LangAGI-Lab/M2WEB-HTML-WM
08
1{2 "best_metric": null,3 "best_model_checkpoint": null,4 "epoch": 1.9751552795031055,5 "eval_steps": 41,6 "global_step": 322,7 "is_hyper_param_search": false,8 "is_local_process_zero": true,9 "is_world_process_zero": true,10 "log_history": [11 {12 "epoch": 0.006211180124223602,13 "grad_norm": 0.3814874589443207,14 "learning_rate": 1.0000000000000002e-06,15 "loss": 1.538,16 "step": 117 },18 {19 "epoch": 0.006211180124223602,20 "eval_loss": 1.570106863975525,21 "eval_runtime": 26.7523,22 "eval_samples_per_second": 2.953,23 "eval_steps_per_second": 0.374,24 "step": 125 },26 {27 "epoch": 0.012422360248447204,28 "grad_norm": 0.34790050983428955,29 "learning_rate": 2.0000000000000003e-06,30 "loss": 1.5515,31 "step": 232 },33 {34 "epoch": 0.018633540372670808,35 "grad_norm": 0.315510630607605,36 "learning_rate": 3e-06,37 "loss": 1.6394,38 "step": 339 },40 {41 "epoch": 0.024844720496894408,42 "grad_norm": 0.333831787109375,43 "learning_rate": 4.000000000000001e-06,44 "loss": 1.5686,45 "step": 446 },47 {48 "epoch": 0.031055900621118012,49 "grad_norm": 0.3356265723705292,50 "learning_rate": 5e-06,51 "loss": 1.6442,52 "step": 553 },54 {55 "epoch": 0.037267080745341616,56 "grad_norm": 0.3302271366119385,57 "learning_rate": 6e-06,58 "loss": 1.6543,59 "step": 660 },61 {62 "epoch": 0.043478260869565216,63 "grad_norm": 0.3428245484828949,64 "learning_rate": 7e-06,65 "loss": 1.583,66 "step": 767 },68 {69 "epoch": 0.049689440993788817,70 "grad_norm": 0.3571615517139435,71 "learning_rate": 8.000000000000001e-06,72 "loss": 1.5863,73 "step": 874 },75 {76 "epoch": 0.055900621118012424,77 "grad_norm": 0.3609354496002197,78 "learning_rate": 9e-06,79 "loss": 1.5861,80 "step": 981 },82 {83 "epoch": 0.062111801242236024,84 "grad_norm": 0.38313618302345276,85 "learning_rate": 1e-05,86 "loss": 1.6357,87 "step": 1088 },89 {90 "epoch": 0.06832298136645963,91 "grad_norm": 0.3678596317768097,92 "learning_rate": 9.99974652980635e-06,93 "loss": 1.6452,94 "step": 1195 },96 {97 "epoch": 0.07453416149068323,98 "grad_norm": 0.36209872364997864,99 "learning_rate": 9.998986144924253e-06,100 "loss": 1.6364,101 "step": 12102 },103 {104 "epoch": 0.08074534161490683,105 "grad_norm": 0.3838407099246979,106 "learning_rate": 9.997718922447669e-06,107 "loss": 1.6456,108 "step": 13109 },110 {111 "epoch": 0.08695652173913043,112 "grad_norm": 0.3796548545360565,113 "learning_rate": 9.995944990857848e-06,114 "loss": 1.5417,115 "step": 14116 },117 {118 "epoch": 0.09316770186335403,119 "grad_norm": 0.393232524394989,120 "learning_rate": 9.993664530010308e-06,121 "loss": 1.5553,122 "step": 15123 },124 {125 "epoch": 0.09937888198757763,126 "grad_norm": 0.40382516384124756,127 "learning_rate": 9.990877771116588e-06,128 "loss": 1.4974,129 "step": 16130 },131 {132 "epoch": 0.10559006211180125,133 "grad_norm": 0.410980224609375,134 "learning_rate": 9.987584996720813e-06,135 "loss": 1.5143,136 "step": 17137 },138 {139 "epoch": 0.11180124223602485,140 "grad_norm": 0.43435031175613403,141 "learning_rate": 9.983786540671052e-06,142 "loss": 1.7078,143 "step": 18144 },145 {146 "epoch": 0.11801242236024845,147 "grad_norm": 0.46852847933769226,148 "learning_rate": 9.979482788085455e-06,149 "loss": 1.5725,150 "step": 19151 },152 {153 "epoch": 0.12422360248447205,154 "grad_norm": 0.44078120589256287,155 "learning_rate": 9.974674175313228e-06,156 "loss": 1.5146,157 "step": 20158 },159 {160 "epoch": 0.13043478260869565,161 "grad_norm": 0.4520568251609802,162 "learning_rate": 9.969361189890373e-06,163 "loss": 1.551,164 "step": 21165 },166 {167 "epoch": 0.13664596273291926,168 "grad_norm": 0.44940102100372314,169 "learning_rate": 9.96354437049027e-06,170 "loss": 1.5154,171 "step": 22172 },173 {174 "epoch": 0.14285714285714285,175 "grad_norm": 0.42880144715309143,176 "learning_rate": 9.957224306869053e-06,177 "loss": 1.5664,178 "step": 23179 },180 {181 "epoch": 0.14906832298136646,182 "grad_norm": 0.4296427071094513,183 "learning_rate": 9.950401639805822e-06,184 "loss": 1.5391,185 "step": 24186 },187 {188 "epoch": 0.15527950310559005,189 "grad_norm": 0.4148462116718292,190 "learning_rate": 9.943077061037672e-06,191 "loss": 1.559,192 "step": 25193 },194 {195 "epoch": 0.16149068322981366,196 "grad_norm": 0.45027458667755127,197 "learning_rate": 9.935251313189564e-06,198 "loss": 1.5797,199 "step": 26200 },201 {202 "epoch": 0.16770186335403728,203 "grad_norm": 0.43325772881507874,204 "learning_rate": 9.92692518969903e-06,205 "loss": 1.5651,206 "step": 27207 },208 {209 "epoch": 0.17391304347826086,210 "grad_norm": 0.4363733232021332,211 "learning_rate": 9.91809953473572e-06,212 "loss": 1.4171,213 "step": 28214 },215 {216 "epoch": 0.18012422360248448,217 "grad_norm": 0.4001283347606659,218 "learning_rate": 9.908775243115822e-06,219 "loss": 1.5172,220 "step": 29221 },222 {223 "epoch": 0.18633540372670807,224 "grad_norm": 0.42918145656585693,225 "learning_rate": 9.89895326021134e-06,226 "loss": 1.4501,227 "step": 30228 },229 {230 "epoch": 0.19254658385093168,231 "grad_norm": 0.3902086019515991,232 "learning_rate": 9.888634581854235e-06,233 "loss": 1.4396,234 "step": 31235 },236 {237 "epoch": 0.19875776397515527,238 "grad_norm": 0.4069195091724396,239 "learning_rate": 9.87782025423547e-06,240 "loss": 1.4399,241 "step": 32242 },243 {244 "epoch": 0.20496894409937888,245 "grad_norm": 0.40982964634895325,246 "learning_rate": 9.86651137379893e-06,247 "loss": 1.4445,248 "step": 33249 },250 {251 "epoch": 0.2111801242236025,252 "grad_norm": 0.4249035716056824,253 "learning_rate": 9.854709087130261e-06,254 "loss": 1.4402,255 "step": 34256 },257 {258 "epoch": 0.21739130434782608,259 "grad_norm": 0.4334266781806946,260 "learning_rate": 9.842414590840618e-06,261 "loss": 1.4633,262 "step": 35263 },264 {265 "epoch": 0.2236024844720497,266 "grad_norm": 0.4241773188114166,267 "learning_rate": 9.829629131445342e-06,268 "loss": 1.5354,269 "step": 36270 },271 {272 "epoch": 0.22981366459627328,273 "grad_norm": 0.45813798904418945,274 "learning_rate": 9.816354005237583e-06,275 "loss": 1.4568,276 "step": 37277 },278 {279 "epoch": 0.2360248447204969,280 "grad_norm": 0.4499090909957886,281 "learning_rate": 9.802590558156863e-06,282 "loss": 1.4483,283 "step": 38284 },285 {286 "epoch": 0.2422360248447205,287 "grad_norm": 0.44764548540115356,288 "learning_rate": 9.78834018565262e-06,289 "loss": 1.4769,290 "step": 39291 },292 {293 "epoch": 0.2484472049689441,294 "grad_norm": 0.4358612298965454,295 "learning_rate": 9.77360433254273e-06,296 "loss": 1.5545,297 "step": 40298 },299 {300 "epoch": 0.2546583850931677,301 "grad_norm": 0.44274434447288513,302 "learning_rate": 9.758384492867004e-06,303 "loss": 1.3615,304 "step": 41305 },306 {307 "epoch": 0.2546583850931677,308 "eval_loss": 1.3996479511260986,309 "eval_runtime": 27.0472,310 "eval_samples_per_second": 2.921,311 "eval_steps_per_second": 0.37,312 "step": 41313 },314 {315 "epoch": 0.2608695652173913,316 "grad_norm": 0.42516258358955383,317 "learning_rate": 9.742682209735727e-06,318 "loss": 1.4802,319 "step": 42320 },321 {322 "epoch": 0.2670807453416149,323 "grad_norm": 0.40934988856315613,324 "learning_rate": 9.726499075173201e-06,325 "loss": 1.3283,326 "step": 43327 },328 {329 "epoch": 0.2732919254658385,330 "grad_norm": 0.40728411078453064,331 "learning_rate": 9.709836729956326e-06,332 "loss": 1.4724,333 "step": 44334 },335 {336 "epoch": 0.2795031055900621,337 "grad_norm": 0.405512273311615,338 "learning_rate": 9.692696863448246e-06,339 "loss": 1.3943,340 "step": 45341 },342 {343 "epoch": 0.2857142857142857,344 "grad_norm": 0.4585839807987213,345 "learning_rate": 9.675081213427076e-06,346 "loss": 1.4038,347 "step": 46348 },349 {350 "epoch": 0.2919254658385093,351 "grad_norm": 0.40490880608558655,352 "learning_rate": 9.656991565909703e-06,353 "loss": 1.3807,354 "step": 47355 },356 {357 "epoch": 0.2981366459627329,358 "grad_norm": 0.4017808735370636,359 "learning_rate": 9.638429754970715e-06,360 "loss": 1.3532,361 "step": 48362 },363 {364 "epoch": 0.30434782608695654,365 "grad_norm": 0.41484227776527405,366 "learning_rate": 9.619397662556434e-06,367 "loss": 1.3379,368 "step": 49369 },370 {371 "epoch": 0.3105590062111801,372 "grad_norm": 0.36943531036376953,373 "learning_rate": 9.599897218294122e-06,374 "loss": 1.393,375 "step": 50376 },377 {378 "epoch": 0.3167701863354037,379 "grad_norm": 0.39943498373031616,380 "learning_rate": 9.57993039929633e-06,381 "loss": 1.3039,382 "step": 51383 },384 {385 "epoch": 0.32298136645962733,386 "grad_norm": 0.374943345785141,387 "learning_rate": 9.55949922996045e-06,388 "loss": 1.355,389 "step": 52390 },391 {392 "epoch": 0.32919254658385094,393 "grad_norm": 0.4175715446472168,394 "learning_rate": 9.538605781763464e-06,395 "loss": 1.3716,396 "step": 53397 },398 {399 "epoch": 0.33540372670807456,400 "grad_norm": 0.37441951036453247,401 "learning_rate": 9.517252173051912e-06,402 "loss": 1.3081,403 "step": 54404 },405 {406 "epoch": 0.3416149068322981,407 "grad_norm": 0.39863765239715576,408 "learning_rate": 9.49544056882713e-06,409 "loss": 1.3942,410 "step": 55411 },412 {413 "epoch": 0.34782608695652173,414 "grad_norm": 0.3854600787162781,415 "learning_rate": 9.473173180525737e-06,416 "loss": 1.2902,417 "step": 56418 },419 {420 "epoch": 0.35403726708074534,421 "grad_norm": 0.42524421215057373,422 "learning_rate": 9.450452265795423e-06,423 "loss": 1.2544,424 "step": 57425 },426 {427 "epoch": 0.36024844720496896,428 "grad_norm": 0.41367658972740173,429 "learning_rate": 9.427280128266049e-06,430 "loss": 1.3194,431 "step": 58432 },433 {434 "epoch": 0.36645962732919257,435 "grad_norm": 0.3787434697151184,436 "learning_rate": 9.403659117316093e-06,437 "loss": 1.3077,438 "step": 59439 },440 {441 "epoch": 0.37267080745341613,442 "grad_norm": 0.4197174608707428,443 "learning_rate": 9.37959162783444e-06,444 "loss": 1.2356,445 "step": 60446 },447 {448 "epoch": 0.37888198757763975,449 "grad_norm": 0.39209407567977905,450 "learning_rate": 9.355080099977579e-06,451 "loss": 1.2168,452 "step": 61453 },454 {455 "epoch": 0.38509316770186336,456 "grad_norm": 0.41228196024894714,457 "learning_rate": 9.330127018922195e-06,458 "loss": 1.2237,459 "step": 62460 },461 {462 "epoch": 0.391304347826087,463 "grad_norm": 0.4241032302379608,464 "learning_rate": 9.3047349146132e-06,465 "loss": 1.3146,466 "step": 63467 },468 {469 "epoch": 0.39751552795031053,470 "grad_norm": 0.4294179677963257,471 "learning_rate": 9.278906361507238e-06,472 "loss": 1.2716,473 "step": 64474 },475 {476 "epoch": 0.40372670807453415,477 "grad_norm": 0.4411190450191498,478 "learning_rate": 9.252643978311649e-06,479 "loss": 1.2735,480 "step": 65481 },482 {483 "epoch": 0.40993788819875776,484 "grad_norm": 0.46513664722442627,485 "learning_rate": 9.225950427718974e-06,486 "loss": 1.2337,487 "step": 66488 },489 {490 "epoch": 0.4161490683229814,491 "grad_norm": 0.36713191866874695,492 "learning_rate": 9.198828416136991e-06,493 "loss": 1.3339,494 "step": 67495 },496 {497 "epoch": 0.422360248447205,498 "grad_norm": 0.351909875869751,499 "learning_rate": 9.171280693414307e-06,500 "loss": 1.358,501 "step": 68502 },503 {504 "epoch": 0.42857142857142855,505 "grad_norm": 0.3643895983695984,506 "learning_rate": 9.14331005256157e-06,507 "loss": 1.2482,508 "step": 69509 },510 {511 "epoch": 0.43478260869565216,512 "grad_norm": 0.4360922873020172,513 "learning_rate": 9.114919329468283e-06,514 "loss": 1.3115,515 "step": 70516 },517 {518 "epoch": 0.4409937888198758,519 "grad_norm": 0.4254709482192993,520 "learning_rate": 9.086111402615274e-06,521 "loss": 1.2351,522 "step": 71523 },524 {525 "epoch": 0.4472049689440994,526 "grad_norm": 0.3786623179912567,527 "learning_rate": 9.056889192782865e-06,528 "loss": 1.3746,529 "step": 72530 },531 {532 "epoch": 0.453416149068323,533 "grad_norm": 0.4647882878780365,534 "learning_rate": 9.02725566275473e-06,535 "loss": 1.2395,536 "step": 73537 },538 {539 "epoch": 0.45962732919254656,540 "grad_norm": 0.4040616452693939,541 "learning_rate": 8.997213817017508e-06,542 "loss": 1.2998,543 "step": 74544 },545 {546 "epoch": 0.4658385093167702,547 "grad_norm": 0.41700279712677,548 "learning_rate": 8.966766701456177e-06,549 "loss": 1.2105,550 "step": 75551 },552 {553 "epoch": 0.4720496894409938,554 "grad_norm": 0.3908318877220154,555 "learning_rate": 8.935917403045251e-06,556 "loss": 1.2011,557 "step": 76558 },559 {560 "epoch": 0.4782608695652174,561 "grad_norm": 0.38885021209716797,562 "learning_rate": 8.90466904953579e-06,563 "loss": 1.2471,564 "step": 77565 },566 {567 "epoch": 0.484472049689441,568 "grad_norm": 0.42335832118988037,569 "learning_rate": 8.873024809138272e-06,570 "loss": 1.2683,571 "step": 78572 },573 {574 "epoch": 0.4906832298136646,575 "grad_norm": 0.39654576778411865,576 "learning_rate": 8.840987890201404e-06,577 "loss": 1.2587,578 "step": 79579 },580 {581 "epoch": 0.4968944099378882,582 "grad_norm": 0.4123787581920624,583 "learning_rate": 8.808561540886796e-06,584 "loss": 1.1794,585 "step": 80586 },587 {588 "epoch": 0.5031055900621118,589 "grad_norm": 0.42227044701576233,590 "learning_rate": 8.77574904883967e-06,591 "loss": 1.2339,592 "step": 81593 },594 {595 "epoch": 0.5093167701863354,596 "grad_norm": 0.4471132159233093,597 "learning_rate": 8.742553740855507e-06,598 "loss": 1.1954,599 "step": 82600 },601 {602 "epoch": 0.5093167701863354,603 "eval_loss": 1.2136852741241455,604 "eval_runtime": 26.7767,605 "eval_samples_per_second": 2.95,606 "eval_steps_per_second": 0.373,607 "step": 82608 },609 {610 "epoch": 0.515527950310559,611 "grad_norm": 0.42590177059173584,612 "learning_rate": 8.708978982542766e-06,613 "loss": 1.2426,614 "step": 83615 },616 {617 "epoch": 0.5217391304347826,618 "grad_norm": 0.39879706501960754,619 "learning_rate": 8.675028177981643e-06,620 "loss": 1.2618,621 "step": 84622 },623 {624 "epoch": 0.5279503105590062,625 "grad_norm": 0.4137211740016937,626 "learning_rate": 8.640704769378943e-06,627 "loss": 1.1735,628 "step": 85629 },630 {631 "epoch": 0.5341614906832298,632 "grad_norm": 0.42463183403015137,633 "learning_rate": 8.606012236719073e-06,634 "loss": 1.1433,635 "step": 86636 },637 {638 "epoch": 0.5403726708074534,639 "grad_norm": 0.42853713035583496,640 "learning_rate": 8.570954097411224e-06,641 "loss": 1.2307,642 "step": 87643 },644 {645 "epoch": 0.546583850931677,646 "grad_norm": 0.408589243888855,647 "learning_rate": 8.535533905932739e-06,648 "loss": 1.1523,649 "step": 88650 },651 {652 "epoch": 0.5527950310559007,653 "grad_norm": 0.42884618043899536,654 "learning_rate": 8.499755253468732e-06,655 "loss": 1.2494,656 "step": 89657 },658 {659 "epoch": 0.5590062111801242,660 "grad_norm": 0.4813741445541382,661 "learning_rate": 8.463621767547998e-06,662 "loss": 1.2417,663 "step": 90664 },665 {666 "epoch": 0.5652173913043478,667 "grad_norm": 0.4021080732345581,668 "learning_rate": 8.4271371116752e-06,669 "loss": 1.2715,670 "step": 91671 },672 {673 "epoch": 0.5714285714285714,674 "grad_norm": 0.3832411468029022,675 "learning_rate": 8.390304984959455e-06,676 "loss": 1.1586,677 "step": 92678 },679 {680 "epoch": 0.577639751552795,681 "grad_norm": 0.4360566735267639,682 "learning_rate": 8.35312912173928e-06,683 "loss": 1.217,684 "step": 93685 },686 {687 "epoch": 0.5838509316770186,688 "grad_norm": 0.44515809416770935,689 "learning_rate": 8.315613291203977e-06,690 "loss": 1.264,691 "step": 94692 },693 {694 "epoch": 0.5900621118012422,695 "grad_norm": 0.41051459312438965,696 "learning_rate": 8.277761297011475e-06,697 "loss": 1.1578,698 "step": 95699 },700 {701 "epoch": 0.5962732919254659,702 "grad_norm": 0.45815277099609375,703 "learning_rate": 8.239576976902694e-06,704 "loss": 1.2084,705 "step": 96706 },707 {708 "epoch": 0.6024844720496895,709 "grad_norm": 0.4373926520347595,710 "learning_rate": 8.20106420231244e-06,711 "loss": 1.1925,712 "step": 97713 },714 {715 "epoch": 0.6086956521739131,716 "grad_norm": 0.444323867559433,717 "learning_rate": 8.162226877976886e-06,718 "loss": 1.2183,719 "step": 98720 },721 {722 "epoch": 0.6149068322981367,723 "grad_norm": 0.4503950774669647,724 "learning_rate": 8.123068941537681e-06,725 "loss": 1.1156,726 "step": 99727 },728 {729 "epoch": 0.6211180124223602,730 "grad_norm": 0.4149700105190277,731 "learning_rate": 8.083594363142717e-06,732 "loss": 1.2379,733 "step": 100734 },735 {736 "epoch": 0.6273291925465838,737 "grad_norm": 0.472960501909256,738 "learning_rate": 8.043807145043604e-06,739 "loss": 1.14,740 "step": 101741 },742 {743 "epoch": 0.6335403726708074,744 "grad_norm": 0.40744665265083313,745 "learning_rate": 8.003711321189895e-06,746 "loss": 1.2446,747 "step": 102748 },749 {750 "epoch": 0.639751552795031,751 "grad_norm": 0.3995897173881531,752 "learning_rate": 7.963310956820085e-06,753 "loss": 1.179,754 "step": 103755 },756 {757 "epoch": 0.6459627329192547,758 "grad_norm": 0.4376934766769409,759 "learning_rate": 7.922610148049445e-06,760 "loss": 1.2818,761 "step": 104762 },763 {764 "epoch": 0.6521739130434783,765 "grad_norm": 0.4294185936450958,766 "learning_rate": 7.881613021454728e-06,767 "loss": 1.1463,768 "step": 105769 },770 {771 "epoch": 0.6583850931677019,772 "grad_norm": 0.46748086810112,773 "learning_rate": 7.84032373365578e-06,774 "loss": 1.1702,775 "step": 106776 },777 {778 "epoch": 0.6645962732919255,779 "grad_norm": 0.47421255707740784,780 "learning_rate": 7.798746470894113e-06,781 "loss": 1.2922,782 "step": 107783 },784 {785 "epoch": 0.6708074534161491,786 "grad_norm": 0.37985584139823914,787 "learning_rate": 7.75688544860846e-06,788 "loss": 1.1581,789 "step": 108790 },791 {792 "epoch": 0.6770186335403726,793 "grad_norm": 0.4336451590061188,794 "learning_rate": 7.714744911007395e-06,795 "loss": 1.1532,796 "step": 109797 },798 {799 "epoch": 0.6832298136645962,800 "grad_norm": 0.4210762679576874,801 "learning_rate": 7.672329130639007e-06,802 "loss": 1.2047,803 "step": 110804 },805 {806 "epoch": 0.6894409937888198,807 "grad_norm": 0.42413201928138733,808 "learning_rate": 7.62964240795772e-06,809 "loss": 1.125,810 "step": 111811 },812 {813 "epoch": 0.6956521739130435,814 "grad_norm": 0.47521328926086426,815 "learning_rate": 7.586689070888284e-06,816 "loss": 1.2143,817 "step": 112818 },819 {820 "epoch": 0.7018633540372671,821 "grad_norm": 0.41499438881874084,822 "learning_rate": 7.543473474386962e-06,823 "loss": 1.07,824 "step": 113825 },826 {827 "epoch": 0.7080745341614907,828 "grad_norm": 0.4027252495288849,829 "learning_rate": 7.500000000000001e-06,830 "loss": 1.262,831 "step": 114832 },833 {834 "epoch": 0.7142857142857143,835 "grad_norm": 0.4009539484977722,836 "learning_rate": 7.4562730554193875e-06,837 "loss": 1.1742,838 "step": 115839 },840 {841 "epoch": 0.7204968944099379,842 "grad_norm": 0.46428173780441284,843 "learning_rate": 7.412297074035968e-06,844 "loss": 1.1629,845 "step": 116846 },847 {848 "epoch": 0.7267080745341615,849 "grad_norm": 0.4374319016933441,850 "learning_rate": 7.368076514489947e-06,851 "loss": 1.1065,852 "step": 117853 },854 {855 "epoch": 0.7329192546583851,856 "grad_norm": 0.42879152297973633,857 "learning_rate": 7.323615860218844e-06,858 "loss": 1.193,859 "step": 118860 },861 {862 "epoch": 0.7391304347826086,863 "grad_norm": 0.4889649748802185,864 "learning_rate": 7.2789196190029155e-06,865 "loss": 1.2618,866 "step": 119867 },868 {869 "epoch": 0.7453416149068323,870 "grad_norm": 0.4490755796432495,871 "learning_rate": 7.2339923225081296e-06,872 "loss": 1.1857,873 "step": 120874 },875 {876 "epoch": 0.7515527950310559,877 "grad_norm": 0.4516308009624481,878 "learning_rate": 7.188838525826702e-06,879 "loss": 1.1076,880 "step": 121881 },882 {883 "epoch": 0.7577639751552795,884 "grad_norm": 0.5366808176040649,885 "learning_rate": 7.143462807015271e-06,886 "loss": 1.1523,887 "step": 122888 },889 {890 "epoch": 0.7639751552795031,891 "grad_norm": 0.4146997332572937,892 "learning_rate": 7.09786976663073e-06,893 "loss": 1.1766,894 "step": 123895 },896 {897 "epoch": 0.7639751552795031,898 "eval_loss": 1.1551172733306885,899 "eval_runtime": 26.7368,900 "eval_samples_per_second": 2.955,901 "eval_steps_per_second": 0.374,902 "step": 123903 },904 {905 "epoch": 0.7701863354037267,906 "grad_norm": 0.4490116536617279,907 "learning_rate": 7.052064027263785e-06,908 "loss": 1.1971,909 "step": 124910 },911 {912 "epoch": 0.7763975155279503,913 "grad_norm": 0.48434358835220337,914 "learning_rate": 7.006050233070289e-06,915 "loss": 1.239,916 "step": 125917 },918 {919 "epoch": 0.782608695652174,920 "grad_norm": 0.4746357798576355,921 "learning_rate": 6.959833049300376e-06,922 "loss": 1.1586,923 "step": 126924 },925 {926 "epoch": 0.7888198757763976,927 "grad_norm": 0.39503100514411926,928 "learning_rate": 6.913417161825449e-06,929 "loss": 1.0894,930 "step": 127931 },932 {933 "epoch": 0.7950310559006211,934 "grad_norm": 0.43696898221969604,935 "learning_rate": 6.8668072766631054e-06,936 "loss": 1.0384,937 "step": 128938 },939 {940 "epoch": 0.8012422360248447,941 "grad_norm": 0.4739917814731598,942 "learning_rate": 6.820008119499992e-06,943 "loss": 1.1741,944 "step": 129945 },946 {947 "epoch": 0.8074534161490683,948 "grad_norm": 0.4144044518470764,949 "learning_rate": 6.773024435212678e-06,950 "loss": 1.1911,951 "step": 130952 },953 {954 "epoch": 0.8136645962732919,955 "grad_norm": 0.4699600636959076,956 "learning_rate": 6.7258609873865895e-06,957 "loss": 1.2041,958 "step": 131959 },960 {961 "epoch": 0.8198757763975155,962 "grad_norm": 0.40221303701400757,963 "learning_rate": 6.678522557833025e-06,964 "loss": 1.1077,965 "step": 132966 },967 {968 "epoch": 0.8260869565217391,969 "grad_norm": 0.4720539152622223,970 "learning_rate": 6.631013946104348e-06,971 "loss": 1.0987,972 "step": 133973 },974 {975 "epoch": 0.8322981366459627,976 "grad_norm": 0.4860423505306244,977 "learning_rate": 6.583339969007364e-06,978 "loss": 1.1455,979 "step": 134980 },981 {982 "epoch": 0.8385093167701864,983 "grad_norm": 0.5083861947059631,984 "learning_rate": 6.5355054601149545e-06,985 "loss": 1.1314,986 "step": 135987 },988 {989 "epoch": 0.84472049689441,990 "grad_norm": 0.5092173218727112,991 "learning_rate": 6.487515269276015e-06,992 "loss": 1.2228,993 "step": 136994 },995 {996 "epoch": 0.8509316770186336,997 "grad_norm": 0.4544651508331299,998 "learning_rate": 6.439374262123731e-06,999 "loss": 1.1177,1000 "step": 1371001 },1002 {1003 "epoch": 0.8571428571428571,1004 "grad_norm": 0.45371013879776,1005 "learning_rate": 6.391087319582264e-06,1006 "loss": 1.1782,1007 "step": 1381008 },1009 {1010 "epoch": 0.8633540372670807,1011 "grad_norm": 0.5025550723075867,1012 "learning_rate": 6.342659337371884e-06,1013 "loss": 1.1594,1014 "step": 1391015 },1016 {1017 "epoch": 0.8695652173913043,1018 "grad_norm": 0.48699814081192017,1019 "learning_rate": 6.294095225512604e-06,1020 "loss": 1.1246,1021 "step": 1401022 },1023 {1024 "epoch": 0.8757763975155279,1025 "grad_norm": 0.442331999540329,1026 "learning_rate": 6.2453999078263596e-06,1027 "loss": 1.1265,1028 "step": 1411029 },1030 {1031 "epoch": 0.8819875776397516,1032 "grad_norm": 0.4538658857345581,1033 "learning_rate": 6.1965783214377895e-06,1034 "loss": 1.2054,1035 "step": 1421036 },1037 {1038 "epoch": 0.8881987577639752,1039 "grad_norm": 0.4796921908855438,1040 "learning_rate": 6.147635416273679e-06,1041 "loss": 1.1621,1042 "step": 1431043 },1044 {1045 "epoch": 0.8944099378881988,1046 "grad_norm": 0.4758358895778656,1047 "learning_rate": 6.0985761545610865e-06,1048 "loss": 1.117,1049 "step": 1441050 },1051 {1052 "epoch": 0.9006211180124224,1053 "grad_norm": 0.44177138805389404,1054 "learning_rate": 6.049405510324237e-06,1055 "loss": 1.1016,1056 "step": 1451057 },1058 {1059 "epoch": 0.906832298136646,1060 "grad_norm": 0.48750707507133484,1061 "learning_rate": 6.000128468880223e-06,1062 "loss": 1.2238,1063 "step": 1461064 },1065 {1066 "epoch": 0.9130434782608695,1067 "grad_norm": 0.53321373462677,1068 "learning_rate": 5.950750026333534e-06,1069 "loss": 1.0768,1070 "step": 1471071 },1072 {1073 "epoch": 0.9192546583850931,1074 "grad_norm": 0.4829699993133545,1075 "learning_rate": 5.90127518906953e-06,1076 "loss": 1.1229,1077 "step": 1481078 },1079 {1080 "epoch": 0.9254658385093167,1081 "grad_norm": 0.5378725528717041,1082 "learning_rate": 5.851708973246841e-06,1083 "loss": 1.076,1084 "step": 1491085 },1086 {1087 "epoch": 0.9316770186335404,1088 "grad_norm": 0.44234123826026917,1089 "learning_rate": 5.8020564042888015e-06,1090 "loss": 1.1543,1091 "step": 1501092 },1093 {1094 "epoch": 0.937888198757764,1095 "grad_norm": 0.41885337233543396,1096 "learning_rate": 5.752322516373916e-06,1097 "loss": 1.1918,1098 "step": 1511099 },1100 {1101 "epoch": 0.9440993788819876,1102 "grad_norm": 0.4357737600803375,1103 "learning_rate": 5.7025123519254644e-06,1104 "loss": 1.1822,1105 "step": 1521106 },1107 {1108 "epoch": 0.9503105590062112,1109 "grad_norm": 0.48525384068489075,1110 "learning_rate": 5.65263096110026e-06,1111 "loss": 1.1569,1112 "step": 1531113 },1114 {1115 "epoch": 0.9565217391304348,1116 "grad_norm": 0.4788309335708618,1117 "learning_rate": 5.6026834012766155e-06,1118 "loss": 1.1826,1119 "step": 1541120 },1121 {1122 "epoch": 0.9627329192546584,1123 "grad_norm": 0.4695623815059662,1124 "learning_rate": 5.5526747365416e-06,1125 "loss": 1.196,1126 "step": 1551127 },1128 {1129 "epoch": 0.968944099378882,1130 "grad_norm": 0.38793641328811646,1131 "learning_rate": 5.502610037177586e-06,1132 "loss": 1.0959,1133 "step": 1561134 },1135 {1136 "epoch": 0.9751552795031055,1137 "grad_norm": 0.4816080629825592,1138 "learning_rate": 5.45249437914819e-06,1139 "loss": 1.1359,1140 "step": 1571141 },1142 {1143 "epoch": 0.9813664596273292,1144 "grad_norm": 0.4505300223827362,1145 "learning_rate": 5.402332843583631e-06,1146 "loss": 1.0793,1147 "step": 1581148 },1149 {1150 "epoch": 0.9875776397515528,1151 "grad_norm": 0.5171789526939392,1152 "learning_rate": 5.35213051626556e-06,1153 "loss": 1.1582,1154 "step": 1591155 },1156 {1157 "epoch": 0.9937888198757764,1158 "grad_norm": 0.4746135473251343,1159 "learning_rate": 5.301892487111431e-06,1160 "loss": 1.13,1161 "step": 1601162 },1163 {1164 "epoch": 1.0,1165 "grad_norm": 0.5038459897041321,1166 "learning_rate": 5.251623849658434e-06,1167 "loss": 1.2033,1168 "step": 1611169 },1170 {1171 "epoch": 1.0062111801242235,1172 "grad_norm": 0.4865545332431793,1173 "learning_rate": 5.201329700547077e-06,1174 "loss": 1.2215,1175 "step": 1621176 },1177 {1178 "epoch": 1.0124223602484472,1179 "grad_norm": 0.44650915265083313,1180 "learning_rate": 5.151015139004445e-06,1181 "loss": 1.1592,1182 "step": 1631183 },1184 {1185 "epoch": 1.0186335403726707,1186 "grad_norm": 0.4772714376449585,1187 "learning_rate": 5.100685266327202e-06,1188 "loss": 1.1361,1189 "step": 1641190 },1191 {1192 "epoch": 1.0186335403726707,1193 "eval_loss": 1.1316967010498047,1194 "eval_runtime": 26.8815,1195 "eval_samples_per_second": 2.939,1196 "eval_steps_per_second": 0.372,1197 "step": 1641198 },1199 {1200 "epoch": 1.0248447204968945,