gaobin/LocateAnything-3B
015
1{2 "best_global_step": null,3 "best_metric": null,4 "best_model_checkpoint": null,5 "epoch": 0.069,6 "eval_steps": 500,7 "global_step": 5000,8 "is_hyper_param_search": false,9 "is_local_process_zero": true,10 "is_world_process_zero": true,11 "log_history": [12 {13 "epoch": 0.0002,14 "grad_norm": 17.97269058227539,15 "learning_rate": 2e-08,16 "loss": 0.8167,17 "step": 118 },19 {20 "epoch": 0.0004,21 "grad_norm": 17.09549331665039,22 "learning_rate": 4e-08,23 "loss": 0.8546,24 "step": 225 },26 {27 "epoch": 0.0006,28 "grad_norm": 16.300983428955078,29 "learning_rate": 6.000000000000001e-08,30 "loss": 0.8395,31 "step": 332 },33 {34 "epoch": 0.0008,35 "grad_norm": 17.731111526489258,36 "learning_rate": 8e-08,37 "loss": 0.8239,38 "step": 439 },40 {41 "epoch": 0.001,42 "grad_norm": 17.372011184692383,43 "learning_rate": 1.0000000000000001e-07,44 "loss": 0.8181,45 "step": 546 },47 {48 "epoch": 0.0012,49 "grad_norm": 17.901464462280273,50 "learning_rate": 1.2000000000000002e-07,51 "loss": 0.8178,52 "step": 653 },54 {55 "epoch": 0.0014,56 "grad_norm": 18.532596588134766,57 "learning_rate": 1.4e-07,58 "loss": 0.7867,59 "step": 760 },61 {62 "epoch": 0.0016,63 "grad_norm": 18.095897674560547,64 "learning_rate": 1.6e-07,65 "loss": 0.827,66 "step": 867 },68 {69 "epoch": 0.0018,70 "grad_norm": 18.920656204223633,71 "learning_rate": 1.8e-07,72 "loss": 0.7756,73 "step": 974 },75 {76 "epoch": 0.002,77 "grad_norm": 17.61746597290039,78 "learning_rate": 2.0000000000000002e-07,79 "loss": 0.8204,80 "step": 1081 },82 {83 "epoch": 0.0022,84 "grad_norm": 17.334989547729492,85 "learning_rate": 2.2e-07,86 "loss": 0.8228,87 "step": 1188 },89 {90 "epoch": 0.0024,91 "grad_norm": 18.79360580444336,92 "learning_rate": 2.4000000000000003e-07,93 "loss": 0.8088,94 "step": 1295 },96 {97 "epoch": 0.0026,98 "grad_norm": 17.502933502197266,99 "learning_rate": 2.6e-07,100 "loss": 0.8062,101 "step": 13102 },103 {104 "epoch": 0.0028,105 "grad_norm": 16.895002365112305,106 "learning_rate": 2.8e-07,107 "loss": 0.8223,108 "step": 14109 },110 {111 "epoch": 0.003,112 "grad_norm": 17.6730899810791,113 "learning_rate": 3.0000000000000004e-07,114 "loss": 0.8076,115 "step": 15116 },117 {118 "epoch": 0.0032,119 "grad_norm": 17.09043312072754,120 "learning_rate": 3.2e-07,121 "loss": 0.7957,122 "step": 16123 },124 {125 "epoch": 0.0034,126 "grad_norm": 17.19906997680664,127 "learning_rate": 3.4000000000000003e-07,128 "loss": 0.7719,129 "step": 17130 },131 {132 "epoch": 0.0036,133 "grad_norm": 18.258024215698242,134 "learning_rate": 3.6e-07,135 "loss": 0.7677,136 "step": 18137 },138 {139 "epoch": 0.0038,140 "grad_norm": 16.483152389526367,141 "learning_rate": 3.8e-07,142 "loss": 0.8058,143 "step": 19144 },145 {146 "epoch": 0.004,147 "grad_norm": 16.420392990112305,148 "learning_rate": 4.0000000000000003e-07,149 "loss": 0.7845,150 "step": 20151 },152 {153 "epoch": 0.0042,154 "grad_norm": 16.849567413330078,155 "learning_rate": 4.2000000000000006e-07,156 "loss": 0.7266,157 "step": 21158 },159 {160 "epoch": 0.0044,161 "grad_norm": 14.625004768371582,162 "learning_rate": 4.4e-07,163 "loss": 0.7221,164 "step": 22165 },166 {167 "epoch": 0.0046,168 "grad_norm": 13.381985664367676,169 "learning_rate": 4.6000000000000004e-07,170 "loss": 0.7395,171 "step": 23172 },173 {174 "epoch": 0.0048,175 "grad_norm": 14.497766494750977,176 "learning_rate": 4.800000000000001e-07,177 "loss": 0.7101,178 "step": 24179 },180 {181 "epoch": 0.005,182 "grad_norm": 13.240507125854492,183 "learning_rate": 5.000000000000001e-07,184 "loss": 0.7618,185 "step": 25186 },187 {188 "epoch": 0.0052,189 "grad_norm": 12.958215713500977,190 "learning_rate": 5.2e-07,191 "loss": 0.7476,192 "step": 26193 },194 {195 "epoch": 0.0054,196 "grad_norm": 13.203178405761719,197 "learning_rate": 5.4e-07,198 "loss": 0.7303,199 "step": 27200 },201 {202 "epoch": 0.0056,203 "grad_norm": 13.179447174072266,204 "learning_rate": 5.6e-07,205 "loss": 0.6991,206 "step": 28207 },208 {209 "epoch": 0.0058,210 "grad_norm": 12.315122604370117,211 "learning_rate": 5.800000000000001e-07,212 "loss": 0.6904,213 "step": 29214 },215 {216 "epoch": 0.006,217 "grad_norm": 6.784637451171875,218 "learning_rate": 6.000000000000001e-07,219 "loss": 0.6437,220 "step": 30221 },222 {223 "epoch": 0.0062,224 "grad_norm": 6.480414390563965,225 "learning_rate": 6.200000000000001e-07,226 "loss": 0.6424,227 "step": 31228 },229 {230 "epoch": 0.0064,231 "grad_norm": 5.988020896911621,232 "learning_rate": 6.4e-07,233 "loss": 0.6354,234 "step": 32235 },236 {237 "epoch": 0.0066,238 "grad_norm": 6.273484230041504,239 "learning_rate": 6.6e-07,240 "loss": 0.5735,241 "step": 33242 },243 {244 "epoch": 0.0068,245 "grad_norm": 6.120683193206787,246 "learning_rate": 6.800000000000001e-07,247 "loss": 0.6357,248 "step": 34249 },250 {251 "epoch": 0.007,252 "grad_norm": 6.0412397384643555,253 "learning_rate": 7.000000000000001e-07,254 "loss": 0.6151,255 "step": 35256 },257 {258 "epoch": 0.0072,259 "grad_norm": 5.526604175567627,260 "learning_rate": 7.2e-07,261 "loss": 0.5895,262 "step": 36263 },264 {265 "epoch": 0.0074,266 "grad_norm": 5.2901716232299805,267 "learning_rate": 7.4e-07,268 "loss": 0.6205,269 "step": 37270 },271 {272 "epoch": 0.0076,273 "grad_norm": 5.131118297576904,274 "learning_rate": 7.6e-07,275 "loss": 0.6118,276 "step": 38277 },278 {279 "epoch": 0.0078,280 "grad_norm": 4.498150825500488,281 "learning_rate": 7.8e-07,282 "loss": 0.598,283 "step": 39284 },285 {286 "epoch": 0.008,287 "grad_norm": 3.5042121410369873,288 "learning_rate": 8.000000000000001e-07,289 "loss": 0.5298,290 "step": 40291 },292 {293 "epoch": 0.0082,294 "grad_norm": 2.486785650253296,295 "learning_rate": 8.200000000000001e-07,296 "loss": 0.5166,297 "step": 41298 },299 {300 "epoch": 0.0084,301 "grad_norm": 2.0915184020996094,302 "learning_rate": 8.400000000000001e-07,303 "loss": 0.4809,304 "step": 42305 },306 {307 "epoch": 0.0086,308 "grad_norm": 2.011510133743286,309 "learning_rate": 8.6e-07,310 "loss": 0.5003,311 "step": 43312 },313 {314 "epoch": 0.0088,315 "grad_norm": 1.8339259624481201,316 "learning_rate": 8.8e-07,317 "loss": 0.4743,318 "step": 44319 },320 {321 "epoch": 0.009,322 "grad_norm": 1.8110532760620117,323 "learning_rate": 9.000000000000001e-07,324 "loss": 0.5289,325 "step": 45326 },327 {328 "epoch": 0.0092,329 "grad_norm": 1.7591607570648193,330 "learning_rate": 9.200000000000001e-07,331 "loss": 0.5078,332 "step": 46333 },334 {335 "epoch": 0.0094,336 "grad_norm": 1.5080705881118774,337 "learning_rate": 9.400000000000001e-07,338 "loss": 0.4792,339 "step": 47340 },341 {342 "epoch": 0.0096,343 "grad_norm": 1.5115655660629272,344 "learning_rate": 9.600000000000001e-07,345 "loss": 0.4529,346 "step": 48347 },348 {349 "epoch": 0.0098,350 "grad_norm": 1.489723563194275,351 "learning_rate": 9.800000000000001e-07,352 "loss": 0.5206,353 "step": 49354 },355 {356 "epoch": 0.01,357 "grad_norm": 1.4680073261260986,358 "learning_rate": 1.0000000000000002e-06,359 "loss": 0.5072,360 "step": 50361 },362 {363 "epoch": 0.0102,364 "grad_norm": 1.371167778968811,365 "learning_rate": 1.02e-06,366 "loss": 0.4273,367 "step": 51368 },369 {370 "epoch": 0.0104,371 "grad_norm": 2.7187724113464355,372 "learning_rate": 1.04e-06,373 "loss": 0.5109,374 "step": 52375 },376 {377 "epoch": 0.0106,378 "grad_norm": 0.9156416058540344,379 "learning_rate": 1.06e-06,380 "loss": 0.4725,381 "step": 53382 },383 {384 "epoch": 0.0108,385 "grad_norm": 0.8322800993919373,386 "learning_rate": 1.08e-06,387 "loss": 0.4608,388 "step": 54389 },390 {391 "epoch": 0.011,392 "grad_norm": 0.652372419834137,393 "learning_rate": 1.1e-06,394 "loss": 0.4015,395 "step": 55396 },397 {398 "epoch": 0.0112,399 "grad_norm": 0.6940380334854126,400 "learning_rate": 1.12e-06,401 "loss": 0.431,402 "step": 56403 },404 {405 "epoch": 0.0114,406 "grad_norm": 0.6117832064628601,407 "learning_rate": 1.14e-06,408 "loss": 0.4102,409 "step": 57410 },411 {412 "epoch": 0.0116,413 "grad_norm": 0.6187627911567688,414 "learning_rate": 1.1600000000000001e-06,415 "loss": 0.4572,416 "step": 58417 },418 {419 "epoch": 0.0118,420 "grad_norm": 0.6884448528289795,421 "learning_rate": 1.1800000000000001e-06,422 "loss": 0.4329,423 "step": 59424 },425 {426 "epoch": 0.012,427 "grad_norm": 0.6142067909240723,428 "learning_rate": 1.2000000000000002e-06,429 "loss": 0.4553,430 "step": 60431 },432 {433 "epoch": 0.0122,434 "grad_norm": 0.611631453037262,435 "learning_rate": 1.2200000000000002e-06,436 "loss": 0.4168,437 "step": 61438 },439 {440 "epoch": 0.0124,441 "grad_norm": 0.8171571493148804,442 "learning_rate": 1.2400000000000002e-06,443 "loss": 0.424,444 "step": 62445 },446 {447 "epoch": 0.0126,448 "grad_norm": 0.5115475654602051,449 "learning_rate": 1.26e-06,450 "loss": 0.3565,451 "step": 63452 },453 {454 "epoch": 0.0128,455 "grad_norm": 0.6233605742454529,456 "learning_rate": 1.28e-06,457 "loss": 0.4151,458 "step": 64459 },460 {461 "epoch": 0.013,462 "grad_norm": 0.47658413648605347,463 "learning_rate": 1.3e-06,464 "loss": 0.3838,465 "step": 65466 },467 {468 "epoch": 0.0132,469 "grad_norm": 0.6022032499313354,470 "learning_rate": 1.32e-06,471 "loss": 0.4034,472 "step": 66473 },474 {475 "epoch": 0.0134,476 "grad_norm": 0.5471862554550171,477 "learning_rate": 1.34e-06,478 "loss": 0.4004,479 "step": 67480 },481 {482 "epoch": 0.0136,483 "grad_norm": 0.4840533435344696,484 "learning_rate": 1.3600000000000001e-06,485 "loss": 0.3914,486 "step": 68487 },488 {489 "epoch": 0.0138,490 "grad_norm": 0.7452861070632935,491 "learning_rate": 1.3800000000000001e-06,492 "loss": 0.4132,493 "step": 69494 },495 {496 "epoch": 0.014,497 "grad_norm": 0.504267692565918,498 "learning_rate": 1.4000000000000001e-06,499 "loss": 0.3992,500 "step": 70501 },502 {503 "epoch": 0.0142,504 "grad_norm": 0.659186840057373,505 "learning_rate": 1.42e-06,506 "loss": 0.4238,507 "step": 71508 },509 {510 "epoch": 0.0144,511 "grad_norm": 0.47695696353912354,512 "learning_rate": 1.44e-06,513 "loss": 0.4181,514 "step": 72515 },516 {517 "epoch": 0.0146,518 "grad_norm": 0.590043306350708,519 "learning_rate": 1.46e-06,520 "loss": 0.4276,521 "step": 73522 },523 {524 "epoch": 0.0148,525 "grad_norm": 0.4621325433254242,526 "learning_rate": 1.48e-06,527 "loss": 0.4124,528 "step": 74529 },530 {531 "epoch": 0.015,532 "grad_norm": 0.5502806305885315,533 "learning_rate": 1.5e-06,534 "loss": 0.4216,535 "step": 75536 },537 {538 "epoch": 0.0152,539 "grad_norm": 0.5027847290039062,540 "learning_rate": 1.52e-06,541 "loss": 0.4263,542 "step": 76543 },544 {545 "epoch": 0.0154,546 "grad_norm": 0.6224135756492615,547 "learning_rate": 1.54e-06,548 "loss": 0.3762,549 "step": 77550 },551 {552 "epoch": 0.0156,553 "grad_norm": 0.5272960066795349,554 "learning_rate": 1.56e-06,555 "loss": 0.3933,556 "step": 78557 },558 {559 "epoch": 0.0158,560 "grad_norm": 0.45190051198005676,561 "learning_rate": 1.5800000000000001e-06,562 "loss": 0.3692,563 "step": 79564 },565 {566 "epoch": 0.016,567 "grad_norm": 0.534759521484375,568 "learning_rate": 1.6000000000000001e-06,569 "loss": 0.3938,570 "step": 80571 },572 {573 "epoch": 0.0162,574 "grad_norm": 0.5399959683418274,575 "learning_rate": 1.6200000000000002e-06,576 "loss": 0.3846,577 "step": 81578 },579 {580 "epoch": 0.0164,581 "grad_norm": 0.4664633274078369,582 "learning_rate": 1.6400000000000002e-06,583 "loss": 0.4163,584 "step": 82585 },586 {587 "epoch": 0.0166,588 "grad_norm": 1.0736939907073975,589 "learning_rate": 1.6600000000000002e-06,590 "loss": 0.4128,591 "step": 83592 },593 {594 "epoch": 0.0168,595 "grad_norm": 0.5936366319656372,596 "learning_rate": 1.6800000000000002e-06,597 "loss": 0.3433,598 "step": 84599 },600 {601 "epoch": 0.017,602 "grad_norm": 0.4042677879333496,603 "learning_rate": 1.7000000000000002e-06,604 "loss": 0.3734,605 "step": 85606 },607 {608 "epoch": 0.0172,609 "grad_norm": 0.43845823407173157,610 "learning_rate": 1.72e-06,611 "loss": 0.3925,612 "step": 86613 },614 {615 "epoch": 0.0174,616 "grad_norm": 0.4913782775402069,617 "learning_rate": 1.74e-06,618 "loss": 0.3947,619 "step": 87620 },621 {622 "epoch": 0.0176,623 "grad_norm": 0.47318845987319946,624 "learning_rate": 1.76e-06,625 "loss": 0.4016,626 "step": 88627 },628 {629 "epoch": 0.0178,630 "grad_norm": 0.5881785154342651,631 "learning_rate": 1.7800000000000001e-06,632 "loss": 0.3652,633 "step": 89634 },635 {636 "epoch": 0.018,637 "grad_norm": 0.551909863948822,638 "learning_rate": 1.8000000000000001e-06,639 "loss": 0.3789,640 "step": 90641 },642 {643 "epoch": 0.0182,644 "grad_norm": 0.48561134934425354,645 "learning_rate": 1.8200000000000002e-06,646 "loss": 0.3613,647 "step": 91648 },649 {650 "epoch": 0.0184,651 "grad_norm": 0.48683464527130127,652 "learning_rate": 1.8400000000000002e-06,653 "loss": 0.3483,654 "step": 92655 },656 {657 "epoch": 0.0186,658 "grad_norm": 0.37006062269210815,659 "learning_rate": 1.8600000000000002e-06,660 "loss": 0.3793,661 "step": 93662 },663 {664 "epoch": 0.0188,665 "grad_norm": 0.44092586636543274,666 "learning_rate": 1.8800000000000002e-06,667 "loss": 0.4029,668 "step": 94669 },670 {671 "epoch": 0.019,672 "grad_norm": 0.6408613920211792,673 "learning_rate": 1.9000000000000002e-06,674 "loss": 0.4543,675 "step": 95676 },677 {678 "epoch": 0.0192,679 "grad_norm": 0.5106256008148193,680 "learning_rate": 1.9200000000000003e-06,681 "loss": 0.3707,682 "step": 96683 },684 {685 "epoch": 0.0194,686 "grad_norm": 0.5857816338539124,687 "learning_rate": 1.94e-06,688 "loss": 0.4017,689 "step": 97690 },691 {692 "epoch": 0.0196,693 "grad_norm": 0.5307949185371399,694 "learning_rate": 1.9600000000000003e-06,695 "loss": 0.3993,696 "step": 98697 },698 {699 "epoch": 0.0198,700 "grad_norm": 0.5702608227729797,701 "learning_rate": 1.98e-06,702 "loss": 0.4083,703 "step": 99704 },705 {706 "epoch": 0.02,707 "grad_norm": 0.566071629524231,708 "learning_rate": 2.0000000000000003e-06,709 "loss": 0.3895,710 "step": 100711 },712 {713 "epoch": 0.0202,714 "grad_norm": 0.43044522404670715,715 "learning_rate": 2.02e-06,716 "loss": 0.383,717 "step": 101718 },719 {720 "epoch": 0.0204,721 "grad_norm": 0.37991130352020264,722 "learning_rate": 2.04e-06,723 "loss": 0.3644,724 "step": 102725 },726 {727 "epoch": 0.0206,728 "grad_norm": 0.45190778374671936,729 "learning_rate": 2.06e-06,730 "loss": 0.3699,731 "step": 103732 },733 {734 "epoch": 0.0208,735 "grad_norm": 0.8487486243247986,736 "learning_rate": 2.08e-06,737 "loss": 0.3237,738 "step": 104739 },740 {741 "epoch": 0.021,742 "grad_norm": 0.5039421319961548,743 "learning_rate": 2.1000000000000002e-06,744 "loss": 0.3652,745 "step": 105746 },747 {748 "epoch": 0.0212,749 "grad_norm": 0.5598636865615845,750 "learning_rate": 2.12e-06,751 "loss": 0.379,752 "step": 106753 },754 {755 "epoch": 0.0214,756 "grad_norm": 0.40302714705467224,757 "learning_rate": 2.1400000000000003e-06,758 "loss": 0.355,759 "step": 107760 },761 {762 "epoch": 0.0216,763 "grad_norm": 0.454899400472641,764 "learning_rate": 2.16e-06,765 "loss": 0.3977,766 "step": 108767 },768 {769 "epoch": 0.0218,770 "grad_norm": 0.5021453499794006,771 "learning_rate": 2.1800000000000003e-06,772 "loss": 0.322,773 "step": 109774 },775 {776 "epoch": 0.022,777 "grad_norm": 0.8794844746589661,778 "learning_rate": 2.2e-06,779 "loss": 0.4249,780 "step": 110781 },782 {783 "epoch": 0.0222,784 "grad_norm": 0.4198489785194397,785 "learning_rate": 2.2200000000000003e-06,786 "loss": 0.3669,787 "step": 111788 },789 {790 "epoch": 0.0224,791 "grad_norm": 0.4584365785121918,792 "learning_rate": 2.24e-06,793 "loss": 0.3622,794 "step": 112795 },796 {797 "epoch": 0.0226,798 "grad_norm": 0.4323468804359436,799 "learning_rate": 2.2600000000000004e-06,800 "loss": 0.3745,801 "step": 113802 },803 {804 "epoch": 0.0228,805 "grad_norm": 0.7490417957305908,806 "learning_rate": 2.28e-06,807 "loss": 0.3832,808 "step": 114809 },810 {811 "epoch": 0.023,812 "grad_norm": 0.4236067235469818,813 "learning_rate": 2.3000000000000004e-06,814 "loss": 0.371,815 "step": 115816 },817 {818 "epoch": 0.0232,819 "grad_norm": 0.48021242022514343,820 "learning_rate": 2.3200000000000002e-06,821 "loss": 0.4046,822 "step": 116823 },824 {825 "epoch": 0.0234,826 "grad_norm": 0.6185507774353027,827 "learning_rate": 2.3400000000000005e-06,828 "loss": 0.3886,829 "step": 117830 },831 {832 "epoch": 0.0236,833 "grad_norm": 0.5042728185653687,834 "learning_rate": 2.3600000000000003e-06,835 "loss": 0.3564,836 "step": 118837 },838 {839 "epoch": 0.0238,840 "grad_norm": 0.4533032178878784,841 "learning_rate": 2.38e-06,842 "loss": 0.3459,843 "step": 119844 },845 {846 "epoch": 0.024,847 "grad_norm": 0.4727221429347992,848 "learning_rate": 2.4000000000000003e-06,849 "loss": 0.4146,850 "step": 120851 },852 {853 "epoch": 0.0242,854 "grad_norm": 0.4660988748073578,855 "learning_rate": 2.42e-06,856 "loss": 0.3662,857 "step": 121858 },859 {860 "epoch": 0.0244,861 "grad_norm": 0.6129406094551086,862 "learning_rate": 2.4400000000000004e-06,863 "loss": 0.4127,864 "step": 122865 },866 {867 "epoch": 0.0246,868 "grad_norm": 0.7617100477218628,869 "learning_rate": 2.46e-06,870 "loss": 0.3333,871 "step": 123872 },873 {874 "epoch": 0.0248,875 "grad_norm": 0.4653623104095459,876 "learning_rate": 2.4800000000000004e-06,877 "loss": 0.3957,878 "step": 124879 },880 {881 "epoch": 0.025,882 "grad_norm": 0.5935105085372925,883 "learning_rate": 2.5e-06,884 "loss": 0.3693,885 "step": 125886 },887 {888 "epoch": 0.0252,889 "grad_norm": 0.4042361080646515,890 "learning_rate": 2.52e-06,891 "loss": 0.3345,892 "step": 126893 },894 {895 "epoch": 0.0254,896 "grad_norm": 0.5434343814849854,897 "learning_rate": 2.5400000000000002e-06,898 "loss": 0.3548,899 "step": 127900 },901 {902 "epoch": 0.0256,903 "grad_norm": 0.4848114252090454,904 "learning_rate": 2.56e-06,905 "loss": 0.3683,906 "step": 128907 },908 {909 "epoch": 0.0258,910 "grad_norm": 0.4947889447212219,911 "learning_rate": 2.5800000000000003e-06,912 "loss": 0.3764,913 "step": 129914 },915 {916 "epoch": 0.026,917 "grad_norm": 0.5272525548934937,918 "learning_rate": 2.6e-06,919 "loss": 0.3847,920 "step": 130921 },922 {923 "epoch": 0.0262,924 "grad_norm": 0.4667666554450989,925 "learning_rate": 2.6200000000000003e-06,926 "loss": 0.3662,927 "step": 131928 },929 {930 "epoch": 0.0264,931 "grad_norm": 0.48736771941185,932 "learning_rate": 2.64e-06,933 "loss": 0.3937,934 "step": 132935 },936 {937 "epoch": 0.0266,938 "grad_norm": 0.4935629367828369,939 "learning_rate": 2.6600000000000004e-06,940 "loss": 0.393,941 "step": 133942 },943 {944 "epoch": 0.0268,945 "grad_norm": 0.42808762192726135,946 "learning_rate": 2.68e-06,947 "loss": 0.3534,948 "step": 134949 },950 {951 "epoch": 0.027,952 "grad_norm": 0.5305779576301575,953 "learning_rate": 2.7000000000000004e-06,954 "loss": 0.3735,955 "step": 135956 },957 {958 "epoch": 0.0272,959 "grad_norm": 0.5294774174690247,960 "learning_rate": 2.7200000000000002e-06,961 "loss": 0.3591,962 "step": 136963 },964 {965 "epoch": 0.0274,966 "grad_norm": 0.4570407569408417,967 "learning_rate": 2.7400000000000004e-06,968 "loss": 0.3968,969 "step": 137970 },971 {972 "epoch": 0.0276,973 "grad_norm": 0.45037782192230225,974 "learning_rate": 2.7600000000000003e-06,975 "loss": 0.3442,976 "step": 138977 },978 {979 "epoch": 0.0278,980 "grad_norm": 0.4974113702774048,981 "learning_rate": 2.7800000000000005e-06,982 "loss": 0.3745,983 "step": 139984 },985 {986 "epoch": 0.028,987 "grad_norm": 0.6906246542930603,988 "learning_rate": 2.8000000000000003e-06,989 "loss": 0.362,990 "step": 140991 },992 {993 "epoch": 0.0282,994 "grad_norm": 0.783227801322937,995 "learning_rate": 2.82e-06,996 "loss": 0.3392,997 "step": 141998 },999 {1000 "epoch": 0.0284,1001 "grad_norm": 0.48793527483940125,1002 "learning_rate": 2.84e-06,1003 "loss": 0.3756,1004 "step": 1421005 },1006 {1007 "epoch": 0.0286,1008 "grad_norm": 0.4634222388267517,1009 "learning_rate": 2.86e-06,1010 "loss": 0.4002,1011 "step": 1431012 },1013 {1014 "epoch": 0.0288,1015 "grad_norm": 0.45307955145835876,1016 "learning_rate": 2.88e-06,1017 "loss": 0.3839,1018 "step": 1441019 },1020 {1021 "epoch": 0.029,1022 "grad_norm": 0.6718571782112122,1023 "learning_rate": 2.9e-06,1024 "loss": 0.3666,1025 "step": 1451026 },1027 {1028 "epoch": 0.0292,1029 "grad_norm": 0.5100830793380737,1030 "learning_rate": 2.92e-06,1031 "loss": 0.3933,1032 "step": 1461033 },1034 {1035 "epoch": 0.0294,1036 "grad_norm": 0.39325565099716187,1037 "learning_rate": 2.9400000000000002e-06,1038 "loss": 0.3643,1039 "step": 1471040 },1041 {1042 "epoch": 0.0296,1043 "grad_norm": 0.5235859155654907,1044 "learning_rate": 2.96e-06,1045 "loss": 0.3282,1046 "step": 1481047 },1048 {1049 "epoch": 0.0298,1050 "grad_norm": 0.5658304691314697,1051 "learning_rate": 2.9800000000000003e-06,1052 "loss": 0.3582,1053 "step": 1491054 },1055 {1056 "epoch": 0.03,1057 "grad_norm": 0.5299038887023926,1058 "learning_rate": 3e-06,1059 "loss": 0.3442,1060 "step": 1501061 },1062 {1063 "epoch": 0.0302,1064 "grad_norm": 0.4664023816585541,1065 "learning_rate": 3.0200000000000003e-06,1066 "loss": 0.3936,1067 "step": 1511068 },1069 {1070 "epoch": 0.0304,1071 "grad_norm": 0.48280373215675354,1072 "learning_rate": 3.04e-06,1073 "loss": 0.3713,1074 "step": 1521075 },1076 {1077 "epoch": 0.0306,1078 "grad_norm": 0.5745037198066711,1079 "learning_rate": 3.0600000000000003e-06,1080 "loss": 0.3647,1081 "step": 1531082 },1083 {1084 "epoch": 0.0308,1085 "grad_norm": 0.5101187825202942,1086 "learning_rate": 3.08e-06,1087 "loss": 0.3605,1088 "step": 1541089 },1090 {1091 "epoch": 0.031,1092 "grad_norm": 0.5530501008033752,1093 "learning_rate": 3.1000000000000004e-06,1094 "loss": 0.3568,1095 "step": 1551096 },1097 {1098 "epoch": 0.0312,1099 "grad_norm": 0.5654841661453247,1100 "learning_rate": 3.12e-06,1101 "loss": 0.3743,1102 "step": 1561103 },1104 {1105 "epoch": 0.0314,1106 "grad_norm": 0.4923531115055084,1107 "learning_rate": 3.1400000000000004e-06,1108 "loss": 0.3801,1109 "step": 1571110 },1111 {1112 "epoch": 0.0316,1113 "grad_norm": 0.5236830711364746,1114 "learning_rate": 3.1600000000000002e-06,1115 "loss": 0.4096,1116 "step": 1581117 },1118 {1119 "epoch": 0.0318,1120 "grad_norm": 0.40548253059387207,1121 "learning_rate": 3.1800000000000005e-06,1122 "loss": 0.3198,1123 "step": 1591124 },1125 {1126 "epoch": 0.032,1127 "grad_norm": 0.49229100346565247,1128 "learning_rate": 3.2000000000000003e-06,1129 "loss": 0.3812,1130 "step": 1601131 },1132 {1133 "epoch": 0.0322,1134 "grad_norm": 0.4936654269695282,1135 "learning_rate": 3.2200000000000005e-06,1136 "loss": 0.3878,1137 "step": 1611138 },1139 {1140 "epoch": 0.0324,1141 "grad_norm": 0.43520358204841614,1142 "learning_rate": 3.2400000000000003e-06,1143 "loss": 0.3647,1144 "step": 1621145 },1146 {1147 "epoch": 0.0326,1148 "grad_norm": 0.5249994397163391,1149 "learning_rate": 3.2600000000000006e-06,1150 "loss": 0.3945,1151 "step": 1631152 },1153 {1154 "epoch": 0.0328,1155 "grad_norm": 0.4600265920162201,1156 "learning_rate": 3.2800000000000004e-06,1157 "loss": 0.4077,1158 "step": 1641159 },1160 {1161 "epoch": 0.033,1162 "grad_norm": 0.5561789274215698,1163 "learning_rate": 3.3000000000000006e-06,1164 "loss": 0.3871,1165 "step": 1651166 },1167 {1168 "epoch": 0.0332,1169 "grad_norm": 0.4665578007698059,1170 "learning_rate": 3.3200000000000004e-06,1171 "loss": 0.3888,1172 "step": 1661173 },1174 {1175 "epoch": 0.0334,1176 "grad_norm": 0.43751949071884155,1177 "learning_rate": 3.3400000000000006e-06,1178 "loss": 0.3595,1179 "step": 1671180 },1181 {1182 "epoch": 0.0336,1183 "grad_norm": 0.573382556438446,1184 "learning_rate": 3.3600000000000004e-06,1185 "loss": 0.3931,1186 "step": 1681187 },1188 {1189 "epoch": 0.0338,1190 "grad_norm": 0.517301082611084,1191 "learning_rate": 3.3800000000000007e-06,1192 "loss": 0.3449,1193 "step": 1691194 },1195 {1196 "epoch": 0.034,1197 "grad_norm": 0.520264208316803,1198 "learning_rate": 3.4000000000000005e-06,1199 "loss": 0.3787,1200 "step": 170