| { | |
| "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": 0.6193506281612009, | |
| "eval_fn": 512, | |
| "eval_fp": 1821, | |
| "eval_loss": 0.736859142780304, | |
| "eval_precision": 0.5103522452272116, | |
| "eval_recall": 0.787551867219917, | |
| "eval_runtime": 69.9291, | |
| "eval_samples_per_second": 66.825, | |
| "eval_steps_per_second": 8.366, | |
| "eval_tn": 442, | |
| "eval_tp": 1898, | |
| "step": 0 | |
| }, | |
| { | |
| "epoch": 0.018026137899954935, | |
| "grad_norm": 32.56014633178711, | |
| "learning_rate": 4.981949458483755e-05, | |
| "loss": 0.8382, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03605227579990987, | |
| "grad_norm": 27.728681564331055, | |
| "learning_rate": 4.963898916967509e-05, | |
| "loss": 0.5565, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.054078413699864804, | |
| "grad_norm": 26.70035171508789, | |
| "learning_rate": 4.945848375451264e-05, | |
| "loss": 0.5545, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.07210455159981974, | |
| "grad_norm": 23.187734603881836, | |
| "learning_rate": 4.927797833935018e-05, | |
| "loss": 0.6027, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.09013068949977468, | |
| "grad_norm": 35.976619720458984, | |
| "learning_rate": 4.909747292418773e-05, | |
| "loss": 0.5741, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.10815682739972961, | |
| "grad_norm": 27.495447158813477, | |
| "learning_rate": 4.891696750902527e-05, | |
| "loss": 0.4941, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.12618296529968454, | |
| "grad_norm": 25.021883010864258, | |
| "learning_rate": 4.873646209386282e-05, | |
| "loss": 0.457, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.14420910319963948, | |
| "grad_norm": 29.016551971435547, | |
| "learning_rate": 4.855595667870036e-05, | |
| "loss": 0.4215, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.16223524109959442, | |
| "grad_norm": 25.652786254882812, | |
| "learning_rate": 4.837545126353791e-05, | |
| "loss": 0.4292, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.18026137899954936, | |
| "grad_norm": 24.254980087280273, | |
| "learning_rate": 4.819494584837546e-05, | |
| "loss": 0.4559, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.19828751689950427, | |
| "grad_norm": 43.745643615722656, | |
| "learning_rate": 4.8014440433213e-05, | |
| "loss": 0.4536, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.21631365479945922, | |
| "grad_norm": 19.265901565551758, | |
| "learning_rate": 4.783393501805055e-05, | |
| "loss": 0.4634, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.23433979269941416, | |
| "grad_norm": 75.13768768310547, | |
| "learning_rate": 4.765342960288809e-05, | |
| "loss": 0.4067, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.25236593059936907, | |
| "grad_norm": 24.337844848632812, | |
| "learning_rate": 4.747292418772563e-05, | |
| "loss": 0.4483, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.270392068499324, | |
| "grad_norm": 27.518775939941406, | |
| "learning_rate": 4.7292418772563177e-05, | |
| "loss": 0.4524, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.28841820639927895, | |
| "grad_norm": 23.281097412109375, | |
| "learning_rate": 4.711191335740072e-05, | |
| "loss": 0.4443, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.3064443442992339, | |
| "grad_norm": 12.933683395385742, | |
| "learning_rate": 4.693140794223827e-05, | |
| "loss": 0.472, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.32447048219918884, | |
| "grad_norm": 16.49010467529297, | |
| "learning_rate": 4.675090252707581e-05, | |
| "loss": 0.4549, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.3424966200991438, | |
| "grad_norm": 19.209016799926758, | |
| "learning_rate": 4.657039711191336e-05, | |
| "loss": 0.4628, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.3605227579990987, | |
| "grad_norm": 19.71845054626465, | |
| "learning_rate": 4.63898916967509e-05, | |
| "loss": 0.4281, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.3785488958990536, | |
| "grad_norm": 19.700483322143555, | |
| "learning_rate": 4.620938628158845e-05, | |
| "loss": 0.4427, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.39657503379900855, | |
| "grad_norm": 19.06282615661621, | |
| "learning_rate": 4.602888086642599e-05, | |
| "loss": 0.4214, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.4146011716989635, | |
| "grad_norm": 14.681564331054688, | |
| "learning_rate": 4.584837545126354e-05, | |
| "loss": 0.4279, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.43262730959891843, | |
| "grad_norm": 28.783143997192383, | |
| "learning_rate": 4.566787003610109e-05, | |
| "loss": 0.4475, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.45065344749887337, | |
| "grad_norm": 14.686984062194824, | |
| "learning_rate": 4.548736462093863e-05, | |
| "loss": 0.4232, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.4686795853988283, | |
| "grad_norm": 15.651535987854004, | |
| "learning_rate": 4.530685920577618e-05, | |
| "loss": 0.4026, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.48670572329878325, | |
| "grad_norm": 17.506153106689453, | |
| "learning_rate": 4.5126353790613716e-05, | |
| "loss": 0.416, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.5047318611987381, | |
| "grad_norm": 23.381561279296875, | |
| "learning_rate": 4.494584837545127e-05, | |
| "loss": 0.385, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.5227579990986931, | |
| "grad_norm": 18.458227157592773, | |
| "learning_rate": 4.4765342960288806e-05, | |
| "loss": 0.4352, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.540784136998648, | |
| "grad_norm": 16.285743713378906, | |
| "learning_rate": 4.458483754512636e-05, | |
| "loss": 0.4191, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.558810274898603, | |
| "grad_norm": 22.14523696899414, | |
| "learning_rate": 4.44043321299639e-05, | |
| "loss": 0.4525, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.5768364127985579, | |
| "grad_norm": 12.94579029083252, | |
| "learning_rate": 4.422382671480145e-05, | |
| "loss": 0.4335, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.5948625506985128, | |
| "grad_norm": 29.218290328979492, | |
| "learning_rate": 4.404332129963899e-05, | |
| "loss": 0.4194, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.6128886885984678, | |
| "grad_norm": 20.067014694213867, | |
| "learning_rate": 4.386281588447654e-05, | |
| "loss": 0.4016, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.6309148264984227, | |
| "grad_norm": 15.315394401550293, | |
| "learning_rate": 4.368231046931408e-05, | |
| "loss": 0.4293, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.6489409643983777, | |
| "grad_norm": 28.87151527404785, | |
| "learning_rate": 4.350180505415163e-05, | |
| "loss": 0.3952, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.6669671022983326, | |
| "grad_norm": 12.91943645477295, | |
| "learning_rate": 4.332129963898917e-05, | |
| "loss": 0.4276, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.6849932401982876, | |
| "grad_norm": 18.917442321777344, | |
| "learning_rate": 4.314079422382672e-05, | |
| "loss": 0.3843, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.7030193780982424, | |
| "grad_norm": 16.344675064086914, | |
| "learning_rate": 4.296028880866426e-05, | |
| "loss": 0.4283, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.7210455159981974, | |
| "grad_norm": 27.420021057128906, | |
| "learning_rate": 4.277978339350181e-05, | |
| "loss": 0.4663, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.7390716538981523, | |
| "grad_norm": 37.13134765625, | |
| "learning_rate": 4.259927797833935e-05, | |
| "loss": 0.4338, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.7570977917981072, | |
| "grad_norm": 23.59578514099121, | |
| "learning_rate": 4.24187725631769e-05, | |
| "loss": 0.3714, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.7751239296980622, | |
| "grad_norm": 30.009662628173828, | |
| "learning_rate": 4.223826714801444e-05, | |
| "loss": 0.3775, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.7931500675980171, | |
| "grad_norm": 13.575807571411133, | |
| "learning_rate": 4.205776173285199e-05, | |
| "loss": 0.4104, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.8111762054979721, | |
| "grad_norm": 18.727609634399414, | |
| "learning_rate": 4.187725631768953e-05, | |
| "loss": 0.3804, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.829202343397927, | |
| "grad_norm": 14.367807388305664, | |
| "learning_rate": 4.169675090252708e-05, | |
| "loss": 0.4124, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.847228481297882, | |
| "grad_norm": 12.945828437805176, | |
| "learning_rate": 4.151624548736462e-05, | |
| "loss": 0.4149, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.8652546191978369, | |
| "grad_norm": 24.02052116394043, | |
| "learning_rate": 4.1335740072202167e-05, | |
| "loss": 0.4093, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.8832807570977917, | |
| "grad_norm": 15.609790802001953, | |
| "learning_rate": 4.115523465703972e-05, | |
| "loss": 0.3874, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.9013068949977467, | |
| "grad_norm": 25.04174041748047, | |
| "learning_rate": 4.0974729241877256e-05, | |
| "loss": 0.391, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.9193330328977016, | |
| "grad_norm": 20.739681243896484, | |
| "learning_rate": 4.079422382671481e-05, | |
| "loss": 0.3939, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.9373591707976566, | |
| "grad_norm": 16.13037872314453, | |
| "learning_rate": 4.0613718411552346e-05, | |
| "loss": 0.4004, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.9553853086976115, | |
| "grad_norm": 21.17761993408203, | |
| "learning_rate": 4.043321299638989e-05, | |
| "loss": 0.3978, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.9734114465975665, | |
| "grad_norm": 27.03806495666504, | |
| "learning_rate": 4.0252707581227436e-05, | |
| "loss": 0.4391, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.9914375844975214, | |
| "grad_norm": 15.054065704345703, | |
| "learning_rate": 4.007220216606498e-05, | |
| "loss": 0.4187, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.0090130689499774, | |
| "grad_norm": 22.64523696899414, | |
| "learning_rate": 3.989169675090253e-05, | |
| "loss": 0.3444, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.0270392068499323, | |
| "grad_norm": 21.697904586791992, | |
| "learning_rate": 3.971119133574007e-05, | |
| "loss": 0.2976, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.0450653447498874, | |
| "grad_norm": 14.764150619506836, | |
| "learning_rate": 3.953068592057762e-05, | |
| "loss": 0.3295, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.0630914826498423, | |
| "grad_norm": 25.92884635925293, | |
| "learning_rate": 3.935018050541516e-05, | |
| "loss": 0.2984, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.0811176205497972, | |
| "grad_norm": 23.327699661254883, | |
| "learning_rate": 3.916967509025271e-05, | |
| "loss": 0.3668, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.099143758449752, | |
| "grad_norm": 27.034189224243164, | |
| "learning_rate": 3.898916967509025e-05, | |
| "loss": 0.2852, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.117169896349707, | |
| "grad_norm": 29.098979949951172, | |
| "learning_rate": 3.88086642599278e-05, | |
| "loss": 0.2913, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.135196034249662, | |
| "grad_norm": 21.875049591064453, | |
| "learning_rate": 3.862815884476535e-05, | |
| "loss": 0.2853, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.153222172149617, | |
| "grad_norm": 15.405293464660645, | |
| "learning_rate": 3.844765342960289e-05, | |
| "loss": 0.3296, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.1712483100495719, | |
| "grad_norm": 21.292566299438477, | |
| "learning_rate": 3.826714801444044e-05, | |
| "loss": 0.2739, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.1892744479495267, | |
| "grad_norm": 17.985347747802734, | |
| "learning_rate": 3.8086642599277976e-05, | |
| "loss": 0.3034, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.2073005858494819, | |
| "grad_norm": 16.177711486816406, | |
| "learning_rate": 3.790613718411553e-05, | |
| "loss": 0.3032, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.2253267237494367, | |
| "grad_norm": 38.261051177978516, | |
| "learning_rate": 3.7725631768953066e-05, | |
| "loss": 0.2819, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.2433528616493916, | |
| "grad_norm": 24.247678756713867, | |
| "learning_rate": 3.754512635379062e-05, | |
| "loss": 0.3527, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.2487607030193781, | |
| "eval_f1": 0.8497959183673469, | |
| "eval_fn": 328, | |
| "eval_fp": 408, | |
| "eval_loss": 0.36085668206214905, | |
| "eval_precision": 0.8361445783132531, | |
| "eval_recall": 0.8639004149377594, | |
| "eval_runtime": 63.9221, | |
| "eval_samples_per_second": 73.105, | |
| "eval_steps_per_second": 9.152, | |
| "eval_tn": 1855, | |
| "eval_tp": 2082, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 1.2613789995493465, | |
| "grad_norm": 26.0637149810791, | |
| "learning_rate": 3.7364620938628155e-05, | |
| "loss": 0.2519, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.2794051374493014, | |
| "grad_norm": 22.626008987426758, | |
| "learning_rate": 3.718411552346571e-05, | |
| "loss": 0.2764, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.2974312753492563, | |
| "grad_norm": 24.119760513305664, | |
| "learning_rate": 3.700361010830325e-05, | |
| "loss": 0.3114, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.3154574132492114, | |
| "grad_norm": 25.81672477722168, | |
| "learning_rate": 3.68231046931408e-05, | |
| "loss": 0.2792, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.3334835511491663, | |
| "grad_norm": 15.591266632080078, | |
| "learning_rate": 3.664259927797834e-05, | |
| "loss": 0.2857, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.3515096890491212, | |
| "grad_norm": 21.272573471069336, | |
| "learning_rate": 3.646209386281589e-05, | |
| "loss": 0.2894, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.3695358269490763, | |
| "grad_norm": 24.08788299560547, | |
| "learning_rate": 3.628158844765343e-05, | |
| "loss": 0.2806, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.3875619648490312, | |
| "grad_norm": 20.197420120239258, | |
| "learning_rate": 3.610108303249098e-05, | |
| "loss": 0.3313, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.405588102748986, | |
| "grad_norm": 22.551584243774414, | |
| "learning_rate": 3.592057761732852e-05, | |
| "loss": 0.3261, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.423614240648941, | |
| "grad_norm": 15.885344505310059, | |
| "learning_rate": 3.574007220216607e-05, | |
| "loss": 0.3198, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.4416403785488958, | |
| "grad_norm": 30.44365882873535, | |
| "learning_rate": 3.555956678700361e-05, | |
| "loss": 0.3143, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.4596665164488507, | |
| "grad_norm": 18.411624908447266, | |
| "learning_rate": 3.537906137184116e-05, | |
| "loss": 0.2787, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.4776926543488058, | |
| "grad_norm": 18.90251922607422, | |
| "learning_rate": 3.51985559566787e-05, | |
| "loss": 0.2858, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.4957187922487607, | |
| "grad_norm": 28.24523162841797, | |
| "learning_rate": 3.5018050541516247e-05, | |
| "loss": 0.2856, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.5137449301487156, | |
| "grad_norm": 20.0897216796875, | |
| "learning_rate": 3.483754512635379e-05, | |
| "loss": 0.324, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.5317710680486707, | |
| "grad_norm": 20.818204879760742, | |
| "learning_rate": 3.4657039711191336e-05, | |
| "loss": 0.2933, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.5497972059486256, | |
| "grad_norm": 25.051939010620117, | |
| "learning_rate": 3.447653429602888e-05, | |
| "loss": 0.2948, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.5678233438485805, | |
| "grad_norm": 32.370933532714844, | |
| "learning_rate": 3.4296028880866426e-05, | |
| "loss": 0.2847, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.5858494817485354, | |
| "grad_norm": 34.086158752441406, | |
| "learning_rate": 3.411552346570397e-05, | |
| "loss": 0.2805, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.6038756196484902, | |
| "grad_norm": 17.575464248657227, | |
| "learning_rate": 3.3935018050541516e-05, | |
| "loss": 0.3013, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.6219017575484451, | |
| "grad_norm": 16.147525787353516, | |
| "learning_rate": 3.375451263537907e-05, | |
| "loss": 0.3257, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.6399278954484002, | |
| "grad_norm": 23.968366622924805, | |
| "learning_rate": 3.3574007220216606e-05, | |
| "loss": 0.3042, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.6579540333483551, | |
| "grad_norm": 24.20454216003418, | |
| "learning_rate": 3.339350180505416e-05, | |
| "loss": 0.3174, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.67598017124831, | |
| "grad_norm": 14.595170021057129, | |
| "learning_rate": 3.3212996389891696e-05, | |
| "loss": 0.305, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.694006309148265, | |
| "grad_norm": 46.18140411376953, | |
| "learning_rate": 3.303249097472924e-05, | |
| "loss": 0.3085, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.71203244704822, | |
| "grad_norm": 25.23206329345703, | |
| "learning_rate": 3.2851985559566786e-05, | |
| "loss": 0.2921, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.7300585849481749, | |
| "grad_norm": 18.097789764404297, | |
| "learning_rate": 3.267148014440433e-05, | |
| "loss": 0.2792, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.7480847228481298, | |
| "grad_norm": 25.6617488861084, | |
| "learning_rate": 3.249097472924188e-05, | |
| "loss": 0.3027, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.7661108607480847, | |
| "grad_norm": 33.289459228515625, | |
| "learning_rate": 3.231046931407942e-05, | |
| "loss": 0.2854, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.7841369986480395, | |
| "grad_norm": 20.507280349731445, | |
| "learning_rate": 3.212996389891697e-05, | |
| "loss": 0.3646, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.8021631365479944, | |
| "grad_norm": 17.23845100402832, | |
| "learning_rate": 3.194945848375451e-05, | |
| "loss": 0.2958, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.8201892744479495, | |
| "grad_norm": 23.676586151123047, | |
| "learning_rate": 3.176895306859206e-05, | |
| "loss": 0.2935, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.8382154123479044, | |
| "grad_norm": 12.614605903625488, | |
| "learning_rate": 3.15884476534296e-05, | |
| "loss": 0.2903, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.8562415502478595, | |
| "grad_norm": 42.599510192871094, | |
| "learning_rate": 3.140794223826715e-05, | |
| "loss": 0.3415, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.8742676881478144, | |
| "grad_norm": 31.382732391357422, | |
| "learning_rate": 3.12274368231047e-05, | |
| "loss": 0.31, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.8922938260477693, | |
| "grad_norm": 41.608890533447266, | |
| "learning_rate": 3.104693140794224e-05, | |
| "loss": 0.2515, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.9103199639477242, | |
| "grad_norm": 28.570653915405273, | |
| "learning_rate": 3.086642599277979e-05, | |
| "loss": 0.325, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.928346101847679, | |
| "grad_norm": 28.502283096313477, | |
| "learning_rate": 3.0685920577617325e-05, | |
| "loss": 0.2759, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.946372239747634, | |
| "grad_norm": 16.24626350402832, | |
| "learning_rate": 3.0505415162454877e-05, | |
| "loss": 0.2995, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.9643983776475888, | |
| "grad_norm": 27.27639389038086, | |
| "learning_rate": 3.032490974729242e-05, | |
| "loss": 0.2787, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.982424515547544, | |
| "grad_norm": 27.928380966186523, | |
| "learning_rate": 3.0144404332129967e-05, | |
| "loss": 0.2762, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 22.15119743347168, | |
| "learning_rate": 2.996389891696751e-05, | |
| "loss": 0.2623, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 2.018026137899955, | |
| "grad_norm": 20.76552963256836, | |
| "learning_rate": 2.9783393501805057e-05, | |
| "loss": 0.1688, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 2.0360522757999098, | |
| "grad_norm": 32.505191802978516, | |
| "learning_rate": 2.9602888086642598e-05, | |
| "loss": 0.1477, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 2.0540784136998647, | |
| "grad_norm": 29.859451293945312, | |
| "learning_rate": 2.9422382671480147e-05, | |
| "loss": 0.1473, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 2.0721045515998195, | |
| "grad_norm": 22.375699996948242, | |
| "learning_rate": 2.924187725631769e-05, | |
| "loss": 0.1156, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.090130689499775, | |
| "grad_norm": 26.230262756347656, | |
| "learning_rate": 2.906137184115524e-05, | |
| "loss": 0.1923, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 2.1081568273997298, | |
| "grad_norm": 29.501089096069336, | |
| "learning_rate": 2.888086642599278e-05, | |
| "loss": 0.1694, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 2.1261829652996846, | |
| "grad_norm": 46.50517654418945, | |
| "learning_rate": 2.870036101083033e-05, | |
| "loss": 0.1945, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 2.1442091031996395, | |
| "grad_norm": 45.4150276184082, | |
| "learning_rate": 2.851985559566787e-05, | |
| "loss": 0.1391, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 2.1622352410995944, | |
| "grad_norm": 19.459104537963867, | |
| "learning_rate": 2.8339350180505413e-05, | |
| "loss": 0.1817, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.1802613789995493, | |
| "grad_norm": 23.73355484008789, | |
| "learning_rate": 2.815884476534296e-05, | |
| "loss": 0.144, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 2.198287516899504, | |
| "grad_norm": 31.736591339111328, | |
| "learning_rate": 2.7978339350180506e-05, | |
| "loss": 0.1532, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 2.216313654799459, | |
| "grad_norm": 17.197023391723633, | |
| "learning_rate": 2.779783393501805e-05, | |
| "loss": 0.2099, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 2.234339792699414, | |
| "grad_norm": 16.515472412109375, | |
| "learning_rate": 2.7617328519855596e-05, | |
| "loss": 0.1429, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 2.2523659305993693, | |
| "grad_norm": 42.32828140258789, | |
| "learning_rate": 2.7436823104693144e-05, | |
| "loss": 0.1326, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.270392068499324, | |
| "grad_norm": 48.43349075317383, | |
| "learning_rate": 2.7256317689530686e-05, | |
| "loss": 0.2316, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 2.288418206399279, | |
| "grad_norm": 21.47289276123047, | |
| "learning_rate": 2.7075812274368234e-05, | |
| "loss": 0.1538, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 2.306444344299234, | |
| "grad_norm": 61.8057861328125, | |
| "learning_rate": 2.6895306859205776e-05, | |
| "loss": 0.1624, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 2.324470482199189, | |
| "grad_norm": 22.287643432617188, | |
| "learning_rate": 2.6714801444043324e-05, | |
| "loss": 0.1518, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 2.3424966200991437, | |
| "grad_norm": 50.18746566772461, | |
| "learning_rate": 2.6534296028880866e-05, | |
| "loss": 0.1767, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.3605227579990986, | |
| "grad_norm": 52.100826263427734, | |
| "learning_rate": 2.6353790613718414e-05, | |
| "loss": 0.1942, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 2.3785488958990535, | |
| "grad_norm": 22.549230575561523, | |
| "learning_rate": 2.617328519855596e-05, | |
| "loss": 0.159, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 2.3965750337990084, | |
| "grad_norm": 33.591896057128906, | |
| "learning_rate": 2.59927797833935e-05, | |
| "loss": 0.2036, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 2.4146011716989637, | |
| "grad_norm": 31.62824058532715, | |
| "learning_rate": 2.581227436823105e-05, | |
| "loss": 0.1464, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 2.4326273095989186, | |
| "grad_norm": 28.01429557800293, | |
| "learning_rate": 2.563176895306859e-05, | |
| "loss": 0.1617, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.4506534474988735, | |
| "grad_norm": 55.43074417114258, | |
| "learning_rate": 2.545126353790614e-05, | |
| "loss": 0.1415, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 2.4686795853988284, | |
| "grad_norm": 20.73956298828125, | |
| "learning_rate": 2.527075812274368e-05, | |
| "loss": 0.1929, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 2.4867057232987833, | |
| "grad_norm": 33.961395263671875, | |
| "learning_rate": 2.509025270758123e-05, | |
| "loss": 0.173, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 2.4975214060387563, | |
| "eval_f1": 0.8451639684516397, | |
| "eval_fn": 374, | |
| "eval_fp": 372, | |
| "eval_loss": 0.4756762683391571, | |
| "eval_precision": 0.845514950166113, | |
| "eval_recall": 0.8448132780082988, | |
| "eval_runtime": 55.8639, | |
| "eval_samples_per_second": 83.65, | |
| "eval_steps_per_second": 10.472, | |
| "eval_tn": 1891, | |
| "eval_tp": 2036, | |
| "step": 1386 | |
| }, | |
| { | |
| "epoch": 2.504731861198738, | |
| "grad_norm": 28.332387924194336, | |
| "learning_rate": 2.4909747292418774e-05, | |
| "loss": 0.1992, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 2.522757999098693, | |
| "grad_norm": 35.882781982421875, | |
| "learning_rate": 2.472924187725632e-05, | |
| "loss": 0.2178, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.540784136998648, | |
| "grad_norm": 15.528435707092285, | |
| "learning_rate": 2.4548736462093864e-05, | |
| "loss": 0.1745, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 2.558810274898603, | |
| "grad_norm": 35.54289627075195, | |
| "learning_rate": 2.436823104693141e-05, | |
| "loss": 0.1731, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 2.576836412798558, | |
| "grad_norm": 394.8744201660156, | |
| "learning_rate": 2.4187725631768953e-05, | |
| "loss": 0.1702, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 2.5948625506985126, | |
| "grad_norm": 27.29697036743164, | |
| "learning_rate": 2.40072202166065e-05, | |
| "loss": 0.1576, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 2.612888688598468, | |
| "grad_norm": 27.070558547973633, | |
| "learning_rate": 2.3826714801444043e-05, | |
| "loss": 0.2202, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.630914826498423, | |
| "grad_norm": 36.06498718261719, | |
| "learning_rate": 2.3646209386281588e-05, | |
| "loss": 0.1796, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 2.6489409643983777, | |
| "grad_norm": 24.6717529296875, | |
| "learning_rate": 2.3465703971119137e-05, | |
| "loss": 0.1501, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 2.6669671022983326, | |
| "grad_norm": 30.312421798706055, | |
| "learning_rate": 2.328519855595668e-05, | |
| "loss": 0.1834, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 2.6849932401982874, | |
| "grad_norm": 27.66963005065918, | |
| "learning_rate": 2.3104693140794227e-05, | |
| "loss": 0.1464, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 2.7030193780982423, | |
| "grad_norm": 28.956432342529297, | |
| "learning_rate": 2.292418772563177e-05, | |
| "loss": 0.1877, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.721045515998197, | |
| "grad_norm": 20.93885040283203, | |
| "learning_rate": 2.2743682310469316e-05, | |
| "loss": 0.1781, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 2.7390716538981525, | |
| "grad_norm": 15.14222526550293, | |
| "learning_rate": 2.2563176895306858e-05, | |
| "loss": 0.1837, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 2.757097791798107, | |
| "grad_norm": 32.12163543701172, | |
| "learning_rate": 2.2382671480144403e-05, | |
| "loss": 0.1547, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 2.7751239296980623, | |
| "grad_norm": 12.95942211151123, | |
| "learning_rate": 2.220216606498195e-05, | |
| "loss": 0.1912, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 2.793150067598017, | |
| "grad_norm": 19.13577651977539, | |
| "learning_rate": 2.2021660649819496e-05, | |
| "loss": 0.1703, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.811176205497972, | |
| "grad_norm": 25.11808204650879, | |
| "learning_rate": 2.184115523465704e-05, | |
| "loss": 0.1786, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 2.829202343397927, | |
| "grad_norm": 38.005741119384766, | |
| "learning_rate": 2.1660649819494586e-05, | |
| "loss": 0.191, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 2.847228481297882, | |
| "grad_norm": 28.062721252441406, | |
| "learning_rate": 2.148014440433213e-05, | |
| "loss": 0.1384, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 2.8652546191978367, | |
| "grad_norm": 38.120235443115234, | |
| "learning_rate": 2.1299638989169676e-05, | |
| "loss": 0.1778, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 2.8832807570977916, | |
| "grad_norm": 25.847713470458984, | |
| "learning_rate": 2.111913357400722e-05, | |
| "loss": 0.1676, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.901306894997747, | |
| "grad_norm": 23.463483810424805, | |
| "learning_rate": 2.0938628158844766e-05, | |
| "loss": 0.173, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 2.9193330328977014, | |
| "grad_norm": 28.62253761291504, | |
| "learning_rate": 2.075812274368231e-05, | |
| "loss": 0.1739, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 2.9373591707976567, | |
| "grad_norm": 16.46048355102539, | |
| "learning_rate": 2.057761732851986e-05, | |
| "loss": 0.1495, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 2.9553853086976116, | |
| "grad_norm": 24.875967025756836, | |
| "learning_rate": 2.0397111913357404e-05, | |
| "loss": 0.1416, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 2.9734114465975665, | |
| "grad_norm": 46.05429458618164, | |
| "learning_rate": 2.0216606498194946e-05, | |
| "loss": 0.1659, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.9914375844975214, | |
| "grad_norm": 19.724300384521484, | |
| "learning_rate": 2.003610108303249e-05, | |
| "loss": 0.1417, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 3.0090130689499777, | |
| "grad_norm": 6.316404342651367, | |
| "learning_rate": 1.9855595667870036e-05, | |
| "loss": 0.1246, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 3.0270392068499326, | |
| "grad_norm": 35.23857879638672, | |
| "learning_rate": 1.967509025270758e-05, | |
| "loss": 0.0565, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 3.0450653447498874, | |
| "grad_norm": 35.49019241333008, | |
| "learning_rate": 1.9494584837545125e-05, | |
| "loss": 0.0669, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 3.0630914826498423, | |
| "grad_norm": 110.1063461303711, | |
| "learning_rate": 1.9314079422382674e-05, | |
| "loss": 0.1124, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.081117620549797, | |
| "grad_norm": 34.99469757080078, | |
| "learning_rate": 1.913357400722022e-05, | |
| "loss": 0.0338, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 3.099143758449752, | |
| "grad_norm": 26.450838088989258, | |
| "learning_rate": 1.8953068592057764e-05, | |
| "loss": 0.0639, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 3.117169896349707, | |
| "grad_norm": 77.21393585205078, | |
| "learning_rate": 1.877256317689531e-05, | |
| "loss": 0.0778, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 3.135196034249662, | |
| "grad_norm": 23.50945281982422, | |
| "learning_rate": 1.8592057761732854e-05, | |
| "loss": 0.0827, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 3.1532221721496168, | |
| "grad_norm": 25.54047203063965, | |
| "learning_rate": 1.84115523465704e-05, | |
| "loss": 0.0737, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.171248310049572, | |
| "grad_norm": 14.785638809204102, | |
| "learning_rate": 1.8231046931407943e-05, | |
| "loss": 0.0707, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 3.189274447949527, | |
| "grad_norm": 19.547487258911133, | |
| "learning_rate": 1.805054151624549e-05, | |
| "loss": 0.0541, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 3.207300585849482, | |
| "grad_norm": 24.53461456298828, | |
| "learning_rate": 1.7870036101083033e-05, | |
| "loss": 0.0774, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 3.2253267237494367, | |
| "grad_norm": 10.091083526611328, | |
| "learning_rate": 1.768953068592058e-05, | |
| "loss": 0.0892, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 3.2433528616493916, | |
| "grad_norm": 12.438738822937012, | |
| "learning_rate": 1.7509025270758123e-05, | |
| "loss": 0.0806, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.2613789995493465, | |
| "grad_norm": 35.19310760498047, | |
| "learning_rate": 1.7328519855595668e-05, | |
| "loss": 0.0587, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 3.2794051374493014, | |
| "grad_norm": 66.27088165283203, | |
| "learning_rate": 1.7148014440433213e-05, | |
| "loss": 0.1149, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 3.2974312753492563, | |
| "grad_norm": 38.344879150390625, | |
| "learning_rate": 1.6967509025270758e-05, | |
| "loss": 0.0885, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 3.315457413249211, | |
| "grad_norm": 24.388492584228516, | |
| "learning_rate": 1.6787003610108303e-05, | |
| "loss": 0.0749, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 3.3334835511491665, | |
| "grad_norm": 13.80942440032959, | |
| "learning_rate": 1.6606498194945848e-05, | |
| "loss": 0.0619, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 3.3515096890491214, | |
| "grad_norm": 21.927907943725586, | |
| "learning_rate": 1.6425992779783393e-05, | |
| "loss": 0.066, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 3.3695358269490763, | |
| "grad_norm": 17.157190322875977, | |
| "learning_rate": 1.624548736462094e-05, | |
| "loss": 0.077, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 3.387561964849031, | |
| "grad_norm": 65.74653625488281, | |
| "learning_rate": 1.6064981949458486e-05, | |
| "loss": 0.0761, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 3.405588102748986, | |
| "grad_norm": 24.20794677734375, | |
| "learning_rate": 1.588447653429603e-05, | |
| "loss": 0.0887, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 3.423614240648941, | |
| "grad_norm": 62.70228958129883, | |
| "learning_rate": 1.5703971119133576e-05, | |
| "loss": 0.0937, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.441640378548896, | |
| "grad_norm": 29.74757957458496, | |
| "learning_rate": 1.552346570397112e-05, | |
| "loss": 0.0636, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 3.4596665164488507, | |
| "grad_norm": 42.17036819458008, | |
| "learning_rate": 1.5342960288808663e-05, | |
| "loss": 0.0871, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 3.4776926543488056, | |
| "grad_norm": 36.15501403808594, | |
| "learning_rate": 1.516245487364621e-05, | |
| "loss": 0.1034, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 3.495718792248761, | |
| "grad_norm": 16.554231643676758, | |
| "learning_rate": 1.4981949458483754e-05, | |
| "loss": 0.0552, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 3.5137449301487154, | |
| "grad_norm": 27.479389190673828, | |
| "learning_rate": 1.4801444043321299e-05, | |
| "loss": 0.0792, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 3.5317710680486707, | |
| "grad_norm": 51.35026550292969, | |
| "learning_rate": 1.4620938628158846e-05, | |
| "loss": 0.0735, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 3.5497972059486256, | |
| "grad_norm": 28.704710006713867, | |
| "learning_rate": 1.444043321299639e-05, | |
| "loss": 0.0746, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 3.5678233438485805, | |
| "grad_norm": 50.08689880371094, | |
| "learning_rate": 1.4259927797833936e-05, | |
| "loss": 0.0678, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 3.5858494817485354, | |
| "grad_norm": 7.874759674072266, | |
| "learning_rate": 1.407942238267148e-05, | |
| "loss": 0.0526, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 3.6038756196484902, | |
| "grad_norm": 75.26201629638672, | |
| "learning_rate": 1.3898916967509026e-05, | |
| "loss": 0.076, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.621901757548445, | |
| "grad_norm": 30.040180206298828, | |
| "learning_rate": 1.3718411552346572e-05, | |
| "loss": 0.0842, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 3.6399278954484, | |
| "grad_norm": 68.3320541381836, | |
| "learning_rate": 1.3537906137184117e-05, | |
| "loss": 0.0782, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 3.6579540333483553, | |
| "grad_norm": 47.21413040161133, | |
| "learning_rate": 1.3357400722021662e-05, | |
| "loss": 0.0938, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 3.67598017124831, | |
| "grad_norm": 51.68861389160156, | |
| "learning_rate": 1.3176895306859207e-05, | |
| "loss": 0.0618, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 3.694006309148265, | |
| "grad_norm": 15.36585807800293, | |
| "learning_rate": 1.299638989169675e-05, | |
| "loss": 0.07, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 3.71203244704822, | |
| "grad_norm": 27.488054275512695, | |
| "learning_rate": 1.2815884476534295e-05, | |
| "loss": 0.0917, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 3.730058584948175, | |
| "grad_norm": 12.570659637451172, | |
| "learning_rate": 1.263537906137184e-05, | |
| "loss": 0.0868, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 3.7462821090581344, | |
| "eval_f1": 0.8361511166771028, | |
| "eval_fn": 407, | |
| "eval_fp": 378, | |
| "eval_loss": 0.7050483822822571, | |
| "eval_precision": 0.8412431751364973, | |
| "eval_recall": 0.8311203319502075, | |
| "eval_runtime": 66.9536, | |
| "eval_samples_per_second": 69.795, | |
| "eval_steps_per_second": 8.737, | |
| "eval_tn": 1885, | |
| "eval_tp": 2003, | |
| "step": 2079 | |
| }, | |
| { | |
| "epoch": 3.7480847228481298, | |
| "grad_norm": 21.334257125854492, | |
| "learning_rate": 1.2454873646209387e-05, | |
| "loss": 0.0991, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 3.7661108607480847, | |
| "grad_norm": 43.5160026550293, | |
| "learning_rate": 1.2274368231046932e-05, | |
| "loss": 0.0589, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 3.7841369986480395, | |
| "grad_norm": 37.79927062988281, | |
| "learning_rate": 1.2093862815884477e-05, | |
| "loss": 0.0828, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.8021631365479944, | |
| "grad_norm": 41.13615417480469, | |
| "learning_rate": 1.1913357400722022e-05, | |
| "loss": 0.0978, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 3.8201892744479498, | |
| "grad_norm": 17.815465927124023, | |
| "learning_rate": 1.1732851985559568e-05, | |
| "loss": 0.0855, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 3.838215412347904, | |
| "grad_norm": 29.20524787902832, | |
| "learning_rate": 1.1552346570397113e-05, | |
| "loss": 0.0737, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 3.8562415502478595, | |
| "grad_norm": 70.78984069824219, | |
| "learning_rate": 1.1371841155234658e-05, | |
| "loss": 0.107, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 3.8742676881478144, | |
| "grad_norm": 11.879849433898926, | |
| "learning_rate": 1.1191335740072201e-05, | |
| "loss": 0.0916, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.8922938260477693, | |
| "grad_norm": 26.21539306640625, | |
| "learning_rate": 1.1010830324909748e-05, | |
| "loss": 0.0862, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 3.910319963947724, | |
| "grad_norm": 18.701679229736328, | |
| "learning_rate": 1.0830324909747293e-05, | |
| "loss": 0.0589, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 3.928346101847679, | |
| "grad_norm": 27.045053482055664, | |
| "learning_rate": 1.0649819494584838e-05, | |
| "loss": 0.0513, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 3.946372239747634, | |
| "grad_norm": 28.006908416748047, | |
| "learning_rate": 1.0469314079422383e-05, | |
| "loss": 0.0476, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 3.964398377647589, | |
| "grad_norm": 36.951419830322266, | |
| "learning_rate": 1.028880866425993e-05, | |
| "loss": 0.0904, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.982424515547544, | |
| "grad_norm": 70.6955337524414, | |
| "learning_rate": 1.0108303249097473e-05, | |
| "loss": 0.1079, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 23.484031677246094, | |
| "learning_rate": 9.927797833935018e-06, | |
| "loss": 0.0614, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 4.018026137899955, | |
| "grad_norm": 19.981300354003906, | |
| "learning_rate": 9.747292418772563e-06, | |
| "loss": 0.0286, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 4.03605227579991, | |
| "grad_norm": 35.96101379394531, | |
| "learning_rate": 9.56678700361011e-06, | |
| "loss": 0.0341, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 4.054078413699865, | |
| "grad_norm": 24.119760513305664, | |
| "learning_rate": 9.386281588447654e-06, | |
| "loss": 0.0344, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 4.0721045515998195, | |
| "grad_norm": 22.60484504699707, | |
| "learning_rate": 9.2057761732852e-06, | |
| "loss": 0.0399, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 4.090130689499775, | |
| "grad_norm": 18.592531204223633, | |
| "learning_rate": 9.025270758122744e-06, | |
| "loss": 0.045, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 4.108156827399729, | |
| "grad_norm": 63.02653884887695, | |
| "learning_rate": 8.84476534296029e-06, | |
| "loss": 0.0531, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 4.126182965299685, | |
| "grad_norm": 36.79388427734375, | |
| "learning_rate": 8.664259927797834e-06, | |
| "loss": 0.0432, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 4.144209103199639, | |
| "grad_norm": 3.6419107913970947, | |
| "learning_rate": 8.483754512635379e-06, | |
| "loss": 0.0728, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 4.162235241099594, | |
| "grad_norm": 4.659115791320801, | |
| "learning_rate": 8.303249097472924e-06, | |
| "loss": 0.0212, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 4.18026137899955, | |
| "grad_norm": 25.78754425048828, | |
| "learning_rate": 8.12274368231047e-06, | |
| "loss": 0.0248, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 4.198287516899504, | |
| "grad_norm": 30.966909408569336, | |
| "learning_rate": 7.942238267148016e-06, | |
| "loss": 0.0703, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 4.2163136547994595, | |
| "grad_norm": 28.843400955200195, | |
| "learning_rate": 7.76173285198556e-06, | |
| "loss": 0.0457, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 4.234339792699414, | |
| "grad_norm": 0.7049341201782227, | |
| "learning_rate": 7.581227436823105e-06, | |
| "loss": 0.0225, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 4.252365930599369, | |
| "grad_norm": 46.65751266479492, | |
| "learning_rate": 7.4007220216606496e-06, | |
| "loss": 0.0337, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 4.270392068499324, | |
| "grad_norm": 82.9363784790039, | |
| "learning_rate": 7.220216606498195e-06, | |
| "loss": 0.0327, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 4.288418206399279, | |
| "grad_norm": 5.899658203125, | |
| "learning_rate": 7.03971119133574e-06, | |
| "loss": 0.0043, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 4.3064443442992335, | |
| "grad_norm": 34.57490921020508, | |
| "learning_rate": 6.859205776173286e-06, | |
| "loss": 0.0307, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 4.324470482199189, | |
| "grad_norm": 12.653234481811523, | |
| "learning_rate": 6.678700361010831e-06, | |
| "loss": 0.0523, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 4.342496620099144, | |
| "grad_norm": 3.5938429832458496, | |
| "learning_rate": 6.498194945848375e-06, | |
| "loss": 0.0306, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 4.360522757999099, | |
| "grad_norm": 20.71294593811035, | |
| "learning_rate": 6.31768953068592e-06, | |
| "loss": 0.049, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 4.378548895899054, | |
| "grad_norm": 5.795297622680664, | |
| "learning_rate": 6.137184115523466e-06, | |
| "loss": 0.0358, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 4.396575033799008, | |
| "grad_norm": 34.56690979003906, | |
| "learning_rate": 5.956678700361011e-06, | |
| "loss": 0.0508, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 4.414601171698964, | |
| "grad_norm": 63.80348587036133, | |
| "learning_rate": 5.776173285198557e-06, | |
| "loss": 0.0511, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 4.432627309598918, | |
| "grad_norm": 2.4131686687469482, | |
| "learning_rate": 5.595667870036101e-06, | |
| "loss": 0.049, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 4.4506534474988735, | |
| "grad_norm": 52.36531448364258, | |
| "learning_rate": 5.4151624548736465e-06, | |
| "loss": 0.022, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 4.468679585398828, | |
| "grad_norm": 0.7890565395355225, | |
| "learning_rate": 5.2346570397111915e-06, | |
| "loss": 0.0281, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 4.486705723298783, | |
| "grad_norm": 70.13034057617188, | |
| "learning_rate": 5.054151624548736e-06, | |
| "loss": 0.0542, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 4.504731861198739, | |
| "grad_norm": 100.14163208007812, | |
| "learning_rate": 4.873646209386281e-06, | |
| "loss": 0.0388, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.522757999098693, | |
| "grad_norm": 0.7780297994613647, | |
| "learning_rate": 4.693140794223827e-06, | |
| "loss": 0.0489, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 4.540784136998648, | |
| "grad_norm": 49.03726577758789, | |
| "learning_rate": 4.512635379061372e-06, | |
| "loss": 0.0548, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 4.558810274898603, | |
| "grad_norm": 54.46559143066406, | |
| "learning_rate": 4.332129963898917e-06, | |
| "loss": 0.038, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 4.576836412798558, | |
| "grad_norm": 34.701942443847656, | |
| "learning_rate": 4.151624548736462e-06, | |
| "loss": 0.0418, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 4.594862550698513, | |
| "grad_norm": 87.46144104003906, | |
| "learning_rate": 3.971119133574008e-06, | |
| "loss": 0.0768, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 4.612888688598468, | |
| "grad_norm": 86.79228210449219, | |
| "learning_rate": 3.7906137184115523e-06, | |
| "loss": 0.0274, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 4.630914826498422, | |
| "grad_norm": 18.837629318237305, | |
| "learning_rate": 3.6101083032490977e-06, | |
| "loss": 0.028, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 4.648940964398378, | |
| "grad_norm": 71.06475067138672, | |
| "learning_rate": 3.429602888086643e-06, | |
| "loss": 0.0434, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 4.666967102298333, | |
| "grad_norm": 36.30558395385742, | |
| "learning_rate": 3.2490974729241876e-06, | |
| "loss": 0.0457, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 4.6849932401982874, | |
| "grad_norm": 26.01364517211914, | |
| "learning_rate": 3.068592057761733e-06, | |
| "loss": 0.0248, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 4.703019378098243, | |
| "grad_norm": 65.48623657226562, | |
| "learning_rate": 2.8880866425992783e-06, | |
| "loss": 0.0402, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 4.721045515998197, | |
| "grad_norm": 54.232330322265625, | |
| "learning_rate": 2.7075812274368233e-06, | |
| "loss": 0.0196, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 4.7390716538981525, | |
| "grad_norm": 0.7030893564224243, | |
| "learning_rate": 2.527075812274368e-06, | |
| "loss": 0.036, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 4.757097791798107, | |
| "grad_norm": 5.075647830963135, | |
| "learning_rate": 2.3465703971119136e-06, | |
| "loss": 0.0163, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 4.775123929698062, | |
| "grad_norm": 30.817026138305664, | |
| "learning_rate": 2.1660649819494585e-06, | |
| "loss": 0.0351, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 4.793150067598017, | |
| "grad_norm": 1.8076817989349365, | |
| "learning_rate": 1.985559566787004e-06, | |
| "loss": 0.0154, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 4.811176205497972, | |
| "grad_norm": 9.753203392028809, | |
| "learning_rate": 1.8050541516245488e-06, | |
| "loss": 0.0244, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 4.829202343397927, | |
| "grad_norm": 20.44203758239746, | |
| "learning_rate": 1.6245487364620938e-06, | |
| "loss": 0.0357, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 4.847228481297882, | |
| "grad_norm": 23.524141311645508, | |
| "learning_rate": 1.4440433212996392e-06, | |
| "loss": 0.0469, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 4.865254619197837, | |
| "grad_norm": 115.92428588867188, | |
| "learning_rate": 1.263537906137184e-06, | |
| "loss": 0.0205, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 4.883280757097792, | |
| "grad_norm": 4.06036376953125, | |
| "learning_rate": 1.0830324909747293e-06, | |
| "loss": 0.0252, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 4.901306894997747, | |
| "grad_norm": 72.22532653808594, | |
| "learning_rate": 9.025270758122744e-07, | |
| "loss": 0.032, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 4.919333032897701, | |
| "grad_norm": 67.36674499511719, | |
| "learning_rate": 7.220216606498196e-07, | |
| "loss": 0.0533, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 4.937359170797657, | |
| "grad_norm": 39.58537292480469, | |
| "learning_rate": 5.415162454873646e-07, | |
| "loss": 0.0687, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 4.955385308697611, | |
| "grad_norm": 1.2294623851776123, | |
| "learning_rate": 3.610108303249098e-07, | |
| "loss": 0.0383, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 4.9734114465975665, | |
| "grad_norm": 58.341033935546875, | |
| "learning_rate": 1.805054151624549e-07, | |
| "loss": 0.0379, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 4.991437584497522, | |
| "grad_norm": 2.680495023727417, | |
| "learning_rate": 0.0, | |
| "loss": 0.014, | |
| "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 | |
| } | |