CodeIsAbstract/HybridModelScratch_lightn_Conti
05
1{2 "best_global_step": null,3 "best_metric": null,4 "best_model_checkpoint": null,5 "epoch": 0.36,6 "eval_steps": 1000,7 "global_step": 50000,8 "is_hyper_param_search": false,9 "is_local_process_zero": true,10 "is_world_process_zero": true,11 "log_history": [12 {13 "epoch": 0.002,14 "grad_norm": 0.5271317362785339,15 "learning_rate": 0.00066,16 "loss": 7.779569091796875,17 "step": 10018 },19 {20 "epoch": 0.004,21 "grad_norm": 0.6387752890586853,22 "learning_rate": 0.0009999976160274871,23 "loss": 6.24430908203125,24 "step": 20025 },26 {27 "epoch": 0.006,28 "grad_norm": 0.42202863097190857,29 "learning_rate": 0.000999977956590203,30 "loss": 5.698394165039063,31 "step": 30032 },33 {34 "epoch": 0.008,35 "grad_norm": 0.42532625794410706,36 "learning_rate": 0.0009999384399149713,37 "loss": 5.405557861328125,38 "step": 40039 },40 {41 "epoch": 0.01,42 "grad_norm": 0.4566122889518738,43 "learning_rate": 0.0009998790675712448,44 "loss": 5.189588623046875,45 "step": 50046 },47 {48 "epoch": 0.012,49 "grad_norm": 0.4048032760620117,50 "learning_rate": 0.000999799841917069,51 "loss": 5.048131713867187,52 "step": 60053 },54 {55 "epoch": 0.014,56 "grad_norm": 0.33951064944267273,57 "learning_rate": 0.0009997007660989881,58 "loss": 4.923414611816407,59 "step": 70060 },61 {62 "epoch": 0.016,63 "grad_norm": 0.3192354440689087,64 "learning_rate": 0.0009995818440519195,65 "loss": 4.7589013671875,66 "step": 80067 },68 {69 "epoch": 0.018,70 "grad_norm": 0.4026377201080322,71 "learning_rate": 0.0009994430804989986,72 "loss": 4.643058166503907,73 "step": 90074 },75 {76 "epoch": 0.02,77 "grad_norm": 0.3453679084777832,78 "learning_rate": 0.0009992844809513896,79 "loss": 4.579913330078125,80 "step": 100081 },82 {83 "epoch": 0.02,84 "eval_accuracy": 0.269652397386301,85 "eval_loss": 4.505138874053955,86 "eval_runtime": 6.375,87 "eval_samples_per_second": 304.469,88 "eval_steps_per_second": 19.137,89 "step": 100090 },91 {92 "epoch": 0.022,93 "grad_norm": 0.3148439824581146,94 "learning_rate": 0.0009991060517080686,95 "loss": 4.5075772094726565,96 "step": 110097 },98 {99 "epoch": 0.024,100 "grad_norm": 0.3168841600418091,101 "learning_rate": 0.0009989077998555718,102 "loss": 4.395697326660156,103 "step": 1200104 },105 {106 "epoch": 0.026,107 "grad_norm": 0.277462899684906,108 "learning_rate": 0.0009986897332677144,109 "loss": 4.3901171875,110 "step": 1300111 },112 {113 "epoch": 0.028,114 "grad_norm": 0.27992936968803406,115 "learning_rate": 0.0009984518606052791,116 "loss": 4.348124084472656,117 "step": 1400118 },119 {120 "epoch": 0.03,121 "grad_norm": 0.2656252980232239,122 "learning_rate": 0.0009981941913156701,123 "loss": 4.3058538818359375,124 "step": 1500125 },126 {127 "epoch": 0.032,128 "grad_norm": 0.36868757009506226,129 "learning_rate": 0.0009979167356325394,130 "loss": 4.28273681640625,131 "step": 1600132 },133 {134 "epoch": 0.034,135 "grad_norm": 0.25333118438720703,136 "learning_rate": 0.0009976195045753797,137 "loss": 4.249351501464844,138 "step": 1700139 },140 {141 "epoch": 0.036,142 "grad_norm": 0.2491893470287323,143 "learning_rate": 0.0009973025099490876,144 "loss": 4.22731689453125,145 "step": 1800146 },147 {148 "epoch": 0.038,149 "grad_norm": 0.28424108028411865,150 "learning_rate": 0.0009969657643434933,151 "loss": 4.169223022460938,152 "step": 1900153 },154 {155 "epoch": 0.04,156 "grad_norm": 0.2666165232658386,157 "learning_rate": 0.0009966092811328612,158 "loss": 4.170299682617188,159 "step": 2000160 },161 {162 "epoch": 0.04,163 "eval_accuracy": 0.3056205014664501,164 "eval_loss": 4.103638648986816,165 "eval_runtime": 6.566,166 "eval_samples_per_second": 295.612,167 "eval_steps_per_second": 18.58,168 "step": 2000169 },170 {171 "epoch": 0.042,172 "grad_norm": 0.3338668942451477,173 "learning_rate": 0.0009962330744753604,174 "loss": 4.163348388671875,175 "step": 2100176 },177 {178 "epoch": 0.044,179 "grad_norm": 0.25480538606643677,180 "learning_rate": 0.0009958371593124988,181 "loss": 4.118557434082032,182 "step": 2200183 },184 {185 "epoch": 0.046,186 "grad_norm": 0.2351720631122589,187 "learning_rate": 0.0009954215513685335,188 "loss": 4.1006707763671875,189 "step": 2300190 },191 {192 "epoch": 0.048,193 "grad_norm": 0.23063364624977112,194 "learning_rate": 0.0009949862671498437,195 "loss": 4.084933166503906,196 "step": 2400197 },198 {199 "epoch": 0.05,200 "grad_norm": 0.270161896944046,201 "learning_rate": 0.000994531323944276,202 "loss": 4.043960876464844,203 "step": 2500204 },205 {206 "epoch": 0.052,207 "grad_norm": 0.23590487241744995,208 "learning_rate": 0.0009940567398204583,209 "loss": 4.043964233398437,210 "step": 2600211 },212 {213 "epoch": 0.054,214 "grad_norm": 0.2604995369911194,215 "learning_rate": 0.000993562533627081,216 "loss": 4.023601379394531,217 "step": 2700218 },219 {220 "epoch": 0.056,221 "grad_norm": 0.21301409602165222,222 "learning_rate": 0.0009930487249921495,223 "loss": 4.006760864257813,224 "step": 2800225 },226 {227 "epoch": 0.058,228 "grad_norm": 0.21575306355953217,229 "learning_rate": 0.0009925153343222043,230 "loss": 3.99057861328125,231 "step": 2900232 },233 {234 "epoch": 0.06,235 "grad_norm": 0.2405770868062973,236 "learning_rate": 0.00099196238280151,237 "loss": 4.004396667480469,238 "step": 3000239 },240 {241 "epoch": 0.06,242 "eval_accuracy": 0.3194532243250246,243 "eval_loss": 3.945106267929077,244 "eval_runtime": 6.283,245 "eval_samples_per_second": 308.927,246 "eval_steps_per_second": 19.417,247 "step": 3000248 },249 {250 "epoch": 0.062,251 "grad_norm": 0.22380003333091736,252 "learning_rate": 0.0009913898923912148,253 "loss": 3.9952774047851562,254 "step": 3100255 },256 {257 "epoch": 0.064,258 "grad_norm": 0.21692751348018646,259 "learning_rate": 0.0009907978858284775,260 "loss": 3.9829229736328124,261 "step": 3200262 },263 {264 "epoch": 0.066,265 "grad_norm": 0.22457344830036163,266 "learning_rate": 0.0009901863866255652,267 "loss": 3.9637200927734373,268 "step": 3300269 },270 {271 "epoch": 0.068,272 "grad_norm": 0.20820939540863037,273 "learning_rate": 0.0009895554190689191,274 "loss": 3.930269470214844,275 "step": 3400276 },277 {278 "epoch": 0.07,279 "grad_norm": 0.1961025595664978,280 "learning_rate": 0.0009889050082181896,281 "loss": 3.91483154296875,282 "step": 3500283 },284 {285 "epoch": 0.072,286 "grad_norm": 0.23347830772399902,287 "learning_rate": 0.0009882351799052416,288 "loss": 3.8973043823242186,289 "step": 3600290 },291 {292 "epoch": 0.074,293 "grad_norm": 0.21384640038013458,294 "learning_rate": 0.0009875459607331285,295 "loss": 3.9072760009765624,296 "step": 3700297 },298 {299 "epoch": 0.076,300 "grad_norm": 0.21380040049552917,301 "learning_rate": 0.000986837378075035,302 "loss": 3.8974929809570313,303 "step": 3800304 },305 {306 "epoch": 0.078,307 "grad_norm": 0.2199064940214157,308 "learning_rate": 0.0009861094600731911,309 "loss": 3.884527587890625,310 "step": 3900311 },312 {313 "epoch": 0.08,314 "grad_norm": 0.2245541661977768,315 "learning_rate": 0.000985362235637753,316 "loss": 3.863822021484375,317 "step": 4000318 },319 {320 "epoch": 0.08,321 "eval_accuracy": 0.3280885939521158,322 "eval_loss": 3.8499362468719482,323 "eval_runtime": 6.3172,324 "eval_samples_per_second": 307.255,325 "eval_steps_per_second": 19.312,326 "step": 4000327 },328 {329 "epoch": 0.082,330 "grad_norm": 0.20376908779144287,331 "learning_rate": 0.000984595734445656,332 "loss": 3.8685012817382813,333 "step": 4100334 },335 {336 "epoch": 0.084,337 "grad_norm": 0.18422065675258636,338 "learning_rate": 0.0009838099869394353,339 "loss": 3.853851318359375,340 "step": 4200341 },342 {343 "epoch": 0.086,344 "grad_norm": 0.19115814566612244,345 "learning_rate": 0.0009830050243260172,346 "loss": 3.8678164672851563,347 "step": 4300348 },349 {350 "epoch": 0.088,351 "grad_norm": 0.20425236225128174,352 "learning_rate": 0.0009821808785754791,353 "loss": 3.8287237548828124,354 "step": 4400355 },356 {357 "epoch": 0.09,358 "grad_norm": 0.1944635510444641,359 "learning_rate": 0.0009813375824197811,360 "loss": 3.8451242065429687,361 "step": 4500362 },363 {364 "epoch": 0.092,365 "grad_norm": 0.17947296798229218,366 "learning_rate": 0.0009804751693514645,367 "loss": 3.845400695800781,368 "step": 4600369 },370 {371 "epoch": 0.094,372 "grad_norm": 0.20076608657836914,373 "learning_rate": 0.0009795936736223222,374 "loss": 3.8257626342773436,375 "step": 4700376 },377 {378 "epoch": 0.096,379 "grad_norm": 0.18554596602916718,380 "learning_rate": 0.0009786931302420386,381 "loss": 3.808670959472656,382 "step": 4800383 },384 {385 "epoch": 0.098,386 "grad_norm": 0.23913995921611786,387 "learning_rate": 0.000977773574976799,388 "loss": 3.810527038574219,389 "step": 4900390 },391 {392 "epoch": 0.1,393 "grad_norm": 0.18414103984832764,394 "learning_rate": 0.0009768350443478686,395 "loss": 3.7990948486328127,396 "step": 5000397 },398 {399 "epoch": 0.1,400 "eval_accuracy": 0.33500293894949945,401 "eval_loss": 3.7702300548553467,402 "eval_runtime": 6.4137,403 "eval_samples_per_second": 302.634,404 "eval_steps_per_second": 19.022,405 "step": 5000406 },407 {408 "epoch": 0.102,409 "grad_norm": 0.20944379270076752,410 "learning_rate": 0.0009758775756301432,411 "loss": 3.798116455078125,412 "step": 5100413 },414 {415 "epoch": 0.104,416 "grad_norm": 0.18056762218475342,417 "learning_rate": 0.0009749012068506673,418 "loss": 3.7997445678710937,419 "step": 5200420 },421 {422 "epoch": 0.106,423 "grad_norm": 0.20922255516052246,424 "learning_rate": 0.000973905976787125,425 "loss": 3.7765731811523438,426 "step": 5300427 },428 {429 "epoch": 0.108,430 "grad_norm": 0.19328580796718597,431 "learning_rate": 0.0009728919249662993,432 "loss": 3.729337463378906,433 "step": 5400434 },435 {436 "epoch": 0.11,437 "grad_norm": 0.2031419277191162,438 "learning_rate": 0.0009718590916625022,439 "loss": 3.744057922363281,440 "step": 5500441 },442 {443 "epoch": 0.112,444 "grad_norm": 0.19743800163269043,445 "learning_rate": 0.0009708075178959756,446 "loss": 3.7054071044921875,447 "step": 5600448 },449 {450 "epoch": 0.114,451 "grad_norm": 0.1741236299276352,452 "learning_rate": 0.0009697372454312618,453 "loss": 3.699843444824219,454 "step": 5700455 },456 {457 "epoch": 0.116,458 "grad_norm": 0.1799750030040741,459 "learning_rate": 0.0009686483167755448,460 "loss": 3.7227572631835937,461 "step": 5800462 },463 {464 "epoch": 0.118,465 "grad_norm": 0.2036087065935135,466 "learning_rate": 0.000967540775176962,467 "loss": 3.6742709350585936,468 "step": 5900469 },470 {471 "epoch": 0.12,472 "grad_norm": 0.21463191509246826,473 "learning_rate": 0.0009664146646228871,474 "loss": 3.6995980834960935,475 "step": 6000476 },477 {478 "epoch": 0.12,479 "eval_accuracy": 0.34141821705074654,480 "eval_loss": 3.711214303970337,481 "eval_runtime": 6.3349,482 "eval_samples_per_second": 306.4,483 "eval_steps_per_second": 19.259,484 "step": 6000485 },486 {487 "epoch": 0.122,488 "grad_norm": 0.18666808307170868,489 "learning_rate": 0.000965270029838182,490 "loss": 3.6821221923828125,491 "step": 6100492 },493 {494 "epoch": 0.124,495 "grad_norm": 0.15751489996910095,496 "learning_rate": 0.0009641069162834215,497 "loss": 3.671753234863281,498 "step": 6200499 },500 {501 "epoch": 0.126,502 "grad_norm": 0.22766178846359253,503 "learning_rate": 0.0009629253701530875,504 "loss": 3.6878753662109376,505 "step": 6300506 },507 {508 "epoch": 0.128,509 "grad_norm": 0.1677054762840271,510 "learning_rate": 0.0009617254383737343,511 "loss": 3.6619857788085937,512 "step": 6400513 },514 {515 "epoch": 0.13,516 "grad_norm": 0.17945168912410736,517 "learning_rate": 0.0009605071686021245,518 "loss": 3.6484750366210936,519 "step": 6500520 },521 {522 "epoch": 0.132,523 "grad_norm": 0.17799174785614014,524 "learning_rate": 0.0009592706092233365,525 "loss": 3.6635867309570314,526 "step": 6600527 },528 {529 "epoch": 0.134,530 "grad_norm": 0.5586189031600952,531 "learning_rate": 0.0009580158093488436,532 "loss": 3.6489688110351564,533 "step": 6700534 },535 {536 "epoch": 0.136,537 "grad_norm": 0.2055712342262268,538 "learning_rate": 0.000956742818814562,539 "loss": 3.6523666381835938,540 "step": 6800541 },542 {543 "epoch": 0.138,544 "grad_norm": 0.1812128871679306,545 "learning_rate": 0.0009554516881788724,546 "loss": 3.6479287719726563,547 "step": 6900548 },549 {550 "epoch": 0.14,551 "grad_norm": 0.1664743423461914,552 "learning_rate": 0.0009541424687206124,553 "loss": 3.6319317626953125,554 "step": 7000555 },556 {557 "epoch": 0.14,558 "eval_accuracy": 0.3448148965923309,559 "eval_loss": 3.6666343212127686,560 "eval_runtime": 6.38,561 "eval_samples_per_second": 304.231,562 "eval_steps_per_second": 19.122,563 "step": 7000564 },565 {566 "epoch": 0.142,567 "grad_norm": 0.15217284858226776,568 "learning_rate": 0.0009528152124370387,569 "loss": 3.6308517456054688,570 "step": 7100571 },572 {573 "epoch": 0.144,574 "grad_norm": 0.16507816314697266,575 "learning_rate": 0.0009514699720417631,576 "loss": 3.6217156982421876,577 "step": 7200578 },579 {580 "epoch": 0.146,581 "grad_norm": 0.2040642648935318,582 "learning_rate": 0.0009501068009626583,583 "loss": 3.632703857421875,584 "step": 7300585 },586 {587 "epoch": 0.148,588 "grad_norm": 0.18437133729457855,589 "learning_rate": 0.0009487257533397361,590 "loss": 3.6346307373046876,591 "step": 7400592 },593 {594 "epoch": 0.15,595 "grad_norm": 0.18760915100574493,596 "learning_rate": 0.0009473268840229971,597 "loss": 3.5973056030273436,598 "step": 7500599 },600 {601 "epoch": 0.152,602 "grad_norm": 0.17388536036014557,603 "learning_rate": 0.0009459102485702528,604 "loss": 3.611794738769531,605 "step": 7600606 },607 {608 "epoch": 0.154,609 "grad_norm": 0.18797264993190765,610 "learning_rate": 0.0009444759032449177,611 "loss": 3.609403381347656,612 "step": 7700613 },614 {615 "epoch": 0.156,616 "grad_norm": 0.18140241503715515,617 "learning_rate": 0.0009430239050137766,618 "loss": 3.6027340698242187,619 "step": 7800620 },621 {622 "epoch": 0.158,623 "grad_norm": 0.17433640360832214,624 "learning_rate": 0.0009415543115447203,625 "loss": 3.6214892578125,626 "step": 7900627 },628 {629 "epoch": 0.16,630 "grad_norm": 0.16605013608932495,631 "learning_rate": 0.0009400671812044565,632 "loss": 3.6061587524414063,633 "step": 8000634 },635 {636 "epoch": 0.16,637 "eval_accuracy": 0.34762277801806923,638 "eval_loss": 3.640489339828491,639 "eval_runtime": 8.0453,640 "eval_samples_per_second": 241.259,641 "eval_steps_per_second": 15.164,642 "step": 8000643 },644 {645 "epoch": 0.162,646 "grad_norm": 0.16548939049243927,647 "learning_rate": 0.0009385625730561914,648 "loss": 3.588617248535156,649 "step": 8100650 },651 {652 "epoch": 0.164,653 "grad_norm": 0.16542939841747284,654 "learning_rate": 0.0009370405468572831,655 "loss": 3.590567626953125,656 "step": 8200657 },658 {659 "epoch": 0.166,660 "grad_norm": 0.18206581473350525,661 "learning_rate": 0.0009355011630568697,662 "loss": 3.636882629394531,663 "step": 8300664 },665 {666 "epoch": 0.168,667 "grad_norm": 0.2654706835746765,668 "learning_rate": 0.0009339444827934676,669 "loss": 3.6797540283203123,670 "step": 8400671 },672 {673 "epoch": 0.17,674 "grad_norm": 0.16825009882450104,675 "learning_rate": 0.0009323705678925427,676 "loss": 3.686376953125,677 "step": 8500678 },679 {680 "epoch": 0.172,681 "grad_norm": 0.18937945365905762,682 "learning_rate": 0.000930779480864057,683 "loss": 3.6942425537109376,684 "step": 8600685 },686 {687 "epoch": 0.174,688 "grad_norm": 0.19819994270801544,689 "learning_rate": 0.0009291712848999833,690 "loss": 3.6582196044921873,691 "step": 8700692 },693 {694 "epoch": 0.176,695 "grad_norm": 0.1557057648897171,696 "learning_rate": 0.0009275460438717975,697 "loss": 3.6659271240234377,698 "step": 8800699 },700 {701 "epoch": 0.178,702 "grad_norm": 0.16359280049800873,703 "learning_rate": 0.0009259038223279413,704 "loss": 3.66092529296875,705 "step": 8900706 },707 {708 "epoch": 0.18,709 "grad_norm": 0.17897452414035797,710 "learning_rate": 0.0009242446854912575,711 "loss": 3.6642529296875,712 "step": 9000713 },714 {715 "epoch": 0.18,716 "eval_accuracy": 0.3497692697794326,717 "eval_loss": 3.617546319961548,718 "eval_runtime": 6.279,719 "eval_samples_per_second": 309.124,720 "eval_steps_per_second": 19.43,721 "step": 9000722 },723 {724 "epoch": 0.182,725 "grad_norm": 0.20515520870685577,726 "learning_rate": 0.0009225686992564015,727 "loss": 3.6445053100585936,728 "step": 9100729 },730 {731 "epoch": 0.184,732 "grad_norm": 0.16788582503795624,733 "learning_rate": 0.0009208759301872229,734 "loss": 3.657896728515625,735 "step": 9200736 },737 {738 "epoch": 0.186,739 "grad_norm": 0.1988508403301239,740 "learning_rate": 0.0009191664455141217,741 "loss": 3.616723937988281,742 "step": 9300743 },744 {745 "epoch": 0.188,746 "grad_norm": 0.1822570413351059,747 "learning_rate": 0.0009174403131313792,748 "loss": 3.6103390502929686,749 "step": 9400750 },751 {752 "epoch": 0.19,753 "grad_norm": 0.15679553151130676,754 "learning_rate": 0.0009156976015944607,755 "loss": 3.624302978515625,756 "step": 9500757 },758 {759 "epoch": 0.192,760 "grad_norm": 0.15683937072753906,761 "learning_rate": 0.0009139383801172929,762 "loss": 3.6558917236328123,763 "step": 9600764 },765 {766 "epoch": 0.194,767 "grad_norm": 0.1553466022014618,768 "learning_rate": 0.0009121627185695152,769 "loss": 3.6198284912109373,770 "step": 9700771 },772 {773 "epoch": 0.196,774 "grad_norm": 0.14881721138954163,775 "learning_rate": 0.0009103706874737043,776 "loss": 3.60408203125,777 "step": 9800778 },779 {780 "epoch": 0.198,781 "grad_norm": 0.1628589779138565,782 "learning_rate": 0.0009085623580025737,783 "loss": 3.6047940063476562,784 "step": 9900785 },786 {787 "epoch": 0.2,788 "grad_norm": 0.15589991211891174,789 "learning_rate": 0.0009067378019761472,790 "loss": 3.599223327636719,791 "step": 10000792 },793 {794 "epoch": 0.2,795 "eval_accuracy": 0.3534593401629882,796 "eval_loss": 3.579674005508423,797 "eval_runtime": 6.3176,798 "eval_samples_per_second": 307.239,799 "eval_steps_per_second": 19.311,800 "step": 10000801 },802 {803 "epoch": 0.202,804 "grad_norm": 0.18158847093582153,805 "learning_rate": 0.0009048970918589055,806 "loss": 3.607147521972656,807 "step": 10100808 },809 {810 "epoch": 0.204,811 "grad_norm": 0.15705344080924988,812 "learning_rate": 0.0009030403007569094,813 "loss": 3.5972610473632813,814 "step": 10200815 },816 {817 "epoch": 0.206,818 "grad_norm": 0.18561947345733643,819 "learning_rate": 0.0009011675024148955,820 "loss": 3.5994192504882814,821 "step": 10300822 },823 {824 "epoch": 0.208,825 "grad_norm": 0.17375363409519196,826 "learning_rate": 0.0008992787712133476,827 "loss": 3.592384948730469,828 "step": 10400829 },830 {831 "epoch": 0.21,832 "grad_norm": 0.14704184234142303,833 "learning_rate": 0.0008973741821655428,834 "loss": 3.5763116455078126,835 "step": 10500836 },837 {838 "epoch": 0.212,839 "grad_norm": 0.1663106381893158,840 "learning_rate": 0.0008954538109145713,841 "loss": 3.59366455078125,842 "step": 10600843 },844 {845 "epoch": 0.214,846 "grad_norm": 0.15191592276096344,847 "learning_rate": 0.0008935177337303337,848 "loss": 3.5883197021484374,849 "step": 10700850 },851 {852 "epoch": 0.216,853 "grad_norm": 0.15566915273666382,854 "learning_rate": 0.0008915660275065107,855 "loss": 3.5549887084960936,856 "step": 10800857 },858 {859 "epoch": 0.218,860 "grad_norm": 0.17490389943122864,861 "learning_rate": 0.0008895987697575093,862 "loss": 3.558960266113281,863 "step": 10900864 },865 {866 "epoch": 0.22,867 "grad_norm": 0.17277678847312927,868 "learning_rate": 0.0008876160386153846,869 "loss": 3.5671484375,870 "step": 11000871 },872 {873 "epoch": 0.22,874 "eval_accuracy": 0.3558709927196726,875 "eval_loss": 3.554290771484375,876 "eval_runtime": 6.8937,877 "eval_samples_per_second": 281.559,878 "eval_steps_per_second": 17.697,879 "step": 11000880 },881 {882 "epoch": 0.222,883 "grad_norm": 0.15808428823947906,884 "learning_rate": 0.0008856179128267363,885 "loss": 3.578575439453125,886 "step": 11100887 },888 {889 "epoch": 0.224,890 "grad_norm": 0.14474111795425415,891 "learning_rate": 0.0008836044717495819,892 "loss": 3.5772747802734375,893 "step": 11200894 },895 {896 "epoch": 0.226,897 "grad_norm": 0.1413886845111847,898 "learning_rate": 0.0008815757953502035,899 "loss": 3.556199645996094,900 "step": 11300901 },902 {903 "epoch": 0.228,904 "grad_norm": 0.1502675712108612,905 "learning_rate": 0.0008795319641999735,906 "loss": 3.570118103027344,907 "step": 11400908 },909 {910 "epoch": 0.23,911 "grad_norm": 0.1517445147037506,912 "learning_rate": 0.0008774730594721534,913 "loss": 3.5563198852539064,914 "step": 11500915 },916 {917 "epoch": 0.232,918 "grad_norm": 0.16406166553497314,919 "learning_rate": 0.0008753991629386706,920 "loss": 3.5616604614257814,921 "step": 11600922 },923 {924 "epoch": 0.234,925 "grad_norm": 0.14764074981212616,926 "learning_rate": 0.0008733103569668701,927 "loss": 3.5495458984375,928 "step": 11700929 },930 {931 "epoch": 0.236,932 "grad_norm": 0.17058512568473816,933 "learning_rate": 0.0008712067245162439,934 "loss": 3.5244882202148435,935 "step": 11800936 },937 {938 "epoch": 0.238,939 "grad_norm": 0.16127121448516846,940 "learning_rate": 0.0008690883491351356,941 "loss": 3.5396826171875,942 "step": 11900943 },944 {945 "epoch": 0.24,946 "grad_norm": 0.14211590588092804,947 "learning_rate": 0.0008669553149574224,948 "loss": 3.522979431152344,949 "step": 12000950 },951 {952 "epoch": 0.24,953 "eval_accuracy": 0.3578128166428224,954 "eval_loss": 3.5307533740997314,955 "eval_runtime": 6.6725,956 "eval_samples_per_second": 290.897,957 "eval_steps_per_second": 18.284,958 "step": 12000959 },960 {961 "epoch": 0.242,962 "grad_norm": 0.1405150443315506,963 "learning_rate": 0.0008648077066991736,964 "loss": 3.527177734375,965 "step": 12100966 },967 {968 "epoch": 0.244,969 "grad_norm": 0.16479545831680298,970 "learning_rate": 0.0008626456096552858,971 "loss": 3.5393338012695312,972 "step": 12200973 },974 {975 "epoch": 0.246,976 "grad_norm": 0.1583888679742813,977 "learning_rate": 0.0008604691096960961,978 "loss": 3.5796047973632814,979 "step": 12300980 },981 {982 "epoch": 0.248,983 "grad_norm": 0.17304493486881256,984 "learning_rate": 0.0008582782932639702,985 "loss": 3.591195068359375,986 "step": 12400987 },988 {989 "epoch": 0.25,990 "grad_norm": 0.15857121348381042,991 "learning_rate": 0.0008560732473698707,992 "loss": 3.584400634765625,993 "step": 12500994 },995 {996 "epoch": 0.252,997 "grad_norm": 0.14658208191394806,998 "learning_rate": 0.0008538540595899008,999 "loss": 3.589564208984375,1000 "step": 126001001 },1002 {1003 "epoch": 0.254,1004 "grad_norm": 0.143401101231575,1005 "learning_rate": 0.0008516208180618252,1006 "loss": 3.5824600219726563,1007 "step": 127001008 },1009 {1010 "epoch": 0.256,1011 "grad_norm": 0.16244202852249146,1012 "learning_rate": 0.0008493736114815718,1013 "loss": 3.574875793457031,1014 "step": 128001015 },1016 {1017 "epoch": 0.258,1018 "grad_norm": 0.1520756632089615,1019 "learning_rate": 0.0008471125290997063,1020 "loss": 3.568282165527344,1021 "step": 129001022 },1023 {1024 "epoch": 0.26,1025 "grad_norm": 0.1582706868648529,1026 "learning_rate": 0.00084483766071789,1027 "loss": 3.5930340576171873,1028 "step": 130001029 },1030 {1031 "epoch": 0.26,1032 "eval_accuracy": 0.35929186944409996,1033 "eval_loss": 3.5190227031707764,1034 "eval_runtime": 6.3791,1035 "eval_samples_per_second": 304.275,1036 "eval_steps_per_second": 19.125,1037 "step": 130001038 },1039 {1040 "epoch": 0.262,1041 "grad_norm": 0.17478469014167786,1042 "learning_rate": 0.0008425490966853113,1043 "loss": 3.5665899658203126,1044 "step": 131001045 },1046 {1047 "epoch": 0.264,1048 "grad_norm": 0.15228988230228424,1049 "learning_rate": 0.0008402469278950989,1050 "loss": 3.5444100952148436,1051 "step": 132001052 },1053 {1054 "epoch": 0.266,1055 "grad_norm": 0.15244163572788239,1056 "learning_rate": 0.0008379312457807102,1057 "loss": 3.5695867919921875,1058 "step": 133001059 },1060 {1061 "epoch": 0.268,1062 "grad_norm": 0.15935847163200378,1063 "learning_rate": 0.0008356021423123017,1064 "loss": 3.5604925537109375,1065 "step": 134001066 },1067 {1068 "epoch": 0.27,1069 "grad_norm": 0.15347428619861603,1070 "learning_rate": 0.000833259709993075,1071 "loss": 3.5551889038085935,1072 "step": 135001073 },1074 {1075 "epoch": 0.272,1076 "grad_norm": 0.21459928154945374,1077 "learning_rate": 0.0008309040418556038,1078 "loss": 3.550736389160156,1079 "step": 136001080 },1081 {1082 "epoch": 0.274,1083 "grad_norm": 0.17246893048286438,1084 "learning_rate": 0.0008285352314581378,1085 "loss": 3.5650180053710936,1086 "step": 137001087 },1088 {1089 "epoch": 0.276,1090 "grad_norm": 0.16454122960567474,1091 "learning_rate": 0.0008261533728808883,1092 "loss": 3.549154968261719,1093 "step": 138001094 },1095 {1096 "epoch": 0.278,1097 "grad_norm": 0.14885519444942474,1098 "learning_rate": 0.000823758560722291,1099 "loss": 3.5430950927734375,1100 "step": 139001101 },1102 {1103 "epoch": 0.28,1104 "grad_norm": 0.1521039605140686,1105 "learning_rate": 0.0008213508900952489,1106 "loss": 3.547527160644531,1107 "step": 140001108 },1109 {1110 "epoch": 0.28,1111 "eval_accuracy": 0.36159261824608735,1112 "eval_loss": 3.4961202144622803,1113 "eval_runtime": 6.3742,1114 "eval_samples_per_second": 304.507,1115 "eval_steps_per_second": 19.14,1116 "step": 140001117 },1118 {1119 "epoch": 0.282,1120 "grad_norm": 0.1499979943037033,1121 "learning_rate": 0.0008189304566233547,1122 "loss": 3.5474667358398437,1123 "step": 141001124 },1125 {1126 "epoch": 0.284,1127 "grad_norm": 0.1552744209766388,1128 "learning_rate": 0.0008164973564370934,1129 "loss": 3.5074334716796876,1130 "step": 142001131 },1132 {1133 "epoch": 0.286,1134 "grad_norm": 0.13462179899215698,1135 "learning_rate": 0.0008140516861700238,1136 "loss": 3.5272781372070314,1137 "step": 143001138 },1139 {1140 "epoch": 0.288,1141 "grad_norm": 0.1381780356168747,1142 "learning_rate": 0.000811593542954941,1143 "loss": 3.5094314575195313,1144 "step": 144001145 },1146 {1147 "epoch": 0.29,1148 "grad_norm": 0.13239678740501404,1149 "learning_rate": 0.000809123024420019,1150 "loss": 3.5180905151367186,1151 "step": 145001152 },1153 {1154 "epoch": 0.292,1155 "grad_norm": 0.15930698812007904,1156 "learning_rate": 0.0008066402286849317,1157 "loss": 3.5262945556640624,1158 "step": 146001159 },1160 {1161 "epoch": 0.294,1162 "grad_norm": 0.1524200737476349,1163 "learning_rate": 0.0008041452543569583,1164 "loss": 3.541943359375,1165 "step": 147001166 },1167 {1168 "epoch": 0.296,1169 "grad_norm": 0.14040064811706543,1170 "learning_rate": 0.0008016382005270651,1171 "loss": 3.543092041015625,1172 "step": 148001173 },1174 {1175 "epoch": 0.298,1176 "grad_norm": 0.14799822866916656,1177 "learning_rate": 0.0007991191667659708,1178 "loss": 3.5315762329101563,1179 "step": 149001180 },1181 {1182 "epoch": 0.3,1183 "grad_norm": 0.1449238359928131,1184 "learning_rate": 0.0007965882531201913,1185 "loss": 3.522008361816406,1186 "step": 150001187 },1188 {1189 "epoch": 0.3,1190 "eval_accuracy": 0.36332372503531274,1191 "eval_loss": 3.4778010845184326,1192 "eval_runtime": 6.6567,1193 "eval_samples_per_second": 291.588,1194 "eval_steps_per_second": 18.328,1195 "step": 150001196 },1197 {1198 "epoch": 0.302,1199 "grad_norm": 0.13593044877052307,1200 "learning_rate": 0.0007940455601080678,