| { | |
| "best_metric": 0.9792746113989638, | |
| "best_model_checkpoint": "videomae-base-finetuned-ucf101\\checkpoint-3328", | |
| "epoch": 3.2490985576923075, | |
| "eval_steps": 500, | |
| "global_step": 3328, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.5015015015015015e-06, | |
| "loss": 3.914, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.003003003003003e-06, | |
| "loss": 3.964, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.504504504504505e-06, | |
| "loss": 3.9988, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.006006006006006e-06, | |
| "loss": 3.967, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.507507507507508e-06, | |
| "loss": 3.9326, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.00900900900901e-06, | |
| "loss": 4.0274, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.051051051051051e-05, | |
| "loss": 3.9367, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.2012012012012012e-05, | |
| "loss": 3.9813, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.3513513513513515e-05, | |
| "loss": 3.898, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.5015015015015016e-05, | |
| "loss": 3.8262, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.6516516516516518e-05, | |
| "loss": 3.8733, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.801801801801802e-05, | |
| "loss": 3.847, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.951951951951952e-05, | |
| "loss": 3.7127, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.102102102102102e-05, | |
| "loss": 3.7771, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.2522522522522523e-05, | |
| "loss": 3.704, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.4024024024024024e-05, | |
| "loss": 3.7173, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.552552552552553e-05, | |
| "loss": 3.6565, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.702702702702703e-05, | |
| "loss": 3.6562, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.852852852852853e-05, | |
| "loss": 3.5141, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.0030030030030033e-05, | |
| "loss": 3.4873, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.153153153153153e-05, | |
| "loss": 3.4563, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.3033033033033035e-05, | |
| "loss": 3.3595, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.453453453453453e-05, | |
| "loss": 3.2937, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.603603603603604e-05, | |
| "loss": 3.1955, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.7537537537537536e-05, | |
| "loss": 3.1335, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.903903903903904e-05, | |
| "loss": 3.0948, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.0540540540540545e-05, | |
| "loss": 2.9545, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.204204204204204e-05, | |
| "loss": 3.0115, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.354354354354355e-05, | |
| "loss": 2.8002, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.5045045045045046e-05, | |
| "loss": 2.7634, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.654654654654655e-05, | |
| "loss": 2.6911, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.804804804804805e-05, | |
| "loss": 2.5804, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.954954954954955e-05, | |
| "loss": 2.6476, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.988313856427379e-05, | |
| "loss": 2.4828, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.971619365609349e-05, | |
| "loss": 2.354, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.9549248747913194e-05, | |
| "loss": 2.5351, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.938230383973289e-05, | |
| "loss": 2.5771, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.921535893155259e-05, | |
| "loss": 2.642, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.904841402337229e-05, | |
| "loss": 2.1664, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.888146911519199e-05, | |
| "loss": 2.1555, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.871452420701168e-05, | |
| "loss": 2.0825, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.8547579298831386e-05, | |
| "loss": 2.2401, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.838063439065109e-05, | |
| "loss": 2.2106, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.8213689482470785e-05, | |
| "loss": 2.0336, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.804674457429049e-05, | |
| "loss": 2.1065, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.7879799666110185e-05, | |
| "loss": 2.0611, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.771285475792989e-05, | |
| "loss": 1.9548, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.7545909849749585e-05, | |
| "loss": 1.8221, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.737896494156929e-05, | |
| "loss": 1.9274, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.7212020033388984e-05, | |
| "loss": 2.2112, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.704507512520868e-05, | |
| "loss": 1.7819, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.6878130217028384e-05, | |
| "loss": 1.8032, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.671118530884808e-05, | |
| "loss": 1.646, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.6544240400667784e-05, | |
| "loss": 1.7348, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.637729549248748e-05, | |
| "loss": 1.664, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.621035058430718e-05, | |
| "loss": 1.6569, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.604340567612688e-05, | |
| "loss": 1.493, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.5876460767946576e-05, | |
| "loss": 1.5023, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.570951585976628e-05, | |
| "loss": 1.3915, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.5542570951585976e-05, | |
| "loss": 1.4739, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.537562604340568e-05, | |
| "loss": 1.5772, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.5208681135225376e-05, | |
| "loss": 1.418, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.504173622704508e-05, | |
| "loss": 1.2303, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.4874791318864775e-05, | |
| "loss": 1.3034, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.470784641068448e-05, | |
| "loss": 1.3198, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.4540901502504175e-05, | |
| "loss": 1.3501, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.437395659432387e-05, | |
| "loss": 1.305, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.4207011686143574e-05, | |
| "loss": 1.063, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.404006677796327e-05, | |
| "loss": 1.2372, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.3873121869782974e-05, | |
| "loss": 1.3926, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.370617696160268e-05, | |
| "loss": 1.0836, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.3539232053422374e-05, | |
| "loss": 1.2272, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.337228714524208e-05, | |
| "loss": 1.2089, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.3205342237061773e-05, | |
| "loss": 0.9882, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.303839732888147e-05, | |
| "loss": 1.1729, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.2871452420701166e-05, | |
| "loss": 1.1109, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.270450751252087e-05, | |
| "loss": 1.1041, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.253756260434057e-05, | |
| "loss": 0.9836, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.237061769616027e-05, | |
| "loss": 0.952, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.220367278797997e-05, | |
| "loss": 1.0417, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.203672787979967e-05, | |
| "loss": 0.8411, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.186978297161937e-05, | |
| "loss": 0.9198, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.170283806343906e-05, | |
| "loss": 1.0148, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "eval_accuracy": 0.6824574389341229, | |
| "eval_loss": 1.1000982522964478, | |
| "eval_runtime": 109.0663, | |
| "eval_samples_per_second": 12.387, | |
| "eval_steps_per_second": 2.072, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.1535893155258765e-05, | |
| "loss": 0.9924, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.136894824707847e-05, | |
| "loss": 0.8616, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.1202003338898165e-05, | |
| "loss": 0.5398, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.103505843071787e-05, | |
| "loss": 0.9011, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.0868113522537564e-05, | |
| "loss": 0.978, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.070116861435727e-05, | |
| "loss": 0.7954, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.0534223706176964e-05, | |
| "loss": 0.6879, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.036727879799666e-05, | |
| "loss": 0.6882, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 4.0200333889816363e-05, | |
| "loss": 0.7091, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 4.003338898163606e-05, | |
| "loss": 1.0599, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.986644407345576e-05, | |
| "loss": 0.7129, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.969949916527546e-05, | |
| "loss": 0.765, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.953255425709516e-05, | |
| "loss": 0.6372, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.936560934891486e-05, | |
| "loss": 0.7401, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.919866444073456e-05, | |
| "loss": 0.5159, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.903171953255426e-05, | |
| "loss": 0.656, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.8864774624373955e-05, | |
| "loss": 0.6563, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.869782971619366e-05, | |
| "loss": 0.7938, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.8530884808013355e-05, | |
| "loss": 0.7291, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.836393989983306e-05, | |
| "loss": 0.7122, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.8196994991652755e-05, | |
| "loss": 0.4916, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.803005008347246e-05, | |
| "loss": 0.4931, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.7863105175292154e-05, | |
| "loss": 0.6749, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.769616026711186e-05, | |
| "loss": 0.6569, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.7529215358931554e-05, | |
| "loss": 0.5455, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.736227045075125e-05, | |
| "loss": 0.5533, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.7195325542570954e-05, | |
| "loss": 0.521, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.702838063439065e-05, | |
| "loss": 0.5858, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.686143572621035e-05, | |
| "loss": 0.3544, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.669449081803005e-05, | |
| "loss": 0.5821, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.652754590984975e-05, | |
| "loss": 0.821, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.6360601001669456e-05, | |
| "loss": 0.6644, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.6193656093489146e-05, | |
| "loss": 0.7287, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.602671118530885e-05, | |
| "loss": 0.6521, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.5859766277128545e-05, | |
| "loss": 0.5092, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.569282136894825e-05, | |
| "loss": 0.5223, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.5525876460767945e-05, | |
| "loss": 0.5667, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.535893155258765e-05, | |
| "loss": 0.6028, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.519198664440735e-05, | |
| "loss": 0.4515, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.502504173622705e-05, | |
| "loss": 0.5617, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.4858096828046744e-05, | |
| "loss": 0.6244, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.469115191986644e-05, | |
| "loss": 0.4726, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.4524207011686144e-05, | |
| "loss": 0.4253, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.435726210350585e-05, | |
| "loss": 0.5008, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.4190317195325544e-05, | |
| "loss": 0.6454, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.402337228714525e-05, | |
| "loss": 0.5715, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.385642737896494e-05, | |
| "loss": 0.5909, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.3689482470784646e-05, | |
| "loss": 0.5649, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.352253756260434e-05, | |
| "loss": 0.667, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.335559265442404e-05, | |
| "loss": 0.6362, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.318864774624374e-05, | |
| "loss": 0.4193, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.302170283806344e-05, | |
| "loss": 0.3466, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.285475792988314e-05, | |
| "loss": 0.3768, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.268781302170284e-05, | |
| "loss": 0.4257, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.252086811352254e-05, | |
| "loss": 0.5538, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.235392320534224e-05, | |
| "loss": 0.3771, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.218697829716194e-05, | |
| "loss": 0.3839, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.202003338898164e-05, | |
| "loss": 0.2901, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.1853088480801334e-05, | |
| "loss": 0.3403, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.168614357262104e-05, | |
| "loss": 0.2758, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.1519198664440734e-05, | |
| "loss": 0.4401, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.135225375626044e-05, | |
| "loss": 0.4628, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.1185308848080134e-05, | |
| "loss": 0.3402, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.101836393989984e-05, | |
| "loss": 0.3673, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.085141903171953e-05, | |
| "loss": 0.3835, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.068447412353923e-05, | |
| "loss": 0.2576, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.051752921535893e-05, | |
| "loss": 0.255, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.035058430717863e-05, | |
| "loss": 0.3277, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 3.0183639398998333e-05, | |
| "loss": 0.3722, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 3.0016694490818032e-05, | |
| "loss": 0.4049, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 2.9849749582637732e-05, | |
| "loss": 0.3366, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.9682804674457432e-05, | |
| "loss": 0.4142, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.9515859766277132e-05, | |
| "loss": 0.2457, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.9348914858096832e-05, | |
| "loss": 0.3911, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.9181969949916528e-05, | |
| "loss": 0.2174, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 2.9015025041736228e-05, | |
| "loss": 0.4427, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 2.8848080133555928e-05, | |
| "loss": 0.3746, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 2.8681135225375628e-05, | |
| "loss": 0.4656, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 2.8514190317195328e-05, | |
| "loss": 0.4516, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 2.8347245409015027e-05, | |
| "loss": 0.2713, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 2.8180300500834727e-05, | |
| "loss": 0.3875, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.8013355592654427e-05, | |
| "loss": 0.328, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.7846410684474123e-05, | |
| "loss": 0.2036, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "eval_accuracy": 0.8734270910436713, | |
| "eval_loss": 0.4224362075328827, | |
| "eval_runtime": 103.8604, | |
| "eval_samples_per_second": 13.008, | |
| "eval_steps_per_second": 2.176, | |
| "step": 1666 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 2.7679465776293823e-05, | |
| "loss": 0.2257, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 2.7512520868113523e-05, | |
| "loss": 0.4037, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 2.7345575959933223e-05, | |
| "loss": 0.1379, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 2.7178631051752923e-05, | |
| "loss": 0.2688, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 2.7011686143572623e-05, | |
| "loss": 0.1742, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 2.6844741235392322e-05, | |
| "loss": 0.1701, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 2.6677796327212022e-05, | |
| "loss": 0.2653, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 2.651085141903172e-05, | |
| "loss": 0.128, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 2.634390651085142e-05, | |
| "loss": 0.1489, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 2.6176961602671118e-05, | |
| "loss": 0.1623, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 2.6010016694490818e-05, | |
| "loss": 0.4388, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 2.5843071786310518e-05, | |
| "loss": 0.1918, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 2.5676126878130218e-05, | |
| "loss": 0.1947, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 2.5509181969949918e-05, | |
| "loss": 0.176, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 2.534223706176962e-05, | |
| "loss": 0.2615, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 2.517529215358932e-05, | |
| "loss": 0.1719, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 2.5008347245409014e-05, | |
| "loss": 0.159, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 2.4841402337228717e-05, | |
| "loss": 0.205, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 2.4674457429048413e-05, | |
| "loss": 0.1687, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 2.4507512520868113e-05, | |
| "loss": 0.0947, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 2.4340567612687813e-05, | |
| "loss": 0.1372, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 2.4173622704507516e-05, | |
| "loss": 0.3354, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 2.4006677796327213e-05, | |
| "loss": 0.2607, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 2.3839732888146912e-05, | |
| "loss": 0.2231, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 2.3672787979966612e-05, | |
| "loss": 0.2651, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 2.3505843071786312e-05, | |
| "loss": 0.1972, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 2.333889816360601e-05, | |
| "loss": 0.3044, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 2.3171953255425712e-05, | |
| "loss": 0.3482, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 2.300500834724541e-05, | |
| "loss": 0.2268, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 2.283806343906511e-05, | |
| "loss": 0.2141, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 2.2671118530884808e-05, | |
| "loss": 0.2173, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 2.2504173622704508e-05, | |
| "loss": 0.3188, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 2.2337228714524207e-05, | |
| "loss": 0.3457, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 2.2170283806343907e-05, | |
| "loss": 0.2314, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 2.2003338898163607e-05, | |
| "loss": 0.3628, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 2.1836393989983307e-05, | |
| "loss": 0.2104, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 2.1669449081803007e-05, | |
| "loss": 0.18, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 2.1502504173622707e-05, | |
| "loss": 0.1475, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 2.1335559265442406e-05, | |
| "loss": 0.0824, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 2.1168614357262103e-05, | |
| "loss": 0.2078, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 2.1001669449081803e-05, | |
| "loss": 0.1055, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 2.0834724540901503e-05, | |
| "loss": 0.2121, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 2.0667779632721206e-05, | |
| "loss": 0.0923, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 2.0500834724540902e-05, | |
| "loss": 0.1223, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 2.0333889816360602e-05, | |
| "loss": 0.0359, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 2.0166944908180302e-05, | |
| "loss": 0.1897, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 2e-05, | |
| "loss": 0.114, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.9833055091819698e-05, | |
| "loss": 0.0897, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.96661101836394e-05, | |
| "loss": 0.4087, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.94991652754591e-05, | |
| "loss": 0.2505, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.93322203672788e-05, | |
| "loss": 0.2011, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.9165275459098497e-05, | |
| "loss": 0.1263, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1.8998330550918197e-05, | |
| "loss": 0.3548, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1.8831385642737897e-05, | |
| "loss": 0.195, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1.8664440734557597e-05, | |
| "loss": 0.0706, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.8497495826377297e-05, | |
| "loss": 0.0723, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.8330550918196996e-05, | |
| "loss": 0.3366, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.8163606010016696e-05, | |
| "loss": 0.1234, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.7996661101836396e-05, | |
| "loss": 0.0387, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.7829716193656093e-05, | |
| "loss": 0.3389, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.7662771285475792e-05, | |
| "loss": 0.093, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.7495826377295492e-05, | |
| "loss": 0.166, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.7328881469115192e-05, | |
| "loss": 0.1487, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.7161936560934895e-05, | |
| "loss": 0.0916, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.6994991652754592e-05, | |
| "loss": 0.2313, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.682804674457429e-05, | |
| "loss": 0.0557, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.666110183639399e-05, | |
| "loss": 0.0394, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.649415692821369e-05, | |
| "loss": 0.0668, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.6327212020033388e-05, | |
| "loss": 0.0283, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.6160267111853087e-05, | |
| "loss": 0.0832, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.599332220367279e-05, | |
| "loss": 0.0682, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.582637729549249e-05, | |
| "loss": 0.1193, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.5659432387312187e-05, | |
| "loss": 0.0482, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.5492487479131887e-05, | |
| "loss": 0.0928, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.5325542570951587e-05, | |
| "loss": 0.1088, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.5158597662771286e-05, | |
| "loss": 0.2721, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.4991652754590985e-05, | |
| "loss": 0.1225, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.4824707846410684e-05, | |
| "loss": 0.0691, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.4657762938230384e-05, | |
| "loss": 0.1112, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.4490818030050086e-05, | |
| "loss": 0.1386, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.4323873121869782e-05, | |
| "loss": 0.0625, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.4156928213689482e-05, | |
| "loss": 0.0712, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1.3989983305509183e-05, | |
| "loss": 0.1442, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "eval_accuracy": 0.9659511472982976, | |
| "eval_loss": 0.123105987906456, | |
| "eval_runtime": 103.3128, | |
| "eval_samples_per_second": 13.077, | |
| "eval_steps_per_second": 2.188, | |
| "step": 2499 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 1.3823038397328883e-05, | |
| "loss": 0.0617, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 1.365609348914858e-05, | |
| "loss": 0.0987, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 1.3489148580968281e-05, | |
| "loss": 0.0165, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 1.3322203672787981e-05, | |
| "loss": 0.0411, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 1.3155258764607681e-05, | |
| "loss": 0.0796, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 1.298831385642738e-05, | |
| "loss": 0.0186, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 1.2821368948247079e-05, | |
| "loss": 0.2463, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 1.2654424040066779e-05, | |
| "loss": 0.071, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 1.2487479131886479e-05, | |
| "loss": 0.0651, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 1.2320534223706177e-05, | |
| "loss": 0.0127, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 1.2153589315525878e-05, | |
| "loss": 0.0181, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 1.1986644407345576e-05, | |
| "loss": 0.0615, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 1.1819699499165276e-05, | |
| "loss": 0.0817, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 1.1652754590984976e-05, | |
| "loss": 0.04, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 1.1485809682804676e-05, | |
| "loss": 0.0942, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 1.1318864774624374e-05, | |
| "loss": 0.0639, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 1.1151919866444074e-05, | |
| "loss": 0.0703, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 1.0984974958263774e-05, | |
| "loss": 0.1239, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 1.0818030050083473e-05, | |
| "loss": 0.0737, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 1.0651085141903172e-05, | |
| "loss": 0.2398, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 1.0484140233722873e-05, | |
| "loss": 0.1577, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 1.0317195325542571e-05, | |
| "loss": 0.0713, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 1.0150250417362271e-05, | |
| "loss": 0.013, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 9.983305509181969e-06, | |
| "loss": 0.0636, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 9.81636060100167e-06, | |
| "loss": 0.0115, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 9.649415692821369e-06, | |
| "loss": 0.0144, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 9.482470784641069e-06, | |
| "loss": 0.0889, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 9.315525876460768e-06, | |
| "loss": 0.012, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 9.148580968280468e-06, | |
| "loss": 0.1484, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 8.981636060100166e-06, | |
| "loss": 0.1009, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 8.814691151919866e-06, | |
| "loss": 0.1242, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 8.647746243739568e-06, | |
| "loss": 0.0582, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 8.480801335559266e-06, | |
| "loss": 0.0601, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 8.313856427378966e-06, | |
| "loss": 0.1516, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 8.146911519198665e-06, | |
| "loss": 0.027, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 7.979966611018365e-06, | |
| "loss": 0.009, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 7.813021702838063e-06, | |
| "loss": 0.0884, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 7.646076794657763e-06, | |
| "loss": 0.0186, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 7.479131886477462e-06, | |
| "loss": 0.116, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 7.312186978297163e-06, | |
| "loss": 0.0139, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 7.145242070116862e-06, | |
| "loss": 0.0104, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 6.978297161936562e-06, | |
| "loss": 0.0116, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 6.811352253756261e-06, | |
| "loss": 0.1064, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 6.6444073455759605e-06, | |
| "loss": 0.1323, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 6.4774624373956595e-06, | |
| "loss": 0.0111, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 6.310517529215359e-06, | |
| "loss": 0.0711, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 6.143572621035059e-06, | |
| "loss": 0.0145, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 5.976627712854758e-06, | |
| "loss": 0.0551, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 5.809682804674458e-06, | |
| "loss": 0.0155, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 5.642737896494158e-06, | |
| "loss": 0.0149, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 5.475792988313857e-06, | |
| "loss": 0.0539, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 5.308848080133557e-06, | |
| "loss": 0.1017, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 5.1419031719532556e-06, | |
| "loss": 0.0503, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 4.974958263772955e-06, | |
| "loss": 0.0305, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 4.808013355592654e-06, | |
| "loss": 0.0082, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 4.641068447412354e-06, | |
| "loss": 0.0249, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 4.474123539232054e-06, | |
| "loss": 0.0087, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 4.307178631051753e-06, | |
| "loss": 0.0109, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 4.140233722871453e-06, | |
| "loss": 0.1134, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 3.973288814691152e-06, | |
| "loss": 0.1474, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 3.8063439065108516e-06, | |
| "loss": 0.0091, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 3.639398998330551e-06, | |
| "loss": 0.0928, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 3.4724540901502504e-06, | |
| "loss": 0.0955, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 3.30550918196995e-06, | |
| "loss": 0.0161, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 3.1385642737896492e-06, | |
| "loss": 0.126, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 2.971619365609349e-06, | |
| "loss": 0.0214, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 2.8046744574290484e-06, | |
| "loss": 0.0387, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 2.6377295492487483e-06, | |
| "loss": 0.0361, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 2.4707846410684477e-06, | |
| "loss": 0.2003, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 2.303839732888147e-06, | |
| "loss": 0.1052, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 2.1368948247078465e-06, | |
| "loss": 0.0349, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 1.969949916527546e-06, | |
| "loss": 0.1133, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 1.8030050083472455e-06, | |
| "loss": 0.0548, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 1.6360601001669449e-06, | |
| "loss": 0.0127, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 1.4691151919866445e-06, | |
| "loss": 0.0068, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 1.302170283806344e-06, | |
| "loss": 0.0347, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 1.1352253756260435e-06, | |
| "loss": 0.0728, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 9.68280467445743e-07, | |
| "loss": 0.0176, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 8.013355592654424e-07, | |
| "loss": 0.032, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 6.343906510851419e-07, | |
| "loss": 0.1883, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 4.674457429048414e-07, | |
| "loss": 0.0333, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 3.005008347245409e-07, | |
| "loss": 0.0118, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 1.335559265442404e-07, | |
| "loss": 0.073, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "eval_accuracy": 0.9792746113989638, | |
| "eval_loss": 0.07213647663593292, | |
| "eval_runtime": 108.752, | |
| "eval_samples_per_second": 12.423, | |
| "eval_steps_per_second": 2.078, | |
| "step": 3328 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "step": 3328, | |
| "total_flos": 2.4877331316268204e+19, | |
| "train_loss": 0.7826446327649487, | |
| "train_runtime": 2917.4147, | |
| "train_samples_per_second": 6.844, | |
| "train_steps_per_second": 1.141 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "eval_accuracy": 0.9792746113989638, | |
| "eval_loss": 0.07213647663593292, | |
| "eval_runtime": 107.9394, | |
| "eval_samples_per_second": 12.516, | |
| "eval_steps_per_second": 2.094, | |
| "step": 3328 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "eval_accuracy": 0.0, | |
| "eval_loss": 7.797203063964844, | |
| "eval_runtime": 417.9481, | |
| "eval_samples_per_second": 11.681, | |
| "eval_steps_per_second": 1.948, | |
| "step": 3328 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "eval_accuracy": 0.9792746113989638, | |
| "eval_loss": 0.07213647663593292, | |
| "eval_runtime": 108.9545, | |
| "eval_samples_per_second": 12.4, | |
| "eval_steps_per_second": 2.074, | |
| "step": 3328 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 3328, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 500, | |
| "total_flos": 2.4877331316268204e+19, | |
| "train_batch_size": 6, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |