internalhell/whisper_small_ru_model_trainer_3ep
011
1{2 "best_global_step": 9500,3 "best_metric": 16.040464106107944,4 "best_model_checkpoint": "/content/drive/MyDrive/models/whisper_small_ru_model_trainer_3ep/checkpoint-9500",5 "epoch": 3.0,6 "eval_steps": 500,7 "global_step": 9894,8 "is_hyper_param_search": false,9 "is_local_process_zero": true,10 "is_world_process_zero": true,11 "log_history": [12 {13 "epoch": 0.007580351728320194,14 "grad_norm": 10.993572235107422,15 "learning_rate": 9.975742874469376e-06,16 "loss": 0.4792,17 "step": 2518 },19 {20 "epoch": 0.015160703456640388,21 "grad_norm": 7.92294979095459,22 "learning_rate": 9.950475035374976e-06,23 "loss": 0.27,24 "step": 5025 },26 {27 "epoch": 0.02274105518496058,28 "grad_norm": 11.214887619018555,29 "learning_rate": 9.925207196280575e-06,30 "loss": 0.3111,31 "step": 7532 },33 {34 "epoch": 0.030321406913280776,35 "grad_norm": 6.483564376831055,36 "learning_rate": 9.899939357186174e-06,37 "loss": 0.2659,38 "step": 10039 },40 {41 "epoch": 0.03790175864160097,42 "grad_norm": 13.091401100158691,43 "learning_rate": 9.874671518091774e-06,44 "loss": 0.2524,45 "step": 12546 },47 {48 "epoch": 0.04548211036992116,49 "grad_norm": 9.941374778747559,50 "learning_rate": 9.849403678997373e-06,51 "loss": 0.2601,52 "step": 15053 },54 {55 "epoch": 0.05306246209824136,56 "grad_norm": 10.436762809753418,57 "learning_rate": 9.824135839902972e-06,58 "loss": 0.2364,59 "step": 17560 },61 {62 "epoch": 0.06064281382656155,63 "grad_norm": 6.529853343963623,64 "learning_rate": 9.798868000808572e-06,65 "loss": 0.2406,66 "step": 20067 },68 {69 "epoch": 0.06822316555488174,70 "grad_norm": 6.7477641105651855,71 "learning_rate": 9.77360016171417e-06,72 "loss": 0.2628,73 "step": 22574 },75 {76 "epoch": 0.07580351728320193,77 "grad_norm": 8.057855606079102,78 "learning_rate": 9.74833232261977e-06,79 "loss": 0.2343,80 "step": 25081 },82 {83 "epoch": 0.08338386901152213,84 "grad_norm": 8.777417182922363,85 "learning_rate": 9.72306448352537e-06,86 "loss": 0.2544,87 "step": 27588 },89 {90 "epoch": 0.09096422073984232,91 "grad_norm": 9.590123176574707,92 "learning_rate": 9.697796644430968e-06,93 "loss": 0.2533,94 "step": 30095 },96 {97 "epoch": 0.09854457246816252,98 "grad_norm": 12.065073013305664,99 "learning_rate": 9.672528805336569e-06,100 "loss": 0.248,101 "step": 325102 },103 {104 "epoch": 0.10612492419648271,105 "grad_norm": 6.656917095184326,106 "learning_rate": 9.647260966242168e-06,107 "loss": 0.245,108 "step": 350109 },110 {111 "epoch": 0.11370527592480291,112 "grad_norm": 13.028727531433105,113 "learning_rate": 9.621993127147768e-06,114 "loss": 0.2245,115 "step": 375116 },117 {118 "epoch": 0.1212856276531231,119 "grad_norm": 9.074626922607422,120 "learning_rate": 9.596725288053367e-06,121 "loss": 0.2464,122 "step": 400123 },124 {125 "epoch": 0.12886597938144329,126 "grad_norm": 7.7365217208862305,127 "learning_rate": 9.571457448958967e-06,128 "loss": 0.2233,129 "step": 425130 },131 {132 "epoch": 0.13644633110976348,133 "grad_norm": 7.931459426879883,134 "learning_rate": 9.546189609864566e-06,135 "loss": 0.2522,136 "step": 450137 },138 {139 "epoch": 0.14402668283808367,140 "grad_norm": 10.054927825927734,141 "learning_rate": 9.520921770770164e-06,142 "loss": 0.2288,143 "step": 475144 },145 {146 "epoch": 0.15160703456640387,147 "grad_norm": 11.330881118774414,148 "learning_rate": 9.495653931675763e-06,149 "loss": 0.2206,150 "step": 500151 },152 {153 "epoch": 0.15160703456640387,154 "eval_cer": 5.496337965229563,155 "eval_loss": 0.26026931405067444,156 "eval_runtime": 4183.6066,157 "eval_samples_per_second": 2.439,158 "eval_ser": 69.43055963932177,159 "eval_steps_per_second": 0.61,160 "eval_wer": 21.266884240601154,161 "step": 500162 },163 {164 "epoch": 0.15918738629472406,165 "grad_norm": 22.062847137451172,166 "learning_rate": 9.470386092581364e-06,167 "loss": 0.2404,168 "step": 525169 },170 {171 "epoch": 0.16676773802304426,172 "grad_norm": 6.779280662536621,173 "learning_rate": 9.445118253486962e-06,174 "loss": 0.2299,175 "step": 550176 },177 {178 "epoch": 0.17434808975136445,179 "grad_norm": 10.592273712158203,180 "learning_rate": 9.419850414392561e-06,181 "loss": 0.2064,182 "step": 575183 },184 {185 "epoch": 0.18192844147968465,186 "grad_norm": 5.647468566894531,187 "learning_rate": 9.394582575298161e-06,188 "loss": 0.1878,189 "step": 600190 },191 {192 "epoch": 0.18950879320800484,193 "grad_norm": 12.663495063781738,194 "learning_rate": 9.36931473620376e-06,195 "loss": 0.2302,196 "step": 625197 },198 {199 "epoch": 0.19708914493632504,200 "grad_norm": 11.353273391723633,201 "learning_rate": 9.34404689710936e-06,202 "loss": 0.263,203 "step": 650204 },205 {206 "epoch": 0.20466949666464523,207 "grad_norm": 10.60062026977539,208 "learning_rate": 9.318779058014959e-06,209 "loss": 0.2276,210 "step": 675211 },212 {213 "epoch": 0.21224984839296543,214 "grad_norm": 6.700550556182861,215 "learning_rate": 9.29351121892056e-06,216 "loss": 0.2314,217 "step": 700218 },219 {220 "epoch": 0.21983020012128562,221 "grad_norm": 11.847448348999023,222 "learning_rate": 9.268243379826158e-06,223 "loss": 0.2282,224 "step": 725225 },226 {227 "epoch": 0.22741055184960582,228 "grad_norm": 10.603355407714844,229 "learning_rate": 9.242975540731759e-06,230 "loss": 0.1968,231 "step": 750232 },233 {234 "epoch": 0.234990903577926,235 "grad_norm": 10.278948783874512,236 "learning_rate": 9.217707701637356e-06,237 "loss": 0.2209,238 "step": 775239 },240 {241 "epoch": 0.2425712553062462,242 "grad_norm": 7.834460735321045,243 "learning_rate": 9.192439862542956e-06,244 "loss": 0.189,245 "step": 800246 },247 {248 "epoch": 0.2501516070345664,249 "grad_norm": 7.9238810539245605,250 "learning_rate": 9.167172023448555e-06,251 "loss": 0.1758,252 "step": 825253 },254 {255 "epoch": 0.25773195876288657,256 "grad_norm": 16.181570053100586,257 "learning_rate": 9.141904184354155e-06,258 "loss": 0.1999,259 "step": 850260 },261 {262 "epoch": 0.2653123104912068,263 "grad_norm": 5.385849475860596,264 "learning_rate": 9.116636345259754e-06,265 "loss": 0.2353,266 "step": 875267 },268 {269 "epoch": 0.27289266221952696,270 "grad_norm": 14.86568832397461,271 "learning_rate": 9.091368506165354e-06,272 "loss": 0.2039,273 "step": 900274 },275 {276 "epoch": 0.2804730139478472,277 "grad_norm": 8.38940715789795,278 "learning_rate": 9.066100667070953e-06,279 "loss": 0.2048,280 "step": 925281 },282 {283 "epoch": 0.28805336567616735,284 "grad_norm": 6.71006965637207,285 "learning_rate": 9.040832827976552e-06,286 "loss": 0.2016,287 "step": 950288 },289 {290 "epoch": 0.2956337174044876,291 "grad_norm": 9.024951934814453,292 "learning_rate": 9.015564988882152e-06,293 "loss": 0.2272,294 "step": 975295 },296 {297 "epoch": 0.30321406913280774,298 "grad_norm": 6.729151725769043,299 "learning_rate": 8.99029714978775e-06,300 "loss": 0.22,301 "step": 1000302 },303 {304 "epoch": 0.30321406913280774,305 "eval_cer": 5.382320373005102,306 "eval_loss": 0.24674250185489655,307 "eval_runtime": 4121.9591,308 "eval_samples_per_second": 2.475,309 "eval_ser": 67.35273939037538,310 "eval_steps_per_second": 0.619,311 "eval_wer": 20.297092948978527,312 "step": 1000313 },314 {315 "epoch": 0.31079442086112796,316 "grad_norm": 9.8275785446167,317 "learning_rate": 8.965029310693351e-06,318 "loss": 0.2513,319 "step": 1025320 },321 {322 "epoch": 0.31837477258944813,323 "grad_norm": 8.723328590393066,324 "learning_rate": 8.93976147159895e-06,325 "loss": 0.2095,326 "step": 1050327 },328 {329 "epoch": 0.32595512431776835,330 "grad_norm": 10.17667007446289,331 "learning_rate": 8.91449363250455e-06,332 "loss": 0.1945,333 "step": 1075334 },335 {336 "epoch": 0.3335354760460885,337 "grad_norm": 9.78806209564209,338 "learning_rate": 8.889225793410147e-06,339 "loss": 0.2357,340 "step": 1100341 },342 {343 "epoch": 0.34111582777440874,344 "grad_norm": 7.773096084594727,345 "learning_rate": 8.863957954315748e-06,346 "loss": 0.1901,347 "step": 1125348 },349 {350 "epoch": 0.3486961795027289,351 "grad_norm": 5.588106155395508,352 "learning_rate": 8.838690115221346e-06,353 "loss": 0.2156,354 "step": 1150355 },356 {357 "epoch": 0.35627653123104913,358 "grad_norm": 9.3190336227417,359 "learning_rate": 8.813422276126947e-06,360 "loss": 0.2014,361 "step": 1175362 },363 {364 "epoch": 0.3638568829593693,365 "grad_norm": 7.115675926208496,366 "learning_rate": 8.788154437032545e-06,367 "loss": 0.2073,368 "step": 1200369 },370 {371 "epoch": 0.3714372346876895,372 "grad_norm": 8.060104370117188,373 "learning_rate": 8.762886597938146e-06,374 "loss": 0.1789,375 "step": 1225376 },377 {378 "epoch": 0.3790175864160097,379 "grad_norm": 8.937747955322266,380 "learning_rate": 8.737618758843744e-06,381 "loss": 0.2115,382 "step": 1250383 },384 {385 "epoch": 0.3865979381443299,386 "grad_norm": 10.151251792907715,387 "learning_rate": 8.712350919749343e-06,388 "loss": 0.216,389 "step": 1275390 },391 {392 "epoch": 0.3941782898726501,393 "grad_norm": 6.023889064788818,394 "learning_rate": 8.687083080654943e-06,395 "loss": 0.1836,396 "step": 1300397 },398 {399 "epoch": 0.4017586416009703,400 "grad_norm": 9.286721229553223,401 "learning_rate": 8.661815241560542e-06,402 "loss": 0.1799,403 "step": 1325404 },405 {406 "epoch": 0.40933899332929047,407 "grad_norm": 10.163074493408203,408 "learning_rate": 8.636547402466143e-06,409 "loss": 0.183,410 "step": 1350411 },412 {413 "epoch": 0.4169193450576107,414 "grad_norm": 11.878681182861328,415 "learning_rate": 8.611279563371741e-06,416 "loss": 0.2121,417 "step": 1375418 },419 {420 "epoch": 0.42449969678593086,421 "grad_norm": 10.150252342224121,422 "learning_rate": 8.58601172427734e-06,423 "loss": 0.1837,424 "step": 1400425 },426 {427 "epoch": 0.4320800485142511,428 "grad_norm": 6.533999443054199,429 "learning_rate": 8.560743885182939e-06,430 "loss": 0.219,431 "step": 1425432 },433 {434 "epoch": 0.43966040024257125,435 "grad_norm": 10.426810264587402,436 "learning_rate": 8.535476046088539e-06,437 "loss": 0.2285,438 "step": 1450439 },440 {441 "epoch": 0.44724075197089147,442 "grad_norm": 10.712567329406738,443 "learning_rate": 8.510208206994138e-06,444 "loss": 0.1837,445 "step": 1475446 },447 {448 "epoch": 0.45482110369921164,449 "grad_norm": 8.469382286071777,450 "learning_rate": 8.484940367899738e-06,451 "loss": 0.1901,452 "step": 1500453 },454 {455 "epoch": 0.45482110369921164,456 "eval_cer": 5.116005507016847,457 "eval_loss": 0.23773027956485748,458 "eval_runtime": 4147.5807,459 "eval_samples_per_second": 2.46,460 "eval_ser": 66.17661472116045,461 "eval_steps_per_second": 0.615,462 "eval_wer": 19.564232703452923,463 "step": 1500464 },465 {466 "epoch": 0.46240145542753186,467 "grad_norm": 11.634538650512695,468 "learning_rate": 8.459672528805337e-06,469 "loss": 0.1774,470 "step": 1525471 },472 {473 "epoch": 0.469981807155852,474 "grad_norm": 9.295145034790039,475 "learning_rate": 8.434404689710937e-06,476 "loss": 0.2293,477 "step": 1550478 },479 {480 "epoch": 0.47756215888417225,481 "grad_norm": 9.195737838745117,482 "learning_rate": 8.409136850616536e-06,483 "loss": 0.195,484 "step": 1575485 },486 {487 "epoch": 0.4851425106124924,488 "grad_norm": 7.1239776611328125,489 "learning_rate": 8.383869011522136e-06,490 "loss": 0.1902,491 "step": 1600492 },493 {494 "epoch": 0.49272286234081264,495 "grad_norm": 8.784886360168457,496 "learning_rate": 8.358601172427735e-06,497 "loss": 0.1708,498 "step": 1625499 },500 {501 "epoch": 0.5003032140691328,502 "grad_norm": 9.43509578704834,503 "learning_rate": 8.333333333333334e-06,504 "loss": 0.1978,505 "step": 1650506 },507 {508 "epoch": 0.507883565797453,509 "grad_norm": 9.499982833862305,510 "learning_rate": 8.308065494238934e-06,511 "loss": 0.2036,512 "step": 1675513 },514 {515 "epoch": 0.5154639175257731,516 "grad_norm": 7.286120414733887,517 "learning_rate": 8.282797655144533e-06,518 "loss": 0.2003,519 "step": 1700520 },521 {522 "epoch": 0.5230442692540934,523 "grad_norm": 6.725291728973389,524 "learning_rate": 8.257529816050132e-06,525 "loss": 0.2139,526 "step": 1725527 },528 {529 "epoch": 0.5306246209824136,530 "grad_norm": 5.5903401374816895,531 "learning_rate": 8.23226197695573e-06,532 "loss": 0.1869,533 "step": 1750534 },535 {536 "epoch": 0.5382049727107338,537 "grad_norm": 7.579101085662842,538 "learning_rate": 8.20699413786133e-06,539 "loss": 0.2097,540 "step": 1775541 },542 {543 "epoch": 0.5457853244390539,544 "grad_norm": 4.244465351104736,545 "learning_rate": 8.18172629876693e-06,546 "loss": 0.1906,547 "step": 1800548 },549 {550 "epoch": 0.5533656761673742,551 "grad_norm": 5.5115580558776855,552 "learning_rate": 8.15645845967253e-06,553 "loss": 0.197,554 "step": 1825555 },556 {557 "epoch": 0.5609460278956944,558 "grad_norm": 6.544187068939209,559 "learning_rate": 8.131190620578128e-06,560 "loss": 0.2061,561 "step": 1850562 },563 {564 "epoch": 0.5685263796240145,565 "grad_norm": 6.87299108505249,566 "learning_rate": 8.105922781483729e-06,567 "loss": 0.2018,568 "step": 1875569 },570 {571 "epoch": 0.5761067313523347,572 "grad_norm": 11.84753704071045,573 "learning_rate": 8.080654942389327e-06,574 "loss": 0.2018,575 "step": 1900576 },577 {578 "epoch": 0.583687083080655,579 "grad_norm": 7.050841331481934,580 "learning_rate": 8.055387103294928e-06,581 "loss": 0.1903,582 "step": 1925583 },584 {585 "epoch": 0.5912674348089751,586 "grad_norm": 7.1467461585998535,587 "learning_rate": 8.030119264200527e-06,588 "loss": 0.2079,589 "step": 1950590 },591 {592 "epoch": 0.5988477865372953,593 "grad_norm": 7.622501373291016,594 "learning_rate": 8.004851425106125e-06,595 "loss": 0.2047,596 "step": 1975597 },598 {599 "epoch": 0.6064281382656155,600 "grad_norm": 10.488551139831543,601 "learning_rate": 7.979583586011726e-06,602 "loss": 0.1969,603 "step": 2000604 },605 {606 "epoch": 0.6064281382656155,607 "eval_cer": 5.075425758775576,608 "eval_loss": 0.22730673849582672,609 "eval_runtime": 4147.4744,610 "eval_samples_per_second": 2.46,611 "eval_ser": 64.32421836714691,612 "eval_steps_per_second": 0.615,613 "eval_wer": 19.050882103576033,614 "step": 2000615 },616 {617 "epoch": 0.6140084899939358,618 "grad_norm": 6.823671817779541,619 "learning_rate": 7.954315746917324e-06,620 "loss": 0.2002,621 "step": 2025622 },623 {624 "epoch": 0.6215888417222559,625 "grad_norm": 7.486941814422607,626 "learning_rate": 7.929047907822923e-06,627 "loss": 0.2046,628 "step": 2050629 },630 {631 "epoch": 0.6291691934505761,632 "grad_norm": 6.544958114624023,633 "learning_rate": 7.903780068728523e-06,634 "loss": 0.1736,635 "step": 2075636 },637 {638 "epoch": 0.6367495451788963,639 "grad_norm": 7.139219760894775,640 "learning_rate": 7.878512229634122e-06,641 "loss": 0.1844,642 "step": 2100643 },644 {645 "epoch": 0.6443298969072165,646 "grad_norm": 10.455595970153809,647 "learning_rate": 7.85324439053972e-06,648 "loss": 0.1916,649 "step": 2125650 },651 {652 "epoch": 0.6519102486355367,653 "grad_norm": 6.0816216468811035,654 "learning_rate": 7.827976551445321e-06,655 "loss": 0.2102,656 "step": 2150657 },658 {659 "epoch": 0.6594906003638569,660 "grad_norm": 4.258129119873047,661 "learning_rate": 7.80270871235092e-06,662 "loss": 0.1785,663 "step": 2175664 },665 {666 "epoch": 0.667070952092177,667 "grad_norm": 8.886266708374023,668 "learning_rate": 7.77744087325652e-06,669 "loss": 0.1916,670 "step": 2200671 },672 {673 "epoch": 0.6746513038204973,674 "grad_norm": 8.997633934020996,675 "learning_rate": 7.752173034162119e-06,676 "loss": 0.2088,677 "step": 2225678 },679 {680 "epoch": 0.6822316555488175,681 "grad_norm": 9.341837882995605,682 "learning_rate": 7.72690519506772e-06,683 "loss": 0.1996,684 "step": 2250685 },686 {687 "epoch": 0.6898120072771377,688 "grad_norm": 11.538143157958984,689 "learning_rate": 7.701637355973318e-06,690 "loss": 0.1766,691 "step": 2275692 },693 {694 "epoch": 0.6973923590054578,695 "grad_norm": 6.78935432434082,696 "learning_rate": 7.676369516878917e-06,697 "loss": 0.1748,698 "step": 2300699 },700 {701 "epoch": 0.7049727107337781,702 "grad_norm": 9.051713943481445,703 "learning_rate": 7.651101677784516e-06,704 "loss": 0.173,705 "step": 2325706 },707 {708 "epoch": 0.7125530624620983,709 "grad_norm": 8.62745189666748,710 "learning_rate": 7.625833838690115e-06,711 "loss": 0.1936,712 "step": 2350713 },714 {715 "epoch": 0.7201334141904184,716 "grad_norm": 9.529070854187012,717 "learning_rate": 7.600565999595715e-06,718 "loss": 0.2004,719 "step": 2375720 },721 {722 "epoch": 0.7277137659187386,723 "grad_norm": 7.745057106018066,724 "learning_rate": 7.575298160501314e-06,725 "loss": 0.1832,726 "step": 2400727 },728 {729 "epoch": 0.7352941176470589,730 "grad_norm": 8.844827651977539,731 "learning_rate": 7.550030321406914e-06,732 "loss": 0.2056,733 "step": 2425734 },735 {736 "epoch": 0.742874469375379,737 "grad_norm": 13.322829246520996,738 "learning_rate": 7.524762482312513e-06,739 "loss": 0.1825,740 "step": 2450741 },742 {743 "epoch": 0.7504548211036992,744 "grad_norm": 10.450922012329102,745 "learning_rate": 7.499494643218113e-06,746 "loss": 0.1798,747 "step": 2475748 },749 {750 "epoch": 0.7580351728320194,751 "grad_norm": 9.188721656799316,752 "learning_rate": 7.474226804123712e-06,753 "loss": 0.1743,754 "step": 2500755 },756 {757 "epoch": 0.7580351728320194,758 "eval_cer": 4.852319288687943,759 "eval_loss": 0.2188422828912735,760 "eval_runtime": 4181.1454,761 "eval_samples_per_second": 2.44,762 "eval_ser": 63.148093697931984,763 "eval_steps_per_second": 0.61,764 "eval_wer": 18.228592002415766,765 "step": 2500766 },767 {768 "epoch": 0.7656155245603395,769 "grad_norm": 5.665472030639648,770 "learning_rate": 7.448958965029312e-06,771 "loss": 0.1979,772 "step": 2525773 },774 {775 "epoch": 0.7731958762886598,776 "grad_norm": 4.752591133117676,777 "learning_rate": 7.4236911259349106e-06,778 "loss": 0.191,779 "step": 2550780 },781 {782 "epoch": 0.78077622801698,783 "grad_norm": 9.350257873535156,784 "learning_rate": 7.39842328684051e-06,785 "loss": 0.1735,786 "step": 2575787 },788 {789 "epoch": 0.7883565797453002,790 "grad_norm": 7.437028408050537,791 "learning_rate": 7.37315544774611e-06,792 "loss": 0.2365,793 "step": 2600794 },795 {796 "epoch": 0.7959369314736203,797 "grad_norm": 9.3013916015625,798 "learning_rate": 7.347887608651709e-06,799 "loss": 0.1888,800 "step": 2625801 },802 {803 "epoch": 0.8035172832019406,804 "grad_norm": 7.452167987823486,805 "learning_rate": 7.322619769557308e-06,806 "loss": 0.1752,807 "step": 2650808 },809 {810 "epoch": 0.8110976349302608,811 "grad_norm": 6.42844820022583,812 "learning_rate": 7.2973519304629074e-06,813 "loss": 0.1991,814 "step": 2675815 },816 {817 "epoch": 0.8186779866585809,818 "grad_norm": 6.838171482086182,819 "learning_rate": 7.272084091368506e-06,820 "loss": 0.1748,821 "step": 2700822 },823 {824 "epoch": 0.8262583383869011,825 "grad_norm": 7.954586505889893,826 "learning_rate": 7.246816252274106e-06,827 "loss": 0.1905,828 "step": 2725829 },830 {831 "epoch": 0.8338386901152214,832 "grad_norm": 6.925525188446045,833 "learning_rate": 7.221548413179705e-06,834 "loss": 0.1784,835 "step": 2750836 },837 {838 "epoch": 0.8414190418435415,839 "grad_norm": 7.18998384475708,840 "learning_rate": 7.196280574085305e-06,841 "loss": 0.1693,842 "step": 2775843 },844 {845 "epoch": 0.8489993935718617,846 "grad_norm": 8.666403770446777,847 "learning_rate": 7.171012734990904e-06,848 "loss": 0.1684,849 "step": 2800850 },851 {852 "epoch": 0.8565797453001819,853 "grad_norm": 7.043132781982422,854 "learning_rate": 7.145744895896504e-06,855 "loss": 0.1781,856 "step": 2825857 },858 {859 "epoch": 0.8641600970285022,860 "grad_norm": 5.09641695022583,861 "learning_rate": 7.120477056802103e-06,862 "loss": 0.1778,863 "step": 2850864 },865 {866 "epoch": 0.8717404487568223,867 "grad_norm": 6.9336748123168945,868 "learning_rate": 7.095209217707703e-06,869 "loss": 0.1607,870 "step": 2875871 },872 {873 "epoch": 0.8793208004851425,874 "grad_norm": 5.951363563537598,875 "learning_rate": 7.069941378613302e-06,876 "loss": 0.1793,877 "step": 2900878 },879 {880 "epoch": 0.8869011522134627,881 "grad_norm": 5.363964557647705,882 "learning_rate": 7.044673539518901e-06,883 "loss": 0.1742,884 "step": 2925885 },886 {887 "epoch": 0.8944815039417829,888 "grad_norm": 8.336238861083984,889 "learning_rate": 7.0194057004245e-06,890 "loss": 0.2014,891 "step": 2950892 },893 {894 "epoch": 0.9020618556701031,895 "grad_norm": 6.5310587882995605,896 "learning_rate": 6.9941378613300994e-06,897 "loss": 0.2074,898 "step": 2975899 },900 {901 "epoch": 0.9096422073984233,902 "grad_norm": 9.294402122497559,903 "learning_rate": 6.968870022235699e-06,904 "loss": 0.1747,905 "step": 3000906 },907 {908 "epoch": 0.9096422073984233,909 "eval_cer": 4.88665599873825,910 "eval_loss": 0.21674878895282745,911 "eval_runtime": 4171.9101,912 "eval_samples_per_second": 2.446,913 "eval_ser": 62.40321474076252,914 "eval_steps_per_second": 0.611,915 "eval_wer": 18.098512212401715,916 "step": 3000917 },918 {919 "epoch": 0.9172225591267434,920 "grad_norm": 8.7633056640625,921 "learning_rate": 6.943602183141298e-06,922 "loss": 0.1575,923 "step": 3025924 },925 {926 "epoch": 0.9248029108550637,927 "grad_norm": 13.642629623413086,928 "learning_rate": 6.918334344046897e-06,929 "loss": 0.1817,930 "step": 3050931 },932 {933 "epoch": 0.9323832625833839,934 "grad_norm": 6.795938968658447,935 "learning_rate": 6.893066504952497e-06,936 "loss": 0.1887,937 "step": 3075938 },939 {940 "epoch": 0.939963614311704,941 "grad_norm": 7.764118671417236,942 "learning_rate": 6.867798665858096e-06,943 "loss": 0.1611,944 "step": 3100945 },946 {947 "epoch": 0.9475439660400242,948 "grad_norm": 7.532794952392578,949 "learning_rate": 6.842530826763696e-06,950 "loss": 0.1869,951 "step": 3125952 },953 {954 "epoch": 0.9551243177683445,955 "grad_norm": 6.741429328918457,956 "learning_rate": 6.817262987669295e-06,957 "loss": 0.1846,958 "step": 3150959 },960 {961 "epoch": 0.9627046694966647,962 "grad_norm": 9.013187408447266,963 "learning_rate": 6.791995148574895e-06,964 "loss": 0.1649,965 "step": 3175966 },967 {968 "epoch": 0.9702850212249848,969 "grad_norm": 11.595109939575195,970 "learning_rate": 6.7667273094804945e-06,971 "loss": 0.1818,972 "step": 3200973 },974 {975 "epoch": 0.977865372953305,976 "grad_norm": 12.244895935058594,977 "learning_rate": 6.741459470386094e-06,978 "loss": 0.1906,979 "step": 3225980 },981 {982 "epoch": 0.9854457246816253,983 "grad_norm": 6.988622188568115,984 "learning_rate": 6.716191631291692e-06,985 "loss": 0.173,986 "step": 3250987 },988 {989 "epoch": 0.9930260764099454,990 "grad_norm": 4.436202526092529,991 "learning_rate": 6.690923792197291e-06,992 "loss": 0.1838,993 "step": 3275994 },995 {996 "epoch": 1.0006064281382656,997 "grad_norm": 6.243381977081299,998 "learning_rate": 6.665655953102891e-06,999 "loss": 0.1718,1000 "step": 33001001 },1002 {1003 "epoch": 1.0081867798665858,1004 "grad_norm": 5.074841022491455,1005 "learning_rate": 6.6403881140084905e-06,1006 "loss": 0.076,1007 "step": 33251008 },1009 {1010 "epoch": 1.015767131594906,1011 "grad_norm": 4.297674655914307,1012 "learning_rate": 6.61512027491409e-06,1013 "loss": 0.078,1014 "step": 33501015 },1016 {1017 "epoch": 1.0233474833232261,1018 "grad_norm": 3.33673095703125,1019 "learning_rate": 6.589852435819689e-06,1020 "loss": 0.0774,1021 "step": 33751022 },1023 {1024 "epoch": 1.0309278350515463,1025 "grad_norm": 7.982945442199707,1026 "learning_rate": 6.564584596725288e-06,1027 "loss": 0.0856,1028 "step": 34001029 },1030 {1031 "epoch": 1.0385081867798667,1032 "grad_norm": 5.4341888427734375,1033 "learning_rate": 6.539316757630888e-06,1034 "loss": 0.0752,1035 "step": 34251036 },1037 {1038 "epoch": 1.0460885385081868,1039 "grad_norm": 3.646911144256592,1040 "learning_rate": 6.514048918536487e-06,1041 "loss": 0.0657,1042 "step": 34501043 },1044 {1045 "epoch": 1.053668890236507,1046 "grad_norm": 2.50636625289917,1047 "learning_rate": 6.488781079442087e-06,1048 "loss": 0.0678,1049 "step": 34751050 },1051 {1052 "epoch": 1.0612492419648272,1053 "grad_norm": 2.662381172180176,1054 "learning_rate": 6.4635132403476865e-06,1055 "loss": 0.077,1056 "step": 35001057 },1058 {1059 "epoch": 1.0612492419648272,1060 "eval_cer": 4.527188431321651,1061 "eval_loss": 0.2141607105731964,1062 "eval_runtime": 4130.8959,1063 "eval_samples_per_second": 2.47,1064 "eval_ser": 60.59982358129962,1065 "eval_steps_per_second": 0.618,1066 "eval_wer": 17.200729375965437,1067 "step": 35001068 },1069 {1070 "epoch": 1.0688295936931473,1071 "grad_norm": 2.4962751865386963,1072 "learning_rate": 6.438245401253286e-06,1073 "loss": 0.0885,1074 "step": 35251075 },1076 {1077 "epoch": 1.0764099454214675,1078 "grad_norm": 5.805286884307861,1079 "learning_rate": 6.4129775621588856e-06,1080 "loss": 0.0644,1081 "step": 35501082 },1083 {1084 "epoch": 1.0839902971497877,1085 "grad_norm": 3.763657331466675,1086 "learning_rate": 6.387709723064483e-06,1087 "loss": 0.0647,1088 "step": 35751089 },1090 {1091 "epoch": 1.0915706488781078,1092 "grad_norm": 3.8962442874908447,1093 "learning_rate": 6.362441883970083e-06,1094 "loss": 0.0759,1095 "step": 36001096 },1097 {1098 "epoch": 1.0991510006064282,1099 "grad_norm": 6.139822483062744,1100 "learning_rate": 6.3371740448756825e-06,1101 "loss": 0.0707,1102 "step": 36251103 },1104 {1105 "epoch": 1.1067313523347484,1106 "grad_norm": 4.11650276184082,1107 "learning_rate": 6.311906205781282e-06,1108 "loss": 0.0699,1109 "step": 36501110 },1111 {1112 "epoch": 1.1143117040630686,1113 "grad_norm": 6.533496379852295,1114 "learning_rate": 6.286638366686882e-06,1115 "loss": 0.069,1116 "step": 36751117 },1118 {1119 "epoch": 1.1218920557913887,1120 "grad_norm": 3.1129257678985596,1121 "learning_rate": 6.261370527592481e-06,1122 "loss": 0.0707,1123 "step": 37001124 },1125 {1126 "epoch": 1.129472407519709,1127 "grad_norm": 3.159837245941162,1128 "learning_rate": 6.23610268849808e-06,1129 "loss": 0.0766,1130 "step": 37251131 },1132 {1133 "epoch": 1.137052759248029,1134 "grad_norm": 6.998505592346191,1135 "learning_rate": 6.210834849403679e-06,1136 "loss": 0.0799,1137 "step": 37501138 },1139 {1140 "epoch": 1.1446331109763492,1141 "grad_norm": 5.051515102386475,1142 "learning_rate": 6.185567010309279e-06,1143 "loss": 0.08,1144 "step": 37751145 },1146 {1147 "epoch": 1.1522134627046694,1148 "grad_norm": 4.849501132965088,1149 "learning_rate": 6.1602991712148785e-06,1150 "loss": 0.0694,1151 "step": 38001152 },1153 {1154 "epoch": 1.1597938144329896,1155 "grad_norm": 4.301568984985352,1156 "learning_rate": 6.135031332120478e-06,1157 "loss": 0.0801,1158 "step": 38251159 },1160 {1161 "epoch": 1.16737416616131,1162 "grad_norm": 7.137290954589844,1163 "learning_rate": 6.1097634930260776e-06,1164 "loss": 0.0707,1165 "step": 38501166 },1167 {1168 "epoch": 1.1749545178896301,1169 "grad_norm": 1.7661166191101074,1170 "learning_rate": 6.084495653931675e-06,1171 "loss": 0.0671,1172 "step": 38751173 },1174 {1175 "epoch": 1.1825348696179503,1176 "grad_norm": 3.448512077331543,1177 "learning_rate": 6.059227814837275e-06,1178 "loss": 0.0689,1179 "step": 39001180 },1181 {1182 "epoch": 1.1901152213462705,1183 "grad_norm": 3.7001595497131348,1184 "learning_rate": 6.0339599757428745e-06,1185 "loss": 0.0764,1186 "step": 39251187 },1188 {1189 "epoch": 1.1976955730745906,1190 "grad_norm": 6.902029037475586,1191 "learning_rate": 6.008692136648474e-06,1192 "loss": 0.0661,1193 "step": 39501194 },1195 {1196 "epoch": 1.2052759248029108,1197 "grad_norm": 4.167758464813232,1198 "learning_rate": 5.983424297554074e-06,1199 "loss": 0.071,1200 "step": 3975