| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.9987492182614135, | |
| "eval_steps": 500, | |
| "global_step": 3196, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.1666666666666667e-07, | |
| "loss": 3.0801, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 3.0098, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.25e-06, | |
| "loss": 3.0501, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 2.9675, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.0833333333333334e-06, | |
| "loss": 3.018, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.5e-06, | |
| "loss": 2.9638, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.916666666666667e-06, | |
| "loss": 2.9484, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 2.954, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.7500000000000005e-06, | |
| "loss": 3.0107, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 2.9987, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.583333333333333e-06, | |
| "loss": 2.9842, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5e-06, | |
| "loss": 2.9954, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.416666666666667e-06, | |
| "loss": 2.9648, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 2.9187, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 6.25e-06, | |
| "loss": 2.9279, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 2.9908, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.083333333333335e-06, | |
| "loss": 2.9122, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 2.9751, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 7.916666666666667e-06, | |
| "loss": 3.0286, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 2.8958, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 8.750000000000001e-06, | |
| "loss": 2.8504, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 2.8388, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.583333333333335e-06, | |
| "loss": 2.91, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1e-05, | |
| "loss": 2.8353, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.0416666666666668e-05, | |
| "loss": 2.8809, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.0833333333333334e-05, | |
| "loss": 2.8605, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.125e-05, | |
| "loss": 2.8641, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.1666666666666668e-05, | |
| "loss": 2.9369, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.2083333333333333e-05, | |
| "loss": 2.9278, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.25e-05, | |
| "loss": 2.8473, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.2916666666666668e-05, | |
| "loss": 2.8363, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 2.808, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.375e-05, | |
| "loss": 2.8614, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.416666666666667e-05, | |
| "loss": 2.8742, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.4583333333333333e-05, | |
| "loss": 2.7467, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.5000000000000002e-05, | |
| "loss": 2.897, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.5416666666666668e-05, | |
| "loss": 2.9228, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.5833333333333333e-05, | |
| "loss": 2.8754, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.6250000000000002e-05, | |
| "loss": 2.88, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "loss": 2.7796, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.7083333333333333e-05, | |
| "loss": 2.7483, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.7500000000000002e-05, | |
| "loss": 2.8297, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.7916666666666667e-05, | |
| "loss": 2.8416, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.8333333333333333e-05, | |
| "loss": 2.8327, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.8750000000000002e-05, | |
| "loss": 2.8062, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.916666666666667e-05, | |
| "loss": 2.7981, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9583333333333333e-05, | |
| "loss": 2.8298, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2e-05, | |
| "loss": 2.8166, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9999979459727323e-05, | |
| "loss": 2.7742, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9999917838993664e-05, | |
| "loss": 2.6957, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9999815138052173e-05, | |
| "loss": 2.8767, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9999671357324743e-05, | |
| "loss": 2.8015, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.999948649740204e-05, | |
| "loss": 2.8407, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.999926055904347e-05, | |
| "loss": 2.78, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.999899354317721e-05, | |
| "loss": 2.7056, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.999868545090017e-05, | |
| "loss": 2.7898, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9998336283478007e-05, | |
| "loss": 2.8308, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9997946042345128e-05, | |
| "loss": 2.6869, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9997514729104664e-05, | |
| "loss": 2.7477, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9997042345528466e-05, | |
| "loss": 2.8106, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9996528893557113e-05, | |
| "loss": 2.6661, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.99959743752999e-05, | |
| "loss": 2.7769, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9995378793034815e-05, | |
| "loss": 2.697, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.999474214920854e-05, | |
| "loss": 2.7514, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9994064446436444e-05, | |
| "loss": 2.7194, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.9993345687502566e-05, | |
| "loss": 2.6859, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.999258587535961e-05, | |
| "loss": 2.73, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.9991785013128922e-05, | |
| "loss": 2.6976, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.9990943104100495e-05, | |
| "loss": 2.7371, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.999006015173293e-05, | |
| "loss": 2.6993, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.9989136159653442e-05, | |
| "loss": 2.6604, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.9988171131657848e-05, | |
| "loss": 2.6928, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.998716507171053e-05, | |
| "loss": 2.6638, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.998611798394444e-05, | |
| "loss": 2.6605, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.998502987266107e-05, | |
| "loss": 2.7528, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.998390074233044e-05, | |
| "loss": 2.7199, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.998273059759108e-05, | |
| "loss": 2.6511, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.998151944325001e-05, | |
| "loss": 2.7285, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.9980267284282718e-05, | |
| "loss": 2.6984, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.9978974125833136e-05, | |
| "loss": 2.6472, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.997763997321364e-05, | |
| "loss": 2.6311, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.9976264831904988e-05, | |
| "loss": 2.6419, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.9974848707556343e-05, | |
| "loss": 2.7605, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.9973391605985226e-05, | |
| "loss": 2.6821, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.997189353317748e-05, | |
| "loss": 2.6946, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.9970354495287275e-05, | |
| "loss": 2.6943, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.996877449863706e-05, | |
| "loss": 2.7136, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.996715354971755e-05, | |
| "loss": 2.6669, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.9965491655187695e-05, | |
| "loss": 2.6809, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.996378882187464e-05, | |
| "loss": 2.625, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.9962045056773724e-05, | |
| "loss": 2.6653, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.9960260367048426e-05, | |
| "loss": 2.5589, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.9958434760030346e-05, | |
| "loss": 2.7113, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.9956568243219182e-05, | |
| "loss": 2.533, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.9954660824282687e-05, | |
| "loss": 2.67, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.9952712511056637e-05, | |
| "loss": 2.6857, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.995072331154481e-05, | |
| "loss": 2.654, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.994869323391895e-05, | |
| "loss": 2.5482, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.994662228651873e-05, | |
| "loss": 2.6769, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.9944510477851705e-05, | |
| "loss": 2.641, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.9942357816593304e-05, | |
| "loss": 2.6606, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.994016431158678e-05, | |
| "loss": 2.662, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.9937929971843168e-05, | |
| "loss": 2.7096, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.993565480654126e-05, | |
| "loss": 2.5603, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.993333882502755e-05, | |
| "loss": 2.5755, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.993098203681623e-05, | |
| "loss": 2.5515, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.992858445158911e-05, | |
| "loss": 2.6096, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.9926146079195597e-05, | |
| "loss": 2.5125, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.992366692965266e-05, | |
| "loss": 2.6472, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.9921147013144782e-05, | |
| "loss": 2.6315, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.9918586340023912e-05, | |
| "loss": 2.5912, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.9915984920809442e-05, | |
| "loss": 2.5412, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.991334276618814e-05, | |
| "loss": 2.5814, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.9910659887014122e-05, | |
| "loss": 2.6188, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.9907936294308802e-05, | |
| "loss": 2.5956, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.9905171999260846e-05, | |
| "loss": 2.5575, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.9902367013226132e-05, | |
| "loss": 2.5507, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.989952134772769e-05, | |
| "loss": 2.6304, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.9896635014455682e-05, | |
| "loss": 2.5472, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.9893708025267305e-05, | |
| "loss": 2.6461, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.98907403921868e-05, | |
| "loss": 2.4726, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.9887732127405367e-05, | |
| "loss": 2.5246, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.9884683243281117e-05, | |
| "loss": 2.5386, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.988159375233903e-05, | |
| "loss": 2.5912, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.9878463667270914e-05, | |
| "loss": 2.5393, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.987529300093532e-05, | |
| "loss": 2.5854, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.987208176635752e-05, | |
| "loss": 2.5409, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.9868829976729444e-05, | |
| "loss": 2.5822, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.9865537645409615e-05, | |
| "loss": 2.6256, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.9862204785923117e-05, | |
| "loss": 2.5346, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.9858831411961513e-05, | |
| "loss": 2.551, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.9855417537382807e-05, | |
| "loss": 2.5035, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.985196317621139e-05, | |
| "loss": 2.5503, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.9848468342637957e-05, | |
| "loss": 2.5373, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.9844933051019475e-05, | |
| "loss": 2.624, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.984135731587912e-05, | |
| "loss": 2.5066, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.9837741151906205e-05, | |
| "loss": 2.5372, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.983408457395613e-05, | |
| "loss": 2.5623, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.9830387597050313e-05, | |
| "loss": 2.566, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.9826650236376136e-05, | |
| "loss": 2.5249, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.982287250728689e-05, | |
| "loss": 2.513, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.981905442530168e-05, | |
| "loss": 2.6131, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.9815196006105402e-05, | |
| "loss": 2.5502, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.9811297265548654e-05, | |
| "loss": 2.5614, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.980735821964767e-05, | |
| "loss": 2.565, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.9803378884584266e-05, | |
| "loss": 2.5482, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.979935927670577e-05, | |
| "loss": 2.5238, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.9795299412524948e-05, | |
| "loss": 2.5186, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.9791199308719943e-05, | |
| "loss": 2.5836, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.978705898213421e-05, | |
| "loss": 2.5315, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.978287844977643e-05, | |
| "loss": 2.4924, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.977865772882046e-05, | |
| "loss": 2.4766, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.9774396836605257e-05, | |
| "loss": 2.5614, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.9770095790634792e-05, | |
| "loss": 2.4726, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.9765754608578e-05, | |
| "loss": 2.4818, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.976137330826869e-05, | |
| "loss": 2.5764, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.9756951907705487e-05, | |
| "loss": 2.4565, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.975249042505174e-05, | |
| "loss": 2.5303, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.974798887863547e-05, | |
| "loss": 2.5989, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.9743447286949274e-05, | |
| "loss": 2.4947, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.973886566865025e-05, | |
| "loss": 2.4949, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.9734244042559948e-05, | |
| "loss": 2.4753, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.972958242766425e-05, | |
| "loss": 2.4441, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.972488084311333e-05, | |
| "loss": 2.4687, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.9720139308221552e-05, | |
| "loss": 2.559, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.97153578424674e-05, | |
| "loss": 2.4826, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.9710536465493396e-05, | |
| "loss": 2.5199, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.970567519710602e-05, | |
| "loss": 2.4394, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.9700774057275624e-05, | |
| "loss": 2.4073, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.9695833066136365e-05, | |
| "loss": 2.4692, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.96908522439861e-05, | |
| "loss": 2.4281, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.9685831611286312e-05, | |
| "loss": 2.5217, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.9680771188662044e-05, | |
| "loss": 2.5191, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.9675670996901783e-05, | |
| "loss": 2.5546, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.9670531056957398e-05, | |
| "loss": 2.4683, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.966535138994404e-05, | |
| "loss": 2.4763, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.9660132017140063e-05, | |
| "loss": 2.3512, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.9654872959986936e-05, | |
| "loss": 2.5075, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.9649574240089152e-05, | |
| "loss": 2.5742, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.9644235879214144e-05, | |
| "loss": 2.4722, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.9638857899292188e-05, | |
| "loss": 2.4669, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.963344032241632e-05, | |
| "loss": 2.5171, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.9627983170842235e-05, | |
| "loss": 2.4252, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.9622486466988216e-05, | |
| "loss": 2.5223, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.961695023343502e-05, | |
| "loss": 2.4036, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.96113744929258e-05, | |
| "loss": 2.5352, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.9605759268365997e-05, | |
| "loss": 2.4282, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.960010458282326e-05, | |
| "loss": 2.4154, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.9594410459527348e-05, | |
| "loss": 2.4524, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.9588676921870028e-05, | |
| "loss": 2.5328, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.9582903993404984e-05, | |
| "loss": 2.4879, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.957709169784773e-05, | |
| "loss": 2.5011, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.9571240059075482e-05, | |
| "loss": 2.5277, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.9565349101127095e-05, | |
| "loss": 2.4501, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.9559418848202947e-05, | |
| "loss": 2.4396, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.9553449324664844e-05, | |
| "loss": 2.4757, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.9547440555035907e-05, | |
| "loss": 2.4308, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.954139256400049e-05, | |
| "loss": 2.3535, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.9535305376404074e-05, | |
| "loss": 2.4537, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.952917901725316e-05, | |
| "loss": 2.409, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.9523013511715153e-05, | |
| "loss": 2.409, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.9516808885118297e-05, | |
| "loss": 2.4207, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.9510565162951538e-05, | |
| "loss": 2.3889, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.950428237086442e-05, | |
| "loss": 2.4098, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.9497960534667e-05, | |
| "loss": 2.5492, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.949159968032972e-05, | |
| "loss": 2.4549, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.9485199833983326e-05, | |
| "loss": 2.4803, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.947876102191873e-05, | |
| "loss": 2.3983, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.947228327058692e-05, | |
| "loss": 2.4165, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.946576660659886e-05, | |
| "loss": 2.4916, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.9459211056725354e-05, | |
| "loss": 2.4041, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.9452616647896963e-05, | |
| "loss": 2.4541, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.9445983407203873e-05, | |
| "loss": 2.4392, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.94393113618958e-05, | |
| "loss": 2.3582, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.9432600539381878e-05, | |
| "loss": 2.4532, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.942585096723052e-05, | |
| "loss": 2.3451, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.941906267316935e-05, | |
| "loss": 2.4774, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.9412235685085034e-05, | |
| "loss": 2.4031, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.9405370031023222e-05, | |
| "loss": 2.3877, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.9398465739188392e-05, | |
| "loss": 2.4408, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.939152283794375e-05, | |
| "loss": 2.408, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.938454135581112e-05, | |
| "loss": 2.388, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.9377521321470806e-05, | |
| "loss": 2.4137, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.9370462763761494e-05, | |
| "loss": 2.4379, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.9363365711680114e-05, | |
| "loss": 2.4674, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.935623019438176e-05, | |
| "loss": 2.4766, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.9349056241179513e-05, | |
| "loss": 2.447, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.9341843881544372e-05, | |
| "loss": 2.4942, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.93345931451051e-05, | |
| "loss": 2.4053, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.9327304061648116e-05, | |
| "loss": 2.4888, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.9319976661117373e-05, | |
| "loss": 2.3857, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.9312610973614238e-05, | |
| "loss": 2.4105, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.9305207029397348e-05, | |
| "loss": 2.3888, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.9297764858882516e-05, | |
| "loss": 2.4371, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.9290284492642582e-05, | |
| "loss": 2.5264, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.9282765961407297e-05, | |
| "loss": 2.4291, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.9275209296063205e-05, | |
| "loss": 2.4449, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.926761452765349e-05, | |
| "loss": 2.364, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.9259981687377872e-05, | |
| "loss": 2.3665, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.925231080659249e-05, | |
| "loss": 2.4422, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.9244601916809727e-05, | |
| "loss": 2.3823, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9236855049698132e-05, | |
| "loss": 2.3205, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9229070237082252e-05, | |
| "loss": 2.4603, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.922124751094252e-05, | |
| "loss": 2.3355, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.921338690341513e-05, | |
| "loss": 2.3201, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.9205488446791873e-05, | |
| "loss": 2.3917, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.9197552173520052e-05, | |
| "loss": 2.3986, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.918957811620231e-05, | |
| "loss": 2.4772, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.91815663075965e-05, | |
| "loss": 2.3752, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.9173516780615575e-05, | |
| "loss": 2.3506, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.916542956832743e-05, | |
| "loss": 2.413, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.9157304703954772e-05, | |
| "loss": 2.3956, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.9149142220874992e-05, | |
| "loss": 2.492, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.9140942152620012e-05, | |
| "loss": 2.3548, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.913270453287616e-05, | |
| "loss": 2.43, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.912442939548403e-05, | |
| "loss": 2.3616, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.911611677443833e-05, | |
| "loss": 2.3568, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.9107766703887764e-05, | |
| "loss": 2.3192, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.9099379218134885e-05, | |
| "loss": 2.3635, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.909095435163593e-05, | |
| "loss": 2.3461, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.9082492139000723e-05, | |
| "loss": 2.4048, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.907399261499248e-05, | |
| "loss": 2.3739, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.9065455814527722e-05, | |
| "loss": 2.3548, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.9056881772676084e-05, | |
| "loss": 2.4062, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.9048270524660197e-05, | |
| "loss": 2.4014, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.903962210585554e-05, | |
| "loss": 2.4357, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.9030936551790292e-05, | |
| "loss": 2.2851, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.9022213898145176e-05, | |
| "loss": 2.3786, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.9013454180753335e-05, | |
| "loss": 2.4144, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.9004657435600165e-05, | |
| "loss": 2.425, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.899582369882317e-05, | |
| "loss": 2.3593, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.8986953006711822e-05, | |
| "loss": 2.4237, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.897804539570742e-05, | |
| "loss": 2.4187, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.8969100902402903e-05, | |
| "loss": 2.3659, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.896011956354274e-05, | |
| "loss": 2.2996, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.895110141602276e-05, | |
| "loss": 2.4063, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.8942046496890013e-05, | |
| "loss": 2.3795, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.893295484334259e-05, | |
| "loss": 2.3961, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.892382649272951e-05, | |
| "loss": 2.351, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.8914661482550523e-05, | |
| "loss": 2.3017, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.8905459850456002e-05, | |
| "loss": 2.3543, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.889622163424675e-05, | |
| "loss": 2.3545, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.888694687187386e-05, | |
| "loss": 2.3739, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.887763560143856e-05, | |
| "loss": 2.4653, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.886828786119206e-05, | |
| "loss": 2.346, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.885890368953539e-05, | |
| "loss": 2.3577, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.8849483125019234e-05, | |
| "loss": 2.398, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.8840026206343786e-05, | |
| "loss": 2.403, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.8830532972358586e-05, | |
| "loss": 2.4021, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.8821003462062354e-05, | |
| "loss": 2.3994, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.8811437714602838e-05, | |
| "loss": 2.3292, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.880183576927665e-05, | |
| "loss": 2.3766, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.879219766552911e-05, | |
| "loss": 2.4219, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.878252344295407e-05, | |
| "loss": 2.4268, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.8772813141293762e-05, | |
| "loss": 2.3674, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.8763066800438638e-05, | |
| "loss": 2.3426, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.8753284460427193e-05, | |
| "loss": 2.434, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.8743466161445823e-05, | |
| "loss": 2.3729, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.8733611943828626e-05, | |
| "loss": 2.3486, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.872372184805727e-05, | |
| "loss": 2.3443, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.8713795914760807e-05, | |
| "loss": 2.3642, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.870383418471551e-05, | |
| "loss": 2.3509, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.869383669884471e-05, | |
| "loss": 2.3292, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.868380349821863e-05, | |
| "loss": 2.357, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.86737346240542e-05, | |
| "loss": 2.3791, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.8663630117714904e-05, | |
| "loss": 2.4309, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.8653490020710606e-05, | |
| "loss": 2.3953, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.8643314374697377e-05, | |
| "loss": 2.3189, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.8633103221477328e-05, | |
| "loss": 2.3911, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.862285660299843e-05, | |
| "loss": 2.3737, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.8612574561354353e-05, | |
| "loss": 2.3227, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.8602257138784284e-05, | |
| "loss": 2.3074, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.859190437767276e-05, | |
| "loss": 2.381, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.858151632054948e-05, | |
| "loss": 2.3292, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.8571093010089155e-05, | |
| "loss": 2.373, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.8560634489111318e-05, | |
| "loss": 2.3828, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.8550140800580136e-05, | |
| "loss": 2.4072, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.853961198760426e-05, | |
| "loss": 2.4374, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.8529048093436617e-05, | |
| "loss": 2.3103, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.851844916147427e-05, | |
| "loss": 2.3273, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.850781523525821e-05, | |
| "loss": 2.292, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.849714635847318e-05, | |
| "loss": 2.4149, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.848644257494751e-05, | |
| "loss": 2.3553, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.8475703928652932e-05, | |
| "loss": 2.3646, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.8464930463704377e-05, | |
| "loss": 2.3025, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.845412222435984e-05, | |
| "loss": 2.3031, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.8443279255020153e-05, | |
| "loss": 2.2799, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.8432401600228823e-05, | |
| "loss": 2.3426, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.8421489304671852e-05, | |
| "loss": 2.3594, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.8410542413177546e-05, | |
| "loss": 2.4183, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.839956097071633e-05, | |
| "loss": 2.3598, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.8388545022400567e-05, | |
| "loss": 2.3948, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.837749461348438e-05, | |
| "loss": 2.3453, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.8366409789363443e-05, | |
| "loss": 2.4202, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.8355290595574825e-05, | |
| "loss": 2.3143, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.834413707779677e-05, | |
| "loss": 2.3925, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.8332949281848552e-05, | |
| "loss": 2.2551, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.832172725369024e-05, | |
| "loss": 2.3615, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.8310471039422537e-05, | |
| "loss": 2.2681, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.8299180685286586e-05, | |
| "loss": 2.2561, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.8287856237663775e-05, | |
| "loss": 2.3803, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.8276497743075556e-05, | |
| "loss": 2.2957, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.826510524818324e-05, | |
| "loss": 2.3533, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.8253678799787825e-05, | |
| "loss": 2.3232, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.8242218444829776e-05, | |
| "loss": 2.3033, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.823072423038886e-05, | |
| "loss": 2.2673, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.8219196203683935e-05, | |
| "loss": 2.3313, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.8207634412072765e-05, | |
| "loss": 2.26, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.819603890305182e-05, | |
| "loss": 2.2909, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.8184409724256084e-05, | |
| "loss": 2.3093, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.8172746923458855e-05, | |
| "loss": 2.3399, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.8161050548571562e-05, | |
| "loss": 2.3171, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.8149320647643544e-05, | |
| "loss": 2.2684, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.813755726886187e-05, | |
| "loss": 2.3293, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.8125760460551152e-05, | |
| "loss": 2.3098, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.8113930271173314e-05, | |
| "loss": 2.3424, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.8102066749327427e-05, | |
| "loss": 2.2803, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.8090169943749477e-05, | |
| "loss": 2.3599, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.8078239903312194e-05, | |
| "loss": 2.3201, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.8066276677024833e-05, | |
| "loss": 2.385, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.8054280314032982e-05, | |
| "loss": 2.4307, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.804225086361836e-05, | |
| "loss": 2.1745, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.8030188375198593e-05, | |
| "loss": 2.3909, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.801809289832705e-05, | |
| "loss": 2.3809, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.8005964482692606e-05, | |
| "loss": 2.3488, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.7993803178119457e-05, | |
| "loss": 2.271, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.79816090345669e-05, | |
| "loss": 2.3261, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.7969382102129153e-05, | |
| "loss": 2.317, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.795712243103511e-05, | |
| "loss": 2.2722, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.794483007164817e-05, | |
| "loss": 2.2294, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.793250507446602e-05, | |
| "loss": 2.4161, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.7920147490120412e-05, | |
| "loss": 2.3576, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.7907757369376984e-05, | |
| "loss": 2.3179, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.7895334763135036e-05, | |
| "loss": 2.306, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.7882879722427293e-05, | |
| "loss": 2.3283, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.7870392298419755e-05, | |
| "loss": 2.3474, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.7857872542411433e-05, | |
| "loss": 2.2914, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.7845320505834176e-05, | |
| "loss": 2.3451, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.7832736240252427e-05, | |
| "loss": 2.256, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.7820119797363038e-05, | |
| "loss": 2.3638, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.7807471228995045e-05, | |
| "loss": 2.2526, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.7794790587109456e-05, | |
| "loss": 2.2797, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.7782077923799043e-05, | |
| "loss": 2.3592, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.776933329128811e-05, | |
| "loss": 2.3266, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.7756556741932315e-05, | |
| "loss": 2.301, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.7743748328218412e-05, | |
| "loss": 2.2971, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.7730908102764065e-05, | |
| "loss": 2.29, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.771803611831762e-05, | |
| "loss": 2.4113, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.7705132427757895e-05, | |
| "loss": 2.2682, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.7692197084093944e-05, | |
| "loss": 2.3538, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.7679230140464877e-05, | |
| "loss": 2.3995, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.7666231650139602e-05, | |
| "loss": 2.2776, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.7653201666516614e-05, | |
| "loss": 2.3188, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.7640140243123813e-05, | |
| "loss": 2.381, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.762704743361823e-05, | |
| "loss": 2.3089, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.7613923291785842e-05, | |
| "loss": 2.3602, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.7600767871541343e-05, | |
| "loss": 2.2753, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.758758122692791e-05, | |
| "loss": 2.2424, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.7574363412117004e-05, | |
| "loss": 2.2396, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.7567742832851094e-05, | |
| "loss": 2.3081, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.75544783645955e-05, | |
| "loss": 2.3722, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.7541182862130793e-05, | |
| "loss": 2.2803, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.752785638007562e-05, | |
| "loss": 2.3004, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.7514498973175907e-05, | |
| "loss": 2.2576, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.7501110696304598e-05, | |
| "loss": 2.3134, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.748769160446147e-05, | |
| "loss": 2.296, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.7474241752772883e-05, | |
| "loss": 2.2324, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.746076119649156e-05, | |
| "loss": 2.2157, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.7447249990996366e-05, | |
| "loss": 2.2719, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.7433708191792065e-05, | |
| "loss": 2.2607, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.742013585450911e-05, | |
| "loss": 2.2322, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.7406533034903402e-05, | |
| "loss": 2.1659, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.7392899788856063e-05, | |
| "loss": 2.2837, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 1.7379236172373214e-05, | |
| "loss": 2.1824, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 1.7365542241585736e-05, | |
| "loss": 2.2136, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 1.7351818052749043e-05, | |
| "loss": 2.2655, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 1.7338063662242852e-05, | |
| "loss": 2.2106, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 1.732427912657095e-05, | |
| "loss": 2.2667, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 1.731046450236096e-05, | |
| "loss": 2.2176, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 1.7296619846364115e-05, | |
| "loss": 2.2638, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 1.7282745215455007e-05, | |
| "loss": 2.2169, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1.7268840666631388e-05, | |
| "loss": 2.2157, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1.7254906257013895e-05, | |
| "loss": 2.2332, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1.724094204384585e-05, | |
| "loss": 2.2029, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1.7226948084492997e-05, | |
| "loss": 2.2371, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.7212924436443285e-05, | |
| "loss": 2.1916, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.7198871157306625e-05, | |
| "loss": 2.1962, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.718478830481465e-05, | |
| "loss": 2.186, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.71706759368205e-05, | |
| "loss": 2.2142, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1.7156534111298534e-05, | |
| "loss": 2.2926, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1.7142362886344156e-05, | |
| "loss": 2.2536, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1.7128162320173523e-05, | |
| "loss": 2.2318, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1.711393247112334e-05, | |
| "loss": 2.2487, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.70996733976506e-05, | |
| "loss": 2.2113, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.708538515833236e-05, | |
| "loss": 2.2863, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.7071067811865477e-05, | |
| "loss": 2.2024, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.70567214170664e-05, | |
| "loss": 2.2314, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1.7042346032870893e-05, | |
| "loss": 2.2278, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1.702794171833383e-05, | |
| "loss": 2.2587, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1.7013508532628907e-05, | |
| "loss": 2.2167, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1.6999046535048446e-05, | |
| "loss": 2.3176, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 1.6984555785003122e-05, | |
| "loss": 2.2578, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 1.697003634202172e-05, | |
| "loss": 2.2541, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 1.6955488265750913e-05, | |
| "loss": 2.199, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 1.6940911615954986e-05, | |
| "loss": 2.326, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 1.692630645251561e-05, | |
| "loss": 2.1777, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 1.6911672835431597e-05, | |
| "loss": 2.279, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 1.689701082481864e-05, | |
| "loss": 2.2798, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 1.6882320480909086e-05, | |
| "loss": 2.2679, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.6867601864051656e-05, | |
| "loss": 2.3151, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.685285503471124e-05, | |
| "loss": 2.1954, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.6838080053468622e-05, | |
| "loss": 2.2319, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.6823276981020217e-05, | |
| "loss": 2.193, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 1.680844587817786e-05, | |
| "loss": 2.2382, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 1.6793586805868535e-05, | |
| "loss": 2.2109, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 1.6778699825134116e-05, | |
| "loss": 2.1958, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 1.6763784997131133e-05, | |
| "loss": 2.1709, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.6748842383130515e-05, | |
| "loss": 2.2905, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.6733872044517334e-05, | |
| "loss": 2.2204, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.6718874042790552e-05, | |
| "loss": 2.1834, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.6703848439562787e-05, | |
| "loss": 2.206, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1.6688795296560033e-05, | |
| "loss": 2.3119, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1.6673714675621418e-05, | |
| "loss": 2.2893, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1.665860663869896e-05, | |
| "loss": 2.274, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1.66434712478573e-05, | |
| "loss": 2.2762, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 1.6628308565273444e-05, | |
| "loss": 2.2691, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 1.661311865323652e-05, | |
| "loss": 2.3091, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 1.659790157414752e-05, | |
| "loss": 2.2129, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 1.6582657390519027e-05, | |
| "loss": 2.2313, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.656738616497499e-05, | |
| "loss": 2.2926, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.6552087960250425e-05, | |
| "loss": 2.2267, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.6536762839191194e-05, | |
| "loss": 2.2603, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.6521410864753735e-05, | |
| "loss": 2.264, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.650603210000479e-05, | |
| "loss": 2.2252, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.649062660812117e-05, | |
| "loss": 2.1176, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.6475194452389467e-05, | |
| "loss": 2.1767, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.645973569620583e-05, | |
| "loss": 2.2042, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.644425040307566e-05, | |
| "loss": 2.2289, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.6428738636613393e-05, | |
| "loss": 2.1852, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.6413200460542212e-05, | |
| "loss": 2.2802, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.639763593869379e-05, | |
| "loss": 2.2578, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.638204513500803e-05, | |
| "loss": 2.2766, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.6366428113532803e-05, | |
| "loss": 2.2064, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.6350784938423695e-05, | |
| "loss": 2.2648, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.6335115673943706e-05, | |
| "loss": 2.2008, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.631942038446304e-05, | |
| "loss": 2.2401, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.6303699134458805e-05, | |
| "loss": 2.2384, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.628795198851474e-05, | |
| "loss": 2.2179, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.627217901132099e-05, | |
| "loss": 2.2748, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.6256380267673807e-05, | |
| "loss": 2.2211, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.6240555822475282e-05, | |
| "loss": 2.2916, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.622470574073311e-05, | |
| "loss": 2.2139, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.620883008756028e-05, | |
| "loss": 2.1887, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.619292892817485e-05, | |
| "loss": 2.2577, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.6177002327899644e-05, | |
| "loss": 2.2832, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.616105035216201e-05, | |
| "loss": 2.2688, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.614507306649353e-05, | |
| "loss": 2.2477, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.6129070536529767e-05, | |
| "loss": 2.2691, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.6113042828009984e-05, | |
| "loss": 2.334, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.6096990006776885e-05, | |
| "loss": 2.2594, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.6080912138776333e-05, | |
| "loss": 2.1936, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 1.606480929005709e-05, | |
| "loss": 2.3006, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 1.6048681526770528e-05, | |
| "loss": 2.2197, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 1.6032528915170388e-05, | |
| "loss": 2.1906, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 1.6016351521612478e-05, | |
| "loss": 2.2047, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.6000149412554407e-05, | |
| "loss": 2.2223, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.5983922654555326e-05, | |
| "loss": 2.2307, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.596767131427564e-05, | |
| "loss": 2.2492, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.5951395458476746e-05, | |
| "loss": 2.2136, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.5935095154020746e-05, | |
| "loss": 2.2776, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.5918770467870174e-05, | |
| "loss": 2.1811, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.5902421467087736e-05, | |
| "loss": 2.2822, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.588604821883602e-05, | |
| "loss": 2.277, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1.586965079037722e-05, | |
| "loss": 2.2251, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1.5853229249072862e-05, | |
| "loss": 2.1909, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1.5836783662383543e-05, | |
| "loss": 2.2319, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1.5820314097868623e-05, | |
| "loss": 2.182, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 1.5803820623185972e-05, | |
| "loss": 2.182, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 1.5787303306091688e-05, | |
| "loss": 2.1612, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 1.5770762214439805e-05, | |
| "loss": 2.1989, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 1.5754197416182033e-05, | |
| "loss": 2.2151, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.5737608979367464e-05, | |
| "loss": 2.2425, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.5720996972142305e-05, | |
| "loss": 2.2705, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.570436146274958e-05, | |
| "loss": 2.2325, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.568770251952888e-05, | |
| "loss": 2.194, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1.5671020210916043e-05, | |
| "loss": 2.2651, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1.5654314605442906e-05, | |
| "loss": 2.2567, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1.563758577173701e-05, | |
| "loss": 2.1368, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1.5620833778521306e-05, | |
| "loss": 2.2524, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1.560405869461391e-05, | |
| "loss": 2.1707, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1.558726058892777e-05, | |
| "loss": 2.1992, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1.557043953047043e-05, | |
| "loss": 2.2327, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1.5553595588343707e-05, | |
| "loss": 2.1169, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1.5536728831743433e-05, | |
| "loss": 2.1454, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1.5519839329959174e-05, | |
| "loss": 2.1905, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1.5502927152373913e-05, | |
| "loss": 2.2778, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1.548599236846381e-05, | |
| "loss": 2.2081, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.546903504779787e-05, | |
| "loss": 2.154, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.5452055260037696e-05, | |
| "loss": 2.2984, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.5435053074937183e-05, | |
| "loss": 2.1696, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.541802856234223e-05, | |
| "loss": 2.2205, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.540098179219047e-05, | |
| "loss": 2.1755, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.5383912834510957e-05, | |
| "loss": 2.1807, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.5366821759423903e-05, | |
| "loss": 2.2996, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.534970863714038e-05, | |
| "loss": 2.1929, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.5332573537962025e-05, | |
| "loss": 2.2536, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.531541653228076e-05, | |
| "loss": 2.1834, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.5298237690578494e-05, | |
| "loss": 2.1474, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.5281037083426856e-05, | |
| "loss": 2.2307, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.5263814781486874e-05, | |
| "loss": 2.1796, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.52465708555087e-05, | |
| "loss": 2.1384, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.522930537633133e-05, | |
| "loss": 2.228, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.5212018414882286e-05, | |
| "loss": 2.1615, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1.5194710042177354e-05, | |
| "loss": 2.3096, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1.517738032932027e-05, | |
| "loss": 2.135, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1.516002934750244e-05, | |
| "loss": 2.2433, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1.5142657168002646e-05, | |
| "loss": 2.1809, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.5125263862186744e-05, | |
| "loss": 2.1299, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.5107849501507383e-05, | |
| "loss": 2.2048, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.5090414157503715e-05, | |
| "loss": 2.1785, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.5072957901801075e-05, | |
| "loss": 2.2413, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.5055480806110716e-05, | |
| "loss": 2.2257, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.5037982942229503e-05, | |
| "loss": 2.2031, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.5020464382039613e-05, | |
| "loss": 2.1196, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.500292519750824e-05, | |
| "loss": 2.1791, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1.4985365460687322e-05, | |
| "loss": 2.183, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1.496778524371321e-05, | |
| "loss": 2.1859, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1.4950184618806395e-05, | |
| "loss": 2.2365, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1.4932563658271205e-05, | |
| "loss": 2.1979, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.49149224344955e-05, | |
| "loss": 2.1962, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.4897261019950398e-05, | |
| "loss": 2.2008, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.4879579487189948e-05, | |
| "loss": 2.1023, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.4861877908850856e-05, | |
| "loss": 2.2018, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 1.4844156357652166e-05, | |
| "loss": 2.2546, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 1.4826414906394979e-05, | |
| "loss": 2.2006, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 1.4808653627962142e-05, | |
| "loss": 2.2682, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 1.4790872595317962e-05, | |
| "loss": 2.1967, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1.4773071881507883e-05, | |
| "loss": 2.2844, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1.475525155965821e-05, | |
| "loss": 2.2192, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1.4737411702975801e-05, | |
| "loss": 2.277, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1.4719552384747754e-05, | |
| "loss": 2.2286, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.4701673678341131e-05, | |
| "loss": 2.1919, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.4683775657202623e-05, | |
| "loss": 2.2597, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.4665858394858286e-05, | |
| "loss": 2.2335, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.4647921964913203e-05, | |
| "loss": 2.2302, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 1.4629966441051208e-05, | |
| "loss": 2.2499, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 1.4611991897034576e-05, | |
| "loss": 2.2377, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 1.4593998406703714e-05, | |
| "loss": 2.1941, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 1.457598604397686e-05, | |
| "loss": 2.2474, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.455795488284978e-05, | |
| "loss": 2.1944, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.4539904997395468e-05, | |
| "loss": 2.1953, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.4521836461763844e-05, | |
| "loss": 2.1912, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.4503749350181427e-05, | |
| "loss": 2.2628, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1.448564373695107e-05, | |
| "loss": 2.1896, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1.4467519696451608e-05, | |
| "loss": 2.2361, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1.4449377303137594e-05, | |
| "loss": 2.1343, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1.4431216631538969e-05, | |
| "loss": 2.1466, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1.4413037756260763e-05, | |
| "loss": 2.2172, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1.4394840751982785e-05, | |
| "loss": 2.1724, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1.437662569345932e-05, | |
| "loss": 2.2752, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1.4358392655518824e-05, | |
| "loss": 2.1815, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 1.4340141713063611e-05, | |
| "loss": 2.2033, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 1.4321872941069548e-05, | |
| "loss": 2.1911, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 1.4303586414585744e-05, | |
| "loss": 2.0747, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 1.4285282208734252e-05, | |
| "loss": 2.1672, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1.4266960398709745e-05, | |
| "loss": 2.1663, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1.4248621059779213e-05, | |
| "loss": 2.2579, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1.423026426728167e-05, | |
| "loss": 2.1535, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1.4211890096627814e-05, | |
| "loss": 2.2365, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 1.4193498623299747e-05, | |
| "loss": 2.1206, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 1.4175089922850633e-05, | |
| "loss": 2.2039, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 1.4156664070904426e-05, | |
| "loss": 2.1958, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 1.4138221143155528e-05, | |
| "loss": 2.1774, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 1.4119761215368493e-05, | |
| "loss": 2.0998, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 1.4101284363377712e-05, | |
| "loss": 2.1687, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 1.4082790663087098e-05, | |
| "loss": 2.2085, | |
| "step": 1232 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 1.4064280190469784e-05, | |
| "loss": 2.1128, | |
| "step": 1234 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 1.4045753021567793e-05, | |
| "loss": 2.1849, | |
| "step": 1236 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 1.4027209232491758e-05, | |
| "loss": 2.167, | |
| "step": 1238 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 1.4008648899420564e-05, | |
| "loss": 2.1118, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 1.3990072098601078e-05, | |
| "loss": 2.2264, | |
| "step": 1242 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 1.3971478906347806e-05, | |
| "loss": 2.2442, | |
| "step": 1244 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 1.3952869399042602e-05, | |
| "loss": 2.1487, | |
| "step": 1246 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 1.3934243653134335e-05, | |
| "loss": 2.1854, | |
| "step": 1248 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 1.391560174513858e-05, | |
| "loss": 2.23, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 1.3896943751637317e-05, | |
| "loss": 2.1585, | |
| "step": 1252 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 1.3878269749278598e-05, | |
| "loss": 2.1637, | |
| "step": 1254 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 1.3859579814776245e-05, | |
| "loss": 2.1457, | |
| "step": 1256 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 1.384087402490953e-05, | |
| "loss": 2.2297, | |
| "step": 1258 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 1.3822152456522852e-05, | |
| "loss": 2.2493, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 1.3803415186525442e-05, | |
| "loss": 2.1561, | |
| "step": 1262 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 1.378466229189102e-05, | |
| "loss": 2.1174, | |
| "step": 1264 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 1.3765893849657502e-05, | |
| "loss": 2.2382, | |
| "step": 1266 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 1.3747109936926674e-05, | |
| "loss": 2.27, | |
| "step": 1268 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 1.3728310630863877e-05, | |
| "loss": 2.1582, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 1.370949600869768e-05, | |
| "loss": 2.1154, | |
| "step": 1272 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 1.3690666147719578e-05, | |
| "loss": 2.2166, | |
| "step": 1274 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1.3671821125283668e-05, | |
| "loss": 2.1755, | |
| "step": 1276 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1.3652961018806328e-05, | |
| "loss": 2.1349, | |
| "step": 1278 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1.363408590576591e-05, | |
| "loss": 2.2237, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1.3615195863702399e-05, | |
| "loss": 2.1227, | |
| "step": 1282 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 1.3596290970217125e-05, | |
| "loss": 2.1984, | |
| "step": 1284 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 1.3577371302972413e-05, | |
| "loss": 2.2181, | |
| "step": 1286 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 1.3558436939691297e-05, | |
| "loss": 2.0844, | |
| "step": 1288 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 1.3539487958157174e-05, | |
| "loss": 2.301, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1.3520524436213485e-05, | |
| "loss": 2.2049, | |
| "step": 1292 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1.3501546451763422e-05, | |
| "loss": 2.1921, | |
| "step": 1294 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1.3482554082769572e-05, | |
| "loss": 2.1569, | |
| "step": 1296 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1.3463547407253625e-05, | |
| "loss": 2.2407, | |
| "step": 1298 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.3444526503296043e-05, | |
| "loss": 2.1856, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.3425491449035734e-05, | |
| "loss": 2.1782, | |
| "step": 1302 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.3406442322669743e-05, | |
| "loss": 2.2341, | |
| "step": 1304 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.3387379202452917e-05, | |
| "loss": 2.2338, | |
| "step": 1306 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1.3368302166697591e-05, | |
| "loss": 2.1966, | |
| "step": 1308 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1.3349211293773272e-05, | |
| "loss": 2.1901, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1.3330106662106308e-05, | |
| "loss": 2.1638, | |
| "step": 1312 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1.3310988350179565e-05, | |
| "loss": 2.2089, | |
| "step": 1314 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1.3291856436532111e-05, | |
| "loss": 2.2215, | |
| "step": 1316 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1.3272710999758891e-05, | |
| "loss": 2.1695, | |
| "step": 1318 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1.3253552118510404e-05, | |
| "loss": 2.1879, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1.3234379871492381e-05, | |
| "loss": 2.1382, | |
| "step": 1322 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1.3215194337465458e-05, | |
| "loss": 2.197, | |
| "step": 1324 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1.3195995595244853e-05, | |
| "loss": 2.1053, | |
| "step": 1326 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1.317678372370004e-05, | |
| "loss": 2.136, | |
| "step": 1328 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1.3157558801754445e-05, | |
| "loss": 2.1793, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.3138320908385094e-05, | |
| "loss": 2.2774, | |
| "step": 1332 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.3119070122622299e-05, | |
| "loss": 2.1748, | |
| "step": 1334 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.3099806523549341e-05, | |
| "loss": 2.1782, | |
| "step": 1336 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.3080530190302132e-05, | |
| "loss": 2.0877, | |
| "step": 1338 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1.3061241202068903e-05, | |
| "loss": 2.1336, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1.3041939638089866e-05, | |
| "loss": 2.1854, | |
| "step": 1342 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1.3022625577656905e-05, | |
| "loss": 2.2033, | |
| "step": 1344 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1.3003299100113226e-05, | |
| "loss": 2.2056, | |
| "step": 1346 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.2983960284853059e-05, | |
| "loss": 2.1738, | |
| "step": 1348 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.2964609211321306e-05, | |
| "loss": 2.1135, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.2945245959013238e-05, | |
| "loss": 2.0968, | |
| "step": 1352 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.292587060747415e-05, | |
| "loss": 2.1987, | |
| "step": 1354 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.2906483236299039e-05, | |
| "loss": 2.1521, | |
| "step": 1356 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.2887083925132284e-05, | |
| "loss": 2.181, | |
| "step": 1358 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.286767275366732e-05, | |
| "loss": 2.0779, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.2848249801646288e-05, | |
| "loss": 2.1716, | |
| "step": 1362 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1.2828815148859743e-05, | |
| "loss": 2.1676, | |
| "step": 1364 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1.2809368875146295e-05, | |
| "loss": 2.1887, | |
| "step": 1366 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1.2789911060392295e-05, | |
| "loss": 2.1923, | |
| "step": 1368 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1.2770441784531507e-05, | |
| "loss": 2.1881, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 1.2750961127544782e-05, | |
| "loss": 2.1515, | |
| "step": 1372 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 1.2731469169459719e-05, | |
| "loss": 2.1685, | |
| "step": 1374 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 1.2711965990350343e-05, | |
| "loss": 2.2019, | |
| "step": 1376 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 1.2692451670336782e-05, | |
| "loss": 2.1065, | |
| "step": 1378 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 1.2672926289584923e-05, | |
| "loss": 2.1505, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 1.2653389928306097e-05, | |
| "loss": 2.1863, | |
| "step": 1382 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 1.2633842666756739e-05, | |
| "loss": 2.134, | |
| "step": 1384 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 1.2614284585238069e-05, | |
| "loss": 2.102, | |
| "step": 1386 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 1.2594715764095751e-05, | |
| "loss": 2.1334, | |
| "step": 1388 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 1.2575136283719567e-05, | |
| "loss": 2.1314, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 1.255554622454309e-05, | |
| "loss": 2.1211, | |
| "step": 1392 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 1.2535945667043357e-05, | |
| "loss": 2.1146, | |
| "step": 1394 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 1.2516334691740526e-05, | |
| "loss": 2.1072, | |
| "step": 1396 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 1.2496713379197547e-05, | |
| "loss": 2.2365, | |
| "step": 1398 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 1.2486898871648552e-05, | |
| "loss": 2.1682, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 1.2467262204393702e-05, | |
| "loss": 2.1776, | |
| "step": 1402 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 1.2447615401491164e-05, | |
| "loss": 2.2121, | |
| "step": 1404 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 1.2427958543651076e-05, | |
| "loss": 2.215, | |
| "step": 1406 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 1.2408291711624878e-05, | |
| "loss": 2.1945, | |
| "step": 1408 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 1.2388614986204992e-05, | |
| "loss": 2.0864, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 1.2368928448224475e-05, | |
| "loss": 2.3703, | |
| "step": 1412 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 1.2349232178556703e-05, | |
| "loss": 2.1804, | |
| "step": 1414 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 1.2329526258115022e-05, | |
| "loss": 2.0809, | |
| "step": 1416 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 1.2309810767852435e-05, | |
| "loss": 2.1397, | |
| "step": 1418 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 1.2290085788761243e-05, | |
| "loss": 2.1471, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 1.2270351401872738e-05, | |
| "loss": 2.1999, | |
| "step": 1422 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 1.2250607688256859e-05, | |
| "loss": 2.1743, | |
| "step": 1424 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 1.223085472902186e-05, | |
| "loss": 2.1734, | |
| "step": 1426 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 1.2211092605313971e-05, | |
| "loss": 2.1747, | |
| "step": 1428 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 1.2191321398317078e-05, | |
| "loss": 2.0526, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 1.2171541189252374e-05, | |
| "loss": 2.2363, | |
| "step": 1432 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 1.2151752059378033e-05, | |
| "loss": 2.2285, | |
| "step": 1434 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 1.213195408998889e-05, | |
| "loss": 2.166, | |
| "step": 1436 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 1.2112147362416076e-05, | |
| "loss": 2.153, | |
| "step": 1438 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 1.2092331958026709e-05, | |
| "loss": 2.0964, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 1.2072507958223551e-05, | |
| "loss": 2.2554, | |
| "step": 1442 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.2052675444444675e-05, | |
| "loss": 2.1397, | |
| "step": 1444 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.2032834498163128e-05, | |
| "loss": 2.0511, | |
| "step": 1446 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.2012985200886602e-05, | |
| "loss": 2.127, | |
| "step": 1448 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.1993127634157089e-05, | |
| "loss": 2.1393, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.1973261879550562e-05, | |
| "loss": 2.1654, | |
| "step": 1452 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.1953388018676618e-05, | |
| "loss": 2.2194, | |
| "step": 1454 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.1933506133178162e-05, | |
| "loss": 2.1934, | |
| "step": 1456 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.1913616304731064e-05, | |
| "loss": 2.1682, | |
| "step": 1458 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.189371861504383e-05, | |
| "loss": 2.1072, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.187381314585725e-05, | |
| "loss": 2.2418, | |
| "step": 1462 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.1853899978944072e-05, | |
| "loss": 2.1983, | |
| "step": 1464 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.183397919610868e-05, | |
| "loss": 2.1958, | |
| "step": 1466 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.1814050879186732e-05, | |
| "loss": 2.2395, | |
| "step": 1468 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.1794115110044842e-05, | |
| "loss": 2.1022, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.1774171970580235e-05, | |
| "loss": 2.0995, | |
| "step": 1472 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.175422154272042e-05, | |
| "loss": 2.1589, | |
| "step": 1474 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.1734263908422837e-05, | |
| "loss": 2.1545, | |
| "step": 1476 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.1714299149674538e-05, | |
| "loss": 2.0923, | |
| "step": 1478 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.1694327348491843e-05, | |
| "loss": 2.1764, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.1674348586919997e-05, | |
| "loss": 2.1114, | |
| "step": 1482 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.1654362947032846e-05, | |
| "loss": 2.1514, | |
| "step": 1484 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.1634370510932487e-05, | |
| "loss": 2.1481, | |
| "step": 1486 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.1614371360748934e-05, | |
| "loss": 2.1236, | |
| "step": 1488 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.1594365578639792e-05, | |
| "loss": 2.2154, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.1574353246789904e-05, | |
| "loss": 2.1229, | |
| "step": 1492 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.1554334447411019e-05, | |
| "loss": 2.1379, | |
| "step": 1494 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.1534309262741457e-05, | |
| "loss": 2.113, | |
| "step": 1496 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.1514277775045768e-05, | |
| "loss": 2.1734, | |
| "step": 1498 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.1494240066614396e-05, | |
| "loss": 2.1214, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.1474196219763345e-05, | |
| "loss": 2.1648, | |
| "step": 1502 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.1454146316833823e-05, | |
| "loss": 2.1457, | |
| "step": 1504 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.1434090440191927e-05, | |
| "loss": 2.121, | |
| "step": 1506 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.1414028672228294e-05, | |
| "loss": 2.1855, | |
| "step": 1508 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.1393961095357761e-05, | |
| "loss": 2.2675, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.1373887792019032e-05, | |
| "loss": 2.1543, | |
| "step": 1512 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.1353808844674321e-05, | |
| "loss": 2.0945, | |
| "step": 1514 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.1333724335809048e-05, | |
| "loss": 2.2255, | |
| "step": 1516 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.1313634347931466e-05, | |
| "loss": 2.172, | |
| "step": 1518 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.1293538963572343e-05, | |
| "loss": 2.2284, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.1273438265284615e-05, | |
| "loss": 2.1856, | |
| "step": 1522 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.1253332335643043e-05, | |
| "loss": 2.1535, | |
| "step": 1524 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.1233221257243888e-05, | |
| "loss": 2.0985, | |
| "step": 1526 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.121310511270455e-05, | |
| "loss": 2.1508, | |
| "step": 1528 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.1192983984663253e-05, | |
| "loss": 2.1423, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.1172857955778684e-05, | |
| "loss": 2.1889, | |
| "step": 1532 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.1152727108729675e-05, | |
| "loss": 2.2269, | |
| "step": 1534 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.1132591526214835e-05, | |
| "loss": 2.1619, | |
| "step": 1536 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.1112451290952238e-05, | |
| "loss": 2.2737, | |
| "step": 1538 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.109230648567907e-05, | |
| "loss": 2.2028, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.1072157193151289e-05, | |
| "loss": 2.097, | |
| "step": 1542 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.105200349614329e-05, | |
| "loss": 2.1606, | |
| "step": 1544 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.1031845477447554e-05, | |
| "loss": 2.2113, | |
| "step": 1546 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.1011683219874324e-05, | |
| "loss": 2.072, | |
| "step": 1548 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.0991516806251252e-05, | |
| "loss": 2.162, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.097134631942307e-05, | |
| "loss": 2.1621, | |
| "step": 1552 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.0951171842251236e-05, | |
| "loss": 2.1583, | |
| "step": 1554 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.0930993457613597e-05, | |
| "loss": 2.2203, | |
| "step": 1556 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.0910811248404064e-05, | |
| "loss": 2.1624, | |
| "step": 1558 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.089062529753225e-05, | |
| "loss": 2.0793, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.0870435687923144e-05, | |
| "loss": 2.2069, | |
| "step": 1562 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.0850242502516761e-05, | |
| "loss": 2.1035, | |
| "step": 1564 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.0830045824267808e-05, | |
| "loss": 2.0654, | |
| "step": 1566 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.0809845736145346e-05, | |
| "loss": 2.2146, | |
| "step": 1568 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.0789642321132427e-05, | |
| "loss": 2.184, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.0769435662225796e-05, | |
| "loss": 2.1924, | |
| "step": 1572 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.0749225842435498e-05, | |
| "loss": 2.09, | |
| "step": 1574 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.0729012944784583e-05, | |
| "loss": 2.1134, | |
| "step": 1576 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.070879705230873e-05, | |
| "loss": 2.1189, | |
| "step": 1578 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.0688578248055934e-05, | |
| "loss": 2.1949, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.0668356615086142e-05, | |
| "loss": 2.0999, | |
| "step": 1582 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.0648132236470924e-05, | |
| "loss": 2.0699, | |
| "step": 1584 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.0627905195293135e-05, | |
| "loss": 2.2001, | |
| "step": 1586 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.060767557464656e-05, | |
| "loss": 2.16, | |
| "step": 1588 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.0587443457635583e-05, | |
| "loss": 2.1827, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.0567208927374845e-05, | |
| "loss": 2.1324, | |
| "step": 1592 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.0546972066988901e-05, | |
| "loss": 2.1059, | |
| "step": 1594 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.0526732959611879e-05, | |
| "loss": 2.1654, | |
| "step": 1596 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.0506491688387128e-05, | |
| "loss": 2.131, | |
| "step": 1598 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.0486248336466904e-05, | |
| "loss": 2.0462, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.0466002987011989e-05, | |
| "loss": 2.1193, | |
| "step": 1602 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.0445755723191397e-05, | |
| "loss": 2.0565, | |
| "step": 1604 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.042550662818198e-05, | |
| "loss": 2.1337, | |
| "step": 1606 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.0405255785168131e-05, | |
| "loss": 1.9967, | |
| "step": 1608 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.0385003277341417e-05, | |
| "loss": 2.0656, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.0364749187900237e-05, | |
| "loss": 2.149, | |
| "step": 1612 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.0344493600049509e-05, | |
| "loss": 2.1519, | |
| "step": 1614 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.0324236597000279e-05, | |
| "loss": 2.1354, | |
| "step": 1616 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.030397826196943e-05, | |
| "loss": 2.1029, | |
| "step": 1618 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.02837186781793e-05, | |
| "loss": 2.0304, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.0263457928857368e-05, | |
| "loss": 2.117, | |
| "step": 1622 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.0243196097235897e-05, | |
| "loss": 2.0862, | |
| "step": 1624 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.0222933266551594e-05, | |
| "loss": 2.0838, | |
| "step": 1626 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.0202669520045278e-05, | |
| "loss": 2.0735, | |
| "step": 1628 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.0182404940961517e-05, | |
| "loss": 2.1451, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.0162139612548312e-05, | |
| "loss": 2.1004, | |
| "step": 1632 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.0141873618056736e-05, | |
| "loss": 2.1186, | |
| "step": 1634 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.01216070407406e-05, | |
| "loss": 2.0799, | |
| "step": 1636 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.0101339963856112e-05, | |
| "loss": 2.0765, | |
| "step": 1638 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.0081072470661522e-05, | |
| "loss": 2.0889, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.0060804644416804e-05, | |
| "loss": 2.1328, | |
| "step": 1642 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.0040536568383285e-05, | |
| "loss": 2.0836, | |
| "step": 1644 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.002026832582334e-05, | |
| "loss": 2.0986, | |
| "step": 1646 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1e-05, | |
| "loss": 2.1852, | |
| "step": 1648 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 9.979731674176667e-06, | |
| "loss": 2.16, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 9.959463431616715e-06, | |
| "loss": 2.1033, | |
| "step": 1652 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 9.939195355583199e-06, | |
| "loss": 2.0556, | |
| "step": 1654 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 9.918927529338481e-06, | |
| "loss": 2.0838, | |
| "step": 1656 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 9.898660036143893e-06, | |
| "loss": 2.0993, | |
| "step": 1658 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 9.8783929592594e-06, | |
| "loss": 2.0997, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 9.858126381943266e-06, | |
| "loss": 2.1296, | |
| "step": 1662 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 9.837860387451691e-06, | |
| "loss": 2.1567, | |
| "step": 1664 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 9.817595059038483e-06, | |
| "loss": 2.0591, | |
| "step": 1666 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 9.797330479954725e-06, | |
| "loss": 2.0996, | |
| "step": 1668 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 9.777066733448408e-06, | |
| "loss": 2.1782, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 9.756803902764108e-06, | |
| "loss": 2.1045, | |
| "step": 1672 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 9.736542071142635e-06, | |
| "loss": 2.2661, | |
| "step": 1674 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 9.716281321820704e-06, | |
| "loss": 2.1533, | |
| "step": 1676 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 9.696021738030575e-06, | |
| "loss": 2.1895, | |
| "step": 1678 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 9.675763402999721e-06, | |
| "loss": 2.0888, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 9.655506399950496e-06, | |
| "loss": 2.1797, | |
| "step": 1682 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 9.635250812099766e-06, | |
| "loss": 2.1408, | |
| "step": 1684 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 9.61499672265859e-06, | |
| "loss": 2.1363, | |
| "step": 1686 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 9.59474421483187e-06, | |
| "loss": 2.062, | |
| "step": 1688 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 9.574493371818021e-06, | |
| "loss": 2.0656, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 9.554244276808608e-06, | |
| "loss": 2.105, | |
| "step": 1692 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 9.53399701298801e-06, | |
| "loss": 2.0811, | |
| "step": 1694 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 9.5137516635331e-06, | |
| "loss": 2.1098, | |
| "step": 1696 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 9.493508311612874e-06, | |
| "loss": 2.131, | |
| "step": 1698 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 9.473267040388128e-06, | |
| "loss": 2.1046, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 9.453027933011099e-06, | |
| "loss": 2.047, | |
| "step": 1702 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 9.432791072625158e-06, | |
| "loss": 2.1337, | |
| "step": 1704 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 9.412556542364419e-06, | |
| "loss": 2.0786, | |
| "step": 1706 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 9.392324425353441e-06, | |
| "loss": 2.1124, | |
| "step": 1708 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 9.372094804706867e-06, | |
| "loss": 2.1203, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 9.351867763529079e-06, | |
| "loss": 2.0938, | |
| "step": 1712 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 9.331643384913865e-06, | |
| "loss": 2.0424, | |
| "step": 1714 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 9.31142175194407e-06, | |
| "loss": 2.0828, | |
| "step": 1716 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 9.291202947691272e-06, | |
| "loss": 2.1315, | |
| "step": 1718 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 9.270987055215422e-06, | |
| "loss": 2.0857, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 9.250774157564503e-06, | |
| "loss": 2.1883, | |
| "step": 1722 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 9.230564337774207e-06, | |
| "loss": 2.0885, | |
| "step": 1724 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 9.210357678867576e-06, | |
| "loss": 2.1303, | |
| "step": 1726 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 9.19015426385466e-06, | |
| "loss": 2.0631, | |
| "step": 1728 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 9.169954175732192e-06, | |
| "loss": 2.1708, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 9.149757497483242e-06, | |
| "loss": 2.0478, | |
| "step": 1732 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 9.129564312076861e-06, | |
| "loss": 2.0841, | |
| "step": 1734 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 9.109374702467752e-06, | |
| "loss": 2.1702, | |
| "step": 1736 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 9.089188751595937e-06, | |
| "loss": 2.156, | |
| "step": 1738 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 9.069006542386405e-06, | |
| "loss": 2.1284, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 9.04882815774877e-06, | |
| "loss": 2.0838, | |
| "step": 1742 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 9.028653680576931e-06, | |
| "loss": 2.1409, | |
| "step": 1744 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 9.00848319374875e-06, | |
| "loss": 2.1278, | |
| "step": 1746 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 8.98831678012568e-06, | |
| "loss": 2.0312, | |
| "step": 1748 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 8.968154522552448e-06, | |
| "loss": 2.0541, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 8.947996503856714e-06, | |
| "loss": 2.0604, | |
| "step": 1752 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 8.927842806848714e-06, | |
| "loss": 2.1058, | |
| "step": 1754 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 8.907693514320935e-06, | |
| "loss": 2.0345, | |
| "step": 1756 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 8.887548709047765e-06, | |
| "loss": 2.1298, | |
| "step": 1758 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 8.867408473785168e-06, | |
| "loss": 2.0751, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 8.84727289127033e-06, | |
| "loss": 2.0627, | |
| "step": 1762 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 8.827142044221316e-06, | |
| "loss": 2.0686, | |
| "step": 1764 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 8.80701601533675e-06, | |
| "loss": 2.0955, | |
| "step": 1766 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 8.786894887295451e-06, | |
| "loss": 2.0525, | |
| "step": 1768 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 8.766778742756117e-06, | |
| "loss": 2.1369, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 8.746667664356957e-06, | |
| "loss": 2.0573, | |
| "step": 1772 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 8.726561734715388e-06, | |
| "loss": 2.0904, | |
| "step": 1774 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 8.70646103642766e-06, | |
| "loss": 2.1056, | |
| "step": 1776 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 8.686365652068536e-06, | |
| "loss": 2.165, | |
| "step": 1778 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 8.666275664190954e-06, | |
| "loss": 2.0661, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 8.646191155325682e-06, | |
| "loss": 2.0031, | |
| "step": 1782 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 8.626112207980974e-06, | |
| "loss": 2.0031, | |
| "step": 1784 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 8.606038904642239e-06, | |
| "loss": 2.1642, | |
| "step": 1786 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 8.585971327771707e-06, | |
| "loss": 2.0883, | |
| "step": 1788 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 8.565909559808076e-06, | |
| "loss": 2.2256, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 8.545853683166179e-06, | |
| "loss": 2.0442, | |
| "step": 1792 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 8.52580378023666e-06, | |
| "loss": 2.1219, | |
| "step": 1794 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 8.505759933385607e-06, | |
| "loss": 2.1003, | |
| "step": 1796 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 8.485722224954237e-06, | |
| "loss": 1.9844, | |
| "step": 1798 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 8.465690737258547e-06, | |
| "loss": 2.1608, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 8.445665552588983e-06, | |
| "loss": 2.0345, | |
| "step": 1802 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 8.4256467532101e-06, | |
| "loss": 2.1158, | |
| "step": 1804 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 8.405634421360207e-06, | |
| "loss": 2.1637, | |
| "step": 1806 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 8.385628639251067e-06, | |
| "loss": 2.1875, | |
| "step": 1808 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 8.365629489067516e-06, | |
| "loss": 2.0898, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 8.345637052967157e-06, | |
| "loss": 2.044, | |
| "step": 1812 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 8.325651413080003e-06, | |
| "loss": 2.1416, | |
| "step": 1814 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 8.30567265150816e-06, | |
| "loss": 2.1302, | |
| "step": 1816 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 8.285700850325467e-06, | |
| "loss": 2.1407, | |
| "step": 1818 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 8.265736091577165e-06, | |
| "loss": 2.0054, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 8.245778457279583e-06, | |
| "loss": 2.1048, | |
| "step": 1822 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 8.225828029419767e-06, | |
| "loss": 2.1311, | |
| "step": 1824 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 8.205884889955163e-06, | |
| "loss": 2.1376, | |
| "step": 1826 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 8.18594912081327e-06, | |
| "loss": 2.0911, | |
| "step": 1828 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 8.166020803891324e-06, | |
| "loss": 2.1059, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 8.146100021055931e-06, | |
| "loss": 2.0526, | |
| "step": 1832 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 8.126186854142752e-06, | |
| "loss": 2.0698, | |
| "step": 1834 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 8.106281384956171e-06, | |
| "loss": 2.0709, | |
| "step": 1836 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 8.086383695268937e-06, | |
| "loss": 2.1658, | |
| "step": 1838 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 8.066493866821843e-06, | |
| "loss": 2.0694, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 8.046611981323387e-06, | |
| "loss": 2.0544, | |
| "step": 1842 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 8.026738120449442e-06, | |
| "loss": 2.0928, | |
| "step": 1844 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 8.006872365842913e-06, | |
| "loss": 2.1327, | |
| "step": 1846 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 7.987014799113398e-06, | |
| "loss": 2.1528, | |
| "step": 1848 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 7.967165501836873e-06, | |
| "loss": 2.111, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 7.947324555555327e-06, | |
| "loss": 2.0708, | |
| "step": 1852 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 7.927492041776452e-06, | |
| "loss": 2.0502, | |
| "step": 1854 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 7.907668041973293e-06, | |
| "loss": 2.0583, | |
| "step": 1856 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 7.887852637583927e-06, | |
| "loss": 2.0626, | |
| "step": 1858 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 7.868045910011115e-06, | |
| "loss": 2.0572, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 7.848247940621967e-06, | |
| "loss": 2.1325, | |
| "step": 1862 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 7.82845881074763e-06, | |
| "loss": 2.0376, | |
| "step": 1864 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 7.808678601682927e-06, | |
| "loss": 2.118, | |
| "step": 1866 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 7.788907394686032e-06, | |
| "loss": 2.2412, | |
| "step": 1868 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 7.769145270978142e-06, | |
| "loss": 2.1014, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 7.749392311743143e-06, | |
| "loss": 2.0906, | |
| "step": 1872 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 7.729648598127263e-06, | |
| "loss": 2.2223, | |
| "step": 1874 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 7.709914211238757e-06, | |
| "loss": 2.0494, | |
| "step": 1876 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 7.690189232147566e-06, | |
| "loss": 2.1574, | |
| "step": 1878 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 7.67047374188498e-06, | |
| "loss": 2.0812, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 7.650767821443302e-06, | |
| "loss": 2.1376, | |
| "step": 1882 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 7.631071551775527e-06, | |
| "loss": 2.1414, | |
| "step": 1884 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 7.611385013795011e-06, | |
| "loss": 2.0865, | |
| "step": 1886 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 7.5917082883751256e-06, | |
| "loss": 2.1289, | |
| "step": 1888 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 7.572041456348925e-06, | |
| "loss": 2.0557, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 7.552384598508837e-06, | |
| "loss": 2.1961, | |
| "step": 1892 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 7.532737795606299e-06, | |
| "loss": 2.1151, | |
| "step": 1894 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 7.513101128351454e-06, | |
| "loss": 2.0157, | |
| "step": 1896 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 7.493474677412795e-06, | |
| "loss": 1.9539, | |
| "step": 1898 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 7.473858523416859e-06, | |
| "loss": 2.1574, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 7.454252746947874e-06, | |
| "loss": 1.9891, | |
| "step": 1902 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 7.434657428547436e-06, | |
| "loss": 2.1115, | |
| "step": 1904 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 7.415072648714186e-06, | |
| "loss": 2.0999, | |
| "step": 1906 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 7.39549848790347e-06, | |
| "loss": 2.0954, | |
| "step": 1908 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 7.3759350265270025e-06, | |
| "loss": 2.0767, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 7.356382344952549e-06, | |
| "loss": 2.0648, | |
| "step": 1912 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 7.336840523503597e-06, | |
| "loss": 2.0533, | |
| "step": 1914 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 7.31730964245901e-06, | |
| "loss": 2.1239, | |
| "step": 1916 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 7.297789782052716e-06, | |
| "loss": 2.0429, | |
| "step": 1918 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 7.278281022473362e-06, | |
| "loss": 2.1153, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 7.2587834438640024e-06, | |
| "loss": 2.1179, | |
| "step": 1922 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 7.239297126321749e-06, | |
| "loss": 2.0549, | |
| "step": 1924 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 7.2198221498974565e-06, | |
| "loss": 1.9891, | |
| "step": 1926 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 7.200358594595391e-06, | |
| "loss": 2.1248, | |
| "step": 1928 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 7.180906540372904e-06, | |
| "loss": 2.1624, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 7.161466067140089e-06, | |
| "loss": 2.0559, | |
| "step": 1932 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 7.142037254759469e-06, | |
| "loss": 2.0102, | |
| "step": 1934 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 7.12262018304567e-06, | |
| "loss": 2.0574, | |
| "step": 1936 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 7.10321493176508e-06, | |
| "loss": 2.1346, | |
| "step": 1938 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 7.083821580635526e-06, | |
| "loss": 2.1182, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 7.064440209325958e-06, | |
| "loss": 2.1391, | |
| "step": 1942 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 7.045070897456105e-06, | |
| "loss": 2.1456, | |
| "step": 1944 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 7.025713724596151e-06, | |
| "loss": 2.0655, | |
| "step": 1946 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 7.006368770266421e-06, | |
| "loss": 2.1471, | |
| "step": 1948 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 6.987036113937045e-06, | |
| "loss": 1.9975, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 6.967715835027629e-06, | |
| "loss": 2.0847, | |
| "step": 1952 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 6.948408012906927e-06, | |
| "loss": 2.0425, | |
| "step": 1954 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 6.929112726892533e-06, | |
| "loss": 2.0219, | |
| "step": 1956 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 6.909830056250527e-06, | |
| "loss": 2.0583, | |
| "step": 1958 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 6.890560080195178e-06, | |
| "loss": 2.1057, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 6.87130287788859e-06, | |
| "loss": 2.1054, | |
| "step": 1962 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 6.852058528440413e-06, | |
| "loss": 2.0865, | |
| "step": 1964 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 6.832827110907472e-06, | |
| "loss": 2.0924, | |
| "step": 1966 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 6.813608704293485e-06, | |
| "loss": 2.0896, | |
| "step": 1968 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 6.794403387548713e-06, | |
| "loss": 2.0256, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 6.775211239569647e-06, | |
| "loss": 2.0805, | |
| "step": 1972 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 6.756032339198675e-06, | |
| "loss": 2.1204, | |
| "step": 1974 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 6.736866765223762e-06, | |
| "loss": 2.0933, | |
| "step": 1976 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 6.717714596378138e-06, | |
| "loss": 2.0873, | |
| "step": 1978 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 6.698575911339955e-06, | |
| "loss": 2.0037, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 6.679450788731971e-06, | |
| "loss": 2.0634, | |
| "step": 1982 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 6.66033930712124e-06, | |
| "loss": 2.0638, | |
| "step": 1984 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 6.641241545018766e-06, | |
| "loss": 2.156, | |
| "step": 1986 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 6.622157580879196e-06, | |
| "loss": 2.134, | |
| "step": 1988 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 6.603087493100495e-06, | |
| "loss": 2.0913, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 6.584031360023631e-06, | |
| "loss": 2.0966, | |
| "step": 1992 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 6.564989259932231e-06, | |
| "loss": 2.0767, | |
| "step": 1994 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 6.545961271052279e-06, | |
| "loss": 1.9789, | |
| "step": 1996 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 6.526947471551799e-06, | |
| "loss": 2.0495, | |
| "step": 1998 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 6.507947939540509e-06, | |
| "loss": 2.0231, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 6.488962753069528e-06, | |
| "loss": 2.1904, | |
| "step": 2002 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 6.469991990131031e-06, | |
| "loss": 2.1106, | |
| "step": 2004 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 6.451035728657953e-06, | |
| "loss": 2.0855, | |
| "step": 2006 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 6.4320940465236455e-06, | |
| "loss": 2.1483, | |
| "step": 2008 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 6.4131670215415755e-06, | |
| "loss": 2.0874, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 6.394254731464991e-06, | |
| "loss": 2.0334, | |
| "step": 2012 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 6.375357253986614e-06, | |
| "loss": 2.0912, | |
| "step": 2014 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 6.356474666738309e-06, | |
| "loss": 2.0309, | |
| "step": 2016 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 6.337607047290774e-06, | |
| "loss": 2.1546, | |
| "step": 2018 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 6.318754473153221e-06, | |
| "loss": 2.0906, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 6.299917021773054e-06, | |
| "loss": 2.0163, | |
| "step": 2022 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 6.281094770535545e-06, | |
| "loss": 2.0965, | |
| "step": 2024 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 6.262287796763538e-06, | |
| "loss": 2.0918, | |
| "step": 2026 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 6.2434961777170985e-06, | |
| "loss": 2.1078, | |
| "step": 2028 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 6.224719990593224e-06, | |
| "loss": 2.1403, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 6.205959312525515e-06, | |
| "loss": 2.0206, | |
| "step": 2032 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 6.187214220583865e-06, | |
| "loss": 2.0553, | |
| "step": 2034 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 6.1684847917741296e-06, | |
| "loss": 2.0536, | |
| "step": 2036 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 6.149771103037821e-06, | |
| "loss": 2.0134, | |
| "step": 2038 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 6.1310732312518005e-06, | |
| "loss": 2.0589, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 6.112391253227939e-06, | |
| "loss": 2.0987, | |
| "step": 2042 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 6.093725245712825e-06, | |
| "loss": 2.0124, | |
| "step": 2044 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 6.075075285387431e-06, | |
| "loss": 2.1084, | |
| "step": 2046 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 6.056441448866817e-06, | |
| "loss": 2.0965, | |
| "step": 2048 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 6.037823812699792e-06, | |
| "loss": 2.1005, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 6.019222453368627e-06, | |
| "loss": 2.0974, | |
| "step": 2052 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 6.000637447288719e-06, | |
| "loss": 2.0383, | |
| "step": 2054 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 5.982068870808289e-06, | |
| "loss": 2.1205, | |
| "step": 2056 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 5.963516800208056e-06, | |
| "loss": 2.0986, | |
| "step": 2058 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 5.9449813117009396e-06, | |
| "loss": 2.026, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 5.926462481431738e-06, | |
| "loss": 2.1155, | |
| "step": 2062 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 5.907960385476821e-06, | |
| "loss": 2.0069, | |
| "step": 2064 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 5.889475099843798e-06, | |
| "loss": 2.1348, | |
| "step": 2066 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 5.87100670047124e-06, | |
| "loss": 2.1213, | |
| "step": 2068 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 5.852555263228334e-06, | |
| "loss": 2.0679, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 5.834120863914592e-06, | |
| "loss": 2.0824, | |
| "step": 2072 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 5.815703578259526e-06, | |
| "loss": 2.062, | |
| "step": 2074 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 5.7973034819223585e-06, | |
| "loss": 2.0782, | |
| "step": 2076 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 5.7789206504916815e-06, | |
| "loss": 2.0913, | |
| "step": 2078 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 5.760555159485173e-06, | |
| "loss": 2.0713, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 5.742207084349274e-06, | |
| "loss": 2.0219, | |
| "step": 2082 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 5.723876500458874e-06, | |
| "loss": 2.0551, | |
| "step": 2084 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 5.705563483117014e-06, | |
| "loss": 2.2002, | |
| "step": 2086 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 5.68726810755456e-06, | |
| "loss": 2.0751, | |
| "step": 2088 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 5.668990448929923e-06, | |
| "loss": 2.0974, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 5.650730582328718e-06, | |
| "loss": 2.0429, | |
| "step": 2092 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 5.6324885827634665e-06, | |
| "loss": 2.0511, | |
| "step": 2094 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 5.6142645251733094e-06, | |
| "loss": 2.0862, | |
| "step": 2096 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 5.5960584844236565e-06, | |
| "loss": 2.0579, | |
| "step": 2098 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 5.577870535305927e-06, | |
| "loss": 2.0178, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 5.559700752537198e-06, | |
| "loss": 2.0191, | |
| "step": 2102 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 5.541549210759939e-06, | |
| "loss": 2.0708, | |
| "step": 2104 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 5.5234159845416664e-06, | |
| "loss": 2.1528, | |
| "step": 2106 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 5.50530114837466e-06, | |
| "loss": 2.0773, | |
| "step": 2108 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 5.4872047766756605e-06, | |
| "loss": 2.0209, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 5.46912694378555e-06, | |
| "loss": 2.0081, | |
| "step": 2112 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 5.451067723969048e-06, | |
| "loss": 2.0386, | |
| "step": 2114 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.43302719141441e-06, | |
| "loss": 2.09, | |
| "step": 2116 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.415005420233141e-06, | |
| "loss": 2.05, | |
| "step": 2118 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.397002484459646e-06, | |
| "loss": 2.0565, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.379018458050974e-06, | |
| "loss": 1.996, | |
| "step": 2122 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 5.361053414886489e-06, | |
| "loss": 1.993, | |
| "step": 2124 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 5.343107428767564e-06, | |
| "loss": 2.0867, | |
| "step": 2126 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 5.32518057341729e-06, | |
| "loss": 2.0426, | |
| "step": 2128 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 5.3072729224801625e-06, | |
| "loss": 2.1515, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 5.289384549521793e-06, | |
| "loss": 2.0868, | |
| "step": 2132 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 5.271515528028592e-06, | |
| "loss": 2.0755, | |
| "step": 2134 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 5.25366593140747e-06, | |
| "loss": 2.0261, | |
| "step": 2136 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 5.235835832985552e-06, | |
| "loss": 2.1133, | |
| "step": 2138 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.218025306009845e-06, | |
| "loss": 2.0651, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.200234423646975e-06, | |
| "loss": 2.1348, | |
| "step": 2142 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.1824632589828465e-06, | |
| "loss": 2.0859, | |
| "step": 2144 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.164711885022382e-06, | |
| "loss": 2.0688, | |
| "step": 2146 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 5.146980374689192e-06, | |
| "loss": 2.0847, | |
| "step": 2148 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 5.129268800825281e-06, | |
| "loss": 2.1682, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 5.11157723619077e-06, | |
| "loss": 2.0623, | |
| "step": 2152 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 5.093905753463567e-06, | |
| "loss": 2.0576, | |
| "step": 2154 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 5.076254425239086e-06, | |
| "loss": 2.0539, | |
| "step": 2156 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 5.058623324029944e-06, | |
| "loss": 2.0917, | |
| "step": 2158 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 5.041012522265673e-06, | |
| "loss": 2.0234, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 5.0234220922923985e-06, | |
| "loss": 2.0465, | |
| "step": 2162 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 5.00585210637257e-06, | |
| "loss": 2.0762, | |
| "step": 2164 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.988302636684654e-06, | |
| "loss": 1.9968, | |
| "step": 2166 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.970773755322823e-06, | |
| "loss": 2.1092, | |
| "step": 2168 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.953265534296678e-06, | |
| "loss": 2.0072, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.935778045530942e-06, | |
| "loss": 2.0648, | |
| "step": 2172 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.91831136086518e-06, | |
| "loss": 2.1264, | |
| "step": 2174 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.900865552053483e-06, | |
| "loss": 2.0266, | |
| "step": 2176 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.8834406907641784e-06, | |
| "loss": 2.0758, | |
| "step": 2178 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.866036848579557e-06, | |
| "loss": 2.0309, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.848654096995542e-06, | |
| "loss": 2.0947, | |
| "step": 2182 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.831292507421434e-06, | |
| "loss": 2.0227, | |
| "step": 2184 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.813952151179582e-06, | |
| "loss": 2.1152, | |
| "step": 2186 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.79663309950512e-06, | |
| "loss": 2.104, | |
| "step": 2188 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.779335423545658e-06, | |
| "loss": 2.0274, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.762059194360984e-06, | |
| "loss": 2.0772, | |
| "step": 2192 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.744804482922799e-06, | |
| "loss": 2.0878, | |
| "step": 2194 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.727571360114395e-06, | |
| "loss": 2.0149, | |
| "step": 2196 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.710359896730379e-06, | |
| "loss": 2.0105, | |
| "step": 2198 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.6931701634763795e-06, | |
| "loss": 2.0797, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.676002230968764e-06, | |
| "loss": 2.1205, | |
| "step": 2202 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.6588561697343286e-06, | |
| "loss": 2.1343, | |
| "step": 2204 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.641732050210032e-06, | |
| "loss": 2.1242, | |
| "step": 2206 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.624629942742695e-06, | |
| "loss": 2.0614, | |
| "step": 2208 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 4.6075499175887015e-06, | |
| "loss": 2.0647, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 4.59049204491373e-06, | |
| "loss": 2.0874, | |
| "step": 2212 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 4.573456394792446e-06, | |
| "loss": 2.117, | |
| "step": 2214 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 4.556443037208238e-06, | |
| "loss": 2.1175, | |
| "step": 2216 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.539452042052901e-06, | |
| "loss": 2.0552, | |
| "step": 2218 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.522483479126367e-06, | |
| "loss": 2.1094, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.505537418136425e-06, | |
| "loss": 2.0207, | |
| "step": 2222 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.488613928698412e-06, | |
| "loss": 2.1713, | |
| "step": 2224 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.471713080334951e-06, | |
| "loss": 2.0321, | |
| "step": 2226 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.454834942475642e-06, | |
| "loss": 2.0425, | |
| "step": 2228 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.437979584456803e-06, | |
| "loss": 2.0765, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.42114707552116e-06, | |
| "loss": 2.1292, | |
| "step": 2232 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 4.404337484817577e-06, | |
| "loss": 2.0476, | |
| "step": 2234 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 4.387550881400772e-06, | |
| "loss": 2.1, | |
| "step": 2236 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 4.370787334231026e-06, | |
| "loss": 2.1019, | |
| "step": 2238 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 4.354046912173905e-06, | |
| "loss": 2.0955, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.3373296839999725e-06, | |
| "loss": 2.0964, | |
| "step": 2242 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.320635718384519e-06, | |
| "loss": 2.0485, | |
| "step": 2244 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.303965083907261e-06, | |
| "loss": 2.024, | |
| "step": 2246 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.287317849052075e-06, | |
| "loss": 2.1409, | |
| "step": 2248 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 4.270694082206716e-06, | |
| "loss": 2.0648, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 4.254093851662518e-06, | |
| "loss": 2.0482, | |
| "step": 2252 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 4.237517225614138e-06, | |
| "loss": 1.9841, | |
| "step": 2254 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 4.2209642721592506e-06, | |
| "loss": 2.0718, | |
| "step": 2256 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 4.204435059298303e-06, | |
| "loss": 2.0425, | |
| "step": 2258 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 4.187929654934196e-06, | |
| "loss": 2.1009, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 4.1714481268720305e-06, | |
| "loss": 2.0425, | |
| "step": 2262 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 4.154990542818828e-06, | |
| "loss": 2.0312, | |
| "step": 2264 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 4.138556970383235e-06, | |
| "loss": 2.0174, | |
| "step": 2266 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 4.12214747707527e-06, | |
| "loss": 2.0914, | |
| "step": 2268 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 4.105762130306018e-06, | |
| "loss": 2.0007, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 4.0894009973873874e-06, | |
| "loss": 2.0731, | |
| "step": 2272 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 4.073064145531798e-06, | |
| "loss": 2.0742, | |
| "step": 2274 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 4.056751641851925e-06, | |
| "loss": 2.1014, | |
| "step": 2276 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 4.040463553360431e-06, | |
| "loss": 2.0327, | |
| "step": 2278 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 4.024199946969668e-06, | |
| "loss": 2.0834, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 4.007960889491419e-06, | |
| "loss": 2.0546, | |
| "step": 2282 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 3.991746447636613e-06, | |
| "loss": 2.1, | |
| "step": 2284 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 3.975556688015069e-06, | |
| "loss": 2.1006, | |
| "step": 2286 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 3.959391677135194e-06, | |
| "loss": 2.1388, | |
| "step": 2288 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 3.94325148140374e-06, | |
| "loss": 1.9756, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 3.927136167125514e-06, | |
| "loss": 2.1165, | |
| "step": 2292 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 3.911045800503101e-06, | |
| "loss": 2.0969, | |
| "step": 2294 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 3.894980447636606e-06, | |
| "loss": 2.0533, | |
| "step": 2296 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 3.878940174523371e-06, | |
| "loss": 2.0705, | |
| "step": 2298 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 3.862925047057721e-06, | |
| "loss": 1.9904, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 3.846935131030667e-06, | |
| "loss": 2.0586, | |
| "step": 2302 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 3.830970492129654e-06, | |
| "loss": 2.1269, | |
| "step": 2304 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 3.815031195938295e-06, | |
| "loss": 2.0169, | |
| "step": 2306 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 3.7991173079360845e-06, | |
| "loss": 2.0113, | |
| "step": 2308 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 3.7832288934981453e-06, | |
| "loss": 2.1444, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 3.7673660178949445e-06, | |
| "loss": 2.0693, | |
| "step": 2312 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 3.751528746292048e-06, | |
| "loss": 2.0509, | |
| "step": 2314 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 3.7357171437498276e-06, | |
| "loss": 2.0359, | |
| "step": 2316 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 3.7199312752232053e-06, | |
| "loss": 2.1342, | |
| "step": 2318 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 3.7041712055613966e-06, | |
| "loss": 2.1406, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 3.6884369995076232e-06, | |
| "loss": 2.1445, | |
| "step": 2322 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 3.6727287216988626e-06, | |
| "loss": 2.0229, | |
| "step": 2324 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 3.6570464366655724e-06, | |
| "loss": 2.0803, | |
| "step": 2326 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 3.6413902088314425e-06, | |
| "loss": 2.0108, | |
| "step": 2328 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 3.625760102513103e-06, | |
| "loss": 2.0553, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 3.610156181919887e-06, | |
| "loss": 2.0145, | |
| "step": 2332 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 3.5945785111535547e-06, | |
| "loss": 2.0302, | |
| "step": 2334 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 3.5790271542080245e-06, | |
| "loss": 2.0322, | |
| "step": 2336 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 3.563502174969117e-06, | |
| "loss": 1.986, | |
| "step": 2338 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 3.548003637214291e-06, | |
| "loss": 2.1194, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 3.5325316046123925e-06, | |
| "loss": 2.0111, | |
| "step": 2342 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 3.5170861407233702e-06, | |
| "loss": 2.0528, | |
| "step": 2344 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 3.5016673089980304e-06, | |
| "loss": 2.1059, | |
| "step": 2346 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 3.48627517277778e-06, | |
| "loss": 1.9447, | |
| "step": 2348 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 3.4709097952943483e-06, | |
| "loss": 2.0537, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 3.4555712396695494e-06, | |
| "loss": 2.0522, | |
| "step": 2352 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 3.4402595689150008e-06, | |
| "loss": 2.1326, | |
| "step": 2354 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 3.424974845931888e-06, | |
| "loss": 2.1185, | |
| "step": 2356 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 3.409717133510683e-06, | |
| "loss": 2.0154, | |
| "step": 2358 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 3.3944864943308953e-06, | |
| "loss": 2.1498, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 3.3792829909608305e-06, | |
| "loss": 2.1767, | |
| "step": 2362 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 3.3641066858573046e-06, | |
| "loss": 2.1465, | |
| "step": 2364 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 3.348957641365408e-06, | |
| "loss": 2.0238, | |
| "step": 2366 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 3.3338359197182366e-06, | |
| "loss": 2.0366, | |
| "step": 2368 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 3.318741583036655e-06, | |
| "loss": 2.0279, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 3.3036746933290166e-06, | |
| "loss": 2.0918, | |
| "step": 2372 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 3.2886353124909267e-06, | |
| "loss": 2.0654, | |
| "step": 2374 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 3.2736235023049867e-06, | |
| "loss": 2.1704, | |
| "step": 2376 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 3.258639324440527e-06, | |
| "loss": 2.0417, | |
| "step": 2378 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 3.2436828404533684e-06, | |
| "loss": 2.0905, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 3.228754111785559e-06, | |
| "loss": 2.0462, | |
| "step": 2382 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 3.213853199765139e-06, | |
| "loss": 2.0241, | |
| "step": 2384 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 3.1989801656058604e-06, | |
| "loss": 2.0508, | |
| "step": 2386 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 3.1841350704069583e-06, | |
| "loss": 2.1177, | |
| "step": 2388 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 3.1693179751529e-06, | |
| "loss": 2.0236, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 3.1545289407131128e-06, | |
| "loss": 2.0454, | |
| "step": 2392 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 3.139768027841764e-06, | |
| "loss": 2.0873, | |
| "step": 2394 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 3.1250352971774824e-06, | |
| "loss": 2.0885, | |
| "step": 2396 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 3.110330809243134e-06, | |
| "loss": 2.092, | |
| "step": 2398 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 3.0956546244455567e-06, | |
| "loss": 2.0555, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 3.0810068030753114e-06, | |
| "loss": 2.1797, | |
| "step": 2402 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 3.066387405306457e-06, | |
| "loss": 2.1085, | |
| "step": 2404 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 3.0517964911962717e-06, | |
| "loss": 2.0191, | |
| "step": 2406 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 3.0372341206850265e-06, | |
| "loss": 2.0661, | |
| "step": 2408 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 3.02270035359573e-06, | |
| "loss": 2.0718, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 3.0081952496338973e-06, | |
| "loss": 2.0744, | |
| "step": 2412 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 2.9937188683872796e-06, | |
| "loss": 2.0369, | |
| "step": 2414 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 2.9792712693256442e-06, | |
| "loss": 2.047, | |
| "step": 2416 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 2.964852511800519e-06, | |
| "loss": 1.9941, | |
| "step": 2418 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 2.9504626550449433e-06, | |
| "loss": 2.0338, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 2.936101758173233e-06, | |
| "loss": 2.0252, | |
| "step": 2422 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 2.9217698801807324e-06, | |
| "loss": 2.0699, | |
| "step": 2424 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 2.9074670799435843e-06, | |
| "loss": 2.0276, | |
| "step": 2426 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 2.893193416218469e-06, | |
| "loss": 2.0658, | |
| "step": 2428 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 2.878948947642376e-06, | |
| "loss": 2.0029, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 2.8647337327323543e-06, | |
| "loss": 1.9765, | |
| "step": 2432 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 2.8505478298852873e-06, | |
| "loss": 2.0633, | |
| "step": 2434 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 2.8363912973776396e-06, | |
| "loss": 2.0589, | |
| "step": 2436 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 2.822264193365212e-06, | |
| "loss": 2.0214, | |
| "step": 2438 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 2.8081665758829256e-06, | |
| "loss": 2.1078, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 2.794098502844558e-06, | |
| "loss": 2.1353, | |
| "step": 2442 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 2.78006003204252e-06, | |
| "loss": 2.0515, | |
| "step": 2444 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 2.7660512211476133e-06, | |
| "loss": 2.0261, | |
| "step": 2446 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 2.7520721277088023e-06, | |
| "loss": 2.0986, | |
| "step": 2448 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 2.738122809152962e-06, | |
| "loss": 2.0835, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 2.724203322784652e-06, | |
| "loss": 1.986, | |
| "step": 2452 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 2.7103137257858867e-06, | |
| "loss": 2.0372, | |
| "step": 2454 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 2.696454075215882e-06, | |
| "loss": 2.0028, | |
| "step": 2456 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 2.6826244280108438e-06, | |
| "loss": 2.0997, | |
| "step": 2458 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 2.6688248409837124e-06, | |
| "loss": 2.0516, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 2.655055370823948e-06, | |
| "loss": 1.9923, | |
| "step": 2462 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 2.6413160740972844e-06, | |
| "loss": 2.1519, | |
| "step": 2464 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 2.627607007245496e-06, | |
| "loss": 2.1369, | |
| "step": 2466 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 2.6139282265861855e-06, | |
| "loss": 2.0219, | |
| "step": 2468 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 2.600279788312524e-06, | |
| "loss": 2.0564, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 2.5866617484930434e-06, | |
| "loss": 2.0527, | |
| "step": 2472 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 2.5730741630713896e-06, | |
| "loss": 2.0417, | |
| "step": 2474 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 2.5595170878661078e-06, | |
| "loss": 2.1441, | |
| "step": 2476 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 2.545990578570404e-06, | |
| "loss": 2.0528, | |
| "step": 2478 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 2.5324946907519123e-06, | |
| "loss": 2.0411, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 2.5190294798524805e-06, | |
| "loss": 1.9793, | |
| "step": 2482 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 2.505595001187926e-06, | |
| "loss": 2.0225, | |
| "step": 2484 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 2.4921913099478203e-06, | |
| "loss": 2.0482, | |
| "step": 2486 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 2.478818461195255e-06, | |
| "loss": 2.0417, | |
| "step": 2488 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 2.465476509866628e-06, | |
| "loss": 2.0944, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 2.4521655107713993e-06, | |
| "loss": 2.01, | |
| "step": 2492 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 2.438885518591878e-06, | |
| "loss": 1.9519, | |
| "step": 2494 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 2.425636587883e-06, | |
| "loss": 1.9499, | |
| "step": 2496 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 2.4124187730720916e-06, | |
| "loss": 2.0696, | |
| "step": 2498 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 2.3992321284586606e-06, | |
| "loss": 2.0389, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 2.3860767082141567e-06, | |
| "loss": 1.9808, | |
| "step": 2502 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 2.372952566381771e-06, | |
| "loss": 2.0378, | |
| "step": 2504 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 2.3598597568761893e-06, | |
| "loss": 2.1372, | |
| "step": 2506 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 2.3467983334833855e-06, | |
| "loss": 2.0144, | |
| "step": 2508 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 2.3337683498604045e-06, | |
| "loss": 2.1156, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 2.3207698595351257e-06, | |
| "loss": 2.0175, | |
| "step": 2512 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 2.307802915906058e-06, | |
| "loss": 2.0258, | |
| "step": 2514 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 2.2948675722421086e-06, | |
| "loss": 2.0199, | |
| "step": 2516 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 2.2819638816823796e-06, | |
| "loss": 2.0457, | |
| "step": 2518 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 2.2690918972359357e-06, | |
| "loss": 2.033, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 2.2562516717815874e-06, | |
| "loss": 2.1335, | |
| "step": 2522 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 2.2434432580676855e-06, | |
| "loss": 1.9865, | |
| "step": 2524 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 2.2306667087118906e-06, | |
| "loss": 2.043, | |
| "step": 2526 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 2.2179220762009624e-06, | |
| "loss": 2.0357, | |
| "step": 2528 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 2.2052094128905454e-06, | |
| "loss": 2.0383, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 2.192528771004958e-06, | |
| "loss": 1.9598, | |
| "step": 2532 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 2.179880202636966e-06, | |
| "loss": 2.0291, | |
| "step": 2534 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 2.1672637597475764e-06, | |
| "loss": 1.9555, | |
| "step": 2536 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 2.154679494165829e-06, | |
| "loss": 2.106, | |
| "step": 2538 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 2.1421274575885674e-06, | |
| "loss": 2.0234, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 2.1296077015802487e-06, | |
| "loss": 2.1185, | |
| "step": 2542 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 2.1171202775727084e-06, | |
| "loss": 2.1265, | |
| "step": 2544 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 2.104665236864969e-06, | |
| "loss": 1.9851, | |
| "step": 2546 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 2.092242630623016e-06, | |
| "loss": 1.959, | |
| "step": 2548 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 2.0798525098795897e-06, | |
| "loss": 1.8975, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 2.067494925533987e-06, | |
| "loss": 1.9805, | |
| "step": 2552 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 2.0551699283518355e-06, | |
| "loss": 2.0533, | |
| "step": 2554 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 2.042877568964897e-06, | |
| "loss": 2.0465, | |
| "step": 2556 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 2.030617897870851e-06, | |
| "loss": 2.041, | |
| "step": 2558 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 2.018390965433098e-06, | |
| "loss": 2.0021, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 2.006196821880544e-06, | |
| "loss": 2.1347, | |
| "step": 2562 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 1.9940355173073924e-06, | |
| "loss": 2.0206, | |
| "step": 2564 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 1.9819071016729506e-06, | |
| "loss": 2.1164, | |
| "step": 2566 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 1.9698116248014088e-06, | |
| "loss": 2.0501, | |
| "step": 2568 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 1.957749136381645e-06, | |
| "loss": 2.058, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 1.945719685967017e-06, | |
| "loss": 2.0223, | |
| "step": 2572 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 1.9337233229751686e-06, | |
| "loss": 2.0018, | |
| "step": 2574 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 1.9217600966878103e-06, | |
| "loss": 2.0471, | |
| "step": 2576 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 1.9098300562505266e-06, | |
| "loss": 2.0002, | |
| "step": 2578 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 1.8979332506725778e-06, | |
| "loss": 2.0919, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 1.886069728826686e-06, | |
| "loss": 2.0476, | |
| "step": 2582 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 1.874239539448851e-06, | |
| "loss": 2.0665, | |
| "step": 2584 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 1.86244273113813e-06, | |
| "loss": 2.0076, | |
| "step": 2586 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 1.8506793523564615e-06, | |
| "loss": 1.9439, | |
| "step": 2588 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 1.8389494514284422e-06, | |
| "loss": 2.0203, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 1.8272530765411445e-06, | |
| "loss": 2.0125, | |
| "step": 2592 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 1.8155902757439181e-06, | |
| "loss": 1.9952, | |
| "step": 2594 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 1.8039610969481836e-06, | |
| "loss": 1.9731, | |
| "step": 2596 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 1.7923655879272395e-06, | |
| "loss": 2.0941, | |
| "step": 2598 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 1.7808037963160684e-06, | |
| "loss": 2.1002, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 1.769275769611143e-06, | |
| "loss": 2.0731, | |
| "step": 2602 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 1.757781555170227e-06, | |
| "loss": 1.9957, | |
| "step": 2604 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 1.7463212002121766e-06, | |
| "loss": 1.9375, | |
| "step": 2606 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 1.7348947518167602e-06, | |
| "loss": 1.9836, | |
| "step": 2608 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 1.7235022569244475e-06, | |
| "loss": 2.1135, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 1.712143762336228e-06, | |
| "loss": 2.0643, | |
| "step": 2612 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 1.7008193147134167e-06, | |
| "loss": 1.9972, | |
| "step": 2614 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 1.6895289605774657e-06, | |
| "loss": 1.9943, | |
| "step": 2616 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 1.6782727463097626e-06, | |
| "loss": 2.064, | |
| "step": 2618 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 1.6670507181514483e-06, | |
| "loss": 2.041, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 1.6558629222032297e-06, | |
| "loss": 2.0379, | |
| "step": 2622 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 1.64470940442518e-06, | |
| "loss": 1.9874, | |
| "step": 2624 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 1.633590210636561e-06, | |
| "loss": 2.0053, | |
| "step": 2626 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 1.6225053865156237e-06, | |
| "loss": 2.036, | |
| "step": 2628 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 1.6114549775994348e-06, | |
| "loss": 2.005, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 1.6004390292836747e-06, | |
| "loss": 2.0534, | |
| "step": 2632 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 1.589457586822456e-06, | |
| "loss": 1.9888, | |
| "step": 2634 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 1.5785106953281492e-06, | |
| "loss": 2.0461, | |
| "step": 2636 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 1.5675983997711797e-06, | |
| "loss": 2.1264, | |
| "step": 2638 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 1.5567207449798517e-06, | |
| "loss": 2.0621, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 1.5458777756401621e-06, | |
| "loss": 2.0534, | |
| "step": 2642 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 1.5350695362956236e-06, | |
| "loss": 2.0692, | |
| "step": 2644 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 1.5242960713470734e-06, | |
| "loss": 1.9893, | |
| "step": 2646 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 1.5135574250524898e-06, | |
| "loss": 2.0268, | |
| "step": 2648 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 1.5028536415268214e-06, | |
| "loss": 1.9606, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 1.492184764741793e-06, | |
| "loss": 2.0561, | |
| "step": 2652 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 1.4815508385257316e-06, | |
| "loss": 2.0315, | |
| "step": 2654 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 1.4709519065633836e-06, | |
| "loss": 1.9875, | |
| "step": 2656 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 1.4603880123957448e-06, | |
| "loss": 2.023, | |
| "step": 2658 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 1.4498591994198664e-06, | |
| "loss": 2.0014, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 1.4393655108886839e-06, | |
| "loss": 2.0192, | |
| "step": 2662 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 1.4289069899108455e-06, | |
| "loss": 1.9682, | |
| "step": 2664 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 1.4184836794505218e-06, | |
| "loss": 2.0563, | |
| "step": 2666 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 1.4080956223272458e-06, | |
| "loss": 1.9578, | |
| "step": 2668 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 1.3977428612157185e-06, | |
| "loss": 1.9601, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 1.3874254386456498e-06, | |
| "loss": 1.9455, | |
| "step": 2672 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 1.3771433970015735e-06, | |
| "loss": 2.0491, | |
| "step": 2674 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 1.366896778522674e-06, | |
| "loss": 2.028, | |
| "step": 2676 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 1.356685625302625e-06, | |
| "loss": 1.956, | |
| "step": 2678 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 1.3465099792893977e-06, | |
| "loss": 2.0295, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 1.3363698822851001e-06, | |
| "loss": 2.0952, | |
| "step": 2682 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 1.3262653759458033e-06, | |
| "loss": 2.0553, | |
| "step": 2684 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 1.3161965017813704e-06, | |
| "loss": 2.0049, | |
| "step": 2686 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 1.3061633011552887e-06, | |
| "loss": 2.0425, | |
| "step": 2688 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 1.2961658152844902e-06, | |
| "loss": 2.0864, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 1.2862040852391955e-06, | |
| "loss": 2.0699, | |
| "step": 2692 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 1.2762781519427324e-06, | |
| "loss": 2.0979, | |
| "step": 2694 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 1.2663880561713771e-06, | |
| "loss": 2.0282, | |
| "step": 2696 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 1.2565338385541792e-06, | |
| "loss": 2.0823, | |
| "step": 2698 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 1.246715539572807e-06, | |
| "loss": 2.0737, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 1.2369331995613664e-06, | |
| "loss": 2.0148, | |
| "step": 2702 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 1.2271868587062407e-06, | |
| "loss": 2.0891, | |
| "step": 2704 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 1.2174765570459335e-06, | |
| "loss": 2.0245, | |
| "step": 2706 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 1.2078023344708911e-06, | |
| "loss": 2.0272, | |
| "step": 2708 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 1.198164230723351e-06, | |
| "loss": 2.04, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 1.1885622853971634e-06, | |
| "loss": 2.0441, | |
| "step": 2712 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 1.1789965379376488e-06, | |
| "loss": 2.0528, | |
| "step": 2714 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 1.1694670276414177e-06, | |
| "loss": 1.9952, | |
| "step": 2716 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 1.159973793656215e-06, | |
| "loss": 2.0524, | |
| "step": 2718 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 1.1505168749807693e-06, | |
| "loss": 2.0553, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 1.1410963104646144e-06, | |
| "loss": 2.0503, | |
| "step": 2722 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 1.1317121388079433e-06, | |
| "loss": 2.0251, | |
| "step": 2724 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 1.122364398561444e-06, | |
| "loss": 2.0251, | |
| "step": 2726 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 1.1130531281261448e-06, | |
| "loss": 2.0333, | |
| "step": 2728 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 1.1037783657532542e-06, | |
| "loss": 2.0168, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 1.0945401495439977e-06, | |
| "loss": 2.0096, | |
| "step": 2732 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 1.0853385174494758e-06, | |
| "loss": 2.0703, | |
| "step": 2734 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 1.0761735072704937e-06, | |
| "loss": 2.0121, | |
| "step": 2736 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 1.0670451566574102e-06, | |
| "loss": 2.0479, | |
| "step": 2738 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 1.0579535031099886e-06, | |
| "loss": 1.9714, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 1.0488985839772404e-06, | |
| "loss": 2.0612, | |
| "step": 2742 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 1.0398804364572646e-06, | |
| "loss": 2.0864, | |
| "step": 2744 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 1.0308990975971012e-06, | |
| "loss": 2.0552, | |
| "step": 2746 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 1.0219546042925842e-06, | |
| "loss": 2.0607, | |
| "step": 2748 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 1.013046993288177e-06, | |
| "loss": 2.0969, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 1.0041763011768335e-06, | |
| "loss": 2.0123, | |
| "step": 2752 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 9.953425643998381e-07, | |
| "loss": 2.1062, | |
| "step": 2754 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 9.86545819246666e-07, | |
| "loss": 2.0423, | |
| "step": 2756 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 9.77786101854825e-07, | |
| "loss": 2.0219, | |
| "step": 2758 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 9.690634482097094e-07, | |
| "loss": 2.05, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 9.603778941444608e-07, | |
| "loss": 2.0129, | |
| "step": 2762 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 9.517294753398066e-07, | |
| "loss": 2.0824, | |
| "step": 2764 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 9.431182273239215e-07, | |
| "loss": 2.0186, | |
| "step": 2766 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 9.345441854722826e-07, | |
| "loss": 1.997, | |
| "step": 2768 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 9.260073850075213e-07, | |
| "loss": 2.0691, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 9.175078609992815e-07, | |
| "loss": 2.0398, | |
| "step": 2772 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 9.090456483640686e-07, | |
| "loss": 2.1227, | |
| "step": 2774 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 9.006207818651169e-07, | |
| "loss": 1.949, | |
| "step": 2776 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 8.92233296112236e-07, | |
| "loss": 2.0316, | |
| "step": 2778 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 8.838832255616736e-07, | |
| "loss": 1.9812, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 8.755706045159751e-07, | |
| "loss": 2.031, | |
| "step": 2782 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 8.67295467123842e-07, | |
| "loss": 2.0845, | |
| "step": 2784 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 8.590578473799905e-07, | |
| "loss": 2.1078, | |
| "step": 2786 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 8.508577791250095e-07, | |
| "loss": 2.0724, | |
| "step": 2788 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 8.426952960452284e-07, | |
| "loss": 2.0203, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 8.345704316725722e-07, | |
| "loss": 2.0057, | |
| "step": 2792 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 8.264832193844274e-07, | |
| "loss": 2.0654, | |
| "step": 2794 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 8.184336924035019e-07, | |
| "loss": 1.9804, | |
| "step": 2796 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 8.10421883797694e-07, | |
| "loss": 2.0329, | |
| "step": 2798 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 8.024478264799484e-07, | |
| "loss": 2.1293, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 7.945115532081271e-07, | |
| "loss": 1.9608, | |
| "step": 2802 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 7.866130965848751e-07, | |
| "loss": 1.9512, | |
| "step": 2804 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 7.787524890574827e-07, | |
| "loss": 2.1307, | |
| "step": 2806 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 7.709297629177525e-07, | |
| "loss": 2.041, | |
| "step": 2808 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 7.631449503018706e-07, | |
| "loss": 2.0908, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 7.553980831902719e-07, | |
| "loss": 2.1424, | |
| "step": 2812 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 7.476891934075126e-07, | |
| "loss": 1.9644, | |
| "step": 2814 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 7.400183126221261e-07, | |
| "loss": 2.0027, | |
| "step": 2816 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 7.32385472346514e-07, | |
| "loss": 2.0594, | |
| "step": 2818 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 7.247907039367996e-07, | |
| "loss": 1.9965, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 7.172340385927045e-07, | |
| "loss": 1.9674, | |
| "step": 2822 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 7.097155073574203e-07, | |
| "loss": 1.97, | |
| "step": 2824 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 7.022351411174866e-07, | |
| "loss": 1.9945, | |
| "step": 2826 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 6.947929706026546e-07, | |
| "loss": 2.0788, | |
| "step": 2828 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 6.873890263857663e-07, | |
| "loss": 2.0386, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 6.800233388826294e-07, | |
| "loss": 1.9913, | |
| "step": 2832 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 6.726959383518871e-07, | |
| "loss": 2.0203, | |
| "step": 2834 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 6.654068548949044e-07, | |
| "loss": 1.9785, | |
| "step": 2836 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 6.581561184556296e-07, | |
| "loss": 1.9909, | |
| "step": 2838 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 6.50943758820487e-07, | |
| "loss": 1.9809, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 6.437698056182429e-07, | |
| "loss": 2.1202, | |
| "step": 2842 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 6.36634288319885e-07, | |
| "loss": 2.0978, | |
| "step": 2844 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 6.295372362385111e-07, | |
| "loss": 2.0348, | |
| "step": 2846 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 6.22478678529197e-07, | |
| "loss": 2.0039, | |
| "step": 2848 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 6.154586441888821e-07, | |
| "loss": 2.0542, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 6.084771620562491e-07, | |
| "loss": 2.0436, | |
| "step": 2852 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 6.015342608116092e-07, | |
| "loss": 2.0156, | |
| "step": 2854 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 5.946299689767798e-07, | |
| "loss": 2.0565, | |
| "step": 2856 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 5.877643149149669e-07, | |
| "loss": 1.9717, | |
| "step": 2858 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 5.80937326830654e-07, | |
| "loss": 2.0914, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 5.741490327694787e-07, | |
| "loss": 2.027, | |
| "step": 2862 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 5.673994606181243e-07, | |
| "loss": 2.0977, | |
| "step": 2864 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 5.606886381041976e-07, | |
| "loss": 2.0128, | |
| "step": 2866 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 5.540165927961294e-07, | |
| "loss": 2.0213, | |
| "step": 2868 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 5.473833521030414e-07, | |
| "loss": 2.0594, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 5.407889432746483e-07, | |
| "loss": 1.9884, | |
| "step": 2872 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 5.342333934011435e-07, | |
| "loss": 2.0562, | |
| "step": 2874 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 5.277167294130802e-07, | |
| "loss": 2.0235, | |
| "step": 2876 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 5.212389780812733e-07, | |
| "loss": 1.9854, | |
| "step": 2878 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 5.148001660166757e-07, | |
| "loss": 1.9631, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 5.084003196702802e-07, | |
| "loss": 2.1452, | |
| "step": 2882 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 5.02039465333004e-07, | |
| "loss": 2.1024, | |
| "step": 2884 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 4.957176291355815e-07, | |
| "loss": 2.0596, | |
| "step": 2886 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 4.894348370484648e-07, | |
| "loss": 2.0524, | |
| "step": 2888 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 4.831911148817036e-07, | |
| "loss": 1.9454, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 4.769864882848496e-07, | |
| "loss": 2.0056, | |
| "step": 2892 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 4.708209827468457e-07, | |
| "loss": 2.0453, | |
| "step": 2894 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 4.646946235959271e-07, | |
| "loss": 1.9995, | |
| "step": 2896 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 4.5860743599951186e-07, | |
| "loss": 2.0489, | |
| "step": 2898 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 4.525594449640958e-07, | |
| "loss": 1.9782, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 4.4655067533515897e-07, | |
| "loss": 2.0753, | |
| "step": 2902 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 4.4058115179705264e-07, | |
| "loss": 1.9671, | |
| "step": 2904 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 4.346508988729059e-07, | |
| "loss": 2.0817, | |
| "step": 2906 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 4.2875994092452024e-07, | |
| "loss": 2.0236, | |
| "step": 2908 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 4.2290830215227397e-07, | |
| "loss": 2.0448, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 4.1709600659501694e-07, | |
| "loss": 1.934, | |
| "step": 2912 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 4.113230781299748e-07, | |
| "loss": 2.0565, | |
| "step": 2914 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 4.055895404726562e-07, | |
| "loss": 2.0337, | |
| "step": 2916 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 3.998954171767422e-07, | |
| "loss": 2.0055, | |
| "step": 2918 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 3.942407316340069e-07, | |
| "loss": 2.043, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 3.886255070742018e-07, | |
| "loss": 2.0616, | |
| "step": 2922 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 3.8304976656498016e-07, | |
| "loss": 2.111, | |
| "step": 2924 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 3.775135330117852e-07, | |
| "loss": 2.0075, | |
| "step": 2926 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 3.720168291577653e-07, | |
| "loss": 2.0582, | |
| "step": 2928 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 3.6655967758368327e-07, | |
| "loss": 1.9975, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 3.611421007078142e-07, | |
| "loss": 2.1098, | |
| "step": 2932 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 3.5576412078585755e-07, | |
| "loss": 2.0389, | |
| "step": 2934 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 3.504257599108485e-07, | |
| "loss": 1.9881, | |
| "step": 2936 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 3.451270400130646e-07, | |
| "loss": 2.0085, | |
| "step": 2938 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 3.3986798285993937e-07, | |
| "loss": 2.1398, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 3.346486100559632e-07, | |
| "loss": 1.9635, | |
| "step": 2942 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 3.2946894304260346e-07, | |
| "loss": 2.0553, | |
| "step": 2944 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 3.2432900309821715e-07, | |
| "loss": 2.066, | |
| "step": 2946 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 3.1922881133795827e-07, | |
| "loss": 2.0327, | |
| "step": 2948 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 3.1416838871368925e-07, | |
| "loss": 2.093, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 3.091477560139056e-07, | |
| "loss": 1.9929, | |
| "step": 2952 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 3.041669338636388e-07, | |
| "loss": 2.0542, | |
| "step": 2954 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 2.992259427243771e-07, | |
| "loss": 1.9884, | |
| "step": 2956 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 2.943248028939838e-07, | |
| "loss": 2.0087, | |
| "step": 2958 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 2.894635345066055e-07, | |
| "loss": 2.0508, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 2.8464215753260196e-07, | |
| "loss": 2.0834, | |
| "step": 2962 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 2.7986069177844857e-07, | |
| "loss": 1.9348, | |
| "step": 2964 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 2.751191568866707e-07, | |
| "loss": 2.0879, | |
| "step": 2966 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 2.704175723357505e-07, | |
| "loss": 2.0293, | |
| "step": 2968 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 2.6575595744005476e-07, | |
| "loss": 2.0565, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 2.6113433134974943e-07, | |
| "loss": 2.0174, | |
| "step": 2972 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 2.565527130507295e-07, | |
| "loss": 2.0263, | |
| "step": 2974 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 2.5201112136453046e-07, | |
| "loss": 2.0199, | |
| "step": 2976 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 2.4750957494826033e-07, | |
| "loss": 1.9831, | |
| "step": 2978 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 2.4304809229451554e-07, | |
| "loss": 2.0736, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 2.386266917313118e-07, | |
| "loss": 2.093, | |
| "step": 2982 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 2.342453914220033e-07, | |
| "loss": 1.9708, | |
| "step": 2984 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 2.299042093652093e-07, | |
| "loss": 2.052, | |
| "step": 2986 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 2.256031633947453e-07, | |
| "loss": 2.0435, | |
| "step": 2988 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 2.2134227117953988e-07, | |
| "loss": 2.0799, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 2.171215502235713e-07, | |
| "loss": 2.0651, | |
| "step": 2992 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 2.1294101786579312e-07, | |
| "loss": 1.9384, | |
| "step": 2994 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 2.0880069128005885e-07, | |
| "loss": 2.0493, | |
| "step": 2996 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 2.0470058747505516e-07, | |
| "loss": 2.0455, | |
| "step": 2998 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 2.0064072329423424e-07, | |
| "loss": 2.0212, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 1.9662111541573713e-07, | |
| "loss": 2.0504, | |
| "step": 3002 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 1.9264178035233395e-07, | |
| "loss": 1.9623, | |
| "step": 3004 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 1.8870273445134813e-07, | |
| "loss": 2.1144, | |
| "step": 3006 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 1.8480399389459892e-07, | |
| "loss": 1.9515, | |
| "step": 3008 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 1.809455746983213e-07, | |
| "loss": 2.0845, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 1.7712749271311392e-07, | |
| "loss": 2.0293, | |
| "step": 3012 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 1.7334976362386458e-07, | |
| "loss": 2.0166, | |
| "step": 3014 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 1.696124029496904e-07, | |
| "loss": 2.0665, | |
| "step": 3016 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 1.6591542604387445e-07, | |
| "loss": 2.0293, | |
| "step": 3018 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 1.62258848093797e-07, | |
| "loss": 2.0101, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 1.5864268412088102e-07, | |
| "loss": 2.0781, | |
| "step": 3022 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 1.5506694898052676e-07, | |
| "loss": 2.0296, | |
| "step": 3024 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 1.5153165736204733e-07, | |
| "loss": 2.0974, | |
| "step": 3026 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 1.4803682378861317e-07, | |
| "loss": 2.0184, | |
| "step": 3028 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 1.4458246261719212e-07, | |
| "loss": 2.0428, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 1.4116858803848944e-07, | |
| "loss": 1.9731, | |
| "step": 3032 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 1.3779521407688458e-07, | |
| "loss": 2.0528, | |
| "step": 3034 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 1.3446235459038558e-07, | |
| "loss": 2.0129, | |
| "step": 3036 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 1.3117002327055927e-07, | |
| "loss": 2.0003, | |
| "step": 3038 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 1.2791823364248113e-07, | |
| "loss": 1.992, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 1.2470699906468097e-07, | |
| "loss": 2.1003, | |
| "step": 3042 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 1.215363327290875e-07, | |
| "loss": 2.0339, | |
| "step": 3044 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 1.1840624766096931e-07, | |
| "loss": 2.0067, | |
| "step": 3046 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 1.1531675671888621e-07, | |
| "loss": 2.0467, | |
| "step": 3048 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 1.1226787259463579e-07, | |
| "loss": 1.9228, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 1.1075866200977226e-07, | |
| "loss": 2.0907, | |
| "step": 3052 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 1.0777071154447039e-07, | |
| "loss": 2.0281, | |
| "step": 3054 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 1.0482339889656524e-07, | |
| "loss": 2.0358, | |
| "step": 3056 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 1.019167361737805e-07, | |
| "loss": 2.0999, | |
| "step": 3058 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 9.90507353168424e-08, | |
| "loss": 2.0524, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 9.62254080994407e-08, | |
| "loss": 1.9762, | |
| "step": 3062 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 9.344076612817221e-08, | |
| "loss": 2.0535, | |
| "step": 3064 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 9.069682084249964e-08, | |
| "loss": 2.048, | |
| "step": 3066 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 8.799358351469945e-08, | |
| "loss": 2.041, | |
| "step": 3068 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 8.53310652498185e-08, | |
| "loss": 2.0062, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 8.270927698562526e-08, | |
| "loss": 2.0123, | |
| "step": 3072 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 8.012822949256981e-08, | |
| "loss": 2.0708, | |
| "step": 3074 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 7.758793337373726e-08, | |
| "loss": 2.0431, | |
| "step": 3076 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 7.508839906479992e-08, | |
| "loss": 2.0011, | |
| "step": 3078 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 7.262963683398295e-08, | |
| "loss": 2.0622, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 7.021165678201658e-08, | |
| "loss": 2.1123, | |
| "step": 3082 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 6.783446884209177e-08, | |
| "loss": 2.1091, | |
| "step": 3084 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 6.549808277982905e-08, | |
| "loss": 2.1527, | |
| "step": 3086 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 6.32025081932286e-08, | |
| "loss": 2.0478, | |
| "step": 3088 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 6.094775451263579e-08, | |
| "loss": 1.9655, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 5.8733831000701294e-08, | |
| "loss": 2.0045, | |
| "step": 3092 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 5.656074675234546e-08, | |
| "loss": 2.0285, | |
| "step": 3094 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 5.442851069471511e-08, | |
| "loss": 2.0561, | |
| "step": 3096 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 5.233713158715348e-08, | |
| "loss": 2.0212, | |
| "step": 3098 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 5.02866180211592e-08, | |
| "loss": 2.0179, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 4.827697842035295e-08, | |
| "loss": 1.9988, | |
| "step": 3102 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 4.630822104044641e-08, | |
| "loss": 2.0096, | |
| "step": 3104 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 4.438035396920004e-08, | |
| "loss": 1.9862, | |
| "step": 3106 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 4.2493385126397554e-08, | |
| "loss": 1.9181, | |
| "step": 3108 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 4.0647322263810407e-08, | |
| "loss": 2.0177, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 3.884217296516446e-08, | |
| "loss": 1.9703, | |
| "step": 3112 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 3.707794464611225e-08, | |
| "loss": 2.0543, | |
| "step": 3114 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 3.535464455419968e-08, | |
| "loss": 1.9482, | |
| "step": 3116 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 3.3672279768838245e-08, | |
| "loss": 1.9763, | |
| "step": 3118 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 3.203085720127286e-08, | |
| "loss": 1.9819, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 3.043038359455741e-08, | |
| "loss": 1.9492, | |
| "step": 3122 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 2.8870865523525916e-08, | |
| "loss": 2.0664, | |
| "step": 3124 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 2.735230939476141e-08, | |
| "loss": 2.0837, | |
| "step": 3126 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 2.58747214465771e-08, | |
| "loss": 2.0272, | |
| "step": 3128 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 2.4438107748985252e-08, | |
| "loss": 2.034, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 2.304247420367167e-08, | |
| "loss": 1.9772, | |
| "step": 3132 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 2.1687826543975722e-08, | |
| "loss": 2.1161, | |
| "step": 3134 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 2.0374170334863662e-08, | |
| "loss": 2.0885, | |
| "step": 3136 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 1.9101510972907576e-08, | |
| "loss": 2.1047, | |
| "step": 3138 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 1.786985368626204e-08, | |
| "loss": 1.9559, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 1.6679203534639697e-08, | |
| "loss": 2.0716, | |
| "step": 3142 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 1.5529565409299063e-08, | |
| "loss": 2.0195, | |
| "step": 3144 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 1.4420944033014528e-08, | |
| "loss": 2.0242, | |
| "step": 3146 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 1.3353343960064158e-08, | |
| "loss": 2.0068, | |
| "step": 3148 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 1.2326769576206376e-08, | |
| "loss": 2.0547, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 1.1341225098665532e-08, | |
| "loss": 2.0266, | |
| "step": 3152 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 1.0396714576113021e-08, | |
| "loss": 1.9689, | |
| "step": 3154 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 9.493241888647309e-09, | |
| "loss": 2.0797, | |
| "step": 3156 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 8.630810747785045e-09, | |
| "loss": 2.0774, | |
| "step": 3158 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 7.80942469643886e-09, | |
| "loss": 2.0255, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 7.029087108909594e-09, | |
| "loss": 2.0273, | |
| "step": 3162 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 6.289801190865197e-09, | |
| "loss": 2.0782, | |
| "step": 3164 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 5.591569979332967e-09, | |
| "loss": 2.0105, | |
| "step": 3166 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 4.9343963426840006e-09, | |
| "loss": 2.0279, | |
| "step": 3168 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 4.318282980625421e-09, | |
| "loss": 2.0475, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 3.743232424182619e-09, | |
| "loss": 2.0418, | |
| "step": 3172 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 3.209247035694807e-09, | |
| "loss": 2.0356, | |
| "step": 3174 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 2.7163290088039198e-09, | |
| "loss": 1.9764, | |
| "step": 3176 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 2.2644803684435114e-09, | |
| "loss": 2.0566, | |
| "step": 3178 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 1.8537029708332045e-09, | |
| "loss": 2.0398, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 1.4839985034675874e-09, | |
| "loss": 2.1187, | |
| "step": 3182 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 1.1553684851139947e-09, | |
| "loss": 1.9731, | |
| "step": 3184 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 8.678142658014033e-10, | |
| "loss": 1.996, | |
| "step": 3186 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 6.213370268204344e-10, | |
| "loss": 2.0838, | |
| "step": 3188 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 4.1593778071002955e-10, | |
| "loss": 1.9909, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 2.516173712630021e-10, | |
| "loss": 2.064, | |
| "step": 3192 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 1.2837647351715554e-10, | |
| "loss": 1.9744, | |
| "step": 3194 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 4.621559375195261e-11, | |
| "loss": 2.0026, | |
| "step": 3196 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "step": 3196, | |
| "total_flos": 1.1926290474572186e+17, | |
| "train_loss": 0.6364963831233142, | |
| "train_runtime": 18349.6257, | |
| "train_samples_per_second": 11.154, | |
| "train_steps_per_second": 0.174 | |
| } | |
| ], | |
| "logging_steps": 2, | |
| "max_steps": 3196, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 200, | |
| "total_flos": 1.1926290474572186e+17, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |