| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 2.0, |
| "eval_steps": 500.0, |
| "global_step": 1876, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0010666666666666667, |
| "grad_norm": 27.125, |
| "learning_rate": 3.5087719298245616e-07, |
| "loss": 1.6233890056610107, |
| "step": 1, |
| "token_acc": 0.702493551160791 |
| }, |
| { |
| "epoch": 0.010666666666666666, |
| "grad_norm": 19.25, |
| "learning_rate": 3.5087719298245615e-06, |
| "loss": 1.4475451575385199, |
| "step": 10, |
| "token_acc": 0.7265573770491803 |
| }, |
| { |
| "epoch": 0.021333333333333333, |
| "grad_norm": 12.1875, |
| "learning_rate": 7.017543859649123e-06, |
| "loss": 1.2004226684570312, |
| "step": 20, |
| "token_acc": 0.7449092613103859 |
| }, |
| { |
| "epoch": 0.032, |
| "grad_norm": 7.5625, |
| "learning_rate": 1.0526315789473684e-05, |
| "loss": 0.9809289932250976, |
| "step": 30, |
| "token_acc": 0.7659592602062558 |
| }, |
| { |
| "epoch": 0.042666666666666665, |
| "grad_norm": 6.1875, |
| "learning_rate": 1.4035087719298246e-05, |
| "loss": 0.8323905944824219, |
| "step": 40, |
| "token_acc": 0.7854956640027206 |
| }, |
| { |
| "epoch": 0.05333333333333334, |
| "grad_norm": 6.59375, |
| "learning_rate": 1.754385964912281e-05, |
| "loss": 0.7661696434020996, |
| "step": 50, |
| "token_acc": 0.7932169071121236 |
| }, |
| { |
| "epoch": 0.064, |
| "grad_norm": 6.15625, |
| "learning_rate": 1.9999865771143805e-05, |
| "loss": 0.7235654354095459, |
| "step": 60, |
| "token_acc": 0.8000837871805614 |
| }, |
| { |
| "epoch": 0.07466666666666667, |
| "grad_norm": 6.5, |
| "learning_rate": 1.9997479580610527e-05, |
| "loss": 0.7175994396209717, |
| "step": 70, |
| "token_acc": 0.8002734808990685 |
| }, |
| { |
| "epoch": 0.08533333333333333, |
| "grad_norm": 5.90625, |
| "learning_rate": 1.9992111345867986e-05, |
| "loss": 0.6924509048461914, |
| "step": 80, |
| "token_acc": 0.8082805238698775 |
| }, |
| { |
| "epoch": 0.096, |
| "grad_norm": 5.3125, |
| "learning_rate": 1.9983762668151084e-05, |
| "loss": 0.6639269828796387, |
| "step": 90, |
| "token_acc": 0.814126705163612 |
| }, |
| { |
| "epoch": 0.10666666666666667, |
| "grad_norm": 5.6875, |
| "learning_rate": 1.9972436037700082e-05, |
| "loss": 0.6697088718414307, |
| "step": 100, |
| "token_acc": 0.8140579955909785 |
| }, |
| { |
| "epoch": 0.11733333333333333, |
| "grad_norm": 5.5, |
| "learning_rate": 1.9958134833017768e-05, |
| "loss": 0.6650930881500244, |
| "step": 110, |
| "token_acc": 0.8157225237449118 |
| }, |
| { |
| "epoch": 0.128, |
| "grad_norm": 5.5625, |
| "learning_rate": 1.9940863319861758e-05, |
| "loss": 0.6610164642333984, |
| "step": 120, |
| "token_acc": 0.8128639945186708 |
| }, |
| { |
| "epoch": 0.13866666666666666, |
| "grad_norm": 6.125, |
| "learning_rate": 1.992062664997209e-05, |
| "loss": 0.627209997177124, |
| "step": 130, |
| "token_acc": 0.8250212765957446 |
| }, |
| { |
| "epoch": 0.14933333333333335, |
| "grad_norm": 6.21875, |
| "learning_rate": 1.9897430859534553e-05, |
| "loss": 0.6545108795166016, |
| "step": 140, |
| "token_acc": 0.8140264586160109 |
| }, |
| { |
| "epoch": 0.16, |
| "grad_norm": 5.0625, |
| "learning_rate": 1.9871282867380245e-05, |
| "loss": 0.6527502536773682, |
| "step": 150, |
| "token_acc": 0.8172938078654548 |
| }, |
| { |
| "epoch": 0.17066666666666666, |
| "grad_norm": 6.09375, |
| "learning_rate": 1.9842190472921802e-05, |
| "loss": 0.6138824462890625, |
| "step": 160, |
| "token_acc": 0.8251860622462788 |
| }, |
| { |
| "epoch": 0.18133333333333335, |
| "grad_norm": 5.5, |
| "learning_rate": 1.9810162353827e-05, |
| "loss": 0.6213672637939454, |
| "step": 170, |
| "token_acc": 0.8211650979394556 |
| }, |
| { |
| "epoch": 0.192, |
| "grad_norm": 6.0, |
| "learning_rate": 1.977520806343039e-05, |
| "loss": 0.5975364685058594, |
| "step": 180, |
| "token_acc": 0.8272704146527601 |
| }, |
| { |
| "epoch": 0.20266666666666666, |
| "grad_norm": 5.59375, |
| "learning_rate": 1.9737338027883727e-05, |
| "loss": 0.6315351963043213, |
| "step": 190, |
| "token_acc": 0.8185825703628349 |
| }, |
| { |
| "epoch": 0.21333333333333335, |
| "grad_norm": 5.15625, |
| "learning_rate": 1.969656354304606e-05, |
| "loss": 0.6197149276733398, |
| "step": 200, |
| "token_acc": 0.8190742151138191 |
| }, |
| { |
| "epoch": 0.224, |
| "grad_norm": 5.8125, |
| "learning_rate": 1.9652896771114416e-05, |
| "loss": 0.6223199367523193, |
| "step": 210, |
| "token_acc": 0.8190856560163154 |
| }, |
| { |
| "epoch": 0.23466666666666666, |
| "grad_norm": 5.03125, |
| "learning_rate": 1.960635073699605e-05, |
| "loss": 0.637688159942627, |
| "step": 220, |
| "token_acc": 0.8177713409439428 |
| }, |
| { |
| "epoch": 0.24533333333333332, |
| "grad_norm": 5.8125, |
| "learning_rate": 1.95569393244234e-05, |
| "loss": 0.6076653003692627, |
| "step": 230, |
| "token_acc": 0.8262690567934328 |
| }, |
| { |
| "epoch": 0.256, |
| "grad_norm": 5.5, |
| "learning_rate": 1.950467727181284e-05, |
| "loss": 0.6383634567260742, |
| "step": 240, |
| "token_acc": 0.8165317426501917 |
| }, |
| { |
| "epoch": 0.26666666666666666, |
| "grad_norm": 5.15625, |
| "learning_rate": 1.9449580167868506e-05, |
| "loss": 0.6294333934783936, |
| "step": 250, |
| "token_acc": 0.819404255319149 |
| }, |
| { |
| "epoch": 0.2773333333333333, |
| "grad_norm": 5.34375, |
| "learning_rate": 1.939166444693252e-05, |
| "loss": 0.6120997428894043, |
| "step": 260, |
| "token_acc": 0.8252212389380531 |
| }, |
| { |
| "epoch": 0.288, |
| "grad_norm": 5.28125, |
| "learning_rate": 1.9330947384082957e-05, |
| "loss": 0.5947658538818359, |
| "step": 270, |
| "token_acc": 0.8248434591301405 |
| }, |
| { |
| "epoch": 0.2986666666666667, |
| "grad_norm": 5.21875, |
| "learning_rate": 1.9267447089981018e-05, |
| "loss": 0.5949877738952637, |
| "step": 280, |
| "token_acc": 0.8277702473737716 |
| }, |
| { |
| "epoch": 0.30933333333333335, |
| "grad_norm": 5.28125, |
| "learning_rate": 1.9201182505469035e-05, |
| "loss": 0.5923934936523437, |
| "step": 290, |
| "token_acc": 0.8287868519146052 |
| }, |
| { |
| "epoch": 0.32, |
| "grad_norm": 6.71875, |
| "learning_rate": 1.9132173395920756e-05, |
| "loss": 0.6005728721618653, |
| "step": 300, |
| "token_acc": 0.8250169721656483 |
| }, |
| { |
| "epoch": 0.33066666666666666, |
| "grad_norm": 5.5625, |
| "learning_rate": 1.9060440345345762e-05, |
| "loss": 0.614538049697876, |
| "step": 310, |
| "token_acc": 0.8181046916094001 |
| }, |
| { |
| "epoch": 0.3413333333333333, |
| "grad_norm": 4.71875, |
| "learning_rate": 1.8986004750249687e-05, |
| "loss": 0.6015767574310302, |
| "step": 320, |
| "token_acc": 0.8247483718176436 |
| }, |
| { |
| "epoch": 0.352, |
| "grad_norm": 5.5625, |
| "learning_rate": 1.890888881325205e-05, |
| "loss": 0.6035000801086425, |
| "step": 330, |
| "token_acc": 0.8262050237610319 |
| }, |
| { |
| "epoch": 0.3626666666666667, |
| "grad_norm": 5.9375, |
| "learning_rate": 1.8829115536463682e-05, |
| "loss": 0.5983945846557617, |
| "step": 340, |
| "token_acc": 0.8245257452574526 |
| }, |
| { |
| "epoch": 0.37333333333333335, |
| "grad_norm": 5.8125, |
| "learning_rate": 1.8746708714625668e-05, |
| "loss": 0.6272837162017822, |
| "step": 350, |
| "token_acc": 0.8214831155608349 |
| }, |
| { |
| "epoch": 0.384, |
| "grad_norm": 5.75, |
| "learning_rate": 1.8661692928011858e-05, |
| "loss": 0.5912134647369385, |
| "step": 360, |
| "token_acc": 0.828818151032848 |
| }, |
| { |
| "epoch": 0.39466666666666667, |
| "grad_norm": 5.65625, |
| "learning_rate": 1.8574093535097062e-05, |
| "loss": 0.573200798034668, |
| "step": 370, |
| "token_acc": 0.8316343866802582 |
| }, |
| { |
| "epoch": 0.4053333333333333, |
| "grad_norm": 5.84375, |
| "learning_rate": 1.8483936664993152e-05, |
| "loss": 0.5843992233276367, |
| "step": 380, |
| "token_acc": 0.8279888315424316 |
| }, |
| { |
| "epoch": 0.416, |
| "grad_norm": 5.5, |
| "learning_rate": 1.839124920965527e-05, |
| "loss": 0.5969943046569824, |
| "step": 390, |
| "token_acc": 0.8256973795435334 |
| }, |
| { |
| "epoch": 0.4266666666666667, |
| "grad_norm": 5.0625, |
| "learning_rate": 1.829605881586051e-05, |
| "loss": 0.5709249019622803, |
| "step": 400, |
| "token_acc": 0.8314263920671243 |
| }, |
| { |
| "epoch": 0.43733333333333335, |
| "grad_norm": 5.4375, |
| "learning_rate": 1.8198393876961446e-05, |
| "loss": 0.5546571731567382, |
| "step": 410, |
| "token_acc": 0.834034515325835 |
| }, |
| { |
| "epoch": 0.448, |
| "grad_norm": 7.5625, |
| "learning_rate": 1.8098283524416982e-05, |
| "loss": 0.5464757919311524, |
| "step": 420, |
| "token_acc": 0.8344332135995322 |
| }, |
| { |
| "epoch": 0.45866666666666667, |
| "grad_norm": 5.3125, |
| "learning_rate": 1.7995757619103012e-05, |
| "loss": 0.5686863899230957, |
| "step": 430, |
| "token_acc": 0.8333759046402724 |
| }, |
| { |
| "epoch": 0.4693333333333333, |
| "grad_norm": 5.0, |
| "learning_rate": 1.7890846742405558e-05, |
| "loss": 0.575782299041748, |
| "step": 440, |
| "token_acc": 0.8317190263623347 |
| }, |
| { |
| "epoch": 0.48, |
| "grad_norm": 5.53125, |
| "learning_rate": 1.7783582187098944e-05, |
| "loss": 0.5736880779266358, |
| "step": 450, |
| "token_acc": 0.8315915710125065 |
| }, |
| { |
| "epoch": 0.49066666666666664, |
| "grad_norm": 5.28125, |
| "learning_rate": 1.7673995948011807e-05, |
| "loss": 0.5740520000457764, |
| "step": 460, |
| "token_acc": 0.8287811539774668 |
| }, |
| { |
| "epoch": 0.5013333333333333, |
| "grad_norm": 5.15625, |
| "learning_rate": 1.75621207124837e-05, |
| "loss": 0.5786960124969482, |
| "step": 470, |
| "token_acc": 0.8299598599368008 |
| }, |
| { |
| "epoch": 0.512, |
| "grad_norm": 5.375, |
| "learning_rate": 1.7447989850615114e-05, |
| "loss": 0.5925831317901611, |
| "step": 480, |
| "token_acc": 0.8259212198221093 |
| }, |
| { |
| "epoch": 0.5226666666666666, |
| "grad_norm": 5.125, |
| "learning_rate": 1.733163740531386e-05, |
| "loss": 0.5791446685791015, |
| "step": 490, |
| "token_acc": 0.8287425149700599 |
| }, |
| { |
| "epoch": 0.5333333333333333, |
| "grad_norm": 5.46875, |
| "learning_rate": 1.7213098082140774e-05, |
| "loss": 0.5816606521606446, |
| "step": 500, |
| "token_acc": 0.8285325856062538 |
| }, |
| { |
| "epoch": 0.544, |
| "grad_norm": 5.34375, |
| "learning_rate": 1.7092407238957737e-05, |
| "loss": 0.5790934562683105, |
| "step": 510, |
| "token_acc": 0.8313161235717309 |
| }, |
| { |
| "epoch": 0.5546666666666666, |
| "grad_norm": 5.15625, |
| "learning_rate": 1.6969600875381172e-05, |
| "loss": 0.5422775268554687, |
| "step": 520, |
| "token_acc": 0.8432008189728715 |
| }, |
| { |
| "epoch": 0.5653333333333334, |
| "grad_norm": 5.125, |
| "learning_rate": 1.684471562204407e-05, |
| "loss": 0.5492764472961426, |
| "step": 530, |
| "token_acc": 0.8374480359718334 |
| }, |
| { |
| "epoch": 0.576, |
| "grad_norm": 5.1875, |
| "learning_rate": 1.6717788729669822e-05, |
| "loss": 0.5672526359558105, |
| "step": 540, |
| "token_acc": 0.8343422286639952 |
| }, |
| { |
| "epoch": 0.5866666666666667, |
| "grad_norm": 5.34375, |
| "learning_rate": 1.658885805796111e-05, |
| "loss": 0.5880655288696289, |
| "step": 550, |
| "token_acc": 0.828699779623665 |
| }, |
| { |
| "epoch": 0.5973333333333334, |
| "grad_norm": 5.40625, |
| "learning_rate": 1.6457962064307104e-05, |
| "loss": 0.5665912628173828, |
| "step": 560, |
| "token_acc": 0.8322487957407251 |
| }, |
| { |
| "epoch": 0.608, |
| "grad_norm": 5.75, |
| "learning_rate": 1.6325139792312414e-05, |
| "loss": 0.5551149368286132, |
| "step": 570, |
| "token_acc": 0.8370577281191807 |
| }, |
| { |
| "epoch": 0.6186666666666667, |
| "grad_norm": 5.3125, |
| "learning_rate": 1.6190430860151196e-05, |
| "loss": 0.5558858394622803, |
| "step": 580, |
| "token_acc": 0.8377869133264817 |
| }, |
| { |
| "epoch": 0.6293333333333333, |
| "grad_norm": 5.3125, |
| "learning_rate": 1.605387544874983e-05, |
| "loss": 0.5704761505126953, |
| "step": 590, |
| "token_acc": 0.83192066451941 |
| }, |
| { |
| "epoch": 0.64, |
| "grad_norm": 4.9375, |
| "learning_rate": 1.591551428980179e-05, |
| "loss": 0.5583576202392578, |
| "step": 600, |
| "token_acc": 0.834020358374695 |
| }, |
| { |
| "epoch": 0.6506666666666666, |
| "grad_norm": 5.71875, |
| "learning_rate": 1.5775388653618217e-05, |
| "loss": 0.5596559524536133, |
| "step": 610, |
| "token_acc": 0.8347915594441585 |
| }, |
| { |
| "epoch": 0.6613333333333333, |
| "grad_norm": 5.28125, |
| "learning_rate": 1.5633540336817804e-05, |
| "loss": 0.5696552276611329, |
| "step": 620, |
| "token_acc": 0.8363047001620746 |
| }, |
| { |
| "epoch": 0.672, |
| "grad_norm": 5.5, |
| "learning_rate": 1.549001164985973e-05, |
| "loss": 0.5665851593017578, |
| "step": 630, |
| "token_acc": 0.8350402031316123 |
| }, |
| { |
| "epoch": 0.6826666666666666, |
| "grad_norm": 5.28125, |
| "learning_rate": 1.5344845404423313e-05, |
| "loss": 0.6030998229980469, |
| "step": 640, |
| "token_acc": 0.8214255373375244 |
| }, |
| { |
| "epoch": 0.6933333333333334, |
| "grad_norm": 5.21875, |
| "learning_rate": 1.5198084900638161e-05, |
| "loss": 0.557160472869873, |
| "step": 650, |
| "token_acc": 0.8347296268088348 |
| }, |
| { |
| "epoch": 0.704, |
| "grad_norm": 5.78125, |
| "learning_rate": 1.5049773914168629e-05, |
| "loss": 0.5760011672973633, |
| "step": 660, |
| "token_acc": 0.8304249812139934 |
| }, |
| { |
| "epoch": 0.7146666666666667, |
| "grad_norm": 5.21875, |
| "learning_rate": 1.4899956683156413e-05, |
| "loss": 0.5680459976196289, |
| "step": 670, |
| "token_acc": 0.8291992103682088 |
| }, |
| { |
| "epoch": 0.7253333333333334, |
| "grad_norm": 6.625, |
| "learning_rate": 1.4748677895025243e-05, |
| "loss": 0.5647514343261719, |
| "step": 680, |
| "token_acc": 0.8340217482930119 |
| }, |
| { |
| "epoch": 0.736, |
| "grad_norm": 5.53125, |
| "learning_rate": 1.4595982673151511e-05, |
| "loss": 0.5595580101013183, |
| "step": 690, |
| "token_acc": 0.8372705642420122 |
| }, |
| { |
| "epoch": 0.7466666666666667, |
| "grad_norm": 5.625, |
| "learning_rate": 1.444191656340491e-05, |
| "loss": 0.551822280883789, |
| "step": 700, |
| "token_acc": 0.8342021187153187 |
| }, |
| { |
| "epoch": 0.7573333333333333, |
| "grad_norm": 5.0625, |
| "learning_rate": 1.4286525520563003e-05, |
| "loss": 0.5694937705993652, |
| "step": 710, |
| "token_acc": 0.8321737654844731 |
| }, |
| { |
| "epoch": 0.768, |
| "grad_norm": 4.90625, |
| "learning_rate": 1.4129855894603885e-05, |
| "loss": 0.5161600112915039, |
| "step": 720, |
| "token_acc": 0.8459770114942529 |
| }, |
| { |
| "epoch": 0.7786666666666666, |
| "grad_norm": 5.0625, |
| "learning_rate": 1.3971954416880897e-05, |
| "loss": 0.5634379863739014, |
| "step": 730, |
| "token_acc": 0.833432531247343 |
| }, |
| { |
| "epoch": 0.7893333333333333, |
| "grad_norm": 4.6875, |
| "learning_rate": 1.3812868186183641e-05, |
| "loss": 0.5540994644165039, |
| "step": 740, |
| "token_acc": 0.8377988807868407 |
| }, |
| { |
| "epoch": 0.8, |
| "grad_norm": 4.9375, |
| "learning_rate": 1.3652644654689355e-05, |
| "loss": 0.5608715057373047, |
| "step": 750, |
| "token_acc": 0.8339112050739957 |
| }, |
| { |
| "epoch": 0.8106666666666666, |
| "grad_norm": 5.40625, |
| "learning_rate": 1.3491331613808909e-05, |
| "loss": 0.5746842384338379, |
| "step": 760, |
| "token_acc": 0.8306396733582851 |
| }, |
| { |
| "epoch": 0.8213333333333334, |
| "grad_norm": 5.21875, |
| "learning_rate": 1.3328977179931595e-05, |
| "loss": 0.5586155891418457, |
| "step": 770, |
| "token_acc": 0.8344169551633877 |
| }, |
| { |
| "epoch": 0.832, |
| "grad_norm": 5.09375, |
| "learning_rate": 1.3165629780072991e-05, |
| "loss": 0.5604053020477295, |
| "step": 780, |
| "token_acc": 0.8320225679603351 |
| }, |
| { |
| "epoch": 0.8426666666666667, |
| "grad_norm": 5.03125, |
| "learning_rate": 1.3001338137430163e-05, |
| "loss": 0.549941110610962, |
| "step": 790, |
| "token_acc": 0.8354163148116872 |
| }, |
| { |
| "epoch": 0.8533333333333334, |
| "grad_norm": 5.21875, |
| "learning_rate": 1.2836151256848562e-05, |
| "loss": 0.5661550521850586, |
| "step": 800, |
| "token_acc": 0.8333191958605479 |
| }, |
| { |
| "epoch": 0.864, |
| "grad_norm": 5.84375, |
| "learning_rate": 1.2670118410204819e-05, |
| "loss": 0.5822794437408447, |
| "step": 810, |
| "token_acc": 0.8300913396481732 |
| }, |
| { |
| "epoch": 0.8746666666666667, |
| "grad_norm": 5.03125, |
| "learning_rate": 1.250328912171001e-05, |
| "loss": 0.5514832019805909, |
| "step": 820, |
| "token_acc": 0.8363110430491747 |
| }, |
| { |
| "epoch": 0.8853333333333333, |
| "grad_norm": 5.5, |
| "learning_rate": 1.2335713153137551e-05, |
| "loss": 0.5658942222595215, |
| "step": 830, |
| "token_acc": 0.8320276303597001 |
| }, |
| { |
| "epoch": 0.896, |
| "grad_norm": 5.59375, |
| "learning_rate": 1.2167440488980309e-05, |
| "loss": 0.5764975547790527, |
| "step": 840, |
| "token_acc": 0.8320033599328014 |
| }, |
| { |
| "epoch": 0.9066666666666666, |
| "grad_norm": 4.71875, |
| "learning_rate": 1.1998521321541214e-05, |
| "loss": 0.5525532722473144, |
| "step": 850, |
| "token_acc": 0.836985477356707 |
| }, |
| { |
| "epoch": 0.9173333333333333, |
| "grad_norm": 5.0625, |
| "learning_rate": 1.1829006035961947e-05, |
| "loss": 0.5769608974456787, |
| "step": 860, |
| "token_acc": 0.832285471537808 |
| }, |
| { |
| "epoch": 0.928, |
| "grad_norm": 5.8125, |
| "learning_rate": 1.165894519519407e-05, |
| "loss": 0.5438988685607911, |
| "step": 870, |
| "token_acc": 0.8355040595399188 |
| }, |
| { |
| "epoch": 0.9386666666666666, |
| "grad_norm": 5.375, |
| "learning_rate": 1.148838952491712e-05, |
| "loss": 0.5477108001708985, |
| "step": 880, |
| "token_acc": 0.8355240696483441 |
| }, |
| { |
| "epoch": 0.9493333333333334, |
| "grad_norm": 4.90625, |
| "learning_rate": 1.1317389898408188e-05, |
| "loss": 0.5485364913940429, |
| "step": 890, |
| "token_acc": 0.832925491519703 |
| }, |
| { |
| "epoch": 0.96, |
| "grad_norm": 5.5625, |
| "learning_rate": 1.1145997321367454e-05, |
| "loss": 0.5587331771850585, |
| "step": 900, |
| "token_acc": 0.8355285376561973 |
| }, |
| { |
| "epoch": 0.9706666666666667, |
| "grad_norm": 5.15625, |
| "learning_rate": 1.0974262916704223e-05, |
| "loss": 0.5252433300018311, |
| "step": 910, |
| "token_acc": 0.8404669260700389 |
| }, |
| { |
| "epoch": 0.9813333333333333, |
| "grad_norm": 5.65625, |
| "learning_rate": 1.0802237909288005e-05, |
| "loss": 0.5581695079803467, |
| "step": 920, |
| "token_acc": 0.838423477222317 |
| }, |
| { |
| "epoch": 0.992, |
| "grad_norm": 5.71875, |
| "learning_rate": 1.062997361066918e-05, |
| "loss": 0.5427685737609863, |
| "step": 930, |
| "token_acc": 0.8367765067440669 |
| }, |
| { |
| "epoch": 1.0021333333333333, |
| "grad_norm": 4.65625, |
| "learning_rate": 1.0457521403773823e-05, |
| "loss": 0.5227243900299072, |
| "step": 940, |
| "token_acc": 0.8432694901537632 |
| }, |
| { |
| "epoch": 1.0128, |
| "grad_norm": 4.84375, |
| "learning_rate": 1.0284932727577203e-05, |
| "loss": 0.4229020595550537, |
| "step": 950, |
| "token_acc": 0.8669341072038238 |
| }, |
| { |
| "epoch": 1.0234666666666667, |
| "grad_norm": 5.03125, |
| "learning_rate": 1.0112259061760605e-05, |
| "loss": 0.43592014312744143, |
| "step": 960, |
| "token_acc": 0.8627684964200477 |
| }, |
| { |
| "epoch": 1.0341333333333333, |
| "grad_norm": 5.46875, |
| "learning_rate": 9.939551911355979e-06, |
| "loss": 0.43202896118164064, |
| "step": 970, |
| "token_acc": 0.8636055574381566 |
| }, |
| { |
| "epoch": 1.0448, |
| "grad_norm": 5.15625, |
| "learning_rate": 9.766862791383054e-06, |
| "loss": 0.4390109062194824, |
| "step": 980, |
| "token_acc": 0.8663960764417385 |
| }, |
| { |
| "epoch": 1.0554666666666668, |
| "grad_norm": 5.28125, |
| "learning_rate": 9.59424321148343e-06, |
| "loss": 0.416988468170166, |
| "step": 990, |
| "token_acc": 0.8692489851150202 |
| }, |
| { |
| "epoch": 1.0661333333333334, |
| "grad_norm": 4.96875, |
| "learning_rate": 9.42174466055634e-06, |
| "loss": 0.4186138153076172, |
| "step": 1000, |
| "token_acc": 0.8688482570061518 |
| }, |
| { |
| "epoch": 1.0768, |
| "grad_norm": 5.28125, |
| "learning_rate": 9.249418591400539e-06, |
| "loss": 0.4354249954223633, |
| "step": 1010, |
| "token_acc": 0.8667799490229396 |
| }, |
| { |
| "epoch": 1.0874666666666666, |
| "grad_norm": 5.15625, |
| "learning_rate": 9.07731640536698e-06, |
| "loss": 0.451226806640625, |
| "step": 1020, |
| "token_acc": 0.8601433994095319 |
| }, |
| { |
| "epoch": 1.0981333333333334, |
| "grad_norm": 5.0, |
| "learning_rate": 8.90548943702686e-06, |
| "loss": 0.4429470539093018, |
| "step": 1030, |
| "token_acc": 0.8604789908804228 |
| }, |
| { |
| "epoch": 1.1088, |
| "grad_norm": 5.4375, |
| "learning_rate": 8.733988938859545e-06, |
| "loss": 0.435594367980957, |
| "step": 1040, |
| "token_acc": 0.8667980126777454 |
| }, |
| { |
| "epoch": 1.1194666666666666, |
| "grad_norm": 8.3125, |
| "learning_rate": 8.562866065965013e-06, |
| "loss": 0.4241774082183838, |
| "step": 1050, |
| "token_acc": 0.8683653601963938 |
| }, |
| { |
| "epoch": 1.1301333333333332, |
| "grad_norm": 5.4375, |
| "learning_rate": 8.39217186080532e-06, |
| "loss": 0.42261734008789065, |
| "step": 1060, |
| "token_acc": 0.8686390532544379 |
| }, |
| { |
| "epoch": 1.1408, |
| "grad_norm": 5.03125, |
| "learning_rate": 8.221957237979686e-06, |
| "loss": 0.41465024948120116, |
| "step": 1070, |
| "token_acc": 0.8682988602785986 |
| }, |
| { |
| "epoch": 1.1514666666666666, |
| "grad_norm": 5.46875, |
| "learning_rate": 8.052272969037702e-06, |
| "loss": 0.4440279006958008, |
| "step": 1080, |
| "token_acc": 0.8630967631736272 |
| }, |
| { |
| "epoch": 1.1621333333333332, |
| "grad_norm": 5.09375, |
| "learning_rate": 7.88316966733522e-06, |
| "loss": 0.4302692413330078, |
| "step": 1090, |
| "token_acc": 0.8661821291150064 |
| }, |
| { |
| "epoch": 1.1728, |
| "grad_norm": 5.0, |
| "learning_rate": 7.71469777293741e-06, |
| "loss": 0.4230166435241699, |
| "step": 1100, |
| "token_acc": 0.8674463937621832 |
| }, |
| { |
| "epoch": 1.1834666666666667, |
| "grad_norm": 5.03125, |
| "learning_rate": 7.546907537573529e-06, |
| "loss": 0.4408688545227051, |
| "step": 1110, |
| "token_acc": 0.8641611190696891 |
| }, |
| { |
| "epoch": 1.1941333333333333, |
| "grad_norm": 5.34375, |
| "learning_rate": 7.379849009647848e-06, |
| "loss": 0.43018126487731934, |
| "step": 1120, |
| "token_acc": 0.8686655405405406 |
| }, |
| { |
| "epoch": 1.2048, |
| "grad_norm": 5.3125, |
| "learning_rate": 7.213572019311235e-06, |
| "loss": 0.4404290199279785, |
| "step": 1130, |
| "token_acc": 0.8649767638360795 |
| }, |
| { |
| "epoch": 1.2154666666666667, |
| "grad_norm": 6.625, |
| "learning_rate": 7.04812616359784e-06, |
| "loss": 0.40485620498657227, |
| "step": 1140, |
| "token_acc": 0.8732010559482245 |
| }, |
| { |
| "epoch": 1.2261333333333333, |
| "grad_norm": 5.46875, |
| "learning_rate": 6.883560791631307e-06, |
| "loss": 0.45158863067626953, |
| "step": 1150, |
| "token_acc": 0.8618751596695904 |
| }, |
| { |
| "epoch": 1.2368000000000001, |
| "grad_norm": 5.59375, |
| "learning_rate": 6.7199249899049435e-06, |
| "loss": 0.4398628234863281, |
| "step": 1160, |
| "token_acc": 0.8650072211366918 |
| }, |
| { |
| "epoch": 1.2474666666666667, |
| "grad_norm": 5.46875, |
| "learning_rate": 6.557267567640217e-06, |
| "loss": 0.4337770462036133, |
| "step": 1170, |
| "token_acc": 0.8668418375995931 |
| }, |
| { |
| "epoch": 1.2581333333333333, |
| "grad_norm": 5.96875, |
| "learning_rate": 6.3956370422279625e-06, |
| "loss": 0.4266934394836426, |
| "step": 1180, |
| "token_acc": 0.8647713851864349 |
| }, |
| { |
| "epoch": 1.2688, |
| "grad_norm": 5.59375, |
| "learning_rate": 6.235081624756629e-06, |
| "loss": 0.40284242630004885, |
| "step": 1190, |
| "token_acc": 0.8689100954294324 |
| }, |
| { |
| "epoch": 1.2794666666666665, |
| "grad_norm": 5.0625, |
| "learning_rate": 6.075649205631904e-06, |
| "loss": 0.43938732147216797, |
| "step": 1200, |
| "token_acc": 0.8652512265268144 |
| }, |
| { |
| "epoch": 1.2901333333333334, |
| "grad_norm": 6.09375, |
| "learning_rate": 5.917387340291969e-06, |
| "loss": 0.43057842254638673, |
| "step": 1210, |
| "token_acc": 0.8647868184134534 |
| }, |
| { |
| "epoch": 1.3008, |
| "grad_norm": 5.53125, |
| "learning_rate": 5.760343235022703e-06, |
| "loss": 0.43517169952392576, |
| "step": 1220, |
| "token_acc": 0.8625830635542682 |
| }, |
| { |
| "epoch": 1.3114666666666666, |
| "grad_norm": 5.75, |
| "learning_rate": 5.604563732876989e-06, |
| "loss": 0.4466823101043701, |
| "step": 1230, |
| "token_acc": 0.8609684210526316 |
| }, |
| { |
| "epoch": 1.3221333333333334, |
| "grad_norm": 5.5625, |
| "learning_rate": 5.450095299702392e-06, |
| "loss": 0.44326114654541016, |
| "step": 1240, |
| "token_acc": 0.8649106462268146 |
| }, |
| { |
| "epoch": 1.3328, |
| "grad_norm": 5.59375, |
| "learning_rate": 5.296984010281367e-06, |
| "loss": 0.40932507514953614, |
| "step": 1250, |
| "token_acc": 0.869938754678462 |
| }, |
| { |
| "epoch": 1.3434666666666666, |
| "grad_norm": 5.15625, |
| "learning_rate": 5.145275534588053e-06, |
| "loss": 0.42442498207092283, |
| "step": 1260, |
| "token_acc": 0.8686963177140913 |
| }, |
| { |
| "epoch": 1.3541333333333334, |
| "grad_norm": 5.125, |
| "learning_rate": 4.995015124165883e-06, |
| "loss": 0.42131719589233396, |
| "step": 1270, |
| "token_acc": 0.8664903600068248 |
| }, |
| { |
| "epoch": 1.3648, |
| "grad_norm": 4.90625, |
| "learning_rate": 4.8462475986299625e-06, |
| "loss": 0.42685232162475584, |
| "step": 1280, |
| "token_acc": 0.8674472267327579 |
| }, |
| { |
| "epoch": 1.3754666666666666, |
| "grad_norm": 5.46875, |
| "learning_rate": 4.69901733229828e-06, |
| "loss": 0.4310049057006836, |
| "step": 1290, |
| "token_acc": 0.8657729509587646 |
| }, |
| { |
| "epoch": 1.3861333333333334, |
| "grad_norm": 5.5, |
| "learning_rate": 4.5533682409557745e-06, |
| "loss": 0.42982044219970705, |
| "step": 1300, |
| "token_acc": 0.8655927725219467 |
| }, |
| { |
| "epoch": 1.3968, |
| "grad_norm": 5.25, |
| "learning_rate": 4.409343768755135e-06, |
| "loss": 0.41706085205078125, |
| "step": 1310, |
| "token_acc": 0.8671621390739792 |
| }, |
| { |
| "epoch": 1.4074666666666666, |
| "grad_norm": 5.375, |
| "learning_rate": 4.266986875258299e-06, |
| "loss": 0.42982168197631837, |
| "step": 1320, |
| "token_acc": 0.8655782082734117 |
| }, |
| { |
| "epoch": 1.4181333333333335, |
| "grad_norm": 5.0, |
| "learning_rate": 4.126340022622506e-06, |
| "loss": 0.40945024490356446, |
| "step": 1330, |
| "token_acc": 0.8718840088180431 |
| }, |
| { |
| "epoch": 1.4288, |
| "grad_norm": 5.84375, |
| "learning_rate": 3.987445162934689e-06, |
| "loss": 0.4395137786865234, |
| "step": 1340, |
| "token_acc": 0.8654642733862616 |
| }, |
| { |
| "epoch": 1.4394666666666667, |
| "grad_norm": 5.09375, |
| "learning_rate": 3.850343725698018e-06, |
| "loss": 0.4194896697998047, |
| "step": 1350, |
| "token_acc": 0.8659327530295272 |
| }, |
| { |
| "epoch": 1.4501333333333333, |
| "grad_norm": 5.40625, |
| "learning_rate": 3.7150766054743393e-06, |
| "loss": 0.4344645500183105, |
| "step": 1360, |
| "token_acc": 0.8671583587656833 |
| }, |
| { |
| "epoch": 1.4607999999999999, |
| "grad_norm": 5.71875, |
| "learning_rate": 3.5816841496861563e-06, |
| "loss": 0.4192943572998047, |
| "step": 1370, |
| "token_acc": 0.8677099367196853 |
| }, |
| { |
| "epoch": 1.4714666666666667, |
| "grad_norm": 5.1875, |
| "learning_rate": 3.4502061465818024e-06, |
| "loss": 0.4389029979705811, |
| "step": 1380, |
| "token_acc": 0.8614711328059249 |
| }, |
| { |
| "epoch": 1.4821333333333333, |
| "grad_norm": 5.25, |
| "learning_rate": 3.3206818133674478e-06, |
| "loss": 0.425324821472168, |
| "step": 1390, |
| "token_acc": 0.8668295148247979 |
| }, |
| { |
| "epoch": 1.4928, |
| "grad_norm": 5.25, |
| "learning_rate": 3.1931497845093753e-06, |
| "loss": 0.4116145133972168, |
| "step": 1400, |
| "token_acc": 0.8701353998126543 |
| }, |
| { |
| "epoch": 1.5034666666666667, |
| "grad_norm": 8.0625, |
| "learning_rate": 3.0676481002101223e-06, |
| "loss": 0.4524849414825439, |
| "step": 1410, |
| "token_acc": 0.8596744107173139 |
| }, |
| { |
| "epoch": 1.5141333333333333, |
| "grad_norm": 6.1875, |
| "learning_rate": 2.9442141950618762e-06, |
| "loss": 0.40929403305053713, |
| "step": 1420, |
| "token_acc": 0.8705962637550115 |
| }, |
| { |
| "epoch": 1.5248, |
| "grad_norm": 5.25, |
| "learning_rate": 2.8228848868804714e-06, |
| "loss": 0.4273798942565918, |
| "step": 1430, |
| "token_acc": 0.8684869739478958 |
| }, |
| { |
| "epoch": 1.5354666666666668, |
| "grad_norm": 5.71875, |
| "learning_rate": 2.7036963657234105e-06, |
| "loss": 0.421265983581543, |
| "step": 1440, |
| "token_acc": 0.8716198925189798 |
| }, |
| { |
| "epoch": 1.5461333333333334, |
| "grad_norm": 5.75, |
| "learning_rate": 2.5866841830950994e-06, |
| "loss": 0.4097298622131348, |
| "step": 1450, |
| "token_acc": 0.8729807855806836 |
| }, |
| { |
| "epoch": 1.5568, |
| "grad_norm": 5.84375, |
| "learning_rate": 2.47188324134255e-06, |
| "loss": 0.4039015293121338, |
| "step": 1460, |
| "token_acc": 0.8734842703298566 |
| }, |
| { |
| "epoch": 1.5674666666666668, |
| "grad_norm": 5.84375, |
| "learning_rate": 2.3593277832447405e-06, |
| "loss": 0.3996889591217041, |
| "step": 1470, |
| "token_acc": 0.8748932536293766 |
| }, |
| { |
| "epoch": 1.5781333333333334, |
| "grad_norm": 6.15625, |
| "learning_rate": 2.2490513817986757e-06, |
| "loss": 0.40860567092895506, |
| "step": 1480, |
| "token_acc": 0.8692588504966465 |
| }, |
| { |
| "epoch": 1.5888, |
| "grad_norm": 6.21875, |
| "learning_rate": 2.141086930205255e-06, |
| "loss": 0.4375762939453125, |
| "step": 1490, |
| "token_acc": 0.8629623316856997 |
| }, |
| { |
| "epoch": 1.5994666666666668, |
| "grad_norm": 5.1875, |
| "learning_rate": 2.035466632057905e-06, |
| "loss": 0.424249267578125, |
| "step": 1500, |
| "token_acc": 0.8679070550597744 |
| }, |
| { |
| "epoch": 1.6101333333333332, |
| "grad_norm": 5.0, |
| "learning_rate": 1.9322219917369e-06, |
| "loss": 0.40370688438415525, |
| "step": 1510, |
| "token_acc": 0.871133586177834 |
| }, |
| { |
| "epoch": 1.6208, |
| "grad_norm": 5.28125, |
| "learning_rate": 1.8313838050122423e-06, |
| "loss": 0.4515178680419922, |
| "step": 1520, |
| "token_acc": 0.8593498005262711 |
| }, |
| { |
| "epoch": 1.6314666666666666, |
| "grad_norm": 5.0625, |
| "learning_rate": 1.7329821498579257e-06, |
| "loss": 0.43173723220825194, |
| "step": 1530, |
| "token_acc": 0.8647822765469825 |
| }, |
| { |
| "epoch": 1.6421333333333332, |
| "grad_norm": 5.125, |
| "learning_rate": 1.6370463774802759e-06, |
| "loss": 0.42908735275268556, |
| "step": 1540, |
| "token_acc": 0.8682573498082659 |
| }, |
| { |
| "epoch": 1.6528, |
| "grad_norm": 5.34375, |
| "learning_rate": 1.5436051035630962e-06, |
| "loss": 0.4312245845794678, |
| "step": 1550, |
| "token_acc": 0.8645736105218498 |
| }, |
| { |
| "epoch": 1.6634666666666666, |
| "grad_norm": 5.65625, |
| "learning_rate": 1.4526861997321794e-06, |
| "loss": 0.4247132301330566, |
| "step": 1560, |
| "token_acc": 0.8694854474188976 |
| }, |
| { |
| "epoch": 1.6741333333333333, |
| "grad_norm": 5.40625, |
| "learning_rate": 1.3643167852417894e-06, |
| "loss": 0.4231447219848633, |
| "step": 1570, |
| "token_acc": 0.8674972640794679 |
| }, |
| { |
| "epoch": 1.6848, |
| "grad_norm": 5.4375, |
| "learning_rate": 1.278523218885518e-06, |
| "loss": 0.40172972679138186, |
| "step": 1580, |
| "token_acc": 0.8752116491703352 |
| }, |
| { |
| "epoch": 1.6954666666666667, |
| "grad_norm": 5.3125, |
| "learning_rate": 1.1953310911340165e-06, |
| "loss": 0.4114649295806885, |
| "step": 1590, |
| "token_acc": 0.8714602524735585 |
| }, |
| { |
| "epoch": 1.7061333333333333, |
| "grad_norm": 5.1875, |
| "learning_rate": 1.1147652165018597e-06, |
| "loss": 0.40552978515625, |
| "step": 1600, |
| "token_acc": 0.8678014905149052 |
| }, |
| { |
| "epoch": 1.7168, |
| "grad_norm": 5.53125, |
| "learning_rate": 1.0368496261459004e-06, |
| "loss": 0.4324502944946289, |
| "step": 1610, |
| "token_acc": 0.862758329818642 |
| }, |
| { |
| "epoch": 1.7274666666666667, |
| "grad_norm": 5.46875, |
| "learning_rate": 9.616075606972597e-07, |
| "loss": 0.4126314163208008, |
| "step": 1620, |
| "token_acc": 0.8694481830417228 |
| }, |
| { |
| "epoch": 1.7381333333333333, |
| "grad_norm": 5.75, |
| "learning_rate": 8.890614633291095e-07, |
| "loss": 0.43532743453979494, |
| "step": 1630, |
| "token_acc": 0.8601766904519198 |
| }, |
| { |
| "epoch": 1.7488000000000001, |
| "grad_norm": 5.15625, |
| "learning_rate": 8.192329730623471e-07, |
| "loss": 0.4172046661376953, |
| "step": 1640, |
| "token_acc": 0.8675934559633806 |
| }, |
| { |
| "epoch": 1.7594666666666665, |
| "grad_norm": 5.34375, |
| "learning_rate": 7.521429183111139e-07, |
| "loss": 0.4211602687835693, |
| "step": 1650, |
| "token_acc": 0.869084099421178 |
| }, |
| { |
| "epoch": 1.7701333333333333, |
| "grad_norm": 5.3125, |
| "learning_rate": 6.878113106700946e-07, |
| "loss": 0.4197425842285156, |
| "step": 1660, |
| "token_acc": 0.8680926916221033 |
| }, |
| { |
| "epoch": 1.7808000000000002, |
| "grad_norm": 5.28125, |
| "learning_rate": 6.262573389454773e-07, |
| "loss": 0.43222551345825194, |
| "step": 1670, |
| "token_acc": 0.8637358935489304 |
| }, |
| { |
| "epoch": 1.7914666666666665, |
| "grad_norm": 6.09375, |
| "learning_rate": 5.674993634313242e-07, |
| "loss": 0.4256401062011719, |
| "step": 1680, |
| "token_acc": 0.8653747652381766 |
| }, |
| { |
| "epoch": 1.8021333333333334, |
| "grad_norm": 5.03125, |
| "learning_rate": 5.115549104330575e-07, |
| "loss": 0.39786996841430666, |
| "step": 1690, |
| "token_acc": 0.8736002714625043 |
| }, |
| { |
| "epoch": 1.8128, |
| "grad_norm": 7.09375, |
| "learning_rate": 4.5844066703973676e-07, |
| "loss": 0.42459850311279296, |
| "step": 1700, |
| "token_acc": 0.8695799745439118 |
| }, |
| { |
| "epoch": 1.8234666666666666, |
| "grad_norm": 5.59375, |
| "learning_rate": 4.081724761466288e-07, |
| "loss": 0.44160709381103513, |
| "step": 1710, |
| "token_acc": 0.8608688288593446 |
| }, |
| { |
| "epoch": 1.8341333333333334, |
| "grad_norm": 5.5, |
| "learning_rate": 3.6076533172959895e-07, |
| "loss": 0.4103675842285156, |
| "step": 1720, |
| "token_acc": 0.8672603545324709 |
| }, |
| { |
| "epoch": 1.8448, |
| "grad_norm": 5.5625, |
| "learning_rate": 3.1623337437272084e-07, |
| "loss": 0.43485474586486816, |
| "step": 1730, |
| "token_acc": 0.8642153532608695 |
| }, |
| { |
| "epoch": 1.8554666666666666, |
| "grad_norm": 5.4375, |
| "learning_rate": 2.745898870504116e-07, |
| "loss": 0.4110551357269287, |
| "step": 1740, |
| "token_acc": 0.8730118009235506 |
| }, |
| { |
| "epoch": 1.8661333333333334, |
| "grad_norm": 5.21875, |
| "learning_rate": 2.3584729116541305e-07, |
| "loss": 0.41416378021240235, |
| "step": 1750, |
| "token_acc": 0.86998136540742 |
| }, |
| { |
| "epoch": 1.8768, |
| "grad_norm": 5.34375, |
| "learning_rate": 2.0001714284372253e-07, |
| "loss": 0.43006277084350586, |
| "step": 1760, |
| "token_acc": 0.8696650229552797 |
| }, |
| { |
| "epoch": 1.8874666666666666, |
| "grad_norm": 5.78125, |
| "learning_rate": 1.6711012948764426e-07, |
| "loss": 0.40975046157836914, |
| "step": 1770, |
| "token_acc": 0.8725116476069462 |
| }, |
| { |
| "epoch": 1.8981333333333335, |
| "grad_norm": 5.625, |
| "learning_rate": 1.371360665879584e-07, |
| "loss": 0.406182861328125, |
| "step": 1780, |
| "token_acc": 0.8718765721952038 |
| }, |
| { |
| "epoch": 1.9088, |
| "grad_norm": 5.25, |
| "learning_rate": 1.1010389479615302e-07, |
| "loss": 0.4136258602142334, |
| "step": 1790, |
| "token_acc": 0.870780770542107 |
| }, |
| { |
| "epoch": 1.9194666666666667, |
| "grad_norm": 6.4375, |
| "learning_rate": 8.60216772576139e-08, |
| "loss": 0.4269017219543457, |
| "step": 1800, |
| "token_acc": 0.8661390519955748 |
| }, |
| { |
| "epoch": 1.9301333333333335, |
| "grad_norm": 5.6875, |
| "learning_rate": 6.489659720655295e-08, |
| "loss": 0.42894792556762695, |
| "step": 1810, |
| "token_acc": 0.8644270122783083 |
| }, |
| { |
| "epoch": 1.9407999999999999, |
| "grad_norm": 5.34375, |
| "learning_rate": 4.673495582339649e-08, |
| "loss": 0.42904911041259763, |
| "step": 1820, |
| "token_acc": 0.8654405474764756 |
| }, |
| { |
| "epoch": 1.9514666666666667, |
| "grad_norm": 5.375, |
| "learning_rate": 3.154217035526874e-08, |
| "loss": 0.42716250419616697, |
| "step": 1830, |
| "token_acc": 0.869757174392936 |
| }, |
| { |
| "epoch": 1.9621333333333333, |
| "grad_norm": 5.28125, |
| "learning_rate": 1.932277250014214e-08, |
| "loss": 0.42876396179199217, |
| "step": 1840, |
| "token_acc": 0.8661888701517707 |
| }, |
| { |
| "epoch": 1.9727999999999999, |
| "grad_norm": 5.5625, |
| "learning_rate": 1.008040705511748e-08, |
| "loss": 0.44731903076171875, |
| "step": 1850, |
| "token_acc": 0.8617075232459848 |
| }, |
| { |
| "epoch": 1.9834666666666667, |
| "grad_norm": 5.40625, |
| "learning_rate": 3.8178308292624585e-09, |
| "loss": 0.4366297721862793, |
| "step": 1860, |
| "token_acc": 0.8642533936651584 |
| }, |
| { |
| "epoch": 1.9941333333333333, |
| "grad_norm": 5.59375, |
| "learning_rate": 5.369118213094382e-10, |
| "loss": 0.4206860542297363, |
| "step": 1870, |
| "token_acc": 0.8676668914362778 |
| } |
| ], |
| "logging_steps": 10, |
| "max_steps": 1876, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 2, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 8.012174212254925e+17, |
| "train_batch_size": 8, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
