laion/glm46-code-feedback-maxeps-131k
0267
1{2 "best_global_step": null,3 "best_metric": null,4 "best_model_checkpoint": null,5 "epoch": 7.0,6 "eval_steps": 500,7 "global_step": 4305,8 "is_hyper_param_search": false,9 "is_local_process_zero": true,10 "is_world_process_zero": true,11 "log_history": [12 {13 "epoch": 0.008130081300813009,14 "grad_norm": 10.729355259341085,15 "learning_rate": 3.7122969837587006e-07,16 "loss": 0.5735,17 "loss_nan_ranks": 0,18 "loss_rank_avg": 0.24004313349723816,19 "step": 5,20 "valid_targets_mean": 2614.5,21 "valid_targets_min": 98822 },23 {24 "epoch": 0.016260162601626018,25 "grad_norm": 9.95614411374737,26 "learning_rate": 8.352668213457077e-07,27 "loss": 0.5844,28 "loss_nan_ranks": 0,29 "loss_rank_avg": 0.3054102659225464,30 "step": 10,31 "valid_targets_mean": 6700.2,32 "valid_targets_min": 102233 },34 {35 "epoch": 0.024390243902439025,36 "grad_norm": 8.490830318504097,37 "learning_rate": 1.2993039443155453e-06,38 "loss": 0.5273,39 "loss_nan_ranks": 0,40 "loss_rank_avg": 0.27064454555511475,41 "step": 15,42 "valid_targets_mean": 4731.8,43 "valid_targets_min": 117144 },45 {46 "epoch": 0.032520325203252036,47 "grad_norm": 5.508858253527907,48 "learning_rate": 1.7633410672853829e-06,49 "loss": 0.5604,50 "loss_nan_ranks": 0,51 "loss_rank_avg": 0.28161105513572693,52 "step": 20,53 "valid_targets_mean": 5939.0,54 "valid_targets_min": 120855 },56 {57 "epoch": 0.04065040650406504,58 "grad_norm": 4.244414584795395,59 "learning_rate": 2.2273781902552207e-06,60 "loss": 0.525,61 "loss_nan_ranks": 0,62 "loss_rank_avg": 0.19481831789016724,63 "step": 25,64 "valid_targets_mean": 2888.2,65 "valid_targets_min": 70866 },67 {68 "epoch": 0.04878048780487805,69 "grad_norm": 2.9008493875224715,70 "learning_rate": 2.691415313225058e-06,71 "loss": 0.4719,72 "loss_nan_ranks": 0,73 "loss_rank_avg": 0.24280954897403717,74 "step": 30,75 "valid_targets_mean": 7361.0,76 "valid_targets_min": 150777 },78 {79 "epoch": 0.056910569105691054,80 "grad_norm": 3.4096382562645484,81 "learning_rate": 3.155452436194896e-06,82 "loss": 0.49,83 "loss_nan_ranks": 0,84 "loss_rank_avg": 0.24371470510959625,85 "step": 35,86 "valid_targets_mean": 3680.0,87 "valid_targets_min": 159988 },89 {90 "epoch": 0.06504065040650407,91 "grad_norm": 2.056871731875376,92 "learning_rate": 3.6194895591647333e-06,93 "loss": 0.446,94 "loss_nan_ranks": 0,95 "loss_rank_avg": 0.14936479926109314,96 "step": 40,97 "valid_targets_mean": 3807.8,98 "valid_targets_min": 109899 },100 {101 "epoch": 0.07317073170731707,102 "grad_norm": 0.9901834399287566,103 "learning_rate": 4.083526682134571e-06,104 "loss": 0.4223,105 "loss_nan_ranks": 0,106 "loss_rank_avg": 0.14713236689567566,107 "step": 45,108 "valid_targets_mean": 4927.4,109 "valid_targets_min": 1369110 },111 {112 "epoch": 0.08130081300813008,113 "grad_norm": 0.9398360387681741,114 "learning_rate": 4.547563805104409e-06,115 "loss": 0.4316,116 "loss_nan_ranks": 0,117 "loss_rank_avg": 0.16155597567558289,118 "step": 50,119 "valid_targets_mean": 4512.9,120 "valid_targets_min": 1585121 },122 {123 "epoch": 0.08943089430894309,124 "grad_norm": 0.8775743692611262,125 "learning_rate": 5.011600928074246e-06,126 "loss": 0.3961,127 "loss_nan_ranks": 0,128 "loss_rank_avg": 0.2303980439901352,129 "step": 55,130 "valid_targets_mean": 4560.0,131 "valid_targets_min": 985132 },133 {134 "epoch": 0.0975609756097561,135 "grad_norm": 0.795569922663752,136 "learning_rate": 5.4756380510440845e-06,137 "loss": 0.3787,138 "loss_nan_ranks": 0,139 "loss_rank_avg": 0.16651669144630432,140 "step": 60,141 "valid_targets_mean": 3114.6,142 "valid_targets_min": 1042143 },144 {145 "epoch": 0.10569105691056911,146 "grad_norm": 0.7519872281053458,147 "learning_rate": 5.939675174013921e-06,148 "loss": 0.3569,149 "loss_nan_ranks": 0,150 "loss_rank_avg": 0.17810176312923431,151 "step": 65,152 "valid_targets_mean": 3038.5,153 "valid_targets_min": 1407154 },155 {156 "epoch": 0.11382113821138211,157 "grad_norm": 0.5792804598494421,158 "learning_rate": 6.403712296983759e-06,159 "loss": 0.3246,160 "loss_nan_ranks": 0,161 "loss_rank_avg": 0.1996963620185852,162 "step": 70,163 "valid_targets_mean": 5235.0,164 "valid_targets_min": 1850165 },166 {167 "epoch": 0.12195121951219512,168 "grad_norm": 0.4824690597797212,169 "learning_rate": 6.867749419953597e-06,170 "loss": 0.3565,171 "loss_nan_ranks": 0,172 "loss_rank_avg": 0.18132978677749634,173 "step": 75,174 "valid_targets_mean": 6870.2,175 "valid_targets_min": 1801176 },177 {178 "epoch": 0.13008130081300814,179 "grad_norm": 0.5501428825070286,180 "learning_rate": 7.331786542923435e-06,181 "loss": 0.3333,182 "loss_nan_ranks": 0,183 "loss_rank_avg": 0.163741797208786,184 "step": 80,185 "valid_targets_mean": 4903.1,186 "valid_targets_min": 1105187 },188 {189 "epoch": 0.13821138211382114,190 "grad_norm": 0.515640618796055,191 "learning_rate": 7.795823665893271e-06,192 "loss": 0.3231,193 "loss_nan_ranks": 0,194 "loss_rank_avg": 0.1189749538898468,195 "step": 85,196 "valid_targets_mean": 4170.2,197 "valid_targets_min": 1259198 },199 {200 "epoch": 0.14634146341463414,201 "grad_norm": 0.6231036421396678,202 "learning_rate": 8.25986078886311e-06,203 "loss": 0.3143,204 "loss_nan_ranks": 0,205 "loss_rank_avg": 0.10459889471530914,206 "step": 90,207 "valid_targets_mean": 2787.4,208 "valid_targets_min": 820209 },210 {211 "epoch": 0.15447154471544716,212 "grad_norm": 0.47776022024330533,213 "learning_rate": 8.723897911832948e-06,214 "loss": 0.3215,215 "loss_nan_ranks": 0,216 "loss_rank_avg": 0.1699979603290558,217 "step": 95,218 "valid_targets_mean": 5565.5,219 "valid_targets_min": 691220 },221 {222 "epoch": 0.16260162601626016,223 "grad_norm": 0.43049029114582227,224 "learning_rate": 9.187935034802784e-06,225 "loss": 0.3178,226 "loss_nan_ranks": 0,227 "loss_rank_avg": 0.16746115684509277,228 "step": 100,229 "valid_targets_mean": 8008.9,230 "valid_targets_min": 3875231 },232 {233 "epoch": 0.17073170731707318,234 "grad_norm": 0.5506518790709487,235 "learning_rate": 9.651972157772623e-06,236 "loss": 0.3086,237 "loss_nan_ranks": 0,238 "loss_rank_avg": 0.17378145456314087,239 "step": 105,240 "valid_targets_mean": 5575.4,241 "valid_targets_min": 1059242 },243 {244 "epoch": 0.17886178861788618,245 "grad_norm": 0.8343481888737161,246 "learning_rate": 1.011600928074246e-05,247 "loss": 0.3039,248 "loss_nan_ranks": 0,249 "loss_rank_avg": 0.13841012120246887,250 "step": 110,251 "valid_targets_mean": 6387.9,252 "valid_targets_min": 1380253 },254 {255 "epoch": 0.18699186991869918,256 "grad_norm": 0.4678403515176022,257 "learning_rate": 1.0580046403712299e-05,258 "loss": 0.3303,259 "loss_nan_ranks": 0,260 "loss_rank_avg": 0.2064119130373001,261 "step": 115,262 "valid_targets_mean": 7768.4,263 "valid_targets_min": 1619264 },265 {266 "epoch": 0.1951219512195122,267 "grad_norm": 0.36446029955073816,268 "learning_rate": 1.1044083526682134e-05,269 "loss": 0.3097,270 "loss_nan_ranks": 0,271 "loss_rank_avg": 0.1493927538394928,272 "step": 120,273 "valid_targets_mean": 8210.1,274 "valid_targets_min": 2419275 },276 {277 "epoch": 0.2032520325203252,278 "grad_norm": 0.5532632575478795,279 "learning_rate": 1.1508120649651972e-05,280 "loss": 0.2989,281 "loss_nan_ranks": 0,282 "loss_rank_avg": 0.1957596093416214,283 "step": 125,284 "valid_targets_mean": 5523.8,285 "valid_targets_min": 1219286 },287 {288 "epoch": 0.21138211382113822,289 "grad_norm": 0.4918145090486274,290 "learning_rate": 1.197215777262181e-05,291 "loss": 0.2847,292 "loss_nan_ranks": 0,293 "loss_rank_avg": 0.1981673538684845,294 "step": 130,295 "valid_targets_mean": 8387.9,296 "valid_targets_min": 1912297 },298 {299 "epoch": 0.21951219512195122,300 "grad_norm": 0.4824040162688919,301 "learning_rate": 1.2436194895591649e-05,302 "loss": 0.3072,303 "loss_nan_ranks": 0,304 "loss_rank_avg": 0.12638956308364868,305 "step": 135,306 "valid_targets_mean": 5431.0,307 "valid_targets_min": 1759308 },309 {310 "epoch": 0.22764227642276422,311 "grad_norm": 0.5178853741428794,312 "learning_rate": 1.2900232018561485e-05,313 "loss": 0.3109,314 "loss_nan_ranks": 0,315 "loss_rank_avg": 0.18286965787410736,316 "step": 140,317 "valid_targets_mean": 4913.8,318 "valid_targets_min": 1029319 },320 {321 "epoch": 0.23577235772357724,322 "grad_norm": 0.5486837832887713,323 "learning_rate": 1.3364269141531323e-05,324 "loss": 0.3227,325 "loss_nan_ranks": 0,326 "loss_rank_avg": 0.26483672857284546,327 "step": 145,328 "valid_targets_mean": 7427.6,329 "valid_targets_min": 1096330 },331 {332 "epoch": 0.24390243902439024,333 "grad_norm": 0.47855051057921927,334 "learning_rate": 1.3828306264501162e-05,335 "loss": 0.2856,336 "loss_nan_ranks": 0,337 "loss_rank_avg": 0.1614745557308197,338 "step": 150,339 "valid_targets_mean": 5602.1,340 "valid_targets_min": 961341 },342 {343 "epoch": 0.25203252032520324,344 "grad_norm": 0.5353775633039806,345 "learning_rate": 1.4292343387471e-05,346 "loss": 0.296,347 "loss_nan_ranks": 0,348 "loss_rank_avg": 0.16262441873550415,349 "step": 155,350 "valid_targets_mean": 4503.0,351 "valid_targets_min": 733352 },353 {354 "epoch": 0.2601626016260163,355 "grad_norm": 0.46885526862806487,356 "learning_rate": 1.4756380510440838e-05,357 "loss": 0.2983,358 "loss_nan_ranks": 0,359 "loss_rank_avg": 0.12933751940727234,360 "step": 160,361 "valid_targets_mean": 5221.4,362 "valid_targets_min": 456363 },364 {365 "epoch": 0.2682926829268293,366 "grad_norm": 0.5510357143973084,367 "learning_rate": 1.5220417633410673e-05,368 "loss": 0.2769,369 "loss_nan_ranks": 0,370 "loss_rank_avg": 0.14275045692920685,371 "step": 165,372 "valid_targets_mean": 3452.2,373 "valid_targets_min": 175374 },375 {376 "epoch": 0.2764227642276423,377 "grad_norm": 0.5986519641807708,378 "learning_rate": 1.5684454756380513e-05,379 "loss": 0.2867,380 "loss_nan_ranks": 0,381 "loss_rank_avg": 0.2011912316083908,382 "step": 170,383 "valid_targets_mean": 5387.8,384 "valid_targets_min": 1825385 },386 {387 "epoch": 0.2845528455284553,388 "grad_norm": 0.7983475971758657,389 "learning_rate": 1.614849187935035e-05,390 "loss": 0.2953,391 "loss_nan_ranks": 0,392 "loss_rank_avg": 0.1250675469636917,393 "step": 175,394 "valid_targets_mean": 2584.2,395 "valid_targets_min": 1308396 },397 {398 "epoch": 0.2926829268292683,399 "grad_norm": 0.5300705164920657,400 "learning_rate": 1.661252900232019e-05,401 "loss": 0.3008,402 "loss_nan_ranks": 0,403 "loss_rank_avg": 0.14062389731407166,404 "step": 180,405 "valid_targets_mean": 4876.6,406 "valid_targets_min": 2482407 },408 {409 "epoch": 0.3008130081300813,410 "grad_norm": 0.4645501589874264,411 "learning_rate": 1.7076566125290022e-05,412 "loss": 0.2959,413 "loss_nan_ranks": 0,414 "loss_rank_avg": 0.10153844952583313,415 "step": 185,416 "valid_targets_mean": 4490.9,417 "valid_targets_min": 938418 },419 {420 "epoch": 0.3089430894308943,421 "grad_norm": 0.5394172146224312,422 "learning_rate": 1.7540603248259862e-05,423 "loss": 0.2887,424 "loss_nan_ranks": 0,425 "loss_rank_avg": 0.1731753945350647,426 "step": 190,427 "valid_targets_mean": 6405.1,428 "valid_targets_min": 1564429 },430 {431 "epoch": 0.3170731707317073,432 "grad_norm": 0.6081161338172066,433 "learning_rate": 1.80046403712297e-05,434 "loss": 0.291,435 "loss_nan_ranks": 0,436 "loss_rank_avg": 0.09163825213909149,437 "step": 195,438 "valid_targets_mean": 2425.0,439 "valid_targets_min": 882440 },441 {442 "epoch": 0.3252032520325203,443 "grad_norm": 0.39700152309765374,444 "learning_rate": 1.846867749419954e-05,445 "loss": 0.3043,446 "loss_nan_ranks": 0,447 "loss_rank_avg": 0.1639854609966278,448 "step": 200,449 "valid_targets_mean": 9026.9,450 "valid_targets_min": 1543451 },452 {453 "epoch": 0.3333333333333333,454 "grad_norm": 1.1686077596478834,455 "learning_rate": 1.8932714617169375e-05,456 "loss": 0.2797,457 "loss_nan_ranks": 0,458 "loss_rank_avg": 0.12743505835533142,459 "step": 205,460 "valid_targets_mean": 4637.0,461 "valid_targets_min": 1690462 },463 {464 "epoch": 0.34146341463414637,465 "grad_norm": 0.5152584843628413,466 "learning_rate": 1.9396751740139212e-05,467 "loss": 0.3006,468 "loss_nan_ranks": 0,469 "loss_rank_avg": 0.11795099079608917,470 "step": 210,471 "valid_targets_mean": 3623.4,472 "valid_targets_min": 1597473 },474 {475 "epoch": 0.34959349593495936,476 "grad_norm": 0.40931960381507093,477 "learning_rate": 1.986078886310905e-05,478 "loss": 0.2783,479 "loss_nan_ranks": 0,480 "loss_rank_avg": 0.19876506924629211,481 "step": 215,482 "valid_targets_mean": 10234.2,483 "valid_targets_min": 795484 },485 {486 "epoch": 0.35772357723577236,487 "grad_norm": 0.5681020072245295,488 "learning_rate": 2.0324825986078888e-05,489 "loss": 0.2613,490 "loss_nan_ranks": 0,491 "loss_rank_avg": 0.138270765542984,492 "step": 220,493 "valid_targets_mean": 4483.9,494 "valid_targets_min": 922495 },496 {497 "epoch": 0.36585365853658536,498 "grad_norm": 0.6620929272244965,499 "learning_rate": 2.0788863109048725e-05,500 "loss": 0.3119,501 "loss_nan_ranks": 0,502 "loss_rank_avg": 0.26536881923675537,503 "step": 225,504 "valid_targets_mean": 4992.5,505 "valid_targets_min": 1766506 },507 {508 "epoch": 0.37398373983739835,509 "grad_norm": 0.473185582494688,510 "learning_rate": 2.125290023201856e-05,511 "loss": 0.2741,512 "loss_nan_ranks": 0,513 "loss_rank_avg": 0.17839691042900085,514 "step": 230,515 "valid_targets_mean": 8155.8,516 "valid_targets_min": 1768517 },518 {519 "epoch": 0.3821138211382114,520 "grad_norm": 0.5676414017591311,521 "learning_rate": 2.17169373549884e-05,522 "loss": 0.2812,523 "loss_nan_ranks": 0,524 "loss_rank_avg": 0.15132203698158264,525 "step": 235,526 "valid_targets_mean": 3835.4,527 "valid_targets_min": 1300528 },529 {530 "epoch": 0.3902439024390244,531 "grad_norm": 0.5829173056048123,532 "learning_rate": 2.2180974477958238e-05,533 "loss": 0.2926,534 "loss_nan_ranks": 0,535 "loss_rank_avg": 0.10389190167188644,536 "step": 240,537 "valid_targets_mean": 3737.4,538 "valid_targets_min": 1126539 },540 {541 "epoch": 0.3983739837398374,542 "grad_norm": 0.4496112265820951,543 "learning_rate": 2.2645011600928078e-05,544 "loss": 0.2652,545 "loss_nan_ranks": 0,546 "loss_rank_avg": 0.1356901228427887,547 "step": 245,548 "valid_targets_mean": 6019.0,549 "valid_targets_min": 1824550 },551 {552 "epoch": 0.4065040650406504,553 "grad_norm": 0.6433613059763399,554 "learning_rate": 2.3109048723897914e-05,555 "loss": 0.2815,556 "loss_nan_ranks": 0,557 "loss_rank_avg": 0.13550019264221191,558 "step": 250,559 "valid_targets_mean": 3801.2,560 "valid_targets_min": 1174561 },562 {563 "epoch": 0.4146341463414634,564 "grad_norm": 0.6396160632853143,565 "learning_rate": 2.357308584686775e-05,566 "loss": 0.2784,567 "loss_nan_ranks": 0,568 "loss_rank_avg": 0.1697913110256195,569 "step": 255,570 "valid_targets_mean": 3961.4,571 "valid_targets_min": 1502572 },573 {574 "epoch": 0.42276422764227645,575 "grad_norm": 0.4542094599236192,576 "learning_rate": 2.4037122969837587e-05,577 "loss": 0.2683,578 "loss_nan_ranks": 0,579 "loss_rank_avg": 0.09232847392559052,580 "step": 260,581 "valid_targets_mean": 4648.8,582 "valid_targets_min": 1023583 },584 {585 "epoch": 0.43089430894308944,586 "grad_norm": 0.5412227898339419,587 "learning_rate": 2.4501160092807427e-05,588 "loss": 0.2728,589 "loss_nan_ranks": 0,590 "loss_rank_avg": 0.1287793219089508,591 "step": 265,592 "valid_targets_mean": 4237.9,593 "valid_targets_min": 1086594 },595 {596 "epoch": 0.43902439024390244,597 "grad_norm": 0.4177869578429799,598 "learning_rate": 2.4965197215777264e-05,599 "loss": 0.253,600 "loss_nan_ranks": 0,601 "loss_rank_avg": 0.1671661138534546,602 "step": 270,603 "valid_targets_mean": 9169.6,604 "valid_targets_min": 4230605 },606 {607 "epoch": 0.44715447154471544,608 "grad_norm": 0.47683007534315536,609 "learning_rate": 2.54292343387471e-05,610 "loss": 0.2598,611 "loss_nan_ranks": 0,612 "loss_rank_avg": 0.11259545385837555,613 "step": 275,614 "valid_targets_mean": 4597.0,615 "valid_targets_min": 1167616 },617 {618 "epoch": 0.45528455284552843,619 "grad_norm": 0.528485160665934,620 "learning_rate": 2.589327146171694e-05,621 "loss": 0.2672,622 "loss_nan_ranks": 0,623 "loss_rank_avg": 0.11031106859445572,624 "step": 280,625 "valid_targets_mean": 5527.1,626 "valid_targets_min": 1031627 },628 {629 "epoch": 0.4634146341463415,630 "grad_norm": 0.6449048798012383,631 "learning_rate": 2.6357308584686777e-05,632 "loss": 0.2784,633 "loss_nan_ranks": 0,634 "loss_rank_avg": 0.16236139833927155,635 "step": 285,636 "valid_targets_mean": 3449.2,637 "valid_targets_min": 900638 },639 {640 "epoch": 0.4715447154471545,641 "grad_norm": 0.47651550237539747,642 "learning_rate": 2.6821345707656617e-05,643 "loss": 0.2671,644 "loss_nan_ranks": 0,645 "loss_rank_avg": 0.16859616339206696,646 "step": 290,647 "valid_targets_mean": 6841.1,648 "valid_targets_min": 1434649 },650 {651 "epoch": 0.4796747967479675,652 "grad_norm": 0.5060381133309334,653 "learning_rate": 2.7285382830626453e-05,654 "loss": 0.2698,655 "loss_nan_ranks": 0,656 "loss_rank_avg": 0.15308891236782074,657 "step": 295,658 "valid_targets_mean": 6108.2,659 "valid_targets_min": 1979660 },661 {662 "epoch": 0.4878048780487805,663 "grad_norm": 0.46570844149760965,664 "learning_rate": 2.774941995359629e-05,665 "loss": 0.2634,666 "loss_nan_ranks": 0,667 "loss_rank_avg": 0.07817001640796661,668 "step": 300,669 "valid_targets_mean": 2998.5,670 "valid_targets_min": 1532671 },672 {673 "epoch": 0.4959349593495935,674 "grad_norm": 0.5748363877093409,675 "learning_rate": 2.8213457076566126e-05,676 "loss": 0.2872,677 "loss_nan_ranks": 0,678 "loss_rank_avg": 0.2178957462310791,679 "step": 305,680 "valid_targets_mean": 4954.2,681 "valid_targets_min": 1530682 },683 {684 "epoch": 0.5040650406504065,685 "grad_norm": 0.4913109666435666,686 "learning_rate": 2.8677494199535966e-05,687 "loss": 0.2593,688 "loss_nan_ranks": 0,689 "loss_rank_avg": 0.08731204271316528,690 "step": 310,691 "valid_targets_mean": 3538.8,692 "valid_targets_min": 814693 },694 {695 "epoch": 0.5121951219512195,696 "grad_norm": 0.4731336417357327,697 "learning_rate": 2.9141531322505803e-05,698 "loss": 0.2499,699 "loss_nan_ranks": 0,700 "loss_rank_avg": 0.10972920060157776,701 "step": 315,702 "valid_targets_mean": 4589.5,703 "valid_targets_min": 716704 },705 {706 "epoch": 0.5203252032520326,707 "grad_norm": 0.544587609060752,708 "learning_rate": 2.9605568445475643e-05,709 "loss": 0.2789,710 "loss_nan_ranks": 0,711 "loss_rank_avg": 0.10782943665981293,712 "step": 320,713 "valid_targets_mean": 3388.5,714 "valid_targets_min": 967715 },716 {717 "epoch": 0.5284552845528455,718 "grad_norm": 0.5206054830318291,719 "learning_rate": 3.006960556844548e-05,720 "loss": 0.2557,721 "loss_nan_ranks": 0,722 "loss_rank_avg": 0.11411543190479279,723 "step": 325,724 "valid_targets_mean": 3479.6,725 "valid_targets_min": 754726 },727 {728 "epoch": 0.5365853658536586,729 "grad_norm": 0.5109963804533619,730 "learning_rate": 3.053364269141532e-05,731 "loss": 0.2803,732 "loss_nan_ranks": 0,733 "loss_rank_avg": 0.11475864052772522,734 "step": 330,735 "valid_targets_mean": 3898.2,736 "valid_targets_min": 639737 },738 {739 "epoch": 0.5447154471544715,740 "grad_norm": 0.49736930558210946,741 "learning_rate": 3.099767981438515e-05,742 "loss": 0.2668,743 "loss_nan_ranks": 0,744 "loss_rank_avg": 0.09260010719299316,745 "step": 335,746 "valid_targets_mean": 3543.0,747 "valid_targets_min": 1186748 },749 {750 "epoch": 0.5528455284552846,751 "grad_norm": 1.4224554880291065,752 "learning_rate": 3.146171693735499e-05,753 "loss": 0.2782,754 "loss_nan_ranks": 0,755 "loss_rank_avg": 0.1166592612862587,756 "step": 340,757 "valid_targets_mean": 6083.1,758 "valid_targets_min": 783759 },760 {761 "epoch": 0.5609756097560976,762 "grad_norm": 0.6507048259502546,763 "learning_rate": 3.1925754060324825e-05,764 "loss": 0.2783,765 "loss_nan_ranks": 0,766 "loss_rank_avg": 0.11932586133480072,767 "step": 345,768 "valid_targets_mean": 2966.4,769 "valid_targets_min": 1495770 },771 {772 "epoch": 0.5691056910569106,773 "grad_norm": 0.5262338877240812,774 "learning_rate": 3.2389791183294665e-05,775 "loss": 0.2748,776 "loss_nan_ranks": 0,777 "loss_rank_avg": 0.18811023235321045,778 "step": 350,779 "valid_targets_mean": 6847.5,780 "valid_targets_min": 1820781 },782 {783 "epoch": 0.5772357723577236,784 "grad_norm": 0.7190553945803088,785 "learning_rate": 3.2853828306264505e-05,786 "loss": 0.2788,787 "loss_nan_ranks": 0,788 "loss_rank_avg": 0.1663609743118286,789 "step": 355,790 "valid_targets_mean": 4037.1,791 "valid_targets_min": 1123792 },793 {794 "epoch": 0.5853658536585366,795 "grad_norm": 0.5338816539006683,796 "learning_rate": 3.3317865429234345e-05,797 "loss": 0.283,798 "loss_nan_ranks": 0,799 "loss_rank_avg": 0.13983416557312012,800 "step": 360,801 "valid_targets_mean": 5041.6,802 "valid_targets_min": 1378803 },804 {805 "epoch": 0.5934959349593496,806 "grad_norm": 0.5021054760120496,807 "learning_rate": 3.378190255220418e-05,808 "loss": 0.2559,809 "loss_nan_ranks": 0,810 "loss_rank_avg": 0.10287587344646454,811 "step": 365,812 "valid_targets_mean": 5181.4,813 "valid_targets_min": 1034814 },815 {816 "epoch": 0.6016260162601627,817 "grad_norm": 0.6030850444597946,818 "learning_rate": 3.424593967517402e-05,819 "loss": 0.2606,820 "loss_nan_ranks": 0,821 "loss_rank_avg": 0.16273638606071472,822 "step": 370,823 "valid_targets_mean": 3852.2,824 "valid_targets_min": 1120825 },826 {827 "epoch": 0.6097560975609756,828 "grad_norm": 0.4200234244730862,829 "learning_rate": 3.470997679814386e-05,830 "loss": 0.2764,831 "loss_nan_ranks": 0,832 "loss_rank_avg": 0.13301149010658264,833 "step": 375,834 "valid_targets_mean": 5999.4,835 "valid_targets_min": 1017836 },837 {838 "epoch": 0.6178861788617886,839 "grad_norm": 0.47437564358950707,840 "learning_rate": 3.517401392111369e-05,841 "loss": 0.2672,842 "loss_nan_ranks": 0,843 "loss_rank_avg": 0.12184874713420868,844 "step": 380,845 "valid_targets_mean": 6132.9,846 "valid_targets_min": 1141847 },848 {849 "epoch": 0.6260162601626016,850 "grad_norm": 0.5720586708854113,851 "learning_rate": 3.563805104408353e-05,852 "loss": 0.2575,853 "loss_nan_ranks": 0,854 "loss_rank_avg": 0.08667431026697159,855 "step": 385,856 "valid_targets_mean": 2480.5,857 "valid_targets_min": 845858 },859 {860 "epoch": 0.6341463414634146,861 "grad_norm": 0.45303244769740036,862 "learning_rate": 3.6102088167053364e-05,863 "loss": 0.2662,864 "loss_nan_ranks": 0,865 "loss_rank_avg": 0.16857236623764038,866 "step": 390,867 "valid_targets_mean": 7660.0,868 "valid_targets_min": 711869 },870 {871 "epoch": 0.6422764227642277,872 "grad_norm": 0.45737182440597945,873 "learning_rate": 3.6566125290023204e-05,874 "loss": 0.2475,875 "loss_nan_ranks": 0,876 "loss_rank_avg": 0.10698920488357544,877 "step": 395,878 "valid_targets_mean": 4812.4,879 "valid_targets_min": 1836880 },881 {882 "epoch": 0.6504065040650406,883 "grad_norm": 0.5668819902013378,884 "learning_rate": 3.7030162412993044e-05,885 "loss": 0.2697,886 "loss_nan_ranks": 0,887 "loss_rank_avg": 0.11894096434116364,888 "step": 400,889 "valid_targets_mean": 3405.1,890 "valid_targets_min": 1117891 },892 {893 "epoch": 0.6585365853658537,894 "grad_norm": 0.4972517400541581,895 "learning_rate": 3.7494199535962884e-05,896 "loss": 0.2606,897 "loss_nan_ranks": 0,898 "loss_rank_avg": 0.14403322339057922,899 "step": 405,900 "valid_targets_mean": 5518.8,901 "valid_targets_min": 1220902 },903 {904 "epoch": 0.6666666666666666,905 "grad_norm": 0.4726743985096405,906 "learning_rate": 3.795823665893272e-05,907 "loss": 0.2464,908 "loss_nan_ranks": 0,909 "loss_rank_avg": 0.1037069782614708,910 "step": 410,911 "valid_targets_mean": 4652.2,912 "valid_targets_min": 971913 },914 {915 "epoch": 0.6747967479674797,916 "grad_norm": 0.5555072614400881,917 "learning_rate": 3.842227378190256e-05,918 "loss": 0.2778,919 "loss_nan_ranks": 0,920 "loss_rank_avg": 0.23938646912574768,921 "step": 415,922 "valid_targets_mean": 6567.6,923 "valid_targets_min": 981924 },925 {926 "epoch": 0.6829268292682927,927 "grad_norm": 0.48406443535669796,928 "learning_rate": 3.888631090487239e-05,929 "loss": 0.2528,930 "loss_nan_ranks": 0,931 "loss_rank_avg": 0.0925922617316246,932 "step": 420,933 "valid_targets_mean": 4397.2,934 "valid_targets_min": 360935 },936 {937 "epoch": 0.6910569105691057,938 "grad_norm": 0.5051976310215145,939 "learning_rate": 3.935034802784223e-05,940 "loss": 0.246,941 "loss_nan_ranks": 0,942 "loss_rank_avg": 0.09235464036464691,943 "step": 425,944 "valid_targets_mean": 4802.4,945 "valid_targets_min": 722946 },947 {948 "epoch": 0.6991869918699187,949 "grad_norm": 0.5776580965911587,950 "learning_rate": 3.981438515081207e-05,951 "loss": 0.2613,952 "loss_nan_ranks": 0,953 "loss_rank_avg": 0.09387768059968948,954 "step": 430,955 "valid_targets_mean": 2986.2,956 "valid_targets_min": 1579957 },958 {959 "epoch": 0.7073170731707317,960 "grad_norm": 0.4136218864471342,961 "learning_rate": 3.9999940813479674e-05,962 "loss": 0.2463,963 "loss_nan_ranks": 0,964 "loss_rank_avg": 0.07906226068735123,965 "step": 435,966 "valid_targets_mean": 3723.9,967 "valid_targets_min": 444968 },969 {970 "epoch": 0.7154471544715447,971 "grad_norm": 0.6531331123865727,972 "learning_rate": 3.999957911934624e-05,973 "loss": 0.2553,974 "loss_nan_ranks": 0,975 "loss_rank_avg": 0.15597087144851685,976 "step": 440,977 "valid_targets_mean": 3835.9,978 "valid_targets_min": 1892979 },980 {981 "epoch": 0.7235772357723578,982 "grad_norm": 0.5556456726719008,983 "learning_rate": 3.9998888618418865e-05,984 "loss": 0.2514,985 "loss_nan_ranks": 0,986 "loss_rank_avg": 0.1279674619436264,987 "step": 445,988 "valid_targets_mean": 4801.5,989 "valid_targets_min": 1082990 },991 {992 "epoch": 0.7317073170731707,993 "grad_norm": 0.435881099780943,994 "learning_rate": 3.999786932204985e-05,995 "loss": 0.2624,996 "loss_nan_ranks": 0,997 "loss_rank_avg": 0.09787143021821976,998 "step": 450,999 "valid_targets_mean": 5274.4,1000 "valid_targets_min": 15371001 },1002 {1003 "epoch": 0.7398373983739838,1004 "grad_norm": 0.5828524476640174,1005 "learning_rate": 3.999652124699712e-05,1006 "loss": 0.2558,1007 "loss_nan_ranks": 0,1008 "loss_rank_avg": 0.10186605900526047,1009 "step": 455,1010 "valid_targets_mean": 2984.5,1011 "valid_targets_min": 7021012 },1013 {1014 "epoch": 0.7479674796747967,1015 "grad_norm": 0.5608325699413138,1016 "learning_rate": 3.999484441542395e-05,1017 "loss": 0.2586,1018 "loss_nan_ranks": 0,1019 "loss_rank_avg": 0.1100446879863739,1020 "step": 460,1021 "valid_targets_mean": 4323.4,1022 "valid_targets_min": 15021023 },1024 {1025 "epoch": 0.7560975609756098,1026 "grad_norm": 0.4896854400312324,1027 "learning_rate": 3.999283885489861e-05,1028 "loss": 0.2519,1029 "loss_nan_ranks": 0,1030 "loss_rank_avg": 0.14972570538520813,1031 "step": 465,1032 "valid_targets_mean": 5723.0,1033 "valid_targets_min": 9091034 },1035 {1036 "epoch": 0.7642276422764228,1037 "grad_norm": 0.4715745180677696,1038 "learning_rate": 3.999050459839389e-05,1039 "loss": 0.2672,1040 "loss_nan_ranks": 0,1041 "loss_rank_avg": 0.16658423840999603,1042 "step": 470,1043 "valid_targets_mean": 5921.8,1044 "valid_targets_min": 16051045 },1046 {1047 "epoch": 0.7723577235772358,1048 "grad_norm": 0.4332253434648535,1049 "learning_rate": 3.998784168428657e-05,1050 "loss": 0.2557,1051 "loss_nan_ranks": 0,1052 "loss_rank_avg": 0.16223764419555664,1053 "step": 475,1054 "valid_targets_mean": 8892.9,1055 "valid_targets_min": 9791056 },1057 {1058 "epoch": 0.7804878048780488,1059 "grad_norm": 0.49416734264895223,1060 "learning_rate": 3.998485015635677e-05,1061 "loss": 0.2413,1062 "loss_nan_ranks": 0,1063 "loss_rank_avg": 0.08383375406265259,1064 "step": 480,1065 "valid_targets_mean": 3830.4,1066 "valid_targets_min": 11981067 },1068 {1069 "epoch": 0.7886178861788617,1070 "grad_norm": 0.44223369864110523,1071 "learning_rate": 3.998153006378727e-05,1072 "loss": 0.2387,1073 "loss_nan_ranks": 0,1074 "loss_rank_avg": 0.13400453329086304,1075 "step": 485,1076 "valid_targets_mean": 7611.5,1077 "valid_targets_min": 10801078 },1079 {1080 "epoch": 0.7967479674796748,1081 "grad_norm": 0.4314334619419397,1082 "learning_rate": 3.997788146116267e-05,1083 "loss": 0.2532,1084 "loss_nan_ranks": 0,1085 "loss_rank_avg": 0.17776936292648315,1086 "step": 490,1087 "valid_targets_mean": 8224.4,1088 "valid_targets_min": 34291089 },1090 {1091 "epoch": 0.8048780487804879,1092 "grad_norm": 0.5567932646873074,1093 "learning_rate": 3.99739044084685e-05,1094 "loss": 0.2734,1095 "loss_nan_ranks": 0,1096 "loss_rank_avg": 0.11946548521518707,1097 "step": 495,1098 "valid_targets_mean": 3261.8,1099 "valid_targets_min": 16131100 },1101 {1102 "epoch": 0.8130081300813008,1103 "grad_norm": 0.448735762829862,1104 "learning_rate": 3.9969598971090225e-05,1105 "loss": 0.2565,1106 "loss_nan_ranks": 0,1107 "loss_rank_avg": 0.11881952732801437,1108 "step": 500,1109 "valid_targets_mean": 5543.5,1110 "valid_targets_min": 8551111 },1112 {1113 "epoch": 0.8211382113821138,1114 "grad_norm": 0.4402840571732114,1115 "learning_rate": 3.99649652198122e-05,1116 "loss": 0.2517,1117 "loss_nan_ranks": 0,1118 "loss_rank_avg": 0.06367374956607819,1119 "step": 505,1120 "valid_targets_mean": 2775.2,1121 "valid_targets_min": 9471122 },1123 {1124 "epoch": 0.8292682926829268,1125 "grad_norm": 0.5282258088763285,1126 "learning_rate": 3.9960003230816456e-05,1127 "loss": 0.265,1128 "loss_nan_ranks": 0,1129 "loss_rank_avg": 0.09785593301057816,1130 "step": 510,1131 "valid_targets_mean": 3915.1,1132 "valid_targets_min": 14161133 },1134 {1135 "epoch": 0.8373983739837398,1136 "grad_norm": 0.47773763553327236,1137 "learning_rate": 3.9954713085681504e-05,1138 "loss": 0.246,1139 "loss_nan_ranks": 0,1140 "loss_rank_avg": 0.14291530847549438,1141 "step": 515,1142 "valid_targets_mean": 5950.1,1143 "valid_targets_min": 15351144 },1145 {1146 "epoch": 0.8455284552845529,1147 "grad_norm": 0.43337914617536744,1148 "learning_rate": 3.994909487138096e-05,1149 "loss": 0.2535,1150 "loss_nan_ranks": 0,1151 "loss_rank_avg": 0.12172655761241913,1152 "step": 520,1153 "valid_targets_mean": 5970.8,1154 "valid_targets_min": 7831155 },1156 {1157 "epoch": 0.8536585365853658,1158 "grad_norm": 0.5378102457615822,1159 "learning_rate": 3.994314868028212e-05,1160 "loss": 0.2593,1161 "loss_nan_ranks": 0,1162 "loss_rank_avg": 0.11482933908700943,1163 "step": 525,1164 "valid_targets_mean": 3605.8,1165 "valid_targets_min": 12161166 },1167 {1168 "epoch": 0.8617886178861789,1169 "grad_norm": 0.5254260728399146,1170 "learning_rate": 3.9936874610144445e-05,1171 "loss": 0.2507,1172 "loss_nan_ranks": 0,1173 "loss_rank_avg": 0.0977395549416542,1174 "step": 530,1175 "valid_targets_mean": 3598.8,1176 "valid_targets_min": 10981177 },1178 {1179 "epoch": 0.8699186991869918,1180 "grad_norm": 0.7023944839667103,1181 "learning_rate": 3.993027276411793e-05,1182 "loss": 0.2769,1183 "loss_nan_ranks": 0,1184 "loss_rank_avg": 0.15339651703834534,1185 "step": 535,1186 "valid_targets_mean": 5156.0,1187 "valid_targets_min": 18231188 },1189 {1190 "epoch": 0.8780487804878049,1191 "grad_norm": 0.5171186618541325,1192 "learning_rate": 3.992334325074148e-05,1193 "loss": 0.265,1194 "loss_nan_ranks": 0,1195 "loss_rank_avg": 0.17998674511909485,1196 "step": 540,1197 "valid_targets_mean": 6728.4,1198 "valid_targets_min": 12721199 },1200 {