adpretko/train-riscv-O2_epoch3_AMD
011
1{2 "best_global_step": null,3 "best_metric": null,4 "best_model_checkpoint": null,5 "epoch": 2.0,6 "eval_steps": 500,7 "global_step": 3886,8 "is_hyper_param_search": false,9 "is_local_process_zero": true,10 "is_world_process_zero": true,11 "log_history": [12 {13 "epoch": 0.0051473426843392095,14 "grad_norm": 0.05525032430887222,15 "learning_rate": 4.6272493573264783e-07,16 "loss": 0.0106,17 "step": 1018 },19 {20 "epoch": 0.010294685368678419,21 "grad_norm": 0.05362959951162338,22 "learning_rate": 9.768637532133676e-07,23 "loss": 0.0106,24 "step": 2025 },26 {27 "epoch": 0.01544202805301763,28 "grad_norm": 0.05603431165218353,29 "learning_rate": 1.4910025706940876e-06,30 "loss": 0.0105,31 "step": 3032 },33 {34 "epoch": 0.020589370737356838,35 "grad_norm": 0.0636371374130249,36 "learning_rate": 2.0051413881748076e-06,37 "loss": 0.0113,38 "step": 4039 },40 {41 "epoch": 0.02573671342169605,42 "grad_norm": 0.06462942063808441,43 "learning_rate": 2.519280205655527e-06,44 "loss": 0.0102,45 "step": 5046 },47 {48 "epoch": 0.03088405610603526,49 "grad_norm": 0.0556134358048439,50 "learning_rate": 3.033419023136247e-06,51 "loss": 0.0111,52 "step": 6053 },54 {55 "epoch": 0.036031398790374466,56 "grad_norm": 0.057751912623643875,57 "learning_rate": 3.547557840616967e-06,58 "loss": 0.0109,59 "step": 7060 },61 {62 "epoch": 0.041178741474713676,63 "grad_norm": 0.059322893619537354,64 "learning_rate": 4.0616966580976866e-06,65 "loss": 0.0112,66 "step": 8067 },68 {69 "epoch": 0.04632608415905289,70 "grad_norm": 0.05676131322979927,71 "learning_rate": 4.575835475578407e-06,72 "loss": 0.0113,73 "step": 9074 },75 {76 "epoch": 0.0514734268433921,77 "grad_norm": 0.06342212855815887,78 "learning_rate": 5.0899742930591265e-06,79 "loss": 0.0115,80 "step": 10081 },82 {83 "epoch": 0.05662076952773131,84 "grad_norm": 0.05883905664086342,85 "learning_rate": 5.604113110539846e-06,86 "loss": 0.011,87 "step": 11088 },89 {90 "epoch": 0.06176811221207052,91 "grad_norm": 0.06119426712393761,92 "learning_rate": 6.1182519280205664e-06,93 "loss": 0.0112,94 "step": 12095 },96 {97 "epoch": 0.06691545489640972,98 "grad_norm": 0.06628178805112839,99 "learning_rate": 6.632390745501286e-06,100 "loss": 0.0114,101 "step": 130102 },103 {104 "epoch": 0.07206279758074893,105 "grad_norm": 0.06540685147047043,106 "learning_rate": 7.1465295629820055e-06,107 "loss": 0.0116,108 "step": 140109 },110 {111 "epoch": 0.07721014026508814,112 "grad_norm": 0.06296146661043167,113 "learning_rate": 7.660668380462726e-06,114 "loss": 0.0112,115 "step": 150116 },117 {118 "epoch": 0.08235748294942735,119 "grad_norm": 0.06895332783460617,120 "learning_rate": 8.174807197943445e-06,121 "loss": 0.0115,122 "step": 160123 },124 {125 "epoch": 0.08750482563376656,126 "grad_norm": 0.08407478034496307,127 "learning_rate": 8.688946015424165e-06,128 "loss": 0.0126,129 "step": 170130 },131 {132 "epoch": 0.09265216831810578,133 "grad_norm": 0.06299558281898499,134 "learning_rate": 9.203084832904885e-06,135 "loss": 0.0116,136 "step": 180137 },138 {139 "epoch": 0.097799511002445,140 "grad_norm": 0.06793331354856491,141 "learning_rate": 9.717223650385606e-06,142 "loss": 0.0115,143 "step": 190144 },145 {146 "epoch": 0.1029468536867842,147 "grad_norm": 0.07241130620241165,148 "learning_rate": 1.0231362467866324e-05,149 "loss": 0.0114,150 "step": 200151 },152 {153 "epoch": 0.10809419637112341,154 "grad_norm": 0.0763164609670639,155 "learning_rate": 1.0745501285347045e-05,156 "loss": 0.0119,157 "step": 210158 },159 {160 "epoch": 0.11324153905546262,161 "grad_norm": 0.06727232038974762,162 "learning_rate": 1.1259640102827764e-05,163 "loss": 0.0116,164 "step": 220165 },166 {167 "epoch": 0.11838888173980183,168 "grad_norm": 0.09049773216247559,169 "learning_rate": 1.1773778920308486e-05,170 "loss": 0.0122,171 "step": 230172 },173 {174 "epoch": 0.12353622442414104,175 "grad_norm": 0.07317323982715607,176 "learning_rate": 1.2287917737789203e-05,177 "loss": 0.012,178 "step": 240179 },180 {181 "epoch": 0.12868356710848025,182 "grad_norm": 0.07453487813472748,183 "learning_rate": 1.2802056555269925e-05,184 "loss": 0.0118,185 "step": 250186 },187 {188 "epoch": 0.13383090979281945,189 "grad_norm": 0.06471797078847885,190 "learning_rate": 1.3316195372750644e-05,191 "loss": 0.0114,192 "step": 260193 },194 {195 "epoch": 0.13897825247715867,196 "grad_norm": 0.08005302399396896,197 "learning_rate": 1.3830334190231362e-05,198 "loss": 0.012,199 "step": 270200 },201 {202 "epoch": 0.14412559516149787,203 "grad_norm": 0.07167572528123856,204 "learning_rate": 1.4344473007712083e-05,205 "loss": 0.0119,206 "step": 280207 },208 {209 "epoch": 0.1492729378458371,210 "grad_norm": 0.08343254774808884,211 "learning_rate": 1.4858611825192803e-05,212 "loss": 0.0126,213 "step": 290214 },215 {216 "epoch": 0.15442028053017628,217 "grad_norm": 0.07320424169301987,218 "learning_rate": 1.5372750642673522e-05,219 "loss": 0.0123,220 "step": 300221 },222 {223 "epoch": 0.1595676232145155,224 "grad_norm": 0.07763416320085526,225 "learning_rate": 1.5886889460154242e-05,226 "loss": 0.0122,227 "step": 310228 },229 {230 "epoch": 0.1647149658988547,231 "grad_norm": 0.07546375691890717,232 "learning_rate": 1.640102827763496e-05,233 "loss": 0.0128,234 "step": 320235 },236 {237 "epoch": 0.16986230858319393,238 "grad_norm": 0.07282799482345581,239 "learning_rate": 1.6915167095115684e-05,240 "loss": 0.0129,241 "step": 330242 },243 {244 "epoch": 0.17500965126753312,245 "grad_norm": 0.07360665500164032,246 "learning_rate": 1.74293059125964e-05,247 "loss": 0.0123,248 "step": 340249 },250 {251 "epoch": 0.18015699395187235,252 "grad_norm": 0.07590780407190323,253 "learning_rate": 1.7943444730077123e-05,254 "loss": 0.0128,255 "step": 350256 },257 {258 "epoch": 0.18530433663621157,259 "grad_norm": 0.06358295679092407,260 "learning_rate": 1.8457583547557843e-05,261 "loss": 0.0124,262 "step": 360263 },264 {265 "epoch": 0.19045167932055077,266 "grad_norm": 0.10541003942489624,267 "learning_rate": 1.8971722365038563e-05,268 "loss": 0.013,269 "step": 370270 },271 {272 "epoch": 0.19559902200489,273 "grad_norm": 0.06901050359010696,274 "learning_rate": 1.9485861182519282e-05,275 "loss": 0.0129,276 "step": 380277 },278 {279 "epoch": 0.20074636468922918,280 "grad_norm": 0.0681707039475441,281 "learning_rate": 2e-05,282 "loss": 0.0131,283 "step": 390284 },285 {286 "epoch": 0.2058937073735684,287 "grad_norm": 0.07603519409894943,288 "learning_rate": 1.999959647024453e-05,289 "loss": 0.0135,290 "step": 400291 },292 {293 "epoch": 0.2110410500579076,294 "grad_norm": 0.06607502698898315,295 "learning_rate": 1.999838591354537e-05,296 "loss": 0.0125,297 "step": 410298 },299 {300 "epoch": 0.21618839274224683,301 "grad_norm": 0.08145863562822342,302 "learning_rate": 1.9996368427601652e-05,303 "loss": 0.0129,304 "step": 420305 },306 {307 "epoch": 0.22133573542658602,308 "grad_norm": 0.06165160983800888,309 "learning_rate": 1.9993544175236498e-05,310 "loss": 0.0124,311 "step": 430312 },313 {314 "epoch": 0.22648307811092525,315 "grad_norm": 0.07120924443006516,316 "learning_rate": 1.998991338438388e-05,317 "loss": 0.013,318 "step": 440319 },320 {321 "epoch": 0.23163042079526444,322 "grad_norm": 0.0669078603386879,323 "learning_rate": 1.9985476348070224e-05,324 "loss": 0.0121,325 "step": 450326 },327 {328 "epoch": 0.23677776347960366,329 "grad_norm": 3.4496610164642334,330 "learning_rate": 1.9980233424390773e-05,331 "loss": 0.021,332 "step": 460333 },334 {335 "epoch": 0.24192510616394286,336 "grad_norm": 0.11881168931722641,337 "learning_rate": 1.9974185036480664e-05,338 "loss": 0.0277,339 "step": 470340 },341 {342 "epoch": 0.24707244884828208,343 "grad_norm": 0.068561851978302,344 "learning_rate": 1.9967331672480798e-05,345 "loss": 0.0164,346 "step": 480347 },348 {349 "epoch": 0.2522197915326213,350 "grad_norm": 0.06290870159864426,351 "learning_rate": 1.995967388549843e-05,352 "loss": 0.0138,353 "step": 490354 },355 {356 "epoch": 0.2573671342169605,357 "grad_norm": 0.06344795227050781,358 "learning_rate": 1.9951212293562547e-05,359 "loss": 0.0137,360 "step": 500361 },362 {363 "epoch": 0.2625144769012997,364 "grad_norm": 0.06535464525222778,365 "learning_rate": 1.994194757957397e-05,366 "loss": 0.013,367 "step": 510368 },369 {370 "epoch": 0.2676618195856389,371 "grad_norm": 0.06390328705310822,372 "learning_rate": 1.9931880491250263e-05,373 "loss": 0.0124,374 "step": 520375 },376 {377 "epoch": 0.27280916226997814,378 "grad_norm": 0.06637264043092728,379 "learning_rate": 1.992101184106535e-05,380 "loss": 0.0125,381 "step": 530382 },383 {384 "epoch": 0.27795650495431734,385 "grad_norm": 0.06743142008781433,386 "learning_rate": 1.990934250618399e-05,387 "loss": 0.0132,388 "step": 540389 },390 {391 "epoch": 0.28310384763865654,392 "grad_norm": 0.053368110209703445,393 "learning_rate": 1.989687342839095e-05,394 "loss": 0.0117,395 "step": 550396 },397 {398 "epoch": 0.28825119032299573,399 "grad_norm": 0.058148909360170364,400 "learning_rate": 1.9883605614015014e-05,401 "loss": 0.0121,402 "step": 560403 },404 {405 "epoch": 0.293398533007335,406 "grad_norm": 0.060990240424871445,407 "learning_rate": 1.986954013384776e-05,408 "loss": 0.0116,409 "step": 570410 },411 {412 "epoch": 0.2985458756916742,413 "grad_norm": 0.06001332029700279,414 "learning_rate": 1.9854678123057144e-05,415 "loss": 0.0122,416 "step": 580417 },418 {419 "epoch": 0.3036932183760134,420 "grad_norm": 0.06884918361902237,421 "learning_rate": 1.9839020781095873e-05,422 "loss": 0.0116,423 "step": 590424 },425 {426 "epoch": 0.30884056106035257,427 "grad_norm": 0.058082133531570435,428 "learning_rate": 1.9822569371604637e-05,429 "loss": 0.012,430 "step": 600431 },432 {433 "epoch": 0.3139879037446918,434 "grad_norm": 0.053921736776828766,435 "learning_rate": 1.9805325222310072e-05,436 "loss": 0.0108,437 "step": 610438 },439 {440 "epoch": 0.319135246429031,441 "grad_norm": 0.059762243181467056,442 "learning_rate": 1.9787289724917657e-05,443 "loss": 0.0117,444 "step": 620445 },446 {447 "epoch": 0.3242825891133702,448 "grad_norm": 0.052163220942020416,449 "learning_rate": 1.9768464334999352e-05,450 "loss": 0.0108,451 "step": 630452 },453 {454 "epoch": 0.3294299317977094,455 "grad_norm": 0.054961204528808594,456 "learning_rate": 1.974885057187617e-05,457 "loss": 0.0114,458 "step": 640459 },460 {461 "epoch": 0.33457727448204866,462 "grad_norm": 0.06342387944459915,463 "learning_rate": 1.9728450018495506e-05,464 "loss": 0.0116,465 "step": 650466 },467 {468 "epoch": 0.33972461716638785,469 "grad_norm": 0.05155491828918457,470 "learning_rate": 1.9707264321303427e-05,471 "loss": 0.0115,472 "step": 660473 },474 {475 "epoch": 0.34487195985072705,476 "grad_norm": 0.05698103830218315,477 "learning_rate": 1.9685295190111777e-05,478 "loss": 0.0113,479 "step": 670480 },481 {482 "epoch": 0.35001930253506625,483 "grad_norm": 0.06619592010974884,484 "learning_rate": 1.9662544397960182e-05,485 "loss": 0.0113,486 "step": 680487 },488 {489 "epoch": 0.3551666452194055,490 "grad_norm": 0.055313605815172195,491 "learning_rate": 1.9639013780972958e-05,492 "loss": 0.0113,493 "step": 690494 },495 {496 "epoch": 0.3603139879037447,497 "grad_norm": 0.06306542456150055,498 "learning_rate": 1.961470523821093e-05,499 "loss": 0.0115,500 "step": 700501 },502 {503 "epoch": 0.3654613305880839,504 "grad_norm": 0.05840103700757027,505 "learning_rate": 1.9589620731518166e-05,506 "loss": 0.0111,507 "step": 710508 },509 {510 "epoch": 0.37060867327242314,511 "grad_norm": 0.05483512952923775,512 "learning_rate": 1.956376228536363e-05,513 "loss": 0.0109,514 "step": 720515 },516 {517 "epoch": 0.37575601595676233,518 "grad_norm": 0.05598577484488487,519 "learning_rate": 1.953713198667782e-05,520 "loss": 0.0116,521 "step": 730522 },523 {524 "epoch": 0.38090335864110153,525 "grad_norm": 0.05302765220403671,526 "learning_rate": 1.950973198468431e-05,527 "loss": 0.0109,528 "step": 740529 },530 {531 "epoch": 0.3860507013254407,532 "grad_norm": 0.05226908251643181,533 "learning_rate": 1.9481564490726327e-05,534 "loss": 0.0116,535 "step": 750536 },537 {538 "epoch": 0.39119804400978,539 "grad_norm": 0.06026271730661392,540 "learning_rate": 1.9452631778088262e-05,541 "loss": 0.0109,542 "step": 760543 },544 {545 "epoch": 0.3963453866941192,546 "grad_norm": 0.058383017778396606,547 "learning_rate": 1.94229361818122e-05,548 "loss": 0.011,549 "step": 770550 },551 {552 "epoch": 0.40149272937845837,553 "grad_norm": 0.05417947843670845,554 "learning_rate": 1.9392480098509488e-05,555 "loss": 0.0111,556 "step": 780557 },558 {559 "epoch": 0.40664007206279756,560 "grad_norm": 0.06370815634727478,561 "learning_rate": 1.9361265986167292e-05,562 "loss": 0.0112,563 "step": 790564 },565 {566 "epoch": 0.4117874147471368,567 "grad_norm": 0.05550752207636833,568 "learning_rate": 1.9329296363950237e-05,569 "loss": 0.0111,570 "step": 800571 },572 {573 "epoch": 0.416934757431476,574 "grad_norm": 0.054070789366960526,575 "learning_rate": 1.929657381199709e-05,576 "loss": 0.0113,577 "step": 810578 },579 {580 "epoch": 0.4220821001158152,581 "grad_norm": 0.06153961271047592,582 "learning_rate": 1.9263100971212533e-05,583 "loss": 0.0115,584 "step": 820585 },586 {587 "epoch": 0.4272294428001544,588 "grad_norm": 0.05265296623110771,589 "learning_rate": 1.922888054305401e-05,590 "loss": 0.0114,591 "step": 830592 },593 {594 "epoch": 0.43237678548449365,595 "grad_norm": 0.051895588636398315,596 "learning_rate": 1.9193915289313726e-05,597 "loss": 0.011,598 "step": 840599 },600 {601 "epoch": 0.43752412816883285,602 "grad_norm": 0.05822490155696869,603 "learning_rate": 1.9158208031895738e-05,604 "loss": 0.0108,605 "step": 850606 },607 {608 "epoch": 0.44267147085317204,609 "grad_norm": 0.05402348190546036,610 "learning_rate": 1.9121761652588217e-05,611 "loss": 0.011,612 "step": 860613 },614 {615 "epoch": 0.44781881353751124,616 "grad_norm": 0.05292493477463722,617 "learning_rate": 1.9084579092830873e-05,618 "loss": 0.0105,619 "step": 870620 },621 {622 "epoch": 0.4529661562218505,623 "grad_norm": 0.052194658666849136,624 "learning_rate": 1.904666335347755e-05,625 "loss": 0.0097,626 "step": 880627 },628 {629 "epoch": 0.4581134989061897,630 "grad_norm": 0.06219863146543503,631 "learning_rate": 1.900801749455406e-05,632 "loss": 0.0104,633 "step": 890634 },635 {636 "epoch": 0.4632608415905289,637 "grad_norm": 0.0552273727953434,638 "learning_rate": 1.8968644635011192e-05,639 "loss": 0.0108,640 "step": 900641 },642 {643 "epoch": 0.4684081842748681,644 "grad_norm": 0.05805448442697525,645 "learning_rate": 1.8928547952473037e-05,646 "loss": 0.0109,647 "step": 910648 },649 {650 "epoch": 0.47355552695920733,651 "grad_norm": 0.05885449796915054,652 "learning_rate": 1.8887730682980484e-05,653 "loss": 0.0101,654 "step": 920655 },656 {657 "epoch": 0.4787028696435465,658 "grad_norm": 0.06097600609064102,659 "learning_rate": 1.8846196120730095e-05,660 "loss": 0.011,661 "step": 930662 },663 {664 "epoch": 0.4838502123278857,665 "grad_norm": 0.057253919541835785,666 "learning_rate": 1.8803947617808217e-05,667 "loss": 0.0113,668 "step": 940669 },670 {671 "epoch": 0.4889975550122249,672 "grad_norm": 0.05612657591700554,673 "learning_rate": 1.8760988583920457e-05,674 "loss": 0.0105,675 "step": 950676 },677 {678 "epoch": 0.49414489769656417,679 "grad_norm": 0.05309102311730385,680 "learning_rate": 1.8717322486116513e-05,681 "loss": 0.0103,682 "step": 960683 },684 {685 "epoch": 0.49929224038090336,686 "grad_norm": 0.06416105479001999,687 "learning_rate": 1.867295284851033e-05,688 "loss": 0.0103,689 "step": 970690 },691 {692 "epoch": 0.5044395830652426,693 "grad_norm": 0.060254644602537155,694 "learning_rate": 1.8627883251995712e-05,695 "loss": 0.0108,696 "step": 980697 },698 {699 "epoch": 0.5095869257495818,700 "grad_norm": 0.057798177003860474,701 "learning_rate": 1.858211733395731e-05,702 "loss": 0.0098,703 "step": 990704 },705 {706 "epoch": 0.514734268433921,707 "grad_norm": 0.05069541186094284,708 "learning_rate": 1.8535658787977076e-05,709 "loss": 0.0108,710 "step": 1000711 },712 {713 "epoch": 0.5198816111182601,714 "grad_norm": 0.05629874765872955,715 "learning_rate": 1.848851136353614e-05,716 "loss": 0.0101,717 "step": 1010718 },719 {720 "epoch": 0.5250289538025994,721 "grad_norm": 0.05504068732261658,722 "learning_rate": 1.8440678865712237e-05,723 "loss": 0.0098,724 "step": 1020725 },726 {727 "epoch": 0.5301762964869386,728 "grad_norm": 0.053771983832120895,729 "learning_rate": 1.8392165154872596e-05,730 "loss": 0.0101,731 "step": 1030732 },733 {734 "epoch": 0.5353236391712778,735 "grad_norm": 0.058356136083602905,736 "learning_rate": 1.8342974146362397e-05,737 "loss": 0.0102,738 "step": 1040739 },740 {741 "epoch": 0.540470981855617,742 "grad_norm": 0.05251409858465195,743 "learning_rate": 1.8293109810188756e-05,744 "loss": 0.0105,745 "step": 1050746 },747 {748 "epoch": 0.5456183245399563,749 "grad_norm": 0.05582083389163017,750 "learning_rate": 1.8242576170700356e-05,751 "loss": 0.0108,752 "step": 1060753 },754 {755 "epoch": 0.5507656672242954,756 "grad_norm": 0.054003242403268814,757 "learning_rate": 1.8191377306262635e-05,758 "loss": 0.0104,759 "step": 1070760 },761 {762 "epoch": 0.5559130099086347,763 "grad_norm": 0.05103457346558571,764 "learning_rate": 1.813951734892864e-05,765 "loss": 0.0102,766 "step": 1080767 },768 {769 "epoch": 0.5610603525929739,770 "grad_norm": 0.04666399583220482,771 "learning_rate": 1.808700048410555e-05,772 "loss": 0.0104,773 "step": 1090774 },775 {776 "epoch": 0.5662076952773131,777 "grad_norm": 0.05002720654010773,778 "learning_rate": 1.80338309502169e-05,779 "loss": 0.0103,780 "step": 1100781 },782 {783 "epoch": 0.5713550379616523,784 "grad_norm": 0.0568714365363121,785 "learning_rate": 1.798001303836048e-05,786 "loss": 0.0107,787 "step": 1110788 },789 {790 "epoch": 0.5765023806459915,791 "grad_norm": 0.051760654896497726,792 "learning_rate": 1.792555109196205e-05,793 "loss": 0.0109,794 "step": 1120795 },796 {797 "epoch": 0.5816497233303307,798 "grad_norm": 0.05566855147480965,799 "learning_rate": 1.7870449506424802e-05,800 "loss": 0.0102,801 "step": 1130802 },803 {804 "epoch": 0.58679706601467,805 "grad_norm": 0.058662049472332,806 "learning_rate": 1.7814712728774602e-05,807 "loss": 0.0098,808 "step": 1140809 },810 {811 "epoch": 0.5919444086990091,812 "grad_norm": 0.04951569065451622,813 "learning_rate": 1.7758345257301097e-05,814 "loss": 0.01,815 "step": 1150816 },817 {818 "epoch": 0.5970917513833484,819 "grad_norm": 0.05501884967088699,820 "learning_rate": 1.770135164119468e-05,821 "loss": 0.0106,822 "step": 1160823 },824 {825 "epoch": 0.6022390940676876,826 "grad_norm": 0.04806562513113022,827 "learning_rate": 1.7643736480179353e-05,828 "loss": 0.0099,829 "step": 1170830 },831 {832 "epoch": 0.6073864367520267,833 "grad_norm": 0.056878462433815,834 "learning_rate": 1.7585504424141483e-05,835 "loss": 0.0101,836 "step": 1180837 },838 {839 "epoch": 0.612533779436366,840 "grad_norm": 0.05829343572258949,841 "learning_rate": 1.752666017275453e-05,842 "loss": 0.0104,843 "step": 1190844 },845 {846 "epoch": 0.6176811221207051,847 "grad_norm": 0.054884593933820724,848 "learning_rate": 1.7467208475099777e-05,849 "loss": 0.0102,850 "step": 1200851 },852 {853 "epoch": 0.6228284648050444,854 "grad_norm": 0.07677657157182693,855 "learning_rate": 1.740715412928302e-05,856 "loss": 0.0101,857 "step": 1210858 },859 {860 "epoch": 0.6279758074893836,861 "grad_norm": 0.058270618319511414,862 "learning_rate": 1.734650198204736e-05,863 "loss": 0.0099,864 "step": 1220865 },866 {867 "epoch": 0.6331231501737228,868 "grad_norm": 0.04892059788107872,869 "learning_rate": 1.7285256928382024e-05,870 "loss": 0.0098,871 "step": 1230872 },873 {874 "epoch": 0.638270492858062,875 "grad_norm": 0.05252835527062416,876 "learning_rate": 1.7223423911127315e-05,877 "loss": 0.0112,878 "step": 1240879 },880 {881 "epoch": 0.6434178355424013,882 "grad_norm": 0.05630745738744736,883 "learning_rate": 1.7161007920575705e-05,884 "loss": 0.0098,885 "step": 1250886 },887 {888 "epoch": 0.6485651782267404,889 "grad_norm": 0.051849983632564545,890 "learning_rate": 1.709801399406907e-05,891 "loss": 0.0097,892 "step": 1260893 },894 {895 "epoch": 0.6537125209110797,896 "grad_norm": 0.05386548116803169,897 "learning_rate": 1.7034447215592168e-05,898 "loss": 0.0097,899 "step": 1270900 },901 {902 "epoch": 0.6588598635954188,903 "grad_norm": 0.051140256226062775,904 "learning_rate": 1.6970312715362304e-05,905 "loss": 0.0101,906 "step": 1280907 },908 {909 "epoch": 0.6640072062797581,910 "grad_norm": 0.047512467950582504,911 "learning_rate": 1.6905615669415326e-05,912 "loss": 0.0105,913 "step": 1290914 },915 {916 "epoch": 0.6691545489640973,917 "grad_norm": 0.051020972430706024,918 "learning_rate": 1.684036129918786e-05,919 "loss": 0.0098,920 "step": 1300921 },922 {923 "epoch": 0.6743018916484365,924 "grad_norm": 0.04566548392176628,925 "learning_rate": 1.6774554871095918e-05,926 "loss": 0.0095,927 "step": 1310928 },929 {930 "epoch": 0.6794492343327757,931 "grad_norm": 0.05308493971824646,932 "learning_rate": 1.6708201696109857e-05,933 "loss": 0.0102,934 "step": 1320935 },936 {937 "epoch": 0.684596577017115,938 "grad_norm": 0.046334341168403625,939 "learning_rate": 1.6641307129325785e-05,940 "loss": 0.0105,941 "step": 1330942 },943 {944 "epoch": 0.6897439197014541,945 "grad_norm": 0.059736642986536026,946 "learning_rate": 1.657387656953333e-05,947 "loss": 0.0103,948 "step": 1340949 },950 {951 "epoch": 0.6948912623857934,952 "grad_norm": 0.055522654205560684,953 "learning_rate": 1.6505915458779958e-05,954 "loss": 0.01,955 "step": 1350956 },957 {958 "epoch": 0.7000386050701325,959 "grad_norm": 0.05038890987634659,960 "learning_rate": 1.6437429281931744e-05,961 "loss": 0.0094,962 "step": 1360963 },964 {965 "epoch": 0.7051859477544717,966 "grad_norm": 0.05290698632597923,967 "learning_rate": 1.636842356623073e-05,968 "loss": 0.0098,969 "step": 1370970 },971 {972 "epoch": 0.710333290438811,973 "grad_norm": 0.053712453693151474,974 "learning_rate": 1.6298903880848836e-05,975 "loss": 0.0102,976 "step": 1380977 },978 {979 "epoch": 0.7154806331231501,980 "grad_norm": 0.04889325425028801,981 "learning_rate": 1.6228875836438386e-05,982 "loss": 0.0099,983 "step": 1390984 },985 {986 "epoch": 0.7206279758074894,987 "grad_norm": 0.05055889114737511,988 "learning_rate": 1.615834508467931e-05,989 "loss": 0.0095,990 "step": 1400991 },992 {993 "epoch": 0.7257753184918286,994 "grad_norm": 0.05285302549600601,995 "learning_rate": 1.608731731782301e-05,996 "loss": 0.0101,997 "step": 1410998 },999 {1000 "epoch": 0.7309226611761678,1001 "grad_norm": 0.05686409771442413,1002 "learning_rate": 1.6015798268232964e-05,1003 "loss": 0.0102,1004 "step": 14201005 },1006 {1007 "epoch": 0.736070003860507,1008 "grad_norm": 0.04935956746339798,1009 "learning_rate": 1.5943793707922086e-05,1010 "loss": 0.0093,1011 "step": 14301012 },1013 {1014 "epoch": 0.7412173465448463,1015 "grad_norm": 0.050160281360149384,1016 "learning_rate": 1.5871309448086903e-05,1017 "loss": 0.0104,1018 "step": 14401019 },1020 {1021 "epoch": 0.7463646892291854,1022 "grad_norm": 0.05499512329697609,1023 "learning_rate": 1.5798351338638548e-05,1024 "loss": 0.0097,1025 "step": 14501026 },1027 {1028 "epoch": 0.7515120319135247,1029 "grad_norm": 0.05228967219591141,1030 "learning_rate": 1.5724925267730625e-05,1031 "loss": 0.0103,1032 "step": 14601033 },1034 {1035 "epoch": 0.7566593745978638,1036 "grad_norm": 0.0571601577103138,1037 "learning_rate": 1.5651037161284032e-05,1038 "loss": 0.0102,1039 "step": 14701040 },1041 {1042 "epoch": 0.7618067172822031,1043 "grad_norm": 0.05126621946692467,1044 "learning_rate": 1.5576692982508665e-05,1045 "loss": 0.0098,1046 "step": 14801047 },1048 {1049 "epoch": 0.7669540599665423,1050 "grad_norm": 0.050559673458337784,1051 "learning_rate": 1.550189873142219e-05,1052 "loss": 0.0099,1053 "step": 14901054 },1055 {1056 "epoch": 0.7721014026508815,1057 "grad_norm": 0.047047846019268036,1058 "learning_rate": 1.542666044436577e-05,1059 "loss": 0.0097,1060 "step": 15001061 },1062 {1063 "epoch": 0.7772487453352207,1064 "grad_norm": 0.051877353340387344,1065 "learning_rate": 1.5350984193516917e-05,1066 "loss": 0.01,1067 "step": 15101068 },1069 {1070 "epoch": 0.78239608801956,1071 "grad_norm": 0.05931961163878441,1072 "learning_rate": 1.5274876086399444e-05,1073 "loss": 0.0096,1074 "step": 15201075 },1076 {1077 "epoch": 0.7875434307038991,1078 "grad_norm": 0.04609573632478714,1079 "learning_rate": 1.5198342265390511e-05,1080 "loss": 0.0095,1081 "step": 15301082 },1083 {1084 "epoch": 0.7926907733882383,1085 "grad_norm": 0.05874916538596153,1086 "learning_rate": 1.512138890722494e-05,1087 "loss": 0.0099,1088 "step": 15401089 },1090 {1091 "epoch": 0.7978381160725775,1092 "grad_norm": 0.05315364524722099,1093 "learning_rate": 1.5044022222496684e-05,1094 "loss": 0.01,1095 "step": 15501096 },1097 {1098 "epoch": 0.8029854587569167,1099 "grad_norm": 0.04722389206290245,1100 "learning_rate": 1.4966248455157622e-05,1101 "loss": 0.0096,1102 "step": 15601103 },1104 {1105 "epoch": 0.808132801441256,1106 "grad_norm": 0.0479586161673069,1107 "learning_rate": 1.4888073882013616e-05,1108 "loss": 0.0094,1109 "step": 15701110 },1111 {1112 "epoch": 0.8132801441255951,1113 "grad_norm": 0.051096055656671524,1114 "learning_rate": 1.480950481221795e-05,1115 "loss": 0.0096,1116 "step": 15801117 },1118 {1119 "epoch": 0.8184274868099344,1120 "grad_norm": 0.0522165410220623,1121 "learning_rate": 1.4730547586762114e-05,1122 "loss": 0.0093,1123 "step": 15901124 },1125 {1126 "epoch": 0.8235748294942736,1127 "grad_norm": 0.05063074454665184,1128 "learning_rate": 1.4651208577964099e-05,1129 "loss": 0.0101,1130 "step": 16001131 },1132 {1133 "epoch": 0.8287221721786128,1134 "grad_norm": 0.060858119279146194,1135 "learning_rate": 1.4571494188954058e-05,1136 "loss": 0.01,1137 "step": 16101138 },1139 {1140 "epoch": 0.833869514862952,1141 "grad_norm": 0.040989868342876434,1142 "learning_rate": 1.4491410853157577e-05,1143 "loss": 0.0095,1144 "step": 16201145 },1146 {1147 "epoch": 0.8390168575472912,1148 "grad_norm": 0.04905058071017265,1149 "learning_rate": 1.4410965033776439e-05,1150 "loss": 0.0099,1151 "step": 16301152 },1153 {1154 "epoch": 0.8441642002316304,1155 "grad_norm": 0.04507426545023918,1156 "learning_rate": 1.4330163223267005e-05,1157 "loss": 0.01,1158 "step": 16401159 },1160 {1161 "epoch": 0.8493115429159697,1162 "grad_norm": 0.048070088028907776,1163 "learning_rate": 1.4249011942816245e-05,1164 "loss": 0.0099,1165 "step": 16501166 },1167 {1168 "epoch": 0.8544588856003088,1169 "grad_norm": 0.04860364645719528,1170 "learning_rate": 1.416751774181543e-05,1171 "loss": 0.0097,1172 "step": 16601173 },1174 {1175 "epoch": 0.859606228284648,1176 "grad_norm": 0.047335829585790634,1177 "learning_rate": 1.4085687197331559e-05,1178 "loss": 0.0097,1179 "step": 16701180 },1181 {1182 "epoch": 0.8647535709689873,1183 "grad_norm": 0.04517504945397377,1184 "learning_rate": 1.4003526913576555e-05,1185 "loss": 0.0103,1186 "step": 16801187 },1188 {1189 "epoch": 0.8699009136533264,1190 "grad_norm": 0.04708757624030113,1191 "learning_rate": 1.392104352137426e-05,1192 "loss": 0.0091,1193 "step": 16901194 },1195 {1196 "epoch": 0.8750482563376657,1197 "grad_norm": 0.04715834558010101,1198 "learning_rate": 1.3838243677625292e-05,1199 "loss": 0.0099,1200 "step": 1700