diff --git "a/trainer_state.json" "b/trainer_state.json" --- "a/trainer_state.json" +++ "b/trainer_state.json" @@ -1,15025 +1,15037 @@ { "best_metric": null, "best_model_checkpoint": null, - "epoch": 0.9999599535461134, + "epoch": 1.0, "eval_steps": 500, - "global_step": 12485, + "global_step": 12498, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, - "learning_rate": 1.6012810248198559e-07, - "loss": 1.6623, + "learning_rate": 1.6e-07, + "loss": 1.6925, "step": 1 }, { "epoch": 0.0, - "learning_rate": 8.00640512409928e-07, - "loss": 1.6771, + "learning_rate": 8.000000000000001e-07, + "loss": 1.6908, "step": 5 }, { "epoch": 0.0, - "learning_rate": 1.601281024819856e-06, - "loss": 1.6806, + "learning_rate": 1.6000000000000001e-06, + "loss": 1.6058, "step": 10 }, { "epoch": 0.0, - "learning_rate": 2.401921537229784e-06, - "loss": 1.6667, + "learning_rate": 2.4000000000000003e-06, + "loss": 1.6401, "step": 15 }, { "epoch": 0.0, - "learning_rate": 3.202562049639712e-06, - "loss": 1.6919, + "learning_rate": 3.2000000000000003e-06, + "loss": 1.5414, "step": 20 }, { "epoch": 0.0, - "learning_rate": 4.00320256204964e-06, - "loss": 1.6029, + "learning_rate": 4.000000000000001e-06, + "loss": 1.6556, "step": 25 }, { "epoch": 0.0, - "learning_rate": 4.803843074459568e-06, - "loss": 1.6437, + "learning_rate": 4.800000000000001e-06, + "loss": 1.6259, "step": 30 }, { "epoch": 0.0, - "learning_rate": 5.604483586869496e-06, - "loss": 1.5762, + "learning_rate": 5.600000000000001e-06, + "loss": 1.5553, "step": 35 }, { "epoch": 0.0, - "learning_rate": 6.405124099279424e-06, - "loss": 1.5614, + "learning_rate": 6.4000000000000006e-06, + "loss": 1.5815, "step": 40 }, { "epoch": 0.0, - "learning_rate": 7.205764611689351e-06, - "loss": 1.4336, + "learning_rate": 7.2e-06, + "loss": 1.4601, "step": 45 }, { "epoch": 0.0, - "learning_rate": 8.00640512409928e-06, - "loss": 1.4509, + "learning_rate": 8.000000000000001e-06, + "loss": 1.3965, "step": 50 }, { "epoch": 0.0, - "learning_rate": 8.807045636509207e-06, - "loss": 1.3651, + "learning_rate": 8.8e-06, + "loss": 1.3905, "step": 55 }, { "epoch": 0.0, - "learning_rate": 9.607686148919135e-06, - "loss": 1.3681, + "learning_rate": 9.600000000000001e-06, + "loss": 1.4426, "step": 60 }, { "epoch": 0.01, - "learning_rate": 1.0408326661329064e-05, - "loss": 1.3529, + "learning_rate": 1.04e-05, + "loss": 1.3069, "step": 65 }, { "epoch": 0.01, - "learning_rate": 1.1208967173738992e-05, - "loss": 1.2711, + "learning_rate": 1.1200000000000001e-05, + "loss": 1.3484, "step": 70 }, { "epoch": 0.01, - "learning_rate": 1.200960768614892e-05, - "loss": 1.2849, + "learning_rate": 1.2e-05, + "loss": 1.3249, "step": 75 }, { "epoch": 0.01, - "learning_rate": 1.2810248198558847e-05, - "loss": 1.2732, + "learning_rate": 1.2800000000000001e-05, + "loss": 1.3055, "step": 80 }, { "epoch": 0.01, - "learning_rate": 1.3610888710968776e-05, - "loss": 1.272, + "learning_rate": 1.3600000000000002e-05, + "loss": 1.246, "step": 85 }, { "epoch": 0.01, - "learning_rate": 1.4411529223378702e-05, - "loss": 1.2058, + "learning_rate": 1.44e-05, + "loss": 1.2003, "step": 90 }, { "epoch": 0.01, - "learning_rate": 1.521216973578863e-05, - "loss": 1.186, + "learning_rate": 1.52e-05, + "loss": 1.2233, "step": 95 }, { "epoch": 0.01, - "learning_rate": 1.601281024819856e-05, - "loss": 1.2082, + "learning_rate": 1.6000000000000003e-05, + "loss": 1.2593, "step": 100 }, { "epoch": 0.01, - "learning_rate": 1.6813450760608486e-05, - "loss": 1.134, + "learning_rate": 1.6800000000000002e-05, + "loss": 1.1586, "step": 105 }, { "epoch": 0.01, - "learning_rate": 1.7614091273018414e-05, - "loss": 1.165, + "learning_rate": 1.76e-05, + "loss": 1.1819, "step": 110 }, { "epoch": 0.01, - "learning_rate": 1.8414731785428343e-05, - "loss": 1.1964, + "learning_rate": 1.84e-05, + "loss": 1.1945, "step": 115 }, { "epoch": 0.01, - "learning_rate": 1.921537229783827e-05, - "loss": 1.1946, + "learning_rate": 1.9200000000000003e-05, + "loss": 1.1622, "step": 120 }, { "epoch": 0.01, - "learning_rate": 2.00160128102482e-05, - "loss": 1.2097, + "learning_rate": 2e-05, + "loss": 1.1232, "step": 125 }, { "epoch": 0.01, - "learning_rate": 2.0816653322658128e-05, - "loss": 1.1597, + "learning_rate": 2.08e-05, + "loss": 1.1616, "step": 130 }, { "epoch": 0.01, - "learning_rate": 2.1617293835068056e-05, - "loss": 1.1705, + "learning_rate": 2.16e-05, + "loss": 1.1588, "step": 135 }, { "epoch": 0.01, - "learning_rate": 2.2417934347477984e-05, - "loss": 1.165, + "learning_rate": 2.2400000000000002e-05, + "loss": 1.223, "step": 140 }, { "epoch": 0.01, - "learning_rate": 2.3218574859887913e-05, - "loss": 1.0927, + "learning_rate": 2.32e-05, + "loss": 1.1289, "step": 145 }, { "epoch": 0.01, - "learning_rate": 2.401921537229784e-05, - "loss": 1.1754, + "learning_rate": 2.4e-05, + "loss": 1.1866, "step": 150 }, { "epoch": 0.01, - "learning_rate": 2.481985588470777e-05, - "loss": 1.0777, + "learning_rate": 2.48e-05, + "loss": 1.0952, "step": 155 }, { "epoch": 0.01, - "learning_rate": 2.5620496397117695e-05, - "loss": 1.1024, + "learning_rate": 2.5600000000000002e-05, + "loss": 1.1546, "step": 160 }, { "epoch": 0.01, - "learning_rate": 2.6421136909527623e-05, - "loss": 1.1271, + "learning_rate": 2.64e-05, + "loss": 1.0909, "step": 165 }, { "epoch": 0.01, - "learning_rate": 2.722177742193755e-05, - "loss": 1.0962, + "learning_rate": 2.7200000000000004e-05, + "loss": 1.1519, "step": 170 }, { "epoch": 0.01, - "learning_rate": 2.802241793434748e-05, - "loss": 1.1478, + "learning_rate": 2.8000000000000003e-05, + "loss": 1.1629, "step": 175 }, { "epoch": 0.01, - "learning_rate": 2.8823058446757405e-05, - "loss": 1.141, + "learning_rate": 2.88e-05, + "loss": 1.0833, "step": 180 }, { "epoch": 0.01, - "learning_rate": 2.9623698959167333e-05, - "loss": 1.0936, + "learning_rate": 2.96e-05, + "loss": 1.0933, "step": 185 }, { "epoch": 0.02, - "learning_rate": 3.042433947157726e-05, - "loss": 1.1054, + "learning_rate": 3.04e-05, + "loss": 1.1094, "step": 190 }, { "epoch": 0.02, - "learning_rate": 3.122497998398719e-05, - "loss": 1.0977, + "learning_rate": 3.12e-05, + "loss": 1.0981, "step": 195 }, { "epoch": 0.02, - "learning_rate": 3.202562049639712e-05, - "loss": 1.1036, + "learning_rate": 3.2000000000000005e-05, + "loss": 1.1094, "step": 200 }, { "epoch": 0.02, - "learning_rate": 3.282626100880705e-05, - "loss": 1.0642, + "learning_rate": 3.2800000000000004e-05, + "loss": 1.1305, "step": 205 }, { "epoch": 0.02, - "learning_rate": 3.362690152121697e-05, - "loss": 1.1187, + "learning_rate": 3.3600000000000004e-05, + "loss": 1.163, "step": 210 }, { "epoch": 0.02, - "learning_rate": 3.44275420336269e-05, - "loss": 1.1119, + "learning_rate": 3.4399999999999996e-05, + "loss": 1.0701, "step": 215 }, { "epoch": 0.02, - "learning_rate": 3.522818254603683e-05, - "loss": 1.032, + "learning_rate": 3.52e-05, + "loss": 1.1082, "step": 220 }, { "epoch": 0.02, - "learning_rate": 3.602882305844676e-05, - "loss": 1.1423, + "learning_rate": 3.6e-05, + "loss": 1.1047, "step": 225 }, { "epoch": 0.02, - "learning_rate": 3.6829463570856685e-05, - "loss": 1.098, + "learning_rate": 3.68e-05, + "loss": 1.0775, "step": 230 }, { "epoch": 0.02, - "learning_rate": 3.763010408326661e-05, - "loss": 1.0508, + "learning_rate": 3.76e-05, + "loss": 1.0998, "step": 235 }, { "epoch": 0.02, - "learning_rate": 3.843074459567654e-05, - "loss": 1.0716, + "learning_rate": 3.8400000000000005e-05, + "loss": 1.0897, "step": 240 }, { "epoch": 0.02, - "learning_rate": 3.923138510808647e-05, - "loss": 1.1234, + "learning_rate": 3.9200000000000004e-05, + "loss": 1.0904, "step": 245 }, { "epoch": 0.02, - "learning_rate": 4.00320256204964e-05, - "loss": 1.0516, + "learning_rate": 4e-05, + "loss": 1.0534, "step": 250 }, { "epoch": 0.02, - "learning_rate": 4.083266613290633e-05, - "loss": 1.0492, + "learning_rate": 4.08e-05, + "loss": 1.1006, "step": 255 }, { "epoch": 0.02, - "learning_rate": 4.1633306645316255e-05, - "loss": 1.0435, + "learning_rate": 4.16e-05, + "loss": 1.0676, "step": 260 }, { "epoch": 0.02, - "learning_rate": 4.2433947157726184e-05, - "loss": 1.0509, + "learning_rate": 4.24e-05, + "loss": 1.1178, "step": 265 }, { "epoch": 0.02, - "learning_rate": 4.323458767013611e-05, - "loss": 1.0632, + "learning_rate": 4.32e-05, + "loss": 1.0918, "step": 270 }, { "epoch": 0.02, - "learning_rate": 4.403522818254604e-05, - "loss": 1.0971, + "learning_rate": 4.4000000000000006e-05, + "loss": 1.0741, "step": 275 }, { "epoch": 0.02, - "learning_rate": 4.483586869495597e-05, - "loss": 1.0586, + "learning_rate": 4.4800000000000005e-05, + "loss": 1.0507, "step": 280 }, { "epoch": 0.02, - "learning_rate": 4.56365092073659e-05, - "loss": 0.9929, + "learning_rate": 4.5600000000000004e-05, + "loss": 1.1007, "step": 285 }, { "epoch": 0.02, - "learning_rate": 4.6437149719775826e-05, - "loss": 1.1185, + "learning_rate": 4.64e-05, + "loss": 1.0899, "step": 290 }, { "epoch": 0.02, - "learning_rate": 4.7237790232185754e-05, - "loss": 1.0587, + "learning_rate": 4.72e-05, + "loss": 1.0392, "step": 295 }, { "epoch": 0.02, - "learning_rate": 4.803843074459568e-05, - "loss": 1.1125, + "learning_rate": 4.8e-05, + "loss": 1.0691, "step": 300 }, { "epoch": 0.02, - "learning_rate": 4.883907125700561e-05, - "loss": 1.0172, + "learning_rate": 4.88e-05, + "loss": 1.0318, "step": 305 }, { "epoch": 0.02, - "learning_rate": 4.963971176941554e-05, - "loss": 1.0572, + "learning_rate": 4.96e-05, + "loss": 1.0926, "step": 310 }, { "epoch": 0.03, - "learning_rate": 5.044035228182547e-05, - "loss": 1.0543, + "learning_rate": 5.0400000000000005e-05, + "loss": 1.0726, "step": 315 }, { "epoch": 0.03, - "learning_rate": 5.124099279423539e-05, - "loss": 1.1266, + "learning_rate": 5.1200000000000004e-05, + "loss": 1.0455, "step": 320 }, { "epoch": 0.03, - "learning_rate": 5.2041633306645324e-05, - "loss": 1.0596, + "learning_rate": 5.2000000000000004e-05, + "loss": 1.0354, "step": 325 }, { "epoch": 0.03, - "learning_rate": 5.2842273819055246e-05, - "loss": 1.0651, + "learning_rate": 5.28e-05, + "loss": 1.0721, "step": 330 }, { "epoch": 0.03, - "learning_rate": 5.3642914331465174e-05, - "loss": 1.0672, + "learning_rate": 5.360000000000001e-05, + "loss": 1.0352, "step": 335 }, { "epoch": 0.03, - "learning_rate": 5.44435548438751e-05, - "loss": 1.1028, + "learning_rate": 5.440000000000001e-05, + "loss": 1.0516, "step": 340 }, { "epoch": 0.03, - "learning_rate": 5.524419535628503e-05, - "loss": 1.0285, + "learning_rate": 5.520000000000001e-05, + "loss": 1.0784, "step": 345 }, { "epoch": 0.03, - "learning_rate": 5.604483586869496e-05, - "loss": 1.0345, + "learning_rate": 5.6000000000000006e-05, + "loss": 1.0527, "step": 350 }, { "epoch": 0.03, - "learning_rate": 5.684547638110489e-05, - "loss": 1.1026, + "learning_rate": 5.68e-05, + "loss": 1.0519, "step": 355 }, { "epoch": 0.03, - "learning_rate": 5.764611689351481e-05, - "loss": 1.105, + "learning_rate": 5.76e-05, + "loss": 1.0054, "step": 360 }, { "epoch": 0.03, - "learning_rate": 5.8446757405924744e-05, - "loss": 1.0429, + "learning_rate": 5.8399999999999997e-05, + "loss": 1.0413, "step": 365 }, { "epoch": 0.03, - "learning_rate": 5.9247397918334666e-05, - "loss": 1.0391, + "learning_rate": 5.92e-05, + "loss": 1.0658, "step": 370 }, { "epoch": 0.03, - "learning_rate": 6.00480384307446e-05, - "loss": 1.0379, + "learning_rate": 6e-05, + "loss": 1.0784, "step": 375 }, { "epoch": 0.03, - "learning_rate": 6.084867894315452e-05, - "loss": 1.0403, + "learning_rate": 6.08e-05, + "loss": 1.0449, "step": 380 }, { "epoch": 0.03, - "learning_rate": 6.164931945556446e-05, - "loss": 1.0939, + "learning_rate": 6.16e-05, + "loss": 1.0649, "step": 385 }, { "epoch": 0.03, - "learning_rate": 6.244995996797439e-05, - "loss": 1.0888, + "learning_rate": 6.24e-05, + "loss": 1.0076, "step": 390 }, { "epoch": 0.03, - "learning_rate": 6.325060048038431e-05, - "loss": 1.1059, + "learning_rate": 6.32e-05, + "loss": 0.975, "step": 395 }, { "epoch": 0.03, - "learning_rate": 6.405124099279424e-05, - "loss": 0.9998, + "learning_rate": 6.400000000000001e-05, + "loss": 1.0469, "step": 400 }, { "epoch": 0.03, - "learning_rate": 6.485188150520416e-05, - "loss": 1.0414, + "learning_rate": 6.48e-05, + "loss": 1.0681, "step": 405 }, { "epoch": 0.03, - "learning_rate": 6.56525220176141e-05, - "loss": 1.0328, + "learning_rate": 6.560000000000001e-05, + "loss": 0.9911, "step": 410 }, { "epoch": 0.03, - "learning_rate": 6.645316253002402e-05, - "loss": 1.1202, + "learning_rate": 6.64e-05, + "loss": 1.0078, "step": 415 }, { "epoch": 0.03, - "learning_rate": 6.725380304243394e-05, - "loss": 1.0936, + "learning_rate": 6.720000000000001e-05, + "loss": 1.0377, "step": 420 }, { "epoch": 0.03, - "learning_rate": 6.805444355484388e-05, - "loss": 1.0689, + "learning_rate": 6.800000000000001e-05, + "loss": 1.0084, "step": 425 }, { "epoch": 0.03, - "learning_rate": 6.88550840672538e-05, - "loss": 1.048, + "learning_rate": 6.879999999999999e-05, + "loss": 1.063, "step": 430 }, { "epoch": 0.03, - "learning_rate": 6.965572457966373e-05, - "loss": 1.0304, + "learning_rate": 6.96e-05, + "loss": 1.0865, "step": 435 }, { "epoch": 0.04, - "learning_rate": 7.045636509207366e-05, - "loss": 1.0755, + "learning_rate": 7.04e-05, + "loss": 1.0285, "step": 440 }, { "epoch": 0.04, - "learning_rate": 7.125700560448359e-05, - "loss": 1.0921, + "learning_rate": 7.12e-05, + "loss": 0.9832, "step": 445 }, { "epoch": 0.04, - "learning_rate": 7.205764611689351e-05, - "loss": 1.008, + "learning_rate": 7.2e-05, + "loss": 1.0256, "step": 450 }, { "epoch": 0.04, - "learning_rate": 7.285828662930345e-05, - "loss": 1.0442, + "learning_rate": 7.280000000000001e-05, + "loss": 1.0703, "step": 455 }, { "epoch": 0.04, - "learning_rate": 7.365892714171337e-05, - "loss": 1.0547, + "learning_rate": 7.36e-05, + "loss": 1.0402, "step": 460 }, { "epoch": 0.04, - "learning_rate": 7.445956765412329e-05, - "loss": 1.0397, + "learning_rate": 7.44e-05, + "loss": 1.0726, "step": 465 }, { "epoch": 0.04, - "learning_rate": 7.526020816653323e-05, - "loss": 1.017, + "learning_rate": 7.52e-05, + "loss": 1.0528, "step": 470 }, { "epoch": 0.04, - "learning_rate": 7.606084867894315e-05, - "loss": 0.9951, + "learning_rate": 7.6e-05, + "loss": 1.0021, "step": 475 }, { "epoch": 0.04, - "learning_rate": 7.686148919135308e-05, - "loss": 1.0623, + "learning_rate": 7.680000000000001e-05, + "loss": 1.0498, "step": 480 }, { "epoch": 0.04, - "learning_rate": 7.7662129703763e-05, - "loss": 1.0354, + "learning_rate": 7.76e-05, + "loss": 1.0084, "step": 485 }, { "epoch": 0.04, - "learning_rate": 7.846277021617294e-05, - "loss": 1.0275, + "learning_rate": 7.840000000000001e-05, + "loss": 0.9683, "step": 490 }, { "epoch": 0.04, - "learning_rate": 7.926341072858286e-05, - "loss": 1.0532, + "learning_rate": 7.920000000000001e-05, + "loss": 1.0322, "step": 495 }, { "epoch": 0.04, - "learning_rate": 8.00640512409928e-05, - "loss": 1.018, + "learning_rate": 8e-05, + "loss": 1.0523, "step": 500 }, { "epoch": 0.04, - "learning_rate": 8.086469175340272e-05, - "loss": 1.0007, + "learning_rate": 8.080000000000001e-05, + "loss": 1.0343, "step": 505 }, { "epoch": 0.04, - "learning_rate": 8.166533226581265e-05, - "loss": 1.0231, + "learning_rate": 8.16e-05, + "loss": 1.1302, "step": 510 }, { "epoch": 0.04, - "learning_rate": 8.246597277822258e-05, - "loss": 1.0209, + "learning_rate": 8.24e-05, + "loss": 1.0469, "step": 515 }, { "epoch": 0.04, - "learning_rate": 8.326661329063251e-05, - "loss": 1.0401, + "learning_rate": 8.32e-05, + "loss": 1.029, "step": 520 }, { "epoch": 0.04, - "learning_rate": 8.406725380304243e-05, - "loss": 1.1141, + "learning_rate": 8.4e-05, + "loss": 0.9907, "step": 525 }, { "epoch": 0.04, - "learning_rate": 8.486789431545237e-05, - "loss": 1.0459, + "learning_rate": 8.48e-05, + "loss": 1.0229, "step": 530 }, { "epoch": 0.04, - "learning_rate": 8.566853482786229e-05, - "loss": 1.0712, + "learning_rate": 8.560000000000001e-05, + "loss": 1.0206, "step": 535 }, { "epoch": 0.04, - "learning_rate": 8.646917534027222e-05, - "loss": 1.0464, + "learning_rate": 8.64e-05, + "loss": 1.0176, "step": 540 }, { "epoch": 0.04, - "learning_rate": 8.726981585268215e-05, - "loss": 1.0192, + "learning_rate": 8.72e-05, + "loss": 1.0113, "step": 545 }, { "epoch": 0.04, - "learning_rate": 8.807045636509208e-05, - "loss": 1.06, + "learning_rate": 8.800000000000001e-05, + "loss": 1.034, "step": 550 }, { "epoch": 0.04, - "learning_rate": 8.8871096877502e-05, - "loss": 1.0569, + "learning_rate": 8.88e-05, + "loss": 1.0229, "step": 555 }, { "epoch": 0.04, - "learning_rate": 8.967173738991194e-05, - "loss": 1.0417, + "learning_rate": 8.960000000000001e-05, + "loss": 1.027, "step": 560 }, { "epoch": 0.05, - "learning_rate": 9.047237790232186e-05, - "loss": 1.0205, + "learning_rate": 9.04e-05, + "loss": 1.0275, "step": 565 }, { "epoch": 0.05, - "learning_rate": 9.12730184147318e-05, - "loss": 1.0838, + "learning_rate": 9.120000000000001e-05, + "loss": 1.0178, "step": 570 }, { "epoch": 0.05, - "learning_rate": 9.207365892714172e-05, - "loss": 1.0194, + "learning_rate": 9.200000000000001e-05, + "loss": 1.0162, "step": 575 }, { "epoch": 0.05, - "learning_rate": 9.287429943955165e-05, - "loss": 1.0663, + "learning_rate": 9.28e-05, + "loss": 0.9921, "step": 580 }, { "epoch": 0.05, - "learning_rate": 9.367493995196157e-05, - "loss": 1.0545, + "learning_rate": 9.360000000000001e-05, + "loss": 1.0447, "step": 585 }, { "epoch": 0.05, - "learning_rate": 9.447558046437151e-05, - "loss": 1.0227, + "learning_rate": 9.44e-05, + "loss": 1.0392, "step": 590 }, { "epoch": 0.05, - "learning_rate": 9.527622097678143e-05, - "loss": 1.0771, + "learning_rate": 9.52e-05, + "loss": 0.9444, "step": 595 }, { "epoch": 0.05, - "learning_rate": 9.607686148919136e-05, - "loss": 1.0796, + "learning_rate": 9.6e-05, + "loss": 1.0127, "step": 600 }, { "epoch": 0.05, - "learning_rate": 9.687750200160129e-05, - "loss": 0.977, + "learning_rate": 9.680000000000001e-05, + "loss": 1.0034, "step": 605 }, { "epoch": 0.05, - "learning_rate": 9.767814251401122e-05, - "loss": 1.0751, + "learning_rate": 9.76e-05, + "loss": 1.0786, "step": 610 }, { "epoch": 0.05, - "learning_rate": 9.847878302642114e-05, - "loss": 1.0557, + "learning_rate": 9.84e-05, + "loss": 1.0179, "step": 615 }, { "epoch": 0.05, - "learning_rate": 9.927942353883108e-05, - "loss": 1.0361, + "learning_rate": 9.92e-05, + "loss": 1.0398, "step": 620 }, { "epoch": 0.05, - "learning_rate": 0.000100080064051241, - "loss": 1.0588, + "learning_rate": 0.0001, + "loss": 1.0504, "step": 625 }, { "epoch": 0.05, - "learning_rate": 0.00010088070456365093, - "loss": 1.0298, + "learning_rate": 0.00010080000000000001, + "loss": 1.0333, "step": 630 }, { "epoch": 0.05, - "learning_rate": 0.00010168134507606087, - "loss": 1.0503, + "learning_rate": 0.0001016, + "loss": 1.0045, "step": 635 }, { "epoch": 0.05, - "learning_rate": 0.00010248198558847078, - "loss": 1.0648, + "learning_rate": 0.00010240000000000001, + "loss": 1.0196, "step": 640 }, { "epoch": 0.05, - "learning_rate": 0.00010328262610088071, - "loss": 1.0462, + "learning_rate": 0.0001032, + "loss": 0.9236, "step": 645 }, { "epoch": 0.05, - "learning_rate": 0.00010408326661329065, - "loss": 0.9897, + "learning_rate": 0.00010400000000000001, + "loss": 1.0332, "step": 650 }, { "epoch": 0.05, - "learning_rate": 0.00010488390712570057, - "loss": 1.0009, + "learning_rate": 0.00010480000000000001, + "loss": 1.0518, "step": 655 }, { "epoch": 0.05, - "learning_rate": 0.00010568454763811049, - "loss": 1.0234, + "learning_rate": 0.0001056, + "loss": 0.9689, "step": 660 }, { "epoch": 0.05, - "learning_rate": 0.00010648518815052043, - "loss": 1.0592, + "learning_rate": 0.00010640000000000001, + "loss": 1.0171, "step": 665 }, { "epoch": 0.05, - "learning_rate": 0.00010728582866293035, - "loss": 1.0466, + "learning_rate": 0.00010720000000000002, + "loss": 0.9967, "step": 670 }, { "epoch": 0.05, - "learning_rate": 0.00010808646917534028, - "loss": 1.1249, + "learning_rate": 0.00010800000000000001, + "loss": 1.0526, "step": 675 }, { "epoch": 0.05, - "learning_rate": 0.0001088871096877502, - "loss": 1.0349, + "learning_rate": 0.00010880000000000002, + "loss": 1.0585, "step": 680 }, { "epoch": 0.05, - "learning_rate": 0.00010968775020016013, - "loss": 1.0189, + "learning_rate": 0.00010960000000000001, + "loss": 0.9914, "step": 685 }, { "epoch": 0.06, - "learning_rate": 0.00011048839071257006, - "loss": 1.0976, + "learning_rate": 0.00011040000000000001, + "loss": 1.0017, "step": 690 }, { "epoch": 0.06, - "learning_rate": 0.00011128903122497998, - "loss": 1.0672, + "learning_rate": 0.00011120000000000002, + "loss": 1.0515, "step": 695 }, { "epoch": 0.06, - "learning_rate": 0.00011208967173738992, - "loss": 1.0155, + "learning_rate": 0.00011200000000000001, + "loss": 1.0085, "step": 700 }, { "epoch": 0.06, - "learning_rate": 0.00011289031224979984, - "loss": 1.0423, + "learning_rate": 0.00011279999999999999, + "loss": 1.0306, "step": 705 }, { "epoch": 0.06, - "learning_rate": 0.00011369095276220978, - "loss": 1.0517, + "learning_rate": 0.0001136, + "loss": 1.0018, "step": 710 }, { "epoch": 0.06, - "learning_rate": 0.0001144915932746197, - "loss": 1.0412, + "learning_rate": 0.0001144, + "loss": 0.9802, "step": 715 }, { "epoch": 0.06, - "learning_rate": 0.00011529223378702962, - "loss": 1.0437, + "learning_rate": 0.0001152, + "loss": 1.0472, "step": 720 }, { "epoch": 0.06, - "learning_rate": 0.00011609287429943955, - "loss": 1.0585, + "learning_rate": 0.000116, + "loss": 1.1019, "step": 725 }, { "epoch": 0.06, - "learning_rate": 0.00011689351481184949, - "loss": 0.9687, + "learning_rate": 0.00011679999999999999, + "loss": 1.0291, "step": 730 }, { "epoch": 0.06, - "learning_rate": 0.0001176941553242594, - "loss": 1.0052, + "learning_rate": 0.0001176, + "loss": 1.055, "step": 735 }, { "epoch": 0.06, - "learning_rate": 0.00011849479583666933, - "loss": 0.9824, + "learning_rate": 0.0001184, + "loss": 0.984, "step": 740 }, { "epoch": 0.06, - "learning_rate": 0.00011929543634907927, - "loss": 0.9841, + "learning_rate": 0.0001192, + "loss": 0.9945, "step": 745 }, { "epoch": 0.06, - "learning_rate": 0.0001200960768614892, - "loss": 1.0231, + "learning_rate": 0.00012, + "loss": 0.999, "step": 750 }, { "epoch": 0.06, - "learning_rate": 0.00012089671737389911, - "loss": 1.077, + "learning_rate": 0.0001208, + "loss": 0.9984, "step": 755 }, { "epoch": 0.06, - "learning_rate": 0.00012169735788630905, - "loss": 1.0367, + "learning_rate": 0.0001216, + "loss": 0.995, "step": 760 }, { "epoch": 0.06, - "learning_rate": 0.00012249799839871898, - "loss": 1.0583, + "learning_rate": 0.0001224, + "loss": 0.9959, "step": 765 }, { "epoch": 0.06, - "learning_rate": 0.00012329863891112893, - "loss": 1.0012, + "learning_rate": 0.0001232, + "loss": 1.0245, "step": 770 }, { "epoch": 0.06, - "learning_rate": 0.00012409927942353882, - "loss": 1.0718, + "learning_rate": 0.000124, + "loss": 1.0563, "step": 775 }, { "epoch": 0.06, - "learning_rate": 0.00012489991993594877, - "loss": 0.989, + "learning_rate": 0.0001248, + "loss": 0.995, "step": 780 }, { "epoch": 0.06, - "learning_rate": 0.0001257005604483587, - "loss": 0.9854, + "learning_rate": 0.00012560000000000002, + "loss": 1.0595, "step": 785 }, { "epoch": 0.06, - "learning_rate": 0.00012650120096076862, - "loss": 0.998, + "learning_rate": 0.0001264, + "loss": 0.9798, "step": 790 }, { "epoch": 0.06, - "learning_rate": 0.00012730184147317854, - "loss": 1.037, + "learning_rate": 0.0001272, + "loss": 0.9994, "step": 795 }, { "epoch": 0.06, - "learning_rate": 0.00012810248198558849, - "loss": 0.9877, + "learning_rate": 0.00012800000000000002, + "loss": 1.0513, "step": 800 }, { "epoch": 0.06, - "learning_rate": 0.0001289031224979984, - "loss": 1.0175, + "learning_rate": 0.00012880000000000001, + "loss": 1.0063, "step": 805 }, { "epoch": 0.06, - "learning_rate": 0.00012970376301040833, - "loss": 1.0293, + "learning_rate": 0.0001296, + "loss": 1.0452, "step": 810 }, { "epoch": 0.07, - "learning_rate": 0.00013050440352281825, - "loss": 1.0239, + "learning_rate": 0.0001304, + "loss": 1.0202, "step": 815 }, { "epoch": 0.07, - "learning_rate": 0.0001313050440352282, - "loss": 1.031, + "learning_rate": 0.00013120000000000002, + "loss": 0.9767, "step": 820 }, { "epoch": 0.07, - "learning_rate": 0.00013210568454763812, - "loss": 1.007, + "learning_rate": 0.000132, + "loss": 1.0252, "step": 825 }, { "epoch": 0.07, - "learning_rate": 0.00013290632506004804, - "loss": 0.9917, + "learning_rate": 0.0001328, + "loss": 1.0517, "step": 830 }, { "epoch": 0.07, - "learning_rate": 0.00013370696557245796, - "loss": 1.0026, + "learning_rate": 0.00013360000000000002, + "loss": 1.0309, "step": 835 }, { "epoch": 0.07, - "learning_rate": 0.00013450760608486789, - "loss": 1.0309, + "learning_rate": 0.00013440000000000001, + "loss": 0.9946, "step": 840 }, { "epoch": 0.07, - "learning_rate": 0.00013530824659727783, - "loss": 1.0287, + "learning_rate": 0.0001352, + "loss": 1.1294, "step": 845 }, { "epoch": 0.07, - "learning_rate": 0.00013610888710968776, - "loss": 1.0503, + "learning_rate": 0.00013600000000000003, + "loss": 1.0405, "step": 850 }, { "epoch": 0.07, - "learning_rate": 0.00013690952762209768, - "loss": 0.9605, + "learning_rate": 0.00013680000000000002, + "loss": 1.0208, "step": 855 }, { "epoch": 0.07, - "learning_rate": 0.0001377101681345076, - "loss": 1.0747, + "learning_rate": 0.00013759999999999998, + "loss": 1.0109, "step": 860 }, { "epoch": 0.07, - "learning_rate": 0.00013851080864691755, - "loss": 1.0208, + "learning_rate": 0.0001384, + "loss": 0.9908, "step": 865 }, { "epoch": 0.07, - "learning_rate": 0.00013931144915932747, - "loss": 1.0154, + "learning_rate": 0.0001392, + "loss": 0.9983, "step": 870 }, { "epoch": 0.07, - "learning_rate": 0.0001401120896717374, - "loss": 1.0408, + "learning_rate": 0.00014, + "loss": 0.9955, "step": 875 }, { "epoch": 0.07, - "learning_rate": 0.0001409127301841473, - "loss": 1.055, + "learning_rate": 0.0001408, + "loss": 0.99, "step": 880 }, { "epoch": 0.07, - "learning_rate": 0.00014171337069655726, - "loss": 0.9696, + "learning_rate": 0.0001416, + "loss": 1.0459, "step": 885 }, { "epoch": 0.07, - "learning_rate": 0.00014251401120896718, - "loss": 1.0553, + "learning_rate": 0.0001424, + "loss": 1.0251, "step": 890 }, { "epoch": 0.07, - "learning_rate": 0.0001433146517213771, - "loss": 0.9832, + "learning_rate": 0.0001432, + "loss": 1.0181, "step": 895 }, { "epoch": 0.07, - "learning_rate": 0.00014411529223378703, - "loss": 1.0525, + "learning_rate": 0.000144, + "loss": 1.0714, "step": 900 }, { "epoch": 0.07, - "learning_rate": 0.00014491593274619698, - "loss": 1.0249, + "learning_rate": 0.0001448, + "loss": 1.016, "step": 905 }, { "epoch": 0.07, - "learning_rate": 0.0001457165732586069, - "loss": 1.0224, + "learning_rate": 0.00014560000000000002, + "loss": 1.088, "step": 910 }, { "epoch": 0.07, - "learning_rate": 0.00014651721377101682, - "loss": 1.0514, + "learning_rate": 0.0001464, + "loss": 0.9982, "step": 915 }, { "epoch": 0.07, - "learning_rate": 0.00014731785428342674, - "loss": 1.0654, + "learning_rate": 0.0001472, + "loss": 1.0553, "step": 920 }, { "epoch": 0.07, - "learning_rate": 0.0001481184947958367, - "loss": 1.0345, + "learning_rate": 0.000148, + "loss": 1.0301, "step": 925 }, { "epoch": 0.07, - "learning_rate": 0.00014891913530824658, - "loss": 1.0993, + "learning_rate": 0.0001488, + "loss": 1.0629, "step": 930 }, { "epoch": 0.07, - "learning_rate": 0.00014971977582065653, - "loss": 1.0389, + "learning_rate": 0.0001496, + "loss": 0.989, "step": 935 }, { "epoch": 0.08, - "learning_rate": 0.00015052041633306645, - "loss": 0.9712, + "learning_rate": 0.0001504, + "loss": 1.0026, "step": 940 }, { "epoch": 0.08, - "learning_rate": 0.0001513210568454764, - "loss": 1.0093, + "learning_rate": 0.00015120000000000002, + "loss": 1.0505, "step": 945 }, { "epoch": 0.08, - "learning_rate": 0.0001521216973578863, - "loss": 1.0086, + "learning_rate": 0.000152, + "loss": 1.0157, "step": 950 }, { "epoch": 0.08, - "learning_rate": 0.00015292233787029625, - "loss": 1.0656, + "learning_rate": 0.0001528, + "loss": 1.0415, "step": 955 }, { "epoch": 0.08, - "learning_rate": 0.00015372297838270617, - "loss": 0.9952, + "learning_rate": 0.00015360000000000002, + "loss": 1.0231, "step": 960 }, { "epoch": 0.08, - "learning_rate": 0.00015452361889511612, - "loss": 0.9933, + "learning_rate": 0.0001544, + "loss": 0.9929, "step": 965 }, { "epoch": 0.08, - "learning_rate": 0.000155324259407526, - "loss": 1.028, + "learning_rate": 0.0001552, + "loss": 0.9767, "step": 970 }, { "epoch": 0.08, - "learning_rate": 0.00015612489991993596, - "loss": 1.0097, + "learning_rate": 0.00015600000000000002, + "loss": 1.0371, "step": 975 }, { "epoch": 0.08, - "learning_rate": 0.00015692554043234588, - "loss": 1.0424, + "learning_rate": 0.00015680000000000002, + "loss": 1.0723, "step": 980 }, { "epoch": 0.08, - "learning_rate": 0.00015772618094475583, - "loss": 1.0332, + "learning_rate": 0.0001576, + "loss": 1.0196, "step": 985 }, { "epoch": 0.08, - "learning_rate": 0.00015852682145716572, - "loss": 1.0749, + "learning_rate": 0.00015840000000000003, + "loss": 1.0009, "step": 990 }, { "epoch": 0.08, - "learning_rate": 0.00015932746196957567, - "loss": 1.0631, + "learning_rate": 0.00015920000000000002, + "loss": 1.0352, "step": 995 }, { "epoch": 0.08, - "learning_rate": 0.0001601281024819856, - "loss": 1.0086, + "learning_rate": 0.00016, + "loss": 1.0163, "step": 1000 }, { "epoch": 0.08, - "learning_rate": 0.00016092874299439554, - "loss": 1.0344, + "learning_rate": 0.0001608, + "loss": 1.0431, "step": 1005 }, { "epoch": 0.08, - "learning_rate": 0.00016172938350680544, - "loss": 1.0144, + "learning_rate": 0.00016160000000000002, + "loss": 1.0825, "step": 1010 }, { "epoch": 0.08, - "learning_rate": 0.00016253002401921539, - "loss": 1.0802, + "learning_rate": 0.00016240000000000002, + "loss": 1.0459, "step": 1015 }, { "epoch": 0.08, - "learning_rate": 0.0001633306645316253, - "loss": 0.989, + "learning_rate": 0.0001632, + "loss": 1.0248, "step": 1020 }, { "epoch": 0.08, - "learning_rate": 0.00016413130504403526, - "loss": 0.99, + "learning_rate": 0.000164, + "loss": 1.0347, "step": 1025 }, { "epoch": 0.08, - "learning_rate": 0.00016493194555644515, - "loss": 0.9939, + "learning_rate": 0.0001648, + "loss": 1.0272, "step": 1030 }, { "epoch": 0.08, - "learning_rate": 0.0001657325860688551, - "loss": 1.0655, + "learning_rate": 0.0001656, + "loss": 1.0378, "step": 1035 }, { "epoch": 0.08, - "learning_rate": 0.00016653322658126502, - "loss": 1.0035, + "learning_rate": 0.0001664, + "loss": 1.0948, "step": 1040 }, { "epoch": 0.08, - "learning_rate": 0.00016733386709367497, - "loss": 0.9928, + "learning_rate": 0.0001672, + "loss": 1.0291, "step": 1045 }, { "epoch": 0.08, - "learning_rate": 0.00016813450760608486, - "loss": 1.0517, + "learning_rate": 0.000168, + "loss": 1.0621, "step": 1050 }, { "epoch": 0.08, - "learning_rate": 0.0001689351481184948, - "loss": 0.9907, + "learning_rate": 0.0001688, + "loss": 1.0189, "step": 1055 }, { "epoch": 0.08, - "learning_rate": 0.00016973578863090473, - "loss": 1.0207, + "learning_rate": 0.0001696, + "loss": 0.9905, "step": 1060 }, { "epoch": 0.09, - "learning_rate": 0.00017053642914331466, - "loss": 0.9894, + "learning_rate": 0.0001704, + "loss": 1.0026, "step": 1065 }, { "epoch": 0.09, - "learning_rate": 0.00017133706965572458, - "loss": 1.0296, + "learning_rate": 0.00017120000000000001, + "loss": 1.0166, "step": 1070 }, { "epoch": 0.09, - "learning_rate": 0.00017213771016813453, - "loss": 0.9827, + "learning_rate": 0.000172, + "loss": 0.9838, "step": 1075 }, { "epoch": 0.09, - "learning_rate": 0.00017293835068054445, - "loss": 0.9915, + "learning_rate": 0.0001728, + "loss": 0.9964, "step": 1080 }, { "epoch": 0.09, - "learning_rate": 0.00017373899119295437, - "loss": 0.9892, + "learning_rate": 0.00017360000000000002, + "loss": 1.0088, "step": 1085 }, { "epoch": 0.09, - "learning_rate": 0.0001745396317053643, - "loss": 0.9987, + "learning_rate": 0.0001744, + "loss": 1.0349, "step": 1090 }, { "epoch": 0.09, - "learning_rate": 0.0001753402722177742, - "loss": 0.9734, + "learning_rate": 0.0001752, + "loss": 0.9985, "step": 1095 }, { "epoch": 0.09, - "learning_rate": 0.00017614091273018416, - "loss": 1.0366, + "learning_rate": 0.00017600000000000002, + "loss": 1.0135, "step": 1100 }, { "epoch": 0.09, - "learning_rate": 0.00017694155324259408, - "loss": 0.9669, + "learning_rate": 0.00017680000000000001, + "loss": 1.0752, "step": 1105 }, { "epoch": 0.09, - "learning_rate": 0.000177742193755004, - "loss": 1.0345, + "learning_rate": 0.0001776, + "loss": 1.0851, "step": 1110 }, { "epoch": 0.09, - "learning_rate": 0.00017854283426741393, - "loss": 1.069, + "learning_rate": 0.0001784, + "loss": 1.0825, "step": 1115 }, { "epoch": 0.09, - "learning_rate": 0.00017934347477982388, - "loss": 0.997, + "learning_rate": 0.00017920000000000002, + "loss": 0.9712, "step": 1120 }, { "epoch": 0.09, - "learning_rate": 0.0001801441152922338, - "loss": 0.9959, + "learning_rate": 0.00018, + "loss": 1.0104, "step": 1125 }, { "epoch": 0.09, - "learning_rate": 0.00018094475580464372, - "loss": 0.9555, + "learning_rate": 0.0001808, + "loss": 1.0264, "step": 1130 }, { "epoch": 0.09, - "learning_rate": 0.00018174539631705364, - "loss": 1.0028, + "learning_rate": 0.00018160000000000002, + "loss": 0.9972, "step": 1135 }, { "epoch": 0.09, - "learning_rate": 0.0001825460368294636, - "loss": 1.0071, + "learning_rate": 0.00018240000000000002, + "loss": 1.0313, "step": 1140 }, { "epoch": 0.09, - "learning_rate": 0.0001833466773418735, - "loss": 1.0238, + "learning_rate": 0.0001832, + "loss": 1.1255, "step": 1145 }, { "epoch": 0.09, - "learning_rate": 0.00018414731785428343, - "loss": 1.0331, + "learning_rate": 0.00018400000000000003, + "loss": 0.9903, "step": 1150 }, { "epoch": 0.09, - "learning_rate": 0.00018494795836669335, - "loss": 1.0205, + "learning_rate": 0.00018480000000000002, + "loss": 0.9698, "step": 1155 }, { "epoch": 0.09, - "learning_rate": 0.0001857485988791033, - "loss": 1.0048, + "learning_rate": 0.0001856, + "loss": 1.0241, "step": 1160 }, { "epoch": 0.09, - "learning_rate": 0.0001865492393915132, - "loss": 0.9968, + "learning_rate": 0.00018640000000000003, + "loss": 0.9969, "step": 1165 }, { "epoch": 0.09, - "learning_rate": 0.00018734987990392315, - "loss": 1.0564, + "learning_rate": 0.00018720000000000002, + "loss": 1.0234, "step": 1170 }, { "epoch": 0.09, - "learning_rate": 0.00018815052041633307, - "loss": 1.0042, + "learning_rate": 0.000188, + "loss": 0.9654, "step": 1175 }, { "epoch": 0.09, - "learning_rate": 0.00018895116092874302, - "loss": 1.0596, + "learning_rate": 0.0001888, + "loss": 1.0585, "step": 1180 }, { "epoch": 0.09, - "learning_rate": 0.0001897518014411529, - "loss": 0.9589, + "learning_rate": 0.0001896, + "loss": 1.0891, "step": 1185 }, { "epoch": 0.1, - "learning_rate": 0.00019055244195356286, - "loss": 1.0561, + "learning_rate": 0.0001904, + "loss": 1.0301, "step": 1190 }, { "epoch": 0.1, - "learning_rate": 0.00019135308246597278, - "loss": 1.0052, + "learning_rate": 0.0001912, + "loss": 0.9896, "step": 1195 }, { "epoch": 0.1, - "learning_rate": 0.00019215372297838273, - "loss": 1.0696, + "learning_rate": 0.000192, + "loss": 0.9893, "step": 1200 }, { "epoch": 0.1, - "learning_rate": 0.00019295436349079262, - "loss": 1.0188, + "learning_rate": 0.0001928, + "loss": 1.0084, "step": 1205 }, { "epoch": 0.1, - "learning_rate": 0.00019375500400320257, - "loss": 0.9802, + "learning_rate": 0.00019360000000000002, + "loss": 1.0332, "step": 1210 }, { "epoch": 0.1, - "learning_rate": 0.0001945556445156125, - "loss": 0.9995, + "learning_rate": 0.0001944, + "loss": 1.0117, "step": 1215 }, { "epoch": 0.1, - "learning_rate": 0.00019535628502802244, - "loss": 1.0077, + "learning_rate": 0.0001952, + "loss": 1.0467, "step": 1220 }, { "epoch": 0.1, - "learning_rate": 0.00019615692554043234, - "loss": 1.0387, + "learning_rate": 0.000196, + "loss": 1.0196, "step": 1225 }, { "epoch": 0.1, - "learning_rate": 0.00019695756605284229, - "loss": 1.0162, + "learning_rate": 0.0001968, + "loss": 1.0205, "step": 1230 }, { "epoch": 0.1, - "learning_rate": 0.0001977582065652522, - "loss": 1.0781, + "learning_rate": 0.0001976, + "loss": 0.9857, "step": 1235 }, { "epoch": 0.1, - "learning_rate": 0.00019855884707766216, - "loss": 0.9585, + "learning_rate": 0.0001984, + "loss": 1.0138, "step": 1240 }, { "epoch": 0.1, - "learning_rate": 0.00019935948759007205, - "loss": 1.0082, + "learning_rate": 0.00019920000000000002, + "loss": 1.0065, "step": 1245 }, { "epoch": 0.1, - "learning_rate": 0.0001999999960911745, - "loss": 0.9976, + "learning_rate": 0.0002, + "loss": 1.0436, "step": 1250 }, { "epoch": 0.1, - "learning_rate": 0.00019999985928231313, - "loss": 1.0361, + "learning_rate": 0.00019999990248777358, + "loss": 1.0583, "step": 1255 }, { "epoch": 0.1, - "learning_rate": 0.000199999527032481, - "loss": 1.0488, + "learning_rate": 0.00019999960995128452, + "loss": 1.0646, "step": 1260 }, { "epoch": 0.1, - "learning_rate": 0.00019999899934232746, - "loss": 0.9971, + "learning_rate": 0.00019999912239110327, + "loss": 1.0488, "step": 1265 }, { "epoch": 0.1, - "learning_rate": 0.00019999827621288384, - "loss": 1.0048, + "learning_rate": 0.0001999984398081807, + "loss": 1.05, "step": 1270 }, { "epoch": 0.1, - "learning_rate": 0.0001999973576455634, - "loss": 1.024, + "learning_rate": 0.00019999756220384808, + "loss": 1.0177, "step": 1275 }, { "epoch": 0.1, - "learning_rate": 0.00019999624364216146, - "loss": 1.0048, + "learning_rate": 0.00019999648957981693, + "loss": 0.9909, "step": 1280 }, { "epoch": 0.1, - "learning_rate": 0.00019999493420485518, - "loss": 1.0489, + "learning_rate": 0.0001999952219381791, + "loss": 1.0648, "step": 1285 }, { "epoch": 0.1, - "learning_rate": 0.00019999342933620375, - "loss": 1.0591, + "learning_rate": 0.0001999937592814068, + "loss": 0.9986, "step": 1290 }, { "epoch": 0.1, - "learning_rate": 0.00019999172903914834, - "loss": 1.0285, + "learning_rate": 0.00019999210161235258, + "loss": 0.9746, "step": 1295 }, { "epoch": 0.1, - "learning_rate": 0.000199989833317012, - "loss": 1.0524, + "learning_rate": 0.00019999024893424934, + "loss": 1.0257, "step": 1300 }, { "epoch": 0.1, - "learning_rate": 0.0001999877421734998, - "loss": 1.0217, + "learning_rate": 0.0001999882012507102, + "loss": 1.0331, "step": 1305 }, { "epoch": 0.1, - "learning_rate": 0.0001999854556126986, - "loss": 0.9936, + "learning_rate": 0.00019998595856572865, + "loss": 0.9908, "step": 1310 }, { "epoch": 0.11, - "learning_rate": 0.0001999829736390774, - "loss": 1.0437, + "learning_rate": 0.0001999835208836785, + "loss": 1.0054, "step": 1315 }, { "epoch": 0.11, - "learning_rate": 0.00019998029625748688, - "loss": 1.0154, + "learning_rate": 0.00019998088820931382, + "loss": 1.0408, "step": 1320 }, { "epoch": 0.11, - "learning_rate": 0.00019997742347315986, - "loss": 0.995, + "learning_rate": 0.00019997806054776896, + "loss": 1.0092, "step": 1325 }, { "epoch": 0.11, - "learning_rate": 0.00019997435529171088, - "loss": 0.9512, + "learning_rate": 0.00019997503790455853, + "loss": 0.9836, "step": 1330 }, { "epoch": 0.11, - "learning_rate": 0.00019997109171913645, - "loss": 1.0562, + "learning_rate": 0.00019997182028557746, + "loss": 1.0519, "step": 1335 }, { "epoch": 0.11, - "learning_rate": 0.00019996763276181496, - "loss": 1.007, + "learning_rate": 0.0001999684076971009, + "loss": 1.0645, "step": 1340 }, { "epoch": 0.11, - "learning_rate": 0.00019996397842650659, - "loss": 1.0412, + "learning_rate": 0.0001999648001457842, + "loss": 1.0933, "step": 1345 }, { "epoch": 0.11, - "learning_rate": 0.00019996012872035344, - "loss": 1.0292, + "learning_rate": 0.00019996099763866298, + "loss": 1.0262, "step": 1350 }, { "epoch": 0.11, - "learning_rate": 0.00019995608365087946, - "loss": 0.9874, + "learning_rate": 0.00019995700018315304, + "loss": 1.0274, "step": 1355 }, { "epoch": 0.11, - "learning_rate": 0.00019995184322599033, - "loss": 1.1157, + "learning_rate": 0.00019995280778705045, + "loss": 1.0472, "step": 1360 }, { "epoch": 0.11, - "learning_rate": 0.0001999474074539736, - "loss": 1.0124, + "learning_rate": 0.00019994842045853136, + "loss": 0.9966, "step": 1365 }, { "epoch": 0.11, - "learning_rate": 0.00019994277634349862, - "loss": 0.9843, + "learning_rate": 0.00019994383820615212, + "loss": 0.9909, "step": 1370 }, { "epoch": 0.11, - "learning_rate": 0.00019993794990361648, - "loss": 0.9999, + "learning_rate": 0.00019993906103884927, + "loss": 1.0132, "step": 1375 }, { "epoch": 0.11, - "learning_rate": 0.00019993292814376003, - "loss": 1.0289, + "learning_rate": 0.0001999340889659395, + "loss": 1.0141, "step": 1380 }, { "epoch": 0.11, - "learning_rate": 0.00019992771107374386, - "loss": 1.0394, + "learning_rate": 0.00019992892199711948, + "loss": 1.0128, "step": 1385 }, { "epoch": 0.11, - "learning_rate": 0.00019992229870376428, - "loss": 0.9849, + "learning_rate": 0.0001999235601424661, + "loss": 1.0671, "step": 1390 }, { "epoch": 0.11, - "learning_rate": 0.0001999166910443993, - "loss": 1.0039, + "learning_rate": 0.0001999180034124363, + "loss": 1.0308, "step": 1395 }, { "epoch": 0.11, - "learning_rate": 0.0001999108881066086, - "loss": 1.0367, + "learning_rate": 0.00019991225181786705, + "loss": 1.0011, "step": 1400 }, { "epoch": 0.11, - "learning_rate": 0.0001999048899017335, - "loss": 0.9905, + "learning_rate": 0.00019990630536997537, + "loss": 1.0435, "step": 1405 }, { "epoch": 0.11, - "learning_rate": 0.00019989869644149698, - "loss": 1.0157, + "learning_rate": 0.0001999001640803583, + "loss": 1.0643, "step": 1410 }, { "epoch": 0.11, - "learning_rate": 0.00019989230773800358, - "loss": 0.9522, + "learning_rate": 0.00019989382796099282, + "loss": 1.032, "step": 1415 }, { "epoch": 0.11, - "learning_rate": 0.0001998857238037395, - "loss": 1.021, + "learning_rate": 0.00019988729702423596, + "loss": 1.0214, "step": 1420 }, { "epoch": 0.11, - "learning_rate": 0.00019987894465157247, - "loss": 1.061, + "learning_rate": 0.0001998805712828246, + "loss": 1.0552, "step": 1425 }, { "epoch": 0.11, - "learning_rate": 0.0001998719702947517, - "loss": 1.0447, + "learning_rate": 0.0001998736507498756, + "loss": 1.0406, "step": 1430 }, { "epoch": 0.11, - "learning_rate": 0.00019986480074690803, - "loss": 0.9754, + "learning_rate": 0.00019986653543888568, + "loss": 0.9748, "step": 1435 }, { "epoch": 0.12, - "learning_rate": 0.00019985743602205368, - "loss": 1.0071, + "learning_rate": 0.00019985922536373146, + "loss": 1.0012, "step": 1440 }, { "epoch": 0.12, - "learning_rate": 0.00019984987613458235, - "loss": 1.0331, + "learning_rate": 0.00019985172053866941, + "loss": 0.9761, "step": 1445 }, { "epoch": 0.12, - "learning_rate": 0.00019984212109926916, - "loss": 1.0518, + "learning_rate": 0.00019984402097833565, + "loss": 0.9752, "step": 1450 }, { "epoch": 0.12, - "learning_rate": 0.00019983417093127068, - "loss": 0.9406, + "learning_rate": 0.00019983612669774632, + "loss": 1.0484, "step": 1455 }, { "epoch": 0.12, - "learning_rate": 0.00019982602564612483, - "loss": 1.0671, + "learning_rate": 0.00019982803771229714, + "loss": 0.9326, "step": 1460 }, { "epoch": 0.12, - "learning_rate": 0.00019981768525975083, - "loss": 0.9864, + "learning_rate": 0.00019981975403776363, + "loss": 1.0509, "step": 1465 }, { "epoch": 0.12, - "learning_rate": 0.00019980914978844923, - "loss": 0.9773, + "learning_rate": 0.000199811275690301, + "loss": 0.9621, "step": 1470 }, { "epoch": 0.12, - "learning_rate": 0.00019980041924890188, - "loss": 1.0339, + "learning_rate": 0.00019980260268644403, + "loss": 0.9773, "step": 1475 }, { "epoch": 0.12, - "learning_rate": 0.00019979149365817182, - "loss": 1.0653, + "learning_rate": 0.0001997937350431073, + "loss": 1.0541, "step": 1480 }, { "epoch": 0.12, - "learning_rate": 0.0001997823730337034, - "loss": 1.0135, + "learning_rate": 0.0001997846727775848, + "loss": 1.0015, "step": 1485 }, { "epoch": 0.12, - "learning_rate": 0.00019977305739332204, - "loss": 1.02, + "learning_rate": 0.00019977541590755023, + "loss": 1.0291, "step": 1490 }, { "epoch": 0.12, - "learning_rate": 0.00019976354675523434, - "loss": 1.0152, + "learning_rate": 0.00019976596445105667, + "loss": 1.0449, "step": 1495 }, { "epoch": 0.12, - "learning_rate": 0.00019975384113802804, - "loss": 0.965, + "learning_rate": 0.00019975631842653684, + "loss": 1.0382, "step": 1500 }, { "epoch": 0.12, - "learning_rate": 0.00019974394056067188, - "loss": 1.0503, + "learning_rate": 0.0001997464778528028, + "loss": 1.0431, "step": 1505 }, { "epoch": 0.12, - "learning_rate": 0.00019973384504251568, - "loss": 1.0126, + "learning_rate": 0.0001997364427490461, + "loss": 0.9537, "step": 1510 }, { "epoch": 0.12, - "learning_rate": 0.00019972355460329028, - "loss": 0.9731, + "learning_rate": 0.0001997262131348377, + "loss": 1.0238, "step": 1515 }, { "epoch": 0.12, - "learning_rate": 0.00019971306926310742, - "loss": 1.0684, + "learning_rate": 0.00019971578903012774, + "loss": 0.9975, "step": 1520 }, { "epoch": 0.12, - "learning_rate": 0.00019970238904245977, - "loss": 0.9686, + "learning_rate": 0.0001997051704552458, + "loss": 1.0763, "step": 1525 }, { "epoch": 0.12, - "learning_rate": 0.00019969151396222087, - "loss": 0.9646, + "learning_rate": 0.00019969435743090076, + "loss": 0.9931, "step": 1530 }, { "epoch": 0.12, - "learning_rate": 0.00019968044404364516, - "loss": 1.0415, + "learning_rate": 0.00019968334997818064, + "loss": 1.0857, "step": 1535 }, { "epoch": 0.12, - "learning_rate": 0.00019966917930836782, - "loss": 0.9598, + "learning_rate": 0.00019967214811855263, + "loss": 1.0309, "step": 1540 }, { "epoch": 0.12, - "learning_rate": 0.00019965771977840474, - "loss": 1.0443, + "learning_rate": 0.0001996607518738631, + "loss": 1.0208, "step": 1545 }, { "epoch": 0.12, - "learning_rate": 0.00019964606547615262, - "loss": 1.0248, + "learning_rate": 0.00019964916126633754, + "loss": 1.015, "step": 1550 }, { "epoch": 0.12, - "learning_rate": 0.00019963421642438877, - "loss": 1.0391, + "learning_rate": 0.00019963737631858045, + "loss": 0.9861, "step": 1555 }, { "epoch": 0.12, - "learning_rate": 0.00019962217264627108, - "loss": 1.0348, + "learning_rate": 0.0001996253970535754, + "loss": 1.055, "step": 1560 }, { "epoch": 0.13, - "learning_rate": 0.00019960993416533808, - "loss": 0.978, + "learning_rate": 0.00019961322349468482, + "loss": 0.9746, "step": 1565 }, { "epoch": 0.13, - "learning_rate": 0.00019959750100550884, - "loss": 1.0232, + "learning_rate": 0.00019960085566565018, + "loss": 0.9777, "step": 1570 }, { "epoch": 0.13, - "learning_rate": 0.00019958487319108288, - "loss": 1.0161, + "learning_rate": 0.00019958829359059175, + "loss": 1.0462, "step": 1575 }, { "epoch": 0.13, - "learning_rate": 0.0001995720507467401, - "loss": 1.0837, + "learning_rate": 0.00019957553729400864, + "loss": 0.973, "step": 1580 }, { "epoch": 0.13, - "learning_rate": 0.0001995590336975409, - "loss": 0.9807, + "learning_rate": 0.00019956258680077876, + "loss": 1.0164, "step": 1585 }, { "epoch": 0.13, - "learning_rate": 0.00019954582206892595, - "loss": 1.0426, + "learning_rate": 0.00019954944213615875, + "loss": 0.9951, "step": 1590 }, { "epoch": 0.13, - "learning_rate": 0.00019953241588671622, - "loss": 1.0027, + "learning_rate": 0.00019953610332578388, + "loss": 1.0109, "step": 1595 }, { "epoch": 0.13, - "learning_rate": 0.00019951881517711291, - "loss": 1.034, + "learning_rate": 0.0001995225703956681, + "loss": 1.0433, "step": 1600 }, { "epoch": 0.13, - "learning_rate": 0.00019950501996669745, - "loss": 0.9549, + "learning_rate": 0.000199508843372204, + "loss": 1.0563, "step": 1605 }, { "epoch": 0.13, - "learning_rate": 0.0001994910302824313, - "loss": 0.9733, + "learning_rate": 0.00019949492228216254, + "loss": 0.9551, "step": 1610 }, { "epoch": 0.13, - "learning_rate": 0.00019947684615165617, - "loss": 0.9949, + "learning_rate": 0.00019948080715269332, + "loss": 0.9498, "step": 1615 }, { "epoch": 0.13, - "learning_rate": 0.00019946246760209362, - "loss": 1.0355, + "learning_rate": 0.00019946649801132427, + "loss": 1.0079, "step": 1620 }, { "epoch": 0.13, - "learning_rate": 0.00019944789466184532, - "loss": 0.9879, + "learning_rate": 0.0001994519948859617, + "loss": 1.109, "step": 1625 }, { "epoch": 0.13, - "learning_rate": 0.00019943312735939275, - "loss": 1.02, + "learning_rate": 0.00019943729780489027, + "loss": 1.0486, "step": 1630 }, { "epoch": 0.13, - "learning_rate": 0.00019941816572359734, - "loss": 1.0153, + "learning_rate": 0.00019942240679677285, + "loss": 0.9946, "step": 1635 }, { "epoch": 0.13, - "learning_rate": 0.00019940300978370031, - "loss": 0.9759, + "learning_rate": 0.00019940732189065058, + "loss": 1.0231, "step": 1640 }, { "epoch": 0.13, - "learning_rate": 0.00019938765956932261, - "loss": 1.027, + "learning_rate": 0.0001993920431159427, + "loss": 0.9942, "step": 1645 }, { "epoch": 0.13, - "learning_rate": 0.00019937211511046493, - "loss": 1.02, + "learning_rate": 0.0001993765705024466, + "loss": 1.0216, "step": 1650 }, { "epoch": 0.13, - "learning_rate": 0.00019935637643750745, - "loss": 1.0289, + "learning_rate": 0.0001993609040803376, + "loss": 1.0245, "step": 1655 }, { "epoch": 0.13, - "learning_rate": 0.00019934044358121012, - "loss": 1.0237, + "learning_rate": 0.00019934504388016908, + "loss": 0.9663, "step": 1660 }, { "epoch": 0.13, - "learning_rate": 0.00019932431657271226, - "loss": 1.0549, + "learning_rate": 0.0001993289899328723, + "loss": 1.0251, "step": 1665 }, { "epoch": 0.13, - "learning_rate": 0.00019930799544353275, - "loss": 1.0109, + "learning_rate": 0.0001993127422697564, + "loss": 1.0622, "step": 1670 }, { "epoch": 0.13, - "learning_rate": 0.00019929148022556977, - "loss": 0.9815, + "learning_rate": 0.00019929630092250825, + "loss": 0.9635, "step": 1675 }, { "epoch": 0.13, - "learning_rate": 0.00019927477095110088, - "loss": 1.0243, + "learning_rate": 0.00019927966592319258, + "loss": 1.0168, "step": 1680 }, { "epoch": 0.13, - "learning_rate": 0.00019925786765278287, - "loss": 0.997, + "learning_rate": 0.00019926283730425164, + "loss": 1.0232, "step": 1685 }, { "epoch": 0.14, - "learning_rate": 0.00019924077036365177, - "loss": 1.012, + "learning_rate": 0.00019924581509850534, + "loss": 1.0397, "step": 1690 }, { "epoch": 0.14, - "learning_rate": 0.00019922347911712277, - "loss": 1.0397, + "learning_rate": 0.00019922859933915122, + "loss": 1.0066, "step": 1695 }, { "epoch": 0.14, - "learning_rate": 0.00019920599394699005, - "loss": 1.0169, + "learning_rate": 0.00019921119005976415, + "loss": 1.0135, "step": 1700 }, { "epoch": 0.14, - "learning_rate": 0.0001991883148874269, - "loss": 1.0549, + "learning_rate": 0.00019919358729429654, + "loss": 1.0455, "step": 1705 }, { "epoch": 0.14, - "learning_rate": 0.0001991704419729854, - "loss": 0.9806, + "learning_rate": 0.000199175791077078, + "loss": 1.0542, "step": 1710 }, { "epoch": 0.14, - "learning_rate": 0.0001991523752385967, - "loss": 1.0343, + "learning_rate": 0.00019915780144281559, + "loss": 1.0315, "step": 1715 }, { "epoch": 0.14, - "learning_rate": 0.00019913411471957058, - "loss": 1.0399, + "learning_rate": 0.00019913961842659344, + "loss": 1.0853, "step": 1720 }, { "epoch": 0.14, - "learning_rate": 0.00019911566045159566, - "loss": 1.0166, + "learning_rate": 0.00019912124206387295, + "loss": 0.9702, "step": 1725 }, { "epoch": 0.14, - "learning_rate": 0.0001990970124707392, - "loss": 1.0175, + "learning_rate": 0.00019910267239049245, + "loss": 1.0132, "step": 1730 }, { "epoch": 0.14, - "learning_rate": 0.00019907817081344699, - "loss": 1.0269, + "learning_rate": 0.00019908390944266735, + "loss": 0.9818, "step": 1735 }, { "epoch": 0.14, - "learning_rate": 0.00019905913551654348, - "loss": 1.0487, + "learning_rate": 0.00019906495325699002, + "loss": 0.9916, "step": 1740 }, { "epoch": 0.14, - "learning_rate": 0.0001990399066172314, - "loss": 1.0073, + "learning_rate": 0.00019904580387042966, + "loss": 0.9861, "step": 1745 }, { "epoch": 0.14, - "learning_rate": 0.00019902048415309205, - "loss": 1.0381, + "learning_rate": 0.0001990264613203322, + "loss": 0.9422, "step": 1750 }, { "epoch": 0.14, - "learning_rate": 0.00019900086816208485, - "loss": 1.028, + "learning_rate": 0.0001990069256444204, + "loss": 1.0313, "step": 1755 }, { "epoch": 0.14, - "learning_rate": 0.0001989810586825476, - "loss": 1.0312, + "learning_rate": 0.00019898719688079357, + "loss": 0.9469, "step": 1760 }, { "epoch": 0.14, - "learning_rate": 0.00019896105575319614, - "loss": 1.0253, + "learning_rate": 0.00019896727506792767, + "loss": 1.029, "step": 1765 }, { "epoch": 0.14, - "learning_rate": 0.00019894085941312448, - "loss": 1.0397, + "learning_rate": 0.00019894716024467502, + "loss": 0.9629, "step": 1770 }, { "epoch": 0.14, - "learning_rate": 0.0001989204697018046, - "loss": 1.0307, + "learning_rate": 0.00019892685245026458, + "loss": 1.0622, "step": 1775 }, { "epoch": 0.14, - "learning_rate": 0.0001988998866590864, - "loss": 1.0626, + "learning_rate": 0.00019890635172430137, + "loss": 1.0073, "step": 1780 }, { "epoch": 0.14, - "learning_rate": 0.0001988791103251976, - "loss": 0.9891, + "learning_rate": 0.00019888565810676688, + "loss": 0.9551, "step": 1785 }, { "epoch": 0.14, - "learning_rate": 0.0001988581407407438, - "loss": 1.0396, + "learning_rate": 0.00019886477163801877, + "loss": 1.0587, "step": 1790 }, { "epoch": 0.14, - "learning_rate": 0.00019883697794670813, - "loss": 1.0318, + "learning_rate": 0.0001988436923587907, + "loss": 1.0609, "step": 1795 }, { "epoch": 0.14, - "learning_rate": 0.00019881562198445146, - "loss": 1.0302, + "learning_rate": 0.00019882242031019241, + "loss": 1.0436, "step": 1800 }, { "epoch": 0.14, - "learning_rate": 0.0001987940728957122, - "loss": 0.9536, + "learning_rate": 0.00019880095553370967, + "loss": 1.0192, "step": 1805 }, { "epoch": 0.14, - "learning_rate": 0.00019877233072260607, - "loss": 1.0017, + "learning_rate": 0.00019877929807120398, + "loss": 1.037, "step": 1810 }, { "epoch": 0.15, - "learning_rate": 0.00019875039550762628, - "loss": 0.9765, + "learning_rate": 0.0001987574479649127, + "loss": 0.9445, "step": 1815 }, { "epoch": 0.15, - "learning_rate": 0.0001987282672936433, - "loss": 0.9631, + "learning_rate": 0.00019873540525744887, + "loss": 0.9386, "step": 1820 }, { "epoch": 0.15, - "learning_rate": 0.00019870594612390478, - "loss": 1.0082, + "learning_rate": 0.00019871316999180117, + "loss": 0.9727, "step": 1825 }, { "epoch": 0.15, - "learning_rate": 0.00019868343204203548, - "loss": 1.0115, + "learning_rate": 0.00019869074221133383, + "loss": 1.0091, "step": 1830 }, { "epoch": 0.15, - "learning_rate": 0.00019866072509203725, - "loss": 1.0289, + "learning_rate": 0.00019866812195978647, + "loss": 1.0104, "step": 1835 }, { "epoch": 0.15, - "learning_rate": 0.00019863782531828874, - "loss": 1.0474, + "learning_rate": 0.00019864530928127415, + "loss": 1.0083, "step": 1840 }, { "epoch": 0.15, - "learning_rate": 0.00019861473276554562, - "loss": 1.02, + "learning_rate": 0.00019862230422028714, + "loss": 1.0241, "step": 1845 }, { "epoch": 0.15, - "learning_rate": 0.00019859144747894026, - "loss": 1.0314, + "learning_rate": 0.00019859910682169094, + "loss": 1.0057, "step": 1850 }, { "epoch": 0.15, - "learning_rate": 0.0001985679695039817, - "loss": 1.0028, + "learning_rate": 0.00019857571713072616, + "loss": 7.4468, "step": 1855 }, { "epoch": 0.15, - "learning_rate": 0.0001985442988865556, - "loss": 1.0052, + "learning_rate": 0.0001985521351930084, + "loss": 8.8254, "step": 1860 }, { "epoch": 0.15, - "learning_rate": 0.00019852043567292413, - "loss": 0.9954, + "learning_rate": 0.00019852836105452822, + "loss": 7.8081, "step": 1865 }, { "epoch": 0.15, - "learning_rate": 0.0001984963799097258, - "loss": 0.9907, + "learning_rate": 0.000198504394761651, + "loss": 7.6345, "step": 1870 }, { "epoch": 0.15, - "learning_rate": 0.00019847213164397553, - "loss": 1.017, + "learning_rate": 0.00019848023636111686, + "loss": 7.5297, "step": 1875 }, { "epoch": 0.15, - "learning_rate": 0.0001984476909230644, - "loss": 0.9711, + "learning_rate": 0.00019845588590004058, + "loss": 7.4831, "step": 1880 }, { "epoch": 0.15, - "learning_rate": 0.00019842305779475968, - "loss": 0.9999, + "learning_rate": 0.00019843134342591157, + "loss": 7.4163, "step": 1885 }, { "epoch": 0.15, - "learning_rate": 0.00019839823230720466, - "loss": 0.997, + "learning_rate": 0.0001984066089865936, + "loss": 7.4007, "step": 1890 }, { "epoch": 0.15, - "learning_rate": 0.00019837321450891864, - "loss": 1.0454, + "learning_rate": 0.00019838168263032488, + "loss": 7.3359, "step": 1895 }, { "epoch": 0.15, - "learning_rate": 0.0001983480044487966, - "loss": 0.9981, + "learning_rate": 0.00019835656440571792, + "loss": 7.3149, "step": 1900 }, { "epoch": 0.15, - "learning_rate": 0.0001983226021761095, - "loss": 1.0104, + "learning_rate": 0.00019833125436175938, + "loss": 7.3384, "step": 1905 }, { "epoch": 0.15, - "learning_rate": 0.00019829700774050377, - "loss": 1.011, + "learning_rate": 0.00019830575254781008, + "loss": 7.2882, "step": 1910 }, { "epoch": 0.15, - "learning_rate": 0.00019827122119200156, - "loss": 1.1083, + "learning_rate": 0.00019828005901360475, + "loss": 7.3616, "step": 1915 }, { "epoch": 0.15, - "learning_rate": 0.00019824524258100042, - "loss": 0.9852, + "learning_rate": 0.0001982541738092521, + "loss": 7.2891, "step": 1920 }, { "epoch": 0.15, - "learning_rate": 0.00019821907195827324, - "loss": 0.969, + "learning_rate": 0.00019822809698523451, + "loss": 7.3301, "step": 1925 }, { "epoch": 0.15, - "learning_rate": 0.00019819270937496822, - "loss": 0.9754, + "learning_rate": 0.0001982018285924083, + "loss": 7.2296, "step": 1930 }, { "epoch": 0.15, - "learning_rate": 0.00019816615488260872, - "loss": 1.0803, + "learning_rate": 0.00019817536868200317, + "loss": 7.2871, "step": 1935 }, { "epoch": 0.16, - "learning_rate": 0.00019813940853309316, - "loss": 0.9769, + "learning_rate": 0.00019814871730562247, + "loss": 7.2735, "step": 1940 }, { "epoch": 0.16, - "learning_rate": 0.00019811247037869497, - "loss": 0.9867, + "learning_rate": 0.00019812187451524282, + "loss": 7.2587, "step": 1945 }, { "epoch": 0.16, - "learning_rate": 0.0001980853404720624, - "loss": 1.0137, + "learning_rate": 0.0001980948403632143, + "loss": 7.2716, "step": 1950 }, { "epoch": 0.16, - "learning_rate": 0.0001980580188662185, - "loss": 0.9856, + "learning_rate": 0.0001980676149022601, + "loss": 7.2557, "step": 1955 }, { "epoch": 0.16, - "learning_rate": 0.0001980305056145609, - "loss": 1.0238, + "learning_rate": 0.00019804019818547651, + "loss": 7.2483, "step": 1960 }, { "epoch": 0.16, - "learning_rate": 0.00019800280077086188, - "loss": 1.0493, + "learning_rate": 0.00019801259026633284, + "loss": 7.2976, "step": 1965 }, { "epoch": 0.16, - "learning_rate": 0.00019797490438926813, - "loss": 1.035, + "learning_rate": 0.00019798479119867134, + "loss": 7.212, "step": 1970 }, { "epoch": 0.16, - "learning_rate": 0.0001979468165243007, - "loss": 0.9897, + "learning_rate": 0.00019795680103670685, + "loss": 7.2595, "step": 1975 }, { "epoch": 0.16, - "learning_rate": 0.00019791853723085483, - "loss": 1.0004, + "learning_rate": 0.00019792861983502718, + "loss": 7.2458, "step": 1980 }, { "epoch": 0.16, - "learning_rate": 0.00019789006656419992, - "loss": 1.003, + "learning_rate": 0.00019790024764859248, + "loss": 7.2721, "step": 1985 }, { "epoch": 0.16, - "learning_rate": 0.00019786140457997943, - "loss": 1.0119, + "learning_rate": 0.00019787168453273544, + "loss": 7.2745, "step": 1990 }, { "epoch": 0.16, - "learning_rate": 0.00019783255133421068, - "loss": 0.9857, + "learning_rate": 0.00019784293054316118, + "loss": 7.2349, "step": 1995 }, { "epoch": 0.16, - "learning_rate": 0.0001978035068832848, - "loss": 0.9844, + "learning_rate": 0.000197813985735947, + "loss": 7.2831, "step": 2000 }, { "epoch": 0.16, - "learning_rate": 0.00019777427128396668, - "loss": 0.9992, + "learning_rate": 0.00019778485016754232, + "loss": 7.2836, "step": 2005 }, { "epoch": 0.16, - "learning_rate": 0.0001977448445933947, - "loss": 1.0199, + "learning_rate": 0.00019775552389476864, + "loss": 7.2821, "step": 2010 }, { "epoch": 0.16, - "learning_rate": 0.00019771522686908073, - "loss": 0.9824, + "learning_rate": 0.00019772600697481933, + "loss": 7.2584, "step": 2015 }, { "epoch": 0.16, - "learning_rate": 0.00019768541816891005, - "loss": 1.023, + "learning_rate": 0.00019769629946525966, + "loss": 7.2901, "step": 2020 }, { "epoch": 0.16, - "learning_rate": 0.00019765541855114112, - "loss": 1.0382, + "learning_rate": 0.0001976664014240265, + "loss": 7.2448, "step": 2025 }, { "epoch": 0.16, - "learning_rate": 0.00019762522807440565, - "loss": 1.0142, + "learning_rate": 0.00019763631290942838, + "loss": 7.2413, "step": 2030 }, { "epoch": 0.16, - "learning_rate": 0.0001975948467977082, - "loss": 1.0201, + "learning_rate": 0.0001976060339801452, + "loss": 7.2749, "step": 2035 }, { "epoch": 0.16, - "learning_rate": 0.00019756427478042634, - "loss": 0.9924, + "learning_rate": 0.0001975755646952283, + "loss": 7.3306, "step": 2040 }, { "epoch": 0.16, - "learning_rate": 0.0001975335120823104, - "loss": 1.0479, + "learning_rate": 0.00019754490511410028, + "loss": 7.2392, "step": 2045 }, { "epoch": 0.16, - "learning_rate": 0.0001975025587634834, - "loss": 1.026, + "learning_rate": 0.00019751405529655475, + "loss": 7.277, "step": 2050 }, { "epoch": 0.16, - "learning_rate": 0.00019747141488444086, - "loss": 0.9987, + "learning_rate": 0.00019748301530275646, + "loss": 7.2549, "step": 2055 }, { "epoch": 0.16, - "learning_rate": 0.00019744008050605078, - "loss": 1.04, + "learning_rate": 0.00019745178519324096, + "loss": 7.2652, "step": 2060 }, { "epoch": 0.17, - "learning_rate": 0.0001974085556895535, - "loss": 1.0257, + "learning_rate": 0.00019742036502891458, + "loss": 7.2313, "step": 2065 }, { "epoch": 0.17, - "learning_rate": 0.00019737684049656143, - "loss": 0.9818, + "learning_rate": 0.00019738875487105434, + "loss": 7.2309, "step": 2070 }, { "epoch": 0.17, - "learning_rate": 0.00019734493498905923, - "loss": 1.0136, + "learning_rate": 0.0001973569547813078, + "loss": 7.268, "step": 2075 }, { "epoch": 0.17, - "learning_rate": 0.00019731283922940333, - "loss": 1.0066, + "learning_rate": 0.0001973249648216929, + "loss": 7.2083, "step": 2080 }, { "epoch": 0.17, - "learning_rate": 0.00019728055328032215, - "loss": 1.0534, + "learning_rate": 0.00019729278505459786, + "loss": 7.2145, "step": 2085 }, { "epoch": 0.17, - "learning_rate": 0.00019724807720491574, - "loss": 1.0247, + "learning_rate": 0.0001972604155427811, + "loss": 7.2356, "step": 2090 }, { "epoch": 0.17, - "learning_rate": 0.00019721541106665572, - "loss": 1.0691, + "learning_rate": 0.00019722785634937112, + "loss": 7.2588, "step": 2095 }, { "epoch": 0.17, - "learning_rate": 0.00019718255492938522, - "loss": 1.0096, + "learning_rate": 0.00019719510753786628, + "loss": 7.2303, "step": 2100 }, { "epoch": 0.17, - "learning_rate": 0.00019714950885731867, - "loss": 1.0042, + "learning_rate": 0.00019716216917213476, + "loss": 7.2577, "step": 2105 }, { "epoch": 0.17, - "learning_rate": 0.00019711627291504175, - "loss": 0.9998, + "learning_rate": 0.00019712904131641444, + "loss": 7.2937, "step": 2110 }, { "epoch": 0.17, - "learning_rate": 0.00019708284716751114, - "loss": 1.0467, + "learning_rate": 0.00019709572403531277, + "loss": 7.2384, "step": 2115 }, { "epoch": 0.17, - "learning_rate": 0.00019704923168005462, - "loss": 0.9959, + "learning_rate": 0.00019706221739380656, + "loss": 7.3166, "step": 2120 }, { "epoch": 0.17, - "learning_rate": 0.00019701542651837065, - "loss": 1.0143, + "learning_rate": 0.00019702852145724197, + "loss": 7.2265, "step": 2125 }, { "epoch": 0.17, - "learning_rate": 0.00019698143174852852, - "loss": 1.0132, + "learning_rate": 0.00019699463629133424, + "loss": 7.2526, "step": 2130 }, { "epoch": 0.17, - "learning_rate": 0.00019694724743696797, - "loss": 0.9966, + "learning_rate": 0.00019696056196216787, + "loss": 7.2019, "step": 2135 }, { "epoch": 0.17, - "learning_rate": 0.0001969128736504993, - "loss": 1.0254, + "learning_rate": 0.00019692629853619603, + "loss": 7.2335, "step": 2140 }, { "epoch": 0.17, - "learning_rate": 0.000196878310456303, - "loss": 1.0267, + "learning_rate": 0.0001968918460802408, + "loss": 7.2781, "step": 2145 }, { "epoch": 0.17, - "learning_rate": 0.0001968435579219299, - "loss": 0.957, + "learning_rate": 0.0001968572046614929, + "loss": 7.2338, "step": 2150 }, { "epoch": 0.17, - "learning_rate": 0.00019680861611530065, - "loss": 1.0538, + "learning_rate": 0.00019682237434751157, + "loss": 7.2244, "step": 2155 }, { "epoch": 0.17, - "learning_rate": 0.0001967734851047061, - "loss": 0.9972, + "learning_rate": 0.00019678735520622443, + "loss": 7.2363, "step": 2160 }, { "epoch": 0.17, - "learning_rate": 0.00019673816495880666, - "loss": 1.0405, + "learning_rate": 0.00019675214730592743, + "loss": 7.28, "step": 2165 }, { "epoch": 0.17, - "learning_rate": 0.00019670265574663244, - "loss": 1.0784, + "learning_rate": 0.0001967167507152845, + "loss": 7.2629, "step": 2170 }, { "epoch": 0.17, - "learning_rate": 0.00019666695753758315, - "loss": 1.0366, + "learning_rate": 0.00019668116550332766, + "loss": 7.2364, "step": 2175 }, { "epoch": 0.17, - "learning_rate": 0.00019663107040142775, - "loss": 1.0377, + "learning_rate": 0.00019664539173945683, + "loss": 7.2588, "step": 2180 }, { - "epoch": 0.18, - "learning_rate": 0.00019659499440830458, - "loss": 1.0154, + "epoch": 0.17, + "learning_rate": 0.00019660942949343953, + "loss": 7.214, "step": 2185 }, { "epoch": 0.18, - "learning_rate": 0.00019655872962872094, - "loss": 1.0345, + "learning_rate": 0.000196573278835411, + "loss": 7.2691, "step": 2190 }, { "epoch": 0.18, - "learning_rate": 0.0001965222761335532, - "loss": 1.0179, + "learning_rate": 0.0001965369398358738, + "loss": 7.239, "step": 2195 }, { "epoch": 0.18, - "learning_rate": 0.00019648563399404654, - "loss": 0.9978, + "learning_rate": 0.00019650041256569792, + "loss": 7.226, "step": 2200 }, { "epoch": 0.18, - "learning_rate": 0.0001964488032818148, - "loss": 0.9969, + "learning_rate": 0.0001964636970961204, + "loss": 7.2564, "step": 2205 }, { "epoch": 0.18, - "learning_rate": 0.00019641178406884033, - "loss": 0.997, + "learning_rate": 0.00019642679349874548, + "loss": 7.2092, "step": 2210 }, { "epoch": 0.18, - "learning_rate": 0.00019637457642747403, - "loss": 1.0187, + "learning_rate": 0.00019638970184554408, + "loss": 7.2154, "step": 2215 }, { "epoch": 0.18, - "learning_rate": 0.00019633718043043493, - "loss": 1.0517, + "learning_rate": 0.0001963524222088541, + "loss": 7.2212, "step": 2220 }, { "epoch": 0.18, - "learning_rate": 0.00019629959615081023, - "loss": 1.0068, + "learning_rate": 0.0001963149546613799, + "loss": 7.2857, "step": 2225 }, { "epoch": 0.18, - "learning_rate": 0.00019626182366205516, - "loss": 0.9404, + "learning_rate": 0.00019627729927619236, + "loss": 7.2489, "step": 2230 }, { "epoch": 0.18, - "learning_rate": 0.00019622386303799267, - "loss": 0.9907, + "learning_rate": 0.00019623945612672872, + "loss": 7.246, "step": 2235 }, { "epoch": 0.18, - "learning_rate": 0.0001961857143528135, - "loss": 1.0724, + "learning_rate": 0.00019620142528679231, + "loss": 7.2828, "step": 2240 }, { "epoch": 0.18, - "learning_rate": 0.00019614737768107595, - "loss": 1.057, + "learning_rate": 0.00019616320683055262, + "loss": 7.2636, "step": 2245 }, { "epoch": 0.18, - "learning_rate": 0.0001961088530977057, - "loss": 0.9985, + "learning_rate": 0.000196124800832545, + "loss": 7.2525, "step": 2250 }, { "epoch": 0.18, - "learning_rate": 0.00019607014067799558, - "loss": 1.0189, + "learning_rate": 0.00019608620736767047, + "loss": 7.2258, "step": 2255 }, { "epoch": 0.18, - "learning_rate": 0.00019603124049760574, - "loss": 1.0346, + "learning_rate": 0.00019604742651119578, + "loss": 7.1844, "step": 2260 }, { "epoch": 0.18, - "learning_rate": 0.00019599215263256309, - "loss": 1.0776, + "learning_rate": 0.00019600845833875306, + "loss": 7.2716, "step": 2265 }, { "epoch": 0.18, - "learning_rate": 0.00019595287715926148, - "loss": 1.0657, + "learning_rate": 0.0001959693029263398, + "loss": 7.2647, "step": 2270 }, { "epoch": 0.18, - "learning_rate": 0.0001959134141544614, - "loss": 0.986, + "learning_rate": 0.0001959299603503186, + "loss": 7.245, "step": 2275 }, { "epoch": 0.18, - "learning_rate": 0.00019587376369528981, - "loss": 0.9688, + "learning_rate": 0.0001958904306874171, + "loss": 7.2232, "step": 2280 }, { "epoch": 0.18, - "learning_rate": 0.00019583392585924007, - "loss": 1.0281, + "learning_rate": 0.00019585071401472784, + "loss": 7.2781, "step": 2285 }, { "epoch": 0.18, - "learning_rate": 0.00019579390072417178, - "loss": 1.0563, + "learning_rate": 0.00019581081040970805, + "loss": 7.2159, "step": 2290 }, { "epoch": 0.18, - "learning_rate": 0.00019575368836831047, - "loss": 0.9866, + "learning_rate": 0.00019577071995017945, + "loss": 7.2941, "step": 2295 }, { "epoch": 0.18, - "learning_rate": 0.0001957132888702477, - "loss": 0.9501, + "learning_rate": 0.0001957304427143283, + "loss": 7.2225, "step": 2300 }, { "epoch": 0.18, - "learning_rate": 0.00019567270230894085, - "loss": 1.0684, + "learning_rate": 0.00019568997878070503, + "loss": 7.2558, "step": 2305 }, { - "epoch": 0.19, - "learning_rate": 0.0001956319287637127, - "loss": 1.049, + "epoch": 0.18, + "learning_rate": 0.00019564932822822425, + "loss": 7.2124, "step": 2310 }, { "epoch": 0.19, - "learning_rate": 0.0001955909683142516, - "loss": 0.9442, + "learning_rate": 0.00019560849113616443, + "loss": 7.2109, "step": 2315 }, { "epoch": 0.19, - "learning_rate": 0.00019554982104061117, - "loss": 0.9767, + "learning_rate": 0.0001955674675841679, + "loss": 7.2294, "step": 2320 }, { "epoch": 0.19, - "learning_rate": 0.00019550848702321014, - "loss": 1.0159, + "learning_rate": 0.00019552625765224056, + "loss": 7.1873, "step": 2325 }, { "epoch": 0.19, - "learning_rate": 0.00019546696634283227, - "loss": 1.0234, + "learning_rate": 0.00019548486142075198, + "loss": 7.2327, "step": 2330 }, { "epoch": 0.19, - "learning_rate": 0.00019542525908062603, - "loss": 1.027, + "learning_rate": 0.00019544327897043484, + "loss": 7.2003, "step": 2335 }, { "epoch": 0.19, - "learning_rate": 0.00019538336531810463, - "loss": 0.9804, + "learning_rate": 0.00019540151038238514, + "loss": 7.1883, "step": 2340 }, { "epoch": 0.19, - "learning_rate": 0.0001953412851371458, - "loss": 0.9403, + "learning_rate": 0.00019535955573806178, + "loss": 7.1803, "step": 2345 }, { "epoch": 0.19, - "learning_rate": 0.00019529901861999158, - "loss": 0.9663, + "learning_rate": 0.00019531741511928663, + "loss": 7.2142, "step": 2350 }, { "epoch": 0.19, - "learning_rate": 0.0001952565658492481, - "loss": 1.0759, + "learning_rate": 0.00019527508860824419, + "loss": 7.2528, "step": 2355 }, { "epoch": 0.19, - "learning_rate": 0.00019521392690788564, - "loss": 1.0261, + "learning_rate": 0.00019523257628748146, + "loss": 7.212, "step": 2360 }, { "epoch": 0.19, - "learning_rate": 0.0001951711018792383, - "loss": 0.9482, + "learning_rate": 0.00019518987823990793, + "loss": 7.1569, "step": 2365 }, { "epoch": 0.19, - "learning_rate": 0.00019512809084700375, - "loss": 1.028, + "learning_rate": 0.00019514699454879517, + "loss": 7.0854, "step": 2370 }, { "epoch": 0.19, - "learning_rate": 0.00019508489389524343, - "loss": 0.9989, + "learning_rate": 0.00019510392529777692, + "loss": 7.1283, "step": 2375 }, { "epoch": 0.19, - "learning_rate": 0.00019504151110838187, - "loss": 1.0761, + "learning_rate": 0.00019506067057084868, + "loss": 7.1222, "step": 2380 }, { "epoch": 0.19, - "learning_rate": 0.00019499794257120702, - "loss": 0.9992, + "learning_rate": 0.00019501723045236785, + "loss": 7.0935, "step": 2385 }, { "epoch": 0.19, - "learning_rate": 0.0001949541883688697, - "loss": 0.9806, + "learning_rate": 0.0001949736050270532, + "loss": 7.0704, "step": 2390 }, { "epoch": 0.19, - "learning_rate": 0.0001949102485868837, - "loss": 1.0263, + "learning_rate": 0.00019492979437998499, + "loss": 7.0341, "step": 2395 }, { "epoch": 0.19, - "learning_rate": 0.00019486612331112553, - "loss": 1.0391, + "learning_rate": 0.00019488579859660473, + "loss": 7.0147, "step": 2400 }, { "epoch": 0.19, - "learning_rate": 0.0001948218126278341, - "loss": 1.0516, + "learning_rate": 0.0001948416177627149, + "loss": 7.008, "step": 2405 }, { "epoch": 0.19, - "learning_rate": 0.00019477731662361078, - "loss": 0.9653, + "learning_rate": 0.00019479725196447898, + "loss": 6.9917, "step": 2410 }, { "epoch": 0.19, - "learning_rate": 0.00019473263538541914, - "loss": 1.0275, + "learning_rate": 0.0001947527012884211, + "loss": 6.9149, "step": 2415 }, { "epoch": 0.19, - "learning_rate": 0.00019468776900058473, - "loss": 0.9902, + "learning_rate": 0.00019470796582142597, + "loss": 6.8777, "step": 2420 }, { "epoch": 0.19, - "learning_rate": 0.00019464271755679499, - "loss": 0.9579, + "learning_rate": 0.00019466304565073873, + "loss": 6.8518, "step": 2425 }, { "epoch": 0.19, - "learning_rate": 0.000194597481142099, - "loss": 0.9679, + "learning_rate": 0.00019461794086396463, + "loss": 6.8689, "step": 2430 }, { - "epoch": 0.2, - "learning_rate": 0.0001945520598449074, - "loss": 0.9681, + "epoch": 0.19, + "learning_rate": 0.00019457265154906907, + "loss": 6.8504, "step": 2435 }, { "epoch": 0.2, - "learning_rate": 0.00019450645375399211, - "loss": 1.0025, + "learning_rate": 0.00019452717779437731, + "loss": 6.7835, "step": 2440 }, { "epoch": 0.2, - "learning_rate": 0.00019446066295848624, - "loss": 0.9259, + "learning_rate": 0.00019448151968857428, + "loss": 6.8541, "step": 2445 }, { "epoch": 0.2, - "learning_rate": 0.00019441468754788387, - "loss": 0.9237, + "learning_rate": 0.00019443567732070437, + "loss": 6.7886, "step": 2450 }, { "epoch": 0.2, - "learning_rate": 0.00019436852761204004, - "loss": 0.9851, + "learning_rate": 0.00019438965078017156, + "loss": 6.7578, "step": 2455 }, { "epoch": 0.2, - "learning_rate": 0.00019432218324117012, - "loss": 1.0229, + "learning_rate": 0.00019434344015673873, + "loss": 6.7578, "step": 2460 }, { "epoch": 0.2, - "learning_rate": 0.00019427565452585026, - "loss": 0.9736, + "learning_rate": 0.00019429704554052793, + "loss": 6.6911, "step": 2465 }, { "epoch": 0.2, - "learning_rate": 0.00019422894155701675, - "loss": 0.9864, + "learning_rate": 0.00019425046702202002, + "loss": 6.6988, "step": 2470 }, { "epoch": 0.2, - "learning_rate": 0.00019418204442596594, - "loss": 1.0071, + "learning_rate": 0.00019420370469205454, + "loss": 6.596, "step": 2475 }, { "epoch": 0.2, - "learning_rate": 0.0001941349632243542, - "loss": 0.9814, + "learning_rate": 0.0001941567586418294, + "loss": 6.6326, "step": 2480 }, { "epoch": 0.2, - "learning_rate": 0.0001940876980441976, - "loss": 1.022, + "learning_rate": 0.0001941096289629009, + "loss": 6.6622, "step": 2485 }, { "epoch": 0.2, - "learning_rate": 0.00019404024897787177, - "loss": 0.97, + "learning_rate": 0.00019406231574718344, + "loss": 6.5711, "step": 2490 }, { "epoch": 0.2, - "learning_rate": 0.00019399261611811186, - "loss": 0.9936, + "learning_rate": 0.0001940148190869494, + "loss": 6.5892, "step": 2495 }, { "epoch": 0.2, - "learning_rate": 0.00019394479955801207, - "loss": 0.9869, + "learning_rate": 0.00019396713907482877, + "loss": 6.573, "step": 2500 }, { "epoch": 0.2, - "learning_rate": 0.00019389679939102564, - "loss": 1.0161, + "learning_rate": 0.00019391927580380934, + "loss": 6.5118, "step": 2505 }, { "epoch": 0.2, - "learning_rate": 0.0001938486157109647, - "loss": 0.9928, + "learning_rate": 0.0001938712293672362, + "loss": 6.5871, "step": 2510 }, { "epoch": 0.2, - "learning_rate": 0.00019380024861200007, - "loss": 0.9997, + "learning_rate": 0.00019382299985881157, + "loss": 6.5794, "step": 2515 }, { "epoch": 0.2, - "learning_rate": 0.00019375169818866103, - "loss": 1.0364, + "learning_rate": 0.0001937745873725948, + "loss": 6.469, "step": 2520 }, { "epoch": 0.2, - "learning_rate": 0.0001937029645358351, - "loss": 1.0031, + "learning_rate": 0.00019372599200300216, + "loss": 6.5333, "step": 2525 }, { "epoch": 0.2, - "learning_rate": 0.0001936540477487679, - "loss": 1.0309, + "learning_rate": 0.00019367721384480643, + "loss": 6.4678, "step": 2530 }, { "epoch": 0.2, - "learning_rate": 0.0001936049479230631, - "loss": 0.9506, + "learning_rate": 0.00019362825299313693, + "loss": 6.4267, "step": 2535 }, { "epoch": 0.2, - "learning_rate": 0.00019355566515468194, - "loss": 0.9846, + "learning_rate": 0.00019357910954347936, + "loss": 6.3952, "step": 2540 }, { "epoch": 0.2, - "learning_rate": 0.0001935061995399433, - "loss": 0.9758, + "learning_rate": 0.0001935297835916754, + "loss": 6.3703, "step": 2545 }, { "epoch": 0.2, - "learning_rate": 0.00019345655117552338, - "loss": 0.9888, + "learning_rate": 0.00019348027523392273, + "loss": 6.4288, "step": 2550 }, { "epoch": 0.2, - "learning_rate": 0.00019340672015845557, - "loss": 0.982, + "learning_rate": 0.0001934305845667748, + "loss": 6.3836, "step": 2555 }, { - "epoch": 0.21, - "learning_rate": 0.00019335670658613023, - "loss": 1.0036, + "epoch": 0.2, + "learning_rate": 0.00019338071168714053, + "loss": 6.3604, "step": 2560 }, { "epoch": 0.21, - "learning_rate": 0.00019330651055629453, - "loss": 1.0092, + "learning_rate": 0.0001933306566922842, + "loss": 6.3361, "step": 2565 }, { "epoch": 0.21, - "learning_rate": 0.0001932561321670522, - "loss": 1.0185, + "learning_rate": 0.00019328041967982536, + "loss": 6.3726, "step": 2570 }, { "epoch": 0.21, - "learning_rate": 0.00019320557151686338, - "loss": 1.0178, + "learning_rate": 0.00019323000074773842, + "loss": 6.3293, "step": 2575 }, { "epoch": 0.21, - "learning_rate": 0.00019315482870454443, - "loss": 0.9791, + "learning_rate": 0.0001931793999943526, + "loss": 6.3296, "step": 2580 }, { "epoch": 0.21, - "learning_rate": 0.00019310390382926777, - "loss": 1.0109, + "learning_rate": 0.00019312861751835184, + "loss": 6.316, "step": 2585 }, { "epoch": 0.21, - "learning_rate": 0.00019305279699056155, - "loss": 0.9849, + "learning_rate": 0.0001930776534187743, + "loss": 6.2917, "step": 2590 }, { "epoch": 0.21, - "learning_rate": 0.0001930015082883097, - "loss": 1.0736, + "learning_rate": 0.00019302650779501245, + "loss": 6.3075, "step": 2595 }, { "epoch": 0.21, - "learning_rate": 0.00019295003782275141, - "loss": 1.0241, + "learning_rate": 0.00019297518074681277, + "loss": 6.3294, "step": 2600 }, { "epoch": 0.21, - "learning_rate": 0.00019289838569448126, - "loss": 0.9893, + "learning_rate": 0.00019292367237427554, + "loss": 6.2665, "step": 2605 }, { "epoch": 0.21, - "learning_rate": 0.00019284655200444879, - "loss": 1.0337, + "learning_rate": 0.00019287198277785473, + "loss": 6.26, "step": 2610 }, { "epoch": 0.21, - "learning_rate": 0.0001927945368539584, - "loss": 1.0693, + "learning_rate": 0.00019282011205835767, + "loss": 6.277, "step": 2615 }, { "epoch": 0.21, - "learning_rate": 0.00019274234034466917, - "loss": 0.9955, + "learning_rate": 0.00019276806031694491, + "loss": 6.2314, "step": 2620 }, { "epoch": 0.21, - "learning_rate": 0.0001926899625785946, - "loss": 0.9341, + "learning_rate": 0.00019271582765513015, + "loss": 6.2353, "step": 2625 }, { "epoch": 0.21, - "learning_rate": 0.00019263740365810245, - "loss": 0.9925, + "learning_rate": 0.00019266341417477977, + "loss": 6.2729, "step": 2630 }, { "epoch": 0.21, - "learning_rate": 0.0001925846636859145, - "loss": 0.949, + "learning_rate": 0.00019261081997811293, + "loss": 6.2684, "step": 2635 }, { "epoch": 0.21, - "learning_rate": 0.00019253174276510646, - "loss": 1.025, + "learning_rate": 0.00019255804516770113, + "loss": 6.2154, "step": 2640 }, { "epoch": 0.21, - "learning_rate": 0.0001924786409991076, - "loss": 0.9874, + "learning_rate": 0.0001925050898464682, + "loss": 6.1966, "step": 2645 }, { "epoch": 0.21, - "learning_rate": 0.00019242535849170068, - "loss": 0.984, + "learning_rate": 0.00019245195411768992, + "loss": 6.2089, "step": 2650 }, { "epoch": 0.21, - "learning_rate": 0.00019237189534702172, - "loss": 0.9974, + "learning_rate": 0.00019239863808499398, + "loss": 6.2215, "step": 2655 }, { "epoch": 0.21, - "learning_rate": 0.0001923182516695597, - "loss": 1.0371, + "learning_rate": 0.00019234514185235974, + "loss": 6.2477, "step": 2660 }, { "epoch": 0.21, - "learning_rate": 0.00019226442756415654, - "loss": 1.0329, + "learning_rate": 0.0001922914655241178, + "loss": 6.1931, "step": 2665 }, { "epoch": 0.21, - "learning_rate": 0.00019221042313600673, - "loss": 1.0193, + "learning_rate": 0.0001922376092049502, + "loss": 6.1564, "step": 2670 }, { "epoch": 0.21, - "learning_rate": 0.0001921562384906572, - "loss": 0.9922, + "learning_rate": 0.00019218357299988998, + "loss": 6.1793, "step": 2675 }, { "epoch": 0.21, - "learning_rate": 0.0001921018737340071, - "loss": 0.9562, + "learning_rate": 0.00019212935701432091, + "loss": 6.2409, "step": 2680 }, { - "epoch": 0.22, - "learning_rate": 0.00019204732897230758, - "loss": 1.073, + "epoch": 0.21, + "learning_rate": 0.0001920749613539774, + "loss": 6.1351, "step": 2685 }, { "epoch": 0.22, - "learning_rate": 0.0001919926043121616, - "loss": 1.0093, + "learning_rate": 0.0001920203861249443, + "loss": 6.2319, "step": 2690 }, { "epoch": 0.22, - "learning_rate": 0.0001919376998605237, - "loss": 1.021, + "learning_rate": 0.00019196563143365663, + "loss": 6.1937, "step": 2695 }, { "epoch": 0.22, - "learning_rate": 0.00019188261572469987, - "loss": 0.9928, + "learning_rate": 0.00019191069738689947, + "loss": 6.1798, "step": 2700 }, { "epoch": 0.22, - "learning_rate": 0.0001918273520123472, - "loss": 1.0269, + "learning_rate": 0.00019185558409180763, + "loss": 6.2213, "step": 2705 }, { "epoch": 0.22, - "learning_rate": 0.00019177190883147376, - "loss": 0.9935, + "learning_rate": 0.00019180029165586553, + "loss": 6.1014, "step": 2710 }, { "epoch": 0.22, - "learning_rate": 0.00019171628629043844, - "loss": 0.9542, + "learning_rate": 0.0001917448201869069, + "loss": 6.2247, "step": 2715 }, { "epoch": 0.22, - "learning_rate": 0.0001916604844979506, - "loss": 1.0454, + "learning_rate": 0.0001916891697931147, + "loss": 6.1906, "step": 2720 }, { "epoch": 0.22, - "learning_rate": 0.00019160450356306994, - "loss": 1.0117, + "learning_rate": 0.00019163334058302077, + "loss": 6.1407, "step": 2725 }, { "epoch": 0.22, - "learning_rate": 0.00019154834359520634, - "loss": 0.9887, + "learning_rate": 0.00019157733266550575, + "loss": 6.1681, "step": 2730 }, { "epoch": 0.22, - "learning_rate": 0.0001914920047041195, - "loss": 1.0235, + "learning_rate": 0.00019152114614979878, + "loss": 6.1832, "step": 2735 }, { "epoch": 0.22, - "learning_rate": 0.00019143548699991888, - "loss": 0.963, + "learning_rate": 0.0001914647811454773, + "loss": 6.0753, "step": 2740 }, { "epoch": 0.22, - "learning_rate": 0.0001913787905930634, - "loss": 1.0095, + "learning_rate": 0.00019140823776246687, + "loss": 6.0713, "step": 2745 }, { "epoch": 0.22, - "learning_rate": 0.00019132191559436117, - "loss": 1.0072, + "learning_rate": 0.00019135151611104085, + "loss": 6.0799, "step": 2750 }, { "epoch": 0.22, - "learning_rate": 0.00019126486211496944, - "loss": 0.9816, + "learning_rate": 0.00019129461630182037, + "loss": 6.081, "step": 2755 }, { "epoch": 0.22, - "learning_rate": 0.0001912076302663942, - "loss": 0.9895, + "learning_rate": 0.000191237538445774, + "loss": 6.058, "step": 2760 }, { "epoch": 0.22, - "learning_rate": 0.00019115022016049014, - "loss": 1.0136, + "learning_rate": 0.00019118028265421746, + "loss": 6.044, "step": 2765 }, { "epoch": 0.22, - "learning_rate": 0.00019109263190946027, - "loss": 0.9968, + "learning_rate": 0.0001911228490388136, + "loss": 6.0452, "step": 2770 }, { "epoch": 0.22, - "learning_rate": 0.00019103486562585572, - "loss": 0.9609, + "learning_rate": 0.00019106523771157194, + "loss": 6.0804, "step": 2775 }, { "epoch": 0.22, - "learning_rate": 0.00019097692142257575, - "loss": 1.0061, + "learning_rate": 0.00019100744878484876, + "loss": 6.0471, "step": 2780 }, { "epoch": 0.22, - "learning_rate": 0.00019091879941286714, - "loss": 0.957, + "learning_rate": 0.0001909494823713465, + "loss": 6.1165, "step": 2785 }, { "epoch": 0.22, - "learning_rate": 0.0001908604997103243, - "loss": 1.0251, + "learning_rate": 0.0001908913385841139, + "loss": 6.0489, "step": 2790 }, { "epoch": 0.22, - "learning_rate": 0.00019080202242888894, - "loss": 1.0293, + "learning_rate": 0.00019083301753654555, + "loss": 6.0567, "step": 2795 }, { "epoch": 0.22, - "learning_rate": 0.00019074336768284972, - "loss": 1.0501, + "learning_rate": 0.00019077451934238174, + "loss": 6.0508, "step": 2800 }, { "epoch": 0.22, - "learning_rate": 0.0001906845355868422, - "loss": 1.081, + "learning_rate": 0.00019071584411570823, + "loss": 6.0439, "step": 2805 }, { - "epoch": 0.23, - "learning_rate": 0.0001906255262558486, - "loss": 1.0319, + "epoch": 0.22, + "learning_rate": 0.00019065699197095615, + "loss": 6.1169, "step": 2810 }, { "epoch": 0.23, - "learning_rate": 0.00019056633980519752, - "loss": 1.0182, + "learning_rate": 0.00019059796302290147, + "loss": 6.0411, "step": 2815 }, { "epoch": 0.23, - "learning_rate": 0.00019050697635056364, - "loss": 0.9714, + "learning_rate": 0.00019053875738666515, + "loss": 5.9977, "step": 2820 }, { "epoch": 0.23, - "learning_rate": 0.00019044743600796763, - "loss": 1.0165, + "learning_rate": 0.00019047937517771262, + "loss": 6.0804, "step": 2825 }, { "epoch": 0.23, - "learning_rate": 0.00019038771889377595, - "loss": 1.016, + "learning_rate": 0.0001904198165118537, + "loss": 6.0493, "step": 2830 }, { "epoch": 0.23, - "learning_rate": 0.00019032782512470037, - "loss": 1.0301, + "learning_rate": 0.00019036008150524237, + "loss": 5.9746, "step": 2835 }, { "epoch": 0.23, - "learning_rate": 0.0001902677548177981, - "loss": 1.041, + "learning_rate": 0.00019030017027437648, + "loss": 5.9989, "step": 2840 }, { "epoch": 0.23, - "learning_rate": 0.00019020750809047127, - "loss": 0.9641, + "learning_rate": 0.0001902400829360976, + "loss": 5.9985, "step": 2845 }, { "epoch": 0.23, - "learning_rate": 0.00019014708506046683, - "loss": 0.9965, + "learning_rate": 0.00019017981960759074, + "loss": 6.0402, "step": 2850 }, { "epoch": 0.23, - "learning_rate": 0.00019008648584587632, - "loss": 0.9584, + "learning_rate": 0.00019011938040638414, + "loss": 5.9829, "step": 2855 }, { "epoch": 0.23, - "learning_rate": 0.00019002571056513556, - "loss": 0.9977, + "learning_rate": 0.00019005876545034897, + "loss": 6.0061, "step": 2860 }, { "epoch": 0.23, - "learning_rate": 0.00018996475933702459, - "loss": 1.0238, + "learning_rate": 0.00018999797485769925, + "loss": 6.0067, "step": 2865 }, { "epoch": 0.23, - "learning_rate": 0.00018990363228066714, - "loss": 0.9942, + "learning_rate": 0.00018993700874699146, + "loss": 5.9176, "step": 2870 }, { "epoch": 0.23, - "learning_rate": 0.0001898423295155308, - "loss": 0.9935, + "learning_rate": 0.00018987586723712453, + "loss": 5.9826, "step": 2875 }, { "epoch": 0.23, - "learning_rate": 0.00018978085116142638, - "loss": 1.0338, + "learning_rate": 0.00018981455044733922, + "loss": 5.9912, "step": 2880 }, { "epoch": 0.23, - "learning_rate": 0.000189719197338508, - "loss": 1.0035, + "learning_rate": 0.00018975305849721835, + "loss": 5.928, "step": 2885 }, { "epoch": 0.23, - "learning_rate": 0.0001896573681672726, - "loss": 1.0074, + "learning_rate": 0.00018969139150668625, + "loss": 5.977, "step": 2890 }, { "epoch": 0.23, - "learning_rate": 0.00018959536376855993, - "loss": 1.0119, + "learning_rate": 0.0001896295495960086, + "loss": 5.9415, "step": 2895 }, { "epoch": 0.23, - "learning_rate": 0.00018953318426355215, - "loss": 1.0241, + "learning_rate": 0.00018956753288579227, + "loss": 5.9975, "step": 2900 }, { "epoch": 0.23, - "learning_rate": 0.00018947082977377363, - "loss": 0.988, + "learning_rate": 0.00018950534149698496, + "loss": 5.9404, "step": 2905 }, { "epoch": 0.23, - "learning_rate": 0.0001894083004210908, - "loss": 0.9722, + "learning_rate": 0.00018944297555087517, + "loss": 5.964, "step": 2910 }, { "epoch": 0.23, - "learning_rate": 0.00018934559632771176, - "loss": 0.9779, + "learning_rate": 0.0001893804351690917, + "loss": 5.9233, "step": 2915 }, { "epoch": 0.23, - "learning_rate": 0.00018928271761618624, - "loss": 1.0269, + "learning_rate": 0.0001893177204736036, + "loss": 5.9423, "step": 2920 }, { "epoch": 0.23, - "learning_rate": 0.00018921966440940512, - "loss": 0.9495, + "learning_rate": 0.00018925483158671984, + "loss": 5.9281, "step": 2925 }, { "epoch": 0.23, - "learning_rate": 0.00018915643683060037, - "loss": 1.0772, + "learning_rate": 0.00018919176863108915, + "loss": 5.903, "step": 2930 }, { - "epoch": 0.24, - "learning_rate": 0.0001890930350033448, - "loss": 1.0201, + "epoch": 0.23, + "learning_rate": 0.0001891285317296997, + "loss": 5.892, "step": 2935 }, { "epoch": 0.24, - "learning_rate": 0.00018902945905155174, - "loss": 1.0191, + "learning_rate": 0.00018906512100587893, + "loss": 6.0272, "step": 2940 }, { "epoch": 0.24, - "learning_rate": 0.00018896570909947475, - "loss": 1.0171, + "learning_rate": 0.00018900153658329324, + "loss": 5.9319, "step": 2945 }, { "epoch": 0.24, - "learning_rate": 0.00018890178527170758, - "loss": 0.9788, + "learning_rate": 0.0001889377785859478, + "loss": 5.9052, "step": 2950 }, { "epoch": 0.24, - "learning_rate": 0.00018883768769318372, - "loss": 1.0263, + "learning_rate": 0.00018887384713818633, + "loss": 5.9463, "step": 2955 }, { "epoch": 0.24, - "learning_rate": 0.00018877341648917633, - "loss": 1.0251, + "learning_rate": 0.0001888097423646907, + "loss": 5.9008, "step": 2960 }, { "epoch": 0.24, - "learning_rate": 0.00018870897178529784, - "loss": 1.028, + "learning_rate": 0.000188745464390481, + "loss": 5.9467, "step": 2965 }, { "epoch": 0.24, - "learning_rate": 0.00018864435370749975, - "loss": 0.9739, + "learning_rate": 0.00018868101334091494, + "loss": 5.941, "step": 2970 }, { "epoch": 0.24, - "learning_rate": 0.00018857956238207244, - "loss": 1.0387, + "learning_rate": 0.00018861638934168783, + "loss": 5.8766, "step": 2975 }, { "epoch": 0.24, - "learning_rate": 0.00018851459793564493, - "loss": 1.0048, + "learning_rate": 0.00018855159251883228, + "loss": 5.9066, "step": 2980 }, { "epoch": 0.24, - "learning_rate": 0.00018844946049518448, - "loss": 1.0366, + "learning_rate": 0.00018848662299871793, + "loss": 5.9495, "step": 2985 }, { "epoch": 0.24, - "learning_rate": 0.00018838415018799655, - "loss": 1.054, + "learning_rate": 0.00018842148090805126, + "loss": 5.8691, "step": 2990 }, { "epoch": 0.24, - "learning_rate": 0.00018831866714172444, - "loss": 1.004, + "learning_rate": 0.00018835616637387522, + "loss": 5.8627, "step": 2995 }, { "epoch": 0.24, - "learning_rate": 0.000188253011484349, - "loss": 1.0201, + "learning_rate": 0.00018829067952356917, + "loss": 5.8325, "step": 3000 }, { "epoch": 0.24, - "learning_rate": 0.00018818718334418844, - "loss": 1.0072, + "learning_rate": 0.0001882250204848485, + "loss": 5.8804, "step": 3005 }, { "epoch": 0.24, - "learning_rate": 0.00018812118284989814, - "loss": 1.0414, + "learning_rate": 0.00018815918938576429, + "loss": 5.8375, "step": 3010 }, { "epoch": 0.24, - "learning_rate": 0.0001880550101304703, - "loss": 1.0388, + "learning_rate": 0.0001880931863547034, + "loss": 5.9393, "step": 3015 }, { "epoch": 0.24, - "learning_rate": 0.00018798866531523361, - "loss": 1.0292, + "learning_rate": 0.0001880270115203878, + "loss": 5.8522, "step": 3020 }, { "epoch": 0.24, - "learning_rate": 0.00018792214853385336, - "loss": 0.9986, + "learning_rate": 0.00018796066501187465, + "loss": 5.8895, "step": 3025 }, { "epoch": 0.24, - "learning_rate": 0.00018785545991633064, - "loss": 0.9647, + "learning_rate": 0.00018789414695855584, + "loss": 5.8154, "step": 3030 }, { "epoch": 0.24, - "learning_rate": 0.00018778859959300257, - "loss": 0.9871, + "learning_rate": 0.00018782745749015785, + "loss": 5.9122, "step": 3035 }, { "epoch": 0.24, - "learning_rate": 0.0001877215676945418, - "loss": 1.0417, + "learning_rate": 0.00018776059673674143, + "loss": 5.8095, "step": 3040 }, { "epoch": 0.24, - "learning_rate": 0.00018765436435195634, - "loss": 1.0166, + "learning_rate": 0.00018769356482870142, + "loss": 5.8542, "step": 3045 }, { "epoch": 0.24, - "learning_rate": 0.00018758698969658917, - "loss": 0.9835, + "learning_rate": 0.00018762636189676642, + "loss": 5.8483, "step": 3050 }, { "epoch": 0.24, - "learning_rate": 0.00018751944386011818, - "loss": 1.0827, + "learning_rate": 0.00018755898807199856, + "loss": 5.8793, "step": 3055 }, { - "epoch": 0.25, - "learning_rate": 0.00018745172697455582, - "loss": 1.0716, + "epoch": 0.24, + "learning_rate": 0.00018749144348579333, + "loss": 5.8655, "step": 3060 }, { "epoch": 0.25, - "learning_rate": 0.0001873838391722488, - "loss": 1.0508, + "learning_rate": 0.00018742372826987915, + "loss": 5.856, "step": 3065 }, { "epoch": 0.25, - "learning_rate": 0.00018731578058587785, - "loss": 0.9823, + "learning_rate": 0.00018735584255631727, + "loss": 5.8795, "step": 3070 }, { "epoch": 0.25, - "learning_rate": 0.0001872475513484576, - "loss": 1.0721, + "learning_rate": 0.0001872877864775014, + "loss": 5.9022, "step": 3075 }, { "epoch": 0.25, - "learning_rate": 0.00018717915159333605, - "loss": 1.013, + "learning_rate": 0.00018721956016615757, + "loss": 5.8075, "step": 3080 }, { "epoch": 0.25, - "learning_rate": 0.00018711058145419457, - "loss": 1.0092, + "learning_rate": 0.00018715116375534371, + "loss": 5.829, "step": 3085 }, { "epoch": 0.25, - "learning_rate": 0.00018704184106504748, - "loss": 0.9949, + "learning_rate": 0.00018708259737844964, + "loss": 5.8623, "step": 3090 }, { "epoch": 0.25, - "learning_rate": 0.00018697293056024186, - "loss": 1.0113, + "learning_rate": 0.0001870138611691965, + "loss": 5.8364, "step": 3095 }, { "epoch": 0.25, - "learning_rate": 0.00018690385007445723, - "loss": 1.0757, + "learning_rate": 0.0001869449552616367, + "loss": 5.7889, "step": 3100 }, { "epoch": 0.25, - "learning_rate": 0.0001868345997427054, - "loss": 0.9902, + "learning_rate": 0.00018687587979015363, + "loss": 5.8653, "step": 3105 }, { "epoch": 0.25, - "learning_rate": 0.00018676517970033006, - "loss": 1.0398, + "learning_rate": 0.00018680663488946134, + "loss": 5.8716, "step": 3110 }, { "epoch": 0.25, - "learning_rate": 0.00018669559008300656, - "loss": 1.0549, + "learning_rate": 0.00018673722069460435, + "loss": 5.9137, "step": 3115 }, { "epoch": 0.25, - "learning_rate": 0.00018662583102674183, - "loss": 1.0059, + "learning_rate": 0.00018666763734095728, + "loss": 5.9688, "step": 3120 }, { "epoch": 0.25, - "learning_rate": 0.00018655590266787372, - "loss": 1.0637, + "learning_rate": 0.00018659788496422467, + "loss": 5.8987, "step": 3125 }, { "epoch": 0.25, - "learning_rate": 0.00018648580514307113, - "loss": 1.0528, + "learning_rate": 0.00018652796370044076, + "loss": 5.8424, "step": 3130 }, { "epoch": 0.25, - "learning_rate": 0.00018641553858933356, - "loss": 0.9809, + "learning_rate": 0.0001864578736859691, + "loss": 5.8277, "step": 3135 }, { "epoch": 0.25, - "learning_rate": 0.0001863451031439908, - "loss": 0.9777, + "learning_rate": 0.00018638761505750231, + "loss": 5.8249, "step": 3140 }, { "epoch": 0.25, - "learning_rate": 0.0001862744989447028, - "loss": 1.0261, + "learning_rate": 0.00018631718795206195, + "loss": 5.7759, "step": 3145 }, { "epoch": 0.25, - "learning_rate": 0.00018620372612945923, - "loss": 1.0005, + "learning_rate": 0.00018624659250699805, + "loss": 5.8483, "step": 3150 }, { "epoch": 0.25, - "learning_rate": 0.00018613278483657942, - "loss": 1.0391, + "learning_rate": 0.00018617582885998907, + "loss": 5.8686, "step": 3155 }, { "epoch": 0.25, - "learning_rate": 0.0001860616752047119, - "loss": 1.0521, + "learning_rate": 0.00018610489714904134, + "loss": 5.8265, "step": 3160 }, { "epoch": 0.25, - "learning_rate": 0.0001859903973728342, - "loss": 1.0034, + "learning_rate": 0.00018603379751248907, + "loss": 5.8335, "step": 3165 }, { "epoch": 0.25, - "learning_rate": 0.0001859189514802526, - "loss": 1.007, + "learning_rate": 0.00018596253008899393, + "loss": 5.779, "step": 3170 }, { "epoch": 0.25, - "learning_rate": 0.0001858473376666019, - "loss": 1.0339, + "learning_rate": 0.00018589109501754484, + "loss": 5.7864, "step": 3175 }, { "epoch": 0.25, - "learning_rate": 0.00018577555607184496, - "loss": 0.9684, + "learning_rate": 0.00018581949243745766, + "loss": 5.7153, "step": 3180 }, { - "epoch": 0.26, - "learning_rate": 0.00018570360683627265, - "loss": 1.0635, + "epoch": 0.25, + "learning_rate": 0.00018574772248837492, + "loss": 5.7103, "step": 3185 }, { "epoch": 0.26, - "learning_rate": 0.0001856314901005035, - "loss": 1.0133, + "learning_rate": 0.00018567578531026557, + "loss": 5.8013, "step": 3190 }, { "epoch": 0.26, - "learning_rate": 0.00018555920600548325, - "loss": 0.9819, + "learning_rate": 0.0001856036810434247, + "loss": 5.7441, "step": 3195 }, { "epoch": 0.26, - "learning_rate": 0.00018548675469248492, - "loss": 0.9929, + "learning_rate": 0.00018553140982847328, + "loss": 5.7557, "step": 3200 }, { "epoch": 0.26, - "learning_rate": 0.00018541413630310824, - "loss": 0.9995, + "learning_rate": 0.00018545897180635782, + "loss": 5.7743, "step": 3205 }, { "epoch": 0.26, - "learning_rate": 0.0001853413509792795, - "loss": 1.0183, + "learning_rate": 0.00018538636711835018, + "loss": 5.7328, "step": 3210 }, { "epoch": 0.26, - "learning_rate": 0.0001852683988632512, - "loss": 1.0172, + "learning_rate": 0.0001853135959060473, + "loss": 5.7897, "step": 3215 }, { "epoch": 0.26, - "learning_rate": 0.0001851952800976019, - "loss": 0.9529, + "learning_rate": 0.00018524065831137073, + "loss": 5.7668, "step": 3220 }, { "epoch": 0.26, - "learning_rate": 0.00018512199482523584, - "loss": 0.9806, + "learning_rate": 0.00018516755447656676, + "loss": 5.7784, "step": 3225 }, { "epoch": 0.26, - "learning_rate": 0.00018504854318938268, - "loss": 0.9458, + "learning_rate": 0.00018509428454420562, + "loss": 5.7874, "step": 3230 }, { "epoch": 0.26, - "learning_rate": 0.00018497492533359715, - "loss": 1.0369, + "learning_rate": 0.0001850208486571817, + "loss": 5.7636, "step": 3235 }, { "epoch": 0.26, - "learning_rate": 0.00018490114140175896, - "loss": 0.9963, + "learning_rate": 0.00018494724695871284, + "loss": 5.76, "step": 3240 }, { "epoch": 0.26, - "learning_rate": 0.00018482719153807233, - "loss": 0.9794, + "learning_rate": 0.0001848734795923404, + "loss": 5.7713, "step": 3245 }, { "epoch": 0.26, - "learning_rate": 0.00018475307588706577, - "loss": 0.996, + "learning_rate": 0.0001847995467019288, + "loss": 5.8089, "step": 3250 }, { "epoch": 0.26, - "learning_rate": 0.00018467879459359185, - "loss": 1.0697, + "learning_rate": 0.0001847254484316652, + "loss": 5.7929, "step": 3255 }, { "epoch": 0.26, - "learning_rate": 0.0001846043478028269, - "loss": 1.0169, + "learning_rate": 0.0001846511849260594, + "loss": 5.7984, "step": 3260 }, { "epoch": 0.26, - "learning_rate": 0.00018452973566027056, - "loss": 1.0606, + "learning_rate": 0.00018457675632994335, + "loss": 5.7461, "step": 3265 }, { "epoch": 0.26, - "learning_rate": 0.00018445495831174581, - "loss": 0.9938, + "learning_rate": 0.00018450216278847106, + "loss": 5.8052, "step": 3270 }, { "epoch": 0.26, - "learning_rate": 0.00018438001590339838, - "loss": 0.9623, + "learning_rate": 0.00018442740444711816, + "loss": 5.8085, "step": 3275 }, { "epoch": 0.26, - "learning_rate": 0.0001843049085816967, - "loss": 1.0117, + "learning_rate": 0.00018435248145168165, + "loss": 5.7897, "step": 3280 }, { "epoch": 0.26, - "learning_rate": 0.00018422963649343145, - "loss": 1.105, + "learning_rate": 0.00018427739394827973, + "loss": 5.7377, "step": 3285 }, { "epoch": 0.26, - "learning_rate": 0.00018415419978571527, - "loss": 1.0052, + "learning_rate": 0.00018420214208335141, + "loss": 5.8137, "step": 3290 }, { "epoch": 0.26, - "learning_rate": 0.00018407859860598266, - "loss": 1.0465, + "learning_rate": 0.00018412672600365623, + "loss": 5.717, "step": 3295 }, { "epoch": 0.26, - "learning_rate": 0.0001840028331019895, - "loss": 1.075, + "learning_rate": 0.00018405114585627393, + "loss": 5.7066, "step": 3300 }, { "epoch": 0.26, - "learning_rate": 0.00018392690342181284, - "loss": 0.9508, + "learning_rate": 0.00018397540178860435, + "loss": 5.7831, "step": 3305 }, { - "epoch": 0.27, - "learning_rate": 0.00018385080971385054, - "loss": 1.0299, + "epoch": 0.26, + "learning_rate": 0.00018389949394836686, + "loss": 5.7129, "step": 3310 }, { "epoch": 0.27, - "learning_rate": 0.00018377455212682117, - "loss": 1.0255, + "learning_rate": 0.0001838234224836004, + "loss": 5.7188, "step": 3315 }, { "epoch": 0.27, - "learning_rate": 0.00018369813080976343, - "loss": 1.0163, + "learning_rate": 0.00018374718754266288, + "loss": 5.7219, "step": 3320 }, { "epoch": 0.27, - "learning_rate": 0.00018362154591203617, - "loss": 1.0076, + "learning_rate": 0.0001836707892742311, + "loss": 5.7743, "step": 3325 }, { "epoch": 0.27, - "learning_rate": 0.0001835447975833178, - "loss": 1.0741, + "learning_rate": 0.00018359422782730032, + "loss": 5.7211, "step": 3330 }, { "epoch": 0.27, - "learning_rate": 0.00018346788597360628, - "loss": 1.0021, + "learning_rate": 0.0001835175033511841, + "loss": 5.7204, "step": 3335 }, { "epoch": 0.27, - "learning_rate": 0.00018339081123321858, - "loss": 1.019, + "learning_rate": 0.00018344061599551398, + "loss": 5.6334, "step": 3340 }, { "epoch": 0.27, - "learning_rate": 0.00018331357351279053, - "loss": 0.9901, + "learning_rate": 0.00018336356591023903, + "loss": 5.7067, "step": 3345 }, { "epoch": 0.27, - "learning_rate": 0.00018323617296327646, - "loss": 1.0591, + "learning_rate": 0.0001832863532456258, + "loss": 5.7076, "step": 3350 }, { "epoch": 0.27, - "learning_rate": 0.00018315860973594904, - "loss": 1.0034, + "learning_rate": 0.00018320897815225787, + "loss": 5.7314, "step": 3355 }, { "epoch": 0.27, - "learning_rate": 0.00018308088398239874, - "loss": 1.0216, + "learning_rate": 0.00018313144078103556, + "loss": 5.6985, "step": 3360 }, { "epoch": 0.27, - "learning_rate": 0.0001830029958545338, - "loss": 0.9797, + "learning_rate": 0.00018305374128317576, + "loss": 5.7099, "step": 3365 }, { "epoch": 0.27, - "learning_rate": 0.0001829249455045797, - "loss": 1.0207, + "learning_rate": 0.00018297587981021144, + "loss": 5.6912, "step": 3370 }, { "epoch": 0.27, - "learning_rate": 0.00018284673308507907, - "loss": 1.0157, + "learning_rate": 0.00018289785651399155, + "loss": 5.7008, "step": 3375 }, { "epoch": 0.27, - "learning_rate": 0.00018276835874889117, - "loss": 1.0922, + "learning_rate": 0.00018281967154668054, + "loss": 5.7486, "step": 3380 }, { "epoch": 0.27, - "learning_rate": 0.00018268982264919186, - "loss": 0.9881, + "learning_rate": 0.00018274132506075828, + "loss": 5.7009, "step": 3385 }, { "epoch": 0.27, - "learning_rate": 0.000182611124939473, - "loss": 1.0647, + "learning_rate": 0.00018266281720901952, + "loss": 5.7126, "step": 3390 }, { "epoch": 0.27, - "learning_rate": 0.00018253226577354234, - "loss": 1.0696, + "learning_rate": 0.0001825841481445738, + "loss": 5.7685, "step": 3395 }, { "epoch": 0.27, - "learning_rate": 0.00018245324530552335, - "loss": 0.9983, + "learning_rate": 0.00018250531802084503, + "loss": 5.7084, "step": 3400 }, { "epoch": 0.27, - "learning_rate": 0.0001823740636898545, - "loss": 1.0794, + "learning_rate": 0.00018242632699157124, + "loss": 5.6336, "step": 3405 }, { "epoch": 0.27, - "learning_rate": 0.0001822947210812894, - "loss": 1.0087, + "learning_rate": 0.0001823471752108042, + "loss": 5.7537, "step": 3410 }, { "epoch": 0.27, - "learning_rate": 0.00018221521763489616, - "loss": 1.0096, + "learning_rate": 0.0001822678628329093, + "loss": 5.7047, "step": 3415 }, { "epoch": 0.27, - "learning_rate": 0.00018213555350605738, - "loss": 1.0247, + "learning_rate": 0.00018218839001256503, + "loss": 5.6344, "step": 3420 }, { "epoch": 0.27, - "learning_rate": 0.0001820557288504696, - "loss": 0.9985, + "learning_rate": 0.00018210875690476282, + "loss": 5.7453, "step": 3425 }, { "epoch": 0.27, - "learning_rate": 0.00018197574382414309, - "loss": 1.0214, + "learning_rate": 0.00018202896366480674, + "loss": 5.7447, "step": 3430 }, { - "epoch": 0.28, - "learning_rate": 0.00018189559858340163, - "loss": 0.9704, + "epoch": 0.27, + "learning_rate": 0.00018194901044831313, + "loss": 5.6894, "step": 3435 }, { "epoch": 0.28, - "learning_rate": 0.00018181529328488205, - "loss": 0.9979, + "learning_rate": 0.00018186889741121024, + "loss": 5.706, "step": 3440 }, { "epoch": 0.28, - "learning_rate": 0.000181734828085534, - "loss": 1.0491, + "learning_rate": 0.00018178862470973815, + "loss": 5.5987, "step": 3445 }, { "epoch": 0.28, - "learning_rate": 0.0001816542031426197, - "loss": 1.0163, + "learning_rate": 0.00018170819250044821, + "loss": 5.7009, "step": 3450 }, { "epoch": 0.28, - "learning_rate": 0.00018157341861371355, - "loss": 0.9504, + "learning_rate": 0.00018162760094020291, + "loss": 5.6811, "step": 3455 }, { "epoch": 0.28, - "learning_rate": 0.00018149247465670186, - "loss": 1.0287, + "learning_rate": 0.00018154685018617554, + "loss": 5.6997, "step": 3460 }, { "epoch": 0.28, - "learning_rate": 0.00018141137142978244, - "loss": 0.9865, + "learning_rate": 0.0001814659403958498, + "loss": 5.6916, "step": 3465 }, { "epoch": 0.28, - "learning_rate": 0.00018133010909146449, - "loss": 1.0198, + "learning_rate": 0.0001813848717270195, + "loss": 5.6774, "step": 3470 }, { "epoch": 0.28, - "learning_rate": 0.00018124868780056814, - "loss": 0.9559, + "learning_rate": 0.00018130364433778846, + "loss": 5.6775, "step": 3475 }, { "epoch": 0.28, - "learning_rate": 0.00018116710771622416, - "loss": 1.0432, + "learning_rate": 0.00018122225838656988, + "loss": 5.6599, "step": 3480 }, { "epoch": 0.28, - "learning_rate": 0.0001810853689978737, - "loss": 0.9388, + "learning_rate": 0.0001811407140320863, + "loss": 5.7068, "step": 3485 }, { "epoch": 0.28, - "learning_rate": 0.0001810034718052679, - "loss": 1.0341, + "learning_rate": 0.00018105901143336913, + "loss": 5.6826, "step": 3490 }, { "epoch": 0.28, - "learning_rate": 0.0001809214162984677, - "loss": 0.9683, + "learning_rate": 0.00018097715074975846, + "loss": 5.6751, "step": 3495 }, { "epoch": 0.28, - "learning_rate": 0.00018083920263784332, - "loss": 1.0284, + "learning_rate": 0.0001808951321409026, + "loss": 5.7074, "step": 3500 }, { "epoch": 0.28, - "learning_rate": 0.00018075683098407426, - "loss": 1.0758, + "learning_rate": 0.0001808129557667579, + "loss": 5.6665, "step": 3505 }, { "epoch": 0.28, - "learning_rate": 0.00018067430149814868, - "loss": 1.0055, + "learning_rate": 0.0001807306217875884, + "loss": 5.74, "step": 3510 }, { "epoch": 0.28, - "learning_rate": 0.0001805916143413632, - "loss": 1.0229, + "learning_rate": 0.00018064813036396547, + "loss": 5.6082, "step": 3515 }, { "epoch": 0.28, - "learning_rate": 0.00018050876967532268, - "loss": 0.9694, + "learning_rate": 0.00018056548165676757, + "loss": 5.6211, "step": 3520 }, { "epoch": 0.28, - "learning_rate": 0.0001804257676619397, - "loss": 1.0602, + "learning_rate": 0.0001804826758271799, + "loss": 5.6662, "step": 3525 }, { "epoch": 0.28, - "learning_rate": 0.0001803426084634345, - "loss": 0.9835, + "learning_rate": 0.00018039971303669407, + "loss": 5.5963, "step": 3530 }, { "epoch": 0.28, - "learning_rate": 0.0001802592922423344, - "loss": 1.019, + "learning_rate": 0.00018031659344710778, + "loss": 5.6565, "step": 3535 }, { "epoch": 0.28, - "learning_rate": 0.00018017581916147365, - "loss": 1.0395, + "learning_rate": 0.00018023331722052458, + "loss": 5.638, "step": 3540 }, { "epoch": 0.28, - "learning_rate": 0.00018009218938399313, - "loss": 1.0598, + "learning_rate": 0.00018014988451935349, + "loss": 5.7129, "step": 3545 }, { "epoch": 0.28, - "learning_rate": 0.00018000840307333993, - "loss": 1.0125, + "learning_rate": 0.00018006629550630865, + "loss": 5.7121, "step": 3550 }, { "epoch": 0.28, - "learning_rate": 0.000179924460393267, - "loss": 0.9905, + "learning_rate": 0.00017998255034440912, + "loss": 5.6157, "step": 3555 }, { - "epoch": 0.29, - "learning_rate": 0.000179840361507833, - "loss": 0.9747, + "epoch": 0.28, + "learning_rate": 0.00017989864919697835, + "loss": 5.6785, "step": 3560 }, { "epoch": 0.29, - "learning_rate": 0.00017975610658140182, - "loss": 1.0479, + "learning_rate": 0.0001798145922276442, + "loss": 5.6483, "step": 3565 }, { "epoch": 0.29, - "learning_rate": 0.00017967169577864236, - "loss": 0.9921, + "learning_rate": 0.00017973037960033824, + "loss": 5.6287, "step": 3570 }, { "epoch": 0.29, - "learning_rate": 0.0001795871292645281, - "loss": 1.0369, + "learning_rate": 0.00017964601147929575, + "loss": 5.7251, "step": 3575 }, { "epoch": 0.29, - "learning_rate": 0.00017950240720433696, - "loss": 1.025, + "learning_rate": 0.0001795614880290551, + "loss": 5.6648, "step": 3580 }, { "epoch": 0.29, - "learning_rate": 0.00017941752976365072, - "loss": 1.0081, + "learning_rate": 0.00017947680941445782, + "loss": 5.6058, "step": 3585 }, { "epoch": 0.29, - "learning_rate": 0.00017933249710835492, - "loss": 0.9758, + "learning_rate": 0.00017939197580064782, + "loss": 5.635, "step": 3590 }, { "epoch": 0.29, - "learning_rate": 0.00017924730940463847, - "loss": 1.0703, + "learning_rate": 0.0001793069873530714, + "loss": 5.6018, "step": 3595 }, { "epoch": 0.29, - "learning_rate": 0.0001791619668189933, - "loss": 1.0327, + "learning_rate": 0.00017922184423747683, + "loss": 5.6442, "step": 3600 }, { "epoch": 0.29, - "learning_rate": 0.00017907646951821396, - "loss": 1.0369, + "learning_rate": 0.000179136546619914, + "loss": 5.6065, "step": 3605 }, { "epoch": 0.29, - "learning_rate": 0.00017899081766939746, - "loss": 1.0738, + "learning_rate": 0.00017905109466673412, + "loss": 5.6272, "step": 3610 }, { "epoch": 0.29, - "learning_rate": 0.00017890501143994286, - "loss": 0.9957, + "learning_rate": 0.00017896548854458942, + "loss": 5.6707, "step": 3615 }, { "epoch": 0.29, - "learning_rate": 0.00017881905099755092, - "loss": 1.0362, + "learning_rate": 0.00017887972842043276, + "loss": 5.6345, "step": 3620 }, { "epoch": 0.29, - "learning_rate": 0.00017873293651022378, - "loss": 1.0233, + "learning_rate": 0.0001787938144615173, + "loss": 5.6612, "step": 3625 }, { "epoch": 0.29, - "learning_rate": 0.00017864666814626468, - "loss": 0.9894, + "learning_rate": 0.00017870774683539632, + "loss": 5.6838, "step": 3630 }, { "epoch": 0.29, - "learning_rate": 0.0001785602460742776, - "loss": 0.9665, + "learning_rate": 0.00017862152570992276, + "loss": 5.6449, "step": 3635 }, { "epoch": 0.29, - "learning_rate": 0.00017847367046316688, - "loss": 1.0345, + "learning_rate": 0.00017853515125324887, + "loss": 5.6459, "step": 3640 }, { "epoch": 0.29, - "learning_rate": 0.00017838694148213698, - "loss": 0.9693, + "learning_rate": 0.00017844862363382595, + "loss": 5.6287, "step": 3645 }, { "epoch": 0.29, - "learning_rate": 0.00017830005930069215, - "loss": 0.9941, + "learning_rate": 0.000178361943020404, + "loss": 5.6405, "step": 3650 }, { "epoch": 0.29, - "learning_rate": 0.00017821302408863594, - "loss": 0.9463, + "learning_rate": 0.00017827510958203147, + "loss": 5.5951, "step": 3655 }, { "epoch": 0.29, - "learning_rate": 0.00017812583601607107, - "loss": 1.0034, + "learning_rate": 0.00017818812348805474, + "loss": 5.5901, "step": 3660 }, { "epoch": 0.29, - "learning_rate": 0.00017803849525339902, - "loss": 1.0336, + "learning_rate": 0.000178100984908118, + "loss": 5.5628, "step": 3665 }, { "epoch": 0.29, - "learning_rate": 0.00017795100197131968, - "loss": 1.0396, + "learning_rate": 0.0001780136940121628, + "loss": 5.6414, "step": 3670 }, { "epoch": 0.29, - "learning_rate": 0.00017786335634083095, - "loss": 0.9742, + "learning_rate": 0.00017792625097042764, + "loss": 5.6175, "step": 3675 }, { "epoch": 0.29, - "learning_rate": 0.0001777755585332286, - "loss": 1.0031, + "learning_rate": 0.00017783865595344796, + "loss": 5.6338, "step": 3680 }, { - "epoch": 0.3, - "learning_rate": 0.00017768760872010573, - "loss": 1.0236, + "epoch": 0.29, + "learning_rate": 0.0001777509091320554, + "loss": 5.6118, "step": 3685 }, { "epoch": 0.3, - "learning_rate": 0.00017759950707335254, - "loss": 1.059, + "learning_rate": 0.00017766301067737773, + "loss": 5.6238, "step": 3690 }, { "epoch": 0.3, - "learning_rate": 0.00017751125376515599, - "loss": 1.0164, + "learning_rate": 0.0001775749607608384, + "loss": 5.6059, "step": 3695 }, { "epoch": 0.3, - "learning_rate": 0.00017742284896799947, - "loss": 0.9809, + "learning_rate": 0.00017748675955415633, + "loss": 5.6437, "step": 3700 }, { "epoch": 0.3, - "learning_rate": 0.00017733429285466242, - "loss": 1.005, + "learning_rate": 0.0001773984072293454, + "loss": 5.6047, "step": 3705 }, { "epoch": 0.3, - "learning_rate": 0.00017724558559821998, - "loss": 1.0229, + "learning_rate": 0.00017730990395871428, + "loss": 5.546, "step": 3710 }, { "epoch": 0.3, - "learning_rate": 0.0001771567273720427, - "loss": 1.0179, + "learning_rate": 0.00017722124991486596, + "loss": 5.5823, "step": 3715 }, { "epoch": 0.3, - "learning_rate": 0.00017706771834979628, - "loss": 0.9715, + "learning_rate": 0.0001771324452706975, + "loss": 5.6231, "step": 3720 }, { "epoch": 0.3, - "learning_rate": 0.00017697855870544101, - "loss": 0.9935, + "learning_rate": 0.00017704349019939967, + "loss": 5.5949, "step": 3725 }, { "epoch": 0.3, - "learning_rate": 0.0001768892486132316, - "loss": 1.0126, + "learning_rate": 0.0001769543848744567, + "loss": 5.6354, "step": 3730 }, { "epoch": 0.3, - "learning_rate": 0.00017679978824771683, - "loss": 1.0283, + "learning_rate": 0.0001768651294696456, + "loss": 5.6574, "step": 3735 }, { "epoch": 0.3, - "learning_rate": 0.00017671017778373913, - "loss": 0.9875, + "learning_rate": 0.0001767757241590364, + "loss": 5.5425, "step": 3740 }, { "epoch": 0.3, - "learning_rate": 0.00017662041739643436, - "loss": 0.9439, + "learning_rate": 0.00017668616911699117, + "loss": 5.6182, "step": 3745 }, { "epoch": 0.3, - "learning_rate": 0.00017653050726123128, - "loss": 1.0358, + "learning_rate": 0.00017659646451816428, + "loss": 5.6352, "step": 3750 }, { "epoch": 0.3, - "learning_rate": 0.00017644044755385139, - "loss": 1.041, + "learning_rate": 0.00017650661053750152, + "loss": 5.5919, "step": 3755 }, { "epoch": 0.3, - "learning_rate": 0.00017635023845030855, - "loss": 1.0085, + "learning_rate": 0.00017641660735024018, + "loss": 5.5688, "step": 3760 }, { "epoch": 0.3, - "learning_rate": 0.00017625988012690845, - "loss": 0.9701, + "learning_rate": 0.0001763264551319085, + "loss": 5.6186, "step": 3765 }, { "epoch": 0.3, - "learning_rate": 0.00017616937276024864, - "loss": 1.0209, + "learning_rate": 0.0001762361540583253, + "loss": 5.5435, "step": 3770 }, { "epoch": 0.3, - "learning_rate": 0.00017607871652721783, - "loss": 1.0484, + "learning_rate": 0.0001761457043055998, + "loss": 5.5826, "step": 3775 }, { "epoch": 0.3, - "learning_rate": 0.00017598791160499563, - "loss": 0.9809, + "learning_rate": 0.0001760551060501311, + "loss": 5.6238, "step": 3780 }, { "epoch": 0.3, - "learning_rate": 0.00017589695817105233, - "loss": 0.9916, + "learning_rate": 0.00017596435946860795, + "loss": 5.6139, "step": 3785 }, { "epoch": 0.3, - "learning_rate": 0.00017580585640314847, - "loss": 0.9982, + "learning_rate": 0.0001758734647380084, + "loss": 5.5699, "step": 3790 }, { "epoch": 0.3, - "learning_rate": 0.00017571460647933446, - "loss": 1.0549, + "learning_rate": 0.0001757824220355994, + "loss": 5.5648, "step": 3795 }, { "epoch": 0.3, - "learning_rate": 0.00017562320857795034, - "loss": 0.9772, + "learning_rate": 0.00017569123153893647, + "loss": 5.6156, "step": 3800 }, { "epoch": 0.3, - "learning_rate": 0.00017553166287762525, - "loss": 0.9786, + "learning_rate": 0.0001755998934258634, + "loss": 5.6007, "step": 3805 }, { - "epoch": 0.31, - "learning_rate": 0.00017543996955727726, - "loss": 0.9888, + "epoch": 0.3, + "learning_rate": 0.0001755084078745118, + "loss": 5.5392, "step": 3810 }, { "epoch": 0.31, - "learning_rate": 0.00017534812879611296, - "loss": 0.9966, + "learning_rate": 0.0001754167750633009, + "loss": 5.5514, "step": 3815 }, { "epoch": 0.31, - "learning_rate": 0.00017525614077362707, - "loss": 1.0129, + "learning_rate": 0.0001753249951709371, + "loss": 5.5463, "step": 3820 }, { "epoch": 0.31, - "learning_rate": 0.0001751640056696021, - "loss": 1.0091, + "learning_rate": 0.0001752330683764136, + "loss": 5.5157, "step": 3825 }, { "epoch": 0.31, - "learning_rate": 0.00017507172366410808, - "loss": 1.0363, + "learning_rate": 0.00017514099485901012, + "loss": 5.5428, "step": 3830 }, { "epoch": 0.31, - "learning_rate": 0.0001749792949375021, - "loss": 1.0161, + "learning_rate": 0.00017504877479829258, + "loss": 5.6118, "step": 3835 }, { "epoch": 0.31, - "learning_rate": 0.00017488671967042798, - "loss": 1.0378, + "learning_rate": 0.0001749564083741126, + "loss": 5.5575, "step": 3840 }, { "epoch": 0.31, - "learning_rate": 0.00017479399804381604, - "loss": 1.0393, + "learning_rate": 0.00017486389576660733, + "loss": 5.5572, "step": 3845 }, { "epoch": 0.31, - "learning_rate": 0.00017470113023888254, - "loss": 1.0281, + "learning_rate": 0.00017477123715619897, + "loss": 5.6342, "step": 3850 }, { "epoch": 0.31, - "learning_rate": 0.00017460811643712948, - "loss": 1.0123, + "learning_rate": 0.00017467843272359448, + "loss": 5.6189, "step": 3855 }, { "epoch": 0.31, - "learning_rate": 0.0001745149568203442, - "loss": 0.9649, + "learning_rate": 0.00017458548264978515, + "loss": 5.7981, "step": 3860 }, { "epoch": 0.31, - "learning_rate": 0.00017442165157059898, - "loss": 0.9769, + "learning_rate": 0.00017449238711604638, + "loss": 5.6369, "step": 3865 }, { "epoch": 0.31, - "learning_rate": 0.00017432820087025084, - "loss": 1.0052, + "learning_rate": 0.00017439914630393725, + "loss": 5.5643, "step": 3870 }, { "epoch": 0.31, - "learning_rate": 0.00017423460490194093, - "loss": 1.0095, + "learning_rate": 0.00017430576039530012, + "loss": 5.5658, "step": 3875 }, { "epoch": 0.31, - "learning_rate": 0.0001741408638485944, - "loss": 1.0797, + "learning_rate": 0.00017421222957226032, + "loss": 5.6123, "step": 3880 }, { "epoch": 0.31, - "learning_rate": 0.0001740469778934199, - "loss": 0.9537, + "learning_rate": 0.00017411855401722588, + "loss": 5.5785, "step": 3885 }, { "epoch": 0.31, - "learning_rate": 0.00017395294721990934, - "loss": 0.9578, + "learning_rate": 0.000174024733912887, + "loss": 5.5507, "step": 3890 }, { "epoch": 0.31, - "learning_rate": 0.00017385877201183747, - "loss": 0.9707, + "learning_rate": 0.00017393076944221584, + "loss": 5.5507, "step": 3895 }, { "epoch": 0.31, - "learning_rate": 0.00017376445245326143, - "loss": 1.0634, + "learning_rate": 0.0001738366607884661, + "loss": 5.5862, "step": 3900 }, { "epoch": 0.31, - "learning_rate": 0.00017366998872852057, - "loss": 1.0357, + "learning_rate": 0.00017374240813517266, + "loss": 5.5678, "step": 3905 }, { "epoch": 0.31, - "learning_rate": 0.00017357538102223598, - "loss": 0.9635, + "learning_rate": 0.00017364801166615124, + "loss": 5.6299, "step": 3910 }, { "epoch": 0.31, - "learning_rate": 0.0001734806295193101, - "loss": 0.9868, + "learning_rate": 0.000173553471565498, + "loss": 5.578, "step": 3915 }, { "epoch": 0.31, - "learning_rate": 0.00017338573440492651, - "loss": 1.0129, + "learning_rate": 0.00017345878801758934, + "loss": 5.5303, "step": 3920 }, { "epoch": 0.31, - "learning_rate": 0.00017329069586454935, - "loss": 0.9904, + "learning_rate": 0.0001733639612070813, + "loss": 5.5605, "step": 3925 }, { "epoch": 0.31, - "learning_rate": 0.00017319551408392315, - "loss": 0.9908, + "learning_rate": 0.00017326899131890925, + "loss": 5.546, "step": 3930 }, { - "epoch": 0.32, - "learning_rate": 0.00017310018924907234, - "loss": 1.0435, + "epoch": 0.31, + "learning_rate": 0.00017317387853828784, + "loss": 5.5608, "step": 3935 }, { "epoch": 0.32, - "learning_rate": 0.00017300472154630102, - "loss": 1.0581, + "learning_rate": 0.00017307862305071018, + "loss": 5.5641, "step": 3940 }, { "epoch": 0.32, - "learning_rate": 0.00017290911116219241, - "loss": 1.0172, + "learning_rate": 0.00017298322504194778, + "loss": 5.5974, "step": 3945 }, { "epoch": 0.32, - "learning_rate": 0.00017281335828360866, - "loss": 1.0003, + "learning_rate": 0.00017288768469805006, + "loss": 5.5818, "step": 3950 }, { "epoch": 0.32, - "learning_rate": 0.00017271746309769037, - "loss": 1.0246, + "learning_rate": 0.00017279200220534407, + "loss": 5.5554, "step": 3955 }, { "epoch": 0.32, - "learning_rate": 0.0001726214257918563, - "loss": 1.0509, + "learning_rate": 0.00017269617775043404, + "loss": 5.4566, "step": 3960 }, { "epoch": 0.32, - "learning_rate": 0.00017252524655380293, - "loss": 1.0787, + "learning_rate": 0.00017260021152020114, + "loss": 5.6025, "step": 3965 }, { "epoch": 0.32, - "learning_rate": 0.00017242892557150423, - "loss": 0.9943, + "learning_rate": 0.00017250410370180295, + "loss": 5.5422, "step": 3970 }, { "epoch": 0.32, - "learning_rate": 0.00017233246303321103, - "loss": 0.9644, + "learning_rate": 0.00017240785448267322, + "loss": 5.5697, "step": 3975 }, { "epoch": 0.32, - "learning_rate": 0.000172235859127451, - "loss": 0.9423, + "learning_rate": 0.0001723114640505215, + "loss": 5.5317, "step": 3980 }, { "epoch": 0.32, - "learning_rate": 0.000172139114043028, - "loss": 1.0108, + "learning_rate": 0.00017221493259333265, + "loss": 5.515, "step": 3985 }, { "epoch": 0.32, - "learning_rate": 0.0001720422279690218, - "loss": 1.0219, + "learning_rate": 0.00017211826029936664, + "loss": 5.4482, "step": 3990 }, { "epoch": 0.32, - "learning_rate": 0.00017194520109478778, - "loss": 1.0224, + "learning_rate": 0.00017202144735715806, + "loss": 5.4939, "step": 3995 }, { "epoch": 0.32, - "learning_rate": 0.00017184803360995642, - "loss": 0.9954, + "learning_rate": 0.0001719244939555159, + "loss": 5.548, "step": 4000 }, { "epoch": 0.32, - "learning_rate": 0.00017175072570443312, - "loss": 0.9735, + "learning_rate": 0.0001718274002835229, + "loss": 5.5757, "step": 4005 }, { "epoch": 0.32, - "learning_rate": 0.00017165327756839765, - "loss": 1.011, + "learning_rate": 0.00017173016653053552, + "loss": 5.5066, "step": 4010 }, { "epoch": 0.32, - "learning_rate": 0.00017155568939230382, - "loss": 0.9807, + "learning_rate": 0.0001716327928861833, + "loss": 5.5814, "step": 4015 }, { "epoch": 0.32, - "learning_rate": 0.00017145796136687925, - "loss": 0.9776, + "learning_rate": 0.00017153527954036875, + "loss": 5.4646, "step": 4020 }, { "epoch": 0.32, - "learning_rate": 0.00017136009368312475, - "loss": 1.0254, + "learning_rate": 0.00017143762668326667, + "loss": 5.4792, "step": 4025 }, { "epoch": 0.32, - "learning_rate": 0.00017126208653231415, - "loss": 1.0374, + "learning_rate": 0.000171339834505324, + "loss": 5.5255, "step": 4030 }, { "epoch": 0.32, - "learning_rate": 0.00017116394010599387, - "loss": 0.9962, + "learning_rate": 0.00017124190319725945, + "loss": 5.5072, "step": 4035 }, { "epoch": 0.32, - "learning_rate": 0.00017106565459598248, - "loss": 0.9293, + "learning_rate": 0.00017114383295006302, + "loss": 5.5271, "step": 4040 }, { "epoch": 0.32, - "learning_rate": 0.00017096723019437043, - "loss": 0.9953, + "learning_rate": 0.00017104562395499562, + "loss": 5.5096, "step": 4045 }, { "epoch": 0.32, - "learning_rate": 0.0001708686670935196, - "loss": 1.0217, + "learning_rate": 0.00017094727640358883, + "loss": 5.4995, "step": 4050 }, { "epoch": 0.32, - "learning_rate": 0.00017076996548606293, - "loss": 0.9944, + "learning_rate": 0.00017084879048764445, + "loss": 5.5561, "step": 4055 }, { - "epoch": 0.33, - "learning_rate": 0.00017067112556490405, - "loss": 0.9905, + "epoch": 0.32, + "learning_rate": 0.00017075016639923405, + "loss": 5.5142, "step": 4060 }, { "epoch": 0.33, - "learning_rate": 0.00017057214752321694, - "loss": 0.9977, + "learning_rate": 0.00017065140433069874, + "loss": 5.5404, "step": 4065 }, { "epoch": 0.33, - "learning_rate": 0.00017047303155444556, - "loss": 1.0112, + "learning_rate": 0.00017055250447464872, + "loss": 5.4158, "step": 4070 }, { "epoch": 0.33, - "learning_rate": 0.00017037377785230335, - "loss": 0.9853, + "learning_rate": 0.00017045346702396287, + "loss": 5.5576, "step": 4075 }, { "epoch": 0.33, - "learning_rate": 0.000170274386610773, - "loss": 0.995, + "learning_rate": 0.00017035429217178847, + "loss": 5.4774, "step": 4080 }, { "epoch": 0.33, - "learning_rate": 0.000170174858024106, - "loss": 1.0208, + "learning_rate": 0.00017025498011154067, + "loss": 5.554, "step": 4085 }, { "epoch": 0.33, - "learning_rate": 0.00017007519228682223, - "loss": 0.9802, + "learning_rate": 0.00017015553103690237, + "loss": 5.4991, "step": 4090 }, { "epoch": 0.33, - "learning_rate": 0.0001699753895937097, - "loss": 1.0323, + "learning_rate": 0.0001700559451418235, + "loss": 5.5104, "step": 4095 }, { "epoch": 0.33, - "learning_rate": 0.00016987545013982396, - "loss": 1.009, + "learning_rate": 0.00016995622262052092, + "loss": 5.5102, "step": 4100 }, { "epoch": 0.33, - "learning_rate": 0.00016977537412048797, - "loss": 0.9981, + "learning_rate": 0.00016985636366747796, + "loss": 5.4761, "step": 4105 }, { "epoch": 0.33, - "learning_rate": 0.0001696751617312916, - "loss": 0.9993, + "learning_rate": 0.00016975636847744397, + "loss": 5.5412, "step": 4110 }, { "epoch": 0.33, - "learning_rate": 0.00016957481316809108, - "loss": 1.0442, + "learning_rate": 0.00016965623724543404, + "loss": 5.473, "step": 4115 }, { "epoch": 0.33, - "learning_rate": 0.00016947432862700896, - "loss": 1.026, + "learning_rate": 0.00016955597016672855, + "loss": 5.4697, "step": 4120 }, { "epoch": 0.33, - "learning_rate": 0.00016937370830443348, - "loss": 0.9961, + "learning_rate": 0.00016945556743687282, + "loss": 5.5064, "step": 4125 }, { "epoch": 0.33, - "learning_rate": 0.00016927295239701823, - "loss": 0.9875, + "learning_rate": 0.00016935502925167673, + "loss": 5.4392, "step": 4130 }, { "epoch": 0.33, - "learning_rate": 0.00016917206110168183, - "loss": 0.9894, + "learning_rate": 0.00016925435580721434, + "loss": 5.4873, "step": 4135 }, { "epoch": 0.33, - "learning_rate": 0.00016907103461560747, - "loss": 0.9365, + "learning_rate": 0.00016915354729982345, + "loss": 5.4398, "step": 4140 }, { "epoch": 0.33, - "learning_rate": 0.00016896987313624258, - "loss": 1.0496, + "learning_rate": 0.00016905260392610533, + "loss": 5.4426, "step": 4145 }, { "epoch": 0.33, - "learning_rate": 0.0001688685768612984, - "loss": 1.032, + "learning_rate": 0.00016895152588292418, + "loss": 5.4588, "step": 4150 }, { "epoch": 0.33, - "learning_rate": 0.00016876714598874969, - "loss": 0.9929, + "learning_rate": 0.00016885031336740699, + "loss": 5.5096, "step": 4155 }, { "epoch": 0.33, - "learning_rate": 0.00016866558071683413, - "loss": 1.0067, + "learning_rate": 0.0001687489665769429, + "loss": 5.4566, "step": 4160 }, { "epoch": 0.33, - "learning_rate": 0.0001685638812440522, - "loss": 1.0791, + "learning_rate": 0.00016864748570918282, + "loss": 5.4565, "step": 4165 }, { "epoch": 0.33, - "learning_rate": 0.0001684620477691666, - "loss": 1.0087, + "learning_rate": 0.00016854587096203942, + "loss": 5.5127, "step": 4170 }, { "epoch": 0.33, - "learning_rate": 0.00016836008049120197, - "loss": 1.0366, + "learning_rate": 0.00016844412253368623, + "loss": 5.5335, "step": 4175 }, { "epoch": 0.33, - "learning_rate": 0.00016825797960944437, - "loss": 0.9811, + "learning_rate": 0.00016834224062255753, + "loss": 5.5111, "step": 4180 }, { - "epoch": 0.34, - "learning_rate": 0.00016815574532344109, - "loss": 0.9584, + "epoch": 0.33, + "learning_rate": 0.00016824022542734804, + "loss": 5.4848, "step": 4185 }, { "epoch": 0.34, - "learning_rate": 0.00016805337783300006, - "loss": 1.0246, + "learning_rate": 0.00016813807714701224, + "loss": 5.5212, "step": 4190 }, { "epoch": 0.34, - "learning_rate": 0.0001679508773381896, - "loss": 0.9929, + "learning_rate": 0.00016803579598076432, + "loss": 5.4978, "step": 4195 }, { "epoch": 0.34, - "learning_rate": 0.00016784824403933796, - "loss": 0.9696, + "learning_rate": 0.0001679333821280776, + "loss": 5.5562, "step": 4200 }, { "epoch": 0.34, - "learning_rate": 0.0001677454781370329, - "loss": 0.998, + "learning_rate": 0.00016783083578868403, + "loss": 5.4776, "step": 4205 }, { "epoch": 0.34, - "learning_rate": 0.00016764257983212146, - "loss": 0.9934, + "learning_rate": 0.00016772815716257412, + "loss": 5.4396, "step": 4210 }, { "epoch": 0.34, - "learning_rate": 0.00016753954932570927, - "loss": 1.0178, + "learning_rate": 0.0001676253464499963, + "loss": 5.4348, "step": 4215 }, { "epoch": 0.34, - "learning_rate": 0.00016743638681916044, - "loss": 1.0223, + "learning_rate": 0.00016752240385145653, + "loss": 5.4475, "step": 4220 }, { "epoch": 0.34, - "learning_rate": 0.00016733309251409714, - "loss": 1.0069, + "learning_rate": 0.00016741932956771815, + "loss": 5.4649, "step": 4225 }, { "epoch": 0.34, - "learning_rate": 0.000167229666612399, - "loss": 1.0343, + "learning_rate": 0.00016731612379980118, + "loss": 5.4329, "step": 4230 }, { "epoch": 0.34, - "learning_rate": 0.00016712610931620293, - "loss": 0.9725, + "learning_rate": 0.00016721278674898207, + "loss": 5.5127, "step": 4235 }, { "epoch": 0.34, - "learning_rate": 0.0001670224208279025, - "loss": 0.9721, + "learning_rate": 0.00016710931861679335, + "loss": 5.4114, "step": 4240 }, { "epoch": 0.34, - "learning_rate": 0.00016691860135014788, - "loss": 1.0202, + "learning_rate": 0.0001670057196050232, + "loss": 5.4805, "step": 4245 }, { "epoch": 0.34, - "learning_rate": 0.00016681465108584512, - "loss": 1.055, + "learning_rate": 0.00016690198991571505, + "loss": 5.4781, "step": 4250 }, { "epoch": 0.34, - "learning_rate": 0.00016671057023815586, - "loss": 1.0144, + "learning_rate": 0.0001667981297511671, + "loss": 5.895, "step": 4255 }, { "epoch": 0.34, - "learning_rate": 0.0001666063590104971, - "loss": 1.0327, + "learning_rate": 0.00016669413931393208, + "loss": 7.1367, "step": 4260 }, { "epoch": 0.34, - "learning_rate": 0.0001665020176065405, - "loss": 0.9623, + "learning_rate": 0.0001665900188068168, + "loss": 7.4144, "step": 4265 }, { "epoch": 0.34, - "learning_rate": 0.00016639754623021225, - "loss": 1.0006, + "learning_rate": 0.0001664857684328817, + "loss": 6.708, "step": 4270 }, { "epoch": 0.34, - "learning_rate": 0.00016629294508569244, - "loss": 0.963, + "learning_rate": 0.00016638138839544046, + "loss": 6.4751, "step": 4275 }, { "epoch": 0.34, - "learning_rate": 0.0001661882143774149, - "loss": 1.0615, + "learning_rate": 0.00016627687889805975, + "loss": 6.4466, "step": 4280 }, { "epoch": 0.34, - "learning_rate": 0.0001660833543100666, - "loss": 0.993, + "learning_rate": 0.00016617224014455858, + "loss": 6.3847, "step": 4285 }, { "epoch": 0.34, - "learning_rate": 0.0001659783650885874, - "loss": 1.0048, + "learning_rate": 0.00016606747233900815, + "loss": 6.216, "step": 4290 }, { "epoch": 0.34, - "learning_rate": 0.0001658732469181695, - "loss": 1.0407, + "learning_rate": 0.00016596257568573128, + "loss": 6.1459, "step": 4295 }, { "epoch": 0.34, - "learning_rate": 0.00016576800000425718, - "loss": 0.9992, + "learning_rate": 0.0001658575503893021, + "loss": 6.0303, "step": 4300 }, { "epoch": 0.34, - "learning_rate": 0.00016566262455254633, - "loss": 0.9846, + "learning_rate": 0.00016575239665454563, + "loss": 5.9808, "step": 4305 }, { - "epoch": 0.35, - "learning_rate": 0.00016555712076898407, - "loss": 0.9688, + "epoch": 0.34, + "learning_rate": 0.00016564711468653733, + "loss": 5.9701, "step": 4310 }, { "epoch": 0.35, - "learning_rate": 0.00016545148885976823, - "loss": 0.9917, + "learning_rate": 0.00016554170469060283, + "loss": 5.8482, "step": 4315 }, { "epoch": 0.35, - "learning_rate": 0.0001653457290313472, - "loss": 0.9855, + "learning_rate": 0.00016543616687231735, + "loss": 5.8348, "step": 4320 }, { "epoch": 0.35, - "learning_rate": 0.00016523984149041928, - "loss": 0.9546, + "learning_rate": 0.00016533050143750547, + "loss": 5.811, "step": 4325 }, { "epoch": 0.35, - "learning_rate": 0.0001651338264439324, - "loss": 0.9517, + "learning_rate": 0.00016522470859224063, + "loss": 5.776, "step": 4330 }, { "epoch": 0.35, - "learning_rate": 0.00016502768409908363, - "loss": 0.9826, + "learning_rate": 0.00016511878854284472, + "loss": 5.7244, "step": 4335 }, { "epoch": 0.35, - "learning_rate": 0.00016492141466331904, - "loss": 1.0067, + "learning_rate": 0.0001650127414958878, + "loss": 5.7335, "step": 4340 }, { "epoch": 0.35, - "learning_rate": 0.0001648150183443328, - "loss": 1.0332, + "learning_rate": 0.00016490656765818745, + "loss": 5.728, "step": 4345 }, { "epoch": 0.35, - "learning_rate": 0.00016470849535006726, - "loss": 0.9814, + "learning_rate": 0.0001648002672368087, + "loss": 5.6689, "step": 4350 }, { "epoch": 0.35, - "learning_rate": 0.00016460184588871227, - "loss": 0.9824, + "learning_rate": 0.0001646938404390633, + "loss": 5.7441, "step": 4355 }, { "epoch": 0.35, - "learning_rate": 0.00016449507016870487, - "loss": 1.0074, + "learning_rate": 0.00016458728747250958, + "loss": 5.6825, "step": 4360 }, { "epoch": 0.35, - "learning_rate": 0.00016438816839872887, - "loss": 1.0079, + "learning_rate": 0.00016448060854495187, + "loss": 5.6087, "step": 4365 }, { "epoch": 0.35, - "learning_rate": 0.00016428114078771442, - "loss": 1.0258, + "learning_rate": 0.00016437380386444015, + "loss": 5.6414, "step": 4370 }, { "epoch": 0.35, - "learning_rate": 0.0001641739875448376, - "loss": 0.9722, + "learning_rate": 0.0001642668736392696, + "loss": 5.619, "step": 4375 }, { "epoch": 0.35, - "learning_rate": 0.00016406670887952004, - "loss": 0.9941, + "learning_rate": 0.00016415981807798048, + "loss": 5.5727, "step": 4380 }, { "epoch": 0.35, - "learning_rate": 0.00016395930500142853, - "loss": 1.0485, + "learning_rate": 0.00016405263738935718, + "loss": 5.5448, "step": 4385 }, { "epoch": 0.35, - "learning_rate": 0.00016385177612047452, - "loss": 1.0775, + "learning_rate": 0.00016394533178242825, + "loss": 5.5716, "step": 4390 }, { "epoch": 0.35, - "learning_rate": 0.0001637441224468138, - "loss": 0.9927, + "learning_rate": 0.00016383790146646588, + "loss": 5.6122, "step": 4395 }, { "epoch": 0.35, - "learning_rate": 0.00016363634419084603, - "loss": 1.0045, + "learning_rate": 0.0001637303466509855, + "loss": 5.5691, "step": 4400 }, { "epoch": 0.35, - "learning_rate": 0.00016352844156321443, - "loss": 1.0311, + "learning_rate": 0.00016362266754574528, + "loss": 5.6606, "step": 4405 }, { "epoch": 0.35, - "learning_rate": 0.00016342041477480522, - "loss": 0.9837, + "learning_rate": 0.00016351486436074579, + "loss": 5.7074, "step": 4410 }, { "epoch": 0.35, - "learning_rate": 0.00016331226403674726, - "loss": 1.0533, + "learning_rate": 0.0001634069373062296, + "loss": 5.6063, "step": 4415 }, { "epoch": 0.35, - "learning_rate": 0.00016320398956041175, - "loss": 1.0454, + "learning_rate": 0.00016329888659268087, + "loss": 5.5416, "step": 4420 }, { "epoch": 0.35, - "learning_rate": 0.00016309559155741165, - "loss": 1.0052, + "learning_rate": 0.00016319071243082485, + "loss": 5.6072, "step": 4425 }, { "epoch": 0.35, - "learning_rate": 0.00016298707023960137, - "loss": 0.968, + "learning_rate": 0.00016308241503162774, + "loss": 5.5678, "step": 4430 }, { - "epoch": 0.36, - "learning_rate": 0.0001628784258190763, - "loss": 1.0541, + "epoch": 0.35, + "learning_rate": 0.00016297399460629586, + "loss": 5.5458, "step": 4435 }, { "epoch": 0.36, - "learning_rate": 0.00016276965850817252, - "loss": 1.0159, + "learning_rate": 0.0001628654513662755, + "loss": 5.636, "step": 4440 }, { "epoch": 0.36, - "learning_rate": 0.00016266076851946616, - "loss": 0.9273, + "learning_rate": 0.00016275678552325266, + "loss": 5.5018, "step": 4445 }, { "epoch": 0.36, - "learning_rate": 0.00016255175606577318, - "loss": 0.9648, + "learning_rate": 0.0001626479972891522, + "loss": 5.5813, "step": 4450 }, { "epoch": 0.36, - "learning_rate": 0.00016244262136014888, - "loss": 1.0197, + "learning_rate": 0.00016253908687613783, + "loss": 5.5326, "step": 4455 }, { "epoch": 0.36, - "learning_rate": 0.0001623333646158875, - "loss": 1.0001, + "learning_rate": 0.00016243005449661144, + "loss": 5.5519, "step": 4460 }, { "epoch": 0.36, - "learning_rate": 0.00016222398604652177, - "loss": 0.9938, + "learning_rate": 0.0001623209003632129, + "loss": 5.5312, "step": 4465 }, { "epoch": 0.36, - "learning_rate": 0.00016211448586582255, - "loss": 1.0316, + "learning_rate": 0.00016221162468881936, + "loss": 5.5857, "step": 4470 }, { "epoch": 0.36, - "learning_rate": 0.00016200486428779833, - "loss": 1.0368, + "learning_rate": 0.0001621022276865452, + "loss": 5.5573, "step": 4475 }, { "epoch": 0.36, - "learning_rate": 0.00016189512152669488, - "loss": 0.9747, + "learning_rate": 0.00016199270956974128, + "loss": 5.6744, "step": 4480 }, { "epoch": 0.36, - "learning_rate": 0.00016178525779699486, - "loss": 1.0056, + "learning_rate": 0.00016188307055199473, + "loss": 5.5639, "step": 4485 }, { "epoch": 0.36, - "learning_rate": 0.00016167527331341724, - "loss": 1.017, + "learning_rate": 0.00016177331084712843, + "loss": 5.4785, "step": 4490 }, { "epoch": 0.36, - "learning_rate": 0.00016156516829091716, - "loss": 1.0333, + "learning_rate": 0.00016166343066920065, + "loss": 5.5764, "step": 4495 }, { "epoch": 0.36, - "learning_rate": 0.0001614549429446852, - "loss": 0.9957, + "learning_rate": 0.0001615534302325046, + "loss": 5.5083, "step": 4500 }, { "epoch": 0.36, - "learning_rate": 0.00016134459749014713, - "loss": 1.0233, + "learning_rate": 0.00016144330975156806, + "loss": 5.5439, "step": 4505 }, { "epoch": 0.36, - "learning_rate": 0.00016123413214296352, - "loss": 0.9988, + "learning_rate": 0.00016133306944115283, + "loss": 5.5313, "step": 4510 }, { "epoch": 0.36, - "learning_rate": 0.00016112354711902917, - "loss": 1.0412, + "learning_rate": 0.0001612227095162545, + "loss": 5.4739, "step": 4515 }, { "epoch": 0.36, - "learning_rate": 0.0001610128426344729, - "loss": 0.982, + "learning_rate": 0.00016111223019210196, + "loss": 5.516, "step": 4520 }, { "epoch": 0.36, - "learning_rate": 0.0001609020189056569, - "loss": 1.0163, + "learning_rate": 0.00016100163168415686, + "loss": 5.5884, "step": 4525 }, { "epoch": 0.36, - "learning_rate": 0.0001607910761491764, - "loss": 1.0435, + "learning_rate": 0.0001608909142081133, + "loss": 5.5522, "step": 4530 }, { "epoch": 0.36, - "learning_rate": 0.00016068001458185936, - "loss": 0.9794, + "learning_rate": 0.0001607800779798975, + "loss": 5.4901, "step": 4535 }, { "epoch": 0.36, - "learning_rate": 0.00016056883442076587, - "loss": 0.9857, + "learning_rate": 0.00016066912321566722, + "loss": 5.5362, "step": 4540 }, { "epoch": 0.36, - "learning_rate": 0.00016045753588318784, - "loss": 0.9271, + "learning_rate": 0.00016055805013181132, + "loss": 5.509, "step": 4545 }, { "epoch": 0.36, - "learning_rate": 0.0001603461191866485, - "loss": 1.0329, + "learning_rate": 0.00016044685894494947, + "loss": 5.4786, "step": 4550 }, { "epoch": 0.36, - "learning_rate": 0.00016023458454890202, - "loss": 0.9991, + "learning_rate": 0.0001603355498719317, + "loss": 5.5791, "step": 4555 }, { - "epoch": 0.37, - "learning_rate": 0.00016012293218793314, - "loss": 0.9671, + "epoch": 0.36, + "learning_rate": 0.00016022412312983797, + "loss": 5.4684, "step": 4560 }, { "epoch": 0.37, - "learning_rate": 0.0001600111623219566, - "loss": 0.9644, + "learning_rate": 0.0001601125789359776, + "loss": 5.5133, "step": 4565 }, { "epoch": 0.37, - "learning_rate": 0.00015989927516941673, - "loss": 1.0097, + "learning_rate": 0.00016000091750788907, + "loss": 5.4883, "step": 4570 }, { "epoch": 0.37, - "learning_rate": 0.00015978727094898733, - "loss": 1.0131, + "learning_rate": 0.00015988913906333946, + "loss": 5.4968, "step": 4575 }, { "epoch": 0.37, - "learning_rate": 0.00015967514987957074, - "loss": 1.0398, + "learning_rate": 0.0001597772438203241, + "loss": 5.4573, "step": 4580 }, { "epoch": 0.37, - "learning_rate": 0.00015956291218029777, - "loss": 0.9566, + "learning_rate": 0.00015966523199706603, + "loss": 5.5303, "step": 4585 }, { "epoch": 0.37, - "learning_rate": 0.00015945055807052724, - "loss": 1.0645, + "learning_rate": 0.00015955310381201576, + "loss": 5.5067, "step": 4590 }, { "epoch": 0.37, - "learning_rate": 0.00015933808776984537, - "loss": 1.0105, + "learning_rate": 0.00015944085948385058, + "loss": 5.5228, "step": 4595 }, { "epoch": 0.37, - "learning_rate": 0.00015922550149806558, - "loss": 1.0144, + "learning_rate": 0.00015932849923147444, + "loss": 5.5218, "step": 4600 }, { "epoch": 0.37, - "learning_rate": 0.00015911279947522776, - "loss": 0.9636, + "learning_rate": 0.00015921602327401732, + "loss": 5.4973, "step": 4605 }, { "epoch": 0.37, - "learning_rate": 0.00015899998192159828, - "loss": 1.0263, + "learning_rate": 0.00015910343183083477, + "loss": 5.5073, "step": 4610 }, { "epoch": 0.37, - "learning_rate": 0.00015888704905766909, - "loss": 1.0258, + "learning_rate": 0.00015899072512150773, + "loss": 5.5194, "step": 4615 }, { "epoch": 0.37, - "learning_rate": 0.00015877400110415763, - "loss": 1.0169, + "learning_rate": 0.00015887790336584177, + "loss": 5.5069, "step": 4620 }, { "epoch": 0.37, - "learning_rate": 0.00015866083828200623, - "loss": 0.9802, + "learning_rate": 0.00015876496678386692, + "loss": 5.5661, "step": 4625 }, { "epoch": 0.37, - "learning_rate": 0.00015854756081238168, - "loss": 0.9569, + "learning_rate": 0.00015865191559583715, + "loss": 5.546, "step": 4630 }, { "epoch": 0.37, - "learning_rate": 0.00015843416891667494, - "loss": 0.9837, + "learning_rate": 0.0001585387500222299, + "loss": 5.5694, "step": 4635 }, { "epoch": 0.37, - "learning_rate": 0.00015832066281650043, - "loss": 0.9745, + "learning_rate": 0.00015842547028374572, + "loss": 5.4905, "step": 4640 }, { "epoch": 0.37, - "learning_rate": 0.000158207042733696, - "loss": 0.9979, + "learning_rate": 0.00015831207660130781, + "loss": 5.4915, "step": 4645 }, { "epoch": 0.37, - "learning_rate": 0.0001580933088903221, - "loss": 1.0243, + "learning_rate": 0.00015819856919606153, + "loss": 5.4949, "step": 4650 }, { "epoch": 0.37, - "learning_rate": 0.0001579794615086616, - "loss": 1.021, + "learning_rate": 0.00015808494828937415, + "loss": 5.4745, "step": 4655 }, { "epoch": 0.37, - "learning_rate": 0.00015786550081121922, - "loss": 1.0081, + "learning_rate": 0.00015797121410283414, + "loss": 5.5438, "step": 4660 }, { "epoch": 0.37, - "learning_rate": 0.0001577514270207212, - "loss": 0.9748, + "learning_rate": 0.00015785736685825107, + "loss": 5.4666, "step": 4665 }, { "epoch": 0.37, - "learning_rate": 0.0001576372403601147, - "loss": 0.9646, + "learning_rate": 0.0001577434067776548, + "loss": 5.5384, "step": 4670 }, { "epoch": 0.37, - "learning_rate": 0.00015752294105256763, - "loss": 0.9974, + "learning_rate": 0.00015762933408329548, + "loss": 5.533, "step": 4675 }, { "epoch": 0.37, - "learning_rate": 0.00015740852932146794, - "loss": 1.0306, + "learning_rate": 0.00015751514899764265, + "loss": 5.3689, "step": 4680 }, { - "epoch": 0.38, - "learning_rate": 0.0001572940053904234, - "loss": 1.0365, + "epoch": 0.37, + "learning_rate": 0.00015740085174338516, + "loss": 5.494, "step": 4685 }, { "epoch": 0.38, - "learning_rate": 0.0001571793694832609, - "loss": 1.0113, + "learning_rate": 0.0001572864425434307, + "loss": 5.5098, "step": 4690 }, { "epoch": 0.38, - "learning_rate": 0.0001570646218240264, - "loss": 1.0079, + "learning_rate": 0.00015717192162090503, + "loss": 5.505, "step": 4695 }, { "epoch": 0.38, - "learning_rate": 0.00015694976263698409, - "loss": 0.9936, + "learning_rate": 0.0001570572891991521, + "loss": 5.4654, "step": 4700 }, { "epoch": 0.38, - "learning_rate": 0.0001568347921466162, - "loss": 0.9878, + "learning_rate": 0.00015694254550173306, + "loss": 5.4573, "step": 4705 }, { "epoch": 0.38, - "learning_rate": 0.0001567197105776225, - "loss": 1.022, + "learning_rate": 0.00015682769075242624, + "loss": 5.5117, "step": 4710 }, { "epoch": 0.38, - "learning_rate": 0.00015660451815491984, - "loss": 1.0436, + "learning_rate": 0.0001567127251752265, + "loss": 5.4355, "step": 4715 }, { "epoch": 0.38, - "learning_rate": 0.0001564892151036417, - "loss": 1.0354, + "learning_rate": 0.00015659764899434475, + "loss": 5.4581, "step": 4720 }, { "epoch": 0.38, - "learning_rate": 0.00015637380164913787, - "loss": 0.9652, + "learning_rate": 0.00015648246243420773, + "loss": 5.4636, "step": 4725 }, { "epoch": 0.38, - "learning_rate": 0.00015625827801697378, - "loss": 1.0546, + "learning_rate": 0.00015636716571945742, + "loss": 5.4019, "step": 4730 }, { "epoch": 0.38, - "learning_rate": 0.00015614264443293026, - "loss": 1.014, + "learning_rate": 0.00015625175907495057, + "loss": 5.3918, "step": 4735 }, { "epoch": 0.38, - "learning_rate": 0.00015602690112300307, - "loss": 0.9892, + "learning_rate": 0.00015613624272575835, + "loss": 5.4166, "step": 4740 }, { "epoch": 0.38, - "learning_rate": 0.0001559110483134023, - "loss": 0.9775, + "learning_rate": 0.00015602061689716593, + "loss": 5.5377, "step": 4745 }, { "epoch": 0.38, - "learning_rate": 0.0001557950862305522, - "loss": 0.9717, + "learning_rate": 0.00015590488181467193, + "loss": 5.5277, "step": 4750 }, { "epoch": 0.38, - "learning_rate": 0.0001556790151010905, - "loss": 1.001, + "learning_rate": 0.00015578903770398803, + "loss": 5.5025, "step": 4755 }, { "epoch": 0.38, - "learning_rate": 0.00015556283515186803, - "loss": 1.0133, + "learning_rate": 0.00015567308479103864, + "loss": 5.384, "step": 4760 }, { "epoch": 0.38, - "learning_rate": 0.00015544654660994836, - "loss": 1.0339, + "learning_rate": 0.00015555702330196023, + "loss": 5.4573, "step": 4765 }, { "epoch": 0.38, - "learning_rate": 0.0001553301497026072, - "loss": 0.9801, + "learning_rate": 0.0001554408534631011, + "loss": 5.5624, "step": 4770 }, { "epoch": 0.38, - "learning_rate": 0.0001552136446573322, - "loss": 0.9978, + "learning_rate": 0.0001553245755010209, + "loss": 5.4106, "step": 4775 }, { "epoch": 0.38, - "learning_rate": 0.00015509703170182225, - "loss": 1.0421, + "learning_rate": 0.00015520818964249, + "loss": 5.51, "step": 4780 }, { "epoch": 0.38, - "learning_rate": 0.00015498031106398713, - "loss": 0.9788, + "learning_rate": 0.00015509169611448932, + "loss": 5.5034, "step": 4785 }, { "epoch": 0.38, - "learning_rate": 0.00015486348297194714, - "loss": 1.0184, + "learning_rate": 0.00015497509514420977, + "loss": 5.5526, "step": 4790 }, { "epoch": 0.38, - "learning_rate": 0.00015474654765403254, - "loss": 0.9744, + "learning_rate": 0.00015485838695905168, + "loss": 5.3577, "step": 4795 }, { "epoch": 0.38, - "learning_rate": 0.00015462950533878317, - "loss": 0.9709, + "learning_rate": 0.00015474157178662457, + "loss": 5.3723, "step": 4800 }, { "epoch": 0.38, - "learning_rate": 0.00015451235625494803, - "loss": 0.9684, + "learning_rate": 0.00015462464985474666, + "loss": 5.4706, "step": 4805 }, { - "epoch": 0.39, - "learning_rate": 0.0001543951006314847, - "loss": 0.969, + "epoch": 0.38, + "learning_rate": 0.0001545076213914442, + "loss": 5.5238, "step": 4810 }, { "epoch": 0.39, - "learning_rate": 0.00015427773869755904, - "loss": 0.9653, + "learning_rate": 0.0001543904866249514, + "loss": 5.365, "step": 4815 }, { "epoch": 0.39, - "learning_rate": 0.00015416027068254468, - "loss": 1.0054, + "learning_rate": 0.00015427324578370962, + "loss": 5.4876, "step": 4820 }, { "epoch": 0.39, - "learning_rate": 0.00015404269681602257, - "loss": 0.9853, + "learning_rate": 0.00015415589909636727, + "loss": 5.4712, "step": 4825 }, { "epoch": 0.39, - "learning_rate": 0.00015392501732778056, - "loss": 0.9986, + "learning_rate": 0.00015403844679177898, + "loss": 5.443, "step": 4830 }, { "epoch": 0.39, - "learning_rate": 0.0001538072324478129, - "loss": 1.0635, + "learning_rate": 0.00015392088909900554, + "loss": 5.4149, "step": 4835 }, { "epoch": 0.39, - "learning_rate": 0.0001536893424063198, - "loss": 1.016, + "learning_rate": 0.00015380322624731314, + "loss": 5.4528, "step": 4840 }, { "epoch": 0.39, - "learning_rate": 0.00015357134743370705, - "loss": 1.0165, + "learning_rate": 0.00015368545846617315, + "loss": 5.4284, "step": 4845 }, { "epoch": 0.39, - "learning_rate": 0.0001534532477605855, - "loss": 1.0234, + "learning_rate": 0.00015356758598526152, + "loss": 5.4387, "step": 4850 }, { "epoch": 0.39, - "learning_rate": 0.0001533350436177706, - "loss": 0.9688, + "learning_rate": 0.00015344960903445842, + "loss": 5.4294, "step": 4855 }, { "epoch": 0.39, - "learning_rate": 0.000153216735236282, - "loss": 0.9882, + "learning_rate": 0.00015333152784384777, + "loss": 5.4483, "step": 4860 }, { "epoch": 0.39, - "learning_rate": 0.00015309832284734313, - "loss": 1.0189, + "learning_rate": 0.00015321334264371673, + "loss": 5.4291, "step": 4865 }, { "epoch": 0.39, - "learning_rate": 0.0001529798066823806, - "loss": 1.0249, + "learning_rate": 0.00015309505366455537, + "loss": 5.4281, "step": 4870 }, { "epoch": 0.39, - "learning_rate": 0.00015286118697302383, - "loss": 1.0142, + "learning_rate": 0.00015297666113705613, + "loss": 5.4046, "step": 4875 }, { "epoch": 0.39, - "learning_rate": 0.00015274246395110473, - "loss": 1.0177, + "learning_rate": 0.00015285816529211338, + "loss": 5.4775, "step": 4880 }, { "epoch": 0.39, - "learning_rate": 0.000152623637848657, - "loss": 0.9523, + "learning_rate": 0.00015273956636082292, + "loss": 5.492, "step": 4885 }, { "epoch": 0.39, - "learning_rate": 0.00015250470889791588, - "loss": 1.0184, + "learning_rate": 0.00015262086457448182, + "loss": 5.3552, "step": 4890 }, { "epoch": 0.39, - "learning_rate": 0.0001523856773313176, - "loss": 0.9952, + "learning_rate": 0.00015250206016458742, + "loss": 5.449, "step": 4895 }, { "epoch": 0.39, - "learning_rate": 0.00015226654338149892, - "loss": 1.0351, + "learning_rate": 0.00015238315336283753, + "loss": 5.4037, "step": 4900 }, { "epoch": 0.39, - "learning_rate": 0.00015214730728129675, - "loss": 0.9598, + "learning_rate": 0.00015226414440112934, + "loss": 5.3913, "step": 4905 }, { "epoch": 0.39, - "learning_rate": 0.00015202796926374758, - "loss": 1.0372, + "learning_rate": 0.00015214503351155954, + "loss": 5.4684, "step": 4910 }, { "epoch": 0.39, - "learning_rate": 0.0001519085295620871, - "loss": 1.0231, + "learning_rate": 0.00015202582092642343, + "loss": 5.4113, "step": 4915 }, { "epoch": 0.39, - "learning_rate": 0.0001517889884097498, - "loss": 1.01, + "learning_rate": 0.00015190650687821474, + "loss": 5.5322, "step": 4920 }, { "epoch": 0.39, - "learning_rate": 0.00015166934604036837, - "loss": 1.0089, + "learning_rate": 0.000151787091599625, + "loss": 5.4471, "step": 4925 }, { "epoch": 0.39, - "learning_rate": 0.0001515496026877733, - "loss": 0.9865, + "learning_rate": 0.00015166757532354325, + "loss": 5.3653, "step": 4930 }, { - "epoch": 0.4, - "learning_rate": 0.0001514297585859926, - "loss": 0.9889, + "epoch": 0.39, + "learning_rate": 0.00015154795828305542, + "loss": 5.5438, "step": 4935 }, { "epoch": 0.4, - "learning_rate": 0.000151309813969251, - "loss": 1.0596, + "learning_rate": 0.00015142824071144397, + "loss": 5.4497, "step": 4940 }, { "epoch": 0.4, - "learning_rate": 0.00015118976907196974, - "loss": 1.0256, + "learning_rate": 0.00015130842284218752, + "loss": 5.4334, "step": 4945 }, { "epoch": 0.4, - "learning_rate": 0.0001510696241287661, - "loss": 0.9728, + "learning_rate": 0.00015118850490896012, + "loss": 5.4315, "step": 4950 }, { "epoch": 0.4, - "learning_rate": 0.00015094937937445283, - "loss": 0.9761, + "learning_rate": 0.00015106848714563112, + "loss": 5.4054, "step": 4955 }, { "epoch": 0.4, - "learning_rate": 0.00015082903504403782, - "loss": 1.0683, + "learning_rate": 0.00015094836978626447, + "loss": 5.364, "step": 4960 }, { "epoch": 0.4, - "learning_rate": 0.00015070859137272352, - "loss": 1.0431, + "learning_rate": 0.00015082815306511843, + "loss": 5.4506, "step": 4965 }, { "epoch": 0.4, - "learning_rate": 0.00015058804859590646, - "loss": 0.9864, + "learning_rate": 0.00015070783721664495, + "loss": 5.4242, "step": 4970 }, { "epoch": 0.4, - "learning_rate": 0.00015046740694917701, - "loss": 1.0694, + "learning_rate": 0.00015058742247548942, + "loss": 5.4281, "step": 4975 }, { "epoch": 0.4, - "learning_rate": 0.00015034666666831873, - "loss": 1.0428, + "learning_rate": 0.00015046690907648998, + "loss": 5.4475, "step": 4980 }, { "epoch": 0.4, - "learning_rate": 0.0001502258279893079, - "loss": 0.9795, + "learning_rate": 0.0001503462972546773, + "loss": 5.4903, "step": 4985 }, { "epoch": 0.4, - "learning_rate": 0.00015010489114831313, - "loss": 1.0092, + "learning_rate": 0.0001502255872452738, + "loss": 5.4362, "step": 4990 }, { "epoch": 0.4, - "learning_rate": 0.0001499838563816949, - "loss": 1.0292, + "learning_rate": 0.00015010477928369364, + "loss": 5.4018, "step": 4995 }, { "epoch": 0.4, - "learning_rate": 0.0001498627239260051, - "loss": 1.0222, + "learning_rate": 0.0001499838736055418, + "loss": 5.3877, "step": 5000 }, { "epoch": 0.4, - "learning_rate": 0.0001497414940179864, - "loss": 0.9425, + "learning_rate": 0.00014986287044661397, + "loss": 5.4277, "step": 5005 }, { "epoch": 0.4, - "learning_rate": 0.00014962016689457216, - "loss": 1.0742, + "learning_rate": 0.00014974177004289588, + "loss": 5.3957, "step": 5010 }, { "epoch": 0.4, - "learning_rate": 0.0001494987427928856, - "loss": 1.0283, + "learning_rate": 0.00014962057263056292, + "loss": 5.4873, "step": 5015 }, { "epoch": 0.4, - "learning_rate": 0.00014937722195023945, - "loss": 1.0046, + "learning_rate": 0.00014949927844597972, + "loss": 5.4437, "step": 5020 }, { "epoch": 0.4, - "learning_rate": 0.0001492556046041356, - "loss": 0.9679, + "learning_rate": 0.00014937788772569953, + "loss": 5.4393, "step": 5025 }, { "epoch": 0.4, - "learning_rate": 0.00014913389099226445, - "loss": 1.026, + "learning_rate": 0.00014925640070646395, + "loss": 5.3273, "step": 5030 }, { "epoch": 0.4, - "learning_rate": 0.0001490120813525047, - "loss": 1.0046, + "learning_rate": 0.00014913481762520245, + "loss": 5.3709, "step": 5035 }, { "epoch": 0.4, - "learning_rate": 0.00014889017592292252, - "loss": 1.0086, + "learning_rate": 0.00014901313871903174, + "loss": 5.4472, "step": 5040 }, { "epoch": 0.4, - "learning_rate": 0.00014876817494177146, - "loss": 0.9956, + "learning_rate": 0.00014889136422525535, + "loss": 5.3416, "step": 5045 }, { "epoch": 0.4, - "learning_rate": 0.00014864607864749173, - "loss": 0.9801, + "learning_rate": 0.00014876949438136347, + "loss": 5.4655, "step": 5050 }, { "epoch": 0.4, - "learning_rate": 0.00014852388727870988, - "loss": 0.9843, + "learning_rate": 0.00014864752942503196, + "loss": 5.4712, "step": 5055 }, { - "epoch": 0.41, - "learning_rate": 0.00014840160107423825, - "loss": 1.0208, + "epoch": 0.4, + "learning_rate": 0.00014852546959412242, + "loss": 5.5033, "step": 5060 }, { "epoch": 0.41, - "learning_rate": 0.00014827922027307451, - "loss": 0.9924, + "learning_rate": 0.00014840331512668128, + "loss": 5.4312, "step": 5065 }, { "epoch": 0.41, - "learning_rate": 0.00014815674511440115, - "loss": 1.0171, + "learning_rate": 0.00014828106626093966, + "loss": 5.4226, "step": 5070 }, { "epoch": 0.41, - "learning_rate": 0.00014803417583758525, - "loss": 1.0138, + "learning_rate": 0.00014815872323531274, + "loss": 5.4053, "step": 5075 }, { "epoch": 0.41, - "learning_rate": 0.0001479115126821777, - "loss": 1.0214, + "learning_rate": 0.00014803628628839934, + "loss": 5.4914, "step": 5080 }, { "epoch": 0.41, - "learning_rate": 0.00014778875588791288, - "loss": 0.9944, + "learning_rate": 0.0001479137556589814, + "loss": 5.4153, "step": 5085 }, { "epoch": 0.41, - "learning_rate": 0.00014766590569470817, - "loss": 0.9592, + "learning_rate": 0.00014779113158602372, + "loss": 5.3742, "step": 5090 }, { "epoch": 0.41, - "learning_rate": 0.00014754296234266358, - "loss": 1.038, + "learning_rate": 0.0001476684143086731, + "loss": 5.4774, "step": 5095 }, { "epoch": 0.41, - "learning_rate": 0.0001474199260720611, - "loss": 0.9811, + "learning_rate": 0.00014754560406625834, + "loss": 5.3977, "step": 5100 }, { "epoch": 0.41, - "learning_rate": 0.00014729679712336435, - "loss": 0.9937, + "learning_rate": 0.00014742270109828937, + "loss": 5.4002, "step": 5105 }, { "epoch": 0.41, - "learning_rate": 0.00014717357573721808, - "loss": 1.0053, + "learning_rate": 0.00014729970564445705, + "loss": 5.4045, "step": 5110 }, { "epoch": 0.41, - "learning_rate": 0.00014705026215444775, - "loss": 0.9606, + "learning_rate": 0.00014717661794463264, + "loss": 5.3889, "step": 5115 }, { "epoch": 0.41, - "learning_rate": 0.00014692685661605887, - "loss": 0.9838, + "learning_rate": 0.00014705343823886717, + "loss": 5.3967, "step": 5120 }, { "epoch": 0.41, - "learning_rate": 0.00014680335936323682, - "loss": 1.0108, + "learning_rate": 0.00014693016676739127, + "loss": 5.3922, "step": 5125 }, { "epoch": 0.41, - "learning_rate": 0.00014667977063734616, - "loss": 1.0723, + "learning_rate": 0.0001468068037706144, + "loss": 5.3768, "step": 5130 }, { "epoch": 0.41, - "learning_rate": 0.00014655609067993022, - "loss": 0.9676, + "learning_rate": 0.00014668334948912453, + "loss": 5.4459, "step": 5135 }, { "epoch": 0.41, - "learning_rate": 0.0001464323197327107, - "loss": 1.0012, + "learning_rate": 0.0001465598041636878, + "loss": 5.4382, "step": 5140 }, { "epoch": 0.41, - "learning_rate": 0.00014630845803758705, - "loss": 0.9945, + "learning_rate": 0.00014643616803524778, + "loss": 5.4477, "step": 5145 }, { "epoch": 0.41, - "learning_rate": 0.00014618450583663608, - "loss": 1.0325, + "learning_rate": 0.00014631244134492507, + "loss": 5.4493, "step": 5150 }, { "epoch": 0.41, - "learning_rate": 0.00014606046337211154, - "loss": 0.9967, + "learning_rate": 0.00014618862433401708, + "loss": 5.4438, "step": 5155 }, { "epoch": 0.41, - "learning_rate": 0.0001459363308864436, - "loss": 1.0093, + "learning_rate": 0.00014606471724399716, + "loss": 5.3437, "step": 5160 }, { "epoch": 0.41, - "learning_rate": 0.00014581210862223826, - "loss": 0.9462, + "learning_rate": 0.00014594072031651453, + "loss": 5.4167, "step": 5165 }, { "epoch": 0.41, - "learning_rate": 0.00014568779682227714, - "loss": 1.016, + "learning_rate": 0.00014581663379339345, + "loss": 5.3682, "step": 5170 }, { "epoch": 0.41, - "learning_rate": 0.00014556339572951673, - "loss": 1.0178, + "learning_rate": 0.000145692457916633, + "loss": 5.4614, "step": 5175 }, { "epoch": 0.41, - "learning_rate": 0.00014543890558708808, - "loss": 1.0203, + "learning_rate": 0.0001455681929284065, + "loss": 5.421, "step": 5180 }, { - "epoch": 0.42, - "learning_rate": 0.00014531432663829628, - "loss": 0.992, + "epoch": 0.41, + "learning_rate": 0.00014544383907106107, + "loss": 5.3822, "step": 5185 }, { "epoch": 0.42, - "learning_rate": 0.00014518965912661998, - "loss": 0.9987, + "learning_rate": 0.00014531939658711716, + "loss": 5.3511, "step": 5190 }, { "epoch": 0.42, - "learning_rate": 0.0001450649032957109, - "loss": 1.0433, + "learning_rate": 0.00014519486571926802, + "loss": 5.4329, "step": 5195 }, { "epoch": 0.42, - "learning_rate": 0.00014494005938939345, - "loss": 1.0053, + "learning_rate": 0.00014507024671037928, + "loss": 5.4036, "step": 5200 }, { "epoch": 0.42, - "learning_rate": 0.000144815127651664, - "loss": 0.9853, + "learning_rate": 0.00014494553980348848, + "loss": 5.3431, "step": 5205 }, { "epoch": 0.42, - "learning_rate": 0.0001446901083266908, - "loss": 0.9895, + "learning_rate": 0.00014482074524180464, + "loss": 5.3705, "step": 5210 }, { "epoch": 0.42, - "learning_rate": 0.00014456500165881318, - "loss": 0.972, + "learning_rate": 0.0001446958632687076, + "loss": 5.4683, "step": 5215 }, { "epoch": 0.42, - "learning_rate": 0.0001444398078925411, - "loss": 0.9335, + "learning_rate": 0.00014457089412774781, + "loss": 5.4509, "step": 5220 }, { "epoch": 0.42, - "learning_rate": 0.00014431452727255484, - "loss": 0.9802, + "learning_rate": 0.0001444458380626456, + "loss": 5.4021, "step": 5225 }, { "epoch": 0.42, - "learning_rate": 0.00014418916004370442, - "loss": 0.9521, + "learning_rate": 0.00014432069531729094, + "loss": 5.4575, "step": 5230 }, { "epoch": 0.42, - "learning_rate": 0.0001440637064510091, - "loss": 0.9779, + "learning_rate": 0.00014419546613574267, + "loss": 5.2872, "step": 5235 }, { "epoch": 0.42, - "learning_rate": 0.000143938166739657, - "loss": 0.9973, + "learning_rate": 0.00014407015076222846, + "loss": 5.3959, "step": 5240 }, { "epoch": 0.42, - "learning_rate": 0.00014381254115500438, - "loss": 1.0067, + "learning_rate": 0.0001439447494411438, + "loss": 5.3322, "step": 5245 }, { "epoch": 0.42, - "learning_rate": 0.00014368682994257554, - "loss": 1.006, + "learning_rate": 0.000143819262417052, + "loss": 5.4487, "step": 5250 }, { "epoch": 0.42, - "learning_rate": 0.00014356103334806203, - "loss": 0.9761, + "learning_rate": 0.00014369368993468343, + "loss": 5.3905, "step": 5255 }, { "epoch": 0.42, - "learning_rate": 0.00014343515161732227, - "loss": 1.0338, + "learning_rate": 0.0001435680322389352, + "loss": 5.4069, "step": 5260 }, { "epoch": 0.42, - "learning_rate": 0.00014330918499638105, - "loss": 0.998, + "learning_rate": 0.0001434422895748704, + "loss": 5.4842, "step": 5265 }, { "epoch": 0.42, - "learning_rate": 0.00014318313373142913, - "loss": 1.008, + "learning_rate": 0.0001433164621877181, + "loss": 5.3584, "step": 5270 }, { "epoch": 0.42, - "learning_rate": 0.0001430569980688227, - "loss": 0.9531, + "learning_rate": 0.0001431905503228724, + "loss": 5.4929, "step": 5275 }, { "epoch": 0.42, - "learning_rate": 0.00014293077825508284, - "loss": 1.0446, + "learning_rate": 0.00014306455422589224, + "loss": 5.4146, "step": 5280 }, { "epoch": 0.42, - "learning_rate": 0.00014280447453689514, - "loss": 1.0368, + "learning_rate": 0.00014293847414250088, + "loss": 5.4017, "step": 5285 }, { "epoch": 0.42, - "learning_rate": 0.00014267808716110912, - "loss": 1.0033, + "learning_rate": 0.00014281231031858523, + "loss": 5.3153, "step": 5290 }, { "epoch": 0.42, - "learning_rate": 0.0001425516163747379, - "loss": 0.9926, + "learning_rate": 0.00014268606300019566, + "loss": 5.4141, "step": 5295 }, { "epoch": 0.42, - "learning_rate": 0.00014242506242495754, - "loss": 1.0096, + "learning_rate": 0.00014255973243354527, + "loss": 5.3551, "step": 5300 }, { "epoch": 0.42, - "learning_rate": 0.00014229842555910663, - "loss": 1.0367, + "learning_rate": 0.00014243331886500953, + "loss": 5.351, "step": 5305 }, { - "epoch": 0.43, - "learning_rate": 0.00014217170602468586, - "loss": 1.0103, + "epoch": 0.42, + "learning_rate": 0.00014230682254112587, + "loss": 5.4383, "step": 5310 }, { "epoch": 0.43, - "learning_rate": 0.0001420449040693575, - "loss": 1.0218, + "learning_rate": 0.0001421802437085931, + "loss": 5.3891, "step": 5315 }, { "epoch": 0.43, - "learning_rate": 0.0001419180199409448, - "loss": 0.9848, + "learning_rate": 0.00014205358261427074, + "loss": 5.3962, "step": 5320 }, { "epoch": 0.43, - "learning_rate": 0.00014179105388743173, - "loss": 1.0506, + "learning_rate": 0.00014192683950517906, + "loss": 5.3912, "step": 5325 }, { "epoch": 0.43, - "learning_rate": 0.00014166400615696231, - "loss": 1.0214, + "learning_rate": 0.000141800014628498, + "loss": 5.3734, "step": 5330 }, { "epoch": 0.43, - "learning_rate": 0.0001415368769978402, - "loss": 0.9996, + "learning_rate": 0.0001416731082315671, + "loss": 5.3091, "step": 5335 }, { "epoch": 0.43, - "learning_rate": 0.00014140966665852827, - "loss": 0.9758, + "learning_rate": 0.00014154612056188495, + "loss": 5.4136, "step": 5340 }, { "epoch": 0.43, - "learning_rate": 0.0001412823753876479, - "loss": 1.007, + "learning_rate": 0.00014141905186710848, + "loss": 5.3723, "step": 5345 }, { "epoch": 0.43, - "learning_rate": 0.00014115500343397878, - "loss": 1.0157, + "learning_rate": 0.0001412919023950527, + "loss": 5.3466, "step": 5350 }, { "epoch": 0.43, - "learning_rate": 0.00014102755104645825, - "loss": 1.0152, + "learning_rate": 0.0001411646723936902, + "loss": 5.3869, "step": 5355 }, { "epoch": 0.43, - "learning_rate": 0.00014090001847418085, - "loss": 0.9908, + "learning_rate": 0.00014103736211115063, + "loss": 5.4564, "step": 5360 }, { "epoch": 0.43, - "learning_rate": 0.00014077240596639781, - "loss": 1.0849, + "learning_rate": 0.00014090997179572014, + "loss": 5.4436, "step": 5365 }, { "epoch": 0.43, - "learning_rate": 0.00014064471377251661, - "loss": 1.0112, + "learning_rate": 0.00014078250169584098, + "loss": 5.327, "step": 5370 }, { "epoch": 0.43, - "learning_rate": 0.00014051694214210048, - "loss": 1.0109, + "learning_rate": 0.00014065495206011104, + "loss": 5.3565, "step": 5375 }, { "epoch": 0.43, - "learning_rate": 0.00014038909132486788, - "loss": 0.9787, + "learning_rate": 0.0001405273231372833, + "loss": 5.4267, "step": 5380 }, { "epoch": 0.43, - "learning_rate": 0.000140261161570692, - "loss": 1.0286, + "learning_rate": 0.00014039961517626535, + "loss": 5.3469, "step": 5385 }, { "epoch": 0.43, - "learning_rate": 0.0001401331531296004, - "loss": 0.956, + "learning_rate": 0.00014027182842611894, + "loss": 5.4414, "step": 5390 }, { "epoch": 0.43, - "learning_rate": 0.00014000506625177434, - "loss": 1.041, + "learning_rate": 0.0001401439631360595, + "loss": 5.3403, "step": 5395 }, { "epoch": 0.43, - "learning_rate": 0.00013987690118754843, - "loss": 0.9605, + "learning_rate": 0.0001400160195554556, + "loss": 5.3864, "step": 5400 }, { "epoch": 0.43, - "learning_rate": 0.00013974865818741004, - "loss": 0.9804, + "learning_rate": 0.00013988799793382856, + "loss": 5.3898, "step": 5405 }, { "epoch": 0.43, - "learning_rate": 0.00013962033750199892, - "loss": 1.0123, + "learning_rate": 0.00013975989852085176, + "loss": 5.3892, "step": 5410 }, { "epoch": 0.43, - "learning_rate": 0.00013949193938210658, - "loss": 0.9885, + "learning_rate": 0.00013963172156635043, + "loss": 5.3623, "step": 5415 }, { "epoch": 0.43, - "learning_rate": 0.00013936346407867595, - "loss": 0.987, + "learning_rate": 0.00013950346732030094, + "loss": 5.3249, "step": 5420 }, { "epoch": 0.43, - "learning_rate": 0.00013923491184280072, - "loss": 0.9485, + "learning_rate": 0.00013937513603283047, + "loss": 5.3231, "step": 5425 }, { "epoch": 0.43, - "learning_rate": 0.00013910628292572504, - "loss": 0.9973, + "learning_rate": 0.00013924672795421637, + "loss": 5.3815, "step": 5430 }, { - "epoch": 0.44, - "learning_rate": 0.00013897757757884278, - "loss": 0.9824, + "epoch": 0.43, + "learning_rate": 0.00013911824333488586, + "loss": 5.3579, "step": 5435 }, { "epoch": 0.44, - "learning_rate": 0.00013884879605369736, - "loss": 1.0232, + "learning_rate": 0.0001389896824254153, + "loss": 5.4289, "step": 5440 }, { "epoch": 0.44, - "learning_rate": 0.0001387199386019809, - "loss": 0.9945, + "learning_rate": 0.00013886104547652993, + "loss": 5.3696, "step": 5445 }, { "epoch": 0.44, - "learning_rate": 0.00013859100547553412, - "loss": 1.0013, + "learning_rate": 0.00013873233273910328, + "loss": 5.3661, "step": 5450 }, { "epoch": 0.44, - "learning_rate": 0.00013846199692634547, - "loss": 0.9724, + "learning_rate": 0.00013860354446415657, + "loss": 5.3924, "step": 5455 }, { "epoch": 0.44, - "learning_rate": 0.00013833291320655087, - "loss": 0.9754, + "learning_rate": 0.00013847468090285854, + "loss": 5.3544, "step": 5460 }, { "epoch": 0.44, - "learning_rate": 0.00013820375456843312, - "loss": 0.9576, + "learning_rate": 0.00013834574230652458, + "loss": 5.3663, "step": 5465 }, { "epoch": 0.44, - "learning_rate": 0.00013807452126442154, - "loss": 0.9857, + "learning_rate": 0.00013821672892661647, + "loss": 5.3811, "step": 5470 }, { "epoch": 0.44, - "learning_rate": 0.0001379452135470913, - "loss": 0.9624, + "learning_rate": 0.00013808764101474192, + "loss": 5.3665, "step": 5475 }, { "epoch": 0.44, - "learning_rate": 0.00013781583166916293, - "loss": 1.0004, + "learning_rate": 0.00013795847882265385, + "loss": 5.3655, "step": 5480 }, { "epoch": 0.44, - "learning_rate": 0.00013768637588350204, - "loss": 1.0191, + "learning_rate": 0.00013782924260225018, + "loss": 5.4591, "step": 5485 }, { "epoch": 0.44, - "learning_rate": 0.0001375568464431187, - "loss": 1.0177, + "learning_rate": 0.0001376999326055731, + "loss": 5.3819, "step": 5490 }, { "epoch": 0.44, - "learning_rate": 0.00013742724360116672, - "loss": 1.0083, + "learning_rate": 0.0001375705490848087, + "loss": 5.3787, "step": 5495 }, { "epoch": 0.44, - "learning_rate": 0.0001372975676109436, - "loss": 0.9726, + "learning_rate": 0.00013744109229228655, + "loss": 5.351, "step": 5500 }, { "epoch": 0.44, - "learning_rate": 0.0001371678187258897, - "loss": 1.0708, + "learning_rate": 0.00013731156248047904, + "loss": 5.3201, "step": 5505 }, { "epoch": 0.44, - "learning_rate": 0.00013703799719958783, - "loss": 0.9997, + "learning_rate": 0.0001371819599020009, + "loss": 5.3506, "step": 5510 }, { "epoch": 0.44, - "learning_rate": 0.00013690810328576287, - "loss": 1.024, + "learning_rate": 0.00013705228480960898, + "loss": 5.3542, "step": 5515 }, { "epoch": 0.44, - "learning_rate": 0.000136778137238281, - "loss": 1.0024, + "learning_rate": 0.00013692253745620128, + "loss": 5.44, "step": 5520 }, { "epoch": 0.44, - "learning_rate": 0.00013664809931114954, - "loss": 1.0056, + "learning_rate": 0.00013679271809481693, + "loss": 5.3697, "step": 5525 }, { "epoch": 0.44, - "learning_rate": 0.00013651798975851626, - "loss": 1.0071, + "learning_rate": 0.00013666282697863546, + "loss": 5.3792, "step": 5530 }, { "epoch": 0.44, - "learning_rate": 0.00013638780883466883, - "loss": 0.9568, + "learning_rate": 0.00013653286436097627, + "loss": 5.391, "step": 5535 }, { "epoch": 0.44, - "learning_rate": 0.00013625755679403452, - "loss": 1.0651, + "learning_rate": 0.0001364028304952982, + "loss": 5.3607, "step": 5540 }, { "epoch": 0.44, - "learning_rate": 0.0001361272338911795, - "loss": 0.9436, + "learning_rate": 0.0001362727256351992, + "loss": 5.3869, "step": 5545 }, { "epoch": 0.44, - "learning_rate": 0.0001359968403808085, - "loss": 0.9846, + "learning_rate": 0.00013614255003441546, + "loss": 5.319, "step": 5550 }, { "epoch": 0.44, - "learning_rate": 0.00013586637651776424, - "loss": 0.9553, + "learning_rate": 0.0001360123039468213, + "loss": 5.3795, "step": 5555 }, { - "epoch": 0.45, - "learning_rate": 0.00013573584255702686, - "loss": 1.0524, + "epoch": 0.44, + "learning_rate": 0.00013588198762642844, + "loss": 5.3611, "step": 5560 }, { "epoch": 0.45, - "learning_rate": 0.0001356052387537136, - "loss": 0.9889, + "learning_rate": 0.0001357516013273855, + "loss": 5.426, "step": 5565 }, { "epoch": 0.45, - "learning_rate": 0.00013547456536307817, - "loss": 1.0355, + "learning_rate": 0.0001356211453039777, + "loss": 5.3739, "step": 5570 }, { "epoch": 0.45, - "learning_rate": 0.00013534382264051025, - "loss": 0.9987, + "learning_rate": 0.00013549061981062616, + "loss": 5.335, "step": 5575 }, { "epoch": 0.45, - "learning_rate": 0.00013521301084153505, - "loss": 0.9989, + "learning_rate": 0.00013536002510188753, + "loss": 5.4603, "step": 5580 }, { "epoch": 0.45, - "learning_rate": 0.00013508213022181276, - "loss": 1.0247, + "learning_rate": 0.00013522936143245344, + "loss": 5.2684, "step": 5585 }, { "epoch": 0.45, - "learning_rate": 0.00013495118103713815, - "loss": 0.9958, + "learning_rate": 0.00013509862905714993, + "loss": 5.3416, "step": 5590 }, { "epoch": 0.45, - "learning_rate": 0.0001348201635434399, - "loss": 1.0002, + "learning_rate": 0.0001349678282309372, + "loss": 5.3608, "step": 5595 }, { "epoch": 0.45, - "learning_rate": 0.00013468907799678022, - "loss": 0.9839, + "learning_rate": 0.00013483695920890875, + "loss": 5.3473, "step": 5600 }, { "epoch": 0.45, - "learning_rate": 0.00013455792465335434, - "loss": 0.9383, + "learning_rate": 0.00013470602224629124, + "loss": 5.4283, "step": 5605 }, { "epoch": 0.45, - "learning_rate": 0.00013442670376949002, - "loss": 1.018, + "learning_rate": 0.00013457501759844373, + "loss": 5.342, "step": 5610 }, { "epoch": 0.45, - "learning_rate": 0.00013429541560164697, - "loss": 1.0052, + "learning_rate": 0.00013444394552085732, + "loss": 5.3108, "step": 5615 }, { "epoch": 0.45, - "learning_rate": 0.0001341640604064164, - "loss": 0.9217, + "learning_rate": 0.00013431280626915467, + "loss": 5.3455, "step": 5620 }, { "epoch": 0.45, - "learning_rate": 0.00013403263844052055, - "loss": 0.9508, + "learning_rate": 0.0001341816000990893, + "loss": 5.3004, "step": 5625 }, { "epoch": 0.45, - "learning_rate": 0.00013390114996081223, - "loss": 1.025, + "learning_rate": 0.00013405032726654533, + "loss": 5.357, "step": 5630 }, { "epoch": 0.45, - "learning_rate": 0.00013376959522427403, - "loss": 1.0271, + "learning_rate": 0.00013391898802753697, + "loss": 5.3273, "step": 5635 }, { "epoch": 0.45, - "learning_rate": 0.00013363797448801823, - "loss": 0.9686, + "learning_rate": 0.00013378758263820777, + "loss": 5.3363, "step": 5640 }, { "epoch": 0.45, - "learning_rate": 0.00013350628800928602, - "loss": 0.9905, + "learning_rate": 0.0001336561113548304, + "loss": 5.43, "step": 5645 }, { "epoch": 0.45, - "learning_rate": 0.00013337453604544714, - "loss": 0.9573, + "learning_rate": 0.00013352457443380601, + "loss": 5.3276, "step": 5650 }, { "epoch": 0.45, - "learning_rate": 0.00013324271885399924, - "loss": 0.9975, + "learning_rate": 0.0001333929721316638, + "loss": 5.342, "step": 5655 }, { "epoch": 0.45, - "learning_rate": 0.00013311083669256745, - "loss": 1.0359, + "learning_rate": 0.0001332613047050604, + "loss": 5.385, "step": 5660 }, { "epoch": 0.45, - "learning_rate": 0.00013297888981890393, - "loss": 0.9516, + "learning_rate": 0.00013312957241077943, + "loss": 5.4021, "step": 5665 }, { "epoch": 0.45, - "learning_rate": 0.00013284687849088732, - "loss": 0.9474, + "learning_rate": 0.00013299777550573118, + "loss": 5.3764, "step": 5670 }, { "epoch": 0.45, - "learning_rate": 0.00013271480296652222, - "loss": 0.9487, + "learning_rate": 0.0001328659142469518, + "loss": 5.3596, "step": 5675 }, { "epoch": 0.45, - "learning_rate": 0.0001325826635039386, - "loss": 0.9709, + "learning_rate": 0.00013273398889160297, + "loss": 5.3697, "step": 5680 }, { - "epoch": 0.46, - "learning_rate": 0.0001324504603613915, - "loss": 1.0228, + "epoch": 0.45, + "learning_rate": 0.00013260199969697146, + "loss": 5.3325, "step": 5685 }, { "epoch": 0.46, - "learning_rate": 0.00013231819379726044, - "loss": 0.9955, + "learning_rate": 0.00013246994692046836, + "loss": 5.334, "step": 5690 }, { "epoch": 0.46, - "learning_rate": 0.0001321858640700488, - "loss": 0.9981, + "learning_rate": 0.00013233783081962897, + "loss": 5.3486, "step": 5695 }, { "epoch": 0.46, - "learning_rate": 0.00013205347143838345, - "loss": 1.0031, + "learning_rate": 0.00013220565165211195, + "loss": 5.349, "step": 5700 }, { "epoch": 0.46, - "learning_rate": 0.00013192101616101422, - "loss": 1.0357, + "learning_rate": 0.00013207340967569904, + "loss": 5.3016, "step": 5705 }, { "epoch": 0.46, - "learning_rate": 0.0001317884984968133, - "loss": 0.9896, + "learning_rate": 0.00013194110514829437, + "loss": 5.2682, "step": 5710 }, { "epoch": 0.46, - "learning_rate": 0.00013165591870477496, - "loss": 1.0353, + "learning_rate": 0.00013180873832792416, + "loss": 5.3565, "step": 5715 }, { "epoch": 0.46, - "learning_rate": 0.0001315232770440147, - "loss": 1.0126, + "learning_rate": 0.00013167630947273607, + "loss": 5.3288, "step": 5720 }, { "epoch": 0.46, - "learning_rate": 0.00013139057377376903, - "loss": 0.9875, + "learning_rate": 0.00013154381884099875, + "loss": 5.353, "step": 5725 }, { "epoch": 0.46, - "learning_rate": 0.000131257809153395, - "loss": 1.0007, + "learning_rate": 0.00013141126669110136, + "loss": 5.3295, "step": 5730 }, { "epoch": 0.46, - "learning_rate": 0.0001311249834423693, - "loss": 0.9883, + "learning_rate": 0.00013127865328155293, + "loss": 5.333, "step": 5735 }, { "epoch": 0.46, - "learning_rate": 0.0001309920969002882, - "loss": 0.9813, + "learning_rate": 0.00013114597887098212, + "loss": 5.3499, "step": 5740 }, { "epoch": 0.46, - "learning_rate": 0.00013085914978686687, - "loss": 0.9861, + "learning_rate": 0.00013101324371813638, + "loss": 5.3163, "step": 5745 }, { "epoch": 0.46, - "learning_rate": 0.0001307261423619387, - "loss": 1.0261, + "learning_rate": 0.00013088044808188184, + "loss": 5.2668, "step": 5750 }, { "epoch": 0.46, - "learning_rate": 0.00013059307488545518, - "loss": 1.0461, + "learning_rate": 0.00013074759222120235, + "loss": 5.4173, "step": 5755 }, { "epoch": 0.46, - "learning_rate": 0.00013045994761748495, - "loss": 0.9783, + "learning_rate": 0.00013061467639519935, + "loss": 5.339, "step": 5760 }, { "epoch": 0.46, - "learning_rate": 0.0001303267608182136, - "loss": 0.9397, + "learning_rate": 0.00013048170086309125, + "loss": 5.3407, "step": 5765 }, { "epoch": 0.46, - "learning_rate": 0.00013019351474794315, - "loss": 1.0485, + "learning_rate": 0.00013034866588421283, + "loss": 5.3295, "step": 5770 }, { "epoch": 0.46, - "learning_rate": 0.0001300602096670913, - "loss": 0.9919, + "learning_rate": 0.0001302155717180148, + "loss": 5.247, "step": 5775 }, { "epoch": 0.46, - "learning_rate": 0.00012992684583619117, - "loss": 0.988, + "learning_rate": 0.00013008241862406337, + "loss": 5.3713, "step": 5780 }, { "epoch": 0.46, - "learning_rate": 0.00012979342351589071, - "loss": 0.9836, + "learning_rate": 0.00012994920686203962, + "loss": 5.3915, "step": 5785 }, { "epoch": 0.46, - "learning_rate": 0.00012965994296695223, - "loss": 1.0327, + "learning_rate": 0.00012981593669173907, + "loss": 5.3389, "step": 5790 }, { "epoch": 0.46, - "learning_rate": 0.00012952640445025167, - "loss": 0.9822, + "learning_rate": 0.0001296826083730711, + "loss": 5.3186, "step": 5795 }, { "epoch": 0.46, - "learning_rate": 0.00012939280822677843, - "loss": 1.0347, + "learning_rate": 0.00012954922216605854, + "loss": 5.3332, "step": 5800 }, { "epoch": 0.46, - "learning_rate": 0.00012925915455763463, - "loss": 0.9648, + "learning_rate": 0.00012941577833083717, + "loss": 5.3104, "step": 5805 }, { - "epoch": 0.47, - "learning_rate": 0.00012912544370403462, - "loss": 0.9878, + "epoch": 0.46, + "learning_rate": 0.00012928227712765504, + "loss": 5.2953, "step": 5810 }, { "epoch": 0.47, - "learning_rate": 0.00012899167592730465, - "loss": 0.9744, + "learning_rate": 0.00012914871881687217, + "loss": 5.3056, "step": 5815 }, { "epoch": 0.47, - "learning_rate": 0.000128857851488882, - "loss": 1.0189, + "learning_rate": 0.00012901510365895993, + "loss": 5.2828, "step": 5820 }, { "epoch": 0.47, - "learning_rate": 0.00012872397065031495, - "loss": 0.9938, + "learning_rate": 0.00012888143191450052, + "loss": 5.3634, "step": 5825 }, { "epoch": 0.47, - "learning_rate": 0.00012859003367326178, - "loss": 0.9825, + "learning_rate": 0.00012874770384418654, + "loss": 5.3822, "step": 5830 }, { "epoch": 0.47, - "learning_rate": 0.00012845604081949066, - "loss": 1.0567, + "learning_rate": 0.00012861391970882046, + "loss": 5.2642, "step": 5835 }, { "epoch": 0.47, - "learning_rate": 0.00012832199235087883, - "loss": 0.9806, + "learning_rate": 0.000128480079769314, + "loss": 5.3891, "step": 5840 }, { "epoch": 0.47, - "learning_rate": 0.00012818788852941236, - "loss": 0.9878, + "learning_rate": 0.00012834618428668782, + "loss": 5.3223, "step": 5845 }, { "epoch": 0.47, - "learning_rate": 0.00012805372961718537, - "loss": 1.025, + "learning_rate": 0.00012821223352207085, + "loss": 5.3218, "step": 5850 }, { "epoch": 0.47, - "learning_rate": 0.00012791951587639972, - "loss": 0.9964, + "learning_rate": 0.0001280782277366998, + "loss": 5.3233, "step": 5855 }, { "epoch": 0.47, - "learning_rate": 0.0001277852475693644, - "loss": 1.0358, + "learning_rate": 0.0001279441671919187, + "loss": 5.3493, "step": 5860 }, { "epoch": 0.47, - "learning_rate": 0.00012765092495849512, - "loss": 1.031, + "learning_rate": 0.0001278100521491785, + "loss": 5.3039, "step": 5865 }, { "epoch": 0.47, - "learning_rate": 0.00012751654830631362, - "loss": 1.0384, + "learning_rate": 0.00012767588287003622, + "loss": 5.3753, "step": 5870 }, { "epoch": 0.47, - "learning_rate": 0.0001273821178754473, - "loss": 1.032, + "learning_rate": 0.0001275416596161548, + "loss": 5.277, "step": 5875 }, { "epoch": 0.47, - "learning_rate": 0.00012724763392862868, - "loss": 0.9451, + "learning_rate": 0.00012740738264930242, + "loss": 5.257, "step": 5880 }, { "epoch": 0.47, - "learning_rate": 0.00012711309672869487, - "loss": 0.9739, + "learning_rate": 0.000127273052231352, + "loss": 5.3391, "step": 5885 }, { "epoch": 0.47, - "learning_rate": 0.00012697850653858704, - "loss": 0.9815, + "learning_rate": 0.00012713866862428067, + "loss": 5.3949, "step": 5890 }, { "epoch": 0.47, - "learning_rate": 0.00012684386362134996, - "loss": 1.0753, + "learning_rate": 0.00012700423209016933, + "loss": 5.3308, "step": 5895 }, { "epoch": 0.47, - "learning_rate": 0.00012670916824013135, - "loss": 1.0437, + "learning_rate": 0.00012686974289120216, + "loss": 5.3034, "step": 5900 }, { "epoch": 0.47, - "learning_rate": 0.00012657442065818164, - "loss": 0.9969, + "learning_rate": 0.00012673520128966592, + "loss": 5.3333, "step": 5905 }, { "epoch": 0.47, - "learning_rate": 0.0001264396211388531, - "loss": 1.0258, + "learning_rate": 0.00012660060754794965, + "loss": 5.2585, "step": 5910 }, { "epoch": 0.47, - "learning_rate": 0.00012630476994559964, - "loss": 0.9922, + "learning_rate": 0.00012646596192854406, + "loss": 5.4027, "step": 5915 }, { "epoch": 0.47, - "learning_rate": 0.0001261698673419761, - "loss": 0.9883, + "learning_rate": 0.00012633126469404103, + "loss": 5.2967, "step": 5920 }, { "epoch": 0.47, - "learning_rate": 0.00012603491359163783, - "loss": 0.9492, + "learning_rate": 0.00012619651610713307, + "loss": 5.3236, "step": 5925 }, { "epoch": 0.47, - "learning_rate": 0.00012589990895834006, - "loss": 0.9831, + "learning_rate": 0.00012606171643061293, + "loss": 5.2916, "step": 5930 }, { - "epoch": 0.48, - "learning_rate": 0.0001257648537059376, - "loss": 0.9471, + "epoch": 0.47, + "learning_rate": 0.0001259268659273729, + "loss": 5.3112, "step": 5935 }, { "epoch": 0.48, - "learning_rate": 0.000125629748098384, - "loss": 0.9642, + "learning_rate": 0.0001257919648604045, + "loss": 5.2214, "step": 5940 }, { "epoch": 0.48, - "learning_rate": 0.0001254945923997315, - "loss": 0.9906, + "learning_rate": 0.00012565701349279772, + "loss": 5.3388, "step": 5945 }, { "epoch": 0.48, - "learning_rate": 0.00012535938687412997, - "loss": 0.9515, + "learning_rate": 0.00012552201208774078, + "loss": 5.3178, "step": 5950 }, { "epoch": 0.48, - "learning_rate": 0.00012522413178582686, - "loss": 1.0128, + "learning_rate": 0.0001253869609085194, + "loss": 5.3739, "step": 5955 }, { "epoch": 0.48, - "learning_rate": 0.00012508882739916628, - "loss": 1.0251, + "learning_rate": 0.00012525186021851642, + "loss": 5.3424, "step": 5960 }, { "epoch": 0.48, - "learning_rate": 0.00012495347397858893, - "loss": 1.0703, + "learning_rate": 0.00012511671028121122, + "loss": 5.2965, "step": 5965 }, { "epoch": 0.48, - "learning_rate": 0.00012481807178863117, - "loss": 0.9637, + "learning_rate": 0.00012498151136017922, + "loss": 5.2235, "step": 5970 }, { "epoch": 0.48, - "learning_rate": 0.00012468262109392467, - "loss": 0.9573, + "learning_rate": 0.0001248462637190914, + "loss": 5.4022, "step": 5975 }, { "epoch": 0.48, - "learning_rate": 0.00012454712215919606, - "loss": 0.955, + "learning_rate": 0.00012471096762171365, + "loss": 5.2702, "step": 5980 }, { "epoch": 0.48, - "learning_rate": 0.00012441157524926605, - "loss": 0.9636, + "learning_rate": 0.00012457562333190652, + "loss": 5.3636, "step": 5985 }, { "epoch": 0.48, - "learning_rate": 0.00012427598062904932, - "loss": 1.0159, + "learning_rate": 0.00012444023111362446, + "loss": 5.2477, "step": 5990 }, { "epoch": 0.48, - "learning_rate": 0.0001241403385635536, - "loss": 0.997, + "learning_rate": 0.00012430479123091542, + "loss": 5.3062, "step": 5995 }, { "epoch": 0.48, - "learning_rate": 0.0001240046493178794, - "loss": 0.9976, + "learning_rate": 0.00012416930394792026, + "loss": 5.2523, "step": 6000 }, { "epoch": 0.48, - "learning_rate": 0.0001238689131572196, - "loss": 0.9621, + "learning_rate": 0.0001240337695288723, + "loss": 5.3176, "step": 6005 }, { "epoch": 0.48, - "learning_rate": 0.00012373313034685853, - "loss": 1.0204, + "learning_rate": 0.00012389818823809685, + "loss": 5.2869, "step": 6010 }, { "epoch": 0.48, - "learning_rate": 0.00012359730115217186, - "loss": 0.9951, + "learning_rate": 0.00012376256034001056, + "loss": 5.2523, "step": 6015 }, { "epoch": 0.48, - "learning_rate": 0.00012346142583862584, - "loss": 1.0484, + "learning_rate": 0.00012362688609912092, + "loss": 5.3344, "step": 6020 }, { "epoch": 0.48, - "learning_rate": 0.0001233255046717769, - "loss": 1.0032, + "learning_rate": 0.000123491165780026, + "loss": 5.2959, "step": 6025 }, { "epoch": 0.48, - "learning_rate": 0.000123189537917271, - "loss": 0.9821, + "learning_rate": 0.0001233553996474135, + "loss": 5.3404, "step": 6030 }, { "epoch": 0.48, - "learning_rate": 0.00012305352584084334, - "loss": 0.9979, + "learning_rate": 0.0001232195879660607, + "loss": 5.3019, "step": 6035 }, { "epoch": 0.48, - "learning_rate": 0.00012291746870831755, - "loss": 1.0086, + "learning_rate": 0.00012308373100083345, + "loss": 5.3154, "step": 6040 }, { "epoch": 0.48, - "learning_rate": 0.00012278136678560544, - "loss": 0.9181, + "learning_rate": 0.00012294782901668612, + "loss": 5.2901, "step": 6045 }, { "epoch": 0.48, - "learning_rate": 0.00012264522033870627, - "loss": 0.9873, + "learning_rate": 0.00012281188227866084, + "loss": 5.3193, "step": 6050 }, { "epoch": 0.48, - "learning_rate": 0.00012250902963370635, - "loss": 1.0164, + "learning_rate": 0.00012267589105188694, + "loss": 5.2879, "step": 6055 }, { - "epoch": 0.49, - "learning_rate": 0.0001223727949367785, - "loss": 0.9907, + "epoch": 0.48, + "learning_rate": 0.00012253985560158062, + "loss": 5.3975, "step": 6060 }, { "epoch": 0.49, - "learning_rate": 0.0001222365165141815, - "loss": 0.9697, + "learning_rate": 0.00012240377619304423, + "loss": 5.3249, "step": 6065 }, { "epoch": 0.49, - "learning_rate": 0.00012210019463225962, - "loss": 1.0193, + "learning_rate": 0.0001222676530916659, + "loss": 5.2623, "step": 6070 }, { "epoch": 0.49, - "learning_rate": 0.00012196382955744204, - "loss": 0.9748, + "learning_rate": 0.00012213148656291898, + "loss": 5.2798, "step": 6075 }, { "epoch": 0.49, - "learning_rate": 0.00012182742155624232, - "loss": 1.0694, + "learning_rate": 0.00012199527687236148, + "loss": 5.2537, "step": 6080 }, { "epoch": 0.49, - "learning_rate": 0.00012169097089525801, - "loss": 0.9873, + "learning_rate": 0.00012185902428563561, + "loss": 5.2317, "step": 6085 }, { "epoch": 0.49, - "learning_rate": 0.00012155447784116996, - "loss": 1.0288, + "learning_rate": 0.00012172272906846722, + "loss": 5.2814, "step": 6090 }, { "epoch": 0.49, - "learning_rate": 0.00012141794266074187, - "loss": 1.0195, + "learning_rate": 0.00012158639148666534, + "loss": 5.3285, "step": 6095 }, { "epoch": 0.49, - "learning_rate": 0.00012128136562081984, - "loss": 1.0326, + "learning_rate": 0.00012145001180612152, + "loss": 5.2792, "step": 6100 }, { "epoch": 0.49, - "learning_rate": 0.00012114474698833174, - "loss": 0.9922, + "learning_rate": 0.00012131359029280958, + "loss": 5.3328, "step": 6105 }, { "epoch": 0.49, - "learning_rate": 0.00012100808703028665, - "loss": 0.9942, + "learning_rate": 0.0001211771272127848, + "loss": 5.224, "step": 6110 }, { "epoch": 0.49, - "learning_rate": 0.00012087138601377459, - "loss": 0.981, + "learning_rate": 0.00012104062283218351, + "loss": 5.2667, "step": 6115 }, { "epoch": 0.49, - "learning_rate": 0.0001207346442059657, - "loss": 0.9879, + "learning_rate": 0.00012090407741722272, + "loss": 5.3292, "step": 6120 }, { "epoch": 0.49, - "learning_rate": 0.00012059786187410984, - "loss": 0.9661, + "learning_rate": 0.00012076749123419928, + "loss": 5.322, "step": 6125 }, { "epoch": 0.49, - "learning_rate": 0.00012046103928553615, - "loss": 1.0337, + "learning_rate": 0.00012063086454948971, + "loss": 5.3504, "step": 6130 }, { "epoch": 0.49, - "learning_rate": 0.00012032417670765237, - "loss": 1.0521, + "learning_rate": 0.00012049419762954943, + "loss": 5.3132, "step": 6135 }, { "epoch": 0.49, - "learning_rate": 0.00012018727440794441, - "loss": 0.9065, + "learning_rate": 0.00012035749074091236, + "loss": 5.2941, "step": 6140 }, { "epoch": 0.49, - "learning_rate": 0.00012005033265397592, - "loss": 0.9723, + "learning_rate": 0.00012022074415019036, + "loss": 5.3279, "step": 6145 }, { "epoch": 0.49, - "learning_rate": 0.00011991335171338745, - "loss": 0.97, + "learning_rate": 0.00012008395812407271, + "loss": 5.3169, "step": 6150 }, { "epoch": 0.49, - "learning_rate": 0.00011977633185389632, - "loss": 1.0342, + "learning_rate": 0.00011994713292932562, + "loss": 5.2837, "step": 6155 }, { "epoch": 0.49, - "learning_rate": 0.00011963927334329586, - "loss": 1.0223, + "learning_rate": 0.0001198102688327917, + "loss": 5.3166, "step": 6160 }, { "epoch": 0.49, - "learning_rate": 0.00011950217644945492, - "loss": 1.0107, + "learning_rate": 0.00011967336610138937, + "loss": 5.2438, "step": 6165 }, { "epoch": 0.49, - "learning_rate": 0.00011936504144031737, - "loss": 0.9509, + "learning_rate": 0.00011953642500211244, + "loss": 5.2971, "step": 6170 }, { "epoch": 0.49, - "learning_rate": 0.00011922786858390154, - "loss": 0.9948, + "learning_rate": 0.00011939944580202956, + "loss": 5.2971, "step": 6175 }, { "epoch": 0.49, - "learning_rate": 0.00011909065814829984, - "loss": 0.9609, + "learning_rate": 0.00011926242876828363, + "loss": 5.281, "step": 6180 }, { - "epoch": 0.5, - "learning_rate": 0.00011895341040167803, - "loss": 1.0529, + "epoch": 0.49, + "learning_rate": 0.0001191253741680914, + "loss": 5.2394, "step": 6185 }, { "epoch": 0.5, - "learning_rate": 0.00011881612561227482, - "loss": 1.0297, + "learning_rate": 0.00011898828226874284, + "loss": 5.2937, "step": 6190 }, { "epoch": 0.5, - "learning_rate": 0.00011867880404840128, - "loss": 0.9812, + "learning_rate": 0.0001188511533376007, + "loss": 5.2404, "step": 6195 }, { "epoch": 0.5, - "learning_rate": 0.00011854144597844043, - "loss": 0.9165, + "learning_rate": 0.00011871398764209988, + "loss": 5.2616, "step": 6200 }, { "epoch": 0.5, - "learning_rate": 0.00011840405167084658, - "loss": 0.9711, + "learning_rate": 0.00011857678544974708, + "loss": 5.304, "step": 6205 }, { "epoch": 0.5, - "learning_rate": 0.00011826662139414483, - "loss": 1.0606, + "learning_rate": 0.00011843954702812008, + "loss": 5.3124, "step": 6210 }, { "epoch": 0.5, - "learning_rate": 0.0001181291554169307, - "loss": 0.996, + "learning_rate": 0.0001183022726448674, + "loss": 5.3032, "step": 6215 }, { "epoch": 0.5, - "learning_rate": 0.00011799165400786937, - "loss": 0.9863, + "learning_rate": 0.00011816496256770759, + "loss": 5.3325, "step": 6220 }, { "epoch": 0.5, - "learning_rate": 0.0001178541174356953, - "loss": 0.9844, + "learning_rate": 0.00011802761706442896, + "loss": 5.1577, "step": 6225 }, { "epoch": 0.5, - "learning_rate": 0.00011771654596921169, - "loss": 1.0437, + "learning_rate": 0.00011789023640288881, + "loss": 5.2629, "step": 6230 }, { "epoch": 0.5, - "learning_rate": 0.00011757893987728992, - "loss": 0.9774, + "learning_rate": 0.00011775282085101298, + "loss": 5.3018, "step": 6235 }, { "epoch": 0.5, - "learning_rate": 0.00011744129942886906, - "loss": 0.9706, + "learning_rate": 0.00011761537067679543, + "loss": 5.3968, "step": 6240 }, { "epoch": 0.5, - "learning_rate": 0.00011730362489295535, - "loss": 1.0428, + "learning_rate": 0.00011747788614829758, + "loss": 5.2864, "step": 6245 }, { "epoch": 0.5, - "learning_rate": 0.00011716591653862152, - "loss": 0.969, + "learning_rate": 0.00011734036753364792, + "loss": 5.3099, "step": 6250 }, { "epoch": 0.5, - "learning_rate": 0.00011702817463500659, - "loss": 1.0099, + "learning_rate": 0.00011720281510104136, + "loss": 5.2449, "step": 6255 }, { "epoch": 0.5, - "learning_rate": 0.00011689039945131503, - "loss": 1.0127, + "learning_rate": 0.00011706522911873878, + "loss": 5.2738, "step": 6260 }, { "epoch": 0.5, - "learning_rate": 0.00011675259125681634, - "loss": 1.0053, + "learning_rate": 0.00011692760985506646, + "loss": 5.3661, "step": 6265 }, { "epoch": 0.5, - "learning_rate": 0.00011661475032084462, - "loss": 0.9798, + "learning_rate": 0.00011678995757841565, + "loss": 5.2363, "step": 6270 }, { "epoch": 0.5, - "learning_rate": 0.00011647687691279786, - "loss": 1.0002, + "learning_rate": 0.00011665227255724193, + "loss": 5.2677, "step": 6275 }, { "epoch": 0.5, - "learning_rate": 0.00011633897130213761, - "loss": 0.9966, + "learning_rate": 0.0001165145550600648, + "loss": 5.3264, "step": 6280 }, { "epoch": 0.5, - "learning_rate": 0.00011620103375838832, - "loss": 0.9581, + "learning_rate": 0.000116376805355467, + "loss": 5.2732, "step": 6285 }, { "epoch": 0.5, - "learning_rate": 0.00011606306455113678, - "loss": 1.0094, + "learning_rate": 0.00011623902371209419, + "loss": 5.3537, "step": 6290 }, { "epoch": 0.5, - "learning_rate": 0.00011592506395003178, - "loss": 1.0082, + "learning_rate": 0.0001161012103986542, + "loss": 5.306, "step": 6295 }, { "epoch": 0.5, - "learning_rate": 0.0001157870322247834, - "loss": 1.0364, + "learning_rate": 0.00011596336568391674, + "loss": 5.2017, "step": 6300 }, { "epoch": 0.5, - "learning_rate": 0.00011564896964516259, - "loss": 0.9476, + "learning_rate": 0.0001158254898367127, + "loss": 5.3281, "step": 6305 }, { - "epoch": 0.51, - "learning_rate": 0.00011551087648100054, - "loss": 0.9407, + "epoch": 0.5, + "learning_rate": 0.00011568758312593371, + "loss": 5.2568, "step": 6310 }, { "epoch": 0.51, - "learning_rate": 0.00011537275300218826, - "loss": 1.0166, + "learning_rate": 0.00011554964582053156, + "loss": 5.2703, "step": 6315 }, { "epoch": 0.51, - "learning_rate": 0.00011523459947867601, - "loss": 1.0379, + "learning_rate": 0.00011541167818951772, + "loss": 5.2938, "step": 6320 }, { "epoch": 0.51, - "learning_rate": 0.00011509641618047273, - "loss": 1.0377, + "learning_rate": 0.00011527368050196281, + "loss": 5.2591, "step": 6325 }, { "epoch": 0.51, - "learning_rate": 0.00011495820337764562, - "loss": 1.0152, + "learning_rate": 0.00011513565302699607, + "loss": 5.3245, "step": 6330 }, { "epoch": 0.51, - "learning_rate": 0.00011481996134031947, - "loss": 0.9655, + "learning_rate": 0.00011499759603380487, + "loss": 5.3193, "step": 6335 }, { "epoch": 0.51, - "learning_rate": 0.00011468169033867623, - "loss": 1.0001, + "learning_rate": 0.00011485950979163403, + "loss": 5.2526, "step": 6340 }, { "epoch": 0.51, - "learning_rate": 0.00011454339064295454, - "loss": 0.9797, + "learning_rate": 0.00011472139456978555, + "loss": 5.2593, "step": 6345 }, { "epoch": 0.51, - "learning_rate": 0.00011440506252344896, - "loss": 0.9823, + "learning_rate": 0.00011458325063761784, + "loss": 5.268, "step": 6350 }, { "epoch": 0.51, - "learning_rate": 0.00011426670625050973, - "loss": 0.9479, + "learning_rate": 0.00011444507826454538, + "loss": 5.245, "step": 6355 }, { "epoch": 0.51, - "learning_rate": 0.0001141283220945421, - "loss": 1.0003, + "learning_rate": 0.00011430687772003805, + "loss": 5.3079, "step": 6360 }, { "epoch": 0.51, - "learning_rate": 0.00011398991032600572, - "loss": 0.9475, + "learning_rate": 0.00011416864927362077, + "loss": 5.3133, "step": 6365 }, { "epoch": 0.51, - "learning_rate": 0.00011385147121541435, - "loss": 0.9971, + "learning_rate": 0.00011403039319487272, + "loss": 5.2596, "step": 6370 }, { "epoch": 0.51, - "learning_rate": 0.00011371300503333504, - "loss": 0.9917, + "learning_rate": 0.00011389210975342714, + "loss": 5.325, "step": 6375 }, { "epoch": 0.51, - "learning_rate": 0.00011357451205038787, - "loss": 0.9768, + "learning_rate": 0.00011375379921897051, + "loss": 5.1978, "step": 6380 }, { "epoch": 0.51, - "learning_rate": 0.00011343599253724526, - "loss": 0.9822, + "learning_rate": 0.00011361546186124219, + "loss": 5.333, "step": 6385 }, { "epoch": 0.51, - "learning_rate": 0.00011329744676463143, - "loss": 1.0197, + "learning_rate": 0.00011347709795003389, + "loss": 5.27, "step": 6390 }, { "epoch": 0.51, - "learning_rate": 0.00011315887500332199, - "loss": 1.0374, + "learning_rate": 0.00011333870775518904, + "loss": 5.2847, "step": 6395 }, { "epoch": 0.51, - "learning_rate": 0.0001130202775241433, - "loss": 0.9708, + "learning_rate": 0.00011320029154660241, + "loss": 5.238, "step": 6400 }, { "epoch": 0.51, - "learning_rate": 0.00011288165459797204, - "loss": 1.0454, + "learning_rate": 0.00011306184959421939, + "loss": 5.2777, "step": 6405 }, { "epoch": 0.51, - "learning_rate": 0.00011274300649573452, - "loss": 0.9838, + "learning_rate": 0.0001129233821680357, + "loss": 5.275, "step": 6410 }, { "epoch": 0.51, - "learning_rate": 0.00011260433348840637, - "loss": 0.9678, + "learning_rate": 0.00011278488953809661, + "loss": 5.2852, "step": 6415 }, { "epoch": 0.51, - "learning_rate": 0.00011246563584701186, - "loss": 1.0421, + "learning_rate": 0.0001126463719744967, + "loss": 5.2595, "step": 6420 }, { "epoch": 0.51, - "learning_rate": 0.00011232691384262333, - "loss": 0.9182, + "learning_rate": 0.00011250782974737899, + "loss": 5.2172, "step": 6425 }, { "epoch": 0.51, - "learning_rate": 0.00011218816774636083, - "loss": 0.9625, + "learning_rate": 0.00011236926312693479, + "loss": 5.2783, "step": 6430 }, { - "epoch": 0.52, - "learning_rate": 0.00011204939782939142, - "loss": 0.9673, + "epoch": 0.51, + "learning_rate": 0.00011223067238340284, + "loss": 5.2711, "step": 6435 }, { "epoch": 0.52, - "learning_rate": 0.00011191060436292878, - "loss": 1.0176, + "learning_rate": 0.00011209205778706898, + "loss": 5.2397, "step": 6440 }, { "epoch": 0.52, - "learning_rate": 0.00011177178761823262, - "loss": 1.0095, + "learning_rate": 0.00011195341960826558, + "loss": 5.2622, "step": 6445 }, { "epoch": 0.52, - "learning_rate": 0.00011163294786660807, - "loss": 0.9861, + "learning_rate": 0.00011181475811737101, + "loss": 5.2977, "step": 6450 }, { "epoch": 0.52, - "learning_rate": 0.00011149408537940529, - "loss": 1.0183, + "learning_rate": 0.00011167607358480907, + "loss": 5.2936, "step": 6455 }, { "epoch": 0.52, - "learning_rate": 0.00011135520042801886, - "loss": 1.051, + "learning_rate": 0.00011153736628104851, + "loss": 5.2288, "step": 6460 }, { "epoch": 0.52, - "learning_rate": 0.00011121629328388725, - "loss": 1.0083, + "learning_rate": 0.00011139863647660247, + "loss": 5.2845, "step": 6465 }, { "epoch": 0.52, - "learning_rate": 0.00011107736421849231, - "loss": 0.94, + "learning_rate": 0.00011125988444202804, + "loss": 5.3017, "step": 6470 }, { "epoch": 0.52, - "learning_rate": 0.00011093841350335872, - "loss": 1.002, + "learning_rate": 0.00011112111044792557, + "loss": 5.1976, "step": 6475 }, { "epoch": 0.52, - "learning_rate": 0.00011079944141005354, - "loss": 1.033, + "learning_rate": 0.00011098231476493828, + "loss": 5.2736, "step": 6480 }, { "epoch": 0.52, - "learning_rate": 0.00011066044821018555, - "loss": 0.9949, + "learning_rate": 0.00011084349766375174, + "loss": 5.347, "step": 6485 }, { "epoch": 0.52, - "learning_rate": 0.00011052143417540473, - "loss": 0.9654, + "learning_rate": 0.0001107046594150932, + "loss": 5.2253, "step": 6490 }, { "epoch": 0.52, - "learning_rate": 0.00011038239957740189, - "loss": 1.0095, + "learning_rate": 0.00011056580028973118, + "loss": 5.3042, "step": 6495 }, { "epoch": 0.52, - "learning_rate": 0.00011024334468790798, - "loss": 1.0326, + "learning_rate": 0.00011042692055847499, + "loss": 5.279, "step": 6500 }, { "epoch": 0.52, - "learning_rate": 0.00011010426977869361, - "loss": 1.0698, + "learning_rate": 0.00011028802049217404, + "loss": 5.2396, "step": 6505 }, { "epoch": 0.52, - "learning_rate": 0.0001099651751215685, - "loss": 0.9698, + "learning_rate": 0.0001101491003617174, + "loss": 5.3328, "step": 6510 }, { "epoch": 0.52, - "learning_rate": 0.00010982606098838097, - "loss": 1.0031, + "learning_rate": 0.00011001016043803331, + "loss": 5.2362, "step": 6515 }, { "epoch": 0.52, - "learning_rate": 0.00010968692765101746, - "loss": 1.0349, + "learning_rate": 0.00010987120099208858, + "loss": 5.2459, "step": 6520 }, { "epoch": 0.52, - "learning_rate": 0.00010954777538140184, - "loss": 1.0119, + "learning_rate": 0.00010973222229488817, + "loss": 5.2246, "step": 6525 }, { "epoch": 0.52, - "learning_rate": 0.00010940860445149508, - "loss": 0.9532, + "learning_rate": 0.00010959322461747446, + "loss": 5.2472, "step": 6530 }, { "epoch": 0.52, - "learning_rate": 0.0001092694151332946, - "loss": 0.9234, + "learning_rate": 0.00010945420823092694, + "loss": 5.3863, "step": 6535 }, { "epoch": 0.52, - "learning_rate": 0.00010913020769883367, - "loss": 0.9893, + "learning_rate": 0.0001093151734063615, + "loss": 5.2757, "step": 6540 }, { "epoch": 0.52, - "learning_rate": 0.0001089909824201811, - "loss": 0.9733, + "learning_rate": 0.00010917612041493016, + "loss": 5.1946, "step": 6545 }, { "epoch": 0.52, - "learning_rate": 0.00010885173956944046, - "loss": 1.0025, + "learning_rate": 0.00010903704952782015, + "loss": 5.2073, "step": 6550 }, { - "epoch": 0.53, - "learning_rate": 0.0001087124794187497, - "loss": 0.9917, + "epoch": 0.52, + "learning_rate": 0.00010889796101625374, + "loss": 5.2399, "step": 6555 }, { - "epoch": 0.53, - "learning_rate": 0.0001085732022402807, - "loss": 0.99, + "epoch": 0.52, + "learning_rate": 0.00010875885515148754, + "loss": 5.2962, "step": 6560 }, { "epoch": 0.53, - "learning_rate": 0.00010843390830623838, - "loss": 0.976, + "learning_rate": 0.00010861973220481202, + "loss": 5.1872, "step": 6565 }, { "epoch": 0.53, - "learning_rate": 0.0001082945978888606, - "loss": 1.011, + "learning_rate": 0.00010848059244755093, + "loss": 5.3185, "step": 6570 }, { "epoch": 0.53, - "learning_rate": 0.00010815527126041738, - "loss": 0.9861, + "learning_rate": 0.00010834143615106081, + "loss": 5.2768, "step": 6575 }, { "epoch": 0.53, - "learning_rate": 0.00010801592869321044, - "loss": 0.9631, + "learning_rate": 0.00010820226358673049, + "loss": 5.2399, "step": 6580 }, { "epoch": 0.53, - "learning_rate": 0.0001078765704595726, - "loss": 1.0231, + "learning_rate": 0.0001080630750259805, + "loss": 5.254, "step": 6585 }, { "epoch": 0.53, - "learning_rate": 0.00010773719683186727, - "loss": 0.9886, + "learning_rate": 0.00010792387074026256, + "loss": 5.2913, "step": 6590 }, { "epoch": 0.53, - "learning_rate": 0.00010759780808248811, - "loss": 0.9973, + "learning_rate": 0.00010778465100105903, + "loss": 5.2854, "step": 6595 }, { "epoch": 0.53, - "learning_rate": 0.00010745840448385815, - "loss": 0.9609, + "learning_rate": 0.00010764541607988252, + "loss": 5.3421, "step": 6600 }, { "epoch": 0.53, - "learning_rate": 0.0001073189863084296, - "loss": 1.0502, + "learning_rate": 0.0001075061662482751, + "loss": 5.223, "step": 6605 }, { "epoch": 0.53, - "learning_rate": 0.00010717955382868299, - "loss": 0.9896, + "learning_rate": 0.00010736690177780801, + "loss": 5.2382, "step": 6610 }, { "epoch": 0.53, - "learning_rate": 0.00010704010731712695, - "loss": 1.0049, + "learning_rate": 0.00010722762294008106, + "loss": 5.252, "step": 6615 }, { "epoch": 0.53, - "learning_rate": 0.00010690064704629748, - "loss": 0.9808, + "learning_rate": 0.00010708833000672205, + "loss": 5.2684, "step": 6620 }, { "epoch": 0.53, - "learning_rate": 0.00010676117328875741, - "loss": 0.9923, + "learning_rate": 0.00010694902324938618, + "loss": 5.2569, "step": 6625 }, { "epoch": 0.53, - "learning_rate": 0.00010662168631709606, - "loss": 1.0161, + "learning_rate": 0.00010680970293975579, + "loss": 5.2781, "step": 6630 }, { "epoch": 0.53, - "learning_rate": 0.00010648218640392848, - "loss": 0.9613, + "learning_rate": 0.00010667036934953944, + "loss": 5.3239, "step": 6635 }, { "epoch": 0.53, - "learning_rate": 0.00010634267382189504, - "loss": 1.0424, + "learning_rate": 0.00010653102275047182, + "loss": 5.2684, "step": 6640 }, { "epoch": 0.53, - "learning_rate": 0.00010620314884366089, - "loss": 0.9958, + "learning_rate": 0.00010639166341431277, + "loss": 5.2535, "step": 6645 }, { "epoch": 0.53, - "learning_rate": 0.00010606361174191531, - "loss": 1.0184, + "learning_rate": 0.00010625229161284715, + "loss": 5.3044, "step": 6650 }, { "epoch": 0.53, - "learning_rate": 0.00010592406278937144, - "loss": 1.0047, + "learning_rate": 0.00010611290761788401, + "loss": 5.3256, "step": 6655 }, { "epoch": 0.53, - "learning_rate": 0.00010578450225876546, - "loss": 0.9699, + "learning_rate": 0.00010597351170125622, + "loss": 5.2388, "step": 6660 }, { "epoch": 0.53, - "learning_rate": 0.00010564493042285617, - "loss": 1.0504, + "learning_rate": 0.00010583410413481994, + "loss": 5.1873, "step": 6665 }, { "epoch": 0.53, - "learning_rate": 0.00010550534755442454, - "loss": 0.9949, + "learning_rate": 0.00010569468519045398, + "loss": 5.2795, "step": 6670 }, { "epoch": 0.53, - "learning_rate": 0.00010536575392627307, - "loss": 0.9858, + "learning_rate": 0.00010555525514005942, + "loss": 5.2921, "step": 6675 }, { - "epoch": 0.54, - "learning_rate": 0.0001052261498112253, - "loss": 0.997, + "epoch": 0.53, + "learning_rate": 0.00010541581425555886, + "loss": 5.319, "step": 6680 }, { - "epoch": 0.54, - "learning_rate": 0.00010508653548212524, - "loss": 1.0086, + "epoch": 0.53, + "learning_rate": 0.00010527636280889623, + "loss": 5.2804, "step": 6685 }, { "epoch": 0.54, - "learning_rate": 0.00010494691121183681, - "loss": 1.0244, + "learning_rate": 0.00010513690107203589, + "loss": 5.1747, "step": 6690 }, { "epoch": 0.54, - "learning_rate": 0.00010480727727324354, - "loss": 1.0777, + "learning_rate": 0.00010499742931696235, + "loss": 5.2341, "step": 6695 }, { "epoch": 0.54, - "learning_rate": 0.00010466763393924763, - "loss": 0.9938, + "learning_rate": 0.00010485794781567961, + "loss": 5.2393, "step": 6700 }, { "epoch": 0.54, - "learning_rate": 0.00010452798148276985, - "loss": 1.0416, + "learning_rate": 0.00010471845684021073, + "loss": 5.1983, "step": 6705 }, { "epoch": 0.54, - "learning_rate": 0.00010438832017674862, - "loss": 1.0135, + "learning_rate": 0.0001045789566625972, + "loss": 5.2315, "step": 6710 }, { "epoch": 0.54, - "learning_rate": 0.00010424865029413975, - "loss": 1.0191, + "learning_rate": 0.00010443944755489854, + "loss": 5.2522, "step": 6715 }, { "epoch": 0.54, - "learning_rate": 0.00010410897210791585, - "loss": 1.0101, + "learning_rate": 0.00010429992978919152, + "loss": 5.2671, "step": 6720 }, { "epoch": 0.54, - "learning_rate": 0.00010396928589106567, - "loss": 0.9814, + "learning_rate": 0.00010416040363756997, + "loss": 5.2968, "step": 6725 }, { "epoch": 0.54, - "learning_rate": 0.00010382959191659367, - "loss": 0.9408, + "learning_rate": 0.00010402086937214399, + "loss": 5.2119, "step": 6730 }, { "epoch": 0.54, - "learning_rate": 0.00010368989045751956, - "loss": 0.9942, + "learning_rate": 0.00010388132726503953, + "loss": 5.2206, "step": 6735 }, { "epoch": 0.54, - "learning_rate": 0.00010355018178687754, - "loss": 1.0073, + "learning_rate": 0.0001037417775883978, + "loss": 5.2255, "step": 6740 }, { "epoch": 0.54, - "learning_rate": 0.00010341046617771603, - "loss": 1.0184, + "learning_rate": 0.0001036022206143748, + "loss": 5.241, "step": 6745 }, { "epoch": 0.54, - "learning_rate": 0.00010327074390309692, - "loss": 0.9546, + "learning_rate": 0.00010346265661514078, + "loss": 5.1686, "step": 6750 }, { "epoch": 0.54, - "learning_rate": 0.00010313101523609517, - "loss": 0.973, + "learning_rate": 0.00010332308586287962, + "loss": 5.2062, "step": 6755 }, { "epoch": 0.54, - "learning_rate": 0.00010299128044979826, - "loss": 1.0636, + "learning_rate": 0.00010318350862978848, + "loss": 5.2481, "step": 6760 }, { "epoch": 0.54, - "learning_rate": 0.00010285153981730556, - "loss": 0.9091, + "learning_rate": 0.000103043925188077, + "loss": 5.3213, "step": 6765 }, { "epoch": 0.54, - "learning_rate": 0.00010271179361172792, - "loss": 1.0056, + "learning_rate": 0.0001029043358099671, + "loss": 5.3188, "step": 6770 }, { "epoch": 0.54, - "learning_rate": 0.00010257204210618705, - "loss": 1.0013, + "learning_rate": 0.00010276474076769216, + "loss": 5.2812, "step": 6775 }, { "epoch": 0.54, - "learning_rate": 0.00010243228557381505, - "loss": 1.0223, + "learning_rate": 0.00010262514033349666, + "loss": 5.2459, "step": 6780 }, { "epoch": 0.54, - "learning_rate": 0.00010229252428775383, - "loss": 1.0397, + "learning_rate": 0.0001024855347796356, + "loss": 5.1943, "step": 6785 }, { "epoch": 0.54, - "learning_rate": 0.00010215275852115459, - "loss": 1.0513, + "learning_rate": 0.00010234592437837396, + "loss": 5.1914, "step": 6790 }, { "epoch": 0.54, - "learning_rate": 0.00010201298854717727, - "loss": 1.0167, + "learning_rate": 0.00010220630940198612, + "loss": 5.2616, "step": 6795 }, { "epoch": 0.54, - "learning_rate": 0.00010187321463899007, - "loss": 1.0278, + "learning_rate": 0.00010206669012275545, + "loss": 5.2873, "step": 6800 }, { - "epoch": 0.55, - "learning_rate": 0.00010173343706976881, - "loss": 1.0044, + "epoch": 0.54, + "learning_rate": 0.00010192706681297365, + "loss": 5.233, "step": 6805 }, { - "epoch": 0.55, - "learning_rate": 0.00010159365611269655, - "loss": 0.9418, + "epoch": 0.54, + "learning_rate": 0.00010178743974494038, + "loss": 5.1955, "step": 6810 }, { "epoch": 0.55, - "learning_rate": 0.00010145387204096295, - "loss": 0.9794, + "learning_rate": 0.00010164780919096248, + "loss": 5.2473, "step": 6815 }, { "epoch": 0.55, - "learning_rate": 0.00010131408512776372, - "loss": 0.9551, + "learning_rate": 0.00010150817542335379, + "loss": 5.2728, "step": 6820 }, { "epoch": 0.55, - "learning_rate": 0.00010117429564630014, - "loss": 0.969, + "learning_rate": 0.00010136853871443417, + "loss": 5.2254, "step": 6825 }, { "epoch": 0.55, - "learning_rate": 0.00010103450386977851, - "loss": 1.0482, + "learning_rate": 0.00010122889933652942, + "loss": 5.2349, "step": 6830 }, { "epoch": 0.55, - "learning_rate": 0.00010089471007140964, - "loss": 0.9877, + "learning_rate": 0.00010108925756197046, + "loss": 5.2407, "step": 6835 }, { "epoch": 0.55, - "learning_rate": 0.00010075491452440826, - "loss": 0.9921, + "learning_rate": 0.0001009496136630929, + "loss": 5.1996, "step": 6840 }, { "epoch": 0.55, - "learning_rate": 0.00010061511750199255, - "loss": 0.9788, + "learning_rate": 0.00010080996791223652, + "loss": 5.2638, "step": 6845 }, { "epoch": 0.55, - "learning_rate": 0.00010047531927738349, - "loss": 1.0605, + "learning_rate": 0.00010067032058174462, + "loss": 5.2154, "step": 6850 }, { "epoch": 0.55, - "learning_rate": 0.00010033552012380453, - "loss": 1.0463, + "learning_rate": 0.0001005306719439637, + "loss": 5.3085, "step": 6855 }, { "epoch": 0.55, - "learning_rate": 0.00010019572031448089, - "loss": 1.0271, + "learning_rate": 0.0001003910222712427, + "loss": 5.2741, "step": 6860 }, { "epoch": 0.55, - "learning_rate": 0.00010005592012263897, - "loss": 1.0287, + "learning_rate": 0.00010025137183593268, + "loss": 5.2422, "step": 6865 }, { "epoch": 0.55, - "learning_rate": 9.991611982150608e-05, - "loss": 0.9803, + "learning_rate": 0.00010011172091038609, + "loss": 5.2517, "step": 6870 }, { "epoch": 0.55, - "learning_rate": 9.977631968430968e-05, - "loss": 1.0218, + "learning_rate": 9.99720697669564e-05, + "loss": 5.2749, "step": 6875 }, { "epoch": 0.55, - "learning_rate": 9.963651998427689e-05, - "loss": 0.9715, + "learning_rate": 9.983241867799752e-05, + "loss": 5.2569, "step": 6880 }, { "epoch": 0.55, - "learning_rate": 9.949672099463393e-05, - "loss": 1.0477, + "learning_rate": 9.96927679158632e-05, + "loss": 5.1778, "step": 6885 }, { "epoch": 0.55, - "learning_rate": 9.935692298860577e-05, - "loss": 0.9935, + "learning_rate": 9.955311775290655e-05, + "loss": 5.2379, "step": 6890 }, { "epoch": 0.55, - "learning_rate": 9.921712623941528e-05, - "loss": 1.0601, + "learning_rate": 9.941346846147955e-05, + "loss": 5.1653, "step": 6895 }, { "epoch": 0.55, - "learning_rate": 9.907733102028307e-05, - "loss": 0.9547, + "learning_rate": 9.927382031393247e-05, + "loss": 5.1989, "step": 6900 }, { "epoch": 0.55, - "learning_rate": 9.89375376044266e-05, - "loss": 0.9689, + "learning_rate": 9.913417358261339e-05, + "loss": 5.3352, "step": 6905 }, { "epoch": 0.55, - "learning_rate": 9.879774626505985e-05, - "loss": 1.025, + "learning_rate": 9.899452853986748e-05, + "loss": 5.1693, "step": 6910 }, { "epoch": 0.55, - "learning_rate": 9.865795727539281e-05, - "loss": 1.0179, + "learning_rate": 9.885488545803683e-05, + "loss": 5.2968, "step": 6915 }, { "epoch": 0.55, - "learning_rate": 9.851817090863073e-05, - "loss": 1.0075, + "learning_rate": 9.871524460945957e-05, + "loss": 5.2328, "step": 6920 }, { "epoch": 0.55, - "learning_rate": 9.837838743797394e-05, - "loss": 0.9798, + "learning_rate": 9.857560626646947e-05, + "loss": 5.2322, "step": 6925 }, { - "epoch": 0.56, - "learning_rate": 9.823860713661692e-05, - "loss": 1.0018, + "epoch": 0.55, + "learning_rate": 9.843597070139539e-05, + "loss": 5.2722, "step": 6930 }, { - "epoch": 0.56, - "learning_rate": 9.809883027774808e-05, - "loss": 1.058, + "epoch": 0.55, + "learning_rate": 9.829633818656097e-05, + "loss": 5.2442, "step": 6935 }, { "epoch": 0.56, - "learning_rate": 9.795905713454902e-05, - "loss": 1.0314, + "learning_rate": 9.815670899428365e-05, + "loss": 5.2272, "step": 6940 }, { "epoch": 0.56, - "learning_rate": 9.781928798019409e-05, - "loss": 1.0081, + "learning_rate": 9.801708339687454e-05, + "loss": 5.195, "step": 6945 }, { "epoch": 0.56, - "learning_rate": 9.767952308784992e-05, - "loss": 1.0202, + "learning_rate": 9.787746166663764e-05, + "loss": 5.1505, "step": 6950 }, { "epoch": 0.56, - "learning_rate": 9.753976273067475e-05, - "loss": 1.013, + "learning_rate": 9.773784407586955e-05, + "loss": 5.3356, "step": 6955 }, { "epoch": 0.56, - "learning_rate": 9.740000718181795e-05, - "loss": 0.9805, + "learning_rate": 9.759823089685868e-05, + "loss": 5.1821, "step": 6960 }, { "epoch": 0.56, - "learning_rate": 9.726025671441948e-05, - "loss": 1.029, + "learning_rate": 9.745862240188487e-05, + "loss": 5.2567, "step": 6965 }, { "epoch": 0.56, - "learning_rate": 9.712051160160944e-05, - "loss": 0.992, + "learning_rate": 9.731901886321876e-05, + "loss": 5.1821, "step": 6970 }, { "epoch": 0.56, - "learning_rate": 9.698077211650736e-05, - "loss": 1.0309, + "learning_rate": 9.717942055312149e-05, + "loss": 5.2792, "step": 6975 }, { "epoch": 0.56, - "learning_rate": 9.684103853222192e-05, - "loss": 0.993, + "learning_rate": 9.703982774384386e-05, + "loss": 5.2593, "step": 6980 }, { "epoch": 0.56, - "learning_rate": 9.670131112185009e-05, - "loss": 1.0155, + "learning_rate": 9.690024070762596e-05, + "loss": 5.2172, "step": 6985 }, { "epoch": 0.56, - "learning_rate": 9.656159015847694e-05, - "loss": 1.0212, + "learning_rate": 9.676065971669664e-05, + "loss": 5.2154, "step": 6990 }, { "epoch": 0.56, - "learning_rate": 9.64218759151748e-05, - "loss": 0.9725, + "learning_rate": 9.662108504327301e-05, + "loss": 5.2237, "step": 6995 }, { "epoch": 0.56, - "learning_rate": 9.62821686650029e-05, - "loss": 0.9735, + "learning_rate": 9.648151695955977e-05, + "loss": 5.243, "step": 7000 }, { "epoch": 0.56, - "learning_rate": 9.614246868100692e-05, - "loss": 0.9362, + "learning_rate": 9.634195573774882e-05, + "loss": 5.2837, "step": 7005 }, { "epoch": 0.56, - "learning_rate": 9.60027762362182e-05, - "loss": 0.9738, + "learning_rate": 9.620240165001868e-05, + "loss": 5.1633, "step": 7010 }, { "epoch": 0.56, - "learning_rate": 9.586309160365341e-05, - "loss": 0.9576, + "learning_rate": 9.606285496853397e-05, + "loss": 5.135, "step": 7015 }, { "epoch": 0.56, - "learning_rate": 9.572341505631392e-05, - "loss": 1.0606, + "learning_rate": 9.592331596544482e-05, + "loss": 5.1669, "step": 7020 }, { "epoch": 0.56, - "learning_rate": 9.558374686718532e-05, - "loss": 1.0058, + "learning_rate": 9.578378491288636e-05, + "loss": 5.2129, "step": 7025 }, { "epoch": 0.56, - "learning_rate": 9.544408730923686e-05, - "loss": 1.0368, + "learning_rate": 9.564426208297833e-05, + "loss": 5.324, "step": 7030 }, { "epoch": 0.56, - "learning_rate": 9.530443665542092e-05, - "loss": 0.987, + "learning_rate": 9.550474774782435e-05, + "loss": 5.1592, "step": 7035 }, { "epoch": 0.56, - "learning_rate": 9.51647951786725e-05, - "loss": 0.9848, + "learning_rate": 9.536524217951148e-05, + "loss": 5.1808, "step": 7040 }, { "epoch": 0.56, - "learning_rate": 9.502516315190859e-05, - "loss": 1.016, + "learning_rate": 9.522574565010965e-05, + "loss": 5.2288, "step": 7045 }, { "epoch": 0.56, - "learning_rate": 9.488554084802778e-05, - "loss": 0.9698, + "learning_rate": 9.508625843167128e-05, + "loss": 5.267, "step": 7050 }, { - "epoch": 0.57, - "learning_rate": 9.474592853990968e-05, - "loss": 1.0255, + "epoch": 0.56, + "learning_rate": 9.494678079623052e-05, + "loss": 5.2181, "step": 7055 }, { - "epoch": 0.57, - "learning_rate": 9.46063265004143e-05, - "loss": 0.983, + "epoch": 0.56, + "learning_rate": 9.480731301580283e-05, + "loss": 5.1698, "step": 7060 }, { "epoch": 0.57, - "learning_rate": 9.446673500238158e-05, - "loss": 1.0292, + "learning_rate": 9.466785536238454e-05, + "loss": 5.1257, "step": 7065 }, { "epoch": 0.57, - "learning_rate": 9.432715431863093e-05, - "loss": 0.9783, + "learning_rate": 9.452840810795214e-05, + "loss": 5.2276, "step": 7070 }, { "epoch": 0.57, - "learning_rate": 9.418758472196049e-05, - "loss": 0.9835, + "learning_rate": 9.438897152446192e-05, + "loss": 5.19, "step": 7075 }, { "epoch": 0.57, - "learning_rate": 9.404802648514695e-05, - "loss": 1.0568, + "learning_rate": 9.424954588384923e-05, + "loss": 5.2089, "step": 7080 }, { "epoch": 0.57, - "learning_rate": 9.390847988094456e-05, - "loss": 0.9744, + "learning_rate": 9.411013145802828e-05, + "loss": 5.1439, "step": 7085 }, { "epoch": 0.57, - "learning_rate": 9.376894518208501e-05, - "loss": 1.0413, + "learning_rate": 9.39707285188912e-05, + "loss": 5.2982, "step": 7090 }, { "epoch": 0.57, - "learning_rate": 9.36294226612766e-05, - "loss": 0.9951, + "learning_rate": 9.383133733830784e-05, + "loss": 5.254, "step": 7095 }, { "epoch": 0.57, - "learning_rate": 9.348991259120388e-05, - "loss": 1.0344, + "learning_rate": 9.369195818812506e-05, + "loss": 5.1566, "step": 7100 }, { "epoch": 0.57, - "learning_rate": 9.335041524452712e-05, - "loss": 0.9945, + "learning_rate": 9.355259134016634e-05, + "loss": 5.2448, "step": 7105 }, { "epoch": 0.57, - "learning_rate": 9.321093089388166e-05, - "loss": 1.0498, + "learning_rate": 9.341323706623109e-05, + "loss": 5.1728, "step": 7110 }, { "epoch": 0.57, - "learning_rate": 9.307145981187743e-05, - "loss": 0.9922, + "learning_rate": 9.327389563809418e-05, + "loss": 5.1837, "step": 7115 }, { "epoch": 0.57, - "learning_rate": 9.293200227109845e-05, - "loss": 0.9978, + "learning_rate": 9.31345673275055e-05, + "loss": 5.192, "step": 7120 }, { "epoch": 0.57, - "learning_rate": 9.279255854410226e-05, - "loss": 0.9785, + "learning_rate": 9.299525240618932e-05, + "loss": 5.2193, "step": 7125 }, { "epoch": 0.57, - "learning_rate": 9.265312890341947e-05, - "loss": 0.9269, + "learning_rate": 9.285595114584384e-05, + "loss": 5.1796, "step": 7130 }, { "epoch": 0.57, - "learning_rate": 9.25137136215531e-05, - "loss": 0.9491, + "learning_rate": 9.271666381814055e-05, + "loss": 5.1956, "step": 7135 }, { "epoch": 0.57, - "learning_rate": 9.237431297097805e-05, - "loss": 1.0114, + "learning_rate": 9.257739069472374e-05, + "loss": 5.2215, "step": 7140 }, { "epoch": 0.57, - "learning_rate": 9.223492722414075e-05, - "loss": 0.9774, + "learning_rate": 9.243813204721017e-05, + "loss": 5.108, "step": 7145 }, { "epoch": 0.57, - "learning_rate": 9.209555665345839e-05, - "loss": 0.9742, + "learning_rate": 9.22988881471882e-05, + "loss": 5.2157, "step": 7150 }, { "epoch": 0.57, - "learning_rate": 9.195620153131862e-05, - "loss": 0.9701, + "learning_rate": 9.215965926621747e-05, + "loss": 5.2697, "step": 7155 }, { "epoch": 0.57, - "learning_rate": 9.181686213007878e-05, - "loss": 0.986, + "learning_rate": 9.202044567582833e-05, + "loss": 5.1905, "step": 7160 }, { "epoch": 0.57, - "learning_rate": 9.167753872206553e-05, - "loss": 0.9826, + "learning_rate": 9.188124764752138e-05, + "loss": 5.2076, "step": 7165 }, { "epoch": 0.57, - "learning_rate": 9.153823157957429e-05, - "loss": 0.9523, + "learning_rate": 9.174206545276677e-05, + "loss": 5.2788, "step": 7170 }, { "epoch": 0.57, - "learning_rate": 9.139894097486862e-05, - "loss": 1.0309, + "learning_rate": 9.160289936300384e-05, + "loss": 5.1886, "step": 7175 }, { - "epoch": 0.58, - "learning_rate": 9.12596671801799e-05, - "loss": 1.0126, + "epoch": 0.57, + "learning_rate": 9.146374964964046e-05, + "loss": 5.2315, "step": 7180 }, { - "epoch": 0.58, - "learning_rate": 9.112041046770653e-05, - "loss": 0.9833, + "epoch": 0.57, + "learning_rate": 9.132461658405265e-05, + "loss": 5.2057, "step": 7185 }, { "epoch": 0.58, - "learning_rate": 9.098117110961357e-05, - "loss": 0.9941, + "learning_rate": 9.118550043758387e-05, + "loss": 5.2174, "step": 7190 }, { "epoch": 0.58, - "learning_rate": 9.084194937803214e-05, - "loss": 0.9879, + "learning_rate": 9.104640148154459e-05, + "loss": 5.1193, "step": 7195 }, { "epoch": 0.58, - "learning_rate": 9.07027455450589e-05, - "loss": 0.9501, + "learning_rate": 9.090731998721189e-05, + "loss": 5.2272, "step": 7200 }, { "epoch": 0.58, - "learning_rate": 9.056355988275562e-05, - "loss": 1.0014, + "learning_rate": 9.076825622582864e-05, + "loss": 5.1942, "step": 7205 }, { "epoch": 0.58, - "learning_rate": 9.042439266314848e-05, - "loss": 0.9966, + "learning_rate": 9.062921046860315e-05, + "loss": 5.1867, "step": 7210 }, { "epoch": 0.58, - "learning_rate": 9.02852441582276e-05, - "loss": 0.9661, + "learning_rate": 9.049018298670863e-05, + "loss": 5.2218, "step": 7215 }, { "epoch": 0.58, - "learning_rate": 9.014611463994657e-05, - "loss": 0.9694, + "learning_rate": 9.035117405128275e-05, + "loss": 5.1633, "step": 7220 }, { "epoch": 0.58, - "learning_rate": 9.000700438022179e-05, - "loss": 0.989, + "learning_rate": 9.02121839334269e-05, + "loss": 5.1908, "step": 7225 }, { "epoch": 0.58, - "learning_rate": 8.986791365093221e-05, - "loss": 0.9708, + "learning_rate": 9.007321290420574e-05, + "loss": 5.2676, "step": 7230 }, { "epoch": 0.58, - "learning_rate": 8.97288427239184e-05, - "loss": 0.9615, + "learning_rate": 8.99342612346468e-05, + "loss": 5.2525, "step": 7235 }, { "epoch": 0.58, - "learning_rate": 8.95897918709823e-05, - "loss": 0.9994, + "learning_rate": 8.97953291957398e-05, + "loss": 5.2256, "step": 7240 }, { "epoch": 0.58, - "learning_rate": 8.945076136388671e-05, - "loss": 1.0126, + "learning_rate": 8.965641705843624e-05, + "loss": 5.1541, "step": 7245 }, { "epoch": 0.58, - "learning_rate": 8.931175147435445e-05, - "loss": 0.9668, + "learning_rate": 8.951752509364873e-05, + "loss": 5.234, "step": 7250 }, { "epoch": 0.58, - "learning_rate": 8.917276247406833e-05, - "loss": 1.0543, + "learning_rate": 8.937865357225048e-05, + "loss": 5.1728, "step": 7255 }, { "epoch": 0.58, - "learning_rate": 8.903379463467008e-05, - "loss": 1.0182, + "learning_rate": 8.923980276507504e-05, + "loss": 5.245, "step": 7260 }, { "epoch": 0.58, - "learning_rate": 8.889484822776024e-05, - "loss": 0.9943, + "learning_rate": 8.910097294291539e-05, + "loss": 5.2453, "step": 7265 }, { "epoch": 0.58, - "learning_rate": 8.875592352489737e-05, - "loss": 0.9826, + "learning_rate": 8.896216437652356e-05, + "loss": 5.2644, "step": 7270 }, { "epoch": 0.58, - "learning_rate": 8.86170207975976e-05, - "loss": 0.9605, + "learning_rate": 8.882337733661032e-05, + "loss": 5.1691, "step": 7275 }, { "epoch": 0.58, - "learning_rate": 8.847814031733422e-05, - "loss": 1.0122, + "learning_rate": 8.868461209384427e-05, + "loss": 5.1859, "step": 7280 }, { "epoch": 0.58, - "learning_rate": 8.833928235553694e-05, - "loss": 0.9643, + "learning_rate": 8.854586891885154e-05, + "loss": 5.1763, "step": 7285 }, { "epoch": 0.58, - "learning_rate": 8.82004471835915e-05, - "loss": 1.0212, + "learning_rate": 8.84071480822153e-05, + "loss": 5.2144, "step": 7290 }, { "epoch": 0.58, - "learning_rate": 8.80616350728391e-05, - "loss": 0.96, + "learning_rate": 8.826844985447507e-05, + "loss": 5.1855, "step": 7295 }, { "epoch": 0.58, - "learning_rate": 8.792284629457576e-05, - "loss": 0.9971, + "learning_rate": 8.812977450612634e-05, + "loss": 5.137, "step": 7300 }, { - "epoch": 0.59, - "learning_rate": 8.778408112005212e-05, - "loss": 1.0003, + "epoch": 0.58, + "learning_rate": 8.799112230761993e-05, + "loss": 5.0849, "step": 7305 }, { - "epoch": 0.59, - "learning_rate": 8.764533982047255e-05, - "loss": 1.0481, + "epoch": 0.58, + "learning_rate": 8.78524935293615e-05, + "loss": 5.1987, "step": 7310 }, { "epoch": 0.59, - "learning_rate": 8.750662266699475e-05, - "loss": 0.9789, + "learning_rate": 8.771388844171113e-05, + "loss": 5.269, "step": 7315 }, { "epoch": 0.59, - "learning_rate": 8.736792993072921e-05, - "loss": 0.9787, + "learning_rate": 8.757530731498261e-05, + "loss": 5.2574, "step": 7320 }, { "epoch": 0.59, - "learning_rate": 8.722926188273879e-05, - "loss": 0.9908, + "learning_rate": 8.743675041944301e-05, + "loss": 5.1733, "step": 7325 }, { "epoch": 0.59, - "learning_rate": 8.709061879403806e-05, - "loss": 0.9774, + "learning_rate": 8.729821802531212e-05, + "loss": 5.1723, "step": 7330 }, { "epoch": 0.59, - "learning_rate": 8.69520009355928e-05, - "loss": 0.9908, + "learning_rate": 8.715971040276209e-05, + "loss": 5.1919, "step": 7335 }, { "epoch": 0.59, - "learning_rate": 8.681340857831944e-05, - "loss": 0.9925, + "learning_rate": 8.702122782191656e-05, + "loss": 5.1621, "step": 7340 }, { "epoch": 0.59, - "learning_rate": 8.667484199308465e-05, - "loss": 1.0515, + "learning_rate": 8.688277055285043e-05, + "loss": 5.1583, "step": 7345 }, { "epoch": 0.59, - "learning_rate": 8.653630145070462e-05, - "loss": 0.9797, + "learning_rate": 8.674433886558928e-05, + "loss": 5.2878, "step": 7350 }, { "epoch": 0.59, - "learning_rate": 8.639778722194481e-05, - "loss": 1.0064, + "learning_rate": 8.660593303010876e-05, + "loss": 5.1771, "step": 7355 }, { "epoch": 0.59, - "learning_rate": 8.625929957751908e-05, - "loss": 0.9866, + "learning_rate": 8.646755331633403e-05, + "loss": 5.1509, "step": 7360 }, { "epoch": 0.59, - "learning_rate": 8.612083878808946e-05, - "loss": 1.0308, + "learning_rate": 8.632919999413939e-05, + "loss": 5.2252, "step": 7365 }, { "epoch": 0.59, - "learning_rate": 8.598240512426542e-05, - "loss": 0.9913, + "learning_rate": 8.619087333334768e-05, + "loss": 5.2268, "step": 7370 }, { "epoch": 0.59, - "learning_rate": 8.58439988566034e-05, - "loss": 0.9912, + "learning_rate": 8.605257360372971e-05, + "loss": 5.2919, "step": 7375 }, { "epoch": 0.59, - "learning_rate": 8.570562025560642e-05, - "loss": 0.9643, + "learning_rate": 8.591430107500376e-05, + "loss": 5.1697, "step": 7380 }, { "epoch": 0.59, - "learning_rate": 8.556726959172326e-05, - "loss": 1.0119, + "learning_rate": 8.577605601683501e-05, + "loss": 5.2065, "step": 7385 }, { "epoch": 0.59, - "learning_rate": 8.542894713534821e-05, - "loss": 0.9721, + "learning_rate": 8.563783869883523e-05, + "loss": 5.1608, "step": 7390 }, { "epoch": 0.59, - "learning_rate": 8.529065315682041e-05, - "loss": 0.9903, + "learning_rate": 8.549964939056194e-05, + "loss": 5.2227, "step": 7395 }, { "epoch": 0.59, - "learning_rate": 8.51523879264233e-05, - "loss": 1.0464, + "learning_rate": 8.536148836151807e-05, + "loss": 5.2012, "step": 7400 }, { "epoch": 0.59, - "learning_rate": 8.501415171438423e-05, - "loss": 1.0484, + "learning_rate": 8.522335588115143e-05, + "loss": 5.1847, "step": 7405 }, { "epoch": 0.59, - "learning_rate": 8.487594479087374e-05, - "loss": 0.9891, + "learning_rate": 8.508525221885414e-05, + "loss": 5.2634, "step": 7410 }, { "epoch": 0.59, - "learning_rate": 8.473776742600516e-05, - "loss": 0.9739, + "learning_rate": 8.494717764396212e-05, + "loss": 5.1305, "step": 7415 }, { "epoch": 0.59, - "learning_rate": 8.459961988983407e-05, - "loss": 0.9437, + "learning_rate": 8.480913242575452e-05, + "loss": 5.208, "step": 7420 }, { "epoch": 0.59, - "learning_rate": 8.446150245235766e-05, - "loss": 0.9939, + "learning_rate": 8.467111683345326e-05, + "loss": 5.2063, "step": 7425 }, { - "epoch": 0.6, - "learning_rate": 8.43234153835145e-05, - "loss": 1.0034, + "epoch": 0.59, + "learning_rate": 8.453313113622257e-05, + "loss": 5.2263, "step": 7430 }, { - "epoch": 0.6, - "learning_rate": 8.41853589531836e-05, - "loss": 1.0366, + "epoch": 0.59, + "learning_rate": 8.439517560316824e-05, + "loss": 5.1407, "step": 7435 }, { "epoch": 0.6, - "learning_rate": 8.404733343118417e-05, - "loss": 0.9527, + "learning_rate": 8.425725050333727e-05, + "loss": 5.1109, "step": 7440 }, { "epoch": 0.6, - "learning_rate": 8.390933908727504e-05, - "loss": 1.0477, + "learning_rate": 8.411935610571742e-05, + "loss": 5.1655, "step": 7445 }, { "epoch": 0.6, - "learning_rate": 8.3771376191154e-05, - "loss": 1.0029, + "learning_rate": 8.398149267923643e-05, + "loss": 5.1677, "step": 7450 }, { "epoch": 0.6, - "learning_rate": 8.363344501245755e-05, - "loss": 0.9906, + "learning_rate": 8.384366049276167e-05, + "loss": 5.2386, "step": 7455 }, { "epoch": 0.6, - "learning_rate": 8.349554582076005e-05, - "loss": 1.0301, + "learning_rate": 8.370585981509962e-05, + "loss": 5.154, "step": 7460 }, { "epoch": 0.6, - "learning_rate": 8.335767888557342e-05, - "loss": 0.975, + "learning_rate": 8.356809091499534e-05, + "loss": 5.2002, "step": 7465 }, { "epoch": 0.6, - "learning_rate": 8.321984447634651e-05, - "loss": 0.9533, + "learning_rate": 8.343035406113185e-05, + "loss": 5.1174, "step": 7470 }, { "epoch": 0.6, - "learning_rate": 8.308204286246454e-05, - "loss": 0.9895, + "learning_rate": 8.329264952212969e-05, + "loss": 5.2218, "step": 7475 }, { "epoch": 0.6, - "learning_rate": 8.29442743132488e-05, - "loss": 0.9387, + "learning_rate": 8.315497756654633e-05, + "loss": 5.2253, "step": 7480 }, { "epoch": 0.6, - "learning_rate": 8.280653909795584e-05, - "loss": 0.937, + "learning_rate": 8.301733846287586e-05, + "loss": 5.2528, "step": 7485 }, { "epoch": 0.6, - "learning_rate": 8.266883748577707e-05, - "loss": 1.0151, + "learning_rate": 8.287973247954813e-05, + "loss": 5.1225, "step": 7490 }, { "epoch": 0.6, - "learning_rate": 8.25311697458382e-05, - "loss": 0.9679, + "learning_rate": 8.274215988492846e-05, + "loss": 5.2537, "step": 7495 }, { "epoch": 0.6, - "learning_rate": 8.239353614719879e-05, - "loss": 0.9468, + "learning_rate": 8.2604620947317e-05, + "loss": 5.0768, "step": 7500 }, { "epoch": 0.6, - "learning_rate": 8.225593695885168e-05, - "loss": 1.0078, + "learning_rate": 8.24671159349484e-05, + "loss": 5.1271, "step": 7505 }, { "epoch": 0.6, - "learning_rate": 8.211837244972247e-05, - "loss": 1.0258, + "learning_rate": 8.2329645115991e-05, + "loss": 5.1748, "step": 7510 }, { "epoch": 0.6, - "learning_rate": 8.19808428886689e-05, - "loss": 0.9323, + "learning_rate": 8.219220875854652e-05, + "loss": 5.1716, "step": 7515 }, { "epoch": 0.6, - "learning_rate": 8.184334854448044e-05, - "loss": 1.0245, + "learning_rate": 8.205480713064946e-05, + "loss": 5.184, "step": 7520 }, { "epoch": 0.6, - "learning_rate": 8.170588968587776e-05, - "loss": 0.9635, + "learning_rate": 8.191744050026666e-05, + "loss": 5.0983, "step": 7525 }, { "epoch": 0.6, - "learning_rate": 8.156846658151223e-05, - "loss": 1.0137, + "learning_rate": 8.178010913529655e-05, + "loss": 5.185, "step": 7530 }, { "epoch": 0.6, - "learning_rate": 8.143107949996522e-05, - "loss": 0.9768, + "learning_rate": 8.164281330356889e-05, + "loss": 5.0998, "step": 7535 }, { "epoch": 0.6, - "learning_rate": 8.129372870974775e-05, - "loss": 1.0253, + "learning_rate": 8.150555327284417e-05, + "loss": 5.2174, "step": 7540 }, { "epoch": 0.6, - "learning_rate": 8.115641447929996e-05, - "loss": 0.9952, + "learning_rate": 8.136832931081299e-05, + "loss": 5.1611, "step": 7545 }, { "epoch": 0.6, - "learning_rate": 8.10191370769904e-05, - "loss": 0.9658, + "learning_rate": 8.123114168509563e-05, + "loss": 5.23, "step": 7550 }, { - "epoch": 0.61, - "learning_rate": 8.088189677111586e-05, - "loss": 1.0142, + "epoch": 0.6, + "learning_rate": 8.109399066324145e-05, + "loss": 5.0806, "step": 7555 }, { - "epoch": 0.61, - "learning_rate": 8.074469382990042e-05, - "loss": 0.9707, + "epoch": 0.6, + "learning_rate": 8.095687651272858e-05, + "loss": 5.1828, "step": 7560 }, { "epoch": 0.61, - "learning_rate": 8.060752852149524e-05, - "loss": 0.9923, + "learning_rate": 8.081979950096311e-05, + "loss": 5.1121, "step": 7565 }, { "epoch": 0.61, - "learning_rate": 8.047040111397794e-05, - "loss": 1.0042, + "learning_rate": 8.06827598952787e-05, + "loss": 5.0954, "step": 7570 }, { "epoch": 0.61, - "learning_rate": 8.033331187535194e-05, - "loss": 0.9887, + "learning_rate": 8.054575796293609e-05, + "loss": 5.1819, "step": 7575 }, { "epoch": 0.61, - "learning_rate": 8.019626107354623e-05, - "loss": 0.9664, + "learning_rate": 8.040879397112257e-05, + "loss": 5.1769, "step": 7580 }, { "epoch": 0.61, - "learning_rate": 8.005924897641461e-05, - "loss": 0.9651, + "learning_rate": 8.027186818695143e-05, + "loss": 5.1569, "step": 7585 }, { "epoch": 0.61, - "learning_rate": 7.992227585173522e-05, - "loss": 1.0394, + "learning_rate": 8.013498087746142e-05, + "loss": 5.1406, "step": 7590 }, { "epoch": 0.61, - "learning_rate": 7.978534196721001e-05, - "loss": 0.9881, + "learning_rate": 7.999813230961623e-05, + "loss": 5.1468, "step": 7595 }, { "epoch": 0.61, - "learning_rate": 7.964844759046427e-05, - "loss": 1.0275, + "learning_rate": 7.986132275030409e-05, + "loss": 5.1436, "step": 7600 }, { "epoch": 0.61, - "learning_rate": 7.951159298904614e-05, - "loss": 1.0433, + "learning_rate": 7.972455246633708e-05, + "loss": 5.2137, "step": 7605 }, { "epoch": 0.61, - "learning_rate": 7.937477843042592e-05, - "loss": 0.9791, + "learning_rate": 7.958782172445063e-05, + "loss": 5.1435, "step": 7610 }, { "epoch": 0.61, - "learning_rate": 7.923800418199564e-05, - "loss": 0.9579, + "learning_rate": 7.945113079130323e-05, + "loss": 5.1857, "step": 7615 }, { "epoch": 0.61, - "learning_rate": 7.910127051106866e-05, - "loss": 1.0581, + "learning_rate": 7.931447993347554e-05, + "loss": 5.0497, "step": 7620 }, { "epoch": 0.61, - "learning_rate": 7.896457768487892e-05, - "loss": 1.0168, + "learning_rate": 7.917786941747018e-05, + "loss": 5.1438, "step": 7625 }, { "epoch": 0.61, - "learning_rate": 7.882792597058066e-05, - "loss": 1.0198, + "learning_rate": 7.904129950971105e-05, + "loss": 5.1055, "step": 7630 }, { "epoch": 0.61, - "learning_rate": 7.869131563524763e-05, - "loss": 0.971, + "learning_rate": 7.890477047654288e-05, + "loss": 5.1625, "step": 7635 }, { "epoch": 0.61, - "learning_rate": 7.855474694587276e-05, - "loss": 1.0065, + "learning_rate": 7.876828258423068e-05, + "loss": 5.224, "step": 7640 }, { "epoch": 0.61, - "learning_rate": 7.841822016936763e-05, - "loss": 0.9909, + "learning_rate": 7.86318360989592e-05, + "loss": 5.0788, "step": 7645 }, { "epoch": 0.61, - "learning_rate": 7.828173557256184e-05, - "loss": 0.9727, + "learning_rate": 7.849543128683242e-05, + "loss": 5.1351, "step": 7650 }, { "epoch": 0.61, - "learning_rate": 7.814529342220266e-05, - "loss": 1.031, + "learning_rate": 7.835906841387314e-05, + "loss": 5.1622, "step": 7655 }, { "epoch": 0.61, - "learning_rate": 7.800889398495425e-05, - "loss": 0.9861, + "learning_rate": 7.822274774602228e-05, + "loss": 5.168, "step": 7660 }, { "epoch": 0.61, - "learning_rate": 7.787253752739742e-05, - "loss": 1.0229, + "learning_rate": 7.808646954913847e-05, + "loss": 5.1767, "step": 7665 }, { "epoch": 0.61, - "learning_rate": 7.773622431602889e-05, - "loss": 0.9508, + "learning_rate": 7.795023408899749e-05, + "loss": 5.1445, "step": 7670 }, { "epoch": 0.61, - "learning_rate": 7.759995461726089e-05, - "loss": 0.9578, + "learning_rate": 7.781404163129186e-05, + "loss": 5.1528, "step": 7675 }, { - "epoch": 0.62, - "learning_rate": 7.746372869742067e-05, - "loss": 0.9634, + "epoch": 0.61, + "learning_rate": 7.767789244163013e-05, + "loss": 5.1595, "step": 7680 }, { - "epoch": 0.62, - "learning_rate": 7.732754682274985e-05, - "loss": 0.9982, + "epoch": 0.61, + "learning_rate": 7.754178678553652e-05, + "loss": 5.1802, "step": 7685 }, { "epoch": 0.62, - "learning_rate": 7.719140925940397e-05, - "loss": 0.9798, + "learning_rate": 7.740572492845038e-05, + "loss": 5.238, "step": 7690 }, { "epoch": 0.62, - "learning_rate": 7.705531627345197e-05, - "loss": 1.0179, + "learning_rate": 7.726970713572555e-05, + "loss": 5.1063, "step": 7695 }, { "epoch": 0.62, - "learning_rate": 7.691926813087564e-05, - "loss": 0.99, + "learning_rate": 7.713373367263004e-05, + "loss": 5.1408, "step": 7700 }, { "epoch": 0.62, - "learning_rate": 7.678326509756928e-05, - "loss": 0.968, + "learning_rate": 7.699780480434526e-05, + "loss": 5.1143, "step": 7705 }, { "epoch": 0.62, - "learning_rate": 7.664730743933884e-05, - "loss": 0.9867, + "learning_rate": 7.686192079596586e-05, + "loss": 5.1146, "step": 7710 }, { "epoch": 0.62, - "learning_rate": 7.651139542190164e-05, - "loss": 0.9621, + "learning_rate": 7.672608191249882e-05, + "loss": 5.1223, "step": 7715 }, { "epoch": 0.62, - "learning_rate": 7.637552931088587e-05, - "loss": 0.9967, + "learning_rate": 7.659028841886317e-05, + "loss": 5.1203, "step": 7720 }, { "epoch": 0.62, - "learning_rate": 7.623970937182987e-05, - "loss": 1.0006, + "learning_rate": 7.645454057988942e-05, + "loss": 5.1058, "step": 7725 }, { "epoch": 0.62, - "learning_rate": 7.610393587018194e-05, - "loss": 0.9694, + "learning_rate": 7.631883866031911e-05, + "loss": 5.1465, "step": 7730 }, { "epoch": 0.62, - "learning_rate": 7.59682090712994e-05, - "loss": 0.9554, + "learning_rate": 7.618318292480413e-05, + "loss": 5.1471, "step": 7735 }, { "epoch": 0.62, - "learning_rate": 7.583252924044846e-05, - "loss": 1.002, + "learning_rate": 7.604757363790631e-05, + "loss": 5.1289, "step": 7740 }, { "epoch": 0.62, - "learning_rate": 7.569689664280346e-05, - "loss": 0.9721, + "learning_rate": 7.591201106409696e-05, + "loss": 5.2459, "step": 7745 }, { "epoch": 0.62, - "learning_rate": 7.556131154344639e-05, - "loss": 1.0177, + "learning_rate": 7.577649546775622e-05, + "loss": 5.1918, "step": 7750 }, { "epoch": 0.62, - "learning_rate": 7.542577420736651e-05, - "loss": 0.9878, + "learning_rate": 7.564102711317267e-05, + "loss": 5.0788, "step": 7755 }, { "epoch": 0.62, - "learning_rate": 7.529028489945966e-05, - "loss": 0.991, + "learning_rate": 7.550560626454272e-05, + "loss": 5.1247, "step": 7760 }, { "epoch": 0.62, - "learning_rate": 7.515484388452784e-05, - "loss": 0.959, + "learning_rate": 7.537023318597009e-05, + "loss": 5.16, "step": 7765 }, { "epoch": 0.62, - "learning_rate": 7.501945142727866e-05, - "loss": 0.9943, + "learning_rate": 7.523490814146546e-05, + "loss": 5.1762, "step": 7770 }, { "epoch": 0.62, - "learning_rate": 7.488410779232477e-05, - "loss": 1.0233, + "learning_rate": 7.509963139494573e-05, + "loss": 5.1602, "step": 7775 }, { "epoch": 0.62, - "learning_rate": 7.474881324418354e-05, - "loss": 0.9885, + "learning_rate": 7.496440321023359e-05, + "loss": 5.0242, "step": 7780 }, { "epoch": 0.62, - "learning_rate": 7.461356804727629e-05, - "loss": 1.0214, + "learning_rate": 7.482922385105713e-05, + "loss": 5.164, "step": 7785 }, { "epoch": 0.62, - "learning_rate": 7.44783724659279e-05, - "loss": 0.984, + "learning_rate": 7.469409358104915e-05, + "loss": 5.1561, "step": 7790 }, { "epoch": 0.62, - "learning_rate": 7.43432267643663e-05, - "loss": 0.92, + "learning_rate": 7.455901266374667e-05, + "loss": 5.0976, "step": 7795 }, { "epoch": 0.62, - "learning_rate": 7.420813120672191e-05, - "loss": 0.9549, + "learning_rate": 7.442398136259054e-05, + "loss": 5.1864, "step": 7800 }, { - "epoch": 0.63, - "learning_rate": 7.407308605702725e-05, - "loss": 0.9376, + "epoch": 0.62, + "learning_rate": 7.428899994092483e-05, + "loss": 5.0696, "step": 7805 }, { - "epoch": 0.63, - "learning_rate": 7.393809157921618e-05, - "loss": 0.9656, + "epoch": 0.62, + "learning_rate": 7.41540686619963e-05, + "loss": 5.1635, "step": 7810 }, { "epoch": 0.63, - "learning_rate": 7.38031480371236e-05, - "loss": 0.9937, + "learning_rate": 7.401918778895395e-05, + "loss": 5.0541, "step": 7815 }, { "epoch": 0.63, - "learning_rate": 7.366825569448487e-05, - "loss": 0.9677, + "learning_rate": 7.388435758484844e-05, + "loss": 5.1099, "step": 7820 }, { "epoch": 0.63, - "learning_rate": 7.353341481493521e-05, - "loss": 0.9915, + "learning_rate": 7.374957831263167e-05, + "loss": 5.0833, "step": 7825 }, { "epoch": 0.63, - "learning_rate": 7.339862566200939e-05, - "loss": 0.9187, + "learning_rate": 7.361485023515618e-05, + "loss": 5.1226, "step": 7830 }, { "epoch": 0.63, - "learning_rate": 7.326388849914095e-05, - "loss": 1.0035, + "learning_rate": 7.348017361517465e-05, + "loss": 5.0266, "step": 7835 }, { "epoch": 0.63, - "learning_rate": 7.312920358966193e-05, - "loss": 1.0323, + "learning_rate": 7.334554871533938e-05, + "loss": 5.2163, "step": 7840 }, { "epoch": 0.63, - "learning_rate": 7.299457119680216e-05, - "loss": 0.9912, + "learning_rate": 7.321097579820192e-05, + "loss": 5.1839, "step": 7845 }, { "epoch": 0.63, - "learning_rate": 7.285999158368883e-05, - "loss": 0.9178, + "learning_rate": 7.307645512621233e-05, + "loss": 5.2666, "step": 7850 }, { "epoch": 0.63, - "learning_rate": 7.272546501334611e-05, - "loss": 1.0096, + "learning_rate": 7.29419869617188e-05, + "loss": 5.13, "step": 7855 }, { "epoch": 0.63, - "learning_rate": 7.259099174869432e-05, - "loss": 0.9555, + "learning_rate": 7.280757156696719e-05, + "loss": 5.1656, "step": 7860 }, { "epoch": 0.63, - "learning_rate": 7.245657205254974e-05, - "loss": 0.9608, + "learning_rate": 7.26732092041003e-05, + "loss": 5.1167, "step": 7865 }, { "epoch": 0.63, - "learning_rate": 7.232220618762386e-05, - "loss": 0.9682, + "learning_rate": 7.253890013515767e-05, + "loss": 5.0893, "step": 7870 }, { "epoch": 0.63, - "learning_rate": 7.2187894416523e-05, - "loss": 1.0021, + "learning_rate": 7.240464462207475e-05, + "loss": 5.1961, "step": 7875 }, { "epoch": 0.63, - "learning_rate": 7.205363700174779e-05, - "loss": 1.0134, + "learning_rate": 7.227044292668273e-05, + "loss": 5.1522, "step": 7880 }, { "epoch": 0.63, - "learning_rate": 7.191943420569259e-05, - "loss": 1.0462, + "learning_rate": 7.213629531070764e-05, + "loss": 5.112, "step": 7885 }, { "epoch": 0.63, - "learning_rate": 7.178528629064498e-05, - "loss": 0.9897, + "learning_rate": 7.200220203577017e-05, + "loss": 5.0136, "step": 7890 }, { "epoch": 0.63, - "learning_rate": 7.165119351878533e-05, - "loss": 0.9915, + "learning_rate": 7.186816336338495e-05, + "loss": 5.1022, "step": 7895 }, { "epoch": 0.63, - "learning_rate": 7.151715615218622e-05, - "loss": 1.0136, + "learning_rate": 7.173417955496024e-05, + "loss": 5.0693, "step": 7900 }, { "epoch": 0.63, - "learning_rate": 7.1383174452812e-05, - "loss": 0.9518, + "learning_rate": 7.16002508717972e-05, + "loss": 5.184, "step": 7905 }, { "epoch": 0.63, - "learning_rate": 7.124924868251815e-05, - "loss": 0.9564, + "learning_rate": 7.146637757508949e-05, + "loss": 5.1052, "step": 7910 }, { "epoch": 0.63, - "learning_rate": 7.111537910305079e-05, - "loss": 0.9621, + "learning_rate": 7.133255992592278e-05, + "loss": 5.1897, "step": 7915 }, { "epoch": 0.63, - "learning_rate": 7.098156597604641e-05, - "loss": 0.9856, + "learning_rate": 7.119879818527421e-05, + "loss": 5.0659, "step": 7920 }, { "epoch": 0.63, - "learning_rate": 7.084780956303095e-05, - "loss": 1.0061, + "learning_rate": 7.106509261401194e-05, + "loss": 5.0322, "step": 7925 }, { - "epoch": 0.64, - "learning_rate": 7.071411012541968e-05, - "loss": 1.0187, + "epoch": 0.63, + "learning_rate": 7.093144347289446e-05, + "loss": 5.1435, "step": 7930 }, { - "epoch": 0.64, - "learning_rate": 7.05804679245164e-05, - "loss": 1.0241, + "epoch": 0.63, + "learning_rate": 7.079785102257028e-05, + "loss": 5.1117, "step": 7935 }, { "epoch": 0.64, - "learning_rate": 7.044688322151313e-05, - "loss": 0.9717, + "learning_rate": 7.066431552357737e-05, + "loss": 5.092, "step": 7940 }, { "epoch": 0.64, - "learning_rate": 7.031335627748947e-05, - "loss": 0.9914, + "learning_rate": 7.053083723634262e-05, + "loss": 5.0251, "step": 7945 }, { "epoch": 0.64, - "learning_rate": 7.01798873534121e-05, - "loss": 0.9919, + "learning_rate": 7.039741642118125e-05, + "loss": 5.1936, "step": 7950 }, { "epoch": 0.64, - "learning_rate": 7.004647671013441e-05, - "loss": 0.9662, + "learning_rate": 7.026405333829659e-05, + "loss": 5.0598, "step": 7955 }, { "epoch": 0.64, - "learning_rate": 6.991312460839579e-05, - "loss": 1.0128, + "learning_rate": 7.013074824777919e-05, + "loss": 5.1014, "step": 7960 }, { "epoch": 0.64, - "learning_rate": 6.977983130882127e-05, - "loss": 1.0064, + "learning_rate": 6.999750140960658e-05, + "loss": 5.1363, "step": 7965 }, { "epoch": 0.64, - "learning_rate": 6.96465970719209e-05, - "loss": 1.0044, + "learning_rate": 6.986431308364268e-05, + "loss": 5.1381, "step": 7970 }, { "epoch": 0.64, - "learning_rate": 6.951342215808935e-05, - "loss": 0.9848, + "learning_rate": 6.97311835296373e-05, + "loss": 5.0667, "step": 7975 }, { "epoch": 0.64, - "learning_rate": 6.938030682760532e-05, - "loss": 0.9924, + "learning_rate": 6.959811300722564e-05, + "loss": 5.14, "step": 7980 }, { "epoch": 0.64, - "learning_rate": 6.92472513406311e-05, - "loss": 1.0566, + "learning_rate": 6.946510177592775e-05, + "loss": 5.0731, "step": 7985 }, { "epoch": 0.64, - "learning_rate": 6.9114255957212e-05, - "loss": 0.9889, + "learning_rate": 6.9332150095148e-05, + "loss": 5.0967, "step": 7990 }, { "epoch": 0.64, - "learning_rate": 6.898132093727579e-05, - "loss": 0.953, + "learning_rate": 6.919925822417476e-05, + "loss": 5.0922, "step": 7995 }, { "epoch": 0.64, - "learning_rate": 6.884844654063238e-05, - "loss": 1.0166, + "learning_rate": 6.906642642217965e-05, + "loss": 5.1258, "step": 8000 }, { "epoch": 0.64, - "learning_rate": 6.871563302697311e-05, - "loss": 1.0436, + "learning_rate": 6.893365494821715e-05, + "loss": 5.0984, "step": 8005 }, { "epoch": 0.64, - "learning_rate": 6.858288065587041e-05, - "loss": 0.9539, + "learning_rate": 6.880094406122407e-05, + "loss": 5.1679, "step": 8010 }, { "epoch": 0.64, - "learning_rate": 6.845018968677715e-05, - "loss": 0.9783, + "learning_rate": 6.866829402001916e-05, + "loss": 5.1904, "step": 8015 }, { "epoch": 0.64, - "learning_rate": 6.831756037902623e-05, - "loss": 0.9968, + "learning_rate": 6.853570508330239e-05, + "loss": 5.1267, "step": 8020 }, { "epoch": 0.64, - "learning_rate": 6.818499299183e-05, - "loss": 0.989, + "learning_rate": 6.840317750965463e-05, + "loss": 5.1585, "step": 8025 }, { "epoch": 0.64, - "learning_rate": 6.805248778427976e-05, - "loss": 0.9989, + "learning_rate": 6.827071155753706e-05, + "loss": 5.1061, "step": 8030 }, { "epoch": 0.64, - "learning_rate": 6.792004501534541e-05, - "loss": 0.9403, + "learning_rate": 6.813830748529064e-05, + "loss": 5.1471, "step": 8035 }, { "epoch": 0.64, - "learning_rate": 6.778766494387476e-05, - "loss": 0.983, + "learning_rate": 6.800596555113575e-05, + "loss": 5.018, "step": 8040 }, { "epoch": 0.64, - "learning_rate": 6.765534782859304e-05, - "loss": 1.0021, + "learning_rate": 6.787368601317142e-05, + "loss": 5.1719, "step": 8045 }, { "epoch": 0.64, - "learning_rate": 6.752309392810245e-05, - "loss": 0.9527, + "learning_rate": 6.774146912937523e-05, + "loss": 4.9763, "step": 8050 }, { - "epoch": 0.65, - "learning_rate": 6.739090350088166e-05, - "loss": 1.0406, + "epoch": 0.64, + "learning_rate": 6.760931515760234e-05, + "loss": 5.1246, "step": 8055 }, { - "epoch": 0.65, - "learning_rate": 6.725877680528534e-05, - "loss": 0.9607, + "epoch": 0.64, + "learning_rate": 6.747722435558534e-05, + "loss": 5.1193, "step": 8060 }, { "epoch": 0.65, - "learning_rate": 6.712671409954353e-05, - "loss": 0.9894, + "learning_rate": 6.734519698093357e-05, + "loss": 5.1495, "step": 8065 }, { "epoch": 0.65, - "learning_rate": 6.699471564176119e-05, - "loss": 0.9848, + "learning_rate": 6.721323329113273e-05, + "loss": 5.0837, "step": 8070 }, { "epoch": 0.65, - "learning_rate": 6.68627816899178e-05, - "loss": 0.9542, + "learning_rate": 6.708133354354428e-05, + "loss": 5.1214, "step": 8075 }, { "epoch": 0.65, - "learning_rate": 6.673091250186665e-05, - "loss": 0.9903, + "learning_rate": 6.694949799540496e-05, + "loss": 5.0417, "step": 8080 }, { "epoch": 0.65, - "learning_rate": 6.65991083353346e-05, - "loss": 1.0099, + "learning_rate": 6.681772690382632e-05, + "loss": 5.1095, "step": 8085 }, { "epoch": 0.65, - "learning_rate": 6.646736944792132e-05, - "loss": 0.9383, + "learning_rate": 6.668602052579424e-05, + "loss": 5.0624, "step": 8090 }, { "epoch": 0.65, - "learning_rate": 6.633569609709896e-05, - "loss": 0.9966, + "learning_rate": 6.655437911816838e-05, + "loss": 5.1461, "step": 8095 }, { "epoch": 0.65, - "learning_rate": 6.620408854021152e-05, - "loss": 0.9414, + "learning_rate": 6.642280293768162e-05, + "loss": 5.0446, "step": 8100 }, { "epoch": 0.65, - "learning_rate": 6.607254703447443e-05, - "loss": 1.0145, + "learning_rate": 6.629129224093968e-05, + "loss": 5.1247, "step": 8105 }, { "epoch": 0.65, - "learning_rate": 6.594107183697411e-05, - "loss": 1.0309, + "learning_rate": 6.615984728442064e-05, + "loss": 5.1121, "step": 8110 }, { "epoch": 0.65, - "learning_rate": 6.580966320466728e-05, - "loss": 1.011, + "learning_rate": 6.602846832447429e-05, + "loss": 5.0423, "step": 8115 }, { "epoch": 0.65, - "learning_rate": 6.567832139438064e-05, - "loss": 1.0131, + "learning_rate": 6.589715561732166e-05, + "loss": 5.1175, "step": 8120 }, { "epoch": 0.65, - "learning_rate": 6.554704666281025e-05, - "loss": 0.9643, + "learning_rate": 6.576590941905474e-05, + "loss": 5.1143, "step": 8125 }, { "epoch": 0.65, - "learning_rate": 6.541583926652102e-05, - "loss": 1.0008, + "learning_rate": 6.563472998563566e-05, + "loss": 5.1638, "step": 8130 }, { "epoch": 0.65, - "learning_rate": 6.52846994619464e-05, - "loss": 1.031, + "learning_rate": 6.550361757289639e-05, + "loss": 5.1362, "step": 8135 }, { "epoch": 0.65, - "learning_rate": 6.515362750538766e-05, - "loss": 0.986, + "learning_rate": 6.537257243653816e-05, + "loss": 5.057, "step": 8140 }, { "epoch": 0.65, - "learning_rate": 6.502262365301343e-05, - "loss": 0.9717, + "learning_rate": 6.52415948321311e-05, + "loss": 5.0888, "step": 8145 }, { "epoch": 0.65, - "learning_rate": 6.489168816085924e-05, - "loss": 1.0134, + "learning_rate": 6.511068501511355e-05, + "loss": 5.1049, "step": 8150 }, { "epoch": 0.65, - "learning_rate": 6.476082128482709e-05, - "loss": 0.9868, + "learning_rate": 6.497984324079163e-05, + "loss": 5.2026, "step": 8155 }, { "epoch": 0.65, - "learning_rate": 6.463002328068486e-05, - "loss": 0.9654, + "learning_rate": 6.48490697643388e-05, + "loss": 5.1499, "step": 8160 }, { "epoch": 0.65, - "learning_rate": 6.449929440406575e-05, - "loss": 0.9674, + "learning_rate": 6.471836484079536e-05, + "loss": 5.0608, "step": 8165 }, { "epoch": 0.65, - "learning_rate": 6.436863491046791e-05, - "loss": 1.0724, + "learning_rate": 6.458772872506785e-05, + "loss": 5.1372, "step": 8170 }, { "epoch": 0.65, - "learning_rate": 6.42380450552539e-05, - "loss": 0.996, + "learning_rate": 6.445716167192862e-05, + "loss": 5.164, "step": 8175 }, { - "epoch": 0.66, - "learning_rate": 6.410752509365014e-05, - "loss": 0.9777, + "epoch": 0.65, + "learning_rate": 6.432666393601534e-05, + "loss": 5.1156, "step": 8180 }, { - "epoch": 0.66, - "learning_rate": 6.39770752807465e-05, - "loss": 1.0446, + "epoch": 0.65, + "learning_rate": 6.419623577183056e-05, + "loss": 5.0476, "step": 8185 }, { "epoch": 0.66, - "learning_rate": 6.384669587149568e-05, - "loss": 1.002, + "learning_rate": 6.406587743374104e-05, + "loss": 5.0751, "step": 8190 }, { "epoch": 0.66, - "learning_rate": 6.371638712071287e-05, - "loss": 0.9678, + "learning_rate": 6.393558917597745e-05, + "loss": 5.1216, "step": 8195 }, { "epoch": 0.66, - "learning_rate": 6.358614928307509e-05, - "loss": 1.0324, + "learning_rate": 6.380537125263375e-05, + "loss": 5.1384, "step": 8200 }, { "epoch": 0.66, - "learning_rate": 6.345598261312075e-05, - "loss": 0.9927, + "learning_rate": 6.367522391766672e-05, + "loss": 5.1592, "step": 8205 }, { "epoch": 0.66, - "learning_rate": 6.332588736524927e-05, - "loss": 1.0909, + "learning_rate": 6.35451474248955e-05, + "loss": 5.0742, "step": 8210 }, { "epoch": 0.66, - "learning_rate": 6.319586379372042e-05, - "loss": 0.9927, + "learning_rate": 6.3415142028001e-05, + "loss": 4.9903, "step": 8215 }, { "epoch": 0.66, - "learning_rate": 6.306591215265387e-05, - "loss": 0.9878, + "learning_rate": 6.328520798052564e-05, + "loss": 5.1529, "step": 8220 }, { "epoch": 0.66, - "learning_rate": 6.293603269602873e-05, - "loss": 0.9656, + "learning_rate": 6.315534553587253e-05, + "loss": 5.091, "step": 8225 }, { "epoch": 0.66, - "learning_rate": 6.280622567768301e-05, - "loss": 0.9712, + "learning_rate": 6.302555494730518e-05, + "loss": 5.0697, "step": 8230 }, { "epoch": 0.66, - "learning_rate": 6.267649135131317e-05, - "loss": 1.0198, + "learning_rate": 6.289583646794694e-05, + "loss": 5.1356, "step": 8235 }, { "epoch": 0.66, - "learning_rate": 6.254682997047365e-05, - "loss": 0.972, + "learning_rate": 6.276619035078066e-05, + "loss": 5.0537, "step": 8240 }, { "epoch": 0.66, - "learning_rate": 6.24172417885762e-05, - "loss": 0.9773, + "learning_rate": 6.263661684864793e-05, + "loss": 5.1565, "step": 8245 }, { "epoch": 0.66, - "learning_rate": 6.228772705888961e-05, - "loss": 0.9874, + "learning_rate": 6.25071162142487e-05, + "loss": 5.1121, "step": 8250 }, { "epoch": 0.66, - "learning_rate": 6.215828603453903e-05, - "loss": 1.0174, + "learning_rate": 6.237768870014096e-05, + "loss": 5.0897, "step": 8255 }, { "epoch": 0.66, - "learning_rate": 6.202891896850569e-05, - "loss": 1.0074, + "learning_rate": 6.224833455873997e-05, + "loss": 5.1256, "step": 8260 }, { "epoch": 0.66, - "learning_rate": 6.189962611362617e-05, - "loss": 0.9681, + "learning_rate": 6.211905404231798e-05, + "loss": 5.0669, "step": 8265 }, { "epoch": 0.66, - "learning_rate": 6.177040772259199e-05, - "loss": 1.0191, + "learning_rate": 6.198984740300355e-05, + "loss": 5.1049, "step": 8270 }, { "epoch": 0.66, - "learning_rate": 6.164126404794924e-05, - "loss": 0.9494, + "learning_rate": 6.186071489278122e-05, + "loss": 5.0825, "step": 8275 }, { "epoch": 0.66, - "learning_rate": 6.151219534209786e-05, - "loss": 1.0061, + "learning_rate": 6.173165676349103e-05, + "loss": 5.1157, "step": 8280 }, { "epoch": 0.66, - "learning_rate": 6.138320185729143e-05, - "loss": 0.9376, + "learning_rate": 6.160267326682785e-05, + "loss": 5.1456, "step": 8285 }, { "epoch": 0.66, - "learning_rate": 6.125428384563635e-05, - "loss": 0.964, + "learning_rate": 6.1473764654341e-05, + "loss": 5.1325, "step": 8290 }, { "epoch": 0.66, - "learning_rate": 6.112544155909164e-05, - "loss": 0.9563, + "learning_rate": 6.134493117743387e-05, + "loss": 5.1065, "step": 8295 }, { "epoch": 0.66, - "learning_rate": 6.099667524946823e-05, - "loss": 0.9903, + "learning_rate": 6.12161730873632e-05, + "loss": 5.0897, "step": 8300 }, { - "epoch": 0.67, - "learning_rate": 6.086798516842858e-05, - "loss": 1.0302, + "epoch": 0.66, + "learning_rate": 6.108749063523876e-05, + "loss": 5.038, "step": 8305 }, { - "epoch": 0.67, - "learning_rate": 6.0739371567486214e-05, - "loss": 0.9309, + "epoch": 0.66, + "learning_rate": 6.095888407202277e-05, + "loss": 5.1049, "step": 8310 }, { "epoch": 0.67, - "learning_rate": 6.061083469800522e-05, - "loss": 1.0136, + "learning_rate": 6.083035364852951e-05, + "loss": 5.1377, "step": 8315 }, { "epoch": 0.67, - "learning_rate": 6.048237481119956e-05, - "loss": 0.9859, + "learning_rate": 6.0701899615424764e-05, + "loss": 5.0632, "step": 8320 }, { "epoch": 0.67, - "learning_rate": 6.0353992158132876e-05, - "loss": 0.9799, + "learning_rate": 6.057352222322525e-05, + "loss": 5.1224, "step": 8325 }, { "epoch": 0.67, - "learning_rate": 6.02256869897178e-05, - "loss": 0.9736, + "learning_rate": 6.044522172229827e-05, + "loss": 5.0995, "step": 8330 }, { "epoch": 0.67, - "learning_rate": 6.0097459556715616e-05, - "loss": 0.9431, + "learning_rate": 6.031699836286123e-05, + "loss": 5.0715, "step": 8335 }, { "epoch": 0.67, - "learning_rate": 5.996931010973557e-05, - "loss": 0.9563, + "learning_rate": 6.018885239498101e-05, + "loss": 5.07, "step": 8340 }, { "epoch": 0.67, - "learning_rate": 5.9841238899234534e-05, - "loss": 0.9767, + "learning_rate": 6.006078406857358e-05, + "loss": 5.1005, "step": 8345 }, { "epoch": 0.67, - "learning_rate": 5.9713246175516504e-05, - "loss": 0.9576, + "learning_rate": 5.993279363340345e-05, + "loss": 5.1621, "step": 8350 }, { "epoch": 0.67, - "learning_rate": 5.958533218873198e-05, - "loss": 0.9644, + "learning_rate": 5.980488133908333e-05, + "loss": 5.0672, "step": 8355 }, { "epoch": 0.67, - "learning_rate": 5.9457497188877766e-05, - "loss": 0.9875, + "learning_rate": 5.967704743507345e-05, + "loss": 5.1307, "step": 8360 }, { "epoch": 0.67, - "learning_rate": 5.9329741425796105e-05, - "loss": 1.0086, + "learning_rate": 5.954929217068119e-05, + "loss": 5.1029, "step": 8365 }, { "epoch": 0.67, - "learning_rate": 5.920206514917446e-05, - "loss": 1.0627, + "learning_rate": 5.942161579506051e-05, + "loss": 5.0517, "step": 8370 }, { "epoch": 0.67, - "learning_rate": 5.907446860854495e-05, - "loss": 0.9868, + "learning_rate": 5.929401855721162e-05, + "loss": 5.0905, "step": 8375 }, { "epoch": 0.67, - "learning_rate": 5.894695205328381e-05, - "loss": 1.0431, + "learning_rate": 5.9166500705980335e-05, + "loss": 5.1807, "step": 8380 }, { "epoch": 0.67, - "learning_rate": 5.8819515732611043e-05, - "loss": 0.9911, + "learning_rate": 5.903906249005757e-05, + "loss": 5.082, "step": 8385 }, { "epoch": 0.67, - "learning_rate": 5.8692159895589736e-05, - "loss": 0.9999, + "learning_rate": 5.891170415797911e-05, + "loss": 5.0097, "step": 8390 }, { "epoch": 0.67, - "learning_rate": 5.856488479112575e-05, - "loss": 0.9605, + "learning_rate": 5.87844259581248e-05, + "loss": 5.0638, "step": 8395 }, { "epoch": 0.67, - "learning_rate": 5.843769066796712e-05, - "loss": 1.0034, + "learning_rate": 5.8657228138718255e-05, + "loss": 5.0707, "step": 8400 }, { "epoch": 0.67, - "learning_rate": 5.831057777470358e-05, - "loss": 1.0017, + "learning_rate": 5.8530110947826274e-05, + "loss": 5.1249, "step": 8405 }, { "epoch": 0.67, - "learning_rate": 5.818354635976627e-05, - "loss": 0.943, + "learning_rate": 5.840307463335856e-05, + "loss": 5.1389, "step": 8410 }, { "epoch": 0.67, - "learning_rate": 5.805659667142687e-05, - "loss": 1.0356, + "learning_rate": 5.827611944306691e-05, + "loss": 5.0893, "step": 8415 }, { "epoch": 0.67, - "learning_rate": 5.7929728957797445e-05, - "loss": 0.9668, + "learning_rate": 5.8149245624545065e-05, + "loss": 5.0862, "step": 8420 }, { "epoch": 0.67, - "learning_rate": 5.7802943466829896e-05, - "loss": 0.9544, + "learning_rate": 5.8022453425227874e-05, + "loss": 5.158, "step": 8425 }, { - "epoch": 0.68, - "learning_rate": 5.7676240446315244e-05, - "loss": 0.9289, + "epoch": 0.67, + "learning_rate": 5.78957430923913e-05, + "loss": 5.0466, "step": 8430 }, { - "epoch": 0.68, - "learning_rate": 5.7549620143883606e-05, - "loss": 0.9715, + "epoch": 0.67, + "learning_rate": 5.776911487315133e-05, + "loss": 5.1084, "step": 8435 }, { "epoch": 0.68, - "learning_rate": 5.742308280700317e-05, - "loss": 1.0343, + "learning_rate": 5.764256901446401e-05, + "loss": 4.9947, "step": 8440 }, { "epoch": 0.68, - "learning_rate": 5.72966286829801e-05, - "loss": 1.0038, + "learning_rate": 5.7516105763124715e-05, + "loss": 5.0802, "step": 8445 }, { "epoch": 0.68, - "learning_rate": 5.717025801895797e-05, - "loss": 1.0361, + "learning_rate": 5.738972536576773e-05, + "loss": 5.0558, "step": 8450 }, { "epoch": 0.68, - "learning_rate": 5.704397106191706e-05, - "loss": 1.0031, + "learning_rate": 5.7263428068865664e-05, + "loss": 5.0511, "step": 8455 }, { "epoch": 0.68, - "learning_rate": 5.691776805867428e-05, - "loss": 0.9737, + "learning_rate": 5.713721411872919e-05, + "loss": 5.082, "step": 8460 }, { "epoch": 0.68, - "learning_rate": 5.679164925588236e-05, - "loss": 0.9894, + "learning_rate": 5.7011083761506344e-05, + "loss": 5.0735, "step": 8465 }, { "epoch": 0.68, - "learning_rate": 5.666561490002939e-05, - "loss": 0.9902, + "learning_rate": 5.688503724318217e-05, + "loss": 5.1189, "step": 8470 }, { "epoch": 0.68, - "learning_rate": 5.653966523743851e-05, - "loss": 0.9249, + "learning_rate": 5.675907480957825e-05, + "loss": 5.0527, "step": 8475 }, { "epoch": 0.68, - "learning_rate": 5.641380051426731e-05, - "loss": 0.968, + "learning_rate": 5.663319670635201e-05, + "loss": 5.1, "step": 8480 }, { "epoch": 0.68, - "learning_rate": 5.628802097650738e-05, - "loss": 1.0266, + "learning_rate": 5.6507403178996685e-05, + "loss": 5.0411, "step": 8485 }, { "epoch": 0.68, - "learning_rate": 5.616232686998385e-05, - "loss": 1.0157, + "learning_rate": 5.63816944728403e-05, + "loss": 5.1244, "step": 8490 }, { "epoch": 0.68, - "learning_rate": 5.6036718440354764e-05, - "loss": 1.0253, + "learning_rate": 5.625607083304565e-05, + "loss": 5.0744, "step": 8495 }, { "epoch": 0.68, - "learning_rate": 5.591119593311086e-05, - "loss": 1.0328, + "learning_rate": 5.6130532504609424e-05, + "loss": 5.0087, "step": 8500 }, { "epoch": 0.68, - "learning_rate": 5.578575959357487e-05, - "loss": 1.0045, + "learning_rate": 5.600507973236223e-05, + "loss": 5.1707, "step": 8505 }, { "epoch": 0.68, - "learning_rate": 5.566040966690115e-05, - "loss": 1.0228, + "learning_rate": 5.587971276096753e-05, + "loss": 5.1334, "step": 8510 }, { "epoch": 0.68, - "learning_rate": 5.553514639807514e-05, - "loss": 0.9882, + "learning_rate": 5.5754431834921603e-05, + "loss": 5.1112, "step": 8515 }, { "epoch": 0.68, - "learning_rate": 5.540997003191301e-05, - "loss": 1.0399, + "learning_rate": 5.5629237198552906e-05, + "loss": 5.1394, "step": 8520 }, { "epoch": 0.68, - "learning_rate": 5.528488081306089e-05, - "loss": 0.9911, + "learning_rate": 5.550412909602158e-05, + "loss": 5.0861, "step": 8525 }, { "epoch": 0.68, - "learning_rate": 5.5159878985994774e-05, - "loss": 1.0162, + "learning_rate": 5.537910777131906e-05, + "loss": 5.0891, "step": 8530 }, { "epoch": 0.68, - "learning_rate": 5.503496479501977e-05, - "loss": 0.9992, + "learning_rate": 5.525417346826742e-05, + "loss": 5.1153, "step": 8535 }, { "epoch": 0.68, - "learning_rate": 5.4910138484269754e-05, - "loss": 1.0232, + "learning_rate": 5.512932643051916e-05, + "loss": 5.169, "step": 8540 }, { "epoch": 0.68, - "learning_rate": 5.478540029770683e-05, - "loss": 1.0048, + "learning_rate": 5.500456690155651e-05, + "loss": 5.0748, "step": 8545 }, { "epoch": 0.68, - "learning_rate": 5.4660750479120826e-05, - "loss": 0.9884, + "learning_rate": 5.487989512469109e-05, + "loss": 5.1572, "step": 8550 }, { - "epoch": 0.69, - "learning_rate": 5.453618927212889e-05, - "loss": 0.9746, + "epoch": 0.68, + "learning_rate": 5.475531134306326e-05, + "loss": 5.0608, "step": 8555 }, { - "epoch": 0.69, - "learning_rate": 5.441171692017504e-05, - "loss": 1.0192, + "epoch": 0.68, + "learning_rate": 5.4630815799641996e-05, + "loss": 5.0609, "step": 8560 }, { "epoch": 0.69, - "learning_rate": 5.4287333666529584e-05, - "loss": 0.9322, + "learning_rate": 5.450640873722395e-05, + "loss": 5.0738, "step": 8565 }, { "epoch": 0.69, - "learning_rate": 5.416303975428871e-05, - "loss": 0.9648, + "learning_rate": 5.438209039843333e-05, + "loss": 5.1029, "step": 8570 }, { "epoch": 0.69, - "learning_rate": 5.403883542637395e-05, - "loss": 0.9857, + "learning_rate": 5.4257861025721315e-05, + "loss": 5.0662, "step": 8575 }, { "epoch": 0.69, - "learning_rate": 5.391472092553176e-05, - "loss": 0.9998, + "learning_rate": 5.4133720861365545e-05, + "loss": 5.1252, "step": 8580 }, { "epoch": 0.69, - "learning_rate": 5.37906964943332e-05, - "loss": 1.0002, + "learning_rate": 5.400967014746976e-05, + "loss": 5.0378, "step": 8585 }, { "epoch": 0.69, - "learning_rate": 5.366676237517305e-05, - "loss": 0.9467, + "learning_rate": 5.388570912596307e-05, + "loss": 5.0326, "step": 8590 }, { "epoch": 0.69, - "learning_rate": 5.354291881026971e-05, - "loss": 1.0204, + "learning_rate": 5.376183803859986e-05, + "loss": 5.215, "step": 8595 }, { "epoch": 0.69, - "learning_rate": 5.341916604166466e-05, - "loss": 0.9047, + "learning_rate": 5.363805712695901e-05, + "loss": 5.0298, "step": 8600 }, { "epoch": 0.69, - "learning_rate": 5.329550431122172e-05, - "loss": 1.0211, + "learning_rate": 5.3514366632443626e-05, + "loss": 4.9864, "step": 8605 }, { "epoch": 0.69, - "learning_rate": 5.317193386062706e-05, - "loss": 0.9516, + "learning_rate": 5.339076679628031e-05, + "loss": 5.0623, "step": 8610 }, { "epoch": 0.69, - "learning_rate": 5.304845493138821e-05, - "loss": 0.9808, + "learning_rate": 5.3267257859519026e-05, + "loss": 5.0605, "step": 8615 }, { "epoch": 0.69, - "learning_rate": 5.2925067764833945e-05, - "loss": 0.9764, + "learning_rate": 5.314384006303241e-05, + "loss": 5.0526, "step": 8620 }, { "epoch": 0.69, - "learning_rate": 5.280177260211375e-05, - "loss": 0.9411, + "learning_rate": 5.302051364751533e-05, + "loss": 4.965, "step": 8625 }, { "epoch": 0.69, - "learning_rate": 5.2678569684197086e-05, - "loss": 0.9533, + "learning_rate": 5.2897278853484464e-05, + "loss": 5.0736, "step": 8630 }, { "epoch": 0.69, - "learning_rate": 5.2555459251873395e-05, - "loss": 0.9951, + "learning_rate": 5.277413592127778e-05, + "loss": 5.0675, "step": 8635 }, { "epoch": 0.69, - "learning_rate": 5.2432441545751244e-05, - "loss": 1.021, + "learning_rate": 5.2651085091054165e-05, + "loss": 5.0029, "step": 8640 }, { "epoch": 0.69, - "learning_rate": 5.230951680625789e-05, - "loss": 0.9877, + "learning_rate": 5.252812660279274e-05, + "loss": 5.1075, "step": 8645 }, { "epoch": 0.69, - "learning_rate": 5.218668527363901e-05, - "loss": 0.9405, + "learning_rate": 5.240526069629265e-05, + "loss": 5.0721, "step": 8650 }, { "epoch": 0.69, - "learning_rate": 5.2063947187958085e-05, - "loss": 0.9534, + "learning_rate": 5.228248761117247e-05, + "loss": 5.0494, "step": 8655 }, { "epoch": 0.69, - "learning_rate": 5.194130278909595e-05, - "loss": 0.9879, + "learning_rate": 5.215980758686978e-05, + "loss": 5.0252, "step": 8660 }, { "epoch": 0.69, - "learning_rate": 5.18187523167504e-05, - "loss": 0.9085, + "learning_rate": 5.203722086264056e-05, + "loss": 5.0636, "step": 8665 }, { "epoch": 0.69, - "learning_rate": 5.1696296010435504e-05, - "loss": 1.0474, + "learning_rate": 5.191472767755889e-05, + "loss": 5.1137, "step": 8670 }, { "epoch": 0.69, - "learning_rate": 5.157393410948146e-05, - "loss": 0.9883, + "learning_rate": 5.179232827051648e-05, + "loss": 5.0657, "step": 8675 }, { - "epoch": 0.7, - "learning_rate": 5.1451666853033885e-05, - "loss": 0.9823, + "epoch": 0.69, + "learning_rate": 5.167002288022209e-05, + "loss": 5.0815, "step": 8680 }, { - "epoch": 0.7, - "learning_rate": 5.132949448005341e-05, - "loss": 1.0241, + "epoch": 0.69, + "learning_rate": 5.154781174520118e-05, + "loss": 5.0423, "step": 8685 }, { "epoch": 0.7, - "learning_rate": 5.12074172293153e-05, - "loss": 1.0298, + "learning_rate": 5.1425695103795225e-05, + "loss": 5.0129, "step": 8690 }, { "epoch": 0.7, - "learning_rate": 5.108543533940877e-05, - "loss": 1.0051, + "learning_rate": 5.13036731941617e-05, + "loss": 5.0075, "step": 8695 }, { "epoch": 0.7, - "learning_rate": 5.0963549048736766e-05, - "loss": 0.9615, + "learning_rate": 5.118174625427302e-05, + "loss": 5.0807, "step": 8700 }, { "epoch": 0.7, - "learning_rate": 5.084175859551542e-05, - "loss": 0.9463, + "learning_rate": 5.105991452191665e-05, + "loss": 5.0181, "step": 8705 }, { "epoch": 0.7, - "learning_rate": 5.072006421777347e-05, - "loss": 1.0006, + "learning_rate": 5.0938178234694133e-05, + "loss": 5.0379, "step": 8710 }, { "epoch": 0.7, - "learning_rate": 5.0598466153351934e-05, - "loss": 1.0038, + "learning_rate": 5.081653763002113e-05, + "loss": 5.0106, "step": 8715 }, { "epoch": 0.7, - "learning_rate": 5.047696463990364e-05, - "loss": 1.023, + "learning_rate": 5.069499294512647e-05, + "loss": 5.1549, "step": 8720 }, { "epoch": 0.7, - "learning_rate": 5.0355559914892535e-05, - "loss": 1.0183, + "learning_rate": 5.0573544417052054e-05, + "loss": 5.1111, "step": 8725 }, { "epoch": 0.7, - "learning_rate": 5.0234252215593626e-05, - "loss": 0.9457, + "learning_rate": 5.045219228265219e-05, + "loss": 5.0949, "step": 8730 }, { "epoch": 0.7, - "learning_rate": 5.011304177909215e-05, - "loss": 1.0031, + "learning_rate": 5.033093677859323e-05, + "loss": 5.0041, "step": 8735 }, { "epoch": 0.7, - "learning_rate": 4.999192884228331e-05, - "loss": 0.982, + "learning_rate": 5.020977814135308e-05, + "loss": 5.0503, "step": 8740 }, { "epoch": 0.7, - "learning_rate": 4.9870913641871775e-05, - "loss": 1.0254, + "learning_rate": 5.008871660722061e-05, + "loss": 5.0935, "step": 8745 }, { "epoch": 0.7, - "learning_rate": 4.974999641437108e-05, - "loss": 0.9864, + "learning_rate": 4.996775241229555e-05, + "loss": 5.1407, "step": 8750 }, { "epoch": 0.7, - "learning_rate": 4.962917739610341e-05, - "loss": 0.9885, + "learning_rate": 4.984688579248756e-05, + "loss": 5.022, "step": 8755 }, { "epoch": 0.7, - "learning_rate": 4.950845682319893e-05, - "loss": 0.9834, + "learning_rate": 4.972611698351618e-05, + "loss": 4.9594, "step": 8760 }, { "epoch": 0.7, - "learning_rate": 4.938783493159548e-05, - "loss": 0.9716, + "learning_rate": 4.960544622090999e-05, + "loss": 5.0411, "step": 8765 }, { "epoch": 0.7, - "learning_rate": 4.9267311957037975e-05, - "loss": 0.9735, + "learning_rate": 4.948487374000665e-05, + "loss": 4.999, "step": 8770 }, { "epoch": 0.7, - "learning_rate": 4.914688813507797e-05, - "loss": 0.9451, + "learning_rate": 4.9364399775951875e-05, + "loss": 5.0029, "step": 8775 }, { "epoch": 0.7, - "learning_rate": 4.902656370107329e-05, - "loss": 0.9367, + "learning_rate": 4.9244024563699366e-05, + "loss": 4.9842, "step": 8780 }, { "epoch": 0.7, - "learning_rate": 4.8906338890187596e-05, - "loss": 0.9877, + "learning_rate": 4.912374833801024e-05, + "loss": 4.9885, "step": 8785 }, { "epoch": 0.7, - "learning_rate": 4.878621393738967e-05, - "loss": 0.9342, + "learning_rate": 4.900357133345259e-05, + "loss": 5.1303, "step": 8790 }, { "epoch": 0.7, - "learning_rate": 4.866618907745323e-05, - "loss": 0.9812, + "learning_rate": 4.8883493784400866e-05, + "loss": 4.9168, "step": 8795 }, { "epoch": 0.7, - "learning_rate": 4.85462645449564e-05, - "loss": 0.9084, + "learning_rate": 4.876351592503572e-05, + "loss": 5.0822, "step": 8800 }, { - "epoch": 0.71, - "learning_rate": 4.8426440574281085e-05, - "loss": 0.9696, + "epoch": 0.7, + "learning_rate": 4.864363798934328e-05, + "loss": 5.0265, "step": 8805 }, { - "epoch": 0.71, - "learning_rate": 4.8306717399612875e-05, - "loss": 0.9665, + "epoch": 0.7, + "learning_rate": 4.852386021111486e-05, + "loss": 5.0752, "step": 8810 }, { "epoch": 0.71, - "learning_rate": 4.818709525494013e-05, - "loss": 0.9902, + "learning_rate": 4.8404182823946445e-05, + "loss": 5.0575, "step": 8815 }, { "epoch": 0.71, - "learning_rate": 4.8067574374053906e-05, - "loss": 1.0006, + "learning_rate": 4.828460606123809e-05, + "loss": 5.0806, "step": 8820 }, { "epoch": 0.71, - "learning_rate": 4.794815499054732e-05, - "loss": 0.9691, + "learning_rate": 4.816513015619387e-05, + "loss": 5.098, "step": 8825 }, { "epoch": 0.71, - "learning_rate": 4.7828837337815015e-05, - "loss": 0.9514, + "learning_rate": 4.804575534182092e-05, + "loss": 5.0798, "step": 8830 }, { "epoch": 0.71, - "learning_rate": 4.770962164905298e-05, - "loss": 0.9683, + "learning_rate": 4.792648185092937e-05, + "loss": 5.0878, "step": 8835 }, { "epoch": 0.71, - "learning_rate": 4.7590508157257873e-05, - "loss": 0.9623, + "learning_rate": 4.78073099161316e-05, + "loss": 5.1384, "step": 8840 }, { "epoch": 0.71, - "learning_rate": 4.747149709522649e-05, - "loss": 0.9291, + "learning_rate": 4.768823976984216e-05, + "loss": 5.008, "step": 8845 }, { "epoch": 0.71, - "learning_rate": 4.735258869555558e-05, - "loss": 1.0139, + "learning_rate": 4.756927164427685e-05, + "loss": 5.0476, "step": 8850 }, { "epoch": 0.71, - "learning_rate": 4.72337831906412e-05, - "loss": 0.9981, + "learning_rate": 4.7450405771452634e-05, + "loss": 4.9724, "step": 8855 }, { "epoch": 0.71, - "learning_rate": 4.7115080812678305e-05, - "loss": 1.0405, + "learning_rate": 4.733164238318703e-05, + "loss": 5.0637, "step": 8860 }, { "epoch": 0.71, - "learning_rate": 4.699648179366034e-05, - "loss": 0.9651, + "learning_rate": 4.721298171109769e-05, + "loss": 5.1059, "step": 8865 }, { "epoch": 0.71, - "learning_rate": 4.6877986365378635e-05, - "loss": 0.9967, + "learning_rate": 4.709442398660198e-05, + "loss": 4.988, "step": 8870 }, { "epoch": 0.71, - "learning_rate": 4.6759594759422163e-05, - "loss": 0.9925, + "learning_rate": 4.697596944091638e-05, + "loss": 5.0242, "step": 8875 }, { "epoch": 0.71, - "learning_rate": 4.6641307207176964e-05, - "loss": 0.9825, + "learning_rate": 4.685761830505625e-05, + "loss": 5.0314, "step": 8880 }, { "epoch": 0.71, - "learning_rate": 4.652312393982568e-05, - "loss": 0.9669, + "learning_rate": 4.673937080983525e-05, + "loss": 5.1087, "step": 8885 }, { "epoch": 0.71, - "learning_rate": 4.640504518834723e-05, - "loss": 0.9905, + "learning_rate": 4.6621227185864935e-05, + "loss": 5.0283, "step": 8890 }, { "epoch": 0.71, - "learning_rate": 4.62870711835161e-05, - "loss": 1.0075, + "learning_rate": 4.650318766355417e-05, + "loss": 5.013, "step": 8895 }, { "epoch": 0.71, - "learning_rate": 4.616920215590219e-05, - "loss": 0.9709, + "learning_rate": 4.638525247310903e-05, + "loss": 5.0661, "step": 8900 }, { "epoch": 0.71, - "learning_rate": 4.605143833587021e-05, - "loss": 0.9869, + "learning_rate": 4.6267421844531856e-05, + "loss": 4.9891, "step": 8905 }, { "epoch": 0.71, - "learning_rate": 4.5933779953579236e-05, - "loss": 0.9953, + "learning_rate": 4.6149696007621235e-05, + "loss": 5.0557, "step": 8910 }, { "epoch": 0.71, - "learning_rate": 4.5816227238982266e-05, - "loss": 1.0112, + "learning_rate": 4.6032075191971324e-05, + "loss": 5.0894, "step": 8915 }, { "epoch": 0.71, - "learning_rate": 4.569878042182584e-05, - "loss": 0.9754, + "learning_rate": 4.5914559626971474e-05, + "loss": 4.9938, "step": 8920 }, { "epoch": 0.71, - "learning_rate": 4.558143973164939e-05, - "loss": 0.9727, + "learning_rate": 4.579714954180583e-05, + "loss": 5.0697, "step": 8925 }, { - "epoch": 0.72, - "learning_rate": 4.5464205397785074e-05, - "loss": 0.9899, + "epoch": 0.71, + "learning_rate": 4.567984516545267e-05, + "loss": 5.056, "step": 8930 }, { - "epoch": 0.72, - "learning_rate": 4.5347077649357116e-05, - "loss": 1.0168, + "epoch": 0.71, + "learning_rate": 4.5562646726684265e-05, + "loss": 5.0946, "step": 8935 }, { "epoch": 0.72, - "learning_rate": 4.523005671528148e-05, - "loss": 0.995, + "learning_rate": 4.544555445406622e-05, + "loss": 5.0095, "step": 8940 }, { "epoch": 0.72, - "learning_rate": 4.511314282426534e-05, - "loss": 0.9485, + "learning_rate": 4.532856857595714e-05, + "loss": 5.0621, "step": 8945 }, { "epoch": 0.72, - "learning_rate": 4.499633620480661e-05, - "loss": 1.0041, + "learning_rate": 4.521168932050802e-05, + "loss": 5.0524, "step": 8950 }, { "epoch": 0.72, - "learning_rate": 4.487963708519358e-05, - "loss": 1.0148, + "learning_rate": 4.5094916915662035e-05, + "loss": 5.0751, "step": 8955 }, { "epoch": 0.72, - "learning_rate": 4.47630456935046e-05, - "loss": 0.9775, + "learning_rate": 4.497825158915391e-05, + "loss": 5.0295, "step": 8960 }, { "epoch": 0.72, - "learning_rate": 4.464656225760722e-05, - "loss": 0.9414, + "learning_rate": 4.486169356850955e-05, + "loss": 4.9998, "step": 8965 }, { "epoch": 0.72, - "learning_rate": 4.453018700515813e-05, - "loss": 0.9903, + "learning_rate": 4.474524308104564e-05, + "loss": 5.0683, "step": 8970 }, { "epoch": 0.72, - "learning_rate": 4.441392016360264e-05, - "loss": 0.98, + "learning_rate": 4.462890035386908e-05, + "loss": 4.9306, "step": 8975 }, { "epoch": 0.72, - "learning_rate": 4.429776196017403e-05, - "loss": 0.9727, + "learning_rate": 4.451266561387666e-05, + "loss": 4.9917, "step": 8980 }, { "epoch": 0.72, - "learning_rate": 4.418171262189339e-05, - "loss": 1.0001, + "learning_rate": 4.43965390877545e-05, + "loss": 5.0562, "step": 8985 }, { "epoch": 0.72, - "learning_rate": 4.406577237556896e-05, - "loss": 0.968, + "learning_rate": 4.4280521001977735e-05, + "loss": 4.9895, "step": 8990 }, { "epoch": 0.72, - "learning_rate": 4.394994144779583e-05, - "loss": 0.9724, + "learning_rate": 4.416461158281002e-05, + "loss": 5.1659, "step": 8995 }, { "epoch": 0.72, - "learning_rate": 4.3834220064955436e-05, - "loss": 1.0192, + "learning_rate": 4.4048811056303094e-05, + "loss": 5.0395, "step": 9000 }, { "epoch": 0.72, - "learning_rate": 4.3718608453215014e-05, - "loss": 0.9411, + "learning_rate": 4.3933119648296236e-05, + "loss": 5.0649, "step": 9005 }, { "epoch": 0.72, - "learning_rate": 4.360310683852739e-05, - "loss": 0.9678, + "learning_rate": 4.3817537584416e-05, + "loss": 5.0206, "step": 9010 }, { "epoch": 0.72, - "learning_rate": 4.348771544663033e-05, - "loss": 1.0292, + "learning_rate": 4.3702065090075686e-05, + "loss": 5.0382, "step": 9015 }, { "epoch": 0.72, - "learning_rate": 4.337243450304623e-05, - "loss": 1.0122, + "learning_rate": 4.35867023904749e-05, + "loss": 4.9994, "step": 9020 }, { "epoch": 0.72, - "learning_rate": 4.325726423308164e-05, - "loss": 0.9539, + "learning_rate": 4.3471449710599134e-05, + "loss": 5.0041, "step": 9025 }, { "epoch": 0.72, - "learning_rate": 4.314220486182668e-05, - "loss": 0.9963, + "learning_rate": 4.33563072752192e-05, + "loss": 5.0569, "step": 9030 }, { "epoch": 0.72, - "learning_rate": 4.3027256614154845e-05, - "loss": 0.9871, + "learning_rate": 4.324127530889114e-05, + "loss": 5.0156, "step": 9035 }, { "epoch": 0.72, - "learning_rate": 4.2912419714722496e-05, - "loss": 0.9596, + "learning_rate": 4.312635403595532e-05, + "loss": 5.0364, "step": 9040 }, { "epoch": 0.72, - "learning_rate": 4.2797694387968224e-05, - "loss": 0.9559, + "learning_rate": 4.3011543680536394e-05, + "loss": 5.0185, "step": 9045 }, { "epoch": 0.72, - "learning_rate": 4.2683080858112657e-05, - "loss": 0.9806, + "learning_rate": 4.289684446654251e-05, + "loss": 5.0593, "step": 9050 }, { - "epoch": 0.73, - "learning_rate": 4.2568579349157945e-05, - "loss": 1.0386, + "epoch": 0.72, + "learning_rate": 4.278225661766533e-05, + "loss": 5.0203, "step": 9055 }, { - "epoch": 0.73, - "learning_rate": 4.2454190084887145e-05, - "loss": 1.04, + "epoch": 0.72, + "learning_rate": 4.266778035737908e-05, + "loss": 5.0338, "step": 9060 }, { "epoch": 0.73, - "learning_rate": 4.233991328886423e-05, - "loss": 0.9856, + "learning_rate": 4.25534159089405e-05, + "loss": 5.0106, "step": 9065 }, { "epoch": 0.73, - "learning_rate": 4.222574918443306e-05, - "loss": 1.0009, + "learning_rate": 4.24391634953882e-05, + "loss": 4.9901, "step": 9070 }, { "epoch": 0.73, - "learning_rate": 4.211169799471743e-05, - "loss": 0.9339, + "learning_rate": 4.232502333954236e-05, + "loss": 5.0219, "step": 9075 }, { "epoch": 0.73, - "learning_rate": 4.199775994262044e-05, - "loss": 1.0096, + "learning_rate": 4.221099566400422e-05, + "loss": 5.0667, "step": 9080 }, { "epoch": 0.73, - "learning_rate": 4.1883935250823935e-05, - "loss": 1.0072, + "learning_rate": 4.20970806911555e-05, + "loss": 5.1011, "step": 9085 }, { "epoch": 0.73, - "learning_rate": 4.1770224141788404e-05, - "loss": 0.9565, + "learning_rate": 4.198327864315843e-05, + "loss": 5.0451, "step": 9090 }, { "epoch": 0.73, - "learning_rate": 4.165662683775228e-05, - "loss": 1.007, + "learning_rate": 4.1869589741954695e-05, + "loss": 5.0588, "step": 9095 }, { "epoch": 0.73, - "learning_rate": 4.154314356073146e-05, - "loss": 0.9592, + "learning_rate": 4.175601420926555e-05, + "loss": 4.9475, "step": 9100 }, { "epoch": 0.73, - "learning_rate": 4.142977453251911e-05, - "loss": 0.9329, + "learning_rate": 4.164255226659093e-05, + "loss": 5.049, "step": 9105 }, { "epoch": 0.73, - "learning_rate": 4.1316519974685077e-05, - "loss": 0.9486, + "learning_rate": 4.152920413520951e-05, + "loss": 5.0313, "step": 9110 }, { "epoch": 0.73, - "learning_rate": 4.120338010857547e-05, - "loss": 1.0156, + "learning_rate": 4.1415970036177776e-05, + "loss": 5.0142, "step": 9115 }, { "epoch": 0.73, - "learning_rate": 4.10903551553123e-05, - "loss": 0.968, + "learning_rate": 4.130285019032992e-05, + "loss": 4.9716, "step": 9120 }, { "epoch": 0.73, - "learning_rate": 4.097744533579288e-05, - "loss": 0.9759, + "learning_rate": 4.118984481827732e-05, + "loss": 5.0336, "step": 9125 }, { "epoch": 0.73, - "learning_rate": 4.086465087068958e-05, - "loss": 0.9978, + "learning_rate": 4.107695414040812e-05, + "loss": 5.0234, "step": 9130 }, { "epoch": 0.73, - "learning_rate": 4.075197198044931e-05, - "loss": 0.9973, + "learning_rate": 4.096417837688666e-05, + "loss": 5.0314, "step": 9135 }, { "epoch": 0.73, - "learning_rate": 4.063940888529312e-05, - "loss": 1.0424, + "learning_rate": 4.085151774765331e-05, + "loss": 5.0649, "step": 9140 }, { "epoch": 0.73, - "learning_rate": 4.0526961805215744e-05, - "loss": 0.9887, + "learning_rate": 4.0738972472423845e-05, + "loss": 5.038, "step": 9145 }, { "epoch": 0.73, - "learning_rate": 4.041463095998509e-05, - "loss": 1.0167, + "learning_rate": 4.0626542770689044e-05, + "loss": 5.1134, "step": 9150 }, { "epoch": 0.73, - "learning_rate": 4.0302416569141996e-05, - "loss": 0.9545, + "learning_rate": 4.051422886171439e-05, + "loss": 5.0124, "step": 9155 }, { "epoch": 0.73, - "learning_rate": 4.019031885199967e-05, - "loss": 0.9923, + "learning_rate": 4.040203096453934e-05, + "loss": 5.0618, "step": 9160 }, { "epoch": 0.73, - "learning_rate": 4.007833802764327e-05, - "loss": 0.9563, + "learning_rate": 4.028994929797738e-05, + "loss": 5.033, "step": 9165 }, { "epoch": 0.73, - "learning_rate": 3.9966474314929526e-05, - "loss": 0.9419, + "learning_rate": 4.017798408061506e-05, + "loss": 5.0358, "step": 9170 }, { "epoch": 0.73, - "learning_rate": 3.985472793248631e-05, - "loss": 1.0117, + "learning_rate": 4.006613553081201e-05, + "loss": 4.9816, "step": 9175 }, { - "epoch": 0.74, - "learning_rate": 3.974309909871208e-05, - "loss": 0.959, + "epoch": 0.73, + "learning_rate": 3.995440386670013e-05, + "loss": 5.1579, "step": 9180 }, { - "epoch": 0.74, - "learning_rate": 3.963158803177564e-05, - "loss": 1.0561, + "epoch": 0.73, + "learning_rate": 3.9842789306183634e-05, + "loss": 5.0535, "step": 9185 }, { "epoch": 0.74, - "learning_rate": 3.9520194949615595e-05, - "loss": 1.0205, + "learning_rate": 3.973129206693811e-05, + "loss": 4.9533, "step": 9190 }, { "epoch": 0.74, - "learning_rate": 3.9408920069940015e-05, - "loss": 0.9808, + "learning_rate": 3.9619912366410464e-05, + "loss": 5.1236, "step": 9195 }, { "epoch": 0.74, - "learning_rate": 3.92977636102259e-05, - "loss": 0.9922, + "learning_rate": 3.950865042181835e-05, + "loss": 5.0415, "step": 9200 }, { "epoch": 0.74, - "learning_rate": 3.918672578771879e-05, - "loss": 1.0167, + "learning_rate": 3.939750645014977e-05, + "loss": 5.125, "step": 9205 }, { "epoch": 0.74, - "learning_rate": 3.907580681943236e-05, - "loss": 0.9868, + "learning_rate": 3.928648066816267e-05, + "loss": 4.9983, "step": 9210 }, { "epoch": 0.74, - "learning_rate": 3.896500692214814e-05, - "loss": 1.0029, + "learning_rate": 3.917557329238443e-05, + "loss": 5.061, "step": 9215 }, { "epoch": 0.74, - "learning_rate": 3.885432631241472e-05, - "loss": 0.9545, + "learning_rate": 3.9064784539111563e-05, + "loss": 5.0054, "step": 9220 }, { "epoch": 0.74, - "learning_rate": 3.874376520654771e-05, - "loss": 0.9708, + "learning_rate": 3.895411462440924e-05, + "loss": 4.8839, "step": 9225 }, { "epoch": 0.74, - "learning_rate": 3.863332382062914e-05, - "loss": 0.9961, + "learning_rate": 3.884356376411089e-05, + "loss": 5.0891, "step": 9230 }, { "epoch": 0.74, - "learning_rate": 3.852300237050691e-05, - "loss": 0.9806, + "learning_rate": 3.873313217381762e-05, + "loss": 5.0029, "step": 9235 }, { "epoch": 0.74, - "learning_rate": 3.84128010717948e-05, - "loss": 0.9967, + "learning_rate": 3.862282006889817e-05, + "loss": 5.0136, "step": 9240 }, { "epoch": 0.74, - "learning_rate": 3.830272013987148e-05, - "loss": 1.0405, + "learning_rate": 3.851262766448801e-05, + "loss": 4.9121, "step": 9245 }, { "epoch": 0.74, - "learning_rate": 3.819275978988055e-05, - "loss": 0.9777, + "learning_rate": 3.840255517548932e-05, + "loss": 5.0205, "step": 9250 }, { "epoch": 0.74, - "learning_rate": 3.808292023672989e-05, - "loss": 0.9295, + "learning_rate": 3.829260281657037e-05, + "loss": 4.9742, "step": 9255 }, { "epoch": 0.74, - "learning_rate": 3.797320169509122e-05, - "loss": 1.0024, + "learning_rate": 3.818277080216514e-05, + "loss": 5.0591, "step": 9260 }, { "epoch": 0.74, - "learning_rate": 3.786360437939994e-05, - "loss": 0.9933, + "learning_rate": 3.807305934647294e-05, + "loss": 4.9452, "step": 9265 }, { "epoch": 0.74, - "learning_rate": 3.7754128503854325e-05, - "loss": 0.9987, + "learning_rate": 3.796346866345788e-05, + "loss": 4.9665, "step": 9270 }, { "epoch": 0.74, - "learning_rate": 3.76447742824154e-05, - "loss": 0.9327, + "learning_rate": 3.7853998966848626e-05, + "loss": 4.9772, "step": 9275 }, { "epoch": 0.74, - "learning_rate": 3.7535541928806476e-05, - "loss": 0.9727, + "learning_rate": 3.774465047013785e-05, + "loss": 4.9244, "step": 9280 }, { "epoch": 0.74, - "learning_rate": 3.7426431656512506e-05, - "loss": 0.9541, + "learning_rate": 3.7635423386581905e-05, + "loss": 5.0349, "step": 9285 }, { "epoch": 0.74, - "learning_rate": 3.731744367878007e-05, - "loss": 0.9857, + "learning_rate": 3.752631792920022e-05, + "loss": 5.0644, "step": 9290 }, { "epoch": 0.74, - "learning_rate": 3.7208578208616654e-05, - "loss": 0.9204, + "learning_rate": 3.741733431077517e-05, + "loss": 5.0834, "step": 9295 }, { "epoch": 0.74, - "learning_rate": 3.709983545879018e-05, - "loss": 1.045, + "learning_rate": 3.730847274385145e-05, + "loss": 4.976, "step": 9300 }, { - "epoch": 0.75, - "learning_rate": 3.69912156418289e-05, - "loss": 0.9924, + "epoch": 0.74, + "learning_rate": 3.719973344073574e-05, + "loss": 5.0478, "step": 9305 }, { - "epoch": 0.75, - "learning_rate": 3.688271897002072e-05, - "loss": 0.9713, + "epoch": 0.74, + "learning_rate": 3.709111661349628e-05, + "loss": 5.0398, "step": 9310 }, { "epoch": 0.75, - "learning_rate": 3.67743456554129e-05, - "loss": 1.0833, + "learning_rate": 3.698262247396242e-05, + "loss": 5.0307, "step": 9315 }, { "epoch": 0.75, - "learning_rate": 3.666609590981162e-05, - "loss": 0.9865, + "learning_rate": 3.6874251233724335e-05, + "loss": 4.9564, "step": 9320 }, { "epoch": 0.75, - "learning_rate": 3.655796994478146e-05, - "loss": 0.9411, + "learning_rate": 3.676600310413233e-05, + "loss": 4.9733, "step": 9325 }, { "epoch": 0.75, - "learning_rate": 3.64499679716452e-05, - "loss": 0.9584, + "learning_rate": 3.665787829629679e-05, + "loss": 5.0066, "step": 9330 }, { "epoch": 0.75, - "learning_rate": 3.6342090201483224e-05, - "loss": 0.9615, + "learning_rate": 3.654987702108752e-05, + "loss": 5.0851, "step": 9335 }, { "epoch": 0.75, - "learning_rate": 3.623433684513321e-05, - "loss": 0.9923, + "learning_rate": 3.644199948913345e-05, + "loss": 5.0005, "step": 9340 }, { "epoch": 0.75, - "learning_rate": 3.612670811318965e-05, - "loss": 1.0191, + "learning_rate": 3.633424591082208e-05, + "loss": 5.004, "step": 9345 }, { "epoch": 0.75, - "learning_rate": 3.601920421600351e-05, - "loss": 1.0379, + "learning_rate": 3.622661649629927e-05, + "loss": 4.9522, "step": 9350 }, { "epoch": 0.75, - "learning_rate": 3.5911825363681686e-05, - "loss": 0.998, + "learning_rate": 3.61191114554687e-05, + "loss": 4.9584, "step": 9355 }, { "epoch": 0.75, - "learning_rate": 3.580457176608678e-05, - "loss": 0.9515, + "learning_rate": 3.6011730997991486e-05, + "loss": 4.949, "step": 9360 }, { "epoch": 0.75, - "learning_rate": 3.569744363283655e-05, - "loss": 1.0181, + "learning_rate": 3.5904475333285814e-05, + "loss": 5.027, "step": 9365 }, { "epoch": 0.75, - "learning_rate": 3.559044117330355e-05, - "loss": 1.0432, + "learning_rate": 3.579734467052634e-05, + "loss": 5.0427, "step": 9370 }, { "epoch": 0.75, - "learning_rate": 3.548356459661476e-05, - "loss": 0.9644, + "learning_rate": 3.569033921864422e-05, + "loss": 4.9723, "step": 9375 }, { "epoch": 0.75, - "learning_rate": 3.537681411165103e-05, - "loss": 0.9789, + "learning_rate": 3.558345918632612e-05, + "loss": 5.043, "step": 9380 }, { "epoch": 0.75, - "learning_rate": 3.527018992704687e-05, - "loss": 0.9636, + "learning_rate": 3.5476704782014315e-05, + "loss": 5.0004, "step": 9385 }, { "epoch": 0.75, - "learning_rate": 3.516369225118992e-05, - "loss": 0.9661, + "learning_rate": 3.53700762139059e-05, + "loss": 4.9569, "step": 9390 }, { "epoch": 0.75, - "learning_rate": 3.5057321292220544e-05, - "loss": 1.0066, + "learning_rate": 3.5263573689952786e-05, + "loss": 5.0294, "step": 9395 }, { "epoch": 0.75, - "learning_rate": 3.495107725803153e-05, - "loss": 1.0663, + "learning_rate": 3.5157197417860856e-05, + "loss": 4.9503, "step": 9400 }, { "epoch": 0.75, - "learning_rate": 3.484496035626745e-05, - "loss": 0.9656, + "learning_rate": 3.5050947605089846e-05, + "loss": 4.9463, "step": 9405 }, { "epoch": 0.75, - "learning_rate": 3.473897079432451e-05, - "loss": 1.0136, + "learning_rate": 3.494482445885291e-05, + "loss": 5.0287, "step": 9410 }, { "epoch": 0.75, - "learning_rate": 3.463310877935013e-05, - "loss": 0.9041, + "learning_rate": 3.483882818611611e-05, + "loss": 4.9799, "step": 9415 }, { "epoch": 0.75, - "learning_rate": 3.452737451824226e-05, - "loss": 0.992, + "learning_rate": 3.4732958993598154e-05, + "loss": 4.9503, "step": 9420 }, { "epoch": 0.75, - "learning_rate": 3.442176821764927e-05, - "loss": 1.0022, + "learning_rate": 3.4627217087769736e-05, + "loss": 4.9665, "step": 9425 }, { - "epoch": 0.76, - "learning_rate": 3.4316290083969495e-05, - "loss": 0.9174, + "epoch": 0.75, + "learning_rate": 3.452160267485356e-05, + "loss": 4.9513, "step": 9430 }, { - "epoch": 0.76, - "learning_rate": 3.421094032335057e-05, - "loss": 0.9784, + "epoch": 0.75, + "learning_rate": 3.441611596082347e-05, + "loss": 4.9225, "step": 9435 }, { "epoch": 0.76, - "learning_rate": 3.410571914168955e-05, - "loss": 1.0017, + "learning_rate": 3.4310757151404415e-05, + "loss": 4.9737, "step": 9440 }, { "epoch": 0.76, - "learning_rate": 3.40006267446319e-05, - "loss": 0.9784, + "learning_rate": 3.420552645207174e-05, + "loss": 4.8908, "step": 9445 }, { "epoch": 0.76, - "learning_rate": 3.389566333757156e-05, - "loss": 0.9562, + "learning_rate": 3.410042406805115e-05, + "loss": 4.9651, "step": 9450 }, { "epoch": 0.76, - "learning_rate": 3.3790829125650346e-05, - "loss": 1.0183, + "learning_rate": 3.399545020431793e-05, + "loss": 5.046, "step": 9455 }, { "epoch": 0.76, - "learning_rate": 3.368612431375745e-05, - "loss": 0.979, + "learning_rate": 3.3890605065596804e-05, + "loss": 5.0278, "step": 9460 }, { "epoch": 0.76, - "learning_rate": 3.358154910652935e-05, - "loss": 1.0309, + "learning_rate": 3.3785888856361414e-05, + "loss": 5.0014, "step": 9465 }, { "epoch": 0.76, - "learning_rate": 3.347710370834916e-05, - "loss": 0.9754, + "learning_rate": 3.368130178083403e-05, + "loss": 5.031, "step": 9470 }, { "epoch": 0.76, - "learning_rate": 3.337278832334616e-05, - "loss": 1.0052, + "learning_rate": 3.3576844042984944e-05, + "loss": 5.0452, "step": 9475 }, { "epoch": 0.76, - "learning_rate": 3.3268603155395706e-05, - "loss": 0.9938, + "learning_rate": 3.3472515846532327e-05, + "loss": 5.0306, "step": 9480 }, { "epoch": 0.76, - "learning_rate": 3.3164548408118575e-05, - "loss": 0.9887, + "learning_rate": 3.336831739494167e-05, + "loss": 5.0094, "step": 9485 }, { "epoch": 0.76, - "learning_rate": 3.306062428488067e-05, - "loss": 0.9894, + "learning_rate": 3.326424889142543e-05, + "loss": 5.0456, "step": 9490 }, { "epoch": 0.76, - "learning_rate": 3.2956830988792616e-05, - "loss": 0.949, + "learning_rate": 3.316031053894267e-05, + "loss": 4.991, "step": 9495 }, { "epoch": 0.76, - "learning_rate": 3.285316872270926e-05, - "loss": 1.0484, + "learning_rate": 3.305650254019851e-05, + "loss": 4.9823, "step": 9500 }, { "epoch": 0.76, - "learning_rate": 3.274963768922946e-05, - "loss": 0.94, + "learning_rate": 3.295282509764405e-05, + "loss": 5.004, "step": 9505 }, { "epoch": 0.76, - "learning_rate": 3.2646238090695555e-05, - "loss": 0.9376, + "learning_rate": 3.2849278413475546e-05, + "loss": 4.9785, "step": 9510 }, { "epoch": 0.76, - "learning_rate": 3.2542970129193004e-05, - "loss": 1.0034, + "learning_rate": 3.274586268963443e-05, + "loss": 4.9196, "step": 9515 }, { "epoch": 0.76, - "learning_rate": 3.243983400655005e-05, - "loss": 1.0006, + "learning_rate": 3.264257812780657e-05, + "loss": 5.0443, "step": 9520 }, { "epoch": 0.76, - "learning_rate": 3.233682992433713e-05, - "loss": 0.9506, + "learning_rate": 3.253942492942221e-05, + "loss": 5.0396, "step": 9525 }, { "epoch": 0.76, - "learning_rate": 3.2233958083866747e-05, - "loss": 0.9542, + "learning_rate": 3.2436403295655246e-05, + "loss": 4.9318, "step": 9530 }, { "epoch": 0.76, - "learning_rate": 3.213121868619291e-05, - "loss": 0.9756, + "learning_rate": 3.233351342742307e-05, + "loss": 4.981, "step": 9535 }, { "epoch": 0.76, - "learning_rate": 3.202861193211076e-05, - "loss": 0.9305, + "learning_rate": 3.223075552538609e-05, + "loss": 4.9344, "step": 9540 }, { "epoch": 0.76, - "learning_rate": 3.192613802215625e-05, - "loss": 1.061, + "learning_rate": 3.2128129789947346e-05, + "loss": 5.0192, "step": 9545 }, { "epoch": 0.76, - "learning_rate": 3.182379715660566e-05, - "loss": 0.9825, + "learning_rate": 3.202563642125215e-05, + "loss": 5.0099, "step": 9550 }, { - "epoch": 0.77, - "learning_rate": 3.1721589535475216e-05, - "loss": 0.9669, + "epoch": 0.76, + "learning_rate": 3.1923275619187554e-05, + "loss": 4.9649, "step": 9555 }, { - "epoch": 0.77, - "learning_rate": 3.161951535852079e-05, - "loss": 1.0307, + "epoch": 0.76, + "learning_rate": 3.18210475833822e-05, + "loss": 5.0004, "step": 9560 }, { "epoch": 0.77, - "learning_rate": 3.1517574825237414e-05, - "loss": 0.9451, + "learning_rate": 3.171895251320576e-05, + "loss": 4.9702, "step": 9565 }, { "epoch": 0.77, - "learning_rate": 3.1415768134858945e-05, - "loss": 0.9048, + "learning_rate": 3.1616990607768594e-05, + "loss": 4.9164, "step": 9570 }, { "epoch": 0.77, - "learning_rate": 3.1314095486357684e-05, - "loss": 0.9757, + "learning_rate": 3.1515162065921276e-05, + "loss": 5.0323, "step": 9575 }, { "epoch": 0.77, - "learning_rate": 3.121255707844386e-05, - "loss": 0.9962, + "learning_rate": 3.1413467086254486e-05, + "loss": 4.9166, "step": 9580 }, { "epoch": 0.77, - "learning_rate": 3.111115310956539e-05, - "loss": 0.9592, + "learning_rate": 3.1311905867098204e-05, + "loss": 4.9246, "step": 9585 }, { "epoch": 0.77, - "learning_rate": 3.1009883777907554e-05, - "loss": 0.953, + "learning_rate": 3.1210478606521664e-05, + "loss": 5.0604, "step": 9590 }, { "epoch": 0.77, - "learning_rate": 3.09087492813923e-05, - "loss": 0.9615, + "learning_rate": 3.110918550233284e-05, + "loss": 4.9613, "step": 9595 }, { "epoch": 0.77, - "learning_rate": 3.080774981767819e-05, - "loss": 0.9818, + "learning_rate": 3.100802675207804e-05, + "loss": 4.9791, "step": 9600 }, { "epoch": 0.77, - "learning_rate": 3.070688558415985e-05, - "loss": 0.9719, + "learning_rate": 3.090700255304163e-05, + "loss": 5.0427, "step": 9605 }, { "epoch": 0.77, - "learning_rate": 3.060615677796751e-05, - "loss": 0.926, + "learning_rate": 3.080611310224539e-05, + "loss": 4.9251, "step": 9610 }, { "epoch": 0.77, - "learning_rate": 3.0505563595966924e-05, - "loss": 0.9681, + "learning_rate": 3.0705358596448486e-05, + "loss": 4.8908, "step": 9615 }, { "epoch": 0.77, - "learning_rate": 3.0405106234758564e-05, - "loss": 1.0099, + "learning_rate": 3.0604739232146816e-05, + "loss": 4.9492, "step": 9620 }, { "epoch": 0.77, - "learning_rate": 3.0304784890677585e-05, - "loss": 0.9947, + "learning_rate": 3.0504255205572795e-05, + "loss": 4.9952, "step": 9625 }, { "epoch": 0.77, - "learning_rate": 3.0204599759793307e-05, - "loss": 1.0186, + "learning_rate": 3.0403906712694784e-05, + "loss": 4.9373, "step": 9630 }, { "epoch": 0.77, - "learning_rate": 3.01045510379087e-05, - "loss": 0.9176, + "learning_rate": 3.0303693949216895e-05, + "loss": 5.0111, "step": 9635 }, { "epoch": 0.77, - "learning_rate": 3.0004638920560345e-05, - "loss": 1.0455, + "learning_rate": 3.0203617110578542e-05, + "loss": 4.9949, "step": 9640 }, { "epoch": 0.77, - "learning_rate": 2.990486360301765e-05, - "loss": 0.9578, + "learning_rate": 3.0103676391954016e-05, + "loss": 4.9145, "step": 9645 }, { "epoch": 0.77, - "learning_rate": 2.9805225280282778e-05, - "loss": 1.0709, + "learning_rate": 3.0003871988252208e-05, + "loss": 4.9915, "step": 9650 }, { "epoch": 0.77, - "learning_rate": 2.970572414709012e-05, - "loss": 1.0115, + "learning_rate": 2.9904204094115994e-05, + "loss": 4.933, "step": 9655 }, { "epoch": 0.77, - "learning_rate": 2.9606360397905842e-05, - "loss": 0.937, + "learning_rate": 2.980467290392227e-05, + "loss": 4.9785, "step": 9660 }, { "epoch": 0.77, - "learning_rate": 2.950713422692779e-05, - "loss": 0.9742, + "learning_rate": 2.9705278611781118e-05, + "loss": 4.9364, "step": 9665 }, { "epoch": 0.77, - "learning_rate": 2.940804582808483e-05, - "loss": 0.9758, + "learning_rate": 2.9606021411535723e-05, + "loss": 5.0327, "step": 9670 }, { "epoch": 0.77, - "learning_rate": 2.9309095395036502e-05, - "loss": 1.0153, + "learning_rate": 2.9506901496761906e-05, + "loss": 4.9414, "step": 9675 }, { - "epoch": 0.78, - "learning_rate": 2.9210283121172787e-05, - "loss": 0.9505, + "epoch": 0.77, + "learning_rate": 2.9407919060767775e-05, + "loss": 4.9666, "step": 9680 }, { - "epoch": 0.78, - "learning_rate": 2.911160919961362e-05, - "loss": 0.9487, + "epoch": 0.77, + "learning_rate": 2.9309074296593207e-05, + "loss": 5.0441, "step": 9685 }, { "epoch": 0.78, - "learning_rate": 2.901307382320857e-05, - "loss": 0.9983, + "learning_rate": 2.9210367397009696e-05, + "loss": 5.0452, "step": 9690 }, { "epoch": 0.78, - "learning_rate": 2.891467718453641e-05, - "loss": 1.0189, + "learning_rate": 2.9111798554519852e-05, + "loss": 4.9842, "step": 9695 }, { "epoch": 0.78, - "learning_rate": 2.88164194759047e-05, - "loss": 0.9822, + "learning_rate": 2.9013367961357007e-05, + "loss": 5.0173, "step": 9700 }, { "epoch": 0.78, - "learning_rate": 2.871830088934957e-05, - "loss": 0.9703, + "learning_rate": 2.8915075809484904e-05, + "loss": 5.0165, "step": 9705 }, { "epoch": 0.78, - "learning_rate": 2.8620321616635193e-05, - "loss": 1.0034, + "learning_rate": 2.8816922290597216e-05, + "loss": 5.0191, "step": 9710 }, { "epoch": 0.78, - "learning_rate": 2.852248184925349e-05, - "loss": 0.9629, + "learning_rate": 2.8718907596117393e-05, + "loss": 4.9786, "step": 9715 }, { "epoch": 0.78, - "learning_rate": 2.8424781778423726e-05, - "loss": 1.0557, + "learning_rate": 2.8621031917197983e-05, + "loss": 5.0337, "step": 9720 }, { "epoch": 0.78, - "learning_rate": 2.8327221595092158e-05, - "loss": 0.9874, + "learning_rate": 2.8523295444720556e-05, + "loss": 4.8536, "step": 9725 }, { "epoch": 0.78, - "learning_rate": 2.8229801489931563e-05, - "loss": 1.0022, + "learning_rate": 2.842569836929504e-05, + "loss": 4.963, "step": 9730 }, { "epoch": 0.78, - "learning_rate": 2.813252165334104e-05, - "loss": 1.0084, + "learning_rate": 2.8328240881259716e-05, + "loss": 4.9628, "step": 9735 }, { "epoch": 0.78, - "learning_rate": 2.803538227544551e-05, - "loss": 1.0106, + "learning_rate": 2.8230923170680423e-05, + "loss": 5.0177, "step": 9740 }, { "epoch": 0.78, - "learning_rate": 2.7938383546095392e-05, - "loss": 0.9991, + "learning_rate": 2.8133745427350522e-05, + "loss": 4.9424, "step": 9745 }, { "epoch": 0.78, - "learning_rate": 2.784152565486624e-05, - "loss": 1.0146, + "learning_rate": 2.8036707840790387e-05, + "loss": 4.9184, "step": 9750 }, { "epoch": 0.78, - "learning_rate": 2.7744808791058253e-05, - "loss": 0.973, + "learning_rate": 2.793981060024703e-05, + "loss": 4.9443, "step": 9755 }, { "epoch": 0.78, - "learning_rate": 2.764823314369612e-05, - "loss": 0.9666, + "learning_rate": 2.7843053894693803e-05, + "loss": 5.0721, "step": 9760 }, { "epoch": 0.78, - "learning_rate": 2.755179890152847e-05, - "loss": 1.0096, + "learning_rate": 2.7746437912829837e-05, + "loss": 4.9967, "step": 9765 }, { "epoch": 0.78, - "learning_rate": 2.7455506253027608e-05, - "loss": 0.9971, + "learning_rate": 2.764996284308006e-05, + "loss": 4.9972, "step": 9770 }, { "epoch": 0.78, - "learning_rate": 2.7359355386389096e-05, - "loss": 0.9343, + "learning_rate": 2.7553628873594352e-05, + "loss": 5.0434, "step": 9775 }, { "epoch": 0.78, - "learning_rate": 2.7263346489531348e-05, - "loss": 0.9365, + "learning_rate": 2.7457436192247566e-05, + "loss": 4.969, "step": 9780 }, { "epoch": 0.78, - "learning_rate": 2.7167479750095325e-05, - "loss": 0.9659, + "learning_rate": 2.736138498663886e-05, + "loss": 4.9752, "step": 9785 }, { "epoch": 0.78, - "learning_rate": 2.7071755355444285e-05, - "loss": 0.9549, + "learning_rate": 2.7265475444091693e-05, + "loss": 5.0462, "step": 9790 }, { "epoch": 0.78, - "learning_rate": 2.697617349266307e-05, - "loss": 0.9743, + "learning_rate": 2.716970775165305e-05, + "loss": 5.0037, "step": 9795 }, { "epoch": 0.78, - "learning_rate": 2.688073434855811e-05, - "loss": 0.9643, + "learning_rate": 2.707408209609339e-05, + "loss": 4.9763, "step": 9800 }, { - "epoch": 0.79, - "learning_rate": 2.678543810965688e-05, - "loss": 0.9791, + "epoch": 0.78, + "learning_rate": 2.697859866390605e-05, + "loss": 4.9082, "step": 9805 }, { - "epoch": 0.79, - "learning_rate": 2.6690284962207446e-05, - "loss": 1.0109, + "epoch": 0.78, + "learning_rate": 2.688325764130718e-05, + "loss": 4.9634, "step": 9810 }, { "epoch": 0.79, - "learning_rate": 2.659527509217842e-05, - "loss": 0.9803, + "learning_rate": 2.6788059214235005e-05, + "loss": 4.9847, "step": 9815 }, { "epoch": 0.79, - "learning_rate": 2.6500408685258206e-05, - "loss": 1.0278, + "learning_rate": 2.669300356834976e-05, + "loss": 4.9768, "step": 9820 }, { "epoch": 0.79, - "learning_rate": 2.6405685926854895e-05, - "loss": 0.9688, + "learning_rate": 2.65980908890332e-05, + "loss": 4.978, "step": 9825 }, { "epoch": 0.79, - "learning_rate": 2.631110700209587e-05, - "loss": 1.0073, + "learning_rate": 2.6503321361388265e-05, + "loss": 4.9786, "step": 9830 }, { "epoch": 0.79, - "learning_rate": 2.6216672095827266e-05, - "loss": 0.983, + "learning_rate": 2.6408695170238728e-05, + "loss": 5.0729, "step": 9835 }, { "epoch": 0.79, - "learning_rate": 2.6122381392613958e-05, - "loss": 0.9412, + "learning_rate": 2.6314212500128722e-05, + "loss": 4.9542, "step": 9840 }, { "epoch": 0.79, - "learning_rate": 2.6028235076738795e-05, - "loss": 0.9504, + "learning_rate": 2.621987353532266e-05, + "loss": 4.939, "step": 9845 }, { "epoch": 0.79, - "learning_rate": 2.5934233332202517e-05, - "loss": 0.9264, + "learning_rate": 2.6125678459804527e-05, + "loss": 4.9835, "step": 9850 }, { "epoch": 0.79, - "learning_rate": 2.5840376342723325e-05, - "loss": 0.989, + "learning_rate": 2.6031627457277784e-05, + "loss": 5.0469, "step": 9855 }, { "epoch": 0.79, - "learning_rate": 2.5746664291736487e-05, - "loss": 1.0181, + "learning_rate": 2.5937720711164826e-05, + "loss": 4.8895, "step": 9860 }, { "epoch": 0.79, - "learning_rate": 2.5653097362393985e-05, - "loss": 0.9605, + "learning_rate": 2.584395840460686e-05, + "loss": 4.9117, "step": 9865 }, { "epoch": 0.79, - "learning_rate": 2.5559675737564236e-05, - "loss": 0.9863, + "learning_rate": 2.575034072046324e-05, + "loss": 5.1073, "step": 9870 }, { "epoch": 0.79, - "learning_rate": 2.546639959983157e-05, - "loss": 0.9989, + "learning_rate": 2.5656867841311373e-05, + "loss": 4.9463, "step": 9875 }, { "epoch": 0.79, - "learning_rate": 2.537326913149606e-05, - "loss": 0.9874, + "learning_rate": 2.5563539949446224e-05, + "loss": 4.9612, "step": 9880 }, { "epoch": 0.79, - "learning_rate": 2.528028451457304e-05, - "loss": 0.9204, + "learning_rate": 2.5470357226880003e-05, + "loss": 4.9686, "step": 9885 }, { "epoch": 0.79, - "learning_rate": 2.518744593079282e-05, - "loss": 0.9713, + "learning_rate": 2.537731985534183e-05, + "loss": 4.9456, "step": 9890 }, { "epoch": 0.79, - "learning_rate": 2.5094753561600304e-05, - "loss": 0.9857, + "learning_rate": 2.5284428016277284e-05, + "loss": 4.9621, "step": 9895 }, { "epoch": 0.79, - "learning_rate": 2.5002207588154557e-05, - "loss": 0.965, + "learning_rate": 2.5191681890848186e-05, + "loss": 4.9526, "step": 9900 }, { "epoch": 0.79, - "learning_rate": 2.490980819132862e-05, - "loss": 0.9601, + "learning_rate": 2.5099081659932154e-05, + "loss": 5.0168, "step": 9905 }, { "epoch": 0.79, - "learning_rate": 2.4817555551709016e-05, - "loss": 0.9968, + "learning_rate": 2.5006627504122337e-05, + "loss": 4.9725, "step": 9910 }, { "epoch": 0.79, - "learning_rate": 2.472544984959545e-05, - "loss": 0.9837, + "learning_rate": 2.491431960372682e-05, + "loss": 4.9492, "step": 9915 }, { "epoch": 0.79, - "learning_rate": 2.4633491265000464e-05, - "loss": 0.9728, + "learning_rate": 2.4822158138768736e-05, + "loss": 4.9784, "step": 9920 }, { "epoch": 0.79, - "learning_rate": 2.454167997764909e-05, - "loss": 0.9543, + "learning_rate": 2.4730143288985365e-05, + "loss": 4.9615, "step": 9925 }, { - "epoch": 0.8, - "learning_rate": 2.4450016166978385e-05, - "loss": 1.0215, + "epoch": 0.79, + "learning_rate": 2.463827523382819e-05, + "loss": 4.9774, "step": 9930 }, { - "epoch": 0.8, - "learning_rate": 2.4358500012137275e-05, - "loss": 0.9852, + "epoch": 0.79, + "learning_rate": 2.4546554152462407e-05, + "loss": 4.9399, "step": 9935 }, { "epoch": 0.8, - "learning_rate": 2.426713169198607e-05, - "loss": 0.9404, + "learning_rate": 2.4454980223766532e-05, + "loss": 4.9834, "step": 9940 }, { "epoch": 0.8, - "learning_rate": 2.417591138509616e-05, - "loss": 0.9842, + "learning_rate": 2.436355362633216e-05, + "loss": 5.0893, "step": 9945 }, { "epoch": 0.8, - "learning_rate": 2.408483926974968e-05, - "loss": 1.0293, + "learning_rate": 2.4272274538463446e-05, + "loss": 4.9906, "step": 9950 }, { "epoch": 0.8, - "learning_rate": 2.3993915523939048e-05, - "loss": 1.0041, + "learning_rate": 2.4181143138176965e-05, + "loss": 5.0692, "step": 9955 }, { "epoch": 0.8, - "learning_rate": 2.390314032536678e-05, - "loss": 0.9252, + "learning_rate": 2.4090159603201225e-05, + "loss": 5.0023, "step": 9960 }, { "epoch": 0.8, - "learning_rate": 2.3812513851445072e-05, - "loss": 0.9616, + "learning_rate": 2.3999324110976408e-05, + "loss": 4.9086, "step": 9965 }, { "epoch": 0.8, - "learning_rate": 2.3722036279295436e-05, - "loss": 0.9426, + "learning_rate": 2.3908636838653876e-05, + "loss": 4.923, "step": 9970 }, { "epoch": 0.8, - "learning_rate": 2.363170778574837e-05, - "loss": 0.9859, + "learning_rate": 2.3818097963095997e-05, + "loss": 4.9943, "step": 9975 }, { "epoch": 0.8, - "learning_rate": 2.354152854734297e-05, - "loss": 1.0323, + "learning_rate": 2.372770766087574e-05, + "loss": 4.964, "step": 9980 }, { "epoch": 0.8, - "learning_rate": 2.3451498740326693e-05, - "loss": 0.9445, + "learning_rate": 2.3637466108276295e-05, + "loss": 4.9611, "step": 9985 }, { "epoch": 0.8, - "learning_rate": 2.3361618540654906e-05, - "loss": 0.9559, + "learning_rate": 2.354737348129077e-05, + "loss": 4.8797, "step": 9990 }, { "epoch": 0.8, - "learning_rate": 2.3271888123990594e-05, - "loss": 0.9383, + "learning_rate": 2.3457429955621747e-05, + "loss": 4.9202, "step": 9995 }, { "epoch": 0.8, - "learning_rate": 2.3182307665704006e-05, - "loss": 1.0412, + "learning_rate": 2.336763570668119e-05, + "loss": 4.9238, "step": 10000 }, { "epoch": 0.8, - "learning_rate": 2.309287734087233e-05, - "loss": 0.9623, + "learning_rate": 2.3277990909589775e-05, + "loss": 5.0442, "step": 10005 }, { "epoch": 0.8, - "learning_rate": 2.300359732427925e-05, - "loss": 1.0196, + "learning_rate": 2.3188495739176796e-05, + "loss": 4.9461, "step": 10010 }, { "epoch": 0.8, - "learning_rate": 2.2914467790414773e-05, - "loss": 0.9434, + "learning_rate": 2.3099150369979726e-05, + "loss": 4.8495, "step": 10015 }, { "epoch": 0.8, - "learning_rate": 2.2825488913474768e-05, - "loss": 0.9648, + "learning_rate": 2.300995497624391e-05, + "loss": 5.0033, "step": 10020 }, { "epoch": 0.8, - "learning_rate": 2.273666086736066e-05, - "loss": 1.0097, + "learning_rate": 2.2920909731922112e-05, + "loss": 4.9713, "step": 10025 }, { "epoch": 0.8, - "learning_rate": 2.264798382567912e-05, - "loss": 0.983, + "learning_rate": 2.2832014810674372e-05, + "loss": 5.0247, "step": 10030 }, { "epoch": 0.8, - "learning_rate": 2.2559457961741604e-05, - "loss": 0.9777, + "learning_rate": 2.274327038586751e-05, + "loss": 5.0136, "step": 10035 }, { "epoch": 0.8, - "learning_rate": 2.2471083448564155e-05, - "loss": 1.033, + "learning_rate": 2.2654676630574866e-05, + "loss": 4.927, "step": 10040 }, { "epoch": 0.8, - "learning_rate": 2.238286045886713e-05, - "loss": 0.9908, + "learning_rate": 2.2566233717575956e-05, + "loss": 4.928, "step": 10045 }, { "epoch": 0.8, - "learning_rate": 2.2294789165074536e-05, - "loss": 0.9577, + "learning_rate": 2.2477941819355986e-05, + "loss": 4.9323, "step": 10050 }, { - "epoch": 0.81, - "learning_rate": 2.220686973931404e-05, - "loss": 0.9806, + "epoch": 0.8, + "learning_rate": 2.238980110810588e-05, + "loss": 4.9816, "step": 10055 }, { - "epoch": 0.81, - "learning_rate": 2.21191023534165e-05, - "loss": 0.9791, + "epoch": 0.8, + "learning_rate": 2.2301811755721492e-05, + "loss": 4.9138, "step": 10060 }, { "epoch": 0.81, - "learning_rate": 2.2031487178915488e-05, - "loss": 0.9874, + "learning_rate": 2.2213973933803613e-05, + "loss": 5.0131, "step": 10065 }, { "epoch": 0.81, - "learning_rate": 2.1944024387047312e-05, - "loss": 0.9961, + "learning_rate": 2.212628781365741e-05, + "loss": 4.9139, "step": 10070 }, { "epoch": 0.81, - "learning_rate": 2.1856714148750256e-05, - "loss": 0.9587, + "learning_rate": 2.203875356629235e-05, + "loss": 4.9379, "step": 10075 }, { "epoch": 0.81, - "learning_rate": 2.1769556634664568e-05, - "loss": 1.0036, + "learning_rate": 2.1951371362421558e-05, + "loss": 4.9107, "step": 10080 }, { "epoch": 0.81, - "learning_rate": 2.1682552015132007e-05, - "loss": 1.069, + "learning_rate": 2.186414137246172e-05, + "loss": 4.9604, "step": 10085 }, { "epoch": 0.81, - "learning_rate": 2.15957004601954e-05, - "loss": 0.9999, + "learning_rate": 2.1777063766532648e-05, + "loss": 4.9101, "step": 10090 }, { "epoch": 0.81, - "learning_rate": 2.150900213959862e-05, - "loss": 0.9763, + "learning_rate": 2.1690138714456954e-05, + "loss": 5.0307, "step": 10095 }, { "epoch": 0.81, - "learning_rate": 2.1422457222785873e-05, - "loss": 0.9952, + "learning_rate": 2.160336638575978e-05, + "loss": 4.89, "step": 10100 }, { "epoch": 0.81, - "learning_rate": 2.1336065878901635e-05, - "loss": 0.9916, + "learning_rate": 2.1516746949668322e-05, + "loss": 4.8689, "step": 10105 }, { "epoch": 0.81, - "learning_rate": 2.1249828276790253e-05, - "loss": 0.9577, + "learning_rate": 2.1430280575111738e-05, + "loss": 4.9359, "step": 10110 }, { "epoch": 0.81, - "learning_rate": 2.1163744584995494e-05, - "loss": 0.9722, + "learning_rate": 2.1343967430720534e-05, + "loss": 4.9257, "step": 10115 }, { "epoch": 0.81, - "learning_rate": 2.1077814971760468e-05, - "loss": 1.0297, + "learning_rate": 2.1257807684826515e-05, + "loss": 4.9616, "step": 10120 }, { "epoch": 0.81, - "learning_rate": 2.0992039605027105e-05, - "loss": 1.0275, + "learning_rate": 2.1171801505462153e-05, + "loss": 4.8092, "step": 10125 }, { "epoch": 0.81, - "learning_rate": 2.0906418652435777e-05, - "loss": 1.0077, + "learning_rate": 2.1085949060360654e-05, + "loss": 4.9157, "step": 10130 }, { "epoch": 0.81, - "learning_rate": 2.082095228132518e-05, - "loss": 0.9626, + "learning_rate": 2.1000250516955178e-05, + "loss": 5.0368, "step": 10135 }, { "epoch": 0.81, - "learning_rate": 2.0735640658731836e-05, - "loss": 0.9889, + "learning_rate": 2.0914706042378916e-05, + "loss": 4.8821, "step": 10140 }, { "epoch": 0.81, - "learning_rate": 2.0650483951389865e-05, - "loss": 1.0404, + "learning_rate": 2.082931580346441e-05, + "loss": 4.8302, "step": 10145 }, { "epoch": 0.81, - "learning_rate": 2.05654823257306e-05, - "loss": 0.9664, + "learning_rate": 2.0744079966743635e-05, + "loss": 4.9077, "step": 10150 }, { "epoch": 0.81, - "learning_rate": 2.0480635947882253e-05, - "loss": 0.9997, + "learning_rate": 2.0658998698447218e-05, + "loss": 4.97, "step": 10155 }, { "epoch": 0.81, - "learning_rate": 2.0395944983669647e-05, - "loss": 1.0282, + "learning_rate": 2.0574072164504478e-05, + "loss": 4.9289, "step": 10160 }, { "epoch": 0.81, - "learning_rate": 2.0311409598613852e-05, - "loss": 1.0173, + "learning_rate": 2.04893005305429e-05, + "loss": 4.831, "step": 10165 }, { "epoch": 0.81, - "learning_rate": 2.02270299579319e-05, - "loss": 0.9882, + "learning_rate": 2.0404683961887926e-05, + "loss": 4.9085, "step": 10170 }, { "epoch": 0.81, - "learning_rate": 2.0142806226536393e-05, - "loss": 0.9626, + "learning_rate": 2.0320222623562568e-05, + "loss": 4.8878, "step": 10175 }, { - "epoch": 0.82, - "learning_rate": 2.0058738569035286e-05, - "loss": 1.0019, + "epoch": 0.81, + "learning_rate": 2.0235916680287015e-05, + "loss": 4.9473, "step": 10180 }, { - "epoch": 0.82, - "learning_rate": 1.9974827149731377e-05, - "loss": 0.9847, + "epoch": 0.81, + "learning_rate": 2.015176629647857e-05, + "loss": 4.904, "step": 10185 }, { "epoch": 0.82, - "learning_rate": 1.9891072132622246e-05, - "loss": 0.9968, + "learning_rate": 2.0067771636250975e-05, + "loss": 4.9011, "step": 10190 }, { "epoch": 0.82, - "learning_rate": 1.9807473681399713e-05, - "loss": 0.9671, + "learning_rate": 1.9983932863414433e-05, + "loss": 4.9563, "step": 10195 }, { "epoch": 0.82, - "learning_rate": 1.972403195944965e-05, - "loss": 0.9926, + "learning_rate": 1.9900250141474953e-05, + "loss": 4.8941, "step": 10200 }, { "epoch": 0.82, - "learning_rate": 1.9640747129851633e-05, - "loss": 1.0598, + "learning_rate": 1.9816723633634417e-05, + "loss": 5.0055, "step": 10205 }, { "epoch": 0.82, - "learning_rate": 1.9557619355378486e-05, - "loss": 0.9608, + "learning_rate": 1.9733353502789855e-05, + "loss": 4.8735, "step": 10210 }, { "epoch": 0.82, - "learning_rate": 1.947464879849622e-05, - "loss": 1.0122, + "learning_rate": 1.965013991153345e-05, + "loss": 4.9426, "step": 10215 }, { "epoch": 0.82, - "learning_rate": 1.9391835621363518e-05, - "loss": 1.0054, + "learning_rate": 1.9567083022152034e-05, + "loss": 4.8942, "step": 10220 }, { "epoch": 0.82, - "learning_rate": 1.9309179985831494e-05, - "loss": 0.9614, + "learning_rate": 1.9484182996626853e-05, + "loss": 4.8889, "step": 10225 }, { "epoch": 0.82, - "learning_rate": 1.922668205344336e-05, - "loss": 0.9838, + "learning_rate": 1.9401439996633274e-05, + "loss": 4.9423, "step": 10230 }, { "epoch": 0.82, - "learning_rate": 1.9144341985434056e-05, - "loss": 0.9346, + "learning_rate": 1.9318854183540313e-05, + "loss": 4.9184, "step": 10235 }, { "epoch": 0.82, - "learning_rate": 1.906215994273004e-05, - "loss": 0.9748, + "learning_rate": 1.9236425718410533e-05, + "loss": 4.9198, "step": 10240 }, { "epoch": 0.82, - "learning_rate": 1.898013608594901e-05, - "loss": 1.0003, + "learning_rate": 1.9154154761999575e-05, + "loss": 4.9385, "step": 10245 }, { "epoch": 0.82, - "learning_rate": 1.8898270575399303e-05, - "loss": 1.0718, + "learning_rate": 1.9072041474755985e-05, + "loss": 4.9499, "step": 10250 }, { "epoch": 0.82, - "learning_rate": 1.8816563571079958e-05, - "loss": 0.9969, + "learning_rate": 1.899008601682064e-05, + "loss": 4.8759, "step": 10255 }, { "epoch": 0.82, - "learning_rate": 1.8735015232680154e-05, - "loss": 1.0109, + "learning_rate": 1.8908288548026855e-05, + "loss": 4.9409, "step": 10260 }, { "epoch": 0.82, - "learning_rate": 1.8653625719578926e-05, - "loss": 0.947, + "learning_rate": 1.8826649227899596e-05, + "loss": 4.8672, "step": 10265 }, { "epoch": 0.82, - "learning_rate": 1.857239519084505e-05, - "loss": 1.0247, + "learning_rate": 1.8745168215655527e-05, + "loss": 4.9549, "step": 10270 }, { "epoch": 0.82, - "learning_rate": 1.8491323805236394e-05, - "loss": 1.0043, + "learning_rate": 1.8663845670202563e-05, + "loss": 4.9283, "step": 10275 }, { "epoch": 0.82, - "learning_rate": 1.8410411721199927e-05, - "loss": 0.9616, + "learning_rate": 1.8582681750139543e-05, + "loss": 4.9455, "step": 10280 }, { "epoch": 0.82, - "learning_rate": 1.8329659096871244e-05, - "loss": 0.9067, + "learning_rate": 1.850167661375599e-05, + "loss": 4.9036, "step": 10285 }, { "epoch": 0.82, - "learning_rate": 1.8249066090074217e-05, - "loss": 0.9895, + "learning_rate": 1.8420830419031665e-05, + "loss": 5.007, "step": 10290 }, { "epoch": 0.82, - "learning_rate": 1.816863285832089e-05, - "loss": 0.9848, + "learning_rate": 1.834014332363645e-05, + "loss": 4.9747, "step": 10295 }, { "epoch": 0.82, - "learning_rate": 1.8088359558810975e-05, - "loss": 0.9577, + "learning_rate": 1.8259615484929915e-05, + "loss": 5.0292, "step": 10300 }, { - "epoch": 0.83, - "learning_rate": 1.8008246348431557e-05, - "loss": 0.9664, + "epoch": 0.82, + "learning_rate": 1.8179247059961058e-05, + "loss": 4.9304, "step": 10305 }, { - "epoch": 0.83, - "learning_rate": 1.792829338375691e-05, - "loss": 0.9858, + "epoch": 0.82, + "learning_rate": 1.80990382054679e-05, + "loss": 5.0413, "step": 10310 }, { "epoch": 0.83, - "learning_rate": 1.784850082104813e-05, - "loss": 1.0126, + "learning_rate": 1.801898907787737e-05, + "loss": 4.9687, "step": 10315 }, { "epoch": 0.83, - "learning_rate": 1.776886881625277e-05, - "loss": 0.9863, + "learning_rate": 1.7939099833304806e-05, + "loss": 4.8533, "step": 10320 }, { "epoch": 0.83, - "learning_rate": 1.7689397525004647e-05, - "loss": 1.0248, + "learning_rate": 1.7859370627553807e-05, + "loss": 4.9037, "step": 10325 }, { "epoch": 0.83, - "learning_rate": 1.7610087102623395e-05, - "loss": 0.977, + "learning_rate": 1.777980161611582e-05, + "loss": 4.9753, "step": 10330 }, { "epoch": 0.83, - "learning_rate": 1.7530937704114314e-05, - "loss": 0.9528, + "learning_rate": 1.770039295416982e-05, + "loss": 4.9331, "step": 10335 }, { "epoch": 0.83, - "learning_rate": 1.7451949484167972e-05, - "loss": 1.0389, + "learning_rate": 1.762114479658219e-05, + "loss": 4.8514, "step": 10340 }, { "epoch": 0.83, - "learning_rate": 1.737312259715993e-05, - "loss": 1.0462, + "learning_rate": 1.7542057297906155e-05, + "loss": 5.0104, "step": 10345 }, { "epoch": 0.83, - "learning_rate": 1.729445719715046e-05, - "loss": 0.9434, + "learning_rate": 1.7463130612381705e-05, + "loss": 5.0094, "step": 10350 }, { "epoch": 0.83, - "learning_rate": 1.721595343788417e-05, - "loss": 1.036, + "learning_rate": 1.7384364893935178e-05, + "loss": 4.9191, "step": 10355 }, { "epoch": 0.83, - "learning_rate": 1.7137611472789783e-05, - "loss": 1.0053, + "learning_rate": 1.7305760296178995e-05, + "loss": 4.964, "step": 10360 }, { "epoch": 0.83, - "learning_rate": 1.7059431454979824e-05, - "loss": 1.0535, + "learning_rate": 1.722731697241131e-05, + "loss": 4.9417, "step": 10365 }, { "epoch": 0.83, - "learning_rate": 1.69814135372503e-05, - "loss": 0.9465, + "learning_rate": 1.7149035075615794e-05, + "loss": 4.9297, "step": 10370 }, { "epoch": 0.83, - "learning_rate": 1.690355787208039e-05, - "loss": 0.9424, + "learning_rate": 1.7070914758461298e-05, + "loss": 4.8719, "step": 10375 }, { "epoch": 0.83, - "learning_rate": 1.6825864611632204e-05, - "loss": 1.0065, + "learning_rate": 1.6992956173301556e-05, + "loss": 4.9464, "step": 10380 }, { "epoch": 0.83, - "learning_rate": 1.674833390775037e-05, - "loss": 0.9769, + "learning_rate": 1.691515947217487e-05, + "loss": 4.934, "step": 10385 }, { "epoch": 0.83, - "learning_rate": 1.667096591196189e-05, - "loss": 1.0145, + "learning_rate": 1.6837524806803773e-05, + "loss": 4.966, "step": 10390 }, { "epoch": 0.83, - "learning_rate": 1.659376077547573e-05, - "loss": 0.9637, + "learning_rate": 1.6760052328594944e-05, + "loss": 4.8967, "step": 10395 }, { "epoch": 0.83, - "learning_rate": 1.651671864918257e-05, - "loss": 0.93, + "learning_rate": 1.6682742188638578e-05, + "loss": 4.9828, "step": 10400 }, { "epoch": 0.83, - "learning_rate": 1.6439839683654547e-05, - "loss": 0.9901, + "learning_rate": 1.6605594537708413e-05, + "loss": 4.8161, "step": 10405 }, { "epoch": 0.83, - "learning_rate": 1.6363124029144793e-05, - "loss": 1.0324, + "learning_rate": 1.652860952626115e-05, + "loss": 4.8736, "step": 10410 }, { "epoch": 0.83, - "learning_rate": 1.6286571835587362e-05, - "loss": 1.0153, + "learning_rate": 1.6451787304436484e-05, + "loss": 4.8816, "step": 10415 }, { "epoch": 0.83, - "learning_rate": 1.6210183252596866e-05, - "loss": 0.9636, + "learning_rate": 1.6375128022056475e-05, + "loss": 4.9506, "step": 10420 }, { "epoch": 0.83, - "learning_rate": 1.6133958429468042e-05, - "loss": 0.9809, + "learning_rate": 1.6298631828625466e-05, + "loss": 4.8099, "step": 10425 }, { - "epoch": 0.84, - "learning_rate": 1.6057897515175656e-05, - "loss": 0.9841, + "epoch": 0.83, + "learning_rate": 1.6222298873329766e-05, + "loss": 5.0062, "step": 10430 }, { - "epoch": 0.84, - "learning_rate": 1.598200065837413e-05, - "loss": 1.0245, + "epoch": 0.83, + "learning_rate": 1.6146129305037293e-05, + "loss": 4.9855, "step": 10435 }, { "epoch": 0.84, - "learning_rate": 1.5906268007397162e-05, - "loss": 0.9906, + "learning_rate": 1.6070123272297355e-05, + "loss": 4.9093, "step": 10440 }, { "epoch": 0.84, - "learning_rate": 1.583069971025767e-05, - "loss": 0.9538, + "learning_rate": 1.5994280923340233e-05, + "loss": 4.9137, "step": 10445 }, { "epoch": 0.84, - "learning_rate": 1.5755295914647218e-05, - "loss": 0.9611, + "learning_rate": 1.591860240607713e-05, + "loss": 4.9075, "step": 10450 }, { "epoch": 0.84, - "learning_rate": 1.5680056767935935e-05, - "loss": 1.0573, + "learning_rate": 1.5843087868099614e-05, + "loss": 4.9025, "step": 10455 }, { "epoch": 0.84, - "learning_rate": 1.5604982417172174e-05, - "loss": 0.9667, + "learning_rate": 1.5767737456679544e-05, + "loss": 4.8669, "step": 10460 }, { "epoch": 0.84, - "learning_rate": 1.5530073009082123e-05, - "loss": 0.9745, + "learning_rate": 1.5692551318768556e-05, + "loss": 4.9615, "step": 10465 }, { "epoch": 0.84, - "learning_rate": 1.5455328690069737e-05, - "loss": 0.9462, + "learning_rate": 1.5617529600998116e-05, + "loss": 4.9402, "step": 10470 }, { "epoch": 0.84, - "learning_rate": 1.5380749606216192e-05, - "loss": 1.0027, + "learning_rate": 1.554267244967885e-05, + "loss": 4.8771, "step": 10475 }, { "epoch": 0.84, - "learning_rate": 1.530633590327979e-05, - "loss": 0.983, + "learning_rate": 1.546798001080054e-05, + "loss": 4.8941, "step": 10480 }, { "epoch": 0.84, - "learning_rate": 1.523208772669562e-05, - "loss": 1.0118, + "learning_rate": 1.539345243003165e-05, + "loss": 4.9978, "step": 10485 }, { "epoch": 0.84, - "learning_rate": 1.515800522157519e-05, - "loss": 0.9697, + "learning_rate": 1.531908985271927e-05, + "loss": 4.9357, "step": 10490 }, { "epoch": 0.84, - "learning_rate": 1.508408853270632e-05, - "loss": 0.9969, + "learning_rate": 1.5244892423888547e-05, + "loss": 4.8976, "step": 10495 }, { "epoch": 0.84, - "learning_rate": 1.501033780455272e-05, - "loss": 0.9931, + "learning_rate": 1.5170860288242638e-05, + "loss": 4.9562, "step": 10500 }, { "epoch": 0.84, - "learning_rate": 1.4936753181253682e-05, - "loss": 1.0153, + "learning_rate": 1.509699359016229e-05, + "loss": 4.9231, "step": 10505 }, { "epoch": 0.84, - "learning_rate": 1.4863334806623941e-05, - "loss": 0.974, + "learning_rate": 1.5023292473705653e-05, + "loss": 5.0185, "step": 10510 }, { "epoch": 0.84, - "learning_rate": 1.4790082824153273e-05, - "loss": 0.9627, + "learning_rate": 1.494975708260793e-05, + "loss": 4.9146, "step": 10515 }, { "epoch": 0.84, - "learning_rate": 1.4716997377006259e-05, - "loss": 0.999, + "learning_rate": 1.4876387560281046e-05, + "loss": 4.8482, "step": 10520 }, { "epoch": 0.84, - "learning_rate": 1.464407860802205e-05, - "loss": 0.9902, + "learning_rate": 1.4803184049813623e-05, + "loss": 4.8852, "step": 10525 }, { "epoch": 0.84, - "learning_rate": 1.4571326659713935e-05, - "loss": 1.0094, + "learning_rate": 1.4730146693970315e-05, + "loss": 4.9723, "step": 10530 }, { "epoch": 0.84, - "learning_rate": 1.4498741674269268e-05, - "loss": 0.9889, + "learning_rate": 1.465727563519188e-05, + "loss": 4.902, "step": 10535 }, { "epoch": 0.84, - "learning_rate": 1.4426323793549024e-05, - "loss": 0.9264, + "learning_rate": 1.4584571015594629e-05, + "loss": 4.9873, "step": 10540 }, { "epoch": 0.84, - "learning_rate": 1.4354073159087632e-05, - "loss": 1.0019, + "learning_rate": 1.4512032976970458e-05, + "loss": 4.8962, "step": 10545 }, { "epoch": 0.84, - "learning_rate": 1.428198991209262e-05, - "loss": 0.9394, + "learning_rate": 1.4439661660786186e-05, + "loss": 4.976, "step": 10550 }, { - "epoch": 0.85, - "learning_rate": 1.421007419344441e-05, - "loss": 1.0235, + "epoch": 0.84, + "learning_rate": 1.436745720818361e-05, + "loss": 4.9049, "step": 10555 }, { - "epoch": 0.85, - "learning_rate": 1.4138326143695945e-05, - "loss": 1.0193, + "epoch": 0.84, + "learning_rate": 1.429541975997908e-05, + "loss": 4.9193, "step": 10560 }, { "epoch": 0.85, - "learning_rate": 1.4066745903072508e-05, - "loss": 0.9779, + "learning_rate": 1.4223549456663222e-05, + "loss": 4.8767, "step": 10565 }, { "epoch": 0.85, - "learning_rate": 1.3995333611471428e-05, - "loss": 1.015, + "learning_rate": 1.4151846438400718e-05, + "loss": 4.9104, "step": 10570 }, { "epoch": 0.85, - "learning_rate": 1.3924089408461782e-05, - "loss": 1.0415, + "learning_rate": 1.4080310845029954e-05, + "loss": 4.9104, "step": 10575 }, { "epoch": 0.85, - "learning_rate": 1.3853013433284134e-05, - "loss": 0.9548, + "learning_rate": 1.4008942816062853e-05, + "loss": 4.8696, "step": 10580 }, { "epoch": 0.85, - "learning_rate": 1.3782105824850212e-05, - "loss": 0.9735, + "learning_rate": 1.3937742490684525e-05, + "loss": 4.8456, "step": 10585 }, { "epoch": 0.85, - "learning_rate": 1.3711366721742769e-05, - "loss": 0.9732, + "learning_rate": 1.3866710007753014e-05, + "loss": 4.8943, "step": 10590 }, { "epoch": 0.85, - "learning_rate": 1.3640796262215172e-05, - "loss": 0.9657, + "learning_rate": 1.3795845505798998e-05, + "loss": 4.9618, "step": 10595 }, { "epoch": 0.85, - "learning_rate": 1.3570394584191216e-05, - "loss": 0.9689, + "learning_rate": 1.372514912302565e-05, + "loss": 4.876, "step": 10600 }, { "epoch": 0.85, - "learning_rate": 1.3500161825264834e-05, - "loss": 1.003, + "learning_rate": 1.3654620997308143e-05, + "loss": 4.9665, "step": 10605 }, { "epoch": 0.85, - "learning_rate": 1.3430098122699775e-05, - "loss": 1.0386, + "learning_rate": 1.3584261266193598e-05, + "loss": 4.9295, "step": 10610 }, { "epoch": 0.85, - "learning_rate": 1.336020361342939e-05, - "loss": 0.9886, + "learning_rate": 1.3514070066900686e-05, + "loss": 4.9928, "step": 10615 }, { "epoch": 0.85, - "learning_rate": 1.3290478434056453e-05, - "loss": 0.9456, + "learning_rate": 1.3444047536319404e-05, + "loss": 4.9679, "step": 10620 }, { "epoch": 0.85, - "learning_rate": 1.3220922720852657e-05, - "loss": 1.01, + "learning_rate": 1.3374193811010849e-05, + "loss": 4.9501, "step": 10625 }, { "epoch": 0.85, - "learning_rate": 1.3151536609758586e-05, - "loss": 1.048, + "learning_rate": 1.3304509027206812e-05, + "loss": 4.9037, "step": 10630 }, { "epoch": 0.85, - "learning_rate": 1.308232023638334e-05, - "loss": 0.9867, + "learning_rate": 1.3234993320809685e-05, + "loss": 4.8972, "step": 10635 }, { "epoch": 0.85, - "learning_rate": 1.3013273736004183e-05, - "loss": 0.9465, + "learning_rate": 1.3165646827392087e-05, + "loss": 4.9805, "step": 10640 }, { "epoch": 0.85, - "learning_rate": 1.2944397243566553e-05, - "loss": 1.0047, + "learning_rate": 1.3096469682196677e-05, + "loss": 4.8897, "step": 10645 }, { "epoch": 0.85, - "learning_rate": 1.2875690893683445e-05, - "loss": 0.9528, + "learning_rate": 1.302746202013574e-05, + "loss": 4.8904, "step": 10650 }, { "epoch": 0.85, - "learning_rate": 1.280715482063546e-05, - "loss": 0.9718, + "learning_rate": 1.2958623975791118e-05, + "loss": 4.8708, "step": 10655 }, { "epoch": 0.85, - "learning_rate": 1.2738789158370335e-05, - "loss": 0.977, + "learning_rate": 1.2889955683413846e-05, + "loss": 4.8514, "step": 10660 }, { "epoch": 0.85, - "learning_rate": 1.2670594040502738e-05, - "loss": 0.9491, + "learning_rate": 1.2821457276923864e-05, + "loss": 4.9443, "step": 10665 }, { "epoch": 0.85, - "learning_rate": 1.2602569600314095e-05, - "loss": 1.0231, + "learning_rate": 1.2753128889909848e-05, + "loss": 4.8813, "step": 10670 }, { "epoch": 0.85, - "learning_rate": 1.2534715970752243e-05, - "loss": 1.0009, + "learning_rate": 1.2684970655628803e-05, + "loss": 4.8926, "step": 10675 }, { - "epoch": 0.86, - "learning_rate": 1.2467033284431107e-05, - "loss": 1.0009, + "epoch": 0.85, + "learning_rate": 1.261698270700602e-05, + "loss": 4.921, "step": 10680 }, { - "epoch": 0.86, - "learning_rate": 1.2399521673630598e-05, - "loss": 0.9906, + "epoch": 0.85, + "learning_rate": 1.254916517663458e-05, + "loss": 4.9065, "step": 10685 }, { "epoch": 0.86, - "learning_rate": 1.233218127029625e-05, - "loss": 0.9816, + "learning_rate": 1.2481518196775256e-05, + "loss": 4.9316, "step": 10690 }, { "epoch": 0.86, - "learning_rate": 1.2265012206038972e-05, - "loss": 0.9863, + "learning_rate": 1.2414041899356199e-05, + "loss": 4.9455, "step": 10695 }, { "epoch": 0.86, - "learning_rate": 1.2198014612134855e-05, - "loss": 0.9872, + "learning_rate": 1.2346736415972715e-05, + "loss": 4.9388, "step": 10700 }, { "epoch": 0.86, - "learning_rate": 1.2131188619524792e-05, - "loss": 0.9804, + "learning_rate": 1.2279601877886927e-05, + "loss": 4.8605, "step": 10705 }, { "epoch": 0.86, - "learning_rate": 1.2064534358814338e-05, - "loss": 1.0112, + "learning_rate": 1.2212638416027599e-05, + "loss": 4.8845, "step": 10710 }, { "epoch": 0.86, - "learning_rate": 1.1998051960273438e-05, - "loss": 0.9994, + "learning_rate": 1.2145846160989871e-05, + "loss": 4.9234, "step": 10715 }, { "epoch": 0.86, - "learning_rate": 1.1931741553836096e-05, - "loss": 0.96, + "learning_rate": 1.2079225243034963e-05, + "loss": 4.9549, "step": 10720 }, { "epoch": 0.86, - "learning_rate": 1.1865603269100223e-05, - "loss": 1.007, + "learning_rate": 1.201277579208998e-05, + "loss": 4.9511, "step": 10725 }, { "epoch": 0.86, - "learning_rate": 1.1799637235327277e-05, - "loss": 0.9648, + "learning_rate": 1.1946497937747547e-05, + "loss": 4.8976, "step": 10730 }, { "epoch": 0.86, - "learning_rate": 1.1733843581442094e-05, - "loss": 0.9964, + "learning_rate": 1.1880391809265757e-05, + "loss": 4.8957, "step": 10735 }, { "epoch": 0.86, - "learning_rate": 1.1668222436032639e-05, - "loss": 1.0408, + "learning_rate": 1.1814457535567658e-05, + "loss": 4.9775, "step": 10740 }, { "epoch": 0.86, - "learning_rate": 1.1602773927349675e-05, - "loss": 1.0114, + "learning_rate": 1.174869524524127e-05, + "loss": 4.8755, "step": 10745 }, { "epoch": 0.86, - "learning_rate": 1.1537498183306594e-05, - "loss": 0.9888, + "learning_rate": 1.1683105066539068e-05, + "loss": 4.924, "step": 10750 }, { "epoch": 0.86, - "learning_rate": 1.1472395331479136e-05, - "loss": 0.9717, + "learning_rate": 1.1617687127378007e-05, + "loss": 4.9669, "step": 10755 }, { "epoch": 0.86, - "learning_rate": 1.1407465499105085e-05, - "loss": 0.9463, + "learning_rate": 1.1552441555339022e-05, + "loss": 4.9046, "step": 10760 }, { "epoch": 0.86, - "learning_rate": 1.1342708813084147e-05, - "loss": 1.0156, + "learning_rate": 1.1487368477666948e-05, + "loss": 4.9565, "step": 10765 }, { "epoch": 0.86, - "learning_rate": 1.1278125399977578e-05, - "loss": 0.9235, + "learning_rate": 1.1422468021270205e-05, + "loss": 4.9145, "step": 10770 }, { "epoch": 0.86, - "learning_rate": 1.1213715386008017e-05, - "loss": 1.0114, + "learning_rate": 1.1357740312720533e-05, + "loss": 4.9316, "step": 10775 }, { "epoch": 0.86, - "learning_rate": 1.1149478897059218e-05, - "loss": 0.9836, + "learning_rate": 1.1293185478252833e-05, + "loss": 4.899, "step": 10780 }, { "epoch": 0.86, - "learning_rate": 1.1085416058675723e-05, - "loss": 0.9689, + "learning_rate": 1.1228803643764762e-05, + "loss": 4.8608, "step": 10785 }, { "epoch": 0.86, - "learning_rate": 1.102152699606277e-05, - "loss": 0.993, + "learning_rate": 1.1164594934816664e-05, + "loss": 4.8999, "step": 10790 }, { "epoch": 0.86, - "learning_rate": 1.095781183408595e-05, - "loss": 0.9783, + "learning_rate": 1.1100559476631234e-05, + "loss": 4.9369, "step": 10795 }, { - "epoch": 0.87, - "learning_rate": 1.0894270697270959e-05, - "loss": 0.9583, + "epoch": 0.86, + "learning_rate": 1.1036697394093287e-05, + "loss": 4.905, "step": 10800 }, { - "epoch": 0.87, - "learning_rate": 1.0830903709803376e-05, - "loss": 0.9851, + "epoch": 0.86, + "learning_rate": 1.097300881174943e-05, + "loss": 4.9203, "step": 10805 }, { - "epoch": 0.87, - "learning_rate": 1.0767710995528468e-05, - "loss": 0.9995, + "epoch": 0.86, + "learning_rate": 1.0909493853808061e-05, + "loss": 4.9572, "step": 10810 }, { "epoch": 0.87, - "learning_rate": 1.070469267795079e-05, - "loss": 0.9973, + "learning_rate": 1.0846152644138829e-05, + "loss": 4.9991, "step": 10815 }, { "epoch": 0.87, - "learning_rate": 1.0641848880234218e-05, - "loss": 0.9429, + "learning_rate": 1.0782985306272608e-05, + "loss": 4.8933, "step": 10820 }, { "epoch": 0.87, - "learning_rate": 1.0579179725201382e-05, - "loss": 1.0226, + "learning_rate": 1.0719991963401087e-05, + "loss": 4.8714, "step": 10825 }, { "epoch": 0.87, - "learning_rate": 1.0516685335333687e-05, - "loss": 1.0218, + "learning_rate": 1.0657172738376786e-05, + "loss": 4.961, "step": 10830 }, { "epoch": 0.87, - "learning_rate": 1.0454365832770963e-05, - "loss": 0.9686, + "learning_rate": 1.059452775371248e-05, + "loss": 4.9085, "step": 10835 }, { "epoch": 0.87, - "learning_rate": 1.039222133931117e-05, - "loss": 0.9714, + "learning_rate": 1.0532057131581242e-05, + "loss": 4.8429, "step": 10840 }, { "epoch": 0.87, - "learning_rate": 1.0330251976410355e-05, - "loss": 0.9371, + "learning_rate": 1.0469760993816057e-05, + "loss": 4.87, "step": 10845 }, { "epoch": 0.87, - "learning_rate": 1.0268457865182147e-05, - "loss": 1.0379, + "learning_rate": 1.0407639461909613e-05, + "loss": 4.9099, "step": 10850 }, { "epoch": 0.87, - "learning_rate": 1.020683912639775e-05, - "loss": 0.9589, + "learning_rate": 1.0345692657014117e-05, + "loss": 4.9344, "step": 10855 }, { "epoch": 0.87, - "learning_rate": 1.014539588048562e-05, - "loss": 0.9598, + "learning_rate": 1.028392069994093e-05, + "loss": 4.8988, "step": 10860 }, { "epoch": 0.87, - "learning_rate": 1.0084128247531144e-05, - "loss": 0.9517, + "learning_rate": 1.0222323711160553e-05, + "loss": 4.9258, "step": 10865 }, { "epoch": 0.87, - "learning_rate": 1.0023036347276594e-05, - "loss": 0.9899, + "learning_rate": 1.0160901810802115e-05, + "loss": 4.8481, "step": 10870 }, { "epoch": 0.87, - "learning_rate": 9.962120299120758e-06, - "loss": 0.9713, + "learning_rate": 1.0099655118653372e-05, + "loss": 4.9742, "step": 10875 }, { "epoch": 0.87, - "learning_rate": 9.901380222118661e-06, - "loss": 1.0293, + "learning_rate": 1.0038583754160314e-05, + "loss": 4.8928, "step": 10880 }, { "epoch": 0.87, - "learning_rate": 9.84081623498151e-06, - "loss": 0.9778, + "learning_rate": 9.977687836427086e-06, + "loss": 4.864, "step": 10885 }, { "epoch": 0.87, - "learning_rate": 9.780428456076285e-06, - "loss": 1.0197, + "learning_rate": 9.916967484215568e-06, + "loss": 4.8993, "step": 10890 }, { "epoch": 0.87, - "learning_rate": 9.720217003425647e-06, - "loss": 0.9659, + "learning_rate": 9.856422815945332e-06, + "loss": 4.9356, "step": 10895 }, { "epoch": 0.87, - "learning_rate": 9.660181994707596e-06, - "loss": 0.9776, + "learning_rate": 9.796053949693262e-06, + "loss": 4.9303, "step": 10900 }, { "epoch": 0.87, - "learning_rate": 9.600323547255285e-06, - "loss": 0.9979, + "learning_rate": 9.735861003193425e-06, + "loss": 4.9313, "step": 10905 }, { "epoch": 0.87, - "learning_rate": 9.540641778056825e-06, - "loss": 0.9804, + "learning_rate": 9.675844093836795e-06, + "loss": 4.9574, "step": 10910 }, { "epoch": 0.87, - "learning_rate": 9.481136803755008e-06, - "loss": 0.9765, + "learning_rate": 9.616003338670986e-06, + "loss": 4.9531, "step": 10915 }, { "epoch": 0.87, - "learning_rate": 9.421808740647087e-06, - "loss": 0.9399, + "learning_rate": 9.556338854400138e-06, + "loss": 4.9305, "step": 10920 }, { - "epoch": 0.88, - "learning_rate": 9.362657704684597e-06, - "loss": 1.0288, + "epoch": 0.87, + "learning_rate": 9.496850757384556e-06, + "loss": 4.9694, "step": 10925 }, { - "epoch": 0.88, - "learning_rate": 9.303683811473019e-06, - "loss": 1.0297, + "epoch": 0.87, + "learning_rate": 9.437539163640618e-06, + "loss": 4.882, "step": 10930 }, { - "epoch": 0.88, - "learning_rate": 9.244887176271688e-06, - "loss": 0.9504, + "epoch": 0.87, + "learning_rate": 9.378404188840384e-06, + "loss": 4.8995, "step": 10935 }, { "epoch": 0.88, - "learning_rate": 9.186267913993474e-06, - "loss": 0.9728, + "learning_rate": 9.319445948311534e-06, + "loss": 4.8903, "step": 10940 }, { "epoch": 0.88, - "learning_rate": 9.127826139204587e-06, - "loss": 0.9356, + "learning_rate": 9.260664557037058e-06, + "loss": 4.9327, "step": 10945 }, { "epoch": 0.88, - "learning_rate": 9.069561966124374e-06, - "loss": 0.9796, + "learning_rate": 9.202060129655054e-06, + "loss": 4.8948, "step": 10950 }, { "epoch": 0.88, - "learning_rate": 9.011475508625079e-06, - "loss": 0.9597, + "learning_rate": 9.143632780458466e-06, + "loss": 4.896, "step": 10955 }, { "epoch": 0.88, - "learning_rate": 8.953566880231557e-06, - "loss": 0.9639, + "learning_rate": 9.085382623394923e-06, + "loss": 4.9398, "step": 10960 }, { "epoch": 0.88, - "learning_rate": 8.895836194121155e-06, - "loss": 0.9818, + "learning_rate": 9.027309772066495e-06, + "loss": 4.8719, "step": 10965 }, { "epoch": 0.88, - "learning_rate": 8.838283563123473e-06, - "loss": 0.9541, + "learning_rate": 8.969414339729398e-06, + "loss": 4.9374, "step": 10970 }, { "epoch": 0.88, - "learning_rate": 8.78090909972008e-06, - "loss": 0.9443, + "learning_rate": 8.911696439293915e-06, + "loss": 4.8813, "step": 10975 }, { "epoch": 0.88, - "learning_rate": 8.723712916044347e-06, - "loss": 1.01, + "learning_rate": 8.854156183324048e-06, + "loss": 4.9082, "step": 10980 }, { "epoch": 0.88, - "learning_rate": 8.666695123881185e-06, - "loss": 1.0075, + "learning_rate": 8.796793684037385e-06, + "loss": 4.8916, "step": 10985 }, { "epoch": 0.88, - "learning_rate": 8.609855834666892e-06, - "loss": 0.9274, + "learning_rate": 8.73960905330481e-06, + "loss": 4.912, "step": 10990 }, { "epoch": 0.88, - "learning_rate": 8.553195159488892e-06, - "loss": 1.036, + "learning_rate": 8.682602402650331e-06, + "loss": 4.8958, "step": 10995 }, { "epoch": 0.88, - "learning_rate": 8.496713209085494e-06, - "loss": 0.963, + "learning_rate": 8.625773843250873e-06, + "loss": 4.8864, "step": 11000 }, { "epoch": 0.88, - "learning_rate": 8.44041009384574e-06, - "loss": 0.9909, + "learning_rate": 8.569123485936014e-06, + "loss": 4.9811, "step": 11005 }, { "epoch": 0.88, - "learning_rate": 8.384285923809155e-06, - "loss": 0.9705, + "learning_rate": 8.512651441187824e-06, + "loss": 4.8969, "step": 11010 }, { "epoch": 0.88, - "learning_rate": 8.328340808665491e-06, - "loss": 1.0247, + "learning_rate": 8.456357819140549e-06, + "loss": 4.9322, "step": 11015 }, { "epoch": 0.88, - "learning_rate": 8.272574857754567e-06, - "loss": 0.9832, + "learning_rate": 8.400242729580587e-06, + "loss": 4.9375, "step": 11020 }, { "epoch": 0.88, - "learning_rate": 8.216988180066087e-06, - "loss": 1.0487, + "learning_rate": 8.344306281946023e-06, + "loss": 4.9152, "step": 11025 }, { "epoch": 0.88, - "learning_rate": 8.161580884239317e-06, - "loss": 0.973, + "learning_rate": 8.28854858532664e-06, + "loss": 4.9962, "step": 11030 }, { "epoch": 0.88, - "learning_rate": 8.106353078563001e-06, - "loss": 0.9151, + "learning_rate": 8.23296974846357e-06, + "loss": 4.8902, "step": 11035 }, { "epoch": 0.88, - "learning_rate": 8.051304870975017e-06, - "loss": 0.909, + "learning_rate": 8.177569879749159e-06, + "loss": 4.8991, "step": 11040 }, { "epoch": 0.88, - "learning_rate": 7.996436369062278e-06, - "loss": 0.9357, + "learning_rate": 8.122349087226654e-06, + "loss": 4.8905, "step": 11045 }, { - "epoch": 0.89, - "learning_rate": 7.941747680060464e-06, - "loss": 0.992, + "epoch": 0.88, + "learning_rate": 8.067307478590125e-06, + "loss": 4.9839, "step": 11050 }, { - "epoch": 0.89, - "learning_rate": 7.887238910853834e-06, - "loss": 0.9508, + "epoch": 0.88, + "learning_rate": 8.01244516118418e-06, + "loss": 4.8983, "step": 11055 }, { - "epoch": 0.89, - "learning_rate": 7.832910167975027e-06, - "loss": 0.983, + "epoch": 0.88, + "learning_rate": 7.95776224200373e-06, + "loss": 4.9262, "step": 11060 }, { "epoch": 0.89, - "learning_rate": 7.778761557604775e-06, - "loss": 1.0066, + "learning_rate": 7.903258827693882e-06, + "loss": 4.883, "step": 11065 }, { "epoch": 0.89, - "learning_rate": 7.7247931855718e-06, - "loss": 0.9621, + "learning_rate": 7.848935024549554e-06, + "loss": 4.9785, "step": 11070 }, { "epoch": 0.89, - "learning_rate": 7.671005157352606e-06, - "loss": 1.0152, + "learning_rate": 7.794790938515517e-06, + "loss": 4.965, "step": 11075 }, { "epoch": 0.89, - "learning_rate": 7.61739757807115e-06, - "loss": 0.9514, + "learning_rate": 7.740826675185919e-06, + "loss": 4.8676, "step": 11080 }, { "epoch": 0.89, - "learning_rate": 7.563970552498745e-06, - "loss": 0.9725, + "learning_rate": 7.687042339804329e-06, + "loss": 4.8561, "step": 11085 }, { "epoch": 0.89, - "learning_rate": 7.5107241850538746e-06, - "loss": 0.9541, + "learning_rate": 7.633438037263274e-06, + "loss": 4.861, "step": 11090 }, { "epoch": 0.89, - "learning_rate": 7.457658579801852e-06, - "loss": 0.9714, + "learning_rate": 7.5800138721043305e-06, + "loss": 5.0399, "step": 11095 }, { "epoch": 0.89, - "learning_rate": 7.404773840454804e-06, - "loss": 1.0041, + "learning_rate": 7.5267699485176425e-06, + "loss": 4.9581, "step": 11100 }, { "epoch": 0.89, - "learning_rate": 7.352070070371286e-06, - "loss": 0.9715, + "learning_rate": 7.473706370341882e-06, + "loss": 4.904, "step": 11105 }, { "epoch": 0.89, - "learning_rate": 7.299547372556226e-06, - "loss": 1.0379, + "learning_rate": 7.420823241064012e-06, + "loss": 4.8102, "step": 11110 }, { "epoch": 0.89, - "learning_rate": 7.2472058496606435e-06, - "loss": 0.9894, + "learning_rate": 7.368120663819056e-06, + "loss": 4.8853, "step": 11115 }, { "epoch": 0.89, - "learning_rate": 7.195045603981421e-06, - "loss": 0.9316, + "learning_rate": 7.3155987413899616e-06, + "loss": 4.9179, "step": 11120 }, { "epoch": 0.89, - "learning_rate": 7.143066737461235e-06, - "loss": 0.9985, + "learning_rate": 7.263257576207261e-06, + "loss": 4.9941, "step": 11125 }, { "epoch": 0.89, - "learning_rate": 7.091269351688245e-06, - "loss": 0.9597, + "learning_rate": 7.211097270349066e-06, + "loss": 4.982, "step": 11130 }, { "epoch": 0.89, - "learning_rate": 7.039653547895852e-06, - "loss": 0.9349, + "learning_rate": 7.159117925540726e-06, + "loss": 4.9743, "step": 11135 }, { "epoch": 0.89, - "learning_rate": 6.988219426962661e-06, - "loss": 1.0419, + "learning_rate": 7.107319643154675e-06, + "loss": 4.9178, "step": 11140 }, { "epoch": 0.89, - "learning_rate": 6.9369670894121495e-06, - "loss": 0.9347, + "learning_rate": 7.055702524210184e-06, + "loss": 4.9412, "step": 11145 }, { "epoch": 0.89, - "learning_rate": 6.8858966354125256e-06, - "loss": 0.9974, + "learning_rate": 7.004266669373338e-06, + "loss": 4.8909, "step": 11150 }, { "epoch": 0.89, - "learning_rate": 6.8350081647765464e-06, - "loss": 1.0033, + "learning_rate": 6.953012178956564e-06, + "loss": 4.945, "step": 11155 }, { "epoch": 0.89, - "learning_rate": 6.7843017769612215e-06, - "loss": 0.9657, + "learning_rate": 6.901939152918691e-06, + "loss": 4.8563, "step": 11160 }, { "epoch": 0.89, - "learning_rate": 6.733777571067779e-06, - "loss": 1.0231, + "learning_rate": 6.851047690864565e-06, + "loss": 4.9574, "step": 11165 }, { "epoch": 0.89, - "learning_rate": 6.683435645841352e-06, - "loss": 0.9478, + "learning_rate": 6.800337892045039e-06, + "loss": 4.9555, "step": 11170 }, { - "epoch": 0.9, - "learning_rate": 6.633276099670816e-06, - "loss": 0.9456, + "epoch": 0.89, + "learning_rate": 6.74980985535657e-06, + "loss": 4.9551, "step": 11175 }, { - "epoch": 0.9, - "learning_rate": 6.5832990305886525e-06, - "loss": 1.0316, + "epoch": 0.89, + "learning_rate": 6.699463679341211e-06, + "loss": 4.966, "step": 11180 }, { - "epoch": 0.9, - "learning_rate": 6.533504536270607e-06, - "loss": 0.9781, + "epoch": 0.89, + "learning_rate": 6.649299462186309e-06, + "loss": 4.8783, "step": 11185 }, { "epoch": 0.9, - "learning_rate": 6.483892714035699e-06, - "loss": 0.9742, + "learning_rate": 6.599317301724362e-06, + "loss": 4.8958, "step": 11190 }, { "epoch": 0.9, - "learning_rate": 6.434463660845891e-06, - "loss": 1.0248, + "learning_rate": 6.549517295432816e-06, + "loss": 4.8527, "step": 11195 }, { "epoch": 0.9, - "learning_rate": 6.385217473305916e-06, - "loss": 1.0073, + "learning_rate": 6.499899540433807e-06, + "loss": 4.9429, "step": 11200 }, { "epoch": 0.9, - "learning_rate": 6.336154247663162e-06, - "loss": 1.0378, + "learning_rate": 6.4504641334941604e-06, + "loss": 4.8417, "step": 11205 }, { "epoch": 0.9, - "learning_rate": 6.287274079807415e-06, - "loss": 0.9732, + "learning_rate": 6.401211171024957e-06, + "loss": 4.9153, "step": 11210 }, { "epoch": 0.9, - "learning_rate": 6.238577065270657e-06, - "loss": 0.9233, + "learning_rate": 6.352140749081547e-06, + "loss": 4.9726, "step": 11215 }, { "epoch": 0.9, - "learning_rate": 6.190063299226945e-06, - "loss": 0.9633, + "learning_rate": 6.303252963363215e-06, + "loss": 4.9163, "step": 11220 }, { "epoch": 0.9, - "learning_rate": 6.141732876492201e-06, - "loss": 1.0043, + "learning_rate": 6.254547909213149e-06, + "loss": 4.9299, "step": 11225 }, { "epoch": 0.9, - "learning_rate": 6.09358589152399e-06, - "loss": 0.9836, + "learning_rate": 6.206025681618077e-06, + "loss": 4.9359, "step": 11230 }, { "epoch": 0.9, - "learning_rate": 6.045622438421405e-06, - "loss": 1.002, + "learning_rate": 6.157686375208227e-06, + "loss": 4.8952, "step": 11235 }, { "epoch": 0.9, - "learning_rate": 5.997842610924786e-06, - "loss": 1.033, + "learning_rate": 6.109530084257042e-06, + "loss": 4.8343, "step": 11240 }, { "epoch": 0.9, - "learning_rate": 5.9502465024156125e-06, - "loss": 1.0052, + "learning_rate": 6.061556902681098e-06, + "loss": 4.8411, "step": 11245 }, { "epoch": 0.9, - "learning_rate": 5.902834205916374e-06, - "loss": 0.9796, + "learning_rate": 6.0137669240398205e-06, + "loss": 4.9183, "step": 11250 }, { "epoch": 0.9, - "learning_rate": 5.8556058140902035e-06, - "loss": 0.9528, + "learning_rate": 5.966160241535335e-06, + "loss": 4.9199, "step": 11255 }, { "epoch": 0.9, - "learning_rate": 5.808561419240865e-06, - "loss": 0.9984, + "learning_rate": 5.918736948012326e-06, + "loss": 4.9842, "step": 11260 }, { "epoch": 0.9, - "learning_rate": 5.761701113312523e-06, - "loss": 0.9883, + "learning_rate": 5.8714971359578154e-06, + "loss": 4.931, "step": 11265 }, { "epoch": 0.9, - "learning_rate": 5.715024987889517e-06, - "loss": 0.9422, + "learning_rate": 5.824440897500994e-06, + "loss": 5.0218, "step": 11270 }, { "epoch": 0.9, - "learning_rate": 5.668533134196286e-06, - "loss": 0.9797, + "learning_rate": 5.777568324413008e-06, + "loss": 4.9182, "step": 11275 }, { "epoch": 0.9, - "learning_rate": 5.6222256430970565e-06, - "loss": 0.9907, + "learning_rate": 5.730879508106857e-06, + "loss": 4.9555, "step": 11280 }, { "epoch": 0.9, - "learning_rate": 5.576102605095779e-06, - "loss": 0.9869, + "learning_rate": 5.6843745396371494e-06, + "loss": 4.922, "step": 11285 }, { "epoch": 0.9, - "learning_rate": 5.530164110335901e-06, - "loss": 0.9949, + "learning_rate": 5.6380535096999364e-06, + "loss": 4.9148, "step": 11290 }, { "epoch": 0.9, - "learning_rate": 5.484410248600158e-06, - "loss": 0.9808, + "learning_rate": 5.591916508632556e-06, + "loss": 4.9045, "step": 11295 }, { - "epoch": 0.91, - "learning_rate": 5.438841109310511e-06, - "loss": 0.9878, + "epoch": 0.9, + "learning_rate": 5.545963626413453e-06, + "loss": 4.9593, "step": 11300 }, { - "epoch": 0.91, - "learning_rate": 5.393456781527817e-06, - "loss": 0.9764, + "epoch": 0.9, + "learning_rate": 5.500194952661986e-06, + "loss": 4.8712, "step": 11305 }, { - "epoch": 0.91, - "learning_rate": 5.348257353951791e-06, - "loss": 1.0054, + "epoch": 0.9, + "learning_rate": 5.4546105766382305e-06, + "loss": 4.8974, "step": 11310 }, { "epoch": 0.91, - "learning_rate": 5.30324291492077e-06, - "loss": 1.0369, + "learning_rate": 5.409210587242885e-06, + "loss": 4.8877, "step": 11315 }, { "epoch": 0.91, - "learning_rate": 5.258413552411479e-06, - "loss": 0.9772, + "learning_rate": 5.363995073017047e-06, + "loss": 5.0066, "step": 11320 }, { "epoch": 0.91, - "learning_rate": 5.213769354039044e-06, - "loss": 1.0084, + "learning_rate": 5.318964122142011e-06, + "loss": 4.8497, "step": 11325 }, { "epoch": 0.91, - "learning_rate": 5.169310407056627e-06, - "loss": 0.9545, + "learning_rate": 5.27411782243914e-06, + "loss": 4.8634, "step": 11330 }, { "epoch": 0.91, - "learning_rate": 5.125036798355332e-06, - "loss": 1.0437, + "learning_rate": 5.22945626136967e-06, + "loss": 4.9107, "step": 11335 }, { "epoch": 0.91, - "learning_rate": 5.080948614464054e-06, - "loss": 0.9981, + "learning_rate": 5.1849795260345945e-06, + "loss": 4.8804, "step": 11340 }, { "epoch": 0.91, - "learning_rate": 5.03704594154929e-06, - "loss": 0.9996, + "learning_rate": 5.140687703174407e-06, + "loss": 4.8933, "step": 11345 }, { "epoch": 0.91, - "learning_rate": 4.993328865414981e-06, - "loss": 0.953, + "learning_rate": 5.0965808791690214e-06, + "loss": 4.9263, "step": 11350 }, { "epoch": 0.91, - "learning_rate": 4.949797471502337e-06, - "loss": 1.0151, + "learning_rate": 5.052659140037464e-06, + "loss": 4.8587, "step": 11355 }, { "epoch": 0.91, - "learning_rate": 4.906451844889615e-06, - "loss": 0.9698, + "learning_rate": 5.008922571437946e-06, + "loss": 4.8832, "step": 11360 }, { "epoch": 0.91, - "learning_rate": 4.863292070292081e-06, - "loss": 1.0068, + "learning_rate": 4.9653712586674176e-06, + "loss": 4.9147, "step": 11365 }, { "epoch": 0.91, - "learning_rate": 4.820318232061727e-06, - "loss": 0.9437, + "learning_rate": 4.922005286661591e-06, + "loss": 4.9498, "step": 11370 }, { "epoch": 0.91, - "learning_rate": 4.777530414187148e-06, - "loss": 0.9551, + "learning_rate": 4.878824739994748e-06, + "loss": 4.9574, "step": 11375 }, { "epoch": 0.91, - "learning_rate": 4.734928700293406e-06, - "loss": 0.9215, + "learning_rate": 4.835829702879491e-06, + "loss": 4.8758, "step": 11380 }, { "epoch": 0.91, - "learning_rate": 4.692513173641821e-06, - "loss": 0.8949, + "learning_rate": 4.793020259166658e-06, + "loss": 4.8707, "step": 11385 }, { "epoch": 0.91, - "learning_rate": 4.6502839171298054e-06, - "loss": 0.9798, + "learning_rate": 4.750396492345122e-06, + "loss": 4.825, "step": 11390 }, { "epoch": 0.91, - "learning_rate": 4.60824101329077e-06, - "loss": 1.0142, + "learning_rate": 4.707958485541664e-06, + "loss": 4.8967, "step": 11395 }, { "epoch": 0.91, - "learning_rate": 4.566384544293867e-06, - "loss": 0.9596, + "learning_rate": 4.66570632152078e-06, + "loss": 4.8722, "step": 11400 }, { "epoch": 0.91, - "learning_rate": 4.5247145919439125e-06, - "loss": 0.9179, + "learning_rate": 4.623640082684522e-06, + "loss": 4.8878, "step": 11405 }, { "epoch": 0.91, - "learning_rate": 4.483231237681185e-06, - "loss": 0.9992, + "learning_rate": 4.5817598510723096e-06, + "loss": 4.8788, "step": 11410 }, { "epoch": 0.91, - "learning_rate": 4.441934562581262e-06, - "loss": 0.9664, + "learning_rate": 4.540065708360886e-06, + "loss": 4.9497, "step": 11415 }, { "epoch": 0.91, - "learning_rate": 4.400824647354873e-06, - "loss": 0.9553, + "learning_rate": 4.498557735863984e-06, + "loss": 4.8755, "step": 11420 }, { - "epoch": 0.92, - "learning_rate": 4.359901572347758e-06, - "loss": 1.025, + "epoch": 0.91, + "learning_rate": 4.45723601453234e-06, + "loss": 4.8915, "step": 11425 }, { - "epoch": 0.92, - "learning_rate": 4.319165417540483e-06, - "loss": 0.9633, + "epoch": 0.91, + "learning_rate": 4.416100624953345e-06, + "loss": 4.8778, "step": 11430 }, { - "epoch": 0.92, - "learning_rate": 4.278616262548318e-06, - "loss": 1.0518, + "epoch": 0.91, + "learning_rate": 4.37515164735115e-06, + "loss": 4.8939, "step": 11435 }, { "epoch": 0.92, - "learning_rate": 4.238254186620993e-06, - "loss": 0.9421, + "learning_rate": 4.334389161586228e-06, + "loss": 4.8668, "step": 11440 }, { "epoch": 0.92, - "learning_rate": 4.198079268642663e-06, - "loss": 0.9826, + "learning_rate": 4.293813247155398e-06, + "loss": 4.9008, "step": 11445 }, { "epoch": 0.92, - "learning_rate": 4.158091587131729e-06, - "loss": 0.9366, + "learning_rate": 4.2534239831916176e-06, + "loss": 4.791, "step": 11450 }, { "epoch": 0.92, - "learning_rate": 4.118291220240567e-06, - "loss": 1.0083, + "learning_rate": 4.213221448463833e-06, + "loss": 4.9013, "step": 11455 }, { "epoch": 0.92, - "learning_rate": 4.07867824575553e-06, - "loss": 0.9672, + "learning_rate": 4.173205721376827e-06, + "loss": 4.8954, "step": 11460 }, { "epoch": 0.92, - "learning_rate": 4.039252741096711e-06, - "loss": 0.9586, + "learning_rate": 4.133376879971018e-06, + "loss": 4.8918, "step": 11465 }, { "epoch": 0.92, - "learning_rate": 4.0000147833177845e-06, - "loss": 0.9389, + "learning_rate": 4.093735001922416e-06, + "loss": 4.888, "step": 11470 }, { "epoch": 0.92, - "learning_rate": 3.960964449105942e-06, - "loss": 0.9834, + "learning_rate": 4.054280164542379e-06, + "loss": 4.9155, "step": 11475 }, { "epoch": 0.92, - "learning_rate": 3.92210181478162e-06, - "loss": 0.9231, + "learning_rate": 4.0150124447774775e-06, + "loss": 4.9098, "step": 11480 }, { "epoch": 0.92, - "learning_rate": 3.883426956298431e-06, - "loss": 1.0144, + "learning_rate": 3.975931919209341e-06, + "loss": 4.8733, "step": 11485 }, { "epoch": 0.92, - "learning_rate": 3.844939949243021e-06, - "loss": 1.0351, + "learning_rate": 3.937038664054604e-06, + "loss": 4.9098, "step": 11490 }, { "epoch": 0.92, - "learning_rate": 3.806640868834832e-06, - "loss": 1.0011, + "learning_rate": 3.898332755164591e-06, + "loss": 4.9134, "step": 11495 }, { "epoch": 0.92, - "learning_rate": 3.768529789926112e-06, - "loss": 1.0573, + "learning_rate": 3.859814268025308e-06, + "loss": 4.8675, "step": 11500 }, { "epoch": 0.92, - "learning_rate": 3.7306067870016226e-06, - "loss": 0.9691, + "learning_rate": 3.821483277757176e-06, + "loss": 4.8103, "step": 11505 }, { "epoch": 0.92, - "learning_rate": 3.6928719341785324e-06, - "loss": 1.0016, + "learning_rate": 3.783339859115065e-06, + "loss": 4.8869, "step": 11510 }, { "epoch": 0.92, - "learning_rate": 3.655325305206314e-06, - "loss": 0.928, + "learning_rate": 3.7453840864879353e-06, + "loss": 4.8634, "step": 11515 }, { "epoch": 0.92, - "learning_rate": 3.617966973466569e-06, - "loss": 1.0141, + "learning_rate": 3.7076160338988086e-06, + "loss": 4.877, "step": 11520 }, { "epoch": 0.92, - "learning_rate": 3.580797011972892e-06, - "loss": 0.9738, + "learning_rate": 3.6700357750046543e-06, + "loss": 4.8789, "step": 11525 }, { "epoch": 0.92, - "learning_rate": 3.5438154933707277e-06, - "loss": 1.0158, + "learning_rate": 3.632643383096157e-06, + "loss": 4.9107, "step": 11530 }, { "epoch": 0.92, - "learning_rate": 3.507022489937184e-06, - "loss": 0.9757, + "learning_rate": 3.5954389310976165e-06, + "loss": 4.8614, "step": 11535 }, { "epoch": 0.92, - "learning_rate": 3.4704180735809834e-06, - "loss": 0.9587, + "learning_rate": 3.558422491566804e-06, + "loss": 4.8355, "step": 11540 }, { "epoch": 0.92, - "learning_rate": 3.4340023158422554e-06, - "loss": 0.9896, + "learning_rate": 3.5215941366948726e-06, + "loss": 4.864, "step": 11545 }, { - "epoch": 0.93, - "learning_rate": 3.397775287892413e-06, - "loss": 0.9588, + "epoch": 0.92, + "learning_rate": 3.484953938306057e-06, + "loss": 4.9117, "step": 11550 }, { - "epoch": 0.93, - "learning_rate": 3.361737060534009e-06, - "loss": 0.9723, + "epoch": 0.92, + "learning_rate": 3.4485019678577646e-06, + "loss": 4.9249, "step": 11555 }, { - "epoch": 0.93, - "learning_rate": 3.3258877042006033e-06, - "loss": 0.9919, + "epoch": 0.92, + "learning_rate": 3.4122382964401845e-06, + "loss": 4.8543, "step": 11560 }, { "epoch": 0.93, - "learning_rate": 3.2902272889566155e-06, - "loss": 0.9571, + "learning_rate": 3.3761629947764105e-06, + "loss": 4.9264, "step": 11565 }, { "epoch": 0.93, - "learning_rate": 3.254755884497207e-06, - "loss": 0.9974, + "learning_rate": 3.3402761332220642e-06, + "loss": 4.8371, "step": 11570 }, { "epoch": 0.93, - "learning_rate": 3.2194735601481453e-06, - "loss": 0.9559, + "learning_rate": 3.3045777817652945e-06, + "loss": 4.9609, "step": 11575 }, { "epoch": 0.93, - "learning_rate": 3.1843803848656486e-06, - "loss": 0.9566, + "learning_rate": 3.2690680100266434e-06, + "loss": 4.8998, "step": 11580 }, { "epoch": 0.93, - "learning_rate": 3.149476427236242e-06, - "loss": 0.9924, + "learning_rate": 3.233746887258837e-06, + "loss": 4.7756, "step": 11585 }, { "epoch": 0.93, - "learning_rate": 3.114761755476647e-06, - "loss": 1.0237, + "learning_rate": 3.1986144823467067e-06, + "loss": 4.834, "step": 11590 }, { "epoch": 0.93, - "learning_rate": 3.08023643743367e-06, - "loss": 0.9739, + "learning_rate": 3.1636708638070112e-06, + "loss": 4.9267, "step": 11595 }, { "epoch": 0.93, - "learning_rate": 3.0459005405840125e-06, - "loss": 0.9681, + "learning_rate": 3.1289160997883703e-06, + "loss": 4.9476, "step": 11600 }, { "epoch": 0.93, - "learning_rate": 3.0117541320341724e-06, - "loss": 0.971, + "learning_rate": 3.0943502580710772e-06, + "loss": 4.8875, "step": 11605 }, { "epoch": 0.93, - "learning_rate": 2.9777972785203444e-06, - "loss": 0.9914, + "learning_rate": 3.059973406066963e-06, + "loss": 4.9482, "step": 11610 }, { "epoch": 0.93, - "learning_rate": 2.944030046408197e-06, - "loss": 0.972, + "learning_rate": 3.0257856108192985e-06, + "loss": 4.8495, "step": 11615 }, { "epoch": 0.93, - "learning_rate": 2.9104525016928284e-06, - "loss": 0.9588, + "learning_rate": 2.99178693900265e-06, + "loss": 4.8479, "step": 11620 }, { "epoch": 0.93, - "learning_rate": 2.8770647099986338e-06, - "loss": 0.9703, + "learning_rate": 2.957977456922734e-06, + "loss": 4.9979, "step": 11625 }, { "epoch": 0.93, - "learning_rate": 2.8438667365791148e-06, - "loss": 0.9954, + "learning_rate": 2.924357230516306e-06, + "loss": 4.8818, "step": 11630 }, { "epoch": 0.93, - "learning_rate": 2.810858646316794e-06, - "loss": 0.95, + "learning_rate": 2.890926325351029e-06, + "loss": 4.9009, "step": 11635 }, { "epoch": 0.93, - "learning_rate": 2.7780405037231226e-06, - "loss": 0.9692, + "learning_rate": 2.85768480662536e-06, + "loss": 4.954, "step": 11640 }, { "epoch": 0.93, - "learning_rate": 2.74541237293825e-06, - "loss": 0.9332, + "learning_rate": 2.8246327391683756e-06, + "loss": 4.8819, "step": 11645 }, { "epoch": 0.93, - "learning_rate": 2.712974317731043e-06, - "loss": 1.0044, + "learning_rate": 2.7917701874396906e-06, + "loss": 4.9285, "step": 11650 }, { "epoch": 0.93, - "learning_rate": 2.680726401498812e-06, - "loss": 0.9082, + "learning_rate": 2.7590972155293047e-06, + "loss": 4.8786, "step": 11655 }, { "epoch": 0.93, - "learning_rate": 2.648668687267286e-06, - "loss": 1.0322, + "learning_rate": 2.726613887157514e-06, + "loss": 4.9518, "step": 11660 }, { "epoch": 0.93, - "learning_rate": 2.6168012376904803e-06, - "loss": 1.0062, + "learning_rate": 2.694320265674777e-06, + "loss": 4.756, "step": 11665 }, { "epoch": 0.93, - "learning_rate": 2.585124115050508e-06, - "loss": 0.9393, + "learning_rate": 2.6622164140615024e-06, + "loss": 4.9822, "step": 11670 }, { - "epoch": 0.94, - "learning_rate": 2.5536373812575674e-06, - "loss": 0.9361, + "epoch": 0.93, + "learning_rate": 2.630302394928086e-06, + "loss": 4.9689, "step": 11675 }, { - "epoch": 0.94, - "learning_rate": 2.522341097849701e-06, - "loss": 0.9596, + "epoch": 0.93, + "learning_rate": 2.598578270514651e-06, + "loss": 4.9152, "step": 11680 }, { - "epoch": 0.94, - "learning_rate": 2.491235325992747e-06, - "loss": 0.9462, + "epoch": 0.93, + "learning_rate": 2.5670441026910076e-06, + "loss": 4.9294, "step": 11685 }, { "epoch": 0.94, - "learning_rate": 2.460320126480242e-06, - "loss": 1.0318, + "learning_rate": 2.535699952956505e-06, + "loss": 4.9018, "step": 11690 }, { "epoch": 0.94, - "learning_rate": 2.4295955597332e-06, - "loss": 0.9517, + "learning_rate": 2.504545882439857e-06, + "loss": 4.9582, "step": 11695 }, { "epoch": 0.94, - "learning_rate": 2.399061685800119e-06, - "loss": 0.9566, + "learning_rate": 2.473581951899184e-06, + "loss": 4.8809, "step": 11700 }, { "epoch": 0.94, - "learning_rate": 2.3687185643567977e-06, - "loss": 1.0078, + "learning_rate": 2.4428082217216706e-06, + "loss": 4.8341, "step": 11705 }, { "epoch": 0.94, - "learning_rate": 2.338566254706176e-06, - "loss": 0.9256, + "learning_rate": 2.4122247519236306e-06, + "loss": 4.9144, "step": 11710 }, { "epoch": 0.94, - "learning_rate": 2.308604815778326e-06, - "loss": 0.9681, + "learning_rate": 2.3818316021503084e-06, + "loss": 4.9294, "step": 11715 }, { "epoch": 0.94, - "learning_rate": 2.278834306130251e-06, - "loss": 0.9627, + "learning_rate": 2.3516288316757897e-06, + "loss": 4.9172, "step": 11720 }, { "epoch": 0.94, - "learning_rate": 2.2492547839457976e-06, - "loss": 1.0808, + "learning_rate": 2.3216164994028457e-06, + "loss": 4.9593, "step": 11725 }, { "epoch": 0.94, - "learning_rate": 2.2198663070355783e-06, - "loss": 0.9964, + "learning_rate": 2.2917946638628674e-06, + "loss": 4.9047, "step": 11730 }, { "epoch": 0.94, - "learning_rate": 2.1906689328367813e-06, - "loss": 0.9477, + "learning_rate": 2.262163383215732e-06, + "loss": 4.857, "step": 11735 }, { "epoch": 0.94, - "learning_rate": 2.1616627184131044e-06, - "loss": 1.0159, + "learning_rate": 2.2327227152496686e-06, + "loss": 4.8838, "step": 11740 }, { "epoch": 0.94, - "learning_rate": 2.132847720454678e-06, - "loss": 0.9994, + "learning_rate": 2.2034727173812163e-06, + "loss": 4.9336, "step": 11745 }, { "epoch": 0.94, - "learning_rate": 2.1042239952778874e-06, - "loss": 0.9595, + "learning_rate": 2.174413446654966e-06, + "loss": 4.9019, "step": 11750 }, { "epoch": 0.94, - "learning_rate": 2.075791598825283e-06, - "loss": 0.9996, + "learning_rate": 2.1455449597436506e-06, + "loss": 4.9563, "step": 11755 }, { "epoch": 0.94, - "learning_rate": 2.0475505866655254e-06, - "loss": 1.0088, + "learning_rate": 2.1168673129478566e-06, + "loss": 4.967, "step": 11760 }, { "epoch": 0.94, - "learning_rate": 2.0195010139931636e-06, - "loss": 0.9687, + "learning_rate": 2.088380562196013e-06, + "loss": 4.925, "step": 11765 }, { "epoch": 0.94, - "learning_rate": 1.991642935628657e-06, - "loss": 0.9467, + "learning_rate": 2.0600847630442345e-06, + "loss": 4.8409, "step": 11770 }, { "epoch": 0.94, - "learning_rate": 1.963976406018175e-06, - "loss": 1.0271, + "learning_rate": 2.031979970676279e-06, + "loss": 4.901, "step": 11775 }, { "epoch": 0.94, - "learning_rate": 1.936501479233521e-06, - "loss": 0.9503, + "learning_rate": 2.0040662399033463e-06, + "loss": 4.874, "step": 11780 }, { "epoch": 0.94, - "learning_rate": 1.9092182089720413e-06, - "loss": 1.0409, + "learning_rate": 1.9763436251640235e-06, + "loss": 4.933, "step": 11785 }, { "epoch": 0.94, - "learning_rate": 1.8821266485565037e-06, - "loss": 0.9267, + "learning_rate": 1.9488121805242177e-06, + "loss": 4.9713, "step": 11790 }, { "epoch": 0.94, - "learning_rate": 1.8552268509349768e-06, - "loss": 0.9623, + "learning_rate": 1.921471959676957e-06, + "loss": 4.9149, "step": 11795 }, { - "epoch": 0.95, - "learning_rate": 1.8285188686807618e-06, - "loss": 1.0058, + "epoch": 0.94, + "learning_rate": 1.8943230159423675e-06, + "loss": 4.9055, "step": 11800 }, { - "epoch": 0.95, - "learning_rate": 1.8020027539922934e-06, - "loss": 1.0079, + "epoch": 0.94, + "learning_rate": 1.867365402267518e-06, + "loss": 4.9412, "step": 11805 }, { - "epoch": 0.95, - "learning_rate": 1.7756785586929836e-06, - "loss": 0.9747, + "epoch": 0.94, + "learning_rate": 1.8405991712263537e-06, + "loss": 4.9321, "step": 11810 }, { "epoch": 0.95, - "learning_rate": 1.7495463342311558e-06, - "loss": 1.0294, + "learning_rate": 1.814024375019574e-06, + "loss": 4.8942, "step": 11815 }, { "epoch": 0.95, - "learning_rate": 1.7236061316799557e-06, - "loss": 0.9998, + "learning_rate": 1.7876410654745325e-06, + "loss": 4.9324, "step": 11820 }, { "epoch": 0.95, - "learning_rate": 1.6978580017372624e-06, - "loss": 0.9508, + "learning_rate": 1.7614492940451033e-06, + "loss": 4.8038, "step": 11825 }, { "epoch": 0.95, - "learning_rate": 1.6723019947255226e-06, - "loss": 0.9758, + "learning_rate": 1.7354491118116934e-06, + "loss": 4.8996, "step": 11830 }, { "epoch": 0.95, - "learning_rate": 1.6469381605917156e-06, - "loss": 0.9113, + "learning_rate": 1.7096405694809858e-06, + "loss": 4.8882, "step": 11835 }, { "epoch": 0.95, - "learning_rate": 1.6217665489072441e-06, - "loss": 0.9946, + "learning_rate": 1.6840237173859519e-06, + "loss": 4.9629, "step": 11840 }, { "epoch": 0.95, - "learning_rate": 1.5967872088678116e-06, - "loss": 0.9599, + "learning_rate": 1.6585986054857173e-06, + "loss": 4.8721, "step": 11845 }, { "epoch": 0.95, - "learning_rate": 1.5720001892933545e-06, - "loss": 0.965, + "learning_rate": 1.6333652833654844e-06, + "loss": 4.8546, "step": 11850 }, { "epoch": 0.95, - "learning_rate": 1.5474055386279328e-06, - "loss": 1.0116, + "learning_rate": 1.6083238002363775e-06, + "loss": 4.8999, "step": 11855 }, { "epoch": 0.95, - "learning_rate": 1.5230033049396297e-06, - "loss": 1.031, + "learning_rate": 1.5834742049354202e-06, + "loss": 4.9587, "step": 11860 }, { "epoch": 0.95, - "learning_rate": 1.4987935359204952e-06, - "loss": 0.9387, + "learning_rate": 1.558816545925401e-06, + "loss": 4.8959, "step": 11865 }, { "epoch": 0.95, - "learning_rate": 1.4747762788863806e-06, - "loss": 0.9921, + "learning_rate": 1.5343508712947874e-06, + "loss": 4.8682, "step": 11870 }, { "epoch": 0.95, - "learning_rate": 1.4509515807769158e-06, - "loss": 0.994, + "learning_rate": 1.5100772287576449e-06, + "loss": 4.8939, "step": 11875 }, { "epoch": 0.95, - "learning_rate": 1.4273194881554096e-06, - "loss": 0.9795, + "learning_rate": 1.4859956656534725e-06, + "loss": 4.8954, "step": 11880 }, { "epoch": 0.95, - "learning_rate": 1.4038800472086945e-06, - "loss": 0.9824, + "learning_rate": 1.4621062289472243e-06, + "loss": 4.8575, "step": 11885 }, { "epoch": 0.95, - "learning_rate": 1.3806333037471141e-06, - "loss": 0.9648, + "learning_rate": 1.4384089652291543e-06, + "loss": 4.8386, "step": 11890 }, { "epoch": 0.95, - "learning_rate": 1.3575793032043926e-06, - "loss": 0.9694, + "learning_rate": 1.4149039207147163e-06, + "loss": 4.8132, "step": 11895 }, { "epoch": 0.95, - "learning_rate": 1.3347180906375656e-06, - "loss": 0.988, + "learning_rate": 1.391591141244486e-06, + "loss": 4.8572, "step": 11900 }, { "epoch": 0.95, - "learning_rate": 1.3120497107268593e-06, - "loss": 0.966, + "learning_rate": 1.3684706722840945e-06, + "loss": 4.941, "step": 11905 }, { "epoch": 0.95, - "learning_rate": 1.2895742077756345e-06, - "loss": 0.9154, + "learning_rate": 1.3455425589240955e-06, + "loss": 4.9598, "step": 11910 }, { "epoch": 0.95, - "learning_rate": 1.2672916257102985e-06, - "loss": 1.0356, + "learning_rate": 1.3228068458799425e-06, + "loss": 4.9379, "step": 11915 }, { "epoch": 0.95, - "learning_rate": 1.2452020080802151e-06, - "loss": 0.9673, + "learning_rate": 1.3002635774918004e-06, + "loss": 4.7972, "step": 11920 }, { - "epoch": 0.96, - "learning_rate": 1.223305398057606e-06, - "loss": 1.0135, + "epoch": 0.95, + "learning_rate": 1.2779127977245897e-06, + "loss": 4.8557, "step": 11925 }, { - "epoch": 0.96, - "learning_rate": 1.2016018384374717e-06, - "loss": 0.9504, + "epoch": 0.95, + "learning_rate": 1.2557545501677758e-06, + "loss": 4.9265, "step": 11930 }, { - "epoch": 0.96, - "learning_rate": 1.1800913716375262e-06, - "loss": 0.9608, + "epoch": 0.95, + "learning_rate": 1.233788878035369e-06, + "loss": 4.9716, "step": 11935 }, { "epoch": 0.96, - "learning_rate": 1.1587740396980851e-06, - "loss": 0.9958, + "learning_rate": 1.2120158241657908e-06, + "loss": 4.8506, "step": 11940 }, { "epoch": 0.96, - "learning_rate": 1.1376498842820105e-06, - "loss": 1.0163, + "learning_rate": 1.190435431021819e-06, + "loss": 4.9209, "step": 11945 }, { "epoch": 0.96, - "learning_rate": 1.1167189466746219e-06, - "loss": 1.0387, + "learning_rate": 1.1690477406905209e-06, + "loss": 4.8247, "step": 11950 }, { "epoch": 0.96, - "learning_rate": 1.0959812677835968e-06, - "loss": 0.9832, + "learning_rate": 1.1478527948830974e-06, + "loss": 4.8924, "step": 11955 }, { "epoch": 0.96, - "learning_rate": 1.0754368881389254e-06, - "loss": 0.9874, + "learning_rate": 1.126850634934884e-06, + "loss": 4.8814, "step": 11960 }, { "epoch": 0.96, - "learning_rate": 1.0550858478927893e-06, - "loss": 0.9675, + "learning_rate": 1.1060413018052273e-06, + "loss": 4.9609, "step": 11965 }, { "epoch": 0.96, - "learning_rate": 1.034928186819506e-06, - "loss": 0.9565, + "learning_rate": 1.08542483607742e-06, + "loss": 4.7887, "step": 11970 }, { "epoch": 0.96, - "learning_rate": 1.0149639443154724e-06, - "loss": 0.9889, + "learning_rate": 1.0650012779586105e-06, + "loss": 4.8089, "step": 11975 }, { "epoch": 0.96, - "learning_rate": 9.951931593990549e-07, - "loss": 1.033, + "learning_rate": 1.0447706672797264e-06, + "loss": 4.9298, "step": 11980 }, { "epoch": 0.96, - "learning_rate": 9.756158707105224e-07, - "loss": 0.9406, + "learning_rate": 1.0247330434954071e-06, + "loss": 4.853, "step": 11985 }, { "epoch": 0.96, - "learning_rate": 9.56232116511957e-07, - "loss": 0.9936, + "learning_rate": 1.0048884456839158e-06, + "loss": 4.9008, "step": 11990 }, { "epoch": 0.96, - "learning_rate": 9.370419346872217e-07, - "loss": 1.0011, + "learning_rate": 9.852369125470829e-07, + "loss": 4.9398, "step": 11995 }, { "epoch": 0.96, - "learning_rate": 9.180453627418372e-07, - "loss": 0.9892, + "learning_rate": 9.657784824101956e-07, + "loss": 4.8871, "step": 12000 }, { "epoch": 0.96, - "learning_rate": 8.992424378029495e-07, - "loss": 0.9783, + "learning_rate": 9.465131932219539e-07, + "loss": 4.8575, "step": 12005 }, { "epoch": 0.96, - "learning_rate": 8.806331966192405e-07, - "loss": 0.963, + "learning_rate": 9.2744108255437e-07, + "loss": 4.9452, "step": 12010 }, { "epoch": 0.96, - "learning_rate": 8.622176755608168e-07, - "loss": 0.9549, + "learning_rate": 9.08562187602735e-07, + "loss": 4.8257, "step": 12015 }, { "epoch": 0.96, - "learning_rate": 8.439959106192108e-07, - "loss": 0.9656, + "learning_rate": 8.898765451855085e-07, + "loss": 4.8414, "step": 12020 }, { "epoch": 0.96, - "learning_rate": 8.259679374072793e-07, - "loss": 0.9597, + "learning_rate": 8.713841917442622e-07, + "loss": 4.9276, "step": 12025 }, { "epoch": 0.96, - "learning_rate": 8.081337911591047e-07, - "loss": 1.0491, + "learning_rate": 8.530851633436033e-07, + "loss": 4.7975, "step": 12030 }, { "epoch": 0.96, - "learning_rate": 7.904935067299724e-07, - "loss": 1.0683, + "learning_rate": 8.349794956711176e-07, + "loss": 4.8384, "step": 12035 }, { "epoch": 0.96, - "learning_rate": 7.730471185962706e-07, - "loss": 0.9804, + "learning_rate": 8.170672240372934e-07, + "loss": 4.8761, "step": 12040 }, { "epoch": 0.96, - "learning_rate": 7.557946608554356e-07, - "loss": 0.9419, + "learning_rate": 7.9934838337542e-07, + "loss": 4.8349, "step": 12045 }, { - "epoch": 0.97, - "learning_rate": 7.387361672258841e-07, - "loss": 0.9527, + "epoch": 0.96, + "learning_rate": 7.818230082415778e-07, + "loss": 4.9221, "step": 12050 }, { - "epoch": 0.97, - "learning_rate": 7.218716710469586e-07, - "loss": 1.0131, + "epoch": 0.96, + "learning_rate": 7.644911328145265e-07, + "loss": 4.8829, "step": 12055 }, { - "epoch": 0.97, - "learning_rate": 7.052012052788381e-07, - "loss": 0.9701, + "epoch": 0.96, + "learning_rate": 7.473527908956834e-07, + "loss": 4.8685, "step": 12060 }, { "epoch": 0.97, - "learning_rate": 6.887248025024829e-07, - "loss": 0.9951, + "learning_rate": 7.304080159089899e-07, + "loss": 4.8311, "step": 12065 }, { "epoch": 0.97, - "learning_rate": 6.724424949195785e-07, - "loss": 0.9704, + "learning_rate": 7.136568409009003e-07, + "loss": 4.8889, "step": 12070 }, { "epoch": 0.97, - "learning_rate": 6.563543143524586e-07, - "loss": 0.9638, + "learning_rate": 6.970992985403047e-07, + "loss": 4.8884, "step": 12075 }, { "epoch": 0.97, - "learning_rate": 6.40460292244105e-07, - "loss": 0.957, + "learning_rate": 6.807354211184613e-07, + "loss": 4.9059, "step": 12080 }, { "epoch": 0.97, - "learning_rate": 6.247604596579471e-07, - "loss": 0.9595, + "learning_rate": 6.645652405489311e-07, + "loss": 4.8943, "step": 12085 }, { "epoch": 0.97, - "learning_rate": 6.092548472779625e-07, - "loss": 1.0176, + "learning_rate": 6.485887883675213e-07, + "loss": 4.9668, "step": 12090 }, { "epoch": 0.97, - "learning_rate": 5.939434854084991e-07, - "loss": 0.9676, + "learning_rate": 6.328060957322191e-07, + "loss": 4.9407, "step": 12095 }, { "epoch": 0.97, - "learning_rate": 5.788264039742641e-07, - "loss": 1.0056, + "learning_rate": 6.172171934231252e-07, + "loss": 4.8871, "step": 12100 }, { "epoch": 0.97, - "learning_rate": 5.639036325203018e-07, - "loss": 1.0322, + "learning_rate": 6.018221118424316e-07, + "loss": 4.9058, "step": 12105 }, { "epoch": 0.97, - "learning_rate": 5.491752002118378e-07, - "loss": 0.971, + "learning_rate": 5.866208810142881e-07, + "loss": 4.9351, "step": 12110 }, { "epoch": 0.97, - "learning_rate": 5.346411358343018e-07, - "loss": 1.0047, + "learning_rate": 5.716135305848358e-07, + "loss": 4.869, "step": 12115 }, { "epoch": 0.97, - "learning_rate": 5.203014677932605e-07, - "loss": 0.9359, + "learning_rate": 5.568000898220626e-07, + "loss": 4.9545, "step": 12120 }, { "epoch": 0.97, - "learning_rate": 5.061562241143403e-07, - "loss": 0.9365, + "learning_rate": 5.421805876158037e-07, + "loss": 4.8983, "step": 12125 }, { "epoch": 0.97, - "learning_rate": 4.922054324431824e-07, - "loss": 0.9505, + "learning_rate": 5.277550524776631e-07, + "loss": 4.907, "step": 12130 }, { "epoch": 0.97, - "learning_rate": 4.784491200453767e-07, - "loss": 1.0046, + "learning_rate": 5.135235125409699e-07, + "loss": 4.9065, "step": 12135 }, { "epoch": 0.97, - "learning_rate": 4.6488731380643913e-07, - "loss": 0.9765, + "learning_rate": 4.99485995560689e-07, + "loss": 4.8111, "step": 12140 }, { "epoch": 0.97, - "learning_rate": 4.5152004023173434e-07, - "loss": 0.9421, + "learning_rate": 4.856425289134326e-07, + "loss": 4.8935, "step": 12145 }, { "epoch": 0.97, - "learning_rate": 4.3834732544641986e-07, - "loss": 0.9633, + "learning_rate": 4.719931395973265e-07, + "loss": 4.9406, "step": 12150 }, { "epoch": 0.97, - "learning_rate": 4.2536919519542415e-07, - "loss": 1.0033, + "learning_rate": 4.585378542320218e-07, + "loss": 4.9192, "step": 12155 }, { "epoch": 0.97, - "learning_rate": 4.125856748433576e-07, - "loss": 1.0371, + "learning_rate": 4.4527669905862767e-07, + "loss": 4.9689, "step": 12160 }, { "epoch": 0.97, - "learning_rate": 3.9999678937450157e-07, - "loss": 0.9734, + "learning_rate": 4.32209699939623e-07, + "loss": 4.9329, "step": 12165 }, { "epoch": 0.97, - "learning_rate": 3.876025633927305e-07, - "loss": 0.9902, + "learning_rate": 4.193368823588673e-07, + "loss": 4.8706, "step": 12170 }, { - "epoch": 0.98, - "learning_rate": 3.754030211214565e-07, - "loss": 0.9486, + "epoch": 0.97, + "learning_rate": 4.066582714214895e-07, + "loss": 4.8466, "step": 12175 }, { - "epoch": 0.98, - "learning_rate": 3.6339818640364066e-07, - "loss": 0.9891, + "epoch": 0.97, + "learning_rate": 3.9417389185388843e-07, + "loss": 4.8855, "step": 12180 }, { - "epoch": 0.98, - "learning_rate": 3.515880827016704e-07, - "loss": 0.958, + "epoch": 0.97, + "learning_rate": 3.8188376800365467e-07, + "loss": 4.8546, "step": 12185 }, { "epoch": 0.98, - "learning_rate": 3.3997273309734903e-07, - "loss": 0.9827, + "learning_rate": 3.697879238395374e-07, + "loss": 4.9741, "step": 12190 }, { "epoch": 0.98, - "learning_rate": 3.2855216029187285e-07, - "loss": 1.0255, + "learning_rate": 3.5788638295137787e-07, + "loss": 4.8376, "step": 12195 }, { "epoch": 0.98, - "learning_rate": 3.173263866057541e-07, - "loss": 0.9761, + "learning_rate": 3.4617916855010925e-07, + "loss": 4.8879, "step": 12200 }, { "epoch": 0.98, - "learning_rate": 3.062954339787871e-07, - "loss": 0.9997, + "learning_rate": 3.346663034676567e-07, + "loss": 4.8881, "step": 12205 }, { "epoch": 0.98, - "learning_rate": 2.954593239699821e-07, - "loss": 1.0075, + "learning_rate": 3.233478101569154e-07, + "loss": 4.8722, "step": 12210 }, { "epoch": 0.98, - "learning_rate": 2.8481807775759817e-07, - "loss": 0.9932, + "learning_rate": 3.1222371069171695e-07, + "loss": 4.9611, "step": 12215 }, { "epoch": 0.98, - "learning_rate": 2.7437171613898807e-07, - "loss": 0.9918, + "learning_rate": 3.012940267667852e-07, + "loss": 4.872, "step": 12220 }, { "epoch": 0.98, - "learning_rate": 2.6412025953066466e-07, - "loss": 1.027, + "learning_rate": 2.9055877969766943e-07, + "loss": 4.8799, "step": 12225 }, { "epoch": 0.98, - "learning_rate": 2.540637279682123e-07, - "loss": 0.9509, + "learning_rate": 2.800179904207223e-07, + "loss": 4.8936, "step": 12230 }, { "epoch": 0.98, - "learning_rate": 2.442021411062312e-07, - "loss": 0.9862, + "learning_rate": 2.6967167949306646e-07, + "loss": 4.8987, "step": 12235 }, { "epoch": 0.98, - "learning_rate": 2.345355182183373e-07, - "loss": 1.0262, + "learning_rate": 2.595198670925392e-07, + "loss": 4.8686, "step": 12240 }, { "epoch": 0.98, - "learning_rate": 2.2506387819708486e-07, - "loss": 0.9292, + "learning_rate": 2.4956257301764765e-07, + "loss": 4.8556, "step": 12245 }, { "epoch": 0.98, - "learning_rate": 2.1578723955397727e-07, - "loss": 0.9703, + "learning_rate": 2.3979981668756924e-07, + "loss": 4.8322, "step": 12250 }, { "epoch": 0.98, - "learning_rate": 2.0670562041938957e-07, - "loss": 0.9865, + "learning_rate": 2.3023161714204044e-07, + "loss": 4.8871, "step": 12255 }, { "epoch": 0.98, - "learning_rate": 1.9781903854254603e-07, - "loss": 0.9788, + "learning_rate": 2.208579930414123e-07, + "loss": 4.9173, "step": 12260 }, { "epoch": 0.98, - "learning_rate": 1.8912751129149807e-07, - "loss": 1.0124, + "learning_rate": 2.1167896266652831e-07, + "loss": 4.9008, "step": 12265 }, { "epoch": 0.98, - "learning_rate": 1.8063105565306882e-07, - "loss": 1.0247, + "learning_rate": 2.0269454391874666e-07, + "loss": 4.8397, "step": 12270 }, { "epoch": 0.98, - "learning_rate": 1.723296882328529e-07, - "loss": 0.9302, + "learning_rate": 1.9390475431989575e-07, + "loss": 4.8498, "step": 12275 }, { "epoch": 0.98, - "learning_rate": 1.642234252551278e-07, - "loss": 1.0086, + "learning_rate": 1.8530961101218547e-07, + "loss": 4.9186, "step": 12280 }, { "epoch": 0.98, - "learning_rate": 1.563122825628871e-07, - "loss": 0.9392, + "learning_rate": 1.7690913075827375e-07, + "loss": 4.8835, "step": 12285 }, { "epoch": 0.98, - "learning_rate": 1.485962756177628e-07, - "loss": 0.9478, + "learning_rate": 1.6870332994114445e-07, + "loss": 4.8807, "step": 12290 }, { "epoch": 0.98, - "learning_rate": 1.4107541950001413e-07, - "loss": 1.0164, + "learning_rate": 1.6069222456410739e-07, + "loss": 4.8744, "step": 12295 }, { - "epoch": 0.99, - "learning_rate": 1.3374972890849436e-07, - "loss": 0.95, + "epoch": 0.98, + "learning_rate": 1.52875830250776e-07, + "loss": 4.8511, "step": 12300 }, { - "epoch": 0.99, - "learning_rate": 1.266192181606285e-07, - "loss": 0.9637, + "epoch": 0.98, + "learning_rate": 1.452541622450454e-07, + "loss": 4.8279, "step": 12305 }, { - "epoch": 0.99, - "learning_rate": 1.1968390119236895e-07, - "loss": 0.9397, + "epoch": 0.98, + "learning_rate": 1.3782723541101439e-07, + "loss": 4.8916, "step": 12310 }, { "epoch": 0.99, - "learning_rate": 1.1294379155818435e-07, - "loss": 1.001, + "learning_rate": 1.3059506423303002e-07, + "loss": 4.9772, "step": 12315 }, { "epoch": 0.99, - "learning_rate": 1.0639890243104856e-07, - "loss": 0.9631, + "learning_rate": 1.2355766281556548e-07, + "loss": 4.8959, "step": 12320 }, { "epoch": 0.99, - "learning_rate": 1.0004924660235171e-07, - "loss": 0.9672, + "learning_rate": 1.1671504488329765e-07, + "loss": 4.8024, "step": 12325 }, { "epoch": 0.99, - "learning_rate": 9.389483648195585e-08, - "loss": 0.9424, + "learning_rate": 1.1006722378098521e-07, + "loss": 4.8367, "step": 12330 }, { "epoch": 0.99, - "learning_rate": 8.793568409810604e-08, - "loss": 0.9719, + "learning_rate": 1.0361421247352398e-07, + "loss": 4.7849, "step": 12335 }, { "epoch": 0.99, - "learning_rate": 8.217180109745259e-08, - "loss": 0.9974, + "learning_rate": 9.735602354584706e-08, + "loss": 4.898, "step": 12340 }, { "epoch": 0.99, - "learning_rate": 7.660319874498445e-08, - "loss": 0.9498, + "learning_rate": 9.12926692029692e-08, + "loss": 4.7917, "step": 12345 }, { "epoch": 0.99, - "learning_rate": 7.122988792406249e-08, - "loss": 0.9245, + "learning_rate": 8.542416126989805e-08, + "loss": 4.9065, "step": 12350 }, { "epoch": 0.99, - "learning_rate": 6.605187913635291e-08, - "loss": 1.0489, + "learning_rate": 7.975051119166743e-08, + "loss": 4.9359, "step": 12355 }, { "epoch": 0.99, - "learning_rate": 6.106918250181615e-08, - "loss": 1.002, + "learning_rate": 7.427173003329291e-08, + "loss": 4.8793, "step": 12360 }, { "epoch": 0.99, - "learning_rate": 5.628180775869574e-08, - "loss": 0.9605, + "learning_rate": 6.89878284797163e-08, + "loss": 4.9946, "step": 12365 }, { "epoch": 0.99, - "learning_rate": 5.168976426349614e-08, - "loss": 0.9628, + "learning_rate": 6.389881683585009e-08, + "loss": 4.7779, "step": 12370 }, { "epoch": 0.99, - "learning_rate": 4.729306099096054e-08, - "loss": 0.9547, + "learning_rate": 5.900470502651079e-08, + "loss": 4.8502, "step": 12375 }, { "epoch": 0.99, - "learning_rate": 4.3091706534070796e-08, - "loss": 1.0496, + "learning_rate": 5.430550259640787e-08, + "loss": 4.8275, "step": 12380 }, { "epoch": 0.99, - "learning_rate": 3.908570910400311e-08, - "loss": 0.9697, + "learning_rate": 4.980121871014376e-08, + "loss": 4.8925, "step": 12385 }, { "epoch": 0.99, - "learning_rate": 3.527507653012796e-08, - "loss": 0.9689, + "learning_rate": 4.5491862152169387e-08, + "loss": 4.8509, "step": 12390 }, { "epoch": 0.99, - "learning_rate": 3.165981625998793e-08, - "loss": 1.0175, + "learning_rate": 4.137744132677313e-08, + "loss": 4.8149, "step": 12395 }, { "epoch": 0.99, - "learning_rate": 2.8239935359297697e-08, - "loss": 0.9772, + "learning_rate": 3.7457964258102995e-08, + "loss": 4.8698, "step": 12400 }, { "epoch": 0.99, - "learning_rate": 2.5015440511910738e-08, - "loss": 0.9482, + "learning_rate": 3.373343859007783e-08, + "loss": 4.8412, "step": 12405 }, { "epoch": 0.99, - "learning_rate": 2.1986338019819308e-08, - "loss": 0.9312, + "learning_rate": 3.020387158645388e-08, + "loss": 4.9293, "step": 12410 }, { "epoch": 0.99, - "learning_rate": 1.9152633803154462e-08, - "loss": 1.0161, + "learning_rate": 2.686927013072493e-08, + "loss": 4.9711, "step": 12415 }, { "epoch": 0.99, - "learning_rate": 1.6514333400130534e-08, - "loss": 0.9715, + "learning_rate": 2.3729640726199987e-08, + "loss": 4.9132, "step": 12420 }, { - "epoch": 1.0, - "learning_rate": 1.4071441967067334e-08, - "loss": 0.9517, + "epoch": 0.99, + "learning_rate": 2.0784989495914454e-08, + "loss": 4.9003, "step": 12425 }, { - "epoch": 1.0, - "learning_rate": 1.1823964278401268e-08, - "loss": 1.0224, + "epoch": 0.99, + "learning_rate": 1.8035322182652357e-08, + "loss": 4.9498, "step": 12430 }, { - "epoch": 1.0, - "learning_rate": 9.771904726607605e-09, - "loss": 1.0485, + "epoch": 0.99, + "learning_rate": 1.5480644148957448e-08, + "loss": 4.9162, "step": 12435 }, { "epoch": 1.0, - "learning_rate": 7.915267322278208e-09, - "loss": 0.9403, + "learning_rate": 1.3120960377055458e-08, + "loss": 4.9288, "step": 12440 }, { "epoch": 1.0, - "learning_rate": 6.254055694043803e-09, - "loss": 1.0126, + "learning_rate": 1.095627546892075e-08, + "loss": 4.8749, "step": 12445 }, { "epoch": 1.0, - "learning_rate": 4.7882730885850935e-09, - "loss": 0.9419, + "learning_rate": 8.986593646209685e-09, + "loss": 4.87, "step": 12450 }, { "epoch": 1.0, - "learning_rate": 3.517922370654958e-09, - "loss": 0.9457, + "learning_rate": 7.21191875028282e-09, + "loss": 4.9153, "step": 12455 }, { "epoch": 1.0, - "learning_rate": 2.443006023045147e-09, - "loss": 0.9417, + "learning_rate": 5.6322542422049266e-09, + "loss": 4.9541, "step": 12460 }, { "epoch": 1.0, - "learning_rate": 1.5635261465751783e-09, - "loss": 1.0143, + "learning_rate": 4.2476032026894656e-09, + "loss": 4.9385, "step": 12465 }, { "epoch": 1.0, - "learning_rate": 8.794844601256458e-10, - "loss": 0.944, + "learning_rate": 3.057968332154104e-09, + "loss": 4.8223, "step": 12470 }, { "epoch": 1.0, - "learning_rate": 3.9088230059380893e-10, - "loss": 0.9389, + "learning_rate": 2.0633519506763067e-09, + "loss": 4.8889, "step": 12475 }, { "epoch": 1.0, - "learning_rate": 9.77206228935934e-11, - "loss": 0.9592, + "learning_rate": 1.2637559980044344e-09, + "loss": 4.783, "step": 12480 }, { "epoch": 1.0, - "learning_rate": 0.0, - "loss": 1.0163, + "learning_rate": 6.591820335355437e-10, + "loss": 4.8535, "step": 12485 }, { "epoch": 1.0, - "eval_loss": 1.0681166648864746, - "eval_runtime": 551.6979, - "eval_samples_per_second": 2.697, - "eval_steps_per_second": 0.337, - "step": 12485 + "learning_rate": 2.496312363486908e-10, + "loss": 4.8704, + "step": 12490 + }, + { + "epoch": 1.0, + "learning_rate": 3.510440516052427e-11, + "loss": 4.8496, + "step": 12495 + }, + { + "epoch": 1.0, + "eval_loss": 4.886576175689697, + "eval_runtime": 553.3141, + "eval_samples_per_second": 2.689, + "eval_steps_per_second": 0.336, + "step": 12498 }, { "epoch": 1.0, - "step": 12485, - "total_flos": 1.8387060424704e+16, - "train_loss": 1.0040549779730794, - "train_runtime": 174545.1979, - "train_samples_per_second": 1.144, - "train_steps_per_second": 0.072 + "step": 12498, + "total_flos": 1.8406206701568e+16, + "train_loss": 4.727149732982012, + "train_runtime": 175008.638, + "train_samples_per_second": 1.143, + "train_steps_per_second": 0.071 } ], "logging_steps": 5, - "max_steps": 12485, + "max_steps": 12498, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 100, - "total_flos": 1.8387060424704e+16, + "total_flos": 1.8406206701568e+16, "train_batch_size": 2, "trial_name": null, "trial_params": null