adpretko/AnghaBench-armv8-O2-native-clang-full-coder-epoch1-AMD
014
1{2 "best_global_step": null,3 "best_metric": null,4 "best_model_checkpoint": null,5 "epoch": 1.0,6 "eval_steps": 500,7 "global_step": 1931,8 "is_hyper_param_search": false,9 "is_local_process_zero": true,10 "is_world_process_zero": true,11 "log_history": [12 {13 "epoch": 0.00518000518000518,14 "grad_norm": 6.292460918426514,15 "learning_rate": 9.278350515463919e-07,16 "loss": 0.5994,17 "step": 1018 },19 {20 "epoch": 0.01036001036001036,21 "grad_norm": 3.469723701477051,22 "learning_rate": 1.9587628865979384e-06,23 "loss": 0.4581,24 "step": 2025 },26 {27 "epoch": 0.01554001554001554,28 "grad_norm": 1.8256299495697021,29 "learning_rate": 2.9896907216494846e-06,30 "loss": 0.254,31 "step": 3032 },33 {34 "epoch": 0.02072002072002072,35 "grad_norm": 0.6593835949897766,36 "learning_rate": 4.020618556701032e-06,37 "loss": 0.1497,38 "step": 4039 },40 {41 "epoch": 0.0259000259000259,42 "grad_norm": 0.4585426449775696,43 "learning_rate": 5.051546391752578e-06,44 "loss": 0.1086,45 "step": 5046 },47 {48 "epoch": 0.03108003108003108,49 "grad_norm": 0.3679659068584442,50 "learning_rate": 6.082474226804124e-06,51 "loss": 0.087,52 "step": 6053 },54 {55 "epoch": 0.03626003626003626,56 "grad_norm": 0.3684946596622467,57 "learning_rate": 7.113402061855671e-06,58 "loss": 0.0775,59 "step": 7060 },61 {62 "epoch": 0.04144004144004144,63 "grad_norm": 0.328337162733078,64 "learning_rate": 8.144329896907216e-06,65 "loss": 0.0686,66 "step": 8067 },68 {69 "epoch": 0.046620046620046623,70 "grad_norm": 1.8907654285430908,71 "learning_rate": 9.175257731958764e-06,72 "loss": 0.0658,73 "step": 9074 },75 {76 "epoch": 0.0518000518000518,77 "grad_norm": 0.5488079786300659,78 "learning_rate": 1.0206185567010309e-05,79 "loss": 0.0631,80 "step": 10081 },82 {83 "epoch": 0.05698005698005698,84 "grad_norm": 0.3149188458919525,85 "learning_rate": 1.1237113402061856e-05,86 "loss": 0.0554,87 "step": 11088 },89 {90 "epoch": 0.06216006216006216,91 "grad_norm": 0.28675776720046997,92 "learning_rate": 1.2268041237113405e-05,93 "loss": 0.0487,94 "step": 12095 },96 {97 "epoch": 0.06734006734006734,98 "grad_norm": 0.20799441635608673,99 "learning_rate": 1.3298969072164948e-05,100 "loss": 0.0462,101 "step": 130102 },103 {104 "epoch": 0.07252007252007252,105 "grad_norm": 0.21834194660186768,106 "learning_rate": 1.4329896907216495e-05,107 "loss": 0.0437,108 "step": 140109 },110 {111 "epoch": 0.0777000777000777,112 "grad_norm": 0.20788688957691193,113 "learning_rate": 1.5360824742268042e-05,114 "loss": 0.0416,115 "step": 150116 },117 {118 "epoch": 0.08288008288008288,119 "grad_norm": 0.19446660578250885,120 "learning_rate": 1.6391752577319588e-05,121 "loss": 0.0395,122 "step": 160123 },124 {125 "epoch": 0.08806008806008805,126 "grad_norm": 0.23992358148097992,127 "learning_rate": 1.7422680412371137e-05,128 "loss": 0.0382,129 "step": 170130 },131 {132 "epoch": 0.09324009324009325,133 "grad_norm": 0.274311363697052,134 "learning_rate": 1.8453608247422682e-05,135 "loss": 0.0374,136 "step": 180137 },138 {139 "epoch": 0.09842009842009843,140 "grad_norm": 0.20221377909183502,141 "learning_rate": 1.9484536082474227e-05,142 "loss": 0.0362,143 "step": 190144 },145 {146 "epoch": 0.1036001036001036,147 "grad_norm": 0.1887311339378357,148 "learning_rate": 1.9999591109366888e-05,149 "loss": 0.0352,150 "step": 200151 },152 {153 "epoch": 0.10878010878010878,154 "grad_norm": 0.16539287567138672,155 "learning_rate": 1.9996320184929093e-05,156 "loss": 0.0345,157 "step": 210158 },159 {160 "epoch": 0.11396011396011396,161 "grad_norm": 0.19868427515029907,162 "learning_rate": 1.9989779405991916e-05,163 "loss": 0.0321,164 "step": 220165 },166 {167 "epoch": 0.11914011914011914,168 "grad_norm": 0.18100829422473907,169 "learning_rate": 1.9979970912082214e-05,170 "loss": 0.0325,171 "step": 230172 },173 {174 "epoch": 0.12432012432012432,175 "grad_norm": 0.16753119230270386,176 "learning_rate": 1.9966897911615417e-05,177 "loss": 0.0321,178 "step": 240179 },180 {181 "epoch": 0.1295001295001295,182 "grad_norm": 0.16504332423210144,183 "learning_rate": 1.9950564680846042e-05,184 "loss": 0.0302,185 "step": 250186 },187 {188 "epoch": 0.13468013468013468,189 "grad_norm": 0.16585540771484375,190 "learning_rate": 1.993097656246892e-05,191 "loss": 0.0299,192 "step": 260193 },194 {195 "epoch": 0.13986013986013987,196 "grad_norm": 0.1773216277360916,197 "learning_rate": 1.9908139963871547e-05,198 "loss": 0.0288,199 "step": 270200 },201 {202 "epoch": 0.14504014504014504,203 "grad_norm": 0.16576074063777924,204 "learning_rate": 1.988206235503821e-05,205 "loss": 0.0292,206 "step": 280207 },208 {209 "epoch": 0.15022015022015023,210 "grad_norm": 0.15523289144039154,211 "learning_rate": 1.98527522661065e-05,212 "loss": 0.0291,213 "step": 290214 },215 {216 "epoch": 0.1554001554001554,217 "grad_norm": 0.16800253093242645,218 "learning_rate": 1.9820219284577052e-05,219 "loss": 0.0284,220 "step": 300221 },222 {223 "epoch": 0.16058016058016059,224 "grad_norm": 0.15856459736824036,225 "learning_rate": 1.9784474052177435e-05,226 "loss": 0.0277,227 "step": 310228 },229 {230 "epoch": 0.16576016576016575,231 "grad_norm": 0.15079988539218903,232 "learning_rate": 1.9745528261381156e-05,233 "loss": 0.0277,234 "step": 320235 },236 {237 "epoch": 0.17094017094017094,238 "grad_norm": 0.18376828730106354,239 "learning_rate": 1.970339465158301e-05,240 "loss": 0.0272,241 "step": 330242 },243 {244 "epoch": 0.1761201761201761,245 "grad_norm": 0.1650630086660385,246 "learning_rate": 1.9658087004931926e-05,247 "loss": 0.0276,248 "step": 340249 },250 {251 "epoch": 0.1813001813001813,252 "grad_norm": 0.14251892268657684,253 "learning_rate": 1.960962014182276e-05,254 "loss": 0.0263,255 "step": 350256 },257 {258 "epoch": 0.1864801864801865,259 "grad_norm": 0.13589179515838623,260 "learning_rate": 1.955800991604846e-05,261 "loss": 0.0267,262 "step": 360263 },264 {265 "epoch": 0.19166019166019166,266 "grad_norm": 0.15240447223186493,267 "learning_rate": 1.9503273209614183e-05,268 "loss": 0.0259,269 "step": 370270 },271 {272 "epoch": 0.19684019684019685,273 "grad_norm": 0.13825932145118713,274 "learning_rate": 1.9445427927215108e-05,275 "loss": 0.0243,276 "step": 380277 },278 {279 "epoch": 0.20202020202020202,280 "grad_norm": 0.13044404983520508,281 "learning_rate": 1.9384492990379703e-05,282 "loss": 0.0254,283 "step": 390284 },285 {286 "epoch": 0.2072002072002072,287 "grad_norm": 0.15161941945552826,288 "learning_rate": 1.9320488331280372e-05,289 "loss": 0.024,290 "step": 400291 },292 {293 "epoch": 0.21238021238021237,294 "grad_norm": 0.12539660930633545,295 "learning_rate": 1.9253434886213548e-05,296 "loss": 0.0247,297 "step": 410298 },299 {300 "epoch": 0.21756021756021757,301 "grad_norm": 0.1758316159248352,302 "learning_rate": 1.9183354588751274e-05,303 "loss": 0.025,304 "step": 420305 },306 {307 "epoch": 0.22274022274022273,308 "grad_norm": 0.1396271139383316,309 "learning_rate": 1.911027036256664e-05,310 "loss": 0.0239,311 "step": 430312 },313 {314 "epoch": 0.22792022792022792,315 "grad_norm": 0.1349797546863556,316 "learning_rate": 1.9034206113935297e-05,317 "loss": 0.0237,318 "step": 440319 },320 {321 "epoch": 0.2331002331002331,322 "grad_norm": 0.13483276963233948,323 "learning_rate": 1.8955186723915573e-05,324 "loss": 0.024,325 "step": 450326 },327 {328 "epoch": 0.23828023828023828,329 "grad_norm": 0.1201694905757904,330 "learning_rate": 1.887323804020975e-05,331 "loss": 0.0226,332 "step": 460333 },334 {335 "epoch": 0.24346024346024345,336 "grad_norm": 0.1224222481250763,337 "learning_rate": 1.878838686870911e-05,338 "loss": 0.0242,339 "step": 470340 },341 {342 "epoch": 0.24864024864024864,343 "grad_norm": 0.11309316009283066,344 "learning_rate": 1.8700660964725583e-05,345 "loss": 0.0221,346 "step": 480347 },348 {349 "epoch": 0.2538202538202538,350 "grad_norm": 0.13578607141971588,351 "learning_rate": 1.8610089023912828e-05,352 "loss": 0.0237,353 "step": 490354 },355 {356 "epoch": 0.259000259000259,357 "grad_norm": 0.13230614364147186,358 "learning_rate": 1.8516700672879706e-05,359 "loss": 0.0224,360 "step": 500361 },362 {363 "epoch": 0.2641802641802642,364 "grad_norm": 0.12713994085788727,365 "learning_rate": 1.8420526459499252e-05,366 "loss": 0.023,367 "step": 510368 },369 {370 "epoch": 0.26936026936026936,371 "grad_norm": 0.1189827024936676,372 "learning_rate": 1.8321597842916282e-05,373 "loss": 0.0219,374 "step": 520375 },376 {377 "epoch": 0.2745402745402745,378 "grad_norm": 0.12820567190647125,379 "learning_rate": 1.821994718325693e-05,380 "loss": 0.0228,381 "step": 530382 },383 {384 "epoch": 0.27972027972027974,385 "grad_norm": 0.11525845527648926,386 "learning_rate": 1.811560773104346e-05,387 "loss": 0.0214,388 "step": 540389 },390 {391 "epoch": 0.2849002849002849,392 "grad_norm": 0.1343483179807663,393 "learning_rate": 1.8008613616317823e-05,394 "loss": 0.0218,395 "step": 550396 },397 {398 "epoch": 0.29008029008029007,399 "grad_norm": 0.11713062226772308,400 "learning_rate": 1.7898999837477528e-05,401 "loss": 0.0216,402 "step": 560403 },404 {405 "epoch": 0.29526029526029524,406 "grad_norm": 0.14427222311496735,407 "learning_rate": 1.7786802249827454e-05,408 "loss": 0.0212,409 "step": 570410 },411 {412 "epoch": 0.30044030044030046,413 "grad_norm": 0.11759313195943832,414 "learning_rate": 1.7672057553851387e-05,415 "loss": 0.0216,416 "step": 580417 },418 {419 "epoch": 0.3056203056203056,420 "grad_norm": 0.11268991976976395,421 "learning_rate": 1.755480328320705e-05,422 "loss": 0.0212,423 "step": 590424 },425 {426 "epoch": 0.3108003108003108,427 "grad_norm": 0.1134837344288826,428 "learning_rate": 1.7435077792448666e-05,429 "loss": 0.0214,430 "step": 600431 },432 {433 "epoch": 0.315980315980316,434 "grad_norm": 0.10657832026481628,435 "learning_rate": 1.731292024448091e-05,436 "loss": 0.0214,437 "step": 610438 },439 {440 "epoch": 0.32116032116032117,441 "grad_norm": 0.11426619440317154,442 "learning_rate": 1.7188370597748553e-05,443 "loss": 0.0211,444 "step": 620445 },446 {447 "epoch": 0.32634032634032634,448 "grad_norm": 0.11127694696187973,449 "learning_rate": 1.706146959316576e-05,450 "loss": 0.0194,451 "step": 630452 },453 {454 "epoch": 0.3315203315203315,455 "grad_norm": 0.11302390694618225,456 "learning_rate": 1.6932258740789553e-05,457 "loss": 0.02,458 "step": 640459 },460 {461 "epoch": 0.3367003367003367,462 "grad_norm": 0.1174977570772171,463 "learning_rate": 1.6800780306241596e-05,464 "loss": 0.0197,465 "step": 650466 },467 {468 "epoch": 0.3418803418803419,469 "grad_norm": 0.12497369199991226,470 "learning_rate": 1.666707729688289e-05,471 "loss": 0.0197,472 "step": 660473 },474 {475 "epoch": 0.34706034706034705,476 "grad_norm": 0.10607603937387466,477 "learning_rate": 1.6531193447745776e-05,478 "loss": 0.0197,479 "step": 670480 },481 {482 "epoch": 0.3522403522403522,483 "grad_norm": 0.1149929016828537,484 "learning_rate": 1.6393173207228e-05,485 "loss": 0.0199,486 "step": 680487 },488 {489 "epoch": 0.35742035742035744,490 "grad_norm": 0.11435063183307648,491 "learning_rate": 1.6253061722553353e-05,492 "loss": 0.0196,493 "step": 690494 },495 {496 "epoch": 0.3626003626003626,497 "grad_norm": 0.11646245419979095,498 "learning_rate": 1.6110904825003754e-05,499 "loss": 0.0199,500 "step": 700501 },502 {503 "epoch": 0.36778036778036777,504 "grad_norm": 0.6431416273117065,505 "learning_rate": 1.596674901492758e-05,506 "loss": 0.0238,507 "step": 710508 },509 {510 "epoch": 0.372960372960373,511 "grad_norm": 6.487484455108643,512 "learning_rate": 1.5820641446529127e-05,513 "loss": 0.0379,514 "step": 720515 },516 {517 "epoch": 0.37814037814037815,518 "grad_norm": 0.19220368564128876,519 "learning_rate": 1.567262991244419e-05,520 "loss": 0.0306,521 "step": 730522 },523 {524 "epoch": 0.3833203833203833,525 "grad_norm": 0.11730585992336273,526 "learning_rate": 1.5522762828106822e-05,527 "loss": 0.022,528 "step": 740529 },530 {531 "epoch": 0.3885003885003885,532 "grad_norm": 0.12174461036920547,533 "learning_rate": 1.5371089215912363e-05,534 "loss": 0.0215,535 "step": 750536 },537 {538 "epoch": 0.3936803936803937,539 "grad_norm": 0.1016201600432396,540 "learning_rate": 1.5217658689181925e-05,541 "loss": 0.0205,542 "step": 760543 },544 {545 "epoch": 0.39886039886039887,546 "grad_norm": 0.09577111154794693,547 "learning_rate": 1.5062521435933586e-05,548 "loss": 0.0198,549 "step": 770550 },551 {552 "epoch": 0.40404040404040403,553 "grad_norm": 0.08802168816328049,554 "learning_rate": 1.4905728202465596e-05,555 "loss": 0.0188,556 "step": 780557 },558 {559 "epoch": 0.4092204092204092,560 "grad_norm": 0.12182024121284485,561 "learning_rate": 1.4747330276756986e-05,562 "loss": 0.0195,563 "step": 790564 },565 {566 "epoch": 0.4144004144004144,567 "grad_norm": 0.09344793111085892,568 "learning_rate": 1.4587379471690937e-05,569 "loss": 0.0202,570 "step": 800571 },572 {573 "epoch": 0.4195804195804196,574 "grad_norm": 0.10629361122846603,575 "learning_rate": 1.4425928108106519e-05,576 "loss": 0.0186,577 "step": 810578 },579 {580 "epoch": 0.42476042476042475,581 "grad_norm": 0.09861776977777481,582 "learning_rate": 1.4263028997684217e-05,583 "loss": 0.0197,584 "step": 820585 },586 {587 "epoch": 0.4299404299404299,588 "grad_norm": 0.10772161930799484,589 "learning_rate": 1.4098735425670931e-05,590 "loss": 0.0193,591 "step": 830592 },593 {594 "epoch": 0.43512043512043513,595 "grad_norm": 0.09898128360509872,596 "learning_rate": 1.393310113345006e-05,597 "loss": 0.0194,598 "step": 840599 },600 {601 "epoch": 0.4403004403004403,602 "grad_norm": 0.09928678721189499,603 "learning_rate": 1.3766180300962393e-05,604 "loss": 0.0183,605 "step": 850606 },607 {608 "epoch": 0.44548044548044546,609 "grad_norm": 0.09649905562400818,610 "learning_rate": 1.3598027528983517e-05,611 "loss": 0.0179,612 "step": 860613 },614 {615 "epoch": 0.4506604506604507,616 "grad_norm": 0.1102428138256073,617 "learning_rate": 1.34286978212636e-05,618 "loss": 0.0193,619 "step": 870620 },621 {622 "epoch": 0.45584045584045585,623 "grad_norm": 0.10131724178791046,624 "learning_rate": 1.325824656653534e-05,625 "loss": 0.0193,626 "step": 880627 },628 {629 "epoch": 0.461020461020461,630 "grad_norm": 0.09785453975200653,631 "learning_rate": 1.308672952039598e-05,632 "loss": 0.0182,633 "step": 890634 },635 {636 "epoch": 0.4662004662004662,637 "grad_norm": 0.11056746542453766,638 "learning_rate": 1.2914202787069345e-05,639 "loss": 0.0183,640 "step": 900641 },642 {643 "epoch": 0.4713804713804714,644 "grad_norm": 0.10242997854948044,645 "learning_rate": 1.2740722801053808e-05,646 "loss": 0.0183,647 "step": 910648 },649 {650 "epoch": 0.47656047656047656,651 "grad_norm": 0.09493660181760788,652 "learning_rate": 1.2566346308662248e-05,653 "loss": 0.0187,654 "step": 920655 },656 {657 "epoch": 0.48174048174048173,658 "grad_norm": 0.09588351845741272,659 "learning_rate": 1.239113034945999e-05,660 "loss": 0.0174,661 "step": 930662 },663 {664 "epoch": 0.4869204869204869,665 "grad_norm": 0.09339374303817749,666 "learning_rate": 1.2215132237606843e-05,667 "loss": 0.0177,668 "step": 940669 },670 {671 "epoch": 0.4921004921004921,672 "grad_norm": 0.09003674238920212,673 "learning_rate": 1.2038409543109295e-05,674 "loss": 0.0176,675 "step": 950676 },677 {678 "epoch": 0.4972804972804973,679 "grad_norm": 0.09645362198352814,680 "learning_rate": 1.186102007298904e-05,681 "loss": 0.0185,682 "step": 960683 },684 {685 "epoch": 0.5024605024605024,686 "grad_norm": 0.09411321580410004,687 "learning_rate": 1.168302185237395e-05,688 "loss": 0.0175,689 "step": 970690 },691 {692 "epoch": 0.5076405076405076,693 "grad_norm": 0.0976066067814827,694 "learning_rate": 1.1504473105517731e-05,695 "loss": 0.017,696 "step": 980697 },698 {699 "epoch": 0.5128205128205128,700 "grad_norm": 0.09469664096832275,701 "learning_rate": 1.1325432236754424e-05,702 "loss": 0.0174,703 "step": 990704 },705 {706 "epoch": 0.518000518000518,707 "grad_norm": 0.09026806801557541,708 "learning_rate": 1.1145957811394006e-05,709 "loss": 0.0174,710 "step": 1000711 },712 {713 "epoch": 0.5231805231805232,714 "grad_norm": 0.09828916192054749,715 "learning_rate": 1.096610853656535e-05,716 "loss": 0.0175,717 "step": 1010718 },719 {720 "epoch": 0.5283605283605284,721 "grad_norm": 0.09940842539072037,722 "learning_rate": 1.0785943242012763e-05,723 "loss": 0.0167,724 "step": 1020725 },726 {727 "epoch": 0.5335405335405335,728 "grad_norm": 0.07944466173648834,729 "learning_rate": 1.0605520860852442e-05,730 "loss": 0.0173,731 "step": 1030732 },733 {734 "epoch": 0.5387205387205387,735 "grad_norm": 0.08528061211109161,736 "learning_rate": 1.0424900410295115e-05,737 "loss": 0.0169,738 "step": 1040739 },740 {741 "epoch": 0.5439005439005439,742 "grad_norm": 0.10138797760009766,743 "learning_rate": 1.0244140972341155e-05,744 "loss": 0.0174,745 "step": 1050746 },747 {748 "epoch": 0.549080549080549,749 "grad_norm": 0.10442786663770676,750 "learning_rate": 1.0063301674454526e-05,751 "loss": 0.0171,752 "step": 1060753 },754 {755 "epoch": 0.5542605542605542,756 "grad_norm": 0.10265690833330154,757 "learning_rate": 9.882441670221846e-06,758 "loss": 0.0162,759 "step": 1070760 },761 {762 "epoch": 0.5594405594405595,763 "grad_norm": 0.10706663131713867,764 "learning_rate": 9.701620120002885e-06,765 "loss": 0.0178,766 "step": 1080767 },768 {769 "epoch": 0.5646205646205646,770 "grad_norm": 0.08483204990625381,771 "learning_rate": 9.520896171578891e-06,772 "loss": 0.0175,773 "step": 1090774 },775 {776 "epoch": 0.5698005698005698,777 "grad_norm": 0.09182880818843842,778 "learning_rate": 9.340328940805003e-06,779 "loss": 0.0174,780 "step": 1100781 },782 {783 "epoch": 0.574980574980575,784 "grad_norm": 0.09123273193836212,785 "learning_rate": 9.159977492273086e-06,786 "loss": 0.0166,787 "step": 1110788 },789 {790 "epoch": 0.5801605801605801,791 "grad_norm": 0.09600038826465607,792 "learning_rate": 8.9799008199914e-06,793 "loss": 0.0166,794 "step": 1120795 },796 {797 "epoch": 0.5853405853405853,798 "grad_norm": 0.07484059780836105,799 "learning_rate": 8.800157828087275e-06,800 "loss": 0.017,801 "step": 1130802 },803 {804 "epoch": 0.5905205905205905,805 "grad_norm": 0.1001143753528595,806 "learning_rate": 8.620807311539258e-06,807 "loss": 0.017,808 "step": 1140809 },810 {811 "epoch": 0.5957005957005957,812 "grad_norm": 0.09151753783226013,813 "learning_rate": 8.441907936944933e-06,814 "loss": 0.0172,815 "step": 1150816 },817 {818 "epoch": 0.6008806008806009,819 "grad_norm": 0.08270428329706192,820 "learning_rate": 8.263518223330698e-06,821 "loss": 0.0164,822 "step": 1160823 },824 {825 "epoch": 0.6060606060606061,826 "grad_norm": 0.07453130185604095,827 "learning_rate": 8.085696523009907e-06,828 "loss": 0.0164,829 "step": 1170830 },831 {832 "epoch": 0.6112406112406112,833 "grad_norm": 0.08895987272262573,834 "learning_rate": 7.908501002495445e-06,835 "loss": 0.0169,836 "step": 1180837 },838 {839 "epoch": 0.6164206164206164,840 "grad_norm": 0.08611361682415009,841 "learning_rate": 7.731989623473144e-06,842 "loss": 0.0155,843 "step": 1190844 },845 {846 "epoch": 0.6216006216006216,847 "grad_norm": 0.09515902400016785,848 "learning_rate": 7.556220123842173e-06,849 "loss": 0.0169,850 "step": 1200851 },852 {853 "epoch": 0.6267806267806267,854 "grad_norm": 0.08863533288240433,855 "learning_rate": 7.38124999882863e-06,856 "loss": 0.0167,857 "step": 1210858 },859 {860 "epoch": 0.631960631960632,861 "grad_norm": 0.0799228847026825,862 "learning_rate": 7.207136482178538e-06,863 "loss": 0.0162,864 "step": 1220865 },866 {867 "epoch": 0.6371406371406372,868 "grad_norm": 0.08676367253065109,869 "learning_rate": 7.033936527436318e-06,870 "loss": 0.017,871 "step": 1230872 },873 {874 "epoch": 0.6423206423206423,875 "grad_norm": 0.08304847776889801,876 "learning_rate": 6.861706789314993e-06,877 "loss": 0.0158,878 "step": 1240879 },880 {881 "epoch": 0.6475006475006475,882 "grad_norm": 0.0753399059176445,883 "learning_rate": 6.6905036051640804e-06,884 "loss": 0.016,885 "step": 1250886 },887 {888 "epoch": 0.6526806526806527,889 "grad_norm": 0.09043081849813461,890 "learning_rate": 6.520382976541313e-06,891 "loss": 0.0159,892 "step": 1260893 },894 {895 "epoch": 0.6578606578606578,896 "grad_norm": 0.08677306771278381,897 "learning_rate": 6.351400550894224e-06,898 "loss": 0.0158,899 "step": 1270900 },901 {902 "epoch": 0.663040663040663,903 "grad_norm": 0.08940693736076355,904 "learning_rate": 6.183611603357513e-06,905 "loss": 0.0159,906 "step": 1280907 },908 {909 "epoch": 0.6682206682206682,910 "grad_norm": 0.0751878097653389,911 "learning_rate": 6.0170710186722605e-06,912 "loss": 0.0161,913 "step": 1290914 },915 {916 "epoch": 0.6734006734006734,917 "grad_norm": 0.0836741104722023,918 "learning_rate": 5.851833273232788e-06,919 "loss": 0.016,920 "step": 1300921 },922 {923 "epoch": 0.6785806785806786,924 "grad_norm": 0.08590537309646606,925 "learning_rate": 5.687952417267115e-06,926 "loss": 0.0157,927 "step": 1310928 },929 {930 "epoch": 0.6837606837606838,931 "grad_norm": 0.09049531072378159,932 "learning_rate": 5.525482057156833e-06,933 "loss": 0.0159,934 "step": 1320935 },936 {937 "epoch": 0.6889406889406889,938 "grad_norm": 0.08685445785522461,939 "learning_rate": 5.364475337902108e-06,940 "loss": 0.0155,941 "step": 1330942 },943 {944 "epoch": 0.6941206941206941,945 "grad_norm": 0.08676782995462418,946 "learning_rate": 5.204984925737689e-06,947 "loss": 0.0166,948 "step": 1340949 },950 {951 "epoch": 0.6993006993006993,952 "grad_norm": 0.09176863729953766,953 "learning_rate": 5.047062990905436e-06,954 "loss": 0.016,955 "step": 1350956 },957 {958 "epoch": 0.7044807044807044,959 "grad_norm": 0.08616431057453156,960 "learning_rate": 4.890761190589157e-06,961 "loss": 0.0156,962 "step": 1360963 },964 {965 "epoch": 0.7096607096607097,966 "grad_norm": 0.07910116016864777,967 "learning_rate": 4.736130652017228e-06,968 "loss": 0.0154,969 "step": 1370970 },971 {972 "epoch": 0.7148407148407149,973 "grad_norm": 0.07988451421260834,974 "learning_rate": 4.5832219557385896e-06,975 "loss": 0.0153,976 "step": 1380977 },978 {979 "epoch": 0.72002072002072,980 "grad_norm": 0.07867439091205597,981 "learning_rate": 4.432085119077536e-06,982 "loss": 0.0153,983 "step": 1390984 },985 {986 "epoch": 0.7252007252007252,987 "grad_norm": 0.08625340461730957,988 "learning_rate": 4.2827695797727835e-06,989 "loss": 0.0153,990 "step": 1400991 },992 {993 "epoch": 0.7303807303807304,994 "grad_norm": 0.07801397144794464,995 "learning_rate": 4.135324179806079e-06,996 "loss": 0.0158,997 "step": 1410998 },999 {1000 "epoch": 0.7355607355607355,1001 "grad_norm": 0.07337366789579391,1002 "learning_rate": 3.989797149425714e-06,1003 "loss": 0.0153,1004 "step": 14201005 },1006 {1007 "epoch": 0.7407407407407407,1008 "grad_norm": 0.07763037085533142,1009 "learning_rate": 3.846236091370119e-06,1010 "loss": 0.0154,1011 "step": 14301012 },1013 {1014 "epoch": 0.745920745920746,1015 "grad_norm": 0.08092360198497772,1016 "learning_rate": 3.704687965296746e-06,1017 "loss": 0.0162,1018 "step": 14401019 },1020 {1021 "epoch": 0.7511007511007511,1022 "grad_norm": 0.08551878482103348,1023 "learning_rate": 3.5651990724212716e-06,1024 "loss": 0.0153,1025 "step": 14501026 },1027 {1028 "epoch": 0.7562807562807563,1029 "grad_norm": 0.07818982750177383,1030 "learning_rate": 3.4278150403722222e-06,1031 "loss": 0.0155,1032 "step": 14601033 },1034 {1035 "epoch": 0.7614607614607615,1036 "grad_norm": 0.08288519084453583,1037 "learning_rate": 3.292580808265897e-06,1038 "loss": 0.0154,1039 "step": 14701040 },1041 {1042 "epoch": 0.7666407666407666,1043 "grad_norm": 0.08305075764656067,1044 "learning_rate": 3.1595406120065174e-06,1045 "loss": 0.0152,1046 "step": 14801047 },1048 {1049 "epoch": 0.7718207718207718,1050 "grad_norm": 0.08365499973297119,1051 "learning_rate": 3.0287379698164245e-06,1052 "loss": 0.0148,1053 "step": 14901054 },1055 {1056 "epoch": 0.777000777000777,1057 "grad_norm": 0.07930707186460495,1058 "learning_rate": 2.900215668000991e-06,1059 "loss": 0.0161,1060 "step": 15001061 },1062 {1063 "epoch": 0.7821807821807821,1064 "grad_norm": 0.07458413392305374,1065 "learning_rate": 2.7740157469529915e-06,1066 "loss": 0.0159,1067 "step": 15101068 },1069 {1070 "epoch": 0.7873607873607874,1071 "grad_norm": 0.08544889092445374,1072 "learning_rate": 2.6501794874009425e-06,1073 "loss": 0.0153,1074 "step": 15201075 },1076 {1077 "epoch": 0.7925407925407926,1078 "grad_norm": 0.0796743705868721,1079 "learning_rate": 2.5287473969059174e-06,1080 "loss": 0.0156,1081 "step": 15301082 },1083 {1084 "epoch": 0.7977207977207977,1085 "grad_norm": 0.089390829205513,1086 "learning_rate": 2.4097591966113155e-06,1087 "loss": 0.0149,1088 "step": 15401089 },1090 {1091 "epoch": 0.8029008029008029,1092 "grad_norm": 0.0870826244354248,1093 "learning_rate": 2.2932538082498225e-06,1094 "loss": 0.0156,1095 "step": 15501096 },1097 {1098 "epoch": 0.8080808080808081,1099 "grad_norm": 0.07120909541845322,1100 "learning_rate": 2.179269341411896e-06,1101 "loss": 0.0152,1102 "step": 15601103 },1104 {1105 "epoch": 0.8132608132608132,1106 "grad_norm": 0.07515786588191986,1107 "learning_rate": 2.0678430810799e-06,1108 "loss": 0.0141,1109 "step": 15701110 },1111 {1112 "epoch": 0.8184408184408184,1113 "grad_norm": 0.07680921256542206,1114 "learning_rate": 1.959011475431952e-06,1115 "loss": 0.0155,1116 "step": 15801117 },1118 {1119 "epoch": 0.8236208236208237,1120 "grad_norm": 0.08255264908075333,1121 "learning_rate": 1.8528101239195394e-06,1122 "loss": 0.0153,1123 "step": 15901124 },1125 {1126 "epoch": 0.8288008288008288,1127 "grad_norm": 0.08244433999061584,1128 "learning_rate": 1.7492737656227032e-06,1129 "loss": 0.0152,1130 "step": 16001131 },1132 {1133 "epoch": 0.833980833980834,1134 "grad_norm": 0.06824195384979248,1135 "learning_rate": 1.6484362678867083e-06,1136 "loss": 0.015,1137 "step": 16101138 },1139 {1140 "epoch": 0.8391608391608392,1141 "grad_norm": 0.07020522654056549,1142 "learning_rate": 1.5503306152438146e-06,1143 "loss": 0.015,1144 "step": 16201145 },1146 {1147 "epoch": 0.8443408443408443,1148 "grad_norm": 0.08058454096317291,1149 "learning_rate": 1.4549888986238658e-06,1150 "loss": 0.0154,1151 "step": 16301152 },1153 {1154 "epoch": 0.8495208495208495,1155 "grad_norm": 0.0823742225766182,1156 "learning_rate": 1.3624423048571434e-06,1157 "loss": 0.0146,1158 "step": 16401159 },1160 {1161 "epoch": 0.8547008547008547,1162 "grad_norm": 0.07870230078697205,1163 "learning_rate": 1.2727211064729862e-06,1164 "loss": 0.0149,1165 "step": 16501166 },1167 {1168 "epoch": 0.8598808598808598,1169 "grad_norm": 0.06677371263504028,1170 "learning_rate": 1.1858546517974511e-06,1171 "loss": 0.014,1172 "step": 16601173 },1174 {1175 "epoch": 0.8650608650608651,1176 "grad_norm": 0.08251772075891495,1177 "learning_rate": 1.1018713553533279e-06,1178 "loss": 0.015,1179 "step": 16701180 },1181 {1182 "epoch": 0.8702408702408703,1183 "grad_norm": 0.07941140979528427,1184 "learning_rate": 1.0207986885655664e-06,1185 "loss": 0.0151,1186 "step": 16801187 },1188 {1189 "epoch": 0.8754208754208754,1190 "grad_norm": 0.07803778350353241,1191 "learning_rate": 9.426631707752243e-07,1192 "loss": 0.015,1193 "step": 16901194 },1195 {1196 "epoch": 0.8806008806008806,1197 "grad_norm": 0.07801060378551483,1198 "learning_rate": 8.674903605648221e-07,1199 "loss": 0.0155,1200 "step": 1700