| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 2.0, |
| "eval_steps": 500, |
| "global_step": 1326, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0015082956259426848, |
| "grad_norm": 50.77140426635742, |
| "learning_rate": 2.9850746268656716e-07, |
| "loss": 0.8260449171066284, |
| "step": 1, |
| "token_acc": 0.8622122762148338 |
| }, |
| { |
| "epoch": 0.015082956259426848, |
| "grad_norm": 20.249711990356445, |
| "learning_rate": 2.9850746268656716e-06, |
| "loss": 0.62702390882704, |
| "step": 10, |
| "token_acc": 0.8779647273464424 |
| }, |
| { |
| "epoch": 0.030165912518853696, |
| "grad_norm": 5.76005220413208, |
| "learning_rate": 5.970149253731343e-06, |
| "loss": 0.2204800844192505, |
| "step": 20, |
| "token_acc": 0.918760037574619 |
| }, |
| { |
| "epoch": 0.04524886877828054, |
| "grad_norm": 4.729307174682617, |
| "learning_rate": 8.955223880597016e-06, |
| "loss": 0.16907016038894654, |
| "step": 30, |
| "token_acc": 0.9293098014497321 |
| }, |
| { |
| "epoch": 0.06033182503770739, |
| "grad_norm": 3.8259971141815186, |
| "learning_rate": 1.1940298507462686e-05, |
| "loss": 0.15654852390289306, |
| "step": 40, |
| "token_acc": 0.9358284188430838 |
| }, |
| { |
| "epoch": 0.07541478129713423, |
| "grad_norm": 2.4985904693603516, |
| "learning_rate": 1.492537313432836e-05, |
| "loss": 0.1476317048072815, |
| "step": 50, |
| "token_acc": 0.940055318878362 |
| }, |
| { |
| "epoch": 0.09049773755656108, |
| "grad_norm": 2.2057321071624756, |
| "learning_rate": 1.791044776119403e-05, |
| "loss": 0.1472379446029663, |
| "step": 60, |
| "token_acc": 0.94014525821898 |
| }, |
| { |
| "epoch": 0.10558069381598793, |
| "grad_norm": 1.6078516244888306, |
| "learning_rate": 1.999971980603894e-05, |
| "loss": 0.14329800605773926, |
| "step": 70, |
| "token_acc": 0.9426169922762381 |
| }, |
| { |
| "epoch": 0.12066365007541478, |
| "grad_norm": 1.6763954162597656, |
| "learning_rate": 1.999473901685785e-05, |
| "loss": 0.14485034942626954, |
| "step": 80, |
| "token_acc": 0.94201612775463 |
| }, |
| { |
| "epoch": 0.13574660633484162, |
| "grad_norm": 1.1592150926589966, |
| "learning_rate": 1.9983535264822027e-05, |
| "loss": 0.14226235151290895, |
| "step": 90, |
| "token_acc": 0.9442544767389544 |
| }, |
| { |
| "epoch": 0.15082956259426847, |
| "grad_norm": 1.7728596925735474, |
| "learning_rate": 1.9966115525654667e-05, |
| "loss": 0.13862955570220947, |
| "step": 100, |
| "token_acc": 0.9451549110085695 |
| }, |
| { |
| "epoch": 0.16591251885369532, |
| "grad_norm": 1.2673591375350952, |
| "learning_rate": 1.9942490645300782e-05, |
| "loss": 0.13428819179534912, |
| "step": 110, |
| "token_acc": 0.9475016801337856 |
| }, |
| { |
| "epoch": 0.18099547511312217, |
| "grad_norm": 2.005850315093994, |
| "learning_rate": 1.9912675333174255e-05, |
| "loss": 0.13490796089172363, |
| "step": 120, |
| "token_acc": 0.947211978636826 |
| }, |
| { |
| "epoch": 0.19607843137254902, |
| "grad_norm": 1.723549246788025, |
| "learning_rate": 1.987668815299942e-05, |
| "loss": 0.13935812711715698, |
| "step": 130, |
| "token_acc": 0.944467493354231 |
| }, |
| { |
| "epoch": 0.21116138763197587, |
| "grad_norm": 1.238308072090149, |
| "learning_rate": 1.9834551511252828e-05, |
| "loss": 0.13570852279663087, |
| "step": 140, |
| "token_acc": 0.9463168996301303 |
| }, |
| { |
| "epoch": 0.22624434389140272, |
| "grad_norm": 1.2276662588119507, |
| "learning_rate": 1.9786291643212466e-05, |
| "loss": 0.13311694860458373, |
| "step": 150, |
| "token_acc": 0.9482371745612148 |
| }, |
| { |
| "epoch": 0.24132730015082957, |
| "grad_norm": 1.313895583152771, |
| "learning_rate": 1.9731938596623012e-05, |
| "loss": 0.13549520969390869, |
| "step": 160, |
| "token_acc": 0.9469860473599416 |
| }, |
| { |
| "epoch": 0.2564102564102564, |
| "grad_norm": 1.1264995336532593, |
| "learning_rate": 1.9671526212987425e-05, |
| "loss": 0.14870965480804443, |
| "step": 170, |
| "token_acc": 0.9486120215247184 |
| }, |
| { |
| "epoch": 0.27149321266968324, |
| "grad_norm": 1.2921221256256104, |
| "learning_rate": 1.9605092106496397e-05, |
| "loss": 0.12742632627487183, |
| "step": 180, |
| "token_acc": 0.9500985392125889 |
| }, |
| { |
| "epoch": 0.2865761689291101, |
| "grad_norm": 1.316508412361145, |
| "learning_rate": 1.9532677640608874e-05, |
| "loss": 0.12348119020462037, |
| "step": 190, |
| "token_acc": 0.950068166325835 |
| }, |
| { |
| "epoch": 0.30165912518853694, |
| "grad_norm": 1.032317876815796, |
| "learning_rate": 1.9454327902298176e-05, |
| "loss": 0.1240010142326355, |
| "step": 200, |
| "token_acc": 0.9513284215226613 |
| }, |
| { |
| "epoch": 0.3167420814479638, |
| "grad_norm": 1.4142050743103027, |
| "learning_rate": 1.9370091673979764e-05, |
| "loss": 0.12045929431915284, |
| "step": 210, |
| "token_acc": 0.9533292817851744 |
| }, |
| { |
| "epoch": 0.33182503770739064, |
| "grad_norm": 0.9892209768295288, |
| "learning_rate": 1.9280021403138154e-05, |
| "loss": 0.12211778163909912, |
| "step": 220, |
| "token_acc": 0.9538232220788002 |
| }, |
| { |
| "epoch": 0.3469079939668175, |
| "grad_norm": 1.2073554992675781, |
| "learning_rate": 1.9184173169671868e-05, |
| "loss": 0.12065562009811401, |
| "step": 230, |
| "token_acc": 0.95251651104532 |
| }, |
| { |
| "epoch": 0.36199095022624433, |
| "grad_norm": 0.8675851225852966, |
| "learning_rate": 1.9082606650976735e-05, |
| "loss": 0.11833696365356446, |
| "step": 240, |
| "token_acc": 0.9540585574863127 |
| }, |
| { |
| "epoch": 0.3770739064856712, |
| "grad_norm": 1.1243841648101807, |
| "learning_rate": 1.8975385084789333e-05, |
| "loss": 0.1197693943977356, |
| "step": 250, |
| "token_acc": 0.9537290121046467 |
| }, |
| { |
| "epoch": 0.39215686274509803, |
| "grad_norm": 0.9350152015686035, |
| "learning_rate": 1.886257522981367e-05, |
| "loss": 0.12061192989349365, |
| "step": 260, |
| "token_acc": 0.9528586661059967 |
| }, |
| { |
| "epoch": 0.4072398190045249, |
| "grad_norm": 0.9414811730384827, |
| "learning_rate": 1.8744247324155604e-05, |
| "loss": 0.12113888263702392, |
| "step": 270, |
| "token_acc": 0.9529830229415511 |
| }, |
| { |
| "epoch": 0.42232277526395173, |
| "grad_norm": 0.7945027947425842, |
| "learning_rate": 1.8620475041590948e-05, |
| "loss": 0.11630624532699585, |
| "step": 280, |
| "token_acc": 0.9544464663942006 |
| }, |
| { |
| "epoch": 0.4374057315233786, |
| "grad_norm": 1.1056071519851685, |
| "learning_rate": 1.8491335445694404e-05, |
| "loss": 0.1149565577507019, |
| "step": 290, |
| "token_acc": 0.9552702724579495 |
| }, |
| { |
| "epoch": 0.45248868778280543, |
| "grad_norm": 0.9025937914848328, |
| "learning_rate": 1.8356908941857942e-05, |
| "loss": 0.12011642456054687, |
| "step": 300, |
| "token_acc": 0.952321223032199 |
| }, |
| { |
| "epoch": 0.4675716440422323, |
| "grad_norm": 1.4455606937408447, |
| "learning_rate": 1.8217279227228503e-05, |
| "loss": 0.11713302135467529, |
| "step": 310, |
| "token_acc": 0.953317924700444 |
| }, |
| { |
| "epoch": 0.48265460030165913, |
| "grad_norm": 0.9116038680076599, |
| "learning_rate": 1.8072533238596134e-05, |
| "loss": 0.1182793378829956, |
| "step": 320, |
| "token_acc": 0.9535011190791577 |
| }, |
| { |
| "epoch": 0.497737556561086, |
| "grad_norm": 0.9311551451683044, |
| "learning_rate": 1.7922761098265096e-05, |
| "loss": 0.11895068883895873, |
| "step": 330, |
| "token_acc": 0.9533265066856625 |
| }, |
| { |
| "epoch": 0.5128205128205128, |
| "grad_norm": 0.7585210204124451, |
| "learning_rate": 1.7768056057941546e-05, |
| "loss": 0.11648373603820801, |
| "step": 340, |
| "token_acc": 0.954956099264881 |
| }, |
| { |
| "epoch": 0.5279034690799397, |
| "grad_norm": 1.0666239261627197, |
| "learning_rate": 1.760851444067282e-05, |
| "loss": 0.11833732128143311, |
| "step": 350, |
| "token_acc": 0.9538295410721173 |
| }, |
| { |
| "epoch": 0.5429864253393665, |
| "grad_norm": 0.7990158796310425, |
| "learning_rate": 1.7444235580874382e-05, |
| "loss": 0.11324186325073242, |
| "step": 360, |
| "token_acc": 0.9567262464722484 |
| }, |
| { |
| "epoch": 0.5580693815987934, |
| "grad_norm": 0.6436813473701477, |
| "learning_rate": 1.727532176248188e-05, |
| "loss": 0.11356878280639648, |
| "step": 370, |
| "token_acc": 0.9565708800396359 |
| }, |
| { |
| "epoch": 0.5731523378582202, |
| "grad_norm": 0.6460765600204468, |
| "learning_rate": 1.7101878155266702e-05, |
| "loss": 0.11457673311233521, |
| "step": 380, |
| "token_acc": 0.9559163059163059 |
| }, |
| { |
| "epoch": 0.5882352941176471, |
| "grad_norm": 1.6263368129730225, |
| "learning_rate": 1.692401274935478e-05, |
| "loss": 0.11588243246078492, |
| "step": 390, |
| "token_acc": 0.9552350930945086 |
| }, |
| { |
| "epoch": 0.6033182503770739, |
| "grad_norm": 0.7785161733627319, |
| "learning_rate": 1.674183628798936e-05, |
| "loss": 0.11478157043457031, |
| "step": 400, |
| "token_acc": 0.9561976946155061 |
| }, |
| { |
| "epoch": 0.6184012066365008, |
| "grad_norm": 1.0025640726089478, |
| "learning_rate": 1.655546219857962e-05, |
| "loss": 0.11230686902999878, |
| "step": 410, |
| "token_acc": 0.9568117392029348 |
| }, |
| { |
| "epoch": 0.6334841628959276, |
| "grad_norm": 0.9032500386238098, |
| "learning_rate": 1.6365006522078057e-05, |
| "loss": 0.11253654956817627, |
| "step": 420, |
| "token_acc": 0.9563058437677205 |
| }, |
| { |
| "epoch": 0.6485671191553545, |
| "grad_norm": 0.8704589605331421, |
| "learning_rate": 1.6170587840730627e-05, |
| "loss": 0.11393698453903198, |
| "step": 430, |
| "token_acc": 0.9554702905972454 |
| }, |
| { |
| "epoch": 0.6636500754147813, |
| "grad_norm": 0.7804033756256104, |
| "learning_rate": 1.59723272042446e-05, |
| "loss": 0.10969747304916382, |
| "step": 440, |
| "token_acc": 0.9561435746661766 |
| }, |
| { |
| "epoch": 0.6787330316742082, |
| "grad_norm": 0.5589516758918762, |
| "learning_rate": 1.5770348054420147e-05, |
| "loss": 0.10877771377563476, |
| "step": 450, |
| "token_acc": 0.9576669661575322 |
| }, |
| { |
| "epoch": 0.693815987933635, |
| "grad_norm": 0.9318929314613342, |
| "learning_rate": 1.556477614829251e-05, |
| "loss": 0.10682339668273926, |
| "step": 460, |
| "token_acc": 0.9583378522845235 |
| }, |
| { |
| "epoch": 0.7088989441930619, |
| "grad_norm": 0.7333630323410034, |
| "learning_rate": 1.5355739479832675e-05, |
| "loss": 0.11077797412872314, |
| "step": 470, |
| "token_acc": 0.9574586215693042 |
| }, |
| { |
| "epoch": 0.7239819004524887, |
| "grad_norm": 0.8250357508659363, |
| "learning_rate": 1.5143368200255294e-05, |
| "loss": 0.10814979076385497, |
| "step": 480, |
| "token_acc": 0.9579977253961083 |
| }, |
| { |
| "epoch": 0.7390648567119156, |
| "grad_norm": 0.7348103523254395, |
| "learning_rate": 1.4927794536983411e-05, |
| "loss": 0.11111488342285156, |
| "step": 490, |
| "token_acc": 0.9566194824725964 |
| }, |
| { |
| "epoch": 0.7541478129713424, |
| "grad_norm": 0.7893050909042358, |
| "learning_rate": 1.4709152711320529e-05, |
| "loss": 0.10905623435974121, |
| "step": 500, |
| "token_acc": 0.9582979240014569 |
| }, |
| { |
| "epoch": 0.7692307692307693, |
| "grad_norm": 0.7968176603317261, |
| "learning_rate": 1.4487578854881217e-05, |
| "loss": 0.10641565322875976, |
| "step": 510, |
| "token_acc": 0.9584265329733899 |
| }, |
| { |
| "epoch": 0.7843137254901961, |
| "grad_norm": 0.7752909660339355, |
| "learning_rate": 1.4263210924832332e-05, |
| "loss": 0.10894688367843627, |
| "step": 520, |
| "token_acc": 0.9575327527684345 |
| }, |
| { |
| "epoch": 0.799396681749623, |
| "grad_norm": 0.7637607455253601, |
| "learning_rate": 1.4036188617997581e-05, |
| "loss": 0.10907313823699952, |
| "step": 530, |
| "token_acc": 0.9576095334084307 |
| }, |
| { |
| "epoch": 0.8144796380090498, |
| "grad_norm": 0.5061221718788147, |
| "learning_rate": 1.3806653283878963e-05, |
| "loss": 0.1043128252029419, |
| "step": 540, |
| "token_acc": 0.9588358307520558 |
| }, |
| { |
| "epoch": 0.8295625942684767, |
| "grad_norm": 0.6120291352272034, |
| "learning_rate": 1.3574747836649197e-05, |
| "loss": 0.10561987161636352, |
| "step": 550, |
| "token_acc": 0.9587829153839257 |
| }, |
| { |
| "epoch": 0.8446455505279035, |
| "grad_norm": 0.775180995464325, |
| "learning_rate": 1.3340616666169925e-05, |
| "loss": 0.10071341991424561, |
| "step": 560, |
| "token_acc": 0.9609576307120258 |
| }, |
| { |
| "epoch": 0.8597285067873304, |
| "grad_norm": 0.54249507188797, |
| "learning_rate": 1.3104405548091188e-05, |
| "loss": 0.10593857765197753, |
| "step": 570, |
| "token_acc": 0.9581673306772909 |
| }, |
| { |
| "epoch": 0.8748114630467572, |
| "grad_norm": 0.615824818611145, |
| "learning_rate": 1.2866261553087977e-05, |
| "loss": 0.10219380855560303, |
| "step": 580, |
| "token_acc": 0.9610485850585893 |
| }, |
| { |
| "epoch": 0.889894419306184, |
| "grad_norm": 0.6161284446716309, |
| "learning_rate": 1.2626332955290564e-05, |
| "loss": 0.10377579927444458, |
| "step": 590, |
| "token_acc": 0.9594988978235897 |
| }, |
| { |
| "epoch": 0.9049773755656109, |
| "grad_norm": 0.5062152743339539, |
| "learning_rate": 1.2384769139965476e-05, |
| "loss": 0.10629792213439941, |
| "step": 600, |
| "token_acc": 0.9591665095775996 |
| }, |
| { |
| "epoch": 0.9200603318250377, |
| "grad_norm": 0.5114654302597046, |
| "learning_rate": 1.2141720510504662e-05, |
| "loss": 0.10493613481521606, |
| "step": 610, |
| "token_acc": 0.959506934685648 |
| }, |
| { |
| "epoch": 0.9351432880844646, |
| "grad_norm": 0.6344050765037537, |
| "learning_rate": 1.1897338394780762e-05, |
| "loss": 0.1010027289390564, |
| "step": 620, |
| "token_acc": 0.9610842985842986 |
| }, |
| { |
| "epoch": 0.9502262443438914, |
| "grad_norm": 0.5088986158370972, |
| "learning_rate": 1.165177495092677e-05, |
| "loss": 0.10696961879730224, |
| "step": 630, |
| "token_acc": 0.9577601718033548 |
| }, |
| { |
| "epoch": 0.9653092006033183, |
| "grad_norm": 0.6246373057365417, |
| "learning_rate": 1.1405183072598735e-05, |
| "loss": 0.10099267959594727, |
| "step": 640, |
| "token_acc": 0.9607030624191263 |
| }, |
| { |
| "epoch": 0.9803921568627451, |
| "grad_norm": 0.6200035810470581, |
| "learning_rate": 1.115771629378054e-05, |
| "loss": 0.10239546298980713, |
| "step": 650, |
| "token_acc": 0.9604871742956376 |
| }, |
| { |
| "epoch": 0.995475113122172, |
| "grad_norm": 0.5683614015579224, |
| "learning_rate": 1.0909528693189974e-05, |
| "loss": 0.10142800807952881, |
| "step": 660, |
| "token_acc": 0.9610971085054089 |
| }, |
| { |
| "epoch": 1.0105580693815988, |
| "grad_norm": 0.5012037754058838, |
| "learning_rate": 1.0660774798345655e-05, |
| "loss": 0.099783855676651, |
| "step": 670, |
| "token_acc": 0.9613417242769011 |
| }, |
| { |
| "epoch": 1.0256410256410255, |
| "grad_norm": 0.4982043504714966, |
| "learning_rate": 1.0411609489354495e-05, |
| "loss": 0.09685867428779601, |
| "step": 680, |
| "token_acc": 0.9626650319223696 |
| }, |
| { |
| "epoch": 1.0407239819004526, |
| "grad_norm": 1.000184416770935, |
| "learning_rate": 1.016218790247969e-05, |
| "loss": 0.09963431358337402, |
| "step": 690, |
| "token_acc": 0.9607164030078942 |
| }, |
| { |
| "epoch": 1.0558069381598794, |
| "grad_norm": 0.7291161417961121, |
| "learning_rate": 9.91266533354913e-06, |
| "loss": 0.0942987084388733, |
| "step": 700, |
| "token_acc": 0.9635701841712294 |
| }, |
| { |
| "epoch": 1.0708898944193062, |
| "grad_norm": 0.5546517372131348, |
| "learning_rate": 9.663197141264556e-06, |
| "loss": 0.09595948457717896, |
| "step": 710, |
| "token_acc": 0.9627618534639628 |
| }, |
| { |
| "epoch": 1.085972850678733, |
| "grad_norm": 0.5681226849555969, |
| "learning_rate": 9.413938650471515e-06, |
| "loss": 0.09452853202819825, |
| "step": 720, |
| "token_acc": 0.9637946320625453 |
| }, |
| { |
| "epoch": 1.10105580693816, |
| "grad_norm": 0.5453760623931885, |
| "learning_rate": 9.165045055450393e-06, |
| "loss": 0.09598196744918823, |
| "step": 730, |
| "token_acc": 0.9625681478999787 |
| }, |
| { |
| "epoch": 1.1161387631975868, |
| "grad_norm": 0.5296013951301575, |
| "learning_rate": 8.916671323288777e-06, |
| "loss": 0.09315053224563599, |
| "step": 740, |
| "token_acc": 0.963243227480827 |
| }, |
| { |
| "epoch": 1.1312217194570136, |
| "grad_norm": 0.5977329611778259, |
| "learning_rate": 8.66897209739523e-06, |
| "loss": 0.09293179512023926, |
| "step": 750, |
| "token_acc": 0.9642896093981272 |
| }, |
| { |
| "epoch": 1.1463046757164403, |
| "grad_norm": 0.4531576931476593, |
| "learning_rate": 8.422101601214643e-06, |
| "loss": 0.09062933921813965, |
| "step": 760, |
| "token_acc": 0.9646416403986372 |
| }, |
| { |
| "epoch": 1.1613876319758674, |
| "grad_norm": 0.5825615525245667, |
| "learning_rate": 8.176213542205016e-06, |
| "loss": 0.09595742225646972, |
| "step": 770, |
| "token_acc": 0.9625474083438685 |
| }, |
| { |
| "epoch": 1.1764705882352942, |
| "grad_norm": 0.7963739037513733, |
| "learning_rate": 7.931461016135534e-06, |
| "loss": 0.0948381781578064, |
| "step": 780, |
| "token_acc": 0.9635484470894551 |
| }, |
| { |
| "epoch": 1.191553544494721, |
| "grad_norm": 0.61602783203125, |
| "learning_rate": 7.687996411765503e-06, |
| "loss": 0.09590113162994385, |
| "step": 790, |
| "token_acc": 0.9624717172505266 |
| }, |
| { |
| "epoch": 1.2066365007541477, |
| "grad_norm": 0.5428170561790466, |
| "learning_rate": 7.445971315963459e-06, |
| "loss": 0.09370912313461303, |
| "step": 800, |
| "token_acc": 0.9632577697093826 |
| }, |
| { |
| "epoch": 1.2217194570135748, |
| "grad_norm": 0.6003525257110596, |
| "learning_rate": 7.205536419325581e-06, |
| "loss": 0.09444937705993653, |
| "step": 810, |
| "token_acc": 0.9637329566392335 |
| }, |
| { |
| "epoch": 1.2368024132730016, |
| "grad_norm": 0.5893547534942627, |
| "learning_rate": 6.966841422352127e-06, |
| "loss": 0.09065158367156982, |
| "step": 820, |
| "token_acc": 0.965184163040017 |
| }, |
| { |
| "epoch": 1.2518853695324283, |
| "grad_norm": 0.5336847305297852, |
| "learning_rate": 6.7300349422403344e-06, |
| "loss": 0.09595112204551696, |
| "step": 830, |
| "token_acc": 0.9626478552087787 |
| }, |
| { |
| "epoch": 1.2669683257918551, |
| "grad_norm": 0.5500826835632324, |
| "learning_rate": 6.495264420351799e-06, |
| "loss": 0.09168100357055664, |
| "step": 840, |
| "token_acc": 0.9643116535410644 |
| }, |
| { |
| "epoch": 1.282051282051282, |
| "grad_norm": 0.5517785549163818, |
| "learning_rate": 6.262676030411985e-06, |
| "loss": 0.09056183099746704, |
| "step": 850, |
| "token_acc": 0.9643301166301581 |
| }, |
| { |
| "epoch": 1.297134238310709, |
| "grad_norm": 0.7318987846374512, |
| "learning_rate": 6.0324145874989606e-06, |
| "loss": 0.0914067804813385, |
| "step": 860, |
| "token_acc": 0.9640134891670902 |
| }, |
| { |
| "epoch": 1.3122171945701357, |
| "grad_norm": 0.6312046647071838, |
| "learning_rate": 5.804623457878088e-06, |
| "loss": 0.09138967990875244, |
| "step": 870, |
| "token_acc": 0.9655435731205585 |
| }, |
| { |
| "epoch": 1.3273001508295625, |
| "grad_norm": 0.5690848231315613, |
| "learning_rate": 5.579444469738771e-06, |
| "loss": 0.09293912649154663, |
| "step": 880, |
| "token_acc": 0.963402983234714 |
| }, |
| { |
| "epoch": 1.3423831070889896, |
| "grad_norm": 0.516086995601654, |
| "learning_rate": 5.357017824888831e-06, |
| "loss": 0.08710741400718688, |
| "step": 890, |
| "token_acc": 0.9655223341757195 |
| }, |
| { |
| "epoch": 1.3574660633484164, |
| "grad_norm": 0.5713684558868408, |
| "learning_rate": 5.137482011461513e-06, |
| "loss": 0.09022394418716431, |
| "step": 900, |
| "token_acc": 0.9650503470524978 |
| }, |
| { |
| "epoch": 1.3725490196078431, |
| "grad_norm": 0.5328916907310486, |
| "learning_rate": 4.920973717689495e-06, |
| "loss": 0.09139166474342346, |
| "step": 910, |
| "token_acc": 0.964425922570514 |
| }, |
| { |
| "epoch": 1.38763197586727, |
| "grad_norm": 0.6854859590530396, |
| "learning_rate": 4.707627746799517e-06, |
| "loss": 0.09102691411972046, |
| "step": 920, |
| "token_acc": 0.9639419712966274 |
| }, |
| { |
| "epoch": 1.4027149321266967, |
| "grad_norm": 0.4496878385543823, |
| "learning_rate": 4.497576933080677e-06, |
| "loss": 0.0911118745803833, |
| "step": 930, |
| "token_acc": 0.9642824308173209 |
| }, |
| { |
| "epoch": 1.4177978883861238, |
| "grad_norm": 0.5049023032188416, |
| "learning_rate": 4.290952059178644e-06, |
| "loss": 0.09250409603118896, |
| "step": 940, |
| "token_acc": 0.9633827640695272 |
| }, |
| { |
| "epoch": 1.4328808446455505, |
| "grad_norm": 0.5796614289283752, |
| "learning_rate": 4.087881774667258e-06, |
| "loss": 0.09458078145980835, |
| "step": 950, |
| "token_acc": 0.9632013078695103 |
| }, |
| { |
| "epoch": 1.4479638009049773, |
| "grad_norm": 0.5066742897033691, |
| "learning_rate": 3.888492515948241e-06, |
| "loss": 0.0892916202545166, |
| "step": 960, |
| "token_acc": 0.9655580487504994 |
| }, |
| { |
| "epoch": 1.4630467571644044, |
| "grad_norm": 0.5400696992874146, |
| "learning_rate": 3.692908427528866e-06, |
| "loss": 0.0880997657775879, |
| "step": 970, |
| "token_acc": 0.966000359625989 |
| }, |
| { |
| "epoch": 1.4781297134238311, |
| "grad_norm": 0.6165044903755188, |
| "learning_rate": 3.5012512847266435e-06, |
| "loss": 0.08752999305725098, |
| "step": 980, |
| "token_acc": 0.9653556766326328 |
| }, |
| { |
| "epoch": 1.493212669683258, |
| "grad_norm": 0.5573943853378296, |
| "learning_rate": 3.313640417849107e-06, |
| "loss": 0.09335318803787232, |
| "step": 990, |
| "token_acc": 0.9641705051838737 |
| }, |
| { |
| "epoch": 1.5082956259426847, |
| "grad_norm": 0.5577435493469238, |
| "learning_rate": 3.130192637895888e-06, |
| "loss": 0.09085619449615479, |
| "step": 1000, |
| "token_acc": 0.9641629636021111 |
| }, |
| { |
| "epoch": 1.5233785822021115, |
| "grad_norm": 0.535598874092102, |
| "learning_rate": 2.9510221638294378e-06, |
| "loss": 0.0922804594039917, |
| "step": 1010, |
| "token_acc": 0.9641638745050259 |
| }, |
| { |
| "epoch": 1.5384615384615383, |
| "grad_norm": 0.48957136273384094, |
| "learning_rate": 2.776240551459558e-06, |
| "loss": 0.08728803992271424, |
| "step": 1020, |
| "token_acc": 0.9655017705779615 |
| }, |
| { |
| "epoch": 1.5535444947209653, |
| "grad_norm": 0.5388123393058777, |
| "learning_rate": 2.6059566239861e-06, |
| "loss": 0.08832274079322815, |
| "step": 1030, |
| "token_acc": 0.9659863945578231 |
| }, |
| { |
| "epoch": 1.5686274509803921, |
| "grad_norm": 0.5348907709121704, |
| "learning_rate": 2.4402764042430317e-06, |
| "loss": 0.087826669216156, |
| "step": 1040, |
| "token_acc": 0.964891315857145 |
| }, |
| { |
| "epoch": 1.5837104072398192, |
| "grad_norm": 0.6047067046165466, |
| "learning_rate": 2.2793030486861212e-06, |
| "loss": 0.08644192218780518, |
| "step": 1050, |
| "token_acc": 0.9671007319846876 |
| }, |
| { |
| "epoch": 1.598793363499246, |
| "grad_norm": 0.5878674983978271, |
| "learning_rate": 2.1231367831652415e-06, |
| "loss": 0.08670351505279542, |
| "step": 1060, |
| "token_acc": 0.9657461539680277 |
| }, |
| { |
| "epoch": 1.6138763197586727, |
| "grad_norm": 0.446951687335968, |
| "learning_rate": 1.971874840521374e-06, |
| "loss": 0.09097723364830017, |
| "step": 1070, |
| "token_acc": 0.9642487991731015 |
| }, |
| { |
| "epoch": 1.6289592760180995, |
| "grad_norm": 0.4840224087238312, |
| "learning_rate": 1.825611400047126e-06, |
| "loss": 0.09004112482070922, |
| "step": 1080, |
| "token_acc": 0.9648130922620778 |
| }, |
| { |
| "epoch": 1.6440422322775263, |
| "grad_norm": 0.4454542100429535, |
| "learning_rate": 1.684437528848456e-06, |
| "loss": 0.08679054975509644, |
| "step": 1090, |
| "token_acc": 0.9660236703356218 |
| }, |
| { |
| "epoch": 1.6591251885369531, |
| "grad_norm": 0.48091331124305725, |
| "learning_rate": 1.5484411251441234e-06, |
| "loss": 0.08844754695892335, |
| "step": 1100, |
| "token_acc": 0.9651457249412957 |
| }, |
| { |
| "epoch": 1.6742081447963801, |
| "grad_norm": 0.5707911252975464, |
| "learning_rate": 1.4177068635381575e-06, |
| "loss": 0.08476131558418273, |
| "step": 1110, |
| "token_acc": 0.9665394775462284 |
| }, |
| { |
| "epoch": 1.689291101055807, |
| "grad_norm": 0.4876456558704376, |
| "learning_rate": 1.2923161422994302e-06, |
| "loss": 0.09168508648872375, |
| "step": 1120, |
| "token_acc": 0.9643400266906675 |
| }, |
| { |
| "epoch": 1.704374057315234, |
| "grad_norm": 0.3989841341972351, |
| "learning_rate": 1.1723470326811582e-06, |
| "loss": 0.08490124940872193, |
| "step": 1130, |
| "token_acc": 0.9661019602931046 |
| }, |
| { |
| "epoch": 1.7194570135746607, |
| "grad_norm": 0.5441372394561768, |
| "learning_rate": 1.0578742303118695e-06, |
| "loss": 0.09091441631317139, |
| "step": 1140, |
| "token_acc": 0.9643065857496067 |
| }, |
| { |
| "epoch": 1.7345399698340875, |
| "grad_norm": 0.5801622271537781, |
| "learning_rate": 9.489690086881321e-07, |
| "loss": 0.08664626479148865, |
| "step": 1150, |
| "token_acc": 0.965849782666272 |
| }, |
| { |
| "epoch": 1.7496229260935143, |
| "grad_norm": 0.5623024106025696, |
| "learning_rate": 8.456991747979782e-07, |
| "loss": 0.08934998512268066, |
| "step": 1160, |
| "token_acc": 0.9652618743122268 |
| }, |
| { |
| "epoch": 1.7647058823529411, |
| "grad_norm": 0.4981709122657776, |
| "learning_rate": 7.48129026902642e-07, |
| "loss": 0.08743290901184082, |
| "step": 1170, |
| "token_acc": 0.9658945262433961 |
| }, |
| { |
| "epoch": 1.779788838612368, |
| "grad_norm": 0.6625328063964844, |
| "learning_rate": 6.563193145029423e-07, |
| "loss": 0.08860676884651184, |
| "step": 1180, |
| "token_acc": 0.9654595396710132 |
| }, |
| { |
| "epoch": 1.7948717948717947, |
| "grad_norm": 0.47374987602233887, |
| "learning_rate": 5.703272005151905e-07, |
| "loss": 0.08183513283729553, |
| "step": 1190, |
| "token_acc": 0.9678513184027385 |
| }, |
| { |
| "epoch": 1.8099547511312217, |
| "grad_norm": 0.4954168200492859, |
| "learning_rate": 4.902062256801954e-07, |
| "loss": 0.08469556570053101, |
| "step": 1200, |
| "token_acc": 0.9671236314584183 |
| }, |
| { |
| "epoch": 1.8250377073906485, |
| "grad_norm": 0.5051677227020264, |
| "learning_rate": 4.1600627522751624e-07, |
| "loss": 0.08613886833190917, |
| "step": 1210, |
| "token_acc": 0.9665093175367148 |
| }, |
| { |
| "epoch": 1.8401206636500755, |
| "grad_norm": 0.4586503803730011, |
| "learning_rate": 3.477735478157429e-07, |
| "loss": 0.082184100151062, |
| "step": 1220, |
| "token_acc": 0.9684546054077819 |
| }, |
| { |
| "epoch": 1.8552036199095023, |
| "grad_norm": 0.4913497865200043, |
| "learning_rate": 2.855505267680947e-07, |
| "loss": 0.09022275805473327, |
| "step": 1230, |
| "token_acc": 0.9645829591042453 |
| }, |
| { |
| "epoch": 1.8702865761689291, |
| "grad_norm": 0.5049121379852295, |
| "learning_rate": 2.2937595362129805e-07, |
| "loss": 0.0863489031791687, |
| "step": 1240, |
| "token_acc": 0.9656973816717019 |
| }, |
| { |
| "epoch": 1.885369532428356, |
| "grad_norm": 0.4359610974788666, |
| "learning_rate": 1.7928480400417837e-07, |
| "loss": 0.08408082127571107, |
| "step": 1250, |
| "token_acc": 0.9673078445341565 |
| }, |
| { |
| "epoch": 1.9004524886877827, |
| "grad_norm": 0.569852352142334, |
| "learning_rate": 1.3530826586100276e-07, |
| "loss": 0.08361083269119263, |
| "step": 1260, |
| "token_acc": 0.9671 |
| }, |
| { |
| "epoch": 1.9155354449472095, |
| "grad_norm": 0.901442289352417, |
| "learning_rate": 9.747372003312106e-08, |
| "loss": 0.08902775645256042, |
| "step": 1270, |
| "token_acc": 0.9657552197082205 |
| }, |
| { |
| "epoch": 1.9306184012066365, |
| "grad_norm": 0.49275603890419006, |
| "learning_rate": 6.580472321099928e-08, |
| "loss": 0.08826894164085389, |
| "step": 1280, |
| "token_acc": 0.9651560396713651 |
| }, |
| { |
| "epoch": 1.9457013574660633, |
| "grad_norm": 0.7015301585197449, |
| "learning_rate": 4.0320993267264354e-08, |
| "loss": 0.08693464994430541, |
| "step": 1290, |
| "token_acc": 0.9661187520966119 |
| }, |
| { |
| "epoch": 1.9607843137254903, |
| "grad_norm": 0.5178602933883667, |
| "learning_rate": 2.1038396979884414e-08, |
| "loss": 0.08611137866973877, |
| "step": 1300, |
| "token_acc": 0.9662187693656383 |
| }, |
| { |
| "epoch": 1.9758672699849171, |
| "grad_norm": 0.5327852368354797, |
| "learning_rate": 7.968940153136872e-09, |
| "loss": 0.08536949157714843, |
| "step": 1310, |
| "token_acc": 0.9665300765118289 |
| }, |
| { |
| "epoch": 1.990950226244344, |
| "grad_norm": 0.5267050266265869, |
| "learning_rate": 1.1207601425078375e-09, |
| "loss": 0.08764583468437195, |
| "step": 1320, |
| "token_acc": 0.9653767820773931 |
| } |
| ], |
| "logging_steps": 10, |
| "max_steps": 1326, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 2, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 7.892974877726999e+18, |
| "train_batch_size": 2, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
