osunlp/WebJudge-7B
9629
1{2 "best_global_step": null,3 "best_metric": null,4 "best_model_checkpoint": null,5 "epoch": 4.972522897585345,6 "eval_steps": 500,7 "global_step": 750,8 "is_hyper_param_search": false,9 "is_local_process_zero": true,10 "is_world_process_zero": true,11 "log_history": [12 {13 "epoch": 0.013322231473771857,14 "grad_norm": 5.780612066318371,15 "learning_rate": 1.3333333333333336e-07,16 "loss": 0.9209,17 "step": 218 },19 {20 "epoch": 0.026644462947543714,21 "grad_norm": 5.803797516781435,22 "learning_rate": 2.666666666666667e-07,23 "loss": 0.9065,24 "step": 425 },26 {27 "epoch": 0.03996669442131557,28 "grad_norm": 5.67259689729213,29 "learning_rate": 4.0000000000000003e-07,30 "loss": 0.9182,31 "step": 632 },33 {34 "epoch": 0.05328892589508743,35 "grad_norm": 5.354376269147714,36 "learning_rate": 5.333333333333335e-07,37 "loss": 0.9233,38 "step": 839 },40 {41 "epoch": 0.06661115736885928,42 "grad_norm": 4.851036441351623,43 "learning_rate": 6.666666666666667e-07,44 "loss": 0.8906,45 "step": 1046 },47 {48 "epoch": 0.07993338884263114,49 "grad_norm": 4.566190641845786,50 "learning_rate": 8.000000000000001e-07,51 "loss": 0.8719,52 "step": 1253 },54 {55 "epoch": 0.093255620316403,56 "grad_norm": 4.013324466824848,57 "learning_rate": 9.333333333333334e-07,58 "loss": 0.8861,59 "step": 1460 },61 {62 "epoch": 0.10657785179017486,63 "grad_norm": 3.540718772081854,64 "learning_rate": 1.066666666666667e-06,65 "loss": 0.8791,66 "step": 1667 },68 {69 "epoch": 0.11990008326394672,70 "grad_norm": 2.88258712818169,71 "learning_rate": 1.2000000000000002e-06,72 "loss": 0.8532,73 "step": 1874 },75 {76 "epoch": 0.13322231473771856,77 "grad_norm": 2.6937271254013613,78 "learning_rate": 1.3333333333333334e-06,79 "loss": 0.8455,80 "step": 2081 },82 {83 "epoch": 0.14654454621149043,84 "grad_norm": 2.3220586555688816,85 "learning_rate": 1.4666666666666669e-06,86 "loss": 0.8159,87 "step": 2288 },89 {90 "epoch": 0.15986677768526228,91 "grad_norm": 3.400225929248788,92 "learning_rate": 1.6000000000000001e-06,93 "loss": 0.8004,94 "step": 2495 },96 {97 "epoch": 0.17318900915903415,98 "grad_norm": 3.5504999091076845,99 "learning_rate": 1.7333333333333336e-06,100 "loss": 0.8084,101 "step": 26102 },103 {104 "epoch": 0.186511240632806,105 "grad_norm": 3.2668071978806057,106 "learning_rate": 1.8666666666666669e-06,107 "loss": 0.8086,108 "step": 28109 },110 {111 "epoch": 0.19983347210657784,112 "grad_norm": 2.4897231787044154,113 "learning_rate": 2.0000000000000003e-06,114 "loss": 0.7821,115 "step": 30116 },117 {118 "epoch": 0.21315570358034971,119 "grad_norm": 2.006655051652645,120 "learning_rate": 2.133333333333334e-06,121 "loss": 0.7767,122 "step": 32123 },124 {125 "epoch": 0.22647793505412156,126 "grad_norm": 1.8054118845570075,127 "learning_rate": 2.266666666666667e-06,128 "loss": 0.764,129 "step": 34130 },131 {132 "epoch": 0.23980016652789343,133 "grad_norm": 2.060437415012226,134 "learning_rate": 2.4000000000000003e-06,135 "loss": 0.7577,136 "step": 36137 },138 {139 "epoch": 0.2531223980016653,140 "grad_norm": 1.822445369621608,141 "learning_rate": 2.5333333333333338e-06,142 "loss": 0.7441,143 "step": 38144 },145 {146 "epoch": 0.2664446294754371,147 "grad_norm": 1.702085750109042,148 "learning_rate": 2.666666666666667e-06,149 "loss": 0.7364,150 "step": 40151 },152 {153 "epoch": 0.279766860949209,154 "grad_norm": 1.4905140822613037,155 "learning_rate": 2.8000000000000003e-06,156 "loss": 0.7252,157 "step": 42158 },159 {160 "epoch": 0.29308909242298087,161 "grad_norm": 1.602012507677594,162 "learning_rate": 2.9333333333333338e-06,163 "loss": 0.727,164 "step": 44165 },166 {167 "epoch": 0.3064113238967527,168 "grad_norm": 1.5462649172094083,169 "learning_rate": 3.066666666666667e-06,170 "loss": 0.7221,171 "step": 46172 },173 {174 "epoch": 0.31973355537052456,175 "grad_norm": 1.4593506403426082,176 "learning_rate": 3.2000000000000003e-06,177 "loss": 0.6922,178 "step": 48179 },180 {181 "epoch": 0.33305578684429643,182 "grad_norm": 1.42990322544597,183 "learning_rate": 3.3333333333333333e-06,184 "loss": 0.6873,185 "step": 50186 },187 {188 "epoch": 0.3463780183180683,189 "grad_norm": 1.4803196580352989,190 "learning_rate": 3.4666666666666672e-06,191 "loss": 0.6975,192 "step": 52193 },194 {195 "epoch": 0.3597002497918401,196 "grad_norm": 1.5152423594984497,197 "learning_rate": 3.6000000000000003e-06,198 "loss": 0.662,199 "step": 54200 },201 {202 "epoch": 0.373022481265612,203 "grad_norm": 1.4183767379176757,204 "learning_rate": 3.7333333333333337e-06,205 "loss": 0.6578,206 "step": 56207 },208 {209 "epoch": 0.38634471273938387,210 "grad_norm": 1.5361891229630817,211 "learning_rate": 3.866666666666667e-06,212 "loss": 0.6655,213 "step": 58214 },215 {216 "epoch": 0.3996669442131557,217 "grad_norm": 1.430106096852744,218 "learning_rate": 4.000000000000001e-06,219 "loss": 0.648,220 "step": 60221 },222 {223 "epoch": 0.41298917568692756,224 "grad_norm": 1.4201912298932542,225 "learning_rate": 4.133333333333333e-06,226 "loss": 0.6449,227 "step": 62228 },229 {230 "epoch": 0.42631140716069943,231 "grad_norm": 1.45821665687646,232 "learning_rate": 4.266666666666668e-06,233 "loss": 0.636,234 "step": 64235 },236 {237 "epoch": 0.43963363863447125,238 "grad_norm": 1.351061303304786,239 "learning_rate": 4.4e-06,240 "loss": 0.6287,241 "step": 66242 },243 {244 "epoch": 0.4529558701082431,245 "grad_norm": 1.4213406831821087,246 "learning_rate": 4.533333333333334e-06,247 "loss": 0.6162,248 "step": 68249 },250 {251 "epoch": 0.466278101582015,252 "grad_norm": 1.5920606382997864,253 "learning_rate": 4.666666666666667e-06,254 "loss": 0.6269,255 "step": 70256 },257 {258 "epoch": 0.47960033305578686,259 "grad_norm": 1.4259278183625448,260 "learning_rate": 4.800000000000001e-06,261 "loss": 0.628,262 "step": 72263 },264 {265 "epoch": 0.4929225645295587,266 "grad_norm": 1.4277383814389801,267 "learning_rate": 4.933333333333334e-06,268 "loss": 0.6047,269 "step": 74270 },271 {272 "epoch": 0.5062447960033306,273 "grad_norm": 1.4056841775145905,274 "learning_rate": 4.999972922944898e-06,275 "loss": 0.5984,276 "step": 76277 },278 {279 "epoch": 0.5195670274771024,280 "grad_norm": 1.3201694715565466,281 "learning_rate": 4.999756310023261e-06,282 "loss": 0.5954,283 "step": 78284 },285 {286 "epoch": 0.5328892589508742,287 "grad_norm": 1.3231909628376382,288 "learning_rate": 4.999323102948655e-06,289 "loss": 0.5954,290 "step": 80291 },292 {293 "epoch": 0.5462114904246461,294 "grad_norm": 1.3735284931415068,295 "learning_rate": 4.998673339256785e-06,296 "loss": 0.5744,297 "step": 82298 },299 {300 "epoch": 0.559533721898418,301 "grad_norm": 1.4504463513541146,302 "learning_rate": 4.997807075247147e-06,303 "loss": 0.593,304 "step": 84305 },306 {307 "epoch": 0.5728559533721899,308 "grad_norm": 1.2740174759395542,309 "learning_rate": 4.996724385978142e-06,310 "loss": 0.5903,311 "step": 86312 },313 {314 "epoch": 0.5861781848459617,315 "grad_norm": 1.3594170816449038,316 "learning_rate": 4.995425365260585e-06,317 "loss": 0.5748,318 "step": 88319 },320 {321 "epoch": 0.5995004163197336,322 "grad_norm": 1.3782163690261147,323 "learning_rate": 4.993910125649561e-06,324 "loss": 0.5814,325 "step": 90326 },327 {328 "epoch": 0.6128226477935054,329 "grad_norm": 1.2584130530987572,330 "learning_rate": 4.992178798434684e-06,331 "loss": 0.5752,332 "step": 92333 },334 {335 "epoch": 0.6261448792672772,336 "grad_norm": 1.4216349386698004,337 "learning_rate": 4.990231533628719e-06,338 "loss": 0.5757,339 "step": 94340 },341 {342 "epoch": 0.6394671107410491,343 "grad_norm": 1.3768977749050733,344 "learning_rate": 4.988068499954578e-06,345 "loss": 0.5555,346 "step": 96347 },348 {349 "epoch": 0.652789342214821,350 "grad_norm": 1.463407945745149,351 "learning_rate": 4.985689884830711e-06,352 "loss": 0.5591,353 "step": 98354 },355 {356 "epoch": 0.6661115736885929,357 "grad_norm": 1.3808427236512926,358 "learning_rate": 4.983095894354858e-06,359 "loss": 0.5588,360 "step": 100361 },362 {363 "epoch": 0.6794338051623647,364 "grad_norm": 1.482231013162315,365 "learning_rate": 4.980286753286196e-06,366 "loss": 0.5418,367 "step": 102368 },369 {370 "epoch": 0.6927560366361366,371 "grad_norm": 1.3778109634949367,372 "learning_rate": 4.97726270502586e-06,373 "loss": 0.5399,374 "step": 104375 },376 {377 "epoch": 0.7060782681099084,378 "grad_norm": 1.4002755485164502,379 "learning_rate": 4.974024011595864e-06,380 "loss": 0.5533,381 "step": 106382 },383 {384 "epoch": 0.7194004995836802,385 "grad_norm": 1.3296620938997752,386 "learning_rate": 4.970570953616383e-06,387 "loss": 0.5438,388 "step": 108389 },390 {391 "epoch": 0.7327227310574521,392 "grad_norm": 1.4458203791375825,393 "learning_rate": 4.966903830281449e-06,394 "loss": 0.5378,395 "step": 110396 },397 {398 "epoch": 0.746044962531224,399 "grad_norm": 1.5136526829998074,400 "learning_rate": 4.9630229593330226e-06,401 "loss": 0.5348,402 "step": 112403 },404 {405 "epoch": 0.7593671940049959,406 "grad_norm": 1.4362377777815807,407 "learning_rate": 4.958928677033465e-06,408 "loss": 0.5267,409 "step": 114410 },411 {412 "epoch": 0.7726894254787677,413 "grad_norm": 1.2730640176398647,414 "learning_rate": 4.954621338136399e-06,415 "loss": 0.5393,416 "step": 116417 },418 {419 "epoch": 0.7860116569525396,420 "grad_norm": 1.3685353603260022,421 "learning_rate": 4.95010131585597e-06,422 "loss": 0.534,423 "step": 118424 },425 {426 "epoch": 0.7993338884263114,427 "grad_norm": 1.2683696145515575,428 "learning_rate": 4.9453690018345144e-06,429 "loss": 0.527,430 "step": 120431 },432 {433 "epoch": 0.8126561199000832,434 "grad_norm": 1.323958192575613,435 "learning_rate": 4.940424806108619e-06,436 "loss": 0.5267,437 "step": 122438 },439 {440 "epoch": 0.8259783513738551,441 "grad_norm": 1.2428318596261736,442 "learning_rate": 4.935269157073597e-06,443 "loss": 0.5149,444 "step": 124445 },446 {447 "epoch": 0.839300582847627,448 "grad_norm": 1.27851729445364,449 "learning_rate": 4.9299025014463665e-06,450 "loss": 0.5228,451 "step": 126452 },453 {454 "epoch": 0.8526228143213989,455 "grad_norm": 1.2913119874277892,456 "learning_rate": 4.924325304226745e-06,457 "loss": 0.5028,458 "step": 128459 },460 {461 "epoch": 0.8659450457951707,462 "grad_norm": 1.3471089811240304,463 "learning_rate": 4.91853804865716e-06,464 "loss": 0.5402,465 "step": 130466 },467 {468 "epoch": 0.8792672772689425,469 "grad_norm": 1.3919989303105873,470 "learning_rate": 4.912541236180779e-06,471 "loss": 0.5208,472 "step": 132473 },474 {475 "epoch": 0.8925895087427144,476 "grad_norm": 1.336135856095439,477 "learning_rate": 4.9063353863980565e-06,478 "loss": 0.5232,479 "step": 134480 },481 {482 "epoch": 0.9059117402164862,483 "grad_norm": 1.535058182009125,484 "learning_rate": 4.899921037021719e-06,485 "loss": 0.5183,486 "step": 136487 },488 {489 "epoch": 0.9192339716902581,490 "grad_norm": 1.4366704774523757,491 "learning_rate": 4.893298743830168e-06,492 "loss": 0.5152,493 "step": 138494 },495 {496 "epoch": 0.93255620316403,497 "grad_norm": 1.4306647802429082,498 "learning_rate": 4.88646908061933e-06,499 "loss": 0.5241,500 "step": 140501 },502 {503 "epoch": 0.9458784346378019,504 "grad_norm": 1.3151003083587773,505 "learning_rate": 4.879432639152935e-06,506 "loss": 0.518,507 "step": 142508 },509 {510 "epoch": 0.9592006661115737,511 "grad_norm": 1.3682779135005043,512 "learning_rate": 4.8721900291112415e-06,513 "loss": 0.51,514 "step": 144515 },516 {517 "epoch": 0.9725228975853455,518 "grad_norm": 1.3896990341168534,519 "learning_rate": 4.864741878038218e-06,520 "loss": 0.5207,521 "step": 146522 },523 {524 "epoch": 0.9858451290591174,525 "grad_norm": 1.2929489978661655,526 "learning_rate": 4.857088831287158e-06,527 "loss": 0.5121,528 "step": 148529 },530 {531 "epoch": 0.9991673605328892,532 "grad_norm": 1.3614193317791738,533 "learning_rate": 4.849231551964771e-06,534 "loss": 0.5016,535 "step": 150536 },537 {538 "epoch": 1.0066611157368859,539 "grad_norm": 1.3229981405906006,540 "learning_rate": 4.841170720873723e-06,541 "loss": 0.2569,542 "step": 152543 },544 {545 "epoch": 1.0199833472106579,546 "grad_norm": 1.2274098346213043,547 "learning_rate": 4.832907036453647e-06,548 "loss": 0.4662,549 "step": 154550 },551 {552 "epoch": 1.0333055786844296,553 "grad_norm": 1.3810724651132364,554 "learning_rate": 4.824441214720629e-06,555 "loss": 0.4503,556 "step": 156557 },558 {559 "epoch": 1.0466278101582014,560 "grad_norm": 1.5094355408493076,561 "learning_rate": 4.815773989205165e-06,562 "loss": 0.4525,563 "step": 158564 },565 {566 "epoch": 1.0599500416319734,567 "grad_norm": 1.191750486186588,568 "learning_rate": 4.806906110888606e-06,569 "loss": 0.4548,570 "step": 160571 },572 {573 "epoch": 1.0732722731057451,574 "grad_norm": 1.2840884507072778,575 "learning_rate": 4.7978383481380865e-06,576 "loss": 0.4552,577 "step": 162578 },579 {580 "epoch": 1.0865945045795171,581 "grad_norm": 1.3818002604555029,582 "learning_rate": 4.788571486639948e-06,583 "loss": 0.452,584 "step": 164585 },586 {587 "epoch": 1.0999167360532889,588 "grad_norm": 1.3200111006279347,589 "learning_rate": 4.779106329331665e-06,590 "loss": 0.45,591 "step": 166592 },593 {594 "epoch": 1.1132389675270609,595 "grad_norm": 1.2755161939993753,596 "learning_rate": 4.769443696332272e-06,597 "loss": 0.4454,598 "step": 168599 },600 {601 "epoch": 1.1265611990008326,602 "grad_norm": 1.3421067926153882,603 "learning_rate": 4.759584424871302e-06,604 "loss": 0.4429,605 "step": 170606 },607 {608 "epoch": 1.1398834304746046,609 "grad_norm": 1.2219457405458125,610 "learning_rate": 4.749529369216246e-06,611 "loss": 0.4481,612 "step": 172613 },614 {615 "epoch": 1.1532056619483764,616 "grad_norm": 1.330574738869651,617 "learning_rate": 4.7392794005985324e-06,618 "loss": 0.4459,619 "step": 174620 },621 {622 "epoch": 1.1665278934221481,623 "grad_norm": 1.2042952174150132,624 "learning_rate": 4.7288354071380415e-06,625 "loss": 0.4339,626 "step": 176627 },628 {629 "epoch": 1.1798501248959201,630 "grad_norm": 1.2535265876319093,631 "learning_rate": 4.7181982937661485e-06,632 "loss": 0.4364,633 "step": 178634 },635 {636 "epoch": 1.1931723563696919,637 "grad_norm": 1.1967067502698956,638 "learning_rate": 4.707368982147318e-06,639 "loss": 0.4484,640 "step": 180641 },642 {643 "epoch": 1.2064945878434639,644 "grad_norm": 1.3022379327320546,645 "learning_rate": 4.696348410599244e-06,646 "loss": 0.4468,647 "step": 182648 },649 {650 "epoch": 1.2198168193172356,651 "grad_norm": 1.3137228151962215,652 "learning_rate": 4.685137534011549e-06,653 "loss": 0.4492,654 "step": 184655 },656 {657 "epoch": 1.2331390507910074,658 "grad_norm": 1.3650226627212705,659 "learning_rate": 4.673737323763048e-06,660 "loss": 0.4389,661 "step": 186662 },663 {664 "epoch": 1.2464612822647794,665 "grad_norm": 1.3122923570081069,666 "learning_rate": 4.662148767637578e-06,667 "loss": 0.4426,668 "step": 188669 },670 {671 "epoch": 1.2597835137385511,672 "grad_norm": 1.3191199275346543,673 "learning_rate": 4.650372869738415e-06,674 "loss": 0.434,675 "step": 190676 },677 {678 "epoch": 1.2731057452123231,679 "grad_norm": 1.4425884017899313,680 "learning_rate": 4.638410650401267e-06,681 "loss": 0.4382,682 "step": 192683 },684 {685 "epoch": 1.2864279766860949,686 "grad_norm": 1.4066578837011166,687 "learning_rate": 4.626263146105875e-06,688 "loss": 0.4473,689 "step": 194690 },691 {692 "epoch": 1.2997502081598669,693 "grad_norm": 1.4430824831613096,694 "learning_rate": 4.613931409386196e-06,695 "loss": 0.4488,696 "step": 196697 },698 {699 "epoch": 1.3130724396336386,700 "grad_norm": 1.2217740909502797,701 "learning_rate": 4.601416508739211e-06,702 "loss": 0.4395,703 "step": 198704 },705 {706 "epoch": 1.3263946711074106,707 "grad_norm": 1.474039226711776,708 "learning_rate": 4.588719528532342e-06,709 "loss": 0.4381,710 "step": 200711 },712 {713 "epoch": 1.3397169025811824,714 "grad_norm": 1.2503538717797444,715 "learning_rate": 4.575841568909494e-06,716 "loss": 0.4317,717 "step": 202718 },719 {720 "epoch": 1.3530391340549541,721 "grad_norm": 1.3172152085291207,722 "learning_rate": 4.562783745695738e-06,723 "loss": 0.4284,724 "step": 204725 },726 {727 "epoch": 1.3663613655287261,728 "grad_norm": 1.2950216606489513,729 "learning_rate": 4.549547190300622e-06,730 "loss": 0.4372,731 "step": 206732 },733 {734 "epoch": 1.3796835970024979,735 "grad_norm": 1.2065789326345406,736 "learning_rate": 4.536133049620143e-06,737 "loss": 0.4376,738 "step": 208739 },740 {741 "epoch": 1.3930058284762699,742 "grad_norm": 1.450309483143858,743 "learning_rate": 4.522542485937369e-06,744 "loss": 0.4368,745 "step": 210746 },747 {748 "epoch": 1.4063280599500416,749 "grad_norm": 1.2856432394840618,750 "learning_rate": 4.508776676821739e-06,751 "loss": 0.4359,752 "step": 212753 },754 {755 "epoch": 1.4196502914238134,756 "grad_norm": 1.303392410991855,757 "learning_rate": 4.494836815027022e-06,758 "loss": 0.437,759 "step": 214760 },761 {762 "epoch": 1.4329725228975854,763 "grad_norm": 1.2374383776516957,764 "learning_rate": 4.4807241083879774e-06,765 "loss": 0.4277,766 "step": 216767 },768 {769 "epoch": 1.4462947543713571,770 "grad_norm": 1.1895403487373037,771 "learning_rate": 4.466439779715696e-06,772 "loss": 0.4219,773 "step": 218774 },775 {776 "epoch": 1.4596169858451291,777 "grad_norm": 1.3959427610193165,778 "learning_rate": 4.451985066691649e-06,779 "loss": 0.4341,780 "step": 220781 },782 {783 "epoch": 1.4729392173189009,784 "grad_norm": 1.2421484766590198,785 "learning_rate": 4.437361221760449e-06,786 "loss": 0.4162,787 "step": 222788 },789 {790 "epoch": 1.4862614487926726,791 "grad_norm": 1.287463815955178,792 "learning_rate": 4.422569512021332e-06,793 "loss": 0.4282,794 "step": 224795 },796 {797 "epoch": 1.4995836802664446,798 "grad_norm": 1.4250139528752677,799 "learning_rate": 4.407611219118363e-06,800 "loss": 0.421,801 "step": 226802 },803 {804 "epoch": 1.5129059117402166,805 "grad_norm": 1.239295099017855,806 "learning_rate": 4.3924876391293915e-06,807 "loss": 0.427,808 "step": 228809 },810 {811 "epoch": 1.5262281432139884,812 "grad_norm": 1.3453909852418124,813 "learning_rate": 4.377200082453748e-06,814 "loss": 0.4357,815 "step": 230816 },817 {818 "epoch": 1.5395503746877601,819 "grad_norm": 1.2197270804139342,820 "learning_rate": 4.361749873698707e-06,821 "loss": 0.4101,822 "step": 232823 },824 {825 "epoch": 1.552872606161532,826 "grad_norm": 1.2833857194816787,827 "learning_rate": 4.346138351564711e-06,828 "loss": 0.424,829 "step": 234830 },831 {832 "epoch": 1.5661948376353039,833 "grad_norm": 1.2293200008377447,834 "learning_rate": 4.330366868729376e-06,835 "loss": 0.421,836 "step": 236837 },838 {839 "epoch": 1.5795170691090759,840 "grad_norm": 1.1926560926173428,841 "learning_rate": 4.3144367917302964e-06,842 "loss": 0.4142,843 "step": 238844 },845 {846 "epoch": 1.5928393005828476,847 "grad_norm": 1.1594494067766803,848 "learning_rate": 4.2983495008466285e-06,849 "loss": 0.4191,850 "step": 240851 },852 {853 "epoch": 1.6061615320566194,854 "grad_norm": 1.224729384745418,855 "learning_rate": 4.2821063899795015e-06,856 "loss": 0.4128,857 "step": 242858 },859 {860 "epoch": 1.6194837635303914,861 "grad_norm": 1.1481228567549495,862 "learning_rate": 4.265708866531238e-06,863 "loss": 0.4279,864 "step": 244865 },866 {867 "epoch": 1.6328059950041633,868 "grad_norm": 1.3467092580505746,869 "learning_rate": 4.249158351283414e-06,870 "loss": 0.4262,871 "step": 246872 },873 {874 "epoch": 1.646128226477935,875 "grad_norm": 1.2776898545321251,876 "learning_rate": 4.232456278273743e-06,877 "loss": 0.4314,878 "step": 248879 },880 {881 "epoch": 1.6594504579517069,882 "grad_norm": 1.2719662910087424,883 "learning_rate": 4.215604094671835e-06,884 "loss": 0.4108,885 "step": 250886 },887 {888 "epoch": 1.6727726894254786,889 "grad_norm": 1.1745562871590098,890 "learning_rate": 4.198603260653792e-06,891 "loss": 0.4165,892 "step": 252893 },894 {895 "epoch": 1.6860949208992506,896 "grad_norm": 1.2455715420366917,897 "learning_rate": 4.181455249275701e-06,898 "loss": 0.4079,899 "step": 254900 },901 {902 "epoch": 1.6994171523730226,903 "grad_norm": 1.3896213959063652,904 "learning_rate": 4.1641615463459926e-06,905 "loss": 0.417,906 "step": 256907 },908 {909 "epoch": 1.7127393838467944,910 "grad_norm": 1.2131393445621887,911 "learning_rate": 4.146723650296701e-06,912 "loss": 0.4116,913 "step": 258914 },915 {916 "epoch": 1.7260616153205661,917 "grad_norm": 1.2101597375627524,918 "learning_rate": 4.129143072053639e-06,919 "loss": 0.4169,920 "step": 260921 },922 {923 "epoch": 1.739383846794338,924 "grad_norm": 1.2983597203629458,925 "learning_rate": 4.111421334905468e-06,926 "loss": 0.4101,927 "step": 262928 },929 {930 "epoch": 1.7527060782681099,931 "grad_norm": 1.1756761204986788,932 "learning_rate": 4.093559974371725e-06,933 "loss": 0.4023,934 "step": 264935 },936 {937 "epoch": 1.7660283097418819,938 "grad_norm": 1.296750722093234,939 "learning_rate": 4.075560538069767e-06,940 "loss": 0.4037,941 "step": 266942 },943 {944 "epoch": 1.7793505412156536,945 "grad_norm": 1.2664686153860956,946 "learning_rate": 4.05742458558068e-06,947 "loss": 0.4005,948 "step": 268949 },950 {951 "epoch": 1.7926727726894254,952 "grad_norm": 1.3144115093925024,953 "learning_rate": 4.039153688314146e-06,954 "loss": 0.4123,955 "step": 270956 },957 {958 "epoch": 1.8059950041631974,959 "grad_norm": 1.177870994913812,960 "learning_rate": 4.020749429372286e-06,961 "loss": 0.4061,962 "step": 272963 },964 {965 "epoch": 1.8193172356369693,966 "grad_norm": 1.1211392036639862,967 "learning_rate": 4.002213403412492e-06,968 "loss": 0.4207,969 "step": 274970 },971 {972 "epoch": 1.832639467110741,973 "grad_norm": 1.1967335338983747,974 "learning_rate": 3.983547216509254e-06,975 "loss": 0.4037,976 "step": 276977 },978 {979 "epoch": 1.8459616985845129,980 "grad_norm": 1.163438902600854,981 "learning_rate": 3.964752486015001e-06,982 "loss": 0.3983,983 "step": 278984 },985 {986 "epoch": 1.8592839300582846,987 "grad_norm": 1.3897690758852341,988 "learning_rate": 3.945830840419966e-06,989 "loss": 0.406,990 "step": 280991 },992 {993 "epoch": 1.8726061615320566,994 "grad_norm": 1.2302319797016965,995 "learning_rate": 3.92678391921108e-06,996 "loss": 0.4102,997 "step": 282998 },999 {1000 "epoch": 1.8859283930058286,1001 "grad_norm": 1.2515743950418428,1002 "learning_rate": 3.907613372729916e-06,1003 "loss": 0.4121,1004 "step": 2841005 },1006 {1007 "epoch": 1.8992506244796004,1008 "grad_norm": 1.2250514633864378,1009 "learning_rate": 3.888320862029699e-06,1010 "loss": 0.4135,1011 "step": 2861012 },1013 {1014 "epoch": 1.9125728559533721,1015 "grad_norm": 1.1786595929578796,1016 "learning_rate": 3.868908058731376e-06,1017 "loss": 0.3961,1018 "step": 2881019 },1020 {1021 "epoch": 1.9258950874271439,1022 "grad_norm": 1.2316483388259516,1023 "learning_rate": 3.849376644878783e-06,1024 "loss": 0.3991,1025 "step": 2901026 },1027 {1028 "epoch": 1.9392173189009159,1029 "grad_norm": 1.2218522002215788,1030 "learning_rate": 3.829728312792895e-06,1031 "loss": 0.4068,1032 "step": 2921033 },1034 {1035 "epoch": 1.9525395503746878,1036 "grad_norm": 1.218981908305007,1037 "learning_rate": 3.8099647649251984e-06,1038 "loss": 0.4116,1039 "step": 2941040 },1041 {1042 "epoch": 1.9658617818484596,1043 "grad_norm": 1.1473329397682062,1044 "learning_rate": 3.790087713710179e-06,1045 "loss": 0.3961,1046 "step": 2961047 },1048 {1049 "epoch": 1.9791840133222314,1050 "grad_norm": 1.15330486401059,1051 "learning_rate": 3.770098881416945e-06,1052 "loss": 0.397,1053 "step": 2981054 },1055 {1056 "epoch": 1.9925062447960034,1057 "grad_norm": 1.1147439818886564,1058 "learning_rate": 3.7500000000000005e-06,1059 "loss": 0.391,1060 "step": 3001061 },1062 {1063 "epoch": 2.0,1064 "grad_norm": 1.1888727583821848,1065 "learning_rate": 3.7297928109491765e-06,1066 "loss": 0.2238,1067 "step": 3021068 },1069 {1070 "epoch": 2.0133222314737718,1071 "grad_norm": 1.1682600742115117,1072 "learning_rate": 3.7094790651387414e-06,1073 "loss": 0.3464,1074 "step": 3041075 },1076 {1077 "epoch": 2.0266444629475435,1078 "grad_norm": 1.2543709475465634,1079 "learning_rate": 3.689060522675689e-06,1080 "loss": 0.3299,1081 "step": 3061082 },1083 {1084 "epoch": 2.0399666944213157,1085 "grad_norm": 1.209782299511866,1086 "learning_rate": 3.668538952747236e-06,1087 "loss": 0.3335,1088 "step": 3081089 },1090 {1091 "epoch": 2.0532889258950875,1092 "grad_norm": 1.2314074580378418,1093 "learning_rate": 3.6479161334675294e-06,1094 "loss": 0.3402,1095 "step": 3101096 },1097 {1098 "epoch": 2.0666111573688593,1099 "grad_norm": 1.089978871118908,1100 "learning_rate": 3.627193851723577e-06,1101 "loss": 0.3282,1102 "step": 3121103 },1104 {1105 "epoch": 2.079933388842631,1106 "grad_norm": 1.1440650029159125,1107 "learning_rate": 3.6063739030204226e-06,1108 "loss": 0.3353,1109 "step": 3141110 },1111 {1112 "epoch": 2.0932556203164028,1113 "grad_norm": 1.1412527172991913,1114 "learning_rate": 3.5854580913255706e-06,1115 "loss": 0.3377,1116 "step": 3161117 },1118 {1119 "epoch": 2.106577851790175,1120 "grad_norm": 1.1374336855151732,1121 "learning_rate": 3.564448228912682e-06,1122 "loss": 0.3303,1123 "step": 3181124 },1125 {1126 "epoch": 2.1199000832639467,1127 "grad_norm": 1.1689768541112975,1128 "learning_rate": 3.543346136204545e-06,1129 "loss": 0.3269,1130 "step": 3201131 },1132 {1133 "epoch": 2.1332223147377185,1134 "grad_norm": 1.1613635619803697,1135 "learning_rate": 3.522153641615345e-06,1136 "loss": 0.3447,1137 "step": 3221138 },1139 {1140 "epoch": 2.1465445462114903,1141 "grad_norm": 1.0764748235217316,1142 "learning_rate": 3.5008725813922383e-06,1143 "loss": 0.3347,1144 "step": 3241145 },1146 {1147 "epoch": 2.1598667776852625,1148 "grad_norm": 1.242351223908071,1149 "learning_rate": 3.4795047994562463e-06,1150 "loss": 0.3337,1151 "step": 3261152 },1153 {1154 "epoch": 2.1731890091590342,1155 "grad_norm": 1.1068446291466676,1156 "learning_rate": 3.458052147242494e-06,1157 "loss": 0.3411,1158 "step": 3281159 },1160 {1161 "epoch": 2.186511240632806,1162 "grad_norm": 1.16808964966109,1163 "learning_rate": 3.436516483539781e-06,1164 "loss": 0.3376,1165 "step": 3301166 },1167 {1168 "epoch": 2.1998334721065778,1169 "grad_norm": 1.1025319948129593,1170 "learning_rate": 3.4148996743295305e-06,1171 "loss": 0.3316,1172 "step": 3321173 },1174 {1175 "epoch": 2.2131557035803495,1176 "grad_norm": 1.1758686501416102,1177 "learning_rate": 3.3932035926241103e-06,1178 "loss": 0.3355,1179 "step": 3341180 },1181 {1182 "epoch": 2.2264779350541217,1183 "grad_norm": 1.1003768444337116,1184 "learning_rate": 3.3714301183045382e-06,1185 "loss": 0.3357,1186 "step": 3361187 },1188 {1189 "epoch": 2.2398001665278935,1190 "grad_norm": 1.0881028666604091,1191 "learning_rate": 3.349581137957604e-06,1192 "loss": 0.3364,1193 "step": 3381194 },1195 {1196 "epoch": 2.2531223980016652,1197 "grad_norm": 1.211964671213877,1198 "learning_rate": 3.3276585447123957e-06,1199 "loss": 0.3353,1200 "step": 340