| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "eval_steps": 200, | |
| "global_step": 2064, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.014534883720930232, | |
| "grad_norm": 5.562132358551025, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 2.7844, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.029069767441860465, | |
| "grad_norm": 6.067043781280518, | |
| "learning_rate": 8.000000000000001e-07, | |
| "loss": 2.7725, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0436046511627907, | |
| "grad_norm": 6.281836986541748, | |
| "learning_rate": 1.2000000000000002e-06, | |
| "loss": 2.7484, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.05813953488372093, | |
| "grad_norm": 5.292967319488525, | |
| "learning_rate": 1.6000000000000001e-06, | |
| "loss": 2.6898, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.07267441860465117, | |
| "grad_norm": 6.3012261390686035, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 2.6246, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0872093023255814, | |
| "grad_norm": 5.181797981262207, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 2.593, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.10174418604651163, | |
| "grad_norm": 6.388678073883057, | |
| "learning_rate": 2.8000000000000003e-06, | |
| "loss": 2.5312, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.11627906976744186, | |
| "grad_norm": 5.643568992614746, | |
| "learning_rate": 3.2000000000000003e-06, | |
| "loss": 2.4826, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1308139534883721, | |
| "grad_norm": 5.568506240844727, | |
| "learning_rate": 3.6000000000000003e-06, | |
| "loss": 2.4131, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.14534883720930233, | |
| "grad_norm": 5.19083833694458, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 2.415, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.15988372093023256, | |
| "grad_norm": 5.537081241607666, | |
| "learning_rate": 4.4e-06, | |
| "loss": 2.3401, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.1744186046511628, | |
| "grad_norm": 5.086550235748291, | |
| "learning_rate": 4.800000000000001e-06, | |
| "loss": 2.3943, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.18895348837209303, | |
| "grad_norm": 7.545917987823486, | |
| "learning_rate": 5.2e-06, | |
| "loss": 2.2582, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.20348837209302326, | |
| "grad_norm": 7.227110862731934, | |
| "learning_rate": 5.600000000000001e-06, | |
| "loss": 2.1587, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.2180232558139535, | |
| "grad_norm": 6.003934860229492, | |
| "learning_rate": 6e-06, | |
| "loss": 2.0893, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.23255813953488372, | |
| "grad_norm": 6.239010810852051, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 2.0743, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.24709302325581395, | |
| "grad_norm": 7.215450763702393, | |
| "learning_rate": 6.800000000000001e-06, | |
| "loss": 2.0095, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.2616279069767442, | |
| "grad_norm": 5.769580364227295, | |
| "learning_rate": 7.2000000000000005e-06, | |
| "loss": 2.0063, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.2761627906976744, | |
| "grad_norm": 7.701093673706055, | |
| "learning_rate": 7.600000000000001e-06, | |
| "loss": 1.9538, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.29069767441860467, | |
| "grad_norm": 8.804961204528809, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 1.9761, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.29069767441860467, | |
| "eval_accuracy": 0.64625, | |
| "eval_accuracy_label_arts, culture, entertainment and media": 0.3157894736842105, | |
| "eval_accuracy_label_conflict, war and peace": 0.8315467075038285, | |
| "eval_accuracy_label_crime, law and justice": 0.7628865979381443, | |
| "eval_accuracy_label_disaster, accident, and emergency incident": 0.7054794520547946, | |
| "eval_accuracy_label_economy, business, and finance": 0.5436893203883495, | |
| "eval_accuracy_label_environment": 0.0, | |
| "eval_accuracy_label_health": 0.5, | |
| "eval_accuracy_label_human interest": 0.0, | |
| "eval_accuracy_label_labour": 0.0, | |
| "eval_accuracy_label_lifestyle and leisure": 0.3333333333333333, | |
| "eval_accuracy_label_politics": 0.4843205574912892, | |
| "eval_accuracy_label_religion": 0.0, | |
| "eval_accuracy_label_science and technology": 0.08333333333333333, | |
| "eval_accuracy_label_society": 0.0, | |
| "eval_accuracy_label_sport": 0.9615384615384616, | |
| "eval_accuracy_label_weather": 0.0, | |
| "eval_f1": 0.6164394441800294, | |
| "eval_loss": 1.4046330451965332, | |
| "eval_precision": 0.6057417123934041, | |
| "eval_recall": 0.64625, | |
| "eval_runtime": 12.4531, | |
| "eval_samples_per_second": 128.482, | |
| "eval_steps_per_second": 8.03, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.30523255813953487, | |
| "grad_norm": 7.173561096191406, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 1.7684, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.31976744186046513, | |
| "grad_norm": 8.086793899536133, | |
| "learning_rate": 8.8e-06, | |
| "loss": 1.7253, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.33430232558139533, | |
| "grad_norm": 8.559186935424805, | |
| "learning_rate": 9.200000000000002e-06, | |
| "loss": 1.795, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.3488372093023256, | |
| "grad_norm": 6.066705226898193, | |
| "learning_rate": 9.600000000000001e-06, | |
| "loss": 1.6181, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.3633720930232558, | |
| "grad_norm": 13.028596878051758, | |
| "learning_rate": 1e-05, | |
| "loss": 1.6328, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.37790697674418605, | |
| "grad_norm": 9.958759307861328, | |
| "learning_rate": 1.04e-05, | |
| "loss": 1.6997, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.39244186046511625, | |
| "grad_norm": 11.519359588623047, | |
| "learning_rate": 1.0800000000000002e-05, | |
| "loss": 1.6046, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.4069767441860465, | |
| "grad_norm": 8.298735618591309, | |
| "learning_rate": 1.1200000000000001e-05, | |
| "loss": 1.4126, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.42151162790697677, | |
| "grad_norm": 10.555624961853027, | |
| "learning_rate": 1.16e-05, | |
| "loss": 1.445, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.436046511627907, | |
| "grad_norm": 8.33277702331543, | |
| "learning_rate": 1.2e-05, | |
| "loss": 1.4088, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.45058139534883723, | |
| "grad_norm": 10.028512001037598, | |
| "learning_rate": 1.2400000000000002e-05, | |
| "loss": 1.3765, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.46511627906976744, | |
| "grad_norm": 11.585610389709473, | |
| "learning_rate": 1.2800000000000001e-05, | |
| "loss": 1.409, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.4796511627906977, | |
| "grad_norm": 13.202520370483398, | |
| "learning_rate": 1.3200000000000002e-05, | |
| "loss": 1.3285, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.4941860465116279, | |
| "grad_norm": 8.829900741577148, | |
| "learning_rate": 1.3600000000000002e-05, | |
| "loss": 1.3261, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.5087209302325582, | |
| "grad_norm": 11.259300231933594, | |
| "learning_rate": 1.4e-05, | |
| "loss": 1.4546, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.5232558139534884, | |
| "grad_norm": 8.918652534484863, | |
| "learning_rate": 1.4400000000000001e-05, | |
| "loss": 1.2455, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.5377906976744186, | |
| "grad_norm": 10.523372650146484, | |
| "learning_rate": 1.48e-05, | |
| "loss": 1.3496, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.5523255813953488, | |
| "grad_norm": 13.649139404296875, | |
| "learning_rate": 1.5200000000000002e-05, | |
| "loss": 1.2688, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.5668604651162791, | |
| "grad_norm": 15.79927921295166, | |
| "learning_rate": 1.5600000000000003e-05, | |
| "loss": 1.22, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.5813953488372093, | |
| "grad_norm": 12.712923049926758, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 1.2153, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.5813953488372093, | |
| "eval_accuracy": 0.689375, | |
| "eval_accuracy_label_arts, culture, entertainment and media": 0.7894736842105263, | |
| "eval_accuracy_label_conflict, war and peace": 0.655436447166922, | |
| "eval_accuracy_label_crime, law and justice": 0.8195876288659794, | |
| "eval_accuracy_label_disaster, accident, and emergency incident": 0.8561643835616438, | |
| "eval_accuracy_label_economy, business, and finance": 0.6407766990291263, | |
| "eval_accuracy_label_environment": 0.2413793103448276, | |
| "eval_accuracy_label_health": 0.8333333333333334, | |
| "eval_accuracy_label_human interest": 0.13636363636363635, | |
| "eval_accuracy_label_labour": 0.0, | |
| "eval_accuracy_label_lifestyle and leisure": 0.6666666666666666, | |
| "eval_accuracy_label_politics": 0.8466898954703833, | |
| "eval_accuracy_label_religion": 0.0, | |
| "eval_accuracy_label_science and technology": 0.375, | |
| "eval_accuracy_label_society": 0.015384615384615385, | |
| "eval_accuracy_label_sport": 0.9615384615384616, | |
| "eval_accuracy_label_weather": 1.0, | |
| "eval_f1": 0.6867853551801604, | |
| "eval_loss": 1.0224822759628296, | |
| "eval_precision": 0.765204058143133, | |
| "eval_recall": 0.689375, | |
| "eval_runtime": 12.5726, | |
| "eval_samples_per_second": 127.261, | |
| "eval_steps_per_second": 7.954, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.5959302325581395, | |
| "grad_norm": 11.235016822814941, | |
| "learning_rate": 1.64e-05, | |
| "loss": 1.2041, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.6104651162790697, | |
| "grad_norm": 10.808572769165039, | |
| "learning_rate": 1.6800000000000002e-05, | |
| "loss": 1.1168, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.625, | |
| "grad_norm": 16.74306869506836, | |
| "learning_rate": 1.72e-05, | |
| "loss": 1.1227, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.6395348837209303, | |
| "grad_norm": 9.389023780822754, | |
| "learning_rate": 1.76e-05, | |
| "loss": 1.1575, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.6540697674418605, | |
| "grad_norm": 15.347270965576172, | |
| "learning_rate": 1.8e-05, | |
| "loss": 0.9979, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.6686046511627907, | |
| "grad_norm": 11.799999237060547, | |
| "learning_rate": 1.8400000000000003e-05, | |
| "loss": 1.0845, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.6831395348837209, | |
| "grad_norm": 12.181020736694336, | |
| "learning_rate": 1.88e-05, | |
| "loss": 1.0833, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.6976744186046512, | |
| "grad_norm": 12.819634437561035, | |
| "learning_rate": 1.9200000000000003e-05, | |
| "loss": 1.137, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.7122093023255814, | |
| "grad_norm": 13.189191818237305, | |
| "learning_rate": 1.9600000000000002e-05, | |
| "loss": 1.0458, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.7267441860465116, | |
| "grad_norm": 9.888482093811035, | |
| "learning_rate": 2e-05, | |
| "loss": 1.0184, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.7412790697674418, | |
| "grad_norm": 9.71278190612793, | |
| "learning_rate": 1.987212276214834e-05, | |
| "loss": 0.98, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.7558139534883721, | |
| "grad_norm": 14.008501052856445, | |
| "learning_rate": 1.9744245524296677e-05, | |
| "loss": 0.9985, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.7703488372093024, | |
| "grad_norm": 16.69378662109375, | |
| "learning_rate": 1.9616368286445014e-05, | |
| "loss": 0.9146, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.7848837209302325, | |
| "grad_norm": 15.569829940795898, | |
| "learning_rate": 1.9488491048593352e-05, | |
| "loss": 0.9717, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.7994186046511628, | |
| "grad_norm": 10.90029239654541, | |
| "learning_rate": 1.936061381074169e-05, | |
| "loss": 1.0832, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.813953488372093, | |
| "grad_norm": 16.751508712768555, | |
| "learning_rate": 1.9232736572890027e-05, | |
| "loss": 0.9345, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.8284883720930233, | |
| "grad_norm": 14.394732475280762, | |
| "learning_rate": 1.9104859335038365e-05, | |
| "loss": 1.0405, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.8430232558139535, | |
| "grad_norm": 11.01040267944336, | |
| "learning_rate": 1.8976982097186702e-05, | |
| "loss": 0.9115, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.8575581395348837, | |
| "grad_norm": 15.872978210449219, | |
| "learning_rate": 1.884910485933504e-05, | |
| "loss": 0.8443, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.872093023255814, | |
| "grad_norm": 19.019338607788086, | |
| "learning_rate": 1.8721227621483378e-05, | |
| "loss": 0.954, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.872093023255814, | |
| "eval_accuracy": 0.723125, | |
| "eval_accuracy_label_arts, culture, entertainment and media": 0.7368421052631579, | |
| "eval_accuracy_label_conflict, war and peace": 0.77947932618683, | |
| "eval_accuracy_label_crime, law and justice": 0.8917525773195877, | |
| "eval_accuracy_label_disaster, accident, and emergency incident": 0.8698630136986302, | |
| "eval_accuracy_label_economy, business, and finance": 0.6213592233009708, | |
| "eval_accuracy_label_environment": 0.3448275862068966, | |
| "eval_accuracy_label_health": 0.8888888888888888, | |
| "eval_accuracy_label_human interest": 0.18181818181818182, | |
| "eval_accuracy_label_labour": 1.0, | |
| "eval_accuracy_label_lifestyle and leisure": 0.5555555555555556, | |
| "eval_accuracy_label_politics": 0.6898954703832753, | |
| "eval_accuracy_label_religion": 0.0, | |
| "eval_accuracy_label_science and technology": 0.25, | |
| "eval_accuracy_label_society": 0.046153846153846156, | |
| "eval_accuracy_label_sport": 0.9615384615384616, | |
| "eval_accuracy_label_weather": 1.0, | |
| "eval_f1": 0.7137964265878213, | |
| "eval_loss": 0.8858376145362854, | |
| "eval_precision": 0.7308873219026191, | |
| "eval_recall": 0.723125, | |
| "eval_runtime": 12.6079, | |
| "eval_samples_per_second": 126.905, | |
| "eval_steps_per_second": 7.932, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.8866279069767442, | |
| "grad_norm": 8.310386657714844, | |
| "learning_rate": 1.8593350383631715e-05, | |
| "loss": 0.8133, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.9011627906976745, | |
| "grad_norm": 12.042872428894043, | |
| "learning_rate": 1.8465473145780053e-05, | |
| "loss": 0.9059, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.9156976744186046, | |
| "grad_norm": 10.83311653137207, | |
| "learning_rate": 1.833759590792839e-05, | |
| "loss": 0.9357, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.9302325581395349, | |
| "grad_norm": 11.619269371032715, | |
| "learning_rate": 1.8209718670076728e-05, | |
| "loss": 0.967, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.9447674418604651, | |
| "grad_norm": 11.699959754943848, | |
| "learning_rate": 1.8081841432225066e-05, | |
| "loss": 0.748, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.9593023255813954, | |
| "grad_norm": 15.047694206237793, | |
| "learning_rate": 1.7953964194373403e-05, | |
| "loss": 0.8928, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.9738372093023255, | |
| "grad_norm": 13.320290565490723, | |
| "learning_rate": 1.782608695652174e-05, | |
| "loss": 0.8786, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.9883720930232558, | |
| "grad_norm": 13.041028022766113, | |
| "learning_rate": 1.769820971867008e-05, | |
| "loss": 0.9478, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.002906976744186, | |
| "grad_norm": 14.132765769958496, | |
| "learning_rate": 1.7570332480818416e-05, | |
| "loss": 0.8231, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.0174418604651163, | |
| "grad_norm": 7.3977837562561035, | |
| "learning_rate": 1.7442455242966754e-05, | |
| "loss": 0.7806, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.0319767441860466, | |
| "grad_norm": 16.572118759155273, | |
| "learning_rate": 1.731457800511509e-05, | |
| "loss": 0.7654, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.0465116279069768, | |
| "grad_norm": 21.040008544921875, | |
| "learning_rate": 1.718670076726343e-05, | |
| "loss": 0.7856, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.0610465116279069, | |
| "grad_norm": 21.33478355407715, | |
| "learning_rate": 1.7058823529411767e-05, | |
| "loss": 0.7323, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.0755813953488371, | |
| "grad_norm": 13.03322696685791, | |
| "learning_rate": 1.6930946291560104e-05, | |
| "loss": 0.7396, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.0901162790697674, | |
| "grad_norm": 19.618709564208984, | |
| "learning_rate": 1.6803069053708442e-05, | |
| "loss": 0.8284, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.1046511627906976, | |
| "grad_norm": 21.11569595336914, | |
| "learning_rate": 1.667519181585678e-05, | |
| "loss": 0.7032, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.119186046511628, | |
| "grad_norm": 13.00526237487793, | |
| "learning_rate": 1.6547314578005117e-05, | |
| "loss": 0.7199, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.1337209302325582, | |
| "grad_norm": 12.57153034210205, | |
| "learning_rate": 1.6419437340153455e-05, | |
| "loss": 0.6688, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.1482558139534884, | |
| "grad_norm": 11.098487854003906, | |
| "learning_rate": 1.6291560102301792e-05, | |
| "loss": 0.6463, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.1627906976744187, | |
| "grad_norm": 12.68520736694336, | |
| "learning_rate": 1.616368286445013e-05, | |
| "loss": 0.6662, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.1627906976744187, | |
| "eval_accuracy": 0.688125, | |
| "eval_accuracy_label_arts, culture, entertainment and media": 0.7894736842105263, | |
| "eval_accuracy_label_conflict, war and peace": 0.6125574272588055, | |
| "eval_accuracy_label_crime, law and justice": 0.845360824742268, | |
| "eval_accuracy_label_disaster, accident, and emergency incident": 0.863013698630137, | |
| "eval_accuracy_label_economy, business, and finance": 0.6504854368932039, | |
| "eval_accuracy_label_environment": 0.4482758620689655, | |
| "eval_accuracy_label_health": 0.7222222222222222, | |
| "eval_accuracy_label_human interest": 0.22727272727272727, | |
| "eval_accuracy_label_labour": 1.0, | |
| "eval_accuracy_label_lifestyle and leisure": 0.4444444444444444, | |
| "eval_accuracy_label_politics": 0.8292682926829268, | |
| "eval_accuracy_label_religion": 0.0, | |
| "eval_accuracy_label_science and technology": 0.5416666666666666, | |
| "eval_accuracy_label_society": 0.23076923076923078, | |
| "eval_accuracy_label_sport": 0.9615384615384616, | |
| "eval_accuracy_label_weather": 1.0, | |
| "eval_f1": 0.7008818694331442, | |
| "eval_loss": 0.9380582571029663, | |
| "eval_precision": 0.7617561903814418, | |
| "eval_recall": 0.688125, | |
| "eval_runtime": 12.5844, | |
| "eval_samples_per_second": 127.142, | |
| "eval_steps_per_second": 7.946, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.177325581395349, | |
| "grad_norm": 12.958664894104004, | |
| "learning_rate": 1.6035805626598467e-05, | |
| "loss": 0.6513, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.191860465116279, | |
| "grad_norm": 13.338906288146973, | |
| "learning_rate": 1.5907928388746805e-05, | |
| "loss": 0.7241, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.2063953488372092, | |
| "grad_norm": 12.350173950195312, | |
| "learning_rate": 1.5780051150895143e-05, | |
| "loss": 0.6857, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.2209302325581395, | |
| "grad_norm": 9.547300338745117, | |
| "learning_rate": 1.565217391304348e-05, | |
| "loss": 0.5884, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.2354651162790697, | |
| "grad_norm": 9.108930587768555, | |
| "learning_rate": 1.5524296675191818e-05, | |
| "loss": 0.5416, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 7.098880767822266, | |
| "learning_rate": 1.5396419437340155e-05, | |
| "loss": 0.5443, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.2645348837209303, | |
| "grad_norm": 15.5491304397583, | |
| "learning_rate": 1.5268542199488493e-05, | |
| "loss": 0.638, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.2790697674418605, | |
| "grad_norm": 14.731600761413574, | |
| "learning_rate": 1.514066496163683e-05, | |
| "loss": 0.6516, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.2936046511627908, | |
| "grad_norm": 6.682343006134033, | |
| "learning_rate": 1.5012787723785167e-05, | |
| "loss": 0.6165, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.308139534883721, | |
| "grad_norm": 14.463234901428223, | |
| "learning_rate": 1.4884910485933506e-05, | |
| "loss": 0.6455, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.322674418604651, | |
| "grad_norm": 14.159065246582031, | |
| "learning_rate": 1.4757033248081842e-05, | |
| "loss": 0.6129, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.3372093023255813, | |
| "grad_norm": 14.65009880065918, | |
| "learning_rate": 1.4629156010230181e-05, | |
| "loss": 0.5702, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.3517441860465116, | |
| "grad_norm": 16.38220977783203, | |
| "learning_rate": 1.4501278772378517e-05, | |
| "loss": 0.6183, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.3662790697674418, | |
| "grad_norm": 9.162236213684082, | |
| "learning_rate": 1.4373401534526856e-05, | |
| "loss": 0.547, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.380813953488372, | |
| "grad_norm": 9.222379684448242, | |
| "learning_rate": 1.4245524296675192e-05, | |
| "loss": 0.5629, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.3953488372093024, | |
| "grad_norm": 8.016491889953613, | |
| "learning_rate": 1.4117647058823532e-05, | |
| "loss": 0.6529, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.4098837209302326, | |
| "grad_norm": 8.260580062866211, | |
| "learning_rate": 1.3989769820971867e-05, | |
| "loss": 0.5309, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.4244186046511627, | |
| "grad_norm": 16.81390953063965, | |
| "learning_rate": 1.3861892583120207e-05, | |
| "loss": 0.6785, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.4389534883720931, | |
| "grad_norm": 11.895804405212402, | |
| "learning_rate": 1.3734015345268543e-05, | |
| "loss": 0.5976, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.4534883720930232, | |
| "grad_norm": 8.548684120178223, | |
| "learning_rate": 1.3606138107416882e-05, | |
| "loss": 0.5554, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.4534883720930232, | |
| "eval_accuracy": 0.7025, | |
| "eval_accuracy_label_arts, culture, entertainment and media": 0.7368421052631579, | |
| "eval_accuracy_label_conflict, war and peace": 0.6477794793261868, | |
| "eval_accuracy_label_crime, law and justice": 0.9020618556701031, | |
| "eval_accuracy_label_disaster, accident, and emergency incident": 0.8561643835616438, | |
| "eval_accuracy_label_economy, business, and finance": 0.6601941747572816, | |
| "eval_accuracy_label_environment": 0.3103448275862069, | |
| "eval_accuracy_label_health": 0.7777777777777778, | |
| "eval_accuracy_label_human interest": 0.36363636363636365, | |
| "eval_accuracy_label_labour": 0.5, | |
| "eval_accuracy_label_lifestyle and leisure": 0.5555555555555556, | |
| "eval_accuracy_label_politics": 0.8083623693379791, | |
| "eval_accuracy_label_religion": 0.0, | |
| "eval_accuracy_label_science and technology": 0.5, | |
| "eval_accuracy_label_society": 0.18461538461538463, | |
| "eval_accuracy_label_sport": 0.9615384615384616, | |
| "eval_accuracy_label_weather": 1.0, | |
| "eval_f1": 0.7123723699365401, | |
| "eval_loss": 0.8791087865829468, | |
| "eval_precision": 0.7628439748212975, | |
| "eval_recall": 0.7025, | |
| "eval_runtime": 12.5844, | |
| "eval_samples_per_second": 127.142, | |
| "eval_steps_per_second": 7.946, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.4680232558139534, | |
| "grad_norm": 13.218795776367188, | |
| "learning_rate": 1.3478260869565218e-05, | |
| "loss": 0.5587, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.4825581395348837, | |
| "grad_norm": 21.222410202026367, | |
| "learning_rate": 1.3350383631713557e-05, | |
| "loss": 0.6128, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.497093023255814, | |
| "grad_norm": 10.608304977416992, | |
| "learning_rate": 1.3222506393861893e-05, | |
| "loss": 0.5334, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.5116279069767442, | |
| "grad_norm": 9.839329719543457, | |
| "learning_rate": 1.3094629156010232e-05, | |
| "loss": 0.4974, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.5261627906976745, | |
| "grad_norm": 16.351816177368164, | |
| "learning_rate": 1.2966751918158568e-05, | |
| "loss": 0.5387, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.5406976744186047, | |
| "grad_norm": 13.274540901184082, | |
| "learning_rate": 1.2838874680306906e-05, | |
| "loss": 0.5928, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.5552325581395348, | |
| "grad_norm": 11.493636131286621, | |
| "learning_rate": 1.2710997442455244e-05, | |
| "loss": 0.4391, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.5697674418604652, | |
| "grad_norm": 7.543900012969971, | |
| "learning_rate": 1.2583120204603581e-05, | |
| "loss": 0.5845, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.5843023255813953, | |
| "grad_norm": 18.52677345275879, | |
| "learning_rate": 1.2455242966751919e-05, | |
| "loss": 0.4857, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.5988372093023255, | |
| "grad_norm": 10.437915802001953, | |
| "learning_rate": 1.2327365728900256e-05, | |
| "loss": 0.4946, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.6133720930232558, | |
| "grad_norm": 9.237434387207031, | |
| "learning_rate": 1.2199488491048594e-05, | |
| "loss": 0.6215, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.627906976744186, | |
| "grad_norm": 15.632594108581543, | |
| "learning_rate": 1.2071611253196932e-05, | |
| "loss": 0.5642, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.6424418604651163, | |
| "grad_norm": 16.443044662475586, | |
| "learning_rate": 1.194373401534527e-05, | |
| "loss": 0.5714, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.6569767441860463, | |
| "grad_norm": 15.662179946899414, | |
| "learning_rate": 1.1815856777493607e-05, | |
| "loss": 0.4438, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.6715116279069768, | |
| "grad_norm": 14.527420043945312, | |
| "learning_rate": 1.1687979539641944e-05, | |
| "loss": 0.5733, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.6860465116279069, | |
| "grad_norm": 20.69162368774414, | |
| "learning_rate": 1.1560102301790282e-05, | |
| "loss": 0.5324, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.7005813953488373, | |
| "grad_norm": 11.502646446228027, | |
| "learning_rate": 1.143222506393862e-05, | |
| "loss": 0.4495, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.7151162790697674, | |
| "grad_norm": 8.692178726196289, | |
| "learning_rate": 1.1304347826086957e-05, | |
| "loss": 0.4218, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.7296511627906976, | |
| "grad_norm": 11.3637113571167, | |
| "learning_rate": 1.1176470588235295e-05, | |
| "loss": 0.5778, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.744186046511628, | |
| "grad_norm": 7.9896721839904785, | |
| "learning_rate": 1.1048593350383632e-05, | |
| "loss": 0.4396, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.744186046511628, | |
| "eval_accuracy": 0.7175, | |
| "eval_accuracy_label_arts, culture, entertainment and media": 0.7894736842105263, | |
| "eval_accuracy_label_conflict, war and peace": 0.663093415007657, | |
| "eval_accuracy_label_crime, law and justice": 0.8195876288659794, | |
| "eval_accuracy_label_disaster, accident, and emergency incident": 0.8835616438356164, | |
| "eval_accuracy_label_economy, business, and finance": 0.6893203883495146, | |
| "eval_accuracy_label_environment": 0.3793103448275862, | |
| "eval_accuracy_label_health": 0.8333333333333334, | |
| "eval_accuracy_label_human interest": 0.4090909090909091, | |
| "eval_accuracy_label_labour": 0.5, | |
| "eval_accuracy_label_lifestyle and leisure": 0.5555555555555556, | |
| "eval_accuracy_label_politics": 0.8362369337979094, | |
| "eval_accuracy_label_religion": 0.0, | |
| "eval_accuracy_label_science and technology": 0.4166666666666667, | |
| "eval_accuracy_label_society": 0.36923076923076925, | |
| "eval_accuracy_label_sport": 0.9615384615384616, | |
| "eval_accuracy_label_weather": 1.0, | |
| "eval_f1": 0.7279616125865503, | |
| "eval_loss": 0.8275023698806763, | |
| "eval_precision": 0.7685711157026855, | |
| "eval_recall": 0.7175, | |
| "eval_runtime": 12.6195, | |
| "eval_samples_per_second": 126.788, | |
| "eval_steps_per_second": 7.924, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.7587209302325582, | |
| "grad_norm": 8.378173828125, | |
| "learning_rate": 1.092071611253197e-05, | |
| "loss": 0.4187, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 1.7732558139534884, | |
| "grad_norm": 22.63587760925293, | |
| "learning_rate": 1.0792838874680308e-05, | |
| "loss": 0.4558, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 1.7877906976744184, | |
| "grad_norm": 16.91942596435547, | |
| "learning_rate": 1.0664961636828645e-05, | |
| "loss": 0.6778, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.802325581395349, | |
| "grad_norm": 11.042551040649414, | |
| "learning_rate": 1.0537084398976983e-05, | |
| "loss": 0.5229, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 1.816860465116279, | |
| "grad_norm": 20.5400390625, | |
| "learning_rate": 1.040920716112532e-05, | |
| "loss": 0.5121, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.8313953488372094, | |
| "grad_norm": 12.495570182800293, | |
| "learning_rate": 1.0281329923273658e-05, | |
| "loss": 0.4737, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.8459302325581395, | |
| "grad_norm": 11.538856506347656, | |
| "learning_rate": 1.0153452685421996e-05, | |
| "loss": 0.3766, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.8604651162790697, | |
| "grad_norm": 8.165515899658203, | |
| "learning_rate": 1.0025575447570333e-05, | |
| "loss": 0.5152, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 1.875, | |
| "grad_norm": 8.704108238220215, | |
| "learning_rate": 9.897698209718671e-06, | |
| "loss": 0.5027, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.8895348837209303, | |
| "grad_norm": 14.96538257598877, | |
| "learning_rate": 9.769820971867009e-06, | |
| "loss": 0.5427, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.9040697674418605, | |
| "grad_norm": 11.60605525970459, | |
| "learning_rate": 9.641943734015346e-06, | |
| "loss": 0.4236, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.9186046511627906, | |
| "grad_norm": 10.070377349853516, | |
| "learning_rate": 9.514066496163684e-06, | |
| "loss": 0.4683, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 1.933139534883721, | |
| "grad_norm": 8.57789134979248, | |
| "learning_rate": 9.386189258312021e-06, | |
| "loss": 0.505, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.947674418604651, | |
| "grad_norm": 14.64469051361084, | |
| "learning_rate": 9.258312020460359e-06, | |
| "loss": 0.4311, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.9622093023255816, | |
| "grad_norm": 16.99574851989746, | |
| "learning_rate": 9.130434782608697e-06, | |
| "loss": 0.4414, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.9767441860465116, | |
| "grad_norm": 14.059187889099121, | |
| "learning_rate": 9.002557544757034e-06, | |
| "loss": 0.5001, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.9912790697674418, | |
| "grad_norm": 7.389652729034424, | |
| "learning_rate": 8.874680306905372e-06, | |
| "loss": 0.4546, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 2.005813953488372, | |
| "grad_norm": 11.378294944763184, | |
| "learning_rate": 8.74680306905371e-06, | |
| "loss": 0.4447, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 2.020348837209302, | |
| "grad_norm": 9.877668380737305, | |
| "learning_rate": 8.618925831202047e-06, | |
| "loss": 0.4254, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 2.0348837209302326, | |
| "grad_norm": 5.385466575622559, | |
| "learning_rate": 8.491048593350385e-06, | |
| "loss": 0.383, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.0348837209302326, | |
| "eval_accuracy": 0.745, | |
| "eval_accuracy_label_arts, culture, entertainment and media": 0.6842105263157895, | |
| "eval_accuracy_label_conflict, war and peace": 0.7840735068912711, | |
| "eval_accuracy_label_crime, law and justice": 0.8865979381443299, | |
| "eval_accuracy_label_disaster, accident, and emergency incident": 0.8767123287671232, | |
| "eval_accuracy_label_economy, business, and finance": 0.7087378640776699, | |
| "eval_accuracy_label_environment": 0.4482758620689655, | |
| "eval_accuracy_label_health": 0.7777777777777778, | |
| "eval_accuracy_label_human interest": 0.4090909090909091, | |
| "eval_accuracy_label_labour": 0.5, | |
| "eval_accuracy_label_lifestyle and leisure": 0.5555555555555556, | |
| "eval_accuracy_label_politics": 0.6898954703832753, | |
| "eval_accuracy_label_religion": 0.0, | |
| "eval_accuracy_label_science and technology": 0.4166666666666667, | |
| "eval_accuracy_label_society": 0.2923076923076923, | |
| "eval_accuracy_label_sport": 0.9615384615384616, | |
| "eval_accuracy_label_weather": 0.0, | |
| "eval_f1": 0.7500828579968577, | |
| "eval_loss": 0.7928646206855774, | |
| "eval_precision": 0.7652956903435105, | |
| "eval_recall": 0.745, | |
| "eval_runtime": 12.617, | |
| "eval_samples_per_second": 126.813, | |
| "eval_steps_per_second": 7.926, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.0494186046511627, | |
| "grad_norm": 10.072980880737305, | |
| "learning_rate": 8.363171355498722e-06, | |
| "loss": 0.3935, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 2.063953488372093, | |
| "grad_norm": 8.466812133789062, | |
| "learning_rate": 8.23529411764706e-06, | |
| "loss": 0.3343, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 2.078488372093023, | |
| "grad_norm": 8.366528511047363, | |
| "learning_rate": 8.107416879795397e-06, | |
| "loss": 0.3865, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 2.0930232558139537, | |
| "grad_norm": 9.42149543762207, | |
| "learning_rate": 7.979539641943735e-06, | |
| "loss": 0.3071, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 2.1075581395348837, | |
| "grad_norm": 7.807155609130859, | |
| "learning_rate": 7.851662404092073e-06, | |
| "loss": 0.3173, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.1220930232558137, | |
| "grad_norm": 9.655524253845215, | |
| "learning_rate": 7.72378516624041e-06, | |
| "loss": 0.4508, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 2.136627906976744, | |
| "grad_norm": 8.439408302307129, | |
| "learning_rate": 7.595907928388747e-06, | |
| "loss": 0.3745, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 2.1511627906976742, | |
| "grad_norm": 8.737176895141602, | |
| "learning_rate": 7.468030690537085e-06, | |
| "loss": 0.3132, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 2.1656976744186047, | |
| "grad_norm": 14.63242244720459, | |
| "learning_rate": 7.340153452685422e-06, | |
| "loss": 0.3235, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 2.1802325581395348, | |
| "grad_norm": 10.674861907958984, | |
| "learning_rate": 7.21227621483376e-06, | |
| "loss": 0.39, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.1947674418604652, | |
| "grad_norm": 21.4650936126709, | |
| "learning_rate": 7.084398976982097e-06, | |
| "loss": 0.3526, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 2.2093023255813953, | |
| "grad_norm": 9.0557222366333, | |
| "learning_rate": 6.956521739130435e-06, | |
| "loss": 0.3315, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 2.2238372093023258, | |
| "grad_norm": 9.88924789428711, | |
| "learning_rate": 6.828644501278773e-06, | |
| "loss": 0.342, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 2.238372093023256, | |
| "grad_norm": 10.470682144165039, | |
| "learning_rate": 6.70076726342711e-06, | |
| "loss": 0.3666, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 2.2529069767441863, | |
| "grad_norm": 10.087296485900879, | |
| "learning_rate": 6.572890025575448e-06, | |
| "loss": 0.3509, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.2674418604651163, | |
| "grad_norm": 9.986245155334473, | |
| "learning_rate": 6.4450127877237854e-06, | |
| "loss": 0.2961, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 2.2819767441860463, | |
| "grad_norm": 4.27475643157959, | |
| "learning_rate": 6.317135549872123e-06, | |
| "loss": 0.3514, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 2.296511627906977, | |
| "grad_norm": 18.000774383544922, | |
| "learning_rate": 6.189258312020461e-06, | |
| "loss": 0.2998, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 2.311046511627907, | |
| "grad_norm": 11.935790061950684, | |
| "learning_rate": 6.061381074168798e-06, | |
| "loss": 0.3849, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 2.3255813953488373, | |
| "grad_norm": 6.844956874847412, | |
| "learning_rate": 5.933503836317136e-06, | |
| "loss": 0.3418, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.3255813953488373, | |
| "eval_accuracy": 0.74375, | |
| "eval_accuracy_label_arts, culture, entertainment and media": 0.7894736842105263, | |
| "eval_accuracy_label_conflict, war and peace": 0.7350689127105666, | |
| "eval_accuracy_label_crime, law and justice": 0.9072164948453608, | |
| "eval_accuracy_label_disaster, accident, and emergency incident": 0.8493150684931506, | |
| "eval_accuracy_label_economy, business, and finance": 0.7864077669902912, | |
| "eval_accuracy_label_environment": 0.4482758620689655, | |
| "eval_accuracy_label_health": 0.7777777777777778, | |
| "eval_accuracy_label_human interest": 0.3181818181818182, | |
| "eval_accuracy_label_labour": 0.5, | |
| "eval_accuracy_label_lifestyle and leisure": 0.5555555555555556, | |
| "eval_accuracy_label_politics": 0.7909407665505227, | |
| "eval_accuracy_label_religion": 0.0, | |
| "eval_accuracy_label_science and technology": 0.4166666666666667, | |
| "eval_accuracy_label_society": 0.18461538461538463, | |
| "eval_accuracy_label_sport": 0.9615384615384616, | |
| "eval_accuracy_label_weather": 0.0, | |
| "eval_f1": 0.7440049917955611, | |
| "eval_loss": 0.804217517375946, | |
| "eval_precision": 0.768551401471047, | |
| "eval_recall": 0.74375, | |
| "eval_runtime": 12.6056, | |
| "eval_samples_per_second": 126.928, | |
| "eval_steps_per_second": 7.933, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.3401162790697674, | |
| "grad_norm": 14.338434219360352, | |
| "learning_rate": 5.8056265984654735e-06, | |
| "loss": 0.2883, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 2.354651162790698, | |
| "grad_norm": 5.656998634338379, | |
| "learning_rate": 5.677749360613811e-06, | |
| "loss": 0.2737, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 2.369186046511628, | |
| "grad_norm": 6.466466903686523, | |
| "learning_rate": 5.549872122762149e-06, | |
| "loss": 0.331, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 2.383720930232558, | |
| "grad_norm": 11.816296577453613, | |
| "learning_rate": 5.421994884910486e-06, | |
| "loss": 0.3733, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 2.3982558139534884, | |
| "grad_norm": 8.958051681518555, | |
| "learning_rate": 5.294117647058824e-06, | |
| "loss": 0.2802, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.4127906976744184, | |
| "grad_norm": 13.985147476196289, | |
| "learning_rate": 5.1662404092071615e-06, | |
| "loss": 0.353, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 2.427325581395349, | |
| "grad_norm": 17.26644515991211, | |
| "learning_rate": 5.038363171355499e-06, | |
| "loss": 0.3399, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 2.441860465116279, | |
| "grad_norm": 13.139063835144043, | |
| "learning_rate": 4.910485933503837e-06, | |
| "loss": 0.2949, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 2.4563953488372094, | |
| "grad_norm": 15.950389862060547, | |
| "learning_rate": 4.782608695652174e-06, | |
| "loss": 0.3742, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 2.4709302325581395, | |
| "grad_norm": 8.10091495513916, | |
| "learning_rate": 4.654731457800512e-06, | |
| "loss": 0.2605, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.4854651162790695, | |
| "grad_norm": 8.73356819152832, | |
| "learning_rate": 4.5268542199488496e-06, | |
| "loss": 0.2077, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 7.236685276031494, | |
| "learning_rate": 4.398976982097187e-06, | |
| "loss": 0.3526, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 2.5145348837209305, | |
| "grad_norm": 6.902716159820557, | |
| "learning_rate": 4.271099744245525e-06, | |
| "loss": 0.2633, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 2.5290697674418605, | |
| "grad_norm": 4.061647891998291, | |
| "learning_rate": 4.143222506393862e-06, | |
| "loss": 0.2682, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 2.5436046511627906, | |
| "grad_norm": 3.789299488067627, | |
| "learning_rate": 4.0153452685422e-06, | |
| "loss": 0.2996, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.558139534883721, | |
| "grad_norm": 7.185976505279541, | |
| "learning_rate": 3.887468030690537e-06, | |
| "loss": 0.323, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 2.572674418604651, | |
| "grad_norm": 9.113693237304688, | |
| "learning_rate": 3.7595907928388748e-06, | |
| "loss": 0.2854, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 2.5872093023255816, | |
| "grad_norm": 10.132606506347656, | |
| "learning_rate": 3.6317135549872124e-06, | |
| "loss": 0.351, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 2.6017441860465116, | |
| "grad_norm": 12.435748100280762, | |
| "learning_rate": 3.50383631713555e-06, | |
| "loss": 0.3271, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 2.616279069767442, | |
| "grad_norm": 15.90312671661377, | |
| "learning_rate": 3.3759590792838876e-06, | |
| "loss": 0.248, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.616279069767442, | |
| "eval_accuracy": 0.7275, | |
| "eval_accuracy_label_arts, culture, entertainment and media": 0.6842105263157895, | |
| "eval_accuracy_label_conflict, war and peace": 0.6891271056661562, | |
| "eval_accuracy_label_crime, law and justice": 0.8814432989690721, | |
| "eval_accuracy_label_disaster, accident, and emergency incident": 0.8698630136986302, | |
| "eval_accuracy_label_economy, business, and finance": 0.7572815533980582, | |
| "eval_accuracy_label_environment": 0.41379310344827586, | |
| "eval_accuracy_label_health": 0.8333333333333334, | |
| "eval_accuracy_label_human interest": 0.4090909090909091, | |
| "eval_accuracy_label_labour": 0.5, | |
| "eval_accuracy_label_lifestyle and leisure": 0.5555555555555556, | |
| "eval_accuracy_label_politics": 0.8013937282229965, | |
| "eval_accuracy_label_religion": 0.0, | |
| "eval_accuracy_label_science and technology": 0.4166666666666667, | |
| "eval_accuracy_label_society": 0.27692307692307694, | |
| "eval_accuracy_label_sport": 0.9615384615384616, | |
| "eval_accuracy_label_weather": 0.0, | |
| "eval_f1": 0.7324830328014649, | |
| "eval_loss": 0.8387188911437988, | |
| "eval_precision": 0.7610492043473545, | |
| "eval_recall": 0.7275, | |
| "eval_runtime": 12.5904, | |
| "eval_samples_per_second": 127.081, | |
| "eval_steps_per_second": 7.943, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.630813953488372, | |
| "grad_norm": 14.707446098327637, | |
| "learning_rate": 3.248081841432225e-06, | |
| "loss": 0.3358, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 2.645348837209302, | |
| "grad_norm": 8.684798240661621, | |
| "learning_rate": 3.120204603580563e-06, | |
| "loss": 0.3082, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 2.6598837209302326, | |
| "grad_norm": 8.125092506408691, | |
| "learning_rate": 2.9923273657289004e-06, | |
| "loss": 0.2818, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 2.6744186046511627, | |
| "grad_norm": 10.351339340209961, | |
| "learning_rate": 2.864450127877238e-06, | |
| "loss": 0.3465, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 2.688953488372093, | |
| "grad_norm": 9.536513328552246, | |
| "learning_rate": 2.7365728900255756e-06, | |
| "loss": 0.3022, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.703488372093023, | |
| "grad_norm": 7.122174263000488, | |
| "learning_rate": 2.6086956521739132e-06, | |
| "loss": 0.3742, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 2.7180232558139537, | |
| "grad_norm": 11.325377464294434, | |
| "learning_rate": 2.480818414322251e-06, | |
| "loss": 0.2912, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 2.7325581395348837, | |
| "grad_norm": 9.651542663574219, | |
| "learning_rate": 2.3529411764705885e-06, | |
| "loss": 0.3156, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 2.7470930232558137, | |
| "grad_norm": 16.223478317260742, | |
| "learning_rate": 2.225063938618926e-06, | |
| "loss": 0.3405, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 2.761627906976744, | |
| "grad_norm": 6.137476921081543, | |
| "learning_rate": 2.0971867007672637e-06, | |
| "loss": 0.1922, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.7761627906976747, | |
| "grad_norm": 9.50839614868164, | |
| "learning_rate": 1.9693094629156013e-06, | |
| "loss": 0.2896, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 2.7906976744186047, | |
| "grad_norm": 8.988372802734375, | |
| "learning_rate": 1.8414322250639389e-06, | |
| "loss": 0.3286, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 2.8052325581395348, | |
| "grad_norm": 9.569466590881348, | |
| "learning_rate": 1.7135549872122765e-06, | |
| "loss": 0.3463, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 2.8197674418604652, | |
| "grad_norm": 7.808605194091797, | |
| "learning_rate": 1.585677749360614e-06, | |
| "loss": 0.3117, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 2.8343023255813953, | |
| "grad_norm": 9.752486228942871, | |
| "learning_rate": 1.4578005115089517e-06, | |
| "loss": 0.3708, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.8488372093023253, | |
| "grad_norm": 6.103060722351074, | |
| "learning_rate": 1.3299232736572893e-06, | |
| "loss": 0.2766, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 2.863372093023256, | |
| "grad_norm": 8.869214057922363, | |
| "learning_rate": 1.2020460358056267e-06, | |
| "loss": 0.2762, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 2.8779069767441863, | |
| "grad_norm": 8.96507740020752, | |
| "learning_rate": 1.0741687979539643e-06, | |
| "loss": 0.2874, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 2.8924418604651163, | |
| "grad_norm": 8.362478256225586, | |
| "learning_rate": 9.462915601023019e-07, | |
| "loss": 0.2944, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 2.9069767441860463, | |
| "grad_norm": 7.523708820343018, | |
| "learning_rate": 8.184143222506395e-07, | |
| "loss": 0.2525, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.9069767441860463, | |
| "eval_accuracy": 0.735, | |
| "eval_accuracy_label_arts, culture, entertainment and media": 0.6842105263157895, | |
| "eval_accuracy_label_conflict, war and peace": 0.7105666156202144, | |
| "eval_accuracy_label_crime, law and justice": 0.8762886597938144, | |
| "eval_accuracy_label_disaster, accident, and emergency incident": 0.8698630136986302, | |
| "eval_accuracy_label_economy, business, and finance": 0.6796116504854369, | |
| "eval_accuracy_label_environment": 0.4482758620689655, | |
| "eval_accuracy_label_health": 0.7222222222222222, | |
| "eval_accuracy_label_human interest": 0.36363636363636365, | |
| "eval_accuracy_label_labour": 0.5, | |
| "eval_accuracy_label_lifestyle and leisure": 0.5555555555555556, | |
| "eval_accuracy_label_politics": 0.8153310104529616, | |
| "eval_accuracy_label_religion": 0.0, | |
| "eval_accuracy_label_science and technology": 0.4583333333333333, | |
| "eval_accuracy_label_society": 0.3384615384615385, | |
| "eval_accuracy_label_sport": 0.9615384615384616, | |
| "eval_accuracy_label_weather": 0.0, | |
| "eval_f1": 0.7412823760546817, | |
| "eval_loss": 0.8137220740318298, | |
| "eval_precision": 0.7696596009813345, | |
| "eval_recall": 0.735, | |
| "eval_runtime": 12.5933, | |
| "eval_samples_per_second": 127.051, | |
| "eval_steps_per_second": 7.941, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.921511627906977, | |
| "grad_norm": 9.854249954223633, | |
| "learning_rate": 6.90537084398977e-07, | |
| "loss": 0.2625, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 2.936046511627907, | |
| "grad_norm": 7.306544780731201, | |
| "learning_rate": 5.626598465473146e-07, | |
| "loss": 0.3285, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 2.9505813953488373, | |
| "grad_norm": 4.267274856567383, | |
| "learning_rate": 4.347826086956522e-07, | |
| "loss": 0.3445, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 2.9651162790697674, | |
| "grad_norm": 15.480578422546387, | |
| "learning_rate": 3.069053708439898e-07, | |
| "loss": 0.3066, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 2.979651162790698, | |
| "grad_norm": 11.125273704528809, | |
| "learning_rate": 1.7902813299232738e-07, | |
| "loss": 0.3026, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.994186046511628, | |
| "grad_norm": 12.110825538635254, | |
| "learning_rate": 5.115089514066497e-08, | |
| "loss": 0.2815, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 2064, | |
| "total_flos": 8687308686630912.0, | |
| "train_loss": 0.8068887479603291, | |
| "train_runtime": 1561.1428, | |
| "train_samples_per_second": 42.294, | |
| "train_steps_per_second": 1.322 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.743125, | |
| "eval_accuracy_label_arts, culture, entertainment and media": 0.6842105263157895, | |
| "eval_accuracy_label_conflict, war and peace": 0.7350689127105666, | |
| "eval_accuracy_label_crime, law and justice": 0.8917525773195877, | |
| "eval_accuracy_label_disaster, accident, and emergency incident": 0.8698630136986302, | |
| "eval_accuracy_label_economy, business, and finance": 0.6893203883495146, | |
| "eval_accuracy_label_environment": 0.4482758620689655, | |
| "eval_accuracy_label_health": 0.7222222222222222, | |
| "eval_accuracy_label_human interest": 0.3181818181818182, | |
| "eval_accuracy_label_labour": 0.5, | |
| "eval_accuracy_label_lifestyle and leisure": 0.5555555555555556, | |
| "eval_accuracy_label_politics": 0.7909407665505227, | |
| "eval_accuracy_label_religion": 0.0, | |
| "eval_accuracy_label_science and technology": 0.4583333333333333, | |
| "eval_accuracy_label_society": 0.35384615384615387, | |
| "eval_accuracy_label_sport": 0.9615384615384616, | |
| "eval_accuracy_label_weather": 0.0, | |
| "eval_f1": 0.747448709907019, | |
| "eval_loss": 0.8030127882957458, | |
| "eval_precision": 0.7695117491320402, | |
| "eval_recall": 0.743125, | |
| "eval_runtime": 12.5023, | |
| "eval_samples_per_second": 127.976, | |
| "eval_steps_per_second": 7.999, | |
| "step": 2064 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 2064, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 1000, | |
| "total_flos": 8687308686630912.0, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |