MLAdaptiveIntelligence/LLaVAction-7B
196
1{2 "best_metric": null,3 "best_model_checkpoint": null,4 "epoch": 0.9999304928060054,5 "eval_steps": 1.0,6 "global_step": 7193,7 "is_hyper_param_search": false,8 "is_local_process_zero": true,9 "is_world_process_zero": true,10 "log_history": [11 {12 "epoch": 0.0,13 "grad_norm": 10.255854128827643,14 "learning_rate": 4.6296296296296295e-08,15 "loss": 0.7089,16 "step": 117 },18 {19 "epoch": 0.0,20 "grad_norm": 8.773792360679519,21 "learning_rate": 9.259259259259259e-08,22 "loss": 0.6414,23 "step": 224 },25 {26 "epoch": 0.0,27 "grad_norm": 10.454717029458765,28 "learning_rate": 1.3888888888888888e-07,29 "loss": 0.6782,30 "step": 331 },32 {33 "epoch": 0.0,34 "grad_norm": 9.402114935002771,35 "learning_rate": 1.8518518518518518e-07,36 "loss": 0.635,37 "step": 438 },39 {40 "epoch": 0.0,41 "grad_norm": 9.169183983416053,42 "learning_rate": 2.3148148148148148e-07,43 "loss": 0.6219,44 "step": 545 },46 {47 "epoch": 0.0,48 "grad_norm": 10.036191772486438,49 "learning_rate": 2.7777777777777776e-07,50 "loss": 0.6553,51 "step": 652 },53 {54 "epoch": 0.0,55 "grad_norm": 11.965713108861875,56 "learning_rate": 3.2407407407407406e-07,57 "loss": 0.7664,58 "step": 759 },60 {61 "epoch": 0.0,62 "grad_norm": 11.154265911383941,63 "learning_rate": 3.7037037037037036e-07,64 "loss": 0.6945,65 "step": 866 },67 {68 "epoch": 0.0,69 "grad_norm": 10.68626688318975,70 "learning_rate": 4.1666666666666667e-07,71 "loss": 0.7317,72 "step": 973 },74 {75 "epoch": 0.0,76 "grad_norm": 13.399415302823204,77 "learning_rate": 4.6296296296296297e-07,78 "loss": 0.6778,79 "step": 1080 },81 {82 "epoch": 0.0,83 "grad_norm": 8.212753191901033,84 "learning_rate": 5.092592592592593e-07,85 "loss": 0.63,86 "step": 1187 },88 {89 "epoch": 0.0,90 "grad_norm": 7.796878855920977,91 "learning_rate": 5.555555555555555e-07,92 "loss": 0.6252,93 "step": 1294 },95 {96 "epoch": 0.0,97 "grad_norm": 11.97829846200218,98 "learning_rate": 6.018518518518519e-07,99 "loss": 0.6076,100 "step": 13101 },102 {103 "epoch": 0.0,104 "grad_norm": 8.612821227689986,105 "learning_rate": 6.481481481481481e-07,106 "loss": 0.6745,107 "step": 14108 },109 {110 "epoch": 0.0,111 "grad_norm": 21.00893106426534,112 "learning_rate": 6.944444444444446e-07,113 "loss": 0.6036,114 "step": 15115 },116 {117 "epoch": 0.0,118 "grad_norm": 9.713683569034302,119 "learning_rate": 7.407407407407407e-07,120 "loss": 0.5631,121 "step": 16122 },123 {124 "epoch": 0.0,125 "grad_norm": 14.72722347265912,126 "learning_rate": 7.870370370370371e-07,127 "loss": 0.5615,128 "step": 17129 },130 {131 "epoch": 0.0,132 "grad_norm": 7.544394214922187,133 "learning_rate": 8.333333333333333e-07,134 "loss": 0.609,135 "step": 18136 },137 {138 "epoch": 0.0,139 "grad_norm": 9.523364730476793,140 "learning_rate": 8.796296296296297e-07,141 "loss": 0.5802,142 "step": 19143 },144 {145 "epoch": 0.0,146 "grad_norm": 5.419257987223273,147 "learning_rate": 9.259259259259259e-07,148 "loss": 0.5041,149 "step": 20150 },151 {152 "epoch": 0.0,153 "grad_norm": 9.77850446169523,154 "learning_rate": 9.722222222222224e-07,155 "loss": 0.5525,156 "step": 21157 },158 {159 "epoch": 0.0,160 "grad_norm": 4.252949025507043,161 "learning_rate": 1.0185185185185185e-06,162 "loss": 0.4768,163 "step": 22164 },165 {166 "epoch": 0.0,167 "grad_norm": 4.7101311808953685,168 "learning_rate": 1.0648148148148149e-06,169 "loss": 0.4981,170 "step": 23171 },172 {173 "epoch": 0.0,174 "grad_norm": 4.53366283698979,175 "learning_rate": 1.111111111111111e-06,176 "loss": 0.5176,177 "step": 24178 },179 {180 "epoch": 0.0,181 "grad_norm": 3.763232944709757,182 "learning_rate": 1.1574074074074076e-06,183 "loss": 0.5139,184 "step": 25185 },186 {187 "epoch": 0.0,188 "grad_norm": 3.7091692609708717,189 "learning_rate": 1.2037037037037037e-06,190 "loss": 0.4981,191 "step": 26192 },193 {194 "epoch": 0.0,195 "grad_norm": 15.016693375984712,196 "learning_rate": 1.25e-06,197 "loss": 0.4577,198 "step": 27199 },200 {201 "epoch": 0.0,202 "grad_norm": 3.6102044253176437,203 "learning_rate": 1.2962962962962962e-06,204 "loss": 0.4989,205 "step": 28206 },207 {208 "epoch": 0.0,209 "grad_norm": 3.078894895654288,210 "learning_rate": 1.3425925925925928e-06,211 "loss": 0.4747,212 "step": 29213 },214 {215 "epoch": 0.0,216 "grad_norm": 2.278891574726967,217 "learning_rate": 1.3888888888888892e-06,218 "loss": 0.4109,219 "step": 30220 },221 {222 "epoch": 0.0,223 "grad_norm": 2.317052004327827,224 "learning_rate": 1.4351851851851853e-06,225 "loss": 0.447,226 "step": 31227 },228 {229 "epoch": 0.0,230 "grad_norm": 2.4476257193527147,231 "learning_rate": 1.4814814814814815e-06,232 "loss": 0.4385,233 "step": 32234 },235 {236 "epoch": 0.0,237 "grad_norm": 3.8040634784992315,238 "learning_rate": 1.527777777777778e-06,239 "loss": 0.435,240 "step": 33241 },242 {243 "epoch": 0.0,244 "grad_norm": 2.610119369335003,245 "learning_rate": 1.5740740740740742e-06,246 "loss": 0.4321,247 "step": 34248 },249 {250 "epoch": 0.0,251 "grad_norm": 2.6263190952334807,252 "learning_rate": 1.6203703703703705e-06,253 "loss": 0.4192,254 "step": 35255 },256 {257 "epoch": 0.01,258 "grad_norm": 3.424885660562162,259 "learning_rate": 1.6666666666666667e-06,260 "loss": 0.4613,261 "step": 36262 },263 {264 "epoch": 0.01,265 "grad_norm": 3.3561962804013494,266 "learning_rate": 1.7129629629629632e-06,267 "loss": 0.458,268 "step": 37269 },270 {271 "epoch": 0.01,272 "grad_norm": 2.520456771934489,273 "learning_rate": 1.7592592592592594e-06,274 "loss": 0.4439,275 "step": 38276 },277 {278 "epoch": 0.01,279 "grad_norm": 3.871885426391512,280 "learning_rate": 1.8055555555555557e-06,281 "loss": 0.4463,282 "step": 39283 },284 {285 "epoch": 0.01,286 "grad_norm": 2.592313829601372,287 "learning_rate": 1.8518518518518519e-06,288 "loss": 0.4371,289 "step": 40290 },291 {292 "epoch": 0.01,293 "grad_norm": 3.5345587197376505,294 "learning_rate": 1.8981481481481484e-06,295 "loss": 0.4288,296 "step": 41297 },298 {299 "epoch": 0.01,300 "grad_norm": 3.2766287781088375,301 "learning_rate": 1.944444444444445e-06,302 "loss": 0.4439,303 "step": 42304 },305 {306 "epoch": 0.01,307 "grad_norm": 2.4974698598721234,308 "learning_rate": 1.9907407407407407e-06,309 "loss": 0.4335,310 "step": 43311 },312 {313 "epoch": 0.01,314 "grad_norm": 2.446775546130796,315 "learning_rate": 2.037037037037037e-06,316 "loss": 0.4276,317 "step": 44318 },319 {320 "epoch": 0.01,321 "grad_norm": 3.610162990317946,322 "learning_rate": 2.0833333333333334e-06,323 "loss": 0.418,324 "step": 45325 },326 {327 "epoch": 0.01,328 "grad_norm": 2.4439951061940675,329 "learning_rate": 2.1296296296296298e-06,330 "loss": 0.4454,331 "step": 46332 },333 {334 "epoch": 0.01,335 "grad_norm": 3.4040054429684474,336 "learning_rate": 2.175925925925926e-06,337 "loss": 0.4367,338 "step": 47339 },340 {341 "epoch": 0.01,342 "grad_norm": 2.312735050532508,343 "learning_rate": 2.222222222222222e-06,344 "loss": 0.4265,345 "step": 48346 },347 {348 "epoch": 0.01,349 "grad_norm": 2.6957192931244576,350 "learning_rate": 2.268518518518519e-06,351 "loss": 0.4032,352 "step": 49353 },354 {355 "epoch": 0.01,356 "grad_norm": 2.598984536491883,357 "learning_rate": 2.314814814814815e-06,358 "loss": 0.4118,359 "step": 50360 },361 {362 "epoch": 0.01,363 "grad_norm": 2.5537662533133347,364 "learning_rate": 2.361111111111111e-06,365 "loss": 0.3956,366 "step": 51367 },368 {369 "epoch": 0.01,370 "grad_norm": 2.6266154138590196,371 "learning_rate": 2.4074074074074075e-06,372 "loss": 0.4259,373 "step": 52374 },375 {376 "epoch": 0.01,377 "grad_norm": 3.477042477617357,378 "learning_rate": 2.453703703703704e-06,379 "loss": 0.4311,380 "step": 53381 },382 {383 "epoch": 0.01,384 "grad_norm": 2.985005093220916,385 "learning_rate": 2.5e-06,386 "loss": 0.468,387 "step": 54388 },389 {390 "epoch": 0.01,391 "grad_norm": 2.8501633741813093,392 "learning_rate": 2.5462962962962966e-06,393 "loss": 0.4334,394 "step": 55395 },396 {397 "epoch": 0.01,398 "grad_norm": 2.731612824284174,399 "learning_rate": 2.5925925925925925e-06,400 "loss": 0.4044,401 "step": 56402 },403 {404 "epoch": 0.01,405 "grad_norm": 2.635429910973552,406 "learning_rate": 2.6388888888888893e-06,407 "loss": 0.4477,408 "step": 57409 },410 {411 "epoch": 0.01,412 "grad_norm": 2.315429583382698,413 "learning_rate": 2.6851851851851856e-06,414 "loss": 0.4545,415 "step": 58416 },417 {418 "epoch": 0.01,419 "grad_norm": 2.8322617435642976,420 "learning_rate": 2.7314814814814816e-06,421 "loss": 0.416,422 "step": 59423 },424 {425 "epoch": 0.01,426 "grad_norm": 2.7183081042824786,427 "learning_rate": 2.7777777777777783e-06,428 "loss": 0.3938,429 "step": 60430 },431 {432 "epoch": 0.01,433 "grad_norm": 2.5562819747238783,434 "learning_rate": 2.8240740740740743e-06,435 "loss": 0.3987,436 "step": 61437 },438 {439 "epoch": 0.01,440 "grad_norm": 3.6937095292787854,441 "learning_rate": 2.8703703703703706e-06,442 "loss": 0.4562,443 "step": 62444 },445 {446 "epoch": 0.01,447 "grad_norm": 2.606990917862641,448 "learning_rate": 2.916666666666667e-06,449 "loss": 0.411,450 "step": 63451 },452 {453 "epoch": 0.01,454 "grad_norm": 2.983323777143174,455 "learning_rate": 2.962962962962963e-06,456 "loss": 0.4274,457 "step": 64458 },459 {460 "epoch": 0.01,461 "grad_norm": 2.5130825647128834,462 "learning_rate": 3.0092592592592597e-06,463 "loss": 0.4109,464 "step": 65465 },466 {467 "epoch": 0.01,468 "grad_norm": 4.471999264480983,469 "learning_rate": 3.055555555555556e-06,470 "loss": 0.4265,471 "step": 66472 },473 {474 "epoch": 0.01,475 "grad_norm": 2.6530405813156763,476 "learning_rate": 3.101851851851852e-06,477 "loss": 0.4201,478 "step": 67479 },480 {481 "epoch": 0.01,482 "grad_norm": 2.124624576669715,483 "learning_rate": 3.1481481481481483e-06,484 "loss": 0.4066,485 "step": 68486 },487 {488 "epoch": 0.01,489 "grad_norm": 2.630607085255591,490 "learning_rate": 3.1944444444444443e-06,491 "loss": 0.3945,492 "step": 69493 },494 {495 "epoch": 0.01,496 "grad_norm": 2.845874367230957,497 "learning_rate": 3.240740740740741e-06,498 "loss": 0.4325,499 "step": 70500 },501 {502 "epoch": 0.01,503 "grad_norm": 8.069961740943874,504 "learning_rate": 3.2870370370370374e-06,505 "loss": 0.4038,506 "step": 71507 },508 {509 "epoch": 0.01,510 "grad_norm": 2.488692512085441,511 "learning_rate": 3.3333333333333333e-06,512 "loss": 0.4638,513 "step": 72514 },515 {516 "epoch": 0.01,517 "grad_norm": 3.0371942256745017,518 "learning_rate": 3.37962962962963e-06,519 "loss": 0.4097,520 "step": 73521 },522 {523 "epoch": 0.01,524 "grad_norm": 3.220511241197115,525 "learning_rate": 3.4259259259259265e-06,526 "loss": 0.4003,527 "step": 74528 },529 {530 "epoch": 0.01,531 "grad_norm": 3.633152204811553,532 "learning_rate": 3.4722222222222224e-06,533 "loss": 0.4464,534 "step": 75535 },536 {537 "epoch": 0.01,538 "grad_norm": 3.4881085145784416,539 "learning_rate": 3.5185185185185187e-06,540 "loss": 0.4104,541 "step": 76542 },543 {544 "epoch": 0.01,545 "grad_norm": 2.7328395195041746,546 "learning_rate": 3.5648148148148147e-06,547 "loss": 0.4023,548 "step": 77549 },550 {551 "epoch": 0.01,552 "grad_norm": 2.952179328952469,553 "learning_rate": 3.6111111111111115e-06,554 "loss": 0.4107,555 "step": 78556 },557 {558 "epoch": 0.01,559 "grad_norm": 3.0145433355840354,560 "learning_rate": 3.657407407407408e-06,561 "loss": 0.3828,562 "step": 79563 },564 {565 "epoch": 0.01,566 "grad_norm": 3.333631472311848,567 "learning_rate": 3.7037037037037037e-06,568 "loss": 0.4452,569 "step": 80570 },571 {572 "epoch": 0.01,573 "grad_norm": 2.69753519175676,574 "learning_rate": 3.7500000000000005e-06,575 "loss": 0.4117,576 "step": 81577 },578 {579 "epoch": 0.01,580 "grad_norm": 3.21176087609264,581 "learning_rate": 3.796296296296297e-06,582 "loss": 0.3869,583 "step": 82584 },585 {586 "epoch": 0.01,587 "grad_norm": 2.4890191961757817,588 "learning_rate": 3.842592592592592e-06,589 "loss": 0.4136,590 "step": 83591 },592 {593 "epoch": 0.01,594 "grad_norm": 2.578718140563557,595 "learning_rate": 3.88888888888889e-06,596 "loss": 0.4368,597 "step": 84598 },599 {600 "epoch": 0.01,601 "grad_norm": 2.466958397571284,602 "learning_rate": 3.935185185185186e-06,603 "loss": 0.4031,604 "step": 85605 },606 {607 "epoch": 0.01,608 "grad_norm": 2.8900685201894576,609 "learning_rate": 3.9814814814814814e-06,610 "loss": 0.4191,611 "step": 86612 },613 {614 "epoch": 0.01,615 "grad_norm": 3.3686680325922955,616 "learning_rate": 4.027777777777779e-06,617 "loss": 0.3907,618 "step": 87619 },620 {621 "epoch": 0.01,622 "grad_norm": 2.7663195782363736,623 "learning_rate": 4.074074074074074e-06,624 "loss": 0.4172,625 "step": 88626 },627 {628 "epoch": 0.01,629 "grad_norm": 2.3780496870898458,630 "learning_rate": 4.1203703703703705e-06,631 "loss": 0.4046,632 "step": 89633 },634 {635 "epoch": 0.01,636 "grad_norm": 3.3868434764372934,637 "learning_rate": 4.166666666666667e-06,638 "loss": 0.4036,639 "step": 90640 },641 {642 "epoch": 0.01,643 "grad_norm": 2.7503898380791716,644 "learning_rate": 4.212962962962963e-06,645 "loss": 0.4393,646 "step": 91647 },648 {649 "epoch": 0.01,650 "grad_norm": 2.4175323073024435,651 "learning_rate": 4.2592592592592596e-06,652 "loss": 0.3995,653 "step": 92654 },655 {656 "epoch": 0.01,657 "grad_norm": 3.2598709968883535,658 "learning_rate": 4.305555555555556e-06,659 "loss": 0.3756,660 "step": 93661 },662 {663 "epoch": 0.01,664 "grad_norm": 2.464987650717689,665 "learning_rate": 4.351851851851852e-06,666 "loss": 0.3961,667 "step": 94668 },669 {670 "epoch": 0.01,671 "grad_norm": 5.0200344520961195,672 "learning_rate": 4.398148148148149e-06,673 "loss": 0.4341,674 "step": 95675 },676 {677 "epoch": 0.01,678 "grad_norm": 3.3934409076301946,679 "learning_rate": 4.444444444444444e-06,680 "loss": 0.3803,681 "step": 96682 },683 {684 "epoch": 0.01,685 "grad_norm": 2.0939684034470716,686 "learning_rate": 4.490740740740741e-06,687 "loss": 0.4096,688 "step": 97689 },690 {691 "epoch": 0.01,692 "grad_norm": 3.3783022516744,693 "learning_rate": 4.537037037037038e-06,694 "loss": 0.4205,695 "step": 98696 },697 {698 "epoch": 0.01,699 "grad_norm": 5.413801759638468,700 "learning_rate": 4.583333333333333e-06,701 "loss": 0.3975,702 "step": 99703 },704 {705 "epoch": 0.01,706 "grad_norm": 2.7398280657726053,707 "learning_rate": 4.62962962962963e-06,708 "loss": 0.4207,709 "step": 100710 },711 {712 "epoch": 0.01,713 "grad_norm": 2.5402510369008784,714 "learning_rate": 4.675925925925927e-06,715 "loss": 0.409,716 "step": 101717 },718 {719 "epoch": 0.01,720 "grad_norm": 2.2447264332340486,721 "learning_rate": 4.722222222222222e-06,722 "loss": 0.396,723 "step": 102724 },725 {726 "epoch": 0.01,727 "grad_norm": 3.55441114778751,728 "learning_rate": 4.768518518518519e-06,729 "loss": 0.4089,730 "step": 103731 },732 {733 "epoch": 0.01,734 "grad_norm": 4.07838457524117,735 "learning_rate": 4.814814814814815e-06,736 "loss": 0.3708,737 "step": 104738 },739 {740 "epoch": 0.01,741 "grad_norm": 2.7851693491114786,742 "learning_rate": 4.861111111111111e-06,743 "loss": 0.4069,744 "step": 105745 },746 {747 "epoch": 0.01,748 "grad_norm": 3.230069779103466,749 "learning_rate": 4.907407407407408e-06,750 "loss": 0.4133,751 "step": 106752 },753 {754 "epoch": 0.01,755 "grad_norm": 2.6078490758442214,756 "learning_rate": 4.953703703703704e-06,757 "loss": 0.3846,758 "step": 107759 },760 {761 "epoch": 0.02,762 "grad_norm": 3.145857591672062,763 "learning_rate": 5e-06,764 "loss": 0.3946,765 "step": 108766 },767 {768 "epoch": 0.02,769 "grad_norm": 2.860373829007912,770 "learning_rate": 5.046296296296297e-06,771 "loss": 0.4048,772 "step": 109773 },774 {775 "epoch": 0.02,776 "grad_norm": 5.181292163771113,777 "learning_rate": 5.092592592592593e-06,778 "loss": 0.3778,779 "step": 110780 },781 {782 "epoch": 0.02,783 "grad_norm": 2.2676312338693894,784 "learning_rate": 5.138888888888889e-06,785 "loss": 0.3934,786 "step": 111787 },788 {789 "epoch": 0.02,790 "grad_norm": 2.5234692901209823,791 "learning_rate": 5.185185185185185e-06,792 "loss": 0.4269,793 "step": 112794 },795 {796 "epoch": 0.02,797 "grad_norm": 2.5337725988428934,798 "learning_rate": 5.231481481481482e-06,799 "loss": 0.4008,800 "step": 113801 },802 {803 "epoch": 0.02,804 "grad_norm": 2.608852953085008,805 "learning_rate": 5.2777777777777785e-06,806 "loss": 0.4367,807 "step": 114808 },809 {810 "epoch": 0.02,811 "grad_norm": 3.2355246857109408,812 "learning_rate": 5.324074074074075e-06,813 "loss": 0.4185,814 "step": 115815 },816 {817 "epoch": 0.02,818 "grad_norm": 2.837771455408418,819 "learning_rate": 5.370370370370371e-06,820 "loss": 0.4104,821 "step": 116822 },823 {824 "epoch": 0.02,825 "grad_norm": 3.383503170967786,826 "learning_rate": 5.416666666666667e-06,827 "loss": 0.4197,828 "step": 117829 },830 {831 "epoch": 0.02,832 "grad_norm": 4.397197479442109,833 "learning_rate": 5.462962962962963e-06,834 "loss": 0.3684,835 "step": 118836 },837 {838 "epoch": 0.02,839 "grad_norm": 3.1374031884442624,840 "learning_rate": 5.5092592592592595e-06,841 "loss": 0.4094,842 "step": 119843 },844 {845 "epoch": 0.02,846 "grad_norm": 3.543344757486579,847 "learning_rate": 5.555555555555557e-06,848 "loss": 0.4067,849 "step": 120850 },851 {852 "epoch": 0.02,853 "grad_norm": 2.1863786999376,854 "learning_rate": 5.601851851851853e-06,855 "loss": 0.3793,856 "step": 121857 },858 {859 "epoch": 0.02,860 "grad_norm": 2.2217766894286863,861 "learning_rate": 5.6481481481481485e-06,862 "loss": 0.3809,863 "step": 122864 },865 {866 "epoch": 0.02,867 "grad_norm": 6.544131417153299,868 "learning_rate": 5.694444444444445e-06,869 "loss": 0.4043,870 "step": 123871 },872 {873 "epoch": 0.02,874 "grad_norm": 3.0111130933671273,875 "learning_rate": 5.740740740740741e-06,876 "loss": 0.3987,877 "step": 124878 },879 {880 "epoch": 0.02,881 "grad_norm": 3.450861166362308,882 "learning_rate": 5.787037037037038e-06,883 "loss": 0.402,884 "step": 125885 },886 {887 "epoch": 0.02,888 "grad_norm": 2.90664853284843,889 "learning_rate": 5.833333333333334e-06,890 "loss": 0.3794,891 "step": 126892 },893 {894 "epoch": 0.02,895 "grad_norm": 3.7115612906279334,896 "learning_rate": 5.8796296296296295e-06,897 "loss": 0.3958,898 "step": 127899 },900 {901 "epoch": 0.02,902 "grad_norm": 2.8251651795903983,903 "learning_rate": 5.925925925925926e-06,904 "loss": 0.4155,905 "step": 128906 },907 {908 "epoch": 0.02,909 "grad_norm": 3.536720329943924,910 "learning_rate": 5.972222222222222e-06,911 "loss": 0.4018,912 "step": 129913 },914 {915 "epoch": 0.02,916 "grad_norm": 3.095105241434553,917 "learning_rate": 6.018518518518519e-06,918 "loss": 0.4101,919 "step": 130920 },921 {922 "epoch": 0.02,923 "grad_norm": 2.3863804556923283,924 "learning_rate": 6.064814814814816e-06,925 "loss": 0.3665,926 "step": 131927 },928 {929 "epoch": 0.02,930 "grad_norm": 2.12988678036126,931 "learning_rate": 6.111111111111112e-06,932 "loss": 0.4129,933 "step": 132934 },935 {936 "epoch": 0.02,937 "grad_norm": 2.6339230413996138,938 "learning_rate": 6.157407407407408e-06,939 "loss": 0.3602,940 "step": 133941 },942 {943 "epoch": 0.02,944 "grad_norm": 3.086883649805019,945 "learning_rate": 6.203703703703704e-06,946 "loss": 0.3925,947 "step": 134948 },949 {950 "epoch": 0.02,951 "grad_norm": 2.8630881890034803,952 "learning_rate": 6.25e-06,953 "loss": 0.3991,954 "step": 135955 },956 {957 "epoch": 0.02,958 "grad_norm": 5.832830470832314,959 "learning_rate": 6.296296296296297e-06,960 "loss": 0.3822,961 "step": 136962 },963 {964 "epoch": 0.02,965 "grad_norm": 3.155554336249525,966 "learning_rate": 6.342592592592594e-06,967 "loss": 0.4152,968 "step": 137969 },970 {971 "epoch": 0.02,972 "grad_norm": 2.210110094234751,973 "learning_rate": 6.3888888888888885e-06,974 "loss": 0.3817,975 "step": 138976 },977 {978 "epoch": 0.02,979 "grad_norm": 2.6684290241753215,980 "learning_rate": 6.435185185185186e-06,981 "loss": 0.4123,982 "step": 139983 },984 {985 "epoch": 0.02,986 "grad_norm": 3.4651775494160404,987 "learning_rate": 6.481481481481482e-06,988 "loss": 0.3952,989 "step": 140990 },991 {992 "epoch": 0.02,993 "grad_norm": 2.7795054216788966,994 "learning_rate": 6.5277777777777784e-06,995 "loss": 0.3871,996 "step": 141997 },998 {999 "epoch": 0.02,1000 "grad_norm": 2.4261457167219835,1001 "learning_rate": 6.574074074074075e-06,1002 "loss": 0.3693,1003 "step": 1421004 },1005 {1006 "epoch": 0.02,1007 "grad_norm": 2.7982447932470698,1008 "learning_rate": 6.620370370370371e-06,1009 "loss": 0.4014,1010 "step": 1431011 },1012 {1013 "epoch": 0.02,1014 "grad_norm": 2.6161920186765126,1015 "learning_rate": 6.666666666666667e-06,1016 "loss": 0.3935,1017 "step": 1441018 },1019 {1020 "epoch": 0.02,1021 "grad_norm": 2.8595145804997073,1022 "learning_rate": 6.712962962962963e-06,1023 "loss": 0.4079,1024 "step": 1451025 },1026 {1027 "epoch": 0.02,1028 "grad_norm": 2.638650439556187,1029 "learning_rate": 6.75925925925926e-06,1030 "loss": 0.4005,1031 "step": 1461032 },1033 {1034 "epoch": 0.02,1035 "grad_norm": 2.6351521116339587,1036 "learning_rate": 6.8055555555555566e-06,1037 "loss": 0.394,1038 "step": 1471039 },1040 {1041 "epoch": 0.02,1042 "grad_norm": 2.5240239772862627,1043 "learning_rate": 6.851851851851853e-06,1044 "loss": 0.4016,1045 "step": 1481046 },1047 {1048 "epoch": 0.02,1049 "grad_norm": 2.3911458158822096,1050 "learning_rate": 6.898148148148148e-06,1051 "loss": 0.3721,1052 "step": 1491053 },1054 {1055 "epoch": 0.02,1056 "grad_norm": 2.554819061478118,1057 "learning_rate": 6.944444444444445e-06,1058 "loss": 0.3861,1059 "step": 1501060 },1061 {1062 "epoch": 0.02,1063 "grad_norm": 2.84413328688025,1064 "learning_rate": 6.990740740740741e-06,1065 "loss": 0.4175,1066 "step": 1511067 },1068 {1069 "epoch": 0.02,1070 "grad_norm": 2.247204058545684,1071 "learning_rate": 7.0370370370370375e-06,1072 "loss": 0.4141,1073 "step": 1521074 },1075 {1076 "epoch": 0.02,1077 "grad_norm": 3.794284021960763,1078 "learning_rate": 7.083333333333335e-06,1079 "loss": 0.4028,1080 "step": 1531081 },1082 {1083 "epoch": 0.02,1084 "grad_norm": 3.8029897805515405,1085 "learning_rate": 7.129629629629629e-06,1086 "loss": 0.4141,1087 "step": 1541088 },1089 {1090 "epoch": 0.02,1091 "grad_norm": 3.0642245306154376,1092 "learning_rate": 7.1759259259259266e-06,1093 "loss": 0.3733,1094 "step": 1551095 },1096 {1097 "epoch": 0.02,1098 "grad_norm": 2.5282627153975894,1099 "learning_rate": 7.222222222222223e-06,1100 "loss": 0.3635,1101 "step": 1561102 },1103 {1104 "epoch": 0.02,1105 "grad_norm": 2.5844471230211523,1106 "learning_rate": 7.268518518518519e-06,1107 "loss": 0.3887,1108 "step": 1571109 },1110 {1111 "epoch": 0.02,1112 "grad_norm": 2.8902502031190025,1113 "learning_rate": 7.314814814814816e-06,1114 "loss": 0.4025,1115 "step": 1581116 },1117 {1118 "epoch": 0.02,1119 "grad_norm": 2.4184674724423885,1120 "learning_rate": 7.361111111111112e-06,1121 "loss": 0.3886,1122 "step": 1591123 },1124 {1125 "epoch": 0.02,1126 "grad_norm": 2.350433379430058,1127 "learning_rate": 7.4074074074074075e-06,1128 "loss": 0.3926,1129 "step": 1601130 },1131 {1132 "epoch": 0.02,1133 "grad_norm": 2.410879023537296,1134 "learning_rate": 7.453703703703704e-06,1135 "loss": 0.3735,1136 "step": 1611137 },1138 {1139 "epoch": 0.02,1140 "grad_norm": 2.774671890082111,1141 "learning_rate": 7.500000000000001e-06,1142 "loss": 0.3964,1143 "step": 1621144 },1145 {1146 "epoch": 0.02,1147 "grad_norm": 5.408550595652813,1148 "learning_rate": 7.546296296296297e-06,1149 "loss": 0.3886,1150 "step": 1631151 },1152 {1153 "epoch": 0.02,1154 "grad_norm": 2.5144844458067306,1155 "learning_rate": 7.592592592592594e-06,1156 "loss": 0.4011,1157 "step": 1641158 },1159 {1160 "epoch": 0.02,1161 "grad_norm": 2.2062221267943753,1162 "learning_rate": 7.638888888888888e-06,1163 "loss": 0.3934,1164 "step": 1651165 },1166 {1167 "epoch": 0.02,1168 "grad_norm": 2.4646673477484202,1169 "learning_rate": 7.685185185185185e-06,1170 "loss": 0.4079,1171 "step": 1661172 },1173 {1174 "epoch": 0.02,1175 "grad_norm": 2.211016478246042,1176 "learning_rate": 7.731481481481483e-06,1177 "loss": 0.3944,1178 "step": 1671179 },1180 {1181 "epoch": 0.02,1182 "grad_norm": 2.8306817341982056,1183 "learning_rate": 7.77777777777778e-06,1184 "loss": 0.3969,1185 "step": 1681186 },1187 {1188 "epoch": 0.02,1189 "grad_norm": 2.0189477377183582,1190 "learning_rate": 7.824074074074076e-06,1191 "loss": 0.3864,1192 "step": 1691193 },1194 {1195 "epoch": 0.02,1196 "grad_norm": 3.1359935744283067,1197 "learning_rate": 7.870370370370372e-06,1198 "loss": 0.3883,1199 "step": 1701200 },