{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 1207, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0008285004142502071, "grad_norm": 23.866117550094145, "learning_rate": 8.264462809917357e-08, "loss": 1.4163, "step": 1 }, { "epoch": 0.004142502071251036, "grad_norm": 22.991207135218495, "learning_rate": 4.132231404958678e-07, "loss": 1.4008, "step": 5 }, { "epoch": 0.008285004142502071, "grad_norm": 8.161072224423009, "learning_rate": 8.264462809917356e-07, "loss": 1.3121, "step": 10 }, { "epoch": 0.012427506213753107, "grad_norm": 10.651287293408194, "learning_rate": 1.2396694214876035e-06, "loss": 1.1732, "step": 15 }, { "epoch": 0.016570008285004142, "grad_norm": 2.9695263532042517, "learning_rate": 1.6528925619834712e-06, "loss": 1.0137, "step": 20 }, { "epoch": 0.020712510356255178, "grad_norm": 3.0649608491168774, "learning_rate": 2.066115702479339e-06, "loss": 0.9608, "step": 25 }, { "epoch": 0.024855012427506214, "grad_norm": 2.4113702925470575, "learning_rate": 2.479338842975207e-06, "loss": 0.9263, "step": 30 }, { "epoch": 0.02899751449875725, "grad_norm": 2.2764880745044525, "learning_rate": 2.8925619834710743e-06, "loss": 0.9069, "step": 35 }, { "epoch": 0.033140016570008285, "grad_norm": 2.307963331742189, "learning_rate": 3.3057851239669424e-06, "loss": 0.8949, "step": 40 }, { "epoch": 0.037282518641259324, "grad_norm": 2.290742694256339, "learning_rate": 3.71900826446281e-06, "loss": 0.8849, "step": 45 }, { "epoch": 0.041425020712510356, "grad_norm": 2.418716943098049, "learning_rate": 4.132231404958678e-06, "loss": 0.8662, "step": 50 }, { "epoch": 0.045567522783761395, "grad_norm": 2.256258952989886, "learning_rate": 4.5454545454545455e-06, "loss": 0.8617, "step": 55 }, { "epoch": 0.04971002485501243, "grad_norm": 2.3643676843983013, "learning_rate": 4.958677685950414e-06, "loss": 0.8434, "step": 60 }, { "epoch": 0.053852526926263466, "grad_norm": 2.51362497881208, "learning_rate": 5.371900826446281e-06, "loss": 0.8586, "step": 65 }, { "epoch": 0.0579950289975145, "grad_norm": 2.409197722940547, "learning_rate": 5.785123966942149e-06, "loss": 0.8391, "step": 70 }, { "epoch": 0.06213753106876554, "grad_norm": 2.504763734639035, "learning_rate": 6.198347107438017e-06, "loss": 0.8323, "step": 75 }, { "epoch": 0.06628003314001657, "grad_norm": 2.342235286421664, "learning_rate": 6.611570247933885e-06, "loss": 0.8366, "step": 80 }, { "epoch": 0.07042253521126761, "grad_norm": 2.601805905814255, "learning_rate": 7.0247933884297525e-06, "loss": 0.83, "step": 85 }, { "epoch": 0.07456503728251865, "grad_norm": 2.4729528650877226, "learning_rate": 7.43801652892562e-06, "loss": 0.8193, "step": 90 }, { "epoch": 0.07870753935376967, "grad_norm": 2.347978873060577, "learning_rate": 7.851239669421489e-06, "loss": 0.8108, "step": 95 }, { "epoch": 0.08285004142502071, "grad_norm": 2.3865550609084005, "learning_rate": 8.264462809917356e-06, "loss": 0.8068, "step": 100 }, { "epoch": 0.08699254349627175, "grad_norm": 2.397866066956603, "learning_rate": 8.677685950413224e-06, "loss": 0.8146, "step": 105 }, { "epoch": 0.09113504556752279, "grad_norm": 2.528140675905014, "learning_rate": 9.090909090909091e-06, "loss": 0.8014, "step": 110 }, { "epoch": 0.09527754763877382, "grad_norm": 2.5772809506040004, "learning_rate": 9.50413223140496e-06, "loss": 0.7929, "step": 115 }, { "epoch": 0.09942004971002485, "grad_norm": 2.4698447608273226, "learning_rate": 9.917355371900828e-06, "loss": 0.7957, "step": 120 }, { "epoch": 0.1035625517812759, "grad_norm": 2.4271647358481077, "learning_rate": 9.999665269535307e-06, "loss": 0.7947, "step": 125 }, { "epoch": 0.10770505385252693, "grad_norm": 2.399585162136438, "learning_rate": 9.998305503833872e-06, "loss": 0.7952, "step": 130 }, { "epoch": 0.11184755592377796, "grad_norm": 2.356625988593114, "learning_rate": 9.995900066492902e-06, "loss": 0.7903, "step": 135 }, { "epoch": 0.115990057995029, "grad_norm": 2.3562051609533894, "learning_rate": 9.992449460742464e-06, "loss": 0.7855, "step": 140 }, { "epoch": 0.12013256006628004, "grad_norm": 2.2315403444753925, "learning_rate": 9.98795440846732e-06, "loss": 0.7721, "step": 145 }, { "epoch": 0.12427506213753108, "grad_norm": 2.3299296938897203, "learning_rate": 9.982415850055902e-06, "loss": 0.7733, "step": 150 }, { "epoch": 0.12841756420878211, "grad_norm": 2.176864153510824, "learning_rate": 9.975834944203581e-06, "loss": 0.7641, "step": 155 }, { "epoch": 0.13256006628003314, "grad_norm": 2.189529036305693, "learning_rate": 9.968213067670265e-06, "loss": 0.7688, "step": 160 }, { "epoch": 0.13670256835128416, "grad_norm": 2.232386403924057, "learning_rate": 9.959551814992364e-06, "loss": 0.7458, "step": 165 }, { "epoch": 0.14084507042253522, "grad_norm": 2.596315903481849, "learning_rate": 9.949852998149217e-06, "loss": 0.7621, "step": 170 }, { "epoch": 0.14498757249378624, "grad_norm": 2.2955324137983615, "learning_rate": 9.939118646184007e-06, "loss": 0.7456, "step": 175 }, { "epoch": 0.1491300745650373, "grad_norm": 2.5048442022915363, "learning_rate": 9.927351004779275e-06, "loss": 0.7536, "step": 180 }, { "epoch": 0.15327257663628832, "grad_norm": 2.373845911290789, "learning_rate": 9.914552535787122e-06, "loss": 0.7334, "step": 185 }, { "epoch": 0.15741507870753935, "grad_norm": 2.490700611604481, "learning_rate": 9.900725916714157e-06, "loss": 0.7385, "step": 190 }, { "epoch": 0.1615575807787904, "grad_norm": 2.4943108442217667, "learning_rate": 9.885874040161373e-06, "loss": 0.7226, "step": 195 }, { "epoch": 0.16570008285004142, "grad_norm": 2.6495649004577753, "learning_rate": 9.87000001321898e-06, "loss": 0.7309, "step": 200 }, { "epoch": 0.16984258492129245, "grad_norm": 2.24352822185874, "learning_rate": 9.853107156816393e-06, "loss": 0.724, "step": 205 }, { "epoch": 0.1739850869925435, "grad_norm": 2.577647983056561, "learning_rate": 9.835199005027477e-06, "loss": 0.7137, "step": 210 }, { "epoch": 0.17812758906379453, "grad_norm": 2.4661359633344566, "learning_rate": 9.816279304331202e-06, "loss": 0.711, "step": 215 }, { "epoch": 0.18227009113504558, "grad_norm": 2.3694444605967147, "learning_rate": 9.79635201282785e-06, "loss": 0.7101, "step": 220 }, { "epoch": 0.1864125932062966, "grad_norm": 2.198135101539898, "learning_rate": 9.775421299410977e-06, "loss": 0.6933, "step": 225 }, { "epoch": 0.19055509527754763, "grad_norm": 2.3950158836366224, "learning_rate": 9.753491542895237e-06, "loss": 0.707, "step": 230 }, { "epoch": 0.19469759734879868, "grad_norm": 2.443771290836774, "learning_rate": 9.730567331100333e-06, "loss": 0.693, "step": 235 }, { "epoch": 0.1988400994200497, "grad_norm": 2.69404610606291, "learning_rate": 9.706653459891207e-06, "loss": 0.6823, "step": 240 }, { "epoch": 0.20298260149130073, "grad_norm": 2.6424681261577265, "learning_rate": 9.681754932174719e-06, "loss": 0.6846, "step": 245 }, { "epoch": 0.2071251035625518, "grad_norm": 2.3720791178993106, "learning_rate": 9.655876956853025e-06, "loss": 0.6829, "step": 250 }, { "epoch": 0.2112676056338028, "grad_norm": 2.246578155524963, "learning_rate": 9.629024947733836e-06, "loss": 0.6802, "step": 255 }, { "epoch": 0.21541010770505387, "grad_norm": 2.532693053271893, "learning_rate": 9.601204522397826e-06, "loss": 0.6811, "step": 260 }, { "epoch": 0.2195526097763049, "grad_norm": 2.3168391897960996, "learning_rate": 9.572421501023403e-06, "loss": 0.6737, "step": 265 }, { "epoch": 0.22369511184755592, "grad_norm": 2.2740027059012102, "learning_rate": 9.5426819051691e-06, "loss": 0.6762, "step": 270 }, { "epoch": 0.22783761391880697, "grad_norm": 2.453598814349311, "learning_rate": 9.511991956513828e-06, "loss": 0.6689, "step": 275 }, { "epoch": 0.231980115990058, "grad_norm": 2.6061851377315857, "learning_rate": 9.480358075555278e-06, "loss": 0.6523, "step": 280 }, { "epoch": 0.23612261806130902, "grad_norm": 2.3350040791986517, "learning_rate": 9.447786880266706e-06, "loss": 0.6626, "step": 285 }, { "epoch": 0.24026512013256007, "grad_norm": 2.16738386427134, "learning_rate": 9.414285184712432e-06, "loss": 0.6526, "step": 290 }, { "epoch": 0.2444076222038111, "grad_norm": 2.353502989995007, "learning_rate": 9.37985999762229e-06, "loss": 0.6503, "step": 295 }, { "epoch": 0.24855012427506215, "grad_norm": 2.559615826375987, "learning_rate": 9.344518520925377e-06, "loss": 0.6409, "step": 300 }, { "epoch": 0.2526926263463132, "grad_norm": 2.2923285566494296, "learning_rate": 9.308268148243355e-06, "loss": 0.628, "step": 305 }, { "epoch": 0.25683512841756423, "grad_norm": 2.20361635294026, "learning_rate": 9.271116463343692e-06, "loss": 0.6401, "step": 310 }, { "epoch": 0.2609776304888152, "grad_norm": 2.3077464669006966, "learning_rate": 9.23307123855307e-06, "loss": 0.6322, "step": 315 }, { "epoch": 0.2651201325600663, "grad_norm": 2.36438437607611, "learning_rate": 9.194140433131397e-06, "loss": 0.6122, "step": 320 }, { "epoch": 0.26926263463131733, "grad_norm": 2.3361508013893255, "learning_rate": 9.154332191606671e-06, "loss": 0.6279, "step": 325 }, { "epoch": 0.27340513670256833, "grad_norm": 2.3055892316817324, "learning_rate": 9.113654842071114e-06, "loss": 0.6136, "step": 330 }, { "epoch": 0.2775476387738194, "grad_norm": 2.150656102185965, "learning_rate": 9.072116894438885e-06, "loss": 0.6132, "step": 335 }, { "epoch": 0.28169014084507044, "grad_norm": 2.1980462175987707, "learning_rate": 9.029727038665765e-06, "loss": 0.6238, "step": 340 }, { "epoch": 0.28583264291632143, "grad_norm": 2.1777751804898915, "learning_rate": 8.986494142931168e-06, "loss": 0.6065, "step": 345 }, { "epoch": 0.2899751449875725, "grad_norm": 2.33402398930558, "learning_rate": 8.94242725178288e-06, "loss": 0.6107, "step": 350 }, { "epoch": 0.29411764705882354, "grad_norm": 2.2919553245717306, "learning_rate": 8.89753558424488e-06, "loss": 0.6065, "step": 355 }, { "epoch": 0.2982601491300746, "grad_norm": 2.421470706741744, "learning_rate": 8.851828531888692e-06, "loss": 0.6123, "step": 360 }, { "epoch": 0.3024026512013256, "grad_norm": 2.226227311801291, "learning_rate": 8.805315656868587e-06, "loss": 0.614, "step": 365 }, { "epoch": 0.30654515327257664, "grad_norm": 2.2094485446668233, "learning_rate": 8.75800668992117e-06, "loss": 0.5961, "step": 370 }, { "epoch": 0.3106876553438277, "grad_norm": 2.172339243964824, "learning_rate": 8.709911528329623e-06, "loss": 0.5977, "step": 375 }, { "epoch": 0.3148301574150787, "grad_norm": 2.202510437920946, "learning_rate": 8.661040233853166e-06, "loss": 0.5913, "step": 380 }, { "epoch": 0.31897265948632975, "grad_norm": 2.4240297905693535, "learning_rate": 8.611403030622074e-06, "loss": 0.5885, "step": 385 }, { "epoch": 0.3231151615575808, "grad_norm": 2.3295900029200447, "learning_rate": 8.561010302998734e-06, "loss": 0.5754, "step": 390 }, { "epoch": 0.3272576636288318, "grad_norm": 2.396863934008288, "learning_rate": 8.509872593405189e-06, "loss": 0.5852, "step": 395 }, { "epoch": 0.33140016570008285, "grad_norm": 2.292182476322964, "learning_rate": 8.458000600117604e-06, "loss": 0.5747, "step": 400 }, { "epoch": 0.3355426677713339, "grad_norm": 2.317074544699955, "learning_rate": 8.40540517502813e-06, "loss": 0.5719, "step": 405 }, { "epoch": 0.3396851698425849, "grad_norm": 2.1839364363051357, "learning_rate": 8.35209732137463e-06, "loss": 0.5718, "step": 410 }, { "epoch": 0.34382767191383595, "grad_norm": 2.252030122787996, "learning_rate": 8.298088191438753e-06, "loss": 0.5697, "step": 415 }, { "epoch": 0.347970173985087, "grad_norm": 2.1248298616099475, "learning_rate": 8.243389084212808e-06, "loss": 0.5523, "step": 420 }, { "epoch": 0.352112676056338, "grad_norm": 2.173293611964785, "learning_rate": 8.188011443035962e-06, "loss": 0.5447, "step": 425 }, { "epoch": 0.35625517812758906, "grad_norm": 2.1954256707470075, "learning_rate": 8.131966853200226e-06, "loss": 0.549, "step": 430 }, { "epoch": 0.3603976801988401, "grad_norm": 2.149878633517254, "learning_rate": 8.075267039526764e-06, "loss": 0.5499, "step": 435 }, { "epoch": 0.36454018227009116, "grad_norm": 2.4205240164300315, "learning_rate": 8.017923863912989e-06, "loss": 0.5575, "step": 440 }, { "epoch": 0.36868268434134216, "grad_norm": 2.0890174160238737, "learning_rate": 7.959949322850994e-06, "loss": 0.5453, "step": 445 }, { "epoch": 0.3728251864125932, "grad_norm": 2.0946555310239887, "learning_rate": 7.901355544917827e-06, "loss": 0.5414, "step": 450 }, { "epoch": 0.37696768848384427, "grad_norm": 2.126619902644704, "learning_rate": 7.842154788238124e-06, "loss": 0.5495, "step": 455 }, { "epoch": 0.38111019055509526, "grad_norm": 2.2360988935530033, "learning_rate": 7.782359437919644e-06, "loss": 0.5438, "step": 460 }, { "epoch": 0.3852526926263463, "grad_norm": 2.1459033718890077, "learning_rate": 7.721982003462255e-06, "loss": 0.5353, "step": 465 }, { "epoch": 0.38939519469759737, "grad_norm": 2.283786387893568, "learning_rate": 7.661035116140856e-06, "loss": 0.5357, "step": 470 }, { "epoch": 0.39353769676884837, "grad_norm": 2.1816854698567, "learning_rate": 7.599531526362873e-06, "loss": 0.5449, "step": 475 }, { "epoch": 0.3976801988400994, "grad_norm": 2.2069133172922255, "learning_rate": 7.537484101000787e-06, "loss": 0.526, "step": 480 }, { "epoch": 0.40182270091135047, "grad_norm": 2.118219362380726, "learning_rate": 7.474905820700334e-06, "loss": 0.534, "step": 485 }, { "epoch": 0.40596520298260147, "grad_norm": 2.264878177122059, "learning_rate": 7.411809777164873e-06, "loss": 0.5165, "step": 490 }, { "epoch": 0.4101077050538525, "grad_norm": 2.1705389482990363, "learning_rate": 7.3482091704165405e-06, "loss": 0.5245, "step": 495 }, { "epoch": 0.4142502071251036, "grad_norm": 2.2941159685489336, "learning_rate": 7.284117306034733e-06, "loss": 0.5103, "step": 500 }, { "epoch": 0.4183927091963546, "grad_norm": 2.2324713585602254, "learning_rate": 7.219547592372512e-06, "loss": 0.5214, "step": 505 }, { "epoch": 0.4225352112676056, "grad_norm": 2.2391139429721236, "learning_rate": 7.15451353775151e-06, "loss": 0.5195, "step": 510 }, { "epoch": 0.4266777133388567, "grad_norm": 2.2449960646697513, "learning_rate": 7.089028747635908e-06, "loss": 0.4994, "step": 515 }, { "epoch": 0.43082021541010773, "grad_norm": 2.4452540985802687, "learning_rate": 7.023106921786118e-06, "loss": 0.5079, "step": 520 }, { "epoch": 0.43496271748135873, "grad_norm": 2.1476468599114065, "learning_rate": 6.956761851392706e-06, "loss": 0.5122, "step": 525 }, { "epoch": 0.4391052195526098, "grad_norm": 2.216754671617839, "learning_rate": 6.890007416191209e-06, "loss": 0.508, "step": 530 }, { "epoch": 0.44324772162386084, "grad_norm": 2.0655580498022537, "learning_rate": 6.822857581558423e-06, "loss": 0.5005, "step": 535 }, { "epoch": 0.44739022369511183, "grad_norm": 2.1974636316633975, "learning_rate": 6.7553263955907755e-06, "loss": 0.4997, "step": 540 }, { "epoch": 0.4515327257663629, "grad_norm": 2.3612483703811127, "learning_rate": 6.687427986165379e-06, "loss": 0.4912, "step": 545 }, { "epoch": 0.45567522783761394, "grad_norm": 2.1051086338026055, "learning_rate": 6.6191765579844205e-06, "loss": 0.4958, "step": 550 }, { "epoch": 0.45981772990886494, "grad_norm": 2.1903547319110013, "learning_rate": 6.550586389603451e-06, "loss": 0.4876, "step": 555 }, { "epoch": 0.463960231980116, "grad_norm": 2.2727306239643523, "learning_rate": 6.481671830444243e-06, "loss": 0.4843, "step": 560 }, { "epoch": 0.46810273405136704, "grad_norm": 2.111227587642084, "learning_rate": 6.412447297792818e-06, "loss": 0.4716, "step": 565 }, { "epoch": 0.47224523612261804, "grad_norm": 2.205352360925359, "learning_rate": 6.3429272737832726e-06, "loss": 0.4784, "step": 570 }, { "epoch": 0.4763877381938691, "grad_norm": 2.16751009903555, "learning_rate": 6.273126302368037e-06, "loss": 0.487, "step": 575 }, { "epoch": 0.48053024026512015, "grad_norm": 2.1849501403147404, "learning_rate": 6.203058986275207e-06, "loss": 0.4727, "step": 580 }, { "epoch": 0.48467274233637114, "grad_norm": 2.2295368151988284, "learning_rate": 6.132739983953579e-06, "loss": 0.4687, "step": 585 }, { "epoch": 0.4888152444076222, "grad_norm": 2.101227396044922, "learning_rate": 6.062184006506027e-06, "loss": 0.4775, "step": 590 }, { "epoch": 0.49295774647887325, "grad_norm": 2.248193658880927, "learning_rate": 5.991405814611855e-06, "loss": 0.4665, "step": 595 }, { "epoch": 0.4971002485501243, "grad_norm": 2.2704834915264662, "learning_rate": 5.920420215438794e-06, "loss": 0.4789, "step": 600 }, { "epoch": 0.5012427506213754, "grad_norm": 2.32381879564274, "learning_rate": 5.849242059545259e-06, "loss": 0.4697, "step": 605 }, { "epoch": 0.5053852526926264, "grad_norm": 2.2576290726260373, "learning_rate": 5.777886237773542e-06, "loss": 0.4593, "step": 610 }, { "epoch": 0.5095277547638773, "grad_norm": 2.135648372066762, "learning_rate": 5.706367678134562e-06, "loss": 0.4668, "step": 615 }, { "epoch": 0.5136702568351285, "grad_norm": 2.07573918775664, "learning_rate": 5.634701342684852e-06, "loss": 0.4632, "step": 620 }, { "epoch": 0.5178127589063795, "grad_norm": 2.1248245211465244, "learning_rate": 5.562902224396416e-06, "loss": 0.4595, "step": 625 }, { "epoch": 0.5219552609776305, "grad_norm": 2.130098034424365, "learning_rate": 5.49098534402012e-06, "loss": 0.4598, "step": 630 }, { "epoch": 0.5260977630488816, "grad_norm": 2.334633169152005, "learning_rate": 5.418965746943281e-06, "loss": 0.4559, "step": 635 }, { "epoch": 0.5302402651201326, "grad_norm": 2.1778621060776193, "learning_rate": 5.34685850004208e-06, "loss": 0.4513, "step": 640 }, { "epoch": 0.5343827671913836, "grad_norm": 2.168236740017704, "learning_rate": 5.2746786885295034e-06, "loss": 0.4556, "step": 645 }, { "epoch": 0.5385252692626347, "grad_norm": 1.993317342719062, "learning_rate": 5.2024414127994325e-06, "loss": 0.4462, "step": 650 }, { "epoch": 0.5426677713338857, "grad_norm": 2.126792629247378, "learning_rate": 5.13016178526756e-06, "loss": 0.4365, "step": 655 }, { "epoch": 0.5468102734051367, "grad_norm": 2.1107942879381727, "learning_rate": 5.057854927209804e-06, "loss": 0.4434, "step": 660 }, { "epoch": 0.5509527754763878, "grad_norm": 2.1123391684515815, "learning_rate": 4.985535965598843e-06, "loss": 0.4489, "step": 665 }, { "epoch": 0.5550952775476388, "grad_norm": 2.20383507837916, "learning_rate": 4.913220029939491e-06, "loss": 0.4426, "step": 670 }, { "epoch": 0.5592377796188898, "grad_norm": 2.024690328428114, "learning_rate": 4.840922249103506e-06, "loss": 0.4372, "step": 675 }, { "epoch": 0.5633802816901409, "grad_norm": 2.1658026493211713, "learning_rate": 4.7686577481645745e-06, "loss": 0.4344, "step": 680 }, { "epoch": 0.5675227837613919, "grad_norm": 2.130458226840472, "learning_rate": 4.696441645234042e-06, "loss": 0.4423, "step": 685 }, { "epoch": 0.5716652858326429, "grad_norm": 2.130335330283906, "learning_rate": 4.624289048298147e-06, "loss": 0.4324, "step": 690 }, { "epoch": 0.575807787903894, "grad_norm": 2.0262507094150077, "learning_rate": 4.55221505205734e-06, "loss": 0.4338, "step": 695 }, { "epoch": 0.579950289975145, "grad_norm": 2.0666888969605277, "learning_rate": 4.480234734768393e-06, "loss": 0.4286, "step": 700 }, { "epoch": 0.584092792046396, "grad_norm": 2.0542477917888453, "learning_rate": 4.408363155089952e-06, "loss": 0.4307, "step": 705 }, { "epoch": 0.5882352941176471, "grad_norm": 2.1776472324502327, "learning_rate": 4.3366153489321855e-06, "loss": 0.4344, "step": 710 }, { "epoch": 0.5923777961888981, "grad_norm": 2.0982654105842475, "learning_rate": 4.265006326311199e-06, "loss": 0.418, "step": 715 }, { "epoch": 0.5965202982601492, "grad_norm": 2.0399148247049363, "learning_rate": 4.1935510682088545e-06, "loss": 0.422, "step": 720 }, { "epoch": 0.6006628003314002, "grad_norm": 2.132392353626558, "learning_rate": 4.122264523438668e-06, "loss": 0.4122, "step": 725 }, { "epoch": 0.6048053024026512, "grad_norm": 2.0436310042803094, "learning_rate": 4.051161605518453e-06, "loss": 0.4233, "step": 730 }, { "epoch": 0.6089478044739023, "grad_norm": 2.0217278892738504, "learning_rate": 3.980257189550316e-06, "loss": 0.4226, "step": 735 }, { "epoch": 0.6130903065451533, "grad_norm": 1.8834806774575015, "learning_rate": 3.909566109108727e-06, "loss": 0.411, "step": 740 }, { "epoch": 0.6172328086164043, "grad_norm": 2.158825355814557, "learning_rate": 3.839103153137247e-06, "loss": 0.4175, "step": 745 }, { "epoch": 0.6213753106876554, "grad_norm": 2.1656769542527363, "learning_rate": 3.768883062854598e-06, "loss": 0.4178, "step": 750 }, { "epoch": 0.6255178127589064, "grad_norm": 2.028679198499099, "learning_rate": 3.6989205286707398e-06, "loss": 0.4169, "step": 755 }, { "epoch": 0.6296603148301574, "grad_norm": 2.0638901993850087, "learning_rate": 3.6292301871135425e-06, "loss": 0.4074, "step": 760 }, { "epoch": 0.6338028169014085, "grad_norm": 2.1149502198751344, "learning_rate": 3.55982661776676e-06, "loss": 0.4126, "step": 765 }, { "epoch": 0.6379453189726595, "grad_norm": 2.242329346685076, "learning_rate": 3.4907243402199013e-06, "loss": 0.4165, "step": 770 }, { "epoch": 0.6420878210439105, "grad_norm": 2.1066940525003326, "learning_rate": 3.4219378110306523e-06, "loss": 0.3997, "step": 775 }, { "epoch": 0.6462303231151616, "grad_norm": 2.169707690258044, "learning_rate": 3.353481420700495e-06, "loss": 0.4104, "step": 780 }, { "epoch": 0.6503728251864126, "grad_norm": 2.098473714909255, "learning_rate": 3.285369490664133e-06, "loss": 0.3956, "step": 785 }, { "epoch": 0.6545153272576636, "grad_norm": 2.1657310992501295, "learning_rate": 3.2176162702933816e-06, "loss": 0.4135, "step": 790 }, { "epoch": 0.6586578293289147, "grad_norm": 2.1452143463915307, "learning_rate": 3.150235933916115e-06, "loss": 0.3934, "step": 795 }, { "epoch": 0.6628003314001657, "grad_norm": 2.052594101769282, "learning_rate": 3.0832425778509235e-06, "loss": 0.3979, "step": 800 }, { "epoch": 0.6669428334714167, "grad_norm": 2.0098945834829824, "learning_rate": 3.0166502174581012e-06, "loss": 0.3967, "step": 805 }, { "epoch": 0.6710853355426678, "grad_norm": 2.042954604946958, "learning_rate": 2.950472784207544e-06, "loss": 0.3945, "step": 810 }, { "epoch": 0.6752278376139188, "grad_norm": 2.0032403215865506, "learning_rate": 2.8847241227642255e-06, "loss": 0.3967, "step": 815 }, { "epoch": 0.6793703396851698, "grad_norm": 2.1221545047500214, "learning_rate": 2.819417988091814e-06, "loss": 0.3868, "step": 820 }, { "epoch": 0.6835128417564209, "grad_norm": 2.130884956811238, "learning_rate": 2.754568042575061e-06, "loss": 0.3922, "step": 825 }, { "epoch": 0.6876553438276719, "grad_norm": 1.9774695256683483, "learning_rate": 2.6901878531615677e-06, "loss": 0.3901, "step": 830 }, { "epoch": 0.6917978458989229, "grad_norm": 2.1120886925275055, "learning_rate": 2.6262908885235046e-06, "loss": 0.3941, "step": 835 }, { "epoch": 0.695940347970174, "grad_norm": 2.045887267583242, "learning_rate": 2.5628905162398797e-06, "loss": 0.3959, "step": 840 }, { "epoch": 0.700082850041425, "grad_norm": 1.8783748619115133, "learning_rate": 2.5000000000000015e-06, "loss": 0.3806, "step": 845 }, { "epoch": 0.704225352112676, "grad_norm": 2.08098731426819, "learning_rate": 2.4376324968286154e-06, "loss": 0.387, "step": 850 }, { "epoch": 0.7083678541839271, "grad_norm": 2.1586747995562705, "learning_rate": 2.375801054333409e-06, "loss": 0.3729, "step": 855 }, { "epoch": 0.7125103562551781, "grad_norm": 2.0279143582578616, "learning_rate": 2.3145186079753685e-06, "loss": 0.3758, "step": 860 }, { "epoch": 0.7166528583264291, "grad_norm": 2.2429697814696854, "learning_rate": 2.253797978362617e-06, "loss": 0.3786, "step": 865 }, { "epoch": 0.7207953603976802, "grad_norm": 2.0360034426541556, "learning_rate": 2.193651868568285e-06, "loss": 0.377, "step": 870 }, { "epoch": 0.7249378624689312, "grad_norm": 1.9725681488817448, "learning_rate": 2.1340928614729445e-06, "loss": 0.3721, "step": 875 }, { "epoch": 0.7290803645401823, "grad_norm": 2.103021145158836, "learning_rate": 2.075133417132223e-06, "loss": 0.3719, "step": 880 }, { "epoch": 0.7332228666114333, "grad_norm": 2.092900322935167, "learning_rate": 2.016785870170079e-06, "loss": 0.3714, "step": 885 }, { "epoch": 0.7373653686826843, "grad_norm": 2.0906479452415336, "learning_rate": 1.9590624271983406e-06, "loss": 0.3749, "step": 890 }, { "epoch": 0.7415078707539354, "grad_norm": 2.096208214143563, "learning_rate": 1.9019751642630252e-06, "loss": 0.37, "step": 895 }, { "epoch": 0.7456503728251864, "grad_norm": 2.0164719015170656, "learning_rate": 1.8455360243179537e-06, "loss": 0.3724, "step": 900 }, { "epoch": 0.7497928748964374, "grad_norm": 1.9827943207162666, "learning_rate": 1.7897568147262323e-06, "loss": 0.3747, "step": 905 }, { "epoch": 0.7539353769676885, "grad_norm": 1.9263424969881737, "learning_rate": 1.7346492047900897e-06, "loss": 0.3635, "step": 910 }, { "epoch": 0.7580778790389395, "grad_norm": 2.2195829689186106, "learning_rate": 1.6802247233095914e-06, "loss": 0.3667, "step": 915 }, { "epoch": 0.7622203811101905, "grad_norm": 2.0103741607119185, "learning_rate": 1.626494756170765e-06, "loss": 0.3692, "step": 920 }, { "epoch": 0.7663628831814416, "grad_norm": 2.1097047260728963, "learning_rate": 1.5734705439636017e-06, "loss": 0.365, "step": 925 }, { "epoch": 0.7705053852526926, "grad_norm": 2.036839326608303, "learning_rate": 1.5211631796304721e-06, "loss": 0.3648, "step": 930 }, { "epoch": 0.7746478873239436, "grad_norm": 1.9414201763962056, "learning_rate": 1.46958360614543e-06, "loss": 0.3656, "step": 935 }, { "epoch": 0.7787903893951947, "grad_norm": 2.047677571212274, "learning_rate": 1.4187426142248723e-06, "loss": 0.3542, "step": 940 }, { "epoch": 0.7829328914664457, "grad_norm": 2.06103898039325, "learning_rate": 1.3686508400700787e-06, "loss": 0.3646, "step": 945 }, { "epoch": 0.7870753935376967, "grad_norm": 2.0587768499958394, "learning_rate": 1.3193187631420462e-06, "loss": 0.3579, "step": 950 }, { "epoch": 0.7912178956089478, "grad_norm": 1.9792392112511976, "learning_rate": 1.2707567039691505e-06, "loss": 0.3607, "step": 955 }, { "epoch": 0.7953603976801988, "grad_norm": 1.9865516347697185, "learning_rate": 1.222974821988024e-06, "loss": 0.3588, "step": 960 }, { "epoch": 0.7995028997514498, "grad_norm": 2.0887989420586055, "learning_rate": 1.1759831134181504e-06, "loss": 0.3604, "step": 965 }, { "epoch": 0.8036454018227009, "grad_norm": 2.1350401974441575, "learning_rate": 1.1297914091706086e-06, "loss": 0.3578, "step": 970 }, { "epoch": 0.8077879038939519, "grad_norm": 2.0023054097994417, "learning_rate": 1.0844093727913868e-06, "loss": 0.352, "step": 975 }, { "epoch": 0.8119304059652029, "grad_norm": 2.006501689438896, "learning_rate": 1.039846498439727e-06, "loss": 0.3508, "step": 980 }, { "epoch": 0.816072908036454, "grad_norm": 2.1675559938665017, "learning_rate": 9.961121089018933e-07, "loss": 0.3513, "step": 985 }, { "epoch": 0.820215410107705, "grad_norm": 1.9789546403226348, "learning_rate": 9.532153536407923e-07, "loss": 0.3548, "step": 990 }, { "epoch": 0.824357912178956, "grad_norm": 2.2136680684771215, "learning_rate": 9.111652068818621e-07, "loss": 0.3515, "step": 995 }, { "epoch": 0.8285004142502072, "grad_norm": 2.0985287351131183, "learning_rate": 8.699704657356195e-07, "loss": 0.3494, "step": 1000 }, { "epoch": 0.8326429163214581, "grad_norm": 2.03602744057189, "learning_rate": 8.296397483572515e-07, "loss": 0.3502, "step": 1005 }, { "epoch": 0.8367854183927091, "grad_norm": 1.9529410237196838, "learning_rate": 7.901814921436624e-07, "loss": 0.3412, "step": 1010 }, { "epoch": 0.8409279204639603, "grad_norm": 2.056866748380538, "learning_rate": 7.516039519683105e-07, "loss": 0.3451, "step": 1015 }, { "epoch": 0.8450704225352113, "grad_norm": 2.1162292783700596, "learning_rate": 7.139151984542636e-07, "loss": 0.346, "step": 1020 }, { "epoch": 0.8492129246064622, "grad_norm": 2.199566140011582, "learning_rate": 6.771231162857722e-07, "loss": 0.3503, "step": 1025 }, { "epoch": 0.8533554266777134, "grad_norm": 2.071853569310024, "learning_rate": 6.412354025587509e-07, "loss": 0.3394, "step": 1030 }, { "epoch": 0.8574979287489644, "grad_norm": 1.9821516702262831, "learning_rate": 6.062595651705111e-07, "loss": 0.3599, "step": 1035 }, { "epoch": 0.8616404308202155, "grad_norm": 2.03084885183303, "learning_rate": 5.722029212490666e-07, "loss": 0.3473, "step": 1040 }, { "epoch": 0.8657829328914665, "grad_norm": 1.9891381030542377, "learning_rate": 5.390725956223531e-07, "loss": 0.3399, "step": 1045 }, { "epoch": 0.8699254349627175, "grad_norm": 1.996858017795371, "learning_rate": 5.068755193276798e-07, "loss": 0.3438, "step": 1050 }, { "epoch": 0.8740679370339686, "grad_norm": 2.0102978075746436, "learning_rate": 4.756184281617121e-07, "loss": 0.3596, "step": 1055 }, { "epoch": 0.8782104391052196, "grad_norm": 2.096782816191255, "learning_rate": 4.4530786127131575e-07, "loss": 0.3426, "step": 1060 }, { "epoch": 0.8823529411764706, "grad_norm": 2.041538694858829, "learning_rate": 4.159501597855287e-07, "loss": 0.3469, "step": 1065 }, { "epoch": 0.8864954432477217, "grad_norm": 2.050624816044249, "learning_rate": 3.8755146548896784e-07, "loss": 0.3453, "step": 1070 }, { "epoch": 0.8906379453189727, "grad_norm": 2.134183615956234, "learning_rate": 3.6011771953693044e-07, "loss": 0.3383, "step": 1075 }, { "epoch": 0.8947804473902237, "grad_norm": 2.0340883063292456, "learning_rate": 3.336546612124758e-07, "loss": 0.3394, "step": 1080 }, { "epoch": 0.8989229494614748, "grad_norm": 2.142834891831593, "learning_rate": 3.081678267257404e-07, "loss": 0.3377, "step": 1085 }, { "epoch": 0.9030654515327258, "grad_norm": 2.1312482485517608, "learning_rate": 2.836625480557265e-07, "loss": 0.3347, "step": 1090 }, { "epoch": 0.9072079536039768, "grad_norm": 2.076260403631579, "learning_rate": 2.601439518348331e-07, "loss": 0.3408, "step": 1095 }, { "epoch": 0.9113504556752279, "grad_norm": 2.063238043755945, "learning_rate": 2.376169582763288e-07, "loss": 0.347, "step": 1100 }, { "epoch": 0.9154929577464789, "grad_norm": 2.045149868081977, "learning_rate": 2.1608628014502364e-07, "loss": 0.338, "step": 1105 }, { "epoch": 0.9196354598177299, "grad_norm": 2.0728146587926535, "learning_rate": 1.955564217713335e-07, "loss": 0.3385, "step": 1110 }, { "epoch": 0.923777961888981, "grad_norm": 1.980028386143511, "learning_rate": 1.7603167810894662e-07, "loss": 0.3397, "step": 1115 }, { "epoch": 0.927920463960232, "grad_norm": 2.040384287235922, "learning_rate": 1.5751613383630128e-07, "loss": 0.337, "step": 1120 }, { "epoch": 0.932062966031483, "grad_norm": 2.0668331131661337, "learning_rate": 1.4001366250204762e-07, "loss": 0.3432, "step": 1125 }, { "epoch": 0.9362054681027341, "grad_norm": 2.1215696014539636, "learning_rate": 1.235279257146804e-07, "loss": 0.348, "step": 1130 }, { "epoch": 0.9403479701739851, "grad_norm": 2.0968064514077187, "learning_rate": 1.080623723765134e-07, "loss": 0.3351, "step": 1135 }, { "epoch": 0.9444904722452361, "grad_norm": 2.0386634215167305, "learning_rate": 9.362023796215036e-08, "loss": 0.335, "step": 1140 }, { "epoch": 0.9486329743164872, "grad_norm": 2.081961716266303, "learning_rate": 8.020454384160437e-08, "loss": 0.345, "step": 1145 }, { "epoch": 0.9527754763877382, "grad_norm": 2.0627875473620443, "learning_rate": 6.78180966482156e-08, "loss": 0.3367, "step": 1150 }, { "epoch": 0.9569179784589892, "grad_norm": 2.0611893408248587, "learning_rate": 5.646348769148491e-08, "loss": 0.3388, "step": 1155 }, { "epoch": 0.9610604805302403, "grad_norm": 2.1119589491438315, "learning_rate": 4.6143092414961396e-08, "loss": 0.3406, "step": 1160 }, { "epoch": 0.9652029826014913, "grad_norm": 2.0229820894857866, "learning_rate": 3.685906989928656e-08, "loss": 0.3446, "step": 1165 }, { "epoch": 0.9693454846727423, "grad_norm": 1.9950176331137996, "learning_rate": 2.861336241050061e-08, "loss": 0.3361, "step": 1170 }, { "epoch": 0.9734879867439934, "grad_norm": 2.051002382020688, "learning_rate": 2.1407694993714755e-08, "loss": 0.3352, "step": 1175 }, { "epoch": 0.9776304888152444, "grad_norm": 2.0952467848619967, "learning_rate": 1.5243575112218744e-08, "loss": 0.3278, "step": 1180 }, { "epoch": 0.9817729908864954, "grad_norm": 2.1473598929894164, "learning_rate": 1.0122292332114814e-08, "loss": 0.3386, "step": 1185 }, { "epoch": 0.9859154929577465, "grad_norm": 2.027802609236831, "learning_rate": 6.044918052531268e-09, "loss": 0.3429, "step": 1190 }, { "epoch": 0.9900579950289975, "grad_norm": 2.088880428637403, "learning_rate": 3.0123052814812203e-09, "loss": 0.3452, "step": 1195 }, { "epoch": 0.9942004971002486, "grad_norm": 2.1283431853256705, "learning_rate": 1.025088457409229e-09, "loss": 0.3321, "step": 1200 }, { "epoch": 0.9983429991714996, "grad_norm": 1.953227804218852, "learning_rate": 8.368331646302353e-11, "loss": 0.3403, "step": 1205 }, { "epoch": 1.0, "eval_loss": 0.33688703179359436, "eval_runtime": 1.1942, "eval_samples_per_second": 2.512, "eval_steps_per_second": 0.837, "step": 1207 }, { "epoch": 1.0, "step": 1207, "total_flos": 252668899491840.0, "train_loss": 0.5252359404283411, "train_runtime": 27071.2973, "train_samples_per_second": 1.427, "train_steps_per_second": 0.045 } ], "logging_steps": 5, "max_steps": 1207, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 100, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 252668899491840.0, "train_batch_size": 2, "trial_name": null, "trial_params": null }