TIGER-Lab/VLM2Vec-LoRA
11554
1{2 "best_metric": null,3 "best_model_checkpoint": null,4 "epoch": 3.091190108191654,5 "eval_steps": 500,6 "global_step": 2000,7 "is_hyper_param_search": false,8 "is_local_process_zero": true,9 "is_world_process_zero": true,10 "log_history": [11 {12 "epoch": 0.0015455950540958269,13 "grad_norm": 472.0,14 "learning_rate": 2.0000000000000002e-07,15 "loss": 26.6042,16 "step": 117 },18 {19 "epoch": 0.0030911901081916537,20 "grad_norm": 490.0,21 "learning_rate": 4.0000000000000003e-07,22 "loss": 27.2068,23 "step": 224 },25 {26 "epoch": 0.00463678516228748,27 "grad_norm": 480.0,28 "learning_rate": 6.000000000000001e-07,29 "loss": 26.9311,30 "step": 331 },32 {33 "epoch": 0.0061823802163833074,34 "grad_norm": 472.0,35 "learning_rate": 8.000000000000001e-07,36 "loss": 26.4758,37 "step": 438 },39 {40 "epoch": 0.0077279752704791345,41 "grad_norm": 504.0,42 "learning_rate": 1.0000000000000002e-06,43 "loss": 27.7404,44 "step": 545 },46 {47 "epoch": 0.00927357032457496,48 "grad_norm": 482.0,49 "learning_rate": 1.2000000000000002e-06,50 "loss": 26.5912,51 "step": 652 },53 {54 "epoch": 0.010819165378670788,55 "grad_norm": 484.0,56 "learning_rate": 1.4000000000000001e-06,57 "loss": 27.0497,58 "step": 759 },60 {61 "epoch": 0.012364760432766615,62 "grad_norm": 482.0,63 "learning_rate": 1.6000000000000001e-06,64 "loss": 26.375,65 "step": 866 },67 {68 "epoch": 0.013910355486862442,69 "grad_norm": 470.0,70 "learning_rate": 1.8000000000000001e-06,71 "loss": 24.9608,72 "step": 973 },74 {75 "epoch": 0.015455950540958269,76 "grad_norm": 484.0,77 "learning_rate": 2.0000000000000003e-06,78 "loss": 24.9394,79 "step": 1080 },81 {82 "epoch": 0.017001545595054096,83 "grad_norm": 464.0,84 "learning_rate": 2.2e-06,85 "loss": 23.9378,86 "step": 1187 },88 {89 "epoch": 0.01854714064914992,90 "grad_norm": 480.0,91 "learning_rate": 2.4000000000000003e-06,92 "loss": 23.005,93 "step": 1294 },95 {96 "epoch": 0.02009273570324575,97 "grad_norm": 448.0,98 "learning_rate": 2.6e-06,99 "loss": 21.464,100 "step": 13101 },102 {103 "epoch": 0.021638330757341576,104 "grad_norm": 430.0,105 "learning_rate": 2.8000000000000003e-06,106 "loss": 20.2247,107 "step": 14108 },109 {110 "epoch": 0.023183925811437404,111 "grad_norm": 416.0,112 "learning_rate": 3e-06,113 "loss": 18.9013,114 "step": 15115 },116 {117 "epoch": 0.02472952086553323,118 "grad_norm": 426.0,119 "learning_rate": 3.2000000000000003e-06,120 "loss": 17.6419,121 "step": 16122 },123 {124 "epoch": 0.02627511591962906,125 "grad_norm": 428.0,126 "learning_rate": 3.4000000000000005e-06,127 "loss": 16.2712,128 "step": 17129 },130 {131 "epoch": 0.027820710973724884,132 "grad_norm": 426.0,133 "learning_rate": 3.6000000000000003e-06,134 "loss": 14.3173,135 "step": 18136 },137 {138 "epoch": 0.02936630602782071,139 "grad_norm": 416.0,140 "learning_rate": 3.8000000000000005e-06,141 "loss": 13.1013,142 "step": 19143 },144 {145 "epoch": 0.030911901081916538,146 "grad_norm": 366.0,147 "learning_rate": 4.000000000000001e-06,148 "loss": 11.5202,149 "step": 20150 },151 {152 "epoch": 0.03245749613601236,153 "grad_norm": 322.0,154 "learning_rate": 4.2000000000000004e-06,155 "loss": 10.0744,156 "step": 21157 },158 {159 "epoch": 0.03400309119010819,160 "grad_norm": 282.0,161 "learning_rate": 4.4e-06,162 "loss": 9.0714,163 "step": 22164 },165 {166 "epoch": 0.03554868624420402,167 "grad_norm": 207.0,168 "learning_rate": 4.600000000000001e-06,169 "loss": 8.0312,170 "step": 23171 },172 {173 "epoch": 0.03709428129829984,174 "grad_norm": 178.0,175 "learning_rate": 4.800000000000001e-06,176 "loss": 7.7583,177 "step": 24178 },179 {180 "epoch": 0.03863987635239567,181 "grad_norm": 190.0,182 "learning_rate": 5e-06,183 "loss": 7.3539,184 "step": 25185 },186 {187 "epoch": 0.0401854714064915,188 "grad_norm": 136.0,189 "learning_rate": 5.2e-06,190 "loss": 6.9746,191 "step": 26192 },193 {194 "epoch": 0.04173106646058733,195 "grad_norm": 121.5,196 "learning_rate": 5.400000000000001e-06,197 "loss": 6.638,198 "step": 27199 },200 {201 "epoch": 0.04327666151468315,202 "grad_norm": 112.0,203 "learning_rate": 5.600000000000001e-06,204 "loss": 6.4899,205 "step": 28206 },207 {208 "epoch": 0.04482225656877898,209 "grad_norm": 102.5,210 "learning_rate": 5.8e-06,211 "loss": 6.2407,212 "step": 29213 },214 {215 "epoch": 0.04636785162287481,216 "grad_norm": 87.0,217 "learning_rate": 6e-06,218 "loss": 5.9863,219 "step": 30220 },221 {222 "epoch": 0.04791344667697063,223 "grad_norm": 84.0,224 "learning_rate": 6.200000000000001e-06,225 "loss": 5.822,226 "step": 31227 },228 {229 "epoch": 0.04945904173106646,230 "grad_norm": 73.5,231 "learning_rate": 6.4000000000000006e-06,232 "loss": 5.7442,233 "step": 32234 },235 {236 "epoch": 0.05100463678516229,237 "grad_norm": 72.0,238 "learning_rate": 6.600000000000001e-06,239 "loss": 5.4181,240 "step": 33241 },242 {243 "epoch": 0.05255023183925812,244 "grad_norm": 67.5,245 "learning_rate": 6.800000000000001e-06,246 "loss": 5.2732,247 "step": 34248 },249 {250 "epoch": 0.05409582689335394,251 "grad_norm": 62.5,252 "learning_rate": 7e-06,253 "loss": 5.0897,254 "step": 35255 },256 {257 "epoch": 0.05564142194744977,258 "grad_norm": 55.5,259 "learning_rate": 7.2000000000000005e-06,260 "loss": 4.8823,261 "step": 36262 },263 {264 "epoch": 0.0571870170015456,265 "grad_norm": 62.5,266 "learning_rate": 7.4e-06,267 "loss": 4.7194,268 "step": 37269 },270 {271 "epoch": 0.05873261205564142,272 "grad_norm": 59.0,273 "learning_rate": 7.600000000000001e-06,274 "loss": 4.5549,275 "step": 38276 },277 {278 "epoch": 0.06027820710973725,279 "grad_norm": 70.5,280 "learning_rate": 7.800000000000002e-06,281 "loss": 4.3491,282 "step": 39283 },284 {285 "epoch": 0.061823802163833076,286 "grad_norm": 54.25,287 "learning_rate": 8.000000000000001e-06,288 "loss": 4.1664,289 "step": 40290 },291 {292 "epoch": 0.0633693972179289,293 "grad_norm": 53.5,294 "learning_rate": 8.2e-06,295 "loss": 3.9786,296 "step": 41297 },298 {299 "epoch": 0.06491499227202473,300 "grad_norm": 52.25,301 "learning_rate": 8.400000000000001e-06,302 "loss": 3.718,303 "step": 42304 },305 {306 "epoch": 0.06646058732612056,307 "grad_norm": 60.0,308 "learning_rate": 8.6e-06,309 "loss": 3.57,310 "step": 43311 },312 {313 "epoch": 0.06800618238021638,314 "grad_norm": 44.5,315 "learning_rate": 8.8e-06,316 "loss": 3.4842,317 "step": 44318 },319 {320 "epoch": 0.0695517774343122,321 "grad_norm": 39.0,322 "learning_rate": 9e-06,323 "loss": 3.0784,324 "step": 45325 },326 {327 "epoch": 0.07109737248840804,328 "grad_norm": 86.5,329 "learning_rate": 9.200000000000002e-06,330 "loss": 3.1007,331 "step": 46332 },333 {334 "epoch": 0.07264296754250386,335 "grad_norm": 74.5,336 "learning_rate": 9.4e-06,337 "loss": 2.9358,338 "step": 47339 },340 {341 "epoch": 0.07418856259659969,342 "grad_norm": 39.0,343 "learning_rate": 9.600000000000001e-06,344 "loss": 2.932,345 "step": 48346 },347 {348 "epoch": 0.07573415765069552,349 "grad_norm": 35.75,350 "learning_rate": 9.800000000000001e-06,351 "loss": 2.7691,352 "step": 49353 },354 {355 "epoch": 0.07727975270479134,356 "grad_norm": 32.5,357 "learning_rate": 1e-05,358 "loss": 2.6701,359 "step": 50360 },361 {362 "epoch": 0.07882534775888717,363 "grad_norm": 28.875,364 "learning_rate": 1.02e-05,365 "loss": 2.4431,366 "step": 51367 },368 {369 "epoch": 0.080370942812983,370 "grad_norm": 30.375,371 "learning_rate": 1.04e-05,372 "loss": 2.5914,373 "step": 52374 },375 {376 "epoch": 0.08191653786707882,377 "grad_norm": 28.875,378 "learning_rate": 1.0600000000000002e-05,379 "loss": 2.4222,380 "step": 53381 },382 {383 "epoch": 0.08346213292117466,384 "grad_norm": 27.375,385 "learning_rate": 1.0800000000000002e-05,386 "loss": 2.1741,387 "step": 54388 },389 {390 "epoch": 0.08500772797527048,391 "grad_norm": 55.75,392 "learning_rate": 1.1000000000000001e-05,393 "loss": 2.1799,394 "step": 55395 },396 {397 "epoch": 0.0865533230293663,398 "grad_norm": 22.75,399 "learning_rate": 1.1200000000000001e-05,400 "loss": 2.0037,401 "step": 56402 },403 {404 "epoch": 0.08809891808346214,405 "grad_norm": 23.0,406 "learning_rate": 1.14e-05,407 "loss": 2.0929,408 "step": 57409 },410 {411 "epoch": 0.08964451313755796,412 "grad_norm": 22.75,413 "learning_rate": 1.16e-05,414 "loss": 1.9233,415 "step": 58416 },417 {418 "epoch": 0.09119010819165378,419 "grad_norm": 22.625,420 "learning_rate": 1.18e-05,421 "loss": 1.8583,422 "step": 59423 },424 {425 "epoch": 0.09273570324574962,426 "grad_norm": 27.0,427 "learning_rate": 1.2e-05,428 "loss": 1.9135,429 "step": 60430 },431 {432 "epoch": 0.09428129829984544,433 "grad_norm": 20.875,434 "learning_rate": 1.22e-05,435 "loss": 1.7249,436 "step": 61437 },438 {439 "epoch": 0.09582689335394126,440 "grad_norm": 20.625,441 "learning_rate": 1.2400000000000002e-05,442 "loss": 1.6811,443 "step": 62444 },445 {446 "epoch": 0.0973724884080371,447 "grad_norm": 26.5,448 "learning_rate": 1.2600000000000001e-05,449 "loss": 1.72,450 "step": 63451 },452 {453 "epoch": 0.09891808346213292,454 "grad_norm": 18.0,455 "learning_rate": 1.2800000000000001e-05,456 "loss": 1.6351,457 "step": 64458 },459 {460 "epoch": 0.10046367851622875,461 "grad_norm": 18.0,462 "learning_rate": 1.3000000000000001e-05,463 "loss": 1.6122,464 "step": 65465 },466 {467 "epoch": 0.10200927357032458,468 "grad_norm": 17.625,469 "learning_rate": 1.3200000000000002e-05,470 "loss": 1.5638,471 "step": 66472 },473 {474 "epoch": 0.1035548686244204,475 "grad_norm": 17.875,476 "learning_rate": 1.3400000000000002e-05,477 "loss": 1.489,478 "step": 67479 },480 {481 "epoch": 0.10510046367851623,482 "grad_norm": 18.75,483 "learning_rate": 1.3600000000000002e-05,484 "loss": 1.5691,485 "step": 68486 },487 {488 "epoch": 0.10664605873261206,489 "grad_norm": 20.5,490 "learning_rate": 1.38e-05,491 "loss": 1.4363,492 "step": 69493 },494 {495 "epoch": 0.10819165378670788,496 "grad_norm": 17.125,497 "learning_rate": 1.4e-05,498 "loss": 1.4473,499 "step": 70500 },501 {502 "epoch": 0.10973724884080371,503 "grad_norm": 17.625,504 "learning_rate": 1.4200000000000001e-05,505 "loss": 1.4345,506 "step": 71507 },508 {509 "epoch": 0.11128284389489954,510 "grad_norm": 27.5,511 "learning_rate": 1.4400000000000001e-05,512 "loss": 1.5177,513 "step": 72514 },515 {516 "epoch": 0.11282843894899536,517 "grad_norm": 16.625,518 "learning_rate": 1.46e-05,519 "loss": 1.3802,520 "step": 73521 },522 {523 "epoch": 0.1143740340030912,524 "grad_norm": 19.625,525 "learning_rate": 1.48e-05,526 "loss": 1.3202,527 "step": 74528 },529 {530 "epoch": 0.11591962905718702,531 "grad_norm": 27.25,532 "learning_rate": 1.5000000000000002e-05,533 "loss": 1.3449,534 "step": 75535 },536 {537 "epoch": 0.11746522411128284,538 "grad_norm": 16.375,539 "learning_rate": 1.5200000000000002e-05,540 "loss": 1.3505,541 "step": 76542 },543 {544 "epoch": 0.11901081916537867,545 "grad_norm": 16.25,546 "learning_rate": 1.54e-05,547 "loss": 1.2425,548 "step": 77549 },550 {551 "epoch": 0.1205564142194745,552 "grad_norm": 15.25,553 "learning_rate": 1.5600000000000003e-05,554 "loss": 1.2216,555 "step": 78556 },557 {558 "epoch": 0.12210200927357033,559 "grad_norm": 15.4375,560 "learning_rate": 1.58e-05,561 "loss": 1.2965,562 "step": 79563 },564 {565 "epoch": 0.12364760432766615,566 "grad_norm": 15.6875,567 "learning_rate": 1.6000000000000003e-05,568 "loss": 1.3441,569 "step": 80570 },571 {572 "epoch": 0.125193199381762,573 "grad_norm": 43.25,574 "learning_rate": 1.62e-05,575 "loss": 1.2836,576 "step": 81577 },578 {579 "epoch": 0.1267387944358578,580 "grad_norm": 18.375,581 "learning_rate": 1.64e-05,582 "loss": 1.2163,583 "step": 82584 },585 {586 "epoch": 0.12828438948995363,587 "grad_norm": 38.25,588 "learning_rate": 1.66e-05,589 "loss": 1.1794,590 "step": 83591 },592 {593 "epoch": 0.12982998454404945,594 "grad_norm": 14.8125,595 "learning_rate": 1.6800000000000002e-05,596 "loss": 1.1239,597 "step": 84598 },599 {600 "epoch": 0.13137557959814528,601 "grad_norm": 13.5625,602 "learning_rate": 1.7e-05,603 "loss": 1.0771,604 "step": 85605 },606 {607 "epoch": 0.13292117465224113,608 "grad_norm": 14.5625,609 "learning_rate": 1.72e-05,610 "loss": 1.0916,611 "step": 86612 },613 {614 "epoch": 0.13446676970633695,615 "grad_norm": 14.5625,616 "learning_rate": 1.7400000000000003e-05,617 "loss": 1.0952,618 "step": 87619 },620 {621 "epoch": 0.13601236476043277,622 "grad_norm": 13.75,623 "learning_rate": 1.76e-05,624 "loss": 1.0659,625 "step": 88626 },627 {628 "epoch": 0.1375579598145286,629 "grad_norm": 13.5625,630 "learning_rate": 1.7800000000000002e-05,631 "loss": 0.9821,632 "step": 89633 },634 {635 "epoch": 0.1391035548686244,636 "grad_norm": 14.75,637 "learning_rate": 1.8e-05,638 "loss": 1.079,639 "step": 90640 },641 {642 "epoch": 0.14064914992272023,643 "grad_norm": 13.5625,644 "learning_rate": 1.8200000000000002e-05,645 "loss": 1.003,646 "step": 91647 },648 {649 "epoch": 0.14219474497681608,650 "grad_norm": 19.875,651 "learning_rate": 1.8400000000000003e-05,652 "loss": 1.1328,653 "step": 92654 },655 {656 "epoch": 0.1437403400309119,657 "grad_norm": 16.875,658 "learning_rate": 1.86e-05,659 "loss": 1.0903,660 "step": 93661 },662 {663 "epoch": 0.14528593508500773,664 "grad_norm": 15.5,665 "learning_rate": 1.88e-05,666 "loss": 1.0609,667 "step": 94668 },669 {670 "epoch": 0.14683153013910355,671 "grad_norm": 13.8125,672 "learning_rate": 1.9e-05,673 "loss": 1.0328,674 "step": 95675 },676 {677 "epoch": 0.14837712519319937,678 "grad_norm": 13.375,679 "learning_rate": 1.9200000000000003e-05,680 "loss": 0.9742,681 "step": 96682 },683 {684 "epoch": 0.14992272024729522,685 "grad_norm": 14.3125,686 "learning_rate": 1.94e-05,687 "loss": 1.0703,688 "step": 97689 },690 {691 "epoch": 0.15146831530139104,692 "grad_norm": 14.3125,693 "learning_rate": 1.9600000000000002e-05,694 "loss": 1.034,695 "step": 98696 },697 {698 "epoch": 0.15301391035548687,699 "grad_norm": 13.3125,700 "learning_rate": 1.98e-05,701 "loss": 0.9953,702 "step": 99703 },704 {705 "epoch": 0.1545595054095827,706 "grad_norm": 13.5,707 "learning_rate": 2e-05,708 "loss": 0.9211,709 "step": 100710 },711 {712 "epoch": 0.1561051004636785,713 "grad_norm": 13.9375,714 "learning_rate": 1.9989473684210526e-05,715 "loss": 1.0385,716 "step": 101717 },718 {719 "epoch": 0.15765069551777433,720 "grad_norm": 13.375,721 "learning_rate": 1.9978947368421054e-05,722 "loss": 0.9443,723 "step": 102724 },725 {726 "epoch": 0.15919629057187018,727 "grad_norm": 23.625,728 "learning_rate": 1.9968421052631582e-05,729 "loss": 0.9293,730 "step": 103731 },732 {733 "epoch": 0.160741885625966,734 "grad_norm": 15.0,735 "learning_rate": 1.9957894736842107e-05,736 "loss": 1.0052,737 "step": 104738 },739 {740 "epoch": 0.16228748068006182,741 "grad_norm": 18.125,742 "learning_rate": 1.994736842105263e-05,743 "loss": 0.9428,744 "step": 105745 },746 {747 "epoch": 0.16383307573415765,748 "grad_norm": 14.1875,749 "learning_rate": 1.993684210526316e-05,750 "loss": 0.9634,751 "step": 106752 },753 {754 "epoch": 0.16537867078825347,755 "grad_norm": 14.875,756 "learning_rate": 1.9926315789473688e-05,757 "loss": 1.0048,758 "step": 107759 },760 {761 "epoch": 0.16692426584234932,762 "grad_norm": 12.875,763 "learning_rate": 1.9915789473684212e-05,764 "loss": 0.9729,765 "step": 108766 },767 {768 "epoch": 0.16846986089644514,769 "grad_norm": 12.6875,770 "learning_rate": 1.990526315789474e-05,771 "loss": 0.9173,772 "step": 109773 },774 {775 "epoch": 0.17001545595054096,776 "grad_norm": 18.0,777 "learning_rate": 1.9894736842105265e-05,778 "loss": 1.0079,779 "step": 110780 },781 {782 "epoch": 0.17156105100463678,783 "grad_norm": 12.9375,784 "learning_rate": 1.988421052631579e-05,785 "loss": 0.9352,786 "step": 111787 },788 {789 "epoch": 0.1731066460587326,790 "grad_norm": 14.125,791 "learning_rate": 1.9873684210526318e-05,792 "loss": 0.9911,793 "step": 112794 },795 {796 "epoch": 0.17465224111282843,797 "grad_norm": 15.25,798 "learning_rate": 1.9863157894736846e-05,799 "loss": 0.9601,800 "step": 113801 },802 {803 "epoch": 0.17619783616692428,804 "grad_norm": 13.0625,805 "learning_rate": 1.985263157894737e-05,806 "loss": 0.8756,807 "step": 114808 },809 {810 "epoch": 0.1777434312210201,811 "grad_norm": 16.75,812 "learning_rate": 1.9842105263157895e-05,813 "loss": 0.9709,814 "step": 115815 },816 {817 "epoch": 0.17928902627511592,818 "grad_norm": 12.9375,819 "learning_rate": 1.9831578947368423e-05,820 "loss": 0.9185,821 "step": 116822 },823 {824 "epoch": 0.18083462132921174,825 "grad_norm": 14.25,826 "learning_rate": 1.982105263157895e-05,827 "loss": 0.98,828 "step": 117829 },830 {831 "epoch": 0.18238021638330756,832 "grad_norm": 13.5,833 "learning_rate": 1.9810526315789476e-05,834 "loss": 1.0079,835 "step": 118836 },837 {838 "epoch": 0.1839258114374034,839 "grad_norm": 12.125,840 "learning_rate": 1.98e-05,841 "loss": 0.9213,842 "step": 119843 },844 {845 "epoch": 0.18547140649149924,846 "grad_norm": 12.125,847 "learning_rate": 1.9789473684210528e-05,848 "loss": 0.9095,849 "step": 120850 },851 {852 "epoch": 0.18701700154559506,853 "grad_norm": 13.75,854 "learning_rate": 1.9778947368421056e-05,855 "loss": 0.8693,856 "step": 121857 },858 {859 "epoch": 0.18856259659969088,860 "grad_norm": 12.8125,861 "learning_rate": 1.976842105263158e-05,862 "loss": 0.897,863 "step": 122864 },865 {866 "epoch": 0.1901081916537867,867 "grad_norm": 11.875,868 "learning_rate": 1.9757894736842105e-05,869 "loss": 0.7896,870 "step": 123871 },872 {873 "epoch": 0.19165378670788252,874 "grad_norm": 12.25,875 "learning_rate": 1.9747368421052633e-05,876 "loss": 0.8458,877 "step": 124878 },879 {880 "epoch": 0.19319938176197837,881 "grad_norm": 11.0625,882 "learning_rate": 1.9736842105263158e-05,883 "loss": 0.8362,884 "step": 125885 },886 {887 "epoch": 0.1947449768160742,888 "grad_norm": 11.0,889 "learning_rate": 1.9726315789473686e-05,890 "loss": 0.8465,891 "step": 126892 },893 {894 "epoch": 0.19629057187017002,895 "grad_norm": 11.0625,896 "learning_rate": 1.9715789473684214e-05,897 "loss": 0.7972,898 "step": 127899 },900 {901 "epoch": 0.19783616692426584,902 "grad_norm": 12.3125,903 "learning_rate": 1.970526315789474e-05,904 "loss": 0.9165,905 "step": 128906 },907 {908 "epoch": 0.19938176197836166,909 "grad_norm": 12.9375,910 "learning_rate": 1.9694736842105263e-05,911 "loss": 0.9013,912 "step": 129913 },914 {915 "epoch": 0.2009273570324575,916 "grad_norm": 12.5,917 "learning_rate": 1.968421052631579e-05,918 "loss": 0.8593,919 "step": 130920 },921 {922 "epoch": 0.20247295208655333,923 "grad_norm": 13.125,924 "learning_rate": 1.967368421052632e-05,925 "loss": 0.8728,926 "step": 131927 },928 {929 "epoch": 0.20401854714064915,930 "grad_norm": 12.5625,931 "learning_rate": 1.9663157894736844e-05,932 "loss": 0.8588,933 "step": 132934 },935 {936 "epoch": 0.20556414219474498,937 "grad_norm": 13.125,938 "learning_rate": 1.965263157894737e-05,939 "loss": 0.837,940 "step": 133941 },942 {943 "epoch": 0.2071097372488408,944 "grad_norm": 12.875,945 "learning_rate": 1.9642105263157897e-05,946 "loss": 0.8504,947 "step": 134948 },949 {950 "epoch": 0.20865533230293662,951 "grad_norm": 15.5625,952 "learning_rate": 1.9631578947368425e-05,953 "loss": 0.86,954 "step": 135955 },956 {957 "epoch": 0.21020092735703247,958 "grad_norm": 13.0,959 "learning_rate": 1.962105263157895e-05,960 "loss": 0.7569,961 "step": 136962 },963 {964 "epoch": 0.2117465224111283,965 "grad_norm": 12.75,966 "learning_rate": 1.9610526315789474e-05,967 "loss": 0.8461,968 "step": 137969 },970 {971 "epoch": 0.2132921174652241,972 "grad_norm": 29.25,973 "learning_rate": 1.9600000000000002e-05,974 "loss": 0.7988,975 "step": 138976 },977 {978 "epoch": 0.21483771251931993,979 "grad_norm": 13.5625,980 "learning_rate": 1.9589473684210527e-05,981 "loss": 0.8452,982 "step": 139983 },984 {985 "epoch": 0.21638330757341576,986 "grad_norm": 14.625,987 "learning_rate": 1.9578947368421055e-05,988 "loss": 0.8358,989 "step": 140990 },991 {992 "epoch": 0.21792890262751158,993 "grad_norm": 11.625,994 "learning_rate": 1.956842105263158e-05,995 "loss": 0.7932,996 "step": 141997 },998 {999 "epoch": 0.21947449768160743,1000 "grad_norm": 12.75,1001 "learning_rate": 1.9557894736842107e-05,1002 "loss": 0.817,1003 "step": 1421004 },1005 {1006 "epoch": 0.22102009273570325,1007 "grad_norm": 16.875,1008 "learning_rate": 1.9547368421052632e-05,1009 "loss": 0.795,1010 "step": 1431011 },1012 {1013 "epoch": 0.22256568778979907,1014 "grad_norm": 13.0625,1015 "learning_rate": 1.953684210526316e-05,1016 "loss": 0.7426,1017 "step": 1441018 },1019 {1020 "epoch": 0.2241112828438949,1021 "grad_norm": 11.3125,1022 "learning_rate": 1.9526315789473688e-05,1023 "loss": 0.76,1024 "step": 1451025 },1026 {1027 "epoch": 0.22565687789799072,1028 "grad_norm": 11.625,1029 "learning_rate": 1.9515789473684213e-05,1030 "loss": 0.8191,1031 "step": 1461032 },1033 {1034 "epoch": 0.22720247295208656,1035 "grad_norm": 12.3125,1036 "learning_rate": 1.9505263157894737e-05,1037 "loss": 0.8164,1038 "step": 1471039 },1040 {1041 "epoch": 0.2287480680061824,1042 "grad_norm": 11.5,1043 "learning_rate": 1.9494736842105265e-05,1044 "loss": 0.8299,1045 "step": 1481046 },1047 {1048 "epoch": 0.2302936630602782,1049 "grad_norm": 16.125,1050 "learning_rate": 1.9484210526315793e-05,1051 "loss": 0.8695,1052 "step": 1491053 },1054 {1055 "epoch": 0.23183925811437403,1056 "grad_norm": 12.5625,1057 "learning_rate": 1.9473684210526318e-05,1058 "loss": 0.8691,1059 "step": 1501060 },1061 {1062 "epoch": 0.23338485316846985,1063 "grad_norm": 12.125,1064 "learning_rate": 1.9463157894736843e-05,1065 "loss": 0.8504,1066 "step": 1511067 },1068 {1069 "epoch": 0.23493044822256567,1070 "grad_norm": 15.5625,1071 "learning_rate": 1.945263157894737e-05,1072 "loss": 0.8583,1073 "step": 1521074 },1075 {1076 "epoch": 0.23647604327666152,1077 "grad_norm": 11.0625,1078 "learning_rate": 1.9442105263157895e-05,1079 "loss": 0.9378,1080 "step": 1531081 },1082 {1083 "epoch": 0.23802163833075735,1084 "grad_norm": 12.6875,1085 "learning_rate": 1.9431578947368423e-05,1086 "loss": 0.9021,1087 "step": 1541088 },1089 {1090 "epoch": 0.23956723338485317,1091 "grad_norm": 12.625,1092 "learning_rate": 1.9421052631578948e-05,1093 "loss": 0.8527,1094 "step": 1551095 },1096 {1097 "epoch": 0.241112828438949,1098 "grad_norm": 11.6875,1099 "learning_rate": 1.9410526315789476e-05,1100 "loss": 0.8327,1101 "step": 1561102 },1103 {1104 "epoch": 0.2426584234930448,1105 "grad_norm": 11.9375,1106 "learning_rate": 1.94e-05,1107 "loss": 0.8003,1108 "step": 1571109 },1110 {1111 "epoch": 0.24420401854714066,1112 "grad_norm": 13.375,1113 "learning_rate": 1.9389473684210525e-05,1114 "loss": 0.6912,1115 "step": 1581116 },1117 {1118 "epoch": 0.24574961360123648,1119 "grad_norm": 10.6875,1120 "learning_rate": 1.9378947368421053e-05,1121 "loss": 0.7362,1122 "step": 1591123 },1124 {1125 "epoch": 0.2472952086553323,1126 "grad_norm": 11.0,1127 "learning_rate": 1.936842105263158e-05,1128 "loss": 0.7379,1129 "step": 1601130 },1131 {1132 "epoch": 0.24884080370942813,1133 "grad_norm": 25.625,1134 "learning_rate": 1.9357894736842106e-05,1135 "loss": 0.7803,1136 "step": 1611137 },1138 {1139 "epoch": 0.250386398763524,1140 "grad_norm": 13.125,1141 "learning_rate": 1.9347368421052634e-05,1142 "loss": 0.8244,1143 "step": 1621144 },1145 {1146 "epoch": 0.25193199381761977,1147 "grad_norm": 13.625,1148 "learning_rate": 1.9336842105263162e-05,1149 "loss": 0.773,1150 "step": 1631151 },1152 {1153 "epoch": 0.2534775888717156,1154 "grad_norm": 11.625,1155 "learning_rate": 1.9326315789473687e-05,1156 "loss": 0.7818,1157 "step": 1641158 },1159 {1160 "epoch": 0.2550231839258114,1161 "grad_norm": 11.5625,1162 "learning_rate": 1.931578947368421e-05,1163 "loss": 0.825,1164 "step": 1651165 },1166 {1167 "epoch": 0.25656877897990726,1168 "grad_norm": 10.75,1169 "learning_rate": 1.930526315789474e-05,1170 "loss": 0.7203,1171 "step": 1661172 },1173 {1174 "epoch": 0.2581143740340031,1175 "grad_norm": 13.0625,1176 "learning_rate": 1.9294736842105264e-05,1177 "loss": 0.8392,1178 "step": 1671179 },1180 {1181 "epoch": 0.2596599690880989,1182 "grad_norm": 12.875,1183 "learning_rate": 1.9284210526315792e-05,1184 "loss": 0.8234,1185 "step": 1681186 },1187 {1188 "epoch": 0.26120556414219476,1189 "grad_norm": 12.5625,1190 "learning_rate": 1.9273684210526317e-05,1191 "loss": 0.8638,1192 "step": 1691193 },1194 {1195 "epoch": 0.26275115919629055,1196 "grad_norm": 10.5625,1197 "learning_rate": 1.9263157894736845e-05,1198 "loss": 0.7969,1199 "step": 1701200 },