| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9995138551288284, | |
| "eval_steps": 500, | |
| "global_step": 1028, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0009722897423432182, | |
| "grad_norm": 22.865641182810197, | |
| "learning_rate": 9.70873786407767e-08, | |
| "loss": 1.3413, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.004861448711716091, | |
| "grad_norm": 21.908798159393672, | |
| "learning_rate": 4.854368932038835e-07, | |
| "loss": 1.3405, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.009722897423432183, | |
| "grad_norm": 8.67999928561487, | |
| "learning_rate": 9.70873786407767e-07, | |
| "loss": 1.227, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.014584346135148274, | |
| "grad_norm": 9.222051962167232, | |
| "learning_rate": 1.4563106796116506e-06, | |
| "loss": 1.0774, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.019445794846864366, | |
| "grad_norm": 3.108065986341407, | |
| "learning_rate": 1.941747572815534e-06, | |
| "loss": 0.9546, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.024307243558580455, | |
| "grad_norm": 2.3305737963317545, | |
| "learning_rate": 2.427184466019418e-06, | |
| "loss": 0.8883, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.02916869227029655, | |
| "grad_norm": 2.166786564817553, | |
| "learning_rate": 2.912621359223301e-06, | |
| "loss": 0.8596, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03403014098201264, | |
| "grad_norm": 2.2200337024713366, | |
| "learning_rate": 3.398058252427185e-06, | |
| "loss": 0.8302, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.03889158969372873, | |
| "grad_norm": 2.266283635100358, | |
| "learning_rate": 3.883495145631068e-06, | |
| "loss": 0.82, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.043753038405444825, | |
| "grad_norm": 2.4391536905162696, | |
| "learning_rate": 4.368932038834952e-06, | |
| "loss": 0.8, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.04861448711716091, | |
| "grad_norm": 2.3258135076743423, | |
| "learning_rate": 4.854368932038836e-06, | |
| "loss": 0.792, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.053475935828877004, | |
| "grad_norm": 2.4765334900301914, | |
| "learning_rate": 5.3398058252427185e-06, | |
| "loss": 0.7804, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.0583373845405931, | |
| "grad_norm": 2.7215966994164704, | |
| "learning_rate": 5.825242718446602e-06, | |
| "loss": 0.764, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.06319883325230918, | |
| "grad_norm": 2.4553563315970854, | |
| "learning_rate": 6.310679611650487e-06, | |
| "loss": 0.7521, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.06806028196402528, | |
| "grad_norm": 2.4367790074569675, | |
| "learning_rate": 6.79611650485437e-06, | |
| "loss": 0.7394, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.07292173067574137, | |
| "grad_norm": 2.335256384006913, | |
| "learning_rate": 7.2815533980582534e-06, | |
| "loss": 0.7134, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.07778317938745746, | |
| "grad_norm": 2.1323333889527953, | |
| "learning_rate": 7.766990291262136e-06, | |
| "loss": 0.7293, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.08264462809917356, | |
| "grad_norm": 2.1774984480511095, | |
| "learning_rate": 8.25242718446602e-06, | |
| "loss": 0.7164, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.08750607681088965, | |
| "grad_norm": 2.2022981209541608, | |
| "learning_rate": 8.737864077669904e-06, | |
| "loss": 0.7011, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.09236752552260574, | |
| "grad_norm": 2.312357582283152, | |
| "learning_rate": 9.223300970873788e-06, | |
| "loss": 0.7079, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.09722897423432182, | |
| "grad_norm": 2.1629606540297406, | |
| "learning_rate": 9.708737864077671e-06, | |
| "loss": 0.7021, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.10209042294603791, | |
| "grad_norm": 2.5455373567567054, | |
| "learning_rate": 9.999884650793835e-06, | |
| "loss": 0.6992, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.10695187165775401, | |
| "grad_norm": 2.3692741540462237, | |
| "learning_rate": 9.998587033345546e-06, | |
| "loss": 0.694, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.1118133203694701, | |
| "grad_norm": 2.1411842110512964, | |
| "learning_rate": 9.995847987378953e-06, | |
| "loss": 0.6975, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.1166747690811862, | |
| "grad_norm": 2.089784612791783, | |
| "learning_rate": 9.991668302745053e-06, | |
| "loss": 0.6838, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.12153621779290229, | |
| "grad_norm": 2.323386252470974, | |
| "learning_rate": 9.98604918472778e-06, | |
| "loss": 0.6777, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.12639766650461837, | |
| "grad_norm": 2.109074650826608, | |
| "learning_rate": 9.97899225369643e-06, | |
| "loss": 0.6804, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.13125911521633446, | |
| "grad_norm": 1.977824454327765, | |
| "learning_rate": 9.970499544638405e-06, | |
| "loss": 0.6828, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.13612056392805055, | |
| "grad_norm": 2.3675044930177314, | |
| "learning_rate": 9.960573506572391e-06, | |
| "loss": 0.6685, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.14098201263976665, | |
| "grad_norm": 2.135635558376172, | |
| "learning_rate": 9.949217001842128e-06, | |
| "loss": 0.6799, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.14584346135148274, | |
| "grad_norm": 2.1238487943590982, | |
| "learning_rate": 9.93643330529103e-06, | |
| "loss": 0.662, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.15070491006319883, | |
| "grad_norm": 2.0658607430202953, | |
| "learning_rate": 9.922226103317802e-06, | |
| "loss": 0.6682, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.15556635877491493, | |
| "grad_norm": 2.3142186772351936, | |
| "learning_rate": 9.906599492813413e-06, | |
| "loss": 0.6465, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.16042780748663102, | |
| "grad_norm": 2.078145593908007, | |
| "learning_rate": 9.889557979979695e-06, | |
| "loss": 0.6493, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.1652892561983471, | |
| "grad_norm": 2.174446805793003, | |
| "learning_rate": 9.871106479029889e-06, | |
| "loss": 0.6642, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.1701507049100632, | |
| "grad_norm": 2.1609913986701184, | |
| "learning_rate": 9.851250310771552e-06, | |
| "loss": 0.6563, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.1750121536217793, | |
| "grad_norm": 2.0544241967243533, | |
| "learning_rate": 9.829995201072217e-06, | |
| "loss": 0.6421, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.1798736023334954, | |
| "grad_norm": 2.1208283810533115, | |
| "learning_rate": 9.807347279208233e-06, | |
| "loss": 0.6427, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.18473505104521148, | |
| "grad_norm": 2.3054596970818593, | |
| "learning_rate": 9.783313076097285e-06, | |
| "loss": 0.6441, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.18959649975692758, | |
| "grad_norm": 2.002111307715066, | |
| "learning_rate": 9.75789952241509e-06, | |
| "loss": 0.6211, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.19445794846864364, | |
| "grad_norm": 2.0437914538851896, | |
| "learning_rate": 9.73111394659682e-06, | |
| "loss": 0.6499, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.19931939718035974, | |
| "grad_norm": 2.0833682723245106, | |
| "learning_rate": 9.702964072723825e-06, | |
| "loss": 0.6286, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.20418084589207583, | |
| "grad_norm": 1.8608505397920612, | |
| "learning_rate": 9.673458018296249e-06, | |
| "loss": 0.6214, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.20904229460379192, | |
| "grad_norm": 1.9565897157549554, | |
| "learning_rate": 9.642604291892227e-06, | |
| "loss": 0.6313, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.21390374331550802, | |
| "grad_norm": 2.0254219704696546, | |
| "learning_rate": 9.610411790714274e-06, | |
| "loss": 0.6294, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.2187651920272241, | |
| "grad_norm": 2.042861634862239, | |
| "learning_rate": 9.576889798023632e-06, | |
| "loss": 0.6062, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.2236266407389402, | |
| "grad_norm": 1.9416084530570248, | |
| "learning_rate": 9.54204798046328e-06, | |
| "loss": 0.6163, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.2284880894506563, | |
| "grad_norm": 2.1101068845725446, | |
| "learning_rate": 9.505896385270397e-06, | |
| "loss": 0.6086, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.2333495381623724, | |
| "grad_norm": 1.9316514600214862, | |
| "learning_rate": 9.468445437379054e-06, | |
| "loss": 0.6163, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.23821098687408848, | |
| "grad_norm": 2.31121577708352, | |
| "learning_rate": 9.42970593641402e-06, | |
| "loss": 0.6083, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.24307243558580457, | |
| "grad_norm": 2.084199435888248, | |
| "learning_rate": 9.389689053576497e-06, | |
| "loss": 0.5875, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.24793388429752067, | |
| "grad_norm": 2.0679175188925742, | |
| "learning_rate": 9.348406328422714e-06, | |
| "loss": 0.5889, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.25279533300923673, | |
| "grad_norm": 1.935193185903998, | |
| "learning_rate": 9.305869665536296e-06, | |
| "loss": 0.5978, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.25765678172095285, | |
| "grad_norm": 1.9881581060186517, | |
| "learning_rate": 9.262091331095375e-06, | |
| "loss": 0.5963, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.2625182304326689, | |
| "grad_norm": 1.9171404527587066, | |
| "learning_rate": 9.217083949335429e-06, | |
| "loss": 0.5826, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.26737967914438504, | |
| "grad_norm": 2.066905479586149, | |
| "learning_rate": 9.170860498908851e-06, | |
| "loss": 0.5868, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.2722411278561011, | |
| "grad_norm": 1.9405147945598347, | |
| "learning_rate": 9.12343430914236e-06, | |
| "loss": 0.5938, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.2771025765678172, | |
| "grad_norm": 2.042477110270263, | |
| "learning_rate": 9.07481905619323e-06, | |
| "loss": 0.5879, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.2819640252795333, | |
| "grad_norm": 2.316892218301555, | |
| "learning_rate": 9.025028759105558e-06, | |
| "loss": 0.5862, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.2868254739912494, | |
| "grad_norm": 1.9826669776669494, | |
| "learning_rate": 8.974077775767618e-06, | |
| "loss": 0.5795, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.2916869227029655, | |
| "grad_norm": 1.9928663954865473, | |
| "learning_rate": 8.921980798771521e-06, | |
| "loss": 0.5711, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.2965483714146816, | |
| "grad_norm": 1.9834504354777311, | |
| "learning_rate": 8.868752851176357e-06, | |
| "loss": 0.5797, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.30140982012639767, | |
| "grad_norm": 1.8922576118910497, | |
| "learning_rate": 8.814409282176029e-06, | |
| "loss": 0.5816, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.30627126883811373, | |
| "grad_norm": 2.054072819946173, | |
| "learning_rate": 8.758965762673065e-06, | |
| "loss": 0.5671, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.31113271754982985, | |
| "grad_norm": 1.9718238348510062, | |
| "learning_rate": 8.70243828075962e-06, | |
| "loss": 0.5582, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.3159941662615459, | |
| "grad_norm": 1.8711561533553072, | |
| "learning_rate": 8.644843137107058e-06, | |
| "loss": 0.5616, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.32085561497326204, | |
| "grad_norm": 2.0374989076214676, | |
| "learning_rate": 8.58619694026536e-06, | |
| "loss": 0.5593, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.3257170636849781, | |
| "grad_norm": 2.505113100765879, | |
| "learning_rate": 8.526516601873764e-06, | |
| "loss": 0.5492, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.3305785123966942, | |
| "grad_norm": 2.210634577584573, | |
| "learning_rate": 8.46581933178401e-06, | |
| "loss": 0.5564, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.3354399611084103, | |
| "grad_norm": 1.9498504768203495, | |
| "learning_rate": 8.404122633097573e-06, | |
| "loss": 0.5446, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.3403014098201264, | |
| "grad_norm": 2.1908777181373567, | |
| "learning_rate": 8.341444297118353e-06, | |
| "loss": 0.5435, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.3451628585318425, | |
| "grad_norm": 1.9450484392786167, | |
| "learning_rate": 8.27780239822224e-06, | |
| "loss": 0.5566, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.3500243072435586, | |
| "grad_norm": 1.9837258801331765, | |
| "learning_rate": 8.213215288645058e-06, | |
| "loss": 0.5267, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.35488575595527466, | |
| "grad_norm": 1.9536552831602378, | |
| "learning_rate": 8.147701593190384e-06, | |
| "loss": 0.5387, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.3597472046669908, | |
| "grad_norm": 1.9999100970286463, | |
| "learning_rate": 8.081280203858767e-06, | |
| "loss": 0.5331, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.36460865337870685, | |
| "grad_norm": 1.9212833886247676, | |
| "learning_rate": 8.01397027439989e-06, | |
| "loss": 0.5349, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.36947010209042297, | |
| "grad_norm": 1.9162389845327823, | |
| "learning_rate": 7.945791214789261e-06, | |
| "loss": 0.5221, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.37433155080213903, | |
| "grad_norm": 1.909711155316506, | |
| "learning_rate": 7.876762685631005e-06, | |
| "loss": 0.5235, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.37919299951385516, | |
| "grad_norm": 2.357077558485306, | |
| "learning_rate": 7.806904592488409e-06, | |
| "loss": 0.5222, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.3840544482255712, | |
| "grad_norm": 1.9488437106922831, | |
| "learning_rate": 7.736237080143788e-06, | |
| "loss": 0.5198, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.3889158969372873, | |
| "grad_norm": 1.9458627487121898, | |
| "learning_rate": 7.664780526789409e-06, | |
| "loss": 0.5145, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.3937773456490034, | |
| "grad_norm": 1.944072611413161, | |
| "learning_rate": 7.592555538151073e-06, | |
| "loss": 0.5136, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.3986387943607195, | |
| "grad_norm": 1.965724726537108, | |
| "learning_rate": 7.519582941546117e-06, | |
| "loss": 0.5235, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.4035002430724356, | |
| "grad_norm": 2.100470738223745, | |
| "learning_rate": 7.445883779877483e-06, | |
| "loss": 0.5094, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.40836169178415166, | |
| "grad_norm": 1.9100636260210968, | |
| "learning_rate": 7.371479305565644e-06, | |
| "loss": 0.5103, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.4132231404958678, | |
| "grad_norm": 1.9333255998991457, | |
| "learning_rate": 7.296390974420102e-06, | |
| "loss": 0.5054, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.41808458920758385, | |
| "grad_norm": 2.080473533066085, | |
| "learning_rate": 7.220640439452236e-06, | |
| "loss": 0.4959, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.42294603791929997, | |
| "grad_norm": 1.981039750107447, | |
| "learning_rate": 7.144249544631279e-06, | |
| "loss": 0.4991, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.42780748663101603, | |
| "grad_norm": 1.9695849449401175, | |
| "learning_rate": 7.067240318585242e-06, | |
| "loss": 0.499, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.43266893534273215, | |
| "grad_norm": 1.9718354318180367, | |
| "learning_rate": 6.989634968248578e-06, | |
| "loss": 0.4999, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.4375303840544482, | |
| "grad_norm": 1.849425307215673, | |
| "learning_rate": 6.911455872458423e-06, | |
| "loss": 0.5024, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.44239183276616434, | |
| "grad_norm": 1.968118907489162, | |
| "learning_rate": 6.832725575501287e-06, | |
| "loss": 0.4928, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.4472532814778804, | |
| "grad_norm": 2.052190888318457, | |
| "learning_rate": 6.753466780612008e-06, | |
| "loss": 0.4883, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.4521147301895965, | |
| "grad_norm": 2.023182275453982, | |
| "learning_rate": 6.673702343426894e-06, | |
| "loss": 0.4917, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.4569761789013126, | |
| "grad_norm": 1.9163489656538493, | |
| "learning_rate": 6.593455265392901e-06, | |
| "loss": 0.4805, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.46183762761302866, | |
| "grad_norm": 2.01148328208047, | |
| "learning_rate": 6.512748687134771e-06, | |
| "loss": 0.4848, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.4666990763247448, | |
| "grad_norm": 1.9062562183206269, | |
| "learning_rate": 6.431605881782043e-06, | |
| "loss": 0.4846, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.47156052503646084, | |
| "grad_norm": 2.1740484017488173, | |
| "learning_rate": 6.3500502482578296e-06, | |
| "loss": 0.4697, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.47642197374817696, | |
| "grad_norm": 1.9616349762922658, | |
| "learning_rate": 6.268105304531353e-06, | |
| "loss": 0.4781, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.48128342245989303, | |
| "grad_norm": 2.0065445987317467, | |
| "learning_rate": 6.185794680836124e-06, | |
| "loss": 0.4682, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.48614487117160915, | |
| "grad_norm": 2.1759803858020716, | |
| "learning_rate": 6.103142112855758e-06, | |
| "loss": 0.4669, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.4910063198833252, | |
| "grad_norm": 1.928661549886414, | |
| "learning_rate": 6.020171434879385e-06, | |
| "loss": 0.4635, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.49586776859504134, | |
| "grad_norm": 2.0042980585449324, | |
| "learning_rate": 5.936906572928625e-06, | |
| "loss": 0.4657, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.5007292173067575, | |
| "grad_norm": 1.9953252819560416, | |
| "learning_rate": 5.8533715378581e-06, | |
| "loss": 0.4642, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.5055906660184735, | |
| "grad_norm": 2.055778442556058, | |
| "learning_rate": 5.769590418431502e-06, | |
| "loss": 0.4586, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.5104521147301896, | |
| "grad_norm": 1.9430537005076507, | |
| "learning_rate": 5.685587374375176e-06, | |
| "loss": 0.4513, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.5153135634419057, | |
| "grad_norm": 1.850811801991838, | |
| "learning_rate": 5.601386629411247e-06, | |
| "loss": 0.4545, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.5201750121536218, | |
| "grad_norm": 1.8848477276717563, | |
| "learning_rate": 5.5170124642723035e-06, | |
| "loss": 0.4459, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.5250364608653378, | |
| "grad_norm": 2.0655655892354985, | |
| "learning_rate": 5.432489209699614e-06, | |
| "loss": 0.4485, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.529897909577054, | |
| "grad_norm": 1.8863958006941897, | |
| "learning_rate": 5.347841239426956e-06, | |
| "loss": 0.4353, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.5347593582887701, | |
| "grad_norm": 1.9547833187632682, | |
| "learning_rate": 5.263092963152021e-06, | |
| "loss": 0.4469, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.5396208070004861, | |
| "grad_norm": 1.9174269856279669, | |
| "learning_rate": 5.178268819497459e-06, | |
| "loss": 0.45, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.5444822557122022, | |
| "grad_norm": 2.0224149703614436, | |
| "learning_rate": 5.0933932689635855e-06, | |
| "loss": 0.4389, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.5493437044239183, | |
| "grad_norm": 1.9813759308607048, | |
| "learning_rate": 5.008490786874775e-06, | |
| "loss": 0.4344, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.5542051531356345, | |
| "grad_norm": 1.8864523302122398, | |
| "learning_rate": 4.923585856321577e-06, | |
| "loss": 0.4385, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.5590666018473505, | |
| "grad_norm": 1.915948818115916, | |
| "learning_rate": 4.8387029611005945e-06, | |
| "loss": 0.4402, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.5639280505590666, | |
| "grad_norm": 1.8637742571775495, | |
| "learning_rate": 4.753866578654171e-06, | |
| "loss": 0.4315, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.5687894992707827, | |
| "grad_norm": 1.9492458299998499, | |
| "learning_rate": 4.669101173011885e-06, | |
| "loss": 0.4262, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.5736509479824988, | |
| "grad_norm": 1.919366061689666, | |
| "learning_rate": 4.584431187735939e-06, | |
| "loss": 0.4329, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.5785123966942148, | |
| "grad_norm": 1.9206859765403341, | |
| "learning_rate": 4.499881038872424e-06, | |
| "loss": 0.4333, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.583373845405931, | |
| "grad_norm": 2.0207639364785663, | |
| "learning_rate": 4.415475107910553e-06, | |
| "loss": 0.4247, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.5882352941176471, | |
| "grad_norm": 1.9003063950335572, | |
| "learning_rate": 4.331237734751813e-06, | |
| "loss": 0.419, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.5930967428293632, | |
| "grad_norm": 1.9589720672960007, | |
| "learning_rate": 4.247193210691164e-06, | |
| "loss": 0.4135, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.5979581915410792, | |
| "grad_norm": 1.8725922589668766, | |
| "learning_rate": 4.1633657714122e-06, | |
| "loss": 0.4203, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.6028196402527953, | |
| "grad_norm": 1.8638498804130523, | |
| "learning_rate": 4.0797795899983984e-06, | |
| "loss": 0.4199, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.6076810889645115, | |
| "grad_norm": 1.896661777937066, | |
| "learning_rate": 3.9964587699623705e-06, | |
| "loss": 0.4083, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.6125425376762275, | |
| "grad_norm": 1.850040585103086, | |
| "learning_rate": 3.913427338295222e-06, | |
| "loss": 0.4132, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.6174039863879436, | |
| "grad_norm": 1.8523931506545597, | |
| "learning_rate": 3.830709238537938e-06, | |
| "loss": 0.4136, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.6222654350996597, | |
| "grad_norm": 1.8618206136881559, | |
| "learning_rate": 3.7483283238768685e-06, | |
| "loss": 0.4089, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.6271268838113758, | |
| "grad_norm": 1.975112166712253, | |
| "learning_rate": 3.6663083502652335e-06, | |
| "loss": 0.4063, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.6319883325230918, | |
| "grad_norm": 1.8034124348015945, | |
| "learning_rate": 3.5846729695727055e-06, | |
| "loss": 0.4022, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.636849781234808, | |
| "grad_norm": 1.8638316469191787, | |
| "learning_rate": 3.503445722764967e-06, | |
| "loss": 0.4087, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.6417112299465241, | |
| "grad_norm": 1.843833767339686, | |
| "learning_rate": 3.4226500331152843e-06, | |
| "loss": 0.4009, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.6465726786582402, | |
| "grad_norm": 1.9595878869963705, | |
| "learning_rate": 3.342309199449991e-06, | |
| "loss": 0.4068, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.6514341273699562, | |
| "grad_norm": 1.8378464923421844, | |
| "learning_rate": 3.262446389429883e-06, | |
| "loss": 0.3986, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.6562955760816723, | |
| "grad_norm": 1.8906212458541776, | |
| "learning_rate": 3.183084632869411e-06, | |
| "loss": 0.3993, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.6611570247933884, | |
| "grad_norm": 1.8741986531721049, | |
| "learning_rate": 3.104246815095653e-06, | |
| "loss": 0.3957, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.6660184735051046, | |
| "grad_norm": 1.9025410789293065, | |
| "learning_rate": 3.0259556703489245e-06, | |
| "loss": 0.3961, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.6708799222168206, | |
| "grad_norm": 1.8317452262969969, | |
| "learning_rate": 2.948233775226975e-06, | |
| "loss": 0.3936, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.6757413709285367, | |
| "grad_norm": 1.791486581405094, | |
| "learning_rate": 2.871103542174637e-06, | |
| "loss": 0.3954, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.6806028196402528, | |
| "grad_norm": 1.8421048675047038, | |
| "learning_rate": 2.794587213020813e-06, | |
| "loss": 0.392, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.6854642683519689, | |
| "grad_norm": 1.8053899992220983, | |
| "learning_rate": 2.7187068525646578e-06, | |
| "loss": 0.3884, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.690325717063685, | |
| "grad_norm": 1.8164816090358933, | |
| "learning_rate": 2.6434843422128225e-06, | |
| "loss": 0.3833, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.6951871657754011, | |
| "grad_norm": 1.8692336513214356, | |
| "learning_rate": 2.5689413736695623e-06, | |
| "loss": 0.3925, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.7000486144871172, | |
| "grad_norm": 1.8131226789865298, | |
| "learning_rate": 2.495099442681574e-06, | |
| "loss": 0.3772, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.7049100631988332, | |
| "grad_norm": 1.7456402646995377, | |
| "learning_rate": 2.4219798428393167e-06, | |
| "loss": 0.3836, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.7097715119105493, | |
| "grad_norm": 1.8417483163838206, | |
| "learning_rate": 2.3496036594366478e-06, | |
| "loss": 0.3767, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.7146329606222654, | |
| "grad_norm": 1.8419144501834137, | |
| "learning_rate": 2.2779917633905075e-06, | |
| "loss": 0.3798, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.7194944093339816, | |
| "grad_norm": 1.697011917829243, | |
| "learning_rate": 2.207164805222441e-06, | |
| "loss": 0.3731, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.7243558580456976, | |
| "grad_norm": 1.7913218626855516, | |
| "learning_rate": 2.1371432091036525e-06, | |
| "loss": 0.3695, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.7292173067574137, | |
| "grad_norm": 1.805255097420597, | |
| "learning_rate": 2.0679471669653596e-06, | |
| "loss": 0.3758, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.7340787554691298, | |
| "grad_norm": 1.67858386866423, | |
| "learning_rate": 1.999596632676087e-06, | |
| "loss": 0.3723, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.7389402041808459, | |
| "grad_norm": 1.6993602534801597, | |
| "learning_rate": 1.93211131628764e-06, | |
| "loss": 0.3684, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.743801652892562, | |
| "grad_norm": 1.6768650553126156, | |
| "learning_rate": 1.865510678351361e-06, | |
| "loss": 0.3664, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.7486631016042781, | |
| "grad_norm": 1.8063986711573161, | |
| "learning_rate": 1.7998139243063523e-06, | |
| "loss": 0.3664, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.7535245503159942, | |
| "grad_norm": 1.8433399163627073, | |
| "learning_rate": 1.7350399989412503e-06, | |
| "loss": 0.3792, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.7583859990277103, | |
| "grad_norm": 1.789706761318329, | |
| "learning_rate": 1.6712075809311801e-06, | |
| "loss": 0.3665, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.7632474477394263, | |
| "grad_norm": 1.874352972700248, | |
| "learning_rate": 1.6083350774514256e-06, | |
| "loss": 0.3588, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.7681088964511424, | |
| "grad_norm": 1.8708699807411355, | |
| "learning_rate": 1.5464406188694176e-06, | |
| "loss": 0.3596, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.7729703451628586, | |
| "grad_norm": 1.7622927599849374, | |
| "learning_rate": 1.4855420535165177e-06, | |
| "loss": 0.3706, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.7778317938745746, | |
| "grad_norm": 1.8033628264638553, | |
| "learning_rate": 1.4256569425411565e-06, | |
| "loss": 0.3627, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.7826932425862907, | |
| "grad_norm": 1.6982941151170463, | |
| "learning_rate": 1.3668025548447645e-06, | |
| "loss": 0.3654, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.7875546912980068, | |
| "grad_norm": 1.822085860012342, | |
| "learning_rate": 1.3089958621019966e-06, | |
| "loss": 0.3566, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.7924161400097229, | |
| "grad_norm": 1.6803332236549615, | |
| "learning_rate": 1.2522535338666487e-06, | |
| "loss": 0.3539, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.797277588721439, | |
| "grad_norm": 1.8018723375209953, | |
| "learning_rate": 1.1965919327647152e-06, | |
| "loss": 0.3624, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.8021390374331551, | |
| "grad_norm": 1.681272890467866, | |
| "learning_rate": 1.1420271097759339e-06, | |
| "loss": 0.3527, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.8070004861448712, | |
| "grad_norm": 1.7619365658437645, | |
| "learning_rate": 1.0885747996052203e-06, | |
| "loss": 0.353, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.8118619348565873, | |
| "grad_norm": 1.7697650095429134, | |
| "learning_rate": 1.0362504161452857e-06, | |
| "loss": 0.3571, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.8167233835683033, | |
| "grad_norm": 1.7834693090793727, | |
| "learning_rate": 9.850690480317837e-07, | |
| "loss": 0.3588, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.8215848322800194, | |
| "grad_norm": 1.6879014168829525, | |
| "learning_rate": 9.350454542922366e-07, | |
| "loss": 0.3485, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.8264462809917356, | |
| "grad_norm": 1.6631076557983835, | |
| "learning_rate": 8.861940600900215e-07, | |
| "loss": 0.3569, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.8313077297034517, | |
| "grad_norm": 1.7470295248017005, | |
| "learning_rate": 8.385289525646211e-07, | |
| "loss": 0.3545, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.8361691784151677, | |
| "grad_norm": 1.6694583459061383, | |
| "learning_rate": 7.920638767693606e-07, | |
| "loss": 0.3459, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.8410306271268838, | |
| "grad_norm": 1.812544085868088, | |
| "learning_rate": 7.468122317077786e-07, | |
| "loss": 0.3497, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.8458920758385999, | |
| "grad_norm": 1.804488776919829, | |
| "learning_rate": 7.027870664698011e-07, | |
| "loss": 0.3447, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.8507535245503159, | |
| "grad_norm": 1.6565427674580868, | |
| "learning_rate": 6.600010764688042e-07, | |
| "loss": 0.3488, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.8556149732620321, | |
| "grad_norm": 1.5660595611029826, | |
| "learning_rate": 6.184665997806832e-07, | |
| "loss": 0.3449, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.8604764219737482, | |
| "grad_norm": 1.628170781974643, | |
| "learning_rate": 5.781956135859446e-07, | |
| "loss": 0.3554, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.8653378706854643, | |
| "grad_norm": 1.6261129189211507, | |
| "learning_rate": 5.39199730715892e-07, | |
| "loss": 0.345, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.8701993193971803, | |
| "grad_norm": 1.6307410900682924, | |
| "learning_rate": 5.01490196303856e-07, | |
| "loss": 0.3456, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.8750607681088964, | |
| "grad_norm": 1.6901295083588348, | |
| "learning_rate": 4.650778845424758e-07, | |
| "loss": 0.3476, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.8799222168206126, | |
| "grad_norm": 1.730644298411666, | |
| "learning_rate": 4.2997329554792965e-07, | |
| "loss": 0.3532, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.8847836655323287, | |
| "grad_norm": 1.5909360839199094, | |
| "learning_rate": 3.961865523320557e-07, | |
| "loss": 0.3406, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.8896451142440447, | |
| "grad_norm": 1.7660527301033595, | |
| "learning_rate": 3.637273978831984e-07, | |
| "loss": 0.3434, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.8945065629557608, | |
| "grad_norm": 1.5431742144466063, | |
| "learning_rate": 3.326051923566559e-07, | |
| "loss": 0.3389, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.8993680116674769, | |
| "grad_norm": 1.5777885568012344, | |
| "learning_rate": 3.028289103755172e-07, | |
| "loss": 0.3438, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.904229460379193, | |
| "grad_norm": 1.6736588557089753, | |
| "learning_rate": 2.744071384426733e-07, | |
| "loss": 0.3456, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.9090909090909091, | |
| "grad_norm": 1.584446915524184, | |
| "learning_rate": 2.473480724647548e-07, | |
| "loss": 0.3519, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.9139523578026252, | |
| "grad_norm": 1.544706705307836, | |
| "learning_rate": 2.216595153886969e-07, | |
| "loss": 0.3504, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.9188138065143413, | |
| "grad_norm": 1.6069910572578268, | |
| "learning_rate": 1.9734887495163114e-07, | |
| "loss": 0.3396, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.9236752552260573, | |
| "grad_norm": 1.5262336921306658, | |
| "learning_rate": 1.7442316154473004e-07, | |
| "loss": 0.3391, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.9285367039377734, | |
| "grad_norm": 1.6151419038210972, | |
| "learning_rate": 1.528889861916477e-07, | |
| "loss": 0.3455, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.9333981526494896, | |
| "grad_norm": 1.7463253293156233, | |
| "learning_rate": 1.3275255864211245e-07, | |
| "loss": 0.342, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.9382596013612057, | |
| "grad_norm": 2.15764657474219, | |
| "learning_rate": 1.1401968558123977e-07, | |
| "loss": 0.3432, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.9431210500729217, | |
| "grad_norm": 1.5836218621495903, | |
| "learning_rate": 9.669576895507515e-08, | |
| "loss": 0.3434, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.9479824987846378, | |
| "grad_norm": 1.6606392678607504, | |
| "learning_rate": 8.078580441285067e-08, | |
| "loss": 0.3488, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.9528439474963539, | |
| "grad_norm": 1.634410010175144, | |
| "learning_rate": 6.629437986640397e-08, | |
| "loss": 0.348, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.95770539620807, | |
| "grad_norm": 1.6947942875406026, | |
| "learning_rate": 5.322567416717106e-08, | |
| "loss": 0.3356, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.9625668449197861, | |
| "grad_norm": 1.5244193912137922, | |
| "learning_rate": 4.158345590114965e-08, | |
| "loss": 0.3461, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.9674282936315022, | |
| "grad_norm": 1.6425083834973517, | |
| "learning_rate": 3.137108230215513e-08, | |
| "loss": 0.3489, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.9722897423432183, | |
| "grad_norm": 1.6739305879871746, | |
| "learning_rate": 2.259149828370999e-08, | |
| "loss": 0.3466, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.9771511910549344, | |
| "grad_norm": 1.6190798474108024, | |
| "learning_rate": 1.5247235589824772e-08, | |
| "loss": 0.3391, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.9820126397666504, | |
| "grad_norm": 1.72920925073275, | |
| "learning_rate": 9.340412064927084e-09, | |
| "loss": 0.3438, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.9868740884783666, | |
| "grad_norm": 1.6092917702604965, | |
| "learning_rate": 4.872731043143453e-09, | |
| "loss": 0.3454, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.9917355371900827, | |
| "grad_norm": 1.5519329352931979, | |
| "learning_rate": 1.845480857116111e-09, | |
| "loss": 0.3391, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.9965969859017987, | |
| "grad_norm": 1.6136231766922748, | |
| "learning_rate": 2.595344664868549e-10, | |
| "loss": 0.3413, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.9995138551288284, | |
| "eval_loss": 0.33695414662361145, | |
| "eval_runtime": 96.5912, | |
| "eval_samples_per_second": 3.127, | |
| "eval_steps_per_second": 0.787, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 0.9995138551288284, | |
| "step": 1028, | |
| "total_flos": 215189941125120.0, | |
| "train_loss": 0.5025305894098393, | |
| "train_runtime": 23224.052, | |
| "train_samples_per_second": 1.417, | |
| "train_steps_per_second": 0.044 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 1028, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 215189941125120.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |