ROBOTIS/GR00T-N1.5_PackageClassification_model
09
1{2 "best_global_step": null,3 "best_metric": null,4 "best_model_checkpoint": null,5 "epoch": 531.25,6 "eval_steps": 500,7 "global_step": 8500,8 "is_hyper_param_search": false,9 "is_local_process_zero": true,10 "is_world_process_zero": true,11 "log_history": [12 {13 "epoch": 0.625,14 "grad_norm": 0.9679650664329529,15 "learning_rate": 1.8e-07,16 "loss": 1.4147,17 "step": 1018 },19 {20 "epoch": 1.25,21 "grad_norm": 0.9715470671653748,22 "learning_rate": 3.8e-07,23 "loss": 1.4102,24 "step": 2025 },26 {27 "epoch": 1.875,28 "grad_norm": 0.9312183260917664,29 "learning_rate": 5.8e-07,30 "loss": 1.4117,31 "step": 3032 },33 {34 "epoch": 2.5,35 "grad_norm": 0.9487130641937256,36 "learning_rate": 7.8e-07,37 "loss": 1.4086,38 "step": 4039 },40 {41 "epoch": 3.125,42 "grad_norm": 0.8825913071632385,43 "learning_rate": 9.8e-07,44 "loss": 1.3962,45 "step": 5046 },47 {48 "epoch": 3.75,49 "grad_norm": 0.8918140530586243,50 "learning_rate": 1.18e-06,51 "loss": 1.3851,52 "step": 6053 },54 {55 "epoch": 4.375,56 "grad_norm": 0.8738917708396912,57 "learning_rate": 1.3800000000000001e-06,58 "loss": 1.3686,59 "step": 7060 },61 {62 "epoch": 5.0,63 "grad_norm": 0.9906870126724243,64 "learning_rate": 1.5800000000000003e-06,65 "loss": 1.3532,66 "step": 8067 },68 {69 "epoch": 5.625,70 "grad_norm": 1.1455429792404175,71 "learning_rate": 1.7800000000000001e-06,72 "loss": 1.3182,73 "step": 9074 },75 {76 "epoch": 6.25,77 "grad_norm": 1.236258864402771,78 "learning_rate": 1.98e-06,79 "loss": 1.2925,80 "step": 10081 },82 {83 "epoch": 6.875,84 "grad_norm": 1.327654242515564,85 "learning_rate": 2.1800000000000003e-06,86 "loss": 1.258,87 "step": 11088 },89 {90 "epoch": 7.5,91 "grad_norm": 1.2924314737319946,92 "learning_rate": 2.38e-06,93 "loss": 1.2277,94 "step": 12095 },96 {97 "epoch": 8.125,98 "grad_norm": 1.167385220527649,99 "learning_rate": 2.5800000000000003e-06,100 "loss": 1.1915,101 "step": 130102 },103 {104 "epoch": 8.75,105 "grad_norm": 0.9448131322860718,106 "learning_rate": 2.78e-06,107 "loss": 1.1615,108 "step": 140109 },110 {111 "epoch": 9.375,112 "grad_norm": 0.6841535568237305,113 "learning_rate": 2.9800000000000003e-06,114 "loss": 1.1314,115 "step": 150116 },117 {118 "epoch": 10.0,119 "grad_norm": 0.41633787751197815,120 "learning_rate": 3.1800000000000005e-06,121 "loss": 1.1138,122 "step": 160123 },124 {125 "epoch": 10.625,126 "grad_norm": 0.3072815537452698,127 "learning_rate": 3.38e-06,128 "loss": 1.1009,129 "step": 170130 },131 {132 "epoch": 11.25,133 "grad_norm": 0.28385525941848755,134 "learning_rate": 3.58e-06,135 "loss": 1.0909,136 "step": 180137 },138 {139 "epoch": 11.875,140 "grad_norm": 0.26593610644340515,141 "learning_rate": 3.7800000000000002e-06,142 "loss": 1.0844,143 "step": 190144 },145 {146 "epoch": 12.5,147 "grad_norm": 0.2532358169555664,148 "learning_rate": 3.98e-06,149 "loss": 1.0767,150 "step": 200151 },152 {153 "epoch": 13.125,154 "grad_norm": 0.2549743354320526,155 "learning_rate": 4.18e-06,156 "loss": 1.0693,157 "step": 210158 },159 {160 "epoch": 13.75,161 "grad_norm": 0.22734442353248596,162 "learning_rate": 4.38e-06,163 "loss": 1.0659,164 "step": 220165 },166 {167 "epoch": 14.375,168 "grad_norm": 0.22863343358039856,169 "learning_rate": 4.58e-06,170 "loss": 1.0584,171 "step": 230172 },173 {174 "epoch": 15.0,175 "grad_norm": 0.22417008876800537,176 "learning_rate": 4.780000000000001e-06,177 "loss": 1.0538,178 "step": 240179 },180 {181 "epoch": 15.625,182 "grad_norm": 0.20044955611228943,183 "learning_rate": 4.98e-06,184 "loss": 1.0501,185 "step": 250186 },187 {188 "epoch": 16.25,189 "grad_norm": 0.20292679965496063,190 "learning_rate": 5.18e-06,191 "loss": 1.0471,192 "step": 260193 },194 {195 "epoch": 16.875,196 "grad_norm": 0.18847720324993134,197 "learning_rate": 5.38e-06,198 "loss": 1.0408,199 "step": 270200 },201 {202 "epoch": 17.5,203 "grad_norm": 0.16966596245765686,204 "learning_rate": 5.580000000000001e-06,205 "loss": 1.0372,206 "step": 280207 },208 {209 "epoch": 18.125,210 "grad_norm": 0.189141184091568,211 "learning_rate": 5.78e-06,212 "loss": 1.0352,213 "step": 290214 },215 {216 "epoch": 18.75,217 "grad_norm": 0.18442021310329437,218 "learning_rate": 5.98e-06,219 "loss": 1.0313,220 "step": 300221 },222 {223 "epoch": 19.375,224 "grad_norm": 0.1687787026166916,225 "learning_rate": 6.18e-06,226 "loss": 1.0297,227 "step": 310228 },229 {230 "epoch": 20.0,231 "grad_norm": 0.20706340670585632,232 "learning_rate": 6.38e-06,233 "loss": 1.0266,234 "step": 320235 },236 {237 "epoch": 20.625,238 "grad_norm": 0.17096994817256927,239 "learning_rate": 6.58e-06,240 "loss": 1.0222,241 "step": 330242 },243 {244 "epoch": 21.25,245 "grad_norm": 0.18136513233184814,246 "learning_rate": 6.78e-06,247 "loss": 1.0181,248 "step": 340249 },250 {251 "epoch": 21.875,252 "grad_norm": 0.18053822219371796,253 "learning_rate": 6.98e-06,254 "loss": 1.0154,255 "step": 350256 },257 {258 "epoch": 22.5,259 "grad_norm": 0.2154332548379898,260 "learning_rate": 7.180000000000001e-06,261 "loss": 1.0116,262 "step": 360263 },264 {265 "epoch": 23.125,266 "grad_norm": 0.23920832574367523,267 "learning_rate": 7.3800000000000005e-06,268 "loss": 1.0056,269 "step": 370270 },271 {272 "epoch": 23.75,273 "grad_norm": 0.24692080914974213,274 "learning_rate": 7.580000000000001e-06,275 "loss": 0.9982,276 "step": 380277 },278 {279 "epoch": 24.375,280 "grad_norm": 0.32462307810783386,281 "learning_rate": 7.78e-06,282 "loss": 0.988,283 "step": 390284 },285 {286 "epoch": 25.0,287 "grad_norm": 0.3690284788608551,288 "learning_rate": 7.98e-06,289 "loss": 0.9721,290 "step": 400291 },292 {293 "epoch": 25.625,294 "grad_norm": 0.44107159972190857,295 "learning_rate": 8.18e-06,296 "loss": 0.9429,297 "step": 410298 },299 {300 "epoch": 26.25,301 "grad_norm": 0.5854696035385132,302 "learning_rate": 8.380000000000001e-06,303 "loss": 0.8986,304 "step": 420305 },306 {307 "epoch": 26.875,308 "grad_norm": 0.7961719632148743,309 "learning_rate": 8.580000000000001e-06,310 "loss": 0.8538,311 "step": 430312 },313 {314 "epoch": 27.5,315 "grad_norm": 0.9634988307952881,316 "learning_rate": 8.78e-06,317 "loss": 0.8043,318 "step": 440319 },320 {321 "epoch": 28.125,322 "grad_norm": 0.8321980834007263,323 "learning_rate": 8.98e-06,324 "loss": 0.7597,325 "step": 450326 },327 {328 "epoch": 28.75,329 "grad_norm": 0.784283459186554,330 "learning_rate": 9.180000000000002e-06,331 "loss": 0.7166,332 "step": 460333 },334 {335 "epoch": 29.375,336 "grad_norm": 0.8551204800605774,337 "learning_rate": 9.38e-06,338 "loss": 0.6809,339 "step": 470340 },341 {342 "epoch": 30.0,343 "grad_norm": 0.774193525314331,344 "learning_rate": 9.58e-06,345 "loss": 0.6511,346 "step": 480347 },348 {349 "epoch": 30.625,350 "grad_norm": 0.8994619846343994,351 "learning_rate": 9.78e-06,352 "loss": 0.6267,353 "step": 490354 },355 {356 "epoch": 31.25,357 "grad_norm": 1.0884357690811157,358 "learning_rate": 9.980000000000001e-06,359 "loss": 0.6061,360 "step": 500361 },362 {363 "epoch": 31.875,364 "grad_norm": 1.0175591707229614,365 "learning_rate": 1.018e-05,366 "loss": 0.5858,367 "step": 510368 },369 {370 "epoch": 32.5,371 "grad_norm": 1.1407521963119507,372 "learning_rate": 1.038e-05,373 "loss": 0.5714,374 "step": 520375 },376 {377 "epoch": 33.125,378 "grad_norm": 1.042667269706726,379 "learning_rate": 1.058e-05,380 "loss": 0.5586,381 "step": 530382 },383 {384 "epoch": 33.75,385 "grad_norm": 0.9843167662620544,386 "learning_rate": 1.0780000000000002e-05,387 "loss": 0.5492,388 "step": 540389 },390 {391 "epoch": 34.375,392 "grad_norm": 1.0415880680084229,393 "learning_rate": 1.098e-05,394 "loss": 0.5431,395 "step": 550396 },397 {398 "epoch": 35.0,399 "grad_norm": 1.1198561191558838,400 "learning_rate": 1.118e-05,401 "loss": 0.5375,402 "step": 560403 },404 {405 "epoch": 35.625,406 "grad_norm": 1.2082597017288208,407 "learning_rate": 1.1380000000000001e-05,408 "loss": 0.5338,409 "step": 570410 },411 {412 "epoch": 36.25,413 "grad_norm": 1.2071870565414429,414 "learning_rate": 1.1580000000000001e-05,415 "loss": 0.5298,416 "step": 580417 },418 {419 "epoch": 36.875,420 "grad_norm": 1.3737869262695312,421 "learning_rate": 1.178e-05,422 "loss": 0.5287,423 "step": 590424 },425 {426 "epoch": 37.5,427 "grad_norm": 1.0583442449569702,428 "learning_rate": 1.198e-05,429 "loss": 0.5243,430 "step": 600431 },432 {433 "epoch": 38.125,434 "grad_norm": 1.0685936212539673,435 "learning_rate": 1.2180000000000002e-05,436 "loss": 0.5229,437 "step": 610438 },439 {440 "epoch": 38.75,441 "grad_norm": 1.2425378561019897,442 "learning_rate": 1.238e-05,443 "loss": 0.5215,444 "step": 620445 },446 {447 "epoch": 39.375,448 "grad_norm": 1.1828131675720215,449 "learning_rate": 1.258e-05,450 "loss": 0.5194,451 "step": 630452 },453 {454 "epoch": 40.0,455 "grad_norm": 1.0766721963882446,456 "learning_rate": 1.278e-05,457 "loss": 0.519,458 "step": 640459 },460 {461 "epoch": 40.625,462 "grad_norm": 1.0481817722320557,463 "learning_rate": 1.2980000000000001e-05,464 "loss": 0.5154,465 "step": 650466 },467 {468 "epoch": 41.25,469 "grad_norm": 1.0572658777236938,470 "learning_rate": 1.3180000000000001e-05,471 "loss": 0.5159,472 "step": 660473 },474 {475 "epoch": 41.875,476 "grad_norm": 1.203803539276123,477 "learning_rate": 1.338e-05,478 "loss": 0.5121,479 "step": 670480 },481 {482 "epoch": 42.5,483 "grad_norm": 0.872924268245697,484 "learning_rate": 1.358e-05,485 "loss": 0.512,486 "step": 680487 },488 {489 "epoch": 43.125,490 "grad_norm": 1.162695050239563,491 "learning_rate": 1.3780000000000002e-05,492 "loss": 0.5064,493 "step": 690494 },495 {496 "epoch": 43.75,497 "grad_norm": 1.1628501415252686,498 "learning_rate": 1.3980000000000002e-05,499 "loss": 0.5082,500 "step": 700501 },502 {503 "epoch": 44.375,504 "grad_norm": 1.217319369316101,505 "learning_rate": 1.4180000000000001e-05,506 "loss": 0.5067,507 "step": 710508 },509 {510 "epoch": 45.0,511 "grad_norm": 1.1384787559509277,512 "learning_rate": 1.4380000000000001e-05,513 "loss": 0.506,514 "step": 720515 },516 {517 "epoch": 45.625,518 "grad_norm": 1.2870819568634033,519 "learning_rate": 1.4580000000000003e-05,520 "loss": 0.505,521 "step": 730522 },523 {524 "epoch": 46.25,525 "grad_norm": 1.1514242887496948,526 "learning_rate": 1.4779999999999999e-05,527 "loss": 0.5064,528 "step": 740529 },530 {531 "epoch": 46.875,532 "grad_norm": 1.144319772720337,533 "learning_rate": 1.4979999999999999e-05,534 "loss": 0.5017,535 "step": 750536 },537 {538 "epoch": 47.5,539 "grad_norm": 1.22275710105896,540 "learning_rate": 1.518e-05,541 "loss": 0.5013,542 "step": 760543 },544 {545 "epoch": 48.125,546 "grad_norm": 1.118972659111023,547 "learning_rate": 1.538e-05,548 "loss": 0.4997,549 "step": 770550 },551 {552 "epoch": 48.75,553 "grad_norm": 1.051975965499878,554 "learning_rate": 1.558e-05,555 "loss": 0.5008,556 "step": 780557 },558 {559 "epoch": 49.375,560 "grad_norm": 1.4525421857833862,561 "learning_rate": 1.578e-05,562 "loss": 0.4996,563 "step": 790564 },565 {566 "epoch": 50.0,567 "grad_norm": 1.183190941810608,568 "learning_rate": 1.598e-05,569 "loss": 0.4963,570 "step": 800571 },572 {573 "epoch": 50.625,574 "grad_norm": 1.2612169981002808,575 "learning_rate": 1.618e-05,576 "loss": 0.4974,577 "step": 810578 },579 {580 "epoch": 51.25,581 "grad_norm": 0.9744483828544617,582 "learning_rate": 1.6380000000000002e-05,583 "loss": 0.4954,584 "step": 820585 },586 {587 "epoch": 51.875,588 "grad_norm": 1.1227869987487793,589 "learning_rate": 1.658e-05,590 "loss": 0.4959,591 "step": 830592 },593 {594 "epoch": 52.5,595 "grad_norm": 1.043039321899414,596 "learning_rate": 1.6780000000000002e-05,597 "loss": 0.4927,598 "step": 840599 },600 {601 "epoch": 53.125,602 "grad_norm": 1.2920984029769897,603 "learning_rate": 1.698e-05,604 "loss": 0.4918,605 "step": 850606 },607 {608 "epoch": 53.75,609 "grad_norm": 1.2774907350540161,610 "learning_rate": 1.718e-05,611 "loss": 0.4907,612 "step": 860613 },614 {615 "epoch": 54.375,616 "grad_norm": 1.1010966300964355,617 "learning_rate": 1.7380000000000003e-05,618 "loss": 0.4914,619 "step": 870620 },621 {622 "epoch": 55.0,623 "grad_norm": 1.0284886360168457,624 "learning_rate": 1.758e-05,625 "loss": 0.4873,626 "step": 880627 },628 {629 "epoch": 55.625,630 "grad_norm": 1.3238409757614136,631 "learning_rate": 1.7780000000000003e-05,632 "loss": 0.4883,633 "step": 890634 },635 {636 "epoch": 56.25,637 "grad_norm": 1.127139687538147,638 "learning_rate": 1.798e-05,639 "loss": 0.4871,640 "step": 900641 },642 {643 "epoch": 56.875,644 "grad_norm": 1.3855187892913818,645 "learning_rate": 1.818e-05,646 "loss": 0.4828,647 "step": 910648 },649 {650 "epoch": 57.5,651 "grad_norm": 1.536786437034607,652 "learning_rate": 1.838e-05,653 "loss": 0.4812,654 "step": 920655 },656 {657 "epoch": 58.125,658 "grad_norm": 2.120702028274536,659 "learning_rate": 1.858e-05,660 "loss": 0.4776,661 "step": 930662 },663 {664 "epoch": 58.75,665 "grad_norm": 1.6191856861114502,666 "learning_rate": 1.878e-05,667 "loss": 0.4761,668 "step": 940669 },670 {671 "epoch": 59.375,672 "grad_norm": 1.6021475791931152,673 "learning_rate": 1.898e-05,674 "loss": 0.4743,675 "step": 950676 },677 {678 "epoch": 60.0,679 "grad_norm": 1.604246735572815,680 "learning_rate": 1.918e-05,681 "loss": 0.4705,682 "step": 960683 },684 {685 "epoch": 60.625,686 "grad_norm": 3.4168691635131836,687 "learning_rate": 1.938e-05,688 "loss": 0.4732,689 "step": 970690 },691 {692 "epoch": 61.25,693 "grad_norm": 1.997071623802185,694 "learning_rate": 1.9580000000000002e-05,695 "loss": 0.4709,696 "step": 980697 },698 {699 "epoch": 61.875,700 "grad_norm": 2.351092576980591,701 "learning_rate": 1.978e-05,702 "loss": 0.4678,703 "step": 990704 },705 {706 "epoch": 62.5,707 "grad_norm": 1.9072397947311401,708 "learning_rate": 1.9980000000000002e-05,709 "loss": 0.469,710 "step": 1000711 },712 {713 "epoch": 63.125,714 "grad_norm": 1.777255892753601,715 "learning_rate": 2.0180000000000003e-05,716 "loss": 0.4642,717 "step": 1010718 },719 {720 "epoch": 63.75,721 "grad_norm": 1.3945808410644531,722 "learning_rate": 2.038e-05,723 "loss": 0.4613,724 "step": 1020725 },726 {727 "epoch": 64.375,728 "grad_norm": 1.5424234867095947,729 "learning_rate": 2.0580000000000003e-05,730 "loss": 0.4621,731 "step": 1030732 },733 {734 "epoch": 65.0,735 "grad_norm": 1.4399698972702026,736 "learning_rate": 2.078e-05,737 "loss": 0.4596,738 "step": 1040739 },740 {741 "epoch": 65.625,742 "grad_norm": 1.5211741924285889,743 "learning_rate": 2.098e-05,744 "loss": 0.4571,745 "step": 1050746 },747 {748 "epoch": 66.25,749 "grad_norm": 1.6879644393920898,750 "learning_rate": 2.118e-05,751 "loss": 0.4535,752 "step": 1060753 },754 {755 "epoch": 66.875,756 "grad_norm": 1.7169924974441528,757 "learning_rate": 2.138e-05,758 "loss": 0.4506,759 "step": 1070760 },761 {762 "epoch": 67.5,763 "grad_norm": 2.5419564247131348,764 "learning_rate": 2.158e-05,765 "loss": 0.4497,766 "step": 1080767 },768 {769 "epoch": 68.125,770 "grad_norm": 1.9424971342086792,771 "learning_rate": 2.178e-05,772 "loss": 0.446,773 "step": 1090774 },775 {776 "epoch": 68.75,777 "grad_norm": 4.030938148498535,778 "learning_rate": 2.198e-05,779 "loss": 0.4459,780 "step": 1100781 },782 {783 "epoch": 69.375,784 "grad_norm": 2.3670308589935303,785 "learning_rate": 2.218e-05,786 "loss": 0.4454,787 "step": 1110788 },789 {790 "epoch": 70.0,791 "grad_norm": 2.562795400619507,792 "learning_rate": 2.2380000000000003e-05,793 "loss": 0.4397,794 "step": 1120795 },796 {797 "epoch": 70.625,798 "grad_norm": 2.4600791931152344,799 "learning_rate": 2.258e-05,800 "loss": 0.4359,801 "step": 1130802 },803 {804 "epoch": 71.25,805 "grad_norm": 2.77167010307312,806 "learning_rate": 2.2780000000000002e-05,807 "loss": 0.4325,808 "step": 1140809 },810 {811 "epoch": 71.875,812 "grad_norm": 3.150618314743042,813 "learning_rate": 2.298e-05,814 "loss": 0.4285,815 "step": 1150816 },817 {818 "epoch": 72.5,819 "grad_norm": 2.5932984352111816,820 "learning_rate": 2.318e-05,821 "loss": 0.4247,822 "step": 1160823 },824 {825 "epoch": 73.125,826 "grad_norm": 2.8484175205230713,827 "learning_rate": 2.3380000000000003e-05,828 "loss": 0.4199,829 "step": 1170830 },831 {832 "epoch": 73.75,833 "grad_norm": 3.6509993076324463,834 "learning_rate": 2.358e-05,835 "loss": 0.4183,836 "step": 1180837 },838 {839 "epoch": 74.375,840 "grad_norm": 5.74982213973999,841 "learning_rate": 2.3780000000000003e-05,842 "loss": 0.4146,843 "step": 1190844 },845 {846 "epoch": 75.0,847 "grad_norm": 5.700360298156738,848 "learning_rate": 2.398e-05,849 "loss": 0.4322,850 "step": 1200851 },852 {853 "epoch": 75.625,854 "grad_norm": 3.3000857830047607,855 "learning_rate": 2.418e-05,856 "loss": 0.4228,857 "step": 1210858 },859 {860 "epoch": 76.25,861 "grad_norm": 2.5206165313720703,862 "learning_rate": 2.438e-05,863 "loss": 0.4072,864 "step": 1220865 },866 {867 "epoch": 76.875,868 "grad_norm": 4.3755083084106445,869 "learning_rate": 2.4580000000000002e-05,870 "loss": 0.4025,871 "step": 1230872 },873 {874 "epoch": 77.5,875 "grad_norm": 3.973033905029297,876 "learning_rate": 2.478e-05,877 "loss": 0.3953,878 "step": 1240879 },880 {881 "epoch": 78.125,882 "grad_norm": 3.5819246768951416,883 "learning_rate": 2.498e-05,884 "loss": 0.3909,885 "step": 1250886 },887 {888 "epoch": 78.75,889 "grad_norm": 3.183445930480957,890 "learning_rate": 2.5180000000000003e-05,891 "loss": 0.3852,892 "step": 1260893 },894 {895 "epoch": 79.375,896 "grad_norm": 3.4347057342529297,897 "learning_rate": 2.5380000000000004e-05,898 "loss": 0.3785,899 "step": 1270900 },901 {902 "epoch": 80.0,903 "grad_norm": 5.1768388748168945,904 "learning_rate": 2.5580000000000002e-05,905 "loss": 0.3717,906 "step": 1280907 },908 {909 "epoch": 80.625,910 "grad_norm": 3.9438490867614746,911 "learning_rate": 2.5779999999999997e-05,912 "loss": 0.3681,913 "step": 1290914 },915 {916 "epoch": 81.25,917 "grad_norm": 4.442440986633301,918 "learning_rate": 2.598e-05,919 "loss": 0.3571,920 "step": 1300921 },922 {923 "epoch": 81.875,924 "grad_norm": 5.6028900146484375,925 "learning_rate": 2.618e-05,926 "loss": 0.3553,927 "step": 1310928 },929 {930 "epoch": 82.5,931 "grad_norm": 5.537353038787842,932 "learning_rate": 2.6379999999999998e-05,933 "loss": 0.3515,934 "step": 1320935 },936 {937 "epoch": 83.125,938 "grad_norm": 6.435239315032959,939 "learning_rate": 2.658e-05,940 "loss": 0.3486,941 "step": 1330942 },943 {944 "epoch": 83.75,945 "grad_norm": 3.413828134536743,946 "learning_rate": 2.678e-05,947 "loss": 0.3373,948 "step": 1340949 },950 {951 "epoch": 84.375,952 "grad_norm": 3.34212327003479,953 "learning_rate": 2.698e-05,954 "loss": 0.3297,955 "step": 1350956 },957 {958 "epoch": 85.0,959 "grad_norm": 3.8223774433135986,960 "learning_rate": 2.718e-05,961 "loss": 0.322,962 "step": 1360963 },964 {965 "epoch": 85.625,966 "grad_norm": 4.211275577545166,967 "learning_rate": 2.738e-05,968 "loss": 0.314,969 "step": 1370970 },971 {972 "epoch": 86.25,973 "grad_norm": 5.035346984863281,974 "learning_rate": 2.758e-05,975 "loss": 0.3089,976 "step": 1380977 },978 {979 "epoch": 86.875,980 "grad_norm": 5.702706813812256,981 "learning_rate": 2.778e-05,982 "loss": 0.3017,983 "step": 1390984 },985 {986 "epoch": 87.5,987 "grad_norm": 3.9658496379852295,988 "learning_rate": 2.798e-05,989 "loss": 0.3002,990 "step": 1400991 },992 {993 "epoch": 88.125,994 "grad_norm": 3.6658170223236084,995 "learning_rate": 2.818e-05,996 "loss": 0.2877,997 "step": 1410998 },999 {1000 "epoch": 88.75,1001 "grad_norm": 4.639245986938477,1002 "learning_rate": 2.8380000000000003e-05,1003 "loss": 0.2773,1004 "step": 14201005 },1006 {1007 "epoch": 89.375,1008 "grad_norm": 4.7095947265625,1009 "learning_rate": 2.858e-05,1010 "loss": 0.2739,1011 "step": 14301012 },1013 {1014 "epoch": 90.0,1015 "grad_norm": 5.2461700439453125,1016 "learning_rate": 2.8780000000000002e-05,1017 "loss": 0.267,1018 "step": 14401019 },1020 {1021 "epoch": 90.625,1022 "grad_norm": 9.186333656311035,1023 "learning_rate": 2.898e-05,1024 "loss": 0.2638,1025 "step": 14501026 },1027 {1028 "epoch": 91.25,1029 "grad_norm": 5.92840051651001,1030 "learning_rate": 2.9180000000000002e-05,1031 "loss": 0.2639,1032 "step": 14601033 },1034 {1035 "epoch": 91.875,1036 "grad_norm": 4.675635814666748,1037 "learning_rate": 2.9380000000000003e-05,1038 "loss": 0.2485,1039 "step": 14701040 },1041 {1042 "epoch": 92.5,1043 "grad_norm": 4.761613845825195,1044 "learning_rate": 2.958e-05,1045 "loss": 0.2392,1046 "step": 14801047 },1048 {1049 "epoch": 93.125,1050 "grad_norm": 5.566701889038086,1051 "learning_rate": 2.9780000000000003e-05,1052 "loss": 0.2341,1053 "step": 14901054 },1055 {1056 "epoch": 93.75,1057 "grad_norm": 4.253500461578369,1058 "learning_rate": 2.998e-05,1059 "loss": 0.2222,1060 "step": 15001061 },1062 {1063 "epoch": 94.375,1064 "grad_norm": 4.602695465087891,1065 "learning_rate": 3.0180000000000002e-05,1066 "loss": 0.2183,1067 "step": 15101068 },1069 {1070 "epoch": 95.0,1071 "grad_norm": 5.938004493713379,1072 "learning_rate": 3.0380000000000004e-05,1073 "loss": 0.213,1074 "step": 15201075 },1076 {1077 "epoch": 95.625,1078 "grad_norm": 6.689606189727783,1079 "learning_rate": 3.058e-05,1080 "loss": 0.2123,1081 "step": 15301082 },1083 {1084 "epoch": 96.25,1085 "grad_norm": 5.87599515914917,1086 "learning_rate": 3.078e-05,1087 "loss": 0.2011,1088 "step": 15401089 },1090 {1091 "epoch": 96.875,1092 "grad_norm": 5.971210956573486,1093 "learning_rate": 3.0980000000000005e-05,1094 "loss": 0.1966,1095 "step": 15501096 },1097 {1098 "epoch": 97.5,1099 "grad_norm": 5.859025001525879,1100 "learning_rate": 3.118e-05,1101 "loss": 0.1946,1102 "step": 15601103 },1104 {1105 "epoch": 98.125,1106 "grad_norm": 5.950936317443848,1107 "learning_rate": 3.138e-05,1108 "loss": 0.1814,1109 "step": 15701110 },1111 {1112 "epoch": 98.75,1113 "grad_norm": 4.768659591674805,1114 "learning_rate": 3.1580000000000006e-05,1115 "loss": 0.1768,1116 "step": 15801117 },1118 {1119 "epoch": 99.375,1120 "grad_norm": 5.677441596984863,1121 "learning_rate": 3.1780000000000004e-05,1122 "loss": 0.1727,1123 "step": 15901124 },1125 {1126 "epoch": 100.0,1127 "grad_norm": 4.38816499710083,1128 "learning_rate": 3.198e-05,1129 "loss": 0.162,1130 "step": 16001131 },1132 {1133 "epoch": 100.625,1134 "grad_norm": 4.571051120758057,1135 "learning_rate": 3.218e-05,1136 "loss": 0.1606,1137 "step": 16101138 },1139 {1140 "epoch": 101.25,1141 "grad_norm": 6.406070709228516,1142 "learning_rate": 3.238e-05,1143 "loss": 0.1494,1144 "step": 16201145 },1146 {1147 "epoch": 101.875,1148 "grad_norm": 4.923046588897705,1149 "learning_rate": 3.2579999999999996e-05,1150 "loss": 0.1479,1151 "step": 16301152 },1153 {1154 "epoch": 102.5,1155 "grad_norm": 6.735141277313232,1156 "learning_rate": 3.278e-05,1157 "loss": 0.1468,1158 "step": 16401159 },1160 {1161 "epoch": 103.125,1162 "grad_norm": 5.201277732849121,1163 "learning_rate": 3.298e-05,1164 "loss": 0.137,1165 "step": 16501166 },1167 {1168 "epoch": 103.75,1169 "grad_norm": 5.127175331115723,1170 "learning_rate": 3.318e-05,1171 "loss": 0.1377,1172 "step": 16601173 },1174 {1175 "epoch": 104.375,1176 "grad_norm": 5.3002471923828125,1177 "learning_rate": 3.338e-05,1178 "loss": 0.1277,1179 "step": 16701180 },1181 {1182 "epoch": 105.0,1183 "grad_norm": 5.290287017822266,1184 "learning_rate": 3.358e-05,1185 "loss": 0.1224,1186 "step": 16801187 },1188 {1189 "epoch": 105.625,1190 "grad_norm": 6.6339311599731445,1191 "learning_rate": 3.378e-05,1192 "loss": 0.121,1193 "step": 16901194 },1195 {1196 "epoch": 106.25,1197 "grad_norm": 6.3351826667785645,1198 "learning_rate": 3.398e-05,1199 "loss": 0.1202,1200 "step": 1700