Balajip30/deberta-hydrangea-leafspot
05
1{2 "best_global_step": 13206,3 "best_metric": 0.9868791709426586,4 "best_model_checkpoint": "/home/myid/bp67339/plant_disease/models/deberta_v3_base_textclf_phase3/checkpoint-13206",5 "epoch": 3.0,6 "eval_steps": 500,7 "global_step": 13206,8 "is_hyper_param_search": false,9 "is_local_process_zero": true,10 "is_world_process_zero": true,11 "log_history": [12 {13 "epoch": 0.011358473421172195,14 "grad_norm": 2.552668809890747,15 "learning_rate": 1.9925791306981678e-05,16 "loss": 0.1196,17 "step": 5018 },19 {20 "epoch": 0.02271694684234439,21 "grad_norm": 0.47541937232017517,22 "learning_rate": 1.985006815084053e-05,23 "loss": 0.0812,24 "step": 10025 },26 {27 "epoch": 0.034075420263516586,28 "grad_norm": 7.045262813568115,29 "learning_rate": 1.977434499469938e-05,30 "loss": 0.0859,31 "step": 15032 },33 {34 "epoch": 0.04543389368468878,35 "grad_norm": 0.03415582329034805,36 "learning_rate": 1.9698621838558232e-05,37 "loss": 0.0847,38 "step": 20039 },40 {41 "epoch": 0.05679236710586097,42 "grad_norm": 0.05752035602927208,43 "learning_rate": 1.9622898682417084e-05,44 "loss": 0.0988,45 "step": 25046 },47 {48 "epoch": 0.06815084052703317,49 "grad_norm": 3.0614030361175537,50 "learning_rate": 1.9547175526275936e-05,51 "loss": 0.1058,52 "step": 30053 },54 {55 "epoch": 0.07950931394820536,56 "grad_norm": 1.2528300285339355,57 "learning_rate": 1.947145237013479e-05,58 "loss": 0.0842,59 "step": 35060 },61 {62 "epoch": 0.09086778736937756,63 "grad_norm": 8.556694984436035,64 "learning_rate": 1.9395729213993642e-05,65 "loss": 0.0777,66 "step": 40067 },68 {69 "epoch": 0.10222626079054975,70 "grad_norm": 20.690099716186523,71 "learning_rate": 1.9320006057852494e-05,72 "loss": 0.0764,73 "step": 45074 },75 {76 "epoch": 0.11358473421172194,77 "grad_norm": 4.0012030601501465,78 "learning_rate": 1.9244282901711345e-05,79 "loss": 0.0814,80 "step": 50081 },82 {83 "epoch": 0.12494320763289414,84 "grad_norm": 0.06608668714761734,85 "learning_rate": 1.9168559745570197e-05,86 "loss": 0.0637,87 "step": 55088 },89 {90 "epoch": 0.13630168105406634,91 "grad_norm": 5.306321620941162,92 "learning_rate": 1.909283658942905e-05,93 "loss": 0.0479,94 "step": 60095 },96 {97 "epoch": 0.14766015447523853,98 "grad_norm": 5.900157451629639,99 "learning_rate": 1.90171134332879e-05,100 "loss": 0.0892,101 "step": 650102 },103 {104 "epoch": 0.15901862789641072,105 "grad_norm": 2.73388934135437,106 "learning_rate": 1.8941390277146755e-05,107 "loss": 0.0721,108 "step": 700109 },110 {111 "epoch": 0.1703771013175829,112 "grad_norm": 6.770120143890381,113 "learning_rate": 1.8865667121005603e-05,114 "loss": 0.082,115 "step": 750116 },117 {118 "epoch": 0.18173557473875512,119 "grad_norm": 0.010230629704892635,120 "learning_rate": 1.8789943964864458e-05,121 "loss": 0.0604,122 "step": 800123 },124 {125 "epoch": 0.1930940481599273,126 "grad_norm": 11.530449867248535,127 "learning_rate": 1.871422080872331e-05,128 "loss": 0.0578,129 "step": 850130 },131 {132 "epoch": 0.2044525215810995,133 "grad_norm": 0.014014088548719883,134 "learning_rate": 1.863849765258216e-05,135 "loss": 0.0574,136 "step": 900137 },138 {139 "epoch": 0.2158109950022717,140 "grad_norm": 0.016662120819091797,141 "learning_rate": 1.8562774496441013e-05,142 "loss": 0.0654,143 "step": 950144 },145 {146 "epoch": 0.22716946842344388,147 "grad_norm": 1.2837393283843994,148 "learning_rate": 1.8487051340299865e-05,149 "loss": 0.0638,150 "step": 1000151 },152 {153 "epoch": 0.2385279418446161,154 "grad_norm": 0.45574066042900085,155 "learning_rate": 1.8411328184158716e-05,156 "loss": 0.0448,157 "step": 1050158 },159 {160 "epoch": 0.24988641526578828,161 "grad_norm": 2.423413038253784,162 "learning_rate": 1.8335605028017568e-05,163 "loss": 0.0535,164 "step": 1100165 },166 {167 "epoch": 0.2612448886869605,168 "grad_norm": 4.537928104400635,169 "learning_rate": 1.8259881871876423e-05,170 "loss": 0.0833,171 "step": 1150172 },173 {174 "epoch": 0.2726033621081327,175 "grad_norm": 6.994387626647949,176 "learning_rate": 1.8184158715735274e-05,177 "loss": 0.06,178 "step": 1200179 },180 {181 "epoch": 0.2839618355293049,182 "grad_norm": 0.01928258314728737,183 "learning_rate": 1.8108435559594126e-05,184 "loss": 0.0498,185 "step": 1250186 },187 {188 "epoch": 0.29532030895047706,189 "grad_norm": 0.4064951241016388,190 "learning_rate": 1.8032712403452977e-05,191 "loss": 0.0491,192 "step": 1300193 },194 {195 "epoch": 0.30667878237164925,196 "grad_norm": 2.6388704776763916,197 "learning_rate": 1.795698924731183e-05,198 "loss": 0.0327,199 "step": 1350200 },201 {202 "epoch": 0.31803725579282144,203 "grad_norm": 2.0852885246276855,204 "learning_rate": 1.788126609117068e-05,205 "loss": 0.0978,206 "step": 1400207 },208 {209 "epoch": 0.3293957292139936,210 "grad_norm": 2.8278560638427734,211 "learning_rate": 1.7805542935029532e-05,212 "loss": 0.0521,213 "step": 1450214 },215 {216 "epoch": 0.3407542026351658,217 "grad_norm": 4.4425177574157715,218 "learning_rate": 1.7729819778888387e-05,219 "loss": 0.0532,220 "step": 1500221 },222 {223 "epoch": 0.352112676056338,224 "grad_norm": 0.0768510028719902,225 "learning_rate": 1.7654096622747235e-05,226 "loss": 0.0486,227 "step": 1550228 },229 {230 "epoch": 0.36347114947751025,231 "grad_norm": 0.4614756405353546,232 "learning_rate": 1.757837346660609e-05,233 "loss": 0.0419,234 "step": 1600235 },236 {237 "epoch": 0.37482962289868244,238 "grad_norm": 0.5945758819580078,239 "learning_rate": 1.7502650310464942e-05,240 "loss": 0.055,241 "step": 1650242 },243 {244 "epoch": 0.3861880963198546,245 "grad_norm": 0.5370413661003113,246 "learning_rate": 1.7426927154323794e-05,247 "loss": 0.0434,248 "step": 1700249 },250 {251 "epoch": 0.3975465697410268,252 "grad_norm": 13.6658296585083,253 "learning_rate": 1.7351203998182645e-05,254 "loss": 0.0329,255 "step": 1750256 },257 {258 "epoch": 0.408905043162199,259 "grad_norm": 0.9185870885848999,260 "learning_rate": 1.72754808420415e-05,261 "loss": 0.0584,262 "step": 1800263 },264 {265 "epoch": 0.4202635165833712,266 "grad_norm": 0.3048367500305176,267 "learning_rate": 1.719975768590035e-05,268 "loss": 0.0561,269 "step": 1850270 },271 {272 "epoch": 0.4316219900045434,273 "grad_norm": 3.6233348846435547,274 "learning_rate": 1.71240345297592e-05,275 "loss": 0.0568,276 "step": 1900277 },278 {279 "epoch": 0.44298046342571556,280 "grad_norm": 7.301445007324219,281 "learning_rate": 1.7048311373618055e-05,282 "loss": 0.0545,283 "step": 1950284 },285 {286 "epoch": 0.45433893684688775,287 "grad_norm": 12.261190414428711,288 "learning_rate": 1.6972588217476907e-05,289 "loss": 0.0347,290 "step": 2000291 },292 {293 "epoch": 0.46569741026806,294 "grad_norm": 0.004870145116001368,295 "learning_rate": 1.6896865061335758e-05,296 "loss": 0.0297,297 "step": 2050298 },299 {300 "epoch": 0.4770558836892322,301 "grad_norm": 0.08898526430130005,302 "learning_rate": 1.682114190519461e-05,303 "loss": 0.0368,304 "step": 2100305 },306 {307 "epoch": 0.4884143571104044,308 "grad_norm": 0.007771740201860666,309 "learning_rate": 1.674541874905346e-05,310 "loss": 0.037,311 "step": 2150312 },313 {314 "epoch": 0.49977283053157656,315 "grad_norm": 0.06338020414113998,316 "learning_rate": 1.6669695592912313e-05,317 "loss": 0.0351,318 "step": 2200319 },320 {321 "epoch": 0.5111313039527488,322 "grad_norm": 0.2080918550491333,323 "learning_rate": 1.6593972436771168e-05,324 "loss": 0.0424,325 "step": 2250326 },327 {328 "epoch": 0.522489777373921,329 "grad_norm": 0.005484211724251509,330 "learning_rate": 1.651824928063002e-05,331 "loss": 0.0498,332 "step": 2300333 },334 {335 "epoch": 0.5338482507950931,336 "grad_norm": 1.5300865173339844,337 "learning_rate": 1.6442526124488868e-05,338 "loss": 0.0309,339 "step": 2350340 },341 {342 "epoch": 0.5452067242162654,343 "grad_norm": 1.4512425661087036,344 "learning_rate": 1.6366802968347723e-05,345 "loss": 0.0397,346 "step": 2400347 },348 {349 "epoch": 0.5565651976374375,350 "grad_norm": 0.07551686465740204,351 "learning_rate": 1.6291079812206574e-05,352 "loss": 0.0368,353 "step": 2450354 },355 {356 "epoch": 0.5679236710586097,357 "grad_norm": 5.5924272537231445,358 "learning_rate": 1.6215356656065426e-05,359 "loss": 0.0169,360 "step": 2500361 },362 {363 "epoch": 0.5792821444797819,364 "grad_norm": 0.03456870838999748,365 "learning_rate": 1.6139633499924277e-05,366 "loss": 0.0397,367 "step": 2550368 },369 {370 "epoch": 0.5906406179009541,371 "grad_norm": 1.605539321899414,372 "learning_rate": 1.6063910343783132e-05,373 "loss": 0.0353,374 "step": 2600375 },376 {377 "epoch": 0.6019990913221263,378 "grad_norm": 2.287083864212036,379 "learning_rate": 1.598818718764198e-05,380 "loss": 0.0233,381 "step": 2650382 },383 {384 "epoch": 0.6133575647432985,385 "grad_norm": 0.02498025633394718,386 "learning_rate": 1.5912464031500836e-05,387 "loss": 0.0389,388 "step": 2700389 },390 {391 "epoch": 0.6247160381644707,392 "grad_norm": 0.009853340685367584,393 "learning_rate": 1.5836740875359687e-05,394 "loss": 0.0343,395 "step": 2750396 },397 {398 "epoch": 0.6360745115856429,399 "grad_norm": 0.006630271207541227,400 "learning_rate": 1.576101771921854e-05,401 "loss": 0.0211,402 "step": 2800403 },404 {405 "epoch": 0.6474329850068151,406 "grad_norm": 9.345346450805664,407 "learning_rate": 1.568529456307739e-05,408 "loss": 0.0374,409 "step": 2850410 },411 {412 "epoch": 0.6587914584279873,413 "grad_norm": 6.142024040222168,414 "learning_rate": 1.5609571406936242e-05,415 "loss": 0.0551,416 "step": 2900417 },418 {419 "epoch": 0.6701499318491595,420 "grad_norm": 0.0021689136046916246,421 "learning_rate": 1.5533848250795093e-05,422 "loss": 0.0223,423 "step": 2950424 },425 {426 "epoch": 0.6815084052703316,427 "grad_norm": 0.12782888114452362,428 "learning_rate": 1.5458125094653945e-05,429 "loss": 0.0397,430 "step": 3000431 },432 {433 "epoch": 0.6928668786915039,434 "grad_norm": 0.021150466054677963,435 "learning_rate": 1.53824019385128e-05,436 "loss": 0.0465,437 "step": 3050438 },439 {440 "epoch": 0.704225352112676,441 "grad_norm": 0.03622226044535637,442 "learning_rate": 1.530667878237165e-05,443 "loss": 0.0245,444 "step": 3100445 },446 {447 "epoch": 0.7155838255338483,448 "grad_norm": 0.01021323911845684,449 "learning_rate": 1.5230955626230503e-05,450 "loss": 0.04,451 "step": 3150452 },453 {454 "epoch": 0.7269422989550205,455 "grad_norm": 0.3106459677219391,456 "learning_rate": 1.5155232470089353e-05,457 "loss": 0.0257,458 "step": 3200459 },460 {461 "epoch": 0.7383007723761926,462 "grad_norm": 0.00275537115521729,463 "learning_rate": 1.5079509313948206e-05,464 "loss": 0.0263,465 "step": 3250466 },467 {468 "epoch": 0.7496592457973649,469 "grad_norm": 2.062044143676758,470 "learning_rate": 1.5003786157807058e-05,471 "loss": 0.0309,472 "step": 3300473 },474 {475 "epoch": 0.761017719218537,476 "grad_norm": 9.71252155303955,477 "learning_rate": 1.4928063001665911e-05,478 "loss": 0.0398,479 "step": 3350480 },481 {482 "epoch": 0.7723761926397092,483 "grad_norm": 0.1981002688407898,484 "learning_rate": 1.4852339845524763e-05,485 "loss": 0.0361,486 "step": 3400487 },488 {489 "epoch": 0.7837346660608814,490 "grad_norm": 5.37298059463501,491 "learning_rate": 1.4776616689383614e-05,492 "loss": 0.028,493 "step": 3450494 },495 {496 "epoch": 0.7950931394820536,497 "grad_norm": 0.01882290095090866,498 "learning_rate": 1.4700893533242466e-05,499 "loss": 0.0326,500 "step": 3500501 },502 {503 "epoch": 0.8064516129032258,504 "grad_norm": 0.035887233912944794,505 "learning_rate": 1.462517037710132e-05,506 "loss": 0.0416,507 "step": 3550508 },509 {510 "epoch": 0.817810086324398,511 "grad_norm": 13.781176567077637,512 "learning_rate": 1.4549447220960171e-05,513 "loss": 0.023,514 "step": 3600515 },516 {517 "epoch": 0.8291685597455702,518 "grad_norm": 0.06990953534841537,519 "learning_rate": 1.4473724064819024e-05,520 "loss": 0.0213,521 "step": 3650522 },523 {524 "epoch": 0.8405270331667424,525 "grad_norm": 0.6063719987869263,526 "learning_rate": 1.4398000908677874e-05,527 "loss": 0.0228,528 "step": 3700529 },530 {531 "epoch": 0.8518855065879146,532 "grad_norm": 4.930353164672852,533 "learning_rate": 1.4322277752536726e-05,534 "loss": 0.0335,535 "step": 3750536 },537 {538 "epoch": 0.8632439800090868,539 "grad_norm": 0.004565690644085407,540 "learning_rate": 1.4246554596395579e-05,541 "loss": 0.0214,542 "step": 3800543 },544 {545 "epoch": 0.874602453430259,546 "grad_norm": 0.17241248488426208,547 "learning_rate": 1.417083144025443e-05,548 "loss": 0.0401,549 "step": 3850550 },551 {552 "epoch": 0.8859609268514311,553 "grad_norm": 5.231073379516602,554 "learning_rate": 1.4095108284113284e-05,555 "loss": 0.0168,556 "step": 3900557 },558 {559 "epoch": 0.8973194002726034,560 "grad_norm": 0.25602054595947266,561 "learning_rate": 1.4019385127972135e-05,562 "loss": 0.0399,563 "step": 3950564 },565 {566 "epoch": 0.9086778736937755,567 "grad_norm": 0.04758666455745697,568 "learning_rate": 1.3943661971830987e-05,569 "loss": 0.0232,570 "step": 4000571 },572 {573 "epoch": 0.9200363471149478,574 "grad_norm": 0.007415900472551584,575 "learning_rate": 1.3867938815689839e-05,576 "loss": 0.0254,577 "step": 4050578 },579 {580 "epoch": 0.93139482053612,581 "grad_norm": 0.05941289663314819,582 "learning_rate": 1.3792215659548692e-05,583 "loss": 0.0182,584 "step": 4100585 },586 {587 "epoch": 0.9427532939572921,588 "grad_norm": 0.2857076823711395,589 "learning_rate": 1.3716492503407543e-05,590 "loss": 0.015,591 "step": 4150592 },593 {594 "epoch": 0.9541117673784644,595 "grad_norm": 6.7106852531433105,596 "learning_rate": 1.3640769347266397e-05,597 "loss": 0.0218,598 "step": 4200599 },600 {601 "epoch": 0.9654702407996365,602 "grad_norm": 2.165870189666748,603 "learning_rate": 1.3565046191125247e-05,604 "loss": 0.0477,605 "step": 4250606 },607 {608 "epoch": 0.9768287142208087,609 "grad_norm": 1.2232965230941772,610 "learning_rate": 1.3489323034984098e-05,611 "loss": 0.0227,612 "step": 4300613 },614 {615 "epoch": 0.9881871876419809,616 "grad_norm": 6.756147384643555,617 "learning_rate": 1.3413599878842951e-05,618 "loss": 0.0273,619 "step": 4350620 },621 {622 "epoch": 0.9995456610631531,623 "grad_norm": 0.07053057104349136,624 "learning_rate": 1.3337876722701803e-05,625 "loss": 0.0217,626 "step": 4400627 },628 {629 "epoch": 1.0,630 "eval_accuracy": 0.9839739413680781,631 "eval_f1_macro": 0.982934850082554,632 "eval_loss": 0.08169186860322952,633 "eval_runtime": 25.3611,634 "eval_samples_per_second": 302.629,635 "eval_steps_per_second": 4.732,636 "step": 4402637 },638 {639 "epoch": 1.0109041344843253,640 "grad_norm": 0.005929551087319851,641 "learning_rate": 1.3262153566560656e-05,642 "loss": 0.0101,643 "step": 4450644 },645 {646 "epoch": 1.0222626079054975,647 "grad_norm": 0.011584239080548286,648 "learning_rate": 1.3186430410419508e-05,649 "loss": 0.012,650 "step": 4500651 },652 {653 "epoch": 1.0336210813266697,654 "grad_norm": 0.06345133483409882,655 "learning_rate": 1.3110707254278358e-05,656 "loss": 0.0295,657 "step": 4550658 },659 {660 "epoch": 1.044979554747842,661 "grad_norm": 0.003701810259371996,662 "learning_rate": 1.3034984098137211e-05,663 "loss": 0.0096,664 "step": 4600665 },666 {667 "epoch": 1.056338028169014,668 "grad_norm": 0.0013922312064096332,669 "learning_rate": 1.2959260941996063e-05,670 "loss": 0.0267,671 "step": 4650672 },673 {674 "epoch": 1.0676965015901863,675 "grad_norm": 0.0026676456909626722,676 "learning_rate": 1.2883537785854916e-05,677 "loss": 0.0145,678 "step": 4700679 },680 {681 "epoch": 1.0790549750113585,682 "grad_norm": 0.0031162630766630173,683 "learning_rate": 1.2807814629713768e-05,684 "loss": 0.0124,685 "step": 4750686 },687 {688 "epoch": 1.0904134484325307,689 "grad_norm": 3.748222589492798,690 "learning_rate": 1.273209147357262e-05,691 "loss": 0.0074,692 "step": 4800693 },694 {695 "epoch": 1.1017719218537028,696 "grad_norm": 0.0014477827353402972,697 "learning_rate": 1.265636831743147e-05,698 "loss": 0.0179,699 "step": 4850700 },701 {702 "epoch": 1.113130395274875,703 "grad_norm": 4.411365985870361,704 "learning_rate": 1.2580645161290324e-05,705 "loss": 0.0181,706 "step": 4900707 },708 {709 "epoch": 1.1244888686960473,710 "grad_norm": 0.05877981334924698,711 "learning_rate": 1.2504922005149176e-05,712 "loss": 0.013,713 "step": 4950714 },715 {716 "epoch": 1.1358473421172195,717 "grad_norm": 0.999143123626709,718 "learning_rate": 1.2429198849008029e-05,719 "loss": 0.0114,720 "step": 5000721 },722 {723 "epoch": 1.1472058155383915,724 "grad_norm": 0.0030792481265962124,725 "learning_rate": 1.2353475692866879e-05,726 "loss": 0.0114,727 "step": 5050728 },729 {730 "epoch": 1.1585642889595638,731 "grad_norm": 0.0014463861007243395,732 "learning_rate": 1.227775253672573e-05,733 "loss": 0.0188,734 "step": 5100735 },736 {737 "epoch": 1.169922762380736,738 "grad_norm": 0.0017361458158120513,739 "learning_rate": 1.2202029380584584e-05,740 "loss": 0.0234,741 "step": 5150742 },743 {744 "epoch": 1.1812812358019082,745 "grad_norm": 13.90465259552002,746 "learning_rate": 1.2126306224443435e-05,747 "loss": 0.0078,748 "step": 5200749 },750 {751 "epoch": 1.1926397092230805,752 "grad_norm": 5.497283935546875,753 "learning_rate": 1.2050583068302289e-05,754 "loss": 0.0124,755 "step": 5250756 },757 {758 "epoch": 1.2039981826442525,759 "grad_norm": 3.7859766483306885,760 "learning_rate": 1.197485991216114e-05,761 "loss": 0.0319,762 "step": 5300763 },764 {765 "epoch": 1.2153566560654248,766 "grad_norm": 0.0024443278089165688,767 "learning_rate": 1.1899136756019992e-05,768 "loss": 0.0177,769 "step": 5350770 },771 {772 "epoch": 1.226715129486597,773 "grad_norm": 0.20258040726184845,774 "learning_rate": 1.1823413599878843e-05,775 "loss": 0.009,776 "step": 5400777 },778 {779 "epoch": 1.2380736029077692,780 "grad_norm": 0.0014519239775836468,781 "learning_rate": 1.1747690443737697e-05,782 "loss": 0.0069,783 "step": 5450784 },785 {786 "epoch": 1.2494320763289415,787 "grad_norm": 0.02450985088944435,788 "learning_rate": 1.1671967287596548e-05,789 "loss": 0.017,790 "step": 5500791 },792 {793 "epoch": 1.2607905497501135,794 "grad_norm": 0.004507457371801138,795 "learning_rate": 1.1596244131455401e-05,796 "loss": 0.01,797 "step": 5550798 },799 {800 "epoch": 1.2721490231712858,801 "grad_norm": 9.089266777038574,802 "learning_rate": 1.1520520975314251e-05,803 "loss": 0.0382,804 "step": 5600805 },806 {807 "epoch": 1.283507496592458,808 "grad_norm": 5.648477077484131,809 "learning_rate": 1.1444797819173103e-05,810 "loss": 0.0165,811 "step": 5650812 },813 {814 "epoch": 1.2948659700136302,815 "grad_norm": 1.2645410299301147,816 "learning_rate": 1.1369074663031956e-05,817 "loss": 0.0163,818 "step": 5700819 },820 {821 "epoch": 1.3062244434348025,822 "grad_norm": 0.0017288923263549805,823 "learning_rate": 1.1293351506890808e-05,824 "loss": 0.0202,825 "step": 5750826 },827 {828 "epoch": 1.3175829168559745,829 "grad_norm": 0.00359937222674489,830 "learning_rate": 1.1217628350749661e-05,831 "loss": 0.0135,832 "step": 5800833 },834 {835 "epoch": 1.3289413902771468,836 "grad_norm": 4.165374279022217,837 "learning_rate": 1.1141905194608513e-05,838 "loss": 0.0132,839 "step": 5850840 },841 {842 "epoch": 1.340299863698319,843 "grad_norm": 0.0018008551560342312,844 "learning_rate": 1.1066182038467364e-05,845 "loss": 0.0112,846 "step": 5900847 },848 {849 "epoch": 1.351658337119491,850 "grad_norm": 0.5573896765708923,851 "learning_rate": 1.0990458882326216e-05,852 "loss": 0.0221,853 "step": 5950854 },855 {856 "epoch": 1.3630168105406633,857 "grad_norm": 1.456154704093933,858 "learning_rate": 1.091473572618507e-05,859 "loss": 0.011,860 "step": 6000861 },862 {863 "epoch": 1.3743752839618355,864 "grad_norm": 3.8196120262145996,865 "learning_rate": 1.083901257004392e-05,866 "loss": 0.0158,867 "step": 6050868 },869 {870 "epoch": 1.3857337573830077,871 "grad_norm": 0.0017874937038868666,872 "learning_rate": 1.0763289413902774e-05,873 "loss": 0.0107,874 "step": 6100875 },876 {877 "epoch": 1.39709223080418,878 "grad_norm": 0.0008583781309425831,879 "learning_rate": 1.0687566257761624e-05,880 "loss": 0.0031,881 "step": 6150882 },883 {884 "epoch": 1.408450704225352,885 "grad_norm": 0.44672343134880066,886 "learning_rate": 1.0611843101620476e-05,887 "loss": 0.0223,888 "step": 6200889 },890 {891 "epoch": 1.4198091776465243,892 "grad_norm": 0.009662467055022717,893 "learning_rate": 1.0536119945479329e-05,894 "loss": 0.0157,895 "step": 6250896 },897 {898 "epoch": 1.4311676510676965,899 "grad_norm": 0.0007439041510224342,900 "learning_rate": 1.046039678933818e-05,901 "loss": 0.0165,902 "step": 6300903 },904 {905 "epoch": 1.4425261244888687,906 "grad_norm": 0.001364586642012,907 "learning_rate": 1.0384673633197034e-05,908 "loss": 0.0054,909 "step": 6350910 },911 {912 "epoch": 1.453884597910041,913 "grad_norm": 0.001398891326971352,914 "learning_rate": 1.0308950477055884e-05,915 "loss": 0.0087,916 "step": 6400917 },918 {919 "epoch": 1.465243071331213,920 "grad_norm": 0.002500066999346018,921 "learning_rate": 1.0233227320914735e-05,922 "loss": 0.0117,923 "step": 6450924 },925 {926 "epoch": 1.4766015447523853,927 "grad_norm": 0.0015706096310168505,928 "learning_rate": 1.0157504164773588e-05,929 "loss": 0.0173,930 "step": 6500931 },932 {933 "epoch": 1.4879600181735575,934 "grad_norm": 0.0015295586781576276,935 "learning_rate": 1.008178100863244e-05,936 "loss": 0.0284,937 "step": 6550938 },939 {940 "epoch": 1.4993184915947297,941 "grad_norm": 2.286747932434082,942 "learning_rate": 1.0006057852491293e-05,943 "loss": 0.011,944 "step": 6600945 },946 {947 "epoch": 1.510676965015902,948 "grad_norm": 0.002311094431206584,949 "learning_rate": 9.930334696350145e-06,950 "loss": 0.0166,951 "step": 6650952 },953 {954 "epoch": 1.522035438437074,955 "grad_norm": 0.0014053047634661198,956 "learning_rate": 9.854611540208996e-06,957 "loss": 0.0174,958 "step": 6700959 },960 {961 "epoch": 1.5333939118582463,962 "grad_norm": 0.012581984512507915,963 "learning_rate": 9.778888384067848e-06,964 "loss": 0.0099,965 "step": 6750966 },967 {968 "epoch": 1.5447523852794185,969 "grad_norm": 0.023738177493214607,970 "learning_rate": 9.703165227926701e-06,971 "loss": 0.0098,972 "step": 6800973 },974 {975 "epoch": 1.5561108587005905,976 "grad_norm": 0.014806082472205162,977 "learning_rate": 9.627442071785553e-06,978 "loss": 0.017,979 "step": 6850980 },981 {982 "epoch": 1.567469332121763,983 "grad_norm": 0.11848437041044235,984 "learning_rate": 9.551718915644405e-06,985 "loss": 0.0051,986 "step": 6900987 },988 {989 "epoch": 1.578827805542935,990 "grad_norm": 2.107661485671997,991 "learning_rate": 9.475995759503258e-06,992 "loss": 0.0066,993 "step": 6950994 },995 {996 "epoch": 1.5901862789641072,997 "grad_norm": 0.08532427251338959,998 "learning_rate": 9.40027260336211e-06,999 "loss": 0.0162,1000 "step": 70001001 },1002 {1003 "epoch": 1.6015447523852795,1004 "grad_norm": 0.0039615873247385025,1005 "learning_rate": 9.324549447220961e-06,1006 "loss": 0.0043,1007 "step": 70501008 },1009 {1010 "epoch": 1.6129032258064515,1011 "grad_norm": 0.0013789839576929808,1012 "learning_rate": 9.248826291079813e-06,1013 "loss": 0.0127,1014 "step": 71001015 },1016 {1017 "epoch": 1.6242616992276238,1018 "grad_norm": 9.550993919372559,1019 "learning_rate": 9.173103134938664e-06,1020 "loss": 0.0106,1021 "step": 71501022 },1023 {1024 "epoch": 1.635620172648796,1025 "grad_norm": 0.0011268022935837507,1026 "learning_rate": 9.097379978797517e-06,1027 "loss": 0.009,1028 "step": 72001029 },1030 {1031 "epoch": 1.6469786460699682,1032 "grad_norm": 0.0006976621225476265,1033 "learning_rate": 9.021656822656369e-06,1034 "loss": 0.015,1035 "step": 72501036 },1037 {1038 "epoch": 1.6583371194911405,1039 "grad_norm": 0.0014887871220707893,1040 "learning_rate": 8.94593366651522e-06,1041 "loss": 0.015,1042 "step": 73001043 },1044 {1045 "epoch": 1.6696955929123125,1046 "grad_norm": 0.0009298041695728898,1047 "learning_rate": 8.870210510374074e-06,1048 "loss": 0.0155,1049 "step": 73501050 },1051 {1052 "epoch": 1.6810540663334848,1053 "grad_norm": 0.287706583738327,1054 "learning_rate": 8.794487354232925e-06,1055 "loss": 0.0116,1056 "step": 74001057 },1058 {1059 "epoch": 1.692412539754657,1060 "grad_norm": 0.051804594695568085,1061 "learning_rate": 8.718764198091777e-06,1062 "loss": 0.0086,1063 "step": 74501064 },1065 {1066 "epoch": 1.703771013175829,1067 "grad_norm": 1.3675575256347656,1068 "learning_rate": 8.643041041950629e-06,1069 "loss": 0.023,1070 "step": 75001071 },1072 {1073 "epoch": 1.7151294865970015,1074 "grad_norm": 0.0012162342900410295,1075 "learning_rate": 8.56731788580948e-06,1076 "loss": 0.0081,1077 "step": 75501078 },1079 {1080 "epoch": 1.7264879600181735,1081 "grad_norm": 0.0013599529629573226,1082 "learning_rate": 8.491594729668334e-06,1083 "loss": 0.0075,1084 "step": 76001085 },1086 {1087 "epoch": 1.7378464334393458,1088 "grad_norm": 0.0011247453512623906,1089 "learning_rate": 8.415871573527185e-06,1090 "loss": 0.0079,1091 "step": 76501092 },1093 {1094 "epoch": 1.749204906860518,1095 "grad_norm": 0.003529078559949994,1096 "learning_rate": 8.340148417386037e-06,1097 "loss": 0.0053,1098 "step": 77001099 },1100 {1101 "epoch": 1.76056338028169,1102 "grad_norm": 13.674657821655273,1103 "learning_rate": 8.26442526124489e-06,1104 "loss": 0.0058,1105 "step": 77501106 },1107 {1108 "epoch": 1.7719218537028625,1109 "grad_norm": 0.0018937140703201294,1110 "learning_rate": 8.188702105103742e-06,1111 "loss": 0.0073,1112 "step": 78001113 },1114 {1115 "epoch": 1.7832803271240345,1116 "grad_norm": 0.0012737379875034094,1117 "learning_rate": 8.112978948962593e-06,1118 "loss": 0.0067,1119 "step": 78501120 },1121 {1122 "epoch": 1.7946388005452067,1123 "grad_norm": 0.0010197096271440387,1124 "learning_rate": 8.037255792821446e-06,1125 "loss": 0.0057,1126 "step": 79001127 },1128 {1129 "epoch": 1.805997273966379,1130 "grad_norm": 0.009366828948259354,1131 "learning_rate": 7.961532636680296e-06,1132 "loss": 0.026,1133 "step": 79501134 },1135 {1136 "epoch": 1.817355747387551,1137 "grad_norm": 0.012473216280341148,1138 "learning_rate": 7.88580948053915e-06,1139 "loss": 0.024,1140 "step": 80001141 },1142 {1143 "epoch": 1.8287142208087233,1144 "grad_norm": 0.26876014471054077,1145 "learning_rate": 7.810086324398001e-06,1146 "loss": 0.0079,1147 "step": 80501148 },1149 {1150 "epoch": 1.8400726942298955,1151 "grad_norm": 0.0026707984507083893,1152 "learning_rate": 7.734363168256853e-06,1153 "loss": 0.0149,1154 "step": 81001155 },1156 {1157 "epoch": 1.8514311676510677,1158 "grad_norm": 0.036166541278362274,1159 "learning_rate": 7.658640012115706e-06,1160 "loss": 0.0062,1161 "step": 81501162 },1163 {1164 "epoch": 1.86278964107224,1165 "grad_norm": 0.013438834808766842,1166 "learning_rate": 7.582916855974558e-06,1167 "loss": 0.0047,1168 "step": 82001169 },1170 {1171 "epoch": 1.874148114493412,1172 "grad_norm": 0.0015038993442431092,1173 "learning_rate": 7.507193699833409e-06,1174 "loss": 0.0155,1175 "step": 82501176 },1177 {1178 "epoch": 1.8855065879145843,1179 "grad_norm": 0.0006626308313570917,1180 "learning_rate": 7.431470543692262e-06,1181 "loss": 0.0014,1182 "step": 83001183 },1184 {1185 "epoch": 1.8968650613357565,1186 "grad_norm": 0.0032690023072063923,1187 "learning_rate": 7.355747387551114e-06,1188 "loss": 0.0103,1189 "step": 83501190 },1191 {1192 "epoch": 1.9082235347569285,1193 "grad_norm": 0.0008181309676729143,1194 "learning_rate": 7.280024231409966e-06,1195 "loss": 0.0184,1196 "step": 84001197 },1198 {1199 "epoch": 1.919582008178101,1200 "grad_norm": 0.0026349183171987534,