rate": 3.7369139632986273e-07, "loss": 1.3152, "step": 36613 }, { "epoch": 91.535, "grad_norm": 7.1875, "learning_rate": 3.7347215041667386e-07, "loss": 1.2777, "step": 36614 }, { "epoch": 91.5375, "grad_norm": 7.5625, "learning_rate": 3.7325296761461685e-07, "loss": 1.3239, "step": 36615 }, { "epoch": 91.54, "grad_norm": 9.5, "learning_rate": 3.730338479251283e-07, "loss": 1.2371, "step": 36616 }, { "epoch": 91.5425, "grad_norm": 9.9375, "learning_rate": 3.7281479134964385e-07, "loss": 1.2098, "step": 36617 }, { "epoch": 91.545, "grad_norm": 8.1875, "learning_rate": 3.725957978896e-07, "loss": 1.276, "step": 36618 }, { "epoch": 91.5475, "grad_norm": 7.125, "learning_rate": 3.7237686754643453e-07, "loss": 1.3747, "step": 36619 }, { "epoch": 91.55, "grad_norm": 7.5, "learning_rate": 3.721580003215808e-07, "loss": 1.2233, "step": 36620 }, { "epoch": 91.5525, "grad_norm": 5.0625, "learning_rate": 3.719391962164731e-07, "loss": 1.3396, "step": 36621 }, { "epoch": 91.555, "grad_norm": 6.375, "learning_rate": 3.7172045523254597e-07, "loss": 1.3734, "step": 36622 }, { "epoch": 91.5575, "grad_norm": 7.84375, "learning_rate": 3.7150177737123596e-07, "loss": 1.096, "step": 36623 }, { "epoch": 91.56, "grad_norm": 6.1875, "learning_rate": 3.712831626339752e-07, "loss": 1.5167, "step": 36624 }, { "epoch": 91.5625, "grad_norm": 7.5, "learning_rate": 3.7106461102219495e-07, "loss": 1.2782, "step": 36625 }, { "epoch": 91.565, "grad_norm": 8.5625, "learning_rate": 3.708461225373305e-07, "loss": 1.1148, "step": 36626 }, { "epoch": 91.5675, "grad_norm": 9.8125, "learning_rate": 3.7062769718081315e-07, "loss": 1.5447, "step": 36627 }, { "epoch": 91.57, "grad_norm": 9.0, "learning_rate": 3.7040933495407606e-07, "loss": 1.3524, "step": 36628 }, { "epoch": 91.5725, "grad_norm": 7.21875, "learning_rate": 3.701910358585492e-07, "loss": 1.2849, "step": 36629 }, { "epoch": 91.575, "grad_norm": 10.3125, "learning_rate": 3.6997279989566594e-07, "loss": 1.3326, "step": 36630 }, { "epoch": 91.5775, "grad_norm": 8.625, "learning_rate": 3.6975462706685503e-07, "loss": 1.525, "step": 36631 }, { "epoch": 91.58, "grad_norm": 7.84375, "learning_rate": 3.695365173735466e-07, "loss": 1.2646, "step": 36632 }, { "epoch": 91.5825, "grad_norm": 6.59375, "learning_rate": 3.693184708171715e-07, "loss": 1.2775, "step": 36633 }, { "epoch": 91.585, "grad_norm": 7.25, "learning_rate": 3.6910048739915994e-07, "loss": 1.4172, "step": 36634 }, { "epoch": 91.5875, "grad_norm": 5.875, "learning_rate": 3.6888256712094063e-07, "loss": 1.5372, "step": 36635 }, { "epoch": 91.59, "grad_norm": 9.0625, "learning_rate": 3.686647099839402e-07, "loss": 1.4329, "step": 36636 }, { "epoch": 91.5925, "grad_norm": 13.25, "learning_rate": 3.684469159895887e-07, "loss": 1.401, "step": 36637 }, { "epoch": 91.595, "grad_norm": 7.09375, "learning_rate": 3.682291851393149e-07, "loss": 1.3034, "step": 36638 }, { "epoch": 91.5975, "grad_norm": 6.8125, "learning_rate": 3.6801151743454444e-07, "loss": 1.4045, "step": 36639 }, { "epoch": 91.6, "grad_norm": 8.4375, "learning_rate": 3.67793912876705e-07, "loss": 1.4448, "step": 36640 }, { "epoch": 91.6025, "grad_norm": 9.125, "learning_rate": 3.6757637146722425e-07, "loss": 1.3947, "step": 36641 }, { "epoch": 91.605, "grad_norm": 5.46875, "learning_rate": 3.673588932075256e-07, "loss": 1.2157, "step": 36642 }, { "epoch": 91.6075, "grad_norm": 6.125, "learning_rate": 3.671414780990379e-07, "loss": 1.2527, "step": 36643 }, { "epoch": 91.61, "grad_norm": 5.40625, "learning_rate": 3.669241261431833e-07, "loss": 1.3811, "step": 36644 }, { "epoch": 91.6125, "grad_norm": 9.6875, "learning_rate": 3.667068373413907e-07, "loss": 1.1712, "step": 36645 }, { "epoch": 91.615, "grad_norm": 6.90625, "learning_rate": 3.6648961169508224e-07, "loss": 1.3959, "step": 36646 }, { "epoch": 91.6175, "grad_norm": 7.65625, "learning_rate": 3.662724492056813e-07, "loss": 1.4074, "step": 36647 }, { "epoch": 91.62, "grad_norm": 6.46875, "learning_rate": 3.6605534987461335e-07, "loss": 1.1508, "step": 36648 }, { "epoch": 91.6225, "grad_norm": 7.75, "learning_rate": 3.658383137033017e-07, "loss": 1.4156, "step": 36649 }, { "epoch": 91.625, "grad_norm": 11.3125, "learning_rate": 3.6562134069316857e-07, "loss": 1.3435, "step": 36650 }, { "epoch": 91.6275, "grad_norm": 7.3125, "learning_rate": 3.6540443084563503e-07, "loss": 1.4393, "step": 36651 }, { "epoch": 91.63, "grad_norm": 8.875, "learning_rate": 3.651875841621244e-07, "loss": 1.2918, "step": 36652 }, { "epoch": 91.6325, "grad_norm": 7.25, "learning_rate": 3.649708006440589e-07, "loss": 1.3184, "step": 36653 }, { "epoch": 91.635, "grad_norm": 9.1875, "learning_rate": 3.6475408029286065e-07, "loss": 1.2539, "step": 36654 }, { "epoch": 91.6375, "grad_norm": 9.0625, "learning_rate": 3.6453742310994745e-07, "loss": 1.3955, "step": 36655 }, { "epoch": 91.64, "grad_norm": 9.625, "learning_rate": 3.643208290967415e-07, "loss": 1.4225, "step": 36656 }, { "epoch": 91.6425, "grad_norm": 8.25, "learning_rate": 3.6410429825466274e-07, "loss": 1.4239, "step": 36657 }, { "epoch": 91.645, "grad_norm": 8.3125, "learning_rate": 3.6388783058513123e-07, "loss": 1.2648, "step": 36658 }, { "epoch": 91.6475, "grad_norm": 7.875, "learning_rate": 3.6367142608956464e-07, "loss": 1.3952, "step": 36659 }, { "epoch": 91.65, "grad_norm": 7.71875, "learning_rate": 3.6345508476938296e-07, "loss": 1.2344, "step": 36660 }, { "epoch": 91.6525, "grad_norm": 8.375, "learning_rate": 3.6323880662600507e-07, "loss": 1.214, "step": 36661 }, { "epoch": 91.655, "grad_norm": 7.0, "learning_rate": 3.630225916608454e-07, "loss": 1.2691, "step": 36662 }, { "epoch": 91.6575, "grad_norm": 8.0, "learning_rate": 3.6280643987532507e-07, "loss": 1.3222, "step": 36663 }, { "epoch": 91.66, "grad_norm": 11.9375, "learning_rate": 3.6259035127086173e-07, "loss": 1.424, "step": 36664 }, { "epoch": 91.6625, "grad_norm": 6.5625, "learning_rate": 3.623743258488688e-07, "loss": 1.4647, "step": 36665 }, { "epoch": 91.665, "grad_norm": 6.875, "learning_rate": 3.6215836361076396e-07, "loss": 1.1664, "step": 36666 }, { "epoch": 91.6675, "grad_norm": 9.375, "learning_rate": 3.619424645579628e-07, "loss": 1.2477, "step": 36667 }, { "epoch": 91.67, "grad_norm": 8.0625, "learning_rate": 3.6172662869188303e-07, "loss": 1.2552, "step": 36668 }, { "epoch": 91.6725, "grad_norm": 7.03125, "learning_rate": 3.6151085601393686e-07, "loss": 1.2458, "step": 36669 }, { "epoch": 91.675, "grad_norm": 6.65625, "learning_rate": 3.6129514652553985e-07, "loss": 1.3198, "step": 36670 }, { "epoch": 91.6775, "grad_norm": 7.78125, "learning_rate": 3.6107950022810644e-07, "loss": 1.3864, "step": 36671 }, { "epoch": 91.68, "grad_norm": 6.9375, "learning_rate": 3.608639171230488e-07, "loss": 1.3614, "step": 36672 }, { "epoch": 91.6825, "grad_norm": 5.84375, "learning_rate": 3.606483972117825e-07, "loss": 1.2709, "step": 36673 }, { "epoch": 91.685, "grad_norm": 7.03125, "learning_rate": 3.6043294049571966e-07, "loss": 1.1756, "step": 36674 }, { "epoch": 91.6875, "grad_norm": 6.3125, "learning_rate": 3.602175469762725e-07, "loss": 1.2161, "step": 36675 }, { "epoch": 91.69, "grad_norm": 9.75, "learning_rate": 3.6000221665485445e-07, "loss": 1.4126, "step": 36676 }, { "epoch": 91.6925, "grad_norm": 5.71875, "learning_rate": 3.597869495328743e-07, "loss": 1.2243, "step": 36677 }, { "epoch": 91.695, "grad_norm": 8.125, "learning_rate": 3.5957174561174537e-07, "loss": 1.2116, "step": 36678 }, { "epoch": 91.6975, "grad_norm": 7.625, "learning_rate": 3.5935660489287874e-07, "loss": 1.178, "step": 36679 }, { "epoch": 91.7, "grad_norm": 6.5625, "learning_rate": 3.591415273776855e-07, "loss": 1.4758, "step": 36680 }, { "epoch": 91.7025, "grad_norm": 8.0625, "learning_rate": 3.5892651306757343e-07, "loss": 1.3219, "step": 36681 }, { "epoch": 91.705, "grad_norm": 8.4375, "learning_rate": 3.587115619639525e-07, "loss": 1.4935, "step": 36682 }, { "epoch": 91.7075, "grad_norm": 10.8125, "learning_rate": 3.5849667406823384e-07, "loss": 1.2922, "step": 36683 }, { "epoch": 91.71, "grad_norm": 7.5, "learning_rate": 3.582818493818252e-07, "loss": 1.312, "step": 36684 }, { "epoch": 91.7125, "grad_norm": 6.4375, "learning_rate": 3.580670879061354e-07, "loss": 1.2844, "step": 36685 }, { "epoch": 91.715, "grad_norm": 9.4375, "learning_rate": 3.578523896425712e-07, "loss": 1.318, "step": 36686 }, { "epoch": 91.7175, "grad_norm": 8.0625, "learning_rate": 3.5763775459254023e-07, "loss": 1.4686, "step": 36687 }, { "epoch": 91.72, "grad_norm": 7.25, "learning_rate": 3.5742318275745147e-07, "loss": 1.3403, "step": 36688 }, { "epoch": 91.7225, "grad_norm": 8.1875, "learning_rate": 3.572086741387093e-07, "loss": 1.2961, "step": 36689 }, { "epoch": 91.725, "grad_norm": 8.25, "learning_rate": 3.569942287377226e-07, "loss": 1.3087, "step": 36690 }, { "epoch": 91.7275, "grad_norm": 5.96875, "learning_rate": 3.567798465558947e-07, "loss": 1.3768, "step": 36691 }, { "epoch": 91.73, "grad_norm": 6.1875, "learning_rate": 3.5656552759463227e-07, "loss": 1.329, "step": 36692 }, { "epoch": 91.7325, "grad_norm": 5.96875, "learning_rate": 3.563512718553397e-07, "loss": 1.3431, "step": 36693 }, { "epoch": 91.735, "grad_norm": 7.75, "learning_rate": 3.5613707933942474e-07, "loss": 1.3265, "step": 36694 }, { "epoch": 91.7375, "grad_norm": 6.84375, "learning_rate": 3.559229500482875e-07, "loss": 1.3742, "step": 36695 }, { "epoch": 91.74, "grad_norm": 8.8125, "learning_rate": 3.557088839833345e-07, "loss": 1.3627, "step": 36696 }, { "epoch": 91.7425, "grad_norm": 9.1875, "learning_rate": 3.5549488114596573e-07, "loss": 1.3547, "step": 36697 }, { "epoch": 91.745, "grad_norm": 6.125, "learning_rate": 3.5528094153758795e-07, "loss": 1.2507, "step": 36698 }, { "epoch": 91.7475, "grad_norm": 4.84375, "learning_rate": 3.550670651596033e-07, "loss": 1.27, "step": 36699 }, { "epoch": 91.75, "grad_norm": 8.1875, "learning_rate": 3.548532520134129e-07, "loss": 1.2791, "step": 36700 }, { "epoch": 91.7525, "grad_norm": 7.40625, "learning_rate": 3.5463950210041676e-07, "loss": 1.324, "step": 36701 }, { "epoch": 91.755, "grad_norm": 8.125, "learning_rate": 3.544258154220193e-07, "loss": 1.2934, "step": 36702 }, { "epoch": 91.7575, "grad_norm": 9.125, "learning_rate": 3.5421219197962044e-07, "loss": 1.2991, "step": 36703 }, { "epoch": 91.76, "grad_norm": 6.84375, "learning_rate": 3.5399863177462024e-07, "loss": 1.3465, "step": 36704 }, { "epoch": 91.7625, "grad_norm": 7.46875, "learning_rate": 3.537851348084187e-07, "loss": 1.4176, "step": 36705 }, { "epoch": 91.765, "grad_norm": 5.1875, "learning_rate": 3.5357170108241576e-07, "loss": 1.2309, "step": 36706 }, { "epoch": 91.7675, "grad_norm": 9.8125, "learning_rate": 3.533583305980104e-07, "loss": 1.261, "step": 36707 }, { "epoch": 91.77, "grad_norm": 6.71875, "learning_rate": 3.5314502335660363e-07, "loss": 1.3122, "step": 36708 }, { "epoch": 91.7725, "grad_norm": 8.75, "learning_rate": 3.529317793595899e-07, "loss": 1.1804, "step": 36709 }, { "epoch": 91.775, "grad_norm": 5.8125, "learning_rate": 3.5271859860837034e-07, "loss": 1.4202, "step": 36710 }, { "epoch": 91.7775, "grad_norm": 5.09375, "learning_rate": 3.525054811043416e-07, "loss": 1.2128, "step": 36711 }, { "epoch": 91.78, "grad_norm": 5.1875, "learning_rate": 3.522924268489003e-07, "loss": 1.2547, "step": 36712 }, { "epoch": 91.7825, "grad_norm": 8.125, "learning_rate": 3.520794358434443e-07, "loss": 1.4282, "step": 36713 }, { "epoch": 91.785, "grad_norm": 8.0, "learning_rate": 3.5186650808936904e-07, "loss": 1.401, "step": 36714 }, { "epoch": 91.7875, "grad_norm": 6.78125, "learning_rate": 3.516536435880724e-07, "loss": 1.3164, "step": 36715 }, { "epoch": 91.79, "grad_norm": 7.96875, "learning_rate": 3.5144084234094544e-07, "loss": 1.3103, "step": 36716 }, { "epoch": 91.7925, "grad_norm": 6.125, "learning_rate": 3.5122810434938813e-07, "loss": 1.3482, "step": 36717 }, { "epoch": 91.795, "grad_norm": 15.5, "learning_rate": 3.5101542961479273e-07, "loss": 1.4594, "step": 36718 }, { "epoch": 91.7975, "grad_norm": 11.25, "learning_rate": 3.508028181385548e-07, "loss": 1.3713, "step": 36719 }, { "epoch": 91.8, "grad_norm": 6.5, "learning_rate": 3.5059026992206645e-07, "loss": 1.3031, "step": 36720 }, { "epoch": 91.8025, "grad_norm": 7.21875, "learning_rate": 3.5037778496672336e-07, "loss": 1.2976, "step": 36721 }, { "epoch": 91.805, "grad_norm": 5.21875, "learning_rate": 3.501653632739155e-07, "loss": 1.2729, "step": 36722 }, { "epoch": 91.8075, "grad_norm": 9.875, "learning_rate": 3.499530048450395e-07, "loss": 1.4521, "step": 36723 }, { "epoch": 91.81, "grad_norm": 7.6875, "learning_rate": 3.497407096814842e-07, "loss": 1.3087, "step": 36724 }, { "epoch": 91.8125, "grad_norm": 8.9375, "learning_rate": 3.4952847778464306e-07, "loss": 1.2246, "step": 36725 }, { "epoch": 91.815, "grad_norm": 6.0625, "learning_rate": 3.4931630915590707e-07, "loss": 1.4404, "step": 36726 }, { "epoch": 91.8175, "grad_norm": 10.625, "learning_rate": 3.491042037966663e-07, "loss": 1.2312, "step": 36727 }, { "epoch": 91.82, "grad_norm": 10.875, "learning_rate": 3.4889216170831187e-07, "loss": 1.3727, "step": 36728 }, { "epoch": 91.8225, "grad_norm": 6.1875, "learning_rate": 3.486801828922359e-07, "loss": 1.362, "step": 36729 }, { "epoch": 91.825, "grad_norm": 8.4375, "learning_rate": 3.484682673498263e-07, "loss": 1.3877, "step": 36730 }, { "epoch": 91.8275, "grad_norm": 7.6875, "learning_rate": 3.4825641508247077e-07, "loss": 1.4271, "step": 36731 }, { "epoch": 91.83, "grad_norm": 6.71875, "learning_rate": 3.4804462609156043e-07, "loss": 1.4062, "step": 36732 }, { "epoch": 91.8325, "grad_norm": 7.0625, "learning_rate": 3.478329003784842e-07, "loss": 1.2923, "step": 36733 }, { "epoch": 91.835, "grad_norm": 6.25, "learning_rate": 3.4762123794462975e-07, "loss": 1.3756, "step": 36734 }, { "epoch": 91.8375, "grad_norm": 9.625, "learning_rate": 3.474096387913828e-07, "loss": 1.3303, "step": 36735 }, { "epoch": 91.84, "grad_norm": 8.0, "learning_rate": 3.4719810292013214e-07, "loss": 1.2195, "step": 36736 }, { "epoch": 91.8425, "grad_norm": 7.34375, "learning_rate": 3.469866303322644e-07, "loss": 1.3833, "step": 36737 }, { "epoch": 91.845, "grad_norm": 5.84375, "learning_rate": 3.467752210291675e-07, "loss": 1.2566, "step": 36738 }, { "epoch": 91.8475, "grad_norm": 6.59375, "learning_rate": 3.4656387501222356e-07, "loss": 1.4206, "step": 36739 }, { "epoch": 91.85, "grad_norm": 6.75, "learning_rate": 3.4635259228282256e-07, "loss": 1.3213, "step": 36740 }, { "epoch": 91.8525, "grad_norm": 6.875, "learning_rate": 3.4614137284234685e-07, "loss": 1.29, "step": 36741 }, { "epoch": 91.855, "grad_norm": 8.3125, "learning_rate": 3.4593021669218184e-07, "loss": 1.3605, "step": 36742 }, { "epoch": 91.8575, "grad_norm": 8.5, "learning_rate": 3.457191238337121e-07, "loss": 1.1606, "step": 36743 }, { "epoch": 91.86, "grad_norm": 7.84375, "learning_rate": 3.45508094268322e-07, "loss": 1.2918, "step": 36744 }, { "epoch": 91.8625, "grad_norm": 11.8125, "learning_rate": 3.452971279973949e-07, "loss": 1.3244, "step": 36745 }, { "epoch": 91.865, "grad_norm": 6.875, "learning_rate": 3.4508622502231195e-07, "loss": 1.1662, "step": 36746 }, { "epoch": 91.8675, "grad_norm": 9.8125, "learning_rate": 3.4487538534445753e-07, "loss": 1.383, "step": 36747 }, { "epoch": 91.87, "grad_norm": 6.0, "learning_rate": 3.446646089652139e-07, "loss": 1.3032, "step": 36748 }, { "epoch": 91.8725, "grad_norm": 6.9375, "learning_rate": 3.4445389588596336e-07, "loss": 1.3489, "step": 36749 }, { "epoch": 91.875, "grad_norm": 9.625, "learning_rate": 3.442432461080858e-07, "loss": 1.2745, "step": 36750 }, { "epoch": 91.8775, "grad_norm": 7.34375, "learning_rate": 3.4403265963296353e-07, "loss": 1.4398, "step": 36751 }, { "epoch": 91.88, "grad_norm": 8.1875, "learning_rate": 3.438221364619776e-07, "loss": 1.2091, "step": 36752 }, { "epoch": 91.8825, "grad_norm": 6.75, "learning_rate": 3.4361167659650695e-07, "loss": 1.1936, "step": 36753 }, { "epoch": 91.885, "grad_norm": 8.5625, "learning_rate": 3.434012800379305e-07, "loss": 1.3769, "step": 36754 }, { "epoch": 91.8875, "grad_norm": 6.8125, "learning_rate": 3.4319094678763046e-07, "loss": 1.3386, "step": 36755 }, { "epoch": 91.89, "grad_norm": 7.96875, "learning_rate": 3.429806768469823e-07, "loss": 1.2201, "step": 36756 }, { "epoch": 91.8925, "grad_norm": 10.5625, "learning_rate": 3.427704702173673e-07, "loss": 1.3148, "step": 36757 }, { "epoch": 91.895, "grad_norm": 7.34375, "learning_rate": 3.4256032690016207e-07, "loss": 1.2068, "step": 36758 }, { "epoch": 91.8975, "grad_norm": 7.40625, "learning_rate": 3.4235024689674657e-07, "loss": 1.4632, "step": 36759 }, { "epoch": 91.9, "grad_norm": 13.1875, "learning_rate": 3.421402302084953e-07, "loss": 1.306, "step": 36760 }, { "epoch": 91.9025, "grad_norm": 19.375, "learning_rate": 3.4193027683678606e-07, "loss": 1.2671, "step": 36761 }, { "epoch": 91.905, "grad_norm": 10.5625, "learning_rate": 3.4172038678299545e-07, "loss": 1.484, "step": 36762 }, { "epoch": 91.9075, "grad_norm": 8.0625, "learning_rate": 3.4151056004849916e-07, "loss": 1.3764, "step": 36763 }, { "epoch": 91.91, "grad_norm": 6.59375, "learning_rate": 3.4130079663467485e-07, "loss": 1.2376, "step": 36764 }, { "epoch": 91.9125, "grad_norm": 9.125, "learning_rate": 3.4109109654289375e-07, "loss": 1.1877, "step": 36765 }, { "epoch": 91.915, "grad_norm": 7.21875, "learning_rate": 3.4088145977453356e-07, "loss": 1.3334, "step": 36766 }, { "epoch": 91.9175, "grad_norm": 8.625, "learning_rate": 3.406718863309688e-07, "loss": 1.4835, "step": 36767 }, { "epoch": 91.92, "grad_norm": 7.25, "learning_rate": 3.404623762135728e-07, "loss": 1.3025, "step": 36768 }, { "epoch": 91.9225, "grad_norm": 6.0, "learning_rate": 3.4025292942371667e-07, "loss": 1.2597, "step": 36769 }, { "epoch": 91.925, "grad_norm": 7.03125, "learning_rate": 3.400435459627782e-07, "loss": 1.3167, "step": 36770 }, { "epoch": 91.9275, "grad_norm": 7.1875, "learning_rate": 3.3983422583212635e-07, "loss": 1.3451, "step": 36771 }, { "epoch": 91.93, "grad_norm": 9.375, "learning_rate": 3.3962496903313544e-07, "loss": 1.2736, "step": 36772 }, { "epoch": 91.9325, "grad_norm": 10.6875, "learning_rate": 3.394157755671756e-07, "loss": 1.3477, "step": 36773 }, { "epoch": 91.935, "grad_norm": 9.75, "learning_rate": 3.392066454356213e-07, "loss": 1.4423, "step": 36774 }, { "epoch": 91.9375, "grad_norm": 7.1875, "learning_rate": 3.389975786398403e-07, "loss": 1.2404, "step": 36775 }, { "epoch": 91.94, "grad_norm": 10.1875, "learning_rate": 3.387885751812048e-07, "loss": 1.2423, "step": 36776 }, { "epoch": 91.9425, "grad_norm": 6.875, "learning_rate": 3.3857963506108373e-07, "loss": 1.4195, "step": 36777 }, { "epoch": 91.945, "grad_norm": 7.59375, "learning_rate": 3.383707582808493e-07, "loss": 1.3249, "step": 36778 }, { "epoch": 91.9475, "grad_norm": 5.40625, "learning_rate": 3.381619448418694e-07, "loss": 1.3307, "step": 36779 }, { "epoch": 91.95, "grad_norm": 6.8125, "learning_rate": 3.379531947455128e-07, "loss": 1.3479, "step": 36780 }, { "epoch": 91.9525, "grad_norm": 8.5625, "learning_rate": 3.377445079931474e-07, "loss": 1.1031, "step": 36781 }, { "epoch": 91.955, "grad_norm": 6.125, "learning_rate": 3.375358845861443e-07, "loss": 1.5232, "step": 36782 }, { "epoch": 91.9575, "grad_norm": 5.53125, "learning_rate": 3.37327324525869e-07, "loss": 1.4194, "step": 36783 }, { "epoch": 91.96, "grad_norm": 8.125, "learning_rate": 3.371188278136883e-07, "loss": 1.3509, "step": 36784 }, { "epoch": 91.9625, "grad_norm": 6.6875, "learning_rate": 3.3691039445096994e-07, "loss": 1.3828, "step": 36785 }, { "epoch": 91.965, "grad_norm": 6.75, "learning_rate": 3.367020244390806e-07, "loss": 1.3528, "step": 36786 }, { "epoch": 91.9675, "grad_norm": 10.125, "learning_rate": 3.364937177793859e-07, "loss": 1.3451, "step": 36787 }, { "epoch": 91.97, "grad_norm": 6.1875, "learning_rate": 3.3628547447325133e-07, "loss": 1.3263, "step": 36788 }, { "epoch": 91.9725, "grad_norm": 7.0625, "learning_rate": 3.360772945220436e-07, "loss": 1.5441, "step": 36789 }, { "epoch": 91.975, "grad_norm": 7.0625, "learning_rate": 3.3586917792712614e-07, "loss": 1.2922, "step": 36790 }, { "epoch": 91.9775, "grad_norm": 7.65625, "learning_rate": 3.356611246898622e-07, "loss": 1.207, "step": 36791 }, { "epoch": 91.98, "grad_norm": 8.875, "learning_rate": 3.3545313481161743e-07, "loss": 1.3345, "step": 36792 }, { "epoch": 91.9825, "grad_norm": 8.6875, "learning_rate": 3.352452082937563e-07, "loss": 1.4366, "step": 36793 }, { "epoch": 91.985, "grad_norm": 5.78125, "learning_rate": 3.3503734513764097e-07, "loss": 1.2803, "step": 36794 }, { "epoch": 91.9875, "grad_norm": 7.84375, "learning_rate": 3.3482954534463264e-07, "loss": 1.2686, "step": 36795 }, { "epoch": 91.99, "grad_norm": 9.6875, "learning_rate": 3.3462180891609573e-07, "loss": 1.5183, "step": 36796 }, { "epoch": 91.9925, "grad_norm": 7.0625, "learning_rate": 3.344141358533914e-07, "loss": 1.3229, "step": 36797 }, { "epoch": 91.995, "grad_norm": 6.8125, "learning_rate": 3.3420652615788194e-07, "loss": 1.2395, "step": 36798 }, { "epoch": 91.9975, "grad_norm": 8.5, "learning_rate": 3.339989798309273e-07, "loss": 1.2953, "step": 36799 }, { "epoch": 92.0, "grad_norm": 7.03125, "learning_rate": 3.3379149687388866e-07, "loss": 1.3427, "step": 36800 }, { "epoch": 92.0025, "grad_norm": 4.96875, "learning_rate": 3.335840772881249e-07, "loss": 1.1784, "step": 36801 }, { "epoch": 92.005, "grad_norm": 11.0625, "learning_rate": 3.333767210749983e-07, "loss": 1.3326, "step": 36802 }, { "epoch": 92.0075, "grad_norm": 9.0625, "learning_rate": 3.331694282358666e-07, "loss": 1.2964, "step": 36803 }, { "epoch": 92.01, "grad_norm": 7.25, "learning_rate": 3.329621987720899e-07, "loss": 1.3352, "step": 36804 }, { "epoch": 92.0125, "grad_norm": 9.25, "learning_rate": 3.3275503268502593e-07, "loss": 1.1909, "step": 36805 }, { "epoch": 92.015, "grad_norm": 6.53125, "learning_rate": 3.3254792997603257e-07, "loss": 1.4142, "step": 36806 }, { "epoch": 92.0175, "grad_norm": 8.875, "learning_rate": 3.323408906464676e-07, "loss": 1.2676, "step": 36807 }, { "epoch": 92.02, "grad_norm": 5.875, "learning_rate": 3.32133914697691e-07, "loss": 1.3376, "step": 36808 }, { "epoch": 92.0225, "grad_norm": 6.25, "learning_rate": 3.319270021310561e-07, "loss": 1.3975, "step": 36809 }, { "epoch": 92.025, "grad_norm": 10.25, "learning_rate": 3.317201529479208e-07, "loss": 1.3414, "step": 36810 }, { "epoch": 92.0275, "grad_norm": 6.8125, "learning_rate": 3.3151336714964067e-07, "loss": 1.3697, "step": 36811 }, { "epoch": 92.03, "grad_norm": 8.6875, "learning_rate": 3.3130664473757344e-07, "loss": 1.3145, "step": 36812 }, { "epoch": 92.0325, "grad_norm": 7.5625, "learning_rate": 3.3109998571307367e-07, "loss": 1.2827, "step": 36813 }, { "epoch": 92.035, "grad_norm": 6.5625, "learning_rate": 3.3089339007749354e-07, "loss": 1.2014, "step": 36814 }, { "epoch": 92.0375, "grad_norm": 6.5625, "learning_rate": 3.306868578321898e-07, "loss": 1.4151, "step": 36815 }, { "epoch": 92.04, "grad_norm": 5.90625, "learning_rate": 3.3048038897851576e-07, "loss": 1.5052, "step": 36816 }, { "epoch": 92.0425, "grad_norm": 6.875, "learning_rate": 3.3027398351782703e-07, "loss": 1.3993, "step": 36817 }, { "epoch": 92.045, "grad_norm": 8.75, "learning_rate": 3.3006764145147365e-07, "loss": 1.2906, "step": 36818 }, { "epoch": 92.0475, "grad_norm": 7.40625, "learning_rate": 3.2986136278081004e-07, "loss": 1.4442, "step": 36819 }, { "epoch": 92.05, "grad_norm": 7.34375, "learning_rate": 3.2965514750718964e-07, "loss": 1.1053, "step": 36820 }, { "epoch": 92.0525, "grad_norm": 6.15625, "learning_rate": 3.294489956319613e-07, "loss": 1.3841, "step": 36821 }, { "epoch": 92.055, "grad_norm": 7.65625, "learning_rate": 3.2924290715647844e-07, "loss": 1.3492, "step": 36822 }, { "epoch": 92.0575, "grad_norm": 5.59375, "learning_rate": 3.2903688208209216e-07, "loss": 1.1511, "step": 36823 }, { "epoch": 92.06, "grad_norm": 8.5, "learning_rate": 3.2883092041015476e-07, "loss": 1.2852, "step": 36824 }, { "epoch": 92.0625, "grad_norm": 9.375, "learning_rate": 3.2862502214201396e-07, "loss": 1.2775, "step": 36825 }, { "epoch": 92.065, "grad_norm": 6.53125, "learning_rate": 3.284191872790188e-07, "loss": 1.379, "step": 36826 }, { "epoch": 92.0675, "grad_norm": 10.1875, "learning_rate": 3.282134158225214e-07, "loss": 1.4133, "step": 36827 }, { "epoch": 92.07, "grad_norm": 6.5, "learning_rate": 3.280077077738708e-07, "loss": 1.4162, "step": 36828 }, { "epoch": 92.0725, "grad_norm": 11.1875, "learning_rate": 3.278020631344137e-07, "loss": 1.3377, "step": 36829 }, { "epoch": 92.075, "grad_norm": 6.78125, "learning_rate": 3.2759648190549997e-07, "loss": 1.2453, "step": 36830 }, { "epoch": 92.0775, "grad_norm": 8.125, "learning_rate": 3.2739096408847537e-07, "loss": 1.2391, "step": 36831 }, { "epoch": 92.08, "grad_norm": 6.34375, "learning_rate": 3.271855096846899e-07, "loss": 1.2741, "step": 36832 }, { "epoch": 92.0825, "grad_norm": 9.5, "learning_rate": 3.2698011869548794e-07, "loss": 1.2232, "step": 36833 }, { "epoch": 92.085, "grad_norm": 7.5625, "learning_rate": 3.2677479112221857e-07, "loss": 1.3351, "step": 36834 }, { "epoch": 92.0875, "grad_norm": 8.5, "learning_rate": 3.265695269662261e-07, "loss": 1.2741, "step": 36835 }, { "epoch": 92.09, "grad_norm": 6.4375, "learning_rate": 3.2636432622885627e-07, "loss": 1.2672, "step": 36836 }, { "epoch": 92.0925, "grad_norm": 10.375, "learning_rate": 3.261591889114546e-07, "loss": 1.4718, "step": 36837 }, { "epoch": 92.095, "grad_norm": 6.28125, "learning_rate": 3.259541150153667e-07, "loss": 1.1979, "step": 36838 }, { "epoch": 92.0975, "grad_norm": 6.1875, "learning_rate": 3.2574910454193697e-07, "loss": 1.3243, "step": 36839 }, { "epoch": 92.1, "grad_norm": 7.9375, "learning_rate": 3.255441574925089e-07, "loss": 1.2434, "step": 36840 }, { "epoch": 92.1025, "grad_norm": 8.4375, "learning_rate": 3.253392738684236e-07, "loss": 1.2713, "step": 36841 }, { "epoch": 92.105, "grad_norm": 9.375, "learning_rate": 3.2513445367102994e-07, "loss": 1.3373, "step": 36842 }, { "epoch": 92.1075, "grad_norm": 8.1875, "learning_rate": 3.249296969016669e-07, "loss": 1.4223, "step": 36843 }, { "epoch": 92.11, "grad_norm": 5.6875, "learning_rate": 3.247250035616778e-07, "loss": 1.3202, "step": 36844 }, { "epoch": 92.1125, "grad_norm": 6.21875, "learning_rate": 3.2452037365240275e-07, "loss": 1.2644, "step": 36845 }, { "epoch": 92.115, "grad_norm": 9.125, "learning_rate": 3.243158071751862e-07, "loss": 1.2313, "step": 36846 }, { "epoch": 92.1175, "grad_norm": 6.6875, "learning_rate": 3.2411130413136817e-07, "loss": 1.2283, "step": 36847 }, { "epoch": 92.12, "grad_norm": 9.375, "learning_rate": 3.2390686452228983e-07, "loss": 1.4764, "step": 36848 }, { "epoch": 92.1225, "grad_norm": 5.25, "learning_rate": 3.2370248834928896e-07, "loss": 1.214, "step": 36849 }, { "epoch": 92.125, "grad_norm": 8.9375, "learning_rate": 3.23498175613709e-07, "loss": 1.312, "step": 36850 }, { "epoch": 92.1275, "grad_norm": 6.375, "learning_rate": 3.232939263168866e-07, "loss": 1.331, "step": 36851 }, { "epoch": 92.13, "grad_norm": 6.03125, "learning_rate": 3.2308974046016296e-07, "loss": 1.2874, "step": 36852 }, { "epoch": 92.1325, "grad_norm": 7.65625, "learning_rate": 3.2288561804487473e-07, "loss": 1.3253, "step": 36853 }, { "epoch": 92.135, "grad_norm": 7.25, "learning_rate": 3.22681559072362e-07, "loss": 1.3505, "step": 36854 }, { "epoch": 92.1375, "grad_norm": 7.25, "learning_rate": 3.2247756354396256e-07, "loss": 1.3238, "step": 36855 }, { "epoch": 92.14, "grad_norm": 9.1875, "learning_rate": 3.222736314610098e-07, "loss": 1.4561, "step": 36856 }, { "epoch": 92.1425, "grad_norm": 9.4375, "learning_rate": 3.2206976282484593e-07, "loss": 1.1829, "step": 36857 }, { "epoch": 92.145, "grad_norm": 5.1875, "learning_rate": 3.218659576368055e-07, "loss": 1.369, "step": 36858 }, { "epoch": 92.1475, "grad_norm": 6.84375, "learning_rate": 3.216622158982252e-07, "loss": 1.29, "step": 36859 }, { "epoch": 92.15, "grad_norm": 9.375, "learning_rate": 3.2145853761043844e-07, "loss": 1.1948, "step": 36860 }, { "epoch": 92.1525, "grad_norm": 6.40625, "learning_rate": 3.21254922774783e-07, "loss": 1.2839, "step": 36861 }, { "epoch": 92.155, "grad_norm": 8.0625, "learning_rate": 3.2105137139259334e-07, "loss": 1.2892, "step": 36862 }, { "epoch": 92.1575, "grad_norm": 6.3125, "learning_rate": 3.20847883465204e-07, "loss": 1.3969, "step": 36863 }, { "epoch": 92.16, "grad_norm": 6.09375, "learning_rate": 3.2064445899394723e-07, "loss": 1.3279, "step": 36864 }, { "epoch": 92.1625, "grad_norm": 7.71875, "learning_rate": 3.2044109798015867e-07, "loss": 1.3246, "step": 36865 }, { "epoch": 92.165, "grad_norm": 8.6875, "learning_rate": 3.202378004251705e-07, "loss": 1.4881, "step": 36866 }, { "epoch": 92.1675, "grad_norm": 9.0625, "learning_rate": 3.2003456633031615e-07, "loss": 1.3071, "step": 36867 }, { "epoch": 92.17, "grad_norm": 7.71875, "learning_rate": 3.198313956969279e-07, "loss": 1.1903, "step": 36868 }, { "epoch": 92.1725, "grad_norm": 6.34375, "learning_rate": 3.1962828852633796e-07, "loss": 1.4855, "step": 36869 }, { "epoch": 92.175, "grad_norm": 7.34375, "learning_rate": 3.194252448198776e-07, "loss": 1.3992, "step": 36870 }, { "epoch": 92.1775, "grad_norm": 11.3125, "learning_rate": 3.1922226457887786e-07, "loss": 1.2059, "step": 36871 }, { "epoch": 92.18, "grad_norm": 8.4375, "learning_rate": 3.190193478046677e-07, "loss": 1.2192, "step": 36872 }, { "epoch": 92.1825, "grad_norm": 7.3125, "learning_rate": 3.188164944985817e-07, "loss": 1.2805, "step": 36873 }, { "epoch": 92.185, "grad_norm": 6.65625, "learning_rate": 3.186137046619475e-07, "loss": 1.3239, "step": 36874 }, { "epoch": 92.1875, "grad_norm": 5.34375, "learning_rate": 3.1841097829609313e-07, "loss": 1.2323, "step": 36875 }, { "epoch": 92.19, "grad_norm": 7.84375, "learning_rate": 3.1820831540234856e-07, "loss": 1.2425, "step": 36876 }, { "epoch": 92.1925, "grad_norm": 7.0625, "learning_rate": 3.1800571598204375e-07, "loss": 1.3378, "step": 36877 }, { "epoch": 92.195, "grad_norm": 7.625, "learning_rate": 3.178031800365067e-07, "loss": 1.1996, "step": 36878 }, { "epoch": 92.1975, "grad_norm": 7.875, "learning_rate": 3.1760070756706284e-07, "loss": 1.3709, "step": 36879 }, { "epoch": 92.2, "grad_norm": 7.65625, "learning_rate": 3.1739829857504235e-07, "loss": 1.3505, "step": 36880 }, { "epoch": 92.2025, "grad_norm": 5.625, "learning_rate": 3.1719595306177077e-07, "loss": 1.157, "step": 36881 }, { "epoch": 92.205, "grad_norm": 9.5625, "learning_rate": 3.1699367102857594e-07, "loss": 1.2196, "step": 36882 }, { "epoch": 92.2075, "grad_norm": 6.1875, "learning_rate": 3.1679145247678123e-07, "loss": 1.2975, "step": 36883 }, { "epoch": 92.21, "grad_norm": 5.71875, "learning_rate": 3.165892974077156e-07, "loss": 1.4422, "step": 36884 }, { "epoch": 92.2125, "grad_norm": 8.125, "learning_rate": 3.1638720582270353e-07, "loss": 1.318, "step": 36885 }, { "epoch": 92.215, "grad_norm": 7.40625, "learning_rate": 3.1618517772306846e-07, "loss": 1.2213, "step": 36886 }, { "epoch": 92.2175, "grad_norm": 6.0, "learning_rate": 3.159832131101348e-07, "loss": 1.3135, "step": 36887 }, { "epoch": 92.22, "grad_norm": 6.96875, "learning_rate": 3.157813119852304e-07, "loss": 1.182, "step": 36888 }, { "epoch": 92.2225, "grad_norm": 8.75, "learning_rate": 3.155794743496743e-07, "loss": 1.3226, "step": 36889 }, { "epoch": 92.225, "grad_norm": 8.8125, "learning_rate": 3.1537770020479195e-07, "loss": 1.321, "step": 36890 }, { "epoch": 92.2275, "grad_norm": 12.3125, "learning_rate": 3.1517598955190576e-07, "loss": 1.3285, "step": 36891 }, { "epoch": 92.23, "grad_norm": 7.5, "learning_rate": 3.1497434239233904e-07, "loss": 1.2403, "step": 36892 }, { "epoch": 92.2325, "grad_norm": 8.3125, "learning_rate": 3.147727587274119e-07, "loss": 1.279, "step": 36893 }, { "epoch": 92.235, "grad_norm": 6.46875, "learning_rate": 3.1457123855844764e-07, "loss": 1.3918, "step": 36894 }, { "epoch": 92.2375, "grad_norm": 7.9375, "learning_rate": 3.143697818867664e-07, "loss": 1.3446, "step": 36895 }, { "epoch": 92.24, "grad_norm": 7.25, "learning_rate": 3.1416838871368925e-07, "loss": 1.3532, "step": 36896 }, { "epoch": 92.2425, "grad_norm": 11.875, "learning_rate": 3.139670590405364e-07, "loss": 1.3338, "step": 36897 }, { "epoch": 92.245, "grad_norm": 6.34375, "learning_rate": 3.1376579286862775e-07, "loss": 1.4106, "step": 36898 }, { "epoch": 92.2475, "grad_norm": 6.65625, "learning_rate": 3.1356459019928455e-07, "loss": 1.2354, "step": 36899 }, { "epoch": 92.25, "grad_norm": 6.1875, "learning_rate": 3.133634510338235e-07, "loss": 1.3822, "step": 36900 }, { "epoch": 92.2525, "grad_norm": 7.28125, "learning_rate": 3.131623753735624e-07, "loss": 1.3486, "step": 36901 }, { "epoch": 92.255, "grad_norm": 7.78125, "learning_rate": 3.129613632198214e-07, "loss": 1.4617, "step": 36902 }, { "epoch": 92.2575, "grad_norm": 5.8125, "learning_rate": 3.127604145739205e-07, "loss": 1.2262, "step": 36903 }, { "epoch": 92.26, "grad_norm": 6.90625, "learning_rate": 3.1255952943717305e-07, "loss": 1.3975, "step": 36904 }, { "epoch": 92.2625, "grad_norm": 9.75, "learning_rate": 3.1235870781089693e-07, "loss": 1.2923, "step": 36905 }, { "epoch": 92.265, "grad_norm": 6.5, "learning_rate": 3.1215794969641e-07, "loss": 1.3555, "step": 36906 }, { "epoch": 92.2675, "grad_norm": 8.0625, "learning_rate": 3.1195725509502893e-07, "loss": 1.4826, "step": 36907 }, { "epoch": 92.27, "grad_norm": 6.8125, "learning_rate": 3.1175662400806716e-07, "loss": 1.3526, "step": 36908 }, { "epoch": 92.2725, "grad_norm": 8.75, "learning_rate": 3.1155605643684137e-07, "loss": 1.3593, "step": 36909 }, { "epoch": 92.275, "grad_norm": 6.5625, "learning_rate": 3.113555523826672e-07, "loss": 1.21, "step": 36910 }, { "epoch": 92.2775, "grad_norm": 6.25, "learning_rate": 3.1115511184685696e-07, "loss": 1.1753, "step": 36911 }, { "epoch": 92.28, "grad_norm": 6.8125, "learning_rate": 3.1095473483072733e-07, "loss": 1.2471, "step": 36912 }, { "epoch": 92.2825, "grad_norm": 9.375, "learning_rate": 3.1075442133559064e-07, "loss": 1.2791, "step": 36913 }, { "epoch": 92.285, "grad_norm": 9.0625, "learning_rate": 3.105541713627602e-07, "loss": 1.3776, "step": 36914 }, { "epoch": 92.2875, "grad_norm": 9.0625, "learning_rate": 3.103539849135484e-07, "loss": 1.2468, "step": 36915 }, { "epoch": 92.29, "grad_norm": 7.53125, "learning_rate": 3.1015386198926856e-07, "loss": 1.472, "step": 36916 }, { "epoch": 92.2925, "grad_norm": 6.625, "learning_rate": 3.0995380259123076e-07, "loss": 1.3692, "step": 36917 }, { "epoch": 92.295, "grad_norm": 6.375, "learning_rate": 3.097538067207495e-07, "loss": 1.375, "step": 36918 }, { "epoch": 92.2975, "grad_norm": 7.53125, "learning_rate": 3.0955387437913485e-07, "loss": 1.2471, "step": 36919 }, { "epoch": 92.3, "grad_norm": 5.25, "learning_rate": 3.093540055676958e-07, "loss": 1.2533, "step": 36920 }, { "epoch": 92.3025, "grad_norm": 7.6875, "learning_rate": 3.091542002877446e-07, "loss": 1.1906, "step": 36921 }, { "epoch": 92.305, "grad_norm": 7.90625, "learning_rate": 3.0895445854059125e-07, "loss": 1.2281, "step": 36922 }, { "epoch": 92.3075, "grad_norm": 7.21875, "learning_rate": 3.087547803275448e-07, "loss": 1.4029, "step": 36923 }, { "epoch": 92.31, "grad_norm": 7.21875, "learning_rate": 3.0855516564991307e-07, "loss": 1.3658, "step": 36924 }, { "epoch": 92.3125, "grad_norm": 7.46875, "learning_rate": 3.083556145090072e-07, "loss": 1.1949, "step": 36925 }, { "epoch": 92.315, "grad_norm": 9.0625, "learning_rate": 3.081561269061328e-07, "loss": 1.4178, "step": 36926 }, { "epoch": 92.3175, "grad_norm": 8.625, "learning_rate": 3.0795670284260007e-07, "loss": 1.3846, "step": 36927 }, { "epoch": 92.32, "grad_norm": 7.96875, "learning_rate": 3.0775734231971443e-07, "loss": 1.4153, "step": 36928 }, { "epoch": 92.3225, "grad_norm": 8.8125, "learning_rate": 3.0755804533878495e-07, "loss": 1.1437, "step": 36929 }, { "epoch": 92.325, "grad_norm": 9.375, "learning_rate": 3.0735881190111616e-07, "loss": 1.279, "step": 36930 }, { "epoch": 92.3275, "grad_norm": 8.375, "learning_rate": 3.0715964200801476e-07, "loss": 1.3116, "step": 36931 }, { "epoch": 92.33, "grad_norm": 7.21875, "learning_rate": 3.069605356607874e-07, "loss": 1.1359, "step": 36932 }, { "epoch": 92.3325, "grad_norm": 5.625, "learning_rate": 3.067614928607388e-07, "loss": 1.3996, "step": 36933 }, { "epoch": 92.335, "grad_norm": 6.15625, "learning_rate": 3.065625136091754e-07, "loss": 1.2301, "step": 36934 }, { "epoch": 92.3375, "grad_norm": 9.1875, "learning_rate": 3.0636359790739757e-07, "loss": 1.3032, "step": 36935 }, { "epoch": 92.34, "grad_norm": 8.8125, "learning_rate": 3.0616474575671297e-07, "loss": 1.4433, "step": 36936 }, { "epoch": 92.3425, "grad_norm": 6.40625, "learning_rate": 3.0596595715842505e-07, "loss": 1.2048, "step": 36937 }, { "epoch": 92.345, "grad_norm": 7.09375, "learning_rate": 3.057672321138372e-07, "loss": 1.2113, "step": 36938 }, { "epoch": 92.3475, "grad_norm": 9.3125, "learning_rate": 3.0556857062424947e-07, "loss": 1.3558, "step": 36939 }, { "epoch": 92.35, "grad_norm": 9.625, "learning_rate": 3.053699726909676e-07, "loss": 1.3585, "step": 36940 }, { "epoch": 92.3525, "grad_norm": 6.0, "learning_rate": 3.051714383152915e-07, "loss": 1.306, "step": 36941 }, { "epoch": 92.355, "grad_norm": 5.8125, "learning_rate": 3.0497296749852354e-07, "loss": 1.3519, "step": 36942 }, { "epoch": 92.3575, "grad_norm": 7.3125, "learning_rate": 3.0477456024196493e-07, "loss": 1.2507, "step": 36943 }, { "epoch": 92.36, "grad_norm": 6.90625, "learning_rate": 3.045762165469168e-07, "loss": 1.3496, "step": 36944 }, { "epoch": 92.3625, "grad_norm": 7.15625, "learning_rate": 3.043779364146793e-07, "loss": 1.3165, "step": 36945 }, { "epoch": 92.365, "grad_norm": 7.625, "learning_rate": 3.0417971984655014e-07, "loss": 1.2862, "step": 36946 }, { "epoch": 92.3675, "grad_norm": 8.6875, "learning_rate": 3.039815668438317e-07, "loss": 1.3705, "step": 36947 }, { "epoch": 92.37, "grad_norm": 7.59375, "learning_rate": 3.03783477407823e-07, "loss": 1.3753, "step": 36948 }, { "epoch": 92.3725, "grad_norm": 8.25, "learning_rate": 3.0358545153982177e-07, "loss": 1.3478, "step": 36949 }, { "epoch": 92.375, "grad_norm": 10.0, "learning_rate": 3.0338748924112483e-07, "loss": 1.3961, "step": 36950 }, { "epoch": 92.3775, "grad_norm": 7.6875, "learning_rate": 3.031895905130322e-07, "loss": 1.465, "step": 36951 }, { "epoch": 92.38, "grad_norm": 6.5, "learning_rate": 3.029917553568407e-07, "loss": 1.513, "step": 36952 }, { "epoch": 92.3825, "grad_norm": 9.0625, "learning_rate": 3.0279398377384805e-07, "loss": 1.3637, "step": 36953 }, { "epoch": 92.385, "grad_norm": 5.96875, "learning_rate": 3.025962757653478e-07, "loss": 1.3536, "step": 36954 }, { "epoch": 92.3875, "grad_norm": 6.15625, "learning_rate": 3.0239863133264103e-07, "loss": 1.184, "step": 36955 }, { "epoch": 92.39, "grad_norm": 7.8125, "learning_rate": 3.0220105047701787e-07, "loss": 1.2408, "step": 36956 }, { "epoch": 92.3925, "grad_norm": 6.84375, "learning_rate": 3.0200353319977837e-07, "loss": 1.2425, "step": 36957 }, { "epoch": 92.395, "grad_norm": 7.15625, "learning_rate": 3.018060795022149e-07, "loss": 1.2425, "step": 36958 }, { "epoch": 92.3975, "grad_norm": 10.0, "learning_rate": 3.0160868938562405e-07, "loss": 1.3871, "step": 36959 }, { "epoch": 92.4, "grad_norm": 6.84375, "learning_rate": 3.0141136285129825e-07, "loss": 1.3166, "step": 36960 }, { "epoch": 92.4025, "grad_norm": 8.3125, "learning_rate": 3.012140999005297e-07, "loss": 1.373, "step": 36961 }, { "epoch": 92.405, "grad_norm": 11.875, "learning_rate": 3.0101690053461417e-07, "loss": 1.3135, "step": 36962 }, { "epoch": 92.4075, "grad_norm": 9.6875, "learning_rate": 3.0081976475484385e-07, "loss": 1.4018, "step": 36963 }, { "epoch": 92.41, "grad_norm": 7.53125, "learning_rate": 3.006226925625122e-07, "loss": 1.3651, "step": 36964 }, { "epoch": 92.4125, "grad_norm": 5.90625, "learning_rate": 3.004256839589093e-07, "loss": 1.2088, "step": 36965 }, { "epoch": 92.415, "grad_norm": 9.9375, "learning_rate": 3.002287389453251e-07, "loss": 1.3632, "step": 36966 }, { "epoch": 92.4175, "grad_norm": 8.5625, "learning_rate": 3.000318575230554e-07, "loss": 1.4962, "step": 36967 }, { "epoch": 92.42, "grad_norm": 8.625, "learning_rate": 2.9983503969338913e-07, "loss": 1.3944, "step": 36968 }, { "epoch": 92.4225, "grad_norm": 7.96875, "learning_rate": 2.996382854576152e-07, "loss": 1.3651, "step": 36969 }, { "epoch": 92.425, "grad_norm": 7.40625, "learning_rate": 2.9944159481702373e-07, "loss": 1.4282, "step": 36970 }, { "epoch": 92.4275, "grad_norm": 6.4375, "learning_rate": 2.992449677729059e-07, "loss": 1.3511, "step": 36971 }, { "epoch": 92.43, "grad_norm": 8.75, "learning_rate": 2.990484043265507e-07, "loss": 1.4274, "step": 36972 }, { "epoch": 92.4325, "grad_norm": 7.90625, "learning_rate": 2.988519044792448e-07, "loss": 1.2274, "step": 36973 }, { "epoch": 92.435, "grad_norm": 9.875, "learning_rate": 2.9865546823227843e-07, "loss": 1.2461, "step": 36974 }, { "epoch": 92.4375, "grad_norm": 9.3125, "learning_rate": 2.9845909558693707e-07, "loss": 1.3714, "step": 36975 }, { "epoch": 92.44, "grad_norm": 10.3125, "learning_rate": 2.982627865445109e-07, "loss": 1.3554, "step": 36976 }, { "epoch": 92.4425, "grad_norm": 8.875, "learning_rate": 2.980665411062855e-07, "loss": 1.1998, "step": 36977 }, { "epoch": 92.445, "grad_norm": 6.0625, "learning_rate": 2.978703592735477e-07, "loss": 1.3885, "step": 36978 }, { "epoch": 92.4475, "grad_norm": 9.875, "learning_rate": 2.976742410475841e-07, "loss": 1.2933, "step": 36979 }, { "epoch": 92.45, "grad_norm": 7.625, "learning_rate": 2.974781864296783e-07, "loss": 1.224, "step": 36980 }, { "epoch": 92.4525, "grad_norm": 9.6875, "learning_rate": 2.9728219542111804e-07, "loss": 1.3352, "step": 36981 }, { "epoch": 92.455, "grad_norm": 6.84375, "learning_rate": 2.9708626802318783e-07, "loss": 1.2018, "step": 36982 }, { "epoch": 92.4575, "grad_norm": 6.5625, "learning_rate": 2.968904042371723e-07, "loss": 1.2064, "step": 36983 }, { "epoch": 92.46, "grad_norm": 6.875, "learning_rate": 2.966946040643548e-07, "loss": 1.1986, "step": 36984 }, { "epoch": 92.4625, "grad_norm": 8.8125, "learning_rate": 2.9649886750601654e-07, "loss": 1.4171, "step": 36985 }, { "epoch": 92.465, "grad_norm": 8.9375, "learning_rate": 2.963031945634454e-07, "loss": 1.4577, "step": 36986 }, { "epoch": 92.4675, "grad_norm": 6.6875, "learning_rate": 2.961075852379225e-07, "loss": 1.2676, "step": 36987 }, { "epoch": 92.47, "grad_norm": 8.9375, "learning_rate": 2.9591203953073024e-07, "loss": 1.24, "step": 36988 }, { "epoch": 92.4725, "grad_norm": 8.125, "learning_rate": 2.9571655744314975e-07, "loss": 1.4188, "step": 36989 }, { "epoch": 92.475, "grad_norm": 6.375, "learning_rate": 2.9552113897646227e-07, "loss": 1.1767, "step": 36990 }, { "epoch": 92.4775, "grad_norm": 4.9375, "learning_rate": 2.953257841319512e-07, "loss": 1.296, "step": 36991 }, { "epoch": 92.48, "grad_norm": 9.9375, "learning_rate": 2.951304929108956e-07, "loss": 1.4119, "step": 36992 }, { "epoch": 92.4825, "grad_norm": 4.46875, "learning_rate": 2.949352653145754e-07, "loss": 1.269, "step": 36993 }, { "epoch": 92.485, "grad_norm": 10.5625, "learning_rate": 2.947401013442719e-07, "loss": 1.3567, "step": 36994 }, { "epoch": 92.4875, "grad_norm": 10.0, "learning_rate": 2.9454500100126406e-07, "loss": 1.1802, "step": 36995 }, { "epoch": 92.49, "grad_norm": 7.1875, "learning_rate": 2.943499642868297e-07, "loss": 1.3464, "step": 36996 }, { "epoch": 92.4925, "grad_norm": 8.375, "learning_rate": 2.941549912022501e-07, "loss": 1.4132, "step": 36997 }, { "epoch": 92.495, "grad_norm": 9.0, "learning_rate": 2.9396008174880195e-07, "loss": 1.2728, "step": 36998 }, { "epoch": 92.4975, "grad_norm": 10.5, "learning_rate": 2.9376523592776206e-07, "loss": 1.2809, "step": 36999 }, { "epoch": 92.5, "grad_norm": 7.8125, "learning_rate": 2.935704537404083e-07, "loss": 1.3097, "step": 37000 }, { "epoch": 92.5025, "grad_norm": 8.375, "learning_rate": 2.9337573518801954e-07, "loss": 1.1564, "step": 37001 }, { "epoch": 92.505, "grad_norm": 5.21875, "learning_rate": 2.931810802718704e-07, "loss": 1.27, "step": 37002 }, { "epoch": 92.5075, "grad_norm": 7.5625, "learning_rate": 2.9298648899323877e-07, "loss": 1.2968, "step": 37003 }, { "epoch": 92.51, "grad_norm": 6.65625, "learning_rate": 2.9279196135339806e-07, "loss": 1.3652, "step": 37004 }, { "epoch": 92.5125, "grad_norm": 8.5625, "learning_rate": 2.9259749735362385e-07, "loss": 1.3616, "step": 37005 }, { "epoch": 92.515, "grad_norm": 5.53125, "learning_rate": 2.924030969951941e-07, "loss": 1.327, "step": 37006 }, { "epoch": 92.5175, "grad_norm": 7.875, "learning_rate": 2.9220876027938106e-07, "loss": 1.3693, "step": 37007 }, { "epoch": 92.52, "grad_norm": 6.28125, "learning_rate": 2.9201448720745706e-07, "loss": 1.1766, "step": 37008 }, { "epoch": 92.5225, "grad_norm": 8.125, "learning_rate": 2.9182027778069886e-07, "loss": 1.3432, "step": 37009 }, { "epoch": 92.525, "grad_norm": 5.21875, "learning_rate": 2.916261320003777e-07, "loss": 1.2492, "step": 37010 }, { "epoch": 92.5275, "grad_norm": 8.8125, "learning_rate": 2.914320498677681e-07, "loss": 1.4355, "step": 37011 }, { "epoch": 92.53, "grad_norm": 9.1875, "learning_rate": 2.912380313841412e-07, "loss": 1.2899, "step": 37012 }, { "epoch": 92.5325, "grad_norm": 6.34375, "learning_rate": 2.9104407655076936e-07, "loss": 1.3725, "step": 37013 }, { "epoch": 92.535, "grad_norm": 7.53125, "learning_rate": 2.9085018536892385e-07, "loss": 1.4069, "step": 37014 }, { "epoch": 92.5375, "grad_norm": 8.75, "learning_rate": 2.906563578398758e-07, "loss": 1.323, "step": 37015 }, { "epoch": 92.54, "grad_norm": 6.21875, "learning_rate": 2.904625939648953e-07, "loss": 1.3891, "step": 37016 }, { "epoch": 92.5425, "grad_norm": 6.375, "learning_rate": 2.902688937452547e-07, "loss": 1.4309, "step": 37017 }, { "epoch": 92.545, "grad_norm": 4.875, "learning_rate": 2.9007525718222295e-07, "loss": 1.1847, "step": 37018 }, { "epoch": 92.5475, "grad_norm": 6.4375, "learning_rate": 2.89881684277068e-07, "loss": 1.4248, "step": 37019 }, { "epoch": 92.55, "grad_norm": 7.15625, "learning_rate": 2.8968817503105984e-07, "loss": 1.3886, "step": 37020 }, { "epoch": 92.5525, "grad_norm": 6.46875, "learning_rate": 2.8949472944546864e-07, "loss": 1.4178, "step": 37021 }, { "epoch": 92.555, "grad_norm": 10.4375, "learning_rate": 2.893013475215622e-07, "loss": 1.4649, "step": 37022 }, { "epoch": 92.5575, "grad_norm": 8.6875, "learning_rate": 2.891080292606052e-07, "loss": 1.3661, "step": 37023 }, { "epoch": 92.56, "grad_norm": 7.71875, "learning_rate": 2.8891477466386987e-07, "loss": 1.4245, "step": 37024 }, { "epoch": 92.5625, "grad_norm": 6.46875, "learning_rate": 2.8872158373261847e-07, "loss": 1.3924, "step": 37025 }, { "epoch": 92.565, "grad_norm": 10.1875, "learning_rate": 2.8852845646812124e-07, "loss": 1.2493, "step": 37026 }, { "epoch": 92.5675, "grad_norm": 7.71875, "learning_rate": 2.883353928716415e-07, "loss": 1.3563, "step": 37027 }, { "epoch": 92.57, "grad_norm": 7.21875, "learning_rate": 2.881423929444471e-07, "loss": 1.4822, "step": 37028 }, { "epoch": 92.5725, "grad_norm": 5.9375, "learning_rate": 2.8794945668780274e-07, "loss": 1.447, "step": 37029 }, { "epoch": 92.575, "grad_norm": 7.65625, "learning_rate": 2.877565841029728e-07, "loss": 1.3565, "step": 37030 }, { "epoch": 92.5775, "grad_norm": 7.21875, "learning_rate": 2.875637751912208e-07, "loss": 1.2163, "step": 37031 }, { "epoch": 92.58, "grad_norm": 7.25, "learning_rate": 2.873710299538146e-07, "loss": 1.357, "step": 37032 }, { "epoch": 92.5825, "grad_norm": 6.09375, "learning_rate": 2.871783483920143e-07, "loss": 1.2814, "step": 37033 }, { "epoch": 92.585, "grad_norm": 7.40625, "learning_rate": 2.8698573050708223e-07, "loss": 1.3649, "step": 37034 }, { "epoch": 92.5875, "grad_norm": 8.3125, "learning_rate": 2.86793176300284e-07, "loss": 1.4942, "step": 37035 }, { "epoch": 92.59, "grad_norm": 8.375, "learning_rate": 2.86600685772882e-07, "loss": 1.3734, "step": 37036 }, { "epoch": 92.5925, "grad_norm": 7.46875, "learning_rate": 2.864082589261374e-07, "loss": 1.2851, "step": 37037 }, { "epoch": 92.595, "grad_norm": 7.46875, "learning_rate": 2.8621589576131037e-07, "loss": 1.3273, "step": 37038 }, { "epoch": 92.5975, "grad_norm": 5.46875, "learning_rate": 2.8602359627966427e-07, "loss": 1.3175, "step": 37039 }, { "epoch": 92.6, "grad_norm": 6.34375, "learning_rate": 2.8583136048245697e-07, "loss": 1.3838, "step": 37040 }, { "epoch": 92.6025, "grad_norm": 7.375, "learning_rate": 2.8563918837095197e-07, "loss": 1.2623, "step": 37041 }, { "epoch": 92.605, "grad_norm": 5.0, "learning_rate": 2.8544707994640706e-07, "loss": 1.2315, "step": 37042 }, { "epoch": 92.6075, "grad_norm": 8.5, "learning_rate": 2.852550352100836e-07, "loss": 1.2869, "step": 37043 }, { "epoch": 92.61, "grad_norm": 8.125, "learning_rate": 2.8506305416323823e-07, "loss": 1.0919, "step": 37044 }, { "epoch": 92.6125, "grad_norm": 7.71875, "learning_rate": 2.8487113680712994e-07, "loss": 1.3272, "step": 37045 }, { "epoch": 92.615, "grad_norm": 7.71875, "learning_rate": 2.8467928314301894e-07, "loss": 1.1674, "step": 37046 }, { "epoch": 92.6175, "grad_norm": 6.8125, "learning_rate": 2.84487493172162e-07, "loss": 1.4089, "step": 37047 }, { "epoch": 92.62, "grad_norm": 7.59375, "learning_rate": 2.842957668958157e-07, "loss": 1.4719, "step": 37048 }, { "epoch": 92.6225, "grad_norm": 9.375, "learning_rate": 2.8410410431523704e-07, "loss": 1.2552, "step": 37049 }, { "epoch": 92.625, "grad_norm": 7.875, "learning_rate": 2.839125054316838e-07, "loss": 1.2334, "step": 37050 }, { "epoch": 92.6275, "grad_norm": 6.375, "learning_rate": 2.8372097024641165e-07, "loss": 1.2687, "step": 37051 }, { "epoch": 92.63, "grad_norm": 6.0, "learning_rate": 2.8352949876067516e-07, "loss": 1.2416, "step": 37052 }, { "epoch": 92.6325, "grad_norm": 8.5625, "learning_rate": 2.833380909757311e-07, "loss": 1.3188, "step": 37053 }, { "epoch": 92.635, "grad_norm": 8.4375, "learning_rate": 2.8314674689283286e-07, "loss": 1.3906, "step": 37054 }, { "epoch": 92.6375, "grad_norm": 5.8125, "learning_rate": 2.829554665132361e-07, "loss": 1.2358, "step": 37055 }, { "epoch": 92.64, "grad_norm": 8.0, "learning_rate": 2.827642498381955e-07, "loss": 1.2096, "step": 37056 }, { "epoch": 92.6425, "grad_norm": 6.96875, "learning_rate": 2.825730968689622e-07, "loss": 1.3483, "step": 37057 }, { "epoch": 92.645, "grad_norm": 10.5, "learning_rate": 2.823820076067918e-07, "loss": 1.4229, "step": 37058 }, { "epoch": 92.6475, "grad_norm": 6.90625, "learning_rate": 2.821909820529356e-07, "loss": 1.267, "step": 37059 }, { "epoch": 92.65, "grad_norm": 9.25, "learning_rate": 2.820000202086459e-07, "loss": 1.2835, "step": 37060 }, { "epoch": 92.6525, "grad_norm": 5.4375, "learning_rate": 2.8180912207517507e-07, "loss": 1.2619, "step": 37061 }, { "epoch": 92.655, "grad_norm": 10.5, "learning_rate": 2.8161828765377543e-07, "loss": 1.3967, "step": 37062 }, { "epoch": 92.6575, "grad_norm": 7.6875, "learning_rate": 2.8142751694569814e-07, "loss": 1.2947, "step": 37063 }, { "epoch": 92.66, "grad_norm": 8.5625, "learning_rate": 2.8123680995219226e-07, "loss": 1.2883, "step": 37064 }, { "epoch": 92.6625, "grad_norm": 6.71875, "learning_rate": 2.810461666745079e-07, "loss": 1.2206, "step": 37065 }, { "epoch": 92.665, "grad_norm": 7.3125, "learning_rate": 2.8085558711389737e-07, "loss": 1.3464, "step": 37066 }, { "epoch": 92.6675, "grad_norm": 9.375, "learning_rate": 2.806650712716086e-07, "loss": 1.1714, "step": 37067 }, { "epoch": 92.67, "grad_norm": 5.21875, "learning_rate": 2.804746191488905e-07, "loss": 1.3908, "step": 37068 }, { "epoch": 92.6725, "grad_norm": 7.6875, "learning_rate": 2.8028423074699217e-07, "loss": 1.2958, "step": 37069 }, { "epoch": 92.675, "grad_norm": 5.125, "learning_rate": 2.8009390606716034e-07, "loss": 1.4063, "step": 37070 }, { "epoch": 92.6775, "grad_norm": 7.46875, "learning_rate": 2.7990364511064516e-07, "loss": 1.3188, "step": 37071 }, { "epoch": 92.68, "grad_norm": 4.9375, "learning_rate": 2.7971344787869114e-07, "loss": 1.1376, "step": 37072 }, { "epoch": 92.6825, "grad_norm": 7.28125, "learning_rate": 2.7952331437254844e-07, "loss": 1.3376, "step": 37073 }, { "epoch": 92.685, "grad_norm": 7.125, "learning_rate": 2.793332445934616e-07, "loss": 1.3998, "step": 37074 }, { "epoch": 92.6875, "grad_norm": 8.75, "learning_rate": 2.791432385426762e-07, "loss": 1.2284, "step": 37075 }, { "epoch": 92.69, "grad_norm": 7.53125, "learning_rate": 2.7895329622143917e-07, "loss": 1.2413, "step": 37076 }, { "epoch": 92.6925, "grad_norm": 8.625, "learning_rate": 2.78763417630995e-07, "loss": 1.3722, "step": 37077 }, { "epoch": 92.695, "grad_norm": 7.53125, "learning_rate": 2.7857360277258937e-07, "loss": 1.1692, "step": 37078 }, { "epoch": 92.6975, "grad_norm": 8.4375, "learning_rate": 2.783838516474657e-07, "loss": 1.2544, "step": 37079 }, { "epoch": 92.7, "grad_norm": 10.625, "learning_rate": 2.781941642568686e-07, "loss": 1.2136, "step": 37080 }, { "epoch": 92.7025, "grad_norm": 7.25, "learning_rate": 2.780045406020415e-07, "loss": 1.5031, "step": 37081 }, { "epoch": 92.705, "grad_norm": 8.25, "learning_rate": 2.7781498068422896e-07, "loss": 1.299, "step": 37082 }, { "epoch": 92.7075, "grad_norm": 7.40625, "learning_rate": 2.7762548450466997e-07, "loss": 1.3597, "step": 37083 }, { "epoch": 92.71, "grad_norm": 8.4375, "learning_rate": 2.774360520646113e-07, "loss": 1.2756, "step": 37084 }, { "epoch": 92.7125, "grad_norm": 5.84375, "learning_rate": 2.772466833652909e-07, "loss": 1.3158, "step": 37085 }, { "epoch": 92.715, "grad_norm": 8.3125, "learning_rate": 2.7705737840795334e-07, "loss": 1.3714, "step": 37086 }, { "epoch": 92.7175, "grad_norm": 8.25, "learning_rate": 2.7686813719383865e-07, "loss": 1.3072, "step": 37087 }, { "epoch": 92.72, "grad_norm": 9.75, "learning_rate": 2.76678959724187e-07, "loss": 1.3592, "step": 37088 }, { "epoch": 92.7225, "grad_norm": 5.71875, "learning_rate": 2.7648984600023964e-07, "loss": 1.3297, "step": 37089 }, { "epoch": 92.725, "grad_norm": 11.9375, "learning_rate": 2.7630079602323447e-07, "loss": 1.4426, "step": 37090 }, { "epoch": 92.7275, "grad_norm": 8.25, "learning_rate": 2.7611180979441155e-07, "loss": 1.4495, "step": 37091 }, { "epoch": 92.73, "grad_norm": 9.6875, "learning_rate": 2.759228873150122e-07, "loss": 1.5416, "step": 37092 }, { "epoch": 92.7325, "grad_norm": 11.75, "learning_rate": 2.7573402858627417e-07, "loss": 1.1925, "step": 37093 }, { "epoch": 92.735, "grad_norm": 8.0, "learning_rate": 2.755452336094322e-07, "loss": 1.2361, "step": 37094 }, { "epoch": 92.7375, "grad_norm": 7.75, "learning_rate": 2.7535650238572744e-07, "loss": 1.292, "step": 37095 }, { "epoch": 92.74, "grad_norm": 6.59375, "learning_rate": 2.7516783491639666e-07, "loss": 1.2668, "step": 37096 }, { "epoch": 92.7425, "grad_norm": 7.34375, "learning_rate": 2.749792312026767e-07, "loss": 1.362, "step": 37097 }, { "epoch": 92.745, "grad_norm": 7.0625, "learning_rate": 2.7479069124580314e-07, "loss": 1.3111, "step": 37098 }, { "epoch": 92.7475, "grad_norm": 8.5625, "learning_rate": 2.746022150470129e-07, "loss": 1.4737, "step": 37099 }, { "epoch": 92.75, "grad_norm": 7.375, "learning_rate": 2.744138026075405e-07, "loss": 1.2287, "step": 37100 }, { "epoch": 92.7525, "grad_norm": 7.8125, "learning_rate": 2.7422545392862267e-07, "loss": 1.3, "step": 37101 }, { "epoch": 92.755, "grad_norm": 8.3125, "learning_rate": 2.740371690114929e-07, "loss": 1.5413, "step": 37102 }, { "epoch": 92.7575, "grad_norm": 9.8125, "learning_rate": 2.73848947857388e-07, "loss": 1.1459, "step": 37103 }, { "epoch": 92.76, "grad_norm": 8.125, "learning_rate": 2.7366079046753925e-07, "loss": 1.3138, "step": 37104 }, { "epoch": 92.7625, "grad_norm": 8.3125, "learning_rate": 2.7347269684318113e-07, "loss": 1.4428, "step": 37105 }, { "epoch": 92.765, "grad_norm": 7.40625, "learning_rate": 2.7328466698554603e-07, "loss": 1.343, "step": 37106 }, { "epoch": 92.7675, "grad_norm": 5.84375, "learning_rate": 2.730967008958685e-07, "loss": 1.2452, "step": 37107 }, { "epoch": 92.77, "grad_norm": 8.125, "learning_rate": 2.729087985753798e-07, "loss": 1.264, "step": 37108 }, { "epoch": 92.7725, "grad_norm": 9.625, "learning_rate": 2.7272096002531e-07, "loss": 1.3445, "step": 37109 }, { "epoch": 92.775, "grad_norm": 6.25, "learning_rate": 2.7253318524689267e-07, "loss": 1.3299, "step": 37110 }, { "epoch": 92.7775, "grad_norm": 7.0625, "learning_rate": 2.7234547424136005e-07, "loss": 1.369, "step": 37111 }, { "epoch": 92.78, "grad_norm": 7.84375, "learning_rate": 2.721578270099412e-07, "loss": 1.2264, "step": 37112 }, { "epoch": 92.7825, "grad_norm": 6.21875, "learning_rate": 2.7197024355386516e-07, "loss": 1.2423, "step": 37113 }, { "epoch": 92.785, "grad_norm": 6.0625, "learning_rate": 2.717827238743642e-07, "loss": 1.2109, "step": 37114 }, { "epoch": 92.7875, "grad_norm": 10.875, "learning_rate": 2.715952679726652e-07, "loss": 1.3381, "step": 37115 }, { "epoch": 92.79, "grad_norm": 6.90625, "learning_rate": 2.714078758499994e-07, "loss": 1.2945, "step": 37116 }, { "epoch": 92.7925, "grad_norm": 6.6875, "learning_rate": 2.712205475075924e-07, "loss": 1.3398, "step": 37117 }, { "epoch": 92.795, "grad_norm": 6.0, "learning_rate": 2.7103328294667553e-07, "loss": 1.4137, "step": 37118 }, { "epoch": 92.7975, "grad_norm": 6.78125, "learning_rate": 2.7084608216847554e-07, "loss": 1.2208, "step": 37119 }, { "epoch": 92.8, "grad_norm": 8.375, "learning_rate": 2.706589451742181e-07, "loss": 1.3194, "step": 37120 }, { "epoch": 92.8025, "grad_norm": 7.3125, "learning_rate": 2.7047187196513223e-07, "loss": 1.1682, "step": 37121 }, { "epoch": 92.805, "grad_norm": 8.5625, "learning_rate": 2.7028486254244256e-07, "loss": 1.3367, "step": 37122 }, { "epoch": 92.8075, "grad_norm": 5.875, "learning_rate": 2.70097916907377e-07, "loss": 1.2121, "step": 37123 }, { "epoch": 92.81, "grad_norm": 6.375, "learning_rate": 2.6991103506116e-07, "loss": 1.2518, "step": 37124 }, { "epoch": 92.8125, "grad_norm": 6.46875, "learning_rate": 2.697242170050152e-07, "loss": 1.2336, "step": 37125 }, { "epoch": 92.815, "grad_norm": 7.96875, "learning_rate": 2.6953746274017145e-07, "loss": 1.2414, "step": 37126 }, { "epoch": 92.8175, "grad_norm": 8.5, "learning_rate": 2.693507722678501e-07, "loss": 1.3775, "step": 37127 }, { "epoch": 92.82, "grad_norm": 8.5, "learning_rate": 2.6916414558927575e-07, "loss": 1.2753, "step": 37128 }, { "epoch": 92.8225, "grad_norm": 10.5625, "learning_rate": 2.6897758270567065e-07, "loss": 1.4522, "step": 37129 }, { "epoch": 92.825, "grad_norm": 5.46875, "learning_rate": 2.687910836182583e-07, "loss": 1.2567, "step": 37130 }, { "epoch": 92.8275, "grad_norm": 6.84375, "learning_rate": 2.686046483282645e-07, "loss": 1.2825, "step": 37131 }, { "epoch": 92.83, "grad_norm": 8.5, "learning_rate": 2.684182768369081e-07, "loss": 1.3721, "step": 37132 }, { "epoch": 92.8325, "grad_norm": 7.0, "learning_rate": 2.682319691454127e-07, "loss": 1.2751, "step": 37133 }, { "epoch": 92.835, "grad_norm": 8.6875, "learning_rate": 2.680457252549984e-07, "loss": 1.4155, "step": 37134 }, { "epoch": 92.8375, "grad_norm": 7.34375, "learning_rate": 2.678595451668875e-07, "loss": 1.3301, "step": 37135 }, { "epoch": 92.84, "grad_norm": 7.25, "learning_rate": 2.6767342888229907e-07, "loss": 1.3631, "step": 37136 }, { "epoch": 92.8425, "grad_norm": 5.3125, "learning_rate": 2.6748737640245434e-07, "loss": 1.4332, "step": 37137 }, { "epoch": 92.845, "grad_norm": 10.5, "learning_rate": 2.673013877285746e-07, "loss": 1.3741, "step": 37138 }, { "epoch": 92.8475, "grad_norm": 7.75, "learning_rate": 2.6711546286187553e-07, "loss": 1.3897, "step": 37139 }, { "epoch": 92.85, "grad_norm": 5.28125, "learning_rate": 2.669296018035772e-07, "loss": 1.4878, "step": 37140 }, { "epoch": 92.8525, "grad_norm": 10.375, "learning_rate": 2.667438045549009e-07, "loss": 1.2564, "step": 37141 }, { "epoch": 92.855, "grad_norm": 8.0, "learning_rate": 2.6655807111706235e-07, "loss": 1.1513, "step": 37142 }, { "epoch": 92.8575, "grad_norm": 7.0625, "learning_rate": 2.6637240149128054e-07, "loss": 1.2618, "step": 37143 }, { "epoch": 92.86, "grad_norm": 6.59375, "learning_rate": 2.6618679567877006e-07, "loss": 1.3307, "step": 37144 }, { "epoch": 92.8625, "grad_norm": 10.4375, "learning_rate": 2.660012536807488e-07, "loss": 1.4474, "step": 37145 }, { "epoch": 92.865, "grad_norm": 10.375, "learning_rate": 2.6581577549843586e-07, "loss": 1.3285, "step": 37146 }, { "epoch": 92.8675, "grad_norm": 7.75, "learning_rate": 2.6563036113304465e-07, "loss": 1.4344, "step": 37147 }, { "epoch": 92.87, "grad_norm": 11.75, "learning_rate": 2.6544501058578976e-07, "loss": 1.1231, "step": 37148 }, { "epoch": 92.8725, "grad_norm": 6.1875, "learning_rate": 2.6525972385788914e-07, "loss": 1.3053, "step": 37149 }, { "epoch": 92.875, "grad_norm": 8.0, "learning_rate": 2.650745009505562e-07, "loss": 1.3675, "step": 37150 }, { "epoch": 92.8775, "grad_norm": 5.96875, "learning_rate": 2.648893418650045e-07, "loss": 1.3828, "step": 37151 }, { "epoch": 92.88, "grad_norm": 10.1875, "learning_rate": 2.647042466024485e-07, "loss": 1.4024, "step": 37152 }, { "epoch": 92.8825, "grad_norm": 8.6875, "learning_rate": 2.6451921516410406e-07, "loss": 1.2843, "step": 37153 }, { "epoch": 92.885, "grad_norm": 6.875, "learning_rate": 2.643342475511801e-07, "loss": 1.3974, "step": 37154 }, { "epoch": 92.8875, "grad_norm": 6.5, "learning_rate": 2.6414934376489123e-07, "loss": 1.2895, "step": 37155 }, { "epoch": 92.89, "grad_norm": 6.9375, "learning_rate": 2.6396450380644976e-07, "loss": 1.4785, "step": 37156 }, { "epoch": 92.8925, "grad_norm": 6.78125, "learning_rate": 2.637797276770682e-07, "loss": 1.2654, "step": 37157 }, { "epoch": 92.895, "grad_norm": 7.65625, "learning_rate": 2.635950153779576e-07, "loss": 1.3878, "step": 37158 }, { "epoch": 92.8975, "grad_norm": 8.375, "learning_rate": 2.6341036691032606e-07, "loss": 1.2582, "step": 37159 }, { "epoch": 92.9, "grad_norm": 6.125, "learning_rate": 2.632257822753881e-07, "loss": 1.2395, "step": 37160 }, { "epoch": 92.9025, "grad_norm": 5.71875, "learning_rate": 2.630412614743527e-07, "loss": 1.3168, "step": 37161 }, { "epoch": 92.905, "grad_norm": 9.875, "learning_rate": 2.62856804508429e-07, "loss": 1.2868, "step": 37162 }, { "epoch": 92.9075, "grad_norm": 7.53125, "learning_rate": 2.626724113788248e-07, "loss": 1.3046, "step": 37163 }, { "epoch": 92.91, "grad_norm": 7.6875, "learning_rate": 2.624880820867526e-07, "loss": 1.2286, "step": 37164 }, { "epoch": 92.9125, "grad_norm": 5.8125, "learning_rate": 2.62303816633418e-07, "loss": 1.3702, "step": 37165 }, { "epoch": 92.915, "grad_norm": 8.3125, "learning_rate": 2.6211961502003005e-07, "loss": 1.314, "step": 37166 }, { "epoch": 92.9175, "grad_norm": 10.8125, "learning_rate": 2.6193547724779555e-07, "loss": 1.2758, "step": 37167 }, { "epoch": 92.92, "grad_norm": 9.0, "learning_rate": 2.617514033179236e-07, "loss": 1.3642, "step": 37168 }, { "epoch": 92.9225, "grad_norm": 6.5625, "learning_rate": 2.615673932316187e-07, "loss": 1.3171, "step": 37169 }, { "epoch": 92.925, "grad_norm": 11.125, "learning_rate": 2.613834469900889e-07, "loss": 1.392, "step": 37170 }, { "epoch": 92.9275, "grad_norm": 7.5625, "learning_rate": 2.6119956459453866e-07, "loss": 1.3084, "step": 37171 }, { "epoch": 92.93, "grad_norm": 7.84375, "learning_rate": 2.6101574604617485e-07, "loss": 1.3053, "step": 37172 }, { "epoch": 92.9325, "grad_norm": 7.90625, "learning_rate": 2.608319913462032e-07, "loss": 1.333, "step": 37173 }, { "epoch": 92.935, "grad_norm": 7.90625, "learning_rate": 2.606483004958249e-07, "loss": 1.3449, "step": 37174 }, { "epoch": 92.9375, "grad_norm": 7.3125, "learning_rate": 2.604646734962479e-07, "loss": 1.3259, "step": 37175 }, { "epoch": 92.94, "grad_norm": 8.5, "learning_rate": 2.6028111034867466e-07, "loss": 1.4652, "step": 37176 }, { "epoch": 92.9425, "grad_norm": 7.125, "learning_rate": 2.6009761105430966e-07, "loss": 1.1642, "step": 37177 }, { "epoch": 92.945, "grad_norm": 5.25, "learning_rate": 2.5991417561435417e-07, "loss": 1.2117, "step": 37178 }, { "epoch": 92.9475, "grad_norm": 6.1875, "learning_rate": 2.597308040300117e-07, "loss": 1.4057, "step": 37179 }, { "epoch": 92.95, "grad_norm": 6.03125, "learning_rate": 2.5954749630248355e-07, "loss": 1.2322, "step": 37180 }, { "epoch": 92.9525, "grad_norm": 6.5, "learning_rate": 2.5936425243297315e-07, "loss": 1.3027, "step": 37181 }, { "epoch": 92.955, "grad_norm": 6.65625, "learning_rate": 2.5918107242267955e-07, "loss": 1.4396, "step": 37182 }, { "epoch": 92.9575, "grad_norm": 6.21875, "learning_rate": 2.589979562728062e-07, "loss": 1.1976, "step": 37183 }, { "epoch": 92.96, "grad_norm": 6.875, "learning_rate": 2.588149039845533e-07, "loss": 1.2573, "step": 37184 }, { "epoch": 92.9625, "grad_norm": 6.15625, "learning_rate": 2.586319155591188e-07, "loss": 1.3926, "step": 37185 }, { "epoch": 92.965, "grad_norm": 8.25, "learning_rate": 2.5844899099770283e-07, "loss": 1.1223, "step": 37186 }, { "epoch": 92.9675, "grad_norm": 8.5, "learning_rate": 2.582661303015066e-07, "loss": 1.3023, "step": 37187 }, { "epoch": 92.97, "grad_norm": 5.46875, "learning_rate": 2.5808333347172807e-07, "loss": 1.2744, "step": 37188 }, { "epoch": 92.9725, "grad_norm": 5.96875, "learning_rate": 2.5790060050956523e-07, "loss": 1.2626, "step": 37189 }, { "epoch": 92.975, "grad_norm": 9.0625, "learning_rate": 2.577179314162148e-07, "loss": 1.3176, "step": 37190 }, { "epoch": 92.9775, "grad_norm": 7.03125, "learning_rate": 2.57535326192877e-07, "loss": 1.4436, "step": 37191 }, { "epoch": 92.98, "grad_norm": 6.78125, "learning_rate": 2.5735278484074865e-07, "loss": 1.298, "step": 37192 }, { "epoch": 92.9825, "grad_norm": 6.75, "learning_rate": 2.571703073610243e-07, "loss": 1.5166, "step": 37193 }, { "epoch": 92.985, "grad_norm": 8.25, "learning_rate": 2.56987893754902e-07, "loss": 1.2166, "step": 37194 }, { "epoch": 92.9875, "grad_norm": 8.4375, "learning_rate": 2.5680554402357617e-07, "loss": 1.524, "step": 37195 }, { "epoch": 92.99, "grad_norm": 6.1875, "learning_rate": 2.566232581682449e-07, "loss": 1.209, "step": 37196 }, { "epoch": 92.9925, "grad_norm": 9.0, "learning_rate": 2.564410361900993e-07, "loss": 1.3436, "step": 37197 }, { "epoch": 92.995, "grad_norm": 6.40625, "learning_rate": 2.5625887809033855e-07, "loss": 1.3709, "step": 37198 }, { "epoch": 92.9975, "grad_norm": 9.0625, "learning_rate": 2.56076783870155e-07, "loss": 1.4651, "step": 37199 }, { "epoch": 93.0, "grad_norm": 10.25, "learning_rate": 2.5589475353073987e-07, "loss": 1.3584, "step": 37200 }, { "epoch": 93.0025, "grad_norm": 7.28125, "learning_rate": 2.5571278707329006e-07, "loss": 1.4566, "step": 37201 }, { "epoch": 93.005, "grad_norm": 8.5, "learning_rate": 2.555308844989968e-07, "loss": 1.2163, "step": 37202 }, { "epoch": 93.0075, "grad_norm": 6.28125, "learning_rate": 2.5534904580905463e-07, "loss": 1.362, "step": 37203 }, { "epoch": 93.01, "grad_norm": 7.03125, "learning_rate": 2.551672710046527e-07, "loss": 1.2718, "step": 37204 }, { "epoch": 93.0125, "grad_norm": 7.25, "learning_rate": 2.5498556008698327e-07, "loss": 1.2, "step": 37205 }, { "epoch": 93.015, "grad_norm": 7.65625, "learning_rate": 2.5480391305723993e-07, "loss": 1.3472, "step": 37206 }, { "epoch": 93.0175, "grad_norm": 8.875, "learning_rate": 2.5462232991661286e-07, "loss": 1.3078, "step": 37207 }, { "epoch": 93.02, "grad_norm": 7.28125, "learning_rate": 2.54440810666291e-07, "loss": 1.3254, "step": 37208 }, { "epoch": 93.0225, "grad_norm": 10.4375, "learning_rate": 2.542593553074646e-07, "loss": 1.4348, "step": 37209 }, { "epoch": 93.025, "grad_norm": 10.0, "learning_rate": 2.540779638413249e-07, "loss": 1.2815, "step": 37210 }, { "epoch": 93.0275, "grad_norm": 8.25, "learning_rate": 2.538966362690598e-07, "loss": 1.2302, "step": 37211 }, { "epoch": 93.03, "grad_norm": 7.28125, "learning_rate": 2.537153725918573e-07, "loss": 1.3453, "step": 37212 }, { "epoch": 93.0325, "grad_norm": 10.8125, "learning_rate": 2.5353417281090754e-07, "loss": 1.3005, "step": 37213 }, { "epoch": 93.035, "grad_norm": 8.625, "learning_rate": 2.5335303692739735e-07, "loss": 1.2488, "step": 37214 }, { "epoch": 93.0375, "grad_norm": 7.3125, "learning_rate": 2.531719649425157e-07, "loss": 1.3721, "step": 37215 }, { "epoch": 93.04, "grad_norm": 7.625, "learning_rate": 2.5299095685744734e-07, "loss": 1.2204, "step": 37216 }, { "epoch": 93.0425, "grad_norm": 5.90625, "learning_rate": 2.5281001267338123e-07, "loss": 1.1835, "step": 37217 }, { "epoch": 93.045, "grad_norm": 6.71875, "learning_rate": 2.526291323915031e-07, "loss": 1.278, "step": 37218 }, { "epoch": 93.0475, "grad_norm": 6.96875, "learning_rate": 2.5244831601299644e-07, "loss": 1.3367, "step": 37219 }, { "epoch": 93.05, "grad_norm": 6.5, "learning_rate": 2.5226756353904925e-07, "loss": 1.2067, "step": 37220 }, { "epoch": 93.0525, "grad_norm": 7.84375, "learning_rate": 2.5208687497084605e-07, "loss": 1.2711, "step": 37221 }, { "epoch": 93.055, "grad_norm": 7.5, "learning_rate": 2.519062503095715e-07, "loss": 1.2084, "step": 37222 }, { "epoch": 93.0575, "grad_norm": 6.4375, "learning_rate": 2.517256895564091e-07, "loss": 1.2623, "step": 37223 }, { "epoch": 93.06, "grad_norm": 8.5, "learning_rate": 2.515451927125423e-07, "loss": 1.1614, "step": 37224 }, { "epoch": 93.0625, "grad_norm": 10.5, "learning_rate": 2.5136475977915686e-07, "loss": 1.2614, "step": 37225 }, { "epoch": 93.065, "grad_norm": 7.3125, "learning_rate": 2.5118439075743295e-07, "loss": 1.3526, "step": 37226 }, { "epoch": 93.0675, "grad_norm": 7.40625, "learning_rate": 2.510040856485541e-07, "loss": 1.4447, "step": 37227 }, { "epoch": 93.07, "grad_norm": 8.0, "learning_rate": 2.5082384445370254e-07, "loss": 1.33, "step": 37228 }, { "epoch": 93.0725, "grad_norm": 8.5, "learning_rate": 2.5064366717405864e-07, "loss": 1.2668, "step": 37229 }, { "epoch": 93.075, "grad_norm": 8.25, "learning_rate": 2.504635538108058e-07, "loss": 1.3809, "step": 37230 }, { "epoch": 93.0775, "grad_norm": 8.375, "learning_rate": 2.502835043651231e-07, "loss": 1.5823, "step": 37231 }, { "epoch": 93.08, "grad_norm": 6.03125, "learning_rate": 2.5010351883819283e-07, "loss": 1.2887, "step": 37232 }, { "epoch": 93.0825, "grad_norm": 6.875, "learning_rate": 2.4992359723119307e-07, "loss": 1.3287, "step": 37233 }, { "epoch": 93.085, "grad_norm": 6.5, "learning_rate": 2.4974373954530395e-07, "loss": 1.3756, "step": 37234 }, { "epoch": 93.0875, "grad_norm": 20.625, "learning_rate": 2.495639457817034e-07, "loss": 1.4798, "step": 37235 }, { "epoch": 93.09, "grad_norm": 8.5, "learning_rate": 2.493842159415738e-07, "loss": 1.3244, "step": 37236 }, { "epoch": 93.0925, "grad_norm": 7.3125, "learning_rate": 2.492045500260898e-07, "loss": 1.2011, "step": 37237 }, { "epoch": 93.095, "grad_norm": 6.1875, "learning_rate": 2.490249480364304e-07, "loss": 1.2114, "step": 37238 }, { "epoch": 93.0975, "grad_norm": 6.4375, "learning_rate": 2.488454099737725e-07, "loss": 1.3631, "step": 37239 }, { "epoch": 93.1, "grad_norm": 9.25, "learning_rate": 2.486659358392951e-07, "loss": 1.3117, "step": 37240 }, { "epoch": 93.1025, "grad_norm": 5.25, "learning_rate": 2.4848652563417396e-07, "loss": 1.2587, "step": 37241 }, { "epoch": 93.105, "grad_norm": 9.375, "learning_rate": 2.483071793595837e-07, "loss": 1.3105, "step": 37242 }, { "epoch": 93.1075, "grad_norm": 7.09375, "learning_rate": 2.4812789701670225e-07, "loss": 1.3701, "step": 37243 }, { "epoch": 93.11, "grad_norm": 10.375, "learning_rate": 2.4794867860670425e-07, "loss": 1.2838, "step": 37244 }, { "epoch": 93.1125, "grad_norm": 8.875, "learning_rate": 2.477695241307643e-07, "loss": 1.3885, "step": 37245 }, { "epoch": 93.115, "grad_norm": 38.5, "learning_rate": 2.4759043359005585e-07, "loss": 1.3909, "step": 37246 }, { "epoch": 93.1175, "grad_norm": 6.21875, "learning_rate": 2.4741140698575586e-07, "loss": 1.1522, "step": 37247 }, { "epoch": 93.12, "grad_norm": 7.53125, "learning_rate": 2.472324443190355e-07, "loss": 1.1851, "step": 37248 }, { "epoch": 93.1225, "grad_norm": 7.8125, "learning_rate": 2.470535455910694e-07, "loss": 1.2865, "step": 37249 }, { "epoch": 93.125, "grad_norm": 6.46875, "learning_rate": 2.468747108030289e-07, "loss": 1.2706, "step": 37250 }, { "epoch": 93.1275, "grad_norm": 5.59375, "learning_rate": 2.4669593995608864e-07, "loss": 1.3183, "step": 37251 }, { "epoch": 93.13, "grad_norm": 5.4375, "learning_rate": 2.4651723305141986e-07, "loss": 1.1636, "step": 37252 }, { "epoch": 93.1325, "grad_norm": 10.0625, "learning_rate": 2.463385900901927e-07, "loss": 1.397, "step": 37253 }, { "epoch": 93.135, "grad_norm": 5.5625, "learning_rate": 2.4616001107357843e-07, "loss": 1.2001, "step": 37254 }, { "epoch": 93.1375, "grad_norm": 7.4375, "learning_rate": 2.459814960027507e-07, "loss": 1.2846, "step": 37255 }, { "epoch": 93.14, "grad_norm": 10.5625, "learning_rate": 2.458030448788784e-07, "loss": 1.225, "step": 37256 }, { "epoch": 93.1425, "grad_norm": 6.46875, "learning_rate": 2.456246577031296e-07, "loss": 1.2873, "step": 37257 }, { "epoch": 93.145, "grad_norm": 7.40625, "learning_rate": 2.4544633447667553e-07, "loss": 1.1738, "step": 37258 }, { "epoch": 93.1475, "grad_norm": 9.75, "learning_rate": 2.4526807520068527e-07, "loss": 1.2431, "step": 37259 }, { "epoch": 93.15, "grad_norm": 6.4375, "learning_rate": 2.450898798763268e-07, "loss": 1.2883, "step": 37260 }, { "epoch": 93.1525, "grad_norm": 5.84375, "learning_rate": 2.449117485047681e-07, "loss": 1.2931, "step": 37261 }, { "epoch": 93.155, "grad_norm": 6.0625, "learning_rate": 2.4473368108717923e-07, "loss": 1.2328, "step": 37262 }, { "epoch": 93.1575, "grad_norm": 7.9375, "learning_rate": 2.4455567762472486e-07, "loss": 1.3567, "step": 37263 }, { "epoch": 93.16, "grad_norm": 8.9375, "learning_rate": 2.4437773811857304e-07, "loss": 1.3361, "step": 37264 }, { "epoch": 93.1625, "grad_norm": 6.53125, "learning_rate": 2.441998625698905e-07, "loss": 1.3841, "step": 37265 }, { "epoch": 93.165, "grad_norm": 7.375, "learning_rate": 2.440220509798441e-07, "loss": 1.3201, "step": 37266 }, { "epoch": 93.1675, "grad_norm": 6.5, "learning_rate": 2.4384430334959743e-07, "loss": 1.2914, "step": 37267 }, { "epoch": 93.17, "grad_norm": 8.0625, "learning_rate": 2.4366661968031836e-07, "loss": 1.2645, "step": 37268 }, { "epoch": 93.1725, "grad_norm": 7.375, "learning_rate": 2.434889999731682e-07, "loss": 1.3986, "step": 37269 }, { "epoch": 93.175, "grad_norm": 7.34375, "learning_rate": 2.433114442293161e-07, "loss": 1.1395, "step": 37270 }, { "epoch": 93.1775, "grad_norm": 7.6875, "learning_rate": 2.431339524499232e-07, "loss": 1.3253, "step": 37271 }, { "epoch": 93.18, "grad_norm": 10.5625, "learning_rate": 2.429565246361532e-07, "loss": 1.4103, "step": 37272 }, { "epoch": 93.1825, "grad_norm": 7.25, "learning_rate": 2.427791607891705e-07, "loss": 1.3376, "step": 37273 }, { "epoch": 93.185, "grad_norm": 8.1875, "learning_rate": 2.426018609101355e-07, "loss": 1.3051, "step": 37274 }, { "epoch": 93.1875, "grad_norm": 10.125, "learning_rate": 2.424246250002138e-07, "loss": 1.2941, "step": 37275 }, { "epoch": 93.19, "grad_norm": 7.4375, "learning_rate": 2.422474530605656e-07, "loss": 1.4349, "step": 37276 }, { "epoch": 93.1925, "grad_norm": 9.0, "learning_rate": 2.420703450923523e-07, "loss": 1.2078, "step": 37277 }, { "epoch": 93.195, "grad_norm": 5.03125, "learning_rate": 2.4189330109673615e-07, "loss": 1.2186, "step": 37278 }, { "epoch": 93.1975, "grad_norm": 8.9375, "learning_rate": 2.417163210748752e-07, "loss": 1.1974, "step": 37279 }, { "epoch": 93.2, "grad_norm": 5.59375, "learning_rate": 2.4153940502793185e-07, "loss": 1.2225, "step": 37280 }, { "epoch": 93.2025, "grad_norm": 5.46875, "learning_rate": 2.413625529570673e-07, "loss": 1.237, "step": 37281 }, { "epoch": 93.205, "grad_norm": 8.25, "learning_rate": 2.411857648634386e-07, "loss": 1.3482, "step": 37282 }, { "epoch": 93.2075, "grad_norm": 8.4375, "learning_rate": 2.4100904074820577e-07, "loss": 1.3392, "step": 37283 }, { "epoch": 93.21, "grad_norm": 6.875, "learning_rate": 2.4083238061252565e-07, "loss": 1.3031, "step": 37284 }, { "epoch": 93.2125, "grad_norm": 6.875, "learning_rate": 2.4065578445755967e-07, "loss": 1.2809, "step": 37285 }, { "epoch": 93.215, "grad_norm": 10.0625, "learning_rate": 2.4047925228446345e-07, "loss": 1.3622, "step": 37286 }, { "epoch": 93.2175, "grad_norm": 6.59375, "learning_rate": 2.40302784094395e-07, "loss": 1.4351, "step": 37287 }, { "epoch": 93.22, "grad_norm": 7.6875, "learning_rate": 2.4012637988850894e-07, "loss": 1.2988, "step": 37288 }, { "epoch": 93.2225, "grad_norm": 8.0, "learning_rate": 2.3995003966796437e-07, "loss": 1.2773, "step": 37289 }, { "epoch": 93.225, "grad_norm": 6.71875, "learning_rate": 2.397737634339181e-07, "loss": 1.1985, "step": 37290 }, { "epoch": 93.2275, "grad_norm": 9.6875, "learning_rate": 2.395975511875248e-07, "loss": 1.4233, "step": 37291 }, { "epoch": 93.23, "grad_norm": 6.46875, "learning_rate": 2.39421402929938e-07, "loss": 1.3568, "step": 37292 }, { "epoch": 93.2325, "grad_norm": 6.625, "learning_rate": 2.392453186623145e-07, "loss": 1.3481, "step": 37293 }, { "epoch": 93.235, "grad_norm": 8.6875, "learning_rate": 2.390692983858078e-07, "loss": 1.2695, "step": 37294 }, { "epoch": 93.2375, "grad_norm": 5.96875, "learning_rate": 2.388933421015727e-07, "loss": 1.2973, "step": 37295 }, { "epoch": 93.24, "grad_norm": 6.6875, "learning_rate": 2.387174498107614e-07, "loss": 1.1722, "step": 37296 }, { "epoch": 93.2425, "grad_norm": 8.0625, "learning_rate": 2.3854162151452865e-07, "loss": 1.2707, "step": 37297 }, { "epoch": 93.245, "grad_norm": 5.9375, "learning_rate": 2.383658572140257e-07, "loss": 1.3164, "step": 37298 }, { "epoch": 93.2475, "grad_norm": 7.34375, "learning_rate": 2.3819015691040504e-07, "loss": 1.2847, "step": 37299 }, { "epoch": 93.25, "grad_norm": 6.65625, "learning_rate": 2.380145206048201e-07, "loss": 1.175, "step": 37300 }, { "epoch": 93.2525, "grad_norm": 11.8125, "learning_rate": 2.3783894829842113e-07, "loss": 1.1918, "step": 37301 }, { "epoch": 93.255, "grad_norm": 5.9375, "learning_rate": 2.3766343999236051e-07, "loss": 1.2248, "step": 37302 }, { "epoch": 93.2575, "grad_norm": 6.78125, "learning_rate": 2.3748799568778514e-07, "loss": 1.4192, "step": 37303 }, { "epoch": 93.26, "grad_norm": 7.96875, "learning_rate": 2.373126153858496e-07, "loss": 1.2873, "step": 37304 }, { "epoch": 93.2625, "grad_norm": 9.5, "learning_rate": 2.3713729908770188e-07, "loss": 1.396, "step": 37305 }, { "epoch": 93.265, "grad_norm": 6.84375, "learning_rate": 2.3696204679449108e-07, "loss": 1.244, "step": 37306 }, { "epoch": 93.2675, "grad_norm": 6.21875, "learning_rate": 2.3678685850736517e-07, "loss": 1.253, "step": 37307 }, { "epoch": 93.27, "grad_norm": 8.8125, "learning_rate": 2.366117342274754e-07, "loss": 1.293, "step": 37308 }, { "epoch": 93.2725, "grad_norm": 7.59375, "learning_rate": 2.364366739559676e-07, "loss": 1.3653, "step": 37309 }, { "epoch": 93.275, "grad_norm": 7.90625, "learning_rate": 2.3626167769399188e-07, "loss": 1.3087, "step": 37310 }, { "epoch": 93.2775, "grad_norm": 6.96875, "learning_rate": 2.3608674544269183e-07, "loss": 1.3044, "step": 37311 }, { "epoch": 93.28, "grad_norm": 8.1875, "learning_rate": 2.359118772032176e-07, "loss": 1.3083, "step": 37312 }, { "epoch": 93.2825, "grad_norm": 8.3125, "learning_rate": 2.357370729767139e-07, "loss": 1.359, "step": 37313 }, { "epoch": 93.285, "grad_norm": 8.0625, "learning_rate": 2.355623327643264e-07, "loss": 1.3442, "step": 37314 }, { "epoch": 93.2875, "grad_norm": 8.5, "learning_rate": 2.3538765656720197e-07, "loss": 1.3671, "step": 37315 }, { "epoch": 93.29, "grad_norm": 7.21875, "learning_rate": 2.352130443864864e-07, "loss": 1.1871, "step": 37316 }, { "epoch": 93.2925, "grad_norm": 6.9375, "learning_rate": 2.350384962233232e-07, "loss": 1.2638, "step": 37317 }, { "epoch": 93.295, "grad_norm": 6.34375, "learning_rate": 2.348640120788559e-07, "loss": 1.312, "step": 37318 }, { "epoch": 93.2975, "grad_norm": 8.9375, "learning_rate": 2.3468959195422914e-07, "loss": 1.3229, "step": 37319 }, { "epoch": 93.3, "grad_norm": 8.25, "learning_rate": 2.3451523585058756e-07, "loss": 1.3706, "step": 37320 }, { "epoch": 93.3025, "grad_norm": 6.90625, "learning_rate": 2.3434094376907357e-07, "loss": 1.2437, "step": 37321 }, { "epoch": 93.305, "grad_norm": 5.5625, "learning_rate": 2.341667157108285e-07, "loss": 1.2766, "step": 37322 }, { "epoch": 93.3075, "grad_norm": 7.65625, "learning_rate": 2.3399255167699586e-07, "loss": 1.1639, "step": 37323 }, { "epoch": 93.31, "grad_norm": 8.4375, "learning_rate": 2.3381845166871698e-07, "loss": 1.3513, "step": 37324 }, { "epoch": 93.3125, "grad_norm": 7.375, "learning_rate": 2.3364441568713424e-07, "loss": 1.2943, "step": 37325 }, { "epoch": 93.315, "grad_norm": 5.15625, "learning_rate": 2.334704437333879e-07, "loss": 1.4945, "step": 37326 }, { "epoch": 93.3175, "grad_norm": 5.96875, "learning_rate": 2.332965358086181e-07, "loss": 1.391, "step": 37327 }, { "epoch": 93.32, "grad_norm": 7.15625, "learning_rate": 2.3312269191396619e-07, "loss": 1.3312, "step": 37328 }, { "epoch": 93.3225, "grad_norm": 6.625, "learning_rate": 2.3294891205057012e-07, "loss": 1.2359, "step": 37329 }, { "epoch": 93.325, "grad_norm": 7.625, "learning_rate": 2.32775196219569e-07, "loss": 1.3403, "step": 37330 }, { "epoch": 93.3275, "grad_norm": 7.625, "learning_rate": 2.3260154442210526e-07, "loss": 1.2702, "step": 37331 }, { "epoch": 93.33, "grad_norm": 8.6875, "learning_rate": 2.324279566593135e-07, "loss": 1.3831, "step": 37332 }, { "epoch": 93.3325, "grad_norm": 8.6875, "learning_rate": 2.3225443293233397e-07, "loss": 1.235, "step": 37333 }, { "epoch": 93.335, "grad_norm": 7.71875, "learning_rate": 2.3208097324230238e-07, "loss": 1.3606, "step": 37334 }, { "epoch": 93.3375, "grad_norm": 9.125, "learning_rate": 2.3190757759035787e-07, "loss": 1.2462, "step": 37335 }, { "epoch": 93.34, "grad_norm": 6.5, "learning_rate": 2.3173424597763615e-07, "loss": 1.2258, "step": 37336 }, { "epoch": 93.3425, "grad_norm": 6.65625, "learning_rate": 2.3156097840527413e-07, "loss": 1.2643, "step": 37337 }, { "epoch": 93.345, "grad_norm": 6.21875, "learning_rate": 2.3138777487440755e-07, "loss": 1.2572, "step": 37338 }, { "epoch": 93.3475, "grad_norm": 12.1875, "learning_rate": 2.3121463538617105e-07, "loss": 1.4314, "step": 37339 }, { "epoch": 93.35, "grad_norm": 9.1875, "learning_rate": 2.3104155994170042e-07, "loss": 1.2208, "step": 37340 }, { "epoch": 93.3525, "grad_norm": 6.71875, "learning_rate": 2.3086854854213137e-07, "loss": 1.2138, "step": 37341 }, { "epoch": 93.355, "grad_norm": 7.71875, "learning_rate": 2.3069560118859636e-07, "loss": 1.1374, "step": 37342 }, { "epoch": 93.3575, "grad_norm": 8.5, "learning_rate": 2.3052271788223112e-07, "loss": 1.2597, "step": 37343 }, { "epoch": 93.36, "grad_norm": 6.5625, "learning_rate": 2.30349898624167e-07, "loss": 1.2066, "step": 37344 }, { "epoch": 93.3625, "grad_norm": 8.75, "learning_rate": 2.3017714341553753e-07, "loss": 1.3519, "step": 37345 }, { "epoch": 93.365, "grad_norm": 10.8125, "learning_rate": 2.3000445225747626e-07, "loss": 1.4708, "step": 37346 }, { "epoch": 93.3675, "grad_norm": 11.25, "learning_rate": 2.2983182515111556e-07, "loss": 1.2731, "step": 37347 }, { "epoch": 93.37, "grad_norm": 7.1875, "learning_rate": 2.2965926209758572e-07, "loss": 1.3659, "step": 37348 }, { "epoch": 93.3725, "grad_norm": 6.75, "learning_rate": 2.2948676309801797e-07, "loss": 1.2275, "step": 37349 }, { "epoch": 93.375, "grad_norm": 6.625, "learning_rate": 2.2931432815354593e-07, "loss": 1.2877, "step": 37350 }, { "epoch": 93.3775, "grad_norm": 6.6875, "learning_rate": 2.2914195726529753e-07, "loss": 1.4486, "step": 37351 }, { "epoch": 93.38, "grad_norm": 8.5625, "learning_rate": 2.289696504344019e-07, "loss": 1.3267, "step": 37352 }, { "epoch": 93.3825, "grad_norm": 5.21875, "learning_rate": 2.2879740766199143e-07, "loss": 1.3644, "step": 37353 }, { "epoch": 93.385, "grad_norm": 10.4375, "learning_rate": 2.2862522894919415e-07, "loss": 1.3563, "step": 37354 }, { "epoch": 93.3875, "grad_norm": 8.25, "learning_rate": 2.2845311429713914e-07, "loss": 1.5141, "step": 37355 }, { "epoch": 93.39, "grad_norm": 7.5, "learning_rate": 2.2828106370695325e-07, "loss": 1.3141, "step": 37356 }, { "epoch": 93.3925, "grad_norm": 9.6875, "learning_rate": 2.2810907717976782e-07, "loss": 1.3308, "step": 37357 }, { "epoch": 93.395, "grad_norm": 7.40625, "learning_rate": 2.279371547167064e-07, "loss": 1.2169, "step": 37358 }, { "epoch": 93.3975, "grad_norm": 7.9375, "learning_rate": 2.277652963188981e-07, "loss": 1.2861, "step": 37359 }, { "epoch": 93.4, "grad_norm": 6.5625, "learning_rate": 2.2759350198746978e-07, "loss": 1.2307, "step": 37360 }, { "epoch": 93.4025, "grad_norm": 7.6875, "learning_rate": 2.274217717235483e-07, "loss": 1.2107, "step": 37361 }, { "epoch": 93.405, "grad_norm": 10.3125, "learning_rate": 2.2725010552825722e-07, "loss": 1.3924, "step": 37362 }, { "epoch": 93.4075, "grad_norm": 5.9375, "learning_rate": 2.2707850340272342e-07, "loss": 1.2973, "step": 37363 }, { "epoch": 93.41, "grad_norm": 7.84375, "learning_rate": 2.2690696534807267e-07, "loss": 1.3211, "step": 37364 }, { "epoch": 93.4125, "grad_norm": 7.25, "learning_rate": 2.267354913654285e-07, "loss": 1.3004, "step": 37365 }, { "epoch": 93.415, "grad_norm": 9.4375, "learning_rate": 2.2656408145591558e-07, "loss": 1.3328, "step": 37366 }, { "epoch": 93.4175, "grad_norm": 9.125, "learning_rate": 2.2639273562065632e-07, "loss": 1.2204, "step": 37367 }, { "epoch": 93.42, "grad_norm": 8.8125, "learning_rate": 2.2622145386077653e-07, "loss": 1.3941, "step": 37368 }, { "epoch": 93.4225, "grad_norm": 6.25, "learning_rate": 2.2605023617739752e-07, "loss": 1.3545, "step": 37369 }, { "epoch": 93.425, "grad_norm": 6.78125, "learning_rate": 2.2587908257164172e-07, "loss": 1.3003, "step": 37370 }, { "epoch": 93.4275, "grad_norm": 9.1875, "learning_rate": 2.2570799304463043e-07, "loss": 1.4555, "step": 37371 }, { "epoch": 93.43, "grad_norm": 6.875, "learning_rate": 2.2553696759748834e-07, "loss": 1.2687, "step": 37372 }, { "epoch": 93.4325, "grad_norm": 6.28125, "learning_rate": 2.2536600623133452e-07, "loss": 1.5416, "step": 37373 }, { "epoch": 93.435, "grad_norm": 8.4375, "learning_rate": 2.2519510894728924e-07, "loss": 1.285, "step": 37374 }, { "epoch": 93.4375, "grad_norm": 9.75, "learning_rate": 2.2502427574647268e-07, "loss": 1.3169, "step": 37375 }, { "epoch": 93.44, "grad_norm": 7.3125, "learning_rate": 2.2485350663000727e-07, "loss": 1.3114, "step": 37376 }, { "epoch": 93.4425, "grad_norm": 8.5, "learning_rate": 2.24682801599011e-07, "loss": 1.3612, "step": 37377 }, { "epoch": 93.445, "grad_norm": 5.90625, "learning_rate": 2.2451216065460301e-07, "loss": 1.3961, "step": 37378 }, { "epoch": 93.4475, "grad_norm": 7.09375, "learning_rate": 2.2434158379790129e-07, "loss": 1.3659, "step": 37379 }, { "epoch": 93.45, "grad_norm": 7.40625, "learning_rate": 2.24171071030026e-07, "loss": 1.4847, "step": 37380 }, { "epoch": 93.4525, "grad_norm": 10.0, "learning_rate": 2.2400062235209407e-07, "loss": 1.3403, "step": 37381 }, { "epoch": 93.455, "grad_norm": 7.34375, "learning_rate": 2.238302377652213e-07, "loss": 1.3513, "step": 37382 }, { "epoch": 93.4575, "grad_norm": 7.875, "learning_rate": 2.2365991727052781e-07, "loss": 1.2763, "step": 37383 }, { "epoch": 93.46, "grad_norm": 9.6875, "learning_rate": 2.2348966086912725e-07, "loss": 1.4921, "step": 37384 }, { "epoch": 93.4625, "grad_norm": 6.71875, "learning_rate": 2.233194685621387e-07, "loss": 1.3354, "step": 37385 }, { "epoch": 93.465, "grad_norm": 7.15625, "learning_rate": 2.2314934035067571e-07, "loss": 1.2778, "step": 37386 }, { "epoch": 93.4675, "grad_norm": 13.5625, "learning_rate": 2.2297927623585513e-07, "loss": 1.2638, "step": 37387 }, { "epoch": 93.47, "grad_norm": 8.6875, "learning_rate": 2.2280927621879056e-07, "loss": 1.3814, "step": 37388 }, { "epoch": 93.4725, "grad_norm": 10.1875, "learning_rate": 2.2263934030059664e-07, "loss": 1.3276, "step": 37389 }, { "epoch": 93.475, "grad_norm": 9.3125, "learning_rate": 2.2246946848238805e-07, "loss": 1.3581, "step": 37390 }, { "epoch": 93.4775, "grad_norm": 4.65625, "learning_rate": 2.2229966076527943e-07, "loss": 1.3257, "step": 37391 }, { "epoch": 93.48, "grad_norm": 6.9375, "learning_rate": 2.2212991715038324e-07, "loss": 1.2641, "step": 37392 }, { "epoch": 93.4825, "grad_norm": 5.4375, "learning_rate": 2.219602376388097e-07, "loss": 1.3423, "step": 37393 }, { "epoch": 93.485, "grad_norm": 5.90625, "learning_rate": 2.2179062223167458e-07, "loss": 1.2983, "step": 37394 }, { "epoch": 93.4875, "grad_norm": 7.8125, "learning_rate": 2.2162107093009034e-07, "loss": 1.2621, "step": 37395 }, { "epoch": 93.49, "grad_norm": 9.0, "learning_rate": 2.2145158373516606e-07, "loss": 1.3075, "step": 37396 }, { "epoch": 93.4925, "grad_norm": 6.65625, "learning_rate": 2.2128216064801311e-07, "loss": 1.3571, "step": 37397 }, { "epoch": 93.495, "grad_norm": 10.0625, "learning_rate": 2.2111280166974392e-07, "loss": 1.325, "step": 37398 }, { "epoch": 93.4975, "grad_norm": 6.6875, "learning_rate": 2.2094350680146647e-07, "loss": 1.0817, "step": 37399 }, { "epoch": 93.5, "grad_norm": 6.15625, "learning_rate": 2.2077427604429435e-07, "loss": 1.236, "step": 37400 }, { "epoch": 93.5025, "grad_norm": 6.90625, "learning_rate": 2.2060510939933332e-07, "loss": 1.2991, "step": 37401 }, { "epoch": 93.505, "grad_norm": 6.9375, "learning_rate": 2.2043600686769361e-07, "loss": 1.4577, "step": 37402 }, { "epoch": 93.5075, "grad_norm": 7.71875, "learning_rate": 2.2026696845048546e-07, "loss": 1.1836, "step": 37403 }, { "epoch": 93.51, "grad_norm": 8.25, "learning_rate": 2.200979941488146e-07, "loss": 1.2764, "step": 37404 }, { "epoch": 93.5125, "grad_norm": 6.96875, "learning_rate": 2.199290839637891e-07, "loss": 1.3953, "step": 37405 }, { "epoch": 93.515, "grad_norm": 15.625, "learning_rate": 2.19760237896518e-07, "loss": 1.3666, "step": 37406 }, { "epoch": 93.5175, "grad_norm": 6.6875, "learning_rate": 2.1959145594810826e-07, "loss": 1.2644, "step": 37407 }, { "epoch": 93.52, "grad_norm": 9.125, "learning_rate": 2.1942273811966563e-07, "loss": 1.2596, "step": 37408 }, { "epoch": 93.5225, "grad_norm": 9.25, "learning_rate": 2.1925408441229368e-07, "loss": 1.4444, "step": 37409 }, { "epoch": 93.525, "grad_norm": 10.625, "learning_rate": 2.1908549482710263e-07, "loss": 1.5649, "step": 37410 }, { "epoch": 93.5275, "grad_norm": 7.25, "learning_rate": 2.1891696936519603e-07, "loss": 1.1881, "step": 37411 }, { "epoch": 93.53, "grad_norm": 6.0625, "learning_rate": 2.1874850802767634e-07, "loss": 1.5188, "step": 37412 }, { "epoch": 93.5325, "grad_norm": 7.71875, "learning_rate": 2.1858011081565156e-07, "loss": 1.4012, "step": 37413 }, { "epoch": 93.535, "grad_norm": 5.0, "learning_rate": 2.1841177773022416e-07, "loss": 1.2581, "step": 37414 }, { "epoch": 93.5375, "grad_norm": 6.0625, "learning_rate": 2.1824350877249767e-07, "loss": 1.3261, "step": 37415 }, { "epoch": 93.54, "grad_norm": 6.84375, "learning_rate": 2.1807530394357457e-07, "loss": 1.3723, "step": 37416 }, { "epoch": 93.5425, "grad_norm": 8.25, "learning_rate": 2.179071632445584e-07, "loss": 1.2492, "step": 37417 }, { "epoch": 93.545, "grad_norm": 8.375, "learning_rate": 2.1773908667655275e-07, "loss": 1.2709, "step": 37418 }, { "epoch": 93.5475, "grad_norm": 5.8125, "learning_rate": 2.175710742406567e-07, "loss": 1.4426, "step": 37419 }, { "epoch": 93.55, "grad_norm": 8.75, "learning_rate": 2.1740312593797274e-07, "loss": 1.3051, "step": 37420 }, { "epoch": 93.5525, "grad_norm": 7.0625, "learning_rate": 2.1723524176960442e-07, "loss": 1.2738, "step": 37421 }, { "epoch": 93.555, "grad_norm": 6.78125, "learning_rate": 2.1706742173664974e-07, "loss": 1.4642, "step": 37422 }, { "epoch": 93.5575, "grad_norm": 6.96875, "learning_rate": 2.1689966584020895e-07, "loss": 1.3288, "step": 37423 }, { "epoch": 93.56, "grad_norm": 9.0625, "learning_rate": 2.1673197408138115e-07, "loss": 1.3989, "step": 37424 }, { "epoch": 93.5625, "grad_norm": 6.84375, "learning_rate": 2.1656434646126878e-07, "loss": 1.3754, "step": 37425 }, { "epoch": 93.565, "grad_norm": 5.90625, "learning_rate": 2.1639678298096988e-07, "loss": 1.269, "step": 37426 }, { "epoch": 93.5675, "grad_norm": 10.0, "learning_rate": 2.1622928364158136e-07, "loss": 1.2414, "step": 37427 }, { "epoch": 93.57, "grad_norm": 5.59375, "learning_rate": 2.160618484442012e-07, "loss": 1.535, "step": 37428 }, { "epoch": 93.5725, "grad_norm": 10.75, "learning_rate": 2.158944773899274e-07, "loss": 1.4019, "step": 37429 }, { "epoch": 93.575, "grad_norm": 7.46875, "learning_rate": 2.1572717047986025e-07, "loss": 1.372, "step": 37430 }, { "epoch": 93.5775, "grad_norm": 8.5, "learning_rate": 2.1555992771509325e-07, "loss": 1.1817, "step": 37431 }, { "epoch": 93.58, "grad_norm": 7.40625, "learning_rate": 2.1539274909672337e-07, "loss": 1.475, "step": 37432 }, { "epoch": 93.5825, "grad_norm": 6.46875, "learning_rate": 2.1522563462584744e-07, "loss": 1.3402, "step": 37433 }, { "epoch": 93.585, "grad_norm": 7.25, "learning_rate": 2.150585843035602e-07, "loss": 1.1645, "step": 37434 }, { "epoch": 93.5875, "grad_norm": 4.28125, "learning_rate": 2.148915981309585e-07, "loss": 1.349, "step": 37435 }, { "epoch": 93.59, "grad_norm": 9.5625, "learning_rate": 2.1472467610913482e-07, "loss": 1.1776, "step": 37436 }, { "epoch": 93.5925, "grad_norm": 5.46875, "learning_rate": 2.14557818239185e-07, "loss": 1.3202, "step": 37437 }, { "epoch": 93.595, "grad_norm": 6.8125, "learning_rate": 2.143910245222025e-07, "loss": 1.297, "step": 37438 }, { "epoch": 93.5975, "grad_norm": 4.71875, "learning_rate": 2.1422429495927988e-07, "loss": 1.3109, "step": 37439 }, { "epoch": 93.6, "grad_norm": 6.90625, "learning_rate": 2.1405762955151178e-07, "loss": 1.3423, "step": 37440 }, { "epoch": 93.6025, "grad_norm": 8.375, "learning_rate": 2.1389102829999066e-07, "loss": 1.3291, "step": 37441 }, { "epoch": 93.605, "grad_norm": 9.75, "learning_rate": 2.1372449120580896e-07, "loss": 1.3634, "step": 37442 }, { "epoch": 93.6075, "grad_norm": 7.90625, "learning_rate": 2.1355801827005695e-07, "loss": 1.1654, "step": 37443 }, { "epoch": 93.61, "grad_norm": 6.8125, "learning_rate": 2.1339160949382597e-07, "loss": 1.4499, "step": 37444 }, { "epoch": 93.6125, "grad_norm": 7.5625, "learning_rate": 2.132252648782096e-07, "loss": 1.4094, "step": 37445 }, { "epoch": 93.615, "grad_norm": 5.78125, "learning_rate": 2.1305898442429696e-07, "loss": 1.408, "step": 37446 }, { "epoch": 93.6175, "grad_norm": 9.3125, "learning_rate": 2.1289276813317717e-07, "loss": 1.3964, "step": 37447 }, { "epoch": 93.62, "grad_norm": 9.0625, "learning_rate": 2.1272661600594046e-07, "loss": 1.3652, "step": 37448 }, { "epoch": 93.6225, "grad_norm": 10.25, "learning_rate": 2.125605280436771e-07, "loss": 1.5698, "step": 37449 }, { "epoch": 93.625, "grad_norm": 7.6875, "learning_rate": 2.123945042474751e-07, "loss": 1.3183, "step": 37450 }, { "epoch": 93.6275, "grad_norm": 6.375, "learning_rate": 2.1222854461842246e-07, "loss": 1.1865, "step": 37451 }, { "epoch": 93.63, "grad_norm": 7.8125, "learning_rate": 2.1206264915760833e-07, "loss": 1.3073, "step": 37452 }, { "epoch": 93.6325, "grad_norm": 6.90625, "learning_rate": 2.118968178661196e-07, "loss": 1.305, "step": 37453 }, { "epoch": 93.635, "grad_norm": 10.5, "learning_rate": 2.117310507450432e-07, "loss": 1.5078, "step": 37454 }, { "epoch": 93.6375, "grad_norm": 7.40625, "learning_rate": 2.11565347795466e-07, "loss": 1.3775, "step": 37455 }, { "epoch": 93.64, "grad_norm": 6.5625, "learning_rate": 2.1139970901847607e-07, "loss": 1.312, "step": 37456 }, { "epoch": 93.6425, "grad_norm": 7.21875, "learning_rate": 2.1123413441515807e-07, "loss": 1.3138, "step": 37457 }, { "epoch": 93.645, "grad_norm": 8.75, "learning_rate": 2.1106862398659555e-07, "loss": 1.3515, "step": 37458 }, { "epoch": 93.6475, "grad_norm": 6.0, "learning_rate": 2.109031777338766e-07, "loss": 1.3516, "step": 37459 }, { "epoch": 93.65, "grad_norm": 11.1875, "learning_rate": 2.1073779565808471e-07, "loss": 1.4176, "step": 37460 }, { "epoch": 93.6525, "grad_norm": 6.5, "learning_rate": 2.1057247776030466e-07, "loss": 1.4299, "step": 37461 }, { "epoch": 93.655, "grad_norm": 8.5625, "learning_rate": 2.1040722404161772e-07, "loss": 1.3411, "step": 37462 }, { "epoch": 93.6575, "grad_norm": 6.25, "learning_rate": 2.1024203450311086e-07, "loss": 1.3289, "step": 37463 }, { "epoch": 93.66, "grad_norm": 7.625, "learning_rate": 2.1007690914586542e-07, "loss": 1.1898, "step": 37464 }, { "epoch": 93.6625, "grad_norm": 5.625, "learning_rate": 2.0991184797096385e-07, "loss": 1.401, "step": 37465 }, { "epoch": 93.665, "grad_norm": 8.125, "learning_rate": 2.0974685097948756e-07, "loss": 1.305, "step": 37466 }, { "epoch": 93.6675, "grad_norm": 7.125, "learning_rate": 2.0958191817252004e-07, "loss": 1.2914, "step": 37467 }, { "epoch": 93.67, "grad_norm": 10.625, "learning_rate": 2.0941704955114162e-07, "loss": 1.2812, "step": 37468 }, { "epoch": 93.6725, "grad_norm": 6.875, "learning_rate": 2.0925224511643361e-07, "loss": 1.2126, "step": 37469 }, { "epoch": 93.675, "grad_norm": 6.25, "learning_rate": 2.0908750486947405e-07, "loss": 1.2715, "step": 37470 }, { "epoch": 93.6775, "grad_norm": 6.1875, "learning_rate": 2.0892282881134762e-07, "loss": 1.2894, "step": 37471 }, { "epoch": 93.68, "grad_norm": 6.71875, "learning_rate": 2.0875821694313014e-07, "loss": 1.3099, "step": 37472 }, { "epoch": 93.6825, "grad_norm": 11.1875, "learning_rate": 2.0859366926590187e-07, "loss": 1.3899, "step": 37473 }, { "epoch": 93.685, "grad_norm": 9.75, "learning_rate": 2.084291857807408e-07, "loss": 1.4234, "step": 37474 }, { "epoch": 93.6875, "grad_norm": 5.40625, "learning_rate": 2.082647664887283e-07, "loss": 1.2332, "step": 37475 }, { "epoch": 93.69, "grad_norm": 5.46875, "learning_rate": 2.0810041139093906e-07, "loss": 1.2121, "step": 37476 }, { "epoch": 93.6925, "grad_norm": 6.4375, "learning_rate": 2.0793612048845113e-07, "loss": 1.1784, "step": 37477 }, { "epoch": 93.695, "grad_norm": 5.8125, "learning_rate": 2.077718937823414e-07, "loss": 1.4374, "step": 37478 }, { "epoch": 93.6975, "grad_norm": 7.1875, "learning_rate": 2.0760773127368904e-07, "loss": 1.2047, "step": 37479 }, { "epoch": 93.7, "grad_norm": 10.0625, "learning_rate": 2.0744363296356872e-07, "loss": 1.217, "step": 37480 }, { "epoch": 93.7025, "grad_norm": 7.875, "learning_rate": 2.0727959885305515e-07, "loss": 1.2296, "step": 37481 }, { "epoch": 93.705, "grad_norm": 6.6875, "learning_rate": 2.0711562894322523e-07, "loss": 1.3691, "step": 37482 }, { "epoch": 93.7075, "grad_norm": 7.21875, "learning_rate": 2.0695172323515256e-07, "loss": 1.3049, "step": 37483 }, { "epoch": 93.71, "grad_norm": 9.3125, "learning_rate": 2.0678788172991294e-07, "loss": 1.3197, "step": 37484 }, { "epoch": 93.7125, "grad_norm": 7.71875, "learning_rate": 2.0662410442857884e-07, "loss": 1.2704, "step": 37485 }, { "epoch": 93.715, "grad_norm": 8.1875, "learning_rate": 2.0646039133222718e-07, "loss": 1.4163, "step": 37486 }, { "epoch": 93.7175, "grad_norm": 5.0625, "learning_rate": 2.0629674244192933e-07, "loss": 1.1698, "step": 37487 }, { "epoch": 93.72, "grad_norm": 7.6875, "learning_rate": 2.0613315775875665e-07, "loss": 1.3866, "step": 37488 }, { "epoch": 93.7225, "grad_norm": 7.84375, "learning_rate": 2.059696372837827e-07, "loss": 1.3009, "step": 37489 }, { "epoch": 93.725, "grad_norm": 8.1875, "learning_rate": 2.0580618101808114e-07, "loss": 1.4103, "step": 37490 }, { "epoch": 93.7275, "grad_norm": 6.0, "learning_rate": 2.0564278896272217e-07, "loss": 1.3334, "step": 37491 }, { "epoch": 93.73, "grad_norm": 5.71875, "learning_rate": 2.0547946111877714e-07, "loss": 1.2201, "step": 37492 }, { "epoch": 93.7325, "grad_norm": 9.5, "learning_rate": 2.0531619748731636e-07, "loss": 1.4264, "step": 37493 }, { "epoch": 93.735, "grad_norm": 6.21875, "learning_rate": 2.0515299806941112e-07, "loss": 1.363, "step": 37494 }, { "epoch": 93.7375, "grad_norm": 9.9375, "learning_rate": 2.0498986286613064e-07, "loss": 1.3665, "step": 37495 }, { "epoch": 93.74, "grad_norm": 6.9375, "learning_rate": 2.0482679187854405e-07, "loss": 1.3258, "step": 37496 }, { "epoch": 93.7425, "grad_norm": 11.5, "learning_rate": 2.0466378510772266e-07, "loss": 1.3559, "step": 37497 }, { "epoch": 93.745, "grad_norm": 9.1875, "learning_rate": 2.0450084255473124e-07, "loss": 1.2779, "step": 37498 }, { "epoch": 93.7475, "grad_norm": 9.625, "learning_rate": 2.0433796422064222e-07, "loss": 1.3494, "step": 37499 }, { "epoch": 93.75, "grad_norm": 8.6875, "learning_rate": 2.0417515010652032e-07, "loss": 1.2188, "step": 37500 }, { "epoch": 93.7525, "grad_norm": 9.9375, "learning_rate": 2.040124002134347e-07, "loss": 1.1265, "step": 37501 }, { "epoch": 93.755, "grad_norm": 7.9375, "learning_rate": 2.0384971454245227e-07, "loss": 1.2118, "step": 37502 }, { "epoch": 93.7575, "grad_norm": 7.75, "learning_rate": 2.0368709309463776e-07, "loss": 1.3585, "step": 37503 }, { "epoch": 93.76, "grad_norm": 6.65625, "learning_rate": 2.0352453587105914e-07, "loss": 1.3862, "step": 37504 }, { "epoch": 93.7625, "grad_norm": 7.46875, "learning_rate": 2.0336204287278228e-07, "loss": 1.4338, "step": 37505 }, { "epoch": 93.765, "grad_norm": 7.625, "learning_rate": 2.0319961410087185e-07, "loss": 1.2456, "step": 37506 }, { "epoch": 93.7675, "grad_norm": 8.625, "learning_rate": 2.030372495563926e-07, "loss": 1.3622, "step": 37507 }, { "epoch": 93.77, "grad_norm": 8.1875, "learning_rate": 2.0287494924040806e-07, "loss": 1.3457, "step": 37508 }, { "epoch": 93.7725, "grad_norm": 4.65625, "learning_rate": 2.0271271315398412e-07, "loss": 1.2708, "step": 37509 }, { "epoch": 93.775, "grad_norm": 8.5625, "learning_rate": 2.025505412981843e-07, "loss": 1.3367, "step": 37510 }, { "epoch": 93.7775, "grad_norm": 9.5, "learning_rate": 2.023884336740689e-07, "loss": 1.4107, "step": 37511 }, { "epoch": 93.78, "grad_norm": 5.40625, "learning_rate": 2.0222639028270486e-07, "loss": 1.2587, "step": 37512 }, { "epoch": 93.7825, "grad_norm": 9.0, "learning_rate": 2.020644111251513e-07, "loss": 1.382, "step": 37513 }, { "epoch": 93.785, "grad_norm": 8.3125, "learning_rate": 2.0190249620247293e-07, "loss": 1.343, "step": 37514 }, { "epoch": 93.7875, "grad_norm": 7.5, "learning_rate": 2.017406455157278e-07, "loss": 1.1569, "step": 37515 }, { "epoch": 93.79, "grad_norm": 6.875, "learning_rate": 2.0157885906597952e-07, "loss": 1.3087, "step": 37516 }, { "epoch": 93.7925, "grad_norm": 6.5, "learning_rate": 2.0141713685428943e-07, "loss": 1.3074, "step": 37517 }, { "epoch": 93.795, "grad_norm": 10.8125, "learning_rate": 2.0125547888171447e-07, "loss": 1.2816, "step": 37518 }, { "epoch": 93.7975, "grad_norm": 8.0, "learning_rate": 2.0109388514931716e-07, "loss": 1.2419, "step": 37519 }, { "epoch": 93.8, "grad_norm": 4.5, "learning_rate": 2.009323556581566e-07, "loss": 1.3914, "step": 37520 }, { "epoch": 93.8025, "grad_norm": 7.8125, "learning_rate": 2.0077089040929197e-07, "loss": 1.3839, "step": 37521 }, { "epoch": 93.805, "grad_norm": 9.875, "learning_rate": 2.0060948940378023e-07, "loss": 1.3827, "step": 37522 }, { "epoch": 93.8075, "grad_norm": 7.25, "learning_rate": 2.0044815264268047e-07, "loss": 1.4207, "step": 37523 }, { "epoch": 93.81, "grad_norm": 5.28125, "learning_rate": 2.0028688012705077e-07, "loss": 1.4057, "step": 37524 }, { "epoch": 93.8125, "grad_norm": 10.9375, "learning_rate": 2.0012567185794808e-07, "loss": 1.338, "step": 37525 }, { "epoch": 93.815, "grad_norm": 8.0, "learning_rate": 1.9996452783642927e-07, "loss": 1.1599, "step": 37526 }, { "epoch": 93.8175, "grad_norm": 5.53125, "learning_rate": 1.9980344806355134e-07, "loss": 1.1243, "step": 37527 }, { "epoch": 93.82, "grad_norm": 7.5625, "learning_rate": 1.9964243254036897e-07, "loss": 1.2607, "step": 37528 }, { "epoch": 93.8225, "grad_norm": 9.75, "learning_rate": 1.994814812679402e-07, "loss": 1.2967, "step": 37529 }, { "epoch": 93.825, "grad_norm": 10.125, "learning_rate": 1.9932059424731641e-07, "loss": 1.3205, "step": 37530 }, { "epoch": 93.8275, "grad_norm": 7.40625, "learning_rate": 1.991597714795568e-07, "loss": 1.3503, "step": 37531 }, { "epoch": 93.83, "grad_norm": 10.625, "learning_rate": 1.9899901296571268e-07, "loss": 1.3413, "step": 37532 }, { "epoch": 93.8325, "grad_norm": 4.46875, "learning_rate": 1.9883831870683768e-07, "loss": 1.407, "step": 37533 }, { "epoch": 93.835, "grad_norm": 9.75, "learning_rate": 1.986776887039865e-07, "loss": 1.2842, "step": 37534 }, { "epoch": 93.8375, "grad_norm": 8.4375, "learning_rate": 1.985171229582139e-07, "loss": 1.1908, "step": 37535 }, { "epoch": 93.84, "grad_norm": 8.0, "learning_rate": 1.9835662147057012e-07, "loss": 1.3135, "step": 37536 }, { "epoch": 93.8425, "grad_norm": 9.5625, "learning_rate": 1.9819618424210763e-07, "loss": 1.2531, "step": 37537 }, { "epoch": 93.845, "grad_norm": 8.375, "learning_rate": 1.9803581127387893e-07, "loss": 1.4215, "step": 37538 }, { "epoch": 93.8475, "grad_norm": 6.59375, "learning_rate": 1.9787550256693544e-07, "loss": 1.3323, "step": 37539 }, { "epoch": 93.85, "grad_norm": 8.25, "learning_rate": 1.977152581223274e-07, "loss": 1.2912, "step": 37540 }, { "epoch": 93.8525, "grad_norm": 5.59375, "learning_rate": 1.9755507794110619e-07, "loss": 1.3627, "step": 37541 }, { "epoch": 93.855, "grad_norm": 6.125, "learning_rate": 1.9739496202432207e-07, "loss": 1.3425, "step": 37542 }, { "epoch": 93.8575, "grad_norm": 5.9375, "learning_rate": 1.9723491037302312e-07, "loss": 1.3047, "step": 37543 }, { "epoch": 93.86, "grad_norm": 9.25, "learning_rate": 1.970749229882607e-07, "loss": 1.3833, "step": 37544 }, { "epoch": 93.8625, "grad_norm": 7.59375, "learning_rate": 1.9691499987108287e-07, "loss": 1.3684, "step": 37545 }, { "epoch": 93.865, "grad_norm": 10.0625, "learning_rate": 1.9675514102253767e-07, "loss": 1.4282, "step": 37546 }, { "epoch": 93.8675, "grad_norm": 9.4375, "learning_rate": 1.9659534644367428e-07, "loss": 1.263, "step": 37547 }, { "epoch": 93.87, "grad_norm": 7.53125, "learning_rate": 1.964356161355374e-07, "loss": 1.3281, "step": 37548 }, { "epoch": 93.8725, "grad_norm": 9.75, "learning_rate": 1.9627595009917732e-07, "loss": 1.4593, "step": 37549 }, { "epoch": 93.875, "grad_norm": 9.4375, "learning_rate": 1.9611634833564096e-07, "loss": 1.2672, "step": 37550 }, { "epoch": 93.8775, "grad_norm": 6.5, "learning_rate": 1.9595681084597306e-07, "loss": 1.2483, "step": 37551 }, { "epoch": 93.88, "grad_norm": 6.25, "learning_rate": 1.9579733763121943e-07, "loss": 1.4909, "step": 37552 }, { "epoch": 93.8825, "grad_norm": 9.0625, "learning_rate": 1.9563792869242482e-07, "loss": 1.2782, "step": 37553 }, { "epoch": 93.885, "grad_norm": 7.75, "learning_rate": 1.9547858403063835e-07, "loss": 1.4326, "step": 37554 }, { "epoch": 93.8875, "grad_norm": 8.625, "learning_rate": 1.9531930364690033e-07, "loss": 1.2992, "step": 37555 }, { "epoch": 93.89, "grad_norm": 6.84375, "learning_rate": 1.9516008754225658e-07, "loss": 1.2249, "step": 37556 }, { "epoch": 93.8925, "grad_norm": 7.1875, "learning_rate": 1.950009357177518e-07, "loss": 1.3644, "step": 37557 }, { "epoch": 93.895, "grad_norm": 8.5, "learning_rate": 1.9484184817442743e-07, "loss": 1.2543, "step": 37558 }, { "epoch": 93.8975, "grad_norm": 7.75, "learning_rate": 1.9468282491332812e-07, "loss": 1.2895, "step": 37559 }, { "epoch": 93.9, "grad_norm": 10.5, "learning_rate": 1.9452386593549534e-07, "loss": 1.3989, "step": 37560 }, { "epoch": 93.9025, "grad_norm": 8.75, "learning_rate": 1.9436497124197262e-07, "loss": 1.3405, "step": 37561 }, { "epoch": 93.905, "grad_norm": 7.4375, "learning_rate": 1.942061408338003e-07, "loss": 1.4193, "step": 37562 }, { "epoch": 93.9075, "grad_norm": 6.9375, "learning_rate": 1.9404737471201973e-07, "loss": 1.3058, "step": 37563 }, { "epoch": 93.91, "grad_norm": 6.0625, "learning_rate": 1.9388867287767232e-07, "loss": 1.1818, "step": 37564 }, { "epoch": 93.9125, "grad_norm": 6.0625, "learning_rate": 1.9373003533179834e-07, "loss": 1.2831, "step": 37565 }, { "epoch": 93.915, "grad_norm": 7.25, "learning_rate": 1.935714620754392e-07, "loss": 1.2176, "step": 37566 }, { "epoch": 93.9175, "grad_norm": 7.59375, "learning_rate": 1.9341295310963182e-07, "loss": 1.4389, "step": 37567 }, { "epoch": 93.92, "grad_norm": 7.125, "learning_rate": 1.932545084354154e-07, "loss": 1.3756, "step": 37568 }, { "epoch": 93.9225, "grad_norm": 10.0625, "learning_rate": 1.9309612805383126e-07, "loss": 1.2815, "step": 37569 }, { "epoch": 93.925, "grad_norm": 4.34375, "learning_rate": 1.9293781196591755e-07, "loss": 1.3595, "step": 37570 }, { "epoch": 93.9275, "grad_norm": 9.375, "learning_rate": 1.9277956017271005e-07, "loss": 1.3639, "step": 37571 }, { "epoch": 93.93, "grad_norm": 7.40625, "learning_rate": 1.9262137267524684e-07, "loss": 1.4071, "step": 37572 }, { "epoch": 93.9325, "grad_norm": 6.8125, "learning_rate": 1.9246324947456596e-07, "loss": 1.3168, "step": 37573 }, { "epoch": 93.935, "grad_norm": 7.3125, "learning_rate": 1.9230519057170327e-07, "loss": 1.4326, "step": 37574 }, { "epoch": 93.9375, "grad_norm": 7.46875, "learning_rate": 1.921471959676957e-07, "loss": 1.2282, "step": 37575 }, { "epoch": 93.94, "grad_norm": 9.75, "learning_rate": 1.91989265663578e-07, "loss": 1.3866, "step": 37576 }, { "epoch": 93.9425, "grad_norm": 7.8125, "learning_rate": 1.918313996603871e-07, "loss": 1.3982, "step": 37577 }, { "epoch": 93.945, "grad_norm": 9.9375, "learning_rate": 1.9167359795915662e-07, "loss": 1.3873, "step": 37578 }, { "epoch": 93.9475, "grad_norm": 6.53125, "learning_rate": 1.915158605609202e-07, "loss": 1.2692, "step": 37579 }, { "epoch": 93.95, "grad_norm": 6.65625, "learning_rate": 1.9135818746671587e-07, "loss": 1.3304, "step": 37580 }, { "epoch": 93.9525, "grad_norm": 10.0625, "learning_rate": 1.9120057867757392e-07, "loss": 1.4143, "step": 37581 }, { "epoch": 93.955, "grad_norm": 6.59375, "learning_rate": 1.9104303419452796e-07, "loss": 1.3587, "step": 37582 }, { "epoch": 93.9575, "grad_norm": 6.875, "learning_rate": 1.9088555401860942e-07, "loss": 1.2654, "step": 37583 }, { "epoch": 93.96, "grad_norm": 8.1875, "learning_rate": 1.9072813815085523e-07, "loss": 1.5942, "step": 37584 }, { "epoch": 93.9625, "grad_norm": 6.375, "learning_rate": 1.9057078659229455e-07, "loss": 1.3573, "step": 37585 }, { "epoch": 93.965, "grad_norm": 11.0, "learning_rate": 1.904134993439588e-07, "loss": 1.3607, "step": 37586 }, { "epoch": 93.9675, "grad_norm": 6.25, "learning_rate": 1.9025627640687937e-07, "loss": 1.1838, "step": 37587 }, { "epoch": 93.97, "grad_norm": 7.96875, "learning_rate": 1.9009911778208766e-07, "loss": 1.3272, "step": 37588 }, { "epoch": 93.9725, "grad_norm": 6.8125, "learning_rate": 1.8994202347061285e-07, "loss": 1.4927, "step": 37589 }, { "epoch": 93.975, "grad_norm": 8.375, "learning_rate": 1.8978499347348745e-07, "loss": 1.363, "step": 37590 }, { "epoch": 93.9775, "grad_norm": 7.875, "learning_rate": 1.896280277917373e-07, "loss": 1.23, "step": 37591 }, { "epoch": 93.98, "grad_norm": 5.03125, "learning_rate": 1.8947112642639376e-07, "loss": 1.4541, "step": 37592 }, { "epoch": 93.9825, "grad_norm": 9.8125, "learning_rate": 1.8931428937848385e-07, "loss": 1.4075, "step": 37593 }, { "epoch": 93.985, "grad_norm": 9.375, "learning_rate": 1.891575166490378e-07, "loss": 1.3258, "step": 37594 }, { "epoch": 93.9875, "grad_norm": 5.4375, "learning_rate": 1.8900080823908262e-07, "loss": 1.4371, "step": 37595 }, { "epoch": 93.99, "grad_norm": 7.71875, "learning_rate": 1.8884416414964524e-07, "loss": 1.4286, "step": 37596 }, { "epoch": 93.9925, "grad_norm": 9.5, "learning_rate": 1.8868758438175373e-07, "loss": 1.444, "step": 37597 }, { "epoch": 93.995, "grad_norm": 7.9375, "learning_rate": 1.885310689364328e-07, "loss": 1.3284, "step": 37598 }, { "epoch": 93.9975, "grad_norm": 4.71875, "learning_rate": 1.8837461781470946e-07, "loss": 1.2711, "step": 37599 }, { "epoch": 94.0, "grad_norm": 9.8125, "learning_rate": 1.8821823101760949e-07, "loss": 1.2796, "step": 37600 }, { "epoch": 94.0025, "grad_norm": 6.625, "learning_rate": 1.880619085461588e-07, "loss": 1.438, "step": 37601 }, { "epoch": 94.005, "grad_norm": 8.125, "learning_rate": 1.8790565040138098e-07, "loss": 1.4404, "step": 37602 }, { "epoch": 94.0075, "grad_norm": 6.03125, "learning_rate": 1.8774945658430078e-07, "loss": 1.3149, "step": 37603 }, { "epoch": 94.01, "grad_norm": 7.03125, "learning_rate": 1.8759332709594403e-07, "loss": 1.4886, "step": 37604 }, { "epoch": 94.0125, "grad_norm": 7.25, "learning_rate": 1.8743726193733215e-07, "loss": 1.1699, "step": 37605 }, { "epoch": 94.015, "grad_norm": 7.0, "learning_rate": 1.8728126110948763e-07, "loss": 1.2776, "step": 37606 }, { "epoch": 94.0175, "grad_norm": 9.9375, "learning_rate": 1.8712532461343636e-07, "loss": 1.3949, "step": 37607 }, { "epoch": 94.02, "grad_norm": 9.0625, "learning_rate": 1.8696945245019638e-07, "loss": 1.3354, "step": 37608 }, { "epoch": 94.0225, "grad_norm": 7.5625, "learning_rate": 1.8681364462079466e-07, "loss": 1.2822, "step": 37609 }, { "epoch": 94.025, "grad_norm": 10.0625, "learning_rate": 1.8665790112624816e-07, "loss": 1.3761, "step": 37610 }, { "epoch": 94.0275, "grad_norm": 7.46875, "learning_rate": 1.865022219675805e-07, "loss": 1.4608, "step": 37611 }, { "epoch": 94.03, "grad_norm": 6.78125, "learning_rate": 1.8634660714581198e-07, "loss": 1.2441, "step": 37612 }, { "epoch": 94.0325, "grad_norm": 8.125, "learning_rate": 1.861910566619607e-07, "loss": 1.2596, "step": 37613 }, { "epoch": 94.035, "grad_norm": 9.4375, "learning_rate": 1.8603557051704912e-07, "loss": 1.2517, "step": 37614 }, { "epoch": 94.0375, "grad_norm": 6.3125, "learning_rate": 1.8588014871209647e-07, "loss": 1.3999, "step": 37615 }, { "epoch": 94.04, "grad_norm": 8.8125, "learning_rate": 1.857247912481197e-07, "loss": 1.2903, "step": 37616 }, { "epoch": 94.0425, "grad_norm": 9.6875, "learning_rate": 1.855694981261391e-07, "loss": 1.314, "step": 37617 }, { "epoch": 94.045, "grad_norm": 8.625, "learning_rate": 1.8541426934717055e-07, "loss": 1.4447, "step": 37618 }, { "epoch": 94.0475, "grad_norm": 9.0625, "learning_rate": 1.852591049122354e-07, "loss": 1.3359, "step": 37619 }, { "epoch": 94.05, "grad_norm": 9.125, "learning_rate": 1.8510400482234848e-07, "loss": 1.389, "step": 37620 }, { "epoch": 94.0525, "grad_norm": 5.40625, "learning_rate": 1.8494896907852556e-07, "loss": 1.1451, "step": 37621 }, { "epoch": 94.055, "grad_norm": 9.8125, "learning_rate": 1.8479399768178584e-07, "loss": 1.4131, "step": 37622 }, { "epoch": 94.0575, "grad_norm": 6.34375, "learning_rate": 1.84639090633143e-07, "loss": 1.2448, "step": 37623 }, { "epoch": 94.06, "grad_norm": 7.78125, "learning_rate": 1.8448424793361398e-07, "loss": 1.3391, "step": 37624 }, { "epoch": 94.0625, "grad_norm": 6.65625, "learning_rate": 1.8432946958421238e-07, "loss": 1.3314, "step": 37625 }, { "epoch": 94.065, "grad_norm": 10.1875, "learning_rate": 1.8417475558595522e-07, "loss": 1.4752, "step": 37626 }, { "epoch": 94.0675, "grad_norm": 12.1875, "learning_rate": 1.840201059398561e-07, "loss": 1.4454, "step": 37627 }, { "epoch": 94.07, "grad_norm": 9.4375, "learning_rate": 1.838655206469264e-07, "loss": 1.2858, "step": 37628 }, { "epoch": 94.0725, "grad_norm": 9.25, "learning_rate": 1.8371099970818206e-07, "loss": 1.2118, "step": 37629 }, { "epoch": 94.075, "grad_norm": 8.4375, "learning_rate": 1.8355654312463665e-07, "loss": 1.251, "step": 37630 }, { "epoch": 94.0775, "grad_norm": 9.25, "learning_rate": 1.8340215089730163e-07, "loss": 1.2829, "step": 37631 }, { "epoch": 94.08, "grad_norm": 6.25, "learning_rate": 1.8324782302718835e-07, "loss": 1.2161, "step": 37632 }, { "epoch": 94.0825, "grad_norm": 6.625, "learning_rate": 1.8309355951530938e-07, "loss": 1.3607, "step": 37633 }, { "epoch": 94.085, "grad_norm": 7.40625, "learning_rate": 1.8293936036267722e-07, "loss": 1.4449, "step": 37634 }, { "epoch": 94.0875, "grad_norm": 8.6875, "learning_rate": 1.827852255703022e-07, "loss": 1.3805, "step": 37635 }, { "epoch": 94.09, "grad_norm": 8.75, "learning_rate": 1.826311551391924e-07, "loss": 1.4075, "step": 37636 }, { "epoch": 94.0925, "grad_norm": 9.4375, "learning_rate": 1.824771490703614e-07, "loss": 1.4323, "step": 37637 }, { "epoch": 94.095, "grad_norm": 13.4375, "learning_rate": 1.8232320736481624e-07, "loss": 1.5075, "step": 37638 }, { "epoch": 94.0975, "grad_norm": 9.5, "learning_rate": 1.821693300235683e-07, "loss": 1.3195, "step": 37639 }, { "epoch": 94.1, "grad_norm": 6.46875, "learning_rate": 1.8201551704762453e-07, "loss": 1.4393, "step": 37640 }, { "epoch": 94.1025, "grad_norm": 8.0, "learning_rate": 1.818617684379953e-07, "loss": 1.2906, "step": 37641 }, { "epoch": 94.105, "grad_norm": 5.84375, "learning_rate": 1.817080841956864e-07, "loss": 1.2271, "step": 37642 }, { "epoch": 94.1075, "grad_norm": 7.40625, "learning_rate": 1.8155446432170709e-07, "loss": 1.2413, "step": 37643 }, { "epoch": 94.11, "grad_norm": 9.1875, "learning_rate": 1.8140090881706207e-07, "loss": 1.4489, "step": 37644 }, { "epoch": 94.1125, "grad_norm": 6.53125, "learning_rate": 1.8124741768276167e-07, "loss": 1.2323, "step": 37645 }, { "epoch": 94.115, "grad_norm": 7.0625, "learning_rate": 1.810939909198106e-07, "loss": 1.3694, "step": 37646 }, { "epoch": 94.1175, "grad_norm": 13.125, "learning_rate": 1.8094062852921256e-07, "loss": 1.1701, "step": 37647 }, { "epoch": 94.12, "grad_norm": 7.5, "learning_rate": 1.8078733051197561e-07, "loss": 1.4176, "step": 37648 }, { "epoch": 94.1225, "grad_norm": 9.0, "learning_rate": 1.806340968691045e-07, "loss": 1.47, "step": 37649 }, { "epoch": 94.125, "grad_norm": 5.75, "learning_rate": 1.8048092760160286e-07, "loss": 1.2932, "step": 37650 }, { "epoch": 94.1275, "grad_norm": 6.46875, "learning_rate": 1.8032782271047545e-07, "loss": 1.2394, "step": 37651 }, { "epoch": 94.13, "grad_norm": 7.875, "learning_rate": 1.801747821967259e-07, "loss": 1.2096, "step": 37652 }, { "epoch": 94.1325, "grad_norm": 6.96875, "learning_rate": 1.8002180606135677e-07, "loss": 1.455, "step": 37653 }, { "epoch": 94.135, "grad_norm": 6.53125, "learning_rate": 1.798688943053728e-07, "loss": 1.3574, "step": 37654 }, { "epoch": 94.1375, "grad_norm": 9.3125, "learning_rate": 1.797160469297743e-07, "loss": 1.293, "step": 37655 }, { "epoch": 94.14, "grad_norm": 6.34375, "learning_rate": 1.79563263935566e-07, "loss": 1.3287, "step": 37656 }, { "epoch": 94.1425, "grad_norm": 6.8125, "learning_rate": 1.7941054532374714e-07, "loss": 1.1762, "step": 37657 }, { "epoch": 94.145, "grad_norm": 4.8125, "learning_rate": 1.792578910953191e-07, "loss": 1.1625, "step": 37658 }, { "epoch": 94.1475, "grad_norm": 8.8125, "learning_rate": 1.7910530125128333e-07, "loss": 1.3016, "step": 37659 }, { "epoch": 94.15, "grad_norm": 6.3125, "learning_rate": 1.7895277579264015e-07, "loss": 1.1229, "step": 37660 }, { "epoch": 94.1525, "grad_norm": 10.25, "learning_rate": 1.7880031472039095e-07, "loss": 1.3479, "step": 37661 }, { "epoch": 94.155, "grad_norm": 8.0625, "learning_rate": 1.786479180355327e-07, "loss": 1.409, "step": 37662 }, { "epoch": 94.1575, "grad_norm": 8.3125, "learning_rate": 1.7849558573906578e-07, "loss": 1.4424, "step": 37663 }, { "epoch": 94.16, "grad_norm": 8.625, "learning_rate": 1.7834331783198933e-07, "loss": 1.392, "step": 37664 }, { "epoch": 94.1625, "grad_norm": 9.5625, "learning_rate": 1.7819111431530034e-07, "loss": 1.2727, "step": 37665 }, { "epoch": 94.165, "grad_norm": 10.625, "learning_rate": 1.7803897518999691e-07, "loss": 1.2574, "step": 37666 }, { "epoch": 94.1675, "grad_norm": 6.09375, "learning_rate": 1.7788690045707824e-07, "loss": 1.25, "step": 37667 }, { "epoch": 94.17, "grad_norm": 5.96875, "learning_rate": 1.7773489011753908e-07, "loss": 1.3633, "step": 37668 }, { "epoch": 94.1725, "grad_norm": 6.1875, "learning_rate": 1.7758294417237755e-07, "loss": 1.2959, "step": 37669 }, { "epoch": 94.175, "grad_norm": 5.875, "learning_rate": 1.7743106262258836e-07, "loss": 1.1471, "step": 37670 }, { "epoch": 94.1775, "grad_norm": 13.75, "learning_rate": 1.7727924546916964e-07, "loss": 1.3743, "step": 37671 }, { "epoch": 94.18, "grad_norm": 6.15625, "learning_rate": 1.7712749271311392e-07, "loss": 1.3339, "step": 37672 }, { "epoch": 94.1825, "grad_norm": 7.71875, "learning_rate": 1.7697580435541705e-07, "loss": 1.3174, "step": 37673 }, { "epoch": 94.185, "grad_norm": 8.5, "learning_rate": 1.7682418039707384e-07, "loss": 1.2801, "step": 37674 }, { "epoch": 94.1875, "grad_norm": 7.3125, "learning_rate": 1.7667262083907789e-07, "loss": 1.3907, "step": 37675 }, { "epoch": 94.19, "grad_norm": 5.125, "learning_rate": 1.7652112568242398e-07, "loss": 1.2421, "step": 37676 }, { "epoch": 94.1925, "grad_norm": 12.125, "learning_rate": 1.7636969492810463e-07, "loss": 1.3152, "step": 37677 }, { "epoch": 94.195, "grad_norm": 6.53125, "learning_rate": 1.762183285771113e-07, "loss": 1.1786, "step": 37678 }, { "epoch": 94.1975, "grad_norm": 9.3125, "learning_rate": 1.7606702663043762e-07, "loss": 1.4315, "step": 37679 }, { "epoch": 94.2, "grad_norm": 7.71875, "learning_rate": 1.7591578908907724e-07, "loss": 1.398, "step": 37680 }, { "epoch": 94.2025, "grad_norm": 8.0625, "learning_rate": 1.7576461595401717e-07, "loss": 1.2451, "step": 37681 }, { "epoch": 94.205, "grad_norm": 7.78125, "learning_rate": 1.7561350722625326e-07, "loss": 1.2243, "step": 37682 }, { "epoch": 94.2075, "grad_norm": 8.375, "learning_rate": 1.7546246290677248e-07, "loss": 1.2354, "step": 37683 }, { "epoch": 94.21, "grad_norm": 9.6875, "learning_rate": 1.753114829965674e-07, "loss": 1.2145, "step": 37684 }, { "epoch": 94.2125, "grad_norm": 8.125, "learning_rate": 1.7516056749662726e-07, "loss": 1.2056, "step": 37685 }, { "epoch": 94.215, "grad_norm": 9.25, "learning_rate": 1.7500971640794118e-07, "loss": 1.4694, "step": 37686 }, { "epoch": 94.2175, "grad_norm": 5.9375, "learning_rate": 1.7485892973149842e-07, "loss": 1.3539, "step": 37687 }, { "epoch": 94.22, "grad_norm": 5.46875, "learning_rate": 1.74708207468286e-07, "loss": 1.4371, "step": 37688 }, { "epoch": 94.2225, "grad_norm": 5.34375, "learning_rate": 1.7455754961929416e-07, "loss": 1.3094, "step": 37689 }, { "epoch": 94.225, "grad_norm": 6.28125, "learning_rate": 1.7440695618551108e-07, "loss": 1.2799, "step": 37690 }, { "epoch": 94.2275, "grad_norm": 8.875, "learning_rate": 1.7425642716792146e-07, "loss": 1.3596, "step": 37691 }, { "epoch": 94.23, "grad_norm": 8.875, "learning_rate": 1.7410596256751343e-07, "loss": 1.3058, "step": 37692 }, { "epoch": 94.2325, "grad_norm": 7.84375, "learning_rate": 1.73955562385274e-07, "loss": 1.3811, "step": 37693 }, { "epoch": 94.235, "grad_norm": 5.96875, "learning_rate": 1.738052266221879e-07, "loss": 1.5056, "step": 37694 }, { "epoch": 94.2375, "grad_norm": 7.84375, "learning_rate": 1.7365495527924326e-07, "loss": 1.3834, "step": 37695 }, { "epoch": 94.24, "grad_norm": 7.375, "learning_rate": 1.735047483574215e-07, "loss": 1.4165, "step": 37696 }, { "epoch": 94.2425, "grad_norm": 6.25, "learning_rate": 1.733546058577096e-07, "loss": 1.2833, "step": 37697 }, { "epoch": 94.245, "grad_norm": 7.59375, "learning_rate": 1.7320452778109342e-07, "loss": 1.2168, "step": 37698 }, { "epoch": 94.2475, "grad_norm": 11.0, "learning_rate": 1.7305451412855445e-07, "loss": 1.3821, "step": 37699 }, { "epoch": 94.25, "grad_norm": 6.9375, "learning_rate": 1.7290456490107522e-07, "loss": 1.3199, "step": 37700 }, { "epoch": 94.2525, "grad_norm": 8.3125, "learning_rate": 1.727546800996427e-07, "loss": 1.3403, "step": 37701 }, { "epoch": 94.255, "grad_norm": 6.71875, "learning_rate": 1.72604859725235e-07, "loss": 1.1865, "step": 37702 }, { "epoch": 94.2575, "grad_norm": 4.34375, "learning_rate": 1.72455103778838e-07, "loss": 1.3002, "step": 37703 }, { "epoch": 94.26, "grad_norm": 6.65625, "learning_rate": 1.7230541226143205e-07, "loss": 1.281, "step": 37704 }, { "epoch": 94.2625, "grad_norm": 7.0, "learning_rate": 1.7215578517399855e-07, "loss": 1.5838, "step": 37705 }, { "epoch": 94.265, "grad_norm": 8.125, "learning_rate": 1.7200622251751786e-07, "loss": 1.2489, "step": 37706 }, { "epoch": 94.2675, "grad_norm": 6.40625, "learning_rate": 1.718567242929714e-07, "loss": 1.154, "step": 37707 }, { "epoch": 94.27, "grad_norm": 6.78125, "learning_rate": 1.7170729050133838e-07, "loss": 1.2722, "step": 37708 }, { "epoch": 94.2725, "grad_norm": 10.9375, "learning_rate": 1.7155792114360026e-07, "loss": 1.3271, "step": 37709 }, { "epoch": 94.275, "grad_norm": 10.25, "learning_rate": 1.7140861622073512e-07, "loss": 1.2615, "step": 37710 }, { "epoch": 94.2775, "grad_norm": 8.625, "learning_rate": 1.7125937573372108e-07, "loss": 1.4044, "step": 37711 }, { "epoch": 94.28, "grad_norm": 8.3125, "learning_rate": 1.7111019968353625e-07, "loss": 1.4627, "step": 37712 }, { "epoch": 94.2825, "grad_norm": 5.09375, "learning_rate": 1.7096108807116095e-07, "loss": 1.3147, "step": 37713 }, { "epoch": 94.285, "grad_norm": 6.84375, "learning_rate": 1.7081204089757108e-07, "loss": 1.3771, "step": 37714 }, { "epoch": 94.2875, "grad_norm": 9.3125, "learning_rate": 1.7066305816374361e-07, "loss": 1.4019, "step": 37715 }, { "epoch": 94.29, "grad_norm": 5.9375, "learning_rate": 1.7051413987065668e-07, "loss": 1.1262, "step": 37716 }, { "epoch": 94.2925, "grad_norm": 9.0, "learning_rate": 1.7036528601928504e-07, "loss": 1.3211, "step": 37717 }, { "epoch": 94.295, "grad_norm": 9.75, "learning_rate": 1.702164966106057e-07, "loss": 1.3803, "step": 37718 }, { "epoch": 94.2975, "grad_norm": 7.5625, "learning_rate": 1.700677716455934e-07, "loss": 1.3155, "step": 37719 }, { "epoch": 94.3, "grad_norm": 7.15625, "learning_rate": 1.699191111252241e-07, "loss": 1.1899, "step": 37720 }, { "epoch": 94.3025, "grad_norm": 10.6875, "learning_rate": 1.6977051505047136e-07, "loss": 1.38, "step": 37721 }, { "epoch": 94.305, "grad_norm": 9.3125, "learning_rate": 1.6962198342230895e-07, "loss": 1.4709, "step": 37722 }, { "epoch": 94.3075, "grad_norm": 7.5, "learning_rate": 1.694735162417116e-07, "loss": 1.3145, "step": 37723 }, { "epoch": 94.31, "grad_norm": 8.9375, "learning_rate": 1.6932511350965298e-07, "loss": 1.3684, "step": 37724 }, { "epoch": 94.3125, "grad_norm": 6.34375, "learning_rate": 1.6917677522710564e-07, "loss": 1.3543, "step": 37725 }, { "epoch": 94.315, "grad_norm": 10.25, "learning_rate": 1.6902850139504213e-07, "loss": 1.2805, "step": 37726 }, { "epoch": 94.3175, "grad_norm": 5.9375, "learning_rate": 1.688802920144328e-07, "loss": 1.3427, "step": 37727 }, { "epoch": 94.32, "grad_norm": 6.4375, "learning_rate": 1.687321470862524e-07, "loss": 1.2973, "step": 37728 }, { "epoch": 94.3225, "grad_norm": 11.6875, "learning_rate": 1.685840666114702e-07, "loss": 1.2907, "step": 37729 }, { "epoch": 94.325, "grad_norm": 9.3125, "learning_rate": 1.6843605059105761e-07, "loss": 1.4293, "step": 37730 }, { "epoch": 94.3275, "grad_norm": 5.46875, "learning_rate": 1.6828809902598498e-07, "loss": 1.4814, "step": 37731 }, { "epoch": 94.33, "grad_norm": 6.78125, "learning_rate": 1.6814021191722042e-07, "loss": 1.2298, "step": 37732 }, { "epoch": 94.3325, "grad_norm": 8.125, "learning_rate": 1.6799238926573758e-07, "loss": 1.2969, "step": 37733 }, { "epoch": 94.335, "grad_norm": 11.1875, "learning_rate": 1.6784463107250238e-07, "loss": 1.4727, "step": 37734 }, { "epoch": 94.3375, "grad_norm": 5.96875, "learning_rate": 1.6769693733848403e-07, "loss": 1.3716, "step": 37735 }, { "epoch": 94.34, "grad_norm": 10.5, "learning_rate": 1.6754930806465063e-07, "loss": 1.4093, "step": 37736 }, { "epoch": 94.3425, "grad_norm": 6.96875, "learning_rate": 1.6740174325197144e-07, "loss": 1.3091, "step": 37737 }, { "epoch": 94.345, "grad_norm": 6.34375, "learning_rate": 1.6725424290141236e-07, "loss": 1.4609, "step": 37738 }, { "epoch": 94.3475, "grad_norm": 7.03125, "learning_rate": 1.6710680701394145e-07, "loss": 1.3633, "step": 37739 }, { "epoch": 94.35, "grad_norm": 5.5, "learning_rate": 1.6695943559052463e-07, "loss": 1.1855, "step": 37740 }, { "epoch": 94.3525, "grad_norm": 8.375, "learning_rate": 1.668121286321278e-07, "loss": 1.2936, "step": 37741 }, { "epoch": 94.355, "grad_norm": 8.125, "learning_rate": 1.6666488613971687e-07, "loss": 1.3042, "step": 37742 }, { "epoch": 94.3575, "grad_norm": 7.59375, "learning_rate": 1.665177081142577e-07, "loss": 1.3957, "step": 37743 }, { "epoch": 94.36, "grad_norm": 7.46875, "learning_rate": 1.6637059455671623e-07, "loss": 1.3719, "step": 37744 }, { "epoch": 94.3625, "grad_norm": 9.5, "learning_rate": 1.6622354546805496e-07, "loss": 1.5131, "step": 37745 }, { "epoch": 94.365, "grad_norm": 6.21875, "learning_rate": 1.6607656084923762e-07, "loss": 1.3169, "step": 37746 }, { "epoch": 94.3675, "grad_norm": 8.8125, "learning_rate": 1.6592964070122898e-07, "loss": 1.3728, "step": 37747 }, { "epoch": 94.37, "grad_norm": 5.9375, "learning_rate": 1.6578278502499378e-07, "loss": 1.3542, "step": 37748 }, { "epoch": 94.3725, "grad_norm": 7.25, "learning_rate": 1.6563599382149243e-07, "loss": 1.3174, "step": 37749 }, { "epoch": 94.375, "grad_norm": 8.8125, "learning_rate": 1.6548926709168634e-07, "loss": 1.2233, "step": 37750 }, { "epoch": 94.3775, "grad_norm": 9.6875, "learning_rate": 1.653426048365403e-07, "loss": 1.3897, "step": 37751 }, { "epoch": 94.38, "grad_norm": 7.78125, "learning_rate": 1.6519600705701465e-07, "loss": 1.2806, "step": 37752 }, { "epoch": 94.3825, "grad_norm": 6.4375, "learning_rate": 1.6504947375407088e-07, "loss": 1.2975, "step": 37753 }, { "epoch": 94.385, "grad_norm": 11.3125, "learning_rate": 1.6490300492866706e-07, "loss": 1.2978, "step": 37754 }, { "epoch": 94.3875, "grad_norm": 13.8125, "learning_rate": 1.64756600581768e-07, "loss": 1.4242, "step": 37755 }, { "epoch": 94.39, "grad_norm": 10.5, "learning_rate": 1.6461026071433074e-07, "loss": 1.2118, "step": 37756 }, { "epoch": 94.3925, "grad_norm": 6.59375, "learning_rate": 1.6446398532731334e-07, "loss": 1.2281, "step": 37757 }, { "epoch": 94.395, "grad_norm": 6.90625, "learning_rate": 1.6431777442167728e-07, "loss": 1.2701, "step": 37758 }, { "epoch": 94.3975, "grad_norm": 10.9375, "learning_rate": 1.641716279983807e-07, "loss": 1.6432, "step": 37759 }, { "epoch": 94.4, "grad_norm": 9.5625, "learning_rate": 1.6402554605838173e-07, "loss": 1.3267, "step": 37760 }, { "epoch": 94.4025, "grad_norm": 8.375, "learning_rate": 1.6387952860263622e-07, "loss": 1.3737, "step": 37761 }, { "epoch": 94.405, "grad_norm": 10.1875, "learning_rate": 1.6373357563210347e-07, "loss": 1.233, "step": 37762 }, { "epoch": 94.4075, "grad_norm": 6.71875, "learning_rate": 1.6358768714774044e-07, "loss": 1.4519, "step": 37763 }, { "epoch": 94.41, "grad_norm": 6.15625, "learning_rate": 1.6344186315050303e-07, "loss": 1.4339, "step": 37764 }, { "epoch": 94.4125, "grad_norm": 7.25, "learning_rate": 1.6329610364134607e-07, "loss": 1.3273, "step": 37765 }, { "epoch": 94.415, "grad_norm": 6.8125, "learning_rate": 1.6315040862122767e-07, "loss": 1.3478, "step": 37766 }, { "epoch": 94.4175, "grad_norm": 6.8125, "learning_rate": 1.630047780911004e-07, "loss": 1.4666, "step": 37767 }, { "epoch": 94.42, "grad_norm": 7.125, "learning_rate": 1.6285921205192012e-07, "loss": 1.2618, "step": 37768 }, { "epoch": 94.4225, "grad_norm": 7.34375, "learning_rate": 1.6271371050464057e-07, "loss": 1.2992, "step": 37769 }, { "epoch": 94.425, "grad_norm": 8.25, "learning_rate": 1.6256827345021764e-07, "loss": 1.5531, "step": 37770 }, { "epoch": 94.4275, "grad_norm": 9.125, "learning_rate": 1.6242290088960388e-07, "loss": 1.3757, "step": 37771 }, { "epoch": 94.43, "grad_norm": 15.0, "learning_rate": 1.6227759282375079e-07, "loss": 1.3627, "step": 37772 }, { "epoch": 94.4325, "grad_norm": 8.125, "learning_rate": 1.6213234925361088e-07, "loss": 1.4003, "step": 37773 }, { "epoch": 94.435, "grad_norm": 7.75, "learning_rate": 1.619871701801401e-07, "loss": 1.2068, "step": 37774 }, { "epoch": 94.4375, "grad_norm": 8.9375, "learning_rate": 1.6184205560428655e-07, "loss": 1.2404, "step": 37775 }, { "epoch": 94.44, "grad_norm": 8.25, "learning_rate": 1.6169700552700284e-07, "loss": 1.3519, "step": 37776 }, { "epoch": 94.4425, "grad_norm": 7.625, "learning_rate": 1.615520199492393e-07, "loss": 1.4314, "step": 37777 }, { "epoch": 94.445, "grad_norm": 5.75, "learning_rate": 1.614070988719485e-07, "loss": 1.2706, "step": 37778 }, { "epoch": 94.4475, "grad_norm": 6.96875, "learning_rate": 1.6126224229607858e-07, "loss": 1.3338, "step": 37779 }, { "epoch": 94.45, "grad_norm": 5.84375, "learning_rate": 1.6111745022257873e-07, "loss": 1.2685, "step": 37780 }, { "epoch": 94.4525, "grad_norm": 8.1875, "learning_rate": 1.6097272265239938e-07, "loss": 1.3348, "step": 37781 }, { "epoch": 94.455, "grad_norm": 7.25, "learning_rate": 1.6082805958648862e-07, "loss": 1.3249, "step": 37782 }, { "epoch": 94.4575, "grad_norm": 6.34375, "learning_rate": 1.606834610257968e-07, "loss": 1.2388, "step": 37783 }, { "epoch": 94.46, "grad_norm": 7.0, "learning_rate": 1.6053892697126984e-07, "loss": 1.3592, "step": 37784 }, { "epoch": 94.4625, "grad_norm": 8.5, "learning_rate": 1.6039445742385583e-07, "loss": 1.3108, "step": 37785 }, { "epoch": 94.465, "grad_norm": 9.75, "learning_rate": 1.6025005238450298e-07, "loss": 1.3686, "step": 37786 }, { "epoch": 94.4675, "grad_norm": 7.46875, "learning_rate": 1.601057118541549e-07, "loss": 1.2931, "step": 37787 }, { "epoch": 94.47, "grad_norm": 6.65625, "learning_rate": 1.599614358337609e-07, "loss": 1.2725, "step": 37788 }, { "epoch": 94.4725, "grad_norm": 5.96875, "learning_rate": 1.5981722432426682e-07, "loss": 1.2318, "step": 37789 }, { "epoch": 94.475, "grad_norm": 7.375, "learning_rate": 1.596730773266175e-07, "loss": 1.2707, "step": 37790 }, { "epoch": 94.4775, "grad_norm": 7.8125, "learning_rate": 1.5952899484175554e-07, "loss": 1.3229, "step": 37791 }, { "epoch": 94.48, "grad_norm": 7.0625, "learning_rate": 1.5938497687062905e-07, "loss": 1.2562, "step": 37792 }, { "epoch": 94.4825, "grad_norm": 6.1875, "learning_rate": 1.592410234141817e-07, "loss": 1.3754, "step": 37793 }, { "epoch": 94.485, "grad_norm": 5.1875, "learning_rate": 1.5909713447335496e-07, "loss": 1.3458, "step": 37794 }, { "epoch": 94.4875, "grad_norm": 7.125, "learning_rate": 1.5895331004909365e-07, "loss": 1.2299, "step": 37795 }, { "epoch": 94.49, "grad_norm": 6.5, "learning_rate": 1.5880955014234144e-07, "loss": 1.2199, "step": 37796 }, { "epoch": 94.4925, "grad_norm": 8.9375, "learning_rate": 1.5866585475403985e-07, "loss": 1.3339, "step": 37797 }, { "epoch": 94.495, "grad_norm": 8.6875, "learning_rate": 1.585222238851314e-07, "loss": 1.3289, "step": 37798 }, { "epoch": 94.4975, "grad_norm": 6.3125, "learning_rate": 1.5837865753655645e-07, "loss": 1.3301, "step": 37799 }, { "epoch": 94.5, "grad_norm": 5.21875, "learning_rate": 1.5823515570925763e-07, "loss": 1.3014, "step": 37800 }, { "epoch": 94.5025, "grad_norm": 9.0, "learning_rate": 1.5809171840417638e-07, "loss": 1.3463, "step": 37801 }, { "epoch": 94.505, "grad_norm": 8.625, "learning_rate": 1.5794834562224969e-07, "loss": 1.3214, "step": 37802 }, { "epoch": 94.5075, "grad_norm": 7.5625, "learning_rate": 1.5780503736442132e-07, "loss": 1.3389, "step": 37803 }, { "epoch": 94.51, "grad_norm": 11.9375, "learning_rate": 1.5766179363162936e-07, "loss": 1.2965, "step": 37804 }, { "epoch": 94.5125, "grad_norm": 6.90625, "learning_rate": 1.5751861442481198e-07, "loss": 1.3322, "step": 37805 }, { "epoch": 94.515, "grad_norm": 7.3125, "learning_rate": 1.5737549974490952e-07, "loss": 1.2529, "step": 37806 }, { "epoch": 94.5175, "grad_norm": 7.53125, "learning_rate": 1.5723244959285787e-07, "loss": 1.2523, "step": 37807 }, { "epoch": 94.52, "grad_norm": 5.59375, "learning_rate": 1.5708946396959856e-07, "loss": 1.3479, "step": 37808 }, { "epoch": 94.5225, "grad_norm": 8.1875, "learning_rate": 1.5694654287606525e-07, "loss": 1.6396, "step": 37809 }, { "epoch": 94.525, "grad_norm": 7.6875, "learning_rate": 1.5680368631319721e-07, "loss": 1.253, "step": 37810 }, { "epoch": 94.5275, "grad_norm": 7.90625, "learning_rate": 1.5666089428193033e-07, "loss": 1.3441, "step": 37811 }, { "epoch": 94.53, "grad_norm": 9.6875, "learning_rate": 1.5651816678320054e-07, "loss": 1.3156, "step": 37812 }, { "epoch": 94.5325, "grad_norm": 7.21875, "learning_rate": 1.5637550381794377e-07, "loss": 1.3562, "step": 37813 }, { "epoch": 94.535, "grad_norm": 7.28125, "learning_rate": 1.562329053870948e-07, "loss": 1.2391, "step": 37814 }, { "epoch": 94.5375, "grad_norm": 6.09375, "learning_rate": 1.5609037149158956e-07, "loss": 1.2874, "step": 37815 }, { "epoch": 94.54, "grad_norm": 7.0625, "learning_rate": 1.5594790213236288e-07, "loss": 1.2214, "step": 37816 }, { "epoch": 94.5425, "grad_norm": 7.78125, "learning_rate": 1.5580549731034622e-07, "loss": 1.2358, "step": 37817 }, { "epoch": 94.545, "grad_norm": 6.53125, "learning_rate": 1.5566315702647438e-07, "loss": 1.3901, "step": 37818 }, { "epoch": 94.5475, "grad_norm": 7.5, "learning_rate": 1.5552088128168218e-07, "loss": 1.3886, "step": 37819 }, { "epoch": 94.55, "grad_norm": 10.9375, "learning_rate": 1.5537867007690111e-07, "loss": 1.2677, "step": 37820 }, { "epoch": 94.5525, "grad_norm": 7.4375, "learning_rate": 1.5523652341306262e-07, "loss": 1.4157, "step": 37821 }, { "epoch": 94.555, "grad_norm": 7.65625, "learning_rate": 1.550944412910993e-07, "loss": 1.4504, "step": 37822 }, { "epoch": 94.5575, "grad_norm": 20.5, "learning_rate": 1.5495242371194485e-07, "loss": 1.3577, "step": 37823 }, { "epoch": 94.56, "grad_norm": 9.0, "learning_rate": 1.5481047067652744e-07, "loss": 1.3398, "step": 37824 }, { "epoch": 94.5625, "grad_norm": 9.0, "learning_rate": 1.546685821857774e-07, "loss": 1.3325, "step": 37825 }, { "epoch": 94.565, "grad_norm": 11.0625, "learning_rate": 1.5452675824062736e-07, "loss": 1.3129, "step": 37826 }, { "epoch": 94.5675, "grad_norm": 8.3125, "learning_rate": 1.5438499884200542e-07, "loss": 1.3598, "step": 37827 }, { "epoch": 94.57, "grad_norm": 10.6875, "learning_rate": 1.5424330399084196e-07, "loss": 1.2975, "step": 37828 }, { "epoch": 94.5725, "grad_norm": 6.125, "learning_rate": 1.5410167368806404e-07, "loss": 1.3288, "step": 37829 }, { "epoch": 94.575, "grad_norm": 7.03125, "learning_rate": 1.5396010793460313e-07, "loss": 1.3867, "step": 37830 }, { "epoch": 94.5775, "grad_norm": 8.0625, "learning_rate": 1.5381860673138517e-07, "loss": 1.2915, "step": 37831 }, { "epoch": 94.58, "grad_norm": 7.90625, "learning_rate": 1.5367717007933826e-07, "loss": 1.3803, "step": 37832 }, { "epoch": 94.5825, "grad_norm": 7.1875, "learning_rate": 1.5353579797938945e-07, "loss": 1.2086, "step": 37833 }, { "epoch": 94.585, "grad_norm": 7.5625, "learning_rate": 1.533944904324669e-07, "loss": 1.329, "step": 37834 }, { "epoch": 94.5875, "grad_norm": 10.1875, "learning_rate": 1.5325324743949543e-07, "loss": 1.5627, "step": 37835 }, { "epoch": 94.59, "grad_norm": 8.4375, "learning_rate": 1.5311206900140095e-07, "loss": 1.4139, "step": 37836 }, { "epoch": 94.5925, "grad_norm": 5.5625, "learning_rate": 1.5297095511911054e-07, "loss": 1.4358, "step": 37837 }, { "epoch": 94.595, "grad_norm": 8.9375, "learning_rate": 1.5282990579354896e-07, "loss": 1.365, "step": 37838 }, { "epoch": 94.5975, "grad_norm": 11.4375, "learning_rate": 1.5268892102563993e-07, "loss": 1.3974, "step": 37839 }, { "epoch": 94.6, "grad_norm": 6.3125, "learning_rate": 1.5254800081630828e-07, "loss": 1.2857, "step": 37840 }, { "epoch": 94.6025, "grad_norm": 8.8125, "learning_rate": 1.524071451664777e-07, "loss": 1.3124, "step": 37841 }, { "epoch": 94.605, "grad_norm": 8.375, "learning_rate": 1.522663540770719e-07, "loss": 1.0991, "step": 37842 }, { "epoch": 94.6075, "grad_norm": 6.1875, "learning_rate": 1.521256275490135e-07, "loss": 1.4615, "step": 37843 }, { "epoch": 94.61, "grad_norm": 8.375, "learning_rate": 1.5198496558322506e-07, "loss": 1.2883, "step": 37844 }, { "epoch": 94.6125, "grad_norm": 6.125, "learning_rate": 1.518443681806303e-07, "loss": 1.3549, "step": 37845 }, { "epoch": 94.615, "grad_norm": 6.53125, "learning_rate": 1.5170383534214962e-07, "loss": 1.2437, "step": 37846 }, { "epoch": 94.6175, "grad_norm": 8.0, "learning_rate": 1.5156336706870335e-07, "loss": 1.2877, "step": 37847 }, { "epoch": 94.62, "grad_norm": 7.5625, "learning_rate": 1.514229633612141e-07, "loss": 1.401, "step": 37848 }, { "epoch": 94.6225, "grad_norm": 7.65625, "learning_rate": 1.5128262422060224e-07, "loss": 1.3611, "step": 37849 }, { "epoch": 94.625, "grad_norm": 6.5, "learning_rate": 1.5114234964778707e-07, "loss": 1.4802, "step": 37850 }, { "epoch": 94.6275, "grad_norm": 7.3125, "learning_rate": 1.5100213964368892e-07, "loss": 1.3412, "step": 37851 }, { "epoch": 94.63, "grad_norm": 9.0, "learning_rate": 1.5086199420922486e-07, "loss": 1.4177, "step": 37852 }, { "epoch": 94.6325, "grad_norm": 5.8125, "learning_rate": 1.5072191334531749e-07, "loss": 1.2168, "step": 37853 }, { "epoch": 94.635, "grad_norm": 10.125, "learning_rate": 1.5058189705288274e-07, "loss": 1.2674, "step": 37854 }, { "epoch": 94.6375, "grad_norm": 5.03125, "learning_rate": 1.5044194533283763e-07, "loss": 1.2747, "step": 37855 }, { "epoch": 94.64, "grad_norm": 5.21875, "learning_rate": 1.5030205818610255e-07, "loss": 1.3533, "step": 37856 }, { "epoch": 94.6425, "grad_norm": 8.125, "learning_rate": 1.5016223561359233e-07, "loss": 1.2959, "step": 37857 }, { "epoch": 94.645, "grad_norm": 6.90625, "learning_rate": 1.5002247761622401e-07, "loss": 1.2039, "step": 37858 }, { "epoch": 94.6475, "grad_norm": 8.5, "learning_rate": 1.4988278419491465e-07, "loss": 1.2366, "step": 37859 }, { "epoch": 94.65, "grad_norm": 6.75, "learning_rate": 1.4974315535058016e-07, "loss": 1.3069, "step": 37860 }, { "epoch": 94.6525, "grad_norm": 8.25, "learning_rate": 1.4960359108413424e-07, "loss": 1.3566, "step": 37861 }, { "epoch": 94.655, "grad_norm": 12.8125, "learning_rate": 1.4946409139649287e-07, "loss": 1.5339, "step": 37862 }, { "epoch": 94.6575, "grad_norm": 7.5625, "learning_rate": 1.4932465628857086e-07, "loss": 1.2669, "step": 37863 }, { "epoch": 94.66, "grad_norm": 8.3125, "learning_rate": 1.4918528576128298e-07, "loss": 1.3218, "step": 37864 }, { "epoch": 94.6625, "grad_norm": 9.0, "learning_rate": 1.4904597981554193e-07, "loss": 1.2587, "step": 37865 }, { "epoch": 94.665, "grad_norm": 9.4375, "learning_rate": 1.4890673845226133e-07, "loss": 1.4141, "step": 37866 }, { "epoch": 94.6675, "grad_norm": 7.3125, "learning_rate": 1.4876756167235273e-07, "loss": 1.2798, "step": 37867 }, { "epoch": 94.67, "grad_norm": 6.28125, "learning_rate": 1.4862844947673093e-07, "loss": 1.2662, "step": 37868 }, { "epoch": 94.6725, "grad_norm": 6.25, "learning_rate": 1.4848940186630745e-07, "loss": 1.1417, "step": 37869 }, { "epoch": 94.675, "grad_norm": 8.5, "learning_rate": 1.4835041884199263e-07, "loss": 1.2478, "step": 37870 }, { "epoch": 94.6775, "grad_norm": 7.03125, "learning_rate": 1.482115004046969e-07, "loss": 1.2064, "step": 37871 }, { "epoch": 94.68, "grad_norm": 5.84375, "learning_rate": 1.4807264655533282e-07, "loss": 1.3168, "step": 37872 }, { "epoch": 94.6825, "grad_norm": 7.6875, "learning_rate": 1.4793385729481081e-07, "loss": 1.4174, "step": 37873 }, { "epoch": 94.685, "grad_norm": 10.0, "learning_rate": 1.477951326240412e-07, "loss": 1.2209, "step": 37874 }, { "epoch": 94.6875, "grad_norm": 6.71875, "learning_rate": 1.4765647254393113e-07, "loss": 1.5421, "step": 37875 }, { "epoch": 94.69, "grad_norm": 10.875, "learning_rate": 1.475178770553909e-07, "loss": 1.3993, "step": 37876 }, { "epoch": 94.6925, "grad_norm": 7.125, "learning_rate": 1.473793461593287e-07, "loss": 1.2222, "step": 37877 }, { "epoch": 94.695, "grad_norm": 7.8125, "learning_rate": 1.4724087985665492e-07, "loss": 1.3394, "step": 37878 }, { "epoch": 94.6975, "grad_norm": 4.59375, "learning_rate": 1.471024781482744e-07, "loss": 1.2714, "step": 37879 }, { "epoch": 94.7, "grad_norm": 5.4375, "learning_rate": 1.469641410350964e-07, "loss": 1.2701, "step": 37880 }, { "epoch": 94.7025, "grad_norm": 7.0, "learning_rate": 1.4682586851802681e-07, "loss": 1.4158, "step": 37881 }, { "epoch": 94.705, "grad_norm": 7.75, "learning_rate": 1.4668766059797168e-07, "loss": 1.2685, "step": 37882 }, { "epoch": 94.7075, "grad_norm": 7.46875, "learning_rate": 1.465495172758391e-07, "loss": 1.3171, "step": 37883 }, { "epoch": 94.71, "grad_norm": 9.25, "learning_rate": 1.4641143855253283e-07, "loss": 1.2991, "step": 37884 }, { "epoch": 94.7125, "grad_norm": 7.40625, "learning_rate": 1.4627342442895986e-07, "loss": 1.2335, "step": 37885 }, { "epoch": 94.715, "grad_norm": 8.0, "learning_rate": 1.4613547490602287e-07, "loss": 1.3732, "step": 37886 }, { "epoch": 94.7175, "grad_norm": 6.375, "learning_rate": 1.4599758998462776e-07, "loss": 1.3093, "step": 37887 }, { "epoch": 94.72, "grad_norm": 7.34375, "learning_rate": 1.4585976966567826e-07, "loss": 1.3067, "step": 37888 }, { "epoch": 94.7225, "grad_norm": 8.8125, "learning_rate": 1.4572201395007922e-07, "loss": 1.3745, "step": 37889 }, { "epoch": 94.725, "grad_norm": 6.03125, "learning_rate": 1.4558432283872993e-07, "loss": 1.2133, "step": 37890 }, { "epoch": 94.7275, "grad_norm": 6.0625, "learning_rate": 1.4544669633253627e-07, "loss": 1.2477, "step": 37891 }, { "epoch": 94.73, "grad_norm": 7.6875, "learning_rate": 1.453091344323998e-07, "loss": 1.1915, "step": 37892 }, { "epoch": 94.7325, "grad_norm": 7.0625, "learning_rate": 1.451716371392231e-07, "loss": 1.3849, "step": 37893 }, { "epoch": 94.735, "grad_norm": 5.5, "learning_rate": 1.4503420445390547e-07, "loss": 1.2625, "step": 37894 }, { "epoch": 94.7375, "grad_norm": 8.625, "learning_rate": 1.4489683637734952e-07, "loss": 1.3372, "step": 37895 }, { "epoch": 94.74, "grad_norm": 6.5, "learning_rate": 1.4475953291045674e-07, "loss": 1.209, "step": 37896 }, { "epoch": 94.7425, "grad_norm": 8.9375, "learning_rate": 1.446222940541242e-07, "loss": 1.4037, "step": 37897 }, { "epoch": 94.745, "grad_norm": 10.0625, "learning_rate": 1.444851198092545e-07, "loss": 1.229, "step": 37898 }, { "epoch": 94.7475, "grad_norm": 6.84375, "learning_rate": 1.443480101767458e-07, "loss": 1.2904, "step": 37899 }, { "epoch": 94.75, "grad_norm": 8.0625, "learning_rate": 1.4421096515749855e-07, "loss": 1.2714, "step": 37900 }, { "epoch": 94.7525, "grad_norm": 8.875, "learning_rate": 1.4407398475240753e-07, "loss": 1.4853, "step": 37901 }, { "epoch": 94.755, "grad_norm": 7.5, "learning_rate": 1.4393706896237314e-07, "loss": 1.1747, "step": 37902 }, { "epoch": 94.7575, "grad_norm": 8.0, "learning_rate": 1.4380021778829468e-07, "loss": 1.266, "step": 37903 }, { "epoch": 94.76, "grad_norm": 8.8125, "learning_rate": 1.4366343123106697e-07, "loss": 1.1149, "step": 37904 }, { "epoch": 94.7625, "grad_norm": 6.1875, "learning_rate": 1.4352670929158708e-07, "loss": 1.1895, "step": 37905 }, { "epoch": 94.765, "grad_norm": 7.28125, "learning_rate": 1.4339005197075208e-07, "loss": 1.3109, "step": 37906 }, { "epoch": 94.7675, "grad_norm": 7.1875, "learning_rate": 1.4325345926945676e-07, "loss": 1.3538, "step": 37907 }, { "epoch": 94.77, "grad_norm": 7.09375, "learning_rate": 1.4311693118859716e-07, "loss": 1.167, "step": 37908 }, { "epoch": 94.7725, "grad_norm": 7.21875, "learning_rate": 1.4298046772906916e-07, "loss": 1.3217, "step": 37909 }, { "epoch": 94.775, "grad_norm": 6.8125, "learning_rate": 1.428440688917665e-07, "loss": 1.3336, "step": 37910 }, { "epoch": 94.7775, "grad_norm": 8.8125, "learning_rate": 1.4270773467758403e-07, "loss": 1.2904, "step": 37911 }, { "epoch": 94.78, "grad_norm": 8.625, "learning_rate": 1.4257146508741436e-07, "loss": 1.3555, "step": 37912 }, { "epoch": 94.7825, "grad_norm": 6.78125, "learning_rate": 1.4243526012215238e-07, "loss": 1.4178, "step": 37913 }, { "epoch": 94.785, "grad_norm": 10.0, "learning_rate": 1.4229911978268952e-07, "loss": 1.3666, "step": 37914 }, { "epoch": 94.7875, "grad_norm": 8.75, "learning_rate": 1.4216304406992067e-07, "loss": 1.2654, "step": 37915 }, { "epoch": 94.79, "grad_norm": 6.53125, "learning_rate": 1.4202703298473508e-07, "loss": 1.3763, "step": 37916 }, { "epoch": 94.7925, "grad_norm": 6.875, "learning_rate": 1.4189108652802541e-07, "loss": 1.2817, "step": 37917 }, { "epoch": 94.795, "grad_norm": 8.0, "learning_rate": 1.417552047006854e-07, "loss": 1.3461, "step": 37918 }, { "epoch": 94.7975, "grad_norm": 6.28125, "learning_rate": 1.4161938750360206e-07, "loss": 1.2474, "step": 37919 }, { "epoch": 94.8, "grad_norm": 15.0625, "learning_rate": 1.4148363493766803e-07, "loss": 1.2795, "step": 37920 }, { "epoch": 94.8025, "grad_norm": 9.875, "learning_rate": 1.4134794700377263e-07, "loss": 1.3894, "step": 37921 }, { "epoch": 94.805, "grad_norm": 7.09375, "learning_rate": 1.4121232370280514e-07, "loss": 1.2297, "step": 37922 }, { "epoch": 94.8075, "grad_norm": 6.0625, "learning_rate": 1.4107676503565594e-07, "loss": 1.3206, "step": 37923 }, { "epoch": 94.81, "grad_norm": 5.875, "learning_rate": 1.4094127100321208e-07, "loss": 1.2406, "step": 37924 }, { "epoch": 94.8125, "grad_norm": 5.1875, "learning_rate": 1.4080584160636402e-07, "loss": 1.3315, "step": 37925 }, { "epoch": 94.815, "grad_norm": 8.625, "learning_rate": 1.4067047684599767e-07, "loss": 1.2999, "step": 37926 }, { "epoch": 94.8175, "grad_norm": 7.125, "learning_rate": 1.405351767230001e-07, "loss": 1.388, "step": 37927 }, { "epoch": 94.82, "grad_norm": 5.625, "learning_rate": 1.403999412382606e-07, "loss": 1.1786, "step": 37928 }, { "epoch": 94.8225, "grad_norm": 8.25, "learning_rate": 1.4026477039266407e-07, "loss": 1.4312, "step": 37929 }, { "epoch": 94.825, "grad_norm": 9.3125, "learning_rate": 1.401296641870975e-07, "loss": 1.3087, "step": 37930 }, { "epoch": 94.8275, "grad_norm": 5.6875, "learning_rate": 1.399946226224458e-07, "loss": 1.2492, "step": 37931 }, { "epoch": 94.83, "grad_norm": 14.5, "learning_rate": 1.3985964569959486e-07, "loss": 1.1724, "step": 37932 }, { "epoch": 94.8325, "grad_norm": 10.25, "learning_rate": 1.397247334194296e-07, "loss": 1.3704, "step": 37933 }, { "epoch": 94.835, "grad_norm": 4.78125, "learning_rate": 1.3958988578283484e-07, "loss": 1.2615, "step": 37934 }, { "epoch": 94.8375, "grad_norm": 7.4375, "learning_rate": 1.3945510279069319e-07, "loss": 1.3301, "step": 37935 }, { "epoch": 94.84, "grad_norm": 5.6875, "learning_rate": 1.3932038444389063e-07, "loss": 1.4843, "step": 37936 }, { "epoch": 94.8425, "grad_norm": 7.40625, "learning_rate": 1.3918573074330756e-07, "loss": 1.2129, "step": 37937 }, { "epoch": 94.845, "grad_norm": 6.84375, "learning_rate": 1.390511416898288e-07, "loss": 1.3511, "step": 37938 }, { "epoch": 94.8475, "grad_norm": 8.9375, "learning_rate": 1.3891661728433592e-07, "loss": 1.3978, "step": 37939 }, { "epoch": 94.85, "grad_norm": 9.125, "learning_rate": 1.3878215752771264e-07, "loss": 1.4038, "step": 37940 }, { "epoch": 94.8525, "grad_norm": 7.1875, "learning_rate": 1.3864776242083712e-07, "loss": 1.2973, "step": 37941 }, { "epoch": 94.855, "grad_norm": 9.4375, "learning_rate": 1.385134319645931e-07, "loss": 1.2407, "step": 37942 }, { "epoch": 94.8575, "grad_norm": 8.1875, "learning_rate": 1.383791661598599e-07, "loss": 1.349, "step": 37943 }, { "epoch": 94.86, "grad_norm": 7.5, "learning_rate": 1.3824496500751904e-07, "loss": 1.1144, "step": 37944 }, { "epoch": 94.8625, "grad_norm": 15.125, "learning_rate": 1.381108285084487e-07, "loss": 1.3197, "step": 37945 }, { "epoch": 94.865, "grad_norm": 9.4375, "learning_rate": 1.3797675666352927e-07, "loss": 1.3933, "step": 37946 }, { "epoch": 94.8675, "grad_norm": 7.34375, "learning_rate": 1.3784274947363895e-07, "loss": 1.4486, "step": 37947 }, { "epoch": 94.87, "grad_norm": 5.75, "learning_rate": 1.3770880693965816e-07, "loss": 1.2133, "step": 37948 }, { "epoch": 94.8725, "grad_norm": 7.09375, "learning_rate": 1.3757492906246394e-07, "loss": 1.3777, "step": 37949 }, { "epoch": 94.875, "grad_norm": 8.3125, "learning_rate": 1.374411158429323e-07, "loss": 1.322, "step": 37950 }, { "epoch": 94.8775, "grad_norm": 9.375, "learning_rate": 1.373073672819425e-07, "loss": 1.1584, "step": 37951 }, { "epoch": 94.88, "grad_norm": 6.65625, "learning_rate": 1.3717368338037163e-07, "loss": 1.4263, "step": 37952 }, { "epoch": 94.8825, "grad_norm": 8.4375, "learning_rate": 1.3704006413909566e-07, "loss": 1.4941, "step": 37953 }, { "epoch": 94.885, "grad_norm": 11.3125, "learning_rate": 1.3690650955898943e-07, "loss": 1.2938, "step": 37954 }, { "epoch": 94.8875, "grad_norm": 8.125, "learning_rate": 1.3677301964093005e-07, "loss": 1.3713, "step": 37955 }, { "epoch": 94.89, "grad_norm": 9.25, "learning_rate": 1.366395943857912e-07, "loss": 1.4089, "step": 37956 }, { "epoch": 94.8925, "grad_norm": 7.15625, "learning_rate": 1.3650623379444895e-07, "loss": 1.4034, "step": 37957 }, { "epoch": 94.895, "grad_norm": 9.4375, "learning_rate": 1.3637293786777695e-07, "loss": 1.4068, "step": 37958 }, { "epoch": 94.8975, "grad_norm": 7.78125, "learning_rate": 1.3623970660665008e-07, "loss": 1.3349, "step": 37959 }, { "epoch": 94.9, "grad_norm": 6.8125, "learning_rate": 1.361065400119399e-07, "loss": 1.2742, "step": 37960 }, { "epoch": 94.9025, "grad_norm": 8.6875, "learning_rate": 1.3597343808452123e-07, "loss": 1.3763, "step": 37961 }, { "epoch": 94.905, "grad_norm": 7.84375, "learning_rate": 1.3584040082526452e-07, "loss": 1.4457, "step": 37962 }, { "epoch": 94.9075, "grad_norm": 6.1875, "learning_rate": 1.357074282350457e-07, "loss": 1.3263, "step": 37963 }, { "epoch": 94.91, "grad_norm": 7.625, "learning_rate": 1.3557452031473296e-07, "loss": 1.2584, "step": 37964 }, { "epoch": 94.9125, "grad_norm": 8.375, "learning_rate": 1.3544167706519895e-07, "loss": 1.3319, "step": 37965 }, { "epoch": 94.915, "grad_norm": 9.0, "learning_rate": 1.353088984873141e-07, "loss": 1.299, "step": 37966 }, { "epoch": 94.9175, "grad_norm": 6.6875, "learning_rate": 1.351761845819499e-07, "loss": 1.3133, "step": 37967 }, { "epoch": 94.92, "grad_norm": 9.9375, "learning_rate": 1.3504353534997682e-07, "loss": 1.4016, "step": 37968 }, { "epoch": 94.9225, "grad_norm": 9.8125, "learning_rate": 1.349109507922619e-07, "loss": 1.3845, "step": 37969 }, { "epoch": 94.925, "grad_norm": 6.65625, "learning_rate": 1.347784309096767e-07, "loss": 1.3781, "step": 37970 }, { "epoch": 94.9275, "grad_norm": 7.0, "learning_rate": 1.3464597570308935e-07, "loss": 1.401, "step": 37971 }, { "epoch": 94.93, "grad_norm": 7.875, "learning_rate": 1.345135851733681e-07, "loss": 1.4, "step": 37972 }, { "epoch": 94.9325, "grad_norm": 8.0625, "learning_rate": 1.3438125932138114e-07, "loss": 1.3823, "step": 37973 }, { "epoch": 94.935, "grad_norm": 6.84375, "learning_rate": 1.3424899814799552e-07, "loss": 1.4623, "step": 37974 }, { "epoch": 94.9375, "grad_norm": 6.15625, "learning_rate": 1.3411680165407948e-07, "loss": 1.3777, "step": 37975 }, { "epoch": 94.94, "grad_norm": 16.0, "learning_rate": 1.3398466984049786e-07, "loss": 1.3353, "step": 37976 }, { "epoch": 94.9425, "grad_norm": 7.21875, "learning_rate": 1.3385260270811773e-07, "loss": 1.2349, "step": 37977 }, { "epoch": 94.945, "grad_norm": 6.1875, "learning_rate": 1.337206002578062e-07, "loss": 1.3444, "step": 37978 }, { "epoch": 94.9475, "grad_norm": 7.53125, "learning_rate": 1.3358866249042812e-07, "loss": 1.1911, "step": 37979 }, { "epoch": 94.95, "grad_norm": 10.375, "learning_rate": 1.3345678940684615e-07, "loss": 1.373, "step": 37980 }, { "epoch": 94.9525, "grad_norm": 7.0, "learning_rate": 1.3332498100792735e-07, "loss": 1.3122, "step": 37981 }, { "epoch": 94.955, "grad_norm": 7.0, "learning_rate": 1.331932372945355e-07, "loss": 1.1952, "step": 37982 }, { "epoch": 94.9575, "grad_norm": 9.0625, "learning_rate": 1.330615582675332e-07, "loss": 1.311, "step": 37983 }, { "epoch": 94.96, "grad_norm": 11.375, "learning_rate": 1.3292994392778535e-07, "loss": 1.2173, "step": 37984 }, { "epoch": 94.9625, "grad_norm": 7.25, "learning_rate": 1.327983942761535e-07, "loss": 1.3946, "step": 37985 }, { "epoch": 94.965, "grad_norm": 5.15625, "learning_rate": 1.3266690931350023e-07, "loss": 1.3796, "step": 37986 }, { "epoch": 94.9675, "grad_norm": 6.1875, "learning_rate": 1.3253548904068825e-07, "loss": 1.353, "step": 37987 }, { "epoch": 94.97, "grad_norm": 8.75, "learning_rate": 1.3240413345857795e-07, "loss": 1.3023, "step": 37988 }, { "epoch": 94.9725, "grad_norm": 9.8125, "learning_rate": 1.3227284256803307e-07, "loss": 1.352, "step": 37989 }, { "epoch": 94.975, "grad_norm": 7.21875, "learning_rate": 1.3214161636991186e-07, "loss": 1.4286, "step": 37990 }, { "epoch": 94.9775, "grad_norm": 6.71875, "learning_rate": 1.3201045486507357e-07, "loss": 1.2695, "step": 37991 }, { "epoch": 94.98, "grad_norm": 6.875, "learning_rate": 1.318793580543809e-07, "loss": 1.3163, "step": 37992 }, { "epoch": 94.9825, "grad_norm": 6.875, "learning_rate": 1.3174832593869315e-07, "loss": 1.2996, "step": 37993 }, { "epoch": 94.985, "grad_norm": 5.1875, "learning_rate": 1.316173585188685e-07, "loss": 1.1761, "step": 37994 }, { "epoch": 94.9875, "grad_norm": 11.8125, "learning_rate": 1.3148645579576403e-07, "loss": 1.4127, "step": 37995 }, { "epoch": 94.99, "grad_norm": 5.46875, "learning_rate": 1.313556177702402e-07, "loss": 1.3723, "step": 37996 }, { "epoch": 94.9925, "grad_norm": 10.0625, "learning_rate": 1.312248444431541e-07, "loss": 1.2334, "step": 37997 }, { "epoch": 94.995, "grad_norm": 9.5, "learning_rate": 1.310941358153639e-07, "loss": 1.2895, "step": 37998 }, { "epoch": 94.9975, "grad_norm": 8.4375, "learning_rate": 1.309634918877234e-07, "loss": 1.3178, "step": 37999 }, { "epoch": 95.0, "grad_norm": 13.5625, "learning_rate": 1.30832912661093e-07, "loss": 1.3173, "step": 38000 }, { "epoch": 95.0025, "grad_norm": 6.9375, "learning_rate": 1.3070239813632645e-07, "loss": 1.2115, "step": 38001 }, { "epoch": 95.005, "grad_norm": 7.96875, "learning_rate": 1.305719483142809e-07, "loss": 1.3624, "step": 38002 }, { "epoch": 95.0075, "grad_norm": 6.84375, "learning_rate": 1.3044156319581004e-07, "loss": 1.3477, "step": 38003 }, { "epoch": 95.01, "grad_norm": 6.25, "learning_rate": 1.3031124278176875e-07, "loss": 1.3589, "step": 38004 }, { "epoch": 95.0125, "grad_norm": 7.5, "learning_rate": 1.3018098707301307e-07, "loss": 1.3272, "step": 38005 }, { "epoch": 95.015, "grad_norm": 8.5625, "learning_rate": 1.3005079607039562e-07, "loss": 1.5212, "step": 38006 }, { "epoch": 95.0175, "grad_norm": 10.875, "learning_rate": 1.2992066977476903e-07, "loss": 1.3859, "step": 38007 }, { "epoch": 95.02, "grad_norm": 6.34375, "learning_rate": 1.297906081869893e-07, "loss": 1.2471, "step": 38008 }, { "epoch": 95.0225, "grad_norm": 10.4375, "learning_rate": 1.2966061130790686e-07, "loss": 1.4488, "step": 38009 }, { "epoch": 95.025, "grad_norm": 5.90625, "learning_rate": 1.2953067913837437e-07, "loss": 1.2407, "step": 38010 }, { "epoch": 95.0275, "grad_norm": 7.5, "learning_rate": 1.2940081167924224e-07, "loss": 1.3265, "step": 38011 }, { "epoch": 95.03, "grad_norm": 6.46875, "learning_rate": 1.2927100893136534e-07, "loss": 1.3205, "step": 38012 }, { "epoch": 95.0325, "grad_norm": 7.5625, "learning_rate": 1.2914127089559192e-07, "loss": 1.394, "step": 38013 }, { "epoch": 95.035, "grad_norm": 7.25, "learning_rate": 1.290115975727746e-07, "loss": 1.2116, "step": 38014 }, { "epoch": 95.0375, "grad_norm": 6.375, "learning_rate": 1.2888198896376048e-07, "loss": 1.2469, "step": 38015 }, { "epoch": 95.04, "grad_norm": 6.90625, "learning_rate": 1.287524450694011e-07, "loss": 1.2472, "step": 38016 }, { "epoch": 95.0425, "grad_norm": 9.1875, "learning_rate": 1.2862296589054691e-07, "loss": 1.2413, "step": 38017 }, { "epoch": 95.045, "grad_norm": 8.3125, "learning_rate": 1.2849355142804497e-07, "loss": 1.3592, "step": 38018 }, { "epoch": 95.0475, "grad_norm": 8.5625, "learning_rate": 1.2836420168274356e-07, "loss": 1.3613, "step": 38019 }, { "epoch": 95.05, "grad_norm": 7.4375, "learning_rate": 1.2823491665549193e-07, "loss": 1.3844, "step": 38020 }, { "epoch": 95.0525, "grad_norm": 11.75, "learning_rate": 1.2810569634713721e-07, "loss": 1.2623, "step": 38021 }, { "epoch": 95.055, "grad_norm": 7.25, "learning_rate": 1.279765407585254e-07, "loss": 1.3089, "step": 38022 }, { "epoch": 95.0575, "grad_norm": 7.34375, "learning_rate": 1.2784744989050578e-07, "loss": 1.2814, "step": 38023 }, { "epoch": 95.06, "grad_norm": 7.21875, "learning_rate": 1.277184237439233e-07, "loss": 1.3088, "step": 38024 }, { "epoch": 95.0625, "grad_norm": 10.875, "learning_rate": 1.2758946231962389e-07, "loss": 1.3263, "step": 38025 }, { "epoch": 95.065, "grad_norm": 5.8125, "learning_rate": 1.2746056561845133e-07, "loss": 1.2833, "step": 38026 }, { "epoch": 95.0675, "grad_norm": 7.84375, "learning_rate": 1.2733173364125273e-07, "loss": 1.325, "step": 38027 }, { "epoch": 95.07, "grad_norm": 7.15625, "learning_rate": 1.2720296638887297e-07, "loss": 1.3534, "step": 38028 }, { "epoch": 95.0725, "grad_norm": 7.96875, "learning_rate": 1.270742638621558e-07, "loss": 1.3027, "step": 38029 }, { "epoch": 95.075, "grad_norm": 6.21875, "learning_rate": 1.2694562606194394e-07, "loss": 1.4102, "step": 38030 }, { "epoch": 95.0775, "grad_norm": 9.0, "learning_rate": 1.2681705298908108e-07, "loss": 1.1936, "step": 38031 }, { "epoch": 95.08, "grad_norm": 8.4375, "learning_rate": 1.2668854464441104e-07, "loss": 1.2224, "step": 38032 }, { "epoch": 95.0825, "grad_norm": 6.21875, "learning_rate": 1.2656010102877647e-07, "loss": 1.3146, "step": 38033 }, { "epoch": 95.085, "grad_norm": 87.5, "learning_rate": 1.264317221430178e-07, "loss": 1.2427, "step": 38034 }, { "epoch": 95.0875, "grad_norm": 8.25, "learning_rate": 1.2630340798797768e-07, "loss": 1.3191, "step": 38035 }, { "epoch": 95.09, "grad_norm": 11.0625, "learning_rate": 1.261751585644977e-07, "loss": 1.2187, "step": 38036 }, { "epoch": 95.0925, "grad_norm": 11.625, "learning_rate": 1.2604697387341823e-07, "loss": 1.2715, "step": 38037 }, { "epoch": 95.095, "grad_norm": 5.6875, "learning_rate": 1.2591885391557867e-07, "loss": 1.2865, "step": 38038 }, { "epoch": 95.0975, "grad_norm": 6.875, "learning_rate": 1.2579079869182165e-07, "loss": 1.2978, "step": 38039 }, { "epoch": 95.1, "grad_norm": 7.78125, "learning_rate": 1.2566280820298427e-07, "loss": 1.3011, "step": 38040 }, { "epoch": 95.1025, "grad_norm": 7.78125, "learning_rate": 1.2553488244990585e-07, "loss": 1.2677, "step": 38041 }, { "epoch": 95.105, "grad_norm": 5.8125, "learning_rate": 1.2540702143342576e-07, "loss": 1.3265, "step": 38042 }, { "epoch": 95.1075, "grad_norm": 10.0625, "learning_rate": 1.2527922515438217e-07, "loss": 1.4395, "step": 38043 }, { "epoch": 95.11, "grad_norm": 7.21875, "learning_rate": 1.2515149361361335e-07, "loss": 1.405, "step": 38044 }, { "epoch": 95.1125, "grad_norm": 7.78125, "learning_rate": 1.2502382681195525e-07, "loss": 1.462, "step": 38045 }, { "epoch": 95.115, "grad_norm": 8.1875, "learning_rate": 1.248962247502461e-07, "loss": 1.3697, "step": 38046 }, { "epoch": 95.1175, "grad_norm": 5.8125, "learning_rate": 1.247686874293219e-07, "loss": 1.2981, "step": 38047 }, { "epoch": 95.12, "grad_norm": 9.75, "learning_rate": 1.246412148500198e-07, "loss": 1.2007, "step": 38048 }, { "epoch": 95.1225, "grad_norm": 6.8125, "learning_rate": 1.245138070131746e-07, "loss": 1.3849, "step": 38049 }, { "epoch": 95.125, "grad_norm": 8.4375, "learning_rate": 1.243864639196213e-07, "loss": 1.2522, "step": 38050 }, { "epoch": 95.1275, "grad_norm": 5.84375, "learning_rate": 1.2425918557019466e-07, "loss": 1.4197, "step": 38051 }, { "epoch": 95.13, "grad_norm": 8.125, "learning_rate": 1.241319719657308e-07, "loss": 1.1766, "step": 38052 }, { "epoch": 95.1325, "grad_norm": 6.9375, "learning_rate": 1.2400482310706118e-07, "loss": 1.4855, "step": 38053 }, { "epoch": 95.135, "grad_norm": 7.375, "learning_rate": 1.2387773899502188e-07, "loss": 1.3059, "step": 38054 }, { "epoch": 95.1375, "grad_norm": 9.6875, "learning_rate": 1.237507196304444e-07, "loss": 1.2148, "step": 38055 }, { "epoch": 95.14, "grad_norm": 7.9375, "learning_rate": 1.2362376501416251e-07, "loss": 1.3134, "step": 38056 }, { "epoch": 95.1425, "grad_norm": 7.0, "learning_rate": 1.234968751470067e-07, "loss": 1.2493, "step": 38057 }, { "epoch": 95.145, "grad_norm": 8.875, "learning_rate": 1.2337005002981185e-07, "loss": 1.2084, "step": 38058 }, { "epoch": 95.1475, "grad_norm": 7.40625, "learning_rate": 1.2324328966340616e-07, "loss": 1.3702, "step": 38059 }, { "epoch": 95.15, "grad_norm": 8.9375, "learning_rate": 1.231165940486234e-07, "loss": 1.1635, "step": 38060 }, { "epoch": 95.1525, "grad_norm": 8.0, "learning_rate": 1.2298996318629186e-07, "loss": 1.2675, "step": 38061 }, { "epoch": 95.155, "grad_norm": 6.34375, "learning_rate": 1.2286339707724304e-07, "loss": 1.2085, "step": 38062 }, { "epoch": 95.1575, "grad_norm": 7.4375, "learning_rate": 1.227368957223074e-07, "loss": 1.4047, "step": 38063 }, { "epoch": 95.16, "grad_norm": 7.78125, "learning_rate": 1.2261045912231318e-07, "loss": 1.403, "step": 38064 }, { "epoch": 95.1625, "grad_norm": 8.8125, "learning_rate": 1.224840872780897e-07, "loss": 1.5022, "step": 38065 }, { "epoch": 95.165, "grad_norm": 6.53125, "learning_rate": 1.2235778019046408e-07, "loss": 1.3455, "step": 38066 }, { "epoch": 95.1675, "grad_norm": 9.125, "learning_rate": 1.2223153786026675e-07, "loss": 1.2689, "step": 38067 }, { "epoch": 95.17, "grad_norm": 5.34375, "learning_rate": 1.2210536028832377e-07, "loss": 1.2498, "step": 38068 }, { "epoch": 95.1725, "grad_norm": 5.875, "learning_rate": 1.219792474754622e-07, "loss": 1.3061, "step": 38069 }, { "epoch": 95.175, "grad_norm": 7.1875, "learning_rate": 1.2185319942251028e-07, "loss": 1.224, "step": 38070 }, { "epoch": 95.1775, "grad_norm": 6.28125, "learning_rate": 1.2172721613029292e-07, "loss": 1.2236, "step": 38071 }, { "epoch": 95.18, "grad_norm": 4.78125, "learning_rate": 1.2160129759963723e-07, "loss": 1.6142, "step": 38072 }, { "epoch": 95.1825, "grad_norm": 7.4375, "learning_rate": 1.2147544383136812e-07, "loss": 1.2098, "step": 38073 }, { "epoch": 95.185, "grad_norm": 6.78125, "learning_rate": 1.2134965482631045e-07, "loss": 1.2752, "step": 38074 }, { "epoch": 95.1875, "grad_norm": 8.8125, "learning_rate": 1.2122393058528803e-07, "loss": 1.4613, "step": 38075 }, { "epoch": 95.19, "grad_norm": 7.71875, "learning_rate": 1.2109827110912687e-07, "loss": 1.302, "step": 38076 }, { "epoch": 95.1925, "grad_norm": 5.53125, "learning_rate": 1.2097267639865073e-07, "loss": 1.3898, "step": 38077 }, { "epoch": 95.195, "grad_norm": 9.5625, "learning_rate": 1.208471464546823e-07, "loss": 1.3046, "step": 38078 }, { "epoch": 95.1975, "grad_norm": 9.3125, "learning_rate": 1.2072168127804428e-07, "loss": 1.5585, "step": 38079 }, { "epoch": 95.2, "grad_norm": 6.46875, "learning_rate": 1.2059628086956044e-07, "loss": 1.3424, "step": 38080 }, { "epoch": 95.2025, "grad_norm": 8.6875, "learning_rate": 1.204709452300501e-07, "loss": 1.1468, "step": 38081 }, { "epoch": 95.205, "grad_norm": 14.8125, "learning_rate": 1.2034567436033927e-07, "loss": 1.1752, "step": 38082 }, { "epoch": 95.2075, "grad_norm": 6.5625, "learning_rate": 1.202204682612451e-07, "loss": 1.4162, "step": 38083 }, { "epoch": 95.21, "grad_norm": 6.78125, "learning_rate": 1.2009532693359027e-07, "loss": 1.2454, "step": 38084 }, { "epoch": 95.2125, "grad_norm": 5.5625, "learning_rate": 1.199702503781963e-07, "loss": 1.3036, "step": 38085 }, { "epoch": 95.215, "grad_norm": 9.25, "learning_rate": 1.198452385958804e-07, "loss": 1.3037, "step": 38086 }, { "epoch": 95.2175, "grad_norm": 5.03125, "learning_rate": 1.197202915874651e-07, "loss": 1.2575, "step": 38087 }, { "epoch": 95.22, "grad_norm": 6.0625, "learning_rate": 1.1959540935376769e-07, "loss": 1.319, "step": 38088 }, { "epoch": 95.2225, "grad_norm": 7.875, "learning_rate": 1.194705918956074e-07, "loss": 1.312, "step": 38089 }, { "epoch": 95.225, "grad_norm": 9.5625, "learning_rate": 1.1934583921380249e-07, "loss": 1.1412, "step": 38090 }, { "epoch": 95.2275, "grad_norm": 9.625, "learning_rate": 1.1922115130917012e-07, "loss": 1.391, "step": 38091 }, { "epoch": 95.23, "grad_norm": 7.84375, "learning_rate": 1.190965281825296e-07, "loss": 1.3777, "step": 38092 }, { "epoch": 95.2325, "grad_norm": 9.1875, "learning_rate": 1.1897196983469694e-07, "loss": 1.3163, "step": 38093 }, { "epoch": 95.235, "grad_norm": 8.375, "learning_rate": 1.1884747626648707e-07, "loss": 1.3803, "step": 38094 }, { "epoch": 95.2375, "grad_norm": 5.8125, "learning_rate": 1.1872304747871933e-07, "loss": 1.2839, "step": 38095 }, { "epoch": 95.24, "grad_norm": 7.78125, "learning_rate": 1.1859868347220749e-07, "loss": 1.4187, "step": 38096 }, { "epoch": 95.2425, "grad_norm": 6.28125, "learning_rate": 1.1847438424776648e-07, "loss": 1.3781, "step": 38097 }, { "epoch": 95.245, "grad_norm": 6.96875, "learning_rate": 1.183501498062123e-07, "loss": 1.3842, "step": 38098 }, { "epoch": 95.2475, "grad_norm": 7.375, "learning_rate": 1.1822598014835984e-07, "loss": 1.2031, "step": 38099 }, { "epoch": 95.25, "grad_norm": 8.875, "learning_rate": 1.1810187527502182e-07, "loss": 1.2711, "step": 38100 }, { "epoch": 95.2525, "grad_norm": 14.875, "learning_rate": 1.1797783518701312e-07, "loss": 1.4491, "step": 38101 }, { "epoch": 95.255, "grad_norm": 7.8125, "learning_rate": 1.1785385988514531e-07, "loss": 1.3247, "step": 38102 }, { "epoch": 95.2575, "grad_norm": 7.6875, "learning_rate": 1.177299493702322e-07, "loss": 1.3859, "step": 38103 }, { "epoch": 95.26, "grad_norm": 8.5, "learning_rate": 1.1760610364308645e-07, "loss": 1.3716, "step": 38104 }, { "epoch": 95.2625, "grad_norm": 9.3125, "learning_rate": 1.1748232270451965e-07, "loss": 1.2302, "step": 38105 }, { "epoch": 95.265, "grad_norm": 9.375, "learning_rate": 1.1735860655534226e-07, "loss": 1.2862, "step": 38106 }, { "epoch": 95.2675, "grad_norm": 11.0625, "learning_rate": 1.1723495519636807e-07, "loss": 1.5318, "step": 38107 }, { "epoch": 95.27, "grad_norm": 7.0, "learning_rate": 1.1711136862840533e-07, "loss": 1.2196, "step": 38108 }, { "epoch": 95.2725, "grad_norm": 6.90625, "learning_rate": 1.169878468522645e-07, "loss": 1.175, "step": 38109 }, { "epoch": 95.275, "grad_norm": 6.375, "learning_rate": 1.1686438986875715e-07, "loss": 1.2753, "step": 38110 }, { "epoch": 95.2775, "grad_norm": 5.8125, "learning_rate": 1.167409976786893e-07, "loss": 1.2484, "step": 38111 }, { "epoch": 95.28, "grad_norm": 7.4375, "learning_rate": 1.1661767028287363e-07, "loss": 1.5272, "step": 38112 }, { "epoch": 95.2825, "grad_norm": 5.90625, "learning_rate": 1.1649440768211617e-07, "loss": 1.4607, "step": 38113 }, { "epoch": 95.285, "grad_norm": 7.59375, "learning_rate": 1.1637120987722628e-07, "loss": 1.2067, "step": 38114 }, { "epoch": 95.2875, "grad_norm": 9.125, "learning_rate": 1.1624807686901108e-07, "loss": 1.1738, "step": 38115 }, { "epoch": 95.29, "grad_norm": 8.4375, "learning_rate": 1.161250086582777e-07, "loss": 1.5304, "step": 38116 }, { "epoch": 95.2925, "grad_norm": 8.5625, "learning_rate": 1.1600200524583327e-07, "loss": 1.1498, "step": 38117 }, { "epoch": 95.295, "grad_norm": 6.46875, "learning_rate": 1.1587906663248494e-07, "loss": 1.2828, "step": 38118 }, { "epoch": 95.2975, "grad_norm": 7.59375, "learning_rate": 1.1575619281903761e-07, "loss": 1.1567, "step": 38119 }, { "epoch": 95.3, "grad_norm": 7.84375, "learning_rate": 1.1563338380629618e-07, "loss": 1.4088, "step": 38120 }, { "epoch": 95.3025, "grad_norm": 7.8125, "learning_rate": 1.1551063959506781e-07, "loss": 1.2659, "step": 38121 }, { "epoch": 95.305, "grad_norm": 8.75, "learning_rate": 1.1538796018615628e-07, "loss": 1.3574, "step": 38122 }, { "epoch": 95.3075, "grad_norm": 7.125, "learning_rate": 1.1526534558036539e-07, "loss": 1.3495, "step": 38123 }, { "epoch": 95.31, "grad_norm": 7.78125, "learning_rate": 1.1514279577849786e-07, "loss": 1.3347, "step": 38124 }, { "epoch": 95.3125, "grad_norm": 6.75, "learning_rate": 1.1502031078136078e-07, "loss": 1.3619, "step": 38125 }, { "epoch": 95.315, "grad_norm": 10.3125, "learning_rate": 1.1489789058975353e-07, "loss": 1.345, "step": 38126 }, { "epoch": 95.3175, "grad_norm": 4.90625, "learning_rate": 1.1477553520448103e-07, "loss": 1.3508, "step": 38127 }, { "epoch": 95.32, "grad_norm": 11.1875, "learning_rate": 1.1465324462634375e-07, "loss": 1.2984, "step": 38128 }, { "epoch": 95.3225, "grad_norm": 8.8125, "learning_rate": 1.1453101885614548e-07, "loss": 1.2876, "step": 38129 }, { "epoch": 95.325, "grad_norm": 8.0, "learning_rate": 1.1440885789468559e-07, "loss": 1.3704, "step": 38130 }, { "epoch": 95.3275, "grad_norm": 7.125, "learning_rate": 1.1428676174276454e-07, "loss": 1.3359, "step": 38131 }, { "epoch": 95.33, "grad_norm": 7.625, "learning_rate": 1.1416473040118504e-07, "loss": 1.3084, "step": 38132 }, { "epoch": 95.3325, "grad_norm": 9.375, "learning_rate": 1.1404276387074532e-07, "loss": 1.3382, "step": 38133 }, { "epoch": 95.335, "grad_norm": 5.125, "learning_rate": 1.1392086215224695e-07, "loss": 1.4015, "step": 38134 }, { "epoch": 95.3375, "grad_norm": 12.25, "learning_rate": 1.13799025246486e-07, "loss": 1.447, "step": 38135 }, { "epoch": 95.34, "grad_norm": 6.75, "learning_rate": 1.1367725315426403e-07, "loss": 1.2714, "step": 38136 }, { "epoch": 95.3425, "grad_norm": 6.4375, "learning_rate": 1.1355554587637818e-07, "loss": 1.3012, "step": 38137 }, { "epoch": 95.345, "grad_norm": 7.5625, "learning_rate": 1.134339034136267e-07, "loss": 1.4027, "step": 38138 }, { "epoch": 95.3475, "grad_norm": 7.15625, "learning_rate": 1.1331232576680673e-07, "loss": 1.3944, "step": 38139 }, { "epoch": 95.35, "grad_norm": 7.59375, "learning_rate": 1.1319081293671541e-07, "loss": 1.3281, "step": 38140 }, { "epoch": 95.3525, "grad_norm": 12.125, "learning_rate": 1.1306936492414989e-07, "loss": 1.2963, "step": 38141 }, { "epoch": 95.355, "grad_norm": 7.8125, "learning_rate": 1.1294798172990507e-07, "loss": 1.3244, "step": 38142 }, { "epoch": 95.3575, "grad_norm": 12.1875, "learning_rate": 1.128266633547781e-07, "loss": 1.3595, "step": 38143 }, { "epoch": 95.36, "grad_norm": 8.5625, "learning_rate": 1.1270540979956501e-07, "loss": 1.2496, "step": 38144 }, { "epoch": 95.3625, "grad_norm": 6.46875, "learning_rate": 1.125842210650585e-07, "loss": 1.4517, "step": 38145 }, { "epoch": 95.365, "grad_norm": 7.6875, "learning_rate": 1.124630971520535e-07, "loss": 1.1944, "step": 38146 }, { "epoch": 95.3675, "grad_norm": 7.0, "learning_rate": 1.1234203806134491e-07, "loss": 1.5291, "step": 38147 }, { "epoch": 95.37, "grad_norm": 7.5, "learning_rate": 1.1222104379372767e-07, "loss": 1.2259, "step": 38148 }, { "epoch": 95.3725, "grad_norm": 9.5, "learning_rate": 1.1210011434999335e-07, "loss": 1.2482, "step": 38149 }, { "epoch": 95.375, "grad_norm": 7.75, "learning_rate": 1.1197924973093466e-07, "loss": 1.2711, "step": 38150 }, { "epoch": 95.3775, "grad_norm": 8.0625, "learning_rate": 1.1185844993734319e-07, "loss": 1.473, "step": 38151 }, { "epoch": 95.38, "grad_norm": 10.1875, "learning_rate": 1.1173771497001273e-07, "loss": 1.2329, "step": 38152 }, { "epoch": 95.3825, "grad_norm": 7.15625, "learning_rate": 1.116170448297349e-07, "loss": 1.4314, "step": 38153 }, { "epoch": 95.385, "grad_norm": 6.3125, "learning_rate": 1.1149643951729905e-07, "loss": 1.3322, "step": 38154 }, { "epoch": 95.3875, "grad_norm": 10.5, "learning_rate": 1.1137589903349677e-07, "loss": 1.3533, "step": 38155 }, { "epoch": 95.39, "grad_norm": 6.53125, "learning_rate": 1.1125542337911854e-07, "loss": 1.2783, "step": 38156 }, { "epoch": 95.3925, "grad_norm": 9.0625, "learning_rate": 1.1113501255495485e-07, "loss": 1.2907, "step": 38157 }, { "epoch": 95.395, "grad_norm": 6.3125, "learning_rate": 1.1101466656179283e-07, "loss": 1.4692, "step": 38158 }, { "epoch": 95.3975, "grad_norm": 8.6875, "learning_rate": 1.1089438540042407e-07, "loss": 1.3233, "step": 38159 }, { "epoch": 95.4, "grad_norm": 8.1875, "learning_rate": 1.1077416907163573e-07, "loss": 1.463, "step": 38160 }, { "epoch": 95.4025, "grad_norm": 5.9375, "learning_rate": 1.1065401757621608e-07, "loss": 1.3284, "step": 38161 }, { "epoch": 95.405, "grad_norm": 5.4375, "learning_rate": 1.1053393091495223e-07, "loss": 1.3441, "step": 38162 }, { "epoch": 95.4075, "grad_norm": 10.1875, "learning_rate": 1.1041390908863358e-07, "loss": 1.3456, "step": 38163 }, { "epoch": 95.41, "grad_norm": 6.53125, "learning_rate": 1.1029395209804505e-07, "loss": 1.3274, "step": 38164 }, { "epoch": 95.4125, "grad_norm": 5.71875, "learning_rate": 1.1017405994397378e-07, "loss": 1.3963, "step": 38165 }, { "epoch": 95.415, "grad_norm": 9.4375, "learning_rate": 1.1005423262720472e-07, "loss": 1.3816, "step": 38166 }, { "epoch": 95.4175, "grad_norm": 5.34375, "learning_rate": 1.0993447014852499e-07, "loss": 1.4674, "step": 38167 }, { "epoch": 95.42, "grad_norm": 7.09375, "learning_rate": 1.0981477250871952e-07, "loss": 1.4375, "step": 38168 }, { "epoch": 95.4225, "grad_norm": 7.1875, "learning_rate": 1.0969513970857216e-07, "loss": 1.3524, "step": 38169 }, { "epoch": 95.425, "grad_norm": 6.875, "learning_rate": 1.0957557174886779e-07, "loss": 1.2995, "step": 38170 }, { "epoch": 95.4275, "grad_norm": 8.875, "learning_rate": 1.0945606863038916e-07, "loss": 1.3025, "step": 38171 }, { "epoch": 95.43, "grad_norm": 8.4375, "learning_rate": 1.0933663035392228e-07, "loss": 1.3893, "step": 38172 }, { "epoch": 95.4325, "grad_norm": 6.625, "learning_rate": 1.0921725692024765e-07, "loss": 1.272, "step": 38173 }, { "epoch": 95.435, "grad_norm": 6.875, "learning_rate": 1.0909794833014909e-07, "loss": 1.3299, "step": 38174 }, { "epoch": 95.4375, "grad_norm": 12.3125, "learning_rate": 1.089787045844093e-07, "loss": 1.3541, "step": 38175 }, { "epoch": 95.44, "grad_norm": 8.125, "learning_rate": 1.0885952568380764e-07, "loss": 1.397, "step": 38176 }, { "epoch": 95.4425, "grad_norm": 10.375, "learning_rate": 1.0874041162912908e-07, "loss": 1.4388, "step": 38177 }, { "epoch": 95.445, "grad_norm": 6.5, "learning_rate": 1.0862136242115073e-07, "loss": 1.3027, "step": 38178 }, { "epoch": 95.4475, "grad_norm": 6.59375, "learning_rate": 1.0850237806065645e-07, "loss": 1.2786, "step": 38179 }, { "epoch": 95.45, "grad_norm": 8.4375, "learning_rate": 1.0838345854842447e-07, "loss": 1.2176, "step": 38180 }, { "epoch": 95.4525, "grad_norm": 7.375, "learning_rate": 1.0826460388523419e-07, "loss": 1.2053, "step": 38181 }, { "epoch": 95.455, "grad_norm": 5.5, "learning_rate": 1.0814581407186497e-07, "loss": 1.3212, "step": 38182 }, { "epoch": 95.4575, "grad_norm": 7.75, "learning_rate": 1.080270891090962e-07, "loss": 1.3134, "step": 38183 }, { "epoch": 95.46, "grad_norm": 9.625, "learning_rate": 1.0790842899770726e-07, "loss": 1.3124, "step": 38184 }, { "epoch": 95.4625, "grad_norm": 9.0, "learning_rate": 1.0778983373847195e-07, "loss": 1.3172, "step": 38185 }, { "epoch": 95.465, "grad_norm": 8.375, "learning_rate": 1.0767130333217302e-07, "loss": 1.4502, "step": 38186 }, { "epoch": 95.4675, "grad_norm": 10.4375, "learning_rate": 1.0755283777958536e-07, "loss": 1.2849, "step": 38187 }, { "epoch": 95.47, "grad_norm": 7.1875, "learning_rate": 1.0743443708148504e-07, "loss": 1.3154, "step": 38188 }, { "epoch": 95.4725, "grad_norm": 8.25, "learning_rate": 1.07316101238647e-07, "loss": 1.1536, "step": 38189 }, { "epoch": 95.475, "grad_norm": 6.65625, "learning_rate": 1.0719783025185059e-07, "loss": 1.325, "step": 38190 }, { "epoch": 95.4775, "grad_norm": 8.1875, "learning_rate": 1.0707962412186856e-07, "loss": 1.2706, "step": 38191 }, { "epoch": 95.48, "grad_norm": 10.5625, "learning_rate": 1.0696148284947694e-07, "loss": 1.3414, "step": 38192 }, { "epoch": 95.4825, "grad_norm": 7.28125, "learning_rate": 1.0684340643544954e-07, "loss": 1.3806, "step": 38193 }, { "epoch": 95.485, "grad_norm": 6.625, "learning_rate": 1.0672539488056133e-07, "loss": 1.4885, "step": 38194 }, { "epoch": 95.4875, "grad_norm": 7.0625, "learning_rate": 1.0660744818558499e-07, "loss": 1.3881, "step": 38195 }, { "epoch": 95.49, "grad_norm": 7.75, "learning_rate": 1.0648956635129548e-07, "loss": 1.379, "step": 38196 }, { "epoch": 95.4925, "grad_norm": 8.125, "learning_rate": 1.0637174937846329e-07, "loss": 1.3627, "step": 38197 }, { "epoch": 95.495, "grad_norm": 8.0, "learning_rate": 1.0625399726786223e-07, "loss": 1.3196, "step": 38198 }, { "epoch": 95.4975, "grad_norm": 6.53125, "learning_rate": 1.0613631002026392e-07, "loss": 1.1329, "step": 38199 }, { "epoch": 95.5, "grad_norm": 5.40625, "learning_rate": 1.0601868763643997e-07, "loss": 1.2818, "step": 38200 }, { "epoch": 95.5025, "grad_norm": 7.96875, "learning_rate": 1.0590113011716197e-07, "loss": 1.3396, "step": 38201 }, { "epoch": 95.505, "grad_norm": 7.53125, "learning_rate": 1.0578363746320041e-07, "loss": 1.3266, "step": 38202 }, { "epoch": 95.5075, "grad_norm": 9.75, "learning_rate": 1.0566620967532471e-07, "loss": 1.3906, "step": 38203 }, { "epoch": 95.51, "grad_norm": 7.40625, "learning_rate": 1.0554884675430532e-07, "loss": 1.3128, "step": 38204 }, { "epoch": 95.5125, "grad_norm": 6.96875, "learning_rate": 1.0543154870091165e-07, "loss": 1.433, "step": 38205 }, { "epoch": 95.515, "grad_norm": 6.9375, "learning_rate": 1.053143155159142e-07, "loss": 1.4282, "step": 38206 }, { "epoch": 95.5175, "grad_norm": 9.625, "learning_rate": 1.05197147200079e-07, "loss": 1.3704, "step": 38207 }, { "epoch": 95.52, "grad_norm": 8.5625, "learning_rate": 1.0508004375417546e-07, "loss": 1.2722, "step": 38208 }, { "epoch": 95.5225, "grad_norm": 6.03125, "learning_rate": 1.0496300517897184e-07, "loss": 1.4235, "step": 38209 }, { "epoch": 95.525, "grad_norm": 5.0, "learning_rate": 1.048460314752331e-07, "loss": 1.2847, "step": 38210 }, { "epoch": 95.5275, "grad_norm": 6.71875, "learning_rate": 1.0472912264372969e-07, "loss": 1.3609, "step": 38211 }, { "epoch": 95.53, "grad_norm": 11.4375, "learning_rate": 1.0461227868522439e-07, "loss": 1.1322, "step": 38212 }, { "epoch": 95.5325, "grad_norm": 6.625, "learning_rate": 1.0449549960048655e-07, "loss": 1.3199, "step": 38213 }, { "epoch": 95.535, "grad_norm": 6.9375, "learning_rate": 1.0437878539028001e-07, "loss": 1.2018, "step": 38214 }, { "epoch": 95.5375, "grad_norm": 7.3125, "learning_rate": 1.0426213605536972e-07, "loss": 1.3086, "step": 38215 }, { "epoch": 95.54, "grad_norm": 7.28125, "learning_rate": 1.0414555159652062e-07, "loss": 1.251, "step": 38216 }, { "epoch": 95.5425, "grad_norm": 7.25, "learning_rate": 1.0402903201449765e-07, "loss": 1.24, "step": 38217 }, { "epoch": 95.545, "grad_norm": 6.96875, "learning_rate": 1.0391257731006466e-07, "loss": 1.2552, "step": 38218 }, { "epoch": 95.5475, "grad_norm": 4.875, "learning_rate": 1.0379618748398434e-07, "loss": 1.2904, "step": 38219 }, { "epoch": 95.55, "grad_norm": 9.4375, "learning_rate": 1.0367986253701945e-07, "loss": 1.4913, "step": 38220 }, { "epoch": 95.5525, "grad_norm": 8.5625, "learning_rate": 1.0356360246993491e-07, "loss": 1.4617, "step": 38221 }, { "epoch": 95.555, "grad_norm": 7.5, "learning_rate": 1.0344740728349012e-07, "loss": 1.4056, "step": 38222 }, { "epoch": 95.5575, "grad_norm": 7.96875, "learning_rate": 1.0333127697844891e-07, "loss": 1.3165, "step": 38223 }, { "epoch": 95.56, "grad_norm": 7.5625, "learning_rate": 1.032152115555718e-07, "loss": 1.346, "step": 38224 }, { "epoch": 95.5625, "grad_norm": 6.53125, "learning_rate": 1.0309921101561926e-07, "loss": 1.373, "step": 38225 }, { "epoch": 95.565, "grad_norm": 6.46875, "learning_rate": 1.0298327535935182e-07, "loss": 1.1576, "step": 38226 }, { "epoch": 95.5675, "grad_norm": 8.125, "learning_rate": 1.0286740458753109e-07, "loss": 1.2362, "step": 38227 }, { "epoch": 95.57, "grad_norm": 8.25, "learning_rate": 1.0275159870091533e-07, "loss": 1.3036, "step": 38228 }, { "epoch": 95.5725, "grad_norm": 7.65625, "learning_rate": 1.0263585770026397e-07, "loss": 1.3041, "step": 38229 }, { "epoch": 95.575, "grad_norm": 6.96875, "learning_rate": 1.0252018158633525e-07, "loss": 1.3789, "step": 38230 }, { "epoch": 95.5775, "grad_norm": 10.1875, "learning_rate": 1.0240457035988749e-07, "loss": 1.4888, "step": 38231 }, { "epoch": 95.58, "grad_norm": 7.4375, "learning_rate": 1.0228902402168118e-07, "loss": 1.1899, "step": 38232 }, { "epoch": 95.5825, "grad_norm": 14.0, "learning_rate": 1.0217354257247125e-07, "loss": 1.4448, "step": 38233 }, { "epoch": 95.585, "grad_norm": 4.9375, "learning_rate": 1.020581260130149e-07, "loss": 1.266, "step": 38234 }, { "epoch": 95.5875, "grad_norm": 7.25, "learning_rate": 1.019427743440693e-07, "loss": 1.5227, "step": 38235 }, { "epoch": 95.59, "grad_norm": 8.4375, "learning_rate": 1.018274875663916e-07, "loss": 1.39, "step": 38236 }, { "epoch": 95.5925, "grad_norm": 6.5625, "learning_rate": 1.0171226568073678e-07, "loss": 1.5198, "step": 38237 }, { "epoch": 95.595, "grad_norm": 6.40625, "learning_rate": 1.0159710868785977e-07, "loss": 1.345, "step": 38238 }, { "epoch": 95.5975, "grad_norm": 9.0625, "learning_rate": 1.0148201658851553e-07, "loss": 1.3011, "step": 38239 }, { "epoch": 95.6, "grad_norm": 7.40625, "learning_rate": 1.0136698938346012e-07, "loss": 1.2906, "step": 38240 }, { "epoch": 95.6025, "grad_norm": 6.96875, "learning_rate": 1.0125202707344628e-07, "loss": 1.2886, "step": 38241 }, { "epoch": 95.605, "grad_norm": 5.5, "learning_rate": 1.0113712965922784e-07, "loss": 1.3278, "step": 38242 }, { "epoch": 95.6075, "grad_norm": 8.6875, "learning_rate": 1.0102229714155864e-07, "loss": 1.6132, "step": 38243 }, { "epoch": 95.61, "grad_norm": 6.75, "learning_rate": 1.0090752952119032e-07, "loss": 1.5016, "step": 38244 }, { "epoch": 95.6125, "grad_norm": 7.96875, "learning_rate": 1.007928267988767e-07, "loss": 1.31, "step": 38245 }, { "epoch": 95.615, "grad_norm": 6.15625, "learning_rate": 1.006781889753683e-07, "loss": 1.4164, "step": 38246 }, { "epoch": 95.6175, "grad_norm": 7.15625, "learning_rate": 1.0056361605141896e-07, "loss": 1.3369, "step": 38247 }, { "epoch": 95.62, "grad_norm": 8.875, "learning_rate": 1.0044910802777807e-07, "loss": 1.2769, "step": 38248 }, { "epoch": 95.6225, "grad_norm": 6.71875, "learning_rate": 1.0033466490519616e-07, "loss": 1.2793, "step": 38249 }, { "epoch": 95.625, "grad_norm": 9.0625, "learning_rate": 1.0022028668442374e-07, "loss": 1.3282, "step": 38250 }, { "epoch": 95.6275, "grad_norm": 7.28125, "learning_rate": 1.0010597336621131e-07, "loss": 1.256, "step": 38251 }, { "epoch": 95.63, "grad_norm": 6.90625, "learning_rate": 9.999172495130827e-08, "loss": 1.3161, "step": 38252 }, { "epoch": 95.6325, "grad_norm": 8.875, "learning_rate": 9.987754144046291e-08, "loss": 1.2648, "step": 38253 }, { "epoch": 95.635, "grad_norm": 5.84375, "learning_rate": 9.976342283442464e-08, "loss": 1.351, "step": 38254 }, { "epoch": 95.6375, "grad_norm": 9.6875, "learning_rate": 9.964936913394063e-08, "loss": 1.2918, "step": 38255 }, { "epoch": 95.64, "grad_norm": 7.28125, "learning_rate": 9.953538033975918e-08, "loss": 1.3418, "step": 38256 }, { "epoch": 95.6425, "grad_norm": 6.25, "learning_rate": 9.942145645262857e-08, "loss": 1.2225, "step": 38257 }, { "epoch": 95.645, "grad_norm": 6.75, "learning_rate": 9.930759747329377e-08, "loss": 1.308, "step": 38258 }, { "epoch": 95.6475, "grad_norm": 9.75, "learning_rate": 9.919380340250306e-08, "loss": 1.2804, "step": 38259 }, { "epoch": 95.65, "grad_norm": 10.1875, "learning_rate": 9.90800742410003e-08, "loss": 1.2973, "step": 38260 }, { "epoch": 95.6525, "grad_norm": 6.5, "learning_rate": 9.896640998953266e-08, "loss": 1.3197, "step": 38261 }, { "epoch": 95.655, "grad_norm": 8.5, "learning_rate": 9.885281064884511e-08, "loss": 1.3407, "step": 38262 }, { "epoch": 95.6575, "grad_norm": 9.0625, "learning_rate": 9.87392762196826e-08, "loss": 1.3378, "step": 38263 }, { "epoch": 95.66, "grad_norm": 9.4375, "learning_rate": 9.862580670278899e-08, "loss": 1.1477, "step": 38264 }, { "epoch": 95.6625, "grad_norm": 6.84375, "learning_rate": 9.851240209890811e-08, "loss": 1.3043, "step": 38265 }, { "epoch": 95.665, "grad_norm": 4.5625, "learning_rate": 9.839906240878494e-08, "loss": 1.3439, "step": 38266 }, { "epoch": 95.6675, "grad_norm": 19.5, "learning_rate": 9.828578763315999e-08, "loss": 1.2958, "step": 38267 }, { "epoch": 95.67, "grad_norm": 7.15625, "learning_rate": 9.81725777727771e-08, "loss": 1.2217, "step": 38268 }, { "epoch": 95.6725, "grad_norm": 7.53125, "learning_rate": 9.805943282837904e-08, "loss": 1.3214, "step": 38269 }, { "epoch": 95.675, "grad_norm": 6.65625, "learning_rate": 9.794635280070741e-08, "loss": 1.3543, "step": 38270 }, { "epoch": 95.6775, "grad_norm": 7.53125, "learning_rate": 9.783333769050273e-08, "loss": 1.3561, "step": 38271 }, { "epoch": 95.68, "grad_norm": 8.3125, "learning_rate": 9.772038749850665e-08, "loss": 1.4172, "step": 38272 }, { "epoch": 95.6825, "grad_norm": 7.40625, "learning_rate": 9.760750222545967e-08, "loss": 1.3823, "step": 38273 }, { "epoch": 95.685, "grad_norm": 7.25, "learning_rate": 9.74946818721012e-08, "loss": 1.3838, "step": 38274 }, { "epoch": 95.6875, "grad_norm": 11.0, "learning_rate": 9.738192643917066e-08, "loss": 1.5518, "step": 38275 }, { "epoch": 95.69, "grad_norm": 11.4375, "learning_rate": 9.726923592740855e-08, "loss": 1.419, "step": 38276 }, { "epoch": 95.6925, "grad_norm": 7.875, "learning_rate": 9.71566103375532e-08, "loss": 1.2969, "step": 38277 }, { "epoch": 95.695, "grad_norm": 9.3125, "learning_rate": 9.704404967034398e-08, "loss": 1.51, "step": 38278 }, { "epoch": 95.6975, "grad_norm": 8.0625, "learning_rate": 9.693155392651588e-08, "loss": 1.2343, "step": 38279 }, { "epoch": 95.7, "grad_norm": 5.09375, "learning_rate": 9.68191231068083e-08, "loss": 1.3191, "step": 38280 }, { "epoch": 95.7025, "grad_norm": 8.75, "learning_rate": 9.670675721195955e-08, "loss": 1.26, "step": 38281 }, { "epoch": 95.705, "grad_norm": 7.3125, "learning_rate": 9.659445624270458e-08, "loss": 1.4253, "step": 38282 }, { "epoch": 95.7075, "grad_norm": 7.5625, "learning_rate": 9.648222019977838e-08, "loss": 1.2603, "step": 38283 }, { "epoch": 95.71, "grad_norm": 8.625, "learning_rate": 9.637004908392033e-08, "loss": 1.239, "step": 38284 }, { "epoch": 95.7125, "grad_norm": 7.25, "learning_rate": 9.625794289586321e-08, "loss": 1.3288, "step": 38285 }, { "epoch": 95.715, "grad_norm": 9.0625, "learning_rate": 9.614590163634307e-08, "loss": 1.2959, "step": 38286 }, { "epoch": 95.7175, "grad_norm": 8.0625, "learning_rate": 9.60339253060938e-08, "loss": 1.3331, "step": 38287 }, { "epoch": 95.72, "grad_norm": 7.71875, "learning_rate": 9.59220139058492e-08, "loss": 1.1058, "step": 38288 }, { "epoch": 95.7225, "grad_norm": 6.6875, "learning_rate": 9.581016743634542e-08, "loss": 1.3505, "step": 38289 }, { "epoch": 95.725, "grad_norm": 8.125, "learning_rate": 9.569838589831181e-08, "loss": 1.33, "step": 38290 }, { "epoch": 95.7275, "grad_norm": 6.96875, "learning_rate": 9.558666929248339e-08, "loss": 1.4092, "step": 38291 }, { "epoch": 95.73, "grad_norm": 7.5625, "learning_rate": 9.547501761959288e-08, "loss": 1.4525, "step": 38292 }, { "epoch": 95.7325, "grad_norm": 6.25, "learning_rate": 9.536343088037303e-08, "loss": 1.4182, "step": 38293 }, { "epoch": 95.735, "grad_norm": 6.0625, "learning_rate": 9.525190907555326e-08, "loss": 1.483, "step": 38294 }, { "epoch": 95.7375, "grad_norm": 7.65625, "learning_rate": 9.51404522058641e-08, "loss": 1.3124, "step": 38295 }, { "epoch": 95.74, "grad_norm": 6.15625, "learning_rate": 9.50290602720394e-08, "loss": 1.2991, "step": 38296 }, { "epoch": 95.7425, "grad_norm": 6.5, "learning_rate": 9.491773327480747e-08, "loss": 1.3547, "step": 38297 }, { "epoch": 95.745, "grad_norm": 9.75, "learning_rate": 9.480647121489883e-08, "loss": 1.3949, "step": 38298 }, { "epoch": 95.7475, "grad_norm": 6.71875, "learning_rate": 9.46952740930418e-08, "loss": 1.2117, "step": 38299 }, { "epoch": 95.75, "grad_norm": 7.84375, "learning_rate": 9.45841419099669e-08, "loss": 1.4142, "step": 38300 }, { "epoch": 95.7525, "grad_norm": 8.25, "learning_rate": 9.447307466640132e-08, "loss": 1.1519, "step": 38301 }, { "epoch": 95.755, "grad_norm": 10.4375, "learning_rate": 9.436207236307448e-08, "loss": 1.3011, "step": 38302 }, { "epoch": 95.7575, "grad_norm": 6.5625, "learning_rate": 9.425113500071359e-08, "loss": 1.2639, "step": 38303 }, { "epoch": 95.76, "grad_norm": 6.9375, "learning_rate": 9.414026258004583e-08, "loss": 1.4698, "step": 38304 }, { "epoch": 95.7625, "grad_norm": 8.9375, "learning_rate": 9.402945510179729e-08, "loss": 1.2815, "step": 38305 }, { "epoch": 95.765, "grad_norm": 9.75, "learning_rate": 9.391871256669516e-08, "loss": 1.3176, "step": 38306 }, { "epoch": 95.7675, "grad_norm": 4.875, "learning_rate": 9.380803497546664e-08, "loss": 1.3267, "step": 38307 }, { "epoch": 95.77, "grad_norm": 7.0, "learning_rate": 9.369742232883561e-08, "loss": 1.2244, "step": 38308 }, { "epoch": 95.7725, "grad_norm": 5.78125, "learning_rate": 9.358687462752814e-08, "loss": 1.2033, "step": 38309 }, { "epoch": 95.775, "grad_norm": 8.8125, "learning_rate": 9.34763918722681e-08, "loss": 1.2568, "step": 38310 }, { "epoch": 95.7775, "grad_norm": 11.3125, "learning_rate": 9.336597406378044e-08, "loss": 1.2787, "step": 38311 }, { "epoch": 95.78, "grad_norm": 6.09375, "learning_rate": 9.32556212027902e-08, "loss": 1.3206, "step": 38312 }, { "epoch": 95.7825, "grad_norm": 6.59375, "learning_rate": 9.314533329001895e-08, "loss": 1.3134, "step": 38313 }, { "epoch": 95.785, "grad_norm": 6.46875, "learning_rate": 9.30351103261895e-08, "loss": 1.2404, "step": 38314 }, { "epoch": 95.7875, "grad_norm": 5.34375, "learning_rate": 9.29249523120257e-08, "loss": 1.2233, "step": 38315 }, { "epoch": 95.79, "grad_norm": 9.625, "learning_rate": 9.281485924825029e-08, "loss": 1.3404, "step": 38316 }, { "epoch": 95.7925, "grad_norm": 11.8125, "learning_rate": 9.270483113558381e-08, "loss": 1.4042, "step": 38317 }, { "epoch": 95.795, "grad_norm": 7.875, "learning_rate": 9.259486797474792e-08, "loss": 1.2766, "step": 38318 }, { "epoch": 95.7975, "grad_norm": 8.6875, "learning_rate": 9.248496976646426e-08, "loss": 1.3318, "step": 38319 }, { "epoch": 95.8, "grad_norm": 10.6875, "learning_rate": 9.237513651145224e-08, "loss": 1.2647, "step": 38320 }, { "epoch": 95.8025, "grad_norm": 10.9375, "learning_rate": 9.226536821043242e-08, "loss": 1.3091, "step": 38321 }, { "epoch": 95.805, "grad_norm": 8.1875, "learning_rate": 9.215566486412309e-08, "loss": 1.2846, "step": 38322 }, { "epoch": 95.8075, "grad_norm": 9.3125, "learning_rate": 9.20460264732459e-08, "loss": 1.4007, "step": 38323 }, { "epoch": 95.81, "grad_norm": 9.8125, "learning_rate": 9.193645303851917e-08, "loss": 1.2972, "step": 38324 }, { "epoch": 95.8125, "grad_norm": 9.5, "learning_rate": 9.1826944560659e-08, "loss": 1.3061, "step": 38325 }, { "epoch": 95.815, "grad_norm": 7.90625, "learning_rate": 9.17175010403859e-08, "loss": 1.225, "step": 38326 }, { "epoch": 95.8175, "grad_norm": 10.5625, "learning_rate": 9.160812247841599e-08, "loss": 1.334, "step": 38327 }, { "epoch": 95.82, "grad_norm": 6.15625, "learning_rate": 9.149880887546759e-08, "loss": 1.234, "step": 38328 }, { "epoch": 95.8225, "grad_norm": 7.15625, "learning_rate": 9.138956023225565e-08, "loss": 1.2936, "step": 38329 }, { "epoch": 95.825, "grad_norm": 6.0, "learning_rate": 9.128037654949629e-08, "loss": 1.3137, "step": 38330 }, { "epoch": 95.8275, "grad_norm": 6.84375, "learning_rate": 9.117125782790781e-08, "loss": 1.1167, "step": 38331 }, { "epoch": 95.83, "grad_norm": 8.875, "learning_rate": 9.10622040682041e-08, "loss": 1.3211, "step": 38332 }, { "epoch": 95.8325, "grad_norm": 7.3125, "learning_rate": 9.095321527109902e-08, "loss": 1.4134, "step": 38333 }, { "epoch": 95.835, "grad_norm": 6.3125, "learning_rate": 9.084429143730866e-08, "loss": 1.4824, "step": 38334 }, { "epoch": 95.8375, "grad_norm": 8.5625, "learning_rate": 9.07354325675458e-08, "loss": 1.1455, "step": 38335 }, { "epoch": 95.84, "grad_norm": 10.75, "learning_rate": 9.062663866252541e-08, "loss": 1.3422, "step": 38336 }, { "epoch": 95.8425, "grad_norm": 8.75, "learning_rate": 9.051790972296027e-08, "loss": 1.349, "step": 38337 }, { "epoch": 95.845, "grad_norm": 9.375, "learning_rate": 9.040924574956422e-08, "loss": 1.4148, "step": 38338 }, { "epoch": 95.8475, "grad_norm": 5.78125, "learning_rate": 9.030064674304784e-08, "loss": 1.3748, "step": 38339 }, { "epoch": 95.85, "grad_norm": 13.25, "learning_rate": 9.019211270412275e-08, "loss": 1.3895, "step": 38340 }, { "epoch": 95.8525, "grad_norm": 6.09375, "learning_rate": 9.008364363350287e-08, "loss": 1.1602, "step": 38341 }, { "epoch": 95.855, "grad_norm": 6.125, "learning_rate": 8.997523953189757e-08, "loss": 1.2298, "step": 38342 }, { "epoch": 95.8575, "grad_norm": 6.71875, "learning_rate": 8.986690040001855e-08, "loss": 1.33, "step": 38343 }, { "epoch": 95.86, "grad_norm": 8.0625, "learning_rate": 8.975862623857523e-08, "loss": 1.3163, "step": 38344 }, { "epoch": 95.8625, "grad_norm": 9.75, "learning_rate": 8.965041704827704e-08, "loss": 1.3847, "step": 38345 }, { "epoch": 95.865, "grad_norm": 8.8125, "learning_rate": 8.954227282983452e-08, "loss": 1.3439, "step": 38346 }, { "epoch": 95.8675, "grad_norm": 7.75, "learning_rate": 8.94341935839571e-08, "loss": 1.4123, "step": 38347 }, { "epoch": 95.87, "grad_norm": 7.625, "learning_rate": 8.932617931135202e-08, "loss": 1.3776, "step": 38348 }, { "epoch": 95.8725, "grad_norm": 7.375, "learning_rate": 8.921823001272756e-08, "loss": 1.4352, "step": 38349 }, { "epoch": 95.875, "grad_norm": 8.0, "learning_rate": 8.911034568879207e-08, "loss": 1.346, "step": 38350 }, { "epoch": 95.8775, "grad_norm": 5.09375, "learning_rate": 8.900252634025275e-08, "loss": 1.123, "step": 38351 }, { "epoch": 95.88, "grad_norm": 7.65625, "learning_rate": 8.889477196781571e-08, "loss": 1.5047, "step": 38352 }, { "epoch": 95.8825, "grad_norm": 9.125, "learning_rate": 8.878708257218815e-08, "loss": 1.3074, "step": 38353 }, { "epoch": 95.885, "grad_norm": 7.59375, "learning_rate": 8.86794581540762e-08, "loss": 1.299, "step": 38354 }, { "epoch": 95.8875, "grad_norm": 9.3125, "learning_rate": 8.857189871418481e-08, "loss": 1.3478, "step": 38355 }, { "epoch": 95.89, "grad_norm": 6.84375, "learning_rate": 8.8464404253219e-08, "loss": 1.3531, "step": 38356 }, { "epoch": 95.8925, "grad_norm": 8.6875, "learning_rate": 8.835697477188487e-08, "loss": 1.4396, "step": 38357 }, { "epoch": 95.895, "grad_norm": 7.40625, "learning_rate": 8.824961027088519e-08, "loss": 1.3318, "step": 38358 }, { "epoch": 95.8975, "grad_norm": 12.9375, "learning_rate": 8.814231075092383e-08, "loss": 1.3276, "step": 38359 }, { "epoch": 95.9, "grad_norm": 33.0, "learning_rate": 8.80350762127058e-08, "loss": 1.6101, "step": 38360 }, { "epoch": 95.9025, "grad_norm": 6.1875, "learning_rate": 8.792790665693274e-08, "loss": 1.3144, "step": 38361 }, { "epoch": 95.905, "grad_norm": 7.5, "learning_rate": 8.782080208430744e-08, "loss": 1.404, "step": 38362 }, { "epoch": 95.9075, "grad_norm": 7.28125, "learning_rate": 8.771376249553265e-08, "loss": 1.3573, "step": 38363 }, { "epoch": 95.91, "grad_norm": 8.75, "learning_rate": 8.760678789131005e-08, "loss": 1.2229, "step": 38364 }, { "epoch": 95.9125, "grad_norm": 9.25, "learning_rate": 8.74998782723402e-08, "loss": 1.3276, "step": 38365 }, { "epoch": 95.915, "grad_norm": 6.0, "learning_rate": 8.739303363932473e-08, "loss": 1.3235, "step": 38366 }, { "epoch": 95.9175, "grad_norm": 6.0, "learning_rate": 8.728625399296309e-08, "loss": 1.3376, "step": 38367 }, { "epoch": 95.92, "grad_norm": 5.5625, "learning_rate": 8.717953933395695e-08, "loss": 1.461, "step": 38368 }, { "epoch": 95.9225, "grad_norm": 9.375, "learning_rate": 8.707288966300576e-08, "loss": 1.2398, "step": 38369 }, { "epoch": 95.925, "grad_norm": 5.25, "learning_rate": 8.696630498080672e-08, "loss": 1.3866, "step": 38370 }, { "epoch": 95.9275, "grad_norm": 7.59375, "learning_rate": 8.685978528805927e-08, "loss": 1.5015, "step": 38371 }, { "epoch": 95.93, "grad_norm": 8.5625, "learning_rate": 8.675333058546397e-08, "loss": 1.332, "step": 38372 }, { "epoch": 95.9325, "grad_norm": 15.75, "learning_rate": 8.664694087371695e-08, "loss": 1.2856, "step": 38373 }, { "epoch": 95.935, "grad_norm": 9.4375, "learning_rate": 8.65406161535154e-08, "loss": 1.3864, "step": 38374 }, { "epoch": 95.9375, "grad_norm": 8.0, "learning_rate": 8.643435642555653e-08, "loss": 1.2599, "step": 38375 }, { "epoch": 95.94, "grad_norm": 6.90625, "learning_rate": 8.63281616905376e-08, "loss": 1.396, "step": 38376 }, { "epoch": 95.9425, "grad_norm": 14.375, "learning_rate": 8.622203194915468e-08, "loss": 1.2513, "step": 38377 }, { "epoch": 95.945, "grad_norm": 10.6875, "learning_rate": 8.61159672021028e-08, "loss": 1.3668, "step": 38378 }, { "epoch": 95.9475, "grad_norm": 8.75, "learning_rate": 8.600996745007916e-08, "loss": 1.3588, "step": 38379 }, { "epoch": 95.95, "grad_norm": 7.1875, "learning_rate": 8.590403269377656e-08, "loss": 1.4735, "step": 38380 }, { "epoch": 95.9525, "grad_norm": 7.78125, "learning_rate": 8.57981629338911e-08, "loss": 1.3396, "step": 38381 }, { "epoch": 95.955, "grad_norm": 7.53125, "learning_rate": 8.569235817111443e-08, "loss": 1.3843, "step": 38382 }, { "epoch": 95.9575, "grad_norm": 6.46875, "learning_rate": 8.558661840614379e-08, "loss": 1.1862, "step": 38383 }, { "epoch": 95.96, "grad_norm": 9.3125, "learning_rate": 8.548094363966974e-08, "loss": 1.3093, "step": 38384 }, { "epoch": 95.9625, "grad_norm": 11.625, "learning_rate": 8.537533387238616e-08, "loss": 1.3053, "step": 38385 }, { "epoch": 95.965, "grad_norm": 6.0625, "learning_rate": 8.526978910498363e-08, "loss": 1.2718, "step": 38386 }, { "epoch": 95.9675, "grad_norm": 7.28125, "learning_rate": 8.516430933815712e-08, "loss": 1.3033, "step": 38387 }, { "epoch": 95.97, "grad_norm": 7.9375, "learning_rate": 8.505889457259719e-08, "loss": 1.331, "step": 38388 }, { "epoch": 95.9725, "grad_norm": 7.5, "learning_rate": 8.495354480899331e-08, "loss": 1.3536, "step": 38389 }, { "epoch": 95.975, "grad_norm": 9.375, "learning_rate": 8.484826004803714e-08, "loss": 1.4014, "step": 38390 }, { "epoch": 95.9775, "grad_norm": 8.5625, "learning_rate": 8.474304029041924e-08, "loss": 1.3942, "step": 38391 }, { "epoch": 95.98, "grad_norm": 9.75, "learning_rate": 8.463788553683017e-08, "loss": 1.3082, "step": 38392 }, { "epoch": 95.9825, "grad_norm": 6.46875, "learning_rate": 8.453279578795603e-08, "loss": 1.4234, "step": 38393 }, { "epoch": 95.985, "grad_norm": 7.71875, "learning_rate": 8.442777104449073e-08, "loss": 1.5464, "step": 38394 }, { "epoch": 95.9875, "grad_norm": 8.375, "learning_rate": 8.432281130711817e-08, "loss": 1.3137, "step": 38395 }, { "epoch": 95.99, "grad_norm": 10.5625, "learning_rate": 8.421791657652889e-08, "loss": 1.3701, "step": 38396 }, { "epoch": 95.9925, "grad_norm": 10.625, "learning_rate": 8.411308685341013e-08, "loss": 1.1688, "step": 38397 }, { "epoch": 95.995, "grad_norm": 7.1875, "learning_rate": 8.400832213845023e-08, "loss": 1.2241, "step": 38398 }, { "epoch": 95.9975, "grad_norm": 7.53125, "learning_rate": 8.390362243233419e-08, "loss": 1.4355, "step": 38399 }, { "epoch": 96.0, "grad_norm": 7.65625, "learning_rate": 8.379898773574924e-08, "loss": 1.2549, "step": 38400 }, { "epoch": 96.0025, "grad_norm": 11.1875, "learning_rate": 8.369441804938038e-08, "loss": 1.4036, "step": 38401 }, { "epoch": 96.005, "grad_norm": 7.65625, "learning_rate": 8.358991337391598e-08, "loss": 1.4574, "step": 38402 }, { "epoch": 96.0075, "grad_norm": 10.0625, "learning_rate": 8.348547371003768e-08, "loss": 1.3855, "step": 38403 }, { "epoch": 96.01, "grad_norm": 8.625, "learning_rate": 8.338109905843273e-08, "loss": 1.2737, "step": 38404 }, { "epoch": 96.0125, "grad_norm": 6.21875, "learning_rate": 8.32767894197839e-08, "loss": 1.5648, "step": 38405 }, { "epoch": 96.015, "grad_norm": 8.0625, "learning_rate": 8.31725447947751e-08, "loss": 1.3809, "step": 38406 }, { "epoch": 96.0175, "grad_norm": 10.25, "learning_rate": 8.306836518409134e-08, "loss": 1.5145, "step": 38407 }, { "epoch": 96.02, "grad_norm": 8.4375, "learning_rate": 8.296425058841429e-08, "loss": 1.2981, "step": 38408 }, { "epoch": 96.0225, "grad_norm": 6.1875, "learning_rate": 8.286020100842674e-08, "loss": 1.5245, "step": 38409 }, { "epoch": 96.025, "grad_norm": 5.65625, "learning_rate": 8.275621644481036e-08, "loss": 1.2828, "step": 38410 }, { "epoch": 96.0275, "grad_norm": 8.0625, "learning_rate": 8.265229689824794e-08, "loss": 1.4437, "step": 38411 }, { "epoch": 96.03, "grad_norm": 9.625, "learning_rate": 8.254844236941895e-08, "loss": 1.2836, "step": 38412 }, { "epoch": 96.0325, "grad_norm": 5.84375, "learning_rate": 8.244465285900727e-08, "loss": 1.46, "step": 38413 }, { "epoch": 96.035, "grad_norm": 10.375, "learning_rate": 8.234092836769126e-08, "loss": 1.2307, "step": 38414 }, { "epoch": 96.0375, "grad_norm": 9.625, "learning_rate": 8.223726889615035e-08, "loss": 1.317, "step": 38415 }, { "epoch": 96.04, "grad_norm": 7.75, "learning_rate": 8.213367444506515e-08, "loss": 1.2628, "step": 38416 }, { "epoch": 96.0425, "grad_norm": 6.40625, "learning_rate": 8.203014501511508e-08, "loss": 1.2023, "step": 38417 }, { "epoch": 96.045, "grad_norm": 6.0625, "learning_rate": 8.192668060697851e-08, "loss": 1.3886, "step": 38418 }, { "epoch": 96.0475, "grad_norm": 8.0625, "learning_rate": 8.182328122133265e-08, "loss": 1.3276, "step": 38419 }, { "epoch": 96.05, "grad_norm": 8.8125, "learning_rate": 8.171994685885698e-08, "loss": 1.479, "step": 38420 }, { "epoch": 96.0525, "grad_norm": 9.0, "learning_rate": 8.161667752022984e-08, "loss": 1.3339, "step": 38421 }, { "epoch": 96.055, "grad_norm": 11.0625, "learning_rate": 8.151347320612624e-08, "loss": 1.353, "step": 38422 }, { "epoch": 96.0575, "grad_norm": 6.15625, "learning_rate": 8.141033391722231e-08, "loss": 1.3314, "step": 38423 }, { "epoch": 96.06, "grad_norm": 10.875, "learning_rate": 8.130725965419528e-08, "loss": 1.3729, "step": 38424 }, { "epoch": 96.0625, "grad_norm": 6.875, "learning_rate": 8.12042504177224e-08, "loss": 1.1986, "step": 38425 }, { "epoch": 96.065, "grad_norm": 8.8125, "learning_rate": 8.110130620847645e-08, "loss": 1.3306, "step": 38426 }, { "epoch": 96.0675, "grad_norm": 9.0625, "learning_rate": 8.099842702713356e-08, "loss": 1.2772, "step": 38427 }, { "epoch": 96.07, "grad_norm": 11.25, "learning_rate": 8.089561287436876e-08, "loss": 1.4348, "step": 38428 }, { "epoch": 96.0725, "grad_norm": 9.0, "learning_rate": 8.079286375085482e-08, "loss": 1.3744, "step": 38429 }, { "epoch": 96.075, "grad_norm": 7.09375, "learning_rate": 8.069017965726566e-08, "loss": 1.3857, "step": 38430 }, { "epoch": 96.0775, "grad_norm": 10.625, "learning_rate": 8.058756059427519e-08, "loss": 1.4571, "step": 38431 }, { "epoch": 96.08, "grad_norm": 6.09375, "learning_rate": 8.04850065625551e-08, "loss": 1.5365, "step": 38432 }, { "epoch": 96.0825, "grad_norm": 5.375, "learning_rate": 8.038251756277926e-08, "loss": 1.3658, "step": 38433 }, { "epoch": 96.085, "grad_norm": 7.125, "learning_rate": 8.028009359561828e-08, "loss": 1.4871, "step": 38434 }, { "epoch": 96.0875, "grad_norm": 7.0625, "learning_rate": 8.017773466174383e-08, "loss": 1.1955, "step": 38435 }, { "epoch": 96.09, "grad_norm": 6.84375, "learning_rate": 8.00754407618276e-08, "loss": 1.2872, "step": 38436 }, { "epoch": 96.0925, "grad_norm": 5.65625, "learning_rate": 7.997321189653905e-08, "loss": 1.2426, "step": 38437 }, { "epoch": 96.095, "grad_norm": 7.125, "learning_rate": 7.987104806654989e-08, "loss": 1.0945, "step": 38438 }, { "epoch": 96.0975, "grad_norm": 7.71875, "learning_rate": 7.976894927252844e-08, "loss": 1.2966, "step": 38439 }, { "epoch": 96.1, "grad_norm": 22.0, "learning_rate": 7.966691551514527e-08, "loss": 1.3264, "step": 38440 }, { "epoch": 96.1025, "grad_norm": 7.90625, "learning_rate": 7.956494679506876e-08, "loss": 1.2731, "step": 38441 }, { "epoch": 96.105, "grad_norm": 7.375, "learning_rate": 7.946304311296726e-08, "loss": 1.2929, "step": 38442 }, { "epoch": 96.1075, "grad_norm": 7.3125, "learning_rate": 7.936120446951023e-08, "loss": 1.2467, "step": 38443 }, { "epoch": 96.11, "grad_norm": 7.78125, "learning_rate": 7.92594308653627e-08, "loss": 1.2852, "step": 38444 }, { "epoch": 96.1125, "grad_norm": 8.0625, "learning_rate": 7.915772230119411e-08, "loss": 1.2272, "step": 38445 }, { "epoch": 96.115, "grad_norm": 9.3125, "learning_rate": 7.905607877766952e-08, "loss": 1.3033, "step": 38446 }, { "epoch": 96.1175, "grad_norm": 8.6875, "learning_rate": 7.895450029545726e-08, "loss": 1.176, "step": 38447 }, { "epoch": 96.12, "grad_norm": 8.1875, "learning_rate": 7.885298685522235e-08, "loss": 1.326, "step": 38448 }, { "epoch": 96.1225, "grad_norm": 5.5625, "learning_rate": 7.875153845762873e-08, "loss": 1.2913, "step": 38449 }, { "epoch": 96.125, "grad_norm": 8.5625, "learning_rate": 7.865015510334473e-08, "loss": 1.2891, "step": 38450 }, { "epoch": 96.1275, "grad_norm": 8.125, "learning_rate": 7.854883679303204e-08, "loss": 1.3763, "step": 38451 }, { "epoch": 96.13, "grad_norm": 6.21875, "learning_rate": 7.844758352735793e-08, "loss": 1.4617, "step": 38452 }, { "epoch": 96.1325, "grad_norm": 6.21875, "learning_rate": 7.834639530698296e-08, "loss": 1.3913, "step": 38453 }, { "epoch": 96.135, "grad_norm": 6.15625, "learning_rate": 7.824527213257105e-08, "loss": 1.274, "step": 38454 }, { "epoch": 96.1375, "grad_norm": 8.75, "learning_rate": 7.814421400478833e-08, "loss": 1.298, "step": 38455 }, { "epoch": 96.14, "grad_norm": 7.53125, "learning_rate": 7.804322092429429e-08, "loss": 1.3009, "step": 38456 }, { "epoch": 96.1425, "grad_norm": 6.28125, "learning_rate": 7.79422928917517e-08, "loss": 1.5673, "step": 38457 }, { "epoch": 96.145, "grad_norm": 15.125, "learning_rate": 7.784142990782229e-08, "loss": 1.1685, "step": 38458 }, { "epoch": 96.1475, "grad_norm": 7.90625, "learning_rate": 7.774063197316661e-08, "loss": 1.4652, "step": 38459 }, { "epoch": 96.15, "grad_norm": 7.90625, "learning_rate": 7.763989908844749e-08, "loss": 1.3098, "step": 38460 }, { "epoch": 96.1525, "grad_norm": 9.4375, "learning_rate": 7.753923125432328e-08, "loss": 1.2645, "step": 38461 }, { "epoch": 96.155, "grad_norm": 4.84375, "learning_rate": 7.743862847145566e-08, "loss": 1.3036, "step": 38462 }, { "epoch": 96.1575, "grad_norm": 9.0625, "learning_rate": 7.733809074050303e-08, "loss": 1.3187, "step": 38463 }, { "epoch": 96.16, "grad_norm": 7.1875, "learning_rate": 7.723761806212371e-08, "loss": 1.658, "step": 38464 }, { "epoch": 96.1625, "grad_norm": 6.78125, "learning_rate": 7.713721043697831e-08, "loss": 1.2609, "step": 38465 }, { "epoch": 96.165, "grad_norm": 7.40625, "learning_rate": 7.703686786572518e-08, "loss": 1.352, "step": 38466 }, { "epoch": 96.1675, "grad_norm": 8.6875, "learning_rate": 7.693659034902046e-08, "loss": 1.5203, "step": 38467 }, { "epoch": 96.17, "grad_norm": 7.84375, "learning_rate": 7.683637788752252e-08, "loss": 1.3288, "step": 38468 }, { "epoch": 96.1725, "grad_norm": 10.125, "learning_rate": 7.673623048188749e-08, "loss": 1.2224, "step": 38469 }, { "epoch": 96.175, "grad_norm": 8.5, "learning_rate": 7.663614813277265e-08, "loss": 1.3459, "step": 38470 }, { "epoch": 96.1775, "grad_norm": 7.46875, "learning_rate": 7.653613084083522e-08, "loss": 1.3228, "step": 38471 }, { "epoch": 96.18, "grad_norm": 7.40625, "learning_rate": 7.643617860672914e-08, "loss": 1.4343, "step": 38472 }, { "epoch": 96.1825, "grad_norm": 8.5625, "learning_rate": 7.633629143111055e-08, "loss": 1.2379, "step": 38473 }, { "epoch": 96.185, "grad_norm": 6.625, "learning_rate": 7.623646931463336e-08, "loss": 1.3725, "step": 38474 }, { "epoch": 96.1875, "grad_norm": 6.78125, "learning_rate": 7.613671225795371e-08, "loss": 1.4187, "step": 38475 }, { "epoch": 96.19, "grad_norm": 7.03125, "learning_rate": 7.603702026172443e-08, "loss": 1.2411, "step": 38476 }, { "epoch": 96.1925, "grad_norm": 9.25, "learning_rate": 7.593739332659944e-08, "loss": 1.3517, "step": 38477 }, { "epoch": 96.195, "grad_norm": 7.71875, "learning_rate": 7.583783145323153e-08, "loss": 1.3341, "step": 38478 }, { "epoch": 96.1975, "grad_norm": 5.78125, "learning_rate": 7.573833464227354e-08, "loss": 1.1802, "step": 38479 }, { "epoch": 96.2, "grad_norm": 5.8125, "learning_rate": 7.563890289437825e-08, "loss": 1.3509, "step": 38480 }, { "epoch": 96.2025, "grad_norm": 8.5, "learning_rate": 7.553953621019628e-08, "loss": 1.2433, "step": 38481 }, { "epoch": 96.205, "grad_norm": 6.5, "learning_rate": 7.544023459038041e-08, "loss": 1.2038, "step": 38482 }, { "epoch": 96.2075, "grad_norm": 6.09375, "learning_rate": 7.534099803558125e-08, "loss": 1.3044, "step": 38483 }, { "epoch": 96.21, "grad_norm": 8.3125, "learning_rate": 7.52418265464494e-08, "loss": 1.4552, "step": 38484 }, { "epoch": 96.2125, "grad_norm": 6.3125, "learning_rate": 7.514272012363543e-08, "loss": 1.2736, "step": 38485 }, { "epoch": 96.215, "grad_norm": 8.9375, "learning_rate": 7.504367876778772e-08, "loss": 1.402, "step": 38486 }, { "epoch": 96.2175, "grad_norm": 7.0, "learning_rate": 7.494470247955688e-08, "loss": 1.3519, "step": 38487 }, { "epoch": 96.22, "grad_norm": 7.75, "learning_rate": 7.484579125959124e-08, "loss": 1.3754, "step": 38488 }, { "epoch": 96.2225, "grad_norm": 8.25, "learning_rate": 7.47469451085392e-08, "loss": 1.2994, "step": 38489 }, { "epoch": 96.225, "grad_norm": 7.09375, "learning_rate": 7.464816402705022e-08, "loss": 1.2823, "step": 38490 }, { "epoch": 96.2275, "grad_norm": 5.9375, "learning_rate": 7.454944801577047e-08, "loss": 1.3906, "step": 38491 }, { "epoch": 96.23, "grad_norm": 12.25, "learning_rate": 7.445079707534608e-08, "loss": 1.5169, "step": 38492 }, { "epoch": 96.2325, "grad_norm": 7.5625, "learning_rate": 7.435221120642655e-08, "loss": 1.2257, "step": 38493 }, { "epoch": 96.235, "grad_norm": 7.84375, "learning_rate": 7.425369040965468e-08, "loss": 1.4547, "step": 38494 }, { "epoch": 96.2375, "grad_norm": 9.125, "learning_rate": 7.415523468567998e-08, "loss": 1.3889, "step": 38495 }, { "epoch": 96.24, "grad_norm": 7.625, "learning_rate": 7.405684403514635e-08, "loss": 1.3808, "step": 38496 }, { "epoch": 96.2425, "grad_norm": 6.28125, "learning_rate": 7.395851845869884e-08, "loss": 1.1765, "step": 38497 }, { "epoch": 96.245, "grad_norm": 7.46875, "learning_rate": 7.386025795698138e-08, "loss": 1.2274, "step": 38498 }, { "epoch": 96.2475, "grad_norm": 9.5625, "learning_rate": 7.3762062530639e-08, "loss": 1.4164, "step": 38499 }, { "epoch": 96.25, "grad_norm": 5.75, "learning_rate": 7.366393218031564e-08, "loss": 1.3079, "step": 38500 }, { "epoch": 96.2525, "grad_norm": 7.875, "learning_rate": 7.356586690665413e-08, "loss": 1.3751, "step": 38501 }, { "epoch": 96.255, "grad_norm": 9.125, "learning_rate": 7.346786671029837e-08, "loss": 1.2933, "step": 38502 }, { "epoch": 96.2575, "grad_norm": 5.84375, "learning_rate": 7.336993159188898e-08, "loss": 1.3541, "step": 38503 }, { "epoch": 96.26, "grad_norm": 8.1875, "learning_rate": 7.327206155206879e-08, "loss": 1.2231, "step": 38504 }, { "epoch": 96.2625, "grad_norm": 9.5625, "learning_rate": 7.317425659148059e-08, "loss": 1.2965, "step": 38505 }, { "epoch": 96.265, "grad_norm": 6.6875, "learning_rate": 7.307651671076499e-08, "loss": 1.2763, "step": 38506 }, { "epoch": 96.2675, "grad_norm": 9.375, "learning_rate": 7.297884191056148e-08, "loss": 1.3194, "step": 38507 }, { "epoch": 96.27, "grad_norm": 5.75, "learning_rate": 7.288123219151178e-08, "loss": 1.1518, "step": 38508 }, { "epoch": 96.2725, "grad_norm": 8.5, "learning_rate": 7.278368755425536e-08, "loss": 1.3583, "step": 38509 }, { "epoch": 96.275, "grad_norm": 7.5, "learning_rate": 7.268620799943171e-08, "loss": 1.3015, "step": 38510 }, { "epoch": 96.2775, "grad_norm": 8.5, "learning_rate": 7.258879352768033e-08, "loss": 1.3617, "step": 38511 }, { "epoch": 96.28, "grad_norm": 6.8125, "learning_rate": 7.24914441396396e-08, "loss": 1.481, "step": 38512 }, { "epoch": 96.2825, "grad_norm": 8.8125, "learning_rate": 7.239415983594788e-08, "loss": 1.2815, "step": 38513 }, { "epoch": 96.285, "grad_norm": 6.375, "learning_rate": 7.229694061724246e-08, "loss": 1.3863, "step": 38514 }, { "epoch": 96.2875, "grad_norm": 6.71875, "learning_rate": 7.21997864841606e-08, "loss": 1.3201, "step": 38515 }, { "epoch": 96.29, "grad_norm": 5.78125, "learning_rate": 7.210269743734067e-08, "loss": 1.2308, "step": 38516 }, { "epoch": 96.2925, "grad_norm": 10.0625, "learning_rate": 7.200567347741771e-08, "loss": 1.504, "step": 38517 }, { "epoch": 96.295, "grad_norm": 7.375, "learning_rate": 7.190871460502791e-08, "loss": 1.2955, "step": 38518 }, { "epoch": 96.2975, "grad_norm": 8.4375, "learning_rate": 7.18118208208074e-08, "loss": 1.3797, "step": 38519 }, { "epoch": 96.3, "grad_norm": 5.90625, "learning_rate": 7.171499212539124e-08, "loss": 1.2191, "step": 38520 }, { "epoch": 96.3025, "grad_norm": 6.71875, "learning_rate": 7.161822851941558e-08, "loss": 1.2996, "step": 38521 }, { "epoch": 96.305, "grad_norm": 6.03125, "learning_rate": 7.152153000351103e-08, "loss": 1.1936, "step": 38522 }, { "epoch": 96.3075, "grad_norm": 7.375, "learning_rate": 7.142489657831597e-08, "loss": 1.3501, "step": 38523 }, { "epoch": 96.31, "grad_norm": 9.0625, "learning_rate": 7.132832824446101e-08, "loss": 1.2614, "step": 38524 }, { "epoch": 96.3125, "grad_norm": 8.3125, "learning_rate": 7.123182500258119e-08, "loss": 1.2695, "step": 38525 }, { "epoch": 96.315, "grad_norm": 7.625, "learning_rate": 7.113538685330712e-08, "loss": 1.3451, "step": 38526 }, { "epoch": 96.3175, "grad_norm": 6.8125, "learning_rate": 7.103901379727274e-08, "loss": 1.2648, "step": 38527 }, { "epoch": 96.32, "grad_norm": 4.71875, "learning_rate": 7.094270583510976e-08, "loss": 1.4557, "step": 38528 }, { "epoch": 96.3225, "grad_norm": 7.1875, "learning_rate": 7.084646296744769e-08, "loss": 1.464, "step": 38529 }, { "epoch": 96.325, "grad_norm": 5.0, "learning_rate": 7.075028519491934e-08, "loss": 1.3714, "step": 38530 }, { "epoch": 96.3275, "grad_norm": 7.84375, "learning_rate": 7.065417251815531e-08, "loss": 1.2773, "step": 38531 }, { "epoch": 96.33, "grad_norm": 6.125, "learning_rate": 7.055812493778513e-08, "loss": 1.2916, "step": 38532 }, { "epoch": 96.3325, "grad_norm": 8.0, "learning_rate": 7.046214245443827e-08, "loss": 1.3635, "step": 38533 }, { "epoch": 96.335, "grad_norm": 8.3125, "learning_rate": 7.036622506874313e-08, "loss": 1.3111, "step": 38534 }, { "epoch": 96.3375, "grad_norm": 8.5625, "learning_rate": 7.027037278133141e-08, "loss": 1.4124, "step": 38535 }, { "epoch": 96.34, "grad_norm": 7.65625, "learning_rate": 7.01745855928293e-08, "loss": 1.1816, "step": 38536 }, { "epoch": 96.3425, "grad_norm": 6.125, "learning_rate": 7.007886350386517e-08, "loss": 1.3825, "step": 38537 }, { "epoch": 96.345, "grad_norm": 8.6875, "learning_rate": 6.998320651506629e-08, "loss": 1.4512, "step": 38538 }, { "epoch": 96.3475, "grad_norm": 6.75, "learning_rate": 6.988761462705995e-08, "loss": 1.3635, "step": 38539 }, { "epoch": 96.35, "grad_norm": 10.9375, "learning_rate": 6.979208784047454e-08, "loss": 1.3312, "step": 38540 }, { "epoch": 96.3525, "grad_norm": 14.4375, "learning_rate": 6.969662615593286e-08, "loss": 1.2761, "step": 38541 }, { "epoch": 96.355, "grad_norm": 7.96875, "learning_rate": 6.960122957406446e-08, "loss": 1.3308, "step": 38542 }, { "epoch": 96.3575, "grad_norm": 10.625, "learning_rate": 6.950589809549213e-08, "loss": 1.5021, "step": 38543 }, { "epoch": 96.36, "grad_norm": 6.59375, "learning_rate": 6.941063172084094e-08, "loss": 1.0695, "step": 38544 }, { "epoch": 96.3625, "grad_norm": 8.5625, "learning_rate": 6.931543045073708e-08, "loss": 1.2649, "step": 38545 }, { "epoch": 96.365, "grad_norm": 6.65625, "learning_rate": 6.922029428580334e-08, "loss": 1.3275, "step": 38546 }, { "epoch": 96.3675, "grad_norm": 11.375, "learning_rate": 6.912522322666482e-08, "loss": 1.309, "step": 38547 }, { "epoch": 96.37, "grad_norm": 9.0625, "learning_rate": 6.903021727394321e-08, "loss": 1.3548, "step": 38548 }, { "epoch": 96.3725, "grad_norm": 7.125, "learning_rate": 6.893527642826248e-08, "loss": 1.1922, "step": 38549 }, { "epoch": 96.375, "grad_norm": 6.84375, "learning_rate": 6.884040069024434e-08, "loss": 1.3818, "step": 38550 }, { "epoch": 96.3775, "grad_norm": 7.84375, "learning_rate": 6.874559006051163e-08, "loss": 1.3187, "step": 38551 }, { "epoch": 96.38, "grad_norm": 6.96875, "learning_rate": 6.865084453968495e-08, "loss": 1.2979, "step": 38552 }, { "epoch": 96.3825, "grad_norm": 7.21875, "learning_rate": 6.855616412838717e-08, "loss": 1.2771, "step": 38553 }, { "epoch": 96.385, "grad_norm": 12.0625, "learning_rate": 6.846154882723555e-08, "loss": 1.4004, "step": 38554 }, { "epoch": 96.3875, "grad_norm": 6.90625, "learning_rate": 6.836699863685403e-08, "loss": 1.3384, "step": 38555 }, { "epoch": 96.39, "grad_norm": 5.4375, "learning_rate": 6.827251355786102e-08, "loss": 1.2427, "step": 38556 }, { "epoch": 96.3925, "grad_norm": 9.9375, "learning_rate": 6.817809359087491e-08, "loss": 1.3901, "step": 38557 }, { "epoch": 96.395, "grad_norm": 8.375, "learning_rate": 6.808373873651741e-08, "loss": 1.1855, "step": 38558 }, { "epoch": 96.3975, "grad_norm": 48.5, "learning_rate": 6.79894489954036e-08, "loss": 1.2483, "step": 38559 }, { "epoch": 96.4, "grad_norm": 7.8125, "learning_rate": 6.78952243681541e-08, "loss": 1.2443, "step": 38560 }, { "epoch": 96.4025, "grad_norm": 7.03125, "learning_rate": 6.780106485538729e-08, "loss": 1.3445, "step": 38561 }, { "epoch": 96.405, "grad_norm": 6.8125, "learning_rate": 6.770697045771823e-08, "loss": 1.3517, "step": 38562 }, { "epoch": 96.4075, "grad_norm": 7.21875, "learning_rate": 6.761294117576423e-08, "loss": 1.2904, "step": 38563 }, { "epoch": 96.41, "grad_norm": 9.3125, "learning_rate": 6.751897701014365e-08, "loss": 1.4546, "step": 38564 }, { "epoch": 96.4125, "grad_norm": 7.34375, "learning_rate": 6.742507796147046e-08, "loss": 1.1864, "step": 38565 }, { "epoch": 96.415, "grad_norm": 7.28125, "learning_rate": 6.733124403036084e-08, "loss": 1.2712, "step": 38566 }, { "epoch": 96.4175, "grad_norm": 6.625, "learning_rate": 6.723747521742985e-08, "loss": 1.2437, "step": 38567 }, { "epoch": 96.42, "grad_norm": 8.1875, "learning_rate": 6.714377152329366e-08, "loss": 1.4176, "step": 38568 }, { "epoch": 96.4225, "grad_norm": 10.8125, "learning_rate": 6.705013294856399e-08, "loss": 1.1755, "step": 38569 }, { "epoch": 96.425, "grad_norm": 7.03125, "learning_rate": 6.695655949385704e-08, "loss": 1.2682, "step": 38570 }, { "epoch": 96.4275, "grad_norm": 6.0625, "learning_rate": 6.686305115978564e-08, "loss": 1.3248, "step": 38571 }, { "epoch": 96.43, "grad_norm": 7.0, "learning_rate": 6.676960794696153e-08, "loss": 1.2481, "step": 38572 }, { "epoch": 96.4325, "grad_norm": 7.75, "learning_rate": 6.667622985599975e-08, "loss": 1.4247, "step": 38573 }, { "epoch": 96.435, "grad_norm": 6.90625, "learning_rate": 6.658291688750984e-08, "loss": 1.3941, "step": 38574 }, { "epoch": 96.4375, "grad_norm": 6.15625, "learning_rate": 6.648966904210463e-08, "loss": 1.3487, "step": 38575 }, { "epoch": 96.44, "grad_norm": 9.25, "learning_rate": 6.639648632039697e-08, "loss": 1.3067, "step": 38576 }, { "epoch": 96.4425, "grad_norm": 11.6875, "learning_rate": 6.630336872299636e-08, "loss": 1.3148, "step": 38577 }, { "epoch": 96.445, "grad_norm": 10.375, "learning_rate": 6.621031625051233e-08, "loss": 1.1297, "step": 38578 }, { "epoch": 96.4475, "grad_norm": 6.625, "learning_rate": 6.611732890355549e-08, "loss": 1.3923, "step": 38579 }, { "epoch": 96.45, "grad_norm": 6.46875, "learning_rate": 6.602440668273758e-08, "loss": 1.4574, "step": 38580 }, { "epoch": 96.4525, "grad_norm": 6.4375, "learning_rate": 6.593154958866588e-08, "loss": 1.4408, "step": 38581 }, { "epoch": 96.455, "grad_norm": 18.0, "learning_rate": 6.58387576219477e-08, "loss": 1.2512, "step": 38582 }, { "epoch": 96.4575, "grad_norm": 5.15625, "learning_rate": 6.574603078319475e-08, "loss": 1.2613, "step": 38583 }, { "epoch": 96.46, "grad_norm": 7.8125, "learning_rate": 6.565336907301212e-08, "loss": 1.2475, "step": 38584 }, { "epoch": 96.4625, "grad_norm": 8.0625, "learning_rate": 6.556077249200931e-08, "loss": 1.3669, "step": 38585 }, { "epoch": 96.465, "grad_norm": 7.875, "learning_rate": 6.546824104079253e-08, "loss": 1.2853, "step": 38586 }, { "epoch": 96.4675, "grad_norm": 9.375, "learning_rate": 6.537577471996792e-08, "loss": 1.2809, "step": 38587 }, { "epoch": 96.47, "grad_norm": 8.375, "learning_rate": 6.52833735301428e-08, "loss": 1.2812, "step": 38588 }, { "epoch": 96.4725, "grad_norm": 8.1875, "learning_rate": 6.519103747192224e-08, "loss": 1.4093, "step": 38589 }, { "epoch": 96.475, "grad_norm": 8.3125, "learning_rate": 6.509876654591018e-08, "loss": 1.3931, "step": 38590 }, { "epoch": 96.4775, "grad_norm": 7.125, "learning_rate": 6.500656075271505e-08, "loss": 1.4642, "step": 38591 }, { "epoch": 96.48, "grad_norm": 7.8125, "learning_rate": 6.491442009293858e-08, "loss": 1.2732, "step": 38592 }, { "epoch": 96.4825, "grad_norm": 10.3125, "learning_rate": 6.482234456718583e-08, "loss": 1.5362, "step": 38593 }, { "epoch": 96.485, "grad_norm": 8.0625, "learning_rate": 6.473033417605856e-08, "loss": 1.3249, "step": 38594 }, { "epoch": 96.4875, "grad_norm": 7.03125, "learning_rate": 6.463838892016406e-08, "loss": 1.2535, "step": 38595 }, { "epoch": 96.49, "grad_norm": 7.1875, "learning_rate": 6.454650880010182e-08, "loss": 1.3669, "step": 38596 }, { "epoch": 96.4925, "grad_norm": 6.90625, "learning_rate": 6.445469381647474e-08, "loss": 1.4308, "step": 38597 }, { "epoch": 96.495, "grad_norm": 7.34375, "learning_rate": 6.436294396988451e-08, "loss": 1.3286, "step": 38598 }, { "epoch": 96.4975, "grad_norm": 16.5, "learning_rate": 6.427125926093403e-08, "loss": 1.2967, "step": 38599 }, { "epoch": 96.5, "grad_norm": 10.0625, "learning_rate": 6.417963969022389e-08, "loss": 1.4495, "step": 38600 }, { "epoch": 96.5025, "grad_norm": 8.0, "learning_rate": 6.408808525835364e-08, "loss": 1.2101, "step": 38601 }, { "epoch": 96.505, "grad_norm": 10.0, "learning_rate": 6.39965959659239e-08, "loss": 1.3142, "step": 38602 }, { "epoch": 96.5075, "grad_norm": 7.375, "learning_rate": 6.39051718135364e-08, "loss": 1.3616, "step": 38603 }, { "epoch": 96.51, "grad_norm": 5.625, "learning_rate": 6.381381280178733e-08, "loss": 1.2178, "step": 38604 }, { "epoch": 96.5125, "grad_norm": 8.1875, "learning_rate": 6.372251893127845e-08, "loss": 1.3455, "step": 38605 }, { "epoch": 96.515, "grad_norm": 12.4375, "learning_rate": 6.363129020260705e-08, "loss": 1.3689, "step": 38606 }, { "epoch": 96.5175, "grad_norm": 7.90625, "learning_rate": 6.354012661637044e-08, "loss": 1.3207, "step": 38607 }, { "epoch": 96.52, "grad_norm": 8.3125, "learning_rate": 6.344902817316811e-08, "loss": 1.1625, "step": 38608 }, { "epoch": 96.5225, "grad_norm": 7.0, "learning_rate": 6.335799487359517e-08, "loss": 1.3097, "step": 38609 }, { "epoch": 96.525, "grad_norm": 7.75, "learning_rate": 6.326702671825114e-08, "loss": 1.2893, "step": 38610 }, { "epoch": 96.5275, "grad_norm": 6.53125, "learning_rate": 6.317612370772997e-08, "loss": 1.3279, "step": 38611 }, { "epoch": 96.53, "grad_norm": 8.125, "learning_rate": 6.308528584262896e-08, "loss": 1.2822, "step": 38612 }, { "epoch": 96.5325, "grad_norm": 8.6875, "learning_rate": 6.29945131235421e-08, "loss": 1.3543, "step": 38613 }, { "epoch": 96.535, "grad_norm": 6.53125, "learning_rate": 6.290380555106557e-08, "loss": 1.3052, "step": 38614 }, { "epoch": 96.5375, "grad_norm": 7.0, "learning_rate": 6.281316312579555e-08, "loss": 1.233, "step": 38615 }, { "epoch": 96.54, "grad_norm": 7.8125, "learning_rate": 6.27225858483238e-08, "loss": 1.433, "step": 38616 }, { "epoch": 96.5425, "grad_norm": 5.75, "learning_rate": 6.263207371924429e-08, "loss": 1.0854, "step": 38617 }, { "epoch": 96.545, "grad_norm": 5.125, "learning_rate": 6.254162673915209e-08, "loss": 1.2491, "step": 38618 }, { "epoch": 96.5475, "grad_norm": 5.6875, "learning_rate": 6.245124490864007e-08, "loss": 1.267, "step": 38619 }, { "epoch": 96.55, "grad_norm": 8.6875, "learning_rate": 6.236092822829887e-08, "loss": 1.368, "step": 38620 }, { "epoch": 96.5525, "grad_norm": 8.25, "learning_rate": 6.227067669872245e-08, "loss": 1.2292, "step": 38621 }, { "epoch": 96.555, "grad_norm": 9.6875, "learning_rate": 6.218049032050145e-08, "loss": 1.4575, "step": 38622 }, { "epoch": 96.5575, "grad_norm": 6.84375, "learning_rate": 6.209036909422761e-08, "loss": 1.1063, "step": 38623 }, { "epoch": 96.56, "grad_norm": 8.3125, "learning_rate": 6.200031302049048e-08, "loss": 1.2667, "step": 38624 }, { "epoch": 96.5625, "grad_norm": 14.1875, "learning_rate": 6.19103220998829e-08, "loss": 1.3219, "step": 38625 }, { "epoch": 96.565, "grad_norm": 8.875, "learning_rate": 6.182039633299331e-08, "loss": 1.36, "step": 38626 }, { "epoch": 96.5675, "grad_norm": 7.59375, "learning_rate": 6.173053572041122e-08, "loss": 1.2904, "step": 38627 }, { "epoch": 96.57, "grad_norm": 7.5, "learning_rate": 6.164074026272615e-08, "loss": 1.2767, "step": 38628 }, { "epoch": 96.5725, "grad_norm": 9.875, "learning_rate": 6.155100996052655e-08, "loss": 1.215, "step": 38629 }, { "epoch": 96.575, "grad_norm": 10.4375, "learning_rate": 6.146134481440192e-08, "loss": 1.3789, "step": 38630 }, { "epoch": 96.5775, "grad_norm": 8.3125, "learning_rate": 6.137174482493846e-08, "loss": 1.4021, "step": 38631 }, { "epoch": 96.58, "grad_norm": 8.25, "learning_rate": 6.12822099927235e-08, "loss": 1.2594, "step": 38632 }, { "epoch": 96.5825, "grad_norm": 10.5, "learning_rate": 6.119274031834432e-08, "loss": 1.4212, "step": 38633 }, { "epoch": 96.585, "grad_norm": 6.5, "learning_rate": 6.110333580238825e-08, "loss": 1.2997, "step": 38634 }, { "epoch": 96.5875, "grad_norm": 7.96875, "learning_rate": 6.101399644544148e-08, "loss": 1.3109, "step": 38635 }, { "epoch": 96.59, "grad_norm": 9.1875, "learning_rate": 6.092472224808909e-08, "loss": 1.3896, "step": 38636 }, { "epoch": 96.5925, "grad_norm": 7.09375, "learning_rate": 6.083551321091619e-08, "loss": 1.1471, "step": 38637 }, { "epoch": 96.595, "grad_norm": 9.4375, "learning_rate": 6.074636933450784e-08, "loss": 1.5799, "step": 38638 }, { "epoch": 96.5975, "grad_norm": 5.78125, "learning_rate": 6.065729061944914e-08, "loss": 1.2087, "step": 38639 }, { "epoch": 96.6, "grad_norm": 9.4375, "learning_rate": 6.056827706632185e-08, "loss": 1.4836, "step": 38640 }, { "epoch": 96.6025, "grad_norm": 12.6875, "learning_rate": 6.047932867571326e-08, "loss": 1.2435, "step": 38641 }, { "epoch": 96.605, "grad_norm": 8.3125, "learning_rate": 6.039044544820404e-08, "loss": 1.395, "step": 38642 }, { "epoch": 96.6075, "grad_norm": 10.875, "learning_rate": 6.030162738437594e-08, "loss": 1.426, "step": 38643 }, { "epoch": 96.61, "grad_norm": 8.0, "learning_rate": 6.021287448481294e-08, "loss": 1.3171, "step": 38644 }, { "epoch": 96.6125, "grad_norm": 11.5, "learning_rate": 6.01241867500979e-08, "loss": 1.3697, "step": 38645 }, { "epoch": 96.615, "grad_norm": 5.4375, "learning_rate": 6.003556418081036e-08, "loss": 1.3312, "step": 38646 }, { "epoch": 96.6175, "grad_norm": 7.25, "learning_rate": 5.994700677753095e-08, "loss": 1.2964, "step": 38647 }, { "epoch": 96.62, "grad_norm": 17.75, "learning_rate": 5.985851454084257e-08, "loss": 1.4715, "step": 38648 }, { "epoch": 96.6225, "grad_norm": 6.6875, "learning_rate": 5.977008747132251e-08, "loss": 1.2665, "step": 38649 }, { "epoch": 96.625, "grad_norm": 7.15625, "learning_rate": 5.968172556955365e-08, "loss": 1.2328, "step": 38650 }, { "epoch": 96.6275, "grad_norm": 9.3125, "learning_rate": 5.9593428836112185e-08, "loss": 1.3619, "step": 38651 }, { "epoch": 96.63, "grad_norm": 7.625, "learning_rate": 5.9505197271578776e-08, "loss": 1.2736, "step": 38652 }, { "epoch": 96.6325, "grad_norm": 7.71875, "learning_rate": 5.9417030876531834e-08, "loss": 1.2373, "step": 38653 }, { "epoch": 96.635, "grad_norm": 6.875, "learning_rate": 5.9328929651548684e-08, "loss": 1.3493, "step": 38654 }, { "epoch": 96.6375, "grad_norm": 10.375, "learning_rate": 5.924089359720775e-08, "loss": 1.318, "step": 38655 }, { "epoch": 96.64, "grad_norm": 6.09375, "learning_rate": 5.915292271408524e-08, "loss": 1.262, "step": 38656 }, { "epoch": 96.6425, "grad_norm": 7.96875, "learning_rate": 5.9065017002758464e-08, "loss": 1.3799, "step": 38657 }, { "epoch": 96.645, "grad_norm": 7.34375, "learning_rate": 5.897717646380363e-08, "loss": 1.3401, "step": 38658 }, { "epoch": 96.6475, "grad_norm": 9.75, "learning_rate": 5.888940109779695e-08, "loss": 1.1307, "step": 38659 }, { "epoch": 96.65, "grad_norm": 10.5, "learning_rate": 5.880169090531351e-08, "loss": 1.4406, "step": 38660 }, { "epoch": 96.6525, "grad_norm": 8.5625, "learning_rate": 5.8714045886928414e-08, "loss": 1.3119, "step": 38661 }, { "epoch": 96.655, "grad_norm": 8.0625, "learning_rate": 5.862646604321565e-08, "loss": 1.3798, "step": 38662 }, { "epoch": 96.6575, "grad_norm": 6.375, "learning_rate": 5.85389513747503e-08, "loss": 1.2873, "step": 38663 }, { "epoch": 96.66, "grad_norm": 5.84375, "learning_rate": 5.845150188210524e-08, "loss": 1.2103, "step": 38664 }, { "epoch": 96.6625, "grad_norm": 8.4375, "learning_rate": 5.836411756585447e-08, "loss": 1.3903, "step": 38665 }, { "epoch": 96.665, "grad_norm": 8.25, "learning_rate": 5.8276798426570856e-08, "loss": 1.3988, "step": 38666 }, { "epoch": 96.6675, "grad_norm": 10.5, "learning_rate": 5.818954446482728e-08, "loss": 1.2971, "step": 38667 }, { "epoch": 96.67, "grad_norm": 5.84375, "learning_rate": 5.8102355681194375e-08, "loss": 1.2812, "step": 38668 }, { "epoch": 96.6725, "grad_norm": 10.8125, "learning_rate": 5.8015232076245044e-08, "loss": 1.3703, "step": 38669 }, { "epoch": 96.675, "grad_norm": 10.3125, "learning_rate": 5.7928173650549924e-08, "loss": 1.3957, "step": 38670 }, { "epoch": 96.6775, "grad_norm": 7.125, "learning_rate": 5.784118040467968e-08, "loss": 1.3428, "step": 38671 }, { "epoch": 96.68, "grad_norm": 6.125, "learning_rate": 5.7754252339204955e-08, "loss": 1.3961, "step": 38672 }, { "epoch": 96.6825, "grad_norm": 6.40625, "learning_rate": 5.76673894546953e-08, "loss": 1.2564, "step": 38673 }, { "epoch": 96.685, "grad_norm": 6.65625, "learning_rate": 5.7580591751720264e-08, "loss": 1.35, "step": 38674 }, { "epoch": 96.6875, "grad_norm": 10.4375, "learning_rate": 5.749385923084938e-08, "loss": 1.2495, "step": 38675 }, { "epoch": 96.69, "grad_norm": 9.5625, "learning_rate": 5.7407191892651095e-08, "loss": 1.2312, "step": 38676 }, { "epoch": 96.6925, "grad_norm": 8.1875, "learning_rate": 5.7320589737692725e-08, "loss": 1.2215, "step": 38677 }, { "epoch": 96.695, "grad_norm": 59.75, "learning_rate": 5.72340527665427e-08, "loss": 1.4869, "step": 38678 }, { "epoch": 96.6975, "grad_norm": 7.5625, "learning_rate": 5.714758097976836e-08, "loss": 1.3519, "step": 38679 }, { "epoch": 96.7, "grad_norm": 6.34375, "learning_rate": 5.7061174377937015e-08, "loss": 1.3463, "step": 38680 }, { "epoch": 96.7025, "grad_norm": 7.875, "learning_rate": 5.697483296161488e-08, "loss": 1.2007, "step": 38681 }, { "epoch": 96.705, "grad_norm": 9.125, "learning_rate": 5.688855673136706e-08, "loss": 1.2017, "step": 38682 }, { "epoch": 96.7075, "grad_norm": 7.71875, "learning_rate": 5.6802345687760884e-08, "loss": 1.1965, "step": 38683 }, { "epoch": 96.71, "grad_norm": 11.375, "learning_rate": 5.671619983136034e-08, "loss": 1.2932, "step": 38684 }, { "epoch": 96.7125, "grad_norm": 7.21875, "learning_rate": 5.663011916272942e-08, "loss": 1.2536, "step": 38685 }, { "epoch": 96.715, "grad_norm": 5.96875, "learning_rate": 5.6544103682434345e-08, "loss": 1.1897, "step": 38686 }, { "epoch": 96.7175, "grad_norm": 5.78125, "learning_rate": 5.645815339103911e-08, "loss": 1.4497, "step": 38687 }, { "epoch": 96.72, "grad_norm": 7.71875, "learning_rate": 5.637226828910436e-08, "loss": 1.2454, "step": 38688 }, { "epoch": 96.7225, "grad_norm": 5.15625, "learning_rate": 5.628644837719632e-08, "loss": 1.2747, "step": 38689 }, { "epoch": 96.725, "grad_norm": 7.34375, "learning_rate": 5.620069365587677e-08, "loss": 1.5133, "step": 38690 }, { "epoch": 96.7275, "grad_norm": 8.25, "learning_rate": 5.6115004125707476e-08, "loss": 1.1915, "step": 38691 }, { "epoch": 96.73, "grad_norm": 6.59375, "learning_rate": 5.6029379787249094e-08, "loss": 1.1518, "step": 38692 }, { "epoch": 96.7325, "grad_norm": 7.9375, "learning_rate": 5.5943820641064515e-08, "loss": 1.3111, "step": 38693 }, { "epoch": 96.735, "grad_norm": 7.34375, "learning_rate": 5.5858326687715514e-08, "loss": 1.2073, "step": 38694 }, { "epoch": 96.7375, "grad_norm": 6.625, "learning_rate": 5.577289792776053e-08, "loss": 1.3247, "step": 38695 }, { "epoch": 96.74, "grad_norm": 9.4375, "learning_rate": 5.568753436176022e-08, "loss": 1.3144, "step": 38696 }, { "epoch": 96.7425, "grad_norm": 7.8125, "learning_rate": 5.5602235990275254e-08, "loss": 1.2108, "step": 38697 }, { "epoch": 96.745, "grad_norm": 13.25, "learning_rate": 5.5517002813862966e-08, "loss": 1.1841, "step": 38698 }, { "epoch": 96.7475, "grad_norm": 7.78125, "learning_rate": 5.543183483308401e-08, "loss": 1.3461, "step": 38699 }, { "epoch": 96.75, "grad_norm": 8.6875, "learning_rate": 5.534673204849572e-08, "loss": 1.2947, "step": 38700 }, { "epoch": 96.7525, "grad_norm": 7.375, "learning_rate": 5.526169446065654e-08, "loss": 1.2199, "step": 38701 }, { "epoch": 96.755, "grad_norm": 7.53125, "learning_rate": 5.517672207012381e-08, "loss": 1.4103, "step": 38702 }, { "epoch": 96.7575, "grad_norm": 10.3125, "learning_rate": 5.509181487745374e-08, "loss": 1.2444, "step": 38703 }, { "epoch": 96.76, "grad_norm": 11.625, "learning_rate": 5.5006972883204776e-08, "loss": 1.3534, "step": 38704 }, { "epoch": 96.7625, "grad_norm": 7.28125, "learning_rate": 5.4922196087932035e-08, "loss": 1.224, "step": 38705 }, { "epoch": 96.765, "grad_norm": 4.90625, "learning_rate": 5.4837484492190615e-08, "loss": 1.2443, "step": 38706 }, { "epoch": 96.7675, "grad_norm": 7.46875, "learning_rate": 5.475283809653675e-08, "loss": 1.2636, "step": 38707 }, { "epoch": 96.77, "grad_norm": 6.5, "learning_rate": 5.4668256901525554e-08, "loss": 1.3083, "step": 38708 }, { "epoch": 96.7725, "grad_norm": 8.3125, "learning_rate": 5.458374090771212e-08, "loss": 1.3036, "step": 38709 }, { "epoch": 96.775, "grad_norm": 7.40625, "learning_rate": 5.449929011564936e-08, "loss": 1.3327, "step": 38710 }, { "epoch": 96.7775, "grad_norm": 10.5625, "learning_rate": 5.4414904525890156e-08, "loss": 1.4159, "step": 38711 }, { "epoch": 96.78, "grad_norm": 6.53125, "learning_rate": 5.4330584138989615e-08, "loss": 1.3265, "step": 38712 }, { "epoch": 96.7825, "grad_norm": 7.0, "learning_rate": 5.4246328955499526e-08, "loss": 1.3389, "step": 38713 }, { "epoch": 96.785, "grad_norm": 8.1875, "learning_rate": 5.416213897597278e-08, "loss": 1.1855, "step": 38714 }, { "epoch": 96.7875, "grad_norm": 4.96875, "learning_rate": 5.407801420096115e-08, "loss": 1.2048, "step": 38715 }, { "epoch": 96.79, "grad_norm": 6.34375, "learning_rate": 5.399395463101531e-08, "loss": 1.2527, "step": 38716 }, { "epoch": 96.7925, "grad_norm": 5.625, "learning_rate": 5.3909960266688156e-08, "loss": 1.3762, "step": 38717 }, { "epoch": 96.795, "grad_norm": 5.8125, "learning_rate": 5.382603110852813e-08, "loss": 1.4036, "step": 38718 }, { "epoch": 96.7975, "grad_norm": 7.40625, "learning_rate": 5.374216715708702e-08, "loss": 1.4215, "step": 38719 }, { "epoch": 96.8, "grad_norm": 7.46875, "learning_rate": 5.365836841291439e-08, "loss": 1.3121, "step": 38720 }, { "epoch": 96.8025, "grad_norm": 7.53125, "learning_rate": 5.3574634876559785e-08, "loss": 1.3062, "step": 38721 }, { "epoch": 96.805, "grad_norm": 6.375, "learning_rate": 5.3490966548571665e-08, "loss": 1.172, "step": 38722 }, { "epoch": 96.8075, "grad_norm": 8.0625, "learning_rate": 5.340736342949848e-08, "loss": 1.2963, "step": 38723 }, { "epoch": 96.81, "grad_norm": 7.875, "learning_rate": 5.332382551988868e-08, "loss": 1.152, "step": 38724 }, { "epoch": 96.8125, "grad_norm": 6.53125, "learning_rate": 5.324035282029072e-08, "loss": 1.4106, "step": 38725 }, { "epoch": 96.815, "grad_norm": 9.6875, "learning_rate": 5.315694533124971e-08, "loss": 1.4, "step": 38726 }, { "epoch": 96.8175, "grad_norm": 6.34375, "learning_rate": 5.3073603053314105e-08, "loss": 1.3142, "step": 38727 }, { "epoch": 96.82, "grad_norm": 7.0625, "learning_rate": 5.2990325987029025e-08, "loss": 1.2865, "step": 38728 }, { "epoch": 96.8225, "grad_norm": 6.5, "learning_rate": 5.2907114132942915e-08, "loss": 1.3171, "step": 38729 }, { "epoch": 96.825, "grad_norm": 9.375, "learning_rate": 5.282396749159868e-08, "loss": 1.4147, "step": 38730 }, { "epoch": 96.8275, "grad_norm": 4.71875, "learning_rate": 5.274088606354255e-08, "loss": 1.0862, "step": 38731 }, { "epoch": 96.83, "grad_norm": 8.5, "learning_rate": 5.265786984931964e-08, "loss": 1.3679, "step": 38732 }, { "epoch": 96.8325, "grad_norm": 6.03125, "learning_rate": 5.2574918849473963e-08, "loss": 1.2567, "step": 38733 }, { "epoch": 96.835, "grad_norm": 4.8125, "learning_rate": 5.249203306454731e-08, "loss": 1.2902, "step": 38734 }, { "epoch": 96.8375, "grad_norm": 9.75, "learning_rate": 5.2409212495085905e-08, "loss": 1.3101, "step": 38735 }, { "epoch": 96.84, "grad_norm": 8.125, "learning_rate": 5.232645714163265e-08, "loss": 1.3427, "step": 38736 }, { "epoch": 96.8425, "grad_norm": 8.0, "learning_rate": 5.224376700472822e-08, "loss": 1.3247, "step": 38737 }, { "epoch": 96.845, "grad_norm": 8.75, "learning_rate": 5.216114208491441e-08, "loss": 1.4091, "step": 38738 }, { "epoch": 96.8475, "grad_norm": 6.03125, "learning_rate": 5.2078582382735224e-08, "loss": 1.3426, "step": 38739 }, { "epoch": 96.85, "grad_norm": 6.71875, "learning_rate": 5.199608789873134e-08, "loss": 1.2383, "step": 38740 }, { "epoch": 96.8525, "grad_norm": 6.96875, "learning_rate": 5.191365863344122e-08, "loss": 1.2196, "step": 38741 }, { "epoch": 96.855, "grad_norm": 8.6875, "learning_rate": 5.1831294587408876e-08, "loss": 1.1826, "step": 38742 }, { "epoch": 96.8575, "grad_norm": 7.34375, "learning_rate": 5.174899576117165e-08, "loss": 1.3701, "step": 38743 }, { "epoch": 96.86, "grad_norm": 7.78125, "learning_rate": 5.166676215526911e-08, "loss": 1.3343, "step": 38744 }, { "epoch": 96.8625, "grad_norm": 7.71875, "learning_rate": 5.158459377024194e-08, "loss": 1.3438, "step": 38745 }, { "epoch": 96.865, "grad_norm": 6.46875, "learning_rate": 5.1502490606628596e-08, "loss": 1.2725, "step": 38746 }, { "epoch": 96.8675, "grad_norm": 7.3125, "learning_rate": 5.142045266496642e-08, "loss": 1.2951, "step": 38747 }, { "epoch": 96.87, "grad_norm": 8.25, "learning_rate": 5.133847994579388e-08, "loss": 1.2455, "step": 38748 }, { "epoch": 96.8725, "grad_norm": 7.84375, "learning_rate": 5.125657244964721e-08, "loss": 1.5554, "step": 38749 }, { "epoch": 96.875, "grad_norm": 6.25, "learning_rate": 5.1174730177064866e-08, "loss": 1.3807, "step": 38750 }, { "epoch": 96.8775, "grad_norm": 7.6875, "learning_rate": 5.109295312858309e-08, "loss": 1.2287, "step": 38751 }, { "epoch": 96.88, "grad_norm": 8.5625, "learning_rate": 5.1011241304738115e-08, "loss": 1.3321, "step": 38752 }, { "epoch": 96.8825, "grad_norm": 6.40625, "learning_rate": 5.092959470606396e-08, "loss": 1.4182, "step": 38753 }, { "epoch": 96.885, "grad_norm": 7.09375, "learning_rate": 5.084801333309797e-08, "loss": 1.5483, "step": 38754 }, { "epoch": 96.8875, "grad_norm": 6.34375, "learning_rate": 5.0766497186375276e-08, "loss": 1.3031, "step": 38755 }, { "epoch": 96.89, "grad_norm": 9.9375, "learning_rate": 5.068504626642767e-08, "loss": 1.3795, "step": 38756 }, { "epoch": 96.8925, "grad_norm": 10.0, "learning_rate": 5.06036605737914e-08, "loss": 1.3165, "step": 38757 }, { "epoch": 96.895, "grad_norm": 7.65625, "learning_rate": 5.0522340108999366e-08, "loss": 1.2416, "step": 38758 }, { "epoch": 96.8975, "grad_norm": 5.96875, "learning_rate": 5.044108487258448e-08, "loss": 1.1722, "step": 38759 }, { "epoch": 96.9, "grad_norm": 9.8125, "learning_rate": 5.035989486508075e-08, "loss": 1.3949, "step": 38760 }, { "epoch": 96.9025, "grad_norm": 5.375, "learning_rate": 5.027877008701776e-08, "loss": 1.3658, "step": 38761 }, { "epoch": 96.905, "grad_norm": 9.6875, "learning_rate": 5.019771053893063e-08, "loss": 1.2178, "step": 38762 }, { "epoch": 96.9075, "grad_norm": 8.375, "learning_rate": 5.011671622134784e-08, "loss": 1.2715, "step": 38763 }, { "epoch": 96.91, "grad_norm": 6.90625, "learning_rate": 5.003578713480228e-08, "loss": 1.303, "step": 38764 }, { "epoch": 96.9125, "grad_norm": 8.125, "learning_rate": 4.995492327982354e-08, "loss": 1.1897, "step": 38765 }, { "epoch": 96.915, "grad_norm": 6.4375, "learning_rate": 4.9874124656943415e-08, "loss": 1.4195, "step": 38766 }, { "epoch": 96.9175, "grad_norm": 9.0625, "learning_rate": 4.9793391266689253e-08, "loss": 1.4124, "step": 38767 }, { "epoch": 96.92, "grad_norm": 5.90625, "learning_rate": 4.9712723109590636e-08, "loss": 1.247, "step": 38768 }, { "epoch": 96.9225, "grad_norm": 8.5, "learning_rate": 4.963212018617825e-08, "loss": 1.3841, "step": 38769 }, { "epoch": 96.925, "grad_norm": 6.09375, "learning_rate": 4.955158249697945e-08, "loss": 1.3526, "step": 38770 }, { "epoch": 96.9275, "grad_norm": 11.1875, "learning_rate": 4.94711100425227e-08, "loss": 1.3399, "step": 38771 }, { "epoch": 96.93, "grad_norm": 7.6875, "learning_rate": 4.939070282333425e-08, "loss": 1.2847, "step": 38772 }, { "epoch": 96.9325, "grad_norm": 7.6875, "learning_rate": 4.931036083994256e-08, "loss": 1.3654, "step": 38773 }, { "epoch": 96.935, "grad_norm": 6.21875, "learning_rate": 4.9230084092873885e-08, "loss": 1.2993, "step": 38774 }, { "epoch": 96.9375, "grad_norm": 5.21875, "learning_rate": 4.914987258265558e-08, "loss": 1.2624, "step": 38775 }, { "epoch": 96.94, "grad_norm": 7.09375, "learning_rate": 4.906972630981055e-08, "loss": 1.27, "step": 38776 }, { "epoch": 96.9425, "grad_norm": 8.4375, "learning_rate": 4.898964527486727e-08, "loss": 1.4026, "step": 38777 }, { "epoch": 96.945, "grad_norm": 6.59375, "learning_rate": 4.8909629478349764e-08, "loss": 1.2922, "step": 38778 }, { "epoch": 96.9475, "grad_norm": 6.34375, "learning_rate": 4.882967892078205e-08, "loss": 1.2372, "step": 38779 }, { "epoch": 96.95, "grad_norm": 8.625, "learning_rate": 4.874979360268928e-08, "loss": 1.2036, "step": 38780 }, { "epoch": 96.9525, "grad_norm": 6.15625, "learning_rate": 4.866997352459435e-08, "loss": 1.3388, "step": 38781 }, { "epoch": 96.955, "grad_norm": 8.0625, "learning_rate": 4.8590218687021294e-08, "loss": 1.4953, "step": 38782 }, { "epoch": 96.9575, "grad_norm": 9.4375, "learning_rate": 4.851052909049192e-08, "loss": 1.3275, "step": 38783 }, { "epoch": 96.96, "grad_norm": 13.1875, "learning_rate": 4.843090473552914e-08, "loss": 1.3992, "step": 38784 }, { "epoch": 96.9625, "grad_norm": 9.3125, "learning_rate": 4.8351345622655866e-08, "loss": 1.2365, "step": 38785 }, { "epoch": 96.965, "grad_norm": 4.375, "learning_rate": 4.82718517523928e-08, "loss": 1.2939, "step": 38786 }, { "epoch": 96.9675, "grad_norm": 6.0625, "learning_rate": 4.819242312526062e-08, "loss": 1.4169, "step": 38787 }, { "epoch": 96.97, "grad_norm": 9.6875, "learning_rate": 4.811305974178004e-08, "loss": 1.5725, "step": 38788 }, { "epoch": 96.9725, "grad_norm": 7.8125, "learning_rate": 4.803376160247286e-08, "loss": 1.1759, "step": 38789 }, { "epoch": 96.975, "grad_norm": 9.875, "learning_rate": 4.795452870785866e-08, "loss": 1.2345, "step": 38790 }, { "epoch": 96.9775, "grad_norm": 5.5625, "learning_rate": 4.787536105845592e-08, "loss": 1.2293, "step": 38791 }, { "epoch": 96.98, "grad_norm": 5.71875, "learning_rate": 4.779625865478421e-08, "loss": 1.2962, "step": 38792 }, { "epoch": 96.9825, "grad_norm": 7.34375, "learning_rate": 4.771722149736091e-08, "loss": 1.1852, "step": 38793 }, { "epoch": 96.985, "grad_norm": 8.25, "learning_rate": 4.763824958670671e-08, "loss": 1.3242, "step": 38794 }, { "epoch": 96.9875, "grad_norm": 16.125, "learning_rate": 4.755934292333675e-08, "loss": 1.2722, "step": 38795 }, { "epoch": 96.99, "grad_norm": 7.15625, "learning_rate": 4.748050150776951e-08, "loss": 1.3757, "step": 38796 }, { "epoch": 96.9925, "grad_norm": 5.5625, "learning_rate": 4.7401725340522344e-08, "loss": 1.278, "step": 38797 }, { "epoch": 96.995, "grad_norm": 6.625, "learning_rate": 4.7323014422111515e-08, "loss": 1.294, "step": 38798 }, { "epoch": 96.9975, "grad_norm": 7.34375, "learning_rate": 4.724436875305216e-08, "loss": 1.2219, "step": 38799 }, { "epoch": 97.0, "grad_norm": 8.625, "learning_rate": 4.716578833386054e-08, "loss": 1.2584, "step": 38800 }, { "epoch": 97.0025, "grad_norm": 5.1875, "learning_rate": 4.7087273165051796e-08, "loss": 1.2081, "step": 38801 }, { "epoch": 97.005, "grad_norm": 8.5625, "learning_rate": 4.7008823247139955e-08, "loss": 1.2623, "step": 38802 }, { "epoch": 97.0075, "grad_norm": 6.65625, "learning_rate": 4.693043858064017e-08, "loss": 1.1475, "step": 38803 }, { "epoch": 97.01, "grad_norm": 7.875, "learning_rate": 4.685211916606647e-08, "loss": 1.2428, "step": 38804 }, { "epoch": 97.0125, "grad_norm": 6.1875, "learning_rate": 4.6773865003932886e-08, "loss": 1.2909, "step": 38805 }, { "epoch": 97.015, "grad_norm": 7.09375, "learning_rate": 4.669567609475012e-08, "loss": 1.2165, "step": 38806 }, { "epoch": 97.0175, "grad_norm": 7.4375, "learning_rate": 4.661755243903332e-08, "loss": 1.434, "step": 38807 }, { "epoch": 97.02, "grad_norm": 7.40625, "learning_rate": 4.653949403729208e-08, "loss": 1.3047, "step": 38808 }, { "epoch": 97.0225, "grad_norm": 7.75, "learning_rate": 4.6461500890040425e-08, "loss": 1.2406, "step": 38809 }, { "epoch": 97.025, "grad_norm": 7.375, "learning_rate": 4.638357299778906e-08, "loss": 1.3994, "step": 38810 }, { "epoch": 97.0275, "grad_norm": 6.1875, "learning_rate": 4.630571036104869e-08, "loss": 1.3789, "step": 38811 }, { "epoch": 97.03, "grad_norm": 8.3125, "learning_rate": 4.6227912980330026e-08, "loss": 1.3082, "step": 38812 }, { "epoch": 97.0325, "grad_norm": 5.875, "learning_rate": 4.6150180856141533e-08, "loss": 1.2764, "step": 38813 }, { "epoch": 97.035, "grad_norm": 6.625, "learning_rate": 4.607251398899504e-08, "loss": 1.3196, "step": 38814 }, { "epoch": 97.0375, "grad_norm": 6.6875, "learning_rate": 4.5994912379399015e-08, "loss": 1.4531, "step": 38815 }, { "epoch": 97.04, "grad_norm": 8.0625, "learning_rate": 4.5917376027861945e-08, "loss": 1.4485, "step": 38816 }, { "epoch": 97.0425, "grad_norm": 10.1875, "learning_rate": 4.583990493489121e-08, "loss": 1.4281, "step": 38817 }, { "epoch": 97.045, "grad_norm": 12.3125, "learning_rate": 4.57624991009964e-08, "loss": 1.3243, "step": 38818 }, { "epoch": 97.0475, "grad_norm": 10.5, "learning_rate": 4.568515852668487e-08, "loss": 1.3498, "step": 38819 }, { "epoch": 97.05, "grad_norm": 6.71875, "learning_rate": 4.56078832124629e-08, "loss": 1.2299, "step": 38820 }, { "epoch": 97.0525, "grad_norm": 7.78125, "learning_rate": 4.553067315883675e-08, "loss": 1.3466, "step": 38821 }, { "epoch": 97.055, "grad_norm": 5.59375, "learning_rate": 4.5453528366313784e-08, "loss": 1.2767, "step": 38822 }, { "epoch": 97.0575, "grad_norm": 14.5625, "learning_rate": 4.537644883539805e-08, "loss": 1.4126, "step": 38823 }, { "epoch": 97.06, "grad_norm": 9.875, "learning_rate": 4.529943456659691e-08, "loss": 1.2302, "step": 38824 }, { "epoch": 97.0625, "grad_norm": 8.3125, "learning_rate": 4.522248556041331e-08, "loss": 1.166, "step": 38825 }, { "epoch": 97.065, "grad_norm": 9.625, "learning_rate": 4.5145601817354615e-08, "loss": 1.536, "step": 38826 }, { "epoch": 97.0675, "grad_norm": 6.59375, "learning_rate": 4.506878333792153e-08, "loss": 1.5197, "step": 38827 }, { "epoch": 97.07, "grad_norm": 7.71875, "learning_rate": 4.4992030122619216e-08, "loss": 1.3592, "step": 38828 }, { "epoch": 97.0725, "grad_norm": 7.96875, "learning_rate": 4.491534217195059e-08, "loss": 1.3712, "step": 38829 }, { "epoch": 97.075, "grad_norm": 7.78125, "learning_rate": 4.483871948641971e-08, "loss": 1.3756, "step": 38830 }, { "epoch": 97.0775, "grad_norm": 5.40625, "learning_rate": 4.4762162066527284e-08, "loss": 1.2906, "step": 38831 }, { "epoch": 97.08, "grad_norm": 6.53125, "learning_rate": 4.468566991277512e-08, "loss": 1.3103, "step": 38832 }, { "epoch": 97.0825, "grad_norm": 6.28125, "learning_rate": 4.460924302566505e-08, "loss": 1.1268, "step": 38833 }, { "epoch": 97.085, "grad_norm": 5.21875, "learning_rate": 4.45328814057e-08, "loss": 1.2191, "step": 38834 }, { "epoch": 97.0875, "grad_norm": 7.375, "learning_rate": 4.445658505337735e-08, "loss": 1.3301, "step": 38835 }, { "epoch": 97.09, "grad_norm": 6.59375, "learning_rate": 4.438035396920004e-08, "loss": 1.4374, "step": 38836 }, { "epoch": 97.0925, "grad_norm": 9.875, "learning_rate": 4.430418815366655e-08, "loss": 1.357, "step": 38837 }, { "epoch": 97.095, "grad_norm": 10.75, "learning_rate": 4.422808760727537e-08, "loss": 1.4954, "step": 38838 }, { "epoch": 97.0975, "grad_norm": 7.03125, "learning_rate": 4.415205233052833e-08, "loss": 1.3274, "step": 38839 }, { "epoch": 97.1, "grad_norm": 8.0, "learning_rate": 4.4076082323920576e-08, "loss": 1.2216, "step": 38840 }, { "epoch": 97.1025, "grad_norm": 7.90625, "learning_rate": 4.400017758795172e-08, "loss": 1.4701, "step": 38841 }, { "epoch": 97.105, "grad_norm": 10.0, "learning_rate": 4.392433812312025e-08, "loss": 1.3769, "step": 38842 }, { "epoch": 97.1075, "grad_norm": 6.78125, "learning_rate": 4.3848563929921316e-08, "loss": 1.2637, "step": 38843 }, { "epoch": 97.11, "grad_norm": 8.5625, "learning_rate": 4.377285500885231e-08, "loss": 1.3105, "step": 38844 }, { "epoch": 97.1125, "grad_norm": 8.3125, "learning_rate": 4.369721136041172e-08, "loss": 1.4502, "step": 38845 }, { "epoch": 97.115, "grad_norm": 12.0, "learning_rate": 4.3621632985092475e-08, "loss": 1.2647, "step": 38846 }, { "epoch": 97.1175, "grad_norm": 5.84375, "learning_rate": 4.354611988339197e-08, "loss": 1.4133, "step": 38847 }, { "epoch": 97.12, "grad_norm": 6.09375, "learning_rate": 4.347067205580424e-08, "loss": 1.2321, "step": 38848 }, { "epoch": 97.1225, "grad_norm": 7.65625, "learning_rate": 4.339528950282557e-08, "loss": 1.4431, "step": 38849 }, { "epoch": 97.125, "grad_norm": 4.84375, "learning_rate": 4.331997222494777e-08, "loss": 1.2756, "step": 38850 }, { "epoch": 97.1275, "grad_norm": 9.4375, "learning_rate": 4.3244720222666014e-08, "loss": 1.3829, "step": 38851 }, { "epoch": 97.13, "grad_norm": 5.78125, "learning_rate": 4.3169533496473237e-08, "loss": 1.1725, "step": 38852 }, { "epoch": 97.1325, "grad_norm": 6.28125, "learning_rate": 4.3094412046862377e-08, "loss": 1.2725, "step": 38853 }, { "epoch": 97.135, "grad_norm": 7.0, "learning_rate": 4.3019355874327486e-08, "loss": 1.2819, "step": 38854 }, { "epoch": 97.1375, "grad_norm": 7.0, "learning_rate": 4.294436497935706e-08, "loss": 1.408, "step": 38855 }, { "epoch": 97.14, "grad_norm": 8.25, "learning_rate": 4.2869439362446254e-08, "loss": 1.3605, "step": 38856 }, { "epoch": 97.1425, "grad_norm": 6.4375, "learning_rate": 4.2794579024085794e-08, "loss": 1.1788, "step": 38857 }, { "epoch": 97.145, "grad_norm": 6.34375, "learning_rate": 4.271978396476417e-08, "loss": 1.3838, "step": 38858 }, { "epoch": 97.1475, "grad_norm": 6.125, "learning_rate": 4.264505418497322e-08, "loss": 1.2932, "step": 38859 }, { "epoch": 97.15, "grad_norm": 9.25, "learning_rate": 4.257038968520366e-08, "loss": 1.4361, "step": 38860 }, { "epoch": 97.1525, "grad_norm": 6.53125, "learning_rate": 4.249579046594399e-08, "loss": 1.2559, "step": 38861 }, { "epoch": 97.155, "grad_norm": 7.90625, "learning_rate": 4.2421256527682695e-08, "loss": 1.3008, "step": 38862 }, { "epoch": 97.1575, "grad_norm": 8.5625, "learning_rate": 4.23467878709094e-08, "loss": 1.3878, "step": 38863 }, { "epoch": 97.16, "grad_norm": 8.375, "learning_rate": 4.2272384496112597e-08, "loss": 1.5827, "step": 38864 }, { "epoch": 97.1625, "grad_norm": 8.875, "learning_rate": 4.2198046403779666e-08, "loss": 1.2019, "step": 38865 }, { "epoch": 97.165, "grad_norm": 7.65625, "learning_rate": 4.21237735943969e-08, "loss": 1.4377, "step": 38866 }, { "epoch": 97.1675, "grad_norm": 7.40625, "learning_rate": 4.2049566068452786e-08, "loss": 1.4542, "step": 38867 }, { "epoch": 97.17, "grad_norm": 6.75, "learning_rate": 4.1975423826433605e-08, "loss": 1.3288, "step": 38868 }, { "epoch": 97.1725, "grad_norm": 6.25, "learning_rate": 4.190134686882452e-08, "loss": 1.3738, "step": 38869 }, { "epoch": 97.175, "grad_norm": 6.4375, "learning_rate": 4.1827335196111817e-08, "loss": 1.1802, "step": 38870 }, { "epoch": 97.1775, "grad_norm": 5.25, "learning_rate": 4.1753388808780656e-08, "loss": 1.4313, "step": 38871 }, { "epoch": 97.18, "grad_norm": 7.375, "learning_rate": 4.1679507707315106e-08, "loss": 1.3159, "step": 38872 }, { "epoch": 97.1825, "grad_norm": 14.875, "learning_rate": 4.160569189220143e-08, "loss": 1.3842, "step": 38873 }, { "epoch": 97.185, "grad_norm": 5.59375, "learning_rate": 4.153194136392147e-08, "loss": 1.3703, "step": 38874 }, { "epoch": 97.1875, "grad_norm": 8.5625, "learning_rate": 4.145825612295928e-08, "loss": 1.2729, "step": 38875 }, { "epoch": 97.19, "grad_norm": 10.0625, "learning_rate": 4.138463616979893e-08, "loss": 1.279, "step": 38876 }, { "epoch": 97.1925, "grad_norm": 8.875, "learning_rate": 4.1311081504922244e-08, "loss": 1.3832, "step": 38877 }, { "epoch": 97.195, "grad_norm": 7.625, "learning_rate": 4.1237592128811064e-08, "loss": 1.4434, "step": 38878 }, { "epoch": 97.1975, "grad_norm": 7.5, "learning_rate": 4.116416804194834e-08, "loss": 1.2348, "step": 38879 }, { "epoch": 97.2, "grad_norm": 10.0625, "learning_rate": 4.109080924481479e-08, "loss": 1.2642, "step": 38880 }, { "epoch": 97.2025, "grad_norm": 12.0625, "learning_rate": 4.101751573789114e-08, "loss": 1.4375, "step": 38881 }, { "epoch": 97.205, "grad_norm": 8.3125, "learning_rate": 4.094428752165813e-08, "loss": 1.2472, "step": 38882 }, { "epoch": 97.2075, "grad_norm": 5.71875, "learning_rate": 4.087112459659648e-08, "loss": 1.3419, "step": 38883 }, { "epoch": 97.21, "grad_norm": 7.78125, "learning_rate": 4.079802696318469e-08, "loss": 1.2788, "step": 38884 }, { "epoch": 97.2125, "grad_norm": 8.875, "learning_rate": 4.0724994621902383e-08, "loss": 1.3004, "step": 38885 }, { "epoch": 97.215, "grad_norm": 8.0, "learning_rate": 4.0652027573228056e-08, "loss": 1.4333, "step": 38886 }, { "epoch": 97.2175, "grad_norm": 7.5625, "learning_rate": 4.057912581764134e-08, "loss": 1.1997, "step": 38887 }, { "epoch": 97.22, "grad_norm": 5.59375, "learning_rate": 4.050628935561851e-08, "loss": 1.3839, "step": 38888 }, { "epoch": 97.2225, "grad_norm": 6.78125, "learning_rate": 4.0433518187638075e-08, "loss": 1.4027, "step": 38889 }, { "epoch": 97.225, "grad_norm": 7.15625, "learning_rate": 4.036081231417743e-08, "loss": 1.2725, "step": 38890 }, { "epoch": 97.2275, "grad_norm": 7.84375, "learning_rate": 4.028817173571287e-08, "loss": 1.258, "step": 38891 }, { "epoch": 97.23, "grad_norm": 10.0625, "learning_rate": 4.0215596452720665e-08, "loss": 1.355, "step": 38892 }, { "epoch": 97.2325, "grad_norm": 8.8125, "learning_rate": 4.0143086465674883e-08, "loss": 1.2414, "step": 38893 }, { "epoch": 97.235, "grad_norm": 7.5, "learning_rate": 4.007064177505404e-08, "loss": 1.4149, "step": 38894 }, { "epoch": 97.2375, "grad_norm": 7.90625, "learning_rate": 3.9998262381331086e-08, "loss": 1.3981, "step": 38895 }, { "epoch": 97.24, "grad_norm": 7.0, "learning_rate": 3.9925948284980086e-08, "loss": 1.3194, "step": 38896 }, { "epoch": 97.2425, "grad_norm": 9.0, "learning_rate": 3.9853699486477327e-08, "loss": 1.454, "step": 38897 }, { "epoch": 97.245, "grad_norm": 9.1875, "learning_rate": 3.978151598629354e-08, "loss": 1.2272, "step": 38898 }, { "epoch": 97.2475, "grad_norm": 8.1875, "learning_rate": 3.970939778490501e-08, "loss": 1.394, "step": 38899 }, { "epoch": 97.25, "grad_norm": 8.4375, "learning_rate": 3.963734488278248e-08, "loss": 1.2399, "step": 38900 }, { "epoch": 97.2525, "grad_norm": 6.65625, "learning_rate": 3.95653572803989e-08, "loss": 1.2052, "step": 38901 }, { "epoch": 97.255, "grad_norm": 8.0625, "learning_rate": 3.949343497822611e-08, "loss": 1.3523, "step": 38902 }, { "epoch": 97.2575, "grad_norm": 9.4375, "learning_rate": 3.942157797673596e-08, "loss": 1.4104, "step": 38903 }, { "epoch": 97.26, "grad_norm": 7.3125, "learning_rate": 3.934978627639918e-08, "loss": 1.5159, "step": 38904 }, { "epoch": 97.2625, "grad_norm": 5.84375, "learning_rate": 3.9278059877687624e-08, "loss": 1.4706, "step": 38905 }, { "epoch": 97.265, "grad_norm": 8.375, "learning_rate": 3.92063987810698e-08, "loss": 1.2501, "step": 38906 }, { "epoch": 97.2675, "grad_norm": 5.09375, "learning_rate": 3.913480298701533e-08, "loss": 1.3465, "step": 38907 }, { "epoch": 97.27, "grad_norm": 7.25, "learning_rate": 3.906327249599495e-08, "loss": 1.376, "step": 38908 }, { "epoch": 97.2725, "grad_norm": 7.46875, "learning_rate": 3.899180730847718e-08, "loss": 1.4327, "step": 38909 }, { "epoch": 97.275, "grad_norm": 5.8125, "learning_rate": 3.892040742493164e-08, "loss": 1.2852, "step": 38910 }, { "epoch": 97.2775, "grad_norm": 7.9375, "learning_rate": 3.884907284582462e-08, "loss": 1.3055, "step": 38911 }, { "epoch": 97.28, "grad_norm": 8.3125, "learning_rate": 3.877780357162353e-08, "loss": 1.2332, "step": 38912 }, { "epoch": 97.2825, "grad_norm": 7.65625, "learning_rate": 3.870659960279688e-08, "loss": 1.1902, "step": 38913 }, { "epoch": 97.285, "grad_norm": 7.21875, "learning_rate": 3.8635460939812075e-08, "loss": 1.2411, "step": 38914 }, { "epoch": 97.2875, "grad_norm": 9.375, "learning_rate": 3.85643875831343e-08, "loss": 1.384, "step": 38915 }, { "epoch": 97.29, "grad_norm": 5.46875, "learning_rate": 3.849337953322874e-08, "loss": 1.2941, "step": 38916 }, { "epoch": 97.2925, "grad_norm": 9.625, "learning_rate": 3.842243679056279e-08, "loss": 1.3505, "step": 38917 }, { "epoch": 97.295, "grad_norm": 7.75, "learning_rate": 3.835155935560053e-08, "loss": 1.2205, "step": 38918 }, { "epoch": 97.2975, "grad_norm": 8.625, "learning_rate": 3.828074722880715e-08, "loss": 1.3105, "step": 38919 }, { "epoch": 97.3, "grad_norm": 6.0625, "learning_rate": 3.82100004106456e-08, "loss": 1.2492, "step": 38920 }, { "epoch": 97.3025, "grad_norm": 8.3125, "learning_rate": 3.813931890158218e-08, "loss": 1.2318, "step": 38921 }, { "epoch": 97.305, "grad_norm": 6.84375, "learning_rate": 3.806870270207763e-08, "loss": 1.1728, "step": 38922 }, { "epoch": 97.3075, "grad_norm": 5.4375, "learning_rate": 3.799815181259603e-08, "loss": 1.2859, "step": 38923 }, { "epoch": 97.31, "grad_norm": 7.34375, "learning_rate": 3.792766623360034e-08, "loss": 1.4173, "step": 38924 }, { "epoch": 97.3125, "grad_norm": 7.1875, "learning_rate": 3.7857245965551294e-08, "loss": 1.2928, "step": 38925 }, { "epoch": 97.315, "grad_norm": 8.5625, "learning_rate": 3.778689100891297e-08, "loss": 1.3803, "step": 38926 }, { "epoch": 97.3175, "grad_norm": 8.3125, "learning_rate": 3.77166013641439e-08, "loss": 1.2973, "step": 38927 }, { "epoch": 97.32, "grad_norm": 5.59375, "learning_rate": 3.764637703170593e-08, "loss": 1.3349, "step": 38928 }, { "epoch": 97.3225, "grad_norm": 7.15625, "learning_rate": 3.7576218012060904e-08, "loss": 1.4044, "step": 38929 }, { "epoch": 97.325, "grad_norm": 6.75, "learning_rate": 3.750612430566625e-08, "loss": 1.4471, "step": 38930 }, { "epoch": 97.3275, "grad_norm": 8.3125, "learning_rate": 3.7436095912983805e-08, "loss": 1.3027, "step": 38931 }, { "epoch": 97.33, "grad_norm": 8.8125, "learning_rate": 3.736613283447099e-08, "loss": 1.3227, "step": 38932 }, { "epoch": 97.3325, "grad_norm": 6.4375, "learning_rate": 3.729623507058744e-08, "loss": 1.3319, "step": 38933 }, { "epoch": 97.335, "grad_norm": 9.1875, "learning_rate": 3.7226402621790556e-08, "loss": 1.3768, "step": 38934 }, { "epoch": 97.3375, "grad_norm": 6.125, "learning_rate": 3.715663548853887e-08, "loss": 1.4764, "step": 38935 }, { "epoch": 97.34, "grad_norm": 5.9375, "learning_rate": 3.7086933671290904e-08, "loss": 1.4105, "step": 38936 }, { "epoch": 97.3425, "grad_norm": 6.03125, "learning_rate": 3.701729717050073e-08, "loss": 1.3647, "step": 38937 }, { "epoch": 97.345, "grad_norm": 7.3125, "learning_rate": 3.694772598662688e-08, "loss": 1.2892, "step": 38938 }, { "epoch": 97.3475, "grad_norm": 7.71875, "learning_rate": 3.6878220120124543e-08, "loss": 1.3812, "step": 38939 }, { "epoch": 97.35, "grad_norm": 6.375, "learning_rate": 3.680877957145112e-08, "loss": 1.3708, "step": 38940 }, { "epoch": 97.3525, "grad_norm": 8.75, "learning_rate": 3.67394043410596e-08, "loss": 1.4466, "step": 38941 }, { "epoch": 97.355, "grad_norm": 4.75, "learning_rate": 3.667009442940628e-08, "loss": 1.2399, "step": 38942 }, { "epoch": 97.3575, "grad_norm": 7.28125, "learning_rate": 3.660084983694523e-08, "loss": 1.4048, "step": 38943 }, { "epoch": 97.36, "grad_norm": 9.0625, "learning_rate": 3.653167056413054e-08, "loss": 1.4093, "step": 38944 }, { "epoch": 97.3625, "grad_norm": 6.4375, "learning_rate": 3.646255661141518e-08, "loss": 1.3637, "step": 38945 }, { "epoch": 97.365, "grad_norm": 6.9375, "learning_rate": 3.639350797925212e-08, "loss": 1.373, "step": 38946 }, { "epoch": 97.3675, "grad_norm": 8.6875, "learning_rate": 3.632452466809433e-08, "loss": 1.3492, "step": 38947 }, { "epoch": 97.37, "grad_norm": 7.46875, "learning_rate": 3.6255606678394786e-08, "loss": 1.239, "step": 38948 }, { "epoch": 97.3725, "grad_norm": 6.71875, "learning_rate": 3.618675401060534e-08, "loss": 1.1567, "step": 38949 }, { "epoch": 97.375, "grad_norm": 7.8125, "learning_rate": 3.6117966665175644e-08, "loss": 1.441, "step": 38950 }, { "epoch": 97.3775, "grad_norm": 6.03125, "learning_rate": 3.604924464255866e-08, "loss": 1.1407, "step": 38951 }, { "epoch": 97.38, "grad_norm": 15.0625, "learning_rate": 3.598058794320402e-08, "loss": 1.2797, "step": 38952 }, { "epoch": 97.3825, "grad_norm": 7.78125, "learning_rate": 3.591199656756139e-08, "loss": 1.351, "step": 38953 }, { "epoch": 97.385, "grad_norm": 9.0625, "learning_rate": 3.58434705160815e-08, "loss": 1.3183, "step": 38954 }, { "epoch": 97.3875, "grad_norm": 8.75, "learning_rate": 3.577500978921289e-08, "loss": 1.3449, "step": 38955 }, { "epoch": 97.39, "grad_norm": 8.25, "learning_rate": 3.5706614387404084e-08, "loss": 1.2859, "step": 38956 }, { "epoch": 97.3925, "grad_norm": 4.65625, "learning_rate": 3.563828431110472e-08, "loss": 1.3888, "step": 38957 }, { "epoch": 97.395, "grad_norm": 10.25, "learning_rate": 3.557001956076111e-08, "loss": 1.3173, "step": 38958 }, { "epoch": 97.3975, "grad_norm": 10.1875, "learning_rate": 3.5501820136822904e-08, "loss": 1.3801, "step": 38959 }, { "epoch": 97.4, "grad_norm": 8.875, "learning_rate": 3.543368603973529e-08, "loss": 1.3232, "step": 38960 }, { "epoch": 97.4025, "grad_norm": 7.03125, "learning_rate": 3.5365617269945695e-08, "loss": 1.3881, "step": 38961 }, { "epoch": 97.405, "grad_norm": 7.53125, "learning_rate": 3.529761382790042e-08, "loss": 1.2455, "step": 38962 }, { "epoch": 97.4075, "grad_norm": 7.03125, "learning_rate": 3.5229675714044676e-08, "loss": 1.4615, "step": 38963 }, { "epoch": 97.41, "grad_norm": 9.0625, "learning_rate": 3.5161802928825874e-08, "loss": 1.4125, "step": 38964 }, { "epoch": 97.4125, "grad_norm": 6.5, "learning_rate": 3.509399547268588e-08, "loss": 1.2866, "step": 38965 }, { "epoch": 97.415, "grad_norm": 6.40625, "learning_rate": 3.502625334607212e-08, "loss": 1.1869, "step": 38966 }, { "epoch": 97.4175, "grad_norm": 8.0625, "learning_rate": 3.495857654942647e-08, "loss": 1.3219, "step": 38967 }, { "epoch": 97.42, "grad_norm": 8.8125, "learning_rate": 3.489096508319412e-08, "loss": 1.356, "step": 38968 }, { "epoch": 97.4225, "grad_norm": 7.84375, "learning_rate": 3.4823418947818044e-08, "loss": 1.3601, "step": 38969 }, { "epoch": 97.425, "grad_norm": 8.375, "learning_rate": 3.475593814374123e-08, "loss": 1.3921, "step": 38970 }, { "epoch": 97.4275, "grad_norm": 6.1875, "learning_rate": 3.468852267140443e-08, "loss": 1.3932, "step": 38971 }, { "epoch": 97.43, "grad_norm": 6.78125, "learning_rate": 3.4621172531251744e-08, "loss": 1.3881, "step": 38972 }, { "epoch": 97.4325, "grad_norm": 12.0, "learning_rate": 3.455388772372281e-08, "loss": 1.3691, "step": 38973 }, { "epoch": 97.435, "grad_norm": 5.625, "learning_rate": 3.448666824926172e-08, "loss": 1.2482, "step": 38974 }, { "epoch": 97.4375, "grad_norm": 10.4375, "learning_rate": 3.4419514108305905e-08, "loss": 1.4521, "step": 38975 }, { "epoch": 97.44, "grad_norm": 7.875, "learning_rate": 3.435242530129723e-08, "loss": 1.2645, "step": 38976 }, { "epoch": 97.4425, "grad_norm": 5.53125, "learning_rate": 3.4285401828675346e-08, "loss": 1.3821, "step": 38977 }, { "epoch": 97.445, "grad_norm": 5.5, "learning_rate": 3.421844369087879e-08, "loss": 1.3084, "step": 38978 }, { "epoch": 97.4475, "grad_norm": 7.09375, "learning_rate": 3.415155088834832e-08, "loss": 1.4732, "step": 38979 }, { "epoch": 97.45, "grad_norm": 9.1875, "learning_rate": 3.408472342152136e-08, "loss": 1.3805, "step": 38980 }, { "epoch": 97.4525, "grad_norm": 6.53125, "learning_rate": 3.4017961290835345e-08, "loss": 1.3895, "step": 38981 }, { "epoch": 97.455, "grad_norm": 7.1875, "learning_rate": 3.395126449672992e-08, "loss": 1.3659, "step": 38982 }, { "epoch": 97.4575, "grad_norm": 9.75, "learning_rate": 3.388463303964029e-08, "loss": 1.2741, "step": 38983 }, { "epoch": 97.46, "grad_norm": 7.75, "learning_rate": 3.381806692000389e-08, "loss": 1.3396, "step": 38984 }, { "epoch": 97.4625, "grad_norm": 6.9375, "learning_rate": 3.375156613825814e-08, "loss": 1.3358, "step": 38985 }, { "epoch": 97.465, "grad_norm": 10.4375, "learning_rate": 3.3685130694838255e-08, "loss": 1.2955, "step": 38986 }, { "epoch": 97.4675, "grad_norm": 6.28125, "learning_rate": 3.3618760590180544e-08, "loss": 1.3751, "step": 38987 }, { "epoch": 97.47, "grad_norm": 8.3125, "learning_rate": 3.3552455824718e-08, "loss": 1.2808, "step": 38988 }, { "epoch": 97.4725, "grad_norm": 7.59375, "learning_rate": 3.348621639888694e-08, "loss": 1.2643, "step": 38989 }, { "epoch": 97.475, "grad_norm": 6.4375, "learning_rate": 3.342004231312257e-08, "loss": 1.2758, "step": 38990 }, { "epoch": 97.4775, "grad_norm": 6.125, "learning_rate": 3.335393356785677e-08, "loss": 1.258, "step": 38991 }, { "epoch": 97.48, "grad_norm": 5.875, "learning_rate": 3.3287890163523626e-08, "loss": 1.2633, "step": 38992 }, { "epoch": 97.4825, "grad_norm": 9.5, "learning_rate": 3.322191210055614e-08, "loss": 1.3645, "step": 38993 }, { "epoch": 97.485, "grad_norm": 6.625, "learning_rate": 3.3155999379387293e-08, "loss": 1.3011, "step": 38994 }, { "epoch": 97.4875, "grad_norm": 12.5, "learning_rate": 3.309015200044785e-08, "loss": 1.2878, "step": 38995 }, { "epoch": 97.49, "grad_norm": 8.4375, "learning_rate": 3.3024369964171907e-08, "loss": 1.3232, "step": 38996 }, { "epoch": 97.4925, "grad_norm": 5.96875, "learning_rate": 3.295865327098913e-08, "loss": 1.2518, "step": 38997 }, { "epoch": 97.495, "grad_norm": 6.59375, "learning_rate": 3.289300192133027e-08, "loss": 1.3773, "step": 38998 }, { "epoch": 97.4975, "grad_norm": 6.9375, "learning_rate": 3.28274159156261e-08, "loss": 1.387, "step": 38999 }, { "epoch": 97.5, "grad_norm": 6.59375, "learning_rate": 3.2761895254306285e-08, "loss": 1.2832, "step": 39000 }, { "epoch": 97.5025, "grad_norm": 6.53125, "learning_rate": 3.269643993780047e-08, "loss": 1.2992, "step": 39001 }, { "epoch": 97.505, "grad_norm": 5.96875, "learning_rate": 3.263104996653832e-08, "loss": 1.2538, "step": 39002 }, { "epoch": 97.5075, "grad_norm": 8.125, "learning_rate": 3.2565725340948374e-08, "loss": 1.2328, "step": 39003 }, { "epoch": 97.51, "grad_norm": 8.0, "learning_rate": 3.2500466061458066e-08, "loss": 1.331, "step": 39004 }, { "epoch": 97.5125, "grad_norm": 7.125, "learning_rate": 3.2435272128495957e-08, "loss": 1.3023, "step": 39005 }, { "epoch": 97.515, "grad_norm": 6.375, "learning_rate": 3.2370143542489465e-08, "loss": 1.4247, "step": 39006 }, { "epoch": 97.5175, "grad_norm": 7.78125, "learning_rate": 3.230508030386492e-08, "loss": 1.3288, "step": 39007 }, { "epoch": 97.52, "grad_norm": 7.0, "learning_rate": 3.224008241304977e-08, "loss": 1.2759, "step": 39008 }, { "epoch": 97.5225, "grad_norm": 5.96875, "learning_rate": 3.217514987047032e-08, "loss": 1.116, "step": 39009 }, { "epoch": 97.525, "grad_norm": 9.6875, "learning_rate": 3.211028267655181e-08, "loss": 1.3641, "step": 39010 }, { "epoch": 97.5275, "grad_norm": 6.84375, "learning_rate": 3.204548083171832e-08, "loss": 1.2327, "step": 39011 }, { "epoch": 97.53, "grad_norm": 9.5625, "learning_rate": 3.1980744336396195e-08, "loss": 1.4027, "step": 39012 }, { "epoch": 97.5325, "grad_norm": 6.46875, "learning_rate": 3.1916073191009536e-08, "loss": 1.3455, "step": 39013 }, { "epoch": 97.535, "grad_norm": 6.875, "learning_rate": 3.1851467395982435e-08, "loss": 1.3196, "step": 39014 }, { "epoch": 97.5375, "grad_norm": 8.25, "learning_rate": 3.178692695173791e-08, "loss": 1.2832, "step": 39015 }, { "epoch": 97.54, "grad_norm": 6.84375, "learning_rate": 3.172245185869893e-08, "loss": 1.4063, "step": 39016 }, { "epoch": 97.5425, "grad_norm": 9.625, "learning_rate": 3.165804211728962e-08, "loss": 1.3217, "step": 39017 }, { "epoch": 97.545, "grad_norm": 9.625, "learning_rate": 3.159369772793075e-08, "loss": 1.3101, "step": 39018 }, { "epoch": 97.5475, "grad_norm": 9.6875, "learning_rate": 3.15294186910442e-08, "loss": 1.2503, "step": 39019 }, { "epoch": 97.55, "grad_norm": 9.0625, "learning_rate": 3.1465205007052965e-08, "loss": 1.3333, "step": 39020 }, { "epoch": 97.5525, "grad_norm": 6.625, "learning_rate": 3.140105667637561e-08, "loss": 1.2915, "step": 39021 }, { "epoch": 97.555, "grad_norm": 10.4375, "learning_rate": 3.133697369943511e-08, "loss": 1.4051, "step": 39022 }, { "epoch": 97.5575, "grad_norm": 8.0, "learning_rate": 3.127295607665004e-08, "loss": 1.1787, "step": 39023 }, { "epoch": 97.56, "grad_norm": 6.78125, "learning_rate": 3.120900380844116e-08, "loss": 1.3719, "step": 39024 }, { "epoch": 97.5625, "grad_norm": 7.875, "learning_rate": 3.114511689522592e-08, "loss": 1.5107, "step": 39025 }, { "epoch": 97.565, "grad_norm": 8.625, "learning_rate": 3.10812953374251e-08, "loss": 1.3912, "step": 39026 }, { "epoch": 97.5675, "grad_norm": 21.125, "learning_rate": 3.1017539135456135e-08, "loss": 1.3788, "step": 39027 }, { "epoch": 97.57, "grad_norm": 5.78125, "learning_rate": 3.095384828973758e-08, "loss": 1.2997, "step": 39028 }, { "epoch": 97.5725, "grad_norm": 7.65625, "learning_rate": 3.0890222800686876e-08, "loss": 1.2734, "step": 39029 }, { "epoch": 97.575, "grad_norm": 7.875, "learning_rate": 3.082666266872036e-08, "loss": 1.2628, "step": 39030 }, { "epoch": 97.5775, "grad_norm": 6.375, "learning_rate": 3.076316789425548e-08, "loss": 1.2165, "step": 39031 }, { "epoch": 97.58, "grad_norm": 8.0625, "learning_rate": 3.0699738477708576e-08, "loss": 1.4122, "step": 39032 }, { "epoch": 97.5825, "grad_norm": 8.5625, "learning_rate": 3.063637441949485e-08, "loss": 1.2872, "step": 39033 }, { "epoch": 97.585, "grad_norm": 9.375, "learning_rate": 3.057307572002954e-08, "loss": 1.4102, "step": 39034 }, { "epoch": 97.5875, "grad_norm": 9.375, "learning_rate": 3.0509842379730094e-08, "loss": 1.2922, "step": 39035 }, { "epoch": 97.59, "grad_norm": 8.75, "learning_rate": 3.044667439900728e-08, "loss": 1.466, "step": 39036 }, { "epoch": 97.5925, "grad_norm": 5.625, "learning_rate": 3.0383571778277445e-08, "loss": 1.3539, "step": 39037 }, { "epoch": 97.595, "grad_norm": 7.0625, "learning_rate": 3.032053451795469e-08, "loss": 1.2851, "step": 39038 }, { "epoch": 97.5975, "grad_norm": 8.9375, "learning_rate": 3.025756261845092e-08, "loss": 1.4244, "step": 39039 }, { "epoch": 97.6, "grad_norm": 7.0625, "learning_rate": 3.019465608018024e-08, "loss": 1.1898, "step": 39040 }, { "epoch": 97.6025, "grad_norm": 7.15625, "learning_rate": 3.013181490355343e-08, "loss": 1.5279, "step": 39041 }, { "epoch": 97.605, "grad_norm": 8.625, "learning_rate": 3.006903908898462e-08, "loss": 1.3355, "step": 39042 }, { "epoch": 97.6075, "grad_norm": 7.15625, "learning_rate": 3.000632863688347e-08, "loss": 1.2776, "step": 39043 }, { "epoch": 97.61, "grad_norm": 11.1875, "learning_rate": 2.994368354766186e-08, "loss": 1.1714, "step": 39044 }, { "epoch": 97.6125, "grad_norm": 6.03125, "learning_rate": 2.9881103821730595e-08, "loss": 1.4699, "step": 39045 }, { "epoch": 97.615, "grad_norm": 7.96875, "learning_rate": 2.981858945950045e-08, "loss": 1.2534, "step": 39046 }, { "epoch": 97.6175, "grad_norm": 9.5, "learning_rate": 2.9756140461379978e-08, "loss": 1.2333, "step": 39047 }, { "epoch": 97.62, "grad_norm": 6.25, "learning_rate": 2.9693756827778862e-08, "loss": 1.1868, "step": 39048 }, { "epoch": 97.6225, "grad_norm": 7.46875, "learning_rate": 2.9631438559106775e-08, "loss": 1.268, "step": 39049 }, { "epoch": 97.625, "grad_norm": 7.0, "learning_rate": 2.9569185655773382e-08, "loss": 1.278, "step": 39050 }, { "epoch": 97.6275, "grad_norm": 6.25, "learning_rate": 2.9506998118183917e-08, "loss": 1.4177, "step": 39051 }, { "epoch": 97.63, "grad_norm": 8.1875, "learning_rate": 2.9444875946746943e-08, "loss": 1.6493, "step": 39052 }, { "epoch": 97.6325, "grad_norm": 9.125, "learning_rate": 2.9382819141872133e-08, "loss": 1.3179, "step": 39053 }, { "epoch": 97.635, "grad_norm": 6.34375, "learning_rate": 2.9320827703963606e-08, "loss": 1.3957, "step": 39054 }, { "epoch": 97.6375, "grad_norm": 10.625, "learning_rate": 2.9258901633428816e-08, "loss": 1.4096, "step": 39055 }, { "epoch": 97.64, "grad_norm": 7.28125, "learning_rate": 2.9197040930674102e-08, "loss": 1.4453, "step": 39056 }, { "epoch": 97.6425, "grad_norm": 8.625, "learning_rate": 2.913524559610359e-08, "loss": 1.2315, "step": 39057 }, { "epoch": 97.645, "grad_norm": 7.0625, "learning_rate": 2.9073515630124727e-08, "loss": 1.3179, "step": 39058 }, { "epoch": 97.6475, "grad_norm": 7.1875, "learning_rate": 2.9011851033139414e-08, "loss": 1.3242, "step": 39059 }, { "epoch": 97.65, "grad_norm": 6.84375, "learning_rate": 2.8950251805553997e-08, "loss": 1.2704, "step": 39060 }, { "epoch": 97.6525, "grad_norm": 8.625, "learning_rate": 2.8888717947772595e-08, "loss": 1.205, "step": 39061 }, { "epoch": 97.655, "grad_norm": 9.1875, "learning_rate": 2.8827249460197104e-08, "loss": 1.3884, "step": 39062 }, { "epoch": 97.6575, "grad_norm": 7.03125, "learning_rate": 2.876584634323054e-08, "loss": 1.3579, "step": 39063 }, { "epoch": 97.66, "grad_norm": 5.53125, "learning_rate": 2.8704508597277024e-08, "loss": 1.2458, "step": 39064 }, { "epoch": 97.6625, "grad_norm": 6.625, "learning_rate": 2.864323622273735e-08, "loss": 1.4206, "step": 39065 }, { "epoch": 97.665, "grad_norm": 9.25, "learning_rate": 2.8582029220013407e-08, "loss": 1.2605, "step": 39066 }, { "epoch": 97.6675, "grad_norm": 6.5625, "learning_rate": 2.8520887589505998e-08, "loss": 1.4121, "step": 39067 }, { "epoch": 97.67, "grad_norm": 7.6875, "learning_rate": 2.8459811331617016e-08, "loss": 1.472, "step": 39068 }, { "epoch": 97.6725, "grad_norm": 10.5, "learning_rate": 2.8398800446747253e-08, "loss": 1.4142, "step": 39069 }, { "epoch": 97.675, "grad_norm": 7.9375, "learning_rate": 2.8337854935295284e-08, "loss": 1.1964, "step": 39070 }, { "epoch": 97.6775, "grad_norm": 7.6875, "learning_rate": 2.8276974797661892e-08, "loss": 1.5049, "step": 39071 }, { "epoch": 97.68, "grad_norm": 4.375, "learning_rate": 2.8216160034244544e-08, "loss": 1.1651, "step": 39072 }, { "epoch": 97.6825, "grad_norm": 7.53125, "learning_rate": 2.8155410645444027e-08, "loss": 1.3786, "step": 39073 }, { "epoch": 97.685, "grad_norm": 9.4375, "learning_rate": 2.8094726631657797e-08, "loss": 1.3472, "step": 39074 }, { "epoch": 97.6875, "grad_norm": 7.0625, "learning_rate": 2.803410799328221e-08, "loss": 1.2347, "step": 39075 }, { "epoch": 97.69, "grad_norm": 8.1875, "learning_rate": 2.7973554730716947e-08, "loss": 1.2428, "step": 39076 }, { "epoch": 97.6925, "grad_norm": 8.1875, "learning_rate": 2.791306684435724e-08, "loss": 1.3396, "step": 39077 }, { "epoch": 97.695, "grad_norm": 9.9375, "learning_rate": 2.7852644334601665e-08, "loss": 1.3152, "step": 39078 }, { "epoch": 97.6975, "grad_norm": 10.4375, "learning_rate": 2.7792287201843237e-08, "loss": 1.1957, "step": 39079 }, { "epoch": 97.7, "grad_norm": 7.21875, "learning_rate": 2.773199544648164e-08, "loss": 1.3103, "step": 39080 }, { "epoch": 97.7025, "grad_norm": 8.0625, "learning_rate": 2.767176906890878e-08, "loss": 1.3535, "step": 39081 }, { "epoch": 97.705, "grad_norm": 9.25, "learning_rate": 2.7611608069521013e-08, "loss": 1.3742, "step": 39082 }, { "epoch": 97.7075, "grad_norm": 6.5625, "learning_rate": 2.7551512448711347e-08, "loss": 1.2432, "step": 39083 }, { "epoch": 97.71, "grad_norm": 7.375, "learning_rate": 2.749148220687614e-08, "loss": 1.2631, "step": 39084 }, { "epoch": 97.7125, "grad_norm": 6.46875, "learning_rate": 2.743151734440841e-08, "loss": 1.3454, "step": 39085 }, { "epoch": 97.715, "grad_norm": 7.59375, "learning_rate": 2.737161786169895e-08, "loss": 1.2874, "step": 39086 }, { "epoch": 97.7175, "grad_norm": 6.8125, "learning_rate": 2.7311783759143006e-08, "loss": 1.2571, "step": 39087 }, { "epoch": 97.72, "grad_norm": 10.8125, "learning_rate": 2.7252015037131373e-08, "loss": 1.2405, "step": 39088 }, { "epoch": 97.7225, "grad_norm": 6.375, "learning_rate": 2.7192311696057073e-08, "loss": 1.3466, "step": 39089 }, { "epoch": 97.725, "grad_norm": 13.3125, "learning_rate": 2.713267373630979e-08, "loss": 1.2972, "step": 39090 }, { "epoch": 97.7275, "grad_norm": 8.375, "learning_rate": 2.7073101158282545e-08, "loss": 1.1265, "step": 39091 }, { "epoch": 97.73, "grad_norm": 6.75, "learning_rate": 2.7013593962365025e-08, "loss": 1.2774, "step": 39092 }, { "epoch": 97.7325, "grad_norm": 8.5, "learning_rate": 2.6954152148946922e-08, "loss": 1.4497, "step": 39093 }, { "epoch": 97.735, "grad_norm": 9.375, "learning_rate": 2.689477571841792e-08, "loss": 1.3302, "step": 39094 }, { "epoch": 97.7375, "grad_norm": 6.53125, "learning_rate": 2.683546467116882e-08, "loss": 1.3464, "step": 39095 }, { "epoch": 97.74, "grad_norm": 7.4375, "learning_rate": 2.677621900758709e-08, "loss": 1.242, "step": 39096 }, { "epoch": 97.7425, "grad_norm": 7.5, "learning_rate": 2.6717038728060197e-08, "loss": 1.2177, "step": 39097 }, { "epoch": 97.745, "grad_norm": 8.375, "learning_rate": 2.6657923832977827e-08, "loss": 1.3821, "step": 39098 }, { "epoch": 97.7475, "grad_norm": 7.4375, "learning_rate": 2.6598874322727452e-08, "loss": 1.315, "step": 39099 }, { "epoch": 97.75, "grad_norm": 7.75, "learning_rate": 2.6539890197695428e-08, "loss": 1.3198, "step": 39100 }, { "epoch": 97.7525, "grad_norm": 5.1875, "learning_rate": 2.648097145826922e-08, "loss": 1.4935, "step": 39101 }, { "epoch": 97.755, "grad_norm": 8.1875, "learning_rate": 2.6422118104834083e-08, "loss": 1.4046, "step": 39102 }, { "epoch": 97.7575, "grad_norm": 6.40625, "learning_rate": 2.6363330137776366e-08, "loss": 1.2839, "step": 39103 }, { "epoch": 97.76, "grad_norm": 7.3125, "learning_rate": 2.6304607557481322e-08, "loss": 1.2448, "step": 39104 }, { "epoch": 97.7625, "grad_norm": 8.0, "learning_rate": 2.6245950364335304e-08, "loss": 1.4096, "step": 39105 }, { "epoch": 97.765, "grad_norm": 9.4375, "learning_rate": 2.6187358558720233e-08, "loss": 1.2882, "step": 39106 }, { "epoch": 97.7675, "grad_norm": 6.15625, "learning_rate": 2.6128832141022464e-08, "loss": 1.364, "step": 39107 }, { "epoch": 97.77, "grad_norm": 6.65625, "learning_rate": 2.6070371111625026e-08, "loss": 1.3149, "step": 39108 }, { "epoch": 97.7725, "grad_norm": 5.84375, "learning_rate": 2.6011975470910945e-08, "loss": 1.2361, "step": 39109 }, { "epoch": 97.775, "grad_norm": 7.0625, "learning_rate": 2.5953645219263245e-08, "loss": 1.2513, "step": 39110 }, { "epoch": 97.7775, "grad_norm": 9.375, "learning_rate": 2.589538035706496e-08, "loss": 1.3327, "step": 39111 }, { "epoch": 97.78, "grad_norm": 7.375, "learning_rate": 2.583718088469689e-08, "loss": 1.2984, "step": 39112 }, { "epoch": 97.7825, "grad_norm": 7.375, "learning_rate": 2.5779046802539842e-08, "loss": 1.2454, "step": 39113 }, { "epoch": 97.785, "grad_norm": 6.875, "learning_rate": 2.5720978110977958e-08, "loss": 1.1022, "step": 39114 }, { "epoch": 97.7875, "grad_norm": 5.40625, "learning_rate": 2.5662974810389817e-08, "loss": 1.2955, "step": 39115 }, { "epoch": 97.79, "grad_norm": 8.5625, "learning_rate": 2.5605036901156233e-08, "loss": 1.3585, "step": 39116 }, { "epoch": 97.7925, "grad_norm": 7.46875, "learning_rate": 2.5547164383655788e-08, "loss": 1.406, "step": 39117 }, { "epoch": 97.795, "grad_norm": 8.5, "learning_rate": 2.5489357258270396e-08, "loss": 1.2415, "step": 39118 }, { "epoch": 97.7975, "grad_norm": 12.3125, "learning_rate": 2.543161552537643e-08, "loss": 1.3151, "step": 39119 }, { "epoch": 97.8, "grad_norm": 7.40625, "learning_rate": 2.537393918535358e-08, "loss": 1.3557, "step": 39120 }, { "epoch": 97.8025, "grad_norm": 8.8125, "learning_rate": 2.5316328238581545e-08, "loss": 1.2724, "step": 39121 }, { "epoch": 97.805, "grad_norm": 6.0625, "learning_rate": 2.525878268543558e-08, "loss": 1.187, "step": 39122 }, { "epoch": 97.8075, "grad_norm": 15.875, "learning_rate": 2.5201302526293157e-08, "loss": 1.1691, "step": 39123 }, { "epoch": 97.81, "grad_norm": 11.5625, "learning_rate": 2.514388776153287e-08, "loss": 1.288, "step": 39124 }, { "epoch": 97.8125, "grad_norm": 6.71875, "learning_rate": 2.5086538391529968e-08, "loss": 1.3064, "step": 39125 }, { "epoch": 97.815, "grad_norm": 8.0, "learning_rate": 2.502925441665971e-08, "loss": 1.1959, "step": 39126 }, { "epoch": 97.8175, "grad_norm": 7.625, "learning_rate": 2.4972035837299567e-08, "loss": 1.2851, "step": 39127 }, { "epoch": 97.82, "grad_norm": 5.46875, "learning_rate": 2.4914882653822582e-08, "loss": 1.267, "step": 39128 }, { "epoch": 97.8225, "grad_norm": 7.65625, "learning_rate": 2.4857794866605113e-08, "loss": 1.2446, "step": 39129 }, { "epoch": 97.825, "grad_norm": 9.75, "learning_rate": 2.4800772476020196e-08, "loss": 1.4725, "step": 39130 }, { "epoch": 97.8275, "grad_norm": 6.9375, "learning_rate": 2.474381548244309e-08, "loss": 1.2514, "step": 39131 }, { "epoch": 97.83, "grad_norm": 7.90625, "learning_rate": 2.468692388624572e-08, "loss": 1.3046, "step": 39132 }, { "epoch": 97.8325, "grad_norm": 6.78125, "learning_rate": 2.4630097687802225e-08, "loss": 1.2429, "step": 39133 }, { "epoch": 97.835, "grad_norm": 9.9375, "learning_rate": 2.4573336887483425e-08, "loss": 1.2345, "step": 39134 }, { "epoch": 97.8375, "grad_norm": 6.25, "learning_rate": 2.451664148566346e-08, "loss": 1.3934, "step": 39135 }, { "epoch": 97.84, "grad_norm": 8.625, "learning_rate": 2.4460011482713153e-08, "loss": 1.3019, "step": 39136 }, { "epoch": 97.8425, "grad_norm": 9.75, "learning_rate": 2.440344687900442e-08, "loss": 1.2592, "step": 39137 }, { "epoch": 97.845, "grad_norm": 7.125, "learning_rate": 2.4346947674906974e-08, "loss": 1.244, "step": 39138 }, { "epoch": 97.8475, "grad_norm": 10.875, "learning_rate": 2.4290513870791622e-08, "loss": 1.2637, "step": 39139 }, { "epoch": 97.85, "grad_norm": 7.375, "learning_rate": 2.423414546702807e-08, "loss": 1.4202, "step": 39140 }, { "epoch": 97.8525, "grad_norm": 8.9375, "learning_rate": 2.4177842463987135e-08, "loss": 1.3129, "step": 39141 }, { "epoch": 97.855, "grad_norm": 7.6875, "learning_rate": 2.4121604862036295e-08, "loss": 1.5878, "step": 39142 }, { "epoch": 97.8575, "grad_norm": 5.625, "learning_rate": 2.406543266154526e-08, "loss": 1.3327, "step": 39143 }, { "epoch": 97.86, "grad_norm": 7.59375, "learning_rate": 2.4009325862881517e-08, "loss": 1.2924, "step": 39144 }, { "epoch": 97.8625, "grad_norm": 9.8125, "learning_rate": 2.3953284466414762e-08, "loss": 1.2827, "step": 39145 }, { "epoch": 97.865, "grad_norm": 7.8125, "learning_rate": 2.389730847251026e-08, "loss": 1.373, "step": 39146 }, { "epoch": 97.8675, "grad_norm": 10.6875, "learning_rate": 2.38413978815355e-08, "loss": 1.4644, "step": 39147 }, { "epoch": 97.87, "grad_norm": 7.40625, "learning_rate": 2.378555269385796e-08, "loss": 1.3492, "step": 39148 }, { "epoch": 97.8725, "grad_norm": 7.3125, "learning_rate": 2.3729772909842907e-08, "loss": 1.3978, "step": 39149 }, { "epoch": 97.875, "grad_norm": 7.78125, "learning_rate": 2.3674058529855603e-08, "loss": 1.3443, "step": 39150 }, { "epoch": 97.8775, "grad_norm": 8.4375, "learning_rate": 2.3618409554262423e-08, "loss": 1.391, "step": 39151 }, { "epoch": 97.88, "grad_norm": 9.3125, "learning_rate": 2.3562825983427517e-08, "loss": 1.4608, "step": 39152 }, { "epoch": 97.8825, "grad_norm": 7.125, "learning_rate": 2.3507307817716153e-08, "loss": 1.2457, "step": 39153 }, { "epoch": 97.885, "grad_norm": 5.0, "learning_rate": 2.3451855057490258e-08, "loss": 1.297, "step": 39154 }, { "epoch": 97.8875, "grad_norm": 5.375, "learning_rate": 2.339646770311621e-08, "loss": 1.2151, "step": 39155 }, { "epoch": 97.89, "grad_norm": 5.75, "learning_rate": 2.334114575495483e-08, "loss": 1.3745, "step": 39156 }, { "epoch": 97.8925, "grad_norm": 8.5, "learning_rate": 2.328588921337027e-08, "loss": 1.2892, "step": 39157 }, { "epoch": 97.895, "grad_norm": 5.90625, "learning_rate": 2.3230698078723356e-08, "loss": 1.216, "step": 39158 }, { "epoch": 97.8975, "grad_norm": 6.4375, "learning_rate": 2.317557235137824e-08, "loss": 1.3838, "step": 39159 }, { "epoch": 97.9, "grad_norm": 5.09375, "learning_rate": 2.312051203169352e-08, "loss": 1.4475, "step": 39160 }, { "epoch": 97.9025, "grad_norm": 6.4375, "learning_rate": 2.3065517120031134e-08, "loss": 1.2408, "step": 39161 }, { "epoch": 97.905, "grad_norm": 7.125, "learning_rate": 2.3010587616753012e-08, "loss": 1.4043, "step": 39162 }, { "epoch": 97.9075, "grad_norm": 7.65625, "learning_rate": 2.2955723522217754e-08, "loss": 1.2753, "step": 39163 }, { "epoch": 97.91, "grad_norm": 5.28125, "learning_rate": 2.2900924836786188e-08, "loss": 1.2381, "step": 39164 }, { "epoch": 97.9125, "grad_norm": 6.59375, "learning_rate": 2.28461915608158e-08, "loss": 1.232, "step": 39165 }, { "epoch": 97.915, "grad_norm": 8.5625, "learning_rate": 2.2791523694667417e-08, "loss": 1.3251, "step": 39166 }, { "epoch": 97.9175, "grad_norm": 9.875, "learning_rate": 2.2736921238697417e-08, "loss": 1.2825, "step": 39167 }, { "epoch": 97.92, "grad_norm": 8.5, "learning_rate": 2.2682384193266625e-08, "loss": 1.3827, "step": 39168 }, { "epoch": 97.9225, "grad_norm": 6.65625, "learning_rate": 2.2627912558729204e-08, "loss": 1.4194, "step": 39169 }, { "epoch": 97.925, "grad_norm": 7.15625, "learning_rate": 2.2573506335444862e-08, "loss": 1.423, "step": 39170 }, { "epoch": 97.9275, "grad_norm": 9.25, "learning_rate": 2.2519165523768872e-08, "loss": 1.2837, "step": 39171 }, { "epoch": 97.93, "grad_norm": 8.25, "learning_rate": 2.2464890124057614e-08, "loss": 1.2415, "step": 39172 }, { "epoch": 97.9325, "grad_norm": 7.96875, "learning_rate": 2.2410680136667474e-08, "loss": 1.1829, "step": 39173 }, { "epoch": 97.935, "grad_norm": 6.3125, "learning_rate": 2.2356535561954827e-08, "loss": 1.2685, "step": 39174 }, { "epoch": 97.9375, "grad_norm": 6.90625, "learning_rate": 2.230245640027273e-08, "loss": 1.386, "step": 39175 }, { "epoch": 97.94, "grad_norm": 7.5625, "learning_rate": 2.2248442651975345e-08, "loss": 1.2574, "step": 39176 }, { "epoch": 97.9425, "grad_norm": 6.78125, "learning_rate": 2.2194494317419048e-08, "loss": 1.2619, "step": 39177 }, { "epoch": 97.945, "grad_norm": 8.1875, "learning_rate": 2.214061139695578e-08, "loss": 1.2723, "step": 39178 }, { "epoch": 97.9475, "grad_norm": 12.3125, "learning_rate": 2.2086793890940817e-08, "loss": 1.2743, "step": 39179 }, { "epoch": 97.95, "grad_norm": 8.4375, "learning_rate": 2.2033041799723877e-08, "loss": 1.4633, "step": 39180 }, { "epoch": 97.9525, "grad_norm": 8.25, "learning_rate": 2.197935512365912e-08, "loss": 1.2931, "step": 39181 }, { "epoch": 97.955, "grad_norm": 10.8125, "learning_rate": 2.1925733863099595e-08, "loss": 1.4714, "step": 39182 }, { "epoch": 97.9575, "grad_norm": 8.5, "learning_rate": 2.187217801839503e-08, "loss": 1.3237, "step": 39183 }, { "epoch": 97.96, "grad_norm": 12.75, "learning_rate": 2.1818687589896248e-08, "loss": 1.3651, "step": 39184 }, { "epoch": 97.9625, "grad_norm": 7.4375, "learning_rate": 2.1765262577956303e-08, "loss": 1.4096, "step": 39185 }, { "epoch": 97.965, "grad_norm": 12.0, "learning_rate": 2.1711902982922696e-08, "loss": 1.3986, "step": 39186 }, { "epoch": 97.9675, "grad_norm": 8.0625, "learning_rate": 2.1658608805147363e-08, "loss": 1.2525, "step": 39187 }, { "epoch": 97.97, "grad_norm": 7.125, "learning_rate": 2.1605380044978918e-08, "loss": 1.3514, "step": 39188 }, { "epoch": 97.9725, "grad_norm": 7.875, "learning_rate": 2.155221670276597e-08, "loss": 1.2887, "step": 39189 }, { "epoch": 97.975, "grad_norm": 8.5625, "learning_rate": 2.1499118778857132e-08, "loss": 1.41, "step": 39190 }, { "epoch": 97.9775, "grad_norm": 10.3125, "learning_rate": 2.144608627360101e-08, "loss": 1.0559, "step": 39191 }, { "epoch": 97.98, "grad_norm": 12.0, "learning_rate": 2.1393119187345103e-08, "loss": 1.2765, "step": 39192 }, { "epoch": 97.9825, "grad_norm": 10.125, "learning_rate": 2.1340217520436913e-08, "loss": 1.4593, "step": 39193 }, { "epoch": 97.985, "grad_norm": 6.96875, "learning_rate": 2.1287381273222828e-08, "loss": 1.307, "step": 39194 }, { "epoch": 97.9875, "grad_norm": 6.375, "learning_rate": 2.1234610446049243e-08, "loss": 1.2686, "step": 39195 }, { "epoch": 97.99, "grad_norm": 8.9375, "learning_rate": 2.118190503926143e-08, "loss": 1.3492, "step": 39196 }, { "epoch": 97.9925, "grad_norm": 7.5625, "learning_rate": 2.1129265053206893e-08, "loss": 1.312, "step": 39197 }, { "epoch": 97.995, "grad_norm": 5.375, "learning_rate": 2.107669048822869e-08, "loss": 1.1923, "step": 39198 }, { "epoch": 97.9975, "grad_norm": 8.0625, "learning_rate": 2.1024181344673213e-08, "loss": 1.3936, "step": 39199 }, { "epoch": 98.0, "grad_norm": 7.84375, "learning_rate": 2.0971737622883515e-08, "loss": 1.3602, "step": 39200 }, { "epoch": 98.0025, "grad_norm": 13.0625, "learning_rate": 2.091935932320377e-08, "loss": 1.3519, "step": 39201 }, { "epoch": 98.005, "grad_norm": 7.21875, "learning_rate": 2.0867046445977033e-08, "loss": 1.2873, "step": 39202 }, { "epoch": 98.0075, "grad_norm": 13.375, "learning_rate": 2.081479899154748e-08, "loss": 1.4076, "step": 39203 }, { "epoch": 98.01, "grad_norm": 7.21875, "learning_rate": 2.0762616960255943e-08, "loss": 1.2623, "step": 39204 }, { "epoch": 98.0125, "grad_norm": 9.1875, "learning_rate": 2.0710500352445485e-08, "loss": 1.3371, "step": 39205 }, { "epoch": 98.015, "grad_norm": 7.3125, "learning_rate": 2.065844916845694e-08, "loss": 1.3261, "step": 39206 }, { "epoch": 98.0175, "grad_norm": 5.21875, "learning_rate": 2.060646340863337e-08, "loss": 1.2871, "step": 39207 }, { "epoch": 98.02, "grad_norm": 8.25, "learning_rate": 2.0554543073313393e-08, "loss": 1.3177, "step": 39208 }, { "epoch": 98.0225, "grad_norm": 8.9375, "learning_rate": 2.0502688162840067e-08, "loss": 1.1503, "step": 39209 }, { "epoch": 98.025, "grad_norm": 6.1875, "learning_rate": 2.0450898677550902e-08, "loss": 1.2543, "step": 39210 }, { "epoch": 98.0275, "grad_norm": 5.71875, "learning_rate": 2.0399174617785623e-08, "loss": 1.3668, "step": 39211 }, { "epoch": 98.03, "grad_norm": 6.75, "learning_rate": 2.0347515983885068e-08, "loss": 1.3398, "step": 39212 }, { "epoch": 98.0325, "grad_norm": 5.84375, "learning_rate": 2.029592277618564e-08, "loss": 1.2808, "step": 39213 }, { "epoch": 98.035, "grad_norm": 8.8125, "learning_rate": 2.024439499502817e-08, "loss": 1.3999, "step": 39214 }, { "epoch": 98.0375, "grad_norm": 7.09375, "learning_rate": 2.0192932640747952e-08, "loss": 1.4774, "step": 39215 }, { "epoch": 98.04, "grad_norm": 8.0625, "learning_rate": 2.01415357136836e-08, "loss": 1.3327, "step": 39216 }, { "epoch": 98.0425, "grad_norm": 7.8125, "learning_rate": 2.0090204214172627e-08, "loss": 1.3164, "step": 39217 }, { "epoch": 98.045, "grad_norm": 8.1875, "learning_rate": 2.003893814255142e-08, "loss": 1.2888, "step": 39218 }, { "epoch": 98.0475, "grad_norm": 5.84375, "learning_rate": 1.998773749915417e-08, "loss": 1.2751, "step": 39219 }, { "epoch": 98.05, "grad_norm": 6.21875, "learning_rate": 1.9936602284318375e-08, "loss": 1.1728, "step": 39220 }, { "epoch": 98.0525, "grad_norm": 7.40625, "learning_rate": 1.9885532498379323e-08, "loss": 1.4581, "step": 39221 }, { "epoch": 98.055, "grad_norm": 7.375, "learning_rate": 1.9834528141671194e-08, "loss": 1.1695, "step": 39222 }, { "epoch": 98.0575, "grad_norm": 9.625, "learning_rate": 1.9783589214528165e-08, "loss": 1.3722, "step": 39223 }, { "epoch": 98.06, "grad_norm": 7.65625, "learning_rate": 1.973271571728441e-08, "loss": 1.348, "step": 39224 }, { "epoch": 98.0625, "grad_norm": 7.90625, "learning_rate": 1.9681907650274113e-08, "loss": 1.3071, "step": 39225 }, { "epoch": 98.065, "grad_norm": 8.4375, "learning_rate": 1.9631165013829224e-08, "loss": 1.4459, "step": 39226 }, { "epoch": 98.0675, "grad_norm": 6.96875, "learning_rate": 1.958048780828392e-08, "loss": 1.4248, "step": 39227 }, { "epoch": 98.07, "grad_norm": 7.25, "learning_rate": 1.9529876033969052e-08, "loss": 1.3256, "step": 39228 }, { "epoch": 98.0725, "grad_norm": 9.625, "learning_rate": 1.9479329691216576e-08, "loss": 1.3297, "step": 39229 }, { "epoch": 98.075, "grad_norm": 9.125, "learning_rate": 1.9428848780358444e-08, "loss": 1.3437, "step": 39230 }, { "epoch": 98.0775, "grad_norm": 6.75, "learning_rate": 1.937843330172551e-08, "loss": 1.159, "step": 39231 }, { "epoch": 98.08, "grad_norm": 7.5, "learning_rate": 1.93280832556475e-08, "loss": 1.3516, "step": 39232 }, { "epoch": 98.0825, "grad_norm": 10.5, "learning_rate": 1.927779864245638e-08, "loss": 1.3175, "step": 39233 }, { "epoch": 98.085, "grad_norm": 8.75, "learning_rate": 1.922757946247966e-08, "loss": 1.2175, "step": 39234 }, { "epoch": 98.0875, "grad_norm": 8.5, "learning_rate": 1.9177425716048193e-08, "loss": 1.4226, "step": 39235 }, { "epoch": 98.09, "grad_norm": 5.78125, "learning_rate": 1.9127337403489486e-08, "loss": 1.4598, "step": 39236 }, { "epoch": 98.0925, "grad_norm": 8.1875, "learning_rate": 1.9077314525133283e-08, "loss": 1.4564, "step": 39237 }, { "epoch": 98.095, "grad_norm": 5.96875, "learning_rate": 1.9027357081305985e-08, "loss": 1.3328, "step": 39238 }, { "epoch": 98.0975, "grad_norm": 8.0625, "learning_rate": 1.897746507233622e-08, "loss": 1.4226, "step": 39239 }, { "epoch": 98.1, "grad_norm": 18.625, "learning_rate": 1.8927638498551502e-08, "loss": 1.2826, "step": 39240 }, { "epoch": 98.1025, "grad_norm": 8.0625, "learning_rate": 1.8877877360277132e-08, "loss": 1.2954, "step": 39241 }, { "epoch": 98.105, "grad_norm": 8.0, "learning_rate": 1.8828181657839506e-08, "loss": 1.3988, "step": 39242 }, { "epoch": 98.1075, "grad_norm": 5.96875, "learning_rate": 1.8778551391565036e-08, "loss": 1.3498, "step": 39243 }, { "epoch": 98.11, "grad_norm": 6.78125, "learning_rate": 1.8728986561780127e-08, "loss": 1.3774, "step": 39244 }, { "epoch": 98.1125, "grad_norm": 7.90625, "learning_rate": 1.8679487168806742e-08, "loss": 1.3494, "step": 39245 }, { "epoch": 98.115, "grad_norm": 7.875, "learning_rate": 1.8630053212972398e-08, "loss": 1.3195, "step": 39246 }, { "epoch": 98.1175, "grad_norm": 6.71875, "learning_rate": 1.858068469460017e-08, "loss": 1.2805, "step": 39247 }, { "epoch": 98.12, "grad_norm": 8.5625, "learning_rate": 1.8531381614013133e-08, "loss": 1.3433, "step": 39248 }, { "epoch": 98.1225, "grad_norm": 7.78125, "learning_rate": 1.8482143971534362e-08, "loss": 1.1113, "step": 39249 }, { "epoch": 98.125, "grad_norm": 10.6875, "learning_rate": 1.8432971767488038e-08, "loss": 1.4021, "step": 39250 }, { "epoch": 98.1275, "grad_norm": 7.0625, "learning_rate": 1.8383865002195024e-08, "loss": 1.362, "step": 39251 }, { "epoch": 98.13, "grad_norm": 6.09375, "learning_rate": 1.8334823675978387e-08, "loss": 1.1841, "step": 39252 }, { "epoch": 98.1325, "grad_norm": 10.0, "learning_rate": 1.8285847789158984e-08, "loss": 1.3278, "step": 39253 }, { "epoch": 98.135, "grad_norm": 5.03125, "learning_rate": 1.823693734205878e-08, "loss": 1.3486, "step": 39254 }, { "epoch": 98.1375, "grad_norm": 8.3125, "learning_rate": 1.818809233499752e-08, "loss": 1.4126, "step": 39255 }, { "epoch": 98.14, "grad_norm": 5.84375, "learning_rate": 1.8139312768294946e-08, "loss": 1.4055, "step": 39256 }, { "epoch": 98.1425, "grad_norm": 9.125, "learning_rate": 1.8090598642271918e-08, "loss": 1.3465, "step": 39257 }, { "epoch": 98.145, "grad_norm": 7.5625, "learning_rate": 1.8041949957248174e-08, "loss": 1.3245, "step": 39258 }, { "epoch": 98.1475, "grad_norm": 7.375, "learning_rate": 1.7993366713541238e-08, "loss": 1.4153, "step": 39259 }, { "epoch": 98.15, "grad_norm": 8.25, "learning_rate": 1.7944848911470857e-08, "loss": 1.3279, "step": 39260 }, { "epoch": 98.1525, "grad_norm": 6.75, "learning_rate": 1.7896396551353446e-08, "loss": 1.408, "step": 39261 }, { "epoch": 98.155, "grad_norm": 8.6875, "learning_rate": 1.7848009633508744e-08, "loss": 1.352, "step": 39262 }, { "epoch": 98.1575, "grad_norm": 7.5625, "learning_rate": 1.779968815825317e-08, "loss": 1.3929, "step": 39263 }, { "epoch": 98.16, "grad_norm": 5.125, "learning_rate": 1.7751432125903133e-08, "loss": 1.2521, "step": 39264 }, { "epoch": 98.1625, "grad_norm": 6.25, "learning_rate": 1.770324153677505e-08, "loss": 1.485, "step": 39265 }, { "epoch": 98.165, "grad_norm": 6.8125, "learning_rate": 1.7655116391185334e-08, "loss": 1.4911, "step": 39266 }, { "epoch": 98.1675, "grad_norm": 7.40625, "learning_rate": 1.7607056689449288e-08, "loss": 1.4148, "step": 39267 }, { "epoch": 98.17, "grad_norm": 9.0, "learning_rate": 1.75590624318811e-08, "loss": 1.3698, "step": 39268 }, { "epoch": 98.1725, "grad_norm": 8.5, "learning_rate": 1.7511133618797193e-08, "loss": 1.3885, "step": 39269 }, { "epoch": 98.175, "grad_norm": 10.625, "learning_rate": 1.7463270250510645e-08, "loss": 1.3564, "step": 39270 }, { "epoch": 98.1775, "grad_norm": 6.21875, "learning_rate": 1.7415472327335647e-08, "loss": 1.2768, "step": 39271 }, { "epoch": 98.18, "grad_norm": 11.1875, "learning_rate": 1.7367739849584174e-08, "loss": 1.2967, "step": 39272 }, { "epoch": 98.1825, "grad_norm": 7.625, "learning_rate": 1.7320072817571532e-08, "loss": 1.4073, "step": 39273 }, { "epoch": 98.185, "grad_norm": 8.4375, "learning_rate": 1.7272471231608578e-08, "loss": 1.258, "step": 39274 }, { "epoch": 98.1875, "grad_norm": 12.4375, "learning_rate": 1.722493509200729e-08, "loss": 1.4568, "step": 39275 }, { "epoch": 98.19, "grad_norm": 7.34375, "learning_rate": 1.7177464399080746e-08, "loss": 1.3973, "step": 39276 }, { "epoch": 98.1925, "grad_norm": 8.5625, "learning_rate": 1.71300591531387e-08, "loss": 1.2219, "step": 39277 }, { "epoch": 98.195, "grad_norm": 10.6875, "learning_rate": 1.7082719354492018e-08, "loss": 1.3572, "step": 39278 }, { "epoch": 98.1975, "grad_norm": 3.890625, "learning_rate": 1.7035445003452668e-08, "loss": 1.2647, "step": 39279 }, { "epoch": 98.2, "grad_norm": 9.5625, "learning_rate": 1.698823610032929e-08, "loss": 1.4286, "step": 39280 }, { "epoch": 98.2025, "grad_norm": 7.9375, "learning_rate": 1.6941092645430535e-08, "loss": 1.3125, "step": 39281 }, { "epoch": 98.205, "grad_norm": 6.59375, "learning_rate": 1.6894014639067258e-08, "loss": 1.2409, "step": 39282 }, { "epoch": 98.2075, "grad_norm": 6.09375, "learning_rate": 1.684700208154699e-08, "loss": 1.2131, "step": 39283 }, { "epoch": 98.21, "grad_norm": 8.75, "learning_rate": 1.680005497317838e-08, "loss": 1.2534, "step": 39284 }, { "epoch": 98.2125, "grad_norm": 9.25, "learning_rate": 1.6753173314268957e-08, "loss": 1.3921, "step": 39285 }, { "epoch": 98.215, "grad_norm": 11.75, "learning_rate": 1.6706357105126247e-08, "loss": 1.1445, "step": 39286 }, { "epoch": 98.2175, "grad_norm": 7.75, "learning_rate": 1.6659606346057787e-08, "loss": 1.2053, "step": 39287 }, { "epoch": 98.22, "grad_norm": 7.8125, "learning_rate": 1.661292103736889e-08, "loss": 1.2855, "step": 39288 }, { "epoch": 98.2225, "grad_norm": 7.0625, "learning_rate": 1.6566301179365974e-08, "loss": 1.4981, "step": 39289 }, { "epoch": 98.225, "grad_norm": 6.9375, "learning_rate": 1.6519746772355462e-08, "loss": 1.1826, "step": 39290 }, { "epoch": 98.2275, "grad_norm": 7.1875, "learning_rate": 1.647325781664155e-08, "loss": 1.2957, "step": 39291 }, { "epoch": 98.23, "grad_norm": 9.75, "learning_rate": 1.642683431252956e-08, "loss": 1.5309, "step": 39292 }, { "epoch": 98.2325, "grad_norm": 8.5, "learning_rate": 1.6380476260323686e-08, "loss": 1.2733, "step": 39293 }, { "epoch": 98.235, "grad_norm": 7.53125, "learning_rate": 1.633418366032813e-08, "loss": 1.2804, "step": 39294 }, { "epoch": 98.2375, "grad_norm": 8.4375, "learning_rate": 1.628795651284598e-08, "loss": 1.4676, "step": 39295 }, { "epoch": 98.24, "grad_norm": 7.5625, "learning_rate": 1.6241794818180333e-08, "loss": 1.2285, "step": 39296 }, { "epoch": 98.2425, "grad_norm": 8.9375, "learning_rate": 1.6195698576634277e-08, "loss": 1.3655, "step": 39297 }, { "epoch": 98.245, "grad_norm": 6.0625, "learning_rate": 1.6149667788508684e-08, "loss": 1.4165, "step": 39298 }, { "epoch": 98.2475, "grad_norm": 8.375, "learning_rate": 1.610370245410775e-08, "loss": 1.3418, "step": 39299 }, { "epoch": 98.25, "grad_norm": 8.5625, "learning_rate": 1.605780257373124e-08, "loss": 1.287, "step": 39300 }, { "epoch": 98.2525, "grad_norm": 6.34375, "learning_rate": 1.6011968147678915e-08, "loss": 1.3136, "step": 39301 }, { "epoch": 98.255, "grad_norm": 5.9375, "learning_rate": 1.596619917625386e-08, "loss": 1.2562, "step": 39302 }, { "epoch": 98.2575, "grad_norm": 6.0625, "learning_rate": 1.5920495659754732e-08, "loss": 1.2091, "step": 39303 }, { "epoch": 98.26, "grad_norm": 9.125, "learning_rate": 1.5874857598481286e-08, "loss": 1.2753, "step": 39304 }, { "epoch": 98.2625, "grad_norm": 8.125, "learning_rate": 1.5829284992732173e-08, "loss": 1.3783, "step": 39305 }, { "epoch": 98.265, "grad_norm": 7.15625, "learning_rate": 1.5783777842806047e-08, "loss": 1.2919, "step": 39306 }, { "epoch": 98.2675, "grad_norm": 9.1875, "learning_rate": 1.5738336149003776e-08, "loss": 1.423, "step": 39307 }, { "epoch": 98.27, "grad_norm": 9.9375, "learning_rate": 1.5692959911620676e-08, "loss": 1.3975, "step": 39308 }, { "epoch": 98.2725, "grad_norm": 10.0, "learning_rate": 1.564764913095429e-08, "loss": 1.2357, "step": 39309 }, { "epoch": 98.275, "grad_norm": 7.84375, "learning_rate": 1.5602403807303266e-08, "loss": 1.2808, "step": 39310 }, { "epoch": 98.2775, "grad_norm": 8.8125, "learning_rate": 1.555722394096293e-08, "loss": 1.3839, "step": 39311 }, { "epoch": 98.28, "grad_norm": 8.3125, "learning_rate": 1.5512109532229703e-08, "loss": 1.2449, "step": 39312 }, { "epoch": 98.2825, "grad_norm": 6.40625, "learning_rate": 1.5467060581400018e-08, "loss": 1.3192, "step": 39313 }, { "epoch": 98.285, "grad_norm": 10.5, "learning_rate": 1.5422077088768082e-08, "loss": 1.5285, "step": 39314 }, { "epoch": 98.2875, "grad_norm": 8.0, "learning_rate": 1.537715905463033e-08, "loss": 1.2838, "step": 39315 }, { "epoch": 98.29, "grad_norm": 9.375, "learning_rate": 1.5332306479279858e-08, "loss": 1.3622, "step": 39316 }, { "epoch": 98.2925, "grad_norm": 6.15625, "learning_rate": 1.5287519363010872e-08, "loss": 1.4038, "step": 39317 }, { "epoch": 98.295, "grad_norm": 9.25, "learning_rate": 1.5242797706117586e-08, "loss": 1.3826, "step": 39318 }, { "epoch": 98.2975, "grad_norm": 9.75, "learning_rate": 1.5198141508893093e-08, "loss": 1.415, "step": 39319 }, { "epoch": 98.3, "grad_norm": 7.71875, "learning_rate": 1.5153550771630498e-08, "loss": 1.3692, "step": 39320 }, { "epoch": 98.3025, "grad_norm": 8.5625, "learning_rate": 1.510902549462068e-08, "loss": 1.3953, "step": 39321 }, { "epoch": 98.305, "grad_norm": 7.1875, "learning_rate": 1.506456567815673e-08, "loss": 1.3683, "step": 39322 }, { "epoch": 98.3075, "grad_norm": 13.6875, "learning_rate": 1.502017132253064e-08, "loss": 1.2395, "step": 39323 }, { "epoch": 98.31, "grad_norm": 8.125, "learning_rate": 1.4975842428032182e-08, "loss": 1.4018, "step": 39324 }, { "epoch": 98.3125, "grad_norm": 7.125, "learning_rate": 1.4931578994952235e-08, "loss": 1.3824, "step": 39325 }, { "epoch": 98.315, "grad_norm": 7.1875, "learning_rate": 1.488738102358167e-08, "loss": 1.3747, "step": 39326 }, { "epoch": 98.3175, "grad_norm": 6.8125, "learning_rate": 1.4843248514210262e-08, "loss": 1.2719, "step": 39327 }, { "epoch": 98.32, "grad_norm": 8.25, "learning_rate": 1.4799181467125557e-08, "loss": 1.3312, "step": 39328 }, { "epoch": 98.3225, "grad_norm": 10.125, "learning_rate": 1.475517988261954e-08, "loss": 1.4006, "step": 39329 }, { "epoch": 98.325, "grad_norm": 8.5, "learning_rate": 1.4711243760978655e-08, "loss": 1.4421, "step": 39330 }, { "epoch": 98.3275, "grad_norm": 7.625, "learning_rate": 1.4667373102490445e-08, "loss": 1.22, "step": 39331 }, { "epoch": 98.33, "grad_norm": 8.0625, "learning_rate": 1.462356790744357e-08, "loss": 1.4094, "step": 39332 }, { "epoch": 98.3325, "grad_norm": 9.0625, "learning_rate": 1.4579828176124466e-08, "loss": 1.3101, "step": 39333 }, { "epoch": 98.335, "grad_norm": 8.3125, "learning_rate": 1.4536153908821793e-08, "loss": 1.3371, "step": 39334 }, { "epoch": 98.3375, "grad_norm": 5.46875, "learning_rate": 1.4492545105819767e-08, "loss": 1.2303, "step": 39335 }, { "epoch": 98.34, "grad_norm": 6.09375, "learning_rate": 1.4449001767404824e-08, "loss": 1.3575, "step": 39336 }, { "epoch": 98.3425, "grad_norm": 8.375, "learning_rate": 1.4405523893863404e-08, "loss": 1.2082, "step": 39337 }, { "epoch": 98.345, "grad_norm": 6.125, "learning_rate": 1.4362111485478613e-08, "loss": 1.2153, "step": 39338 }, { "epoch": 98.3475, "grad_norm": 6.84375, "learning_rate": 1.431876454253689e-08, "loss": 1.2435, "step": 39339 }, { "epoch": 98.35, "grad_norm": 9.6875, "learning_rate": 1.4275483065321338e-08, "loss": 1.3948, "step": 39340 }, { "epoch": 98.3525, "grad_norm": 9.25, "learning_rate": 1.4232267054116178e-08, "loss": 1.2543, "step": 39341 }, { "epoch": 98.355, "grad_norm": 7.90625, "learning_rate": 1.4189116509205625e-08, "loss": 1.2583, "step": 39342 }, { "epoch": 98.3575, "grad_norm": 11.4375, "learning_rate": 1.4146031430870566e-08, "loss": 1.4476, "step": 39343 }, { "epoch": 98.36, "grad_norm": 6.84375, "learning_rate": 1.4103011819395218e-08, "loss": 1.1968, "step": 39344 }, { "epoch": 98.3625, "grad_norm": 9.25, "learning_rate": 1.4060057675060468e-08, "loss": 1.3718, "step": 39345 }, { "epoch": 98.365, "grad_norm": 5.78125, "learning_rate": 1.4017168998148312e-08, "loss": 1.2359, "step": 39346 }, { "epoch": 98.3675, "grad_norm": 7.84375, "learning_rate": 1.3974345788940747e-08, "loss": 1.371, "step": 39347 }, { "epoch": 98.37, "grad_norm": 5.6875, "learning_rate": 1.3931588047716438e-08, "loss": 1.2226, "step": 39348 }, { "epoch": 98.3725, "grad_norm": 9.0625, "learning_rate": 1.3888895774758494e-08, "loss": 1.2306, "step": 39349 }, { "epoch": 98.375, "grad_norm": 9.0625, "learning_rate": 1.3846268970344467e-08, "loss": 1.3545, "step": 39350 }, { "epoch": 98.3775, "grad_norm": 7.21875, "learning_rate": 1.3803707634754138e-08, "loss": 1.2695, "step": 39351 }, { "epoch": 98.38, "grad_norm": 6.53125, "learning_rate": 1.376121176826728e-08, "loss": 1.3922, "step": 39352 }, { "epoch": 98.3825, "grad_norm": 16.75, "learning_rate": 1.371878137116367e-08, "loss": 1.3718, "step": 39353 }, { "epoch": 98.385, "grad_norm": 8.0625, "learning_rate": 1.3676416443718643e-08, "loss": 1.4016, "step": 39354 }, { "epoch": 98.3875, "grad_norm": 8.8125, "learning_rate": 1.3634116986211975e-08, "loss": 1.3233, "step": 39355 }, { "epoch": 98.39, "grad_norm": 8.3125, "learning_rate": 1.3591882998920113e-08, "loss": 1.2839, "step": 39356 }, { "epoch": 98.3925, "grad_norm": 8.625, "learning_rate": 1.3549714482120612e-08, "loss": 1.4245, "step": 39357 }, { "epoch": 98.395, "grad_norm": 7.9375, "learning_rate": 1.3507611436089918e-08, "loss": 1.1459, "step": 39358 }, { "epoch": 98.3975, "grad_norm": 6.09375, "learning_rate": 1.3465573861102255e-08, "loss": 1.2509, "step": 39359 }, { "epoch": 98.4, "grad_norm": 8.0625, "learning_rate": 1.3423601757436289e-08, "loss": 1.2422, "step": 39360 }, { "epoch": 98.4025, "grad_norm": 5.96875, "learning_rate": 1.3381695125364025e-08, "loss": 1.3953, "step": 39361 }, { "epoch": 98.405, "grad_norm": 6.15625, "learning_rate": 1.333985396516302e-08, "loss": 1.3489, "step": 39362 }, { "epoch": 98.4075, "grad_norm": 7.6875, "learning_rate": 1.3298078277105275e-08, "loss": 1.255, "step": 39363 }, { "epoch": 98.41, "grad_norm": 8.5625, "learning_rate": 1.325636806146613e-08, "loss": 1.3373, "step": 39364 }, { "epoch": 98.4125, "grad_norm": 7.59375, "learning_rate": 1.3214723318518697e-08, "loss": 1.2898, "step": 39365 }, { "epoch": 98.415, "grad_norm": 6.3125, "learning_rate": 1.317314404853609e-08, "loss": 1.3356, "step": 39366 }, { "epoch": 98.4175, "grad_norm": 6.125, "learning_rate": 1.3131630251790317e-08, "loss": 1.2477, "step": 39367 }, { "epoch": 98.42, "grad_norm": 5.09375, "learning_rate": 1.309018192855449e-08, "loss": 1.2985, "step": 39368 }, { "epoch": 98.4225, "grad_norm": 7.59375, "learning_rate": 1.3048799079099505e-08, "loss": 1.3082, "step": 39369 }, { "epoch": 98.425, "grad_norm": 10.875, "learning_rate": 1.3007481703696257e-08, "loss": 1.2666, "step": 39370 }, { "epoch": 98.4275, "grad_norm": 14.0, "learning_rate": 1.2966229802616748e-08, "loss": 1.2893, "step": 39371 }, { "epoch": 98.43, "grad_norm": 6.09375, "learning_rate": 1.2925043376130764e-08, "loss": 1.2946, "step": 39372 }, { "epoch": 98.4325, "grad_norm": 6.46875, "learning_rate": 1.28839224245092e-08, "loss": 1.3321, "step": 39373 }, { "epoch": 98.435, "grad_norm": 6.9375, "learning_rate": 1.2842866948020727e-08, "loss": 1.2788, "step": 39374 }, { "epoch": 98.4375, "grad_norm": 7.71875, "learning_rate": 1.2801876946935133e-08, "loss": 1.4278, "step": 39375 }, { "epoch": 98.44, "grad_norm": 8.625, "learning_rate": 1.276095242151998e-08, "loss": 1.4905, "step": 39376 }, { "epoch": 98.4425, "grad_norm": 6.9375, "learning_rate": 1.272009337204394e-08, "loss": 1.374, "step": 39377 }, { "epoch": 98.445, "grad_norm": 6.875, "learning_rate": 1.26792997987768e-08, "loss": 1.2789, "step": 39378 }, { "epoch": 98.4475, "grad_norm": 5.53125, "learning_rate": 1.2638571701982794e-08, "loss": 1.2262, "step": 39379 }, { "epoch": 98.45, "grad_norm": 6.78125, "learning_rate": 1.2597909081931702e-08, "loss": 1.5896, "step": 39380 }, { "epoch": 98.4525, "grad_norm": 11.0, "learning_rate": 1.2557311938888872e-08, "loss": 1.198, "step": 39381 }, { "epoch": 98.455, "grad_norm": 7.5, "learning_rate": 1.2516780273120754e-08, "loss": 1.2783, "step": 39382 }, { "epoch": 98.4575, "grad_norm": 6.34375, "learning_rate": 1.2476314084892694e-08, "loss": 1.1168, "step": 39383 }, { "epoch": 98.46, "grad_norm": 9.25, "learning_rate": 1.2435913374471142e-08, "loss": 1.4065, "step": 39384 }, { "epoch": 98.4625, "grad_norm": 5.53125, "learning_rate": 1.2395578142119225e-08, "loss": 1.2927, "step": 39385 }, { "epoch": 98.465, "grad_norm": 7.40625, "learning_rate": 1.2355308388102284e-08, "loss": 1.1804, "step": 39386 }, { "epoch": 98.4675, "grad_norm": 8.5625, "learning_rate": 1.231510411268344e-08, "loss": 1.2565, "step": 39387 }, { "epoch": 98.47, "grad_norm": 8.625, "learning_rate": 1.227496531612804e-08, "loss": 1.4006, "step": 39388 }, { "epoch": 98.4725, "grad_norm": 9.9375, "learning_rate": 1.2234891998698096e-08, "loss": 1.2054, "step": 39389 }, { "epoch": 98.475, "grad_norm": 6.90625, "learning_rate": 1.2194884160656728e-08, "loss": 1.4901, "step": 39390 }, { "epoch": 98.4775, "grad_norm": 6.96875, "learning_rate": 1.2154941802265952e-08, "loss": 1.251, "step": 39391 }, { "epoch": 98.48, "grad_norm": 7.53125, "learning_rate": 1.2115064923787778e-08, "loss": 1.236, "step": 39392 }, { "epoch": 98.4825, "grad_norm": 9.3125, "learning_rate": 1.207525352548311e-08, "loss": 1.1679, "step": 39393 }, { "epoch": 98.485, "grad_norm": 9.5, "learning_rate": 1.203550760761396e-08, "loss": 1.4386, "step": 39394 }, { "epoch": 98.4875, "grad_norm": 7.125, "learning_rate": 1.1995827170440122e-08, "loss": 1.2925, "step": 39395 }, { "epoch": 98.49, "grad_norm": 5.90625, "learning_rate": 1.1956212214221385e-08, "loss": 1.2793, "step": 39396 }, { "epoch": 98.4925, "grad_norm": 7.5625, "learning_rate": 1.1916662739218655e-08, "loss": 1.2342, "step": 39397 }, { "epoch": 98.495, "grad_norm": 8.0625, "learning_rate": 1.187717874569061e-08, "loss": 1.3492, "step": 39398 }, { "epoch": 98.4975, "grad_norm": 7.9375, "learning_rate": 1.1837760233895934e-08, "loss": 1.3098, "step": 39399 }, { "epoch": 98.5, "grad_norm": 9.0, "learning_rate": 1.179840720409331e-08, "loss": 1.3741, "step": 39400 }, { "epoch": 98.5025, "grad_norm": 8.5, "learning_rate": 1.1759119656540308e-08, "loss": 1.3734, "step": 39401 }, { "epoch": 98.505, "grad_norm": 6.53125, "learning_rate": 1.17198975914945e-08, "loss": 1.2467, "step": 39402 }, { "epoch": 98.5075, "grad_norm": 8.3125, "learning_rate": 1.168074100921346e-08, "loss": 1.3475, "step": 39403 }, { "epoch": 98.51, "grad_norm": 6.03125, "learning_rate": 1.1641649909954755e-08, "loss": 1.3302, "step": 39404 }, { "epoch": 98.5125, "grad_norm": 6.75, "learning_rate": 1.160262429397263e-08, "loss": 1.2518, "step": 39405 }, { "epoch": 98.515, "grad_norm": 10.375, "learning_rate": 1.1563664161524657e-08, "loss": 1.3067, "step": 39406 }, { "epoch": 98.5175, "grad_norm": 7.84375, "learning_rate": 1.1524769512865075e-08, "loss": 1.3735, "step": 39407 }, { "epoch": 98.52, "grad_norm": 6.625, "learning_rate": 1.1485940348249235e-08, "loss": 1.2867, "step": 39408 }, { "epoch": 98.5225, "grad_norm": 7.6875, "learning_rate": 1.144717666793249e-08, "loss": 1.2588, "step": 39409 }, { "epoch": 98.525, "grad_norm": 8.25, "learning_rate": 1.1408478472167972e-08, "loss": 1.4639, "step": 39410 }, { "epoch": 98.5275, "grad_norm": 5.75, "learning_rate": 1.1369845761209918e-08, "loss": 1.3952, "step": 39411 }, { "epoch": 98.53, "grad_norm": 6.1875, "learning_rate": 1.1331278535311464e-08, "loss": 1.2536, "step": 39412 }, { "epoch": 98.5325, "grad_norm": 6.28125, "learning_rate": 1.1292776794725735e-08, "loss": 1.2443, "step": 39413 }, { "epoch": 98.535, "grad_norm": 7.84375, "learning_rate": 1.1254340539703646e-08, "loss": 1.2869, "step": 39414 }, { "epoch": 98.5375, "grad_norm": 8.375, "learning_rate": 1.1215969770499435e-08, "loss": 1.3922, "step": 39415 }, { "epoch": 98.54, "grad_norm": 8.3125, "learning_rate": 1.1177664487362905e-08, "loss": 1.227, "step": 39416 }, { "epoch": 98.5425, "grad_norm": 7.4375, "learning_rate": 1.1139424690547185e-08, "loss": 1.3309, "step": 39417 }, { "epoch": 98.545, "grad_norm": 7.53125, "learning_rate": 1.1101250380300965e-08, "loss": 1.284, "step": 39418 }, { "epoch": 98.5475, "grad_norm": 7.25, "learning_rate": 1.1063141556874047e-08, "loss": 1.3724, "step": 39419 }, { "epoch": 98.55, "grad_norm": 9.1875, "learning_rate": 1.102509822051845e-08, "loss": 1.2463, "step": 39420 }, { "epoch": 98.5525, "grad_norm": 8.8125, "learning_rate": 1.0987120371482862e-08, "loss": 1.2556, "step": 39421 }, { "epoch": 98.555, "grad_norm": 7.03125, "learning_rate": 1.0949208010015977e-08, "loss": 1.3794, "step": 39422 }, { "epoch": 98.5575, "grad_norm": 7.5, "learning_rate": 1.0911361136365373e-08, "loss": 1.5103, "step": 39423 }, { "epoch": 98.56, "grad_norm": 5.3125, "learning_rate": 1.087357975078085e-08, "loss": 1.1669, "step": 39424 }, { "epoch": 98.5625, "grad_norm": 9.8125, "learning_rate": 1.0835863853509988e-08, "loss": 1.3619, "step": 39425 }, { "epoch": 98.565, "grad_norm": 8.25, "learning_rate": 1.0798213444798145e-08, "loss": 1.3558, "step": 39426 }, { "epoch": 98.5675, "grad_norm": 7.40625, "learning_rate": 1.0760628524894012e-08, "loss": 1.2919, "step": 39427 }, { "epoch": 98.57, "grad_norm": 7.75, "learning_rate": 1.072310909404406e-08, "loss": 1.1998, "step": 39428 }, { "epoch": 98.5725, "grad_norm": 11.0, "learning_rate": 1.0685655152492536e-08, "loss": 1.2622, "step": 39429 }, { "epoch": 98.575, "grad_norm": 5.1875, "learning_rate": 1.064826670048702e-08, "loss": 1.3122, "step": 39430 }, { "epoch": 98.5775, "grad_norm": 6.40625, "learning_rate": 1.0610943738271761e-08, "loss": 1.2773, "step": 39431 }, { "epoch": 98.58, "grad_norm": 9.1875, "learning_rate": 1.057368626609101e-08, "loss": 1.4767, "step": 39432 }, { "epoch": 98.5825, "grad_norm": 12.625, "learning_rate": 1.0536494284189015e-08, "loss": 1.4753, "step": 39433 }, { "epoch": 98.585, "grad_norm": 7.125, "learning_rate": 1.0499367792811132e-08, "loss": 1.2455, "step": 39434 }, { "epoch": 98.5875, "grad_norm": 8.625, "learning_rate": 1.0462306792198285e-08, "loss": 1.3086, "step": 39435 }, { "epoch": 98.59, "grad_norm": 8.1875, "learning_rate": 1.042531128259583e-08, "loss": 1.2523, "step": 39436 }, { "epoch": 98.5925, "grad_norm": 8.125, "learning_rate": 1.0388381264244684e-08, "loss": 1.3699, "step": 39437 }, { "epoch": 98.595, "grad_norm": 7.03125, "learning_rate": 1.035151673738799e-08, "loss": 1.2782, "step": 39438 }, { "epoch": 98.5975, "grad_norm": 8.8125, "learning_rate": 1.0314717702266663e-08, "loss": 1.2666, "step": 39439 }, { "epoch": 98.6, "grad_norm": 6.9375, "learning_rate": 1.0277984159122734e-08, "loss": 1.4335, "step": 39440 }, { "epoch": 98.6025, "grad_norm": 9.8125, "learning_rate": 1.0241316108196009e-08, "loss": 1.1929, "step": 39441 }, { "epoch": 98.605, "grad_norm": 7.5625, "learning_rate": 1.0204713549727407e-08, "loss": 1.3374, "step": 39442 }, { "epoch": 98.6075, "grad_norm": 6.6875, "learning_rate": 1.0168176483957848e-08, "loss": 1.1463, "step": 39443 }, { "epoch": 98.61, "grad_norm": 8.5, "learning_rate": 1.0131704911124917e-08, "loss": 1.2666, "step": 39444 }, { "epoch": 98.6125, "grad_norm": 8.375, "learning_rate": 1.0095298831469535e-08, "loss": 1.3523, "step": 39445 }, { "epoch": 98.615, "grad_norm": 9.5, "learning_rate": 1.0058958245229289e-08, "loss": 1.3027, "step": 39446 }, { "epoch": 98.6175, "grad_norm": 7.40625, "learning_rate": 1.0022683152642876e-08, "loss": 1.3495, "step": 39447 }, { "epoch": 98.62, "grad_norm": 7.84375, "learning_rate": 9.986473553946774e-09, "loss": 1.3399, "step": 39448 }, { "epoch": 98.6225, "grad_norm": 7.28125, "learning_rate": 9.95032944938079e-09, "loss": 1.3836, "step": 39449 }, { "epoch": 98.625, "grad_norm": 7.1875, "learning_rate": 9.914250839180296e-09, "loss": 1.276, "step": 39450 }, { "epoch": 98.6275, "grad_norm": 7.8125, "learning_rate": 9.878237723582873e-09, "loss": 1.4051, "step": 39451 }, { "epoch": 98.63, "grad_norm": 5.625, "learning_rate": 9.842290102823893e-09, "loss": 1.29, "step": 39452 }, { "epoch": 98.6325, "grad_norm": 15.1875, "learning_rate": 9.806407977137611e-09, "loss": 1.4118, "step": 39453 }, { "epoch": 98.635, "grad_norm": 5.15625, "learning_rate": 9.770591346761616e-09, "loss": 1.5012, "step": 39454 }, { "epoch": 98.6375, "grad_norm": 7.3125, "learning_rate": 9.734840211930164e-09, "loss": 1.2391, "step": 39455 }, { "epoch": 98.64, "grad_norm": 8.4375, "learning_rate": 9.699154572877511e-09, "loss": 1.2928, "step": 39456 }, { "epoch": 98.6425, "grad_norm": 7.9375, "learning_rate": 9.663534429836807e-09, "loss": 1.2791, "step": 39457 }, { "epoch": 98.645, "grad_norm": 10.6875, "learning_rate": 9.627979783042307e-09, "loss": 1.2635, "step": 39458 }, { "epoch": 98.6475, "grad_norm": 8.9375, "learning_rate": 9.592490632727158e-09, "loss": 1.4162, "step": 39459 }, { "epoch": 98.65, "grad_norm": 10.125, "learning_rate": 9.557066979123398e-09, "loss": 1.2934, "step": 39460 }, { "epoch": 98.6525, "grad_norm": 5.875, "learning_rate": 9.521708822464171e-09, "loss": 1.1988, "step": 39461 }, { "epoch": 98.655, "grad_norm": 5.78125, "learning_rate": 9.486416162980405e-09, "loss": 1.3441, "step": 39462 }, { "epoch": 98.6575, "grad_norm": 7.09375, "learning_rate": 9.451189000905247e-09, "loss": 1.282, "step": 39463 }, { "epoch": 98.66, "grad_norm": 6.6875, "learning_rate": 9.416027336466294e-09, "loss": 1.3133, "step": 39464 }, { "epoch": 98.6625, "grad_norm": 6.5625, "learning_rate": 9.38093116989669e-09, "loss": 1.3426, "step": 39465 }, { "epoch": 98.665, "grad_norm": 8.5625, "learning_rate": 9.345900501427363e-09, "loss": 1.3588, "step": 39466 }, { "epoch": 98.6675, "grad_norm": 4.875, "learning_rate": 9.310935331284799e-09, "loss": 1.3167, "step": 39467 }, { "epoch": 98.67, "grad_norm": 7.96875, "learning_rate": 9.276035659701032e-09, "loss": 1.2415, "step": 39468 }, { "epoch": 98.6725, "grad_norm": 5.875, "learning_rate": 9.24120148690255e-09, "loss": 1.2177, "step": 39469 }, { "epoch": 98.675, "grad_norm": 11.9375, "learning_rate": 9.206432813120281e-09, "loss": 1.3467, "step": 39470 }, { "epoch": 98.6775, "grad_norm": 7.84375, "learning_rate": 9.171729638579595e-09, "loss": 1.2338, "step": 39471 }, { "epoch": 98.68, "grad_norm": 6.25, "learning_rate": 9.137091963510314e-09, "loss": 1.3238, "step": 39472 }, { "epoch": 98.6825, "grad_norm": 8.75, "learning_rate": 9.10251978813781e-09, "loss": 1.4059, "step": 39473 }, { "epoch": 98.685, "grad_norm": 9.25, "learning_rate": 9.068013112689678e-09, "loss": 1.616, "step": 39474 }, { "epoch": 98.6875, "grad_norm": 8.875, "learning_rate": 9.033571937391294e-09, "loss": 1.5529, "step": 39475 }, { "epoch": 98.69, "grad_norm": 7.1875, "learning_rate": 8.999196262469146e-09, "loss": 1.3822, "step": 39476 }, { "epoch": 98.6925, "grad_norm": 10.0625, "learning_rate": 8.964886088148606e-09, "loss": 1.2699, "step": 39477 }, { "epoch": 98.695, "grad_norm": 4.625, "learning_rate": 8.93064141465505e-09, "loss": 1.2981, "step": 39478 }, { "epoch": 98.6975, "grad_norm": 14.375, "learning_rate": 8.896462242211635e-09, "loss": 1.4773, "step": 39479 }, { "epoch": 98.7, "grad_norm": 6.6875, "learning_rate": 8.862348571043733e-09, "loss": 1.2383, "step": 39480 }, { "epoch": 98.7025, "grad_norm": 5.375, "learning_rate": 8.82830040137339e-09, "loss": 1.3051, "step": 39481 }, { "epoch": 98.705, "grad_norm": 7.625, "learning_rate": 8.794317733425983e-09, "loss": 1.407, "step": 39482 }, { "epoch": 98.7075, "grad_norm": 5.90625, "learning_rate": 8.760400567423555e-09, "loss": 1.3392, "step": 39483 }, { "epoch": 98.71, "grad_norm": 9.1875, "learning_rate": 8.72654890358704e-09, "loss": 1.2063, "step": 39484 }, { "epoch": 98.7125, "grad_norm": 6.9375, "learning_rate": 8.692762742139594e-09, "loss": 1.2406, "step": 39485 }, { "epoch": 98.715, "grad_norm": 8.625, "learning_rate": 8.65904208330326e-09, "loss": 1.1954, "step": 39486 }, { "epoch": 98.7175, "grad_norm": 7.21875, "learning_rate": 8.625386927297862e-09, "loss": 1.3625, "step": 39487 }, { "epoch": 98.72, "grad_norm": 7.90625, "learning_rate": 8.591797274344338e-09, "loss": 1.3243, "step": 39488 }, { "epoch": 98.7225, "grad_norm": 7.9375, "learning_rate": 8.558273124663618e-09, "loss": 1.2437, "step": 39489 }, { "epoch": 98.725, "grad_norm": 5.4375, "learning_rate": 8.524814478474419e-09, "loss": 1.4887, "step": 39490 }, { "epoch": 98.7275, "grad_norm": 7.4375, "learning_rate": 8.491421335997674e-09, "loss": 1.2164, "step": 39491 }, { "epoch": 98.73, "grad_norm": 7.6875, "learning_rate": 8.458093697450987e-09, "loss": 1.389, "step": 39492 }, { "epoch": 98.7325, "grad_norm": 8.875, "learning_rate": 8.424831563053072e-09, "loss": 1.3244, "step": 39493 }, { "epoch": 98.735, "grad_norm": 7.71875, "learning_rate": 8.391634933022641e-09, "loss": 1.3865, "step": 39494 }, { "epoch": 98.7375, "grad_norm": 7.59375, "learning_rate": 8.35850380757508e-09, "loss": 1.4837, "step": 39495 }, { "epoch": 98.74, "grad_norm": 7.3125, "learning_rate": 8.325438186931323e-09, "loss": 1.3087, "step": 39496 }, { "epoch": 98.7425, "grad_norm": 10.8125, "learning_rate": 8.292438071305642e-09, "loss": 1.2337, "step": 39497 }, { "epoch": 98.745, "grad_norm": 7.21875, "learning_rate": 8.259503460914531e-09, "loss": 1.2955, "step": 39498 }, { "epoch": 98.7475, "grad_norm": 6.875, "learning_rate": 8.226634355973372e-09, "loss": 1.3465, "step": 39499 }, { "epoch": 98.75, "grad_norm": 8.4375, "learning_rate": 8.193830756699773e-09, "loss": 1.2942, "step": 39500 }, { "epoch": 98.7525, "grad_norm": 6.28125, "learning_rate": 8.161092663308001e-09, "loss": 1.2742, "step": 39501 }, { "epoch": 98.755, "grad_norm": 10.0625, "learning_rate": 8.128420076011223e-09, "loss": 1.2576, "step": 39502 }, { "epoch": 98.7575, "grad_norm": 6.1875, "learning_rate": 8.095812995024821e-09, "loss": 1.3591, "step": 39503 }, { "epoch": 98.76, "grad_norm": 9.75, "learning_rate": 8.063271420563068e-09, "loss": 1.3793, "step": 39504 }, { "epoch": 98.7625, "grad_norm": 6.15625, "learning_rate": 8.030795352838016e-09, "loss": 1.3383, "step": 39505 }, { "epoch": 98.765, "grad_norm": 10.625, "learning_rate": 7.99838479206283e-09, "loss": 1.4839, "step": 39506 }, { "epoch": 98.7675, "grad_norm": 7.625, "learning_rate": 7.966039738451781e-09, "loss": 1.3334, "step": 39507 }, { "epoch": 98.77, "grad_norm": 8.0625, "learning_rate": 7.933760192214702e-09, "loss": 1.1732, "step": 39508 }, { "epoch": 98.7725, "grad_norm": 8.9375, "learning_rate": 7.901546153563643e-09, "loss": 1.4645, "step": 39509 }, { "epoch": 98.775, "grad_norm": 8.5, "learning_rate": 7.869397622710662e-09, "loss": 1.1964, "step": 39510 }, { "epoch": 98.7775, "grad_norm": 10.625, "learning_rate": 7.837314599865587e-09, "loss": 1.3885, "step": 39511 }, { "epoch": 98.78, "grad_norm": 8.625, "learning_rate": 7.80529708523936e-09, "loss": 1.3224, "step": 39512 }, { "epoch": 98.7825, "grad_norm": 6.34375, "learning_rate": 7.773345079041816e-09, "loss": 1.3421, "step": 39513 }, { "epoch": 98.785, "grad_norm": 7.9375, "learning_rate": 7.741458581482786e-09, "loss": 1.3319, "step": 39514 }, { "epoch": 98.7875, "grad_norm": 5.40625, "learning_rate": 7.70963759277099e-09, "loss": 1.3031, "step": 39515 }, { "epoch": 98.79, "grad_norm": 4.90625, "learning_rate": 7.677882113114043e-09, "loss": 1.3766, "step": 39516 }, { "epoch": 98.7925, "grad_norm": 9.4375, "learning_rate": 7.646192142721775e-09, "loss": 1.4603, "step": 39517 }, { "epoch": 98.795, "grad_norm": 7.8125, "learning_rate": 7.614567681800688e-09, "loss": 1.4683, "step": 39518 }, { "epoch": 98.7975, "grad_norm": 6.40625, "learning_rate": 7.583008730559505e-09, "loss": 1.3522, "step": 39519 }, { "epoch": 98.8, "grad_norm": 10.125, "learning_rate": 7.551515289203615e-09, "loss": 1.3578, "step": 39520 }, { "epoch": 98.8025, "grad_norm": 6.28125, "learning_rate": 7.52008735793952e-09, "loss": 1.191, "step": 39521 }, { "epoch": 98.805, "grad_norm": 8.0625, "learning_rate": 7.488724936974834e-09, "loss": 1.1978, "step": 39522 }, { "epoch": 98.8075, "grad_norm": 6.78125, "learning_rate": 7.457428026513836e-09, "loss": 1.2662, "step": 39523 }, { "epoch": 98.81, "grad_norm": 6.53125, "learning_rate": 7.426196626761917e-09, "loss": 1.3004, "step": 39524 }, { "epoch": 98.8125, "grad_norm": 8.5625, "learning_rate": 7.395030737924469e-09, "loss": 1.1865, "step": 39525 }, { "epoch": 98.815, "grad_norm": 8.625, "learning_rate": 7.363930360204663e-09, "loss": 1.2711, "step": 39526 }, { "epoch": 98.8175, "grad_norm": 8.125, "learning_rate": 7.33289549380678e-09, "loss": 1.2835, "step": 39527 }, { "epoch": 98.82, "grad_norm": 7.65625, "learning_rate": 7.3019261389351005e-09, "loss": 1.3175, "step": 39528 }, { "epoch": 98.8225, "grad_norm": 8.1875, "learning_rate": 7.2710222957927955e-09, "loss": 1.274, "step": 39529 }, { "epoch": 98.825, "grad_norm": 8.3125, "learning_rate": 7.240183964580816e-09, "loss": 1.4003, "step": 39530 }, { "epoch": 98.8275, "grad_norm": 7.53125, "learning_rate": 7.209411145502332e-09, "loss": 1.3334, "step": 39531 }, { "epoch": 98.83, "grad_norm": 4.3125, "learning_rate": 7.178703838759405e-09, "loss": 1.1659, "step": 39532 }, { "epoch": 98.8325, "grad_norm": 11.9375, "learning_rate": 7.148062044552984e-09, "loss": 1.2922, "step": 39533 }, { "epoch": 98.835, "grad_norm": 7.875, "learning_rate": 7.11748576308402e-09, "loss": 1.2012, "step": 39534 }, { "epoch": 98.8375, "grad_norm": 8.3125, "learning_rate": 7.0869749945523534e-09, "loss": 1.3513, "step": 39535 }, { "epoch": 98.84, "grad_norm": 5.71875, "learning_rate": 7.056529739158935e-09, "loss": 1.4278, "step": 39536 }, { "epoch": 98.8425, "grad_norm": 6.25, "learning_rate": 7.026149997103604e-09, "loss": 1.5199, "step": 39537 }, { "epoch": 98.845, "grad_norm": 6.28125, "learning_rate": 6.995835768583981e-09, "loss": 1.2972, "step": 39538 }, { "epoch": 98.8475, "grad_norm": 9.25, "learning_rate": 6.965587053801015e-09, "loss": 1.3238, "step": 39539 }, { "epoch": 98.85, "grad_norm": 7.46875, "learning_rate": 6.935403852950107e-09, "loss": 1.2877, "step": 39540 }, { "epoch": 98.8525, "grad_norm": 7.78125, "learning_rate": 6.905286166232206e-09, "loss": 1.3258, "step": 39541 }, { "epoch": 98.855, "grad_norm": 8.9375, "learning_rate": 6.875233993842711e-09, "loss": 1.4362, "step": 39542 }, { "epoch": 98.8575, "grad_norm": 6.90625, "learning_rate": 6.845247335979244e-09, "loss": 1.3517, "step": 39543 }, { "epoch": 98.86, "grad_norm": 8.5, "learning_rate": 6.8153261928383115e-09, "loss": 1.2445, "step": 39544 }, { "epoch": 98.8625, "grad_norm": 9.6875, "learning_rate": 6.785470564616426e-09, "loss": 1.3994, "step": 39545 }, { "epoch": 98.865, "grad_norm": 8.125, "learning_rate": 6.7556804515089835e-09, "loss": 1.4339, "step": 39546 }, { "epoch": 98.8675, "grad_norm": 8.0625, "learning_rate": 6.725955853712496e-09, "loss": 1.2619, "step": 39547 }, { "epoch": 98.87, "grad_norm": 10.5, "learning_rate": 6.696296771419031e-09, "loss": 1.3751, "step": 39548 }, { "epoch": 98.8725, "grad_norm": 6.34375, "learning_rate": 6.666703204825098e-09, "loss": 1.4428, "step": 39549 }, { "epoch": 98.875, "grad_norm": 7.8125, "learning_rate": 6.6371751541249865e-09, "loss": 1.6474, "step": 39550 }, { "epoch": 98.8775, "grad_norm": 8.125, "learning_rate": 6.607712619511874e-09, "loss": 1.3077, "step": 39551 }, { "epoch": 98.88, "grad_norm": 10.375, "learning_rate": 6.5783156011778315e-09, "loss": 1.2806, "step": 39552 }, { "epoch": 98.8825, "grad_norm": 8.75, "learning_rate": 6.548984099317146e-09, "loss": 1.3789, "step": 39553 }, { "epoch": 98.885, "grad_norm": 8.6875, "learning_rate": 6.519718114120776e-09, "loss": 1.2396, "step": 39554 }, { "epoch": 98.8875, "grad_norm": 9.125, "learning_rate": 6.490517645781902e-09, "loss": 1.3163, "step": 39555 }, { "epoch": 98.89, "grad_norm": 8.4375, "learning_rate": 6.46138269449037e-09, "loss": 1.1832, "step": 39556 }, { "epoch": 98.8925, "grad_norm": 7.0, "learning_rate": 6.43231326043825e-09, "loss": 1.3298, "step": 39557 }, { "epoch": 98.895, "grad_norm": 8.375, "learning_rate": 6.4033093438164995e-09, "loss": 1.377, "step": 39558 }, { "epoch": 98.8975, "grad_norm": 7.84375, "learning_rate": 6.374370944814967e-09, "loss": 1.3505, "step": 39559 }, { "epoch": 98.9, "grad_norm": 7.125, "learning_rate": 6.345498063622391e-09, "loss": 1.561, "step": 39560 }, { "epoch": 98.9025, "grad_norm": 8.25, "learning_rate": 6.316690700428618e-09, "loss": 1.3772, "step": 39561 }, { "epoch": 98.905, "grad_norm": 5.9375, "learning_rate": 6.287948855423498e-09, "loss": 1.2869, "step": 39562 }, { "epoch": 98.9075, "grad_norm": 7.03125, "learning_rate": 6.2592725287946576e-09, "loss": 1.2698, "step": 39563 }, { "epoch": 98.91, "grad_norm": 6.125, "learning_rate": 6.2306617207297245e-09, "loss": 1.3409, "step": 39564 }, { "epoch": 98.9125, "grad_norm": 9.0625, "learning_rate": 6.202116431416328e-09, "loss": 1.4038, "step": 39565 }, { "epoch": 98.915, "grad_norm": 4.9375, "learning_rate": 6.173636661042093e-09, "loss": 1.4339, "step": 39566 }, { "epoch": 98.9175, "grad_norm": 8.75, "learning_rate": 6.145222409794649e-09, "loss": 1.2709, "step": 39567 }, { "epoch": 98.92, "grad_norm": 7.125, "learning_rate": 6.116873677858293e-09, "loss": 1.3014, "step": 39568 }, { "epoch": 98.9225, "grad_norm": 4.84375, "learning_rate": 6.088590465419542e-09, "loss": 1.2147, "step": 39569 }, { "epoch": 98.925, "grad_norm": 9.3125, "learning_rate": 6.060372772663803e-09, "loss": 1.3223, "step": 39570 }, { "epoch": 98.9275, "grad_norm": 7.0, "learning_rate": 6.032220599777594e-09, "loss": 1.3662, "step": 39571 }, { "epoch": 98.93, "grad_norm": 7.46875, "learning_rate": 6.004133946942992e-09, "loss": 1.33, "step": 39572 }, { "epoch": 98.9325, "grad_norm": 9.1875, "learning_rate": 5.976112814346513e-09, "loss": 1.327, "step": 39573 }, { "epoch": 98.935, "grad_norm": 9.375, "learning_rate": 5.948157202169125e-09, "loss": 1.3827, "step": 39574 }, { "epoch": 98.9375, "grad_norm": 9.0, "learning_rate": 5.920267110597344e-09, "loss": 1.3595, "step": 39575 }, { "epoch": 98.94, "grad_norm": 6.625, "learning_rate": 5.892442539811028e-09, "loss": 1.3317, "step": 39576 }, { "epoch": 98.9425, "grad_norm": 5.4375, "learning_rate": 5.864683489994471e-09, "loss": 1.2294, "step": 39577 }, { "epoch": 98.945, "grad_norm": 8.125, "learning_rate": 5.836989961329753e-09, "loss": 1.4194, "step": 39578 }, { "epoch": 98.9475, "grad_norm": 8.375, "learning_rate": 5.809361953996728e-09, "loss": 1.4431, "step": 39579 }, { "epoch": 98.95, "grad_norm": 7.6875, "learning_rate": 5.781799468177473e-09, "loss": 1.4186, "step": 39580 }, { "epoch": 98.9525, "grad_norm": 7.28125, "learning_rate": 5.754302504054065e-09, "loss": 1.3828, "step": 39581 }, { "epoch": 98.955, "grad_norm": 5.21875, "learning_rate": 5.72687106180414e-09, "loss": 1.1956, "step": 39582 }, { "epoch": 98.9575, "grad_norm": 8.875, "learning_rate": 5.699505141609774e-09, "loss": 1.3551, "step": 39583 }, { "epoch": 98.96, "grad_norm": 6.875, "learning_rate": 5.6722047436497115e-09, "loss": 1.2362, "step": 39584 }, { "epoch": 98.9625, "grad_norm": 7.03125, "learning_rate": 5.644969868101591e-09, "loss": 1.295, "step": 39585 }, { "epoch": 98.965, "grad_norm": 8.375, "learning_rate": 5.617800515146376e-09, "loss": 1.3862, "step": 39586 }, { "epoch": 98.9675, "grad_norm": 7.46875, "learning_rate": 5.5906966849594844e-09, "loss": 1.3316, "step": 39587 }, { "epoch": 98.97, "grad_norm": 6.78125, "learning_rate": 5.563658377720771e-09, "loss": 1.3336, "step": 39588 }, { "epoch": 98.9725, "grad_norm": 8.3125, "learning_rate": 5.536685593606761e-09, "loss": 1.2347, "step": 39589 }, { "epoch": 98.975, "grad_norm": 12.75, "learning_rate": 5.509778332795091e-09, "loss": 1.4349, "step": 39590 }, { "epoch": 98.9775, "grad_norm": 9.3125, "learning_rate": 5.4829365954600645e-09, "loss": 1.4018, "step": 39591 }, { "epoch": 98.98, "grad_norm": 6.75, "learning_rate": 5.456160381779319e-09, "loss": 1.2799, "step": 39592 }, { "epoch": 98.9825, "grad_norm": 5.875, "learning_rate": 5.429449691928268e-09, "loss": 1.4404, "step": 39593 }, { "epoch": 98.985, "grad_norm": 10.875, "learning_rate": 5.402804526082328e-09, "loss": 1.4973, "step": 39594 }, { "epoch": 98.9875, "grad_norm": 6.5625, "learning_rate": 5.376224884414693e-09, "loss": 1.2752, "step": 39595 }, { "epoch": 98.99, "grad_norm": 20.75, "learning_rate": 5.34971076710078e-09, "loss": 1.0972, "step": 39596 }, { "epoch": 98.9925, "grad_norm": 7.84375, "learning_rate": 5.323262174314891e-09, "loss": 1.351, "step": 39597 }, { "epoch": 98.995, "grad_norm": 8.75, "learning_rate": 5.296879106229114e-09, "loss": 1.2564, "step": 39598 }, { "epoch": 98.9975, "grad_norm": 8.5625, "learning_rate": 5.27056156301664e-09, "loss": 1.2026, "step": 39599 }, { "epoch": 99.0, "grad_norm": 9.0625, "learning_rate": 5.2443095448506674e-09, "loss": 1.2361, "step": 39600 }, { "epoch": 99.0025, "grad_norm": 6.9375, "learning_rate": 5.218123051903279e-09, "loss": 1.4324, "step": 39601 }, { "epoch": 99.005, "grad_norm": 4.75, "learning_rate": 5.192002084346559e-09, "loss": 1.266, "step": 39602 }, { "epoch": 99.0075, "grad_norm": 11.875, "learning_rate": 5.165946642350373e-09, "loss": 1.404, "step": 39603 }, { "epoch": 99.01, "grad_norm": 4.71875, "learning_rate": 5.1399567260856935e-09, "loss": 1.3605, "step": 39604 }, { "epoch": 99.0125, "grad_norm": 6.625, "learning_rate": 5.114032335724605e-09, "loss": 1.3908, "step": 39605 }, { "epoch": 99.015, "grad_norm": 6.6875, "learning_rate": 5.088173471434754e-09, "loss": 1.3222, "step": 39606 }, { "epoch": 99.0175, "grad_norm": 6.53125, "learning_rate": 5.062380133388223e-09, "loss": 1.2892, "step": 39607 }, { "epoch": 99.02, "grad_norm": 8.6875, "learning_rate": 5.036652321751545e-09, "loss": 1.2252, "step": 39608 }, { "epoch": 99.0225, "grad_norm": 6.21875, "learning_rate": 5.010990036695695e-09, "loss": 1.2983, "step": 39609 }, { "epoch": 99.025, "grad_norm": 9.75, "learning_rate": 4.985393278386097e-09, "loss": 1.3943, "step": 39610 }, { "epoch": 99.0275, "grad_norm": 7.25, "learning_rate": 4.959862046993724e-09, "loss": 1.1841, "step": 39611 }, { "epoch": 99.03, "grad_norm": 8.0625, "learning_rate": 4.9343963426840006e-09, "loss": 1.4335, "step": 39612 }, { "epoch": 99.0325, "grad_norm": 6.875, "learning_rate": 4.908996165624569e-09, "loss": 1.2698, "step": 39613 }, { "epoch": 99.035, "grad_norm": 8.0, "learning_rate": 4.883661515981963e-09, "loss": 1.3225, "step": 39614 }, { "epoch": 99.0375, "grad_norm": 5.59375, "learning_rate": 4.858392393922717e-09, "loss": 1.4582, "step": 39615 }, { "epoch": 99.04, "grad_norm": 6.65625, "learning_rate": 4.833188799610033e-09, "loss": 1.313, "step": 39616 }, { "epoch": 99.0425, "grad_norm": 6.59375, "learning_rate": 4.808050733212666e-09, "loss": 1.2535, "step": 39617 }, { "epoch": 99.045, "grad_norm": 6.5625, "learning_rate": 4.782978194892707e-09, "loss": 1.4103, "step": 39618 }, { "epoch": 99.0475, "grad_norm": 5.46875, "learning_rate": 4.75797118481669e-09, "loss": 1.3323, "step": 39619 }, { "epoch": 99.05, "grad_norm": 7.25, "learning_rate": 4.733029703146708e-09, "loss": 1.3191, "step": 39620 }, { "epoch": 99.0525, "grad_norm": 10.5, "learning_rate": 4.708153750047073e-09, "loss": 1.2109, "step": 39621 }, { "epoch": 99.055, "grad_norm": 8.25, "learning_rate": 4.683343325679879e-09, "loss": 1.1878, "step": 39622 }, { "epoch": 99.0575, "grad_norm": 7.65625, "learning_rate": 4.6585984302094374e-09, "loss": 1.518, "step": 39623 }, { "epoch": 99.06, "grad_norm": 8.75, "learning_rate": 4.633919063797843e-09, "loss": 1.2739, "step": 39624 }, { "epoch": 99.0625, "grad_norm": 9.5, "learning_rate": 4.609305226606076e-09, "loss": 1.2594, "step": 39625 }, { "epoch": 99.065, "grad_norm": 6.625, "learning_rate": 4.584756918795119e-09, "loss": 1.204, "step": 39626 }, { "epoch": 99.0675, "grad_norm": 7.8125, "learning_rate": 4.560274140525955e-09, "loss": 1.4565, "step": 39627 }, { "epoch": 99.07, "grad_norm": 9.0, "learning_rate": 4.535856891960677e-09, "loss": 1.2878, "step": 39628 }, { "epoch": 99.0725, "grad_norm": 9.75, "learning_rate": 4.511505173258046e-09, "loss": 1.3904, "step": 39629 }, { "epoch": 99.075, "grad_norm": 8.5, "learning_rate": 4.487218984577935e-09, "loss": 1.2849, "step": 39630 }, { "epoch": 99.0775, "grad_norm": 6.21875, "learning_rate": 4.462998326080214e-09, "loss": 1.4688, "step": 39631 }, { "epoch": 99.08, "grad_norm": 7.59375, "learning_rate": 4.438843197922538e-09, "loss": 1.3478, "step": 39632 }, { "epoch": 99.0825, "grad_norm": 8.375, "learning_rate": 4.414753600263666e-09, "loss": 1.4494, "step": 39633 }, { "epoch": 99.085, "grad_norm": 6.0, "learning_rate": 4.390729533262361e-09, "loss": 1.2375, "step": 39634 }, { "epoch": 99.0875, "grad_norm": 7.0625, "learning_rate": 4.366770997075165e-09, "loss": 1.3607, "step": 39635 }, { "epoch": 99.09, "grad_norm": 7.84375, "learning_rate": 4.342877991858618e-09, "loss": 1.2189, "step": 39636 }, { "epoch": 99.0925, "grad_norm": 10.0625, "learning_rate": 4.319050517770373e-09, "loss": 1.3273, "step": 39637 }, { "epoch": 99.095, "grad_norm": 10.25, "learning_rate": 4.295288574966972e-09, "loss": 1.472, "step": 39638 }, { "epoch": 99.0975, "grad_norm": 9.5, "learning_rate": 4.271592163603844e-09, "loss": 1.2625, "step": 39639 }, { "epoch": 99.1, "grad_norm": 7.28125, "learning_rate": 4.247961283835311e-09, "loss": 1.2037, "step": 39640 }, { "epoch": 99.1025, "grad_norm": 11.375, "learning_rate": 4.2243959358179156e-09, "loss": 1.2047, "step": 39641 }, { "epoch": 99.105, "grad_norm": 8.4375, "learning_rate": 4.200896119704867e-09, "loss": 1.3583, "step": 39642 }, { "epoch": 99.1075, "grad_norm": 6.28125, "learning_rate": 4.177461835651598e-09, "loss": 1.2555, "step": 39643 }, { "epoch": 99.11, "grad_norm": 9.125, "learning_rate": 4.154093083809096e-09, "loss": 1.4167, "step": 39644 }, { "epoch": 99.1125, "grad_norm": 8.3125, "learning_rate": 4.130789864333906e-09, "loss": 1.3269, "step": 39645 }, { "epoch": 99.115, "grad_norm": 8.6875, "learning_rate": 4.107552177377016e-09, "loss": 1.2358, "step": 39646 }, { "epoch": 99.1175, "grad_norm": 10.9375, "learning_rate": 4.084380023090528e-09, "loss": 1.3651, "step": 39647 }, { "epoch": 99.12, "grad_norm": 6.90625, "learning_rate": 4.061273401627653e-09, "loss": 1.5253, "step": 39648 }, { "epoch": 99.1225, "grad_norm": 4.96875, "learning_rate": 4.0382323131382684e-09, "loss": 1.4456, "step": 39649 }, { "epoch": 99.125, "grad_norm": 7.71875, "learning_rate": 4.015256757774477e-09, "loss": 1.3339, "step": 39650 }, { "epoch": 99.1275, "grad_norm": 8.1875, "learning_rate": 3.992346735686159e-09, "loss": 1.3166, "step": 39651 }, { "epoch": 99.13, "grad_norm": 8.25, "learning_rate": 3.969502247024304e-09, "loss": 1.2912, "step": 39652 }, { "epoch": 99.1325, "grad_norm": 8.5625, "learning_rate": 3.9467232919387924e-09, "loss": 1.4048, "step": 39653 }, { "epoch": 99.135, "grad_norm": 7.40625, "learning_rate": 3.924009870578393e-09, "loss": 1.1679, "step": 39654 }, { "epoch": 99.1375, "grad_norm": 7.96875, "learning_rate": 3.901361983091878e-09, "loss": 1.265, "step": 39655 }, { "epoch": 99.14, "grad_norm": 8.875, "learning_rate": 3.878779629628016e-09, "loss": 1.3903, "step": 39656 }, { "epoch": 99.1425, "grad_norm": 8.375, "learning_rate": 3.8562628103355756e-09, "loss": 1.3664, "step": 39657 }, { "epoch": 99.145, "grad_norm": 7.875, "learning_rate": 3.833811525359999e-09, "loss": 1.3149, "step": 39658 }, { "epoch": 99.1475, "grad_norm": 9.1875, "learning_rate": 3.811425774851163e-09, "loss": 1.2694, "step": 39659 }, { "epoch": 99.15, "grad_norm": 8.3125, "learning_rate": 3.789105558954509e-09, "loss": 1.3978, "step": 39660 }, { "epoch": 99.1525, "grad_norm": 7.25, "learning_rate": 3.7668508778165855e-09, "loss": 1.3155, "step": 39661 }, { "epoch": 99.155, "grad_norm": 8.125, "learning_rate": 3.744661731581722e-09, "loss": 1.3709, "step": 39662 }, { "epoch": 99.1575, "grad_norm": 10.6875, "learning_rate": 3.722538120398689e-09, "loss": 1.3385, "step": 39663 }, { "epoch": 99.16, "grad_norm": 6.875, "learning_rate": 3.7004800444095933e-09, "loss": 1.317, "step": 39664 }, { "epoch": 99.1625, "grad_norm": 9.8125, "learning_rate": 3.6784875037598757e-09, "loss": 1.2669, "step": 39665 }, { "epoch": 99.165, "grad_norm": 5.03125, "learning_rate": 3.6565604985949744e-09, "loss": 1.2504, "step": 39666 }, { "epoch": 99.1675, "grad_norm": 5.0625, "learning_rate": 3.6346990290569983e-09, "loss": 1.1586, "step": 39667 }, { "epoch": 99.17, "grad_norm": 9.0625, "learning_rate": 3.6129030952902767e-09, "loss": 1.2901, "step": 39668 }, { "epoch": 99.1725, "grad_norm": 7.375, "learning_rate": 3.5911726974369177e-09, "loss": 1.3613, "step": 39669 }, { "epoch": 99.175, "grad_norm": 7.59375, "learning_rate": 3.56950783564014e-09, "loss": 1.171, "step": 39670 }, { "epoch": 99.1775, "grad_norm": 8.4375, "learning_rate": 3.5479085100409427e-09, "loss": 1.3859, "step": 39671 }, { "epoch": 99.18, "grad_norm": 10.3125, "learning_rate": 3.526374720782544e-09, "loss": 1.3958, "step": 39672 }, { "epoch": 99.1825, "grad_norm": 8.8125, "learning_rate": 3.5049064680048315e-09, "loss": 1.3202, "step": 39673 }, { "epoch": 99.185, "grad_norm": 8.0, "learning_rate": 3.4835037518488047e-09, "loss": 1.3364, "step": 39674 }, { "epoch": 99.1875, "grad_norm": 9.375, "learning_rate": 3.462166572454351e-09, "loss": 1.3226, "step": 39675 }, { "epoch": 99.19, "grad_norm": 8.6875, "learning_rate": 3.4408949299624683e-09, "loss": 1.3544, "step": 39676 }, { "epoch": 99.1925, "grad_norm": 8.0, "learning_rate": 3.4196888245130456e-09, "loss": 1.2268, "step": 39677 }, { "epoch": 99.195, "grad_norm": 6.75, "learning_rate": 3.3985482562426397e-09, "loss": 1.4462, "step": 39678 }, { "epoch": 99.1975, "grad_norm": 8.3125, "learning_rate": 3.3774732252922492e-09, "loss": 1.475, "step": 39679 }, { "epoch": 99.2, "grad_norm": 9.375, "learning_rate": 3.3564637317984318e-09, "loss": 1.2987, "step": 39680 }, { "epoch": 99.2025, "grad_norm": 9.375, "learning_rate": 3.3355197758999647e-09, "loss": 1.1652, "step": 39681 }, { "epoch": 99.205, "grad_norm": 9.0, "learning_rate": 3.314641357733406e-09, "loss": 1.3087, "step": 39682 }, { "epoch": 99.2075, "grad_norm": 11.4375, "learning_rate": 3.2938284774364228e-09, "loss": 1.3491, "step": 39683 }, { "epoch": 99.21, "grad_norm": 7.46875, "learning_rate": 3.2730811351444624e-09, "loss": 1.3951, "step": 39684 }, { "epoch": 99.2125, "grad_norm": 8.875, "learning_rate": 3.2523993309940827e-09, "loss": 1.2888, "step": 39685 }, { "epoch": 99.215, "grad_norm": 7.75, "learning_rate": 3.231783065121841e-09, "loss": 1.2587, "step": 39686 }, { "epoch": 99.2175, "grad_norm": 7.96875, "learning_rate": 3.2112323376620736e-09, "loss": 1.3363, "step": 39687 }, { "epoch": 99.22, "grad_norm": 6.15625, "learning_rate": 3.190747148748008e-09, "loss": 1.2567, "step": 39688 }, { "epoch": 99.2225, "grad_norm": 8.5, "learning_rate": 3.1703274985162015e-09, "loss": 1.3029, "step": 39689 }, { "epoch": 99.225, "grad_norm": 9.4375, "learning_rate": 3.1499733870998804e-09, "loss": 1.3024, "step": 39690 }, { "epoch": 99.2275, "grad_norm": 7.6875, "learning_rate": 3.1296848146322722e-09, "loss": 1.3238, "step": 39691 }, { "epoch": 99.23, "grad_norm": 7.3125, "learning_rate": 3.109461781246603e-09, "loss": 1.3957, "step": 39692 }, { "epoch": 99.2325, "grad_norm": 9.3125, "learning_rate": 3.08930428707388e-09, "loss": 1.3207, "step": 39693 }, { "epoch": 99.235, "grad_norm": 6.15625, "learning_rate": 3.0692123322495493e-09, "loss": 1.3107, "step": 39694 }, { "epoch": 99.2375, "grad_norm": 5.3125, "learning_rate": 3.0491859169023975e-09, "loss": 1.3398, "step": 39695 }, { "epoch": 99.24, "grad_norm": 5.8125, "learning_rate": 3.0292250411645406e-09, "loss": 1.2012, "step": 39696 }, { "epoch": 99.2425, "grad_norm": 6.125, "learning_rate": 3.0093297051669857e-09, "loss": 1.2498, "step": 39697 }, { "epoch": 99.245, "grad_norm": 9.25, "learning_rate": 2.9894999090407384e-09, "loss": 1.2782, "step": 39698 }, { "epoch": 99.2475, "grad_norm": 9.3125, "learning_rate": 2.969735652914585e-09, "loss": 1.3601, "step": 39699 }, { "epoch": 99.25, "grad_norm": 6.09375, "learning_rate": 2.9500369369195313e-09, "loss": 1.3084, "step": 39700 }, { "epoch": 99.2525, "grad_norm": 7.21875, "learning_rate": 2.9304037611843637e-09, "loss": 1.4072, "step": 39701 }, { "epoch": 99.255, "grad_norm": 7.5, "learning_rate": 2.910836125836758e-09, "loss": 1.221, "step": 39702 }, { "epoch": 99.2575, "grad_norm": 4.9375, "learning_rate": 2.8913340310054994e-09, "loss": 1.1606, "step": 39703 }, { "epoch": 99.26, "grad_norm": 7.09375, "learning_rate": 2.8718974768182638e-09, "loss": 1.2286, "step": 39704 }, { "epoch": 99.2625, "grad_norm": 8.625, "learning_rate": 2.852526463402727e-09, "loss": 1.1512, "step": 39705 }, { "epoch": 99.265, "grad_norm": 6.71875, "learning_rate": 2.833220990886565e-09, "loss": 1.2588, "step": 39706 }, { "epoch": 99.2675, "grad_norm": 6.78125, "learning_rate": 2.813981059396342e-09, "loss": 1.3069, "step": 39707 }, { "epoch": 99.27, "grad_norm": 7.34375, "learning_rate": 2.7948066690564045e-09, "loss": 1.1846, "step": 39708 }, { "epoch": 99.2725, "grad_norm": 5.5625, "learning_rate": 2.7756978199944273e-09, "loss": 1.2939, "step": 39709 }, { "epoch": 99.275, "grad_norm": 5.78125, "learning_rate": 2.7566545123347553e-09, "loss": 1.125, "step": 39710 }, { "epoch": 99.2775, "grad_norm": 5.5, "learning_rate": 2.737676746201734e-09, "loss": 1.3555, "step": 39711 }, { "epoch": 99.28, "grad_norm": 5.875, "learning_rate": 2.7187645217219283e-09, "loss": 1.3926, "step": 39712 }, { "epoch": 99.2825, "grad_norm": 7.4375, "learning_rate": 2.6999178390163526e-09, "loss": 1.3392, "step": 39713 }, { "epoch": 99.285, "grad_norm": 4.25, "learning_rate": 2.6811366982104625e-09, "loss": 1.3129, "step": 39714 }, { "epoch": 99.2875, "grad_norm": 7.21875, "learning_rate": 2.6624210994274925e-09, "loss": 1.4506, "step": 39715 }, { "epoch": 99.29, "grad_norm": 8.5, "learning_rate": 2.643771042789567e-09, "loss": 1.4562, "step": 39716 }, { "epoch": 99.2925, "grad_norm": 10.1875, "learning_rate": 2.6251865284177004e-09, "loss": 1.3724, "step": 39717 }, { "epoch": 99.295, "grad_norm": 7.46875, "learning_rate": 2.6066675564362374e-09, "loss": 1.3014, "step": 39718 }, { "epoch": 99.2975, "grad_norm": 8.75, "learning_rate": 2.5882141269650827e-09, "loss": 1.3847, "step": 39719 }, { "epoch": 99.3, "grad_norm": 6.90625, "learning_rate": 2.5698262401263607e-09, "loss": 1.3258, "step": 39720 }, { "epoch": 99.3025, "grad_norm": 8.0625, "learning_rate": 2.551503896038865e-09, "loss": 1.3687, "step": 39721 }, { "epoch": 99.305, "grad_norm": 7.34375, "learning_rate": 2.53324709482361e-09, "loss": 1.185, "step": 39722 }, { "epoch": 99.3075, "grad_norm": 15.375, "learning_rate": 2.5150558366004997e-09, "loss": 1.4265, "step": 39723 }, { "epoch": 99.31, "grad_norm": 5.5, "learning_rate": 2.496930121488328e-09, "loss": 1.2783, "step": 39724 }, { "epoch": 99.3125, "grad_norm": 9.625, "learning_rate": 2.478869949606999e-09, "loss": 1.3376, "step": 39725 }, { "epoch": 99.315, "grad_norm": 8.0625, "learning_rate": 2.4608753210730862e-09, "loss": 1.3142, "step": 39726 }, { "epoch": 99.3175, "grad_norm": 7.53125, "learning_rate": 2.4429462360064935e-09, "loss": 1.3799, "step": 39727 }, { "epoch": 99.32, "grad_norm": 8.625, "learning_rate": 2.4250826945226847e-09, "loss": 1.2725, "step": 39728 }, { "epoch": 99.3225, "grad_norm": 8.625, "learning_rate": 2.407284696741563e-09, "loss": 1.397, "step": 39729 }, { "epoch": 99.325, "grad_norm": 6.75, "learning_rate": 2.389552242777482e-09, "loss": 1.3904, "step": 39730 }, { "epoch": 99.3275, "grad_norm": 8.0625, "learning_rate": 2.371885332747015e-09, "loss": 1.3847, "step": 39731 }, { "epoch": 99.33, "grad_norm": 6.8125, "learning_rate": 2.354283966766735e-09, "loss": 1.4404, "step": 39732 }, { "epoch": 99.3325, "grad_norm": 8.75, "learning_rate": 2.3367481449509955e-09, "loss": 1.3809, "step": 39733 }, { "epoch": 99.335, "grad_norm": 6.96875, "learning_rate": 2.3192778674163697e-09, "loss": 1.2623, "step": 39734 }, { "epoch": 99.3375, "grad_norm": 9.6875, "learning_rate": 2.3018731342761004e-09, "loss": 1.29, "step": 39735 }, { "epoch": 99.34, "grad_norm": 10.9375, "learning_rate": 2.2845339456445402e-09, "loss": 1.23, "step": 39736 }, { "epoch": 99.3425, "grad_norm": 7.75, "learning_rate": 2.267260301636043e-09, "loss": 1.3025, "step": 39737 }, { "epoch": 99.345, "grad_norm": 6.0, "learning_rate": 2.25005220236274e-09, "loss": 1.2025, "step": 39738 }, { "epoch": 99.3475, "grad_norm": 7.625, "learning_rate": 2.2329096479389854e-09, "loss": 1.2754, "step": 39739 }, { "epoch": 99.35, "grad_norm": 7.6875, "learning_rate": 2.215832638474691e-09, "loss": 1.5437, "step": 39740 }, { "epoch": 99.3525, "grad_norm": 9.25, "learning_rate": 2.1988211740842093e-09, "loss": 1.3649, "step": 39741 }, { "epoch": 99.355, "grad_norm": 6.875, "learning_rate": 2.181875254878563e-09, "loss": 1.2304, "step": 39742 }, { "epoch": 99.3575, "grad_norm": 6.0, "learning_rate": 2.1649948809676634e-09, "loss": 1.3892, "step": 39743 }, { "epoch": 99.36, "grad_norm": 6.25, "learning_rate": 2.148180052462534e-09, "loss": 1.2947, "step": 39744 }, { "epoch": 99.3625, "grad_norm": 6.46875, "learning_rate": 2.1314307694753068e-09, "loss": 1.2055, "step": 39745 }, { "epoch": 99.365, "grad_norm": 9.0, "learning_rate": 2.1147470321136733e-09, "loss": 1.2698, "step": 39746 }, { "epoch": 99.3675, "grad_norm": 5.5625, "learning_rate": 2.0981288404875455e-09, "loss": 1.2851, "step": 39747 }, { "epoch": 99.37, "grad_norm": 9.9375, "learning_rate": 2.0815761947057254e-09, "loss": 1.4063, "step": 39748 }, { "epoch": 99.3725, "grad_norm": 13.0625, "learning_rate": 2.0650890948770154e-09, "loss": 1.4252, "step": 39749 }, { "epoch": 99.375, "grad_norm": 7.90625, "learning_rate": 2.0486675411102165e-09, "loss": 1.3606, "step": 39750 }, { "epoch": 99.3775, "grad_norm": 9.0625, "learning_rate": 2.0323115335108e-09, "loss": 1.0865, "step": 39751 }, { "epoch": 99.38, "grad_norm": 8.75, "learning_rate": 2.0160210721886788e-09, "loss": 1.3268, "step": 39752 }, { "epoch": 99.3825, "grad_norm": 4.625, "learning_rate": 1.999796157249323e-09, "loss": 1.2895, "step": 39753 }, { "epoch": 99.385, "grad_norm": 7.6875, "learning_rate": 1.9836367887993146e-09, "loss": 1.3015, "step": 39754 }, { "epoch": 99.3875, "grad_norm": 6.6875, "learning_rate": 1.967542966944125e-09, "loss": 1.2478, "step": 39755 }, { "epoch": 99.39, "grad_norm": 6.96875, "learning_rate": 1.9515146917892248e-09, "loss": 1.3079, "step": 39756 }, { "epoch": 99.3925, "grad_norm": 9.0625, "learning_rate": 1.9355519634411955e-09, "loss": 1.1116, "step": 39757 }, { "epoch": 99.395, "grad_norm": 7.53125, "learning_rate": 1.919654782002178e-09, "loss": 1.2396, "step": 39758 }, { "epoch": 99.3975, "grad_norm": 7.65625, "learning_rate": 1.9038231475787538e-09, "loss": 1.3163, "step": 39759 }, { "epoch": 99.4, "grad_norm": 7.375, "learning_rate": 1.888057060274173e-09, "loss": 1.2611, "step": 39760 }, { "epoch": 99.4025, "grad_norm": 9.8125, "learning_rate": 1.8723565201905768e-09, "loss": 1.5645, "step": 39761 }, { "epoch": 99.405, "grad_norm": 6.78125, "learning_rate": 1.8567215274323259e-09, "loss": 1.246, "step": 39762 }, { "epoch": 99.4075, "grad_norm": 6.53125, "learning_rate": 1.8411520821004502e-09, "loss": 1.2086, "step": 39763 }, { "epoch": 99.41, "grad_norm": 10.25, "learning_rate": 1.825648184298201e-09, "loss": 1.3311, "step": 39764 }, { "epoch": 99.4125, "grad_norm": 9.4375, "learning_rate": 1.8102098341277185e-09, "loss": 1.1747, "step": 39765 }, { "epoch": 99.415, "grad_norm": 6.5625, "learning_rate": 1.7948370316889229e-09, "loss": 1.3541, "step": 39766 }, { "epoch": 99.4175, "grad_norm": 6.34375, "learning_rate": 1.7795297770828446e-09, "loss": 1.3547, "step": 39767 }, { "epoch": 99.42, "grad_norm": 7.0625, "learning_rate": 1.7642880704094033e-09, "loss": 1.2223, "step": 39768 }, { "epoch": 99.4225, "grad_norm": 7.9375, "learning_rate": 1.7491119117696299e-09, "loss": 1.2802, "step": 39769 }, { "epoch": 99.425, "grad_norm": 8.875, "learning_rate": 1.7340013012634437e-09, "loss": 1.4237, "step": 39770 }, { "epoch": 99.4275, "grad_norm": 6.625, "learning_rate": 1.718956238988545e-09, "loss": 1.3644, "step": 39771 }, { "epoch": 99.43, "grad_norm": 8.25, "learning_rate": 1.7039767250437433e-09, "loss": 1.2695, "step": 39772 }, { "epoch": 99.4325, "grad_norm": 9.375, "learning_rate": 1.6890627595278487e-09, "loss": 1.3424, "step": 39773 }, { "epoch": 99.435, "grad_norm": 9.5, "learning_rate": 1.6742143425385605e-09, "loss": 1.5135, "step": 39774 }, { "epoch": 99.4375, "grad_norm": 9.3125, "learning_rate": 1.6594314741724682e-09, "loss": 1.3404, "step": 39775 }, { "epoch": 99.44, "grad_norm": 9.8125, "learning_rate": 1.6447141545272717e-09, "loss": 1.3132, "step": 39776 }, { "epoch": 99.4425, "grad_norm": 6.875, "learning_rate": 1.6300623836984497e-09, "loss": 1.4213, "step": 39777 }, { "epoch": 99.445, "grad_norm": 7.125, "learning_rate": 1.6154761617837022e-09, "loss": 1.2468, "step": 39778 }, { "epoch": 99.4475, "grad_norm": 8.625, "learning_rate": 1.6009554888785083e-09, "loss": 1.329, "step": 39779 }, { "epoch": 99.45, "grad_norm": 11.0, "learning_rate": 1.5865003650761268e-09, "loss": 1.3749, "step": 39780 }, { "epoch": 99.4525, "grad_norm": 6.46875, "learning_rate": 1.572110790473147e-09, "loss": 1.3882, "step": 39781 }, { "epoch": 99.455, "grad_norm": 11.4375, "learning_rate": 1.5577867651628276e-09, "loss": 1.3231, "step": 39782 }, { "epoch": 99.4575, "grad_norm": 6.03125, "learning_rate": 1.5435282892395375e-09, "loss": 1.4191, "step": 39783 }, { "epoch": 99.46, "grad_norm": 7.375, "learning_rate": 1.5293353627965358e-09, "loss": 1.621, "step": 39784 }, { "epoch": 99.4625, "grad_norm": 13.3125, "learning_rate": 1.515207985928191e-09, "loss": 1.3162, "step": 39785 }, { "epoch": 99.465, "grad_norm": 5.40625, "learning_rate": 1.5011461587255415e-09, "loss": 1.128, "step": 39786 }, { "epoch": 99.4675, "grad_norm": 7.6875, "learning_rate": 1.4871498812807362e-09, "loss": 1.2618, "step": 39787 }, { "epoch": 99.47, "grad_norm": 11.4375, "learning_rate": 1.4732191536870332e-09, "loss": 1.4365, "step": 39788 }, { "epoch": 99.4725, "grad_norm": 13.8125, "learning_rate": 1.459353976033251e-09, "loss": 1.4229, "step": 39789 }, { "epoch": 99.475, "grad_norm": 7.875, "learning_rate": 1.445554348413758e-09, "loss": 1.2812, "step": 39790 }, { "epoch": 99.4775, "grad_norm": 6.46875, "learning_rate": 1.4318202709151518e-09, "loss": 1.2109, "step": 39791 }, { "epoch": 99.48, "grad_norm": 5.53125, "learning_rate": 1.4181517436306913e-09, "loss": 1.2244, "step": 39792 }, { "epoch": 99.4825, "grad_norm": 7.71875, "learning_rate": 1.404548766648084e-09, "loss": 1.2774, "step": 39793 }, { "epoch": 99.485, "grad_norm": 8.125, "learning_rate": 1.3910113400572578e-09, "loss": 1.3401, "step": 39794 }, { "epoch": 99.4875, "grad_norm": 5.4375, "learning_rate": 1.3775394639470307e-09, "loss": 1.2915, "step": 39795 }, { "epoch": 99.49, "grad_norm": 8.1875, "learning_rate": 1.3641331384062207e-09, "loss": 1.3603, "step": 39796 }, { "epoch": 99.4925, "grad_norm": 9.0, "learning_rate": 1.350792363521425e-09, "loss": 1.3233, "step": 39797 }, { "epoch": 99.495, "grad_norm": 6.625, "learning_rate": 1.337517139380351e-09, "loss": 1.1207, "step": 39798 }, { "epoch": 99.4975, "grad_norm": 10.375, "learning_rate": 1.3243074660707066e-09, "loss": 1.3692, "step": 39799 }, { "epoch": 99.5, "grad_norm": 9.9375, "learning_rate": 1.3111633436779792e-09, "loss": 1.4746, "step": 39800 }, { "epoch": 99.5025, "grad_norm": 7.21875, "learning_rate": 1.298084772289876e-09, "loss": 1.2881, "step": 39801 }, { "epoch": 99.505, "grad_norm": 24.125, "learning_rate": 1.2850717519918843e-09, "loss": 1.5319, "step": 39802 }, { "epoch": 99.5075, "grad_norm": 7.53125, "learning_rate": 1.2721242828683811e-09, "loss": 1.4011, "step": 39803 }, { "epoch": 99.51, "grad_norm": 12.0625, "learning_rate": 1.2592423650048535e-09, "loss": 1.2986, "step": 39804 }, { "epoch": 99.5125, "grad_norm": 8.9375, "learning_rate": 1.2464259984856786e-09, "loss": 1.367, "step": 39805 }, { "epoch": 99.515, "grad_norm": 6.75, "learning_rate": 1.233675183394123e-09, "loss": 1.4086, "step": 39806 }, { "epoch": 99.5175, "grad_norm": 9.0625, "learning_rate": 1.2209899198145637e-09, "loss": 1.2407, "step": 39807 }, { "epoch": 99.52, "grad_norm": 6.71875, "learning_rate": 1.2083702078302673e-09, "loss": 1.1581, "step": 39808 }, { "epoch": 99.5225, "grad_norm": 7.5, "learning_rate": 1.1958160475245006e-09, "loss": 1.3209, "step": 39809 }, { "epoch": 99.525, "grad_norm": 11.5, "learning_rate": 1.18332743897831e-09, "loss": 1.3061, "step": 39810 }, { "epoch": 99.5275, "grad_norm": 7.3125, "learning_rate": 1.170904382273852e-09, "loss": 1.3156, "step": 39811 }, { "epoch": 99.53, "grad_norm": 8.3125, "learning_rate": 1.158546877493283e-09, "loss": 1.4649, "step": 39812 }, { "epoch": 99.5325, "grad_norm": 5.875, "learning_rate": 1.1462549247165388e-09, "loss": 1.3057, "step": 39813 }, { "epoch": 99.535, "grad_norm": 8.75, "learning_rate": 1.1340285240246662e-09, "loss": 1.3063, "step": 39814 }, { "epoch": 99.5375, "grad_norm": 9.0, "learning_rate": 1.1218676754987112e-09, "loss": 1.2841, "step": 39815 }, { "epoch": 99.54, "grad_norm": 7.78125, "learning_rate": 1.1097723792174998e-09, "loss": 1.2965, "step": 39816 }, { "epoch": 99.5425, "grad_norm": 6.15625, "learning_rate": 1.0977426352598574e-09, "loss": 1.2616, "step": 39817 }, { "epoch": 99.545, "grad_norm": 5.03125, "learning_rate": 1.0857784437057206e-09, "loss": 1.3241, "step": 39818 }, { "epoch": 99.5475, "grad_norm": 8.0, "learning_rate": 1.073879804633915e-09, "loss": 1.2382, "step": 39819 }, { "epoch": 99.55, "grad_norm": 7.125, "learning_rate": 1.062046718121046e-09, "loss": 1.3341, "step": 39820 }, { "epoch": 99.5525, "grad_norm": 9.125, "learning_rate": 1.050279184244829e-09, "loss": 1.1838, "step": 39821 }, { "epoch": 99.555, "grad_norm": 7.21875, "learning_rate": 1.03857720308298e-09, "loss": 1.2709, "step": 39822 }, { "epoch": 99.5575, "grad_norm": 8.5625, "learning_rate": 1.0269407747121041e-09, "loss": 1.5829, "step": 39823 }, { "epoch": 99.56, "grad_norm": 9.375, "learning_rate": 1.0153698992088069e-09, "loss": 1.5852, "step": 39824 }, { "epoch": 99.5625, "grad_norm": 9.5625, "learning_rate": 1.003864576647473e-09, "loss": 1.4098, "step": 39825 }, { "epoch": 99.565, "grad_norm": 6.59375, "learning_rate": 9.924248071058184e-10, "loss": 1.1387, "step": 39826 }, { "epoch": 99.5675, "grad_norm": 7.34375, "learning_rate": 9.81050590658228e-10, "loss": 1.2697, "step": 39827 }, { "epoch": 99.57, "grad_norm": 6.875, "learning_rate": 9.69741927376866e-10, "loss": 1.2151, "step": 39828 }, { "epoch": 99.5725, "grad_norm": 7.6875, "learning_rate": 9.584988173394483e-10, "loss": 1.3665, "step": 39829 }, { "epoch": 99.575, "grad_norm": 6.78125, "learning_rate": 9.47321260617029e-10, "loss": 1.241, "step": 39830 }, { "epoch": 99.5775, "grad_norm": 7.0625, "learning_rate": 9.362092572839932e-10, "loss": 1.5208, "step": 39831 }, { "epoch": 99.58, "grad_norm": 6.90625, "learning_rate": 9.251628074136154e-10, "loss": 1.3912, "step": 39832 }, { "epoch": 99.5825, "grad_norm": 7.375, "learning_rate": 9.1418191107695e-10, "loss": 1.2194, "step": 39833 }, { "epoch": 99.585, "grad_norm": 7.78125, "learning_rate": 9.032665683472719e-10, "loss": 1.4113, "step": 39834 }, { "epoch": 99.5875, "grad_norm": 5.78125, "learning_rate": 8.924167792956351e-10, "loss": 1.3278, "step": 39835 }, { "epoch": 99.59, "grad_norm": 7.40625, "learning_rate": 8.81632543993094e-10, "loss": 1.2985, "step": 39836 }, { "epoch": 99.5925, "grad_norm": 7.71875, "learning_rate": 8.709138625095925e-10, "loss": 1.3242, "step": 39837 }, { "epoch": 99.595, "grad_norm": 14.9375, "learning_rate": 8.602607349172953e-10, "loss": 1.2655, "step": 39838 }, { "epoch": 99.5975, "grad_norm": 11.5625, "learning_rate": 8.496731612850362e-10, "loss": 1.3253, "step": 39839 }, { "epoch": 99.6, "grad_norm": 8.1875, "learning_rate": 8.391511416816489e-10, "loss": 1.3889, "step": 39840 }, { "epoch": 99.6025, "grad_norm": 6.46875, "learning_rate": 8.286946761770776e-10, "loss": 1.3277, "step": 39841 }, { "epoch": 99.605, "grad_norm": 40.75, "learning_rate": 8.183037648390457e-10, "loss": 1.4193, "step": 39842 }, { "epoch": 99.6075, "grad_norm": 6.84375, "learning_rate": 8.079784077352771e-10, "loss": 1.3277, "step": 39843 }, { "epoch": 99.61, "grad_norm": 9.3125, "learning_rate": 7.977186049357156e-10, "loss": 1.4602, "step": 39844 }, { "epoch": 99.6125, "grad_norm": 7.4375, "learning_rate": 7.875243565058644e-10, "loss": 1.3087, "step": 39845 }, { "epoch": 99.615, "grad_norm": 9.125, "learning_rate": 7.773956625123369e-10, "loss": 1.3486, "step": 39846 }, { "epoch": 99.6175, "grad_norm": 6.375, "learning_rate": 7.673325230228567e-10, "loss": 1.1274, "step": 39847 }, { "epoch": 99.62, "grad_norm": 6.5, "learning_rate": 7.573349381018169e-10, "loss": 1.3452, "step": 39848 }, { "epoch": 99.6225, "grad_norm": 6.75, "learning_rate": 7.474029078158307e-10, "loss": 1.2514, "step": 39849 }, { "epoch": 99.625, "grad_norm": 8.4375, "learning_rate": 7.375364322292911e-10, "loss": 1.504, "step": 39850 }, { "epoch": 99.6275, "grad_norm": 7.9375, "learning_rate": 7.277355114077012e-10, "loss": 1.3216, "step": 39851 }, { "epoch": 99.63, "grad_norm": 6.375, "learning_rate": 7.18000145415454e-10, "loss": 1.305, "step": 39852 }, { "epoch": 99.6325, "grad_norm": 6.84375, "learning_rate": 7.083303343147218e-10, "loss": 1.1985, "step": 39853 }, { "epoch": 99.635, "grad_norm": 7.90625, "learning_rate": 6.987260781710082e-10, "loss": 1.4817, "step": 39854 }, { "epoch": 99.6375, "grad_norm": 8.4375, "learning_rate": 6.891873770453749e-10, "loss": 1.2977, "step": 39855 }, { "epoch": 99.64, "grad_norm": 8.6875, "learning_rate": 6.797142310022154e-10, "loss": 1.4103, "step": 39856 }, { "epoch": 99.6425, "grad_norm": 6.03125, "learning_rate": 6.703066401025915e-10, "loss": 1.3078, "step": 39857 }, { "epoch": 99.645, "grad_norm": 7.375, "learning_rate": 6.609646044075657e-10, "loss": 1.3529, "step": 39858 }, { "epoch": 99.6475, "grad_norm": 9.75, "learning_rate": 6.516881239793105e-10, "loss": 1.4022, "step": 39859 }, { "epoch": 99.65, "grad_norm": 9.5, "learning_rate": 6.424771988788881e-10, "loss": 1.2133, "step": 39860 }, { "epoch": 99.6525, "grad_norm": 6.0, "learning_rate": 6.333318291662504e-10, "loss": 1.2881, "step": 39861 }, { "epoch": 99.655, "grad_norm": 6.28125, "learning_rate": 6.242520149013498e-10, "loss": 1.2462, "step": 39862 }, { "epoch": 99.6575, "grad_norm": 7.15625, "learning_rate": 6.152377561430278e-10, "loss": 1.2168, "step": 39863 }, { "epoch": 99.66, "grad_norm": 6.71875, "learning_rate": 6.062890529512366e-10, "loss": 1.447, "step": 39864 }, { "epoch": 99.6625, "grad_norm": 6.3125, "learning_rate": 5.97405905384818e-10, "loss": 1.4955, "step": 39865 }, { "epoch": 99.665, "grad_norm": 11.875, "learning_rate": 5.885883135015036e-10, "loss": 1.3462, "step": 39866 }, { "epoch": 99.6675, "grad_norm": 8.5, "learning_rate": 5.798362773590249e-10, "loss": 1.2552, "step": 39867 }, { "epoch": 99.67, "grad_norm": 7.15625, "learning_rate": 5.711497970162238e-10, "loss": 1.2938, "step": 39868 }, { "epoch": 99.6725, "grad_norm": 7.625, "learning_rate": 5.625288725275013e-10, "loss": 1.3569, "step": 39869 }, { "epoch": 99.675, "grad_norm": 10.4375, "learning_rate": 5.539735039505889e-10, "loss": 1.3832, "step": 39870 }, { "epoch": 99.6775, "grad_norm": 6.3125, "learning_rate": 5.454836913421079e-10, "loss": 1.422, "step": 39871 }, { "epoch": 99.68, "grad_norm": 7.34375, "learning_rate": 5.370594347575697e-10, "loss": 1.3893, "step": 39872 }, { "epoch": 99.6825, "grad_norm": 8.125, "learning_rate": 5.28700734251375e-10, "loss": 1.2326, "step": 39873 }, { "epoch": 99.685, "grad_norm": 8.1875, "learning_rate": 5.204075898790351e-10, "loss": 1.5774, "step": 39874 }, { "epoch": 99.6875, "grad_norm": 8.75, "learning_rate": 5.121800016949508e-10, "loss": 1.4546, "step": 39875 }, { "epoch": 99.69, "grad_norm": 8.25, "learning_rate": 5.040179697524128e-10, "loss": 1.2809, "step": 39876 }, { "epoch": 99.6925, "grad_norm": 6.75, "learning_rate": 4.959214941058221e-10, "loss": 1.1992, "step": 39877 }, { "epoch": 99.695, "grad_norm": 6.25, "learning_rate": 4.878905748084695e-10, "loss": 1.2099, "step": 39878 }, { "epoch": 99.6975, "grad_norm": 7.375, "learning_rate": 4.799252119114251e-10, "loss": 1.2377, "step": 39879 }, { "epoch": 99.7, "grad_norm": 8.4375, "learning_rate": 4.720254054679796e-10, "loss": 1.4069, "step": 39880 }, { "epoch": 99.7025, "grad_norm": 8.25, "learning_rate": 4.6419115552920333e-10, "loss": 1.3594, "step": 39881 }, { "epoch": 99.705, "grad_norm": 7.0, "learning_rate": 4.5642246214838705e-10, "loss": 1.1544, "step": 39882 }, { "epoch": 99.7075, "grad_norm": 6.8125, "learning_rate": 4.4871932537327023e-10, "loss": 1.316, "step": 39883 }, { "epoch": 99.71, "grad_norm": 7.375, "learning_rate": 4.4108174525825386e-10, "loss": 1.4444, "step": 39884 }, { "epoch": 99.7125, "grad_norm": 8.1875, "learning_rate": 4.335097218499673e-10, "loss": 1.2551, "step": 39885 }, { "epoch": 99.715, "grad_norm": 6.6875, "learning_rate": 4.260032551994808e-10, "loss": 1.1334, "step": 39886 }, { "epoch": 99.7175, "grad_norm": 8.25, "learning_rate": 4.185623453567544e-10, "loss": 1.4349, "step": 39887 }, { "epoch": 99.72, "grad_norm": 9.9375, "learning_rate": 4.1118699236841753e-10, "loss": 1.4052, "step": 39888 }, { "epoch": 99.7225, "grad_norm": 10.125, "learning_rate": 4.0387719628554033e-10, "loss": 1.3984, "step": 39889 }, { "epoch": 99.725, "grad_norm": 7.71875, "learning_rate": 3.9663295715364205e-10, "loss": 1.2117, "step": 39890 }, { "epoch": 99.7275, "grad_norm": 7.4375, "learning_rate": 3.8945427502268264e-10, "loss": 1.2099, "step": 39891 }, { "epoch": 99.73, "grad_norm": 6.46875, "learning_rate": 3.8234114993707105e-10, "loss": 1.2707, "step": 39892 }, { "epoch": 99.7325, "grad_norm": 7.84375, "learning_rate": 3.752935819445469e-10, "loss": 1.4196, "step": 39893 }, { "epoch": 99.735, "grad_norm": 9.3125, "learning_rate": 3.683115710928498e-10, "loss": 1.4557, "step": 39894 }, { "epoch": 99.7375, "grad_norm": 9.125, "learning_rate": 3.613951174252783e-10, "loss": 1.3216, "step": 39895 }, { "epoch": 99.74, "grad_norm": 7.34375, "learning_rate": 3.5454422098735176e-10, "loss": 1.4079, "step": 39896 }, { "epoch": 99.7425, "grad_norm": 7.34375, "learning_rate": 3.4775888182680964e-10, "loss": 1.4368, "step": 39897 }, { "epoch": 99.745, "grad_norm": 6.875, "learning_rate": 3.4103909998473016e-10, "loss": 1.1295, "step": 39898 }, { "epoch": 99.7475, "grad_norm": 8.8125, "learning_rate": 3.343848755077428e-10, "loss": 1.5636, "step": 39899 }, { "epoch": 99.75, "grad_norm": 5.3125, "learning_rate": 3.277962084369257e-10, "loss": 1.2562, "step": 39900 }, { "epoch": 99.7525, "grad_norm": 8.5625, "learning_rate": 3.212730988177981e-10, "loss": 1.158, "step": 39901 }, { "epoch": 99.755, "grad_norm": 5.8125, "learning_rate": 3.1481554669254844e-10, "loss": 1.4218, "step": 39902 }, { "epoch": 99.7575, "grad_norm": 6.21875, "learning_rate": 3.0842355210336515e-10, "loss": 1.2832, "step": 39903 }, { "epoch": 99.76, "grad_norm": 9.0, "learning_rate": 3.0209711509132657e-10, "loss": 1.3948, "step": 39904 }, { "epoch": 99.7625, "grad_norm": 12.0625, "learning_rate": 2.958362356986211e-10, "loss": 1.3214, "step": 39905 }, { "epoch": 99.765, "grad_norm": 6.3125, "learning_rate": 2.8964091396632697e-10, "loss": 1.271, "step": 39906 }, { "epoch": 99.7675, "grad_norm": 5.46875, "learning_rate": 2.835111499344123e-10, "loss": 1.2193, "step": 39907 }, { "epoch": 99.77, "grad_norm": 7.0625, "learning_rate": 2.7744694364506554e-10, "loss": 1.3123, "step": 39908 }, { "epoch": 99.7725, "grad_norm": 8.375, "learning_rate": 2.71448295134924e-10, "loss": 1.3491, "step": 39909 }, { "epoch": 99.775, "grad_norm": 9.0, "learning_rate": 2.655152044461762e-10, "loss": 1.2963, "step": 39910 }, { "epoch": 99.7775, "grad_norm": 8.375, "learning_rate": 2.5964767161656966e-10, "loss": 1.4469, "step": 39911 }, { "epoch": 99.78, "grad_norm": 7.9375, "learning_rate": 2.538456966838521e-10, "loss": 1.2262, "step": 39912 }, { "epoch": 99.7825, "grad_norm": 7.6875, "learning_rate": 2.4810927968688115e-10, "loss": 1.2983, "step": 39913 }, { "epoch": 99.785, "grad_norm": 5.28125, "learning_rate": 2.424384206634045e-10, "loss": 1.3394, "step": 39914 }, { "epoch": 99.7875, "grad_norm": 6.21875, "learning_rate": 2.368331196500595e-10, "loss": 1.2218, "step": 39915 }, { "epoch": 99.79, "grad_norm": 6.53125, "learning_rate": 2.3129337668459372e-10, "loss": 1.3359, "step": 39916 }, { "epoch": 99.7925, "grad_norm": 6.4375, "learning_rate": 2.258191918025343e-10, "loss": 1.2507, "step": 39917 }, { "epoch": 99.795, "grad_norm": 7.125, "learning_rate": 2.2041056503940838e-10, "loss": 1.259, "step": 39918 }, { "epoch": 99.7975, "grad_norm": 5.03125, "learning_rate": 2.150674964307431e-10, "loss": 1.3445, "step": 39919 }, { "epoch": 99.8, "grad_norm": 8.1875, "learning_rate": 2.0978998601206558e-10, "loss": 1.4476, "step": 39920 }, { "epoch": 99.8025, "grad_norm": 6.5625, "learning_rate": 2.0457803381779273e-10, "loss": 1.361, "step": 39921 }, { "epoch": 99.805, "grad_norm": 9.0, "learning_rate": 1.9943163988345172e-10, "loss": 1.1661, "step": 39922 }, { "epoch": 99.8075, "grad_norm": 5.625, "learning_rate": 1.9435080424012875e-10, "loss": 1.3215, "step": 39923 }, { "epoch": 99.81, "grad_norm": 9.5625, "learning_rate": 1.8933552692224078e-10, "loss": 1.4048, "step": 39924 }, { "epoch": 99.8125, "grad_norm": 6.5, "learning_rate": 1.843858079642047e-10, "loss": 1.2763, "step": 39925 }, { "epoch": 99.815, "grad_norm": 7.71875, "learning_rate": 1.795016473959965e-10, "loss": 1.2691, "step": 39926 }, { "epoch": 99.8175, "grad_norm": 5.875, "learning_rate": 1.7468304525092294e-10, "loss": 1.1848, "step": 39927 }, { "epoch": 99.82, "grad_norm": 8.625, "learning_rate": 1.6993000156118044e-10, "loss": 1.3548, "step": 39928 }, { "epoch": 99.8225, "grad_norm": 7.71875, "learning_rate": 1.6524251635674505e-10, "loss": 1.2533, "step": 39929 }, { "epoch": 99.825, "grad_norm": 8.875, "learning_rate": 1.6062058966870297e-10, "loss": 1.2612, "step": 39930 }, { "epoch": 99.8275, "grad_norm": 5.8125, "learning_rate": 1.5606422152703027e-10, "loss": 1.3675, "step": 39931 }, { "epoch": 99.83, "grad_norm": 6.78125, "learning_rate": 1.5157341196281317e-10, "loss": 1.2289, "step": 39932 }, { "epoch": 99.8325, "grad_norm": 8.6875, "learning_rate": 1.4714816100491745e-10, "loss": 1.3548, "step": 39933 }, { "epoch": 99.835, "grad_norm": 6.71875, "learning_rate": 1.4278846868109876e-10, "loss": 1.4246, "step": 39934 }, { "epoch": 99.8375, "grad_norm": 8.125, "learning_rate": 1.3849433502133302e-10, "loss": 1.3495, "step": 39935 }, { "epoch": 99.84, "grad_norm": 8.625, "learning_rate": 1.342657600544861e-10, "loss": 1.3127, "step": 39936 }, { "epoch": 99.8425, "grad_norm": 8.625, "learning_rate": 1.3010274380609312e-10, "loss": 1.3983, "step": 39937 }, { "epoch": 99.845, "grad_norm": 6.3125, "learning_rate": 1.2600528630501984e-10, "loss": 1.4262, "step": 39938 }, { "epoch": 99.8475, "grad_norm": 7.3125, "learning_rate": 1.2197338757791166e-10, "loss": 1.3859, "step": 39939 }, { "epoch": 99.85, "grad_norm": 5.1875, "learning_rate": 1.1800704765030367e-10, "loss": 1.2905, "step": 39940 }, { "epoch": 99.8525, "grad_norm": 10.8125, "learning_rate": 1.1410626654884127e-10, "loss": 1.2996, "step": 39941 }, { "epoch": 99.855, "grad_norm": 6.4375, "learning_rate": 1.1027104429905954e-10, "loss": 1.376, "step": 39942 }, { "epoch": 99.8575, "grad_norm": 9.5, "learning_rate": 1.0650138092760387e-10, "loss": 1.4038, "step": 39943 }, { "epoch": 99.86, "grad_norm": 9.4375, "learning_rate": 1.027972764566787e-10, "loss": 1.4319, "step": 39944 }, { "epoch": 99.8625, "grad_norm": 6.6875, "learning_rate": 9.915873091070894e-11, "loss": 1.4082, "step": 39945 }, { "epoch": 99.865, "grad_norm": 6.28125, "learning_rate": 9.558574431633994e-11, "loss": 1.2839, "step": 39946 }, { "epoch": 99.8675, "grad_norm": 7.1875, "learning_rate": 9.207831669355571e-11, "loss": 1.2404, "step": 39947 }, { "epoch": 99.87, "grad_norm": 6.6875, "learning_rate": 8.86364480678914e-11, "loss": 1.3086, "step": 39948 }, { "epoch": 99.8725, "grad_norm": 6.125, "learning_rate": 8.526013846044123e-11, "loss": 1.3111, "step": 39949 }, { "epoch": 99.875, "grad_norm": 7.5625, "learning_rate": 8.19493878945199e-11, "loss": 1.2928, "step": 39950 }, { "epoch": 99.8775, "grad_norm": 6.53125, "learning_rate": 7.870419639122163e-11, "loss": 1.2424, "step": 39951 }, { "epoch": 99.88, "grad_norm": 8.375, "learning_rate": 7.552456397053042e-11, "loss": 1.3555, "step": 39952 }, { "epoch": 99.8825, "grad_norm": 9.0625, "learning_rate": 7.2410490655761e-11, "loss": 1.3282, "step": 39953 }, { "epoch": 99.885, "grad_norm": 10.0625, "learning_rate": 6.936197646578713e-11, "loss": 1.3494, "step": 39954 }, { "epoch": 99.8875, "grad_norm": 7.84375, "learning_rate": 6.637902142059282e-11, "loss": 1.3334, "step": 39955 }, { "epoch": 99.89, "grad_norm": 8.3125, "learning_rate": 6.34616255401621e-11, "loss": 1.3838, "step": 39956 }, { "epoch": 99.8925, "grad_norm": 7.65625, "learning_rate": 6.060978884336876e-11, "loss": 1.3259, "step": 39957 }, { "epoch": 99.895, "grad_norm": 5.96875, "learning_rate": 5.782351134908659e-11, "loss": 1.4248, "step": 39958 }, { "epoch": 99.8975, "grad_norm": 9.5, "learning_rate": 5.5102793076189376e-11, "loss": 1.4096, "step": 39959 }, { "epoch": 99.9, "grad_norm": 6.9375, "learning_rate": 5.244763404133046e-11, "loss": 1.3318, "step": 39960 }, { "epoch": 99.9025, "grad_norm": 8.5625, "learning_rate": 4.9858034262273425e-11, "loss": 1.4567, "step": 39961 }, { "epoch": 99.905, "grad_norm": 8.625, "learning_rate": 4.73339937556716e-11, "loss": 1.3914, "step": 39962 }, { "epoch": 99.9075, "grad_norm": 9.75, "learning_rate": 4.487551253928857e-11, "loss": 1.4796, "step": 39963 }, { "epoch": 99.91, "grad_norm": 7.375, "learning_rate": 4.2482590628667444e-11, "loss": 1.3551, "step": 39964 }, { "epoch": 99.9125, "grad_norm": 8.5, "learning_rate": 4.0155228039351345e-11, "loss": 1.3322, "step": 39965 }, { "epoch": 99.915, "grad_norm": 6.59375, "learning_rate": 3.789342478577318e-11, "loss": 1.3555, "step": 39966 }, { "epoch": 99.9175, "grad_norm": 5.78125, "learning_rate": 3.569718088458629e-11, "loss": 1.4366, "step": 39967 }, { "epoch": 99.92, "grad_norm": 8.625, "learning_rate": 3.3566496349113355e-11, "loss": 1.3237, "step": 39968 }, { "epoch": 99.9225, "grad_norm": 5.65625, "learning_rate": 3.1501371192677045e-11, "loss": 1.4223, "step": 39969 }, { "epoch": 99.925, "grad_norm": 11.0625, "learning_rate": 2.950180543082049e-11, "loss": 1.2732, "step": 39970 }, { "epoch": 99.9275, "grad_norm": 10.0, "learning_rate": 2.7567799073535685e-11, "loss": 1.4529, "step": 39971 }, { "epoch": 99.93, "grad_norm": 12.25, "learning_rate": 2.569935213636576e-11, "loss": 1.372, "step": 39972 }, { "epoch": 99.9325, "grad_norm": 7.375, "learning_rate": 2.3896464631523175e-11, "loss": 1.3632, "step": 39973 }, { "epoch": 99.935, "grad_norm": 10.875, "learning_rate": 2.2159136567889705e-11, "loss": 1.3151, "step": 39974 }, { "epoch": 99.9375, "grad_norm": 9.5, "learning_rate": 2.0487367959898253e-11, "loss": 1.208, "step": 39975 }, { "epoch": 99.94, "grad_norm": 6.8125, "learning_rate": 1.8881158817540824e-11, "loss": 1.4415, "step": 39976 }, { "epoch": 99.9425, "grad_norm": 8.75, "learning_rate": 1.7340509150809425e-11, "loss": 1.3406, "step": 39977 }, { "epoch": 99.945, "grad_norm": 6.84375, "learning_rate": 1.586541897080629e-11, "loss": 1.3986, "step": 39978 }, { "epoch": 99.9475, "grad_norm": 6.28125, "learning_rate": 1.44558882864132e-11, "loss": 1.2824, "step": 39979 }, { "epoch": 99.95, "grad_norm": 8.4375, "learning_rate": 1.311191710651194e-11, "loss": 1.3633, "step": 39980 }, { "epoch": 99.9525, "grad_norm": 11.375, "learning_rate": 1.1833505441094517e-11, "loss": 1.1338, "step": 39981 }, { "epoch": 99.955, "grad_norm": 9.375, "learning_rate": 1.062065329682227e-11, "loss": 1.4018, "step": 39982 }, { "epoch": 99.9575, "grad_norm": 9.5625, "learning_rate": 9.473360683687205e-12, "loss": 1.3332, "step": 39983 }, { "epoch": 99.96, "grad_norm": 9.6875, "learning_rate": 8.39162760835066e-12, "loss": 1.3838, "step": 39984 }, { "epoch": 99.9625, "grad_norm": 10.375, "learning_rate": 7.375454077473976e-12, "loss": 1.3525, "step": 39985 }, { "epoch": 99.965, "grad_norm": 7.84375, "learning_rate": 6.424840097718488e-12, "loss": 1.3461, "step": 39986 }, { "epoch": 99.9675, "grad_norm": 6.25, "learning_rate": 5.539785676855758e-12, "loss": 1.279, "step": 39987 }, { "epoch": 99.97, "grad_norm": 4.09375, "learning_rate": 4.720290818216455e-12, "loss": 1.3555, "step": 39988 }, { "epoch": 99.9725, "grad_norm": 7.40625, "learning_rate": 3.966355529572141e-12, "loss": 1.1809, "step": 39989 }, { "epoch": 99.975, "grad_norm": 9.375, "learning_rate": 3.2779798142534848e-12, "loss": 1.3423, "step": 39990 }, { "epoch": 99.9775, "grad_norm": 9.375, "learning_rate": 2.655163676701378e-12, "loss": 1.3454, "step": 39991 }, { "epoch": 99.98, "grad_norm": 8.9375, "learning_rate": 2.0979071224669357e-12, "loss": 1.343, "step": 39992 }, { "epoch": 99.9825, "grad_norm": 5.28125, "learning_rate": 1.6062101537706042e-12, "loss": 1.4798, "step": 39993 }, { "epoch": 99.985, "grad_norm": 8.0625, "learning_rate": 1.1800727739430528e-12, "loss": 1.2864, "step": 39994 }, { "epoch": 99.9875, "grad_norm": 8.9375, "learning_rate": 8.194949874251734e-13, "loss": 1.2962, "step": 39995 }, { "epoch": 99.99, "grad_norm": 7.71875, "learning_rate": 5.244767942169659e-13, "loss": 1.1616, "step": 39996 }, { "epoch": 99.9925, "grad_norm": 7.4375, "learning_rate": 2.9501819764909955e-13, "loss": 1.1847, "step": 39997 }, { "epoch": 99.995, "grad_norm": 7.1875, "learning_rate": 1.3111919994202028e-13, "loss": 1.5085, "step": 39998 }, { "epoch": 99.9975, "grad_norm": 9.6875, "learning_rate": 3.277979998550507e-14, "loss": 1.3943, "step": 39999 }, { "epoch": 100.0, "grad_norm": 6.375, "learning_rate": 0.0, "loss": 1.3369, "step": 40000 }, { "epoch": 100.0, "step": 40000, "total_flos": 9.340920528896e+16, "train_loss": 1.44269131193161, "train_runtime": 18705.1667, "train_samples_per_second": 68.43, "train_steps_per_second": 2.138 } ], "logging_steps": 1.0, "max_steps": 40000, "num_input_tokens_seen": 0, "num_train_epochs": 100, "save_steps": 25000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 9.340920528896e+16, "train_batch_size": 2, "trial_name": null, "trial_params": null }