edithram23/Redaction
010
1{2 "best_metric": 1.9801934740826255e-06,3 "best_model_checkpoint": "Models/t5-base-class-gen/checkpoint-10400",4 "epoch": 4.9440298507462686,5 "eval_steps": 100,6 "global_step": 10600,7 "is_hyper_param_search": false,8 "is_local_process_zero": true,9 "is_world_process_zero": true,10 "log_history": [11 {12 "epoch": 0.04664179104477612,13 "grad_norm": 1.2865939140319824,14 "learning_rate": 3.9630597014925376e-05,15 "loss": 0.437,16 "step": 10017 },18 {19 "epoch": 0.04664179104477612,20 "eval_loss": 0.020251411944627762,21 "eval_runtime": 0.1587,22 "eval_samples_per_second": 189.027,23 "eval_steps_per_second": 25.204,24 "step": 10025 },26 {27 "epoch": 0.09328358208955224,28 "grad_norm": 0.69548100233078,29 "learning_rate": 3.925746268656717e-05,30 "loss": 0.0333,31 "step": 20032 },33 {34 "epoch": 0.09328358208955224,35 "eval_loss": 0.0021759807132184505,36 "eval_runtime": 0.1592,37 "eval_samples_per_second": 188.446,38 "eval_steps_per_second": 25.126,39 "step": 20040 },41 {42 "epoch": 0.13992537313432835,43 "grad_norm": 0.07560934871435165,44 "learning_rate": 3.888432835820896e-05,45 "loss": 0.0203,46 "step": 30047 },48 {49 "epoch": 0.13992537313432835,50 "eval_loss": 0.004563063848763704,51 "eval_runtime": 0.1618,52 "eval_samples_per_second": 185.371,53 "eval_steps_per_second": 24.716,54 "step": 30055 },56 {57 "epoch": 0.1865671641791045,58 "grad_norm": 0.646364688873291,59 "learning_rate": 3.851119402985075e-05,60 "loss": 0.0144,61 "step": 40062 },63 {64 "epoch": 0.1865671641791045,65 "eval_loss": 0.00161658669821918,66 "eval_runtime": 0.1567,67 "eval_samples_per_second": 191.49,68 "eval_steps_per_second": 25.532,69 "step": 40070 },71 {72 "epoch": 0.2332089552238806,73 "grad_norm": 0.30101820826530457,74 "learning_rate": 3.813805970149254e-05,75 "loss": 0.0079,76 "step": 50077 },78 {79 "epoch": 0.2332089552238806,80 "eval_loss": 0.0010551114100962877,81 "eval_runtime": 0.1559,82 "eval_samples_per_second": 192.377,83 "eval_steps_per_second": 25.65,84 "step": 50085 },86 {87 "epoch": 0.2798507462686567,88 "grad_norm": 0.5621537566184998,89 "learning_rate": 3.7764925373134334e-05,90 "loss": 0.0076,91 "step": 60092 },93 {94 "epoch": 0.2798507462686567,95 "eval_loss": 0.0002234852290712297,96 "eval_runtime": 0.1571,97 "eval_samples_per_second": 191.018,98 "eval_steps_per_second": 25.469,99 "step": 600100 },101 {102 "epoch": 0.32649253731343286,103 "grad_norm": 0.10140960663557053,104 "learning_rate": 3.7391791044776125e-05,105 "loss": 0.0092,106 "step": 700107 },108 {109 "epoch": 0.32649253731343286,110 "eval_loss": 0.00010653692879714072,111 "eval_runtime": 0.1572,112 "eval_samples_per_second": 190.855,113 "eval_steps_per_second": 25.447,114 "step": 700115 },116 {117 "epoch": 0.373134328358209,118 "grad_norm": 0.02672586776316166,119 "learning_rate": 3.701865671641792e-05,120 "loss": 0.0053,121 "step": 800122 },123 {124 "epoch": 0.373134328358209,125 "eval_loss": 0.00011367371189408004,126 "eval_runtime": 0.1569,127 "eval_samples_per_second": 191.155,128 "eval_steps_per_second": 25.487,129 "step": 800130 },131 {132 "epoch": 0.4197761194029851,133 "grad_norm": 0.09280095249414444,134 "learning_rate": 3.66455223880597e-05,135 "loss": 0.0054,136 "step": 900137 },138 {139 "epoch": 0.4197761194029851,140 "eval_loss": 0.00013640630641020834,141 "eval_runtime": 0.1588,142 "eval_samples_per_second": 188.886,143 "eval_steps_per_second": 25.185,144 "step": 900145 },146 {147 "epoch": 0.4664179104477612,148 "grad_norm": 0.034318458288908005,149 "learning_rate": 3.62723880597015e-05,150 "loss": 0.0067,151 "step": 1000152 },153 {154 "epoch": 0.4664179104477612,155 "eval_loss": 7.585491403006017e-05,156 "eval_runtime": 0.1557,157 "eval_samples_per_second": 192.664,158 "eval_steps_per_second": 25.689,159 "step": 1000160 },161 {162 "epoch": 0.5130597014925373,163 "grad_norm": 0.018935877829790115,164 "learning_rate": 3.5899253731343285e-05,165 "loss": 0.0035,166 "step": 1100167 },168 {169 "epoch": 0.5130597014925373,170 "eval_loss": 4.489553612074815e-05,171 "eval_runtime": 0.1574,172 "eval_samples_per_second": 190.631,173 "eval_steps_per_second": 25.418,174 "step": 1100175 },176 {177 "epoch": 0.5597014925373134,178 "grad_norm": 0.8913596868515015,179 "learning_rate": 3.5526119402985077e-05,180 "loss": 0.0056,181 "step": 1200182 },183 {184 "epoch": 0.5597014925373134,185 "eval_loss": 4.118124343222007e-05,186 "eval_runtime": 0.1589,187 "eval_samples_per_second": 188.831,188 "eval_steps_per_second": 25.177,189 "step": 1200190 },191 {192 "epoch": 0.6063432835820896,193 "grad_norm": 0.017305633053183556,194 "learning_rate": 3.515298507462687e-05,195 "loss": 0.0041,196 "step": 1300197 },198 {199 "epoch": 0.6063432835820896,200 "eval_loss": 4.481070573092438e-05,201 "eval_runtime": 0.1552,202 "eval_samples_per_second": 193.329,203 "eval_steps_per_second": 25.777,204 "step": 1300205 },206 {207 "epoch": 0.6529850746268657,208 "grad_norm": 0.009413465857505798,209 "learning_rate": 3.477985074626866e-05,210 "loss": 0.0029,211 "step": 1400212 },213 {214 "epoch": 0.6529850746268657,215 "eval_loss": 2.9594228180940263e-05,216 "eval_runtime": 0.1654,217 "eval_samples_per_second": 181.422,218 "eval_steps_per_second": 24.19,219 "step": 1400220 },221 {222 "epoch": 0.6996268656716418,223 "grad_norm": 0.004255557898432016,224 "learning_rate": 3.440671641791045e-05,225 "loss": 0.0028,226 "step": 1500227 },228 {229 "epoch": 0.6996268656716418,230 "eval_loss": 3.3944801543839276e-05,231 "eval_runtime": 0.1543,232 "eval_samples_per_second": 194.435,233 "eval_steps_per_second": 25.925,234 "step": 1500235 },236 {237 "epoch": 0.746268656716418,238 "grad_norm": 0.6249216198921204,239 "learning_rate": 3.403358208955224e-05,240 "loss": 0.003,241 "step": 1600242 },243 {244 "epoch": 0.746268656716418,245 "eval_loss": 3.595968155423179e-05,246 "eval_runtime": 0.1621,247 "eval_samples_per_second": 185.072,248 "eval_steps_per_second": 24.676,249 "step": 1600250 },251 {252 "epoch": 0.792910447761194,253 "grad_norm": 0.014389215037226677,254 "learning_rate": 3.3660447761194034e-05,255 "loss": 0.002,256 "step": 1700257 },258 {259 "epoch": 0.792910447761194,260 "eval_loss": 3.822403959929943e-05,261 "eval_runtime": 0.1593,262 "eval_samples_per_second": 188.284,263 "eval_steps_per_second": 25.105,264 "step": 1700265 },266 {267 "epoch": 0.8395522388059702,268 "grad_norm": 0.1081342026591301,269 "learning_rate": 3.328731343283582e-05,270 "loss": 0.0039,271 "step": 1800272 },273 {274 "epoch": 0.8395522388059702,275 "eval_loss": 3.7738984246971086e-05,276 "eval_runtime": 0.1558,277 "eval_samples_per_second": 192.583,278 "eval_steps_per_second": 25.678,279 "step": 1800280 },281 {282 "epoch": 0.8861940298507462,283 "grad_norm": 1.123632550239563,284 "learning_rate": 3.291417910447762e-05,285 "loss": 0.0035,286 "step": 1900287 },288 {289 "epoch": 0.8861940298507462,290 "eval_loss": 2.4694763851584867e-05,291 "eval_runtime": 0.157,292 "eval_samples_per_second": 191.065,293 "eval_steps_per_second": 25.475,294 "step": 1900295 },296 {297 "epoch": 0.9328358208955224,298 "grad_norm": 0.00756313418969512,299 "learning_rate": 3.25410447761194e-05,300 "loss": 0.0027,301 "step": 2000302 },303 {304 "epoch": 0.9328358208955224,305 "eval_loss": 5.045727448305115e-05,306 "eval_runtime": 0.1664,307 "eval_samples_per_second": 180.272,308 "eval_steps_per_second": 24.036,309 "step": 2000310 },311 {312 "epoch": 0.9794776119402985,313 "grad_norm": 0.0035852426663041115,314 "learning_rate": 3.2167910447761194e-05,315 "loss": 0.0026,316 "step": 2100317 },318 {319 "epoch": 0.9794776119402985,320 "eval_loss": 2.137782212230377e-05,321 "eval_runtime": 0.1552,322 "eval_samples_per_second": 193.332,323 "eval_steps_per_second": 25.778,324 "step": 2100325 },326 {327 "epoch": 1.0261194029850746,328 "grad_norm": 0.01669035293161869,329 "learning_rate": 3.1794776119402986e-05,330 "loss": 0.0008,331 "step": 2200332 },333 {334 "epoch": 1.0261194029850746,335 "eval_loss": 2.3261545720743015e-05,336 "eval_runtime": 0.1554,337 "eval_samples_per_second": 193.073,338 "eval_steps_per_second": 25.743,339 "step": 2200340 },341 {342 "epoch": 1.0727611940298507,343 "grad_norm": 0.002784071722999215,344 "learning_rate": 3.142164179104478e-05,345 "loss": 0.0016,346 "step": 2300347 },348 {349 "epoch": 1.0727611940298507,350 "eval_loss": 1.877884460554924e-05,351 "eval_runtime": 0.1597,352 "eval_samples_per_second": 187.795,353 "eval_steps_per_second": 25.039,354 "step": 2300355 },356 {357 "epoch": 1.1194029850746268,358 "grad_norm": 0.010812056250870228,359 "learning_rate": 3.104850746268657e-05,360 "loss": 0.0013,361 "step": 2400362 },363 {364 "epoch": 1.1194029850746268,365 "eval_loss": 2.032700831477996e-05,366 "eval_runtime": 0.157,367 "eval_samples_per_second": 191.1,368 "eval_steps_per_second": 25.48,369 "step": 2400370 },371 {372 "epoch": 1.166044776119403,373 "grad_norm": 0.08882497251033783,374 "learning_rate": 3.067537313432836e-05,375 "loss": 0.0023,376 "step": 2500377 },378 {379 "epoch": 1.166044776119403,380 "eval_loss": 2.7997491997666657e-05,381 "eval_runtime": 0.1583,382 "eval_samples_per_second": 189.547,383 "eval_steps_per_second": 25.273,384 "step": 2500385 },386 {387 "epoch": 1.212686567164179,388 "grad_norm": 0.5638116002082825,389 "learning_rate": 3.030223880597015e-05,390 "loss": 0.0014,391 "step": 2600392 },393 {394 "epoch": 1.212686567164179,395 "eval_loss": 1.3365392987907398e-05,396 "eval_runtime": 0.1568,397 "eval_samples_per_second": 191.368,398 "eval_steps_per_second": 25.516,399 "step": 2600400 },401 {402 "epoch": 1.2593283582089552,403 "grad_norm": 0.6982096433639526,404 "learning_rate": 2.9929104477611944e-05,405 "loss": 0.0009,406 "step": 2700407 },408 {409 "epoch": 1.2593283582089552,410 "eval_loss": 1.1131261089758482e-05,411 "eval_runtime": 0.1571,412 "eval_samples_per_second": 191.001,413 "eval_steps_per_second": 25.467,414 "step": 2700415 },416 {417 "epoch": 1.3059701492537314,418 "grad_norm": 0.2126922756433487,419 "learning_rate": 2.9555970149253732e-05,420 "loss": 0.0007,421 "step": 2800422 },423 {424 "epoch": 1.3059701492537314,425 "eval_loss": 1.0051960998680443e-05,426 "eval_runtime": 0.1557,427 "eval_samples_per_second": 192.64,428 "eval_steps_per_second": 25.685,429 "step": 2800430 },431 {432 "epoch": 1.3526119402985075,433 "grad_norm": 0.07693888247013092,434 "learning_rate": 2.9182835820895527e-05,435 "loss": 0.0011,436 "step": 2900437 },438 {439 "epoch": 1.3526119402985075,440 "eval_loss": 8.506859558110591e-06,441 "eval_runtime": 0.1606,442 "eval_samples_per_second": 186.795,443 "eval_steps_per_second": 24.906,444 "step": 2900445 },446 {447 "epoch": 1.3992537313432836,448 "grad_norm": 0.005555067211389542,449 "learning_rate": 2.8809701492537315e-05,450 "loss": 0.0012,451 "step": 3000452 },453 {454 "epoch": 1.3992537313432836,455 "eval_loss": 1.0461569218023214e-05,456 "eval_runtime": 0.1562,457 "eval_samples_per_second": 192.09,458 "eval_steps_per_second": 25.612,459 "step": 3000460 },461 {462 "epoch": 1.4458955223880596,463 "grad_norm": 0.15883490443229675,464 "learning_rate": 2.8436567164179106e-05,465 "loss": 0.0015,466 "step": 3100467 },468 {469 "epoch": 1.4458955223880596,470 "eval_loss": 9.19290141609963e-06,471 "eval_runtime": 0.1574,472 "eval_samples_per_second": 190.61,473 "eval_steps_per_second": 25.415,474 "step": 3100475 },476 {477 "epoch": 1.4925373134328357,478 "grad_norm": 0.0036452861968427896,479 "learning_rate": 2.806716417910448e-05,480 "loss": 0.0017,481 "step": 3200482 },483 {484 "epoch": 1.4925373134328357,485 "eval_loss": 7.644109246029984e-06,486 "eval_runtime": 0.1573,487 "eval_samples_per_second": 190.724,488 "eval_steps_per_second": 25.43,489 "step": 3200490 },491 {492 "epoch": 1.539179104477612,493 "grad_norm": 0.1588873416185379,494 "learning_rate": 2.769402985074627e-05,495 "loss": 0.0009,496 "step": 3300497 },498 {499 "epoch": 1.539179104477612,500 "eval_loss": 9.155748557532206e-06,501 "eval_runtime": 0.1602,502 "eval_samples_per_second": 187.322,503 "eval_steps_per_second": 24.976,504 "step": 3300505 },506 {507 "epoch": 1.585820895522388,508 "grad_norm": 0.004792116116732359,509 "learning_rate": 2.7320895522388062e-05,510 "loss": 0.0018,511 "step": 3400512 },513 {514 "epoch": 1.585820895522388,515 "eval_loss": 8.091868039628025e-06,516 "eval_runtime": 0.1578,517 "eval_samples_per_second": 190.11,518 "eval_steps_per_second": 25.348,519 "step": 3400520 },521 {522 "epoch": 1.6324626865671643,523 "grad_norm": 2.761561870574951,524 "learning_rate": 2.694776119402985e-05,525 "loss": 0.0011,526 "step": 3500527 },528 {529 "epoch": 1.6324626865671643,530 "eval_loss": 8.603728929301724e-06,531 "eval_runtime": 0.1552,532 "eval_samples_per_second": 193.265,533 "eval_steps_per_second": 25.769,534 "step": 3500535 },536 {537 "epoch": 1.6791044776119404,538 "grad_norm": 0.001969451317563653,539 "learning_rate": 2.6574626865671645e-05,540 "loss": 0.0012,541 "step": 3600542 },543 {544 "epoch": 1.6791044776119404,545 "eval_loss": 8.435232302872464e-06,546 "eval_runtime": 0.1553,547 "eval_samples_per_second": 193.184,548 "eval_steps_per_second": 25.758,549 "step": 3600550 },551 {552 "epoch": 1.7257462686567164,553 "grad_norm": 0.005350825376808643,554 "learning_rate": 2.6201492537313434e-05,555 "loss": 0.0008,556 "step": 3700557 },558 {559 "epoch": 1.7257462686567164,560 "eval_loss": 7.432493021042319e-06,561 "eval_runtime": 0.1563,562 "eval_samples_per_second": 191.964,563 "eval_steps_per_second": 25.595,564 "step": 3700565 },566 {567 "epoch": 1.7723880597014925,568 "grad_norm": 0.004110433626919985,569 "learning_rate": 2.5828358208955225e-05,570 "loss": 0.0014,571 "step": 3800572 },573 {574 "epoch": 1.7723880597014925,575 "eval_loss": 7.303169240913121e-06,576 "eval_runtime": 0.1569,577 "eval_samples_per_second": 191.177,578 "eval_steps_per_second": 25.49,579 "step": 3800580 },581 {582 "epoch": 1.8190298507462686,583 "grad_norm": 0.0039463951252400875,584 "learning_rate": 2.5455223880597017e-05,585 "loss": 0.0015,586 "step": 3900587 },588 {589 "epoch": 1.8190298507462686,590 "eval_loss": 9.76303817878943e-06,591 "eval_runtime": 0.1571,592 "eval_samples_per_second": 190.973,593 "eval_steps_per_second": 25.463,594 "step": 3900595 },596 {597 "epoch": 1.8656716417910446,598 "grad_norm": 0.004805909004062414,599 "learning_rate": 2.508208955223881e-05,600 "loss": 0.0012,601 "step": 4000602 },603 {604 "epoch": 1.8656716417910446,605 "eval_loss": 8.14036320662126e-06,606 "eval_runtime": 0.157,607 "eval_samples_per_second": 191.027,608 "eval_steps_per_second": 25.47,609 "step": 4000610 },611 {612 "epoch": 1.912313432835821,613 "grad_norm": 0.0031389566138386726,614 "learning_rate": 2.4708955223880597e-05,615 "loss": 0.0004,616 "step": 4100617 },618 {619 "epoch": 1.912313432835821,620 "eval_loss": 6.821030183346011e-06,621 "eval_runtime": 0.1614,622 "eval_samples_per_second": 185.9,623 "eval_steps_per_second": 24.787,624 "step": 4100625 },626 {627 "epoch": 1.9589552238805972,628 "grad_norm": 0.0017104181461036205,629 "learning_rate": 2.433582089552239e-05,630 "loss": 0.0008,631 "step": 4200632 },633 {634 "epoch": 1.9589552238805972,635 "eval_loss": 7.834916686988436e-06,636 "eval_runtime": 0.1615,637 "eval_samples_per_second": 185.757,638 "eval_steps_per_second": 24.768,639 "step": 4200640 },641 {642 "epoch": 2.0055970149253732,643 "grad_norm": 0.013449718244373798,644 "learning_rate": 2.396268656716418e-05,645 "loss": 0.001,646 "step": 4300647 },648 {649 "epoch": 2.0055970149253732,650 "eval_loss": 7.582441867270973e-06,651 "eval_runtime": 0.1686,652 "eval_samples_per_second": 177.913,653 "eval_steps_per_second": 23.722,654 "step": 4300655 },656 {657 "epoch": 2.0522388059701493,658 "grad_norm": 0.0017907052533701062,659 "learning_rate": 2.358955223880597e-05,660 "loss": 0.0007,661 "step": 4400662 },663 {664 "epoch": 2.0522388059701493,665 "eval_loss": 6.949214366613887e-06,666 "eval_runtime": 0.1537,667 "eval_samples_per_second": 195.234,668 "eval_steps_per_second": 26.031,669 "step": 4400670 },671 {672 "epoch": 2.0988805970149254,673 "grad_norm": 0.001830141874961555,674 "learning_rate": 2.3216417910447763e-05,675 "loss": 0.0004,676 "step": 4500677 },678 {679 "epoch": 2.0988805970149254,680 "eval_loss": 7.074564564391039e-06,681 "eval_runtime": 0.1577,682 "eval_samples_per_second": 190.227,683 "eval_steps_per_second": 25.364,684 "step": 4500685 },686 {687 "epoch": 2.1455223880597014,688 "grad_norm": 0.0030846677254885435,689 "learning_rate": 2.2843283582089555e-05,690 "loss": 0.0006,691 "step": 4600692 },693 {694 "epoch": 2.1455223880597014,695 "eval_loss": 6.291250883805333e-06,696 "eval_runtime": 0.1566,697 "eval_samples_per_second": 191.563,698 "eval_steps_per_second": 25.542,699 "step": 4600700 },701 {702 "epoch": 2.1921641791044775,703 "grad_norm": 0.24780245125293732,704 "learning_rate": 2.2470149253731343e-05,705 "loss": 0.0003,706 "step": 4700707 },708 {709 "epoch": 2.1921641791044775,710 "eval_loss": 6.1347077462414745e-06,711 "eval_runtime": 0.1575,712 "eval_samples_per_second": 190.525,713 "eval_steps_per_second": 25.403,714 "step": 4700715 },716 {717 "epoch": 2.2388059701492535,718 "grad_norm": 0.006296923384070396,719 "learning_rate": 2.2097014925373138e-05,720 "loss": 0.0003,721 "step": 4800722 },723 {724 "epoch": 2.2388059701492535,725 "eval_loss": 5.517637873708736e-06,726 "eval_runtime": 0.1579,727 "eval_samples_per_second": 189.97,728 "eval_steps_per_second": 25.329,729 "step": 4800730 },731 {732 "epoch": 2.28544776119403,733 "grad_norm": 0.0017776070162653923,734 "learning_rate": 2.1723880597014926e-05,735 "loss": 0.0007,736 "step": 4900737 },738 {739 "epoch": 2.28544776119403,740 "eval_loss": 5.534998763323529e-06,741 "eval_runtime": 0.158,742 "eval_samples_per_second": 189.867,743 "eval_steps_per_second": 25.316,744 "step": 4900745 },746 {747 "epoch": 2.332089552238806,748 "grad_norm": 0.0026376782916486263,749 "learning_rate": 2.1350746268656717e-05,750 "loss": 0.0004,751 "step": 5000752 },753 {754 "epoch": 2.332089552238806,755 "eval_loss": 5.3646053856937215e-06,756 "eval_runtime": 0.158,757 "eval_samples_per_second": 189.825,758 "eval_steps_per_second": 25.31,759 "step": 5000760 },761 {762 "epoch": 2.378731343283582,763 "grad_norm": 0.008729691617190838,764 "learning_rate": 2.097761194029851e-05,765 "loss": 0.0009,766 "step": 5100767 },768 {769 "epoch": 2.378731343283582,770 "eval_loss": 5.699926987290382e-06,771 "eval_runtime": 0.1586,772 "eval_samples_per_second": 189.183,773 "eval_steps_per_second": 25.224,774 "step": 5100775 },776 {777 "epoch": 2.425373134328358,778 "grad_norm": 0.0009663441451266408,779 "learning_rate": 2.06044776119403e-05,780 "loss": 0.0002,781 "step": 5200782 },783 {784 "epoch": 2.425373134328358,785 "eval_loss": 5.018339379603276e-06,786 "eval_runtime": 0.1609,787 "eval_samples_per_second": 186.434,788 "eval_steps_per_second": 24.858,789 "step": 5200790 },791 {792 "epoch": 2.4720149253731343,793 "grad_norm": 0.0009370720363222063,794 "learning_rate": 2.023134328358209e-05,795 "loss": 0.0003,796 "step": 5300797 },798 {799 "epoch": 2.4720149253731343,800 "eval_loss": 4.319118488638196e-06,801 "eval_runtime": 0.1594,802 "eval_samples_per_second": 188.258,803 "eval_steps_per_second": 25.101,804 "step": 5300805 },806 {807 "epoch": 2.5186567164179103,808 "grad_norm": 0.00313656241632998,809 "learning_rate": 1.9858208955223884e-05,810 "loss": 0.0008,811 "step": 5400812 },813 {814 "epoch": 2.5186567164179103,815 "eval_loss": 4.456574970390648e-06,816 "eval_runtime": 0.1551,817 "eval_samples_per_second": 193.476,818 "eval_steps_per_second": 25.797,819 "step": 5400820 },821 {822 "epoch": 2.5652985074626864,823 "grad_norm": 0.0011074623325839639,824 "learning_rate": 1.9485074626865675e-05,825 "loss": 0.001,826 "step": 5500827 },828 {829 "epoch": 2.5652985074626864,830 "eval_loss": 3.924767952412367e-06,831 "eval_runtime": 0.1544,832 "eval_samples_per_second": 194.285,833 "eval_steps_per_second": 25.905,834 "step": 5500835 },836 {837 "epoch": 2.611940298507463,838 "grad_norm": 0.02510717324912548,839 "learning_rate": 1.9111940298507467e-05,840 "loss": 0.0003,841 "step": 5600842 },843 {844 "epoch": 2.611940298507463,845 "eval_loss": 4.046037247462664e-06,846 "eval_runtime": 0.1553,847 "eval_samples_per_second": 193.136,848 "eval_steps_per_second": 25.752,849 "step": 5600850 },851 {852 "epoch": 2.658582089552239,853 "grad_norm": 0.023706955835223198,854 "learning_rate": 1.8738805970149255e-05,855 "loss": 0.0003,856 "step": 5700857 },858 {859 "epoch": 2.658582089552239,860 "eval_loss": 4.111709131393582e-06,861 "eval_runtime": 0.1558,862 "eval_samples_per_second": 192.543,863 "eval_steps_per_second": 25.672,864 "step": 5700865 },866 {867 "epoch": 2.705223880597015,868 "grad_norm": 0.004163182340562344,869 "learning_rate": 1.8365671641791047e-05,870 "loss": 0.0006,871 "step": 5800872 },873 {874 "epoch": 2.705223880597015,875 "eval_loss": 4.339735369285336e-06,876 "eval_runtime": 0.1607,877 "eval_samples_per_second": 186.695,878 "eval_steps_per_second": 24.893,879 "step": 5800880 },881 {882 "epoch": 2.751865671641791,883 "grad_norm": 0.002963052364066243,884 "learning_rate": 1.7992537313432835e-05,885 "loss": 0.0005,886 "step": 5900887 },888 {889 "epoch": 2.751865671641791,890 "eval_loss": 3.9786968955013435e-06,891 "eval_runtime": 0.1575,892 "eval_samples_per_second": 190.477,893 "eval_steps_per_second": 25.397,894 "step": 5900895 },896 {897 "epoch": 2.798507462686567,898 "grad_norm": 0.045626550912857056,899 "learning_rate": 1.7619402985074627e-05,900 "loss": 0.0008,901 "step": 6000902 },903 {904 "epoch": 2.798507462686567,905 "eval_loss": 4.052158146805596e-06,906 "eval_runtime": 0.1579,907 "eval_samples_per_second": 189.965,908 "eval_steps_per_second": 25.329,909 "step": 6000910 },911 {912 "epoch": 2.845149253731343,913 "grad_norm": 0.008129669353365898,914 "learning_rate": 1.7246268656716418e-05,915 "loss": 0.0004,916 "step": 6100917 },918 {919 "epoch": 2.845149253731343,920 "eval_loss": 4.024254849355202e-06,921 "eval_runtime": 0.158,922 "eval_samples_per_second": 189.912,923 "eval_steps_per_second": 25.322,924 "step": 6100925 },926 {927 "epoch": 2.8917910447761193,928 "grad_norm": 0.002264247043058276,929 "learning_rate": 1.687313432835821e-05,930 "loss": 0.0007,931 "step": 6200932 },933 {934 "epoch": 2.8917910447761193,935 "eval_loss": 3.7680488276237156e-06,936 "eval_runtime": 0.1559,937 "eval_samples_per_second": 192.455,938 "eval_steps_per_second": 25.661,939 "step": 6200940 },941 {942 "epoch": 2.9384328358208958,943 "grad_norm": 0.013498159125447273,944 "learning_rate": 1.65e-05,945 "loss": 0.0002,946 "step": 6300947 },948 {949 "epoch": 2.9384328358208958,950 "eval_loss": 3.534241386660142e-06,951 "eval_runtime": 0.1562,952 "eval_samples_per_second": 192.066,953 "eval_steps_per_second": 25.609,954 "step": 6300955 },956 {957 "epoch": 2.9850746268656714,958 "grad_norm": 0.000992327812127769,959 "learning_rate": 1.6126865671641793e-05,960 "loss": 0.0009,961 "step": 6400962 },963 {964 "epoch": 2.9850746268656714,965 "eval_loss": 3.539219051162945e-06,966 "eval_runtime": 0.1581,967 "eval_samples_per_second": 189.704,968 "eval_steps_per_second": 25.294,969 "step": 6400970 },971 {972 "epoch": 3.031716417910448,973 "grad_norm": 0.002681414596736431,974 "learning_rate": 1.575373134328358e-05,975 "loss": 0.0003,976 "step": 6500977 },978 {979 "epoch": 3.031716417910448,980 "eval_loss": 3.239377292629797e-06,981 "eval_runtime": 0.1572,982 "eval_samples_per_second": 190.818,983 "eval_steps_per_second": 25.442,984 "step": 6500985 },986 {987 "epoch": 3.078358208955224,988 "grad_norm": 0.06693683564662933,989 "learning_rate": 1.5380597014925373e-05,990 "loss": 0.0003,991 "step": 6600992 },993 {994 "epoch": 3.078358208955224,995 "eval_loss": 3.1195118026516866e-06,996 "eval_runtime": 0.1554,997 "eval_samples_per_second": 192.996,998 "eval_steps_per_second": 25.733,999 "step": 66001000 },1001 {1002 "epoch": 3.125,1003 "grad_norm": 0.007976911030709743,1004 "learning_rate": 1.5007462686567164e-05,1005 "loss": 0.0002,1006 "step": 67001007 },1008 {1009 "epoch": 3.125,1010 "eval_loss": 3.158217623422388e-06,1011 "eval_runtime": 0.1613,1012 "eval_samples_per_second": 186.004,1013 "eval_steps_per_second": 24.801,1014 "step": 67001015 },1016 {1017 "epoch": 3.171641791044776,1018 "grad_norm": 0.0018902523443102837,1019 "learning_rate": 1.4634328358208956e-05,1020 "loss": 0.0002,1021 "step": 68001022 },1023 {1024 "epoch": 3.171641791044776,1025 "eval_loss": 3.020254553121049e-06,1026 "eval_runtime": 0.1619,1027 "eval_samples_per_second": 185.322,1028 "eval_steps_per_second": 24.71,1029 "step": 68001030 },1031 {1032 "epoch": 3.218283582089552,1033 "grad_norm": 0.0031356920953840017,1034 "learning_rate": 1.4261194029850747e-05,1035 "loss": 0.0002,1036 "step": 69001037 },1038 {1039 "epoch": 3.218283582089552,1040 "eval_loss": 2.9624054604937555e-06,1041 "eval_runtime": 0.1556,1042 "eval_samples_per_second": 192.812,1043 "eval_steps_per_second": 25.708,1044 "step": 69001045 },1046 {1047 "epoch": 3.264925373134328,1048 "grad_norm": 0.005576052237302065,1049 "learning_rate": 1.3888059701492537e-05,1050 "loss": 0.0005,1051 "step": 70001052 },1053 {1054 "epoch": 3.264925373134328,1055 "eval_loss": 2.9549014470831025e-06,1056 "eval_runtime": 0.1578,1057 "eval_samples_per_second": 190.15,1058 "eval_steps_per_second": 25.353,1059 "step": 70001060 },1061 {1062 "epoch": 3.3115671641791042,1063 "grad_norm": 0.0018602815689519048,1064 "learning_rate": 1.3514925373134329e-05,1065 "loss": 0.0002,1066 "step": 71001067 },1068 {1069 "epoch": 3.3115671641791042,1070 "eval_loss": 2.8336680770735256e-06,1071 "eval_runtime": 0.1552,1072 "eval_samples_per_second": 193.246,1073 "eval_steps_per_second": 25.766,1074 "step": 71001075 },1076 {1077 "epoch": 3.3582089552238807,1078 "grad_norm": 0.002030579838901758,1079 "learning_rate": 1.314179104477612e-05,1080 "loss": 0.0003,1081 "step": 72001082 },1083 {1084 "epoch": 3.3582089552238807,1085 "eval_loss": 2.7239029805059545e-06,1086 "eval_runtime": 0.1624,1087 "eval_samples_per_second": 184.688,1088 "eval_steps_per_second": 24.625,1089 "step": 72001090 },1091 {1092 "epoch": 3.404850746268657,1093 "grad_norm": 0.027261830866336823,1094 "learning_rate": 1.276865671641791e-05,1095 "loss": 0.0003,1096 "step": 73001097 },1098 {1099 "epoch": 3.404850746268657,1100 "eval_loss": 2.706149643927347e-06,1101 "eval_runtime": 0.1586,1102 "eval_samples_per_second": 189.213,1103 "eval_steps_per_second": 25.228,1104 "step": 73001105 },1106 {1107 "epoch": 3.451492537313433,1108 "grad_norm": 0.005287382751703262,1109 "learning_rate": 1.2395522388059702e-05,1110 "loss": 0.0002,1111 "step": 74001112 },1113 {1114 "epoch": 3.451492537313433,1115 "eval_loss": 2.5531167011649814e-06,1116 "eval_runtime": 0.1543,1117 "eval_samples_per_second": 194.469,1118 "eval_steps_per_second": 25.929,1119 "step": 74001120 },1121 {1122 "epoch": 3.498134328358209,1123 "grad_norm": 0.0021771446336060762,1124 "learning_rate": 1.2022388059701493e-05,1125 "loss": 0.0003,1126 "step": 75001127 },1128 {1129 "epoch": 3.498134328358209,1130 "eval_loss": 2.8221681986906333e-06,1131 "eval_runtime": 0.1601,1132 "eval_samples_per_second": 187.36,1133 "eval_steps_per_second": 24.981,1134 "step": 75001135 },1136 {1137 "epoch": 3.544776119402985,1138 "grad_norm": 0.0027294817846268415,1139 "learning_rate": 1.1649253731343283e-05,1140 "loss": 0.0003,1141 "step": 76001142 },1143 {1144 "epoch": 3.544776119402985,1145 "eval_loss": 2.784717707982054e-06,1146 "eval_runtime": 0.1555,1147 "eval_samples_per_second": 192.876,1148 "eval_steps_per_second": 25.717,1149 "step": 76001150 },1151 {1152 "epoch": 3.591417910447761,1153 "grad_norm": 0.0013641875702887774,1154 "learning_rate": 1.1276119402985075e-05,1155 "loss": 0.0003,1156 "step": 77001157 },1158 {1159 "epoch": 3.591417910447761,1160 "eval_loss": 2.9222812827356393e-06,1161 "eval_runtime": 0.1639,1162 "eval_samples_per_second": 183.034,1163 "eval_steps_per_second": 24.405,1164 "step": 77001165 },1166 {1167 "epoch": 3.638059701492537,1168 "grad_norm": 0.08378789573907852,1169 "learning_rate": 1.0902985074626867e-05,1170 "loss": 0.0002,1171 "step": 78001172 },1173 {1174 "epoch": 3.638059701492537,1175 "eval_loss": 2.9144475774955936e-06,1176 "eval_runtime": 0.1585,1177 "eval_samples_per_second": 189.266,1178 "eval_steps_per_second": 25.235,1179 "step": 78001180 },1181 {1182 "epoch": 3.6847014925373136,1183 "grad_norm": 0.004190696403384209,1184 "learning_rate": 1.0529850746268656e-05,1185 "loss": 0.0002,1186 "step": 79001187 },1188 {1189 "epoch": 3.6847014925373136,1190 "eval_loss": 2.752276259343489e-06,1191 "eval_runtime": 0.1641,1192 "eval_samples_per_second": 182.838,1193 "eval_steps_per_second": 24.378,1194 "step": 79001195 },1196 {1197 "epoch": 3.7313432835820897,1198 "grad_norm": 0.0028494582511484623,1199 "learning_rate": 1.0156716417910448e-05,1200 "loss": 0.0005,