verbalyze/Adaptor_Conv_Text2Text
06
1{2 "best_metric": null,3 "best_model_checkpoint": null,4 "epoch": 0.9619348632678302,5 "eval_steps": 500,6 "global_step": 14000,7 "is_hyper_param_search": false,8 "is_local_process_zero": true,9 "is_world_process_zero": true,10 "log_history": [11 {12 "epoch": 6.87096330905593e-05,13 "grad_norm": 7.6275835037231445,14 "learning_rate": 3.4340659340659344e-08,15 "loss": 2.8946,16 "step": 117 },18 {19 "epoch": 0.0001374192661811186,20 "grad_norm": 7.034836769104004,21 "learning_rate": 6.868131868131869e-08,22 "loss": 2.9183,23 "step": 224 },25 {26 "epoch": 0.00020612889927167788,27 "grad_norm": 6.99338960647583,28 "learning_rate": 1.0302197802197802e-07,29 "loss": 2.8106,30 "step": 331 },32 {33 "epoch": 0.0002748385323622372,34 "grad_norm": 7.4205002784729,35 "learning_rate": 1.3736263736263737e-07,36 "loss": 2.9096,37 "step": 438 },39 {40 "epoch": 0.0003435481654527965,41 "grad_norm": 7.320061683654785,42 "learning_rate": 1.717032967032967e-07,43 "loss": 2.8319,44 "step": 545 },46 {47 "epoch": 0.00041225779854335576,48 "grad_norm": 7.4554290771484375,49 "learning_rate": 2.0604395604395605e-07,50 "loss": 2.9632,51 "step": 652 },53 {54 "epoch": 0.0004809674316339151,55 "grad_norm": 7.403331756591797,56 "learning_rate": 2.403846153846154e-07,57 "loss": 2.9145,58 "step": 759 },60 {61 "epoch": 0.0005496770647244744,62 "grad_norm": 7.208906650543213,63 "learning_rate": 2.7472527472527475e-07,64 "loss": 2.9067,65 "step": 866 },67 {68 "epoch": 0.0006183866978150337,69 "grad_norm": 7.182309150695801,70 "learning_rate": 3.0906593406593407e-07,71 "loss": 2.9476,72 "step": 973 },74 {75 "epoch": 0.000687096330905593,76 "grad_norm": 7.336084365844727,77 "learning_rate": 3.434065934065934e-07,78 "loss": 2.9436,79 "step": 1080 },81 {82 "epoch": 0.0007558059639961522,83 "grad_norm": 7.305691719055176,84 "learning_rate": 3.7774725274725277e-07,85 "loss": 2.8734,86 "step": 1187 },88 {89 "epoch": 0.0008245155970867115,90 "grad_norm": 6.958437919616699,91 "learning_rate": 4.120879120879121e-07,92 "loss": 2.8335,93 "step": 1294 },95 {96 "epoch": 0.0008932252301772709,97 "grad_norm": 7.204424858093262,98 "learning_rate": 4.464285714285714e-07,99 "loss": 2.8405,100 "step": 13101 },102 {103 "epoch": 0.0009619348632678302,104 "grad_norm": 7.613925933837891,105 "learning_rate": 4.807692307692308e-07,106 "loss": 2.8819,107 "step": 14108 },109 {110 "epoch": 0.0010306444963583896,111 "grad_norm": 7.293791770935059,112 "learning_rate": 5.151098901098902e-07,113 "loss": 2.9231,114 "step": 15115 },116 {117 "epoch": 0.0010993541294489488,118 "grad_norm": 7.24744176864624,119 "learning_rate": 5.494505494505495e-07,120 "loss": 2.8779,121 "step": 16122 },123 {124 "epoch": 0.001168063762539508,125 "grad_norm": 7.415509223937988,126 "learning_rate": 5.837912087912088e-07,127 "loss": 2.9154,128 "step": 17129 },130 {131 "epoch": 0.0012367733956300674,132 "grad_norm": 7.4035139083862305,133 "learning_rate": 6.181318681318681e-07,134 "loss": 2.8272,135 "step": 18136 },137 {138 "epoch": 0.0013054830287206266,139 "grad_norm": 7.272813320159912,140 "learning_rate": 6.524725274725276e-07,141 "loss": 2.8737,142 "step": 19143 },144 {145 "epoch": 0.001374192661811186,146 "grad_norm": 7.2420196533203125,147 "learning_rate": 6.868131868131868e-07,148 "loss": 2.9522,149 "step": 20150 },151 {152 "epoch": 0.0014429022949017452,153 "grad_norm": 7.265440940856934,154 "learning_rate": 7.211538461538462e-07,155 "loss": 2.8594,156 "step": 21157 },158 {159 "epoch": 0.0015116119279923045,160 "grad_norm": 7.195631504058838,161 "learning_rate": 7.554945054945055e-07,162 "loss": 2.821,163 "step": 22164 },165 {166 "epoch": 0.0015803215610828637,167 "grad_norm": 7.36093807220459,168 "learning_rate": 7.898351648351649e-07,169 "loss": 2.9065,170 "step": 23171 },172 {173 "epoch": 0.001649031194173423,174 "grad_norm": 6.964420318603516,175 "learning_rate": 8.241758241758242e-07,176 "loss": 2.8252,177 "step": 24178 },179 {180 "epoch": 0.0017177408272639825,181 "grad_norm": 7.27252197265625,182 "learning_rate": 8.585164835164836e-07,183 "loss": 2.8766,184 "step": 25185 },186 {187 "epoch": 0.0017864504603545418,188 "grad_norm": 6.662872314453125,189 "learning_rate": 8.928571428571428e-07,190 "loss": 2.7526,191 "step": 26192 },193 {194 "epoch": 0.001855160093445101,195 "grad_norm": 7.0713210105896,196 "learning_rate": 9.271978021978023e-07,197 "loss": 2.7311,198 "step": 27199 },200 {201 "epoch": 0.0019238697265356603,202 "grad_norm": 7.623928070068359,203 "learning_rate": 9.615384615384617e-07,204 "loss": 2.8237,205 "step": 28206 },207 {208 "epoch": 0.0019925793596262194,209 "grad_norm": 7.193939208984375,210 "learning_rate": 9.958791208791208e-07,211 "loss": 2.8225,212 "step": 29213 },214 {215 "epoch": 0.002061288992716779,216 "grad_norm": 6.923072338104248,217 "learning_rate": 1.0302197802197803e-06,218 "loss": 2.6622,219 "step": 30220 },221 {222 "epoch": 0.0021299986258073384,223 "grad_norm": 7.405568599700928,224 "learning_rate": 1.0645604395604397e-06,225 "loss": 2.7985,226 "step": 31227 },228 {229 "epoch": 0.0021987082588978977,230 "grad_norm": 6.8562541007995605,231 "learning_rate": 1.098901098901099e-06,232 "loss": 2.7624,233 "step": 32234 },235 {236 "epoch": 0.002267417891988457,237 "grad_norm": 7.247007369995117,238 "learning_rate": 1.1332417582417583e-06,239 "loss": 2.8215,240 "step": 33241 },242 {243 "epoch": 0.002336127525079016,244 "grad_norm": 7.152578830718994,245 "learning_rate": 1.1675824175824176e-06,246 "loss": 2.7058,247 "step": 34248 },249 {250 "epoch": 0.0024048371581695755,251 "grad_norm": 6.76224946975708,252 "learning_rate": 1.201923076923077e-06,253 "loss": 2.6896,254 "step": 35255 },256 {257 "epoch": 0.0024735467912601347,258 "grad_norm": 6.8790364265441895,259 "learning_rate": 1.2362637362637363e-06,260 "loss": 2.6491,261 "step": 36262 },263 {264 "epoch": 0.002542256424350694,265 "grad_norm": 6.814023494720459,266 "learning_rate": 1.2706043956043956e-06,267 "loss": 2.5844,268 "step": 37269 },270 {271 "epoch": 0.0026109660574412533,272 "grad_norm": 6.636133670806885,273 "learning_rate": 1.3049450549450551e-06,274 "loss": 2.6174,275 "step": 38276 },277 {278 "epoch": 0.0026796756905318126,279 "grad_norm": 6.30716609954834,280 "learning_rate": 1.3392857142857143e-06,281 "loss": 2.4916,282 "step": 39283 },284 {285 "epoch": 0.002748385323622372,286 "grad_norm": 6.623904228210449,287 "learning_rate": 1.3736263736263736e-06,288 "loss": 2.5194,289 "step": 40290 },291 {292 "epoch": 0.002817094956712931,293 "grad_norm": 6.534739971160889,294 "learning_rate": 1.4079670329670331e-06,295 "loss": 2.4827,296 "step": 41297 },298 {299 "epoch": 0.0028858045898034904,300 "grad_norm": 6.654708385467529,301 "learning_rate": 1.4423076923076924e-06,302 "loss": 2.4983,303 "step": 42304 },305 {306 "epoch": 0.0029545142228940497,307 "grad_norm": 6.537683486938477,308 "learning_rate": 1.4766483516483518e-06,309 "loss": 2.5157,310 "step": 43311 },312 {313 "epoch": 0.003023223855984609,314 "grad_norm": 6.166847229003906,315 "learning_rate": 1.510989010989011e-06,316 "loss": 2.444,317 "step": 44318 },319 {320 "epoch": 0.0030919334890751682,321 "grad_norm": 6.101149082183838,322 "learning_rate": 1.5453296703296704e-06,323 "loss": 2.3854,324 "step": 45325 },326 {327 "epoch": 0.0031606431221657275,328 "grad_norm": 6.729682922363281,329 "learning_rate": 1.5796703296703297e-06,330 "loss": 2.4451,331 "step": 46332 },333 {334 "epoch": 0.0032293527552562868,335 "grad_norm": 6.204443454742432,336 "learning_rate": 1.6140109890109893e-06,337 "loss": 2.4385,338 "step": 47339 },340 {341 "epoch": 0.003298062388346846,342 "grad_norm": 5.8173298835754395,343 "learning_rate": 1.6483516483516484e-06,344 "loss": 2.3583,345 "step": 48346 },347 {348 "epoch": 0.0033667720214374053,349 "grad_norm": 5.775567531585693,350 "learning_rate": 1.6826923076923077e-06,351 "loss": 2.2726,352 "step": 49353 },354 {355 "epoch": 0.003435481654527965,356 "grad_norm": 5.820885181427002,357 "learning_rate": 1.7170329670329672e-06,358 "loss": 2.3267,359 "step": 50360 },361 {362 "epoch": 0.0035041912876185243,363 "grad_norm": 5.634654998779297,364 "learning_rate": 1.7513736263736266e-06,365 "loss": 2.2839,366 "step": 51367 },368 {369 "epoch": 0.0035729009207090836,370 "grad_norm": 5.820955276489258,371 "learning_rate": 1.7857142857142857e-06,372 "loss": 2.2362,373 "step": 52374 },375 {376 "epoch": 0.003641610553799643,377 "grad_norm": 5.6372389793396,378 "learning_rate": 1.8200549450549452e-06,379 "loss": 2.2835,380 "step": 53381 },382 {383 "epoch": 0.003710320186890202,384 "grad_norm": 5.373723983764648,385 "learning_rate": 1.8543956043956045e-06,386 "loss": 2.1533,387 "step": 54388 },389 {390 "epoch": 0.0037790298199807614,391 "grad_norm": 5.111402988433838,392 "learning_rate": 1.8887362637362636e-06,393 "loss": 2.2055,394 "step": 55395 },396 {397 "epoch": 0.0038477394530713207,398 "grad_norm": 4.865420818328857,399 "learning_rate": 1.9230769230769234e-06,400 "loss": 2.1924,401 "step": 56402 },403 {404 "epoch": 0.0039164490861618795,405 "grad_norm": 4.456362724304199,406 "learning_rate": 1.9574175824175823e-06,407 "loss": 2.1377,408 "step": 57409 },410 {411 "epoch": 0.003985158719252439,412 "grad_norm": 4.122003078460693,413 "learning_rate": 1.9917582417582416e-06,414 "loss": 2.0542,415 "step": 58416 },417 {418 "epoch": 0.004053868352342998,419 "grad_norm": 3.8278002738952637,420 "learning_rate": 2.0260989010989014e-06,421 "loss": 2.0219,422 "step": 59423 },424 {425 "epoch": 0.004122577985433558,426 "grad_norm": 3.7971367835998535,427 "learning_rate": 2.0604395604395607e-06,428 "loss": 2.0652,429 "step": 60430 },431 {432 "epoch": 0.0041912876185241175,433 "grad_norm": 3.7625250816345215,434 "learning_rate": 2.0947802197802196e-06,435 "loss": 2.0637,436 "step": 61437 },438 {439 "epoch": 0.004259997251614677,440 "grad_norm": 3.3542184829711914,441 "learning_rate": 2.1291208791208793e-06,442 "loss": 2.002,443 "step": 62444 },445 {446 "epoch": 0.004328706884705236,447 "grad_norm": 3.2720158100128174,448 "learning_rate": 2.1634615384615387e-06,449 "loss": 1.9667,450 "step": 63451 },452 {453 "epoch": 0.004397416517795795,454 "grad_norm": 3.4174327850341797,455 "learning_rate": 2.197802197802198e-06,456 "loss": 2.0516,457 "step": 64458 },459 {460 "epoch": 0.004466126150886355,461 "grad_norm": 3.233170986175537,462 "learning_rate": 2.2321428571428573e-06,463 "loss": 2.0174,464 "step": 65465 },466 {467 "epoch": 0.004534835783976914,468 "grad_norm": 3.1018035411834717,469 "learning_rate": 2.2664835164835166e-06,470 "loss": 1.9514,471 "step": 66472 },473 {474 "epoch": 0.004603545417067473,475 "grad_norm": 3.185926675796509,476 "learning_rate": 2.300824175824176e-06,477 "loss": 1.9775,478 "step": 67479 },480 {481 "epoch": 0.004672255050158032,482 "grad_norm": 3.1349759101867676,483 "learning_rate": 2.3351648351648353e-06,484 "loss": 1.9103,485 "step": 68486 },487 {488 "epoch": 0.004740964683248592,489 "grad_norm": 3.2016353607177734,490 "learning_rate": 2.3695054945054946e-06,491 "loss": 1.8872,492 "step": 69493 },494 {495 "epoch": 0.004809674316339151,496 "grad_norm": 2.82181453704834,497 "learning_rate": 2.403846153846154e-06,498 "loss": 1.8656,499 "step": 70500 },501 {502 "epoch": 0.00487838394942971,503 "grad_norm": 2.9412529468536377,504 "learning_rate": 2.4381868131868133e-06,505 "loss": 1.8396,506 "step": 71507 },508 {509 "epoch": 0.0049470935825202695,510 "grad_norm": 2.7847046852111816,511 "learning_rate": 2.4725274725274726e-06,512 "loss": 1.9568,513 "step": 72514 },515 {516 "epoch": 0.005015803215610829,517 "grad_norm": 2.9551219940185547,518 "learning_rate": 2.506868131868132e-06,519 "loss": 1.843,520 "step": 73521 },522 {523 "epoch": 0.005084512848701388,524 "grad_norm": 2.636174440383911,525 "learning_rate": 2.5412087912087912e-06,526 "loss": 1.8051,527 "step": 74528 },529 {530 "epoch": 0.005153222481791947,531 "grad_norm": 2.5625147819519043,532 "learning_rate": 2.5755494505494505e-06,533 "loss": 1.8249,534 "step": 75535 },536 {537 "epoch": 0.005221932114882507,538 "grad_norm": 2.8047335147857666,539 "learning_rate": 2.6098901098901103e-06,540 "loss": 1.8585,541 "step": 76542 },543 {544 "epoch": 0.005290641747973066,545 "grad_norm": 2.790449619293213,546 "learning_rate": 2.644230769230769e-06,547 "loss": 1.8639,548 "step": 77549 },550 {551 "epoch": 0.005359351381063625,552 "grad_norm": 2.593555450439453,553 "learning_rate": 2.6785714285714285e-06,554 "loss": 1.7297,555 "step": 78556 },557 {558 "epoch": 0.005428061014154184,559 "grad_norm": 2.4444713592529297,560 "learning_rate": 2.7129120879120883e-06,561 "loss": 1.7266,562 "step": 79563 },564 {565 "epoch": 0.005496770647244744,566 "grad_norm": 2.5306320190429688,567 "learning_rate": 2.747252747252747e-06,568 "loss": 1.7997,569 "step": 80570 },571 {572 "epoch": 0.005565480280335303,573 "grad_norm": 2.3893392086029053,574 "learning_rate": 2.7815934065934065e-06,575 "loss": 1.8138,576 "step": 81577 },578 {579 "epoch": 0.005634189913425862,580 "grad_norm": 2.4648749828338623,581 "learning_rate": 2.8159340659340662e-06,582 "loss": 1.736,583 "step": 82584 },585 {586 "epoch": 0.0057028995465164215,587 "grad_norm": 2.3402011394500732,588 "learning_rate": 2.850274725274725e-06,589 "loss": 1.7413,590 "step": 83591 },592 {593 "epoch": 0.005771609179606981,594 "grad_norm": 2.384143352508545,595 "learning_rate": 2.884615384615385e-06,596 "loss": 1.7028,597 "step": 84598 },599 {600 "epoch": 0.00584031881269754,601 "grad_norm": 2.4274344444274902,602 "learning_rate": 2.918956043956044e-06,603 "loss": 1.7012,604 "step": 85605 },606 {607 "epoch": 0.005909028445788099,608 "grad_norm": 2.3654894828796387,609 "learning_rate": 2.9532967032967035e-06,610 "loss": 1.6972,611 "step": 86612 },613 {614 "epoch": 0.005977738078878659,615 "grad_norm": 2.293001413345337,616 "learning_rate": 2.987637362637363e-06,617 "loss": 1.6364,618 "step": 87619 },620 {621 "epoch": 0.006046447711969218,622 "grad_norm": 2.3870837688446045,623 "learning_rate": 3.021978021978022e-06,624 "loss": 1.6952,625 "step": 88626 },627 {628 "epoch": 0.006115157345059777,629 "grad_norm": 2.3209028244018555,630 "learning_rate": 3.0563186813186815e-06,631 "loss": 1.6679,632 "step": 89633 },634 {635 "epoch": 0.0061838669781503364,636 "grad_norm": 2.3537614345550537,637 "learning_rate": 3.090659340659341e-06,638 "loss": 1.6645,639 "step": 90640 },641 {642 "epoch": 0.006252576611240896,643 "grad_norm": 2.4990780353546143,644 "learning_rate": 3.125e-06,645 "loss": 1.7019,646 "step": 91647 },648 {649 "epoch": 0.006321286244331455,650 "grad_norm": 2.2731547355651855,651 "learning_rate": 3.1593406593406595e-06,652 "loss": 1.6265,653 "step": 92654 },655 {656 "epoch": 0.006389995877422014,657 "grad_norm": 2.4347567558288574,658 "learning_rate": 3.193681318681319e-06,659 "loss": 1.6253,660 "step": 93661 },662 {663 "epoch": 0.0064587055105125735,664 "grad_norm": 2.36327862739563,665 "learning_rate": 3.2280219780219785e-06,666 "loss": 1.6279,667 "step": 94668 },669 {670 "epoch": 0.006527415143603133,671 "grad_norm": 2.1690657138824463,672 "learning_rate": 3.262362637362638e-06,673 "loss": 1.6204,674 "step": 95675 },676 {677 "epoch": 0.006596124776693692,678 "grad_norm": 2.274386167526245,679 "learning_rate": 3.2967032967032968e-06,680 "loss": 1.6423,681 "step": 96682 },683 {684 "epoch": 0.006664834409784251,685 "grad_norm": 2.2845969200134277,686 "learning_rate": 3.331043956043956e-06,687 "loss": 1.5863,688 "step": 97689 },690 {691 "epoch": 0.006733544042874811,692 "grad_norm": 2.2257418632507324,693 "learning_rate": 3.3653846153846154e-06,694 "loss": 1.6801,695 "step": 98696 },697 {698 "epoch": 0.006802253675965371,699 "grad_norm": 2.1797356605529785,700 "learning_rate": 3.3997252747252747e-06,701 "loss": 1.6156,702 "step": 99703 },704 {705 "epoch": 0.00687096330905593,706 "grad_norm": 2.4152626991271973,707 "learning_rate": 3.4340659340659345e-06,708 "loss": 1.5866,709 "step": 100710 },711 {712 "epoch": 0.006939672942146489,713 "grad_norm": 2.4223873615264893,714 "learning_rate": 3.468406593406594e-06,715 "loss": 1.538,716 "step": 101717 },718 {719 "epoch": 0.007008382575237049,720 "grad_norm": 2.350939989089966,721 "learning_rate": 3.502747252747253e-06,722 "loss": 1.6015,723 "step": 102724 },725 {726 "epoch": 0.007077092208327608,727 "grad_norm": 2.2195897102355957,728 "learning_rate": 3.537087912087912e-06,729 "loss": 1.5623,730 "step": 103731 },732 {733 "epoch": 0.007145801841418167,734 "grad_norm": 2.3513755798339844,735 "learning_rate": 3.5714285714285714e-06,736 "loss": 1.6151,737 "step": 104738 },739 {740 "epoch": 0.007214511474508726,741 "grad_norm": 2.2731540203094482,742 "learning_rate": 3.6057692307692307e-06,743 "loss": 1.5388,744 "step": 105745 },746 {747 "epoch": 0.007283221107599286,748 "grad_norm": 2.3015456199645996,749 "learning_rate": 3.6401098901098904e-06,750 "loss": 1.5513,751 "step": 106752 },753 {754 "epoch": 0.007351930740689845,755 "grad_norm": 2.541792154312134,756 "learning_rate": 3.6744505494505498e-06,757 "loss": 1.5543,758 "step": 107759 },760 {761 "epoch": 0.007420640373780404,762 "grad_norm": 2.2383248805999756,763 "learning_rate": 3.708791208791209e-06,764 "loss": 1.5505,765 "step": 108766 },767 {768 "epoch": 0.0074893500068709635,769 "grad_norm": 2.3158953189849854,770 "learning_rate": 3.7431318681318684e-06,771 "loss": 1.5215,772 "step": 109773 },774 {775 "epoch": 0.007558059639961523,776 "grad_norm": 2.24996280670166,777 "learning_rate": 3.7774725274725273e-06,778 "loss": 1.5197,779 "step": 110780 },781 {782 "epoch": 0.007626769273052082,783 "grad_norm": 2.339571237564087,784 "learning_rate": 3.8118131868131875e-06,785 "loss": 1.5122,786 "step": 111787 },788 {789 "epoch": 0.007695478906142641,790 "grad_norm": 2.3899333477020264,791 "learning_rate": 3.846153846153847e-06,792 "loss": 1.5806,793 "step": 112794 },795 {796 "epoch": 0.007764188539233201,797 "grad_norm": 2.2707226276397705,798 "learning_rate": 3.880494505494506e-06,799 "loss": 1.4875,800 "step": 113801 },802 {803 "epoch": 0.007832898172323759,804 "grad_norm": 2.4128918647766113,805 "learning_rate": 3.914835164835165e-06,806 "loss": 1.5629,807 "step": 114808 },809 {810 "epoch": 0.007901607805414318,811 "grad_norm": 2.2566280364990234,812 "learning_rate": 3.949175824175824e-06,813 "loss": 1.492,814 "step": 115815 },816 {817 "epoch": 0.007970317438504878,818 "grad_norm": 2.4572346210479736,819 "learning_rate": 3.983516483516483e-06,820 "loss": 1.5491,821 "step": 116822 },823 {824 "epoch": 0.008039027071595437,825 "grad_norm": 2.351505994796753,826 "learning_rate": 4.017857142857143e-06,827 "loss": 1.4869,828 "step": 117829 },830 {831 "epoch": 0.008107736704685996,832 "grad_norm": 2.407135009765625,833 "learning_rate": 4.052197802197803e-06,834 "loss": 1.4832,835 "step": 118836 },837 {838 "epoch": 0.008176446337776557,839 "grad_norm": 2.2351458072662354,840 "learning_rate": 4.086538461538462e-06,841 "loss": 1.4449,842 "step": 119843 },844 {845 "epoch": 0.008245155970867116,846 "grad_norm": 2.301877975463867,847 "learning_rate": 4.120879120879121e-06,848 "loss": 1.435,849 "step": 120850 },851 {852 "epoch": 0.008313865603957676,853 "grad_norm": 2.291778564453125,854 "learning_rate": 4.15521978021978e-06,855 "loss": 1.4471,856 "step": 121857 },858 {859 "epoch": 0.008382575237048235,860 "grad_norm": 2.071810483932495,861 "learning_rate": 4.189560439560439e-06,862 "loss": 1.4251,863 "step": 122864 },865 {866 "epoch": 0.008451284870138794,867 "grad_norm": 1.9987494945526123,868 "learning_rate": 4.2239010989011e-06,869 "loss": 1.4582,870 "step": 123871 },872 {873 "epoch": 0.008519994503229354,874 "grad_norm": 1.9596329927444458,875 "learning_rate": 4.258241758241759e-06,876 "loss": 1.4143,877 "step": 124878 },879 {880 "epoch": 0.008588704136319913,881 "grad_norm": 1.8985395431518555,882 "learning_rate": 4.292582417582418e-06,883 "loss": 1.4898,884 "step": 125885 },886 {887 "epoch": 0.008657413769410472,888 "grad_norm": 1.9674025774002075,889 "learning_rate": 4.326923076923077e-06,890 "loss": 1.4418,891 "step": 126892 },893 {894 "epoch": 0.008726123402501031,895 "grad_norm": 1.9395891427993774,896 "learning_rate": 4.361263736263736e-06,897 "loss": 1.4103,898 "step": 127899 },900 {901 "epoch": 0.00879483303559159,902 "grad_norm": 1.9848326444625854,903 "learning_rate": 4.395604395604396e-06,904 "loss": 1.3934,905 "step": 128906 },907 {908 "epoch": 0.00886354266868215,909 "grad_norm": 2.5897297859191895,910 "learning_rate": 4.429945054945056e-06,911 "loss": 1.3345,912 "step": 129913 },914 {915 "epoch": 0.00893225230177271,916 "grad_norm": 2.467331647872925,917 "learning_rate": 4.464285714285715e-06,918 "loss": 1.3537,919 "step": 130920 },921 {922 "epoch": 0.009000961934863268,923 "grad_norm": 2.2849247455596924,924 "learning_rate": 4.4986263736263735e-06,925 "loss": 1.3731,926 "step": 131927 },928 {929 "epoch": 0.009069671567953828,930 "grad_norm": 2.3435752391815186,931 "learning_rate": 4.532967032967033e-06,932 "loss": 1.4271,933 "step": 132934 },935 {936 "epoch": 0.009138381201044387,937 "grad_norm": 1.9016616344451904,938 "learning_rate": 4.567307692307692e-06,939 "loss": 1.3948,940 "step": 133941 },942 {943 "epoch": 0.009207090834134946,944 "grad_norm": 1.9436383247375488,945 "learning_rate": 4.601648351648352e-06,946 "loss": 1.3657,947 "step": 134948 },949 {950 "epoch": 0.009275800467225506,951 "grad_norm": 1.8262126445770264,952 "learning_rate": 4.635989010989012e-06,953 "loss": 1.3794,954 "step": 135955 },956 {957 "epoch": 0.009344510100316065,958 "grad_norm": 1.9528446197509766,959 "learning_rate": 4.6703296703296706e-06,960 "loss": 1.4398,961 "step": 136962 },963 {964 "epoch": 0.009413219733406624,965 "grad_norm": 1.868327021598816,966 "learning_rate": 4.7046703296703295e-06,967 "loss": 1.3879,968 "step": 137969 },970 {971 "epoch": 0.009481929366497183,972 "grad_norm": 1.9784241914749146,973 "learning_rate": 4.739010989010989e-06,974 "loss": 1.4418,975 "step": 138976 },977 {978 "epoch": 0.009550638999587743,979 "grad_norm": 1.8264296054840088,980 "learning_rate": 4.773351648351648e-06,981 "loss": 1.4034,982 "step": 139983 },984 {985 "epoch": 0.009619348632678302,986 "grad_norm": 1.8559573888778687,987 "learning_rate": 4.807692307692308e-06,988 "loss": 1.4029,989 "step": 140990 },991 {992 "epoch": 0.009688058265768861,993 "grad_norm": 1.7852126359939575,994 "learning_rate": 4.842032967032968e-06,995 "loss": 1.4524,996 "step": 141997 },998 {999 "epoch": 0.00975676789885942,1000 "grad_norm": 1.877097725868225,1001 "learning_rate": 4.8763736263736265e-06,1002 "loss": 1.3647,1003 "step": 1421004 },1005 {1006 "epoch": 0.00982547753194998,1007 "grad_norm": 1.8444008827209473,1008 "learning_rate": 4.910714285714286e-06,1009 "loss": 1.3675,1010 "step": 1431011 },1012 {1013 "epoch": 0.009894187165040539,1014 "grad_norm": 1.8507466316223145,1015 "learning_rate": 4.945054945054945e-06,1016 "loss": 1.3751,1017 "step": 1441018 },1019 {1020 "epoch": 0.009962896798131098,1021 "grad_norm": 1.7089701890945435,1022 "learning_rate": 4.979395604395604e-06,1023 "loss": 1.3548,1024 "step": 1451025 },1026 {1027 "epoch": 0.010031606431221658,1028 "grad_norm": 1.8487894535064697,1029 "learning_rate": 5.013736263736264e-06,1030 "loss": 1.3926,1031 "step": 1461032 },1033 {1034 "epoch": 0.010100316064312217,1035 "grad_norm": 1.7713396549224854,1036 "learning_rate": 5.0480769230769235e-06,1037 "loss": 1.3107,1038 "step": 1471039 },1040 {1041 "epoch": 0.010169025697402776,1042 "grad_norm": 1.950148344039917,1043 "learning_rate": 5.0824175824175824e-06,1044 "loss": 1.3245,1045 "step": 1481046 },1047 {1048 "epoch": 0.010237735330493335,1049 "grad_norm": 1.8153810501098633,1050 "learning_rate": 5.116758241758242e-06,1051 "loss": 1.3405,1052 "step": 1491053 },1054 {1055 "epoch": 0.010306444963583895,1056 "grad_norm": 1.7532216310501099,1057 "learning_rate": 5.151098901098901e-06,1058 "loss": 1.3503,1059 "step": 1501060 },1061 {1062 "epoch": 0.010375154596674454,1063 "grad_norm": 1.7673825025558472,1064 "learning_rate": 5.185439560439561e-06,1065 "loss": 1.3175,1066 "step": 1511067 },1068 {1069 "epoch": 0.010443864229765013,1070 "grad_norm": 1.9625242948532104,1071 "learning_rate": 5.219780219780221e-06,1072 "loss": 1.3695,1073 "step": 1521074 },1075 {1076 "epoch": 0.010512573862855572,1077 "grad_norm": 1.8338172435760498,1078 "learning_rate": 5.2541208791208795e-06,1079 "loss": 1.3307,1080 "step": 1531081 },1082 {1083 "epoch": 0.010581283495946132,1084 "grad_norm": 1.8488733768463135,1085 "learning_rate": 5.288461538461538e-06,1086 "loss": 1.3603,1087 "step": 1541088 },1089 {1090 "epoch": 0.010649993129036691,1091 "grad_norm": 1.7387433052062988,1092 "learning_rate": 5.322802197802198e-06,1093 "loss": 1.3519,1094 "step": 1551095 },1096 {1097 "epoch": 0.01071870276212725,1098 "grad_norm": 1.8353899717330933,1099 "learning_rate": 5.357142857142857e-06,1100 "loss": 1.3633,1101 "step": 1561102 },1103 {1104 "epoch": 0.01078741239521781,1105 "grad_norm": 1.8494092226028442,1106 "learning_rate": 5.391483516483517e-06,1107 "loss": 1.3698,1108 "step": 1571109 },1110 {1111 "epoch": 0.010856122028308369,1112 "grad_norm": 1.6814545392990112,1113 "learning_rate": 5.4258241758241765e-06,1114 "loss": 1.3351,1115 "step": 1581116 },1117 {1118 "epoch": 0.010924831661398928,1119 "grad_norm": 1.776051640510559,1120 "learning_rate": 5.4601648351648354e-06,1121 "loss": 1.3166,1122 "step": 1591123 },1124 {1125 "epoch": 0.010993541294489487,1126 "grad_norm": 1.714782953262329,1127 "learning_rate": 5.494505494505494e-06,1128 "loss": 1.3375,1129 "step": 1601130 },1131 {1132 "epoch": 0.011062250927580047,1133 "grad_norm": 1.8178660869598389,1134 "learning_rate": 5.528846153846154e-06,1135 "loss": 1.2968,1136 "step": 1611137 },1138 {1139 "epoch": 0.011130960560670606,1140 "grad_norm": 1.8437119722366333,1141 "learning_rate": 5.563186813186813e-06,1142 "loss": 1.3047,1143 "step": 1621144 },1145 {1146 "epoch": 0.011199670193761165,1147 "grad_norm": 1.729608178138733,1148 "learning_rate": 5.597527472527473e-06,1149 "loss": 1.3025,1150 "step": 1631151 },1152 {1153 "epoch": 0.011268379826851724,1154 "grad_norm": 1.9262492656707764,1155 "learning_rate": 5.6318681318681325e-06,1156 "loss": 1.3427,1157 "step": 1641158 },1159 {1160 "epoch": 0.011337089459942284,1161 "grad_norm": 1.7662638425827026,1162 "learning_rate": 5.666208791208791e-06,1163 "loss": 1.2914,1164 "step": 1651165 },1166 {1167 "epoch": 0.011405799093032843,1168 "grad_norm": 1.7553950548171997,1169 "learning_rate": 5.70054945054945e-06,1170 "loss": 1.3458,1171 "step": 1661172 },1173 {1174 "epoch": 0.011474508726123402,1175 "grad_norm": 1.7810910940170288,1176 "learning_rate": 5.73489010989011e-06,1177 "loss": 1.3387,1178 "step": 1671179 },1180 {1181 "epoch": 0.011543218359213962,1182 "grad_norm": 1.875128149986267,1183 "learning_rate": 5.76923076923077e-06,1184 "loss": 1.3337,1185 "step": 1681186 },1187 {1188 "epoch": 0.01161192799230452,1189 "grad_norm": 1.884323000907898,1190 "learning_rate": 5.803571428571429e-06,1191 "loss": 1.3702,1192 "step": 1691193 },1194 {1195 "epoch": 0.01168063762539508,1196 "grad_norm": 1.7443432807922363,1197 "learning_rate": 5.837912087912088e-06,1198 "loss": 1.2652,1199 "step": 1701200 },