DCAgent/a1-stack_bash
0154
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": 4662,8 "is_hyper_param_search": false,9 "is_local_process_zero": true,10 "is_world_process_zero": true,11 "log_history": [12 {13 "epoch": 0.0075075075075075074,14 "grad_norm": 11.018023139973177,15 "learning_rate": 3.426124197002142e-07,16 "loss": 0.6962,17 "loss_nan_ranks": 0,18 "loss_rank_avg": 0.666359007358551,19 "step": 5,20 "valid_targets_mean": 3406.2,21 "valid_targets_min": 59022 },23 {24 "epoch": 0.015015015015015015,25 "grad_norm": 14.300498234778246,26 "learning_rate": 7.708779443254819e-07,27 "loss": 0.6812,28 "loss_nan_ranks": 0,29 "loss_rank_avg": 0.7260710597038269,30 "step": 10,31 "valid_targets_mean": 2929.3,32 "valid_targets_min": 54033 },34 {35 "epoch": 0.02252252252252252,36 "grad_norm": 10.308334513580103,37 "learning_rate": 1.1991434689507496e-06,38 "loss": 0.694,39 "loss_nan_ranks": 0,40 "loss_rank_avg": 0.7622815370559692,41 "step": 15,42 "valid_targets_mean": 4942.4,43 "valid_targets_min": 47444 },45 {46 "epoch": 0.03003003003003003,47 "grad_norm": 9.480467405179247,48 "learning_rate": 1.6274089935760173e-06,49 "loss": 0.63,50 "loss_nan_ranks": 0,51 "loss_rank_avg": 0.5915040969848633,52 "step": 20,53 "valid_targets_mean": 4988.7,54 "valid_targets_min": 50655 },56 {57 "epoch": 0.03753753753753754,58 "grad_norm": 6.98322327787682,59 "learning_rate": 2.055674518201285e-06,60 "loss": 0.634,61 "loss_nan_ranks": 0,62 "loss_rank_avg": 0.6035254001617432,63 "step": 25,64 "valid_targets_mean": 4144.8,65 "valid_targets_min": 40666 },67 {68 "epoch": 0.04504504504504504,69 "grad_norm": 4.519545202299929,70 "learning_rate": 2.4839400428265524e-06,71 "loss": 0.6108,72 "loss_nan_ranks": 0,73 "loss_rank_avg": 0.6693063974380493,74 "step": 30,75 "valid_targets_mean": 2645.9,76 "valid_targets_min": 81177 },78 {79 "epoch": 0.052552552552552555,80 "grad_norm": 2.254075439509915,81 "learning_rate": 2.9122055674518203e-06,82 "loss": 0.583,83 "loss_nan_ranks": 0,84 "loss_rank_avg": 0.6287423968315125,85 "step": 35,86 "valid_targets_mean": 5337.9,87 "valid_targets_min": 42488 },89 {90 "epoch": 0.06006006006006006,91 "grad_norm": 1.5869006029617074,92 "learning_rate": 3.3404710920770882e-06,93 "loss": 0.4981,94 "loss_nan_ranks": 0,95 "loss_rank_avg": 0.5136668682098389,96 "step": 40,97 "valid_targets_mean": 4675.1,98 "valid_targets_min": 68399 },100 {101 "epoch": 0.06756756756756757,102 "grad_norm": 1.2871585570908246,103 "learning_rate": 3.7687366167023558e-06,104 "loss": 0.498,105 "loss_nan_ranks": 0,106 "loss_rank_avg": 0.4761298894882202,107 "step": 45,108 "valid_targets_mean": 4580.2,109 "valid_targets_min": 695110 },111 {112 "epoch": 0.07507507507507508,113 "grad_norm": 1.1073292407135484,114 "learning_rate": 4.197002141327624e-06,115 "loss": 0.4618,116 "loss_nan_ranks": 0,117 "loss_rank_avg": 0.4800295829772949,118 "step": 50,119 "valid_targets_mean": 3829.7,120 "valid_targets_min": 408121 },122 {123 "epoch": 0.08258258258258258,124 "grad_norm": 0.8919504665868666,125 "learning_rate": 4.625267665952891e-06,126 "loss": 0.5237,127 "loss_nan_ranks": 0,128 "loss_rank_avg": 0.525561511516571,129 "step": 55,130 "valid_targets_mean": 5515.5,131 "valid_targets_min": 571132 },133 {134 "epoch": 0.09009009009009009,135 "grad_norm": 0.865780829661676,136 "learning_rate": 5.05353319057816e-06,137 "loss": 0.5163,138 "loss_nan_ranks": 0,139 "loss_rank_avg": 0.5415364503860474,140 "step": 60,141 "valid_targets_mean": 3433.2,142 "valid_targets_min": 511143 },144 {145 "epoch": 0.09759759759759759,146 "grad_norm": 0.5872913252913201,147 "learning_rate": 5.481798715203427e-06,148 "loss": 0.4707,149 "loss_nan_ranks": 0,150 "loss_rank_avg": 0.3888802230358124,151 "step": 65,152 "valid_targets_mean": 5232.6,153 "valid_targets_min": 664154 },155 {156 "epoch": 0.10510510510510511,157 "grad_norm": 0.5994653626627001,158 "learning_rate": 5.910064239828695e-06,159 "loss": 0.4821,160 "loss_nan_ranks": 0,161 "loss_rank_avg": 0.43450668454170227,162 "step": 70,163 "valid_targets_mean": 5187.1,164 "valid_targets_min": 460165 },166 {167 "epoch": 0.11261261261261261,168 "grad_norm": 0.6662883071321841,169 "learning_rate": 6.338329764453962e-06,170 "loss": 0.4331,171 "loss_nan_ranks": 0,172 "loss_rank_avg": 0.3964645564556122,173 "step": 75,174 "valid_targets_mean": 3182.0,175 "valid_targets_min": 880176 },177 {178 "epoch": 0.12012012012012012,179 "grad_norm": 0.6301518751447736,180 "learning_rate": 6.76659528907923e-06,181 "loss": 0.4374,182 "loss_nan_ranks": 0,183 "loss_rank_avg": 0.37995022535324097,184 "step": 80,185 "valid_targets_mean": 3907.1,186 "valid_targets_min": 773187 },188 {189 "epoch": 0.12762762762762764,190 "grad_norm": 0.6479909629866574,191 "learning_rate": 7.194860813704497e-06,192 "loss": 0.4749,193 "loss_nan_ranks": 0,194 "loss_rank_avg": 0.43532025814056396,195 "step": 85,196 "valid_targets_mean": 3381.8,197 "valid_targets_min": 727198 },199 {200 "epoch": 0.13513513513513514,201 "grad_norm": 0.601954910384077,202 "learning_rate": 7.623126338329765e-06,203 "loss": 0.449,204 "loss_nan_ranks": 0,205 "loss_rank_avg": 0.37232670187950134,206 "step": 90,207 "valid_targets_mean": 3859.4,208 "valid_targets_min": 681209 },210 {211 "epoch": 0.14264264264264265,212 "grad_norm": 0.6010699211352588,213 "learning_rate": 8.051391862955033e-06,214 "loss": 0.4036,215 "loss_nan_ranks": 0,216 "loss_rank_avg": 0.3912065625190735,217 "step": 95,218 "valid_targets_mean": 4532.2,219 "valid_targets_min": 525220 },221 {222 "epoch": 0.15015015015015015,223 "grad_norm": 0.6776934870921543,224 "learning_rate": 8.4796573875803e-06,225 "loss": 0.4009,226 "loss_nan_ranks": 0,227 "loss_rank_avg": 0.3918367922306061,228 "step": 100,229 "valid_targets_mean": 3923.7,230 "valid_targets_min": 631231 },232 {233 "epoch": 0.15765765765765766,234 "grad_norm": 0.5815397148408697,235 "learning_rate": 8.907922912205568e-06,236 "loss": 0.422,237 "loss_nan_ranks": 0,238 "loss_rank_avg": 0.40705981850624084,239 "step": 105,240 "valid_targets_mean": 4907.4,241 "valid_targets_min": 624242 },243 {244 "epoch": 0.16516516516516516,245 "grad_norm": 0.6207231229898003,246 "learning_rate": 9.336188436830836e-06,247 "loss": 0.4197,248 "loss_nan_ranks": 0,249 "loss_rank_avg": 0.35407790541648865,250 "step": 110,251 "valid_targets_mean": 4749.1,252 "valid_targets_min": 357253 },254 {255 "epoch": 0.17267267267267267,256 "grad_norm": 0.7161182478604338,257 "learning_rate": 9.764453961456105e-06,258 "loss": 0.4271,259 "loss_nan_ranks": 0,260 "loss_rank_avg": 0.3748543858528137,261 "step": 115,262 "valid_targets_mean": 2599.0,263 "valid_targets_min": 495264 },265 {266 "epoch": 0.18018018018018017,267 "grad_norm": 0.6433782885571062,268 "learning_rate": 1.019271948608137e-05,269 "loss": 0.4129,270 "loss_nan_ranks": 0,271 "loss_rank_avg": 0.44952958822250366,272 "step": 120,273 "valid_targets_mean": 3456.8,274 "valid_targets_min": 521275 },276 {277 "epoch": 0.18768768768768768,278 "grad_norm": 0.608760987576289,279 "learning_rate": 1.062098501070664e-05,280 "loss": 0.396,281 "loss_nan_ranks": 0,282 "loss_rank_avg": 0.4421440064907074,283 "step": 125,284 "valid_targets_mean": 4304.8,285 "valid_targets_min": 676286 },287 {288 "epoch": 0.19519519519519518,289 "grad_norm": 0.6056247540086898,290 "learning_rate": 1.1049250535331907e-05,291 "loss": 0.356,292 "loss_nan_ranks": 0,293 "loss_rank_avg": 0.3524633049964905,294 "step": 130,295 "valid_targets_mean": 4378.4,296 "valid_targets_min": 699297 },298 {299 "epoch": 0.20270270270270271,300 "grad_norm": 0.7470542876873747,301 "learning_rate": 1.1477516059957175e-05,302 "loss": 0.3928,303 "loss_nan_ranks": 0,304 "loss_rank_avg": 0.3509212136268616,305 "step": 135,306 "valid_targets_mean": 3348.1,307 "valid_targets_min": 580308 },309 {310 "epoch": 0.21021021021021022,311 "grad_norm": 0.5297517332166896,312 "learning_rate": 1.1905781584582443e-05,313 "loss": 0.3971,314 "loss_nan_ranks": 0,315 "loss_rank_avg": 0.3493044376373291,316 "step": 140,317 "valid_targets_mean": 4820.6,318 "valid_targets_min": 588319 },320 {321 "epoch": 0.21771771771771772,322 "grad_norm": 0.8236972811131019,323 "learning_rate": 1.233404710920771e-05,324 "loss": 0.4229,325 "loss_nan_ranks": 0,326 "loss_rank_avg": 0.4382990002632141,327 "step": 145,328 "valid_targets_mean": 4910.1,329 "valid_targets_min": 704330 },331 {332 "epoch": 0.22522522522522523,333 "grad_norm": 0.5641188867988419,334 "learning_rate": 1.2762312633832978e-05,335 "loss": 0.4053,336 "loss_nan_ranks": 0,337 "loss_rank_avg": 0.4189315438270569,338 "step": 150,339 "valid_targets_mean": 4847.2,340 "valid_targets_min": 479341 },342 {343 "epoch": 0.23273273273273273,344 "grad_norm": 0.759436996305121,345 "learning_rate": 1.3190578158458245e-05,346 "loss": 0.3995,347 "loss_nan_ranks": 0,348 "loss_rank_avg": 0.4512358009815216,349 "step": 155,350 "valid_targets_mean": 3609.1,351 "valid_targets_min": 660352 },353 {354 "epoch": 0.24024024024024024,355 "grad_norm": 0.5776616654260357,356 "learning_rate": 1.3618843683083513e-05,357 "loss": 0.3982,358 "loss_nan_ranks": 0,359 "loss_rank_avg": 0.366862416267395,360 "step": 160,361 "valid_targets_mean": 4489.5,362 "valid_targets_min": 709363 },364 {365 "epoch": 0.24774774774774774,366 "grad_norm": 0.5621030834623711,367 "learning_rate": 1.404710920770878e-05,368 "loss": 0.3562,369 "loss_nan_ranks": 0,370 "loss_rank_avg": 0.3245847225189209,371 "step": 165,372 "valid_targets_mean": 3611.0,373 "valid_targets_min": 857374 },375 {376 "epoch": 0.2552552552552553,377 "grad_norm": 0.6383475705856398,378 "learning_rate": 1.4475374732334048e-05,379 "loss": 0.3633,380 "loss_nan_ranks": 0,381 "loss_rank_avg": 0.3641152083873749,382 "step": 170,383 "valid_targets_mean": 3776.7,384 "valid_targets_min": 273385 },386 {387 "epoch": 0.2627627627627628,388 "grad_norm": 0.6387883870614447,389 "learning_rate": 1.4903640256959315e-05,390 "loss": 0.3892,391 "loss_nan_ranks": 0,392 "loss_rank_avg": 0.4122806787490845,393 "step": 175,394 "valid_targets_mean": 4106.8,395 "valid_targets_min": 367396 },397 {398 "epoch": 0.2702702702702703,399 "grad_norm": 0.5859730062305724,400 "learning_rate": 1.5331905781584583e-05,401 "loss": 0.4145,402 "loss_nan_ranks": 0,403 "loss_rank_avg": 0.4545392692089081,404 "step": 180,405 "valid_targets_mean": 5412.8,406 "valid_targets_min": 485407 },408 {409 "epoch": 0.2777777777777778,410 "grad_norm": 0.6291566957212851,411 "learning_rate": 1.576017130620985e-05,412 "loss": 0.3586,413 "loss_nan_ranks": 0,414 "loss_rank_avg": 0.34801965951919556,415 "step": 185,416 "valid_targets_mean": 3683.8,417 "valid_targets_min": 388418 },419 {420 "epoch": 0.2852852852852853,421 "grad_norm": 0.7714952807971188,422 "learning_rate": 1.618843683083512e-05,423 "loss": 0.4031,424 "loss_nan_ranks": 0,425 "loss_rank_avg": 0.33357149362564087,426 "step": 190,427 "valid_targets_mean": 3992.8,428 "valid_targets_min": 645429 },430 {431 "epoch": 0.2927927927927928,432 "grad_norm": 0.6022688124054402,433 "learning_rate": 1.6616702355460387e-05,434 "loss": 0.3632,435 "loss_nan_ranks": 0,436 "loss_rank_avg": 0.32963526248931885,437 "step": 195,438 "valid_targets_mean": 4479.6,439 "valid_targets_min": 376440 },441 {442 "epoch": 0.3003003003003003,443 "grad_norm": 0.5461125797479052,444 "learning_rate": 1.7044967880085653e-05,445 "loss": 0.3609,446 "loss_nan_ranks": 0,447 "loss_rank_avg": 0.3471255898475647,448 "step": 200,449 "valid_targets_mean": 5124.2,450 "valid_targets_min": 2440451 },452 {453 "epoch": 0.3078078078078078,454 "grad_norm": 0.6479231652954496,455 "learning_rate": 1.7473233404710924e-05,456 "loss": 0.3709,457 "loss_nan_ranks": 0,458 "loss_rank_avg": 0.36097437143325806,459 "step": 205,460 "valid_targets_mean": 3874.2,461 "valid_targets_min": 536462 },463 {464 "epoch": 0.3153153153153153,465 "grad_norm": 0.6307321400936947,466 "learning_rate": 1.790149892933619e-05,467 "loss": 0.356,468 "loss_nan_ranks": 0,469 "loss_rank_avg": 0.32168492674827576,470 "step": 210,471 "valid_targets_mean": 4215.9,472 "valid_targets_min": 750473 },474 {475 "epoch": 0.3228228228228228,476 "grad_norm": 0.7342733098843338,477 "learning_rate": 1.8329764453961457e-05,478 "loss": 0.3716,479 "loss_nan_ranks": 0,480 "loss_rank_avg": 0.3192724585533142,481 "step": 215,482 "valid_targets_mean": 3300.5,483 "valid_targets_min": 465484 },485 {486 "epoch": 0.3303303303303303,487 "grad_norm": 0.6269217295173055,488 "learning_rate": 1.8758029978586724e-05,489 "loss": 0.382,490 "loss_nan_ranks": 0,491 "loss_rank_avg": 0.393771767616272,492 "step": 220,493 "valid_targets_mean": 3780.5,494 "valid_targets_min": 520495 },496 {497 "epoch": 0.33783783783783783,498 "grad_norm": 0.6699586606597604,499 "learning_rate": 1.9186295503211994e-05,500 "loss": 0.3367,501 "loss_nan_ranks": 0,502 "loss_rank_avg": 0.3441130518913269,503 "step": 225,504 "valid_targets_mean": 3570.4,505 "valid_targets_min": 367506 },507 {508 "epoch": 0.34534534534534533,509 "grad_norm": 0.6212030440388314,510 "learning_rate": 1.961456102783726e-05,511 "loss": 0.3249,512 "loss_nan_ranks": 0,513 "loss_rank_avg": 0.28860020637512207,514 "step": 230,515 "valid_targets_mean": 4043.8,516 "valid_targets_min": 797517 },518 {519 "epoch": 0.35285285285285284,520 "grad_norm": 0.5933247879185318,521 "learning_rate": 2.004282655246253e-05,522 "loss": 0.3663,523 "loss_nan_ranks": 0,524 "loss_rank_avg": 0.2963605523109436,525 "step": 235,526 "valid_targets_mean": 4519.4,527 "valid_targets_min": 679528 },529 {530 "epoch": 0.36036036036036034,531 "grad_norm": 0.5915857140141431,532 "learning_rate": 2.0471092077087794e-05,533 "loss": 0.3707,534 "loss_nan_ranks": 0,535 "loss_rank_avg": 0.3590105175971985,536 "step": 240,537 "valid_targets_mean": 4801.9,538 "valid_targets_min": 786539 },540 {541 "epoch": 0.36786786786786785,542 "grad_norm": 0.6926624493340766,543 "learning_rate": 2.089935760171306e-05,544 "loss": 0.3699,545 "loss_nan_ranks": 0,546 "loss_rank_avg": 0.3375079035758972,547 "step": 245,548 "valid_targets_mean": 3064.8,549 "valid_targets_min": 636550 },551 {552 "epoch": 0.37537537537537535,553 "grad_norm": 0.5625060900455121,554 "learning_rate": 2.1327623126338334e-05,555 "loss": 0.3717,556 "loss_nan_ranks": 0,557 "loss_rank_avg": 0.2878052592277527,558 "step": 250,559 "valid_targets_mean": 4984.9,560 "valid_targets_min": 389561 },562 {563 "epoch": 0.38288288288288286,564 "grad_norm": 0.4895346484151352,565 "learning_rate": 2.17558886509636e-05,566 "loss": 0.3319,567 "loss_nan_ranks": 0,568 "loss_rank_avg": 0.3224838972091675,569 "step": 255,570 "valid_targets_mean": 5919.2,571 "valid_targets_min": 631572 },573 {574 "epoch": 0.39039039039039036,575 "grad_norm": 0.6186907667129198,576 "learning_rate": 2.2184154175588867e-05,577 "loss": 0.3133,578 "loss_nan_ranks": 0,579 "loss_rank_avg": 0.3299029767513275,580 "step": 260,581 "valid_targets_mean": 4151.9,582 "valid_targets_min": 694583 },584 {585 "epoch": 0.3978978978978979,586 "grad_norm": 0.637540257330486,587 "learning_rate": 2.2612419700214137e-05,588 "loss": 0.3034,589 "loss_nan_ranks": 0,590 "loss_rank_avg": 0.31127655506134033,591 "step": 265,592 "valid_targets_mean": 4565.4,593 "valid_targets_min": 667594 },595 {596 "epoch": 0.40540540540540543,597 "grad_norm": 0.6482997719326944,598 "learning_rate": 2.3040685224839404e-05,599 "loss": 0.371,600 "loss_nan_ranks": 0,601 "loss_rank_avg": 0.3371477723121643,602 "step": 270,603 "valid_targets_mean": 4799.7,604 "valid_targets_min": 565605 },606 {607 "epoch": 0.41291291291291293,608 "grad_norm": 0.630240697831865,609 "learning_rate": 2.346895074946467e-05,610 "loss": 0.4042,611 "loss_nan_ranks": 0,612 "loss_rank_avg": 0.3549198508262634,613 "step": 275,614 "valid_targets_mean": 3973.2,615 "valid_targets_min": 571616 },617 {618 "epoch": 0.42042042042042044,619 "grad_norm": 0.5378005948550142,620 "learning_rate": 2.3897216274089937e-05,621 "loss": 0.3568,622 "loss_nan_ranks": 0,623 "loss_rank_avg": 0.3140275776386261,624 "step": 280,625 "valid_targets_mean": 4524.4,626 "valid_targets_min": 466627 },628 {629 "epoch": 0.42792792792792794,630 "grad_norm": 0.5496707584360776,631 "learning_rate": 2.4325481798715207e-05,632 "loss": 0.3439,633 "loss_nan_ranks": 0,634 "loss_rank_avg": 0.27984702587127686,635 "step": 285,636 "valid_targets_mean": 5251.4,637 "valid_targets_min": 871638 },639 {640 "epoch": 0.43543543543543545,641 "grad_norm": 0.6054529261217243,642 "learning_rate": 2.4753747323340474e-05,643 "loss": 0.3399,644 "loss_nan_ranks": 0,645 "loss_rank_avg": 0.2717384994029999,646 "step": 290,647 "valid_targets_mean": 5978.3,648 "valid_targets_min": 665649 },650 {651 "epoch": 0.44294294294294295,652 "grad_norm": 0.60946853866193,653 "learning_rate": 2.518201284796574e-05,654 "loss": 0.3324,655 "loss_nan_ranks": 0,656 "loss_rank_avg": 0.31978553533554077,657 "step": 295,658 "valid_targets_mean": 3610.3,659 "valid_targets_min": 729660 },661 {662 "epoch": 0.45045045045045046,663 "grad_norm": 0.7112237553385803,664 "learning_rate": 2.5610278372591007e-05,665 "loss": 0.365,666 "loss_nan_ranks": 0,667 "loss_rank_avg": 0.36318349838256836,668 "step": 300,669 "valid_targets_mean": 2997.6,670 "valid_targets_min": 398671 },672 {673 "epoch": 0.45795795795795796,674 "grad_norm": 0.5754561363093669,675 "learning_rate": 2.6038543897216277e-05,676 "loss": 0.3475,677 "loss_nan_ranks": 0,678 "loss_rank_avg": 0.3775876760482788,679 "step": 305,680 "valid_targets_mean": 5485.2,681 "valid_targets_min": 291682 },683 {684 "epoch": 0.46546546546546547,685 "grad_norm": 0.5225836627231101,686 "learning_rate": 2.6466809421841544e-05,687 "loss": 0.3327,688 "loss_nan_ranks": 0,689 "loss_rank_avg": 0.2630726099014282,690 "step": 310,691 "valid_targets_mean": 5477.4,692 "valid_targets_min": 1953693 },694 {695 "epoch": 0.47297297297297297,696 "grad_norm": 0.6631066150226813,697 "learning_rate": 2.689507494646681e-05,698 "loss": 0.3341,699 "loss_nan_ranks": 0,700 "loss_rank_avg": 0.4078267812728882,701 "step": 315,702 "valid_targets_mean": 4227.4,703 "valid_targets_min": 691704 },705 {706 "epoch": 0.4804804804804805,707 "grad_norm": 0.5625056894327084,708 "learning_rate": 2.732334047109208e-05,709 "loss": 0.3307,710 "loss_nan_ranks": 0,711 "loss_rank_avg": 0.34400713443756104,712 "step": 320,713 "valid_targets_mean": 6082.6,714 "valid_targets_min": 756715 },716 {717 "epoch": 0.487987987987988,718 "grad_norm": 0.5720940827291631,719 "learning_rate": 2.7751605995717347e-05,720 "loss": 0.3544,721 "loss_nan_ranks": 0,722 "loss_rank_avg": 0.364154189825058,723 "step": 325,724 "valid_targets_mean": 5058.0,725 "valid_targets_min": 1324726 },727 {728 "epoch": 0.4954954954954955,729 "grad_norm": 0.6142764560167397,730 "learning_rate": 2.8179871520342614e-05,731 "loss": 0.3567,732 "loss_nan_ranks": 0,733 "loss_rank_avg": 0.3340011239051819,734 "step": 330,735 "valid_targets_mean": 3809.8,736 "valid_targets_min": 606737 },738 {739 "epoch": 0.503003003003003,740 "grad_norm": 0.6862489554890606,741 "learning_rate": 2.860813704496788e-05,742 "loss": 0.3377,743 "loss_nan_ranks": 0,744 "loss_rank_avg": 0.347086638212204,745 "step": 335,746 "valid_targets_mean": 3381.4,747 "valid_targets_min": 529748 },749 {750 "epoch": 0.5105105105105106,751 "grad_norm": 0.5884162448896875,752 "learning_rate": 2.903640256959315e-05,753 "loss": 0.3156,754 "loss_nan_ranks": 0,755 "loss_rank_avg": 0.3551441431045532,756 "step": 340,757 "valid_targets_mean": 4280.0,758 "valid_targets_min": 656759 },760 {761 "epoch": 0.5180180180180181,762 "grad_norm": 0.5985519809481195,763 "learning_rate": 2.9464668094218417e-05,764 "loss": 0.3718,765 "loss_nan_ranks": 0,766 "loss_rank_avg": 0.3538270592689514,767 "step": 345,768 "valid_targets_mean": 4933.6,769 "valid_targets_min": 679770 },771 {772 "epoch": 0.5255255255255256,773 "grad_norm": 0.6608489868770677,774 "learning_rate": 2.9892933618843684e-05,775 "loss": 0.3821,776 "loss_nan_ranks": 0,777 "loss_rank_avg": 0.3977230191230774,778 "step": 350,779 "valid_targets_mean": 4619.8,780 "valid_targets_min": 617781 },782 {783 "epoch": 0.5330330330330331,784 "grad_norm": 0.6666435654761541,785 "learning_rate": 3.032119914346895e-05,786 "loss": 0.3205,787 "loss_nan_ranks": 0,788 "loss_rank_avg": 0.3276807963848114,789 "step": 355,790 "valid_targets_mean": 3228.0,791 "valid_targets_min": 396792 },793 {794 "epoch": 0.5405405405405406,795 "grad_norm": 0.5700169845584041,796 "learning_rate": 3.0749464668094224e-05,797 "loss": 0.3373,798 "loss_nan_ranks": 0,799 "loss_rank_avg": 0.252112478017807,800 "step": 360,801 "valid_targets_mean": 4043.2,802 "valid_targets_min": 470803 },804 {805 "epoch": 0.5480480480480481,806 "grad_norm": 0.5969317196396643,807 "learning_rate": 3.117773019271949e-05,808 "loss": 0.3344,809 "loss_nan_ranks": 0,810 "loss_rank_avg": 0.257426917552948,811 "step": 365,812 "valid_targets_mean": 3501.2,813 "valid_targets_min": 637814 },815 {816 "epoch": 0.5555555555555556,817 "grad_norm": 0.5338582932799799,818 "learning_rate": 3.160599571734476e-05,819 "loss": 0.3469,820 "loss_nan_ranks": 0,821 "loss_rank_avg": 0.4118390679359436,822 "step": 370,823 "valid_targets_mean": 6022.6,824 "valid_targets_min": 668825 },826 {827 "epoch": 0.5630630630630631,828 "grad_norm": 0.7156868770693144,829 "learning_rate": 3.2034261241970024e-05,830 "loss": 0.342,831 "loss_nan_ranks": 0,832 "loss_rank_avg": 0.33580997586250305,833 "step": 375,834 "valid_targets_mean": 3146.6,835 "valid_targets_min": 580836 },837 {838 "epoch": 0.5705705705705706,839 "grad_norm": 0.6105442252909425,840 "learning_rate": 3.246252676659529e-05,841 "loss": 0.3246,842 "loss_nan_ranks": 0,843 "loss_rank_avg": 0.29976391792297363,844 "step": 380,845 "valid_targets_mean": 4423.6,846 "valid_targets_min": 579847 },848 {849 "epoch": 0.5780780780780781,850 "grad_norm": 0.48535869781119606,851 "learning_rate": 3.289079229122056e-05,852 "loss": 0.3211,853 "loss_nan_ranks": 0,854 "loss_rank_avg": 0.2958953380584717,855 "step": 385,856 "valid_targets_mean": 5096.5,857 "valid_targets_min": 1111858 },859 {860 "epoch": 0.5855855855855856,861 "grad_norm": 0.5293285996606892,862 "learning_rate": 3.3319057815845824e-05,863 "loss": 0.3012,864 "loss_nan_ranks": 0,865 "loss_rank_avg": 0.29280033707618713,866 "step": 390,867 "valid_targets_mean": 4133.6,868 "valid_targets_min": 664869 },870 {871 "epoch": 0.5930930930930931,872 "grad_norm": 0.6364074446274751,873 "learning_rate": 3.37473233404711e-05,874 "loss": 0.3546,875 "loss_nan_ranks": 0,876 "loss_rank_avg": 0.3336215019226074,877 "step": 395,878 "valid_targets_mean": 3474.1,879 "valid_targets_min": 277880 },881 {882 "epoch": 0.6006006006006006,883 "grad_norm": 0.7064277061832516,884 "learning_rate": 3.4175588865096364e-05,885 "loss": 0.3238,886 "loss_nan_ranks": 0,887 "loss_rank_avg": 0.38183677196502686,888 "step": 400,889 "valid_targets_mean": 4343.8,890 "valid_targets_min": 831891 },892 {893 "epoch": 0.6081081081081081,894 "grad_norm": 0.5234255969114723,895 "learning_rate": 3.460385438972163e-05,896 "loss": 0.3259,897 "loss_nan_ranks": 0,898 "loss_rank_avg": 0.3263946771621704,899 "step": 405,900 "valid_targets_mean": 5912.4,901 "valid_targets_min": 525902 },903 {904 "epoch": 0.6156156156156156,905 "grad_norm": 0.6859056535544392,906 "learning_rate": 3.50321199143469e-05,907 "loss": 0.3613,908 "loss_nan_ranks": 0,909 "loss_rank_avg": 0.4518062472343445,910 "step": 410,911 "valid_targets_mean": 4294.0,912 "valid_targets_min": 672913 },914 {915 "epoch": 0.6231231231231231,916 "grad_norm": 0.593812126409495,917 "learning_rate": 3.5460385438972164e-05,918 "loss": 0.3236,919 "loss_nan_ranks": 0,920 "loss_rank_avg": 0.284976601600647,921 "step": 415,922 "valid_targets_mean": 4542.8,923 "valid_targets_min": 732924 },925 {926 "epoch": 0.6306306306306306,927 "grad_norm": 0.5542761631226861,928 "learning_rate": 3.588865096359743e-05,929 "loss": 0.3575,930 "loss_nan_ranks": 0,931 "loss_rank_avg": 0.3369911313056946,932 "step": 420,933 "valid_targets_mean": 4173.8,934 "valid_targets_min": 623935 },936 {937 "epoch": 0.6381381381381381,938 "grad_norm": 0.694482077890855,939 "learning_rate": 3.63169164882227e-05,940 "loss": 0.331,941 "loss_nan_ranks": 0,942 "loss_rank_avg": 0.35457199811935425,943 "step": 425,944 "valid_targets_mean": 3347.3,945 "valid_targets_min": 323946 },947 {948 "epoch": 0.6456456456456456,949 "grad_norm": 0.5087776401105766,950 "learning_rate": 3.674518201284797e-05,951 "loss": 0.3077,952 "loss_nan_ranks": 0,953 "loss_rank_avg": 0.2870054244995117,954 "step": 430,955 "valid_targets_mean": 5496.8,956 "valid_targets_min": 995957 },958 {959 "epoch": 0.6531531531531531,960 "grad_norm": 0.5710934865190025,961 "learning_rate": 3.717344753747324e-05,962 "loss": 0.3449,963 "loss_nan_ranks": 0,964 "loss_rank_avg": 0.3406960964202881,965 "step": 435,966 "valid_targets_mean": 4272.1,967 "valid_targets_min": 578968 },969 {970 "epoch": 0.6606606606606606,971 "grad_norm": 0.6532417856194169,972 "learning_rate": 3.7601713062098504e-05,973 "loss": 0.3358,974 "loss_nan_ranks": 0,975 "loss_rank_avg": 0.3105970621109009,976 "step": 440,977 "valid_targets_mean": 4052.7,978 "valid_targets_min": 469979 },980 {981 "epoch": 0.6681681681681682,982 "grad_norm": 0.5810179731496691,983 "learning_rate": 3.802997858672377e-05,984 "loss": 0.3534,985 "loss_nan_ranks": 0,986 "loss_rank_avg": 0.2967357933521271,987 "step": 445,988 "valid_targets_mean": 5399.5,989 "valid_targets_min": 481990 },991 {992 "epoch": 0.6756756756756757,993 "grad_norm": 0.4933051723343659,994 "learning_rate": 3.8458244111349045e-05,995 "loss": 0.307,996 "loss_nan_ranks": 0,997 "loss_rank_avg": 0.2656002640724182,998 "step": 450,999 "valid_targets_mean": 5050.2,1000 "valid_targets_min": 5361001 },1002 {1003 "epoch": 0.6831831831831832,1004 "grad_norm": 0.5117823141215931,1005 "learning_rate": 3.8886509635974305e-05,1006 "loss": 0.3345,1007 "loss_nan_ranks": 0,1008 "loss_rank_avg": 0.3394744396209717,1009 "step": 455,1010 "valid_targets_mean": 4780.2,1011 "valid_targets_min": 6981012 },1013 {1014 "epoch": 0.6906906906906907,1015 "grad_norm": 0.5513150421712305,1016 "learning_rate": 3.931477516059957e-05,1017 "loss": 0.3344,1018 "loss_nan_ranks": 0,1019 "loss_rank_avg": 0.38678789138793945,1020 "step": 460,1021 "valid_targets_mean": 4995.6,1022 "valid_targets_min": 7871023 },1024 {1025 "epoch": 0.6981981981981982,1026 "grad_norm": 0.5223578821175074,1027 "learning_rate": 3.974304068522484e-05,1028 "loss": 0.3301,1029 "loss_nan_ranks": 0,1030 "loss_rank_avg": 0.3217771053314209,1031 "step": 465,1032 "valid_targets_mean": 5331.4,1033 "valid_targets_min": 10071034 },1035 {1036 "epoch": 0.7057057057057057,1037 "grad_norm": 0.5938737922502441,1038 "learning_rate": 3.999997756656772e-05,1039 "loss": 0.3306,1040 "loss_nan_ranks": 0,1041 "loss_rank_avg": 0.31824183464050293,1042 "step": 470,1043 "valid_targets_mean": 4223.0,1044 "valid_targets_min": 4521045 },1046 {1047 "epoch": 0.7132132132132132,1048 "grad_norm": 0.49626310595338613,1049 "learning_rate": 3.9999725191032436e-05,1050 "loss": 0.2875,1051 "loss_nan_ranks": 0,1052 "loss_rank_avg": 0.26925772428512573,1053 "step": 475,1054 "valid_targets_mean": 5763.9,1055 "valid_targets_min": 7611056 },1057 {1058 "epoch": 0.7207207207207207,1059 "grad_norm": 0.6276161448251935,1060 "learning_rate": 3.999919240172184e-05,1061 "loss": 0.3228,1062 "loss_nan_ranks": 0,1063 "loss_rank_avg": 0.325122594833374,1064 "step": 480,1065 "valid_targets_mean": 4395.4,1066 "valid_targets_min": 6621067 },1068 {1069 "epoch": 0.7282282282282282,1070 "grad_norm": 0.598897987227093,1071 "learning_rate": 3.9998379206106104e-05,1072 "loss": 0.2941,1073 "loss_nan_ranks": 0,1074 "loss_rank_avg": 0.2764396369457245,1075 "step": 485,1076 "valid_targets_mean": 5031.5,1077 "valid_targets_min": 8361078 },1079 {1080 "epoch": 0.7357357357357357,1081 "grad_norm": 0.5920738798268582,1082 "learning_rate": 3.9997285615586944e-05,1083 "loss": 0.3449,1084 "loss_nan_ranks": 0,1085 "loss_rank_avg": 0.33547401428222656,1086 "step": 490,1087 "valid_targets_mean": 4218.7,1088 "valid_targets_min": 4161089 },1090 {1091 "epoch": 0.7432432432432432,1092 "grad_norm": 0.5388292737396405,1093 "learning_rate": 3.999591164549747e-05,1094 "loss": 0.3216,1095 "loss_nan_ranks": 0,1096 "loss_rank_avg": 0.30026373267173767,1097 "step": 495,1098 "valid_targets_mean": 5046.2,1099 "valid_targets_min": 8851100 },1101 {1102 "epoch": 0.7507507507507507,1103 "grad_norm": 0.7138413019632255,1104 "learning_rate": 3.999425731510195e-05,1105 "loss": 0.3161,1106 "loss_nan_ranks": 0,1107 "loss_rank_avg": 0.3095625340938568,1108 "step": 500,1109 "valid_targets_mean": 3115.5,1110 "valid_targets_min": 6961111 },1112 {1113 "epoch": 0.7582582582582582,1114 "grad_norm": 0.5065067605354636,1115 "learning_rate": 3.9992322647595543e-05,1116 "loss": 0.3132,1117 "loss_nan_ranks": 0,1118 "loss_rank_avg": 0.2668853998184204,1119 "step": 505,1120 "valid_targets_mean": 4890.2,1121 "valid_targets_min": 5831122 },1123 {1124 "epoch": 0.7657657657657657,1125 "grad_norm": 0.48772270569522996,1126 "learning_rate": 3.999010767010401e-05,1127 "loss": 0.2792,1128 "loss_nan_ranks": 0,1129 "loss_rank_avg": 0.21127644181251526,1130 "step": 510,1131 "valid_targets_mean": 4030.5,1132 "valid_targets_min": 6201133 },1134 {1135 "epoch": 0.7732732732732732,1136 "grad_norm": 0.688910852981923,1137 "learning_rate": 3.998761241368328e-05,1138 "loss": 0.3356,1139 "loss_nan_ranks": 0,1140 "loss_rank_avg": 0.31102752685546875,1141 "step": 515,1142 "valid_targets_mean": 2963.6,1143 "valid_targets_min": 6621144 },1145 {1146 "epoch": 0.7807807807807807,1147 "grad_norm": 0.590399156229966,1148 "learning_rate": 3.9984836913319044e-05,1149 "loss": 0.3022,1150 "loss_nan_ranks": 0,1151 "loss_rank_avg": 0.295174777507782,1152 "step": 520,1153 "valid_targets_mean": 3820.8,1154 "valid_targets_min": 5471155 },1156 {1157 "epoch": 0.7882882882882883,1158 "grad_norm": 0.7498553089205983,1159 "learning_rate": 3.998178120792627e-05,1160 "loss": 0.3289,1161 "loss_nan_ranks": 0,1162 "loss_rank_avg": 0.29107627272605896,1163 "step": 525,1164 "valid_targets_mean": 4623.9,1165 "valid_targets_min": 4631166 },1167 {1168 "epoch": 0.7957957957957958,1169 "grad_norm": 0.541430521248251,1170 "learning_rate": 3.9978445340348646e-05,1171 "loss": 0.3114,1172 "loss_nan_ranks": 0,1173 "loss_rank_avg": 0.35057809948921204,1174 "step": 530,1175 "valid_targets_mean": 4895.9,1176 "valid_targets_min": 8101177 },1178 {1179 "epoch": 0.8033033033033034,1180 "grad_norm": 0.6145616694631061,1181 "learning_rate": 3.997482935735796e-05,1182 "loss": 0.2989,1183 "loss_nan_ranks": 0,1184 "loss_rank_avg": 0.2825467586517334,1185 "step": 535,1186 "valid_targets_mean": 4079.4,1187 "valid_targets_min": 5931188 },1189 {1190 "epoch": 0.8108108108108109,1191 "grad_norm": 0.6457216047110463,1192 "learning_rate": 3.997093330965349e-05,1193 "loss": 0.3289,1194 "loss_nan_ranks": 0,1195 "loss_rank_avg": 0.37978658080101013,1196 "step": 540,1197 "valid_targets_mean": 3493.6,1198 "valid_targets_min": 5131199 },1200 {