arthoho66/model_test_002
08
1{2 "best_metric": 95.5,3 "best_model_checkpoint": "./medium_TH/checkpoint-3000",4 "epoch": 88.23529411764706,5 "global_step": 3000,6 "is_hyper_param_search": false,7 "is_local_process_zero": true,8 "is_world_process_zero": true,9 "log_history": [10 {11 "epoch": 0.74,12 "learning_rate": 4.4e-07,13 "loss": 0.8659,14 "step": 2515 },16 {17 "epoch": 1.47,18 "learning_rate": 9.400000000000001e-07,19 "loss": 0.7368,20 "step": 5021 },22 {23 "epoch": 2.21,24 "learning_rate": 1.44e-06,25 "loss": 0.5972,26 "step": 7527 },28 {29 "epoch": 2.94,30 "learning_rate": 1.94e-06,31 "loss": 0.5492,32 "step": 10033 },34 {35 "epoch": 3.68,36 "learning_rate": 2.4400000000000004e-06,37 "loss": 0.4563,38 "step": 12539 },40 {41 "epoch": 4.41,42 "learning_rate": 2.9400000000000002e-06,43 "loss": 0.4365,44 "step": 15045 },46 {47 "epoch": 5.15,48 "learning_rate": 3.44e-06,49 "loss": 0.5196,50 "step": 17551 },52 {53 "epoch": 5.88,54 "learning_rate": 3.94e-06,55 "loss": 0.3682,56 "step": 20057 },58 {59 "epoch": 6.62,60 "learning_rate": 4.440000000000001e-06,61 "loss": 0.2693,62 "step": 22563 },64 {65 "epoch": 7.35,66 "learning_rate": 4.94e-06,67 "loss": 0.2433,68 "step": 25069 },70 {71 "epoch": 8.09,72 "learning_rate": 5.4400000000000004e-06,73 "loss": 0.2578,74 "step": 27575 },76 {77 "epoch": 8.82,78 "learning_rate": 5.94e-06,79 "loss": 0.173,80 "step": 30081 },82 {83 "epoch": 9.56,84 "learning_rate": 6.440000000000001e-06,85 "loss": 0.1173,86 "step": 32587 },88 {89 "epoch": 10.29,90 "learning_rate": 6.9400000000000005e-06,91 "loss": 0.1491,92 "step": 35093 },94 {95 "epoch": 11.03,96 "learning_rate": 7.440000000000001e-06,97 "loss": 0.1102,98 "step": 37599 },100 {101 "epoch": 11.76,102 "learning_rate": 7.94e-06,103 "loss": 0.1325,104 "step": 400105 },106 {107 "epoch": 12.5,108 "learning_rate": 8.44e-06,109 "loss": 0.1195,110 "step": 425111 },112 {113 "epoch": 13.24,114 "learning_rate": 8.94e-06,115 "loss": 0.0834,116 "step": 450117 },118 {119 "epoch": 13.97,120 "learning_rate": 9.440000000000001e-06,121 "loss": 0.1089,122 "step": 475123 },124 {125 "epoch": 14.71,126 "learning_rate": 9.940000000000001e-06,127 "loss": 0.0594,128 "step": 500129 },130 {131 "epoch": 15.44,132 "learning_rate": 9.937142857142858e-06,133 "loss": 0.0973,134 "step": 525135 },136 {137 "epoch": 16.18,138 "learning_rate": 9.865714285714285e-06,139 "loss": 0.0344,140 "step": 550141 },142 {143 "epoch": 16.91,144 "learning_rate": 9.794285714285714e-06,145 "loss": 0.0628,146 "step": 575147 },148 {149 "epoch": 17.65,150 "learning_rate": 9.722857142857143e-06,151 "loss": 0.0611,152 "step": 600153 },154 {155 "epoch": 18.38,156 "learning_rate": 9.651428571428572e-06,157 "loss": 0.054,158 "step": 625159 },160 {161 "epoch": 19.12,162 "learning_rate": 9.58e-06,163 "loss": 0.069,164 "step": 650165 },166 {167 "epoch": 19.85,168 "learning_rate": 9.508571428571429e-06,169 "loss": 0.0491,170 "step": 675171 },172 {173 "epoch": 20.59,174 "learning_rate": 9.437142857142858e-06,175 "loss": 0.0958,176 "step": 700177 },178 {179 "epoch": 21.32,180 "learning_rate": 9.365714285714287e-06,181 "loss": 0.0477,182 "step": 725183 },184 {185 "epoch": 22.06,186 "learning_rate": 9.294285714285714e-06,187 "loss": 0.0318,188 "step": 750189 },190 {191 "epoch": 22.79,192 "learning_rate": 9.222857142857143e-06,193 "loss": 0.0691,194 "step": 775195 },196 {197 "epoch": 23.53,198 "learning_rate": 9.151428571428572e-06,199 "loss": 0.0317,200 "step": 800201 },202 {203 "epoch": 24.26,204 "learning_rate": 9.080000000000001e-06,205 "loss": 0.0221,206 "step": 825207 },208 {209 "epoch": 25.0,210 "learning_rate": 9.00857142857143e-06,211 "loss": 0.039,212 "step": 850213 },214 {215 "epoch": 25.74,216 "learning_rate": 8.937142857142857e-06,217 "loss": 0.0344,218 "step": 875219 },220 {221 "epoch": 26.47,222 "learning_rate": 8.865714285714287e-06,223 "loss": 0.0168,224 "step": 900225 },226 {227 "epoch": 27.21,228 "learning_rate": 8.794285714285716e-06,229 "loss": 0.0346,230 "step": 925231 },232 {233 "epoch": 27.94,234 "learning_rate": 8.722857142857145e-06,235 "loss": 0.0437,236 "step": 950237 },238 {239 "epoch": 28.68,240 "learning_rate": 8.651428571428572e-06,241 "loss": 0.0197,242 "step": 975243 },244 {245 "epoch": 29.41,246 "learning_rate": 8.580000000000001e-06,247 "loss": 0.0303,248 "step": 1000249 },250 {251 "epoch": 29.41,252 "eval_loss": 0.3286643922328949,253 "eval_runtime": 10.3915,254 "eval_samples_per_second": 2.213,255 "eval_steps_per_second": 0.289,256 "eval_wer": 106.0,257 "step": 1000258 },259 {260 "epoch": 30.15,261 "learning_rate": 8.50857142857143e-06,262 "loss": 0.0517,263 "step": 1025264 },265 {266 "epoch": 30.88,267 "learning_rate": 8.437142857142859e-06,268 "loss": 0.0381,269 "step": 1050270 },271 {272 "epoch": 31.62,273 "learning_rate": 8.365714285714286e-06,274 "loss": 0.0309,275 "step": 1075276 },277 {278 "epoch": 32.35,279 "learning_rate": 8.294285714285715e-06,280 "loss": 0.0078,281 "step": 1100282 },283 {284 "epoch": 33.09,285 "learning_rate": 8.222857142857144e-06,286 "loss": 0.0297,287 "step": 1125288 },289 {290 "epoch": 33.82,291 "learning_rate": 8.151428571428572e-06,292 "loss": 0.0247,293 "step": 1150294 },295 {296 "epoch": 34.56,297 "learning_rate": 8.08e-06,298 "loss": 0.0098,299 "step": 1175300 },301 {302 "epoch": 35.29,303 "learning_rate": 8.00857142857143e-06,304 "loss": 0.0083,305 "step": 1200306 },307 {308 "epoch": 36.03,309 "learning_rate": 7.937142857142857e-06,310 "loss": 0.0296,311 "step": 1225312 },313 {314 "epoch": 36.76,315 "learning_rate": 7.865714285714286e-06,316 "loss": 0.021,317 "step": 1250318 },319 {320 "epoch": 37.5,321 "learning_rate": 7.794285714285715e-06,322 "loss": 0.0403,323 "step": 1275324 },325 {326 "epoch": 38.24,327 "learning_rate": 7.722857142857142e-06,328 "loss": 0.0225,329 "step": 1300330 },331 {332 "epoch": 38.97,333 "learning_rate": 7.651428571428571e-06,334 "loss": 0.0295,335 "step": 1325336 },337 {338 "epoch": 39.71,339 "learning_rate": 7.58e-06,340 "loss": 0.0144,341 "step": 1350342 },343 {344 "epoch": 40.44,345 "learning_rate": 7.508571428571429e-06,346 "loss": 0.0271,347 "step": 1375348 },349 {350 "epoch": 41.18,351 "learning_rate": 7.4371428571428575e-06,352 "loss": 0.0225,353 "step": 1400354 },355 {356 "epoch": 41.91,357 "learning_rate": 7.365714285714286e-06,358 "loss": 0.0242,359 "step": 1425360 },361 {362 "epoch": 42.65,363 "learning_rate": 7.294285714285715e-06,364 "loss": 0.0138,365 "step": 1450366 },367 {368 "epoch": 43.38,369 "learning_rate": 7.222857142857144e-06,370 "loss": 0.0106,371 "step": 1475372 },373 {374 "epoch": 44.12,375 "learning_rate": 7.151428571428573e-06,376 "loss": 0.0221,377 "step": 1500378 },379 {380 "epoch": 44.85,381 "learning_rate": 7.08e-06,382 "loss": 0.0304,383 "step": 1525384 },385 {386 "epoch": 45.59,387 "learning_rate": 7.008571428571429e-06,388 "loss": 0.0224,389 "step": 1550390 },391 {392 "epoch": 46.32,393 "learning_rate": 6.937142857142858e-06,394 "loss": 0.0208,395 "step": 1575396 },397 {398 "epoch": 47.06,399 "learning_rate": 6.865714285714287e-06,400 "loss": 0.0057,401 "step": 1600402 },403 {404 "epoch": 47.79,405 "learning_rate": 6.794285714285714e-06,406 "loss": 0.0181,407 "step": 1625408 },409 {410 "epoch": 48.53,411 "learning_rate": 6.722857142857143e-06,412 "loss": 0.0118,413 "step": 1650414 },415 {416 "epoch": 49.26,417 "learning_rate": 6.651428571428572e-06,418 "loss": 0.0158,419 "step": 1675420 },421 {422 "epoch": 50.0,423 "learning_rate": 6.5800000000000005e-06,424 "loss": 0.015,425 "step": 1700426 },427 {428 "epoch": 50.74,429 "learning_rate": 6.5085714285714295e-06,430 "loss": 0.0095,431 "step": 1725432 },433 {434 "epoch": 51.47,435 "learning_rate": 6.437142857142858e-06,436 "loss": 0.015,437 "step": 1750438 },439 {440 "epoch": 52.21,441 "learning_rate": 6.365714285714286e-06,442 "loss": 0.0099,443 "step": 1775444 },445 {446 "epoch": 52.94,447 "learning_rate": 6.294285714285715e-06,448 "loss": 0.0111,449 "step": 1800450 },451 {452 "epoch": 53.68,453 "learning_rate": 6.222857142857144e-06,454 "loss": 0.0165,455 "step": 1825456 },457 {458 "epoch": 54.41,459 "learning_rate": 6.151428571428571e-06,460 "loss": 0.0097,461 "step": 1850462 },463 {464 "epoch": 55.15,465 "learning_rate": 6.08e-06,466 "loss": 0.0042,467 "step": 1875468 },469 {470 "epoch": 55.88,471 "learning_rate": 6.008571428571429e-06,472 "loss": 0.0222,473 "step": 1900474 },475 {476 "epoch": 56.62,477 "learning_rate": 5.937142857142858e-06,478 "loss": 0.009,479 "step": 1925480 },481 {482 "epoch": 57.35,483 "learning_rate": 5.865714285714286e-06,484 "loss": 0.0052,485 "step": 1950486 },487 {488 "epoch": 58.09,489 "learning_rate": 5.794285714285715e-06,490 "loss": 0.0126,491 "step": 1975492 },493 {494 "epoch": 58.82,495 "learning_rate": 5.722857142857144e-06,496 "loss": 0.0142,497 "step": 2000498 },499 {500 "epoch": 58.82,501 "eval_loss": 0.3790155053138733,502 "eval_runtime": 10.2927,503 "eval_samples_per_second": 2.235,504 "eval_steps_per_second": 0.291,505 "eval_wer": 99.5,506 "step": 2000507 },508 {509 "epoch": 59.56,510 "learning_rate": 5.651428571428572e-06,511 "loss": 0.0117,512 "step": 2025513 },514 {515 "epoch": 60.29,516 "learning_rate": 5.580000000000001e-06,517 "loss": 0.0079,518 "step": 2050519 },520 {521 "epoch": 61.03,522 "learning_rate": 5.508571428571429e-06,523 "loss": 0.0044,524 "step": 2075525 },526 {527 "epoch": 61.76,528 "learning_rate": 5.437142857142857e-06,529 "loss": 0.007,530 "step": 2100531 },532 {533 "epoch": 62.5,534 "learning_rate": 5.365714285714286e-06,535 "loss": 0.021,536 "step": 2125537 },538 {539 "epoch": 63.24,540 "learning_rate": 5.294285714285715e-06,541 "loss": 0.0102,542 "step": 2150543 },544 {545 "epoch": 63.97,546 "learning_rate": 5.2228571428571425e-06,547 "loss": 0.0035,548 "step": 2175549 },550 {551 "epoch": 64.71,552 "learning_rate": 5.1514285714285715e-06,553 "loss": 0.0072,554 "step": 2200555 },556 {557 "epoch": 65.44,558 "learning_rate": 5.0800000000000005e-06,559 "loss": 0.0032,560 "step": 2225561 },562 {563 "epoch": 66.18,564 "learning_rate": 5.0085714285714295e-06,565 "loss": 0.0047,566 "step": 2250567 },568 {569 "epoch": 66.91,570 "learning_rate": 4.937142857142858e-06,571 "loss": 0.003,572 "step": 2275573 },574 {575 "epoch": 67.65,576 "learning_rate": 4.865714285714287e-06,577 "loss": 0.0114,578 "step": 2300579 },580 {581 "epoch": 68.38,582 "learning_rate": 4.794285714285715e-06,583 "loss": 0.0073,584 "step": 2325585 },586 {587 "epoch": 69.12,588 "learning_rate": 4.722857142857144e-06,589 "loss": 0.0119,590 "step": 2350591 },592 {593 "epoch": 69.85,594 "learning_rate": 4.651428571428572e-06,595 "loss": 0.0045,596 "step": 2375597 },598 {599 "epoch": 70.59,600 "learning_rate": 4.58e-06,601 "loss": 0.0147,602 "step": 2400603 },604 {605 "epoch": 71.32,606 "learning_rate": 4.508571428571429e-06,607 "loss": 0.0081,608 "step": 2425609 },610 {611 "epoch": 72.06,612 "learning_rate": 4.437142857142857e-06,613 "loss": 0.0159,614 "step": 2450615 },616 {617 "epoch": 72.79,618 "learning_rate": 4.3657142857142855e-06,619 "loss": 0.0032,620 "step": 2475621 },622 {623 "epoch": 73.53,624 "learning_rate": 4.2942857142857146e-06,625 "loss": 0.0081,626 "step": 2500627 },628 {629 "epoch": 74.26,630 "learning_rate": 4.222857142857143e-06,631 "loss": 0.0116,632 "step": 2525633 },634 {635 "epoch": 75.0,636 "learning_rate": 4.151428571428572e-06,637 "loss": 0.0012,638 "step": 2550639 },640 {641 "epoch": 75.74,642 "learning_rate": 4.08e-06,643 "loss": 0.0018,644 "step": 2575645 },646 {647 "epoch": 76.47,648 "learning_rate": 4.008571428571429e-06,649 "loss": 0.0059,650 "step": 2600651 },652 {653 "epoch": 77.21,654 "learning_rate": 3.937142857142858e-06,655 "loss": 0.0032,656 "step": 2625657 },658 {659 "epoch": 77.94,660 "learning_rate": 3.865714285714286e-06,661 "loss": 0.0089,662 "step": 2650663 },664 {665 "epoch": 78.68,666 "learning_rate": 3.7942857142857147e-06,667 "loss": 0.0044,668 "step": 2675669 },670 {671 "epoch": 79.41,672 "learning_rate": 3.722857142857143e-06,673 "loss": 0.0049,674 "step": 2700675 },676 {677 "epoch": 80.15,678 "learning_rate": 3.651428571428572e-06,679 "loss": 0.0038,680 "step": 2725681 },682 {683 "epoch": 80.88,684 "learning_rate": 3.58e-06,685 "loss": 0.0053,686 "step": 2750687 },688 {689 "epoch": 81.62,690 "learning_rate": 3.508571428571429e-06,691 "loss": 0.004,692 "step": 2775693 },694 {695 "epoch": 82.35,696 "learning_rate": 3.437142857142857e-06,697 "loss": 0.0495,698 "step": 2800699 },700 {701 "epoch": 83.09,702 "learning_rate": 3.3657142857142862e-06,703 "loss": 0.0035,704 "step": 2825705 },706 {707 "epoch": 83.82,708 "learning_rate": 3.2942857142857144e-06,709 "loss": 0.0044,710 "step": 2850711 },712 {713 "epoch": 84.56,714 "learning_rate": 3.222857142857143e-06,715 "loss": 0.005,716 "step": 2875717 },718 {719 "epoch": 85.29,720 "learning_rate": 3.151428571428572e-06,721 "loss": 0.0102,722 "step": 2900723 },724 {725 "epoch": 86.03,726 "learning_rate": 3.08e-06,727 "loss": 0.0054,728 "step": 2925729 },730 {731 "epoch": 86.76,732 "learning_rate": 3.008571428571429e-06,733 "loss": 0.0095,734 "step": 2950735 },736 {737 "epoch": 87.5,738 "learning_rate": 2.9371428571428573e-06,739 "loss": 0.0065,740 "step": 2975741 },742 {743 "epoch": 88.24,744 "learning_rate": 2.865714285714286e-06,745 "loss": 0.0032,746 "step": 3000747 },748 {749 "epoch": 88.24,750 "eval_loss": 0.3970164954662323,751 "eval_runtime": 10.2993,752 "eval_samples_per_second": 2.233,753 "eval_steps_per_second": 0.291,754 "eval_wer": 95.5,755 "step": 3000756 }757 ],758 "max_steps": 4000,759 "num_train_epochs": 118,760 "total_flos": 1.801163540791296e+19,761 "trial_name": null,762 "trial_params": null763}764 