| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.12040939193257075, | |
| "eval_steps": 200, | |
| "global_step": 500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0012040939193257074, | |
| "grad_norm": 4.470886707305908, | |
| "learning_rate": 1.9975915221579964e-05, | |
| "loss": 2.6688, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.002408187838651415, | |
| "grad_norm": 4.067092418670654, | |
| "learning_rate": 1.9951830443159926e-05, | |
| "loss": 2.2509, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.003612281757977122, | |
| "grad_norm": 4.133108615875244, | |
| "learning_rate": 1.9927745664739885e-05, | |
| "loss": 2.3732, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.00481637567730283, | |
| "grad_norm": 3.4890763759613037, | |
| "learning_rate": 1.9903660886319848e-05, | |
| "loss": 2.3632, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.006020469596628537, | |
| "grad_norm": 4.1045308113098145, | |
| "learning_rate": 1.987957610789981e-05, | |
| "loss": 2.5203, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.007224563515954244, | |
| "grad_norm": 4.26784086227417, | |
| "learning_rate": 1.985549132947977e-05, | |
| "loss": 2.3349, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.008428657435279952, | |
| "grad_norm": 4.144766330718994, | |
| "learning_rate": 1.983140655105973e-05, | |
| "loss": 2.4106, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.00963275135460566, | |
| "grad_norm": 3.9538934230804443, | |
| "learning_rate": 1.9807321772639694e-05, | |
| "loss": 2.3192, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.010836845273931367, | |
| "grad_norm": 3.9219865798950195, | |
| "learning_rate": 1.9783236994219656e-05, | |
| "loss": 2.2012, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.012040939193257074, | |
| "grad_norm": 3.391493797302246, | |
| "learning_rate": 1.9759152215799615e-05, | |
| "loss": 2.4064, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.013245033112582781, | |
| "grad_norm": 4.393350124359131, | |
| "learning_rate": 1.9735067437379577e-05, | |
| "loss": 2.1513, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.014449127031908489, | |
| "grad_norm": 3.6243207454681396, | |
| "learning_rate": 1.971098265895954e-05, | |
| "loss": 2.2788, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.015653220951234198, | |
| "grad_norm": 3.642468214035034, | |
| "learning_rate": 1.9686897880539502e-05, | |
| "loss": 2.2244, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.016857314870559904, | |
| "grad_norm": 4.0894670486450195, | |
| "learning_rate": 1.966281310211946e-05, | |
| "loss": 2.3622, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.018061408789885613, | |
| "grad_norm": 4.033819198608398, | |
| "learning_rate": 1.9638728323699423e-05, | |
| "loss": 2.3803, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.01926550270921132, | |
| "grad_norm": 3.535987138748169, | |
| "learning_rate": 1.9614643545279385e-05, | |
| "loss": 2.2161, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.020469596628537028, | |
| "grad_norm": 3.541868209838867, | |
| "learning_rate": 1.9590558766859348e-05, | |
| "loss": 2.1721, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.021673690547862733, | |
| "grad_norm": 4.147072792053223, | |
| "learning_rate": 1.9566473988439307e-05, | |
| "loss": 2.3239, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.022877784467188442, | |
| "grad_norm": 3.4949986934661865, | |
| "learning_rate": 1.954238921001927e-05, | |
| "loss": 2.265, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.024081878386514148, | |
| "grad_norm": 3.793950319290161, | |
| "learning_rate": 1.951830443159923e-05, | |
| "loss": 2.152, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.025285972305839857, | |
| "grad_norm": 3.9355053901672363, | |
| "learning_rate": 1.949421965317919e-05, | |
| "loss": 2.2534, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.026490066225165563, | |
| "grad_norm": 3.255175828933716, | |
| "learning_rate": 1.9470134874759156e-05, | |
| "loss": 2.2971, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.027694160144491272, | |
| "grad_norm": 3.650298595428467, | |
| "learning_rate": 1.9446050096339115e-05, | |
| "loss": 2.1228, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.028898254063816978, | |
| "grad_norm": 3.1906814575195312, | |
| "learning_rate": 1.9421965317919077e-05, | |
| "loss": 2.0995, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.030102347983142687, | |
| "grad_norm": 3.8122494220733643, | |
| "learning_rate": 1.939788053949904e-05, | |
| "loss": 2.1651, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.031306441902468396, | |
| "grad_norm": 3.8269336223602295, | |
| "learning_rate": 1.9373795761079e-05, | |
| "loss": 2.1731, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.0325105358217941, | |
| "grad_norm": 3.75238037109375, | |
| "learning_rate": 1.934971098265896e-05, | |
| "loss": 2.3071, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.03371462974111981, | |
| "grad_norm": 3.538330078125, | |
| "learning_rate": 1.9325626204238923e-05, | |
| "loss": 2.3015, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.034918723660445516, | |
| "grad_norm": 3.497131586074829, | |
| "learning_rate": 1.9301541425818882e-05, | |
| "loss": 2.128, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.036122817579771226, | |
| "grad_norm": 3.6173276901245117, | |
| "learning_rate": 1.9277456647398845e-05, | |
| "loss": 2.1792, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.03732691149909693, | |
| "grad_norm": 3.2987892627716064, | |
| "learning_rate": 1.9253371868978807e-05, | |
| "loss": 2.0246, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.03853100541842264, | |
| "grad_norm": 3.1787831783294678, | |
| "learning_rate": 1.922928709055877e-05, | |
| "loss": 2.2108, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.039735099337748346, | |
| "grad_norm": 3.5422236919403076, | |
| "learning_rate": 1.920520231213873e-05, | |
| "loss": 2.1738, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.040939193257074055, | |
| "grad_norm": 3.7987539768218994, | |
| "learning_rate": 1.918111753371869e-05, | |
| "loss": 2.1161, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.04214328717639976, | |
| "grad_norm": 3.2058522701263428, | |
| "learning_rate": 1.9157032755298653e-05, | |
| "loss": 2.0808, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.04334738109572547, | |
| "grad_norm": 3.00519061088562, | |
| "learning_rate": 1.9132947976878615e-05, | |
| "loss": 2.1412, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.044551475015051176, | |
| "grad_norm": 3.4471330642700195, | |
| "learning_rate": 1.9108863198458578e-05, | |
| "loss": 2.1695, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.045755568934376885, | |
| "grad_norm": 3.394496440887451, | |
| "learning_rate": 1.9084778420038536e-05, | |
| "loss": 1.9532, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.04695966285370259, | |
| "grad_norm": 3.03004789352417, | |
| "learning_rate": 1.90606936416185e-05, | |
| "loss": 2.0659, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.048163756773028296, | |
| "grad_norm": 3.4260365962982178, | |
| "learning_rate": 1.903660886319846e-05, | |
| "loss": 2.0792, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.048163756773028296, | |
| "eval_loss": 2.1430513858795166, | |
| "eval_runtime": 16.4051, | |
| "eval_samples_per_second": 6.096, | |
| "eval_steps_per_second": 0.792, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.049367850692354005, | |
| "grad_norm": 4.670680999755859, | |
| "learning_rate": 1.901252408477842e-05, | |
| "loss": 2.0952, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.050571944611679714, | |
| "grad_norm": 3.510042667388916, | |
| "learning_rate": 1.8988439306358382e-05, | |
| "loss": 2.195, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.05177603853100542, | |
| "grad_norm": 3.0459847450256348, | |
| "learning_rate": 1.8964354527938345e-05, | |
| "loss": 2.2117, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.052980132450331126, | |
| "grad_norm": 4.36016321182251, | |
| "learning_rate": 1.8940269749518304e-05, | |
| "loss": 2.1191, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.054184226369656835, | |
| "grad_norm": 3.0498242378234863, | |
| "learning_rate": 1.891618497109827e-05, | |
| "loss": 2.0838, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.055388320288982544, | |
| "grad_norm": 3.218038558959961, | |
| "learning_rate": 1.889210019267823e-05, | |
| "loss": 2.1118, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.056592414208308246, | |
| "grad_norm": 3.3144683837890625, | |
| "learning_rate": 1.886801541425819e-05, | |
| "loss": 2.2176, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.057796508127633955, | |
| "grad_norm": 3.2364652156829834, | |
| "learning_rate": 1.8843930635838153e-05, | |
| "loss": 2.112, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.059000602046959665, | |
| "grad_norm": 3.291278839111328, | |
| "learning_rate": 1.8819845857418112e-05, | |
| "loss": 2.144, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.060204695966285374, | |
| "grad_norm": 3.65297794342041, | |
| "learning_rate": 1.8795761078998074e-05, | |
| "loss": 2.2597, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.061408789885611076, | |
| "grad_norm": 3.2321982383728027, | |
| "learning_rate": 1.8771676300578037e-05, | |
| "loss": 2.1618, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.06261288380493679, | |
| "grad_norm": 3.352842330932617, | |
| "learning_rate": 1.8747591522158e-05, | |
| "loss": 2.006, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.0638169777242625, | |
| "grad_norm": 3.5657215118408203, | |
| "learning_rate": 1.8723506743737958e-05, | |
| "loss": 2.2253, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.0650210716435882, | |
| "grad_norm": 3.060060739517212, | |
| "learning_rate": 1.869942196531792e-05, | |
| "loss": 2.1187, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.06622516556291391, | |
| "grad_norm": 3.473719835281372, | |
| "learning_rate": 1.8675337186897883e-05, | |
| "loss": 2.0299, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.06742925948223961, | |
| "grad_norm": 3.1167919635772705, | |
| "learning_rate": 1.8651252408477845e-05, | |
| "loss": 2.0381, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.06863335340156532, | |
| "grad_norm": 3.815816640853882, | |
| "learning_rate": 1.8627167630057804e-05, | |
| "loss": 2.1624, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.06983744732089103, | |
| "grad_norm": 3.2820959091186523, | |
| "learning_rate": 1.8603082851637766e-05, | |
| "loss": 2.0819, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.07104154124021674, | |
| "grad_norm": 3.568885087966919, | |
| "learning_rate": 1.857899807321773e-05, | |
| "loss": 2.0749, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.07224563515954245, | |
| "grad_norm": 3.424076795578003, | |
| "learning_rate": 1.855491329479769e-05, | |
| "loss": 2.129, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07344972907886815, | |
| "grad_norm": 3.2800493240356445, | |
| "learning_rate": 1.853082851637765e-05, | |
| "loss": 2.2067, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.07465382299819386, | |
| "grad_norm": 3.487868547439575, | |
| "learning_rate": 1.8506743737957612e-05, | |
| "loss": 2.124, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.07585791691751957, | |
| "grad_norm": 3.3999245166778564, | |
| "learning_rate": 1.8482658959537575e-05, | |
| "loss": 1.9888, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.07706201083684527, | |
| "grad_norm": 3.973482370376587, | |
| "learning_rate": 1.8458574181117533e-05, | |
| "loss": 2.0592, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.07826610475617098, | |
| "grad_norm": 2.9601657390594482, | |
| "learning_rate": 1.8434489402697496e-05, | |
| "loss": 2.1022, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.07947019867549669, | |
| "grad_norm": 3.260118246078491, | |
| "learning_rate": 1.8410404624277458e-05, | |
| "loss": 1.9763, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.0806742925948224, | |
| "grad_norm": 3.509838819503784, | |
| "learning_rate": 1.838631984585742e-05, | |
| "loss": 2.0284, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.08187838651414811, | |
| "grad_norm": 4.363494396209717, | |
| "learning_rate": 1.8362235067437383e-05, | |
| "loss": 2.0479, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.08308248043347381, | |
| "grad_norm": 3.2578630447387695, | |
| "learning_rate": 1.8338150289017342e-05, | |
| "loss": 2.0488, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.08428657435279951, | |
| "grad_norm": 3.2846531867980957, | |
| "learning_rate": 1.8314065510597304e-05, | |
| "loss": 2.0876, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.08549066827212523, | |
| "grad_norm": 3.3275203704833984, | |
| "learning_rate": 1.8289980732177266e-05, | |
| "loss": 2.0564, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.08669476219145093, | |
| "grad_norm": 3.1368625164031982, | |
| "learning_rate": 1.8265895953757225e-05, | |
| "loss": 2.1533, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.08789885611077664, | |
| "grad_norm": 3.3824191093444824, | |
| "learning_rate": 1.8241811175337188e-05, | |
| "loss": 2.1821, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.08910295003010235, | |
| "grad_norm": 3.5150134563446045, | |
| "learning_rate": 1.821772639691715e-05, | |
| "loss": 2.0292, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.09030704394942805, | |
| "grad_norm": 3.421921730041504, | |
| "learning_rate": 1.8193641618497112e-05, | |
| "loss": 1.9862, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.09151113786875377, | |
| "grad_norm": 3.616887092590332, | |
| "learning_rate": 1.8169556840077075e-05, | |
| "loss": 2.0158, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.09271523178807947, | |
| "grad_norm": 5.063056945800781, | |
| "learning_rate": 1.8145472061657034e-05, | |
| "loss": 2.0579, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.09391932570740517, | |
| "grad_norm": 3.5242559909820557, | |
| "learning_rate": 1.8121387283236996e-05, | |
| "loss": 2.0272, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.09512341962673089, | |
| "grad_norm": 3.2852962017059326, | |
| "learning_rate": 1.809730250481696e-05, | |
| "loss": 2.077, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.09632751354605659, | |
| "grad_norm": 3.710927963256836, | |
| "learning_rate": 1.8073217726396917e-05, | |
| "loss": 2.1271, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.09632751354605659, | |
| "eval_loss": 2.0655810832977295, | |
| "eval_runtime": 16.3755, | |
| "eval_samples_per_second": 6.107, | |
| "eval_steps_per_second": 0.794, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.0975316074653823, | |
| "grad_norm": 3.5019216537475586, | |
| "learning_rate": 1.804913294797688e-05, | |
| "loss": 2.1081, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.09873570138470801, | |
| "grad_norm": 3.5533690452575684, | |
| "learning_rate": 1.8025048169556842e-05, | |
| "loss": 2.0751, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.09993979530403371, | |
| "grad_norm": 3.4970240592956543, | |
| "learning_rate": 1.8000963391136804e-05, | |
| "loss": 2.066, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.10114388922335943, | |
| "grad_norm": 3.0926427841186523, | |
| "learning_rate": 1.7976878612716763e-05, | |
| "loss": 2.0516, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.10234798314268513, | |
| "grad_norm": 3.747452974319458, | |
| "learning_rate": 1.7952793834296726e-05, | |
| "loss": 2.0721, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.10355207706201083, | |
| "grad_norm": 3.3113677501678467, | |
| "learning_rate": 1.7928709055876688e-05, | |
| "loss": 2.1527, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.10475617098133655, | |
| "grad_norm": 3.357912063598633, | |
| "learning_rate": 1.7904624277456647e-05, | |
| "loss": 2.0113, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.10596026490066225, | |
| "grad_norm": 3.023893356323242, | |
| "learning_rate": 1.7880539499036613e-05, | |
| "loss": 2.1332, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.10716435881998795, | |
| "grad_norm": 3.3027355670928955, | |
| "learning_rate": 1.785645472061657e-05, | |
| "loss": 1.9699, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.10836845273931367, | |
| "grad_norm": 5.3524932861328125, | |
| "learning_rate": 1.7832369942196534e-05, | |
| "loss": 2.0182, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.10957254665863937, | |
| "grad_norm": 3.200258731842041, | |
| "learning_rate": 1.7808285163776496e-05, | |
| "loss": 2.007, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.11077664057796509, | |
| "grad_norm": 3.286268949508667, | |
| "learning_rate": 1.7784200385356455e-05, | |
| "loss": 2.0907, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.11198073449729079, | |
| "grad_norm": 3.15291428565979, | |
| "learning_rate": 1.7760115606936417e-05, | |
| "loss": 2.0468, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.11318482841661649, | |
| "grad_norm": 3.3798069953918457, | |
| "learning_rate": 1.773603082851638e-05, | |
| "loss": 1.9927, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.11438892233594221, | |
| "grad_norm": 3.4220967292785645, | |
| "learning_rate": 1.771194605009634e-05, | |
| "loss": 2.1326, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.11559301625526791, | |
| "grad_norm": 3.379628896713257, | |
| "learning_rate": 1.76878612716763e-05, | |
| "loss": 1.9202, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.11679711017459361, | |
| "grad_norm": 3.3020846843719482, | |
| "learning_rate": 1.7663776493256263e-05, | |
| "loss": 2.1176, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.11800120409391933, | |
| "grad_norm": 3.2711665630340576, | |
| "learning_rate": 1.7639691714836226e-05, | |
| "loss": 2.0865, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.11920529801324503, | |
| "grad_norm": 3.239253520965576, | |
| "learning_rate": 1.7615606936416188e-05, | |
| "loss": 1.9284, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.12040939193257075, | |
| "grad_norm": 3.4960460662841797, | |
| "learning_rate": 1.7591522157996147e-05, | |
| "loss": 2.0088, | |
| "step": 500 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 4152, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 529407992473776.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |