tz579/example_asr_wav2vec2
06
1{2 "best_metric": null,3 "best_model_checkpoint": null,4 "epoch": 2.0,5 "eval_steps": 1000,6 "global_step": 12776,7 "is_hyper_param_search": false,8 "is_local_process_zero": true,9 "is_world_process_zero": true,10 "log_history": [11 {12 "epoch": 0.00015654351909830932,13 "grad_norm": Infinity,14 "learning_rate": 0.0,15 "loss": 10.2049,16 "step": 117 },18 {19 "epoch": 0.00031308703819661864,20 "grad_norm": Infinity,21 "learning_rate": 0.0,22 "loss": 10.1719,23 "step": 224 },25 {26 "epoch": 0.000469630557294928,27 "grad_norm": 10.289196014404297,28 "learning_rate": 6e-07,29 "loss": 11.8275,30 "step": 331 },32 {33 "epoch": 0.0006261740763932373,34 "grad_norm": 8.432656288146973,35 "learning_rate": 1.2e-06,36 "loss": 9.669,37 "step": 438 },39 {40 "epoch": 0.0007827175954915466,41 "grad_norm": 10.236042022705078,42 "learning_rate": 1.8e-06,43 "loss": 11.8878,44 "step": 545 },46 {47 "epoch": 0.000939261114589856,48 "grad_norm": 9.023536682128906,49 "learning_rate": 2.4e-06,50 "loss": 11.0993,51 "step": 652 },53 {54 "epoch": 0.0010958046336881652,55 "grad_norm": 9.157316207885742,56 "learning_rate": 2.9999999999999997e-06,57 "loss": 11.0304,58 "step": 759 },60 {61 "epoch": 0.0012523481527864746,62 "grad_norm": 15.224575996398926,63 "learning_rate": 3.6e-06,64 "loss": 18.5492,65 "step": 866 },67 {68 "epoch": 0.001408891671884784,69 "grad_norm": 8.25251579284668,70 "learning_rate": 4.2e-06,71 "loss": 9.8113,72 "step": 973 },74 {75 "epoch": 0.0015654351909830933,76 "grad_norm": 8.493621826171875,77 "learning_rate": 4.8e-06,78 "loss": 10.5777,79 "step": 1080 },81 {82 "epoch": 0.0017219787100814026,83 "grad_norm": 8.078389167785645,84 "learning_rate": 5.399999999999999e-06,85 "loss": 9.6918,86 "step": 1187 },88 {89 "epoch": 0.001878522229179712,90 "grad_norm": 12.197300910949707,91 "learning_rate": 5.999999999999999e-06,92 "loss": 15.5513,93 "step": 1294 },95 {96 "epoch": 0.002035065748278021,97 "grad_norm": 14.715069770812988,98 "learning_rate": 6.599999999999999e-06,99 "loss": 21.1878,100 "step": 13101 },102 {103 "epoch": 0.0021916092673763305,104 "grad_norm": 14.680959701538086,105 "learning_rate": 7.2e-06,106 "loss": 20.1223,107 "step": 14108 },109 {110 "epoch": 0.00234815278647464,111 "grad_norm": Infinity,112 "learning_rate": 7.2e-06,113 "loss": 25.9204,114 "step": 15115 },116 {117 "epoch": 0.002504696305572949,118 "grad_norm": 7.316311359405518,119 "learning_rate": 7.799999999999998e-06,120 "loss": 9.0609,121 "step": 16122 },123 {124 "epoch": 0.0026612398246712585,125 "grad_norm": 14.48494815826416,126 "learning_rate": 8.4e-06,127 "loss": 19.6195,128 "step": 17129 },130 {131 "epoch": 0.002817783343769568,132 "grad_norm": 24.607234954833984,133 "learning_rate": 8.999999999999999e-06,134 "loss": 24.8781,135 "step": 18136 },137 {138 "epoch": 0.002974326862867877,139 "grad_norm": 10.66619873046875,140 "learning_rate": 9.6e-06,141 "loss": 14.3883,142 "step": 19143 },144 {145 "epoch": 0.0031308703819661866,146 "grad_norm": 9.02160930633545,147 "learning_rate": 1.02e-05,148 "loss": 11.2116,149 "step": 20150 },151 {152 "epoch": 0.003287413901064496,153 "grad_norm": 8.273414611816406,154 "learning_rate": 1.0799999999999998e-05,155 "loss": 10.0936,156 "step": 21157 },158 {159 "epoch": 0.0034439574201628053,160 "grad_norm": 9.83197021484375,161 "learning_rate": 1.14e-05,162 "loss": 12.5294,163 "step": 22164 },165 {166 "epoch": 0.0036005009392611146,167 "grad_norm": 11.494529724121094,168 "learning_rate": 1.1999999999999999e-05,169 "loss": 15.1262,170 "step": 23171 },172 {173 "epoch": 0.003757044458359424,174 "grad_norm": 8.389466285705566,175 "learning_rate": 1.26e-05,176 "loss": 11.8818,177 "step": 24178 },179 {180 "epoch": 0.003913587977457733,181 "grad_norm": 11.484485626220703,182 "learning_rate": 1.3199999999999997e-05,183 "loss": 16.7747,184 "step": 25185 },186 {187 "epoch": 0.004070131496556042,188 "grad_norm": 11.260869026184082,189 "learning_rate": 1.3799999999999998e-05,190 "loss": 15.3859,191 "step": 26192 },193 {194 "epoch": 0.004226675015654352,195 "grad_norm": 17.069162368774414,196 "learning_rate": 1.44e-05,197 "loss": 19.4706,198 "step": 27199 },200 {201 "epoch": 0.004383218534752661,202 "grad_norm": 8.778854370117188,203 "learning_rate": 1.4999999999999999e-05,204 "loss": 13.0765,205 "step": 28206 },207 {208 "epoch": 0.004539762053850971,209 "grad_norm": 10.239326477050781,210 "learning_rate": 1.5599999999999996e-05,211 "loss": 14.7478,212 "step": 29213 },214 {215 "epoch": 0.00469630557294928,216 "grad_norm": 9.703652381896973,217 "learning_rate": 1.6199999999999997e-05,218 "loss": 13.5911,219 "step": 30220 },221 {222 "epoch": 0.004852849092047589,223 "grad_norm": 10.830334663391113,224 "learning_rate": 1.68e-05,225 "loss": 16.7388,226 "step": 31227 },228 {229 "epoch": 0.005009392611145898,230 "grad_norm": 14.102766990661621,231 "learning_rate": 1.74e-05,232 "loss": 19.3178,233 "step": 32234 },235 {236 "epoch": 0.005165936130244208,237 "grad_norm": 12.113805770874023,238 "learning_rate": 1.7999999999999997e-05,239 "loss": 17.4729,240 "step": 33241 },242 {243 "epoch": 0.005322479649342517,244 "grad_norm": 9.666101455688477,245 "learning_rate": 1.8599999999999998e-05,246 "loss": 14.8432,247 "step": 34248 },249 {250 "epoch": 0.005479023168440827,251 "grad_norm": 10.271428108215332,252 "learning_rate": 1.92e-05,253 "loss": 15.5869,254 "step": 35255 },256 {257 "epoch": 0.005635566687539136,258 "grad_norm": 27.508230209350586,259 "learning_rate": 1.98e-05,260 "loss": 13.2456,261 "step": 36262 },263 {264 "epoch": 0.0057921102066374455,265 "grad_norm": 10.40356159210205,266 "learning_rate": 2.04e-05,267 "loss": 15.4974,268 "step": 37269 },270 {271 "epoch": 0.005948653725735754,272 "grad_norm": 10.521185874938965,273 "learning_rate": 2.1e-05,274 "loss": 16.509,275 "step": 38276 },277 {278 "epoch": 0.006105197244834064,279 "grad_norm": 9.262508392333984,280 "learning_rate": 2.1599999999999996e-05,281 "loss": 12.5824,282 "step": 39283 },284 {285 "epoch": 0.006261740763932373,286 "grad_norm": 9.204741477966309,287 "learning_rate": 2.2199999999999998e-05,288 "loss": 13.576,289 "step": 40290 },291 {292 "epoch": 0.006418284283030683,293 "grad_norm": 6.91443395614624,294 "learning_rate": 2.28e-05,295 "loss": 11.8722,296 "step": 41297 },298 {299 "epoch": 0.006574827802128992,300 "grad_norm": 7.96762752532959,301 "learning_rate": 2.34e-05,302 "loss": 13.263,303 "step": 42304 },305 {306 "epoch": 0.006731371321227302,307 "grad_norm": 9.377372741699219,308 "learning_rate": 2.3999999999999997e-05,309 "loss": 12.1481,310 "step": 43311 },312 {313 "epoch": 0.0068879148403256105,314 "grad_norm": 6.440145015716553,315 "learning_rate": 2.4599999999999998e-05,316 "loss": 10.7579,317 "step": 44318 },319 {320 "epoch": 0.007044458359423919,321 "grad_norm": 5.9850077629089355,322 "learning_rate": 2.52e-05,323 "loss": 9.8462,324 "step": 45325 },326 {327 "epoch": 0.007201001878522229,328 "grad_norm": 10.260824203491211,329 "learning_rate": 2.5799999999999997e-05,330 "loss": 9.8883,331 "step": 46332 },333 {334 "epoch": 0.007357545397620538,335 "grad_norm": 10.321931838989258,336 "learning_rate": 2.6399999999999995e-05,337 "loss": 8.8922,338 "step": 47339 },340 {341 "epoch": 0.007514088916718848,342 "grad_norm": 5.632627964019775,343 "learning_rate": 2.6999999999999996e-05,344 "loss": 8.3128,345 "step": 48346 },347 {348 "epoch": 0.007670632435817157,349 "grad_norm": 4.798144817352295,350 "learning_rate": 2.7599999999999997e-05,351 "loss": 7.1669,352 "step": 49353 },354 {355 "epoch": 0.007827175954915467,356 "grad_norm": 4.642441272735596,357 "learning_rate": 2.8199999999999998e-05,358 "loss": 6.2127,359 "step": 50360 },361 {362 "epoch": 0.007983719474013776,363 "grad_norm": 12.44339370727539,364 "learning_rate": 2.88e-05,365 "loss": 11.9784,366 "step": 51367 },368 {369 "epoch": 0.008140262993112084,370 "grad_norm": 11.49549674987793,371 "learning_rate": 2.94e-05,372 "loss": 10.976,373 "step": 52374 },375 {376 "epoch": 0.008296806512210394,377 "grad_norm": 16.98020362854004,378 "learning_rate": 2.9999999999999997e-05,379 "loss": 15.17,380 "step": 53381 },382 {383 "epoch": 0.008453350031308704,384 "grad_norm": 8.904106140136719,385 "learning_rate": 3.06e-05,386 "loss": 8.8607,387 "step": 54388 },389 {390 "epoch": 0.008609893550407014,391 "grad_norm": 9.727025985717773,392 "learning_rate": 3.119999999999999e-05,393 "loss": 9.6081,394 "step": 55395 },396 {397 "epoch": 0.008766437069505322,398 "grad_norm": 9.460471153259277,399 "learning_rate": 3.1799999999999994e-05,400 "loss": 9.9255,401 "step": 56402 },403 {404 "epoch": 0.008922980588603632,405 "grad_norm": 9.601983070373535,406 "learning_rate": 3.2399999999999995e-05,407 "loss": 9.9013,408 "step": 57409 },410 {411 "epoch": 0.009079524107701941,412 "grad_norm": 10.275938034057617,413 "learning_rate": 3.2999999999999996e-05,414 "loss": 11.9712,415 "step": 58416 },417 {418 "epoch": 0.009236067626800251,419 "grad_norm": 9.154601097106934,420 "learning_rate": 3.36e-05,421 "loss": 10.6476,422 "step": 59423 },424 {425 "epoch": 0.00939261114589856,426 "grad_norm": 8.198284149169922,427 "learning_rate": 3.42e-05,428 "loss": 9.7442,429 "step": 60430 },431 {432 "epoch": 0.009549154664996869,433 "grad_norm": 10.912142753601074,434 "learning_rate": 3.48e-05,435 "loss": 11.3455,436 "step": 61437 },438 {439 "epoch": 0.009705698184095179,440 "grad_norm": 6.028509616851807,441 "learning_rate": 3.539999999999999e-05,442 "loss": 7.7892,443 "step": 62444 },445 {446 "epoch": 0.009862241703193489,447 "grad_norm": 10.75712776184082,448 "learning_rate": 3.5999999999999994e-05,449 "loss": 10.4904,450 "step": 63451 },452 {453 "epoch": 0.010018785222291797,454 "grad_norm": 11.000633239746094,455 "learning_rate": 3.6599999999999995e-05,456 "loss": 9.4008,457 "step": 64458 },459 {460 "epoch": 0.010175328741390106,461 "grad_norm": 7.76064395904541,462 "learning_rate": 3.7199999999999996e-05,463 "loss": 7.2246,464 "step": 65465 },466 {467 "epoch": 0.010331872260488416,468 "grad_norm": 23.41922950744629,469 "learning_rate": 3.78e-05,470 "loss": 13.3124,471 "step": 66472 },473 {474 "epoch": 0.010488415779586726,475 "grad_norm": 11.076509475708008,476 "learning_rate": 3.84e-05,477 "loss": 7.637,478 "step": 67479 },480 {481 "epoch": 0.010644959298685034,482 "grad_norm": 29.545185089111328,483 "learning_rate": 3.9e-05,484 "loss": 14.3584,485 "step": 68486 },487 {488 "epoch": 0.010801502817783344,489 "grad_norm": 17.31962776184082,490 "learning_rate": 3.96e-05,491 "loss": 8.6004,492 "step": 69493 },494 {495 "epoch": 0.010958046336881654,496 "grad_norm": 11.55864143371582,497 "learning_rate": 4.02e-05,498 "loss": 6.3439,499 "step": 70500 },501 {502 "epoch": 0.011114589855979962,503 "grad_norm": 25.8043212890625,504 "learning_rate": 4.08e-05,505 "loss": 10.298,506 "step": 71507 },508 {509 "epoch": 0.011271133375078271,510 "grad_norm": 33.71727752685547,511 "learning_rate": 4.14e-05,512 "loss": 12.2877,513 "step": 72514 },515 {516 "epoch": 0.011427676894176581,517 "grad_norm": 20.209596633911133,518 "learning_rate": 4.2e-05,519 "loss": 7.7747,520 "step": 73521 },522 {523 "epoch": 0.011584220413274891,524 "grad_norm": 20.540761947631836,525 "learning_rate": 4.259999999999999e-05,526 "loss": 7.6235,527 "step": 74528 },529 {530 "epoch": 0.011740763932373199,531 "grad_norm": Infinity,532 "learning_rate": 4.259999999999999e-05,533 "loss": 12.327,534 "step": 75535 },536 {537 "epoch": 0.011897307451471509,538 "grad_norm": 19.35993766784668,539 "learning_rate": 4.319999999999999e-05,540 "loss": 7.1177,541 "step": 76542 },543 {544 "epoch": 0.012053850970569819,545 "grad_norm": 36.380619049072266,546 "learning_rate": 4.3799999999999994e-05,547 "loss": 10.6781,548 "step": 77549 },550 {551 "epoch": 0.012210394489668128,552 "grad_norm": 45.644840240478516,553 "learning_rate": 4.4399999999999995e-05,554 "loss": 12.4375,555 "step": 78556 },557 {558 "epoch": 0.012366938008766436,559 "grad_norm": 27.72957992553711,560 "learning_rate": 4.4999999999999996e-05,561 "loss": 8.1513,562 "step": 79563 },564 {565 "epoch": 0.012523481527864746,566 "grad_norm": 35.70543670654297,567 "learning_rate": 4.56e-05,568 "loss": 9.3938,569 "step": 80570 },571 {572 "epoch": 0.012680025046963056,573 "grad_norm": 51.08866500854492,574 "learning_rate": 4.62e-05,575 "loss": 12.0758,576 "step": 81577 },578 {579 "epoch": 0.012836568566061366,580 "grad_norm": 38.18144989013672,581 "learning_rate": 4.68e-05,582 "loss": 9.2489,583 "step": 82584 },585 {586 "epoch": 0.012993112085159674,587 "grad_norm": 25.754817962646484,588 "learning_rate": 4.7399999999999993e-05,589 "loss": 6.9829,590 "step": 83591 },592 {593 "epoch": 0.013149655604257984,594 "grad_norm": 33.58719253540039,595 "learning_rate": 4.7999999999999994e-05,596 "loss": 8.0805,597 "step": 84598 },599 {600 "epoch": 0.013306199123356293,601 "grad_norm": 19.882631301879883,602 "learning_rate": 4.8599999999999995e-05,603 "loss": 5.7876,604 "step": 85605 },606 {607 "epoch": 0.013462742642454603,608 "grad_norm": 28.052623748779297,609 "learning_rate": 4.9199999999999997e-05,610 "loss": 6.844,611 "step": 86612 },613 {614 "epoch": 0.013619286161552911,615 "grad_norm": 24.705547332763672,616 "learning_rate": 4.98e-05,617 "loss": 6.2694,618 "step": 87619 },620 {621 "epoch": 0.013775829680651221,622 "grad_norm": 27.596519470214844,623 "learning_rate": 5.04e-05,624 "loss": 6.6974,625 "step": 88626 },627 {628 "epoch": 0.01393237319974953,629 "grad_norm": 24.197368621826172,630 "learning_rate": 5.1e-05,631 "loss": 5.9251,632 "step": 89633 },634 {635 "epoch": 0.014088916718847839,636 "grad_norm": 33.817039489746094,637 "learning_rate": 5.1599999999999994e-05,638 "loss": 7.2591,639 "step": 90640 },641 {642 "epoch": 0.014245460237946149,643 "grad_norm": 28.664337158203125,644 "learning_rate": 5.2199999999999995e-05,645 "loss": 6.3747,646 "step": 91647 },648 {649 "epoch": 0.014402003757044458,650 "grad_norm": 19.04326057434082,651 "learning_rate": 5.279999999999999e-05,652 "loss": 5.2411,653 "step": 92654 },655 {656 "epoch": 0.014558547276142768,657 "grad_norm": 19.95840835571289,658 "learning_rate": 5.339999999999999e-05,659 "loss": 5.2001,660 "step": 93661 },662 {663 "epoch": 0.014715090795241076,664 "grad_norm": 17.750680923461914,665 "learning_rate": 5.399999999999999e-05,666 "loss": 4.9905,667 "step": 94668 },669 {670 "epoch": 0.014871634314339386,671 "grad_norm": 17.195247650146484,672 "learning_rate": 5.459999999999999e-05,673 "loss": 4.8285,674 "step": 95675 },676 {677 "epoch": 0.015028177833437696,678 "grad_norm": 12.27951431274414,679 "learning_rate": 5.519999999999999e-05,680 "loss": 4.311,681 "step": 96682 },683 {684 "epoch": 0.015184721352536006,685 "grad_norm": 10.93501091003418,686 "learning_rate": 5.5799999999999994e-05,687 "loss": 4.0848,688 "step": 97689 },690 {691 "epoch": 0.015341264871634314,692 "grad_norm": 8.672501564025879,693 "learning_rate": 5.6399999999999995e-05,694 "loss": 3.961,695 "step": 98696 },697 {698 "epoch": 0.015497808390732623,699 "grad_norm": 7.686964988708496,700 "learning_rate": 5.6999999999999996e-05,701 "loss": 3.8516,702 "step": 99703 },704 {705 "epoch": 0.015654351909830933,706 "grad_norm": 5.318123817443848,707 "learning_rate": 5.76e-05,708 "loss": 3.6987,709 "step": 100710 },711 {712 "epoch": 0.01581089542892924,713 "grad_norm": 13.304719924926758,714 "learning_rate": 5.82e-05,715 "loss": 4.1017,716 "step": 101717 },718 {719 "epoch": 0.015967438948027553,720 "grad_norm": 12.244338989257812,721 "learning_rate": 5.88e-05,722 "loss": 3.9793,723 "step": 102724 },725 {726 "epoch": 0.01612398246712586,727 "grad_norm": 11.684955596923828,728 "learning_rate": 5.94e-05,729 "loss": 3.9138,730 "step": 103731 },732 {733 "epoch": 0.01628052598622417,734 "grad_norm": 36.793663024902344,735 "learning_rate": 5.9999999999999995e-05,736 "loss": 6.8625,737 "step": 104738 },739 {740 "epoch": 0.01643706950532248,741 "grad_norm": 20.531827926635742,742 "learning_rate": 6.0599999999999996e-05,743 "loss": 4.8527,744 "step": 105745 },746 {747 "epoch": 0.01659361302442079,748 "grad_norm": 12.65087604522705,749 "learning_rate": 6.12e-05,750 "loss": 4.0059,751 "step": 106752 },753 {754 "epoch": 0.0167501565435191,755 "grad_norm": 9.573142051696777,756 "learning_rate": 6.18e-05,757 "loss": 3.7414,758 "step": 107759 },760 {761 "epoch": 0.016906700062617408,762 "grad_norm": 13.197820663452148,763 "learning_rate": 6.239999999999999e-05,764 "loss": 4.1523,765 "step": 108766 },767 {768 "epoch": 0.017063243581715716,769 "grad_norm": 10.532169342041016,770 "learning_rate": 6.299999999999999e-05,771 "loss": 3.8368,772 "step": 109773 },774 {775 "epoch": 0.017219787100814028,776 "grad_norm": 12.9858980178833,777 "learning_rate": 6.359999999999999e-05,778 "loss": 4.1369,779 "step": 110780 },781 {782 "epoch": 0.017376330619912336,783 "grad_norm": 29.88274574279785,784 "learning_rate": 6.419999999999999e-05,785 "loss": 5.8008,786 "step": 111787 },788 {789 "epoch": 0.017532874139010644,790 "grad_norm": 17.681074142456055,791 "learning_rate": 6.479999999999999e-05,792 "loss": 4.4349,793 "step": 112794 },795 {796 "epoch": 0.017689417658108955,797 "grad_norm": 14.819682121276855,798 "learning_rate": 6.539999999999999e-05,799 "loss": 4.3329,800 "step": 113801 },802 {803 "epoch": 0.017845961177207263,804 "grad_norm": 12.398885726928711,805 "learning_rate": 6.599999999999999e-05,806 "loss": 4.0054,807 "step": 114808 },809 {810 "epoch": 0.01800250469630557,811 "grad_norm": 11.930230140686035,812 "learning_rate": 6.659999999999999e-05,813 "loss": 3.9655,814 "step": 115815 },816 {817 "epoch": 0.018159048215403883,818 "grad_norm": 8.862527847290039,819 "learning_rate": 6.72e-05,820 "loss": 3.6875,821 "step": 116822 },823 {824 "epoch": 0.01831559173450219,825 "grad_norm": 21.191814422607422,826 "learning_rate": 6.78e-05,827 "loss": 4.9808,828 "step": 117829 },830 {831 "epoch": 0.018472135253600502,832 "grad_norm": 5.6609392166137695,833 "learning_rate": 6.84e-05,834 "loss": 3.446,835 "step": 118836 },837 {838 "epoch": 0.01862867877269881,839 "grad_norm": 15.799832344055176,840 "learning_rate": 6.9e-05,841 "loss": 4.3345,842 "step": 119843 },844 {845 "epoch": 0.01878522229179712,846 "grad_norm": 12.569608688354492,847 "learning_rate": 6.96e-05,848 "loss": 4.0145,849 "step": 120850 },851 {852 "epoch": 0.01894176581089543,853 "grad_norm": 6.250307559967041,854 "learning_rate": 7.02e-05,855 "loss": 3.531,856 "step": 121857 },858 {859 "epoch": 0.019098309329993738,860 "grad_norm": 9.823454856872559,861 "learning_rate": 7.079999999999999e-05,862 "loss": 3.8133,863 "step": 122864 },865 {866 "epoch": 0.019254852849092046,867 "grad_norm": 14.932891845703125,868 "learning_rate": 7.139999999999999e-05,869 "loss": 4.0484,870 "step": 123871 },872 {873 "epoch": 0.019411396368190358,874 "grad_norm": 8.835564613342285,875 "learning_rate": 7.199999999999999e-05,876 "loss": 3.775,877 "step": 124878 },879 {880 "epoch": 0.019567939887288666,881 "grad_norm": 2.77044939994812,882 "learning_rate": 7.259999999999999e-05,883 "loss": 3.385,884 "step": 125885 },886 {887 "epoch": 0.019724483406386977,888 "grad_norm": 19.207578659057617,889 "learning_rate": 7.319999999999999e-05,890 "loss": 4.4818,891 "step": 126892 },893 {894 "epoch": 0.019881026925485285,895 "grad_norm": 20.31056785583496,896 "learning_rate": 7.379999999999999e-05,897 "loss": 4.5493,898 "step": 127899 },900 {901 "epoch": 0.020037570444583593,902 "grad_norm": 9.711579322814941,903 "learning_rate": 7.439999999999999e-05,904 "loss": 3.7879,905 "step": 128906 },907 {908 "epoch": 0.020194113963681905,909 "grad_norm": 19.968996047973633,910 "learning_rate": 7.5e-05,911 "loss": 4.6096,912 "step": 129913 },914 {915 "epoch": 0.020350657482780213,916 "grad_norm": 13.640237808227539,917 "learning_rate": 7.56e-05,918 "loss": 4.1364,919 "step": 130920 },921 {922 "epoch": 0.02050720100187852,923 "grad_norm": 18.92345428466797,924 "learning_rate": 7.62e-05,925 "loss": 4.4219,926 "step": 131927 },928 {929 "epoch": 0.020663744520976832,930 "grad_norm": 5.510997295379639,931 "learning_rate": 7.68e-05,932 "loss": 3.615,933 "step": 132934 },935 {936 "epoch": 0.02082028804007514,937 "grad_norm": 11.286620140075684,938 "learning_rate": 7.74e-05,939 "loss": 4.0149,940 "step": 133941 },942 {943 "epoch": 0.020976831559173452,944 "grad_norm": 8.938575744628906,945 "learning_rate": 7.8e-05,946 "loss": 3.8435,947 "step": 134948 },949 {950 "epoch": 0.02113337507827176,951 "grad_norm": 8.202083587646484,952 "learning_rate": 7.86e-05,953 "loss": 3.8606,954 "step": 135955 },956 {957 "epoch": 0.021289918597370068,958 "grad_norm": 7.211292266845703,959 "learning_rate": 7.92e-05,960 "loss": 3.6637,961 "step": 136962 },963 {964 "epoch": 0.02144646211646838,965 "grad_norm": 6.565310001373291,966 "learning_rate": 7.98e-05,967 "loss": 3.602,968 "step": 137969 },970 {971 "epoch": 0.021603005635566688,972 "grad_norm": 5.222198009490967,973 "learning_rate": 8.04e-05,974 "loss": 3.6108,975 "step": 138976 },977 {978 "epoch": 0.021759549154664996,979 "grad_norm": 9.35428237915039,980 "learning_rate": 8.1e-05,981 "loss": 3.831,982 "step": 139983 },984 {985 "epoch": 0.021916092673763307,986 "grad_norm": 9.887560844421387,987 "learning_rate": 8.16e-05,988 "loss": 3.8316,989 "step": 140990 },991 {992 "epoch": 0.022072636192861615,993 "grad_norm": 8.195279121398926,994 "learning_rate": 8.22e-05,995 "loss": 3.642,996 "step": 141997 },998 {999 "epoch": 0.022229179711959923,1000 "grad_norm": 3.218400478363037,1001 "learning_rate": 8.28e-05,1002 "loss": 3.44,1003 "step": 1421004 },1005 {1006 "epoch": 0.022385723231058235,1007 "grad_norm": 2.472031354904175,1008 "learning_rate": 8.34e-05,1009 "loss": 3.4347,1010 "step": 1431011 },1012 {1013 "epoch": 0.022542266750156543,1014 "grad_norm": 2.141615390777588,1015 "learning_rate": 8.4e-05,1016 "loss": 3.3126,1017 "step": 1441018 },1019 {1020 "epoch": 0.022698810269254854,1021 "grad_norm": 3.026726007461548,1022 "learning_rate": 8.459999999999998e-05,1023 "loss": 3.3058,1024 "step": 1451025 },1026 {1027 "epoch": 0.022855353788353162,1028 "grad_norm": 3.966043472290039,1029 "learning_rate": 8.519999999999998e-05,1030 "loss": 3.3199,1031 "step": 1461032 },1033 {1034 "epoch": 0.02301189730745147,1035 "grad_norm": 3.906219720840454,1036 "learning_rate": 8.579999999999998e-05,1037 "loss": 3.2163,1038 "step": 1471039 },1040 {1041 "epoch": 0.023168440826549782,1042 "grad_norm": 2.3450818061828613,1043 "learning_rate": 8.639999999999999e-05,1044 "loss": 3.1645,1045 "step": 1481046 },1047 {1048 "epoch": 0.02332498434564809,1049 "grad_norm": 3.949995756149292,1050 "learning_rate": 8.699999999999999e-05,1051 "loss": 3.1981,1052 "step": 1491053 },1054 {1055 "epoch": 0.023481527864746398,1056 "grad_norm": 2.897047996520996,1057 "learning_rate": 8.759999999999999e-05,1058 "loss": 3.0328,1059 "step": 1501060 },1061 {1062 "epoch": 0.02363807138384471,1063 "grad_norm": 9.815628051757812,1064 "learning_rate": 8.819999999999999e-05,1065 "loss": 3.4079,1066 "step": 1511067 },1068 {1069 "epoch": 0.023794614902943018,1070 "grad_norm": 8.956120491027832,1071 "learning_rate": 8.879999999999999e-05,1072 "loss": 3.4144,1073 "step": 1521074 },1075 {1076 "epoch": 0.02395115842204133,1077 "grad_norm": 5.622773170471191,1078 "learning_rate": 8.939999999999999e-05,1079 "loss": 3.2383,1080 "step": 1531081 },1082 {1083 "epoch": 0.024107701941139637,1084 "grad_norm": 10.2916259765625,1085 "learning_rate": 8.999999999999999e-05,1086 "loss": 3.6527,1087 "step": 1541088 },1089 {1090 "epoch": 0.024264245460237945,1091 "grad_norm": 3.466508388519287,1092 "learning_rate": 9.059999999999999e-05,1093 "loss": 3.1562,1094 "step": 1551095 },1096 {1097 "epoch": 0.024420788979336257,1098 "grad_norm": 3.6558587551116943,1099 "learning_rate": 9.12e-05,1100 "loss": 3.1937,1101 "step": 1561102 },1103 {1104 "epoch": 0.024577332498434565,1105 "grad_norm": 1.539302945137024,1106 "learning_rate": 9.18e-05,1107 "loss": 3.0899,1108 "step": 1571109 },1110 {1111 "epoch": 0.024733876017532873,1112 "grad_norm": 4.4443230628967285,1113 "learning_rate": 9.24e-05,1114 "loss": 3.3159,1115 "step": 1581116 },1117 {1118 "epoch": 0.024890419536631184,1119 "grad_norm": 3.2545831203460693,1120 "learning_rate": 9.3e-05,1121 "loss": 3.1318,1122 "step": 1591123 },1124 {1125 "epoch": 0.025046963055729492,1126 "grad_norm": 3.192456007003784,1127 "learning_rate": 9.36e-05,1128 "loss": 3.2075,1129 "step": 1601130 },1131 {1132 "epoch": 0.0252035065748278,1133 "grad_norm": 1.9067258834838867,1134 "learning_rate": 9.419999999999999e-05,1135 "loss": 3.1137,1136 "step": 1611137 },1138 {1139 "epoch": 0.025360050093926112,1140 "grad_norm": 1.6764792203903198,1141 "learning_rate": 9.479999999999999e-05,1142 "loss": 3.109,1143 "step": 1621144 },1145 {1146 "epoch": 0.02551659361302442,1147 "grad_norm": 25.346240997314453,1148 "learning_rate": 9.539999999999999e-05,1149 "loss": 4.8797,1150 "step": 1631151 },1152 {1153 "epoch": 0.02567313713212273,1154 "grad_norm": 1.1866326332092285,1155 "learning_rate": 9.599999999999999e-05,1156 "loss": 3.0522,1157 "step": 1641158 },1159 {1160 "epoch": 0.02582968065122104,1161 "grad_norm": 1.4010124206542969,1162 "learning_rate": 9.659999999999999e-05,1163 "loss": 3.0789,1164 "step": 1651165 },1166 {1167 "epoch": 0.025986224170319348,1168 "grad_norm": 8.419206619262695,1169 "learning_rate": 9.719999999999999e-05,1170 "loss": 3.3802,1171 "step": 1661172 },1173 {1174 "epoch": 0.02614276768941766,1175 "grad_norm": 8.86681842803955,1176 "learning_rate": 9.779999999999999e-05,1177 "loss": 3.5289,1178 "step": 1671179 },1180 {1181 "epoch": 0.026299311208515967,1182 "grad_norm": 23.243457794189453,1183 "learning_rate": 9.839999999999999e-05,1184 "loss": 4.5939,1185 "step": 1681186 },1187 {1188 "epoch": 0.026455854727614275,1189 "grad_norm": 2.583829164505005,1190 "learning_rate": 9.9e-05,1191 "loss": 3.0461,1192 "step": 1691193 },1194 {1195 "epoch": 0.026612398246712587,1196 "grad_norm": 3.8143973350524902,1197 "learning_rate": 9.96e-05,1198 "loss": 3.1916,1199 "step": 1701200 },