golaxy/ReDI_Decomposition
07
1{2 "best_global_step": null,3 "best_metric": null,4 "best_model_checkpoint": null,5 "epoch": 3.0,6 "eval_steps": 500,7 "global_step": 1278,8 "is_hyper_param_search": false,9 "is_local_process_zero": true,10 "is_world_process_zero": true,11 "log_history": [12 {13 "epoch": 0.023508668821627974,14 "grad_norm": 84.37700653076172,15 "learning_rate": 7.031250000000001e-06,16 "loss": 115.85,17 "step": 1018 },19 {20 "epoch": 0.04701733764325595,21 "grad_norm": 106.4520034790039,22 "learning_rate": 1.484375e-05,23 "loss": 116.5781,24 "step": 2025 },26 {27 "epoch": 0.07052600646488393,28 "grad_norm": 130.89254760742188,29 "learning_rate": 2.2656250000000002e-05,30 "loss": 103.3281,31 "step": 3032 },33 {34 "epoch": 0.0940346752865119,35 "grad_norm": 37.089256286621094,36 "learning_rate": 3.0468750000000002e-05,37 "loss": 60.0406,38 "step": 4039 },40 {41 "epoch": 0.11754334410813988,42 "grad_norm": 44.051109313964844,43 "learning_rate": 3.828125e-05,44 "loss": 39.75,45 "step": 5046 },47 {48 "epoch": 0.14105201292976785,49 "grad_norm": 43.561981201171875,50 "learning_rate": 4.609375e-05,51 "loss": 24.7984,52 "step": 6053 },54 {55 "epoch": 0.16456068175139582,56 "grad_norm": 34.622520446777344,57 "learning_rate": 5.3906250000000006e-05,58 "loss": 11.9723,59 "step": 7060 },61 {62 "epoch": 0.1880693505730238,63 "grad_norm": 22.26849937438965,64 "learning_rate": 6.171875e-05,65 "loss": 5.4578,66 "step": 8067 },68 {69 "epoch": 0.2115780193946518,70 "grad_norm": 9.059814453125,71 "learning_rate": 6.953125e-05,72 "loss": 3.992,73 "step": 9074 },75 {76 "epoch": 0.23508668821627976,77 "grad_norm": 4.332883834838867,78 "learning_rate": 7.734375e-05,79 "loss": 3.1058,80 "step": 10081 },82 {83 "epoch": 0.2585953570379077,84 "grad_norm": 6.844908237457275,85 "learning_rate": 8.515625e-05,86 "loss": 2.6654,87 "step": 11088 },89 {90 "epoch": 0.2821040258595357,91 "grad_norm": 4.539117813110352,92 "learning_rate": 9.296875e-05,93 "loss": 2.422,94 "step": 12095 },96 {97 "epoch": 0.3056126946811637,98 "grad_norm": 3.098025321960449,99 "learning_rate": 9.999981342914437e-05,100 "loss": 2.1623,101 "step": 130102 },103 {104 "epoch": 0.32912136350279164,105 "grad_norm": 1.3452321290969849,106 "learning_rate": 9.997742661115932e-05,107 "loss": 2.1429,108 "step": 140109 },110 {111 "epoch": 0.35263003232441964,112 "grad_norm": 0.37226417660713196,113 "learning_rate": 9.991774476447404e-05,114 "loss": 2.1347,115 "step": 150116 },117 {118 "epoch": 0.3761387011460476,119 "grad_norm": 3.523218870162964,120 "learning_rate": 9.982081242591919e-05,121 "loss": 2.0512,122 "step": 160123 },124 {125 "epoch": 0.3996473699676756,126 "grad_norm": 0.33414289355278015,127 "learning_rate": 9.968670193003843e-05,128 "loss": 2.047,129 "step": 170130 },131 {132 "epoch": 0.4231560387893036,133 "grad_norm": 0.24106919765472412,134 "learning_rate": 9.951551335510978e-05,135 "loss": 2.1046,136 "step": 180137 },138 {139 "epoch": 0.4466647076109315,140 "grad_norm": 0.328752726316452,141 "learning_rate": 9.930737444846331e-05,142 "loss": 2.1397,143 "step": 190144 },145 {146 "epoch": 0.4701733764325595,147 "grad_norm": 1.1409285068511963,148 "learning_rate": 9.906244053115143e-05,149 "loss": 2.1377,150 "step": 200151 },152 {153 "epoch": 0.49368204525418746,154 "grad_norm": 0.0690290704369545,155 "learning_rate": 9.87808943820424e-05,156 "loss": 2.0992,157 "step": 210158 },159 {160 "epoch": 0.5171907140758154,161 "grad_norm": 0.20527280867099762,162 "learning_rate": 9.846294610142398e-05,163 "loss": 2.1339,164 "step": 220165 },166 {167 "epoch": 0.5406993828974435,168 "grad_norm": 0.3991909623146057,169 "learning_rate": 9.810883295421864e-05,170 "loss": 2.1176,171 "step": 230172 },173 {174 "epoch": 0.5642080517190714,175 "grad_norm": 0.08950258791446686,176 "learning_rate": 9.771881919292765e-05,177 "loss": 2.0895,178 "step": 240179 },180 {181 "epoch": 0.5877167205406993,182 "grad_norm": 0.17848969995975494,183 "learning_rate": 9.729319586043591e-05,184 "loss": 2.0877,185 "step": 250186 },187 {188 "epoch": 0.6112253893623274,189 "grad_norm": 0.30160897970199585,190 "learning_rate": 9.683228057282483e-05,191 "loss": 2.0648,192 "step": 260193 },194 {195 "epoch": 0.6347340581839553,196 "grad_norm": 0.11883804202079773,197 "learning_rate": 9.63364172823554e-05,198 "loss": 2.0591,199 "step": 270200 },201 {202 "epoch": 0.6582427270055833,203 "grad_norm": 0.1619461327791214,204 "learning_rate": 9.580597602079802e-05,205 "loss": 2.1386,206 "step": 280207 },208 {209 "epoch": 0.6817513958272113,210 "grad_norm": 0.11473017930984497,211 "learning_rate": 9.524135262330098e-05,212 "loss": 2.1028,213 "step": 290214 },215 {216 "epoch": 0.7052600646488393,217 "grad_norm": 0.059143248945474625,218 "learning_rate": 9.464296843300342e-05,219 "loss": 2.0881,220 "step": 300221 },222 {223 "epoch": 0.7287687334704672,224 "grad_norm": 0.1624162495136261,225 "learning_rate": 9.401126998661328e-05,226 "loss": 2.0971,227 "step": 310228 },229 {230 "epoch": 0.7522774022920952,231 "grad_norm": 0.15460653603076935,232 "learning_rate": 9.334672868118491e-05,233 "loss": 2.1154,234 "step": 320235 },236 {237 "epoch": 0.7757860711137232,238 "grad_norm": 0.4213317334651947,239 "learning_rate": 9.26498404223449e-05,240 "loss": 2.1113,241 "step": 330242 },243 {244 "epoch": 0.7992947399353512,245 "grad_norm": 0.14529550075531006,246 "learning_rate": 9.192112525422868e-05,247 "loss": 2.0867,248 "step": 340249 },250 {251 "epoch": 0.8228034087569791,252 "grad_norm": 0.06652400642633438,253 "learning_rate": 9.116112697140418e-05,254 "loss": 2.0807,255 "step": 350256 },257 {258 "epoch": 0.8463120775786072,259 "grad_norm": 0.12574820220470428,260 "learning_rate": 9.037041271307188e-05,261 "loss": 2.1711,262 "step": 360263 },264 {265 "epoch": 0.8698207464002351,266 "grad_norm": 0.10869150608778,267 "learning_rate": 8.954957253984426e-05,268 "loss": 2.1252,269 "step": 370270 },271 {272 "epoch": 0.893329415221863,273 "grad_norm": 0.143524169921875,274 "learning_rate": 8.869921899342056e-05,275 "loss": 2.0468,276 "step": 380277 },278 {279 "epoch": 0.916838084043491,280 "grad_norm": 0.27682727575302124,281 "learning_rate": 8.781998663948513e-05,282 "loss": 2.1023,283 "step": 390284 },285 {286 "epoch": 0.940346752865119,287 "grad_norm": 0.08671136200428009,288 "learning_rate": 8.691253159417074e-05,289 "loss": 2.0991,290 "step": 400291 },292 {293 "epoch": 0.963855421686747,294 "grad_norm": 0.14743109047412872,295 "learning_rate": 8.597753103444016e-05,296 "loss": 2.0858,297 "step": 410298 },299 {300 "epoch": 0.9873640905083749,301 "grad_norm": 0.14369799196720123,302 "learning_rate": 8.501568269275126e-05,303 "loss": 2.1057,304 "step": 420305 },306 {307 "epoch": 1.0094034675286512,308 "grad_norm": 0.09220755100250244,309 "learning_rate": 8.40277043363831e-05,310 "loss": 1.9405,311 "step": 430312 },313 {314 "epoch": 1.0329121363502791,315 "grad_norm": 0.25697803497314453,316 "learning_rate": 8.301433323181076e-05,317 "loss": 2.1221,318 "step": 440319 },320 {321 "epoch": 1.056420805171907,322 "grad_norm": 0.1321459412574768,323 "learning_rate": 8.19763255945298e-05,324 "loss": 2.1576,325 "step": 450326 },327 {328 "epoch": 1.079929473993535,329 "grad_norm": 0.11310122162103653,330 "learning_rate": 8.091445602473972e-05,331 "loss": 2.0483,332 "step": 460333 },334 {335 "epoch": 1.1034381428151632,336 "grad_norm": 0.08170254528522491,337 "learning_rate": 7.982951692930829e-05,338 "loss": 2.1367,339 "step": 470340 },341 {342 "epoch": 1.1269468116367911,343 "grad_norm": 0.022492246702313423,344 "learning_rate": 7.87223179304479e-05,345 "loss": 2.1435,346 "step": 480347 },348 {349 "epoch": 1.150455480458419,350 "grad_norm": 0.015893638134002686,351 "learning_rate": 7.759368526154509e-05,352 "loss": 2.0666,353 "step": 490354 },355 {356 "epoch": 1.173964149280047,357 "grad_norm": 0.053727954626083374,358 "learning_rate": 7.644446115059425e-05,359 "loss": 2.0886,360 "step": 500361 },362 {363 "epoch": 1.197472818101675,364 "grad_norm": 0.12717482447624207,365 "learning_rate": 7.527550319169546e-05,366 "loss": 2.1418,367 "step": 510368 },369 {370 "epoch": 1.2209814869233029,371 "grad_norm": 0.030752327293157578,372 "learning_rate": 7.408768370508576e-05,373 "loss": 2.0688,374 "step": 520375 },376 {377 "epoch": 1.244490155744931,378 "grad_norm": 0.07915918529033661,379 "learning_rate": 7.288188908618102e-05,380 "loss": 2.0857,381 "step": 530382 },383 {384 "epoch": 1.267998824566559,385 "grad_norm": 0.09853250533342361,386 "learning_rate": 7.165901914411435e-05,387 "loss": 2.0392,388 "step": 540389 },390 {391 "epoch": 1.291507493388187,392 "grad_norm": 0.21571961045265198,393 "learning_rate": 7.041998643026511e-05,394 "loss": 2.0921,395 "step": 550396 },397 {398 "epoch": 1.3150161622098149,399 "grad_norm": 0.10290564596652985,400 "learning_rate": 6.916571555727852e-05,401 "loss": 2.0686,402 "step": 560403 },404 {405 "epoch": 1.3385248310314428,406 "grad_norm": 0.08689925819635391,407 "learning_rate": 6.789714250908533e-05,408 "loss": 2.0845,409 "step": 570410 },411 {412 "epoch": 1.3620334998530708,413 "grad_norm": 0.1267048567533493,414 "learning_rate": 6.661521394243533e-05,415 "loss": 2.1465,416 "step": 580417 },418 {419 "epoch": 1.3855421686746987,420 "grad_norm": 0.18292003870010376,421 "learning_rate": 6.532088648046677e-05,422 "loss": 2.067,423 "step": 590424 },425 {426 "epoch": 1.4090508374963266,427 "grad_norm": 0.051083244383335114,428 "learning_rate": 6.401512599883818e-05,429 "loss": 2.0697,430 "step": 600431 },432 {433 "epoch": 1.4325595063179548,434 "grad_norm": 0.0492679663002491,435 "learning_rate": 6.269890690495593e-05,436 "loss": 2.1064,437 "step": 610438 },439 {440 "epoch": 1.4560681751395828,441 "grad_norm": 0.051078781485557556,442 "learning_rate": 6.137321141083468e-05,443 "loss": 2.0533,444 "step": 620445 },446 {447 "epoch": 1.4795768439612107,448 "grad_norm": 0.11715124547481537,449 "learning_rate": 6.003902880013416e-05,450 "loss": 2.1156,451 "step": 630452 },453 {454 "epoch": 1.5030855127828386,455 "grad_norm": 0.041975561529397964,456 "learning_rate": 5.869735468991854e-05,457 "loss": 2.0543,458 "step": 640459 },460 {461 "epoch": 1.5265941816044668,462 "grad_norm": 0.10740803927183151,463 "learning_rate": 5.73491902876897e-05,464 "loss": 2.0539,465 "step": 650466 },467 {468 "epoch": 1.5501028504260947,469 "grad_norm": 0.07230595499277115,470 "learning_rate": 5.599554164424859e-05,471 "loss": 2.1078,472 "step": 660473 },474 {475 "epoch": 1.5736115192477227,476 "grad_norm": 0.010545202530920506,477 "learning_rate": 5.463741890294257e-05,478 "loss": 2.1434,479 "step": 670480 },481 {482 "epoch": 1.5971201880693506,483 "grad_norm": 0.0882532000541687,484 "learning_rate": 5.327583554585842e-05,485 "loss": 2.0696,486 "step": 680487 },488 {489 "epoch": 1.6206288568909786,490 "grad_norm": 0.15730935335159302,491 "learning_rate": 5.191180763752418e-05,492 "loss": 2.0743,493 "step": 690494 },495 {496 "epoch": 1.6441375257126065,497 "grad_norm": 0.16760171949863434,498 "learning_rate": 5.054635306668355e-05,499 "loss": 2.1303,500 "step": 700501 },502 {503 "epoch": 1.6676461945342345,504 "grad_norm": 0.12746253609657288,505 "learning_rate": 4.9180490786709335e-05,506 "loss": 2.1359,507 "step": 710508 },509 {510 "epoch": 1.6911548633558624,511 "grad_norm": 0.08576823770999908,512 "learning_rate": 4.781524005522216e-05,513 "loss": 2.0884,514 "step": 720515 },516 {517 "epoch": 1.7146635321774903,518 "grad_norm": 0.01966356858611107,519 "learning_rate": 4.645161967348223e-05,520 "loss": 2.1102,521 "step": 730522 },523 {524 "epoch": 1.7381722009991183,525 "grad_norm": 0.1140172928571701,526 "learning_rate": 4.509064722612169e-05,527 "loss": 2.134,528 "step": 740529 },530 {531 "epoch": 1.7616808698207465,532 "grad_norm": 0.11263474822044373,533 "learning_rate": 4.373333832178478e-05,534 "loss": 2.212,535 "step": 750536 },537 {538 "epoch": 1.7851895386423744,539 "grad_norm": 0.01955031044781208,540 "learning_rate": 4.2380705835242716e-05,541 "loss": 2.0876,542 "step": 760543 },544 {545 "epoch": 1.8086982074640023,546 "grad_norm": 0.02556225284934044,547 "learning_rate": 4.10337591515484e-05,548 "loss": 2.064,549 "step": 770550 },551 {552 "epoch": 1.8322068762856303,553 "grad_norm": 0.04094263166189194,554 "learning_rate": 3.969350341279566e-05,555 "loss": 2.1206,556 "step": 780557 },558 {559 "epoch": 1.8557155451072584,560 "grad_norm": 0.0680101215839386,561 "learning_rate": 3.8360938768044405e-05,562 "loss": 2.0975,563 "step": 790564 },565 {566 "epoch": 1.8792242139288864,567 "grad_norm": 0.08410263061523438,568 "learning_rate": 3.7037059626971944e-05,569 "loss": 2.0519,570 "step": 800571 },572 {573 "epoch": 1.9027328827505143,574 "grad_norm": 0.02342092990875244,575 "learning_rate": 3.572285391780734e-05,576 "loss": 2.019,577 "step": 810578 },579 {580 "epoch": 1.9262415515721423,581 "grad_norm": 0.02327684499323368,582 "learning_rate": 3.441930235010211e-05,583 "loss": 2.0966,584 "step": 820585 },586 {587 "epoch": 1.9497502203937702,588 "grad_norm": 0.0357813723385334,589 "learning_rate": 3.312737768288827e-05,590 "loss": 2.0924,591 "step": 830592 },593 {594 "epoch": 1.9732588892153982,595 "grad_norm": 0.07186655700206757,596 "learning_rate": 3.184804399876886e-05,597 "loss": 2.0315,598 "step": 840599 },600 {601 "epoch": 1.996767558037026,602 "grad_norm": 0.09240284562110901,603 "learning_rate": 3.0582255984483535e-05,604 "loss": 2.0409,605 "step": 850606 },607 {608 "epoch": 2.0188069350573024,609 "grad_norm": 0.04462951049208641,610 "learning_rate": 2.9330958218485392e-05,611 "loss": 1.9696,612 "step": 860613 },614 {615 "epoch": 2.0423156038789303,616 "grad_norm": 0.029340475797653198,617 "learning_rate": 2.8095084466061367e-05,618 "loss": 2.1055,619 "step": 870620 },621 {622 "epoch": 2.0658242727005582,623 "grad_norm": 0.048022590577602386,624 "learning_rate": 2.6875556982521248e-05,625 "loss": 2.0865,626 "step": 880627 },628 {629 "epoch": 2.089332941522186,630 "grad_norm": 0.04670249670743942,631 "learning_rate": 2.5673285824976575e-05,632 "loss": 2.1081,633 "step": 890634 },635 {636 "epoch": 2.112841610343814,637 "grad_norm": 0.015516964718699455,638 "learning_rate": 2.4489168173221567e-05,639 "loss": 2.1023,640 "step": 900641 },642 {643 "epoch": 2.136350279165442,644 "grad_norm": 0.06762902438640594,645 "learning_rate": 2.33240876602242e-05,646 "loss": 2.0929,647 "step": 910648 },649 {650 "epoch": 2.15985894798707,651 "grad_norm": 0.014990041963756084,652 "learning_rate": 2.2178913712726045e-05,653 "loss": 2.0817,654 "step": 920655 },656 {657 "epoch": 2.1833676168086984,658 "grad_norm": 0.07376622408628464,659 "learning_rate": 2.1054500902443423e-05,660 "loss": 2.0868,661 "step": 930662 },663 {664 "epoch": 2.2068762856303263,665 "grad_norm": 0.03299334645271301,666 "learning_rate": 1.9951688308354133e-05,667 "loss": 2.1123,668 "step": 940669 },670 {671 "epoch": 2.2303849544519543,672 "grad_norm": 0.05409041792154312,673 "learning_rate": 1.887129889054503e-05,674 "loss": 2.051,675 "step": 950676 },677 {678 "epoch": 2.2538936232735822,679 "grad_norm": 0.041158534586429596,680 "learning_rate": 1.7814138876088566e-05,681 "loss": 2.0458,682 "step": 960683 },684 {685 "epoch": 2.27740229209521,686 "grad_norm": 0.03300878778100014,687 "learning_rate": 1.6780997157405846e-05,688 "loss": 2.0565,689 "step": 970690 },691 {692 "epoch": 2.300910960916838,693 "grad_norm": 0.009902500547468662,694 "learning_rate": 1.5772644703565565e-05,695 "loss": 2.0952,696 "step": 980697 },698 {699 "epoch": 2.324419629738466,700 "grad_norm": 0.44653964042663574,701 "learning_rate": 1.4789833984957962e-05,702 "loss": 2.0386,703 "step": 990704 },705 {706 "epoch": 2.347928298560094,707 "grad_norm": 0.4100548326969147,708 "learning_rate": 1.3833298411773276e-05,709 "loss": 2.0977,710 "step": 1000711 },712 {713 "epoch": 2.371436967381722,714 "grad_norm": 0.00745609262958169,715 "learning_rate": 1.2903751786703482e-05,716 "loss": 2.0503,717 "step": 1010718 },719 {720 "epoch": 2.39494563620335,721 "grad_norm": 0.05212165042757988,722 "learning_rate": 1.2001887772276076e-05,723 "loss": 2.1437,724 "step": 1020725 },726 {727 "epoch": 2.418454305024978,728 "grad_norm": 0.0367308035492897,729 "learning_rate": 1.1128379373217058e-05,730 "loss": 2.0911,731 "step": 1030732 },733 {734 "epoch": 2.4419629738466058,735 "grad_norm": 0.02135683037340641,736 "learning_rate": 1.028387843422952e-05,737 "loss": 2.1183,738 "step": 1040739 },740 {741 "epoch": 2.465471642668234,742 "grad_norm": 0.02447247877717018,743 "learning_rate": 9.469015153562855e-06,744 "loss": 2.1205,745 "step": 1050746 },747 {748 "epoch": 2.488980311489862,749 "grad_norm": 0.011414138600230217,750 "learning_rate": 8.684397612735024e-06,751 "loss": 2.0027,752 "step": 1060753 },754 {755 "epoch": 2.51248898031149,756 "grad_norm": 0.0028556822799146175,757 "learning_rate": 7.930611322759462e-06,758 "loss": 2.0682,759 "step": 1070760 },761 {762 "epoch": 2.535997649133118,763 "grad_norm": 0.008712991140782833,764 "learning_rate": 7.2082187872147935e-06,765 "loss": 2.0895,766 "step": 1080767 },768 {769 "epoch": 2.559506317954746,770 "grad_norm": 0.03148069605231285,771 "learning_rate": 6.517759082483543e-06,772 "loss": 2.1298,773 "step": 1090774 },775 {776 "epoch": 2.583014986776374,777 "grad_norm": 0.03679489344358444,778 "learning_rate": 5.8597474554731745e-06,779 "loss": 2.0837,780 "step": 1100781 },782 {783 "epoch": 2.606523655598002,784 "grad_norm": 0.05236874148249626,785 "learning_rate": 5.234674939119538e-06,786 "loss": 2.1486,787 "step": 1110788 },789 {790 "epoch": 2.6300323244196298,791 "grad_norm": 0.03369889780879021,792 "learning_rate": 4.643007985959641e-06,793 "loss": 2.0959,794 "step": 1120795 },796 {797 "epoch": 2.6535409932412577,798 "grad_norm": 0.006195698864758015,799 "learning_rate": 4.085188120047362e-06,800 "loss": 2.0439,801 "step": 1130802 },803 {804 "epoch": 2.6770496620628856,805 "grad_norm": 0.03900829330086708,806 "learning_rate": 3.5616316074715995e-06,807 "loss": 2.111,808 "step": 1140809 },810 {811 "epoch": 2.7005583308845136,812 "grad_norm": 0.00790059007704258,813 "learning_rate": 3.0727291457229303e-06,814 "loss": 2.1078,815 "step": 1150816 },817 {818 "epoch": 2.7240669997061415,819 "grad_norm": 0.017095139250159264,820 "learning_rate": 2.618845572140527e-06,821 "loss": 2.1248,822 "step": 1160823 },824 {825 "epoch": 2.7475756685277695,826 "grad_norm": 0.006935155484825373,827 "learning_rate": 2.2003195916567855e-06,828 "loss": 2.0573,829 "step": 1170830 },831 {832 "epoch": 2.7710843373493974,833 "grad_norm": 0.01475539617240429,834 "learning_rate": 1.8174635240431282e-06,835 "loss": 2.0218,836 "step": 1180837 },838 {839 "epoch": 2.7945930061710254,840 "grad_norm": 0.08390510082244873,841 "learning_rate": 1.4705630708451712e-06,842 "loss": 2.0729,843 "step": 1190844 },845 {846 "epoch": 2.8181016749926533,847 "grad_norm": 0.04823232442140579,848 "learning_rate": 1.1598771021816058e-06,849 "loss": 2.0432,850 "step": 1200851 },852 {853 "epoch": 2.8416103438142817,854 "grad_norm": 0.017384668812155724,855 "learning_rate": 8.856374635655695e-07,856 "loss": 2.1005,857 "step": 1210858 },859 {860 "epoch": 2.8651190126359096,861 "grad_norm": 0.016273437067866325,862 "learning_rate": 6.480488028928422e-07,863 "loss": 2.1043,864 "step": 1220865 },866 {867 "epoch": 2.8886276814575376,868 "grad_norm": 0.004406425170600414,869 "learning_rate": 4.47288417725944e-07,870 "loss": 2.1214,871 "step": 1230872 },873 {874 "epoch": 2.9121363502791655,875 "grad_norm": 0.04330357536673546,876 "learning_rate": 2.8350612298801427e-07,877 "loss": 2.0679,878 "step": 1240879 },880 {881 "epoch": 2.9356450191007935,882 "grad_norm": 0.005912340711802244,883 "learning_rate": 1.5682413916531802e-07,884 "loss": 2.1937,885 "step": 1250886 },887 {888 "epoch": 2.9591536879224214,889 "grad_norm": 0.021025869995355606,890 "learning_rate": 6.733700110174157e-08,891 "loss": 2.1239,892 "step": 1260893 },894 {895 "epoch": 2.9826623567440493,896 "grad_norm": 0.006763559300452471,897 "learning_rate": 1.511148745335089e-08,898 "loss": 2.1434,899 "step": 1270900 },901 {902 "epoch": 3.0,903 "step": 1278,904 "total_flos": 2.0454733263908045e+17,905 "train_loss": 5.729273645344288,906 "train_runtime": 5128.0315,907 "train_samples_per_second": 1.991,908 "train_steps_per_second": 0.249909 }910 ],911 "logging_steps": 10,912 "max_steps": 1278,913 "num_input_tokens_seen": 0,914 "num_train_epochs": 3,915 "save_steps": 500,916 "stateful_callbacks": {917 "TrainerControl": {918 "args": {919 "should_epoch_stop": false,920 "should_evaluate": false,921 "should_log": false,922 "should_save": true,923 "should_training_stop": true924 },925 "attributes": {}926 }927 },928 "total_flos": 2.0454733263908045e+17,929 "train_batch_size": 1,930 "trial_name": null,931 "trial_params": null932}933 