| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.543115878703137, | |
| "eval_steps": 1000000, | |
| "global_step": 232807, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 9.9789, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 9.1523, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.5e-06, | |
| "loss": 8.7253, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 8.2358, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.5e-06, | |
| "loss": 7.7474, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3e-06, | |
| "loss": 7.3656, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.5e-06, | |
| "loss": 7.1347, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 6.996, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.5e-06, | |
| "loss": 6.9122, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5e-06, | |
| "loss": 6.8432, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 6.7914, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6e-06, | |
| "loss": 6.742, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 6.7043, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 7e-06, | |
| "loss": 6.667, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 6.6395, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 6.609, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.5e-06, | |
| "loss": 6.5723, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9e-06, | |
| "loss": 6.557, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.5e-06, | |
| "loss": 6.5284, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1e-05, | |
| "loss": 6.5054, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.999453217269587e-06, | |
| "loss": 6.4907, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.998906434539172e-06, | |
| "loss": 6.4621, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.998359651808758e-06, | |
| "loss": 6.4478, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.997812869078344e-06, | |
| "loss": 6.432, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.99726608634793e-06, | |
| "loss": 6.4166, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.996719303617516e-06, | |
| "loss": 6.3981, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.996172520887101e-06, | |
| "loss": 6.39, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.995625738156687e-06, | |
| "loss": 6.369, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.995078955426273e-06, | |
| "loss": 6.3661, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.994532172695859e-06, | |
| "loss": 6.3465, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.993985389965445e-06, | |
| "loss": 6.3371, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.99343860723503e-06, | |
| "loss": 6.3304, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.992891824504616e-06, | |
| "loss": 6.3201, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.992345041774202e-06, | |
| "loss": 6.3124, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.991798259043786e-06, | |
| "loss": 6.3031, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.991251476313374e-06, | |
| "loss": 6.2961, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.99070469358296e-06, | |
| "loss": 6.2811, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.990157910852545e-06, | |
| "loss": 6.2795, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.98961112812213e-06, | |
| "loss": 6.2809, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.989064345391715e-06, | |
| "loss": 6.2676, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.988517562661302e-06, | |
| "loss": 6.2549, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.987970779930888e-06, | |
| "loss": 6.2502, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.987423997200472e-06, | |
| "loss": 6.2518, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.986877214470058e-06, | |
| "loss": 6.2446, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.986330431739644e-06, | |
| "loss": 6.2418, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.985783649009231e-06, | |
| "loss": 6.2318, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.985236866278816e-06, | |
| "loss": 6.2242, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.984690083548401e-06, | |
| "loss": 6.227, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.984143300817987e-06, | |
| "loss": 6.2216, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.983596518087575e-06, | |
| "loss": 6.2106, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.98304973535716e-06, | |
| "loss": 6.2116, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.982502952626745e-06, | |
| "loss": 6.2045, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.98195616989633e-06, | |
| "loss": 6.208, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.981409387165916e-06, | |
| "loss": 6.1964, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.980862604435504e-06, | |
| "loss": 6.188, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.980315821705088e-06, | |
| "loss": 6.1911, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.979769038974673e-06, | |
| "loss": 6.1826, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.97922225624426e-06, | |
| "loss": 6.1864, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.978675473513845e-06, | |
| "loss": 6.1786, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.97812869078343e-06, | |
| "loss": 6.1727, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.977581908053017e-06, | |
| "loss": 6.1738, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.977035125322602e-06, | |
| "loss": 6.1743, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.976488342592188e-06, | |
| "loss": 6.1656, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.975941559861774e-06, | |
| "loss": 6.1678, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.97539477713136e-06, | |
| "loss": 6.1622, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.974847994400946e-06, | |
| "loss": 6.1577, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.974301211670531e-06, | |
| "loss": 6.1546, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.973754428940117e-06, | |
| "loss": 6.1512, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.973207646209703e-06, | |
| "loss": 6.1472, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.972660863479289e-06, | |
| "loss": 6.1437, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.972114080748875e-06, | |
| "loss": 6.1449, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.97156729801846e-06, | |
| "loss": 6.1439, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.971020515288046e-06, | |
| "loss": 6.1354, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.970473732557632e-06, | |
| "loss": 6.1392, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.969926949827218e-06, | |
| "loss": 6.1352, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.969380167096804e-06, | |
| "loss": 6.1343, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.96883338436639e-06, | |
| "loss": 6.1332, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.968286601635975e-06, | |
| "loss": 6.1197, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.967739818905561e-06, | |
| "loss": 6.121, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.967193036175147e-06, | |
| "loss": 6.1193, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.966646253444732e-06, | |
| "loss": 6.1211, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.966099470714318e-06, | |
| "loss": 6.1134, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.965552687983904e-06, | |
| "loss": 6.1204, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.96500590525349e-06, | |
| "loss": 6.1137, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.964459122523074e-06, | |
| "loss": 6.1153, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 9.963912339792661e-06, | |
| "loss": 6.1102, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.963365557062247e-06, | |
| "loss": 6.0987, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.962818774331833e-06, | |
| "loss": 6.1063, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 9.962271991601417e-06, | |
| "loss": 6.1057, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 9.961725208871003e-06, | |
| "loss": 6.1006, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.96117842614059e-06, | |
| "loss": 6.103, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.960631643410176e-06, | |
| "loss": 6.0981, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.96008486067976e-06, | |
| "loss": 6.0965, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.959538077949346e-06, | |
| "loss": 6.0923, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.958991295218932e-06, | |
| "loss": 6.0916, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.95844451248852e-06, | |
| "loss": 6.088, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.957897729758103e-06, | |
| "loss": 6.0888, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.95735094702769e-06, | |
| "loss": 6.0908, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.956804164297275e-06, | |
| "loss": 6.0877, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.956257381566863e-06, | |
| "loss": 6.0863, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.955710598836447e-06, | |
| "loss": 6.0814, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.955163816106032e-06, | |
| "loss": 6.0778, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.954617033375618e-06, | |
| "loss": 6.0746, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.954070250645204e-06, | |
| "loss": 6.0759, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.953523467914792e-06, | |
| "loss": 6.0719, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.952976685184376e-06, | |
| "loss": 6.0784, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.952429902453961e-06, | |
| "loss": 6.0724, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 9.951883119723547e-06, | |
| "loss": 6.073, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 9.951336336993133e-06, | |
| "loss": 6.0703, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 9.950789554262719e-06, | |
| "loss": 6.0664, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 9.950242771532305e-06, | |
| "loss": 6.0693, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 9.94969598880189e-06, | |
| "loss": 6.0664, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 9.949149206071476e-06, | |
| "loss": 6.0627, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 9.948602423341062e-06, | |
| "loss": 6.0608, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 9.948055640610648e-06, | |
| "loss": 6.0606, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 9.947508857880234e-06, | |
| "loss": 6.0548, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 9.94696207514982e-06, | |
| "loss": 6.0465, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 9.946415292419405e-06, | |
| "loss": 6.0435, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 9.945868509688991e-06, | |
| "loss": 6.0328, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 9.945321726958577e-06, | |
| "loss": 6.0353, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 9.944774944228163e-06, | |
| "loss": 6.0196, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 9.944228161497748e-06, | |
| "loss": 6.0076, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 9.943681378767334e-06, | |
| "loss": 5.9971, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 9.94313459603692e-06, | |
| "loss": 5.9783, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 9.942587813306506e-06, | |
| "loss": 5.9617, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 9.94204103057609e-06, | |
| "loss": 5.9502, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 9.941494247845677e-06, | |
| "loss": 5.928, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 9.940947465115263e-06, | |
| "loss": 5.921, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 9.940400682384849e-06, | |
| "loss": 5.9109, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 9.939853899654433e-06, | |
| "loss": 5.8914, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 9.93930711692402e-06, | |
| "loss": 5.8828, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 9.938760334193606e-06, | |
| "loss": 5.8726, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 9.938213551463192e-06, | |
| "loss": 5.8681, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 9.937666768732778e-06, | |
| "loss": 5.8591, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.937119986002362e-06, | |
| "loss": 5.8543, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.93657320327195e-06, | |
| "loss": 5.8458, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.936026420541535e-06, | |
| "loss": 5.8387, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.935479637811121e-06, | |
| "loss": 5.8297, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 9.934932855080705e-06, | |
| "loss": 5.8223, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 9.934386072350291e-06, | |
| "loss": 5.8208, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 9.933839289619878e-06, | |
| "loss": 5.8106, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 9.933292506889464e-06, | |
| "loss": 5.8008, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 9.932745724159048e-06, | |
| "loss": 5.7954, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 9.932198941428634e-06, | |
| "loss": 5.7938, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 9.93165215869822e-06, | |
| "loss": 5.7845, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 9.931105375967807e-06, | |
| "loss": 5.7746, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 9.930558593237391e-06, | |
| "loss": 5.7726, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.930011810506977e-06, | |
| "loss": 5.7701, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.929465027776563e-06, | |
| "loss": 5.7648, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 9.928918245046149e-06, | |
| "loss": 5.7566, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 9.928371462315735e-06, | |
| "loss": 5.7437, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 9.92782467958532e-06, | |
| "loss": 5.7401, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 9.927277896854906e-06, | |
| "loss": 5.7386, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 9.926731114124492e-06, | |
| "loss": 5.7292, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 9.926184331394078e-06, | |
| "loss": 5.7264, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 9.925637548663664e-06, | |
| "loss": 5.7148, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 9.92509076593325e-06, | |
| "loss": 5.7053, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 9.924543983202835e-06, | |
| "loss": 5.7018, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.923997200472421e-06, | |
| "loss": 5.6953, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.923450417742007e-06, | |
| "loss": 5.6929, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 9.922903635011593e-06, | |
| "loss": 5.6906, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 9.922356852281178e-06, | |
| "loss": 5.6735, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 9.921810069550764e-06, | |
| "loss": 5.6663, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 9.92126328682035e-06, | |
| "loss": 5.657, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 9.920716504089936e-06, | |
| "loss": 5.6512, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.920169721359522e-06, | |
| "loss": 5.6392, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.919622938629107e-06, | |
| "loss": 5.6294, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 9.919076155898693e-06, | |
| "loss": 5.6277, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 9.918529373168279e-06, | |
| "loss": 5.6061, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 9.917982590437865e-06, | |
| "loss": 5.6036, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 9.91743580770745e-06, | |
| "loss": 5.5902, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 9.916889024977036e-06, | |
| "loss": 5.5646, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 9.916342242246622e-06, | |
| "loss": 5.5324, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 9.915795459516208e-06, | |
| "loss": 5.493, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 9.915248676785794e-06, | |
| "loss": 5.4518, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 9.914701894055378e-06, | |
| "loss": 5.4038, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 9.914155111324965e-06, | |
| "loss": 5.3678, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 9.913608328594551e-06, | |
| "loss": 5.3252, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 9.913061545864137e-06, | |
| "loss": 5.2908, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 9.912514763133721e-06, | |
| "loss": 5.2526, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 9.911967980403307e-06, | |
| "loss": 5.2146, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 9.911421197672894e-06, | |
| "loss": 5.1818, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.91087441494248e-06, | |
| "loss": 5.157, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.910327632212064e-06, | |
| "loss": 5.1218, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.90978084948165e-06, | |
| "loss": 5.0903, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.909234066751237e-06, | |
| "loss": 5.0619, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.908687284020823e-06, | |
| "loss": 5.0413, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.908140501290409e-06, | |
| "loss": 5.0108, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.907593718559993e-06, | |
| "loss": 4.991, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.907046935829579e-06, | |
| "loss": 4.9627, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.906500153099166e-06, | |
| "loss": 4.9401, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.905953370368752e-06, | |
| "loss": 4.9129, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.905406587638336e-06, | |
| "loss": 4.8912, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.904859804907922e-06, | |
| "loss": 4.8729, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.904313022177508e-06, | |
| "loss": 4.8533, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.903766239447095e-06, | |
| "loss": 4.828, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.90321945671668e-06, | |
| "loss": 4.8151, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.902672673986265e-06, | |
| "loss": 4.7884, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.902125891255851e-06, | |
| "loss": 4.7728, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.901579108525437e-06, | |
| "loss": 4.7488, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 9.901032325795023e-06, | |
| "loss": 4.7387, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 9.900485543064608e-06, | |
| "loss": 4.7159, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 9.899938760334194e-06, | |
| "loss": 4.6981, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 9.89939197760378e-06, | |
| "loss": 4.6833, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 9.898845194873366e-06, | |
| "loss": 4.6696, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 9.898298412142952e-06, | |
| "loss": 4.647, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 9.897751629412537e-06, | |
| "loss": 4.6336, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 9.897204846682123e-06, | |
| "loss": 4.6117, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 9.896658063951709e-06, | |
| "loss": 4.5993, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 9.896111281221295e-06, | |
| "loss": 4.5817, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 9.89556449849088e-06, | |
| "loss": 4.5709, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 9.895017715760466e-06, | |
| "loss": 4.5481, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 9.894470933030052e-06, | |
| "loss": 4.5362, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 9.893924150299638e-06, | |
| "loss": 4.5212, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 9.893377367569224e-06, | |
| "loss": 4.5116, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 9.89283058483881e-06, | |
| "loss": 4.4934, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 9.892283802108395e-06, | |
| "loss": 4.4762, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 9.891737019377981e-06, | |
| "loss": 4.4654, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 9.891190236647567e-06, | |
| "loss": 4.4501, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 9.890643453917153e-06, | |
| "loss": 4.4265, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 9.890096671186738e-06, | |
| "loss": 4.3906, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 9.889549888456324e-06, | |
| "loss": 4.3579, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 9.88900310572591e-06, | |
| "loss": 4.3283, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 9.888456322995496e-06, | |
| "loss": 4.3051, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 9.887909540265082e-06, | |
| "loss": 4.2768, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 9.887362757534666e-06, | |
| "loss": 4.256, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 9.886815974804253e-06, | |
| "loss": 4.2377, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 9.886269192073839e-06, | |
| "loss": 4.2195, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 9.885722409343425e-06, | |
| "loss": 4.2069, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 9.885175626613009e-06, | |
| "loss": 4.1831, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 9.884628843882595e-06, | |
| "loss": 4.177, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 9.884082061152182e-06, | |
| "loss": 4.1553, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 9.883535278421768e-06, | |
| "loss": 4.1407, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.882988495691352e-06, | |
| "loss": 4.128, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.882441712960938e-06, | |
| "loss": 4.1119, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 9.881894930230524e-06, | |
| "loss": 4.0982, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 9.881348147500111e-06, | |
| "loss": 4.087, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 9.880801364769695e-06, | |
| "loss": 4.0627, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 9.880254582039281e-06, | |
| "loss": 4.0603, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 9.879707799308867e-06, | |
| "loss": 4.0494, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 9.879161016578454e-06, | |
| "loss": 4.0341, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 9.878614233848038e-06, | |
| "loss": 4.013, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 9.878067451117624e-06, | |
| "loss": 4.0131, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 9.87752066838721e-06, | |
| "loss": 3.9992, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 9.876973885656796e-06, | |
| "loss": 3.9831, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 9.876427102926383e-06, | |
| "loss": 3.974, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 9.875880320195967e-06, | |
| "loss": 3.9645, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 9.875333537465553e-06, | |
| "loss": 3.9593, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 9.874786754735139e-06, | |
| "loss": 3.9428, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 9.874239972004725e-06, | |
| "loss": 3.9294, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 9.87369318927431e-06, | |
| "loss": 3.9229, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 9.873146406543896e-06, | |
| "loss": 3.9173, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 9.872599623813482e-06, | |
| "loss": 3.8997, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 9.872052841083068e-06, | |
| "loss": 3.8965, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 9.871506058352654e-06, | |
| "loss": 3.8906, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 9.87095927562224e-06, | |
| "loss": 3.8752, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 9.870412492891825e-06, | |
| "loss": 3.8704, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 9.869865710161411e-06, | |
| "loss": 3.8609, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 9.869318927430997e-06, | |
| "loss": 3.8468, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 9.868772144700583e-06, | |
| "loss": 3.8462, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 9.868225361970168e-06, | |
| "loss": 3.8337, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 9.867678579239754e-06, | |
| "loss": 3.8273, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 9.86713179650934e-06, | |
| "loss": 3.8206, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 9.866585013778926e-06, | |
| "loss": 3.8159, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 9.866038231048512e-06, | |
| "loss": 3.8037, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 9.865491448318097e-06, | |
| "loss": 3.8, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 9.864944665587683e-06, | |
| "loss": 3.7908, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 9.864397882857269e-06, | |
| "loss": 3.7823, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 9.863851100126855e-06, | |
| "loss": 3.7739, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 9.86330431739644e-06, | |
| "loss": 3.7645, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 9.862757534666025e-06, | |
| "loss": 3.7634, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 9.862210751935612e-06, | |
| "loss": 3.7555, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 9.861663969205198e-06, | |
| "loss": 3.7403, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 9.861117186474784e-06, | |
| "loss": 3.7337, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 9.86057040374437e-06, | |
| "loss": 3.7341, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 9.860023621013954e-06, | |
| "loss": 3.7221, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 9.859476838283541e-06, | |
| "loss": 3.7147, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 9.858930055553127e-06, | |
| "loss": 3.7164, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 9.858383272822713e-06, | |
| "loss": 3.7024, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 9.857836490092297e-06, | |
| "loss": 3.7001, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 9.857289707361883e-06, | |
| "loss": 3.6919, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 9.85674292463147e-06, | |
| "loss": 3.694, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 9.856196141901056e-06, | |
| "loss": 3.683, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 9.85564935917064e-06, | |
| "loss": 3.673, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 9.855102576440226e-06, | |
| "loss": 3.6787, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 9.854555793709812e-06, | |
| "loss": 3.6687, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 9.854009010979399e-06, | |
| "loss": 3.6555, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 9.853462228248983e-06, | |
| "loss": 3.6559, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 9.852915445518569e-06, | |
| "loss": 3.6547, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 9.852368662788155e-06, | |
| "loss": 3.6402, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 9.851821880057742e-06, | |
| "loss": 3.6434, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 9.851275097327326e-06, | |
| "loss": 3.637, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 9.850728314596912e-06, | |
| "loss": 3.6329, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.850181531866498e-06, | |
| "loss": 3.6219, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.849634749136084e-06, | |
| "loss": 3.6213, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 9.84908796640567e-06, | |
| "loss": 3.6129, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 9.848541183675255e-06, | |
| "loss": 3.6005, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 9.847994400944841e-06, | |
| "loss": 3.6033, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 9.847447618214427e-06, | |
| "loss": 3.5977, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 9.846900835484013e-06, | |
| "loss": 3.5927, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 9.846354052753598e-06, | |
| "loss": 3.5835, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 9.845807270023184e-06, | |
| "loss": 3.5834, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 9.84526048729277e-06, | |
| "loss": 3.578, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 9.844713704562356e-06, | |
| "loss": 3.5689, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 9.844166921831942e-06, | |
| "loss": 3.5622, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 9.843620139101527e-06, | |
| "loss": 3.5635, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 9.843073356371113e-06, | |
| "loss": 3.5589, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 9.842526573640699e-06, | |
| "loss": 3.5576, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 9.841979790910285e-06, | |
| "loss": 3.5478, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 9.84143300817987e-06, | |
| "loss": 3.5401, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 9.840886225449456e-06, | |
| "loss": 3.539, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 9.840339442719042e-06, | |
| "loss": 3.5334, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 9.839792659988628e-06, | |
| "loss": 3.5374, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 9.839245877258214e-06, | |
| "loss": 3.5259, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 9.8386990945278e-06, | |
| "loss": 3.5181, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 9.838152311797385e-06, | |
| "loss": 3.5202, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 9.83760552906697e-06, | |
| "loss": 3.5114, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 9.837058746336557e-06, | |
| "loss": 3.5079, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 9.836511963606143e-06, | |
| "loss": 3.504, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 9.835965180875729e-06, | |
| "loss": 3.4994, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 9.835418398145313e-06, | |
| "loss": 3.4921, | |
| "step": 160500 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 9.8348716154149e-06, | |
| "loss": 3.4915, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 9.834324832684486e-06, | |
| "loss": 3.4893, | |
| "step": 161500 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 9.833778049954072e-06, | |
| "loss": 3.4816, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 9.833231267223656e-06, | |
| "loss": 3.482, | |
| "step": 162500 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 9.832684484493242e-06, | |
| "loss": 3.4697, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 9.832137701762829e-06, | |
| "loss": 3.4759, | |
| "step": 163500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 9.831590919032415e-06, | |
| "loss": 3.463, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 9.831044136302e-06, | |
| "loss": 3.4663, | |
| "step": 164500 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 9.830497353571585e-06, | |
| "loss": 3.4497, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 9.82995057084117e-06, | |
| "loss": 3.4613, | |
| "step": 165500 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 9.829403788110758e-06, | |
| "loss": 3.4535, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 9.828857005380344e-06, | |
| "loss": 3.4469, | |
| "step": 166500 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 9.828310222649928e-06, | |
| "loss": 3.45, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 9.827763439919514e-06, | |
| "loss": 3.4426, | |
| "step": 167500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 9.8272166571891e-06, | |
| "loss": 3.442, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 9.826669874458687e-06, | |
| "loss": 3.435, | |
| "step": 168500 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 9.826123091728271e-06, | |
| "loss": 3.4416, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 9.825576308997857e-06, | |
| "loss": 3.4207, | |
| "step": 169500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 9.825029526267443e-06, | |
| "loss": 3.4284, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 9.824482743537029e-06, | |
| "loss": 3.4224, | |
| "step": 170500 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 9.823935960806614e-06, | |
| "loss": 3.4198, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 9.8233891780762e-06, | |
| "loss": 3.4088, | |
| "step": 171500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 9.822842395345786e-06, | |
| "loss": 3.4116, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 9.822295612615372e-06, | |
| "loss": 3.4076, | |
| "step": 172500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 9.821748829884957e-06, | |
| "loss": 3.4004, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 9.821202047154543e-06, | |
| "loss": 3.4015, | |
| "step": 173500 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 9.820655264424129e-06, | |
| "loss": 3.3981, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 9.820108481693715e-06, | |
| "loss": 3.388, | |
| "step": 174500 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 9.8195616989633e-06, | |
| "loss": 3.3849, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 9.819014916232886e-06, | |
| "loss": 3.3866, | |
| "step": 175500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 9.818468133502472e-06, | |
| "loss": 3.3866, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 9.817921350772058e-06, | |
| "loss": 3.384, | |
| "step": 176500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 9.817374568041644e-06, | |
| "loss": 3.3784, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 9.81682778531123e-06, | |
| "loss": 3.3695, | |
| "step": 177500 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 9.816281002580815e-06, | |
| "loss": 3.3745, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 9.815734219850401e-06, | |
| "loss": 3.3657, | |
| "step": 178500 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 9.815187437119987e-06, | |
| "loss": 3.3622, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 9.814640654389573e-06, | |
| "loss": 3.3645, | |
| "step": 179500 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 9.814093871659159e-06, | |
| "loss": 3.3589, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 9.813547088928744e-06, | |
| "loss": 3.3476, | |
| "step": 180500 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 9.81300030619833e-06, | |
| "loss": 3.3516, | |
| "step": 181000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 9.812453523467916e-06, | |
| "loss": 3.3472, | |
| "step": 181500 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 9.811906740737502e-06, | |
| "loss": 3.346, | |
| "step": 182000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 9.811359958007088e-06, | |
| "loss": 3.3466, | |
| "step": 182500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 9.810813175276673e-06, | |
| "loss": 3.3428, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 9.810266392546257e-06, | |
| "loss": 3.3339, | |
| "step": 183500 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 9.809719609815845e-06, | |
| "loss": 3.33, | |
| "step": 184000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 9.80917282708543e-06, | |
| "loss": 3.3309, | |
| "step": 184500 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 9.808626044355017e-06, | |
| "loss": 3.3301, | |
| "step": 185000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 9.8080792616246e-06, | |
| "loss": 3.3218, | |
| "step": 185500 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 9.807532478894186e-06, | |
| "loss": 3.324, | |
| "step": 186000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 9.806985696163774e-06, | |
| "loss": 3.3227, | |
| "step": 186500 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 9.80643891343336e-06, | |
| "loss": 3.3153, | |
| "step": 187000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 9.805892130702944e-06, | |
| "loss": 3.3113, | |
| "step": 187500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 9.80534534797253e-06, | |
| "loss": 3.3049, | |
| "step": 188000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 9.804798565242117e-06, | |
| "loss": 3.3052, | |
| "step": 188500 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 9.804251782511703e-06, | |
| "loss": 3.3019, | |
| "step": 189000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 9.803704999781287e-06, | |
| "loss": 3.2976, | |
| "step": 189500 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 9.803158217050873e-06, | |
| "loss": 3.3023, | |
| "step": 190000 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 9.802611434320459e-06, | |
| "loss": 3.2856, | |
| "step": 190500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 9.802064651590046e-06, | |
| "loss": 3.2954, | |
| "step": 191000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 9.80151786885963e-06, | |
| "loss": 3.2899, | |
| "step": 191500 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 9.800971086129216e-06, | |
| "loss": 3.2856, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 9.800424303398802e-06, | |
| "loss": 3.2892, | |
| "step": 192500 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 9.799877520668388e-06, | |
| "loss": 3.2874, | |
| "step": 193000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 9.799330737937975e-06, | |
| "loss": 3.2739, | |
| "step": 193500 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 9.798783955207559e-06, | |
| "loss": 3.2769, | |
| "step": 194000 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 9.798237172477145e-06, | |
| "loss": 3.2698, | |
| "step": 194500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 9.79769038974673e-06, | |
| "loss": 3.2753, | |
| "step": 195000 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 9.797143607016316e-06, | |
| "loss": 3.2699, | |
| "step": 195500 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 9.796596824285902e-06, | |
| "loss": 3.2654, | |
| "step": 196000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 9.796050041555488e-06, | |
| "loss": 3.266, | |
| "step": 196500 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 9.795503258825074e-06, | |
| "loss": 3.2639, | |
| "step": 197000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 9.79495647609466e-06, | |
| "loss": 3.2581, | |
| "step": 197500 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 9.794409693364245e-06, | |
| "loss": 3.2602, | |
| "step": 198000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 9.793862910633831e-06, | |
| "loss": 3.2563, | |
| "step": 198500 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 9.793316127903417e-06, | |
| "loss": 3.2546, | |
| "step": 199000 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 9.792769345173003e-06, | |
| "loss": 3.246, | |
| "step": 199500 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 9.792222562442589e-06, | |
| "loss": 3.2517, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 9.791675779712174e-06, | |
| "loss": 3.2438, | |
| "step": 200500 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 9.79112899698176e-06, | |
| "loss": 3.242, | |
| "step": 201000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 9.790582214251346e-06, | |
| "loss": 3.2359, | |
| "step": 201500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 9.790035431520932e-06, | |
| "loss": 3.2367, | |
| "step": 202000 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 9.789488648790518e-06, | |
| "loss": 3.2376, | |
| "step": 202500 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 9.788941866060103e-06, | |
| "loss": 3.2377, | |
| "step": 203000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 9.78839508332969e-06, | |
| "loss": 3.231, | |
| "step": 203500 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 9.787848300599275e-06, | |
| "loss": 3.2267, | |
| "step": 204000 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 9.78730151786886e-06, | |
| "loss": 3.2245, | |
| "step": 204500 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 9.786754735138447e-06, | |
| "loss": 3.2228, | |
| "step": 205000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 9.786207952408032e-06, | |
| "loss": 3.2179, | |
| "step": 205500 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 9.785661169677618e-06, | |
| "loss": 3.2125, | |
| "step": 206000 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 9.785114386947204e-06, | |
| "loss": 3.2171, | |
| "step": 206500 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 9.78456760421679e-06, | |
| "loss": 3.2146, | |
| "step": 207000 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 9.784020821486376e-06, | |
| "loss": 3.2083, | |
| "step": 207500 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 9.783474038755961e-06, | |
| "loss": 3.2097, | |
| "step": 208000 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 9.782927256025545e-06, | |
| "loss": 3.2157, | |
| "step": 208500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 9.782380473295133e-06, | |
| "loss": 3.2016, | |
| "step": 209000 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 9.781833690564719e-06, | |
| "loss": 3.1991, | |
| "step": 209500 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 9.781286907834304e-06, | |
| "loss": 3.1978, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 9.780740125103889e-06, | |
| "loss": 3.1969, | |
| "step": 210500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 9.780193342373474e-06, | |
| "loss": 3.1988, | |
| "step": 211000 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 9.779646559643062e-06, | |
| "loss": 3.1923, | |
| "step": 211500 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 9.779099776912648e-06, | |
| "loss": 3.1875, | |
| "step": 212000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 9.778552994182232e-06, | |
| "loss": 3.1816, | |
| "step": 212500 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 9.778006211451818e-06, | |
| "loss": 3.1837, | |
| "step": 213000 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 9.777459428721405e-06, | |
| "loss": 3.1773, | |
| "step": 213500 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 9.77691264599099e-06, | |
| "loss": 3.1877, | |
| "step": 214000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 9.776365863260575e-06, | |
| "loss": 3.1792, | |
| "step": 214500 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 9.77581908053016e-06, | |
| "loss": 3.1706, | |
| "step": 215000 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 9.775272297799747e-06, | |
| "loss": 3.1767, | |
| "step": 215500 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 9.774725515069334e-06, | |
| "loss": 3.1675, | |
| "step": 216000 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 9.774178732338918e-06, | |
| "loss": 3.1594, | |
| "step": 216500 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 9.773631949608504e-06, | |
| "loss": 3.1742, | |
| "step": 217000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 9.77308516687809e-06, | |
| "loss": 3.1645, | |
| "step": 217500 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 9.772538384147675e-06, | |
| "loss": 3.1644, | |
| "step": 218000 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 9.771991601417261e-06, | |
| "loss": 3.1616, | |
| "step": 218500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 9.771444818686847e-06, | |
| "loss": 3.1551, | |
| "step": 219000 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 9.770898035956433e-06, | |
| "loss": 3.1572, | |
| "step": 219500 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 9.770351253226019e-06, | |
| "loss": 3.1575, | |
| "step": 220000 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 9.769804470495604e-06, | |
| "loss": 3.1529, | |
| "step": 220500 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 9.76925768776519e-06, | |
| "loss": 3.1432, | |
| "step": 221000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 9.768710905034776e-06, | |
| "loss": 3.147, | |
| "step": 221500 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 9.768164122304362e-06, | |
| "loss": 3.1408, | |
| "step": 222000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 9.767617339573948e-06, | |
| "loss": 3.1431, | |
| "step": 222500 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.767070556843533e-06, | |
| "loss": 3.1372, | |
| "step": 223000 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.76652377411312e-06, | |
| "loss": 3.1349, | |
| "step": 223500 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 9.765976991382705e-06, | |
| "loss": 3.1371, | |
| "step": 224000 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 9.76543020865229e-06, | |
| "loss": 3.1335, | |
| "step": 224500 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 9.764883425921877e-06, | |
| "loss": 3.1328, | |
| "step": 225000 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 9.764336643191462e-06, | |
| "loss": 3.122, | |
| "step": 225500 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 9.763789860461048e-06, | |
| "loss": 3.1268, | |
| "step": 226000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 9.763243077730634e-06, | |
| "loss": 3.1273, | |
| "step": 226500 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 9.76269629500022e-06, | |
| "loss": 3.1256, | |
| "step": 227000 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 9.762149512269806e-06, | |
| "loss": 3.1171, | |
| "step": 227500 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 9.761602729539391e-06, | |
| "loss": 3.1158, | |
| "step": 228000 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 9.761055946808977e-06, | |
| "loss": 3.1178, | |
| "step": 228500 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 9.760509164078563e-06, | |
| "loss": 3.1131, | |
| "step": 229000 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 9.759962381348149e-06, | |
| "loss": 3.112, | |
| "step": 229500 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 9.759415598617735e-06, | |
| "loss": 3.1148, | |
| "step": 230000 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 9.75886881588732e-06, | |
| "loss": 3.1016, | |
| "step": 230500 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 9.758322033156904e-06, | |
| "loss": 3.0997, | |
| "step": 231000 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 9.757775250426492e-06, | |
| "loss": 3.0994, | |
| "step": 231500 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 9.757228467696078e-06, | |
| "loss": 3.1081, | |
| "step": 232000 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 9.756681684965663e-06, | |
| "loss": 3.0964, | |
| "step": 232500 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "step": 232807, | |
| "total_flos": 2.95255140264942e+18, | |
| "train_loss": 4.630338809797302, | |
| "train_runtime": 197999.0445, | |
| "train_samples_per_second": 2958.983, | |
| "train_steps_per_second": 46.235 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 9154400, | |
| "num_train_epochs": 100, | |
| "save_steps": 1000000, | |
| "total_flos": 2.95255140264942e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |