| { |
| "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": 9.75, |
| "learning_rate": 3.5087719298245616e-07, |
| "loss": 0.06797242164611816, |
| "step": 1, |
| "token_acc": 0.970873786407767 |
| }, |
| { |
| "epoch": 0.010666666666666666, |
| "grad_norm": 3.375, |
| "learning_rate": 3.5087719298245615e-06, |
| "loss": 0.04619172546598646, |
| "step": 10, |
| "token_acc": 0.9840170484816196 |
| }, |
| { |
| "epoch": 0.021333333333333333, |
| "grad_norm": 1.3515625, |
| "learning_rate": 7.017543859649123e-06, |
| "loss": 0.03136685192584991, |
| "step": 20, |
| "token_acc": 0.9862650602409638 |
| }, |
| { |
| "epoch": 0.032, |
| "grad_norm": 2.234375, |
| "learning_rate": 1.0526315789473684e-05, |
| "loss": 0.035260409116744995, |
| "step": 30, |
| "token_acc": 0.9865771812080537 |
| }, |
| { |
| "epoch": 0.042666666666666665, |
| "grad_norm": 2.90625, |
| "learning_rate": 1.4035087719298246e-05, |
| "loss": 0.03275148570537567, |
| "step": 40, |
| "token_acc": 0.985594237695078 |
| }, |
| { |
| "epoch": 0.05333333333333334, |
| "grad_norm": 2.296875, |
| "learning_rate": 1.754385964912281e-05, |
| "loss": 0.02667125463485718, |
| "step": 50, |
| "token_acc": 0.9887181949111858 |
| }, |
| { |
| "epoch": 0.064, |
| "grad_norm": 1.8203125, |
| "learning_rate": 1.9999865771143805e-05, |
| "loss": 0.042026185989379884, |
| "step": 60, |
| "token_acc": 0.9822286263208453 |
| }, |
| { |
| "epoch": 0.07466666666666667, |
| "grad_norm": 3.046875, |
| "learning_rate": 1.9997479580610527e-05, |
| "loss": 0.041328877210617065, |
| "step": 70, |
| "token_acc": 0.9809270883631096 |
| }, |
| { |
| "epoch": 0.08533333333333333, |
| "grad_norm": 2.140625, |
| "learning_rate": 1.9992111345867986e-05, |
| "loss": 0.03837056159973144, |
| "step": 80, |
| "token_acc": 0.9831162566328991 |
| }, |
| { |
| "epoch": 0.096, |
| "grad_norm": 1.59375, |
| "learning_rate": 1.9983762668151084e-05, |
| "loss": 0.032483524084091185, |
| "step": 90, |
| "token_acc": 0.9863896848137536 |
| }, |
| { |
| "epoch": 0.10666666666666667, |
| "grad_norm": 1.8671875, |
| "learning_rate": 1.9972436037700082e-05, |
| "loss": 0.03307614922523498, |
| "step": 100, |
| "token_acc": 0.9855560905151661 |
| }, |
| { |
| "epoch": 0.11733333333333333, |
| "grad_norm": 1.734375, |
| "learning_rate": 1.9958134833017768e-05, |
| "loss": 0.0380557119846344, |
| "step": 110, |
| "token_acc": 0.98277117204562 |
| }, |
| { |
| "epoch": 0.128, |
| "grad_norm": 3.53125, |
| "learning_rate": 1.9940863319861758e-05, |
| "loss": 0.047079157829284665, |
| "step": 120, |
| "token_acc": 0.9812545061283345 |
| }, |
| { |
| "epoch": 0.13866666666666666, |
| "grad_norm": 2.625, |
| "learning_rate": 1.992062664997209e-05, |
| "loss": 0.039805743098258975, |
| "step": 130, |
| "token_acc": 0.9844460397224216 |
| }, |
| { |
| "epoch": 0.14933333333333335, |
| "grad_norm": 3.03125, |
| "learning_rate": 1.9897430859534553e-05, |
| "loss": 0.029691782593727113, |
| "step": 140, |
| "token_acc": 0.9876781831360232 |
| }, |
| { |
| "epoch": 0.16, |
| "grad_norm": 2.203125, |
| "learning_rate": 1.9871282867380245e-05, |
| "loss": 0.03704167008399963, |
| "step": 150, |
| "token_acc": 0.9851104707012488 |
| }, |
| { |
| "epoch": 0.17066666666666666, |
| "grad_norm": 2.828125, |
| "learning_rate": 1.9842190472921802e-05, |
| "loss": 0.03181208968162537, |
| "step": 160, |
| "token_acc": 0.9877049180327869 |
| }, |
| { |
| "epoch": 0.18133333333333335, |
| "grad_norm": 2.0, |
| "learning_rate": 1.9810162353827e-05, |
| "loss": 0.03544786870479584, |
| "step": 170, |
| "token_acc": 0.9834333733493398 |
| }, |
| { |
| "epoch": 0.192, |
| "grad_norm": 2.453125, |
| "learning_rate": 1.977520806343039e-05, |
| "loss": 0.02816782593727112, |
| "step": 180, |
| "token_acc": 0.9887935145445875 |
| }, |
| { |
| "epoch": 0.20266666666666666, |
| "grad_norm": 2.140625, |
| "learning_rate": 1.9737338027883727e-05, |
| "loss": 0.033563745021820066, |
| "step": 190, |
| "token_acc": 0.9869910864851843 |
| }, |
| { |
| "epoch": 0.21333333333333335, |
| "grad_norm": 1.1015625, |
| "learning_rate": 1.969656354304606e-05, |
| "loss": 0.037265950441360475, |
| "step": 200, |
| "token_acc": 0.983140655105973 |
| }, |
| { |
| "epoch": 0.224, |
| "grad_norm": 2.1875, |
| "learning_rate": 1.9652896771114416e-05, |
| "loss": 0.03406996726989746, |
| "step": 210, |
| "token_acc": 0.9867565615217915 |
| }, |
| { |
| "epoch": 0.23466666666666666, |
| "grad_norm": 1.875, |
| "learning_rate": 1.960635073699605e-05, |
| "loss": 0.036304399371147156, |
| "step": 220, |
| "token_acc": 0.9850169163847269 |
| }, |
| { |
| "epoch": 0.24533333333333332, |
| "grad_norm": 2.0625, |
| "learning_rate": 1.95569393244234e-05, |
| "loss": 0.03020932674407959, |
| "step": 230, |
| "token_acc": 0.9867437936852254 |
| }, |
| { |
| "epoch": 0.256, |
| "grad_norm": 3.046875, |
| "learning_rate": 1.950467727181284e-05, |
| "loss": 0.03292643427848816, |
| "step": 240, |
| "token_acc": 0.986077772443591 |
| }, |
| { |
| "epoch": 0.26666666666666666, |
| "grad_norm": 1.28125, |
| "learning_rate": 1.9449580167868506e-05, |
| "loss": 0.03676349222660065, |
| "step": 250, |
| "token_acc": 0.983389504092441 |
| }, |
| { |
| "epoch": 0.2773333333333333, |
| "grad_norm": 1.2421875, |
| "learning_rate": 1.939166444693252e-05, |
| "loss": 0.03957515358924866, |
| "step": 260, |
| "token_acc": 0.982388419782871 |
| }, |
| { |
| "epoch": 0.288, |
| "grad_norm": 1.875, |
| "learning_rate": 1.9330947384082957e-05, |
| "loss": 0.03727341294288635, |
| "step": 270, |
| "token_acc": 0.9828998073217726 |
| }, |
| { |
| "epoch": 0.2986666666666667, |
| "grad_norm": 1.0, |
| "learning_rate": 1.9267447089981018e-05, |
| "loss": 0.04420234560966492, |
| "step": 280, |
| "token_acc": 0.9821299203091041 |
| }, |
| { |
| "epoch": 0.30933333333333335, |
| "grad_norm": 2.046875, |
| "learning_rate": 1.9201182505469035e-05, |
| "loss": 0.03257325291633606, |
| "step": 290, |
| "token_acc": 0.986077772443591 |
| }, |
| { |
| "epoch": 0.32, |
| "grad_norm": 2.265625, |
| "learning_rate": 1.9132173395920756e-05, |
| "loss": 0.04046020805835724, |
| "step": 300, |
| "token_acc": 0.9863243761996161 |
| }, |
| { |
| "epoch": 0.33066666666666666, |
| "grad_norm": 1.6640625, |
| "learning_rate": 1.9060440345345762e-05, |
| "loss": 0.0318397045135498, |
| "step": 310, |
| "token_acc": 0.9875 |
| }, |
| { |
| "epoch": 0.3413333333333333, |
| "grad_norm": 2.4375, |
| "learning_rate": 1.8986004750249687e-05, |
| "loss": 0.030359810590744017, |
| "step": 320, |
| "token_acc": 0.9879576107899807 |
| }, |
| { |
| "epoch": 0.352, |
| "grad_norm": 2.59375, |
| "learning_rate": 1.890888881325205e-05, |
| "loss": 0.03387542963027954, |
| "step": 330, |
| "token_acc": 0.9860710854947167 |
| }, |
| { |
| "epoch": 0.3626666666666667, |
| "grad_norm": 1.2109375, |
| "learning_rate": 1.8829115536463682e-05, |
| "loss": 0.02995218336582184, |
| "step": 340, |
| "token_acc": 0.9878048780487805 |
| }, |
| { |
| "epoch": 0.37333333333333335, |
| "grad_norm": 1.5546875, |
| "learning_rate": 1.8746708714625668e-05, |
| "loss": 0.03572494685649872, |
| "step": 350, |
| "token_acc": 0.9858173076923077 |
| }, |
| { |
| "epoch": 0.384, |
| "grad_norm": 1.796875, |
| "learning_rate": 1.8661692928011858e-05, |
| "loss": 0.027254277467727663, |
| "step": 360, |
| "token_acc": 0.987460815047022 |
| }, |
| { |
| "epoch": 0.39466666666666667, |
| "grad_norm": 2.640625, |
| "learning_rate": 1.8574093535097062e-05, |
| "loss": 0.03353258669376373, |
| "step": 370, |
| "token_acc": 0.984844840028867 |
| }, |
| { |
| "epoch": 0.4053333333333333, |
| "grad_norm": 1.6484375, |
| "learning_rate": 1.8483936664993152e-05, |
| "loss": 0.03917534947395325, |
| "step": 380, |
| "token_acc": 0.9830917874396136 |
| }, |
| { |
| "epoch": 0.416, |
| "grad_norm": 1.5703125, |
| "learning_rate": 1.839124920965527e-05, |
| "loss": 0.03518342077732086, |
| "step": 390, |
| "token_acc": 0.9841346153846153 |
| }, |
| { |
| "epoch": 0.4266666666666667, |
| "grad_norm": 1.7890625, |
| "learning_rate": 1.829605881586051e-05, |
| "loss": 0.03069465160369873, |
| "step": 400, |
| "token_acc": 0.9872504209766658 |
| }, |
| { |
| "epoch": 0.43733333333333335, |
| "grad_norm": 1.5546875, |
| "learning_rate": 1.8198393876961446e-05, |
| "loss": 0.02751390039920807, |
| "step": 410, |
| "token_acc": 0.9887613582018173 |
| }, |
| { |
| "epoch": 0.448, |
| "grad_norm": 1.6640625, |
| "learning_rate": 1.8098283524416982e-05, |
| "loss": 0.026150995492935182, |
| "step": 420, |
| "token_acc": 0.9882606612362242 |
| }, |
| { |
| "epoch": 0.45866666666666667, |
| "grad_norm": 1.828125, |
| "learning_rate": 1.7995757619103012e-05, |
| "loss": 0.03289719820022583, |
| "step": 430, |
| "token_acc": 0.9862617498192335 |
| }, |
| { |
| "epoch": 0.4693333333333333, |
| "grad_norm": 1.921875, |
| "learning_rate": 1.7890846742405558e-05, |
| "loss": 0.035283339023590085, |
| "step": 440, |
| "token_acc": 0.98579340235974 |
| }, |
| { |
| "epoch": 0.48, |
| "grad_norm": 2.203125, |
| "learning_rate": 1.7783582187098944e-05, |
| "loss": 0.03278760313987732, |
| "step": 450, |
| "token_acc": 0.9858716475095786 |
| }, |
| { |
| "epoch": 0.49066666666666664, |
| "grad_norm": 1.6796875, |
| "learning_rate": 1.7673995948011807e-05, |
| "loss": 0.030364534258842467, |
| "step": 460, |
| "token_acc": 0.9877873563218391 |
| }, |
| { |
| "epoch": 0.5013333333333333, |
| "grad_norm": 1.40625, |
| "learning_rate": 1.75621207124837e-05, |
| "loss": 0.02944081127643585, |
| "step": 470, |
| "token_acc": 0.9857522337599613 |
| }, |
| { |
| "epoch": 0.512, |
| "grad_norm": 2.359375, |
| "learning_rate": 1.7447989850615114e-05, |
| "loss": 0.03358876705169678, |
| "step": 480, |
| "token_acc": 0.9858343337334934 |
| }, |
| { |
| "epoch": 0.5226666666666666, |
| "grad_norm": 1.9609375, |
| "learning_rate": 1.733163740531386e-05, |
| "loss": 0.03116101920604706, |
| "step": 490, |
| "token_acc": 0.9863603732950467 |
| }, |
| { |
| "epoch": 0.5333333333333333, |
| "grad_norm": 0.91015625, |
| "learning_rate": 1.7213098082140774e-05, |
| "loss": 0.030902951955795288, |
| "step": 500, |
| "token_acc": 0.9863276565123531 |
| }, |
| { |
| "epoch": 0.544, |
| "grad_norm": 1.1953125, |
| "learning_rate": 1.7092407238957737e-05, |
| "loss": 0.03449364304542542, |
| "step": 510, |
| "token_acc": 0.988502994011976 |
| }, |
| { |
| "epoch": 0.5546666666666666, |
| "grad_norm": 1.1875, |
| "learning_rate": 1.6969600875381172e-05, |
| "loss": 0.0299816370010376, |
| "step": 520, |
| "token_acc": 0.9869109947643979 |
| }, |
| { |
| "epoch": 0.5653333333333334, |
| "grad_norm": 1.2265625, |
| "learning_rate": 1.684471562204407e-05, |
| "loss": 0.033460429310798644, |
| "step": 530, |
| "token_acc": 0.9860744297719087 |
| }, |
| { |
| "epoch": 0.576, |
| "grad_norm": 1.6875, |
| "learning_rate": 1.6717788729669822e-05, |
| "loss": 0.029380664229393005, |
| "step": 540, |
| "token_acc": 0.9881756756756757 |
| }, |
| { |
| "epoch": 0.5866666666666667, |
| "grad_norm": 2.21875, |
| "learning_rate": 1.658885805796111e-05, |
| "loss": 0.0348785400390625, |
| "step": 550, |
| "token_acc": 0.9853047458443749 |
| }, |
| { |
| "epoch": 0.5973333333333334, |
| "grad_norm": 1.1015625, |
| "learning_rate": 1.6457962064307104e-05, |
| "loss": 0.025776177644729614, |
| "step": 560, |
| "token_acc": 0.9886883273164861 |
| }, |
| { |
| "epoch": 0.608, |
| "grad_norm": 2.453125, |
| "learning_rate": 1.6325139792312414e-05, |
| "loss": 0.03230432271957397, |
| "step": 570, |
| "token_acc": 0.9868641031764986 |
| }, |
| { |
| "epoch": 0.6186666666666667, |
| "grad_norm": 2.265625, |
| "learning_rate": 1.6190430860151196e-05, |
| "loss": 0.031005895137786864, |
| "step": 580, |
| "token_acc": 0.9841612670986322 |
| }, |
| { |
| "epoch": 0.6293333333333333, |
| "grad_norm": 2.546875, |
| "learning_rate": 1.605387544874983e-05, |
| "loss": 0.03286699056625366, |
| "step": 590, |
| "token_acc": 0.9843033083796184 |
| }, |
| { |
| "epoch": 0.64, |
| "grad_norm": 1.515625, |
| "learning_rate": 1.591551428980179e-05, |
| "loss": 0.032039952278137204, |
| "step": 600, |
| "token_acc": 0.9858648778150455 |
| }, |
| { |
| "epoch": 0.6506666666666666, |
| "grad_norm": 1.3515625, |
| "learning_rate": 1.5775388653618217e-05, |
| "loss": 0.03594903945922852, |
| "step": 610, |
| "token_acc": 0.9849318344893566 |
| }, |
| { |
| "epoch": 0.6613333333333333, |
| "grad_norm": 2.625, |
| "learning_rate": 1.5633540336817804e-05, |
| "loss": 0.027490669488906862, |
| "step": 620, |
| "token_acc": 0.985925572519084 |
| }, |
| { |
| "epoch": 0.672, |
| "grad_norm": 1.75, |
| "learning_rate": 1.549001164985973e-05, |
| "loss": 0.03038315773010254, |
| "step": 630, |
| "token_acc": 0.9857865574560347 |
| }, |
| { |
| "epoch": 0.6826666666666666, |
| "grad_norm": 2.109375, |
| "learning_rate": 1.5344845404423313e-05, |
| "loss": 0.0375783383846283, |
| "step": 640, |
| "token_acc": 0.981414433985035 |
| }, |
| { |
| "epoch": 0.6933333333333334, |
| "grad_norm": 1.3515625, |
| "learning_rate": 1.5198084900638161e-05, |
| "loss": 0.03005017340183258, |
| "step": 650, |
| "token_acc": 0.9848082951531227 |
| }, |
| { |
| "epoch": 0.704, |
| "grad_norm": 0.7265625, |
| "learning_rate": 1.5049773914168629e-05, |
| "loss": 0.031009536981582642, |
| "step": 660, |
| "token_acc": 0.9857899807321773 |
| }, |
| { |
| "epoch": 0.7146666666666667, |
| "grad_norm": 2.34375, |
| "learning_rate": 1.4899956683156413e-05, |
| "loss": 0.033309900760650636, |
| "step": 670, |
| "token_acc": 0.98503861003861 |
| }, |
| { |
| "epoch": 0.7253333333333334, |
| "grad_norm": 1.9921875, |
| "learning_rate": 1.4748677895025243e-05, |
| "loss": 0.03143962621688843, |
| "step": 680, |
| "token_acc": 0.9863013698630136 |
| }, |
| { |
| "epoch": 0.736, |
| "grad_norm": 1.5703125, |
| "learning_rate": 1.4595982673151511e-05, |
| "loss": 0.028191816806793214, |
| "step": 690, |
| "token_acc": 0.9877785765636233 |
| }, |
| { |
| "epoch": 0.7466666666666667, |
| "grad_norm": 1.9921875, |
| "learning_rate": 1.444191656340491e-05, |
| "loss": 0.02878325283527374, |
| "step": 700, |
| "token_acc": 0.9867437936852254 |
| }, |
| { |
| "epoch": 0.7573333333333333, |
| "grad_norm": 0.875, |
| "learning_rate": 1.4286525520563003e-05, |
| "loss": 0.028445839881896973, |
| "step": 710, |
| "token_acc": 0.9870192307692308 |
| }, |
| { |
| "epoch": 0.768, |
| "grad_norm": 1.921875, |
| "learning_rate": 1.4129855894603885e-05, |
| "loss": 0.02481076419353485, |
| "step": 720, |
| "token_acc": 0.9894636015325671 |
| }, |
| { |
| "epoch": 0.7786666666666666, |
| "grad_norm": 1.703125, |
| "learning_rate": 1.3971954416880897e-05, |
| "loss": 0.032581061124801636, |
| "step": 730, |
| "token_acc": 0.9865222623345367 |
| }, |
| { |
| "epoch": 0.7893333333333333, |
| "grad_norm": 1.28125, |
| "learning_rate": 1.3812868186183641e-05, |
| "loss": 0.03237102627754211, |
| "step": 740, |
| "token_acc": 0.9857865574560347 |
| }, |
| { |
| "epoch": 0.8, |
| "grad_norm": 0.83984375, |
| "learning_rate": 1.3652644654689355e-05, |
| "loss": 0.027326399087905885, |
| "step": 750, |
| "token_acc": 0.9896634615384615 |
| }, |
| { |
| "epoch": 0.8106666666666666, |
| "grad_norm": 2.59375, |
| "learning_rate": 1.3491331613808909e-05, |
| "loss": 0.032716572284698486, |
| "step": 760, |
| "token_acc": 0.9857831325301205 |
| }, |
| { |
| "epoch": 0.8213333333333334, |
| "grad_norm": 2.171875, |
| "learning_rate": 1.3328977179931595e-05, |
| "loss": 0.03095829486846924, |
| "step": 770, |
| "token_acc": 0.9869848156182213 |
| }, |
| { |
| "epoch": 0.832, |
| "grad_norm": 1.0625, |
| "learning_rate": 1.3165629780072991e-05, |
| "loss": 0.029185777902603148, |
| "step": 780, |
| "token_acc": 0.9876960193003619 |
| }, |
| { |
| "epoch": 0.8426666666666667, |
| "grad_norm": 1.1328125, |
| "learning_rate": 1.3001338137430163e-05, |
| "loss": 0.024865221977233887, |
| "step": 790, |
| "token_acc": 0.9870720612880057 |
| }, |
| { |
| "epoch": 0.8533333333333334, |
| "grad_norm": 1.15625, |
| "learning_rate": 1.2836151256848562e-05, |
| "loss": 0.029257166385650634, |
| "step": 800, |
| "token_acc": 0.9872779644743159 |
| }, |
| { |
| "epoch": 0.864, |
| "grad_norm": 3.296875, |
| "learning_rate": 1.2670118410204819e-05, |
| "loss": 0.031032463908195494, |
| "step": 810, |
| "token_acc": 0.9853153586904189 |
| }, |
| { |
| "epoch": 0.8746666666666667, |
| "grad_norm": 1.5234375, |
| "learning_rate": 1.250328912171001e-05, |
| "loss": 0.027763417363166808, |
| "step": 820, |
| "token_acc": 0.9882719004308281 |
| }, |
| { |
| "epoch": 0.8853333333333333, |
| "grad_norm": 1.40625, |
| "learning_rate": 1.2335713153137551e-05, |
| "loss": 0.03198724389076233, |
| "step": 830, |
| "token_acc": 0.9862947823996153 |
| }, |
| { |
| "epoch": 0.896, |
| "grad_norm": 1.4453125, |
| "learning_rate": 1.2167440488980309e-05, |
| "loss": 0.02678609490394592, |
| "step": 840, |
| "token_acc": 0.9867053420352913 |
| }, |
| { |
| "epoch": 0.9066666666666666, |
| "grad_norm": 2.390625, |
| "learning_rate": 1.1998521321541214e-05, |
| "loss": 0.028643500804901124, |
| "step": 850, |
| "token_acc": 0.9868483978957436 |
| }, |
| { |
| "epoch": 0.9173333333333333, |
| "grad_norm": 0.8203125, |
| "learning_rate": 1.1829006035961947e-05, |
| "loss": 0.033641707897186277, |
| "step": 860, |
| "token_acc": 0.983855421686747 |
| }, |
| { |
| "epoch": 0.928, |
| "grad_norm": 0.75390625, |
| "learning_rate": 1.165894519519407e-05, |
| "loss": 0.02538594901561737, |
| "step": 870, |
| "token_acc": 0.9887290167865708 |
| }, |
| { |
| "epoch": 0.9386666666666666, |
| "grad_norm": 0.92578125, |
| "learning_rate": 1.148838952491712e-05, |
| "loss": 0.029299235343933104, |
| "step": 880, |
| "token_acc": 0.9872350674373795 |
| }, |
| { |
| "epoch": 0.9493333333333334, |
| "grad_norm": 1.4296875, |
| "learning_rate": 1.1317389898408188e-05, |
| "loss": 0.027896964550018312, |
| "step": 890, |
| "token_acc": 0.9877668505636843 |
| }, |
| { |
| "epoch": 0.96, |
| "grad_norm": 1.1796875, |
| "learning_rate": 1.1145997321367454e-05, |
| "loss": 0.02889745831489563, |
| "step": 900, |
| "token_acc": 0.9889316650625601 |
| }, |
| { |
| "epoch": 0.9706666666666667, |
| "grad_norm": 2.203125, |
| "learning_rate": 1.0974262916704223e-05, |
| "loss": 0.030401945114135742, |
| "step": 910, |
| "token_acc": 0.986557849255881 |
| }, |
| { |
| "epoch": 0.9813333333333333, |
| "grad_norm": 1.9375, |
| "learning_rate": 1.0802237909288005e-05, |
| "loss": 0.029643896222114562, |
| "step": 920, |
| "token_acc": 0.9869942196531792 |
| }, |
| { |
| "epoch": 0.992, |
| "grad_norm": 2.90625, |
| "learning_rate": 1.062997361066918e-05, |
| "loss": 0.02902018129825592, |
| "step": 930, |
| "token_acc": 0.9874396135265701 |
| }, |
| { |
| "epoch": 1.0021333333333333, |
| "grad_norm": 1.0234375, |
| "learning_rate": 1.0457521403773823e-05, |
| "loss": 0.02213967442512512, |
| "step": 940, |
| "token_acc": 0.9900153609831029 |
| }, |
| { |
| "epoch": 1.0128, |
| "grad_norm": 1.3125, |
| "learning_rate": 1.0284932727577203e-05, |
| "loss": 0.010565431416034698, |
| "step": 950, |
| "token_acc": 0.9968839884947267 |
| }, |
| { |
| "epoch": 1.0234666666666667, |
| "grad_norm": 1.1640625, |
| "learning_rate": 1.0112259061760605e-05, |
| "loss": 0.01075158715248108, |
| "step": 960, |
| "token_acc": 0.9956741167988464 |
| }, |
| { |
| "epoch": 1.0341333333333333, |
| "grad_norm": 2.484375, |
| "learning_rate": 9.939551911355979e-06, |
| "loss": 0.016012924909591674, |
| "step": 970, |
| "token_acc": 0.9942335415665545 |
| }, |
| { |
| "epoch": 1.0448, |
| "grad_norm": 1.578125, |
| "learning_rate": 9.766862791383054e-06, |
| "loss": 0.012686330080032348, |
| "step": 980, |
| "token_acc": 0.995164410058027 |
| }, |
| { |
| "epoch": 1.0554666666666668, |
| "grad_norm": 0.9609375, |
| "learning_rate": 9.59424321148343e-06, |
| "loss": 0.006662490963935852, |
| "step": 990, |
| "token_acc": 0.9975984630163305 |
| }, |
| { |
| "epoch": 1.0661333333333334, |
| "grad_norm": 0.765625, |
| "learning_rate": 9.42174466055634e-06, |
| "loss": 0.0062740519642829895, |
| "step": 1000, |
| "token_acc": 0.9980694980694981 |
| }, |
| { |
| "epoch": 1.0768, |
| "grad_norm": 2.515625, |
| "learning_rate": 9.249418591400539e-06, |
| "loss": 0.011450471729040146, |
| "step": 1010, |
| "token_acc": 0.9964046021093 |
| }, |
| { |
| "epoch": 1.0874666666666666, |
| "grad_norm": 2.75, |
| "learning_rate": 9.07731640536698e-06, |
| "loss": 0.014336982369422912, |
| "step": 1020, |
| "token_acc": 0.9954172696575012 |
| }, |
| { |
| "epoch": 1.0981333333333334, |
| "grad_norm": 0.77734375, |
| "learning_rate": 8.90548943702686e-06, |
| "loss": 0.009051916003227235, |
| "step": 1030, |
| "token_acc": 0.9963820549927641 |
| }, |
| { |
| "epoch": 1.1088, |
| "grad_norm": 3.40625, |
| "learning_rate": 8.733988938859545e-06, |
| "loss": 0.014770987629890441, |
| "step": 1040, |
| "token_acc": 0.9946962391513983 |
| }, |
| { |
| "epoch": 1.1194666666666666, |
| "grad_norm": 0.40625, |
| "learning_rate": 8.562866065965013e-06, |
| "loss": 0.0071306928992271425, |
| "step": 1050, |
| "token_acc": 0.9971112181030333 |
| }, |
| { |
| "epoch": 1.1301333333333332, |
| "grad_norm": 1.5703125, |
| "learning_rate": 8.39217186080532e-06, |
| "loss": 0.010715489089488984, |
| "step": 1060, |
| "token_acc": 0.9963916285783017 |
| }, |
| { |
| "epoch": 1.1408, |
| "grad_norm": 0.8125, |
| "learning_rate": 8.221957237979686e-06, |
| "loss": 0.011085398495197296, |
| "step": 1070, |
| "token_acc": 0.9956374212312167 |
| }, |
| { |
| "epoch": 1.1514666666666666, |
| "grad_norm": 1.859375, |
| "learning_rate": 8.052272969037702e-06, |
| "loss": 0.011836430430412293, |
| "step": 1080, |
| "token_acc": 0.9942154736080984 |
| }, |
| { |
| "epoch": 1.1621333333333332, |
| "grad_norm": 2.5625, |
| "learning_rate": 7.88316966733522e-06, |
| "loss": 0.00989045649766922, |
| "step": 1090, |
| "token_acc": 0.9959281437125749 |
| }, |
| { |
| "epoch": 1.1728, |
| "grad_norm": 2.25, |
| "learning_rate": 7.71469777293741e-06, |
| "loss": 0.012199420481920242, |
| "step": 1100, |
| "token_acc": 0.995438175270108 |
| }, |
| { |
| "epoch": 1.1834666666666667, |
| "grad_norm": 1.4765625, |
| "learning_rate": 7.546907537573529e-06, |
| "loss": 0.01276719719171524, |
| "step": 1110, |
| "token_acc": 0.9954216867469879 |
| }, |
| { |
| "epoch": 1.1941333333333333, |
| "grad_norm": 1.171875, |
| "learning_rate": 7.379849009647848e-06, |
| "loss": 0.008200515061616898, |
| "step": 1120, |
| "token_acc": 0.9975851243660951 |
| }, |
| { |
| "epoch": 1.2048, |
| "grad_norm": 1.7734375, |
| "learning_rate": 7.213572019311235e-06, |
| "loss": 0.010789988934993744, |
| "step": 1130, |
| "token_acc": 0.9956751561749159 |
| }, |
| { |
| "epoch": 1.2154666666666667, |
| "grad_norm": 2.390625, |
| "learning_rate": 7.04812616359784e-06, |
| "loss": 0.00852215737104416, |
| "step": 1140, |
| "token_acc": 0.9968958930276982 |
| }, |
| { |
| "epoch": 1.2261333333333333, |
| "grad_norm": 2.375, |
| "learning_rate": 6.883560791631307e-06, |
| "loss": 0.010482875257730484, |
| "step": 1150, |
| "token_acc": 0.9961371318203767 |
| }, |
| { |
| "epoch": 1.2368000000000001, |
| "grad_norm": 0.95703125, |
| "learning_rate": 6.7199249899049435e-06, |
| "loss": 0.012379509210586549, |
| "step": 1160, |
| "token_acc": 0.9954249939802552 |
| }, |
| { |
| "epoch": 1.2474666666666667, |
| "grad_norm": 2.953125, |
| "learning_rate": 6.557267567640217e-06, |
| "loss": 0.010692089051008224, |
| "step": 1170, |
| "token_acc": 0.9959271681839962 |
| }, |
| { |
| "epoch": 1.2581333333333333, |
| "grad_norm": 0.65234375, |
| "learning_rate": 6.3956370422279625e-06, |
| "loss": 0.0074609115719795225, |
| "step": 1180, |
| "token_acc": 0.997589780670041 |
| }, |
| { |
| "epoch": 1.2688, |
| "grad_norm": 2.9375, |
| "learning_rate": 6.235081624756629e-06, |
| "loss": 0.009892877936363221, |
| "step": 1190, |
| "token_acc": 0.9968636911942099 |
| }, |
| { |
| "epoch": 1.2794666666666665, |
| "grad_norm": 1.765625, |
| "learning_rate": 6.075649205631904e-06, |
| "loss": 0.011288212984800339, |
| "step": 1200, |
| "token_acc": 0.9954414587332053 |
| }, |
| { |
| "epoch": 1.2901333333333334, |
| "grad_norm": 2.421875, |
| "learning_rate": 5.917387340291969e-06, |
| "loss": 0.009337464720010758, |
| "step": 1210, |
| "token_acc": 0.9973595775324052 |
| }, |
| { |
| "epoch": 1.3008, |
| "grad_norm": 3.578125, |
| "learning_rate": 5.760343235022703e-06, |
| "loss": 0.017143648862838746, |
| "step": 1220, |
| "token_acc": 0.9942321557317952 |
| }, |
| { |
| "epoch": 1.3114666666666666, |
| "grad_norm": 2.15625, |
| "learning_rate": 5.604563732876989e-06, |
| "loss": 0.013416722416877747, |
| "step": 1230, |
| "token_acc": 0.9958887545344619 |
| }, |
| { |
| "epoch": 1.3221333333333334, |
| "grad_norm": 3.375, |
| "learning_rate": 5.450095299702392e-06, |
| "loss": 0.017079284787178038, |
| "step": 1240, |
| "token_acc": 0.9930238152513832 |
| }, |
| { |
| "epoch": 1.3328, |
| "grad_norm": 1.8515625, |
| "learning_rate": 5.296984010281367e-06, |
| "loss": 0.009312470257282258, |
| "step": 1250, |
| "token_acc": 0.9961722488038277 |
| }, |
| { |
| "epoch": 1.3434666666666666, |
| "grad_norm": 1.8828125, |
| "learning_rate": 5.145275534588053e-06, |
| "loss": 0.012822112441062928, |
| "step": 1260, |
| "token_acc": 0.995438175270108 |
| }, |
| { |
| "epoch": 1.3541333333333334, |
| "grad_norm": 0.92578125, |
| "learning_rate": 4.995015124165883e-06, |
| "loss": 0.0064118683338165285, |
| "step": 1270, |
| "token_acc": 0.9976047904191617 |
| }, |
| { |
| "epoch": 1.3648, |
| "grad_norm": 0.7265625, |
| "learning_rate": 4.8462475986299625e-06, |
| "loss": 0.015998595952987672, |
| "step": 1280, |
| "token_acc": 0.9946898382814385 |
| }, |
| { |
| "epoch": 1.3754666666666666, |
| "grad_norm": 1.2109375, |
| "learning_rate": 4.69901733229828e-06, |
| "loss": 0.011354514211416245, |
| "step": 1290, |
| "token_acc": 0.9952049868137137 |
| }, |
| { |
| "epoch": 1.3861333333333334, |
| "grad_norm": 1.421875, |
| "learning_rate": 4.5533682409557745e-06, |
| "loss": 0.01483256220817566, |
| "step": 1300, |
| "token_acc": 0.995706106870229 |
| }, |
| { |
| "epoch": 1.3968, |
| "grad_norm": 3.078125, |
| "learning_rate": 4.409343768755135e-06, |
| "loss": 0.009935888648033141, |
| "step": 1310, |
| "token_acc": 0.99592032637389 |
| }, |
| { |
| "epoch": 1.4074666666666666, |
| "grad_norm": 1.2578125, |
| "learning_rate": 4.266986875258299e-06, |
| "loss": 0.013563062250614166, |
| "step": 1320, |
| "token_acc": 0.9956730769230769 |
| }, |
| { |
| "epoch": 1.4181333333333335, |
| "grad_norm": 1.203125, |
| "learning_rate": 4.126340022622506e-06, |
| "loss": 0.011859767884016038, |
| "step": 1330, |
| "token_acc": 0.9961713328547499 |
| }, |
| { |
| "epoch": 1.4288, |
| "grad_norm": 0.73046875, |
| "learning_rate": 3.987445162934689e-06, |
| "loss": 0.007380545139312744, |
| "step": 1340, |
| "token_acc": 0.9973627427475426 |
| }, |
| { |
| "epoch": 1.4394666666666667, |
| "grad_norm": 1.359375, |
| "learning_rate": 3.850343725698018e-06, |
| "loss": 0.011476223170757294, |
| "step": 1350, |
| "token_acc": 0.9949640287769784 |
| }, |
| { |
| "epoch": 1.4501333333333333, |
| "grad_norm": 1.3046875, |
| "learning_rate": 3.7150766054743393e-06, |
| "loss": 0.008989098668098449, |
| "step": 1360, |
| "token_acc": 0.9966370405957242 |
| }, |
| { |
| "epoch": 1.4607999999999999, |
| "grad_norm": 1.1171875, |
| "learning_rate": 3.5816841496861563e-06, |
| "loss": 0.008315403014421463, |
| "step": 1370, |
| "token_acc": 0.9968772519817439 |
| }, |
| { |
| "epoch": 1.4714666666666667, |
| "grad_norm": 1.3203125, |
| "learning_rate": 3.4502061465818024e-06, |
| "loss": 0.012508057057857513, |
| "step": 1380, |
| "token_acc": 0.9963776865491427 |
| }, |
| { |
| "epoch": 1.4821333333333333, |
| "grad_norm": 1.9453125, |
| "learning_rate": 3.3206818133674478e-06, |
| "loss": 0.013706037402153015, |
| "step": 1390, |
| "token_acc": 0.9961436490720655 |
| }, |
| { |
| "epoch": 1.4928, |
| "grad_norm": 2.171875, |
| "learning_rate": 3.1931497845093753e-06, |
| "loss": 0.010326726734638214, |
| "step": 1400, |
| "token_acc": 0.9968936678614098 |
| }, |
| { |
| "epoch": 1.5034666666666667, |
| "grad_norm": 0.4296875, |
| "learning_rate": 3.0676481002101223e-06, |
| "loss": 0.00966293215751648, |
| "step": 1410, |
| "token_acc": 0.9968697327233326 |
| }, |
| { |
| "epoch": 1.5141333333333333, |
| "grad_norm": 1.71875, |
| "learning_rate": 2.9442141950618762e-06, |
| "loss": 0.010416048765182494, |
| "step": 1420, |
| "token_acc": 0.9968862275449102 |
| }, |
| { |
| "epoch": 1.5248, |
| "grad_norm": 1.7578125, |
| "learning_rate": 2.8228848868804714e-06, |
| "loss": 0.010529063642024994, |
| "step": 1430, |
| "token_acc": 0.9954337899543378 |
| }, |
| { |
| "epoch": 1.5354666666666668, |
| "grad_norm": 1.078125, |
| "learning_rate": 2.7036963657234105e-06, |
| "loss": 0.009810931980609894, |
| "step": 1440, |
| "token_acc": 0.9966208061790973 |
| }, |
| { |
| "epoch": 1.5461333333333334, |
| "grad_norm": 1.9609375, |
| "learning_rate": 2.5866841830950994e-06, |
| "loss": 0.010760371387004853, |
| "step": 1450, |
| "token_acc": 0.9954501915708812 |
| }, |
| { |
| "epoch": 1.5568, |
| "grad_norm": 1.0625, |
| "learning_rate": 2.47188324134255e-06, |
| "loss": 0.011224661022424698, |
| "step": 1460, |
| "token_acc": 0.9966434907695996 |
| }, |
| { |
| "epoch": 1.5674666666666668, |
| "grad_norm": 1.03125, |
| "learning_rate": 2.3593277832447405e-06, |
| "loss": 0.01038598120212555, |
| "step": 1470, |
| "token_acc": 0.9959552700452058 |
| }, |
| { |
| "epoch": 1.5781333333333334, |
| "grad_norm": 0.80078125, |
| "learning_rate": 2.2490513817986757e-06, |
| "loss": 0.010531876236200333, |
| "step": 1480, |
| "token_acc": 0.9968734968734969 |
| }, |
| { |
| "epoch": 1.5888, |
| "grad_norm": 1.8515625, |
| "learning_rate": 2.141086930205255e-06, |
| "loss": 0.010695824027061462, |
| "step": 1490, |
| "token_acc": 0.9956782713085234 |
| }, |
| { |
| "epoch": 1.5994666666666668, |
| "grad_norm": 1.5703125, |
| "learning_rate": 2.035466632057905e-06, |
| "loss": 0.007252989709377289, |
| "step": 1500, |
| "token_acc": 0.9966483121857793 |
| }, |
| { |
| "epoch": 1.6101333333333332, |
| "grad_norm": 0.18359375, |
| "learning_rate": 1.9322219917369e-06, |
| "loss": 0.007471411675214768, |
| "step": 1510, |
| "token_acc": 0.9971133028626413 |
| }, |
| { |
| "epoch": 1.6208, |
| "grad_norm": 2.15625, |
| "learning_rate": 1.8313838050122423e-06, |
| "loss": 0.015065881609916686, |
| "step": 1520, |
| "token_acc": 0.9946885562530179 |
| }, |
| { |
| "epoch": 1.6314666666666666, |
| "grad_norm": 2.46875, |
| "learning_rate": 1.7329821498579257e-06, |
| "loss": 0.013307811319828033, |
| "step": 1530, |
| "token_acc": 0.9947000722717417 |
| }, |
| { |
| "epoch": 1.6421333333333332, |
| "grad_norm": 1.375, |
| "learning_rate": 1.6370463774802759e-06, |
| "loss": 0.008927606046199799, |
| "step": 1540, |
| "token_acc": 0.9973513123043583 |
| }, |
| { |
| "epoch": 1.6528, |
| "grad_norm": 2.296875, |
| "learning_rate": 1.5436051035630962e-06, |
| "loss": 0.010501272231340408, |
| "step": 1550, |
| "token_acc": 0.9963820549927641 |
| }, |
| { |
| "epoch": 1.6634666666666666, |
| "grad_norm": 1.8046875, |
| "learning_rate": 1.4526861997321794e-06, |
| "loss": 0.013346484303474427, |
| "step": 1560, |
| "token_acc": 0.9951969260326609 |
| }, |
| { |
| "epoch": 1.6741333333333333, |
| "grad_norm": 1.1015625, |
| "learning_rate": 1.3643167852417894e-06, |
| "loss": 0.011042323708534241, |
| "step": 1570, |
| "token_acc": 0.9959359311498924 |
| }, |
| { |
| "epoch": 1.6848, |
| "grad_norm": 1.6953125, |
| "learning_rate": 1.278523218885518e-06, |
| "loss": 0.008789537847042084, |
| "step": 1580, |
| "token_acc": 0.9968877184582237 |
| }, |
| { |
| "epoch": 1.6954666666666667, |
| "grad_norm": 1.5546875, |
| "learning_rate": 1.1953310911340165e-06, |
| "loss": 0.009821423888206482, |
| "step": 1590, |
| "token_acc": 0.9949507093051214 |
| }, |
| { |
| "epoch": 1.7061333333333333, |
| "grad_norm": 2.09375, |
| "learning_rate": 1.1147652165018597e-06, |
| "loss": 0.009412066638469696, |
| "step": 1600, |
| "token_acc": 0.9966378482228626 |
| }, |
| { |
| "epoch": 1.7168, |
| "grad_norm": 0.703125, |
| "learning_rate": 1.0368496261459004e-06, |
| "loss": 0.011984483152627946, |
| "step": 1610, |
| "token_acc": 0.995906573561281 |
| }, |
| { |
| "epoch": 1.7274666666666667, |
| "grad_norm": 1.1875, |
| "learning_rate": 9.616075606972597e-07, |
| "loss": 0.007756148278713226, |
| "step": 1620, |
| "token_acc": 0.9971181556195965 |
| }, |
| { |
| "epoch": 1.7381333333333333, |
| "grad_norm": 2.046875, |
| "learning_rate": 8.890614633291095e-07, |
| "loss": 0.00872180163860321, |
| "step": 1630, |
| "token_acc": 0.9966313763233878 |
| }, |
| { |
| "epoch": 1.7488000000000001, |
| "grad_norm": 2.546875, |
| "learning_rate": 8.192329730623471e-07, |
| "loss": 0.011189891397953034, |
| "step": 1640, |
| "token_acc": 0.9973765800143095 |
| }, |
| { |
| "epoch": 1.7594666666666665, |
| "grad_norm": 1.3671875, |
| "learning_rate": 7.521429183111139e-07, |
| "loss": 0.011723892390727996, |
| "step": 1650, |
| "token_acc": 0.9959036144578314 |
| }, |
| { |
| "epoch": 1.7701333333333333, |
| "grad_norm": 2.3125, |
| "learning_rate": 6.878113106700946e-07, |
| "loss": 0.007294324040412903, |
| "step": 1660, |
| "token_acc": 0.9973493975903615 |
| }, |
| { |
| "epoch": 1.7808000000000002, |
| "grad_norm": 0.42578125, |
| "learning_rate": 6.262573389454773e-07, |
| "loss": 0.008933399617671967, |
| "step": 1670, |
| "token_acc": 0.9963741841914431 |
| }, |
| { |
| "epoch": 1.7914666666666665, |
| "grad_norm": 1.1796875, |
| "learning_rate": 5.674993634313242e-07, |
| "loss": 0.012564325332641601, |
| "step": 1680, |
| "token_acc": 0.9956720365472469 |
| }, |
| { |
| "epoch": 1.8021333333333334, |
| "grad_norm": 0.5625, |
| "learning_rate": 5.115549104330575e-07, |
| "loss": 0.00787913054227829, |
| "step": 1690, |
| "token_acc": 0.9976082276967233 |
| }, |
| { |
| "epoch": 1.8128, |
| "grad_norm": 1.1875, |
| "learning_rate": 4.5844066703973676e-07, |
| "loss": 0.007952670753002166, |
| "step": 1700, |
| "token_acc": 0.9973583093179635 |
| }, |
| { |
| "epoch": 1.8234666666666666, |
| "grad_norm": 1.0546875, |
| "learning_rate": 4.081724761466288e-07, |
| "loss": 0.013583377003669739, |
| "step": 1710, |
| "token_acc": 0.9946898382814385 |
| }, |
| { |
| "epoch": 1.8341333333333334, |
| "grad_norm": 0.9609375, |
| "learning_rate": 3.6076533172959895e-07, |
| "loss": 0.008094549179077148, |
| "step": 1720, |
| "token_acc": 0.9973519499277804 |
| }, |
| { |
| "epoch": 1.8448, |
| "grad_norm": 1.3046875, |
| "learning_rate": 3.1623337437272084e-07, |
| "loss": 0.010456332564353943, |
| "step": 1730, |
| "token_acc": 0.9963864129125511 |
| }, |
| { |
| "epoch": 1.8554666666666666, |
| "grad_norm": 1.375, |
| "learning_rate": 2.745898870504116e-07, |
| "loss": 0.009122979640960694, |
| "step": 1740, |
| "token_acc": 0.9980750721847931 |
| }, |
| { |
| "epoch": 1.8661333333333334, |
| "grad_norm": 2.0625, |
| "learning_rate": 2.3584729116541305e-07, |
| "loss": 0.010391271114349366, |
| "step": 1750, |
| "token_acc": 0.9958986731001206 |
| }, |
| { |
| "epoch": 1.8768, |
| "grad_norm": 1.171875, |
| "learning_rate": 2.0001714284372253e-07, |
| "loss": 0.009891322255134583, |
| "step": 1760, |
| "token_acc": 0.9959193470955353 |
| }, |
| { |
| "epoch": 1.8874666666666666, |
| "grad_norm": 1.359375, |
| "learning_rate": 1.6711012948764426e-07, |
| "loss": 0.010584476590156554, |
| "step": 1770, |
| "token_acc": 0.9956948098541019 |
| }, |
| { |
| "epoch": 1.8981333333333335, |
| "grad_norm": 2.65625, |
| "learning_rate": 1.371360665879584e-07, |
| "loss": 0.006497502326965332, |
| "step": 1780, |
| "token_acc": 0.9978401727861771 |
| }, |
| { |
| "epoch": 1.9088, |
| "grad_norm": 1.265625, |
| "learning_rate": 1.1010389479615302e-07, |
| "loss": 0.011741268634796142, |
| "step": 1790, |
| "token_acc": 0.9944698244770378 |
| }, |
| { |
| "epoch": 1.9194666666666667, |
| "grad_norm": 2.234375, |
| "learning_rate": 8.60216772576139e-08, |
| "loss": 0.010211998224258423, |
| "step": 1800, |
| "token_acc": 0.9961566178236848 |
| }, |
| { |
| "epoch": 1.9301333333333335, |
| "grad_norm": 1.5546875, |
| "learning_rate": 6.489659720655295e-08, |
| "loss": 0.008484911918640137, |
| "step": 1810, |
| "token_acc": 0.9968712394705175 |
| }, |
| { |
| "epoch": 1.9407999999999999, |
| "grad_norm": 2.0, |
| "learning_rate": 4.673495582339649e-08, |
| "loss": 0.011189290881156921, |
| "step": 1820, |
| "token_acc": 0.9954050785973397 |
| }, |
| { |
| "epoch": 1.9514666666666667, |
| "grad_norm": 1.5, |
| "learning_rate": 3.154217035526874e-08, |
| "loss": 0.012943205237388612, |
| "step": 1830, |
| "token_acc": 0.9949604031677466 |
| }, |
| { |
| "epoch": 1.9621333333333333, |
| "grad_norm": 2.078125, |
| "learning_rate": 1.932277250014214e-08, |
| "loss": 0.014319291710853577, |
| "step": 1840, |
| "token_acc": 0.9944883776659478 |
| }, |
| { |
| "epoch": 1.9727999999999999, |
| "grad_norm": 1.1640625, |
| "learning_rate": 1.008040705511748e-08, |
| "loss": 0.0120982825756073, |
| "step": 1850, |
| "token_acc": 0.9949016751638747 |
| }, |
| { |
| "epoch": 1.9834666666666667, |
| "grad_norm": 1.4609375, |
| "learning_rate": 3.8178308292624585e-09, |
| "loss": 0.00735490694642067, |
| "step": 1860, |
| "token_acc": 0.997832891885384 |
| }, |
| { |
| "epoch": 1.9941333333333333, |
| "grad_norm": 1.859375, |
| "learning_rate": 5.369118213094382e-10, |
| "loss": 0.009274058043956757, |
| "step": 1870, |
| "token_acc": 0.9959075589792971 |
| } |
| ], |
| "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": 7.528224876952289e+17, |
| "train_batch_size": 8, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
