dzungpham/graphcodebert-code-classification
0
1{2 "best_global_step": 2000,3 "best_metric": 0.797186028145479,4 "best_model_checkpoint": "./output_checkpoints/graphcodebert-rdrop/checkpoint-2000",5 "epoch": 0.128,6 "eval_steps": 1000,7 "global_step": 2000,8 "is_hyper_param_search": false,9 "is_local_process_zero": true,10 "is_world_process_zero": true,11 "log_history": [12 {13 "epoch": 0.00064,14 "grad_norm": 1.8688431978225708,15 "learning_rate": 3.840409643695328e-08,16 "loss": 0.7168,17 "step": 1018 },19 {20 "epoch": 0.00128,21 "grad_norm": 1.835353970527649,22 "learning_rate": 8.10753147002347e-08,23 "loss": 0.7179,24 "step": 2025 },26 {27 "epoch": 0.00192,28 "grad_norm": 1.2541388273239136,29 "learning_rate": 1.2374653296351612e-07,30 "loss": 0.7177,31 "step": 3032 },33 {34 "epoch": 0.00256,35 "grad_norm": 1.1381633281707764,36 "learning_rate": 1.6641775122679754e-07,37 "loss": 0.7312,38 "step": 4039 },40 {41 "epoch": 0.0032,42 "grad_norm": 1.6313824653625488,43 "learning_rate": 2.0908896949007894e-07,44 "loss": 0.7167,45 "step": 5046 },47 {48 "epoch": 0.00384,49 "grad_norm": 1.092464566230774,50 "learning_rate": 2.517601877533604e-07,51 "loss": 0.7201,52 "step": 6053 },54 {55 "epoch": 0.00448,56 "grad_norm": 1.164099097251892,57 "learning_rate": 2.944314060166418e-07,58 "loss": 0.7195,59 "step": 7060 },61 {62 "epoch": 0.00512,63 "grad_norm": 1.9283920526504517,64 "learning_rate": 3.371026242799232e-07,65 "loss": 0.7223,66 "step": 8067 },68 {69 "epoch": 0.00576,70 "grad_norm": 1.8716129064559937,71 "learning_rate": 3.7977384254320464e-07,72 "loss": 0.7151,73 "step": 9074 },75 {76 "epoch": 0.0064,77 "grad_norm": 1.5499262809753418,78 "learning_rate": 4.22445060806486e-07,79 "loss": 0.7211,80 "step": 10081 },82 {83 "epoch": 0.00704,84 "grad_norm": 1.375114917755127,85 "learning_rate": 4.651162790697675e-07,86 "loss": 0.7247,87 "step": 11088 },89 {90 "epoch": 0.00768,91 "grad_norm": 1.384915828704834,92 "learning_rate": 5.077874973330489e-07,93 "loss": 0.7156,94 "step": 12095 },96 {97 "epoch": 0.00832,98 "grad_norm": 1.6376659870147705,99 "learning_rate": 5.504587155963304e-07,100 "loss": 0.7196,101 "step": 130102 },103 {104 "epoch": 0.00896,105 "grad_norm": 1.709489345550537,106 "learning_rate": 5.931299338596117e-07,107 "loss": 0.7215,108 "step": 140109 },110 {111 "epoch": 0.0096,112 "grad_norm": 1.350512146949768,113 "learning_rate": 6.358011521228932e-07,114 "loss": 0.71,115 "step": 150116 },117 {118 "epoch": 0.01024,119 "grad_norm": 2.33050537109375,120 "learning_rate": 6.784723703861745e-07,121 "loss": 0.7191,122 "step": 160123 },124 {125 "epoch": 0.01088,126 "grad_norm": 1.0042874813079834,127 "learning_rate": 7.21143588649456e-07,128 "loss": 0.72,129 "step": 170130 },131 {132 "epoch": 0.01152,133 "grad_norm": 1.1835744380950928,134 "learning_rate": 7.638148069127374e-07,135 "loss": 0.7122,136 "step": 180137 },138 {139 "epoch": 0.01216,140 "grad_norm": 1.949506402015686,141 "learning_rate": 8.064860251760189e-07,142 "loss": 0.7091,143 "step": 190144 },145 {146 "epoch": 0.0128,147 "grad_norm": 1.139626383781433,148 "learning_rate": 8.491572434393003e-07,149 "loss": 0.7046,150 "step": 200151 },152 {153 "epoch": 0.01344,154 "grad_norm": 1.7734779119491577,155 "learning_rate": 8.918284617025817e-07,156 "loss": 0.7135,157 "step": 210158 },159 {160 "epoch": 0.01408,161 "grad_norm": 1.3442974090576172,162 "learning_rate": 9.344996799658632e-07,163 "loss": 0.7127,164 "step": 220165 },166 {167 "epoch": 0.01472,168 "grad_norm": 1.6148335933685303,169 "learning_rate": 9.771708982291445e-07,170 "loss": 0.7168,171 "step": 230172 },173 {174 "epoch": 0.01536,175 "grad_norm": 0.9220213294029236,176 "learning_rate": 1.0198421164924258e-06,177 "loss": 0.7109,178 "step": 240179 },180 {181 "epoch": 0.016,182 "grad_norm": 1.5631815195083618,183 "learning_rate": 1.0625133347557074e-06,184 "loss": 0.7149,185 "step": 250186 },187 {188 "epoch": 0.01664,189 "grad_norm": 1.5052822828292847,190 "learning_rate": 1.1051845530189888e-06,191 "loss": 0.7112,192 "step": 260193 },194 {195 "epoch": 0.01728,196 "grad_norm": 1.6714903116226196,197 "learning_rate": 1.1478557712822702e-06,198 "loss": 0.7128,199 "step": 270200 },201 {202 "epoch": 0.01792,203 "grad_norm": 1.6667262315750122,204 "learning_rate": 1.1905269895455517e-06,205 "loss": 0.7091,206 "step": 280207 },208 {209 "epoch": 0.01856,210 "grad_norm": 1.6861907243728638,211 "learning_rate": 1.233198207808833e-06,212 "loss": 0.7078,213 "step": 290214 },215 {216 "epoch": 0.0192,217 "grad_norm": 1.6259212493896484,218 "learning_rate": 1.2758694260721145e-06,219 "loss": 0.7087,220 "step": 300221 },222 {223 "epoch": 0.01984,224 "grad_norm": 1.207320213317871,225 "learning_rate": 1.318540644335396e-06,226 "loss": 0.7087,227 "step": 310228 },229 {230 "epoch": 0.02048,231 "grad_norm": 1.087124228477478,232 "learning_rate": 1.3612118625986772e-06,233 "loss": 0.7174,234 "step": 320235 },236 {237 "epoch": 0.02112,238 "grad_norm": 1.5668721199035645,239 "learning_rate": 1.4038830808619588e-06,240 "loss": 0.7066,241 "step": 330242 },243 {244 "epoch": 0.02176,245 "grad_norm": 1.5332859754562378,246 "learning_rate": 1.4465542991252401e-06,247 "loss": 0.7115,248 "step": 340249 },250 {251 "epoch": 0.0224,252 "grad_norm": 2.092994451522827,253 "learning_rate": 1.4892255173885215e-06,254 "loss": 0.7085,255 "step": 350256 },257 {258 "epoch": 0.02304,259 "grad_norm": 1.266851544380188,260 "learning_rate": 1.531896735651803e-06,261 "loss": 0.7091,262 "step": 360263 },264 {265 "epoch": 0.02368,266 "grad_norm": 1.290616512298584,267 "learning_rate": 1.5745679539150842e-06,268 "loss": 0.7103,269 "step": 370270 },271 {272 "epoch": 0.02432,273 "grad_norm": 1.2045000791549683,274 "learning_rate": 1.6172391721783658e-06,275 "loss": 0.705,276 "step": 380277 },278 {279 "epoch": 0.02496,280 "grad_norm": 1.828832983970642,281 "learning_rate": 1.6599103904416472e-06,282 "loss": 0.7038,283 "step": 390284 },285 {286 "epoch": 0.0256,287 "grad_norm": 1.7793196439743042,288 "learning_rate": 1.7025816087049288e-06,289 "loss": 0.7085,290 "step": 400291 },292 {293 "epoch": 0.02624,294 "grad_norm": 0.8949472904205322,295 "learning_rate": 1.7452528269682101e-06,296 "loss": 0.7045,297 "step": 410298 },299 {300 "epoch": 0.02688,301 "grad_norm": 1.3071945905685425,302 "learning_rate": 1.7879240452314913e-06,303 "loss": 0.702,304 "step": 420305 },306 {307 "epoch": 0.02752,308 "grad_norm": 1.7497148513793945,309 "learning_rate": 1.8305952634947729e-06,310 "loss": 0.7073,311 "step": 430312 },313 {314 "epoch": 0.02816,315 "grad_norm": 1.0856297016143799,316 "learning_rate": 1.8732664817580542e-06,317 "loss": 0.7008,318 "step": 440319 },320 {321 "epoch": 0.0288,322 "grad_norm": 1.369019627571106,323 "learning_rate": 1.915937700021336e-06,324 "loss": 0.7059,325 "step": 450326 },327 {328 "epoch": 0.02944,329 "grad_norm": 2.0961010456085205,330 "learning_rate": 1.958608918284617e-06,331 "loss": 0.6952,332 "step": 460333 },334 {335 "epoch": 0.03008,336 "grad_norm": 1.6076347827911377,337 "learning_rate": 2.0012801365478988e-06,338 "loss": 0.7074,339 "step": 470340 },341 {342 "epoch": 0.03072,343 "grad_norm": 1.3129311800003052,344 "learning_rate": 2.04395135481118e-06,345 "loss": 0.6994,346 "step": 480347 },348 {349 "epoch": 0.03136,350 "grad_norm": 1.3621476888656616,351 "learning_rate": 2.0866225730744615e-06,352 "loss": 0.696,353 "step": 490354 },355 {356 "epoch": 0.032,357 "grad_norm": 0.9786806106567383,358 "learning_rate": 2.129293791337743e-06,359 "loss": 0.7011,360 "step": 500361 },362 {363 "epoch": 0.03264,364 "grad_norm": 1.4611176252365112,365 "learning_rate": 2.1719650096010242e-06,366 "loss": 0.699,367 "step": 510368 },369 {370 "epoch": 0.03328,371 "grad_norm": 1.0675945281982422,372 "learning_rate": 2.214636227864306e-06,373 "loss": 0.6906,374 "step": 520375 },376 {377 "epoch": 0.03392,378 "grad_norm": 1.6564017534255981,379 "learning_rate": 2.257307446127587e-06,380 "loss": 0.6923,381 "step": 530382 },383 {384 "epoch": 0.03456,385 "grad_norm": 1.228119134902954,386 "learning_rate": 2.2999786643908685e-06,387 "loss": 0.6928,388 "step": 540389 },390 {391 "epoch": 0.0352,392 "grad_norm": 1.9868593215942383,393 "learning_rate": 2.34264988265415e-06,394 "loss": 0.6912,395 "step": 550396 },397 {398 "epoch": 0.03584,399 "grad_norm": 1.2531176805496216,400 "learning_rate": 2.3853211009174317e-06,401 "loss": 0.6923,402 "step": 560403 },404 {405 "epoch": 0.03648,406 "grad_norm": 1.413602352142334,407 "learning_rate": 2.427992319180713e-06,408 "loss": 0.6993,409 "step": 570410 },411 {412 "epoch": 0.03712,413 "grad_norm": 0.9655390977859497,414 "learning_rate": 2.470663537443994e-06,415 "loss": 0.7017,416 "step": 580417 },418 {419 "epoch": 0.03776,420 "grad_norm": 1.962438941001892,421 "learning_rate": 2.5133347557072756e-06,422 "loss": 0.6933,423 "step": 590424 },425 {426 "epoch": 0.0384,427 "grad_norm": 1.1099931001663208,428 "learning_rate": 2.556005973970557e-06,429 "loss": 0.6925,430 "step": 600431 },432 {433 "epoch": 0.03904,434 "grad_norm": 1.1766624450683594,435 "learning_rate": 2.5986771922338383e-06,436 "loss": 0.6894,437 "step": 610438 },439 {440 "epoch": 0.03968,441 "grad_norm": 1.574353814125061,442 "learning_rate": 2.64134841049712e-06,443 "loss": 0.6923,444 "step": 620445 },446 {447 "epoch": 0.04032,448 "grad_norm": 1.194074273109436,449 "learning_rate": 2.6840196287604015e-06,450 "loss": 0.6855,451 "step": 630452 },453 {454 "epoch": 0.04096,455 "grad_norm": 1.1750593185424805,456 "learning_rate": 2.7266908470236826e-06,457 "loss": 0.7002,458 "step": 640459 },460 {461 "epoch": 0.0416,462 "grad_norm": 1.3165347576141357,463 "learning_rate": 2.7693620652869642e-06,464 "loss": 0.6893,465 "step": 650466 },467 {468 "epoch": 0.04224,469 "grad_norm": 1.1079384088516235,470 "learning_rate": 2.812033283550246e-06,471 "loss": 0.6852,472 "step": 660473 },474 {475 "epoch": 0.04288,476 "grad_norm": 1.231327772140503,477 "learning_rate": 2.8547045018135274e-06,478 "loss": 0.695,479 "step": 670480 },481 {482 "epoch": 0.04352,483 "grad_norm": 1.966036319732666,484 "learning_rate": 2.897375720076808e-06,485 "loss": 0.6833,486 "step": 680487 },488 {489 "epoch": 0.04416,490 "grad_norm": 1.0460131168365479,491 "learning_rate": 2.9400469383400897e-06,492 "loss": 0.6802,493 "step": 690494 },495 {496 "epoch": 0.0448,497 "grad_norm": 1.2445200681686401,498 "learning_rate": 2.9827181566033713e-06,499 "loss": 0.6883,500 "step": 700501 },502 {503 "epoch": 0.04544,504 "grad_norm": 1.1759636402130127,505 "learning_rate": 3.025389374866653e-06,506 "loss": 0.682,507 "step": 710508 },509 {510 "epoch": 0.04608,511 "grad_norm": 1.2478561401367188,512 "learning_rate": 3.068060593129934e-06,513 "loss": 0.6802,514 "step": 720515 },516 {517 "epoch": 0.04672,518 "grad_norm": 0.8945108652114868,519 "learning_rate": 3.1107318113932156e-06,520 "loss": 0.6873,521 "step": 730522 },523 {524 "epoch": 0.04736,525 "grad_norm": 1.6810317039489746,526 "learning_rate": 3.153403029656497e-06,527 "loss": 0.6828,528 "step": 740529 },530 {531 "epoch": 0.048,532 "grad_norm": 2.882283926010132,533 "learning_rate": 3.1960742479197783e-06,534 "loss": 0.6797,535 "step": 750536 },537 {538 "epoch": 0.04864,539 "grad_norm": 1.575766921043396,540 "learning_rate": 3.23874546618306e-06,541 "loss": 0.6826,542 "step": 760543 },544 {545 "epoch": 0.04928,546 "grad_norm": 1.8044737577438354,547 "learning_rate": 3.2814166844463415e-06,548 "loss": 0.6837,549 "step": 770550 },551 {552 "epoch": 0.04992,553 "grad_norm": 1.4755513668060303,554 "learning_rate": 3.324087902709623e-06,555 "loss": 0.6815,556 "step": 780557 },558 {559 "epoch": 0.05056,560 "grad_norm": 2.0735654830932617,561 "learning_rate": 3.3667591209729038e-06,562 "loss": 0.6773,563 "step": 790564 },565 {566 "epoch": 0.0512,567 "grad_norm": 0.9823655486106873,568 "learning_rate": 3.4094303392361854e-06,569 "loss": 0.6689,570 "step": 800571 },572 {573 "epoch": 0.05184,574 "grad_norm": 1.0731920003890991,575 "learning_rate": 3.452101557499467e-06,576 "loss": 0.6689,577 "step": 810578 },579 {580 "epoch": 0.05248,581 "grad_norm": 1.1081087589263916,582 "learning_rate": 3.494772775762748e-06,583 "loss": 0.6694,584 "step": 820585 },586 {587 "epoch": 0.05312,588 "grad_norm": 1.2130361795425415,589 "learning_rate": 3.5374439940260297e-06,590 "loss": 0.682,591 "step": 830592 },593 {594 "epoch": 0.05376,595 "grad_norm": 1.170345664024353,596 "learning_rate": 3.5801152122893113e-06,597 "loss": 0.6756,598 "step": 840599 },600 {601 "epoch": 0.0544,602 "grad_norm": 1.2610124349594116,603 "learning_rate": 3.622786430552593e-06,604 "loss": 0.67,605 "step": 850606 },607 {608 "epoch": 0.05504,609 "grad_norm": 1.0138545036315918,610 "learning_rate": 3.665457648815874e-06,611 "loss": 0.6807,612 "step": 860613 },614 {615 "epoch": 0.05568,616 "grad_norm": 1.0937036275863647,617 "learning_rate": 3.7081288670791556e-06,618 "loss": 0.6774,619 "step": 870620 },621 {622 "epoch": 0.05632,623 "grad_norm": 1.119221568107605,624 "learning_rate": 3.750800085342437e-06,625 "loss": 0.6728,626 "step": 880627 },628 {629 "epoch": 0.05696,630 "grad_norm": 1.9351646900177002,631 "learning_rate": 3.793471303605718e-06,632 "loss": 0.6664,633 "step": 890634 },635 {636 "epoch": 0.0576,637 "grad_norm": 1.6208665370941162,638 "learning_rate": 3.836142521869e-06,639 "loss": 0.6653,640 "step": 900641 },642 {643 "epoch": 0.05824,644 "grad_norm": 1.3460793495178223,645 "learning_rate": 3.8788137401322815e-06,646 "loss": 0.6651,647 "step": 910648 },649 {650 "epoch": 0.05888,651 "grad_norm": 1.8987629413604736,652 "learning_rate": 3.921484958395563e-06,653 "loss": 0.6694,654 "step": 920655 },656 {657 "epoch": 0.05952,658 "grad_norm": 1.3637055158615112,659 "learning_rate": 3.964156176658844e-06,660 "loss": 0.6691,661 "step": 930662 },663 {664 "epoch": 0.06016,665 "grad_norm": 1.3319580554962158,666 "learning_rate": 4.006827394922125e-06,667 "loss": 0.6749,668 "step": 940669 },670 {671 "epoch": 0.0608,672 "grad_norm": 1.3574756383895874,673 "learning_rate": 4.049498613185407e-06,674 "loss": 0.6688,675 "step": 950676 },677 {678 "epoch": 0.06144,679 "grad_norm": 1.136319637298584,680 "learning_rate": 4.092169831448688e-06,681 "loss": 0.6657,682 "step": 960683 },684 {685 "epoch": 0.06208,686 "grad_norm": 1.7666376829147339,687 "learning_rate": 4.134841049711969e-06,688 "loss": 0.6667,689 "step": 970690 },691 {692 "epoch": 0.06272,693 "grad_norm": 1.332014560699463,694 "learning_rate": 4.177512267975251e-06,695 "loss": 0.6715,696 "step": 980697 },698 {699 "epoch": 0.06336,700 "grad_norm": 1.3127824068069458,701 "learning_rate": 4.220183486238532e-06,702 "loss": 0.6737,703 "step": 990704 },705 {706 "epoch": 0.064,707 "grad_norm": 1.293678641319275,708 "learning_rate": 4.262854704501814e-06,709 "loss": 0.6557,710 "step": 1000711 },712 {713 "epoch": 0.064,714 "eval_accuracy": 0.76589,715 "eval_loss": 0.6282070279121399,716 "eval_macro_f1": 0.7657238614598489,717 "eval_precision": 0.7708542319194018,718 "eval_recall": 0.7687514011652705,719 "eval_runtime": 806.5969,720 "eval_samples_per_second": 123.978,721 "eval_steps_per_second": 1.938,722 "step": 1000723 },724 {725 "epoch": 0.06464,726 "grad_norm": 1.2433074712753296,727 "learning_rate": 4.3055259227650956e-06,728 "loss": 0.6583,729 "step": 1010730 },731 {732 "epoch": 0.06528,733 "grad_norm": 1.5115686655044556,734 "learning_rate": 4.348197141028377e-06,735 "loss": 0.6659,736 "step": 1020737 },738 {739 "epoch": 0.06592,740 "grad_norm": 1.3560088872909546,741 "learning_rate": 4.390868359291658e-06,742 "loss": 0.6675,743 "step": 1030744 },745 {746 "epoch": 0.06656,747 "grad_norm": 0.8229928016662598,748 "learning_rate": 4.4335395775549394e-06,749 "loss": 0.6629,750 "step": 1040751 },752 {753 "epoch": 0.0672,754 "grad_norm": 1.5138262510299683,755 "learning_rate": 4.476210795818221e-06,756 "loss": 0.6635,757 "step": 1050758 },759 {760 "epoch": 0.06784,761 "grad_norm": 1.1732233762741089,762 "learning_rate": 4.518882014081503e-06,763 "loss": 0.6531,764 "step": 1060765 },766 {767 "epoch": 0.06848,768 "grad_norm": 1.6118066310882568,769 "learning_rate": 4.561553232344783e-06,770 "loss": 0.6644,771 "step": 1070772 },773 {774 "epoch": 0.06912,775 "grad_norm": 0.9575800895690918,776 "learning_rate": 4.604224450608065e-06,777 "loss": 0.6542,778 "step": 1080779 },780 {781 "epoch": 0.06976,782 "grad_norm": 1.157271146774292,783 "learning_rate": 4.6468956688713465e-06,784 "loss": 0.6505,785 "step": 1090786 },787 {788 "epoch": 0.0704,789 "grad_norm": 1.4502589702606201,790 "learning_rate": 4.689566887134628e-06,791 "loss": 0.6725,792 "step": 1100793 },794 {795 "epoch": 0.07104,796 "grad_norm": 1.122251033782959,797 "learning_rate": 4.73223810539791e-06,798 "loss": 0.6625,799 "step": 1110800 },801 {802 "epoch": 0.07168,803 "grad_norm": 0.9529566764831543,804 "learning_rate": 4.774909323661191e-06,805 "loss": 0.6684,806 "step": 1120807 },808 {809 "epoch": 0.07232,810 "grad_norm": 1.6433424949645996,811 "learning_rate": 4.817580541924473e-06,812 "loss": 0.6599,813 "step": 1130814 },815 {816 "epoch": 0.07296,817 "grad_norm": 1.6565576791763306,818 "learning_rate": 4.8602517601877535e-06,819 "loss": 0.6591,820 "step": 1140821 },822 {823 "epoch": 0.0736,824 "grad_norm": 1.3955042362213135,825 "learning_rate": 4.902922978451035e-06,826 "loss": 0.6649,827 "step": 1150828 },829 {830 "epoch": 0.07424,831 "grad_norm": 1.530819296836853,832 "learning_rate": 4.945594196714317e-06,833 "loss": 0.65,834 "step": 1160835 },836 {837 "epoch": 0.07488,838 "grad_norm": 1.0729092359542847,839 "learning_rate": 4.9882654149775974e-06,840 "loss": 0.6603,841 "step": 1170842 },843 {844 "epoch": 0.07552,845 "grad_norm": 1.6307648420333862,846 "learning_rate": 5.030936633240879e-06,847 "loss": 0.6702,848 "step": 1180849 },850 {851 "epoch": 0.07616,852 "grad_norm": 1.324406623840332,853 "learning_rate": 5.073607851504161e-06,854 "loss": 0.6634,855 "step": 1190856 },857 {858 "epoch": 0.0768,859 "grad_norm": 1.1919140815734863,860 "learning_rate": 5.116279069767442e-06,861 "loss": 0.6586,862 "step": 1200863 },864 {865 "epoch": 0.07744,866 "grad_norm": 1.0658751726150513,867 "learning_rate": 5.158950288030724e-06,868 "loss": 0.6573,869 "step": 1210870 },871 {872 "epoch": 0.07808,873 "grad_norm": 1.7762951850891113,874 "learning_rate": 5.201621506294005e-06,875 "loss": 0.6596,876 "step": 1220877 },878 {879 "epoch": 0.07872,880 "grad_norm": 1.1911667585372925,881 "learning_rate": 5.244292724557286e-06,882 "loss": 0.651,883 "step": 1230884 },885 {886 "epoch": 0.07936,887 "grad_norm": 1.442029595375061,888 "learning_rate": 5.286963942820568e-06,889 "loss": 0.6598,890 "step": 1240891 },892 {893 "epoch": 0.08,894 "grad_norm": 0.9955100417137146,895 "learning_rate": 5.329635161083849e-06,896 "loss": 0.6587,897 "step": 1250898 },899 {900 "epoch": 0.08064,901 "grad_norm": 1.1801563501358032,902 "learning_rate": 5.372306379347131e-06,903 "loss": 0.6669,904 "step": 1260905 },906 {907 "epoch": 0.08128,908 "grad_norm": 1.8548506498336792,909 "learning_rate": 5.414977597610412e-06,910 "loss": 0.6534,911 "step": 1270912 },913 {914 "epoch": 0.08192,915 "grad_norm": 1.8297194242477417,916 "learning_rate": 5.457648815873694e-06,917 "loss": 0.668,918 "step": 1280919 },920 {921 "epoch": 0.08256,922 "grad_norm": 1.3471113443374634,923 "learning_rate": 5.5003200341369755e-06,924 "loss": 0.659,925 "step": 1290926 },927 {928 "epoch": 0.0832,929 "grad_norm": 1.4015250205993652,930 "learning_rate": 5.542991252400256e-06,931 "loss": 0.6602,932 "step": 1300933 },934 {935 "epoch": 0.08384,936 "grad_norm": 1.0796362161636353,937 "learning_rate": 5.585662470663538e-06,938 "loss": 0.6652,939 "step": 1310940 },941 {942 "epoch": 0.08448,943 "grad_norm": 2.0007293224334717,944 "learning_rate": 5.628333688926819e-06,945 "loss": 0.6573,946 "step": 1320947 },948 {949 "epoch": 0.08512,950 "grad_norm": 1.3140546083450317,951 "learning_rate": 5.671004907190101e-06,952 "loss": 0.6642,953 "step": 1330954 },955 {956 "epoch": 0.08576,957 "grad_norm": 1.6801965236663818,958 "learning_rate": 5.7136761254533826e-06,959 "loss": 0.6527,960 "step": 1340961 },962 {963 "epoch": 0.0864,964 "grad_norm": 1.9289913177490234,965 "learning_rate": 5.756347343716664e-06,966 "loss": 0.6567,967 "step": 1350968 },969 {970 "epoch": 0.08704,971 "grad_norm": 1.420436143875122,972 "learning_rate": 5.799018561979946e-06,973 "loss": 0.6479,974 "step": 1360975 },976 {977 "epoch": 0.08768,978 "grad_norm": 1.3068914413452148,979 "learning_rate": 5.841689780243226e-06,980 "loss": 0.6522,981 "step": 1370982 },983 {984 "epoch": 0.08832,985 "grad_norm": 1.8793392181396484,986 "learning_rate": 5.884360998506507e-06,987 "loss": 0.6491,988 "step": 1380989 },990 {991 "epoch": 0.08896,992 "grad_norm": 1.3998247385025024,993 "learning_rate": 5.927032216769789e-06,994 "loss": 0.6638,995 "step": 1390996 },997 {998 "epoch": 0.0896,999 "grad_norm": 1.371824026107788,1000 "learning_rate": 5.96970343503307e-06,1001 "loss": 0.6517,1002 "step": 14001003 },1004 {1005 "epoch": 0.09024,1006 "grad_norm": 2.0050251483917236,1007 "learning_rate": 6.012374653296352e-06,1008 "loss": 0.6556,1009 "step": 14101010 },1011 {1012 "epoch": 0.09088,1013 "grad_norm": 1.499855399131775,1014 "learning_rate": 6.0550458715596335e-06,1015 "loss": 0.6597,1016 "step": 14201017 },1018 {1019 "epoch": 0.09152,1020 "grad_norm": 1.2066516876220703,1021 "learning_rate": 6.097717089822915e-06,1022 "loss": 0.6546,1023 "step": 14301024 },1025 {1026 "epoch": 0.09216,1027 "grad_norm": 1.3586395978927612,1028 "learning_rate": 6.140388308086196e-06,1029 "loss": 0.6434,1030 "step": 14401031 },1032 {1033 "epoch": 0.0928,1034 "grad_norm": 1.1413036584854126,1035 "learning_rate": 6.183059526349477e-06,1036 "loss": 0.6554,1037 "step": 14501038 },1039 {1040 "epoch": 0.09344,1041 "grad_norm": 1.1844472885131836,1042 "learning_rate": 6.225730744612759e-06,1043 "loss": 0.6357,1044 "step": 14601045 },1046 {1047 "epoch": 0.09408,1048 "grad_norm": 1.7203835248947144,1049 "learning_rate": 6.2684019628760406e-06,1050 "loss": 0.6546,1051 "step": 14701052 },1053 {1054 "epoch": 0.09472,1055 "grad_norm": 1.1869184970855713,1056 "learning_rate": 6.311073181139322e-06,1057 "loss": 0.6376,1058 "step": 14801059 },1060 {1061 "epoch": 0.09536,1062 "grad_norm": 1.3234835863113403,1063 "learning_rate": 6.353744399402604e-06,1064 "loss": 0.6494,1065 "step": 14901066 },1067 {1068 "epoch": 0.096,1069 "grad_norm": 1.1924010515213013,1070 "learning_rate": 6.396415617665885e-06,1071 "loss": 0.652,1072 "step": 15001073 },1074 {1075 "epoch": 0.09664,1076 "grad_norm": 1.4505853652954102,1077 "learning_rate": 6.439086835929166e-06,1078 "loss": 0.6606,1079 "step": 15101080 },1081 {1082 "epoch": 0.09728,1083 "grad_norm": 1.6351200342178345,1084 "learning_rate": 6.481758054192448e-06,1085 "loss": 0.6567,1086 "step": 15201087 },1088 {1089 "epoch": 0.09792,1090 "grad_norm": 1.4423909187316895,1091 "learning_rate": 6.524429272455729e-06,1092 "loss": 0.653,1093 "step": 15301094 },1095 {1096 "epoch": 0.09856,1097 "grad_norm": 0.8749169111251831,1098 "learning_rate": 6.567100490719011e-06,1099 "loss": 0.6417,1100 "step": 15401101 },1102 {1103 "epoch": 0.0992,1104 "grad_norm": 1.2707208395004272,1105 "learning_rate": 6.609771708982292e-06,1106 "loss": 0.6563,1107 "step": 15501108 },1109 {1110 "epoch": 0.09984,1111 "grad_norm": 1.163254976272583,1112 "learning_rate": 6.652442927245574e-06,1113 "loss": 0.6425,1114 "step": 15601115 },1116 {1117 "epoch": 0.10048,1118 "grad_norm": 1.4674128293991089,1119 "learning_rate": 6.6951141455088555e-06,1120 "loss": 0.6561,1121 "step": 15701122 },1123 {1124 "epoch": 0.10112,1125 "grad_norm": 1.6261919736862183,1126 "learning_rate": 6.737785363772135e-06,1127 "loss": 0.6473,1128 "step": 15801129 },1130 {1131 "epoch": 0.10176,1132 "grad_norm": 1.2146090269088745,1133 "learning_rate": 6.780456582035417e-06,1134 "loss": 0.6467,1135 "step": 15901136 },1137 {1138 "epoch": 0.1024,1139 "grad_norm": 1.3596712350845337,1140 "learning_rate": 6.8231278002986985e-06,1141 "loss": 0.6431,1142 "step": 16001143 },1144 {1145 "epoch": 0.10304,1146 "grad_norm": 1.7861741781234741,1147 "learning_rate": 6.86579901856198e-06,1148 "loss": 0.6586,1149 "step": 16101150 },1151 {1152 "epoch": 0.10368,1153 "grad_norm": 1.456587553024292,1154 "learning_rate": 6.908470236825262e-06,1155 "loss": 0.6496,1156 "step": 16201157 },1158 {1159 "epoch": 0.10432,1160 "grad_norm": 1.5652499198913574,1161 "learning_rate": 6.951141455088543e-06,1162 "loss": 0.6436,1163 "step": 16301164 },1165 {1166 "epoch": 0.10496,1167 "grad_norm": 1.1411017179489136,1168 "learning_rate": 6.993812673351825e-06,1169 "loss": 0.6572,1170 "step": 16401171 },1172 {1173 "epoch": 0.1056,1174 "grad_norm": 1.3867056369781494,1175 "learning_rate": 7.036483891615106e-06,1176 "loss": 0.6425,1177 "step": 16501178 },1179 {1180 "epoch": 0.10624,1181 "grad_norm": 1.2714463472366333,1182 "learning_rate": 7.079155109878387e-06,1183 "loss": 0.6558,1184 "step": 16601185 },1186 {1187 "epoch": 0.10688,1188 "grad_norm": 1.32753586769104,1189 "learning_rate": 7.121826328141669e-06,1190 "loss": 0.6537,1191 "step": 16701192 },1193 {1194 "epoch": 0.10752,1195 "grad_norm": 1.5559014081954956,1196 "learning_rate": 7.16449754640495e-06,1197 "loss": 0.6613,1198 "step": 16801199 },1200 {