| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 1506, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 41.35739895402257, | |
| "learning_rate": 7.193423539345941e-06, | |
| "loss": 0.5141, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 10.443694874625423, | |
| "learning_rate": 9.358859796204429e-06, | |
| "loss": 0.4195, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 5.720021591145773, | |
| "learning_rate": 1.0625558804168632e-05, | |
| "loss": 0.2851, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 7.59250257002781, | |
| "learning_rate": 1.1524296053062918e-05, | |
| "loss": 0.2174, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 11.521550428114184, | |
| "learning_rate": 1.2221410821833392e-05, | |
| "loss": 0.1817, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 6.027998448017936, | |
| "learning_rate": 1.2790995061027121e-05, | |
| "loss": 0.1886, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 4.929816146115093, | |
| "learning_rate": 1.3272571673439616e-05, | |
| "loss": 0.1553, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 2.3881515971871345, | |
| "learning_rate": 1.3689732309921406e-05, | |
| "loss": 0.129, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 5.441118636680569, | |
| "learning_rate": 1.4057694068991321e-05, | |
| "loss": 0.1433, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 6.085213496930297, | |
| "learning_rate": 1.4386847078691883e-05, | |
| "loss": 0.1092, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 4.713361075579101, | |
| "learning_rate": 1.4684602194465794e-05, | |
| "loss": 0.1231, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 3.8259500358417924, | |
| "learning_rate": 1.495643131788561e-05, | |
| "loss": 0.0697, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 2.67920682727699, | |
| "learning_rate": 1.5206489871327869e-05, | |
| "loss": 0.084, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 8.189491553913532, | |
| "learning_rate": 1.54380079302981e-05, | |
| "loss": 0.1023, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 12.000369384166694, | |
| "learning_rate": 1.5653546086656083e-05, | |
| "loss": 0.0972, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 2.3726942012239953, | |
| "learning_rate": 1.5855168566779895e-05, | |
| "loss": 0.1036, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 2.0242117206977044, | |
| "learning_rate": 1.604456377435124e-05, | |
| "loss": 0.1081, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 2.2261934744219967, | |
| "learning_rate": 1.6223130325849813e-05, | |
| "loss": 0.082, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 4.70421073268978, | |
| "learning_rate": 1.6392039793463407e-05, | |
| "loss": 0.0744, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 12.064244277626111, | |
| "learning_rate": 1.6552283335550368e-05, | |
| "loss": 0.0934, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 2.150990470911233, | |
| "learning_rate": 1.67047069382623e-05, | |
| "loss": 0.0737, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 4.074258308547766, | |
| "learning_rate": 1.6850038451324284e-05, | |
| "loss": 0.0841, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 2.6764310841274237, | |
| "learning_rate": 1.6988908609137504e-05, | |
| "loss": 0.0821, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 3.1391283710987063, | |
| "learning_rate": 1.71218675747441e-05, | |
| "loss": 0.0747, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 3.1103728720325754, | |
| "learning_rate": 1.7249398104320845e-05, | |
| "loss": 0.0907, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 5.354173276825422, | |
| "learning_rate": 1.7371926128186358e-05, | |
| "loss": 0.0765, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 2.8677808267225005, | |
| "learning_rate": 1.7489829333814013e-05, | |
| "loss": 0.0777, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 2.6124301773304874, | |
| "learning_rate": 1.760344418715659e-05, | |
| "loss": 0.0897, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 2.915238130503912, | |
| "learning_rate": 1.7713071721324668e-05, | |
| "loss": 0.0866, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 4.061775729833908, | |
| "learning_rate": 1.781898234351457e-05, | |
| "loss": 0.0804, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 1.5648704857906242, | |
| "learning_rate": 1.7921419853452233e-05, | |
| "loss": 0.0834, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 2.225837991191409, | |
| "learning_rate": 1.8020604823638384e-05, | |
| "loss": 0.0646, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 4.041622096074851, | |
| "learning_rate": 1.811673745928848e-05, | |
| "loss": 0.0636, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 0.21100008000919218, | |
| "learning_rate": 1.821000003120973e-05, | |
| "loss": 0.0656, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 2.2923304434569443, | |
| "learning_rate": 1.8300558955927067e-05, | |
| "loss": 0.0756, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 3.6524729986322013, | |
| "learning_rate": 1.83885665827083e-05, | |
| "loss": 0.0466, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 2.1793660025907284, | |
| "learning_rate": 1.847416273569235e-05, | |
| "loss": 0.0704, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 4.363503091034068, | |
| "learning_rate": 1.8557476050321896e-05, | |
| "loss": 0.0712, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 3.0942807753734476, | |
| "learning_rate": 1.863862513615056e-05, | |
| "loss": 0.0831, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 1.789230871814937, | |
| "learning_rate": 1.8717719592408857e-05, | |
| "loss": 0.078, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 3.4361028432853677, | |
| "learning_rate": 1.879486089815082e-05, | |
| "loss": 0.0663, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 3.690439378170989, | |
| "learning_rate": 1.8870143195120794e-05, | |
| "loss": 0.0738, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 2.159404934400987, | |
| "learning_rate": 1.8943653978491198e-05, | |
| "loss": 0.0768, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 0.8042070009727055, | |
| "learning_rate": 1.901547470818277e-05, | |
| "loss": 0.0777, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 1.70758347995601, | |
| "learning_rate": 1.9085681351478775e-05, | |
| "loss": 0.05, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 1.2113560599237037, | |
| "learning_rate": 1.9154344865995993e-05, | |
| "loss": 0.0715, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 2.5816071085486527, | |
| "learning_rate": 1.9221531630710657e-05, | |
| "loss": 0.0688, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 1.4447841193953872, | |
| "learning_rate": 1.9287303831602588e-05, | |
| "loss": 0.0659, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 1.2668557011358101, | |
| "learning_rate": 1.9351719807533285e-05, | |
| "loss": 0.0515, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 2.0353893651728865, | |
| "learning_rate": 1.9414834361179333e-05, | |
| "loss": 0.0687, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 2.005863891715338, | |
| "learning_rate": 1.947669903917393e-05, | |
| "loss": 0.0459, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 1.4330303529877615, | |
| "learning_rate": 1.9537362385044847e-05, | |
| "loss": 0.0557, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 1.8683959370360874, | |
| "learning_rate": 1.959687016805845e-05, | |
| "loss": 0.0656, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 1.5963373312962135, | |
| "learning_rate": 1.9655265590672502e-05, | |
| "loss": 0.0519, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 1.824110595633705, | |
| "learning_rate": 1.9712589476953243e-05, | |
| "loss": 0.0557, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 2.024299349665376, | |
| "learning_rate": 1.976888044401508e-05, | |
| "loss": 0.0617, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 1.0490238284158413, | |
| "learning_rate": 1.98241750582861e-05, | |
| "loss": 0.0529, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 1.2602675746455845, | |
| "learning_rate": 1.9878507978183157e-05, | |
| "loss": 0.0617, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 1.7353304044327778, | |
| "learning_rate": 1.9931912084590654e-05, | |
| "loss": 0.0464, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 2.043608410331547, | |
| "learning_rate": 1.998441860037306e-05, | |
| "loss": 0.0716, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 0.9522735357022469, | |
| "learning_rate": 1.997786386275595e-05, | |
| "loss": 0.0496, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 1.3740017908881452, | |
| "learning_rate": 1.9940970300682533e-05, | |
| "loss": 0.0686, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 1.1533525092583308, | |
| "learning_rate": 1.9904076738609114e-05, | |
| "loss": 0.0562, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 1.564671647601224, | |
| "learning_rate": 1.9867183176535695e-05, | |
| "loss": 0.0553, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 0.8996693997352794, | |
| "learning_rate": 1.9830289614462276e-05, | |
| "loss": 0.0524, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 1.9340857910489522, | |
| "learning_rate": 1.979339605238886e-05, | |
| "loss": 0.0534, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 1.892682744806536, | |
| "learning_rate": 1.9756502490315442e-05, | |
| "loss": 0.0625, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 1.2063853192894494, | |
| "learning_rate": 1.9719608928242023e-05, | |
| "loss": 0.0519, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 1.984312822464147, | |
| "learning_rate": 1.9682715366168604e-05, | |
| "loss": 0.0494, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 1.6705446030280595, | |
| "learning_rate": 1.9645821804095185e-05, | |
| "loss": 0.059, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 1.7422620817223426, | |
| "learning_rate": 1.960892824202177e-05, | |
| "loss": 0.0454, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 1.3182961229762868, | |
| "learning_rate": 1.957203467994835e-05, | |
| "loss": 0.0581, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 1.330360819481426, | |
| "learning_rate": 1.9535141117874932e-05, | |
| "loss": 0.0575, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 1.9840741868184866, | |
| "learning_rate": 1.9498247555801517e-05, | |
| "loss": 0.063, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 1.6018064731760029, | |
| "learning_rate": 1.9461353993728094e-05, | |
| "loss": 0.0519, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 1.391561342963203, | |
| "learning_rate": 1.9424460431654675e-05, | |
| "loss": 0.0572, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 1.0319629863193043, | |
| "learning_rate": 1.938756686958126e-05, | |
| "loss": 0.0619, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 1.7398111190442345, | |
| "learning_rate": 1.935067330750784e-05, | |
| "loss": 0.05, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 1.675180749962208, | |
| "learning_rate": 1.9313779745434422e-05, | |
| "loss": 0.0654, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 1.486943722740635, | |
| "learning_rate": 1.9276886183361007e-05, | |
| "loss": 0.0463, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 1.1444156835752686, | |
| "learning_rate": 1.9239992621287588e-05, | |
| "loss": 0.0598, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 1.1260777714975718, | |
| "learning_rate": 1.920309905921417e-05, | |
| "loss": 0.039, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 1.328313971146618, | |
| "learning_rate": 1.916620549714075e-05, | |
| "loss": 0.0591, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 0.8805517771854091, | |
| "learning_rate": 1.912931193506733e-05, | |
| "loss": 0.0392, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 0.9704226543415952, | |
| "learning_rate": 1.9092418372993916e-05, | |
| "loss": 0.056, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 1.4328262810995938, | |
| "learning_rate": 1.9055524810920497e-05, | |
| "loss": 0.0602, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 2.6133255890647167, | |
| "learning_rate": 1.9018631248847078e-05, | |
| "loss": 0.0507, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 1.4574352859865667, | |
| "learning_rate": 1.898173768677366e-05, | |
| "loss": 0.0667, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 1.9116869451062006, | |
| "learning_rate": 1.894484412470024e-05, | |
| "loss": 0.0533, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 0.8152499611885836, | |
| "learning_rate": 1.890795056262682e-05, | |
| "loss": 0.0534, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 2.9440345318883696, | |
| "learning_rate": 1.8871057000553406e-05, | |
| "loss": 0.0419, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 2.7588896552457074, | |
| "learning_rate": 1.8834163438479987e-05, | |
| "loss": 0.0418, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 1.8886319208504987, | |
| "learning_rate": 1.8797269876406568e-05, | |
| "loss": 0.0397, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 1.8820615671771377, | |
| "learning_rate": 1.876037631433315e-05, | |
| "loss": 0.0454, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 1.2868214865586614, | |
| "learning_rate": 1.872348275225973e-05, | |
| "loss": 0.0373, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 2.016876778284122, | |
| "learning_rate": 1.8686589190186315e-05, | |
| "loss": 0.0619, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 0.8992764743601639, | |
| "learning_rate": 1.8649695628112896e-05, | |
| "loss": 0.0403, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 0.42416507318131186, | |
| "learning_rate": 1.8612802066039477e-05, | |
| "loss": 0.067, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 1.8601423331120188, | |
| "learning_rate": 1.8575908503966062e-05, | |
| "loss": 0.066, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 1.3893250379275068, | |
| "learning_rate": 1.8539014941892643e-05, | |
| "loss": 0.0437, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 2.3240219930504376, | |
| "learning_rate": 1.850212137981922e-05, | |
| "loss": 0.0488, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 0.8849896953543195, | |
| "learning_rate": 1.8465227817745805e-05, | |
| "loss": 0.0472, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 0.8655212477880712, | |
| "learning_rate": 1.8428334255672386e-05, | |
| "loss": 0.0546, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 1.2016665548799828, | |
| "learning_rate": 1.8391440693598967e-05, | |
| "loss": 0.0956, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 1.492659569010543, | |
| "learning_rate": 1.8354547131525552e-05, | |
| "loss": 0.0449, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 0.4224837643985645, | |
| "learning_rate": 1.8317653569452133e-05, | |
| "loss": 0.0545, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 1.8405573730480247, | |
| "learning_rate": 1.8280760007378714e-05, | |
| "loss": 0.0328, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 0.3988661237020132, | |
| "learning_rate": 1.8243866445305295e-05, | |
| "loss": 0.0451, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 0.5573396543240564, | |
| "learning_rate": 1.8206972883231876e-05, | |
| "loss": 0.0502, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 0.7855566721285819, | |
| "learning_rate": 1.817007932115846e-05, | |
| "loss": 0.0341, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 1.497005175064917, | |
| "learning_rate": 1.8133185759085042e-05, | |
| "loss": 0.0513, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 1.650975953086994, | |
| "learning_rate": 1.8096292197011623e-05, | |
| "loss": 0.0473, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 0.8418054071334755, | |
| "learning_rate": 1.8059398634938204e-05, | |
| "loss": 0.0444, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 1.33651471077315, | |
| "learning_rate": 1.8022505072864785e-05, | |
| "loss": 0.0476, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 1.1529220423121023, | |
| "learning_rate": 1.7985611510791367e-05, | |
| "loss": 0.0379, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 0.6912013035706749, | |
| "learning_rate": 1.794871794871795e-05, | |
| "loss": 0.0526, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 2.806793867054328, | |
| "learning_rate": 1.7911824386644532e-05, | |
| "loss": 0.047, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 3.323409487413899, | |
| "learning_rate": 1.7874930824571113e-05, | |
| "loss": 0.0455, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 1.0877783561592873, | |
| "learning_rate": 1.7838037262497695e-05, | |
| "loss": 0.0555, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 1.9158165193439103, | |
| "learning_rate": 1.7801143700424276e-05, | |
| "loss": 0.0458, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 1.0863638939536142, | |
| "learning_rate": 1.776425013835086e-05, | |
| "loss": 0.0462, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 2.3398031954044978, | |
| "learning_rate": 1.772735657627744e-05, | |
| "loss": 0.032, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 0.9272917966743286, | |
| "learning_rate": 1.7690463014204022e-05, | |
| "loss": 0.0464, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 1.4608044079217914, | |
| "learning_rate": 1.7653569452130607e-05, | |
| "loss": 0.0544, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 2.0799767984416713, | |
| "learning_rate": 1.7616675890057188e-05, | |
| "loss": 0.0629, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 2.438242481318271, | |
| "learning_rate": 1.757978232798377e-05, | |
| "loss": 0.0586, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 0.22666963149041416, | |
| "learning_rate": 1.754288876591035e-05, | |
| "loss": 0.0419, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 1.6906855856145777, | |
| "learning_rate": 1.750599520383693e-05, | |
| "loss": 0.0417, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 0.8417294291584847, | |
| "learning_rate": 1.7469101641763513e-05, | |
| "loss": 0.0496, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 2.384648750559231, | |
| "learning_rate": 1.7432208079690097e-05, | |
| "loss": 0.037, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 1.0969098360012755, | |
| "learning_rate": 1.7395314517616678e-05, | |
| "loss": 0.0392, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 4.265947825753472, | |
| "learning_rate": 1.735842095554326e-05, | |
| "loss": 0.0434, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 1.0043877066833349, | |
| "learning_rate": 1.732152739346984e-05, | |
| "loss": 0.0423, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 0.3834891081373628, | |
| "learning_rate": 1.728463383139642e-05, | |
| "loss": 0.0423, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 0.8917610552257311, | |
| "learning_rate": 1.7247740269323006e-05, | |
| "loss": 0.0491, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 1.7777935423085822, | |
| "learning_rate": 1.7210846707249587e-05, | |
| "loss": 0.0465, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 2.882686457165806, | |
| "learning_rate": 1.717395314517617e-05, | |
| "loss": 0.0366, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 1.4070944899475468, | |
| "learning_rate": 1.713705958310275e-05, | |
| "loss": 0.0364, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 1.368387839772382, | |
| "learning_rate": 1.710016602102933e-05, | |
| "loss": 0.0302, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 1.2202288141019488, | |
| "learning_rate": 1.7063272458955912e-05, | |
| "loss": 0.0455, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 1.7419694841079256, | |
| "learning_rate": 1.7026378896882496e-05, | |
| "loss": 0.0402, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 1.5792094366611027, | |
| "learning_rate": 1.6989485334809077e-05, | |
| "loss": 0.0364, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 1.0092569099059323, | |
| "learning_rate": 1.695259177273566e-05, | |
| "loss": 0.0335, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 1.5374731387141227, | |
| "learning_rate": 1.691569821066224e-05, | |
| "loss": 0.0428, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 1.6268956058587998, | |
| "learning_rate": 1.687880464858882e-05, | |
| "loss": 0.0483, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 1.6628348966888566, | |
| "learning_rate": 1.6841911086515402e-05, | |
| "loss": 0.0487, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 8.037103442054311, | |
| "learning_rate": 1.6805017524441987e-05, | |
| "loss": 0.0679, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 1.620270154467724, | |
| "learning_rate": 1.6768123962368568e-05, | |
| "loss": 0.0437, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 1.0436875513091504, | |
| "learning_rate": 1.6731230400295152e-05, | |
| "loss": 0.0346, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.8374706362800531, | |
| "learning_rate": 1.6694336838221733e-05, | |
| "loss": 0.0517, | |
| "step": 1500 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 6024, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 500, | |
| "total_flos": 0.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |