SurplusDeficit/GeLM
0
1{2 "best_metric": null,3 "best_model_checkpoint": null,4 "epoch": 5.0,5 "eval_steps": 500,6 "global_step": 2620,7 "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": 1.2658227848101266e-07,14 "loss": 8.0625,15 "step": 116 },17 {18 "epoch": 0.0,19 "learning_rate": 2.5316455696202533e-07,20 "loss": 8.0547,21 "step": 222 },23 {24 "epoch": 0.01,25 "learning_rate": 3.79746835443038e-07,26 "loss": 8.0625,27 "step": 328 },29 {30 "epoch": 0.01,31 "learning_rate": 5.063291139240507e-07,32 "loss": 8.4531,33 "step": 434 },35 {36 "epoch": 0.01,37 "learning_rate": 6.329113924050634e-07,38 "loss": 7.9844,39 "step": 540 },41 {42 "epoch": 0.01,43 "learning_rate": 7.59493670886076e-07,44 "loss": 8.0781,45 "step": 646 },47 {48 "epoch": 0.01,49 "learning_rate": 8.860759493670887e-07,50 "loss": 8.0938,51 "step": 752 },53 {54 "epoch": 0.02,55 "learning_rate": 1.0126582278481013e-06,56 "loss": 7.9688,57 "step": 858 },59 {60 "epoch": 0.02,61 "learning_rate": 1.139240506329114e-06,62 "loss": 7.9531,63 "step": 964 },65 {66 "epoch": 0.02,67 "learning_rate": 1.2658227848101267e-06,68 "loss": 7.9453,69 "step": 1070 },71 {72 "epoch": 0.02,73 "learning_rate": 1.3924050632911392e-06,74 "loss": 7.9062,75 "step": 1176 },77 {78 "epoch": 0.02,79 "learning_rate": 1.518987341772152e-06,80 "loss": 7.5703,81 "step": 1282 },83 {84 "epoch": 0.02,85 "learning_rate": 1.6455696202531647e-06,86 "loss": 7.5156,87 "step": 1388 },89 {90 "epoch": 0.03,91 "learning_rate": 1.7721518987341774e-06,92 "loss": 7.3906,93 "step": 1494 },95 {96 "epoch": 0.03,97 "learning_rate": 1.8987341772151901e-06,98 "loss": 7.2109,99 "step": 15100 },101 {102 "epoch": 0.03,103 "learning_rate": 2.0253164556962026e-06,104 "loss": 7.0156,105 "step": 16106 },107 {108 "epoch": 0.03,109 "learning_rate": 2.1518987341772153e-06,110 "loss": 6.7969,111 "step": 17112 },113 {114 "epoch": 0.03,115 "learning_rate": 2.278481012658228e-06,116 "loss": 6.5703,117 "step": 18118 },119 {120 "epoch": 0.04,121 "learning_rate": 2.4050632911392408e-06,122 "loss": 6.3125,123 "step": 19124 },125 {126 "epoch": 0.04,127 "learning_rate": 2.5316455696202535e-06,128 "loss": 6.1016,129 "step": 20130 },131 {132 "epoch": 0.04,133 "learning_rate": 2.6582278481012658e-06,134 "loss": 6.0234,135 "step": 21136 },137 {138 "epoch": 0.04,139 "learning_rate": 2.7848101265822785e-06,140 "loss": 5.9922,141 "step": 22142 },143 {144 "epoch": 0.04,145 "learning_rate": 2.9113924050632912e-06,146 "loss": 5.6953,147 "step": 23148 },149 {150 "epoch": 0.05,151 "learning_rate": 3.037974683544304e-06,152 "loss": 5.5781,153 "step": 24154 },155 {156 "epoch": 0.05,157 "learning_rate": 3.164556962025317e-06,158 "loss": 5.5156,159 "step": 25160 },161 {162 "epoch": 0.05,163 "learning_rate": 3.2911392405063294e-06,164 "loss": 5.3125,165 "step": 26166 },167 {168 "epoch": 0.05,169 "learning_rate": 3.417721518987342e-06,170 "loss": 5.1953,171 "step": 27172 },173 {174 "epoch": 0.05,175 "learning_rate": 3.544303797468355e-06,176 "loss": 5.0234,177 "step": 28178 },179 {180 "epoch": 0.06,181 "learning_rate": 3.6708860759493675e-06,182 "loss": 4.9766,183 "step": 29184 },185 {186 "epoch": 0.06,187 "learning_rate": 3.7974683544303802e-06,188 "loss": 4.8672,189 "step": 30190 },191 {192 "epoch": 0.06,193 "learning_rate": 3.924050632911393e-06,194 "loss": 4.8672,195 "step": 31196 },197 {198 "epoch": 0.06,199 "learning_rate": 4.050632911392405e-06,200 "loss": 4.6406,201 "step": 32202 },203 {204 "epoch": 0.06,205 "learning_rate": 4.177215189873418e-06,206 "loss": 4.5859,207 "step": 33208 },209 {210 "epoch": 0.06,211 "learning_rate": 4.303797468354431e-06,212 "loss": 4.7031,213 "step": 34214 },215 {216 "epoch": 0.07,217 "learning_rate": 4.430379746835443e-06,218 "loss": 4.3203,219 "step": 35220 },221 {222 "epoch": 0.07,223 "learning_rate": 4.556962025316456e-06,224 "loss": 4.5703,225 "step": 36226 },227 {228 "epoch": 0.07,229 "learning_rate": 4.683544303797468e-06,230 "loss": 4.6016,231 "step": 37232 },233 {234 "epoch": 0.07,235 "learning_rate": 4.8101265822784815e-06,236 "loss": 4.5391,237 "step": 38238 },239 {240 "epoch": 0.07,241 "learning_rate": 4.936708860759495e-06,242 "loss": 4.3672,243 "step": 39244 },245 {246 "epoch": 0.08,247 "learning_rate": 5.063291139240507e-06,248 "loss": 4.4297,249 "step": 40250 },251 {252 "epoch": 0.08,253 "learning_rate": 5.189873417721519e-06,254 "loss": 4.3594,255 "step": 41256 },257 {258 "epoch": 0.08,259 "learning_rate": 5.3164556962025316e-06,260 "loss": 4.1016,261 "step": 42262 },263 {264 "epoch": 0.08,265 "learning_rate": 5.443037974683545e-06,266 "loss": 4.2266,267 "step": 43268 },269 {270 "epoch": 0.08,271 "learning_rate": 5.569620253164557e-06,272 "loss": 4.1484,273 "step": 44274 },275 {276 "epoch": 0.09,277 "learning_rate": 5.69620253164557e-06,278 "loss": 4.3398,279 "step": 45280 },281 {282 "epoch": 0.09,283 "learning_rate": 5.8227848101265824e-06,284 "loss": 4.2344,285 "step": 46286 },287 {288 "epoch": 0.09,289 "learning_rate": 5.949367088607595e-06,290 "loss": 4.2539,291 "step": 47292 },293 {294 "epoch": 0.09,295 "learning_rate": 6.075949367088608e-06,296 "loss": 4.1641,297 "step": 48298 },299 {300 "epoch": 0.09,301 "learning_rate": 6.20253164556962e-06,302 "loss": 4.3359,303 "step": 49304 },305 {306 "epoch": 0.1,307 "learning_rate": 6.329113924050634e-06,308 "loss": 4.1172,309 "step": 50310 },311 {312 "epoch": 0.1,313 "learning_rate": 6.4556962025316464e-06,314 "loss": 4.3125,315 "step": 51316 },317 {318 "epoch": 0.1,319 "learning_rate": 6.582278481012659e-06,320 "loss": 4.1016,321 "step": 52322 },323 {324 "epoch": 0.1,325 "learning_rate": 6.708860759493672e-06,326 "loss": 4.3906,327 "step": 53328 },329 {330 "epoch": 0.1,331 "learning_rate": 6.835443037974684e-06,332 "loss": 4.082,333 "step": 54334 },335 {336 "epoch": 0.1,337 "learning_rate": 6.962025316455697e-06,338 "loss": 4.2188,339 "step": 55340 },341 {342 "epoch": 0.11,343 "learning_rate": 7.08860759493671e-06,344 "loss": 4.1953,345 "step": 56346 },347 {348 "epoch": 0.11,349 "learning_rate": 7.215189873417722e-06,350 "loss": 4.1719,351 "step": 57352 },353 {354 "epoch": 0.11,355 "learning_rate": 7.341772151898735e-06,356 "loss": 4.1953,357 "step": 58358 },359 {360 "epoch": 0.11,361 "learning_rate": 7.468354430379747e-06,362 "loss": 4.3477,363 "step": 59364 },365 {366 "epoch": 0.11,367 "learning_rate": 7.5949367088607605e-06,368 "loss": 4.1328,369 "step": 60370 },371 {372 "epoch": 0.12,373 "learning_rate": 7.721518987341773e-06,374 "loss": 4.1406,375 "step": 61376 },377 {378 "epoch": 0.12,379 "learning_rate": 7.848101265822786e-06,380 "loss": 4.1094,381 "step": 62382 },383 {384 "epoch": 0.12,385 "learning_rate": 7.974683544303799e-06,386 "loss": 4.4297,387 "step": 63388 },389 {390 "epoch": 0.12,391 "learning_rate": 8.10126582278481e-06,392 "loss": 4.1875,393 "step": 64394 },395 {396 "epoch": 0.12,397 "learning_rate": 8.227848101265824e-06,398 "loss": 4.3438,399 "step": 65400 },401 {402 "epoch": 0.13,403 "learning_rate": 8.354430379746837e-06,404 "loss": 4.25,405 "step": 66406 },407 {408 "epoch": 0.13,409 "learning_rate": 8.481012658227848e-06,410 "loss": 4.0586,411 "step": 67412 },413 {414 "epoch": 0.13,415 "learning_rate": 8.607594936708861e-06,416 "loss": 4.1797,417 "step": 68418 },419 {420 "epoch": 0.13,421 "learning_rate": 8.734177215189874e-06,422 "loss": 4.1953,423 "step": 69424 },425 {426 "epoch": 0.13,427 "learning_rate": 8.860759493670886e-06,428 "loss": 3.9336,429 "step": 70430 },431 {432 "epoch": 0.14,433 "learning_rate": 8.987341772151899e-06,434 "loss": 4.1953,435 "step": 71436 },437 {438 "epoch": 0.14,439 "learning_rate": 9.113924050632912e-06,440 "loss": 4.1797,441 "step": 72442 },443 {444 "epoch": 0.14,445 "learning_rate": 9.240506329113925e-06,446 "loss": 4.0938,447 "step": 73448 },449 {450 "epoch": 0.14,451 "learning_rate": 9.367088607594937e-06,452 "loss": 4.2109,453 "step": 74454 },455 {456 "epoch": 0.14,457 "learning_rate": 9.49367088607595e-06,458 "loss": 3.9883,459 "step": 75460 },461 {462 "epoch": 0.15,463 "learning_rate": 9.620253164556963e-06,464 "loss": 4.1094,465 "step": 76466 },467 {468 "epoch": 0.15,469 "learning_rate": 9.746835443037975e-06,470 "loss": 4.2383,471 "step": 77472 },473 {474 "epoch": 0.15,475 "learning_rate": 9.87341772151899e-06,476 "loss": 4.1875,477 "step": 78478 },479 {480 "epoch": 0.15,481 "learning_rate": 1e-05,482 "loss": 4.3672,483 "step": 79484 },485 {486 "epoch": 0.15,487 "learning_rate": 9.999996178530757e-06,488 "loss": 4.1172,489 "step": 80490 },491 {492 "epoch": 0.15,493 "learning_rate": 9.999984714128867e-06,494 "loss": 4.2344,495 "step": 81496 },497 {498 "epoch": 0.16,499 "learning_rate": 9.999965606811854e-06,500 "loss": 4.1367,501 "step": 82502 },503 {504 "epoch": 0.16,505 "learning_rate": 9.999938856608927e-06,506 "loss": 3.918,507 "step": 83508 },509 {510 "epoch": 0.16,511 "learning_rate": 9.999904463560974e-06,512 "loss": 4.2266,513 "step": 84514 },515 {516 "epoch": 0.16,517 "learning_rate": 9.99986242772057e-06,518 "loss": 4.2109,519 "step": 85520 },521 {522 "epoch": 0.16,523 "learning_rate": 9.999812749151968e-06,524 "loss": 3.8516,525 "step": 86526 },527 {528 "epoch": 0.17,529 "learning_rate": 9.999755427931108e-06,530 "loss": 4.25,531 "step": 87532 },533 {534 "epoch": 0.17,535 "learning_rate": 9.99969046414561e-06,536 "loss": 3.8398,537 "step": 88538 },539 {540 "epoch": 0.17,541 "learning_rate": 9.999617857894777e-06,542 "loss": 4.1953,543 "step": 89544 },545 {546 "epoch": 0.17,547 "learning_rate": 9.999537609289592e-06,548 "loss": 4.25,549 "step": 90550 },551 {552 "epoch": 0.17,553 "learning_rate": 9.999449718452724e-06,554 "loss": 3.8867,555 "step": 91556 },557 {558 "epoch": 0.18,559 "learning_rate": 9.999354185518525e-06,560 "loss": 3.9805,561 "step": 92562 },563 {564 "epoch": 0.18,565 "learning_rate": 9.99925101063302e-06,566 "loss": 3.9883,567 "step": 93568 },569 {570 "epoch": 0.18,571 "learning_rate": 9.999140193953923e-06,572 "loss": 4.1562,573 "step": 94574 },575 {576 "epoch": 0.18,577 "learning_rate": 9.999021735650626e-06,578 "loss": 4.0859,579 "step": 95580 },581 {582 "epoch": 0.18,583 "learning_rate": 9.998895635904205e-06,584 "loss": 4.1641,585 "step": 96586 },587 {588 "epoch": 0.19,589 "learning_rate": 9.998761894907414e-06,590 "loss": 3.75,591 "step": 97592 },593 {594 "epoch": 0.19,595 "learning_rate": 9.998620512864686e-06,596 "loss": 4.1406,597 "step": 98598 },599 {600 "epoch": 0.19,601 "learning_rate": 9.998471489992138e-06,602 "loss": 3.8789,603 "step": 99604 },605 {606 "epoch": 0.19,607 "learning_rate": 9.998314826517564e-06,608 "loss": 4.1641,609 "step": 100610 },611 {612 "epoch": 0.19,613 "learning_rate": 9.998150522680437e-06,614 "loss": 3.9688,615 "step": 101616 },617 {618 "epoch": 0.19,619 "learning_rate": 9.997978578731908e-06,620 "loss": 3.8438,621 "step": 102622 },623 {624 "epoch": 0.2,625 "learning_rate": 9.997798994934812e-06,626 "loss": 4.0312,627 "step": 103628 },629 {630 "epoch": 0.2,631 "learning_rate": 9.997611771563661e-06,632 "loss": 3.8672,633 "step": 104634 },635 {636 "epoch": 0.2,637 "learning_rate": 9.997416908904634e-06,638 "loss": 3.8711,639 "step": 105640 },641 {642 "epoch": 0.2,643 "learning_rate": 9.997214407255602e-06,644 "loss": 4.1367,645 "step": 106646 },647 {648 "epoch": 0.2,649 "learning_rate": 9.997004266926105e-06,650 "loss": 4.0469,651 "step": 107652 },653 {654 "epoch": 0.21,655 "learning_rate": 9.99678648823736e-06,656 "loss": 3.9844,657 "step": 108658 },659 {660 "epoch": 0.21,661 "learning_rate": 9.996561071522264e-06,662 "loss": 3.8945,663 "step": 109664 },665 {666 "epoch": 0.21,667 "learning_rate": 9.996328017125381e-06,668 "loss": 4.1719,669 "step": 110670 },671 {672 "epoch": 0.21,673 "learning_rate": 9.996087325402959e-06,674 "loss": 3.7891,675 "step": 111676 },677 {678 "epoch": 0.21,679 "learning_rate": 9.995838996722916e-06,680 "loss": 3.8594,681 "step": 112682 },683 {684 "epoch": 0.22,685 "learning_rate": 9.995583031464841e-06,686 "loss": 3.8125,687 "step": 113688 },689 {690 "epoch": 0.22,691 "learning_rate": 9.995319430020004e-06,692 "loss": 4.1172,693 "step": 114694 },695 {696 "epoch": 0.22,697 "learning_rate": 9.995048192791341e-06,698 "loss": 4.0195,699 "step": 115700 },701 {702 "epoch": 0.22,703 "learning_rate": 9.99476932019346e-06,704 "loss": 4.0664,705 "step": 116706 },707 {708 "epoch": 0.22,709 "learning_rate": 9.994482812652645e-06,710 "loss": 4.0938,711 "step": 117712 },713 {714 "epoch": 0.23,715 "learning_rate": 9.994188670606845e-06,716 "loss": 4.0469,717 "step": 118718 },719 {720 "epoch": 0.23,721 "learning_rate": 9.993886894505687e-06,722 "loss": 3.8047,723 "step": 119724 },725 {726 "epoch": 0.23,727 "learning_rate": 9.993577484810457e-06,728 "loss": 4.0625,729 "step": 120730 },731 {732 "epoch": 0.23,733 "learning_rate": 9.993260441994116e-06,734 "loss": 4.0039,735 "step": 121736 },737 {738 "epoch": 0.23,739 "learning_rate": 9.992935766541295e-06,740 "loss": 3.7656,741 "step": 122742 },743 {744 "epoch": 0.23,745 "learning_rate": 9.992603458948282e-06,746 "loss": 3.9766,747 "step": 123748 },749 {750 "epoch": 0.24,751 "learning_rate": 9.992263519723046e-06,752 "loss": 3.7422,753 "step": 124754 },755 {756 "epoch": 0.24,757 "learning_rate": 9.99191594938521e-06,758 "loss": 4.1953,759 "step": 125760 },761 {762 "epoch": 0.24,763 "learning_rate": 9.991560748466068e-06,764 "loss": 4.0039,765 "step": 126766 },767 {768 "epoch": 0.24,769 "learning_rate": 9.99119791750857e-06,770 "loss": 3.9453,771 "step": 127772 },773 {774 "epoch": 0.24,775 "learning_rate": 9.990827457067342e-06,776 "loss": 3.875,777 "step": 128778 },779 {780 "epoch": 0.25,781 "learning_rate": 9.990449367708662e-06,782 "loss": 4.1016,783 "step": 129784 },785 {786 "epoch": 0.25,787 "learning_rate": 9.990063650010473e-06,788 "loss": 3.9727,789 "step": 130790 },791 {792 "epoch": 0.25,793 "learning_rate": 9.989670304562378e-06,794 "loss": 3.9922,795 "step": 131796 },797 {798 "epoch": 0.25,799 "learning_rate": 9.98926933196564e-06,800 "loss": 3.7188,801 "step": 132802 },803 {804 "epoch": 0.25,805 "learning_rate": 9.988860732833183e-06,806 "loss": 4.375,807 "step": 133808 },809 {810 "epoch": 0.26,811 "learning_rate": 9.988444507789584e-06,812 "loss": 3.6367,813 "step": 134814 },815 {816 "epoch": 0.26,817 "learning_rate": 9.988020657471078e-06,818 "loss": 3.8633,819 "step": 135820 },821 {822 "epoch": 0.26,823 "learning_rate": 9.987589182525562e-06,824 "loss": 3.8555,825 "step": 136826 },827 {828 "epoch": 0.26,829 "learning_rate": 9.98715008361258e-06,830 "loss": 3.9688,831 "step": 137832 },833 {834 "epoch": 0.26,835 "learning_rate": 9.986703361403334e-06,836 "loss": 3.8008,837 "step": 138838 },839 {840 "epoch": 0.27,841 "learning_rate": 9.98624901658068e-06,842 "loss": 3.9062,843 "step": 139844 },845 {846 "epoch": 0.27,847 "learning_rate": 9.98578704983912e-06,848 "loss": 3.7344,849 "step": 140850 },851 {852 "epoch": 0.27,853 "learning_rate": 9.985317461884815e-06,854 "loss": 3.7617,855 "step": 141856 },857 {858 "epoch": 0.27,859 "learning_rate": 9.984840253435569e-06,860 "loss": 3.9844,861 "step": 142862 },863 {864 "epoch": 0.27,865 "learning_rate": 9.984355425220835e-06,866 "loss": 4.0859,867 "step": 143868 },869 {870 "epoch": 0.27,871 "learning_rate": 9.983862977981717e-06,872 "loss": 3.75,873 "step": 144874 },875 {876 "epoch": 0.28,877 "learning_rate": 9.983362912470967e-06,878 "loss": 3.9492,879 "step": 145880 },881 {882 "epoch": 0.28,883 "learning_rate": 9.982855229452976e-06,884 "loss": 3.9375,885 "step": 146886 },887 {888 "epoch": 0.28,889 "learning_rate": 9.982339929703782e-06,890 "loss": 3.8516,891 "step": 147892 },893 {894 "epoch": 0.28,895 "learning_rate": 9.981817014011066e-06,896 "loss": 3.8477,897 "step": 148898 },899 {900 "epoch": 0.28,901 "learning_rate": 9.98128648317415e-06,902 "loss": 3.8516,903 "step": 149904 },905 {906 "epoch": 0.29,907 "learning_rate": 9.980748338003999e-06,908 "loss": 3.8008,909 "step": 150910 },911 {912 "epoch": 0.29,913 "learning_rate": 9.980202579323212e-06,914 "loss": 3.8867,915 "step": 151916 },917 {918 "epoch": 0.29,919 "learning_rate": 9.97964920796603e-06,920 "loss": 3.918,921 "step": 152922 },923 {924 "epoch": 0.29,925 "learning_rate": 9.979088224778336e-06,926 "loss": 3.5586,927 "step": 153928 },929 {930 "epoch": 0.29,931 "learning_rate": 9.978519630617632e-06,932 "loss": 3.9609,933 "step": 154934 },935 {936 "epoch": 0.3,937 "learning_rate": 9.977943426353067e-06,938 "loss": 3.5508,939 "step": 155940 },941 {942 "epoch": 0.3,943 "learning_rate": 9.977359612865424e-06,944 "loss": 3.8711,945 "step": 156946 },947 {948 "epoch": 0.3,949 "learning_rate": 9.976768191047109e-06,950 "loss": 3.7422,951 "step": 157952 },953 {954 "epoch": 0.3,955 "learning_rate": 9.976169161802164e-06,956 "loss": 3.8477,957 "step": 158958 },959 {960 "epoch": 0.3,961 "learning_rate": 9.975562526046257e-06,962 "loss": 3.582,963 "step": 159964 },965 {966 "epoch": 0.31,967 "learning_rate": 9.974948284706684e-06,968 "loss": 3.8164,969 "step": 160970 },971 {972 "epoch": 0.31,973 "learning_rate": 9.974326438722367e-06,974 "loss": 3.7188,975 "step": 161976 },977 {978 "epoch": 0.31,979 "learning_rate": 9.973696989043852e-06,980 "loss": 4.1992,981 "step": 162982 },983 {984 "epoch": 0.31,985 "learning_rate": 9.973059936633308e-06,986 "loss": 3.8047,987 "step": 163988 },989 {990 "epoch": 0.31,991 "learning_rate": 9.972415282464524e-06,992 "loss": 3.7305,993 "step": 164994 },995 {996 "epoch": 0.31,997 "learning_rate": 9.971763027522913e-06,998 "loss": 3.8906,999 "step": 1651000 },1001 {1002 "epoch": 0.32,1003 "learning_rate": 9.971103172805504e-06,1004 "loss": 3.5859,1005 "step": 1661006 },1007 {1008 "epoch": 0.32,1009 "learning_rate": 9.97043571932094e-06,1010 "loss": 3.9258,1011 "step": 1671012 },1013 {1014 "epoch": 0.32,1015 "learning_rate": 9.969760668089485e-06,1016 "loss": 3.8047,1017 "step": 1681018 },1019 {1020 "epoch": 0.32,1021 "learning_rate": 9.969078020143013e-06,1022 "loss": 3.9531,1023 "step": 1691024 },1025 {1026 "epoch": 0.32,1027 "learning_rate": 9.968387776525009e-06,1028 "loss": 4.0352,1029 "step": 1701030 },1031 {1032 "epoch": 0.33,1033 "learning_rate": 9.967689938290574e-06,1034 "loss": 3.7422,1035 "step": 1711036 },1037 {1038 "epoch": 0.33,1039 "learning_rate": 9.966984506506415e-06,1040 "loss": 3.8125,1041 "step": 1721042 },1043 {1044 "epoch": 0.33,1045 "learning_rate": 9.966271482250845e-06,1046 "loss": 3.668,1047 "step": 1731048 },1049 {1050 "epoch": 0.33,1051 "learning_rate": 9.965550866613782e-06,1052 "loss": 3.7031,1053 "step": 1741054 },1055 {1056 "epoch": 0.33,1057 "learning_rate": 9.964822660696753e-06,1058 "loss": 3.668,1059 "step": 1751060 },1061 {1062 "epoch": 0.34,1063 "learning_rate": 9.964086865612886e-06,1064 "loss": 3.9883,1065 "step": 1761066 },1067 {1068 "epoch": 0.34,1069 "learning_rate": 9.963343482486907e-06,1070 "loss": 3.5977,1071 "step": 1771072 },1073 {1074 "epoch": 0.34,1075 "learning_rate": 9.96259251245514e-06,1076 "loss": 3.7891,1077 "step": 1781078 },1079 {1080 "epoch": 0.34,1081 "learning_rate": 9.96183395666551e-06,1082 "loss": 3.8945,1083 "step": 1791084 },1085 {1086 "epoch": 0.34,1087 "learning_rate": 9.961067816277537e-06,1088 "loss": 3.6172,1089 "step": 1801090 },1091 {1092 "epoch": 0.35,1093 "learning_rate": 9.960294092462332e-06,1094 "loss": 3.4766,1095 "step": 1811096 },1097 {1098 "epoch": 0.35,1099 "learning_rate": 9.959512786402603e-06,1100 "loss": 3.7305,1101 "step": 1821102 },1103 {1104 "epoch": 0.35,1105 "learning_rate": 9.958723899292642e-06,1106 "loss": 3.7852,1107 "step": 1831108 },1109 {1110 "epoch": 0.35,1111 "learning_rate": 9.957927432338332e-06,1112 "loss": 3.7188,1113 "step": 1841114 },1115 {1116 "epoch": 0.35,1117 "learning_rate": 9.957123386757144e-06,1118 "loss": 3.7031,1119 "step": 1851120 },1121 {1122 "epoch": 0.35,1123 "learning_rate": 9.956311763778131e-06,1124 "loss": 3.5898,1125 "step": 1861126 },1127 {1128 "epoch": 0.36,1129 "learning_rate": 9.95549256464193e-06,1130 "loss": 3.8477,1131 "step": 1871132 },1133 {1134 "epoch": 0.36,1135 "learning_rate": 9.954665790600762e-06,1136 "loss": 3.4219,1137 "step": 1881138 },1139 {1140 "epoch": 0.36,1141 "learning_rate": 9.953831442918418e-06,1142 "loss": 3.7852,1143 "step": 1891144 },1145 {1146 "epoch": 0.36,1147 "learning_rate": 9.952989522870275e-06,1148 "loss": 3.7305,1149 "step": 1901150 },1151 {1152 "epoch": 0.36,1153 "learning_rate": 9.952140031743282e-06,1154 "loss": 3.75,1155 "step": 1911156 },1157 {1158 "epoch": 0.37,1159 "learning_rate": 9.951282970835959e-06,1160 "loss": 3.6523,1161 "step": 1921162 },1163 {1164 "epoch": 0.37,1165 "learning_rate": 9.950418341458398e-06,1166 "loss": 3.5859,1167 "step": 1931168 },1169 {1170 "epoch": 0.37,1171 "learning_rate": 9.949546144932265e-06,1172 "loss": 3.707,1173 "step": 1941174 },1175 {1176 "epoch": 0.37,1177 "learning_rate": 9.948666382590785e-06,1178 "loss": 3.6094,1179 "step": 1951180 },1181 {1182 "epoch": 0.37,1183 "learning_rate": 9.947779055778752e-06,1184 "loss": 3.7227,1185 "step": 1961186 },1187 {1188 "epoch": 0.38,1189 "learning_rate": 9.946884165852526e-06,1190 "loss": 3.7617,1191 "step": 1971192 },1193 {1194 "epoch": 0.38,1195 "learning_rate": 9.945981714180021e-06,1196 "loss": 3.6797,1197 "step": 1981198 },1199 {1200 "epoch": 0.38,