| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.9959587795514246, | |
| "eval_steps": 500, | |
| "global_step": 1236, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.0526315789473683e-07, | |
| "loss": 2.1959, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.5789473684210525e-07, | |
| "loss": 2.1258, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.631578947368421e-07, | |
| "loss": 2.1826, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.684210526315789e-07, | |
| "loss": 2.1599, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.7368421052631574e-07, | |
| "loss": 2.238, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 5.789473684210526e-07, | |
| "loss": 2.2262, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 6.842105263157895e-07, | |
| "loss": 2.1216, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 7.368421052631578e-07, | |
| "loss": 2.1524, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 8.421052631578947e-07, | |
| "loss": 2.1345, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.473684210526315e-07, | |
| "loss": 2.1243, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.0526315789473683e-06, | |
| "loss": 2.2935, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.1578947368421053e-06, | |
| "loss": 2.1534, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.263157894736842e-06, | |
| "loss": 2.2208, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.368421052631579e-06, | |
| "loss": 2.1108, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.4736842105263156e-06, | |
| "loss": 2.1267, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.5789473684210526e-06, | |
| "loss": 2.1529, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.6842105263157893e-06, | |
| "loss": 2.2229, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.7894736842105262e-06, | |
| "loss": 2.1263, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.894736842105263e-06, | |
| "loss": 2.1313, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2e-06, | |
| "loss": 2.0866, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9999862464405374e-06, | |
| "loss": 2.1889, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.999944986140471e-06, | |
| "loss": 2.2336, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9998762202347525e-06, | |
| "loss": 2.1225, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9997799506149334e-06, | |
| "loss": 2.1098, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.9996561799291147e-06, | |
| "loss": 2.183, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.9995049115818703e-06, | |
| "loss": 2.2008, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.9993261497341573e-06, | |
| "loss": 2.1518, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.9991198993031988e-06, | |
| "loss": 2.1438, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.99888616596235e-06, | |
| "loss": 2.2554, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.9986249561409413e-06, | |
| "loss": 2.1589, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.998336277024103e-06, | |
| "loss": 2.059, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.9980201365525657e-06, | |
| "loss": 2.2238, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.997676543422442e-06, | |
| "loss": 2.1098, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.997305507084991e-06, | |
| "loss": 2.1444, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.9969070377463515e-06, | |
| "loss": 2.1842, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.9964811463672683e-06, | |
| "loss": 2.2062, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.996027844662785e-06, | |
| "loss": 2.0729, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.995547145101926e-06, | |
| "loss": 2.1952, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.9950390609073516e-06, | |
| "loss": 2.1698, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.994503606054994e-06, | |
| "loss": 2.0591, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.9939407952736733e-06, | |
| "loss": 2.0508, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.993350644044693e-06, | |
| "loss": 2.0566, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.992733168601413e-06, | |
| "loss": 2.042, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.992088385928803e-06, | |
| "loss": 2.1241, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.991416313762978e-06, | |
| "loss": 2.2268, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.9907169705907057e-06, | |
| "loss": 2.123, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.9899903756489025e-06, | |
| "loss": 2.0804, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.989236548924102e-06, | |
| "loss": 2.1322, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.988455511151906e-06, | |
| "loss": 2.1986, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.987647283816412e-06, | |
| "loss": 2.0547, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.9868118891496263e-06, | |
| "loss": 2.1752, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.9859493501308496e-06, | |
| "loss": 2.1216, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.985059690486045e-06, | |
| "loss": 2.1593, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.984142934687186e-06, | |
| "loss": 2.016, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.9831991079515832e-06, | |
| "loss": 2.1072, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.9822282362411916e-06, | |
| "loss": 2.1336, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.981230346261894e-06, | |
| "loss": 2.0016, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.980205465462769e-06, | |
| "loss": 2.1503, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.9791536220353354e-06, | |
| "loss": 2.2006, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.978074844912774e-06, | |
| "loss": 2.048, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.9769691637691368e-06, | |
| "loss": 2.0971, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.9758366090185254e-06, | |
| "loss": 2.0509, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.9746772118142586e-06, | |
| "loss": 2.1021, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.9734910040480134e-06, | |
| "loss": 2.0307, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.9722780183489473e-06, | |
| "loss": 2.0414, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.9710382880828024e-06, | |
| "loss": 2.128, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.969771847350987e-06, | |
| "loss": 2.0941, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.968478730989636e-06, | |
| "loss": 2.1418, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.967158974568656e-06, | |
| "loss": 2.1382, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.9658126143907428e-06, | |
| "loss": 2.0852, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.964439687490386e-06, | |
| "loss": 2.1183, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.9630402316328505e-06, | |
| "loss": 1.9974, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.9616142853131337e-06, | |
| "loss": 2.0584, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.960161887754911e-06, | |
| "loss": 2.0791, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.9586830789094544e-06, | |
| "loss": 2.0694, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.9571778994545353e-06, | |
| "loss": 2.0454, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.9556463907933034e-06, | |
| "loss": 2.1072, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.9540885950531503e-06, | |
| "loss": 2.063, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.952504555084548e-06, | |
| "loss": 2.0433, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.9508943144598726e-06, | |
| "loss": 1.9556, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.949257917472204e-06, | |
| "loss": 1.9947, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.9475954091341094e-06, | |
| "loss": 2.0189, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.945906835176403e-06, | |
| "loss": 2.0664, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.9441922420468896e-06, | |
| "loss": 2.1918, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.9424516769090862e-06, | |
| "loss": 1.9924, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.9406851876409253e-06, | |
| "loss": 2.1053, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.938892822833437e-06, | |
| "loss": 2.1167, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.937074631789413e-06, | |
| "loss": 2.0673, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.9352306645220514e-06, | |
| "loss": 2.2079, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.9333609717535785e-06, | |
| "loss": 2.0725, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.9314656049138556e-06, | |
| "loss": 2.032, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.929544616138964e-06, | |
| "loss": 2.1356, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.9275980582697706e-06, | |
| "loss": 2.0379, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9256259848504737e-06, | |
| "loss": 2.0668, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9236284501271313e-06, | |
| "loss": 2.0651, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.921605509046169e-06, | |
| "loss": 2.1157, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.9195572172528676e-06, | |
| "loss": 2.098, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.9174836310898332e-06, | |
| "loss": 2.0669, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.9153848075954463e-06, | |
| "loss": 2.026, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.9132608045022953e-06, | |
| "loss": 2.0655, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.9121893788370484e-06, | |
| "loss": 1.9918, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.910027716109015e-06, | |
| "loss": 1.9431, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.90784102114037e-06, | |
| "loss": 2.0387, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.9056293540807919e-06, | |
| "loss": 2.1631, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.9033927757668691e-06, | |
| "loss": 2.0848, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.9011313477204271e-06, | |
| "loss": 2.0461, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.898845132146837e-06, | |
| "loss": 2.0879, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.8965341919333014e-06, | |
| "loss": 2.1088, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.8941985906471283e-06, | |
| "loss": 2.0245, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.89183839253398e-06, | |
| "loss": 1.9938, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.889453662516106e-06, | |
| "loss": 2.0197, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.8870444661905594e-06, | |
| "loss": 1.9899, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.8846108698273895e-06, | |
| "loss": 1.9828, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.8821529403678208e-06, | |
| "loss": 1.9724, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.8796707454224122e-06, | |
| "loss": 1.9798, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.8771643532691938e-06, | |
| "loss": 2.108, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.8746338328517938e-06, | |
| "loss": 2.0505, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.8720792537775378e-06, | |
| "loss": 1.9334, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.8695006863155363e-06, | |
| "loss": 2.0817, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.8668982013947508e-06, | |
| "loss": 2.0406, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.864271870602044e-06, | |
| "loss": 1.9474, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.8616217661802089e-06, | |
| "loss": 2.0711, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.8589479610259833e-06, | |
| "loss": 2.0697, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.8562505286880436e-06, | |
| "loss": 2.0769, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.8535295433649819e-06, | |
| "loss": 2.0401, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.850785079903265e-06, | |
| "loss": 2.1115, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.8480172137951753e-06, | |
| "loss": 2.0819, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.8452260211767357e-06, | |
| "loss": 2.0084, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.8424115788256134e-06, | |
| "loss": 2.0482, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.8395739641590081e-06, | |
| "loss": 2.0274, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.8367132552315249e-06, | |
| "loss": 1.9554, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.8338295307330239e-06, | |
| "loss": 2.0611, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.8309228699864581e-06, | |
| "loss": 2.0884, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.8279933529456902e-06, | |
| "loss": 2.0598, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.8250410601932935e-06, | |
| "loss": 2.1424, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.8220660729383364e-06, | |
| "loss": 2.0487, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.8190684730141467e-06, | |
| "loss": 1.9051, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.8160483428760625e-06, | |
| "loss": 2.0287, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.8130057655991622e-06, | |
| "loss": 2.0952, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.8099408248759808e-06, | |
| "loss": 2.0877, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.8068536050142079e-06, | |
| "loss": 1.9685, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.8037441909343665e-06, | |
| "loss": 1.9782, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.80061266816748e-06, | |
| "loss": 2.0603, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.7974591228527177e-06, | |
| "loss": 2.0778, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.794283641735025e-06, | |
| "loss": 2.0055, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.7910863121627394e-06, | |
| "loss": 2.0357, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.7878672220851852e-06, | |
| "loss": 2.0374, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.7846264600502558e-06, | |
| "loss": 2.0203, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.7813641152019786e-06, | |
| "loss": 2.1824, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.7780802772780605e-06, | |
| "loss": 1.9724, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.7747750366074228e-06, | |
| "loss": 1.9461, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.7714484841077123e-06, | |
| "loss": 2.0833, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.7681007112828059e-06, | |
| "loss": 1.8861, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.7647318102202874e-06, | |
| "loss": 1.9626, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.76134187358892e-06, | |
| "loss": 1.9566, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.7579309946360929e-06, | |
| "loss": 1.8845, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.7544992671852597e-06, | |
| "loss": 1.9537, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.7510467856333554e-06, | |
| "loss": 2.0696, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.7475736449482003e-06, | |
| "loss": 2.0027, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.7440799406658886e-06, | |
| "loss": 1.999, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 1.7405657688881594e-06, | |
| "loss": 2.0281, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 1.737031226279754e-06, | |
| "loss": 1.9971, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 1.7334764100657556e-06, | |
| "loss": 1.9807, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1.729901418028917e-06, | |
| "loss": 1.9029, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.7263063485069694e-06, | |
| "loss": 1.9552, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.7226913003899175e-06, | |
| "loss": 1.9035, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1.7190563731173198e-06, | |
| "loss": 2.0661, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.715401666675553e-06, | |
| "loss": 1.9804, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.711727281595062e-06, | |
| "loss": 1.9396, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1.708033318947594e-06, | |
| "loss": 1.9572, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 1.704319880343419e-06, | |
| "loss": 1.8574, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 1.7005870679285343e-06, | |
| "loss": 1.95, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 1.6968349843818548e-06, | |
| "loss": 1.9327, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.6930637329123888e-06, | |
| "loss": 1.8264, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.689273417256399e-06, | |
| "loss": 2.0653, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 1.6854641416745493e-06, | |
| "loss": 2.0006, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 1.681636010949036e-06, | |
| "loss": 1.9048, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.6777891303807057e-06, | |
| "loss": 2.0127, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1.67392360578616e-06, | |
| "loss": 1.9885, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1.6700395434948437e-06, | |
| "loss": 1.9216, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 1.6661370503461205e-06, | |
| "loss": 1.9481, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.6622162336863329e-06, | |
| "loss": 1.9496, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.658277201365851e-06, | |
| "loss": 2.0493, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.654320061736106e-06, | |
| "loss": 1.9908, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.6503449236466077e-06, | |
| "loss": 1.8431, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.6463518964419528e-06, | |
| "loss": 1.9419, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.6423410899588146e-06, | |
| "loss": 1.9064, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.638312614522925e-06, | |
| "loss": 2.0264, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.6342665809460363e-06, | |
| "loss": 2.0588, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.6302031005228757e-06, | |
| "loss": 2.0816, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.626122285028082e-06, | |
| "loss": 1.896, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.6220242467131325e-06, | |
| "loss": 1.9603, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.6179090983032545e-06, | |
| "loss": 1.9241, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.6137769529943247e-06, | |
| "loss": 2.0254, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 1.609627924449756e-06, | |
| "loss": 2.0094, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.6054621267973688e-06, | |
| "loss": 1.8623, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.6012796746262557e-06, | |
| "loss": 1.9158, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.5970806829836256e-06, | |
| "loss": 1.9625, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1.592865267371641e-06, | |
| "loss": 2.0567, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1.58863354374424e-06, | |
| "loss": 2.0132, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 1.5843856285039484e-06, | |
| "loss": 1.9242, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.5801216384986752e-06, | |
| "loss": 1.9029, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.5758416910185015e-06, | |
| "loss": 1.9612, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1.5715459037924506e-06, | |
| "loss": 2.0309, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1.5672343949852533e-06, | |
| "loss": 1.9158, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1.5629072831940948e-06, | |
| "loss": 1.9896, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1.5585646874453537e-06, | |
| "loss": 1.8897, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1.5542067271913276e-06, | |
| "loss": 2.0121, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.5498335223069482e-06, | |
| "loss": 1.9133, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.5454451930864816e-06, | |
| "loss": 1.9221, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.5410418602402217e-06, | |
| "loss": 2.0094, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.536623644891169e-06, | |
| "loss": 2.0667, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.5321906685716983e-06, | |
| "loss": 2.0361, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.5277430532202167e-06, | |
| "loss": 1.9785, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1.523280921177808e-06, | |
| "loss": 1.9187, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.5188043951848697e-06, | |
| "loss": 1.8864, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.514313598377734e-06, | |
| "loss": 1.9596, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.5098086542852832e-06, | |
| "loss": 1.9671, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1.5052896868255502e-06, | |
| "loss": 2.0107, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1.50075682030231e-06, | |
| "loss": 1.9892, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.4962101794016615e-06, | |
| "loss": 1.9368, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 1.4916498891885962e-06, | |
| "loss": 1.9081, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 1.4870760751035598e-06, | |
| "loss": 2.0494, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1.482488862959e-06, | |
| "loss": 1.9624, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1.4778883789359067e-06, | |
| "loss": 1.9493, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.4732747495803416e-06, | |
| "loss": 1.9449, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 1.4686481017999552e-06, | |
| "loss": 1.9648, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 1.464008562860499e-06, | |
| "loss": 2.0374, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.459356260382322e-06, | |
| "loss": 2.0578, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1.4546913223368615e-06, | |
| "loss": 1.9133, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1.4500138770431236e-06, | |
| "loss": 1.961, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1.445324053164152e-06, | |
| "loss": 1.9423, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 1.44062197970349e-06, | |
| "loss": 1.9526, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 1.4359077860016313e-06, | |
| "loss": 1.8559, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1.4311816017324629e-06, | |
| "loss": 1.9737, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 1.4264435568996975e-06, | |
| "loss": 1.9352, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 1.4216937818332975e-06, | |
| "loss": 1.9109, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 1.4169324071858908e-06, | |
| "loss": 1.9183, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 1.412159563929176e-06, | |
| "loss": 1.891, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 1.4073753833503204e-06, | |
| "loss": 1.8555, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 1.4025799970483475e-06, | |
| "loss": 1.9341, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 1.397773536930519e-06, | |
| "loss": 1.9083, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 1.3929561352087054e-06, | |
| "loss": 1.9301, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 1.3881279243957482e-06, | |
| "loss": 2.042, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 1.3832890373018164e-06, | |
| "loss": 1.8688, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 1.3784396070307534e-06, | |
| "loss": 2.0596, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1.3735797669764136e-06, | |
| "loss": 1.8923, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1.3687096508189955e-06, | |
| "loss": 1.9361, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 1.3638293925213637e-06, | |
| "loss": 1.9087, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1.3589391263253636e-06, | |
| "loss": 1.9374, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1.3540389867481288e-06, | |
| "loss": 1.9463, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.3491291085783813e-06, | |
| "loss": 1.9529, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1.3442096268727243e-06, | |
| "loss": 1.9956, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1.3392806769519257e-06, | |
| "loss": 1.9595, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1.3343423943971974e-06, | |
| "loss": 1.9829, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1.3293949150464648e-06, | |
| "loss": 1.9166, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1.3244383749906307e-06, | |
| "loss": 2.0054, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.3194729105698317e-06, | |
| "loss": 1.8892, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.3144986583696887e-06, | |
| "loss": 1.9351, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1.309515755217548e-06, | |
| "loss": 1.9215, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.3045243381787194e-06, | |
| "loss": 1.902, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.2995245445527052e-06, | |
| "loss": 1.9517, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.294516511869423e-06, | |
| "loss": 1.9034, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1.2895003778854238e-06, | |
| "loss": 1.978, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 1.2844762805801005e-06, | |
| "loss": 2.0102, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 1.2794443581518965e-06, | |
| "loss": 2.0112, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 1.2744047490144997e-06, | |
| "loss": 1.9159, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 1.2693575917930385e-06, | |
| "loss": 1.8738, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 1.2643030253202667e-06, | |
| "loss": 1.9698, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 1.2592411886327447e-06, | |
| "loss": 1.9181, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 1.254172220967017e-06, | |
| "loss": 1.9527, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 1.2490962617557799e-06, | |
| "loss": 2.0047, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 1.2440134506240468e-06, | |
| "loss": 1.928, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 1.2389239273853083e-06, | |
| "loss": 1.9579, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 1.233827832037685e-06, | |
| "loss": 2.0036, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 1.228725304760078e-06, | |
| "loss": 1.9702, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 1.2236164859083115e-06, | |
| "loss": 1.973, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 1.2185015160112739e-06, | |
| "loss": 1.9291, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 1.21338053576705e-06, | |
| "loss": 2.0149, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.2082536860390527e-06, | |
| "loss": 2.0611, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.203121107852147e-06, | |
| "loss": 1.9571, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.1979829423887714e-06, | |
| "loss": 1.9215, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.192839330985054e-06, | |
| "loss": 1.9835, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.1876904151269261e-06, | |
| "loss": 1.9197, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.1825363364462284e-06, | |
| "loss": 1.9465, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.1773772367168158e-06, | |
| "loss": 2.0307, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.1722132578506583e-06, | |
| "loss": 1.9428, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.1670445418939365e-06, | |
| "loss": 2.0145, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.1618712310231354e-06, | |
| "loss": 1.924, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.1566934675411319e-06, | |
| "loss": 1.9517, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.151511393873282e-06, | |
| "loss": 1.9754, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.1463251525635023e-06, | |
| "loss": 1.9828, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.1411348862703493e-06, | |
| "loss": 1.9448, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.1359407377630954e-06, | |
| "loss": 1.9983, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.1307428499178012e-06, | |
| "loss": 2.0485, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.1255413657133854e-06, | |
| "loss": 1.9675, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.120336428227693e-06, | |
| "loss": 1.9289, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.1151281806335585e-06, | |
| "loss": 2.031, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.1099167661948675e-06, | |
| "loss": 2.0032, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.1047023282626168e-06, | |
| "loss": 1.8983, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.0994850102709712e-06, | |
| "loss": 1.9464, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.0942649557333167e-06, | |
| "loss": 1.9024, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.0890423082383146e-06, | |
| "loss": 1.9065, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.0838172114459503e-06, | |
| "loss": 1.8741, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.078589809083583e-06, | |
| "loss": 2.0164, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.0733602449419908e-06, | |
| "loss": 1.907, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.0681286628714165e-06, | |
| "loss": 1.9996, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.0628952067776103e-06, | |
| "loss": 1.9657, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.0576600206178717e-06, | |
| "loss": 2.0442, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.052423248397089e-06, | |
| "loss": 1.9115, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.0471850341637779e-06, | |
| "loss": 1.8589, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.0419455220061205e-06, | |
| "loss": 1.7948, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.0367048560480011e-06, | |
| "loss": 1.9377, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.0314631804450413e-06, | |
| "loss": 1.8379, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.026220639380635e-06, | |
| "loss": 1.9493, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.020977377061983e-06, | |
| "loss": 1.832, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.0157335377161256e-06, | |
| "loss": 1.9286, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.0104892655859749e-06, | |
| "loss": 1.8405, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.0052447049263486e-06, | |
| "loss": 1.9792, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1e-06, | |
| "loss": 1.8744, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 9.947552950736513e-07, | |
| "loss": 1.7692, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 9.89510734414025e-07, | |
| "loss": 1.854, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 9.842664622838746e-07, | |
| "loss": 1.9847, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 9.790226229380172e-07, | |
| "loss": 2.0078, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 9.73779360619365e-07, | |
| "loss": 1.9792, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 9.68536819554959e-07, | |
| "loss": 1.993, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 9.632951439519992e-07, | |
| "loss": 1.9626, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 9.580544779938798e-07, | |
| "loss": 1.9082, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 9.528149658362222e-07, | |
| "loss": 1.8499, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 9.475767516029112e-07, | |
| "loss": 1.8829, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 9.423399793821282e-07, | |
| "loss": 1.9563, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 9.371047932223896e-07, | |
| "loss": 1.8991, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 9.318713371285835e-07, | |
| "loss": 1.985, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 9.266397550580094e-07, | |
| "loss": 1.8339, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 9.214101909164171e-07, | |
| "loss": 1.8133, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 9.161827885540496e-07, | |
| "loss": 1.9, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 9.109576917616854e-07, | |
| "loss": 1.8889, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 9.057350442666832e-07, | |
| "loss": 1.8454, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 9.005149897290287e-07, | |
| "loss": 1.9678, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 8.952976717373831e-07, | |
| "loss": 1.8224, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 8.900832338051327e-07, | |
| "loss": 1.8635, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 8.848718193664418e-07, | |
| "loss": 1.9417, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 8.796635717723072e-07, | |
| "loss": 1.8558, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 8.744586342866149e-07, | |
| "loss": 1.9656, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 8.69257150082199e-07, | |
| "loss": 1.8963, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 8.640592622369046e-07, | |
| "loss": 1.9479, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 8.588651137296506e-07, | |
| "loss": 1.9209, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 8.536748474364978e-07, | |
| "loss": 1.8947, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 8.48488606126718e-07, | |
| "loss": 1.9776, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 8.433065324588682e-07, | |
| "loss": 1.9054, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 8.381287689768646e-07, | |
| "loss": 1.8856, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 8.329554581060634e-07, | |
| "loss": 1.999, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 8.277867421493418e-07, | |
| "loss": 1.9128, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 8.226227632831842e-07, | |
| "loss": 1.9865, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 8.174636635537717e-07, | |
| "loss": 1.9372, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 8.123095848730739e-07, | |
| "loss": 1.8931, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 8.07160669014946e-07, | |
| "loss": 1.8591, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 8.020170576112291e-07, | |
| "loss": 1.8493, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 7.968788921478533e-07, | |
| "loss": 1.9029, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 7.917463139609475e-07, | |
| "loss": 1.8815, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 7.8661946423295e-07, | |
| "loss": 1.9633, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 7.814984839887257e-07, | |
| "loss": 1.9075, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 7.763835140916881e-07, | |
| "loss": 1.8922, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 7.712746952399221e-07, | |
| "loss": 1.9037, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 7.661721679623151e-07, | |
| "loss": 1.9065, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 7.610760726146918e-07, | |
| "loss": 1.9652, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 7.559865493759532e-07, | |
| "loss": 1.9843, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 7.509037382442201e-07, | |
| "loss": 1.8977, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 7.45827779032983e-07, | |
| "loss": 1.9443, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 7.407588113672551e-07, | |
| "loss": 1.9343, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 7.356969746797335e-07, | |
| "loss": 1.9663, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 7.306424082069614e-07, | |
| "loss": 1.8752, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 7.255952509855001e-07, | |
| "loss": 1.8767, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 7.205556418481035e-07, | |
| "loss": 1.8254, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 7.155237194198996e-07, | |
| "loss": 1.8957, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 7.104996221145767e-07, | |
| "loss": 1.8839, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 7.054834881305771e-07, | |
| "loss": 1.8521, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 7.004754554472949e-07, | |
| "loss": 1.9679, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 6.954756618212802e-07, | |
| "loss": 1.8759, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 6.904842447824517e-07, | |
| "loss": 1.9296, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 6.855013416303112e-07, | |
| "loss": 1.8545, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 6.805270894301681e-07, | |
| "loss": 1.8324, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 6.755616250093693e-07, | |
| "loss": 1.9205, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 6.706050849535352e-07, | |
| "loss": 1.7527, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 6.656576056028027e-07, | |
| "loss": 2.0347, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 6.607193230480743e-07, | |
| "loss": 1.8517, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 6.557903731272756e-07, | |
| "loss": 1.8878, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 6.508708914216185e-07, | |
| "loss": 1.8004, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 6.459610132518714e-07, | |
| "loss": 1.8932, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 6.410608736746366e-07, | |
| "loss": 2.0047, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 6.361706074786363e-07, | |
| "loss": 1.8754, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 6.312903491810046e-07, | |
| "loss": 1.8231, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 6.264202330235869e-07, | |
| "loss": 1.9687, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 6.21560392969247e-07, | |
| "loss": 1.9207, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 6.167109626981832e-07, | |
| "loss": 1.9462, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 6.118720756042516e-07, | |
| "loss": 1.8702, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 6.070438647912944e-07, | |
| "loss": 2.0097, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 6.022264630694806e-07, | |
| "loss": 1.8375, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 5.974200029516525e-07, | |
| "loss": 1.863, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 5.926246166496798e-07, | |
| "loss": 1.9128, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 5.87840436070824e-07, | |
| "loss": 1.8964, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 5.830675928141091e-07, | |
| "loss": 1.8985, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 5.783062181667025e-07, | |
| "loss": 1.8863, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 5.735564431003026e-07, | |
| "loss": 1.8901, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 5.688183982675369e-07, | |
| "loss": 1.8665, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 5.640922139983686e-07, | |
| "loss": 1.8827, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 5.593780202965099e-07, | |
| "loss": 1.9644, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 5.546759468358478e-07, | |
| "loss": 1.9868, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 5.499861229568763e-07, | |
| "loss": 1.9224, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.453086776631384e-07, | |
| "loss": 1.8728, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 5.406437396176781e-07, | |
| "loss": 1.8951, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 5.359914371395011e-07, | |
| "loss": 1.936, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 5.313518982000447e-07, | |
| "loss": 1.9036, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.267252504196586e-07, | |
| "loss": 1.9023, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.221116210640932e-07, | |
| "loss": 1.833, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 5.175111370410002e-07, | |
| "loss": 1.9128, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 5.129239248964403e-07, | |
| "loss": 1.9986, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 5.083501108114039e-07, | |
| "loss": 1.8244, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 5.037898205983387e-07, | |
| "loss": 1.8555, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.992431796976902e-07, | |
| "loss": 1.9078, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.9471031317445e-07, | |
| "loss": 1.8335, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.901913457147169e-07, | |
| "loss": 1.9092, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.856864016222661e-07, | |
| "loss": 1.8267, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.811956048151306e-07, | |
| "loss": 1.8613, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.767190788221921e-07, | |
| "loss": 1.8242, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.722569467797837e-07, | |
| "loss": 1.8874, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.678093314283018e-07, | |
| "loss": 1.9682, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 4.633763551088312e-07, | |
| "loss": 1.8454, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 4.58958139759778e-07, | |
| "loss": 1.7379, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.545548069135182e-07, | |
| "loss": 1.9102, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.501664776930516e-07, | |
| "loss": 1.8493, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.457932728086721e-07, | |
| "loss": 1.9161, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 4.414353125546463e-07, | |
| "loss": 1.857, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.370927168059052e-07, | |
| "loss": 1.9713, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.327656050147467e-07, | |
| "loss": 1.9553, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 4.2845409620754937e-07, | |
| "loss": 1.9113, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 4.2415830898149864e-07, | |
| "loss": 1.871, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 4.198783615013246e-07, | |
| "loss": 1.9381, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 4.156143714960516e-07, | |
| "loss": 1.8421, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 4.113664562557599e-07, | |
| "loss": 1.7979, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 4.071347326283592e-07, | |
| "loss": 1.7993, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 4.029193170163745e-07, | |
| "loss": 1.7727, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 3.987203253737442e-07, | |
| "loss": 1.921, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 3.945378732026311e-07, | |
| "loss": 1.9242, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 3.9037207555024396e-07, | |
| "loss": 1.8395, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 3.8622304700567487e-07, | |
| "loss": 1.8791, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 3.820909016967452e-07, | |
| "loss": 1.9393, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 3.7797575328686724e-07, | |
| "loss": 1.8406, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 3.7387771497191777e-07, | |
| "loss": 1.9665, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 3.6979689947712457e-07, | |
| "loss": 1.7523, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 3.657334190539637e-07, | |
| "loss": 1.8894, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 3.616873854770751e-07, | |
| "loss": 1.9143, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 3.5765891004118545e-07, | |
| "loss": 1.9404, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 3.5364810355804753e-07, | |
| "loss": 1.8843, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 3.496550763533924e-07, | |
| "loss": 1.8803, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 3.4567993826389417e-07, | |
| "loss": 1.9427, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 3.417227986341491e-07, | |
| "loss": 1.9263, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 3.377837663136676e-07, | |
| "loss": 1.8607, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 3.3386294965387985e-07, | |
| "loss": 1.8372, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 3.2996045650515635e-07, | |
| "loss": 1.9187, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 3.2607639421384015e-07, | |
| "loss": 1.9061, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 3.222108696192942e-07, | |
| "loss": 1.8791, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 3.183639890509641e-07, | |
| "loss": 1.8556, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 3.145358583254505e-07, | |
| "loss": 1.8458, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 3.1072658274360066e-07, | |
| "loss": 1.747, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 3.06936267087611e-07, | |
| "loss": 1.8214, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 3.031650156181452e-07, | |
| "loss": 1.7996, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 2.9941293207146565e-07, | |
| "loss": 1.8987, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 2.95680119656581e-07, | |
| "loss": 1.9444, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 2.9196668105240595e-07, | |
| "loss": 1.9235, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 2.8827271840493806e-07, | |
| "loss": 1.994, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 2.8459833332444706e-07, | |
| "loss": 1.9091, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 2.809436268826804e-07, | |
| "loss": 1.8419, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 2.773086996100826e-07, | |
| "loss": 1.8717, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 2.736936514930307e-07, | |
| "loss": 1.9248, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 2.7009858197108304e-07, | |
| "loss": 1.7921, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 2.665235899342445e-07, | |
| "loss": 1.8654, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 2.629687737202463e-07, | |
| "loss": 1.9847, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 2.5943423111184025e-07, | |
| "loss": 1.8849, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 2.5592005933411123e-07, | |
| "loss": 1.7962, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 2.5242635505179956e-07, | |
| "loss": 1.8999, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 2.4895321436664454e-07, | |
| "loss": 1.9158, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 2.4550073281474016e-07, | |
| "loss": 1.8853, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 2.4206900536390694e-07, | |
| "loss": 1.9685, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 2.3865812641107994e-07, | |
| "loss": 1.8479, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 2.352681897797123e-07, | |
| "loss": 1.8929, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 2.31899288717194e-07, | |
| "loss": 1.8901, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 2.2855151589228728e-07, | |
| "loss": 1.917, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 2.2522496339257725e-07, | |
| "loss": 1.8749, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 2.2191972272193927e-07, | |
| "loss": 1.8659, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 2.1863588479802164e-07, | |
| "loss": 1.8686, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 2.1537353994974427e-07, | |
| "loss": 1.8086, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 2.121327779148151e-07, | |
| "loss": 1.8242, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 2.0891368783726082e-07, | |
| "loss": 1.9317, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 2.057163582649748e-07, | |
| "loss": 1.6616, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 2.025408771472823e-07, | |
| "loss": 2.0278, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 1.9938733183251988e-07, | |
| "loss": 1.8284, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 1.9625580906563344e-07, | |
| "loss": 1.8216, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 1.931463949857921e-07, | |
| "loss": 1.9233, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 1.9005917512401904e-07, | |
| "loss": 1.8586, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 1.8699423440083784e-07, | |
| "loss": 1.9473, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 1.8395165712393757e-07, | |
| "loss": 1.8949, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 1.8093152698585323e-07, | |
| "loss": 1.8405, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 1.7793392706166366e-07, | |
| "loss": 1.8654, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 1.7495893980670652e-07, | |
| "loss": 1.8732, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 1.7200664705431e-07, | |
| "loss": 1.7413, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 1.6907713001354207e-07, | |
| "loss": 1.8582, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 1.661704692669761e-07, | |
| "loss": 1.9428, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 1.6328674476847514e-07, | |
| "loss": 1.7569, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 1.6042603584099178e-07, | |
| "loss": 1.8232, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 1.5758842117438687e-07, | |
| "loss": 1.8977, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 1.5477397882326415e-07, | |
| "loss": 1.9244, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 1.5198278620482451e-07, | |
| "loss": 1.9544, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 1.4921492009673508e-07, | |
| "loss": 1.9015, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 1.46470456635018e-07, | |
| "loss": 1.8531, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 1.4374947131195613e-07, | |
| "loss": 1.7529, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 1.4105203897401652e-07, | |
| "loss": 1.9829, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 1.3837823381979096e-07, | |
| "loss": 1.8249, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 1.3572812939795596e-07, | |
| "loss": 1.971, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 1.3310179860524917e-07, | |
| "loss": 1.9932, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 1.3049931368446377e-07, | |
| "loss": 1.9642, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 1.279207462224622e-07, | |
| "loss": 1.8701, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 1.2536616714820613e-07, | |
| "loss": 1.7716, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 1.2283564673080604e-07, | |
| "loss": 1.8446, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 1.2032925457758792e-07, | |
| "loss": 1.8007, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 1.178470596321789e-07, | |
| "loss": 1.9133, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 1.1538913017261054e-07, | |
| "loss": 1.7986, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 1.1295553380944067e-07, | |
| "loss": 1.8212, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 1.1054633748389397e-07, | |
| "loss": 1.8235, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 1.0816160746602033e-07, | |
| "loss": 1.9137, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 1.0580140935287174e-07, | |
| "loss": 1.7471, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 1.0346580806669858e-07, | |
| "loss": 1.9121, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 1.0115486785316318e-07, | |
| "loss": 1.8075, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 9.886865227957275e-08, | |
| "loss": 1.989, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 9.660722423313105e-08, | |
| "loss": 1.8621, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 9.437064591920818e-08, | |
| "loss": 1.8096, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 9.215897885962986e-08, | |
| "loss": 1.9778, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 8.997228389098488e-08, | |
| "loss": 1.8672, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 8.78106211629518e-08, | |
| "loss": 1.9536, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 8.567405013664453e-08, | |
| "loss": 1.8732, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 8.356262958297611e-08, | |
| "loss": 1.8178, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 8.147641758104296e-08, | |
| "loss": 1.8722, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 7.941547151652673e-08, | |
| "loss": 1.7503, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 7.737984808011588e-08, | |
| "loss": 1.8077, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 7.536960326594621e-08, | |
| "loss": 1.8949, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 7.338479237006124e-08, | |
| "loss": 1.8243, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 7.14254699888901e-08, | |
| "loss": 1.8965, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 6.949169001774657e-08, | |
| "loss": 1.8391, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 6.758350564934623e-08, | |
| "loss": 1.9538, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 6.57009693723436e-08, | |
| "loss": 1.8083, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 6.384413296988777e-08, | |
| "loss": 1.8027, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 6.201304751819858e-08, | |
| "loss": 1.8294, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 6.020776338516132e-08, | |
| "loss": 1.8743, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 5.8428330228941225e-08, | |
| "loss": 1.9289, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 5.66747969966177e-08, | |
| "loss": 1.9257, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 5.4947211922838134e-08, | |
| "loss": 1.8246, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 5.324562252849052e-08, | |
| "loss": 1.8949, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 5.157007561939653e-08, | |
| "loss": 1.7944, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 4.9920617285024503e-08, | |
| "loss": 1.8897, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 4.829729289722084e-08, | |
| "loss": 1.8992, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 4.670014710896253e-08, | |
| "loss": 1.8639, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 4.5129223853128674e-08, | |
| "loss": 1.9125, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 4.358456634129226e-08, | |
| "loss": 1.7916, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 4.206621706253111e-08, | |
| "loss": 1.8905, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 4.0574217782259335e-08, | |
| "loss": 1.8401, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 3.910860954107853e-08, | |
| "loss": 1.9283, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 3.766943265364908e-08, | |
| "loss": 1.874, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 3.625672670758051e-08, | |
| "loss": 1.8394, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 3.487053056234357e-08, | |
| "loss": 1.8679, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 3.3510882348200207e-08, | |
| "loss": 1.8873, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 3.2177819465155744e-08, | |
| "loss": 1.9823, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 3.087137858192945e-08, | |
| "loss": 1.8502, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 2.9591595634945844e-08, | |
| "loss": 1.942, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 2.833850582734687e-08, | |
| "loss": 1.8499, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 2.7112143628022832e-08, | |
| "loss": 1.909, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 2.591254277066457e-08, | |
| "loss": 1.929, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 2.473973625283554e-08, | |
| "loss": 1.7558, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 2.3593756335063997e-08, | |
| "loss": 1.9228, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 2.2474634539955906e-08, | |
| "loss": 1.8687, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 2.1382401651327763e-08, | |
| "loss": 1.9711, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 2.031708771335927e-08, | |
| "loss": 1.789, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 1.9278722029767657e-08, | |
| "loss": 1.8015, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 1.8267333163001443e-08, | |
| "loss": 1.9756, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 1.7282948933454276e-08, | |
| "loss": 1.7515, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 1.6325596418700238e-08, | |
| "loss": 1.8489, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 1.5395301952748853e-08, | |
| "loss": 1.9228, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 1.4492091125320683e-08, | |
| "loss": 1.7926, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 1.361598878114334e-08, | |
| "loss": 1.9389, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 1.2767019019268132e-08, | |
| "loss": 1.8759, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 1.1945205192407382e-08, | |
| "loss": 1.8408, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 1.1150569906291596e-08, | |
| "loss": 1.8249, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 1.0383135019048195e-08, | |
| "loss": 1.8251, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 9.642921640599876e-09, | |
| "loss": 1.8421, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 8.929950132084085e-09, | |
| "loss": 1.8591, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 8.244240105292898e-09, | |
| "loss": 1.9311, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 7.585810422133466e-09, | |
| "loss": 1.772, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 6.9546791941096365e-09, | |
| "loss": 1.8463, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 6.3508637818228124e-09, | |
| "loss": 1.8765, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 5.77438079449577e-09, | |
| "loss": 1.8936, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 5.2252460895141345e-09, | |
| "loss": 1.7933, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 4.703474771991845e-09, | |
| "loss": 1.8547, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 4.20908119435448e-09, | |
| "loss": 1.7349, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 3.742078955945027e-09, | |
| "loss": 1.9617, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 3.302480902649618e-09, | |
| "loss": 1.861, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 2.890299126544038e-09, | |
| "loss": 1.8998, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 2.5055449655615457e-09, | |
| "loss": 1.9103, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 2.1482290031805684e-09, | |
| "loss": 1.8103, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 1.8183610681338224e-09, | |
| "loss": 1.7133, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 1.5159502341377528e-09, | |
| "loss": 1.8372, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 1.2410048196432876e-09, | |
| "loss": 1.8701, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 9.93532387606355e-10, | |
| "loss": 1.7707, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 7.735397452808268e-10, | |
| "loss": 1.9447, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 5.810329440303352e-10, | |
| "loss": 1.9433, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 4.1601727916251717e-10, | |
| "loss": 1.9547, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 2.784972897827975e-10, | |
| "loss": 1.9523, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 1.6847675866971112e-10, | |
| "loss": 1.873, | |
| "step": 1232 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 8.595871217143003e-11, | |
| "loss": 1.8817, | |
| "step": 1234 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 3.094542012138657e-11, | |
| "loss": 1.7426, | |
| "step": 1236 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "step": 1236, | |
| "total_flos": 5.429005242282803e+16, | |
| "train_loss": 1.9528567034835569, | |
| "train_runtime": 28672.908, | |
| "train_samples_per_second": 2.761, | |
| "train_steps_per_second": 0.043 | |
| } | |
| ], | |
| "logging_steps": 2, | |
| "max_steps": 1236, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 9000, | |
| "total_flos": 5.429005242282803e+16, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |