{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.2754977651361235, "eval_steps": 500, "global_step": 700, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.016253555465258026, "grad_norm": 2.077186346054077, "learning_rate": 4.9998691031433496e-05, "loss": 2.8778, "step": 5 }, { "epoch": 0.03250711093051605, "grad_norm": 2.073504686355591, "learning_rate": 4.999476426280588e-05, "loss": 2.6432, "step": 10 }, { "epoch": 0.048760666395774074, "grad_norm": 2.1281931400299072, "learning_rate": 4.998822010531848e-05, "loss": 2.3167, "step": 15 }, { "epoch": 0.0650142218610321, "grad_norm": 1.090219497680664, "learning_rate": 4.997905924425903e-05, "loss": 2.1533, "step": 20 }, { "epoch": 0.08126777732629012, "grad_norm": 1.007896900177002, "learning_rate": 4.996728263892985e-05, "loss": 2.0267, "step": 25 }, { "epoch": 0.09752133279154815, "grad_norm": 0.9982665777206421, "learning_rate": 4.995289152254744e-05, "loss": 1.9352, "step": 30 }, { "epoch": 0.11377488825680618, "grad_norm": 0.8844298720359802, "learning_rate": 4.9935887402113315e-05, "loss": 1.9486, "step": 35 }, { "epoch": 0.1300284437220642, "grad_norm": 0.9337536692619324, "learning_rate": 4.991627205825621e-05, "loss": 1.9228, "step": 40 }, { "epoch": 0.14628199918732224, "grad_norm": 0.9377800822257996, "learning_rate": 4.9894047545045605e-05, "loss": 1.835, "step": 45 }, { "epoch": 0.16253555465258024, "grad_norm": 0.8525241017341614, "learning_rate": 4.986921618977664e-05, "loss": 1.8157, "step": 50 }, { "epoch": 0.17878911011783827, "grad_norm": 0.872871458530426, "learning_rate": 4.984178059272638e-05, "loss": 1.8811, "step": 55 }, { "epoch": 0.1950426655830963, "grad_norm": 0.9125804305076599, "learning_rate": 4.981174362688158e-05, "loss": 1.8242, "step": 60 }, { "epoch": 0.21129622104835433, "grad_norm": 0.7474733591079712, "learning_rate": 4.977910843763777e-05, "loss": 1.7808, "step": 65 }, { "epoch": 0.22754977651361236, "grad_norm": 1.0248199701309204, "learning_rate": 4.974387844246987e-05, "loss": 1.8512, "step": 70 }, { "epoch": 0.24380333197887039, "grad_norm": 0.9032835960388184, "learning_rate": 4.970605733057441e-05, "loss": 1.8172, "step": 75 }, { "epoch": 0.2600568874441284, "grad_norm": 0.8550340533256531, "learning_rate": 4.9665649062483115e-05, "loss": 1.7418, "step": 80 }, { "epoch": 0.27631044290938644, "grad_norm": 0.9429016709327698, "learning_rate": 4.96226578696482e-05, "loss": 1.8265, "step": 85 }, { "epoch": 0.2925639983746445, "grad_norm": 0.975885272026062, "learning_rate": 4.957708825399927e-05, "loss": 1.7943, "step": 90 }, { "epoch": 0.3088175538399025, "grad_norm": 0.827629804611206, "learning_rate": 4.9528944987471884e-05, "loss": 1.8067, "step": 95 }, { "epoch": 0.3250711093051605, "grad_norm": 1.0019093751907349, "learning_rate": 4.9478233111507856e-05, "loss": 1.7874, "step": 100 }, { "epoch": 0.3413246647704185, "grad_norm": 1.0811312198638916, "learning_rate": 4.9424957936527295e-05, "loss": 1.7395, "step": 105 }, { "epoch": 0.35757822023567654, "grad_norm": 0.9461565613746643, "learning_rate": 4.936912504137257e-05, "loss": 1.7833, "step": 110 }, { "epoch": 0.37383177570093457, "grad_norm": 1.0386009216308594, "learning_rate": 4.9310740272724055e-05, "loss": 1.7569, "step": 115 }, { "epoch": 0.3900853311661926, "grad_norm": 0.9916568994522095, "learning_rate": 4.924980974448791e-05, "loss": 1.7584, "step": 120 }, { "epoch": 0.4063388866314506, "grad_norm": 1.240400791168213, "learning_rate": 4.918633983715581e-05, "loss": 1.7438, "step": 125 }, { "epoch": 0.42259244209670865, "grad_norm": 1.1441287994384766, "learning_rate": 4.912033719713687e-05, "loss": 1.7046, "step": 130 }, { "epoch": 0.4388459975619667, "grad_norm": 1.1687984466552734, "learning_rate": 4.905180873606157e-05, "loss": 1.738, "step": 135 }, { "epoch": 0.4550995530272247, "grad_norm": 0.9126319885253906, "learning_rate": 4.8980761630058014e-05, "loss": 1.7738, "step": 140 }, { "epoch": 0.47135310849248274, "grad_norm": 1.172781229019165, "learning_rate": 4.8907203319000495e-05, "loss": 1.6642, "step": 145 }, { "epoch": 0.48760666395774077, "grad_norm": 1.1217776536941528, "learning_rate": 4.883114150573037e-05, "loss": 1.7359, "step": 150 }, { "epoch": 0.5038602194229987, "grad_norm": 1.1024829149246216, "learning_rate": 4.8752584155249444e-05, "loss": 1.6707, "step": 155 }, { "epoch": 0.5201137748882568, "grad_norm": 1.042017936706543, "learning_rate": 4.8671539493885934e-05, "loss": 1.709, "step": 160 }, { "epoch": 0.5363673303535148, "grad_norm": 1.1834771633148193, "learning_rate": 4.8588016008432945e-05, "loss": 1.7282, "step": 165 }, { "epoch": 0.5526208858187729, "grad_norm": 1.0311099290847778, "learning_rate": 4.850202244525987e-05, "loss": 1.774, "step": 170 }, { "epoch": 0.5688744412840309, "grad_norm": 1.2995566129684448, "learning_rate": 4.8413567809396376e-05, "loss": 1.714, "step": 175 }, { "epoch": 0.585127996749289, "grad_norm": 1.23491370677948, "learning_rate": 4.8322661363589507e-05, "loss": 1.7717, "step": 180 }, { "epoch": 0.6013815522145469, "grad_norm": 1.1809489727020264, "learning_rate": 4.822931262733367e-05, "loss": 1.722, "step": 185 }, { "epoch": 0.617635107679805, "grad_norm": 0.988572359085083, "learning_rate": 4.813353137587377e-05, "loss": 1.7007, "step": 190 }, { "epoch": 0.633888663145063, "grad_norm": 1.0803310871124268, "learning_rate": 4.803532763918162e-05, "loss": 1.6824, "step": 195 }, { "epoch": 0.650142218610321, "grad_norm": 1.02713143825531, "learning_rate": 4.793471170090555e-05, "loss": 1.6453, "step": 200 }, { "epoch": 0.666395774075579, "grad_norm": 1.2023831605911255, "learning_rate": 4.783169409729362e-05, "loss": 1.6974, "step": 205 }, { "epoch": 0.682649329540837, "grad_norm": 1.0009326934814453, "learning_rate": 4.772628561609022e-05, "loss": 1.7094, "step": 210 }, { "epoch": 0.6989028850060951, "grad_norm": 1.269020438194275, "learning_rate": 4.761849729540643e-05, "loss": 1.6832, "step": 215 }, { "epoch": 0.7151564404713531, "grad_norm": 1.3129128217697144, "learning_rate": 4.7508340422564134e-05, "loss": 1.6847, "step": 220 }, { "epoch": 0.7314099959366112, "grad_norm": 1.0837377309799194, "learning_rate": 4.7395826532914054e-05, "loss": 1.7215, "step": 225 }, { "epoch": 0.7476635514018691, "grad_norm": 1.1211832761764526, "learning_rate": 4.728096740862778e-05, "loss": 1.7658, "step": 230 }, { "epoch": 0.7639171068671272, "grad_norm": 1.1422042846679688, "learning_rate": 4.716377507746397e-05, "loss": 1.7098, "step": 235 }, { "epoch": 0.7801706623323852, "grad_norm": 1.2925424575805664, "learning_rate": 4.704426181150884e-05, "loss": 1.7504, "step": 240 }, { "epoch": 0.7964242177976433, "grad_norm": 1.282771348953247, "learning_rate": 4.6922440125891064e-05, "loss": 1.72, "step": 245 }, { "epoch": 0.8126777732629012, "grad_norm": 1.2494529485702515, "learning_rate": 4.6798322777471216e-05, "loss": 1.6581, "step": 250 }, { "epoch": 0.8289313287281593, "grad_norm": 1.2736073732376099, "learning_rate": 4.667192276350591e-05, "loss": 1.6825, "step": 255 }, { "epoch": 0.8451848841934173, "grad_norm": 1.1477563381195068, "learning_rate": 4.654325332028676e-05, "loss": 1.6889, "step": 260 }, { "epoch": 0.8614384396586753, "grad_norm": 1.2415803670883179, "learning_rate": 4.641232792175428e-05, "loss": 1.6586, "step": 265 }, { "epoch": 0.8776919951239334, "grad_norm": 1.3159024715423584, "learning_rate": 4.6279160278086994e-05, "loss": 1.7102, "step": 270 }, { "epoch": 0.8939455505891913, "grad_norm": 1.2742581367492676, "learning_rate": 4.614376433426565e-05, "loss": 1.7476, "step": 275 }, { "epoch": 0.9101991060544494, "grad_norm": 1.34221351146698, "learning_rate": 4.6006154268613014e-05, "loss": 1.7497, "step": 280 }, { "epoch": 0.9264526615197074, "grad_norm": 1.1669361591339111, "learning_rate": 4.586634449130911e-05, "loss": 1.6693, "step": 285 }, { "epoch": 0.9427062169849655, "grad_norm": 1.2765443325042725, "learning_rate": 4.572434964288226e-05, "loss": 1.5862, "step": 290 }, { "epoch": 0.9589597724502235, "grad_norm": 1.2429810762405396, "learning_rate": 4.55801845926759e-05, "loss": 1.6875, "step": 295 }, { "epoch": 0.9752133279154815, "grad_norm": 1.2103233337402344, "learning_rate": 4.543386443729157e-05, "loss": 1.7294, "step": 300 }, { "epoch": 0.9914668833807395, "grad_norm": 1.3512628078460693, "learning_rate": 4.528540449900798e-05, "loss": 1.709, "step": 305 }, { "epoch": 1.0077204388459975, "grad_norm": 1.2039848566055298, "learning_rate": 4.513482032417656e-05, "loss": 1.6632, "step": 310 }, { "epoch": 1.0239739943112556, "grad_norm": 1.3805475234985352, "learning_rate": 4.498212768159341e-05, "loss": 1.695, "step": 315 }, { "epoch": 1.0402275497765137, "grad_norm": 1.3265151977539062, "learning_rate": 4.48273425608481e-05, "loss": 1.6653, "step": 320 }, { "epoch": 1.0564811052417715, "grad_norm": 1.015641212463379, "learning_rate": 4.4670481170649214e-05, "loss": 1.652, "step": 325 }, { "epoch": 1.0727346607070296, "grad_norm": 1.2359689474105835, "learning_rate": 4.451155993712711e-05, "loss": 1.6457, "step": 330 }, { "epoch": 1.0889882161722877, "grad_norm": 1.3188092708587646, "learning_rate": 4.435059550211371e-05, "loss": 1.6512, "step": 335 }, { "epoch": 1.1052417716375458, "grad_norm": 1.323879361152649, "learning_rate": 4.4187604721399877e-05, "loss": 1.6338, "step": 340 }, { "epoch": 1.1214953271028036, "grad_norm": 1.263047218322754, "learning_rate": 4.40226046629703e-05, "loss": 1.6508, "step": 345 }, { "epoch": 1.1377488825680617, "grad_norm": 1.265649676322937, "learning_rate": 4.3855612605216174e-05, "loss": 1.6447, "step": 350 }, { "epoch": 1.1540024380333198, "grad_norm": 1.2379087209701538, "learning_rate": 4.3686646035125855e-05, "loss": 1.6522, "step": 355 }, { "epoch": 1.170255993498578, "grad_norm": 1.200422763824463, "learning_rate": 4.351572264645366e-05, "loss": 1.6935, "step": 360 }, { "epoch": 1.1865095489638358, "grad_norm": 1.3982837200164795, "learning_rate": 4.334286033786704e-05, "loss": 1.6792, "step": 365 }, { "epoch": 1.2027631044290938, "grad_norm": 1.3108696937561035, "learning_rate": 4.3168077211072263e-05, "loss": 1.6498, "step": 370 }, { "epoch": 1.219016659894352, "grad_norm": 1.3713555335998535, "learning_rate": 4.2991391568918825e-05, "loss": 1.6675, "step": 375 }, { "epoch": 1.23527021535961, "grad_norm": 1.406441330909729, "learning_rate": 4.281282191348289e-05, "loss": 1.6328, "step": 380 }, { "epoch": 1.2515237708248679, "grad_norm": 1.3209528923034668, "learning_rate": 4.2632386944129707e-05, "loss": 1.7064, "step": 385 }, { "epoch": 1.267777326290126, "grad_norm": 1.4870327711105347, "learning_rate": 4.245010555555554e-05, "loss": 1.7161, "step": 390 }, { "epoch": 1.284030881755384, "grad_norm": 1.4530566930770874, "learning_rate": 4.2265996835809016e-05, "loss": 1.6267, "step": 395 }, { "epoch": 1.300284437220642, "grad_norm": 1.4400850534439087, "learning_rate": 4.2080080064292304e-05, "loss": 1.6029, "step": 400 }, { "epoch": 1.3165379926859, "grad_norm": 1.389769434928894, "learning_rate": 4.189237470974219e-05, "loss": 1.6298, "step": 405 }, { "epoch": 1.332791548151158, "grad_norm": 1.2579468488693237, "learning_rate": 4.1702900428191374e-05, "loss": 1.6965, "step": 410 }, { "epoch": 1.3490451036164162, "grad_norm": 1.4243718385696411, "learning_rate": 4.151167706091017e-05, "loss": 1.5955, "step": 415 }, { "epoch": 1.3652986590816742, "grad_norm": 1.395219087600708, "learning_rate": 4.131872463232872e-05, "loss": 1.6369, "step": 420 }, { "epoch": 1.3815522145469321, "grad_norm": 1.4617631435394287, "learning_rate": 4.1124063347940135e-05, "loss": 1.6105, "step": 425 }, { "epoch": 1.3978057700121902, "grad_norm": 1.1881122589111328, "learning_rate": 4.092771359218461e-05, "loss": 1.6372, "step": 430 }, { "epoch": 1.414059325477448, "grad_norm": 1.330495834350586, "learning_rate": 4.0729695926314815e-05, "loss": 1.5817, "step": 435 }, { "epoch": 1.4303128809427061, "grad_norm": 1.3312103748321533, "learning_rate": 4.053003108624276e-05, "loss": 1.6602, "step": 440 }, { "epoch": 1.4465664364079642, "grad_norm": 1.3933234214782715, "learning_rate": 4.03287399803684e-05, "loss": 1.63, "step": 445 }, { "epoch": 1.4628199918732223, "grad_norm": 1.600183129310608, "learning_rate": 4.0125843687390166e-05, "loss": 1.6614, "step": 450 }, { "epoch": 1.4790735473384804, "grad_norm": 1.4344749450683594, "learning_rate": 3.992136345409765e-05, "loss": 1.6492, "step": 455 }, { "epoch": 1.4953271028037383, "grad_norm": 1.4546074867248535, "learning_rate": 3.9715320693146655e-05, "loss": 1.6063, "step": 460 }, { "epoch": 1.5115806582689963, "grad_norm": 1.2904654741287231, "learning_rate": 3.9507736980817e-05, "loss": 1.6618, "step": 465 }, { "epoch": 1.5278342137342542, "grad_norm": 1.8645012378692627, "learning_rate": 3.9298634054753026e-05, "loss": 1.697, "step": 470 }, { "epoch": 1.5440877691995123, "grad_norm": 1.4428379535675049, "learning_rate": 3.908803381168732e-05, "loss": 1.6741, "step": 475 }, { "epoch": 1.5603413246647704, "grad_norm": 1.4418410062789917, "learning_rate": 3.887595830514775e-05, "loss": 1.7026, "step": 480 }, { "epoch": 1.5765948801300285, "grad_norm": 1.3783613443374634, "learning_rate": 3.8662429743148046e-05, "loss": 1.6692, "step": 485 }, { "epoch": 1.5928484355952865, "grad_norm": 1.3248358964920044, "learning_rate": 3.844747048586228e-05, "loss": 1.6315, "step": 490 }, { "epoch": 1.6091019910605446, "grad_norm": 1.3751637935638428, "learning_rate": 3.823110304328331e-05, "loss": 1.6716, "step": 495 }, { "epoch": 1.6253555465258025, "grad_norm": 1.3097206354141235, "learning_rate": 3.801335007286564e-05, "loss": 1.5766, "step": 500 }, { "epoch": 1.6416091019910606, "grad_norm": 1.402269959449768, "learning_rate": 3.779423437715274e-05, "loss": 1.6405, "step": 505 }, { "epoch": 1.6578626574563184, "grad_norm": 1.6543242931365967, "learning_rate": 3.757377890138927e-05, "loss": 1.6307, "step": 510 }, { "epoch": 1.6741162129215765, "grad_norm": 1.4377750158309937, "learning_rate": 3.7352006731118264e-05, "loss": 1.588, "step": 515 }, { "epoch": 1.6903697683868346, "grad_norm": 1.578800916671753, "learning_rate": 3.712894108976372e-05, "loss": 1.6307, "step": 520 }, { "epoch": 1.7066233238520927, "grad_norm": 1.4415676593780518, "learning_rate": 3.690460533619866e-05, "loss": 1.6854, "step": 525 }, { "epoch": 1.7228768793173508, "grad_norm": 1.5168665647506714, "learning_rate": 3.667902296229905e-05, "loss": 1.599, "step": 530 }, { "epoch": 1.7391304347826086, "grad_norm": 1.3709524869918823, "learning_rate": 3.645221759048384e-05, "loss": 1.6427, "step": 535 }, { "epoch": 1.7553839902478667, "grad_norm": 1.4154709577560425, "learning_rate": 3.622421297124122e-05, "loss": 1.5486, "step": 540 }, { "epoch": 1.7716375457131246, "grad_norm": 1.5655745267868042, "learning_rate": 3.599503298064154e-05, "loss": 1.6065, "step": 545 }, { "epoch": 1.7878911011783827, "grad_norm": 1.5691167116165161, "learning_rate": 3.576470161783712e-05, "loss": 1.6194, "step": 550 }, { "epoch": 1.8041446566436408, "grad_norm": 1.5229332447052002, "learning_rate": 3.5533243002549046e-05, "loss": 1.6312, "step": 555 }, { "epoch": 1.8203982121088988, "grad_norm": 2.0803072452545166, "learning_rate": 3.5300681372541476e-05, "loss": 1.5872, "step": 560 }, { "epoch": 1.836651767574157, "grad_norm": 1.4445246458053589, "learning_rate": 3.5067041081083496e-05, "loss": 1.6226, "step": 565 }, { "epoch": 1.852905323039415, "grad_norm": 1.559645652770996, "learning_rate": 3.483234659439889e-05, "loss": 1.5703, "step": 570 }, { "epoch": 1.8691588785046729, "grad_norm": 1.3045098781585693, "learning_rate": 3.459662248910411e-05, "loss": 1.5574, "step": 575 }, { "epoch": 1.885412433969931, "grad_norm": 1.410343050956726, "learning_rate": 3.435989344963471e-05, "loss": 1.6836, "step": 580 }, { "epoch": 1.9016659894351888, "grad_norm": 1.5801265239715576, "learning_rate": 3.41221842656604e-05, "loss": 1.6366, "step": 585 }, { "epoch": 1.917919544900447, "grad_norm": 1.440772533416748, "learning_rate": 3.3883519829489154e-05, "loss": 1.6395, "step": 590 }, { "epoch": 1.934173100365705, "grad_norm": 1.3845847845077515, "learning_rate": 3.3643925133460564e-05, "loss": 1.6139, "step": 595 }, { "epoch": 1.950426655830963, "grad_norm": 1.3879539966583252, "learning_rate": 3.3403425267328715e-05, "loss": 1.5887, "step": 600 }, { "epoch": 1.9666802112962212, "grad_norm": 1.3762508630752563, "learning_rate": 3.316204541563479e-05, "loss": 1.636, "step": 605 }, { "epoch": 1.982933766761479, "grad_norm": 1.6764074563980103, "learning_rate": 3.2919810855069864e-05, "loss": 1.6699, "step": 610 }, { "epoch": 1.9991873222267371, "grad_norm": 1.5237112045288086, "learning_rate": 3.267674695182798e-05, "loss": 1.6194, "step": 615 }, { "epoch": 2.015440877691995, "grad_norm": 1.7229068279266357, "learning_rate": 3.243287915894987e-05, "loss": 1.6077, "step": 620 }, { "epoch": 2.031694433157253, "grad_norm": 1.5503779649734497, "learning_rate": 3.218823301365754e-05, "loss": 1.6024, "step": 625 }, { "epoch": 2.047947988622511, "grad_norm": 1.584635853767395, "learning_rate": 3.1942834134680124e-05, "loss": 1.5911, "step": 630 }, { "epoch": 2.0642015440877692, "grad_norm": 1.4763562679290771, "learning_rate": 3.1696708219571126e-05, "loss": 1.6143, "step": 635 }, { "epoch": 2.0804550995530273, "grad_norm": 1.5131165981292725, "learning_rate": 3.144988104201745e-05, "loss": 1.6069, "step": 640 }, { "epoch": 2.0967086550182854, "grad_norm": 1.595162272453308, "learning_rate": 3.120237844914044e-05, "loss": 1.6128, "step": 645 }, { "epoch": 2.112962210483543, "grad_norm": 1.9360445737838745, "learning_rate": 3.095422635878923e-05, "loss": 1.6004, "step": 650 }, { "epoch": 2.129215765948801, "grad_norm": 1.4424818754196167, "learning_rate": 3.0705450756826706e-05, "loss": 1.6084, "step": 655 }, { "epoch": 2.145469321414059, "grad_norm": 1.6683636903762817, "learning_rate": 3.045607769440829e-05, "loss": 1.6063, "step": 660 }, { "epoch": 2.1617228768793173, "grad_norm": 1.9396454095840454, "learning_rate": 3.0206133285254017e-05, "loss": 1.6374, "step": 665 }, { "epoch": 2.1779764323445754, "grad_norm": 1.9183346033096313, "learning_rate": 2.9955643702913865e-05, "loss": 1.529, "step": 670 }, { "epoch": 2.1942299878098335, "grad_norm": 1.61932373046875, "learning_rate": 2.9704635178027012e-05, "loss": 1.5607, "step": 675 }, { "epoch": 2.2104835432750916, "grad_norm": 1.7259231805801392, "learning_rate": 2.9453133995574955e-05, "loss": 1.5854, "step": 680 }, { "epoch": 2.2267370987403496, "grad_norm": 1.675529956817627, "learning_rate": 2.9201166492129088e-05, "loss": 1.5563, "step": 685 }, { "epoch": 2.2429906542056073, "grad_norm": 1.628509283065796, "learning_rate": 2.8948759053092754e-05, "loss": 1.5781, "step": 690 }, { "epoch": 2.2592442096708654, "grad_norm": 1.5093809366226196, "learning_rate": 2.869593810993824e-05, "loss": 1.5759, "step": 695 }, { "epoch": 2.2754977651361235, "grad_norm": 1.8209433555603027, "learning_rate": 2.844273013743896e-05, "loss": 1.6089, "step": 700 } ], "logging_steps": 5, "max_steps": 1535, "num_input_tokens_seen": 0, "num_train_epochs": 5, "save_steps": 100, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 3.608456376263639e+17, "train_batch_size": 2, "trial_name": null, "trial_params": null }