dzungpham/graphcodebert-code-classification
0
1{2 "best_global_step": 1350,3 "best_metric": 0.5986622073578596,4 "best_model_checkpoint": "output_checkpoints/graphcodebert-best/checkpoint-1350",5 "epoch": 1.3209393346379648,6 "eval_steps": 50,7 "global_step": 1350,8 "is_hyper_param_search": false,9 "is_local_process_zero": true,10 "is_world_process_zero": true,11 "log_history": [12 {13 "epoch": 0.009784735812133072,14 "grad_norm": 165951.1875,15 "learning_rate": 4.411764705882353e-08,16 "loss": 0.8416,17 "step": 1018 },19 {20 "epoch": 0.019569471624266144,21 "grad_norm": 206132.796875,22 "learning_rate": 9.313725490196079e-08,23 "loss": 0.8378,24 "step": 2025 },26 {27 "epoch": 0.029354207436399216,28 "grad_norm": 171197.53125,29 "learning_rate": 1.4215686274509803e-07,30 "loss": 0.8388,31 "step": 3032 },33 {34 "epoch": 0.03913894324853229,35 "grad_norm": 206661.828125,36 "learning_rate": 1.9117647058823527e-07,37 "loss": 0.847,38 "step": 4039 },40 {41 "epoch": 0.04892367906066536,42 "grad_norm": 230268.96875,43 "learning_rate": 2.4019607843137255e-07,44 "loss": 0.8411,45 "step": 5046 },47 {48 "epoch": 0.05870841487279843,49 "grad_norm": 133248.28125,50 "learning_rate": 2.8921568627450984e-07,51 "loss": 0.8412,52 "step": 6053 },54 {55 "epoch": 0.0684931506849315,56 "grad_norm": 147768.828125,57 "learning_rate": 3.3823529411764707e-07,58 "loss": 0.8431,59 "step": 7060 },61 {62 "epoch": 0.07827788649706457,63 "grad_norm": 167327.53125,64 "learning_rate": 3.872549019607843e-07,65 "loss": 0.8431,66 "step": 8067 },68 {69 "epoch": 0.08806262230919765,70 "grad_norm": 151943.75,71 "learning_rate": 4.3627450980392154e-07,72 "loss": 0.8311,73 "step": 9074 },75 {76 "epoch": 0.09784735812133072,77 "grad_norm": 149971.59375,78 "learning_rate": 4.852941176470588e-07,79 "loss": 0.8412,80 "step": 10081 },82 {83 "epoch": 0.10763209393346379,84 "grad_norm": 162329.40625,85 "learning_rate": 5.34313725490196e-07,86 "loss": 0.8356,87 "step": 11088 },89 {90 "epoch": 0.11741682974559686,91 "grad_norm": 158769.765625,92 "learning_rate": 5.833333333333334e-07,93 "loss": 0.8351,94 "step": 12095 },96 {97 "epoch": 0.12720156555772993,98 "grad_norm": 157725.59375,99 "learning_rate": 6.323529411764706e-07,100 "loss": 0.8331,101 "step": 130102 },103 {104 "epoch": 0.136986301369863,105 "grad_norm": 205733.0,106 "learning_rate": 6.813725490196079e-07,107 "loss": 0.8366,108 "step": 140109 },110 {111 "epoch": 0.14677103718199608,112 "grad_norm": 172581.328125,113 "learning_rate": 7.30392156862745e-07,114 "loss": 0.831,115 "step": 150116 },117 {118 "epoch": 0.15655577299412915,119 "grad_norm": 137428.0,120 "learning_rate": 7.794117647058823e-07,121 "loss": 0.8326,122 "step": 160123 },124 {125 "epoch": 0.16634050880626222,126 "grad_norm": 151189.203125,127 "learning_rate": 8.284313725490196e-07,128 "loss": 0.8281,129 "step": 170130 },131 {132 "epoch": 0.1761252446183953,133 "grad_norm": 145620.625,134 "learning_rate": 8.774509803921568e-07,135 "loss": 0.8252,136 "step": 180137 },138 {139 "epoch": 0.18590998043052837,140 "grad_norm": 169938.84375,141 "learning_rate": 9.264705882352941e-07,142 "loss": 0.8308,143 "step": 190144 },145 {146 "epoch": 0.19569471624266144,147 "grad_norm": 138760.015625,148 "learning_rate": 9.754901960784313e-07,149 "loss": 0.824,150 "step": 200151 },152 {153 "epoch": 0.2054794520547945,154 "grad_norm": 107872.125,155 "learning_rate": 9.99631294374483e-07,156 "loss": 0.8254,157 "step": 210158 },159 {160 "epoch": 0.21526418786692758,161 "grad_norm": 113610.8359375,162 "learning_rate": 9.966849112204936e-07,163 "loss": 0.8267,164 "step": 220165 },166 {167 "epoch": 0.22504892367906065,168 "grad_norm": 172441.515625,169 "learning_rate": 9.908095200725339e-07,170 "loss": 0.8198,171 "step": 230172 },173 {174 "epoch": 0.23483365949119372,175 "grad_norm": 128079.9140625,176 "learning_rate": 9.82039768787326e-07,177 "loss": 0.8143,178 "step": 240179 },180 {181 "epoch": 0.2446183953033268,182 "grad_norm": 98149.8984375,183 "learning_rate": 9.70427373595903e-07,184 "loss": 0.8191,185 "step": 250186 },187 {188 "epoch": 0.25440313111545987,189 "grad_norm": 147377.4375,190 "learning_rate": 9.560408141270531e-07,191 "loss": 0.8178,192 "step": 260193 },194 {195 "epoch": 0.26418786692759294,196 "grad_norm": 156439.40625,197 "learning_rate": 9.389649295750781e-07,198 "loss": 0.8085,199 "step": 270200 },201 {202 "epoch": 0.273972602739726,203 "grad_norm": 131892.015625,204 "learning_rate": 9.193004183933174e-07,205 "loss": 0.8172,206 "step": 280207 },208 {209 "epoch": 0.2837573385518591,210 "grad_norm": 132474.5,211 "learning_rate": 8.971632444637971e-07,212 "loss": 0.8136,213 "step": 290214 },215 {216 "epoch": 0.29354207436399216,217 "grad_norm": 112619.109375,218 "learning_rate": 8.726839532448928e-07,219 "loss": 0.8065,220 "step": 300221 },222 {223 "epoch": 0.30332681017612523,224 "grad_norm": 94225.5703125,225 "learning_rate": 8.460069019297606e-07,226 "loss": 0.813,227 "step": 310228 },229 {230 "epoch": 0.3131115459882583,231 "grad_norm": 117812.8203125,232 "learning_rate": 8.172894081553822e-07,233 "loss": 0.8022,234 "step": 320235 },236 {237 "epoch": 0.32289628180039137,238 "grad_norm": 143287.125,239 "learning_rate": 7.86700822282382e-07,240 "loss": 0.8061,241 "step": 330242 },243 {244 "epoch": 0.33268101761252444,245 "grad_norm": 107456.25,246 "learning_rate": 7.544215287164917e-07,247 "loss": 0.8006,248 "step": 340249 },250 {251 "epoch": 0.3424657534246575,252 "grad_norm": 101521.171875,253 "learning_rate": 7.206418821609812e-07,254 "loss": 0.8055,255 "step": 350256 },257 {258 "epoch": 0.3522504892367906,259 "grad_norm": 118311.5234375,260 "learning_rate": 6.855610850731017e-07,261 "loss": 0.8081,262 "step": 360263 },264 {265 "epoch": 0.36203522504892366,266 "grad_norm": 89621.828125,267 "learning_rate": 6.493860129443046e-07,268 "loss": 0.8058,269 "step": 370270 },271 {272 "epoch": 0.37181996086105673,273 "grad_norm": 143168.65625,274 "learning_rate": 6.123299943316975e-07,275 "loss": 0.8028,276 "step": 380277 },278 {279 "epoch": 0.3816046966731898,280 "grad_norm": 118598.2890625,281 "learning_rate": 5.746115528350295e-07,282 "loss": 0.8032,283 "step": 390284 },285 {286 "epoch": 0.3913894324853229,287 "grad_norm": 111807.8203125,288 "learning_rate": 5.364531184379139e-07,289 "loss": 0.8014,290 "step": 400291 },292 {293 "epoch": 0.40117416829745595,294 "grad_norm": 124106.1328125,295 "learning_rate": 4.980797158126566e-07,296 "loss": 0.8045,297 "step": 410298 },299 {300 "epoch": 0.410958904109589,301 "grad_norm": 102746.4765625,302 "learning_rate": 4.597176373239061e-07,303 "loss": 0.7996,304 "step": 420305 },306 {307 "epoch": 0.4207436399217221,308 "grad_norm": 108317.5859375,309 "learning_rate": 4.215931085565761e-07,310 "loss": 0.8037,311 "step": 430312 },313 {314 "epoch": 0.43052837573385516,315 "grad_norm": 113342.515625,316 "learning_rate": 3.8393095423757764e-07,317 "loss": 0.8047,318 "step": 440319 },320 {321 "epoch": 0.44031311154598823,322 "grad_norm": 111908.90625,323 "learning_rate": 3.469532724185699e-07,324 "loss": 0.8021,325 "step": 450326 },327 {328 "epoch": 0.4500978473581213,329 "grad_norm": 85004.15625,330 "learning_rate": 3.1087812473822845e-07,331 "loss": 0.8012,332 "step": 460333 },334 {335 "epoch": 0.4598825831702544,336 "grad_norm": 98335.1796875,337 "learning_rate": 2.7591825048770643e-07,338 "loss": 0.7977,339 "step": 470340 },341 {342 "epoch": 0.46966731898238745,343 "grad_norm": 110674.03125,344 "learning_rate": 2.422798120625886e-07,345 "loss": 0.7979,346 "step": 480347 },348 {349 "epoch": 0.4794520547945205,350 "grad_norm": 114861.5390625,351 "learning_rate": 2.1016117919955512e-07,352 "loss": 0.8014,353 "step": 490354 },355 {356 "epoch": 0.4892367906066536,357 "grad_norm": 130109.5703125,358 "learning_rate": 1.7975175916725031e-07,359 "loss": 0.7994,360 "step": 500361 },362 {363 "epoch": 0.49902152641878667,364 "grad_norm": 93286.4296875,365 "learning_rate": 1.5123087980985606e-07,366 "loss": 0.8039,367 "step": 510368 },369 {370 "epoch": 0.5088062622309197,371 "grad_norm": 123646.703125,372 "learning_rate": 1.2476673203018728e-07,373 "loss": 0.7917,374 "step": 520375 },376 {377 "epoch": 0.5185909980430529,378 "grad_norm": 98060.2265625,379 "learning_rate": 1.0051537794861287e-07,380 "loss": 0.7945,381 "step": 530382 },383 {384 "epoch": 0.5283757338551859,385 "grad_norm": 91181.1796875,386 "learning_rate": 7.861983058679871e-08,387 "loss": 0.7952,388 "step": 540389 },390 {391 "epoch": 0.538160469667319,392 "grad_norm": 97605.3984375,393 "learning_rate": 5.920921050348626e-08,394 "loss": 0.7963,395 "step": 550396 },397 {398 "epoch": 0.538160469667319,399 "eval_accuracy": 0.767,400 "eval_loss": 0.662277102470398,401 "eval_macro_f1": 0.5341406895117674,402 "eval_precision": 0.6105222734254993,403 "eval_recall": 0.5415245482510056,404 "eval_runtime": 16.9722,405 "eval_samples_per_second": 58.92,406 "eval_steps_per_second": 0.059,407 "step": 550408 },409 {410 "epoch": 0.547945205479452,411 "grad_norm": 101627.8046875,412 "learning_rate": 4.23979843557199e-08,413 "loss": 0.7962,414 "step": 560415 },416 {417 "epoch": 0.5577299412915852,418 "grad_norm": 73889.578125,419 "learning_rate": 2.8285289875818407e-08,420 "loss": 0.794,421 "step": 570422 },423 {424 "epoch": 0.5675146771037182,425 "grad_norm": 87191.0546875,426 "learning_rate": 1.695435124477512e-08,427 "loss": 0.7984,428 "step": 580429 },430 {431 "epoch": 0.5772994129158513,432 "grad_norm": 105470.234375,433 "learning_rate": 8.471988309699807e-09,434 "loss": 0.7924,435 "step": 590436 },437 {438 "epoch": 0.5870841487279843,439 "grad_norm": 81710.7265625,440 "learning_rate": 2.8882225395023275e-09,441 "loss": 0.8005,442 "step": 600443 },444 {445 "epoch": 0.5870841487279843,446 "eval_accuracy": 0.767,447 "eval_loss": 0.6623347401618958,448 "eval_macro_f1": 0.5341406895117674,449 "eval_precision": 0.6105222734254993,450 "eval_recall": 0.5415245482510056,451 "eval_runtime": 17.0174,452 "eval_samples_per_second": 58.763,453 "eval_steps_per_second": 0.059,454 "step": 600455 },456 {457 "epoch": 0.5968688845401174,458 "grad_norm": 97212.125,459 "learning_rate": 2.35982042542604e-10,460 "loss": 0.7981,461 "step": 610462 },463 {464 "epoch": 0.6066536203522505,465 "grad_norm": 100700.640625,466 "learning_rate": 9.994690926142082e-07,467 "loss": 0.8006,468 "step": 620469 },470 {471 "epoch": 0.6164383561643836,472 "grad_norm": 88023.046875,473 "learning_rate": 9.962287406393883e-07,474 "loss": 0.7976,475 "step": 630476 },477 {478 "epoch": 0.6262230919765166,479 "grad_norm": 118936.609375,480 "learning_rate": 9.900620707609317e-07,481 "loss": 0.7961,482 "step": 640483 },484 {485 "epoch": 0.6360078277886497,486 "grad_norm": 91811.3984375,487 "learning_rate": 9.810054485397043e-07,488 "loss": 0.7904,489 "step": 650490 },491 {492 "epoch": 0.6360078277886497,493 "eval_accuracy": 0.757,494 "eval_loss": 0.6629467606544495,495 "eval_macro_f1": 0.5299579283330915,496 "eval_precision": 0.5844492268452096,497 "eval_recall": 0.5366881936388663,498 "eval_runtime": 17.0504,499 "eval_samples_per_second": 58.65,500 "eval_steps_per_second": 0.059,501 "step": 650502 },503 {504 "epoch": 0.6457925636007827,505 "grad_norm": 118632.46875,506 "learning_rate": 9.69112281917613e-07,507 "loss": 0.7927,508 "step": 660509 },510 {511 "epoch": 0.6555772994129159,512 "grad_norm": 70413.0390625,513 "learning_rate": 9.544527062648345e-07,514 "loss": 0.785,515 "step": 670516 },517 {518 "epoch": 0.6653620352250489,519 "grad_norm": 84746.4453125,520 "learning_rate": 9.371131707834789e-07,521 "loss": 0.7861,522 "step": 680523 },524 {525 "epoch": 0.675146771037182,526 "grad_norm": 92019.0390625,527 "learning_rate": 9.171959287067115e-07,528 "loss": 0.7892,529 "step": 690530 },531 {532 "epoch": 0.684931506849315,533 "grad_norm": 70649.2578125,534 "learning_rate": 8.948184342996868e-07,535 "loss": 0.789,536 "step": 700537 },538 {539 "epoch": 0.684931506849315,540 "eval_accuracy": 0.751,541 "eval_loss": 0.6648359894752502,542 "eval_macro_f1": 0.5514410631387237,543 "eval_precision": 0.589223697650664,544 "eval_recall": 0.5504123598294002,545 "eval_runtime": 17.0124,546 "eval_samples_per_second": 58.781,547 "eval_steps_per_second": 0.059,548 "step": 700549 },550 {551 "epoch": 0.6947162426614482,552 "grad_norm": 80338.796875,553 "learning_rate": 8.701126502182503e-07,554 "loss": 0.7907,555 "step": 710556 },557 {558 "epoch": 0.7045009784735812,559 "grad_norm": 69257.328125,560 "learning_rate": 8.432242693099946e-07,561 "loss": 0.7881,562 "step": 720563 },564 {565 "epoch": 0.7142857142857143,566 "grad_norm": 95345.15625,567 "learning_rate": 8.14311855446795e-07,568 "loss": 0.7849,569 "step": 730570 },571 {572 "epoch": 0.7240704500978473,573 "grad_norm": 94407.8984375,574 "learning_rate": 7.835459084554374e-07,575 "loss": 0.7868,576 "step": 740577 },578 {579 "epoch": 0.7338551859099804,580 "grad_norm": 74418.0625,581 "learning_rate": 7.511078586605484e-07,582 "loss": 0.783,583 "step": 750584 },585 {586 "epoch": 0.7338551859099804,587 "eval_accuracy": 0.742,588 "eval_loss": 0.6660248637199402,589 "eval_macro_f1": 0.5532467532467532,590 "eval_precision": 0.579727791757989,591 "eval_recall": 0.5510154613293626,592 "eval_runtime": 16.9303,593 "eval_samples_per_second": 59.066,594 "eval_steps_per_second": 0.059,595 "step": 750596 },597 {598 "epoch": 0.7436399217221135,599 "grad_norm": 95142.0546875,600 "learning_rate": 7.171889969691225e-07,601 "loss": 0.7861,602 "step": 760603 },604 {605 "epoch": 0.7534246575342466,606 "grad_norm": 118907.203125,607 "learning_rate": 6.819893468060643e-07,608 "loss": 0.7878,609 "step": 770610 },611 {612 "epoch": 0.7632093933463796,613 "grad_norm": 66268.78125,614 "learning_rate": 6.457164845530662e-07,615 "loss": 0.7859,616 "step": 780617 },618 {619 "epoch": 0.7729941291585127,620 "grad_norm": 81464.265625,621 "learning_rate": 6.085843154468354e-07,622 "loss": 0.7819,623 "step": 790624 },625 {626 "epoch": 0.7827788649706457,627 "grad_norm": 87098.984375,628 "learning_rate": 5.708118121553318e-07,629 "loss": 0.7801,630 "step": 800631 },632 {633 "epoch": 0.7827788649706457,634 "eval_accuracy": 0.734,635 "eval_loss": 0.667039155960083,636 "eval_macro_f1": 0.5514001038862917,637 "eval_precision": 0.5710358542434345,638 "eval_recall": 0.5490647598270917,639 "eval_runtime": 17.0236,640 "eval_samples_per_second": 58.742,641 "eval_steps_per_second": 0.059,642 "step": 800643 },644 {645 "epoch": 0.7925636007827789,646 "grad_norm": 70466.7109375,647 "learning_rate": 5.326217234707852e-07,648 "loss": 0.7782,649 "step": 810650 },651 {652 "epoch": 0.8023483365949119,653 "grad_norm": 78038.6015625,654 "learning_rate": 4.942392607344717e-07,655 "loss": 0.786,656 "step": 820657 },658 {659 "epoch": 0.812133072407045,660 "grad_norm": 91227.28125,661 "learning_rate": 4.558907697395553e-07,662 "loss": 0.7781,663 "step": 830664 },665 {666 "epoch": 0.821917808219178,667 "grad_norm": 85470.5625,668 "learning_rate": 4.1780239594393807e-07,669 "loss": 0.7806,670 "step": 840671 },672 {673 "epoch": 0.8317025440313112,674 "grad_norm": 73229.96875,675 "learning_rate": 3.8019875086450787e-07,676 "loss": 0.7823,677 "step": 850678 },679 {680 "epoch": 0.8317025440313112,681 "eval_accuracy": 0.725,682 "eval_loss": 0.668379545211792,683 "eval_macro_f1": 0.5541786430494829,684 "eval_precision": 0.5667714002224962,685 "eval_recall": 0.5512665131499213,686 "eval_runtime": 16.9299,687 "eval_samples_per_second": 59.067,688 "eval_steps_per_second": 0.059,689 "step": 850690 },691 {692 "epoch": 0.8414872798434442,693 "grad_norm": 84347.0546875,694 "learning_rate": 3.4330158751721504e-07,695 "loss": 0.7805,696 "step": 860697 },698 {699 "epoch": 0.8512720156555773,700 "grad_norm": 60984.21875,701 "learning_rate": 3.0732849271405547e-07,702 "loss": 0.779,703 "step": 870704 },705 {706 "epoch": 0.8610567514677103,707 "grad_norm": 72854.8046875,708 "learning_rate": 2.7249160392863903e-07,709 "loss": 0.7811,710 "step": 880711 },712 {713 "epoch": 0.8708414872798435,714 "grad_norm": 69136.65625,715 "learning_rate": 2.3899635829713077e-07,716 "loss": 0.7811,717 "step": 890718 },719 {720 "epoch": 0.8806262230919765,721 "grad_norm": 74345.7421875,722 "learning_rate": 2.0704028113185306e-07,723 "loss": 0.7845,724 "step": 900725 },726 {727 "epoch": 0.8806262230919765,728 "eval_accuracy": 0.726,729 "eval_loss": 0.6684988737106323,730 "eval_macro_f1": 0.5584519389438943,731 "eval_precision": 0.5706882639048261,732 "eval_recall": 0.555107317439156,733 "eval_runtime": 16.9904,734 "eval_samples_per_second": 58.857,735 "eval_steps_per_second": 0.059,736 "step": 900737 },738 {739 "epoch": 0.8904109589041096,740 "grad_norm": 75862.8046875,741 "learning_rate": 1.768118210918219e-07,742 "loss": 0.7821,743 "step": 910744 },745 {746 "epoch": 0.9001956947162426,747 "grad_norm": 60382.25,748 "learning_rate": 1.484892388793511e-07,749 "loss": 0.7811,750 "step": 920751 },752 {753 "epoch": 0.9099804305283757,754 "grad_norm": 67346.859375,755 "learning_rate": 1.2223955601620633e-07,756 "loss": 0.781,757 "step": 930758 },759 {760 "epoch": 0.9197651663405088,761 "grad_norm": 81895.7265625,762 "learning_rate": 9.821756989850016e-08,763 "loss": 0.7816,764 "step": 940765 },766 {767 "epoch": 0.9295499021526419,768 "grad_norm": 87715.9453125,769 "learning_rate": 7.656494093865984e-08,770 "loss": 0.7815,771 "step": 950772 },773 {774 "epoch": 0.9295499021526419,775 "eval_accuracy": 0.723,776 "eval_loss": 0.6687072515487671,777 "eval_macro_f1": 0.5562540149399822,778 "eval_precision": 0.5672044579297467,779 "eval_recall": 0.553176815508654,780 "eval_runtime": 16.997,781 "eval_samples_per_second": 58.834,782 "eval_steps_per_second": 0.059,783 "step": 950784 },785 {786 "epoch": 0.9393346379647749,787 "grad_norm": 62333.75390625,788 "learning_rate": 5.740935717769707e-08,789 "loss": 0.775,790 "step": 960791 },792 {793 "epoch": 0.949119373776908,794 "grad_norm": 67404.265625,795 "learning_rate": 4.0863781294158185e-08,796 "loss": 0.78,797 "step": 970798 },799 {800 "epoch": 0.958904109589041,801 "grad_norm": 75798.1328125,802 "learning_rate": 2.702578445022852e-08,803 "loss": 0.7808,804 "step": 980805 },806 {807 "epoch": 0.9686888454011742,808 "grad_norm": 77483.9609375,809 "learning_rate": 1.597697090337763e-08,810 "loss": 0.7804,811 "step": 990812 },813 {814 "epoch": 0.9784735812133072,815 "grad_norm": 82261.78125,816 "learning_rate": 7.782496776675695e-09,817 "loss": 0.7795,818 "step": 1000819 },820 {821 "epoch": 0.9784735812133072,822 "eval_accuracy": 0.723,823 "eval_loss": 0.6687466502189636,824 "eval_macro_f1": 0.5579874225879597,825 "eval_precision": 0.5685547947184422,826 "eval_recall": 0.5547754673315212,827 "eval_runtime": 16.934,828 "eval_samples_per_second": 59.053,829 "eval_steps_per_second": 0.059,830 "step": 1000831 },832 {833 "epoch": 0.9882583170254403,834 "grad_norm": 68659.8671875,835 "learning_rate": 2.4906858256491524e-09,836 "loss": 0.78,837 "step": 1010838 },839 {840 "epoch": 0.9980430528375733,841 "grad_norm": 92298.78125,842 "learning_rate": 1.3274446754157898e-10,843 "loss": 0.7824,844 "step": 1020845 },846 {847 "epoch": 1.0078277886497065,848 "grad_norm": 47534.20703125,849 "learning_rate": 2.607950146993215e-08,850 "loss": 0.7439,851 "step": 1030852 },853 {854 "epoch": 1.0176125244618395,855 "grad_norm": 43369.4765625,856 "learning_rate": 2.09146573606756e-08,857 "loss": 0.741,858 "step": 1040859 },860 {861 "epoch": 1.0273972602739727,862 "grad_norm": 48618.94140625,863 "learning_rate": 1.6308405330324294e-08,864 "loss": 0.742,865 "step": 1050866 },867 {868 "epoch": 1.0273972602739727,869 "eval_accuracy": 0.723,870 "eval_loss": 0.668785035610199,871 "eval_macro_f1": 0.5597017416418834,872 "eval_precision": 0.5698832417582418,873 "eval_recall": 0.5563741191543882,874 "eval_runtime": 17.6735,875 "eval_samples_per_second": 56.582,876 "eval_steps_per_second": 0.057,877 "step": 1050878 },879 {880 "epoch": 1.0371819960861057,881 "grad_norm": 47853.890625,882 "learning_rate": 1.2266116062696951e-08,883 "loss": 0.7434,884 "step": 1060885 },886 {887 "epoch": 1.0469667318982387,888 "grad_norm": 58555.92578125,889 "learning_rate": 8.792502686258752e-09,890 "loss": 0.7448,891 "step": 1070892 },893 {894 "epoch": 1.0567514677103718,895 "grad_norm": 54508.24609375,896 "learning_rate": 5.891615278823537e-09,897 "loss": 0.7406,898 "step": 1080899 },900 {901 "epoch": 1.0665362035225048,902 "grad_norm": 51657.953125,903 "learning_rate": 3.5668361453450313e-09,904 "loss": 0.7427,905 "step": 1090906 },907 {908 "epoch": 1.076320939334638,909 "grad_norm": 54486.4921875,910 "learning_rate": 1.8208758743000207e-09,911 "loss": 0.7444,912 "step": 1100913 },914 {915 "epoch": 1.076320939334638,916 "eval_accuracy": 0.723,917 "eval_loss": 0.6687813997268677,918 "eval_macro_f1": 0.5597017416418834,919 "eval_precision": 0.5698832417582418,920 "eval_recall": 0.5563741191543882,921 "eval_runtime": 17.4461,922 "eval_samples_per_second": 57.319,923 "eval_steps_per_second": 0.057,924 "step": 1100925 },926 {927 "epoch": 1.086105675146771,928 "grad_norm": 42684.90625,929 "learning_rate": 6.557701772635371e-10,930 "loss": 0.7402,931 "step": 1110932 },933 {934 "epoch": 1.095890410958904,935 "grad_norm": 43652.2265625,936 "learning_rate": 7.287751536050324e-11,937 "loss": 0.742,938 "step": 1120939 },940 {941 "epoch": 1.1056751467710373,942 "grad_norm": 55561.12890625,943 "learning_rate": 9.999271224846395e-07,944 "loss": 0.7405,945 "step": 1130946 },947 {948 "epoch": 1.1154598825831703,949 "grad_norm": 57599.58984375,950 "learning_rate": 9.993442298227364e-07,951 "loss": 0.7397,952 "step": 1140953 },954 {955 "epoch": 1.1252446183953033,956 "grad_norm": 43188.921875,957 "learning_rate": 9.981791241257e-07,958 "loss": 0.7405,959 "step": 1150960 },961 {962 "epoch": 1.1252446183953033,963 "eval_accuracy": 0.723,964 "eval_loss": 0.6686130166053772,965 "eval_macro_f1": 0.5630742537166292,966 "eval_precision": 0.572477811481856,967 "eval_recall": 0.5595714228001223,968 "eval_runtime": 17.3251,969 "eval_samples_per_second": 57.72,970 "eval_steps_per_second": 0.058,971 "step": 1150972 },973 {974 "epoch": 1.1350293542074363,975 "grad_norm": 42596.6796875,976 "learning_rate": 9.964331638546548e-07,977 "loss": 0.7397,978 "step": 1160979 },980 {981 "epoch": 1.1448140900195694,982 "grad_norm": 60584.2890625,983 "learning_rate": 9.941083847211764e-07,984 "loss": 0.7417,985 "step": 1170986 },987 {988 "epoch": 1.1545988258317026,989 "grad_norm": 61568.48046875,990 "learning_rate": 9.912074973137411e-07,991 "loss": 0.7405,992 "step": 1180993 },994 {995 "epoch": 1.1643835616438356,996 "grad_norm": 50988.54296875,997 "learning_rate": 9.87733883937303e-07,998 "loss": 0.7381,999 "step": 11901000 },1001 {1002 "epoch": 1.1741682974559686,1003 "grad_norm": 48960.01171875,1004 "learning_rate": 9.836915946696757e-07,1005 "loss": 0.7377,1006 "step": 12001007 },1008 {1009 "epoch": 1.1741682974559686,1010 "eval_accuracy": 0.716,1011 "eval_loss": 0.6701745986938477,1012 "eval_macro_f1": 0.5807350718065004,1013 "eval_precision": 0.582848916337505,1014 "eval_recall": 0.5790466956386239,1015 "eval_runtime": 17.3208,1016 "eval_samples_per_second": 57.734,1017 "eval_steps_per_second": 0.058,1018 "step": 12001019 },1020 {1021 "epoch": 1.1839530332681019,1022 "grad_norm": 50420.14453125,1023 "learning_rate": 9.790853426393242e-07,1024 "loss": 0.7393,1025 "step": 12101026 },1027 {1028 "epoch": 1.1937377690802349,1029 "grad_norm": 39574.37890625,1030 "learning_rate": 9.73920498530068e-07,1031 "loss": 0.7388,1032 "step": 12201033 },1034 {1035 "epoch": 1.203522504892368,1036 "grad_norm": 51037.01953125,1037 "learning_rate": 9.68203084319102e-07,1038 "loss": 0.739,1039 "step": 12301040 },1041 {1042 "epoch": 1.213307240704501,1043 "grad_norm": 40847.125,1044 "learning_rate": 9.619397662556433e-07,1045 "loss": 0.7377,1046 "step": 12401047 },1048 {1049 "epoch": 1.223091976516634,1050 "grad_norm": 54462.5703125,1051 "learning_rate": 9.55137847088381e-07,1052 "loss": 0.7361,1053 "step": 12501054 },1055 {1056 "epoch": 1.223091976516634,1057 "eval_accuracy": 0.701,1058 "eval_loss": 0.671897292137146,1059 "eval_macro_f1": 0.5884212859150606,1060 "eval_precision": 0.5853845179540225,1061 "eval_recall": 0.5933739633291203,1062 "eval_runtime": 17.4826,1063 "eval_samples_per_second": 57.2,1064 "eval_steps_per_second": 0.057,1065 "step": 12501066 },1067 {1068 "epoch": 1.2328767123287672,1069 "grad_norm": 49697.23046875,1070 "learning_rate": 9.478052575507982e-07,1071 "loss": 0.737,1072 "step": 12601073 },1074 {1075 "epoch": 1.2426614481409002,1076 "grad_norm": 40874.4921875,1077 "learning_rate": 9.39950547114292e-07,1078 "loss": 0.7343,1079 "step": 12701080 },1081 {1082 "epoch": 1.2524461839530332,1083 "grad_norm": 63225.41796875,1084 "learning_rate": 9.315828740198713e-07,1085 "loss": 0.7346,1086 "step": 12801087 },1088 {1089 "epoch": 1.2622309197651664,1090 "grad_norm": 42103.6328125,1091 "learning_rate": 9.227119946000589e-07,1092 "loss": 0.733,1093 "step": 12901094 },1095 {1096 "epoch": 1.2720156555772995,1097 "grad_norm": 53617.17578125,1098 "learning_rate": 9.133482519034428e-07,1099 "loss": 0.7334,1100 "step": 13001101 },1102 {1103 "epoch": 1.2720156555772995,1104 "eval_accuracy": 0.699,1105 "eval_loss": 0.6714372038841248,1106 "eval_macro_f1": 0.586855697342288,1107 "eval_precision": 0.5837743615591398,1108 "eval_recall": 0.592086962042119,1109 "eval_runtime": 17.5108,1110 "eval_samples_per_second": 57.108,1111 "eval_steps_per_second": 0.057,1112 "step": 13001113 },1114 {1115 "epoch": 1.2818003913894325,1116 "grad_norm": 49828.37890625,1117 "learning_rate": 9.035025636351454e-07,1118 "loss": 0.7325,1119 "step": 13101120 },1121 {1122 "epoch": 1.2915851272015655,1123 "grad_norm": 52261.57421875,1124 "learning_rate": 8.931864094272663e-07,1125 "loss": 0.7311,1126 "step": 13201127 },1128 {1129 "epoch": 1.3013698630136985,1130 "grad_norm": 44072.09765625,1131 "learning_rate": 8.824118174541462e-07,1132 "loss": 0.7327,1133 "step": 13301134 },1135 {1136 "epoch": 1.3111545988258317,1137 "grad_norm": 46144.7578125,1138 "learning_rate": 8.711913504080533e-07,1139 "loss": 0.73,1140 "step": 13401141 },1142 {1143 "epoch": 1.3209393346379648,1144 "grad_norm": 40000.671875,1145 "learning_rate": 8.595380908516452e-07,1146 "loss": 0.7327,1147 "step": 13501148 },1149 {1150 "epoch": 1.3209393346379648,1151 "eval_accuracy": 0.697,1152 "eval_loss": 0.6719282865524292,1153 "eval_macro_f1": 0.5986622073578596,1154 "eval_precision": 0.5941197969142022,1155 "eval_recall": 0.6099837826295225,1156 "eval_runtime": 17.375,1157 "eval_samples_per_second": 57.554,1158 "eval_steps_per_second": 0.058,1159 "step": 13501160 }1161 ],1162 "logging_steps": 10,1163 "max_steps": 2044,1164 "num_input_tokens_seen": 0,1165 "num_train_epochs": 2,1166 "save_steps": 50,1167 "stateful_callbacks": {1168 "EarlyStoppingCallback": {1169 "args": {1170 "early_stopping_patience": 3,1171 "early_stopping_threshold": 0.01172 },1173 "attributes": {1174 "early_stopping_patience_counter": 01175 }1176 },1177 "TrainerControl": {1178 "args": {1179 "should_epoch_stop": false,1180 "should_evaluate": false,1181 "should_log": false,1182 "should_save": true,1183 "should_training_stop": false1184 },1185 "attributes": {}1186 }1187 },1188 "total_flos": 1.8180605569898496e+17,1189 "train_batch_size": 512,1190 "trial_name": null,1191 "trial_params": null1192}1193 