OpenGVLab/VideoChat-Flash-Qwen2-7B_res448
131.1k
1{2 "best_metric": null,3 "best_model_checkpoint": null,4 "epoch": 0.9999461468038128,5 "eval_steps": 500,6 "global_step": 4642,7 "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": 0.3680746555328369,14 "learning_rate": 7.142857142857144e-08,15 "loss": 0.6038,16 "step": 117 },18 {19 "epoch": 0.0,20 "grad_norm": 0.5339280366897583,21 "learning_rate": 1.4285714285714287e-07,22 "loss": 0.6378,23 "step": 224 },25 {26 "epoch": 0.0,27 "grad_norm": 0.68046635389328,28 "learning_rate": 2.142857142857143e-07,29 "loss": 0.6492,30 "step": 331 },32 {33 "epoch": 0.0,34 "grad_norm": 0.4668383002281189,35 "learning_rate": 2.8571428571428575e-07,36 "loss": 0.5478,37 "step": 438 },39 {40 "epoch": 0.0,41 "grad_norm": 0.7353125810623169,42 "learning_rate": 3.5714285714285716e-07,43 "loss": 0.5802,44 "step": 545 },46 {47 "epoch": 0.0,48 "grad_norm": 0.4221579134464264,49 "learning_rate": 4.285714285714286e-07,50 "loss": 0.5863,51 "step": 652 },53 {54 "epoch": 0.0,55 "grad_norm": 0.35981830954551697,56 "learning_rate": 5.000000000000001e-07,57 "loss": 0.5881,58 "step": 759 },60 {61 "epoch": 0.0,62 "grad_norm": 0.49171018600463867,63 "learning_rate": 5.714285714285715e-07,64 "loss": 0.6196,65 "step": 866 },67 {68 "epoch": 0.0,69 "grad_norm": 0.26764529943466187,70 "learning_rate": 6.428571428571428e-07,71 "loss": 0.6026,72 "step": 973 },74 {75 "epoch": 0.0,76 "grad_norm": 0.31956347823143005,77 "learning_rate": 7.142857142857143e-07,78 "loss": 0.6077,79 "step": 1080 },81 {82 "epoch": 0.0,83 "grad_norm": 0.46464401483535767,84 "learning_rate": 7.857142857142857e-07,85 "loss": 0.6144,86 "step": 1187 },88 {89 "epoch": 0.0,90 "grad_norm": 0.41506412625312805,91 "learning_rate": 8.571428571428572e-07,92 "loss": 0.5684,93 "step": 1294 },95 {96 "epoch": 0.0,97 "grad_norm": 0.448373407125473,98 "learning_rate": 9.285714285714287e-07,99 "loss": 0.5785,100 "step": 13101 },102 {103 "epoch": 0.0,104 "grad_norm": 0.3355347514152527,105 "learning_rate": 1.0000000000000002e-06,106 "loss": 0.5688,107 "step": 14108 },109 {110 "epoch": 0.0,111 "grad_norm": 0.4548545181751251,112 "learning_rate": 1.0714285714285714e-06,113 "loss": 0.574,114 "step": 15115 },116 {117 "epoch": 0.0,118 "grad_norm": 0.4471697211265564,119 "learning_rate": 1.142857142857143e-06,120 "loss": 0.5526,121 "step": 16122 },123 {124 "epoch": 0.0,125 "grad_norm": 0.5379983186721802,126 "learning_rate": 1.2142857142857144e-06,127 "loss": 0.5503,128 "step": 17129 },130 {131 "epoch": 0.0,132 "grad_norm": 0.3886550962924957,133 "learning_rate": 1.2857142857142856e-06,134 "loss": 0.6074,135 "step": 18136 },137 {138 "epoch": 0.0,139 "grad_norm": 0.4560534656047821,140 "learning_rate": 1.3571428571428572e-06,141 "loss": 0.6262,142 "step": 19143 },144 {145 "epoch": 0.0,146 "grad_norm": 0.3122997283935547,147 "learning_rate": 1.4285714285714286e-06,148 "loss": 0.6103,149 "step": 20150 },151 {152 "epoch": 0.0,153 "grad_norm": 0.35939767956733704,154 "learning_rate": 1.5e-06,155 "loss": 0.6164,156 "step": 21157 },158 {159 "epoch": 0.0,160 "grad_norm": 0.3751821517944336,161 "learning_rate": 1.5714285714285714e-06,162 "loss": 0.5534,163 "step": 22164 },165 {166 "epoch": 0.0,167 "grad_norm": 0.5395365953445435,168 "learning_rate": 1.642857142857143e-06,169 "loss": 0.5911,170 "step": 23171 },172 {173 "epoch": 0.01,174 "grad_norm": 0.46072208881378174,175 "learning_rate": 1.7142857142857145e-06,176 "loss": 0.5986,177 "step": 24178 },179 {180 "epoch": 0.01,181 "grad_norm": 0.35585564374923706,182 "learning_rate": 1.7857142857142859e-06,183 "loss": 0.5726,184 "step": 25185 },186 {187 "epoch": 0.01,188 "grad_norm": 0.5622196197509766,189 "learning_rate": 1.8571428571428573e-06,190 "loss": 0.6092,191 "step": 26192 },193 {194 "epoch": 0.01,195 "grad_norm": 0.4780106246471405,196 "learning_rate": 1.928571428571429e-06,197 "loss": 0.6318,198 "step": 27199 },200 {201 "epoch": 0.01,202 "grad_norm": 0.4055005609989166,203 "learning_rate": 2.0000000000000003e-06,204 "loss": 0.5731,205 "step": 28206 },207 {208 "epoch": 0.01,209 "grad_norm": 0.29330089688301086,210 "learning_rate": 2.0714285714285717e-06,211 "loss": 0.577,212 "step": 29213 },214 {215 "epoch": 0.01,216 "grad_norm": 0.4011281132698059,217 "learning_rate": 2.1428571428571427e-06,218 "loss": 0.5771,219 "step": 30220 },221 {222 "epoch": 0.01,223 "grad_norm": 0.5358087420463562,224 "learning_rate": 2.2142857142857146e-06,225 "loss": 0.5324,226 "step": 31227 },228 {229 "epoch": 0.01,230 "grad_norm": 0.39781442284584045,231 "learning_rate": 2.285714285714286e-06,232 "loss": 0.5668,233 "step": 32234 },235 {236 "epoch": 0.01,237 "grad_norm": 0.44512811303138733,238 "learning_rate": 2.3571428571428574e-06,239 "loss": 0.6232,240 "step": 33241 },242 {243 "epoch": 0.01,244 "grad_norm": 0.3510986566543579,245 "learning_rate": 2.428571428571429e-06,246 "loss": 0.5816,247 "step": 34248 },249 {250 "epoch": 0.01,251 "grad_norm": 0.39098355174064636,252 "learning_rate": 2.5e-06,253 "loss": 0.5486,254 "step": 35255 },256 {257 "epoch": 0.01,258 "grad_norm": 0.4460495412349701,259 "learning_rate": 2.571428571428571e-06,260 "loss": 0.6256,261 "step": 36262 },263 {264 "epoch": 0.01,265 "grad_norm": 0.5601059794425964,266 "learning_rate": 2.642857142857143e-06,267 "loss": 0.5408,268 "step": 37269 },270 {271 "epoch": 0.01,272 "grad_norm": 0.543770432472229,273 "learning_rate": 2.7142857142857144e-06,274 "loss": 0.5843,275 "step": 38276 },277 {278 "epoch": 0.01,279 "grad_norm": 0.45234617590904236,280 "learning_rate": 2.785714285714286e-06,281 "loss": 0.6491,282 "step": 39283 },284 {285 "epoch": 0.01,286 "grad_norm": 0.35524260997772217,287 "learning_rate": 2.8571428571428573e-06,288 "loss": 0.5765,289 "step": 40290 },291 {292 "epoch": 0.01,293 "grad_norm": 0.3411543071269989,294 "learning_rate": 2.928571428571429e-06,295 "loss": 0.5184,296 "step": 41297 },298 {299 "epoch": 0.01,300 "grad_norm": 0.2239224910736084,301 "learning_rate": 3e-06,302 "loss": 0.5921,303 "step": 42304 },305 {306 "epoch": 0.01,307 "grad_norm": 0.4779617190361023,308 "learning_rate": 3.071428571428572e-06,309 "loss": 0.5981,310 "step": 43311 },312 {313 "epoch": 0.01,314 "grad_norm": 0.43023017048835754,315 "learning_rate": 3.142857142857143e-06,316 "loss": 0.5822,317 "step": 44318 },319 {320 "epoch": 0.01,321 "grad_norm": 0.5614896416664124,322 "learning_rate": 3.2142857142857147e-06,323 "loss": 0.5585,324 "step": 45325 },326 {327 "epoch": 0.01,328 "grad_norm": 0.35685351490974426,329 "learning_rate": 3.285714285714286e-06,330 "loss": 0.5623,331 "step": 46332 },333 {334 "epoch": 0.01,335 "grad_norm": 0.2451944649219513,336 "learning_rate": 3.357142857142857e-06,337 "loss": 0.4998,338 "step": 47339 },340 {341 "epoch": 0.01,342 "grad_norm": 0.24035461246967316,343 "learning_rate": 3.428571428571429e-06,344 "loss": 0.5864,345 "step": 48346 },347 {348 "epoch": 0.01,349 "grad_norm": 0.29300373792648315,350 "learning_rate": 3.5e-06,351 "loss": 0.5861,352 "step": 49353 },354 {355 "epoch": 0.01,356 "grad_norm": 0.3160554766654968,357 "learning_rate": 3.5714285714285718e-06,358 "loss": 0.5769,359 "step": 50360 },361 {362 "epoch": 0.01,363 "grad_norm": 0.20753253996372223,364 "learning_rate": 3.642857142857143e-06,365 "loss": 0.5592,366 "step": 51367 },368 {369 "epoch": 0.01,370 "grad_norm": 0.29364365339279175,371 "learning_rate": 3.7142857142857146e-06,372 "loss": 0.5855,373 "step": 52374 },375 {376 "epoch": 0.01,377 "grad_norm": 0.25966310501098633,378 "learning_rate": 3.785714285714286e-06,379 "loss": 0.5714,380 "step": 53381 },382 {383 "epoch": 0.01,384 "grad_norm": 0.5970475673675537,385 "learning_rate": 3.857142857142858e-06,386 "loss": 0.5955,387 "step": 54388 },389 {390 "epoch": 0.01,391 "grad_norm": 0.2742187976837158,392 "learning_rate": 3.928571428571429e-06,393 "loss": 0.5982,394 "step": 55395 },396 {397 "epoch": 0.01,398 "grad_norm": 0.21851637959480286,399 "learning_rate": 4.000000000000001e-06,400 "loss": 0.5086,401 "step": 56402 },403 {404 "epoch": 0.01,405 "grad_norm": 0.31623247265815735,406 "learning_rate": 4.071428571428572e-06,407 "loss": 0.5507,408 "step": 57409 },410 {411 "epoch": 0.01,412 "grad_norm": 0.21017701923847198,413 "learning_rate": 4.1428571428571435e-06,414 "loss": 0.5681,415 "step": 58416 },417 {418 "epoch": 0.01,419 "grad_norm": 0.23857641220092773,420 "learning_rate": 4.2142857142857145e-06,421 "loss": 0.5873,422 "step": 59423 },424 {425 "epoch": 0.01,426 "grad_norm": 0.330123633146286,427 "learning_rate": 4.2857142857142855e-06,428 "loss": 0.5507,429 "step": 60430 },431 {432 "epoch": 0.01,433 "grad_norm": 0.2590518295764923,434 "learning_rate": 4.357142857142857e-06,435 "loss": 0.5563,436 "step": 61437 },438 {439 "epoch": 0.01,440 "grad_norm": 0.22396299242973328,441 "learning_rate": 4.428571428571429e-06,442 "loss": 0.5208,443 "step": 62444 },445 {446 "epoch": 0.01,447 "grad_norm": 0.20799732208251953,448 "learning_rate": 4.5e-06,449 "loss": 0.5636,450 "step": 63451 },452 {453 "epoch": 0.01,454 "grad_norm": 0.28667014837265015,455 "learning_rate": 4.571428571428572e-06,456 "loss": 0.5126,457 "step": 64458 },459 {460 "epoch": 0.01,461 "grad_norm": 0.3183256983757019,462 "learning_rate": 4.642857142857144e-06,463 "loss": 0.4917,464 "step": 65465 },466 {467 "epoch": 0.01,468 "grad_norm": 0.34055784344673157,469 "learning_rate": 4.714285714285715e-06,470 "loss": 0.5869,471 "step": 66472 },473 {474 "epoch": 0.01,475 "grad_norm": 0.18996137380599976,476 "learning_rate": 4.785714285714287e-06,477 "loss": 0.5898,478 "step": 67479 },480 {481 "epoch": 0.01,482 "grad_norm": 0.19642199575901031,483 "learning_rate": 4.857142857142858e-06,484 "loss": 0.5024,485 "step": 68486 },487 {488 "epoch": 0.01,489 "grad_norm": 0.2719160318374634,490 "learning_rate": 4.928571428571429e-06,491 "loss": 0.5239,492 "step": 69493 },494 {495 "epoch": 0.02,496 "grad_norm": 0.2251090109348297,497 "learning_rate": 5e-06,498 "loss": 0.5327,499 "step": 70500 },501 {502 "epoch": 0.02,503 "grad_norm": 0.15708671510219574,504 "learning_rate": 5.071428571428571e-06,505 "loss": 0.5446,506 "step": 71507 },508 {509 "epoch": 0.02,510 "grad_norm": 0.2416897416114807,511 "learning_rate": 5.142857142857142e-06,512 "loss": 0.5972,513 "step": 72514 },515 {516 "epoch": 0.02,517 "grad_norm": 0.30218374729156494,518 "learning_rate": 5.214285714285715e-06,519 "loss": 0.5723,520 "step": 73521 },522 {523 "epoch": 0.02,524 "grad_norm": 0.23136214911937714,525 "learning_rate": 5.285714285714286e-06,526 "loss": 0.5919,527 "step": 74528 },529 {530 "epoch": 0.02,531 "grad_norm": 0.29520007967948914,532 "learning_rate": 5.357142857142857e-06,533 "loss": 0.5443,534 "step": 75535 },536 {537 "epoch": 0.02,538 "grad_norm": 0.2675969898700714,539 "learning_rate": 5.428571428571429e-06,540 "loss": 0.6078,541 "step": 76542 },543 {544 "epoch": 0.02,545 "grad_norm": 0.21040533483028412,546 "learning_rate": 5.500000000000001e-06,547 "loss": 0.5131,548 "step": 77549 },550 {551 "epoch": 0.02,552 "grad_norm": 0.21507872641086578,553 "learning_rate": 5.571428571428572e-06,554 "loss": 0.5537,555 "step": 78556 },557 {558 "epoch": 0.02,559 "grad_norm": 0.3713940680027008,560 "learning_rate": 5.6428571428571435e-06,561 "loss": 0.5677,562 "step": 79563 },564 {565 "epoch": 0.02,566 "grad_norm": 0.2338705062866211,567 "learning_rate": 5.7142857142857145e-06,568 "loss": 0.5583,569 "step": 80570 },571 {572 "epoch": 0.02,573 "grad_norm": 0.24568618834018707,574 "learning_rate": 5.785714285714286e-06,575 "loss": 0.591,576 "step": 81577 },578 {579 "epoch": 0.02,580 "grad_norm": 0.2607351541519165,581 "learning_rate": 5.857142857142858e-06,582 "loss": 0.5854,583 "step": 82584 },585 {586 "epoch": 0.02,587 "grad_norm": 0.25233450531959534,588 "learning_rate": 5.928571428571429e-06,589 "loss": 0.5435,590 "step": 83591 },592 {593 "epoch": 0.02,594 "grad_norm": 0.1901504397392273,595 "learning_rate": 6e-06,596 "loss": 0.5189,597 "step": 84598 },599 {600 "epoch": 0.02,601 "grad_norm": 0.20455455780029297,602 "learning_rate": 6.071428571428571e-06,603 "loss": 0.5372,604 "step": 85605 },606 {607 "epoch": 0.02,608 "grad_norm": 0.2110891193151474,609 "learning_rate": 6.142857142857144e-06,610 "loss": 0.5386,611 "step": 86612 },613 {614 "epoch": 0.02,615 "grad_norm": 0.18980112671852112,616 "learning_rate": 6.214285714285715e-06,617 "loss": 0.5834,618 "step": 87619 },620 {621 "epoch": 0.02,622 "grad_norm": 0.24649843573570251,623 "learning_rate": 6.285714285714286e-06,624 "loss": 0.5769,625 "step": 88626 },627 {628 "epoch": 0.02,629 "grad_norm": 0.20015066862106323,630 "learning_rate": 6.357142857142858e-06,631 "loss": 0.5213,632 "step": 89633 },634 {635 "epoch": 0.02,636 "grad_norm": 0.23394432663917542,637 "learning_rate": 6.4285714285714295e-06,638 "loss": 0.5351,639 "step": 90640 },641 {642 "epoch": 0.02,643 "grad_norm": 0.23542846739292145,644 "learning_rate": 6.5000000000000004e-06,645 "loss": 0.5527,646 "step": 91647 },648 {649 "epoch": 0.02,650 "grad_norm": 0.2840578258037567,651 "learning_rate": 6.571428571428572e-06,652 "loss": 0.5917,653 "step": 92654 },655 {656 "epoch": 0.02,657 "grad_norm": 0.5794204473495483,658 "learning_rate": 6.642857142857143e-06,659 "loss": 0.5693,660 "step": 93661 },662 {663 "epoch": 0.02,664 "grad_norm": 0.2579974830150604,665 "learning_rate": 6.714285714285714e-06,666 "loss": 0.577,667 "step": 94668 },669 {670 "epoch": 0.02,671 "grad_norm": 0.23646292090415955,672 "learning_rate": 6.785714285714287e-06,673 "loss": 0.5327,674 "step": 95675 },676 {677 "epoch": 0.02,678 "grad_norm": 0.3467201888561249,679 "learning_rate": 6.857142857142858e-06,680 "loss": 0.5163,681 "step": 96682 },683 {684 "epoch": 0.02,685 "grad_norm": 0.1848195195198059,686 "learning_rate": 6.928571428571429e-06,687 "loss": 0.5614,688 "step": 97689 },690 {691 "epoch": 0.02,692 "grad_norm": 0.20233601331710815,693 "learning_rate": 7e-06,694 "loss": 0.5993,695 "step": 98696 },697 {698 "epoch": 0.02,699 "grad_norm": 0.2592422366142273,700 "learning_rate": 7.0714285714285726e-06,701 "loss": 0.5568,702 "step": 99703 },704 {705 "epoch": 0.02,706 "grad_norm": 0.19915878772735596,707 "learning_rate": 7.1428571428571436e-06,708 "loss": 0.5331,709 "step": 100710 },711 {712 "epoch": 0.02,713 "grad_norm": 0.24805685877799988,714 "learning_rate": 7.2142857142857145e-06,715 "loss": 0.5241,716 "step": 101717 },718 {719 "epoch": 0.02,720 "grad_norm": 0.2214994579553604,721 "learning_rate": 7.285714285714286e-06,722 "loss": 0.5219,723 "step": 102724 },725 {726 "epoch": 0.02,727 "grad_norm": 0.2977463901042938,728 "learning_rate": 7.357142857142858e-06,729 "loss": 0.5836,730 "step": 103731 },732 {733 "epoch": 0.02,734 "grad_norm": 0.23825155198574066,735 "learning_rate": 7.428571428571429e-06,736 "loss": 0.5452,737 "step": 104738 },739 {740 "epoch": 0.02,741 "grad_norm": 0.22910349071025848,742 "learning_rate": 7.500000000000001e-06,743 "loss": 0.5525,744 "step": 105745 },746 {747 "epoch": 0.02,748 "grad_norm": 0.24861909449100494,749 "learning_rate": 7.571428571428572e-06,750 "loss": 0.5602,751 "step": 106752 },753 {754 "epoch": 0.02,755 "grad_norm": 0.220360666513443,756 "learning_rate": 7.642857142857143e-06,757 "loss": 0.4968,758 "step": 107759 },760 {761 "epoch": 0.02,762 "grad_norm": 0.29663270711898804,763 "learning_rate": 7.714285714285716e-06,764 "loss": 0.547,765 "step": 108766 },767 {768 "epoch": 0.02,769 "grad_norm": 0.15902388095855713,770 "learning_rate": 7.785714285714287e-06,771 "loss": 0.5593,772 "step": 109773 },774 {775 "epoch": 0.02,776 "grad_norm": 0.25408726930618286,777 "learning_rate": 7.857142857142858e-06,778 "loss": 0.5129,779 "step": 110780 },781 {782 "epoch": 0.02,783 "grad_norm": 0.25450989603996277,784 "learning_rate": 7.928571428571429e-06,785 "loss": 0.5568,786 "step": 111787 },788 {789 "epoch": 0.02,790 "grad_norm": 0.2113712877035141,791 "learning_rate": 8.000000000000001e-06,792 "loss": 0.5311,793 "step": 112794 },795 {796 "epoch": 0.02,797 "grad_norm": 0.2673487663269043,798 "learning_rate": 8.071428571428572e-06,799 "loss": 0.5063,800 "step": 113801 },802 {803 "epoch": 0.02,804 "grad_norm": 0.17971846461296082,805 "learning_rate": 8.142857142857143e-06,806 "loss": 0.4959,807 "step": 114808 },809 {810 "epoch": 0.02,811 "grad_norm": 0.27486327290534973,812 "learning_rate": 8.214285714285714e-06,813 "loss": 0.504,814 "step": 115815 },816 {817 "epoch": 0.02,818 "grad_norm": 0.3731400966644287,819 "learning_rate": 8.285714285714287e-06,820 "loss": 0.5262,821 "step": 116822 },823 {824 "epoch": 0.03,825 "grad_norm": 0.1998678743839264,826 "learning_rate": 8.357142857142858e-06,827 "loss": 0.6066,828 "step": 117829 },830 {831 "epoch": 0.03,832 "grad_norm": 0.18095743656158447,833 "learning_rate": 8.428571428571429e-06,834 "loss": 0.581,835 "step": 118836 },837 {838 "epoch": 0.03,839 "grad_norm": 0.20576633512973785,840 "learning_rate": 8.5e-06,841 "loss": 0.5646,842 "step": 119843 },844 {845 "epoch": 0.03,846 "grad_norm": 0.21952424943447113,847 "learning_rate": 8.571428571428571e-06,848 "loss": 0.5274,849 "step": 120850 },851 {852 "epoch": 0.03,853 "grad_norm": 0.22617046535015106,854 "learning_rate": 8.642857142857144e-06,855 "loss": 0.4918,856 "step": 121857 },858 {859 "epoch": 0.03,860 "grad_norm": 0.22353151440620422,861 "learning_rate": 8.714285714285715e-06,862 "loss": 0.5383,863 "step": 122864 },865 {866 "epoch": 0.03,867 "grad_norm": 0.24257732927799225,868 "learning_rate": 8.785714285714286e-06,869 "loss": 0.4734,870 "step": 123871 },872 {873 "epoch": 0.03,874 "grad_norm": 0.16320379078388214,875 "learning_rate": 8.857142857142858e-06,876 "loss": 0.5033,877 "step": 124878 },879 {880 "epoch": 0.03,881 "grad_norm": 0.21186141669750214,882 "learning_rate": 8.92857142857143e-06,883 "loss": 0.5116,884 "step": 125885 },886 {887 "epoch": 0.03,888 "grad_norm": 0.1727321892976761,889 "learning_rate": 9e-06,890 "loss": 0.4887,891 "step": 126892 },893 {894 "epoch": 0.03,895 "grad_norm": 0.17333361506462097,896 "learning_rate": 9.071428571428573e-06,897 "loss": 0.5629,898 "step": 127899 },900 {901 "epoch": 0.03,902 "grad_norm": 0.20159348845481873,903 "learning_rate": 9.142857142857144e-06,904 "loss": 0.5855,905 "step": 128906 },907 {908 "epoch": 0.03,909 "grad_norm": 0.25432631373405457,910 "learning_rate": 9.214285714285715e-06,911 "loss": 0.5565,912 "step": 129913 },914 {915 "epoch": 0.03,916 "grad_norm": 0.18436311185359955,917 "learning_rate": 9.285714285714288e-06,918 "loss": 0.4746,919 "step": 130920 },921 {922 "epoch": 0.03,923 "grad_norm": 0.22167499363422394,924 "learning_rate": 9.357142857142859e-06,925 "loss": 0.5437,926 "step": 131927 },928 {929 "epoch": 0.03,930 "grad_norm": 0.29192057251930237,931 "learning_rate": 9.42857142857143e-06,932 "loss": 0.5443,933 "step": 132934 },935 {936 "epoch": 0.03,937 "grad_norm": 0.1628040224313736,938 "learning_rate": 9.5e-06,939 "loss": 0.5563,940 "step": 133941 },942 {943 "epoch": 0.03,944 "grad_norm": 0.32334551215171814,945 "learning_rate": 9.571428571428573e-06,946 "loss": 0.575,947 "step": 134948 },949 {950 "epoch": 0.03,951 "grad_norm": 0.272955983877182,952 "learning_rate": 9.642857142857144e-06,953 "loss": 0.5363,954 "step": 135955 },956 {957 "epoch": 0.03,958 "grad_norm": 0.2314363420009613,959 "learning_rate": 9.714285714285715e-06,960 "loss": 0.5356,961 "step": 136962 },963 {964 "epoch": 0.03,965 "grad_norm": 0.18768808245658875,966 "learning_rate": 9.785714285714286e-06,967 "loss": 0.5053,968 "step": 137969 },970 {971 "epoch": 0.03,972 "grad_norm": 0.22900734841823578,973 "learning_rate": 9.857142857142859e-06,974 "loss": 0.5594,975 "step": 138976 },977 {978 "epoch": 0.03,979 "grad_norm": 0.1723155379295349,980 "learning_rate": 9.92857142857143e-06,981 "loss": 0.5104,982 "step": 139983 },984 {985 "epoch": 0.03,986 "grad_norm": 0.2596263885498047,987 "learning_rate": 1e-05,988 "loss": 0.5271,989 "step": 140990 },991 {992 "epoch": 0.03,993 "grad_norm": 0.15986420214176178,994 "learning_rate": 9.999998782612734e-06,995 "loss": 0.5382,996 "step": 141997 },998 {999 "epoch": 0.03,1000 "grad_norm": 0.33412984013557434,1001 "learning_rate": 9.999995130451526e-06,1002 "loss": 0.4807,1003 "step": 1421004 },1005 {1006 "epoch": 0.03,1007 "grad_norm": 0.20340685546398163,1008 "learning_rate": 9.999989043518153e-06,1009 "loss": 0.519,1010 "step": 1431011 },1012 {1013 "epoch": 0.03,1014 "grad_norm": 0.18798081576824188,1015 "learning_rate": 9.999980521815582e-06,1016 "loss": 0.5347,1017 "step": 1441018 },1019 {1020 "epoch": 0.03,1021 "grad_norm": 0.20350557565689087,1022 "learning_rate": 9.99996956534796e-06,1023 "loss": 0.4913,1024 "step": 1451025 },1026 {1027 "epoch": 0.03,1028 "grad_norm": 0.2547079026699066,1029 "learning_rate": 9.999956174120626e-06,1030 "loss": 0.5284,1031 "step": 1461032 },1033 {1034 "epoch": 0.03,1035 "grad_norm": 0.26818037033081055,1036 "learning_rate": 9.999940348140098e-06,1037 "loss": 0.5597,1038 "step": 1471039 },1040 {1041 "epoch": 0.03,1042 "grad_norm": 0.1871444284915924,1043 "learning_rate": 9.999922087414084e-06,1044 "loss": 0.4857,1045 "step": 1481046 },1047 {1048 "epoch": 0.03,1049 "grad_norm": 0.24267414212226868,1050 "learning_rate": 9.999901391951474e-06,1051 "loss": 0.5243,1052 "step": 1491053 },1054 {1055 "epoch": 0.03,1056 "grad_norm": 0.22753533720970154,1057 "learning_rate": 9.99987826176235e-06,1058 "loss": 0.4868,1059 "step": 1501060 },1061 {1062 "epoch": 0.03,1063 "grad_norm": 0.17949774861335754,1064 "learning_rate": 9.99985269685797e-06,1065 "loss": 0.5477,1066 "step": 1511067 },1068 {1069 "epoch": 0.03,1070 "grad_norm": 0.24117450416088104,1071 "learning_rate": 9.999824697250786e-06,1072 "loss": 0.5583,1073 "step": 1521074 },1075 {1076 "epoch": 0.03,1077 "grad_norm": 0.2758869230747223,1078 "learning_rate": 9.999794262954432e-06,1079 "loss": 0.6049,1080 "step": 1531081 },1082 {1083 "epoch": 0.03,1084 "grad_norm": 0.22826828062534332,1085 "learning_rate": 9.999761393983728e-06,1086 "loss": 0.5437,1087 "step": 1541088 },1089 {1090 "epoch": 0.03,1091 "grad_norm": 0.24656014144420624,1092 "learning_rate": 9.999726090354683e-06,1093 "loss": 0.5417,1094 "step": 1551095 },1096 {1097 "epoch": 0.03,1098 "grad_norm": 0.1714806854724884,1099 "learning_rate": 9.999688352084482e-06,1100 "loss": 0.5189,1101 "step": 1561102 },1103 {1104 "epoch": 0.03,1105 "grad_norm": 0.17295528948307037,1106 "learning_rate": 9.999648179191505e-06,1107 "loss": 0.5478,1108 "step": 1571109 },1110 {1111 "epoch": 0.03,1112 "grad_norm": 0.19324244558811188,1113 "learning_rate": 9.999605571695317e-06,1114 "loss": 0.5664,1115 "step": 1581116 },1117 {1118 "epoch": 0.03,1119 "grad_norm": 0.20624053478240967,1120 "learning_rate": 9.999560529616661e-06,1121 "loss": 0.5087,1122 "step": 1591123 },1124 {1125 "epoch": 0.03,1126 "grad_norm": 0.26294004917144775,1127 "learning_rate": 9.999513052977473e-06,1128 "loss": 0.6106,1129 "step": 1601130 },1131 {1132 "epoch": 0.03,1133 "grad_norm": 0.19021935760974884,1134 "learning_rate": 9.999463141800873e-06,1135 "loss": 0.4975,1136 "step": 1611137 },1138 {1139 "epoch": 0.03,1140 "grad_norm": 0.17529787123203278,1141 "learning_rate": 9.999410796111163e-06,1142 "loss": 0.5204,1143 "step": 1621144 },1145 {1146 "epoch": 0.04,1147 "grad_norm": 0.19823302328586578,1148 "learning_rate": 9.999356015933834e-06,1149 "loss": 0.5312,1150 "step": 1631151 },1152 {1153 "epoch": 0.04,1154 "grad_norm": 0.2933864891529083,1155 "learning_rate": 9.999298801295564e-06,1156 "loss": 0.5123,1157 "step": 1641158 },1159 {1160 "epoch": 0.04,1161 "grad_norm": 0.19900120794773102,1162 "learning_rate": 9.99923915222421e-06,1163 "loss": 0.5824,1164 "step": 1651165 },1166 {1167 "epoch": 0.04,1168 "grad_norm": 0.18184617161750793,1169 "learning_rate": 9.99917706874882e-06,1170 "loss": 0.5623,1171 "step": 1661172 },1173 {1174 "epoch": 0.04,1175 "grad_norm": 0.1705755591392517,1176 "learning_rate": 9.999112550899627e-06,1177 "loss": 0.5458,1178 "step": 1671179 },1180 {1181 "epoch": 0.04,1182 "grad_norm": 0.3029863238334656,1183 "learning_rate": 9.999045598708047e-06,1184 "loss": 0.54,1185 "step": 1681186 },1187 {1188 "epoch": 0.04,1189 "grad_norm": 0.1956048458814621,1190 "learning_rate": 9.998976212206683e-06,1191 "loss": 0.5387,1192 "step": 1691193 },1194 {1195 "epoch": 0.04,1196 "grad_norm": 0.1850360929965973,1197 "learning_rate": 9.998904391429323e-06,1198 "loss": 0.5085,1199 "step": 1701200 },