Agents-X/PyVision-Video-7B-SFT
013
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": 6885,8 "is_hyper_param_search": false,9 "is_local_process_zero": true,10 "is_world_process_zero": true,11 "log_history": [12 {13 "epoch": 0.0014524328249818446,14 "grad_norm": 4.328955480739728,15 "learning_rate": 1.3062409288824383e-07,16 "loss": 0.9607,17 "step": 1018 },19 {20 "epoch": 0.002904865649963689,21 "grad_norm": 4.469323164876104,22 "learning_rate": 2.757619738751814e-07,23 "loss": 0.9859,24 "step": 2025 },26 {27 "epoch": 0.004357298474945534,28 "grad_norm": 4.000416594025176,29 "learning_rate": 4.2089985486211904e-07,30 "loss": 0.9872,31 "step": 3032 },33 {34 "epoch": 0.005809731299927378,35 "grad_norm": 3.1566001029759914,36 "learning_rate": 5.660377358490567e-07,37 "loss": 0.9191,38 "step": 4039 },40 {41 "epoch": 0.007262164124909223,42 "grad_norm": 2.000776925354802,43 "learning_rate": 7.111756168359943e-07,44 "loss": 0.866,45 "step": 5046 },47 {48 "epoch": 0.008714596949891068,49 "grad_norm": 2.03383269865318,50 "learning_rate": 8.563134978229319e-07,51 "loss": 0.8475,52 "step": 6053 },54 {55 "epoch": 0.010167029774872912,56 "grad_norm": 1.981671850063017,57 "learning_rate": 1.0014513788098695e-06,58 "loss": 0.8145,59 "step": 7060 },61 {62 "epoch": 0.011619462599854757,63 "grad_norm": 1.9935447101504142,64 "learning_rate": 1.146589259796807e-06,65 "loss": 0.7874,66 "step": 8067 },68 {69 "epoch": 0.013071895424836602,70 "grad_norm": 1.696794144473072,71 "learning_rate": 1.2917271407837448e-06,72 "loss": 0.7606,73 "step": 9074 },75 {76 "epoch": 0.014524328249818447,77 "grad_norm": 1.8441704167155635,78 "learning_rate": 1.4368650217706823e-06,79 "loss": 0.7505,80 "step": 10081 },82 {83 "epoch": 0.01597676107480029,84 "grad_norm": 1.6167640330505846,85 "learning_rate": 1.5820029027576197e-06,86 "loss": 0.7432,87 "step": 11088 },89 {90 "epoch": 0.017429193899782137,91 "grad_norm": 1.7310300613256226,92 "learning_rate": 1.7271407837445576e-06,93 "loss": 0.7502,94 "step": 12095 },96 {97 "epoch": 0.01888162672476398,98 "grad_norm": 1.5504171157690307,99 "learning_rate": 1.872278664731495e-06,100 "loss": 0.7075,101 "step": 130102 },103 {104 "epoch": 0.020334059549745823,105 "grad_norm": 1.5001595551333269,106 "learning_rate": 2.0174165457184327e-06,107 "loss": 0.7242,108 "step": 140109 },110 {111 "epoch": 0.02178649237472767,112 "grad_norm": 1.7680255328873922,113 "learning_rate": 2.1625544267053704e-06,114 "loss": 0.7299,115 "step": 150116 },117 {118 "epoch": 0.023238925199709513,119 "grad_norm": 1.9776874021989124,120 "learning_rate": 2.307692307692308e-06,121 "loss": 0.7074,122 "step": 160123 },124 {125 "epoch": 0.024691358024691357,126 "grad_norm": 1.645294675336186,127 "learning_rate": 2.4528301886792453e-06,128 "loss": 0.7003,129 "step": 170130 },131 {132 "epoch": 0.026143790849673203,133 "grad_norm": 1.903626800669526,134 "learning_rate": 2.597968069666183e-06,135 "loss": 0.6935,136 "step": 180137 },138 {139 "epoch": 0.027596223674655047,140 "grad_norm": 1.6296522016767983,141 "learning_rate": 2.7431059506531207e-06,142 "loss": 0.7099,143 "step": 190144 },145 {146 "epoch": 0.029048656499636893,147 "grad_norm": 1.5624745122869332,148 "learning_rate": 2.8882438316400583e-06,149 "loss": 0.7082,150 "step": 200151 },152 {153 "epoch": 0.030501089324618737,154 "grad_norm": 1.5327148829437787,155 "learning_rate": 3.033381712626996e-06,156 "loss": 0.6847,157 "step": 210158 },159 {160 "epoch": 0.03195352214960058,161 "grad_norm": 1.4217156007581908,162 "learning_rate": 3.1785195936139337e-06,163 "loss": 0.6997,164 "step": 220165 },166 {167 "epoch": 0.03340595497458242,168 "grad_norm": 1.678714535521671,169 "learning_rate": 3.323657474600871e-06,170 "loss": 0.6922,171 "step": 230172 },173 {174 "epoch": 0.034858387799564274,175 "grad_norm": 1.6893028132334575,176 "learning_rate": 3.4687953555878086e-06,177 "loss": 0.6764,178 "step": 240179 },180 {181 "epoch": 0.03631082062454612,182 "grad_norm": 1.6842923668045748,183 "learning_rate": 3.6139332365747467e-06,184 "loss": 0.6838,185 "step": 250186 },187 {188 "epoch": 0.03776325344952796,189 "grad_norm": 2.0758637079489306,190 "learning_rate": 3.759071117561684e-06,191 "loss": 0.6961,192 "step": 260193 },194 {195 "epoch": 0.0392156862745098,196 "grad_norm": 1.651886885559497,197 "learning_rate": 3.904208998548621e-06,198 "loss": 0.6619,199 "step": 270200 },201 {202 "epoch": 0.04066811909949165,203 "grad_norm": 1.6813735734416895,204 "learning_rate": 4.049346879535559e-06,205 "loss": 0.691,206 "step": 280207 },208 {209 "epoch": 0.04212055192447349,210 "grad_norm": 1.8001370749006687,211 "learning_rate": 4.194484760522497e-06,212 "loss": 0.6646,213 "step": 290214 },215 {216 "epoch": 0.04357298474945534,217 "grad_norm": 1.8255351447030483,218 "learning_rate": 4.339622641509435e-06,219 "loss": 0.6595,220 "step": 300221 },222 {223 "epoch": 0.04502541757443718,224 "grad_norm": 1.7918481140936697,225 "learning_rate": 4.484760522496372e-06,226 "loss": 0.6555,227 "step": 310228 },229 {230 "epoch": 0.04647785039941903,231 "grad_norm": 1.6697318257583398,232 "learning_rate": 4.629898403483309e-06,233 "loss": 0.6734,234 "step": 320235 },236 {237 "epoch": 0.04793028322440087,238 "grad_norm": 1.5656777878920214,239 "learning_rate": 4.775036284470247e-06,240 "loss": 0.6511,241 "step": 330242 },243 {244 "epoch": 0.04938271604938271,245 "grad_norm": 1.6515736055504289,246 "learning_rate": 4.920174165457185e-06,247 "loss": 0.6651,248 "step": 340249 },250 {251 "epoch": 0.050835148874364564,252 "grad_norm": 1.6517233906536315,253 "learning_rate": 5.065312046444122e-06,254 "loss": 0.665,255 "step": 350256 },257 {258 "epoch": 0.05228758169934641,259 "grad_norm": 1.6987223199576384,260 "learning_rate": 5.210449927431061e-06,261 "loss": 0.6632,262 "step": 360263 },264 {265 "epoch": 0.05374001452432825,266 "grad_norm": 1.578744968443496,267 "learning_rate": 5.355587808417998e-06,268 "loss": 0.665,269 "step": 370270 },271 {272 "epoch": 0.05519244734931009,273 "grad_norm": 1.4975426293081397,274 "learning_rate": 5.500725689404935e-06,275 "loss": 0.6511,276 "step": 380277 },278 {279 "epoch": 0.05664488017429194,280 "grad_norm": 1.7386717568110297,281 "learning_rate": 5.645863570391873e-06,282 "loss": 0.6676,283 "step": 390284 },285 {286 "epoch": 0.05809731299927379,287 "grad_norm": 1.5916583497500596,288 "learning_rate": 5.7910014513788105e-06,289 "loss": 0.6635,290 "step": 400291 },292 {293 "epoch": 0.05954974582425563,294 "grad_norm": 1.6931617934865184,295 "learning_rate": 5.936139332365748e-06,296 "loss": 0.6668,297 "step": 410298 },299 {300 "epoch": 0.06100217864923747,301 "grad_norm": 1.5616372247201953,302 "learning_rate": 6.081277213352685e-06,303 "loss": 0.6685,304 "step": 420305 },306 {307 "epoch": 0.06245461147421932,308 "grad_norm": 1.5424914283941253,309 "learning_rate": 6.226415094339623e-06,310 "loss": 0.659,311 "step": 430312 },313 {314 "epoch": 0.06390704429920116,315 "grad_norm": 1.6468311050594455,316 "learning_rate": 6.37155297532656e-06,317 "loss": 0.6453,318 "step": 440319 },320 {321 "epoch": 0.06535947712418301,322 "grad_norm": 1.5765402125957226,323 "learning_rate": 6.5166908563134976e-06,324 "loss": 0.6598,325 "step": 450326 },327 {328 "epoch": 0.06681190994916485,329 "grad_norm": 1.7349394887283642,330 "learning_rate": 6.6618287373004365e-06,331 "loss": 0.6619,332 "step": 460333 },334 {335 "epoch": 0.0682643427741467,336 "grad_norm": 1.6385635232751372,337 "learning_rate": 6.806966618287374e-06,338 "loss": 0.6692,339 "step": 470340 },341 {342 "epoch": 0.06971677559912855,343 "grad_norm": 1.4945507177883908,344 "learning_rate": 6.952104499274311e-06,345 "loss": 0.6484,346 "step": 480347 },348 {349 "epoch": 0.07116920842411038,350 "grad_norm": 1.583857774726375,351 "learning_rate": 7.097242380261249e-06,352 "loss": 0.657,353 "step": 490354 },355 {356 "epoch": 0.07262164124909223,357 "grad_norm": 1.8780189334850588,358 "learning_rate": 7.242380261248186e-06,359 "loss": 0.6601,360 "step": 500361 },362 {363 "epoch": 0.07407407407407407,364 "grad_norm": 1.5153409007972507,365 "learning_rate": 7.387518142235124e-06,366 "loss": 0.6542,367 "step": 510368 },369 {370 "epoch": 0.07552650689905592,371 "grad_norm": 1.5243833834622142,372 "learning_rate": 7.532656023222062e-06,373 "loss": 0.6476,374 "step": 520375 },376 {377 "epoch": 0.07697893972403776,378 "grad_norm": 1.6429693792028686,379 "learning_rate": 7.677793904208998e-06,380 "loss": 0.6451,381 "step": 530382 },383 {384 "epoch": 0.0784313725490196,385 "grad_norm": 1.802860360098263,386 "learning_rate": 7.822931785195936e-06,387 "loss": 0.6527,388 "step": 540389 },390 {391 "epoch": 0.07988380537400146,392 "grad_norm": 1.6594363957156038,393 "learning_rate": 7.968069666182874e-06,394 "loss": 0.661,395 "step": 550396 },397 {398 "epoch": 0.0813362381989833,399 "grad_norm": 1.5938255936259151,400 "learning_rate": 8.113207547169812e-06,401 "loss": 0.6547,402 "step": 560403 },404 {405 "epoch": 0.08278867102396514,406 "grad_norm": 1.3939924292770436,407 "learning_rate": 8.25834542815675e-06,408 "loss": 0.6609,409 "step": 570410 },411 {412 "epoch": 0.08424110384894698,413 "grad_norm": 1.5321796462771227,414 "learning_rate": 8.403483309143687e-06,415 "loss": 0.6419,416 "step": 580417 },418 {419 "epoch": 0.08569353667392883,420 "grad_norm": 1.5907007682060863,421 "learning_rate": 8.548621190130625e-06,422 "loss": 0.625,423 "step": 590424 },425 {426 "epoch": 0.08714596949891068,427 "grad_norm": 1.6048966671231157,428 "learning_rate": 8.693759071117563e-06,429 "loss": 0.658,430 "step": 600431 },432 {433 "epoch": 0.08859840232389252,434 "grad_norm": 1.457751877262412,435 "learning_rate": 8.8388969521045e-06,436 "loss": 0.6456,437 "step": 610438 },439 {440 "epoch": 0.09005083514887437,441 "grad_norm": 1.3925725985786772,442 "learning_rate": 8.984034833091438e-06,443 "loss": 0.6494,444 "step": 620445 },446 {447 "epoch": 0.0915032679738562,448 "grad_norm": 1.6476815627809678,449 "learning_rate": 9.129172714078376e-06,450 "loss": 0.6604,451 "step": 630452 },453 {454 "epoch": 0.09295570079883805,455 "grad_norm": 1.4844043302240553,456 "learning_rate": 9.274310595065312e-06,457 "loss": 0.6462,458 "step": 640459 },460 {461 "epoch": 0.0944081336238199,462 "grad_norm": 1.5541257847812342,463 "learning_rate": 9.41944847605225e-06,464 "loss": 0.6464,465 "step": 650466 },467 {468 "epoch": 0.09586056644880174,469 "grad_norm": 1.5339956751582804,470 "learning_rate": 9.564586357039188e-06,471 "loss": 0.6471,472 "step": 660473 },474 {475 "epoch": 0.09731299927378359,476 "grad_norm": 1.550006983868159,477 "learning_rate": 9.709724238026126e-06,478 "loss": 0.6519,479 "step": 670480 },481 {482 "epoch": 0.09876543209876543,483 "grad_norm": 1.298622779401985,484 "learning_rate": 9.854862119013063e-06,485 "loss": 0.6508,486 "step": 680487 },488 {489 "epoch": 0.10021786492374728,490 "grad_norm": 1.4545201677417376,491 "learning_rate": 1e-05,492 "loss": 0.6483,493 "step": 690494 },495 {496 "epoch": 0.10167029774872913,497 "grad_norm": 1.7514454450540817,498 "learning_rate": 9.999935728859667e-06,499 "loss": 0.6517,500 "step": 700501 },502 {503 "epoch": 0.10312273057371096,504 "grad_norm": 1.3010290416328456,505 "learning_rate": 9.999742917090981e-06,506 "loss": 0.6435,507 "step": 710508 },509 {510 "epoch": 0.10457516339869281,511 "grad_norm": 1.5222737445349914,512 "learning_rate": 9.999421569650833e-06,513 "loss": 0.6355,514 "step": 720515 },516 {517 "epoch": 0.10602759622367465,518 "grad_norm": 1.5758824439402839,519 "learning_rate": 9.99897169480057e-06,520 "loss": 0.6414,521 "step": 730522 },523 {524 "epoch": 0.1074800290486565,525 "grad_norm": 1.3245458819453462,526 "learning_rate": 9.99839330410578e-06,527 "loss": 0.6416,528 "step": 740529 },530 {531 "epoch": 0.10893246187363835,532 "grad_norm": 1.4753577499137038,533 "learning_rate": 9.997686412435996e-06,534 "loss": 0.6381,535 "step": 750536 },537 {538 "epoch": 0.11038489469862019,539 "grad_norm": 1.4578988593383,540 "learning_rate": 9.99685103796431e-06,541 "loss": 0.6369,542 "step": 760543 },544 {545 "epoch": 0.11183732752360204,546 "grad_norm": 1.389881220599468,547 "learning_rate": 9.99588720216691e-06,548 "loss": 0.6622,549 "step": 770550 },551 {552 "epoch": 0.11328976034858387,553 "grad_norm": 1.2318560606230133,554 "learning_rate": 9.994794929822527e-06,555 "loss": 0.6279,556 "step": 780557 },558 {559 "epoch": 0.11474219317356572,560 "grad_norm": 1.355472620629438,561 "learning_rate": 9.993574249011797e-06,562 "loss": 0.641,563 "step": 790564 },565 {566 "epoch": 0.11619462599854757,567 "grad_norm": 1.4379602146139996,568 "learning_rate": 9.992225191116538e-06,569 "loss": 0.6439,570 "step": 800571 },572 {573 "epoch": 0.11764705882352941,574 "grad_norm": 1.4777958226910466,575 "learning_rate": 9.990747790818946e-06,576 "loss": 0.6457,577 "step": 810578 },579 {580 "epoch": 0.11909949164851126,581 "grad_norm": 1.2895229336241503,582 "learning_rate": 9.989142086100703e-06,583 "loss": 0.6483,584 "step": 820585 },586 {587 "epoch": 0.1205519244734931,588 "grad_norm": 1.4811460587250382,589 "learning_rate": 9.987408118241995e-06,590 "loss": 0.6509,591 "step": 830592 },593 {594 "epoch": 0.12200435729847495,595 "grad_norm": 1.3189208191268318,596 "learning_rate": 9.985545931820463e-06,597 "loss": 0.6181,598 "step": 840599 },600 {601 "epoch": 0.12345679012345678,602 "grad_norm": 1.3731300368595278,603 "learning_rate": 9.983555574710043e-06,604 "loss": 0.6274,605 "step": 850606 },607 {608 "epoch": 0.12490922294843863,609 "grad_norm": 1.4055775942483093,610 "learning_rate": 9.981437098079743e-06,611 "loss": 0.6398,612 "step": 860613 },614 {615 "epoch": 0.12636165577342048,616 "grad_norm": 1.3307192435974602,617 "learning_rate": 9.979190556392326e-06,618 "loss": 0.6393,619 "step": 870620 },621 {622 "epoch": 0.12781408859840232,623 "grad_norm": 1.5622917958142868,624 "learning_rate": 9.976816007402912e-06,625 "loss": 0.6456,626 "step": 880627 },628 {629 "epoch": 0.12926652142338416,630 "grad_norm": 1.390636406480548,631 "learning_rate": 9.974313512157488e-06,632 "loss": 0.6288,633 "step": 890634 },635 {636 "epoch": 0.13071895424836602,637 "grad_norm": 1.4427250843896926,638 "learning_rate": 9.971683134991344e-06,639 "loss": 0.6266,640 "step": 900641 },642 {643 "epoch": 0.13217138707334786,644 "grad_norm": 1.4098179198178282,645 "learning_rate": 9.968924943527418e-06,646 "loss": 0.6411,647 "step": 910648 },649 {650 "epoch": 0.1336238198983297,651 "grad_norm": 1.4962238363929918,652 "learning_rate": 9.96603900867455e-06,653 "loss": 0.6315,654 "step": 920655 },656 {657 "epoch": 0.13507625272331156,658 "grad_norm": 1.3209044251278015,659 "learning_rate": 9.963025404625673e-06,660 "loss": 0.6423,661 "step": 930662 },663 {664 "epoch": 0.1365286855482934,665 "grad_norm": 1.39955503516968,666 "learning_rate": 9.959884208855893e-06,667 "loss": 0.6361,668 "step": 940669 },670 {671 "epoch": 0.13798111837327523,672 "grad_norm": 1.5348970475105241,673 "learning_rate": 9.956615502120504e-06,674 "loss": 0.6241,675 "step": 950676 },677 {678 "epoch": 0.1394335511982571,679 "grad_norm": 1.48874630945738,680 "learning_rate": 9.953219368452908e-06,681 "loss": 0.631,682 "step": 960683 },684 {685 "epoch": 0.14088598402323893,686 "grad_norm": 1.310857282598366,687 "learning_rate": 9.949695895162464e-06,688 "loss": 0.627,689 "step": 970690 },691 {692 "epoch": 0.14233841684822077,693 "grad_norm": 1.3619342578169393,694 "learning_rate": 9.946045172832224e-06,695 "loss": 0.6387,696 "step": 980697 },698 {699 "epoch": 0.1437908496732026,700 "grad_norm": 1.4936986486504984,701 "learning_rate": 9.942267295316625e-06,702 "loss": 0.6331,703 "step": 990704 },705 {706 "epoch": 0.14524328249818447,707 "grad_norm": 1.32511584393411,708 "learning_rate": 9.938362359739068e-06,709 "loss": 0.626,710 "step": 1000711 },712 {713 "epoch": 0.1466957153231663,714 "grad_norm": 1.3291454266011833,715 "learning_rate": 9.934330466489414e-06,716 "loss": 0.6451,717 "step": 1010718 },719 {720 "epoch": 0.14814814814814814,721 "grad_norm": 1.3289648153139675,722 "learning_rate": 9.930171719221418e-06,723 "loss": 0.6333,724 "step": 1020725 },726 {727 "epoch": 0.14960058097313,728 "grad_norm": 1.3388955314518605,729 "learning_rate": 9.925886224850047e-06,730 "loss": 0.6329,731 "step": 1030732 },733 {734 "epoch": 0.15105301379811184,735 "grad_norm": 1.3788458990043229,736 "learning_rate": 9.921474093548748e-06,737 "loss": 0.6308,738 "step": 1040739 },740 {741 "epoch": 0.15250544662309368,742 "grad_norm": 1.2630947233952987,743 "learning_rate": 9.916935438746604e-06,744 "loss": 0.6366,745 "step": 1050746 },747 {748 "epoch": 0.1539578794480755,749 "grad_norm": 1.2586848110727198,750 "learning_rate": 9.912270377125424e-06,751 "loss": 0.6224,752 "step": 1060753 },754 {755 "epoch": 0.15541031227305738,756 "grad_norm": 1.5648142512317709,757 "learning_rate": 9.90747902861674e-06,758 "loss": 0.6261,759 "step": 1070760 },761 {762 "epoch": 0.1568627450980392,763 "grad_norm": 1.477705850244199,764 "learning_rate": 9.902561516398723e-06,765 "loss": 0.6207,766 "step": 1080767 },768 {769 "epoch": 0.15831517792302105,770 "grad_norm": 1.2950681154644361,771 "learning_rate": 9.897517966893023e-06,772 "loss": 0.6218,773 "step": 1090774 },775 {776 "epoch": 0.15976761074800291,777 "grad_norm": 1.4613516139089748,778 "learning_rate": 9.892348509761509e-06,779 "loss": 0.6237,780 "step": 1100781 },782 {783 "epoch": 0.16122004357298475,784 "grad_norm": 1.2641419484176866,785 "learning_rate": 9.887053277902943e-06,786 "loss": 0.6425,787 "step": 1110788 },789 {790 "epoch": 0.1626724763979666,791 "grad_norm": 1.2419109246681843,792 "learning_rate": 9.881632407449561e-06,793 "loss": 0.6423,794 "step": 1120795 },796 {797 "epoch": 0.16412490922294845,798 "grad_norm": 1.4096648257937974,799 "learning_rate": 9.876086037763575e-06,800 "loss": 0.6383,801 "step": 1130802 },803 {804 "epoch": 0.1655773420479303,805 "grad_norm": 1.2574892255736747,806 "learning_rate": 9.870414311433585e-06,807 "loss": 0.6059,808 "step": 1140809 },810 {811 "epoch": 0.16702977487291212,812 "grad_norm": 1.2716145459010044,813 "learning_rate": 9.86461737427092e-06,814 "loss": 0.6098,815 "step": 1150816 },817 {818 "epoch": 0.16848220769789396,819 "grad_norm": 1.1998298755084313,820 "learning_rate": 9.858695375305885e-06,821 "loss": 0.6214,822 "step": 1160823 },824 {825 "epoch": 0.16993464052287582,826 "grad_norm": 1.4281449888166444,827 "learning_rate": 9.852648466783927e-06,828 "loss": 0.6241,829 "step": 1170830 },831 {832 "epoch": 0.17138707334785766,833 "grad_norm": 1.4071764477667867,834 "learning_rate": 9.84647680416173e-06,835 "loss": 0.6474,836 "step": 1180837 },838 {839 "epoch": 0.1728395061728395,840 "grad_norm": 1.2174453861834778,841 "learning_rate": 9.840180546103215e-06,842 "loss": 0.6326,843 "step": 1190844 },845 {846 "epoch": 0.17429193899782136,847 "grad_norm": 1.3029300772595094,848 "learning_rate": 9.833759854475453e-06,849 "loss": 0.6185,850 "step": 1200851 },852 {853 "epoch": 0.1757443718228032,854 "grad_norm": 1.271112016193465,855 "learning_rate": 9.827214894344514e-06,856 "loss": 0.6301,857 "step": 1210858 },859 {860 "epoch": 0.17719680464778503,861 "grad_norm": 1.2997276991719462,862 "learning_rate": 9.82054583397122e-06,863 "loss": 0.6317,864 "step": 1220865 },866 {867 "epoch": 0.1786492374727669,868 "grad_norm": 1.2096030387104992,869 "learning_rate": 9.813752844806814e-06,870 "loss": 0.6159,871 "step": 1230872 },873 {874 "epoch": 0.18010167029774873,875 "grad_norm": 1.2973416257944899,876 "learning_rate": 9.806836101488561e-06,877 "loss": 0.6289,878 "step": 1240879 },880 {881 "epoch": 0.18155410312273057,882 "grad_norm": 1.3197440048632956,883 "learning_rate": 9.799795781835253e-06,884 "loss": 0.6088,885 "step": 1250886 },887 {888 "epoch": 0.1830065359477124,889 "grad_norm": 1.2535036782710556,890 "learning_rate": 9.79263206684264e-06,891 "loss": 0.6206,892 "step": 1260893 },894 {895 "epoch": 0.18445896877269427,896 "grad_norm": 1.3190252094745194,897 "learning_rate": 9.785345140678775e-06,898 "loss": 0.6149,899 "step": 1270900 },901 {902 "epoch": 0.1859114015976761,903 "grad_norm": 1.3148617882447478,904 "learning_rate": 9.777935190679277e-06,905 "loss": 0.6134,906 "step": 1280907 },908 {909 "epoch": 0.18736383442265794,910 "grad_norm": 1.3368521794263946,911 "learning_rate": 9.770402407342524e-06,912 "loss": 0.6258,913 "step": 1290914 },915 {916 "epoch": 0.1888162672476398,917 "grad_norm": 1.3941700458180073,918 "learning_rate": 9.762746984324743e-06,919 "loss": 0.6191,920 "step": 1300921 },922 {923 "epoch": 0.19026870007262164,924 "grad_norm": 1.3152403546822757,925 "learning_rate": 9.754969118435043e-06,926 "loss": 0.6446,927 "step": 1310928 },929 {930 "epoch": 0.19172113289760348,931 "grad_norm": 1.3013626770341264,932 "learning_rate": 9.747069009630347e-06,933 "loss": 0.6312,934 "step": 1320935 },936 {937 "epoch": 0.19317356572258534,938 "grad_norm": 1.3966383885583535,939 "learning_rate": 9.739046861010255e-06,940 "loss": 0.6207,941 "step": 1330942 },943 {944 "epoch": 0.19462599854756718,945 "grad_norm": 1.1439991746974036,946 "learning_rate": 9.730902878811825e-06,947 "loss": 0.6144,948 "step": 1340949 },950 {951 "epoch": 0.19607843137254902,952 "grad_norm": 1.3540894709055364,953 "learning_rate": 9.722637272404263e-06,954 "loss": 0.6044,955 "step": 1350956 },957 {958 "epoch": 0.19753086419753085,959 "grad_norm": 1.100639588271217,960 "learning_rate": 9.71425025428355e-06,961 "loss": 0.6036,962 "step": 1360963 },964 {965 "epoch": 0.19898329702251272,966 "grad_norm": 1.1874319432290736,967 "learning_rate": 9.705742040066977e-06,968 "loss": 0.6039,969 "step": 1370970 },971 {972 "epoch": 0.20043572984749455,973 "grad_norm": 1.1767671647303808,974 "learning_rate": 9.697112848487591e-06,975 "loss": 0.6376,976 "step": 1380977 },978 {979 "epoch": 0.2018881626724764,980 "grad_norm": 1.135879944041461,981 "learning_rate": 9.688362901388586e-06,982 "loss": 0.6035,983 "step": 1390984 },985 {986 "epoch": 0.20334059549745825,987 "grad_norm": 1.2315910796359388,988 "learning_rate": 9.679492423717596e-06,989 "loss": 0.6098,990 "step": 1400991 },992 {993 "epoch": 0.2047930283224401,994 "grad_norm": 1.4949408462288012,995 "learning_rate": 9.670501643520904e-06,996 "loss": 0.6203,997 "step": 1410998 },999 {1000 "epoch": 0.20624546114742193,1001 "grad_norm": 1.3180181445795711,1002 "learning_rate": 9.66139079193759e-06,1003 "loss": 0.6286,1004 "step": 14201005 },1006 {1007 "epoch": 0.20769789397240376,1008 "grad_norm": 1.2616556885045909,1009 "learning_rate": 9.652160103193583e-06,1010 "loss": 0.6274,1011 "step": 14301012 },1013 {1014 "epoch": 0.20915032679738563,1015 "grad_norm": 1.3174449455574337,1016 "learning_rate": 9.642809814595637e-06,1017 "loss": 0.6136,1018 "step": 14401019 },1020 {1021 "epoch": 0.21060275962236746,1022 "grad_norm": 1.296735377133819,1023 "learning_rate": 9.633340166525238e-06,1024 "loss": 0.6145,1025 "step": 14501026 },1027 {1028 "epoch": 0.2120551924473493,1029 "grad_norm": 1.2502497833244608,1030 "learning_rate": 9.62375140243242e-06,1031 "loss": 0.6031,1032 "step": 14601033 },1034 {1035 "epoch": 0.21350762527233116,1036 "grad_norm": 1.2288830705505374,1037 "learning_rate": 9.6140437688295e-06,1038 "loss": 0.6128,1039 "step": 14701040 },1041 {1042 "epoch": 0.214960058097313,1043 "grad_norm": 1.1119473380240397,1044 "learning_rate": 9.604217515284753e-06,1045 "loss": 0.6171,1046 "step": 14801047 },1048 {1049 "epoch": 0.21641249092229484,1050 "grad_norm": 1.2070397164389806,1051 "learning_rate": 9.594272894415986e-06,1052 "loss": 0.6238,1053 "step": 14901054 },1055 {1056 "epoch": 0.2178649237472767,1057 "grad_norm": 1.3345637205372078,1058 "learning_rate": 9.584210161884049e-06,1059 "loss": 0.6163,1060 "step": 15001061 },1062 {1063 "epoch": 0.21931735657225854,1064 "grad_norm": 1.1385043759036517,1065 "learning_rate": 9.57402957638626e-06,1066 "loss": 0.6083,1067 "step": 15101068 },1069 {1070 "epoch": 0.22076978939724037,1071 "grad_norm": 1.1936988121465326,1072 "learning_rate": 9.563731399649756e-06,1073 "loss": 0.5992,1074 "step": 15201075 },1076 {1077 "epoch": 0.2222222222222222,1078 "grad_norm": 1.4103572503621762,1079 "learning_rate": 9.553315896424758e-06,1080 "loss": 0.6054,1081 "step": 15301082 },1083 {1084 "epoch": 0.22367465504720407,1085 "grad_norm": 1.3209719950503893,1086 "learning_rate": 9.54278333447778e-06,1087 "loss": 0.596,1088 "step": 15401089 },1090 {1091 "epoch": 0.2251270878721859,1092 "grad_norm": 1.1693016501696898,1093 "learning_rate": 9.532133984584721e-06,1094 "loss": 0.6323,1095 "step": 15501096 },1097 {1098 "epoch": 0.22657952069716775,1099 "grad_norm": 1.1691510921859125,1100 "learning_rate": 9.521368120523931e-06,1101 "loss": 0.6027,1102 "step": 15601103 },1104 {1105 "epoch": 0.2280319535221496,1106 "grad_norm": 1.2114364957172101,1107 "learning_rate": 9.510486019069154e-06,1108 "loss": 0.6245,1109 "step": 15701110 },1111 {1112 "epoch": 0.22948438634713145,1113 "grad_norm": 1.265123327235345,1114 "learning_rate": 9.499487959982415e-06,1115 "loss": 0.6189,1116 "step": 15801117 },1118 {1119 "epoch": 0.23093681917211328,1120 "grad_norm": 1.3773059483594046,1121 "learning_rate": 9.488374226006836e-06,1122 "loss": 0.6106,1123 "step": 15901124 },1125 {1126 "epoch": 0.23238925199709515,1127 "grad_norm": 1.2737618179619303,1128 "learning_rate": 9.477145102859357e-06,1129 "loss": 0.6115,1130 "step": 16001131 },1132 {1133 "epoch": 0.23384168482207698,1134 "grad_norm": 1.3066121502077,1135 "learning_rate": 9.4658008792234e-06,1136 "loss": 0.609,1137 "step": 16101138 },1139 {1140 "epoch": 0.23529411764705882,1141 "grad_norm": 1.242518893517758,1142 "learning_rate": 9.45434184674144e-06,1143 "loss": 0.6,1144 "step": 16201145 },1146 {1147 "epoch": 0.23674655047204066,1148 "grad_norm": 1.2493334973003818,1149 "learning_rate": 9.442768300007511e-06,1150 "loss": 0.6144,1151 "step": 16301152 },1153 {1154 "epoch": 0.23819898329702252,1155 "grad_norm": 1.2775874117960886,1156 "learning_rate": 9.431080536559631e-06,1157 "loss": 0.6245,1158 "step": 16401159 },1160 {1161 "epoch": 0.23965141612200436,1162 "grad_norm": 1.247039996382283,1163 "learning_rate": 9.419278856872154e-06,1164 "loss": 0.6279,1165 "step": 16501166 },1167 {1168 "epoch": 0.2411038489469862,1169 "grad_norm": 1.302601682600637,1170 "learning_rate": 9.407363564348047e-06,1171 "loss": 0.5933,1172 "step": 16601173 },1174 {1175 "epoch": 0.24255628177196806,1176 "grad_norm": 1.431347455463815,1177 "learning_rate": 9.39533496531108e-06,1178 "loss": 0.6171,1179 "step": 16701180 },1181 {1182 "epoch": 0.2440087145969499,1183 "grad_norm": 1.2527655662771335,1184 "learning_rate": 9.38319336899797e-06,1185 "loss": 0.6099,1186 "step": 16801187 },1188 {1189 "epoch": 0.24546114742193173,1190 "grad_norm": 1.205551788839019,1191 "learning_rate": 9.370939087550407e-06,1192 "loss": 0.6077,1193 "step": 16901194 },1195 {1196 "epoch": 0.24691358024691357,1197 "grad_norm": 1.332981320431861,1198 "learning_rate": 9.358572436007052e-06,1199 "loss": 0.6126,1200 "step": 1700