| { |
| "best_metric": 0.7036260157126459, |
| "best_model_checkpoint": "hq_fer2013notest/checkpoint-3530", |
| "epoch": 9.998585572842998, |
| "global_step": 3530, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.03, |
| "learning_rate": 2.8328611898017e-07, |
| "loss": 1.9499, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 5.6657223796034e-07, |
| "loss": 1.9546, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 8.4985835694051e-07, |
| "loss": 1.9464, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 1.13314447592068e-06, |
| "loss": 1.9487, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 1.41643059490085e-06, |
| "loss": 1.9466, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 1.69971671388102e-06, |
| "loss": 1.9392, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 1.98300283286119e-06, |
| "loss": 1.9351, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 2.26628895184136e-06, |
| "loss": 1.9287, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 2.54957507082153e-06, |
| "loss": 1.9207, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 2.8328611898017e-06, |
| "loss": 1.9098, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 3.11614730878187e-06, |
| "loss": 1.9077, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 3.39943342776204e-06, |
| "loss": 1.8908, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 3.68271954674221e-06, |
| "loss": 1.8814, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 3.96600566572238e-06, |
| "loss": 1.8634, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 4.24929178470255e-06, |
| "loss": 1.8403, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 4.53257790368272e-06, |
| "loss": 1.8193, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 4.81586402266289e-06, |
| "loss": 1.7788, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 5.09915014164306e-06, |
| "loss": 1.7874, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 5.382436260623229e-06, |
| "loss": 1.7342, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 5.6657223796034e-06, |
| "loss": 1.714, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 5.9490084985835695e-06, |
| "loss": 1.6659, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 6.23229461756374e-06, |
| "loss": 1.6448, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 6.51558073654391e-06, |
| "loss": 1.6183, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 6.79886685552408e-06, |
| "loss": 1.5849, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 7.082152974504249e-06, |
| "loss": 1.5351, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 7.36543909348442e-06, |
| "loss": 1.5024, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 7.64872521246459e-06, |
| "loss": 1.4676, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 7.93201133144476e-06, |
| "loss": 1.4434, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 8.215297450424929e-06, |
| "loss": 1.4305, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 8.4985835694051e-06, |
| "loss": 1.3976, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 8.78186968838527e-06, |
| "loss": 1.3898, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 9.06515580736544e-06, |
| "loss": 1.3764, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 9.34844192634561e-06, |
| "loss": 1.3302, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 9.63172804532578e-06, |
| "loss": 1.2724, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 9.91501416430595e-06, |
| "loss": 1.2982, |
| "step": 350 |
| }, |
| { |
| "epoch": 1.0, |
| "eval_accuracy": 0.5635447068187848, |
| "eval_f1": 0.5167583882016407, |
| "eval_loss": 1.270811676979065, |
| "eval_precision": 0.5106863863183857, |
| "eval_recall": 0.5635447068187848, |
| "eval_runtime": 355.6788, |
| "eval_samples_per_second": 31.79, |
| "eval_steps_per_second": 0.995, |
| "step": 353 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 9.977966635190433e-06, |
| "loss": 1.3363, |
| "step": 360 |
| }, |
| { |
| "epoch": 1.05, |
| "learning_rate": 9.946490399748192e-06, |
| "loss": 1.2513, |
| "step": 370 |
| }, |
| { |
| "epoch": 1.08, |
| "learning_rate": 9.91501416430595e-06, |
| "loss": 1.2351, |
| "step": 380 |
| }, |
| { |
| "epoch": 1.1, |
| "learning_rate": 9.883537928863708e-06, |
| "loss": 1.2467, |
| "step": 390 |
| }, |
| { |
| "epoch": 1.13, |
| "learning_rate": 9.852061693421467e-06, |
| "loss": 1.1845, |
| "step": 400 |
| }, |
| { |
| "epoch": 1.16, |
| "learning_rate": 9.820585457979227e-06, |
| "loss": 1.1763, |
| "step": 410 |
| }, |
| { |
| "epoch": 1.19, |
| "learning_rate": 9.789109222536986e-06, |
| "loss": 1.1951, |
| "step": 420 |
| }, |
| { |
| "epoch": 1.22, |
| "learning_rate": 9.757632987094745e-06, |
| "loss": 1.1537, |
| "step": 430 |
| }, |
| { |
| "epoch": 1.25, |
| "learning_rate": 9.726156751652504e-06, |
| "loss": 1.1759, |
| "step": 440 |
| }, |
| { |
| "epoch": 1.27, |
| "learning_rate": 9.694680516210261e-06, |
| "loss": 1.1568, |
| "step": 450 |
| }, |
| { |
| "epoch": 1.3, |
| "learning_rate": 9.66320428076802e-06, |
| "loss": 1.1592, |
| "step": 460 |
| }, |
| { |
| "epoch": 1.33, |
| "learning_rate": 9.63172804532578e-06, |
| "loss": 1.1321, |
| "step": 470 |
| }, |
| { |
| "epoch": 1.36, |
| "learning_rate": 9.600251809883539e-06, |
| "loss": 1.1427, |
| "step": 480 |
| }, |
| { |
| "epoch": 1.39, |
| "learning_rate": 9.568775574441298e-06, |
| "loss": 1.0947, |
| "step": 490 |
| }, |
| { |
| "epoch": 1.42, |
| "learning_rate": 9.537299338999057e-06, |
| "loss": 1.1334, |
| "step": 500 |
| }, |
| { |
| "epoch": 1.44, |
| "learning_rate": 9.505823103556814e-06, |
| "loss": 1.1015, |
| "step": 510 |
| }, |
| { |
| "epoch": 1.47, |
| "learning_rate": 9.474346868114575e-06, |
| "loss": 1.0936, |
| "step": 520 |
| }, |
| { |
| "epoch": 1.5, |
| "learning_rate": 9.442870632672332e-06, |
| "loss": 1.0957, |
| "step": 530 |
| }, |
| { |
| "epoch": 1.53, |
| "learning_rate": 9.411394397230093e-06, |
| "loss": 1.0667, |
| "step": 540 |
| }, |
| { |
| "epoch": 1.56, |
| "learning_rate": 9.37991816178785e-06, |
| "loss": 1.1095, |
| "step": 550 |
| }, |
| { |
| "epoch": 1.59, |
| "learning_rate": 9.34844192634561e-06, |
| "loss": 1.0726, |
| "step": 560 |
| }, |
| { |
| "epoch": 1.61, |
| "learning_rate": 9.316965690903369e-06, |
| "loss": 1.0828, |
| "step": 570 |
| }, |
| { |
| "epoch": 1.64, |
| "learning_rate": 9.285489455461128e-06, |
| "loss": 1.0642, |
| "step": 580 |
| }, |
| { |
| "epoch": 1.67, |
| "learning_rate": 9.254013220018887e-06, |
| "loss": 1.0758, |
| "step": 590 |
| }, |
| { |
| "epoch": 1.7, |
| "learning_rate": 9.222536984576646e-06, |
| "loss": 1.0554, |
| "step": 600 |
| }, |
| { |
| "epoch": 1.73, |
| "learning_rate": 9.191060749134404e-06, |
| "loss": 1.0505, |
| "step": 610 |
| }, |
| { |
| "epoch": 1.76, |
| "learning_rate": 9.159584513692163e-06, |
| "loss": 1.0214, |
| "step": 620 |
| }, |
| { |
| "epoch": 1.78, |
| "learning_rate": 9.128108278249922e-06, |
| "loss": 1.0502, |
| "step": 630 |
| }, |
| { |
| "epoch": 1.81, |
| "learning_rate": 9.096632042807681e-06, |
| "loss": 1.0673, |
| "step": 640 |
| }, |
| { |
| "epoch": 1.84, |
| "learning_rate": 9.06515580736544e-06, |
| "loss": 1.0627, |
| "step": 650 |
| }, |
| { |
| "epoch": 1.87, |
| "learning_rate": 9.033679571923199e-06, |
| "loss": 1.03, |
| "step": 660 |
| }, |
| { |
| "epoch": 1.9, |
| "learning_rate": 9.002203336480958e-06, |
| "loss": 1.047, |
| "step": 670 |
| }, |
| { |
| "epoch": 1.93, |
| "learning_rate": 8.970727101038716e-06, |
| "loss": 1.0401, |
| "step": 680 |
| }, |
| { |
| "epoch": 1.95, |
| "learning_rate": 8.939250865596475e-06, |
| "loss": 1.0169, |
| "step": 690 |
| }, |
| { |
| "epoch": 1.98, |
| "learning_rate": 8.907774630154234e-06, |
| "loss": 1.0218, |
| "step": 700 |
| }, |
| { |
| "epoch": 2.0, |
| "eval_accuracy": 0.6411072786769259, |
| "eval_f1": 0.6300997297814149, |
| "eval_loss": 1.0159069299697876, |
| "eval_precision": 0.639687066833172, |
| "eval_recall": 0.6411072786769259, |
| "eval_runtime": 348.5034, |
| "eval_samples_per_second": 32.444, |
| "eval_steps_per_second": 1.016, |
| "step": 706 |
| }, |
| { |
| "epoch": 2.01, |
| "learning_rate": 8.876298394711993e-06, |
| "loss": 1.051, |
| "step": 710 |
| }, |
| { |
| "epoch": 2.04, |
| "learning_rate": 8.844822159269752e-06, |
| "loss": 1.0255, |
| "step": 720 |
| }, |
| { |
| "epoch": 2.07, |
| "learning_rate": 8.813345923827511e-06, |
| "loss": 0.9958, |
| "step": 730 |
| }, |
| { |
| "epoch": 2.1, |
| "learning_rate": 8.78186968838527e-06, |
| "loss": 0.9825, |
| "step": 740 |
| }, |
| { |
| "epoch": 2.12, |
| "learning_rate": 8.75039345294303e-06, |
| "loss": 0.9707, |
| "step": 750 |
| }, |
| { |
| "epoch": 2.15, |
| "learning_rate": 8.718917217500787e-06, |
| "loss": 0.9753, |
| "step": 760 |
| }, |
| { |
| "epoch": 2.18, |
| "learning_rate": 8.687440982058548e-06, |
| "loss": 0.9378, |
| "step": 770 |
| }, |
| { |
| "epoch": 2.21, |
| "learning_rate": 8.655964746616305e-06, |
| "loss": 0.9686, |
| "step": 780 |
| }, |
| { |
| "epoch": 2.24, |
| "learning_rate": 8.624488511174064e-06, |
| "loss": 1.0152, |
| "step": 790 |
| }, |
| { |
| "epoch": 2.27, |
| "learning_rate": 8.593012275731823e-06, |
| "loss": 0.9509, |
| "step": 800 |
| }, |
| { |
| "epoch": 2.29, |
| "learning_rate": 8.561536040289582e-06, |
| "loss": 0.9493, |
| "step": 810 |
| }, |
| { |
| "epoch": 2.32, |
| "learning_rate": 8.53005980484734e-06, |
| "loss": 0.9555, |
| "step": 820 |
| }, |
| { |
| "epoch": 2.35, |
| "learning_rate": 8.4985835694051e-06, |
| "loss": 0.9328, |
| "step": 830 |
| }, |
| { |
| "epoch": 2.38, |
| "learning_rate": 8.467107333962858e-06, |
| "loss": 0.9853, |
| "step": 840 |
| }, |
| { |
| "epoch": 2.41, |
| "learning_rate": 8.435631098520619e-06, |
| "loss": 0.9353, |
| "step": 850 |
| }, |
| { |
| "epoch": 2.44, |
| "learning_rate": 8.404154863078376e-06, |
| "loss": 0.958, |
| "step": 860 |
| }, |
| { |
| "epoch": 2.46, |
| "learning_rate": 8.372678627636135e-06, |
| "loss": 0.974, |
| "step": 870 |
| }, |
| { |
| "epoch": 2.49, |
| "learning_rate": 8.341202392193894e-06, |
| "loss": 0.8983, |
| "step": 880 |
| }, |
| { |
| "epoch": 2.52, |
| "learning_rate": 8.309726156751653e-06, |
| "loss": 0.9215, |
| "step": 890 |
| }, |
| { |
| "epoch": 2.55, |
| "learning_rate": 8.278249921309413e-06, |
| "loss": 0.9315, |
| "step": 900 |
| }, |
| { |
| "epoch": 2.58, |
| "learning_rate": 8.246773685867172e-06, |
| "loss": 0.9416, |
| "step": 910 |
| }, |
| { |
| "epoch": 2.61, |
| "learning_rate": 8.215297450424929e-06, |
| "loss": 0.9368, |
| "step": 920 |
| }, |
| { |
| "epoch": 2.63, |
| "learning_rate": 8.183821214982688e-06, |
| "loss": 0.9287, |
| "step": 930 |
| }, |
| { |
| "epoch": 2.66, |
| "learning_rate": 8.152344979540447e-06, |
| "loss": 0.9257, |
| "step": 940 |
| }, |
| { |
| "epoch": 2.69, |
| "learning_rate": 8.120868744098206e-06, |
| "loss": 0.9561, |
| "step": 950 |
| }, |
| { |
| "epoch": 2.72, |
| "learning_rate": 8.089392508655966e-06, |
| "loss": 0.8878, |
| "step": 960 |
| }, |
| { |
| "epoch": 2.75, |
| "learning_rate": 8.057916273213725e-06, |
| "loss": 0.9317, |
| "step": 970 |
| }, |
| { |
| "epoch": 2.78, |
| "learning_rate": 8.026440037771484e-06, |
| "loss": 0.8958, |
| "step": 980 |
| }, |
| { |
| "epoch": 2.8, |
| "learning_rate": 7.994963802329241e-06, |
| "loss": 0.9433, |
| "step": 990 |
| }, |
| { |
| "epoch": 2.83, |
| "learning_rate": 7.963487566887e-06, |
| "loss": 0.9559, |
| "step": 1000 |
| }, |
| { |
| "epoch": 2.86, |
| "learning_rate": 7.93201133144476e-06, |
| "loss": 0.9449, |
| "step": 1010 |
| }, |
| { |
| "epoch": 2.89, |
| "learning_rate": 7.900535096002518e-06, |
| "loss": 0.8942, |
| "step": 1020 |
| }, |
| { |
| "epoch": 2.92, |
| "learning_rate": 7.869058860560278e-06, |
| "loss": 0.9393, |
| "step": 1030 |
| }, |
| { |
| "epoch": 2.94, |
| "learning_rate": 7.837582625118037e-06, |
| "loss": 0.9477, |
| "step": 1040 |
| }, |
| { |
| "epoch": 2.97, |
| "learning_rate": 7.806106389675794e-06, |
| "loss": 0.9437, |
| "step": 1050 |
| }, |
| { |
| "epoch": 3.0, |
| "eval_accuracy": 0.6631290351109932, |
| "eval_f1": 0.6556352912293997, |
| "eval_loss": 0.9452024698257446, |
| "eval_precision": 0.669817068462312, |
| "eval_recall": 0.6631290351109932, |
| "eval_runtime": 342.9112, |
| "eval_samples_per_second": 32.974, |
| "eval_steps_per_second": 1.032, |
| "step": 1059 |
| }, |
| { |
| "epoch": 3.0, |
| "learning_rate": 7.774630154233555e-06, |
| "loss": 0.9938, |
| "step": 1060 |
| }, |
| { |
| "epoch": 3.03, |
| "learning_rate": 7.743153918791312e-06, |
| "loss": 0.9038, |
| "step": 1070 |
| }, |
| { |
| "epoch": 3.06, |
| "learning_rate": 7.711677683349073e-06, |
| "loss": 0.9207, |
| "step": 1080 |
| }, |
| { |
| "epoch": 3.09, |
| "learning_rate": 7.68020144790683e-06, |
| "loss": 0.889, |
| "step": 1090 |
| }, |
| { |
| "epoch": 3.12, |
| "learning_rate": 7.64872521246459e-06, |
| "loss": 0.831, |
| "step": 1100 |
| }, |
| { |
| "epoch": 3.14, |
| "learning_rate": 7.617248977022349e-06, |
| "loss": 0.8713, |
| "step": 1110 |
| }, |
| { |
| "epoch": 3.17, |
| "learning_rate": 7.585772741580108e-06, |
| "loss": 0.8606, |
| "step": 1120 |
| }, |
| { |
| "epoch": 3.2, |
| "learning_rate": 7.554296506137867e-06, |
| "loss": 0.8634, |
| "step": 1130 |
| }, |
| { |
| "epoch": 3.23, |
| "learning_rate": 7.522820270695625e-06, |
| "loss": 0.8653, |
| "step": 1140 |
| }, |
| { |
| "epoch": 3.26, |
| "learning_rate": 7.491344035253384e-06, |
| "loss": 0.8661, |
| "step": 1150 |
| }, |
| { |
| "epoch": 3.29, |
| "learning_rate": 7.4598677998111434e-06, |
| "loss": 0.8929, |
| "step": 1160 |
| }, |
| { |
| "epoch": 3.31, |
| "learning_rate": 7.428391564368902e-06, |
| "loss": 0.8978, |
| "step": 1170 |
| }, |
| { |
| "epoch": 3.34, |
| "learning_rate": 7.396915328926662e-06, |
| "loss": 0.8828, |
| "step": 1180 |
| }, |
| { |
| "epoch": 3.37, |
| "learning_rate": 7.36543909348442e-06, |
| "loss": 0.8668, |
| "step": 1190 |
| }, |
| { |
| "epoch": 3.4, |
| "learning_rate": 7.333962858042178e-06, |
| "loss": 0.8403, |
| "step": 1200 |
| }, |
| { |
| "epoch": 3.43, |
| "learning_rate": 7.302486622599938e-06, |
| "loss": 0.8483, |
| "step": 1210 |
| }, |
| { |
| "epoch": 3.46, |
| "learning_rate": 7.271010387157696e-06, |
| "loss": 0.7935, |
| "step": 1220 |
| }, |
| { |
| "epoch": 3.48, |
| "learning_rate": 7.239534151715455e-06, |
| "loss": 0.865, |
| "step": 1230 |
| }, |
| { |
| "epoch": 3.51, |
| "learning_rate": 7.208057916273215e-06, |
| "loss": 0.8383, |
| "step": 1240 |
| }, |
| { |
| "epoch": 3.54, |
| "learning_rate": 7.176581680830973e-06, |
| "loss": 0.8755, |
| "step": 1250 |
| }, |
| { |
| "epoch": 3.57, |
| "learning_rate": 7.145105445388733e-06, |
| "loss": 0.8768, |
| "step": 1260 |
| }, |
| { |
| "epoch": 3.6, |
| "learning_rate": 7.113629209946491e-06, |
| "loss": 0.8303, |
| "step": 1270 |
| }, |
| { |
| "epoch": 3.63, |
| "learning_rate": 7.082152974504249e-06, |
| "loss": 0.8629, |
| "step": 1280 |
| }, |
| { |
| "epoch": 3.65, |
| "learning_rate": 7.050676739062009e-06, |
| "loss": 0.8743, |
| "step": 1290 |
| }, |
| { |
| "epoch": 3.68, |
| "learning_rate": 7.0192005036197675e-06, |
| "loss": 0.8477, |
| "step": 1300 |
| }, |
| { |
| "epoch": 3.71, |
| "learning_rate": 6.987724268177527e-06, |
| "loss": 0.87, |
| "step": 1310 |
| }, |
| { |
| "epoch": 3.74, |
| "learning_rate": 6.956248032735286e-06, |
| "loss": 0.8687, |
| "step": 1320 |
| }, |
| { |
| "epoch": 3.77, |
| "learning_rate": 6.924771797293044e-06, |
| "loss": 0.8403, |
| "step": 1330 |
| }, |
| { |
| "epoch": 3.79, |
| "learning_rate": 6.893295561850803e-06, |
| "loss": 0.8619, |
| "step": 1340 |
| }, |
| { |
| "epoch": 3.82, |
| "learning_rate": 6.861819326408562e-06, |
| "loss": 0.8407, |
| "step": 1350 |
| }, |
| { |
| "epoch": 3.85, |
| "learning_rate": 6.830343090966321e-06, |
| "loss": 0.878, |
| "step": 1360 |
| }, |
| { |
| "epoch": 3.88, |
| "learning_rate": 6.79886685552408e-06, |
| "loss": 0.8613, |
| "step": 1370 |
| }, |
| { |
| "epoch": 3.91, |
| "learning_rate": 6.767390620081839e-06, |
| "loss": 0.8423, |
| "step": 1380 |
| }, |
| { |
| "epoch": 3.94, |
| "learning_rate": 6.735914384639598e-06, |
| "loss": 0.8278, |
| "step": 1390 |
| }, |
| { |
| "epoch": 3.96, |
| "learning_rate": 6.704438149197356e-06, |
| "loss": 0.8436, |
| "step": 1400 |
| }, |
| { |
| "epoch": 3.99, |
| "learning_rate": 6.672961913755115e-06, |
| "loss": 0.8282, |
| "step": 1410 |
| }, |
| { |
| "epoch": 4.0, |
| "eval_accuracy": 0.6828513310338729, |
| "eval_f1": 0.6743032487410993, |
| "eval_loss": 0.8873406648635864, |
| "eval_precision": 0.6798039204557976, |
| "eval_recall": 0.6828513310338729, |
| "eval_runtime": 346.7232, |
| "eval_samples_per_second": 32.611, |
| "eval_steps_per_second": 1.021, |
| "step": 1412 |
| }, |
| { |
| "epoch": 4.02, |
| "learning_rate": 6.641485678312874e-06, |
| "loss": 0.8709, |
| "step": 1420 |
| }, |
| { |
| "epoch": 4.05, |
| "learning_rate": 6.6100094428706325e-06, |
| "loss": 0.7873, |
| "step": 1430 |
| }, |
| { |
| "epoch": 4.08, |
| "learning_rate": 6.5785332074283925e-06, |
| "loss": 0.8373, |
| "step": 1440 |
| }, |
| { |
| "epoch": 4.11, |
| "learning_rate": 6.547056971986151e-06, |
| "loss": 0.8404, |
| "step": 1450 |
| }, |
| { |
| "epoch": 4.14, |
| "learning_rate": 6.51558073654391e-06, |
| "loss": 0.7976, |
| "step": 1460 |
| }, |
| { |
| "epoch": 4.16, |
| "learning_rate": 6.484104501101669e-06, |
| "loss": 0.8124, |
| "step": 1470 |
| }, |
| { |
| "epoch": 4.19, |
| "learning_rate": 6.452628265659427e-06, |
| "loss": 0.8127, |
| "step": 1480 |
| }, |
| { |
| "epoch": 4.22, |
| "learning_rate": 6.421152030217187e-06, |
| "loss": 0.7845, |
| "step": 1490 |
| }, |
| { |
| "epoch": 4.25, |
| "learning_rate": 6.3896757947749454e-06, |
| "loss": 0.785, |
| "step": 1500 |
| }, |
| { |
| "epoch": 4.28, |
| "learning_rate": 6.358199559332704e-06, |
| "loss": 0.8259, |
| "step": 1510 |
| }, |
| { |
| "epoch": 4.31, |
| "learning_rate": 6.326723323890464e-06, |
| "loss": 0.8127, |
| "step": 1520 |
| }, |
| { |
| "epoch": 4.33, |
| "learning_rate": 6.295247088448222e-06, |
| "loss": 0.8197, |
| "step": 1530 |
| }, |
| { |
| "epoch": 4.36, |
| "learning_rate": 6.263770853005982e-06, |
| "loss": 0.7851, |
| "step": 1540 |
| }, |
| { |
| "epoch": 4.39, |
| "learning_rate": 6.23229461756374e-06, |
| "loss": 0.8336, |
| "step": 1550 |
| }, |
| { |
| "epoch": 4.42, |
| "learning_rate": 6.200818382121498e-06, |
| "loss": 0.7848, |
| "step": 1560 |
| }, |
| { |
| "epoch": 4.45, |
| "learning_rate": 6.169342146679258e-06, |
| "loss": 0.8269, |
| "step": 1570 |
| }, |
| { |
| "epoch": 4.48, |
| "learning_rate": 6.137865911237017e-06, |
| "loss": 0.7919, |
| "step": 1580 |
| }, |
| { |
| "epoch": 4.5, |
| "learning_rate": 6.106389675794775e-06, |
| "loss": 0.7737, |
| "step": 1590 |
| }, |
| { |
| "epoch": 4.53, |
| "learning_rate": 6.074913440352535e-06, |
| "loss": 0.7451, |
| "step": 1600 |
| }, |
| { |
| "epoch": 4.56, |
| "learning_rate": 6.043437204910293e-06, |
| "loss": 0.8239, |
| "step": 1610 |
| }, |
| { |
| "epoch": 4.59, |
| "learning_rate": 6.011960969468052e-06, |
| "loss": 0.8159, |
| "step": 1620 |
| }, |
| { |
| "epoch": 4.62, |
| "learning_rate": 5.980484734025811e-06, |
| "loss": 0.8604, |
| "step": 1630 |
| }, |
| { |
| "epoch": 4.64, |
| "learning_rate": 5.9490084985835695e-06, |
| "loss": 0.7953, |
| "step": 1640 |
| }, |
| { |
| "epoch": 4.67, |
| "learning_rate": 5.917532263141329e-06, |
| "loss": 0.7837, |
| "step": 1650 |
| }, |
| { |
| "epoch": 4.7, |
| "learning_rate": 5.886056027699088e-06, |
| "loss": 0.8048, |
| "step": 1660 |
| }, |
| { |
| "epoch": 4.73, |
| "learning_rate": 5.854579792256847e-06, |
| "loss": 0.7574, |
| "step": 1670 |
| }, |
| { |
| "epoch": 4.76, |
| "learning_rate": 5.823103556814605e-06, |
| "loss": 0.751, |
| "step": 1680 |
| }, |
| { |
| "epoch": 4.79, |
| "learning_rate": 5.791627321372364e-06, |
| "loss": 0.8196, |
| "step": 1690 |
| }, |
| { |
| "epoch": 4.81, |
| "learning_rate": 5.760151085930123e-06, |
| "loss": 0.7627, |
| "step": 1700 |
| }, |
| { |
| "epoch": 4.84, |
| "learning_rate": 5.728674850487882e-06, |
| "loss": 0.8042, |
| "step": 1710 |
| }, |
| { |
| "epoch": 4.87, |
| "learning_rate": 5.6971986150456415e-06, |
| "loss": 0.7741, |
| "step": 1720 |
| }, |
| { |
| "epoch": 4.9, |
| "learning_rate": 5.6657223796034e-06, |
| "loss": 0.8194, |
| "step": 1730 |
| }, |
| { |
| "epoch": 4.93, |
| "learning_rate": 5.634246144161158e-06, |
| "loss": 0.8013, |
| "step": 1740 |
| }, |
| { |
| "epoch": 4.96, |
| "learning_rate": 5.602769908718918e-06, |
| "loss": 0.792, |
| "step": 1750 |
| }, |
| { |
| "epoch": 4.98, |
| "learning_rate": 5.571293673276676e-06, |
| "loss": 0.7717, |
| "step": 1760 |
| }, |
| { |
| "epoch": 5.0, |
| "eval_accuracy": 0.6884231007340585, |
| "eval_f1": 0.6835180422317746, |
| "eval_loss": 0.8611695766448975, |
| "eval_precision": 0.6887727666592812, |
| "eval_recall": 0.6884231007340585, |
| "eval_runtime": 343.597, |
| "eval_samples_per_second": 32.908, |
| "eval_steps_per_second": 1.03, |
| "step": 1765 |
| }, |
| { |
| "epoch": 5.01, |
| "learning_rate": 5.5398174378344345e-06, |
| "loss": 0.8065, |
| "step": 1770 |
| }, |
| { |
| "epoch": 5.04, |
| "learning_rate": 5.5083412023921945e-06, |
| "loss": 0.7874, |
| "step": 1780 |
| }, |
| { |
| "epoch": 5.07, |
| "learning_rate": 5.476864966949953e-06, |
| "loss": 0.7351, |
| "step": 1790 |
| }, |
| { |
| "epoch": 5.1, |
| "learning_rate": 5.445388731507713e-06, |
| "loss": 0.7611, |
| "step": 1800 |
| }, |
| { |
| "epoch": 5.13, |
| "learning_rate": 5.413912496065471e-06, |
| "loss": 0.7424, |
| "step": 1810 |
| }, |
| { |
| "epoch": 5.16, |
| "learning_rate": 5.382436260623229e-06, |
| "loss": 0.7288, |
| "step": 1820 |
| }, |
| { |
| "epoch": 5.18, |
| "learning_rate": 5.350960025180989e-06, |
| "loss": 0.776, |
| "step": 1830 |
| }, |
| { |
| "epoch": 5.21, |
| "learning_rate": 5.3194837897387474e-06, |
| "loss": 0.7324, |
| "step": 1840 |
| }, |
| { |
| "epoch": 5.24, |
| "learning_rate": 5.288007554296507e-06, |
| "loss": 0.7687, |
| "step": 1850 |
| }, |
| { |
| "epoch": 5.27, |
| "learning_rate": 5.256531318854266e-06, |
| "loss": 0.7938, |
| "step": 1860 |
| }, |
| { |
| "epoch": 5.3, |
| "learning_rate": 5.225055083412024e-06, |
| "loss": 0.742, |
| "step": 1870 |
| }, |
| { |
| "epoch": 5.33, |
| "learning_rate": 5.193578847969784e-06, |
| "loss": 0.7933, |
| "step": 1880 |
| }, |
| { |
| "epoch": 5.35, |
| "learning_rate": 5.162102612527542e-06, |
| "loss": 0.7356, |
| "step": 1890 |
| }, |
| { |
| "epoch": 5.38, |
| "learning_rate": 5.130626377085301e-06, |
| "loss": 0.7827, |
| "step": 1900 |
| }, |
| { |
| "epoch": 5.41, |
| "learning_rate": 5.09915014164306e-06, |
| "loss": 0.7322, |
| "step": 1910 |
| }, |
| { |
| "epoch": 5.44, |
| "learning_rate": 5.067673906200819e-06, |
| "loss": 0.7323, |
| "step": 1920 |
| }, |
| { |
| "epoch": 5.47, |
| "learning_rate": 5.036197670758578e-06, |
| "loss": 0.7896, |
| "step": 1930 |
| }, |
| { |
| "epoch": 5.5, |
| "learning_rate": 5.004721435316337e-06, |
| "loss": 0.7531, |
| "step": 1940 |
| }, |
| { |
| "epoch": 5.52, |
| "learning_rate": 4.973245199874096e-06, |
| "loss": 0.7156, |
| "step": 1950 |
| }, |
| { |
| "epoch": 5.55, |
| "learning_rate": 4.941768964431854e-06, |
| "loss": 0.7514, |
| "step": 1960 |
| }, |
| { |
| "epoch": 5.58, |
| "learning_rate": 4.910292728989613e-06, |
| "loss": 0.7274, |
| "step": 1970 |
| }, |
| { |
| "epoch": 5.61, |
| "learning_rate": 4.878816493547372e-06, |
| "loss": 0.7872, |
| "step": 1980 |
| }, |
| { |
| "epoch": 5.64, |
| "learning_rate": 4.847340258105131e-06, |
| "loss": 0.7255, |
| "step": 1990 |
| }, |
| { |
| "epoch": 5.66, |
| "learning_rate": 4.81586402266289e-06, |
| "loss": 0.7428, |
| "step": 2000 |
| }, |
| { |
| "epoch": 5.69, |
| "learning_rate": 4.784387787220649e-06, |
| "loss": 0.7371, |
| "step": 2010 |
| }, |
| { |
| "epoch": 5.72, |
| "learning_rate": 4.752911551778407e-06, |
| "loss": 0.7241, |
| "step": 2020 |
| }, |
| { |
| "epoch": 5.75, |
| "learning_rate": 4.721435316336166e-06, |
| "loss": 0.7763, |
| "step": 2030 |
| }, |
| { |
| "epoch": 5.78, |
| "learning_rate": 4.689959080893925e-06, |
| "loss": 0.7072, |
| "step": 2040 |
| }, |
| { |
| "epoch": 5.81, |
| "learning_rate": 4.6584828454516844e-06, |
| "loss": 0.753, |
| "step": 2050 |
| }, |
| { |
| "epoch": 5.83, |
| "learning_rate": 4.6270066100094435e-06, |
| "loss": 0.7878, |
| "step": 2060 |
| }, |
| { |
| "epoch": 5.86, |
| "learning_rate": 4.595530374567202e-06, |
| "loss": 0.7597, |
| "step": 2070 |
| }, |
| { |
| "epoch": 5.89, |
| "learning_rate": 4.564054139124961e-06, |
| "loss": 0.747, |
| "step": 2080 |
| }, |
| { |
| "epoch": 5.92, |
| "learning_rate": 4.53257790368272e-06, |
| "loss": 0.7393, |
| "step": 2090 |
| }, |
| { |
| "epoch": 5.95, |
| "learning_rate": 4.501101668240479e-06, |
| "loss": 0.7316, |
| "step": 2100 |
| }, |
| { |
| "epoch": 5.98, |
| "learning_rate": 4.469625432798237e-06, |
| "loss": 0.7678, |
| "step": 2110 |
| }, |
| { |
| "epoch": 6.0, |
| "eval_accuracy": 0.698505350667728, |
| "eval_f1": 0.6966177547304468, |
| "eval_loss": 0.8472527265548706, |
| "eval_precision": 0.6989184452261918, |
| "eval_recall": 0.698505350667728, |
| "eval_runtime": 344.0128, |
| "eval_samples_per_second": 32.868, |
| "eval_steps_per_second": 1.029, |
| "step": 2118 |
| }, |
| { |
| "epoch": 6.01, |
| "learning_rate": 4.4381491973559965e-06, |
| "loss": 0.7287, |
| "step": 2120 |
| }, |
| { |
| "epoch": 6.03, |
| "learning_rate": 4.406672961913756e-06, |
| "loss": 0.6859, |
| "step": 2130 |
| }, |
| { |
| "epoch": 6.06, |
| "learning_rate": 4.375196726471515e-06, |
| "loss": 0.7264, |
| "step": 2140 |
| }, |
| { |
| "epoch": 6.09, |
| "learning_rate": 4.343720491029274e-06, |
| "loss": 0.7404, |
| "step": 2150 |
| }, |
| { |
| "epoch": 6.12, |
| "learning_rate": 4.312244255587032e-06, |
| "loss": 0.7106, |
| "step": 2160 |
| }, |
| { |
| "epoch": 6.15, |
| "learning_rate": 4.280768020144791e-06, |
| "loss": 0.7143, |
| "step": 2170 |
| }, |
| { |
| "epoch": 6.18, |
| "learning_rate": 4.24929178470255e-06, |
| "loss": 0.7384, |
| "step": 2180 |
| }, |
| { |
| "epoch": 6.2, |
| "learning_rate": 4.217815549260309e-06, |
| "loss": 0.7204, |
| "step": 2190 |
| }, |
| { |
| "epoch": 6.23, |
| "learning_rate": 4.186339313818068e-06, |
| "loss": 0.7265, |
| "step": 2200 |
| }, |
| { |
| "epoch": 6.26, |
| "learning_rate": 4.154863078375827e-06, |
| "loss": 0.6953, |
| "step": 2210 |
| }, |
| { |
| "epoch": 6.29, |
| "learning_rate": 4.123386842933586e-06, |
| "loss": 0.6832, |
| "step": 2220 |
| }, |
| { |
| "epoch": 6.32, |
| "learning_rate": 4.091910607491344e-06, |
| "loss": 0.7231, |
| "step": 2230 |
| }, |
| { |
| "epoch": 6.35, |
| "learning_rate": 4.060434372049103e-06, |
| "loss": 0.7582, |
| "step": 2240 |
| }, |
| { |
| "epoch": 6.37, |
| "learning_rate": 4.028958136606862e-06, |
| "loss": 0.7243, |
| "step": 2250 |
| }, |
| { |
| "epoch": 6.4, |
| "learning_rate": 3.997481901164621e-06, |
| "loss": 0.7149, |
| "step": 2260 |
| }, |
| { |
| "epoch": 6.43, |
| "learning_rate": 3.96600566572238e-06, |
| "loss": 0.7101, |
| "step": 2270 |
| }, |
| { |
| "epoch": 6.46, |
| "learning_rate": 3.934529430280139e-06, |
| "loss": 0.6958, |
| "step": 2280 |
| }, |
| { |
| "epoch": 6.49, |
| "learning_rate": 3.903053194837897e-06, |
| "loss": 0.7126, |
| "step": 2290 |
| }, |
| { |
| "epoch": 6.51, |
| "learning_rate": 3.871576959395656e-06, |
| "loss": 0.6794, |
| "step": 2300 |
| }, |
| { |
| "epoch": 6.54, |
| "learning_rate": 3.840100723953415e-06, |
| "loss": 0.7204, |
| "step": 2310 |
| }, |
| { |
| "epoch": 6.57, |
| "learning_rate": 3.8086244885111744e-06, |
| "loss": 0.7066, |
| "step": 2320 |
| }, |
| { |
| "epoch": 6.6, |
| "learning_rate": 3.7771482530689335e-06, |
| "loss": 0.6775, |
| "step": 2330 |
| }, |
| { |
| "epoch": 6.63, |
| "learning_rate": 3.745672017626692e-06, |
| "loss": 0.736, |
| "step": 2340 |
| }, |
| { |
| "epoch": 6.66, |
| "learning_rate": 3.714195782184451e-06, |
| "loss": 0.6766, |
| "step": 2350 |
| }, |
| { |
| "epoch": 6.68, |
| "learning_rate": 3.68271954674221e-06, |
| "loss": 0.7436, |
| "step": 2360 |
| }, |
| { |
| "epoch": 6.71, |
| "learning_rate": 3.651243311299969e-06, |
| "loss": 0.7107, |
| "step": 2370 |
| }, |
| { |
| "epoch": 6.74, |
| "learning_rate": 3.6197670758577273e-06, |
| "loss": 0.7264, |
| "step": 2380 |
| }, |
| { |
| "epoch": 6.77, |
| "learning_rate": 3.5882908404154864e-06, |
| "loss": 0.7182, |
| "step": 2390 |
| }, |
| { |
| "epoch": 6.8, |
| "learning_rate": 3.5568146049732455e-06, |
| "loss": 0.7273, |
| "step": 2400 |
| }, |
| { |
| "epoch": 6.83, |
| "learning_rate": 3.5253383695310046e-06, |
| "loss": 0.7511, |
| "step": 2410 |
| }, |
| { |
| "epoch": 6.85, |
| "learning_rate": 3.4938621340887633e-06, |
| "loss": 0.6866, |
| "step": 2420 |
| }, |
| { |
| "epoch": 6.88, |
| "learning_rate": 3.462385898646522e-06, |
| "loss": 0.6801, |
| "step": 2430 |
| }, |
| { |
| "epoch": 6.91, |
| "learning_rate": 3.430909663204281e-06, |
| "loss": 0.7376, |
| "step": 2440 |
| }, |
| { |
| "epoch": 6.94, |
| "learning_rate": 3.39943342776204e-06, |
| "loss": 0.6749, |
| "step": 2450 |
| }, |
| { |
| "epoch": 6.97, |
| "learning_rate": 3.367957192319799e-06, |
| "loss": 0.6817, |
| "step": 2460 |
| }, |
| { |
| "epoch": 7.0, |
| "learning_rate": 3.3364809568775576e-06, |
| "loss": 0.7096, |
| "step": 2470 |
| }, |
| { |
| "epoch": 7.0, |
| "eval_accuracy": 0.7017776598567259, |
| "eval_f1": 0.6988712533287156, |
| "eval_loss": 0.8362669944763184, |
| "eval_precision": 0.7000656535425815, |
| "eval_recall": 0.7017776598567259, |
| "eval_runtime": 344.8245, |
| "eval_samples_per_second": 32.791, |
| "eval_steps_per_second": 1.027, |
| "step": 2471 |
| }, |
| { |
| "epoch": 7.03, |
| "learning_rate": 3.3050047214353163e-06, |
| "loss": 0.7373, |
| "step": 2480 |
| }, |
| { |
| "epoch": 7.05, |
| "learning_rate": 3.2735284859930754e-06, |
| "loss": 0.7116, |
| "step": 2490 |
| }, |
| { |
| "epoch": 7.08, |
| "learning_rate": 3.2420522505508345e-06, |
| "loss": 0.7172, |
| "step": 2500 |
| }, |
| { |
| "epoch": 7.11, |
| "learning_rate": 3.2105760151085936e-06, |
| "loss": 0.7007, |
| "step": 2510 |
| }, |
| { |
| "epoch": 7.14, |
| "learning_rate": 3.179099779666352e-06, |
| "loss": 0.6765, |
| "step": 2520 |
| }, |
| { |
| "epoch": 7.17, |
| "learning_rate": 3.147623544224111e-06, |
| "loss": 0.7166, |
| "step": 2530 |
| }, |
| { |
| "epoch": 7.2, |
| "learning_rate": 3.11614730878187e-06, |
| "loss": 0.7143, |
| "step": 2540 |
| }, |
| { |
| "epoch": 7.22, |
| "learning_rate": 3.084671073339629e-06, |
| "loss": 0.7011, |
| "step": 2550 |
| }, |
| { |
| "epoch": 7.25, |
| "learning_rate": 3.0531948378973874e-06, |
| "loss": 0.6641, |
| "step": 2560 |
| }, |
| { |
| "epoch": 7.28, |
| "learning_rate": 3.0217186024551465e-06, |
| "loss": 0.6599, |
| "step": 2570 |
| }, |
| { |
| "epoch": 7.31, |
| "learning_rate": 2.9902423670129056e-06, |
| "loss": 0.635, |
| "step": 2580 |
| }, |
| { |
| "epoch": 7.34, |
| "learning_rate": 2.9587661315706643e-06, |
| "loss": 0.6624, |
| "step": 2590 |
| }, |
| { |
| "epoch": 7.36, |
| "learning_rate": 2.9272898961284234e-06, |
| "loss": 0.709, |
| "step": 2600 |
| }, |
| { |
| "epoch": 7.39, |
| "learning_rate": 2.895813660686182e-06, |
| "loss": 0.666, |
| "step": 2610 |
| }, |
| { |
| "epoch": 7.42, |
| "learning_rate": 2.864337425243941e-06, |
| "loss": 0.6904, |
| "step": 2620 |
| }, |
| { |
| "epoch": 7.45, |
| "learning_rate": 2.8328611898017e-06, |
| "loss": 0.6696, |
| "step": 2630 |
| }, |
| { |
| "epoch": 7.48, |
| "learning_rate": 2.801384954359459e-06, |
| "loss": 0.7035, |
| "step": 2640 |
| }, |
| { |
| "epoch": 7.51, |
| "learning_rate": 2.7699087189172173e-06, |
| "loss": 0.669, |
| "step": 2650 |
| }, |
| { |
| "epoch": 7.53, |
| "learning_rate": 2.7384324834749764e-06, |
| "loss": 0.6821, |
| "step": 2660 |
| }, |
| { |
| "epoch": 7.56, |
| "learning_rate": 2.7069562480327355e-06, |
| "loss": 0.6913, |
| "step": 2670 |
| }, |
| { |
| "epoch": 7.59, |
| "learning_rate": 2.6754800125904946e-06, |
| "loss": 0.6461, |
| "step": 2680 |
| }, |
| { |
| "epoch": 7.62, |
| "learning_rate": 2.6440037771482537e-06, |
| "loss": 0.6753, |
| "step": 2690 |
| }, |
| { |
| "epoch": 7.65, |
| "learning_rate": 2.612527541706012e-06, |
| "loss": 0.6653, |
| "step": 2700 |
| }, |
| { |
| "epoch": 7.68, |
| "learning_rate": 2.581051306263771e-06, |
| "loss": 0.6927, |
| "step": 2710 |
| }, |
| { |
| "epoch": 7.7, |
| "learning_rate": 2.54957507082153e-06, |
| "loss": 0.6661, |
| "step": 2720 |
| }, |
| { |
| "epoch": 7.73, |
| "learning_rate": 2.518098835379289e-06, |
| "loss": 0.6691, |
| "step": 2730 |
| }, |
| { |
| "epoch": 7.76, |
| "learning_rate": 2.486622599937048e-06, |
| "loss": 0.6467, |
| "step": 2740 |
| }, |
| { |
| "epoch": 7.79, |
| "learning_rate": 2.4551463644948066e-06, |
| "loss": 0.6769, |
| "step": 2750 |
| }, |
| { |
| "epoch": 7.82, |
| "learning_rate": 2.4236701290525653e-06, |
| "loss": 0.6513, |
| "step": 2760 |
| }, |
| { |
| "epoch": 7.85, |
| "learning_rate": 2.3921938936103244e-06, |
| "loss": 0.6915, |
| "step": 2770 |
| }, |
| { |
| "epoch": 7.87, |
| "learning_rate": 2.360717658168083e-06, |
| "loss": 0.6643, |
| "step": 2780 |
| }, |
| { |
| "epoch": 7.9, |
| "learning_rate": 2.3292414227258422e-06, |
| "loss": 0.7081, |
| "step": 2790 |
| }, |
| { |
| "epoch": 7.93, |
| "learning_rate": 2.297765187283601e-06, |
| "loss": 0.6743, |
| "step": 2800 |
| }, |
| { |
| "epoch": 7.96, |
| "learning_rate": 2.26628895184136e-06, |
| "loss": 0.6687, |
| "step": 2810 |
| }, |
| { |
| "epoch": 7.99, |
| "learning_rate": 2.2348127163991187e-06, |
| "loss": 0.6803, |
| "step": 2820 |
| }, |
| { |
| "epoch": 8.0, |
| "eval_accuracy": 0.7036349164234545, |
| "eval_f1": 0.7019307239036613, |
| "eval_loss": 0.8332926630973816, |
| "eval_precision": 0.703637760158709, |
| "eval_recall": 0.7036349164234545, |
| "eval_runtime": 347.9791, |
| "eval_samples_per_second": 32.493, |
| "eval_steps_per_second": 1.017, |
| "step": 2824 |
| }, |
| { |
| "epoch": 8.02, |
| "learning_rate": 2.203336480956878e-06, |
| "loss": 0.7287, |
| "step": 2830 |
| }, |
| { |
| "epoch": 8.05, |
| "learning_rate": 2.171860245514637e-06, |
| "loss": 0.6467, |
| "step": 2840 |
| }, |
| { |
| "epoch": 8.07, |
| "learning_rate": 2.1403840100723956e-06, |
| "loss": 0.6718, |
| "step": 2850 |
| }, |
| { |
| "epoch": 8.1, |
| "learning_rate": 2.1089077746301547e-06, |
| "loss": 0.6039, |
| "step": 2860 |
| }, |
| { |
| "epoch": 8.13, |
| "learning_rate": 2.0774315391879134e-06, |
| "loss": 0.6294, |
| "step": 2870 |
| }, |
| { |
| "epoch": 8.16, |
| "learning_rate": 2.045955303745672e-06, |
| "loss": 0.639, |
| "step": 2880 |
| }, |
| { |
| "epoch": 8.19, |
| "learning_rate": 2.014479068303431e-06, |
| "loss": 0.6642, |
| "step": 2890 |
| }, |
| { |
| "epoch": 8.21, |
| "learning_rate": 1.98300283286119e-06, |
| "loss": 0.6429, |
| "step": 2900 |
| }, |
| { |
| "epoch": 8.24, |
| "learning_rate": 1.9515265974189485e-06, |
| "loss": 0.6578, |
| "step": 2910 |
| }, |
| { |
| "epoch": 8.27, |
| "learning_rate": 1.9200503619767076e-06, |
| "loss": 0.6357, |
| "step": 2920 |
| }, |
| { |
| "epoch": 8.3, |
| "learning_rate": 1.8885741265344667e-06, |
| "loss": 0.6641, |
| "step": 2930 |
| }, |
| { |
| "epoch": 8.33, |
| "learning_rate": 1.8570978910922254e-06, |
| "loss": 0.6736, |
| "step": 2940 |
| }, |
| { |
| "epoch": 8.36, |
| "learning_rate": 1.8256216556499845e-06, |
| "loss": 0.6217, |
| "step": 2950 |
| }, |
| { |
| "epoch": 8.38, |
| "learning_rate": 1.7941454202077432e-06, |
| "loss": 0.624, |
| "step": 2960 |
| }, |
| { |
| "epoch": 8.41, |
| "learning_rate": 1.7626691847655023e-06, |
| "loss": 0.7125, |
| "step": 2970 |
| }, |
| { |
| "epoch": 8.44, |
| "learning_rate": 1.731192949323261e-06, |
| "loss": 0.6653, |
| "step": 2980 |
| }, |
| { |
| "epoch": 8.47, |
| "learning_rate": 1.69971671388102e-06, |
| "loss": 0.6701, |
| "step": 2990 |
| }, |
| { |
| "epoch": 8.5, |
| "learning_rate": 1.6682404784387788e-06, |
| "loss": 0.6604, |
| "step": 3000 |
| }, |
| { |
| "epoch": 8.53, |
| "learning_rate": 1.6367642429965377e-06, |
| "loss": 0.6618, |
| "step": 3010 |
| }, |
| { |
| "epoch": 8.55, |
| "learning_rate": 1.6052880075542968e-06, |
| "loss": 0.6569, |
| "step": 3020 |
| }, |
| { |
| "epoch": 8.58, |
| "learning_rate": 1.5738117721120555e-06, |
| "loss": 0.6474, |
| "step": 3030 |
| }, |
| { |
| "epoch": 8.61, |
| "learning_rate": 1.5423355366698146e-06, |
| "loss": 0.6674, |
| "step": 3040 |
| }, |
| { |
| "epoch": 8.64, |
| "learning_rate": 1.5108593012275733e-06, |
| "loss": 0.6612, |
| "step": 3050 |
| }, |
| { |
| "epoch": 8.67, |
| "learning_rate": 1.4793830657853322e-06, |
| "loss": 0.7034, |
| "step": 3060 |
| }, |
| { |
| "epoch": 8.7, |
| "learning_rate": 1.447906830343091e-06, |
| "loss": 0.6367, |
| "step": 3070 |
| }, |
| { |
| "epoch": 8.72, |
| "learning_rate": 1.41643059490085e-06, |
| "loss": 0.6622, |
| "step": 3080 |
| }, |
| { |
| "epoch": 8.75, |
| "learning_rate": 1.3849543594586086e-06, |
| "loss": 0.6402, |
| "step": 3090 |
| }, |
| { |
| "epoch": 8.78, |
| "learning_rate": 1.3534781240163677e-06, |
| "loss": 0.6936, |
| "step": 3100 |
| }, |
| { |
| "epoch": 8.81, |
| "learning_rate": 1.3220018885741268e-06, |
| "loss": 0.6325, |
| "step": 3110 |
| }, |
| { |
| "epoch": 8.84, |
| "learning_rate": 1.2905256531318855e-06, |
| "loss": 0.6744, |
| "step": 3120 |
| }, |
| { |
| "epoch": 8.87, |
| "learning_rate": 1.2590494176896444e-06, |
| "loss": 0.6392, |
| "step": 3130 |
| }, |
| { |
| "epoch": 8.89, |
| "learning_rate": 1.2275731822474033e-06, |
| "loss": 0.6942, |
| "step": 3140 |
| }, |
| { |
| "epoch": 8.92, |
| "learning_rate": 1.1960969468051622e-06, |
| "loss": 0.6734, |
| "step": 3150 |
| }, |
| { |
| "epoch": 8.95, |
| "learning_rate": 1.1646207113629211e-06, |
| "loss": 0.646, |
| "step": 3160 |
| }, |
| { |
| "epoch": 8.98, |
| "learning_rate": 1.13314447592068e-06, |
| "loss": 0.6521, |
| "step": 3170 |
| }, |
| { |
| "epoch": 9.0, |
| "eval_accuracy": 0.7050499690457239, |
| "eval_f1": 0.7028178160001216, |
| "eval_loss": 0.8309338688850403, |
| "eval_precision": 0.7039313795096374, |
| "eval_recall": 0.7050499690457239, |
| "eval_runtime": 344.0588, |
| "eval_samples_per_second": 32.864, |
| "eval_steps_per_second": 1.029, |
| "step": 3177 |
| }, |
| { |
| "epoch": 9.01, |
| "learning_rate": 1.101668240478439e-06, |
| "loss": 0.6957, |
| "step": 3180 |
| }, |
| { |
| "epoch": 9.04, |
| "learning_rate": 1.0701920050361978e-06, |
| "loss": 0.6276, |
| "step": 3190 |
| }, |
| { |
| "epoch": 9.07, |
| "learning_rate": 1.0387157695939567e-06, |
| "loss": 0.6519, |
| "step": 3200 |
| }, |
| { |
| "epoch": 9.09, |
| "learning_rate": 1.0072395341517156e-06, |
| "loss": 0.6492, |
| "step": 3210 |
| }, |
| { |
| "epoch": 9.12, |
| "learning_rate": 9.757632987094743e-07, |
| "loss": 0.6355, |
| "step": 3220 |
| }, |
| { |
| "epoch": 9.15, |
| "learning_rate": 9.442870632672334e-07, |
| "loss": 0.686, |
| "step": 3230 |
| }, |
| { |
| "epoch": 9.18, |
| "learning_rate": 9.128108278249923e-07, |
| "loss": 0.6562, |
| "step": 3240 |
| }, |
| { |
| "epoch": 9.21, |
| "learning_rate": 8.813345923827512e-07, |
| "loss": 0.6595, |
| "step": 3250 |
| }, |
| { |
| "epoch": 9.23, |
| "learning_rate": 8.4985835694051e-07, |
| "loss": 0.6408, |
| "step": 3260 |
| }, |
| { |
| "epoch": 9.26, |
| "learning_rate": 8.183821214982688e-07, |
| "loss": 0.6152, |
| "step": 3270 |
| }, |
| { |
| "epoch": 9.29, |
| "learning_rate": 7.869058860560277e-07, |
| "loss": 0.6425, |
| "step": 3280 |
| }, |
| { |
| "epoch": 9.32, |
| "learning_rate": 7.554296506137866e-07, |
| "loss": 0.6303, |
| "step": 3290 |
| }, |
| { |
| "epoch": 9.35, |
| "learning_rate": 7.239534151715455e-07, |
| "loss": 0.6264, |
| "step": 3300 |
| }, |
| { |
| "epoch": 9.38, |
| "learning_rate": 6.924771797293043e-07, |
| "loss": 0.636, |
| "step": 3310 |
| }, |
| { |
| "epoch": 9.4, |
| "learning_rate": 6.610009442870634e-07, |
| "loss": 0.649, |
| "step": 3320 |
| }, |
| { |
| "epoch": 9.43, |
| "learning_rate": 6.295247088448222e-07, |
| "loss": 0.6765, |
| "step": 3330 |
| }, |
| { |
| "epoch": 9.46, |
| "learning_rate": 5.980484734025811e-07, |
| "loss": 0.653, |
| "step": 3340 |
| }, |
| { |
| "epoch": 9.49, |
| "learning_rate": 5.6657223796034e-07, |
| "loss": 0.6591, |
| "step": 3350 |
| }, |
| { |
| "epoch": 9.52, |
| "learning_rate": 5.350960025180989e-07, |
| "loss": 0.6361, |
| "step": 3360 |
| }, |
| { |
| "epoch": 9.55, |
| "learning_rate": 5.036197670758578e-07, |
| "loss": 0.6508, |
| "step": 3370 |
| }, |
| { |
| "epoch": 9.57, |
| "learning_rate": 4.721435316336167e-07, |
| "loss": 0.6659, |
| "step": 3380 |
| }, |
| { |
| "epoch": 9.6, |
| "learning_rate": 4.406672961913756e-07, |
| "loss": 0.6058, |
| "step": 3390 |
| }, |
| { |
| "epoch": 9.63, |
| "learning_rate": 4.091910607491344e-07, |
| "loss": 0.6354, |
| "step": 3400 |
| }, |
| { |
| "epoch": 9.66, |
| "learning_rate": 3.777148253068933e-07, |
| "loss": 0.6355, |
| "step": 3410 |
| }, |
| { |
| "epoch": 9.69, |
| "learning_rate": 3.4623858986465216e-07, |
| "loss": 0.6074, |
| "step": 3420 |
| }, |
| { |
| "epoch": 9.72, |
| "learning_rate": 3.147623544224111e-07, |
| "loss": 0.6205, |
| "step": 3430 |
| }, |
| { |
| "epoch": 9.74, |
| "learning_rate": 2.8328611898017e-07, |
| "loss": 0.6658, |
| "step": 3440 |
| }, |
| { |
| "epoch": 9.77, |
| "learning_rate": 2.518098835379289e-07, |
| "loss": 0.6252, |
| "step": 3450 |
| }, |
| { |
| "epoch": 9.8, |
| "learning_rate": 2.203336480956878e-07, |
| "loss": 0.6479, |
| "step": 3460 |
| }, |
| { |
| "epoch": 9.83, |
| "learning_rate": 1.8885741265344666e-07, |
| "loss": 0.6563, |
| "step": 3470 |
| }, |
| { |
| "epoch": 9.86, |
| "learning_rate": 1.5738117721120555e-07, |
| "loss": 0.6521, |
| "step": 3480 |
| }, |
| { |
| "epoch": 9.89, |
| "learning_rate": 1.2590494176896445e-07, |
| "loss": 0.6251, |
| "step": 3490 |
| }, |
| { |
| "epoch": 9.91, |
| "learning_rate": 9.442870632672333e-08, |
| "loss": 0.6095, |
| "step": 3500 |
| }, |
| { |
| "epoch": 9.94, |
| "learning_rate": 6.295247088448222e-08, |
| "loss": 0.6169, |
| "step": 3510 |
| }, |
| { |
| "epoch": 9.97, |
| "learning_rate": 3.147623544224111e-08, |
| "loss": 0.6726, |
| "step": 3520 |
| }, |
| { |
| "epoch": 10.0, |
| "learning_rate": 0.0, |
| "loss": 0.6671, |
| "step": 3530 |
| }, |
| { |
| "epoch": 10.0, |
| "eval_accuracy": 0.7052268506235075, |
| "eval_f1": 0.7036260157126459, |
| "eval_loss": 0.829365074634552, |
| "eval_precision": 0.7048074435355876, |
| "eval_recall": 0.7052268506235075, |
| "eval_runtime": 345.8224, |
| "eval_samples_per_second": 32.696, |
| "eval_steps_per_second": 1.024, |
| "step": 3530 |
| }, |
| { |
| "epoch": 10.0, |
| "step": 3530, |
| "total_flos": 3.5044145509209285e+19, |
| "train_loss": 0.8871098300890612, |
| "train_runtime": 28088.7735, |
| "train_samples_per_second": 16.101, |
| "train_steps_per_second": 0.126 |
| } |
| ], |
| "max_steps": 3530, |
| "num_train_epochs": 10, |
| "total_flos": 3.5044145509209285e+19, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
