{ "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": [ { "cos_sim": 0.03186357393860817, "distill_lambda": 0.05000000447034836, "distill_loss": 0.9681364297866821, "distill_only": 0.0, "epoch": 0.0015082956259426848, "grad_norm": 211.61996459960938, "learning_rate": 2.9850746268656716e-07, "loss": 2.372626781463623, "sft_loss": 0.025737762451171875, "step": 1, "token_acc": 0.8626918158567775 }, { "cos_sim": 0.03638706273502774, "distill_lambda": 0.050000001986821495, "distill_loss": 0.9636128743489584, "distill_only": 0.0, "epoch": 0.015082956259426848, "grad_norm": 27.790311813354492, "learning_rate": 2.9850746268656716e-06, "loss": 2.1743456522623696, "sft_loss": 0.019767655266655818, "step": 10, "token_acc": 0.8781167646462599 }, { "cos_sim": 0.14173474311828613, "distill_lambda": 0.05, "distill_loss": 0.8582653045654297, "distill_only": 0.0, "epoch": 0.030165912518853696, "grad_norm": 13.423632621765137, "learning_rate": 5.970149253731343e-06, "loss": 1.6065895080566406, "sft_loss": 0.007292652130126953, "step": 20, "token_acc": 0.9173358383079301 }, { "cos_sim": 0.5156546592712402, "distill_lambda": 0.05, "distill_loss": 0.4843453407287598, "distill_only": 0.0, "epoch": 0.04524886877828054, "grad_norm": 11.553773880004883, "learning_rate": 8.955223880597016e-06, "loss": 0.9485068321228027, "sft_loss": 0.0054235696792602536, "step": 30, "token_acc": 0.9289473684210526 }, { "cos_sim": 0.7724837779998779, "distill_lambda": 0.05, "distill_loss": 0.22751622200012206, "distill_only": 0.0, "epoch": 0.06033182503770739, "grad_norm": 7.349006175994873, "learning_rate": 1.1940298507462686e-05, "loss": 0.5208699703216553, "sft_loss": 0.004901373386383056, "step": 40, "token_acc": 0.9352286844397482 }, { "cos_sim": 0.8874543190002442, "distill_lambda": 0.05, "distill_loss": 0.11254562139511108, "distill_only": 0.0, "epoch": 0.07541478129713423, "grad_norm": 4.914926052093506, "learning_rate": 1.492537313432836e-05, "loss": 0.32853567600250244, "sft_loss": 0.004639458656311035, "step": 50, "token_acc": 0.9395466395371018 }, { "cos_sim": 0.9269522666931153, "distill_lambda": 0.05, "distill_loss": 0.0730478286743164, "distill_only": 0.0, "epoch": 0.09049773755656108, "grad_norm": 8.202323913574219, "learning_rate": 1.791044776119403e-05, "loss": 0.26580114364624025, "sft_loss": 0.004653894901275634, "step": 60, "token_acc": 0.9396150014461548 }, { "cos_sim": 0.9346080780029297, "distill_lambda": 0.05, "distill_loss": 0.06539193391799927, "distill_only": 0.0, "epoch": 0.10558069381598793, "grad_norm": 1.661431074142456, "learning_rate": 1.999971980603894e-05, "loss": 0.25142083168029783, "sft_loss": 0.004587304592132568, "step": 70, "token_acc": 0.9416931697713161 }, { "cos_sim": 0.9475898742675781, "distill_lambda": 0.05, "distill_loss": 0.052410244941711426, "distill_only": 0.0, "epoch": 0.12066365007541478, "grad_norm": 1.8265098333358765, "learning_rate": 1.999473901685785e-05, "loss": 0.23054256439208984, "sft_loss": 0.0045839428901672365, "step": 80, "token_acc": 0.9412239983523708 }, { "cos_sim": 0.9495115280151367, "distill_lambda": 0.05, "distill_loss": 0.05048844814300537, "distill_only": 0.0, "epoch": 0.13574660633484162, "grad_norm": 1.4613008499145508, "learning_rate": 1.9983535264822027e-05, "loss": 0.24433209896087646, "sft_loss": 0.0051109552383422855, "step": 90, "token_acc": 0.9402801523740965 }, { "cos_sim": 0.9408574104309082, "distill_lambda": 0.05, "distill_loss": 0.05914261341094971, "distill_only": 0.0, "epoch": 0.15082956259426847, "grad_norm": 1.3204941749572754, "learning_rate": 1.9966115525654667e-05, "loss": 0.23413853645324706, "sft_loss": 0.004359698295593262, "step": 100, "token_acc": 0.9443052808906467 }, { "cos_sim": 0.9505158424377441, "distill_lambda": 0.05, "distill_loss": 0.04948410391807556, "distill_only": 0.0, "epoch": 0.16591251885369532, "grad_norm": 1.1803853511810303, "learning_rate": 1.9942490645300782e-05, "loss": 0.21255049705505372, "sft_loss": 0.00416799783706665, "step": 110, "token_acc": 0.9469546598315177 }, { "cos_sim": 0.9514935493469239, "distill_lambda": 0.05, "distill_loss": 0.048506435751914975, "distill_only": 0.0, "epoch": 0.18099547511312217, "grad_norm": 1.406587839126587, "learning_rate": 1.9912675333174255e-05, "loss": 0.2096571922302246, "sft_loss": 0.0041264653205871586, "step": 120, "token_acc": 0.9481497965412004 }, { "cos_sim": 0.9463747978210449, "distill_lambda": 0.05, "distill_loss": 0.05362526774406433, "distill_only": 0.0, "epoch": 0.19607843137254902, "grad_norm": 1.3852708339691162, "learning_rate": 1.987668815299942e-05, "loss": 0.21961355209350586, "sft_loss": 0.00418165922164917, "step": 130, "token_acc": 0.9473255581677653 }, { "cos_sim": 0.9499053955078125, "distill_lambda": 0.05, "distill_loss": 0.050094664096832275, "distill_only": 0.0, "epoch": 0.21116138763197587, "grad_norm": 1.0918831825256348, "learning_rate": 1.9834551511252828e-05, "loss": 0.23451168537139894, "sft_loss": 0.004823756217956543, "step": 140, "token_acc": 0.9448166248273199 }, { "cos_sim": 0.9528703689575195, "distill_lambda": 0.05, "distill_loss": 0.047129732370376584, "distill_only": 0.0, "epoch": 0.22624434389140272, "grad_norm": 2.272617816925049, "learning_rate": 1.9786291643212466e-05, "loss": 0.20551390647888185, "sft_loss": 0.004065823554992676, "step": 150, "token_acc": 0.9483908646604986 }, { "cos_sim": 0.9500127792358398, "distill_lambda": 0.05, "distill_loss": 0.04998720288276672, "distill_only": 0.0, "epoch": 0.24132730015082957, "grad_norm": 0.9511111974716187, "learning_rate": 1.9731938596623012e-05, "loss": 0.2126539468765259, "sft_loss": 0.004146075248718262, "step": 160, "token_acc": 0.9475028118065477 }, { "cos_sim": 0.949603271484375, "distill_lambda": 0.05, "distill_loss": 0.050396817922592166, "distill_only": 0.0, "epoch": 0.2564102564102564, "grad_norm": 0.7958192229270935, "learning_rate": 1.9671526212987425e-05, "loss": 0.20801222324371338, "sft_loss": 0.0039805412292480465, "step": 170, "token_acc": 0.9501974115459078 }, { "cos_sim": 0.9550053596496582, "distill_lambda": 0.05, "distill_loss": 0.04499462842941284, "distill_only": 0.0, "epoch": 0.27149321266968324, "grad_norm": 0.9181481003761292, "learning_rate": 1.9605092106496397e-05, "loss": 0.20015151500701905, "sft_loss": 0.004005002975463867, "step": 180, "token_acc": 0.9495118171834334 }, { "cos_sim": 0.952204418182373, "distill_lambda": 0.05, "distill_loss": 0.04779554009437561, "distill_only": 0.0, "epoch": 0.2865761689291101, "grad_norm": 1.0268183946609497, "learning_rate": 1.9532677640608874e-05, "loss": 0.2003624200820923, "sft_loss": 0.0038715481758117678, "step": 190, "token_acc": 0.950470967342133 }, { "cos_sim": 0.9559913635253906, "distill_lambda": 0.05, "distill_loss": 0.04400867223739624, "distill_only": 0.0, "epoch": 0.30165912518853694, "grad_norm": 0.9260615706443787, "learning_rate": 1.9454327902298176e-05, "loss": 0.19288719892501832, "sft_loss": 0.00382729172706604, "step": 200, "token_acc": 0.9516792651420917 }, { "cos_sim": 0.9555576324462891, "distill_lambda": 0.05, "distill_loss": 0.04444245994091034, "distill_only": 0.0, "epoch": 0.3167420814479638, "grad_norm": 1.2006932497024536, "learning_rate": 1.9370091673979764e-05, "loss": 0.1899334669113159, "sft_loss": 0.0037132978439331056, "step": 210, "token_acc": 0.9536409393358578 }, { "cos_sim": 0.9535330772399903, "distill_lambda": 0.05, "distill_loss": 0.04646695256233215, "distill_only": 0.0, "epoch": 0.33182503770739064, "grad_norm": 0.713481068611145, "learning_rate": 1.9280021403138154e-05, "loss": 0.19584884643554687, "sft_loss": 0.003796929121017456, "step": 220, "token_acc": 0.9538078554305734 }, { "cos_sim": 0.9567322731018066, "distill_lambda": 0.05, "distill_loss": 0.043267744779586795, "distill_only": 0.0, "epoch": 0.3469079939668175, "grad_norm": 1.056697964668274, "learning_rate": 1.9184173169671868e-05, "loss": 0.18983482122421264, "sft_loss": 0.0037689507007598877, "step": 230, "token_acc": 0.9520122328138725 }, { "cos_sim": 0.9581542015075684, "distill_lambda": 0.05, "distill_loss": 0.0418458491563797, "distill_only": 0.0, "epoch": 0.36199095022624433, "grad_norm": 0.7865216135978699, "learning_rate": 1.9082606650976735e-05, "loss": 0.18423471450805665, "sft_loss": 0.003665041923522949, "step": 240, "token_acc": 0.9545187653733238 }, { "cos_sim": 0.9589271545410156, "distill_lambda": 0.05, "distill_loss": 0.04107289016246796, "distill_only": 0.0, "epoch": 0.3770739064856712, "grad_norm": 0.6734248995780945, "learning_rate": 1.8975385084789333e-05, "loss": 0.18630998134613036, "sft_loss": 0.0037685424089431764, "step": 250, "token_acc": 0.9535825849277626 }, { "cos_sim": 0.9578152656555176, "distill_lambda": 0.05, "distill_loss": 0.04218478798866272, "distill_only": 0.0, "epoch": 0.39215686274509803, "grad_norm": 0.9397635459899902, "learning_rate": 1.886257522981367e-05, "loss": 0.18721728324890136, "sft_loss": 0.003741300106048584, "step": 260, "token_acc": 0.9535795275354048 }, { "cos_sim": 0.9539778709411622, "distill_lambda": 0.05, "distill_loss": 0.0460222065448761, "distill_only": 0.0, "epoch": 0.4072398190045249, "grad_norm": 0.6943348050117493, "learning_rate": 1.8744247324155604e-05, "loss": 0.1930428147315979, "sft_loss": 0.0037314772605895998, "step": 270, "token_acc": 0.9541790279740917 }, { "cos_sim": 0.9539613723754883, "distill_lambda": 0.05, "distill_loss": 0.04603857696056366, "distill_only": 0.0, "epoch": 0.42232277526395173, "grad_norm": 0.9831451773643494, "learning_rate": 1.8620475041590948e-05, "loss": 0.18914786577224732, "sft_loss": 0.0036089420318603516, "step": 280, "token_acc": 0.9551176874198264 }, { "cos_sim": 0.9597013473510743, "distill_lambda": 0.05, "distill_loss": 0.040298661589622496, "distill_only": 0.0, "epoch": 0.4374057315233786, "grad_norm": 0.649888813495636, "learning_rate": 1.8491335445694404e-05, "loss": 0.17802318334579467, "sft_loss": 0.0035482913255691527, "step": 290, "token_acc": 0.9559825808227161 }, { "cos_sim": 0.9558124542236328, "distill_lambda": 0.05, "distill_loss": 0.04418749511241913, "distill_only": 0.0, "epoch": 0.45248868778280543, "grad_norm": 0.7827683091163635, "learning_rate": 1.8356908941857942e-05, "loss": 0.1903221368789673, "sft_loss": 0.003738191723823547, "step": 300, "token_acc": 0.9527934362364607 }, { "cos_sim": 0.9548751831054687, "distill_lambda": 0.05, "distill_loss": 0.04512483775615692, "distill_only": 0.0, "epoch": 0.4675716440422323, "grad_norm": 0.8502541780471802, "learning_rate": 1.8217279227228503e-05, "loss": 0.187775182723999, "sft_loss": 0.00361173152923584, "step": 310, "token_acc": 0.9548689252478559 }, { "cos_sim": 0.9588627815246582, "distill_lambda": 0.05, "distill_loss": 0.041137173771858215, "distill_only": 0.0, "epoch": 0.48265460030165913, "grad_norm": 0.6845568418502808, "learning_rate": 1.8072533238596134e-05, "loss": 0.1827320337295532, "sft_loss": 0.0036535173654556274, "step": 320, "token_acc": 0.954643036587037 }, { "cos_sim": 0.9571377754211425, "distill_lambda": 0.05, "distill_loss": 0.04286225140094757, "distill_only": 0.0, "epoch": 0.497737556561086, "grad_norm": 0.5921292901039124, "learning_rate": 1.7922761098265096e-05, "loss": 0.18519527912139894, "sft_loss": 0.0036442399024963377, "step": 330, "token_acc": 0.9540685336063992 }, { "cos_sim": 0.9566804885864257, "distill_lambda": 0.05, "distill_loss": 0.04331963062286377, "distill_only": 0.0, "epoch": 0.5128205128205128, "grad_norm": 0.7089025378227234, "learning_rate": 1.7768056057941546e-05, "loss": 0.1843092679977417, "sft_loss": 0.0035936832427978516, "step": 340, "token_acc": 0.9545115348586126 }, { "cos_sim": 0.9552593231201172, "distill_lambda": 0.05, "distill_loss": 0.04474072754383087, "distill_only": 0.0, "epoch": 0.5279034690799397, "grad_norm": 0.9024905562400818, "learning_rate": 1.760851444067282e-05, "loss": 0.189734148979187, "sft_loss": 0.0036921560764312743, "step": 350, "token_acc": 0.9544465869649055 }, { "cos_sim": 0.9582769393920898, "distill_lambda": 0.05, "distill_loss": 0.04172297418117523, "distill_only": 0.0, "epoch": 0.5429864253393665, "grad_norm": 0.8202751874923706, "learning_rate": 1.7444235580874382e-05, "loss": 0.1781691789627075, "sft_loss": 0.003481638431549072, "step": 360, "token_acc": 0.9575075338424989 }, { "cos_sim": 0.9562713623046875, "distill_lambda": 0.05, "distill_loss": 0.043728628754615785, "distill_only": 0.0, "epoch": 0.5580693815987934, "grad_norm": 0.6490883827209473, "learning_rate": 1.727532176248188e-05, "loss": 0.1826735258102417, "sft_loss": 0.00352211594581604, "step": 370, "token_acc": 0.9564315352697096 }, { "cos_sim": 0.9532212257385254, "distill_lambda": 0.05, "distill_loss": 0.046778762340545656, "distill_only": 0.0, "epoch": 0.5731523378582202, "grad_norm": 0.5874303579330444, "learning_rate": 1.7101878155266702e-05, "loss": 0.18791269063949584, "sft_loss": 0.0035333335399627686, "step": 380, "token_acc": 0.9563780663780663 }, { "cos_sim": 0.9561857223510742, "distill_lambda": 0.05, "distill_loss": 0.04381435215473175, "distill_only": 0.0, "epoch": 0.5882352941176471, "grad_norm": 0.7648326754570007, "learning_rate": 1.692401274935478e-05, "loss": 0.1842123746871948, "sft_loss": 0.0035659193992614744, "step": 390, "token_acc": 0.9560894414329484 }, { "cos_sim": 0.95799560546875, "distill_lambda": 0.05, "distill_loss": 0.042004448175430295, "distill_only": 0.0, "epoch": 0.6033182503770739, "grad_norm": 0.6999265551567078, "learning_rate": 1.674183628798936e-05, "loss": 0.17877480983734131, "sft_loss": 0.003486490249633789, "step": 400, "token_acc": 0.9573503868624664 }, { "cos_sim": 0.9572476387023926, "distill_lambda": 0.05, "distill_loss": 0.04275233149528503, "distill_only": 0.0, "epoch": 0.6184012066365008, "grad_norm": 0.7718906402587891, "learning_rate": 1.655546219857962e-05, "loss": 0.18074029684066772, "sft_loss": 0.003510516881942749, "step": 410, "token_acc": 0.9562963299831734 }, { "cos_sim": 0.9566576957702637, "distill_lambda": 0.05, "distill_loss": 0.04334227740764618, "distill_only": 0.0, "epoch": 0.6334841628959276, "grad_norm": 0.8230934143066406, "learning_rate": 1.6365006522078057e-05, "loss": 0.18042057752609253, "sft_loss": 0.0034710288047790526, "step": 420, "token_acc": 0.956474428726877 }, { "cos_sim": 0.956779670715332, "distill_lambda": 0.05, "distill_loss": 0.04322041571140289, "distill_only": 0.0, "epoch": 0.6485671191553545, "grad_norm": 0.7724851369857788, "learning_rate": 1.6170587840730627e-05, "loss": 0.18108007907867432, "sft_loss": 0.0034977316856384276, "step": 430, "token_acc": 0.9561123255422902 }, { "cos_sim": 0.9577300071716308, "distill_lambda": 0.05, "distill_loss": 0.042270025610923766, "distill_only": 0.0, "epoch": 0.6636500754147813, "grad_norm": 0.5802809596061707, "learning_rate": 1.59723272042446e-05, "loss": 0.17707055807113647, "sft_loss": 0.0034199535846710207, "step": 440, "token_acc": 0.9575064314590224 }, { "cos_sim": 0.9593218803405762, "distill_lambda": 0.05, "distill_loss": 0.04067819714546204, "distill_only": 0.0, "epoch": 0.6787330316742082, "grad_norm": 0.7270879149436951, "learning_rate": 1.5770348054420147e-05, "loss": 0.1737990379333496, "sft_loss": 0.0033973097801208497, "step": 450, "token_acc": 0.9572926025756214 }, { "cos_sim": 0.9563617706298828, "distill_lambda": 0.05, "distill_loss": 0.04363825023174286, "distill_only": 0.0, "epoch": 0.693815987933635, "grad_norm": 0.973190188407898, "learning_rate": 1.556477614829251e-05, "loss": 0.17760031223297118, "sft_loss": 0.003368097543716431, "step": 460, "token_acc": 0.9580279813457695 }, { "cos_sim": 0.9571840286254882, "distill_lambda": 0.05, "distill_loss": 0.04281598031520843, "distill_only": 0.0, "epoch": 0.7088989441930619, "grad_norm": 0.7574863433837891, "learning_rate": 1.5355739479832675e-05, "loss": 0.18030910491943358, "sft_loss": 0.003493860363960266, "step": 470, "token_acc": 0.9569184135892319 }, { "cos_sim": 0.9575754165649414, "distill_lambda": 0.05, "distill_loss": 0.042424646019935605, "distill_only": 0.0, "epoch": 0.7239819004524887, "grad_norm": 0.9733440279960632, "learning_rate": 1.5143368200255294e-05, "loss": 0.17481188774108886, "sft_loss": 0.003341639041900635, "step": 480, "token_acc": 0.9575770794709209 }, { "cos_sim": 0.9553702354431153, "distill_lambda": 0.05, "distill_loss": 0.044629788398742674, "distill_only": 0.0, "epoch": 0.7390648567119156, "grad_norm": 0.7730650901794434, "learning_rate": 1.4927794536983411e-05, "loss": 0.18388822078704833, "sft_loss": 0.003515017032623291, "step": 490, "token_acc": 0.9564489371928246 }, { "cos_sim": 0.9576074600219726, "distill_lambda": 0.05, "distill_loss": 0.04239253103733063, "distill_only": 0.0, "epoch": 0.7541478129713424, "grad_norm": 0.6918530464172363, "learning_rate": 1.4709152711320529e-05, "loss": 0.17626463174819945, "sft_loss": 0.003388643264770508, "step": 500, "token_acc": 0.9582675731455627 }, { "cos_sim": 0.9582037925720215, "distill_lambda": 0.05, "distill_loss": 0.04179618954658508, "distill_only": 0.0, "epoch": 0.7692307692307693, "grad_norm": 0.6755518913269043, "learning_rate": 1.4487578854881217e-05, "loss": 0.17306239604949952, "sft_loss": 0.0033183902502059936, "step": 510, "token_acc": 0.9585653682992673 }, { "cos_sim": 0.9581752777099609, "distill_lambda": 0.05, "distill_loss": 0.04182468056678772, "distill_only": 0.0, "epoch": 0.7843137254901961, "grad_norm": 0.6448860168457031, "learning_rate": 1.4263210924832332e-05, "loss": 0.1756377935409546, "sft_loss": 0.003397446870803833, "step": 520, "token_acc": 0.9570566281159286 }, { "cos_sim": 0.9568546295166016, "distill_lambda": 0.05, "distill_loss": 0.04314538836479187, "distill_only": 0.0, "epoch": 0.799396681749623, "grad_norm": 0.6447641253471375, "learning_rate": 1.4036188617997581e-05, "loss": 0.17713675498962403, "sft_loss": 0.003378254175186157, "step": 530, "token_acc": 0.9574316713108436 }, { "cos_sim": 0.9590435028076172, "distill_lambda": 0.05, "distill_loss": 0.040956521034240724, "distill_only": 0.0, "epoch": 0.8144796380090498, "grad_norm": 0.5363443493843079, "learning_rate": 1.3806653283878963e-05, "loss": 0.1698854923248291, "sft_loss": 0.0032610952854156493, "step": 540, "token_acc": 0.9596377301529898 }, { "cos_sim": 0.9582330703735351, "distill_lambda": 0.05, "distill_loss": 0.04176701903343201, "distill_only": 0.0, "epoch": 0.8295625942684767, "grad_norm": 0.5748865008354187, "learning_rate": 1.3574747836649197e-05, "loss": 0.1734813928604126, "sft_loss": 0.0033329427242279053, "step": 550, "token_acc": 0.9590220282750994 }, { "cos_sim": 0.9610880851745606, "distill_lambda": 0.05, "distill_loss": 0.03891191482543945, "distill_only": 0.0, "epoch": 0.8446455505279035, "grad_norm": 0.8104287385940552, "learning_rate": 1.3340616666169925e-05, "loss": 0.1647010326385498, "sft_loss": 0.0032013118267059325, "step": 560, "token_acc": 0.9608597801624319 }, { "cos_sim": 0.9566938400268554, "distill_lambda": 0.05, "distill_loss": 0.04330608248710632, "distill_only": 0.0, "epoch": 0.8597285067873304, "grad_norm": 0.6440488696098328, "learning_rate": 1.3104405548091188e-05, "loss": 0.1752454161643982, "sft_loss": 0.0033111155033111574, "step": 570, "token_acc": 0.9586294820717132 }, { "cos_sim": 0.9602250099182129, "distill_lambda": 0.05, "distill_loss": 0.03977500200271607, "distill_only": 0.0, "epoch": 0.8748114630467572, "grad_norm": 0.5489897727966309, "learning_rate": 1.2866261553087977e-05, "loss": 0.16527974605560303, "sft_loss": 0.0031762421131134033, "step": 580, "token_acc": 0.9606163061769565 }, { "cos_sim": 0.9600116729736328, "distill_lambda": 0.05, "distill_loss": 0.03998832404613495, "distill_only": 0.0, "epoch": 0.889894419306184, "grad_norm": 0.6131101250648499, "learning_rate": 1.2626332955290564e-05, "loss": 0.16689059734344483, "sft_loss": 0.0032159149646759032, "step": 590, "token_acc": 0.9592660436523953 }, { "cos_sim": 0.9574895858764648, "distill_lambda": 0.05, "distill_loss": 0.042510432004928586, "distill_only": 0.0, "epoch": 0.9049773755656109, "grad_norm": 0.6989392042160034, "learning_rate": 1.2384769139965476e-05, "loss": 0.17427356243133546, "sft_loss": 0.0033205270767211912, "step": 600, "token_acc": 0.9589200005800211 }, { "cos_sim": 0.960666275024414, "distill_lambda": 0.05, "distill_loss": 0.03933373689651489, "distill_only": 0.0, "epoch": 0.9200603318250377, "grad_norm": 0.614352285861969, "learning_rate": 1.2141720510504662e-05, "loss": 0.16761900186538697, "sft_loss": 0.003271406888961792, "step": 610, "token_acc": 0.9594916600476568 }, { "cos_sim": 0.9618660926818847, "distill_lambda": 0.05, "distill_loss": 0.03813394904136658, "distill_only": 0.0, "epoch": 0.9351432880844646, "grad_norm": 0.6415292024612427, "learning_rate": 1.1897338394780762e-05, "loss": 0.1617180347442627, "sft_loss": 0.0031469911336898804, "step": 620, "token_acc": 0.9609395109395109 }, { "cos_sim": 0.9555173873901367, "distill_lambda": 0.05, "distill_loss": 0.04448263049125671, "distill_only": 0.0, "epoch": 0.9502262443438914, "grad_norm": 0.526348888874054, "learning_rate": 1.165177495092677e-05, "loss": 0.17745519876480104, "sft_loss": 0.0033213436603546143, "step": 630, "token_acc": 0.9584856927273783 }, { "cos_sim": 0.9585472106933594, "distill_lambda": 0.05, "distill_loss": 0.041452842950820926, "distill_only": 0.0, "epoch": 0.9653092006033183, "grad_norm": 0.647962749004364, "learning_rate": 1.1405183072598735e-05, "loss": 0.16704519987106323, "sft_loss": 0.003147520124912262, "step": 640, "token_acc": 0.9602409267360897 }, { "cos_sim": 0.9609892845153809, "distill_lambda": 0.05, "distill_loss": 0.039010798931121825, "distill_only": 0.0, "epoch": 0.9803921568627451, "grad_norm": 0.5828487873077393, "learning_rate": 1.115771629378054e-05, "loss": 0.16383100748062135, "sft_loss": 0.0031691789627075195, "step": 650, "token_acc": 0.9605183235472767 }, { "cos_sim": 0.9588005065917968, "distill_lambda": 0.05, "distill_loss": 0.04119955003261566, "distill_only": 0.0, "epoch": 0.995475113122172, "grad_norm": 0.6482275724411011, "learning_rate": 1.0909528693189974e-05, "loss": 0.16884152889251708, "sft_loss": 0.003216320276260376, "step": 660, "token_acc": 0.9604709050205864 }, { "cos_sim": 0.9610332489013672, "distill_lambda": 0.05, "distill_loss": 0.038966789841651917, "distill_only": 0.0, "epoch": 1.0105580693815988, "grad_norm": 0.5082425475120544, "learning_rate": 1.0660774798345655e-05, "loss": 0.16101839542388915, "sft_loss": 0.0030834853649139406, "step": 670, "token_acc": 0.9619462442260595 }, { "cos_sim": 0.9571125030517578, "distill_lambda": 0.05, "distill_loss": 0.042887437343597415, "distill_only": 0.0, "epoch": 1.0256410256410255, "grad_norm": 0.5533788204193115, "learning_rate": 1.0411609489354495e-05, "loss": 0.16550883054733276, "sft_loss": 0.00302777886390686, "step": 680, "token_acc": 0.9631733975269509 }, { "cos_sim": 0.9615534782409668, "distill_lambda": 0.05, "distill_loss": 0.03844650983810425, "distill_only": 0.0, "epoch": 1.0407239819004526, "grad_norm": 0.9731346964836121, "learning_rate": 1.016218790247969e-05, "loss": 0.16046689748764037, "sft_loss": 0.0030922651290893554, "step": 690, "token_acc": 0.961387250772255 }, { "cos_sim": 0.9571468353271484, "distill_lambda": 0.05, "distill_loss": 0.04285323321819305, "distill_only": 0.0, "epoch": 1.0558069381598794, "grad_norm": 0.7573302984237671, "learning_rate": 9.91266533354913e-06, "loss": 0.16289539337158204, "sft_loss": 0.0029478192329406737, "step": 700, "token_acc": 0.9636635141861623 }, { "cos_sim": 0.9576352119445801, "distill_lambda": 0.05, "distill_loss": 0.04236477613449097, "distill_only": 0.0, "epoch": 1.0708898944193062, "grad_norm": 0.6214897632598877, "learning_rate": 9.663197141264556e-06, "loss": 0.16363902091979982, "sft_loss": 0.0029954805970191954, "step": 710, "token_acc": 0.962718180891794 }, { "cos_sim": 0.9617841720581055, "distill_lambda": 0.05, "distill_loss": 0.038215866684913634, "distill_only": 0.0, "epoch": 1.085972850678733, "grad_norm": 0.5831133127212524, "learning_rate": 9.413938650471515e-06, "loss": 0.1560113787651062, "sft_loss": 0.002964562177658081, "step": 720, "token_acc": 0.9631175949061014 }, { "cos_sim": 0.9575637817382813, "distill_lambda": 0.05, "distill_loss": 0.04243628680706024, "distill_only": 0.0, "epoch": 1.10105580693816, "grad_norm": 0.5310092568397522, "learning_rate": 9.165045055450393e-06, "loss": 0.16315915584564208, "sft_loss": 0.0029769092798233032, "step": 730, "token_acc": 0.9631772911400116 }, { "cos_sim": 0.958311653137207, "distill_lambda": 0.05, "distill_loss": 0.041688340902328494, "distill_only": 0.0, "epoch": 1.1161387631975868, "grad_norm": 0.6231887340545654, "learning_rate": 8.916671323288777e-06, "loss": 0.1598809003829956, "sft_loss": 0.0029118612408638002, "step": 740, "token_acc": 0.963957658997463 }, { "cos_sim": 0.9586709976196289, "distill_lambda": 0.05, "distill_loss": 0.04132902026176453, "distill_only": 0.0, "epoch": 1.1312217194570136, "grad_norm": 0.6037419438362122, "learning_rate": 8.66897209739523e-06, "loss": 0.15905189514160156, "sft_loss": 0.0029039204120635985, "step": 750, "token_acc": 0.9639000981568328 }, { "cos_sim": 0.9582222938537598, "distill_lambda": 0.05, "distill_loss": 0.041777703166008, "distill_only": 0.0, "epoch": 1.1463046757164403, "grad_norm": 0.5308356285095215, "learning_rate": 8.422101601214643e-06, "loss": 0.15644775629043578, "sft_loss": 0.0028001070022583008, "step": 760, "token_acc": 0.9649918807909065 }, { "cos_sim": 0.9610555648803711, "distill_lambda": 0.05, "distill_loss": 0.03894454836845398, "distill_only": 0.0, "epoch": 1.1613876319758674, "grad_norm": 0.5524260401725769, "learning_rate": 8.176213542205016e-06, "loss": 0.1585470199584961, "sft_loss": 0.003007367253303528, "step": 770, "token_acc": 0.9628490978048501 }, { "cos_sim": 0.9591650009155274, "distill_lambda": 0.05, "distill_loss": 0.04083501696586609, "distill_only": 0.0, "epoch": 1.1764705882352942, "grad_norm": 0.7071956396102905, "learning_rate": 7.931461016135534e-06, "loss": 0.15953245162963867, "sft_loss": 0.0029436379671096803, "step": 780, "token_acc": 0.9639978922571446 }, { "cos_sim": 0.9610062599182129, "distill_lambda": 0.05, "distill_loss": 0.0389937698841095, "distill_only": 0.0, "epoch": 1.191553544494721, "grad_norm": 0.5717310309410095, "learning_rate": 7.687996411765503e-06, "loss": 0.15766040086746216, "sft_loss": 0.0029771983623504638, "step": 790, "token_acc": 0.9624405086993836 }, { "cos_sim": 0.9629199028015136, "distill_lambda": 0.05, "distill_loss": 0.037080177664756776, "distill_only": 0.0, "epoch": 1.2066365007541477, "grad_norm": 0.5949327349662781, "learning_rate": 7.445971315963459e-06, "loss": 0.15318033695220948, "sft_loss": 0.0029328763484954834, "step": 800, "token_acc": 0.9633663827212214 }, { "cos_sim": 0.9606420516967773, "distill_lambda": 0.05, "distill_loss": 0.03935791254043579, "distill_only": 0.0, "epoch": 1.2217194570135748, "grad_norm": 0.5595757961273193, "learning_rate": 7.205536419325581e-06, "loss": 0.1577266812324524, "sft_loss": 0.0029610633850097657, "step": 810, "token_acc": 0.9631494902346149 }, { "cos_sim": 0.9623119354248046, "distill_lambda": 0.05, "distill_loss": 0.037688106298446655, "distill_only": 0.0, "epoch": 1.2368024132730016, "grad_norm": 0.6127773523330688, "learning_rate": 6.966841422352127e-06, "loss": 0.15100159645080566, "sft_loss": 0.0028343945741653442, "step": 820, "token_acc": 0.9652296541161842 }, { "cos_sim": 0.9584258079528809, "distill_lambda": 0.05, "distill_loss": 0.04157418310642243, "distill_only": 0.0, "epoch": 1.2518853695324283, "grad_norm": 0.5361637473106384, "learning_rate": 6.7300349422403344e-06, "loss": 0.16251258850097655, "sft_loss": 0.002999809384346008, "step": 830, "token_acc": 0.9619352999857489 }, { "cos_sim": 0.9567733764648437, "distill_lambda": 0.05, "distill_loss": 0.043226572871208194, "distill_only": 0.0, "epoch": 1.2669683257918551, "grad_norm": 0.598453938961029, "learning_rate": 6.495264420351799e-06, "loss": 0.16118255853652955, "sft_loss": 0.0028756260871887207, "step": 840, "token_acc": 0.964183483102864 }, { "cos_sim": 0.9598235130310059, "distill_lambda": 0.05, "distill_loss": 0.04017650187015533, "distill_only": 0.0, "epoch": 1.282051282051282, "grad_norm": 0.5505294799804688, "learning_rate": 6.262676030411985e-06, "loss": 0.1548150897026062, "sft_loss": 0.002829146385192871, "step": 850, "token_acc": 0.9647297377301521 }, { "cos_sim": 0.9622200965881348, "distill_lambda": 0.05, "distill_loss": 0.03777988851070404, "distill_only": 0.0, "epoch": 1.297134238310709, "grad_norm": 0.761867344379425, "learning_rate": 6.0324145874989606e-06, "loss": 0.15253689289093017, "sft_loss": 0.0028777837753295897, "step": 860, "token_acc": 0.9640442863522274 }, { "cos_sim": 0.958277416229248, "distill_lambda": 0.05, "distill_loss": 0.04172258675098419, "distill_only": 0.0, "epoch": 1.3122171945701357, "grad_norm": 0.6454381942749023, "learning_rate": 5.804623457878088e-06, "loss": 0.15913302898406984, "sft_loss": 0.002886778116226196, "step": 870, "token_acc": 0.964312429871587 }, { "cos_sim": 0.9604253768920898, "distill_lambda": 0.05, "distill_loss": 0.039574623107910156, "distill_only": 0.0, "epoch": 1.3273001508295625, "grad_norm": 0.5452210307121277, "learning_rate": 5.579444469738771e-06, "loss": 0.15630502700805665, "sft_loss": 0.002905800938606262, "step": 880, "token_acc": 0.9635262573964497 }, { "cos_sim": 0.9618984222412109, "distill_lambda": 0.05, "distill_loss": 0.03810164928436279, "distill_only": 0.0, "epoch": 1.3423831070889896, "grad_norm": 0.7685712575912476, "learning_rate": 5.357017824888831e-06, "loss": 0.14914145469665527, "sft_loss": 0.0027555882930755614, "step": 890, "token_acc": 0.9653910532016148 }, { "cos_sim": 0.9619590759277343, "distill_lambda": 0.05, "distill_loss": 0.03804098665714264, "distill_only": 0.0, "epoch": 1.3574660633484164, "grad_norm": 0.5975191593170166, "learning_rate": 5.137482011461513e-06, "loss": 0.15076780319213867, "sft_loss": 0.002809444069862366, "step": 900, "token_acc": 0.9649688793300094 }, { "cos_sim": 0.9614360809326172, "distill_lambda": 0.05, "distill_loss": 0.0385638564825058, "distill_only": 0.0, "epoch": 1.3725490196078431, "grad_norm": 0.5798740386962891, "learning_rate": 4.920973717689495e-06, "loss": 0.1523332715034485, "sft_loss": 0.0028322219848632814, "step": 910, "token_acc": 0.9649694993054297 }, { "cos_sim": 0.957396125793457, "distill_lambda": 0.05, "distill_loss": 0.04260391592979431, "distill_only": 0.0, "epoch": 1.38763197586727, "grad_norm": 0.6167722344398499, "learning_rate": 4.707627746799517e-06, "loss": 0.1587763786315918, "sft_loss": 0.002831566333770752, "step": 920, "token_acc": 0.9646261253556824 }, { "cos_sim": 0.9588289260864258, "distill_lambda": 0.05, "distill_loss": 0.04117104709148407, "distill_only": 0.0, "epoch": 1.4027149321266967, "grad_norm": 0.5638949871063232, "learning_rate": 4.497576933080677e-06, "loss": 0.1576831817626953, "sft_loss": 0.002869047224521637, "step": 930, "token_acc": 0.9636541938647298 }, { "cos_sim": 0.9586837768554688, "distill_lambda": 0.05, "distill_loss": 0.041316288709640506, "distill_only": 0.0, "epoch": 1.4177978883861238, "grad_norm": 0.5917676091194153, "learning_rate": 4.290952059178644e-06, "loss": 0.1582876443862915, "sft_loss": 0.002880674600601196, "step": 940, "token_acc": 0.9639145496535797 }, { "cos_sim": 0.9565746307373046, "distill_lambda": 0.05, "distill_loss": 0.043425387144088744, "distill_only": 0.0, "epoch": 1.4328808446455505, "grad_norm": 0.649226188659668, "learning_rate": 4.087881774667258e-06, "loss": 0.16422024965286255, "sft_loss": 0.0029606133699417113, "step": 950, "token_acc": 0.9628297540313592 }, { "cos_sim": 0.9604790687561036, "distill_lambda": 0.05, "distill_loss": 0.03952091634273529, "distill_only": 0.0, "epoch": 1.4479638009049773, "grad_norm": 0.5247393846511841, "learning_rate": 3.888492515948241e-06, "loss": 0.15257443189620973, "sft_loss": 0.002791905403137207, "step": 960, "token_acc": 0.9648818123136508 }, { "cos_sim": 0.9606393814086914, "distill_lambda": 0.05, "distill_loss": 0.0393606573343277, "distill_only": 0.0, "epoch": 1.4630467571644044, "grad_norm": 0.5353207588195801, "learning_rate": 3.692908427528866e-06, "loss": 0.15109989643096924, "sft_loss": 0.002753838896751404, "step": 970, "token_acc": 0.9657755933828818 }, { "cos_sim": 0.9582902908325195, "distill_lambda": 0.05, "distill_loss": 0.04170980453491211, "distill_only": 0.0, "epoch": 1.4781297134238311, "grad_norm": 0.5182374119758606, "learning_rate": 3.5012512847266435e-06, "loss": 0.15429224967956542, "sft_loss": 0.0027361422777175902, "step": 980, "token_acc": 0.9657219602277063 }, { "cos_sim": 0.9585134506225585, "distill_lambda": 0.05, "distill_loss": 0.041486534476280215, "distill_only": 0.0, "epoch": 1.493212669683258, "grad_norm": 0.5777459144592285, "learning_rate": 3.313640417849107e-06, "loss": 0.16000924110412598, "sft_loss": 0.002925962209701538, "step": 990, "token_acc": 0.9639754534951762 }, { "cos_sim": 0.9615019798278809, "distill_lambda": 0.05, "distill_loss": 0.03849804997444153, "distill_only": 0.0, "epoch": 1.5082956259426847, "grad_norm": 0.639433741569519, "learning_rate": 3.130192637895888e-06, "loss": 0.15193912982940674, "sft_loss": 0.0028231948614120483, "step": 1000, "token_acc": 0.9649682902627378 }, { "cos_sim": 0.9563374519348145, "distill_lambda": 0.05, "distill_loss": 0.04366253316402435, "distill_only": 0.0, "epoch": 1.5233785822021115, "grad_norm": 0.5977683663368225, "learning_rate": 2.9510221638294378e-06, "loss": 0.1627751111984253, "sft_loss": 0.0029035955667495728, "step": 1010, "token_acc": 0.964194334450198 }, { "cos_sim": 0.9659231185913086, "distill_lambda": 0.05, "distill_loss": 0.0340768575668335, "distill_only": 0.0, "epoch": 1.5384615384615383, "grad_norm": 0.5716925859451294, "learning_rate": 2.776240551459558e-06, "loss": 0.14257134199142457, "sft_loss": 0.0027515113353729246, "step": 1020, "token_acc": 0.9651652806486244 }, { "cos_sim": 0.958721923828125, "distill_lambda": 0.05, "distill_loss": 0.041278117895126344, "distill_only": 0.0, "epoch": 1.5535444947209653, "grad_norm": 0.5039287805557251, "learning_rate": 2.6059566239861e-06, "loss": 0.15441036224365234, "sft_loss": 0.002761417627334595, "step": 1030, "token_acc": 0.9661985061285093 }, { "cos_sim": 0.959202766418457, "distill_lambda": 0.05, "distill_loss": 0.04079728126525879, "distill_only": 0.0, "epoch": 1.5686274509803921, "grad_norm": 0.5184313058853149, "learning_rate": 2.4402764042430317e-06, "loss": 0.1532346487045288, "sft_loss": 0.0027487188577651977, "step": 1040, "token_acc": 0.9650577077251207 }, { "cos_sim": 0.961637020111084, "distill_lambda": 0.05, "distill_loss": 0.03836305141448974, "distill_only": 0.0, "epoch": 1.5837104072398192, "grad_norm": 0.7458616495132446, "learning_rate": 2.2793030486861212e-06, "loss": 0.14818975925445557, "sft_loss": 0.002712777256965637, "step": 1050, "token_acc": 0.9668444572581808 }, { "cos_sim": 0.9626045227050781, "distill_lambda": 0.05, "distill_loss": 0.03739550113677979, "distill_only": 0.0, "epoch": 1.598793363499246, "grad_norm": 0.5712209343910217, "learning_rate": 2.1231367831652415e-06, "loss": 0.14667056798934935, "sft_loss": 0.0027136802673339844, "step": 1060, "token_acc": 0.9656194052315541 }, { "cos_sim": 0.9599955558776856, "distill_lambda": 0.05, "distill_loss": 0.04000440239906311, "distill_only": 0.0, "epoch": 1.6138763197586727, "grad_norm": 0.5938213467597961, "learning_rate": 1.971874840521374e-06, "loss": 0.15558981895446777, "sft_loss": 0.0028619617223739624, "step": 1070, "token_acc": 0.9639599927038366 }, { "cos_sim": 0.9599279403686524, "distill_lambda": 0.05, "distill_loss": 0.04007203280925751, "distill_only": 0.0, "epoch": 1.6289592760180995, "grad_norm": 0.9566603899002075, "learning_rate": 1.825611400047126e-06, "loss": 0.15478403568267823, "sft_loss": 0.0028333991765975954, "step": 1080, "token_acc": 0.9649584326492646 }, { "cos_sim": 0.9623710632324218, "distill_lambda": 0.05, "distill_loss": 0.03762889504432678, "distill_only": 0.0, "epoch": 1.6440422322775263, "grad_norm": 0.5412837266921997, "learning_rate": 1.684437528848456e-06, "loss": 0.14730063676834107, "sft_loss": 0.0027217000722885134, "step": 1090, "token_acc": 0.9654343012790901 }, { "cos_sim": 0.9615623474121093, "distill_lambda": 0.05, "distill_loss": 0.03843761682510376, "distill_only": 0.0, "epoch": 1.6591251885369531, "grad_norm": 0.5399158596992493, "learning_rate": 1.5484411251441234e-06, "loss": 0.15010616779327393, "sft_loss": 0.0027689367532730104, "step": 1100, "token_acc": 0.9653759381186979 }, { "cos_sim": 0.9638436317443848, "distill_lambda": 0.05, "distill_loss": 0.036156341433525085, "distill_only": 0.0, "epoch": 1.6742081447963801, "grad_norm": 0.6399282217025757, "learning_rate": 1.4177068635381575e-06, "loss": 0.1425804853439331, "sft_loss": 0.002647823095321655, "step": 1110, "token_acc": 0.9672080357434041 }, { "cos_sim": 0.9571516036987304, "distill_lambda": 0.05, "distill_loss": 0.04284847378730774, "distill_only": 0.0, "epoch": 1.689291101055807, "grad_norm": 0.5090851783752441, "learning_rate": 1.2923161422994302e-06, "loss": 0.16026003360748292, "sft_loss": 0.002865701913833618, "step": 1120, "token_acc": 0.9648365972502405 }, { "cos_sim": 0.9603984832763672, "distill_lambda": 0.05, "distill_loss": 0.039601612091064456, "distill_only": 0.0, "epoch": 1.704374057315234, "grad_norm": 0.46891576051712036, "learning_rate": 1.1723470326811582e-06, "loss": 0.14807002544403075, "sft_loss": 0.0026471078395843505, "step": 1130, "token_acc": 0.9664151061564477 }, { "cos_sim": 0.9597763061523438, "distill_lambda": 0.05, "distill_loss": 0.040223723649978636, "distill_only": 0.0, "epoch": 1.7194570135746607, "grad_norm": 0.5507431626319885, "learning_rate": 1.0578742303118695e-06, "loss": 0.15549242496490479, "sft_loss": 0.0028479516506195067, "step": 1140, "token_acc": 0.9641867086236607 }, { "cos_sim": 0.9618812561035156, "distill_lambda": 0.05, "distill_loss": 0.03811879158020019, "distill_only": 0.0, "epoch": 1.7345399698340875, "grad_norm": 0.5255997776985168, "learning_rate": 9.489690086881321e-07, "loss": 0.1471501588821411, "sft_loss": 0.002692502737045288, "step": 1150, "token_acc": 0.9657874647514294 }, { "cos_sim": 0.95782470703125, "distill_lambda": 0.05, "distill_loss": 0.04217531681060791, "distill_only": 0.0, "epoch": 1.7496229260935143, "grad_norm": 0.5514457821846008, "learning_rate": 8.456991747979782e-07, "loss": 0.15675148963928223, "sft_loss": 0.00278971791267395, "step": 1160, "token_acc": 0.9654551943610029 }, { "cos_sim": 0.9652828216552735, "distill_lambda": 0.05, "distill_loss": 0.03471708297729492, "distill_only": 0.0, "epoch": 1.7647058823529411, "grad_norm": 0.9719615578651428, "learning_rate": 7.48129026902642e-07, "loss": 0.14254264831542968, "sft_loss": 0.0027186036109924316, "step": 1170, "token_acc": 0.9664103285504392 }, { "cos_sim": 0.9611705780029297, "distill_lambda": 0.05, "distill_loss": 0.03882928490638733, "distill_only": 0.0, "epoch": 1.779788838612368, "grad_norm": 0.7314273715019226, "learning_rate": 6.563193145029423e-07, "loss": 0.1511031985282898, "sft_loss": 0.0027805104851722716, "step": 1180, "token_acc": 0.965696334417327 }, { "cos_sim": 0.9645143508911133, "distill_lambda": 0.05, "distill_loss": 0.035485664010047914, "distill_only": 0.0, "epoch": 1.7948717948717947, "grad_norm": 0.5017873048782349, "learning_rate": 5.703272005151905e-07, "loss": 0.13897626399993895, "sft_loss": 0.0025687247514724733, "step": 1190, "token_acc": 0.9678674654050475 }, { "cos_sim": 0.9612783432006836, "distill_lambda": 0.05, "distill_loss": 0.03872166574001312, "distill_only": 0.0, "epoch": 1.8099547511312217, "grad_norm": 0.5210822820663452, "learning_rate": 4.902062256801954e-07, "loss": 0.14710806608200072, "sft_loss": 0.002661043405532837, "step": 1200, "token_acc": 0.9672804843617655 }, { "cos_sim": 0.9607630729675293, "distill_lambda": 0.05, "distill_loss": 0.039236992597579956, "distill_only": 0.0, "epoch": 1.8250377073906485, "grad_norm": 0.48953771591186523, "learning_rate": 4.1600627522751624e-07, "loss": 0.14924311637878418, "sft_loss": 0.002701997756958008, "step": 1210, "token_acc": 0.9663704800691102 }, { "cos_sim": 0.9623642921447754, "distill_lambda": 0.05, "distill_loss": 0.037635689973831175, "distill_only": 0.0, "epoch": 1.8401206636500755, "grad_norm": 0.4910859167575836, "learning_rate": 3.477735478157429e-07, "loss": 0.14284021854400636, "sft_loss": 0.002581971883773804, "step": 1220, "token_acc": 0.9684232013315328 }, { "cos_sim": 0.9573232650756835, "distill_lambda": 0.05, "distill_loss": 0.04267663657665253, "distill_only": 0.0, "epoch": 1.8552036199095023, "grad_norm": 0.47646966576576233, "learning_rate": 2.855505267680947e-07, "loss": 0.15834914445877074, "sft_loss": 0.002814579010009766, "step": 1230, "token_acc": 0.9647476199029998 }, { "cos_sim": 0.9598515510559082, "distill_lambda": 0.05, "distill_loss": 0.04014853835105896, "distill_only": 0.0, "epoch": 1.8702865761689291, "grad_norm": 0.5589637160301208, "learning_rate": 2.2937595362129805e-07, "loss": 0.15024152994155884, "sft_loss": 0.0026876211166381838, "step": 1240, "token_acc": 0.9658862034239678 }, { "cos_sim": 0.9612709045410156, "distill_lambda": 0.05, "distill_loss": 0.03872906267642975, "distill_only": 0.0, "epoch": 1.885369532428356, "grad_norm": 0.4848785698413849, "learning_rate": 1.7928480400417837e-07, "loss": 0.14641841650009155, "sft_loss": 0.0026391223073005676, "step": 1250, "token_acc": 0.9672920129818728 }, { "cos_sim": 0.9641980171203614, "distill_lambda": 0.05, "distill_loss": 0.035801997780799864, "distill_only": 0.0, "epoch": 1.9004524886877827, "grad_norm": 1.1431437730789185, "learning_rate": 1.3530826586100276e-07, "loss": 0.14064620733261107, "sft_loss": 0.002605094015598297, "step": 1260, "token_acc": 0.9675166666666667 }, { "cos_sim": 0.9595614433288574, "distill_lambda": 0.05, "distill_loss": 0.04043857753276825, "distill_only": 0.0, "epoch": 1.9155354449472095, "grad_norm": 0.4684814512729645, "learning_rate": 9.747372003312106e-08, "loss": 0.15321109294891358, "sft_loss": 0.002765917778015137, "step": 1270, "token_acc": 0.9654640225968958 }, { "cos_sim": 0.9601219177246094, "distill_lambda": 0.05, "distill_loss": 0.03987819254398346, "distill_only": 0.0, "epoch": 1.9306184012066365, "grad_norm": 0.5522783398628235, "learning_rate": 6.580472321099928e-08, "loss": 0.1522770881652832, "sft_loss": 0.0027647495269775392, "step": 1280, "token_acc": 0.9654809611486748 }, { "cos_sim": 0.9622014045715332, "distill_lambda": 0.05, "distill_loss": 0.037798523902893066, "distill_only": 0.0, "epoch": 1.9457013574660633, "grad_norm": 0.5565215349197388, "learning_rate": 4.0320993267264354e-08, "loss": 0.1477041721343994, "sft_loss": 0.002725829184055328, "step": 1290, "token_acc": 0.9660229069823166 }, { "cos_sim": 0.9629398345947265, "distill_lambda": 0.05, "distill_loss": 0.037060147523880003, "distill_only": 0.0, "epoch": 1.9607843137254903, "grad_norm": 0.48291346430778503, "learning_rate": 2.1038396979884414e-08, "loss": 0.1460159420967102, "sft_loss": 0.0027099907398223876, "step": 1300, "token_acc": 0.9656141835824731 }, { "cos_sim": 0.9619705200195312, "distill_lambda": 0.05, "distill_loss": 0.03802947998046875, "distill_only": 0.0, "epoch": 1.9758672699849171, "grad_norm": 0.5083843469619751, "learning_rate": 7.968940153136872e-09, "loss": 0.14663779735565186, "sft_loss": 0.002680957317352295, "step": 1310, "token_acc": 0.966562089829369 }, { "cos_sim": 0.9584096908569336, "distill_lambda": 0.05, "distill_loss": 0.04159027636051178, "distill_only": 0.0, "epoch": 1.990950226244344, "grad_norm": 0.5097407698631287, "learning_rate": 1.1207601425078375e-09, "loss": 0.15432920455932617, "sft_loss": 0.002743273973464966, "step": 1320, "token_acc": 0.9660405823338614 } ], "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": 8.089131042604057e+18, "train_batch_size": 2, "trial_name": null, "trial_params": null }