YieldInc/pythoncode
012
1{2 "best_metric": null,3 "best_model_checkpoint": null,4 "epoch": 2.496,5 "eval_steps": 156,6 "global_step": 312,7 "is_hyper_param_search": false,8 "is_local_process_zero": true,9 "is_world_process_zero": true,10 "log_history": [11 {12 "epoch": 0.01,13 "learning_rate": 0.0001,14 "loss": 1.2612,15 "step": 116 },17 {18 "epoch": 0.02,19 "learning_rate": 0.0002,20 "loss": 0.5839,21 "step": 222 },23 {24 "epoch": 0.02,25 "learning_rate": 0.0001994638069705094,26 "loss": 1.2718,27 "step": 328 },29 {30 "epoch": 0.03,31 "learning_rate": 0.00019892761394101877,32 "loss": 0.9663,33 "step": 434 },35 {36 "epoch": 0.04,37 "learning_rate": 0.00019839142091152817,38 "loss": 0.8353,39 "step": 540 },41 {42 "epoch": 0.05,43 "learning_rate": 0.00019785522788203755,44 "loss": 0.4235,45 "step": 646 },47 {48 "epoch": 0.06,49 "learning_rate": 0.00019731903485254693,50 "loss": 0.3499,51 "step": 752 },53 {54 "epoch": 0.06,55 "learning_rate": 0.0001967828418230563,56 "loss": 0.8885,57 "step": 858 },59 {60 "epoch": 0.07,61 "learning_rate": 0.00019624664879356568,62 "loss": 0.6318,63 "step": 964 },65 {66 "epoch": 0.08,67 "learning_rate": 0.00019571045576407506,68 "loss": 0.7978,69 "step": 1070 },71 {72 "epoch": 0.09,73 "learning_rate": 0.00019517426273458447,74 "loss": 0.3304,75 "step": 1176 },77 {78 "epoch": 0.1,79 "learning_rate": 0.00019463806970509384,80 "loss": 0.8138,81 "step": 1282 },83 {84 "epoch": 0.1,85 "learning_rate": 0.00019410187667560322,86 "loss": 0.3049,87 "step": 1388 },89 {90 "epoch": 0.11,91 "learning_rate": 0.0001935656836461126,92 "loss": 0.9467,93 "step": 1494 },95 {96 "epoch": 0.12,97 "learning_rate": 0.00019302949061662198,98 "loss": 0.4926,99 "step": 15100 },101 {102 "epoch": 0.13,103 "learning_rate": 0.00019249329758713138,104 "loss": 0.2896,105 "step": 16106 },107 {108 "epoch": 0.14,109 "learning_rate": 0.00019195710455764076,110 "loss": 0.314,111 "step": 17112 },113 {114 "epoch": 0.14,115 "learning_rate": 0.00019142091152815014,116 "loss": 0.6319,117 "step": 18118 },119 {120 "epoch": 0.15,121 "learning_rate": 0.00019088471849865952,122 "loss": 0.8047,123 "step": 19124 },125 {126 "epoch": 0.16,127 "learning_rate": 0.0001903485254691689,128 "loss": 0.6664,129 "step": 20130 },131 {132 "epoch": 0.17,133 "learning_rate": 0.00018981233243967828,134 "loss": 0.7509,135 "step": 21136 },137 {138 "epoch": 0.18,139 "learning_rate": 0.00018927613941018768,140 "loss": 0.572,141 "step": 22142 },143 {144 "epoch": 0.18,145 "learning_rate": 0.00018873994638069706,146 "loss": 0.6726,147 "step": 23148 },149 {150 "epoch": 0.19,151 "learning_rate": 0.00018820375335120644,152 "loss": 0.2929,153 "step": 24154 },155 {156 "epoch": 0.2,157 "learning_rate": 0.00018766756032171581,158 "loss": 0.5821,159 "step": 25160 },161 {162 "epoch": 0.21,163 "learning_rate": 0.0001871313672922252,164 "loss": 0.5325,165 "step": 26166 },167 {168 "epoch": 0.22,169 "learning_rate": 0.00018659517426273457,170 "loss": 0.4568,171 "step": 27172 },173 {174 "epoch": 0.22,175 "learning_rate": 0.00018605898123324398,176 "loss": 0.2517,177 "step": 28178 },179 {180 "epoch": 0.23,181 "learning_rate": 0.00018552278820375335,182 "loss": 0.5343,183 "step": 29184 },185 {186 "epoch": 0.24,187 "learning_rate": 0.00018498659517426273,188 "loss": 0.3203,189 "step": 30190 },191 {192 "epoch": 0.25,193 "learning_rate": 0.0001844504021447721,194 "loss": 0.0,195 "step": 31196 },197 {198 "epoch": 0.26,199 "learning_rate": 0.0001839142091152815,200 "loss": 0.2905,201 "step": 32202 },203 {204 "epoch": 0.26,205 "learning_rate": 0.0001833780160857909,206 "loss": 0.5738,207 "step": 33208 },209 {210 "epoch": 0.27,211 "learning_rate": 0.0001828418230563003,212 "loss": 0.5445,213 "step": 34214 },215 {216 "epoch": 0.28,217 "learning_rate": 0.00018230563002680968,218 "loss": 0.6075,219 "step": 35220 },221 {222 "epoch": 0.29,223 "learning_rate": 0.00018176943699731906,224 "loss": 0.6869,225 "step": 36226 },227 {228 "epoch": 0.3,229 "learning_rate": 0.00018123324396782843,230 "loss": 0.2808,231 "step": 37232 },233 {234 "epoch": 0.3,235 "learning_rate": 0.0001806970509383378,236 "loss": 0.6005,237 "step": 38238 },239 {240 "epoch": 0.31,241 "learning_rate": 0.0001801608579088472,242 "loss": 0.5839,243 "step": 39244 },245 {246 "epoch": 0.32,247 "learning_rate": 0.0001796246648793566,248 "loss": 0.5602,249 "step": 40250 },251 {252 "epoch": 0.33,253 "learning_rate": 0.00017908847184986597,254 "loss": 0.5986,255 "step": 41256 },257 {258 "epoch": 0.34,259 "learning_rate": 0.00017855227882037535,260 "loss": 0.3701,261 "step": 42262 },263 {264 "epoch": 0.34,265 "learning_rate": 0.00017801608579088473,266 "loss": 0.2545,267 "step": 43268 },269 {270 "epoch": 0.35,271 "learning_rate": 0.0001774798927613941,272 "loss": 0.6094,273 "step": 44274 },275 {276 "epoch": 0.36,277 "learning_rate": 0.0001769436997319035,278 "loss": 0.0,279 "step": 45280 },281 {282 "epoch": 0.37,283 "learning_rate": 0.0001764075067024129,284 "loss": 0.4668,285 "step": 46286 },287 {288 "epoch": 0.38,289 "learning_rate": 0.00017587131367292227,290 "loss": 0.2302,291 "step": 47292 },293 {294 "epoch": 0.38,295 "learning_rate": 0.00017533512064343165,296 "loss": 0.2882,297 "step": 48298 },299 {300 "epoch": 0.39,301 "learning_rate": 0.00017479892761394103,302 "loss": 0.2486,303 "step": 49304 },305 {306 "epoch": 0.4,307 "learning_rate": 0.0001742627345844504,308 "loss": 0.2614,309 "step": 50310 },311 {312 "epoch": 0.41,313 "learning_rate": 0.0001737265415549598,314 "loss": 0.237,315 "step": 51316 },317 {318 "epoch": 0.42,319 "learning_rate": 0.0001731903485254692,320 "loss": 0.5986,321 "step": 52322 },323 {324 "epoch": 0.42,325 "learning_rate": 0.00017265415549597856,326 "loss": 0.0,327 "step": 53328 },329 {330 "epoch": 0.43,331 "learning_rate": 0.00017211796246648794,332 "loss": 0.5648,333 "step": 54334 },335 {336 "epoch": 0.44,337 "learning_rate": 0.00017158176943699732,338 "loss": 0.5096,339 "step": 55340 },341 {342 "epoch": 0.45,343 "learning_rate": 0.00017104557640750673,344 "loss": 0.5139,345 "step": 56346 },347 {348 "epoch": 0.46,349 "learning_rate": 0.0001705093833780161,350 "loss": 0.2226,351 "step": 57352 },353 {354 "epoch": 0.46,355 "learning_rate": 0.00016997319034852548,356 "loss": 0.4839,357 "step": 58358 },359 {360 "epoch": 0.47,361 "learning_rate": 0.00016943699731903486,362 "loss": 0.5375,363 "step": 59364 },365 {366 "epoch": 0.48,367 "learning_rate": 0.00016890080428954424,368 "loss": 0.5792,369 "step": 60370 },371 {372 "epoch": 0.49,373 "learning_rate": 0.00016836461126005362,374 "loss": 0.6151,375 "step": 61376 },377 {378 "epoch": 0.5,379 "learning_rate": 0.00016782841823056302,380 "loss": 0.5358,381 "step": 62382 },383 {384 "epoch": 0.5,385 "learning_rate": 0.0001672922252010724,386 "loss": 0.4993,387 "step": 63388 },389 {390 "epoch": 0.51,391 "learning_rate": 0.00016675603217158178,392 "loss": 0.2054,393 "step": 64394 },395 {396 "epoch": 0.52,397 "learning_rate": 0.00016621983914209116,398 "loss": 0.2973,399 "step": 65400 },401 {402 "epoch": 0.53,403 "learning_rate": 0.00016568364611260053,404 "loss": 0.4578,405 "step": 66406 },407 {408 "epoch": 0.54,409 "learning_rate": 0.00016514745308310994,410 "loss": 0.4827,411 "step": 67412 },413 {414 "epoch": 0.54,415 "learning_rate": 0.00016461126005361932,416 "loss": 0.2362,417 "step": 68418 },419 {420 "epoch": 0.55,421 "learning_rate": 0.0001640750670241287,422 "loss": 0.2993,423 "step": 69424 },425 {426 "epoch": 0.56,427 "learning_rate": 0.00016353887399463807,428 "loss": 0.6205,429 "step": 70430 },431 {432 "epoch": 0.57,433 "learning_rate": 0.00016300268096514745,434 "loss": 0.5475,435 "step": 71436 },437 {438 "epoch": 0.58,439 "learning_rate": 0.00016246648793565683,440 "loss": 0.4761,441 "step": 72442 },443 {444 "epoch": 0.58,445 "learning_rate": 0.00016193029490616624,446 "loss": 0.2454,447 "step": 73448 },449 {450 "epoch": 0.59,451 "learning_rate": 0.00016139410187667561,452 "loss": 0.559,453 "step": 74454 },455 {456 "epoch": 0.6,457 "learning_rate": 0.000160857908847185,458 "loss": 0.4535,459 "step": 75460 },461 {462 "epoch": 0.61,463 "learning_rate": 0.00016032171581769437,464 "loss": 0.5401,465 "step": 76466 },467 {468 "epoch": 0.62,469 "learning_rate": 0.00015978552278820375,470 "loss": 0.4684,471 "step": 77472 },473 {474 "epoch": 0.62,475 "learning_rate": 0.00015924932975871313,476 "loss": 0.2701,477 "step": 78478 },479 {480 "epoch": 0.63,481 "learning_rate": 0.00015871313672922253,482 "loss": 0.2443,483 "step": 79484 },485 {486 "epoch": 0.64,487 "learning_rate": 0.0001581769436997319,488 "loss": 0.3073,489 "step": 80490 },491 {492 "epoch": 0.65,493 "learning_rate": 0.0001576407506702413,494 "loss": 0.3229,495 "step": 81496 },497 {498 "epoch": 0.66,499 "learning_rate": 0.00015710455764075067,500 "loss": 0.4567,501 "step": 82502 },503 {504 "epoch": 0.66,505 "learning_rate": 0.00015656836461126004,506 "loss": 0.3031,507 "step": 83508 },509 {510 "epoch": 0.67,511 "learning_rate": 0.00015603217158176945,512 "loss": 0.2604,513 "step": 84514 },515 {516 "epoch": 0.68,517 "learning_rate": 0.00015549597855227883,518 "loss": 0.5654,519 "step": 85520 },521 {522 "epoch": 0.69,523 "learning_rate": 0.0001549597855227882,524 "loss": 0.2723,525 "step": 86526 },527 {528 "epoch": 0.7,529 "learning_rate": 0.00015442359249329758,530 "loss": 0.5312,531 "step": 87532 },533 {534 "epoch": 0.7,535 "learning_rate": 0.00015388739946380696,536 "loss": 0.5638,537 "step": 88538 },539 {540 "epoch": 0.71,541 "learning_rate": 0.00015335120643431634,542 "loss": 0.2928,543 "step": 89544 },545 {546 "epoch": 0.72,547 "learning_rate": 0.00015281501340482574,548 "loss": 0.5724,549 "step": 90550 },551 {552 "epoch": 0.73,553 "learning_rate": 0.00015227882037533512,554 "loss": 0.5019,555 "step": 91556 },557 {558 "epoch": 0.74,559 "learning_rate": 0.0001517426273458445,560 "loss": 0.5651,561 "step": 92562 },563 {564 "epoch": 0.74,565 "learning_rate": 0.00015120643431635388,566 "loss": 0.5588,567 "step": 93568 },569 {570 "epoch": 0.75,571 "learning_rate": 0.00015067024128686326,572 "loss": 0.6167,573 "step": 94574 },575 {576 "epoch": 0.76,577 "learning_rate": 0.00015013404825737266,578 "loss": 0.5652,579 "step": 95580 },581 {582 "epoch": 0.77,583 "learning_rate": 0.00014959785522788207,584 "loss": 0.0,585 "step": 96586 },587 {588 "epoch": 0.78,589 "learning_rate": 0.00014906166219839145,590 "loss": 0.2309,591 "step": 97592 },593 {594 "epoch": 0.78,595 "learning_rate": 0.00014852546916890082,596 "loss": 0.5339,597 "step": 98598 },599 {600 "epoch": 0.79,601 "learning_rate": 0.0001479892761394102,602 "loss": 0.522,603 "step": 99604 },605 {606 "epoch": 0.8,607 "learning_rate": 0.00014745308310991958,608 "loss": 0.2473,609 "step": 100610 },611 {612 "epoch": 0.81,613 "learning_rate": 0.00014691689008042896,614 "loss": 0.2674,615 "step": 101616 },617 {618 "epoch": 0.82,619 "learning_rate": 0.00014638069705093836,620 "loss": 0.4833,621 "step": 102622 },623 {624 "epoch": 0.82,625 "learning_rate": 0.00014584450402144774,626 "loss": 0.5153,627 "step": 103628 },629 {630 "epoch": 0.83,631 "learning_rate": 0.00014530831099195712,632 "loss": 0.2366,633 "step": 104634 },635 {636 "epoch": 0.84,637 "learning_rate": 0.0001447721179624665,638 "loss": 0.5037,639 "step": 105640 },641 {642 "epoch": 0.85,643 "learning_rate": 0.00014423592493297588,644 "loss": 0.4913,645 "step": 106646 },647 {648 "epoch": 0.86,649 "learning_rate": 0.00014369973190348528,650 "loss": 0.4979,651 "step": 107652 },653 {654 "epoch": 0.86,655 "learning_rate": 0.00014316353887399466,656 "loss": 0.518,657 "step": 108658 },659 {660 "epoch": 0.87,661 "learning_rate": 0.00014262734584450404,662 "loss": 0.0,663 "step": 109664 },665 {666 "epoch": 0.88,667 "learning_rate": 0.00014209115281501342,668 "loss": 0.5866,669 "step": 110670 },671 {672 "epoch": 0.89,673 "learning_rate": 0.0001415549597855228,674 "loss": 0.4525,675 "step": 111676 },677 {678 "epoch": 0.9,679 "learning_rate": 0.00014101876675603217,680 "loss": 0.2802,681 "step": 112682 },683 {684 "epoch": 0.9,685 "learning_rate": 0.00014048257372654158,686 "loss": 0.4686,687 "step": 113688 },689 {690 "epoch": 0.91,691 "learning_rate": 0.00013994638069705096,692 "loss": 0.4963,693 "step": 114694 },695 {696 "epoch": 0.92,697 "learning_rate": 0.00013941018766756033,698 "loss": 0.4853,699 "step": 115700 },701 {702 "epoch": 0.93,703 "learning_rate": 0.0001388739946380697,704 "loss": 0.6237,705 "step": 116706 },707 {708 "epoch": 0.94,709 "learning_rate": 0.0001383378016085791,710 "loss": 0.582,711 "step": 117712 },713 {714 "epoch": 0.94,715 "learning_rate": 0.0001378016085790885,716 "loss": 0.5814,717 "step": 118718 },719 {720 "epoch": 0.95,721 "learning_rate": 0.00013726541554959787,722 "loss": 0.4702,723 "step": 119724 },725 {726 "epoch": 0.96,727 "learning_rate": 0.00013672922252010725,728 "loss": 0.2196,729 "step": 120730 },731 {732 "epoch": 0.97,733 "learning_rate": 0.00013619302949061663,734 "loss": 0.5335,735 "step": 121736 },737 {738 "epoch": 0.98,739 "learning_rate": 0.000135656836461126,740 "loss": 0.5734,741 "step": 122742 },743 {744 "epoch": 0.98,745 "learning_rate": 0.00013512064343163539,746 "loss": 0.0,747 "step": 123748 },749 {750 "epoch": 0.99,751 "learning_rate": 0.0001345844504021448,752 "loss": 0.385,753 "step": 124754 },755 {756 "epoch": 1.0,757 "learning_rate": 0.00013404825737265417,758 "loss": 0.4994,759 "step": 125760 },761 {762 "epoch": 1.01,763 "learning_rate": 0.00013351206434316355,764 "loss": 0.2143,765 "step": 126766 },767 {768 "epoch": 1.02,769 "learning_rate": 0.00013297587131367293,770 "loss": 0.609,771 "step": 127772 },773 {774 "epoch": 1.02,775 "learning_rate": 0.0001324396782841823,776 "loss": 0.2692,777 "step": 128778 },779 {780 "epoch": 1.03,781 "learning_rate": 0.00013190348525469168,782 "loss": 0.2581,783 "step": 129784 },785 {786 "epoch": 1.04,787 "learning_rate": 0.0001313672922252011,788 "loss": 0.3353,789 "step": 130790 },791 {792 "epoch": 1.05,793 "learning_rate": 0.00013083109919571046,794 "loss": 0.4328,795 "step": 131796 },797 {798 "epoch": 1.06,799 "learning_rate": 0.00013029490616621984,800 "loss": 0.2286,801 "step": 132802 },803 {804 "epoch": 1.06,805 "learning_rate": 0.00012975871313672922,806 "loss": 0.4495,807 "step": 133808 },809 {810 "epoch": 1.07,811 "learning_rate": 0.0001292225201072386,812 "loss": 0.5591,813 "step": 134814 },815 {816 "epoch": 1.08,817 "learning_rate": 0.000128686327077748,818 "loss": 0.2977,819 "step": 135820 },821 {822 "epoch": 1.09,823 "learning_rate": 0.00012815013404825738,824 "loss": 0.4241,825 "step": 136826 },827 {828 "epoch": 1.1,829 "learning_rate": 0.00012761394101876676,830 "loss": 0.3706,831 "step": 137832 },833 {834 "epoch": 1.1,835 "learning_rate": 0.00012707774798927614,836 "loss": 0.4937,837 "step": 138838 },839 {840 "epoch": 1.11,841 "learning_rate": 0.00012654155495978552,842 "loss": 0.4193,843 "step": 139844 },845 {846 "epoch": 1.12,847 "learning_rate": 0.0001260053619302949,848 "loss": 0.4165,849 "step": 140850 },851 {852 "epoch": 1.13,853 "learning_rate": 0.0001254691689008043,854 "loss": 0.4585,855 "step": 141856 },857 {858 "epoch": 1.14,859 "learning_rate": 0.00012493297587131368,860 "loss": 0.4031,861 "step": 142862 },863 {864 "epoch": 1.14,865 "learning_rate": 0.00012439678284182306,866 "loss": 0.1778,867 "step": 143868 },869 {870 "epoch": 1.15,871 "learning_rate": 0.00012386058981233243,872 "loss": 0.4911,873 "step": 144874 },875 {876 "epoch": 1.16,877 "learning_rate": 0.0001233243967828418,878 "loss": 0.489,879 "step": 145880 },881 {882 "epoch": 1.17,883 "learning_rate": 0.00012278820375335122,884 "loss": 0.3833,885 "step": 146886 },887 {888 "epoch": 1.18,889 "learning_rate": 0.0001222520107238606,890 "loss": 0.4249,891 "step": 147892 },893 {894 "epoch": 1.18,895 "learning_rate": 0.00012171581769436997,896 "loss": 0.1813,897 "step": 148898 },899 {900 "epoch": 1.19,901 "learning_rate": 0.00012117962466487935,902 "loss": 0.2181,903 "step": 149904 },905 {906 "epoch": 1.2,907 "learning_rate": 0.00012064343163538874,908 "loss": 0.4004,909 "step": 150910 },911 {912 "epoch": 1.21,913 "learning_rate": 0.00012010723860589812,914 "loss": 0.5599,915 "step": 151916 },917 {918 "epoch": 1.22,919 "learning_rate": 0.0001195710455764075,920 "loss": 0.2171,921 "step": 152922 },923 {924 "epoch": 1.22,925 "learning_rate": 0.00011903485254691689,926 "loss": 0.4072,927 "step": 153928 },929 {930 "epoch": 1.23,931 "learning_rate": 0.00011849865951742627,932 "loss": 0.4229,933 "step": 154934 },935 {936 "epoch": 1.24,937 "learning_rate": 0.00011796246648793565,938 "loss": 0.3922,939 "step": 155940 },941 {942 "epoch": 1.25,943 "learning_rate": 0.00011742627345844504,944 "loss": 0.4483,945 "step": 156946 },947 {948 "epoch": 1.25,949 "eval_loss": 0.4745557904243469,950 "eval_runtime": 507.5138,951 "eval_samples_per_second": 1.97,952 "eval_steps_per_second": 0.124,953 "step": 156954 },955 {956 "epoch": 1.26,957 "learning_rate": 0.00011689008042895442,958 "loss": 0.3986,959 "step": 157960 },961 {962 "epoch": 1.26,963 "learning_rate": 0.00011635388739946382,964 "loss": 0.3715,965 "step": 158966 },967 {968 "epoch": 1.27,969 "learning_rate": 0.0001158176943699732,970 "loss": 0.3707,971 "step": 159972 },973 {974 "epoch": 1.28,975 "learning_rate": 0.00011528150134048259,976 "loss": 0.2226,977 "step": 160978 },979 {980 "epoch": 1.29,981 "learning_rate": 0.00011474530831099197,982 "loss": 0.1913,983 "step": 161984 },985 {986 "epoch": 1.3,987 "learning_rate": 0.00011420911528150135,988 "loss": 0.0,989 "step": 162990 },991 {992 "epoch": 1.3,993 "learning_rate": 0.00011367292225201074,994 "loss": 0.4847,995 "step": 163996 },997 {998 "epoch": 1.31,999 "learning_rate": 0.00011313672922252012,1000 "loss": 0.0,1001 "step": 1641002 },1003 {1004 "epoch": 1.32,1005 "learning_rate": 0.00011260053619302951,1006 "loss": 0.2191,1007 "step": 1651008 },1009 {1010 "epoch": 1.33,1011 "learning_rate": 0.00011206434316353889,1012 "loss": 0.4678,1013 "step": 1661014 },1015 {1016 "epoch": 1.34,1017 "learning_rate": 0.00011152815013404827,1018 "loss": 0.4135,1019 "step": 1671020 },1021 {1022 "epoch": 1.34,1023 "learning_rate": 0.00011099195710455766,1024 "loss": 0.4971,1025 "step": 1681026 },1027 {1028 "epoch": 1.35,1029 "learning_rate": 0.00011045576407506704,1030 "loss": 0.2149,1031 "step": 1691032 },1033 {1034 "epoch": 1.36,1035 "learning_rate": 0.00010991957104557641,1036 "loss": 0.2638,1037 "step": 1701038 },1039 {1040 "epoch": 1.37,1041 "learning_rate": 0.0001093833780160858,1042 "loss": 0.5197,1043 "step": 1711044 },1045 {1046 "epoch": 1.38,1047 "learning_rate": 0.00010884718498659518,1048 "loss": 0.4409,1049 "step": 1721050 },1051 {1052 "epoch": 1.38,1053 "learning_rate": 0.00010831099195710456,1054 "loss": 0.4943,1055 "step": 1731056 },1057 {1058 "epoch": 1.39,1059 "learning_rate": 0.00010777479892761395,1060 "loss": 0.4369,1061 "step": 1741062 },1063 {1064 "epoch": 1.4,1065 "learning_rate": 0.00010723860589812333,1066 "loss": 0.281,1067 "step": 1751068 },1069 {1070 "epoch": 1.41,1071 "learning_rate": 0.00010670241286863272,1072 "loss": 0.5053,1073 "step": 1761074 },1075 {1076 "epoch": 1.42,1077 "learning_rate": 0.0001061662198391421,1078 "loss": 0.4787,1079 "step": 1771080 },1081 {1082 "epoch": 1.42,1083 "learning_rate": 0.00010563002680965148,1084 "loss": 0.4353,1085 "step": 1781086 },1087 {1088 "epoch": 1.43,1089 "learning_rate": 0.00010509383378016087,1090 "loss": 0.437,1091 "step": 1791092 },1093 {1094 "epoch": 1.44,1095 "learning_rate": 0.00010455764075067025,1096 "loss": 0.2177,1097 "step": 1801098 },1099 {1100 "epoch": 1.45,1101 "learning_rate": 0.00010402144772117963,1102 "loss": 0.0,1103 "step": 1811104 },1105 {1106 "epoch": 1.46,1107 "learning_rate": 0.00010348525469168902,1108 "loss": 0.2128,1109 "step": 1821110 },1111 {1112 "epoch": 1.46,1113 "learning_rate": 0.0001029490616621984,1114 "loss": 0.622,1115 "step": 1831116 },1117 {1118 "epoch": 1.47,1119 "learning_rate": 0.00010241286863270778,1120 "loss": 0.0,1121 "step": 1841122 },1123 {1124 "epoch": 1.48,1125 "learning_rate": 0.00010187667560321717,1126 "loss": 0.285,1127 "step": 1851128 },1129 {1130 "epoch": 1.49,1131 "learning_rate": 0.00010134048257372655,1132 "loss": 0.4149,1133 "step": 1861134 },1135 {1136 "epoch": 1.5,1137 "learning_rate": 0.00010080428954423592,1138 "loss": 0.426,1139 "step": 1871140 },1141 {1142 "epoch": 1.5,1143 "learning_rate": 0.00010026809651474532,1144 "loss": 0.5177,1145 "step": 1881146 },1147 {1148 "epoch": 1.51,1149 "learning_rate": 9.97319034852547e-05,1150 "loss": 0.5009,1151 "step": 1891152 },1153 {1154 "epoch": 1.52,1155 "learning_rate": 9.919571045576409e-05,1156 "loss": 0.2128,1157 "step": 1901158 },1159 {1160 "epoch": 1.53,1161 "learning_rate": 9.865951742627346e-05,1162 "loss": 0.2701,1163 "step": 1911164 },1165 {1166 "epoch": 1.54,1167 "learning_rate": 9.812332439678284e-05,1168 "loss": 0.4293,1169 "step": 1921170 },1171 {1172 "epoch": 1.54,1173 "learning_rate": 9.758713136729223e-05,1174 "loss": 0.4286,1175 "step": 1931176 },1177 {1178 "epoch": 1.55,1179 "learning_rate": 9.705093833780161e-05,1180 "loss": 0.4106,1181 "step": 1941182 },1183 {1184 "epoch": 1.56,1185 "learning_rate": 9.651474530831099e-05,1186 "loss": 0.6385,1187 "step": 1951188 },1189 {1190 "epoch": 1.57,1191 "learning_rate": 9.597855227882038e-05,1192 "loss": 0.2073,1193 "step": 1961194 },1195 {1196 "epoch": 1.58,1197 "learning_rate": 9.544235924932976e-05,1198 "loss": 0.5092,1199 "step": 1971200 },