CLMBR/binding-c-command-transformer-0
020
1{2 "best_metric": 3.842787981033325,3 "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/binding-c-command/transformer/0/checkpoints/checkpoint-992160",4 "epoch": 0.025000606015738065,5 "eval_steps": 10,6 "global_step": 1297440,7 "is_hyper_param_search": false,8 "is_local_process_zero": true,9 "is_world_process_zero": true,10 "log_history": [11 {12 "epoch": 0.0,13 "learning_rate": 4.999998362119627e-05,14 "loss": 10.9624,15 "step": 116 },17 {18 "epoch": 0.0,19 "learning_rate": 4.999161405248948e-05,20 "loss": 6.822,21 "step": 51222 },23 {24 "epoch": 0.0,25 "learning_rate": 4.998322810497896e-05,26 "loss": 6.1893,27 "step": 102428 },29 {30 "epoch": 0.0,31 "learning_rate": 4.997484215746844e-05,32 "loss": 5.9774,33 "step": 153634 },35 {36 "epoch": 0.0,37 "learning_rate": 4.996645620995792e-05,38 "loss": 5.8082,39 "step": 204840 },41 {42 "epoch": 0.0,43 "learning_rate": 4.99580702624474e-05,44 "loss": 5.6938,45 "step": 256046 },47 {48 "epoch": 0.0,49 "learning_rate": 4.994968431493688e-05,50 "loss": 5.5921,51 "step": 307252 },53 {54 "epoch": 0.0,55 "learning_rate": 4.994129836742636e-05,56 "loss": 5.5277,57 "step": 358458 },59 {60 "epoch": 0.0,61 "learning_rate": 4.993291241991584e-05,62 "loss": 5.4338,63 "step": 409664 },65 {66 "epoch": 0.0,67 "learning_rate": 4.992452647240532e-05,68 "loss": 5.4008,69 "step": 460870 },71 {72 "epoch": 0.0,73 "learning_rate": 4.99161405248948e-05,74 "loss": 5.3401,75 "step": 512076 },77 {78 "epoch": 0.0,79 "learning_rate": 4.990775457738428e-05,80 "loss": 5.3007,81 "step": 563282 },83 {84 "epoch": 0.0,85 "learning_rate": 4.989938500867749e-05,86 "loss": 5.2593,87 "step": 614488 },89 {90 "epoch": 0.0,91 "learning_rate": 4.989099906116697e-05,92 "loss": 5.1918,93 "step": 665694 },95 {96 "epoch": 0.0,97 "learning_rate": 4.988261311365645e-05,98 "loss": 5.1743,99 "step": 7168100 },101 {102 "epoch": 0.0,103 "learning_rate": 4.987422716614593e-05,104 "loss": 5.1266,105 "step": 7680106 },107 {108 "epoch": 0.0,109 "learning_rate": 4.986584121863541e-05,110 "loss": 5.1042,111 "step": 8192112 },113 {114 "epoch": 0.0,115 "learning_rate": 4.985745527112489e-05,116 "loss": 5.0762,117 "step": 8704118 },119 {120 "epoch": 0.0,121 "learning_rate": 4.984906932361437e-05,122 "loss": 5.0383,123 "step": 9216124 },125 {126 "epoch": 0.0,127 "learning_rate": 4.984068337610385e-05,128 "loss": 5.0149,129 "step": 9728130 },131 {132 "epoch": 0.0,133 "learning_rate": 4.983229742859333e-05,134 "loss": 5.0018,135 "step": 10240136 },137 {138 "epoch": 0.0,139 "learning_rate": 4.9823927859886547e-05,140 "loss": 4.972,141 "step": 10752142 },143 {144 "epoch": 0.0,145 "learning_rate": 4.9815541912376026e-05,146 "loss": 4.9446,147 "step": 11264148 },149 {150 "epoch": 0.0,151 "learning_rate": 4.9807155964865506e-05,152 "loss": 4.9231,153 "step": 11776154 },155 {156 "epoch": 0.0,157 "learning_rate": 4.9798770017354986e-05,158 "loss": 4.9142,159 "step": 12288160 },161 {162 "epoch": 0.0,163 "learning_rate": 4.9790400448648195e-05,164 "loss": 4.8885,165 "step": 12800166 },167 {168 "epoch": 0.0,169 "learning_rate": 4.9782014501137675e-05,170 "loss": 4.8615,171 "step": 13312172 },173 {174 "epoch": 0.0,175 "learning_rate": 4.9773628553627155e-05,176 "loss": 4.8521,177 "step": 13824178 },179 {180 "epoch": 0.0,181 "learning_rate": 4.9765242606116635e-05,182 "loss": 4.8317,183 "step": 14336184 },185 {186 "epoch": 0.0,187 "learning_rate": 4.9756873037409844e-05,188 "loss": 4.8177,189 "step": 14848190 },191 {192 "epoch": 0.01,193 "learning_rate": 4.9748487089899324e-05,194 "loss": 4.8048,195 "step": 15360196 },197 {198 "epoch": 0.01,199 "learning_rate": 4.9740101142388804e-05,200 "loss": 4.7885,201 "step": 15872202 },203 {204 "epoch": 0.01,205 "learning_rate": 4.9731715194878284e-05,206 "loss": 4.7788,207 "step": 16384208 },209 {210 "epoch": 0.01,211 "learning_rate": 4.97233456261715e-05,212 "loss": 4.7606,213 "step": 16896214 },215 {216 "epoch": 0.01,217 "learning_rate": 4.971495967866098e-05,218 "loss": 4.7554,219 "step": 17408220 },221 {222 "epoch": 0.01,223 "learning_rate": 4.970657373115046e-05,224 "loss": 4.756,225 "step": 17920226 },227 {228 "epoch": 0.01,229 "learning_rate": 4.969818778363994e-05,230 "loss": 4.7316,231 "step": 18432232 },233 {234 "epoch": 0.01,235 "learning_rate": 4.968981821493315e-05,236 "loss": 4.6953,237 "step": 18944238 },239 {240 "epoch": 0.01,241 "learning_rate": 4.968143226742263e-05,242 "loss": 4.7007,243 "step": 19456244 },245 {246 "epoch": 0.01,247 "learning_rate": 4.967304631991211e-05,248 "loss": 4.6801,249 "step": 19968250 },251 {252 "epoch": 0.01,253 "learning_rate": 4.966466037240159e-05,254 "loss": 4.6846,255 "step": 20480256 },257 {258 "epoch": 0.01,259 "learning_rate": 4.96562908036948e-05,260 "loss": 4.6714,261 "step": 20992262 },263 {264 "epoch": 0.01,265 "learning_rate": 4.964790485618428e-05,266 "loss": 4.6626,267 "step": 21504268 },269 {270 "epoch": 0.01,271 "learning_rate": 4.963951890867376e-05,272 "loss": 4.6429,273 "step": 22016274 },275 {276 "epoch": 0.01,277 "learning_rate": 4.963113296116324e-05,278 "loss": 4.6519,279 "step": 22528280 },281 {282 "epoch": 0.01,283 "learning_rate": 4.962274701365272e-05,284 "loss": 4.6346,285 "step": 23040286 },287 {288 "epoch": 0.01,289 "learning_rate": 4.9614377444945934e-05,290 "loss": 4.6331,291 "step": 23552292 },293 {294 "epoch": 0.01,295 "learning_rate": 4.9605991497435414e-05,296 "loss": 4.623,297 "step": 24064298 },299 {300 "epoch": 0.01,301 "learning_rate": 4.9597605549924894e-05,302 "loss": 4.6016,303 "step": 24576304 },305 {306 "epoch": 0.01,307 "learning_rate": 4.95892359812181e-05,308 "loss": 4.6024,309 "step": 25088310 },311 {312 "epoch": 0.01,313 "learning_rate": 4.958085003370758e-05,314 "loss": 4.6049,315 "step": 25600316 },317 {318 "epoch": 0.01,319 "learning_rate": 4.957246408619706e-05,320 "loss": 4.5845,321 "step": 26112322 },323 {324 "epoch": 0.01,325 "learning_rate": 4.956407813868654e-05,326 "loss": 4.5752,327 "step": 26624328 },329 {330 "epoch": 0.01,331 "learning_rate": 4.9555692191176016e-05,332 "loss": 4.5654,333 "step": 27136334 },335 {336 "epoch": 0.01,337 "learning_rate": 4.9547306243665496e-05,338 "loss": 4.5651,339 "step": 27648340 },341 {342 "epoch": 0.01,343 "learning_rate": 4.9538920296154976e-05,344 "loss": 4.5488,345 "step": 28160346 },347 {348 "epoch": 0.01,349 "learning_rate": 4.9530534348644456e-05,350 "loss": 4.5677,351 "step": 28672352 },353 {354 "epoch": 0.01,355 "learning_rate": 4.952216477993767e-05,356 "loss": 4.5365,357 "step": 29184358 },359 {360 "epoch": 0.01,361 "learning_rate": 4.951377883242715e-05,362 "loss": 4.5446,363 "step": 29696364 },365 {366 "epoch": 0.01,367 "learning_rate": 4.950539288491663e-05,368 "loss": 4.5359,369 "step": 30208370 },371 {372 "epoch": 0.01,373 "learning_rate": 4.949700693740611e-05,374 "loss": 4.5111,375 "step": 30720376 },377 {378 "epoch": 0.01,379 "learning_rate": 4.948862098989559e-05,380 "loss": 4.5192,381 "step": 31232382 },383 {384 "epoch": 0.01,385 "learning_rate": 4.94802514211888e-05,386 "loss": 4.5095,387 "step": 31744388 },389 {390 "epoch": 0.01,391 "learning_rate": 4.947186547367828e-05,392 "loss": 4.4953,393 "step": 32256394 },395 {396 "epoch": 0.01,397 "learning_rate": 4.946347952616776e-05,398 "loss": 4.4871,399 "step": 32768400 },401 {402 "epoch": 0.01,403 "learning_rate": 4.945509357865724e-05,404 "loss": 4.5026,405 "step": 33280406 },407 {408 "epoch": 0.01,409 "learning_rate": 4.944672400995045e-05,410 "loss": 4.482,411 "step": 33792412 },413 {414 "epoch": 0.01,415 "learning_rate": 4.943833806243993e-05,416 "loss": 4.4749,417 "step": 34304418 },419 {420 "epoch": 0.01,421 "learning_rate": 4.942995211492941e-05,422 "loss": 4.464,423 "step": 34816424 },425 {426 "epoch": 0.01,427 "learning_rate": 4.9421582546222625e-05,428 "loss": 4.4767,429 "step": 35328430 },431 {432 "epoch": 0.01,433 "learning_rate": 4.9413196598712105e-05,434 "loss": 4.481,435 "step": 35840436 },437 {438 "epoch": 0.01,439 "learning_rate": 4.9404810651201585e-05,440 "loss": 4.466,441 "step": 36352442 },443 {444 "epoch": 0.01,445 "learning_rate": 4.9396424703691065e-05,446 "loss": 4.4608,447 "step": 36864448 },449 {450 "epoch": 0.01,451 "learning_rate": 4.9388038756180545e-05,452 "loss": 4.4643,453 "step": 37376454 },455 {456 "epoch": 0.01,457 "learning_rate": 4.9379652808670025e-05,458 "loss": 4.4623,459 "step": 37888460 },461 {462 "epoch": 0.01,463 "learning_rate": 4.9371266861159505e-05,464 "loss": 4.4513,465 "step": 38400466 },467 {468 "epoch": 0.01,469 "learning_rate": 4.9362897292452714e-05,470 "loss": 4.4343,471 "step": 38912472 },473 {474 "epoch": 0.01,475 "learning_rate": 4.9354511344942194e-05,476 "loss": 4.4415,477 "step": 39424478 },479 {480 "epoch": 0.01,481 "learning_rate": 4.9346125397431674e-05,482 "loss": 4.4303,483 "step": 39936484 },485 {486 "epoch": 0.01,487 "learning_rate": 4.9337739449921154e-05,488 "loss": 4.4284,489 "step": 40448490 },491 {492 "epoch": 0.01,493 "learning_rate": 4.9329353502410634e-05,494 "loss": 4.417,495 "step": 40960496 },497 {498 "epoch": 0.01,499 "learning_rate": 4.9320967554900114e-05,500 "loss": 4.4235,501 "step": 41472502 },503 {504 "epoch": 0.01,505 "learning_rate": 4.9312581607389594e-05,506 "loss": 4.428,507 "step": 41984508 },509 {510 "epoch": 0.01,511 "learning_rate": 4.9304195659879074e-05,512 "loss": 4.4195,513 "step": 42496514 },515 {516 "epoch": 0.01,517 "learning_rate": 4.929582609117229e-05,518 "loss": 4.3942,519 "step": 43008520 },521 {522 "epoch": 0.01,523 "learning_rate": 4.92874565224655e-05,524 "loss": 4.3995,525 "step": 43520526 },527 {528 "epoch": 0.01,529 "learning_rate": 4.927907057495498e-05,530 "loss": 4.4045,531 "step": 44032532 },533 {534 "epoch": 0.01,535 "learning_rate": 4.927068462744446e-05,536 "loss": 4.3984,537 "step": 44544538 },539 {540 "epoch": 0.01,541 "learning_rate": 4.926229867993394e-05,542 "loss": 4.3933,543 "step": 45056544 },545 {546 "epoch": 0.01,547 "learning_rate": 4.925391273242342e-05,548 "loss": 4.3903,549 "step": 45568550 },551 {552 "epoch": 0.02,553 "learning_rate": 4.92455267849129e-05,554 "loss": 4.3846,555 "step": 46080556 },557 {558 "epoch": 0.02,559 "learning_rate": 4.923714083740238e-05,560 "loss": 4.3713,561 "step": 46592562 },563 {564 "epoch": 0.02,565 "learning_rate": 4.922877126869559e-05,566 "loss": 4.3764,567 "step": 47104568 },569 {570 "epoch": 0.02,571 "learning_rate": 4.922038532118507e-05,572 "loss": 4.3763,573 "step": 47616574 },575 {576 "epoch": 0.02,577 "learning_rate": 4.921199937367455e-05,578 "loss": 4.3682,579 "step": 48128580 },581 {582 "epoch": 0.02,583 "learning_rate": 4.920361342616403e-05,584 "loss": 4.3771,585 "step": 48640586 },587 {588 "epoch": 0.02,589 "learning_rate": 4.919522747865351e-05,590 "loss": 4.3572,591 "step": 49152592 },593 {594 "epoch": 0.02,595 "learning_rate": 4.918684153114299e-05,596 "loss": 4.358,597 "step": 49664598 },599 {600 "epoch": 0.02,601 "learning_rate": 4.917845558363247e-05,602 "loss": 4.3554,603 "step": 50176604 },605 {606 "epoch": 0.02,607 "learning_rate": 4.917006963612195e-05,608 "loss": 4.3545,609 "step": 50688610 },611 {612 "epoch": 0.02,613 "learning_rate": 4.916171644621889e-05,614 "loss": 4.3583,615 "step": 51200616 },617 {618 "epoch": 0.02,619 "learning_rate": 4.915333049870837e-05,620 "loss": 4.3416,621 "step": 51712622 },623 {624 "epoch": 0.02,625 "learning_rate": 4.914494455119785e-05,626 "loss": 4.3392,627 "step": 52224628 },629 {630 "epoch": 0.02,631 "learning_rate": 4.9136558603687325e-05,632 "loss": 4.3352,633 "step": 52736634 },635 {636 "epoch": 0.02,637 "learning_rate": 4.9128172656176805e-05,638 "loss": 4.34,639 "step": 53248640 },641 {642 "epoch": 0.02,643 "learning_rate": 4.9119786708666285e-05,644 "loss": 4.3212,645 "step": 53760646 },647 {648 "epoch": 0.02,649 "learning_rate": 4.91114171399595e-05,650 "loss": 4.3129,651 "step": 54272652 },653 {654 "epoch": 0.02,655 "learning_rate": 4.910303119244898e-05,656 "loss": 4.324,657 "step": 54784658 },659 {660 "epoch": 0.02,661 "learning_rate": 4.909464524493846e-05,662 "loss": 4.3285,663 "step": 55296664 },665 {666 "epoch": 0.02,667 "learning_rate": 4.908625929742794e-05,668 "loss": 4.3214,669 "step": 55808670 },671 {672 "epoch": 0.02,673 "learning_rate": 4.907787334991742e-05,674 "loss": 4.32,675 "step": 56320676 },677 {678 "epoch": 0.02,679 "learning_rate": 4.90694874024069e-05,680 "loss": 4.3071,681 "step": 56832682 },683 {684 "epoch": 0.02,685 "learning_rate": 4.906110145489638e-05,686 "loss": 4.3175,687 "step": 57344688 },689 {690 "epoch": 0.02,691 "learning_rate": 4.905273188618959e-05,692 "loss": 4.3005,693 "step": 57856694 },695 {696 "epoch": 0.02,697 "learning_rate": 4.904434593867907e-05,698 "loss": 4.3082,699 "step": 58368700 },701 {702 "epoch": 0.02,703 "learning_rate": 4.903595999116855e-05,704 "loss": 4.3083,705 "step": 58880706 },707 {708 "epoch": 0.02,709 "learning_rate": 4.902757404365803e-05,710 "loss": 4.3094,711 "step": 59392712 },713 {714 "epoch": 0.02,715 "learning_rate": 4.901920447495124e-05,716 "loss": 4.2979,717 "step": 59904718 },719 {720 "epoch": 0.02,721 "learning_rate": 4.901081852744072e-05,722 "loss": 4.2957,723 "step": 60416724 },725 {726 "epoch": 0.02,727 "learning_rate": 4.90024325799302e-05,728 "loss": 4.2959,729 "step": 60928730 },731 {732 "epoch": 0.02,733 "learning_rate": 4.8994046632419686e-05,734 "loss": 4.2984,735 "step": 61440736 },737 {738 "epoch": 0.02,739 "learning_rate": 4.8985660684909166e-05,740 "loss": 4.2861,741 "step": 61952742 },743 {744 "epoch": 0.02,745 "learning_rate": 4.8977274737398646e-05,746 "loss": 4.2889,747 "step": 62464748 },749 {750 "epoch": 0.02,751 "learning_rate": 4.8968888789888125e-05,752 "loss": 4.285,753 "step": 62976754 },755 {756 "epoch": 0.02,757 "learning_rate": 4.8960519221181335e-05,758 "loss": 4.2868,759 "step": 63488760 },761 {762 "epoch": 0.02,763 "learning_rate": 4.8952133273670815e-05,764 "loss": 4.2715,765 "step": 64000766 },767 {768 "epoch": 0.02,769 "learning_rate": 4.8943747326160294e-05,770 "loss": 4.2681,771 "step": 64512772 },773 {774 "epoch": 0.02,775 "learning_rate": 4.8935361378649774e-05,776 "loss": 4.274,777 "step": 65024778 },779 {780 "epoch": 0.02,781 "learning_rate": 4.8926991809942984e-05,782 "loss": 4.2763,783 "step": 65536784 },785 {786 "epoch": 0.02,787 "learning_rate": 4.8918605862432463e-05,788 "loss": 4.2667,789 "step": 66048790 },791 {792 "epoch": 0.02,793 "learning_rate": 4.8910219914921943e-05,794 "loss": 4.2809,795 "step": 66560796 },797 {798 "epoch": 0.02,799 "learning_rate": 4.8901833967411423e-05,800 "loss": 4.2703,801 "step": 67072802 },803 {804 "epoch": 0.02,805 "learning_rate": 4.889346439870464e-05,806 "loss": 4.2557,807 "step": 67584808 },809 {810 "epoch": 0.02,811 "learning_rate": 4.888507845119412e-05,812 "loss": 4.2638,813 "step": 68096814 },815 {816 "epoch": 0.02,817 "learning_rate": 4.88766925036836e-05,818 "loss": 4.2722,819 "step": 68608820 },821 {822 "epoch": 0.02,823 "learning_rate": 4.886830655617308e-05,824 "loss": 4.2618,825 "step": 69120826 },827 {828 "epoch": 0.02,829 "learning_rate": 4.885993698746629e-05,830 "loss": 4.2588,831 "step": 69632832 },833 {834 "epoch": 0.02,835 "learning_rate": 4.885155103995577e-05,836 "loss": 4.2492,837 "step": 70144838 },839 {840 "epoch": 0.02,841 "learning_rate": 4.884316509244525e-05,842 "loss": 4.257,843 "step": 70656844 },845 {846 "epoch": 0.02,847 "learning_rate": 4.883479552373846e-05,848 "loss": 4.2614,849 "step": 71168850 },851 {852 "epoch": 0.02,853 "learning_rate": 4.882640957622794e-05,854 "loss": 4.2466,855 "step": 71680856 },857 {858 "epoch": 0.02,859 "learning_rate": 4.881802362871742e-05,860 "loss": 4.2398,861 "step": 72192862 },863 {864 "epoch": 0.02,865 "learning_rate": 4.88096376812069e-05,866 "loss": 4.2453,867 "step": 72704868 },869 {870 "epoch": 0.02,871 "learning_rate": 4.880125173369638e-05,872 "loss": 4.2382,873 "step": 73216874 },875 {876 "epoch": 0.02,877 "learning_rate": 4.8792882164989586e-05,878 "loss": 4.2353,879 "step": 73728880 },881 {882 "epoch": 0.02,883 "learning_rate": 4.878449621747907e-05,884 "loss": 4.2438,885 "step": 74240886 },887 {888 "epoch": 0.02,889 "learning_rate": 4.877611026996855e-05,890 "loss": 4.2317,891 "step": 74752892 },893 {894 "epoch": 0.02,895 "learning_rate": 4.876772432245803e-05,896 "loss": 4.2344,897 "step": 75264898 },899 {900 "epoch": 0.02,901 "learning_rate": 4.875933837494751e-05,902 "loss": 4.2301,903 "step": 75776904 },905 {906 "epoch": 0.02,907 "learning_rate": 4.8750952427436986e-05,908 "loss": 4.2256,909 "step": 76288910 },911 {912 "epoch": 0.03,913 "eval_loss": 4.193907260894775,914 "eval_runtime": 305.9942,915 "eval_samples_per_second": 1247.053,916 "eval_steps_per_second": 38.971,917 "step": 76320918 },919 {920 "epoch": 1.0,921 "learning_rate": 4.8742566479926466e-05,922 "loss": 4.2149,923 "step": 76800924 },925 {926 "epoch": 1.0,927 "learning_rate": 4.8734180532415946e-05,928 "loss": 4.2221,929 "step": 77312930 },931 {932 "epoch": 1.0,933 "learning_rate": 4.872581096370916e-05,934 "loss": 4.2301,935 "step": 77824936 },937 {938 "epoch": 1.0,939 "learning_rate": 4.8717425016198635e-05,940 "loss": 4.219,941 "step": 78336942 },943 {944 "epoch": 1.0,945 "learning_rate": 4.8709039068688115e-05,946 "loss": 4.2201,947 "step": 78848948 },949 {950 "epoch": 1.0,951 "learning_rate": 4.8700653121177595e-05,952 "loss": 4.2136,953 "step": 79360954 },955 {956 "epoch": 1.0,957 "learning_rate": 4.869228355247081e-05,958 "loss": 4.2104,959 "step": 79872960 },961 {962 "epoch": 1.0,963 "learning_rate": 4.868389760496029e-05,964 "loss": 4.1966,965 "step": 80384966 },967 {968 "epoch": 1.0,969 "learning_rate": 4.8675528036253507e-05,970 "loss": 4.2225,971 "step": 80896972 },973 {974 "epoch": 1.0,975 "learning_rate": 4.8667142088742986e-05,976 "loss": 4.2065,977 "step": 81408978 },979 {980 "epoch": 1.0,981 "learning_rate": 4.865875614123246e-05,982 "loss": 4.2163,983 "step": 81920984 },985 {986 "epoch": 1.0,987 "learning_rate": 4.865037019372194e-05,988 "loss": 4.2125,989 "step": 82432990 },991 {992 "epoch": 1.0,993 "learning_rate": 4.8642000625015155e-05,994 "loss": 4.1956,995 "step": 82944996 },997 {998 "epoch": 1.0,999 "learning_rate": 4.8633614677504635e-05,1000 "loss": 4.1966,1001 "step": 834561002 },1003 {1004 "epoch": 1.0,1005 "learning_rate": 4.862522872999411e-05,1006 "loss": 4.1915,1007 "step": 839681008 },1009 {1010 "epoch": 1.0,1011 "learning_rate": 4.861684278248359e-05,1012 "loss": 4.1838,1013 "step": 844801014 },1015 {1016 "epoch": 1.0,1017 "learning_rate": 4.860845683497307e-05,1018 "loss": 4.1903,1019 "step": 849921020 },1021 {1022 "epoch": 1.0,1023 "learning_rate": 4.860007088746255e-05,1024 "loss": 4.1828,1025 "step": 855041026 },1027 {1028 "epoch": 1.0,1029 "learning_rate": 4.859168493995203e-05,1030 "loss": 4.1796,1031 "step": 860161032 },1033 {1034 "epoch": 1.0,1035 "learning_rate": 4.858329899244151e-05,1036 "loss": 4.2006,1037 "step": 865281038 },1039 {1040 "epoch": 1.0,1041 "learning_rate": 4.8574913044930995e-05,1042 "loss": 4.1851,1043 "step": 870401044 },1045 {1046 "epoch": 1.0,1047 "learning_rate": 4.8566527097420475e-05,1048 "loss": 4.1773,1049 "step": 875521050 },1051 {1052 "epoch": 1.0,1053 "learning_rate": 4.8558141149909955e-05,1054 "loss": 4.175,1055 "step": 880641056 },1057 {1058 "epoch": 1.0,1059 "learning_rate": 4.8549771581203164e-05,1060 "loss": 4.1887,1061 "step": 885761062 },1063 {1064 "epoch": 1.0,1065 "learning_rate": 4.854140201249637e-05,1066 "loss": 4.167,1067 "step": 890881068 },1069 {1070 "epoch": 1.0,1071 "learning_rate": 4.853301606498585e-05,1072 "loss": 4.163,1073 "step": 896001074 },1075 {1076 "epoch": 1.0,1077 "learning_rate": 4.852463011747533e-05,1078 "loss": 4.1674,1079 "step": 901121080 },1081 {1082 "epoch": 1.0,1083 "learning_rate": 4.851624416996481e-05,1084 "loss": 4.168,1085 "step": 906241086 },1087 {1088 "epoch": 1.0,1089 "learning_rate": 4.850785822245429e-05,1090 "loss": 4.1622,1091 "step": 911361092 },1093 {1094 "epoch": 1.01,1095 "learning_rate": 4.849947227494377e-05,1096 "loss": 4.1625,1097 "step": 916481098 },1099 {1100 "epoch": 1.01,1101 "learning_rate": 4.849110270623698e-05,1102 "loss": 4.1634,1103 "step": 921601104 },1105 {1106 "epoch": 1.01,1107 "learning_rate": 4.848271675872646e-05,1108 "loss": 4.1675,1109 "step": 926721110 },1111 {1112 "epoch": 1.01,1113 "learning_rate": 4.847433081121595e-05,1114 "loss": 4.1523,1115 "step": 931841116 },1117 {1118 "epoch": 1.01,1119 "learning_rate": 4.846594486370543e-05,1120 "loss": 4.1616,1121 "step": 936961122 },1123 {1124 "epoch": 1.01,1125 "learning_rate": 4.845755891619491e-05,1126 "loss": 4.1674,1127 "step": 942081128 },1129 {1130 "epoch": 1.01,1131 "learning_rate": 4.844918934748812e-05,1132 "loss": 4.1636,1133 "step": 947201134 },1135 {1136 "epoch": 1.01,1137 "learning_rate": 4.84408033999776e-05,1138 "loss": 4.1358,1139 "step": 952321140 },1141 {1142 "epoch": 1.01,1143 "learning_rate": 4.843241745246708e-05,1144 "loss": 4.1439,1145 "step": 957441146 },1147 {1148 "epoch": 1.01,1149 "learning_rate": 4.842403150495656e-05,1150 "loss": 4.1424,1151 "step": 962561152 },1153 {1154 "epoch": 1.01,1155 "learning_rate": 4.841564555744604e-05,1156 "loss": 4.1458,1157 "step": 967681158 },1159 {1160 "epoch": 1.01,1161 "learning_rate": 4.840725960993552e-05,1162 "loss": 4.1399,1163 "step": 972801164 },1165 {1166 "epoch": 1.01,1167 "learning_rate": 4.8398873662425e-05,1168 "loss": 4.146,1169 "step": 977921170 },1171 {1172 "epoch": 1.01,1173 "learning_rate": 4.839050409371821e-05,1174 "loss": 4.1351,1175 "step": 983041176 },1177 {1178 "epoch": 1.01,1179 "learning_rate": 4.8382118146207687e-05,1180 "loss": 4.1434,1181 "step": 988161182 },1183 {1184 "epoch": 1.01,1185 "learning_rate": 4.8373732198697167e-05,1186 "loss": 4.1413,1187 "step": 993281188 },1189 {1190 "epoch": 1.01,1191 "learning_rate": 4.8365346251186646e-05,1192 "loss": 4.1428,1193 "step": 998401194 },1195 {1196 "epoch": 1.01,1197 "learning_rate": 4.8356960303676126e-05,1198 "loss": 4.1404,1199 "step": 1003521200 },