DCAgent/a1-stack_cpp
023
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": 4410,8 "is_hyper_param_search": false,9 "is_local_process_zero": true,10 "is_world_process_zero": true,11 "log_history": [12 {13 "epoch": 0.007936507936507936,14 "grad_norm": 13.936445999727159,15 "learning_rate": 3.6281179138322e-07,16 "loss": 0.5831,17 "loss_nan_ranks": 0,18 "loss_rank_avg": 0.5446114540100098,19 "step": 5,20 "valid_targets_mean": 5181.8,21 "valid_targets_min": 133322 },23 {24 "epoch": 0.015873015873015872,25 "grad_norm": 14.14405406674101,26 "learning_rate": 8.163265306122449e-07,27 "loss": 0.6026,28 "loss_nan_ranks": 0,29 "loss_rank_avg": 0.6077768802642822,30 "step": 10,31 "valid_targets_mean": 5303.6,32 "valid_targets_min": 94833 },34 {35 "epoch": 0.023809523809523808,36 "grad_norm": 13.385853949153486,37 "learning_rate": 1.26984126984127e-06,38 "loss": 0.5824,39 "loss_nan_ranks": 0,40 "loss_rank_avg": 0.5799890160560608,41 "step": 15,42 "valid_targets_mean": 5301.1,43 "valid_targets_min": 100044 },45 {46 "epoch": 0.031746031746031744,47 "grad_norm": 10.791756912090966,48 "learning_rate": 1.723356009070295e-06,49 "loss": 0.5901,50 "loss_nan_ranks": 0,51 "loss_rank_avg": 0.5597653985023499,52 "step": 20,53 "valid_targets_mean": 4823.9,54 "valid_targets_min": 26355 },56 {57 "epoch": 0.03968253968253968,58 "grad_norm": 6.581205621105729,59 "learning_rate": 2.17687074829932e-06,60 "loss": 0.5323,61 "loss_nan_ranks": 0,62 "loss_rank_avg": 0.4685205817222595,63 "step": 25,64 "valid_targets_mean": 4813.8,65 "valid_targets_min": 33066 },67 {68 "epoch": 0.047619047619047616,69 "grad_norm": 3.7853786833152236,70 "learning_rate": 2.6303854875283447e-06,71 "loss": 0.4573,72 "loss_nan_ranks": 0,73 "loss_rank_avg": 0.4164578914642334,74 "step": 30,75 "valid_targets_mean": 5283.2,76 "valid_targets_min": 53877 },78 {79 "epoch": 0.05555555555555555,80 "grad_norm": 2.602330607184181,81 "learning_rate": 3.08390022675737e-06,82 "loss": 0.423,83 "loss_nan_ranks": 0,84 "loss_rank_avg": 0.39225059747695923,85 "step": 35,86 "valid_targets_mean": 4700.6,87 "valid_targets_min": 239088 },89 {90 "epoch": 0.06349206349206349,91 "grad_norm": 1.4233023861991225,92 "learning_rate": 3.537414965986395e-06,93 "loss": 0.3809,94 "loss_nan_ranks": 0,95 "loss_rank_avg": 0.35967135429382324,96 "step": 40,97 "valid_targets_mean": 5445.3,98 "valid_targets_min": 65799 },100 {101 "epoch": 0.07142857142857142,102 "grad_norm": 1.0170115963512993,103 "learning_rate": 3.99092970521542e-06,104 "loss": 0.402,105 "loss_nan_ranks": 0,106 "loss_rank_avg": 0.3903147876262665,107 "step": 45,108 "valid_targets_mean": 6464.0,109 "valid_targets_min": 3085110 },111 {112 "epoch": 0.07936507936507936,113 "grad_norm": 0.8370897006461435,114 "learning_rate": 4.444444444444444e-06,115 "loss": 0.3794,116 "loss_nan_ranks": 0,117 "loss_rank_avg": 0.33025485277175903,118 "step": 50,119 "valid_targets_mean": 5718.5,120 "valid_targets_min": 975121 },122 {123 "epoch": 0.0873015873015873,124 "grad_norm": 0.7252658896074926,125 "learning_rate": 4.897959183673469e-06,126 "loss": 0.3611,127 "loss_nan_ranks": 0,128 "loss_rank_avg": 0.34258341789245605,129 "step": 55,130 "valid_targets_mean": 5999.1,131 "valid_targets_min": 989132 },133 {134 "epoch": 0.09523809523809523,135 "grad_norm": 0.756722250222215,136 "learning_rate": 5.3514739229024945e-06,137 "loss": 0.3478,138 "loss_nan_ranks": 0,139 "loss_rank_avg": 0.3988223075866699,140 "step": 60,141 "valid_targets_mean": 4506.9,142 "valid_targets_min": 290143 },144 {145 "epoch": 0.10317460317460317,146 "grad_norm": 0.568685223882473,147 "learning_rate": 5.80498866213152e-06,148 "loss": 0.3448,149 "loss_nan_ranks": 0,150 "loss_rank_avg": 0.3268760144710541,151 "step": 65,152 "valid_targets_mean": 6105.9,153 "valid_targets_min": 2717154 },155 {156 "epoch": 0.1111111111111111,157 "grad_norm": 0.5731290626935265,158 "learning_rate": 6.258503401360545e-06,159 "loss": 0.3215,160 "loss_nan_ranks": 0,161 "loss_rank_avg": 0.3184238374233246,162 "step": 70,163 "valid_targets_mean": 5088.2,164 "valid_targets_min": 420165 },166 {167 "epoch": 0.11904761904761904,168 "grad_norm": 0.5739988755146004,169 "learning_rate": 6.71201814058957e-06,170 "loss": 0.3124,171 "loss_nan_ranks": 0,172 "loss_rank_avg": 0.34080153703689575,173 "step": 75,174 "valid_targets_mean": 5674.6,175 "valid_targets_min": 1923176 },177 {178 "epoch": 0.12698412698412698,179 "grad_norm": 0.5021784940463249,180 "learning_rate": 7.165532879818595e-06,181 "loss": 0.3008,182 "loss_nan_ranks": 0,183 "loss_rank_avg": 0.3233030438423157,184 "step": 80,185 "valid_targets_mean": 5277.2,186 "valid_targets_min": 291187 },188 {189 "epoch": 0.1349206349206349,190 "grad_norm": 0.5521401363464609,191 "learning_rate": 7.61904761904762e-06,192 "loss": 0.3086,193 "loss_nan_ranks": 0,194 "loss_rank_avg": 0.2872031331062317,195 "step": 85,196 "valid_targets_mean": 5470.0,197 "valid_targets_min": 737198 },199 {200 "epoch": 0.14285714285714285,201 "grad_norm": 0.4751497127812398,202 "learning_rate": 8.072562358276645e-06,203 "loss": 0.3222,204 "loss_nan_ranks": 0,205 "loss_rank_avg": 0.3212955594062805,206 "step": 90,207 "valid_targets_mean": 6301.1,208 "valid_targets_min": 2854209 },210 {211 "epoch": 0.15079365079365079,212 "grad_norm": 0.43696107468178264,213 "learning_rate": 8.52607709750567e-06,214 "loss": 0.2895,215 "loss_nan_ranks": 0,216 "loss_rank_avg": 0.2621079683303833,217 "step": 95,218 "valid_targets_mean": 5617.8,219 "valid_targets_min": 266220 },221 {222 "epoch": 0.15873015873015872,223 "grad_norm": 0.5242468737414164,224 "learning_rate": 8.979591836734695e-06,225 "loss": 0.3018,226 "loss_nan_ranks": 0,227 "loss_rank_avg": 0.3428232669830322,228 "step": 100,229 "valid_targets_mean": 5715.6,230 "valid_targets_min": 2266231 },232 {233 "epoch": 0.16666666666666666,234 "grad_norm": 0.4723266645052996,235 "learning_rate": 9.43310657596372e-06,236 "loss": 0.2767,237 "loss_nan_ranks": 0,238 "loss_rank_avg": 0.27712035179138184,239 "step": 105,240 "valid_targets_mean": 5371.8,241 "valid_targets_min": 662242 },243 {244 "epoch": 0.1746031746031746,245 "grad_norm": 0.49793096409078624,246 "learning_rate": 9.886621315192746e-06,247 "loss": 0.2986,248 "loss_nan_ranks": 0,249 "loss_rank_avg": 0.2675686478614807,250 "step": 110,251 "valid_targets_mean": 4750.8,252 "valid_targets_min": 662253 },254 {255 "epoch": 0.18253968253968253,256 "grad_norm": 0.4902776042402625,257 "learning_rate": 1.034013605442177e-05,258 "loss": 0.2876,259 "loss_nan_ranks": 0,260 "loss_rank_avg": 0.30433300137519836,261 "step": 115,262 "valid_targets_mean": 5146.1,263 "valid_targets_min": 940264 },265 {266 "epoch": 0.19047619047619047,267 "grad_norm": 0.465412157163947,268 "learning_rate": 1.0793650793650794e-05,269 "loss": 0.2969,270 "loss_nan_ranks": 0,271 "loss_rank_avg": 0.2920241057872772,272 "step": 120,273 "valid_targets_mean": 5453.2,274 "valid_targets_min": 841275 },276 {277 "epoch": 0.1984126984126984,278 "grad_norm": 0.4355039639514751,279 "learning_rate": 1.124716553287982e-05,280 "loss": 0.2503,281 "loss_nan_ranks": 0,282 "loss_rank_avg": 0.2529776692390442,283 "step": 125,284 "valid_targets_mean": 5727.6,285 "valid_targets_min": 2696286 },287 {288 "epoch": 0.20634920634920634,289 "grad_norm": 0.4996682572655225,290 "learning_rate": 1.1700680272108845e-05,291 "loss": 0.2826,292 "loss_nan_ranks": 0,293 "loss_rank_avg": 0.2821199297904968,294 "step": 130,295 "valid_targets_mean": 5396.6,296 "valid_targets_min": 2020297 },298 {299 "epoch": 0.21428571428571427,300 "grad_norm": 0.43094951290178424,301 "learning_rate": 1.215419501133787e-05,302 "loss": 0.2583,303 "loss_nan_ranks": 0,304 "loss_rank_avg": 0.24614138901233673,305 "step": 135,306 "valid_targets_mean": 5842.7,307 "valid_targets_min": 286308 },309 {310 "epoch": 0.2222222222222222,311 "grad_norm": 0.47723431745558176,312 "learning_rate": 1.2607709750566895e-05,313 "loss": 0.2845,314 "loss_nan_ranks": 0,315 "loss_rank_avg": 0.2890702486038208,316 "step": 140,317 "valid_targets_mean": 5354.1,318 "valid_targets_min": 2307319 },320 {321 "epoch": 0.23015873015873015,322 "grad_norm": 0.4617350150011481,323 "learning_rate": 1.3061224489795918e-05,324 "loss": 0.2586,325 "loss_nan_ranks": 0,326 "loss_rank_avg": 0.2510530948638916,327 "step": 145,328 "valid_targets_mean": 5552.1,329 "valid_targets_min": 608330 },331 {332 "epoch": 0.23809523809523808,333 "grad_norm": 0.45824070397194494,334 "learning_rate": 1.3514739229024945e-05,335 "loss": 0.2652,336 "loss_nan_ranks": 0,337 "loss_rank_avg": 0.26986801624298096,338 "step": 150,339 "valid_targets_mean": 5441.2,340 "valid_targets_min": 1175341 },342 {343 "epoch": 0.24603174603174602,344 "grad_norm": 0.4392573652714541,345 "learning_rate": 1.3968253968253968e-05,346 "loss": 0.2548,347 "loss_nan_ranks": 0,348 "loss_rank_avg": 0.2496776133775711,349 "step": 155,350 "valid_targets_mean": 5731.4,351 "valid_targets_min": 1393352 },353 {354 "epoch": 0.25396825396825395,355 "grad_norm": 0.4536964041682826,356 "learning_rate": 1.4421768707482994e-05,357 "loss": 0.2374,358 "loss_nan_ranks": 0,359 "loss_rank_avg": 0.21673960983753204,360 "step": 160,361 "valid_targets_mean": 6079.4,362 "valid_targets_min": 300363 },364 {365 "epoch": 0.2619047619047619,366 "grad_norm": 0.5619742860376928,367 "learning_rate": 1.4875283446712018e-05,368 "loss": 0.2578,369 "loss_nan_ranks": 0,370 "loss_rank_avg": 0.2804248631000519,371 "step": 165,372 "valid_targets_mean": 5392.2,373 "valid_targets_min": 2214374 },375 {376 "epoch": 0.2698412698412698,377 "grad_norm": 0.4588222248723898,378 "learning_rate": 1.5328798185941044e-05,379 "loss": 0.2643,380 "loss_nan_ranks": 0,381 "loss_rank_avg": 0.24595819413661957,382 "step": 170,383 "valid_targets_mean": 5367.2,384 "valid_targets_min": 1806385 },386 {387 "epoch": 0.2777777777777778,388 "grad_norm": 0.4837210875448078,389 "learning_rate": 1.578231292517007e-05,390 "loss": 0.2465,391 "loss_nan_ranks": 0,392 "loss_rank_avg": 0.26808232069015503,393 "step": 175,394 "valid_targets_mean": 4694.2,395 "valid_targets_min": 1970396 },397 {398 "epoch": 0.2857142857142857,399 "grad_norm": 0.4642135470304616,400 "learning_rate": 1.6235827664399097e-05,401 "loss": 0.2568,402 "loss_nan_ranks": 0,403 "loss_rank_avg": 0.25557032227516174,404 "step": 180,405 "valid_targets_mean": 5724.9,406 "valid_targets_min": 1981407 },408 {409 "epoch": 0.29365079365079366,410 "grad_norm": 0.4404678599566564,411 "learning_rate": 1.668934240362812e-05,412 "loss": 0.2319,413 "loss_nan_ranks": 0,414 "loss_rank_avg": 0.2403090000152588,415 "step": 185,416 "valid_targets_mean": 5580.2,417 "valid_targets_min": 1807418 },419 {420 "epoch": 0.30158730158730157,421 "grad_norm": 0.461513428421082,422 "learning_rate": 1.7142857142857142e-05,423 "loss": 0.2717,424 "loss_nan_ranks": 0,425 "loss_rank_avg": 0.2342744767665863,426 "step": 190,427 "valid_targets_mean": 5209.1,428 "valid_targets_min": 528429 },430 {431 "epoch": 0.30952380952380953,432 "grad_norm": 0.4616028237565255,433 "learning_rate": 1.759637188208617e-05,434 "loss": 0.2573,435 "loss_nan_ranks": 0,436 "loss_rank_avg": 0.2381066232919693,437 "step": 195,438 "valid_targets_mean": 4881.2,439 "valid_targets_min": 1007440 },441 {442 "epoch": 0.31746031746031744,443 "grad_norm": 0.5048131717477712,444 "learning_rate": 1.8049886621315194e-05,445 "loss": 0.2497,446 "loss_nan_ranks": 0,447 "loss_rank_avg": 0.24961230158805847,448 "step": 200,449 "valid_targets_mean": 4766.8,450 "valid_targets_min": 760451 },452 {453 "epoch": 0.3253968253968254,454 "grad_norm": 0.4608218526938904,455 "learning_rate": 1.8503401360544218e-05,456 "loss": 0.2423,457 "loss_nan_ranks": 0,458 "loss_rank_avg": 0.22085292637348175,459 "step": 205,460 "valid_targets_mean": 5626.6,461 "valid_targets_min": 311462 },463 {464 "epoch": 0.3333333333333333,465 "grad_norm": 0.5276409706041367,466 "learning_rate": 1.8956916099773243e-05,467 "loss": 0.2558,468 "loss_nan_ranks": 0,469 "loss_rank_avg": 0.2512398362159729,470 "step": 210,471 "valid_targets_mean": 5207.8,472 "valid_targets_min": 901473 },474 {475 "epoch": 0.3412698412698413,476 "grad_norm": 0.44125944724862465,477 "learning_rate": 1.941043083900227e-05,478 "loss": 0.2319,479 "loss_nan_ranks": 0,480 "loss_rank_avg": 0.23039595782756805,481 "step": 215,482 "valid_targets_mean": 6150.5,483 "valid_targets_min": 2819484 },485 {486 "epoch": 0.3492063492063492,487 "grad_norm": 0.8225576873165226,488 "learning_rate": 1.9863945578231295e-05,489 "loss": 0.2654,490 "loss_nan_ranks": 0,491 "loss_rank_avg": 0.2745077311992645,492 "step": 220,493 "valid_targets_mean": 4586.4,494 "valid_targets_min": 889495 },496 {497 "epoch": 0.35714285714285715,498 "grad_norm": 0.44559302418983976,499 "learning_rate": 2.031746031746032e-05,500 "loss": 0.2447,501 "loss_nan_ranks": 0,502 "loss_rank_avg": 0.27263548970222473,503 "step": 225,504 "valid_targets_mean": 5901.4,505 "valid_targets_min": 1813506 },507 {508 "epoch": 0.36507936507936506,509 "grad_norm": 0.4702793229346846,510 "learning_rate": 2.0770975056689343e-05,511 "loss": 0.2457,512 "loss_nan_ranks": 0,513 "loss_rank_avg": 0.2387492060661316,514 "step": 230,515 "valid_targets_mean": 5740.0,516 "valid_targets_min": 1593517 },518 {519 "epoch": 0.373015873015873,520 "grad_norm": 0.5771251589812603,521 "learning_rate": 2.122448979591837e-05,522 "loss": 0.2577,523 "loss_nan_ranks": 0,524 "loss_rank_avg": 0.27772706747055054,525 "step": 235,526 "valid_targets_mean": 5892.1,527 "valid_targets_min": 807528 },529 {530 "epoch": 0.38095238095238093,531 "grad_norm": 0.4716600277551803,532 "learning_rate": 2.1678004535147395e-05,533 "loss": 0.2366,534 "loss_nan_ranks": 0,535 "loss_rank_avg": 0.24430419504642487,536 "step": 240,537 "valid_targets_mean": 5370.8,538 "valid_targets_min": 300539 },540 {541 "epoch": 0.3888888888888889,542 "grad_norm": 0.4834321959641482,543 "learning_rate": 2.213151927437642e-05,544 "loss": 0.2335,545 "loss_nan_ranks": 0,546 "loss_rank_avg": 0.2554604709148407,547 "step": 245,548 "valid_targets_mean": 5334.6,549 "valid_targets_min": 800550 },551 {552 "epoch": 0.3968253968253968,553 "grad_norm": 0.4468651198346908,554 "learning_rate": 2.2585034013605444e-05,555 "loss": 0.2326,556 "loss_nan_ranks": 0,557 "loss_rank_avg": 0.22178535163402557,558 "step": 250,559 "valid_targets_mean": 5966.1,560 "valid_targets_min": 1724561 },562 {563 "epoch": 0.40476190476190477,564 "grad_norm": 0.4750302614876995,565 "learning_rate": 2.3038548752834472e-05,566 "loss": 0.2402,567 "loss_nan_ranks": 0,568 "loss_rank_avg": 0.26613864302635193,569 "step": 255,570 "valid_targets_mean": 5677.9,571 "valid_targets_min": 1011572 },573 {574 "epoch": 0.4126984126984127,575 "grad_norm": 0.4468674852795431,576 "learning_rate": 2.3492063492063496e-05,577 "loss": 0.2425,578 "loss_nan_ranks": 0,579 "loss_rank_avg": 0.2535470724105835,580 "step": 260,581 "valid_targets_mean": 5571.2,582 "valid_targets_min": 3116583 },584 {585 "epoch": 0.42063492063492064,586 "grad_norm": 0.4435957536640986,587 "learning_rate": 2.394557823129252e-05,588 "loss": 0.2403,589 "loss_nan_ranks": 0,590 "loss_rank_avg": 0.24253222346305847,591 "step": 265,592 "valid_targets_mean": 5520.6,593 "valid_targets_min": 2481594 },595 {596 "epoch": 0.42857142857142855,597 "grad_norm": 0.48192905697263166,598 "learning_rate": 2.439909297052154e-05,599 "loss": 0.2405,600 "loss_nan_ranks": 0,601 "loss_rank_avg": 0.22654183208942413,602 "step": 270,603 "valid_targets_mean": 5040.6,604 "valid_targets_min": 1020605 },606 {607 "epoch": 0.4365079365079365,608 "grad_norm": 0.4654909902022295,609 "learning_rate": 2.4852607709750566e-05,610 "loss": 0.2397,611 "loss_nan_ranks": 0,612 "loss_rank_avg": 0.22321158647537231,613 "step": 275,614 "valid_targets_mean": 4938.6,615 "valid_targets_min": 554616 },617 {618 "epoch": 0.4444444444444444,619 "grad_norm": 0.47136311558318184,620 "learning_rate": 2.5306122448979597e-05,621 "loss": 0.256,622 "loss_nan_ranks": 0,623 "loss_rank_avg": 0.25398164987564087,624 "step": 280,625 "valid_targets_mean": 5719.1,626 "valid_targets_min": 1427627 },628 {629 "epoch": 0.4523809523809524,630 "grad_norm": 0.5143544056685512,631 "learning_rate": 2.5759637188208618e-05,632 "loss": 0.2538,633 "loss_nan_ranks": 0,634 "loss_rank_avg": 0.28001242876052856,635 "step": 285,636 "valid_targets_mean": 4379.2,637 "valid_targets_min": 275638 },639 {640 "epoch": 0.4603174603174603,641 "grad_norm": 0.40992433170257936,642 "learning_rate": 2.6213151927437642e-05,643 "loss": 0.2107,644 "loss_nan_ranks": 0,645 "loss_rank_avg": 0.19944259524345398,646 "step": 290,647 "valid_targets_mean": 6113.2,648 "valid_targets_min": 1496649 },650 {651 "epoch": 0.46825396825396826,652 "grad_norm": 0.7108778765163672,653 "learning_rate": 2.6666666666666667e-05,654 "loss": 0.2316,655 "loss_nan_ranks": 0,656 "loss_rank_avg": 0.2676636576652527,657 "step": 295,658 "valid_targets_mean": 4878.0,659 "valid_targets_min": 1397660 },661 {662 "epoch": 0.47619047619047616,663 "grad_norm": 0.7424501325702221,664 "learning_rate": 2.7120181405895694e-05,665 "loss": 0.2345,666 "loss_nan_ranks": 0,667 "loss_rank_avg": 0.27534204721450806,668 "step": 300,669 "valid_targets_mean": 4737.1,670 "valid_targets_min": 1195671 },672 {673 "epoch": 0.48412698412698413,674 "grad_norm": 0.4554032952042926,675 "learning_rate": 2.757369614512472e-05,676 "loss": 0.2423,677 "loss_nan_ranks": 0,678 "loss_rank_avg": 0.23061871528625488,679 "step": 305,680 "valid_targets_mean": 5816.9,681 "valid_targets_min": 935682 },683 {684 "epoch": 0.49206349206349204,685 "grad_norm": 0.43918142641279273,686 "learning_rate": 2.8027210884353743e-05,687 "loss": 0.246,688 "loss_nan_ranks": 0,689 "loss_rank_avg": 0.2266923487186432,690 "step": 310,691 "valid_targets_mean": 5512.4,692 "valid_targets_min": 674693 },694 {695 "epoch": 0.5,696 "grad_norm": 0.46398515425879433,697 "learning_rate": 2.8480725623582767e-05,698 "loss": 0.2234,699 "loss_nan_ranks": 0,700 "loss_rank_avg": 0.21582689881324768,701 "step": 315,702 "valid_targets_mean": 5874.5,703 "valid_targets_min": 1816704 },705 {706 "epoch": 0.5079365079365079,707 "grad_norm": 0.4043326688329922,708 "learning_rate": 2.893424036281179e-05,709 "loss": 0.2239,710 "loss_nan_ranks": 0,711 "loss_rank_avg": 0.21339790523052216,712 "step": 320,713 "valid_targets_mean": 5820.1,714 "valid_targets_min": 282715 },716 {717 "epoch": 0.5158730158730159,718 "grad_norm": 0.502316882086023,719 "learning_rate": 2.938775510204082e-05,720 "loss": 0.2237,721 "loss_nan_ranks": 0,722 "loss_rank_avg": 0.20219871401786804,723 "step": 325,724 "valid_targets_mean": 5578.6,725 "valid_targets_min": 1540726 },727 {728 "epoch": 0.5238095238095238,729 "grad_norm": 0.9835515697061404,730 "learning_rate": 2.9841269841269844e-05,731 "loss": 0.2323,732 "loss_nan_ranks": 0,733 "loss_rank_avg": 0.2298191487789154,734 "step": 330,735 "valid_targets_mean": 4911.0,736 "valid_targets_min": 632737 },738 {739 "epoch": 0.5317460317460317,740 "grad_norm": 0.4647735663644707,741 "learning_rate": 3.0294784580498868e-05,742 "loss": 0.2339,743 "loss_nan_ranks": 0,744 "loss_rank_avg": 0.22781239449977875,745 "step": 335,746 "valid_targets_mean": 5532.8,747 "valid_targets_min": 1535748 },749 {750 "epoch": 0.5396825396825397,751 "grad_norm": 0.5438389697935898,752 "learning_rate": 3.074829931972789e-05,753 "loss": 0.2375,754 "loss_nan_ranks": 0,755 "loss_rank_avg": 0.22640562057495117,756 "step": 340,757 "valid_targets_mean": 4285.8,758 "valid_targets_min": 727759 },760 {761 "epoch": 0.5476190476190477,762 "grad_norm": 0.4334285919558895,763 "learning_rate": 3.1201814058956924e-05,764 "loss": 0.2399,765 "loss_nan_ranks": 0,766 "loss_rank_avg": 0.21913409233093262,767 "step": 345,768 "valid_targets_mean": 5222.7,769 "valid_targets_min": 285770 },771 {772 "epoch": 0.5555555555555556,773 "grad_norm": 0.4738405937337707,774 "learning_rate": 3.1655328798185945e-05,775 "loss": 0.2378,776 "loss_nan_ranks": 0,777 "loss_rank_avg": 0.25035279989242554,778 "step": 350,779 "valid_targets_mean": 5209.8,780 "valid_targets_min": 833781 },782 {783 "epoch": 0.5634920634920635,784 "grad_norm": 0.4555272170875938,785 "learning_rate": 3.2108843537414965e-05,786 "loss": 0.2223,787 "loss_nan_ranks": 0,788 "loss_rank_avg": 0.21295906603336334,789 "step": 355,790 "valid_targets_mean": 4887.6,791 "valid_targets_min": 845792 },793 {794 "epoch": 0.5714285714285714,795 "grad_norm": 0.4607956603180466,796 "learning_rate": 3.256235827664399e-05,797 "loss": 0.2242,798 "loss_nan_ranks": 0,799 "loss_rank_avg": 0.23422232270240784,800 "step": 360,801 "valid_targets_mean": 5982.9,802 "valid_targets_min": 3791803 },804 {805 "epoch": 0.5793650793650794,806 "grad_norm": 0.46019446827222377,807 "learning_rate": 3.3015873015873014e-05,808 "loss": 0.234,809 "loss_nan_ranks": 0,810 "loss_rank_avg": 0.21903209388256073,811 "step": 365,812 "valid_targets_mean": 5330.9,813 "valid_targets_min": 2257814 },815 {816 "epoch": 0.5873015873015873,817 "grad_norm": 0.6042052076393369,818 "learning_rate": 3.346938775510204e-05,819 "loss": 0.2358,820 "loss_nan_ranks": 0,821 "loss_rank_avg": 0.2550661861896515,822 "step": 370,823 "valid_targets_mean": 4556.8,824 "valid_targets_min": 763825 },826 {827 "epoch": 0.5952380952380952,828 "grad_norm": 0.44853135001478606,829 "learning_rate": 3.392290249433107e-05,830 "loss": 0.2178,831 "loss_nan_ranks": 0,832 "loss_rank_avg": 0.202229842543602,833 "step": 375,834 "valid_targets_mean": 5618.4,835 "valid_targets_min": 287836 },837 {838 "epoch": 0.6031746031746031,839 "grad_norm": 0.4114084841092905,840 "learning_rate": 3.437641723356009e-05,841 "loss": 0.2336,842 "loss_nan_ranks": 0,843 "loss_rank_avg": 0.208021879196167,844 "step": 380,845 "valid_targets_mean": 6323.6,846 "valid_targets_min": 3070847 },848 {849 "epoch": 0.6111111111111112,850 "grad_norm": 0.4907016504547875,851 "learning_rate": 3.482993197278912e-05,852 "loss": 0.2096,853 "loss_nan_ranks": 0,854 "loss_rank_avg": 0.2262699007987976,855 "step": 385,856 "valid_targets_mean": 5458.8,857 "valid_targets_min": 1233858 },859 {860 "epoch": 0.6190476190476191,861 "grad_norm": 0.4470894371245897,862 "learning_rate": 3.5283446712018146e-05,863 "loss": 0.2147,864 "loss_nan_ranks": 0,865 "loss_rank_avg": 0.2284412682056427,866 "step": 390,867 "valid_targets_mean": 5275.5,868 "valid_targets_min": 614869 },870 {871 "epoch": 0.626984126984127,872 "grad_norm": 0.8292394623205982,873 "learning_rate": 3.573696145124717e-05,874 "loss": 0.2177,875 "loss_nan_ranks": 0,876 "loss_rank_avg": 0.23321247100830078,877 "step": 395,878 "valid_targets_mean": 4546.7,879 "valid_targets_min": 1120880 },881 {882 "epoch": 0.6349206349206349,883 "grad_norm": 0.5101519952218712,884 "learning_rate": 3.6190476190476195e-05,885 "loss": 0.2029,886 "loss_nan_ranks": 0,887 "loss_rank_avg": 0.22552572190761566,888 "step": 400,889 "valid_targets_mean": 4713.0,890 "valid_targets_min": 403891 },892 {893 "epoch": 0.6428571428571429,894 "grad_norm": 0.4782175383328679,895 "learning_rate": 3.6643990929705216e-05,896 "loss": 0.2336,897 "loss_nan_ranks": 0,898 "loss_rank_avg": 0.22066287696361542,899 "step": 405,900 "valid_targets_mean": 5557.4,901 "valid_targets_min": 1431902 },903 {904 "epoch": 0.6507936507936508,905 "grad_norm": 0.6813010668498629,906 "learning_rate": 3.7097505668934243e-05,907 "loss": 0.2304,908 "loss_nan_ranks": 0,909 "loss_rank_avg": 0.23460529744625092,910 "step": 410,911 "valid_targets_mean": 4301.2,912 "valid_targets_min": 290913 },914 {915 "epoch": 0.6587301587301587,916 "grad_norm": 0.42762135214795316,917 "learning_rate": 3.755102040816327e-05,918 "loss": 0.2218,919 "loss_nan_ranks": 0,920 "loss_rank_avg": 0.22290262579917908,921 "step": 415,922 "valid_targets_mean": 5696.5,923 "valid_targets_min": 1242924 },925 {926 "epoch": 0.6666666666666666,927 "grad_norm": 0.7113003892096883,928 "learning_rate": 3.800453514739229e-05,929 "loss": 0.2307,930 "loss_nan_ranks": 0,931 "loss_rank_avg": 0.24848781526088715,932 "step": 420,933 "valid_targets_mean": 5861.1,934 "valid_targets_min": 1829935 },936 {937 "epoch": 0.6746031746031746,938 "grad_norm": 0.4223786613285171,939 "learning_rate": 3.845804988662132e-05,940 "loss": 0.2361,941 "loss_nan_ranks": 0,942 "loss_rank_avg": 0.23021358251571655,943 "step": 425,944 "valid_targets_mean": 5603.6,945 "valid_targets_min": 2005946 },947 {948 "epoch": 0.6825396825396826,949 "grad_norm": 0.46482319003042916,950 "learning_rate": 3.891156462585034e-05,951 "loss": 0.2369,952 "loss_nan_ranks": 0,953 "loss_rank_avg": 0.2257668375968933,954 "step": 430,955 "valid_targets_mean": 5665.9,956 "valid_targets_min": 936957 },958 {959 "epoch": 0.6904761904761905,960 "grad_norm": 0.49517195170111594,961 "learning_rate": 3.936507936507937e-05,962 "loss": 0.2219,963 "loss_nan_ranks": 0,964 "loss_rank_avg": 0.24368128180503845,965 "step": 435,966 "valid_targets_mean": 5276.9,967 "valid_targets_min": 965968 },969 {970 "epoch": 0.6984126984126984,971 "grad_norm": 0.4533449896226927,972 "learning_rate": 3.9818594104308396e-05,973 "loss": 0.2227,974 "loss_nan_ranks": 0,975 "loss_rank_avg": 0.2182699739933014,976 "step": 440,977 "valid_targets_mean": 5190.2,978 "valid_targets_min": 609979 },980 {981 "epoch": 0.7063492063492064,982 "grad_norm": 0.47604972418524916,983 "learning_rate": 3.999994361288785e-05,984 "loss": 0.2316,985 "loss_nan_ranks": 0,986 "loss_rank_avg": 0.27464592456817627,987 "step": 445,988 "valid_targets_mean": 5377.2,989 "valid_targets_min": 299990 },991 {992 "epoch": 0.7142857142857143,993 "grad_norm": 0.4843309348827549,994 "learning_rate": 3.9999599026131644e-05,995 "loss": 0.2078,996 "loss_nan_ranks": 0,997 "loss_rank_avg": 0.21664546430110931,998 "step": 450,999 "valid_targets_mean": 5402.9,1000 "valid_targets_min": 13591001 },1002 {1003 "epoch": 0.7222222222222222,1004 "grad_norm": 0.4382635205661119,1005 "learning_rate": 3.999894118418342e-05,1006 "loss": 0.2176,1007 "loss_nan_ranks": 0,1008 "loss_rank_avg": 0.22355255484580994,1009 "step": 455,1010 "valid_targets_mean": 5052.1,1011 "valid_targets_min": 20691012 },1013 {1014 "epoch": 0.7301587301587301,1015 "grad_norm": 0.472980100407786,1016 "learning_rate": 3.999797009734697e-05,1017 "loss": 0.2286,1018 "loss_nan_ranks": 0,1019 "loss_rank_avg": 0.22827665507793427,1020 "step": 460,1021 "valid_targets_mean": 5404.2,1022 "valid_targets_min": 17441023 },1024 {1025 "epoch": 0.7380952380952381,1026 "grad_norm": 0.43558241143554155,1027 "learning_rate": 3.999668578083253e-05,1028 "loss": 0.2344,1029 "loss_nan_ranks": 0,1030 "loss_rank_avg": 0.22950908541679382,1031 "step": 465,1032 "valid_targets_mean": 5786.8,1033 "valid_targets_min": 6881034 },1035 {1036 "epoch": 0.746031746031746,1037 "grad_norm": 0.39301407866116134,1038 "learning_rate": 3.9995088254756434e-05,1039 "loss": 0.2171,1040 "loss_nan_ranks": 0,1041 "loss_rank_avg": 0.20144161581993103,1042 "step": 470,1043 "valid_targets_mean": 6067.9,1044 "valid_targets_min": 34771045 },1046 {1047 "epoch": 0.753968253968254,1048 "grad_norm": 0.4335866805247727,1049 "learning_rate": 3.999317754414084e-05,1050 "loss": 0.2162,1051 "loss_nan_ranks": 0,1052 "loss_rank_avg": 0.24468550086021423,1053 "step": 475,1054 "valid_targets_mean": 5383.6,1055 "valid_targets_min": 2971056 },1057 {1058 "epoch": 0.7619047619047619,1059 "grad_norm": 0.43530076929025974,1060 "learning_rate": 3.999095367891337e-05,1061 "loss": 0.2064,1062 "loss_nan_ranks": 0,1063 "loss_rank_avg": 0.2087598741054535,1064 "step": 480,1065 "valid_targets_mean": 5412.7,1066 "valid_targets_min": 2411067 },1068 {1069 "epoch": 0.7698412698412699,1070 "grad_norm": 0.42581385510950404,1071 "learning_rate": 3.9988416693906563e-05,1072 "loss": 0.2149,1073 "loss_nan_ranks": 0,1074 "loss_rank_avg": 0.20418357849121094,1075 "step": 485,1076 "valid_targets_mean": 5400.4,1077 "valid_targets_min": 6181078 },1079 {1080 "epoch": 0.7777777777777778,1081 "grad_norm": 0.6201401443531581,1082 "learning_rate": 3.9985566628857425e-05,1083 "loss": 0.2286,1084 "loss_nan_ranks": 0,1085 "loss_rank_avg": 0.22904717922210693,1086 "step": 490,1087 "valid_targets_mean": 5283.9,1088 "valid_targets_min": 16411089 },1090 {1091 "epoch": 0.7857142857142857,1092 "grad_norm": 0.40573600003328514,1093 "learning_rate": 3.998240352840672e-05,1094 "loss": 0.2369,1095 "loss_nan_ranks": 0,1096 "loss_rank_avg": 0.19750481843948364,1097 "step": 495,1098 "valid_targets_mean": 5820.5,1099 "valid_targets_min": 25271100 },1101 {1102 "epoch": 0.7936507936507936,1103 "grad_norm": 0.6282176812699868,1104 "learning_rate": 3.997892744209833e-05,1105 "loss": 0.2157,1106 "loss_nan_ranks": 0,1107 "loss_rank_avg": 0.21052823960781097,1108 "step": 500,1109 "valid_targets_mean": 3286.4,1110 "valid_targets_min": 2841111 },1112 {1113 "epoch": 0.8015873015873016,1114 "grad_norm": 0.4150664416041001,1115 "learning_rate": 3.997513842437845e-05,1116 "loss": 0.2168,1117 "loss_nan_ranks": 0,1118 "loss_rank_avg": 0.20114190876483917,1119 "step": 505,1120 "valid_targets_mean": 6746.6,1121 "valid_targets_min": 35031122 },1123 {1124 "epoch": 0.8095238095238095,1125 "grad_norm": 0.5580131451084269,1126 "learning_rate": 3.997103653459475e-05,1127 "loss": 0.2255,1128 "loss_nan_ranks": 0,1129 "loss_rank_avg": 0.21351292729377747,1130 "step": 510,1131 "valid_targets_mean": 5427.2,1132 "valid_targets_min": 13101133 },1134 {1135 "epoch": 0.8174603174603174,1136 "grad_norm": 0.4539249289508295,1137 "learning_rate": 3.996662183699541e-05,1138 "loss": 0.2029,1139 "loss_nan_ranks": 0,1140 "loss_rank_avg": 0.22894957661628723,1141 "step": 515,1142 "valid_targets_mean": 6587.3,1143 "valid_targets_min": 29871144 },1145 {1146 "epoch": 0.8253968253968254,1147 "grad_norm": 0.4606135797915741,1148 "learning_rate": 3.996189440072818e-05,1149 "loss": 0.2235,1150 "loss_nan_ranks": 0,1151 "loss_rank_avg": 0.23994050920009613,1152 "step": 520,1153 "valid_targets_mean": 5603.4,1154 "valid_targets_min": 9631155 },1156 {1157 "epoch": 0.8333333333333334,1158 "grad_norm": 0.46485471964434133,1159 "learning_rate": 3.9956854299839246e-05,1160 "loss": 0.2399,1161 "loss_nan_ranks": 0,1162 "loss_rank_avg": 0.24385607242584229,1163 "step": 525,1164 "valid_targets_mean": 5554.9,1165 "valid_targets_min": 19981166 },1167 {1168 "epoch": 0.8412698412698413,1169 "grad_norm": 0.5070540327503203,1170 "learning_rate": 3.9951501613272076e-05,1171 "loss": 0.2238,1172 "loss_nan_ranks": 0,1173 "loss_rank_avg": 0.2311379760503769,1174 "step": 530,1175 "valid_targets_mean": 5792.5,1176 "valid_targets_min": 29501177 },1178 {1179 "epoch": 0.8492063492063492,1180 "grad_norm": 0.4509301237412979,1181 "learning_rate": 3.994583642486618e-05,1182 "loss": 0.2155,1183 "loss_nan_ranks": 0,1184 "loss_rank_avg": 0.22014960646629333,1185 "step": 535,1186 "valid_targets_mean": 5551.4,1187 "valid_targets_min": 24541188 },1189 {1190 "epoch": 0.8571428571428571,1191 "grad_norm": 0.37450581843880576,1192 "learning_rate": 3.993985882335584e-05,1193 "loss": 0.2102,1194 "loss_nan_ranks": 0,1195 "loss_rank_avg": 0.17562690377235413,1196 "step": 540,1197 "valid_targets_mean": 6399.9,1198 "valid_targets_min": 8651199 },1200 {