| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.2408187838651415, | |
| "eval_steps": 200, | |
| "global_step": 1000, | |
| "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 | |
| }, | |
| { | |
| "epoch": 0.12161348585189645, | |
| "grad_norm": 3.337407350540161, | |
| "learning_rate": 1.756743737957611e-05, | |
| "loss": 1.9687, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.12281757977122215, | |
| "grad_norm": 3.534827709197998, | |
| "learning_rate": 1.754335260115607e-05, | |
| "loss": 2.0273, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.12402167369054787, | |
| "grad_norm": 3.6207938194274902, | |
| "learning_rate": 1.7519267822736034e-05, | |
| "loss": 2.0458, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.12522576760987358, | |
| "grad_norm": 3.396012544631958, | |
| "learning_rate": 1.7495183044315993e-05, | |
| "loss": 1.9185, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.12642986152919927, | |
| "grad_norm": 3.001236915588379, | |
| "learning_rate": 1.7471098265895955e-05, | |
| "loss": 1.9407, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.127633955448525, | |
| "grad_norm": 3.1318376064300537, | |
| "learning_rate": 1.7447013487475918e-05, | |
| "loss": 1.8984, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.1288380493678507, | |
| "grad_norm": 3.4541585445404053, | |
| "learning_rate": 1.7422928709055877e-05, | |
| "loss": 1.8846, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.1300421432871764, | |
| "grad_norm": 3.311082363128662, | |
| "learning_rate": 1.739884393063584e-05, | |
| "loss": 2.0015, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.1312462372065021, | |
| "grad_norm": 3.2366561889648438, | |
| "learning_rate": 1.73747591522158e-05, | |
| "loss": 2.0176, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.13245033112582782, | |
| "grad_norm": 3.123307943344116, | |
| "learning_rate": 1.735067437379576e-05, | |
| "loss": 1.9731, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.1336544250451535, | |
| "grad_norm": 3.776921033859253, | |
| "learning_rate": 1.7326589595375726e-05, | |
| "loss": 2.0484, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.13485851896447923, | |
| "grad_norm": 2.959716796875, | |
| "learning_rate": 1.7302504816955685e-05, | |
| "loss": 1.9689, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.13606261288380495, | |
| "grad_norm": 3.527384042739868, | |
| "learning_rate": 1.7278420038535647e-05, | |
| "loss": 1.9488, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.13726670680313063, | |
| "grad_norm": 3.0703189373016357, | |
| "learning_rate": 1.725433526011561e-05, | |
| "loss": 2.1226, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.13847080072245635, | |
| "grad_norm": 3.1028363704681396, | |
| "learning_rate": 1.723025048169557e-05, | |
| "loss": 1.8966, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.13967489464178207, | |
| "grad_norm": 3.340517044067383, | |
| "learning_rate": 1.720616570327553e-05, | |
| "loss": 2.2156, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.14087898856110775, | |
| "grad_norm": 3.2740213871002197, | |
| "learning_rate": 1.7182080924855493e-05, | |
| "loss": 1.9445, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.14208308248043347, | |
| "grad_norm": 3.240690231323242, | |
| "learning_rate": 1.7157996146435455e-05, | |
| "loss": 2.0295, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.1432871763997592, | |
| "grad_norm": 3.821340799331665, | |
| "learning_rate": 1.7133911368015418e-05, | |
| "loss": 2.1401, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.1444912703190849, | |
| "grad_norm": 3.103550910949707, | |
| "learning_rate": 1.7109826589595377e-05, | |
| "loss": 2.1164, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.1444912703190849, | |
| "eval_loss": 2.017059564590454, | |
| "eval_runtime": 16.4324, | |
| "eval_samples_per_second": 6.086, | |
| "eval_steps_per_second": 0.791, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.1456953642384106, | |
| "grad_norm": 3.4971117973327637, | |
| "learning_rate": 1.708574181117534e-05, | |
| "loss": 1.9864, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.1468994581577363, | |
| "grad_norm": 3.324803590774536, | |
| "learning_rate": 1.70616570327553e-05, | |
| "loss": 2.0402, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.14810355207706202, | |
| "grad_norm": 3.302614450454712, | |
| "learning_rate": 1.703757225433526e-05, | |
| "loss": 1.9494, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.1493076459963877, | |
| "grad_norm": 3.3090734481811523, | |
| "learning_rate": 1.7013487475915223e-05, | |
| "loss": 2.0748, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.15051173991571343, | |
| "grad_norm": 3.559049129486084, | |
| "learning_rate": 1.6989402697495185e-05, | |
| "loss": 1.9038, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.15171583383503914, | |
| "grad_norm": 3.5149178504943848, | |
| "learning_rate": 1.6965317919075147e-05, | |
| "loss": 2.0261, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.15291992775436483, | |
| "grad_norm": 3.835693120956421, | |
| "learning_rate": 1.6941233140655106e-05, | |
| "loss": 1.9453, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.15412402167369055, | |
| "grad_norm": 3.521132469177246, | |
| "learning_rate": 1.691714836223507e-05, | |
| "loss": 2.0138, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.15532811559301626, | |
| "grad_norm": 3.2369840145111084, | |
| "learning_rate": 1.689306358381503e-05, | |
| "loss": 2.0285, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.15653220951234195, | |
| "grad_norm": 3.1592392921447754, | |
| "learning_rate": 1.686897880539499e-05, | |
| "loss": 1.9912, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.15773630343166767, | |
| "grad_norm": 3.2069106101989746, | |
| "learning_rate": 1.6844894026974952e-05, | |
| "loss": 2.0159, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.15894039735099338, | |
| "grad_norm": 3.318230390548706, | |
| "learning_rate": 1.6820809248554915e-05, | |
| "loss": 2.0412, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.16014449127031907, | |
| "grad_norm": 3.549443244934082, | |
| "learning_rate": 1.6796724470134877e-05, | |
| "loss": 2.0014, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.1613485851896448, | |
| "grad_norm": 3.32999324798584, | |
| "learning_rate": 1.677263969171484e-05, | |
| "loss": 2.0303, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.1625526791089705, | |
| "grad_norm": 3.262946367263794, | |
| "learning_rate": 1.6748554913294798e-05, | |
| "loss": 1.9883, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.16375677302829622, | |
| "grad_norm": 3.484685182571411, | |
| "learning_rate": 1.672447013487476e-05, | |
| "loss": 1.9695, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.1649608669476219, | |
| "grad_norm": 3.4177358150482178, | |
| "learning_rate": 1.6700385356454723e-05, | |
| "loss": 2.0088, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.16616496086694763, | |
| "grad_norm": 3.447498321533203, | |
| "learning_rate": 1.6676300578034682e-05, | |
| "loss": 2.0813, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.16736905478627334, | |
| "grad_norm": 3.152740240097046, | |
| "learning_rate": 1.6652215799614644e-05, | |
| "loss": 1.9988, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.16857314870559903, | |
| "grad_norm": 3.8948824405670166, | |
| "learning_rate": 1.6628131021194607e-05, | |
| "loss": 2.0801, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.16977724262492475, | |
| "grad_norm": 3.81358003616333, | |
| "learning_rate": 1.660404624277457e-05, | |
| "loss": 1.944, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.17098133654425046, | |
| "grad_norm": 2.980236053466797, | |
| "learning_rate": 1.657996146435453e-05, | |
| "loss": 1.9151, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.17218543046357615, | |
| "grad_norm": 3.041680335998535, | |
| "learning_rate": 1.655587668593449e-05, | |
| "loss": 1.9486, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.17338952438290187, | |
| "grad_norm": 2.898974657058716, | |
| "learning_rate": 1.6531791907514452e-05, | |
| "loss": 2.1119, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.17459361830222758, | |
| "grad_norm": 3.161224603652954, | |
| "learning_rate": 1.6507707129094415e-05, | |
| "loss": 2.036, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.17579771222155327, | |
| "grad_norm": 3.2449426651000977, | |
| "learning_rate": 1.6483622350674374e-05, | |
| "loss": 2.0635, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.177001806140879, | |
| "grad_norm": 3.2805328369140625, | |
| "learning_rate": 1.6459537572254336e-05, | |
| "loss": 1.8022, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.1782059000602047, | |
| "grad_norm": 3.491149663925171, | |
| "learning_rate": 1.64354527938343e-05, | |
| "loss": 1.9832, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.1794099939795304, | |
| "grad_norm": 3.423267126083374, | |
| "learning_rate": 1.641136801541426e-05, | |
| "loss": 1.9574, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.1806140878988561, | |
| "grad_norm": 3.1914217472076416, | |
| "learning_rate": 1.638728323699422e-05, | |
| "loss": 1.9283, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.18181818181818182, | |
| "grad_norm": 3.2903149127960205, | |
| "learning_rate": 1.6363198458574182e-05, | |
| "loss": 2.1174, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.18302227573750754, | |
| "grad_norm": 3.113159656524658, | |
| "learning_rate": 1.6339113680154144e-05, | |
| "loss": 1.8794, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.18422636965683323, | |
| "grad_norm": 3.261596918106079, | |
| "learning_rate": 1.6315028901734103e-05, | |
| "loss": 2.0853, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.18543046357615894, | |
| "grad_norm": 2.9525296688079834, | |
| "learning_rate": 1.629094412331407e-05, | |
| "loss": 1.8622, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.18663455749548466, | |
| "grad_norm": 3.2103638648986816, | |
| "learning_rate": 1.6266859344894028e-05, | |
| "loss": 2.0536, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.18783865141481035, | |
| "grad_norm": 3.5312676429748535, | |
| "learning_rate": 1.624277456647399e-05, | |
| "loss": 1.9387, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.18904274533413606, | |
| "grad_norm": 3.277223825454712, | |
| "learning_rate": 1.6218689788053953e-05, | |
| "loss": 1.9804, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.19024683925346178, | |
| "grad_norm": 3.207287549972534, | |
| "learning_rate": 1.619460500963391e-05, | |
| "loss": 1.8782, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.19145093317278747, | |
| "grad_norm": 3.401834487915039, | |
| "learning_rate": 1.6170520231213874e-05, | |
| "loss": 1.9383, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.19265502709211318, | |
| "grad_norm": 3.5186078548431396, | |
| "learning_rate": 1.6146435452793836e-05, | |
| "loss": 2.0963, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.19265502709211318, | |
| "eval_loss": 1.979533076286316, | |
| "eval_runtime": 16.4348, | |
| "eval_samples_per_second": 6.085, | |
| "eval_steps_per_second": 0.791, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.1938591210114389, | |
| "grad_norm": 3.0080161094665527, | |
| "learning_rate": 1.6122350674373795e-05, | |
| "loss": 1.951, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.1950632149307646, | |
| "grad_norm": 3.124155044555664, | |
| "learning_rate": 1.6098265895953758e-05, | |
| "loss": 1.8663, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.1962673088500903, | |
| "grad_norm": 3.6262383460998535, | |
| "learning_rate": 1.607418111753372e-05, | |
| "loss": 1.9478, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.19747140276941602, | |
| "grad_norm": 3.3047947883605957, | |
| "learning_rate": 1.6050096339113682e-05, | |
| "loss": 1.9203, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.1986754966887417, | |
| "grad_norm": 3.0261447429656982, | |
| "learning_rate": 1.6026011560693645e-05, | |
| "loss": 1.8988, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.19987959060806743, | |
| "grad_norm": 4.233884334564209, | |
| "learning_rate": 1.6001926782273604e-05, | |
| "loss": 2.0327, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.20108368452739314, | |
| "grad_norm": 2.9169118404388428, | |
| "learning_rate": 1.5977842003853566e-05, | |
| "loss": 1.9264, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.20228777844671886, | |
| "grad_norm": 3.0078282356262207, | |
| "learning_rate": 1.5953757225433528e-05, | |
| "loss": 1.8821, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.20349187236604455, | |
| "grad_norm": 3.4188835620880127, | |
| "learning_rate": 1.592967244701349e-05, | |
| "loss": 1.933, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.20469596628537026, | |
| "grad_norm": 4.739987850189209, | |
| "learning_rate": 1.590558766859345e-05, | |
| "loss": 1.9182, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.20590006020469598, | |
| "grad_norm": 3.1810977458953857, | |
| "learning_rate": 1.5881502890173412e-05, | |
| "loss": 1.984, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.20710415412402167, | |
| "grad_norm": 3.174739360809326, | |
| "learning_rate": 1.5857418111753374e-05, | |
| "loss": 1.7719, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.20830824804334738, | |
| "grad_norm": 3.379767656326294, | |
| "learning_rate": 1.5833333333333333e-05, | |
| "loss": 1.9481, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.2095123419626731, | |
| "grad_norm": 3.3487260341644287, | |
| "learning_rate": 1.5809248554913295e-05, | |
| "loss": 1.9416, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.2107164358819988, | |
| "grad_norm": 3.4879958629608154, | |
| "learning_rate": 1.5785163776493258e-05, | |
| "loss": 2.0463, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.2119205298013245, | |
| "grad_norm": 3.2338194847106934, | |
| "learning_rate": 1.5761078998073217e-05, | |
| "loss": 1.9441, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.21312462372065022, | |
| "grad_norm": 3.122405529022217, | |
| "learning_rate": 1.5736994219653182e-05, | |
| "loss": 2.0644, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.2143287176399759, | |
| "grad_norm": 3.0773510932922363, | |
| "learning_rate": 1.571290944123314e-05, | |
| "loss": 1.9415, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.21553281155930162, | |
| "grad_norm": 3.004040241241455, | |
| "learning_rate": 1.5688824662813104e-05, | |
| "loss": 1.9305, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.21673690547862734, | |
| "grad_norm": 3.547109603881836, | |
| "learning_rate": 1.5664739884393066e-05, | |
| "loss": 2.1088, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.21794099939795303, | |
| "grad_norm": 3.1982204914093018, | |
| "learning_rate": 1.5640655105973025e-05, | |
| "loss": 1.8667, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.21914509331727874, | |
| "grad_norm": 3.381781578063965, | |
| "learning_rate": 1.5616570327552987e-05, | |
| "loss": 1.9783, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.22034918723660446, | |
| "grad_norm": 2.9775896072387695, | |
| "learning_rate": 1.559248554913295e-05, | |
| "loss": 2.0211, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.22155328115593018, | |
| "grad_norm": 2.864551067352295, | |
| "learning_rate": 1.5568400770712912e-05, | |
| "loss": 1.8579, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.22275737507525586, | |
| "grad_norm": 3.0532050132751465, | |
| "learning_rate": 1.5544315992292874e-05, | |
| "loss": 1.9398, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.22396146899458158, | |
| "grad_norm": 2.859631061553955, | |
| "learning_rate": 1.5520231213872833e-05, | |
| "loss": 1.8625, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.2251655629139073, | |
| "grad_norm": 3.1373536586761475, | |
| "learning_rate": 1.5496146435452796e-05, | |
| "loss": 1.9003, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.22636965683323299, | |
| "grad_norm": 3.3248465061187744, | |
| "learning_rate": 1.5472061657032758e-05, | |
| "loss": 2.0517, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.2275737507525587, | |
| "grad_norm": 3.5990936756134033, | |
| "learning_rate": 1.5447976878612717e-05, | |
| "loss": 1.8975, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.22877784467188442, | |
| "grad_norm": 3.4047725200653076, | |
| "learning_rate": 1.542389210019268e-05, | |
| "loss": 1.9786, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.2299819385912101, | |
| "grad_norm": 3.3326022624969482, | |
| "learning_rate": 1.539980732177264e-05, | |
| "loss": 1.7848, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.23118603251053582, | |
| "grad_norm": 3.2942848205566406, | |
| "learning_rate": 1.5375722543352604e-05, | |
| "loss": 1.8549, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.23239012642986154, | |
| "grad_norm": 3.4602601528167725, | |
| "learning_rate": 1.5351637764932563e-05, | |
| "loss": 1.917, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.23359422034918723, | |
| "grad_norm": 3.591327428817749, | |
| "learning_rate": 1.5327552986512525e-05, | |
| "loss": 1.9457, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.23479831426851294, | |
| "grad_norm": 3.215808868408203, | |
| "learning_rate": 1.5303468208092487e-05, | |
| "loss": 1.9261, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.23600240818783866, | |
| "grad_norm": 3.5032927989959717, | |
| "learning_rate": 1.5279383429672446e-05, | |
| "loss": 1.967, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.23720650210716435, | |
| "grad_norm": 3.1476144790649414, | |
| "learning_rate": 1.525529865125241e-05, | |
| "loss": 2.0448, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.23841059602649006, | |
| "grad_norm": 3.046126365661621, | |
| "learning_rate": 1.5231213872832371e-05, | |
| "loss": 1.9086, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.23961468994581578, | |
| "grad_norm": 3.0403099060058594, | |
| "learning_rate": 1.5207129094412332e-05, | |
| "loss": 1.9671, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.2408187838651415, | |
| "grad_norm": 3.524573802947998, | |
| "learning_rate": 1.5183044315992294e-05, | |
| "loss": 1.9532, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.2408187838651415, | |
| "eval_loss": 1.9571956396102905, | |
| "eval_runtime": 16.4441, | |
| "eval_samples_per_second": 6.081, | |
| "eval_steps_per_second": 0.791, | |
| "step": 1000 | |
| } | |
| ], | |
| "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": 1054167504365664.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |