SushantGautam/CodeGeneration
0100
1{2 "best_metric": null,3 "best_model_checkpoint": null,4 "epoch": 10.0,5 "global_step": 4390,6 "is_hyper_param_search": false,7 "is_local_process_zero": true,8 "is_world_process_zero": true,9 "log_history": [10 {11 "epoch": 0.0,12 "learning_rate": 4.998861047835991e-05,13 "loss": 11.7092,14 "step": 115 },16 {17 "epoch": 0.0,18 "learning_rate": 4.9977220956719826e-05,19 "loss": 9.6548,20 "step": 221 },22 {23 "epoch": 0.01,24 "learning_rate": 4.996583143507973e-05,25 "loss": 8.4051,26 "step": 327 },28 {29 "epoch": 0.01,30 "learning_rate": 4.9954441913439635e-05,31 "loss": 9.5686,32 "step": 433 },34 {35 "epoch": 0.01,36 "learning_rate": 4.9943052391799543e-05,37 "loss": 8.9303,38 "step": 539 },40 {41 "epoch": 0.01,42 "learning_rate": 4.993166287015946e-05,43 "loss": 7.7487,44 "step": 645 },46 {47 "epoch": 0.02,48 "learning_rate": 4.992027334851937e-05,49 "loss": 6.5083,50 "step": 751 },52 {53 "epoch": 0.02,54 "learning_rate": 4.990888382687927e-05,55 "loss": 8.1923,56 "step": 857 },58 {59 "epoch": 0.02,60 "learning_rate": 4.989749430523918e-05,61 "loss": 8.0454,62 "step": 963 },64 {65 "epoch": 0.02,66 "learning_rate": 4.988610478359909e-05,67 "loss": 7.6701,68 "step": 1069 },70 {71 "epoch": 0.03,72 "learning_rate": 4.9874715261959e-05,73 "loss": 7.1795,74 "step": 1175 },76 {77 "epoch": 0.03,78 "learning_rate": 4.9863325740318914e-05,79 "loss": 7.6715,80 "step": 1281 },82 {83 "epoch": 0.03,84 "learning_rate": 4.9851936218678816e-05,85 "loss": 7.9991,86 "step": 1387 },88 {89 "epoch": 0.03,90 "learning_rate": 4.9840546697038724e-05,91 "loss": 6.5221,92 "step": 1493 },94 {95 "epoch": 0.03,96 "learning_rate": 4.982915717539863e-05,97 "loss": 7.3411,98 "step": 1599 },100 {101 "epoch": 0.04,102 "learning_rate": 4.981776765375855e-05,103 "loss": 5.8229,104 "step": 16105 },106 {107 "epoch": 0.04,108 "learning_rate": 4.9806378132118455e-05,109 "loss": 6.8379,110 "step": 17111 },112 {113 "epoch": 0.04,114 "learning_rate": 4.9794988610478364e-05,115 "loss": 7.0664,116 "step": 18117 },118 {119 "epoch": 0.04,120 "learning_rate": 4.978359908883827e-05,121 "loss": 7.0716,122 "step": 19123 },124 {125 "epoch": 0.05,126 "learning_rate": 4.977220956719818e-05,127 "loss": 8.7517,128 "step": 20129 },130 {131 "epoch": 0.05,132 "learning_rate": 4.976082004555809e-05,133 "loss": 6.6135,134 "step": 21135 },136 {137 "epoch": 0.05,138 "learning_rate": 4.9749430523918e-05,139 "loss": 6.1742,140 "step": 22141 },142 {143 "epoch": 0.05,144 "learning_rate": 4.9738041002277905e-05,145 "loss": 8.0588,146 "step": 23147 },148 {149 "epoch": 0.05,150 "learning_rate": 4.972665148063781e-05,151 "loss": 7.9152,152 "step": 24153 },154 {155 "epoch": 0.06,156 "learning_rate": 4.971526195899772e-05,157 "loss": 5.1515,158 "step": 25159 },160 {161 "epoch": 0.06,162 "learning_rate": 4.9703872437357636e-05,163 "loss": 5.6302,164 "step": 26165 },166 {167 "epoch": 0.06,168 "learning_rate": 4.9692482915717544e-05,169 "loss": 9.6288,170 "step": 27171 },172 {173 "epoch": 0.06,174 "learning_rate": 4.968109339407745e-05,175 "loss": 5.228,176 "step": 28177 },178 {179 "epoch": 0.07,180 "learning_rate": 4.966970387243736e-05,181 "loss": 6.7267,182 "step": 29183 },184 {185 "epoch": 0.07,186 "learning_rate": 4.965831435079727e-05,187 "loss": 5.9025,188 "step": 30189 },190 {191 "epoch": 0.07,192 "learning_rate": 4.964692482915718e-05,193 "loss": 6.6566,194 "step": 31195 },196 {197 "epoch": 0.07,198 "learning_rate": 4.963553530751709e-05,199 "loss": 6.449,200 "step": 32201 },202 {203 "epoch": 0.08,204 "learning_rate": 4.9624145785876994e-05,205 "loss": 5.6756,206 "step": 33207 },208 {209 "epoch": 0.08,210 "learning_rate": 4.96127562642369e-05,211 "loss": 6.2954,212 "step": 34213 },214 {215 "epoch": 0.08,216 "learning_rate": 4.960136674259681e-05,217 "loss": 6.7187,218 "step": 35219 },220 {221 "epoch": 0.08,222 "learning_rate": 4.9589977220956725e-05,223 "loss": 7.1454,224 "step": 36225 },226 {227 "epoch": 0.08,228 "learning_rate": 4.957858769931663e-05,229 "loss": 5.7555,230 "step": 37231 },232 {233 "epoch": 0.09,234 "learning_rate": 4.956719817767654e-05,235 "loss": 6.1015,236 "step": 38237 },238 {239 "epoch": 0.09,240 "learning_rate": 4.955580865603645e-05,241 "loss": 6.5185,242 "step": 39243 },244 {245 "epoch": 0.09,246 "learning_rate": 4.954441913439636e-05,247 "loss": 4.8291,248 "step": 40249 },250 {251 "epoch": 0.09,252 "learning_rate": 4.9533029612756266e-05,253 "loss": 5.2706,254 "step": 41255 },256 {257 "epoch": 0.1,258 "learning_rate": 4.952164009111618e-05,259 "loss": 7.173,260 "step": 42261 },262 {263 "epoch": 0.1,264 "learning_rate": 4.951025056947608e-05,265 "loss": 8.1812,266 "step": 43267 },268 {269 "epoch": 0.1,270 "learning_rate": 4.949886104783599e-05,271 "loss": 7.6324,272 "step": 44273 },274 {275 "epoch": 0.1,276 "learning_rate": 4.94874715261959e-05,277 "loss": 7.6471,278 "step": 45279 },280 {281 "epoch": 0.1,282 "learning_rate": 4.9476082004555814e-05,283 "loss": 6.6633,284 "step": 46285 },286 {287 "epoch": 0.11,288 "learning_rate": 4.946469248291572e-05,289 "loss": 6.1763,290 "step": 47291 },292 {293 "epoch": 0.11,294 "learning_rate": 4.945330296127563e-05,295 "loss": 4.8749,296 "step": 48297 },298 {299 "epoch": 0.11,300 "learning_rate": 4.944191343963554e-05,301 "loss": 7.3018,302 "step": 49303 },304 {305 "epoch": 0.11,306 "learning_rate": 4.9430523917995447e-05,307 "loss": 6.4854,308 "step": 50309 },310 {311 "epoch": 0.12,312 "learning_rate": 4.9419134396355355e-05,313 "loss": 5.4195,314 "step": 51315 },316 {317 "epoch": 0.12,318 "learning_rate": 4.940774487471526e-05,319 "loss": 6.822,320 "step": 52321 },322 {323 "epoch": 0.12,324 "learning_rate": 4.939635535307517e-05,325 "loss": 7.2982,326 "step": 53327 },328 {329 "epoch": 0.12,330 "learning_rate": 4.938496583143508e-05,331 "loss": 6.6605,332 "step": 54333 },334 {335 "epoch": 0.13,336 "learning_rate": 4.937357630979499e-05,337 "loss": 4.8833,338 "step": 55339 },340 {341 "epoch": 0.13,342 "learning_rate": 4.93621867881549e-05,343 "loss": 6.4155,344 "step": 56345 },346 {347 "epoch": 0.13,348 "learning_rate": 4.935079726651481e-05,349 "loss": 4.1882,350 "step": 57351 },352 {353 "epoch": 0.13,354 "learning_rate": 4.933940774487472e-05,355 "loss": 6.7254,356 "step": 58357 },358 {359 "epoch": 0.13,360 "learning_rate": 4.932801822323463e-05,361 "loss": 6.5771,362 "step": 59363 },364 {365 "epoch": 0.14,366 "learning_rate": 4.9316628701594535e-05,367 "loss": 5.0929,368 "step": 60369 },370 {371 "epoch": 0.14,372 "learning_rate": 4.9305239179954444e-05,373 "loss": 8.5807,374 "step": 61375 },376 {377 "epoch": 0.14,378 "learning_rate": 4.929384965831435e-05,379 "loss": 4.8718,380 "step": 62381 },382 {383 "epoch": 0.14,384 "learning_rate": 4.928246013667426e-05,385 "loss": 6.4467,386 "step": 63387 },388 {389 "epoch": 0.15,390 "learning_rate": 4.927107061503417e-05,391 "loss": 5.6606,392 "step": 64393 },394 {395 "epoch": 0.15,396 "learning_rate": 4.9259681093394076e-05,397 "loss": 6.4872,398 "step": 65399 },400 {401 "epoch": 0.15,402 "learning_rate": 4.924829157175399e-05,403 "loss": 8.0231,404 "step": 66405 },406 {407 "epoch": 0.15,408 "learning_rate": 4.92369020501139e-05,409 "loss": 6.0967,410 "step": 67411 },412 {413 "epoch": 0.15,414 "learning_rate": 4.922551252847381e-05,415 "loss": 6.8669,416 "step": 68417 },418 {419 "epoch": 0.16,420 "learning_rate": 4.9214123006833716e-05,421 "loss": 6.5913,422 "step": 69423 },424 {425 "epoch": 0.16,426 "learning_rate": 4.9202733485193624e-05,427 "loss": 5.0972,428 "step": 70429 },430 {431 "epoch": 0.16,432 "learning_rate": 4.919134396355353e-05,433 "loss": 5.1875,434 "step": 71435 },436 {437 "epoch": 0.16,438 "learning_rate": 4.917995444191344e-05,439 "loss": 5.143,440 "step": 72441 },442 {443 "epoch": 0.17,444 "learning_rate": 4.9168564920273356e-05,445 "loss": 4.976,446 "step": 73447 },448 {449 "epoch": 0.17,450 "learning_rate": 4.915717539863326e-05,451 "loss": 6.2076,452 "step": 74453 },454 {455 "epoch": 0.17,456 "learning_rate": 4.9145785876993165e-05,457 "loss": 6.3557,458 "step": 75459 },460 {461 "epoch": 0.17,462 "learning_rate": 4.913439635535308e-05,463 "loss": 3.9665,464 "step": 76465 },466 {467 "epoch": 0.18,468 "learning_rate": 4.912300683371299e-05,469 "loss": 6.9845,470 "step": 77471 },472 {473 "epoch": 0.18,474 "learning_rate": 4.91116173120729e-05,475 "loss": 5.5932,476 "step": 78477 },478 {479 "epoch": 0.18,480 "learning_rate": 4.9100227790432805e-05,481 "loss": 8.17,482 "step": 79483 },484 {485 "epoch": 0.18,486 "learning_rate": 4.908883826879271e-05,487 "loss": 6.9583,488 "step": 80489 },490 {491 "epoch": 0.18,492 "learning_rate": 4.907744874715262e-05,493 "loss": 5.9453,494 "step": 81495 },496 {497 "epoch": 0.19,498 "learning_rate": 4.906605922551253e-05,499 "loss": 6.1392,500 "step": 82501 },502 {503 "epoch": 0.19,504 "learning_rate": 4.9054669703872444e-05,505 "loss": 5.7155,506 "step": 83507 },508 {509 "epoch": 0.19,510 "learning_rate": 4.9043280182232346e-05,511 "loss": 6.3088,512 "step": 84513 },514 {515 "epoch": 0.19,516 "learning_rate": 4.9031890660592254e-05,517 "loss": 7.3777,518 "step": 85519 },520 {521 "epoch": 0.2,522 "learning_rate": 4.902050113895217e-05,523 "loss": 6.2157,524 "step": 86525 },526 {527 "epoch": 0.2,528 "learning_rate": 4.900911161731208e-05,529 "loss": 5.0033,530 "step": 87531 },532 {533 "epoch": 0.2,534 "learning_rate": 4.8997722095671985e-05,535 "loss": 4.8096,536 "step": 88537 },538 {539 "epoch": 0.2,540 "learning_rate": 4.8986332574031894e-05,541 "loss": 5.0981,542 "step": 89543 },544 {545 "epoch": 0.21,546 "learning_rate": 4.89749430523918e-05,547 "loss": 7.8127,548 "step": 90549 },550 {551 "epoch": 0.21,552 "learning_rate": 4.896355353075171e-05,553 "loss": 5.5104,554 "step": 91555 },556 {557 "epoch": 0.21,558 "learning_rate": 4.895216400911162e-05,559 "loss": 4.2981,560 "step": 92561 },562 {563 "epoch": 0.21,564 "learning_rate": 4.894077448747153e-05,565 "loss": 5.3174,566 "step": 93567 },568 {569 "epoch": 0.21,570 "learning_rate": 4.8929384965831435e-05,571 "loss": 5.5535,572 "step": 94573 },574 {575 "epoch": 0.22,576 "learning_rate": 4.891799544419134e-05,577 "loss": 4.1204,578 "step": 95579 },580 {581 "epoch": 0.22,582 "learning_rate": 4.890660592255126e-05,583 "loss": 5.5789,584 "step": 96585 },586 {587 "epoch": 0.22,588 "learning_rate": 4.8895216400911166e-05,589 "loss": 6.7247,590 "step": 97591 },592 {593 "epoch": 0.22,594 "learning_rate": 4.8883826879271074e-05,595 "loss": 7.0594,596 "step": 98597 },598 {599 "epoch": 0.23,600 "learning_rate": 4.8872437357630976e-05,601 "loss": 5.1639,602 "step": 99603 },604 {605 "epoch": 0.23,606 "learning_rate": 4.886104783599089e-05,607 "loss": 6.2181,608 "step": 100609 },610 {611 "epoch": 0.23,612 "learning_rate": 4.88496583143508e-05,613 "loss": 4.1885,614 "step": 101615 },616 {617 "epoch": 0.23,618 "learning_rate": 4.883826879271071e-05,619 "loss": 5.3262,620 "step": 102621 },622 {623 "epoch": 0.23,624 "learning_rate": 4.882687927107062e-05,625 "loss": 7.0901,626 "step": 103627 },628 {629 "epoch": 0.24,630 "learning_rate": 4.8815489749430524e-05,631 "loss": 6.3874,632 "step": 104633 },634 {635 "epoch": 0.24,636 "learning_rate": 4.880410022779043e-05,637 "loss": 5.7896,638 "step": 105639 },640 {641 "epoch": 0.24,642 "learning_rate": 4.879271070615035e-05,643 "loss": 5.2289,644 "step": 106645 },646 {647 "epoch": 0.24,648 "learning_rate": 4.8781321184510255e-05,649 "loss": 4.9152,650 "step": 107651 },652 {653 "epoch": 0.25,654 "learning_rate": 4.876993166287016e-05,655 "loss": 4.0898,656 "step": 108657 },658 {659 "epoch": 0.25,660 "learning_rate": 4.875854214123007e-05,661 "loss": 5.399,662 "step": 109663 },664 {665 "epoch": 0.25,666 "learning_rate": 4.874715261958998e-05,667 "loss": 5.256,668 "step": 110669 },670 {671 "epoch": 0.25,672 "learning_rate": 4.873576309794989e-05,673 "loss": 7.3301,674 "step": 111675 },676 {677 "epoch": 0.26,678 "learning_rate": 4.8724373576309796e-05,679 "loss": 5.971,680 "step": 112681 },682 {683 "epoch": 0.26,684 "learning_rate": 4.871298405466971e-05,685 "loss": 5.5653,686 "step": 113687 },688 {689 "epoch": 0.26,690 "learning_rate": 4.870159453302961e-05,691 "loss": 6.7108,692 "step": 114693 },694 {695 "epoch": 0.26,696 "learning_rate": 4.869020501138952e-05,697 "loss": 5.104,698 "step": 115699 },700 {701 "epoch": 0.26,702 "learning_rate": 4.8678815489749436e-05,703 "loss": 3.1726,704 "step": 116705 },706 {707 "epoch": 0.27,708 "learning_rate": 4.8667425968109344e-05,709 "loss": 6.0224,710 "step": 117711 },712 {713 "epoch": 0.27,714 "learning_rate": 4.865603644646925e-05,715 "loss": 6.4886,716 "step": 118717 },718 {719 "epoch": 0.27,720 "learning_rate": 4.864464692482916e-05,721 "loss": 5.3235,722 "step": 119723 },724 {725 "epoch": 0.27,726 "learning_rate": 4.863325740318907e-05,727 "loss": 4.7159,728 "step": 120729 },730 {731 "epoch": 0.28,732 "learning_rate": 4.8621867881548977e-05,733 "loss": 6.0486,734 "step": 121735 },736 {737 "epoch": 0.28,738 "learning_rate": 4.8610478359908885e-05,739 "loss": 4.8432,740 "step": 122741 },742 {743 "epoch": 0.28,744 "learning_rate": 4.85990888382688e-05,745 "loss": 3.8479,746 "step": 123747 },748 {749 "epoch": 0.28,750 "learning_rate": 4.85876993166287e-05,751 "loss": 5.1927,752 "step": 124753 },754 {755 "epoch": 0.28,756 "learning_rate": 4.857630979498861e-05,757 "loss": 4.9575,758 "step": 125759 },760 {761 "epoch": 0.29,762 "learning_rate": 4.8564920273348524e-05,763 "loss": 6.6181,764 "step": 126765 },766 {767 "epoch": 0.29,768 "learning_rate": 4.855353075170843e-05,769 "loss": 3.5296,770 "step": 127771 },772 {773 "epoch": 0.29,774 "learning_rate": 4.854214123006834e-05,775 "loss": 5.9652,776 "step": 128777 },778 {779 "epoch": 0.29,780 "learning_rate": 4.853075170842825e-05,781 "loss": 3.7124,782 "step": 129783 },784 {785 "epoch": 0.3,786 "learning_rate": 4.851936218678816e-05,787 "loss": 4.6301,788 "step": 130789 },790 {791 "epoch": 0.3,792 "learning_rate": 4.8507972665148065e-05,793 "loss": 4.6318,794 "step": 131795 },796 {797 "epoch": 0.3,798 "learning_rate": 4.8496583143507974e-05,799 "loss": 4.9169,800 "step": 132801 },802 {803 "epoch": 0.3,804 "learning_rate": 4.848519362186789e-05,805 "loss": 5.6412,806 "step": 133807 },808 {809 "epoch": 0.31,810 "learning_rate": 4.847380410022779e-05,811 "loss": 5.703,812 "step": 134813 },814 {815 "epoch": 0.31,816 "learning_rate": 4.84624145785877e-05,817 "loss": 5.5157,818 "step": 135819 },820 {821 "epoch": 0.31,822 "learning_rate": 4.845102505694761e-05,823 "loss": 5.302,824 "step": 136825 },826 {827 "epoch": 0.31,828 "learning_rate": 4.843963553530752e-05,829 "loss": 8.4144,830 "step": 137831 },832 {833 "epoch": 0.31,834 "learning_rate": 4.842824601366743e-05,835 "loss": 4.4277,836 "step": 138837 },838 {839 "epoch": 0.32,840 "learning_rate": 4.841685649202734e-05,841 "loss": 3.3021,842 "step": 139843 },844 {845 "epoch": 0.32,846 "learning_rate": 4.8405466970387246e-05,847 "loss": 5.3697,848 "step": 140849 },850 {851 "epoch": 0.32,852 "learning_rate": 4.8394077448747154e-05,853 "loss": 6.1515,854 "step": 141855 },856 {857 "epoch": 0.32,858 "learning_rate": 4.838268792710706e-05,859 "loss": 5.0783,860 "step": 142861 },862 {863 "epoch": 0.33,864 "learning_rate": 4.837129840546698e-05,865 "loss": 5.6476,866 "step": 143867 },868 {869 "epoch": 0.33,870 "learning_rate": 4.835990888382688e-05,871 "loss": 6.0066,872 "step": 144873 },874 {875 "epoch": 0.33,876 "learning_rate": 4.834851936218679e-05,877 "loss": 5.6522,878 "step": 145879 },880 {881 "epoch": 0.33,882 "learning_rate": 4.83371298405467e-05,883 "loss": 6.2684,884 "step": 146885 },886 {887 "epoch": 0.33,888 "learning_rate": 4.832574031890661e-05,889 "loss": 6.2464,890 "step": 147891 },892 {893 "epoch": 0.34,894 "learning_rate": 4.831435079726652e-05,895 "loss": 8.025,896 "step": 148897 },898 {899 "epoch": 0.34,900 "learning_rate": 4.830296127562643e-05,901 "loss": 5.7292,902 "step": 149903 },904 {905 "epoch": 0.34,906 "learning_rate": 4.8291571753986335e-05,907 "loss": 6.6286,908 "step": 150909 },910 {911 "epoch": 0.34,912 "learning_rate": 4.828018223234624e-05,913 "loss": 5.2295,914 "step": 151915 },916 {917 "epoch": 0.35,918 "learning_rate": 4.826879271070615e-05,919 "loss": 3.3221,920 "step": 152921 },922 {923 "epoch": 0.35,924 "learning_rate": 4.8257403189066066e-05,925 "loss": 5.561,926 "step": 153927 },928 {929 "epoch": 0.35,930 "learning_rate": 4.824601366742597e-05,931 "loss": 4.0204,932 "step": 154933 },934 {935 "epoch": 0.35,936 "learning_rate": 4.8234624145785876e-05,937 "loss": 5.223,938 "step": 155939 },940 {941 "epoch": 0.36,942 "learning_rate": 4.8223234624145784e-05,943 "loss": 5.8434,944 "step": 156945 },946 {947 "epoch": 0.36,948 "learning_rate": 4.82118451025057e-05,949 "loss": 4.7687,950 "step": 157951 },952 {953 "epoch": 0.36,954 "learning_rate": 4.820045558086561e-05,955 "loss": 5.6896,956 "step": 158957 },958 {959 "epoch": 0.36,960 "learning_rate": 4.8189066059225516e-05,961 "loss": 4.4732,962 "step": 159963 },964 {965 "epoch": 0.36,966 "learning_rate": 4.8177676537585424e-05,967 "loss": 5.0196,968 "step": 160969 },970 {971 "epoch": 0.37,972 "learning_rate": 4.816628701594533e-05,973 "loss": 5.2394,974 "step": 161975 },976 {977 "epoch": 0.37,978 "learning_rate": 4.815489749430524e-05,979 "loss": 4.4934,980 "step": 162981 },982 {983 "epoch": 0.37,984 "learning_rate": 4.8143507972665155e-05,985 "loss": 3.838,986 "step": 163987 },988 {989 "epoch": 0.37,990 "learning_rate": 4.813211845102506e-05,991 "loss": 4.5055,992 "step": 164993 },994 {995 "epoch": 0.38,996 "learning_rate": 4.8120728929384965e-05,997 "loss": 6.3589,998 "step": 165999 },1000 {1001 "epoch": 0.38,1002 "learning_rate": 4.810933940774487e-05,1003 "loss": 4.5191,1004 "step": 1661005 },1006 {1007 "epoch": 0.38,1008 "learning_rate": 4.809794988610479e-05,1009 "loss": 5.4652,1010 "step": 1671011 },1012 {1013 "epoch": 0.38,1014 "learning_rate": 4.8086560364464696e-05,1015 "loss": 5.786,1016 "step": 1681017 },1018 {1019 "epoch": 0.38,1020 "learning_rate": 4.8075170842824604e-05,1021 "loss": 5.3748,1022 "step": 1691023 },1024 {1025 "epoch": 0.39,1026 "learning_rate": 4.806378132118451e-05,1027 "loss": 5.2441,1028 "step": 1701029 },1030 {1031 "epoch": 0.39,1032 "learning_rate": 4.805239179954442e-05,1033 "loss": 8.0993,1034 "step": 1711035 },1036 {1037 "epoch": 0.39,1038 "learning_rate": 4.804100227790433e-05,1039 "loss": 6.6524,1040 "step": 1721041 },1042 {1043 "epoch": 0.39,1044 "learning_rate": 4.8029612756264244e-05,1045 "loss": 5.0897,1046 "step": 1731047 },1048 {1049 "epoch": 0.4,1050 "learning_rate": 4.801822323462415e-05,1051 "loss": 5.6209,1052 "step": 1741053 },1054 {1055 "epoch": 0.4,1056 "learning_rate": 4.8006833712984054e-05,1057 "loss": 6.0429,1058 "step": 1751059 },1060 {1061 "epoch": 0.4,1062 "learning_rate": 4.799544419134396e-05,1063 "loss": 5.6825,1064 "step": 1761065 },1066 {1067 "epoch": 0.4,1068 "learning_rate": 4.798405466970388e-05,1069 "loss": 4.1911,1070 "step": 1771071 },1072 {1073 "epoch": 0.41,1074 "learning_rate": 4.7972665148063785e-05,1075 "loss": 6.3678,1076 "step": 1781077 },1078 {1079 "epoch": 0.41,1080 "learning_rate": 4.796127562642369e-05,1081 "loss": 5.4064,1082 "step": 1791083 },1084 {1085 "epoch": 0.41,1086 "learning_rate": 4.79498861047836e-05,1087 "loss": 6.096,1088 "step": 1801089 },1090 {1091 "epoch": 0.41,1092 "learning_rate": 4.793849658314351e-05,1093 "loss": 6.7477,1094 "step": 1811095 },1096 {1097 "epoch": 0.41,1098 "learning_rate": 4.792710706150342e-05,1099 "loss": 5.9499,1100 "step": 1821101 },1102 {1103 "epoch": 0.42,1104 "learning_rate": 4.791571753986333e-05,1105 "loss": 5.4382,1106 "step": 1831107 },1108 {1109 "epoch": 0.42,1110 "learning_rate": 4.790432801822324e-05,1111 "loss": 6.2821,1112 "step": 1841113 },1114 {1115 "epoch": 0.42,1116 "learning_rate": 4.789293849658314e-05,1117 "loss": 4.3839,1118 "step": 1851119 },1120 {1121 "epoch": 0.42,1122 "learning_rate": 4.788154897494305e-05,1123 "loss": 6.8907,1124 "step": 1861125 },1126 {1127 "epoch": 0.43,1128 "learning_rate": 4.7870159453302966e-05,1129 "loss": 4.5461,1130 "step": 1871131 },1132 {1133 "epoch": 0.43,1134 "learning_rate": 4.7858769931662874e-05,1135 "loss": 6.7334,1136 "step": 1881137 },1138 {1139 "epoch": 0.43,1140 "learning_rate": 4.784738041002278e-05,1141 "loss": 4.8625,1142 "step": 1891143 },1144 {1145 "epoch": 0.43,1146 "learning_rate": 4.783599088838269e-05,1147 "loss": 5.446,1148 "step": 1901149 },1150 {1151 "epoch": 0.44,1152 "learning_rate": 4.78246013667426e-05,1153 "loss": 4.1151,1154 "step": 1911155 },1156 {1157 "epoch": 0.44,1158 "learning_rate": 4.781321184510251e-05,1159 "loss": 3.6497,1160 "step": 1921161 },1162 {1163 "epoch": 0.44,1164 "learning_rate": 4.780182232346242e-05,1165 "loss": 5.479,1166 "step": 1931167 },1168 {1169 "epoch": 0.44,1170 "learning_rate": 4.779043280182233e-05,1171 "loss": 5.2026,1172 "step": 1941173 },1174 {1175 "epoch": 0.44,1176 "learning_rate": 4.777904328018223e-05,1177 "loss": 5.0487,1178 "step": 1951179 },1180 {1181 "epoch": 0.45,1182 "learning_rate": 4.776765375854214e-05,1183 "loss": 5.3173,1184 "step": 1961185 },1186 {1187 "epoch": 0.45,1188 "learning_rate": 4.7756264236902054e-05,1189 "loss": 5.2775,1190 "step": 1971191 },1192 {1193 "epoch": 0.45,1194 "learning_rate": 4.774487471526196e-05,1195 "loss": 5.9495,1196 "step": 1981197 },1198 {1199 "epoch": 0.45,1200 "learning_rate": 4.773348519362187e-05,