diff --git "a/trainer_state.json" "b/trainer_state.json" deleted file mode 100644--- "a/trainer_state.json" +++ /dev/null @@ -1,22273 +0,0 @@ -{ - "best_metric": null, - "best_model_checkpoint": null, - "epoch": 3.0, - "global_step": 3708, - "is_hyper_param_search": false, - "is_local_process_zero": true, - "is_world_process_zero": true, - "log_history": [ - { - "epoch": 0.0, - "learning_rate": 1.7857142857142858e-07, - "loss": 1.9067, - "step": 1 - }, - { - "epoch": 0.0, - "learning_rate": 3.5714285714285716e-07, - "loss": 1.8722, - "step": 2 - }, - { - "epoch": 0.0, - "learning_rate": 5.357142857142857e-07, - "loss": 1.8473, - "step": 3 - }, - { - "epoch": 0.0, - "learning_rate": 7.142857142857143e-07, - "loss": 1.7228, - "step": 4 - }, - { - "epoch": 0.0, - "learning_rate": 8.928571428571429e-07, - "loss": 1.6887, - "step": 5 - }, - { - "epoch": 0.0, - "learning_rate": 1.0714285714285714e-06, - "loss": 1.8274, - "step": 6 - }, - { - "epoch": 0.01, - "learning_rate": 1.25e-06, - "loss": 1.8355, - "step": 7 - }, - { - "epoch": 0.01, - "learning_rate": 1.4285714285714286e-06, - "loss": 1.8505, - "step": 8 - }, - { - "epoch": 0.01, - "learning_rate": 1.6071428571428574e-06, - "loss": 1.8948, - "step": 9 - }, - { - "epoch": 0.01, - "learning_rate": 1.7857142857142859e-06, - "loss": 1.9172, - "step": 10 - }, - { - "epoch": 0.01, - "learning_rate": 1.9642857142857144e-06, - "loss": 1.8426, - "step": 11 - }, - { - "epoch": 0.01, - "learning_rate": 2.1428571428571427e-06, - "loss": 1.8447, - "step": 12 - }, - { - "epoch": 0.01, - "learning_rate": 2.321428571428572e-06, - "loss": 1.9126, - "step": 13 - }, - { - "epoch": 0.01, - "learning_rate": 2.5e-06, - "loss": 2.0015, - "step": 14 - }, - { - "epoch": 0.01, - "learning_rate": 2.6785714285714285e-06, - "loss": 1.7946, - "step": 15 - }, - { - "epoch": 0.01, - "learning_rate": 2.8571428571428573e-06, - "loss": 1.6832, - "step": 16 - }, - { - "epoch": 0.01, - "learning_rate": 3.0357142857142856e-06, - "loss": 1.7436, - "step": 17 - }, - { - "epoch": 0.01, - "learning_rate": 3.2142857142857147e-06, - "loss": 1.6714, - "step": 18 - }, - { - "epoch": 0.02, - "learning_rate": 3.3928571428571435e-06, - "loss": 1.8393, - "step": 19 - }, - { - "epoch": 0.02, - "learning_rate": 3.5714285714285718e-06, - "loss": 1.6286, - "step": 20 - }, - { - "epoch": 0.02, - "learning_rate": 3.7500000000000005e-06, - "loss": 1.653, - "step": 21 - }, - { - "epoch": 0.02, - "learning_rate": 3.928571428571429e-06, - "loss": 1.8719, - "step": 22 - }, - { - "epoch": 0.02, - "learning_rate": 4.107142857142857e-06, - "loss": 1.7195, - "step": 23 - }, - { - "epoch": 0.02, - "learning_rate": 4.2857142857142855e-06, - "loss": 1.6012, - "step": 24 - }, - { - "epoch": 0.02, - "learning_rate": 4.464285714285715e-06, - "loss": 1.8671, - "step": 25 - }, - { - "epoch": 0.02, - "learning_rate": 4.642857142857144e-06, - "loss": 1.7655, - "step": 26 - }, - { - "epoch": 0.02, - "learning_rate": 4.821428571428572e-06, - "loss": 1.4681, - "step": 27 - }, - { - "epoch": 0.02, - "learning_rate": 5e-06, - "loss": 1.775, - "step": 28 - }, - { - "epoch": 0.02, - "learning_rate": 5.1785714285714296e-06, - "loss": 1.7809, - "step": 29 - }, - { - "epoch": 0.02, - "learning_rate": 5.357142857142857e-06, - "loss": 1.6159, - "step": 30 - }, - { - "epoch": 0.03, - "learning_rate": 5.535714285714286e-06, - "loss": 1.6685, - "step": 31 - }, - { - "epoch": 0.03, - "learning_rate": 5.7142857142857145e-06, - "loss": 1.4895, - "step": 32 - }, - { - "epoch": 0.03, - "learning_rate": 5.892857142857144e-06, - "loss": 1.6808, - "step": 33 - }, - { - "epoch": 0.03, - "learning_rate": 6.071428571428571e-06, - "loss": 1.7442, - "step": 34 - }, - { - "epoch": 0.03, - "learning_rate": 6.25e-06, - "loss": 1.6427, - "step": 35 - }, - { - "epoch": 0.03, - "learning_rate": 6.4285714285714295e-06, - "loss": 1.748, - "step": 36 - }, - { - "epoch": 0.03, - "learning_rate": 6.607142857142858e-06, - "loss": 1.7085, - "step": 37 - }, - { - "epoch": 0.03, - "learning_rate": 6.785714285714287e-06, - "loss": 1.6381, - "step": 38 - }, - { - "epoch": 0.03, - "learning_rate": 6.964285714285714e-06, - "loss": 1.5926, - "step": 39 - }, - { - "epoch": 0.03, - "learning_rate": 7.1428571428571436e-06, - "loss": 1.7404, - "step": 40 - }, - { - "epoch": 0.03, - "learning_rate": 7.321428571428572e-06, - "loss": 1.6152, - "step": 41 - }, - { - "epoch": 0.03, - "learning_rate": 7.500000000000001e-06, - "loss": 1.6195, - "step": 42 - }, - { - "epoch": 0.03, - "learning_rate": 7.67857142857143e-06, - "loss": 1.6178, - "step": 43 - }, - { - "epoch": 0.04, - "learning_rate": 7.857142857142858e-06, - "loss": 1.6502, - "step": 44 - }, - { - "epoch": 0.04, - "learning_rate": 8.035714285714286e-06, - "loss": 1.5838, - "step": 45 - }, - { - "epoch": 0.04, - "learning_rate": 8.214285714285714e-06, - "loss": 1.692, - "step": 46 - }, - { - "epoch": 0.04, - "learning_rate": 8.392857142857144e-06, - "loss": 1.6212, - "step": 47 - }, - { - "epoch": 0.04, - "learning_rate": 8.571428571428571e-06, - "loss": 1.5756, - "step": 48 - }, - { - "epoch": 0.04, - "learning_rate": 8.750000000000001e-06, - "loss": 1.6996, - "step": 49 - }, - { - "epoch": 0.04, - "learning_rate": 8.92857142857143e-06, - "loss": 1.5539, - "step": 50 - }, - { - "epoch": 0.04, - "learning_rate": 9.107142857142858e-06, - "loss": 1.4984, - "step": 51 - }, - { - "epoch": 0.04, - "learning_rate": 9.285714285714288e-06, - "loss": 1.6488, - "step": 52 - }, - { - "epoch": 0.04, - "learning_rate": 9.464285714285714e-06, - "loss": 1.693, - "step": 53 - }, - { - "epoch": 0.04, - "learning_rate": 9.642857142857144e-06, - "loss": 1.7048, - "step": 54 - }, - { - "epoch": 0.04, - "learning_rate": 9.821428571428573e-06, - "loss": 1.6943, - "step": 55 - }, - { - "epoch": 0.05, - "learning_rate": 1e-05, - "loss": 1.7987, - "step": 56 - }, - { - "epoch": 0.05, - "learning_rate": 1.0178571428571429e-05, - "loss": 1.5656, - "step": 57 - }, - { - "epoch": 0.05, - "learning_rate": 1.0357142857142859e-05, - "loss": 1.7579, - "step": 58 - }, - { - "epoch": 0.05, - "learning_rate": 1.0535714285714287e-05, - "loss": 1.6027, - "step": 59 - }, - { - "epoch": 0.05, - "learning_rate": 1.0714285714285714e-05, - "loss": 1.4865, - "step": 60 - }, - { - "epoch": 0.05, - "learning_rate": 1.0892857142857142e-05, - "loss": 1.8245, - "step": 61 - }, - { - "epoch": 0.05, - "learning_rate": 1.1071428571428572e-05, - "loss": 1.6553, - "step": 62 - }, - { - "epoch": 0.05, - "learning_rate": 1.125e-05, - "loss": 1.6048, - "step": 63 - }, - { - "epoch": 0.05, - "learning_rate": 1.1428571428571429e-05, - "loss": 1.6447, - "step": 64 - }, - { - "epoch": 0.05, - "learning_rate": 1.1607142857142859e-05, - "loss": 1.6939, - "step": 65 - }, - { - "epoch": 0.05, - "learning_rate": 1.1785714285714287e-05, - "loss": 1.6523, - "step": 66 - }, - { - "epoch": 0.05, - "learning_rate": 1.1964285714285716e-05, - "loss": 1.5621, - "step": 67 - }, - { - "epoch": 0.06, - "learning_rate": 1.2142857142857142e-05, - "loss": 1.5823, - "step": 68 - }, - { - "epoch": 0.06, - "learning_rate": 1.2321428571428572e-05, - "loss": 1.6247, - "step": 69 - }, - { - "epoch": 0.06, - "learning_rate": 1.25e-05, - "loss": 1.6321, - "step": 70 - }, - { - "epoch": 0.06, - "learning_rate": 1.2678571428571429e-05, - "loss": 1.6937, - "step": 71 - }, - { - "epoch": 0.06, - "learning_rate": 1.2857142857142859e-05, - "loss": 1.7602, - "step": 72 - }, - { - "epoch": 0.06, - "learning_rate": 1.3035714285714287e-05, - "loss": 1.5329, - "step": 73 - }, - { - "epoch": 0.06, - "learning_rate": 1.3214285714285716e-05, - "loss": 1.6822, - "step": 74 - }, - { - "epoch": 0.06, - "learning_rate": 1.3392857142857142e-05, - "loss": 1.6911, - "step": 75 - }, - { - "epoch": 0.06, - "learning_rate": 1.3571428571428574e-05, - "loss": 1.6515, - "step": 76 - }, - { - "epoch": 0.06, - "learning_rate": 1.375e-05, - "loss": 1.6842, - "step": 77 - }, - { - "epoch": 0.06, - "learning_rate": 1.3928571428571429e-05, - "loss": 1.6601, - "step": 78 - }, - { - "epoch": 0.06, - "learning_rate": 1.4107142857142859e-05, - "loss": 1.609, - "step": 79 - }, - { - "epoch": 0.06, - "learning_rate": 1.4285714285714287e-05, - "loss": 1.8505, - "step": 80 - }, - { - "epoch": 0.07, - "learning_rate": 1.4464285714285715e-05, - "loss": 1.6627, - "step": 81 - }, - { - "epoch": 0.07, - "learning_rate": 1.4642857142857144e-05, - "loss": 1.4739, - "step": 82 - }, - { - "epoch": 0.07, - "learning_rate": 1.4821428571428574e-05, - "loss": 1.5807, - "step": 83 - }, - { - "epoch": 0.07, - "learning_rate": 1.5000000000000002e-05, - "loss": 1.616, - "step": 84 - }, - { - "epoch": 0.07, - "learning_rate": 1.5178571428571429e-05, - "loss": 1.7028, - "step": 85 - }, - { - "epoch": 0.07, - "learning_rate": 1.535714285714286e-05, - "loss": 1.5748, - "step": 86 - }, - { - "epoch": 0.07, - "learning_rate": 1.553571428571429e-05, - "loss": 1.5767, - "step": 87 - }, - { - "epoch": 0.07, - "learning_rate": 1.5714285714285715e-05, - "loss": 1.5114, - "step": 88 - }, - { - "epoch": 0.07, - "learning_rate": 1.5892857142857142e-05, - "loss": 1.6526, - "step": 89 - }, - { - "epoch": 0.07, - "learning_rate": 1.6071428571428572e-05, - "loss": 1.5282, - "step": 90 - }, - { - "epoch": 0.07, - "learning_rate": 1.6250000000000002e-05, - "loss": 1.6632, - "step": 91 - }, - { - "epoch": 0.07, - "learning_rate": 1.642857142857143e-05, - "loss": 1.6151, - "step": 92 - }, - { - "epoch": 0.08, - "learning_rate": 1.660714285714286e-05, - "loss": 1.6835, - "step": 93 - }, - { - "epoch": 0.08, - "learning_rate": 1.678571428571429e-05, - "loss": 1.5505, - "step": 94 - }, - { - "epoch": 0.08, - "learning_rate": 1.6964285714285715e-05, - "loss": 1.5607, - "step": 95 - }, - { - "epoch": 0.08, - "learning_rate": 1.7142857142857142e-05, - "loss": 1.4759, - "step": 96 - }, - { - "epoch": 0.08, - "learning_rate": 1.7321428571428572e-05, - "loss": 1.6432, - "step": 97 - }, - { - "epoch": 0.08, - "learning_rate": 1.7500000000000002e-05, - "loss": 1.6049, - "step": 98 - }, - { - "epoch": 0.08, - "learning_rate": 1.767857142857143e-05, - "loss": 1.6423, - "step": 99 - }, - { - "epoch": 0.08, - "learning_rate": 1.785714285714286e-05, - "loss": 1.5777, - "step": 100 - }, - { - "epoch": 0.08, - "learning_rate": 1.803571428571429e-05, - "loss": 1.5935, - "step": 101 - }, - { - "epoch": 0.08, - "learning_rate": 1.8214285714285715e-05, - "loss": 1.6558, - "step": 102 - }, - { - "epoch": 0.08, - "learning_rate": 1.8392857142857142e-05, - "loss": 1.7385, - "step": 103 - }, - { - "epoch": 0.08, - "learning_rate": 1.8571428571428575e-05, - "loss": 1.58, - "step": 104 - }, - { - "epoch": 0.08, - "learning_rate": 1.8750000000000002e-05, - "loss": 1.5839, - "step": 105 - }, - { - "epoch": 0.09, - "learning_rate": 1.892857142857143e-05, - "loss": 1.7108, - "step": 106 - }, - { - "epoch": 0.09, - "learning_rate": 1.910714285714286e-05, - "loss": 1.5698, - "step": 107 - }, - { - "epoch": 0.09, - "learning_rate": 1.928571428571429e-05, - "loss": 1.6364, - "step": 108 - }, - { - "epoch": 0.09, - "learning_rate": 1.9464285714285715e-05, - "loss": 1.7472, - "step": 109 - }, - { - "epoch": 0.09, - "learning_rate": 1.9642857142857145e-05, - "loss": 1.667, - "step": 110 - }, - { - "epoch": 0.09, - "learning_rate": 1.9821428571428575e-05, - "loss": 1.5784, - "step": 111 - }, - { - "epoch": 0.09, - "learning_rate": 2e-05, - "loss": 1.5402, - "step": 112 - }, - { - "epoch": 0.09, - "learning_rate": 1.9999996183806777e-05, - "loss": 1.6774, - "step": 113 - }, - { - "epoch": 0.09, - "learning_rate": 1.999998473523002e-05, - "loss": 1.7228, - "step": 114 - }, - { - "epoch": 0.09, - "learning_rate": 1.999996565427847e-05, - "loss": 1.7846, - "step": 115 - }, - { - "epoch": 0.09, - "learning_rate": 1.9999938940966684e-05, - "loss": 1.6966, - "step": 116 - }, - { - "epoch": 0.09, - "learning_rate": 1.9999904595315053e-05, - "loss": 1.5938, - "step": 117 - }, - { - "epoch": 0.1, - "learning_rate": 1.9999862617349797e-05, - "loss": 1.4902, - "step": 118 - }, - { - "epoch": 0.1, - "learning_rate": 1.9999813007102947e-05, - "loss": 1.7071, - "step": 119 - }, - { - "epoch": 0.1, - "learning_rate": 1.999975576461237e-05, - "loss": 1.5693, - "step": 120 - }, - { - "epoch": 0.1, - "learning_rate": 1.9999690889921763e-05, - "loss": 1.5868, - "step": 121 - }, - { - "epoch": 0.1, - "learning_rate": 1.9999618383080625e-05, - "loss": 1.6915, - "step": 122 - }, - { - "epoch": 0.1, - "learning_rate": 1.9999538244144312e-05, - "loss": 1.6496, - "step": 123 - }, - { - "epoch": 0.1, - "learning_rate": 1.999945047317398e-05, - "loss": 1.6834, - "step": 124 - }, - { - "epoch": 0.1, - "learning_rate": 1.999935507023662e-05, - "loss": 1.475, - "step": 125 - }, - { - "epoch": 0.1, - "learning_rate": 1.999925203540505e-05, - "loss": 1.7546, - "step": 126 - }, - { - "epoch": 0.1, - "learning_rate": 1.9999141368757913e-05, - "loss": 1.6926, - "step": 127 - }, - { - "epoch": 0.1, - "learning_rate": 1.999902307037967e-05, - "loss": 1.5127, - "step": 128 - }, - { - "epoch": 0.1, - "learning_rate": 1.9998897140360608e-05, - "loss": 1.6158, - "step": 129 - }, - { - "epoch": 0.11, - "learning_rate": 1.999876357879684e-05, - "loss": 1.6991, - "step": 130 - }, - { - "epoch": 0.11, - "learning_rate": 1.999862238579032e-05, - "loss": 1.6578, - "step": 131 - }, - { - "epoch": 0.11, - "learning_rate": 1.9998473561448797e-05, - "loss": 1.7583, - "step": 132 - }, - { - "epoch": 0.11, - "learning_rate": 1.9998317105885865e-05, - "loss": 1.5801, - "step": 133 - }, - { - "epoch": 0.11, - "learning_rate": 1.9998153019220935e-05, - "loss": 1.6904, - "step": 134 - }, - { - "epoch": 0.11, - "learning_rate": 1.999798130157925e-05, - "loss": 1.7465, - "step": 135 - }, - { - "epoch": 0.11, - "learning_rate": 1.9997801953091863e-05, - "loss": 1.5472, - "step": 136 - }, - { - "epoch": 0.11, - "learning_rate": 1.9997614973895665e-05, - "loss": 1.5386, - "step": 137 - }, - { - "epoch": 0.11, - "learning_rate": 1.9997420364133363e-05, - "loss": 1.619, - "step": 138 - }, - { - "epoch": 0.11, - "learning_rate": 1.9997218123953495e-05, - "loss": 1.6212, - "step": 139 - }, - { - "epoch": 0.11, - "learning_rate": 1.9997008253510416e-05, - "loss": 1.6644, - "step": 140 - }, - { - "epoch": 0.11, - "learning_rate": 1.9996790752964305e-05, - "loss": 1.6402, - "step": 141 - }, - { - "epoch": 0.11, - "learning_rate": 1.999656562248117e-05, - "loss": 1.6133, - "step": 142 - }, - { - "epoch": 0.12, - "learning_rate": 1.9996332862232843e-05, - "loss": 1.6191, - "step": 143 - }, - { - "epoch": 0.12, - "learning_rate": 1.9996092472396964e-05, - "loss": 1.722, - "step": 144 - }, - { - "epoch": 0.12, - "learning_rate": 1.999584445315702e-05, - "loss": 1.6977, - "step": 145 - }, - { - "epoch": 0.12, - "learning_rate": 1.99955888047023e-05, - "loss": 1.4735, - "step": 146 - }, - { - "epoch": 0.12, - "learning_rate": 1.999532552722793e-05, - "loss": 1.4238, - "step": 147 - }, - { - "epoch": 0.12, - "learning_rate": 1.999505462093485e-05, - "loss": 1.6099, - "step": 148 - }, - { - "epoch": 0.12, - "learning_rate": 1.999477608602983e-05, - "loss": 1.6256, - "step": 149 - }, - { - "epoch": 0.12, - "learning_rate": 1.9994489922725454e-05, - "loss": 1.5015, - "step": 150 - }, - { - "epoch": 0.12, - "learning_rate": 1.9994196131240137e-05, - "loss": 1.6188, - "step": 151 - }, - { - "epoch": 0.12, - "learning_rate": 1.9993894711798112e-05, - "loss": 1.6702, - "step": 152 - }, - { - "epoch": 0.12, - "learning_rate": 1.999358566462943e-05, - "loss": 1.6748, - "step": 153 - }, - { - "epoch": 0.12, - "learning_rate": 1.9993268989969974e-05, - "loss": 1.6675, - "step": 154 - }, - { - "epoch": 0.13, - "learning_rate": 1.9992944688061434e-05, - "loss": 1.7273, - "step": 155 - }, - { - "epoch": 0.13, - "learning_rate": 1.9992612759151336e-05, - "loss": 1.6801, - "step": 156 - }, - { - "epoch": 0.13, - "learning_rate": 1.9992273203493023e-05, - "loss": 1.5497, - "step": 157 - }, - { - "epoch": 0.13, - "learning_rate": 1.999192602134565e-05, - "loss": 1.3976, - "step": 158 - }, - { - "epoch": 0.13, - "learning_rate": 1.9991571212974204e-05, - "loss": 1.6548, - "step": 159 - }, - { - "epoch": 0.13, - "learning_rate": 1.9991208778649485e-05, - "loss": 1.6958, - "step": 160 - }, - { - "epoch": 0.13, - "learning_rate": 1.9990838718648122e-05, - "loss": 1.6118, - "step": 161 - }, - { - "epoch": 0.13, - "learning_rate": 1.9990461033252556e-05, - "loss": 1.7866, - "step": 162 - }, - { - "epoch": 0.13, - "learning_rate": 1.999007572275105e-05, - "loss": 1.6955, - "step": 163 - }, - { - "epoch": 0.13, - "learning_rate": 1.9989682787437695e-05, - "loss": 1.6124, - "step": 164 - }, - { - "epoch": 0.13, - "learning_rate": 1.9989282227612382e-05, - "loss": 1.6802, - "step": 165 - }, - { - "epoch": 0.13, - "learning_rate": 1.9988874043580848e-05, - "loss": 1.5961, - "step": 166 - }, - { - "epoch": 0.14, - "learning_rate": 1.998845823565462e-05, - "loss": 1.7018, - "step": 167 - }, - { - "epoch": 0.14, - "learning_rate": 1.998803480415107e-05, - "loss": 1.8051, - "step": 168 - }, - { - "epoch": 0.14, - "learning_rate": 1.9987603749393375e-05, - "loss": 1.6751, - "step": 169 - }, - { - "epoch": 0.14, - "learning_rate": 1.998716507171053e-05, - "loss": 1.6671, - "step": 170 - }, - { - "epoch": 0.14, - "learning_rate": 1.998671877143735e-05, - "loss": 1.5918, - "step": 171 - }, - { - "epoch": 0.14, - "learning_rate": 1.9986264848914475e-05, - "loss": 1.2644, - "step": 172 - }, - { - "epoch": 0.14, - "learning_rate": 1.9985803304488347e-05, - "loss": 1.6401, - "step": 173 - }, - { - "epoch": 0.14, - "learning_rate": 1.998533413851124e-05, - "loss": 1.6257, - "step": 174 - }, - { - "epoch": 0.14, - "learning_rate": 1.998485735134124e-05, - "loss": 1.659, - "step": 175 - }, - { - "epoch": 0.14, - "learning_rate": 1.9984372943342248e-05, - "loss": 1.6952, - "step": 176 - }, - { - "epoch": 0.14, - "learning_rate": 1.998388091488398e-05, - "loss": 1.4936, - "step": 177 - }, - { - "epoch": 0.14, - "learning_rate": 1.998338126634198e-05, - "loss": 1.445, - "step": 178 - }, - { - "epoch": 0.14, - "learning_rate": 1.998287399809759e-05, - "loss": 1.6422, - "step": 179 - }, - { - "epoch": 0.15, - "learning_rate": 1.998235911053798e-05, - "loss": 1.6674, - "step": 180 - }, - { - "epoch": 0.15, - "learning_rate": 1.9981836604056136e-05, - "loss": 1.7948, - "step": 181 - }, - { - "epoch": 0.15, - "learning_rate": 1.9981306479050848e-05, - "loss": 1.4816, - "step": 182 - }, - { - "epoch": 0.15, - "learning_rate": 1.9980768735926734e-05, - "loss": 1.6839, - "step": 183 - }, - { - "epoch": 0.15, - "learning_rate": 1.998022337509421e-05, - "loss": 1.6497, - "step": 184 - }, - { - "epoch": 0.15, - "learning_rate": 1.9979670396969532e-05, - "loss": 1.607, - "step": 185 - }, - { - "epoch": 0.15, - "learning_rate": 1.9979109801974743e-05, - "loss": 1.5922, - "step": 186 - }, - { - "epoch": 0.15, - "learning_rate": 1.9978541590537714e-05, - "loss": 1.7237, - "step": 187 - }, - { - "epoch": 0.15, - "learning_rate": 1.9977965763092126e-05, - "loss": 1.7199, - "step": 188 - }, - { - "epoch": 0.15, - "learning_rate": 1.9977382320077474e-05, - "loss": 1.6322, - "step": 189 - }, - { - "epoch": 0.15, - "learning_rate": 1.9976791261939064e-05, - "loss": 1.5814, - "step": 190 - }, - { - "epoch": 0.15, - "learning_rate": 1.997619258912801e-05, - "loss": 1.4485, - "step": 191 - }, - { - "epoch": 0.16, - "learning_rate": 1.9975586302101247e-05, - "loss": 1.5581, - "step": 192 - }, - { - "epoch": 0.16, - "learning_rate": 1.9974972401321518e-05, - "loss": 1.48, - "step": 193 - }, - { - "epoch": 0.16, - "learning_rate": 1.997435088725737e-05, - "loss": 1.5821, - "step": 194 - }, - { - "epoch": 0.16, - "learning_rate": 1.997372176038317e-05, - "loss": 1.785, - "step": 195 - }, - { - "epoch": 0.16, - "learning_rate": 1.9973085021179092e-05, - "loss": 1.605, - "step": 196 - }, - { - "epoch": 0.16, - "learning_rate": 1.997244067013112e-05, - "loss": 1.5186, - "step": 197 - }, - { - "epoch": 0.16, - "learning_rate": 1.997178870773105e-05, - "loss": 1.4923, - "step": 198 - }, - { - "epoch": 0.16, - "learning_rate": 1.9971129134476474e-05, - "loss": 1.4974, - "step": 199 - }, - { - "epoch": 0.16, - "learning_rate": 1.997046195087082e-05, - "loss": 1.5932, - "step": 200 - }, - { - "epoch": 0.16, - "learning_rate": 1.9969787157423293e-05, - "loss": 1.5272, - "step": 201 - }, - { - "epoch": 0.16, - "learning_rate": 1.9969104754648937e-05, - "loss": 1.5369, - "step": 202 - }, - { - "epoch": 0.16, - "learning_rate": 1.9968414743068572e-05, - "loss": 1.5134, - "step": 203 - }, - { - "epoch": 0.17, - "learning_rate": 1.9967717123208855e-05, - "loss": 1.578, - "step": 204 - }, - { - "epoch": 0.17, - "learning_rate": 1.996701189560223e-05, - "loss": 1.5785, - "step": 205 - }, - { - "epoch": 0.17, - "learning_rate": 1.9966299060786953e-05, - "loss": 1.4609, - "step": 206 - }, - { - "epoch": 0.17, - "learning_rate": 1.996557861930709e-05, - "loss": 1.5453, - "step": 207 - }, - { - "epoch": 0.17, - "learning_rate": 1.9964850571712507e-05, - "loss": 1.6202, - "step": 208 - }, - { - "epoch": 0.17, - "learning_rate": 1.996411491855888e-05, - "loss": 1.4516, - "step": 209 - }, - { - "epoch": 0.17, - "learning_rate": 1.996337166040769e-05, - "loss": 1.6497, - "step": 210 - }, - { - "epoch": 0.17, - "learning_rate": 1.9962620797826214e-05, - "loss": 1.6425, - "step": 211 - }, - { - "epoch": 0.17, - "learning_rate": 1.9961862331387545e-05, - "loss": 1.4863, - "step": 212 - }, - { - "epoch": 0.17, - "learning_rate": 1.996109626167057e-05, - "loss": 1.6865, - "step": 213 - }, - { - "epoch": 0.17, - "learning_rate": 1.996032258925999e-05, - "loss": 1.5926, - "step": 214 - }, - { - "epoch": 0.17, - "learning_rate": 1.9959541314746293e-05, - "loss": 1.5904, - "step": 215 - }, - { - "epoch": 0.17, - "learning_rate": 1.9958752438725782e-05, - "loss": 1.7331, - "step": 216 - }, - { - "epoch": 0.18, - "learning_rate": 1.995795596180056e-05, - "loss": 1.5389, - "step": 217 - }, - { - "epoch": 0.18, - "learning_rate": 1.995715188457852e-05, - "loss": 1.7142, - "step": 218 - }, - { - "epoch": 0.18, - "learning_rate": 1.995634020767338e-05, - "loss": 1.4996, - "step": 219 - }, - { - "epoch": 0.18, - "learning_rate": 1.995552093170463e-05, - "loss": 1.5868, - "step": 220 - }, - { - "epoch": 0.18, - "learning_rate": 1.995469405729758e-05, - "loss": 1.7172, - "step": 221 - }, - { - "epoch": 0.18, - "learning_rate": 1.9953859585083326e-05, - "loss": 1.5691, - "step": 222 - }, - { - "epoch": 0.18, - "learning_rate": 1.9953017515698775e-05, - "loss": 1.5533, - "step": 223 - }, - { - "epoch": 0.18, - "learning_rate": 1.9952167849786622e-05, - "loss": 1.6101, - "step": 224 - }, - { - "epoch": 0.18, - "learning_rate": 1.995131058799537e-05, - "loss": 1.6822, - "step": 225 - }, - { - "epoch": 0.18, - "learning_rate": 1.9950445730979314e-05, - "loss": 1.5921, - "step": 226 - }, - { - "epoch": 0.18, - "learning_rate": 1.994957327939855e-05, - "loss": 1.5084, - "step": 227 - }, - { - "epoch": 0.18, - "learning_rate": 1.994869323391895e-05, - "loss": 1.584, - "step": 228 - }, - { - "epoch": 0.19, - "learning_rate": 1.994780559521222e-05, - "loss": 1.6358, - "step": 229 - }, - { - "epoch": 0.19, - "learning_rate": 1.994691036395583e-05, - "loss": 1.6532, - "step": 230 - }, - { - "epoch": 0.19, - "learning_rate": 1.994600754083305e-05, - "loss": 1.7814, - "step": 231 - }, - { - "epoch": 0.19, - "learning_rate": 1.9945097126532955e-05, - "loss": 1.678, - "step": 232 - }, - { - "epoch": 0.19, - "learning_rate": 1.9944179121750412e-05, - "loss": 1.7164, - "step": 233 - }, - { - "epoch": 0.19, - "learning_rate": 1.994325352718607e-05, - "loss": 1.5054, - "step": 234 - }, - { - "epoch": 0.19, - "learning_rate": 1.9942320343546387e-05, - "loss": 1.7299, - "step": 235 - }, - { - "epoch": 0.19, - "learning_rate": 1.9941379571543597e-05, - "loss": 1.5628, - "step": 236 - }, - { - "epoch": 0.19, - "learning_rate": 1.994043121189574e-05, - "loss": 1.7332, - "step": 237 - }, - { - "epoch": 0.19, - "learning_rate": 1.9939475265326636e-05, - "loss": 1.4742, - "step": 238 - }, - { - "epoch": 0.19, - "learning_rate": 1.9938511732565903e-05, - "loss": 1.5157, - "step": 239 - }, - { - "epoch": 0.19, - "learning_rate": 1.9937540614348944e-05, - "loss": 1.6368, - "step": 240 - }, - { - "epoch": 0.19, - "learning_rate": 1.9936561911416956e-05, - "loss": 1.548, - "step": 241 - }, - { - "epoch": 0.2, - "learning_rate": 1.9935575624516922e-05, - "loss": 1.7297, - "step": 242 - }, - { - "epoch": 0.2, - "learning_rate": 1.9934581754401616e-05, - "loss": 1.6475, - "step": 243 - }, - { - "epoch": 0.2, - "learning_rate": 1.9933580301829596e-05, - "loss": 1.4693, - "step": 244 - }, - { - "epoch": 0.2, - "learning_rate": 1.993257126756521e-05, - "loss": 1.7086, - "step": 245 - }, - { - "epoch": 0.2, - "learning_rate": 1.9931554652378594e-05, - "loss": 1.4321, - "step": 246 - }, - { - "epoch": 0.2, - "learning_rate": 1.9930530457045663e-05, - "loss": 1.7032, - "step": 247 - }, - { - "epoch": 0.2, - "learning_rate": 1.9929498682348127e-05, - "loss": 1.6443, - "step": 248 - }, - { - "epoch": 0.2, - "learning_rate": 1.9928459329073474e-05, - "loss": 1.6528, - "step": 249 - }, - { - "epoch": 0.2, - "learning_rate": 1.992741239801498e-05, - "loss": 1.6617, - "step": 250 - }, - { - "epoch": 0.2, - "learning_rate": 1.9926357889971702e-05, - "loss": 1.7758, - "step": 251 - }, - { - "epoch": 0.2, - "learning_rate": 1.992529580574848e-05, - "loss": 1.6625, - "step": 252 - }, - { - "epoch": 0.2, - "learning_rate": 1.9924226146155944e-05, - "loss": 1.5701, - "step": 253 - }, - { - "epoch": 0.21, - "learning_rate": 1.992314891201049e-05, - "loss": 1.6148, - "step": 254 - }, - { - "epoch": 0.21, - "learning_rate": 1.9922064104134317e-05, - "loss": 1.5484, - "step": 255 - }, - { - "epoch": 0.21, - "learning_rate": 1.9920971723355378e-05, - "loss": 1.5157, - "step": 256 - }, - { - "epoch": 0.21, - "learning_rate": 1.991987177050743e-05, - "loss": 1.5569, - "step": 257 - }, - { - "epoch": 0.21, - "learning_rate": 1.991876424643e-05, - "loss": 1.4864, - "step": 258 - }, - { - "epoch": 0.21, - "learning_rate": 1.9917649151968385e-05, - "loss": 1.6305, - "step": 259 - }, - { - "epoch": 0.21, - "learning_rate": 1.9916526487973678e-05, - "loss": 1.5377, - "step": 260 - }, - { - "epoch": 0.21, - "learning_rate": 1.9915396255302732e-05, - "loss": 1.6798, - "step": 261 - }, - { - "epoch": 0.21, - "learning_rate": 1.9914258454818185e-05, - "loss": 1.5498, - "step": 262 - }, - { - "epoch": 0.21, - "learning_rate": 1.9913113087388457e-05, - "loss": 1.5671, - "step": 263 - }, - { - "epoch": 0.21, - "learning_rate": 1.9911960153887732e-05, - "loss": 1.6456, - "step": 264 - }, - { - "epoch": 0.21, - "learning_rate": 1.991079965519597e-05, - "loss": 1.543, - "step": 265 - }, - { - "epoch": 0.22, - "learning_rate": 1.990963159219891e-05, - "loss": 1.2485, - "step": 266 - }, - { - "epoch": 0.22, - "learning_rate": 1.990845596578807e-05, - "loss": 1.6158, - "step": 267 - }, - { - "epoch": 0.22, - "learning_rate": 1.9907272776860728e-05, - "loss": 1.6109, - "step": 268 - }, - { - "epoch": 0.22, - "learning_rate": 1.9906082026319936e-05, - "loss": 1.4691, - "step": 269 - }, - { - "epoch": 0.22, - "learning_rate": 1.9904883715074525e-05, - "loss": 1.6512, - "step": 270 - }, - { - "epoch": 0.22, - "learning_rate": 1.9903677844039097e-05, - "loss": 1.6145, - "step": 271 - }, - { - "epoch": 0.22, - "learning_rate": 1.9902464414134008e-05, - "loss": 1.5317, - "step": 272 - }, - { - "epoch": 0.22, - "learning_rate": 1.9901243426285403e-05, - "loss": 1.5934, - "step": 273 - }, - { - "epoch": 0.22, - "learning_rate": 1.9900014881425186e-05, - "loss": 1.7579, - "step": 274 - }, - { - "epoch": 0.22, - "learning_rate": 1.9898778780491028e-05, - "loss": 1.5637, - "step": 275 - }, - { - "epoch": 0.22, - "learning_rate": 1.989753512442637e-05, - "loss": 1.605, - "step": 276 - }, - { - "epoch": 0.22, - "learning_rate": 1.9896283914180422e-05, - "loss": 1.4395, - "step": 277 - }, - { - "epoch": 0.22, - "learning_rate": 1.989502515070815e-05, - "loss": 1.6155, - "step": 278 - }, - { - "epoch": 0.23, - "learning_rate": 1.9893758834970294e-05, - "loss": 1.5235, - "step": 279 - }, - { - "epoch": 0.23, - "learning_rate": 1.989248496793335e-05, - "loss": 1.6919, - "step": 280 - }, - { - "epoch": 0.23, - "learning_rate": 1.9891203550569595e-05, - "loss": 1.5518, - "step": 281 - }, - { - "epoch": 0.23, - "learning_rate": 1.988991458385704e-05, - "loss": 1.6092, - "step": 282 - }, - { - "epoch": 0.23, - "learning_rate": 1.9888618068779486e-05, - "loss": 1.6179, - "step": 283 - }, - { - "epoch": 0.23, - "learning_rate": 1.9887314006326482e-05, - "loss": 1.6147, - "step": 284 - }, - { - "epoch": 0.23, - "learning_rate": 1.9886002397493332e-05, - "loss": 1.5904, - "step": 285 - }, - { - "epoch": 0.23, - "learning_rate": 1.9884683243281117e-05, - "loss": 1.7253, - "step": 286 - }, - { - "epoch": 0.23, - "learning_rate": 1.9883356544696656e-05, - "loss": 1.626, - "step": 287 - }, - { - "epoch": 0.23, - "learning_rate": 1.9882022302752544e-05, - "loss": 1.6888, - "step": 288 - }, - { - "epoch": 0.23, - "learning_rate": 1.988068051846712e-05, - "loss": 1.5804, - "step": 289 - }, - { - "epoch": 0.23, - "learning_rate": 1.9879331192864492e-05, - "loss": 1.62, - "step": 290 - }, - { - "epoch": 0.24, - "learning_rate": 1.9877974326974515e-05, - "loss": 1.6635, - "step": 291 - }, - { - "epoch": 0.24, - "learning_rate": 1.98766099218328e-05, - "loss": 1.6511, - "step": 292 - }, - { - "epoch": 0.24, - "learning_rate": 1.987523797848071e-05, - "loss": 1.5598, - "step": 293 - }, - { - "epoch": 0.24, - "learning_rate": 1.987385849796538e-05, - "loss": 1.622, - "step": 294 - }, - { - "epoch": 0.24, - "learning_rate": 1.9872471481339666e-05, - "loss": 1.7325, - "step": 295 - }, - { - "epoch": 0.24, - "learning_rate": 1.9871076929662202e-05, - "loss": 1.8229, - "step": 296 - }, - { - "epoch": 0.24, - "learning_rate": 1.9869674843997357e-05, - "loss": 1.651, - "step": 297 - }, - { - "epoch": 0.24, - "learning_rate": 1.9868265225415263e-05, - "loss": 1.6443, - "step": 298 - }, - { - "epoch": 0.24, - "learning_rate": 1.9866848074991793e-05, - "loss": 1.5512, - "step": 299 - }, - { - "epoch": 0.24, - "learning_rate": 1.9865423393808573e-05, - "loss": 1.4969, - "step": 300 - }, - { - "epoch": 0.24, - "learning_rate": 1.9863991182952974e-05, - "loss": 1.7485, - "step": 301 - }, - { - "epoch": 0.24, - "learning_rate": 1.986255144351811e-05, - "loss": 1.7143, - "step": 302 - }, - { - "epoch": 0.25, - "learning_rate": 1.9861104176602852e-05, - "loss": 1.6244, - "step": 303 - }, - { - "epoch": 0.25, - "learning_rate": 1.9859649383311807e-05, - "loss": 1.459, - "step": 304 - }, - { - "epoch": 0.25, - "learning_rate": 1.985818706475533e-05, - "loss": 1.6803, - "step": 305 - }, - { - "epoch": 0.25, - "learning_rate": 1.985671722204952e-05, - "loss": 1.5307, - "step": 306 - }, - { - "epoch": 0.25, - "learning_rate": 1.9855239856316218e-05, - "loss": 1.4453, - "step": 307 - }, - { - "epoch": 0.25, - "learning_rate": 1.9853754968683003e-05, - "loss": 1.602, - "step": 308 - }, - { - "epoch": 0.25, - "learning_rate": 1.9852262560283203e-05, - "loss": 1.7416, - "step": 309 - }, - { - "epoch": 0.25, - "learning_rate": 1.985076263225588e-05, - "loss": 1.6605, - "step": 310 - }, - { - "epoch": 0.25, - "learning_rate": 1.9849255185745835e-05, - "loss": 1.5733, - "step": 311 - }, - { - "epoch": 0.25, - "learning_rate": 1.984774022190361e-05, - "loss": 1.6527, - "step": 312 - }, - { - "epoch": 0.25, - "learning_rate": 1.984621774188549e-05, - "loss": 1.7347, - "step": 313 - }, - { - "epoch": 0.25, - "learning_rate": 1.984468774685348e-05, - "loss": 1.6994, - "step": 314 - }, - { - "epoch": 0.25, - "learning_rate": 1.9843150237975343e-05, - "loss": 1.6765, - "step": 315 - }, - { - "epoch": 0.26, - "learning_rate": 1.984160521642456e-05, - "loss": 1.58, - "step": 316 - }, - { - "epoch": 0.26, - "learning_rate": 1.9840052683380343e-05, - "loss": 1.722, - "step": 317 - }, - { - "epoch": 0.26, - "learning_rate": 1.9838492640027658e-05, - "loss": 1.5887, - "step": 318 - }, - { - "epoch": 0.26, - "learning_rate": 1.9836925087557187e-05, - "loss": 1.6167, - "step": 319 - }, - { - "epoch": 0.26, - "learning_rate": 1.9835350027165342e-05, - "loss": 1.481, - "step": 320 - }, - { - "epoch": 0.26, - "learning_rate": 1.9833767460054273e-05, - "loss": 1.6998, - "step": 321 - }, - { - "epoch": 0.26, - "learning_rate": 1.9832177387431854e-05, - "loss": 1.5802, - "step": 322 - }, - { - "epoch": 0.26, - "learning_rate": 1.9830579810511694e-05, - "loss": 1.5881, - "step": 323 - }, - { - "epoch": 0.26, - "learning_rate": 1.9828974730513125e-05, - "loss": 1.566, - "step": 324 - }, - { - "epoch": 0.26, - "learning_rate": 1.9827362148661203e-05, - "loss": 1.5894, - "step": 325 - }, - { - "epoch": 0.26, - "learning_rate": 1.982574206618671e-05, - "loss": 1.5789, - "step": 326 - }, - { - "epoch": 0.26, - "learning_rate": 1.982411448432617e-05, - "loss": 1.7475, - "step": 327 - }, - { - "epoch": 0.27, - "learning_rate": 1.98224794043218e-05, - "loss": 1.5694, - "step": 328 - }, - { - "epoch": 0.27, - "learning_rate": 1.982083682742156e-05, - "loss": 1.6314, - "step": 329 - }, - { - "epoch": 0.27, - "learning_rate": 1.9819186754879137e-05, - "loss": 1.4827, - "step": 330 - }, - { - "epoch": 0.27, - "learning_rate": 1.9817529187953918e-05, - "loss": 1.6933, - "step": 331 - }, - { - "epoch": 0.27, - "learning_rate": 1.981586412791103e-05, - "loss": 1.6055, - "step": 332 - }, - { - "epoch": 0.27, - "learning_rate": 1.981419157602131e-05, - "loss": 1.5742, - "step": 333 - }, - { - "epoch": 0.27, - "learning_rate": 1.9812511533561314e-05, - "loss": 1.514, - "step": 334 - }, - { - "epoch": 0.27, - "learning_rate": 1.981082400181331e-05, - "loss": 1.6631, - "step": 335 - }, - { - "epoch": 0.27, - "learning_rate": 1.9809128982065296e-05, - "loss": 1.7324, - "step": 336 - }, - { - "epoch": 0.27, - "learning_rate": 1.9807426475610972e-05, - "loss": 1.6263, - "step": 337 - }, - { - "epoch": 0.27, - "learning_rate": 1.980571648374976e-05, - "loss": 1.6199, - "step": 338 - }, - { - "epoch": 0.27, - "learning_rate": 1.9803999007786783e-05, - "loss": 1.6602, - "step": 339 - }, - { - "epoch": 0.28, - "learning_rate": 1.9802274049032898e-05, - "loss": 1.6846, - "step": 340 - }, - { - "epoch": 0.28, - "learning_rate": 1.980054160880465e-05, - "loss": 1.5962, - "step": 341 - }, - { - "epoch": 0.28, - "learning_rate": 1.9798801688424307e-05, - "loss": 1.6799, - "step": 342 - }, - { - "epoch": 0.28, - "learning_rate": 1.9797054289219845e-05, - "loss": 1.542, - "step": 343 - }, - { - "epoch": 0.28, - "learning_rate": 1.9795299412524948e-05, - "loss": 1.4955, - "step": 344 - }, - { - "epoch": 0.28, - "learning_rate": 1.9793537059679e-05, - "loss": 1.5883, - "step": 345 - }, - { - "epoch": 0.28, - "learning_rate": 1.97917672320271e-05, - "loss": 1.6325, - "step": 346 - }, - { - "epoch": 0.28, - "learning_rate": 1.9789989930920056e-05, - "loss": 1.5959, - "step": 347 - }, - { - "epoch": 0.28, - "learning_rate": 1.978820515771436e-05, - "loss": 1.4601, - "step": 348 - }, - { - "epoch": 0.28, - "learning_rate": 1.9786412913772225e-05, - "loss": 1.5219, - "step": 349 - }, - { - "epoch": 0.28, - "learning_rate": 1.9784613200461568e-05, - "loss": 1.5457, - "step": 350 - }, - { - "epoch": 0.28, - "learning_rate": 1.978280601915599e-05, - "loss": 1.4609, - "step": 351 - }, - { - "epoch": 0.28, - "learning_rate": 1.9780991371234803e-05, - "loss": 1.4623, - "step": 352 - }, - { - "epoch": 0.29, - "learning_rate": 1.977916925808302e-05, - "loss": 1.6013, - "step": 353 - }, - { - "epoch": 0.29, - "learning_rate": 1.977733968109135e-05, - "loss": 1.4784, - "step": 354 - }, - { - "epoch": 0.29, - "learning_rate": 1.977550264165619e-05, - "loss": 1.6438, - "step": 355 - }, - { - "epoch": 0.29, - "learning_rate": 1.9773658141179642e-05, - "loss": 1.8001, - "step": 356 - }, - { - "epoch": 0.29, - "learning_rate": 1.9771806181069504e-05, - "loss": 1.5841, - "step": 357 - }, - { - "epoch": 0.29, - "learning_rate": 1.9769946762739262e-05, - "loss": 1.4975, - "step": 358 - }, - { - "epoch": 0.29, - "learning_rate": 1.9768079887608093e-05, - "loss": 1.6086, - "step": 359 - }, - { - "epoch": 0.29, - "learning_rate": 1.976620555710087e-05, - "loss": 1.6058, - "step": 360 - }, - { - "epoch": 0.29, - "learning_rate": 1.9764323772648153e-05, - "loss": 1.6068, - "step": 361 - }, - { - "epoch": 0.29, - "learning_rate": 1.9762434535686196e-05, - "loss": 1.6563, - "step": 362 - }, - { - "epoch": 0.29, - "learning_rate": 1.9760537847656932e-05, - "loss": 1.5562, - "step": 363 - }, - { - "epoch": 0.29, - "learning_rate": 1.9758633710007995e-05, - "loss": 1.467, - "step": 364 - }, - { - "epoch": 0.3, - "learning_rate": 1.975672212419269e-05, - "loss": 1.5169, - "step": 365 - }, - { - "epoch": 0.3, - "learning_rate": 1.9754803091670013e-05, - "loss": 1.593, - "step": 366 - }, - { - "epoch": 0.3, - "learning_rate": 1.9752876613904645e-05, - "loss": 1.4927, - "step": 367 - }, - { - "epoch": 0.3, - "learning_rate": 1.9750942692366953e-05, - "loss": 1.6924, - "step": 368 - }, - { - "epoch": 0.3, - "learning_rate": 1.9749001328532973e-05, - "loss": 1.577, - "step": 369 - }, - { - "epoch": 0.3, - "learning_rate": 1.9747052523884435e-05, - "loss": 1.6532, - "step": 370 - }, - { - "epoch": 0.3, - "learning_rate": 1.9745096279908737e-05, - "loss": 1.5701, - "step": 371 - }, - { - "epoch": 0.3, - "learning_rate": 1.9743132598098965e-05, - "loss": 1.6504, - "step": 372 - }, - { - "epoch": 0.3, - "learning_rate": 1.9741161479953872e-05, - "loss": 1.4996, - "step": 373 - }, - { - "epoch": 0.3, - "learning_rate": 1.9739182926977895e-05, - "loss": 1.5383, - "step": 374 - }, - { - "epoch": 0.3, - "learning_rate": 1.973719694068114e-05, - "loss": 1.7158, - "step": 375 - }, - { - "epoch": 0.3, - "learning_rate": 1.973520352257939e-05, - "loss": 1.4944, - "step": 376 - }, - { - "epoch": 0.31, - "learning_rate": 1.97332026741941e-05, - "loss": 1.5549, - "step": 377 - }, - { - "epoch": 0.31, - "learning_rate": 1.973119439705239e-05, - "loss": 1.6078, - "step": 378 - }, - { - "epoch": 0.31, - "learning_rate": 1.9729178692687058e-05, - "loss": 1.7062, - "step": 379 - }, - { - "epoch": 0.31, - "learning_rate": 1.972715556263657e-05, - "loss": 1.5386, - "step": 380 - }, - { - "epoch": 0.31, - "learning_rate": 1.972512500844505e-05, - "loss": 1.6967, - "step": 381 - }, - { - "epoch": 0.31, - "learning_rate": 1.9723087031662304e-05, - "loss": 1.445, - "step": 382 - }, - { - "epoch": 0.31, - "learning_rate": 1.972104163384379e-05, - "loss": 1.5644, - "step": 383 - }, - { - "epoch": 0.31, - "learning_rate": 1.9718988816550633e-05, - "loss": 1.6197, - "step": 384 - }, - { - "epoch": 0.31, - "learning_rate": 1.9716928581349622e-05, - "loss": 1.4858, - "step": 385 - }, - { - "epoch": 0.31, - "learning_rate": 1.9714860929813217e-05, - "loss": 1.5565, - "step": 386 - }, - { - "epoch": 0.31, - "learning_rate": 1.9712785863519516e-05, - "loss": 1.5098, - "step": 387 - }, - { - "epoch": 0.31, - "learning_rate": 1.97107033840523e-05, - "loss": 1.6171, - "step": 388 - }, - { - "epoch": 0.31, - "learning_rate": 1.9708613493000997e-05, - "loss": 1.5721, - "step": 389 - }, - { - "epoch": 0.32, - "learning_rate": 1.9706516191960687e-05, - "loss": 1.7274, - "step": 390 - }, - { - "epoch": 0.32, - "learning_rate": 1.9704411482532116e-05, - "loss": 1.5431, - "step": 391 - }, - { - "epoch": 0.32, - "learning_rate": 1.970229936632168e-05, - "loss": 1.5385, - "step": 392 - }, - { - "epoch": 0.32, - "learning_rate": 1.970017984494142e-05, - "loss": 1.6541, - "step": 393 - }, - { - "epoch": 0.32, - "learning_rate": 1.9698052920009046e-05, - "loss": 1.3581, - "step": 394 - }, - { - "epoch": 0.32, - "learning_rate": 1.9695918593147908e-05, - "loss": 1.4411, - "step": 395 - }, - { - "epoch": 0.32, - "learning_rate": 1.9693776865987e-05, - "loss": 1.5492, - "step": 396 - }, - { - "epoch": 0.32, - "learning_rate": 1.9691627740160978e-05, - "loss": 1.6808, - "step": 397 - }, - { - "epoch": 0.32, - "learning_rate": 1.9689471217310134e-05, - "loss": 1.5448, - "step": 398 - }, - { - "epoch": 0.32, - "learning_rate": 1.968730729908041e-05, - "loss": 1.4919, - "step": 399 - }, - { - "epoch": 0.32, - "learning_rate": 1.9685135987123396e-05, - "loss": 1.5776, - "step": 400 - }, - { - "epoch": 0.32, - "learning_rate": 1.9682957283096317e-05, - "loss": 1.5923, - "step": 401 - }, - { - "epoch": 0.33, - "learning_rate": 1.9680771188662044e-05, - "loss": 1.6841, - "step": 402 - }, - { - "epoch": 0.33, - "learning_rate": 1.9678577705489092e-05, - "loss": 1.7185, - "step": 403 - }, - { - "epoch": 0.33, - "learning_rate": 1.9676376835251608e-05, - "loss": 1.5113, - "step": 404 - }, - { - "epoch": 0.33, - "learning_rate": 1.9674168579629384e-05, - "loss": 1.6404, - "step": 405 - }, - { - "epoch": 0.33, - "learning_rate": 1.9671952940307845e-05, - "loss": 1.6501, - "step": 406 - }, - { - "epoch": 0.33, - "learning_rate": 1.9669729918978054e-05, - "loss": 1.629, - "step": 407 - }, - { - "epoch": 0.33, - "learning_rate": 1.9667499517336704e-05, - "loss": 1.6265, - "step": 408 - }, - { - "epoch": 0.33, - "learning_rate": 1.9665261737086125e-05, - "loss": 1.5197, - "step": 409 - }, - { - "epoch": 0.33, - "learning_rate": 1.966301657993428e-05, - "loss": 1.6681, - "step": 410 - }, - { - "epoch": 0.33, - "learning_rate": 1.9660764047594756e-05, - "loss": 1.7421, - "step": 411 - }, - { - "epoch": 0.33, - "learning_rate": 1.9658504141786775e-05, - "loss": 1.6098, - "step": 412 - }, - { - "epoch": 0.33, - "learning_rate": 1.9656236864235182e-05, - "loss": 1.5717, - "step": 413 - }, - { - "epoch": 0.33, - "learning_rate": 1.9653962216670453e-05, - "loss": 1.6651, - "step": 414 - }, - { - "epoch": 0.34, - "learning_rate": 1.965168020082869e-05, - "loss": 1.7182, - "step": 415 - }, - { - "epoch": 0.34, - "learning_rate": 1.9649390818451607e-05, - "loss": 1.4928, - "step": 416 - }, - { - "epoch": 0.34, - "learning_rate": 1.9647094071286554e-05, - "loss": 1.6534, - "step": 417 - }, - { - "epoch": 0.34, - "learning_rate": 1.96447899610865e-05, - "loss": 1.7108, - "step": 418 - }, - { - "epoch": 0.34, - "learning_rate": 1.9642478489610028e-05, - "loss": 1.5155, - "step": 419 - }, - { - "epoch": 0.34, - "learning_rate": 1.9640159658621344e-05, - "loss": 1.6454, - "step": 420 - }, - { - "epoch": 0.34, - "learning_rate": 1.9637833469890264e-05, - "loss": 1.5031, - "step": 421 - }, - { - "epoch": 0.34, - "learning_rate": 1.963549992519223e-05, - "loss": 1.4747, - "step": 422 - }, - { - "epoch": 0.34, - "learning_rate": 1.9633159026308296e-05, - "loss": 1.6183, - "step": 423 - }, - { - "epoch": 0.34, - "learning_rate": 1.963081077502512e-05, - "loss": 1.5567, - "step": 424 - }, - { - "epoch": 0.34, - "learning_rate": 1.9628455173134985e-05, - "loss": 1.5225, - "step": 425 - }, - { - "epoch": 0.34, - "learning_rate": 1.962609222243577e-05, - "loss": 1.7058, - "step": 426 - }, - { - "epoch": 0.35, - "learning_rate": 1.9623721924730975e-05, - "loss": 1.7531, - "step": 427 - }, - { - "epoch": 0.35, - "learning_rate": 1.9621344281829703e-05, - "loss": 1.6401, - "step": 428 - }, - { - "epoch": 0.35, - "learning_rate": 1.961895929554666e-05, - "loss": 1.5045, - "step": 429 - }, - { - "epoch": 0.35, - "learning_rate": 1.9616566967702164e-05, - "loss": 1.6029, - "step": 430 - }, - { - "epoch": 0.35, - "learning_rate": 1.9614167300122126e-05, - "loss": 1.5846, - "step": 431 - }, - { - "epoch": 0.35, - "learning_rate": 1.961176029463807e-05, - "loss": 1.6231, - "step": 432 - }, - { - "epoch": 0.35, - "learning_rate": 1.9609345953087115e-05, - "loss": 1.5313, - "step": 433 - }, - { - "epoch": 0.35, - "learning_rate": 1.960692427731198e-05, - "loss": 1.5396, - "step": 434 - }, - { - "epoch": 0.35, - "learning_rate": 1.9604495269160977e-05, - "loss": 1.4325, - "step": 435 - }, - { - "epoch": 0.35, - "learning_rate": 1.9602058930488025e-05, - "loss": 1.5985, - "step": 436 - }, - { - "epoch": 0.35, - "learning_rate": 1.959961526315263e-05, - "loss": 1.5668, - "step": 437 - }, - { - "epoch": 0.35, - "learning_rate": 1.9597164269019888e-05, - "loss": 1.5379, - "step": 438 - }, - { - "epoch": 0.36, - "learning_rate": 1.95947059499605e-05, - "loss": 1.6883, - "step": 439 - }, - { - "epoch": 0.36, - "learning_rate": 1.9592240307850748e-05, - "loss": 1.5937, - "step": 440 - }, - { - "epoch": 0.36, - "learning_rate": 1.9589767344572505e-05, - "loss": 1.5392, - "step": 441 - }, - { - "epoch": 0.36, - "learning_rate": 1.9587287062013226e-05, - "loss": 1.5143, - "step": 442 - }, - { - "epoch": 0.36, - "learning_rate": 1.958479946206597e-05, - "loss": 1.5949, - "step": 443 - }, - { - "epoch": 0.36, - "learning_rate": 1.9582304546629363e-05, - "loss": 1.5551, - "step": 444 - }, - { - "epoch": 0.36, - "learning_rate": 1.9579802317607618e-05, - "loss": 1.5439, - "step": 445 - }, - { - "epoch": 0.36, - "learning_rate": 1.957729277691054e-05, - "loss": 1.576, - "step": 446 - }, - { - "epoch": 0.36, - "learning_rate": 1.95747759264535e-05, - "loss": 1.7002, - "step": 447 - }, - { - "epoch": 0.36, - "learning_rate": 1.957225176815746e-05, - "loss": 1.595, - "step": 448 - }, - { - "epoch": 0.36, - "learning_rate": 1.9569720303948955e-05, - "loss": 1.7449, - "step": 449 - }, - { - "epoch": 0.36, - "learning_rate": 1.95671815357601e-05, - "loss": 1.6706, - "step": 450 - }, - { - "epoch": 0.36, - "learning_rate": 1.956463546552857e-05, - "loss": 1.5549, - "step": 451 - }, - { - "epoch": 0.37, - "learning_rate": 1.956208209519763e-05, - "loss": 1.7118, - "step": 452 - }, - { - "epoch": 0.37, - "learning_rate": 1.955952142671612e-05, - "loss": 1.6668, - "step": 453 - }, - { - "epoch": 0.37, - "learning_rate": 1.9556953462038425e-05, - "loss": 1.5327, - "step": 454 - }, - { - "epoch": 0.37, - "learning_rate": 1.9554378203124525e-05, - "loss": 1.5816, - "step": 455 - }, - { - "epoch": 0.37, - "learning_rate": 1.9551795651939956e-05, - "loss": 1.5778, - "step": 456 - }, - { - "epoch": 0.37, - "learning_rate": 1.954920581045582e-05, - "loss": 1.6379, - "step": 457 - }, - { - "epoch": 0.37, - "learning_rate": 1.9546608680648784e-05, - "loss": 1.565, - "step": 458 - }, - { - "epoch": 0.37, - "learning_rate": 1.9544004264501077e-05, - "loss": 1.7145, - "step": 459 - }, - { - "epoch": 0.37, - "learning_rate": 1.954139256400049e-05, - "loss": 1.5366, - "step": 460 - }, - { - "epoch": 0.37, - "learning_rate": 1.9538773581140377e-05, - "loss": 1.6652, - "step": 461 - }, - { - "epoch": 0.37, - "learning_rate": 1.953614731791964e-05, - "loss": 1.5889, - "step": 462 - }, - { - "epoch": 0.37, - "learning_rate": 1.9533513776342753e-05, - "loss": 1.3155, - "step": 463 - }, - { - "epoch": 0.38, - "learning_rate": 1.953087295841973e-05, - "loss": 1.7268, - "step": 464 - }, - { - "epoch": 0.38, - "learning_rate": 1.952822486616615e-05, - "loss": 1.4518, - "step": 465 - }, - { - "epoch": 0.38, - "learning_rate": 1.9525569501603137e-05, - "loss": 1.7402, - "step": 466 - }, - { - "epoch": 0.38, - "learning_rate": 1.952290686675737e-05, - "loss": 1.4279, - "step": 467 - }, - { - "epoch": 0.38, - "learning_rate": 1.9520236963661068e-05, - "loss": 1.6124, - "step": 468 - }, - { - "epoch": 0.38, - "learning_rate": 1.9517559794352013e-05, - "loss": 1.6838, - "step": 469 - }, - { - "epoch": 0.38, - "learning_rate": 1.951487536087352e-05, - "loss": 1.4889, - "step": 470 - }, - { - "epoch": 0.38, - "learning_rate": 1.9512183665274452e-05, - "loss": 1.6696, - "step": 471 - }, - { - "epoch": 0.38, - "learning_rate": 1.9509484709609214e-05, - "loss": 1.5193, - "step": 472 - }, - { - "epoch": 0.38, - "learning_rate": 1.9506778495937758e-05, - "loss": 1.6827, - "step": 473 - }, - { - "epoch": 0.38, - "learning_rate": 1.9504065026325567e-05, - "loss": 1.4621, - "step": 474 - }, - { - "epoch": 0.38, - "learning_rate": 1.9501344302843667e-05, - "loss": 1.493, - "step": 475 - }, - { - "epoch": 0.39, - "learning_rate": 1.9498616327568617e-05, - "loss": 1.4611, - "step": 476 - }, - { - "epoch": 0.39, - "learning_rate": 1.949588110258252e-05, - "loss": 1.4707, - "step": 477 - }, - { - "epoch": 0.39, - "learning_rate": 1.9493138629972998e-05, - "loss": 1.5931, - "step": 478 - }, - { - "epoch": 0.39, - "learning_rate": 1.9490388911833213e-05, - "loss": 1.6163, - "step": 479 - }, - { - "epoch": 0.39, - "learning_rate": 1.948763195026186e-05, - "loss": 1.4232, - "step": 480 - }, - { - "epoch": 0.39, - "learning_rate": 1.9484867747363153e-05, - "loss": 1.54, - "step": 481 - }, - { - "epoch": 0.39, - "learning_rate": 1.9482096305246846e-05, - "loss": 1.5112, - "step": 482 - }, - { - "epoch": 0.39, - "learning_rate": 1.9479317626028202e-05, - "loss": 1.6015, - "step": 483 - }, - { - "epoch": 0.39, - "learning_rate": 1.9476531711828027e-05, - "loss": 1.5471, - "step": 484 - }, - { - "epoch": 0.39, - "learning_rate": 1.9473738564772628e-05, - "loss": 1.5595, - "step": 485 - }, - { - "epoch": 0.39, - "learning_rate": 1.947093818699385e-05, - "loss": 1.4943, - "step": 486 - }, - { - "epoch": 0.39, - "learning_rate": 1.9468130580629047e-05, - "loss": 1.7047, - "step": 487 - }, - { - "epoch": 0.39, - "learning_rate": 1.946531574782109e-05, - "loss": 1.5044, - "step": 488 - }, - { - "epoch": 0.4, - "learning_rate": 1.9462493690718373e-05, - "loss": 1.4776, - "step": 489 - }, - { - "epoch": 0.4, - "learning_rate": 1.9459664411474793e-05, - "loss": 1.5203, - "step": 490 - }, - { - "epoch": 0.4, - "learning_rate": 1.9456827912249768e-05, - "loss": 1.6122, - "step": 491 - }, - { - "epoch": 0.4, - "learning_rate": 1.945398419520823e-05, - "loss": 1.4943, - "step": 492 - }, - { - "epoch": 0.4, - "learning_rate": 1.9451133262520604e-05, - "loss": 1.5644, - "step": 493 - }, - { - "epoch": 0.4, - "learning_rate": 1.944827511636284e-05, - "loss": 1.5666, - "step": 494 - }, - { - "epoch": 0.4, - "learning_rate": 1.944540975891638e-05, - "loss": 1.635, - "step": 495 - }, - { - "epoch": 0.4, - "learning_rate": 1.944253719236818e-05, - "loss": 1.58, - "step": 496 - }, - { - "epoch": 0.4, - "learning_rate": 1.9439657418910687e-05, - "loss": 1.4817, - "step": 497 - }, - { - "epoch": 0.4, - "learning_rate": 1.9436770440741862e-05, - "loss": 1.5677, - "step": 498 - }, - { - "epoch": 0.4, - "learning_rate": 1.9433876260065157e-05, - "loss": 1.4748, - "step": 499 - }, - { - "epoch": 0.4, - "learning_rate": 1.9430974879089522e-05, - "loss": 1.5838, - "step": 500 - }, - { - "epoch": 0.41, - "learning_rate": 1.94280663000294e-05, - "loss": 1.601, - "step": 501 - }, - { - "epoch": 0.41, - "learning_rate": 1.9425150525104738e-05, - "loss": 1.6091, - "step": 502 - }, - { - "epoch": 0.41, - "learning_rate": 1.9422227556540957e-05, - "loss": 1.4875, - "step": 503 - }, - { - "epoch": 0.41, - "learning_rate": 1.941929739656899e-05, - "loss": 1.5109, - "step": 504 - }, - { - "epoch": 0.41, - "learning_rate": 1.9416360047425246e-05, - "loss": 1.507, - "step": 505 - }, - { - "epoch": 0.41, - "learning_rate": 1.9413415511351622e-05, - "loss": 1.6741, - "step": 506 - }, - { - "epoch": 0.41, - "learning_rate": 1.9410463790595496e-05, - "loss": 1.5454, - "step": 507 - }, - { - "epoch": 0.41, - "learning_rate": 1.940750488740975e-05, - "loss": 1.6019, - "step": 508 - }, - { - "epoch": 0.41, - "learning_rate": 1.9404538804052718e-05, - "loss": 1.4606, - "step": 509 - }, - { - "epoch": 0.41, - "learning_rate": 1.9401565542788238e-05, - "loss": 1.5908, - "step": 510 - }, - { - "epoch": 0.41, - "learning_rate": 1.9398585105885613e-05, - "loss": 1.6558, - "step": 511 - }, - { - "epoch": 0.41, - "learning_rate": 1.9395597495619634e-05, - "loss": 1.535, - "step": 512 - }, - { - "epoch": 0.42, - "learning_rate": 1.9392602714270555e-05, - "loss": 1.5881, - "step": 513 - }, - { - "epoch": 0.42, - "learning_rate": 1.938960076412411e-05, - "loss": 1.3882, - "step": 514 - }, - { - "epoch": 0.42, - "learning_rate": 1.9386591647471508e-05, - "loss": 1.6159, - "step": 515 - }, - { - "epoch": 0.42, - "learning_rate": 1.9383575366609416e-05, - "loss": 1.7844, - "step": 516 - }, - { - "epoch": 0.42, - "learning_rate": 1.938055192383998e-05, - "loss": 1.5926, - "step": 517 - }, - { - "epoch": 0.42, - "learning_rate": 1.9377521321470806e-05, - "loss": 1.4945, - "step": 518 - }, - { - "epoch": 0.42, - "learning_rate": 1.937448356181497e-05, - "loss": 1.6406, - "step": 519 - }, - { - "epoch": 0.42, - "learning_rate": 1.9371438647191007e-05, - "loss": 1.4952, - "step": 520 - }, - { - "epoch": 0.42, - "learning_rate": 1.936838657992291e-05, - "loss": 1.5827, - "step": 521 - }, - { - "epoch": 0.42, - "learning_rate": 1.9365327362340142e-05, - "loss": 1.5993, - "step": 522 - }, - { - "epoch": 0.42, - "learning_rate": 1.9362260996777608e-05, - "loss": 1.6468, - "step": 523 - }, - { - "epoch": 0.42, - "learning_rate": 1.9359187485575678e-05, - "loss": 1.4345, - "step": 524 - }, - { - "epoch": 0.42, - "learning_rate": 1.935610683108018e-05, - "loss": 1.7226, - "step": 525 - }, - { - "epoch": 0.43, - "learning_rate": 1.9353019035642385e-05, - "loss": 1.4115, - "step": 526 - }, - { - "epoch": 0.43, - "learning_rate": 1.934992410161902e-05, - "loss": 1.482, - "step": 527 - }, - { - "epoch": 0.43, - "learning_rate": 1.934682203137225e-05, - "loss": 1.6858, - "step": 528 - }, - { - "epoch": 0.43, - "learning_rate": 1.9343712827269705e-05, - "loss": 1.5908, - "step": 529 - }, - { - "epoch": 0.43, - "learning_rate": 1.9340596491684443e-05, - "loss": 1.7148, - "step": 530 - }, - { - "epoch": 0.43, - "learning_rate": 1.933747302699498e-05, - "loss": 1.4624, - "step": 531 - }, - { - "epoch": 0.43, - "learning_rate": 1.9334342435585257e-05, - "loss": 1.4726, - "step": 532 - }, - { - "epoch": 0.43, - "learning_rate": 1.933120471984466e-05, - "loss": 1.467, - "step": 533 - }, - { - "epoch": 0.43, - "learning_rate": 1.9328059882168026e-05, - "loss": 1.4646, - "step": 534 - }, - { - "epoch": 0.43, - "learning_rate": 1.932490792495561e-05, - "loss": 1.6675, - "step": 535 - }, - { - "epoch": 0.43, - "learning_rate": 1.9321748850613104e-05, - "loss": 1.6211, - "step": 536 - }, - { - "epoch": 0.43, - "learning_rate": 1.9318582661551643e-05, - "loss": 1.5121, - "step": 537 - }, - { - "epoch": 0.44, - "learning_rate": 1.9315409360187784e-05, - "loss": 1.4222, - "step": 538 - }, - { - "epoch": 0.44, - "learning_rate": 1.9312228948943504e-05, - "loss": 1.5154, - "step": 539 - }, - { - "epoch": 0.44, - "learning_rate": 1.9309041430246228e-05, - "loss": 1.6363, - "step": 540 - }, - { - "epoch": 0.44, - "learning_rate": 1.9305846806528784e-05, - "loss": 1.695, - "step": 541 - }, - { - "epoch": 0.44, - "learning_rate": 1.930264508022944e-05, - "loss": 1.5745, - "step": 542 - }, - { - "epoch": 0.44, - "learning_rate": 1.929943625379187e-05, - "loss": 1.5454, - "step": 543 - }, - { - "epoch": 0.44, - "learning_rate": 1.9296220329665177e-05, - "loss": 1.6461, - "step": 544 - }, - { - "epoch": 0.44, - "learning_rate": 1.929299731030388e-05, - "loss": 1.5995, - "step": 545 - }, - { - "epoch": 0.44, - "learning_rate": 1.9289767198167918e-05, - "loss": 1.5414, - "step": 546 - }, - { - "epoch": 0.44, - "learning_rate": 1.9286529995722624e-05, - "loss": 1.6214, - "step": 547 - }, - { - "epoch": 0.44, - "learning_rate": 1.9283285705438764e-05, - "loss": 1.6373, - "step": 548 - }, - { - "epoch": 0.44, - "learning_rate": 1.9280034329792503e-05, - "loss": 1.7244, - "step": 549 - }, - { - "epoch": 0.44, - "learning_rate": 1.927677587126542e-05, - "loss": 1.3249, - "step": 550 - }, - { - "epoch": 0.45, - "learning_rate": 1.9273510332344495e-05, - "loss": 1.4731, - "step": 551 - }, - { - "epoch": 0.45, - "learning_rate": 1.9270237715522112e-05, - "loss": 1.6399, - "step": 552 - }, - { - "epoch": 0.45, - "learning_rate": 1.9266958023296058e-05, - "loss": 1.3381, - "step": 553 - }, - { - "epoch": 0.45, - "learning_rate": 1.9263671258169524e-05, - "loss": 1.6866, - "step": 554 - }, - { - "epoch": 0.45, - "learning_rate": 1.9260377422651094e-05, - "loss": 1.6581, - "step": 555 - }, - { - "epoch": 0.45, - "learning_rate": 1.925707651925475e-05, - "loss": 1.5061, - "step": 556 - }, - { - "epoch": 0.45, - "learning_rate": 1.9253768550499873e-05, - "loss": 1.404, - "step": 557 - }, - { - "epoch": 0.45, - "learning_rate": 1.925045351891123e-05, - "loss": 1.5398, - "step": 558 - }, - { - "epoch": 0.45, - "learning_rate": 1.9247131427018977e-05, - "loss": 1.5863, - "step": 559 - }, - { - "epoch": 0.45, - "learning_rate": 1.924380227735867e-05, - "loss": 1.5012, - "step": 560 - }, - { - "epoch": 0.45, - "learning_rate": 1.924046607247124e-05, - "loss": 1.615, - "step": 561 - }, - { - "epoch": 0.45, - "learning_rate": 1.923712281490301e-05, - "loss": 1.532, - "step": 562 - }, - { - "epoch": 0.46, - "learning_rate": 1.9233772507205682e-05, - "loss": 1.5933, - "step": 563 - }, - { - "epoch": 0.46, - "learning_rate": 1.923041515193634e-05, - "loss": 1.3475, - "step": 564 - }, - { - "epoch": 0.46, - "learning_rate": 1.9227050751657448e-05, - "loss": 1.6313, - "step": 565 - }, - { - "epoch": 0.46, - "learning_rate": 1.922367930893685e-05, - "loss": 1.5558, - "step": 566 - }, - { - "epoch": 0.46, - "learning_rate": 1.9220300826347752e-05, - "loss": 1.5801, - "step": 567 - }, - { - "epoch": 0.46, - "learning_rate": 1.921691530646875e-05, - "loss": 1.5042, - "step": 568 - }, - { - "epoch": 0.46, - "learning_rate": 1.9213522751883803e-05, - "loss": 1.5519, - "step": 569 - }, - { - "epoch": 0.46, - "learning_rate": 1.921012316518224e-05, - "loss": 1.6382, - "step": 570 - }, - { - "epoch": 0.46, - "learning_rate": 1.9206716548958754e-05, - "loss": 1.5461, - "step": 571 - }, - { - "epoch": 0.46, - "learning_rate": 1.9203302905813407e-05, - "loss": 1.4358, - "step": 572 - }, - { - "epoch": 0.46, - "learning_rate": 1.9199882238351622e-05, - "loss": 1.439, - "step": 573 - }, - { - "epoch": 0.46, - "learning_rate": 1.9196454549184188e-05, - "loss": 1.7074, - "step": 574 - }, - { - "epoch": 0.47, - "learning_rate": 1.9193019840927252e-05, - "loss": 1.7553, - "step": 575 - }, - { - "epoch": 0.47, - "learning_rate": 1.918957811620231e-05, - "loss": 1.6474, - "step": 576 - }, - { - "epoch": 0.47, - "learning_rate": 1.918612937763622e-05, - "loss": 1.5455, - "step": 577 - }, - { - "epoch": 0.47, - "learning_rate": 1.9182673627861198e-05, - "loss": 1.6045, - "step": 578 - }, - { - "epoch": 0.47, - "learning_rate": 1.9179210869514796e-05, - "loss": 1.6658, - "step": 579 - }, - { - "epoch": 0.47, - "learning_rate": 1.917574110523994e-05, - "loss": 1.4631, - "step": 580 - }, - { - "epoch": 0.47, - "learning_rate": 1.917226433768487e-05, - "loss": 1.3291, - "step": 581 - }, - { - "epoch": 0.47, - "learning_rate": 1.9168780569503202e-05, - "loss": 1.6936, - "step": 582 - }, - { - "epoch": 0.47, - "learning_rate": 1.916528980335388e-05, - "loss": 1.5625, - "step": 583 - }, - { - "epoch": 0.47, - "learning_rate": 1.9161792041901193e-05, - "loss": 1.4438, - "step": 584 - }, - { - "epoch": 0.47, - "learning_rate": 1.9158287287814762e-05, - "loss": 1.5263, - "step": 585 - }, - { - "epoch": 0.47, - "learning_rate": 1.9154775543769558e-05, - "loss": 1.5367, - "step": 586 - }, - { - "epoch": 0.47, - "learning_rate": 1.9151256812445872e-05, - "loss": 1.4222, - "step": 587 - }, - { - "epoch": 0.48, - "learning_rate": 1.9147731096529343e-05, - "loss": 1.4522, - "step": 588 - }, - { - "epoch": 0.48, - "learning_rate": 1.9144198398710927e-05, - "loss": 1.761, - "step": 589 - }, - { - "epoch": 0.48, - "learning_rate": 1.914065872168692e-05, - "loss": 1.6548, - "step": 590 - }, - { - "epoch": 0.48, - "learning_rate": 1.9137112068158942e-05, - "loss": 1.6458, - "step": 591 - }, - { - "epoch": 0.48, - "learning_rate": 1.9133558440833928e-05, - "loss": 1.5989, - "step": 592 - }, - { - "epoch": 0.48, - "learning_rate": 1.9129997842424153e-05, - "loss": 1.5292, - "step": 593 - }, - { - "epoch": 0.48, - "learning_rate": 1.91264302756472e-05, - "loss": 1.5206, - "step": 594 - }, - { - "epoch": 0.48, - "learning_rate": 1.912285574322597e-05, - "loss": 1.5348, - "step": 595 - }, - { - "epoch": 0.48, - "learning_rate": 1.911927424788869e-05, - "loss": 1.5651, - "step": 596 - }, - { - "epoch": 0.48, - "learning_rate": 1.911568579236889e-05, - "loss": 1.4727, - "step": 597 - }, - { - "epoch": 0.48, - "learning_rate": 1.9112090379405422e-05, - "loss": 1.546, - "step": 598 - }, - { - "epoch": 0.48, - "learning_rate": 1.9108488011742444e-05, - "loss": 1.6373, - "step": 599 - }, - { - "epoch": 0.49, - "learning_rate": 1.910487869212942e-05, - "loss": 1.5112, - "step": 600 - }, - { - "epoch": 0.49, - "learning_rate": 1.9101262423321123e-05, - "loss": 1.3084, - "step": 601 - }, - { - "epoch": 0.49, - "learning_rate": 1.9097639208077633e-05, - "loss": 1.5369, - "step": 602 - }, - { - "epoch": 0.49, - "learning_rate": 1.9094009049164318e-05, - "loss": 1.5971, - "step": 603 - }, - { - "epoch": 0.49, - "learning_rate": 1.9090371949351868e-05, - "loss": 1.6469, - "step": 604 - }, - { - "epoch": 0.49, - "learning_rate": 1.908672791141625e-05, - "loss": 1.5293, - "step": 605 - }, - { - "epoch": 0.49, - "learning_rate": 1.908307693813874e-05, - "loss": 1.4721, - "step": 606 - }, - { - "epoch": 0.49, - "learning_rate": 1.9079419032305892e-05, - "loss": 1.6315, - "step": 607 - }, - { - "epoch": 0.49, - "learning_rate": 1.9075754196709574e-05, - "loss": 1.4915, - "step": 608 - }, - { - "epoch": 0.49, - "learning_rate": 1.907208243414692e-05, - "loss": 1.6837, - "step": 609 - }, - { - "epoch": 0.49, - "learning_rate": 1.9068403747420365e-05, - "loss": 1.5794, - "step": 610 - }, - { - "epoch": 0.49, - "learning_rate": 1.9064718139337624e-05, - "loss": 1.5364, - "step": 611 - }, - { - "epoch": 0.5, - "learning_rate": 1.90610256127117e-05, - "loss": 1.6605, - "step": 612 - }, - { - "epoch": 0.5, - "learning_rate": 1.9057326170360866e-05, - "loss": 1.4934, - "step": 613 - }, - { - "epoch": 0.5, - "learning_rate": 1.9053619815108684e-05, - "loss": 1.6583, - "step": 614 - }, - { - "epoch": 0.5, - "learning_rate": 1.9049906549783982e-05, - "loss": 1.4754, - "step": 615 - }, - { - "epoch": 0.5, - "learning_rate": 1.904618637722087e-05, - "loss": 1.6617, - "step": 616 - }, - { - "epoch": 0.5, - "learning_rate": 1.9042459300258733e-05, - "loss": 1.5868, - "step": 617 - }, - { - "epoch": 0.5, - "learning_rate": 1.9038725321742212e-05, - "loss": 1.5668, - "step": 618 - }, - { - "epoch": 0.5, - "learning_rate": 1.9034984444521227e-05, - "loss": 1.584, - "step": 619 - }, - { - "epoch": 0.5, - "learning_rate": 1.9031236671450963e-05, - "loss": 1.5652, - "step": 620 - }, - { - "epoch": 0.5, - "learning_rate": 1.902748200539186e-05, - "loss": 1.6257, - "step": 621 - }, - { - "epoch": 0.5, - "learning_rate": 1.902372044920963e-05, - "loss": 1.5139, - "step": 622 - }, - { - "epoch": 0.5, - "learning_rate": 1.901995200577523e-05, - "loss": 1.5986, - "step": 623 - }, - { - "epoch": 0.5, - "learning_rate": 1.9016176677964887e-05, - "loss": 1.5699, - "step": 624 - }, - { - "epoch": 0.51, - "learning_rate": 1.901239446866008e-05, - "loss": 1.6853, - "step": 625 - }, - { - "epoch": 0.51, - "learning_rate": 1.900860538074753e-05, - "loss": 1.5847, - "step": 626 - }, - { - "epoch": 0.51, - "learning_rate": 1.9004809417119223e-05, - "loss": 1.483, - "step": 627 - }, - { - "epoch": 0.51, - "learning_rate": 1.9001006580672378e-05, - "loss": 1.6349, - "step": 628 - }, - { - "epoch": 0.51, - "learning_rate": 1.8997196874309472e-05, - "loss": 1.5402, - "step": 629 - }, - { - "epoch": 0.51, - "learning_rate": 1.899338030093822e-05, - "loss": 1.4086, - "step": 630 - }, - { - "epoch": 0.51, - "learning_rate": 1.898955686347157e-05, - "loss": 1.4702, - "step": 631 - }, - { - "epoch": 0.51, - "learning_rate": 1.898572656482773e-05, - "loss": 1.4065, - "step": 632 - }, - { - "epoch": 0.51, - "learning_rate": 1.8981889407930124e-05, - "loss": 1.6262, - "step": 633 - }, - { - "epoch": 0.51, - "learning_rate": 1.897804539570742e-05, - "loss": 1.51, - "step": 634 - }, - { - "epoch": 0.51, - "learning_rate": 1.8974194531093513e-05, - "loss": 1.5159, - "step": 635 - }, - { - "epoch": 0.51, - "learning_rate": 1.897033681702754e-05, - "loss": 1.6486, - "step": 636 - }, - { - "epoch": 0.52, - "learning_rate": 1.896647225645385e-05, - "loss": 1.5124, - "step": 637 - }, - { - "epoch": 0.52, - "learning_rate": 1.896260085232203e-05, - "loss": 1.6366, - "step": 638 - }, - { - "epoch": 0.52, - "learning_rate": 1.8958722607586883e-05, - "loss": 1.5849, - "step": 639 - }, - { - "epoch": 0.52, - "learning_rate": 1.8954837525208432e-05, - "loss": 1.4595, - "step": 640 - }, - { - "epoch": 0.52, - "learning_rate": 1.8950945608151928e-05, - "loss": 1.7082, - "step": 641 - }, - { - "epoch": 0.52, - "learning_rate": 1.894704685938783e-05, - "loss": 1.5452, - "step": 642 - }, - { - "epoch": 0.52, - "learning_rate": 1.8943141281891813e-05, - "loss": 1.5916, - "step": 643 - }, - { - "epoch": 0.52, - "learning_rate": 1.8939228878644766e-05, - "loss": 1.5577, - "step": 644 - }, - { - "epoch": 0.52, - "learning_rate": 1.8935309652632783e-05, - "loss": 1.5362, - "step": 645 - }, - { - "epoch": 0.52, - "learning_rate": 1.8931383606847173e-05, - "loss": 1.6325, - "step": 646 - }, - { - "epoch": 0.52, - "learning_rate": 1.8927450744284443e-05, - "loss": 1.4357, - "step": 647 - }, - { - "epoch": 0.52, - "learning_rate": 1.892351106794631e-05, - "loss": 1.5345, - "step": 648 - }, - { - "epoch": 0.53, - "learning_rate": 1.8919564580839683e-05, - "loss": 1.3827, - "step": 649 - }, - { - "epoch": 0.53, - "learning_rate": 1.8915611285976672e-05, - "loss": 1.5981, - "step": 650 - }, - { - "epoch": 0.53, - "learning_rate": 1.891165118637459e-05, - "loss": 1.5473, - "step": 651 - }, - { - "epoch": 0.53, - "learning_rate": 1.8907684285055933e-05, - "loss": 1.5762, - "step": 652 - }, - { - "epoch": 0.53, - "learning_rate": 1.8903710585048395e-05, - "loss": 1.5338, - "step": 653 - }, - { - "epoch": 0.53, - "learning_rate": 1.889973008938486e-05, - "loss": 1.6663, - "step": 654 - }, - { - "epoch": 0.53, - "learning_rate": 1.8895742801103393e-05, - "loss": 1.5324, - "step": 655 - }, - { - "epoch": 0.53, - "learning_rate": 1.8891748723247246e-05, - "loss": 1.4536, - "step": 656 - }, - { - "epoch": 0.53, - "learning_rate": 1.8887747858864856e-05, - "loss": 1.4978, - "step": 657 - }, - { - "epoch": 0.53, - "learning_rate": 1.888374021100984e-05, - "loss": 1.6171, - "step": 658 - }, - { - "epoch": 0.53, - "learning_rate": 1.887972578274098e-05, - "loss": 1.5503, - "step": 659 - }, - { - "epoch": 0.53, - "learning_rate": 1.887570457712225e-05, - "loss": 1.5991, - "step": 660 - }, - { - "epoch": 0.53, - "learning_rate": 1.887167659722279e-05, - "loss": 1.4478, - "step": 661 - }, - { - "epoch": 0.54, - "learning_rate": 1.886764184611691e-05, - "loss": 1.5581, - "step": 662 - }, - { - "epoch": 0.54, - "learning_rate": 1.8863600326884085e-05, - "loss": 1.6666, - "step": 663 - }, - { - "epoch": 0.54, - "learning_rate": 1.885955204260896e-05, - "loss": 1.4746, - "step": 664 - }, - { - "epoch": 0.54, - "learning_rate": 1.885549699638134e-05, - "loss": 1.5832, - "step": 665 - }, - { - "epoch": 0.54, - "learning_rate": 1.8851435191296197e-05, - "loss": 1.4203, - "step": 666 - }, - { - "epoch": 0.54, - "learning_rate": 1.8847366630453655e-05, - "loss": 1.4715, - "step": 667 - }, - { - "epoch": 0.54, - "learning_rate": 1.8843291316958995e-05, - "loss": 1.4083, - "step": 668 - }, - { - "epoch": 0.54, - "learning_rate": 1.8839209253922658e-05, - "loss": 1.5644, - "step": 669 - }, - { - "epoch": 0.54, - "learning_rate": 1.883512044446023e-05, - "loss": 1.4014, - "step": 670 - }, - { - "epoch": 0.54, - "learning_rate": 1.883102489169245e-05, - "loss": 1.546, - "step": 671 - }, - { - "epoch": 0.54, - "learning_rate": 1.88269225987452e-05, - "loss": 1.405, - "step": 672 - }, - { - "epoch": 0.54, - "learning_rate": 1.882281356874951e-05, - "loss": 1.422, - "step": 673 - }, - { - "epoch": 0.55, - "learning_rate": 1.8818697804841547e-05, - "loss": 1.4789, - "step": 674 - }, - { - "epoch": 0.55, - "learning_rate": 1.8814575310162628e-05, - "loss": 1.6684, - "step": 675 - }, - { - "epoch": 0.55, - "learning_rate": 1.8810446087859195e-05, - "loss": 1.4463, - "step": 676 - }, - { - "epoch": 0.55, - "learning_rate": 1.880631014108283e-05, - "loss": 1.5032, - "step": 677 - }, - { - "epoch": 0.55, - "learning_rate": 1.880216747299025e-05, - "loss": 1.4675, - "step": 678 - }, - { - "epoch": 0.55, - "learning_rate": 1.87980180867433e-05, - "loss": 1.6153, - "step": 679 - }, - { - "epoch": 0.55, - "learning_rate": 1.879386198550895e-05, - "loss": 1.5017, - "step": 680 - }, - { - "epoch": 0.55, - "learning_rate": 1.878969917245929e-05, - "loss": 1.5965, - "step": 681 - }, - { - "epoch": 0.55, - "learning_rate": 1.8785529650771552e-05, - "loss": 1.4735, - "step": 682 - }, - { - "epoch": 0.55, - "learning_rate": 1.878135342362807e-05, - "loss": 1.4951, - "step": 683 - }, - { - "epoch": 0.55, - "learning_rate": 1.8777170494216303e-05, - "loss": 1.4836, - "step": 684 - }, - { - "epoch": 0.55, - "learning_rate": 1.8772980865728824e-05, - "loss": 1.4588, - "step": 685 - }, - { - "epoch": 0.56, - "learning_rate": 1.8768784541363322e-05, - "loss": 1.6328, - "step": 686 - }, - { - "epoch": 0.56, - "learning_rate": 1.8764581524322588e-05, - "loss": 1.5416, - "step": 687 - }, - { - "epoch": 0.56, - "learning_rate": 1.8760371817814534e-05, - "loss": 1.5191, - "step": 688 - }, - { - "epoch": 0.56, - "learning_rate": 1.8756155425052163e-05, - "loss": 1.6145, - "step": 689 - }, - { - "epoch": 0.56, - "learning_rate": 1.8751932349253595e-05, - "loss": 1.4063, - "step": 690 - }, - { - "epoch": 0.56, - "learning_rate": 1.8747702593642044e-05, - "loss": 1.6896, - "step": 691 - }, - { - "epoch": 0.56, - "learning_rate": 1.8743466161445823e-05, - "loss": 1.6175, - "step": 692 - }, - { - "epoch": 0.56, - "learning_rate": 1.8739223055898338e-05, - "loss": 1.4995, - "step": 693 - }, - { - "epoch": 0.56, - "learning_rate": 1.873497328023809e-05, - "loss": 1.3836, - "step": 694 - }, - { - "epoch": 0.56, - "learning_rate": 1.873071683770868e-05, - "loss": 1.4563, - "step": 695 - }, - { - "epoch": 0.56, - "learning_rate": 1.872645373155878e-05, - "loss": 1.5516, - "step": 696 - }, - { - "epoch": 0.56, - "learning_rate": 1.8722183965042162e-05, - "loss": 1.4189, - "step": 697 - }, - { - "epoch": 0.56, - "learning_rate": 1.8717907541417678e-05, - "loss": 1.5791, - "step": 698 - }, - { - "epoch": 0.57, - "learning_rate": 1.8713624463949255e-05, - "loss": 1.517, - "step": 699 - }, - { - "epoch": 0.57, - "learning_rate": 1.8709334735905908e-05, - "loss": 1.4743, - "step": 700 - }, - { - "epoch": 0.57, - "learning_rate": 1.8705038360561724e-05, - "loss": 1.4619, - "step": 701 - }, - { - "epoch": 0.57, - "learning_rate": 1.8700735341195857e-05, - "loss": 1.4308, - "step": 702 - }, - { - "epoch": 0.57, - "learning_rate": 1.869642568109254e-05, - "loss": 1.5176, - "step": 703 - }, - { - "epoch": 0.57, - "learning_rate": 1.8692109383541074e-05, - "loss": 1.3905, - "step": 704 - }, - { - "epoch": 0.57, - "learning_rate": 1.8687786451835827e-05, - "loss": 1.5235, - "step": 705 - }, - { - "epoch": 0.57, - "learning_rate": 1.8683456889276216e-05, - "loss": 1.5915, - "step": 706 - }, - { - "epoch": 0.57, - "learning_rate": 1.8679120699166745e-05, - "loss": 1.4823, - "step": 707 - }, - { - "epoch": 0.57, - "learning_rate": 1.8674777884816952e-05, - "loss": 1.5024, - "step": 708 - }, - { - "epoch": 0.57, - "learning_rate": 1.8670428449541442e-05, - "loss": 1.2594, - "step": 709 - }, - { - "epoch": 0.57, - "learning_rate": 1.866607239665988e-05, - "loss": 1.4048, - "step": 710 - }, - { - "epoch": 0.58, - "learning_rate": 1.866170972949696e-05, - "loss": 1.424, - "step": 711 - }, - { - "epoch": 0.58, - "learning_rate": 1.865734045138245e-05, - "loss": 1.5463, - "step": 712 - }, - { - "epoch": 0.58, - "learning_rate": 1.8652964565651147e-05, - "loss": 1.5443, - "step": 713 - }, - { - "epoch": 0.58, - "learning_rate": 1.8648582075642892e-05, - "loss": 1.4677, - "step": 714 - }, - { - "epoch": 0.58, - "learning_rate": 1.8644192984702583e-05, - "loss": 1.5592, - "step": 715 - }, - { - "epoch": 0.58, - "learning_rate": 1.8639797296180133e-05, - "loss": 1.6481, - "step": 716 - }, - { - "epoch": 0.58, - "learning_rate": 1.8635395013430508e-05, - "loss": 1.4641, - "step": 717 - }, - { - "epoch": 0.58, - "learning_rate": 1.863098613981369e-05, - "loss": 1.5463, - "step": 718 - }, - { - "epoch": 0.58, - "learning_rate": 1.8626570678694714e-05, - "loss": 1.5762, - "step": 719 - }, - { - "epoch": 0.58, - "learning_rate": 1.8622148633443626e-05, - "loss": 1.744, - "step": 720 - }, - { - "epoch": 0.58, - "learning_rate": 1.8617720007435497e-05, - "loss": 1.4325, - "step": 721 - }, - { - "epoch": 0.58, - "learning_rate": 1.8613284804050433e-05, - "loss": 1.6106, - "step": 722 - }, - { - "epoch": 0.58, - "learning_rate": 1.860884302667355e-05, - "loss": 1.5402, - "step": 723 - }, - { - "epoch": 0.59, - "learning_rate": 1.860439467869498e-05, - "loss": 1.6977, - "step": 724 - }, - { - "epoch": 0.59, - "learning_rate": 1.859993976350988e-05, - "loss": 1.3925, - "step": 725 - }, - { - "epoch": 0.59, - "learning_rate": 1.8595478284518413e-05, - "loss": 1.5088, - "step": 726 - }, - { - "epoch": 0.59, - "learning_rate": 1.8591010245125744e-05, - "loss": 1.4322, - "step": 727 - }, - { - "epoch": 0.59, - "learning_rate": 1.8586535648742065e-05, - "loss": 1.694, - "step": 728 - }, - { - "epoch": 0.59, - "learning_rate": 1.858205449878255e-05, - "loss": 1.6109, - "step": 729 - }, - { - "epoch": 0.59, - "learning_rate": 1.8577566798667397e-05, - "loss": 1.6199, - "step": 730 - }, - { - "epoch": 0.59, - "learning_rate": 1.8573072551821782e-05, - "loss": 1.4657, - "step": 731 - }, - { - "epoch": 0.59, - "learning_rate": 1.8568571761675893e-05, - "loss": 1.6134, - "step": 732 - }, - { - "epoch": 0.59, - "learning_rate": 1.856406443166491e-05, - "loss": 1.7281, - "step": 733 - }, - { - "epoch": 0.59, - "learning_rate": 1.8559550565228993e-05, - "loss": 1.5124, - "step": 734 - }, - { - "epoch": 0.59, - "learning_rate": 1.8555030165813303e-05, - "loss": 1.6102, - "step": 735 - }, - { - "epoch": 0.6, - "learning_rate": 1.8550503236867986e-05, - "loss": 1.6052, - "step": 736 - }, - { - "epoch": 0.6, - "learning_rate": 1.8545969781848172e-05, - "loss": 1.3372, - "step": 737 - }, - { - "epoch": 0.6, - "learning_rate": 1.854142980421396e-05, - "loss": 1.4238, - "step": 738 - }, - { - "epoch": 0.6, - "learning_rate": 1.853688330743044e-05, - "loss": 1.551, - "step": 739 - }, - { - "epoch": 0.6, - "learning_rate": 1.8532330294967678e-05, - "loss": 1.5199, - "step": 740 - }, - { - "epoch": 0.6, - "learning_rate": 1.8527770770300707e-05, - "loss": 1.5015, - "step": 741 - }, - { - "epoch": 0.6, - "learning_rate": 1.8523204736909526e-05, - "loss": 1.6076, - "step": 742 - }, - { - "epoch": 0.6, - "learning_rate": 1.8518632198279117e-05, - "loss": 1.4446, - "step": 743 - }, - { - "epoch": 0.6, - "learning_rate": 1.8514053157899417e-05, - "loss": 1.6826, - "step": 744 - }, - { - "epoch": 0.6, - "learning_rate": 1.8509467619265323e-05, - "loss": 1.6, - "step": 745 - }, - { - "epoch": 0.6, - "learning_rate": 1.8504875585876696e-05, - "loss": 1.5509, - "step": 746 - }, - { - "epoch": 0.6, - "learning_rate": 1.8500277061238354e-05, - "loss": 1.517, - "step": 747 - }, - { - "epoch": 0.61, - "learning_rate": 1.849567204886007e-05, - "loss": 1.6361, - "step": 748 - }, - { - "epoch": 0.61, - "learning_rate": 1.8491060552256565e-05, - "loss": 1.5211, - "step": 749 - }, - { - "epoch": 0.61, - "learning_rate": 1.848644257494751e-05, - "loss": 1.2962, - "step": 750 - }, - { - "epoch": 0.61, - "learning_rate": 1.848181812045753e-05, - "loss": 1.4599, - "step": 751 - }, - { - "epoch": 0.61, - "learning_rate": 1.8477187192316182e-05, - "loss": 1.4868, - "step": 752 - }, - { - "epoch": 0.61, - "learning_rate": 1.8472549794057972e-05, - "loss": 1.2506, - "step": 753 - }, - { - "epoch": 0.61, - "learning_rate": 1.846790592922234e-05, - "loss": 1.6276, - "step": 754 - }, - { - "epoch": 0.61, - "learning_rate": 1.8463255601353663e-05, - "loss": 1.4601, - "step": 755 - }, - { - "epoch": 0.61, - "learning_rate": 1.8458598814001255e-05, - "loss": 1.5172, - "step": 756 - }, - { - "epoch": 0.61, - "learning_rate": 1.8453935570719347e-05, - "loss": 1.3831, - "step": 757 - }, - { - "epoch": 0.61, - "learning_rate": 1.8449265875067116e-05, - "loss": 1.494, - "step": 758 - }, - { - "epoch": 0.61, - "learning_rate": 1.8444589730608648e-05, - "loss": 1.5422, - "step": 759 - }, - { - "epoch": 0.61, - "learning_rate": 1.8439907140912962e-05, - "loss": 1.4577, - "step": 760 - }, - { - "epoch": 0.62, - "learning_rate": 1.843521810955399e-05, - "loss": 1.6428, - "step": 761 - }, - { - "epoch": 0.62, - "learning_rate": 1.8430522640110577e-05, - "loss": 1.4929, - "step": 762 - }, - { - "epoch": 0.62, - "learning_rate": 1.8425820736166492e-05, - "loss": 1.5278, - "step": 763 - }, - { - "epoch": 0.62, - "learning_rate": 1.842111240131041e-05, - "loss": 1.5503, - "step": 764 - }, - { - "epoch": 0.62, - "learning_rate": 1.841639763913591e-05, - "loss": 1.4129, - "step": 765 - }, - { - "epoch": 0.62, - "learning_rate": 1.8411676453241485e-05, - "loss": 1.5804, - "step": 766 - }, - { - "epoch": 0.62, - "learning_rate": 1.8406948847230522e-05, - "loss": 1.478, - "step": 767 - }, - { - "epoch": 0.62, - "learning_rate": 1.8402214824711318e-05, - "loss": 1.2323, - "step": 768 - }, - { - "epoch": 0.62, - "learning_rate": 1.839747438929706e-05, - "loss": 1.4618, - "step": 769 - }, - { - "epoch": 0.62, - "learning_rate": 1.839272754460583e-05, - "loss": 1.5452, - "step": 770 - }, - { - "epoch": 0.62, - "learning_rate": 1.83879742942606e-05, - "loss": 1.5762, - "step": 771 - }, - { - "epoch": 0.62, - "learning_rate": 1.8383214641889244e-05, - "loss": 1.5168, - "step": 772 - }, - { - "epoch": 0.63, - "learning_rate": 1.8378448591124505e-05, - "loss": 1.5598, - "step": 773 - }, - { - "epoch": 0.63, - "learning_rate": 1.837367614560402e-05, - "loss": 1.3882, - "step": 774 - }, - { - "epoch": 0.63, - "learning_rate": 1.8368897308970302e-05, - "loss": 1.6454, - "step": 775 - }, - { - "epoch": 0.63, - "learning_rate": 1.8364112084870745e-05, - "loss": 1.5565, - "step": 776 - }, - { - "epoch": 0.63, - "learning_rate": 1.8359320476957616e-05, - "loss": 1.5322, - "step": 777 - }, - { - "epoch": 0.63, - "learning_rate": 1.8354522488888056e-05, - "loss": 1.7092, - "step": 778 - }, - { - "epoch": 0.63, - "learning_rate": 1.8349718124324075e-05, - "loss": 1.5669, - "step": 779 - }, - { - "epoch": 0.63, - "learning_rate": 1.8344907386932552e-05, - "loss": 1.6404, - "step": 780 - }, - { - "epoch": 0.63, - "learning_rate": 1.8340090280385222e-05, - "loss": 1.4757, - "step": 781 - }, - { - "epoch": 0.63, - "learning_rate": 1.833526680835869e-05, - "loss": 1.5354, - "step": 782 - }, - { - "epoch": 0.63, - "learning_rate": 1.8330436974534423e-05, - "loss": 1.4585, - "step": 783 - }, - { - "epoch": 0.63, - "learning_rate": 1.8325600782598724e-05, - "loss": 1.4592, - "step": 784 - }, - { - "epoch": 0.64, - "learning_rate": 1.8320758236242772e-05, - "loss": 1.3812, - "step": 785 - }, - { - "epoch": 0.64, - "learning_rate": 1.8315909339162576e-05, - "loss": 1.5175, - "step": 786 - }, - { - "epoch": 0.64, - "learning_rate": 1.831105409505901e-05, - "loss": 1.6107, - "step": 787 - }, - { - "epoch": 0.64, - "learning_rate": 1.830619250763778e-05, - "loss": 1.5187, - "step": 788 - }, - { - "epoch": 0.64, - "learning_rate": 1.830132458060944e-05, - "loss": 1.4704, - "step": 789 - }, - { - "epoch": 0.64, - "learning_rate": 1.8296450317689377e-05, - "loss": 1.4188, - "step": 790 - }, - { - "epoch": 0.64, - "learning_rate": 1.829156972259782e-05, - "loss": 1.5359, - "step": 791 - }, - { - "epoch": 0.64, - "learning_rate": 1.8286682799059827e-05, - "loss": 1.5452, - "step": 792 - }, - { - "epoch": 0.64, - "learning_rate": 1.8281789550805283e-05, - "loss": 1.5036, - "step": 793 - }, - { - "epoch": 0.64, - "learning_rate": 1.827688998156891e-05, - "loss": 1.5094, - "step": 794 - }, - { - "epoch": 0.64, - "learning_rate": 1.827198409509024e-05, - "loss": 1.478, - "step": 795 - }, - { - "epoch": 0.64, - "learning_rate": 1.8267071895113645e-05, - "loss": 1.5969, - "step": 796 - }, - { - "epoch": 0.64, - "learning_rate": 1.8262153385388303e-05, - "loss": 1.5304, - "step": 797 - }, - { - "epoch": 0.65, - "learning_rate": 1.8257228569668206e-05, - "loss": 1.6215, - "step": 798 - }, - { - "epoch": 0.65, - "learning_rate": 1.8252297451712167e-05, - "loss": 1.4482, - "step": 799 - }, - { - "epoch": 0.65, - "learning_rate": 1.824736003528381e-05, - "loss": 1.6214, - "step": 800 - }, - { - "epoch": 0.65, - "learning_rate": 1.8242416324151554e-05, - "loss": 1.3365, - "step": 801 - }, - { - "epoch": 0.65, - "learning_rate": 1.8237466322088635e-05, - "loss": 1.5775, - "step": 802 - }, - { - "epoch": 0.65, - "learning_rate": 1.8232510032873087e-05, - "loss": 1.6282, - "step": 803 - }, - { - "epoch": 0.65, - "learning_rate": 1.822754746028774e-05, - "loss": 1.5358, - "step": 804 - }, - { - "epoch": 0.65, - "learning_rate": 1.8222578608120217e-05, - "loss": 1.5782, - "step": 805 - }, - { - "epoch": 0.65, - "learning_rate": 1.8217603480162942e-05, - "loss": 1.484, - "step": 806 - }, - { - "epoch": 0.65, - "learning_rate": 1.8212622080213128e-05, - "loss": 1.5165, - "step": 807 - }, - { - "epoch": 0.65, - "learning_rate": 1.8207634412072765e-05, - "loss": 1.5655, - "step": 808 - }, - { - "epoch": 0.65, - "learning_rate": 1.820264047954864e-05, - "loss": 1.485, - "step": 809 - }, - { - "epoch": 0.66, - "learning_rate": 1.8197640286452312e-05, - "loss": 1.3877, - "step": 810 - }, - { - "epoch": 0.66, - "learning_rate": 1.8192633836600123e-05, - "loss": 1.5959, - "step": 811 - }, - { - "epoch": 0.66, - "learning_rate": 1.8187621133813187e-05, - "loss": 1.4381, - "step": 812 - }, - { - "epoch": 0.66, - "learning_rate": 1.8182602181917395e-05, - "loss": 1.4785, - "step": 813 - }, - { - "epoch": 0.66, - "learning_rate": 1.8177576984743406e-05, - "loss": 1.5112, - "step": 814 - }, - { - "epoch": 0.66, - "learning_rate": 1.817254554612664e-05, - "loss": 1.5413, - "step": 815 - }, - { - "epoch": 0.66, - "learning_rate": 1.816750786990729e-05, - "loss": 1.4776, - "step": 816 - }, - { - "epoch": 0.66, - "learning_rate": 1.8162463959930303e-05, - "loss": 1.5384, - "step": 817 - }, - { - "epoch": 0.66, - "learning_rate": 1.8157413820045385e-05, - "loss": 1.5127, - "step": 818 - }, - { - "epoch": 0.66, - "learning_rate": 1.8152357454107002e-05, - "loss": 1.47, - "step": 819 - }, - { - "epoch": 0.66, - "learning_rate": 1.814729486597436e-05, - "loss": 1.572, - "step": 820 - }, - { - "epoch": 0.66, - "learning_rate": 1.8142226059511432e-05, - "loss": 1.6226, - "step": 821 - }, - { - "epoch": 0.67, - "learning_rate": 1.813715103858692e-05, - "loss": 1.4954, - "step": 822 - }, - { - "epoch": 0.67, - "learning_rate": 1.813206980707428e-05, - "loss": 1.6259, - "step": 823 - }, - { - "epoch": 0.67, - "learning_rate": 1.81269823688517e-05, - "loss": 1.5969, - "step": 824 - }, - { - "epoch": 0.67, - "learning_rate": 1.8121888727802113e-05, - "loss": 1.4289, - "step": 825 - }, - { - "epoch": 0.67, - "learning_rate": 1.8116788887813182e-05, - "loss": 1.4716, - "step": 826 - }, - { - "epoch": 0.67, - "learning_rate": 1.81116828527773e-05, - "loss": 1.6285, - "step": 827 - }, - { - "epoch": 0.67, - "learning_rate": 1.8106570626591594e-05, - "loss": 1.2992, - "step": 828 - }, - { - "epoch": 0.67, - "learning_rate": 1.8101452213157906e-05, - "loss": 1.608, - "step": 829 - }, - { - "epoch": 0.67, - "learning_rate": 1.8096327616382815e-05, - "loss": 1.6087, - "step": 830 - }, - { - "epoch": 0.67, - "learning_rate": 1.8091196840177608e-05, - "loss": 1.4181, - "step": 831 - }, - { - "epoch": 0.67, - "learning_rate": 1.8086059888458287e-05, - "loss": 1.6262, - "step": 832 - }, - { - "epoch": 0.67, - "learning_rate": 1.808091676514558e-05, - "loss": 1.5991, - "step": 833 - }, - { - "epoch": 0.67, - "learning_rate": 1.807576747416491e-05, - "loss": 1.5068, - "step": 834 - }, - { - "epoch": 0.68, - "learning_rate": 1.8070612019446422e-05, - "loss": 1.5457, - "step": 835 - }, - { - "epoch": 0.68, - "learning_rate": 1.8065450404924956e-05, - "loss": 1.5433, - "step": 836 - }, - { - "epoch": 0.68, - "learning_rate": 1.8060282634540053e-05, - "loss": 1.5104, - "step": 837 - }, - { - "epoch": 0.68, - "learning_rate": 1.8055108712235956e-05, - "loss": 1.5509, - "step": 838 - }, - { - "epoch": 0.68, - "learning_rate": 1.8049928641961603e-05, - "loss": 1.4747, - "step": 839 - }, - { - "epoch": 0.68, - "learning_rate": 1.8044742427670627e-05, - "loss": 1.5867, - "step": 840 - }, - { - "epoch": 0.68, - "learning_rate": 1.803955007332134e-05, - "loss": 1.5129, - "step": 841 - }, - { - "epoch": 0.68, - "learning_rate": 1.803435158287675e-05, - "loss": 1.6435, - "step": 842 - }, - { - "epoch": 0.68, - "learning_rate": 1.802914696030455e-05, - "loss": 1.5901, - "step": 843 - }, - { - "epoch": 0.68, - "learning_rate": 1.8023936209577107e-05, - "loss": 1.4455, - "step": 844 - }, - { - "epoch": 0.68, - "learning_rate": 1.8018719334671465e-05, - "loss": 1.5525, - "step": 845 - }, - { - "epoch": 0.68, - "learning_rate": 1.801349633956935e-05, - "loss": 1.3374, - "step": 846 - }, - { - "epoch": 0.69, - "learning_rate": 1.800826722825715e-05, - "loss": 1.5279, - "step": 847 - }, - { - "epoch": 0.69, - "learning_rate": 1.800303200472592e-05, - "loss": 1.5674, - "step": 848 - }, - { - "epoch": 0.69, - "learning_rate": 1.7997790672971392e-05, - "loss": 1.6139, - "step": 849 - }, - { - "epoch": 0.69, - "learning_rate": 1.7992543236993952e-05, - "loss": 1.5659, - "step": 850 - }, - { - "epoch": 0.69, - "learning_rate": 1.798728970079864e-05, - "loss": 1.625, - "step": 851 - }, - { - "epoch": 0.69, - "learning_rate": 1.798203006839517e-05, - "loss": 1.5035, - "step": 852 - }, - { - "epoch": 0.69, - "learning_rate": 1.7976764343797885e-05, - "loss": 1.5179, - "step": 853 - }, - { - "epoch": 0.69, - "learning_rate": 1.7971492531025792e-05, - "loss": 1.7088, - "step": 854 - }, - { - "epoch": 0.69, - "learning_rate": 1.7966214634102545e-05, - "loss": 1.5373, - "step": 855 - }, - { - "epoch": 0.69, - "learning_rate": 1.796093065705644e-05, - "loss": 1.5922, - "step": 856 - }, - { - "epoch": 0.69, - "learning_rate": 1.7955640603920404e-05, - "loss": 1.3572, - "step": 857 - }, - { - "epoch": 0.69, - "learning_rate": 1.7950344478732024e-05, - "loss": 1.569, - "step": 858 - }, - { - "epoch": 0.69, - "learning_rate": 1.7945042285533495e-05, - "loss": 1.4733, - "step": 859 - }, - { - "epoch": 0.7, - "learning_rate": 1.7939734028371663e-05, - "loss": 1.5448, - "step": 860 - }, - { - "epoch": 0.7, - "learning_rate": 1.7934419711297992e-05, - "loss": 1.3813, - "step": 861 - }, - { - "epoch": 0.7, - "learning_rate": 1.7929099338368573e-05, - "loss": 1.5062, - "step": 862 - }, - { - "epoch": 0.7, - "learning_rate": 1.7923772913644123e-05, - "loss": 1.5383, - "step": 863 - }, - { - "epoch": 0.7, - "learning_rate": 1.7918440441189976e-05, - "loss": 1.5439, - "step": 864 - }, - { - "epoch": 0.7, - "learning_rate": 1.7913101925076075e-05, - "loss": 1.4657, - "step": 865 - }, - { - "epoch": 0.7, - "learning_rate": 1.7907757369376984e-05, - "loss": 1.6374, - "step": 866 - }, - { - "epoch": 0.7, - "learning_rate": 1.7902406778171882e-05, - "loss": 1.4602, - "step": 867 - }, - { - "epoch": 0.7, - "learning_rate": 1.789705015554454e-05, - "loss": 1.4257, - "step": 868 - }, - { - "epoch": 0.7, - "learning_rate": 1.789168750558334e-05, - "loss": 1.5034, - "step": 869 - }, - { - "epoch": 0.7, - "learning_rate": 1.7886318832381264e-05, - "loss": 1.5895, - "step": 870 - }, - { - "epoch": 0.7, - "learning_rate": 1.788094414003589e-05, - "loss": 1.4696, - "step": 871 - }, - { - "epoch": 0.71, - "learning_rate": 1.7875563432649394e-05, - "loss": 1.6711, - "step": 872 - }, - { - "epoch": 0.71, - "learning_rate": 1.7870176714328537e-05, - "loss": 1.515, - "step": 873 - }, - { - "epoch": 0.71, - "learning_rate": 1.786478398918467e-05, - "loss": 1.4375, - "step": 874 - }, - { - "epoch": 0.71, - "learning_rate": 1.7859385261333736e-05, - "loss": 1.4576, - "step": 875 - }, - { - "epoch": 0.71, - "learning_rate": 1.7853980534896242e-05, - "loss": 1.5451, - "step": 876 - }, - { - "epoch": 0.71, - "learning_rate": 1.7848569813997293e-05, - "loss": 1.3663, - "step": 877 - }, - { - "epoch": 0.71, - "learning_rate": 1.7843153102766554e-05, - "loss": 1.596, - "step": 878 - }, - { - "epoch": 0.71, - "learning_rate": 1.783773040533827e-05, - "loss": 1.6045, - "step": 879 - }, - { - "epoch": 0.71, - "learning_rate": 1.783230172585126e-05, - "loss": 1.6328, - "step": 880 - }, - { - "epoch": 0.71, - "learning_rate": 1.7826867068448895e-05, - "loss": 1.4852, - "step": 881 - }, - { - "epoch": 0.71, - "learning_rate": 1.7821426437279112e-05, - "loss": 1.4556, - "step": 882 - }, - { - "epoch": 0.71, - "learning_rate": 1.781597983649442e-05, - "loss": 1.5214, - "step": 883 - }, - { - "epoch": 0.72, - "learning_rate": 1.7810527270251868e-05, - "loss": 1.5582, - "step": 884 - }, - { - "epoch": 0.72, - "learning_rate": 1.7805068742713073e-05, - "loss": 1.5739, - "step": 885 - }, - { - "epoch": 0.72, - "learning_rate": 1.7799604258044185e-05, - "loss": 1.5095, - "step": 886 - }, - { - "epoch": 0.72, - "learning_rate": 1.7794133820415916e-05, - "loss": 1.4503, - "step": 887 - }, - { - "epoch": 0.72, - "learning_rate": 1.7788657434003514e-05, - "loss": 1.5505, - "step": 888 - }, - { - "epoch": 0.72, - "learning_rate": 1.778317510298677e-05, - "loss": 1.4311, - "step": 889 - }, - { - "epoch": 0.72, - "learning_rate": 1.7777686831550008e-05, - "loss": 1.5378, - "step": 890 - }, - { - "epoch": 0.72, - "learning_rate": 1.777219262388209e-05, - "loss": 1.7353, - "step": 891 - }, - { - "epoch": 0.72, - "learning_rate": 1.7766692484176408e-05, - "loss": 1.5913, - "step": 892 - }, - { - "epoch": 0.72, - "learning_rate": 1.7761186416630882e-05, - "loss": 1.4492, - "step": 893 - }, - { - "epoch": 0.72, - "learning_rate": 1.775567442544795e-05, - "loss": 1.5626, - "step": 894 - }, - { - "epoch": 0.72, - "learning_rate": 1.775015651483459e-05, - "loss": 1.3199, - "step": 895 - }, - { - "epoch": 0.72, - "learning_rate": 1.774463268900227e-05, - "loss": 1.5394, - "step": 896 - }, - { - "epoch": 0.73, - "learning_rate": 1.7739102952166995e-05, - "loss": 1.49, - "step": 897 - }, - { - "epoch": 0.73, - "learning_rate": 1.773356730854927e-05, - "loss": 1.5154, - "step": 898 - }, - { - "epoch": 0.73, - "learning_rate": 1.7728025762374117e-05, - "loss": 1.5566, - "step": 899 - }, - { - "epoch": 0.73, - "learning_rate": 1.7722478317871053e-05, - "loss": 1.438, - "step": 900 - }, - { - "epoch": 0.73, - "learning_rate": 1.7716924979274105e-05, - "loss": 1.6369, - "step": 901 - }, - { - "epoch": 0.73, - "learning_rate": 1.7711365750821797e-05, - "loss": 1.5315, - "step": 902 - }, - { - "epoch": 0.73, - "learning_rate": 1.7705800636757146e-05, - "loss": 1.4361, - "step": 903 - }, - { - "epoch": 0.73, - "learning_rate": 1.7700229641327655e-05, - "loss": 1.4099, - "step": 904 - }, - { - "epoch": 0.73, - "learning_rate": 1.7694652768785335e-05, - "loss": 1.563, - "step": 905 - }, - { - "epoch": 0.73, - "learning_rate": 1.7689070023386662e-05, - "loss": 1.4415, - "step": 906 - }, - { - "epoch": 0.73, - "learning_rate": 1.7683481409392605e-05, - "loss": 1.5086, - "step": 907 - }, - { - "epoch": 0.73, - "learning_rate": 1.7677886931068615e-05, - "loss": 1.3464, - "step": 908 - }, - { - "epoch": 0.74, - "learning_rate": 1.7672286592684605e-05, - "loss": 1.6203, - "step": 909 - }, - { - "epoch": 0.74, - "learning_rate": 1.7666680398514978e-05, - "loss": 1.5396, - "step": 910 - }, - { - "epoch": 0.74, - "learning_rate": 1.7661068352838593e-05, - "loss": 1.3804, - "step": 911 - }, - { - "epoch": 0.74, - "learning_rate": 1.7655450459938786e-05, - "loss": 1.4931, - "step": 912 - }, - { - "epoch": 0.74, - "learning_rate": 1.764982672410334e-05, - "loss": 1.4645, - "step": 913 - }, - { - "epoch": 0.74, - "learning_rate": 1.764419714962452e-05, - "loss": 1.3662, - "step": 914 - }, - { - "epoch": 0.74, - "learning_rate": 1.7638561740799025e-05, - "loss": 1.5265, - "step": 915 - }, - { - "epoch": 0.74, - "learning_rate": 1.7632920501928014e-05, - "loss": 1.3809, - "step": 916 - }, - { - "epoch": 0.74, - "learning_rate": 1.7627273437317113e-05, - "loss": 1.6059, - "step": 917 - }, - { - "epoch": 0.74, - "learning_rate": 1.7621620551276366e-05, - "loss": 1.518, - "step": 918 - }, - { - "epoch": 0.74, - "learning_rate": 1.7615961848120282e-05, - "loss": 1.5075, - "step": 919 - }, - { - "epoch": 0.74, - "learning_rate": 1.76102973321678e-05, - "loss": 1.6026, - "step": 920 - }, - { - "epoch": 0.75, - "learning_rate": 1.7604627007742296e-05, - "loss": 1.4265, - "step": 921 - }, - { - "epoch": 0.75, - "learning_rate": 1.759895087917158e-05, - "loss": 1.5754, - "step": 922 - }, - { - "epoch": 0.75, - "learning_rate": 1.7593268950787896e-05, - "loss": 1.4336, - "step": 923 - }, - { - "epoch": 0.75, - "learning_rate": 1.758758122692791e-05, - "loss": 1.4706, - "step": 924 - }, - { - "epoch": 0.75, - "learning_rate": 1.7581887711932713e-05, - "loss": 1.5535, - "step": 925 - }, - { - "epoch": 0.75, - "learning_rate": 1.7576188410147814e-05, - "loss": 1.3634, - "step": 926 - }, - { - "epoch": 0.75, - "learning_rate": 1.757048332592314e-05, - "loss": 1.5515, - "step": 927 - }, - { - "epoch": 0.75, - "learning_rate": 1.7564772463613036e-05, - "loss": 1.5909, - "step": 928 - }, - { - "epoch": 0.75, - "learning_rate": 1.7559055827576248e-05, - "loss": 1.6042, - "step": 929 - }, - { - "epoch": 0.75, - "learning_rate": 1.7553333422175933e-05, - "loss": 1.541, - "step": 930 - }, - { - "epoch": 0.75, - "learning_rate": 1.7547605251779657e-05, - "loss": 1.4997, - "step": 931 - }, - { - "epoch": 0.75, - "learning_rate": 1.754187132075938e-05, - "loss": 1.6326, - "step": 932 - }, - { - "epoch": 0.75, - "learning_rate": 1.7536131633491457e-05, - "loss": 1.4602, - "step": 933 - }, - { - "epoch": 0.76, - "learning_rate": 1.753038619435664e-05, - "loss": 1.5302, - "step": 934 - }, - { - "epoch": 0.76, - "learning_rate": 1.752463500774007e-05, - "loss": 1.4208, - "step": 935 - }, - { - "epoch": 0.76, - "learning_rate": 1.7518878078031276e-05, - "loss": 1.462, - "step": 936 - }, - { - "epoch": 0.76, - "learning_rate": 1.7513115409624165e-05, - "loss": 1.4025, - "step": 937 - }, - { - "epoch": 0.76, - "learning_rate": 1.7507347006917037e-05, - "loss": 1.4377, - "step": 938 - }, - { - "epoch": 0.76, - "learning_rate": 1.750157287431255e-05, - "loss": 1.6667, - "step": 939 - }, - { - "epoch": 0.76, - "learning_rate": 1.7495793016217754e-05, - "loss": 1.4998, - "step": 940 - }, - { - "epoch": 0.76, - "learning_rate": 1.7490007437044053e-05, - "loss": 1.3958, - "step": 941 - }, - { - "epoch": 0.76, - "learning_rate": 1.7484216141207228e-05, - "loss": 1.4415, - "step": 942 - }, - { - "epoch": 0.76, - "learning_rate": 1.7478419133127423e-05, - "loss": 1.5889, - "step": 943 - }, - { - "epoch": 0.76, - "learning_rate": 1.747261641722913e-05, - "loss": 1.4574, - "step": 944 - }, - { - "epoch": 0.76, - "learning_rate": 1.7466807997941214e-05, - "loss": 1.3469, - "step": 945 - }, - { - "epoch": 0.77, - "learning_rate": 1.7460993879696877e-05, - "loss": 1.6043, - "step": 946 - }, - { - "epoch": 0.77, - "learning_rate": 1.745517406693369e-05, - "loss": 1.492, - "step": 947 - }, - { - "epoch": 0.77, - "learning_rate": 1.7449348564093545e-05, - "loss": 1.531, - "step": 948 - }, - { - "epoch": 0.77, - "learning_rate": 1.7443517375622706e-05, - "loss": 1.5882, - "step": 949 - }, - { - "epoch": 0.77, - "learning_rate": 1.743768050597175e-05, - "loss": 1.7165, - "step": 950 - }, - { - "epoch": 0.77, - "learning_rate": 1.7431837959595605e-05, - "loss": 1.3356, - "step": 951 - }, - { - "epoch": 0.77, - "learning_rate": 1.742598974095353e-05, - "loss": 1.4238, - "step": 952 - }, - { - "epoch": 0.77, - "learning_rate": 1.742013585450911e-05, - "loss": 1.3477, - "step": 953 - }, - { - "epoch": 0.77, - "learning_rate": 1.741427630473026e-05, - "loss": 1.5347, - "step": 954 - }, - { - "epoch": 0.77, - "learning_rate": 1.7408411096089207e-05, - "loss": 1.542, - "step": 955 - }, - { - "epoch": 0.77, - "learning_rate": 1.740254023306251e-05, - "loss": 1.4162, - "step": 956 - }, - { - "epoch": 0.77, - "learning_rate": 1.739666372013104e-05, - "loss": 1.4142, - "step": 957 - }, - { - "epoch": 0.78, - "learning_rate": 1.739078156177998e-05, - "loss": 1.7346, - "step": 958 - }, - { - "epoch": 0.78, - "learning_rate": 1.7384893762498816e-05, - "loss": 1.6337, - "step": 959 - }, - { - "epoch": 0.78, - "learning_rate": 1.7379000326781348e-05, - "loss": 1.5078, - "step": 960 - }, - { - "epoch": 0.78, - "learning_rate": 1.7373101259125672e-05, - "loss": 1.508, - "step": 961 - }, - { - "epoch": 0.78, - "learning_rate": 1.736719656403418e-05, - "loss": 1.4282, - "step": 962 - }, - { - "epoch": 0.78, - "learning_rate": 1.7361286246013568e-05, - "loss": 1.5222, - "step": 963 - }, - { - "epoch": 0.78, - "learning_rate": 1.7355370309574822e-05, - "loss": 1.403, - "step": 964 - }, - { - "epoch": 0.78, - "learning_rate": 1.734944875923321e-05, - "loss": 1.5627, - "step": 965 - }, - { - "epoch": 0.78, - "learning_rate": 1.7343521599508285e-05, - "loss": 1.6168, - "step": 966 - }, - { - "epoch": 0.78, - "learning_rate": 1.733758883492389e-05, - "loss": 1.3577, - "step": 967 - }, - { - "epoch": 0.78, - "learning_rate": 1.733165047000814e-05, - "loss": 1.4574, - "step": 968 - }, - { - "epoch": 0.78, - "learning_rate": 1.7325706509293422e-05, - "loss": 1.6162, - "step": 969 - }, - { - "epoch": 0.78, - "learning_rate": 1.7319756957316392e-05, - "loss": 1.6909, - "step": 970 - }, - { - "epoch": 0.79, - "learning_rate": 1.7313801818617983e-05, - "loss": 1.4009, - "step": 971 - }, - { - "epoch": 0.79, - "learning_rate": 1.730784109774339e-05, - "loss": 1.4353, - "step": 972 - }, - { - "epoch": 0.79, - "learning_rate": 1.730187479924206e-05, - "loss": 1.4807, - "step": 973 - }, - { - "epoch": 0.79, - "learning_rate": 1.729590292766771e-05, - "loss": 1.4043, - "step": 974 - }, - { - "epoch": 0.79, - "learning_rate": 1.7289925487578294e-05, - "loss": 1.4513, - "step": 975 - }, - { - "epoch": 0.79, - "learning_rate": 1.728394248353603e-05, - "loss": 1.319, - "step": 976 - }, - { - "epoch": 0.79, - "learning_rate": 1.727795392010738e-05, - "loss": 1.4372, - "step": 977 - }, - { - "epoch": 0.79, - "learning_rate": 1.727195980186304e-05, - "loss": 1.3204, - "step": 978 - }, - { - "epoch": 0.79, - "learning_rate": 1.7265960133377956e-05, - "loss": 1.4769, - "step": 979 - }, - { - "epoch": 0.79, - "learning_rate": 1.725995491923131e-05, - "loss": 1.5308, - "step": 980 - }, - { - "epoch": 0.79, - "learning_rate": 1.7253944164006512e-05, - "loss": 1.5332, - "step": 981 - }, - { - "epoch": 0.79, - "learning_rate": 1.72479278722912e-05, - "loss": 1.5143, - "step": 982 - }, - { - "epoch": 0.8, - "learning_rate": 1.7241906048677247e-05, - "loss": 1.3385, - "step": 983 - }, - { - "epoch": 0.8, - "learning_rate": 1.7235878697760733e-05, - "loss": 1.3622, - "step": 984 - }, - { - "epoch": 0.8, - "learning_rate": 1.722984582414197e-05, - "loss": 1.4534, - "step": 985 - }, - { - "epoch": 0.8, - "learning_rate": 1.7223807432425477e-05, - "loss": 1.5287, - "step": 986 - }, - { - "epoch": 0.8, - "learning_rate": 1.721776352721999e-05, - "loss": 1.5158, - "step": 987 - }, - { - "epoch": 0.8, - "learning_rate": 1.7211714113138457e-05, - "loss": 1.3505, - "step": 988 - }, - { - "epoch": 0.8, - "learning_rate": 1.7205659194798013e-05, - "loss": 1.3256, - "step": 989 - }, - { - "epoch": 0.8, - "learning_rate": 1.7199598776820013e-05, - "loss": 1.4627, - "step": 990 - }, - { - "epoch": 0.8, - "learning_rate": 1.7193532863830002e-05, - "loss": 1.5268, - "step": 991 - }, - { - "epoch": 0.8, - "learning_rate": 1.7187461460457716e-05, - "loss": 1.4632, - "step": 992 - }, - { - "epoch": 0.8, - "learning_rate": 1.718138457133709e-05, - "loss": 1.4982, - "step": 993 - }, - { - "epoch": 0.8, - "learning_rate": 1.7175302201106232e-05, - "loss": 1.4433, - "step": 994 - }, - { - "epoch": 0.81, - "learning_rate": 1.7169214354407452e-05, - "loss": 1.3765, - "step": 995 - }, - { - "epoch": 0.81, - "learning_rate": 1.716312103588722e-05, - "loss": 1.4781, - "step": 996 - }, - { - "epoch": 0.81, - "learning_rate": 1.71570222501962e-05, - "loss": 1.5851, - "step": 997 - }, - { - "epoch": 0.81, - "learning_rate": 1.715091800198922e-05, - "loss": 1.5367, - "step": 998 - }, - { - "epoch": 0.81, - "learning_rate": 1.7144808295925272e-05, - "loss": 1.6048, - "step": 999 - }, - { - "epoch": 0.81, - "learning_rate": 1.713869313666753e-05, - "loss": 1.5089, - "step": 1000 - }, - { - "epoch": 0.81, - "learning_rate": 1.713257252888331e-05, - "loss": 1.5381, - "step": 1001 - }, - { - "epoch": 0.81, - "learning_rate": 1.7126446477244097e-05, - "loss": 1.3708, - "step": 1002 - }, - { - "epoch": 0.81, - "learning_rate": 1.712031498642554e-05, - "loss": 1.3399, - "step": 1003 - }, - { - "epoch": 0.81, - "learning_rate": 1.711417806110742e-05, - "loss": 1.4434, - "step": 1004 - }, - { - "epoch": 0.81, - "learning_rate": 1.710803570597368e-05, - "loss": 1.4828, - "step": 1005 - }, - { - "epoch": 0.81, - "learning_rate": 1.71018879257124e-05, - "loss": 1.5829, - "step": 1006 - }, - { - "epoch": 0.81, - "learning_rate": 1.7095734725015807e-05, - "loss": 1.5607, - "step": 1007 - }, - { - "epoch": 0.82, - "learning_rate": 1.708957610858026e-05, - "loss": 1.5237, - "step": 1008 - }, - { - "epoch": 0.82, - "learning_rate": 1.7083412081106253e-05, - "loss": 1.4524, - "step": 1009 - }, - { - "epoch": 0.82, - "learning_rate": 1.7077242647298405e-05, - "loss": 1.3794, - "step": 1010 - }, - { - "epoch": 0.82, - "learning_rate": 1.7071067811865477e-05, - "loss": 1.5726, - "step": 1011 - }, - { - "epoch": 0.82, - "learning_rate": 1.7064887579520336e-05, - "loss": 1.6461, - "step": 1012 - }, - { - "epoch": 0.82, - "learning_rate": 1.705870195497997e-05, - "loss": 1.3625, - "step": 1013 - }, - { - "epoch": 0.82, - "learning_rate": 1.7052510942965494e-05, - "loss": 1.5195, - "step": 1014 - }, - { - "epoch": 0.82, - "learning_rate": 1.7046314548202124e-05, - "loss": 1.4059, - "step": 1015 - }, - { - "epoch": 0.82, - "learning_rate": 1.7040112775419187e-05, - "loss": 1.6696, - "step": 1016 - }, - { - "epoch": 0.82, - "learning_rate": 1.7033905629350118e-05, - "loss": 1.5966, - "step": 1017 - }, - { - "epoch": 0.82, - "learning_rate": 1.702769311473245e-05, - "loss": 1.5028, - "step": 1018 - }, - { - "epoch": 0.82, - "learning_rate": 1.7021475236307818e-05, - "loss": 1.4793, - "step": 1019 - }, - { - "epoch": 0.83, - "learning_rate": 1.7015251998821938e-05, - "loss": 1.5452, - "step": 1020 - }, - { - "epoch": 0.83, - "learning_rate": 1.7009023407024633e-05, - "loss": 1.5367, - "step": 1021 - }, - { - "epoch": 0.83, - "learning_rate": 1.7002789465669806e-05, - "loss": 1.5466, - "step": 1022 - }, - { - "epoch": 0.83, - "learning_rate": 1.6996550179515435e-05, - "loss": 1.4684, - "step": 1023 - }, - { - "epoch": 0.83, - "learning_rate": 1.6990305553323584e-05, - "loss": 1.5703, - "step": 1024 - }, - { - "epoch": 0.83, - "learning_rate": 1.6984055591860402e-05, - "loss": 1.6161, - "step": 1025 - }, - { - "epoch": 0.83, - "learning_rate": 1.6977800299896093e-05, - "loss": 1.4194, - "step": 1026 - }, - { - "epoch": 0.83, - "learning_rate": 1.6971539682204945e-05, - "loss": 1.4733, - "step": 1027 - }, - { - "epoch": 0.83, - "learning_rate": 1.6965273743565296e-05, - "loss": 1.5273, - "step": 1028 - }, - { - "epoch": 0.83, - "learning_rate": 1.6959002488759556e-05, - "loss": 1.3824, - "step": 1029 - }, - { - "epoch": 0.83, - "learning_rate": 1.6952725922574188e-05, - "loss": 1.4391, - "step": 1030 - }, - { - "epoch": 0.83, - "learning_rate": 1.6946444049799708e-05, - "loss": 1.4768, - "step": 1031 - }, - { - "epoch": 0.83, - "learning_rate": 1.6940156875230688e-05, - "loss": 1.3247, - "step": 1032 - }, - { - "epoch": 0.84, - "learning_rate": 1.6933864403665744e-05, - "loss": 1.6031, - "step": 1033 - }, - { - "epoch": 0.84, - "learning_rate": 1.6927566639907527e-05, - "loss": 1.5628, - "step": 1034 - }, - { - "epoch": 0.84, - "learning_rate": 1.6921263588762738e-05, - "loss": 1.5056, - "step": 1035 - }, - { - "epoch": 0.84, - "learning_rate": 1.691495525504211e-05, - "loss": 1.4527, - "step": 1036 - }, - { - "epoch": 0.84, - "learning_rate": 1.6908641643560405e-05, - "loss": 1.6131, - "step": 1037 - }, - { - "epoch": 0.84, - "learning_rate": 1.6902322759136412e-05, - "loss": 1.4024, - "step": 1038 - }, - { - "epoch": 0.84, - "learning_rate": 1.6895998606592955e-05, - "loss": 1.3497, - "step": 1039 - }, - { - "epoch": 0.84, - "learning_rate": 1.688966919075687e-05, - "loss": 1.2815, - "step": 1040 - }, - { - "epoch": 0.84, - "learning_rate": 1.6883334516459005e-05, - "loss": 1.5157, - "step": 1041 - }, - { - "epoch": 0.84, - "learning_rate": 1.6876994588534234e-05, - "loss": 1.3215, - "step": 1042 - }, - { - "epoch": 0.84, - "learning_rate": 1.6870649411821437e-05, - "loss": 1.4728, - "step": 1043 - }, - { - "epoch": 0.84, - "learning_rate": 1.6864298991163492e-05, - "loss": 1.6463, - "step": 1044 - }, - { - "epoch": 0.85, - "learning_rate": 1.685794333140729e-05, - "loss": 1.5932, - "step": 1045 - }, - { - "epoch": 0.85, - "learning_rate": 1.6851582437403715e-05, - "loss": 1.4843, - "step": 1046 - }, - { - "epoch": 0.85, - "learning_rate": 1.6845216314007646e-05, - "loss": 1.5697, - "step": 1047 - }, - { - "epoch": 0.85, - "learning_rate": 1.6838844966077955e-05, - "loss": 1.4736, - "step": 1048 - }, - { - "epoch": 0.85, - "learning_rate": 1.6832468398477503e-05, - "loss": 1.6764, - "step": 1049 - }, - { - "epoch": 0.85, - "learning_rate": 1.682608661607313e-05, - "loss": 1.4785, - "step": 1050 - }, - { - "epoch": 0.85, - "learning_rate": 1.681969962373566e-05, - "loss": 1.5377, - "step": 1051 - }, - { - "epoch": 0.85, - "learning_rate": 1.681330742633989e-05, - "loss": 1.5148, - "step": 1052 - }, - { - "epoch": 0.85, - "learning_rate": 1.68069100287646e-05, - "loss": 1.5352, - "step": 1053 - }, - { - "epoch": 0.85, - "learning_rate": 1.680050743589252e-05, - "loss": 1.3609, - "step": 1054 - }, - { - "epoch": 0.85, - "learning_rate": 1.6794099652610368e-05, - "loss": 1.5015, - "step": 1055 - }, - { - "epoch": 0.85, - "learning_rate": 1.67876866838088e-05, - "loss": 1.5087, - "step": 1056 - }, - { - "epoch": 0.86, - "learning_rate": 1.6781268534382448e-05, - "loss": 1.6179, - "step": 1057 - }, - { - "epoch": 0.86, - "learning_rate": 1.677484520922989e-05, - "loss": 1.5206, - "step": 1058 - }, - { - "epoch": 0.86, - "learning_rate": 1.6768416713253662e-05, - "loss": 1.5597, - "step": 1059 - }, - { - "epoch": 0.86, - "learning_rate": 1.6761983051360232e-05, - "loss": 1.5088, - "step": 1060 - }, - { - "epoch": 0.86, - "learning_rate": 1.6755544228460022e-05, - "loss": 1.4013, - "step": 1061 - }, - { - "epoch": 0.86, - "learning_rate": 1.674910024946739e-05, - "loss": 1.4615, - "step": 1062 - }, - { - "epoch": 0.86, - "learning_rate": 1.6742651119300634e-05, - "loss": 1.3706, - "step": 1063 - }, - { - "epoch": 0.86, - "learning_rate": 1.6736196842881977e-05, - "loss": 1.4208, - "step": 1064 - }, - { - "epoch": 0.86, - "learning_rate": 1.672973742513757e-05, - "loss": 1.5143, - "step": 1065 - }, - { - "epoch": 0.86, - "learning_rate": 1.6723272870997494e-05, - "loss": 1.5176, - "step": 1066 - }, - { - "epoch": 0.86, - "learning_rate": 1.671680318539574e-05, - "loss": 1.4858, - "step": 1067 - }, - { - "epoch": 0.86, - "learning_rate": 1.671032837327023e-05, - "loss": 1.577, - "step": 1068 - }, - { - "epoch": 0.86, - "learning_rate": 1.6703848439562787e-05, - "loss": 1.5518, - "step": 1069 - }, - { - "epoch": 0.87, - "learning_rate": 1.6697363389219147e-05, - "loss": 1.5682, - "step": 1070 - }, - { - "epoch": 0.87, - "learning_rate": 1.669087322718895e-05, - "loss": 1.4899, - "step": 1071 - }, - { - "epoch": 0.87, - "learning_rate": 1.668437795842574e-05, - "loss": 1.5309, - "step": 1072 - }, - { - "epoch": 0.87, - "learning_rate": 1.6677877587886956e-05, - "loss": 1.5037, - "step": 1073 - }, - { - "epoch": 0.87, - "learning_rate": 1.6671372120533935e-05, - "loss": 1.499, - "step": 1074 - }, - { - "epoch": 0.87, - "learning_rate": 1.66648615613319e-05, - "loss": 1.3936, - "step": 1075 - }, - { - "epoch": 0.87, - "learning_rate": 1.6658345915249954e-05, - "loss": 1.5097, - "step": 1076 - }, - { - "epoch": 0.87, - "learning_rate": 1.6651825187261103e-05, - "loss": 1.5118, - "step": 1077 - }, - { - "epoch": 0.87, - "learning_rate": 1.6645299382342207e-05, - "loss": 1.5608, - "step": 1078 - }, - { - "epoch": 0.87, - "learning_rate": 1.6638768505474023e-05, - "loss": 1.5721, - "step": 1079 - }, - { - "epoch": 0.87, - "learning_rate": 1.6632232561641158e-05, - "loss": 1.6079, - "step": 1080 - }, - { - "epoch": 0.87, - "learning_rate": 1.66256915558321e-05, - "loss": 1.4034, - "step": 1081 - }, - { - "epoch": 0.88, - "learning_rate": 1.66191454930392e-05, - "loss": 1.6607, - "step": 1082 - }, - { - "epoch": 0.88, - "learning_rate": 1.661259437825867e-05, - "loss": 1.5616, - "step": 1083 - }, - { - "epoch": 0.88, - "learning_rate": 1.6606038216490566e-05, - "loss": 1.4304, - "step": 1084 - }, - { - "epoch": 0.88, - "learning_rate": 1.6599477012738806e-05, - "loss": 1.5558, - "step": 1085 - }, - { - "epoch": 0.88, - "learning_rate": 1.659291077201116e-05, - "loss": 1.4881, - "step": 1086 - }, - { - "epoch": 0.88, - "learning_rate": 1.6586339499319228e-05, - "loss": 1.5662, - "step": 1087 - }, - { - "epoch": 0.88, - "learning_rate": 1.657976319967846e-05, - "loss": 1.499, - "step": 1088 - }, - { - "epoch": 0.88, - "learning_rate": 1.6573181878108154e-05, - "loss": 1.3515, - "step": 1089 - }, - { - "epoch": 0.88, - "learning_rate": 1.6566595539631417e-05, - "loss": 1.6315, - "step": 1090 - }, - { - "epoch": 0.88, - "learning_rate": 1.65600041892752e-05, - "loss": 1.4261, - "step": 1091 - }, - { - "epoch": 0.88, - "learning_rate": 1.6553407832070275e-05, - "loss": 1.3242, - "step": 1092 - }, - { - "epoch": 0.88, - "learning_rate": 1.6546806473051244e-05, - "loss": 1.3031, - "step": 1093 - }, - { - "epoch": 0.89, - "learning_rate": 1.6540200117256507e-05, - "loss": 1.3639, - "step": 1094 - }, - { - "epoch": 0.89, - "learning_rate": 1.65335887697283e-05, - "loss": 1.3961, - "step": 1095 - }, - { - "epoch": 0.89, - "learning_rate": 1.6526972435512655e-05, - "loss": 1.4269, - "step": 1096 - }, - { - "epoch": 0.89, - "learning_rate": 1.6520351119659414e-05, - "loss": 1.5195, - "step": 1097 - }, - { - "epoch": 0.89, - "learning_rate": 1.6513724827222225e-05, - "loss": 1.4967, - "step": 1098 - }, - { - "epoch": 0.89, - "learning_rate": 1.6507093563258523e-05, - "loss": 1.5039, - "step": 1099 - }, - { - "epoch": 0.89, - "learning_rate": 1.6500457332829553e-05, - "loss": 1.4771, - "step": 1100 - }, - { - "epoch": 0.89, - "learning_rate": 1.6493816141000335e-05, - "loss": 1.5145, - "step": 1101 - }, - { - "epoch": 0.89, - "learning_rate": 1.648716999283969e-05, - "loss": 1.6113, - "step": 1102 - }, - { - "epoch": 0.89, - "learning_rate": 1.6480518893420214e-05, - "loss": 1.4997, - "step": 1103 - }, - { - "epoch": 0.89, - "learning_rate": 1.647386284781828e-05, - "loss": 1.4223, - "step": 1104 - }, - { - "epoch": 0.89, - "learning_rate": 1.646720186111404e-05, - "loss": 1.4543, - "step": 1105 - }, - { - "epoch": 0.89, - "learning_rate": 1.6460535938391412e-05, - "loss": 1.5138, - "step": 1106 - }, - { - "epoch": 0.9, - "learning_rate": 1.6453865084738097e-05, - "loss": 1.5518, - "step": 1107 - }, - { - "epoch": 0.9, - "learning_rate": 1.644718930524554e-05, - "loss": 1.4373, - "step": 1108 - }, - { - "epoch": 0.9, - "learning_rate": 1.6440508605008954e-05, - "loss": 1.428, - "step": 1109 - }, - { - "epoch": 0.9, - "learning_rate": 1.6433822989127314e-05, - "loss": 1.4613, - "step": 1110 - }, - { - "epoch": 0.9, - "learning_rate": 1.6427132462703333e-05, - "loss": 1.387, - "step": 1111 - }, - { - "epoch": 0.9, - "learning_rate": 1.6420437030843482e-05, - "loss": 1.6015, - "step": 1112 - }, - { - "epoch": 0.9, - "learning_rate": 1.6413736698657973e-05, - "loss": 1.543, - "step": 1113 - }, - { - "epoch": 0.9, - "learning_rate": 1.6407031471260762e-05, - "loss": 1.4991, - "step": 1114 - }, - { - "epoch": 0.9, - "learning_rate": 1.640032135376953e-05, - "loss": 1.619, - "step": 1115 - }, - { - "epoch": 0.9, - "learning_rate": 1.639360635130571e-05, - "loss": 1.5852, - "step": 1116 - }, - { - "epoch": 0.9, - "learning_rate": 1.6386886468994437e-05, - "loss": 1.2775, - "step": 1117 - }, - { - "epoch": 0.9, - "learning_rate": 1.63801617119646e-05, - "loss": 1.5418, - "step": 1118 - }, - { - "epoch": 0.91, - "learning_rate": 1.6373432085348777e-05, - "loss": 1.513, - "step": 1119 - }, - { - "epoch": 0.91, - "learning_rate": 1.636669759428329e-05, - "loss": 1.6944, - "step": 1120 - }, - { - "epoch": 0.91, - "learning_rate": 1.635995824390816e-05, - "loss": 1.4034, - "step": 1121 - }, - { - "epoch": 0.91, - "learning_rate": 1.6353214039367123e-05, - "loss": 1.5966, - "step": 1122 - }, - { - "epoch": 0.91, - "learning_rate": 1.634646498580761e-05, - "loss": 1.6214, - "step": 1123 - }, - { - "epoch": 0.91, - "learning_rate": 1.6339711088380765e-05, - "loss": 1.4191, - "step": 1124 - }, - { - "epoch": 0.91, - "learning_rate": 1.633295235224142e-05, - "loss": 1.4739, - "step": 1125 - }, - { - "epoch": 0.91, - "learning_rate": 1.6326188782548106e-05, - "loss": 1.4302, - "step": 1126 - }, - { - "epoch": 0.91, - "learning_rate": 1.631942038446304e-05, - "loss": 1.4086, - "step": 1127 - }, - { - "epoch": 0.91, - "learning_rate": 1.6312647163152128e-05, - "loss": 1.5646, - "step": 1128 - }, - { - "epoch": 0.91, - "learning_rate": 1.630586912378495e-05, - "loss": 1.4407, - "step": 1129 - }, - { - "epoch": 0.91, - "learning_rate": 1.6299086271534764e-05, - "loss": 1.5806, - "step": 1130 - }, - { - "epoch": 0.92, - "learning_rate": 1.629229861157851e-05, - "loss": 1.4927, - "step": 1131 - }, - { - "epoch": 0.92, - "learning_rate": 1.6285506149096794e-05, - "loss": 1.3568, - "step": 1132 - }, - { - "epoch": 0.92, - "learning_rate": 1.627870888927389e-05, - "loss": 1.4433, - "step": 1133 - }, - { - "epoch": 0.92, - "learning_rate": 1.6271906837297715e-05, - "loss": 1.3651, - "step": 1134 - }, - { - "epoch": 0.92, - "learning_rate": 1.6265099998359868e-05, - "loss": 1.6317, - "step": 1135 - }, - { - "epoch": 0.92, - "learning_rate": 1.625828837765559e-05, - "loss": 1.4478, - "step": 1136 - }, - { - "epoch": 0.92, - "learning_rate": 1.6251471980383776e-05, - "loss": 1.5549, - "step": 1137 - }, - { - "epoch": 0.92, - "learning_rate": 1.6244650811746954e-05, - "loss": 1.4948, - "step": 1138 - }, - { - "epoch": 0.92, - "learning_rate": 1.6237824876951314e-05, - "loss": 1.5983, - "step": 1139 - }, - { - "epoch": 0.92, - "learning_rate": 1.6230994181206674e-05, - "loss": 1.3836, - "step": 1140 - }, - { - "epoch": 0.92, - "learning_rate": 1.6224158729726475e-05, - "loss": 1.5278, - "step": 1141 - }, - { - "epoch": 0.92, - "learning_rate": 1.6217318527727802e-05, - "loss": 1.5151, - "step": 1142 - }, - { - "epoch": 0.92, - "learning_rate": 1.6210473580431366e-05, - "loss": 1.4873, - "step": 1143 - }, - { - "epoch": 0.93, - "learning_rate": 1.620362389306149e-05, - "loss": 1.3992, - "step": 1144 - }, - { - "epoch": 0.93, - "learning_rate": 1.6196769470846122e-05, - "loss": 1.5369, - "step": 1145 - }, - { - "epoch": 0.93, - "learning_rate": 1.6189910319016826e-05, - "loss": 1.4299, - "step": 1146 - }, - { - "epoch": 0.93, - "learning_rate": 1.6183046442808764e-05, - "loss": 1.3765, - "step": 1147 - }, - { - "epoch": 0.93, - "learning_rate": 1.6176177847460716e-05, - "loss": 1.4537, - "step": 1148 - }, - { - "epoch": 0.93, - "learning_rate": 1.6169304538215057e-05, - "loss": 1.4938, - "step": 1149 - }, - { - "epoch": 0.93, - "learning_rate": 1.6162426520317765e-05, - "loss": 1.4805, - "step": 1150 - }, - { - "epoch": 0.93, - "learning_rate": 1.6155543799018413e-05, - "loss": 1.3976, - "step": 1151 - }, - { - "epoch": 0.93, - "learning_rate": 1.6148656379570144e-05, - "loss": 1.3411, - "step": 1152 - }, - { - "epoch": 0.93, - "learning_rate": 1.6141764267229723e-05, - "loss": 1.45, - "step": 1153 - }, - { - "epoch": 0.93, - "learning_rate": 1.6134867467257462e-05, - "loss": 1.4446, - "step": 1154 - }, - { - "epoch": 0.93, - "learning_rate": 1.6127965984917273e-05, - "loss": 1.3652, - "step": 1155 - }, - { - "epoch": 0.94, - "learning_rate": 1.612105982547663e-05, - "loss": 1.5378, - "step": 1156 - }, - { - "epoch": 0.94, - "learning_rate": 1.6114148994206582e-05, - "loss": 1.423, - "step": 1157 - }, - { - "epoch": 0.94, - "learning_rate": 1.6107233496381742e-05, - "loss": 1.5992, - "step": 1158 - }, - { - "epoch": 0.94, - "learning_rate": 1.6100313337280288e-05, - "loss": 1.3489, - "step": 1159 - }, - { - "epoch": 0.94, - "learning_rate": 1.6093388522183948e-05, - "loss": 1.4148, - "step": 1160 - }, - { - "epoch": 0.94, - "learning_rate": 1.6086459056378015e-05, - "loss": 1.4885, - "step": 1161 - }, - { - "epoch": 0.94, - "learning_rate": 1.6079524945151316e-05, - "loss": 1.5592, - "step": 1162 - }, - { - "epoch": 0.94, - "learning_rate": 1.607258619379624e-05, - "loss": 1.5012, - "step": 1163 - }, - { - "epoch": 0.94, - "learning_rate": 1.606564280760871e-05, - "loss": 1.5044, - "step": 1164 - }, - { - "epoch": 0.94, - "learning_rate": 1.6058694791888185e-05, - "loss": 1.3981, - "step": 1165 - }, - { - "epoch": 0.94, - "learning_rate": 1.6051742151937655e-05, - "loss": 1.4692, - "step": 1166 - }, - { - "epoch": 0.94, - "learning_rate": 1.6044784893063647e-05, - "loss": 1.5158, - "step": 1167 - }, - { - "epoch": 0.94, - "learning_rate": 1.603782302057621e-05, - "loss": 1.4907, - "step": 1168 - }, - { - "epoch": 0.95, - "learning_rate": 1.6030856539788916e-05, - "loss": 1.4301, - "step": 1169 - }, - { - "epoch": 0.95, - "learning_rate": 1.6023885456018852e-05, - "loss": 1.5399, - "step": 1170 - }, - { - "epoch": 0.95, - "learning_rate": 1.6016909774586616e-05, - "loss": 1.3746, - "step": 1171 - }, - { - "epoch": 0.95, - "learning_rate": 1.6009929500816318e-05, - "loss": 1.5652, - "step": 1172 - }, - { - "epoch": 0.95, - "learning_rate": 1.6002944640035577e-05, - "loss": 1.2885, - "step": 1173 - }, - { - "epoch": 0.95, - "learning_rate": 1.59959551975755e-05, - "loss": 1.2036, - "step": 1174 - }, - { - "epoch": 0.95, - "learning_rate": 1.5988961178770707e-05, - "loss": 1.4767, - "step": 1175 - }, - { - "epoch": 0.95, - "learning_rate": 1.5981962588959304e-05, - "loss": 1.4155, - "step": 1176 - }, - { - "epoch": 0.95, - "learning_rate": 1.5974959433482877e-05, - "loss": 1.4911, - "step": 1177 - }, - { - "epoch": 0.95, - "learning_rate": 1.5967951717686513e-05, - "loss": 1.6203, - "step": 1178 - }, - { - "epoch": 0.95, - "learning_rate": 1.596093944691877e-05, - "loss": 1.5049, - "step": 1179 - }, - { - "epoch": 0.95, - "learning_rate": 1.595392262653168e-05, - "loss": 1.312, - "step": 1180 - }, - { - "epoch": 0.96, - "learning_rate": 1.5946901261880755e-05, - "loss": 1.4362, - "step": 1181 - }, - { - "epoch": 0.96, - "learning_rate": 1.593987535832497e-05, - "loss": 1.401, - "step": 1182 - }, - { - "epoch": 0.96, - "learning_rate": 1.593284492122677e-05, - "loss": 1.4818, - "step": 1183 - }, - { - "epoch": 0.96, - "learning_rate": 1.5925809955952054e-05, - "loss": 1.6743, - "step": 1184 - }, - { - "epoch": 0.96, - "learning_rate": 1.5918770467870174e-05, - "loss": 1.5337, - "step": 1185 - }, - { - "epoch": 0.96, - "learning_rate": 1.5911726462353947e-05, - "loss": 1.6183, - "step": 1186 - }, - { - "epoch": 0.96, - "learning_rate": 1.590467794477963e-05, - "loss": 1.386, - "step": 1187 - }, - { - "epoch": 0.96, - "learning_rate": 1.5897624920526914e-05, - "loss": 1.5765, - "step": 1188 - }, - { - "epoch": 0.96, - "learning_rate": 1.5890567394978954e-05, - "loss": 1.483, - "step": 1189 - }, - { - "epoch": 0.96, - "learning_rate": 1.5883505373522317e-05, - "loss": 1.471, - "step": 1190 - }, - { - "epoch": 0.96, - "learning_rate": 1.5876438861547016e-05, - "loss": 1.4263, - "step": 1191 - }, - { - "epoch": 0.96, - "learning_rate": 1.5869367864446478e-05, - "loss": 1.379, - "step": 1192 - }, - { - "epoch": 0.97, - "learning_rate": 1.586229238761757e-05, - "loss": 1.4728, - "step": 1193 - }, - { - "epoch": 0.97, - "learning_rate": 1.5855212436460563e-05, - "loss": 1.6278, - "step": 1194 - }, - { - "epoch": 0.97, - "learning_rate": 1.5848128016379152e-05, - "loss": 1.379, - "step": 1195 - }, - { - "epoch": 0.97, - "learning_rate": 1.5841039132780444e-05, - "loss": 1.4435, - "step": 1196 - }, - { - "epoch": 0.97, - "learning_rate": 1.5833945791074943e-05, - "loss": 1.6049, - "step": 1197 - }, - { - "epoch": 0.97, - "learning_rate": 1.582684799667656e-05, - "loss": 1.4534, - "step": 1198 - }, - { - "epoch": 0.97, - "learning_rate": 1.581974575500261e-05, - "loss": 1.4875, - "step": 1199 - }, - { - "epoch": 0.97, - "learning_rate": 1.5812639071473804e-05, - "loss": 1.3534, - "step": 1200 - }, - { - "epoch": 0.97, - "learning_rate": 1.5805527951514224e-05, - "loss": 1.5325, - "step": 1201 - }, - { - "epoch": 0.97, - "learning_rate": 1.5798412400551365e-05, - "loss": 1.5207, - "step": 1202 - }, - { - "epoch": 0.97, - "learning_rate": 1.579129242401608e-05, - "loss": 1.5594, - "step": 1203 - }, - { - "epoch": 0.97, - "learning_rate": 1.5784168027342616e-05, - "loss": 1.4524, - "step": 1204 - }, - { - "epoch": 0.97, - "learning_rate": 1.5777039215968592e-05, - "loss": 1.5025, - "step": 1205 - }, - { - "epoch": 0.98, - "learning_rate": 1.576990599533498e-05, - "loss": 1.5904, - "step": 1206 - }, - { - "epoch": 0.98, - "learning_rate": 1.5762768370886136e-05, - "loss": 1.5331, - "step": 1207 - }, - { - "epoch": 0.98, - "learning_rate": 1.5755626348069775e-05, - "loss": 1.4736, - "step": 1208 - }, - { - "epoch": 0.98, - "learning_rate": 1.5748479932336957e-05, - "loss": 1.3984, - "step": 1209 - }, - { - "epoch": 0.98, - "learning_rate": 1.574132912914211e-05, - "loss": 1.4351, - "step": 1210 - }, - { - "epoch": 0.98, - "learning_rate": 1.5734173943943002e-05, - "loss": 1.5117, - "step": 1211 - }, - { - "epoch": 0.98, - "learning_rate": 1.572701438220074e-05, - "loss": 1.3251, - "step": 1212 - }, - { - "epoch": 0.98, - "learning_rate": 1.5719850449379783e-05, - "loss": 1.4178, - "step": 1213 - }, - { - "epoch": 0.98, - "learning_rate": 1.5712682150947926e-05, - "loss": 1.4426, - "step": 1214 - }, - { - "epoch": 0.98, - "learning_rate": 1.5705509492376283e-05, - "loss": 1.4601, - "step": 1215 - }, - { - "epoch": 0.98, - "learning_rate": 1.5698332479139307e-05, - "loss": 1.5112, - "step": 1216 - }, - { - "epoch": 0.98, - "learning_rate": 1.5691151116714776e-05, - "loss": 1.2767, - "step": 1217 - }, - { - "epoch": 0.99, - "learning_rate": 1.568396541058378e-05, - "loss": 1.5352, - "step": 1218 - }, - { - "epoch": 0.99, - "learning_rate": 1.567677536623073e-05, - "loss": 1.4237, - "step": 1219 - }, - { - "epoch": 0.99, - "learning_rate": 1.566958098914334e-05, - "loss": 1.5295, - "step": 1220 - }, - { - "epoch": 0.99, - "learning_rate": 1.566238228481264e-05, - "loss": 1.4693, - "step": 1221 - }, - { - "epoch": 0.99, - "learning_rate": 1.565517925873296e-05, - "loss": 1.4285, - "step": 1222 - }, - { - "epoch": 0.99, - "learning_rate": 1.564797191640193e-05, - "loss": 1.4853, - "step": 1223 - }, - { - "epoch": 0.99, - "learning_rate": 1.5640760263320468e-05, - "loss": 1.4835, - "step": 1224 - }, - { - "epoch": 0.99, - "learning_rate": 1.5633544304992785e-05, - "loss": 1.3076, - "step": 1225 - }, - { - "epoch": 0.99, - "learning_rate": 1.562632404692639e-05, - "loss": 1.4609, - "step": 1226 - }, - { - "epoch": 0.99, - "learning_rate": 1.5619099494632047e-05, - "loss": 1.4708, - "step": 1227 - }, - { - "epoch": 0.99, - "learning_rate": 1.5611870653623826e-05, - "loss": 1.3178, - "step": 1228 - }, - { - "epoch": 0.99, - "learning_rate": 1.5604637529419052e-05, - "loss": 1.405, - "step": 1229 - }, - { - "epoch": 1.0, - "learning_rate": 1.5597400127538324e-05, - "loss": 1.4628, - "step": 1230 - }, - { - "epoch": 1.0, - "learning_rate": 1.5590158453505513e-05, - "loss": 1.4308, - "step": 1231 - }, - { - "epoch": 1.0, - "learning_rate": 1.5582912512847742e-05, - "loss": 1.4323, - "step": 1232 - }, - { - "epoch": 1.0, - "learning_rate": 1.5575662311095388e-05, - "loss": 1.3509, - "step": 1233 - }, - { - "epoch": 1.0, - "learning_rate": 1.5568407853782086e-05, - "loss": 1.3018, - "step": 1234 - }, - { - "epoch": 1.0, - "learning_rate": 1.556114914644473e-05, - "loss": 1.525, - "step": 1235 - }, - { - "epoch": 1.0, - "learning_rate": 1.5553886194623426e-05, - "loss": 1.4646, - "step": 1236 - }, - { - "epoch": 1.0, - "learning_rate": 1.554661900386156e-05, - "loss": 1.3257, - "step": 1237 - }, - { - "epoch": 1.0, - "learning_rate": 1.5539347579705718e-05, - "loss": 1.3756, - "step": 1238 - }, - { - "epoch": 1.0, - "learning_rate": 1.5532071927705735e-05, - "loss": 1.3837, - "step": 1239 - }, - { - "epoch": 1.0, - "learning_rate": 1.5524792053414676e-05, - "loss": 1.23, - "step": 1240 - }, - { - "epoch": 1.0, - "learning_rate": 1.5517507962388817e-05, - "loss": 1.3752, - "step": 1241 - }, - { - "epoch": 1.0, - "learning_rate": 1.551021966018766e-05, - "loss": 1.3779, - "step": 1242 - }, - { - "epoch": 1.01, - "learning_rate": 1.5502927152373913e-05, - "loss": 1.2191, - "step": 1243 - }, - { - "epoch": 1.01, - "learning_rate": 1.5495630444513512e-05, - "loss": 1.4238, - "step": 1244 - }, - { - "epoch": 1.01, - "learning_rate": 1.5488329542175576e-05, - "loss": 1.5856, - "step": 1245 - }, - { - "epoch": 1.01, - "learning_rate": 1.548102445093244e-05, - "loss": 1.2487, - "step": 1246 - }, - { - "epoch": 1.01, - "learning_rate": 1.5473715176359627e-05, - "loss": 1.3221, - "step": 1247 - }, - { - "epoch": 1.01, - "learning_rate": 1.5466401724035863e-05, - "loss": 1.4664, - "step": 1248 - }, - { - "epoch": 1.01, - "learning_rate": 1.5459084099543056e-05, - "loss": 1.3097, - "step": 1249 - }, - { - "epoch": 1.01, - "learning_rate": 1.5451762308466302e-05, - "loss": 1.3631, - "step": 1250 - }, - { - "epoch": 1.01, - "learning_rate": 1.5444436356393872e-05, - "loss": 1.3673, - "step": 1251 - }, - { - "epoch": 1.01, - "learning_rate": 1.5437106248917218e-05, - "loss": 1.3058, - "step": 1252 - }, - { - "epoch": 1.01, - "learning_rate": 1.5429771991630958e-05, - "loss": 1.2596, - "step": 1253 - }, - { - "epoch": 1.01, - "learning_rate": 1.5422433590132883e-05, - "loss": 1.2705, - "step": 1254 - }, - { - "epoch": 1.02, - "learning_rate": 1.5415091050023946e-05, - "loss": 1.2774, - "step": 1255 - }, - { - "epoch": 1.02, - "learning_rate": 1.540774437690825e-05, - "loss": 1.2011, - "step": 1256 - }, - { - "epoch": 1.02, - "learning_rate": 1.540039357639307e-05, - "loss": 1.3794, - "step": 1257 - }, - { - "epoch": 1.02, - "learning_rate": 1.5393038654088814e-05, - "loss": 1.3274, - "step": 1258 - }, - { - "epoch": 1.02, - "learning_rate": 1.5385679615609045e-05, - "loss": 1.2789, - "step": 1259 - }, - { - "epoch": 1.02, - "learning_rate": 1.5378316466570466e-05, - "loss": 1.1074, - "step": 1260 - }, - { - "epoch": 1.02, - "learning_rate": 1.537094921259292e-05, - "loss": 1.1044, - "step": 1261 - }, - { - "epoch": 1.02, - "learning_rate": 1.536357785929937e-05, - "loss": 1.3489, - "step": 1262 - }, - { - "epoch": 1.02, - "learning_rate": 1.535620241231593e-05, - "loss": 1.2226, - "step": 1263 - }, - { - "epoch": 1.02, - "learning_rate": 1.5348822877271818e-05, - "loss": 1.3996, - "step": 1264 - }, - { - "epoch": 1.02, - "learning_rate": 1.5341439259799386e-05, - "loss": 1.3555, - "step": 1265 - }, - { - "epoch": 1.02, - "learning_rate": 1.533405156553409e-05, - "loss": 1.1923, - "step": 1266 - }, - { - "epoch": 1.03, - "learning_rate": 1.53266598001145e-05, - "loss": 1.249, - "step": 1267 - }, - { - "epoch": 1.03, - "learning_rate": 1.531926396918231e-05, - "loss": 1.1102, - "step": 1268 - }, - { - "epoch": 1.03, - "learning_rate": 1.5311864078382304e-05, - "loss": 1.2044, - "step": 1269 - }, - { - "epoch": 1.03, - "learning_rate": 1.530446013336235e-05, - "loss": 1.2682, - "step": 1270 - }, - { - "epoch": 1.03, - "learning_rate": 1.5297052139773435e-05, - "loss": 1.2467, - "step": 1271 - }, - { - "epoch": 1.03, - "learning_rate": 1.5289640103269626e-05, - "loss": 1.3872, - "step": 1272 - }, - { - "epoch": 1.03, - "learning_rate": 1.5282224029508076e-05, - "loss": 1.24, - "step": 1273 - }, - { - "epoch": 1.03, - "learning_rate": 1.5274803924149012e-05, - "loss": 1.3839, - "step": 1274 - }, - { - "epoch": 1.03, - "learning_rate": 1.5267379792855752e-05, - "loss": 1.338, - "step": 1275 - }, - { - "epoch": 1.03, - "learning_rate": 1.5259951641294677e-05, - "loss": 1.3302, - "step": 1276 - }, - { - "epoch": 1.03, - "learning_rate": 1.5252519475135241e-05, - "loss": 1.1035, - "step": 1277 - }, - { - "epoch": 1.03, - "learning_rate": 1.524508330004996e-05, - "loss": 1.3846, - "step": 1278 - }, - { - "epoch": 1.03, - "learning_rate": 1.5237643121714412e-05, - "loss": 1.2931, - "step": 1279 - }, - { - "epoch": 1.04, - "learning_rate": 1.5230198945807226e-05, - "loss": 1.1985, - "step": 1280 - }, - { - "epoch": 1.04, - "learning_rate": 1.5222750778010085e-05, - "loss": 1.3393, - "step": 1281 - }, - { - "epoch": 1.04, - "learning_rate": 1.521529862400772e-05, - "loss": 1.2797, - "step": 1282 - }, - { - "epoch": 1.04, - "learning_rate": 1.5207842489487901e-05, - "loss": 1.2973, - "step": 1283 - }, - { - "epoch": 1.04, - "learning_rate": 1.5200382380141436e-05, - "loss": 1.3093, - "step": 1284 - }, - { - "epoch": 1.04, - "learning_rate": 1.5192918301662174e-05, - "loss": 1.3208, - "step": 1285 - }, - { - "epoch": 1.04, - "learning_rate": 1.5185450259746985e-05, - "loss": 1.4209, - "step": 1286 - }, - { - "epoch": 1.04, - "learning_rate": 1.5177978260095772e-05, - "loss": 1.3959, - "step": 1287 - }, - { - "epoch": 1.04, - "learning_rate": 1.5170502308411444e-05, - "loss": 1.2304, - "step": 1288 - }, - { - "epoch": 1.04, - "learning_rate": 1.5163022410399944e-05, - "loss": 1.2693, - "step": 1289 - }, - { - "epoch": 1.04, - "learning_rate": 1.515553857177022e-05, - "loss": 1.311, - "step": 1290 - }, - { - "epoch": 1.04, - "learning_rate": 1.5148050798234222e-05, - "loss": 1.2597, - "step": 1291 - }, - { - "epoch": 1.05, - "learning_rate": 1.5140559095506908e-05, - "loss": 1.4796, - "step": 1292 - }, - { - "epoch": 1.05, - "learning_rate": 1.513306346930624e-05, - "loss": 1.2865, - "step": 1293 - }, - { - "epoch": 1.05, - "learning_rate": 1.5125563925353166e-05, - "loss": 1.3235, - "step": 1294 - }, - { - "epoch": 1.05, - "learning_rate": 1.5118060469371628e-05, - "loss": 1.3794, - "step": 1295 - }, - { - "epoch": 1.05, - "learning_rate": 1.5110553107088555e-05, - "loss": 1.4342, - "step": 1296 - }, - { - "epoch": 1.05, - "learning_rate": 1.5103041844233857e-05, - "loss": 1.2605, - "step": 1297 - }, - { - "epoch": 1.05, - "learning_rate": 1.5095526686540418e-05, - "loss": 1.2953, - "step": 1298 - }, - { - "epoch": 1.05, - "learning_rate": 1.5088007639744096e-05, - "loss": 1.3283, - "step": 1299 - }, - { - "epoch": 1.05, - "learning_rate": 1.5080484709583715e-05, - "loss": 1.3284, - "step": 1300 - }, - { - "epoch": 1.05, - "learning_rate": 1.5072957901801075e-05, - "loss": 1.2956, - "step": 1301 - }, - { - "epoch": 1.05, - "learning_rate": 1.5065427222140918e-05, - "loss": 1.2595, - "step": 1302 - }, - { - "epoch": 1.05, - "learning_rate": 1.505789267635096e-05, - "loss": 1.4398, - "step": 1303 - }, - { - "epoch": 1.06, - "learning_rate": 1.5050354270181846e-05, - "loss": 1.4387, - "step": 1304 - }, - { - "epoch": 1.06, - "learning_rate": 1.5042812009387184e-05, - "loss": 1.26, - "step": 1305 - }, - { - "epoch": 1.06, - "learning_rate": 1.5035265899723519e-05, - "loss": 1.2065, - "step": 1306 - }, - { - "epoch": 1.06, - "learning_rate": 1.5027715946950334e-05, - "loss": 1.2984, - "step": 1307 - }, - { - "epoch": 1.06, - "learning_rate": 1.5020162156830048e-05, - "loss": 1.3145, - "step": 1308 - }, - { - "epoch": 1.06, - "learning_rate": 1.5012604535127994e-05, - "loss": 1.2204, - "step": 1309 - }, - { - "epoch": 1.06, - "learning_rate": 1.5005043087612452e-05, - "loss": 1.3114, - "step": 1310 - }, - { - "epoch": 1.06, - "learning_rate": 1.4997477820054606e-05, - "loss": 1.2276, - "step": 1311 - }, - { - "epoch": 1.06, - "learning_rate": 1.4989908738228567e-05, - "loss": 1.3488, - "step": 1312 - }, - { - "epoch": 1.06, - "learning_rate": 1.498233584791134e-05, - "loss": 1.2212, - "step": 1313 - }, - { - "epoch": 1.06, - "learning_rate": 1.4974759154882858e-05, - "loss": 1.103, - "step": 1314 - }, - { - "epoch": 1.06, - "learning_rate": 1.4967178664925937e-05, - "loss": 1.3404, - "step": 1315 - }, - { - "epoch": 1.06, - "learning_rate": 1.4959594383826307e-05, - "loss": 1.3247, - "step": 1316 - }, - { - "epoch": 1.07, - "learning_rate": 1.4952006317372578e-05, - "loss": 1.2695, - "step": 1317 - }, - { - "epoch": 1.07, - "learning_rate": 1.4944414471356263e-05, - "loss": 1.3802, - "step": 1318 - }, - { - "epoch": 1.07, - "learning_rate": 1.4936818851571748e-05, - "loss": 1.3236, - "step": 1319 - }, - { - "epoch": 1.07, - "learning_rate": 1.4929219463816302e-05, - "loss": 1.269, - "step": 1320 - }, - { - "epoch": 1.07, - "learning_rate": 1.4921616313890073e-05, - "loss": 1.2295, - "step": 1321 - }, - { - "epoch": 1.07, - "learning_rate": 1.4914009407596082e-05, - "loss": 1.1723, - "step": 1322 - }, - { - "epoch": 1.07, - "learning_rate": 1.490639875074021e-05, - "loss": 1.3142, - "step": 1323 - }, - { - "epoch": 1.07, - "learning_rate": 1.4898784349131204e-05, - "loss": 1.2877, - "step": 1324 - }, - { - "epoch": 1.07, - "learning_rate": 1.4891166208580673e-05, - "loss": 1.3196, - "step": 1325 - }, - { - "epoch": 1.07, - "learning_rate": 1.4883544334903073e-05, - "loss": 1.3066, - "step": 1326 - }, - { - "epoch": 1.07, - "learning_rate": 1.4875918733915712e-05, - "loss": 1.399, - "step": 1327 - }, - { - "epoch": 1.07, - "learning_rate": 1.4868289411438747e-05, - "loss": 1.3897, - "step": 1328 - }, - { - "epoch": 1.08, - "learning_rate": 1.4860656373295173e-05, - "loss": 1.3262, - "step": 1329 - }, - { - "epoch": 1.08, - "learning_rate": 1.4853019625310813e-05, - "loss": 1.396, - "step": 1330 - }, - { - "epoch": 1.08, - "learning_rate": 1.4845379173314331e-05, - "loss": 1.2679, - "step": 1331 - }, - { - "epoch": 1.08, - "learning_rate": 1.4837735023137216e-05, - "loss": 1.4434, - "step": 1332 - }, - { - "epoch": 1.08, - "learning_rate": 1.483008718061378e-05, - "loss": 1.2654, - "step": 1333 - }, - { - "epoch": 1.08, - "learning_rate": 1.4822435651581151e-05, - "loss": 1.2958, - "step": 1334 - }, - { - "epoch": 1.08, - "learning_rate": 1.4814780441879273e-05, - "loss": 1.2051, - "step": 1335 - }, - { - "epoch": 1.08, - "learning_rate": 1.4807121557350896e-05, - "loss": 1.3245, - "step": 1336 - }, - { - "epoch": 1.08, - "learning_rate": 1.4799459003841575e-05, - "loss": 1.2884, - "step": 1337 - }, - { - "epoch": 1.08, - "learning_rate": 1.4791792787199667e-05, - "loss": 1.3105, - "step": 1338 - }, - { - "epoch": 1.08, - "learning_rate": 1.4784122913276331e-05, - "loss": 1.2905, - "step": 1339 - }, - { - "epoch": 1.08, - "learning_rate": 1.4776449387925507e-05, - "loss": 1.1693, - "step": 1340 - }, - { - "epoch": 1.08, - "learning_rate": 1.4768772217003922e-05, - "loss": 1.2371, - "step": 1341 - }, - { - "epoch": 1.09, - "learning_rate": 1.4761091406371097e-05, - "loss": 1.3205, - "step": 1342 - }, - { - "epoch": 1.09, - "learning_rate": 1.4753406961889319e-05, - "loss": 1.2199, - "step": 1343 - }, - { - "epoch": 1.09, - "learning_rate": 1.4745718889423652e-05, - "loss": 1.4258, - "step": 1344 - }, - { - "epoch": 1.09, - "learning_rate": 1.4738027194841932e-05, - "loss": 1.4262, - "step": 1345 - }, - { - "epoch": 1.09, - "learning_rate": 1.4730331884014762e-05, - "loss": 1.394, - "step": 1346 - }, - { - "epoch": 1.09, - "learning_rate": 1.4722632962815495e-05, - "loss": 1.2745, - "step": 1347 - }, - { - "epoch": 1.09, - "learning_rate": 1.4714930437120244e-05, - "loss": 1.2861, - "step": 1348 - }, - { - "epoch": 1.09, - "learning_rate": 1.4707224312807874e-05, - "loss": 1.2129, - "step": 1349 - }, - { - "epoch": 1.09, - "learning_rate": 1.4699514595760006e-05, - "loss": 1.2161, - "step": 1350 - }, - { - "epoch": 1.09, - "learning_rate": 1.4691801291860985e-05, - "loss": 1.3331, - "step": 1351 - }, - { - "epoch": 1.09, - "learning_rate": 1.4684084406997903e-05, - "loss": 1.3173, - "step": 1352 - }, - { - "epoch": 1.09, - "learning_rate": 1.4676363947060589e-05, - "loss": 1.4099, - "step": 1353 - }, - { - "epoch": 1.1, - "learning_rate": 1.4668639917941591e-05, - "loss": 1.3756, - "step": 1354 - }, - { - "epoch": 1.1, - "learning_rate": 1.4660912325536193e-05, - "loss": 1.2639, - "step": 1355 - }, - { - "epoch": 1.1, - "learning_rate": 1.4653181175742388e-05, - "loss": 1.284, - "step": 1356 - }, - { - "epoch": 1.1, - "learning_rate": 1.4645446474460892e-05, - "loss": 1.2632, - "step": 1357 - }, - { - "epoch": 1.1, - "learning_rate": 1.463770822759512e-05, - "loss": 1.3524, - "step": 1358 - }, - { - "epoch": 1.1, - "learning_rate": 1.4629966441051208e-05, - "loss": 1.3948, - "step": 1359 - }, - { - "epoch": 1.1, - "learning_rate": 1.4622221120737985e-05, - "loss": 1.2708, - "step": 1360 - }, - { - "epoch": 1.1, - "learning_rate": 1.4614472272566982e-05, - "loss": 1.2345, - "step": 1361 - }, - { - "epoch": 1.1, - "learning_rate": 1.4606719902452413e-05, - "loss": 1.2611, - "step": 1362 - }, - { - "epoch": 1.1, - "learning_rate": 1.4598964016311188e-05, - "loss": 1.3656, - "step": 1363 - }, - { - "epoch": 1.1, - "learning_rate": 1.45912046200629e-05, - "loss": 1.361, - "step": 1364 - }, - { - "epoch": 1.1, - "learning_rate": 1.4583441719629824e-05, - "loss": 1.4177, - "step": 1365 - }, - { - "epoch": 1.11, - "learning_rate": 1.45756753209369e-05, - "loss": 1.1536, - "step": 1366 - }, - { - "epoch": 1.11, - "learning_rate": 1.4567905429911745e-05, - "loss": 1.2561, - "step": 1367 - }, - { - "epoch": 1.11, - "learning_rate": 1.4560132052484646e-05, - "loss": 1.2445, - "step": 1368 - }, - { - "epoch": 1.11, - "learning_rate": 1.4552355194588535e-05, - "loss": 1.2569, - "step": 1369 - }, - { - "epoch": 1.11, - "learning_rate": 1.4544574862159013e-05, - "loss": 1.1896, - "step": 1370 - }, - { - "epoch": 1.11, - "learning_rate": 1.4536791061134339e-05, - "loss": 1.3314, - "step": 1371 - }, - { - "epoch": 1.11, - "learning_rate": 1.4529003797455402e-05, - "loss": 1.2821, - "step": 1372 - }, - { - "epoch": 1.11, - "learning_rate": 1.4521213077065746e-05, - "loss": 1.398, - "step": 1373 - }, - { - "epoch": 1.11, - "learning_rate": 1.4513418905911546e-05, - "loss": 1.3335, - "step": 1374 - }, - { - "epoch": 1.11, - "learning_rate": 1.4505621289941619e-05, - "loss": 1.1488, - "step": 1375 - }, - { - "epoch": 1.11, - "learning_rate": 1.4497820235107405e-05, - "loss": 1.3706, - "step": 1376 - }, - { - "epoch": 1.11, - "learning_rate": 1.449001574736297e-05, - "loss": 1.1735, - "step": 1377 - }, - { - "epoch": 1.11, - "learning_rate": 1.4482207832665003e-05, - "loss": 1.2635, - "step": 1378 - }, - { - "epoch": 1.12, - "learning_rate": 1.4474396496972806e-05, - "loss": 1.3881, - "step": 1379 - }, - { - "epoch": 1.12, - "learning_rate": 1.446658174624829e-05, - "loss": 1.3339, - "step": 1380 - }, - { - "epoch": 1.12, - "learning_rate": 1.4458763586455975e-05, - "loss": 1.2122, - "step": 1381 - }, - { - "epoch": 1.12, - "learning_rate": 1.4450942023562987e-05, - "loss": 1.3406, - "step": 1382 - }, - { - "epoch": 1.12, - "learning_rate": 1.4443117063539039e-05, - "loss": 1.2246, - "step": 1383 - }, - { - "epoch": 1.12, - "learning_rate": 1.4435288712356443e-05, - "loss": 1.2415, - "step": 1384 - }, - { - "epoch": 1.12, - "learning_rate": 1.4427456975990107e-05, - "loss": 1.3344, - "step": 1385 - }, - { - "epoch": 1.12, - "learning_rate": 1.4419621860417503e-05, - "loss": 1.2786, - "step": 1386 - }, - { - "epoch": 1.12, - "learning_rate": 1.4411783371618704e-05, - "loss": 1.2828, - "step": 1387 - }, - { - "epoch": 1.12, - "learning_rate": 1.4403941515576344e-05, - "loss": 1.3637, - "step": 1388 - }, - { - "epoch": 1.12, - "learning_rate": 1.439609629827563e-05, - "loss": 1.4899, - "step": 1389 - }, - { - "epoch": 1.12, - "learning_rate": 1.4388247725704338e-05, - "loss": 1.1886, - "step": 1390 - }, - { - "epoch": 1.13, - "learning_rate": 1.4380395803852798e-05, - "loss": 1.3284, - "step": 1391 - }, - { - "epoch": 1.13, - "learning_rate": 1.43725405387139e-05, - "loss": 1.1073, - "step": 1392 - }, - { - "epoch": 1.13, - "learning_rate": 1.4364681936283089e-05, - "loss": 1.2911, - "step": 1393 - }, - { - "epoch": 1.13, - "learning_rate": 1.4356820002558354e-05, - "loss": 1.253, - "step": 1394 - }, - { - "epoch": 1.13, - "learning_rate": 1.4348954743540222e-05, - "loss": 1.2565, - "step": 1395 - }, - { - "epoch": 1.13, - "learning_rate": 1.4341086165231768e-05, - "loss": 1.5576, - "step": 1396 - }, - { - "epoch": 1.13, - "learning_rate": 1.4333214273638593e-05, - "loss": 1.2922, - "step": 1397 - }, - { - "epoch": 1.13, - "learning_rate": 1.432533907476883e-05, - "loss": 1.3993, - "step": 1398 - }, - { - "epoch": 1.13, - "learning_rate": 1.431746057463313e-05, - "loss": 1.2497, - "step": 1399 - }, - { - "epoch": 1.13, - "learning_rate": 1.4309578779244678e-05, - "loss": 1.2999, - "step": 1400 - }, - { - "epoch": 1.13, - "learning_rate": 1.4301693694619157e-05, - "loss": 1.1271, - "step": 1401 - }, - { - "epoch": 1.13, - "learning_rate": 1.4293805326774768e-05, - "loss": 1.2511, - "step": 1402 - }, - { - "epoch": 1.14, - "learning_rate": 1.4285913681732227e-05, - "loss": 1.0751, - "step": 1403 - }, - { - "epoch": 1.14, - "learning_rate": 1.4278018765514734e-05, - "loss": 1.2795, - "step": 1404 - }, - { - "epoch": 1.14, - "learning_rate": 1.4270120584147995e-05, - "loss": 1.1887, - "step": 1405 - }, - { - "epoch": 1.14, - "learning_rate": 1.426221914366021e-05, - "loss": 1.3304, - "step": 1406 - }, - { - "epoch": 1.14, - "learning_rate": 1.4254314450082061e-05, - "loss": 1.303, - "step": 1407 - }, - { - "epoch": 1.14, - "learning_rate": 1.424640650944672e-05, - "loss": 1.362, - "step": 1408 - }, - { - "epoch": 1.14, - "learning_rate": 1.423849532778983e-05, - "loss": 1.2102, - "step": 1409 - }, - { - "epoch": 1.14, - "learning_rate": 1.423058091114951e-05, - "loss": 1.4888, - "step": 1410 - }, - { - "epoch": 1.14, - "learning_rate": 1.422266326556635e-05, - "loss": 1.2955, - "step": 1411 - }, - { - "epoch": 1.14, - "learning_rate": 1.4214742397083399e-05, - "loss": 1.1469, - "step": 1412 - }, - { - "epoch": 1.14, - "learning_rate": 1.4206818311746175e-05, - "loss": 1.3351, - "step": 1413 - }, - { - "epoch": 1.14, - "learning_rate": 1.4198891015602648e-05, - "loss": 1.3655, - "step": 1414 - }, - { - "epoch": 1.14, - "learning_rate": 1.4190960514703232e-05, - "loss": 1.3834, - "step": 1415 - }, - { - "epoch": 1.15, - "learning_rate": 1.4183026815100788e-05, - "loss": 1.2687, - "step": 1416 - }, - { - "epoch": 1.15, - "learning_rate": 1.4175089922850633e-05, - "loss": 1.2112, - "step": 1417 - }, - { - "epoch": 1.15, - "learning_rate": 1.4167149844010501e-05, - "loss": 1.2796, - "step": 1418 - }, - { - "epoch": 1.15, - "learning_rate": 1.4159206584640573e-05, - "loss": 1.2547, - "step": 1419 - }, - { - "epoch": 1.15, - "learning_rate": 1.4151260150803445e-05, - "loss": 1.1697, - "step": 1420 - }, - { - "epoch": 1.15, - "learning_rate": 1.4143310548564149e-05, - "loss": 1.2586, - "step": 1421 - }, - { - "epoch": 1.15, - "learning_rate": 1.4135357783990124e-05, - "loss": 1.3323, - "step": 1422 - }, - { - "epoch": 1.15, - "learning_rate": 1.412740186315123e-05, - "loss": 1.3101, - "step": 1423 - }, - { - "epoch": 1.15, - "learning_rate": 1.4119442792119729e-05, - "loss": 1.2092, - "step": 1424 - }, - { - "epoch": 1.15, - "learning_rate": 1.4111480576970298e-05, - "loss": 1.2342, - "step": 1425 - }, - { - "epoch": 1.15, - "learning_rate": 1.4103515223780002e-05, - "loss": 1.43, - "step": 1426 - }, - { - "epoch": 1.15, - "learning_rate": 1.409554673862831e-05, - "loss": 1.2522, - "step": 1427 - }, - { - "epoch": 1.16, - "learning_rate": 1.4087575127597076e-05, - "loss": 1.3456, - "step": 1428 - }, - { - "epoch": 1.16, - "learning_rate": 1.4079600396770539e-05, - "loss": 1.3116, - "step": 1429 - }, - { - "epoch": 1.16, - "learning_rate": 1.4071622552235327e-05, - "loss": 1.2096, - "step": 1430 - }, - { - "epoch": 1.16, - "learning_rate": 1.4063641600080435e-05, - "loss": 1.3543, - "step": 1431 - }, - { - "epoch": 1.16, - "learning_rate": 1.405565754639724e-05, - "loss": 1.2252, - "step": 1432 - }, - { - "epoch": 1.16, - "learning_rate": 1.4047670397279473e-05, - "loss": 1.2572, - "step": 1433 - }, - { - "epoch": 1.16, - "learning_rate": 1.4039680158823237e-05, - "loss": 1.1978, - "step": 1434 - }, - { - "epoch": 1.16, - "learning_rate": 1.4031686837126992e-05, - "loss": 1.3075, - "step": 1435 - }, - { - "epoch": 1.16, - "learning_rate": 1.4023690438291552e-05, - "loss": 1.1352, - "step": 1436 - }, - { - "epoch": 1.16, - "learning_rate": 1.4015690968420076e-05, - "loss": 1.245, - "step": 1437 - }, - { - "epoch": 1.16, - "learning_rate": 1.4007688433618065e-05, - "loss": 1.0651, - "step": 1438 - }, - { - "epoch": 1.16, - "learning_rate": 1.3999682839993366e-05, - "loss": 1.2017, - "step": 1439 - }, - { - "epoch": 1.17, - "learning_rate": 1.399167419365616e-05, - "loss": 1.436, - "step": 1440 - }, - { - "epoch": 1.17, - "learning_rate": 1.398366250071895e-05, - "loss": 1.3677, - "step": 1441 - }, - { - "epoch": 1.17, - "learning_rate": 1.3975647767296575e-05, - "loss": 1.2367, - "step": 1442 - }, - { - "epoch": 1.17, - "learning_rate": 1.3967629999506185e-05, - "loss": 1.3621, - "step": 1443 - }, - { - "epoch": 1.17, - "learning_rate": 1.395960920346725e-05, - "loss": 1.2077, - "step": 1444 - }, - { - "epoch": 1.17, - "learning_rate": 1.3951585385301557e-05, - "loss": 1.2966, - "step": 1445 - }, - { - "epoch": 1.17, - "learning_rate": 1.3943558551133186e-05, - "loss": 1.3493, - "step": 1446 - }, - { - "epoch": 1.17, - "learning_rate": 1.3935528707088534e-05, - "loss": 1.3436, - "step": 1447 - }, - { - "epoch": 1.17, - "learning_rate": 1.3927495859296287e-05, - "loss": 1.2854, - "step": 1448 - }, - { - "epoch": 1.17, - "learning_rate": 1.391946001388742e-05, - "loss": 1.2901, - "step": 1449 - }, - { - "epoch": 1.17, - "learning_rate": 1.3911421176995206e-05, - "loss": 1.38, - "step": 1450 - }, - { - "epoch": 1.17, - "learning_rate": 1.3903379354755195e-05, - "loss": 1.1856, - "step": 1451 - }, - { - "epoch": 1.17, - "learning_rate": 1.3895334553305215e-05, - "loss": 1.25, - "step": 1452 - }, - { - "epoch": 1.18, - "learning_rate": 1.3887286778785372e-05, - "loss": 1.1983, - "step": 1453 - }, - { - "epoch": 1.18, - "learning_rate": 1.3879236037338035e-05, - "loss": 1.4348, - "step": 1454 - }, - { - "epoch": 1.18, - "learning_rate": 1.3871182335107843e-05, - "loss": 1.2935, - "step": 1455 - }, - { - "epoch": 1.18, - "learning_rate": 1.386312567824169e-05, - "loss": 1.4197, - "step": 1456 - }, - { - "epoch": 1.18, - "learning_rate": 1.3855066072888733e-05, - "loss": 1.4155, - "step": 1457 - }, - { - "epoch": 1.18, - "learning_rate": 1.3847003525200371e-05, - "loss": 1.2207, - "step": 1458 - }, - { - "epoch": 1.18, - "learning_rate": 1.3838938041330252e-05, - "loss": 1.3946, - "step": 1459 - }, - { - "epoch": 1.18, - "learning_rate": 1.3830869627434267e-05, - "loss": 1.1823, - "step": 1460 - }, - { - "epoch": 1.18, - "learning_rate": 1.3822798289670538e-05, - "loss": 1.1481, - "step": 1461 - }, - { - "epoch": 1.18, - "learning_rate": 1.3814724034199425e-05, - "loss": 1.3665, - "step": 1462 - }, - { - "epoch": 1.18, - "learning_rate": 1.3806646867183507e-05, - "loss": 1.1852, - "step": 1463 - }, - { - "epoch": 1.18, - "learning_rate": 1.3798566794787598e-05, - "loss": 1.3645, - "step": 1464 - }, - { - "epoch": 1.19, - "learning_rate": 1.3790483823178715e-05, - "loss": 1.3123, - "step": 1465 - }, - { - "epoch": 1.19, - "learning_rate": 1.3782397958526097e-05, - "loss": 1.1629, - "step": 1466 - }, - { - "epoch": 1.19, - "learning_rate": 1.3774309207001184e-05, - "loss": 1.3024, - "step": 1467 - }, - { - "epoch": 1.19, - "learning_rate": 1.3766217574777634e-05, - "loss": 1.2773, - "step": 1468 - }, - { - "epoch": 1.19, - "learning_rate": 1.3758123068031283e-05, - "loss": 1.3669, - "step": 1469 - }, - { - "epoch": 1.19, - "learning_rate": 1.3750025692940174e-05, - "loss": 1.1358, - "step": 1470 - }, - { - "epoch": 1.19, - "learning_rate": 1.374192545568454e-05, - "loss": 1.4698, - "step": 1471 - }, - { - "epoch": 1.19, - "learning_rate": 1.3733822362446791e-05, - "loss": 1.3388, - "step": 1472 - }, - { - "epoch": 1.19, - "learning_rate": 1.3725716419411523e-05, - "loss": 1.2536, - "step": 1473 - }, - { - "epoch": 1.19, - "learning_rate": 1.3717607632765506e-05, - "loss": 1.2306, - "step": 1474 - }, - { - "epoch": 1.19, - "learning_rate": 1.370949600869768e-05, - "loss": 1.3681, - "step": 1475 - }, - { - "epoch": 1.19, - "learning_rate": 1.3701381553399147e-05, - "loss": 1.1437, - "step": 1476 - }, - { - "epoch": 1.19, - "learning_rate": 1.3693264273063171e-05, - "loss": 1.2117, - "step": 1477 - }, - { - "epoch": 1.2, - "learning_rate": 1.3685144173885179e-05, - "loss": 1.1646, - "step": 1478 - }, - { - "epoch": 1.2, - "learning_rate": 1.3677021262062741e-05, - "loss": 1.3466, - "step": 1479 - }, - { - "epoch": 1.2, - "learning_rate": 1.3668895543795581e-05, - "loss": 1.205, - "step": 1480 - }, - { - "epoch": 1.2, - "learning_rate": 1.3660767025285558e-05, - "loss": 1.2733, - "step": 1481 - }, - { - "epoch": 1.2, - "learning_rate": 1.3652635712736672e-05, - "loss": 1.1806, - "step": 1482 - }, - { - "epoch": 1.2, - "learning_rate": 1.3644501612355052e-05, - "loss": 1.0976, - "step": 1483 - }, - { - "epoch": 1.2, - "learning_rate": 1.3636364730348963e-05, - "loss": 1.3505, - "step": 1484 - }, - { - "epoch": 1.2, - "learning_rate": 1.3628225072928789e-05, - "loss": 1.3497, - "step": 1485 - }, - { - "epoch": 1.2, - "learning_rate": 1.3620082646307028e-05, - "loss": 1.2189, - "step": 1486 - }, - { - "epoch": 1.2, - "learning_rate": 1.361193745669829e-05, - "loss": 1.1427, - "step": 1487 - }, - { - "epoch": 1.2, - "learning_rate": 1.3603789510319306e-05, - "loss": 1.2241, - "step": 1488 - }, - { - "epoch": 1.2, - "learning_rate": 1.35956388133889e-05, - "loss": 1.3371, - "step": 1489 - }, - { - "epoch": 1.21, - "learning_rate": 1.3587485372128e-05, - "loss": 1.317, - "step": 1490 - }, - { - "epoch": 1.21, - "learning_rate": 1.3579329192759625e-05, - "loss": 1.2237, - "step": 1491 - }, - { - "epoch": 1.21, - "learning_rate": 1.357117028150889e-05, - "loss": 1.2483, - "step": 1492 - }, - { - "epoch": 1.21, - "learning_rate": 1.3563008644602988e-05, - "loss": 1.381, - "step": 1493 - }, - { - "epoch": 1.21, - "learning_rate": 1.35548442882712e-05, - "loss": 1.2234, - "step": 1494 - }, - { - "epoch": 1.21, - "learning_rate": 1.3546677218744872e-05, - "loss": 1.1998, - "step": 1495 - }, - { - "epoch": 1.21, - "learning_rate": 1.3538507442257433e-05, - "loss": 1.2937, - "step": 1496 - }, - { - "epoch": 1.21, - "learning_rate": 1.3530334965044372e-05, - "loss": 1.2623, - "step": 1497 - }, - { - "epoch": 1.21, - "learning_rate": 1.3522159793343232e-05, - "loss": 1.4073, - "step": 1498 - }, - { - "epoch": 1.21, - "learning_rate": 1.3513981933393627e-05, - "loss": 1.2629, - "step": 1499 - }, - { - "epoch": 1.21, - "learning_rate": 1.3505801391437215e-05, - "loss": 1.2979, - "step": 1500 - }, - { - "epoch": 1.21, - "learning_rate": 1.3497618173717701e-05, - "loss": 1.3064, - "step": 1501 - }, - { - "epoch": 1.22, - "learning_rate": 1.348943228648083e-05, - "loss": 1.3723, - "step": 1502 - }, - { - "epoch": 1.22, - "learning_rate": 1.3481243735974393e-05, - "loss": 1.1934, - "step": 1503 - }, - { - "epoch": 1.22, - "learning_rate": 1.3473052528448203e-05, - "loss": 1.2988, - "step": 1504 - }, - { - "epoch": 1.22, - "learning_rate": 1.346485867015411e-05, - "loss": 1.2986, - "step": 1505 - }, - { - "epoch": 1.22, - "learning_rate": 1.3456662167345982e-05, - "loss": 1.2374, - "step": 1506 - }, - { - "epoch": 1.22, - "learning_rate": 1.3448463026279706e-05, - "loss": 1.474, - "step": 1507 - }, - { - "epoch": 1.22, - "learning_rate": 1.3440261253213185e-05, - "loss": 1.2753, - "step": 1508 - }, - { - "epoch": 1.22, - "learning_rate": 1.3432056854406326e-05, - "loss": 1.2408, - "step": 1509 - }, - { - "epoch": 1.22, - "learning_rate": 1.3423849836121043e-05, - "loss": 1.2734, - "step": 1510 - }, - { - "epoch": 1.22, - "learning_rate": 1.3415640204621254e-05, - "loss": 1.2643, - "step": 1511 - }, - { - "epoch": 1.22, - "learning_rate": 1.3407427966172866e-05, - "loss": 1.325, - "step": 1512 - }, - { - "epoch": 1.22, - "learning_rate": 1.3399213127043776e-05, - "loss": 1.3092, - "step": 1513 - }, - { - "epoch": 1.22, - "learning_rate": 1.3390995693503862e-05, - "loss": 1.0985, - "step": 1514 - }, - { - "epoch": 1.23, - "learning_rate": 1.3382775671824995e-05, - "loss": 1.3207, - "step": 1515 - }, - { - "epoch": 1.23, - "learning_rate": 1.3374553068281005e-05, - "loss": 1.428, - "step": 1516 - }, - { - "epoch": 1.23, - "learning_rate": 1.3366327889147708e-05, - "loss": 1.3639, - "step": 1517 - }, - { - "epoch": 1.23, - "learning_rate": 1.3358100140702876e-05, - "loss": 1.1925, - "step": 1518 - }, - { - "epoch": 1.23, - "learning_rate": 1.3349869829226244e-05, - "loss": 1.1085, - "step": 1519 - }, - { - "epoch": 1.23, - "learning_rate": 1.33416369609995e-05, - "loss": 1.2113, - "step": 1520 - }, - { - "epoch": 1.23, - "learning_rate": 1.3333401542306293e-05, - "loss": 1.2529, - "step": 1521 - }, - { - "epoch": 1.23, - "learning_rate": 1.3325163579432211e-05, - "loss": 1.2183, - "step": 1522 - }, - { - "epoch": 1.23, - "learning_rate": 1.3316923078664786e-05, - "loss": 1.3315, - "step": 1523 - }, - { - "epoch": 1.23, - "learning_rate": 1.3308680046293487e-05, - "loss": 1.3445, - "step": 1524 - }, - { - "epoch": 1.23, - "learning_rate": 1.3300434488609714e-05, - "loss": 1.3047, - "step": 1525 - }, - { - "epoch": 1.23, - "learning_rate": 1.3292186411906794e-05, - "loss": 1.1564, - "step": 1526 - }, - { - "epoch": 1.24, - "learning_rate": 1.3283935822479977e-05, - "loss": 1.2511, - "step": 1527 - }, - { - "epoch": 1.24, - "learning_rate": 1.3275682726626436e-05, - "loss": 1.2258, - "step": 1528 - }, - { - "epoch": 1.24, - "learning_rate": 1.3267427130645252e-05, - "loss": 1.2621, - "step": 1529 - }, - { - "epoch": 1.24, - "learning_rate": 1.325916904083741e-05, - "loss": 1.243, - "step": 1530 - }, - { - "epoch": 1.24, - "learning_rate": 1.3250908463505808e-05, - "loss": 1.3881, - "step": 1531 - }, - { - "epoch": 1.24, - "learning_rate": 1.3242645404955236e-05, - "loss": 1.1725, - "step": 1532 - }, - { - "epoch": 1.24, - "learning_rate": 1.3234379871492381e-05, - "loss": 1.2794, - "step": 1533 - }, - { - "epoch": 1.24, - "learning_rate": 1.3226111869425817e-05, - "loss": 1.5134, - "step": 1534 - }, - { - "epoch": 1.24, - "learning_rate": 1.3217841405066003e-05, - "loss": 1.2675, - "step": 1535 - }, - { - "epoch": 1.24, - "learning_rate": 1.3209568484725275e-05, - "loss": 1.3483, - "step": 1536 - }, - { - "epoch": 1.24, - "learning_rate": 1.3201293114717847e-05, - "loss": 1.2615, - "step": 1537 - }, - { - "epoch": 1.24, - "learning_rate": 1.31930153013598e-05, - "loss": 1.2191, - "step": 1538 - }, - { - "epoch": 1.25, - "learning_rate": 1.3184735050969085e-05, - "loss": 1.3458, - "step": 1539 - }, - { - "epoch": 1.25, - "learning_rate": 1.3176452369865504e-05, - "loss": 1.3313, - "step": 1540 - }, - { - "epoch": 1.25, - "learning_rate": 1.3168167264370721e-05, - "loss": 1.284, - "step": 1541 - }, - { - "epoch": 1.25, - "learning_rate": 1.315987974080825e-05, - "loss": 1.169, - "step": 1542 - }, - { - "epoch": 1.25, - "learning_rate": 1.3151589805503448e-05, - "loss": 1.2868, - "step": 1543 - }, - { - "epoch": 1.25, - "learning_rate": 1.3143297464783516e-05, - "loss": 1.2795, - "step": 1544 - }, - { - "epoch": 1.25, - "learning_rate": 1.3135002724977486e-05, - "loss": 1.1662, - "step": 1545 - }, - { - "epoch": 1.25, - "learning_rate": 1.3126705592416227e-05, - "loss": 1.2601, - "step": 1546 - }, - { - "epoch": 1.25, - "learning_rate": 1.3118406073432426e-05, - "loss": 1.3559, - "step": 1547 - }, - { - "epoch": 1.25, - "learning_rate": 1.31101041743606e-05, - "loss": 1.1709, - "step": 1548 - }, - { - "epoch": 1.25, - "learning_rate": 1.3101799901537082e-05, - "loss": 1.2765, - "step": 1549 - }, - { - "epoch": 1.25, - "learning_rate": 1.3093493261300012e-05, - "loss": 1.31, - "step": 1550 - }, - { - "epoch": 1.25, - "learning_rate": 1.3085184259989336e-05, - "loss": 1.1647, - "step": 1551 - }, - { - "epoch": 1.26, - "learning_rate": 1.3076872903946807e-05, - "loss": 1.3096, - "step": 1552 - }, - { - "epoch": 1.26, - "learning_rate": 1.3068559199515973e-05, - "loss": 1.1986, - "step": 1553 - }, - { - "epoch": 1.26, - "learning_rate": 1.3060243153042176e-05, - "loss": 1.2067, - "step": 1554 - }, - { - "epoch": 1.26, - "learning_rate": 1.3051924770872542e-05, - "loss": 1.4188, - "step": 1555 - }, - { - "epoch": 1.26, - "learning_rate": 1.304360405935598e-05, - "loss": 1.2241, - "step": 1556 - }, - { - "epoch": 1.26, - "learning_rate": 1.3035281024843185e-05, - "loss": 1.4098, - "step": 1557 - }, - { - "epoch": 1.26, - "learning_rate": 1.302695567368661e-05, - "loss": 1.483, - "step": 1558 - }, - { - "epoch": 1.26, - "learning_rate": 1.3018628012240493e-05, - "loss": 1.1673, - "step": 1559 - }, - { - "epoch": 1.26, - "learning_rate": 1.3010298046860821e-05, - "loss": 1.2976, - "step": 1560 - }, - { - "epoch": 1.26, - "learning_rate": 1.300196578390535e-05, - "loss": 1.2414, - "step": 1561 - }, - { - "epoch": 1.26, - "learning_rate": 1.2993631229733584e-05, - "loss": 1.2727, - "step": 1562 - }, - { - "epoch": 1.26, - "learning_rate": 1.298529439070677e-05, - "loss": 1.2951, - "step": 1563 - }, - { - "epoch": 1.27, - "learning_rate": 1.2976955273187914e-05, - "loss": 1.2318, - "step": 1564 - }, - { - "epoch": 1.27, - "learning_rate": 1.2968613883541752e-05, - "loss": 1.3143, - "step": 1565 - }, - { - "epoch": 1.27, - "learning_rate": 1.296027022813475e-05, - "loss": 1.2364, - "step": 1566 - }, - { - "epoch": 1.27, - "learning_rate": 1.2951924313335113e-05, - "loss": 1.3078, - "step": 1567 - }, - { - "epoch": 1.27, - "learning_rate": 1.2943576145512766e-05, - "loss": 1.2174, - "step": 1568 - }, - { - "epoch": 1.27, - "learning_rate": 1.2935225731039349e-05, - "loss": 1.2376, - "step": 1569 - }, - { - "epoch": 1.27, - "learning_rate": 1.2926873076288222e-05, - "loss": 1.3129, - "step": 1570 - }, - { - "epoch": 1.27, - "learning_rate": 1.2918518187634458e-05, - "loss": 1.4206, - "step": 1571 - }, - { - "epoch": 1.27, - "learning_rate": 1.291016107145483e-05, - "loss": 1.4149, - "step": 1572 - }, - { - "epoch": 1.27, - "learning_rate": 1.2901801734127806e-05, - "loss": 1.2031, - "step": 1573 - }, - { - "epoch": 1.27, - "learning_rate": 1.2893440182033559e-05, - "loss": 1.3816, - "step": 1574 - }, - { - "epoch": 1.27, - "learning_rate": 1.2885076421553953e-05, - "loss": 1.3969, - "step": 1575 - }, - { - "epoch": 1.28, - "learning_rate": 1.2876710459072527e-05, - "loss": 1.3869, - "step": 1576 - }, - { - "epoch": 1.28, - "learning_rate": 1.2868342300974508e-05, - "loss": 1.1395, - "step": 1577 - }, - { - "epoch": 1.28, - "learning_rate": 1.2859971953646802e-05, - "loss": 1.3634, - "step": 1578 - }, - { - "epoch": 1.28, - "learning_rate": 1.2851599423477974e-05, - "loss": 1.2956, - "step": 1579 - }, - { - "epoch": 1.28, - "learning_rate": 1.2843224716858271e-05, - "loss": 1.3105, - "step": 1580 - }, - { - "epoch": 1.28, - "learning_rate": 1.2834847840179586e-05, - "loss": 1.4745, - "step": 1581 - }, - { - "epoch": 1.28, - "learning_rate": 1.2826468799835481e-05, - "loss": 1.2004, - "step": 1582 - }, - { - "epoch": 1.28, - "learning_rate": 1.2818087602221158e-05, - "loss": 1.4127, - "step": 1583 - }, - { - "epoch": 1.28, - "learning_rate": 1.2809704253733473e-05, - "loss": 1.3166, - "step": 1584 - }, - { - "epoch": 1.28, - "learning_rate": 1.280131876077092e-05, - "loss": 1.3878, - "step": 1585 - }, - { - "epoch": 1.28, - "learning_rate": 1.2792931129733637e-05, - "loss": 1.3171, - "step": 1586 - }, - { - "epoch": 1.28, - "learning_rate": 1.2784541367023382e-05, - "loss": 1.4039, - "step": 1587 - }, - { - "epoch": 1.28, - "learning_rate": 1.2776149479043544e-05, - "loss": 1.2209, - "step": 1588 - }, - { - "epoch": 1.29, - "learning_rate": 1.2767755472199145e-05, - "loss": 1.1777, - "step": 1589 - }, - { - "epoch": 1.29, - "learning_rate": 1.2759359352896809e-05, - "loss": 1.2394, - "step": 1590 - }, - { - "epoch": 1.29, - "learning_rate": 1.2750961127544782e-05, - "loss": 1.1719, - "step": 1591 - }, - { - "epoch": 1.29, - "learning_rate": 1.2742560802552912e-05, - "loss": 1.3132, - "step": 1592 - }, - { - "epoch": 1.29, - "learning_rate": 1.273415838433265e-05, - "loss": 1.3201, - "step": 1593 - }, - { - "epoch": 1.29, - "learning_rate": 1.2725753879297053e-05, - "loss": 1.5524, - "step": 1594 - }, - { - "epoch": 1.29, - "learning_rate": 1.2717347293860755e-05, - "loss": 1.2224, - "step": 1595 - }, - { - "epoch": 1.29, - "learning_rate": 1.2708938634439994e-05, - "loss": 1.2818, - "step": 1596 - }, - { - "epoch": 1.29, - "learning_rate": 1.270052790745258e-05, - "loss": 1.2856, - "step": 1597 - }, - { - "epoch": 1.29, - "learning_rate": 1.2692115119317907e-05, - "loss": 1.3504, - "step": 1598 - }, - { - "epoch": 1.29, - "learning_rate": 1.2683700276456938e-05, - "loss": 1.4196, - "step": 1599 - }, - { - "epoch": 1.29, - "learning_rate": 1.2675283385292212e-05, - "loss": 1.3761, - "step": 1600 - }, - { - "epoch": 1.3, - "learning_rate": 1.2666864452247816e-05, - "loss": 1.1733, - "step": 1601 - }, - { - "epoch": 1.3, - "learning_rate": 1.265844348374941e-05, - "loss": 1.2145, - "step": 1602 - }, - { - "epoch": 1.3, - "learning_rate": 1.2650020486224204e-05, - "loss": 1.313, - "step": 1603 - }, - { - "epoch": 1.3, - "learning_rate": 1.2641595466100952e-05, - "loss": 1.1817, - "step": 1604 - }, - { - "epoch": 1.3, - "learning_rate": 1.263316842980996e-05, - "loss": 1.2449, - "step": 1605 - }, - { - "epoch": 1.3, - "learning_rate": 1.2624739383783064e-05, - "loss": 1.3036, - "step": 1606 - }, - { - "epoch": 1.3, - "learning_rate": 1.2616308334453636e-05, - "loss": 1.2257, - "step": 1607 - }, - { - "epoch": 1.3, - "learning_rate": 1.2607875288256584e-05, - "loss": 1.1835, - "step": 1608 - }, - { - "epoch": 1.3, - "learning_rate": 1.2599440251628328e-05, - "loss": 1.2733, - "step": 1609 - }, - { - "epoch": 1.3, - "learning_rate": 1.259100323100682e-05, - "loss": 1.3295, - "step": 1610 - }, - { - "epoch": 1.3, - "learning_rate": 1.2582564232831517e-05, - "loss": 1.234, - "step": 1611 - }, - { - "epoch": 1.3, - "learning_rate": 1.2574123263543387e-05, - "loss": 1.3011, - "step": 1612 - }, - { - "epoch": 1.31, - "learning_rate": 1.2565680329584908e-05, - "loss": 1.2249, - "step": 1613 - }, - { - "epoch": 1.31, - "learning_rate": 1.2557235437400053e-05, - "loss": 1.3021, - "step": 1614 - }, - { - "epoch": 1.31, - "learning_rate": 1.254878859343429e-05, - "loss": 1.3101, - "step": 1615 - }, - { - "epoch": 1.31, - "learning_rate": 1.2540339804134572e-05, - "loss": 1.1994, - "step": 1616 - }, - { - "epoch": 1.31, - "learning_rate": 1.2531889075949344e-05, - "loss": 1.3648, - "step": 1617 - }, - { - "epoch": 1.31, - "learning_rate": 1.2523436415328534e-05, - "loss": 1.2115, - "step": 1618 - }, - { - "epoch": 1.31, - "learning_rate": 1.2514981828723531e-05, - "loss": 1.3502, - "step": 1619 - }, - { - "epoch": 1.31, - "learning_rate": 1.2506525322587207e-05, - "loss": 1.177, - "step": 1620 - }, - { - "epoch": 1.31, - "learning_rate": 1.2498066903373893e-05, - "loss": 1.3265, - "step": 1621 - }, - { - "epoch": 1.31, - "learning_rate": 1.2489606577539379e-05, - "loss": 1.1701, - "step": 1622 - }, - { - "epoch": 1.31, - "learning_rate": 1.2481144351540916e-05, - "loss": 1.265, - "step": 1623 - }, - { - "epoch": 1.31, - "learning_rate": 1.2472680231837204e-05, - "loss": 1.4224, - "step": 1624 - }, - { - "epoch": 1.31, - "learning_rate": 1.2464214224888384e-05, - "loss": 1.295, - "step": 1625 - }, - { - "epoch": 1.32, - "learning_rate": 1.2455746337156036e-05, - "loss": 1.2511, - "step": 1626 - }, - { - "epoch": 1.32, - "learning_rate": 1.2447276575103182e-05, - "loss": 1.4193, - "step": 1627 - }, - { - "epoch": 1.32, - "learning_rate": 1.2438804945194272e-05, - "loss": 1.4153, - "step": 1628 - }, - { - "epoch": 1.32, - "learning_rate": 1.2430331453895182e-05, - "loss": 1.1959, - "step": 1629 - }, - { - "epoch": 1.32, - "learning_rate": 1.2421856107673205e-05, - "loss": 1.3909, - "step": 1630 - }, - { - "epoch": 1.32, - "learning_rate": 1.2413378912997058e-05, - "loss": 1.2553, - "step": 1631 - }, - { - "epoch": 1.32, - "learning_rate": 1.240489987633686e-05, - "loss": 1.1953, - "step": 1632 - }, - { - "epoch": 1.32, - "learning_rate": 1.2396419004164137e-05, - "loss": 1.4734, - "step": 1633 - }, - { - "epoch": 1.32, - "learning_rate": 1.2387936302951822e-05, - "loss": 1.3504, - "step": 1634 - }, - { - "epoch": 1.32, - "learning_rate": 1.2379451779174243e-05, - "loss": 1.3828, - "step": 1635 - }, - { - "epoch": 1.32, - "learning_rate": 1.237096543930711e-05, - "loss": 1.1549, - "step": 1636 - }, - { - "epoch": 1.32, - "learning_rate": 1.236247728982753e-05, - "loss": 1.3343, - "step": 1637 - }, - { - "epoch": 1.33, - "learning_rate": 1.2353987337213984e-05, - "loss": 1.2328, - "step": 1638 - }, - { - "epoch": 1.33, - "learning_rate": 1.2345495587946331e-05, - "loss": 1.3521, - "step": 1639 - }, - { - "epoch": 1.33, - "learning_rate": 1.233700204850581e-05, - "loss": 1.2576, - "step": 1640 - }, - { - "epoch": 1.33, - "learning_rate": 1.232850672537501e-05, - "loss": 1.3657, - "step": 1641 - }, - { - "epoch": 1.33, - "learning_rate": 1.2320009625037894e-05, - "loss": 1.2928, - "step": 1642 - }, - { - "epoch": 1.33, - "learning_rate": 1.2311510753979776e-05, - "loss": 1.3001, - "step": 1643 - }, - { - "epoch": 1.33, - "learning_rate": 1.2303010118687322e-05, - "loss": 1.3022, - "step": 1644 - }, - { - "epoch": 1.33, - "learning_rate": 1.2294507725648545e-05, - "loss": 1.3075, - "step": 1645 - }, - { - "epoch": 1.33, - "learning_rate": 1.2286003581352805e-05, - "loss": 1.3915, - "step": 1646 - }, - { - "epoch": 1.33, - "learning_rate": 1.227749769229079e-05, - "loss": 1.2422, - "step": 1647 - }, - { - "epoch": 1.33, - "learning_rate": 1.226899006495452e-05, - "loss": 1.4356, - "step": 1648 - }, - { - "epoch": 1.33, - "learning_rate": 1.226048070583735e-05, - "loss": 1.149, - "step": 1649 - }, - { - "epoch": 1.33, - "learning_rate": 1.2251969621433947e-05, - "loss": 1.2658, - "step": 1650 - }, - { - "epoch": 1.34, - "learning_rate": 1.2243456818240307e-05, - "loss": 1.2499, - "step": 1651 - }, - { - "epoch": 1.34, - "learning_rate": 1.2234942302753723e-05, - "loss": 1.2894, - "step": 1652 - }, - { - "epoch": 1.34, - "learning_rate": 1.2226426081472806e-05, - "loss": 1.1907, - "step": 1653 - }, - { - "epoch": 1.34, - "learning_rate": 1.2217908160897465e-05, - "loss": 1.1893, - "step": 1654 - }, - { - "epoch": 1.34, - "learning_rate": 1.2209388547528901e-05, - "loss": 1.2255, - "step": 1655 - }, - { - "epoch": 1.34, - "learning_rate": 1.2200867247869621e-05, - "loss": 1.1617, - "step": 1656 - }, - { - "epoch": 1.34, - "learning_rate": 1.2192344268423407e-05, - "loss": 1.3943, - "step": 1657 - }, - { - "epoch": 1.34, - "learning_rate": 1.2183819615695322e-05, - "loss": 1.41, - "step": 1658 - }, - { - "epoch": 1.34, - "learning_rate": 1.2175293296191712e-05, - "loss": 1.3122, - "step": 1659 - }, - { - "epoch": 1.34, - "learning_rate": 1.2166765316420195e-05, - "loss": 1.333, - "step": 1660 - }, - { - "epoch": 1.34, - "learning_rate": 1.2158235682889659e-05, - "loss": 1.3695, - "step": 1661 - }, - { - "epoch": 1.34, - "learning_rate": 1.2149704402110243e-05, - "loss": 1.1909, - "step": 1662 - }, - { - "epoch": 1.35, - "learning_rate": 1.2141171480593352e-05, - "loss": 1.2625, - "step": 1663 - }, - { - "epoch": 1.35, - "learning_rate": 1.2132636924851644e-05, - "loss": 1.2284, - "step": 1664 - }, - { - "epoch": 1.35, - "learning_rate": 1.2124100741399016e-05, - "loss": 1.4324, - "step": 1665 - }, - { - "epoch": 1.35, - "learning_rate": 1.211556293675062e-05, - "loss": 1.3858, - "step": 1666 - }, - { - "epoch": 1.35, - "learning_rate": 1.2107023517422836e-05, - "loss": 1.1409, - "step": 1667 - }, - { - "epoch": 1.35, - "learning_rate": 1.209848248993328e-05, - "loss": 1.2336, - "step": 1668 - }, - { - "epoch": 1.35, - "learning_rate": 1.208993986080079e-05, - "loss": 1.2898, - "step": 1669 - }, - { - "epoch": 1.35, - "learning_rate": 1.2081395636545432e-05, - "loss": 1.1649, - "step": 1670 - }, - { - "epoch": 1.35, - "learning_rate": 1.207284982368849e-05, - "loss": 1.2317, - "step": 1671 - }, - { - "epoch": 1.35, - "learning_rate": 1.2064302428752461e-05, - "loss": 1.3948, - "step": 1672 - }, - { - "epoch": 1.35, - "learning_rate": 1.2055753458261042e-05, - "loss": 1.2319, - "step": 1673 - }, - { - "epoch": 1.35, - "learning_rate": 1.204720291873914e-05, - "loss": 1.275, - "step": 1674 - }, - { - "epoch": 1.36, - "learning_rate": 1.2038650816712858e-05, - "loss": 1.2867, - "step": 1675 - }, - { - "epoch": 1.36, - "learning_rate": 1.2030097158709487e-05, - "loss": 1.173, - "step": 1676 - }, - { - "epoch": 1.36, - "learning_rate": 1.2021541951257514e-05, - "loss": 1.3089, - "step": 1677 - }, - { - "epoch": 1.36, - "learning_rate": 1.2012985200886602e-05, - "loss": 1.1743, - "step": 1678 - }, - { - "epoch": 1.36, - "learning_rate": 1.2004426914127593e-05, - "loss": 1.2601, - "step": 1679 - }, - { - "epoch": 1.36, - "learning_rate": 1.1995867097512504e-05, - "loss": 1.0967, - "step": 1680 - }, - { - "epoch": 1.36, - "learning_rate": 1.1987305757574516e-05, - "loss": 1.3141, - "step": 1681 - }, - { - "epoch": 1.36, - "learning_rate": 1.1978742900847975e-05, - "loss": 1.3616, - "step": 1682 - }, - { - "epoch": 1.36, - "learning_rate": 1.1970178533868384e-05, - "loss": 1.3077, - "step": 1683 - }, - { - "epoch": 1.36, - "learning_rate": 1.19616126631724e-05, - "loss": 1.3554, - "step": 1684 - }, - { - "epoch": 1.36, - "learning_rate": 1.1953045295297826e-05, - "loss": 1.3538, - "step": 1685 - }, - { - "epoch": 1.36, - "learning_rate": 1.1944476436783608e-05, - "loss": 1.2264, - "step": 1686 - }, - { - "epoch": 1.36, - "learning_rate": 1.1935906094169827e-05, - "loss": 1.5312, - "step": 1687 - }, - { - "epoch": 1.37, - "learning_rate": 1.1927334273997704e-05, - "loss": 1.3409, - "step": 1688 - }, - { - "epoch": 1.37, - "learning_rate": 1.1918760982809583e-05, - "loss": 1.307, - "step": 1689 - }, - { - "epoch": 1.37, - "learning_rate": 1.191018622714893e-05, - "loss": 1.264, - "step": 1690 - }, - { - "epoch": 1.37, - "learning_rate": 1.1901610013560327e-05, - "loss": 1.3514, - "step": 1691 - }, - { - "epoch": 1.37, - "learning_rate": 1.189303234858948e-05, - "loss": 1.3729, - "step": 1692 - }, - { - "epoch": 1.37, - "learning_rate": 1.1884453238783185e-05, - "loss": 1.2775, - "step": 1693 - }, - { - "epoch": 1.37, - "learning_rate": 1.1875872690689357e-05, - "loss": 1.3189, - "step": 1694 - }, - { - "epoch": 1.37, - "learning_rate": 1.1867290710856993e-05, - "loss": 1.5188, - "step": 1695 - }, - { - "epoch": 1.37, - "learning_rate": 1.1858707305836204e-05, - "loss": 1.3531, - "step": 1696 - }, - { - "epoch": 1.37, - "learning_rate": 1.185012248217817e-05, - "loss": 1.0528, - "step": 1697 - }, - { - "epoch": 1.37, - "learning_rate": 1.1841536246435154e-05, - "loss": 1.192, - "step": 1698 - }, - { - "epoch": 1.37, - "learning_rate": 1.1832948605160515e-05, - "loss": 1.418, - "step": 1699 - }, - { - "epoch": 1.38, - "learning_rate": 1.1824359564908667e-05, - "loss": 1.1702, - "step": 1700 - }, - { - "epoch": 1.38, - "learning_rate": 1.1815769132235096e-05, - "loss": 1.2854, - "step": 1701 - }, - { - "epoch": 1.38, - "learning_rate": 1.1807177313696351e-05, - "loss": 1.2736, - "step": 1702 - }, - { - "epoch": 1.38, - "learning_rate": 1.1798584115850045e-05, - "loss": 1.2223, - "step": 1703 - }, - { - "epoch": 1.38, - "learning_rate": 1.1789989545254836e-05, - "loss": 1.4509, - "step": 1704 - }, - { - "epoch": 1.38, - "learning_rate": 1.1781393608470432e-05, - "loss": 1.1913, - "step": 1705 - }, - { - "epoch": 1.38, - "learning_rate": 1.1772796312057586e-05, - "loss": 1.323, - "step": 1706 - }, - { - "epoch": 1.38, - "learning_rate": 1.1764197662578087e-05, - "loss": 1.2922, - "step": 1707 - }, - { - "epoch": 1.38, - "learning_rate": 1.1755597666594751e-05, - "loss": 1.2594, - "step": 1708 - }, - { - "epoch": 1.38, - "learning_rate": 1.1746996330671432e-05, - "loss": 1.2521, - "step": 1709 - }, - { - "epoch": 1.38, - "learning_rate": 1.1738393661373004e-05, - "loss": 1.2561, - "step": 1710 - }, - { - "epoch": 1.38, - "learning_rate": 1.1729789665265354e-05, - "loss": 1.3669, - "step": 1711 - }, - { - "epoch": 1.39, - "learning_rate": 1.1721184348915384e-05, - "loss": 1.203, - "step": 1712 - }, - { - "epoch": 1.39, - "learning_rate": 1.1712577718891003e-05, - "loss": 1.2659, - "step": 1713 - }, - { - "epoch": 1.39, - "learning_rate": 1.1703969781761124e-05, - "loss": 1.2732, - "step": 1714 - }, - { - "epoch": 1.39, - "learning_rate": 1.1695360544095661e-05, - "loss": 1.1817, - "step": 1715 - }, - { - "epoch": 1.39, - "learning_rate": 1.1686750012465512e-05, - "loss": 1.2933, - "step": 1716 - }, - { - "epoch": 1.39, - "learning_rate": 1.1678138193442572e-05, - "loss": 1.4199, - "step": 1717 - }, - { - "epoch": 1.39, - "learning_rate": 1.1669525093599711e-05, - "loss": 1.2217, - "step": 1718 - }, - { - "epoch": 1.39, - "learning_rate": 1.1660910719510778e-05, - "loss": 1.34, - "step": 1719 - }, - { - "epoch": 1.39, - "learning_rate": 1.1652295077750599e-05, - "loss": 1.1909, - "step": 1720 - }, - { - "epoch": 1.39, - "learning_rate": 1.1643678174894968e-05, - "loss": 1.2131, - "step": 1721 - }, - { - "epoch": 1.39, - "learning_rate": 1.1635060017520635e-05, - "loss": 1.3318, - "step": 1722 - }, - { - "epoch": 1.39, - "learning_rate": 1.1626440612205306e-05, - "loss": 1.3549, - "step": 1723 - }, - { - "epoch": 1.39, - "learning_rate": 1.161781996552765e-05, - "loss": 1.2762, - "step": 1724 - }, - { - "epoch": 1.4, - "learning_rate": 1.1609198084067275e-05, - "loss": 1.2308, - "step": 1725 - }, - { - "epoch": 1.4, - "learning_rate": 1.1600574974404736e-05, - "loss": 1.215, - "step": 1726 - }, - { - "epoch": 1.4, - "learning_rate": 1.1591950643121524e-05, - "loss": 1.3206, - "step": 1727 - }, - { - "epoch": 1.4, - "learning_rate": 1.1583325096800062e-05, - "loss": 1.2189, - "step": 1728 - }, - { - "epoch": 1.4, - "learning_rate": 1.1574698342023699e-05, - "loss": 1.3781, - "step": 1729 - }, - { - "epoch": 1.4, - "learning_rate": 1.1566070385376705e-05, - "loss": 1.2022, - "step": 1730 - }, - { - "epoch": 1.4, - "learning_rate": 1.155744123344427e-05, - "loss": 1.1843, - "step": 1731 - }, - { - "epoch": 1.4, - "learning_rate": 1.1548810892812506e-05, - "loss": 1.446, - "step": 1732 - }, - { - "epoch": 1.4, - "learning_rate": 1.1540179370068411e-05, - "loss": 1.207, - "step": 1733 - }, - { - "epoch": 1.4, - "learning_rate": 1.1531546671799903e-05, - "loss": 1.2482, - "step": 1734 - }, - { - "epoch": 1.4, - "learning_rate": 1.1522912804595789e-05, - "loss": 1.2068, - "step": 1735 - }, - { - "epoch": 1.4, - "learning_rate": 1.1514277775045768e-05, - "loss": 1.342, - "step": 1736 - }, - { - "epoch": 1.41, - "learning_rate": 1.1505641589740431e-05, - "loss": 1.17, - "step": 1737 - }, - { - "epoch": 1.41, - "learning_rate": 1.1497004255271249e-05, - "loss": 1.1938, - "step": 1738 - }, - { - "epoch": 1.41, - "learning_rate": 1.1488365778230568e-05, - "loss": 1.3173, - "step": 1739 - }, - { - "epoch": 1.41, - "learning_rate": 1.1479726165211609e-05, - "loss": 1.176, - "step": 1740 - }, - { - "epoch": 1.41, - "learning_rate": 1.1471085422808452e-05, - "loss": 1.2527, - "step": 1741 - }, - { - "epoch": 1.41, - "learning_rate": 1.1462443557616056e-05, - "loss": 1.3521, - "step": 1742 - }, - { - "epoch": 1.41, - "learning_rate": 1.1453800576230221e-05, - "loss": 1.1173, - "step": 1743 - }, - { - "epoch": 1.41, - "learning_rate": 1.1445156485247604e-05, - "loss": 1.2163, - "step": 1744 - }, - { - "epoch": 1.41, - "learning_rate": 1.1436511291265706e-05, - "loss": 1.1429, - "step": 1745 - }, - { - "epoch": 1.41, - "learning_rate": 1.1427865000882881e-05, - "loss": 1.1757, - "step": 1746 - }, - { - "epoch": 1.41, - "learning_rate": 1.1419217620698309e-05, - "loss": 1.374, - "step": 1747 - }, - { - "epoch": 1.41, - "learning_rate": 1.1410569157312e-05, - "loss": 1.3696, - "step": 1748 - }, - { - "epoch": 1.42, - "learning_rate": 1.1401919617324802e-05, - "loss": 1.3353, - "step": 1749 - }, - { - "epoch": 1.42, - "learning_rate": 1.1393269007338375e-05, - "loss": 1.333, - "step": 1750 - }, - { - "epoch": 1.42, - "learning_rate": 1.1384617333955197e-05, - "loss": 1.3202, - "step": 1751 - }, - { - "epoch": 1.42, - "learning_rate": 1.1375964603778562e-05, - "loss": 1.2973, - "step": 1752 - }, - { - "epoch": 1.42, - "learning_rate": 1.1367310823412572e-05, - "loss": 1.2942, - "step": 1753 - }, - { - "epoch": 1.42, - "learning_rate": 1.1358655999462119e-05, - "loss": 1.3389, - "step": 1754 - }, - { - "epoch": 1.42, - "learning_rate": 1.1350000138532902e-05, - "loss": 1.2647, - "step": 1755 - }, - { - "epoch": 1.42, - "learning_rate": 1.1341343247231411e-05, - "loss": 1.436, - "step": 1756 - }, - { - "epoch": 1.42, - "learning_rate": 1.1332685332164917e-05, - "loss": 1.1815, - "step": 1757 - }, - { - "epoch": 1.42, - "learning_rate": 1.1324026399941476e-05, - "loss": 1.3243, - "step": 1758 - }, - { - "epoch": 1.42, - "learning_rate": 1.1315366457169918e-05, - "loss": 1.3291, - "step": 1759 - }, - { - "epoch": 1.42, - "learning_rate": 1.1306705510459852e-05, - "loss": 1.3163, - "step": 1760 - }, - { - "epoch": 1.42, - "learning_rate": 1.1298043566421644e-05, - "loss": 1.1549, - "step": 1761 - }, - { - "epoch": 1.43, - "learning_rate": 1.128938063166642e-05, - "loss": 1.3216, - "step": 1762 - }, - { - "epoch": 1.43, - "learning_rate": 1.128071671280607e-05, - "loss": 1.2101, - "step": 1763 - }, - { - "epoch": 1.43, - "learning_rate": 1.1272051816453233e-05, - "loss": 1.1888, - "step": 1764 - }, - { - "epoch": 1.43, - "learning_rate": 1.1263385949221294e-05, - "loss": 1.2364, - "step": 1765 - }, - { - "epoch": 1.43, - "learning_rate": 1.1254719117724374e-05, - "loss": 1.1801, - "step": 1766 - }, - { - "epoch": 1.43, - "learning_rate": 1.1246051328577335e-05, - "loss": 1.3838, - "step": 1767 - }, - { - "epoch": 1.43, - "learning_rate": 1.1237382588395766e-05, - "loss": 1.4147, - "step": 1768 - }, - { - "epoch": 1.43, - "learning_rate": 1.1228712903795994e-05, - "loss": 1.2246, - "step": 1769 - }, - { - "epoch": 1.43, - "learning_rate": 1.1220042281395042e-05, - "loss": 1.192, - "step": 1770 - }, - { - "epoch": 1.43, - "learning_rate": 1.121137072781068e-05, - "loss": 1.0873, - "step": 1771 - }, - { - "epoch": 1.43, - "learning_rate": 1.1202698249661364e-05, - "loss": 1.2984, - "step": 1772 - }, - { - "epoch": 1.43, - "learning_rate": 1.1194024853566264e-05, - "loss": 1.309, - "step": 1773 - }, - { - "epoch": 1.44, - "learning_rate": 1.1185350546145253e-05, - "loss": 1.2296, - "step": 1774 - }, - { - "epoch": 1.44, - "learning_rate": 1.11766753340189e-05, - "loss": 1.2463, - "step": 1775 - }, - { - "epoch": 1.44, - "learning_rate": 1.1167999223808459e-05, - "loss": 1.2479, - "step": 1776 - }, - { - "epoch": 1.44, - "learning_rate": 1.1159322222135873e-05, - "loss": 1.1834, - "step": 1777 - }, - { - "epoch": 1.44, - "learning_rate": 1.1150644335623766e-05, - "loss": 1.3382, - "step": 1778 - }, - { - "epoch": 1.44, - "learning_rate": 1.1141965570895435e-05, - "loss": 1.1867, - "step": 1779 - }, - { - "epoch": 1.44, - "learning_rate": 1.1133285934574849e-05, - "loss": 1.2344, - "step": 1780 - }, - { - "epoch": 1.44, - "learning_rate": 1.1124605433286643e-05, - "loss": 1.2292, - "step": 1781 - }, - { - "epoch": 1.44, - "learning_rate": 1.1115924073656112e-05, - "loss": 1.3918, - "step": 1782 - }, - { - "epoch": 1.44, - "learning_rate": 1.1107241862309203e-05, - "loss": 1.3281, - "step": 1783 - }, - { - "epoch": 1.44, - "learning_rate": 1.1098558805872512e-05, - "loss": 1.3019, - "step": 1784 - }, - { - "epoch": 1.44, - "learning_rate": 1.108987491097329e-05, - "loss": 1.3533, - "step": 1785 - }, - { - "epoch": 1.44, - "learning_rate": 1.1081190184239418e-05, - "loss": 1.2985, - "step": 1786 - }, - { - "epoch": 1.45, - "learning_rate": 1.1072504632299414e-05, - "loss": 1.117, - "step": 1787 - }, - { - "epoch": 1.45, - "learning_rate": 1.106381826178243e-05, - "loss": 1.2594, - "step": 1788 - }, - { - "epoch": 1.45, - "learning_rate": 1.1055131079318237e-05, - "loss": 1.2506, - "step": 1789 - }, - { - "epoch": 1.45, - "learning_rate": 1.1046443091537232e-05, - "loss": 1.3494, - "step": 1790 - }, - { - "epoch": 1.45, - "learning_rate": 1.1037754305070414e-05, - "loss": 1.2628, - "step": 1791 - }, - { - "epoch": 1.45, - "learning_rate": 1.1029064726549413e-05, - "loss": 1.2248, - "step": 1792 - }, - { - "epoch": 1.45, - "learning_rate": 1.102037436260644e-05, - "loss": 1.3959, - "step": 1793 - }, - { - "epoch": 1.45, - "learning_rate": 1.1011683219874324e-05, - "loss": 1.3645, - "step": 1794 - }, - { - "epoch": 1.45, - "learning_rate": 1.1002991304986476e-05, - "loss": 1.3835, - "step": 1795 - }, - { - "epoch": 1.45, - "learning_rate": 1.0994298624576907e-05, - "loss": 1.3202, - "step": 1796 - }, - { - "epoch": 1.45, - "learning_rate": 1.0985605185280201e-05, - "loss": 1.1679, - "step": 1797 - }, - { - "epoch": 1.45, - "learning_rate": 1.0976910993731531e-05, - "loss": 1.396, - "step": 1798 - }, - { - "epoch": 1.46, - "learning_rate": 1.0968216056566635e-05, - "loss": 1.2276, - "step": 1799 - }, - { - "epoch": 1.46, - "learning_rate": 1.0959520380421831e-05, - "loss": 1.2917, - "step": 1800 - }, - { - "epoch": 1.46, - "learning_rate": 1.0950823971933993e-05, - "loss": 1.2959, - "step": 1801 - }, - { - "epoch": 1.46, - "learning_rate": 1.0942126837740555e-05, - "loss": 1.3225, - "step": 1802 - }, - { - "epoch": 1.46, - "learning_rate": 1.0933428984479506e-05, - "loss": 1.1973, - "step": 1803 - }, - { - "epoch": 1.46, - "learning_rate": 1.0924730418789384e-05, - "loss": 1.1976, - "step": 1804 - }, - { - "epoch": 1.46, - "learning_rate": 1.091603114730927e-05, - "loss": 1.123, - "step": 1805 - }, - { - "epoch": 1.46, - "learning_rate": 1.0907331176678784e-05, - "loss": 1.2973, - "step": 1806 - }, - { - "epoch": 1.46, - "learning_rate": 1.0898630513538083e-05, - "loss": 1.2842, - "step": 1807 - }, - { - "epoch": 1.46, - "learning_rate": 1.0889929164527844e-05, - "loss": 1.3496, - "step": 1808 - }, - { - "epoch": 1.46, - "learning_rate": 1.0881227136289276e-05, - "loss": 1.3128, - "step": 1809 - }, - { - "epoch": 1.46, - "learning_rate": 1.0872524435464104e-05, - "loss": 1.2843, - "step": 1810 - }, - { - "epoch": 1.47, - "learning_rate": 1.0863821068694562e-05, - "loss": 1.3135, - "step": 1811 - }, - { - "epoch": 1.47, - "learning_rate": 1.08551170426234e-05, - "loss": 1.2957, - "step": 1812 - }, - { - "epoch": 1.47, - "learning_rate": 1.0846412363893868e-05, - "loss": 1.1616, - "step": 1813 - }, - { - "epoch": 1.47, - "learning_rate": 1.0837707039149709e-05, - "loss": 1.1189, - "step": 1814 - }, - { - "epoch": 1.47, - "learning_rate": 1.0829001075035166e-05, - "loss": 1.41, - "step": 1815 - }, - { - "epoch": 1.47, - "learning_rate": 1.0820294478194962e-05, - "loss": 1.3185, - "step": 1816 - }, - { - "epoch": 1.47, - "learning_rate": 1.0811587255274313e-05, - "loss": 1.1754, - "step": 1817 - }, - { - "epoch": 1.47, - "learning_rate": 1.080287941291891e-05, - "loss": 1.3523, - "step": 1818 - }, - { - "epoch": 1.47, - "learning_rate": 1.0794170957774913e-05, - "loss": 1.3047, - "step": 1819 - }, - { - "epoch": 1.47, - "learning_rate": 1.0785461896488947e-05, - "loss": 1.2813, - "step": 1820 - }, - { - "epoch": 1.47, - "learning_rate": 1.077675223570811e-05, - "loss": 1.1853, - "step": 1821 - }, - { - "epoch": 1.47, - "learning_rate": 1.0768041982079946e-05, - "loss": 1.2212, - "step": 1822 - }, - { - "epoch": 1.47, - "learning_rate": 1.0759331142252463e-05, - "loss": 1.3181, - "step": 1823 - }, - { - "epoch": 1.48, - "learning_rate": 1.075061972287411e-05, - "loss": 1.4139, - "step": 1824 - }, - { - "epoch": 1.48, - "learning_rate": 1.0741907730593776e-05, - "loss": 1.2979, - "step": 1825 - }, - { - "epoch": 1.48, - "learning_rate": 1.0733195172060791e-05, - "loss": 1.2802, - "step": 1826 - }, - { - "epoch": 1.48, - "learning_rate": 1.0724482053924914e-05, - "loss": 1.2952, - "step": 1827 - }, - { - "epoch": 1.48, - "learning_rate": 1.0715768382836338e-05, - "loss": 1.2556, - "step": 1828 - }, - { - "epoch": 1.48, - "learning_rate": 1.0707054165445674e-05, - "loss": 1.1701, - "step": 1829 - }, - { - "epoch": 1.48, - "learning_rate": 1.0698339408403944e-05, - "loss": 1.3003, - "step": 1830 - }, - { - "epoch": 1.48, - "learning_rate": 1.068962411836259e-05, - "loss": 1.408, - "step": 1831 - }, - { - "epoch": 1.48, - "learning_rate": 1.0680908301973461e-05, - "loss": 1.3269, - "step": 1832 - }, - { - "epoch": 1.48, - "learning_rate": 1.0672191965888803e-05, - "loss": 1.2066, - "step": 1833 - }, - { - "epoch": 1.48, - "learning_rate": 1.0663475116761258e-05, - "loss": 1.2777, - "step": 1834 - }, - { - "epoch": 1.48, - "learning_rate": 1.0654757761243865e-05, - "loss": 1.4303, - "step": 1835 - }, - { - "epoch": 1.49, - "learning_rate": 1.0646039905990046e-05, - "loss": 1.4886, - "step": 1836 - }, - { - "epoch": 1.49, - "learning_rate": 1.0637321557653604e-05, - "loss": 1.1979, - "step": 1837 - }, - { - "epoch": 1.49, - "learning_rate": 1.062860272288872e-05, - "loss": 1.2422, - "step": 1838 - }, - { - "epoch": 1.49, - "learning_rate": 1.0619883408349947e-05, - "loss": 1.3414, - "step": 1839 - }, - { - "epoch": 1.49, - "learning_rate": 1.06111636206922e-05, - "loss": 1.3405, - "step": 1840 - }, - { - "epoch": 1.49, - "learning_rate": 1.060244336657076e-05, - "loss": 1.3371, - "step": 1841 - }, - { - "epoch": 1.49, - "learning_rate": 1.0593722652641263e-05, - "loss": 1.2828, - "step": 1842 - }, - { - "epoch": 1.49, - "learning_rate": 1.058500148555969e-05, - "loss": 1.1793, - "step": 1843 - }, - { - "epoch": 1.49, - "learning_rate": 1.057627987198238e-05, - "loss": 1.2421, - "step": 1844 - }, - { - "epoch": 1.49, - "learning_rate": 1.0567557818565997e-05, - "loss": 1.1845, - "step": 1845 - }, - { - "epoch": 1.49, - "learning_rate": 1.0558835331967559e-05, - "loss": 1.3135, - "step": 1846 - }, - { - "epoch": 1.49, - "learning_rate": 1.0550112418844396e-05, - "loss": 1.2039, - "step": 1847 - }, - { - "epoch": 1.5, - "learning_rate": 1.0541389085854177e-05, - "loss": 1.2987, - "step": 1848 - }, - { - "epoch": 1.5, - "learning_rate": 1.0532665339654883e-05, - "loss": 1.3111, - "step": 1849 - }, - { - "epoch": 1.5, - "learning_rate": 1.0523941186904823e-05, - "loss": 1.3208, - "step": 1850 - }, - { - "epoch": 1.5, - "learning_rate": 1.05152166342626e-05, - "loss": 1.2987, - "step": 1851 - }, - { - "epoch": 1.5, - "learning_rate": 1.0506491688387128e-05, - "loss": 1.2831, - "step": 1852 - }, - { - "epoch": 1.5, - "learning_rate": 1.049776635593763e-05, - "loss": 1.2717, - "step": 1853 - }, - { - "epoch": 1.5, - "learning_rate": 1.0489040643573609e-05, - "loss": 1.2855, - "step": 1854 - }, - { - "epoch": 1.5, - "learning_rate": 1.0480314557954872e-05, - "loss": 1.2114, - "step": 1855 - }, - { - "epoch": 1.5, - "learning_rate": 1.0471588105741506e-05, - "loss": 1.3478, - "step": 1856 - }, - { - "epoch": 1.5, - "learning_rate": 1.046286129359387e-05, - "loss": 1.4471, - "step": 1857 - }, - { - "epoch": 1.5, - "learning_rate": 1.045413412817261e-05, - "loss": 1.2665, - "step": 1858 - }, - { - "epoch": 1.5, - "learning_rate": 1.0445406616138632e-05, - "loss": 1.2159, - "step": 1859 - }, - { - "epoch": 1.5, - "learning_rate": 1.043667876415311e-05, - "loss": 1.1814, - "step": 1860 - }, - { - "epoch": 1.51, - "learning_rate": 1.0427950578877482e-05, - "loss": 1.214, - "step": 1861 - }, - { - "epoch": 1.51, - "learning_rate": 1.0419222066973433e-05, - "loss": 1.3007, - "step": 1862 - }, - { - "epoch": 1.51, - "learning_rate": 1.0410493235102902e-05, - "loss": 1.2205, - "step": 1863 - }, - { - "epoch": 1.51, - "learning_rate": 1.0401764089928073e-05, - "loss": 1.3227, - "step": 1864 - }, - { - "epoch": 1.51, - "learning_rate": 1.0393034638111361e-05, - "loss": 1.2348, - "step": 1865 - }, - { - "epoch": 1.51, - "learning_rate": 1.0384304886315426e-05, - "loss": 1.3248, - "step": 1866 - }, - { - "epoch": 1.51, - "learning_rate": 1.0375574841203151e-05, - "loss": 1.2559, - "step": 1867 - }, - { - "epoch": 1.51, - "learning_rate": 1.0366844509437645e-05, - "loss": 1.3379, - "step": 1868 - }, - { - "epoch": 1.51, - "learning_rate": 1.035811389768223e-05, - "loss": 1.2945, - "step": 1869 - }, - { - "epoch": 1.51, - "learning_rate": 1.0349383012600448e-05, - "loss": 1.2464, - "step": 1870 - }, - { - "epoch": 1.51, - "learning_rate": 1.0340651860856052e-05, - "loss": 1.2913, - "step": 1871 - }, - { - "epoch": 1.51, - "learning_rate": 1.0331920449112991e-05, - "loss": 1.1412, - "step": 1872 - }, - { - "epoch": 1.52, - "learning_rate": 1.0323188784035416e-05, - "loss": 1.3293, - "step": 1873 - }, - { - "epoch": 1.52, - "learning_rate": 1.0314456872287672e-05, - "loss": 1.1675, - "step": 1874 - }, - { - "epoch": 1.52, - "learning_rate": 1.030572472053429e-05, - "loss": 1.2915, - "step": 1875 - }, - { - "epoch": 1.52, - "learning_rate": 1.0296992335439987e-05, - "loss": 1.3968, - "step": 1876 - }, - { - "epoch": 1.52, - "learning_rate": 1.0288259723669657e-05, - "loss": 1.1862, - "step": 1877 - }, - { - "epoch": 1.52, - "learning_rate": 1.0279526891888365e-05, - "loss": 1.3007, - "step": 1878 - }, - { - "epoch": 1.52, - "learning_rate": 1.0270793846761347e-05, - "loss": 1.2248, - "step": 1879 - }, - { - "epoch": 1.52, - "learning_rate": 1.0262060594954e-05, - "loss": 1.2337, - "step": 1880 - }, - { - "epoch": 1.52, - "learning_rate": 1.025332714313188e-05, - "loss": 1.2659, - "step": 1881 - }, - { - "epoch": 1.52, - "learning_rate": 1.0244593497960696e-05, - "loss": 1.1729, - "step": 1882 - }, - { - "epoch": 1.52, - "learning_rate": 1.0235859666106302e-05, - "loss": 1.313, - "step": 1883 - }, - { - "epoch": 1.52, - "learning_rate": 1.0227125654234695e-05, - "loss": 1.3337, - "step": 1884 - }, - { - "epoch": 1.53, - "learning_rate": 1.0218391469012012e-05, - "loss": 1.3193, - "step": 1885 - }, - { - "epoch": 1.53, - "learning_rate": 1.020965711710452e-05, - "loss": 1.4333, - "step": 1886 - }, - { - "epoch": 1.53, - "learning_rate": 1.0200922605178611e-05, - "loss": 1.3685, - "step": 1887 - }, - { - "epoch": 1.53, - "learning_rate": 1.0192187939900811e-05, - "loss": 1.3515, - "step": 1888 - }, - { - "epoch": 1.53, - "learning_rate": 1.0183453127937745e-05, - "loss": 1.2917, - "step": 1889 - }, - { - "epoch": 1.53, - "learning_rate": 1.0174718175956164e-05, - "loss": 1.3016, - "step": 1890 - }, - { - "epoch": 1.53, - "learning_rate": 1.0165983090622917e-05, - "loss": 1.1813, - "step": 1891 - }, - { - "epoch": 1.53, - "learning_rate": 1.015724787860496e-05, - "loss": 1.3397, - "step": 1892 - }, - { - "epoch": 1.53, - "learning_rate": 1.0148512546569348e-05, - "loss": 1.1592, - "step": 1893 - }, - { - "epoch": 1.53, - "learning_rate": 1.0139777101183221e-05, - "loss": 1.3559, - "step": 1894 - }, - { - "epoch": 1.53, - "learning_rate": 1.013104154911381e-05, - "loss": 1.3773, - "step": 1895 - }, - { - "epoch": 1.53, - "learning_rate": 1.0122305897028423e-05, - "loss": 1.2473, - "step": 1896 - }, - { - "epoch": 1.53, - "learning_rate": 1.0113570151594448e-05, - "loss": 1.326, - "step": 1897 - }, - { - "epoch": 1.54, - "learning_rate": 1.0104834319479343e-05, - "loss": 1.1308, - "step": 1898 - }, - { - "epoch": 1.54, - "learning_rate": 1.0096098407350639e-05, - "loss": 1.2109, - "step": 1899 - }, - { - "epoch": 1.54, - "learning_rate": 1.0087362421875912e-05, - "loss": 1.2446, - "step": 1900 - }, - { - "epoch": 1.54, - "learning_rate": 1.0078626369722812e-05, - "loss": 1.2596, - "step": 1901 - }, - { - "epoch": 1.54, - "learning_rate": 1.0069890257559027e-05, - "loss": 1.3862, - "step": 1902 - }, - { - "epoch": 1.54, - "learning_rate": 1.0061154092052294e-05, - "loss": 1.2027, - "step": 1903 - }, - { - "epoch": 1.54, - "learning_rate": 1.0052417879870399e-05, - "loss": 1.1844, - "step": 1904 - }, - { - "epoch": 1.54, - "learning_rate": 1.004368162768115e-05, - "loss": 1.297, - "step": 1905 - }, - { - "epoch": 1.54, - "learning_rate": 1.0034945342152395e-05, - "loss": 1.2983, - "step": 1906 - }, - { - "epoch": 1.54, - "learning_rate": 1.0026209029952005e-05, - "loss": 1.3104, - "step": 1907 - }, - { - "epoch": 1.54, - "learning_rate": 1.001747269774787e-05, - "loss": 1.1799, - "step": 1908 - }, - { - "epoch": 1.54, - "learning_rate": 1.0008736352207895e-05, - "loss": 1.2357, - "step": 1909 - }, - { - "epoch": 1.55, - "learning_rate": 1e-05, - "loss": 1.4556, - "step": 1910 - }, - { - "epoch": 1.55, - "learning_rate": 9.991263647792108e-06, - "loss": 1.2267, - "step": 1911 - }, - { - "epoch": 1.55, - "learning_rate": 9.982527302252135e-06, - "loss": 1.2388, - "step": 1912 - }, - { - "epoch": 1.55, - "learning_rate": 9.973790970048e-06, - "loss": 1.3472, - "step": 1913 - }, - { - "epoch": 1.55, - "learning_rate": 9.965054657847606e-06, - "loss": 1.2285, - "step": 1914 - }, - { - "epoch": 1.55, - "learning_rate": 9.956318372318852e-06, - "loss": 1.3635, - "step": 1915 - }, - { - "epoch": 1.55, - "learning_rate": 9.947582120129604e-06, - "loss": 1.3718, - "step": 1916 - }, - { - "epoch": 1.55, - "learning_rate": 9.938845907947708e-06, - "loss": 1.2181, - "step": 1917 - }, - { - "epoch": 1.55, - "learning_rate": 9.930109742440976e-06, - "loss": 1.2794, - "step": 1918 - }, - { - "epoch": 1.55, - "learning_rate": 9.921373630277191e-06, - "loss": 1.2711, - "step": 1919 - }, - { - "epoch": 1.55, - "learning_rate": 9.912637578124092e-06, - "loss": 1.1453, - "step": 1920 - }, - { - "epoch": 1.55, - "learning_rate": 9.903901592649363e-06, - "loss": 1.302, - "step": 1921 - }, - { - "epoch": 1.56, - "learning_rate": 9.895165680520658e-06, - "loss": 1.3414, - "step": 1922 - }, - { - "epoch": 1.56, - "learning_rate": 9.886429848405556e-06, - "loss": 1.1807, - "step": 1923 - }, - { - "epoch": 1.56, - "learning_rate": 9.87769410297158e-06, - "loss": 1.197, - "step": 1924 - }, - { - "epoch": 1.56, - "learning_rate": 9.868958450886194e-06, - "loss": 1.1237, - "step": 1925 - }, - { - "epoch": 1.56, - "learning_rate": 9.860222898816782e-06, - "loss": 1.099, - "step": 1926 - }, - { - "epoch": 1.56, - "learning_rate": 9.851487453430657e-06, - "loss": 1.1589, - "step": 1927 - }, - { - "epoch": 1.56, - "learning_rate": 9.84275212139504e-06, - "loss": 1.2546, - "step": 1928 - }, - { - "epoch": 1.56, - "learning_rate": 9.834016909377086e-06, - "loss": 1.3236, - "step": 1929 - }, - { - "epoch": 1.56, - "learning_rate": 9.825281824043838e-06, - "loss": 1.2516, - "step": 1930 - }, - { - "epoch": 1.56, - "learning_rate": 9.816546872062257e-06, - "loss": 1.2312, - "step": 1931 - }, - { - "epoch": 1.56, - "learning_rate": 9.807812060099192e-06, - "loss": 1.2853, - "step": 1932 - }, - { - "epoch": 1.56, - "learning_rate": 9.79907739482139e-06, - "loss": 1.179, - "step": 1933 - }, - { - "epoch": 1.56, - "learning_rate": 9.790342882895484e-06, - "loss": 1.3582, - "step": 1934 - }, - { - "epoch": 1.57, - "learning_rate": 9.78160853098799e-06, - "loss": 1.2171, - "step": 1935 - }, - { - "epoch": 1.57, - "learning_rate": 9.772874345765307e-06, - "loss": 1.4213, - "step": 1936 - }, - { - "epoch": 1.57, - "learning_rate": 9.764140333893701e-06, - "loss": 1.1197, - "step": 1937 - }, - { - "epoch": 1.57, - "learning_rate": 9.755406502039307e-06, - "loss": 1.2688, - "step": 1938 - }, - { - "epoch": 1.57, - "learning_rate": 9.746672856868124e-06, - "loss": 1.276, - "step": 1939 - }, - { - "epoch": 1.57, - "learning_rate": 9.737939405046002e-06, - "loss": 1.3728, - "step": 1940 - }, - { - "epoch": 1.57, - "learning_rate": 9.729206153238658e-06, - "loss": 1.307, - "step": 1941 - }, - { - "epoch": 1.57, - "learning_rate": 9.72047310811164e-06, - "loss": 1.242, - "step": 1942 - }, - { - "epoch": 1.57, - "learning_rate": 9.711740276330345e-06, - "loss": 1.1343, - "step": 1943 - }, - { - "epoch": 1.57, - "learning_rate": 9.703007664560015e-06, - "loss": 1.3874, - "step": 1944 - }, - { - "epoch": 1.57, - "learning_rate": 9.694275279465715e-06, - "loss": 1.2966, - "step": 1945 - }, - { - "epoch": 1.57, - "learning_rate": 9.685543127712331e-06, - "loss": 1.3564, - "step": 1946 - }, - { - "epoch": 1.58, - "learning_rate": 9.676811215964588e-06, - "loss": 1.2492, - "step": 1947 - }, - { - "epoch": 1.58, - "learning_rate": 9.668079550887012e-06, - "loss": 1.2201, - "step": 1948 - }, - { - "epoch": 1.58, - "learning_rate": 9.659348139143953e-06, - "loss": 1.2791, - "step": 1949 - }, - { - "epoch": 1.58, - "learning_rate": 9.650616987399553e-06, - "loss": 1.2906, - "step": 1950 - }, - { - "epoch": 1.58, - "learning_rate": 9.641886102317772e-06, - "loss": 1.3411, - "step": 1951 - }, - { - "epoch": 1.58, - "learning_rate": 9.633155490562358e-06, - "loss": 1.244, - "step": 1952 - }, - { - "epoch": 1.58, - "learning_rate": 9.62442515879685e-06, - "loss": 1.2407, - "step": 1953 - }, - { - "epoch": 1.58, - "learning_rate": 9.615695113684576e-06, - "loss": 1.2462, - "step": 1954 - }, - { - "epoch": 1.58, - "learning_rate": 9.606965361888642e-06, - "loss": 1.1232, - "step": 1955 - }, - { - "epoch": 1.58, - "learning_rate": 9.59823591007193e-06, - "loss": 1.2615, - "step": 1956 - }, - { - "epoch": 1.58, - "learning_rate": 9.589506764897097e-06, - "loss": 1.3553, - "step": 1957 - }, - { - "epoch": 1.58, - "learning_rate": 9.580777933026568e-06, - "loss": 1.2972, - "step": 1958 - }, - { - "epoch": 1.58, - "learning_rate": 9.572049421122521e-06, - "loss": 1.3411, - "step": 1959 - }, - { - "epoch": 1.59, - "learning_rate": 9.563321235846894e-06, - "loss": 1.2159, - "step": 1960 - }, - { - "epoch": 1.59, - "learning_rate": 9.554593383861372e-06, - "loss": 1.2964, - "step": 1961 - }, - { - "epoch": 1.59, - "learning_rate": 9.545865871827396e-06, - "loss": 1.3115, - "step": 1962 - }, - { - "epoch": 1.59, - "learning_rate": 9.537138706406135e-06, - "loss": 1.2905, - "step": 1963 - }, - { - "epoch": 1.59, - "learning_rate": 9.528411894258495e-06, - "loss": 1.2186, - "step": 1964 - }, - { - "epoch": 1.59, - "learning_rate": 9.519685442045128e-06, - "loss": 1.3558, - "step": 1965 - }, - { - "epoch": 1.59, - "learning_rate": 9.510959356426393e-06, - "loss": 1.4378, - "step": 1966 - }, - { - "epoch": 1.59, - "learning_rate": 9.502233644062373e-06, - "loss": 1.3221, - "step": 1967 - }, - { - "epoch": 1.59, - "learning_rate": 9.493508311612874e-06, - "loss": 1.3417, - "step": 1968 - }, - { - "epoch": 1.59, - "learning_rate": 9.484783365737405e-06, - "loss": 1.3947, - "step": 1969 - }, - { - "epoch": 1.59, - "learning_rate": 9.476058813095182e-06, - "loss": 1.267, - "step": 1970 - }, - { - "epoch": 1.59, - "learning_rate": 9.467334660345115e-06, - "loss": 1.3085, - "step": 1971 - }, - { - "epoch": 1.6, - "learning_rate": 9.458610914145826e-06, - "loss": 1.391, - "step": 1972 - }, - { - "epoch": 1.6, - "learning_rate": 9.449887581155606e-06, - "loss": 1.2241, - "step": 1973 - }, - { - "epoch": 1.6, - "learning_rate": 9.441164668032446e-06, - "loss": 1.1832, - "step": 1974 - }, - { - "epoch": 1.6, - "learning_rate": 9.432442181434005e-06, - "loss": 1.4677, - "step": 1975 - }, - { - "epoch": 1.6, - "learning_rate": 9.423720128017626e-06, - "loss": 1.2551, - "step": 1976 - }, - { - "epoch": 1.6, - "learning_rate": 9.414998514440312e-06, - "loss": 1.2863, - "step": 1977 - }, - { - "epoch": 1.6, - "learning_rate": 9.40627734735874e-06, - "loss": 1.2912, - "step": 1978 - }, - { - "epoch": 1.6, - "learning_rate": 9.397556633429242e-06, - "loss": 1.1924, - "step": 1979 - }, - { - "epoch": 1.6, - "learning_rate": 9.388836379307802e-06, - "loss": 1.1942, - "step": 1980 - }, - { - "epoch": 1.6, - "learning_rate": 9.380116591650056e-06, - "loss": 1.2418, - "step": 1981 - }, - { - "epoch": 1.6, - "learning_rate": 9.371397277111284e-06, - "loss": 1.251, - "step": 1982 - }, - { - "epoch": 1.6, - "learning_rate": 9.3626784423464e-06, - "loss": 1.2541, - "step": 1983 - }, - { - "epoch": 1.61, - "learning_rate": 9.353960094009959e-06, - "loss": 1.3102, - "step": 1984 - }, - { - "epoch": 1.61, - "learning_rate": 9.345242238756136e-06, - "loss": 1.2063, - "step": 1985 - }, - { - "epoch": 1.61, - "learning_rate": 9.336524883238744e-06, - "loss": 1.2118, - "step": 1986 - }, - { - "epoch": 1.61, - "learning_rate": 9.3278080341112e-06, - "loss": 1.3086, - "step": 1987 - }, - { - "epoch": 1.61, - "learning_rate": 9.31909169802654e-06, - "loss": 1.2193, - "step": 1988 - }, - { - "epoch": 1.61, - "learning_rate": 9.31037588163741e-06, - "loss": 1.3106, - "step": 1989 - }, - { - "epoch": 1.61, - "learning_rate": 9.301660591596059e-06, - "loss": 1.2823, - "step": 1990 - }, - { - "epoch": 1.61, - "learning_rate": 9.292945834554331e-06, - "loss": 1.3499, - "step": 1991 - }, - { - "epoch": 1.61, - "learning_rate": 9.284231617163665e-06, - "loss": 1.2374, - "step": 1992 - }, - { - "epoch": 1.61, - "learning_rate": 9.275517946075088e-06, - "loss": 1.3761, - "step": 1993 - }, - { - "epoch": 1.61, - "learning_rate": 9.26680482793921e-06, - "loss": 1.2809, - "step": 1994 - }, - { - "epoch": 1.61, - "learning_rate": 9.258092269406227e-06, - "loss": 1.2416, - "step": 1995 - }, - { - "epoch": 1.61, - "learning_rate": 9.249380277125894e-06, - "loss": 1.3151, - "step": 1996 - }, - { - "epoch": 1.62, - "learning_rate": 9.24066885774754e-06, - "loss": 1.2849, - "step": 1997 - }, - { - "epoch": 1.62, - "learning_rate": 9.231958017920057e-06, - "loss": 1.186, - "step": 1998 - }, - { - "epoch": 1.62, - "learning_rate": 9.223247764291894e-06, - "loss": 1.3388, - "step": 1999 - }, - { - "epoch": 1.62, - "learning_rate": 9.214538103511053e-06, - "loss": 1.2728, - "step": 2000 - }, - { - "epoch": 1.62, - "learning_rate": 9.20582904222509e-06, - "loss": 1.3318, - "step": 2001 - }, - { - "epoch": 1.62, - "learning_rate": 9.197120587081092e-06, - "loss": 1.1358, - "step": 2002 - }, - { - "epoch": 1.62, - "learning_rate": 9.18841274472569e-06, - "loss": 1.278, - "step": 2003 - }, - { - "epoch": 1.62, - "learning_rate": 9.179705521805042e-06, - "loss": 1.4646, - "step": 2004 - }, - { - "epoch": 1.62, - "learning_rate": 9.17099892496484e-06, - "loss": 1.1685, - "step": 2005 - }, - { - "epoch": 1.62, - "learning_rate": 9.162292960850296e-06, - "loss": 1.2531, - "step": 2006 - }, - { - "epoch": 1.62, - "learning_rate": 9.153587636106134e-06, - "loss": 1.3381, - "step": 2007 - }, - { - "epoch": 1.62, - "learning_rate": 9.1448829573766e-06, - "loss": 1.185, - "step": 2008 - }, - { - "epoch": 1.63, - "learning_rate": 9.13617893130544e-06, - "loss": 1.2841, - "step": 2009 - }, - { - "epoch": 1.63, - "learning_rate": 9.127475564535898e-06, - "loss": 1.1833, - "step": 2010 - }, - { - "epoch": 1.63, - "learning_rate": 9.118772863710728e-06, - "loss": 1.286, - "step": 2011 - }, - { - "epoch": 1.63, - "learning_rate": 9.110070835472161e-06, - "loss": 1.3566, - "step": 2012 - }, - { - "epoch": 1.63, - "learning_rate": 9.101369486461924e-06, - "loss": 1.2843, - "step": 2013 - }, - { - "epoch": 1.63, - "learning_rate": 9.092668823321218e-06, - "loss": 1.2617, - "step": 2014 - }, - { - "epoch": 1.63, - "learning_rate": 9.083968852690735e-06, - "loss": 1.4831, - "step": 2015 - }, - { - "epoch": 1.63, - "learning_rate": 9.075269581210618e-06, - "loss": 1.2992, - "step": 2016 - }, - { - "epoch": 1.63, - "learning_rate": 9.066571015520498e-06, - "loss": 1.2388, - "step": 2017 - }, - { - "epoch": 1.63, - "learning_rate": 9.05787316225945e-06, - "loss": 1.3373, - "step": 2018 - }, - { - "epoch": 1.63, - "learning_rate": 9.04917602806601e-06, - "loss": 1.2792, - "step": 2019 - }, - { - "epoch": 1.63, - "learning_rate": 9.04047961957817e-06, - "loss": 1.295, - "step": 2020 - }, - { - "epoch": 1.64, - "learning_rate": 9.031783943433366e-06, - "loss": 1.3178, - "step": 2021 - }, - { - "epoch": 1.64, - "learning_rate": 9.02308900626847e-06, - "loss": 1.3059, - "step": 2022 - }, - { - "epoch": 1.64, - "learning_rate": 9.0143948147198e-06, - "loss": 1.208, - "step": 2023 - }, - { - "epoch": 1.64, - "learning_rate": 9.005701375423097e-06, - "loss": 1.3449, - "step": 2024 - }, - { - "epoch": 1.64, - "learning_rate": 8.997008695013527e-06, - "loss": 1.2922, - "step": 2025 - }, - { - "epoch": 1.64, - "learning_rate": 8.98831678012568e-06, - "loss": 1.0861, - "step": 2026 - }, - { - "epoch": 1.64, - "learning_rate": 8.979625637393565e-06, - "loss": 1.2307, - "step": 2027 - }, - { - "epoch": 1.64, - "learning_rate": 8.970935273450589e-06, - "loss": 1.3288, - "step": 2028 - }, - { - "epoch": 1.64, - "learning_rate": 8.962245694929586e-06, - "loss": 1.1626, - "step": 2029 - }, - { - "epoch": 1.64, - "learning_rate": 8.953556908462773e-06, - "loss": 1.2599, - "step": 2030 - }, - { - "epoch": 1.64, - "learning_rate": 8.944868920681765e-06, - "loss": 1.2051, - "step": 2031 - }, - { - "epoch": 1.64, - "learning_rate": 8.936181738217571e-06, - "loss": 1.2943, - "step": 2032 - }, - { - "epoch": 1.64, - "learning_rate": 8.927495367700587e-06, - "loss": 1.2598, - "step": 2033 - }, - { - "epoch": 1.65, - "learning_rate": 8.918809815760585e-06, - "loss": 1.2036, - "step": 2034 - }, - { - "epoch": 1.65, - "learning_rate": 8.91012508902671e-06, - "loss": 1.2142, - "step": 2035 - }, - { - "epoch": 1.65, - "learning_rate": 8.90144119412749e-06, - "loss": 1.0803, - "step": 2036 - }, - { - "epoch": 1.65, - "learning_rate": 8.8927581376908e-06, - "loss": 1.2221, - "step": 2037 - }, - { - "epoch": 1.65, - "learning_rate": 8.88407592634389e-06, - "loss": 1.3216, - "step": 2038 - }, - { - "epoch": 1.65, - "learning_rate": 8.875394566713358e-06, - "loss": 1.4816, - "step": 2039 - }, - { - "epoch": 1.65, - "learning_rate": 8.866714065425154e-06, - "loss": 1.346, - "step": 2040 - }, - { - "epoch": 1.65, - "learning_rate": 8.85803442910457e-06, - "loss": 1.2034, - "step": 2041 - }, - { - "epoch": 1.65, - "learning_rate": 8.849355664376237e-06, - "loss": 1.3037, - "step": 2042 - }, - { - "epoch": 1.65, - "learning_rate": 8.840677777864129e-06, - "loss": 1.1387, - "step": 2043 - }, - { - "epoch": 1.65, - "learning_rate": 8.832000776191544e-06, - "loss": 1.2285, - "step": 2044 - }, - { - "epoch": 1.65, - "learning_rate": 8.823324665981103e-06, - "loss": 1.2436, - "step": 2045 - }, - { - "epoch": 1.66, - "learning_rate": 8.81464945385475e-06, - "loss": 1.2069, - "step": 2046 - }, - { - "epoch": 1.66, - "learning_rate": 8.805975146433738e-06, - "loss": 1.2635, - "step": 2047 - }, - { - "epoch": 1.66, - "learning_rate": 8.79730175033864e-06, - "loss": 1.3093, - "step": 2048 - }, - { - "epoch": 1.66, - "learning_rate": 8.788629272189325e-06, - "loss": 1.2586, - "step": 2049 - }, - { - "epoch": 1.66, - "learning_rate": 8.779957718604956e-06, - "loss": 1.1703, - "step": 2050 - }, - { - "epoch": 1.66, - "learning_rate": 8.771287096204011e-06, - "loss": 1.2712, - "step": 2051 - }, - { - "epoch": 1.66, - "learning_rate": 8.762617411604236e-06, - "loss": 1.1708, - "step": 2052 - }, - { - "epoch": 1.66, - "learning_rate": 8.753948671422669e-06, - "loss": 1.1916, - "step": 2053 - }, - { - "epoch": 1.66, - "learning_rate": 8.74528088227563e-06, - "loss": 1.2649, - "step": 2054 - }, - { - "epoch": 1.66, - "learning_rate": 8.73661405077871e-06, - "loss": 1.1738, - "step": 2055 - }, - { - "epoch": 1.66, - "learning_rate": 8.72794818354677e-06, - "loss": 1.1296, - "step": 2056 - }, - { - "epoch": 1.66, - "learning_rate": 8.719283287193932e-06, - "loss": 1.2301, - "step": 2057 - }, - { - "epoch": 1.67, - "learning_rate": 8.710619368333584e-06, - "loss": 1.3467, - "step": 2058 - }, - { - "epoch": 1.67, - "learning_rate": 8.70195643357836e-06, - "loss": 1.3224, - "step": 2059 - }, - { - "epoch": 1.67, - "learning_rate": 8.693294489540151e-06, - "loss": 1.2553, - "step": 2060 - }, - { - "epoch": 1.67, - "learning_rate": 8.684633542830084e-06, - "loss": 1.2104, - "step": 2061 - }, - { - "epoch": 1.67, - "learning_rate": 8.675973600058529e-06, - "loss": 1.2463, - "step": 2062 - }, - { - "epoch": 1.67, - "learning_rate": 8.667314667835088e-06, - "loss": 1.3071, - "step": 2063 - }, - { - "epoch": 1.67, - "learning_rate": 8.658656752768592e-06, - "loss": 1.1131, - "step": 2064 - }, - { - "epoch": 1.67, - "learning_rate": 8.6499998614671e-06, - "loss": 1.1256, - "step": 2065 - }, - { - "epoch": 1.67, - "learning_rate": 8.641344000537883e-06, - "loss": 1.26, - "step": 2066 - }, - { - "epoch": 1.67, - "learning_rate": 8.632689176587433e-06, - "loss": 1.239, - "step": 2067 - }, - { - "epoch": 1.67, - "learning_rate": 8.62403539622144e-06, - "loss": 1.2572, - "step": 2068 - }, - { - "epoch": 1.67, - "learning_rate": 8.615382666044806e-06, - "loss": 1.1415, - "step": 2069 - }, - { - "epoch": 1.67, - "learning_rate": 8.60673099266163e-06, - "loss": 1.2859, - "step": 2070 - }, - { - "epoch": 1.68, - "learning_rate": 8.5980803826752e-06, - "loss": 1.3625, - "step": 2071 - }, - { - "epoch": 1.68, - "learning_rate": 8.589430842688002e-06, - "loss": 1.3811, - "step": 2072 - }, - { - "epoch": 1.68, - "learning_rate": 8.580782379301696e-06, - "loss": 1.1991, - "step": 2073 - }, - { - "epoch": 1.68, - "learning_rate": 8.572134999117122e-06, - "loss": 1.269, - "step": 2074 - }, - { - "epoch": 1.68, - "learning_rate": 8.563488708734296e-06, - "loss": 1.2087, - "step": 2075 - }, - { - "epoch": 1.68, - "learning_rate": 8.554843514752402e-06, - "loss": 1.1202, - "step": 2076 - }, - { - "epoch": 1.68, - "learning_rate": 8.546199423769786e-06, - "loss": 1.1487, - "step": 2077 - }, - { - "epoch": 1.68, - "learning_rate": 8.537556442383945e-06, - "loss": 1.2464, - "step": 2078 - }, - { - "epoch": 1.68, - "learning_rate": 8.528914577191548e-06, - "loss": 1.4482, - "step": 2079 - }, - { - "epoch": 1.68, - "learning_rate": 8.520273834788395e-06, - "loss": 1.1445, - "step": 2080 - }, - { - "epoch": 1.68, - "learning_rate": 8.511634221769434e-06, - "loss": 1.3359, - "step": 2081 - }, - { - "epoch": 1.68, - "learning_rate": 8.502995744728755e-06, - "loss": 1.4168, - "step": 2082 - }, - { - "epoch": 1.69, - "learning_rate": 8.494358410259572e-06, - "loss": 1.1879, - "step": 2083 - }, - { - "epoch": 1.69, - "learning_rate": 8.485722224954237e-06, - "loss": 1.1967, - "step": 2084 - }, - { - "epoch": 1.69, - "learning_rate": 8.477087195404214e-06, - "loss": 1.0367, - "step": 2085 - }, - { - "epoch": 1.69, - "learning_rate": 8.468453328200098e-06, - "loss": 1.3085, - "step": 2086 - }, - { - "epoch": 1.69, - "learning_rate": 8.45982062993159e-06, - "loss": 1.2964, - "step": 2087 - }, - { - "epoch": 1.69, - "learning_rate": 8.451189107187498e-06, - "loss": 1.1648, - "step": 2088 - }, - { - "epoch": 1.69, - "learning_rate": 8.442558766555732e-06, - "loss": 1.3886, - "step": 2089 - }, - { - "epoch": 1.69, - "learning_rate": 8.4339296146233e-06, - "loss": 1.31, - "step": 2090 - }, - { - "epoch": 1.69, - "learning_rate": 8.425301657976306e-06, - "loss": 1.2303, - "step": 2091 - }, - { - "epoch": 1.69, - "learning_rate": 8.41667490319994e-06, - "loss": 1.1818, - "step": 2092 - }, - { - "epoch": 1.69, - "learning_rate": 8.408049356878478e-06, - "loss": 1.3147, - "step": 2093 - }, - { - "epoch": 1.69, - "learning_rate": 8.399425025595265e-06, - "loss": 1.3568, - "step": 2094 - }, - { - "epoch": 1.69, - "learning_rate": 8.390801915932729e-06, - "loss": 1.2653, - "step": 2095 - }, - { - "epoch": 1.7, - "learning_rate": 8.382180034472353e-06, - "loss": 1.2303, - "step": 2096 - }, - { - "epoch": 1.7, - "learning_rate": 8.373559387794699e-06, - "loss": 1.3284, - "step": 2097 - }, - { - "epoch": 1.7, - "learning_rate": 8.364939982479372e-06, - "loss": 1.3294, - "step": 2098 - }, - { - "epoch": 1.7, - "learning_rate": 8.356321825105037e-06, - "loss": 1.2835, - "step": 2099 - }, - { - "epoch": 1.7, - "learning_rate": 8.3477049222494e-06, - "loss": 1.3436, - "step": 2100 - }, - { - "epoch": 1.7, - "learning_rate": 8.339089280489224e-06, - "loss": 1.356, - "step": 2101 - }, - { - "epoch": 1.7, - "learning_rate": 8.330474906400292e-06, - "loss": 1.3123, - "step": 2102 - }, - { - "epoch": 1.7, - "learning_rate": 8.321861806557431e-06, - "loss": 1.2175, - "step": 2103 - }, - { - "epoch": 1.7, - "learning_rate": 8.313249987534491e-06, - "loss": 1.3075, - "step": 2104 - }, - { - "epoch": 1.7, - "learning_rate": 8.304639455904344e-06, - "loss": 1.3268, - "step": 2105 - }, - { - "epoch": 1.7, - "learning_rate": 8.29603021823888e-06, - "loss": 1.241, - "step": 2106 - }, - { - "epoch": 1.7, - "learning_rate": 8.287422281109e-06, - "loss": 1.2451, - "step": 2107 - }, - { - "epoch": 1.71, - "learning_rate": 8.278815651084618e-06, - "loss": 1.3183, - "step": 2108 - }, - { - "epoch": 1.71, - "learning_rate": 8.270210334734649e-06, - "loss": 1.3347, - "step": 2109 - }, - { - "epoch": 1.71, - "learning_rate": 8.261606338626998e-06, - "loss": 1.2126, - "step": 2110 - }, - { - "epoch": 1.71, - "learning_rate": 8.25300366932857e-06, - "loss": 1.2341, - "step": 2111 - }, - { - "epoch": 1.71, - "learning_rate": 8.244402333405252e-06, - "loss": 1.2113, - "step": 2112 - }, - { - "epoch": 1.71, - "learning_rate": 8.23580233742192e-06, - "loss": 1.1667, - "step": 2113 - }, - { - "epoch": 1.71, - "learning_rate": 8.227203687942414e-06, - "loss": 1.328, - "step": 2114 - }, - { - "epoch": 1.71, - "learning_rate": 8.218606391529568e-06, - "loss": 1.3621, - "step": 2115 - }, - { - "epoch": 1.71, - "learning_rate": 8.210010454745166e-06, - "loss": 1.2181, - "step": 2116 - }, - { - "epoch": 1.71, - "learning_rate": 8.201415884149957e-06, - "loss": 1.1729, - "step": 2117 - }, - { - "epoch": 1.71, - "learning_rate": 8.19282268630365e-06, - "loss": 1.2232, - "step": 2118 - }, - { - "epoch": 1.71, - "learning_rate": 8.184230867764908e-06, - "loss": 1.1594, - "step": 2119 - }, - { - "epoch": 1.72, - "learning_rate": 8.17564043509134e-06, - "loss": 1.1721, - "step": 2120 - }, - { - "epoch": 1.72, - "learning_rate": 8.167051394839487e-06, - "loss": 1.3526, - "step": 2121 - }, - { - "epoch": 1.72, - "learning_rate": 8.158463753564846e-06, - "loss": 1.2141, - "step": 2122 - }, - { - "epoch": 1.72, - "learning_rate": 8.149877517821834e-06, - "loss": 1.1534, - "step": 2123 - }, - { - "epoch": 1.72, - "learning_rate": 8.141292694163798e-06, - "loss": 1.2557, - "step": 2124 - }, - { - "epoch": 1.72, - "learning_rate": 8.132709289143008e-06, - "loss": 1.2057, - "step": 2125 - }, - { - "epoch": 1.72, - "learning_rate": 8.12412730931065e-06, - "loss": 1.1648, - "step": 2126 - }, - { - "epoch": 1.72, - "learning_rate": 8.115546761216822e-06, - "loss": 1.1236, - "step": 2127 - }, - { - "epoch": 1.72, - "learning_rate": 8.106967651410524e-06, - "loss": 1.2082, - "step": 2128 - }, - { - "epoch": 1.72, - "learning_rate": 8.098389986439671e-06, - "loss": 1.4652, - "step": 2129 - }, - { - "epoch": 1.72, - "learning_rate": 8.089813772851073e-06, - "loss": 1.3103, - "step": 2130 - }, - { - "epoch": 1.72, - "learning_rate": 8.081239017190422e-06, - "loss": 1.3811, - "step": 2131 - }, - { - "epoch": 1.72, - "learning_rate": 8.0726657260023e-06, - "loss": 1.2938, - "step": 2132 - }, - { - "epoch": 1.73, - "learning_rate": 8.064093905830175e-06, - "loss": 1.3933, - "step": 2133 - }, - { - "epoch": 1.73, - "learning_rate": 8.055523563216397e-06, - "loss": 1.1729, - "step": 2134 - }, - { - "epoch": 1.73, - "learning_rate": 8.046954704702176e-06, - "loss": 1.242, - "step": 2135 - }, - { - "epoch": 1.73, - "learning_rate": 8.038387336827602e-06, - "loss": 1.2286, - "step": 2136 - }, - { - "epoch": 1.73, - "learning_rate": 8.029821466131617e-06, - "loss": 1.2198, - "step": 2137 - }, - { - "epoch": 1.73, - "learning_rate": 8.021257099152028e-06, - "loss": 1.2875, - "step": 2138 - }, - { - "epoch": 1.73, - "learning_rate": 8.012694242425485e-06, - "loss": 1.1678, - "step": 2139 - }, - { - "epoch": 1.73, - "learning_rate": 8.004132902487499e-06, - "loss": 1.2509, - "step": 2140 - }, - { - "epoch": 1.73, - "learning_rate": 7.99557308587241e-06, - "loss": 1.27, - "step": 2141 - }, - { - "epoch": 1.73, - "learning_rate": 7.987014799113398e-06, - "loss": 1.251, - "step": 2142 - }, - { - "epoch": 1.73, - "learning_rate": 7.978458048742487e-06, - "loss": 1.173, - "step": 2143 - }, - { - "epoch": 1.73, - "learning_rate": 7.969902841290515e-06, - "loss": 1.1624, - "step": 2144 - }, - { - "epoch": 1.74, - "learning_rate": 7.961349183287146e-06, - "loss": 1.3468, - "step": 2145 - }, - { - "epoch": 1.74, - "learning_rate": 7.952797081260863e-06, - "loss": 1.2805, - "step": 2146 - }, - { - "epoch": 1.74, - "learning_rate": 7.944246541738963e-06, - "loss": 1.2206, - "step": 2147 - }, - { - "epoch": 1.74, - "learning_rate": 7.935697571247545e-06, - "loss": 1.2933, - "step": 2148 - }, - { - "epoch": 1.74, - "learning_rate": 7.927150176311511e-06, - "loss": 1.256, - "step": 2149 - }, - { - "epoch": 1.74, - "learning_rate": 7.91860436345457e-06, - "loss": 1.2133, - "step": 2150 - }, - { - "epoch": 1.74, - "learning_rate": 7.910060139199212e-06, - "loss": 1.3436, - "step": 2151 - }, - { - "epoch": 1.74, - "learning_rate": 7.901517510066724e-06, - "loss": 1.3874, - "step": 2152 - }, - { - "epoch": 1.74, - "learning_rate": 7.892976482577165e-06, - "loss": 1.2281, - "step": 2153 - }, - { - "epoch": 1.74, - "learning_rate": 7.884437063249383e-06, - "loss": 1.3191, - "step": 2154 - }, - { - "epoch": 1.74, - "learning_rate": 7.875899258600986e-06, - "loss": 1.3745, - "step": 2155 - }, - { - "epoch": 1.74, - "learning_rate": 7.867363075148362e-06, - "loss": 1.1763, - "step": 2156 - }, - { - "epoch": 1.75, - "learning_rate": 7.85882851940665e-06, - "loss": 1.1485, - "step": 2157 - }, - { - "epoch": 1.75, - "learning_rate": 7.85029559788976e-06, - "loss": 1.1275, - "step": 2158 - }, - { - "epoch": 1.75, - "learning_rate": 7.841764317110345e-06, - "loss": 1.2011, - "step": 2159 - }, - { - "epoch": 1.75, - "learning_rate": 7.833234683579806e-06, - "loss": 1.2273, - "step": 2160 - }, - { - "epoch": 1.75, - "learning_rate": 7.82470670380829e-06, - "loss": 1.194, - "step": 2161 - }, - { - "epoch": 1.75, - "learning_rate": 7.816180384304684e-06, - "loss": 1.3193, - "step": 2162 - }, - { - "epoch": 1.75, - "learning_rate": 7.8076557315766e-06, - "loss": 1.2212, - "step": 2163 - }, - { - "epoch": 1.75, - "learning_rate": 7.79913275213038e-06, - "loss": 1.3354, - "step": 2164 - }, - { - "epoch": 1.75, - "learning_rate": 7.7906114524711e-06, - "loss": 1.1544, - "step": 2165 - }, - { - "epoch": 1.75, - "learning_rate": 7.782091839102537e-06, - "loss": 1.1905, - "step": 2166 - }, - { - "epoch": 1.75, - "learning_rate": 7.773573918527196e-06, - "loss": 1.225, - "step": 2167 - }, - { - "epoch": 1.75, - "learning_rate": 7.76505769724628e-06, - "loss": 1.3354, - "step": 2168 - }, - { - "epoch": 1.75, - "learning_rate": 7.756543181759696e-06, - "loss": 1.3528, - "step": 2169 - }, - { - "epoch": 1.76, - "learning_rate": 7.748030378566056e-06, - "loss": 1.2948, - "step": 2170 - }, - { - "epoch": 1.76, - "learning_rate": 7.739519294162652e-06, - "loss": 1.2775, - "step": 2171 - }, - { - "epoch": 1.76, - "learning_rate": 7.73100993504548e-06, - "loss": 1.2526, - "step": 2172 - }, - { - "epoch": 1.76, - "learning_rate": 7.722502307709212e-06, - "loss": 1.2959, - "step": 2173 - }, - { - "epoch": 1.76, - "learning_rate": 7.713996418647196e-06, - "loss": 1.2519, - "step": 2174 - }, - { - "epoch": 1.76, - "learning_rate": 7.705492274351457e-06, - "loss": 1.3104, - "step": 2175 - }, - { - "epoch": 1.76, - "learning_rate": 7.696989881312682e-06, - "loss": 1.3216, - "step": 2176 - }, - { - "epoch": 1.76, - "learning_rate": 7.688489246020227e-06, - "loss": 1.3298, - "step": 2177 - }, - { - "epoch": 1.76, - "learning_rate": 7.679990374962106e-06, - "loss": 1.3053, - "step": 2178 - }, - { - "epoch": 1.76, - "learning_rate": 7.67149327462499e-06, - "loss": 1.1486, - "step": 2179 - }, - { - "epoch": 1.76, - "learning_rate": 7.662997951494193e-06, - "loss": 1.1826, - "step": 2180 - }, - { - "epoch": 1.76, - "learning_rate": 7.65450441205367e-06, - "loss": 1.107, - "step": 2181 - }, - { - "epoch": 1.77, - "learning_rate": 7.646012662786017e-06, - "loss": 1.2511, - "step": 2182 - }, - { - "epoch": 1.77, - "learning_rate": 7.637522710172473e-06, - "loss": 1.2768, - "step": 2183 - }, - { - "epoch": 1.77, - "learning_rate": 7.629034560692894e-06, - "loss": 1.2143, - "step": 2184 - }, - { - "epoch": 1.77, - "learning_rate": 7.620548220825759e-06, - "loss": 1.1738, - "step": 2185 - }, - { - "epoch": 1.77, - "learning_rate": 7.612063697048179e-06, - "loss": 1.4226, - "step": 2186 - }, - { - "epoch": 1.77, - "learning_rate": 7.603580995835865e-06, - "loss": 1.3658, - "step": 2187 - }, - { - "epoch": 1.77, - "learning_rate": 7.595100123663143e-06, - "loss": 1.2491, - "step": 2188 - }, - { - "epoch": 1.77, - "learning_rate": 7.586621087002945e-06, - "loss": 1.3785, - "step": 2189 - }, - { - "epoch": 1.77, - "learning_rate": 7.578143892326797e-06, - "loss": 1.1837, - "step": 2190 - }, - { - "epoch": 1.77, - "learning_rate": 7.569668546104823e-06, - "loss": 1.2947, - "step": 2191 - }, - { - "epoch": 1.77, - "learning_rate": 7.561195054805729e-06, - "loss": 1.2039, - "step": 2192 - }, - { - "epoch": 1.77, - "learning_rate": 7.55272342489682e-06, - "loss": 1.2938, - "step": 2193 - }, - { - "epoch": 1.78, - "learning_rate": 7.544253662843966e-06, - "loss": 1.237, - "step": 2194 - }, - { - "epoch": 1.78, - "learning_rate": 7.535785775111619e-06, - "loss": 1.3141, - "step": 2195 - }, - { - "epoch": 1.78, - "learning_rate": 7.527319768162798e-06, - "loss": 1.2452, - "step": 2196 - }, - { - "epoch": 1.78, - "learning_rate": 7.518855648459084e-06, - "loss": 1.2264, - "step": 2197 - }, - { - "epoch": 1.78, - "learning_rate": 7.510393422460623e-06, - "loss": 1.2785, - "step": 2198 - }, - { - "epoch": 1.78, - "learning_rate": 7.50193309662611e-06, - "loss": 1.2435, - "step": 2199 - }, - { - "epoch": 1.78, - "learning_rate": 7.493474677412795e-06, - "loss": 1.2183, - "step": 2200 - }, - { - "epoch": 1.78, - "learning_rate": 7.48501817127647e-06, - "loss": 1.2198, - "step": 2201 - }, - { - "epoch": 1.78, - "learning_rate": 7.47656358467147e-06, - "loss": 1.345, - "step": 2202 - }, - { - "epoch": 1.78, - "learning_rate": 7.468110924050658e-06, - "loss": 1.1803, - "step": 2203 - }, - { - "epoch": 1.78, - "learning_rate": 7.459660195865431e-06, - "loss": 1.3496, - "step": 2204 - }, - { - "epoch": 1.78, - "learning_rate": 7.451211406565716e-06, - "loss": 1.2677, - "step": 2205 - }, - { - "epoch": 1.78, - "learning_rate": 7.442764562599951e-06, - "loss": 1.5747, - "step": 2206 - }, - { - "epoch": 1.79, - "learning_rate": 7.434319670415092e-06, - "loss": 1.3595, - "step": 2207 - }, - { - "epoch": 1.79, - "learning_rate": 7.425876736456614e-06, - "loss": 1.1032, - "step": 2208 - }, - { - "epoch": 1.79, - "learning_rate": 7.417435767168486e-06, - "loss": 1.2115, - "step": 2209 - }, - { - "epoch": 1.79, - "learning_rate": 7.408996768993184e-06, - "loss": 1.2549, - "step": 2210 - }, - { - "epoch": 1.79, - "learning_rate": 7.400559748371676e-06, - "loss": 1.2295, - "step": 2211 - }, - { - "epoch": 1.79, - "learning_rate": 7.3921247117434225e-06, - "loss": 1.2261, - "step": 2212 - }, - { - "epoch": 1.79, - "learning_rate": 7.383691665546368e-06, - "loss": 1.2624, - "step": 2213 - }, - { - "epoch": 1.79, - "learning_rate": 7.375260616216939e-06, - "loss": 1.268, - "step": 2214 - }, - { - "epoch": 1.79, - "learning_rate": 7.366831570190041e-06, - "loss": 1.4602, - "step": 2215 - }, - { - "epoch": 1.79, - "learning_rate": 7.358404533899048e-06, - "loss": 1.2844, - "step": 2216 - }, - { - "epoch": 1.79, - "learning_rate": 7.349979513775799e-06, - "loss": 1.3486, - "step": 2217 - }, - { - "epoch": 1.79, - "learning_rate": 7.341556516250594e-06, - "loss": 1.2786, - "step": 2218 - }, - { - "epoch": 1.8, - "learning_rate": 7.3331355477521886e-06, - "loss": 1.1573, - "step": 2219 - }, - { - "epoch": 1.8, - "learning_rate": 7.324716614707794e-06, - "loss": 1.2455, - "step": 2220 - }, - { - "epoch": 1.8, - "learning_rate": 7.316299723543061e-06, - "loss": 1.3431, - "step": 2221 - }, - { - "epoch": 1.8, - "learning_rate": 7.307884880682095e-06, - "loss": 1.3045, - "step": 2222 - }, - { - "epoch": 1.8, - "learning_rate": 7.2994720925474214e-06, - "loss": 1.3273, - "step": 2223 - }, - { - "epoch": 1.8, - "learning_rate": 7.291061365560009e-06, - "loss": 1.05, - "step": 2224 - }, - { - "epoch": 1.8, - "learning_rate": 7.282652706139246e-06, - "loss": 1.2508, - "step": 2225 - }, - { - "epoch": 1.8, - "learning_rate": 7.274246120702951e-06, - "loss": 1.1524, - "step": 2226 - }, - { - "epoch": 1.8, - "learning_rate": 7.2658416156673525e-06, - "loss": 1.3006, - "step": 2227 - }, - { - "epoch": 1.8, - "learning_rate": 7.257439197447089e-06, - "loss": 1.2965, - "step": 2228 - }, - { - "epoch": 1.8, - "learning_rate": 7.24903887245522e-06, - "loss": 1.1817, - "step": 2229 - }, - { - "epoch": 1.8, - "learning_rate": 7.240640647103192e-06, - "loss": 1.219, - "step": 2230 - }, - { - "epoch": 1.81, - "learning_rate": 7.2322445278008555e-06, - "loss": 1.2045, - "step": 2231 - }, - { - "epoch": 1.81, - "learning_rate": 7.2238505209564565e-06, - "loss": 1.3683, - "step": 2232 - }, - { - "epoch": 1.81, - "learning_rate": 7.215458632976624e-06, - "loss": 1.2616, - "step": 2233 - }, - { - "epoch": 1.81, - "learning_rate": 7.20706887026637e-06, - "loss": 1.1971, - "step": 2234 - }, - { - "epoch": 1.81, - "learning_rate": 7.1986812392290805e-06, - "loss": 1.2818, - "step": 2235 - }, - { - "epoch": 1.81, - "learning_rate": 7.19029574626653e-06, - "loss": 1.1902, - "step": 2236 - }, - { - "epoch": 1.81, - "learning_rate": 7.181912397778844e-06, - "loss": 1.2619, - "step": 2237 - }, - { - "epoch": 1.81, - "learning_rate": 7.173531200164523e-06, - "loss": 1.2743, - "step": 2238 - }, - { - "epoch": 1.81, - "learning_rate": 7.165152159820417e-06, - "loss": 1.1503, - "step": 2239 - }, - { - "epoch": 1.81, - "learning_rate": 7.156775283141733e-06, - "loss": 1.2914, - "step": 2240 - }, - { - "epoch": 1.81, - "learning_rate": 7.148400576522028e-06, - "loss": 1.27, - "step": 2241 - }, - { - "epoch": 1.81, - "learning_rate": 7.140028046353202e-06, - "loss": 1.2119, - "step": 2242 - }, - { - "epoch": 1.81, - "learning_rate": 7.131657699025493e-06, - "loss": 1.2693, - "step": 2243 - }, - { - "epoch": 1.82, - "learning_rate": 7.123289540927476e-06, - "loss": 1.4255, - "step": 2244 - }, - { - "epoch": 1.82, - "learning_rate": 7.114923578446051e-06, - "loss": 1.1576, - "step": 2245 - }, - { - "epoch": 1.82, - "learning_rate": 7.106559817966444e-06, - "loss": 1.2584, - "step": 2246 - }, - { - "epoch": 1.82, - "learning_rate": 7.098198265872198e-06, - "loss": 1.2587, - "step": 2247 - }, - { - "epoch": 1.82, - "learning_rate": 7.089838928545177e-06, - "loss": 1.1544, - "step": 2248 - }, - { - "epoch": 1.82, - "learning_rate": 7.081481812365542e-06, - "loss": 1.2927, - "step": 2249 - }, - { - "epoch": 1.82, - "learning_rate": 7.0731269237117775e-06, - "loss": 1.3326, - "step": 2250 - }, - { - "epoch": 1.82, - "learning_rate": 7.064774268960654e-06, - "loss": 1.3319, - "step": 2251 - }, - { - "epoch": 1.82, - "learning_rate": 7.056423854487236e-06, - "loss": 1.2588, - "step": 2252 - }, - { - "epoch": 1.82, - "learning_rate": 7.048075686664888e-06, - "loss": 1.3291, - "step": 2253 - }, - { - "epoch": 1.82, - "learning_rate": 7.039729771865252e-06, - "loss": 1.2765, - "step": 2254 - }, - { - "epoch": 1.82, - "learning_rate": 7.0313861164582535e-06, - "loss": 1.3025, - "step": 2255 - }, - { - "epoch": 1.83, - "learning_rate": 7.0230447268120905e-06, - "loss": 1.323, - "step": 2256 - }, - { - "epoch": 1.83, - "learning_rate": 7.014705609293232e-06, - "loss": 1.3384, - "step": 2257 - }, - { - "epoch": 1.83, - "learning_rate": 7.006368770266421e-06, - "loss": 1.282, - "step": 2258 - }, - { - "epoch": 1.83, - "learning_rate": 6.998034216094653e-06, - "loss": 1.3316, - "step": 2259 - }, - { - "epoch": 1.83, - "learning_rate": 6.989701953139181e-06, - "loss": 1.2729, - "step": 2260 - }, - { - "epoch": 1.83, - "learning_rate": 6.981371987759511e-06, - "loss": 1.1962, - "step": 2261 - }, - { - "epoch": 1.83, - "learning_rate": 6.973044326313392e-06, - "loss": 1.2742, - "step": 2262 - }, - { - "epoch": 1.83, - "learning_rate": 6.96471897515682e-06, - "loss": 1.1766, - "step": 2263 - }, - { - "epoch": 1.83, - "learning_rate": 6.95639594064402e-06, - "loss": 1.3189, - "step": 2264 - }, - { - "epoch": 1.83, - "learning_rate": 6.94807522912746e-06, - "loss": 1.2807, - "step": 2265 - }, - { - "epoch": 1.83, - "learning_rate": 6.939756846957827e-06, - "loss": 1.3125, - "step": 2266 - }, - { - "epoch": 1.83, - "learning_rate": 6.93144080048403e-06, - "loss": 1.2388, - "step": 2267 - }, - { - "epoch": 1.83, - "learning_rate": 6.923127096053195e-06, - "loss": 1.261, - "step": 2268 - }, - { - "epoch": 1.84, - "learning_rate": 6.914815740010668e-06, - "loss": 1.157, - "step": 2269 - }, - { - "epoch": 1.84, - "learning_rate": 6.906506738699994e-06, - "loss": 1.3809, - "step": 2270 - }, - { - "epoch": 1.84, - "learning_rate": 6.898200098462918e-06, - "loss": 1.3011, - "step": 2271 - }, - { - "epoch": 1.84, - "learning_rate": 6.889895825639401e-06, - "loss": 1.2797, - "step": 2272 - }, - { - "epoch": 1.84, - "learning_rate": 6.881593926567578e-06, - "loss": 1.2227, - "step": 2273 - }, - { - "epoch": 1.84, - "learning_rate": 6.873294407583777e-06, - "loss": 1.1626, - "step": 2274 - }, - { - "epoch": 1.84, - "learning_rate": 6.864997275022518e-06, - "loss": 1.4339, - "step": 2275 - }, - { - "epoch": 1.84, - "learning_rate": 6.856702535216488e-06, - "loss": 1.3233, - "step": 2276 - }, - { - "epoch": 1.84, - "learning_rate": 6.848410194496555e-06, - "loss": 1.2554, - "step": 2277 - }, - { - "epoch": 1.84, - "learning_rate": 6.840120259191751e-06, - "loss": 1.3598, - "step": 2278 - }, - { - "epoch": 1.84, - "learning_rate": 6.831832735629281e-06, - "loss": 1.3138, - "step": 2279 - }, - { - "epoch": 1.84, - "learning_rate": 6.823547630134497e-06, - "loss": 1.2989, - "step": 2280 - }, - { - "epoch": 1.85, - "learning_rate": 6.815264949030917e-06, - "loss": 1.3989, - "step": 2281 - }, - { - "epoch": 1.85, - "learning_rate": 6.806984698640202e-06, - "loss": 1.1697, - "step": 2282 - }, - { - "epoch": 1.85, - "learning_rate": 6.798706885282158e-06, - "loss": 1.2855, - "step": 2283 - }, - { - "epoch": 1.85, - "learning_rate": 6.790431515274729e-06, - "loss": 1.1274, - "step": 2284 - }, - { - "epoch": 1.85, - "learning_rate": 6.782158594934e-06, - "loss": 1.1652, - "step": 2285 - }, - { - "epoch": 1.85, - "learning_rate": 6.773888130574183e-06, - "loss": 1.2293, - "step": 2286 - }, - { - "epoch": 1.85, - "learning_rate": 6.7656201285076195e-06, - "loss": 1.3827, - "step": 2287 - }, - { - "epoch": 1.85, - "learning_rate": 6.757354595044765e-06, - "loss": 1.3252, - "step": 2288 - }, - { - "epoch": 1.85, - "learning_rate": 6.7490915364941955e-06, - "loss": 1.3072, - "step": 2289 - }, - { - "epoch": 1.85, - "learning_rate": 6.740830959162592e-06, - "loss": 1.3423, - "step": 2290 - }, - { - "epoch": 1.85, - "learning_rate": 6.732572869354753e-06, - "loss": 1.2447, - "step": 2291 - }, - { - "epoch": 1.85, - "learning_rate": 6.724317273373564e-06, - "loss": 1.2226, - "step": 2292 - }, - { - "epoch": 1.86, - "learning_rate": 6.716064177520024e-06, - "loss": 1.3086, - "step": 2293 - }, - { - "epoch": 1.86, - "learning_rate": 6.70781358809321e-06, - "loss": 1.1298, - "step": 2294 - }, - { - "epoch": 1.86, - "learning_rate": 6.699565511390289e-06, - "loss": 1.1995, - "step": 2295 - }, - { - "epoch": 1.86, - "learning_rate": 6.691319953706517e-06, - "loss": 1.3186, - "step": 2296 - }, - { - "epoch": 1.86, - "learning_rate": 6.683076921335217e-06, - "loss": 1.2574, - "step": 2297 - }, - { - "epoch": 1.86, - "learning_rate": 6.674836420567793e-06, - "loss": 1.4406, - "step": 2298 - }, - { - "epoch": 1.86, - "learning_rate": 6.666598457693707e-06, - "loss": 1.3022, - "step": 2299 - }, - { - "epoch": 1.86, - "learning_rate": 6.658363039000501e-06, - "loss": 1.3075, - "step": 2300 - }, - { - "epoch": 1.86, - "learning_rate": 6.6501301707737596e-06, - "loss": 1.1677, - "step": 2301 - }, - { - "epoch": 1.86, - "learning_rate": 6.641899859297128e-06, - "loss": 1.3032, - "step": 2302 - }, - { - "epoch": 1.86, - "learning_rate": 6.633672110852295e-06, - "loss": 1.2669, - "step": 2303 - }, - { - "epoch": 1.86, - "learning_rate": 6.625446931718998e-06, - "loss": 1.1005, - "step": 2304 - }, - { - "epoch": 1.86, - "learning_rate": 6.617224328175009e-06, - "loss": 1.3469, - "step": 2305 - }, - { - "epoch": 1.87, - "learning_rate": 6.609004306496139e-06, - "loss": 1.2014, - "step": 2306 - }, - { - "epoch": 1.87, - "learning_rate": 6.600786872956227e-06, - "loss": 1.1978, - "step": 2307 - }, - { - "epoch": 1.87, - "learning_rate": 6.592572033827136e-06, - "loss": 1.3481, - "step": 2308 - }, - { - "epoch": 1.87, - "learning_rate": 6.584359795378747e-06, - "loss": 1.3222, - "step": 2309 - }, - { - "epoch": 1.87, - "learning_rate": 6.57615016387896e-06, - "loss": 1.2362, - "step": 2310 - }, - { - "epoch": 1.87, - "learning_rate": 6.567943145593678e-06, - "loss": 1.341, - "step": 2311 - }, - { - "epoch": 1.87, - "learning_rate": 6.55973874678682e-06, - "loss": 1.22, - "step": 2312 - }, - { - "epoch": 1.87, - "learning_rate": 6.551536973720298e-06, - "loss": 1.1412, - "step": 2313 - }, - { - "epoch": 1.87, - "learning_rate": 6.54333783265402e-06, - "loss": 1.1662, - "step": 2314 - }, - { - "epoch": 1.87, - "learning_rate": 6.535141329845891e-06, - "loss": 1.3961, - "step": 2315 - }, - { - "epoch": 1.87, - "learning_rate": 6.526947471551799e-06, - "loss": 1.3724, - "step": 2316 - }, - { - "epoch": 1.87, - "learning_rate": 6.518756264025608e-06, - "loss": 1.1791, - "step": 2317 - }, - { - "epoch": 1.88, - "learning_rate": 6.510567713519171e-06, - "loss": 1.3833, - "step": 2318 - }, - { - "epoch": 1.88, - "learning_rate": 6.502381826282304e-06, - "loss": 1.1687, - "step": 2319 - }, - { - "epoch": 1.88, - "learning_rate": 6.4941986085627895e-06, - "loss": 1.227, - "step": 2320 - }, - { - "epoch": 1.88, - "learning_rate": 6.4860180666063734e-06, - "loss": 1.2505, - "step": 2321 - }, - { - "epoch": 1.88, - "learning_rate": 6.47784020665677e-06, - "loss": 1.3076, - "step": 2322 - }, - { - "epoch": 1.88, - "learning_rate": 6.469665034955631e-06, - "loss": 1.249, - "step": 2323 - }, - { - "epoch": 1.88, - "learning_rate": 6.461492557742568e-06, - "loss": 1.2093, - "step": 2324 - }, - { - "epoch": 1.88, - "learning_rate": 6.453322781255131e-06, - "loss": 1.3032, - "step": 2325 - }, - { - "epoch": 1.88, - "learning_rate": 6.4451557117288055e-06, - "loss": 1.168, - "step": 2326 - }, - { - "epoch": 1.88, - "learning_rate": 6.436991355397014e-06, - "loss": 1.2965, - "step": 2327 - }, - { - "epoch": 1.88, - "learning_rate": 6.428829718491112e-06, - "loss": 1.4417, - "step": 2328 - }, - { - "epoch": 1.88, - "learning_rate": 6.420670807240376e-06, - "loss": 1.2515, - "step": 2329 - }, - { - "epoch": 1.89, - "learning_rate": 6.412514627872003e-06, - "loss": 1.0659, - "step": 2330 - }, - { - "epoch": 1.89, - "learning_rate": 6.404361186611103e-06, - "loss": 1.2193, - "step": 2331 - }, - { - "epoch": 1.89, - "learning_rate": 6.396210489680699e-06, - "loss": 1.2162, - "step": 2332 - }, - { - "epoch": 1.89, - "learning_rate": 6.3880625433017115e-06, - "loss": 1.1897, - "step": 2333 - }, - { - "epoch": 1.89, - "learning_rate": 6.379917353692977e-06, - "loss": 1.4141, - "step": 2334 - }, - { - "epoch": 1.89, - "learning_rate": 6.371774927071212e-06, - "loss": 1.1874, - "step": 2335 - }, - { - "epoch": 1.89, - "learning_rate": 6.363635269651037e-06, - "loss": 1.2022, - "step": 2336 - }, - { - "epoch": 1.89, - "learning_rate": 6.355498387644949e-06, - "loss": 1.2049, - "step": 2337 - }, - { - "epoch": 1.89, - "learning_rate": 6.347364287263331e-06, - "loss": 1.3088, - "step": 2338 - }, - { - "epoch": 1.89, - "learning_rate": 6.339232974714447e-06, - "loss": 1.2532, - "step": 2339 - }, - { - "epoch": 1.89, - "learning_rate": 6.331104456204423e-06, - "loss": 1.3595, - "step": 2340 - }, - { - "epoch": 1.89, - "learning_rate": 6.322978737937263e-06, - "loss": 1.2719, - "step": 2341 - }, - { - "epoch": 1.89, - "learning_rate": 6.314855826114823e-06, - "loss": 1.2401, - "step": 2342 - }, - { - "epoch": 1.9, - "learning_rate": 6.306735726936832e-06, - "loss": 1.3146, - "step": 2343 - }, - { - "epoch": 1.9, - "learning_rate": 6.298618446600856e-06, - "loss": 1.4177, - "step": 2344 - }, - { - "epoch": 1.9, - "learning_rate": 6.290503991302324e-06, - "loss": 1.3257, - "step": 2345 - }, - { - "epoch": 1.9, - "learning_rate": 6.282392367234495e-06, - "loss": 1.2679, - "step": 2346 - }, - { - "epoch": 1.9, - "learning_rate": 6.274283580588479e-06, - "loss": 1.3343, - "step": 2347 - }, - { - "epoch": 1.9, - "learning_rate": 6.266177637553211e-06, - "loss": 1.3779, - "step": 2348 - }, - { - "epoch": 1.9, - "learning_rate": 6.2580745443154624e-06, - "loss": 1.4191, - "step": 2349 - }, - { - "epoch": 1.9, - "learning_rate": 6.249974307059826e-06, - "loss": 1.3653, - "step": 2350 - }, - { - "epoch": 1.9, - "learning_rate": 6.24187693196872e-06, - "loss": 1.3717, - "step": 2351 - }, - { - "epoch": 1.9, - "learning_rate": 6.23378242522237e-06, - "loss": 1.261, - "step": 2352 - }, - { - "epoch": 1.9, - "learning_rate": 6.225690792998818e-06, - "loss": 1.3418, - "step": 2353 - }, - { - "epoch": 1.9, - "learning_rate": 6.217602041473907e-06, - "loss": 1.4191, - "step": 2354 - }, - { - "epoch": 1.91, - "learning_rate": 6.209516176821289e-06, - "loss": 1.1256, - "step": 2355 - }, - { - "epoch": 1.91, - "learning_rate": 6.201433205212403e-06, - "loss": 1.3266, - "step": 2356 - }, - { - "epoch": 1.91, - "learning_rate": 6.193353132816493e-06, - "loss": 1.2958, - "step": 2357 - }, - { - "epoch": 1.91, - "learning_rate": 6.185275965800579e-06, - "loss": 1.3118, - "step": 2358 - }, - { - "epoch": 1.91, - "learning_rate": 6.177201710329466e-06, - "loss": 1.1899, - "step": 2359 - }, - { - "epoch": 1.91, - "learning_rate": 6.169130372565737e-06, - "loss": 1.2168, - "step": 2360 - }, - { - "epoch": 1.91, - "learning_rate": 6.161061958669752e-06, - "loss": 1.2175, - "step": 2361 - }, - { - "epoch": 1.91, - "learning_rate": 6.152996474799633e-06, - "loss": 1.1926, - "step": 2362 - }, - { - "epoch": 1.91, - "learning_rate": 6.1449339271112715e-06, - "loss": 1.2349, - "step": 2363 - }, - { - "epoch": 1.91, - "learning_rate": 6.136874321758311e-06, - "loss": 1.1138, - "step": 2364 - }, - { - "epoch": 1.91, - "learning_rate": 6.128817664892161e-06, - "loss": 1.2257, - "step": 2365 - }, - { - "epoch": 1.91, - "learning_rate": 6.1207639626619675e-06, - "loss": 1.1387, - "step": 2366 - }, - { - "epoch": 1.92, - "learning_rate": 6.1127132212146326e-06, - "loss": 1.4322, - "step": 2367 - }, - { - "epoch": 1.92, - "learning_rate": 6.1046654466947885e-06, - "loss": 1.3023, - "step": 2368 - }, - { - "epoch": 1.92, - "learning_rate": 6.096620645244808e-06, - "loss": 1.3513, - "step": 2369 - }, - { - "epoch": 1.92, - "learning_rate": 6.088578823004796e-06, - "loss": 1.3451, - "step": 2370 - }, - { - "epoch": 1.92, - "learning_rate": 6.080539986112581e-06, - "loss": 1.1805, - "step": 2371 - }, - { - "epoch": 1.92, - "learning_rate": 6.072504140703714e-06, - "loss": 1.1541, - "step": 2372 - }, - { - "epoch": 1.92, - "learning_rate": 6.0644712929114665e-06, - "loss": 1.3276, - "step": 2373 - }, - { - "epoch": 1.92, - "learning_rate": 6.056441448866817e-06, - "loss": 1.2355, - "step": 2374 - }, - { - "epoch": 1.92, - "learning_rate": 6.0484146146984475e-06, - "loss": 1.2892, - "step": 2375 - }, - { - "epoch": 1.92, - "learning_rate": 6.040390796532751e-06, - "loss": 1.1339, - "step": 2376 - }, - { - "epoch": 1.92, - "learning_rate": 6.032370000493819e-06, - "loss": 1.3633, - "step": 2377 - }, - { - "epoch": 1.92, - "learning_rate": 6.024352232703427e-06, - "loss": 1.3364, - "step": 2378 - }, - { - "epoch": 1.92, - "learning_rate": 6.016337499281051e-06, - "loss": 1.1378, - "step": 2379 - }, - { - "epoch": 1.93, - "learning_rate": 6.008325806343842e-06, - "loss": 1.2612, - "step": 2380 - }, - { - "epoch": 1.93, - "learning_rate": 6.000317160006637e-06, - "loss": 1.1573, - "step": 2381 - }, - { - "epoch": 1.93, - "learning_rate": 5.9923115663819365e-06, - "loss": 1.3269, - "step": 2382 - }, - { - "epoch": 1.93, - "learning_rate": 5.9843090315799294e-06, - "loss": 1.2674, - "step": 2383 - }, - { - "epoch": 1.93, - "learning_rate": 5.976309561708452e-06, - "loss": 1.1373, - "step": 2384 - }, - { - "epoch": 1.93, - "learning_rate": 5.968313162873007e-06, - "loss": 1.2568, - "step": 2385 - }, - { - "epoch": 1.93, - "learning_rate": 5.9603198411767646e-06, - "loss": 1.2131, - "step": 2386 - }, - { - "epoch": 1.93, - "learning_rate": 5.9523296027205305e-06, - "loss": 1.2476, - "step": 2387 - }, - { - "epoch": 1.93, - "learning_rate": 5.944342453602764e-06, - "loss": 1.0683, - "step": 2388 - }, - { - "epoch": 1.93, - "learning_rate": 5.936358399919566e-06, - "loss": 1.3532, - "step": 2389 - }, - { - "epoch": 1.93, - "learning_rate": 5.9283774477646775e-06, - "loss": 1.2798, - "step": 2390 - }, - { - "epoch": 1.93, - "learning_rate": 5.920399603229463e-06, - "loss": 1.1823, - "step": 2391 - }, - { - "epoch": 1.94, - "learning_rate": 5.912424872402928e-06, - "loss": 1.3543, - "step": 2392 - }, - { - "epoch": 1.94, - "learning_rate": 5.904453261371692e-06, - "loss": 1.3989, - "step": 2393 - }, - { - "epoch": 1.94, - "learning_rate": 5.896484776219999e-06, - "loss": 1.2263, - "step": 2394 - }, - { - "epoch": 1.94, - "learning_rate": 5.888519423029706e-06, - "loss": 1.2623, - "step": 2395 - }, - { - "epoch": 1.94, - "learning_rate": 5.880557207880275e-06, - "loss": 1.3281, - "step": 2396 - }, - { - "epoch": 1.94, - "learning_rate": 5.872598136848775e-06, - "loss": 1.2752, - "step": 2397 - }, - { - "epoch": 1.94, - "learning_rate": 5.86464221600988e-06, - "loss": 1.22, - "step": 2398 - }, - { - "epoch": 1.94, - "learning_rate": 5.856689451435853e-06, - "loss": 1.1712, - "step": 2399 - }, - { - "epoch": 1.94, - "learning_rate": 5.848739849196556e-06, - "loss": 1.2013, - "step": 2400 - }, - { - "epoch": 1.94, - "learning_rate": 5.840793415359431e-06, - "loss": 1.1786, - "step": 2401 - }, - { - "epoch": 1.94, - "learning_rate": 5.8328501559895e-06, - "loss": 1.1271, - "step": 2402 - }, - { - "epoch": 1.94, - "learning_rate": 5.824910077149372e-06, - "loss": 1.3429, - "step": 2403 - }, - { - "epoch": 1.94, - "learning_rate": 5.816973184899213e-06, - "loss": 1.2005, - "step": 2404 - }, - { - "epoch": 1.95, - "learning_rate": 5.809039485296772e-06, - "loss": 1.2689, - "step": 2405 - }, - { - "epoch": 1.95, - "learning_rate": 5.801108984397355e-06, - "loss": 1.1279, - "step": 2406 - }, - { - "epoch": 1.95, - "learning_rate": 5.793181688253824e-06, - "loss": 1.1911, - "step": 2407 - }, - { - "epoch": 1.95, - "learning_rate": 5.785257602916603e-06, - "loss": 1.0759, - "step": 2408 - }, - { - "epoch": 1.95, - "learning_rate": 5.777336734433654e-06, - "loss": 1.3562, - "step": 2409 - }, - { - "epoch": 1.95, - "learning_rate": 5.7694190888504964e-06, - "loss": 1.2889, - "step": 2410 - }, - { - "epoch": 1.95, - "learning_rate": 5.761504672210174e-06, - "loss": 1.1882, - "step": 2411 - }, - { - "epoch": 1.95, - "learning_rate": 5.753593490553282e-06, - "loss": 1.2115, - "step": 2412 - }, - { - "epoch": 1.95, - "learning_rate": 5.745685549917943e-06, - "loss": 1.4138, - "step": 2413 - }, - { - "epoch": 1.95, - "learning_rate": 5.737780856339791e-06, - "loss": 1.1659, - "step": 2414 - }, - { - "epoch": 1.95, - "learning_rate": 5.729879415852009e-06, - "loss": 1.1213, - "step": 2415 - }, - { - "epoch": 1.95, - "learning_rate": 5.72198123448527e-06, - "loss": 1.18, - "step": 2416 - }, - { - "epoch": 1.96, - "learning_rate": 5.714086318267776e-06, - "loss": 1.2609, - "step": 2417 - }, - { - "epoch": 1.96, - "learning_rate": 5.706194673225234e-06, - "loss": 1.2662, - "step": 2418 - }, - { - "epoch": 1.96, - "learning_rate": 5.698306305380848e-06, - "loss": 1.2722, - "step": 2419 - }, - { - "epoch": 1.96, - "learning_rate": 5.690421220755329e-06, - "loss": 1.166, - "step": 2420 - }, - { - "epoch": 1.96, - "learning_rate": 5.682539425366869e-06, - "loss": 1.3241, - "step": 2421 - }, - { - "epoch": 1.96, - "learning_rate": 5.674660925231171e-06, - "loss": 1.2351, - "step": 2422 - }, - { - "epoch": 1.96, - "learning_rate": 5.66678572636141e-06, - "loss": 1.2609, - "step": 2423 - }, - { - "epoch": 1.96, - "learning_rate": 5.658913834768234e-06, - "loss": 1.3402, - "step": 2424 - }, - { - "epoch": 1.96, - "learning_rate": 5.651045256459782e-06, - "loss": 1.3034, - "step": 2425 - }, - { - "epoch": 1.96, - "learning_rate": 5.643179997441651e-06, - "loss": 1.3236, - "step": 2426 - }, - { - "epoch": 1.96, - "learning_rate": 5.6353180637169135e-06, - "loss": 1.2588, - "step": 2427 - }, - { - "epoch": 1.96, - "learning_rate": 5.6274594612861024e-06, - "loss": 1.4581, - "step": 2428 - }, - { - "epoch": 1.97, - "learning_rate": 5.619604196147203e-06, - "loss": 1.2467, - "step": 2429 - }, - { - "epoch": 1.97, - "learning_rate": 5.611752274295665e-06, - "loss": 1.3019, - "step": 2430 - }, - { - "epoch": 1.97, - "learning_rate": 5.603903701724371e-06, - "loss": 1.2771, - "step": 2431 - }, - { - "epoch": 1.97, - "learning_rate": 5.5960584844236565e-06, - "loss": 1.3705, - "step": 2432 - }, - { - "epoch": 1.97, - "learning_rate": 5.5882166283813e-06, - "loss": 1.2907, - "step": 2433 - }, - { - "epoch": 1.97, - "learning_rate": 5.580378139582498e-06, - "loss": 1.1505, - "step": 2434 - }, - { - "epoch": 1.97, - "learning_rate": 5.572543024009897e-06, - "loss": 1.1563, - "step": 2435 - }, - { - "epoch": 1.97, - "learning_rate": 5.564711287643558e-06, - "loss": 1.3778, - "step": 2436 - }, - { - "epoch": 1.97, - "learning_rate": 5.556882936460966e-06, - "loss": 1.3097, - "step": 2437 - }, - { - "epoch": 1.97, - "learning_rate": 5.549057976437018e-06, - "loss": 1.2873, - "step": 2438 - }, - { - "epoch": 1.97, - "learning_rate": 5.541236413544027e-06, - "loss": 1.172, - "step": 2439 - }, - { - "epoch": 1.97, - "learning_rate": 5.533418253751714e-06, - "loss": 1.2579, - "step": 2440 - }, - { - "epoch": 1.97, - "learning_rate": 5.525603503027197e-06, - "loss": 1.2874, - "step": 2441 - }, - { - "epoch": 1.98, - "learning_rate": 5.517792167334999e-06, - "loss": 1.2548, - "step": 2442 - }, - { - "epoch": 1.98, - "learning_rate": 5.50998425263703e-06, - "loss": 1.2928, - "step": 2443 - }, - { - "epoch": 1.98, - "learning_rate": 5.502179764892597e-06, - "loss": 1.2571, - "step": 2444 - }, - { - "epoch": 1.98, - "learning_rate": 5.4943787100583855e-06, - "loss": 1.2911, - "step": 2445 - }, - { - "epoch": 1.98, - "learning_rate": 5.486581094088458e-06, - "loss": 1.274, - "step": 2446 - }, - { - "epoch": 1.98, - "learning_rate": 5.478786922934263e-06, - "loss": 1.3367, - "step": 2447 - }, - { - "epoch": 1.98, - "learning_rate": 5.470996202544604e-06, - "loss": 1.2255, - "step": 2448 - }, - { - "epoch": 1.98, - "learning_rate": 5.463208938865665e-06, - "loss": 1.3066, - "step": 2449 - }, - { - "epoch": 1.98, - "learning_rate": 5.455425137840987e-06, - "loss": 1.2197, - "step": 2450 - }, - { - "epoch": 1.98, - "learning_rate": 5.447644805411467e-06, - "loss": 1.1085, - "step": 2451 - }, - { - "epoch": 1.98, - "learning_rate": 5.43986794751536e-06, - "loss": 1.4356, - "step": 2452 - }, - { - "epoch": 1.98, - "learning_rate": 5.432094570088256e-06, - "loss": 1.3747, - "step": 2453 - }, - { - "epoch": 1.99, - "learning_rate": 5.424324679063102e-06, - "loss": 1.2649, - "step": 2454 - }, - { - "epoch": 1.99, - "learning_rate": 5.416558280370181e-06, - "loss": 1.3871, - "step": 2455 - }, - { - "epoch": 1.99, - "learning_rate": 5.408795379937097e-06, - "loss": 1.1373, - "step": 2456 - }, - { - "epoch": 1.99, - "learning_rate": 5.401035983688815e-06, - "loss": 1.2575, - "step": 2457 - }, - { - "epoch": 1.99, - "learning_rate": 5.3932800975475905e-06, - "loss": 1.274, - "step": 2458 - }, - { - "epoch": 1.99, - "learning_rate": 5.385527727433024e-06, - "loss": 1.2572, - "step": 2459 - }, - { - "epoch": 1.99, - "learning_rate": 5.377778879262017e-06, - "loss": 1.253, - "step": 2460 - }, - { - "epoch": 1.99, - "learning_rate": 5.370033558948793e-06, - "loss": 1.2804, - "step": 2461 - }, - { - "epoch": 1.99, - "learning_rate": 5.362291772404884e-06, - "loss": 1.3168, - "step": 2462 - }, - { - "epoch": 1.99, - "learning_rate": 5.3545535255391104e-06, - "loss": 1.1645, - "step": 2463 - }, - { - "epoch": 1.99, - "learning_rate": 5.346818824257614e-06, - "loss": 1.3627, - "step": 2464 - }, - { - "epoch": 1.99, - "learning_rate": 5.3390876744638075e-06, - "loss": 1.2285, - "step": 2465 - }, - { - "epoch": 2.0, - "learning_rate": 5.331360082058407e-06, - "loss": 1.2659, - "step": 2466 - }, - { - "epoch": 2.0, - "learning_rate": 5.323636052939416e-06, - "loss": 1.6618, - "step": 2467 - }, - { - "epoch": 2.0, - "learning_rate": 5.3159155930021e-06, - "loss": 1.118, - "step": 2468 - }, - { - "epoch": 2.0, - "learning_rate": 5.308198708139023e-06, - "loss": 1.2527, - "step": 2469 - }, - { - "epoch": 2.0, - "learning_rate": 5.300485404239999e-06, - "loss": 1.2897, - "step": 2470 - }, - { - "epoch": 2.0, - "learning_rate": 5.292775687192123e-06, - "loss": 1.2892, - "step": 2471 - }, - { - "epoch": 2.0, - "learning_rate": 5.285069562879759e-06, - "loss": 1.2559, - "step": 2472 - }, - { - "epoch": 2.0, - "learning_rate": 5.277367037184509e-06, - "loss": 1.1998, - "step": 2473 - }, - { - "epoch": 2.0, - "learning_rate": 5.269668115985243e-06, - "loss": 1.1651, - "step": 2474 - }, - { - "epoch": 2.0, - "learning_rate": 5.261972805158069e-06, - "loss": 1.0898, - "step": 2475 - }, - { - "epoch": 2.0, - "learning_rate": 5.254281110576349e-06, - "loss": 1.2899, - "step": 2476 - }, - { - "epoch": 2.0, - "learning_rate": 5.246593038110688e-06, - "loss": 1.1918, - "step": 2477 - }, - { - "epoch": 2.0, - "learning_rate": 5.238908593628904e-06, - "loss": 1.151, - "step": 2478 - }, - { - "epoch": 2.01, - "learning_rate": 5.2312277829960804e-06, - "loss": 1.1847, - "step": 2479 - }, - { - "epoch": 2.01, - "learning_rate": 5.223550612074497e-06, - "loss": 1.2245, - "step": 2480 - }, - { - "epoch": 2.01, - "learning_rate": 5.21587708672367e-06, - "loss": 1.1795, - "step": 2481 - }, - { - "epoch": 2.01, - "learning_rate": 5.208207212800335e-06, - "loss": 1.3079, - "step": 2482 - }, - { - "epoch": 2.01, - "learning_rate": 5.200540996158429e-06, - "loss": 1.1204, - "step": 2483 - }, - { - "epoch": 2.01, - "learning_rate": 5.19287844264911e-06, - "loss": 1.1726, - "step": 2484 - }, - { - "epoch": 2.01, - "learning_rate": 5.1852195581207275e-06, - "loss": 1.1541, - "step": 2485 - }, - { - "epoch": 2.01, - "learning_rate": 5.177564348418851e-06, - "loss": 1.1888, - "step": 2486 - }, - { - "epoch": 2.01, - "learning_rate": 5.16991281938622e-06, - "loss": 1.103, - "step": 2487 - }, - { - "epoch": 2.01, - "learning_rate": 5.162264976862784e-06, - "loss": 1.2111, - "step": 2488 - }, - { - "epoch": 2.01, - "learning_rate": 5.154620826685674e-06, - "loss": 1.2294, - "step": 2489 - }, - { - "epoch": 2.01, - "learning_rate": 5.146980374689192e-06, - "loss": 1.1673, - "step": 2490 - }, - { - "epoch": 2.02, - "learning_rate": 5.139343626704831e-06, - "loss": 1.217, - "step": 2491 - }, - { - "epoch": 2.02, - "learning_rate": 5.131710588561253e-06, - "loss": 1.2711, - "step": 2492 - }, - { - "epoch": 2.02, - "learning_rate": 5.124081266084287e-06, - "loss": 1.2833, - "step": 2493 - }, - { - "epoch": 2.02, - "learning_rate": 5.116455665096931e-06, - "loss": 1.1294, - "step": 2494 - }, - { - "epoch": 2.02, - "learning_rate": 5.10883379141933e-06, - "loss": 1.2041, - "step": 2495 - }, - { - "epoch": 2.02, - "learning_rate": 5.1012156508688e-06, - "loss": 1.2086, - "step": 2496 - }, - { - "epoch": 2.02, - "learning_rate": 5.093601249259794e-06, - "loss": 1.2484, - "step": 2497 - }, - { - "epoch": 2.02, - "learning_rate": 5.08599059240392e-06, - "loss": 1.3159, - "step": 2498 - }, - { - "epoch": 2.02, - "learning_rate": 5.078383686109927e-06, - "loss": 1.239, - "step": 2499 - }, - { - "epoch": 2.02, - "learning_rate": 5.070780536183698e-06, - "loss": 1.1855, - "step": 2500 - }, - { - "epoch": 2.02, - "learning_rate": 5.0631811484282555e-06, - "loss": 1.1216, - "step": 2501 - }, - { - "epoch": 2.02, - "learning_rate": 5.055585528643739e-06, - "loss": 1.2015, - "step": 2502 - }, - { - "epoch": 2.03, - "learning_rate": 5.047993682627421e-06, - "loss": 1.1623, - "step": 2503 - }, - { - "epoch": 2.03, - "learning_rate": 5.040405616173698e-06, - "loss": 1.1716, - "step": 2504 - }, - { - "epoch": 2.03, - "learning_rate": 5.032821335074065e-06, - "loss": 1.3491, - "step": 2505 - }, - { - "epoch": 2.03, - "learning_rate": 5.025240845117145e-06, - "loss": 1.2104, - "step": 2506 - }, - { - "epoch": 2.03, - "learning_rate": 5.01766415208866e-06, - "loss": 1.0785, - "step": 2507 - }, - { - "epoch": 2.03, - "learning_rate": 5.0100912617714345e-06, - "loss": 1.1319, - "step": 2508 - }, - { - "epoch": 2.03, - "learning_rate": 5.002522179945395e-06, - "loss": 1.1395, - "step": 2509 - }, - { - "epoch": 2.03, - "learning_rate": 4.99495691238755e-06, - "loss": 1.1652, - "step": 2510 - }, - { - "epoch": 2.03, - "learning_rate": 4.987395464872011e-06, - "loss": 1.2205, - "step": 2511 - }, - { - "epoch": 2.03, - "learning_rate": 4.979837843169959e-06, - "loss": 1.0703, - "step": 2512 - }, - { - "epoch": 2.03, - "learning_rate": 4.972284053049668e-06, - "loss": 0.9383, - "step": 2513 - }, - { - "epoch": 2.03, - "learning_rate": 4.964734100276481e-06, - "loss": 1.1527, - "step": 2514 - }, - { - "epoch": 2.03, - "learning_rate": 4.957187990612816e-06, - "loss": 1.3221, - "step": 2515 - }, - { - "epoch": 2.04, - "learning_rate": 4.949645729818158e-06, - "loss": 1.2647, - "step": 2516 - }, - { - "epoch": 2.04, - "learning_rate": 4.942107323649043e-06, - "loss": 1.1507, - "step": 2517 - }, - { - "epoch": 2.04, - "learning_rate": 4.934572777859081e-06, - "loss": 1.1016, - "step": 2518 - }, - { - "epoch": 2.04, - "learning_rate": 4.9270420981989295e-06, - "loss": 1.1606, - "step": 2519 - }, - { - "epoch": 2.04, - "learning_rate": 4.9195152904162865e-06, - "loss": 1.122, - "step": 2520 - }, - { - "epoch": 2.04, - "learning_rate": 4.911992360255908e-06, - "loss": 1.1159, - "step": 2521 - }, - { - "epoch": 2.04, - "learning_rate": 4.904473313459586e-06, - "loss": 1.2182, - "step": 2522 - }, - { - "epoch": 2.04, - "learning_rate": 4.896958155766147e-06, - "loss": 1.2111, - "step": 2523 - }, - { - "epoch": 2.04, - "learning_rate": 4.889446892911447e-06, - "loss": 1.2957, - "step": 2524 - }, - { - "epoch": 2.04, - "learning_rate": 4.881939530628372e-06, - "loss": 1.1382, - "step": 2525 - }, - { - "epoch": 2.04, - "learning_rate": 4.874436074646839e-06, - "loss": 1.302, - "step": 2526 - }, - { - "epoch": 2.04, - "learning_rate": 4.866936530693763e-06, - "loss": 1.2031, - "step": 2527 - }, - { - "epoch": 2.05, - "learning_rate": 4.859440904493093e-06, - "loss": 1.1108, - "step": 2528 - }, - { - "epoch": 2.05, - "learning_rate": 4.851949201765781e-06, - "loss": 1.1194, - "step": 2529 - }, - { - "epoch": 2.05, - "learning_rate": 4.844461428229782e-06, - "loss": 1.0807, - "step": 2530 - }, - { - "epoch": 2.05, - "learning_rate": 4.836977589600058e-06, - "loss": 1.3026, - "step": 2531 - }, - { - "epoch": 2.05, - "learning_rate": 4.829497691588557e-06, - "loss": 1.2146, - "step": 2532 - }, - { - "epoch": 2.05, - "learning_rate": 4.822021739904235e-06, - "loss": 1.1443, - "step": 2533 - }, - { - "epoch": 2.05, - "learning_rate": 4.814549740253017e-06, - "loss": 1.1694, - "step": 2534 - }, - { - "epoch": 2.05, - "learning_rate": 4.8070816983378285e-06, - "loss": 1.2121, - "step": 2535 - }, - { - "epoch": 2.05, - "learning_rate": 4.7996176198585655e-06, - "loss": 1.2164, - "step": 2536 - }, - { - "epoch": 2.05, - "learning_rate": 4.7921575105121024e-06, - "loss": 1.055, - "step": 2537 - }, - { - "epoch": 2.05, - "learning_rate": 4.784701375992285e-06, - "loss": 1.3024, - "step": 2538 - }, - { - "epoch": 2.05, - "learning_rate": 4.777249221989918e-06, - "loss": 1.1489, - "step": 2539 - }, - { - "epoch": 2.06, - "learning_rate": 4.769801054192776e-06, - "loss": 1.1593, - "step": 2540 - }, - { - "epoch": 2.06, - "learning_rate": 4.762356878285592e-06, - "loss": 1.1282, - "step": 2541 - }, - { - "epoch": 2.06, - "learning_rate": 4.754916699950038e-06, - "loss": 1.2205, - "step": 2542 - }, - { - "epoch": 2.06, - "learning_rate": 4.7474805248647604e-06, - "loss": 1.1642, - "step": 2543 - }, - { - "epoch": 2.06, - "learning_rate": 4.740048358705323e-06, - "loss": 1.217, - "step": 2544 - }, - { - "epoch": 2.06, - "learning_rate": 4.732620207144253e-06, - "loss": 0.9738, - "step": 2545 - }, - { - "epoch": 2.06, - "learning_rate": 4.7251960758509915e-06, - "loss": 1.1254, - "step": 2546 - }, - { - "epoch": 2.06, - "learning_rate": 4.7177759704919295e-06, - "loss": 1.2478, - "step": 2547 - }, - { - "epoch": 2.06, - "learning_rate": 4.710359896730379e-06, - "loss": 1.1425, - "step": 2548 - }, - { - "epoch": 2.06, - "learning_rate": 4.702947860226564e-06, - "loss": 1.2718, - "step": 2549 - }, - { - "epoch": 2.06, - "learning_rate": 4.695539866637653e-06, - "loss": 1.1795, - "step": 2550 - }, - { - "epoch": 2.06, - "learning_rate": 4.6881359216177e-06, - "loss": 1.2617, - "step": 2551 - }, - { - "epoch": 2.06, - "learning_rate": 4.680736030817687e-06, - "loss": 1.2789, - "step": 2552 - }, - { - "epoch": 2.07, - "learning_rate": 4.673340199885501e-06, - "loss": 1.1125, - "step": 2553 - }, - { - "epoch": 2.07, - "learning_rate": 4.665948434465917e-06, - "loss": 1.3381, - "step": 2554 - }, - { - "epoch": 2.07, - "learning_rate": 4.658560740200622e-06, - "loss": 1.3991, - "step": 2555 - }, - { - "epoch": 2.07, - "learning_rate": 4.6511771227281835e-06, - "loss": 1.2726, - "step": 2556 - }, - { - "epoch": 2.07, - "learning_rate": 4.643797587684072e-06, - "loss": 1.1761, - "step": 2557 - }, - { - "epoch": 2.07, - "learning_rate": 4.636422140700631e-06, - "loss": 1.249, - "step": 2558 - }, - { - "epoch": 2.07, - "learning_rate": 4.629050787407085e-06, - "loss": 1.1933, - "step": 2559 - }, - { - "epoch": 2.07, - "learning_rate": 4.6216835334295385e-06, - "loss": 1.2821, - "step": 2560 - }, - { - "epoch": 2.07, - "learning_rate": 4.614320384390959e-06, - "loss": 1.2444, - "step": 2561 - }, - { - "epoch": 2.07, - "learning_rate": 4.60696134591119e-06, - "loss": 1.1475, - "step": 2562 - }, - { - "epoch": 2.07, - "learning_rate": 4.599606423606933e-06, - "loss": 1.1273, - "step": 2563 - }, - { - "epoch": 2.07, - "learning_rate": 4.59225562309175e-06, - "loss": 1.1582, - "step": 2564 - }, - { - "epoch": 2.08, - "learning_rate": 4.58490894997606e-06, - "loss": 1.286, - "step": 2565 - }, - { - "epoch": 2.08, - "learning_rate": 4.57756640986712e-06, - "loss": 1.1656, - "step": 2566 - }, - { - "epoch": 2.08, - "learning_rate": 4.570228008369043e-06, - "loss": 1.1137, - "step": 2567 - }, - { - "epoch": 2.08, - "learning_rate": 4.562893751082786e-06, - "loss": 1.1325, - "step": 2568 - }, - { - "epoch": 2.08, - "learning_rate": 4.55556364360613e-06, - "loss": 1.0912, - "step": 2569 - }, - { - "epoch": 2.08, - "learning_rate": 4.548237691533699e-06, - "loss": 1.1227, - "step": 2570 - }, - { - "epoch": 2.08, - "learning_rate": 4.540915900456943e-06, - "loss": 1.2314, - "step": 2571 - }, - { - "epoch": 2.08, - "learning_rate": 4.533598275964139e-06, - "loss": 1.3098, - "step": 2572 - }, - { - "epoch": 2.08, - "learning_rate": 4.526284823640375e-06, - "loss": 1.1927, - "step": 2573 - }, - { - "epoch": 2.08, - "learning_rate": 4.518975549067565e-06, - "loss": 1.2935, - "step": 2574 - }, - { - "epoch": 2.08, - "learning_rate": 4.51167045782443e-06, - "loss": 1.1796, - "step": 2575 - }, - { - "epoch": 2.08, - "learning_rate": 4.504369555486493e-06, - "loss": 1.2814, - "step": 2576 - }, - { - "epoch": 2.08, - "learning_rate": 4.497072847626087e-06, - "loss": 1.2588, - "step": 2577 - }, - { - "epoch": 2.09, - "learning_rate": 4.489780339812343e-06, - "loss": 1.1779, - "step": 2578 - }, - { - "epoch": 2.09, - "learning_rate": 4.482492037611184e-06, - "loss": 1.3332, - "step": 2579 - }, - { - "epoch": 2.09, - "learning_rate": 4.475207946585328e-06, - "loss": 1.117, - "step": 2580 - }, - { - "epoch": 2.09, - "learning_rate": 4.467928072294266e-06, - "loss": 1.2159, - "step": 2581 - }, - { - "epoch": 2.09, - "learning_rate": 4.460652420294289e-06, - "loss": 1.3787, - "step": 2582 - }, - { - "epoch": 2.09, - "learning_rate": 4.453380996138446e-06, - "loss": 1.2608, - "step": 2583 - }, - { - "epoch": 2.09, - "learning_rate": 4.446113805376575e-06, - "loss": 1.2294, - "step": 2584 - }, - { - "epoch": 2.09, - "learning_rate": 4.438850853555275e-06, - "loss": 1.2438, - "step": 2585 - }, - { - "epoch": 2.09, - "learning_rate": 4.431592146217914e-06, - "loss": 1.3712, - "step": 2586 - }, - { - "epoch": 2.09, - "learning_rate": 4.424337688904617e-06, - "loss": 1.2531, - "step": 2587 - }, - { - "epoch": 2.09, - "learning_rate": 4.417087487152263e-06, - "loss": 1.1473, - "step": 2588 - }, - { - "epoch": 2.09, - "learning_rate": 4.409841546494489e-06, - "loss": 1.1117, - "step": 2589 - }, - { - "epoch": 2.1, - "learning_rate": 4.402599872461678e-06, - "loss": 0.9537, - "step": 2590 - }, - { - "epoch": 2.1, - "learning_rate": 4.395362470580952e-06, - "loss": 1.2554, - "step": 2591 - }, - { - "epoch": 2.1, - "learning_rate": 4.388129346376177e-06, - "loss": 1.0302, - "step": 2592 - }, - { - "epoch": 2.1, - "learning_rate": 4.380900505367953e-06, - "loss": 1.1459, - "step": 2593 - }, - { - "epoch": 2.1, - "learning_rate": 4.373675953073613e-06, - "loss": 1.2976, - "step": 2594 - }, - { - "epoch": 2.1, - "learning_rate": 4.366455695007215e-06, - "loss": 1.1769, - "step": 2595 - }, - { - "epoch": 2.1, - "learning_rate": 4.359239736679534e-06, - "loss": 1.1067, - "step": 2596 - }, - { - "epoch": 2.1, - "learning_rate": 4.352028083598074e-06, - "loss": 1.0579, - "step": 2597 - }, - { - "epoch": 2.1, - "learning_rate": 4.344820741267042e-06, - "loss": 1.1556, - "step": 2598 - }, - { - "epoch": 2.1, - "learning_rate": 4.337617715187361e-06, - "loss": 1.1279, - "step": 2599 - }, - { - "epoch": 2.1, - "learning_rate": 4.330419010856661e-06, - "loss": 1.2146, - "step": 2600 - }, - { - "epoch": 2.1, - "learning_rate": 4.323224633769272e-06, - "loss": 1.2498, - "step": 2601 - }, - { - "epoch": 2.11, - "learning_rate": 4.3160345894162225e-06, - "loss": 1.1894, - "step": 2602 - }, - { - "epoch": 2.11, - "learning_rate": 4.3088488832852245e-06, - "loss": 1.1117, - "step": 2603 - }, - { - "epoch": 2.11, - "learning_rate": 4.301667520860693e-06, - "loss": 1.2099, - "step": 2604 - }, - { - "epoch": 2.11, - "learning_rate": 4.294490507623722e-06, - "loss": 1.1551, - "step": 2605 - }, - { - "epoch": 2.11, - "learning_rate": 4.287317849052075e-06, - "loss": 1.2329, - "step": 2606 - }, - { - "epoch": 2.11, - "learning_rate": 4.280149550620218e-06, - "loss": 1.1825, - "step": 2607 - }, - { - "epoch": 2.11, - "learning_rate": 4.272985617799262e-06, - "loss": 1.1043, - "step": 2608 - }, - { - "epoch": 2.11, - "learning_rate": 4.265826056057006e-06, - "loss": 1.275, - "step": 2609 - }, - { - "epoch": 2.11, - "learning_rate": 4.258670870857894e-06, - "loss": 1.2871, - "step": 2610 - }, - { - "epoch": 2.11, - "learning_rate": 4.251520067663044e-06, - "loss": 1.2538, - "step": 2611 - }, - { - "epoch": 2.11, - "learning_rate": 4.244373651930232e-06, - "loss": 1.2276, - "step": 2612 - }, - { - "epoch": 2.11, - "learning_rate": 4.237231629113863e-06, - "loss": 1.273, - "step": 2613 - }, - { - "epoch": 2.11, - "learning_rate": 4.2300940046650244e-06, - "loss": 1.1729, - "step": 2614 - }, - { - "epoch": 2.12, - "learning_rate": 4.222960784031413e-06, - "loss": 1.3797, - "step": 2615 - }, - { - "epoch": 2.12, - "learning_rate": 4.215831972657383e-06, - "loss": 1.2014, - "step": 2616 - }, - { - "epoch": 2.12, - "learning_rate": 4.208707575983921e-06, - "loss": 1.1187, - "step": 2617 - }, - { - "epoch": 2.12, - "learning_rate": 4.201587599448638e-06, - "loss": 1.2762, - "step": 2618 - }, - { - "epoch": 2.12, - "learning_rate": 4.194472048485778e-06, - "loss": 1.0266, - "step": 2619 - }, - { - "epoch": 2.12, - "learning_rate": 4.187360928526198e-06, - "loss": 1.1903, - "step": 2620 - }, - { - "epoch": 2.12, - "learning_rate": 4.180254244997391e-06, - "loss": 1.153, - "step": 2621 - }, - { - "epoch": 2.12, - "learning_rate": 4.173152003323442e-06, - "loss": 1.29, - "step": 2622 - }, - { - "epoch": 2.12, - "learning_rate": 4.16605420892506e-06, - "loss": 1.0924, - "step": 2623 - }, - { - "epoch": 2.12, - "learning_rate": 4.158960867219561e-06, - "loss": 1.1327, - "step": 2624 - }, - { - "epoch": 2.12, - "learning_rate": 4.15187198362085e-06, - "loss": 1.0018, - "step": 2625 - }, - { - "epoch": 2.12, - "learning_rate": 4.1447875635394384e-06, - "loss": 1.1622, - "step": 2626 - }, - { - "epoch": 2.13, - "learning_rate": 4.137707612382436e-06, - "loss": 1.0979, - "step": 2627 - }, - { - "epoch": 2.13, - "learning_rate": 4.130632135553522e-06, - "loss": 1.1179, - "step": 2628 - }, - { - "epoch": 2.13, - "learning_rate": 4.123561138452988e-06, - "loss": 1.1846, - "step": 2629 - }, - { - "epoch": 2.13, - "learning_rate": 4.116494626477684e-06, - "loss": 0.9, - "step": 2630 - }, - { - "epoch": 2.13, - "learning_rate": 4.10943260502105e-06, - "loss": 1.2842, - "step": 2631 - }, - { - "epoch": 2.13, - "learning_rate": 4.102375079473086e-06, - "loss": 1.1916, - "step": 2632 - }, - { - "epoch": 2.13, - "learning_rate": 4.095322055220374e-06, - "loss": 1.1887, - "step": 2633 - }, - { - "epoch": 2.13, - "learning_rate": 4.088273537646058e-06, - "loss": 1.3723, - "step": 2634 - }, - { - "epoch": 2.13, - "learning_rate": 4.081229532129826e-06, - "loss": 1.1113, - "step": 2635 - }, - { - "epoch": 2.13, - "learning_rate": 4.074190044047952e-06, - "loss": 1.244, - "step": 2636 - }, - { - "epoch": 2.13, - "learning_rate": 4.067155078773233e-06, - "loss": 1.2092, - "step": 2637 - }, - { - "epoch": 2.13, - "learning_rate": 4.0601246416750304e-06, - "loss": 1.092, - "step": 2638 - }, - { - "epoch": 2.14, - "learning_rate": 4.05309873811925e-06, - "loss": 1.2013, - "step": 2639 - }, - { - "epoch": 2.14, - "learning_rate": 4.046077373468325e-06, - "loss": 1.1136, - "step": 2640 - }, - { - "epoch": 2.14, - "learning_rate": 4.039060553081238e-06, - "loss": 1.1546, - "step": 2641 - }, - { - "epoch": 2.14, - "learning_rate": 4.032048282313489e-06, - "loss": 1.2175, - "step": 2642 - }, - { - "epoch": 2.14, - "learning_rate": 4.025040566517123e-06, - "loss": 1.0243, - "step": 2643 - }, - { - "epoch": 2.14, - "learning_rate": 4.0180374110407005e-06, - "loss": 1.1583, - "step": 2644 - }, - { - "epoch": 2.14, - "learning_rate": 4.011038821229294e-06, - "loss": 1.1083, - "step": 2645 - }, - { - "epoch": 2.14, - "learning_rate": 4.004044802424504e-06, - "loss": 1.1891, - "step": 2646 - }, - { - "epoch": 2.14, - "learning_rate": 3.997055359964428e-06, - "loss": 1.2297, - "step": 2647 - }, - { - "epoch": 2.14, - "learning_rate": 3.9900704991836824e-06, - "loss": 1.1716, - "step": 2648 - }, - { - "epoch": 2.14, - "learning_rate": 3.983090225413384e-06, - "loss": 1.1815, - "step": 2649 - }, - { - "epoch": 2.14, - "learning_rate": 3.976114543981148e-06, - "loss": 1.303, - "step": 2650 - }, - { - "epoch": 2.14, - "learning_rate": 3.969143460211084e-06, - "loss": 1.3235, - "step": 2651 - }, - { - "epoch": 2.15, - "learning_rate": 3.96217697942379e-06, - "loss": 1.0828, - "step": 2652 - }, - { - "epoch": 2.15, - "learning_rate": 3.9552151069363545e-06, - "loss": 1.0734, - "step": 2653 - }, - { - "epoch": 2.15, - "learning_rate": 3.948257848062351e-06, - "loss": 1.0794, - "step": 2654 - }, - { - "epoch": 2.15, - "learning_rate": 3.941305208111822e-06, - "loss": 1.3172, - "step": 2655 - }, - { - "epoch": 2.15, - "learning_rate": 3.934357192391294e-06, - "loss": 1.2133, - "step": 2656 - }, - { - "epoch": 2.15, - "learning_rate": 3.927413806203761e-06, - "loss": 1.2618, - "step": 2657 - }, - { - "epoch": 2.15, - "learning_rate": 3.920475054848687e-06, - "loss": 1.296, - "step": 2658 - }, - { - "epoch": 2.15, - "learning_rate": 3.91354094362199e-06, - "loss": 1.0353, - "step": 2659 - }, - { - "epoch": 2.15, - "learning_rate": 3.906611477816054e-06, - "loss": 1.2593, - "step": 2660 - }, - { - "epoch": 2.15, - "learning_rate": 3.899686662719717e-06, - "loss": 1.1429, - "step": 2661 - }, - { - "epoch": 2.15, - "learning_rate": 3.892766503618261e-06, - "loss": 1.1587, - "step": 2662 - }, - { - "epoch": 2.15, - "learning_rate": 3.88585100579342e-06, - "loss": 1.1894, - "step": 2663 - }, - { - "epoch": 2.16, - "learning_rate": 3.878940174523371e-06, - "loss": 1.2002, - "step": 2664 - }, - { - "epoch": 2.16, - "learning_rate": 3.872034015082727e-06, - "loss": 1.1127, - "step": 2665 - }, - { - "epoch": 2.16, - "learning_rate": 3.865132532742541e-06, - "loss": 1.1687, - "step": 2666 - }, - { - "epoch": 2.16, - "learning_rate": 3.85823573277028e-06, - "loss": 1.254, - "step": 2667 - }, - { - "epoch": 2.16, - "learning_rate": 3.851343620429858e-06, - "loss": 1.2035, - "step": 2668 - }, - { - "epoch": 2.16, - "learning_rate": 3.844456200981595e-06, - "loss": 1.0783, - "step": 2669 - }, - { - "epoch": 2.16, - "learning_rate": 3.837573479682236e-06, - "loss": 1.1539, - "step": 2670 - }, - { - "epoch": 2.16, - "learning_rate": 3.8306954617849445e-06, - "loss": 1.1888, - "step": 2671 - }, - { - "epoch": 2.16, - "learning_rate": 3.823822152539286e-06, - "loss": 1.2228, - "step": 2672 - }, - { - "epoch": 2.16, - "learning_rate": 3.81695355719124e-06, - "loss": 1.324, - "step": 2673 - }, - { - "epoch": 2.16, - "learning_rate": 3.810089680983178e-06, - "loss": 1.2208, - "step": 2674 - }, - { - "epoch": 2.16, - "learning_rate": 3.8032305291538775e-06, - "loss": 1.2254, - "step": 2675 - }, - { - "epoch": 2.17, - "learning_rate": 3.796376106938513e-06, - "loss": 1.2654, - "step": 2676 - }, - { - "epoch": 2.17, - "learning_rate": 3.7895264195686366e-06, - "loss": 1.1646, - "step": 2677 - }, - { - "epoch": 2.17, - "learning_rate": 3.782681472272198e-06, - "loss": 0.9526, - "step": 2678 - }, - { - "epoch": 2.17, - "learning_rate": 3.775841270273527e-06, - "loss": 1.343, - "step": 2679 - }, - { - "epoch": 2.17, - "learning_rate": 3.769005818793329e-06, - "loss": 1.2781, - "step": 2680 - }, - { - "epoch": 2.17, - "learning_rate": 3.762175123048687e-06, - "loss": 1.2978, - "step": 2681 - }, - { - "epoch": 2.17, - "learning_rate": 3.7553491882530456e-06, - "loss": 1.1272, - "step": 2682 - }, - { - "epoch": 2.17, - "learning_rate": 3.748528019616231e-06, - "loss": 1.1456, - "step": 2683 - }, - { - "epoch": 2.17, - "learning_rate": 3.741711622344414e-06, - "loss": 1.1215, - "step": 2684 - }, - { - "epoch": 2.17, - "learning_rate": 3.734900001640135e-06, - "loss": 1.2508, - "step": 2685 - }, - { - "epoch": 2.17, - "learning_rate": 3.7280931627022876e-06, - "loss": 1.2225, - "step": 2686 - }, - { - "epoch": 2.17, - "learning_rate": 3.721291110726115e-06, - "loss": 1.1982, - "step": 2687 - }, - { - "epoch": 2.17, - "learning_rate": 3.714493850903208e-06, - "loss": 1.0766, - "step": 2688 - }, - { - "epoch": 2.18, - "learning_rate": 3.707701388421491e-06, - "loss": 1.148, - "step": 2689 - }, - { - "epoch": 2.18, - "learning_rate": 3.7009137284652386e-06, - "loss": 1.2755, - "step": 2690 - }, - { - "epoch": 2.18, - "learning_rate": 3.6941308762150575e-06, - "loss": 1.1433, - "step": 2691 - }, - { - "epoch": 2.18, - "learning_rate": 3.687352836847874e-06, - "loss": 1.2337, - "step": 2692 - }, - { - "epoch": 2.18, - "learning_rate": 3.680579615536961e-06, - "loss": 1.2624, - "step": 2693 - }, - { - "epoch": 2.18, - "learning_rate": 3.6738112174518936e-06, - "loss": 1.1671, - "step": 2694 - }, - { - "epoch": 2.18, - "learning_rate": 3.6670476477585827e-06, - "loss": 1.3304, - "step": 2695 - }, - { - "epoch": 2.18, - "learning_rate": 3.6602889116192374e-06, - "loss": 1.123, - "step": 2696 - }, - { - "epoch": 2.18, - "learning_rate": 3.653535014192391e-06, - "loss": 1.1937, - "step": 2697 - }, - { - "epoch": 2.18, - "learning_rate": 3.6467859606328825e-06, - "loss": 1.2665, - "step": 2698 - }, - { - "epoch": 2.18, - "learning_rate": 3.640041756091839e-06, - "loss": 1.1759, - "step": 2699 - }, - { - "epoch": 2.18, - "learning_rate": 3.633302405716712e-06, - "loss": 1.1955, - "step": 2700 - }, - { - "epoch": 2.19, - "learning_rate": 3.6265679146512256e-06, - "loss": 1.1659, - "step": 2701 - }, - { - "epoch": 2.19, - "learning_rate": 3.6198382880354043e-06, - "loss": 1.2522, - "step": 2702 - }, - { - "epoch": 2.19, - "learning_rate": 3.6131135310055653e-06, - "loss": 1.0931, - "step": 2703 - }, - { - "epoch": 2.19, - "learning_rate": 3.606393648694294e-06, - "loss": 1.1952, - "step": 2704 - }, - { - "epoch": 2.19, - "learning_rate": 3.5996786462304724e-06, - "loss": 1.1516, - "step": 2705 - }, - { - "epoch": 2.19, - "learning_rate": 3.592968528739239e-06, - "loss": 1.2166, - "step": 2706 - }, - { - "epoch": 2.19, - "learning_rate": 3.586263301342029e-06, - "loss": 1.1735, - "step": 2707 - }, - { - "epoch": 2.19, - "learning_rate": 3.5795629691565205e-06, - "loss": 1.2276, - "step": 2708 - }, - { - "epoch": 2.19, - "learning_rate": 3.572867537296669e-06, - "loss": 1.1303, - "step": 2709 - }, - { - "epoch": 2.19, - "learning_rate": 3.5661770108726914e-06, - "loss": 1.1399, - "step": 2710 - }, - { - "epoch": 2.19, - "learning_rate": 3.559491394991048e-06, - "loss": 1.3359, - "step": 2711 - }, - { - "epoch": 2.19, - "learning_rate": 3.5528106947544626e-06, - "loss": 1.1199, - "step": 2712 - }, - { - "epoch": 2.19, - "learning_rate": 3.546134915261905e-06, - "loss": 1.199, - "step": 2713 - }, - { - "epoch": 2.2, - "learning_rate": 3.5394640616085874e-06, - "loss": 1.2683, - "step": 2714 - }, - { - "epoch": 2.2, - "learning_rate": 3.5327981388859657e-06, - "loss": 1.0963, - "step": 2715 - }, - { - "epoch": 2.2, - "learning_rate": 3.5261371521817247e-06, - "loss": 1.3118, - "step": 2716 - }, - { - "epoch": 2.2, - "learning_rate": 3.5194811065797917e-06, - "loss": 1.2364, - "step": 2717 - }, - { - "epoch": 2.2, - "learning_rate": 3.5128300071603117e-06, - "loss": 1.1456, - "step": 2718 - }, - { - "epoch": 2.2, - "learning_rate": 3.5061838589996654e-06, - "loss": 1.0908, - "step": 2719 - }, - { - "epoch": 2.2, - "learning_rate": 3.4995426671704493e-06, - "loss": 1.2468, - "step": 2720 - }, - { - "epoch": 2.2, - "learning_rate": 3.4929064367414767e-06, - "loss": 1.0807, - "step": 2721 - }, - { - "epoch": 2.2, - "learning_rate": 3.48627517277778e-06, - "loss": 1.0754, - "step": 2722 - }, - { - "epoch": 2.2, - "learning_rate": 3.479648880340588e-06, - "loss": 1.1362, - "step": 2723 - }, - { - "epoch": 2.2, - "learning_rate": 3.473027564487347e-06, - "loss": 1.2362, - "step": 2724 - }, - { - "epoch": 2.2, - "learning_rate": 3.4664112302717047e-06, - "loss": 1.2529, - "step": 2725 - }, - { - "epoch": 2.21, - "learning_rate": 3.4597998827434964e-06, - "loss": 1.1742, - "step": 2726 - }, - { - "epoch": 2.21, - "learning_rate": 3.4531935269487614e-06, - "loss": 1.1954, - "step": 2727 - }, - { - "epoch": 2.21, - "learning_rate": 3.446592167929725e-06, - "loss": 1.0174, - "step": 2728 - }, - { - "epoch": 2.21, - "learning_rate": 3.439995810724801e-06, - "loss": 1.1354, - "step": 2729 - }, - { - "epoch": 2.21, - "learning_rate": 3.433404460368587e-06, - "loss": 1.2362, - "step": 2730 - }, - { - "epoch": 2.21, - "learning_rate": 3.4268181218918486e-06, - "loss": 1.1494, - "step": 2731 - }, - { - "epoch": 2.21, - "learning_rate": 3.4202368003215404e-06, - "loss": 1.2693, - "step": 2732 - }, - { - "epoch": 2.21, - "learning_rate": 3.4136605006807768e-06, - "loss": 1.2324, - "step": 2733 - }, - { - "epoch": 2.21, - "learning_rate": 3.4070892279888444e-06, - "loss": 1.234, - "step": 2734 - }, - { - "epoch": 2.21, - "learning_rate": 3.4005229872611944e-06, - "loss": 1.0152, - "step": 2735 - }, - { - "epoch": 2.21, - "learning_rate": 3.3939617835094353e-06, - "loss": 1.1861, - "step": 2736 - }, - { - "epoch": 2.21, - "learning_rate": 3.3874056217413333e-06, - "loss": 1.1596, - "step": 2737 - }, - { - "epoch": 2.22, - "learning_rate": 3.3808545069608e-06, - "loss": 1.3704, - "step": 2738 - }, - { - "epoch": 2.22, - "learning_rate": 3.3743084441679007e-06, - "loss": 1.0827, - "step": 2739 - }, - { - "epoch": 2.22, - "learning_rate": 3.3677674383588476e-06, - "loss": 1.1825, - "step": 2740 - }, - { - "epoch": 2.22, - "learning_rate": 3.3612314945259827e-06, - "loss": 1.2536, - "step": 2741 - }, - { - "epoch": 2.22, - "learning_rate": 3.3547006176577934e-06, - "loss": 1.1954, - "step": 2742 - }, - { - "epoch": 2.22, - "learning_rate": 3.348174812738899e-06, - "loss": 1.2612, - "step": 2743 - }, - { - "epoch": 2.22, - "learning_rate": 3.341654084750047e-06, - "loss": 1.181, - "step": 2744 - }, - { - "epoch": 2.22, - "learning_rate": 3.335138438668104e-06, - "loss": 1.167, - "step": 2745 - }, - { - "epoch": 2.22, - "learning_rate": 3.3286278794660665e-06, - "loss": 1.0838, - "step": 2746 - }, - { - "epoch": 2.22, - "learning_rate": 3.322122412113047e-06, - "loss": 1.184, - "step": 2747 - }, - { - "epoch": 2.22, - "learning_rate": 3.3156220415742624e-06, - "loss": 1.0462, - "step": 2748 - }, - { - "epoch": 2.22, - "learning_rate": 3.309126772811052e-06, - "loss": 1.3089, - "step": 2749 - }, - { - "epoch": 2.22, - "learning_rate": 3.302636610780855e-06, - "loss": 1.1065, - "step": 2750 - }, - { - "epoch": 2.23, - "learning_rate": 3.296151560437214e-06, - "loss": 1.1528, - "step": 2751 - }, - { - "epoch": 2.23, - "learning_rate": 3.2896716267297725e-06, - "loss": 1.4117, - "step": 2752 - }, - { - "epoch": 2.23, - "learning_rate": 3.2831968146042613e-06, - "loss": 1.1412, - "step": 2753 - }, - { - "epoch": 2.23, - "learning_rate": 3.2767271290025127e-06, - "loss": 1.377, - "step": 2754 - }, - { - "epoch": 2.23, - "learning_rate": 3.270262574862434e-06, - "loss": 1.1565, - "step": 2755 - }, - { - "epoch": 2.23, - "learning_rate": 3.2638031571180263e-06, - "loss": 1.2675, - "step": 2756 - }, - { - "epoch": 2.23, - "learning_rate": 3.257348880699367e-06, - "loss": 1.2377, - "step": 2757 - }, - { - "epoch": 2.23, - "learning_rate": 3.250899750532609e-06, - "loss": 1.0762, - "step": 2758 - }, - { - "epoch": 2.23, - "learning_rate": 3.2444557715399816e-06, - "loss": 1.1364, - "step": 2759 - }, - { - "epoch": 2.23, - "learning_rate": 3.238016948639772e-06, - "loss": 1.1972, - "step": 2760 - }, - { - "epoch": 2.23, - "learning_rate": 3.2315832867463413e-06, - "loss": 1.2602, - "step": 2761 - }, - { - "epoch": 2.23, - "learning_rate": 3.225154790770112e-06, - "loss": 1.0987, - "step": 2762 - }, - { - "epoch": 2.24, - "learning_rate": 3.2187314656175517e-06, - "loss": 1.2276, - "step": 2763 - }, - { - "epoch": 2.24, - "learning_rate": 3.2123133161912035e-06, - "loss": 1.1046, - "step": 2764 - }, - { - "epoch": 2.24, - "learning_rate": 3.2059003473896353e-06, - "loss": 1.0262, - "step": 2765 - }, - { - "epoch": 2.24, - "learning_rate": 3.1994925641074793e-06, - "loss": 1.1489, - "step": 2766 - }, - { - "epoch": 2.24, - "learning_rate": 3.1930899712354037e-06, - "loss": 1.3458, - "step": 2767 - }, - { - "epoch": 2.24, - "learning_rate": 3.1866925736601097e-06, - "loss": 1.2207, - "step": 2768 - }, - { - "epoch": 2.24, - "learning_rate": 3.1803003762643445e-06, - "loss": 1.2397, - "step": 2769 - }, - { - "epoch": 2.24, - "learning_rate": 3.1739133839268698e-06, - "loss": 1.1192, - "step": 2770 - }, - { - "epoch": 2.24, - "learning_rate": 3.1675316015225e-06, - "loss": 1.2752, - "step": 2771 - }, - { - "epoch": 2.24, - "learning_rate": 3.1611550339220455e-06, - "loss": 1.3605, - "step": 2772 - }, - { - "epoch": 2.24, - "learning_rate": 3.1547836859923553e-06, - "loss": 1.2283, - "step": 2773 - }, - { - "epoch": 2.24, - "learning_rate": 3.148417562596289e-06, - "loss": 1.2635, - "step": 2774 - }, - { - "epoch": 2.25, - "learning_rate": 3.1420566685927125e-06, - "loss": 1.2242, - "step": 2775 - }, - { - "epoch": 2.25, - "learning_rate": 3.1357010088365104e-06, - "loss": 1.1617, - "step": 2776 - }, - { - "epoch": 2.25, - "learning_rate": 3.129350588178566e-06, - "loss": 1.2321, - "step": 2777 - }, - { - "epoch": 2.25, - "learning_rate": 3.123005411465766e-06, - "loss": 1.1145, - "step": 2778 - }, - { - "epoch": 2.25, - "learning_rate": 3.1166654835409982e-06, - "loss": 1.1382, - "step": 2779 - }, - { - "epoch": 2.25, - "learning_rate": 3.110330809243134e-06, - "loss": 0.9494, - "step": 2780 - }, - { - "epoch": 2.25, - "learning_rate": 3.104001393407049e-06, - "loss": 1.1705, - "step": 2781 - }, - { - "epoch": 2.25, - "learning_rate": 3.0976772408635904e-06, - "loss": 1.2002, - "step": 2782 - }, - { - "epoch": 2.25, - "learning_rate": 3.091358356439599e-06, - "loss": 1.2612, - "step": 2783 - }, - { - "epoch": 2.25, - "learning_rate": 3.085044744957896e-06, - "loss": 1.3506, - "step": 2784 - }, - { - "epoch": 2.25, - "learning_rate": 3.078736411237262e-06, - "loss": 1.0218, - "step": 2785 - }, - { - "epoch": 2.25, - "learning_rate": 3.072433360092476e-06, - "loss": 1.1304, - "step": 2786 - }, - { - "epoch": 2.25, - "learning_rate": 3.066135596334259e-06, - "loss": 1.1392, - "step": 2787 - }, - { - "epoch": 2.26, - "learning_rate": 3.059843124769312e-06, - "loss": 1.2022, - "step": 2788 - }, - { - "epoch": 2.26, - "learning_rate": 3.0535559502002944e-06, - "loss": 1.1627, - "step": 2789 - }, - { - "epoch": 2.26, - "learning_rate": 3.0472740774258157e-06, - "loss": 1.2965, - "step": 2790 - }, - { - "epoch": 2.26, - "learning_rate": 3.0409975112404512e-06, - "loss": 1.2052, - "step": 2791 - }, - { - "epoch": 2.26, - "learning_rate": 3.0347262564347057e-06, - "loss": 1.3182, - "step": 2792 - }, - { - "epoch": 2.26, - "learning_rate": 3.02846031779506e-06, - "loss": 1.1393, - "step": 2793 - }, - { - "epoch": 2.26, - "learning_rate": 3.0221997001039083e-06, - "loss": 1.2262, - "step": 2794 - }, - { - "epoch": 2.26, - "learning_rate": 3.0159444081395994e-06, - "loss": 1.0854, - "step": 2795 - }, - { - "epoch": 2.26, - "learning_rate": 3.009694446676419e-06, - "loss": 1.1661, - "step": 2796 - }, - { - "epoch": 2.26, - "learning_rate": 3.003449820484571e-06, - "loss": 1.1382, - "step": 2797 - }, - { - "epoch": 2.26, - "learning_rate": 2.997210534330198e-06, - "loss": 1.2788, - "step": 2798 - }, - { - "epoch": 2.26, - "learning_rate": 2.9909765929753674e-06, - "loss": 1.2479, - "step": 2799 - }, - { - "epoch": 2.27, - "learning_rate": 2.9847480011780607e-06, - "loss": 1.1286, - "step": 2800 - }, - { - "epoch": 2.27, - "learning_rate": 2.9785247636921853e-06, - "loss": 1.2937, - "step": 2801 - }, - { - "epoch": 2.27, - "learning_rate": 2.9723068852675495e-06, - "loss": 1.1684, - "step": 2802 - }, - { - "epoch": 2.27, - "learning_rate": 2.9660943706498847e-06, - "loss": 1.058, - "step": 2803 - }, - { - "epoch": 2.27, - "learning_rate": 2.959887224580815e-06, - "loss": 1.1207, - "step": 2804 - }, - { - "epoch": 2.27, - "learning_rate": 2.9536854517978784e-06, - "loss": 1.2968, - "step": 2805 - }, - { - "epoch": 2.27, - "learning_rate": 2.9474890570345083e-06, - "loss": 1.1937, - "step": 2806 - }, - { - "epoch": 2.27, - "learning_rate": 2.9412980450200303e-06, - "loss": 1.175, - "step": 2807 - }, - { - "epoch": 2.27, - "learning_rate": 2.93511242047967e-06, - "loss": 1.1662, - "step": 2808 - }, - { - "epoch": 2.27, - "learning_rate": 2.9289321881345257e-06, - "loss": 1.081, - "step": 2809 - }, - { - "epoch": 2.27, - "learning_rate": 2.922757352701595e-06, - "loss": 1.2353, - "step": 2810 - }, - { - "epoch": 2.27, - "learning_rate": 2.916587918893753e-06, - "loss": 1.1912, - "step": 2811 - }, - { - "epoch": 2.28, - "learning_rate": 2.9104238914197447e-06, - "loss": 0.9942, - "step": 2812 - }, - { - "epoch": 2.28, - "learning_rate": 2.904265274984196e-06, - "loss": 1.2722, - "step": 2813 - }, - { - "epoch": 2.28, - "learning_rate": 2.8981120742876014e-06, - "loss": 1.1474, - "step": 2814 - }, - { - "epoch": 2.28, - "learning_rate": 2.8919642940263214e-06, - "loss": 1.1204, - "step": 2815 - }, - { - "epoch": 2.28, - "learning_rate": 2.8858219388925836e-06, - "loss": 1.3213, - "step": 2816 - }, - { - "epoch": 2.28, - "learning_rate": 2.8796850135744627e-06, - "loss": 1.1777, - "step": 2817 - }, - { - "epoch": 2.28, - "learning_rate": 2.8735535227559054e-06, - "loss": 1.2276, - "step": 2818 - }, - { - "epoch": 2.28, - "learning_rate": 2.867427471116695e-06, - "loss": 1.1322, - "step": 2819 - }, - { - "epoch": 2.28, - "learning_rate": 2.861306863332475e-06, - "loss": 1.3159, - "step": 2820 - }, - { - "epoch": 2.28, - "learning_rate": 2.855191704074728e-06, - "loss": 1.1568, - "step": 2821 - }, - { - "epoch": 2.28, - "learning_rate": 2.849081998010782e-06, - "loss": 1.2192, - "step": 2822 - }, - { - "epoch": 2.28, - "learning_rate": 2.8429777498038023e-06, - "loss": 1.2619, - "step": 2823 - }, - { - "epoch": 2.28, - "learning_rate": 2.8368789641127814e-06, - "loss": 1.0891, - "step": 2824 - }, - { - "epoch": 2.29, - "learning_rate": 2.8307856455925508e-06, - "loss": 1.2036, - "step": 2825 - }, - { - "epoch": 2.29, - "learning_rate": 2.824697798893772e-06, - "loss": 1.5487, - "step": 2826 - }, - { - "epoch": 2.29, - "learning_rate": 2.8186154286629118e-06, - "loss": 1.1545, - "step": 2827 - }, - { - "epoch": 2.29, - "learning_rate": 2.8125385395422856e-06, - "loss": 1.1918, - "step": 2828 - }, - { - "epoch": 2.29, - "learning_rate": 2.8064671361700003e-06, - "loss": 1.172, - "step": 2829 - }, - { - "epoch": 2.29, - "learning_rate": 2.8004012231799905e-06, - "loss": 1.2229, - "step": 2830 - }, - { - "epoch": 2.29, - "learning_rate": 2.794340805201989e-06, - "loss": 1.2055, - "step": 2831 - }, - { - "epoch": 2.29, - "learning_rate": 2.788285886861547e-06, - "loss": 1.0384, - "step": 2832 - }, - { - "epoch": 2.29, - "learning_rate": 2.7822364727800123e-06, - "loss": 1.3245, - "step": 2833 - }, - { - "epoch": 2.29, - "learning_rate": 2.776192567574524e-06, - "loss": 1.0756, - "step": 2834 - }, - { - "epoch": 2.29, - "learning_rate": 2.7701541758580343e-06, - "loss": 1.2646, - "step": 2835 - }, - { - "epoch": 2.29, - "learning_rate": 2.7641213022392698e-06, - "loss": 1.2008, - "step": 2836 - }, - { - "epoch": 2.3, - "learning_rate": 2.7580939513227556e-06, - "loss": 1.3135, - "step": 2837 - }, - { - "epoch": 2.3, - "learning_rate": 2.7520721277088023e-06, - "loss": 1.148, - "step": 2838 - }, - { - "epoch": 2.3, - "learning_rate": 2.7460558359934907e-06, - "loss": 1.0834, - "step": 2839 - }, - { - "epoch": 2.3, - "learning_rate": 2.740045080768694e-06, - "loss": 1.1726, - "step": 2840 - }, - { - "epoch": 2.3, - "learning_rate": 2.734039866622048e-06, - "loss": 1.0611, - "step": 2841 - }, - { - "epoch": 2.3, - "learning_rate": 2.728040198136965e-06, - "loss": 1.1233, - "step": 2842 - }, - { - "epoch": 2.3, - "learning_rate": 2.722046079892626e-06, - "loss": 1.1032, - "step": 2843 - }, - { - "epoch": 2.3, - "learning_rate": 2.7160575164639722e-06, - "loss": 1.3142, - "step": 2844 - }, - { - "epoch": 2.3, - "learning_rate": 2.710074512421711e-06, - "loss": 1.1082, - "step": 2845 - }, - { - "epoch": 2.3, - "learning_rate": 2.7040970723322945e-06, - "loss": 1.2092, - "step": 2846 - }, - { - "epoch": 2.3, - "learning_rate": 2.6981252007579395e-06, - "loss": 1.1862, - "step": 2847 - }, - { - "epoch": 2.3, - "learning_rate": 2.692158902256613e-06, - "loss": 1.2653, - "step": 2848 - }, - { - "epoch": 2.31, - "learning_rate": 2.686198181382016e-06, - "loss": 1.1028, - "step": 2849 - }, - { - "epoch": 2.31, - "learning_rate": 2.6802430426836113e-06, - "loss": 1.2936, - "step": 2850 - }, - { - "epoch": 2.31, - "learning_rate": 2.6742934907065833e-06, - "loss": 1.1231, - "step": 2851 - }, - { - "epoch": 2.31, - "learning_rate": 2.668349529991865e-06, - "loss": 1.2234, - "step": 2852 - }, - { - "epoch": 2.31, - "learning_rate": 2.662411165076112e-06, - "loss": 1.0622, - "step": 2853 - }, - { - "epoch": 2.31, - "learning_rate": 2.6564784004917156e-06, - "loss": 1.1404, - "step": 2854 - }, - { - "epoch": 2.31, - "learning_rate": 2.650551240766794e-06, - "loss": 1.1333, - "step": 2855 - }, - { - "epoch": 2.31, - "learning_rate": 2.644629690425178e-06, - "loss": 1.1608, - "step": 2856 - }, - { - "epoch": 2.31, - "learning_rate": 2.6387137539864326e-06, - "loss": 1.2512, - "step": 2857 - }, - { - "epoch": 2.31, - "learning_rate": 2.632803435965823e-06, - "loss": 1.2802, - "step": 2858 - }, - { - "epoch": 2.31, - "learning_rate": 2.626898740874332e-06, - "loss": 1.1876, - "step": 2859 - }, - { - "epoch": 2.31, - "learning_rate": 2.620999673218656e-06, - "loss": 1.3147, - "step": 2860 - }, - { - "epoch": 2.31, - "learning_rate": 2.615106237501186e-06, - "loss": 1.3008, - "step": 2861 - }, - { - "epoch": 2.32, - "learning_rate": 2.6092184382200216e-06, - "loss": 1.1373, - "step": 2862 - }, - { - "epoch": 2.32, - "learning_rate": 2.603336279868959e-06, - "loss": 1.1831, - "step": 2863 - }, - { - "epoch": 2.32, - "learning_rate": 2.597459766937489e-06, - "loss": 1.0573, - "step": 2864 - }, - { - "epoch": 2.32, - "learning_rate": 2.5915889039107967e-06, - "loss": 1.1659, - "step": 2865 - }, - { - "epoch": 2.32, - "learning_rate": 2.585723695269745e-06, - "loss": 1.2382, - "step": 2866 - }, - { - "epoch": 2.32, - "learning_rate": 2.5798641454908945e-06, - "loss": 1.1173, - "step": 2867 - }, - { - "epoch": 2.32, - "learning_rate": 2.574010259046473e-06, - "loss": 1.1606, - "step": 2868 - }, - { - "epoch": 2.32, - "learning_rate": 2.568162040404397e-06, - "loss": 1.2141, - "step": 2869 - }, - { - "epoch": 2.32, - "learning_rate": 2.5623194940282526e-06, - "loss": 1.2727, - "step": 2870 - }, - { - "epoch": 2.32, - "learning_rate": 2.5564826243772965e-06, - "loss": 1.0827, - "step": 2871 - }, - { - "epoch": 2.32, - "learning_rate": 2.5506514359064562e-06, - "loss": 1.1992, - "step": 2872 - }, - { - "epoch": 2.32, - "learning_rate": 2.544825933066315e-06, - "loss": 1.2062, - "step": 2873 - }, - { - "epoch": 2.33, - "learning_rate": 2.5390061203031234e-06, - "loss": 1.25, - "step": 2874 - }, - { - "epoch": 2.33, - "learning_rate": 2.533192002058792e-06, - "loss": 1.3069, - "step": 2875 - }, - { - "epoch": 2.33, - "learning_rate": 2.5273835827708738e-06, - "loss": 1.2269, - "step": 2876 - }, - { - "epoch": 2.33, - "learning_rate": 2.5215808668725815e-06, - "loss": 1.2341, - "step": 2877 - }, - { - "epoch": 2.33, - "learning_rate": 2.515783858792772e-06, - "loss": 1.2191, - "step": 2878 - }, - { - "epoch": 2.33, - "learning_rate": 2.5099925629559495e-06, - "loss": 1.0972, - "step": 2879 - }, - { - "epoch": 2.33, - "learning_rate": 2.504206983782248e-06, - "loss": 1.1902, - "step": 2880 - }, - { - "epoch": 2.33, - "learning_rate": 2.4984271256874504e-06, - "loss": 1.0813, - "step": 2881 - }, - { - "epoch": 2.33, - "learning_rate": 2.492652993082967e-06, - "loss": 1.1447, - "step": 2882 - }, - { - "epoch": 2.33, - "learning_rate": 2.4868845903758366e-06, - "loss": 1.2228, - "step": 2883 - }, - { - "epoch": 2.33, - "learning_rate": 2.4811219219687275e-06, - "loss": 1.2066, - "step": 2884 - }, - { - "epoch": 2.33, - "learning_rate": 2.475364992259932e-06, - "loss": 1.2241, - "step": 2885 - }, - { - "epoch": 2.33, - "learning_rate": 2.469613805643362e-06, - "loss": 1.2069, - "step": 2886 - }, - { - "epoch": 2.34, - "learning_rate": 2.463868366508546e-06, - "loss": 1.1531, - "step": 2887 - }, - { - "epoch": 2.34, - "learning_rate": 2.4581286792406223e-06, - "loss": 1.1092, - "step": 2888 - }, - { - "epoch": 2.34, - "learning_rate": 2.452394748220346e-06, - "loss": 1.1622, - "step": 2889 - }, - { - "epoch": 2.34, - "learning_rate": 2.446666577824068e-06, - "loss": 1.2961, - "step": 2890 - }, - { - "epoch": 2.34, - "learning_rate": 2.440944172423756e-06, - "loss": 1.1437, - "step": 2891 - }, - { - "epoch": 2.34, - "learning_rate": 2.435227536386967e-06, - "loss": 1.0714, - "step": 2892 - }, - { - "epoch": 2.34, - "learning_rate": 2.4295166740768605e-06, - "loss": 1.3431, - "step": 2893 - }, - { - "epoch": 2.34, - "learning_rate": 2.4238115898521897e-06, - "loss": 1.1435, - "step": 2894 - }, - { - "epoch": 2.34, - "learning_rate": 2.4181122880672903e-06, - "loss": 1.252, - "step": 2895 - }, - { - "epoch": 2.34, - "learning_rate": 2.4124187730720916e-06, - "loss": 1.3622, - "step": 2896 - }, - { - "epoch": 2.34, - "learning_rate": 2.4067310492121064e-06, - "loss": 1.0496, - "step": 2897 - }, - { - "epoch": 2.34, - "learning_rate": 2.4010491208284226e-06, - "loss": 1.0793, - "step": 2898 - }, - { - "epoch": 2.35, - "learning_rate": 2.3953729922577075e-06, - "loss": 1.1588, - "step": 2899 - }, - { - "epoch": 2.35, - "learning_rate": 2.389702667832202e-06, - "loss": 1.1668, - "step": 2900 - }, - { - "epoch": 2.35, - "learning_rate": 2.3840381518797182e-06, - "loss": 1.1615, - "step": 2901 - }, - { - "epoch": 2.35, - "learning_rate": 2.3783794487236367e-06, - "loss": 1.1537, - "step": 2902 - }, - { - "epoch": 2.35, - "learning_rate": 2.372726562682891e-06, - "loss": 1.3515, - "step": 2903 - }, - { - "epoch": 2.35, - "learning_rate": 2.367079498071989e-06, - "loss": 1.0111, - "step": 2904 - }, - { - "epoch": 2.35, - "learning_rate": 2.3614382592009822e-06, - "loss": 1.2441, - "step": 2905 - }, - { - "epoch": 2.35, - "learning_rate": 2.355802850375486e-06, - "loss": 1.185, - "step": 2906 - }, - { - "epoch": 2.35, - "learning_rate": 2.350173275896661e-06, - "loss": 1.1507, - "step": 2907 - }, - { - "epoch": 2.35, - "learning_rate": 2.344549540061217e-06, - "loss": 1.0944, - "step": 2908 - }, - { - "epoch": 2.35, - "learning_rate": 2.33893164716141e-06, - "loss": 1.1102, - "step": 2909 - }, - { - "epoch": 2.35, - "learning_rate": 2.3333196014850246e-06, - "loss": 1.227, - "step": 2910 - }, - { - "epoch": 2.36, - "learning_rate": 2.3277134073153963e-06, - "loss": 1.183, - "step": 2911 - }, - { - "epoch": 2.36, - "learning_rate": 2.322113068931391e-06, - "loss": 1.1999, - "step": 2912 - }, - { - "epoch": 2.36, - "learning_rate": 2.316518590607395e-06, - "loss": 1.1329, - "step": 2913 - }, - { - "epoch": 2.36, - "learning_rate": 2.310929976613341e-06, - "loss": 1.3664, - "step": 2914 - }, - { - "epoch": 2.36, - "learning_rate": 2.3053472312146683e-06, - "loss": 1.1305, - "step": 2915 - }, - { - "epoch": 2.36, - "learning_rate": 2.2997703586723475e-06, - "loss": 1.1465, - "step": 2916 - }, - { - "epoch": 2.36, - "learning_rate": 2.2941993632428595e-06, - "loss": 1.2007, - "step": 2917 - }, - { - "epoch": 2.36, - "learning_rate": 2.2886342491782056e-06, - "loss": 1.2281, - "step": 2918 - }, - { - "epoch": 2.36, - "learning_rate": 2.2830750207258977e-06, - "loss": 1.2058, - "step": 2919 - }, - { - "epoch": 2.36, - "learning_rate": 2.277521682128947e-06, - "loss": 1.2269, - "step": 2920 - }, - { - "epoch": 2.36, - "learning_rate": 2.271974237625886e-06, - "loss": 1.278, - "step": 2921 - }, - { - "epoch": 2.36, - "learning_rate": 2.2664326914507305e-06, - "loss": 1.0236, - "step": 2922 - }, - { - "epoch": 2.36, - "learning_rate": 2.2608970478330063e-06, - "loss": 1.1397, - "step": 2923 - }, - { - "epoch": 2.37, - "learning_rate": 2.2553673109977326e-06, - "loss": 1.2241, - "step": 2924 - }, - { - "epoch": 2.37, - "learning_rate": 2.2498434851654125e-06, - "loss": 1.0855, - "step": 2925 - }, - { - "epoch": 2.37, - "learning_rate": 2.24432557455205e-06, - "loss": 1.1, - "step": 2926 - }, - { - "epoch": 2.37, - "learning_rate": 2.2388135833691182e-06, - "loss": 1.1115, - "step": 2927 - }, - { - "epoch": 2.37, - "learning_rate": 2.233307515823594e-06, - "loss": 1.0859, - "step": 2928 - }, - { - "epoch": 2.37, - "learning_rate": 2.227807376117912e-06, - "loss": 1.2954, - "step": 2929 - }, - { - "epoch": 2.37, - "learning_rate": 2.2223131684499932e-06, - "loss": 1.2655, - "step": 2930 - }, - { - "epoch": 2.37, - "learning_rate": 2.2168248970132333e-06, - "loss": 1.1096, - "step": 2931 - }, - { - "epoch": 2.37, - "learning_rate": 2.2113425659964873e-06, - "loss": 1.1912, - "step": 2932 - }, - { - "epoch": 2.37, - "learning_rate": 2.205866179584084e-06, - "loss": 1.1789, - "step": 2933 - }, - { - "epoch": 2.37, - "learning_rate": 2.200395741955815e-06, - "loss": 1.125, - "step": 2934 - }, - { - "epoch": 2.37, - "learning_rate": 2.1949312572869287e-06, - "loss": 1.0783, - "step": 2935 - }, - { - "epoch": 2.38, - "learning_rate": 2.189472729748132e-06, - "loss": 1.1818, - "step": 2936 - }, - { - "epoch": 2.38, - "learning_rate": 2.1840201635055823e-06, - "loss": 1.2237, - "step": 2937 - }, - { - "epoch": 2.38, - "learning_rate": 2.1785735627208913e-06, - "loss": 1.2091, - "step": 2938 - }, - { - "epoch": 2.38, - "learning_rate": 2.1731329315511107e-06, - "loss": 1.2077, - "step": 2939 - }, - { - "epoch": 2.38, - "learning_rate": 2.1676982741487427e-06, - "loss": 1.222, - "step": 2940 - }, - { - "epoch": 2.38, - "learning_rate": 2.162269594661729e-06, - "loss": 1.1159, - "step": 2941 - }, - { - "epoch": 2.38, - "learning_rate": 2.1568468972334465e-06, - "loss": 1.3479, - "step": 2942 - }, - { - "epoch": 2.38, - "learning_rate": 2.151430186002711e-06, - "loss": 1.1948, - "step": 2943 - }, - { - "epoch": 2.38, - "learning_rate": 2.146019465103759e-06, - "loss": 1.1809, - "step": 2944 - }, - { - "epoch": 2.38, - "learning_rate": 2.140614738666267e-06, - "loss": 1.2298, - "step": 2945 - }, - { - "epoch": 2.38, - "learning_rate": 2.135216010815331e-06, - "loss": 0.9459, - "step": 2946 - }, - { - "epoch": 2.38, - "learning_rate": 2.129823285671465e-06, - "loss": 1.1176, - "step": 2947 - }, - { - "epoch": 2.39, - "learning_rate": 2.1244365673506074e-06, - "loss": 1.1678, - "step": 2948 - }, - { - "epoch": 2.39, - "learning_rate": 2.1190558599641097e-06, - "loss": 1.3229, - "step": 2949 - }, - { - "epoch": 2.39, - "learning_rate": 2.113681167618736e-06, - "loss": 1.3054, - "step": 2950 - }, - { - "epoch": 2.39, - "learning_rate": 2.108312494416663e-06, - "loss": 1.2042, - "step": 2951 - }, - { - "epoch": 2.39, - "learning_rate": 2.102949844455462e-06, - "loss": 1.14, - "step": 2952 - }, - { - "epoch": 2.39, - "learning_rate": 2.0975932218281203e-06, - "loss": 1.2829, - "step": 2953 - }, - { - "epoch": 2.39, - "learning_rate": 2.092242630623016e-06, - "loss": 1.1766, - "step": 2954 - }, - { - "epoch": 2.39, - "learning_rate": 2.0868980749239277e-06, - "loss": 1.2294, - "step": 2955 - }, - { - "epoch": 2.39, - "learning_rate": 2.081559558810028e-06, - "loss": 1.2329, - "step": 2956 - }, - { - "epoch": 2.39, - "learning_rate": 2.0762270863558774e-06, - "loss": 1.2206, - "step": 2957 - }, - { - "epoch": 2.39, - "learning_rate": 2.070900661631429e-06, - "loss": 1.1934, - "step": 2958 - }, - { - "epoch": 2.39, - "learning_rate": 2.0655802887020106e-06, - "loss": 1.1996, - "step": 2959 - }, - { - "epoch": 2.39, - "learning_rate": 2.060265971628338e-06, - "loss": 1.2041, - "step": 2960 - }, - { - "epoch": 2.4, - "learning_rate": 2.0549577144665067e-06, - "loss": 1.192, - "step": 2961 - }, - { - "epoch": 2.4, - "learning_rate": 2.049655521267978e-06, - "loss": 1.191, - "step": 2962 - }, - { - "epoch": 2.4, - "learning_rate": 2.044359396079595e-06, - "loss": 1.1315, - "step": 2963 - }, - { - "epoch": 2.4, - "learning_rate": 2.0390693429435626e-06, - "loss": 1.1492, - "step": 2964 - }, - { - "epoch": 2.4, - "learning_rate": 2.0337853658974573e-06, - "loss": 1.2017, - "step": 2965 - }, - { - "epoch": 2.4, - "learning_rate": 2.02850746897421e-06, - "loss": 1.2996, - "step": 2966 - }, - { - "epoch": 2.4, - "learning_rate": 2.0232356562021183e-06, - "loss": 1.426, - "step": 2967 - }, - { - "epoch": 2.4, - "learning_rate": 2.017969931604835e-06, - "loss": 1.2426, - "step": 2968 - }, - { - "epoch": 2.4, - "learning_rate": 2.0127102992013605e-06, - "loss": 1.229, - "step": 2969 - }, - { - "epoch": 2.4, - "learning_rate": 2.0074567630060514e-06, - "loss": 1.2308, - "step": 2970 - }, - { - "epoch": 2.4, - "learning_rate": 2.002209327028609e-06, - "loss": 1.1706, - "step": 2971 - }, - { - "epoch": 2.4, - "learning_rate": 1.9969679952740805e-06, - "loss": 1.2618, - "step": 2972 - }, - { - "epoch": 2.41, - "learning_rate": 1.9917327717428546e-06, - "loss": 1.2788, - "step": 2973 - }, - { - "epoch": 2.41, - "learning_rate": 1.986503660430652e-06, - "loss": 1.1624, - "step": 2974 - }, - { - "epoch": 2.41, - "learning_rate": 1.9812806653285366e-06, - "loss": 1.1493, - "step": 2975 - }, - { - "epoch": 2.41, - "learning_rate": 1.9760637904228953e-06, - "loss": 1.2322, - "step": 2976 - }, - { - "epoch": 2.41, - "learning_rate": 1.9708530396954484e-06, - "loss": 1.1345, - "step": 2977 - }, - { - "epoch": 2.41, - "learning_rate": 1.9656484171232506e-06, - "loss": 1.1859, - "step": 2978 - }, - { - "epoch": 2.41, - "learning_rate": 1.9604499266786627e-06, - "loss": 1.2352, - "step": 2979 - }, - { - "epoch": 2.41, - "learning_rate": 1.955257572329379e-06, - "loss": 1.164, - "step": 2980 - }, - { - "epoch": 2.41, - "learning_rate": 1.9500713580384e-06, - "loss": 1.1184, - "step": 2981 - }, - { - "epoch": 2.41, - "learning_rate": 1.944891287764047e-06, - "loss": 1.0865, - "step": 2982 - }, - { - "epoch": 2.41, - "learning_rate": 1.939717365459952e-06, - "loss": 1.2667, - "step": 2983 - }, - { - "epoch": 2.41, - "learning_rate": 1.934549595075045e-06, - "loss": 1.2165, - "step": 2984 - }, - { - "epoch": 2.42, - "learning_rate": 1.9293879805535785e-06, - "loss": 1.1375, - "step": 2985 - }, - { - "epoch": 2.42, - "learning_rate": 1.9242325258350892e-06, - "loss": 1.1235, - "step": 2986 - }, - { - "epoch": 2.42, - "learning_rate": 1.919083234854422e-06, - "loss": 1.2749, - "step": 2987 - }, - { - "epoch": 2.42, - "learning_rate": 1.9139401115417156e-06, - "loss": 1.2056, - "step": 2988 - }, - { - "epoch": 2.42, - "learning_rate": 1.9088031598223965e-06, - "loss": 1.2692, - "step": 2989 - }, - { - "epoch": 2.42, - "learning_rate": 1.9036723836171899e-06, - "loss": 1.2319, - "step": 2990 - }, - { - "epoch": 2.42, - "learning_rate": 1.8985477868420944e-06, - "loss": 1.2516, - "step": 2991 - }, - { - "epoch": 2.42, - "learning_rate": 1.8934293734084108e-06, - "loss": 1.0878, - "step": 2992 - }, - { - "epoch": 2.42, - "learning_rate": 1.8883171472227024e-06, - "loss": 1.2057, - "step": 2993 - }, - { - "epoch": 2.42, - "learning_rate": 1.8832111121868202e-06, - "loss": 1.164, - "step": 2994 - }, - { - "epoch": 2.42, - "learning_rate": 1.87811127219789e-06, - "loss": 1.0183, - "step": 2995 - }, - { - "epoch": 2.42, - "learning_rate": 1.8730176311483017e-06, - "loss": 1.3838, - "step": 2996 - }, - { - "epoch": 2.42, - "learning_rate": 1.867930192925722e-06, - "loss": 1.2683, - "step": 2997 - }, - { - "epoch": 2.43, - "learning_rate": 1.862848961413083e-06, - "loss": 1.1899, - "step": 2998 - }, - { - "epoch": 2.43, - "learning_rate": 1.8577739404885675e-06, - "loss": 1.3525, - "step": 2999 - }, - { - "epoch": 2.43, - "learning_rate": 1.8527051340256397e-06, - "loss": 1.1829, - "step": 3000 - }, - { - "epoch": 2.43, - "learning_rate": 1.8476425458930003e-06, - "loss": 1.1494, - "step": 3001 - }, - { - "epoch": 2.43, - "learning_rate": 1.8425861799546175e-06, - "loss": 1.287, - "step": 3002 - }, - { - "epoch": 2.43, - "learning_rate": 1.8375360400697007e-06, - "loss": 1.1104, - "step": 3003 - }, - { - "epoch": 2.43, - "learning_rate": 1.8324921300927123e-06, - "loss": 1.1262, - "step": 3004 - }, - { - "epoch": 2.43, - "learning_rate": 1.8274544538733629e-06, - "loss": 1.2537, - "step": 3005 - }, - { - "epoch": 2.43, - "learning_rate": 1.8224230152565958e-06, - "loss": 1.0156, - "step": 3006 - }, - { - "epoch": 2.43, - "learning_rate": 1.8173978180826068e-06, - "loss": 1.2801, - "step": 3007 - }, - { - "epoch": 2.43, - "learning_rate": 1.8123788661868146e-06, - "loss": 1.1252, - "step": 3008 - }, - { - "epoch": 2.43, - "learning_rate": 1.8073661633998796e-06, - "loss": 1.1458, - "step": 3009 - }, - { - "epoch": 2.44, - "learning_rate": 1.8023597135476923e-06, - "loss": 1.2469, - "step": 3010 - }, - { - "epoch": 2.44, - "learning_rate": 1.7973595204513638e-06, - "loss": 1.1726, - "step": 3011 - }, - { - "epoch": 2.44, - "learning_rate": 1.7923655879272395e-06, - "loss": 1.1686, - "step": 3012 - }, - { - "epoch": 2.44, - "learning_rate": 1.7873779197868746e-06, - "loss": 1.2089, - "step": 3013 - }, - { - "epoch": 2.44, - "learning_rate": 1.7823965198370607e-06, - "loss": 1.1477, - "step": 3014 - }, - { - "epoch": 2.44, - "learning_rate": 1.7774213918797868e-06, - "loss": 1.2079, - "step": 3015 - }, - { - "epoch": 2.44, - "learning_rate": 1.772452539712265e-06, - "loss": 1.1489, - "step": 3016 - }, - { - "epoch": 2.44, - "learning_rate": 1.7674899671269175e-06, - "loss": 1.129, - "step": 3017 - }, - { - "epoch": 2.44, - "learning_rate": 1.7625336779113678e-06, - "loss": 1.366, - "step": 3018 - }, - { - "epoch": 2.44, - "learning_rate": 1.7575836758484478e-06, - "loss": 1.0975, - "step": 3019 - }, - { - "epoch": 2.44, - "learning_rate": 1.752639964716193e-06, - "loss": 1.122, - "step": 3020 - }, - { - "epoch": 2.44, - "learning_rate": 1.747702548287832e-06, - "loss": 1.3125, - "step": 3021 - }, - { - "epoch": 2.44, - "learning_rate": 1.7427714303317966e-06, - "loss": 1.2897, - "step": 3022 - }, - { - "epoch": 2.45, - "learning_rate": 1.7378466146117e-06, - "loss": 1.1252, - "step": 3023 - }, - { - "epoch": 2.45, - "learning_rate": 1.7329281048863588e-06, - "loss": 1.0944, - "step": 3024 - }, - { - "epoch": 2.45, - "learning_rate": 1.7280159049097622e-06, - "loss": 1.2356, - "step": 3025 - }, - { - "epoch": 2.45, - "learning_rate": 1.7231100184310955e-06, - "loss": 1.2456, - "step": 3026 - }, - { - "epoch": 2.45, - "learning_rate": 1.7182104491947193e-06, - "loss": 1.1785, - "step": 3027 - }, - { - "epoch": 2.45, - "learning_rate": 1.7133172009401755e-06, - "loss": 1.2575, - "step": 3028 - }, - { - "epoch": 2.45, - "learning_rate": 1.7084302774021832e-06, - "loss": 1.1174, - "step": 3029 - }, - { - "epoch": 2.45, - "learning_rate": 1.7035496823106247e-06, - "loss": 1.0943, - "step": 3030 - }, - { - "epoch": 2.45, - "learning_rate": 1.6986754193905618e-06, - "loss": 1.0756, - "step": 3031 - }, - { - "epoch": 2.45, - "learning_rate": 1.6938074923622227e-06, - "loss": 1.0809, - "step": 3032 - }, - { - "epoch": 2.45, - "learning_rate": 1.688945904940993e-06, - "loss": 1.2481, - "step": 3033 - }, - { - "epoch": 2.45, - "learning_rate": 1.6840906608374264e-06, - "loss": 1.1378, - "step": 3034 - }, - { - "epoch": 2.46, - "learning_rate": 1.6792417637572324e-06, - "loss": 1.2115, - "step": 3035 - }, - { - "epoch": 2.46, - "learning_rate": 1.674399217401278e-06, - "loss": 1.3145, - "step": 3036 - }, - { - "epoch": 2.46, - "learning_rate": 1.6695630254655826e-06, - "loss": 1.2484, - "step": 3037 - }, - { - "epoch": 2.46, - "learning_rate": 1.6647331916413102e-06, - "loss": 1.3023, - "step": 3038 - }, - { - "epoch": 2.46, - "learning_rate": 1.6599097196147818e-06, - "loss": 1.212, - "step": 3039 - }, - { - "epoch": 2.46, - "learning_rate": 1.6550926130674527e-06, - "loss": 1.1351, - "step": 3040 - }, - { - "epoch": 2.46, - "learning_rate": 1.6502818756759275e-06, - "loss": 0.9676, - "step": 3041 - }, - { - "epoch": 2.46, - "learning_rate": 1.6454775111119458e-06, - "loss": 1.2051, - "step": 3042 - }, - { - "epoch": 2.46, - "learning_rate": 1.6406795230423855e-06, - "loss": 1.1125, - "step": 3043 - }, - { - "epoch": 2.46, - "learning_rate": 1.6358879151292595e-06, - "loss": 1.2991, - "step": 3044 - }, - { - "epoch": 2.46, - "learning_rate": 1.6311026910297013e-06, - "loss": 1.2118, - "step": 3045 - }, - { - "epoch": 2.46, - "learning_rate": 1.6263238543959825e-06, - "loss": 1.1312, - "step": 3046 - }, - { - "epoch": 2.47, - "learning_rate": 1.621551408875498e-06, - "loss": 1.2365, - "step": 3047 - }, - { - "epoch": 2.47, - "learning_rate": 1.6167853581107584e-06, - "loss": 1.3287, - "step": 3048 - }, - { - "epoch": 2.47, - "learning_rate": 1.6120257057393996e-06, - "loss": 1.3287, - "step": 3049 - }, - { - "epoch": 2.47, - "learning_rate": 1.607272455394172e-06, - "loss": 1.1086, - "step": 3050 - }, - { - "epoch": 2.47, - "learning_rate": 1.602525610702943e-06, - "loss": 1.1485, - "step": 3051 - }, - { - "epoch": 2.47, - "learning_rate": 1.597785175288683e-06, - "loss": 1.178, - "step": 3052 - }, - { - "epoch": 2.47, - "learning_rate": 1.593051152769478e-06, - "loss": 1.2129, - "step": 3053 - }, - { - "epoch": 2.47, - "learning_rate": 1.588323546758519e-06, - "loss": 1.0935, - "step": 3054 - }, - { - "epoch": 2.47, - "learning_rate": 1.5836023608640927e-06, - "loss": 1.2113, - "step": 3055 - }, - { - "epoch": 2.47, - "learning_rate": 1.5788875986895936e-06, - "loss": 1.3559, - "step": 3056 - }, - { - "epoch": 2.47, - "learning_rate": 1.5741792638335096e-06, - "loss": 1.1986, - "step": 3057 - }, - { - "epoch": 2.47, - "learning_rate": 1.569477359889424e-06, - "loss": 1.3229, - "step": 3058 - }, - { - "epoch": 2.47, - "learning_rate": 1.5647818904460143e-06, - "loss": 1.2306, - "step": 3059 - }, - { - "epoch": 2.48, - "learning_rate": 1.5600928590870402e-06, - "loss": 1.227, - "step": 3060 - }, - { - "epoch": 2.48, - "learning_rate": 1.5554102693913543e-06, - "loss": 1.1405, - "step": 3061 - }, - { - "epoch": 2.48, - "learning_rate": 1.5507341249328867e-06, - "loss": 1.1667, - "step": 3062 - }, - { - "epoch": 2.48, - "learning_rate": 1.5460644292806527e-06, - "loss": 1.308, - "step": 3063 - }, - { - "epoch": 2.48, - "learning_rate": 1.5414011859987488e-06, - "loss": 1.1761, - "step": 3064 - }, - { - "epoch": 2.48, - "learning_rate": 1.5367443986463382e-06, - "loss": 1.2762, - "step": 3065 - }, - { - "epoch": 2.48, - "learning_rate": 1.5320940707776633e-06, - "loss": 1.1403, - "step": 3066 - }, - { - "epoch": 2.48, - "learning_rate": 1.5274502059420303e-06, - "loss": 1.0538, - "step": 3067 - }, - { - "epoch": 2.48, - "learning_rate": 1.5228128076838201e-06, - "loss": 1.3167, - "step": 3068 - }, - { - "epoch": 2.48, - "learning_rate": 1.5181818795424741e-06, - "loss": 1.3031, - "step": 3069 - }, - { - "epoch": 2.48, - "learning_rate": 1.5135574250524898e-06, - "loss": 1.1231, - "step": 3070 - }, - { - "epoch": 2.48, - "learning_rate": 1.5089394477434382e-06, - "loss": 1.1709, - "step": 3071 - }, - { - "epoch": 2.49, - "learning_rate": 1.5043279511399334e-06, - "loss": 1.3344, - "step": 3072 - }, - { - "epoch": 2.49, - "learning_rate": 1.4997229387616474e-06, - "loss": 1.3782, - "step": 3073 - }, - { - "epoch": 2.49, - "learning_rate": 1.4951244141233067e-06, - "loss": 1.2354, - "step": 3074 - }, - { - "epoch": 2.49, - "learning_rate": 1.4905323807346794e-06, - "loss": 1.1779, - "step": 3075 - }, - { - "epoch": 2.49, - "learning_rate": 1.4859468421005863e-06, - "loss": 1.1792, - "step": 3076 - }, - { - "epoch": 2.49, - "learning_rate": 1.481367801720881e-06, - "loss": 1.1694, - "step": 3077 - }, - { - "epoch": 2.49, - "learning_rate": 1.4767952630904746e-06, - "loss": 1.2722, - "step": 3078 - }, - { - "epoch": 2.49, - "learning_rate": 1.4722292296992957e-06, - "loss": 1.0935, - "step": 3079 - }, - { - "epoch": 2.49, - "learning_rate": 1.467669705032323e-06, - "loss": 1.0384, - "step": 3080 - }, - { - "epoch": 2.49, - "learning_rate": 1.4631166925695617e-06, - "loss": 1.1863, - "step": 3081 - }, - { - "epoch": 2.49, - "learning_rate": 1.458570195786042e-06, - "loss": 1.1043, - "step": 3082 - }, - { - "epoch": 2.49, - "learning_rate": 1.4540302181518306e-06, - "loss": 1.0384, - "step": 3083 - }, - { - "epoch": 2.5, - "learning_rate": 1.4494967631320123e-06, - "loss": 1.1939, - "step": 3084 - }, - { - "epoch": 2.5, - "learning_rate": 1.4449698341866958e-06, - "loss": 1.157, - "step": 3085 - }, - { - "epoch": 2.5, - "learning_rate": 1.4404494347710096e-06, - "loss": 1.1961, - "step": 3086 - }, - { - "epoch": 2.5, - "learning_rate": 1.435935568335094e-06, - "loss": 1.1875, - "step": 3087 - }, - { - "epoch": 2.5, - "learning_rate": 1.4314282383241097e-06, - "loss": 1.205, - "step": 3088 - }, - { - "epoch": 2.5, - "learning_rate": 1.4269274481782203e-06, - "loss": 1.1717, - "step": 3089 - }, - { - "epoch": 2.5, - "learning_rate": 1.422433201332607e-06, - "loss": 1.1658, - "step": 3090 - }, - { - "epoch": 2.5, - "learning_rate": 1.4179455012174503e-06, - "loss": 1.1449, - "step": 3091 - }, - { - "epoch": 2.5, - "learning_rate": 1.4134643512579383e-06, - "loss": 1.1364, - "step": 3092 - }, - { - "epoch": 2.5, - "learning_rate": 1.4089897548742581e-06, - "loss": 1.1702, - "step": 3093 - }, - { - "epoch": 2.5, - "learning_rate": 1.4045217154815926e-06, - "loss": 1.3786, - "step": 3094 - }, - { - "epoch": 2.5, - "learning_rate": 1.4000602364901218e-06, - "loss": 1.143, - "step": 3095 - }, - { - "epoch": 2.5, - "learning_rate": 1.3956053213050224e-06, - "loss": 1.2091, - "step": 3096 - }, - { - "epoch": 2.51, - "learning_rate": 1.3911569733264541e-06, - "loss": 1.1168, - "step": 3097 - }, - { - "epoch": 2.51, - "learning_rate": 1.3867151959495684e-06, - "loss": 0.9922, - "step": 3098 - }, - { - "epoch": 2.51, - "learning_rate": 1.3822799925645036e-06, - "loss": 1.2234, - "step": 3099 - }, - { - "epoch": 2.51, - "learning_rate": 1.3778513665563786e-06, - "loss": 1.1293, - "step": 3100 - }, - { - "epoch": 2.51, - "learning_rate": 1.3734293213052885e-06, - "loss": 1.108, - "step": 3101 - }, - { - "epoch": 2.51, - "learning_rate": 1.36901386018631e-06, - "loss": 1.1236, - "step": 3102 - }, - { - "epoch": 2.51, - "learning_rate": 1.364604986569499e-06, - "loss": 1.2793, - "step": 3103 - }, - { - "epoch": 2.51, - "learning_rate": 1.3602027038198695e-06, - "loss": 1.1267, - "step": 3104 - }, - { - "epoch": 2.51, - "learning_rate": 1.35580701529742e-06, - "loss": 1.2945, - "step": 3105 - }, - { - "epoch": 2.51, - "learning_rate": 1.3514179243571069e-06, - "loss": 1.218, - "step": 3106 - }, - { - "epoch": 2.51, - "learning_rate": 1.3470354343488568e-06, - "loss": 1.2613, - "step": 3107 - }, - { - "epoch": 2.51, - "learning_rate": 1.3426595486175542e-06, - "loss": 1.3523, - "step": 3108 - }, - { - "epoch": 2.52, - "learning_rate": 1.3382902705030432e-06, - "loss": 1.2384, - "step": 3109 - }, - { - "epoch": 2.52, - "learning_rate": 1.3339276033401283e-06, - "loss": 1.0767, - "step": 3110 - }, - { - "epoch": 2.52, - "learning_rate": 1.3295715504585604e-06, - "loss": 1.2409, - "step": 3111 - }, - { - "epoch": 2.52, - "learning_rate": 1.3252221151830513e-06, - "loss": 1.1923, - "step": 3112 - }, - { - "epoch": 2.52, - "learning_rate": 1.3208793008332587e-06, - "loss": 1.0467, - "step": 3113 - }, - { - "epoch": 2.52, - "learning_rate": 1.3165431107237836e-06, - "loss": 1.3638, - "step": 3114 - }, - { - "epoch": 2.52, - "learning_rate": 1.3122135481641784e-06, - "loss": 1.1619, - "step": 3115 - }, - { - "epoch": 2.52, - "learning_rate": 1.3078906164589278e-06, - "loss": 1.1821, - "step": 3116 - }, - { - "epoch": 2.52, - "learning_rate": 1.3035743189074612e-06, - "loss": 1.1615, - "step": 3117 - }, - { - "epoch": 2.52, - "learning_rate": 1.299264658804148e-06, - "loss": 1.1649, - "step": 3118 - }, - { - "epoch": 2.52, - "learning_rate": 1.2949616394382802e-06, - "loss": 1.216, - "step": 3119 - }, - { - "epoch": 2.52, - "learning_rate": 1.290665264094093e-06, - "loss": 1.1782, - "step": 3120 - }, - { - "epoch": 2.53, - "learning_rate": 1.2863755360507458e-06, - "loss": 1.1259, - "step": 3121 - }, - { - "epoch": 2.53, - "learning_rate": 1.2820924585823236e-06, - "loss": 1.2723, - "step": 3122 - }, - { - "epoch": 2.53, - "learning_rate": 1.2778160349578405e-06, - "loss": 1.2029, - "step": 3123 - }, - { - "epoch": 2.53, - "learning_rate": 1.2735462684412225e-06, - "loss": 1.1055, - "step": 3124 - }, - { - "epoch": 2.53, - "learning_rate": 1.2692831622913249e-06, - "loss": 1.1469, - "step": 3125 - }, - { - "epoch": 2.53, - "learning_rate": 1.2650267197619114e-06, - "loss": 1.181, - "step": 3126 - }, - { - "epoch": 2.53, - "learning_rate": 1.2607769441016648e-06, - "loss": 1.218, - "step": 3127 - }, - { - "epoch": 2.53, - "learning_rate": 1.2565338385541792e-06, - "loss": 1.2239, - "step": 3128 - }, - { - "epoch": 2.53, - "learning_rate": 1.2522974063579552e-06, - "loss": 1.0997, - "step": 3129 - }, - { - "epoch": 2.53, - "learning_rate": 1.2480676507464051e-06, - "loss": 1.1899, - "step": 3130 - }, - { - "epoch": 2.53, - "learning_rate": 1.2438445749478379e-06, - "loss": 1.2462, - "step": 3131 - }, - { - "epoch": 2.53, - "learning_rate": 1.2396281821854684e-06, - "loss": 1.0247, - "step": 3132 - }, - { - "epoch": 2.53, - "learning_rate": 1.2354184756774147e-06, - "loss": 1.2345, - "step": 3133 - }, - { - "epoch": 2.54, - "learning_rate": 1.2312154586366799e-06, - "loss": 1.2361, - "step": 3134 - }, - { - "epoch": 2.54, - "learning_rate": 1.2270191342711768e-06, - "loss": 1.3638, - "step": 3135 - }, - { - "epoch": 2.54, - "learning_rate": 1.2228295057836981e-06, - "loss": 0.9582, - "step": 3136 - }, - { - "epoch": 2.54, - "learning_rate": 1.2186465763719324e-06, - "loss": 1.4414, - "step": 3137 - }, - { - "epoch": 2.54, - "learning_rate": 1.2144703492284504e-06, - "loss": 1.0821, - "step": 3138 - }, - { - "epoch": 2.54, - "learning_rate": 1.2103008275407113e-06, - "loss": 1.2393, - "step": 3139 - }, - { - "epoch": 2.54, - "learning_rate": 1.2061380144910572e-06, - "loss": 1.0564, - "step": 3140 - }, - { - "epoch": 2.54, - "learning_rate": 1.2019819132567023e-06, - "loss": 1.237, - "step": 3141 - }, - { - "epoch": 2.54, - "learning_rate": 1.1978325270097512e-06, - "loss": 1.0626, - "step": 3142 - }, - { - "epoch": 2.54, - "learning_rate": 1.1936898589171708e-06, - "loss": 1.2909, - "step": 3143 - }, - { - "epoch": 2.54, - "learning_rate": 1.1895539121408062e-06, - "loss": 1.1765, - "step": 3144 - }, - { - "epoch": 2.54, - "learning_rate": 1.1854246898373744e-06, - "loss": 1.139, - "step": 3145 - }, - { - "epoch": 2.55, - "learning_rate": 1.1813021951584537e-06, - "loss": 1.1875, - "step": 3146 - }, - { - "epoch": 2.55, - "learning_rate": 1.1771864312504943e-06, - "loss": 1.1659, - "step": 3147 - }, - { - "epoch": 2.55, - "learning_rate": 1.1730774012548008e-06, - "loss": 1.3073, - "step": 3148 - }, - { - "epoch": 2.55, - "learning_rate": 1.1689751083075507e-06, - "loss": 1.2235, - "step": 3149 - }, - { - "epoch": 2.55, - "learning_rate": 1.1648795555397719e-06, - "loss": 1.3586, - "step": 3150 - }, - { - "epoch": 2.55, - "learning_rate": 1.1607907460773426e-06, - "loss": 1.1434, - "step": 3151 - }, - { - "epoch": 2.55, - "learning_rate": 1.156708683041008e-06, - "loss": 1.1946, - "step": 3152 - }, - { - "epoch": 2.55, - "learning_rate": 1.1526333695463488e-06, - "loss": 1.2285, - "step": 3153 - }, - { - "epoch": 2.55, - "learning_rate": 1.148564808703806e-06, - "loss": 1.1049, - "step": 3154 - }, - { - "epoch": 2.55, - "learning_rate": 1.1445030036186644e-06, - "loss": 1.1787, - "step": 3155 - }, - { - "epoch": 2.55, - "learning_rate": 1.140447957391042e-06, - "loss": 1.2781, - "step": 3156 - }, - { - "epoch": 2.55, - "learning_rate": 1.1363996731159188e-06, - "loss": 1.2288, - "step": 3157 - }, - { - "epoch": 2.56, - "learning_rate": 1.1323581538830919e-06, - "loss": 1.2373, - "step": 3158 - }, - { - "epoch": 2.56, - "learning_rate": 1.1283234027772095e-06, - "loss": 1.259, - "step": 3159 - }, - { - "epoch": 2.56, - "learning_rate": 1.1242954228777513e-06, - "loss": 1.2147, - "step": 3160 - }, - { - "epoch": 2.56, - "learning_rate": 1.1202742172590219e-06, - "loss": 1.4239, - "step": 3161 - }, - { - "epoch": 2.56, - "learning_rate": 1.116259788990166e-06, - "loss": 1.3156, - "step": 3162 - }, - { - "epoch": 2.56, - "learning_rate": 1.112252141135144e-06, - "loss": 1.1757, - "step": 3163 - }, - { - "epoch": 2.56, - "learning_rate": 1.1082512767527564e-06, - "loss": 1.2801, - "step": 3164 - }, - { - "epoch": 2.56, - "learning_rate": 1.10425719889661e-06, - "loss": 1.1406, - "step": 3165 - }, - { - "epoch": 2.56, - "learning_rate": 1.1002699106151426e-06, - "loss": 1.1087, - "step": 3166 - }, - { - "epoch": 2.56, - "learning_rate": 1.096289414951608e-06, - "loss": 1.2848, - "step": 3167 - }, - { - "epoch": 2.56, - "learning_rate": 1.09231571494407e-06, - "loss": 1.3059, - "step": 3168 - }, - { - "epoch": 2.56, - "learning_rate": 1.088348813625414e-06, - "loss": 1.2787, - "step": 3169 - }, - { - "epoch": 2.56, - "learning_rate": 1.08438871402333e-06, - "loss": 1.1109, - "step": 3170 - }, - { - "epoch": 2.57, - "learning_rate": 1.0804354191603195e-06, - "loss": 1.2464, - "step": 3171 - }, - { - "epoch": 2.57, - "learning_rate": 1.076488932053693e-06, - "loss": 1.13, - "step": 3172 - }, - { - "epoch": 2.57, - "learning_rate": 1.0725492557155581e-06, - "loss": 1.2377, - "step": 3173 - }, - { - "epoch": 2.57, - "learning_rate": 1.0686163931528305e-06, - "loss": 1.1776, - "step": 3174 - }, - { - "epoch": 2.57, - "learning_rate": 1.0646903473672199e-06, - "loss": 1.0109, - "step": 3175 - }, - { - "epoch": 2.57, - "learning_rate": 1.0607711213552374e-06, - "loss": 1.0184, - "step": 3176 - }, - { - "epoch": 2.57, - "learning_rate": 1.0568587181081891e-06, - "loss": 1.1687, - "step": 3177 - }, - { - "epoch": 2.57, - "learning_rate": 1.0529531406121719e-06, - "loss": 1.1104, - "step": 3178 - }, - { - "epoch": 2.57, - "learning_rate": 1.0490543918480745e-06, - "loss": 1.1719, - "step": 3179 - }, - { - "epoch": 2.57, - "learning_rate": 1.04516247479157e-06, - "loss": 1.1753, - "step": 3180 - }, - { - "epoch": 2.57, - "learning_rate": 1.0412773924131202e-06, - "loss": 1.0799, - "step": 3181 - }, - { - "epoch": 2.57, - "learning_rate": 1.0373991476779732e-06, - "loss": 1.2467, - "step": 3182 - }, - { - "epoch": 2.58, - "learning_rate": 1.0335277435461521e-06, - "loss": 1.0739, - "step": 3183 - }, - { - "epoch": 2.58, - "learning_rate": 1.0296631829724624e-06, - "loss": 1.1995, - "step": 3184 - }, - { - "epoch": 2.58, - "learning_rate": 1.0258054689064878e-06, - "loss": 1.1545, - "step": 3185 - }, - { - "epoch": 2.58, - "learning_rate": 1.0219546042925842e-06, - "loss": 1.2378, - "step": 3186 - }, - { - "epoch": 2.58, - "learning_rate": 1.0181105920698797e-06, - "loss": 0.9623, - "step": 3187 - }, - { - "epoch": 2.58, - "learning_rate": 1.0142734351722715e-06, - "loss": 1.2381, - "step": 3188 - }, - { - "epoch": 2.58, - "learning_rate": 1.0104431365284307e-06, - "loss": 1.3917, - "step": 3189 - }, - { - "epoch": 2.58, - "learning_rate": 1.006619699061785e-06, - "loss": 1.2756, - "step": 3190 - }, - { - "epoch": 2.58, - "learning_rate": 1.0028031256905292e-06, - "loss": 1.2105, - "step": 3191 - }, - { - "epoch": 2.58, - "learning_rate": 9.989934193276218e-07, - "loss": 1.2063, - "step": 3192 - }, - { - "epoch": 2.58, - "learning_rate": 9.951905828807785e-07, - "loss": 1.2113, - "step": 3193 - }, - { - "epoch": 2.58, - "learning_rate": 9.913946192524705e-07, - "loss": 1.0809, - "step": 3194 - }, - { - "epoch": 2.58, - "learning_rate": 9.876055313399225e-07, - "loss": 1.2564, - "step": 3195 - }, - { - "epoch": 2.59, - "learning_rate": 9.838233220351145e-07, - "loss": 1.3729, - "step": 3196 - }, - { - "epoch": 2.59, - "learning_rate": 9.80047994224773e-07, - "loss": 1.3311, - "step": 3197 - }, - { - "epoch": 2.59, - "learning_rate": 9.762795507903732e-07, - "loss": 1.0119, - "step": 3198 - }, - { - "epoch": 2.59, - "learning_rate": 9.72517994608142e-07, - "loss": 1.2609, - "step": 3199 - }, - { - "epoch": 2.59, - "learning_rate": 9.687633285490395e-07, - "loss": 1.2287, - "step": 3200 - }, - { - "epoch": 2.59, - "learning_rate": 9.650155554787755e-07, - "loss": 1.1925, - "step": 3201 - }, - { - "epoch": 2.59, - "learning_rate": 9.61274678257791e-07, - "loss": 1.308, - "step": 3202 - }, - { - "epoch": 2.59, - "learning_rate": 9.575406997412695e-07, - "loss": 1.214, - "step": 3203 - }, - { - "epoch": 2.59, - "learning_rate": 9.538136227791318e-07, - "loss": 1.3956, - "step": 3204 - }, - { - "epoch": 2.59, - "learning_rate": 9.500934502160197e-07, - "loss": 1.1626, - "step": 3205 - }, - { - "epoch": 2.59, - "learning_rate": 9.463801848913201e-07, - "loss": 1.201, - "step": 3206 - }, - { - "epoch": 2.59, - "learning_rate": 9.426738296391358e-07, - "loss": 1.1217, - "step": 3207 - }, - { - "epoch": 2.6, - "learning_rate": 9.389743872883017e-07, - "loss": 1.207, - "step": 3208 - }, - { - "epoch": 2.6, - "learning_rate": 9.35281860662377e-07, - "loss": 1.2745, - "step": 3209 - }, - { - "epoch": 2.6, - "learning_rate": 9.315962525796374e-07, - "loss": 1.0763, - "step": 3210 - }, - { - "epoch": 2.6, - "learning_rate": 9.279175658530848e-07, - "loss": 1.1991, - "step": 3211 - }, - { - "epoch": 2.6, - "learning_rate": 9.242458032904311e-07, - "loss": 1.1878, - "step": 3212 - }, - { - "epoch": 2.6, - "learning_rate": 9.205809676941091e-07, - "loss": 1.1184, - "step": 3213 - }, - { - "epoch": 2.6, - "learning_rate": 9.169230618612646e-07, - "loss": 1.1767, - "step": 3214 - }, - { - "epoch": 2.6, - "learning_rate": 9.132720885837509e-07, - "loss": 1.12, - "step": 3215 - }, - { - "epoch": 2.6, - "learning_rate": 9.096280506481337e-07, - "loss": 1.1685, - "step": 3216 - }, - { - "epoch": 2.6, - "learning_rate": 9.059909508356812e-07, - "loss": 1.0719, - "step": 3217 - }, - { - "epoch": 2.6, - "learning_rate": 9.023607919223699e-07, - "loss": 1.2591, - "step": 3218 - }, - { - "epoch": 2.6, - "learning_rate": 8.987375766788786e-07, - "loss": 1.2193, - "step": 3219 - }, - { - "epoch": 2.61, - "learning_rate": 8.951213078705811e-07, - "loss": 1.1034, - "step": 3220 - }, - { - "epoch": 2.61, - "learning_rate": 8.915119882575585e-07, - "loss": 1.2297, - "step": 3221 - }, - { - "epoch": 2.61, - "learning_rate": 8.879096205945792e-07, - "loss": 1.314, - "step": 3222 - }, - { - "epoch": 2.61, - "learning_rate": 8.843142076311128e-07, - "loss": 1.159, - "step": 3223 - }, - { - "epoch": 2.61, - "learning_rate": 8.807257521113144e-07, - "loss": 1.2789, - "step": 3224 - }, - { - "epoch": 2.61, - "learning_rate": 8.771442567740318e-07, - "loss": 1.0523, - "step": 3225 - }, - { - "epoch": 2.61, - "learning_rate": 8.735697243528041e-07, - "loss": 1.0565, - "step": 3226 - }, - { - "epoch": 2.61, - "learning_rate": 8.700021575758478e-07, - "loss": 1.2606, - "step": 3227 - }, - { - "epoch": 2.61, - "learning_rate": 8.664415591660724e-07, - "loss": 1.2051, - "step": 3228 - }, - { - "epoch": 2.61, - "learning_rate": 8.62887931841061e-07, - "loss": 1.3226, - "step": 3229 - }, - { - "epoch": 2.61, - "learning_rate": 8.593412783130805e-07, - "loss": 1.2758, - "step": 3230 - }, - { - "epoch": 2.61, - "learning_rate": 8.558016012890757e-07, - "loss": 1.1833, - "step": 3231 - }, - { - "epoch": 2.61, - "learning_rate": 8.522689034706611e-07, - "loss": 1.064, - "step": 3232 - }, - { - "epoch": 2.62, - "learning_rate": 8.487431875541319e-07, - "loss": 1.0611, - "step": 3233 - }, - { - "epoch": 2.62, - "learning_rate": 8.452244562304446e-07, - "loss": 1.2167, - "step": 3234 - }, - { - "epoch": 2.62, - "learning_rate": 8.417127121852387e-07, - "loss": 1.2214, - "step": 3235 - }, - { - "epoch": 2.62, - "learning_rate": 8.3820795809881e-07, - "loss": 1.2342, - "step": 3236 - }, - { - "epoch": 2.62, - "learning_rate": 8.347101966461202e-07, - "loss": 1.2245, - "step": 3237 - }, - { - "epoch": 2.62, - "learning_rate": 8.312194304967991e-07, - "loss": 1.2831, - "step": 3238 - }, - { - "epoch": 2.62, - "learning_rate": 8.277356623151323e-07, - "loss": 1.1516, - "step": 3239 - }, - { - "epoch": 2.62, - "learning_rate": 8.24258894760066e-07, - "loss": 1.1812, - "step": 3240 - }, - { - "epoch": 2.62, - "learning_rate": 8.207891304852045e-07, - "loss": 1.2012, - "step": 3241 - }, - { - "epoch": 2.62, - "learning_rate": 8.173263721388059e-07, - "loss": 1.0498, - "step": 3242 - }, - { - "epoch": 2.62, - "learning_rate": 8.138706223637827e-07, - "loss": 1.2493, - "step": 3243 - }, - { - "epoch": 2.62, - "learning_rate": 8.10421883797694e-07, - "loss": 1.1227, - "step": 3244 - }, - { - "epoch": 2.63, - "learning_rate": 8.069801590727511e-07, - "loss": 1.295, - "step": 3245 - }, - { - "epoch": 2.63, - "learning_rate": 8.035454508158136e-07, - "loss": 1.0979, - "step": 3246 - }, - { - "epoch": 2.63, - "learning_rate": 8.001177616483801e-07, - "loss": 1.1702, - "step": 3247 - }, - { - "epoch": 2.63, - "learning_rate": 7.96697094186597e-07, - "loss": 1.191, - "step": 3248 - }, - { - "epoch": 2.63, - "learning_rate": 7.93283451041249e-07, - "loss": 1.1959, - "step": 3249 - }, - { - "epoch": 2.63, - "learning_rate": 7.898768348177643e-07, - "loss": 1.081, - "step": 3250 - }, - { - "epoch": 2.63, - "learning_rate": 7.864772481161986e-07, - "loss": 1.2192, - "step": 3251 - }, - { - "epoch": 2.63, - "learning_rate": 7.830846935312508e-07, - "loss": 1.0959, - "step": 3252 - }, - { - "epoch": 2.63, - "learning_rate": 7.796991736522508e-07, - "loss": 1.259, - "step": 3253 - }, - { - "epoch": 2.63, - "learning_rate": 7.763206910631549e-07, - "loss": 1.0687, - "step": 3254 - }, - { - "epoch": 2.63, - "learning_rate": 7.729492483425539e-07, - "loss": 1.2457, - "step": 3255 - }, - { - "epoch": 2.63, - "learning_rate": 7.695848480636614e-07, - "loss": 1.1427, - "step": 3256 - }, - { - "epoch": 2.64, - "learning_rate": 7.662274927943202e-07, - "loss": 1.1535, - "step": 3257 - }, - { - "epoch": 2.64, - "learning_rate": 7.62877185096993e-07, - "loss": 1.0674, - "step": 3258 - }, - { - "epoch": 2.64, - "learning_rate": 7.595339275287628e-07, - "loss": 1.3145, - "step": 3259 - }, - { - "epoch": 2.64, - "learning_rate": 7.561977226413341e-07, - "loss": 1.1106, - "step": 3260 - }, - { - "epoch": 2.64, - "learning_rate": 7.528685729810259e-07, - "loss": 1.1449, - "step": 3261 - }, - { - "epoch": 2.64, - "learning_rate": 7.495464810887743e-07, - "loss": 1.2775, - "step": 3262 - }, - { - "epoch": 2.64, - "learning_rate": 7.462314495001288e-07, - "loss": 1.1112, - "step": 3263 - }, - { - "epoch": 2.64, - "learning_rate": 7.429234807452501e-07, - "loss": 1.0478, - "step": 3264 - }, - { - "epoch": 2.64, - "learning_rate": 7.396225773489086e-07, - "loss": 1.1659, - "step": 3265 - }, - { - "epoch": 2.64, - "learning_rate": 7.363287418304788e-07, - "loss": 1.1438, - "step": 3266 - }, - { - "epoch": 2.64, - "learning_rate": 7.330419767039432e-07, - "loss": 1.0917, - "step": 3267 - }, - { - "epoch": 2.64, - "learning_rate": 7.297622844778917e-07, - "loss": 1.2017, - "step": 3268 - }, - { - "epoch": 2.64, - "learning_rate": 7.264896676555077e-07, - "loss": 1.1032, - "step": 3269 - }, - { - "epoch": 2.65, - "learning_rate": 7.23224128734582e-07, - "loss": 1.1734, - "step": 3270 - }, - { - "epoch": 2.65, - "learning_rate": 7.199656702074975e-07, - "loss": 1.4405, - "step": 3271 - }, - { - "epoch": 2.65, - "learning_rate": 7.167142945612393e-07, - "loss": 1.2168, - "step": 3272 - }, - { - "epoch": 2.65, - "learning_rate": 7.13470004277379e-07, - "loss": 1.2839, - "step": 3273 - }, - { - "epoch": 2.65, - "learning_rate": 7.102328018320859e-07, - "loss": 1.0187, - "step": 3274 - }, - { - "epoch": 2.65, - "learning_rate": 7.070026896961201e-07, - "loss": 1.2202, - "step": 3275 - }, - { - "epoch": 2.65, - "learning_rate": 7.037796703348232e-07, - "loss": 1.0935, - "step": 3276 - }, - { - "epoch": 2.65, - "learning_rate": 7.005637462081327e-07, - "loss": 1.177, - "step": 3277 - }, - { - "epoch": 2.65, - "learning_rate": 6.973549197705632e-07, - "loss": 1.1651, - "step": 3278 - }, - { - "epoch": 2.65, - "learning_rate": 6.941531934712165e-07, - "loss": 1.2006, - "step": 3279 - }, - { - "epoch": 2.65, - "learning_rate": 6.909585697537758e-07, - "loss": 1.1946, - "step": 3280 - }, - { - "epoch": 2.65, - "learning_rate": 6.877710510564972e-07, - "loss": 1.1401, - "step": 3281 - }, - { - "epoch": 2.66, - "learning_rate": 6.845906398122215e-07, - "loss": 1.2839, - "step": 3282 - }, - { - "epoch": 2.66, - "learning_rate": 6.814173384483592e-07, - "loss": 1.2239, - "step": 3283 - }, - { - "epoch": 2.66, - "learning_rate": 6.782511493868949e-07, - "loss": 1.234, - "step": 3284 - }, - { - "epoch": 2.66, - "learning_rate": 6.750920750443923e-07, - "loss": 1.0927, - "step": 3285 - }, - { - "epoch": 2.66, - "learning_rate": 6.71940117831975e-07, - "loss": 1.198, - "step": 3286 - }, - { - "epoch": 2.66, - "learning_rate": 6.687952801553399e-07, - "loss": 1.0925, - "step": 3287 - }, - { - "epoch": 2.66, - "learning_rate": 6.656575644147467e-07, - "loss": 1.2378, - "step": 3288 - }, - { - "epoch": 2.66, - "learning_rate": 6.625269730050232e-07, - "loss": 1.2063, - "step": 3289 - }, - { - "epoch": 2.66, - "learning_rate": 6.594035083155581e-07, - "loss": 1.2455, - "step": 3290 - }, - { - "epoch": 2.66, - "learning_rate": 6.562871727302966e-07, - "loss": 1.2066, - "step": 3291 - }, - { - "epoch": 2.66, - "learning_rate": 6.531779686277528e-07, - "loss": 1.1673, - "step": 3292 - }, - { - "epoch": 2.66, - "learning_rate": 6.500758983809852e-07, - "loss": 1.195, - "step": 3293 - }, - { - "epoch": 2.67, - "learning_rate": 6.469809643576175e-07, - "loss": 1.2203, - "step": 3294 - }, - { - "epoch": 2.67, - "learning_rate": 6.438931689198225e-07, - "loss": 1.172, - "step": 3295 - }, - { - "epoch": 2.67, - "learning_rate": 6.408125144243238e-07, - "loss": 1.1591, - "step": 3296 - }, - { - "epoch": 2.67, - "learning_rate": 6.377390032223974e-07, - "loss": 1.3616, - "step": 3297 - }, - { - "epoch": 2.67, - "learning_rate": 6.346726376598611e-07, - "loss": 1.1919, - "step": 3298 - }, - { - "epoch": 2.67, - "learning_rate": 6.316134200770919e-07, - "loss": 1.2695, - "step": 3299 - }, - { - "epoch": 2.67, - "learning_rate": 6.285613528089962e-07, - "loss": 1.2426, - "step": 3300 - }, - { - "epoch": 2.67, - "learning_rate": 6.25516438185031e-07, - "loss": 1.1707, - "step": 3301 - }, - { - "epoch": 2.67, - "learning_rate": 6.22478678529197e-07, - "loss": 1.0535, - "step": 3302 - }, - { - "epoch": 2.67, - "learning_rate": 6.19448076160023e-07, - "loss": 1.2372, - "step": 3303 - }, - { - "epoch": 2.67, - "learning_rate": 6.164246333905865e-07, - "loss": 1.1067, - "step": 3304 - }, - { - "epoch": 2.67, - "learning_rate": 6.13408352528495e-07, - "loss": 1.0874, - "step": 3305 - }, - { - "epoch": 2.67, - "learning_rate": 6.103992358758892e-07, - "loss": 1.191, - "step": 3306 - }, - { - "epoch": 2.68, - "learning_rate": 6.073972857294463e-07, - "loss": 1.1643, - "step": 3307 - }, - { - "epoch": 2.68, - "learning_rate": 6.04402504380368e-07, - "loss": 1.1004, - "step": 3308 - }, - { - "epoch": 2.68, - "learning_rate": 6.014148941143882e-07, - "loss": 1.0571, - "step": 3309 - }, - { - "epoch": 2.68, - "learning_rate": 5.98434457211765e-07, - "loss": 1.1697, - "step": 3310 - }, - { - "epoch": 2.68, - "learning_rate": 5.954611959472845e-07, - "loss": 1.0623, - "step": 3311 - }, - { - "epoch": 2.68, - "learning_rate": 5.924951125902545e-07, - "loss": 1.3264, - "step": 3312 - }, - { - "epoch": 2.68, - "learning_rate": 5.89536209404502e-07, - "loss": 1.2206, - "step": 3313 - }, - { - "epoch": 2.68, - "learning_rate": 5.865844886483818e-07, - "loss": 1.1501, - "step": 3314 - }, - { - "epoch": 2.68, - "learning_rate": 5.83639952574756e-07, - "loss": 1.0906, - "step": 3315 - }, - { - "epoch": 2.68, - "learning_rate": 5.807026034310104e-07, - "loss": 1.3306, - "step": 3316 - }, - { - "epoch": 2.68, - "learning_rate": 5.777724434590448e-07, - "loss": 1.2115, - "step": 3317 - }, - { - "epoch": 2.68, - "learning_rate": 5.748494748952671e-07, - "loss": 1.1975, - "step": 3318 - }, - { - "epoch": 2.69, - "learning_rate": 5.719336999706027e-07, - "loss": 1.3621, - "step": 3319 - }, - { - "epoch": 2.69, - "learning_rate": 5.690251209104802e-07, - "loss": 1.1653, - "step": 3320 - }, - { - "epoch": 2.69, - "learning_rate": 5.66123739934844e-07, - "loss": 1.1754, - "step": 3321 - }, - { - "epoch": 2.69, - "learning_rate": 5.632295592581394e-07, - "loss": 1.1875, - "step": 3322 - }, - { - "epoch": 2.69, - "learning_rate": 5.603425810893137e-07, - "loss": 1.0885, - "step": 3323 - }, - { - "epoch": 2.69, - "learning_rate": 5.574628076318245e-07, - "loss": 1.2677, - "step": 3324 - }, - { - "epoch": 2.69, - "learning_rate": 5.545902410836224e-07, - "loss": 1.0298, - "step": 3325 - }, - { - "epoch": 2.69, - "learning_rate": 5.517248836371624e-07, - "loss": 1.1955, - "step": 3326 - }, - { - "epoch": 2.69, - "learning_rate": 5.488667374793966e-07, - "loss": 1.2805, - "step": 3327 - }, - { - "epoch": 2.69, - "learning_rate": 5.460158047917719e-07, - "loss": 1.0669, - "step": 3328 - }, - { - "epoch": 2.69, - "learning_rate": 5.431720877502322e-07, - "loss": 1.3431, - "step": 3329 - }, - { - "epoch": 2.69, - "learning_rate": 5.403355885252104e-07, - "loss": 1.2105, - "step": 3330 - }, - { - "epoch": 2.69, - "learning_rate": 5.375063092816313e-07, - "loss": 1.2208, - "step": 3331 - }, - { - "epoch": 2.7, - "learning_rate": 5.346842521789142e-07, - "loss": 1.1947, - "step": 3332 - }, - { - "epoch": 2.7, - "learning_rate": 5.318694193709573e-07, - "loss": 1.3115, - "step": 3333 - }, - { - "epoch": 2.7, - "learning_rate": 5.29061813006152e-07, - "loss": 1.095, - "step": 3334 - }, - { - "epoch": 2.7, - "learning_rate": 5.262614352273732e-07, - "loss": 1.2003, - "step": 3335 - }, - { - "epoch": 2.7, - "learning_rate": 5.234682881719766e-07, - "loss": 1.2959, - "step": 3336 - }, - { - "epoch": 2.7, - "learning_rate": 5.206823739717981e-07, - "loss": 1.2385, - "step": 3337 - }, - { - "epoch": 2.7, - "learning_rate": 5.179036947531568e-07, - "loss": 1.0673, - "step": 3338 - }, - { - "epoch": 2.7, - "learning_rate": 5.151322526368496e-07, - "loss": 1.2148, - "step": 3339 - }, - { - "epoch": 2.7, - "learning_rate": 5.123680497381444e-07, - "loss": 1.2455, - "step": 3340 - }, - { - "epoch": 2.7, - "learning_rate": 5.096110881667904e-07, - "loss": 1.2426, - "step": 3341 - }, - { - "epoch": 2.7, - "learning_rate": 5.068613700270064e-07, - "loss": 1.2661, - "step": 3342 - }, - { - "epoch": 2.7, - "learning_rate": 5.041188974174826e-07, - "loss": 1.1448, - "step": 3343 - }, - { - "epoch": 2.71, - "learning_rate": 5.013836724313836e-07, - "loss": 1.2599, - "step": 3344 - }, - { - "epoch": 2.71, - "learning_rate": 4.98655697156335e-07, - "loss": 1.1351, - "step": 3345 - }, - { - "epoch": 2.71, - "learning_rate": 4.959349736744357e-07, - "loss": 1.0227, - "step": 3346 - }, - { - "epoch": 2.71, - "learning_rate": 4.932215040622446e-07, - "loss": 1.1979, - "step": 3347 - }, - { - "epoch": 2.71, - "learning_rate": 4.905152903907873e-07, - "loss": 1.2408, - "step": 3348 - }, - { - "epoch": 2.71, - "learning_rate": 4.878163347255505e-07, - "loss": 1.1593, - "step": 3349 - }, - { - "epoch": 2.71, - "learning_rate": 4.851246391264819e-07, - "loss": 1.0801, - "step": 3350 - }, - { - "epoch": 2.71, - "learning_rate": 4.824402056479893e-07, - "loss": 1.2415, - "step": 3351 - }, - { - "epoch": 2.71, - "learning_rate": 4.797630363389339e-07, - "loss": 1.1718, - "step": 3352 - }, - { - "epoch": 2.71, - "learning_rate": 4.770931332426343e-07, - "loss": 1.3582, - "step": 3353 - }, - { - "epoch": 2.71, - "learning_rate": 4.7443049839686617e-07, - "loss": 1.2353, - "step": 3354 - }, - { - "epoch": 2.71, - "learning_rate": 4.7177513383385055e-07, - "loss": 1.2919, - "step": 3355 - }, - { - "epoch": 2.72, - "learning_rate": 4.6912704158027065e-07, - "loss": 1.222, - "step": 3356 - }, - { - "epoch": 2.72, - "learning_rate": 4.664862236572487e-07, - "loss": 1.1172, - "step": 3357 - }, - { - "epoch": 2.72, - "learning_rate": 4.6385268208036125e-07, - "loss": 1.2163, - "step": 3358 - }, - { - "epoch": 2.72, - "learning_rate": 4.6122641885962715e-07, - "loss": 1.0303, - "step": 3359 - }, - { - "epoch": 2.72, - "learning_rate": 4.5860743599951186e-07, - "loss": 1.1205, - "step": 3360 - }, - { - "epoch": 2.72, - "learning_rate": 4.559957354989264e-07, - "loss": 1.0419, - "step": 3361 - }, - { - "epoch": 2.72, - "learning_rate": 4.5339131935121737e-07, - "loss": 1.1156, - "step": 3362 - }, - { - "epoch": 2.72, - "learning_rate": 4.5079418954418143e-07, - "loss": 1.2089, - "step": 3363 - }, - { - "epoch": 2.72, - "learning_rate": 4.482043480600451e-07, - "loss": 1.084, - "step": 3364 - }, - { - "epoch": 2.72, - "learning_rate": 4.4562179687547504e-07, - "loss": 1.0983, - "step": 3365 - }, - { - "epoch": 2.72, - "learning_rate": 4.430465379615767e-07, - "loss": 1.2637, - "step": 3366 - }, - { - "epoch": 2.72, - "learning_rate": 4.4047857328388457e-07, - "loss": 1.1793, - "step": 3367 - }, - { - "epoch": 2.72, - "learning_rate": 4.379179048023707e-07, - "loss": 1.25, - "step": 3368 - }, - { - "epoch": 2.73, - "learning_rate": 4.35364534471433e-07, - "loss": 1.2015, - "step": 3369 - }, - { - "epoch": 2.73, - "learning_rate": 4.328184642399036e-07, - "loss": 1.0798, - "step": 3370 - }, - { - "epoch": 2.73, - "learning_rate": 4.302796960510458e-07, - "loss": 1.1379, - "step": 3371 - }, - { - "epoch": 2.73, - "learning_rate": 4.2774823184254075e-07, - "loss": 1.1921, - "step": 3372 - }, - { - "epoch": 2.73, - "learning_rate": 4.2522407354650184e-07, - "loss": 1.279, - "step": 3373 - }, - { - "epoch": 2.73, - "learning_rate": 4.2270722308946356e-07, - "loss": 1.096, - "step": 3374 - }, - { - "epoch": 2.73, - "learning_rate": 4.201976823923837e-07, - "loss": 1.2263, - "step": 3375 - }, - { - "epoch": 2.73, - "learning_rate": 4.176954533706412e-07, - "loss": 1.1903, - "step": 3376 - }, - { - "epoch": 2.73, - "learning_rate": 4.152005379340307e-07, - "loss": 1.1315, - "step": 3377 - }, - { - "epoch": 2.73, - "learning_rate": 4.127129379867734e-07, - "loss": 1.1552, - "step": 3378 - }, - { - "epoch": 2.73, - "learning_rate": 4.1023265542749825e-07, - "loss": 1.2503, - "step": 3379 - }, - { - "epoch": 2.73, - "learning_rate": 4.077596921492533e-07, - "loss": 1.1214, - "step": 3380 - }, - { - "epoch": 2.74, - "learning_rate": 4.052940500395008e-07, - "loss": 1.1628, - "step": 3381 - }, - { - "epoch": 2.74, - "learning_rate": 4.028357309801123e-07, - "loss": 1.2079, - "step": 3382 - }, - { - "epoch": 2.74, - "learning_rate": 4.0038473684737453e-07, - "loss": 1.0909, - "step": 3383 - }, - { - "epoch": 2.74, - "learning_rate": 3.9794106951197677e-07, - "loss": 1.2185, - "step": 3384 - }, - { - "epoch": 2.74, - "learning_rate": 3.9550473083902475e-07, - "loss": 1.134, - "step": 3385 - }, - { - "epoch": 2.74, - "learning_rate": 3.930757226880233e-07, - "loss": 1.2201, - "step": 3386 - }, - { - "epoch": 2.74, - "learning_rate": 3.906540469128861e-07, - "loss": 1.0801, - "step": 3387 - }, - { - "epoch": 2.74, - "learning_rate": 3.8823970536193133e-07, - "loss": 1.2288, - "step": 3388 - }, - { - "epoch": 2.74, - "learning_rate": 3.8583269987787607e-07, - "loss": 1.4702, - "step": 3389 - }, - { - "epoch": 2.74, - "learning_rate": 3.834330322978397e-07, - "loss": 1.2662, - "step": 3390 - }, - { - "epoch": 2.74, - "learning_rate": 3.810407044533415e-07, - "loss": 1.1307, - "step": 3391 - }, - { - "epoch": 2.74, - "learning_rate": 3.786557181702988e-07, - "loss": 1.1885, - "step": 3392 - }, - { - "epoch": 2.75, - "learning_rate": 3.762780752690265e-07, - "loss": 1.3015, - "step": 3393 - }, - { - "epoch": 2.75, - "learning_rate": 3.739077775642319e-07, - "loss": 1.1393, - "step": 3394 - }, - { - "epoch": 2.75, - "learning_rate": 3.7154482686501904e-07, - "loss": 1.0995, - "step": 3395 - }, - { - "epoch": 2.75, - "learning_rate": 3.6918922497488185e-07, - "loss": 1.2034, - "step": 3396 - }, - { - "epoch": 2.75, - "learning_rate": 3.668409736917067e-07, - "loss": 1.2243, - "step": 3397 - }, - { - "epoch": 2.75, - "learning_rate": 3.645000748077709e-07, - "loss": 1.3606, - "step": 3398 - }, - { - "epoch": 2.75, - "learning_rate": 3.621665301097377e-07, - "loss": 1.2144, - "step": 3399 - }, - { - "epoch": 2.75, - "learning_rate": 3.598403413786611e-07, - "loss": 1.3035, - "step": 3400 - }, - { - "epoch": 2.75, - "learning_rate": 3.575215103899743e-07, - "loss": 1.0898, - "step": 3401 - }, - { - "epoch": 2.75, - "learning_rate": 3.552100389135016e-07, - "loss": 1.1503, - "step": 3402 - }, - { - "epoch": 2.75, - "learning_rate": 3.5290592871344844e-07, - "loss": 1.091, - "step": 3403 - }, - { - "epoch": 2.75, - "learning_rate": 3.5060918154839696e-07, - "loss": 1.1714, - "step": 3404 - }, - { - "epoch": 2.75, - "learning_rate": 3.4831979917131495e-07, - "loss": 1.2825, - "step": 3405 - }, - { - "epoch": 2.76, - "learning_rate": 3.4603778332954807e-07, - "loss": 1.2636, - "step": 3406 - }, - { - "epoch": 2.76, - "learning_rate": 3.4376313576481966e-07, - "loss": 1.1347, - "step": 3407 - }, - { - "epoch": 2.76, - "learning_rate": 3.414958582132277e-07, - "loss": 1.3028, - "step": 3408 - }, - { - "epoch": 2.76, - "learning_rate": 3.3923595240524575e-07, - "loss": 1.1145, - "step": 3409 - }, - { - "epoch": 2.76, - "learning_rate": 3.3698342006572294e-07, - "loss": 1.1226, - "step": 3410 - }, - { - "epoch": 2.76, - "learning_rate": 3.347382629138762e-07, - "loss": 1.1554, - "step": 3411 - }, - { - "epoch": 2.76, - "learning_rate": 3.3250048266329826e-07, - "loss": 1.143, - "step": 3412 - }, - { - "epoch": 2.76, - "learning_rate": 3.3027008102194835e-07, - "loss": 1.2404, - "step": 3413 - }, - { - "epoch": 2.76, - "learning_rate": 3.280470596921559e-07, - "loss": 1.1588, - "step": 3414 - }, - { - "epoch": 2.76, - "learning_rate": 3.258314203706181e-07, - "loss": 0.9839, - "step": 3415 - }, - { - "epoch": 2.76, - "learning_rate": 3.236231647483945e-07, - "loss": 1.2553, - "step": 3416 - }, - { - "epoch": 2.76, - "learning_rate": 3.2142229451091025e-07, - "loss": 1.0705, - "step": 3417 - }, - { - "epoch": 2.77, - "learning_rate": 3.1922881133795827e-07, - "loss": 1.2067, - "step": 3418 - }, - { - "epoch": 2.77, - "learning_rate": 3.1704271690368494e-07, - "loss": 1.0487, - "step": 3419 - }, - { - "epoch": 2.77, - "learning_rate": 3.148640128766056e-07, - "loss": 1.1252, - "step": 3420 - }, - { - "epoch": 2.77, - "learning_rate": 3.12692700919589e-07, - "loss": 1.1183, - "step": 3421 - }, - { - "epoch": 2.77, - "learning_rate": 3.105287826898673e-07, - "loss": 1.2043, - "step": 3422 - }, - { - "epoch": 2.77, - "learning_rate": 3.0837225983902376e-07, - "loss": 1.2392, - "step": 3423 - }, - { - "epoch": 2.77, - "learning_rate": 3.0622313401300173e-07, - "loss": 1.3706, - "step": 3424 - }, - { - "epoch": 2.77, - "learning_rate": 3.0408140685209585e-07, - "loss": 1.1874, - "step": 3425 - }, - { - "epoch": 2.77, - "learning_rate": 3.019470799909552e-07, - "loss": 1.0893, - "step": 3426 - }, - { - "epoch": 2.77, - "learning_rate": 2.998201550585811e-07, - "loss": 1.1885, - "step": 3427 - }, - { - "epoch": 2.77, - "learning_rate": 2.9770063367832393e-07, - "loss": 1.1761, - "step": 3428 - }, - { - "epoch": 2.77, - "learning_rate": 2.955885174678852e-07, - "loss": 1.1207, - "step": 3429 - }, - { - "epoch": 2.78, - "learning_rate": 2.934838080393154e-07, - "loss": 1.126, - "step": 3430 - }, - { - "epoch": 2.78, - "learning_rate": 2.9138650699900627e-07, - "loss": 1.072, - "step": 3431 - }, - { - "epoch": 2.78, - "learning_rate": 2.8929661594770173e-07, - "loss": 1.0604, - "step": 3432 - }, - { - "epoch": 2.78, - "learning_rate": 2.872141364804848e-07, - "loss": 1.2393, - "step": 3433 - }, - { - "epoch": 2.78, - "learning_rate": 2.851390701867873e-07, - "loss": 1.2702, - "step": 3434 - }, - { - "epoch": 2.78, - "learning_rate": 2.83071418650378e-07, - "loss": 1.1504, - "step": 3435 - }, - { - "epoch": 2.78, - "learning_rate": 2.8101118344936893e-07, - "loss": 1.253, - "step": 3436 - }, - { - "epoch": 2.78, - "learning_rate": 2.789583661562134e-07, - "loss": 1.3285, - "step": 3437 - }, - { - "epoch": 2.78, - "learning_rate": 2.76912968337697e-07, - "loss": 1.1758, - "step": 3438 - }, - { - "epoch": 2.78, - "learning_rate": 2.748749915549487e-07, - "loss": 1.265, - "step": 3439 - }, - { - "epoch": 2.78, - "learning_rate": 2.7284443736343203e-07, - "loss": 1.1386, - "step": 3440 - }, - { - "epoch": 2.78, - "learning_rate": 2.7082130731294174e-07, - "loss": 1.1628, - "step": 3441 - }, - { - "epoch": 2.78, - "learning_rate": 2.688056029476116e-07, - "loss": 1.1846, - "step": 3442 - }, - { - "epoch": 2.79, - "learning_rate": 2.66797325805902e-07, - "loss": 1.2168, - "step": 3443 - }, - { - "epoch": 2.79, - "learning_rate": 2.6479647742061133e-07, - "loss": 1.1049, - "step": 3444 - }, - { - "epoch": 2.79, - "learning_rate": 2.628030593188613e-07, - "loss": 1.2085, - "step": 3445 - }, - { - "epoch": 2.79, - "learning_rate": 2.6081707302210714e-07, - "loss": 1.1596, - "step": 3446 - }, - { - "epoch": 2.79, - "learning_rate": 2.588385200461307e-07, - "loss": 1.1347, - "step": 3447 - }, - { - "epoch": 2.79, - "learning_rate": 2.5686740190103754e-07, - "loss": 1.138, - "step": 3448 - }, - { - "epoch": 2.79, - "learning_rate": 2.549037200912641e-07, - "loss": 1.1714, - "step": 3449 - }, - { - "epoch": 2.79, - "learning_rate": 2.52947476115567e-07, - "loss": 1.2033, - "step": 3450 - }, - { - "epoch": 2.79, - "learning_rate": 2.509986714670276e-07, - "loss": 1.2234, - "step": 3451 - }, - { - "epoch": 2.79, - "learning_rate": 2.4905730763305046e-07, - "loss": 1.1616, - "step": 3452 - }, - { - "epoch": 2.79, - "learning_rate": 2.471233860953559e-07, - "loss": 1.1522, - "step": 3453 - }, - { - "epoch": 2.79, - "learning_rate": 2.4519690832998985e-07, - "loss": 1.2102, - "step": 3454 - }, - { - "epoch": 2.8, - "learning_rate": 2.432778758073129e-07, - "loss": 1.2686, - "step": 3455 - }, - { - "epoch": 2.8, - "learning_rate": 2.4136628999200775e-07, - "loss": 1.2856, - "step": 3456 - }, - { - "epoch": 2.8, - "learning_rate": 2.394621523430685e-07, - "loss": 1.1855, - "step": 3457 - }, - { - "epoch": 2.8, - "learning_rate": 2.37565464313807e-07, - "loss": 1.07, - "step": 3458 - }, - { - "epoch": 2.8, - "learning_rate": 2.356762273518498e-07, - "loss": 1.1005, - "step": 3459 - }, - { - "epoch": 2.8, - "learning_rate": 2.3379444289913344e-07, - "loss": 1.306, - "step": 3460 - }, - { - "epoch": 2.8, - "learning_rate": 2.3192011239191014e-07, - "loss": 1.3277, - "step": 3461 - }, - { - "epoch": 2.8, - "learning_rate": 2.3005323726074002e-07, - "loss": 1.312, - "step": 3462 - }, - { - "epoch": 2.8, - "learning_rate": 2.2819381893049662e-07, - "loss": 1.1136, - "step": 3463 - }, - { - "epoch": 2.8, - "learning_rate": 2.263418588203581e-07, - "loss": 1.1796, - "step": 3464 - }, - { - "epoch": 2.8, - "learning_rate": 2.2449735834381258e-07, - "loss": 1.3325, - "step": 3465 - }, - { - "epoch": 2.8, - "learning_rate": 2.2266031890865293e-07, - "loss": 1.2863, - "step": 3466 - }, - { - "epoch": 2.81, - "learning_rate": 2.2083074191698195e-07, - "loss": 1.1532, - "step": 3467 - }, - { - "epoch": 2.81, - "learning_rate": 2.1900862876519935e-07, - "loss": 1.1728, - "step": 3468 - }, - { - "epoch": 2.81, - "learning_rate": 2.1719398084401376e-07, - "loss": 1.2022, - "step": 3469 - }, - { - "epoch": 2.81, - "learning_rate": 2.153867995384351e-07, - "loss": 1.2278, - "step": 3470 - }, - { - "epoch": 2.81, - "learning_rate": 2.1358708622777556e-07, - "loss": 1.3607, - "step": 3471 - }, - { - "epoch": 2.81, - "learning_rate": 2.1179484228564308e-07, - "loss": 1.137, - "step": 3472 - }, - { - "epoch": 2.81, - "learning_rate": 2.100100690799478e-07, - "loss": 1.0344, - "step": 3473 - }, - { - "epoch": 2.81, - "learning_rate": 2.0823276797290015e-07, - "loss": 1.1882, - "step": 3474 - }, - { - "epoch": 2.81, - "learning_rate": 2.0646294032100167e-07, - "loss": 1.133, - "step": 3475 - }, - { - "epoch": 2.81, - "learning_rate": 2.0470058747505516e-07, - "loss": 1.1054, - "step": 3476 - }, - { - "epoch": 2.81, - "learning_rate": 2.0294571078015578e-07, - "loss": 1.2127, - "step": 3477 - }, - { - "epoch": 2.81, - "learning_rate": 2.0119831157569435e-07, - "loss": 1.2871, - "step": 3478 - }, - { - "epoch": 2.81, - "learning_rate": 1.9945839119535293e-07, - "loss": 1.0838, - "step": 3479 - }, - { - "epoch": 2.82, - "learning_rate": 1.9772595096710477e-07, - "loss": 1.1148, - "step": 3480 - }, - { - "epoch": 2.82, - "learning_rate": 1.960009922132178e-07, - "loss": 1.2333, - "step": 3481 - }, - { - "epoch": 2.82, - "learning_rate": 1.9428351625024433e-07, - "loss": 1.1004, - "step": 3482 - }, - { - "epoch": 2.82, - "learning_rate": 1.9257352438902922e-07, - "loss": 1.3024, - "step": 3483 - }, - { - "epoch": 2.82, - "learning_rate": 1.908710179347051e-07, - "loss": 1.1327, - "step": 3484 - }, - { - "epoch": 2.82, - "learning_rate": 1.891759981866903e-07, - "loss": 1.3407, - "step": 3485 - }, - { - "epoch": 2.82, - "learning_rate": 1.8748846643869e-07, - "loss": 1.2057, - "step": 3486 - }, - { - "epoch": 2.82, - "learning_rate": 1.858084239786917e-07, - "loss": 1.3413, - "step": 3487 - }, - { - "epoch": 2.82, - "learning_rate": 1.8413587208897077e-07, - "loss": 1.2708, - "step": 3488 - }, - { - "epoch": 2.82, - "learning_rate": 1.8247081204608386e-07, - "loss": 1.1568, - "step": 3489 - }, - { - "epoch": 2.82, - "learning_rate": 1.8081324512086663e-07, - "loss": 1.1464, - "step": 3490 - }, - { - "epoch": 2.82, - "learning_rate": 1.791631725784404e-07, - "loss": 1.2326, - "step": 3491 - }, - { - "epoch": 2.83, - "learning_rate": 1.7752059567820335e-07, - "loss": 1.1003, - "step": 3492 - }, - { - "epoch": 2.83, - "learning_rate": 1.758855156738326e-07, - "loss": 1.1823, - "step": 3493 - }, - { - "epoch": 2.83, - "learning_rate": 1.742579338132877e-07, - "loss": 1.2416, - "step": 3494 - }, - { - "epoch": 2.83, - "learning_rate": 1.726378513387994e-07, - "loss": 1.1577, - "step": 3495 - }, - { - "epoch": 2.83, - "learning_rate": 1.7102526948687748e-07, - "loss": 1.118, - "step": 3496 - }, - { - "epoch": 2.83, - "learning_rate": 1.6942018948830742e-07, - "loss": 1.166, - "step": 3497 - }, - { - "epoch": 2.83, - "learning_rate": 1.6782261256814703e-07, - "loss": 1.1314, - "step": 3498 - }, - { - "epoch": 2.83, - "learning_rate": 1.6623253994572986e-07, - "loss": 1.5226, - "step": 3499 - }, - { - "epoch": 2.83, - "learning_rate": 1.6464997283466067e-07, - "loss": 1.2905, - "step": 3500 - }, - { - "epoch": 2.83, - "learning_rate": 1.630749124428166e-07, - "loss": 1.2503, - "step": 3501 - }, - { - "epoch": 2.83, - "learning_rate": 1.6150735997234268e-07, - "loss": 1.2521, - "step": 3502 - }, - { - "epoch": 2.83, - "learning_rate": 1.5994731661965635e-07, - "loss": 1.1981, - "step": 3503 - }, - { - "epoch": 2.83, - "learning_rate": 1.5839478357544514e-07, - "loss": 1.194, - "step": 3504 - }, - { - "epoch": 2.84, - "learning_rate": 1.5684976202465786e-07, - "loss": 1.1495, - "step": 3505 - }, - { - "epoch": 2.84, - "learning_rate": 1.55312253146519e-07, - "loss": 1.1926, - "step": 3506 - }, - { - "epoch": 2.84, - "learning_rate": 1.5378225811451208e-07, - "loss": 0.9964, - "step": 3507 - }, - { - "epoch": 2.84, - "learning_rate": 1.5225977809639082e-07, - "loss": 1.3099, - "step": 3508 - }, - { - "epoch": 2.84, - "learning_rate": 1.5074481425416786e-07, - "loss": 1.1585, - "step": 3509 - }, - { - "epoch": 2.84, - "learning_rate": 1.492373677441228e-07, - "loss": 1.1233, - "step": 3510 - }, - { - "epoch": 2.84, - "learning_rate": 1.4773743971679965e-07, - "loss": 1.0769, - "step": 3511 - }, - { - "epoch": 2.84, - "learning_rate": 1.462450313169983e-07, - "loss": 1.2018, - "step": 3512 - }, - { - "epoch": 2.84, - "learning_rate": 1.447601436837842e-07, - "loss": 1.1104, - "step": 3513 - }, - { - "epoch": 2.84, - "learning_rate": 1.4328277795048085e-07, - "loss": 1.19, - "step": 3514 - }, - { - "epoch": 2.84, - "learning_rate": 1.4181293524466955e-07, - "loss": 1.2206, - "step": 3515 - }, - { - "epoch": 2.84, - "learning_rate": 1.4035061668819404e-07, - "loss": 1.311, - "step": 3516 - }, - { - "epoch": 2.85, - "learning_rate": 1.3889582339714937e-07, - "loss": 1.3148, - "step": 3517 - }, - { - "epoch": 2.85, - "learning_rate": 1.3744855648189171e-07, - "loss": 1.2876, - "step": 3518 - }, - { - "epoch": 2.85, - "learning_rate": 1.3600881704702863e-07, - "loss": 1.0483, - "step": 3519 - }, - { - "epoch": 2.85, - "learning_rate": 1.3457660619142887e-07, - "loss": 1.2989, - "step": 3520 - }, - { - "epoch": 2.85, - "learning_rate": 1.3315192500820806e-07, - "loss": 1.2144, - "step": 3521 - }, - { - "epoch": 2.85, - "learning_rate": 1.317347745847386e-07, - "loss": 1.2323, - "step": 3522 - }, - { - "epoch": 2.85, - "learning_rate": 1.303251560026453e-07, - "loss": 1.1653, - "step": 3523 - }, - { - "epoch": 2.85, - "learning_rate": 1.2892307033780305e-07, - "loss": 1.257, - "step": 3524 - }, - { - "epoch": 2.85, - "learning_rate": 1.27528518660337e-07, - "loss": 1.159, - "step": 3525 - }, - { - "epoch": 2.85, - "learning_rate": 1.261415020346235e-07, - "loss": 1.252, - "step": 3526 - }, - { - "epoch": 2.85, - "learning_rate": 1.2476202151928685e-07, - "loss": 1.0884, - "step": 3527 - }, - { - "epoch": 2.85, - "learning_rate": 1.233900781672026e-07, - "loss": 1.0297, - "step": 3528 - }, - { - "epoch": 2.86, - "learning_rate": 1.2202567302548652e-07, - "loss": 1.1886, - "step": 3529 - }, - { - "epoch": 2.86, - "learning_rate": 1.2066880713550888e-07, - "loss": 1.3311, - "step": 3530 - }, - { - "epoch": 2.86, - "learning_rate": 1.193194815328802e-07, - "loss": 1.0548, - "step": 3531 - }, - { - "epoch": 2.86, - "learning_rate": 1.1797769724745889e-07, - "loss": 1.1563, - "step": 3532 - }, - { - "epoch": 2.86, - "learning_rate": 1.1664345530334576e-07, - "loss": 1.18, - "step": 3533 - }, - { - "epoch": 2.86, - "learning_rate": 1.1531675671888621e-07, - "loss": 1.3143, - "step": 3534 - }, - { - "epoch": 2.86, - "learning_rate": 1.1399760250666803e-07, - "loss": 1.1333, - "step": 3535 - }, - { - "epoch": 2.86, - "learning_rate": 1.126859936735203e-07, - "loss": 1.0795, - "step": 3536 - }, - { - "epoch": 2.86, - "learning_rate": 1.1138193122051444e-07, - "loss": 1.0982, - "step": 3537 - }, - { - "epoch": 2.86, - "learning_rate": 1.1008541614296097e-07, - "loss": 1.2988, - "step": 3538 - }, - { - "epoch": 2.86, - "learning_rate": 1.0879644943040946e-07, - "loss": 1.3072, - "step": 3539 - }, - { - "epoch": 2.86, - "learning_rate": 1.0751503206665071e-07, - "loss": 1.142, - "step": 3540 - }, - { - "epoch": 2.86, - "learning_rate": 1.0624116502970905e-07, - "loss": 1.1339, - "step": 3541 - }, - { - "epoch": 2.87, - "learning_rate": 1.0497484929185232e-07, - "loss": 1.3341, - "step": 3542 - }, - { - "epoch": 2.87, - "learning_rate": 1.037160858195807e-07, - "loss": 1.082, - "step": 3543 - }, - { - "epoch": 2.87, - "learning_rate": 1.0246487557362905e-07, - "loss": 1.1251, - "step": 3544 - }, - { - "epoch": 2.87, - "learning_rate": 1.0122121950897235e-07, - "loss": 1.0938, - "step": 3545 - }, - { - "epoch": 2.87, - "learning_rate": 9.998511857481574e-08, - "loss": 1.2421, - "step": 3546 - }, - { - "epoch": 2.87, - "learning_rate": 9.87565737145979e-08, - "loss": 1.1536, - "step": 3547 - }, - { - "epoch": 2.87, - "learning_rate": 9.753558586599432e-08, - "loss": 1.1969, - "step": 3548 - }, - { - "epoch": 2.87, - "learning_rate": 9.632215596090733e-08, - "loss": 1.3668, - "step": 3549 - }, - { - "epoch": 2.87, - "learning_rate": 9.511628492547609e-08, - "loss": 1.1429, - "step": 3550 - }, - { - "epoch": 2.87, - "learning_rate": 9.391797368006661e-08, - "loss": 1.3498, - "step": 3551 - }, - { - "epoch": 2.87, - "learning_rate": 9.272722313927618e-08, - "loss": 1.3509, - "step": 3552 - }, - { - "epoch": 2.87, - "learning_rate": 9.154403421193226e-08, - "loss": 1.2437, - "step": 3553 - }, - { - "epoch": 2.88, - "learning_rate": 9.036840780109024e-08, - "loss": 1.1241, - "step": 3554 - }, - { - "epoch": 2.88, - "learning_rate": 8.92003448040335e-08, - "loss": 1.0443, - "step": 3555 - }, - { - "epoch": 2.88, - "learning_rate": 8.803984611227223e-08, - "loss": 1.4045, - "step": 3556 - }, - { - "epoch": 2.88, - "learning_rate": 8.688691261154569e-08, - "loss": 1.2782, - "step": 3557 - }, - { - "epoch": 2.88, - "learning_rate": 8.574154518181555e-08, - "loss": 1.1435, - "step": 3558 - }, - { - "epoch": 2.88, - "learning_rate": 8.460374469727029e-08, - "loss": 1.1359, - "step": 3559 - }, - { - "epoch": 2.88, - "learning_rate": 8.347351202632525e-08, - "loss": 1.1665, - "step": 3560 - }, - { - "epoch": 2.88, - "learning_rate": 8.235084803161597e-08, - "loss": 1.229, - "step": 3561 - }, - { - "epoch": 2.88, - "learning_rate": 8.123575357000257e-08, - "loss": 1.1466, - "step": 3562 - }, - { - "epoch": 2.88, - "learning_rate": 8.012822949256981e-08, - "loss": 1.1816, - "step": 3563 - }, - { - "epoch": 2.88, - "learning_rate": 7.902827664462154e-08, - "loss": 1.1301, - "step": 3564 - }, - { - "epoch": 2.88, - "learning_rate": 7.793589586568617e-08, - "loss": 1.3679, - "step": 3565 - }, - { - "epoch": 2.89, - "learning_rate": 7.685108798950903e-08, - "loss": 1.0455, - "step": 3566 - }, - { - "epoch": 2.89, - "learning_rate": 7.577385384405778e-08, - "loss": 1.0191, - "step": 3567 - }, - { - "epoch": 2.89, - "learning_rate": 7.470419425152031e-08, - "loss": 1.2242, - "step": 3568 - }, - { - "epoch": 2.89, - "learning_rate": 7.364211002830024e-08, - "loss": 1.3142, - "step": 3569 - }, - { - "epoch": 2.89, - "learning_rate": 7.258760198502246e-08, - "loss": 1.1583, - "step": 3570 - }, - { - "epoch": 2.89, - "learning_rate": 7.154067092652761e-08, - "loss": 1.1515, - "step": 3571 - }, - { - "epoch": 2.89, - "learning_rate": 7.050131765187541e-08, - "loss": 1.1635, - "step": 3572 - }, - { - "epoch": 2.89, - "learning_rate": 6.946954295433906e-08, - "loss": 1.0995, - "step": 3573 - }, - { - "epoch": 2.89, - "learning_rate": 6.844534762140864e-08, - "loss": 1.2054, - "step": 3574 - }, - { - "epoch": 2.89, - "learning_rate": 6.742873243479109e-08, - "loss": 1.0971, - "step": 3575 - }, - { - "epoch": 2.89, - "learning_rate": 6.641969817040572e-08, - "loss": 1.2271, - "step": 3576 - }, - { - "epoch": 2.89, - "learning_rate": 6.541824559838539e-08, - "loss": 1.2077, - "step": 3577 - }, - { - "epoch": 2.89, - "learning_rate": 6.442437548307867e-08, - "loss": 1.1863, - "step": 3578 - }, - { - "epoch": 2.9, - "learning_rate": 6.343808858304546e-08, - "loss": 1.1725, - "step": 3579 - }, - { - "epoch": 2.9, - "learning_rate": 6.245938565105803e-08, - "loss": 1.2843, - "step": 3580 - }, - { - "epoch": 2.9, - "learning_rate": 6.148826743409997e-08, - "loss": 1.1286, - "step": 3581 - }, - { - "epoch": 2.9, - "learning_rate": 6.052473467336617e-08, - "loss": 1.1009, - "step": 3582 - }, - { - "epoch": 2.9, - "learning_rate": 5.956878810426281e-08, - "loss": 1.2386, - "step": 3583 - }, - { - "epoch": 2.9, - "learning_rate": 5.862042845640403e-08, - "loss": 1.325, - "step": 3584 - }, - { - "epoch": 2.9, - "learning_rate": 5.767965645361529e-08, - "loss": 1.2652, - "step": 3585 - }, - { - "epoch": 2.9, - "learning_rate": 5.674647281393109e-08, - "loss": 1.1757, - "step": 3586 - }, - { - "epoch": 2.9, - "learning_rate": 5.5820878249591706e-08, - "loss": 1.2131, - "step": 3587 - }, - { - "epoch": 2.9, - "learning_rate": 5.490287346704648e-08, - "loss": 1.1482, - "step": 3588 - }, - { - "epoch": 2.9, - "learning_rate": 5.3992459166953836e-08, - "loss": 1.1165, - "step": 3589 - }, - { - "epoch": 2.9, - "learning_rate": 5.308963604417572e-08, - "loss": 1.2039, - "step": 3590 - }, - { - "epoch": 2.91, - "learning_rate": 5.219440478778204e-08, - "loss": 1.1854, - "step": 3591 - }, - { - "epoch": 2.91, - "learning_rate": 5.1306766081048456e-08, - "loss": 1.174, - "step": 3592 - }, - { - "epoch": 2.91, - "learning_rate": 5.042672060145415e-08, - "loss": 1.1246, - "step": 3593 - }, - { - "epoch": 2.91, - "learning_rate": 4.9554269020685165e-08, - "loss": 1.2183, - "step": 3594 - }, - { - "epoch": 2.91, - "learning_rate": 4.868941200462884e-08, - "loss": 1.2135, - "step": 3595 - }, - { - "epoch": 2.91, - "learning_rate": 4.783215021337828e-08, - "loss": 1.1635, - "step": 3596 - }, - { - "epoch": 2.91, - "learning_rate": 4.698248430122898e-08, - "loss": 1.2315, - "step": 3597 - }, - { - "epoch": 2.91, - "learning_rate": 4.6140414916676646e-08, - "loss": 1.2133, - "step": 3598 - }, - { - "epoch": 2.91, - "learning_rate": 4.5305942702424945e-08, - "loss": 1.1334, - "step": 3599 - }, - { - "epoch": 2.91, - "learning_rate": 4.447906829537219e-08, - "loss": 1.0716, - "step": 3600 - }, - { - "epoch": 2.91, - "learning_rate": 4.365979232662243e-08, - "loss": 1.2123, - "step": 3601 - }, - { - "epoch": 2.91, - "learning_rate": 4.2848115421477706e-08, - "loss": 1.0648, - "step": 3602 - }, - { - "epoch": 2.92, - "learning_rate": 4.204403819944247e-08, - "loss": 1.239, - "step": 3603 - }, - { - "epoch": 2.92, - "learning_rate": 4.124756127421914e-08, - "loss": 1.2208, - "step": 3604 - }, - { - "epoch": 2.92, - "learning_rate": 4.045868525370811e-08, - "loss": 1.228, - "step": 3605 - }, - { - "epoch": 2.92, - "learning_rate": 3.96774107400133e-08, - "loss": 1.1289, - "step": 3606 - }, - { - "epoch": 2.92, - "learning_rate": 3.890373832942995e-08, - "loss": 1.1494, - "step": 3607 - }, - { - "epoch": 2.92, - "learning_rate": 3.813766861245682e-08, - "loss": 1.1713, - "step": 3608 - }, - { - "epoch": 2.92, - "learning_rate": 3.737920217378843e-08, - "loss": 1.2268, - "step": 3609 - }, - { - "epoch": 2.92, - "learning_rate": 3.6628339592313935e-08, - "loss": 1.1603, - "step": 3610 - }, - { - "epoch": 2.92, - "learning_rate": 3.588508144112157e-08, - "loss": 1.3584, - "step": 3611 - }, - { - "epoch": 2.92, - "learning_rate": 3.514942828749535e-08, - "loss": 1.1048, - "step": 3612 - }, - { - "epoch": 2.92, - "learning_rate": 3.4421380692912784e-08, - "loss": 1.2641, - "step": 3613 - }, - { - "epoch": 2.92, - "learning_rate": 3.37009392130494e-08, - "loss": 1.2067, - "step": 3614 - }, - { - "epoch": 2.92, - "learning_rate": 3.2988104397773115e-08, - "loss": 1.1812, - "step": 3615 - }, - { - "epoch": 2.93, - "learning_rate": 3.22828767911465e-08, - "loss": 1.1069, - "step": 3616 - }, - { - "epoch": 2.93, - "learning_rate": 3.158525693142678e-08, - "loss": 1.1134, - "step": 3617 - }, - { - "epoch": 2.93, - "learning_rate": 3.0895245351065803e-08, - "loss": 1.1609, - "step": 3618 - }, - { - "epoch": 2.93, - "learning_rate": 3.021284257670565e-08, - "loss": 1.1479, - "step": 3619 - }, - { - "epoch": 2.93, - "learning_rate": 2.95380491291819e-08, - "loss": 1.0582, - "step": 3620 - }, - { - "epoch": 2.93, - "learning_rate": 2.8870865523525916e-08, - "loss": 1.3158, - "step": 3621 - }, - { - "epoch": 2.93, - "learning_rate": 2.8211292268954806e-08, - "loss": 1.3221, - "step": 3622 - }, - { - "epoch": 2.93, - "learning_rate": 2.7559329868881435e-08, - "loss": 1.2574, - "step": 3623 - }, - { - "epoch": 2.93, - "learning_rate": 2.6914978820908876e-08, - "loss": 1.4373, - "step": 3624 - }, - { - "epoch": 2.93, - "learning_rate": 2.6278239616831513e-08, - "loss": 1.2538, - "step": 3625 - }, - { - "epoch": 2.93, - "learning_rate": 2.564911274263171e-08, - "loss": 1.166, - "step": 3626 - }, - { - "epoch": 2.93, - "learning_rate": 2.5027598678484255e-08, - "loss": 1.1635, - "step": 3627 - }, - { - "epoch": 2.94, - "learning_rate": 2.441369789875303e-08, - "loss": 1.1832, - "step": 3628 - }, - { - "epoch": 2.94, - "learning_rate": 2.380741087199101e-08, - "loss": 1.1743, - "step": 3629 - }, - { - "epoch": 2.94, - "learning_rate": 2.320873806093804e-08, - "loss": 1.2072, - "step": 3630 - }, - { - "epoch": 2.94, - "learning_rate": 2.261767992252639e-08, - "loss": 1.3097, - "step": 3631 - }, - { - "epoch": 2.94, - "learning_rate": 2.2034236907874096e-08, - "loss": 1.132, - "step": 3632 - }, - { - "epoch": 2.94, - "learning_rate": 2.1458409462286056e-08, - "loss": 1.2615, - "step": 3633 - }, - { - "epoch": 2.94, - "learning_rate": 2.089019802525849e-08, - "loss": 1.15, - "step": 3634 - }, - { - "epoch": 2.94, - "learning_rate": 2.0329603030468936e-08, - "loss": 1.2048, - "step": 3635 - }, - { - "epoch": 2.94, - "learning_rate": 1.9776624905788468e-08, - "loss": 1.3012, - "step": 3636 - }, - { - "epoch": 2.94, - "learning_rate": 1.9231264073270583e-08, - "loss": 1.1857, - "step": 3637 - }, - { - "epoch": 2.94, - "learning_rate": 1.8693520949153444e-08, - "loss": 1.1619, - "step": 3638 - }, - { - "epoch": 2.94, - "learning_rate": 1.8163395943866514e-08, - "loss": 1.2497, - "step": 3639 - }, - { - "epoch": 2.94, - "learning_rate": 1.764088946201947e-08, - "loss": 1.1742, - "step": 3640 - }, - { - "epoch": 2.95, - "learning_rate": 1.7126001902409984e-08, - "loss": 1.2342, - "step": 3641 - }, - { - "epoch": 2.95, - "learning_rate": 1.6618733658021468e-08, - "loss": 1.2585, - "step": 3642 - }, - { - "epoch": 2.95, - "learning_rate": 1.6119085116018673e-08, - "loss": 1.1732, - "step": 3643 - }, - { - "epoch": 2.95, - "learning_rate": 1.5627056657753214e-08, - "loss": 1.0005, - "step": 3644 - }, - { - "epoch": 2.95, - "learning_rate": 1.514264865876136e-08, - "loss": 1.1433, - "step": 3645 - }, - { - "epoch": 2.95, - "learning_rate": 1.4665861488761813e-08, - "loss": 1.1087, - "step": 3646 - }, - { - "epoch": 2.95, - "learning_rate": 1.41966955116557e-08, - "loss": 1.1723, - "step": 3647 - }, - { - "epoch": 2.95, - "learning_rate": 1.3735151085528808e-08, - "loss": 1.24, - "step": 3648 - }, - { - "epoch": 2.95, - "learning_rate": 1.3281228562650461e-08, - "loss": 1.3475, - "step": 3649 - }, - { - "epoch": 2.95, - "learning_rate": 1.2834928289472415e-08, - "loss": 1.0585, - "step": 3650 - }, - { - "epoch": 2.95, - "learning_rate": 1.2396250606626636e-08, - "loss": 1.2242, - "step": 3651 - }, - { - "epoch": 2.95, - "learning_rate": 1.1965195848929744e-08, - "loss": 1.1566, - "step": 3652 - }, - { - "epoch": 2.96, - "learning_rate": 1.154176434537968e-08, - "loss": 1.1605, - "step": 3653 - }, - { - "epoch": 2.96, - "learning_rate": 1.1125956419155704e-08, - "loss": 1.1876, - "step": 3654 - }, - { - "epoch": 2.96, - "learning_rate": 1.0717772387617287e-08, - "loss": 1.0997, - "step": 3655 - }, - { - "epoch": 2.96, - "learning_rate": 1.0317212562307443e-08, - "loss": 1.1562, - "step": 3656 - }, - { - "epoch": 2.96, - "learning_rate": 9.9242772489494e-09, - "loss": 1.0848, - "step": 3657 - }, - { - "epoch": 2.96, - "learning_rate": 9.538966747444366e-09, - "loss": 1.2348, - "step": 3658 - }, - { - "epoch": 2.96, - "learning_rate": 9.161281351879326e-09, - "loss": 1.0837, - "step": 3659 - }, - { - "epoch": 2.96, - "learning_rate": 8.79122135051591e-09, - "loss": 1.1939, - "step": 3660 - }, - { - "epoch": 2.96, - "learning_rate": 8.4287870257993e-09, - "loss": 1.2582, - "step": 3661 - }, - { - "epoch": 2.96, - "learning_rate": 8.073978654351555e-09, - "loss": 1.1915, - "step": 3662 - }, - { - "epoch": 2.96, - "learning_rate": 7.726796506979384e-09, - "loss": 1.054, - "step": 3663 - }, - { - "epoch": 2.96, - "learning_rate": 7.387240848663046e-09, - "loss": 1.0872, - "step": 3664 - }, - { - "epoch": 2.97, - "learning_rate": 7.055311938566345e-09, - "loss": 1.2847, - "step": 3665 - }, - { - "epoch": 2.97, - "learning_rate": 6.731010030028851e-09, - "loss": 1.3215, - "step": 3666 - }, - { - "epoch": 2.97, - "learning_rate": 6.414335370570346e-09, - "loss": 1.1195, - "step": 3667 - }, - { - "epoch": 2.97, - "learning_rate": 6.105288201889714e-09, - "loss": 1.1477, - "step": 3668 - }, - { - "epoch": 2.97, - "learning_rate": 5.803868759863829e-09, - "loss": 1.1077, - "step": 3669 - }, - { - "epoch": 2.97, - "learning_rate": 5.510077274547554e-09, - "loss": 1.0697, - "step": 3670 - }, - { - "epoch": 2.97, - "learning_rate": 5.223913970172634e-09, - "loss": 1.3245, - "step": 3671 - }, - { - "epoch": 2.97, - "learning_rate": 4.945379065152134e-09, - "loss": 1.2907, - "step": 3672 - }, - { - "epoch": 2.97, - "learning_rate": 4.674472772073779e-09, - "loss": 1.061, - "step": 3673 - }, - { - "epoch": 2.97, - "learning_rate": 4.411195297702176e-09, - "loss": 1.1527, - "step": 3674 - }, - { - "epoch": 2.97, - "learning_rate": 4.155546842983249e-09, - "loss": 1.114, - "step": 3675 - }, - { - "epoch": 2.97, - "learning_rate": 3.907527603036476e-09, - "loss": 1.1769, - "step": 3676 - }, - { - "epoch": 2.97, - "learning_rate": 3.6671377671604337e-09, - "loss": 1.1011, - "step": 3677 - }, - { - "epoch": 2.98, - "learning_rate": 3.4343775188294683e-09, - "loss": 1.1608, - "step": 3678 - }, - { - "epoch": 2.98, - "learning_rate": 3.209247035694807e-09, - "loss": 1.3841, - "step": 3679 - }, - { - "epoch": 2.98, - "learning_rate": 2.9917464895856673e-09, - "loss": 1.1951, - "step": 3680 - }, - { - "epoch": 2.98, - "learning_rate": 2.781876046505927e-09, - "loss": 1.249, - "step": 3681 - }, - { - "epoch": 2.98, - "learning_rate": 2.5796358666374533e-09, - "loss": 1.0905, - "step": 3682 - }, - { - "epoch": 2.98, - "learning_rate": 2.385026104336774e-09, - "loss": 1.3697, - "step": 3683 - }, - { - "epoch": 2.98, - "learning_rate": 2.1980469081395173e-09, - "loss": 1.1802, - "step": 3684 - }, - { - "epoch": 2.98, - "learning_rate": 2.0186984207537506e-09, - "loss": 1.1041, - "step": 3685 - }, - { - "epoch": 2.98, - "learning_rate": 1.8469807790655325e-09, - "loss": 1.0854, - "step": 3686 - }, - { - "epoch": 2.98, - "learning_rate": 1.6828941141366906e-09, - "loss": 1.1079, - "step": 3687 - }, - { - "epoch": 2.98, - "learning_rate": 1.526438551204823e-09, - "loss": 1.1294, - "step": 3688 - }, - { - "epoch": 2.98, - "learning_rate": 1.3776142096821876e-09, - "loss": 1.187, - "step": 3689 - }, - { - "epoch": 2.99, - "learning_rate": 1.2364212031579226e-09, - "loss": 1.1456, - "step": 3690 - }, - { - "epoch": 2.99, - "learning_rate": 1.1028596393947155e-09, - "loss": 1.1511, - "step": 3691 - }, - { - "epoch": 2.99, - "learning_rate": 9.769296203332446e-10, - "loss": 1.1856, - "step": 3692 - }, - { - "epoch": 2.99, - "learning_rate": 8.586312420877374e-10, - "loss": 1.1594, - "step": 3693 - }, - { - "epoch": 2.99, - "learning_rate": 7.479645949481918e-10, - "loss": 1.1531, - "step": 3694 - }, - { - "epoch": 2.99, - "learning_rate": 6.449297633792651e-10, - "loss": 1.0876, - "step": 3695 - }, - { - "epoch": 2.99, - "learning_rate": 5.495268260213848e-10, - "loss": 0.843, - "step": 3696 - }, - { - "epoch": 2.99, - "learning_rate": 4.617558556896384e-10, - "loss": 1.1532, - "step": 3697 - }, - { - "epoch": 2.99, - "learning_rate": 3.816169193748831e-10, - "loss": 1.2176, - "step": 3698 - }, - { - "epoch": 2.99, - "learning_rate": 3.091100782415257e-10, - "loss": 1.1557, - "step": 3699 - }, - { - "epoch": 2.99, - "learning_rate": 2.442353876297432e-10, - "loss": 1.1749, - "step": 3700 - }, - { - "epoch": 2.99, - "learning_rate": 1.869928970543722e-10, - "loss": 1.1999, - "step": 3701 - }, - { - "epoch": 3.0, - "learning_rate": 1.3738265020490914e-10, - "loss": 1.2779, - "step": 3702 - }, - { - "epoch": 3.0, - "learning_rate": 9.540468494662059e-11, - "loss": 1.2068, - "step": 3703 - }, - { - "epoch": 3.0, - "learning_rate": 6.10590333172123e-11, - "loss": 1.2245, - "step": 3704 - }, - { - "epoch": 3.0, - "learning_rate": 3.4345721532380585e-11, - "loss": 1.1575, - "step": 3705 - }, - { - "epoch": 3.0, - "learning_rate": 1.5264769980261052e-11, - "loss": 1.3206, - "step": 3706 - }, - { - "epoch": 3.0, - "learning_rate": 3.816193223649123e-12, - "loss": 1.2228, - "step": 3707 - }, - { - "epoch": 3.0, - "learning_rate": 0.0, - "loss": 1.1879, - "step": 3708 - }, - { - "epoch": 3.0, - "step": 3708, - "total_flos": 3.2705004138450125e+17, - "train_loss": 1.339754913472435, - "train_runtime": 4083.1615, - "train_samples_per_second": 43.577, - "train_steps_per_second": 0.908 - } - ], - "max_steps": 3708, - "num_train_epochs": 3, - "total_flos": 3.2705004138450125e+17, - "trial_name": null, - "trial_params": null -}