ddz16/CamInject-8B
020
1{2 "best_global_step": null,3 "best_metric": null,4 "best_model_checkpoint": null,5 "epoch": 2.0,6 "eval_steps": 500,7 "global_step": 1326,8 "is_hyper_param_search": false,9 "is_local_process_zero": true,10 "is_world_process_zero": true,11 "log_history": [12 {13 "epoch": 0.0015082956259426848,14 "grad_norm": 102.95948028564453,15 "learning_rate": 2.2388059701492537e-07,16 "loss": 0.9106237292289734,17 "step": 1,18 "token_acc": 0.882033248081841519 },20 {21 "epoch": 0.015082956259426848,22 "grad_norm": 86.42596435546875,23 "learning_rate": 2.2388059701492535e-06,24 "loss": 0.7358827590942383,25 "step": 10,26 "token_acc": 0.888117440367592427 },28 {29 "epoch": 0.030165912518853696,30 "grad_norm": 8.159674644470215,31 "learning_rate": 4.477611940298507e-06,32 "loss": 0.22912912368774413,33 "step": 20,34 "token_acc": 0.914275324990151835 },36 {37 "epoch": 0.04524886877828054,38 "grad_norm": 5.822969913482666,39 "learning_rate": 6.716417910447762e-06,40 "loss": 0.16711633205413817,41 "step": 30,42 "token_acc": 0.93153167349511543 },44 {45 "epoch": 0.06033182503770739,46 "grad_norm": 3.3466780185699463,47 "learning_rate": 8.955223880597014e-06,48 "loss": 0.18157639503479003,49 "step": 40,50 "token_acc": 0.935357198954748651 },52 {53 "epoch": 0.07541478129713423,54 "grad_norm": 11.269857406616211,55 "learning_rate": 1.119402985074627e-05,56 "loss": 0.1465624213218689,57 "step": 50,58 "token_acc": 0.940834234119666859 },60 {61 "epoch": 0.09049773755656108,62 "grad_norm": 4.60915470123291,63 "learning_rate": 1.3432835820895523e-05,64 "loss": 0.14280350208282472,65 "step": 60,66 "token_acc": 0.94281261047016167 },68 {69 "epoch": 0.10558069381598793,70 "grad_norm": 5.3014421463012695,71 "learning_rate": 1.4999789854529206e-05,72 "loss": 0.14221720695495604,73 "step": 70,74 "token_acc": 0.943389368468877875 },76 {77 "epoch": 0.12066365007541478,78 "grad_norm": 2.8947246074676514,79 "learning_rate": 1.4996054262643387e-05,80 "loss": 0.140693998336792,81 "step": 80,82 "token_acc": 0.943441960678696583 },84 {85 "epoch": 0.13574660633484162,86 "grad_norm": 27.04346466064453,87 "learning_rate": 1.498765144861652e-05,88 "loss": 0.14304617643356324,89 "step": 90,90 "token_acc": 0.943504604217283191 },92 {93 "epoch": 0.15082956259426847,94 "grad_norm": 2.744962215423584,95 "learning_rate": 1.4974586644241e-05,96 "loss": 0.1404696822166443,97 "step": 100,98 "token_acc": 0.94455431040796999 },100 {101 "epoch": 0.16591251885369532,102 "grad_norm": 2.598541259765625,103 "learning_rate": 1.4956867983975587e-05,104 "loss": 0.13427730798721313,105 "step": 110,106 "token_acc": 0.9465170435897036107 },108 {109 "epoch": 0.18099547511312217,110 "grad_norm": 2.129733085632324,111 "learning_rate": 1.493450649988069e-05,112 "loss": 0.13516327142715454,113 "step": 120,114 "token_acc": 0.9463536368260427115 },116 {117 "epoch": 0.19607843137254902,118 "grad_norm": 1.705326795578003,119 "learning_rate": 1.4907516114749562e-05,120 "loss": 0.13605421781539917,121 "step": 130,122 "token_acc": 0.9467877502727454123 },124 {125 "epoch": 0.21116138763197587,126 "grad_norm": 1.479612112045288,127 "learning_rate": 1.487591363343962e-05,128 "loss": 0.1342160224914551,129 "step": 140,130 "token_acc": 0.9476240697553512131 },132 {133 "epoch": 0.22624434389140272,134 "grad_norm": 1.3483514785766602,135 "learning_rate": 1.4839718732409347e-05,136 "loss": 0.13110628128051757,137 "step": 150,138 "token_acc": 0.9483754956505702139 },140 {141 "epoch": 0.24132730015082957,142 "grad_norm": 1.838326334953308,143 "learning_rate": 1.4798953947467258e-05,144 "loss": 0.13084776401519777,145 "step": 160,146 "token_acc": 0.9485363406997599147 },148 {149 "epoch": 0.2564102564102564,150 "grad_norm": 1.149587631225586,151 "learning_rate": 1.4753644659740567e-05,152 "loss": 0.1279315710067749,153 "step": 170,154 "token_acc": 0.9509291300172259155 },156 {157 "epoch": 0.27149321266968324,158 "grad_norm": 1.747382402420044,159 "learning_rate": 1.4703819079872297e-05,160 "loss": 0.12598202228546143,161 "step": 180,162 "token_acc": 0.9506852612417446163 },164 {165 "epoch": 0.2865761689291101,166 "grad_norm": 1.8707523345947266,167 "learning_rate": 1.4649508230456655e-05,168 "loss": 0.12346329689025878,169 "step": 190,170 "token_acc": 0.9505019520356943171 },172 {173 "epoch": 0.30165912518853694,174 "grad_norm": 0.9066284894943237,175 "learning_rate": 1.459074592672363e-05,176 "loss": 0.12236473560333253,177 "step": 200,178 "token_acc": 0.9514878958951296179 },180 {181 "epoch": 0.3167420814479638,182 "grad_norm": 1.7287721633911133,183 "learning_rate": 1.4527568755484822e-05,184 "loss": 0.11908667087554932,185 "step": 210,186 "token_acc": 0.9531890358873669187 },188 {189 "epoch": 0.33182503770739064,190 "grad_norm": 0.9215026497840881,191 "learning_rate": 1.4460016052353615e-05,192 "loss": 0.12045620679855347,193 "step": 220,194 "token_acc": 0.9537771221341201195 },196 {197 "epoch": 0.3469079939668175,198 "grad_norm": 0.9963810443878174,199 "learning_rate": 1.43881298772539e-05,200 "loss": 0.12107706069946289,201 "step": 230,202 "token_acc": 0.952223704330286203 },204 {205 "epoch": 0.36199095022624433,206 "grad_norm": 3.457219362258911,207 "learning_rate": 1.4311954988232549e-05,208 "loss": 0.11647148132324218,209 "step": 240,210 "token_acc": 0.9550424502102673211 },212 {213 "epoch": 0.3770739064856712,214 "grad_norm": 0.9486528038978577,215 "learning_rate": 1.4231538813591998e-05,216 "loss": 0.12029435634613037,217 "step": 250,218 "token_acc": 0.9532897305739946219 },220 {221 "epoch": 0.39215686274509803,222 "grad_norm": 0.8173623085021973,223 "learning_rate": 1.4146931422360251e-05,224 "loss": 0.11927635669708252,225 "step": 260,226 "token_acc": 0.9536245813747428227 },228 {229 "epoch": 0.4072398190045249,230 "grad_norm": 0.7389430999755859,231 "learning_rate": 1.4058185493116703e-05,232 "loss": 0.11926217079162597,233 "step": 270,234 "token_acc": 0.954318820770103235 },236 {237 "epoch": 0.42232277526395173,238 "grad_norm": 0.9154466390609741,239 "learning_rate": 1.396535628119321e-05,240 "loss": 0.11332788467407226,241 "step": 280,242 "token_acc": 0.9558336565138271243 },244 {245 "epoch": 0.4374057315233786,246 "grad_norm": 1.3126405477523804,247 "learning_rate": 1.3868501584270802e-05,248 "loss": 0.11628217697143554,249 "step": 290,250 "token_acc": 0.9555292936815011251 },252 {253 "epoch": 0.45248868778280543,254 "grad_norm": 0.8111864328384399,255 "learning_rate": 1.3767681706393457e-05,256 "loss": 0.11811476945877075,257 "step": 300,258 "token_acc": 0.9534279727296875259 },260 {261 "epoch": 0.4675716440422323,262 "grad_norm": 0.8737362027168274,263 "learning_rate": 1.3662959420421375e-05,264 "loss": 0.11337666511535645,265 "step": 310,266 "token_acc": 0.9552794842162885267 },268 {269 "epoch": 0.48265460030165913,270 "grad_norm": 1.1029027700424194,271 "learning_rate": 1.35543999289471e-05,272 "loss": 0.11433804035186768,273 "step": 320,274 "token_acc": 0.9556174728604272275 },276 {277 "epoch": 0.497737556561086,278 "grad_norm": 0.6877908110618591,279 "learning_rate": 1.3442070823698822e-05,280 "loss": 0.11494088172912598,281 "step": 330,282 "token_acc": 0.9556713117551905283 },284 {285 "epoch": 0.5128205128205128,286 "grad_norm": 1.1351238489151,287 "learning_rate": 1.332604204345616e-05,288 "loss": 0.11259114742279053,289 "step": 340,290 "token_acc": 0.9564803200863725291 },292 {293 "epoch": 0.5279034690799397,294 "grad_norm": 1.1910680532455444,295 "learning_rate": 1.3206385830504614e-05,296 "loss": 0.1151119589805603,297 "step": 350,298 "token_acc": 0.9553413035094486299 },300 {301 "epoch": 0.5429864253393665,302 "grad_norm": 0.99884432554245,303 "learning_rate": 1.3083176685655786e-05,304 "loss": 0.10849617719650269,305 "step": 360,306 "token_acc": 0.9579699284085654307 },308 {309 "epoch": 0.5580693815987934,310 "grad_norm": 0.6889309883117676,311 "learning_rate": 1.295649132186141e-05,312 "loss": 0.11037075519561768,313 "step": 370,314 "token_acc": 0.9574533969158358315 },316 {317 "epoch": 0.5731523378582202,318 "grad_norm": 0.6279752254486084,319 "learning_rate": 1.2826408616450027e-05,320 "loss": 0.112148118019104,321 "step": 380,322 "token_acc": 0.957041847041847323 },324 {325 "epoch": 0.5882352941176471,326 "grad_norm": 0.7004132270812988,327 "learning_rate": 1.2693009562016084e-05,328 "loss": 0.1127288818359375,329 "step": 390,330 "token_acc": 0.956207282583078331 },332 {333 "epoch": 0.6033182503770739,334 "grad_norm": 0.7210747599601746,335 "learning_rate": 1.2556377215992019e-05,336 "loss": 0.11080955266952515,337 "step": 400,338 "token_acc": 0.9577135638717827339 },340 {341 "epoch": 0.6184012066365008,342 "grad_norm": 0.8611845374107361,343 "learning_rate": 1.2416596648934716e-05,344 "loss": 0.10928909778594971,345 "step": 410,346 "token_acc": 0.9575848530325769347 },348 {349 "epoch": 0.6334841628959276,350 "grad_norm": 0.7873753309249878,351 "learning_rate": 1.2273754891558542e-05,352 "loss": 0.11083755493164063,353 "step": 420,354 "token_acc": 0.9573786571441708355 },356 {357 "epoch": 0.6485671191553545,358 "grad_norm": 0.8866596221923828,359 "learning_rate": 1.212794088054797e-05,360 "loss": 0.11016602516174316,361 "step": 430,362 "token_acc": 0.9571059510524788363 },364 {365 "epoch": 0.6636500754147813,366 "grad_norm": 0.5996086001396179,367 "learning_rate": 1.1979245403183451e-05,368 "loss": 0.10762960910797119,369 "step": 440,370 "token_acc": 0.9582261423496263371 },372 {373 "epoch": 0.6787330316742082,374 "grad_norm": 0.8173403739929199,375 "learning_rate": 1.182776104081511e-05,376 "loss": 0.1057842493057251,377 "step": 450,378 "token_acc": 0.9589248277927523379 },380 {381 "epoch": 0.693815987933635,382 "grad_norm": 0.8952581882476807,383 "learning_rate": 1.167358211121938e-05,384 "loss": 0.10421741008758545,385 "step": 460,386 "token_acc": 0.959809739243605387 },388 {389 "epoch": 0.7088989441930619,390 "grad_norm": 0.8190169334411621,391 "learning_rate": 1.1516804609874507e-05,392 "loss": 0.10857417583465576,393 "step": 470,394 "token_acc": 0.9578637775543585395 },396 {397 "epoch": 0.7239819004524887,398 "grad_norm": 0.8312342762947083,399 "learning_rate": 1.1357526150191471e-05,400 "loss": 0.10505218505859375,401 "step": 480,402 "token_acc": 0.958963653076168403 },404 {405 "epoch": 0.7390648567119156,406 "grad_norm": 0.6930398941040039,407 "learning_rate": 1.1195845902737558e-05,408 "loss": 0.10784991979598998,409 "step": 490,410 "token_acc": 0.9578753158963705411 },412 {413 "epoch": 0.7541478129713424,414 "grad_norm": 0.7167505025863647,415 "learning_rate": 1.1031864533490397e-05,416 "loss": 0.10691661834716797,417 "step": 500,418 "token_acc": 0.9590415199708632419 },420 {421 "epoch": 0.7692307692307693,422 "grad_norm": 0.9776201844215393,423 "learning_rate": 1.0865684141160912e-05,424 "loss": 0.10396910905838012,425 "step": 510,426 "token_acc": 0.9596143463170074427 },428 {429 "epoch": 0.7843137254901961,430 "grad_norm": 0.6151344180107117,431 "learning_rate": 1.0697408193624248e-05,432 "loss": 0.10538628101348876,433 "step": 520,434 "token_acc": 0.9587614615490946435 },436 {437 "epoch": 0.799396681749623,438 "grad_norm": 0.6072873473167419,439 "learning_rate": 1.0527141463498186e-05,440 "loss": 0.10715539455413818,441 "step": 530,442 "token_acc": 0.9581875852255884443 },444 {445 "epoch": 0.8144796380090498,446 "grad_norm": 0.6179074048995972,447 "learning_rate": 1.0354989962909223e-05,448 "loss": 0.10194606781005859,449 "step": 540,450 "token_acc": 0.9605654176952467451 },452 {453 "epoch": 0.8295625942684767,454 "grad_norm": 0.6735125184059143,455 "learning_rate": 1.0181060877486896e-05,456 "loss": 0.10430541038513183,457 "step": 550,458 "token_acc": 0.9597842006157157459 },460 {461 "epoch": 0.8446455505279035,462 "grad_norm": 0.9804912209510803,463 "learning_rate": 1.0005462499627443e-05,464 "loss": 0.09949544668197632,465 "step": 560,466 "token_acc": 0.9620176783326266467 },468 {469 "epoch": 0.8597285067873304,470 "grad_norm": 0.7045010328292847,471 "learning_rate": 9.82830416106839e-06,472 "loss": 0.10367201566696167,473 "step": 570,474 "token_acc": 0.96001593625498475 },476 {477 "epoch": 0.8748114630467572,478 "grad_norm": 0.7430729866027832,479 "learning_rate": 9.649696164815983e-06,480 "loss": 0.09992576837539673,481 "step": 580,482 "token_acc": 0.9614963024717089483 },484 {485 "epoch": 0.889894419306184,486 "grad_norm": 0.6927557587623596,487 "learning_rate": 9.469749716467922e-06,488 "loss": 0.10162270069122314,489 "step": 590,490 "token_acc": 0.9600887950572822491 },492 {493 "epoch": 0.9049773755656109,494 "grad_norm": 0.7048885226249695,495 "learning_rate": 9.288576854974107e-06,496 "loss": 0.1029390573501587,497 "step": 600,498 "token_acc": 0.9599495381581428499 },500 {501 "epoch": 0.9200603318250377,502 "grad_norm": 0.7786235809326172,503 "learning_rate": 9.106290382878496e-06,504 "loss": 0.10250804424285889,505 "step": 610,506 "token_acc": 0.9600262723773446507 },508 {509 "epoch": 0.9351432880844646,510 "grad_norm": 0.6432821154594421,511 "learning_rate": 8.923003796085571e-06,512 "loss": 0.09980993270874024,513 "step": 620,514 "token_acc": 0.9616151866151866515 },516 {517 "epoch": 0.9502262443438914,518 "grad_norm": 0.4739595353603363,519 "learning_rate": 8.738831213195076e-06,520 "loss": 0.10440936088562011,521 "step": 630,522 "token_acc": 0.9596175053688548523 },524 {525 "epoch": 0.9653092006033183,526 "grad_norm": 0.5999048352241516,527 "learning_rate": 8.553887304449051e-06,528 "loss": 0.09771686196327209,529 "step": 640,530 "token_acc": 0.9621664920820753531 },532 {533 "epoch": 0.9803921568627451,534 "grad_norm": 0.6257745027542114,535 "learning_rate": 8.368287220335405e-06,536 "loss": 0.09986623525619506,537 "step": 650,538 "token_acc": 0.9612191817091594539 },540 {541 "epoch": 0.995475113122172,542 "grad_norm": 0.6971683502197266,543 "learning_rate": 8.18214651989248e-06,544 "loss": 0.10077552795410157,545 "step": 660,546 "token_acc": 0.9610657983311677547 },548 {549 "epoch": 1.0105580693815988,550 "grad_norm": 0.5287856459617615,551 "learning_rate": 7.99558109875924e-06,552 "loss": 0.09711869359016419,553 "step": 670,554 "token_acc": 0.9632017856589268555 },556 {557 "epoch": 1.0256410256410255,558 "grad_norm": 0.5113791227340698,559 "learning_rate": 7.808707117015871e-06,560 "loss": 0.09611727595329285,561 "step": 680,562 "token_acc": 0.9628295031473811563 },564 {565 "epoch": 1.0407239819004526,566 "grad_norm": 0.9405093193054199,567 "learning_rate": 7.621640926859767e-06,568 "loss": 0.09766757488250732,569 "step": 690,570 "token_acc": 0.961746076320634571 },572 {573 "epoch": 1.0558069381598794,574 "grad_norm": 0.7931028604507446,575 "learning_rate": 7.434499000161847e-06,576 "loss": 0.09249839186668396,577 "step": 700,578 "token_acc": 0.9642857142857143579 },580 {581 "epoch": 1.0708898944193062,582 "grad_norm": 0.5711884498596191,583 "learning_rate": 7.247397855948417e-06,584 "loss": 0.09391047954559326,585 "step": 710,586 "token_acc": 0.9633587119502715587 },588 {589 "epoch": 1.085972850678733,590 "grad_norm": 0.5200867056846619,591 "learning_rate": 7.060453987853635e-06,592 "loss": 0.09218854904174804,593 "step": 720,594 "token_acc": 0.9643104698960264595 },596 {597 "epoch": 1.10105580693816,598 "grad_norm": 0.5937605500221252,599 "learning_rate": 6.873783791587794e-06,600 "loss": 0.09332010746002198,601 "step": 730,602 "token_acc": 0.963314348369019603 },604 {605 "epoch": 1.1161387631975868,606 "grad_norm": 0.5634381771087646,607 "learning_rate": 6.6875034924665825e-06,608 "loss": 0.09174969792366028,609 "step": 740,610 "token_acc": 0.9643075846382644611 },612 {613 "epoch": 1.1312217194570136,614 "grad_norm": 0.575001060962677,615 "learning_rate": 6.501729073046421e-06,616 "loss": 0.09115147590637207,617 "step": 750,618 "token_acc": 0.96492840783385619 },620 {621 "epoch": 1.1463046757164403,622 "grad_norm": 0.5640040040016174,623 "learning_rate": 6.316576200910982e-06,624 "loss": 0.08881928324699402,625 "step": 760,626 "token_acc": 0.9655490814149712627 },628 {629 "epoch": 1.1613876319758674,630 "grad_norm": 0.6095544099807739,631 "learning_rate": 6.1321601566537614e-06,632 "loss": 0.09392951130867004,633 "step": 770,634 "token_acc": 0.9636105045397081635 },636 {637 "epoch": 1.1764705882352942,638 "grad_norm": 0.6617202758789062,639 "learning_rate": 5.94859576210165e-06,640 "loss": 0.09372889995574951,641 "step": 780,642 "token_acc": 0.9637654206186845643 },644 {645 "epoch": 1.191553544494721,646 "grad_norm": 1.2394925355911255,647 "learning_rate": 5.765997308824127e-06,648 "loss": 0.09473921060562134,649 "step": 790,650 "token_acc": 0.9626901771085278651 },652 {653 "epoch": 1.2066365007541477,654 "grad_norm": 0.5932070016860962,655 "learning_rate": 5.584478486972594e-06,656 "loss": 0.09192266464233398,657 "step": 800,658 "token_acc": 0.9635680926003507659 },660 {661 "epoch": 1.2217194570135748,662 "grad_norm": 0.9952248930931091,663 "learning_rate": 5.404152314494185e-06,664 "loss": 0.09385214447975158,665 "step": 810,666 "token_acc": 0.9637176022601646667 },668 {669 "epoch": 1.2368024132730016,670 "grad_norm": 0.6110525727272034,671 "learning_rate": 5.225131066764096e-06,672 "loss": 0.08856210112571716,673 "step": 820,674 "token_acc": 0.9655632553414105675 },676 {677 "epoch": 1.2518853695324283,678 "grad_norm": 0.6847214698791504,679 "learning_rate": 5.04752620668025e-06,680 "loss": 0.09417049884796143,681 "step": 830,682 "token_acc": 0.9632891549095055683 },684 {685 "epoch": 1.2669683257918551,686 "grad_norm": 0.6929960250854492,687 "learning_rate": 4.871448315263849e-06,688 "loss": 0.09022506475448608,689 "step": 840,690 "token_acc": 0.9648385764536664691 },692 {693 "epoch": 1.282051282051282,694 "grad_norm": 0.5659875869750977,695 "learning_rate": 4.697007022808989e-06,696 "loss": 0.08967013955116272,697 "step": 850,698 "token_acc": 0.9652773666449589699 },700 {701 "epoch": 1.297134238310709,702 "grad_norm": 0.819989025592804,703 "learning_rate": 4.52431094062422e-06,704 "loss": 0.09020506143569947,705 "step": 860,706 "token_acc": 0.964875810350934707 },708 {709 "epoch": 1.3122171945701357,710 "grad_norm": 0.6961911916732788,711 "learning_rate": 4.353467593408566e-06,712 "loss": 0.09065240621566772,713 "step": 870,714 "token_acc": 0.964779952624361715 },716 {717 "epoch": 1.3273001508295625,718 "grad_norm": 0.711963951587677,719 "learning_rate": 4.1845833523040775e-06,720 "loss": 0.09238415956497192,721 "step": 880,722 "token_acc": 0.9640809911242604723 },724 {725 "epoch": 1.3423831070889896,726 "grad_norm": 0.5482915639877319,727 "learning_rate": 4.0177633686666225e-06,728 "loss": 0.08626269102096558,729 "step": 890,730 "token_acc": 0.9664413009944534731 },732 {733 "epoch": 1.3574660633484164,734 "grad_norm": 0.6547002792358398,735 "learning_rate": 3.853111508596134e-06,736 "loss": 0.08824948668479919,737 "step": 900,738 "token_acc": 0.9652458695864699739 },740 {741 "epoch": 1.3725490196078431,742 "grad_norm": 0.7393091917037964,743 "learning_rate": 3.6907302882671214e-06,744 "loss": 0.0905347228050232,745 "step": 910,746 "token_acc": 0.965180890257897747 },748 {749 "epoch": 1.38763197586727,750 "grad_norm": 0.7055743932723999,751 "learning_rate": 3.5307208100996376e-06,752 "loss": 0.08957746028900146,753 "step": 920,754 "token_acc": 0.965201436723524755 },756 {757 "epoch": 1.4027149321266967,758 "grad_norm": 0.5165733695030212,759 "learning_rate": 3.373182699810508e-06,760 "loss": 0.0900674045085907,761 "step": 930,762 "token_acc": 0.9648800220649076763 },764 {765 "epoch": 1.4177978883861238,766 "grad_norm": 0.6470308303833008,767 "learning_rate": 3.2182140443839834e-06,768 "loss": 0.09126996994018555,769 "step": 940,770 "token_acc": 0.9645374984806127771 },772 {773 "epoch": 1.4328808446455505,774 "grad_norm": 0.7675370573997498,775 "learning_rate": 3.0659113310004437e-06,776 "loss": 0.09289331436157226,777 "step": 950,778 "token_acc": 0.9638552426246563779 },780 {781 "epoch": 1.4479638009049773,782 "grad_norm": 0.5420223474502563,783 "learning_rate": 2.9163693869611806e-06,784 "loss": 0.08824996948242188,785 "step": 960,786 "token_acc": 0.9658039529093536787 },788 {789 "epoch": 1.4630467571644044,790 "grad_norm": 0.49305713176727295,791 "learning_rate": 2.7696813206466494e-06,792 "loss": 0.0872989296913147,793 "step": 970,794 "token_acc": 0.966210141452889795 },796 {797 "epoch": 1.4781297134238311,798 "grad_norm": 0.5599616169929504,799 "learning_rate": 2.6259384635449823e-06,800 "loss": 0.08567045331001281,801 "step": 980,802 "token_acc": 0.9661950765380095803 },804 {805 "epoch": 1.493212669683258,806 "grad_norm": 0.5709689259529114,807 "learning_rate": 2.48523031338683e-06,808 "loss": 0.09245221614837647,809 "step": 990,810 "token_acc": 0.9648156761541808811 },812 {813 "epoch": 1.5082956259426847,814 "grad_norm": 0.6077988147735596,815 "learning_rate": 2.347644478421916e-06,816 "loss": 0.09017724990844726,817 "step": 1000,818 "token_acc": 0.9644074377669443819 },820 {821 "epoch": 1.5233785822021115,822 "grad_norm": 0.5771783590316772,823 "learning_rate": 2.2132666228720783e-06,824 "loss": 0.09128254652023315,825 "step": 1010,826 "token_acc": 0.964498933901919827 },828 {829 "epoch": 1.5384615384615383,830 "grad_norm": 0.5295813083648682,831 "learning_rate": 2.0821804135946685e-06,832 "loss": 0.0867614507675171,833 "step": 1020,834 "token_acc": 0.9657581438574565835 },836 {837 "epoch": 1.5535444947209653,838 "grad_norm": 0.6218671202659607,839 "learning_rate": 1.954467467989575e-06,840 "loss": 0.08816643357276917,841 "step": 1030,842 "token_acc": 0.9664409193521507843 },844 {845 "epoch": 1.5686274509803921,846 "grad_norm": 0.5615459084510803,847 "learning_rate": 1.8302073031822736e-06,848 "loss": 0.08652478456497192,849 "step": 1040,850 "token_acc": 0.9657081486635708851 },852 {853 "epoch": 1.5837104072398192,854 "grad_norm": 0.6350222229957581,855 "learning_rate": 1.7094772865145908e-06,856 "loss": 0.08563481569290161,857 "step": 1050,858 "token_acc": 0.9670686976438743859 },860 {861 "epoch": 1.598793363499246,862 "grad_norm": 0.611201286315918,863 "learning_rate": 1.5923525873739312e-06,864 "loss": 0.08522043824195862,865 "step": 1060,866 "token_acc": 0.9665700207551056867 },868 {869 "epoch": 1.6138763197586727,870 "grad_norm": 0.5586323142051697,871 "learning_rate": 1.4789061303910304e-06,872 "loss": 0.09052041172981262,873 "step": 1070,874 "token_acc": 0.9644768042804159875 },876 {877 "epoch": 1.6289592760180995,878 "grad_norm": 0.6737985014915466,879 "learning_rate": 1.3692085500353444e-06,880 "loss": 0.08938039541244507,881 "step": 1080,882 "token_acc": 0.965438055926981883 },884 {885 "epoch": 1.6440422322775263,886 "grad_norm": 0.5793459415435791,887 "learning_rate": 1.263328146636342e-06,888 "loss": 0.08627030849456788,889 "step": 1090,890 "token_acc": 0.9662307459500789891 },892 {893 "epoch": 1.6591251885369531,894 "grad_norm": 0.553979754447937,895 "learning_rate": 1.1613308438580924e-06,896 "loss": 0.08699219822883605,897 "step": 1100,898 "token_acc": 0.9655294135702994899 },900 {901 "epoch": 1.6742081447963801,902 "grad_norm": 0.6702016592025757,903 "learning_rate": 1.063280147653618e-06,904 "loss": 0.08392992615699768,905 "step": 1110,906 "token_acc": 0.9671428105534358907 },908 {909 "epoch": 1.689291101055807,910 "grad_norm": 0.5012071132659912,911 "learning_rate": 9.692371067245726e-07,912 "loss": 0.09083313941955566,913 "step": 1120,914 "token_acc": 0.9637969026411347915 },916 {917 "epoch": 1.704374057315234,918 "grad_norm": 0.5176061987876892,919 "learning_rate": 8.792602745108685e-07,920 "loss": 0.0841116487979889,921 "step": 1130,922 "token_acc": 0.9662585332247761923 },924 {925 "epoch": 1.7194570135746607,926 "grad_norm": 0.7949459552764893,927 "learning_rate": 7.93405672733902e-07,928 "loss": 0.08959006071090699,929 "step": 1140,930 "token_acc": 0.9651756949127145931 },932 {933 "epoch": 1.7345399698340875,934 "grad_norm": 0.4745083749294281,935 "learning_rate": 7.117267565160989e-07,936 "loss": 0.08536668419837952,937 "step": 1150,938 "token_acc": 0.966504120772119939 },940 {941 "epoch": 1.7496229260935143,942 "grad_norm": 0.6467885375022888,943 "learning_rate": 6.342743810984836e-07,944 "loss": 0.08896956443786622,945 "step": 1160,946 "token_acc": 0.9651875204473128947 },948 {949 "epoch": 1.7647058823529411,950 "grad_norm": 0.6844164729118347,951 "learning_rate": 5.610967701769814e-07,952 "loss": 0.08681308031082154,953 "step": 1170,954 "token_acc": 0.966941761230423955 },956 {957 "epoch": 1.779788838612368,958 "grad_norm": 0.6476861238479614,959 "learning_rate": 4.922394858772067e-07,960 "loss": 0.087920880317688,961 "step": 1180,962 "token_acc": 0.9660436333785875963 },964 {965 "epoch": 1.7948717948717947,966 "grad_norm": 0.5709762573242188,967 "learning_rate": 4.2774540038639286e-07,968 "loss": 0.0817031741142273,969 "step": 1190,970 "token_acc": 0.9678674654050475971 },972 {973 "epoch": 1.8099547511312217,974 "grad_norm": 0.5973171591758728,975 "learning_rate": 3.676546692601465e-07,976 "loss": 0.08370765447616577,977 "step": 1200,978 "token_acc": 0.9672491137810961979 },980 {981 "epoch": 1.8250377073906485,982 "grad_norm": 0.775956392288208,983 "learning_rate": 3.1200470642063716e-07,984 "loss": 0.0849980354309082,985 "step": 1210,986 "token_acc": 0.9670338146365544987 },988 {989 "epoch": 1.8401206636500755,990 "grad_norm": 0.5492123365402222,991 "learning_rate": 2.6083016086180715e-07,992 "loss": 0.08112341165542603,993 "step": 1220,994 "token_acc": 0.9692397073140093995 },996 {997 "epoch": 1.8552036199095023,998 "grad_norm": 0.5747104287147522,999 "learning_rate": 2.14162895076071e-07,1000 "loss": 0.0888913631439209,1001 "step": 1230,1002 "token_acc": 0.9648673732111851003 },1004 {1005 "epoch": 1.8702865761689291,1006 "grad_norm": 0.5583189725875854,1007 "learning_rate": 1.7203196521597354e-07,1008 "loss": 0.08525767922401428,1009 "step": 1240,1010 "token_acc": 0.96670443101711991011 },1012 {1013 "epoch": 1.885369532428356,1014 "grad_norm": 0.5149325728416443,1015 "learning_rate": 1.3446360300313375e-07,1016 "loss": 0.08332073092460632,1017 "step": 1250,1018 "token_acc": 0.96779862265495131019 },1020 {1021 "epoch": 1.9004524886877827,1022 "grad_norm": 0.6395967602729797,1023 "learning_rate": 1.0148119939575207e-07,1024 "loss": 0.0832521915435791,1025 "step": 1260,1026 "token_acc": 0.96748333333333341027 },1028 {1029 "epoch": 1.9155354449472095,1030 "grad_norm": 0.5390537977218628,1031 "learning_rate": 7.310529002484079e-08,1032 "loss": 0.0882462739944458,1033 "step": 1270,1034 "token_acc": 0.96574065985265421035 },1036 {1037 "epoch": 1.9306184012066365,1038 "grad_norm": 0.5799115896224976,1039 "learning_rate": 4.9353542408249465e-08,1040 "loss": 0.08766170740127563,1041 "step": 1280,1042 "token_acc": 0.96535718153827111043 },1044 {1045 "epoch": 1.9457013574660633,1046 "grad_norm": 0.5990166664123535,1047 "learning_rate": 3.024074495044826e-08,1048 "loss": 0.0859516441822052,1049 "step": 1290,1050 "token_acc": 0.96669382278238371051 },1052 {1053 "epoch": 1.9607843137254903,1054 "grad_norm": 1.4371193647384644,1055 "learning_rate": 1.5778797734913307e-08,1056 "loss": 0.08548210859298706,1057 "step": 1300,1058 "token_acc": 0.96620365472105921059 },1060 {1061 "epoch": 1.9758672699849171,1062 "grad_norm": 0.5660617351531982,1063 "learning_rate": 5.976705114852654e-09,1064 "loss": 0.08374635577201843,1065 "step": 1310,1066 "token_acc": 0.96710631622755061067 },1068 {1069 "epoch": 1.990950226244344,1070 "grad_norm": 0.6013316512107849,1071 "learning_rate": 8.40570106880878e-10,1072 "loss": 0.08703867197036744,1073 "step": 1320,1074 "token_acc": 0.96588971863920951075 }1076 ],1077 "logging_steps": 10,1078 "max_steps": 1326,1079 "num_input_tokens_seen": 0,1080 "num_train_epochs": 2,1081 "save_steps": 500,1082 "stateful_callbacks": {1083 "TrainerControl": {1084 "args": {1085 "should_epoch_stop": false,1086 "should_evaluate": false,1087 "should_log": false,1088 "should_save": true,1089 "should_training_stop": true1090 },1091 "attributes": {}1092 }1093 },1094 "total_flos": 1.5869318005205762e+19,1095 "train_batch_size": 2,1096 "trial_name": null,1097 "trial_params": null1098}1099 