RobotisSW/clean_table_60k
024
1{2 "best_global_step": null,3 "best_metric": null,4 "best_model_checkpoint": null,5 "epoch": 0.6,6 "eval_steps": 500,7 "global_step": 60000,8 "is_hyper_param_search": false,9 "is_local_process_zero": true,10 "is_world_process_zero": true,11 "log_history": [12 {13 "grad_norm": 0.926266074180603,14 "learning_rate": 1.8e-07,15 "loss": 1.4438,16 "step": 1017 },18 {19 "grad_norm": 1.0640310049057007,20 "learning_rate": 3.8e-07,21 "loss": 1.4395,22 "step": 2023 },24 {25 "grad_norm": 0.9207378625869751,26 "learning_rate": 5.8e-07,27 "loss": 1.4342,28 "step": 3029 },30 {31 "grad_norm": 0.8789049386978149,32 "learning_rate": 7.8e-07,33 "loss": 1.4412,34 "step": 4035 },36 {37 "grad_norm": 0.7615863680839539,38 "learning_rate": 9.8e-07,39 "loss": 1.4241,40 "step": 5041 },42 {43 "grad_norm": 1.0124503374099731,44 "learning_rate": 1.18e-06,45 "loss": 1.376,46 "step": 6047 },48 {49 "grad_norm": 0.8852088451385498,50 "learning_rate": 1.3800000000000001e-06,51 "loss": 1.3341,52 "step": 7053 },54 {55 "grad_norm": 1.1166917085647583,56 "learning_rate": 1.5800000000000003e-06,57 "loss": 1.287,58 "step": 8059 },60 {61 "grad_norm": 0.8359726071357727,62 "learning_rate": 1.7800000000000001e-06,63 "loss": 1.2729,64 "step": 9065 },66 {67 "grad_norm": 0.7087129354476929,68 "learning_rate": 1.98e-06,69 "loss": 1.2444,70 "step": 10071 },72 {73 "grad_norm": 0.8514461517333984,74 "learning_rate": 2.1800000000000003e-06,75 "loss": 1.2798,76 "step": 11077 },78 {79 "grad_norm": 1.2569960355758667,80 "learning_rate": 2.38e-06,81 "loss": 1.2389,82 "step": 12083 },84 {85 "grad_norm": 1.465353012084961,86 "learning_rate": 2.5800000000000003e-06,87 "loss": 1.2177,88 "step": 13089 },90 {91 "grad_norm": 2.254542589187622,92 "learning_rate": 2.78e-06,93 "loss": 1.2076,94 "step": 14095 },96 {97 "grad_norm": 1.0675272941589355,98 "learning_rate": 2.9800000000000003e-06,99 "loss": 1.1924,100 "step": 150101 },102 {103 "grad_norm": 1.3136998414993286,104 "learning_rate": 3.1800000000000005e-06,105 "loss": 1.1717,106 "step": 160107 },108 {109 "grad_norm": 1.7947893142700195,110 "learning_rate": 3.38e-06,111 "loss": 1.1604,112 "step": 170113 },114 {115 "grad_norm": 1.1186102628707886,116 "learning_rate": 3.58e-06,117 "loss": 1.1338,118 "step": 180119 },120 {121 "grad_norm": 1.1510117053985596,122 "learning_rate": 3.7800000000000002e-06,123 "loss": 1.1316,124 "step": 190125 },126 {127 "grad_norm": 1.0984177589416504,128 "learning_rate": 3.98e-06,129 "loss": 1.1084,130 "step": 200131 },132 {133 "grad_norm": 1.1552013158798218,134 "learning_rate": 4.18e-06,135 "loss": 1.1082,136 "step": 210137 },138 {139 "grad_norm": 1.0869241952896118,140 "learning_rate": 4.38e-06,141 "loss": 1.1082,142 "step": 220143 },144 {145 "grad_norm": 1.1602262258529663,146 "learning_rate": 4.58e-06,147 "loss": 1.111,148 "step": 230149 },150 {151 "grad_norm": 1.7610721588134766,152 "learning_rate": 4.780000000000001e-06,153 "loss": 1.0903,154 "step": 240155 },156 {157 "grad_norm": 1.2524158954620361,158 "learning_rate": 4.98e-06,159 "loss": 1.0972,160 "step": 250161 },162 {163 "grad_norm": 1.145118236541748,164 "learning_rate": 5.18e-06,165 "loss": 1.0818,166 "step": 260167 },168 {169 "grad_norm": 1.0128108263015747,170 "learning_rate": 5.38e-06,171 "loss": 1.0724,172 "step": 270173 },174 {175 "grad_norm": 1.351258635520935,176 "learning_rate": 5.580000000000001e-06,177 "loss": 1.0873,178 "step": 280179 },180 {181 "grad_norm": 1.1126656532287598,182 "learning_rate": 5.78e-06,183 "loss": 1.0808,184 "step": 290185 },186 {187 "grad_norm": 1.5978598594665527,188 "learning_rate": 5.98e-06,189 "loss": 1.0694,190 "step": 300191 },192 {193 "grad_norm": 1.2440999746322632,194 "learning_rate": 6.18e-06,195 "loss": 1.0426,196 "step": 310197 },198 {199 "grad_norm": 1.4261150360107422,200 "learning_rate": 6.38e-06,201 "loss": 1.0258,202 "step": 320203 },204 {205 "grad_norm": 1.6032270193099976,206 "learning_rate": 6.58e-06,207 "loss": 1.0008,208 "step": 330209 },210 {211 "grad_norm": 1.5931432247161865,212 "learning_rate": 6.78e-06,213 "loss": 0.9674,214 "step": 340215 },216 {217 "grad_norm": 1.5402252674102783,218 "learning_rate": 6.98e-06,219 "loss": 0.962,220 "step": 350221 },222 {223 "grad_norm": 1.5855474472045898,224 "learning_rate": 7.180000000000001e-06,225 "loss": 0.9422,226 "step": 360227 },228 {229 "grad_norm": 1.9631766080856323,230 "learning_rate": 7.3800000000000005e-06,231 "loss": 0.9142,232 "step": 370233 },234 {235 "grad_norm": 1.8945468664169312,236 "learning_rate": 7.580000000000001e-06,237 "loss": 0.8874,238 "step": 380239 },240 {241 "grad_norm": 1.8411067724227905,242 "learning_rate": 7.78e-06,243 "loss": 0.8613,244 "step": 390245 },246 {247 "grad_norm": 1.7587705850601196,248 "learning_rate": 7.98e-06,249 "loss": 0.8239,250 "step": 400251 },252 {253 "grad_norm": 1.9953256845474243,254 "learning_rate": 8.18e-06,255 "loss": 0.7818,256 "step": 410257 },258 {259 "grad_norm": 1.5480085611343384,260 "learning_rate": 8.380000000000001e-06,261 "loss": 0.744,262 "step": 420263 },264 {265 "grad_norm": 2.051807165145874,266 "learning_rate": 8.580000000000001e-06,267 "loss": 0.7039,268 "step": 430269 },270 {271 "grad_norm": 1.8827357292175293,272 "learning_rate": 8.78e-06,273 "loss": 0.6799,274 "step": 440275 },276 {277 "grad_norm": 2.2053680419921875,278 "learning_rate": 8.98e-06,279 "loss": 0.6308,280 "step": 450281 },282 {283 "grad_norm": 2.365339994430542,284 "learning_rate": 9.180000000000002e-06,285 "loss": 0.601,286 "step": 460287 },288 {289 "grad_norm": 2.4449708461761475,290 "learning_rate": 9.38e-06,291 "loss": 0.5866,292 "step": 470293 },294 {295 "grad_norm": 2.2257940769195557,296 "learning_rate": 9.58e-06,297 "loss": 0.544,298 "step": 480299 },300 {301 "grad_norm": 2.1845319271087646,302 "learning_rate": 9.78e-06,303 "loss": 0.5304,304 "step": 490305 },306 {307 "grad_norm": 1.8719619512557983,308 "learning_rate": 9.980000000000001e-06,309 "loss": 0.4801,310 "step": 500311 },312 {313 "grad_norm": 2.23685884475708,314 "learning_rate": 1.018e-05,315 "loss": 0.4509,316 "step": 510317 },318 {319 "grad_norm": 2.2305705547332764,320 "learning_rate": 1.038e-05,321 "loss": 0.4353,322 "step": 520323 },324 {325 "grad_norm": 2.8304765224456787,326 "learning_rate": 1.058e-05,327 "loss": 0.4199,328 "step": 530329 },330 {331 "grad_norm": 2.6049983501434326,332 "learning_rate": 1.0780000000000002e-05,333 "loss": 0.3829,334 "step": 540335 },336 {337 "grad_norm": 2.4628560543060303,338 "learning_rate": 1.098e-05,339 "loss": 0.3924,340 "step": 550341 },342 {343 "grad_norm": 3.0925612449645996,344 "learning_rate": 1.118e-05,345 "loss": 0.3626,346 "step": 560347 },348 {349 "grad_norm": 2.828927516937256,350 "learning_rate": 1.1380000000000001e-05,351 "loss": 0.3571,352 "step": 570353 },354 {355 "grad_norm": 2.2817559242248535,356 "learning_rate": 1.1580000000000001e-05,357 "loss": 0.3344,358 "step": 580359 },360 {361 "grad_norm": 2.1635260581970215,362 "learning_rate": 1.178e-05,363 "loss": 0.3044,364 "step": 590365 },366 {367 "grad_norm": 2.560818672180176,368 "learning_rate": 1.198e-05,369 "loss": 0.3004,370 "step": 600371 },372 {373 "grad_norm": 3.1146340370178223,374 "learning_rate": 1.2180000000000002e-05,375 "loss": 0.2827,376 "step": 610377 },378 {379 "grad_norm": 2.7388076782226562,380 "learning_rate": 1.238e-05,381 "loss": 0.302,382 "step": 620383 },384 {385 "grad_norm": 2.4787237644195557,386 "learning_rate": 1.258e-05,387 "loss": 0.2454,388 "step": 630389 },390 {391 "grad_norm": 2.952630043029785,392 "learning_rate": 1.278e-05,393 "loss": 0.2523,394 "step": 640395 },396 {397 "grad_norm": 2.6432156562805176,398 "learning_rate": 1.2980000000000001e-05,399 "loss": 0.2293,400 "step": 650401 },402 {403 "grad_norm": 2.473900318145752,404 "learning_rate": 1.3180000000000001e-05,405 "loss": 0.2358,406 "step": 660407 },408 {409 "grad_norm": 2.6395559310913086,410 "learning_rate": 1.338e-05,411 "loss": 0.2321,412 "step": 670413 },414 {415 "grad_norm": 2.060318946838379,416 "learning_rate": 1.358e-05,417 "loss": 0.2122,418 "step": 680419 },420 {421 "grad_norm": 2.3797101974487305,422 "learning_rate": 1.3780000000000002e-05,423 "loss": 0.2131,424 "step": 690425 },426 {427 "grad_norm": 2.041769027709961,428 "learning_rate": 1.3980000000000002e-05,429 "loss": 0.2049,430 "step": 700431 },432 {433 "grad_norm": 2.726841688156128,434 "learning_rate": 1.4180000000000001e-05,435 "loss": 0.2173,436 "step": 710437 },438 {439 "grad_norm": 1.9584563970565796,440 "learning_rate": 1.4380000000000001e-05,441 "loss": 0.197,442 "step": 720443 },444 {445 "grad_norm": 2.1229758262634277,446 "learning_rate": 1.4580000000000003e-05,447 "loss": 0.2028,448 "step": 730449 },450 {451 "grad_norm": 2.2316715717315674,452 "learning_rate": 1.4779999999999999e-05,453 "loss": 0.1857,454 "step": 740455 },456 {457 "grad_norm": 3.030278205871582,458 "learning_rate": 1.4979999999999999e-05,459 "loss": 0.2036,460 "step": 750461 },462 {463 "grad_norm": 2.147660732269287,464 "learning_rate": 1.518e-05,465 "loss": 0.1753,466 "step": 760467 },468 {469 "grad_norm": 2.1291427612304688,470 "learning_rate": 1.538e-05,471 "loss": 0.1953,472 "step": 770473 },474 {475 "grad_norm": 2.3734099864959717,476 "learning_rate": 1.558e-05,477 "loss": 0.1811,478 "step": 780479 },480 {481 "grad_norm": 2.0445516109466553,482 "learning_rate": 1.578e-05,483 "loss": 0.1846,484 "step": 790485 },486 {487 "grad_norm": 2.276503801345825,488 "learning_rate": 1.598e-05,489 "loss": 0.1781,490 "step": 800491 },492 {493 "grad_norm": 2.500742197036743,494 "learning_rate": 1.618e-05,495 "loss": 0.1903,496 "step": 810497 },498 {499 "grad_norm": 1.7059069871902466,500 "learning_rate": 1.6380000000000002e-05,501 "loss": 0.1798,502 "step": 820503 },504 {505 "grad_norm": 2.3454294204711914,506 "learning_rate": 1.658e-05,507 "loss": 0.1775,508 "step": 830509 },510 {511 "grad_norm": 2.3162667751312256,512 "learning_rate": 1.6780000000000002e-05,513 "loss": 0.1608,514 "step": 840515 },516 {517 "grad_norm": 1.9290380477905273,518 "learning_rate": 1.698e-05,519 "loss": 0.1591,520 "step": 850521 },522 {523 "grad_norm": 1.9581745862960815,524 "learning_rate": 1.718e-05,525 "loss": 0.1661,526 "step": 860527 },528 {529 "grad_norm": 2.471449375152588,530 "learning_rate": 1.7380000000000003e-05,531 "loss": 0.1685,532 "step": 870533 },534 {535 "grad_norm": 2.3924989700317383,536 "learning_rate": 1.758e-05,537 "loss": 0.1626,538 "step": 880539 },540 {541 "grad_norm": 2.1243438720703125,542 "learning_rate": 1.7780000000000003e-05,543 "loss": 0.1702,544 "step": 890545 },546 {547 "grad_norm": 2.313042402267456,548 "learning_rate": 1.798e-05,549 "loss": 0.1661,550 "step": 900551 },552 {553 "grad_norm": 2.1570398807525635,554 "learning_rate": 1.818e-05,555 "loss": 0.1663,556 "step": 910557 },558 {559 "grad_norm": 2.036301612854004,560 "learning_rate": 1.838e-05,561 "loss": 0.1446,562 "step": 920563 },564 {565 "grad_norm": 1.9931927919387817,566 "learning_rate": 1.858e-05,567 "loss": 0.1615,568 "step": 930569 },570 {571 "grad_norm": 1.81403648853302,572 "learning_rate": 1.878e-05,573 "loss": 0.1689,574 "step": 940575 },576 {577 "grad_norm": 2.0347235202789307,578 "learning_rate": 1.898e-05,579 "loss": 0.156,580 "step": 950581 },582 {583 "grad_norm": 2.040017604827881,584 "learning_rate": 1.918e-05,585 "loss": 0.1376,586 "step": 960587 },588 {589 "grad_norm": 1.9952945709228516,590 "learning_rate": 1.938e-05,591 "loss": 0.153,592 "step": 970593 },594 {595 "grad_norm": 2.2728240489959717,596 "learning_rate": 1.9580000000000002e-05,597 "loss": 0.1355,598 "step": 980599 },600 {601 "grad_norm": 2.187716245651245,602 "learning_rate": 1.978e-05,603 "loss": 0.1315,604 "step": 990605 },606 {607 "grad_norm": 1.5979294776916504,608 "learning_rate": 1.9980000000000002e-05,609 "loss": 0.1487,610 "step": 1000611 },612 {613 "grad_norm": 2.006293773651123,614 "learning_rate": 2.0180000000000003e-05,615 "loss": 0.127,616 "step": 1010617 },618 {619 "grad_norm": 1.860627293586731,620 "learning_rate": 2.038e-05,621 "loss": 0.1404,622 "step": 1020623 },624 {625 "grad_norm": 2.1587531566619873,626 "learning_rate": 2.0580000000000003e-05,627 "loss": 0.1466,628 "step": 1030629 },630 {631 "grad_norm": 1.531379222869873,632 "learning_rate": 2.078e-05,633 "loss": 0.1331,634 "step": 1040635 },636 {637 "grad_norm": 2.1895930767059326,638 "learning_rate": 2.098e-05,639 "loss": 0.1232,640 "step": 1050641 },642 {643 "grad_norm": 1.7790223360061646,644 "learning_rate": 2.118e-05,645 "loss": 0.1397,646 "step": 1060647 },648 {649 "grad_norm": 1.8040825128555298,650 "learning_rate": 2.138e-05,651 "loss": 0.1427,652 "step": 1070653 },654 {655 "grad_norm": 1.5641379356384277,656 "learning_rate": 2.158e-05,657 "loss": 0.1426,658 "step": 1080659 },660 {661 "grad_norm": 2.0407676696777344,662 "learning_rate": 2.178e-05,663 "loss": 0.1416,664 "step": 1090665 },666 {667 "grad_norm": 1.5827224254608154,668 "learning_rate": 2.198e-05,669 "loss": 0.1253,670 "step": 1100671 },672 {673 "grad_norm": 2.0948946475982666,674 "learning_rate": 2.218e-05,675 "loss": 0.1442,676 "step": 1110677 },678 {679 "grad_norm": 1.7712286710739136,680 "learning_rate": 2.2380000000000003e-05,681 "loss": 0.1337,682 "step": 1120683 },684 {685 "grad_norm": 1.352770447731018,686 "learning_rate": 2.258e-05,687 "loss": 0.1467,688 "step": 1130689 },690 {691 "grad_norm": 1.789804220199585,692 "learning_rate": 2.2780000000000002e-05,693 "loss": 0.1326,694 "step": 1140695 },696 {697 "grad_norm": 1.6404792070388794,698 "learning_rate": 2.298e-05,699 "loss": 0.1334,700 "step": 1150701 },702 {703 "grad_norm": 1.9014173746109009,704 "learning_rate": 2.318e-05,705 "loss": 0.1269,706 "step": 1160707 },708 {709 "grad_norm": 1.6180914640426636,710 "learning_rate": 2.3380000000000003e-05,711 "loss": 0.1231,712 "step": 1170713 },714 {715 "grad_norm": 1.6588298082351685,716 "learning_rate": 2.358e-05,717 "loss": 0.1258,718 "step": 1180719 },720 {721 "grad_norm": 1.6564593315124512,722 "learning_rate": 2.3780000000000003e-05,723 "loss": 0.122,724 "step": 1190725 },726 {727 "grad_norm": 1.6317307949066162,728 "learning_rate": 2.398e-05,729 "loss": 0.1341,730 "step": 1200731 },732 {733 "grad_norm": 1.6443976163864136,734 "learning_rate": 2.418e-05,735 "loss": 0.1138,736 "step": 1210737 },738 {739 "grad_norm": 1.5773783922195435,740 "learning_rate": 2.438e-05,741 "loss": 0.1214,742 "step": 1220743 },744 {745 "grad_norm": 1.4465614557266235,746 "learning_rate": 2.4580000000000002e-05,747 "loss": 0.1214,748 "step": 1230749 },750 {751 "grad_norm": 1.6430245637893677,752 "learning_rate": 2.478e-05,753 "loss": 0.115,754 "step": 1240755 },756 {757 "grad_norm": 1.7517200708389282,758 "learning_rate": 2.498e-05,759 "loss": 0.1172,760 "step": 1250761 },762 {763 "grad_norm": 1.3756705522537231,764 "learning_rate": 2.5180000000000003e-05,765 "loss": 0.123,766 "step": 1260767 },768 {769 "grad_norm": 1.6590244770050049,770 "learning_rate": 2.5380000000000004e-05,771 "loss": 0.1198,772 "step": 1270773 },774 {775 "grad_norm": 1.6160808801651,776 "learning_rate": 2.5580000000000002e-05,777 "loss": 0.1115,778 "step": 1280779 },780 {781 "grad_norm": 1.2673935890197754,782 "learning_rate": 2.5779999999999997e-05,783 "loss": 0.1154,784 "step": 1290785 },786 {787 "grad_norm": 1.4574971199035645,788 "learning_rate": 2.598e-05,789 "loss": 0.1136,790 "step": 1300791 },792 {793 "grad_norm": 1.2327100038528442,794 "learning_rate": 2.618e-05,795 "loss": 0.1172,796 "step": 1310797 },798 {799 "grad_norm": 1.3506114482879639,800 "learning_rate": 2.6379999999999998e-05,801 "loss": 0.1287,802 "step": 1320803 },804 {805 "grad_norm": 1.7393161058425903,806 "learning_rate": 2.658e-05,807 "loss": 0.1195,808 "step": 1330809 },810 {811 "grad_norm": 1.0298277139663696,812 "learning_rate": 2.678e-05,813 "loss": 0.1124,814 "step": 1340815 },816 {817 "grad_norm": 1.3873794078826904,818 "learning_rate": 2.698e-05,819 "loss": 0.1088,820 "step": 1350821 },822 {823 "grad_norm": 1.1892354488372803,824 "learning_rate": 2.718e-05,825 "loss": 0.1098,826 "step": 1360827 },828 {829 "grad_norm": 1.558323621749878,830 "learning_rate": 2.738e-05,831 "loss": 0.1094,832 "step": 1370833 },834 {835 "grad_norm": 1.1874359846115112,836 "learning_rate": 2.758e-05,837 "loss": 0.1178,838 "step": 1380839 },840 {841 "grad_norm": 1.5623217821121216,842 "learning_rate": 2.778e-05,843 "loss": 0.1179,844 "step": 1390845 },846 {847 "grad_norm": 1.7260891199111938,848 "learning_rate": 2.798e-05,849 "loss": 0.1052,850 "step": 1400851 },852 {853 "grad_norm": 1.6796718835830688,854 "learning_rate": 2.818e-05,855 "loss": 0.1219,856 "step": 1410857 },858 {859 "grad_norm": 1.7223145961761475,860 "learning_rate": 2.8380000000000003e-05,861 "loss": 0.1226,862 "step": 1420863 },864 {865 "grad_norm": 1.462886929512024,866 "learning_rate": 2.858e-05,867 "loss": 0.1107,868 "step": 1430869 },870 {871 "grad_norm": 1.3069902658462524,872 "learning_rate": 2.8780000000000002e-05,873 "loss": 0.1177,874 "step": 1440875 },876 {877 "grad_norm": 1.3319061994552612,878 "learning_rate": 2.898e-05,879 "loss": 0.099,880 "step": 1450881 },882 {883 "grad_norm": 1.3519988059997559,884 "learning_rate": 2.9180000000000002e-05,885 "loss": 0.1116,886 "step": 1460887 },888 {889 "grad_norm": 1.008862853050232,890 "learning_rate": 2.9380000000000003e-05,891 "loss": 0.1095,892 "step": 1470893 },894 {895 "grad_norm": 1.6593354940414429,896 "learning_rate": 2.958e-05,897 "loss": 0.1175,898 "step": 1480899 },900 {901 "grad_norm": 1.3916335105895996,902 "learning_rate": 2.9780000000000003e-05,903 "loss": 0.1181,904 "step": 1490905 },906 {907 "grad_norm": 1.650828242301941,908 "learning_rate": 2.998e-05,909 "loss": 0.1146,910 "step": 1500911 },912 {913 "grad_norm": 1.2685716152191162,914 "learning_rate": 3.0180000000000002e-05,915 "loss": 0.1176,916 "step": 1510917 },918 {919 "grad_norm": 1.510684609413147,920 "learning_rate": 3.0380000000000004e-05,921 "loss": 0.112,922 "step": 1520923 },924 {925 "grad_norm": 1.432042121887207,926 "learning_rate": 3.058e-05,927 "loss": 0.1113,928 "step": 1530929 },930 {931 "grad_norm": 1.7365586757659912,932 "learning_rate": 3.078e-05,933 "loss": 0.1009,934 "step": 1540935 },936 {937 "grad_norm": 1.3169975280761719,938 "learning_rate": 3.0980000000000005e-05,939 "loss": 0.1088,940 "step": 1550941 },942 {943 "grad_norm": 1.4847275018692017,944 "learning_rate": 3.118e-05,945 "loss": 0.11,946 "step": 1560947 },948 {949 "grad_norm": 1.5091370344161987,950 "learning_rate": 3.138e-05,951 "loss": 0.1107,952 "step": 1570953 },954 {955 "grad_norm": 1.3874123096466064,956 "learning_rate": 3.1580000000000006e-05,957 "loss": 0.1096,958 "step": 1580959 },960 {961 "grad_norm": 1.6959023475646973,962 "learning_rate": 3.1780000000000004e-05,963 "loss": 0.1058,964 "step": 1590965 },966 {967 "grad_norm": 1.6006110906600952,968 "learning_rate": 3.198e-05,969 "loss": 0.1139,970 "step": 1600971 },972 {973 "grad_norm": 1.3088146448135376,974 "learning_rate": 3.218e-05,975 "loss": 0.1072,976 "step": 1610977 },978 {979 "grad_norm": 1.1733874082565308,980 "learning_rate": 3.238e-05,981 "loss": 0.1047,982 "step": 1620983 },984 {985 "grad_norm": 1.708408236503601,986 "learning_rate": 3.2579999999999996e-05,987 "loss": 0.1037,988 "step": 1630989 },990 {991 "grad_norm": 1.2776528596878052,992 "learning_rate": 3.278e-05,993 "loss": 0.1053,994 "step": 1640995 },996 {997 "grad_norm": 1.4028071165084839,998 "learning_rate": 3.298e-05,999 "loss": 0.1046,1000 "step": 16501001 },1002 {1003 "grad_norm": 1.1378097534179688,1004 "learning_rate": 3.318e-05,1005 "loss": 0.1055,1006 "step": 16601007 },1008 {1009 "grad_norm": 1.6558804512023926,1010 "learning_rate": 3.338e-05,1011 "loss": 0.1093,1012 "step": 16701013 },1014 {1015 "grad_norm": 1.1552854776382446,1016 "learning_rate": 3.358e-05,1017 "loss": 0.0966,1018 "step": 16801019 },1020 {1021 "grad_norm": 1.4640536308288574,1022 "learning_rate": 3.378e-05,1023 "loss": 0.1098,1024 "step": 16901025 },1026 {1027 "grad_norm": 1.322249174118042,1028 "learning_rate": 3.398e-05,1029 "loss": 0.103,1030 "step": 17001031 },1032 {1033 "grad_norm": 1.3498573303222656,1034 "learning_rate": 3.418e-05,1035 "loss": 0.1097,1036 "step": 17101037 },1038 {1039 "grad_norm": 1.262957215309143,1040 "learning_rate": 3.438e-05,1041 "loss": 0.1034,1042 "step": 17201043 },1044 {1045 "grad_norm": 1.4252980947494507,1046 "learning_rate": 3.4580000000000004e-05,1047 "loss": 0.1028,1048 "step": 17301049 },1050 {1051 "grad_norm": 1.2226507663726807,1052 "learning_rate": 3.478e-05,1053 "loss": 0.0976,1054 "step": 17401055 },1056 {1057 "grad_norm": 1.3647156953811646,1058 "learning_rate": 3.498e-05,1059 "loss": 0.1039,1060 "step": 17501061 },1062 {1063 "grad_norm": 1.2517162561416626,1064 "learning_rate": 3.518e-05,1065 "loss": 0.1066,1066 "step": 17601067 },1068 {1069 "grad_norm": 0.861154317855835,1070 "learning_rate": 3.5380000000000003e-05,1071 "loss": 0.0944,1072 "step": 17701073 },1074 {1075 "grad_norm": 1.3860739469528198,1076 "learning_rate": 3.558e-05,1077 "loss": 0.1087,1078 "step": 17801079 },1080 {1081 "grad_norm": 1.2621777057647705,1082 "learning_rate": 3.578e-05,1083 "loss": 0.0947,1084 "step": 17901085 },1086 {1087 "grad_norm": 1.1672967672348022,1088 "learning_rate": 3.5980000000000004e-05,1089 "loss": 0.0979,1090 "step": 18001091 },1092 {1093 "grad_norm": 1.1447454690933228,1094 "learning_rate": 3.618e-05,1095 "loss": 0.0967,1096 "step": 18101097 },1098 {1099 "grad_norm": 1.183205485343933,1100 "learning_rate": 3.638e-05,1101 "loss": 0.0962,1102 "step": 18201103 },1104 {1105 "grad_norm": 1.0532835721969604,1106 "learning_rate": 3.6580000000000006e-05,1107 "loss": 0.0954,1108 "step": 18301109 },1110 {1111 "grad_norm": 1.1037495136260986,1112 "learning_rate": 3.6780000000000004e-05,1113 "loss": 0.0971,1114 "step": 18401115 },1116 {1117 "grad_norm": 1.0365946292877197,1118 "learning_rate": 3.698e-05,1119 "loss": 0.0842,1120 "step": 18501121 },1122 {1123 "grad_norm": 1.3786957263946533,1124 "learning_rate": 3.7180000000000007e-05,1125 "loss": 0.0956,1126 "step": 18601127 },1128 {1129 "grad_norm": 1.1241066455841064,1130 "learning_rate": 3.7380000000000005e-05,1131 "loss": 0.0938,1132 "step": 18701133 },1134 {1135 "grad_norm": 1.217797040939331,1136 "learning_rate": 3.758e-05,1137 "loss": 0.0885,1138 "step": 18801139 },1140 {1141 "grad_norm": 1.19753098487854,1142 "learning_rate": 3.778000000000001e-05,1143 "loss": 0.0976,1144 "step": 18901145 },1146 {1147 "grad_norm": 1.3078787326812744,1148 "learning_rate": 3.7980000000000006e-05,1149 "loss": 0.101,1150 "step": 19001151 },1152 {1153 "grad_norm": 1.1819777488708496,1154 "learning_rate": 3.818e-05,1155 "loss": 0.0995,1156 "step": 19101157 },1158 {1159 "grad_norm": 1.1380770206451416,1160 "learning_rate": 3.838e-05,1161 "loss": 0.1035,1162 "step": 19201163 },1164 {1165 "grad_norm": 1.1521198749542236,1166 "learning_rate": 3.858e-05,1167 "loss": 0.1001,1168 "step": 19301169 },1170 {1171 "grad_norm": 1.1119909286499023,1172 "learning_rate": 3.878e-05,1173 "loss": 0.1035,1174 "step": 19401175 },1176 {1177 "grad_norm": 1.0273287296295166,1178 "learning_rate": 3.898e-05,1179 "loss": 0.1018,1180 "step": 19501181 },1182 {1183 "grad_norm": 1.1000126600265503,1184 "learning_rate": 3.918e-05,1185 "loss": 0.1093,1186 "step": 19601187 },1188 {1189 "grad_norm": 0.8933774828910828,1190 "learning_rate": 3.938e-05,1191 "loss": 0.1103,1192 "step": 19701193 },1194 {1195 "grad_norm": 1.1168996095657349,1196 "learning_rate": 3.958e-05,1197 "loss": 0.0951,1198 "step": 19801199 },1200 {