dipikakhullar/olmo-code-python3-text-only
04
1{2 "best_global_step": 5900,3 "best_metric": 1.0105234384536743,4 "best_model_checkpoint": "./outputs/checkpoint-5900",5 "epoch": 2.176326530612245,6 "eval_steps": 100,7 "global_step": 6000,8 "is_hyper_param_search": false,9 "is_local_process_zero": true,10 "is_world_process_zero": true,11 "log_history": [12 {13 "epoch": 0.0007256235827664399,14 "grad_norm": 0.88671875,15 "learning_rate": 2.0000000000000003e-06,16 "loss": 1.5116,17 "step": 218 },19 {20 "epoch": 0.0014512471655328798,21 "grad_norm": 1.8046875,22 "learning_rate": 6e-06,23 "loss": 1.7699,24 "step": 425 },26 {27 "epoch": 0.0021768707482993197,28 "grad_norm": 0.7890625,29 "learning_rate": 1e-05,30 "loss": 1.4788,31 "step": 632 },33 {34 "epoch": 0.0029024943310657597,35 "grad_norm": 1.3671875,36 "learning_rate": 1.4000000000000001e-05,37 "loss": 1.6688,38 "step": 839 },40 {41 "epoch": 0.0036281179138321997,42 "grad_norm": 1.7421875,43 "learning_rate": 1.8e-05,44 "loss": 1.798,45 "step": 1046 },47 {48 "epoch": 0.004353741496598639,49 "grad_norm": 1.171875,50 "learning_rate": 2.2000000000000003e-05,51 "loss": 1.6686,52 "step": 1253 },54 {55 "epoch": 0.005079365079365079,56 "grad_norm": 1.1484375,57 "learning_rate": 2.6000000000000002e-05,58 "loss": 1.8021,59 "step": 1460 },61 {62 "epoch": 0.005804988662131519,63 "grad_norm": 0.7890625,64 "learning_rate": 3e-05,65 "loss": 1.6768,66 "step": 1667 },68 {69 "epoch": 0.006530612244897959,70 "grad_norm": 0.83203125,71 "learning_rate": 3.4000000000000007e-05,72 "loss": 1.4667,73 "step": 1874 },75 {76 "epoch": 0.0072562358276643995,77 "grad_norm": 0.98828125,78 "learning_rate": 3.8e-05,79 "loss": 1.6074,80 "step": 2081 },82 {83 "epoch": 0.00798185941043084,84 "grad_norm": 0.90234375,85 "learning_rate": 4.2e-05,86 "loss": 1.6353,87 "step": 2288 },89 {90 "epoch": 0.008707482993197279,91 "grad_norm": 0.92578125,92 "learning_rate": 4.600000000000001e-05,93 "loss": 1.5452,94 "step": 2495 },96 {97 "epoch": 0.00943310657596372,98 "grad_norm": 0.87109375,99 "learning_rate": 5e-05,100 "loss": 1.6205,101 "step": 26102 },103 {104 "epoch": 0.010158730158730159,105 "grad_norm": 0.61328125,106 "learning_rate": 5.4000000000000005e-05,107 "loss": 1.4568,108 "step": 28109 },110 {111 "epoch": 0.010884353741496598,112 "grad_norm": 0.69921875,113 "learning_rate": 5.8e-05,114 "loss": 1.4901,115 "step": 30116 },117 {118 "epoch": 0.011609977324263039,119 "grad_norm": 0.98046875,120 "learning_rate": 6.2e-05,121 "loss": 1.422,122 "step": 32123 },124 {125 "epoch": 0.012335600907029478,126 "grad_norm": 0.984375,127 "learning_rate": 6.6e-05,128 "loss": 1.406,129 "step": 34130 },131 {132 "epoch": 0.013061224489795919,133 "grad_norm": 0.69921875,134 "learning_rate": 7e-05,135 "loss": 1.2822,136 "step": 36137 },138 {139 "epoch": 0.013786848072562358,140 "grad_norm": 0.86328125,141 "learning_rate": 7.4e-05,142 "loss": 1.3677,143 "step": 38144 },145 {146 "epoch": 0.014512471655328799,147 "grad_norm": 0.78125,148 "learning_rate": 7.800000000000001e-05,149 "loss": 1.3597,150 "step": 40151 },152 {153 "epoch": 0.015238095238095238,154 "grad_norm": 0.828125,155 "learning_rate": 8.2e-05,156 "loss": 1.263,157 "step": 42158 },159 {160 "epoch": 0.01596371882086168,161 "grad_norm": 0.75390625,162 "learning_rate": 8.6e-05,163 "loss": 1.314,164 "step": 44165 },166 {167 "epoch": 0.016689342403628116,168 "grad_norm": 0.46484375,169 "learning_rate": 9e-05,170 "loss": 1.2784,171 "step": 46172 },173 {174 "epoch": 0.017414965986394557,175 "grad_norm": 0.65625,176 "learning_rate": 9.4e-05,177 "loss": 1.1945,178 "step": 48179 },180 {181 "epoch": 0.018140589569160998,182 "grad_norm": 0.5703125,183 "learning_rate": 9.8e-05,184 "loss": 1.2637,185 "step": 50186 },187 {188 "epoch": 0.01886621315192744,189 "grad_norm": 0.625,190 "learning_rate": 0.00010200000000000001,191 "loss": 1.1792,192 "step": 52193 },194 {195 "epoch": 0.019591836734693877,196 "grad_norm": 0.349609375,197 "learning_rate": 0.00010600000000000002,198 "loss": 1.1901,199 "step": 54200 },201 {202 "epoch": 0.020317460317460317,203 "grad_norm": 0.392578125,204 "learning_rate": 0.00011000000000000002,205 "loss": 1.1808,206 "step": 56207 },208 {209 "epoch": 0.02104308390022676,210 "grad_norm": 0.34765625,211 "learning_rate": 0.00011399999999999999,212 "loss": 1.134,213 "step": 58214 },215 {216 "epoch": 0.021768707482993196,217 "grad_norm": 0.326171875,218 "learning_rate": 0.000118,219 "loss": 1.1071,220 "step": 60221 },222 {223 "epoch": 0.022494331065759637,224 "grad_norm": 0.255859375,225 "learning_rate": 0.000122,226 "loss": 1.1066,227 "step": 62228 },229 {230 "epoch": 0.023219954648526078,231 "grad_norm": 0.30859375,232 "learning_rate": 0.000126,233 "loss": 1.1922,234 "step": 64235 },236 {237 "epoch": 0.02394557823129252,238 "grad_norm": 0.2578125,239 "learning_rate": 0.00013000000000000002,240 "loss": 1.1028,241 "step": 66242 },243 {244 "epoch": 0.024671201814058956,245 "grad_norm": 0.369140625,246 "learning_rate": 0.000134,247 "loss": 1.0819,248 "step": 68249 },250 {251 "epoch": 0.025396825396825397,252 "grad_norm": 0.275390625,253 "learning_rate": 0.000138,254 "loss": 1.1744,255 "step": 70256 },257 {258 "epoch": 0.026122448979591838,259 "grad_norm": 0.26171875,260 "learning_rate": 0.000142,261 "loss": 1.0784,262 "step": 72263 },264 {265 "epoch": 0.026848072562358275,266 "grad_norm": 0.21875,267 "learning_rate": 0.000146,268 "loss": 1.0396,269 "step": 74270 },271 {272 "epoch": 0.027573696145124716,273 "grad_norm": 0.2333984375,274 "learning_rate": 0.00015000000000000001,275 "loss": 1.2014,276 "step": 76277 },278 {279 "epoch": 0.028299319727891157,280 "grad_norm": 0.23828125,281 "learning_rate": 0.000154,282 "loss": 1.1024,283 "step": 78284 },285 {286 "epoch": 0.029024943310657598,287 "grad_norm": 0.28125,288 "learning_rate": 0.00015800000000000002,289 "loss": 0.9752,290 "step": 80291 },292 {293 "epoch": 0.029750566893424035,294 "grad_norm": 0.2373046875,295 "learning_rate": 0.000162,296 "loss": 1.0902,297 "step": 82298 },299 {300 "epoch": 0.030476190476190476,301 "grad_norm": 0.19921875,302 "learning_rate": 0.000166,303 "loss": 1.0058,304 "step": 84305 },306 {307 "epoch": 0.031201814058956917,308 "grad_norm": 0.2197265625,309 "learning_rate": 0.00017,310 "loss": 1.0682,311 "step": 86312 },313 {314 "epoch": 0.03192743764172336,315 "grad_norm": 0.2119140625,316 "learning_rate": 0.000174,317 "loss": 1.0874,318 "step": 88319 },320 {321 "epoch": 0.0326530612244898,322 "grad_norm": 0.458984375,323 "learning_rate": 0.00017800000000000002,324 "loss": 1.0739,325 "step": 90326 },327 {328 "epoch": 0.03337868480725623,329 "grad_norm": 0.212890625,330 "learning_rate": 0.000182,331 "loss": 1.0259,332 "step": 92333 },334 {335 "epoch": 0.034104308390022674,336 "grad_norm": 0.2470703125,337 "learning_rate": 0.00018600000000000002,338 "loss": 1.0279,339 "step": 94340 },341 {342 "epoch": 0.034829931972789115,343 "grad_norm": 0.2001953125,344 "learning_rate": 0.00019,345 "loss": 1.0263,346 "step": 96347 },348 {349 "epoch": 0.035555555555555556,350 "grad_norm": 0.251953125,351 "learning_rate": 0.000194,352 "loss": 1.0206,353 "step": 98354 },355 {356 "epoch": 0.036281179138321996,357 "grad_norm": 0.2353515625,358 "learning_rate": 0.00019800000000000002,359 "loss": 1.1275,360 "step": 100361 },362 {363 "epoch": 0.036281179138321996,364 "eval_loss": 1.0672887563705444,365 "eval_runtime": 99.6091,366 "eval_samples_per_second": 18.071,367 "eval_steps_per_second": 0.572,368 "step": 100369 },370 {371 "epoch": 0.03700680272108844,372 "grad_norm": 0.267578125,373 "learning_rate": 0.0001999755231917758,374 "loss": 1.1141,375 "step": 102376 },377 {378 "epoch": 0.03773242630385488,379 "grad_norm": 0.2080078125,380 "learning_rate": 0.00019992656957532737,381 "loss": 1.0543,382 "step": 104383 },384 {385 "epoch": 0.03845804988662131,386 "grad_norm": 0.20703125,387 "learning_rate": 0.00019987761595887898,388 "loss": 1.0389,389 "step": 106390 },391 {392 "epoch": 0.03918367346938775,393 "grad_norm": 0.1982421875,394 "learning_rate": 0.00019982866234243056,395 "loss": 1.0699,396 "step": 108397 },398 {399 "epoch": 0.039909297052154194,400 "grad_norm": 0.20703125,401 "learning_rate": 0.00019977970872598214,402 "loss": 1.0756,403 "step": 110404 },405 {406 "epoch": 0.040634920634920635,407 "grad_norm": 0.1845703125,408 "learning_rate": 0.00019973075510953373,409 "loss": 1.0309,410 "step": 112411 },412 {413 "epoch": 0.041360544217687076,414 "grad_norm": 0.1884765625,415 "learning_rate": 0.0001996818014930853,416 "loss": 0.9821,417 "step": 114418 },419 {420 "epoch": 0.04208616780045352,421 "grad_norm": 0.2353515625,422 "learning_rate": 0.0001996328478766369,423 "loss": 1.0897,424 "step": 116425 },426 {427 "epoch": 0.04281179138321996,428 "grad_norm": 0.2080078125,429 "learning_rate": 0.0001995838942601885,430 "loss": 1.0576,431 "step": 118432 },433 {434 "epoch": 0.04353741496598639,435 "grad_norm": 0.2001953125,436 "learning_rate": 0.00019953494064374008,437 "loss": 1.0482,438 "step": 120439 },440 {441 "epoch": 0.04426303854875283,442 "grad_norm": 0.259765625,443 "learning_rate": 0.00019948598702729164,444 "loss": 1.0018,445 "step": 122446 },447 {448 "epoch": 0.04498866213151927,449 "grad_norm": 0.1865234375,450 "learning_rate": 0.00019943703341084322,451 "loss": 1.0637,452 "step": 124453 },454 {455 "epoch": 0.045714285714285714,456 "grad_norm": 0.1806640625,457 "learning_rate": 0.00019938807979439483,458 "loss": 1.0471,459 "step": 126460 },461 {462 "epoch": 0.046439909297052155,463 "grad_norm": 0.1875,464 "learning_rate": 0.0001993391261779464,465 "loss": 0.9958,466 "step": 128467 },468 {469 "epoch": 0.047165532879818596,470 "grad_norm": 0.201171875,471 "learning_rate": 0.000199290172561498,472 "loss": 1.0396,473 "step": 130474 },475 {476 "epoch": 0.04789115646258504,477 "grad_norm": 0.2333984375,478 "learning_rate": 0.00019924121894504958,479 "loss": 1.0407,480 "step": 132481 },482 {483 "epoch": 0.04861678004535147,484 "grad_norm": 0.201171875,485 "learning_rate": 0.00019919226532860116,486 "loss": 1.043,487 "step": 134488 },489 {490 "epoch": 0.04934240362811791,491 "grad_norm": 0.193359375,492 "learning_rate": 0.00019914331171215274,493 "loss": 1.0859,494 "step": 136495 },496 {497 "epoch": 0.05006802721088435,498 "grad_norm": 0.181640625,499 "learning_rate": 0.00019909435809570435,500 "loss": 1.1251,501 "step": 138502 },503 {504 "epoch": 0.050793650793650794,505 "grad_norm": 0.2431640625,506 "learning_rate": 0.0001990454044792559,507 "loss": 1.0666,508 "step": 140509 },510 {511 "epoch": 0.051519274376417235,512 "grad_norm": 0.203125,513 "learning_rate": 0.0001989964508628075,514 "loss": 0.9962,515 "step": 142516 },517 {518 "epoch": 0.052244897959183675,519 "grad_norm": 0.1923828125,520 "learning_rate": 0.00019894749724635907,521 "loss": 1.1262,522 "step": 144523 },524 {525 "epoch": 0.052970521541950116,526 "grad_norm": 0.1875,527 "learning_rate": 0.00019889854362991068,528 "loss": 1.0174,529 "step": 146530 },531 {532 "epoch": 0.05369614512471655,533 "grad_norm": 0.181640625,534 "learning_rate": 0.00019884959001346226,535 "loss": 1.0509,536 "step": 148537 },538 {539 "epoch": 0.05442176870748299,540 "grad_norm": 0.21875,541 "learning_rate": 0.00019880063639701382,542 "loss": 1.05,543 "step": 150544 },545 {546 "epoch": 0.05514739229024943,547 "grad_norm": 0.2109375,548 "learning_rate": 0.00019875168278056543,549 "loss": 1.134,550 "step": 152551 },552 {553 "epoch": 0.05587301587301587,554 "grad_norm": 0.1650390625,555 "learning_rate": 0.000198702729164117,556 "loss": 0.9511,557 "step": 154558 },559 {560 "epoch": 0.056598639455782314,561 "grad_norm": 0.1748046875,562 "learning_rate": 0.0001986537755476686,563 "loss": 0.9989,564 "step": 156565 },566 {567 "epoch": 0.057324263038548755,568 "grad_norm": 0.162109375,569 "learning_rate": 0.0001986048219312202,570 "loss": 0.9466,571 "step": 158572 },573 {574 "epoch": 0.058049886621315196,575 "grad_norm": 0.162109375,576 "learning_rate": 0.00019855586831477175,577 "loss": 0.9605,578 "step": 160579 },580 {581 "epoch": 0.05877551020408163,582 "grad_norm": 0.2001953125,583 "learning_rate": 0.00019850691469832334,584 "loss": 1.0511,585 "step": 162586 },587 {588 "epoch": 0.05950113378684807,589 "grad_norm": 0.205078125,590 "learning_rate": 0.00019845796108187492,591 "loss": 1.0575,592 "step": 164593 },594 {595 "epoch": 0.06022675736961451,596 "grad_norm": 0.189453125,597 "learning_rate": 0.00019840900746542653,598 "loss": 1.0211,599 "step": 166600 },601 {602 "epoch": 0.06095238095238095,603 "grad_norm": 0.1923828125,604 "learning_rate": 0.0001983600538489781,605 "loss": 1.0435,606 "step": 168607 },608 {609 "epoch": 0.06167800453514739,610 "grad_norm": 0.2021484375,611 "learning_rate": 0.00019831110023252967,612 "loss": 0.9651,613 "step": 170614 },615 {616 "epoch": 0.062403628117913834,617 "grad_norm": 0.17578125,618 "learning_rate": 0.00019826214661608128,619 "loss": 0.9683,620 "step": 172621 },622 {623 "epoch": 0.06312925170068028,624 "grad_norm": 0.1796875,625 "learning_rate": 0.00019821319299963286,626 "loss": 1.0064,627 "step": 174628 },629 {630 "epoch": 0.06385487528344672,631 "grad_norm": 0.1875,632 "learning_rate": 0.00019816423938318444,633 "loss": 1.0669,634 "step": 176635 },636 {637 "epoch": 0.06458049886621316,638 "grad_norm": 0.1796875,639 "learning_rate": 0.00019811528576673605,640 "loss": 1.0931,641 "step": 178642 },643 {644 "epoch": 0.0653061224489796,645 "grad_norm": 0.19921875,646 "learning_rate": 0.0001980663321502876,647 "loss": 1.04,648 "step": 180649 },650 {651 "epoch": 0.06603174603174604,652 "grad_norm": 0.1884765625,653 "learning_rate": 0.0001980173785338392,654 "loss": 0.9917,655 "step": 182656 },657 {658 "epoch": 0.06675736961451247,659 "grad_norm": 0.166015625,660 "learning_rate": 0.00019796842491739077,661 "loss": 0.9985,662 "step": 184663 },664 {665 "epoch": 0.0674829931972789,666 "grad_norm": 0.1728515625,667 "learning_rate": 0.00019791947130094238,668 "loss": 0.9511,669 "step": 186670 },671 {672 "epoch": 0.06820861678004535,673 "grad_norm": 0.169921875,674 "learning_rate": 0.00019787051768449396,675 "loss": 0.9866,676 "step": 188677 },678 {679 "epoch": 0.06893424036281179,680 "grad_norm": 0.1982421875,681 "learning_rate": 0.00019782156406804552,682 "loss": 1.0879,683 "step": 190684 },685 {686 "epoch": 0.06965986394557823,687 "grad_norm": 0.181640625,688 "learning_rate": 0.00019777261045159713,689 "loss": 1.006,690 "step": 192691 },692 {693 "epoch": 0.07038548752834467,694 "grad_norm": 0.1787109375,695 "learning_rate": 0.0001977236568351487,696 "loss": 1.0737,697 "step": 194698 },699 {700 "epoch": 0.07111111111111111,701 "grad_norm": 0.203125,702 "learning_rate": 0.0001976747032187003,703 "loss": 0.9719,704 "step": 196705 },706 {707 "epoch": 0.07183673469387755,708 "grad_norm": 0.21484375,709 "learning_rate": 0.0001976257496022519,710 "loss": 1.0539,711 "step": 198712 },713 {714 "epoch": 0.07256235827664399,715 "grad_norm": 0.19921875,716 "learning_rate": 0.00019757679598580345,717 "loss": 1.0863,718 "step": 200719 },720 {721 "epoch": 0.07256235827664399,722 "eval_loss": 1.0431190729141235,723 "eval_runtime": 99.2841,724 "eval_samples_per_second": 18.13,725 "eval_steps_per_second": 0.574,726 "step": 200727 },728 {729 "epoch": 0.07328798185941043,730 "grad_norm": 0.2021484375,731 "learning_rate": 0.00019752784236935504,732 "loss": 1.1022,733 "step": 202734 },735 {736 "epoch": 0.07401360544217687,737 "grad_norm": 0.177734375,738 "learning_rate": 0.00019747888875290662,739 "loss": 0.9781,740 "step": 204741 },742 {743 "epoch": 0.07473922902494332,744 "grad_norm": 0.21875,745 "learning_rate": 0.00019742993513645823,746 "loss": 1.0735,747 "step": 206748 },749 {750 "epoch": 0.07546485260770976,751 "grad_norm": 0.1865234375,752 "learning_rate": 0.0001973809815200098,753 "loss": 0.957,754 "step": 208755 },756 {757 "epoch": 0.0761904761904762,758 "grad_norm": 0.158203125,759 "learning_rate": 0.00019733202790356137,760 "loss": 1.0229,761 "step": 210762 },763 {764 "epoch": 0.07691609977324262,765 "grad_norm": 0.1962890625,766 "learning_rate": 0.00019728307428711298,767 "loss": 0.9876,768 "step": 212769 },770 {771 "epoch": 0.07764172335600907,772 "grad_norm": 0.1884765625,773 "learning_rate": 0.00019723412067066456,774 "loss": 1.0047,775 "step": 214776 },777 {778 "epoch": 0.0783673469387755,779 "grad_norm": 0.1875,780 "learning_rate": 0.00019718516705421614,781 "loss": 0.9995,782 "step": 216783 },784 {785 "epoch": 0.07909297052154195,786 "grad_norm": 0.1923828125,787 "learning_rate": 0.00019713621343776772,788 "loss": 1.0447,789 "step": 218790 },791 {792 "epoch": 0.07981859410430839,793 "grad_norm": 0.193359375,794 "learning_rate": 0.0001970872598213193,795 "loss": 1.0969,796 "step": 220797 },798 {799 "epoch": 0.08054421768707483,800 "grad_norm": 0.203125,801 "learning_rate": 0.0001970383062048709,802 "loss": 0.9948,803 "step": 222804 },805 {806 "epoch": 0.08126984126984127,807 "grad_norm": 0.1591796875,808 "learning_rate": 0.00019698935258842247,809 "loss": 0.9642,810 "step": 224811 },812 {813 "epoch": 0.08199546485260771,814 "grad_norm": 0.19140625,815 "learning_rate": 0.00019694039897197408,816 "loss": 1.0414,817 "step": 226818 },819 {820 "epoch": 0.08272108843537415,821 "grad_norm": 0.1943359375,822 "learning_rate": 0.00019689144535552563,823 "loss": 0.9293,824 "step": 228825 },826 {827 "epoch": 0.08344671201814059,828 "grad_norm": 0.2021484375,829 "learning_rate": 0.00019684249173907722,830 "loss": 0.9743,831 "step": 230832 },833 {834 "epoch": 0.08417233560090703,835 "grad_norm": 0.189453125,836 "learning_rate": 0.00019679353812262883,837 "loss": 1.0206,838 "step": 232839 },840 {841 "epoch": 0.08489795918367347,842 "grad_norm": 0.1787109375,843 "learning_rate": 0.0001967445845061804,844 "loss": 0.9956,845 "step": 234846 },847 {848 "epoch": 0.08562358276643992,849 "grad_norm": 0.185546875,850 "learning_rate": 0.000196695630889732,851 "loss": 1.0204,852 "step": 236853 },854 {855 "epoch": 0.08634920634920636,856 "grad_norm": 0.177734375,857 "learning_rate": 0.00019664667727328357,858 "loss": 0.9644,859 "step": 238860 },861 {862 "epoch": 0.08707482993197278,863 "grad_norm": 0.1748046875,864 "learning_rate": 0.00019659772365683515,865 "loss": 0.9574,866 "step": 240867 },868 {869 "epoch": 0.08780045351473922,870 "grad_norm": 0.181640625,871 "learning_rate": 0.00019654877004038674,872 "loss": 1.0699,873 "step": 242874 },875 {876 "epoch": 0.08852607709750566,877 "grad_norm": 0.193359375,878 "learning_rate": 0.00019649981642393832,879 "loss": 1.0396,880 "step": 244881 },882 {883 "epoch": 0.0892517006802721,884 "grad_norm": 0.1611328125,885 "learning_rate": 0.00019645086280748993,886 "loss": 0.9685,887 "step": 246888 },889 {890 "epoch": 0.08997732426303855,891 "grad_norm": 0.162109375,892 "learning_rate": 0.00019640190919104148,893 "loss": 0.9963,894 "step": 248895 },896 {897 "epoch": 0.09070294784580499,898 "grad_norm": 0.181640625,899 "learning_rate": 0.00019635295557459307,900 "loss": 1.0085,901 "step": 250902 },903 {904 "epoch": 0.09142857142857143,905 "grad_norm": 0.216796875,906 "learning_rate": 0.00019630400195814468,907 "loss": 1.0444,908 "step": 252909 },910 {911 "epoch": 0.09215419501133787,912 "grad_norm": 0.1591796875,913 "learning_rate": 0.00019625504834169626,914 "loss": 0.9959,915 "step": 254916 },917 {918 "epoch": 0.09287981859410431,919 "grad_norm": 0.189453125,920 "learning_rate": 0.00019620609472524784,921 "loss": 1.0705,922 "step": 256923 },924 {925 "epoch": 0.09360544217687075,926 "grad_norm": 0.1796875,927 "learning_rate": 0.00019615714110879942,928 "loss": 1.0762,929 "step": 258930 },931 {932 "epoch": 0.09433106575963719,933 "grad_norm": 0.1826171875,934 "learning_rate": 0.000196108187492351,935 "loss": 1.0227,936 "step": 260937 },938 {939 "epoch": 0.09505668934240363,940 "grad_norm": 0.2080078125,941 "learning_rate": 0.00019605923387590259,942 "loss": 1.0503,943 "step": 262944 },945 {946 "epoch": 0.09578231292517007,947 "grad_norm": 0.1796875,948 "learning_rate": 0.00019601028025945417,949 "loss": 1.0013,950 "step": 264951 },952 {953 "epoch": 0.09650793650793651,954 "grad_norm": 0.1689453125,955 "learning_rate": 0.00019596132664300578,956 "loss": 1.0844,957 "step": 266958 },959 {960 "epoch": 0.09723356009070294,961 "grad_norm": 0.1767578125,962 "learning_rate": 0.00019591237302655733,963 "loss": 0.9824,964 "step": 268965 },966 {967 "epoch": 0.09795918367346938,968 "grad_norm": 0.1533203125,969 "learning_rate": 0.00019586341941010892,970 "loss": 0.9688,971 "step": 270972 },973 {974 "epoch": 0.09868480725623582,975 "grad_norm": 0.169921875,976 "learning_rate": 0.00019581446579366052,977 "loss": 0.923,978 "step": 272979 },980 {981 "epoch": 0.09941043083900226,982 "grad_norm": 0.2197265625,983 "learning_rate": 0.0001957655121772121,984 "loss": 0.9817,985 "step": 274986 },987 {988 "epoch": 0.1001360544217687,989 "grad_norm": 0.1728515625,990 "learning_rate": 0.0001957165585607637,991 "loss": 0.9691,992 "step": 276993 },994 {995 "epoch": 0.10086167800453515,996 "grad_norm": 0.1806640625,997 "learning_rate": 0.00019566760494431527,998 "loss": 1.019,999 "step": 2781000 },1001 {1002 "epoch": 0.10158730158730159,1003 "grad_norm": 0.1767578125,1004 "learning_rate": 0.00019561865132786685,1005 "loss": 1.008,1006 "step": 2801007 },1008 {1009 "epoch": 0.10231292517006803,1010 "grad_norm": 0.162109375,1011 "learning_rate": 0.00019556969771141844,1012 "loss": 0.9477,1013 "step": 2821014 },1015 {1016 "epoch": 0.10303854875283447,1017 "grad_norm": 0.1748046875,1018 "learning_rate": 0.00019552074409497002,1019 "loss": 0.962,1020 "step": 2841021 },1022 {1023 "epoch": 0.10376417233560091,1024 "grad_norm": 0.169921875,1025 "learning_rate": 0.0001954717904785216,1026 "loss": 0.9685,1027 "step": 2861028 },1029 {1030 "epoch": 0.10448979591836735,1031 "grad_norm": 0.1650390625,1032 "learning_rate": 0.00019542283686207318,1033 "loss": 1.015,1034 "step": 2881035 },1036 {1037 "epoch": 0.10521541950113379,1038 "grad_norm": 0.166015625,1039 "learning_rate": 0.00019537388324562477,1040 "loss": 0.9902,1041 "step": 2901042 },1043 {1044 "epoch": 0.10594104308390023,1045 "grad_norm": 0.16796875,1046 "learning_rate": 0.00019532492962917637,1047 "loss": 1.0309,1048 "step": 2921049 },1050 {1051 "epoch": 0.10666666666666667,1052 "grad_norm": 0.1953125,1053 "learning_rate": 0.00019527597601272796,1054 "loss": 1.0179,1055 "step": 2941056 },1057 {1058 "epoch": 0.1073922902494331,1059 "grad_norm": 0.16796875,1060 "learning_rate": 0.0001952270223962795,1061 "loss": 1.0208,1062 "step": 2961063 },1064 {1065 "epoch": 0.10811791383219954,1066 "grad_norm": 0.16796875,1067 "learning_rate": 0.00019517806877983112,1068 "loss": 1.0252,1069 "step": 2981070 },1071 {1072 "epoch": 0.10884353741496598,1073 "grad_norm": 0.171875,1074 "learning_rate": 0.0001951291151633827,1075 "loss": 1.0185,1076 "step": 3001077 },1078 {1079 "epoch": 0.10884353741496598,1080 "eval_loss": 1.036349892616272,1081 "eval_runtime": 91.1673,1082 "eval_samples_per_second": 19.744,1083 "eval_steps_per_second": 0.625,1084 "step": 3001085 },1086 {1087 "epoch": 0.10956916099773242,1088 "grad_norm": 0.177734375,1089 "learning_rate": 0.00019508016154693429,1090 "loss": 0.9886,1091 "step": 3021092 },1093 {1094 "epoch": 0.11029478458049886,1095 "grad_norm": 0.1689453125,1096 "learning_rate": 0.00019503120793048587,1097 "loss": 1.0495,1098 "step": 3041099 },1100 {1101 "epoch": 0.1110204081632653,1102 "grad_norm": 0.1533203125,1103 "learning_rate": 0.00019498225431403745,1104 "loss": 1.0057,1105 "step": 3061106 },1107 {1108 "epoch": 0.11174603174603175,1109 "grad_norm": 0.1728515625,1110 "learning_rate": 0.00019493330069758903,1111 "loss": 1.042,1112 "step": 3081113 },1114 {1115 "epoch": 0.11247165532879819,1116 "grad_norm": 0.1767578125,1117 "learning_rate": 0.00019488434708114062,1118 "loss": 1.0619,1119 "step": 3101120 },1121 {1122 "epoch": 0.11319727891156463,1123 "grad_norm": 0.150390625,1124 "learning_rate": 0.00019483539346469222,1125 "loss": 1.0055,1126 "step": 3121127 },1128 {1129 "epoch": 0.11392290249433107,1130 "grad_norm": 0.1826171875,1131 "learning_rate": 0.0001947864398482438,1132 "loss": 0.9609,1133 "step": 3141134 },1135 {1136 "epoch": 0.11464852607709751,1137 "grad_norm": 0.1689453125,1138 "learning_rate": 0.00019473748623179536,1139 "loss": 1.0478,1140 "step": 3161141 },1142 {1143 "epoch": 0.11537414965986395,1144 "grad_norm": 0.1826171875,1145 "learning_rate": 0.00019468853261534697,1146 "loss": 1.0008,1147 "step": 3181148 },1149 {1150 "epoch": 0.11609977324263039,1151 "grad_norm": 0.15625,1152 "learning_rate": 0.00019463957899889855,1153 "loss": 0.9464,1154 "step": 3201155 },1156 {1157 "epoch": 0.11682539682539683,1158 "grad_norm": 0.16015625,1159 "learning_rate": 0.00019459062538245014,1160 "loss": 1.0368,1161 "step": 3221162 },1163 {1164 "epoch": 0.11755102040816326,1165 "grad_norm": 0.171875,1166 "learning_rate": 0.00019454167176600172,1167 "loss": 1.0134,1168 "step": 3241169 },1170 {1171 "epoch": 0.1182766439909297,1172 "grad_norm": 0.15625,1173 "learning_rate": 0.0001944927181495533,1174 "loss": 0.9533,1175 "step": 3261176 },1177 {1178 "epoch": 0.11900226757369614,1179 "grad_norm": 0.1689453125,1180 "learning_rate": 0.00019444376453310488,1181 "loss": 0.9526,1182 "step": 3281183 },1184 {1185 "epoch": 0.11972789115646258,1186 "grad_norm": 0.18359375,1187 "learning_rate": 0.00019439481091665647,1188 "loss": 0.9875,1189 "step": 3301190 },1191 {1192 "epoch": 0.12045351473922902,1193 "grad_norm": 0.1787109375,1194 "learning_rate": 0.00019434585730020807,1195 "loss": 1.0052,1196 "step": 3321197 },1198 {1199 "epoch": 0.12117913832199546,1200 "grad_norm": 0.1943359375,