| { | |
| "best_metric": 0.9879931389365352, | |
| "best_model_checkpoint": "videomae-base-finetuned-numbers-augmented/checkpoint-2816", | |
| "epoch": 3.25, | |
| "eval_steps": 500, | |
| "global_step": 2816, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0035511363636363635, | |
| "grad_norm": 11.368436813354492, | |
| "learning_rate": 1.7730496453900712e-06, | |
| "loss": 2.4086, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.007102272727272727, | |
| "grad_norm": 13.567573547363281, | |
| "learning_rate": 3.5460992907801423e-06, | |
| "loss": 2.4404, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01065340909090909, | |
| "grad_norm": 10.261510848999023, | |
| "learning_rate": 5.319148936170213e-06, | |
| "loss": 2.4437, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.014204545454545454, | |
| "grad_norm": 10.864927291870117, | |
| "learning_rate": 7.092198581560285e-06, | |
| "loss": 2.4537, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.01775568181818182, | |
| "grad_norm": 13.574213027954102, | |
| "learning_rate": 8.865248226950355e-06, | |
| "loss": 2.3604, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.02130681818181818, | |
| "grad_norm": 12.087525367736816, | |
| "learning_rate": 1.0638297872340426e-05, | |
| "loss": 2.4028, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.024857954545454544, | |
| "grad_norm": 16.888160705566406, | |
| "learning_rate": 1.2411347517730498e-05, | |
| "loss": 2.3285, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.028409090909090908, | |
| "grad_norm": 11.21091365814209, | |
| "learning_rate": 1.418439716312057e-05, | |
| "loss": 2.4292, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.03196022727272727, | |
| "grad_norm": 12.686086654663086, | |
| "learning_rate": 1.595744680851064e-05, | |
| "loss": 2.3094, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.03551136363636364, | |
| "grad_norm": 13.735329627990723, | |
| "learning_rate": 1.773049645390071e-05, | |
| "loss": 2.2478, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0390625, | |
| "grad_norm": 15.364631652832031, | |
| "learning_rate": 1.950354609929078e-05, | |
| "loss": 2.2576, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.04261363636363636, | |
| "grad_norm": 17.786842346191406, | |
| "learning_rate": 2.1276595744680852e-05, | |
| "loss": 2.2132, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.04616477272727273, | |
| "grad_norm": 15.442058563232422, | |
| "learning_rate": 2.3049645390070924e-05, | |
| "loss": 2.1968, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.04971590909090909, | |
| "grad_norm": 18.52874755859375, | |
| "learning_rate": 2.4822695035460995e-05, | |
| "loss": 1.9613, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.053267045454545456, | |
| "grad_norm": 16.437816619873047, | |
| "learning_rate": 2.6595744680851064e-05, | |
| "loss": 2.055, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.056818181818181816, | |
| "grad_norm": 16.25989532470703, | |
| "learning_rate": 2.836879432624114e-05, | |
| "loss": 2.2668, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.060369318181818184, | |
| "grad_norm": 16.64414405822754, | |
| "learning_rate": 3.0141843971631207e-05, | |
| "loss": 2.0752, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.06392045454545454, | |
| "grad_norm": 22.03922462463379, | |
| "learning_rate": 3.191489361702128e-05, | |
| "loss": 1.9947, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.06747159090909091, | |
| "grad_norm": 20.077898025512695, | |
| "learning_rate": 3.3687943262411347e-05, | |
| "loss": 1.8888, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.07102272727272728, | |
| "grad_norm": 17.019630432128906, | |
| "learning_rate": 3.546099290780142e-05, | |
| "loss": 1.8263, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.07457386363636363, | |
| "grad_norm": 22.599655151367188, | |
| "learning_rate": 3.723404255319149e-05, | |
| "loss": 1.6843, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.078125, | |
| "grad_norm": 17.443727493286133, | |
| "learning_rate": 3.900709219858156e-05, | |
| "loss": 1.7081, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.08167613636363637, | |
| "grad_norm": 12.519112586975098, | |
| "learning_rate": 4.078014184397163e-05, | |
| "loss": 1.4262, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.08522727272727272, | |
| "grad_norm": 21.003129959106445, | |
| "learning_rate": 4.2553191489361704e-05, | |
| "loss": 1.6321, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.08877840909090909, | |
| "grad_norm": 9.726280212402344, | |
| "learning_rate": 4.432624113475177e-05, | |
| "loss": 1.6712, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.09232954545454546, | |
| "grad_norm": 24.422882080078125, | |
| "learning_rate": 4.609929078014185e-05, | |
| "loss": 2.0967, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.09588068181818182, | |
| "grad_norm": 11.92298698425293, | |
| "learning_rate": 4.787234042553192e-05, | |
| "loss": 1.7163, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.09943181818181818, | |
| "grad_norm": 14.245928764343262, | |
| "learning_rate": 4.964539007092199e-05, | |
| "loss": 1.9799, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.10298295454545454, | |
| "grad_norm": 14.423157691955566, | |
| "learning_rate": 4.9842146803472775e-05, | |
| "loss": 1.6602, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.10653409090909091, | |
| "grad_norm": 11.99425983428955, | |
| "learning_rate": 4.9644830307813736e-05, | |
| "loss": 1.6844, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.11008522727272728, | |
| "grad_norm": 13.40038776397705, | |
| "learning_rate": 4.94475138121547e-05, | |
| "loss": 1.6082, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.11363636363636363, | |
| "grad_norm": 9.363014221191406, | |
| "learning_rate": 4.925019731649566e-05, | |
| "loss": 1.7191, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.1171875, | |
| "grad_norm": 16.138887405395508, | |
| "learning_rate": 4.905288082083662e-05, | |
| "loss": 1.621, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.12073863636363637, | |
| "grad_norm": 13.552618980407715, | |
| "learning_rate": 4.885556432517759e-05, | |
| "loss": 1.4569, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.12428977272727272, | |
| "grad_norm": 16.55213737487793, | |
| "learning_rate": 4.865824782951855e-05, | |
| "loss": 1.5886, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.1278409090909091, | |
| "grad_norm": 4.656749248504639, | |
| "learning_rate": 4.8460931333859514e-05, | |
| "loss": 1.2231, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.13139204545454544, | |
| "grad_norm": 9.768326759338379, | |
| "learning_rate": 4.8263614838200475e-05, | |
| "loss": 1.65, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.13494318181818182, | |
| "grad_norm": 20.023557662963867, | |
| "learning_rate": 4.806629834254144e-05, | |
| "loss": 1.4166, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.13849431818181818, | |
| "grad_norm": 18.180370330810547, | |
| "learning_rate": 4.78689818468824e-05, | |
| "loss": 1.1376, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.14204545454545456, | |
| "grad_norm": 16.621734619140625, | |
| "learning_rate": 4.767166535122337e-05, | |
| "loss": 1.5926, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.1455965909090909, | |
| "grad_norm": 17.98001480102539, | |
| "learning_rate": 4.747434885556433e-05, | |
| "loss": 1.3632, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.14914772727272727, | |
| "grad_norm": 20.42987060546875, | |
| "learning_rate": 4.727703235990529e-05, | |
| "loss": 1.3245, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.15269886363636365, | |
| "grad_norm": 11.056243896484375, | |
| "learning_rate": 4.707971586424625e-05, | |
| "loss": 1.2668, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.15625, | |
| "grad_norm": 13.91126537322998, | |
| "learning_rate": 4.6882399368587215e-05, | |
| "loss": 1.0435, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.15980113636363635, | |
| "grad_norm": 19.263099670410156, | |
| "learning_rate": 4.6685082872928176e-05, | |
| "loss": 1.3369, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.16335227272727273, | |
| "grad_norm": 23.511869430541992, | |
| "learning_rate": 4.6487766377269145e-05, | |
| "loss": 1.1933, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.1669034090909091, | |
| "grad_norm": 14.228583335876465, | |
| "learning_rate": 4.629044988161011e-05, | |
| "loss": 1.591, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.17045454545454544, | |
| "grad_norm": 19.068967819213867, | |
| "learning_rate": 4.609313338595106e-05, | |
| "loss": 0.9233, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.17400568181818182, | |
| "grad_norm": 11.136780738830566, | |
| "learning_rate": 4.589581689029203e-05, | |
| "loss": 1.0833, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.17755681818181818, | |
| "grad_norm": 10.600316047668457, | |
| "learning_rate": 4.569850039463299e-05, | |
| "loss": 1.1883, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.18110795454545456, | |
| "grad_norm": 14.483977317810059, | |
| "learning_rate": 4.550118389897396e-05, | |
| "loss": 1.2224, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.1846590909090909, | |
| "grad_norm": 20.28068733215332, | |
| "learning_rate": 4.530386740331492e-05, | |
| "loss": 0.9271, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.18821022727272727, | |
| "grad_norm": 8.706443786621094, | |
| "learning_rate": 4.5106550907655884e-05, | |
| "loss": 0.876, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.19176136363636365, | |
| "grad_norm": 24.446735382080078, | |
| "learning_rate": 4.4909234411996846e-05, | |
| "loss": 0.9443, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.1953125, | |
| "grad_norm": 18.790678024291992, | |
| "learning_rate": 4.471191791633781e-05, | |
| "loss": 1.5365, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.19886363636363635, | |
| "grad_norm": 12.385162353515625, | |
| "learning_rate": 4.451460142067877e-05, | |
| "loss": 1.1072, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.20241477272727273, | |
| "grad_norm": 10.875085830688477, | |
| "learning_rate": 4.431728492501974e-05, | |
| "loss": 1.3267, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.2059659090909091, | |
| "grad_norm": 23.155399322509766, | |
| "learning_rate": 4.41199684293607e-05, | |
| "loss": 1.0875, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.20951704545454544, | |
| "grad_norm": 16.206789016723633, | |
| "learning_rate": 4.392265193370166e-05, | |
| "loss": 1.532, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.21306818181818182, | |
| "grad_norm": 10.199226379394531, | |
| "learning_rate": 4.372533543804262e-05, | |
| "loss": 1.3542, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.21661931818181818, | |
| "grad_norm": 24.252439498901367, | |
| "learning_rate": 4.3528018942383585e-05, | |
| "loss": 0.9932, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.22017045454545456, | |
| "grad_norm": 21.493459701538086, | |
| "learning_rate": 4.333070244672455e-05, | |
| "loss": 0.9927, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.2237215909090909, | |
| "grad_norm": 14.551782608032227, | |
| "learning_rate": 4.3133385951065515e-05, | |
| "loss": 0.86, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.22727272727272727, | |
| "grad_norm": 36.133609771728516, | |
| "learning_rate": 4.293606945540648e-05, | |
| "loss": 0.7792, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.23082386363636365, | |
| "grad_norm": 13.04199504852295, | |
| "learning_rate": 4.273875295974744e-05, | |
| "loss": 0.9251, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.234375, | |
| "grad_norm": 13.920279502868652, | |
| "learning_rate": 4.25414364640884e-05, | |
| "loss": 1.0642, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.23792613636363635, | |
| "grad_norm": 4.325495719909668, | |
| "learning_rate": 4.234411996842936e-05, | |
| "loss": 0.9275, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.24147727272727273, | |
| "grad_norm": 23.056415557861328, | |
| "learning_rate": 4.2146803472770324e-05, | |
| "loss": 1.1907, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.2450284090909091, | |
| "grad_norm": 11.812889099121094, | |
| "learning_rate": 4.194948697711129e-05, | |
| "loss": 1.0329, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.24857954545454544, | |
| "grad_norm": 11.360099792480469, | |
| "learning_rate": 4.1752170481452254e-05, | |
| "loss": 0.8968, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "eval_accuracy": 0.6878216123499142, | |
| "eval_f1": 0.6885242103977541, | |
| "eval_loss": 0.8689462542533875, | |
| "eval_precision": 0.742302721464857, | |
| "eval_recall": 0.688613815159033, | |
| "eval_runtime": 653.0033, | |
| "eval_samples_per_second": 0.893, | |
| "eval_steps_per_second": 0.224, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 1.0021306818181819, | |
| "grad_norm": 5.516427040100098, | |
| "learning_rate": 4.1554853985793216e-05, | |
| "loss": 0.9278, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.0056818181818181, | |
| "grad_norm": 8.10287094116211, | |
| "learning_rate": 4.135753749013418e-05, | |
| "loss": 0.8833, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.0092329545454546, | |
| "grad_norm": 21.60080337524414, | |
| "learning_rate": 4.116022099447514e-05, | |
| "loss": 0.6148, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.0127840909090908, | |
| "grad_norm": 17.968292236328125, | |
| "learning_rate": 4.09629044988161e-05, | |
| "loss": 0.8788, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.0163352272727273, | |
| "grad_norm": 8.538472175598145, | |
| "learning_rate": 4.076558800315707e-05, | |
| "loss": 0.778, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.0198863636363635, | |
| "grad_norm": 4.343443870544434, | |
| "learning_rate": 4.056827150749803e-05, | |
| "loss": 0.5165, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.0234375, | |
| "grad_norm": 6.1510009765625, | |
| "learning_rate": 4.037095501183899e-05, | |
| "loss": 0.8024, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.0269886363636365, | |
| "grad_norm": 23.44464111328125, | |
| "learning_rate": 4.0173638516179955e-05, | |
| "loss": 1.0074, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.0305397727272727, | |
| "grad_norm": 16.371910095214844, | |
| "learning_rate": 3.997632202052092e-05, | |
| "loss": 0.9794, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.0340909090909092, | |
| "grad_norm": 8.532549858093262, | |
| "learning_rate": 3.977900552486188e-05, | |
| "loss": 0.8435, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.0376420454545454, | |
| "grad_norm": 15.13945484161377, | |
| "learning_rate": 3.958168902920285e-05, | |
| "loss": 0.9073, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.0411931818181819, | |
| "grad_norm": 12.980558395385742, | |
| "learning_rate": 3.938437253354381e-05, | |
| "loss": 0.8077, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.0447443181818181, | |
| "grad_norm": 7.141788005828857, | |
| "learning_rate": 3.9187056037884764e-05, | |
| "loss": 0.9853, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.0482954545454546, | |
| "grad_norm": 5.774048328399658, | |
| "learning_rate": 3.898973954222573e-05, | |
| "loss": 0.5617, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.0518465909090908, | |
| "grad_norm": 30.009641647338867, | |
| "learning_rate": 3.8792423046566694e-05, | |
| "loss": 0.8913, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.0553977272727273, | |
| "grad_norm": 11.07997989654541, | |
| "learning_rate": 3.8595106550907656e-05, | |
| "loss": 0.6674, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.0589488636363635, | |
| "grad_norm": 13.393827438354492, | |
| "learning_rate": 3.8397790055248625e-05, | |
| "loss": 0.5832, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.0625, | |
| "grad_norm": 10.827837944030762, | |
| "learning_rate": 3.8200473559589587e-05, | |
| "loss": 0.6509, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.0660511363636365, | |
| "grad_norm": 27.103227615356445, | |
| "learning_rate": 3.800315706393054e-05, | |
| "loss": 0.7135, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.0696022727272727, | |
| "grad_norm": 16.949966430664062, | |
| "learning_rate": 3.780584056827151e-05, | |
| "loss": 1.0985, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.0731534090909092, | |
| "grad_norm": 11.158037185668945, | |
| "learning_rate": 3.760852407261247e-05, | |
| "loss": 0.7335, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.0767045454545454, | |
| "grad_norm": 26.55095863342285, | |
| "learning_rate": 3.7411207576953434e-05, | |
| "loss": 0.7143, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.0802556818181819, | |
| "grad_norm": 7.920712947845459, | |
| "learning_rate": 3.72138910812944e-05, | |
| "loss": 0.8197, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.0838068181818181, | |
| "grad_norm": 2.0704779624938965, | |
| "learning_rate": 3.7016574585635364e-05, | |
| "loss": 0.7739, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.0873579545454546, | |
| "grad_norm": 29.551998138427734, | |
| "learning_rate": 3.681925808997632e-05, | |
| "loss": 0.9224, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.0909090909090908, | |
| "grad_norm": 9.859967231750488, | |
| "learning_rate": 3.662194159431729e-05, | |
| "loss": 0.8151, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.0944602272727273, | |
| "grad_norm": 10.744651794433594, | |
| "learning_rate": 3.642462509865825e-05, | |
| "loss": 0.4832, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.0980113636363635, | |
| "grad_norm": 3.085749864578247, | |
| "learning_rate": 3.622730860299921e-05, | |
| "loss": 0.5775, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.1015625, | |
| "grad_norm": 11.848700523376465, | |
| "learning_rate": 3.602999210734018e-05, | |
| "loss": 0.4641, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.1051136363636365, | |
| "grad_norm": 5.603400230407715, | |
| "learning_rate": 3.583267561168114e-05, | |
| "loss": 0.5069, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.1086647727272727, | |
| "grad_norm": 8.615111351013184, | |
| "learning_rate": 3.5635359116022096e-05, | |
| "loss": 0.5031, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.1122159090909092, | |
| "grad_norm": 23.804479598999023, | |
| "learning_rate": 3.5438042620363065e-05, | |
| "loss": 0.7348, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.1157670454545454, | |
| "grad_norm": 1.017351508140564, | |
| "learning_rate": 3.5240726124704027e-05, | |
| "loss": 0.7015, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.1193181818181819, | |
| "grad_norm": 14.092010498046875, | |
| "learning_rate": 3.504340962904499e-05, | |
| "loss": 0.5592, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.1228693181818181, | |
| "grad_norm": 20.209379196166992, | |
| "learning_rate": 3.484609313338596e-05, | |
| "loss": 0.7704, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.1264204545454546, | |
| "grad_norm": 0.6501548290252686, | |
| "learning_rate": 3.464877663772691e-05, | |
| "loss": 0.5026, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.1299715909090908, | |
| "grad_norm": 1.0749989748001099, | |
| "learning_rate": 3.4451460142067874e-05, | |
| "loss": 0.5371, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.1335227272727273, | |
| "grad_norm": 10.37779712677002, | |
| "learning_rate": 3.425414364640884e-05, | |
| "loss": 0.4032, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.1370738636363638, | |
| "grad_norm": 30.61161231994629, | |
| "learning_rate": 3.4056827150749804e-05, | |
| "loss": 0.6612, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.140625, | |
| "grad_norm": 28.0980167388916, | |
| "learning_rate": 3.3859510655090766e-05, | |
| "loss": 0.5713, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.1441761363636362, | |
| "grad_norm": 25.723461151123047, | |
| "learning_rate": 3.3662194159431734e-05, | |
| "loss": 0.6669, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.1477272727272727, | |
| "grad_norm": 7.138387203216553, | |
| "learning_rate": 3.346487766377269e-05, | |
| "loss": 0.5781, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.1512784090909092, | |
| "grad_norm": 27.993236541748047, | |
| "learning_rate": 3.326756116811365e-05, | |
| "loss": 0.6033, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.1548295454545454, | |
| "grad_norm": 3.16493558883667, | |
| "learning_rate": 3.307024467245462e-05, | |
| "loss": 0.787, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.1583806818181819, | |
| "grad_norm": 11.943394660949707, | |
| "learning_rate": 3.287292817679558e-05, | |
| "loss": 0.657, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.1619318181818181, | |
| "grad_norm": 3.7604823112487793, | |
| "learning_rate": 3.267561168113654e-05, | |
| "loss": 0.4096, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.1654829545454546, | |
| "grad_norm": 21.66828727722168, | |
| "learning_rate": 3.247829518547751e-05, | |
| "loss": 0.5554, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.1690340909090908, | |
| "grad_norm": 3.944326162338257, | |
| "learning_rate": 3.2280978689818467e-05, | |
| "loss": 0.8685, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.1725852272727273, | |
| "grad_norm": 4.541408061981201, | |
| "learning_rate": 3.208366219415943e-05, | |
| "loss": 0.651, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.1761363636363638, | |
| "grad_norm": 16.059410095214844, | |
| "learning_rate": 3.18863456985004e-05, | |
| "loss": 0.5313, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.1796875, | |
| "grad_norm": 29.15842056274414, | |
| "learning_rate": 3.168902920284136e-05, | |
| "loss": 0.6873, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 1.1832386363636362, | |
| "grad_norm": 27.368764877319336, | |
| "learning_rate": 3.149171270718232e-05, | |
| "loss": 0.7953, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 1.1867897727272727, | |
| "grad_norm": 12.318428039550781, | |
| "learning_rate": 3.129439621152329e-05, | |
| "loss": 0.316, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.1903409090909092, | |
| "grad_norm": 7.381816864013672, | |
| "learning_rate": 3.1097079715864244e-05, | |
| "loss": 0.5909, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 1.1938920454545454, | |
| "grad_norm": 56.74608612060547, | |
| "learning_rate": 3.0899763220205206e-05, | |
| "loss": 0.5496, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.1974431818181819, | |
| "grad_norm": 5.5123090744018555, | |
| "learning_rate": 3.0702446724546174e-05, | |
| "loss": 0.4415, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.2009943181818181, | |
| "grad_norm": 3.0995726585388184, | |
| "learning_rate": 3.0505130228887136e-05, | |
| "loss": 0.4572, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.2045454545454546, | |
| "grad_norm": 1.3388721942901611, | |
| "learning_rate": 3.03078137332281e-05, | |
| "loss": 0.4261, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 1.2080965909090908, | |
| "grad_norm": 3.432847738265991, | |
| "learning_rate": 3.0110497237569063e-05, | |
| "loss": 0.5804, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.2116477272727273, | |
| "grad_norm": 5.881500244140625, | |
| "learning_rate": 2.9913180741910025e-05, | |
| "loss": 0.3265, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.2151988636363638, | |
| "grad_norm": 27.702701568603516, | |
| "learning_rate": 2.971586424625099e-05, | |
| "loss": 0.8524, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.21875, | |
| "grad_norm": 3.5202479362487793, | |
| "learning_rate": 2.951854775059195e-05, | |
| "loss": 0.7208, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 1.2223011363636362, | |
| "grad_norm": 14.859701156616211, | |
| "learning_rate": 2.9321231254932913e-05, | |
| "loss": 0.6345, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.2258522727272727, | |
| "grad_norm": 2.571312427520752, | |
| "learning_rate": 2.912391475927388e-05, | |
| "loss": 0.3549, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.2294034090909092, | |
| "grad_norm": 8.97351360321045, | |
| "learning_rate": 2.892659826361484e-05, | |
| "loss": 0.507, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.2329545454545454, | |
| "grad_norm": 13.665780067443848, | |
| "learning_rate": 2.8729281767955802e-05, | |
| "loss": 0.4561, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.2365056818181819, | |
| "grad_norm": 13.264731407165527, | |
| "learning_rate": 2.8531965272296767e-05, | |
| "loss": 0.539, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 1.2400568181818181, | |
| "grad_norm": 29.683090209960938, | |
| "learning_rate": 2.833464877663773e-05, | |
| "loss": 0.5715, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.2436079545454546, | |
| "grad_norm": 8.478317260742188, | |
| "learning_rate": 2.813733228097869e-05, | |
| "loss": 0.183, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 1.2471590909090908, | |
| "grad_norm": 28.569978713989258, | |
| "learning_rate": 2.7940015785319656e-05, | |
| "loss": 0.5002, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "eval_accuracy": 0.8542024013722127, | |
| "eval_f1": 0.8530975388761288, | |
| "eval_loss": 0.4373819828033447, | |
| "eval_precision": 0.8718298587182326, | |
| "eval_recall": 0.853515201868564, | |
| "eval_runtime": 695.9457, | |
| "eval_samples_per_second": 0.838, | |
| "eval_steps_per_second": 0.21, | |
| "step": 1408 | |
| }, | |
| { | |
| "epoch": 2.000710227272727, | |
| "grad_norm": 25.39379119873047, | |
| "learning_rate": 2.7742699289660618e-05, | |
| "loss": 0.6607, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 2.0042613636363638, | |
| "grad_norm": 3.422687530517578, | |
| "learning_rate": 2.754538279400158e-05, | |
| "loss": 0.5038, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 2.0078125, | |
| "grad_norm": 1.3007397651672363, | |
| "learning_rate": 2.7348066298342545e-05, | |
| "loss": 0.3052, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 2.0113636363636362, | |
| "grad_norm": 15.327472686767578, | |
| "learning_rate": 2.7150749802683506e-05, | |
| "loss": 0.2933, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 2.014914772727273, | |
| "grad_norm": 23.62551498413086, | |
| "learning_rate": 2.6953433307024468e-05, | |
| "loss": 0.4165, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.018465909090909, | |
| "grad_norm": 8.549898147583008, | |
| "learning_rate": 2.6756116811365433e-05, | |
| "loss": 0.3321, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 2.0220170454545454, | |
| "grad_norm": 25.098161697387695, | |
| "learning_rate": 2.6558800315706395e-05, | |
| "loss": 0.4109, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 2.0255681818181817, | |
| "grad_norm": 9.298599243164062, | |
| "learning_rate": 2.6361483820047357e-05, | |
| "loss": 0.4261, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 2.0291193181818183, | |
| "grad_norm": 66.2896957397461, | |
| "learning_rate": 2.6164167324388322e-05, | |
| "loss": 0.3508, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 2.0326704545454546, | |
| "grad_norm": 1.1833637952804565, | |
| "learning_rate": 2.5966850828729284e-05, | |
| "loss": 0.1169, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.036221590909091, | |
| "grad_norm": 32.40926742553711, | |
| "learning_rate": 2.5769534333070246e-05, | |
| "loss": 0.7922, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 2.039772727272727, | |
| "grad_norm": 31.9284610748291, | |
| "learning_rate": 2.557221783741121e-05, | |
| "loss": 0.3044, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 2.0433238636363638, | |
| "grad_norm": 0.32966673374176025, | |
| "learning_rate": 2.5374901341752172e-05, | |
| "loss": 0.5598, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 2.046875, | |
| "grad_norm": 3.2222678661346436, | |
| "learning_rate": 2.5177584846093134e-05, | |
| "loss": 0.5305, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 2.0504261363636362, | |
| "grad_norm": 5.157206058502197, | |
| "learning_rate": 2.4980268350434096e-05, | |
| "loss": 0.3806, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.053977272727273, | |
| "grad_norm": 29.749786376953125, | |
| "learning_rate": 2.478295185477506e-05, | |
| "loss": 0.6572, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 2.057528409090909, | |
| "grad_norm": 5.4849534034729, | |
| "learning_rate": 2.4585635359116023e-05, | |
| "loss": 0.2828, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 2.0610795454545454, | |
| "grad_norm": 1.8682303428649902, | |
| "learning_rate": 2.4388318863456985e-05, | |
| "loss": 0.275, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 2.0646306818181817, | |
| "grad_norm": 31.073657989501953, | |
| "learning_rate": 2.419100236779795e-05, | |
| "loss": 0.2635, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 2.0681818181818183, | |
| "grad_norm": 2.873807668685913, | |
| "learning_rate": 2.399368587213891e-05, | |
| "loss": 0.2189, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.0717329545454546, | |
| "grad_norm": 6.81038761138916, | |
| "learning_rate": 2.3796369376479873e-05, | |
| "loss": 0.2335, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 2.075284090909091, | |
| "grad_norm": 5.240432262420654, | |
| "learning_rate": 2.359905288082084e-05, | |
| "loss": 0.513, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 2.078835227272727, | |
| "grad_norm": 2.8783762454986572, | |
| "learning_rate": 2.34017363851618e-05, | |
| "loss": 0.2719, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 2.0823863636363638, | |
| "grad_norm": 4.599855422973633, | |
| "learning_rate": 2.3204419889502762e-05, | |
| "loss": 0.2012, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 2.0859375, | |
| "grad_norm": 1.6575523614883423, | |
| "learning_rate": 2.3007103393843727e-05, | |
| "loss": 0.1299, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.0894886363636362, | |
| "grad_norm": 0.3831326961517334, | |
| "learning_rate": 2.280978689818469e-05, | |
| "loss": 0.1766, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 2.093039772727273, | |
| "grad_norm": 1.1560420989990234, | |
| "learning_rate": 2.261247040252565e-05, | |
| "loss": 0.3844, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 2.096590909090909, | |
| "grad_norm": 1.8366397619247437, | |
| "learning_rate": 2.2415153906866616e-05, | |
| "loss": 0.3113, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 2.1001420454545454, | |
| "grad_norm": 10.303657531738281, | |
| "learning_rate": 2.2217837411207578e-05, | |
| "loss": 0.1612, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 2.1036931818181817, | |
| "grad_norm": 35.21874237060547, | |
| "learning_rate": 2.202052091554854e-05, | |
| "loss": 0.454, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.1072443181818183, | |
| "grad_norm": 4.324901580810547, | |
| "learning_rate": 2.1823204419889505e-05, | |
| "loss": 0.5713, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 2.1107954545454546, | |
| "grad_norm": 4.857918739318848, | |
| "learning_rate": 2.1625887924230466e-05, | |
| "loss": 0.4748, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 2.114346590909091, | |
| "grad_norm": 0.15146464109420776, | |
| "learning_rate": 2.1428571428571428e-05, | |
| "loss": 0.3261, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 2.117897727272727, | |
| "grad_norm": 19.752939224243164, | |
| "learning_rate": 2.1231254932912393e-05, | |
| "loss": 0.2406, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 2.1214488636363638, | |
| "grad_norm": 18.78827476501465, | |
| "learning_rate": 2.1033938437253355e-05, | |
| "loss": 0.4846, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.125, | |
| "grad_norm": 0.27601099014282227, | |
| "learning_rate": 2.0836621941594317e-05, | |
| "loss": 0.2026, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 2.1285511363636362, | |
| "grad_norm": 0.5743858218193054, | |
| "learning_rate": 2.0639305445935282e-05, | |
| "loss": 0.2579, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 2.132102272727273, | |
| "grad_norm": 0.10733279585838318, | |
| "learning_rate": 2.0441988950276244e-05, | |
| "loss": 0.2348, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 2.135653409090909, | |
| "grad_norm": 2.2115983963012695, | |
| "learning_rate": 2.0244672454617206e-05, | |
| "loss": 0.2632, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 2.1392045454545454, | |
| "grad_norm": 37.26658248901367, | |
| "learning_rate": 2.004735595895817e-05, | |
| "loss": 0.4384, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.1427556818181817, | |
| "grad_norm": 10.85006332397461, | |
| "learning_rate": 1.9850039463299132e-05, | |
| "loss": 0.2217, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 2.1463068181818183, | |
| "grad_norm": 4.717857360839844, | |
| "learning_rate": 1.9652722967640098e-05, | |
| "loss": 0.2748, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 2.1498579545454546, | |
| "grad_norm": 2.3380446434020996, | |
| "learning_rate": 1.945540647198106e-05, | |
| "loss": 0.1738, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 2.153409090909091, | |
| "grad_norm": 1.6607507467269897, | |
| "learning_rate": 1.925808997632202e-05, | |
| "loss": 0.2641, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 2.156960227272727, | |
| "grad_norm": 12.202836036682129, | |
| "learning_rate": 1.9060773480662986e-05, | |
| "loss": 0.2995, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.1605113636363638, | |
| "grad_norm": 20.316972732543945, | |
| "learning_rate": 1.8863456985003948e-05, | |
| "loss": 0.1573, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 2.1640625, | |
| "grad_norm": 34.030296325683594, | |
| "learning_rate": 1.866614048934491e-05, | |
| "loss": 0.266, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 2.1676136363636362, | |
| "grad_norm": 4.90464973449707, | |
| "learning_rate": 1.8468823993685875e-05, | |
| "loss": 0.1064, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 2.171164772727273, | |
| "grad_norm": 38.46381378173828, | |
| "learning_rate": 1.8271507498026837e-05, | |
| "loss": 0.2679, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 2.174715909090909, | |
| "grad_norm": 52.19645690917969, | |
| "learning_rate": 1.80741910023678e-05, | |
| "loss": 0.166, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.1782670454545454, | |
| "grad_norm": 15.714051246643066, | |
| "learning_rate": 1.7876874506708764e-05, | |
| "loss": 0.4399, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 2.1818181818181817, | |
| "grad_norm": 50.66694641113281, | |
| "learning_rate": 1.7679558011049725e-05, | |
| "loss": 0.1816, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 2.1853693181818183, | |
| "grad_norm": 4.8143205642700195, | |
| "learning_rate": 1.7482241515390687e-05, | |
| "loss": 0.5087, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 2.1889204545454546, | |
| "grad_norm": 1.34380042552948, | |
| "learning_rate": 1.7284925019731652e-05, | |
| "loss": 0.4257, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 2.192471590909091, | |
| "grad_norm": 6.62140417098999, | |
| "learning_rate": 1.7087608524072614e-05, | |
| "loss": 0.1767, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.196022727272727, | |
| "grad_norm": 34.63706588745117, | |
| "learning_rate": 1.6890292028413576e-05, | |
| "loss": 0.3301, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 2.1995738636363638, | |
| "grad_norm": 3.23504638671875, | |
| "learning_rate": 1.669297553275454e-05, | |
| "loss": 0.2174, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 2.203125, | |
| "grad_norm": 16.90635871887207, | |
| "learning_rate": 1.64956590370955e-05, | |
| "loss": 0.4572, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 2.2066761363636362, | |
| "grad_norm": 20.676467895507812, | |
| "learning_rate": 1.6298342541436465e-05, | |
| "loss": 0.3199, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 2.210227272727273, | |
| "grad_norm": 0.31672224402427673, | |
| "learning_rate": 1.610102604577743e-05, | |
| "loss": 0.2, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.213778409090909, | |
| "grad_norm": 8.79170036315918, | |
| "learning_rate": 1.5903709550118388e-05, | |
| "loss": 0.1675, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 2.2173295454545454, | |
| "grad_norm": 2.0761942863464355, | |
| "learning_rate": 1.5706393054459353e-05, | |
| "loss": 0.2614, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 2.2208806818181817, | |
| "grad_norm": 0.06360641866922379, | |
| "learning_rate": 1.550907655880032e-05, | |
| "loss": 0.2335, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 2.2244318181818183, | |
| "grad_norm": 18.36467170715332, | |
| "learning_rate": 1.5311760063141277e-05, | |
| "loss": 0.5066, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 2.2279829545454546, | |
| "grad_norm": 9.642960548400879, | |
| "learning_rate": 1.5114443567482242e-05, | |
| "loss": 0.2035, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.231534090909091, | |
| "grad_norm": 6.050065040588379, | |
| "learning_rate": 1.4917127071823205e-05, | |
| "loss": 0.1467, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 2.235085227272727, | |
| "grad_norm": 0.06556593626737595, | |
| "learning_rate": 1.4719810576164167e-05, | |
| "loss": 0.1243, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 2.2386363636363638, | |
| "grad_norm": 0.29696065187454224, | |
| "learning_rate": 1.452249408050513e-05, | |
| "loss": 0.2465, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 2.2421875, | |
| "grad_norm": 1.2069284915924072, | |
| "learning_rate": 1.4325177584846094e-05, | |
| "loss": 0.191, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 2.2457386363636362, | |
| "grad_norm": 2.228632688522339, | |
| "learning_rate": 1.4127861089187056e-05, | |
| "loss": 0.1718, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.249289772727273, | |
| "grad_norm": 1.17184317111969, | |
| "learning_rate": 1.393054459352802e-05, | |
| "loss": 0.3627, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "eval_accuracy": 0.9622641509433962, | |
| "eval_f1": 0.9617697305162668, | |
| "eval_loss": 0.1109052374958992, | |
| "eval_precision": 0.9647490495721144, | |
| "eval_recall": 0.9614389495077621, | |
| "eval_runtime": 696.0893, | |
| "eval_samples_per_second": 0.838, | |
| "eval_steps_per_second": 0.21, | |
| "step": 2112 | |
| }, | |
| { | |
| "epoch": 3.002840909090909, | |
| "grad_norm": 0.03763706982135773, | |
| "learning_rate": 1.3733228097868983e-05, | |
| "loss": 0.0171, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 3.0063920454545454, | |
| "grad_norm": 1.5585871934890747, | |
| "learning_rate": 1.3535911602209945e-05, | |
| "loss": 0.2023, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 3.0099431818181817, | |
| "grad_norm": 20.643203735351562, | |
| "learning_rate": 1.3338595106550908e-05, | |
| "loss": 0.1212, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 3.0134943181818183, | |
| "grad_norm": 0.19967862963676453, | |
| "learning_rate": 1.3141278610891871e-05, | |
| "loss": 0.0201, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.0170454545454546, | |
| "grad_norm": 3.6144466400146484, | |
| "learning_rate": 1.2943962115232833e-05, | |
| "loss": 0.1455, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 3.020596590909091, | |
| "grad_norm": 7.9842610359191895, | |
| "learning_rate": 1.2746645619573797e-05, | |
| "loss": 0.042, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 3.024147727272727, | |
| "grad_norm": 3.346548080444336, | |
| "learning_rate": 1.254932912391476e-05, | |
| "loss": 0.0365, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 3.0276988636363638, | |
| "grad_norm": 0.10210000723600388, | |
| "learning_rate": 1.2352012628255722e-05, | |
| "loss": 0.118, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 3.03125, | |
| "grad_norm": 2.003911018371582, | |
| "learning_rate": 1.2154696132596685e-05, | |
| "loss": 0.0461, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.0348011363636362, | |
| "grad_norm": 0.06448382884263992, | |
| "learning_rate": 1.1957379636937649e-05, | |
| "loss": 0.1276, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 3.038352272727273, | |
| "grad_norm": 59.72112274169922, | |
| "learning_rate": 1.176006314127861e-05, | |
| "loss": 0.1744, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 3.041903409090909, | |
| "grad_norm": 26.68014907836914, | |
| "learning_rate": 1.1562746645619574e-05, | |
| "loss": 0.0988, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 3.0454545454545454, | |
| "grad_norm": 1.2907896041870117, | |
| "learning_rate": 1.1365430149960538e-05, | |
| "loss": 0.0432, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 3.0490056818181817, | |
| "grad_norm": 2.807907819747925, | |
| "learning_rate": 1.11681136543015e-05, | |
| "loss": 0.1671, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.0525568181818183, | |
| "grad_norm": 0.22834616899490356, | |
| "learning_rate": 1.0970797158642463e-05, | |
| "loss": 0.0643, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 3.0561079545454546, | |
| "grad_norm": 0.23886850476264954, | |
| "learning_rate": 1.0773480662983426e-05, | |
| "loss": 0.1166, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 3.059659090909091, | |
| "grad_norm": 2.190918445587158, | |
| "learning_rate": 1.0576164167324388e-05, | |
| "loss": 0.0236, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 3.063210227272727, | |
| "grad_norm": 1.075537919998169, | |
| "learning_rate": 1.0378847671665353e-05, | |
| "loss": 0.1185, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 3.0667613636363638, | |
| "grad_norm": 34.861881256103516, | |
| "learning_rate": 1.0181531176006315e-05, | |
| "loss": 0.1225, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 3.0703125, | |
| "grad_norm": 0.2916108965873718, | |
| "learning_rate": 9.984214680347277e-06, | |
| "loss": 0.0877, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 3.0738636363636362, | |
| "grad_norm": 24.542476654052734, | |
| "learning_rate": 9.786898184688242e-06, | |
| "loss": 0.1007, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 3.077414772727273, | |
| "grad_norm": 1.0566986799240112, | |
| "learning_rate": 9.589581689029204e-06, | |
| "loss": 0.0563, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 3.080965909090909, | |
| "grad_norm": 20.17405128479004, | |
| "learning_rate": 9.392265193370165e-06, | |
| "loss": 0.0374, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 3.0845170454545454, | |
| "grad_norm": 0.07401897758245468, | |
| "learning_rate": 9.194948697711129e-06, | |
| "loss": 0.0495, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 3.0880681818181817, | |
| "grad_norm": 4.210412502288818, | |
| "learning_rate": 8.997632202052092e-06, | |
| "loss": 0.1265, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 3.0916193181818183, | |
| "grad_norm": 27.83361053466797, | |
| "learning_rate": 8.800315706393054e-06, | |
| "loss": 0.2213, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 3.0951704545454546, | |
| "grad_norm": 0.04297183081507683, | |
| "learning_rate": 8.602999210734018e-06, | |
| "loss": 0.1469, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 3.098721590909091, | |
| "grad_norm": 0.7070327401161194, | |
| "learning_rate": 8.405682715074981e-06, | |
| "loss": 0.0219, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 3.102272727272727, | |
| "grad_norm": 0.22129130363464355, | |
| "learning_rate": 8.208366219415943e-06, | |
| "loss": 0.0115, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.1058238636363638, | |
| "grad_norm": 0.8595547080039978, | |
| "learning_rate": 8.011049723756906e-06, | |
| "loss": 0.0124, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 3.109375, | |
| "grad_norm": 0.13839711248874664, | |
| "learning_rate": 7.81373322809787e-06, | |
| "loss": 0.0865, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 3.1129261363636362, | |
| "grad_norm": 0.04325387626886368, | |
| "learning_rate": 7.6164167324388314e-06, | |
| "loss": 0.0793, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 3.116477272727273, | |
| "grad_norm": 3.90751051902771, | |
| "learning_rate": 7.419100236779796e-06, | |
| "loss": 0.1025, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 3.120028409090909, | |
| "grad_norm": 0.03920091316103935, | |
| "learning_rate": 7.2217837411207575e-06, | |
| "loss": 0.2589, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 3.1235795454545454, | |
| "grad_norm": 0.2701418995857239, | |
| "learning_rate": 7.02446724546172e-06, | |
| "loss": 0.0066, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 3.1271306818181817, | |
| "grad_norm": 0.3685694634914398, | |
| "learning_rate": 6.827150749802684e-06, | |
| "loss": 0.0537, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 3.1306818181818183, | |
| "grad_norm": 0.03167586028575897, | |
| "learning_rate": 6.629834254143646e-06, | |
| "loss": 0.2329, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 3.1342329545454546, | |
| "grad_norm": 24.94156265258789, | |
| "learning_rate": 6.4325177584846105e-06, | |
| "loss": 0.113, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 3.137784090909091, | |
| "grad_norm": 2.3878488540649414, | |
| "learning_rate": 6.235201262825572e-06, | |
| "loss": 0.0289, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 3.141335227272727, | |
| "grad_norm": 0.1500893086194992, | |
| "learning_rate": 6.037884767166536e-06, | |
| "loss": 0.0164, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 3.1448863636363638, | |
| "grad_norm": 1.0166605710983276, | |
| "learning_rate": 5.840568271507498e-06, | |
| "loss": 0.0665, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 3.1484375, | |
| "grad_norm": 8.010841369628906, | |
| "learning_rate": 5.643251775848461e-06, | |
| "loss": 0.2883, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 3.1519886363636362, | |
| "grad_norm": 11.449795722961426, | |
| "learning_rate": 5.445935280189424e-06, | |
| "loss": 0.0892, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 3.155539772727273, | |
| "grad_norm": 41.279239654541016, | |
| "learning_rate": 5.248618784530387e-06, | |
| "loss": 0.0529, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 3.159090909090909, | |
| "grad_norm": 0.06051742658019066, | |
| "learning_rate": 5.05130228887135e-06, | |
| "loss": 0.1401, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 3.1626420454545454, | |
| "grad_norm": 0.08330044150352478, | |
| "learning_rate": 4.853985793212313e-06, | |
| "loss": 0.0411, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 3.1661931818181817, | |
| "grad_norm": 44.56931686401367, | |
| "learning_rate": 4.656669297553276e-06, | |
| "loss": 0.0657, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 3.1697443181818183, | |
| "grad_norm": 1.3760157823562622, | |
| "learning_rate": 4.459352801894238e-06, | |
| "loss": 0.1188, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 3.1732954545454546, | |
| "grad_norm": 0.1809810996055603, | |
| "learning_rate": 4.262036306235202e-06, | |
| "loss": 0.0305, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 3.176846590909091, | |
| "grad_norm": 1.571562647819519, | |
| "learning_rate": 4.064719810576164e-06, | |
| "loss": 0.0448, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 3.180397727272727, | |
| "grad_norm": 0.04679562523961067, | |
| "learning_rate": 3.867403314917127e-06, | |
| "loss": 0.0876, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 3.1839488636363638, | |
| "grad_norm": 0.40420079231262207, | |
| "learning_rate": 3.67008681925809e-06, | |
| "loss": 0.0399, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 3.1875, | |
| "grad_norm": 0.8966354727745056, | |
| "learning_rate": 3.472770323599053e-06, | |
| "loss": 0.0747, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 3.1910511363636362, | |
| "grad_norm": 0.499515563249588, | |
| "learning_rate": 3.2754538279400157e-06, | |
| "loss": 0.1019, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 3.194602272727273, | |
| "grad_norm": 0.09785713255405426, | |
| "learning_rate": 3.0781373322809787e-06, | |
| "loss": 0.032, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 3.198153409090909, | |
| "grad_norm": 0.037965867668390274, | |
| "learning_rate": 2.8808208366219414e-06, | |
| "loss": 0.0693, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 3.2017045454545454, | |
| "grad_norm": 0.5521709322929382, | |
| "learning_rate": 2.683504340962905e-06, | |
| "loss": 0.0761, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 3.2052556818181817, | |
| "grad_norm": 0.09314560890197754, | |
| "learning_rate": 2.4861878453038674e-06, | |
| "loss": 0.1744, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 3.2088068181818183, | |
| "grad_norm": 12.886544227600098, | |
| "learning_rate": 2.2888713496448305e-06, | |
| "loss": 0.1549, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 3.2123579545454546, | |
| "grad_norm": 0.18088915944099426, | |
| "learning_rate": 2.091554853985793e-06, | |
| "loss": 0.186, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 3.215909090909091, | |
| "grad_norm": 0.18304170668125153, | |
| "learning_rate": 1.8942383583267563e-06, | |
| "loss": 0.0137, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 3.219460227272727, | |
| "grad_norm": 5.272530555725098, | |
| "learning_rate": 1.6969218626677192e-06, | |
| "loss": 0.0369, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 3.2230113636363638, | |
| "grad_norm": 0.17552484571933746, | |
| "learning_rate": 1.499605367008682e-06, | |
| "loss": 0.0675, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 3.2265625, | |
| "grad_norm": 0.28992605209350586, | |
| "learning_rate": 1.3022888713496448e-06, | |
| "loss": 0.0744, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 3.2301136363636362, | |
| "grad_norm": 0.17767734825611115, | |
| "learning_rate": 1.1049723756906078e-06, | |
| "loss": 0.1551, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 3.233664772727273, | |
| "grad_norm": 0.43689748644828796, | |
| "learning_rate": 9.076558800315707e-07, | |
| "loss": 0.1754, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 3.237215909090909, | |
| "grad_norm": 0.956597089767456, | |
| "learning_rate": 7.103393843725336e-07, | |
| "loss": 0.0069, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 3.2407670454545454, | |
| "grad_norm": 25.514036178588867, | |
| "learning_rate": 5.130228887134964e-07, | |
| "loss": 0.0997, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 3.2443181818181817, | |
| "grad_norm": 3.5862650871276855, | |
| "learning_rate": 3.1570639305445935e-07, | |
| "loss": 0.1085, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 3.2478693181818183, | |
| "grad_norm": 0.035997312515974045, | |
| "learning_rate": 1.1838989739542227e-07, | |
| "loss": 0.0289, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "eval_accuracy": 0.9879931389365352, | |
| "eval_f1": 0.9879774481091409, | |
| "eval_loss": 0.037413839250802994, | |
| "eval_precision": 0.9880848996629442, | |
| "eval_recall": 0.9879997453526865, | |
| "eval_runtime": 692.1711, | |
| "eval_samples_per_second": 0.842, | |
| "eval_steps_per_second": 0.211, | |
| "step": 2816 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "step": 2816, | |
| "total_flos": 1.4036798084625727e+19, | |
| "train_loss": 0.6578278151044453, | |
| "train_runtime": 18231.9061, | |
| "train_samples_per_second": 0.618, | |
| "train_steps_per_second": 0.154 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "eval_accuracy": 0.9558620689655173, | |
| "eval_f1": 0.9562337987051189, | |
| "eval_loss": 0.14939993619918823, | |
| "eval_precision": 0.9567608190198278, | |
| "eval_recall": 0.956465718392097, | |
| "eval_runtime": 911.348, | |
| "eval_samples_per_second": 0.796, | |
| "eval_steps_per_second": 0.2, | |
| "step": 2816 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "eval_accuracy": 0.9558620689655173, | |
| "eval_f1": 0.9562337987051189, | |
| "eval_loss": 0.14939993619918823, | |
| "eval_precision": 0.9567608190198278, | |
| "eval_recall": 0.956465718392097, | |
| "eval_runtime": 930.9063, | |
| "eval_samples_per_second": 0.779, | |
| "eval_steps_per_second": 0.196, | |
| "step": 2816 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 2816, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 500, | |
| "total_flos": 1.4036798084625727e+19, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |