| { | |
| "best_metric": 0.9820535451603413, | |
| "best_model_checkpoint": "wav2vec2-base-ft-keyword-spotting/checkpoint-1995", | |
| "epoch": 4.996869129618034, | |
| "eval_steps": 500, | |
| "global_step": 1995, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.025046963055729492, | |
| "grad_norm": 1.2714543342590332, | |
| "learning_rate": 1.5e-06, | |
| "loss": 2.4697, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.050093926111458985, | |
| "grad_norm": 1.3721944093704224, | |
| "learning_rate": 3e-06, | |
| "loss": 2.454, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.07514088916718847, | |
| "grad_norm": 1.3193830251693726, | |
| "learning_rate": 4.5e-06, | |
| "loss": 2.4043, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.10018785222291797, | |
| "grad_norm": 1.7374777793884277, | |
| "learning_rate": 6e-06, | |
| "loss": 2.3255, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.12523481527864747, | |
| "grad_norm": 2.109724760055542, | |
| "learning_rate": 7.5e-06, | |
| "loss": 2.216, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.15028177833437695, | |
| "grad_norm": 2.4082186222076416, | |
| "learning_rate": 9e-06, | |
| "loss": 2.057, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.17532874139010646, | |
| "grad_norm": 2.3386483192443848, | |
| "learning_rate": 1.05e-05, | |
| "loss": 1.8432, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.20037570444583594, | |
| "grad_norm": 2.037081241607666, | |
| "learning_rate": 1.2e-05, | |
| "loss": 1.7691, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.22542266750156542, | |
| "grad_norm": 2.0078208446502686, | |
| "learning_rate": 1.3500000000000001e-05, | |
| "loss": 1.6564, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.25046963055729493, | |
| "grad_norm": 1.5335993766784668, | |
| "learning_rate": 1.5e-05, | |
| "loss": 1.6115, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.27551659361302444, | |
| "grad_norm": 0.93914395570755, | |
| "learning_rate": 1.65e-05, | |
| "loss": 1.6253, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.3005635566687539, | |
| "grad_norm": 1.5028908252716064, | |
| "learning_rate": 1.8e-05, | |
| "loss": 1.5118, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.3256105197244834, | |
| "grad_norm": 0.5246427655220032, | |
| "learning_rate": 1.95e-05, | |
| "loss": 1.5028, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.3506574827802129, | |
| "grad_norm": 1.8485475778579712, | |
| "learning_rate": 2.1e-05, | |
| "loss": 1.5388, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.37570444583594237, | |
| "grad_norm": 1.7445107698440552, | |
| "learning_rate": 2.25e-05, | |
| "loss": 1.4807, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.4007514088916719, | |
| "grad_norm": 1.8197436332702637, | |
| "learning_rate": 2.4e-05, | |
| "loss": 1.4336, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.4257983719474014, | |
| "grad_norm": 2.3059756755828857, | |
| "learning_rate": 2.55e-05, | |
| "loss": 1.4139, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.45084533500313084, | |
| "grad_norm": 2.768622398376465, | |
| "learning_rate": 2.7000000000000002e-05, | |
| "loss": 1.3435, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.47589229805886035, | |
| "grad_norm": 2.9464476108551025, | |
| "learning_rate": 2.8499999999999998e-05, | |
| "loss": 1.3293, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.5009392611145899, | |
| "grad_norm": 4.094380855560303, | |
| "learning_rate": 3e-05, | |
| "loss": 1.2875, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.5259862241703194, | |
| "grad_norm": 3.006328582763672, | |
| "learning_rate": 2.9832869080779945e-05, | |
| "loss": 1.2165, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.5510331872260489, | |
| "grad_norm": 2.451118230819702, | |
| "learning_rate": 2.9665738161559886e-05, | |
| "loss": 1.1742, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.5760801502817783, | |
| "grad_norm": 5.200439929962158, | |
| "learning_rate": 2.9498607242339834e-05, | |
| "loss": 1.1409, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.6011271133375078, | |
| "grad_norm": 4.16045618057251, | |
| "learning_rate": 2.933147632311978e-05, | |
| "loss": 1.1796, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.6261740763932373, | |
| "grad_norm": 5.874985218048096, | |
| "learning_rate": 2.916434540389972e-05, | |
| "loss": 1.0734, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.6512210394489668, | |
| "grad_norm": 3.755768060684204, | |
| "learning_rate": 2.8997214484679665e-05, | |
| "loss": 1.0903, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.6762680025046963, | |
| "grad_norm": 4.306619644165039, | |
| "learning_rate": 2.8830083565459613e-05, | |
| "loss": 0.9576, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.7013149655604258, | |
| "grad_norm": 4.983129978179932, | |
| "learning_rate": 2.8662952646239554e-05, | |
| "loss": 0.8808, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.7263619286161553, | |
| "grad_norm": 3.138037919998169, | |
| "learning_rate": 2.84958217270195e-05, | |
| "loss": 0.8472, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.7514088916718847, | |
| "grad_norm": 4.114541530609131, | |
| "learning_rate": 2.8328690807799443e-05, | |
| "loss": 0.7928, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.7764558547276142, | |
| "grad_norm": 3.26853609085083, | |
| "learning_rate": 2.8161559888579388e-05, | |
| "loss": 0.8008, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.8015028177833438, | |
| "grad_norm": 5.10264253616333, | |
| "learning_rate": 2.7994428969359332e-05, | |
| "loss": 0.7371, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.8265497808390733, | |
| "grad_norm": 2.795498847961426, | |
| "learning_rate": 2.7827298050139277e-05, | |
| "loss": 0.6737, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.8515967438948028, | |
| "grad_norm": 3.0811805725097656, | |
| "learning_rate": 2.7660167130919218e-05, | |
| "loss": 0.6421, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.8766437069505323, | |
| "grad_norm": 3.24672794342041, | |
| "learning_rate": 2.7493036211699166e-05, | |
| "loss": 0.6054, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.9016906700062617, | |
| "grad_norm": 2.978050947189331, | |
| "learning_rate": 2.732590529247911e-05, | |
| "loss": 0.5662, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.9267376330619912, | |
| "grad_norm": 1.892045259475708, | |
| "learning_rate": 2.7158774373259055e-05, | |
| "loss": 0.5699, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.9517845961177207, | |
| "grad_norm": 2.637104034423828, | |
| "learning_rate": 2.6991643454038996e-05, | |
| "loss": 0.5033, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.9768315591734502, | |
| "grad_norm": 2.7039406299591064, | |
| "learning_rate": 2.6824512534818944e-05, | |
| "loss": 0.5147, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.9993738259236068, | |
| "eval_accuracy": 0.9664607237422771, | |
| "eval_loss": 0.3694746196269989, | |
| "eval_runtime": 21.9401, | |
| "eval_samples_per_second": 309.844, | |
| "eval_steps_per_second": 9.708, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 1.0018785222291797, | |
| "grad_norm": 3.884601593017578, | |
| "learning_rate": 2.665738161559889e-05, | |
| "loss": 0.5237, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.0269254852849092, | |
| "grad_norm": 3.843261957168579, | |
| "learning_rate": 2.649025069637883e-05, | |
| "loss": 0.4463, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.0519724483406387, | |
| "grad_norm": 7.544467926025391, | |
| "learning_rate": 2.6323119777158774e-05, | |
| "loss": 0.4465, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.0770194113963683, | |
| "grad_norm": 4.039328098297119, | |
| "learning_rate": 2.6155988857938722e-05, | |
| "loss": 0.4229, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.1020663744520978, | |
| "grad_norm": 2.7726874351501465, | |
| "learning_rate": 2.5988857938718663e-05, | |
| "loss": 0.4179, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.127113337507827, | |
| "grad_norm": 1.9234434366226196, | |
| "learning_rate": 2.5838440111420614e-05, | |
| "loss": 0.3878, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.1521603005635566, | |
| "grad_norm": 5.885218143463135, | |
| "learning_rate": 2.567130919220056e-05, | |
| "loss": 0.3734, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.177207263619286, | |
| "grad_norm": 4.204211235046387, | |
| "learning_rate": 2.55041782729805e-05, | |
| "loss": 0.3569, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.2022542266750156, | |
| "grad_norm": 3.173961639404297, | |
| "learning_rate": 2.5337047353760448e-05, | |
| "loss": 0.3598, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.227301189730745, | |
| "grad_norm": 2.828176975250244, | |
| "learning_rate": 2.5169916434540392e-05, | |
| "loss": 0.3744, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.2523481527864746, | |
| "grad_norm": 3.787830352783203, | |
| "learning_rate": 2.5002785515320333e-05, | |
| "loss": 0.3437, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.277395115842204, | |
| "grad_norm": 2.385486364364624, | |
| "learning_rate": 2.4835654596100278e-05, | |
| "loss": 0.3273, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.3024420788979336, | |
| "grad_norm": 2.802664279937744, | |
| "learning_rate": 2.4668523676880226e-05, | |
| "loss": 0.3129, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.3274890419536631, | |
| "grad_norm": 4.155833721160889, | |
| "learning_rate": 2.4501392757660167e-05, | |
| "loss": 0.3631, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.3525360050093926, | |
| "grad_norm": 4.156424522399902, | |
| "learning_rate": 2.433426183844011e-05, | |
| "loss": 0.2902, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.3775829680651221, | |
| "grad_norm": 2.7472970485687256, | |
| "learning_rate": 2.4167130919220056e-05, | |
| "loss": 0.3067, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.4026299311208517, | |
| "grad_norm": 2.5559287071228027, | |
| "learning_rate": 2.4e-05, | |
| "loss": 0.2819, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.4276768941765812, | |
| "grad_norm": 2.6116209030151367, | |
| "learning_rate": 2.3832869080779945e-05, | |
| "loss": 0.2995, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.4527238572323107, | |
| "grad_norm": 1.80980384349823, | |
| "learning_rate": 2.366573816155989e-05, | |
| "loss": 0.2733, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.4777708202880402, | |
| "grad_norm": 3.6102912425994873, | |
| "learning_rate": 2.349860724233983e-05, | |
| "loss": 0.3052, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.5028177833437697, | |
| "grad_norm": 2.1520848274230957, | |
| "learning_rate": 2.333147632311978e-05, | |
| "loss": 0.2605, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.5278647463994992, | |
| "grad_norm": 1.7574571371078491, | |
| "learning_rate": 2.3164345403899723e-05, | |
| "loss": 0.2805, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.5529117094552287, | |
| "grad_norm": 4.775511741638184, | |
| "learning_rate": 2.2997214484679665e-05, | |
| "loss": 0.2532, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.577958672510958, | |
| "grad_norm": 3.570802688598633, | |
| "learning_rate": 2.283008356545961e-05, | |
| "loss": 0.2524, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.6030056355666875, | |
| "grad_norm": 2.678054094314575, | |
| "learning_rate": 2.2662952646239557e-05, | |
| "loss": 0.2355, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.628052598622417, | |
| "grad_norm": 2.604247808456421, | |
| "learning_rate": 2.2495821727019498e-05, | |
| "loss": 0.257, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.6530995616781465, | |
| "grad_norm": 2.318293809890747, | |
| "learning_rate": 2.2328690807799443e-05, | |
| "loss": 0.2606, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.678146524733876, | |
| "grad_norm": 3.355053186416626, | |
| "learning_rate": 2.2161559888579387e-05, | |
| "loss": 0.2644, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.7031934877896056, | |
| "grad_norm": 3.6672332286834717, | |
| "learning_rate": 2.1994428969359335e-05, | |
| "loss": 0.2521, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.7282404508453348, | |
| "grad_norm": 5.440925121307373, | |
| "learning_rate": 2.1827298050139276e-05, | |
| "loss": 0.2612, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.7532874139010644, | |
| "grad_norm": 2.679652214050293, | |
| "learning_rate": 2.166016713091922e-05, | |
| "loss": 0.2291, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.7783343769567939, | |
| "grad_norm": 2.6038177013397217, | |
| "learning_rate": 2.1493036211699166e-05, | |
| "loss": 0.227, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.8033813400125234, | |
| "grad_norm": 3.784883499145508, | |
| "learning_rate": 2.1325905292479107e-05, | |
| "loss": 0.2404, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.8284283030682529, | |
| "grad_norm": 3.544004201889038, | |
| "learning_rate": 2.1158774373259055e-05, | |
| "loss": 0.2314, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.8534752661239824, | |
| "grad_norm": 1.186355471611023, | |
| "learning_rate": 2.0991643454039e-05, | |
| "loss": 0.2236, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.878522229179712, | |
| "grad_norm": 1.5770840644836426, | |
| "learning_rate": 2.082451253481894e-05, | |
| "loss": 0.2087, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.9035691922354414, | |
| "grad_norm": 2.307018995285034, | |
| "learning_rate": 2.0657381615598885e-05, | |
| "loss": 0.1937, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.928616155291171, | |
| "grad_norm": 3.615679979324341, | |
| "learning_rate": 2.0490250696378833e-05, | |
| "loss": 0.2061, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.9536631183469004, | |
| "grad_norm": 2.979759931564331, | |
| "learning_rate": 2.0323119777158774e-05, | |
| "loss": 0.2321, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.97871008140263, | |
| "grad_norm": 3.7641780376434326, | |
| "learning_rate": 2.015598885793872e-05, | |
| "loss": 0.2219, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.9987476518472134, | |
| "eval_accuracy": 0.9767578699617535, | |
| "eval_loss": 0.12758757174015045, | |
| "eval_runtime": 22.3439, | |
| "eval_samples_per_second": 304.245, | |
| "eval_steps_per_second": 9.533, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 2.0037570444583594, | |
| "grad_norm": 3.2667317390441895, | |
| "learning_rate": 1.9988857938718663e-05, | |
| "loss": 0.2024, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.028804007514089, | |
| "grad_norm": 2.963759422302246, | |
| "learning_rate": 1.9821727019498608e-05, | |
| "loss": 0.1808, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 2.0538509705698185, | |
| "grad_norm": 4.141463279724121, | |
| "learning_rate": 1.9654596100278552e-05, | |
| "loss": 0.194, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 2.078897933625548, | |
| "grad_norm": 4.703100681304932, | |
| "learning_rate": 1.9487465181058497e-05, | |
| "loss": 0.1963, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 2.1039448966812775, | |
| "grad_norm": 2.3061559200286865, | |
| "learning_rate": 1.9320334261838438e-05, | |
| "loss": 0.182, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 2.128991859737007, | |
| "grad_norm": 2.588878631591797, | |
| "learning_rate": 1.9153203342618386e-05, | |
| "loss": 0.197, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 2.1540388227927365, | |
| "grad_norm": 2.036632776260376, | |
| "learning_rate": 1.898607242339833e-05, | |
| "loss": 0.1939, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 2.179085785848466, | |
| "grad_norm": 2.999091625213623, | |
| "learning_rate": 1.8818941504178272e-05, | |
| "loss": 0.1776, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 2.2041327489041955, | |
| "grad_norm": 4.586688041687012, | |
| "learning_rate": 1.8651810584958216e-05, | |
| "loss": 0.2066, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 2.229179711959925, | |
| "grad_norm": 2.8724615573883057, | |
| "learning_rate": 1.8484679665738164e-05, | |
| "loss": 0.1814, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 2.254226675015654, | |
| "grad_norm": 3.051309585571289, | |
| "learning_rate": 1.8317548746518105e-05, | |
| "loss": 0.2035, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.279273638071384, | |
| "grad_norm": 2.697582721710205, | |
| "learning_rate": 1.815041782729805e-05, | |
| "loss": 0.2259, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 2.304320601127113, | |
| "grad_norm": 2.4961538314819336, | |
| "learning_rate": 1.7983286908077995e-05, | |
| "loss": 0.2069, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 2.329367564182843, | |
| "grad_norm": 4.160647392272949, | |
| "learning_rate": 1.781615598885794e-05, | |
| "loss": 0.1876, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 2.354414527238572, | |
| "grad_norm": 4.850248336791992, | |
| "learning_rate": 1.7649025069637884e-05, | |
| "loss": 0.1525, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 2.3794614902943017, | |
| "grad_norm": 2.7647998332977295, | |
| "learning_rate": 1.7481894150417828e-05, | |
| "loss": 0.1663, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.404508453350031, | |
| "grad_norm": 2.985328435897827, | |
| "learning_rate": 1.731476323119777e-05, | |
| "loss": 0.212, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 2.4295554164057607, | |
| "grad_norm": 2.1071982383728027, | |
| "learning_rate": 1.7147632311977717e-05, | |
| "loss": 0.1694, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 2.45460237946149, | |
| "grad_norm": 1.330075740814209, | |
| "learning_rate": 1.6980501392757662e-05, | |
| "loss": 0.1798, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 2.4796493425172197, | |
| "grad_norm": 3.0519397258758545, | |
| "learning_rate": 1.6813370473537606e-05, | |
| "loss": 0.1719, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 2.504696305572949, | |
| "grad_norm": 3.780168056488037, | |
| "learning_rate": 1.6646239554317548e-05, | |
| "loss": 0.174, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.5297432686286787, | |
| "grad_norm": 2.939603090286255, | |
| "learning_rate": 1.6479108635097496e-05, | |
| "loss": 0.1765, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 2.554790231684408, | |
| "grad_norm": 3.981560230255127, | |
| "learning_rate": 1.631197771587744e-05, | |
| "loss": 0.1951, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 2.5798371947401377, | |
| "grad_norm": 3.61944842338562, | |
| "learning_rate": 1.614484679665738e-05, | |
| "loss": 0.1394, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 2.6048841577958672, | |
| "grad_norm": 2.8106541633605957, | |
| "learning_rate": 1.5977715877437326e-05, | |
| "loss": 0.1767, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 2.6299311208515967, | |
| "grad_norm": 0.978993833065033, | |
| "learning_rate": 1.581058495821727e-05, | |
| "loss": 0.1438, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.6549780839073263, | |
| "grad_norm": 3.861020803451538, | |
| "learning_rate": 1.5643454038997215e-05, | |
| "loss": 0.1966, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 2.6800250469630558, | |
| "grad_norm": 6.368480682373047, | |
| "learning_rate": 1.547632311977716e-05, | |
| "loss": 0.1735, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 2.7050720100187853, | |
| "grad_norm": 2.9076144695281982, | |
| "learning_rate": 1.5309192200557104e-05, | |
| "loss": 0.1771, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 2.730118973074515, | |
| "grad_norm": 1.3361045122146606, | |
| "learning_rate": 1.5142061281337047e-05, | |
| "loss": 0.1594, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 2.7551659361302443, | |
| "grad_norm": 2.077012300491333, | |
| "learning_rate": 1.4974930362116992e-05, | |
| "loss": 0.1476, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.780212899185974, | |
| "grad_norm": 3.1275434494018555, | |
| "learning_rate": 1.4807799442896936e-05, | |
| "loss": 0.1965, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 2.8052598622417033, | |
| "grad_norm": 1.4369275569915771, | |
| "learning_rate": 1.464066852367688e-05, | |
| "loss": 0.1709, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 2.830306825297433, | |
| "grad_norm": 2.649810791015625, | |
| "learning_rate": 1.4473537604456825e-05, | |
| "loss": 0.1672, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 2.8553537883531623, | |
| "grad_norm": 2.733898401260376, | |
| "learning_rate": 1.4306406685236768e-05, | |
| "loss": 0.1614, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 2.8804007514088914, | |
| "grad_norm": 2.4066076278686523, | |
| "learning_rate": 1.4139275766016714e-05, | |
| "loss": 0.1798, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.9054477144646214, | |
| "grad_norm": 2.8622891902923584, | |
| "learning_rate": 1.3972144846796657e-05, | |
| "loss": 0.2037, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 2.9304946775203504, | |
| "grad_norm": 3.267059564590454, | |
| "learning_rate": 1.3805013927576602e-05, | |
| "loss": 0.1649, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 2.9555416405760804, | |
| "grad_norm": 1.7078909873962402, | |
| "learning_rate": 1.3637883008356546e-05, | |
| "loss": 0.1882, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 2.9805886036318094, | |
| "grad_norm": 4.093703746795654, | |
| "learning_rate": 1.3470752089136491e-05, | |
| "loss": 0.196, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 2.9981214777708205, | |
| "eval_accuracy": 0.980876728449544, | |
| "eval_loss": 0.09253185242414474, | |
| "eval_runtime": 23.1702, | |
| "eval_samples_per_second": 293.394, | |
| "eval_steps_per_second": 9.193, | |
| "step": 1197 | |
| }, | |
| { | |
| "epoch": 3.005635566687539, | |
| "grad_norm": 1.693821907043457, | |
| "learning_rate": 1.3303621169916434e-05, | |
| "loss": 0.1862, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 3.0306825297432685, | |
| "grad_norm": 3.7635231018066406, | |
| "learning_rate": 1.313649025069638e-05, | |
| "loss": 0.1663, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 3.055729492798998, | |
| "grad_norm": 1.3760706186294556, | |
| "learning_rate": 1.2969359331476323e-05, | |
| "loss": 0.1484, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 3.0807764558547275, | |
| "grad_norm": 3.305560350418091, | |
| "learning_rate": 1.2802228412256267e-05, | |
| "loss": 0.174, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 3.105823418910457, | |
| "grad_norm": 3.514143228530884, | |
| "learning_rate": 1.2635097493036212e-05, | |
| "loss": 0.1642, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 3.1308703819661865, | |
| "grad_norm": 2.355562210083008, | |
| "learning_rate": 1.2467966573816157e-05, | |
| "loss": 0.1697, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 3.155917345021916, | |
| "grad_norm": 1.1649140119552612, | |
| "learning_rate": 1.2300835654596101e-05, | |
| "loss": 0.173, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 3.1809643080776455, | |
| "grad_norm": 2.151714324951172, | |
| "learning_rate": 1.2133704735376046e-05, | |
| "loss": 0.1686, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 3.206011271133375, | |
| "grad_norm": 1.4168139696121216, | |
| "learning_rate": 1.1966573816155989e-05, | |
| "loss": 0.1604, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 3.2310582341891045, | |
| "grad_norm": 1.7652188539505005, | |
| "learning_rate": 1.1799442896935935e-05, | |
| "loss": 0.1311, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 3.256105197244834, | |
| "grad_norm": 3.6942124366760254, | |
| "learning_rate": 1.1632311977715878e-05, | |
| "loss": 0.1641, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 3.2811521603005636, | |
| "grad_norm": 3.0993881225585938, | |
| "learning_rate": 1.1465181058495822e-05, | |
| "loss": 0.1075, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 3.306199123356293, | |
| "grad_norm": 2.4146475791931152, | |
| "learning_rate": 1.1298050139275767e-05, | |
| "loss": 0.1708, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 3.3312460864120226, | |
| "grad_norm": 2.550797700881958, | |
| "learning_rate": 1.1130919220055711e-05, | |
| "loss": 0.1377, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 3.356293049467752, | |
| "grad_norm": 1.8913064002990723, | |
| "learning_rate": 1.0963788300835654e-05, | |
| "loss": 0.1567, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 3.3813400125234816, | |
| "grad_norm": 2.369048833847046, | |
| "learning_rate": 1.0796657381615599e-05, | |
| "loss": 0.1363, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 3.406386975579211, | |
| "grad_norm": 3.636960029602051, | |
| "learning_rate": 1.0629526462395543e-05, | |
| "loss": 0.1527, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 3.4314339386349406, | |
| "grad_norm": 3.4311366081237793, | |
| "learning_rate": 1.0462395543175486e-05, | |
| "loss": 0.1285, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 3.45648090169067, | |
| "grad_norm": 0.938562273979187, | |
| "learning_rate": 1.0295264623955432e-05, | |
| "loss": 0.175, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 3.4815278647463996, | |
| "grad_norm": 3.5569591522216797, | |
| "learning_rate": 1.0128133704735375e-05, | |
| "loss": 0.1461, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 3.506574827802129, | |
| "grad_norm": 1.755241870880127, | |
| "learning_rate": 9.961002785515322e-06, | |
| "loss": 0.1548, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 3.5316217908578587, | |
| "grad_norm": 3.8491158485412598, | |
| "learning_rate": 9.793871866295264e-06, | |
| "loss": 0.1507, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 3.5566687539135877, | |
| "grad_norm": 1.9301509857177734, | |
| "learning_rate": 9.626740947075209e-06, | |
| "loss": 0.1525, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 3.5817157169693177, | |
| "grad_norm": 3.393623113632202, | |
| "learning_rate": 9.459610027855154e-06, | |
| "loss": 0.1532, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 3.6067626800250467, | |
| "grad_norm": 3.466665267944336, | |
| "learning_rate": 9.292479108635098e-06, | |
| "loss": 0.1489, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 3.6318096430807767, | |
| "grad_norm": 0.6595897674560547, | |
| "learning_rate": 9.125348189415041e-06, | |
| "loss": 0.14, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 3.6568566061365058, | |
| "grad_norm": 1.5920480489730835, | |
| "learning_rate": 8.958217270194987e-06, | |
| "loss": 0.1535, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 3.6819035691922357, | |
| "grad_norm": 2.3759074211120605, | |
| "learning_rate": 8.79108635097493e-06, | |
| "loss": 0.1352, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 3.706950532247965, | |
| "grad_norm": 3.0922329425811768, | |
| "learning_rate": 8.623955431754875e-06, | |
| "loss": 0.1517, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 3.7319974953036943, | |
| "grad_norm": 6.3821210861206055, | |
| "learning_rate": 8.45682451253482e-06, | |
| "loss": 0.1335, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 3.757044458359424, | |
| "grad_norm": 3.3234517574310303, | |
| "learning_rate": 8.289693593314764e-06, | |
| "loss": 0.1547, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.7820914214151533, | |
| "grad_norm": 1.0703407526016235, | |
| "learning_rate": 8.122562674094707e-06, | |
| "loss": 0.1488, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 3.807138384470883, | |
| "grad_norm": 4.2433905601501465, | |
| "learning_rate": 7.955431754874653e-06, | |
| "loss": 0.1546, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 3.8321853475266123, | |
| "grad_norm": 3.813833475112915, | |
| "learning_rate": 7.788300835654596e-06, | |
| "loss": 0.1178, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 3.857232310582342, | |
| "grad_norm": 2.304304599761963, | |
| "learning_rate": 7.621169916434541e-06, | |
| "loss": 0.1388, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 3.8822792736380713, | |
| "grad_norm": 2.777663230895996, | |
| "learning_rate": 7.454038997214485e-06, | |
| "loss": 0.1385, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 3.907326236693801, | |
| "grad_norm": 3.856813430786133, | |
| "learning_rate": 7.2869080779944286e-06, | |
| "loss": 0.1248, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 3.9323731997495304, | |
| "grad_norm": 1.7799324989318848, | |
| "learning_rate": 7.119777158774373e-06, | |
| "loss": 0.1573, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 3.95742016280526, | |
| "grad_norm": 2.822288751602173, | |
| "learning_rate": 6.952646239554318e-06, | |
| "loss": 0.1469, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 3.9824671258609894, | |
| "grad_norm": 2.2535672187805176, | |
| "learning_rate": 6.785515320334261e-06, | |
| "loss": 0.1388, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.9788172992056487, | |
| "eval_loss": 0.09761285036802292, | |
| "eval_runtime": 22.1195, | |
| "eval_samples_per_second": 307.331, | |
| "eval_steps_per_second": 9.63, | |
| "step": 1597 | |
| }, | |
| { | |
| "epoch": 4.007514088916719, | |
| "grad_norm": 2.963806390762329, | |
| "learning_rate": 6.618384401114206e-06, | |
| "loss": 0.1605, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 4.032561051972448, | |
| "grad_norm": 1.9512594938278198, | |
| "learning_rate": 6.4512534818941505e-06, | |
| "loss": 0.1419, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 4.057608015028178, | |
| "grad_norm": 1.9299558401107788, | |
| "learning_rate": 6.284122562674095e-06, | |
| "loss": 0.1437, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 4.082654978083907, | |
| "grad_norm": 3.1459672451019287, | |
| "learning_rate": 6.116991643454039e-06, | |
| "loss": 0.1376, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 4.107701941139637, | |
| "grad_norm": 0.6871834993362427, | |
| "learning_rate": 5.949860724233983e-06, | |
| "loss": 0.1167, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 4.132748904195366, | |
| "grad_norm": 3.4506194591522217, | |
| "learning_rate": 5.782729805013928e-06, | |
| "loss": 0.1589, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 4.157795867251096, | |
| "grad_norm": 2.1339712142944336, | |
| "learning_rate": 5.615598885793872e-06, | |
| "loss": 0.1517, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 4.182842830306825, | |
| "grad_norm": 2.481374502182007, | |
| "learning_rate": 5.448467966573816e-06, | |
| "loss": 0.1726, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 4.207889793362555, | |
| "grad_norm": 3.550483465194702, | |
| "learning_rate": 5.281337047353761e-06, | |
| "loss": 0.1422, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 4.232936756418284, | |
| "grad_norm": 2.749545097351074, | |
| "learning_rate": 5.114206128133705e-06, | |
| "loss": 0.1275, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 4.257983719474014, | |
| "grad_norm": 4.703949928283691, | |
| "learning_rate": 4.947075208913649e-06, | |
| "loss": 0.1421, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 4.283030682529743, | |
| "grad_norm": 2.5116734504699707, | |
| "learning_rate": 4.7799442896935936e-06, | |
| "loss": 0.1101, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 4.308077645585473, | |
| "grad_norm": 3.076810121536255, | |
| "learning_rate": 4.612813370473538e-06, | |
| "loss": 0.0948, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 4.333124608641202, | |
| "grad_norm": 2.641812324523926, | |
| "learning_rate": 4.445682451253482e-06, | |
| "loss": 0.1119, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 4.358171571696932, | |
| "grad_norm": 2.6420016288757324, | |
| "learning_rate": 4.278551532033426e-06, | |
| "loss": 0.124, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 4.383218534752661, | |
| "grad_norm": 2.386286973953247, | |
| "learning_rate": 4.111420612813371e-06, | |
| "loss": 0.1498, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 4.408265497808391, | |
| "grad_norm": 1.9823393821716309, | |
| "learning_rate": 3.944289693593315e-06, | |
| "loss": 0.1205, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 4.43331246086412, | |
| "grad_norm": 1.9473960399627686, | |
| "learning_rate": 3.7771587743732592e-06, | |
| "loss": 0.1333, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 4.45835942391985, | |
| "grad_norm": 3.5217528343200684, | |
| "learning_rate": 3.6100278551532034e-06, | |
| "loss": 0.1328, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 4.483406386975579, | |
| "grad_norm": 1.426268219947815, | |
| "learning_rate": 3.4428969359331475e-06, | |
| "loss": 0.101, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 4.508453350031308, | |
| "grad_norm": 1.8567904233932495, | |
| "learning_rate": 3.275766016713092e-06, | |
| "loss": 0.1341, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 4.533500313087038, | |
| "grad_norm": 2.341360330581665, | |
| "learning_rate": 3.108635097493036e-06, | |
| "loss": 0.1096, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 4.558547276142768, | |
| "grad_norm": 2.002462387084961, | |
| "learning_rate": 2.9415041782729803e-06, | |
| "loss": 0.1397, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 4.583594239198497, | |
| "grad_norm": 1.7507251501083374, | |
| "learning_rate": 2.774373259052925e-06, | |
| "loss": 0.1081, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 4.608641202254226, | |
| "grad_norm": 1.9063440561294556, | |
| "learning_rate": 2.607242339832869e-06, | |
| "loss": 0.1053, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 4.633688165309956, | |
| "grad_norm": 3.434494733810425, | |
| "learning_rate": 2.4401114206128136e-06, | |
| "loss": 0.1492, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 4.658735128365686, | |
| "grad_norm": 2.072505235671997, | |
| "learning_rate": 2.2729805013927577e-06, | |
| "loss": 0.1578, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 4.683782091421415, | |
| "grad_norm": 1.0633418560028076, | |
| "learning_rate": 2.1058495821727023e-06, | |
| "loss": 0.1165, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 4.708829054477144, | |
| "grad_norm": 1.8562265634536743, | |
| "learning_rate": 1.9387186629526464e-06, | |
| "loss": 0.1282, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 4.733876017532874, | |
| "grad_norm": 1.3848246335983276, | |
| "learning_rate": 1.7715877437325906e-06, | |
| "loss": 0.1274, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 4.758922980588603, | |
| "grad_norm": 3.5492846965789795, | |
| "learning_rate": 1.604456824512535e-06, | |
| "loss": 0.1334, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 4.783969943644333, | |
| "grad_norm": 3.0531601905822754, | |
| "learning_rate": 1.437325905292479e-06, | |
| "loss": 0.1341, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 4.809016906700062, | |
| "grad_norm": 2.4751946926116943, | |
| "learning_rate": 1.2701949860724234e-06, | |
| "loss": 0.0987, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 4.834063869755792, | |
| "grad_norm": 1.2282112836837769, | |
| "learning_rate": 1.1030640668523677e-06, | |
| "loss": 0.1427, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 4.859110832811521, | |
| "grad_norm": 0.9631487131118774, | |
| "learning_rate": 9.35933147632312e-07, | |
| "loss": 0.1296, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 4.884157795867251, | |
| "grad_norm": 1.5988798141479492, | |
| "learning_rate": 7.688022284122563e-07, | |
| "loss": 0.1258, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 4.90920475892298, | |
| "grad_norm": 2.9330708980560303, | |
| "learning_rate": 6.016713091922006e-07, | |
| "loss": 0.1258, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 4.93425172197871, | |
| "grad_norm": 2.6305460929870605, | |
| "learning_rate": 4.3454038997214486e-07, | |
| "loss": 0.1257, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 4.959298685034439, | |
| "grad_norm": 1.8302196264266968, | |
| "learning_rate": 2.6740947075208915e-07, | |
| "loss": 0.1186, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 4.984345648090169, | |
| "grad_norm": 3.9666271209716797, | |
| "learning_rate": 1.0027855153203343e-07, | |
| "loss": 0.1444, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 4.996869129618034, | |
| "eval_accuracy": 0.9820535451603413, | |
| "eval_loss": 0.0860319584608078, | |
| "eval_runtime": 22.6177, | |
| "eval_samples_per_second": 300.561, | |
| "eval_steps_per_second": 9.417, | |
| "step": 1995 | |
| }, | |
| { | |
| "epoch": 4.996869129618034, | |
| "step": 1995, | |
| "total_flos": 2.3180194781952e+18, | |
| "train_loss": 0.41004490450510106, | |
| "train_runtime": 2324.9845, | |
| "train_samples_per_second": 109.88, | |
| "train_steps_per_second": 0.858 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 1995, | |
| "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": 2.3180194781952e+18, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |