| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.991437584497522, | |
| "eval_steps": 693, | |
| "global_step": 2770, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0, | |
| "eval_f1": 0.14772727272727273, | |
| "eval_loss": 0.8850772380828857, | |
| "eval_precision": 0.08080808080808081, | |
| "eval_recall": 0.859504132231405, | |
| "eval_runtime": 99.3343, | |
| "eval_samples_per_second": 47.043, | |
| "eval_steps_per_second": 5.889, | |
| "step": 0 | |
| }, | |
| { | |
| "epoch": 0.018026137899954935, | |
| "grad_norm": 6.000438690185547, | |
| "learning_rate": 4.981949458483755e-05, | |
| "loss": 0.3931, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03605227579990987, | |
| "grad_norm": 47.497467041015625, | |
| "learning_rate": 4.963898916967509e-05, | |
| "loss": 0.2607, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.054078413699864804, | |
| "grad_norm": 13.464141845703125, | |
| "learning_rate": 4.945848375451264e-05, | |
| "loss": 0.1813, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.07210455159981974, | |
| "grad_norm": 15.57222843170166, | |
| "learning_rate": 4.927797833935018e-05, | |
| "loss": 0.2618, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.09013068949977468, | |
| "grad_norm": 11.892234802246094, | |
| "learning_rate": 4.909747292418773e-05, | |
| "loss": 0.1559, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.10815682739972961, | |
| "grad_norm": 6.340595722198486, | |
| "learning_rate": 4.891696750902527e-05, | |
| "loss": 0.1648, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.12618296529968454, | |
| "grad_norm": 10.196683883666992, | |
| "learning_rate": 4.873646209386282e-05, | |
| "loss": 0.1472, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.14420910319963948, | |
| "grad_norm": 11.511515617370605, | |
| "learning_rate": 4.855595667870036e-05, | |
| "loss": 0.2108, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.16223524109959442, | |
| "grad_norm": 8.681073188781738, | |
| "learning_rate": 4.837545126353791e-05, | |
| "loss": 0.1561, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.18026137899954936, | |
| "grad_norm": 41.062801361083984, | |
| "learning_rate": 4.819494584837546e-05, | |
| "loss": 0.1563, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.19828751689950427, | |
| "grad_norm": 19.93891716003418, | |
| "learning_rate": 4.8014440433213e-05, | |
| "loss": 0.1881, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.21631365479945922, | |
| "grad_norm": 21.90877914428711, | |
| "learning_rate": 4.783393501805055e-05, | |
| "loss": 0.1262, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.23433979269941416, | |
| "grad_norm": 14.123740196228027, | |
| "learning_rate": 4.765342960288809e-05, | |
| "loss": 0.1535, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.25236593059936907, | |
| "grad_norm": 30.671157836914062, | |
| "learning_rate": 4.747292418772563e-05, | |
| "loss": 0.1458, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.270392068499324, | |
| "grad_norm": 7.560745716094971, | |
| "learning_rate": 4.7292418772563177e-05, | |
| "loss": 0.1797, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.28841820639927895, | |
| "grad_norm": 12.448927879333496, | |
| "learning_rate": 4.711191335740072e-05, | |
| "loss": 0.1898, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.3064443442992339, | |
| "grad_norm": 20.040008544921875, | |
| "learning_rate": 4.693140794223827e-05, | |
| "loss": 0.1698, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.32447048219918884, | |
| "grad_norm": 7.592446804046631, | |
| "learning_rate": 4.675090252707581e-05, | |
| "loss": 0.156, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.3424966200991438, | |
| "grad_norm": 18.830904006958008, | |
| "learning_rate": 4.657039711191336e-05, | |
| "loss": 0.1536, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.3605227579990987, | |
| "grad_norm": 3.684938430786133, | |
| "learning_rate": 4.63898916967509e-05, | |
| "loss": 0.1152, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.3785488958990536, | |
| "grad_norm": 13.293309211730957, | |
| "learning_rate": 4.620938628158845e-05, | |
| "loss": 0.2116, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.39657503379900855, | |
| "grad_norm": 14.630786895751953, | |
| "learning_rate": 4.602888086642599e-05, | |
| "loss": 0.1549, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.4146011716989635, | |
| "grad_norm": 20.37642478942871, | |
| "learning_rate": 4.584837545126354e-05, | |
| "loss": 0.1679, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.43262730959891843, | |
| "grad_norm": 13.183340072631836, | |
| "learning_rate": 4.566787003610109e-05, | |
| "loss": 0.1581, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.45065344749887337, | |
| "grad_norm": 21.777263641357422, | |
| "learning_rate": 4.548736462093863e-05, | |
| "loss": 0.1618, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.4686795853988283, | |
| "grad_norm": 20.683944702148438, | |
| "learning_rate": 4.530685920577618e-05, | |
| "loss": 0.1606, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.48670572329878325, | |
| "grad_norm": 30.506752014160156, | |
| "learning_rate": 4.5126353790613716e-05, | |
| "loss": 0.1831, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.5047318611987381, | |
| "grad_norm": 22.953815460205078, | |
| "learning_rate": 4.494584837545127e-05, | |
| "loss": 0.1565, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.5227579990986931, | |
| "grad_norm": 6.177365303039551, | |
| "learning_rate": 4.4765342960288806e-05, | |
| "loss": 0.1287, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.540784136998648, | |
| "grad_norm": 8.373098373413086, | |
| "learning_rate": 4.458483754512636e-05, | |
| "loss": 0.1553, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.558810274898603, | |
| "grad_norm": 10.07934284210205, | |
| "learning_rate": 4.44043321299639e-05, | |
| "loss": 0.1868, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.5768364127985579, | |
| "grad_norm": 18.686702728271484, | |
| "learning_rate": 4.422382671480145e-05, | |
| "loss": 0.1576, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.5948625506985128, | |
| "grad_norm": 10.679413795471191, | |
| "learning_rate": 4.404332129963899e-05, | |
| "loss": 0.1409, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.6128886885984678, | |
| "grad_norm": 11.381563186645508, | |
| "learning_rate": 4.386281588447654e-05, | |
| "loss": 0.2043, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.6309148264984227, | |
| "grad_norm": 40.68292236328125, | |
| "learning_rate": 4.368231046931408e-05, | |
| "loss": 0.1288, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.6489409643983777, | |
| "grad_norm": 17.46027946472168, | |
| "learning_rate": 4.350180505415163e-05, | |
| "loss": 0.1537, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.6669671022983326, | |
| "grad_norm": 13.656320571899414, | |
| "learning_rate": 4.332129963898917e-05, | |
| "loss": 0.1201, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.6849932401982876, | |
| "grad_norm": 20.412939071655273, | |
| "learning_rate": 4.314079422382672e-05, | |
| "loss": 0.1321, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.7030193780982424, | |
| "grad_norm": 9.957433700561523, | |
| "learning_rate": 4.296028880866426e-05, | |
| "loss": 0.141, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.7210455159981974, | |
| "grad_norm": 9.682558059692383, | |
| "learning_rate": 4.277978339350181e-05, | |
| "loss": 0.1607, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.7390716538981523, | |
| "grad_norm": 9.733131408691406, | |
| "learning_rate": 4.259927797833935e-05, | |
| "loss": 0.1037, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.7570977917981072, | |
| "grad_norm": 15.76230525970459, | |
| "learning_rate": 4.24187725631769e-05, | |
| "loss": 0.2101, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.7751239296980622, | |
| "grad_norm": 22.166545867919922, | |
| "learning_rate": 4.223826714801444e-05, | |
| "loss": 0.1459, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.7931500675980171, | |
| "grad_norm": 17.87278175354004, | |
| "learning_rate": 4.205776173285199e-05, | |
| "loss": 0.147, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.8111762054979721, | |
| "grad_norm": 5.201169490814209, | |
| "learning_rate": 4.187725631768953e-05, | |
| "loss": 0.1247, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.829202343397927, | |
| "grad_norm": 8.68251895904541, | |
| "learning_rate": 4.169675090252708e-05, | |
| "loss": 0.1172, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.847228481297882, | |
| "grad_norm": 14.086528778076172, | |
| "learning_rate": 4.151624548736462e-05, | |
| "loss": 0.111, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.8652546191978369, | |
| "grad_norm": 5.421584606170654, | |
| "learning_rate": 4.1335740072202167e-05, | |
| "loss": 0.1264, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.8832807570977917, | |
| "grad_norm": 7.700821399688721, | |
| "learning_rate": 4.115523465703972e-05, | |
| "loss": 0.1278, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.9013068949977467, | |
| "grad_norm": 9.820999145507812, | |
| "learning_rate": 4.0974729241877256e-05, | |
| "loss": 0.1294, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.9193330328977016, | |
| "grad_norm": 15.016349792480469, | |
| "learning_rate": 4.079422382671481e-05, | |
| "loss": 0.1607, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.9373591707976566, | |
| "grad_norm": 10.43041706085205, | |
| "learning_rate": 4.0613718411552346e-05, | |
| "loss": 0.1091, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.9553853086976115, | |
| "grad_norm": 26.99540901184082, | |
| "learning_rate": 4.043321299638989e-05, | |
| "loss": 0.1442, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.9734114465975665, | |
| "grad_norm": 8.318906784057617, | |
| "learning_rate": 4.0252707581227436e-05, | |
| "loss": 0.1233, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.9914375844975214, | |
| "grad_norm": 8.85923957824707, | |
| "learning_rate": 4.007220216606498e-05, | |
| "loss": 0.1466, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.0090130689499774, | |
| "grad_norm": 4.69356107711792, | |
| "learning_rate": 3.989169675090253e-05, | |
| "loss": 0.1433, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.0270392068499323, | |
| "grad_norm": 17.66107177734375, | |
| "learning_rate": 3.971119133574007e-05, | |
| "loss": 0.0944, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.0450653447498874, | |
| "grad_norm": 4.721229553222656, | |
| "learning_rate": 3.953068592057762e-05, | |
| "loss": 0.1372, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.0630914826498423, | |
| "grad_norm": 6.2178955078125, | |
| "learning_rate": 3.935018050541516e-05, | |
| "loss": 0.1412, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.0811176205497972, | |
| "grad_norm": 6.32495641708374, | |
| "learning_rate": 3.916967509025271e-05, | |
| "loss": 0.0892, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.099143758449752, | |
| "grad_norm": 12.221674919128418, | |
| "learning_rate": 3.898916967509025e-05, | |
| "loss": 0.1333, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.117169896349707, | |
| "grad_norm": 1.7088432312011719, | |
| "learning_rate": 3.88086642599278e-05, | |
| "loss": 0.1077, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.135196034249662, | |
| "grad_norm": 17.136337280273438, | |
| "learning_rate": 3.862815884476535e-05, | |
| "loss": 0.1381, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.153222172149617, | |
| "grad_norm": 5.424814224243164, | |
| "learning_rate": 3.844765342960289e-05, | |
| "loss": 0.1228, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.1712483100495719, | |
| "grad_norm": 18.775466918945312, | |
| "learning_rate": 3.826714801444044e-05, | |
| "loss": 0.099, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.1892744479495267, | |
| "grad_norm": 8.731025695800781, | |
| "learning_rate": 3.8086642599277976e-05, | |
| "loss": 0.1324, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.2073005858494819, | |
| "grad_norm": 9.53122615814209, | |
| "learning_rate": 3.790613718411553e-05, | |
| "loss": 0.1085, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.2253267237494367, | |
| "grad_norm": 17.20035743713379, | |
| "learning_rate": 3.7725631768953066e-05, | |
| "loss": 0.0774, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.2433528616493916, | |
| "grad_norm": 18.292253494262695, | |
| "learning_rate": 3.754512635379062e-05, | |
| "loss": 0.0965, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.2487607030193781, | |
| "eval_f1": 0.7416879795396419, | |
| "eval_loss": 0.10558413714170456, | |
| "eval_precision": 0.6921241050119332, | |
| "eval_recall": 0.7988980716253443, | |
| "eval_runtime": 115.001, | |
| "eval_samples_per_second": 40.634, | |
| "eval_steps_per_second": 5.087, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 1.2613789995493465, | |
| "grad_norm": 6.922005653381348, | |
| "learning_rate": 3.7364620938628155e-05, | |
| "loss": 0.0906, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.2794051374493014, | |
| "grad_norm": 33.214107513427734, | |
| "learning_rate": 3.718411552346571e-05, | |
| "loss": 0.072, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.2974312753492563, | |
| "grad_norm": 9.25637149810791, | |
| "learning_rate": 3.700361010830325e-05, | |
| "loss": 0.1597, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.3154574132492114, | |
| "grad_norm": 10.336318016052246, | |
| "learning_rate": 3.68231046931408e-05, | |
| "loss": 0.1063, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.3334835511491663, | |
| "grad_norm": 11.632000923156738, | |
| "learning_rate": 3.664259927797834e-05, | |
| "loss": 0.1175, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.3515096890491212, | |
| "grad_norm": 12.909192085266113, | |
| "learning_rate": 3.646209386281589e-05, | |
| "loss": 0.1285, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.3695358269490763, | |
| "grad_norm": 7.545075416564941, | |
| "learning_rate": 3.628158844765343e-05, | |
| "loss": 0.0864, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.3875619648490312, | |
| "grad_norm": 25.248464584350586, | |
| "learning_rate": 3.610108303249098e-05, | |
| "loss": 0.0769, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.405588102748986, | |
| "grad_norm": 10.5347318649292, | |
| "learning_rate": 3.592057761732852e-05, | |
| "loss": 0.088, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.423614240648941, | |
| "grad_norm": 13.101734161376953, | |
| "learning_rate": 3.574007220216607e-05, | |
| "loss": 0.1084, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.4416403785488958, | |
| "grad_norm": 12.188843727111816, | |
| "learning_rate": 3.555956678700361e-05, | |
| "loss": 0.1101, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.4596665164488507, | |
| "grad_norm": 11.552268981933594, | |
| "learning_rate": 3.537906137184116e-05, | |
| "loss": 0.0785, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.4776926543488058, | |
| "grad_norm": 10.269721031188965, | |
| "learning_rate": 3.51985559566787e-05, | |
| "loss": 0.1161, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.4957187922487607, | |
| "grad_norm": 12.09333610534668, | |
| "learning_rate": 3.5018050541516247e-05, | |
| "loss": 0.1113, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.5137449301487156, | |
| "grad_norm": 5.4690141677856445, | |
| "learning_rate": 3.483754512635379e-05, | |
| "loss": 0.1046, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.5317710680486707, | |
| "grad_norm": 6.296072959899902, | |
| "learning_rate": 3.4657039711191336e-05, | |
| "loss": 0.1202, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.5497972059486256, | |
| "grad_norm": 8.854303359985352, | |
| "learning_rate": 3.447653429602888e-05, | |
| "loss": 0.0929, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.5678233438485805, | |
| "grad_norm": 23.987024307250977, | |
| "learning_rate": 3.4296028880866426e-05, | |
| "loss": 0.1326, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.5858494817485354, | |
| "grad_norm": 7.229393005371094, | |
| "learning_rate": 3.411552346570397e-05, | |
| "loss": 0.1169, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.6038756196484902, | |
| "grad_norm": 5.463610649108887, | |
| "learning_rate": 3.3935018050541516e-05, | |
| "loss": 0.0809, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.6219017575484451, | |
| "grad_norm": 5.9659624099731445, | |
| "learning_rate": 3.375451263537907e-05, | |
| "loss": 0.1265, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.6399278954484002, | |
| "grad_norm": 17.582469940185547, | |
| "learning_rate": 3.3574007220216606e-05, | |
| "loss": 0.0846, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.6579540333483551, | |
| "grad_norm": 13.633155822753906, | |
| "learning_rate": 3.339350180505416e-05, | |
| "loss": 0.0999, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.67598017124831, | |
| "grad_norm": 14.496593475341797, | |
| "learning_rate": 3.3212996389891696e-05, | |
| "loss": 0.0903, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.694006309148265, | |
| "grad_norm": 7.594356536865234, | |
| "learning_rate": 3.303249097472924e-05, | |
| "loss": 0.1021, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.71203244704822, | |
| "grad_norm": 20.628849029541016, | |
| "learning_rate": 3.2851985559566786e-05, | |
| "loss": 0.1131, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.7300585849481749, | |
| "grad_norm": 8.25106430053711, | |
| "learning_rate": 3.267148014440433e-05, | |
| "loss": 0.0872, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.7480847228481298, | |
| "grad_norm": 17.08070182800293, | |
| "learning_rate": 3.249097472924188e-05, | |
| "loss": 0.0663, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.7661108607480847, | |
| "grad_norm": 38.33162307739258, | |
| "learning_rate": 3.231046931407942e-05, | |
| "loss": 0.142, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.7841369986480395, | |
| "grad_norm": 17.89957618713379, | |
| "learning_rate": 3.212996389891697e-05, | |
| "loss": 0.0942, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.8021631365479944, | |
| "grad_norm": 9.691226959228516, | |
| "learning_rate": 3.194945848375451e-05, | |
| "loss": 0.0902, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.8201892744479495, | |
| "grad_norm": 9.764283180236816, | |
| "learning_rate": 3.176895306859206e-05, | |
| "loss": 0.1108, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.8382154123479044, | |
| "grad_norm": 13.829035758972168, | |
| "learning_rate": 3.15884476534296e-05, | |
| "loss": 0.0881, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.8562415502478595, | |
| "grad_norm": 8.372889518737793, | |
| "learning_rate": 3.140794223826715e-05, | |
| "loss": 0.1272, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.8742676881478144, | |
| "grad_norm": 12.576041221618652, | |
| "learning_rate": 3.12274368231047e-05, | |
| "loss": 0.1076, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.8922938260477693, | |
| "grad_norm": 12.995355606079102, | |
| "learning_rate": 3.104693140794224e-05, | |
| "loss": 0.0932, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.9103199639477242, | |
| "grad_norm": 6.626877784729004, | |
| "learning_rate": 3.086642599277979e-05, | |
| "loss": 0.105, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.928346101847679, | |
| "grad_norm": 11.209733009338379, | |
| "learning_rate": 3.0685920577617325e-05, | |
| "loss": 0.1331, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.946372239747634, | |
| "grad_norm": 7.832246780395508, | |
| "learning_rate": 3.0505415162454877e-05, | |
| "loss": 0.087, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.9643983776475888, | |
| "grad_norm": 9.489190101623535, | |
| "learning_rate": 3.032490974729242e-05, | |
| "loss": 0.0904, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.982424515547544, | |
| "grad_norm": 14.940125465393066, | |
| "learning_rate": 3.0144404332129967e-05, | |
| "loss": 0.0959, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 8.826193809509277, | |
| "learning_rate": 2.996389891696751e-05, | |
| "loss": 0.1276, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 2.018026137899955, | |
| "grad_norm": 4.580510139465332, | |
| "learning_rate": 2.9783393501805057e-05, | |
| "loss": 0.0492, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 2.0360522757999098, | |
| "grad_norm": 16.128124237060547, | |
| "learning_rate": 2.9602888086642598e-05, | |
| "loss": 0.064, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 2.0540784136998647, | |
| "grad_norm": 6.763059139251709, | |
| "learning_rate": 2.9422382671480147e-05, | |
| "loss": 0.076, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 2.0721045515998195, | |
| "grad_norm": 10.386236190795898, | |
| "learning_rate": 2.924187725631769e-05, | |
| "loss": 0.0776, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.090130689499775, | |
| "grad_norm": 8.901439666748047, | |
| "learning_rate": 2.906137184115524e-05, | |
| "loss": 0.0972, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 2.1081568273997298, | |
| "grad_norm": 8.78127384185791, | |
| "learning_rate": 2.888086642599278e-05, | |
| "loss": 0.1042, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 2.1261829652996846, | |
| "grad_norm": 4.073227405548096, | |
| "learning_rate": 2.870036101083033e-05, | |
| "loss": 0.059, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 2.1442091031996395, | |
| "grad_norm": 22.62556266784668, | |
| "learning_rate": 2.851985559566787e-05, | |
| "loss": 0.0736, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 2.1622352410995944, | |
| "grad_norm": 3.826328754425049, | |
| "learning_rate": 2.8339350180505413e-05, | |
| "loss": 0.0513, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.1802613789995493, | |
| "grad_norm": 9.261476516723633, | |
| "learning_rate": 2.815884476534296e-05, | |
| "loss": 0.0604, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 2.198287516899504, | |
| "grad_norm": 18.781217575073242, | |
| "learning_rate": 2.7978339350180506e-05, | |
| "loss": 0.0622, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 2.216313654799459, | |
| "grad_norm": 23.580307006835938, | |
| "learning_rate": 2.779783393501805e-05, | |
| "loss": 0.0652, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 2.234339792699414, | |
| "grad_norm": 15.79100227355957, | |
| "learning_rate": 2.7617328519855596e-05, | |
| "loss": 0.0676, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 2.2523659305993693, | |
| "grad_norm": 1.7981843948364258, | |
| "learning_rate": 2.7436823104693144e-05, | |
| "loss": 0.0671, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.270392068499324, | |
| "grad_norm": 26.794170379638672, | |
| "learning_rate": 2.7256317689530686e-05, | |
| "loss": 0.0825, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 2.288418206399279, | |
| "grad_norm": 21.756696701049805, | |
| "learning_rate": 2.7075812274368234e-05, | |
| "loss": 0.0518, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 2.306444344299234, | |
| "grad_norm": 5.695476055145264, | |
| "learning_rate": 2.6895306859205776e-05, | |
| "loss": 0.0642, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 2.324470482199189, | |
| "grad_norm": 15.769753456115723, | |
| "learning_rate": 2.6714801444043324e-05, | |
| "loss": 0.1, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 2.3424966200991437, | |
| "grad_norm": 3.572789192199707, | |
| "learning_rate": 2.6534296028880866e-05, | |
| "loss": 0.0516, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.3605227579990986, | |
| "grad_norm": 16.55797004699707, | |
| "learning_rate": 2.6353790613718414e-05, | |
| "loss": 0.0676, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 2.3785488958990535, | |
| "grad_norm": 5.78305196762085, | |
| "learning_rate": 2.617328519855596e-05, | |
| "loss": 0.0713, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 2.3965750337990084, | |
| "grad_norm": 15.535674095153809, | |
| "learning_rate": 2.59927797833935e-05, | |
| "loss": 0.0756, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 2.4146011716989637, | |
| "grad_norm": 7.25338888168335, | |
| "learning_rate": 2.581227436823105e-05, | |
| "loss": 0.0658, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 2.4326273095989186, | |
| "grad_norm": 7.954975605010986, | |
| "learning_rate": 2.563176895306859e-05, | |
| "loss": 0.062, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.4506534474988735, | |
| "grad_norm": 16.88165855407715, | |
| "learning_rate": 2.545126353790614e-05, | |
| "loss": 0.0428, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 2.4686795853988284, | |
| "grad_norm": 5.712924003601074, | |
| "learning_rate": 2.527075812274368e-05, | |
| "loss": 0.0594, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 2.4867057232987833, | |
| "grad_norm": 26.46906280517578, | |
| "learning_rate": 2.509025270758123e-05, | |
| "loss": 0.1073, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 2.4975214060387563, | |
| "eval_f1": 0.7364532019704434, | |
| "eval_loss": 0.11697933077812195, | |
| "eval_precision": 0.6659242761692651, | |
| "eval_recall": 0.8236914600550964, | |
| "eval_runtime": 100.7721, | |
| "eval_samples_per_second": 46.372, | |
| "eval_steps_per_second": 5.805, | |
| "step": 1386 | |
| }, | |
| { | |
| "epoch": 2.504731861198738, | |
| "grad_norm": 1.9493204355239868, | |
| "learning_rate": 2.4909747292418774e-05, | |
| "loss": 0.0622, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 2.522757999098693, | |
| "grad_norm": 4.658102989196777, | |
| "learning_rate": 2.472924187725632e-05, | |
| "loss": 0.0843, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.540784136998648, | |
| "grad_norm": 24.372589111328125, | |
| "learning_rate": 2.4548736462093864e-05, | |
| "loss": 0.0555, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 2.558810274898603, | |
| "grad_norm": 35.2315673828125, | |
| "learning_rate": 2.436823104693141e-05, | |
| "loss": 0.0752, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 2.576836412798558, | |
| "grad_norm": 8.754782676696777, | |
| "learning_rate": 2.4187725631768953e-05, | |
| "loss": 0.0817, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 2.5948625506985126, | |
| "grad_norm": 20.37566566467285, | |
| "learning_rate": 2.40072202166065e-05, | |
| "loss": 0.0651, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 2.612888688598468, | |
| "grad_norm": 8.432345390319824, | |
| "learning_rate": 2.3826714801444043e-05, | |
| "loss": 0.0719, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.630914826498423, | |
| "grad_norm": 20.364051818847656, | |
| "learning_rate": 2.3646209386281588e-05, | |
| "loss": 0.0682, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 2.6489409643983777, | |
| "grad_norm": 20.46283721923828, | |
| "learning_rate": 2.3465703971119137e-05, | |
| "loss": 0.0929, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 2.6669671022983326, | |
| "grad_norm": 7.231199264526367, | |
| "learning_rate": 2.328519855595668e-05, | |
| "loss": 0.0655, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 2.6849932401982874, | |
| "grad_norm": 3.337161064147949, | |
| "learning_rate": 2.3104693140794227e-05, | |
| "loss": 0.0288, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 2.7030193780982423, | |
| "grad_norm": 8.78416633605957, | |
| "learning_rate": 2.292418772563177e-05, | |
| "loss": 0.0797, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.721045515998197, | |
| "grad_norm": 13.156060218811035, | |
| "learning_rate": 2.2743682310469316e-05, | |
| "loss": 0.0533, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 2.7390716538981525, | |
| "grad_norm": 27.271133422851562, | |
| "learning_rate": 2.2563176895306858e-05, | |
| "loss": 0.0669, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 2.757097791798107, | |
| "grad_norm": 25.35082244873047, | |
| "learning_rate": 2.2382671480144403e-05, | |
| "loss": 0.0709, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 2.7751239296980623, | |
| "grad_norm": 1.3115813732147217, | |
| "learning_rate": 2.220216606498195e-05, | |
| "loss": 0.0545, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 2.793150067598017, | |
| "grad_norm": 9.926563262939453, | |
| "learning_rate": 2.2021660649819496e-05, | |
| "loss": 0.0345, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.811176205497972, | |
| "grad_norm": 12.350533485412598, | |
| "learning_rate": 2.184115523465704e-05, | |
| "loss": 0.1106, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 2.829202343397927, | |
| "grad_norm": 16.243236541748047, | |
| "learning_rate": 2.1660649819494586e-05, | |
| "loss": 0.1022, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 2.847228481297882, | |
| "grad_norm": 3.126918315887451, | |
| "learning_rate": 2.148014440433213e-05, | |
| "loss": 0.0695, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 2.8652546191978367, | |
| "grad_norm": 7.89870548248291, | |
| "learning_rate": 2.1299638989169676e-05, | |
| "loss": 0.0714, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 2.8832807570977916, | |
| "grad_norm": 17.361522674560547, | |
| "learning_rate": 2.111913357400722e-05, | |
| "loss": 0.0546, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.901306894997747, | |
| "grad_norm": 28.539026260375977, | |
| "learning_rate": 2.0938628158844766e-05, | |
| "loss": 0.0736, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 2.9193330328977014, | |
| "grad_norm": 9.696992874145508, | |
| "learning_rate": 2.075812274368231e-05, | |
| "loss": 0.0543, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 2.9373591707976567, | |
| "grad_norm": 14.558252334594727, | |
| "learning_rate": 2.057761732851986e-05, | |
| "loss": 0.0459, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 2.9553853086976116, | |
| "grad_norm": 15.250856399536133, | |
| "learning_rate": 2.0397111913357404e-05, | |
| "loss": 0.0668, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 2.9734114465975665, | |
| "grad_norm": 7.706424713134766, | |
| "learning_rate": 2.0216606498194946e-05, | |
| "loss": 0.0526, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.9914375844975214, | |
| "grad_norm": 6.1349945068359375, | |
| "learning_rate": 2.003610108303249e-05, | |
| "loss": 0.0816, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 3.0090130689499777, | |
| "grad_norm": 9.719901084899902, | |
| "learning_rate": 1.9855595667870036e-05, | |
| "loss": 0.0315, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 3.0270392068499326, | |
| "grad_norm": 14.534510612487793, | |
| "learning_rate": 1.967509025270758e-05, | |
| "loss": 0.0473, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 3.0450653447498874, | |
| "grad_norm": 46.903160095214844, | |
| "learning_rate": 1.9494584837545125e-05, | |
| "loss": 0.0152, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 3.0630914826498423, | |
| "grad_norm": 7.3955254554748535, | |
| "learning_rate": 1.9314079422382674e-05, | |
| "loss": 0.037, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.081117620549797, | |
| "grad_norm": 17.802396774291992, | |
| "learning_rate": 1.913357400722022e-05, | |
| "loss": 0.0472, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 3.099143758449752, | |
| "grad_norm": 1.0125623941421509, | |
| "learning_rate": 1.8953068592057764e-05, | |
| "loss": 0.0217, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 3.117169896349707, | |
| "grad_norm": 2.3023953437805176, | |
| "learning_rate": 1.877256317689531e-05, | |
| "loss": 0.0285, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 3.135196034249662, | |
| "grad_norm": 0.2563769221305847, | |
| "learning_rate": 1.8592057761732854e-05, | |
| "loss": 0.08, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 3.1532221721496168, | |
| "grad_norm": 6.048330783843994, | |
| "learning_rate": 1.84115523465704e-05, | |
| "loss": 0.0241, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.171248310049572, | |
| "grad_norm": 19.981657028198242, | |
| "learning_rate": 1.8231046931407943e-05, | |
| "loss": 0.0411, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 3.189274447949527, | |
| "grad_norm": 18.629526138305664, | |
| "learning_rate": 1.805054151624549e-05, | |
| "loss": 0.0352, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 3.207300585849482, | |
| "grad_norm": 51.33302307128906, | |
| "learning_rate": 1.7870036101083033e-05, | |
| "loss": 0.041, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 3.2253267237494367, | |
| "grad_norm": 15.880636215209961, | |
| "learning_rate": 1.768953068592058e-05, | |
| "loss": 0.0453, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 3.2433528616493916, | |
| "grad_norm": 1.6619783639907837, | |
| "learning_rate": 1.7509025270758123e-05, | |
| "loss": 0.0467, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.2613789995493465, | |
| "grad_norm": 28.439903259277344, | |
| "learning_rate": 1.7328519855595668e-05, | |
| "loss": 0.0347, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 3.2794051374493014, | |
| "grad_norm": 3.802276134490967, | |
| "learning_rate": 1.7148014440433213e-05, | |
| "loss": 0.0607, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 3.2974312753492563, | |
| "grad_norm": 0.17207615077495575, | |
| "learning_rate": 1.6967509025270758e-05, | |
| "loss": 0.0229, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 3.315457413249211, | |
| "grad_norm": 4.585143089294434, | |
| "learning_rate": 1.6787003610108303e-05, | |
| "loss": 0.0572, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 3.3334835511491665, | |
| "grad_norm": 0.3470168709754944, | |
| "learning_rate": 1.6606498194945848e-05, | |
| "loss": 0.0095, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 3.3515096890491214, | |
| "grad_norm": 1.0657894611358643, | |
| "learning_rate": 1.6425992779783393e-05, | |
| "loss": 0.056, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 3.3695358269490763, | |
| "grad_norm": 1.310930848121643, | |
| "learning_rate": 1.624548736462094e-05, | |
| "loss": 0.0278, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 3.387561964849031, | |
| "grad_norm": 0.5731173157691956, | |
| "learning_rate": 1.6064981949458486e-05, | |
| "loss": 0.0493, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 3.405588102748986, | |
| "grad_norm": 1.0355983972549438, | |
| "learning_rate": 1.588447653429603e-05, | |
| "loss": 0.0676, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 3.423614240648941, | |
| "grad_norm": 2.357179880142212, | |
| "learning_rate": 1.5703971119133576e-05, | |
| "loss": 0.0616, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.441640378548896, | |
| "grad_norm": 23.674598693847656, | |
| "learning_rate": 1.552346570397112e-05, | |
| "loss": 0.0522, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 3.4596665164488507, | |
| "grad_norm": 7.019822597503662, | |
| "learning_rate": 1.5342960288808663e-05, | |
| "loss": 0.0306, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 3.4776926543488056, | |
| "grad_norm": 10.89382553100586, | |
| "learning_rate": 1.516245487364621e-05, | |
| "loss": 0.0395, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 3.495718792248761, | |
| "grad_norm": 16.17826271057129, | |
| "learning_rate": 1.4981949458483754e-05, | |
| "loss": 0.0211, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 3.5137449301487154, | |
| "grad_norm": 21.532861709594727, | |
| "learning_rate": 1.4801444043321299e-05, | |
| "loss": 0.044, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 3.5317710680486707, | |
| "grad_norm": 0.7629399299621582, | |
| "learning_rate": 1.4620938628158846e-05, | |
| "loss": 0.0345, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 3.5497972059486256, | |
| "grad_norm": 42.46355438232422, | |
| "learning_rate": 1.444043321299639e-05, | |
| "loss": 0.0385, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 3.5678233438485805, | |
| "grad_norm": 33.24613571166992, | |
| "learning_rate": 1.4259927797833936e-05, | |
| "loss": 0.0555, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 3.5858494817485354, | |
| "grad_norm": 14.760307312011719, | |
| "learning_rate": 1.407942238267148e-05, | |
| "loss": 0.0309, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 3.6038756196484902, | |
| "grad_norm": 0.26270392537117004, | |
| "learning_rate": 1.3898916967509026e-05, | |
| "loss": 0.0397, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.621901757548445, | |
| "grad_norm": 18.96925926208496, | |
| "learning_rate": 1.3718411552346572e-05, | |
| "loss": 0.0607, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 3.6399278954484, | |
| "grad_norm": 0.8502946496009827, | |
| "learning_rate": 1.3537906137184117e-05, | |
| "loss": 0.022, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 3.6579540333483553, | |
| "grad_norm": 14.483598709106445, | |
| "learning_rate": 1.3357400722021662e-05, | |
| "loss": 0.0273, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 3.67598017124831, | |
| "grad_norm": 19.76103401184082, | |
| "learning_rate": 1.3176895306859207e-05, | |
| "loss": 0.0156, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 3.694006309148265, | |
| "grad_norm": 103.1634292602539, | |
| "learning_rate": 1.299638989169675e-05, | |
| "loss": 0.033, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 3.71203244704822, | |
| "grad_norm": 19.32097816467285, | |
| "learning_rate": 1.2815884476534295e-05, | |
| "loss": 0.0202, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 3.730058584948175, | |
| "grad_norm": 0.7152062654495239, | |
| "learning_rate": 1.263537906137184e-05, | |
| "loss": 0.0373, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 3.7462821090581344, | |
| "eval_f1": 0.7580025608194623, | |
| "eval_loss": 0.22068411111831665, | |
| "eval_precision": 0.7081339712918661, | |
| "eval_recall": 0.8154269972451791, | |
| "eval_runtime": 99.058, | |
| "eval_samples_per_second": 47.174, | |
| "eval_steps_per_second": 5.906, | |
| "step": 2079 | |
| }, | |
| { | |
| "epoch": 3.7480847228481298, | |
| "grad_norm": 15.258056640625, | |
| "learning_rate": 1.2454873646209387e-05, | |
| "loss": 0.0325, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 3.7661108607480847, | |
| "grad_norm": 13.003891944885254, | |
| "learning_rate": 1.2274368231046932e-05, | |
| "loss": 0.0655, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 3.7841369986480395, | |
| "grad_norm": 3.3672218322753906, | |
| "learning_rate": 1.2093862815884477e-05, | |
| "loss": 0.0595, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.8021631365479944, | |
| "grad_norm": 10.780056953430176, | |
| "learning_rate": 1.1913357400722022e-05, | |
| "loss": 0.0607, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 3.8201892744479498, | |
| "grad_norm": 4.282696723937988, | |
| "learning_rate": 1.1732851985559568e-05, | |
| "loss": 0.0222, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 3.838215412347904, | |
| "grad_norm": 7.597197532653809, | |
| "learning_rate": 1.1552346570397113e-05, | |
| "loss": 0.0425, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 3.8562415502478595, | |
| "grad_norm": 29.276180267333984, | |
| "learning_rate": 1.1371841155234658e-05, | |
| "loss": 0.0217, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 3.8742676881478144, | |
| "grad_norm": 1.948074221611023, | |
| "learning_rate": 1.1191335740072201e-05, | |
| "loss": 0.017, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.8922938260477693, | |
| "grad_norm": 0.19444149732589722, | |
| "learning_rate": 1.1010830324909748e-05, | |
| "loss": 0.0109, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 3.910319963947724, | |
| "grad_norm": 4.235020160675049, | |
| "learning_rate": 1.0830324909747293e-05, | |
| "loss": 0.0181, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 3.928346101847679, | |
| "grad_norm": 30.39529037475586, | |
| "learning_rate": 1.0649819494584838e-05, | |
| "loss": 0.0092, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 3.946372239747634, | |
| "grad_norm": 34.26523208618164, | |
| "learning_rate": 1.0469314079422383e-05, | |
| "loss": 0.0596, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 3.964398377647589, | |
| "grad_norm": 6.3887553215026855, | |
| "learning_rate": 1.028880866425993e-05, | |
| "loss": 0.0315, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.982424515547544, | |
| "grad_norm": 32.655189514160156, | |
| "learning_rate": 1.0108303249097473e-05, | |
| "loss": 0.0334, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 3.0682740211486816, | |
| "learning_rate": 9.927797833935018e-06, | |
| "loss": 0.067, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 4.018026137899955, | |
| "grad_norm": 3.4062814712524414, | |
| "learning_rate": 9.747292418772563e-06, | |
| "loss": 0.02, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 4.03605227579991, | |
| "grad_norm": 57.081993103027344, | |
| "learning_rate": 9.56678700361011e-06, | |
| "loss": 0.013, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 4.054078413699865, | |
| "grad_norm": 1.0301337242126465, | |
| "learning_rate": 9.386281588447654e-06, | |
| "loss": 0.0161, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 4.0721045515998195, | |
| "grad_norm": 0.3438373804092407, | |
| "learning_rate": 9.2057761732852e-06, | |
| "loss": 0.025, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 4.090130689499775, | |
| "grad_norm": 2.7622809410095215, | |
| "learning_rate": 9.025270758122744e-06, | |
| "loss": 0.0086, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 4.108156827399729, | |
| "grad_norm": 29.382837295532227, | |
| "learning_rate": 8.84476534296029e-06, | |
| "loss": 0.0174, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 4.126182965299685, | |
| "grad_norm": 7.100313663482666, | |
| "learning_rate": 8.664259927797834e-06, | |
| "loss": 0.0257, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 4.144209103199639, | |
| "grad_norm": 31.98218536376953, | |
| "learning_rate": 8.483754512635379e-06, | |
| "loss": 0.033, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 4.162235241099594, | |
| "grad_norm": 0.8779153823852539, | |
| "learning_rate": 8.303249097472924e-06, | |
| "loss": 0.0052, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 4.18026137899955, | |
| "grad_norm": 18.081058502197266, | |
| "learning_rate": 8.12274368231047e-06, | |
| "loss": 0.0114, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 4.198287516899504, | |
| "grad_norm": 261.2298278808594, | |
| "learning_rate": 7.942238267148016e-06, | |
| "loss": 0.0315, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 4.2163136547994595, | |
| "grad_norm": 5.500637531280518, | |
| "learning_rate": 7.76173285198556e-06, | |
| "loss": 0.0358, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 4.234339792699414, | |
| "grad_norm": 0.1317097395658493, | |
| "learning_rate": 7.581227436823105e-06, | |
| "loss": 0.0258, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 4.252365930599369, | |
| "grad_norm": 3.035848617553711, | |
| "learning_rate": 7.4007220216606496e-06, | |
| "loss": 0.0143, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 4.270392068499324, | |
| "grad_norm": 0.05077278986573219, | |
| "learning_rate": 7.220216606498195e-06, | |
| "loss": 0.0181, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 4.288418206399279, | |
| "grad_norm": 7.1182379722595215, | |
| "learning_rate": 7.03971119133574e-06, | |
| "loss": 0.03, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 4.3064443442992335, | |
| "grad_norm": 0.2915518581867218, | |
| "learning_rate": 6.859205776173286e-06, | |
| "loss": 0.0215, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 4.324470482199189, | |
| "grad_norm": 29.823713302612305, | |
| "learning_rate": 6.678700361010831e-06, | |
| "loss": 0.013, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 4.342496620099144, | |
| "grad_norm": 24.919235229492188, | |
| "learning_rate": 6.498194945848375e-06, | |
| "loss": 0.03, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 4.360522757999099, | |
| "grad_norm": 28.06072235107422, | |
| "learning_rate": 6.31768953068592e-06, | |
| "loss": 0.0197, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 4.378548895899054, | |
| "grad_norm": 0.1475232094526291, | |
| "learning_rate": 6.137184115523466e-06, | |
| "loss": 0.0457, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 4.396575033799008, | |
| "grad_norm": 11.063607215881348, | |
| "learning_rate": 5.956678700361011e-06, | |
| "loss": 0.0206, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 4.414601171698964, | |
| "grad_norm": 18.570436477661133, | |
| "learning_rate": 5.776173285198557e-06, | |
| "loss": 0.0347, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 4.432627309598918, | |
| "grad_norm": 0.21391639113426208, | |
| "learning_rate": 5.595667870036101e-06, | |
| "loss": 0.0127, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 4.4506534474988735, | |
| "grad_norm": 0.0873202458024025, | |
| "learning_rate": 5.4151624548736465e-06, | |
| "loss": 0.0091, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 4.468679585398828, | |
| "grad_norm": 0.1600690335035324, | |
| "learning_rate": 5.2346570397111915e-06, | |
| "loss": 0.0291, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 4.486705723298783, | |
| "grad_norm": 0.2069604992866516, | |
| "learning_rate": 5.054151624548736e-06, | |
| "loss": 0.0293, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 4.504731861198739, | |
| "grad_norm": 0.8192632794380188, | |
| "learning_rate": 4.873646209386281e-06, | |
| "loss": 0.0236, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.522757999098693, | |
| "grad_norm": 4.89054536819458, | |
| "learning_rate": 4.693140794223827e-06, | |
| "loss": 0.0276, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 4.540784136998648, | |
| "grad_norm": 24.810901641845703, | |
| "learning_rate": 4.512635379061372e-06, | |
| "loss": 0.0275, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 4.558810274898603, | |
| "grad_norm": 0.39144259691238403, | |
| "learning_rate": 4.332129963898917e-06, | |
| "loss": 0.0221, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 4.576836412798558, | |
| "grad_norm": 20.9676513671875, | |
| "learning_rate": 4.151624548736462e-06, | |
| "loss": 0.0184, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 4.594862550698513, | |
| "grad_norm": 0.15349972248077393, | |
| "learning_rate": 3.971119133574008e-06, | |
| "loss": 0.0086, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 4.612888688598468, | |
| "grad_norm": 1.4720628261566162, | |
| "learning_rate": 3.7906137184115523e-06, | |
| "loss": 0.0234, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 4.630914826498422, | |
| "grad_norm": 11.18973445892334, | |
| "learning_rate": 3.6101083032490977e-06, | |
| "loss": 0.0324, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 4.648940964398378, | |
| "grad_norm": 0.07954315841197968, | |
| "learning_rate": 3.429602888086643e-06, | |
| "loss": 0.0117, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 4.666967102298333, | |
| "grad_norm": 0.24165672063827515, | |
| "learning_rate": 3.2490974729241876e-06, | |
| "loss": 0.0121, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 4.6849932401982874, | |
| "grad_norm": 10.629122734069824, | |
| "learning_rate": 3.068592057761733e-06, | |
| "loss": 0.0246, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 4.703019378098243, | |
| "grad_norm": 0.5089478492736816, | |
| "learning_rate": 2.8880866425992783e-06, | |
| "loss": 0.0011, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 4.721045515998197, | |
| "grad_norm": 0.08004222065210342, | |
| "learning_rate": 2.7075812274368233e-06, | |
| "loss": 0.0364, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 4.7390716538981525, | |
| "grad_norm": 0.03793829679489136, | |
| "learning_rate": 2.527075812274368e-06, | |
| "loss": 0.0207, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 4.757097791798107, | |
| "grad_norm": 13.989105224609375, | |
| "learning_rate": 2.3465703971119136e-06, | |
| "loss": 0.0211, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 4.775123929698062, | |
| "grad_norm": 8.655593872070312, | |
| "learning_rate": 2.1660649819494585e-06, | |
| "loss": 0.0167, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 4.793150067598017, | |
| "grad_norm": 0.12756069004535675, | |
| "learning_rate": 1.985559566787004e-06, | |
| "loss": 0.019, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 4.811176205497972, | |
| "grad_norm": 23.192468643188477, | |
| "learning_rate": 1.8050541516245488e-06, | |
| "loss": 0.0074, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 4.829202343397927, | |
| "grad_norm": 0.17219115793704987, | |
| "learning_rate": 1.6245487364620938e-06, | |
| "loss": 0.0223, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 4.847228481297882, | |
| "grad_norm": 37.450740814208984, | |
| "learning_rate": 1.4440433212996392e-06, | |
| "loss": 0.0479, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 4.865254619197837, | |
| "grad_norm": 0.5785008668899536, | |
| "learning_rate": 1.263537906137184e-06, | |
| "loss": 0.0304, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 4.883280757097792, | |
| "grad_norm": 3.550365686416626, | |
| "learning_rate": 1.0830324909747293e-06, | |
| "loss": 0.0153, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 4.901306894997747, | |
| "grad_norm": 16.135068893432617, | |
| "learning_rate": 9.025270758122744e-07, | |
| "loss": 0.0437, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 4.919333032897701, | |
| "grad_norm": 0.8536310195922852, | |
| "learning_rate": 7.220216606498196e-07, | |
| "loss": 0.0159, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 4.937359170797657, | |
| "grad_norm": 0.09728287905454636, | |
| "learning_rate": 5.415162454873646e-07, | |
| "loss": 0.0116, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 4.955385308697611, | |
| "grad_norm": 2.123337745666504, | |
| "learning_rate": 3.610108303249098e-07, | |
| "loss": 0.0278, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 4.9734114465975665, | |
| "grad_norm": 1.2163937091827393, | |
| "learning_rate": 1.805054151624549e-07, | |
| "loss": 0.0012, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 4.991437584497522, | |
| "grad_norm": 0.2870337963104248, | |
| "learning_rate": 0.0, | |
| "loss": 0.0009, | |
| "step": 2770 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 2770, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 9.629766539300045e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |