{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.49173976535569175, "eval_steps": 500, "global_step": 99000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 4.9670683369261796e-05, "grad_norm": 4.901270389556885, "learning_rate": 2.9801324503311263e-08, "loss": 0.29717721939086916, "step": 10 }, { "epoch": 9.934136673852359e-05, "grad_norm": 4.774776935577393, "learning_rate": 6.291390728476823e-08, "loss": 0.271336030960083, "step": 20 }, { "epoch": 0.0001490120501077854, "grad_norm": 6.1027984619140625, "learning_rate": 9.602649006622517e-08, "loss": 0.29968786239624023, "step": 30 }, { "epoch": 0.00019868273347704719, "grad_norm": 7.096562385559082, "learning_rate": 1.2913907284768212e-07, "loss": 0.26782639026641847, "step": 40 }, { "epoch": 0.000248353416846309, "grad_norm": 4.739919185638428, "learning_rate": 1.6225165562913908e-07, "loss": 0.276702356338501, "step": 50 }, { "epoch": 0.0002980241002155708, "grad_norm": 6.672008037567139, "learning_rate": 1.9536423841059604e-07, "loss": 0.27533023357391356, "step": 60 }, { "epoch": 0.00034769478358483255, "grad_norm": 6.635137557983398, "learning_rate": 2.2847682119205302e-07, "loss": 0.3171017408370972, "step": 70 }, { "epoch": 0.00039736546695409437, "grad_norm": 6.985750675201416, "learning_rate": 2.6158940397350995e-07, "loss": 0.285334038734436, "step": 80 }, { "epoch": 0.00044703615032335613, "grad_norm": 5.501302719116211, "learning_rate": 2.9470198675496694e-07, "loss": 0.2702797889709473, "step": 90 }, { "epoch": 0.000496706833692618, "grad_norm": 4.2764434814453125, "learning_rate": 3.278145695364238e-07, "loss": 0.2735286235809326, "step": 100 }, { "epoch": 0.0005463775170618798, "grad_norm": 4.287500858306885, "learning_rate": 3.609271523178808e-07, "loss": 0.25853862762451174, "step": 110 }, { "epoch": 0.0005960482004311416, "grad_norm": 5.4883198738098145, "learning_rate": 3.940397350993378e-07, "loss": 0.28962202072143556, "step": 120 }, { "epoch": 0.0006457188838004033, "grad_norm": 4.478418827056885, "learning_rate": 4.271523178807947e-07, "loss": 0.26436734199523926, "step": 130 }, { "epoch": 0.0006953895671696651, "grad_norm": 4.08826208114624, "learning_rate": 4.602649006622517e-07, "loss": 0.25210568904876707, "step": 140 }, { "epoch": 0.0007450602505389269, "grad_norm": 5.705537796020508, "learning_rate": 4.933774834437086e-07, "loss": 0.2487276554107666, "step": 150 }, { "epoch": 0.0007947309339081887, "grad_norm": 4.603724479675293, "learning_rate": 5.264900662251656e-07, "loss": 0.2741535663604736, "step": 160 }, { "epoch": 0.0008444016172774504, "grad_norm": 4.102601528167725, "learning_rate": 5.596026490066225e-07, "loss": 0.2587513208389282, "step": 170 }, { "epoch": 0.0008940723006467123, "grad_norm": 3.9175188541412354, "learning_rate": 5.927152317880795e-07, "loss": 0.24495558738708495, "step": 180 }, { "epoch": 0.0009437429840159741, "grad_norm": 3.6219232082366943, "learning_rate": 6.258278145695365e-07, "loss": 0.2088637351989746, "step": 190 }, { "epoch": 0.000993413667385236, "grad_norm": 4.794059753417969, "learning_rate": 6.589403973509935e-07, "loss": 0.22896394729614258, "step": 200 }, { "epoch": 0.0010430843507544976, "grad_norm": 3.9685440063476562, "learning_rate": 6.920529801324503e-07, "loss": 0.20394153594970704, "step": 210 }, { "epoch": 0.0010927550341237595, "grad_norm": 4.257049083709717, "learning_rate": 7.251655629139073e-07, "loss": 0.21307835578918458, "step": 220 }, { "epoch": 0.0011424257174930212, "grad_norm": 3.678680896759033, "learning_rate": 7.582781456953643e-07, "loss": 0.23205828666687012, "step": 230 }, { "epoch": 0.0011920964008622832, "grad_norm": 4.08661413192749, "learning_rate": 7.913907284768213e-07, "loss": 0.2123100757598877, "step": 240 }, { "epoch": 0.0012417670842315449, "grad_norm": 4.023794651031494, "learning_rate": 8.245033112582782e-07, "loss": 0.2124955415725708, "step": 250 }, { "epoch": 0.0012914377676008066, "grad_norm": 4.135656356811523, "learning_rate": 8.576158940397352e-07, "loss": 0.1973767399787903, "step": 260 }, { "epoch": 0.0013411084509700685, "grad_norm": 3.5855722427368164, "learning_rate": 8.907284768211922e-07, "loss": 0.2176358222961426, "step": 270 }, { "epoch": 0.0013907791343393302, "grad_norm": 3.690850019454956, "learning_rate": 9.238410596026491e-07, "loss": 0.22189640998840332, "step": 280 }, { "epoch": 0.0014404498177085921, "grad_norm": 3.2823073863983154, "learning_rate": 9.56953642384106e-07, "loss": 0.2006618022918701, "step": 290 }, { "epoch": 0.0014901205010778538, "grad_norm": 3.786754846572876, "learning_rate": 9.90066225165563e-07, "loss": 0.19255210161209108, "step": 300 }, { "epoch": 0.0015397911844471156, "grad_norm": 3.0756802558898926, "learning_rate": 1.02317880794702e-06, "loss": 0.19744149446487427, "step": 310 }, { "epoch": 0.0015894618678163775, "grad_norm": 5.836268424987793, "learning_rate": 1.056291390728477e-06, "loss": 0.20446815490722656, "step": 320 }, { "epoch": 0.0016391325511856392, "grad_norm": 3.0814549922943115, "learning_rate": 1.089403973509934e-06, "loss": 0.19874093532562256, "step": 330 }, { "epoch": 0.001688803234554901, "grad_norm": 4.48346471786499, "learning_rate": 1.1225165562913907e-06, "loss": 0.2032982587814331, "step": 340 }, { "epoch": 0.0017384739179241628, "grad_norm": 3.9577476978302, "learning_rate": 1.1556291390728477e-06, "loss": 0.20114972591400146, "step": 350 }, { "epoch": 0.0017881446012934245, "grad_norm": 3.4688820838928223, "learning_rate": 1.1887417218543047e-06, "loss": 0.20399303436279298, "step": 360 }, { "epoch": 0.0018378152846626865, "grad_norm": 4.167360305786133, "learning_rate": 1.2218543046357617e-06, "loss": 0.1976668953895569, "step": 370 }, { "epoch": 0.0018874859680319482, "grad_norm": 2.326749563217163, "learning_rate": 1.2549668874172185e-06, "loss": 0.20687432289123536, "step": 380 }, { "epoch": 0.0019371566514012099, "grad_norm": 3.1341328620910645, "learning_rate": 1.2880794701986757e-06, "loss": 0.2024765968322754, "step": 390 }, { "epoch": 0.001986827334770472, "grad_norm": 2.6481590270996094, "learning_rate": 1.3211920529801324e-06, "loss": 0.2004005193710327, "step": 400 }, { "epoch": 0.0020364980181397335, "grad_norm": 3.2916314601898193, "learning_rate": 1.3543046357615894e-06, "loss": 0.19647223949432374, "step": 410 }, { "epoch": 0.002086168701508995, "grad_norm": 3.4507641792297363, "learning_rate": 1.3874172185430464e-06, "loss": 0.18551559448242189, "step": 420 }, { "epoch": 0.0021358393848782574, "grad_norm": 2.152679920196533, "learning_rate": 1.4205298013245034e-06, "loss": 0.1984817147254944, "step": 430 }, { "epoch": 0.002185510068247519, "grad_norm": 3.8054752349853516, "learning_rate": 1.4536423841059604e-06, "loss": 0.20621390342712403, "step": 440 }, { "epoch": 0.0022351807516167808, "grad_norm": 3.86279559135437, "learning_rate": 1.4867549668874174e-06, "loss": 0.2061779499053955, "step": 450 }, { "epoch": 0.0022848514349860425, "grad_norm": 2.6772871017456055, "learning_rate": 1.5198675496688741e-06, "loss": 0.20490944385528564, "step": 460 }, { "epoch": 0.002334522118355304, "grad_norm": 2.034043312072754, "learning_rate": 1.5529801324503313e-06, "loss": 0.1755089282989502, "step": 470 }, { "epoch": 0.0023841928017245663, "grad_norm": 2.332104444503784, "learning_rate": 1.5860927152317881e-06, "loss": 0.20633258819580078, "step": 480 }, { "epoch": 0.002433863485093828, "grad_norm": 3.3892877101898193, "learning_rate": 1.6192052980132453e-06, "loss": 0.19218671321868896, "step": 490 }, { "epoch": 0.0024835341684630897, "grad_norm": 3.5697853565216064, "learning_rate": 1.652317880794702e-06, "loss": 0.2081817626953125, "step": 500 }, { "epoch": 0.0025332048518323515, "grad_norm": 2.3731367588043213, "learning_rate": 1.685430463576159e-06, "loss": 0.19222270250320433, "step": 510 }, { "epoch": 0.002582875535201613, "grad_norm": 2.4798941612243652, "learning_rate": 1.718543046357616e-06, "loss": 0.19549782276153566, "step": 520 }, { "epoch": 0.0026325462185708753, "grad_norm": 2.793424129486084, "learning_rate": 1.751655629139073e-06, "loss": 0.19979052543640136, "step": 530 }, { "epoch": 0.002682216901940137, "grad_norm": 3.191100835800171, "learning_rate": 1.78476821192053e-06, "loss": 0.2025853395462036, "step": 540 }, { "epoch": 0.0027318875853093987, "grad_norm": 1.7298438549041748, "learning_rate": 1.817880794701987e-06, "loss": 0.2074094295501709, "step": 550 }, { "epoch": 0.0027815582686786604, "grad_norm": 2.918518543243408, "learning_rate": 1.8509933774834438e-06, "loss": 0.1848048448562622, "step": 560 }, { "epoch": 0.002831228952047922, "grad_norm": 1.900816798210144, "learning_rate": 1.884105960264901e-06, "loss": 0.193809175491333, "step": 570 }, { "epoch": 0.0028808996354171843, "grad_norm": 1.9395045042037964, "learning_rate": 1.9172185430463578e-06, "loss": 0.196250581741333, "step": 580 }, { "epoch": 0.002930570318786446, "grad_norm": 2.019723415374756, "learning_rate": 1.950331125827815e-06, "loss": 0.1892547607421875, "step": 590 }, { "epoch": 0.0029802410021557077, "grad_norm": 3.260744571685791, "learning_rate": 1.9834437086092717e-06, "loss": 0.19586299657821654, "step": 600 }, { "epoch": 0.0030299116855249694, "grad_norm": 2.1798887252807617, "learning_rate": 2.0165562913907285e-06, "loss": 0.18963403701782228, "step": 610 }, { "epoch": 0.003079582368894231, "grad_norm": 3.566533327102661, "learning_rate": 2.0496688741721857e-06, "loss": 0.20769989490509033, "step": 620 }, { "epoch": 0.0031292530522634933, "grad_norm": 4.243162155151367, "learning_rate": 2.0827814569536425e-06, "loss": 0.18993401527404785, "step": 630 }, { "epoch": 0.003178923735632755, "grad_norm": 2.696281909942627, "learning_rate": 2.1158940397350993e-06, "loss": 0.18575761318206788, "step": 640 }, { "epoch": 0.0032285944190020167, "grad_norm": 2.396850347518921, "learning_rate": 2.1490066225165565e-06, "loss": 0.2081833600997925, "step": 650 }, { "epoch": 0.0032782651023712784, "grad_norm": 3.6090505123138428, "learning_rate": 2.1821192052980132e-06, "loss": 0.18878703117370604, "step": 660 }, { "epoch": 0.00332793578574054, "grad_norm": 2.408635139465332, "learning_rate": 2.2152317880794704e-06, "loss": 0.19299166202545165, "step": 670 }, { "epoch": 0.003377606469109802, "grad_norm": 3.48075270652771, "learning_rate": 2.248344370860927e-06, "loss": 0.19772933721542357, "step": 680 }, { "epoch": 0.003427277152479064, "grad_norm": 2.138187885284424, "learning_rate": 2.281456953642384e-06, "loss": 0.20521624088287355, "step": 690 }, { "epoch": 0.0034769478358483256, "grad_norm": 3.3127388954162598, "learning_rate": 2.314569536423841e-06, "loss": 0.1893216848373413, "step": 700 }, { "epoch": 0.0035266185192175874, "grad_norm": 1.9119617938995361, "learning_rate": 2.347682119205298e-06, "loss": 0.19215447902679444, "step": 710 }, { "epoch": 0.003576289202586849, "grad_norm": 2.291130542755127, "learning_rate": 2.380794701986755e-06, "loss": 0.19218604564666747, "step": 720 }, { "epoch": 0.0036259598859561108, "grad_norm": 2.707551956176758, "learning_rate": 2.413907284768212e-06, "loss": 0.18764326572418213, "step": 730 }, { "epoch": 0.003675630569325373, "grad_norm": 2.8797173500061035, "learning_rate": 2.447019867549669e-06, "loss": 0.20091838836669923, "step": 740 }, { "epoch": 0.0037253012526946346, "grad_norm": 3.9512650966644287, "learning_rate": 2.480132450331126e-06, "loss": 0.1914796471595764, "step": 750 }, { "epoch": 0.0037749719360638963, "grad_norm": 3.001540422439575, "learning_rate": 2.513245033112583e-06, "loss": 0.19033777713775635, "step": 760 }, { "epoch": 0.003824642619433158, "grad_norm": 3.623997926712036, "learning_rate": 2.54635761589404e-06, "loss": 0.181196928024292, "step": 770 }, { "epoch": 0.0038743133028024197, "grad_norm": 1.9239459037780762, "learning_rate": 2.5794701986754966e-06, "loss": 0.19171653985977172, "step": 780 }, { "epoch": 0.0039239839861716815, "grad_norm": 1.9946714639663696, "learning_rate": 2.612582781456954e-06, "loss": 0.19114352464675904, "step": 790 }, { "epoch": 0.003973654669540944, "grad_norm": 1.9327802658081055, "learning_rate": 2.645695364238411e-06, "loss": 0.19364423751831056, "step": 800 }, { "epoch": 0.004023325352910206, "grad_norm": 1.8982785940170288, "learning_rate": 2.6788079470198682e-06, "loss": 0.19063801765441896, "step": 810 }, { "epoch": 0.004072996036279467, "grad_norm": 2.2981839179992676, "learning_rate": 2.7119205298013246e-06, "loss": 0.1760612964630127, "step": 820 }, { "epoch": 0.004122666719648729, "grad_norm": 3.774841070175171, "learning_rate": 2.7450331125827818e-06, "loss": 0.18398233652114868, "step": 830 }, { "epoch": 0.00417233740301799, "grad_norm": 2.886324405670166, "learning_rate": 2.778145695364239e-06, "loss": 0.19504904747009277, "step": 840 }, { "epoch": 0.004222008086387253, "grad_norm": 1.7509461641311646, "learning_rate": 2.8112582781456953e-06, "loss": 0.20129125118255614, "step": 850 }, { "epoch": 0.004271678769756515, "grad_norm": 2.3611929416656494, "learning_rate": 2.8443708609271525e-06, "loss": 0.1928415536880493, "step": 860 }, { "epoch": 0.004321349453125776, "grad_norm": 2.874192237854004, "learning_rate": 2.8774834437086097e-06, "loss": 0.1941465377807617, "step": 870 }, { "epoch": 0.004371020136495038, "grad_norm": 2.394481897354126, "learning_rate": 2.910596026490066e-06, "loss": 0.1841208219528198, "step": 880 }, { "epoch": 0.004420690819864299, "grad_norm": 3.6327292919158936, "learning_rate": 2.9437086092715233e-06, "loss": 0.1830369472503662, "step": 890 }, { "epoch": 0.0044703615032335615, "grad_norm": 2.699270009994507, "learning_rate": 2.9768211920529805e-06, "loss": 0.1947508215904236, "step": 900 }, { "epoch": 0.004520032186602824, "grad_norm": 2.001339912414551, "learning_rate": 3.0099337748344377e-06, "loss": 0.1880634069442749, "step": 910 }, { "epoch": 0.004569702869972085, "grad_norm": 2.702805519104004, "learning_rate": 3.043046357615894e-06, "loss": 0.18173846006393432, "step": 920 }, { "epoch": 0.004619373553341347, "grad_norm": 2.357896566390991, "learning_rate": 3.076158940397351e-06, "loss": 0.1867668867111206, "step": 930 }, { "epoch": 0.004669044236710608, "grad_norm": 2.557943820953369, "learning_rate": 3.1092715231788084e-06, "loss": 0.19335694313049318, "step": 940 }, { "epoch": 0.0047187149200798705, "grad_norm": 1.7541555166244507, "learning_rate": 3.142384105960265e-06, "loss": 0.17601248025894164, "step": 950 }, { "epoch": 0.004768385603449133, "grad_norm": 1.9294695854187012, "learning_rate": 3.175496688741722e-06, "loss": 0.18531960248947144, "step": 960 }, { "epoch": 0.004818056286818394, "grad_norm": 2.9450271129608154, "learning_rate": 3.208609271523179e-06, "loss": 0.18386194705963135, "step": 970 }, { "epoch": 0.004867726970187656, "grad_norm": 2.49463152885437, "learning_rate": 3.241721854304636e-06, "loss": 0.19056713581085205, "step": 980 }, { "epoch": 0.004917397653556917, "grad_norm": 1.7094453573226929, "learning_rate": 3.274834437086093e-06, "loss": 0.18888884782791138, "step": 990 }, { "epoch": 0.0049670683369261795, "grad_norm": 5.2370076179504395, "learning_rate": 3.30794701986755e-06, "loss": 0.19682384729385377, "step": 1000 }, { "epoch": 0.005016739020295442, "grad_norm": 2.3159408569335938, "learning_rate": 3.3410596026490067e-06, "loss": 0.19584348201751708, "step": 1010 }, { "epoch": 0.005066409703664703, "grad_norm": 2.086446523666382, "learning_rate": 3.374172185430464e-06, "loss": 0.1884816288948059, "step": 1020 }, { "epoch": 0.005116080387033965, "grad_norm": 2.1157519817352295, "learning_rate": 3.407284768211921e-06, "loss": 0.18591091632843018, "step": 1030 }, { "epoch": 0.005165751070403226, "grad_norm": 3.9170103073120117, "learning_rate": 3.440397350993378e-06, "loss": 0.20570816993713378, "step": 1040 }, { "epoch": 0.0052154217537724885, "grad_norm": 2.287094831466675, "learning_rate": 3.4735099337748346e-06, "loss": 0.1860327124595642, "step": 1050 }, { "epoch": 0.005265092437141751, "grad_norm": 2.2751879692077637, "learning_rate": 3.506622516556292e-06, "loss": 0.18944721221923827, "step": 1060 }, { "epoch": 0.005314763120511012, "grad_norm": 3.782728910446167, "learning_rate": 3.539735099337749e-06, "loss": 0.17933911085128784, "step": 1070 }, { "epoch": 0.005364433803880274, "grad_norm": 2.8770065307617188, "learning_rate": 3.5728476821192054e-06, "loss": 0.16828861236572265, "step": 1080 }, { "epoch": 0.005414104487249535, "grad_norm": 1.9759399890899658, "learning_rate": 3.6059602649006626e-06, "loss": 0.18992364406585693, "step": 1090 }, { "epoch": 0.0054637751706187974, "grad_norm": 1.7432211637496948, "learning_rate": 3.6390728476821198e-06, "loss": 0.18240160942077638, "step": 1100 }, { "epoch": 0.00551344585398806, "grad_norm": 1.6426855325698853, "learning_rate": 3.672185430463576e-06, "loss": 0.19775619506835937, "step": 1110 }, { "epoch": 0.005563116537357321, "grad_norm": 2.234701156616211, "learning_rate": 3.7052980132450333e-06, "loss": 0.18301160335540773, "step": 1120 }, { "epoch": 0.005612787220726583, "grad_norm": 2.6606202125549316, "learning_rate": 3.7384105960264905e-06, "loss": 0.174980092048645, "step": 1130 }, { "epoch": 0.005662457904095844, "grad_norm": 4.484838008880615, "learning_rate": 3.7715231788079477e-06, "loss": 0.18564072847366334, "step": 1140 }, { "epoch": 0.005712128587465106, "grad_norm": 2.9550058841705322, "learning_rate": 3.804635761589404e-06, "loss": 0.18100292682647706, "step": 1150 }, { "epoch": 0.0057617992708343686, "grad_norm": 3.566561698913574, "learning_rate": 3.837748344370861e-06, "loss": 0.2025855302810669, "step": 1160 }, { "epoch": 0.00581146995420363, "grad_norm": 2.428694248199463, "learning_rate": 3.8708609271523184e-06, "loss": 0.17671436071395874, "step": 1170 }, { "epoch": 0.005861140637572892, "grad_norm": 1.4793609380722046, "learning_rate": 3.903973509933775e-06, "loss": 0.18337156772613525, "step": 1180 }, { "epoch": 0.005910811320942153, "grad_norm": 2.798325777053833, "learning_rate": 3.937086092715232e-06, "loss": 0.17586904764175415, "step": 1190 }, { "epoch": 0.005960482004311415, "grad_norm": 2.048156261444092, "learning_rate": 3.970198675496689e-06, "loss": 0.17542786598205568, "step": 1200 }, { "epoch": 0.0060101526876806775, "grad_norm": 3.105327606201172, "learning_rate": 4.0033112582781455e-06, "loss": 0.19959595203399658, "step": 1210 }, { "epoch": 0.006059823371049939, "grad_norm": 2.558241367340088, "learning_rate": 4.036423841059603e-06, "loss": 0.20176634788513184, "step": 1220 }, { "epoch": 0.006109494054419201, "grad_norm": 2.749053716659546, "learning_rate": 4.06953642384106e-06, "loss": 0.1958804488182068, "step": 1230 }, { "epoch": 0.006159164737788462, "grad_norm": 3.022240400314331, "learning_rate": 4.102649006622516e-06, "loss": 0.18517439365386962, "step": 1240 }, { "epoch": 0.006208835421157724, "grad_norm": 2.458303689956665, "learning_rate": 4.1357615894039735e-06, "loss": 0.17090067863464356, "step": 1250 }, { "epoch": 0.0062585061045269865, "grad_norm": 2.8880913257598877, "learning_rate": 4.168874172185431e-06, "loss": 0.19426097869873046, "step": 1260 }, { "epoch": 0.006308176787896248, "grad_norm": 2.2417047023773193, "learning_rate": 4.201986754966888e-06, "loss": 0.20269179344177246, "step": 1270 }, { "epoch": 0.00635784747126551, "grad_norm": 1.9885133504867554, "learning_rate": 4.235099337748344e-06, "loss": 0.18810005187988282, "step": 1280 }, { "epoch": 0.006407518154634771, "grad_norm": 2.2882509231567383, "learning_rate": 4.2682119205298014e-06, "loss": 0.18725730180740358, "step": 1290 }, { "epoch": 0.006457188838004033, "grad_norm": 5.343250274658203, "learning_rate": 4.301324503311259e-06, "loss": 0.18258235454559327, "step": 1300 }, { "epoch": 0.0065068595213732955, "grad_norm": 2.3401052951812744, "learning_rate": 4.334437086092716e-06, "loss": 0.17985385656356812, "step": 1310 }, { "epoch": 0.006556530204742557, "grad_norm": 1.5908796787261963, "learning_rate": 4.367549668874172e-06, "loss": 0.17719674110412598, "step": 1320 }, { "epoch": 0.006606200888111819, "grad_norm": 2.955873489379883, "learning_rate": 4.400662251655629e-06, "loss": 0.1920550584793091, "step": 1330 }, { "epoch": 0.00665587157148108, "grad_norm": 1.9207123517990112, "learning_rate": 4.4337748344370866e-06, "loss": 0.1567138671875, "step": 1340 }, { "epoch": 0.006705542254850342, "grad_norm": 2.4794676303863525, "learning_rate": 4.466887417218544e-06, "loss": 0.18398749828338623, "step": 1350 }, { "epoch": 0.006755212938219604, "grad_norm": 3.0565361976623535, "learning_rate": 4.5e-06, "loss": 0.18242337703704833, "step": 1360 }, { "epoch": 0.006804883621588866, "grad_norm": 2.3667056560516357, "learning_rate": 4.533112582781457e-06, "loss": 0.17548438310623168, "step": 1370 }, { "epoch": 0.006854554304958128, "grad_norm": 3.991516351699829, "learning_rate": 4.5662251655629145e-06, "loss": 0.1812723994255066, "step": 1380 }, { "epoch": 0.006904224988327389, "grad_norm": 3.1555089950561523, "learning_rate": 4.599337748344372e-06, "loss": 0.18013391494750977, "step": 1390 }, { "epoch": 0.006953895671696651, "grad_norm": 2.9473206996917725, "learning_rate": 4.632450331125828e-06, "loss": 0.18811894655227662, "step": 1400 }, { "epoch": 0.007003566355065913, "grad_norm": 3.460946798324585, "learning_rate": 4.665562913907285e-06, "loss": 0.18089066743850707, "step": 1410 }, { "epoch": 0.007053237038435175, "grad_norm": 3.395982503890991, "learning_rate": 4.6986754966887425e-06, "loss": 0.1825493335723877, "step": 1420 }, { "epoch": 0.007102907721804437, "grad_norm": 1.7729791402816772, "learning_rate": 4.7317880794702e-06, "loss": 0.17387770414352416, "step": 1430 }, { "epoch": 0.007152578405173698, "grad_norm": 3.1970808506011963, "learning_rate": 4.764900662251656e-06, "loss": 0.17198877334594725, "step": 1440 }, { "epoch": 0.00720224908854296, "grad_norm": 2.301589012145996, "learning_rate": 4.798013245033113e-06, "loss": 0.19783320426940917, "step": 1450 }, { "epoch": 0.0072519197719122215, "grad_norm": 2.817408323287964, "learning_rate": 4.83112582781457e-06, "loss": 0.18564193248748778, "step": 1460 }, { "epoch": 0.007301590455281484, "grad_norm": 1.8399732112884521, "learning_rate": 4.864238410596027e-06, "loss": 0.16991117000579833, "step": 1470 }, { "epoch": 0.007351261138650746, "grad_norm": 3.026158094406128, "learning_rate": 4.897350993377484e-06, "loss": 0.1591586947441101, "step": 1480 }, { "epoch": 0.007400931822020007, "grad_norm": 2.1211276054382324, "learning_rate": 4.930463576158941e-06, "loss": 0.19459903240203857, "step": 1490 }, { "epoch": 0.007450602505389269, "grad_norm": 2.178318500518799, "learning_rate": 4.963576158940398e-06, "loss": 0.18723661899566652, "step": 1500 }, { "epoch": 0.0075002731887585305, "grad_norm": 1.8238352537155151, "learning_rate": 4.996688741721855e-06, "loss": 0.16536825895309448, "step": 1510 }, { "epoch": 0.007549943872127793, "grad_norm": 1.6106855869293213, "learning_rate": 5.029801324503312e-06, "loss": 0.18371682167053222, "step": 1520 }, { "epoch": 0.007599614555497055, "grad_norm": 2.2653934955596924, "learning_rate": 5.062913907284769e-06, "loss": 0.1908413887023926, "step": 1530 }, { "epoch": 0.007649285238866316, "grad_norm": 4.213798999786377, "learning_rate": 5.096026490066226e-06, "loss": 0.17016336917877198, "step": 1540 }, { "epoch": 0.007698955922235578, "grad_norm": 2.5297701358795166, "learning_rate": 5.1291390728476835e-06, "loss": 0.1851850152015686, "step": 1550 }, { "epoch": 0.0077486266056048395, "grad_norm": 2.5567612648010254, "learning_rate": 5.162251655629139e-06, "loss": 0.16841502189636232, "step": 1560 }, { "epoch": 0.007798297288974102, "grad_norm": 3.4279279708862305, "learning_rate": 5.195364238410596e-06, "loss": 0.1767507553100586, "step": 1570 }, { "epoch": 0.007847967972343363, "grad_norm": 3.709191083908081, "learning_rate": 5.228476821192053e-06, "loss": 0.1708056688308716, "step": 1580 }, { "epoch": 0.007897638655712625, "grad_norm": 2.7237648963928223, "learning_rate": 5.2615894039735106e-06, "loss": 0.18479905128479004, "step": 1590 }, { "epoch": 0.007947309339081887, "grad_norm": 1.848931908607483, "learning_rate": 5.294701986754968e-06, "loss": 0.18338754177093505, "step": 1600 }, { "epoch": 0.00799698002245115, "grad_norm": 2.1490585803985596, "learning_rate": 5.327814569536425e-06, "loss": 0.1670324206352234, "step": 1610 }, { "epoch": 0.008046650705820411, "grad_norm": 3.4938628673553467, "learning_rate": 5.3609271523178805e-06, "loss": 0.19210712909698485, "step": 1620 }, { "epoch": 0.008096321389189672, "grad_norm": 2.2048985958099365, "learning_rate": 5.394039735099338e-06, "loss": 0.18928823471069336, "step": 1630 }, { "epoch": 0.008145992072558934, "grad_norm": 2.6025733947753906, "learning_rate": 5.427152317880795e-06, "loss": 0.16721415519714355, "step": 1640 }, { "epoch": 0.008195662755928196, "grad_norm": 3.087392807006836, "learning_rate": 5.460264900662252e-06, "loss": 0.1940237283706665, "step": 1650 }, { "epoch": 0.008245333439297458, "grad_norm": 2.3141720294952393, "learning_rate": 5.493377483443709e-06, "loss": 0.17843434810638428, "step": 1660 }, { "epoch": 0.00829500412266672, "grad_norm": 3.11641526222229, "learning_rate": 5.5264900662251665e-06, "loss": 0.1763406753540039, "step": 1670 }, { "epoch": 0.00834467480603598, "grad_norm": 1.8845865726470947, "learning_rate": 5.559602649006624e-06, "loss": 0.18380764722824097, "step": 1680 }, { "epoch": 0.008394345489405243, "grad_norm": 3.968796730041504, "learning_rate": 5.592715231788079e-06, "loss": 0.17430617809295654, "step": 1690 }, { "epoch": 0.008444016172774505, "grad_norm": 1.735971450805664, "learning_rate": 5.625827814569536e-06, "loss": 0.1857773780822754, "step": 1700 }, { "epoch": 0.008493686856143767, "grad_norm": 3.357771635055542, "learning_rate": 5.6589403973509936e-06, "loss": 0.17025994062423705, "step": 1710 }, { "epoch": 0.00854335753951303, "grad_norm": 2.4030866622924805, "learning_rate": 5.692052980132451e-06, "loss": 0.19262006282806396, "step": 1720 }, { "epoch": 0.00859302822288229, "grad_norm": 2.9310948848724365, "learning_rate": 5.725165562913908e-06, "loss": 0.18524959087371826, "step": 1730 }, { "epoch": 0.008642698906251552, "grad_norm": 2.462550163269043, "learning_rate": 5.758278145695365e-06, "loss": 0.1758629322052002, "step": 1740 }, { "epoch": 0.008692369589620814, "grad_norm": 3.330195426940918, "learning_rate": 5.7913907284768215e-06, "loss": 0.1725791096687317, "step": 1750 }, { "epoch": 0.008742040272990076, "grad_norm": 3.0388636589050293, "learning_rate": 5.824503311258279e-06, "loss": 0.17465840578079223, "step": 1760 }, { "epoch": 0.008791710956359338, "grad_norm": 3.212313652038574, "learning_rate": 5.857615894039735e-06, "loss": 0.17417666912078858, "step": 1770 }, { "epoch": 0.008841381639728599, "grad_norm": 2.8090872764587402, "learning_rate": 5.890728476821192e-06, "loss": 0.21028003692626954, "step": 1780 }, { "epoch": 0.008891052323097861, "grad_norm": 1.3673946857452393, "learning_rate": 5.9238410596026494e-06, "loss": 0.17181005477905273, "step": 1790 }, { "epoch": 0.008940723006467123, "grad_norm": 2.0812013149261475, "learning_rate": 5.956953642384107e-06, "loss": 0.17320625782012938, "step": 1800 }, { "epoch": 0.008990393689836385, "grad_norm": 2.6714887619018555, "learning_rate": 5.990066225165564e-06, "loss": 0.18040544986724855, "step": 1810 }, { "epoch": 0.009040064373205647, "grad_norm": 1.975069284439087, "learning_rate": 6.02317880794702e-06, "loss": 0.17286829948425292, "step": 1820 }, { "epoch": 0.009089735056574908, "grad_norm": 4.027335166931152, "learning_rate": 6.056291390728477e-06, "loss": 0.1897881031036377, "step": 1830 }, { "epoch": 0.00913940573994417, "grad_norm": 2.075819730758667, "learning_rate": 6.089403973509935e-06, "loss": 0.1899643898010254, "step": 1840 }, { "epoch": 0.009189076423313432, "grad_norm": 2.9417898654937744, "learning_rate": 6.122516556291391e-06, "loss": 0.17459083795547486, "step": 1850 }, { "epoch": 0.009238747106682694, "grad_norm": 3.4720942974090576, "learning_rate": 6.155629139072848e-06, "loss": 0.18561891317367554, "step": 1860 }, { "epoch": 0.009288417790051956, "grad_norm": 1.9221569299697876, "learning_rate": 6.188741721854305e-06, "loss": 0.16674094200134276, "step": 1870 }, { "epoch": 0.009338088473421217, "grad_norm": 6.564807891845703, "learning_rate": 6.221854304635762e-06, "loss": 0.18576953411102295, "step": 1880 }, { "epoch": 0.009387759156790479, "grad_norm": 1.6799699068069458, "learning_rate": 6.254966887417219e-06, "loss": 0.16049432754516602, "step": 1890 }, { "epoch": 0.009437429840159741, "grad_norm": 1.4184718132019043, "learning_rate": 6.288079470198676e-06, "loss": 0.19208366870880128, "step": 1900 }, { "epoch": 0.009487100523529003, "grad_norm": 2.559979200363159, "learning_rate": 6.321192052980133e-06, "loss": 0.1843846321105957, "step": 1910 }, { "epoch": 0.009536771206898265, "grad_norm": 2.71077036857605, "learning_rate": 6.3543046357615905e-06, "loss": 0.21124677658081054, "step": 1920 }, { "epoch": 0.009586441890267526, "grad_norm": 2.360442876815796, "learning_rate": 6.387417218543047e-06, "loss": 0.18002129793167115, "step": 1930 }, { "epoch": 0.009636112573636788, "grad_norm": 1.9702231884002686, "learning_rate": 6.420529801324504e-06, "loss": 0.1686561346054077, "step": 1940 }, { "epoch": 0.00968578325700605, "grad_norm": 1.7616103887557983, "learning_rate": 6.45364238410596e-06, "loss": 0.1721489667892456, "step": 1950 }, { "epoch": 0.009735453940375312, "grad_norm": 3.43619966506958, "learning_rate": 6.4867549668874176e-06, "loss": 0.18258501291275026, "step": 1960 }, { "epoch": 0.009785124623744574, "grad_norm": 3.576558828353882, "learning_rate": 6.519867549668875e-06, "loss": 0.18773183822631836, "step": 1970 }, { "epoch": 0.009834795307113835, "grad_norm": 2.4696295261383057, "learning_rate": 6.552980132450332e-06, "loss": 0.18054547309875488, "step": 1980 }, { "epoch": 0.009884465990483097, "grad_norm": 3.8781003952026367, "learning_rate": 6.586092715231789e-06, "loss": 0.18373600244522095, "step": 1990 }, { "epoch": 0.009934136673852359, "grad_norm": 3.9460840225219727, "learning_rate": 6.619205298013246e-06, "loss": 0.16504924297332763, "step": 2000 }, { "epoch": 0.009983807357221621, "grad_norm": 2.025362253189087, "learning_rate": 6.652317880794702e-06, "loss": 0.17502574920654296, "step": 2010 }, { "epoch": 0.010033478040590883, "grad_norm": 1.6949776411056519, "learning_rate": 6.685430463576159e-06, "loss": 0.17076375484466552, "step": 2020 }, { "epoch": 0.010083148723960144, "grad_norm": 3.0462610721588135, "learning_rate": 6.718543046357616e-06, "loss": 0.17878905534744263, "step": 2030 }, { "epoch": 0.010132819407329406, "grad_norm": 1.4910180568695068, "learning_rate": 6.7516556291390735e-06, "loss": 0.17256510257720947, "step": 2040 }, { "epoch": 0.010182490090698668, "grad_norm": 2.795171022415161, "learning_rate": 6.784768211920531e-06, "loss": 0.1784762978553772, "step": 2050 }, { "epoch": 0.01023216077406793, "grad_norm": 3.0944666862487793, "learning_rate": 6.817880794701988e-06, "loss": 0.18711752891540528, "step": 2060 }, { "epoch": 0.010281831457437192, "grad_norm": 2.4873673915863037, "learning_rate": 6.850993377483445e-06, "loss": 0.195280921459198, "step": 2070 }, { "epoch": 0.010331502140806453, "grad_norm": 2.1496129035949707, "learning_rate": 6.8841059602649006e-06, "loss": 0.17016714811325073, "step": 2080 }, { "epoch": 0.010381172824175715, "grad_norm": 2.9776406288146973, "learning_rate": 6.917218543046358e-06, "loss": 0.1847905158996582, "step": 2090 }, { "epoch": 0.010430843507544977, "grad_norm": 4.823115348815918, "learning_rate": 6.950331125827815e-06, "loss": 0.19009037017822267, "step": 2100 }, { "epoch": 0.010480514190914239, "grad_norm": 1.9519308805465698, "learning_rate": 6.983443708609272e-06, "loss": 0.1778056502342224, "step": 2110 }, { "epoch": 0.010530184874283501, "grad_norm": 2.630695104598999, "learning_rate": 7.016556291390729e-06, "loss": 0.17983213663101197, "step": 2120 }, { "epoch": 0.010579855557652762, "grad_norm": 2.1870250701904297, "learning_rate": 7.0496688741721865e-06, "loss": 0.2071401596069336, "step": 2130 }, { "epoch": 0.010629526241022024, "grad_norm": 2.2603466510772705, "learning_rate": 7.082781456953644e-06, "loss": 0.17749457359313964, "step": 2140 }, { "epoch": 0.010679196924391286, "grad_norm": 1.8912417888641357, "learning_rate": 7.115894039735099e-06, "loss": 0.17831871509552003, "step": 2150 }, { "epoch": 0.010728867607760548, "grad_norm": 2.3413286209106445, "learning_rate": 7.1490066225165564e-06, "loss": 0.18799715042114257, "step": 2160 }, { "epoch": 0.01077853829112981, "grad_norm": 1.4445513486862183, "learning_rate": 7.182119205298014e-06, "loss": 0.17708225250244142, "step": 2170 }, { "epoch": 0.01082820897449907, "grad_norm": 2.9110445976257324, "learning_rate": 7.215231788079471e-06, "loss": 0.1809491753578186, "step": 2180 }, { "epoch": 0.010877879657868333, "grad_norm": 1.7447422742843628, "learning_rate": 7.248344370860928e-06, "loss": 0.18451323509216308, "step": 2190 }, { "epoch": 0.010927550341237595, "grad_norm": 2.7776272296905518, "learning_rate": 7.281456953642385e-06, "loss": 0.18350493907928467, "step": 2200 }, { "epoch": 0.010977221024606857, "grad_norm": 2.650736093521118, "learning_rate": 7.314569536423842e-06, "loss": 0.1838235855102539, "step": 2210 }, { "epoch": 0.01102689170797612, "grad_norm": 3.3094468116760254, "learning_rate": 7.347682119205298e-06, "loss": 0.17390596866607666, "step": 2220 }, { "epoch": 0.01107656239134538, "grad_norm": 1.4791520833969116, "learning_rate": 7.380794701986755e-06, "loss": 0.17292146682739257, "step": 2230 }, { "epoch": 0.011126233074714642, "grad_norm": 2.2489101886749268, "learning_rate": 7.413907284768212e-06, "loss": 0.16858043670654296, "step": 2240 }, { "epoch": 0.011175903758083904, "grad_norm": 1.479506492614746, "learning_rate": 7.4470198675496695e-06, "loss": 0.1839114546775818, "step": 2250 }, { "epoch": 0.011225574441453166, "grad_norm": 2.3024728298187256, "learning_rate": 7.480132450331127e-06, "loss": 0.1723141074180603, "step": 2260 }, { "epoch": 0.011275245124822428, "grad_norm": 2.308689594268799, "learning_rate": 7.513245033112584e-06, "loss": 0.17530457973480223, "step": 2270 }, { "epoch": 0.011324915808191689, "grad_norm": 1.8987630605697632, "learning_rate": 7.54635761589404e-06, "loss": 0.18625223636627197, "step": 2280 }, { "epoch": 0.01137458649156095, "grad_norm": 2.243648052215576, "learning_rate": 7.5794701986754975e-06, "loss": 0.16756192445755005, "step": 2290 }, { "epoch": 0.011424257174930213, "grad_norm": 2.62277889251709, "learning_rate": 7.612582781456954e-06, "loss": 0.17523298263549805, "step": 2300 }, { "epoch": 0.011473927858299475, "grad_norm": 2.6969313621520996, "learning_rate": 7.645695364238411e-06, "loss": 0.1843332529067993, "step": 2310 }, { "epoch": 0.011523598541668737, "grad_norm": 2.7692887783050537, "learning_rate": 7.678807947019868e-06, "loss": 0.19569106101989747, "step": 2320 }, { "epoch": 0.011573269225037998, "grad_norm": 1.7062021493911743, "learning_rate": 7.711920529801325e-06, "loss": 0.1815157413482666, "step": 2330 }, { "epoch": 0.01162293990840726, "grad_norm": 5.356148719787598, "learning_rate": 7.745033112582781e-06, "loss": 0.17086052894592285, "step": 2340 }, { "epoch": 0.011672610591776522, "grad_norm": 3.33719801902771, "learning_rate": 7.778145695364238e-06, "loss": 0.18090696334838868, "step": 2350 }, { "epoch": 0.011722281275145784, "grad_norm": 3.0900793075561523, "learning_rate": 7.811258278145695e-06, "loss": 0.18333261013031005, "step": 2360 }, { "epoch": 0.011771951958515046, "grad_norm": 2.135249614715576, "learning_rate": 7.844370860927153e-06, "loss": 0.17205777168273925, "step": 2370 }, { "epoch": 0.011821622641884307, "grad_norm": 2.3547425270080566, "learning_rate": 7.87748344370861e-06, "loss": 0.18577401638031005, "step": 2380 }, { "epoch": 0.011871293325253569, "grad_norm": 1.8727818727493286, "learning_rate": 7.910596026490067e-06, "loss": 0.17891082763671876, "step": 2390 }, { "epoch": 0.01192096400862283, "grad_norm": 2.3855531215667725, "learning_rate": 7.943708609271524e-06, "loss": 0.1763103723526001, "step": 2400 }, { "epoch": 0.011970634691992093, "grad_norm": 1.387784719467163, "learning_rate": 7.97682119205298e-06, "loss": 0.1762932538986206, "step": 2410 }, { "epoch": 0.012020305375361355, "grad_norm": 1.034947156906128, "learning_rate": 8.009933774834437e-06, "loss": 0.1866842269897461, "step": 2420 }, { "epoch": 0.012069976058730615, "grad_norm": 2.5323398113250732, "learning_rate": 8.043046357615894e-06, "loss": 0.16892561912536622, "step": 2430 }, { "epoch": 0.012119646742099878, "grad_norm": 2.989145278930664, "learning_rate": 8.076158940397351e-06, "loss": 0.1839211940765381, "step": 2440 }, { "epoch": 0.01216931742546914, "grad_norm": 1.756300926208496, "learning_rate": 8.109271523178808e-06, "loss": 0.1910154104232788, "step": 2450 }, { "epoch": 0.012218988108838402, "grad_norm": 2.668762683868408, "learning_rate": 8.142384105960266e-06, "loss": 0.18085395097732543, "step": 2460 }, { "epoch": 0.012268658792207664, "grad_norm": 3.873810052871704, "learning_rate": 8.175496688741723e-06, "loss": 0.18682827949523925, "step": 2470 }, { "epoch": 0.012318329475576924, "grad_norm": 2.312880277633667, "learning_rate": 8.208609271523178e-06, "loss": 0.19024269580841063, "step": 2480 }, { "epoch": 0.012368000158946187, "grad_norm": 1.3593415021896362, "learning_rate": 8.241721854304635e-06, "loss": 0.16449612379074097, "step": 2490 }, { "epoch": 0.012417670842315449, "grad_norm": 2.937798500061035, "learning_rate": 8.274834437086093e-06, "loss": 0.17878899574279786, "step": 2500 }, { "epoch": 0.012467341525684711, "grad_norm": 2.913886547088623, "learning_rate": 8.30794701986755e-06, "loss": 0.1829519271850586, "step": 2510 }, { "epoch": 0.012517012209053973, "grad_norm": 1.671430230140686, "learning_rate": 8.341059602649007e-06, "loss": 0.18153033256530762, "step": 2520 }, { "epoch": 0.012566682892423233, "grad_norm": 1.66941499710083, "learning_rate": 8.374172185430464e-06, "loss": 0.18168771266937256, "step": 2530 }, { "epoch": 0.012616353575792496, "grad_norm": 2.3386270999908447, "learning_rate": 8.407284768211921e-06, "loss": 0.18447788953781127, "step": 2540 }, { "epoch": 0.012666024259161758, "grad_norm": 2.0375492572784424, "learning_rate": 8.440397350993379e-06, "loss": 0.19668658971786498, "step": 2550 }, { "epoch": 0.01271569494253102, "grad_norm": 2.4055545330047607, "learning_rate": 8.473509933774834e-06, "loss": 0.17970207929611207, "step": 2560 }, { "epoch": 0.012765365625900282, "grad_norm": 2.064282178878784, "learning_rate": 8.506622516556291e-06, "loss": 0.16650962829589844, "step": 2570 }, { "epoch": 0.012815036309269542, "grad_norm": 2.3335351943969727, "learning_rate": 8.539735099337749e-06, "loss": 0.17502758502960206, "step": 2580 }, { "epoch": 0.012864706992638805, "grad_norm": 2.0668752193450928, "learning_rate": 8.572847682119206e-06, "loss": 0.18390549421310426, "step": 2590 }, { "epoch": 0.012914377676008067, "grad_norm": 2.047877550125122, "learning_rate": 8.605960264900663e-06, "loss": 0.2016974925994873, "step": 2600 }, { "epoch": 0.012964048359377329, "grad_norm": 2.7247772216796875, "learning_rate": 8.63907284768212e-06, "loss": 0.18642885684967042, "step": 2610 }, { "epoch": 0.013013719042746591, "grad_norm": 6.118054389953613, "learning_rate": 8.672185430463577e-06, "loss": 0.18229581117630006, "step": 2620 }, { "epoch": 0.013063389726115851, "grad_norm": 1.9164704084396362, "learning_rate": 8.705298013245035e-06, "loss": 0.18297843933105468, "step": 2630 }, { "epoch": 0.013113060409485114, "grad_norm": 1.9237078428268433, "learning_rate": 8.73841059602649e-06, "loss": 0.18488857746124268, "step": 2640 }, { "epoch": 0.013162731092854376, "grad_norm": 1.656651258468628, "learning_rate": 8.771523178807947e-06, "loss": 0.17938733100891113, "step": 2650 }, { "epoch": 0.013212401776223638, "grad_norm": 2.0605554580688477, "learning_rate": 8.804635761589404e-06, "loss": 0.1645503282546997, "step": 2660 }, { "epoch": 0.0132620724595929, "grad_norm": 1.8292603492736816, "learning_rate": 8.837748344370862e-06, "loss": 0.19101364612579347, "step": 2670 }, { "epoch": 0.01331174314296216, "grad_norm": 2.741668701171875, "learning_rate": 8.870860927152319e-06, "loss": 0.1789675235748291, "step": 2680 }, { "epoch": 0.013361413826331422, "grad_norm": 1.1822489500045776, "learning_rate": 8.903973509933776e-06, "loss": 0.19115128517150878, "step": 2690 }, { "epoch": 0.013411084509700685, "grad_norm": 1.3517659902572632, "learning_rate": 8.937086092715233e-06, "loss": 0.18000019788742067, "step": 2700 }, { "epoch": 0.013460755193069947, "grad_norm": 2.8536813259124756, "learning_rate": 8.97019867549669e-06, "loss": 0.1957855463027954, "step": 2710 }, { "epoch": 0.013510425876439207, "grad_norm": 1.4317512512207031, "learning_rate": 9.003311258278146e-06, "loss": 0.185702908039093, "step": 2720 }, { "epoch": 0.01356009655980847, "grad_norm": 3.7853920459747314, "learning_rate": 9.036423841059603e-06, "loss": 0.1870926022529602, "step": 2730 }, { "epoch": 0.013609767243177731, "grad_norm": 4.703758239746094, "learning_rate": 9.06953642384106e-06, "loss": 0.18276008367538452, "step": 2740 }, { "epoch": 0.013659437926546994, "grad_norm": 1.6708961725234985, "learning_rate": 9.102649006622518e-06, "loss": 0.1756451368331909, "step": 2750 }, { "epoch": 0.013709108609916256, "grad_norm": 3.231356382369995, "learning_rate": 9.135761589403975e-06, "loss": 0.1866975784301758, "step": 2760 }, { "epoch": 0.013758779293285516, "grad_norm": 1.9459664821624756, "learning_rate": 9.168874172185432e-06, "loss": 0.1776055097579956, "step": 2770 }, { "epoch": 0.013808449976654778, "grad_norm": 1.9123119115829468, "learning_rate": 9.201986754966889e-06, "loss": 0.18591654300689697, "step": 2780 }, { "epoch": 0.01385812066002404, "grad_norm": 1.29712975025177, "learning_rate": 9.235099337748346e-06, "loss": 0.16000210046768187, "step": 2790 }, { "epoch": 0.013907791343393303, "grad_norm": 4.1224822998046875, "learning_rate": 9.268211920529802e-06, "loss": 0.19867550134658812, "step": 2800 }, { "epoch": 0.013957462026762565, "grad_norm": 0.9826055765151978, "learning_rate": 9.301324503311259e-06, "loss": 0.18331130743026733, "step": 2810 }, { "epoch": 0.014007132710131825, "grad_norm": 6.333941459655762, "learning_rate": 9.334437086092716e-06, "loss": 0.20352435111999512, "step": 2820 }, { "epoch": 0.014056803393501087, "grad_norm": 3.268662691116333, "learning_rate": 9.367549668874173e-06, "loss": 0.1821104884147644, "step": 2830 }, { "epoch": 0.01410647407687035, "grad_norm": 5.303905963897705, "learning_rate": 9.40066225165563e-06, "loss": 0.1844209313392639, "step": 2840 }, { "epoch": 0.014156144760239612, "grad_norm": 4.225063323974609, "learning_rate": 9.433774834437088e-06, "loss": 0.19134098291397095, "step": 2850 }, { "epoch": 0.014205815443608874, "grad_norm": 2.2520272731781006, "learning_rate": 9.466887417218545e-06, "loss": 0.1822751522064209, "step": 2860 }, { "epoch": 0.014255486126978134, "grad_norm": 2.0338294506073, "learning_rate": 9.5e-06, "loss": 0.19041919708251953, "step": 2870 }, { "epoch": 0.014305156810347396, "grad_norm": 2.232076406478882, "learning_rate": 9.533112582781458e-06, "loss": 0.17710930109024048, "step": 2880 }, { "epoch": 0.014354827493716658, "grad_norm": 1.5226856470108032, "learning_rate": 9.566225165562915e-06, "loss": 0.17154742479324342, "step": 2890 }, { "epoch": 0.01440449817708592, "grad_norm": 1.588347315788269, "learning_rate": 9.599337748344372e-06, "loss": 0.1862330436706543, "step": 2900 }, { "epoch": 0.014454168860455183, "grad_norm": 2.339529275894165, "learning_rate": 9.63245033112583e-06, "loss": 0.18834686279296875, "step": 2910 }, { "epoch": 0.014503839543824443, "grad_norm": 2.3268871307373047, "learning_rate": 9.665562913907287e-06, "loss": 0.18093793392181395, "step": 2920 }, { "epoch": 0.014553510227193705, "grad_norm": 2.2513318061828613, "learning_rate": 9.698675496688742e-06, "loss": 0.18054587841033937, "step": 2930 }, { "epoch": 0.014603180910562967, "grad_norm": 2.4553403854370117, "learning_rate": 9.7317880794702e-06, "loss": 0.174951171875, "step": 2940 }, { "epoch": 0.01465285159393223, "grad_norm": 3.5175304412841797, "learning_rate": 9.764900662251656e-06, "loss": 0.20368194580078125, "step": 2950 }, { "epoch": 0.014702522277301492, "grad_norm": 1.5899333953857422, "learning_rate": 9.798013245033114e-06, "loss": 0.18064996004104614, "step": 2960 }, { "epoch": 0.014752192960670752, "grad_norm": 2.110138177871704, "learning_rate": 9.83112582781457e-06, "loss": 0.1805877685546875, "step": 2970 }, { "epoch": 0.014801863644040014, "grad_norm": 1.4651800394058228, "learning_rate": 9.864238410596028e-06, "loss": 0.18488636016845703, "step": 2980 }, { "epoch": 0.014851534327409276, "grad_norm": 1.733068585395813, "learning_rate": 9.897350993377485e-06, "loss": 0.18526778221130372, "step": 2990 }, { "epoch": 0.014901205010778538, "grad_norm": 2.9246044158935547, "learning_rate": 9.93046357615894e-06, "loss": 0.19750728607177734, "step": 3000 }, { "epoch": 0.0149508756941478, "grad_norm": 3.156139373779297, "learning_rate": 9.963576158940398e-06, "loss": 0.19816428422927856, "step": 3010 }, { "epoch": 0.015000546377517061, "grad_norm": 1.8836030960083008, "learning_rate": 9.996688741721855e-06, "loss": 0.18277287483215332, "step": 3020 }, { "epoch": 0.015050217060886323, "grad_norm": 1.8550848960876465, "learning_rate": 1.0029801324503312e-05, "loss": 0.18352826833724975, "step": 3030 }, { "epoch": 0.015099887744255585, "grad_norm": 2.934332847595215, "learning_rate": 1.0062913907284768e-05, "loss": 0.18847784996032715, "step": 3040 }, { "epoch": 0.015149558427624847, "grad_norm": 1.8866370916366577, "learning_rate": 1.0096026490066227e-05, "loss": 0.17600018978118898, "step": 3050 }, { "epoch": 0.01519922911099411, "grad_norm": 1.5904643535614014, "learning_rate": 1.0129139072847682e-05, "loss": 0.17463159561157227, "step": 3060 }, { "epoch": 0.01524889979436337, "grad_norm": 3.3409969806671143, "learning_rate": 1.0162251655629141e-05, "loss": 0.18307161331176758, "step": 3070 }, { "epoch": 0.015298570477732632, "grad_norm": 2.627317428588867, "learning_rate": 1.0195364238410597e-05, "loss": 0.18472886085510254, "step": 3080 }, { "epoch": 0.015348241161101894, "grad_norm": 1.7248982191085815, "learning_rate": 1.0228476821192055e-05, "loss": 0.1849926471710205, "step": 3090 }, { "epoch": 0.015397911844471156, "grad_norm": 2.2031733989715576, "learning_rate": 1.0261589403973511e-05, "loss": 0.18167083263397216, "step": 3100 }, { "epoch": 0.015447582527840419, "grad_norm": 1.8046129941940308, "learning_rate": 1.0294701986754966e-05, "loss": 0.18709464073181153, "step": 3110 }, { "epoch": 0.015497253211209679, "grad_norm": 1.8180136680603027, "learning_rate": 1.0327814569536425e-05, "loss": 0.19029703140258789, "step": 3120 }, { "epoch": 0.015546923894578941, "grad_norm": 1.8701063394546509, "learning_rate": 1.0360927152317881e-05, "loss": 0.18794244527816772, "step": 3130 }, { "epoch": 0.015596594577948203, "grad_norm": 2.552666187286377, "learning_rate": 1.039403973509934e-05, "loss": 0.184382426738739, "step": 3140 }, { "epoch": 0.015646265261317464, "grad_norm": 3.249128580093384, "learning_rate": 1.0427152317880795e-05, "loss": 0.18306152820587157, "step": 3150 }, { "epoch": 0.015695935944686726, "grad_norm": 2.020355701446533, "learning_rate": 1.046026490066225e-05, "loss": 0.18529982566833497, "step": 3160 }, { "epoch": 0.015745606628055988, "grad_norm": 2.1964938640594482, "learning_rate": 1.049337748344371e-05, "loss": 0.16631340980529785, "step": 3170 }, { "epoch": 0.01579527731142525, "grad_norm": 3.1487207412719727, "learning_rate": 1.0526490066225165e-05, "loss": 0.17396063804626466, "step": 3180 }, { "epoch": 0.015844947994794512, "grad_norm": 2.1085164546966553, "learning_rate": 1.0559602649006624e-05, "loss": 0.17292239665985107, "step": 3190 }, { "epoch": 0.015894618678163774, "grad_norm": 1.870263934135437, "learning_rate": 1.059271523178808e-05, "loss": 0.18283443450927733, "step": 3200 }, { "epoch": 0.015944289361533037, "grad_norm": 3.3897039890289307, "learning_rate": 1.0625827814569538e-05, "loss": 0.18037118911743164, "step": 3210 }, { "epoch": 0.0159939600449023, "grad_norm": 1.9457917213439941, "learning_rate": 1.0658940397350994e-05, "loss": 0.17297500371932983, "step": 3220 }, { "epoch": 0.01604363072827156, "grad_norm": 1.4785962104797363, "learning_rate": 1.069205298013245e-05, "loss": 0.19152796268463135, "step": 3230 }, { "epoch": 0.016093301411640823, "grad_norm": 4.244647979736328, "learning_rate": 1.0725165562913908e-05, "loss": 0.1917176604270935, "step": 3240 }, { "epoch": 0.01614297209501008, "grad_norm": 1.3154908418655396, "learning_rate": 1.0758278145695364e-05, "loss": 0.1923983573913574, "step": 3250 }, { "epoch": 0.016192642778379344, "grad_norm": 2.002685546875, "learning_rate": 1.0791390728476823e-05, "loss": 0.18203485012054443, "step": 3260 }, { "epoch": 0.016242313461748606, "grad_norm": 1.4298230409622192, "learning_rate": 1.0824503311258278e-05, "loss": 0.18492965698242186, "step": 3270 }, { "epoch": 0.016291984145117868, "grad_norm": 1.589779019355774, "learning_rate": 1.0857615894039737e-05, "loss": 0.1913064956665039, "step": 3280 }, { "epoch": 0.01634165482848713, "grad_norm": 1.1774101257324219, "learning_rate": 1.0890728476821193e-05, "loss": 0.1926274061203003, "step": 3290 }, { "epoch": 0.016391325511856392, "grad_norm": 1.907808542251587, "learning_rate": 1.0923841059602648e-05, "loss": 0.16846878528594972, "step": 3300 }, { "epoch": 0.016440996195225654, "grad_norm": 3.1571271419525146, "learning_rate": 1.0956953642384107e-05, "loss": 0.1865251898765564, "step": 3310 }, { "epoch": 0.016490666878594917, "grad_norm": 1.1461679935455322, "learning_rate": 1.0990066225165563e-05, "loss": 0.1811055064201355, "step": 3320 }, { "epoch": 0.01654033756196418, "grad_norm": 4.195809841156006, "learning_rate": 1.1023178807947021e-05, "loss": 0.19288461208343505, "step": 3330 }, { "epoch": 0.01659000824533344, "grad_norm": 2.624876022338867, "learning_rate": 1.1056291390728477e-05, "loss": 0.1845552682876587, "step": 3340 }, { "epoch": 0.0166396789287027, "grad_norm": 2.936798095703125, "learning_rate": 1.1089403973509936e-05, "loss": 0.5221993446350097, "step": 3350 }, { "epoch": 0.01668934961207196, "grad_norm": 2.2101831436157227, "learning_rate": 1.1122516556291391e-05, "loss": 0.20033981800079345, "step": 3360 }, { "epoch": 0.016739020295441224, "grad_norm": 1.6262238025665283, "learning_rate": 1.1155629139072849e-05, "loss": 0.21598553657531738, "step": 3370 }, { "epoch": 0.016788690978810486, "grad_norm": 3.315415382385254, "learning_rate": 1.1188741721854306e-05, "loss": 0.19819328784942628, "step": 3380 }, { "epoch": 0.016838361662179748, "grad_norm": 2.0930416584014893, "learning_rate": 1.1221854304635761e-05, "loss": 0.1824117660522461, "step": 3390 }, { "epoch": 0.01688803234554901, "grad_norm": 2.3413522243499756, "learning_rate": 1.125496688741722e-05, "loss": 0.20024561882019043, "step": 3400 }, { "epoch": 0.016937703028918272, "grad_norm": 1.1119698286056519, "learning_rate": 1.1288079470198676e-05, "loss": 0.1878279209136963, "step": 3410 }, { "epoch": 0.016987373712287535, "grad_norm": 1.506826639175415, "learning_rate": 1.1321192052980135e-05, "loss": 0.17250553369522095, "step": 3420 }, { "epoch": 0.017037044395656797, "grad_norm": 2.0375940799713135, "learning_rate": 1.135430463576159e-05, "loss": 0.1902737259864807, "step": 3430 }, { "epoch": 0.01708671507902606, "grad_norm": 1.257516860961914, "learning_rate": 1.1387417218543047e-05, "loss": 0.1775859236717224, "step": 3440 }, { "epoch": 0.017136385762395318, "grad_norm": 1.026654839515686, "learning_rate": 1.1420529801324504e-05, "loss": 0.18622901439666747, "step": 3450 }, { "epoch": 0.01718605644576458, "grad_norm": 2.50437593460083, "learning_rate": 1.145364238410596e-05, "loss": 0.18855082988739014, "step": 3460 }, { "epoch": 0.017235727129133842, "grad_norm": 2.2275784015655518, "learning_rate": 1.1486754966887419e-05, "loss": 0.18335413932800293, "step": 3470 }, { "epoch": 0.017285397812503104, "grad_norm": 1.4574965238571167, "learning_rate": 1.1519867549668874e-05, "loss": 0.19104597568511963, "step": 3480 }, { "epoch": 0.017335068495872366, "grad_norm": 2.309854745864868, "learning_rate": 1.1552980132450332e-05, "loss": 0.20363006591796876, "step": 3490 }, { "epoch": 0.017384739179241628, "grad_norm": 4.3640851974487305, "learning_rate": 1.1586092715231789e-05, "loss": 0.19278345108032227, "step": 3500 }, { "epoch": 0.01743440986261089, "grad_norm": 1.7568119764328003, "learning_rate": 1.1619205298013246e-05, "loss": 0.19056177139282227, "step": 3510 }, { "epoch": 0.017484080545980153, "grad_norm": 1.558046817779541, "learning_rate": 1.1652317880794703e-05, "loss": 0.2008971691131592, "step": 3520 }, { "epoch": 0.017533751229349415, "grad_norm": 1.7508635520935059, "learning_rate": 1.168543046357616e-05, "loss": 0.18216084241867064, "step": 3530 }, { "epoch": 0.017583421912718677, "grad_norm": 1.7093795537948608, "learning_rate": 1.1718543046357618e-05, "loss": 0.17815396785736085, "step": 3540 }, { "epoch": 0.017633092596087935, "grad_norm": 1.0686417818069458, "learning_rate": 1.1751655629139073e-05, "loss": 0.17890329360961915, "step": 3550 }, { "epoch": 0.017682763279457198, "grad_norm": 3.1284048557281494, "learning_rate": 1.178476821192053e-05, "loss": 0.18507553339004518, "step": 3560 }, { "epoch": 0.01773243396282646, "grad_norm": 3.162768840789795, "learning_rate": 1.1817880794701987e-05, "loss": 0.16619131565093995, "step": 3570 }, { "epoch": 0.017782104646195722, "grad_norm": 1.9076405763626099, "learning_rate": 1.1850993377483445e-05, "loss": 0.17408595085144044, "step": 3580 }, { "epoch": 0.017831775329564984, "grad_norm": 1.202549934387207, "learning_rate": 1.1884105960264902e-05, "loss": 0.1833592653274536, "step": 3590 }, { "epoch": 0.017881446012934246, "grad_norm": 3.560203790664673, "learning_rate": 1.1917218543046359e-05, "loss": 0.18326954841613768, "step": 3600 }, { "epoch": 0.01793111669630351, "grad_norm": 2.0534822940826416, "learning_rate": 1.1950331125827816e-05, "loss": 0.1972375512123108, "step": 3610 }, { "epoch": 0.01798078737967277, "grad_norm": 1.6020872592926025, "learning_rate": 1.1983443708609272e-05, "loss": 0.17817931175231932, "step": 3620 }, { "epoch": 0.018030458063042033, "grad_norm": 2.0773584842681885, "learning_rate": 1.2016556291390729e-05, "loss": 0.19503461122512816, "step": 3630 }, { "epoch": 0.018080128746411295, "grad_norm": 3.766469717025757, "learning_rate": 1.2049668874172186e-05, "loss": 0.1799681544303894, "step": 3640 }, { "epoch": 0.018129799429780553, "grad_norm": 1.3046866655349731, "learning_rate": 1.2082781456953643e-05, "loss": 0.17981308698654175, "step": 3650 }, { "epoch": 0.018179470113149816, "grad_norm": 2.4708404541015625, "learning_rate": 1.21158940397351e-05, "loss": 0.1770862817764282, "step": 3660 }, { "epoch": 0.018229140796519078, "grad_norm": 1.2854769229888916, "learning_rate": 1.2149006622516558e-05, "loss": 0.19599409103393556, "step": 3670 }, { "epoch": 0.01827881147988834, "grad_norm": 4.593092918395996, "learning_rate": 1.2182119205298015e-05, "loss": 0.18509241342544555, "step": 3680 }, { "epoch": 0.018328482163257602, "grad_norm": 1.926474690437317, "learning_rate": 1.2215231788079472e-05, "loss": 0.17143242359161376, "step": 3690 }, { "epoch": 0.018378152846626864, "grad_norm": 1.0870530605316162, "learning_rate": 1.2248344370860928e-05, "loss": 0.1823495388031006, "step": 3700 }, { "epoch": 0.018427823529996126, "grad_norm": 3.989715099334717, "learning_rate": 1.2281456953642385e-05, "loss": 0.1855185866355896, "step": 3710 }, { "epoch": 0.01847749421336539, "grad_norm": 1.276634693145752, "learning_rate": 1.2314569536423842e-05, "loss": 0.1973806381225586, "step": 3720 }, { "epoch": 0.01852716489673465, "grad_norm": 2.3657877445220947, "learning_rate": 1.23476821192053e-05, "loss": 0.172166109085083, "step": 3730 }, { "epoch": 0.018576835580103913, "grad_norm": 3.665963649749756, "learning_rate": 1.2380794701986756e-05, "loss": 0.20155985355377198, "step": 3740 }, { "epoch": 0.01862650626347317, "grad_norm": 2.2763442993164062, "learning_rate": 1.2413907284768212e-05, "loss": 0.18565049171447753, "step": 3750 }, { "epoch": 0.018676176946842434, "grad_norm": 3.1490914821624756, "learning_rate": 1.244701986754967e-05, "loss": 0.19103680849075316, "step": 3760 }, { "epoch": 0.018725847630211696, "grad_norm": 1.623007893562317, "learning_rate": 1.2480132450331126e-05, "loss": 0.18467895984649657, "step": 3770 }, { "epoch": 0.018775518313580958, "grad_norm": 3.9740612506866455, "learning_rate": 1.2513245033112583e-05, "loss": 0.18200567960739136, "step": 3780 }, { "epoch": 0.01882518899695022, "grad_norm": 1.4589102268218994, "learning_rate": 1.254635761589404e-05, "loss": 0.18682761192321778, "step": 3790 }, { "epoch": 0.018874859680319482, "grad_norm": 2.0185797214508057, "learning_rate": 1.2579470198675498e-05, "loss": 0.18410794734954833, "step": 3800 }, { "epoch": 0.018924530363688744, "grad_norm": 3.118875741958618, "learning_rate": 1.2612582781456955e-05, "loss": 0.19398970603942872, "step": 3810 }, { "epoch": 0.018974201047058006, "grad_norm": 1.5565133094787598, "learning_rate": 1.264569536423841e-05, "loss": 0.1756715178489685, "step": 3820 }, { "epoch": 0.01902387173042727, "grad_norm": 3.9159209728240967, "learning_rate": 1.267880794701987e-05, "loss": 0.20220003128051758, "step": 3830 }, { "epoch": 0.01907354241379653, "grad_norm": 2.339905261993408, "learning_rate": 1.2711920529801325e-05, "loss": 0.18066842555999757, "step": 3840 }, { "epoch": 0.01912321309716579, "grad_norm": 1.4503462314605713, "learning_rate": 1.2745033112582784e-05, "loss": 0.18167765140533448, "step": 3850 }, { "epoch": 0.01917288378053505, "grad_norm": 1.7537838220596313, "learning_rate": 1.277814569536424e-05, "loss": 0.18141448497772217, "step": 3860 }, { "epoch": 0.019222554463904314, "grad_norm": 2.0678136348724365, "learning_rate": 1.2811258278145697e-05, "loss": 0.18028171062469484, "step": 3870 }, { "epoch": 0.019272225147273576, "grad_norm": 1.5099399089813232, "learning_rate": 1.2844370860927154e-05, "loss": 0.17978010177612305, "step": 3880 }, { "epoch": 0.019321895830642838, "grad_norm": 2.392169237136841, "learning_rate": 1.287748344370861e-05, "loss": 0.1776036500930786, "step": 3890 }, { "epoch": 0.0193715665140121, "grad_norm": 1.2238954305648804, "learning_rate": 1.2910596026490068e-05, "loss": 0.1893895149230957, "step": 3900 }, { "epoch": 0.019421237197381362, "grad_norm": 1.9165122509002686, "learning_rate": 1.2943708609271524e-05, "loss": 0.18912961483001708, "step": 3910 }, { "epoch": 0.019470907880750624, "grad_norm": 1.9419690370559692, "learning_rate": 1.2976821192052983e-05, "loss": 0.18101351261138915, "step": 3920 }, { "epoch": 0.019520578564119886, "grad_norm": 1.5333714485168457, "learning_rate": 1.3009933774834438e-05, "loss": 0.17350373268127442, "step": 3930 }, { "epoch": 0.01957024924748915, "grad_norm": 3.992816925048828, "learning_rate": 1.3043046357615897e-05, "loss": 0.18603839874267578, "step": 3940 }, { "epoch": 0.019619919930858407, "grad_norm": 1.593489408493042, "learning_rate": 1.3076158940397352e-05, "loss": 0.16893066167831422, "step": 3950 }, { "epoch": 0.01966959061422767, "grad_norm": 2.9280264377593994, "learning_rate": 1.3109271523178808e-05, "loss": 0.1838160514831543, "step": 3960 }, { "epoch": 0.01971926129759693, "grad_norm": 1.0762102603912354, "learning_rate": 1.3142384105960267e-05, "loss": 0.1836058974266052, "step": 3970 }, { "epoch": 0.019768931980966194, "grad_norm": 2.8995108604431152, "learning_rate": 1.3175496688741722e-05, "loss": 0.18762834072113038, "step": 3980 }, { "epoch": 0.019818602664335456, "grad_norm": 4.401794910430908, "learning_rate": 1.3208609271523181e-05, "loss": 0.18991607427597046, "step": 3990 }, { "epoch": 0.019868273347704718, "grad_norm": 3.8974339962005615, "learning_rate": 1.3241721854304637e-05, "loss": 0.19488385915756226, "step": 4000 }, { "epoch": 0.01991794403107398, "grad_norm": 1.5920697450637817, "learning_rate": 1.3274834437086096e-05, "loss": 0.1631417989730835, "step": 4010 }, { "epoch": 0.019967614714443242, "grad_norm": 2.052913188934326, "learning_rate": 1.3307947019867551e-05, "loss": 0.18227431774139405, "step": 4020 }, { "epoch": 0.020017285397812504, "grad_norm": 1.3711425065994263, "learning_rate": 1.3341059602649007e-05, "loss": 0.18092360496520996, "step": 4030 }, { "epoch": 0.020066956081181767, "grad_norm": 1.6977415084838867, "learning_rate": 1.3374172185430466e-05, "loss": 0.17300416231155397, "step": 4040 }, { "epoch": 0.020116626764551025, "grad_norm": 3.1753745079040527, "learning_rate": 1.3407284768211921e-05, "loss": 0.19415032863616943, "step": 4050 }, { "epoch": 0.020166297447920287, "grad_norm": 2.880204439163208, "learning_rate": 1.344039735099338e-05, "loss": 0.18640899658203125, "step": 4060 }, { "epoch": 0.02021596813128955, "grad_norm": 4.243243217468262, "learning_rate": 1.3473509933774835e-05, "loss": 0.18785024881362916, "step": 4070 }, { "epoch": 0.02026563881465881, "grad_norm": 4.697863578796387, "learning_rate": 1.3506622516556291e-05, "loss": 0.17452749013900756, "step": 4080 }, { "epoch": 0.020315309498028074, "grad_norm": 1.80364990234375, "learning_rate": 1.353973509933775e-05, "loss": 0.17987821102142335, "step": 4090 }, { "epoch": 0.020364980181397336, "grad_norm": 1.3665533065795898, "learning_rate": 1.3572847682119205e-05, "loss": 0.18346667289733887, "step": 4100 }, { "epoch": 0.020414650864766598, "grad_norm": 1.4037001132965088, "learning_rate": 1.3605960264900664e-05, "loss": 0.18313882350921631, "step": 4110 }, { "epoch": 0.02046432154813586, "grad_norm": 2.612327814102173, "learning_rate": 1.363907284768212e-05, "loss": 0.1762984037399292, "step": 4120 }, { "epoch": 0.020513992231505122, "grad_norm": 2.199845790863037, "learning_rate": 1.3672185430463579e-05, "loss": 0.17278604507446288, "step": 4130 }, { "epoch": 0.020563662914874384, "grad_norm": 1.5565985441207886, "learning_rate": 1.3705298013245034e-05, "loss": 0.17784904241561889, "step": 4140 }, { "epoch": 0.020613333598243643, "grad_norm": 2.160515069961548, "learning_rate": 1.373841059602649e-05, "loss": 0.18784053325653077, "step": 4150 }, { "epoch": 0.020663004281612905, "grad_norm": 2.341576337814331, "learning_rate": 1.3771523178807949e-05, "loss": 0.18644143342971803, "step": 4160 }, { "epoch": 0.020712674964982167, "grad_norm": 2.0756490230560303, "learning_rate": 1.3804635761589404e-05, "loss": 0.19552795886993407, "step": 4170 }, { "epoch": 0.02076234564835143, "grad_norm": 2.330216884613037, "learning_rate": 1.3837748344370863e-05, "loss": 0.17452402114868165, "step": 4180 }, { "epoch": 0.020812016331720692, "grad_norm": 2.0819590091705322, "learning_rate": 1.3870860927152318e-05, "loss": 0.185872220993042, "step": 4190 }, { "epoch": 0.020861687015089954, "grad_norm": 1.2811914682388306, "learning_rate": 1.3903973509933777e-05, "loss": 0.17623027563095092, "step": 4200 }, { "epoch": 0.020911357698459216, "grad_norm": 1.9838154315948486, "learning_rate": 1.3937086092715233e-05, "loss": 0.18179011344909668, "step": 4210 }, { "epoch": 0.020961028381828478, "grad_norm": 1.1842217445373535, "learning_rate": 1.3970198675496688e-05, "loss": 0.19668999910354615, "step": 4220 }, { "epoch": 0.02101069906519774, "grad_norm": 1.477606177330017, "learning_rate": 1.4003311258278147e-05, "loss": 0.18854626417160034, "step": 4230 }, { "epoch": 0.021060369748567002, "grad_norm": 1.310619592666626, "learning_rate": 1.4036423841059603e-05, "loss": 0.19349594116210939, "step": 4240 }, { "epoch": 0.02111004043193626, "grad_norm": 4.131065845489502, "learning_rate": 1.4069536423841062e-05, "loss": 0.18577840328216552, "step": 4250 }, { "epoch": 0.021159711115305523, "grad_norm": 2.2833690643310547, "learning_rate": 1.4102649006622517e-05, "loss": 0.17440168857574462, "step": 4260 }, { "epoch": 0.021209381798674785, "grad_norm": 2.530653715133667, "learning_rate": 1.4135761589403976e-05, "loss": 0.18711636066436768, "step": 4270 }, { "epoch": 0.021259052482044048, "grad_norm": 1.4933055639266968, "learning_rate": 1.4168874172185431e-05, "loss": 0.17468180656433105, "step": 4280 }, { "epoch": 0.02130872316541331, "grad_norm": 4.097238540649414, "learning_rate": 1.4201986754966887e-05, "loss": 0.17553427219390869, "step": 4290 }, { "epoch": 0.021358393848782572, "grad_norm": 3.42136287689209, "learning_rate": 1.4235099337748346e-05, "loss": 0.17184096574783325, "step": 4300 }, { "epoch": 0.021408064532151834, "grad_norm": 1.7677042484283447, "learning_rate": 1.4268211920529801e-05, "loss": 0.1688260555267334, "step": 4310 }, { "epoch": 0.021457735215521096, "grad_norm": 1.3072869777679443, "learning_rate": 1.430132450331126e-05, "loss": 0.18450546264648438, "step": 4320 }, { "epoch": 0.021507405898890358, "grad_norm": 2.6121156215667725, "learning_rate": 1.4334437086092716e-05, "loss": 0.1744043231010437, "step": 4330 }, { "epoch": 0.02155707658225962, "grad_norm": 1.4036321640014648, "learning_rate": 1.4367549668874175e-05, "loss": 0.1666831374168396, "step": 4340 }, { "epoch": 0.02160674726562888, "grad_norm": 1.8476784229278564, "learning_rate": 1.440066225165563e-05, "loss": 0.1944596290588379, "step": 4350 }, { "epoch": 0.02165641794899814, "grad_norm": 1.831693172454834, "learning_rate": 1.4433774834437086e-05, "loss": 0.1786064624786377, "step": 4360 }, { "epoch": 0.021706088632367403, "grad_norm": 1.9149588346481323, "learning_rate": 1.4466887417218545e-05, "loss": 0.1935401439666748, "step": 4370 }, { "epoch": 0.021755759315736665, "grad_norm": 2.446338415145874, "learning_rate": 1.45e-05, "loss": 0.19188815355300903, "step": 4380 }, { "epoch": 0.021805429999105928, "grad_norm": 2.763719320297241, "learning_rate": 1.4533112582781459e-05, "loss": 0.18605217933654786, "step": 4390 }, { "epoch": 0.02185510068247519, "grad_norm": 1.8102272748947144, "learning_rate": 1.4566225165562914e-05, "loss": 0.20171017646789552, "step": 4400 }, { "epoch": 0.021904771365844452, "grad_norm": 1.545214295387268, "learning_rate": 1.4599337748344372e-05, "loss": 0.18247838020324708, "step": 4410 }, { "epoch": 0.021954442049213714, "grad_norm": 2.4472062587738037, "learning_rate": 1.4632450331125829e-05, "loss": 0.17189887762069703, "step": 4420 }, { "epoch": 0.022004112732582976, "grad_norm": 5.533267974853516, "learning_rate": 1.4665562913907286e-05, "loss": 0.18658413887023925, "step": 4430 }, { "epoch": 0.02205378341595224, "grad_norm": 2.4620585441589355, "learning_rate": 1.4698675496688743e-05, "loss": 0.17720479965209962, "step": 4440 }, { "epoch": 0.022103454099321497, "grad_norm": 1.3521543741226196, "learning_rate": 1.4731788079470199e-05, "loss": 0.18161754608154296, "step": 4450 }, { "epoch": 0.02215312478269076, "grad_norm": 3.0568931102752686, "learning_rate": 1.4764900662251658e-05, "loss": 0.18362152576446533, "step": 4460 }, { "epoch": 0.02220279546606002, "grad_norm": 3.39493989944458, "learning_rate": 1.4798013245033113e-05, "loss": 0.17443255186080933, "step": 4470 }, { "epoch": 0.022252466149429283, "grad_norm": 2.2855563163757324, "learning_rate": 1.483112582781457e-05, "loss": 0.18411847352981567, "step": 4480 }, { "epoch": 0.022302136832798546, "grad_norm": 2.5773496627807617, "learning_rate": 1.4864238410596028e-05, "loss": 0.1733729362487793, "step": 4490 }, { "epoch": 0.022351807516167808, "grad_norm": 3.864358425140381, "learning_rate": 1.4897350993377485e-05, "loss": 0.1782907724380493, "step": 4500 }, { "epoch": 0.02240147819953707, "grad_norm": 1.375651240348816, "learning_rate": 1.4930463576158942e-05, "loss": 0.1713019609451294, "step": 4510 }, { "epoch": 0.022451148882906332, "grad_norm": 1.3490885496139526, "learning_rate": 1.4963576158940397e-05, "loss": 0.17239396572113036, "step": 4520 }, { "epoch": 0.022500819566275594, "grad_norm": 1.723838210105896, "learning_rate": 1.4996688741721856e-05, "loss": 0.1922934651374817, "step": 4530 }, { "epoch": 0.022550490249644856, "grad_norm": 1.4078688621520996, "learning_rate": 1.5029801324503312e-05, "loss": 0.19912900924682617, "step": 4540 }, { "epoch": 0.022600160933014115, "grad_norm": 1.3611547946929932, "learning_rate": 1.5062913907284769e-05, "loss": 0.18313374519348144, "step": 4550 }, { "epoch": 0.022649831616383377, "grad_norm": 2.3439412117004395, "learning_rate": 1.5096026490066226e-05, "loss": 0.18990919589996338, "step": 4560 }, { "epoch": 0.02269950229975264, "grad_norm": 3.6604363918304443, "learning_rate": 1.5129139072847683e-05, "loss": 0.16953084468841553, "step": 4570 }, { "epoch": 0.0227491729831219, "grad_norm": 2.3739607334136963, "learning_rate": 1.516225165562914e-05, "loss": 0.18331968784332275, "step": 4580 }, { "epoch": 0.022798843666491164, "grad_norm": 1.2837337255477905, "learning_rate": 1.5195364238410598e-05, "loss": 0.18896515369415284, "step": 4590 }, { "epoch": 0.022848514349860426, "grad_norm": 2.4605743885040283, "learning_rate": 1.5228476821192055e-05, "loss": 0.18917088508605956, "step": 4600 }, { "epoch": 0.022898185033229688, "grad_norm": 3.2989394664764404, "learning_rate": 1.5261589403973512e-05, "loss": 0.1984776735305786, "step": 4610 }, { "epoch": 0.02294785571659895, "grad_norm": 2.3021836280822754, "learning_rate": 1.5294701986754966e-05, "loss": 0.17536404132843017, "step": 4620 }, { "epoch": 0.022997526399968212, "grad_norm": 2.805008888244629, "learning_rate": 1.5327814569536427e-05, "loss": 0.17060699462890624, "step": 4630 }, { "epoch": 0.023047197083337474, "grad_norm": 1.398053526878357, "learning_rate": 1.536092715231788e-05, "loss": 0.17602884769439697, "step": 4640 }, { "epoch": 0.023096867766706733, "grad_norm": 1.8528401851654053, "learning_rate": 1.539403973509934e-05, "loss": 0.18861098289489747, "step": 4650 }, { "epoch": 0.023146538450075995, "grad_norm": 2.6060314178466797, "learning_rate": 1.5427152317880795e-05, "loss": 0.1811823606491089, "step": 4660 }, { "epoch": 0.023196209133445257, "grad_norm": 1.3528269529342651, "learning_rate": 1.5460264900662252e-05, "loss": 0.16673653125762938, "step": 4670 }, { "epoch": 0.02324587981681452, "grad_norm": 1.129175066947937, "learning_rate": 1.549337748344371e-05, "loss": 0.15974375009536743, "step": 4680 }, { "epoch": 0.02329555050018378, "grad_norm": 3.320842742919922, "learning_rate": 1.5526490066225166e-05, "loss": 0.18307993412017823, "step": 4690 }, { "epoch": 0.023345221183553044, "grad_norm": 1.8923377990722656, "learning_rate": 1.5559602649006624e-05, "loss": 0.1712352752685547, "step": 4700 }, { "epoch": 0.023394891866922306, "grad_norm": 4.719736099243164, "learning_rate": 1.559271523178808e-05, "loss": 0.19337048530578613, "step": 4710 }, { "epoch": 0.023444562550291568, "grad_norm": 1.7653902769088745, "learning_rate": 1.5625827814569538e-05, "loss": 0.18556073904037476, "step": 4720 }, { "epoch": 0.02349423323366083, "grad_norm": 1.4848212003707886, "learning_rate": 1.5658940397350995e-05, "loss": 0.18999094963073732, "step": 4730 }, { "epoch": 0.023543903917030092, "grad_norm": 2.7478420734405518, "learning_rate": 1.569205298013245e-05, "loss": 0.18315911293029785, "step": 4740 }, { "epoch": 0.02359357460039935, "grad_norm": 1.8283861875534058, "learning_rate": 1.572516556291391e-05, "loss": 0.17825062274932862, "step": 4750 }, { "epoch": 0.023643245283768613, "grad_norm": 1.04271399974823, "learning_rate": 1.5758278145695363e-05, "loss": 0.18558740615844727, "step": 4760 }, { "epoch": 0.023692915967137875, "grad_norm": 1.1154428720474243, "learning_rate": 1.5791390728476824e-05, "loss": 0.18958668708801268, "step": 4770 }, { "epoch": 0.023742586650507137, "grad_norm": 2.2958524227142334, "learning_rate": 1.5824503311258278e-05, "loss": 0.18551766872406006, "step": 4780 }, { "epoch": 0.0237922573338764, "grad_norm": 2.542163133621216, "learning_rate": 1.585761589403974e-05, "loss": 0.187781023979187, "step": 4790 }, { "epoch": 0.02384192801724566, "grad_norm": 1.8248882293701172, "learning_rate": 1.5890728476821192e-05, "loss": 0.18838313817977906, "step": 4800 }, { "epoch": 0.023891598700614924, "grad_norm": 3.8700053691864014, "learning_rate": 1.592384105960265e-05, "loss": 0.1855283260345459, "step": 4810 }, { "epoch": 0.023941269383984186, "grad_norm": 1.7982794046401978, "learning_rate": 1.5956953642384107e-05, "loss": 0.16432201862335205, "step": 4820 }, { "epoch": 0.023990940067353448, "grad_norm": 4.214598655700684, "learning_rate": 1.5990066225165564e-05, "loss": 0.1803242564201355, "step": 4830 }, { "epoch": 0.02404061075072271, "grad_norm": 2.4653894901275635, "learning_rate": 1.602317880794702e-05, "loss": 0.17772159576416016, "step": 4840 }, { "epoch": 0.02409028143409197, "grad_norm": 2.261542797088623, "learning_rate": 1.6056291390728478e-05, "loss": 0.18457852602005004, "step": 4850 }, { "epoch": 0.02413995211746123, "grad_norm": 4.4585466384887695, "learning_rate": 1.6089403973509935e-05, "loss": 0.1882317066192627, "step": 4860 }, { "epoch": 0.024189622800830493, "grad_norm": 2.931868314743042, "learning_rate": 1.6122516556291393e-05, "loss": 0.18489526510238646, "step": 4870 }, { "epoch": 0.024239293484199755, "grad_norm": 1.425565481185913, "learning_rate": 1.615562913907285e-05, "loss": 0.17813045978546144, "step": 4880 }, { "epoch": 0.024288964167569017, "grad_norm": 1.843508005142212, "learning_rate": 1.6188741721854307e-05, "loss": 0.1960810661315918, "step": 4890 }, { "epoch": 0.02433863485093828, "grad_norm": 5.8394341468811035, "learning_rate": 1.622185430463576e-05, "loss": 0.1925906777381897, "step": 4900 }, { "epoch": 0.02438830553430754, "grad_norm": 2.6312947273254395, "learning_rate": 1.625496688741722e-05, "loss": 0.18620507717132567, "step": 4910 }, { "epoch": 0.024437976217676804, "grad_norm": 2.8575057983398438, "learning_rate": 1.6288079470198675e-05, "loss": 0.1771397829055786, "step": 4920 }, { "epoch": 0.024487646901046066, "grad_norm": 1.5042877197265625, "learning_rate": 1.6321192052980136e-05, "loss": 0.2014847755432129, "step": 4930 }, { "epoch": 0.024537317584415328, "grad_norm": 1.1081721782684326, "learning_rate": 1.635430463576159e-05, "loss": 0.1708313226699829, "step": 4940 }, { "epoch": 0.024586988267784587, "grad_norm": 1.6378979682922363, "learning_rate": 1.6387417218543047e-05, "loss": 0.18033827543258668, "step": 4950 }, { "epoch": 0.02463665895115385, "grad_norm": 1.8712351322174072, "learning_rate": 1.6420529801324504e-05, "loss": 0.1932552933692932, "step": 4960 }, { "epoch": 0.02468632963452311, "grad_norm": 1.660624623298645, "learning_rate": 1.645364238410596e-05, "loss": 0.16470872163772582, "step": 4970 }, { "epoch": 0.024736000317892373, "grad_norm": 5.020325660705566, "learning_rate": 1.648675496688742e-05, "loss": 0.19426305294036866, "step": 4980 }, { "epoch": 0.024785671001261635, "grad_norm": 2.30584716796875, "learning_rate": 1.6519867549668876e-05, "loss": 0.18807424306869508, "step": 4990 }, { "epoch": 0.024835341684630897, "grad_norm": 2.546746015548706, "learning_rate": 1.6552980132450333e-05, "loss": 0.17840919494628907, "step": 5000 }, { "epoch": 0.02488501236800016, "grad_norm": 3.004467248916626, "learning_rate": 1.658609271523179e-05, "loss": 0.18264486789703369, "step": 5010 }, { "epoch": 0.024934683051369422, "grad_norm": 1.5215213298797607, "learning_rate": 1.6619205298013247e-05, "loss": 0.18726651668548583, "step": 5020 }, { "epoch": 0.024984353734738684, "grad_norm": 1.0139532089233398, "learning_rate": 1.6652317880794704e-05, "loss": 0.17367167472839357, "step": 5030 }, { "epoch": 0.025034024418107946, "grad_norm": 1.5561110973358154, "learning_rate": 1.668543046357616e-05, "loss": 0.1881392240524292, "step": 5040 }, { "epoch": 0.025083695101477205, "grad_norm": 1.3861910104751587, "learning_rate": 1.671854304635762e-05, "loss": 0.18514833450317383, "step": 5050 }, { "epoch": 0.025133365784846467, "grad_norm": 1.1477420330047607, "learning_rate": 1.6751655629139073e-05, "loss": 0.18297908306121827, "step": 5060 }, { "epoch": 0.02518303646821573, "grad_norm": 2.2814126014709473, "learning_rate": 1.678476821192053e-05, "loss": 0.17513140439987182, "step": 5070 }, { "epoch": 0.02523270715158499, "grad_norm": 2.790719509124756, "learning_rate": 1.6817880794701987e-05, "loss": 0.17375082969665528, "step": 5080 }, { "epoch": 0.025282377834954253, "grad_norm": 3.2675998210906982, "learning_rate": 1.6850993377483444e-05, "loss": 0.19240210056304932, "step": 5090 }, { "epoch": 0.025332048518323515, "grad_norm": 1.439772605895996, "learning_rate": 1.68841059602649e-05, "loss": 0.18473100662231445, "step": 5100 }, { "epoch": 0.025381719201692778, "grad_norm": 2.1480071544647217, "learning_rate": 1.691721854304636e-05, "loss": 0.19117534160614014, "step": 5110 }, { "epoch": 0.02543138988506204, "grad_norm": 2.2216718196868896, "learning_rate": 1.6950331125827816e-05, "loss": 0.19107242822647094, "step": 5120 }, { "epoch": 0.025481060568431302, "grad_norm": 4.087093830108643, "learning_rate": 1.6983443708609273e-05, "loss": 0.18036739826202391, "step": 5130 }, { "epoch": 0.025530731251800564, "grad_norm": 3.4149210453033447, "learning_rate": 1.701655629139073e-05, "loss": 0.19189417362213135, "step": 5140 }, { "epoch": 0.025580401935169823, "grad_norm": 2.0039639472961426, "learning_rate": 1.7049668874172187e-05, "loss": 0.17406562566757203, "step": 5150 }, { "epoch": 0.025630072618539085, "grad_norm": 1.947262167930603, "learning_rate": 1.7082781456953645e-05, "loss": 0.1775000810623169, "step": 5160 }, { "epoch": 0.025679743301908347, "grad_norm": 3.2433552742004395, "learning_rate": 1.7115894039735102e-05, "loss": 0.1813387989997864, "step": 5170 }, { "epoch": 0.02572941398527761, "grad_norm": 2.579951286315918, "learning_rate": 1.714900662251656e-05, "loss": 0.16986905336380004, "step": 5180 }, { "epoch": 0.02577908466864687, "grad_norm": 1.5216988325119019, "learning_rate": 1.7182119205298016e-05, "loss": 0.18459784984588623, "step": 5190 }, { "epoch": 0.025828755352016133, "grad_norm": 1.8185160160064697, "learning_rate": 1.7215231788079473e-05, "loss": 0.1966113805770874, "step": 5200 }, { "epoch": 0.025878426035385396, "grad_norm": 1.3568289279937744, "learning_rate": 1.7248344370860927e-05, "loss": 0.1883610725402832, "step": 5210 }, { "epoch": 0.025928096718754658, "grad_norm": 1.5655676126480103, "learning_rate": 1.7281456953642384e-05, "loss": 0.19285691976547242, "step": 5220 }, { "epoch": 0.02597776740212392, "grad_norm": 2.1123387813568115, "learning_rate": 1.731456953642384e-05, "loss": 0.17895874977111817, "step": 5230 }, { "epoch": 0.026027438085493182, "grad_norm": 1.651623010635376, "learning_rate": 1.73476821192053e-05, "loss": 0.19343414306640624, "step": 5240 }, { "epoch": 0.02607710876886244, "grad_norm": 1.3095844984054565, "learning_rate": 1.7380794701986756e-05, "loss": 0.17728674411773682, "step": 5250 }, { "epoch": 0.026126779452231703, "grad_norm": 1.781893253326416, "learning_rate": 1.7413907284768213e-05, "loss": 0.1858449935913086, "step": 5260 }, { "epoch": 0.026176450135600965, "grad_norm": 1.0437121391296387, "learning_rate": 1.744701986754967e-05, "loss": 0.17732579708099366, "step": 5270 }, { "epoch": 0.026226120818970227, "grad_norm": 3.822406768798828, "learning_rate": 1.7480132450331128e-05, "loss": 0.19679372310638427, "step": 5280 }, { "epoch": 0.02627579150233949, "grad_norm": 3.3338968753814697, "learning_rate": 1.7513245033112585e-05, "loss": 0.18438608646392823, "step": 5290 }, { "epoch": 0.02632546218570875, "grad_norm": 1.619061827659607, "learning_rate": 1.7546357615894042e-05, "loss": 0.1915043354034424, "step": 5300 }, { "epoch": 0.026375132869078013, "grad_norm": 1.890722632408142, "learning_rate": 1.75794701986755e-05, "loss": 0.1878333568572998, "step": 5310 }, { "epoch": 0.026424803552447276, "grad_norm": 1.8500640392303467, "learning_rate": 1.7612582781456956e-05, "loss": 0.19285329580307006, "step": 5320 }, { "epoch": 0.026474474235816538, "grad_norm": 2.2553329467773438, "learning_rate": 1.764569536423841e-05, "loss": 0.1820528268814087, "step": 5330 }, { "epoch": 0.0265241449191858, "grad_norm": 2.2740800380706787, "learning_rate": 1.767880794701987e-05, "loss": 0.1888440728187561, "step": 5340 }, { "epoch": 0.02657381560255506, "grad_norm": 3.4372267723083496, "learning_rate": 1.7711920529801325e-05, "loss": 0.19743905067443848, "step": 5350 }, { "epoch": 0.02662348628592432, "grad_norm": 1.3318332433700562, "learning_rate": 1.7745033112582785e-05, "loss": 0.17245767116546631, "step": 5360 }, { "epoch": 0.026673156969293583, "grad_norm": 2.2477972507476807, "learning_rate": 1.777814569536424e-05, "loss": 0.17963836193084717, "step": 5370 }, { "epoch": 0.026722827652662845, "grad_norm": 2.318594217300415, "learning_rate": 1.7811258278145696e-05, "loss": 0.19476101398468018, "step": 5380 }, { "epoch": 0.026772498336032107, "grad_norm": 1.486621379852295, "learning_rate": 1.7844370860927153e-05, "loss": 0.18895423412322998, "step": 5390 }, { "epoch": 0.02682216901940137, "grad_norm": 1.5213308334350586, "learning_rate": 1.787748344370861e-05, "loss": 0.1948537826538086, "step": 5400 }, { "epoch": 0.02687183970277063, "grad_norm": 1.4316673278808594, "learning_rate": 1.7910596026490068e-05, "loss": 0.17958195209503175, "step": 5410 }, { "epoch": 0.026921510386139894, "grad_norm": 2.2641541957855225, "learning_rate": 1.7943708609271525e-05, "loss": 0.17565008401870727, "step": 5420 }, { "epoch": 0.026971181069509156, "grad_norm": 0.8466399908065796, "learning_rate": 1.7976821192052982e-05, "loss": 0.18056776523590087, "step": 5430 }, { "epoch": 0.027020851752878414, "grad_norm": 1.5260937213897705, "learning_rate": 1.800993377483444e-05, "loss": 0.17928318977355956, "step": 5440 }, { "epoch": 0.027070522436247677, "grad_norm": 1.2432365417480469, "learning_rate": 1.8043046357615897e-05, "loss": 0.1869424104690552, "step": 5450 }, { "epoch": 0.02712019311961694, "grad_norm": 3.2125978469848633, "learning_rate": 1.8076158940397354e-05, "loss": 0.19767470359802247, "step": 5460 }, { "epoch": 0.0271698638029862, "grad_norm": 1.1317777633666992, "learning_rate": 1.8109271523178808e-05, "loss": 0.18766345977783203, "step": 5470 }, { "epoch": 0.027219534486355463, "grad_norm": 2.9523227214813232, "learning_rate": 1.8142384105960268e-05, "loss": 0.19121065139770507, "step": 5480 }, { "epoch": 0.027269205169724725, "grad_norm": 3.769292116165161, "learning_rate": 1.8175496688741722e-05, "loss": 0.18120768070220947, "step": 5490 }, { "epoch": 0.027318875853093987, "grad_norm": 2.3400533199310303, "learning_rate": 1.8208609271523182e-05, "loss": 0.1910192847251892, "step": 5500 }, { "epoch": 0.02736854653646325, "grad_norm": 1.361035943031311, "learning_rate": 1.8241721854304636e-05, "loss": 0.18747018575668334, "step": 5510 }, { "epoch": 0.02741821721983251, "grad_norm": 3.0645618438720703, "learning_rate": 1.8274834437086097e-05, "loss": 0.191718590259552, "step": 5520 }, { "epoch": 0.027467887903201774, "grad_norm": 3.1527726650238037, "learning_rate": 1.830794701986755e-05, "loss": 0.1704928159713745, "step": 5530 }, { "epoch": 0.027517558586571032, "grad_norm": 1.3480923175811768, "learning_rate": 1.8341059602649008e-05, "loss": 0.1920831561088562, "step": 5540 }, { "epoch": 0.027567229269940294, "grad_norm": 3.072476863861084, "learning_rate": 1.8374172185430465e-05, "loss": 0.18282728195190429, "step": 5550 }, { "epoch": 0.027616899953309557, "grad_norm": 3.5245187282562256, "learning_rate": 1.8407284768211922e-05, "loss": 0.19609930515289306, "step": 5560 }, { "epoch": 0.02766657063667882, "grad_norm": 3.018420457839966, "learning_rate": 1.844039735099338e-05, "loss": 0.17348480224609375, "step": 5570 }, { "epoch": 0.02771624132004808, "grad_norm": 4.268692970275879, "learning_rate": 1.8473509933774837e-05, "loss": 0.19556944370269774, "step": 5580 }, { "epoch": 0.027765912003417343, "grad_norm": 1.7281346321105957, "learning_rate": 1.850662251655629e-05, "loss": 0.20262715816497803, "step": 5590 }, { "epoch": 0.027815582686786605, "grad_norm": 1.2462327480316162, "learning_rate": 1.853973509933775e-05, "loss": 0.1939528226852417, "step": 5600 }, { "epoch": 0.027865253370155867, "grad_norm": 1.1245447397232056, "learning_rate": 1.8572847682119205e-05, "loss": 0.20099425315856934, "step": 5610 }, { "epoch": 0.02791492405352513, "grad_norm": 2.090320348739624, "learning_rate": 1.8605960264900665e-05, "loss": 0.19059369564056397, "step": 5620 }, { "epoch": 0.02796459473689439, "grad_norm": 3.7829039096832275, "learning_rate": 1.863907284768212e-05, "loss": 0.188185715675354, "step": 5630 }, { "epoch": 0.02801426542026365, "grad_norm": 1.1379507780075073, "learning_rate": 1.867218543046358e-05, "loss": 0.18149299621582032, "step": 5640 }, { "epoch": 0.028063936103632912, "grad_norm": 3.188553810119629, "learning_rate": 1.8705298013245034e-05, "loss": 0.17856278419494628, "step": 5650 }, { "epoch": 0.028113606787002175, "grad_norm": 2.3142080307006836, "learning_rate": 1.873841059602649e-05, "loss": 0.17563855648040771, "step": 5660 }, { "epoch": 0.028163277470371437, "grad_norm": 2.1785945892333984, "learning_rate": 1.8771523178807948e-05, "loss": 0.1837291717529297, "step": 5670 }, { "epoch": 0.0282129481537407, "grad_norm": 1.2971378564834595, "learning_rate": 1.8804635761589405e-05, "loss": 0.17460645437240602, "step": 5680 }, { "epoch": 0.02826261883710996, "grad_norm": 2.887206792831421, "learning_rate": 1.8837748344370862e-05, "loss": 0.1831148862838745, "step": 5690 }, { "epoch": 0.028312289520479223, "grad_norm": 1.0440592765808105, "learning_rate": 1.887086092715232e-05, "loss": 0.18533517122268678, "step": 5700 }, { "epoch": 0.028361960203848485, "grad_norm": 1.546286940574646, "learning_rate": 1.8903973509933777e-05, "loss": 0.18208394050598145, "step": 5710 }, { "epoch": 0.028411630887217747, "grad_norm": 2.3942229747772217, "learning_rate": 1.8937086092715234e-05, "loss": 0.19123371839523315, "step": 5720 }, { "epoch": 0.02846130157058701, "grad_norm": 5.706197261810303, "learning_rate": 1.8970198675496688e-05, "loss": 0.1767780065536499, "step": 5730 }, { "epoch": 0.028510972253956268, "grad_norm": 1.1395833492279053, "learning_rate": 1.900331125827815e-05, "loss": 0.18258469104766845, "step": 5740 }, { "epoch": 0.02856064293732553, "grad_norm": 3.1035654544830322, "learning_rate": 1.9036423841059602e-05, "loss": 0.20923476219177245, "step": 5750 }, { "epoch": 0.028610313620694793, "grad_norm": 3.100940704345703, "learning_rate": 1.9069536423841063e-05, "loss": 0.18142032623291016, "step": 5760 }, { "epoch": 0.028659984304064055, "grad_norm": 2.4104886054992676, "learning_rate": 1.9102649006622517e-05, "loss": 0.1897432327270508, "step": 5770 }, { "epoch": 0.028709654987433317, "grad_norm": 3.6743531227111816, "learning_rate": 1.9135761589403977e-05, "loss": 0.18745782375335693, "step": 5780 }, { "epoch": 0.02875932567080258, "grad_norm": 1.0521204471588135, "learning_rate": 1.916887417218543e-05, "loss": 0.1888510465621948, "step": 5790 }, { "epoch": 0.02880899635417184, "grad_norm": 1.83640456199646, "learning_rate": 1.9201986754966888e-05, "loss": 0.1850576639175415, "step": 5800 }, { "epoch": 0.028858667037541103, "grad_norm": 3.602574586868286, "learning_rate": 1.9235099337748345e-05, "loss": 0.18199291229248046, "step": 5810 }, { "epoch": 0.028908337720910365, "grad_norm": 1.4718270301818848, "learning_rate": 1.9268211920529803e-05, "loss": 0.17659734487533568, "step": 5820 }, { "epoch": 0.028958008404279627, "grad_norm": 1.4230375289916992, "learning_rate": 1.930132450331126e-05, "loss": 0.18600201606750488, "step": 5830 }, { "epoch": 0.029007679087648886, "grad_norm": 2.987978219985962, "learning_rate": 1.9334437086092717e-05, "loss": 0.17540154457092286, "step": 5840 }, { "epoch": 0.02905734977101815, "grad_norm": 3.466312885284424, "learning_rate": 1.9367549668874174e-05, "loss": 0.19361597299575806, "step": 5850 }, { "epoch": 0.02910702045438741, "grad_norm": 2.289588212966919, "learning_rate": 1.940066225165563e-05, "loss": 0.1790560245513916, "step": 5860 }, { "epoch": 0.029156691137756673, "grad_norm": 2.5651659965515137, "learning_rate": 1.9433774834437085e-05, "loss": 0.18151025772094725, "step": 5870 }, { "epoch": 0.029206361821125935, "grad_norm": 3.6089413166046143, "learning_rate": 1.9466887417218546e-05, "loss": 0.2127222537994385, "step": 5880 }, { "epoch": 0.029256032504495197, "grad_norm": 2.278566598892212, "learning_rate": 1.95e-05, "loss": 0.18963286876678467, "step": 5890 }, { "epoch": 0.02930570318786446, "grad_norm": 1.9733635187149048, "learning_rate": 1.953311258278146e-05, "loss": 0.18508970737457275, "step": 5900 }, { "epoch": 0.02935537387123372, "grad_norm": 1.6679989099502563, "learning_rate": 1.9566225165562914e-05, "loss": 0.16768676042556763, "step": 5910 }, { "epoch": 0.029405044554602983, "grad_norm": 2.4258005619049072, "learning_rate": 1.959933774834437e-05, "loss": 0.196936297416687, "step": 5920 }, { "epoch": 0.029454715237972245, "grad_norm": 2.221240282058716, "learning_rate": 1.963245033112583e-05, "loss": 0.18119330406188966, "step": 5930 }, { "epoch": 0.029504385921341504, "grad_norm": 3.9504804611206055, "learning_rate": 1.9665562913907286e-05, "loss": 0.17794029712677, "step": 5940 }, { "epoch": 0.029554056604710766, "grad_norm": 1.507676124572754, "learning_rate": 1.9698675496688743e-05, "loss": 0.17473435401916504, "step": 5950 }, { "epoch": 0.02960372728808003, "grad_norm": 3.0456509590148926, "learning_rate": 1.97317880794702e-05, "loss": 0.1880318284034729, "step": 5960 }, { "epoch": 0.02965339797144929, "grad_norm": 1.5272202491760254, "learning_rate": 1.9764900662251657e-05, "loss": 0.17652642726898193, "step": 5970 }, { "epoch": 0.029703068654818553, "grad_norm": 1.9416542053222656, "learning_rate": 1.9798013245033114e-05, "loss": 0.19829347133636474, "step": 5980 }, { "epoch": 0.029752739338187815, "grad_norm": 1.1851716041564941, "learning_rate": 1.983112582781457e-05, "loss": 0.1890525460243225, "step": 5990 }, { "epoch": 0.029802410021557077, "grad_norm": 1.4757286310195923, "learning_rate": 1.986423841059603e-05, "loss": 0.17229740619659423, "step": 6000 }, { "epoch": 0.02985208070492634, "grad_norm": 2.468862295150757, "learning_rate": 1.9897350993377486e-05, "loss": 0.18764499425888062, "step": 6010 }, { "epoch": 0.0299017513882956, "grad_norm": 2.0162980556488037, "learning_rate": 1.9930463576158943e-05, "loss": 0.18922858238220214, "step": 6020 }, { "epoch": 0.029951422071664863, "grad_norm": 3.268242835998535, "learning_rate": 1.9963576158940397e-05, "loss": 0.19193854331970214, "step": 6030 }, { "epoch": 0.030001092755034122, "grad_norm": 3.460871696472168, "learning_rate": 1.9996688741721858e-05, "loss": 0.20183877944946288, "step": 6040 }, { "epoch": 0.030050763438403384, "grad_norm": 1.5077201128005981, "learning_rate": 1.999999989518763e-05, "loss": 0.1694555401802063, "step": 6050 }, { "epoch": 0.030100434121772646, "grad_norm": 1.4503179788589478, "learning_rate": 1.999999953287326e-05, "loss": 0.1884917736053467, "step": 6060 }, { "epoch": 0.03015010480514191, "grad_norm": 1.2710117101669312, "learning_rate": 1.9999998911762927e-05, "loss": 0.177032732963562, "step": 6070 }, { "epoch": 0.03019977548851117, "grad_norm": 3.2853927612304688, "learning_rate": 1.999999803185664e-05, "loss": 0.19335122108459474, "step": 6080 }, { "epoch": 0.030249446171880433, "grad_norm": 2.6263139247894287, "learning_rate": 1.9999996893154424e-05, "loss": 0.18506855964660646, "step": 6090 }, { "epoch": 0.030299116855249695, "grad_norm": 3.352389335632324, "learning_rate": 1.9999995495656313e-05, "loss": 0.18883568048477173, "step": 6100 }, { "epoch": 0.030348787538618957, "grad_norm": 1.8068441152572632, "learning_rate": 1.9999993839362338e-05, "loss": 0.1873695135116577, "step": 6110 }, { "epoch": 0.03039845822198822, "grad_norm": 1.8545129299163818, "learning_rate": 1.999999192427254e-05, "loss": 0.17891402244567872, "step": 6120 }, { "epoch": 0.03044812890535748, "grad_norm": 2.9026780128479004, "learning_rate": 1.9999989750386977e-05, "loss": 0.18817405700683593, "step": 6130 }, { "epoch": 0.03049779958872674, "grad_norm": 1.6350668668746948, "learning_rate": 1.9999987317705697e-05, "loss": 0.19859167337417602, "step": 6140 }, { "epoch": 0.030547470272096002, "grad_norm": 2.9034335613250732, "learning_rate": 1.9999984626228766e-05, "loss": 0.19867751598358155, "step": 6150 }, { "epoch": 0.030597140955465264, "grad_norm": 2.5729033946990967, "learning_rate": 1.9999981675956252e-05, "loss": 0.2093266010284424, "step": 6160 }, { "epoch": 0.030646811638834526, "grad_norm": 3.260625123977661, "learning_rate": 1.9999978466888238e-05, "loss": 0.19229673147201537, "step": 6170 }, { "epoch": 0.03069648232220379, "grad_norm": 1.8501306772232056, "learning_rate": 1.9999974999024797e-05, "loss": 0.1863693118095398, "step": 6180 }, { "epoch": 0.03074615300557305, "grad_norm": 1.2402381896972656, "learning_rate": 1.9999971272366028e-05, "loss": 0.1977843999862671, "step": 6190 }, { "epoch": 0.030795823688942313, "grad_norm": 2.158047914505005, "learning_rate": 1.9999967286912022e-05, "loss": 0.1903085231781006, "step": 6200 }, { "epoch": 0.030845494372311575, "grad_norm": 2.744081974029541, "learning_rate": 1.999996304266288e-05, "loss": 0.18399020433425903, "step": 6210 }, { "epoch": 0.030895165055680837, "grad_norm": 1.1725471019744873, "learning_rate": 1.9999958539618717e-05, "loss": 0.19473793506622314, "step": 6220 }, { "epoch": 0.0309448357390501, "grad_norm": 1.128821611404419, "learning_rate": 1.9999953777779652e-05, "loss": 0.17267992496490478, "step": 6230 }, { "epoch": 0.030994506422419358, "grad_norm": 1.7352782487869263, "learning_rate": 1.99999487571458e-05, "loss": 0.1739220142364502, "step": 6240 }, { "epoch": 0.03104417710578862, "grad_norm": 1.0200051069259644, "learning_rate": 1.9999943477717298e-05, "loss": 0.17993378639221191, "step": 6250 }, { "epoch": 0.031093847789157882, "grad_norm": 3.8990819454193115, "learning_rate": 1.9999937939494277e-05, "loss": 0.19100871086120605, "step": 6260 }, { "epoch": 0.031143518472527144, "grad_norm": 1.3653457164764404, "learning_rate": 1.999993214247688e-05, "loss": 0.17852108478546141, "step": 6270 }, { "epoch": 0.031193189155896407, "grad_norm": 2.874274730682373, "learning_rate": 1.9999926086665266e-05, "loss": 0.20142054557800293, "step": 6280 }, { "epoch": 0.03124285983926567, "grad_norm": 1.4667178392410278, "learning_rate": 1.999991977205958e-05, "loss": 0.18970229625701904, "step": 6290 }, { "epoch": 0.03129253052263493, "grad_norm": 1.4533590078353882, "learning_rate": 1.9999913198659997e-05, "loss": 0.19565414190292357, "step": 6300 }, { "epoch": 0.03134220120600419, "grad_norm": 1.4822006225585938, "learning_rate": 1.9999906366466678e-05, "loss": 0.18434765338897705, "step": 6310 }, { "epoch": 0.03139187188937345, "grad_norm": 3.417120933532715, "learning_rate": 1.9999899275479802e-05, "loss": 0.19339420795440673, "step": 6320 }, { "epoch": 0.03144154257274272, "grad_norm": 2.1752419471740723, "learning_rate": 1.9999891925699557e-05, "loss": 0.19237428903579712, "step": 6330 }, { "epoch": 0.031491213256111976, "grad_norm": 2.9983530044555664, "learning_rate": 1.9999884317126128e-05, "loss": 0.18837794065475463, "step": 6340 }, { "epoch": 0.03154088393948124, "grad_norm": 2.441739320755005, "learning_rate": 1.9999876449759715e-05, "loss": 0.18233842849731446, "step": 6350 }, { "epoch": 0.0315905546228505, "grad_norm": 1.2167118787765503, "learning_rate": 1.9999868323600515e-05, "loss": 0.17658932209014894, "step": 6360 }, { "epoch": 0.031640225306219766, "grad_norm": 1.571290373802185, "learning_rate": 1.999985993864875e-05, "loss": 0.1813891649246216, "step": 6370 }, { "epoch": 0.031689895989589024, "grad_norm": 1.673170804977417, "learning_rate": 1.9999851294904628e-05, "loss": 0.17944542169570923, "step": 6380 }, { "epoch": 0.03173956667295828, "grad_norm": 2.111201763153076, "learning_rate": 1.9999842392368375e-05, "loss": 0.18344727754592896, "step": 6390 }, { "epoch": 0.03178923735632755, "grad_norm": 3.1604137420654297, "learning_rate": 1.999983323104022e-05, "loss": 0.21127738952636718, "step": 6400 }, { "epoch": 0.03183890803969681, "grad_norm": 1.7897100448608398, "learning_rate": 1.9999823810920403e-05, "loss": 0.1965092897415161, "step": 6410 }, { "epoch": 0.03188857872306607, "grad_norm": 1.7564324140548706, "learning_rate": 1.9999814132009166e-05, "loss": 0.18340462446212769, "step": 6420 }, { "epoch": 0.03193824940643533, "grad_norm": 2.5695676803588867, "learning_rate": 1.9999804194306765e-05, "loss": 0.17891623973846435, "step": 6430 }, { "epoch": 0.0319879200898046, "grad_norm": 2.280308246612549, "learning_rate": 1.9999793997813447e-05, "loss": 0.1929874062538147, "step": 6440 }, { "epoch": 0.032037590773173856, "grad_norm": 1.6075693368911743, "learning_rate": 1.999978354252948e-05, "loss": 0.1801946520805359, "step": 6450 }, { "epoch": 0.03208726145654312, "grad_norm": 1.0032427310943604, "learning_rate": 1.999977282845514e-05, "loss": 0.193949556350708, "step": 6460 }, { "epoch": 0.03213693213991238, "grad_norm": 1.5277541875839233, "learning_rate": 1.99997618555907e-05, "loss": 0.17704315185546876, "step": 6470 }, { "epoch": 0.032186602823281646, "grad_norm": 0.8996288180351257, "learning_rate": 1.9999750623936437e-05, "loss": 0.20104215145111085, "step": 6480 }, { "epoch": 0.032236273506650905, "grad_norm": 3.0303688049316406, "learning_rate": 1.9999739133492654e-05, "loss": 0.18994584083557128, "step": 6490 }, { "epoch": 0.03228594419002016, "grad_norm": 4.776918411254883, "learning_rate": 1.9999727384259646e-05, "loss": 0.1794612169265747, "step": 6500 }, { "epoch": 0.03233561487338943, "grad_norm": 1.513665795326233, "learning_rate": 1.999971537623771e-05, "loss": 0.18645414113998413, "step": 6510 }, { "epoch": 0.03238528555675869, "grad_norm": 2.6776444911956787, "learning_rate": 1.999970310942716e-05, "loss": 0.209816312789917, "step": 6520 }, { "epoch": 0.03243495624012795, "grad_norm": 2.2672574520111084, "learning_rate": 1.9999690583828315e-05, "loss": 0.19576339721679686, "step": 6530 }, { "epoch": 0.03248462692349721, "grad_norm": 2.047037363052368, "learning_rate": 1.9999677799441498e-05, "loss": 0.18727402687072753, "step": 6540 }, { "epoch": 0.03253429760686648, "grad_norm": 1.7498048543930054, "learning_rate": 1.999966475626704e-05, "loss": 0.18896055221557617, "step": 6550 }, { "epoch": 0.032583968290235736, "grad_norm": 2.299344062805176, "learning_rate": 1.999965145430528e-05, "loss": 0.18938617706298827, "step": 6560 }, { "epoch": 0.032633638973605, "grad_norm": 1.7651188373565674, "learning_rate": 1.9999637893556563e-05, "loss": 0.19028282165527344, "step": 6570 }, { "epoch": 0.03268330965697426, "grad_norm": 1.346317172050476, "learning_rate": 1.9999624074021237e-05, "loss": 0.18447561264038087, "step": 6580 }, { "epoch": 0.03273298034034352, "grad_norm": 2.7909209728240967, "learning_rate": 1.999960999569966e-05, "loss": 0.18659781217575072, "step": 6590 }, { "epoch": 0.032782651023712785, "grad_norm": 1.4024304151535034, "learning_rate": 1.9999595658592196e-05, "loss": 0.1896750807762146, "step": 6600 }, { "epoch": 0.03283232170708204, "grad_norm": 1.4841336011886597, "learning_rate": 1.9999581062699217e-05, "loss": 0.1775990128517151, "step": 6610 }, { "epoch": 0.03288199239045131, "grad_norm": 1.6550289392471313, "learning_rate": 1.99995662080211e-05, "loss": 0.1882711410522461, "step": 6620 }, { "epoch": 0.03293166307382057, "grad_norm": 5.056297779083252, "learning_rate": 1.9999551094558232e-05, "loss": 0.1922849655151367, "step": 6630 }, { "epoch": 0.03298133375718983, "grad_norm": 3.346773386001587, "learning_rate": 1.9999535722311005e-05, "loss": 0.19225859642028809, "step": 6640 }, { "epoch": 0.03303100444055909, "grad_norm": 1.9558876752853394, "learning_rate": 1.999952009127981e-05, "loss": 0.18541008234024048, "step": 6650 }, { "epoch": 0.03308067512392836, "grad_norm": 1.8062920570373535, "learning_rate": 1.9999504201465053e-05, "loss": 0.19442158937454224, "step": 6660 }, { "epoch": 0.033130345807297616, "grad_norm": 2.751258134841919, "learning_rate": 1.9999488052867156e-05, "loss": 0.19509785175323485, "step": 6670 }, { "epoch": 0.03318001649066688, "grad_norm": 1.5248245000839233, "learning_rate": 1.9999471645486524e-05, "loss": 0.182175874710083, "step": 6680 }, { "epoch": 0.03322968717403614, "grad_norm": 2.3416507244110107, "learning_rate": 1.9999454979323584e-05, "loss": 0.16830968856811523, "step": 6690 }, { "epoch": 0.0332793578574054, "grad_norm": 1.7074943780899048, "learning_rate": 1.9999438054378774e-05, "loss": 0.19703226089477538, "step": 6700 }, { "epoch": 0.033329028540774665, "grad_norm": 1.9197473526000977, "learning_rate": 1.9999420870652527e-05, "loss": 0.18079004287719727, "step": 6710 }, { "epoch": 0.03337869922414392, "grad_norm": 2.5770509243011475, "learning_rate": 1.9999403428145287e-05, "loss": 0.18488471508026122, "step": 6720 }, { "epoch": 0.03342836990751319, "grad_norm": 4.997267723083496, "learning_rate": 1.9999385726857508e-05, "loss": 0.17437491416931153, "step": 6730 }, { "epoch": 0.03347804059088245, "grad_norm": 3.196343421936035, "learning_rate": 1.9999367766789647e-05, "loss": 0.19010136127471924, "step": 6740 }, { "epoch": 0.03352771127425171, "grad_norm": 1.4225190877914429, "learning_rate": 1.999934954794217e-05, "loss": 0.18985238075256347, "step": 6750 }, { "epoch": 0.03357738195762097, "grad_norm": 1.7537826299667358, "learning_rate": 1.9999331070315544e-05, "loss": 0.2013699531555176, "step": 6760 }, { "epoch": 0.03362705264099024, "grad_norm": 2.6435132026672363, "learning_rate": 1.9999312333910254e-05, "loss": 0.21661365032196045, "step": 6770 }, { "epoch": 0.033676723324359496, "grad_norm": 2.857698440551758, "learning_rate": 1.999929333872678e-05, "loss": 0.1802372455596924, "step": 6780 }, { "epoch": 0.033726394007728755, "grad_norm": 3.1176044940948486, "learning_rate": 1.999927408476561e-05, "loss": 0.1864077091217041, "step": 6790 }, { "epoch": 0.03377606469109802, "grad_norm": 4.361641883850098, "learning_rate": 1.9999254572027252e-05, "loss": 0.1953536629676819, "step": 6800 }, { "epoch": 0.03382573537446728, "grad_norm": 3.735250473022461, "learning_rate": 1.9999234800512207e-05, "loss": 0.19222012758255005, "step": 6810 }, { "epoch": 0.033875406057836545, "grad_norm": 1.0061638355255127, "learning_rate": 1.9999214770220986e-05, "loss": 0.1744908094406128, "step": 6820 }, { "epoch": 0.033925076741205804, "grad_norm": 2.2941415309906006, "learning_rate": 1.9999194481154105e-05, "loss": 0.1828018307685852, "step": 6830 }, { "epoch": 0.03397474742457507, "grad_norm": 2.213592290878296, "learning_rate": 1.9999173933312095e-05, "loss": 0.1735187292098999, "step": 6840 }, { "epoch": 0.03402441810794433, "grad_norm": 1.2101788520812988, "learning_rate": 1.999915312669548e-05, "loss": 0.19215950965881348, "step": 6850 }, { "epoch": 0.03407408879131359, "grad_norm": 1.219704508781433, "learning_rate": 1.9999132061304803e-05, "loss": 0.19014577865600585, "step": 6860 }, { "epoch": 0.03412375947468285, "grad_norm": 1.4860621690750122, "learning_rate": 1.999911073714061e-05, "loss": 0.1848066568374634, "step": 6870 }, { "epoch": 0.03417343015805212, "grad_norm": 4.250628471374512, "learning_rate": 1.9999089154203452e-05, "loss": 0.19131821393966675, "step": 6880 }, { "epoch": 0.034223100841421376, "grad_norm": 2.2336935997009277, "learning_rate": 1.9999067312493886e-05, "loss": 0.1933651328086853, "step": 6890 }, { "epoch": 0.034272771524790635, "grad_norm": 1.193811058998108, "learning_rate": 1.999904521201248e-05, "loss": 0.18997987508773803, "step": 6900 }, { "epoch": 0.0343224422081599, "grad_norm": 3.781595230102539, "learning_rate": 1.9999022852759804e-05, "loss": 0.20240976810455322, "step": 6910 }, { "epoch": 0.03437211289152916, "grad_norm": 1.7223721742630005, "learning_rate": 1.9999000234736438e-05, "loss": 0.17756097316741942, "step": 6920 }, { "epoch": 0.034421783574898425, "grad_norm": 9.254674911499023, "learning_rate": 1.999897735794296e-05, "loss": 0.18922746181488037, "step": 6930 }, { "epoch": 0.034471454258267684, "grad_norm": 2.861865282058716, "learning_rate": 1.9998954222379976e-05, "loss": 0.1908287763595581, "step": 6940 }, { "epoch": 0.03452112494163695, "grad_norm": 2.3734092712402344, "learning_rate": 1.9998930828048074e-05, "loss": 0.193984317779541, "step": 6950 }, { "epoch": 0.03457079562500621, "grad_norm": 2.0750656127929688, "learning_rate": 1.9998907174947864e-05, "loss": 0.17678375244140626, "step": 6960 }, { "epoch": 0.034620466308375474, "grad_norm": 4.574565887451172, "learning_rate": 1.9998883263079952e-05, "loss": 0.1884387254714966, "step": 6970 }, { "epoch": 0.03467013699174473, "grad_norm": 1.7910422086715698, "learning_rate": 1.9998859092444966e-05, "loss": 0.18136060237884521, "step": 6980 }, { "epoch": 0.03471980767511399, "grad_norm": 1.9014009237289429, "learning_rate": 1.9998834663043526e-05, "loss": 0.18519384860992433, "step": 6990 }, { "epoch": 0.034769478358483256, "grad_norm": 4.275418281555176, "learning_rate": 1.9998809974876267e-05, "loss": 0.17952609062194824, "step": 7000 }, { "epoch": 0.034819149041852515, "grad_norm": 1.5870976448059082, "learning_rate": 1.9998785027943825e-05, "loss": 0.19175338745117188, "step": 7010 }, { "epoch": 0.03486881972522178, "grad_norm": 2.1649701595306396, "learning_rate": 1.9998759822246846e-05, "loss": 0.18207430839538574, "step": 7020 }, { "epoch": 0.03491849040859104, "grad_norm": 3.2826590538024902, "learning_rate": 1.9998734357785984e-05, "loss": 0.19307277202606202, "step": 7030 }, { "epoch": 0.034968161091960305, "grad_norm": 4.718287944793701, "learning_rate": 1.99987086345619e-05, "loss": 0.19331049919128418, "step": 7040 }, { "epoch": 0.035017831775329564, "grad_norm": 2.0324461460113525, "learning_rate": 1.9998682652575253e-05, "loss": 0.1875785231590271, "step": 7050 }, { "epoch": 0.03506750245869883, "grad_norm": 2.980271100997925, "learning_rate": 1.999865641182672e-05, "loss": 0.18289573192596437, "step": 7060 }, { "epoch": 0.03511717314206809, "grad_norm": 3.379962921142578, "learning_rate": 1.9998629912316976e-05, "loss": 0.19473283290863036, "step": 7070 }, { "epoch": 0.035166843825437354, "grad_norm": 3.543607234954834, "learning_rate": 1.9998603154046714e-05, "loss": 0.19347970485687255, "step": 7080 }, { "epoch": 0.03521651450880661, "grad_norm": 2.017616033554077, "learning_rate": 1.9998576137016622e-05, "loss": 0.18155897855758668, "step": 7090 }, { "epoch": 0.03526618519217587, "grad_norm": 1.4588793516159058, "learning_rate": 1.99985488612274e-05, "loss": 0.19225786924362182, "step": 7100 }, { "epoch": 0.03531585587554514, "grad_norm": 2.3607888221740723, "learning_rate": 1.9998521326679753e-05, "loss": 0.19087581634521483, "step": 7110 }, { "epoch": 0.035365526558914395, "grad_norm": 2.538808584213257, "learning_rate": 1.9998493533374396e-05, "loss": 0.1803775429725647, "step": 7120 }, { "epoch": 0.03541519724228366, "grad_norm": 1.9267258644104004, "learning_rate": 1.9998465481312045e-05, "loss": 0.19174269437789918, "step": 7130 }, { "epoch": 0.03546486792565292, "grad_norm": 3.4450161457061768, "learning_rate": 1.9998437170493427e-05, "loss": 0.190806245803833, "step": 7140 }, { "epoch": 0.035514538609022185, "grad_norm": 4.797665596008301, "learning_rate": 1.9998408600919278e-05, "loss": 0.19383248090744018, "step": 7150 }, { "epoch": 0.035564209292391444, "grad_norm": 3.2758524417877197, "learning_rate": 1.9998379772590333e-05, "loss": 0.18686105012893678, "step": 7160 }, { "epoch": 0.03561387997576071, "grad_norm": 1.4992021322250366, "learning_rate": 1.9998350685507337e-05, "loss": 0.19652796983718873, "step": 7170 }, { "epoch": 0.03566355065912997, "grad_norm": 2.6989572048187256, "learning_rate": 1.999832133967105e-05, "loss": 0.19291999340057372, "step": 7180 }, { "epoch": 0.03571322134249923, "grad_norm": 1.2524843215942383, "learning_rate": 1.9998291735082226e-05, "loss": 0.17045941352844238, "step": 7190 }, { "epoch": 0.03576289202586849, "grad_norm": 1.0930969715118408, "learning_rate": 1.999826187174163e-05, "loss": 0.19631989002227784, "step": 7200 }, { "epoch": 0.03581256270923775, "grad_norm": 2.6451263427734375, "learning_rate": 1.9998231749650036e-05, "loss": 0.18894153833389282, "step": 7210 }, { "epoch": 0.03586223339260702, "grad_norm": 2.421973943710327, "learning_rate": 1.999820136880823e-05, "loss": 0.20023541450500487, "step": 7220 }, { "epoch": 0.035911904075976275, "grad_norm": 1.8694618940353394, "learning_rate": 1.999817072921699e-05, "loss": 0.18774312734603882, "step": 7230 }, { "epoch": 0.03596157475934554, "grad_norm": 0.8615121245384216, "learning_rate": 1.9998139830877106e-05, "loss": 0.18795033693313598, "step": 7240 }, { "epoch": 0.0360112454427148, "grad_norm": 1.1376062631607056, "learning_rate": 1.999810867378939e-05, "loss": 0.18418809175491332, "step": 7250 }, { "epoch": 0.036060916126084065, "grad_norm": 3.9085192680358887, "learning_rate": 1.999807725795464e-05, "loss": 0.19342494010925293, "step": 7260 }, { "epoch": 0.036110586809453324, "grad_norm": 2.109203815460205, "learning_rate": 1.9998045583373668e-05, "loss": 0.19614911079406738, "step": 7270 }, { "epoch": 0.03616025749282259, "grad_norm": 3.1635963916778564, "learning_rate": 1.99980136500473e-05, "loss": 0.18319724798202514, "step": 7280 }, { "epoch": 0.03620992817619185, "grad_norm": 3.138136386871338, "learning_rate": 1.9997981457976358e-05, "loss": 0.1969326376914978, "step": 7290 }, { "epoch": 0.03625959885956111, "grad_norm": 2.911313533782959, "learning_rate": 1.999794900716167e-05, "loss": 0.187428617477417, "step": 7300 }, { "epoch": 0.03630926954293037, "grad_norm": 1.0615603923797607, "learning_rate": 1.999791629760409e-05, "loss": 0.19818978309631347, "step": 7310 }, { "epoch": 0.03635894022629963, "grad_norm": 1.765286922454834, "learning_rate": 1.9997883329304452e-05, "loss": 0.197043514251709, "step": 7320 }, { "epoch": 0.0364086109096689, "grad_norm": 1.3903865814208984, "learning_rate": 1.9997850102263613e-05, "loss": 0.2005133867263794, "step": 7330 }, { "epoch": 0.036458281593038155, "grad_norm": 1.8683730363845825, "learning_rate": 1.9997816616482435e-05, "loss": 0.18196537494659423, "step": 7340 }, { "epoch": 0.03650795227640742, "grad_norm": 2.5690550804138184, "learning_rate": 1.9997782871961785e-05, "loss": 0.18556376695632934, "step": 7350 }, { "epoch": 0.03655762295977668, "grad_norm": 1.6464706659317017, "learning_rate": 1.999774886870253e-05, "loss": 0.18307666778564452, "step": 7360 }, { "epoch": 0.036607293643145945, "grad_norm": 1.7213579416275024, "learning_rate": 1.9997714606705555e-05, "loss": 0.2024366855621338, "step": 7370 }, { "epoch": 0.036656964326515204, "grad_norm": 2.755427360534668, "learning_rate": 1.999768008597175e-05, "loss": 0.16558196544647216, "step": 7380 }, { "epoch": 0.03670663500988446, "grad_norm": 2.2359704971313477, "learning_rate": 1.9997645306502e-05, "loss": 0.1817310094833374, "step": 7390 }, { "epoch": 0.03675630569325373, "grad_norm": 2.9269142150878906, "learning_rate": 1.9997610268297212e-05, "loss": 0.19598331451416015, "step": 7400 }, { "epoch": 0.03680597637662299, "grad_norm": 1.204959750175476, "learning_rate": 1.999757497135829e-05, "loss": 0.1892489790916443, "step": 7410 }, { "epoch": 0.03685564705999225, "grad_norm": 3.686612844467163, "learning_rate": 1.9997539415686145e-05, "loss": 0.18446264266967774, "step": 7420 }, { "epoch": 0.03690531774336151, "grad_norm": 1.3834911584854126, "learning_rate": 1.9997503601281697e-05, "loss": 0.1864295482635498, "step": 7430 }, { "epoch": 0.03695498842673078, "grad_norm": 2.797651529312134, "learning_rate": 1.9997467528145882e-05, "loss": 0.1854156494140625, "step": 7440 }, { "epoch": 0.037004659110100036, "grad_norm": 1.8775084018707275, "learning_rate": 1.9997431196279625e-05, "loss": 0.18444132804870605, "step": 7450 }, { "epoch": 0.0370543297934693, "grad_norm": 5.172884464263916, "learning_rate": 1.9997394605683868e-05, "loss": 0.17829747200012208, "step": 7460 }, { "epoch": 0.03710400047683856, "grad_norm": 1.4751524925231934, "learning_rate": 1.999735775635956e-05, "loss": 0.1907474637031555, "step": 7470 }, { "epoch": 0.037153671160207825, "grad_norm": 2.1286251544952393, "learning_rate": 1.999732064830765e-05, "loss": 0.17594099044799805, "step": 7480 }, { "epoch": 0.037203341843577084, "grad_norm": 2.274862051010132, "learning_rate": 1.9997283281529106e-05, "loss": 0.18588297367095946, "step": 7490 }, { "epoch": 0.03725301252694634, "grad_norm": 1.190251111984253, "learning_rate": 1.9997245656024885e-05, "loss": 0.1907365083694458, "step": 7500 }, { "epoch": 0.03730268321031561, "grad_norm": 0.7568497657775879, "learning_rate": 1.999720777179597e-05, "loss": 0.19010186195373535, "step": 7510 }, { "epoch": 0.03735235389368487, "grad_norm": 1.9553078413009644, "learning_rate": 1.9997169628843336e-05, "loss": 0.17743825912475586, "step": 7520 }, { "epoch": 0.03740202457705413, "grad_norm": 1.73178231716156, "learning_rate": 1.9997131227167972e-05, "loss": 0.1995189070701599, "step": 7530 }, { "epoch": 0.03745169526042339, "grad_norm": 2.070929527282715, "learning_rate": 1.999709256677087e-05, "loss": 0.19391705989837646, "step": 7540 }, { "epoch": 0.03750136594379266, "grad_norm": 1.9667948484420776, "learning_rate": 1.9997053647653037e-05, "loss": 0.18957546949386597, "step": 7550 }, { "epoch": 0.037551036627161916, "grad_norm": 1.746222734451294, "learning_rate": 1.999701446981547e-05, "loss": 0.1941380500793457, "step": 7560 }, { "epoch": 0.03760070731053118, "grad_norm": 1.1525993347167969, "learning_rate": 1.999697503325919e-05, "loss": 0.17805712223052977, "step": 7570 }, { "epoch": 0.03765037799390044, "grad_norm": 3.8541407585144043, "learning_rate": 1.9996935337985215e-05, "loss": 0.18108220100402833, "step": 7580 }, { "epoch": 0.0377000486772697, "grad_norm": 2.7613773345947266, "learning_rate": 1.999689538399457e-05, "loss": 0.18975496292114258, "step": 7590 }, { "epoch": 0.037749719360638964, "grad_norm": 1.4162300825119019, "learning_rate": 1.9996855171288302e-05, "loss": 0.18108139038085938, "step": 7600 }, { "epoch": 0.03779939004400822, "grad_norm": 1.6876250505447388, "learning_rate": 1.9996814699867434e-05, "loss": 0.1970057010650635, "step": 7610 }, { "epoch": 0.03784906072737749, "grad_norm": 1.6302893161773682, "learning_rate": 1.999677396973302e-05, "loss": 0.18873549699783326, "step": 7620 }, { "epoch": 0.03789873141074675, "grad_norm": 2.5567734241485596, "learning_rate": 1.9996732980886117e-05, "loss": 0.17361085414886473, "step": 7630 }, { "epoch": 0.03794840209411601, "grad_norm": 0.9883072376251221, "learning_rate": 1.9996691733327786e-05, "loss": 0.18857173919677733, "step": 7640 }, { "epoch": 0.03799807277748527, "grad_norm": 1.5234135389328003, "learning_rate": 1.9996650227059092e-05, "loss": 0.18541548252105713, "step": 7650 }, { "epoch": 0.03804774346085454, "grad_norm": 1.3695720434188843, "learning_rate": 1.9996608462081107e-05, "loss": 0.17836403846740723, "step": 7660 }, { "epoch": 0.038097414144223796, "grad_norm": 1.3183900117874146, "learning_rate": 1.999656643839492e-05, "loss": 0.21072287559509278, "step": 7670 }, { "epoch": 0.03814708482759306, "grad_norm": 3.1013240814208984, "learning_rate": 1.999652415600161e-05, "loss": 0.1971733331680298, "step": 7680 }, { "epoch": 0.03819675551096232, "grad_norm": 0.8369402289390564, "learning_rate": 1.9996481614902273e-05, "loss": 0.18611875772476197, "step": 7690 }, { "epoch": 0.03824642619433158, "grad_norm": 1.2909796237945557, "learning_rate": 1.999643881509801e-05, "loss": 0.1957031011581421, "step": 7700 }, { "epoch": 0.038296096877700844, "grad_norm": 1.2962678670883179, "learning_rate": 1.9996395756589934e-05, "loss": 0.1823320746421814, "step": 7710 }, { "epoch": 0.0383457675610701, "grad_norm": 1.9351861476898193, "learning_rate": 1.999635243937915e-05, "loss": 0.18881609439849853, "step": 7720 }, { "epoch": 0.03839543824443937, "grad_norm": 0.7827179431915283, "learning_rate": 1.9996308863466788e-05, "loss": 0.17822039127349854, "step": 7730 }, { "epoch": 0.03844510892780863, "grad_norm": 3.7589685916900635, "learning_rate": 1.9996265028853972e-05, "loss": 0.19447224140167235, "step": 7740 }, { "epoch": 0.03849477961117789, "grad_norm": 1.3540079593658447, "learning_rate": 1.9996220935541834e-05, "loss": 0.1969710111618042, "step": 7750 }, { "epoch": 0.03854445029454715, "grad_norm": 1.4027432203292847, "learning_rate": 1.999617658353152e-05, "loss": 0.20385704040527344, "step": 7760 }, { "epoch": 0.03859412097791642, "grad_norm": 2.7583699226379395, "learning_rate": 1.9996131972824172e-05, "loss": 0.18813978433609008, "step": 7770 }, { "epoch": 0.038643791661285676, "grad_norm": 2.591571092605591, "learning_rate": 1.9996087103420955e-05, "loss": 0.1892712116241455, "step": 7780 }, { "epoch": 0.038693462344654934, "grad_norm": 2.0215842723846436, "learning_rate": 1.9996041975323014e-05, "loss": 0.17643401622772217, "step": 7790 }, { "epoch": 0.0387431330280242, "grad_norm": 0.7700796127319336, "learning_rate": 1.999599658853153e-05, "loss": 0.1897421360015869, "step": 7800 }, { "epoch": 0.03879280371139346, "grad_norm": 1.2896193265914917, "learning_rate": 1.9995950943047675e-05, "loss": 0.18837263584136962, "step": 7810 }, { "epoch": 0.038842474394762724, "grad_norm": 1.0837624073028564, "learning_rate": 1.999590503887263e-05, "loss": 0.18199763298034669, "step": 7820 }, { "epoch": 0.03889214507813198, "grad_norm": 2.400818109512329, "learning_rate": 1.9995858876007577e-05, "loss": 0.17886013984680177, "step": 7830 }, { "epoch": 0.03894181576150125, "grad_norm": 1.629488468170166, "learning_rate": 1.9995812454453717e-05, "loss": 0.19708781242370604, "step": 7840 }, { "epoch": 0.03899148644487051, "grad_norm": 1.9306994676589966, "learning_rate": 1.999576577421225e-05, "loss": 0.1784777045249939, "step": 7850 }, { "epoch": 0.03904115712823977, "grad_norm": 2.240954875946045, "learning_rate": 1.9995718835284385e-05, "loss": 0.19481163024902343, "step": 7860 }, { "epoch": 0.03909082781160903, "grad_norm": 5.152160167694092, "learning_rate": 1.9995671637671336e-05, "loss": 0.19424468278884888, "step": 7870 }, { "epoch": 0.0391404984949783, "grad_norm": 3.434250593185425, "learning_rate": 1.9995624181374322e-05, "loss": 0.18221142292022705, "step": 7880 }, { "epoch": 0.039190169178347556, "grad_norm": 1.0663762092590332, "learning_rate": 1.9995576466394575e-05, "loss": 0.19644631147384645, "step": 7890 }, { "epoch": 0.039239839861716815, "grad_norm": 1.068454623222351, "learning_rate": 1.9995528492733328e-05, "loss": 0.2006126880645752, "step": 7900 }, { "epoch": 0.03928951054508608, "grad_norm": 1.343475341796875, "learning_rate": 1.999548026039182e-05, "loss": 0.1712511658668518, "step": 7910 }, { "epoch": 0.03933918122845534, "grad_norm": 1.5640902519226074, "learning_rate": 1.9995431769371305e-05, "loss": 0.18141376972198486, "step": 7920 }, { "epoch": 0.039388851911824604, "grad_norm": 0.8841328024864197, "learning_rate": 1.9995383019673033e-05, "loss": 0.18365185260772704, "step": 7930 }, { "epoch": 0.03943852259519386, "grad_norm": 2.8507869243621826, "learning_rate": 1.999533401129827e-05, "loss": 0.18967721462249756, "step": 7940 }, { "epoch": 0.03948819327856313, "grad_norm": 1.7014814615249634, "learning_rate": 1.999528474424828e-05, "loss": 0.1852262496948242, "step": 7950 }, { "epoch": 0.03953786396193239, "grad_norm": 2.630533218383789, "learning_rate": 1.999523521852434e-05, "loss": 0.19309080839157106, "step": 7960 }, { "epoch": 0.03958753464530165, "grad_norm": 1.318984031677246, "learning_rate": 1.999518543412773e-05, "loss": 0.2044673442840576, "step": 7970 }, { "epoch": 0.03963720532867091, "grad_norm": 4.804283142089844, "learning_rate": 1.999513539105974e-05, "loss": 0.19194091558456422, "step": 7980 }, { "epoch": 0.03968687601204017, "grad_norm": 1.1907895803451538, "learning_rate": 1.9995085089321667e-05, "loss": 0.19696660041809083, "step": 7990 }, { "epoch": 0.039736546695409436, "grad_norm": 1.4592751264572144, "learning_rate": 1.9995034528914813e-05, "loss": 0.18930039405822754, "step": 8000 }, { "epoch": 0.039786217378778695, "grad_norm": 1.6850457191467285, "learning_rate": 1.999498370984048e-05, "loss": 0.1902056336402893, "step": 8010 }, { "epoch": 0.03983588806214796, "grad_norm": 1.4203131198883057, "learning_rate": 1.9994932632099985e-05, "loss": 0.19842348098754883, "step": 8020 }, { "epoch": 0.03988555874551722, "grad_norm": 2.294044017791748, "learning_rate": 1.9994881295694657e-05, "loss": 0.19943960905075073, "step": 8030 }, { "epoch": 0.039935229428886485, "grad_norm": 1.7936129570007324, "learning_rate": 1.999482970062582e-05, "loss": 0.18306283950805663, "step": 8040 }, { "epoch": 0.03998490011225574, "grad_norm": 1.101064682006836, "learning_rate": 1.9994777846894807e-05, "loss": 0.19348469972610474, "step": 8050 }, { "epoch": 0.04003457079562501, "grad_norm": 1.3392682075500488, "learning_rate": 1.9994725734502962e-05, "loss": 0.19878337383270264, "step": 8060 }, { "epoch": 0.04008424147899427, "grad_norm": 1.5036557912826538, "learning_rate": 1.9994673363451636e-05, "loss": 0.19430913925170898, "step": 8070 }, { "epoch": 0.04013391216236353, "grad_norm": 3.406937599182129, "learning_rate": 1.9994620733742177e-05, "loss": 0.17944064140319824, "step": 8080 }, { "epoch": 0.04018358284573279, "grad_norm": 1.2924362421035767, "learning_rate": 1.9994567845375957e-05, "loss": 0.18609817028045655, "step": 8090 }, { "epoch": 0.04023325352910205, "grad_norm": 3.778667449951172, "learning_rate": 1.999451469835434e-05, "loss": 0.18725446462631226, "step": 8100 }, { "epoch": 0.040282924212471316, "grad_norm": 1.7813550233840942, "learning_rate": 1.99944612926787e-05, "loss": 0.20286085605621337, "step": 8110 }, { "epoch": 0.040332594895840575, "grad_norm": 1.1545593738555908, "learning_rate": 1.9994407628350418e-05, "loss": 0.19091154336929322, "step": 8120 }, { "epoch": 0.04038226557920984, "grad_norm": 1.7082860469818115, "learning_rate": 1.999435370537089e-05, "loss": 0.20665473937988282, "step": 8130 }, { "epoch": 0.0404319362625791, "grad_norm": 0.9693362712860107, "learning_rate": 1.99942995237415e-05, "loss": 0.18016085624694825, "step": 8140 }, { "epoch": 0.040481606945948365, "grad_norm": 1.8497751951217651, "learning_rate": 1.999424508346366e-05, "loss": 0.20025787353515626, "step": 8150 }, { "epoch": 0.04053127762931762, "grad_norm": 0.8482716679573059, "learning_rate": 1.9994190384538777e-05, "loss": 0.19478864669799806, "step": 8160 }, { "epoch": 0.04058094831268689, "grad_norm": 1.3001701831817627, "learning_rate": 1.9994135426968263e-05, "loss": 0.19023027420043945, "step": 8170 }, { "epoch": 0.04063061899605615, "grad_norm": 2.3757762908935547, "learning_rate": 1.9994080210753548e-05, "loss": 0.17946259975433348, "step": 8180 }, { "epoch": 0.040680289679425406, "grad_norm": 1.872725009918213, "learning_rate": 1.9994024735896052e-05, "loss": 0.18996318578720092, "step": 8190 }, { "epoch": 0.04072996036279467, "grad_norm": 1.4241015911102295, "learning_rate": 1.9993969002397212e-05, "loss": 0.18268651962280275, "step": 8200 }, { "epoch": 0.04077963104616393, "grad_norm": 2.53377103805542, "learning_rate": 1.9993913010258478e-05, "loss": 0.18460774421691895, "step": 8210 }, { "epoch": 0.040829301729533196, "grad_norm": 3.5073516368865967, "learning_rate": 1.999385675948129e-05, "loss": 0.1849143385887146, "step": 8220 }, { "epoch": 0.040878972412902455, "grad_norm": 1.8302332162857056, "learning_rate": 1.9993800250067106e-05, "loss": 0.19175937175750732, "step": 8230 }, { "epoch": 0.04092864309627172, "grad_norm": 2.149559259414673, "learning_rate": 1.9993743482017393e-05, "loss": 0.19386571645736694, "step": 8240 }, { "epoch": 0.04097831377964098, "grad_norm": 1.176457405090332, "learning_rate": 1.9993686455333616e-05, "loss": 0.1827727198600769, "step": 8250 }, { "epoch": 0.041027984463010245, "grad_norm": 0.8271085023880005, "learning_rate": 1.999362917001725e-05, "loss": 0.16927241086959838, "step": 8260 }, { "epoch": 0.0410776551463795, "grad_norm": 2.9169387817382812, "learning_rate": 1.999357162606978e-05, "loss": 0.17849330902099608, "step": 8270 }, { "epoch": 0.04112732582974877, "grad_norm": 2.051633358001709, "learning_rate": 1.9993513823492695e-05, "loss": 0.2097078323364258, "step": 8280 }, { "epoch": 0.04117699651311803, "grad_norm": 1.2946205139160156, "learning_rate": 1.9993455762287492e-05, "loss": 0.1911860942840576, "step": 8290 }, { "epoch": 0.041226667196487286, "grad_norm": 1.2050282955169678, "learning_rate": 1.9993397442455672e-05, "loss": 0.20627667903900146, "step": 8300 }, { "epoch": 0.04127633787985655, "grad_norm": 3.0086560249328613, "learning_rate": 1.9993338863998744e-05, "loss": 0.1880814790725708, "step": 8310 }, { "epoch": 0.04132600856322581, "grad_norm": 1.0439400672912598, "learning_rate": 1.999328002691822e-05, "loss": 0.18997159004211425, "step": 8320 }, { "epoch": 0.041375679246595076, "grad_norm": 1.5454521179199219, "learning_rate": 1.9993220931215633e-05, "loss": 0.18920937776565552, "step": 8330 }, { "epoch": 0.041425349929964335, "grad_norm": 1.585253357887268, "learning_rate": 1.9993161576892507e-05, "loss": 0.19531573057174684, "step": 8340 }, { "epoch": 0.0414750206133336, "grad_norm": 1.2780126333236694, "learning_rate": 1.999310196395037e-05, "loss": 0.18173519372940064, "step": 8350 }, { "epoch": 0.04152469129670286, "grad_norm": 3.4169275760650635, "learning_rate": 1.999304209239078e-05, "loss": 0.19064805507659913, "step": 8360 }, { "epoch": 0.041574361980072125, "grad_norm": 2.08524227142334, "learning_rate": 1.9992981962215277e-05, "loss": 0.18499770164489746, "step": 8370 }, { "epoch": 0.041624032663441383, "grad_norm": 1.4797579050064087, "learning_rate": 1.9992921573425415e-05, "loss": 0.19890213012695312, "step": 8380 }, { "epoch": 0.04167370334681064, "grad_norm": 1.5397199392318726, "learning_rate": 1.9992860926022764e-05, "loss": 0.19090006351470948, "step": 8390 }, { "epoch": 0.04172337403017991, "grad_norm": 0.9949977993965149, "learning_rate": 1.999280002000889e-05, "loss": 0.1832919478416443, "step": 8400 }, { "epoch": 0.041773044713549166, "grad_norm": 1.982430338859558, "learning_rate": 1.999273885538537e-05, "loss": 0.18228607177734374, "step": 8410 }, { "epoch": 0.04182271539691843, "grad_norm": 3.4752249717712402, "learning_rate": 1.9992677432153783e-05, "loss": 0.20453004837036132, "step": 8420 }, { "epoch": 0.04187238608028769, "grad_norm": 1.362601399421692, "learning_rate": 1.9992615750315723e-05, "loss": 0.20635857582092285, "step": 8430 }, { "epoch": 0.041922056763656956, "grad_norm": 2.476029396057129, "learning_rate": 1.999255380987279e-05, "loss": 0.17570955753326417, "step": 8440 }, { "epoch": 0.041971727447026215, "grad_norm": 2.7348616123199463, "learning_rate": 1.999249161082658e-05, "loss": 0.20034077167510986, "step": 8450 }, { "epoch": 0.04202139813039548, "grad_norm": 2.0817365646362305, "learning_rate": 1.99924291531787e-05, "loss": 0.1986673355102539, "step": 8460 }, { "epoch": 0.04207106881376474, "grad_norm": 2.6181232929229736, "learning_rate": 1.9992366436930774e-05, "loss": 0.19557913541793823, "step": 8470 }, { "epoch": 0.042120739497134005, "grad_norm": 2.5368611812591553, "learning_rate": 1.9992303462084425e-05, "loss": 0.20365481376647948, "step": 8480 }, { "epoch": 0.042170410180503264, "grad_norm": 1.2511171102523804, "learning_rate": 1.9992240228641278e-05, "loss": 0.18873921632766724, "step": 8490 }, { "epoch": 0.04222008086387252, "grad_norm": 2.924605131149292, "learning_rate": 1.9992176736602974e-05, "loss": 0.19384504556655885, "step": 8500 }, { "epoch": 0.04226975154724179, "grad_norm": 1.8399345874786377, "learning_rate": 1.999211298597115e-05, "loss": 0.17864437103271485, "step": 8510 }, { "epoch": 0.042319422230611047, "grad_norm": 1.0766328573226929, "learning_rate": 1.999204897674746e-05, "loss": 0.19539803266525269, "step": 8520 }, { "epoch": 0.04236909291398031, "grad_norm": 2.0864450931549072, "learning_rate": 1.9991984708933563e-05, "loss": 0.1783444881439209, "step": 8530 }, { "epoch": 0.04241876359734957, "grad_norm": 2.2235374450683594, "learning_rate": 1.9991920182531116e-05, "loss": 0.19218909740447998, "step": 8540 }, { "epoch": 0.042468434280718836, "grad_norm": 1.2135193347930908, "learning_rate": 1.9991855397541796e-05, "loss": 0.18230321407318115, "step": 8550 }, { "epoch": 0.042518104964088095, "grad_norm": 1.5291516780853271, "learning_rate": 1.9991790353967272e-05, "loss": 0.18230921030044556, "step": 8560 }, { "epoch": 0.04256777564745736, "grad_norm": 0.8094042539596558, "learning_rate": 1.9991725051809233e-05, "loss": 0.19261181354522705, "step": 8570 }, { "epoch": 0.04261744633082662, "grad_norm": 2.262481451034546, "learning_rate": 1.999165949106937e-05, "loss": 0.1884602427482605, "step": 8580 }, { "epoch": 0.04266711701419588, "grad_norm": 1.343259572982788, "learning_rate": 1.9991593671749373e-05, "loss": 0.19103624820709228, "step": 8590 }, { "epoch": 0.042716787697565144, "grad_norm": 3.4696857929229736, "learning_rate": 1.999152759385095e-05, "loss": 0.18146387338638306, "step": 8600 }, { "epoch": 0.0427664583809344, "grad_norm": 1.26375412940979, "learning_rate": 1.999146125737581e-05, "loss": 0.19554804563522338, "step": 8610 }, { "epoch": 0.04281612906430367, "grad_norm": 1.508449673652649, "learning_rate": 1.999139466232567e-05, "loss": 0.18346610069274902, "step": 8620 }, { "epoch": 0.04286579974767293, "grad_norm": 2.9258673191070557, "learning_rate": 1.9991327808702253e-05, "loss": 0.19551327228546142, "step": 8630 }, { "epoch": 0.04291547043104219, "grad_norm": 1.3575226068496704, "learning_rate": 1.999126069650729e-05, "loss": 0.19322624206542968, "step": 8640 }, { "epoch": 0.04296514111441145, "grad_norm": 1.2577259540557861, "learning_rate": 1.999119332574252e-05, "loss": 0.19382684230804442, "step": 8650 }, { "epoch": 0.043014811797780717, "grad_norm": 2.0918850898742676, "learning_rate": 1.999112569640968e-05, "loss": 0.1925571084022522, "step": 8660 }, { "epoch": 0.043064482481149975, "grad_norm": 1.5131981372833252, "learning_rate": 1.999105780851053e-05, "loss": 0.19223004579544067, "step": 8670 }, { "epoch": 0.04311415316451924, "grad_norm": 2.1632022857666016, "learning_rate": 1.999098966204682e-05, "loss": 0.1969390869140625, "step": 8680 }, { "epoch": 0.0431638238478885, "grad_norm": 2.0157337188720703, "learning_rate": 1.999092125702031e-05, "loss": 0.1983858108520508, "step": 8690 }, { "epoch": 0.04321349453125776, "grad_norm": 1.473118543624878, "learning_rate": 1.9990852593432778e-05, "loss": 0.17815330028533935, "step": 8700 }, { "epoch": 0.043263165214627024, "grad_norm": 1.1298192739486694, "learning_rate": 1.9990783671285996e-05, "loss": 0.182272469997406, "step": 8710 }, { "epoch": 0.04331283589799628, "grad_norm": 1.092848539352417, "learning_rate": 1.999071449058175e-05, "loss": 0.18506252765655518, "step": 8720 }, { "epoch": 0.04336250658136555, "grad_norm": 1.4078537225723267, "learning_rate": 1.9990645051321832e-05, "loss": 0.17807412147521973, "step": 8730 }, { "epoch": 0.04341217726473481, "grad_norm": 2.7101807594299316, "learning_rate": 1.9990575353508038e-05, "loss": 0.18734285831451417, "step": 8740 }, { "epoch": 0.04346184794810407, "grad_norm": 3.1990253925323486, "learning_rate": 1.999050539714217e-05, "loss": 0.19575960636138917, "step": 8750 }, { "epoch": 0.04351151863147333, "grad_norm": 1.583275556564331, "learning_rate": 1.9990435182226037e-05, "loss": 0.19004437923431397, "step": 8760 }, { "epoch": 0.0435611893148426, "grad_norm": 2.2574689388275146, "learning_rate": 1.9990364708761462e-05, "loss": 0.1871953010559082, "step": 8770 }, { "epoch": 0.043610859998211855, "grad_norm": 1.1294900178909302, "learning_rate": 1.999029397675026e-05, "loss": 0.1980406880378723, "step": 8780 }, { "epoch": 0.043660530681581114, "grad_norm": 1.7366975545883179, "learning_rate": 1.999022298619427e-05, "loss": 0.18857386112213134, "step": 8790 }, { "epoch": 0.04371020136495038, "grad_norm": 2.068629026412964, "learning_rate": 1.9990151737095328e-05, "loss": 0.17628488540649415, "step": 8800 }, { "epoch": 0.04375987204831964, "grad_norm": 2.175743818283081, "learning_rate": 1.9990080229455275e-05, "loss": 0.19206042289733888, "step": 8810 }, { "epoch": 0.043809542731688904, "grad_norm": 3.163944721221924, "learning_rate": 1.999000846327596e-05, "loss": 0.2032494068145752, "step": 8820 }, { "epoch": 0.04385921341505816, "grad_norm": 1.0735634565353394, "learning_rate": 1.9989936438559245e-05, "loss": 0.1741531252861023, "step": 8830 }, { "epoch": 0.04390888409842743, "grad_norm": 2.4950461387634277, "learning_rate": 1.9989864155306992e-05, "loss": 0.19488863945007323, "step": 8840 }, { "epoch": 0.04395855478179669, "grad_norm": 0.7572214007377625, "learning_rate": 1.998979161352107e-05, "loss": 0.2076941967010498, "step": 8850 }, { "epoch": 0.04400822546516595, "grad_norm": 0.7296691536903381, "learning_rate": 1.9989718813203358e-05, "loss": 0.20033662319183348, "step": 8860 }, { "epoch": 0.04405789614853521, "grad_norm": 2.127110719680786, "learning_rate": 1.998964575435574e-05, "loss": 0.19645562171936035, "step": 8870 }, { "epoch": 0.04410756683190448, "grad_norm": 2.071153163909912, "learning_rate": 1.9989572436980106e-05, "loss": 0.1765273094177246, "step": 8880 }, { "epoch": 0.044157237515273735, "grad_norm": 1.9359387159347534, "learning_rate": 1.998949886107836e-05, "loss": 0.19582984447479249, "step": 8890 }, { "epoch": 0.044206908198642994, "grad_norm": 1.803133487701416, "learning_rate": 1.998942502665239e-05, "loss": 0.18552467823028565, "step": 8900 }, { "epoch": 0.04425657888201226, "grad_norm": 1.570164442062378, "learning_rate": 1.9989350933704125e-05, "loss": 0.19702503681182862, "step": 8910 }, { "epoch": 0.04430624956538152, "grad_norm": 2.393195867538452, "learning_rate": 1.9989276582235472e-05, "loss": 0.1961916208267212, "step": 8920 }, { "epoch": 0.044355920248750784, "grad_norm": 1.5722287893295288, "learning_rate": 1.998920197224835e-05, "loss": 0.18366637229919433, "step": 8930 }, { "epoch": 0.04440559093212004, "grad_norm": 2.64998722076416, "learning_rate": 1.998912710374471e-05, "loss": 0.17593650817871093, "step": 8940 }, { "epoch": 0.04445526161548931, "grad_norm": 2.4210309982299805, "learning_rate": 1.998905197672647e-05, "loss": 0.1853591561317444, "step": 8950 }, { "epoch": 0.04450493229885857, "grad_norm": 3.034041404724121, "learning_rate": 1.9988976591195582e-05, "loss": 0.1812192440032959, "step": 8960 }, { "epoch": 0.04455460298222783, "grad_norm": 2.405548572540283, "learning_rate": 1.9988900947153994e-05, "loss": 0.19798171520233154, "step": 8970 }, { "epoch": 0.04460427366559709, "grad_norm": 1.5005993843078613, "learning_rate": 1.998882504460367e-05, "loss": 0.20678448677062988, "step": 8980 }, { "epoch": 0.04465394434896635, "grad_norm": 1.3125015497207642, "learning_rate": 1.9988748883546568e-05, "loss": 0.19659849405288696, "step": 8990 }, { "epoch": 0.044703615032335615, "grad_norm": 4.282323837280273, "learning_rate": 1.9988672463984663e-05, "loss": 0.18867688179016112, "step": 9000 }, { "epoch": 0.044753285715704874, "grad_norm": 3.1316280364990234, "learning_rate": 1.998859578591993e-05, "loss": 0.1859598159790039, "step": 9010 }, { "epoch": 0.04480295639907414, "grad_norm": 2.3281803131103516, "learning_rate": 1.9988518849354356e-05, "loss": 0.1871855616569519, "step": 9020 }, { "epoch": 0.0448526270824434, "grad_norm": 1.01481032371521, "learning_rate": 1.9988441654289928e-05, "loss": 0.17842001914978028, "step": 9030 }, { "epoch": 0.044902297765812664, "grad_norm": 1.7525341510772705, "learning_rate": 1.9988364200728646e-05, "loss": 0.18859814405441283, "step": 9040 }, { "epoch": 0.04495196844918192, "grad_norm": 1.4282052516937256, "learning_rate": 1.9988286488672516e-05, "loss": 0.18514339923858641, "step": 9050 }, { "epoch": 0.04500163913255119, "grad_norm": 4.548031330108643, "learning_rate": 1.9988208518123547e-05, "loss": 0.19169044494628906, "step": 9060 }, { "epoch": 0.04505130981592045, "grad_norm": 1.3518379926681519, "learning_rate": 1.9988130289083757e-05, "loss": 0.1834939479827881, "step": 9070 }, { "epoch": 0.04510098049928971, "grad_norm": 1.3124542236328125, "learning_rate": 1.9988051801555174e-05, "loss": 0.19361294507980348, "step": 9080 }, { "epoch": 0.04515065118265897, "grad_norm": 3.359520435333252, "learning_rate": 1.9987973055539823e-05, "loss": 0.205430269241333, "step": 9090 }, { "epoch": 0.04520032186602823, "grad_norm": 1.5646183490753174, "learning_rate": 1.9987894051039745e-05, "loss": 0.18313835859298705, "step": 9100 }, { "epoch": 0.045249992549397496, "grad_norm": 4.038293838500977, "learning_rate": 1.9987814788056988e-05, "loss": 0.1769040822982788, "step": 9110 }, { "epoch": 0.045299663232766754, "grad_norm": 1.2264066934585571, "learning_rate": 1.99877352665936e-05, "loss": 0.1925806522369385, "step": 9120 }, { "epoch": 0.04534933391613602, "grad_norm": 2.6065077781677246, "learning_rate": 1.9987655486651638e-05, "loss": 0.20659205913543702, "step": 9130 }, { "epoch": 0.04539900459950528, "grad_norm": 0.8215966820716858, "learning_rate": 1.9987575448233166e-05, "loss": 0.18780891895294188, "step": 9140 }, { "epoch": 0.045448675282874544, "grad_norm": 1.7694783210754395, "learning_rate": 1.998749515134026e-05, "loss": 0.17732566595077515, "step": 9150 }, { "epoch": 0.0454983459662438, "grad_norm": 1.9075610637664795, "learning_rate": 1.9987414595974993e-05, "loss": 0.18061659336090088, "step": 9160 }, { "epoch": 0.04554801664961307, "grad_norm": 2.0313427448272705, "learning_rate": 1.9987333782139454e-05, "loss": 0.1971130609512329, "step": 9170 }, { "epoch": 0.04559768733298233, "grad_norm": 1.2280527353286743, "learning_rate": 1.9987252709835734e-05, "loss": 0.19804737567901612, "step": 9180 }, { "epoch": 0.045647358016351586, "grad_norm": 1.3429787158966064, "learning_rate": 1.9987171379065927e-05, "loss": 0.16522595882415772, "step": 9190 }, { "epoch": 0.04569702869972085, "grad_norm": 1.756858229637146, "learning_rate": 1.998708978983214e-05, "loss": 0.1794728994369507, "step": 9200 }, { "epoch": 0.04574669938309011, "grad_norm": 5.122588157653809, "learning_rate": 1.9987007942136488e-05, "loss": 0.20626215934753417, "step": 9210 }, { "epoch": 0.045796370066459376, "grad_norm": 2.2754838466644287, "learning_rate": 1.998692583598108e-05, "loss": 0.1936350703239441, "step": 9220 }, { "epoch": 0.045846040749828634, "grad_norm": 1.7017879486083984, "learning_rate": 1.9986843471368052e-05, "loss": 0.17975643873214722, "step": 9230 }, { "epoch": 0.0458957114331979, "grad_norm": 1.947868824005127, "learning_rate": 1.998676084829953e-05, "loss": 0.18420917987823487, "step": 9240 }, { "epoch": 0.04594538211656716, "grad_norm": 2.935121536254883, "learning_rate": 1.998667796677765e-05, "loss": 0.18633233308792113, "step": 9250 }, { "epoch": 0.045995052799936424, "grad_norm": 1.8366669416427612, "learning_rate": 1.9986594826804563e-05, "loss": 0.19657456874847412, "step": 9260 }, { "epoch": 0.04604472348330568, "grad_norm": 4.181309223175049, "learning_rate": 1.9986511428382416e-05, "loss": 0.17783241271972655, "step": 9270 }, { "epoch": 0.04609439416667495, "grad_norm": 1.2781078815460205, "learning_rate": 1.998642777151337e-05, "loss": 0.1894156217575073, "step": 9280 }, { "epoch": 0.04614406485004421, "grad_norm": 2.5235369205474854, "learning_rate": 1.998634385619959e-05, "loss": 0.1841883659362793, "step": 9290 }, { "epoch": 0.046193735533413466, "grad_norm": 2.0019123554229736, "learning_rate": 1.998625968244324e-05, "loss": 0.195229434967041, "step": 9300 }, { "epoch": 0.04624340621678273, "grad_norm": 1.7492727041244507, "learning_rate": 1.998617525024651e-05, "loss": 0.19998369216918946, "step": 9310 }, { "epoch": 0.04629307690015199, "grad_norm": 2.5883750915527344, "learning_rate": 1.9986090559611576e-05, "loss": 0.1838996648788452, "step": 9320 }, { "epoch": 0.046342747583521256, "grad_norm": 1.6113817691802979, "learning_rate": 1.9986005610540635e-05, "loss": 0.19552382230758666, "step": 9330 }, { "epoch": 0.046392418266890514, "grad_norm": 5.804068565368652, "learning_rate": 1.9985920403035887e-05, "loss": 0.20243921279907226, "step": 9340 }, { "epoch": 0.04644208895025978, "grad_norm": 2.753303289413452, "learning_rate": 1.998583493709953e-05, "loss": 0.18856725692749024, "step": 9350 }, { "epoch": 0.04649175963362904, "grad_norm": 1.509912133216858, "learning_rate": 1.9985749212733783e-05, "loss": 0.19611293077468872, "step": 9360 }, { "epoch": 0.046541430316998304, "grad_norm": 1.7880443334579468, "learning_rate": 1.9985663229940862e-05, "loss": 0.17913534641265869, "step": 9370 }, { "epoch": 0.04659110100036756, "grad_norm": 2.424729585647583, "learning_rate": 1.9985576988722987e-05, "loss": 0.20413331985473632, "step": 9380 }, { "epoch": 0.04664077168373682, "grad_norm": 2.4936726093292236, "learning_rate": 1.99854904890824e-05, "loss": 0.18437224626541138, "step": 9390 }, { "epoch": 0.04669044236710609, "grad_norm": 3.8576433658599854, "learning_rate": 1.998540373102133e-05, "loss": 0.19166767597198486, "step": 9400 }, { "epoch": 0.046740113050475346, "grad_norm": 1.7579423189163208, "learning_rate": 1.998531671454203e-05, "loss": 0.18554091453552246, "step": 9410 }, { "epoch": 0.04678978373384461, "grad_norm": 3.286877155303955, "learning_rate": 1.9985229439646745e-05, "loss": 0.2053157329559326, "step": 9420 }, { "epoch": 0.04683945441721387, "grad_norm": 3.0591254234313965, "learning_rate": 1.998514190633774e-05, "loss": 0.1685953378677368, "step": 9430 }, { "epoch": 0.046889125100583136, "grad_norm": 3.199199914932251, "learning_rate": 1.9985054114617277e-05, "loss": 0.18840140104293823, "step": 9440 }, { "epoch": 0.046938795783952394, "grad_norm": 1.1536264419555664, "learning_rate": 1.9984966064487627e-05, "loss": 0.18658983707427979, "step": 9450 }, { "epoch": 0.04698846646732166, "grad_norm": 1.806703805923462, "learning_rate": 1.998487775595107e-05, "loss": 0.18897318840026855, "step": 9460 }, { "epoch": 0.04703813715069092, "grad_norm": 1.2678346633911133, "learning_rate": 1.998478918900989e-05, "loss": 0.18861501216888427, "step": 9470 }, { "epoch": 0.047087807834060184, "grad_norm": 1.7472634315490723, "learning_rate": 1.998470036366638e-05, "loss": 0.18906453847885132, "step": 9480 }, { "epoch": 0.04713747851742944, "grad_norm": 2.0839126110076904, "learning_rate": 1.9984611279922842e-05, "loss": 0.19320703744888307, "step": 9490 }, { "epoch": 0.0471871492007987, "grad_norm": 1.7950834035873413, "learning_rate": 1.998452193778158e-05, "loss": 0.1785499095916748, "step": 9500 }, { "epoch": 0.04723681988416797, "grad_norm": 1.1295043230056763, "learning_rate": 1.9984432337244903e-05, "loss": 0.18180323839187623, "step": 9510 }, { "epoch": 0.047286490567537226, "grad_norm": 1.6410095691680908, "learning_rate": 1.9984342478315134e-05, "loss": 0.18364341259002687, "step": 9520 }, { "epoch": 0.04733616125090649, "grad_norm": 1.8153092861175537, "learning_rate": 1.9984252360994593e-05, "loss": 0.19316798448562622, "step": 9530 }, { "epoch": 0.04738583193427575, "grad_norm": 0.8338620066642761, "learning_rate": 1.9984161985285615e-05, "loss": 0.17457011938095093, "step": 9540 }, { "epoch": 0.047435502617645016, "grad_norm": 1.4612367153167725, "learning_rate": 1.998407135119054e-05, "loss": 0.18430919647216798, "step": 9550 }, { "epoch": 0.047485173301014275, "grad_norm": 1.1518187522888184, "learning_rate": 1.9983980458711715e-05, "loss": 0.18069422245025635, "step": 9560 }, { "epoch": 0.04753484398438354, "grad_norm": 2.4197402000427246, "learning_rate": 1.998388930785149e-05, "loss": 0.20041396617889404, "step": 9570 }, { "epoch": 0.0475845146677528, "grad_norm": 2.136801242828369, "learning_rate": 1.9983797898612217e-05, "loss": 0.18307583332061766, "step": 9580 }, { "epoch": 0.04763418535112206, "grad_norm": 1.3404631614685059, "learning_rate": 1.9983706230996276e-05, "loss": 0.1965999960899353, "step": 9590 }, { "epoch": 0.04768385603449132, "grad_norm": 2.7403838634490967, "learning_rate": 1.9983614305006032e-05, "loss": 0.1836625576019287, "step": 9600 }, { "epoch": 0.04773352671786058, "grad_norm": 2.2054405212402344, "learning_rate": 1.9983522120643864e-05, "loss": 0.19102761745452881, "step": 9610 }, { "epoch": 0.04778319740122985, "grad_norm": 3.5969865322113037, "learning_rate": 1.9983429677912155e-05, "loss": 0.1902795910835266, "step": 9620 }, { "epoch": 0.047832868084599106, "grad_norm": 1.496014952659607, "learning_rate": 1.99833369768133e-05, "loss": 0.17808406352996825, "step": 9630 }, { "epoch": 0.04788253876796837, "grad_norm": 3.2215986251831055, "learning_rate": 1.99832440173497e-05, "loss": 0.19681552648544312, "step": 9640 }, { "epoch": 0.04793220945133763, "grad_norm": 4.208413124084473, "learning_rate": 1.998315079952376e-05, "loss": 0.19061732292175293, "step": 9650 }, { "epoch": 0.047981880134706896, "grad_norm": 4.00034761428833, "learning_rate": 1.9983057323337888e-05, "loss": 0.20171923637390138, "step": 9660 }, { "epoch": 0.048031550818076155, "grad_norm": 63.53104782104492, "learning_rate": 1.998296358879451e-05, "loss": 6.974680328369141, "step": 9670 }, { "epoch": 0.04808122150144542, "grad_norm": 34.46577072143555, "learning_rate": 1.9982869595896044e-05, "loss": 1.3094185829162597, "step": 9680 }, { "epoch": 0.04813089218481468, "grad_norm": 7.02289342880249, "learning_rate": 1.998277534464493e-05, "loss": 0.3053770065307617, "step": 9690 }, { "epoch": 0.04818056286818394, "grad_norm": 2.3407013416290283, "learning_rate": 1.9982680835043605e-05, "loss": 0.25963635444641114, "step": 9700 }, { "epoch": 0.0482302335515532, "grad_norm": 3.196854591369629, "learning_rate": 1.9982586067094512e-05, "loss": 0.24773223400115968, "step": 9710 }, { "epoch": 0.04827990423492246, "grad_norm": 4.068754196166992, "learning_rate": 1.9982491040800106e-05, "loss": 0.2494257688522339, "step": 9720 }, { "epoch": 0.04832957491829173, "grad_norm": 10.205211639404297, "learning_rate": 1.9982395756162843e-05, "loss": 0.2641646385192871, "step": 9730 }, { "epoch": 0.048379245601660986, "grad_norm": 5.911604881286621, "learning_rate": 1.9982300213185196e-05, "loss": 0.26866366863250735, "step": 9740 }, { "epoch": 0.04842891628503025, "grad_norm": 6.743490219116211, "learning_rate": 1.998220441186963e-05, "loss": 0.24028384685516357, "step": 9750 }, { "epoch": 0.04847858696839951, "grad_norm": 3.68603515625, "learning_rate": 1.998210835221863e-05, "loss": 0.22679948806762695, "step": 9760 }, { "epoch": 0.048528257651768776, "grad_norm": 5.785937786102295, "learning_rate": 1.9982012034234675e-05, "loss": 0.2382153034210205, "step": 9770 }, { "epoch": 0.048577928335138035, "grad_norm": 2.5278844833374023, "learning_rate": 1.9981915457920265e-05, "loss": 0.22983598709106445, "step": 9780 }, { "epoch": 0.04862759901850729, "grad_norm": 3.6016738414764404, "learning_rate": 1.9981818623277896e-05, "loss": 0.22831976413726807, "step": 9790 }, { "epoch": 0.04867726970187656, "grad_norm": 2.3504199981689453, "learning_rate": 1.9981721530310075e-05, "loss": 0.20172529220581054, "step": 9800 }, { "epoch": 0.04872694038524582, "grad_norm": 2.0631892681121826, "learning_rate": 1.9981624179019317e-05, "loss": 0.22408840656280518, "step": 9810 }, { "epoch": 0.04877661106861508, "grad_norm": 4.1862311363220215, "learning_rate": 1.9981526569408137e-05, "loss": 0.21481914520263673, "step": 9820 }, { "epoch": 0.04882628175198434, "grad_norm": 2.8760759830474854, "learning_rate": 1.998142870147906e-05, "loss": 0.19957789182662963, "step": 9830 }, { "epoch": 0.04887595243535361, "grad_norm": 3.640606641769409, "learning_rate": 1.9981330575234623e-05, "loss": 0.21046152114868164, "step": 9840 }, { "epoch": 0.048925623118722866, "grad_norm": 8.376635551452637, "learning_rate": 1.9981232190677364e-05, "loss": 0.21513218879699708, "step": 9850 }, { "epoch": 0.04897529380209213, "grad_norm": 3.5450661182403564, "learning_rate": 1.998113354780983e-05, "loss": 0.2186683177947998, "step": 9860 }, { "epoch": 0.04902496448546139, "grad_norm": 3.2633495330810547, "learning_rate": 1.998103464663457e-05, "loss": 0.1979755640029907, "step": 9870 }, { "epoch": 0.049074635168830656, "grad_norm": 1.194901466369629, "learning_rate": 1.9980935487154153e-05, "loss": 0.1963602066040039, "step": 9880 }, { "epoch": 0.049124305852199915, "grad_norm": 2.2986533641815186, "learning_rate": 1.9980836069371133e-05, "loss": 0.23478055000305176, "step": 9890 }, { "epoch": 0.049173976535569174, "grad_norm": 6.471606254577637, "learning_rate": 1.9980736393288092e-05, "loss": 0.20367045402526857, "step": 9900 }, { "epoch": 0.04922364721893844, "grad_norm": 1.6313825845718384, "learning_rate": 1.9980636458907606e-05, "loss": 0.1979822635650635, "step": 9910 }, { "epoch": 0.0492733179023077, "grad_norm": 3.395371437072754, "learning_rate": 1.998053626623226e-05, "loss": 0.20702190399169923, "step": 9920 }, { "epoch": 0.04932298858567696, "grad_norm": 5.208441257476807, "learning_rate": 1.998043581526465e-05, "loss": 0.20352942943573, "step": 9930 }, { "epoch": 0.04937265926904622, "grad_norm": 3.9624617099761963, "learning_rate": 1.998033510600737e-05, "loss": 0.1989293336868286, "step": 9940 }, { "epoch": 0.04942232995241549, "grad_norm": 4.367539405822754, "learning_rate": 1.9980234138463034e-05, "loss": 0.18272075653076172, "step": 9950 }, { "epoch": 0.049472000635784746, "grad_norm": 3.039595365524292, "learning_rate": 1.9980132912634253e-05, "loss": 0.19371578693389893, "step": 9960 }, { "epoch": 0.04952167131915401, "grad_norm": 5.237715721130371, "learning_rate": 1.998003142852364e-05, "loss": 0.19813950061798097, "step": 9970 }, { "epoch": 0.04957134200252327, "grad_norm": 2.81891131401062, "learning_rate": 1.997992968613383e-05, "loss": 0.21813230514526366, "step": 9980 }, { "epoch": 0.04962101268589253, "grad_norm": 4.272536754608154, "learning_rate": 1.9979827685467452e-05, "loss": 0.20208392143249512, "step": 9990 }, { "epoch": 0.049670683369261795, "grad_norm": 2.8223915100097656, "learning_rate": 1.9979725426527147e-05, "loss": 0.19441179037094117, "step": 10000 }, { "epoch": 0.049720354052631054, "grad_norm": 5.145959377288818, "learning_rate": 1.997962290931556e-05, "loss": 0.20359091758728026, "step": 10010 }, { "epoch": 0.04977002473600032, "grad_norm": 3.8380959033966064, "learning_rate": 1.997952013383534e-05, "loss": 0.19262031316757203, "step": 10020 }, { "epoch": 0.04981969541936958, "grad_norm": 4.360852241516113, "learning_rate": 1.9979417100089157e-05, "loss": 0.19963674545288085, "step": 10030 }, { "epoch": 0.049869366102738844, "grad_norm": 3.0843093395233154, "learning_rate": 1.9979313808079673e-05, "loss": 0.19763617515563964, "step": 10040 }, { "epoch": 0.0499190367861081, "grad_norm": 6.045949935913086, "learning_rate": 1.997921025780956e-05, "loss": 0.2050089120864868, "step": 10050 }, { "epoch": 0.04996870746947737, "grad_norm": 2.525954484939575, "learning_rate": 1.9979106449281494e-05, "loss": 0.20092363357543946, "step": 10060 }, { "epoch": 0.050018378152846626, "grad_norm": 4.644025802612305, "learning_rate": 1.997900238249817e-05, "loss": 0.20151734352111816, "step": 10070 }, { "epoch": 0.05006804883621589, "grad_norm": 6.752222061157227, "learning_rate": 1.9978898057462272e-05, "loss": 0.19930708408355713, "step": 10080 }, { "epoch": 0.05011771951958515, "grad_norm": 7.793459415435791, "learning_rate": 1.9978793474176505e-05, "loss": 0.19151742458343507, "step": 10090 }, { "epoch": 0.05016739020295441, "grad_norm": 1.478395700454712, "learning_rate": 1.9978688632643578e-05, "loss": 0.19918529987335204, "step": 10100 }, { "epoch": 0.050217060886323675, "grad_norm": 5.800583839416504, "learning_rate": 1.99785835328662e-05, "loss": 0.19828251600265503, "step": 10110 }, { "epoch": 0.050266731569692934, "grad_norm": 2.7432851791381836, "learning_rate": 1.9978478174847097e-05, "loss": 0.19561400413513183, "step": 10120 }, { "epoch": 0.0503164022530622, "grad_norm": 7.890352249145508, "learning_rate": 1.9978372558588984e-05, "loss": 0.19781334400177003, "step": 10130 }, { "epoch": 0.05036607293643146, "grad_norm": 5.853646278381348, "learning_rate": 1.9978266684094605e-05, "loss": 0.21229760646820067, "step": 10140 }, { "epoch": 0.050415743619800724, "grad_norm": 1.9440377950668335, "learning_rate": 1.9978160551366693e-05, "loss": 0.20202314853668213, "step": 10150 }, { "epoch": 0.05046541430316998, "grad_norm": 2.5874626636505127, "learning_rate": 1.9978054160408e-05, "loss": 0.17273520231246947, "step": 10160 }, { "epoch": 0.05051508498653925, "grad_norm": 2.447329521179199, "learning_rate": 1.997794751122128e-05, "loss": 0.18241032361984252, "step": 10170 }, { "epoch": 0.05056475566990851, "grad_norm": 4.707355499267578, "learning_rate": 1.997784060380929e-05, "loss": 0.20254387855529785, "step": 10180 }, { "epoch": 0.050614426353277765, "grad_norm": 3.6908161640167236, "learning_rate": 1.9977733438174796e-05, "loss": 0.19967867136001588, "step": 10190 }, { "epoch": 0.05066409703664703, "grad_norm": 1.963938593864441, "learning_rate": 1.997762601432057e-05, "loss": 0.1944124698638916, "step": 10200 }, { "epoch": 0.05071376772001629, "grad_norm": 2.4183175563812256, "learning_rate": 1.9977518332249398e-05, "loss": 0.20306320190429689, "step": 10210 }, { "epoch": 0.050763438403385555, "grad_norm": 2.78562331199646, "learning_rate": 1.997741039196406e-05, "loss": 0.20060629844665528, "step": 10220 }, { "epoch": 0.050813109086754814, "grad_norm": 4.847365856170654, "learning_rate": 1.9977302193467355e-05, "loss": 0.2074354410171509, "step": 10230 }, { "epoch": 0.05086277977012408, "grad_norm": 1.9056421518325806, "learning_rate": 1.997719373676208e-05, "loss": 0.1892165422439575, "step": 10240 }, { "epoch": 0.05091245045349334, "grad_norm": 7.2208781242370605, "learning_rate": 1.9977085021851046e-05, "loss": 0.1777960777282715, "step": 10250 }, { "epoch": 0.050962121136862604, "grad_norm": 1.5843255519866943, "learning_rate": 1.997697604873706e-05, "loss": 0.20712656974792482, "step": 10260 }, { "epoch": 0.05101179182023186, "grad_norm": 4.201554775238037, "learning_rate": 1.9976866817422948e-05, "loss": 0.19532965421676635, "step": 10270 }, { "epoch": 0.05106146250360113, "grad_norm": 5.373702049255371, "learning_rate": 1.997675732791153e-05, "loss": 0.1802466869354248, "step": 10280 }, { "epoch": 0.05111113318697039, "grad_norm": 6.3112077713012695, "learning_rate": 1.9976647580205646e-05, "loss": 0.17435803413391113, "step": 10290 }, { "epoch": 0.051160803870339645, "grad_norm": 3.1935863494873047, "learning_rate": 1.9976537574308134e-05, "loss": 0.20024497509002687, "step": 10300 }, { "epoch": 0.05121047455370891, "grad_norm": 3.8979625701904297, "learning_rate": 1.997642731022184e-05, "loss": 0.17130939960479735, "step": 10310 }, { "epoch": 0.05126014523707817, "grad_norm": 1.8476386070251465, "learning_rate": 1.9976316787949622e-05, "loss": 0.19144608974456787, "step": 10320 }, { "epoch": 0.051309815920447435, "grad_norm": 0.8853944540023804, "learning_rate": 1.9976206007494334e-05, "loss": 0.17416481971740722, "step": 10330 }, { "epoch": 0.051359486603816694, "grad_norm": 2.7029545307159424, "learning_rate": 1.9976094968858847e-05, "loss": 0.18255542516708373, "step": 10340 }, { "epoch": 0.05140915728718596, "grad_norm": 1.168501853942871, "learning_rate": 1.9975983672046034e-05, "loss": 0.17869644165039061, "step": 10350 }, { "epoch": 0.05145882797055522, "grad_norm": 4.167721271514893, "learning_rate": 1.997587211705877e-05, "loss": 0.18797585964202881, "step": 10360 }, { "epoch": 0.051508498653924484, "grad_norm": 3.0645434856414795, "learning_rate": 1.997576030389995e-05, "loss": 0.19772076606750488, "step": 10370 }, { "epoch": 0.05155816933729374, "grad_norm": 3.2186038494110107, "learning_rate": 1.9975648232572466e-05, "loss": 0.18453891277313234, "step": 10380 }, { "epoch": 0.051607840020663, "grad_norm": 2.1690475940704346, "learning_rate": 1.9975535903079218e-05, "loss": 0.1889283299446106, "step": 10390 }, { "epoch": 0.05165751070403227, "grad_norm": 4.062684535980225, "learning_rate": 1.9975423315423108e-05, "loss": 0.1919172525405884, "step": 10400 }, { "epoch": 0.051707181387401525, "grad_norm": 1.9569443464279175, "learning_rate": 1.9975310469607056e-05, "loss": 0.19161975383758545, "step": 10410 }, { "epoch": 0.05175685207077079, "grad_norm": 1.781265139579773, "learning_rate": 1.9975197365633976e-05, "loss": 0.19059581756591798, "step": 10420 }, { "epoch": 0.05180652275414005, "grad_norm": 1.391137957572937, "learning_rate": 1.9975084003506802e-05, "loss": 0.16670093536376954, "step": 10430 }, { "epoch": 0.051856193437509315, "grad_norm": 1.5601227283477783, "learning_rate": 1.9974970383228465e-05, "loss": 0.18084900379180907, "step": 10440 }, { "epoch": 0.051905864120878574, "grad_norm": 1.9555320739746094, "learning_rate": 1.9974856504801906e-05, "loss": 0.19257738590240478, "step": 10450 }, { "epoch": 0.05195553480424784, "grad_norm": 1.3168854713439941, "learning_rate": 1.9974742368230073e-05, "loss": 0.19218252897262572, "step": 10460 }, { "epoch": 0.0520052054876171, "grad_norm": 4.515407085418701, "learning_rate": 1.9974627973515912e-05, "loss": 0.19093421697616578, "step": 10470 }, { "epoch": 0.052054876170986364, "grad_norm": 6.495380401611328, "learning_rate": 1.9974513320662393e-05, "loss": 0.18028569221496582, "step": 10480 }, { "epoch": 0.05210454685435562, "grad_norm": 4.90662145614624, "learning_rate": 1.997439840967248e-05, "loss": 0.2015596866607666, "step": 10490 }, { "epoch": 0.05215421753772488, "grad_norm": 3.7082009315490723, "learning_rate": 1.9974283240549147e-05, "loss": 0.20074453353881835, "step": 10500 }, { "epoch": 0.05220388822109415, "grad_norm": 3.3475215435028076, "learning_rate": 1.9974167813295377e-05, "loss": 0.19413859844207765, "step": 10510 }, { "epoch": 0.052253558904463406, "grad_norm": 2.35263991355896, "learning_rate": 1.997405212791415e-05, "loss": 0.2039710283279419, "step": 10520 }, { "epoch": 0.05230322958783267, "grad_norm": 5.2689080238342285, "learning_rate": 1.9973936184408465e-05, "loss": 0.18637676239013673, "step": 10530 }, { "epoch": 0.05235290027120193, "grad_norm": 6.480616092681885, "learning_rate": 1.997381998278132e-05, "loss": 0.18808149099349974, "step": 10540 }, { "epoch": 0.052402570954571195, "grad_norm": 2.2921228408813477, "learning_rate": 1.997370352303573e-05, "loss": 0.1827620506286621, "step": 10550 }, { "epoch": 0.052452241637940454, "grad_norm": 1.0856393575668335, "learning_rate": 1.99735868051747e-05, "loss": 0.1670300006866455, "step": 10560 }, { "epoch": 0.05250191232130972, "grad_norm": 1.560765266418457, "learning_rate": 1.9973469829201253e-05, "loss": 0.19990341663360595, "step": 10570 }, { "epoch": 0.05255158300467898, "grad_norm": 2.0328238010406494, "learning_rate": 1.9973352595118416e-05, "loss": 0.19698100090026854, "step": 10580 }, { "epoch": 0.05260125368804824, "grad_norm": 4.345996379852295, "learning_rate": 1.9973235102929224e-05, "loss": 0.1852494716644287, "step": 10590 }, { "epoch": 0.0526509243714175, "grad_norm": 5.05408239364624, "learning_rate": 1.9973117352636723e-05, "loss": 0.2014796257019043, "step": 10600 }, { "epoch": 0.05270059505478676, "grad_norm": 3.62317156791687, "learning_rate": 1.9972999344243948e-05, "loss": 0.19845401048660277, "step": 10610 }, { "epoch": 0.05275026573815603, "grad_norm": 1.9605445861816406, "learning_rate": 1.9972881077753967e-05, "loss": 0.18104143142700196, "step": 10620 }, { "epoch": 0.052799936421525286, "grad_norm": 4.990301132202148, "learning_rate": 1.9972762553169827e-05, "loss": 0.17512983083724976, "step": 10630 }, { "epoch": 0.05284960710489455, "grad_norm": 0.7975099086761475, "learning_rate": 1.9972643770494606e-05, "loss": 0.1711282730102539, "step": 10640 }, { "epoch": 0.05289927778826381, "grad_norm": 2.0918643474578857, "learning_rate": 1.9972524729731374e-05, "loss": 0.2020557403564453, "step": 10650 }, { "epoch": 0.052948948471633075, "grad_norm": 5.0733537673950195, "learning_rate": 1.997240543088321e-05, "loss": 0.17977864742279054, "step": 10660 }, { "epoch": 0.052998619155002334, "grad_norm": 7.4161295890808105, "learning_rate": 1.9972285873953204e-05, "loss": 0.19110180139541627, "step": 10670 }, { "epoch": 0.0530482898383716, "grad_norm": 3.523524045944214, "learning_rate": 1.9972166058944445e-05, "loss": 0.17702988386154175, "step": 10680 }, { "epoch": 0.05309796052174086, "grad_norm": 4.156544208526611, "learning_rate": 1.9972045985860046e-05, "loss": 0.21025161743164061, "step": 10690 }, { "epoch": 0.05314763120511012, "grad_norm": 4.003136157989502, "learning_rate": 1.9971925654703098e-05, "loss": 0.1892772912979126, "step": 10700 }, { "epoch": 0.05319730188847938, "grad_norm": 1.2433468103408813, "learning_rate": 1.997180506547673e-05, "loss": 0.1878098726272583, "step": 10710 }, { "epoch": 0.05324697257184864, "grad_norm": 4.178988933563232, "learning_rate": 1.9971684218184056e-05, "loss": 0.19199620485305785, "step": 10720 }, { "epoch": 0.05329664325521791, "grad_norm": 3.68572998046875, "learning_rate": 1.99715631128282e-05, "loss": 0.1886314034461975, "step": 10730 }, { "epoch": 0.053346313938587166, "grad_norm": 2.1659398078918457, "learning_rate": 1.9971441749412304e-05, "loss": 0.18853622674942017, "step": 10740 }, { "epoch": 0.05339598462195643, "grad_norm": 2.0852982997894287, "learning_rate": 1.9971320127939503e-05, "loss": 0.18914496898651123, "step": 10750 }, { "epoch": 0.05344565530532569, "grad_norm": 0.8609249591827393, "learning_rate": 1.9971198248412948e-05, "loss": 0.18060746192932128, "step": 10760 }, { "epoch": 0.053495325988694956, "grad_norm": 1.5282251834869385, "learning_rate": 1.9971076110835786e-05, "loss": 0.18510217666625978, "step": 10770 }, { "epoch": 0.053544996672064214, "grad_norm": 1.9267898797988892, "learning_rate": 1.997095371521119e-05, "loss": 0.19943259954452514, "step": 10780 }, { "epoch": 0.05359466735543347, "grad_norm": 8.051063537597656, "learning_rate": 1.9970831061542317e-05, "loss": 0.1807202696800232, "step": 10790 }, { "epoch": 0.05364433803880274, "grad_norm": 3.299224615097046, "learning_rate": 1.9970708149832346e-05, "loss": 0.18114001750946046, "step": 10800 }, { "epoch": 0.053694008722172, "grad_norm": 6.553736686706543, "learning_rate": 1.9970584980084464e-05, "loss": 0.20626358985900878, "step": 10810 }, { "epoch": 0.05374367940554126, "grad_norm": 2.235957622528076, "learning_rate": 1.9970461552301844e-05, "loss": 0.19737441539764405, "step": 10820 }, { "epoch": 0.05379335008891052, "grad_norm": 5.152070045471191, "learning_rate": 1.9970337866487687e-05, "loss": 0.22060661315917968, "step": 10830 }, { "epoch": 0.05384302077227979, "grad_norm": 1.3990485668182373, "learning_rate": 1.99702139226452e-05, "loss": 0.1962510108947754, "step": 10840 }, { "epoch": 0.053892691455649046, "grad_norm": 3.9368836879730225, "learning_rate": 1.997008972077758e-05, "loss": 0.20865540504455565, "step": 10850 }, { "epoch": 0.05394236213901831, "grad_norm": 1.8725979328155518, "learning_rate": 1.9969965260888053e-05, "loss": 0.19004602432250978, "step": 10860 }, { "epoch": 0.05399203282238757, "grad_norm": 1.7474936246871948, "learning_rate": 1.996984054297983e-05, "loss": 0.19754770994186402, "step": 10870 }, { "epoch": 0.05404170350575683, "grad_norm": 3.9942197799682617, "learning_rate": 1.9969715567056145e-05, "loss": 0.2048717498779297, "step": 10880 }, { "epoch": 0.054091374189126094, "grad_norm": 2.558885335922241, "learning_rate": 1.9969590333120225e-05, "loss": 0.19372254610061646, "step": 10890 }, { "epoch": 0.05414104487249535, "grad_norm": 4.812727451324463, "learning_rate": 1.996946484117532e-05, "loss": 0.19069074392318724, "step": 10900 }, { "epoch": 0.05419071555586462, "grad_norm": 2.012700319290161, "learning_rate": 1.9969339091224675e-05, "loss": 0.1813439130783081, "step": 10910 }, { "epoch": 0.05424038623923388, "grad_norm": 1.787481665611267, "learning_rate": 1.9969213083271535e-05, "loss": 0.1804707646369934, "step": 10920 }, { "epoch": 0.05429005692260314, "grad_norm": 2.524777889251709, "learning_rate": 1.9969086817319177e-05, "loss": 0.20035624504089355, "step": 10930 }, { "epoch": 0.0543397276059724, "grad_norm": 1.890337347984314, "learning_rate": 1.996896029337086e-05, "loss": 0.17740432024002076, "step": 10940 }, { "epoch": 0.05438939828934167, "grad_norm": 2.424029588699341, "learning_rate": 1.9968833511429855e-05, "loss": 0.1786879301071167, "step": 10950 }, { "epoch": 0.054439068972710926, "grad_norm": 1.8825136423110962, "learning_rate": 1.9968706471499447e-05, "loss": 0.19382525682449342, "step": 10960 }, { "epoch": 0.05448873965608019, "grad_norm": 3.4243369102478027, "learning_rate": 1.9968579173582926e-05, "loss": 0.1934976100921631, "step": 10970 }, { "epoch": 0.05453841033944945, "grad_norm": 2.2182319164276123, "learning_rate": 1.9968451617683584e-05, "loss": 0.19752198457717896, "step": 10980 }, { "epoch": 0.05458808102281871, "grad_norm": 2.90773606300354, "learning_rate": 1.9968323803804722e-05, "loss": 0.19050254821777343, "step": 10990 }, { "epoch": 0.054637751706187974, "grad_norm": 1.6558336019515991, "learning_rate": 1.996819573194965e-05, "loss": 0.1704142212867737, "step": 11000 }, { "epoch": 0.05468742238955723, "grad_norm": 3.5275752544403076, "learning_rate": 1.9968067402121674e-05, "loss": 0.18242104053497316, "step": 11010 }, { "epoch": 0.0547370930729265, "grad_norm": 3.085569143295288, "learning_rate": 1.996793881432413e-05, "loss": 0.18988957405090331, "step": 11020 }, { "epoch": 0.05478676375629576, "grad_norm": 2.5139315128326416, "learning_rate": 1.9967809968560332e-05, "loss": 0.18420255184173584, "step": 11030 }, { "epoch": 0.05483643443966502, "grad_norm": 2.624666690826416, "learning_rate": 1.996768086483362e-05, "loss": 0.18815579414367675, "step": 11040 }, { "epoch": 0.05488610512303428, "grad_norm": 1.4355363845825195, "learning_rate": 1.9967551503147336e-05, "loss": 0.19381728172302246, "step": 11050 }, { "epoch": 0.05493577580640355, "grad_norm": 1.3324685096740723, "learning_rate": 1.9967421883504826e-05, "loss": 0.1805586814880371, "step": 11060 }, { "epoch": 0.054985446489772806, "grad_norm": 6.5803680419921875, "learning_rate": 1.996729200590945e-05, "loss": 0.19694304466247559, "step": 11070 }, { "epoch": 0.055035117173142065, "grad_norm": 1.6343886852264404, "learning_rate": 1.996716187036456e-05, "loss": 0.18930087089538575, "step": 11080 }, { "epoch": 0.05508478785651133, "grad_norm": 3.3162825107574463, "learning_rate": 1.9967031476873527e-05, "loss": 0.18085918426513672, "step": 11090 }, { "epoch": 0.05513445853988059, "grad_norm": 3.1708641052246094, "learning_rate": 1.9966900825439733e-05, "loss": 0.1808319330215454, "step": 11100 }, { "epoch": 0.055184129223249855, "grad_norm": 1.3324929475784302, "learning_rate": 1.996676991606655e-05, "loss": 0.19102125167846679, "step": 11110 }, { "epoch": 0.05523379990661911, "grad_norm": 4.379794120788574, "learning_rate": 1.9966638748757366e-05, "loss": 0.19763798713684083, "step": 11120 }, { "epoch": 0.05528347058998838, "grad_norm": 2.001664161682129, "learning_rate": 1.9966507323515586e-05, "loss": 0.17993093729019166, "step": 11130 }, { "epoch": 0.05533314127335764, "grad_norm": 2.6899538040161133, "learning_rate": 1.9966375640344596e-05, "loss": 0.17523202896118165, "step": 11140 }, { "epoch": 0.0553828119567269, "grad_norm": 3.6475255489349365, "learning_rate": 1.9966243699247815e-05, "loss": 0.186035680770874, "step": 11150 }, { "epoch": 0.05543248264009616, "grad_norm": 1.2796047925949097, "learning_rate": 1.9966111500228655e-05, "loss": 0.17724682092666627, "step": 11160 }, { "epoch": 0.05548215332346543, "grad_norm": 1.8433866500854492, "learning_rate": 1.9965979043290537e-05, "loss": 0.18595399856567382, "step": 11170 }, { "epoch": 0.055531824006834686, "grad_norm": 2.4447672367095947, "learning_rate": 1.9965846328436887e-05, "loss": 0.1889857292175293, "step": 11180 }, { "epoch": 0.055581494690203945, "grad_norm": 0.8989034295082092, "learning_rate": 1.996571335567114e-05, "loss": 0.16773312091827391, "step": 11190 }, { "epoch": 0.05563116537357321, "grad_norm": 5.300826072692871, "learning_rate": 1.9965580124996743e-05, "loss": 0.19548311233520507, "step": 11200 }, { "epoch": 0.05568083605694247, "grad_norm": 1.5416316986083984, "learning_rate": 1.9965446636417134e-05, "loss": 0.16813089847564697, "step": 11210 }, { "epoch": 0.055730506740311735, "grad_norm": 1.4540801048278809, "learning_rate": 1.9965312889935775e-05, "loss": 0.18227038383483887, "step": 11220 }, { "epoch": 0.05578017742368099, "grad_norm": 2.9949429035186768, "learning_rate": 1.9965178885556127e-05, "loss": 0.19275989532470703, "step": 11230 }, { "epoch": 0.05582984810705026, "grad_norm": 4.08969783782959, "learning_rate": 1.9965044623281655e-05, "loss": 0.1906532883644104, "step": 11240 }, { "epoch": 0.05587951879041952, "grad_norm": 1.849341630935669, "learning_rate": 1.9964910103115836e-05, "loss": 0.17282457351684571, "step": 11250 }, { "epoch": 0.05592918947378878, "grad_norm": 2.4814536571502686, "learning_rate": 1.9964775325062145e-05, "loss": 0.19305059909820557, "step": 11260 }, { "epoch": 0.05597886015715804, "grad_norm": 2.377920389175415, "learning_rate": 1.9964640289124083e-05, "loss": 0.18912378549575806, "step": 11270 }, { "epoch": 0.0560285308405273, "grad_norm": 3.5492300987243652, "learning_rate": 1.9964504995305132e-05, "loss": 0.19272377490997314, "step": 11280 }, { "epoch": 0.056078201523896566, "grad_norm": 2.0737695693969727, "learning_rate": 1.99643694436088e-05, "loss": 0.1858086347579956, "step": 11290 }, { "epoch": 0.056127872207265825, "grad_norm": 4.259186744689941, "learning_rate": 1.9964233634038593e-05, "loss": 0.19496572017669678, "step": 11300 }, { "epoch": 0.05617754289063509, "grad_norm": 1.975564956665039, "learning_rate": 1.9964097566598024e-05, "loss": 0.17815237045288085, "step": 11310 }, { "epoch": 0.05622721357400435, "grad_norm": 2.4119603633880615, "learning_rate": 1.996396124129062e-05, "loss": 0.1728588342666626, "step": 11320 }, { "epoch": 0.056276884257373615, "grad_norm": 3.514799118041992, "learning_rate": 1.99638246581199e-05, "loss": 0.19769442081451416, "step": 11330 }, { "epoch": 0.05632655494074287, "grad_norm": 1.3493077754974365, "learning_rate": 1.996368781708941e-05, "loss": 0.1804884195327759, "step": 11340 }, { "epoch": 0.05637622562411214, "grad_norm": 2.0474233627319336, "learning_rate": 1.9963550718202685e-05, "loss": 0.18659764528274536, "step": 11350 }, { "epoch": 0.0564258963074814, "grad_norm": 1.5863031148910522, "learning_rate": 1.9963413361463272e-05, "loss": 0.18806719779968262, "step": 11360 }, { "epoch": 0.05647556699085066, "grad_norm": 2.9400317668914795, "learning_rate": 1.9963275746874724e-05, "loss": 0.16983331441879274, "step": 11370 }, { "epoch": 0.05652523767421992, "grad_norm": 1.532012701034546, "learning_rate": 1.996313787444061e-05, "loss": 0.17935314178466796, "step": 11380 }, { "epoch": 0.05657490835758918, "grad_norm": 2.746480941772461, "learning_rate": 1.996299974416449e-05, "loss": 0.1869114637374878, "step": 11390 }, { "epoch": 0.056624579040958446, "grad_norm": 5.6814422607421875, "learning_rate": 1.996286135604995e-05, "loss": 0.18434606790542601, "step": 11400 }, { "epoch": 0.056674249724327705, "grad_norm": 2.5970399379730225, "learning_rate": 1.996272271010056e-05, "loss": 0.1792165994644165, "step": 11410 }, { "epoch": 0.05672392040769697, "grad_norm": 2.246572256088257, "learning_rate": 1.996258380631991e-05, "loss": 0.19713904857635497, "step": 11420 }, { "epoch": 0.05677359109106623, "grad_norm": 2.4854984283447266, "learning_rate": 1.9962444644711597e-05, "loss": 0.19026331901550292, "step": 11430 }, { "epoch": 0.056823261774435495, "grad_norm": 4.8888630867004395, "learning_rate": 1.9962305225279224e-05, "loss": 0.2058253526687622, "step": 11440 }, { "epoch": 0.056872932457804753, "grad_norm": 1.6347169876098633, "learning_rate": 1.9962165548026398e-05, "loss": 0.18952717781066894, "step": 11450 }, { "epoch": 0.05692260314117402, "grad_norm": 1.6090458631515503, "learning_rate": 1.9962025612956732e-05, "loss": 0.18434903621673585, "step": 11460 }, { "epoch": 0.05697227382454328, "grad_norm": 1.9903364181518555, "learning_rate": 1.996188542007385e-05, "loss": 0.18865965604782103, "step": 11470 }, { "epoch": 0.057021944507912536, "grad_norm": 8.222525596618652, "learning_rate": 1.9961744969381377e-05, "loss": 0.19856867790222169, "step": 11480 }, { "epoch": 0.0570716151912818, "grad_norm": 4.565370559692383, "learning_rate": 1.9961604260882952e-05, "loss": 0.1924126386642456, "step": 11490 }, { "epoch": 0.05712128587465106, "grad_norm": 2.229748010635376, "learning_rate": 1.996146329458221e-05, "loss": 0.18376510143280028, "step": 11500 }, { "epoch": 0.057170956558020326, "grad_norm": 1.8786078691482544, "learning_rate": 1.9961322070482806e-05, "loss": 0.1887008547782898, "step": 11510 }, { "epoch": 0.057220627241389585, "grad_norm": 2.411583423614502, "learning_rate": 1.9961180588588392e-05, "loss": 0.1917503833770752, "step": 11520 }, { "epoch": 0.05727029792475885, "grad_norm": 2.8629345893859863, "learning_rate": 1.996103884890263e-05, "loss": 0.19327574968338013, "step": 11530 }, { "epoch": 0.05731996860812811, "grad_norm": 2.0672190189361572, "learning_rate": 1.9960896851429187e-05, "loss": 0.20025410652160644, "step": 11540 }, { "epoch": 0.057369639291497375, "grad_norm": 3.235623359680176, "learning_rate": 1.996075459617174e-05, "loss": 0.18599112033843995, "step": 11550 }, { "epoch": 0.057419309974866634, "grad_norm": 2.2566428184509277, "learning_rate": 1.996061208313397e-05, "loss": 0.19602956771850585, "step": 11560 }, { "epoch": 0.0574689806582359, "grad_norm": 2.980665922164917, "learning_rate": 1.996046931231956e-05, "loss": 0.18469457626342772, "step": 11570 }, { "epoch": 0.05751865134160516, "grad_norm": 1.617379069328308, "learning_rate": 1.9960326283732212e-05, "loss": 0.1779223084449768, "step": 11580 }, { "epoch": 0.057568322024974417, "grad_norm": 3.200010299682617, "learning_rate": 1.9960182997375624e-05, "loss": 0.19278723001480103, "step": 11590 }, { "epoch": 0.05761799270834368, "grad_norm": 1.8591700792312622, "learning_rate": 1.9960039453253505e-05, "loss": 0.20155200958251954, "step": 11600 }, { "epoch": 0.05766766339171294, "grad_norm": 1.394624948501587, "learning_rate": 1.995989565136957e-05, "loss": 0.19243301153182985, "step": 11610 }, { "epoch": 0.057717334075082206, "grad_norm": 2.083488941192627, "learning_rate": 1.9959751591727538e-05, "loss": 0.18867433071136475, "step": 11620 }, { "epoch": 0.057767004758451465, "grad_norm": 1.3689640760421753, "learning_rate": 1.9959607274331143e-05, "loss": 0.17438679933547974, "step": 11630 }, { "epoch": 0.05781667544182073, "grad_norm": 2.332979917526245, "learning_rate": 1.9959462699184112e-05, "loss": 0.17963968515396117, "step": 11640 }, { "epoch": 0.05786634612518999, "grad_norm": 3.9746406078338623, "learning_rate": 1.9959317866290194e-05, "loss": 0.19851118326187134, "step": 11650 }, { "epoch": 0.057916016808559255, "grad_norm": 1.3907946348190308, "learning_rate": 1.9959172775653136e-05, "loss": 0.1762930631637573, "step": 11660 }, { "epoch": 0.057965687491928514, "grad_norm": 2.8181333541870117, "learning_rate": 1.995902742727669e-05, "loss": 0.1952437400817871, "step": 11670 }, { "epoch": 0.05801535817529777, "grad_norm": 2.299896240234375, "learning_rate": 1.9958881821164616e-05, "loss": 0.19252214431762696, "step": 11680 }, { "epoch": 0.05806502885866704, "grad_norm": 4.705321311950684, "learning_rate": 1.9958735957320687e-05, "loss": 0.18294528722763062, "step": 11690 }, { "epoch": 0.0581146995420363, "grad_norm": 4.295843124389648, "learning_rate": 1.9958589835748675e-05, "loss": 0.18320562839508056, "step": 11700 }, { "epoch": 0.05816437022540556, "grad_norm": 3.080634832382202, "learning_rate": 1.995844345645236e-05, "loss": 0.18878769874572754, "step": 11710 }, { "epoch": 0.05821404090877482, "grad_norm": 2.313220977783203, "learning_rate": 1.9958296819435534e-05, "loss": 0.1819305419921875, "step": 11720 }, { "epoch": 0.058263711592144087, "grad_norm": 2.8737242221832275, "learning_rate": 1.995814992470199e-05, "loss": 0.18961172103881835, "step": 11730 }, { "epoch": 0.058313382275513345, "grad_norm": 3.0714099407196045, "learning_rate": 1.9958002772255533e-05, "loss": 0.18628530502319335, "step": 11740 }, { "epoch": 0.05836305295888261, "grad_norm": 3.308664083480835, "learning_rate": 1.9957855362099966e-05, "loss": 0.18899638652801515, "step": 11750 }, { "epoch": 0.05841272364225187, "grad_norm": 3.1781649589538574, "learning_rate": 1.9957707694239106e-05, "loss": 0.18490247726440429, "step": 11760 }, { "epoch": 0.058462394325621135, "grad_norm": 1.9054657220840454, "learning_rate": 1.9957559768676777e-05, "loss": 0.19117902517318724, "step": 11770 }, { "epoch": 0.058512065008990394, "grad_norm": 2.414731025695801, "learning_rate": 1.9957411585416803e-05, "loss": 0.18151715993881226, "step": 11780 }, { "epoch": 0.05856173569235965, "grad_norm": 2.7231128215789795, "learning_rate": 1.995726314446302e-05, "loss": 0.19090330600738525, "step": 11790 }, { "epoch": 0.05861140637572892, "grad_norm": 1.0931742191314697, "learning_rate": 1.995711444581927e-05, "loss": 0.1960313320159912, "step": 11800 }, { "epoch": 0.05866107705909818, "grad_norm": 1.5834863185882568, "learning_rate": 1.9956965489489403e-05, "loss": 0.18831613063812255, "step": 11810 }, { "epoch": 0.05871074774246744, "grad_norm": 3.6748781204223633, "learning_rate": 1.995681627547727e-05, "loss": 0.18419270515441893, "step": 11820 }, { "epoch": 0.0587604184258367, "grad_norm": 2.383744955062866, "learning_rate": 1.995666680378674e-05, "loss": 0.18480443954467773, "step": 11830 }, { "epoch": 0.05881008910920597, "grad_norm": 2.3934500217437744, "learning_rate": 1.9956517074421674e-05, "loss": 0.19455255270004274, "step": 11840 }, { "epoch": 0.058859759792575225, "grad_norm": 2.6027750968933105, "learning_rate": 1.9956367087385947e-05, "loss": 0.19643479585647583, "step": 11850 }, { "epoch": 0.05890943047594449, "grad_norm": 3.093099594116211, "learning_rate": 1.9956216842683448e-05, "loss": 0.18576955795288086, "step": 11860 }, { "epoch": 0.05895910115931375, "grad_norm": 1.588609218597412, "learning_rate": 1.995606634031806e-05, "loss": 0.19142663478851318, "step": 11870 }, { "epoch": 0.05900877184268301, "grad_norm": 3.3718857765197754, "learning_rate": 1.9955915580293674e-05, "loss": 0.1907932996749878, "step": 11880 }, { "epoch": 0.059058442526052274, "grad_norm": 0.7232437133789062, "learning_rate": 1.9955764562614196e-05, "loss": 0.1964329719543457, "step": 11890 }, { "epoch": 0.05910811320942153, "grad_norm": 1.1793677806854248, "learning_rate": 1.9955613287283538e-05, "loss": 0.19542447328567505, "step": 11900 }, { "epoch": 0.0591577838927908, "grad_norm": 0.8697139024734497, "learning_rate": 1.995546175430561e-05, "loss": 0.1836245059967041, "step": 11910 }, { "epoch": 0.05920745457616006, "grad_norm": 2.38850736618042, "learning_rate": 1.9955309963684336e-05, "loss": 0.1805813193321228, "step": 11920 }, { "epoch": 0.05925712525952932, "grad_norm": 3.1422762870788574, "learning_rate": 1.995515791542364e-05, "loss": 0.18455114364624023, "step": 11930 }, { "epoch": 0.05930679594289858, "grad_norm": 2.1355748176574707, "learning_rate": 1.995500560952746e-05, "loss": 0.1960806965827942, "step": 11940 }, { "epoch": 0.05935646662626785, "grad_norm": 3.5272955894470215, "learning_rate": 1.995485304599974e-05, "loss": 0.19669474363327027, "step": 11950 }, { "epoch": 0.059406137309637105, "grad_norm": 2.6885123252868652, "learning_rate": 1.9954700224844425e-05, "loss": 0.20634090900421143, "step": 11960 }, { "epoch": 0.05945580799300637, "grad_norm": 2.1994566917419434, "learning_rate": 1.995454714606547e-05, "loss": 0.19152956008911132, "step": 11970 }, { "epoch": 0.05950547867637563, "grad_norm": 5.623774528503418, "learning_rate": 1.995439380966684e-05, "loss": 0.18533297777175903, "step": 11980 }, { "epoch": 0.05955514935974489, "grad_norm": 2.3207592964172363, "learning_rate": 1.9954240215652497e-05, "loss": 0.19525905847549438, "step": 11990 }, { "epoch": 0.059604820043114154, "grad_norm": 5.598517894744873, "learning_rate": 1.995408636402642e-05, "loss": 0.19785131216049195, "step": 12000 }, { "epoch": 0.05965449072648341, "grad_norm": 3.514256715774536, "learning_rate": 1.9953932254792592e-05, "loss": 0.18512542247772218, "step": 12010 }, { "epoch": 0.05970416140985268, "grad_norm": 2.7381482124328613, "learning_rate": 1.9953777887955e-05, "loss": 0.18680815696716307, "step": 12020 }, { "epoch": 0.05975383209322194, "grad_norm": 3.3716301918029785, "learning_rate": 1.9953623263517636e-05, "loss": 0.19853196144104004, "step": 12030 }, { "epoch": 0.0598035027765912, "grad_norm": 1.1183518171310425, "learning_rate": 1.9953468381484507e-05, "loss": 0.20080099105834961, "step": 12040 }, { "epoch": 0.05985317345996046, "grad_norm": 3.4464261531829834, "learning_rate": 1.9953313241859616e-05, "loss": 0.1914297342300415, "step": 12050 }, { "epoch": 0.05990284414332973, "grad_norm": 1.9181591272354126, "learning_rate": 1.995315784464698e-05, "loss": 0.19494636058807374, "step": 12060 }, { "epoch": 0.059952514826698985, "grad_norm": 1.6161350011825562, "learning_rate": 1.9953002189850626e-05, "loss": 0.18777129650115967, "step": 12070 }, { "epoch": 0.060002185510068244, "grad_norm": 1.0025947093963623, "learning_rate": 1.9952846277474572e-05, "loss": 0.19381361007690429, "step": 12080 }, { "epoch": 0.06005185619343751, "grad_norm": 4.114737033843994, "learning_rate": 1.995269010752286e-05, "loss": 0.192064106464386, "step": 12090 }, { "epoch": 0.06010152687680677, "grad_norm": 2.890122652053833, "learning_rate": 1.995253367999953e-05, "loss": 0.1909679651260376, "step": 12100 }, { "epoch": 0.060151197560176034, "grad_norm": 1.4449480772018433, "learning_rate": 1.995237699490863e-05, "loss": 0.18480526208877562, "step": 12110 }, { "epoch": 0.06020086824354529, "grad_norm": 6.30374002456665, "learning_rate": 1.9952220052254218e-05, "loss": 0.20561800003051758, "step": 12120 }, { "epoch": 0.06025053892691456, "grad_norm": 2.6956605911254883, "learning_rate": 1.995206285204035e-05, "loss": 0.19729410409927367, "step": 12130 }, { "epoch": 0.06030020961028382, "grad_norm": 3.58213472366333, "learning_rate": 1.9951905394271094e-05, "loss": 0.1831960916519165, "step": 12140 }, { "epoch": 0.06034988029365308, "grad_norm": 1.7465636730194092, "learning_rate": 1.995174767895053e-05, "loss": 0.2032681941986084, "step": 12150 }, { "epoch": 0.06039955097702234, "grad_norm": 2.951094388961792, "learning_rate": 1.995158970608274e-05, "loss": 0.1779225468635559, "step": 12160 }, { "epoch": 0.06044922166039161, "grad_norm": 2.053384780883789, "learning_rate": 1.9951431475671804e-05, "loss": 0.1791072368621826, "step": 12170 }, { "epoch": 0.060498892343760866, "grad_norm": 2.8575005531311035, "learning_rate": 1.9951272987721826e-05, "loss": 0.18072330951690674, "step": 12180 }, { "epoch": 0.060548563027130124, "grad_norm": 3.8326942920684814, "learning_rate": 1.9951114242236906e-05, "loss": 0.18391190767288207, "step": 12190 }, { "epoch": 0.06059823371049939, "grad_norm": 2.4187235832214355, "learning_rate": 1.995095523922115e-05, "loss": 0.1826360821723938, "step": 12200 }, { "epoch": 0.06064790439386865, "grad_norm": 1.2354031801223755, "learning_rate": 1.995079597867867e-05, "loss": 0.17270867824554442, "step": 12210 }, { "epoch": 0.060697575077237914, "grad_norm": 1.5850189924240112, "learning_rate": 1.9950636460613596e-05, "loss": 0.16989301443099974, "step": 12220 }, { "epoch": 0.06074724576060717, "grad_norm": 2.4397270679473877, "learning_rate": 1.9950476685030052e-05, "loss": 0.18611016273498535, "step": 12230 }, { "epoch": 0.06079691644397644, "grad_norm": 2.219385862350464, "learning_rate": 1.9950316651932167e-05, "loss": 0.1860527753829956, "step": 12240 }, { "epoch": 0.0608465871273457, "grad_norm": 1.9963260889053345, "learning_rate": 1.995015636132409e-05, "loss": 0.2004610300064087, "step": 12250 }, { "epoch": 0.06089625781071496, "grad_norm": 1.7567542791366577, "learning_rate": 1.994999581320997e-05, "loss": 0.1852891445159912, "step": 12260 }, { "epoch": 0.06094592849408422, "grad_norm": 2.5415875911712646, "learning_rate": 1.9949835007593954e-05, "loss": 0.19123475551605223, "step": 12270 }, { "epoch": 0.06099559917745348, "grad_norm": 1.8038532733917236, "learning_rate": 1.9949673944480214e-05, "loss": 0.20365192890167236, "step": 12280 }, { "epoch": 0.061045269860822746, "grad_norm": 3.993732452392578, "learning_rate": 1.994951262387291e-05, "loss": 0.1934351921081543, "step": 12290 }, { "epoch": 0.061094940544192004, "grad_norm": 1.40989351272583, "learning_rate": 1.994935104577622e-05, "loss": 0.18438655138015747, "step": 12300 }, { "epoch": 0.06114461122756127, "grad_norm": 2.7375824451446533, "learning_rate": 1.9949189210194328e-05, "loss": 0.18893117904663087, "step": 12310 }, { "epoch": 0.06119428191093053, "grad_norm": 3.264561176300049, "learning_rate": 1.9949027117131415e-05, "loss": 0.21677815914154053, "step": 12320 }, { "epoch": 0.061243952594299794, "grad_norm": 1.780203938484192, "learning_rate": 1.994886476659168e-05, "loss": 0.17740567922592163, "step": 12330 }, { "epoch": 0.06129362327766905, "grad_norm": 2.57875394821167, "learning_rate": 1.9948702158579333e-05, "loss": 0.1729417324066162, "step": 12340 }, { "epoch": 0.06134329396103832, "grad_norm": 2.084005832672119, "learning_rate": 1.9948539293098567e-05, "loss": 0.18471813201904297, "step": 12350 }, { "epoch": 0.06139296464440758, "grad_norm": 1.722626805305481, "learning_rate": 1.9948376170153606e-05, "loss": 0.18721516132354737, "step": 12360 }, { "epoch": 0.06144263532777684, "grad_norm": 1.3687351942062378, "learning_rate": 1.9948212789748667e-05, "loss": 0.1937612771987915, "step": 12370 }, { "epoch": 0.0614923060111461, "grad_norm": 2.5697102546691895, "learning_rate": 1.9948049151887987e-05, "loss": 0.18438904285430907, "step": 12380 }, { "epoch": 0.06154197669451536, "grad_norm": 1.202816128730774, "learning_rate": 1.9947885256575792e-05, "loss": 0.18654220104217528, "step": 12390 }, { "epoch": 0.061591647377884626, "grad_norm": 3.825728416442871, "learning_rate": 1.9947721103816325e-05, "loss": 0.168682861328125, "step": 12400 }, { "epoch": 0.061641318061253884, "grad_norm": 1.3174723386764526, "learning_rate": 1.994755669361384e-05, "loss": 0.2032010078430176, "step": 12410 }, { "epoch": 0.06169098874462315, "grad_norm": 2.7250587940216064, "learning_rate": 1.9947392025972584e-05, "loss": 0.18170409202575682, "step": 12420 }, { "epoch": 0.06174065942799241, "grad_norm": 2.866619110107422, "learning_rate": 1.994722710089682e-05, "loss": 0.19453351497650145, "step": 12430 }, { "epoch": 0.061790330111361674, "grad_norm": 2.77901554107666, "learning_rate": 1.9947061918390825e-05, "loss": 0.19382824897766113, "step": 12440 }, { "epoch": 0.06184000079473093, "grad_norm": 1.6234337091445923, "learning_rate": 1.9946896478458862e-05, "loss": 0.187169086933136, "step": 12450 }, { "epoch": 0.0618896714781002, "grad_norm": 1.7409019470214844, "learning_rate": 1.994673078110522e-05, "loss": 0.1847991704940796, "step": 12460 }, { "epoch": 0.06193934216146946, "grad_norm": 1.0363420248031616, "learning_rate": 1.9946564826334184e-05, "loss": 0.18921250104904175, "step": 12470 }, { "epoch": 0.061989012844838716, "grad_norm": 1.6374191045761108, "learning_rate": 1.9946398614150048e-05, "loss": 0.1929882287979126, "step": 12480 }, { "epoch": 0.06203868352820798, "grad_norm": 3.5618643760681152, "learning_rate": 1.994623214455712e-05, "loss": 0.1967092752456665, "step": 12490 }, { "epoch": 0.06208835421157724, "grad_norm": 1.9377866983413696, "learning_rate": 1.9946065417559697e-05, "loss": 0.1883978486061096, "step": 12500 }, { "epoch": 0.062138024894946506, "grad_norm": 2.3825137615203857, "learning_rate": 1.99458984331621e-05, "loss": 0.18225908279418945, "step": 12510 }, { "epoch": 0.062187695578315764, "grad_norm": 2.3769476413726807, "learning_rate": 1.9945731191368655e-05, "loss": 0.1765909433364868, "step": 12520 }, { "epoch": 0.06223736626168503, "grad_norm": 6.51134729385376, "learning_rate": 1.9945563692183686e-05, "loss": 0.1987217903137207, "step": 12530 }, { "epoch": 0.06228703694505429, "grad_norm": 4.364565372467041, "learning_rate": 1.9945395935611523e-05, "loss": 0.20085396766662597, "step": 12540 }, { "epoch": 0.062336707628423554, "grad_norm": 4.373596668243408, "learning_rate": 1.9945227921656517e-05, "loss": 0.2011564254760742, "step": 12550 }, { "epoch": 0.06238637831179281, "grad_norm": 2.6492178440093994, "learning_rate": 1.9945059650323007e-05, "loss": 0.19576597213745117, "step": 12560 }, { "epoch": 0.06243604899516208, "grad_norm": 2.8484339714050293, "learning_rate": 1.9944891121615356e-05, "loss": 0.19143091440200805, "step": 12570 }, { "epoch": 0.06248571967853134, "grad_norm": 0.7480109333992004, "learning_rate": 1.9944722335537922e-05, "loss": 0.20010035037994384, "step": 12580 }, { "epoch": 0.0625353903619006, "grad_norm": 3.402941942214966, "learning_rate": 1.9944553292095067e-05, "loss": 0.19352350234985352, "step": 12590 }, { "epoch": 0.06258506104526985, "grad_norm": 2.492265462875366, "learning_rate": 1.9944383991291178e-05, "loss": 0.19598495960235596, "step": 12600 }, { "epoch": 0.06263473172863912, "grad_norm": 4.091788291931152, "learning_rate": 1.9944214433130626e-05, "loss": 0.19773731231689454, "step": 12610 }, { "epoch": 0.06268440241200839, "grad_norm": 2.761965751647949, "learning_rate": 1.9944044617617798e-05, "loss": 0.1970674753189087, "step": 12620 }, { "epoch": 0.06273407309537765, "grad_norm": 2.433762311935425, "learning_rate": 1.9943874544757103e-05, "loss": 0.17999515533447266, "step": 12630 }, { "epoch": 0.0627837437787469, "grad_norm": 1.0200916528701782, "learning_rate": 1.9943704214552925e-05, "loss": 0.17099218368530272, "step": 12640 }, { "epoch": 0.06283341446211617, "grad_norm": 1.112032413482666, "learning_rate": 1.9943533627009685e-05, "loss": 0.18659577369689942, "step": 12650 }, { "epoch": 0.06288308514548543, "grad_norm": 1.1213034391403198, "learning_rate": 1.9943362782131794e-05, "loss": 0.17623168230056763, "step": 12660 }, { "epoch": 0.0629327558288547, "grad_norm": 3.0616793632507324, "learning_rate": 1.9943191679923668e-05, "loss": 0.18245605230331421, "step": 12670 }, { "epoch": 0.06298242651222395, "grad_norm": 2.857860565185547, "learning_rate": 1.994302032038974e-05, "loss": 0.1742384433746338, "step": 12680 }, { "epoch": 0.06303209719559322, "grad_norm": 0.9841245412826538, "learning_rate": 1.994284870353445e-05, "loss": 0.18376080989837645, "step": 12690 }, { "epoch": 0.06308176787896248, "grad_norm": 2.8538780212402344, "learning_rate": 1.9942676829362226e-05, "loss": 0.17738571166992187, "step": 12700 }, { "epoch": 0.06313143856233173, "grad_norm": 2.492175817489624, "learning_rate": 1.9942504697877525e-05, "loss": 0.20648536682128907, "step": 12710 }, { "epoch": 0.063181109245701, "grad_norm": 4.293769836425781, "learning_rate": 1.9942332309084804e-05, "loss": 0.18305684328079225, "step": 12720 }, { "epoch": 0.06323077992907027, "grad_norm": 4.084159851074219, "learning_rate": 1.994215966298852e-05, "loss": 0.1888105869293213, "step": 12730 }, { "epoch": 0.06328045061243953, "grad_norm": 4.263660430908203, "learning_rate": 1.9941986759593135e-05, "loss": 0.1859895706176758, "step": 12740 }, { "epoch": 0.06333012129580878, "grad_norm": 2.2701756954193115, "learning_rate": 1.9941813598903134e-05, "loss": 0.21073989868164061, "step": 12750 }, { "epoch": 0.06337979197917805, "grad_norm": 1.3938764333724976, "learning_rate": 1.9941640180922997e-05, "loss": 0.17860349416732788, "step": 12760 }, { "epoch": 0.06342946266254731, "grad_norm": 1.7759230136871338, "learning_rate": 1.9941466505657208e-05, "loss": 0.19995617866516113, "step": 12770 }, { "epoch": 0.06347913334591657, "grad_norm": 2.6225414276123047, "learning_rate": 1.994129257311026e-05, "loss": 0.1806251287460327, "step": 12780 }, { "epoch": 0.06352880402928583, "grad_norm": 2.5657401084899902, "learning_rate": 1.994111838328666e-05, "loss": 0.19961471557617189, "step": 12790 }, { "epoch": 0.0635784747126551, "grad_norm": 1.6054635047912598, "learning_rate": 1.9940943936190915e-05, "loss": 0.201645827293396, "step": 12800 }, { "epoch": 0.06362814539602436, "grad_norm": 1.8911857604980469, "learning_rate": 1.9940769231827536e-05, "loss": 0.1943213701248169, "step": 12810 }, { "epoch": 0.06367781607939361, "grad_norm": 3.562560796737671, "learning_rate": 1.9940594270201047e-05, "loss": 0.19509330987930298, "step": 12820 }, { "epoch": 0.06372748676276288, "grad_norm": 4.087702751159668, "learning_rate": 1.9940419051315975e-05, "loss": 0.18979339599609374, "step": 12830 }, { "epoch": 0.06377715744613215, "grad_norm": 2.3296613693237305, "learning_rate": 1.9940243575176856e-05, "loss": 0.17892112731933593, "step": 12840 }, { "epoch": 0.06382682812950141, "grad_norm": 2.3296544551849365, "learning_rate": 1.994006784178823e-05, "loss": 0.20620315074920653, "step": 12850 }, { "epoch": 0.06387649881287066, "grad_norm": 1.7174785137176514, "learning_rate": 1.993989185115464e-05, "loss": 0.1931580901145935, "step": 12860 }, { "epoch": 0.06392616949623993, "grad_norm": 2.610215663909912, "learning_rate": 1.9939715603280652e-05, "loss": 0.19722052812576293, "step": 12870 }, { "epoch": 0.0639758401796092, "grad_norm": 3.639677047729492, "learning_rate": 1.9939539098170818e-05, "loss": 0.21047589778900147, "step": 12880 }, { "epoch": 0.06402551086297845, "grad_norm": 1.4046729803085327, "learning_rate": 1.993936233582971e-05, "loss": 0.1784581184387207, "step": 12890 }, { "epoch": 0.06407518154634771, "grad_norm": 2.367210865020752, "learning_rate": 1.99391853162619e-05, "loss": 0.18700376749038697, "step": 12900 }, { "epoch": 0.06412485222971698, "grad_norm": 2.4448862075805664, "learning_rate": 1.993900803947197e-05, "loss": 0.1955185651779175, "step": 12910 }, { "epoch": 0.06417452291308624, "grad_norm": 1.3318227529525757, "learning_rate": 1.9938830505464507e-05, "loss": 0.18967885971069337, "step": 12920 }, { "epoch": 0.0642241935964555, "grad_norm": 0.9591392278671265, "learning_rate": 1.9938652714244112e-05, "loss": 0.19302446842193605, "step": 12930 }, { "epoch": 0.06427386427982476, "grad_norm": 1.6380339860916138, "learning_rate": 1.9938474665815376e-05, "loss": 0.19074240922927857, "step": 12940 }, { "epoch": 0.06432353496319403, "grad_norm": 2.425166606903076, "learning_rate": 1.9938296360182913e-05, "loss": 0.19011720418930053, "step": 12950 }, { "epoch": 0.06437320564656329, "grad_norm": 4.601326942443848, "learning_rate": 1.9938117797351336e-05, "loss": 0.1831658124923706, "step": 12960 }, { "epoch": 0.06442287632993254, "grad_norm": 3.1355161666870117, "learning_rate": 1.9937938977325268e-05, "loss": 0.20010266304016114, "step": 12970 }, { "epoch": 0.06447254701330181, "grad_norm": 0.8103299140930176, "learning_rate": 1.993775990010933e-05, "loss": 0.20275635719299318, "step": 12980 }, { "epoch": 0.06452221769667107, "grad_norm": 1.3596128225326538, "learning_rate": 1.9937580565708168e-05, "loss": 0.17581582069396973, "step": 12990 }, { "epoch": 0.06457188838004033, "grad_norm": 1.4516867399215698, "learning_rate": 1.9937400974126416e-05, "loss": 0.18726931810379027, "step": 13000 }, { "epoch": 0.06462155906340959, "grad_norm": 1.4253952503204346, "learning_rate": 1.993722112536872e-05, "loss": 0.19053930044174194, "step": 13010 }, { "epoch": 0.06467122974677886, "grad_norm": 1.5627474784851074, "learning_rate": 1.9937041019439736e-05, "loss": 0.19208685159683228, "step": 13020 }, { "epoch": 0.06472090043014812, "grad_norm": 2.1072826385498047, "learning_rate": 1.993686065634413e-05, "loss": 0.1849515199661255, "step": 13030 }, { "epoch": 0.06477057111351738, "grad_norm": 1.1709280014038086, "learning_rate": 1.9936680036086562e-05, "loss": 0.18679075241088866, "step": 13040 }, { "epoch": 0.06482024179688664, "grad_norm": 1.6746759414672852, "learning_rate": 1.9936499158671713e-05, "loss": 0.194447922706604, "step": 13050 }, { "epoch": 0.0648699124802559, "grad_norm": 0.9550370573997498, "learning_rate": 1.9936318024104254e-05, "loss": 0.17267098426818847, "step": 13060 }, { "epoch": 0.06491958316362517, "grad_norm": 1.5117518901824951, "learning_rate": 1.993613663238889e-05, "loss": 0.19594351053237916, "step": 13070 }, { "epoch": 0.06496925384699442, "grad_norm": 1.7799700498580933, "learning_rate": 1.9935954983530295e-05, "loss": 0.17979393005371094, "step": 13080 }, { "epoch": 0.06501892453036369, "grad_norm": 1.8537554740905762, "learning_rate": 1.9935773077533187e-05, "loss": 0.18684072494506837, "step": 13090 }, { "epoch": 0.06506859521373295, "grad_norm": 1.7734979391098022, "learning_rate": 1.9935590914402264e-05, "loss": 0.18197849988937378, "step": 13100 }, { "epoch": 0.0651182658971022, "grad_norm": 3.3661460876464844, "learning_rate": 1.9935408494142244e-05, "loss": 0.18127028942108153, "step": 13110 }, { "epoch": 0.06516793658047147, "grad_norm": 2.5453553199768066, "learning_rate": 1.993522581675785e-05, "loss": 0.19951801300048827, "step": 13120 }, { "epoch": 0.06521760726384074, "grad_norm": 2.6221678256988525, "learning_rate": 1.9935042882253803e-05, "loss": 0.18018369674682616, "step": 13130 }, { "epoch": 0.06526727794721, "grad_norm": 3.1127448081970215, "learning_rate": 1.993485969063484e-05, "loss": 0.18518062829971313, "step": 13140 }, { "epoch": 0.06531694863057926, "grad_norm": 4.725893974304199, "learning_rate": 1.9934676241905707e-05, "loss": 0.19226233959197997, "step": 13150 }, { "epoch": 0.06536661931394852, "grad_norm": 2.159822940826416, "learning_rate": 1.9934492536071145e-05, "loss": 0.20082509517669678, "step": 13160 }, { "epoch": 0.06541628999731779, "grad_norm": 2.2730445861816406, "learning_rate": 1.993430857313591e-05, "loss": 0.18901736736297609, "step": 13170 }, { "epoch": 0.06546596068068704, "grad_norm": 3.5600788593292236, "learning_rate": 1.9934124353104764e-05, "loss": 0.1913991928100586, "step": 13180 }, { "epoch": 0.0655156313640563, "grad_norm": 2.9012598991394043, "learning_rate": 1.9933939875982472e-05, "loss": 0.18940240144729614, "step": 13190 }, { "epoch": 0.06556530204742557, "grad_norm": 0.8424009084701538, "learning_rate": 1.9933755141773812e-05, "loss": 0.18636469841003417, "step": 13200 }, { "epoch": 0.06561497273079483, "grad_norm": 2.5360162258148193, "learning_rate": 1.9933570150483565e-05, "loss": 0.18812384605407714, "step": 13210 }, { "epoch": 0.06566464341416409, "grad_norm": 3.3500802516937256, "learning_rate": 1.9933384902116514e-05, "loss": 0.1793673515319824, "step": 13220 }, { "epoch": 0.06571431409753335, "grad_norm": 0.9552027583122253, "learning_rate": 1.9933199396677457e-05, "loss": 0.19656813144683838, "step": 13230 }, { "epoch": 0.06576398478090262, "grad_norm": 3.810826301574707, "learning_rate": 1.9933013634171195e-05, "loss": 0.18138128519058228, "step": 13240 }, { "epoch": 0.06581365546427188, "grad_norm": 2.2809970378875732, "learning_rate": 1.9932827614602533e-05, "loss": 0.18860812187194825, "step": 13250 }, { "epoch": 0.06586332614764114, "grad_norm": 1.257433533668518, "learning_rate": 1.9932641337976285e-05, "loss": 0.18652677536010742, "step": 13260 }, { "epoch": 0.0659129968310104, "grad_norm": 1.0018810033798218, "learning_rate": 1.9932454804297276e-05, "loss": 0.1888627290725708, "step": 13270 }, { "epoch": 0.06596266751437967, "grad_norm": 2.6319360733032227, "learning_rate": 1.993226801357033e-05, "loss": 0.202469801902771, "step": 13280 }, { "epoch": 0.06601233819774892, "grad_norm": 2.0137109756469727, "learning_rate": 1.993208096580028e-05, "loss": 0.1959981679916382, "step": 13290 }, { "epoch": 0.06606200888111818, "grad_norm": 1.9217708110809326, "learning_rate": 1.9931893660991965e-05, "loss": 0.19477503299713134, "step": 13300 }, { "epoch": 0.06611167956448745, "grad_norm": 0.885433554649353, "learning_rate": 1.9931706099150235e-05, "loss": 0.16863222122192384, "step": 13310 }, { "epoch": 0.06616135024785672, "grad_norm": 1.6662060022354126, "learning_rate": 1.993151828027995e-05, "loss": 0.1746702551841736, "step": 13320 }, { "epoch": 0.06621102093122597, "grad_norm": 2.646873712539673, "learning_rate": 1.9931330204385962e-05, "loss": 0.19925718307495116, "step": 13330 }, { "epoch": 0.06626069161459523, "grad_norm": 4.4168314933776855, "learning_rate": 1.9931141871473143e-05, "loss": 0.17900443077087402, "step": 13340 }, { "epoch": 0.0663103622979645, "grad_norm": 1.056782603263855, "learning_rate": 1.9930953281546363e-05, "loss": 0.1867438793182373, "step": 13350 }, { "epoch": 0.06636003298133376, "grad_norm": 0.9754116535186768, "learning_rate": 1.993076443461051e-05, "loss": 0.18852940797805787, "step": 13360 }, { "epoch": 0.06640970366470302, "grad_norm": 2.1764895915985107, "learning_rate": 1.993057533067046e-05, "loss": 0.18732963800430297, "step": 13370 }, { "epoch": 0.06645937434807228, "grad_norm": 1.356523871421814, "learning_rate": 1.9930385969731118e-05, "loss": 0.20059328079223632, "step": 13380 }, { "epoch": 0.06650904503144155, "grad_norm": 3.9977498054504395, "learning_rate": 1.993019635179738e-05, "loss": 0.18206593990325928, "step": 13390 }, { "epoch": 0.0665587157148108, "grad_norm": 1.2055078744888306, "learning_rate": 1.9930006476874152e-05, "loss": 0.17696871757507324, "step": 13400 }, { "epoch": 0.06660838639818006, "grad_norm": 5.28648042678833, "learning_rate": 1.992981634496635e-05, "loss": 0.1906710147857666, "step": 13410 }, { "epoch": 0.06665805708154933, "grad_norm": 1.421715497970581, "learning_rate": 1.9929625956078893e-05, "loss": 0.2031856060028076, "step": 13420 }, { "epoch": 0.0667077277649186, "grad_norm": 3.375626564025879, "learning_rate": 1.9929435310216708e-05, "loss": 0.18122308254241942, "step": 13430 }, { "epoch": 0.06675739844828785, "grad_norm": 2.9042117595672607, "learning_rate": 1.992924440738473e-05, "loss": 0.1993652582168579, "step": 13440 }, { "epoch": 0.06680706913165711, "grad_norm": 0.9751995205879211, "learning_rate": 1.9929053247587897e-05, "loss": 0.18629698753356932, "step": 13450 }, { "epoch": 0.06685673981502638, "grad_norm": 1.2755382061004639, "learning_rate": 1.9928861830831163e-05, "loss": 0.19721686840057373, "step": 13460 }, { "epoch": 0.06690641049839564, "grad_norm": 1.7321959733963013, "learning_rate": 1.9928670157119474e-05, "loss": 0.2072441577911377, "step": 13470 }, { "epoch": 0.0669560811817649, "grad_norm": 1.2296913862228394, "learning_rate": 1.9928478226457796e-05, "loss": 0.17458829879760743, "step": 13480 }, { "epoch": 0.06700575186513416, "grad_norm": 3.1790168285369873, "learning_rate": 1.992828603885109e-05, "loss": 0.1920534372329712, "step": 13490 }, { "epoch": 0.06705542254850343, "grad_norm": 1.4508981704711914, "learning_rate": 1.9928093594304334e-05, "loss": 0.18822555541992186, "step": 13500 }, { "epoch": 0.06710509323187268, "grad_norm": 1.6423066854476929, "learning_rate": 1.9927900892822508e-05, "loss": 0.18808701038360595, "step": 13510 }, { "epoch": 0.06715476391524194, "grad_norm": 1.4030452966690063, "learning_rate": 1.9927707934410598e-05, "loss": 0.18925466537475585, "step": 13520 }, { "epoch": 0.06720443459861121, "grad_norm": 2.36429762840271, "learning_rate": 1.99275147190736e-05, "loss": 0.18859080076217652, "step": 13530 }, { "epoch": 0.06725410528198048, "grad_norm": 3.8015522956848145, "learning_rate": 1.9927321246816516e-05, "loss": 0.19595552682876588, "step": 13540 }, { "epoch": 0.06730377596534973, "grad_norm": 0.8076359629631042, "learning_rate": 1.9927127517644343e-05, "loss": 0.17982170581817628, "step": 13550 }, { "epoch": 0.06735344664871899, "grad_norm": 3.902090549468994, "learning_rate": 1.9926933531562107e-05, "loss": 0.1733607053756714, "step": 13560 }, { "epoch": 0.06740311733208826, "grad_norm": 4.143814563751221, "learning_rate": 1.9926739288574818e-05, "loss": 0.18203777074813843, "step": 13570 }, { "epoch": 0.06745278801545751, "grad_norm": 1.68954336643219, "learning_rate": 1.992654478868751e-05, "loss": 0.20276787281036376, "step": 13580 }, { "epoch": 0.06750245869882678, "grad_norm": 2.0649666786193848, "learning_rate": 1.9926350031905217e-05, "loss": 0.17707483768463134, "step": 13590 }, { "epoch": 0.06755212938219604, "grad_norm": 1.2656066417694092, "learning_rate": 1.992615501823297e-05, "loss": 0.17393157482147217, "step": 13600 }, { "epoch": 0.0676018000655653, "grad_norm": 1.6838281154632568, "learning_rate": 1.9925959747675827e-05, "loss": 0.2053096055984497, "step": 13610 }, { "epoch": 0.06765147074893456, "grad_norm": 1.299180269241333, "learning_rate": 1.9925764220238835e-05, "loss": 0.18235766887664795, "step": 13620 }, { "epoch": 0.06770114143230382, "grad_norm": 3.800302743911743, "learning_rate": 1.9925568435927055e-05, "loss": 0.19645581245422364, "step": 13630 }, { "epoch": 0.06775081211567309, "grad_norm": 2.672978401184082, "learning_rate": 1.992537239474556e-05, "loss": 0.17433935403823853, "step": 13640 }, { "epoch": 0.06780048279904236, "grad_norm": 1.4725679159164429, "learning_rate": 1.9925176096699414e-05, "loss": 0.18480850458145143, "step": 13650 }, { "epoch": 0.06785015348241161, "grad_norm": 0.9966779947280884, "learning_rate": 1.99249795417937e-05, "loss": 0.17928322553634643, "step": 13660 }, { "epoch": 0.06789982416578087, "grad_norm": 2.5115437507629395, "learning_rate": 1.992478273003351e-05, "loss": 0.18386589288711547, "step": 13670 }, { "epoch": 0.06794949484915014, "grad_norm": 0.910858154296875, "learning_rate": 1.992458566142393e-05, "loss": 0.2011904239654541, "step": 13680 }, { "epoch": 0.06799916553251939, "grad_norm": 2.2666540145874023, "learning_rate": 1.9924388335970068e-05, "loss": 0.18390367031097413, "step": 13690 }, { "epoch": 0.06804883621588866, "grad_norm": 4.016231536865234, "learning_rate": 1.992419075367702e-05, "loss": 0.1954594373703003, "step": 13700 }, { "epoch": 0.06809850689925792, "grad_norm": 1.7470957040786743, "learning_rate": 1.992399291454991e-05, "loss": 0.18811867237091065, "step": 13710 }, { "epoch": 0.06814817758262719, "grad_norm": 4.529282093048096, "learning_rate": 1.992379481859385e-05, "loss": 0.175227153301239, "step": 13720 }, { "epoch": 0.06819784826599644, "grad_norm": 1.8461692333221436, "learning_rate": 1.9923596465813974e-05, "loss": 0.18793058395385742, "step": 13730 }, { "epoch": 0.0682475189493657, "grad_norm": 2.592710494995117, "learning_rate": 1.9923397856215413e-05, "loss": 0.18740379810333252, "step": 13740 }, { "epoch": 0.06829718963273497, "grad_norm": 1.2972633838653564, "learning_rate": 1.9923198989803302e-05, "loss": 0.20622467994689941, "step": 13750 }, { "epoch": 0.06834686031610424, "grad_norm": 0.8721946477890015, "learning_rate": 1.9922999866582792e-05, "loss": 0.19391148090362548, "step": 13760 }, { "epoch": 0.06839653099947349, "grad_norm": 2.289083242416382, "learning_rate": 1.9922800486559034e-05, "loss": 0.20001087188720704, "step": 13770 }, { "epoch": 0.06844620168284275, "grad_norm": 2.1939616203308105, "learning_rate": 1.9922600849737193e-05, "loss": 0.17571030855178832, "step": 13780 }, { "epoch": 0.06849587236621202, "grad_norm": 3.14127779006958, "learning_rate": 1.992240095612243e-05, "loss": 0.18404510021209716, "step": 13790 }, { "epoch": 0.06854554304958127, "grad_norm": 1.5727224349975586, "learning_rate": 1.992220080571992e-05, "loss": 0.17691464424133302, "step": 13800 }, { "epoch": 0.06859521373295054, "grad_norm": 0.6278261542320251, "learning_rate": 1.992200039853484e-05, "loss": 0.19145004749298095, "step": 13810 }, { "epoch": 0.0686448844163198, "grad_norm": 0.8297731280326843, "learning_rate": 1.9921799734572383e-05, "loss": 0.1812778115272522, "step": 13820 }, { "epoch": 0.06869455509968907, "grad_norm": 1.4181832075119019, "learning_rate": 1.9921598813837733e-05, "loss": 0.18897600173950196, "step": 13830 }, { "epoch": 0.06874422578305832, "grad_norm": 1.5835415124893188, "learning_rate": 1.99213976363361e-05, "loss": 0.18671640157699584, "step": 13840 }, { "epoch": 0.06879389646642758, "grad_norm": 1.9518193006515503, "learning_rate": 1.9921196202072683e-05, "loss": 0.19159324169158937, "step": 13850 }, { "epoch": 0.06884356714979685, "grad_norm": 1.4662102460861206, "learning_rate": 1.99209945110527e-05, "loss": 0.18480989933013917, "step": 13860 }, { "epoch": 0.06889323783316612, "grad_norm": 2.9791388511657715, "learning_rate": 1.9920792563281368e-05, "loss": 0.18956735134124755, "step": 13870 }, { "epoch": 0.06894290851653537, "grad_norm": 1.172623872756958, "learning_rate": 1.9920590358763908e-05, "loss": 0.1865631341934204, "step": 13880 }, { "epoch": 0.06899257919990463, "grad_norm": 1.2064582109451294, "learning_rate": 1.992038789750556e-05, "loss": 0.18236145973205567, "step": 13890 }, { "epoch": 0.0690422498832739, "grad_norm": 3.1987667083740234, "learning_rate": 1.9920185179511566e-05, "loss": 0.19648606777191163, "step": 13900 }, { "epoch": 0.06909192056664315, "grad_norm": 2.5814449787139893, "learning_rate": 1.991998220478717e-05, "loss": 0.18787283897399903, "step": 13910 }, { "epoch": 0.06914159125001242, "grad_norm": 2.763178586959839, "learning_rate": 1.9919778973337616e-05, "loss": 0.19080042839050293, "step": 13920 }, { "epoch": 0.06919126193338168, "grad_norm": 1.8893474340438843, "learning_rate": 1.9919575485168176e-05, "loss": 0.19025740623474122, "step": 13930 }, { "epoch": 0.06924093261675095, "grad_norm": 1.2842098474502563, "learning_rate": 1.991937174028411e-05, "loss": 0.17388895750045777, "step": 13940 }, { "epoch": 0.0692906033001202, "grad_norm": 1.758651614189148, "learning_rate": 1.991916773869069e-05, "loss": 0.19198260307312012, "step": 13950 }, { "epoch": 0.06934027398348946, "grad_norm": 1.07144033908844, "learning_rate": 1.99189634803932e-05, "loss": 0.19602725505828858, "step": 13960 }, { "epoch": 0.06938994466685873, "grad_norm": 2.6804232597351074, "learning_rate": 1.9918758965396923e-05, "loss": 0.19859511852264405, "step": 13970 }, { "epoch": 0.06943961535022798, "grad_norm": 1.445979356765747, "learning_rate": 1.991855419370715e-05, "loss": 0.19579837322235108, "step": 13980 }, { "epoch": 0.06948928603359725, "grad_norm": 1.101590871810913, "learning_rate": 1.9918349165329186e-05, "loss": 0.19125726222991943, "step": 13990 }, { "epoch": 0.06953895671696651, "grad_norm": 1.3133987188339233, "learning_rate": 1.9918143880268327e-05, "loss": 0.1967920780181885, "step": 14000 }, { "epoch": 0.06958862740033578, "grad_norm": 1.5287173986434937, "learning_rate": 1.99179383385299e-05, "loss": 0.19707354307174682, "step": 14010 }, { "epoch": 0.06963829808370503, "grad_norm": 1.9226405620574951, "learning_rate": 1.9917732540119214e-05, "loss": 0.18707576990127564, "step": 14020 }, { "epoch": 0.0696879687670743, "grad_norm": 2.2983291149139404, "learning_rate": 1.9917526485041594e-05, "loss": 0.18748159408569337, "step": 14030 }, { "epoch": 0.06973763945044356, "grad_norm": 1.6805639266967773, "learning_rate": 1.991732017330238e-05, "loss": 0.18544961214065553, "step": 14040 }, { "epoch": 0.06978731013381283, "grad_norm": 1.0950853824615479, "learning_rate": 1.9917113604906907e-05, "loss": 0.1907804012298584, "step": 14050 }, { "epoch": 0.06983698081718208, "grad_norm": 3.835371255874634, "learning_rate": 1.991690677986052e-05, "loss": 0.19440077543258666, "step": 14060 }, { "epoch": 0.06988665150055134, "grad_norm": 2.347822427749634, "learning_rate": 1.9916699698168573e-05, "loss": 0.19644676446914672, "step": 14070 }, { "epoch": 0.06993632218392061, "grad_norm": 2.422818660736084, "learning_rate": 1.9916492359836424e-05, "loss": 0.1814340829849243, "step": 14080 }, { "epoch": 0.06998599286728986, "grad_norm": 2.3592495918273926, "learning_rate": 1.9916284764869444e-05, "loss": 0.18274400234222413, "step": 14090 }, { "epoch": 0.07003566355065913, "grad_norm": 0.9712285399436951, "learning_rate": 1.9916076913273e-05, "loss": 0.1837260127067566, "step": 14100 }, { "epoch": 0.07008533423402839, "grad_norm": 3.230379104614258, "learning_rate": 1.9915868805052472e-05, "loss": 0.1807236909866333, "step": 14110 }, { "epoch": 0.07013500491739766, "grad_norm": 1.0739943981170654, "learning_rate": 1.991566044021325e-05, "loss": 0.18948066234588623, "step": 14120 }, { "epoch": 0.07018467560076691, "grad_norm": 0.9765630960464478, "learning_rate": 1.9915451818760717e-05, "loss": 0.1778656244277954, "step": 14130 }, { "epoch": 0.07023434628413618, "grad_norm": 0.7649897336959839, "learning_rate": 1.9915242940700277e-05, "loss": 0.19276832342147826, "step": 14140 }, { "epoch": 0.07028401696750544, "grad_norm": 0.6448673605918884, "learning_rate": 1.991503380603734e-05, "loss": 0.18210015296936036, "step": 14150 }, { "epoch": 0.07033368765087471, "grad_norm": 1.8433374166488647, "learning_rate": 1.9914824414777313e-05, "loss": 0.19046812057495116, "step": 14160 }, { "epoch": 0.07038335833424396, "grad_norm": 2.1656711101531982, "learning_rate": 1.991461476692562e-05, "loss": 0.1911451816558838, "step": 14170 }, { "epoch": 0.07043302901761322, "grad_norm": 1.893424153327942, "learning_rate": 1.991440486248768e-05, "loss": 0.19109961986541749, "step": 14180 }, { "epoch": 0.07048269970098249, "grad_norm": 1.9133002758026123, "learning_rate": 1.991419470146893e-05, "loss": 0.18251742124557496, "step": 14190 }, { "epoch": 0.07053237038435174, "grad_norm": 1.7863258123397827, "learning_rate": 1.9913984283874804e-05, "loss": 0.18401122093200684, "step": 14200 }, { "epoch": 0.07058204106772101, "grad_norm": 2.138580560684204, "learning_rate": 1.9913773609710756e-05, "loss": 0.19376572370529174, "step": 14210 }, { "epoch": 0.07063171175109027, "grad_norm": 1.4574238061904907, "learning_rate": 1.9913562678982232e-05, "loss": 0.19450030326843262, "step": 14220 }, { "epoch": 0.07068138243445954, "grad_norm": 2.3346357345581055, "learning_rate": 1.991335149169469e-05, "loss": 0.16652426719665528, "step": 14230 }, { "epoch": 0.07073105311782879, "grad_norm": 1.9305787086486816, "learning_rate": 1.9913140047853598e-05, "loss": 0.1841769814491272, "step": 14240 }, { "epoch": 0.07078072380119806, "grad_norm": 1.7583990097045898, "learning_rate": 1.9912928347464428e-05, "loss": 0.17805899381637574, "step": 14250 }, { "epoch": 0.07083039448456732, "grad_norm": 2.8117122650146484, "learning_rate": 1.991271639053266e-05, "loss": 0.17941590547561645, "step": 14260 }, { "epoch": 0.07088006516793659, "grad_norm": 2.1584930419921875, "learning_rate": 1.9912504177063775e-05, "loss": 0.2092052221298218, "step": 14270 }, { "epoch": 0.07092973585130584, "grad_norm": 2.1332857608795166, "learning_rate": 1.9912291707063268e-05, "loss": 0.18781375885009766, "step": 14280 }, { "epoch": 0.0709794065346751, "grad_norm": 1.9138884544372559, "learning_rate": 1.9912078980536638e-05, "loss": 0.19516621828079223, "step": 14290 }, { "epoch": 0.07102907721804437, "grad_norm": 1.6165169477462769, "learning_rate": 1.991186599748939e-05, "loss": 0.18714874982833862, "step": 14300 }, { "epoch": 0.07107874790141362, "grad_norm": 2.175591230392456, "learning_rate": 1.9911652757927036e-05, "loss": 0.2007305860519409, "step": 14310 }, { "epoch": 0.07112841858478289, "grad_norm": 0.9564386010169983, "learning_rate": 1.991143926185509e-05, "loss": 0.18629223108291626, "step": 14320 }, { "epoch": 0.07117808926815215, "grad_norm": 1.005439043045044, "learning_rate": 1.9911225509279083e-05, "loss": 0.18222904205322266, "step": 14330 }, { "epoch": 0.07122775995152142, "grad_norm": 2.0008251667022705, "learning_rate": 1.9911011500204546e-05, "loss": 0.1974201202392578, "step": 14340 }, { "epoch": 0.07127743063489067, "grad_norm": 1.7552082538604736, "learning_rate": 1.9910797234637014e-05, "loss": 0.18463799953460694, "step": 14350 }, { "epoch": 0.07132710131825994, "grad_norm": 1.1456776857376099, "learning_rate": 1.9910582712582038e-05, "loss": 0.1801755905151367, "step": 14360 }, { "epoch": 0.0713767720016292, "grad_norm": 1.079565167427063, "learning_rate": 1.9910367934045167e-05, "loss": 0.1908494234085083, "step": 14370 }, { "epoch": 0.07142644268499845, "grad_norm": 1.6519339084625244, "learning_rate": 1.9910152899031954e-05, "loss": 0.1897681713104248, "step": 14380 }, { "epoch": 0.07147611336836772, "grad_norm": 2.070291519165039, "learning_rate": 1.990993760754797e-05, "loss": 0.18537230491638185, "step": 14390 }, { "epoch": 0.07152578405173698, "grad_norm": 2.765611410140991, "learning_rate": 1.9909722059598787e-05, "loss": 0.17961974143981935, "step": 14400 }, { "epoch": 0.07157545473510625, "grad_norm": 1.4152265787124634, "learning_rate": 1.9909506255189983e-05, "loss": 0.20209414958953859, "step": 14410 }, { "epoch": 0.0716251254184755, "grad_norm": 3.090080499649048, "learning_rate": 1.9909290194327138e-05, "loss": 0.20293183326721193, "step": 14420 }, { "epoch": 0.07167479610184477, "grad_norm": 0.7573899030685425, "learning_rate": 1.990907387701585e-05, "loss": 0.1892648458480835, "step": 14430 }, { "epoch": 0.07172446678521403, "grad_norm": 2.5052313804626465, "learning_rate": 1.9908857303261713e-05, "loss": 0.1862254858016968, "step": 14440 }, { "epoch": 0.0717741374685833, "grad_norm": 3.3568997383117676, "learning_rate": 1.9908640473070334e-05, "loss": 0.17943320274353028, "step": 14450 }, { "epoch": 0.07182380815195255, "grad_norm": 1.9363595247268677, "learning_rate": 1.990842338644732e-05, "loss": 0.18326830863952637, "step": 14460 }, { "epoch": 0.07187347883532182, "grad_norm": 2.0202393531799316, "learning_rate": 1.9908206043398295e-05, "loss": 0.19094529151916503, "step": 14470 }, { "epoch": 0.07192314951869108, "grad_norm": 2.159661293029785, "learning_rate": 1.9907988443928884e-05, "loss": 0.1825115442276001, "step": 14480 }, { "epoch": 0.07197282020206033, "grad_norm": 2.887866973876953, "learning_rate": 1.9907770588044716e-05, "loss": 0.20799486637115477, "step": 14490 }, { "epoch": 0.0720224908854296, "grad_norm": 0.806067168712616, "learning_rate": 1.9907552475751424e-05, "loss": 0.19190454483032227, "step": 14500 }, { "epoch": 0.07207216156879886, "grad_norm": 2.4008257389068604, "learning_rate": 1.9907334107054663e-05, "loss": 0.20917997360229493, "step": 14510 }, { "epoch": 0.07212183225216813, "grad_norm": 0.768426775932312, "learning_rate": 1.9907115481960076e-05, "loss": 0.1925137996673584, "step": 14520 }, { "epoch": 0.07217150293553738, "grad_norm": 2.4414639472961426, "learning_rate": 1.9906896600473325e-05, "loss": 0.19720526933670043, "step": 14530 }, { "epoch": 0.07222117361890665, "grad_norm": 1.21886146068573, "learning_rate": 1.990667746260007e-05, "loss": 0.17743299007415772, "step": 14540 }, { "epoch": 0.07227084430227591, "grad_norm": 1.3514058589935303, "learning_rate": 1.990645806834599e-05, "loss": 0.19218206405639648, "step": 14550 }, { "epoch": 0.07232051498564518, "grad_norm": 1.6678884029388428, "learning_rate": 1.9906238417716758e-05, "loss": 0.191861355304718, "step": 14560 }, { "epoch": 0.07237018566901443, "grad_norm": 1.2979952096939087, "learning_rate": 1.9906018510718056e-05, "loss": 0.19257442951202391, "step": 14570 }, { "epoch": 0.0724198563523837, "grad_norm": 1.1607797145843506, "learning_rate": 1.990579834735558e-05, "loss": 0.18460909128189087, "step": 14580 }, { "epoch": 0.07246952703575296, "grad_norm": 3.922914743423462, "learning_rate": 1.9905577927635025e-05, "loss": 0.1818099856376648, "step": 14590 }, { "epoch": 0.07251919771912221, "grad_norm": 1.8802053928375244, "learning_rate": 1.9905357251562097e-05, "loss": 0.19501549005508423, "step": 14600 }, { "epoch": 0.07256886840249148, "grad_norm": 3.3477911949157715, "learning_rate": 1.9905136319142504e-05, "loss": 0.18106937408447266, "step": 14610 }, { "epoch": 0.07261853908586074, "grad_norm": 0.9317293167114258, "learning_rate": 1.9904915130381966e-05, "loss": 0.1955704689025879, "step": 14620 }, { "epoch": 0.07266820976923001, "grad_norm": 1.1757832765579224, "learning_rate": 1.990469368528621e-05, "loss": 0.17678589820861818, "step": 14630 }, { "epoch": 0.07271788045259926, "grad_norm": 1.631170630455017, "learning_rate": 1.9904471983860962e-05, "loss": 0.18489494323730468, "step": 14640 }, { "epoch": 0.07276755113596853, "grad_norm": 3.0231359004974365, "learning_rate": 1.990425002611196e-05, "loss": 0.1852772831916809, "step": 14650 }, { "epoch": 0.0728172218193378, "grad_norm": 1.7666987180709839, "learning_rate": 1.9904027812044952e-05, "loss": 0.18059489727020264, "step": 14660 }, { "epoch": 0.07286689250270706, "grad_norm": 0.9649761915206909, "learning_rate": 1.9903805341665685e-05, "loss": 0.1864149570465088, "step": 14670 }, { "epoch": 0.07291656318607631, "grad_norm": 2.1300668716430664, "learning_rate": 1.990358261497992e-05, "loss": 0.186387038230896, "step": 14680 }, { "epoch": 0.07296623386944558, "grad_norm": 4.113840579986572, "learning_rate": 1.9903359631993418e-05, "loss": 0.2048922061920166, "step": 14690 }, { "epoch": 0.07301590455281484, "grad_norm": 0.5205636620521545, "learning_rate": 1.990313639271195e-05, "loss": 0.2032397747039795, "step": 14700 }, { "epoch": 0.0730655752361841, "grad_norm": 2.654641628265381, "learning_rate": 1.9902912897141294e-05, "loss": 0.1755729079246521, "step": 14710 }, { "epoch": 0.07311524591955336, "grad_norm": 1.281758427619934, "learning_rate": 1.9902689145287236e-05, "loss": 0.17743165493011476, "step": 14720 }, { "epoch": 0.07316491660292262, "grad_norm": 1.8848062753677368, "learning_rate": 1.990246513715556e-05, "loss": 0.18822593688964845, "step": 14730 }, { "epoch": 0.07321458728629189, "grad_norm": 1.9293937683105469, "learning_rate": 1.9902240872752073e-05, "loss": 0.18778345584869385, "step": 14740 }, { "epoch": 0.07326425796966114, "grad_norm": 2.634308099746704, "learning_rate": 1.990201635208257e-05, "loss": 0.19324085712432862, "step": 14750 }, { "epoch": 0.07331392865303041, "grad_norm": 1.2372832298278809, "learning_rate": 1.990179157515287e-05, "loss": 0.18176205158233644, "step": 14760 }, { "epoch": 0.07336359933639967, "grad_norm": 1.9027855396270752, "learning_rate": 1.990156654196878e-05, "loss": 0.1804281711578369, "step": 14770 }, { "epoch": 0.07341327001976893, "grad_norm": 1.1614606380462646, "learning_rate": 1.9901341252536132e-05, "loss": 0.19878162145614625, "step": 14780 }, { "epoch": 0.07346294070313819, "grad_norm": 1.6227315664291382, "learning_rate": 1.990111570686075e-05, "loss": 0.17981793880462646, "step": 14790 }, { "epoch": 0.07351261138650746, "grad_norm": 1.0434335470199585, "learning_rate": 1.990088990494848e-05, "loss": 0.19463348388671875, "step": 14800 }, { "epoch": 0.07356228206987672, "grad_norm": 3.0587029457092285, "learning_rate": 1.9900663846805156e-05, "loss": 0.18701555728912353, "step": 14810 }, { "epoch": 0.07361195275324597, "grad_norm": 0.9409241080284119, "learning_rate": 1.9900437532436635e-05, "loss": 0.1790233850479126, "step": 14820 }, { "epoch": 0.07366162343661524, "grad_norm": 1.128328561782837, "learning_rate": 1.9900210961848772e-05, "loss": 0.18082261085510254, "step": 14830 }, { "epoch": 0.0737112941199845, "grad_norm": 2.2466986179351807, "learning_rate": 1.989998413504743e-05, "loss": 0.19308803081512452, "step": 14840 }, { "epoch": 0.07376096480335377, "grad_norm": 1.0193777084350586, "learning_rate": 1.9899757052038477e-05, "loss": 0.18389549255371093, "step": 14850 }, { "epoch": 0.07381063548672302, "grad_norm": 1.440751552581787, "learning_rate": 1.9899529712827794e-05, "loss": 0.17239117622375488, "step": 14860 }, { "epoch": 0.07386030617009229, "grad_norm": 1.5818763971328735, "learning_rate": 1.989930211742126e-05, "loss": 0.18233163356781007, "step": 14870 }, { "epoch": 0.07390997685346155, "grad_norm": 3.748000383377075, "learning_rate": 1.9899074265824773e-05, "loss": 0.18414782285690307, "step": 14880 }, { "epoch": 0.0739596475368308, "grad_norm": 1.8887110948562622, "learning_rate": 1.9898846158044218e-05, "loss": 0.19784544706344603, "step": 14890 }, { "epoch": 0.07400931822020007, "grad_norm": 1.4882935285568237, "learning_rate": 1.9898617794085507e-05, "loss": 0.18032407760620117, "step": 14900 }, { "epoch": 0.07405898890356934, "grad_norm": 1.277713418006897, "learning_rate": 1.9898389173954546e-05, "loss": 0.17793147563934325, "step": 14910 }, { "epoch": 0.0741086595869386, "grad_norm": 1.7420002222061157, "learning_rate": 1.9898160297657253e-05, "loss": 0.1904569149017334, "step": 14920 }, { "epoch": 0.07415833027030785, "grad_norm": 2.7686142921447754, "learning_rate": 1.9897931165199556e-05, "loss": 0.20639798641204835, "step": 14930 }, { "epoch": 0.07420800095367712, "grad_norm": 1.4699971675872803, "learning_rate": 1.9897701776587376e-05, "loss": 0.19380710124969483, "step": 14940 }, { "epoch": 0.07425767163704639, "grad_norm": 1.2504957914352417, "learning_rate": 1.9897472131826655e-05, "loss": 0.17137205600738525, "step": 14950 }, { "epoch": 0.07430734232041565, "grad_norm": 0.9363964796066284, "learning_rate": 1.9897242230923334e-05, "loss": 0.20341670513153076, "step": 14960 }, { "epoch": 0.0743570130037849, "grad_norm": 3.41458797454834, "learning_rate": 1.9897012073883366e-05, "loss": 0.18297083377838136, "step": 14970 }, { "epoch": 0.07440668368715417, "grad_norm": 1.9635573625564575, "learning_rate": 1.9896781660712703e-05, "loss": 0.191111958026886, "step": 14980 }, { "epoch": 0.07445635437052343, "grad_norm": 2.7019972801208496, "learning_rate": 1.9896550991417308e-05, "loss": 0.1964726686477661, "step": 14990 }, { "epoch": 0.07450602505389269, "grad_norm": 1.930614709854126, "learning_rate": 1.9896320066003158e-05, "loss": 0.19335832595825195, "step": 15000 }, { "epoch": 0.07455569573726195, "grad_norm": 1.3863584995269775, "learning_rate": 1.989608888447622e-05, "loss": 0.19221282005310059, "step": 15010 }, { "epoch": 0.07460536642063122, "grad_norm": 1.9934608936309814, "learning_rate": 1.9895857446842483e-05, "loss": 0.18962500095367432, "step": 15020 }, { "epoch": 0.07465503710400048, "grad_norm": 1.0710999965667725, "learning_rate": 1.9895625753107932e-05, "loss": 0.19927819967269897, "step": 15030 }, { "epoch": 0.07470470778736973, "grad_norm": 2.4942777156829834, "learning_rate": 1.9895393803278566e-05, "loss": 0.1966660737991333, "step": 15040 }, { "epoch": 0.074754378470739, "grad_norm": 1.1804163455963135, "learning_rate": 1.9895161597360386e-05, "loss": 0.19138920307159424, "step": 15050 }, { "epoch": 0.07480404915410827, "grad_norm": 1.3114608526229858, "learning_rate": 1.9894929135359403e-05, "loss": 0.1866985321044922, "step": 15060 }, { "epoch": 0.07485371983747753, "grad_norm": 2.878434896469116, "learning_rate": 1.9894696417281634e-05, "loss": 0.18670456409454345, "step": 15070 }, { "epoch": 0.07490339052084678, "grad_norm": 1.5365194082260132, "learning_rate": 1.9894463443133102e-05, "loss": 0.19524154663085938, "step": 15080 }, { "epoch": 0.07495306120421605, "grad_norm": 2.1228394508361816, "learning_rate": 1.989423021291983e-05, "loss": 0.1900258779525757, "step": 15090 }, { "epoch": 0.07500273188758531, "grad_norm": 4.456286430358887, "learning_rate": 1.989399672664786e-05, "loss": 0.198387610912323, "step": 15100 }, { "epoch": 0.07505240257095457, "grad_norm": 3.628023862838745, "learning_rate": 1.9893762984323234e-05, "loss": 0.19553552865982055, "step": 15110 }, { "epoch": 0.07510207325432383, "grad_norm": 2.2937328815460205, "learning_rate": 1.9893528985952e-05, "loss": 0.1923731803894043, "step": 15120 }, { "epoch": 0.0751517439376931, "grad_norm": 3.1050517559051514, "learning_rate": 1.989329473154021e-05, "loss": 0.17807470560073851, "step": 15130 }, { "epoch": 0.07520141462106236, "grad_norm": 1.777683138847351, "learning_rate": 1.9893060221093937e-05, "loss": 0.18776981830596923, "step": 15140 }, { "epoch": 0.07525108530443161, "grad_norm": 0.9673828482627869, "learning_rate": 1.989282545461924e-05, "loss": 0.18261284828186036, "step": 15150 }, { "epoch": 0.07530075598780088, "grad_norm": 2.5867364406585693, "learning_rate": 1.9892590432122195e-05, "loss": 0.17741700410842895, "step": 15160 }, { "epoch": 0.07535042667117015, "grad_norm": 2.0553138256073, "learning_rate": 1.989235515360889e-05, "loss": 0.18111784458160402, "step": 15170 }, { "epoch": 0.0754000973545394, "grad_norm": 1.5107802152633667, "learning_rate": 1.989211961908541e-05, "loss": 0.20167121887207032, "step": 15180 }, { "epoch": 0.07544976803790866, "grad_norm": 2.087348222732544, "learning_rate": 1.9891883828557853e-05, "loss": 0.18763628005981445, "step": 15190 }, { "epoch": 0.07549943872127793, "grad_norm": 2.5923142433166504, "learning_rate": 1.9891647782032323e-05, "loss": 0.19933270215988158, "step": 15200 }, { "epoch": 0.0755491094046472, "grad_norm": 1.2898558378219604, "learning_rate": 1.989141147951492e-05, "loss": 0.18621160984039306, "step": 15210 }, { "epoch": 0.07559878008801645, "grad_norm": 0.8836215734481812, "learning_rate": 1.989117492101177e-05, "loss": 0.20940356254577636, "step": 15220 }, { "epoch": 0.07564845077138571, "grad_norm": 1.5464638471603394, "learning_rate": 1.9890938106528986e-05, "loss": 0.18481786251068116, "step": 15230 }, { "epoch": 0.07569812145475498, "grad_norm": 1.6441603899002075, "learning_rate": 1.9890701036072702e-05, "loss": 0.19101032018661498, "step": 15240 }, { "epoch": 0.07574779213812424, "grad_norm": 1.066701889038086, "learning_rate": 1.9890463709649052e-05, "loss": 0.17850910425186156, "step": 15250 }, { "epoch": 0.0757974628214935, "grad_norm": 1.6904391050338745, "learning_rate": 1.9890226127264182e-05, "loss": 0.1796804189682007, "step": 15260 }, { "epoch": 0.07584713350486276, "grad_norm": 1.2223893404006958, "learning_rate": 1.9889988288924234e-05, "loss": 0.17934961318969728, "step": 15270 }, { "epoch": 0.07589680418823203, "grad_norm": 0.9793955087661743, "learning_rate": 1.988975019463536e-05, "loss": 0.18608120679855347, "step": 15280 }, { "epoch": 0.07594647487160128, "grad_norm": 2.789933443069458, "learning_rate": 1.9889511844403733e-05, "loss": 0.2028334379196167, "step": 15290 }, { "epoch": 0.07599614555497054, "grad_norm": 1.3333269357681274, "learning_rate": 1.9889273238235517e-05, "loss": 0.20368390083312987, "step": 15300 }, { "epoch": 0.07604581623833981, "grad_norm": 2.52007794380188, "learning_rate": 1.9889034376136885e-05, "loss": 0.18883426189422609, "step": 15310 }, { "epoch": 0.07609548692170907, "grad_norm": 1.458486557006836, "learning_rate": 1.9888795258114017e-05, "loss": 0.177216100692749, "step": 15320 }, { "epoch": 0.07614515760507833, "grad_norm": 1.5596415996551514, "learning_rate": 1.9888555884173107e-05, "loss": 0.18019208908081055, "step": 15330 }, { "epoch": 0.07619482828844759, "grad_norm": 1.5551730394363403, "learning_rate": 1.9888316254320346e-05, "loss": 0.18106881380081177, "step": 15340 }, { "epoch": 0.07624449897181686, "grad_norm": 2.773003578186035, "learning_rate": 1.9888076368561933e-05, "loss": 0.16290316581726075, "step": 15350 }, { "epoch": 0.07629416965518612, "grad_norm": 1.949914574623108, "learning_rate": 1.9887836226904082e-05, "loss": 0.19127850532531737, "step": 15360 }, { "epoch": 0.07634384033855537, "grad_norm": 1.462386965751648, "learning_rate": 1.9887595829353007e-05, "loss": 0.18307560682296753, "step": 15370 }, { "epoch": 0.07639351102192464, "grad_norm": 1.9003816843032837, "learning_rate": 1.9887355175914925e-05, "loss": 0.18970208168029784, "step": 15380 }, { "epoch": 0.0764431817052939, "grad_norm": 0.8539119958877563, "learning_rate": 1.9887114266596064e-05, "loss": 0.17933449745178223, "step": 15390 }, { "epoch": 0.07649285238866316, "grad_norm": 1.9453743696212769, "learning_rate": 1.9886873101402665e-05, "loss": 0.18189718723297119, "step": 15400 }, { "epoch": 0.07654252307203242, "grad_norm": 0.9385855197906494, "learning_rate": 1.9886631680340968e-05, "loss": 0.18951058387756348, "step": 15410 }, { "epoch": 0.07659219375540169, "grad_norm": 1.0601192712783813, "learning_rate": 1.9886390003417215e-05, "loss": 0.1886405825614929, "step": 15420 }, { "epoch": 0.07664186443877095, "grad_norm": 2.212061643600464, "learning_rate": 1.9886148070637663e-05, "loss": 0.18737744092941283, "step": 15430 }, { "epoch": 0.0766915351221402, "grad_norm": 3.414231061935425, "learning_rate": 1.9885905882008573e-05, "loss": 0.19776198863983155, "step": 15440 }, { "epoch": 0.07674120580550947, "grad_norm": 1.3826007843017578, "learning_rate": 1.9885663437536218e-05, "loss": 0.18246755599975586, "step": 15450 }, { "epoch": 0.07679087648887874, "grad_norm": 1.6042609214782715, "learning_rate": 1.9885420737226865e-05, "loss": 0.19105851650238037, "step": 15460 }, { "epoch": 0.076840547172248, "grad_norm": 0.9024907350540161, "learning_rate": 1.98851777810868e-05, "loss": 0.19209680557250977, "step": 15470 }, { "epoch": 0.07689021785561725, "grad_norm": 3.547079086303711, "learning_rate": 1.9884934569122308e-05, "loss": 0.19204394817352294, "step": 15480 }, { "epoch": 0.07693988853898652, "grad_norm": 3.9185903072357178, "learning_rate": 1.988469110133968e-05, "loss": 0.1815481662750244, "step": 15490 }, { "epoch": 0.07698955922235579, "grad_norm": 1.8372896909713745, "learning_rate": 1.988444737774523e-05, "loss": 0.18048118352890014, "step": 15500 }, { "epoch": 0.07703922990572504, "grad_norm": 2.058715343475342, "learning_rate": 1.9884203398345246e-05, "loss": 0.18908361196517945, "step": 15510 }, { "epoch": 0.0770889005890943, "grad_norm": 1.1540462970733643, "learning_rate": 1.988395916314606e-05, "loss": 0.1964450478553772, "step": 15520 }, { "epoch": 0.07713857127246357, "grad_norm": 2.8943943977355957, "learning_rate": 1.988371467215398e-05, "loss": 0.19384195804595947, "step": 15530 }, { "epoch": 0.07718824195583283, "grad_norm": 2.227259874343872, "learning_rate": 1.9883469925375342e-05, "loss": 0.18043018579483033, "step": 15540 }, { "epoch": 0.07723791263920209, "grad_norm": 1.4769800901412964, "learning_rate": 1.9883224922816472e-05, "loss": 0.16241177320480346, "step": 15550 }, { "epoch": 0.07728758332257135, "grad_norm": 1.3731080293655396, "learning_rate": 1.988297966448372e-05, "loss": 0.19052164554595946, "step": 15560 }, { "epoch": 0.07733725400594062, "grad_norm": 1.1700022220611572, "learning_rate": 1.9882734150383424e-05, "loss": 0.19407005310058595, "step": 15570 }, { "epoch": 0.07738692468930987, "grad_norm": 2.0452311038970947, "learning_rate": 1.9882488380521944e-05, "loss": 0.1912243127822876, "step": 15580 }, { "epoch": 0.07743659537267913, "grad_norm": 2.309688091278076, "learning_rate": 1.988224235490564e-05, "loss": 0.1779821753501892, "step": 15590 }, { "epoch": 0.0774862660560484, "grad_norm": 3.9585211277008057, "learning_rate": 1.9881996073540874e-05, "loss": 0.1739181399345398, "step": 15600 }, { "epoch": 0.07753593673941767, "grad_norm": 1.2653050422668457, "learning_rate": 1.9881749536434023e-05, "loss": 0.1895080804824829, "step": 15610 }, { "epoch": 0.07758560742278692, "grad_norm": 0.9042441248893738, "learning_rate": 1.988150274359147e-05, "loss": 0.18482087850570678, "step": 15620 }, { "epoch": 0.07763527810615618, "grad_norm": 2.835636615753174, "learning_rate": 1.9881255695019596e-05, "loss": 0.1929654598236084, "step": 15630 }, { "epoch": 0.07768494878952545, "grad_norm": 0.7934901714324951, "learning_rate": 1.98810083907248e-05, "loss": 0.18053710460662842, "step": 15640 }, { "epoch": 0.07773461947289471, "grad_norm": 2.3808510303497314, "learning_rate": 1.9880760830713477e-05, "loss": 0.20655226707458496, "step": 15650 }, { "epoch": 0.07778429015626397, "grad_norm": 1.4632786512374878, "learning_rate": 1.988051301499204e-05, "loss": 0.17952826023101806, "step": 15660 }, { "epoch": 0.07783396083963323, "grad_norm": 3.3597869873046875, "learning_rate": 1.98802649435669e-05, "loss": 0.18191443681716918, "step": 15670 }, { "epoch": 0.0778836315230025, "grad_norm": 1.0658780336380005, "learning_rate": 1.988001661644447e-05, "loss": 0.1782119870185852, "step": 15680 }, { "epoch": 0.07793330220637175, "grad_norm": 1.9039359092712402, "learning_rate": 1.9879768033631187e-05, "loss": 0.181870698928833, "step": 15690 }, { "epoch": 0.07798297288974101, "grad_norm": 0.794613778591156, "learning_rate": 1.987951919513348e-05, "loss": 0.1908489227294922, "step": 15700 }, { "epoch": 0.07803264357311028, "grad_norm": 2.6545541286468506, "learning_rate": 1.9879270100957785e-05, "loss": 0.17968647480010985, "step": 15710 }, { "epoch": 0.07808231425647955, "grad_norm": 2.6754660606384277, "learning_rate": 1.9879020751110557e-05, "loss": 0.20441799163818358, "step": 15720 }, { "epoch": 0.0781319849398488, "grad_norm": 0.8536524176597595, "learning_rate": 1.9878771145598242e-05, "loss": 0.1905885696411133, "step": 15730 }, { "epoch": 0.07818165562321806, "grad_norm": 1.4985594749450684, "learning_rate": 1.98785212844273e-05, "loss": 0.18444948196411132, "step": 15740 }, { "epoch": 0.07823132630658733, "grad_norm": 2.8323659896850586, "learning_rate": 1.98782711676042e-05, "loss": 0.18317911624908448, "step": 15750 }, { "epoch": 0.0782809969899566, "grad_norm": 1.0966695547103882, "learning_rate": 1.9878020795135415e-05, "loss": 0.18007150888442994, "step": 15760 }, { "epoch": 0.07833066767332585, "grad_norm": 1.2528870105743408, "learning_rate": 1.9877770167027426e-05, "loss": 0.18367767333984375, "step": 15770 }, { "epoch": 0.07838033835669511, "grad_norm": 1.278324842453003, "learning_rate": 1.9877519283286713e-05, "loss": 0.17425124645233153, "step": 15780 }, { "epoch": 0.07843000904006438, "grad_norm": 2.37483549118042, "learning_rate": 1.9877268143919774e-05, "loss": 0.196613609790802, "step": 15790 }, { "epoch": 0.07847967972343363, "grad_norm": 2.2874419689178467, "learning_rate": 1.9877016748933108e-05, "loss": 0.19662129878997803, "step": 15800 }, { "epoch": 0.0785293504068029, "grad_norm": 1.802225947380066, "learning_rate": 1.9876765098333217e-05, "loss": 0.18245668411254884, "step": 15810 }, { "epoch": 0.07857902109017216, "grad_norm": 1.126485824584961, "learning_rate": 1.987651319212662e-05, "loss": 0.17886769771575928, "step": 15820 }, { "epoch": 0.07862869177354143, "grad_norm": 0.9199432134628296, "learning_rate": 1.987626103031983e-05, "loss": 0.16029558181762696, "step": 15830 }, { "epoch": 0.07867836245691068, "grad_norm": 2.9150350093841553, "learning_rate": 1.9876008612919376e-05, "loss": 0.19044046401977538, "step": 15840 }, { "epoch": 0.07872803314027994, "grad_norm": 0.9824492335319519, "learning_rate": 1.987575593993179e-05, "loss": 0.19056652784347533, "step": 15850 }, { "epoch": 0.07877770382364921, "grad_norm": 2.0828535556793213, "learning_rate": 1.9875503011363612e-05, "loss": 0.19589900970458984, "step": 15860 }, { "epoch": 0.07882737450701847, "grad_norm": 1.6572400331497192, "learning_rate": 1.9875249827221385e-05, "loss": 0.19816066026687623, "step": 15870 }, { "epoch": 0.07887704519038773, "grad_norm": 2.1295061111450195, "learning_rate": 1.9874996387511665e-05, "loss": 0.18103116750717163, "step": 15880 }, { "epoch": 0.07892671587375699, "grad_norm": 0.8833894729614258, "learning_rate": 1.9874742692241013e-05, "loss": 0.181893253326416, "step": 15890 }, { "epoch": 0.07897638655712626, "grad_norm": 2.8187453746795654, "learning_rate": 1.9874488741415984e-05, "loss": 0.18204604387283324, "step": 15900 }, { "epoch": 0.07902605724049551, "grad_norm": 1.5533338785171509, "learning_rate": 1.9874234535043157e-05, "loss": 0.17757985591888428, "step": 15910 }, { "epoch": 0.07907572792386477, "grad_norm": 2.145047187805176, "learning_rate": 1.9873980073129116e-05, "loss": 0.19095816612243652, "step": 15920 }, { "epoch": 0.07912539860723404, "grad_norm": 1.2922476530075073, "learning_rate": 1.9873725355680435e-05, "loss": 0.18250372409820556, "step": 15930 }, { "epoch": 0.0791750692906033, "grad_norm": 1.0079563856124878, "learning_rate": 1.9873470382703716e-05, "loss": 0.17403353452682496, "step": 15940 }, { "epoch": 0.07922473997397256, "grad_norm": 1.03610360622406, "learning_rate": 1.9873215154205552e-05, "loss": 0.18069887161254883, "step": 15950 }, { "epoch": 0.07927441065734182, "grad_norm": 2.3143653869628906, "learning_rate": 1.9872959670192548e-05, "loss": 0.17881251573562623, "step": 15960 }, { "epoch": 0.07932408134071109, "grad_norm": 2.1399996280670166, "learning_rate": 1.987270393067132e-05, "loss": 0.18059566020965576, "step": 15970 }, { "epoch": 0.07937375202408034, "grad_norm": 2.592663526535034, "learning_rate": 1.9872447935648482e-05, "loss": 0.19731857776641845, "step": 15980 }, { "epoch": 0.0794234227074496, "grad_norm": 2.233246088027954, "learning_rate": 1.9872191685130664e-05, "loss": 0.19616286754608153, "step": 15990 }, { "epoch": 0.07947309339081887, "grad_norm": 1.8473234176635742, "learning_rate": 1.987193517912449e-05, "loss": 0.19145607948303223, "step": 16000 }, { "epoch": 0.07952276407418814, "grad_norm": 2.055518627166748, "learning_rate": 1.9871678417636606e-05, "loss": 0.19355506896972657, "step": 16010 }, { "epoch": 0.07957243475755739, "grad_norm": 0.9023445844650269, "learning_rate": 1.9871421400673653e-05, "loss": 0.19902389049530028, "step": 16020 }, { "epoch": 0.07962210544092665, "grad_norm": 1.0863306522369385, "learning_rate": 1.9871164128242284e-05, "loss": 0.19814441204071045, "step": 16030 }, { "epoch": 0.07967177612429592, "grad_norm": 1.7981183528900146, "learning_rate": 1.9870906600349156e-05, "loss": 0.1972690463066101, "step": 16040 }, { "epoch": 0.07972144680766519, "grad_norm": 1.327301025390625, "learning_rate": 1.9870648817000934e-05, "loss": 0.1710134506225586, "step": 16050 }, { "epoch": 0.07977111749103444, "grad_norm": 1.7824559211730957, "learning_rate": 1.9870390778204288e-05, "loss": 0.1884487748146057, "step": 16060 }, { "epoch": 0.0798207881744037, "grad_norm": 1.6767020225524902, "learning_rate": 1.98701324839659e-05, "loss": 0.18823232650756835, "step": 16070 }, { "epoch": 0.07987045885777297, "grad_norm": 3.5594286918640137, "learning_rate": 1.986987393429245e-05, "loss": 0.18678972721099854, "step": 16080 }, { "epoch": 0.07992012954114222, "grad_norm": 3.138714551925659, "learning_rate": 1.9869615129190632e-05, "loss": 0.1916707396507263, "step": 16090 }, { "epoch": 0.07996980022451149, "grad_norm": 0.8349475860595703, "learning_rate": 1.986935606866714e-05, "loss": 0.20039613246917726, "step": 16100 }, { "epoch": 0.08001947090788075, "grad_norm": 1.3937724828720093, "learning_rate": 1.9869096752728688e-05, "loss": 0.17467427253723145, "step": 16110 }, { "epoch": 0.08006914159125002, "grad_norm": 1.3640767335891724, "learning_rate": 1.986883718138198e-05, "loss": 0.18616969585418702, "step": 16120 }, { "epoch": 0.08011881227461927, "grad_norm": 2.0638411045074463, "learning_rate": 1.9868577354633727e-05, "loss": 0.18680024147033691, "step": 16130 }, { "epoch": 0.08016848295798853, "grad_norm": 2.2514452934265137, "learning_rate": 1.986831727249066e-05, "loss": 0.1873570203781128, "step": 16140 }, { "epoch": 0.0802181536413578, "grad_norm": 2.5240020751953125, "learning_rate": 1.9868056934959514e-05, "loss": 0.1948472023010254, "step": 16150 }, { "epoch": 0.08026782432472707, "grad_norm": 1.523435354232788, "learning_rate": 1.986779634204702e-05, "loss": 0.179129695892334, "step": 16160 }, { "epoch": 0.08031749500809632, "grad_norm": 3.9299511909484863, "learning_rate": 1.9867535493759925e-05, "loss": 0.1807323694229126, "step": 16170 }, { "epoch": 0.08036716569146558, "grad_norm": 1.6709332466125488, "learning_rate": 1.986727439010498e-05, "loss": 0.16557481288909912, "step": 16180 }, { "epoch": 0.08041683637483485, "grad_norm": 2.3807311058044434, "learning_rate": 1.9867013031088942e-05, "loss": 0.1969142436981201, "step": 16190 }, { "epoch": 0.0804665070582041, "grad_norm": 1.4711202383041382, "learning_rate": 1.9866751416718574e-05, "loss": 0.18687175512313842, "step": 16200 }, { "epoch": 0.08051617774157337, "grad_norm": 1.0664535760879517, "learning_rate": 1.9866489547000643e-05, "loss": 0.18729720115661622, "step": 16210 }, { "epoch": 0.08056584842494263, "grad_norm": 3.5515265464782715, "learning_rate": 1.9866227421941934e-05, "loss": 0.1947142004966736, "step": 16220 }, { "epoch": 0.0806155191083119, "grad_norm": 0.7198452949523926, "learning_rate": 1.9865965041549226e-05, "loss": 0.18537235260009766, "step": 16230 }, { "epoch": 0.08066518979168115, "grad_norm": 4.877874851226807, "learning_rate": 1.9865702405829305e-05, "loss": 0.18670767545700073, "step": 16240 }, { "epoch": 0.08071486047505042, "grad_norm": 3.2744057178497314, "learning_rate": 1.9865439514788978e-05, "loss": 0.19200528860092164, "step": 16250 }, { "epoch": 0.08076453115841968, "grad_norm": 2.572171211242676, "learning_rate": 1.986517636843504e-05, "loss": 0.2097715377807617, "step": 16260 }, { "epoch": 0.08081420184178893, "grad_norm": 3.2553248405456543, "learning_rate": 1.9864912966774305e-05, "loss": 0.1749875783920288, "step": 16270 }, { "epoch": 0.0808638725251582, "grad_norm": 2.8652517795562744, "learning_rate": 1.986464930981359e-05, "loss": 0.18816912174224854, "step": 16280 }, { "epoch": 0.08091354320852746, "grad_norm": 2.1358699798583984, "learning_rate": 1.9864385397559716e-05, "loss": 0.19930628538131714, "step": 16290 }, { "epoch": 0.08096321389189673, "grad_norm": 1.039469599723816, "learning_rate": 1.9864121230019514e-05, "loss": 0.18321609497070312, "step": 16300 }, { "epoch": 0.08101288457526598, "grad_norm": 5.676634788513184, "learning_rate": 1.986385680719982e-05, "loss": 0.22061524391174317, "step": 16310 }, { "epoch": 0.08106255525863525, "grad_norm": 1.8209503889083862, "learning_rate": 1.986359212910748e-05, "loss": 0.1845106601715088, "step": 16320 }, { "epoch": 0.08111222594200451, "grad_norm": 1.5099867582321167, "learning_rate": 1.9863327195749342e-05, "loss": 0.18242278099060058, "step": 16330 }, { "epoch": 0.08116189662537378, "grad_norm": 1.3538789749145508, "learning_rate": 1.986306200713226e-05, "loss": 0.18204896450042723, "step": 16340 }, { "epoch": 0.08121156730874303, "grad_norm": 1.621077060699463, "learning_rate": 1.98627965632631e-05, "loss": 0.19356533288955688, "step": 16350 }, { "epoch": 0.0812612379921123, "grad_norm": 1.1784679889678955, "learning_rate": 1.9862530864148732e-05, "loss": 0.1862929344177246, "step": 16360 }, { "epoch": 0.08131090867548156, "grad_norm": 3.4592297077178955, "learning_rate": 1.9862264909796028e-05, "loss": 0.17951959371566772, "step": 16370 }, { "epoch": 0.08136057935885081, "grad_norm": 1.9779828786849976, "learning_rate": 1.9861998700211877e-05, "loss": 0.19352452754974364, "step": 16380 }, { "epoch": 0.08141025004222008, "grad_norm": 2.3044393062591553, "learning_rate": 1.9861732235403162e-05, "loss": 0.18419814109802246, "step": 16390 }, { "epoch": 0.08145992072558934, "grad_norm": 4.175529479980469, "learning_rate": 1.9861465515376784e-05, "loss": 0.1964043617248535, "step": 16400 }, { "epoch": 0.08150959140895861, "grad_norm": 2.065455198287964, "learning_rate": 1.9861198540139643e-05, "loss": 0.19261572360992432, "step": 16410 }, { "epoch": 0.08155926209232786, "grad_norm": 0.8536353707313538, "learning_rate": 1.9860931309698647e-05, "loss": 0.18483657836914064, "step": 16420 }, { "epoch": 0.08160893277569713, "grad_norm": 1.2228338718414307, "learning_rate": 1.9860663824060716e-05, "loss": 0.17523926496505737, "step": 16430 }, { "epoch": 0.08165860345906639, "grad_norm": 1.5354890823364258, "learning_rate": 1.9860396083232767e-05, "loss": 0.1945727586746216, "step": 16440 }, { "epoch": 0.08170827414243566, "grad_norm": 1.4985754489898682, "learning_rate": 1.9860128087221733e-05, "loss": 0.18396962881088258, "step": 16450 }, { "epoch": 0.08175794482580491, "grad_norm": 1.281240463256836, "learning_rate": 1.985985983603455e-05, "loss": 0.18860366344451904, "step": 16460 }, { "epoch": 0.08180761550917418, "grad_norm": 1.6614755392074585, "learning_rate": 1.985959132967816e-05, "loss": 0.1903761625289917, "step": 16470 }, { "epoch": 0.08185728619254344, "grad_norm": 1.5934151411056519, "learning_rate": 1.9859322568159508e-05, "loss": 0.18411684036254883, "step": 16480 }, { "epoch": 0.08190695687591269, "grad_norm": 3.0820086002349854, "learning_rate": 1.9859053551485553e-05, "loss": 0.18264999389648437, "step": 16490 }, { "epoch": 0.08195662755928196, "grad_norm": 2.5937254428863525, "learning_rate": 1.985878427966326e-05, "loss": 0.1797635316848755, "step": 16500 }, { "epoch": 0.08200629824265122, "grad_norm": 0.955431342124939, "learning_rate": 1.985851475269959e-05, "loss": 0.1937873125076294, "step": 16510 }, { "epoch": 0.08205596892602049, "grad_norm": 0.9011479616165161, "learning_rate": 1.9858244970601525e-05, "loss": 0.18954010009765626, "step": 16520 }, { "epoch": 0.08210563960938974, "grad_norm": 1.7992167472839355, "learning_rate": 1.985797493337604e-05, "loss": 0.18747715950012206, "step": 16530 }, { "epoch": 0.082155310292759, "grad_norm": 0.7556397318840027, "learning_rate": 1.985770464103013e-05, "loss": 0.194907283782959, "step": 16540 }, { "epoch": 0.08220498097612827, "grad_norm": 0.9578019380569458, "learning_rate": 1.9857434093570788e-05, "loss": 0.19748167991638182, "step": 16550 }, { "epoch": 0.08225465165949754, "grad_norm": 2.708660364151001, "learning_rate": 1.9857163291005012e-05, "loss": 0.19194111824035645, "step": 16560 }, { "epoch": 0.08230432234286679, "grad_norm": 1.2403182983398438, "learning_rate": 1.9856892233339814e-05, "loss": 0.17842557430267333, "step": 16570 }, { "epoch": 0.08235399302623606, "grad_norm": 1.4454847574234009, "learning_rate": 1.985662092058221e-05, "loss": 0.18191802501678467, "step": 16580 }, { "epoch": 0.08240366370960532, "grad_norm": 2.2608280181884766, "learning_rate": 1.9856349352739223e-05, "loss": 0.20415263175964354, "step": 16590 }, { "epoch": 0.08245333439297457, "grad_norm": 1.169264554977417, "learning_rate": 1.9856077529817873e-05, "loss": 0.2028017520904541, "step": 16600 }, { "epoch": 0.08250300507634384, "grad_norm": 2.466099739074707, "learning_rate": 1.98558054518252e-05, "loss": 0.19234131574630736, "step": 16610 }, { "epoch": 0.0825526757597131, "grad_norm": 2.0841896533966064, "learning_rate": 1.9855533118768246e-05, "loss": 0.17781647443771362, "step": 16620 }, { "epoch": 0.08260234644308237, "grad_norm": 1.8552354574203491, "learning_rate": 1.9855260530654056e-05, "loss": 0.21008586883544922, "step": 16630 }, { "epoch": 0.08265201712645162, "grad_norm": 1.015738844871521, "learning_rate": 1.9854987687489687e-05, "loss": 0.20153470039367677, "step": 16640 }, { "epoch": 0.08270168780982089, "grad_norm": 2.5800411701202393, "learning_rate": 1.98547145892822e-05, "loss": 0.1897251844406128, "step": 16650 }, { "epoch": 0.08275135849319015, "grad_norm": 0.5756033658981323, "learning_rate": 1.9854441236038663e-05, "loss": 0.18832690715789796, "step": 16660 }, { "epoch": 0.0828010291765594, "grad_norm": 1.7339658737182617, "learning_rate": 1.9854167627766146e-05, "loss": 0.19356920719146728, "step": 16670 }, { "epoch": 0.08285069985992867, "grad_norm": 1.3172472715377808, "learning_rate": 1.9853893764471735e-05, "loss": 0.1889823079109192, "step": 16680 }, { "epoch": 0.08290037054329794, "grad_norm": 1.012932300567627, "learning_rate": 1.9853619646162517e-05, "loss": 0.19094974994659425, "step": 16690 }, { "epoch": 0.0829500412266672, "grad_norm": 0.9426286220550537, "learning_rate": 1.985334527284558e-05, "loss": 0.209495210647583, "step": 16700 }, { "epoch": 0.08299971191003645, "grad_norm": 3.0086376667022705, "learning_rate": 1.9853070644528033e-05, "loss": 0.16969101428985595, "step": 16710 }, { "epoch": 0.08304938259340572, "grad_norm": 1.0354896783828735, "learning_rate": 1.985279576121698e-05, "loss": 0.1761239528656006, "step": 16720 }, { "epoch": 0.08309905327677498, "grad_norm": 3.367408275604248, "learning_rate": 1.985252062291953e-05, "loss": 0.18636019229888917, "step": 16730 }, { "epoch": 0.08314872396014425, "grad_norm": 1.9348340034484863, "learning_rate": 1.985224522964281e-05, "loss": 0.1900080919265747, "step": 16740 }, { "epoch": 0.0831983946435135, "grad_norm": 1.3027700185775757, "learning_rate": 1.9851969581393946e-05, "loss": 0.18026657104492189, "step": 16750 }, { "epoch": 0.08324806532688277, "grad_norm": 2.272655487060547, "learning_rate": 1.985169367818007e-05, "loss": 0.2082054615020752, "step": 16760 }, { "epoch": 0.08329773601025203, "grad_norm": 1.5870115756988525, "learning_rate": 1.9851417520008323e-05, "loss": 0.16784918308258057, "step": 16770 }, { "epoch": 0.08334740669362128, "grad_norm": 2.230679750442505, "learning_rate": 1.985114110688585e-05, "loss": 0.20352113246917725, "step": 16780 }, { "epoch": 0.08339707737699055, "grad_norm": 1.5069175958633423, "learning_rate": 1.985086443881981e-05, "loss": 0.17240667343139648, "step": 16790 }, { "epoch": 0.08344674806035982, "grad_norm": 1.3928520679473877, "learning_rate": 1.9850587515817353e-05, "loss": 0.19824886322021484, "step": 16800 }, { "epoch": 0.08349641874372908, "grad_norm": 1.143185019493103, "learning_rate": 1.9850310337885657e-05, "loss": 0.18322151899337769, "step": 16810 }, { "epoch": 0.08354608942709833, "grad_norm": 1.116145133972168, "learning_rate": 1.9850032905031887e-05, "loss": 0.1853644847869873, "step": 16820 }, { "epoch": 0.0835957601104676, "grad_norm": 1.6791704893112183, "learning_rate": 1.984975521726323e-05, "loss": 0.18162285089492797, "step": 16830 }, { "epoch": 0.08364543079383686, "grad_norm": 1.4114789962768555, "learning_rate": 1.9849477274586862e-05, "loss": 0.18935371637344361, "step": 16840 }, { "epoch": 0.08369510147720613, "grad_norm": 1.9551212787628174, "learning_rate": 1.984919907700999e-05, "loss": 0.18417582511901856, "step": 16850 }, { "epoch": 0.08374477216057538, "grad_norm": 1.5016852617263794, "learning_rate": 1.98489206245398e-05, "loss": 0.1824796676635742, "step": 16860 }, { "epoch": 0.08379444284394465, "grad_norm": 1.7784934043884277, "learning_rate": 1.984864191718351e-05, "loss": 0.17699041366577148, "step": 16870 }, { "epoch": 0.08384411352731391, "grad_norm": 1.301695704460144, "learning_rate": 1.9848362954948323e-05, "loss": 0.19044322967529298, "step": 16880 }, { "epoch": 0.08389378421068316, "grad_norm": 1.4464081525802612, "learning_rate": 1.984808373784146e-05, "loss": 0.18919991254806517, "step": 16890 }, { "epoch": 0.08394345489405243, "grad_norm": 1.2123790979385376, "learning_rate": 1.9847804265870155e-05, "loss": 0.17656445503234863, "step": 16900 }, { "epoch": 0.0839931255774217, "grad_norm": 1.9899603128433228, "learning_rate": 1.9847524539041635e-05, "loss": 0.19443095922470094, "step": 16910 }, { "epoch": 0.08404279626079096, "grad_norm": 1.4716112613677979, "learning_rate": 1.984724455736314e-05, "loss": 0.18131670951843262, "step": 16920 }, { "epoch": 0.08409246694416021, "grad_norm": 1.5083571672439575, "learning_rate": 1.9846964320841908e-05, "loss": 0.19176369905471802, "step": 16930 }, { "epoch": 0.08414213762752948, "grad_norm": 2.4099786281585693, "learning_rate": 1.9846683829485207e-05, "loss": 0.19101083278656006, "step": 16940 }, { "epoch": 0.08419180831089874, "grad_norm": 2.9526255130767822, "learning_rate": 1.984640308330028e-05, "loss": 0.19261339902877808, "step": 16950 }, { "epoch": 0.08424147899426801, "grad_norm": 2.0505664348602295, "learning_rate": 1.9846122082294407e-05, "loss": 0.18192012310028077, "step": 16960 }, { "epoch": 0.08429114967763726, "grad_norm": 1.6297962665557861, "learning_rate": 1.9845840826474853e-05, "loss": 0.1904342293739319, "step": 16970 }, { "epoch": 0.08434082036100653, "grad_norm": 1.1731430292129517, "learning_rate": 1.9845559315848896e-05, "loss": 0.20237331390380858, "step": 16980 }, { "epoch": 0.08439049104437579, "grad_norm": 0.7706189155578613, "learning_rate": 1.984527755042382e-05, "loss": 0.19446173906326295, "step": 16990 }, { "epoch": 0.08444016172774504, "grad_norm": 1.63225519657135, "learning_rate": 1.984499553020692e-05, "loss": 0.18195765018463134, "step": 17000 }, { "epoch": 0.08448983241111431, "grad_norm": 1.501947045326233, "learning_rate": 1.98447132552055e-05, "loss": 0.18612098693847656, "step": 17010 }, { "epoch": 0.08453950309448358, "grad_norm": 1.366693377494812, "learning_rate": 1.9844430725426854e-05, "loss": 0.19152088165283204, "step": 17020 }, { "epoch": 0.08458917377785284, "grad_norm": 1.4312058687210083, "learning_rate": 1.98441479408783e-05, "loss": 0.19810253381729126, "step": 17030 }, { "epoch": 0.08463884446122209, "grad_norm": 1.0917071104049683, "learning_rate": 1.9843864901567158e-05, "loss": 0.18874142169952393, "step": 17040 }, { "epoch": 0.08468851514459136, "grad_norm": 3.9053359031677246, "learning_rate": 1.9843581607500746e-05, "loss": 0.17970674037933348, "step": 17050 }, { "epoch": 0.08473818582796062, "grad_norm": 1.8148237466812134, "learning_rate": 1.98432980586864e-05, "loss": 0.1826530933380127, "step": 17060 }, { "epoch": 0.08478785651132988, "grad_norm": 0.8493493795394897, "learning_rate": 1.984301425513146e-05, "loss": 0.19696935415267944, "step": 17070 }, { "epoch": 0.08483752719469914, "grad_norm": 1.1563369035720825, "learning_rate": 1.9842730196843272e-05, "loss": 0.1870563268661499, "step": 17080 }, { "epoch": 0.08488719787806841, "grad_norm": 2.3204092979431152, "learning_rate": 1.984244588382918e-05, "loss": 0.18134623765945435, "step": 17090 }, { "epoch": 0.08493686856143767, "grad_norm": 1.8303477764129639, "learning_rate": 1.9842161316096546e-05, "loss": 0.19648828506469726, "step": 17100 }, { "epoch": 0.08498653924480692, "grad_norm": 0.8544474244117737, "learning_rate": 1.9841876493652738e-05, "loss": 0.1783301830291748, "step": 17110 }, { "epoch": 0.08503620992817619, "grad_norm": 2.089724540710449, "learning_rate": 1.984159141650512e-05, "loss": 0.18370389938354492, "step": 17120 }, { "epoch": 0.08508588061154546, "grad_norm": 2.503777503967285, "learning_rate": 1.9841306084661074e-05, "loss": 0.17581639289855958, "step": 17130 }, { "epoch": 0.08513555129491472, "grad_norm": 1.6904548406600952, "learning_rate": 1.9841020498127984e-05, "loss": 0.1886984348297119, "step": 17140 }, { "epoch": 0.08518522197828397, "grad_norm": 2.2832443714141846, "learning_rate": 1.9840734656913242e-05, "loss": 0.18576626777648925, "step": 17150 }, { "epoch": 0.08523489266165324, "grad_norm": 0.8588850498199463, "learning_rate": 1.984044856102424e-05, "loss": 0.202805233001709, "step": 17160 }, { "epoch": 0.0852845633450225, "grad_norm": 3.2647857666015625, "learning_rate": 1.9840162210468388e-05, "loss": 0.19346851110458374, "step": 17170 }, { "epoch": 0.08533423402839176, "grad_norm": 0.8257107734680176, "learning_rate": 1.9839875605253096e-05, "loss": 0.18376375436782838, "step": 17180 }, { "epoch": 0.08538390471176102, "grad_norm": 1.1438173055648804, "learning_rate": 1.9839588745385778e-05, "loss": 0.2027841329574585, "step": 17190 }, { "epoch": 0.08543357539513029, "grad_norm": 1.9983561038970947, "learning_rate": 1.983930163087386e-05, "loss": 0.19206299781799316, "step": 17200 }, { "epoch": 0.08548324607849955, "grad_norm": 1.0400742292404175, "learning_rate": 1.9839014261724774e-05, "loss": 0.18585350513458251, "step": 17210 }, { "epoch": 0.0855329167618688, "grad_norm": 1.2125412225723267, "learning_rate": 1.9838726637945952e-05, "loss": 0.18565863370895386, "step": 17220 }, { "epoch": 0.08558258744523807, "grad_norm": 0.6527097821235657, "learning_rate": 1.983843875954484e-05, "loss": 0.17626487016677855, "step": 17230 }, { "epoch": 0.08563225812860734, "grad_norm": 1.2139068841934204, "learning_rate": 1.983815062652889e-05, "loss": 0.18970649242401122, "step": 17240 }, { "epoch": 0.0856819288119766, "grad_norm": 1.1164665222167969, "learning_rate": 1.983786223890556e-05, "loss": 0.19685132503509523, "step": 17250 }, { "epoch": 0.08573159949534585, "grad_norm": 2.438662528991699, "learning_rate": 1.9837573596682308e-05, "loss": 0.19368069171905516, "step": 17260 }, { "epoch": 0.08578127017871512, "grad_norm": 0.8166346549987793, "learning_rate": 1.983728469986661e-05, "loss": 0.19534077644348144, "step": 17270 }, { "epoch": 0.08583094086208438, "grad_norm": 0.9355737566947937, "learning_rate": 1.9836995548465935e-05, "loss": 0.190015709400177, "step": 17280 }, { "epoch": 0.08588061154545364, "grad_norm": 3.428576707839966, "learning_rate": 1.9836706142487773e-05, "loss": 0.19650626182556152, "step": 17290 }, { "epoch": 0.0859302822288229, "grad_norm": 2.1343564987182617, "learning_rate": 1.9836416481939615e-05, "loss": 0.19192924499511718, "step": 17300 }, { "epoch": 0.08597995291219217, "grad_norm": 0.9177645444869995, "learning_rate": 1.983612656682895e-05, "loss": 0.1859757661819458, "step": 17310 }, { "epoch": 0.08602962359556143, "grad_norm": 3.440089464187622, "learning_rate": 1.983583639716328e-05, "loss": 0.18940310478210448, "step": 17320 }, { "epoch": 0.08607929427893068, "grad_norm": 1.5075346231460571, "learning_rate": 1.9835545972950124e-05, "loss": 0.1714792013168335, "step": 17330 }, { "epoch": 0.08612896496229995, "grad_norm": 2.5535731315612793, "learning_rate": 1.9835255294196992e-05, "loss": 0.19363278150558472, "step": 17340 }, { "epoch": 0.08617863564566922, "grad_norm": 2.10414719581604, "learning_rate": 1.9834964360911405e-05, "loss": 0.17967815399169923, "step": 17350 }, { "epoch": 0.08622830632903848, "grad_norm": 2.234698534011841, "learning_rate": 1.9834673173100898e-05, "loss": 0.1975816011428833, "step": 17360 }, { "epoch": 0.08627797701240773, "grad_norm": 1.3111653327941895, "learning_rate": 1.9834381730773002e-05, "loss": 0.18234225511550903, "step": 17370 }, { "epoch": 0.086327647695777, "grad_norm": 1.59418785572052, "learning_rate": 1.9834090033935263e-05, "loss": 0.17539273500442504, "step": 17380 }, { "epoch": 0.08637731837914626, "grad_norm": 1.368044376373291, "learning_rate": 1.9833798082595227e-05, "loss": 0.19156378507614136, "step": 17390 }, { "epoch": 0.08642698906251552, "grad_norm": 1.9844205379486084, "learning_rate": 1.983350587676045e-05, "loss": 0.18329402208328247, "step": 17400 }, { "epoch": 0.08647665974588478, "grad_norm": 1.4205445051193237, "learning_rate": 1.9833213416438494e-05, "loss": 0.19185178279876708, "step": 17410 }, { "epoch": 0.08652633042925405, "grad_norm": 0.9568859934806824, "learning_rate": 1.9832920701636932e-05, "loss": 0.19462950229644777, "step": 17420 }, { "epoch": 0.08657600111262331, "grad_norm": 1.2731821537017822, "learning_rate": 1.9832627732363332e-05, "loss": 0.18028607368469238, "step": 17430 }, { "epoch": 0.08662567179599256, "grad_norm": 2.198117256164551, "learning_rate": 1.9832334508625283e-05, "loss": 0.19784605503082275, "step": 17440 }, { "epoch": 0.08667534247936183, "grad_norm": 1.0116539001464844, "learning_rate": 1.9832041030430368e-05, "loss": 0.17895538806915284, "step": 17450 }, { "epoch": 0.0867250131627311, "grad_norm": 0.7380485534667969, "learning_rate": 1.9831747297786186e-05, "loss": 0.19199666976928711, "step": 17460 }, { "epoch": 0.08677468384610035, "grad_norm": 0.622434675693512, "learning_rate": 1.9831453310700334e-05, "loss": 0.19315088987350465, "step": 17470 }, { "epoch": 0.08682435452946961, "grad_norm": 2.541839122772217, "learning_rate": 1.983115906918043e-05, "loss": 0.19100337028503417, "step": 17480 }, { "epoch": 0.08687402521283888, "grad_norm": 1.9450818300247192, "learning_rate": 1.9830864573234075e-05, "loss": 0.18701536655426027, "step": 17490 }, { "epoch": 0.08692369589620814, "grad_norm": 3.7708451747894287, "learning_rate": 1.98305698228689e-05, "loss": 0.18810561895370484, "step": 17500 }, { "epoch": 0.0869733665795774, "grad_norm": 1.119361162185669, "learning_rate": 1.9830274818092533e-05, "loss": 0.19785404205322266, "step": 17510 }, { "epoch": 0.08702303726294666, "grad_norm": 1.4103567600250244, "learning_rate": 1.98299795589126e-05, "loss": 0.19577107429504395, "step": 17520 }, { "epoch": 0.08707270794631593, "grad_norm": 2.5042176246643066, "learning_rate": 1.9829684045336758e-05, "loss": 0.17620885372161865, "step": 17530 }, { "epoch": 0.0871223786296852, "grad_norm": 1.2911103963851929, "learning_rate": 1.982938827737264e-05, "loss": 0.1828914165496826, "step": 17540 }, { "epoch": 0.08717204931305444, "grad_norm": 0.958665668964386, "learning_rate": 1.9829092255027906e-05, "loss": 0.18248200416564941, "step": 17550 }, { "epoch": 0.08722171999642371, "grad_norm": 1.88141930103302, "learning_rate": 1.9828795978310215e-05, "loss": 0.19477081298828125, "step": 17560 }, { "epoch": 0.08727139067979298, "grad_norm": 1.8061318397521973, "learning_rate": 1.982849944722724e-05, "loss": 0.18910791873931884, "step": 17570 }, { "epoch": 0.08732106136316223, "grad_norm": 1.8581048250198364, "learning_rate": 1.9828202661786645e-05, "loss": 0.165451979637146, "step": 17580 }, { "epoch": 0.0873707320465315, "grad_norm": 2.5324325561523438, "learning_rate": 1.982790562199612e-05, "loss": 0.18832653760910034, "step": 17590 }, { "epoch": 0.08742040272990076, "grad_norm": 1.4143543243408203, "learning_rate": 1.9827608327863352e-05, "loss": 0.17484722137451172, "step": 17600 }, { "epoch": 0.08747007341327002, "grad_norm": 1.4369949102401733, "learning_rate": 1.982731077939603e-05, "loss": 0.18831379413604737, "step": 17610 }, { "epoch": 0.08751974409663928, "grad_norm": 1.340205192565918, "learning_rate": 1.9827012976601857e-05, "loss": 0.17862926721572875, "step": 17620 }, { "epoch": 0.08756941478000854, "grad_norm": 1.0042799711227417, "learning_rate": 1.9826714919488538e-05, "loss": 0.16621553897857666, "step": 17630 }, { "epoch": 0.08761908546337781, "grad_norm": 1.6296405792236328, "learning_rate": 1.9826416608063788e-05, "loss": 0.18695275783538817, "step": 17640 }, { "epoch": 0.08766875614674707, "grad_norm": 2.180309295654297, "learning_rate": 1.9826118042335326e-05, "loss": 0.16752777099609376, "step": 17650 }, { "epoch": 0.08771842683011633, "grad_norm": 1.067280650138855, "learning_rate": 1.9825819222310886e-05, "loss": 0.1854785442352295, "step": 17660 }, { "epoch": 0.08776809751348559, "grad_norm": 1.7899067401885986, "learning_rate": 1.982552014799819e-05, "loss": 0.17566679716110228, "step": 17670 }, { "epoch": 0.08781776819685486, "grad_norm": 1.6369632482528687, "learning_rate": 1.9825220819404985e-05, "loss": 0.2026191234588623, "step": 17680 }, { "epoch": 0.08786743888022411, "grad_norm": 1.407593846321106, "learning_rate": 1.9824921236539014e-05, "loss": 0.19429807662963866, "step": 17690 }, { "epoch": 0.08791710956359337, "grad_norm": 1.7249488830566406, "learning_rate": 1.9824621399408036e-05, "loss": 0.19635467529296874, "step": 17700 }, { "epoch": 0.08796678024696264, "grad_norm": 1.078071117401123, "learning_rate": 1.9824321308019804e-05, "loss": 0.1932634472846985, "step": 17710 }, { "epoch": 0.0880164509303319, "grad_norm": 1.9257292747497559, "learning_rate": 1.9824020962382084e-05, "loss": 0.18977800607681275, "step": 17720 }, { "epoch": 0.08806612161370116, "grad_norm": 1.3427071571350098, "learning_rate": 1.9823720362502653e-05, "loss": 0.1973031997680664, "step": 17730 }, { "epoch": 0.08811579229707042, "grad_norm": 0.9495793581008911, "learning_rate": 1.982341950838929e-05, "loss": 0.18239715099334716, "step": 17740 }, { "epoch": 0.08816546298043969, "grad_norm": 2.3207855224609375, "learning_rate": 1.9823118400049777e-05, "loss": 0.17575087547302246, "step": 17750 }, { "epoch": 0.08821513366380895, "grad_norm": 1.358628273010254, "learning_rate": 1.9822817037491914e-05, "loss": 0.18402599096298217, "step": 17760 }, { "epoch": 0.0882648043471782, "grad_norm": 1.0636749267578125, "learning_rate": 1.9822515420723497e-05, "loss": 0.1777154326438904, "step": 17770 }, { "epoch": 0.08831447503054747, "grad_norm": 1.364957571029663, "learning_rate": 1.9822213549752325e-05, "loss": 0.19103143215179444, "step": 17780 }, { "epoch": 0.08836414571391674, "grad_norm": 2.356905937194824, "learning_rate": 1.9821911424586216e-05, "loss": 0.18804486989974975, "step": 17790 }, { "epoch": 0.08841381639728599, "grad_norm": 1.1161432266235352, "learning_rate": 1.9821609045232994e-05, "loss": 0.18961172103881835, "step": 17800 }, { "epoch": 0.08846348708065525, "grad_norm": 1.7270790338516235, "learning_rate": 1.9821306411700475e-05, "loss": 0.18145058155059815, "step": 17810 }, { "epoch": 0.08851315776402452, "grad_norm": 1.2416013479232788, "learning_rate": 1.9821003523996496e-05, "loss": 0.18281195163726807, "step": 17820 }, { "epoch": 0.08856282844739378, "grad_norm": 1.7210253477096558, "learning_rate": 1.9820700382128894e-05, "loss": 0.18971676826477052, "step": 17830 }, { "epoch": 0.08861249913076304, "grad_norm": 1.6020097732543945, "learning_rate": 1.9820396986105518e-05, "loss": 0.1746999740600586, "step": 17840 }, { "epoch": 0.0886621698141323, "grad_norm": 0.8780081272125244, "learning_rate": 1.9820093335934213e-05, "loss": 0.1885896682739258, "step": 17850 }, { "epoch": 0.08871184049750157, "grad_norm": 0.6476821899414062, "learning_rate": 1.981978943162284e-05, "loss": 0.183785343170166, "step": 17860 }, { "epoch": 0.08876151118087082, "grad_norm": 2.8995954990386963, "learning_rate": 1.9819485273179267e-05, "loss": 0.18153505325317382, "step": 17870 }, { "epoch": 0.08881118186424009, "grad_norm": 2.6294071674346924, "learning_rate": 1.9819180860611363e-05, "loss": 0.18648321628570558, "step": 17880 }, { "epoch": 0.08886085254760935, "grad_norm": 3.0315895080566406, "learning_rate": 1.9818876193927006e-05, "loss": 0.20034208297729492, "step": 17890 }, { "epoch": 0.08891052323097862, "grad_norm": 2.3931970596313477, "learning_rate": 1.9818571273134083e-05, "loss": 0.19588128328323365, "step": 17900 }, { "epoch": 0.08896019391434787, "grad_norm": 1.4309629201889038, "learning_rate": 1.981826609824048e-05, "loss": 0.17971220016479492, "step": 17910 }, { "epoch": 0.08900986459771713, "grad_norm": 1.0960497856140137, "learning_rate": 1.98179606692541e-05, "loss": 0.18669577836990356, "step": 17920 }, { "epoch": 0.0890595352810864, "grad_norm": 1.381799340248108, "learning_rate": 1.9817654986182845e-05, "loss": 0.21031503677368163, "step": 17930 }, { "epoch": 0.08910920596445566, "grad_norm": 2.479253053665161, "learning_rate": 1.981734904903463e-05, "loss": 0.19135115146636963, "step": 17940 }, { "epoch": 0.08915887664782492, "grad_norm": 1.8015313148498535, "learning_rate": 1.9817042857817362e-05, "loss": 0.19417133331298828, "step": 17950 }, { "epoch": 0.08920854733119418, "grad_norm": 2.4427855014801025, "learning_rate": 1.981673641253898e-05, "loss": 0.17187917232513428, "step": 17960 }, { "epoch": 0.08925821801456345, "grad_norm": 1.824781060218811, "learning_rate": 1.98164297132074e-05, "loss": 0.19347509145736694, "step": 17970 }, { "epoch": 0.0893078886979327, "grad_norm": 1.1413757801055908, "learning_rate": 1.981612275983057e-05, "loss": 0.18402304649353027, "step": 17980 }, { "epoch": 0.08935755938130197, "grad_norm": 3.7868521213531494, "learning_rate": 1.9815815552416428e-05, "loss": 0.19676856994628905, "step": 17990 }, { "epoch": 0.08940723006467123, "grad_norm": 1.3764392137527466, "learning_rate": 1.9815508090972926e-05, "loss": 0.1850198030471802, "step": 18000 }, { "epoch": 0.0894569007480405, "grad_norm": 1.825618863105774, "learning_rate": 1.9815200375508025e-05, "loss": 0.1767406463623047, "step": 18010 }, { "epoch": 0.08950657143140975, "grad_norm": 1.4561611413955688, "learning_rate": 1.9814892406029682e-05, "loss": 0.19261536598205567, "step": 18020 }, { "epoch": 0.08955624211477901, "grad_norm": 3.2025389671325684, "learning_rate": 1.981458418254587e-05, "loss": 0.1923828363418579, "step": 18030 }, { "epoch": 0.08960591279814828, "grad_norm": 0.9234296679496765, "learning_rate": 1.9814275705064566e-05, "loss": 0.19253556728363036, "step": 18040 }, { "epoch": 0.08965558348151755, "grad_norm": 1.968117594718933, "learning_rate": 1.9813966973593755e-05, "loss": 0.18600175380706788, "step": 18050 }, { "epoch": 0.0897052541648868, "grad_norm": 1.6558189392089844, "learning_rate": 1.9813657988141418e-05, "loss": 0.18652673959732055, "step": 18060 }, { "epoch": 0.08975492484825606, "grad_norm": 0.8103030920028687, "learning_rate": 1.9813348748715567e-05, "loss": 0.19555429220199586, "step": 18070 }, { "epoch": 0.08980459553162533, "grad_norm": 2.1627252101898193, "learning_rate": 1.981303925532419e-05, "loss": 0.19613361358642578, "step": 18080 }, { "epoch": 0.08985426621499458, "grad_norm": 1.379837989807129, "learning_rate": 1.9812729507975305e-05, "loss": 0.19977283477783203, "step": 18090 }, { "epoch": 0.08990393689836385, "grad_norm": 3.0787715911865234, "learning_rate": 1.9812419506676924e-05, "loss": 0.18467967510223388, "step": 18100 }, { "epoch": 0.08995360758173311, "grad_norm": 2.498459815979004, "learning_rate": 1.9812109251437076e-05, "loss": 0.1948505997657776, "step": 18110 }, { "epoch": 0.09000327826510238, "grad_norm": 1.3860889673233032, "learning_rate": 1.981179874226378e-05, "loss": 0.19131466150283813, "step": 18120 }, { "epoch": 0.09005294894847163, "grad_norm": 1.3576852083206177, "learning_rate": 1.9811487979165082e-05, "loss": 0.1782729983329773, "step": 18130 }, { "epoch": 0.0901026196318409, "grad_norm": 1.4867138862609863, "learning_rate": 1.9811176962149016e-05, "loss": 0.18917417526245117, "step": 18140 }, { "epoch": 0.09015229031521016, "grad_norm": 1.6593135595321655, "learning_rate": 1.981086569122364e-05, "loss": 0.18309588432312013, "step": 18150 }, { "epoch": 0.09020196099857943, "grad_norm": 1.8647058010101318, "learning_rate": 1.9810554166397004e-05, "loss": 0.1874840021133423, "step": 18160 }, { "epoch": 0.09025163168194868, "grad_norm": 0.9646362662315369, "learning_rate": 1.981024238767717e-05, "loss": 0.19778017997741698, "step": 18170 }, { "epoch": 0.09030130236531794, "grad_norm": 1.4640140533447266, "learning_rate": 1.9809930355072204e-05, "loss": 0.1887320399284363, "step": 18180 }, { "epoch": 0.09035097304868721, "grad_norm": 1.8337754011154175, "learning_rate": 1.9809618068590187e-05, "loss": 0.18834383487701417, "step": 18190 }, { "epoch": 0.09040064373205646, "grad_norm": 2.09721040725708, "learning_rate": 1.98093055282392e-05, "loss": 0.1929265022277832, "step": 18200 }, { "epoch": 0.09045031441542573, "grad_norm": 1.4792778491973877, "learning_rate": 1.9808992734027332e-05, "loss": 0.18037885427474976, "step": 18210 }, { "epoch": 0.09049998509879499, "grad_norm": 1.0556236505508423, "learning_rate": 1.980867968596267e-05, "loss": 0.16871201992034912, "step": 18220 }, { "epoch": 0.09054965578216426, "grad_norm": 1.3317930698394775, "learning_rate": 1.980836638405333e-05, "loss": 0.19206526279449462, "step": 18230 }, { "epoch": 0.09059932646553351, "grad_norm": 1.2320984601974487, "learning_rate": 1.9808052828307403e-05, "loss": 0.18618180751800537, "step": 18240 }, { "epoch": 0.09064899714890277, "grad_norm": 2.1671717166900635, "learning_rate": 1.9807739018733018e-05, "loss": 0.20461368560791016, "step": 18250 }, { "epoch": 0.09069866783227204, "grad_norm": 2.4000959396362305, "learning_rate": 1.980742495533829e-05, "loss": 0.18188412189483644, "step": 18260 }, { "epoch": 0.09074833851564129, "grad_norm": 1.854724407196045, "learning_rate": 1.9807110638131345e-05, "loss": 0.18608739376068115, "step": 18270 }, { "epoch": 0.09079800919901056, "grad_norm": 1.571001410484314, "learning_rate": 1.980679606712032e-05, "loss": 0.1895371198654175, "step": 18280 }, { "epoch": 0.09084767988237982, "grad_norm": 1.5733325481414795, "learning_rate": 1.980648124231336e-05, "loss": 0.19117498397827148, "step": 18290 }, { "epoch": 0.09089735056574909, "grad_norm": 1.8337050676345825, "learning_rate": 1.9806166163718605e-05, "loss": 0.19505071640014648, "step": 18300 }, { "epoch": 0.09094702124911834, "grad_norm": 2.6597886085510254, "learning_rate": 1.9805850831344213e-05, "loss": 0.18733854293823243, "step": 18310 }, { "epoch": 0.0909966919324876, "grad_norm": 1.7883470058441162, "learning_rate": 1.9805535245198347e-05, "loss": 0.18612520694732665, "step": 18320 }, { "epoch": 0.09104636261585687, "grad_norm": 3.1793618202209473, "learning_rate": 1.9805219405289168e-05, "loss": 0.18588970899581908, "step": 18330 }, { "epoch": 0.09109603329922614, "grad_norm": 2.3894574642181396, "learning_rate": 1.9804903311624853e-05, "loss": 0.2028188705444336, "step": 18340 }, { "epoch": 0.09114570398259539, "grad_norm": 1.377704381942749, "learning_rate": 1.9804586964213584e-05, "loss": 0.19028995037078858, "step": 18350 }, { "epoch": 0.09119537466596465, "grad_norm": 2.510540008544922, "learning_rate": 1.980427036306355e-05, "loss": 0.17894382476806642, "step": 18360 }, { "epoch": 0.09124504534933392, "grad_norm": 1.9563764333724976, "learning_rate": 1.980395350818294e-05, "loss": 0.1957486629486084, "step": 18370 }, { "epoch": 0.09129471603270317, "grad_norm": 1.2175160646438599, "learning_rate": 1.980363639957995e-05, "loss": 0.19381377696990967, "step": 18380 }, { "epoch": 0.09134438671607244, "grad_norm": 2.659876823425293, "learning_rate": 1.9803319037262795e-05, "loss": 0.1882026433944702, "step": 18390 }, { "epoch": 0.0913940573994417, "grad_norm": 1.4424166679382324, "learning_rate": 1.9803001421239686e-05, "loss": 0.1697993040084839, "step": 18400 }, { "epoch": 0.09144372808281097, "grad_norm": 2.6233317852020264, "learning_rate": 1.9802683551518842e-05, "loss": 0.16503541469573973, "step": 18410 }, { "epoch": 0.09149339876618022, "grad_norm": 1.109499216079712, "learning_rate": 1.9802365428108486e-05, "loss": 0.1895276665687561, "step": 18420 }, { "epoch": 0.09154306944954949, "grad_norm": 0.9607769250869751, "learning_rate": 1.980204705101686e-05, "loss": 0.18477263450622558, "step": 18430 }, { "epoch": 0.09159274013291875, "grad_norm": 1.5097761154174805, "learning_rate": 1.9801728420252192e-05, "loss": 0.20451507568359376, "step": 18440 }, { "epoch": 0.09164241081628802, "grad_norm": 0.8380489349365234, "learning_rate": 1.9801409535822737e-05, "loss": 0.18320300579071044, "step": 18450 }, { "epoch": 0.09169208149965727, "grad_norm": 0.8002621531486511, "learning_rate": 1.9801090397736743e-05, "loss": 0.17129185199737548, "step": 18460 }, { "epoch": 0.09174175218302653, "grad_norm": 0.9636737704277039, "learning_rate": 1.9800771006002468e-05, "loss": 0.17925286293029785, "step": 18470 }, { "epoch": 0.0917914228663958, "grad_norm": 2.134417772293091, "learning_rate": 1.980045136062818e-05, "loss": 0.17133719921112062, "step": 18480 }, { "epoch": 0.09184109354976505, "grad_norm": 1.5464272499084473, "learning_rate": 1.9800131461622157e-05, "loss": 0.18537087440490724, "step": 18490 }, { "epoch": 0.09189076423313432, "grad_norm": 1.2726203203201294, "learning_rate": 1.9799811308992666e-05, "loss": 0.1792527675628662, "step": 18500 }, { "epoch": 0.09194043491650358, "grad_norm": 1.1405675411224365, "learning_rate": 1.9799490902748e-05, "loss": 0.1717861294746399, "step": 18510 }, { "epoch": 0.09199010559987285, "grad_norm": 2.459214210510254, "learning_rate": 1.9799170242896455e-05, "loss": 0.19008554220199586, "step": 18520 }, { "epoch": 0.0920397762832421, "grad_norm": 1.8859755992889404, "learning_rate": 1.979884932944632e-05, "loss": 0.19573774337768554, "step": 18530 }, { "epoch": 0.09208944696661137, "grad_norm": 1.9014164209365845, "learning_rate": 1.9798528162405906e-05, "loss": 0.1863764762878418, "step": 18540 }, { "epoch": 0.09213911764998063, "grad_norm": 2.343672752380371, "learning_rate": 1.9798206741783523e-05, "loss": 0.1781807065010071, "step": 18550 }, { "epoch": 0.0921887883333499, "grad_norm": 1.7178869247436523, "learning_rate": 1.9797885067587484e-05, "loss": 0.18618881702423096, "step": 18560 }, { "epoch": 0.09223845901671915, "grad_norm": 2.6065194606781006, "learning_rate": 1.9797563139826122e-05, "loss": 0.18904086351394653, "step": 18570 }, { "epoch": 0.09228812970008841, "grad_norm": 2.4204599857330322, "learning_rate": 1.979724095850777e-05, "loss": 0.18611665964126586, "step": 18580 }, { "epoch": 0.09233780038345768, "grad_norm": 1.962519645690918, "learning_rate": 1.9796918523640755e-05, "loss": 0.1876244902610779, "step": 18590 }, { "epoch": 0.09238747106682693, "grad_norm": 1.004422664642334, "learning_rate": 1.979659583523343e-05, "loss": 0.19134321212768554, "step": 18600 }, { "epoch": 0.0924371417501962, "grad_norm": 2.287411689758301, "learning_rate": 1.9796272893294145e-05, "loss": 0.18628374338150025, "step": 18610 }, { "epoch": 0.09248681243356546, "grad_norm": 1.801259994506836, "learning_rate": 1.9795949697831257e-05, "loss": 0.18974312543869018, "step": 18620 }, { "epoch": 0.09253648311693473, "grad_norm": 3.6421666145324707, "learning_rate": 1.9795626248853125e-05, "loss": 0.1804498553276062, "step": 18630 }, { "epoch": 0.09258615380030398, "grad_norm": 2.076324701309204, "learning_rate": 1.979530254636813e-05, "loss": 0.1857425332069397, "step": 18640 }, { "epoch": 0.09263582448367325, "grad_norm": 1.8642290830612183, "learning_rate": 1.979497859038464e-05, "loss": 0.19108171463012696, "step": 18650 }, { "epoch": 0.09268549516704251, "grad_norm": 1.2727351188659668, "learning_rate": 1.9794654380911045e-05, "loss": 0.1787927746772766, "step": 18660 }, { "epoch": 0.09273516585041176, "grad_norm": 2.583841562271118, "learning_rate": 1.979432991795573e-05, "loss": 0.18453336954116822, "step": 18670 }, { "epoch": 0.09278483653378103, "grad_norm": 1.4869447946548462, "learning_rate": 1.9794005201527096e-05, "loss": 0.18301503658294677, "step": 18680 }, { "epoch": 0.0928345072171503, "grad_norm": 1.502569317817688, "learning_rate": 1.979368023163355e-05, "loss": 0.17772167921066284, "step": 18690 }, { "epoch": 0.09288417790051956, "grad_norm": 3.3094077110290527, "learning_rate": 1.979335500828349e-05, "loss": 0.1866968870162964, "step": 18700 }, { "epoch": 0.09293384858388881, "grad_norm": 3.8661949634552, "learning_rate": 1.9793029531485345e-05, "loss": 0.1977898120880127, "step": 18710 }, { "epoch": 0.09298351926725808, "grad_norm": 1.306839942932129, "learning_rate": 1.979270380124753e-05, "loss": 0.16854522228240967, "step": 18720 }, { "epoch": 0.09303318995062734, "grad_norm": 3.5718817710876465, "learning_rate": 1.9792377817578482e-05, "loss": 0.1916907548904419, "step": 18730 }, { "epoch": 0.09308286063399661, "grad_norm": 0.6208097338676453, "learning_rate": 1.979205158048663e-05, "loss": 0.19000016450881957, "step": 18740 }, { "epoch": 0.09313253131736586, "grad_norm": 1.4071255922317505, "learning_rate": 1.9791725089980422e-05, "loss": 0.1896461009979248, "step": 18750 }, { "epoch": 0.09318220200073513, "grad_norm": 2.9089109897613525, "learning_rate": 1.9791398346068304e-05, "loss": 0.20540232658386232, "step": 18760 }, { "epoch": 0.09323187268410439, "grad_norm": 1.7569185495376587, "learning_rate": 1.9791071348758736e-05, "loss": 0.18523125648498534, "step": 18770 }, { "epoch": 0.09328154336747364, "grad_norm": 1.2213188409805298, "learning_rate": 1.9790744098060178e-05, "loss": 0.19389853477478028, "step": 18780 }, { "epoch": 0.09333121405084291, "grad_norm": 1.3544822931289673, "learning_rate": 1.9790416593981095e-05, "loss": 0.19907305240631104, "step": 18790 }, { "epoch": 0.09338088473421217, "grad_norm": 0.991907000541687, "learning_rate": 1.9790088836529975e-05, "loss": 0.1931004762649536, "step": 18800 }, { "epoch": 0.09343055541758144, "grad_norm": 1.4078477621078491, "learning_rate": 1.978976082571529e-05, "loss": 0.19160289764404298, "step": 18810 }, { "epoch": 0.09348022610095069, "grad_norm": 1.059682011604309, "learning_rate": 1.9789432561545525e-05, "loss": 0.1864662766456604, "step": 18820 }, { "epoch": 0.09352989678431996, "grad_norm": 2.020815372467041, "learning_rate": 1.9789104044029187e-05, "loss": 0.19066139459609985, "step": 18830 }, { "epoch": 0.09357956746768922, "grad_norm": 1.3727291822433472, "learning_rate": 1.978877527317477e-05, "loss": 0.17024117708206177, "step": 18840 }, { "epoch": 0.09362923815105849, "grad_norm": 2.230408191680908, "learning_rate": 1.978844624899079e-05, "loss": 0.20249247550964355, "step": 18850 }, { "epoch": 0.09367890883442774, "grad_norm": 2.420214891433716, "learning_rate": 1.978811697148575e-05, "loss": 0.17750751972198486, "step": 18860 }, { "epoch": 0.093728579517797, "grad_norm": 3.1555323600769043, "learning_rate": 1.978778744066818e-05, "loss": 0.19626102447509766, "step": 18870 }, { "epoch": 0.09377825020116627, "grad_norm": 0.7277570366859436, "learning_rate": 1.978745765654661e-05, "loss": 0.19422913789749147, "step": 18880 }, { "epoch": 0.09382792088453552, "grad_norm": 2.3732755184173584, "learning_rate": 1.978712761912957e-05, "loss": 0.17833590507507324, "step": 18890 }, { "epoch": 0.09387759156790479, "grad_norm": 1.735433578491211, "learning_rate": 1.97867973284256e-05, "loss": 0.1849447965621948, "step": 18900 }, { "epoch": 0.09392726225127405, "grad_norm": 2.0134782791137695, "learning_rate": 1.978646678444325e-05, "loss": 0.1859198570251465, "step": 18910 }, { "epoch": 0.09397693293464332, "grad_norm": 1.6935995817184448, "learning_rate": 1.978613598719108e-05, "loss": 0.1782681465148926, "step": 18920 }, { "epoch": 0.09402660361801257, "grad_norm": 2.0036003589630127, "learning_rate": 1.9785804936677643e-05, "loss": 0.19263455867767335, "step": 18930 }, { "epoch": 0.09407627430138184, "grad_norm": 0.8577083945274353, "learning_rate": 1.9785473632911507e-05, "loss": 0.17916882038116455, "step": 18940 }, { "epoch": 0.0941259449847511, "grad_norm": 2.0036962032318115, "learning_rate": 1.978514207590125e-05, "loss": 0.19383001327514648, "step": 18950 }, { "epoch": 0.09417561566812037, "grad_norm": 1.7930432558059692, "learning_rate": 1.9784810265655446e-05, "loss": 0.17691335678100586, "step": 18960 }, { "epoch": 0.09422528635148962, "grad_norm": 1.7219318151474, "learning_rate": 1.9784478202182692e-05, "loss": 0.18928194046020508, "step": 18970 }, { "epoch": 0.09427495703485889, "grad_norm": 1.3888752460479736, "learning_rate": 1.9784145885491576e-05, "loss": 0.187248694896698, "step": 18980 }, { "epoch": 0.09432462771822815, "grad_norm": 1.0839478969573975, "learning_rate": 1.9783813315590694e-05, "loss": 0.1851428508758545, "step": 18990 }, { "epoch": 0.0943742984015974, "grad_norm": 1.3137403726577759, "learning_rate": 1.978348049248866e-05, "loss": 0.20286965370178223, "step": 19000 }, { "epoch": 0.09442396908496667, "grad_norm": 1.8519103527069092, "learning_rate": 1.9783147416194086e-05, "loss": 0.16546566486358644, "step": 19010 }, { "epoch": 0.09447363976833593, "grad_norm": 1.249864935874939, "learning_rate": 1.978281408671559e-05, "loss": 0.20179197788238526, "step": 19020 }, { "epoch": 0.0945233104517052, "grad_norm": 2.286134958267212, "learning_rate": 1.9782480504061796e-05, "loss": 0.19154083728790283, "step": 19030 }, { "epoch": 0.09457298113507445, "grad_norm": 3.453500747680664, "learning_rate": 1.9782146668241342e-05, "loss": 0.19584180116653443, "step": 19040 }, { "epoch": 0.09462265181844372, "grad_norm": 1.980674147605896, "learning_rate": 1.9781812579262868e-05, "loss": 0.1897769570350647, "step": 19050 }, { "epoch": 0.09467232250181298, "grad_norm": 1.0245825052261353, "learning_rate": 1.9781478237135015e-05, "loss": 0.1880699872970581, "step": 19060 }, { "epoch": 0.09472199318518223, "grad_norm": 2.1092796325683594, "learning_rate": 1.978114364186644e-05, "loss": 0.18769688606262208, "step": 19070 }, { "epoch": 0.0947716638685515, "grad_norm": 2.139331817626953, "learning_rate": 1.9780808793465795e-05, "loss": 0.19350805282592773, "step": 19080 }, { "epoch": 0.09482133455192077, "grad_norm": 1.4447884559631348, "learning_rate": 1.9780473691941754e-05, "loss": 0.2021420955657959, "step": 19090 }, { "epoch": 0.09487100523529003, "grad_norm": 1.5098649263381958, "learning_rate": 1.978013833730299e-05, "loss": 0.19822995662689208, "step": 19100 }, { "epoch": 0.09492067591865928, "grad_norm": 1.2802132368087769, "learning_rate": 1.9779802729558174e-05, "loss": 0.17190784215927124, "step": 19110 }, { "epoch": 0.09497034660202855, "grad_norm": 3.2099287509918213, "learning_rate": 1.9779466868716e-05, "loss": 0.18019359111785888, "step": 19120 }, { "epoch": 0.09502001728539781, "grad_norm": 1.2463291883468628, "learning_rate": 1.977913075478515e-05, "loss": 0.19305791854858398, "step": 19130 }, { "epoch": 0.09506968796876708, "grad_norm": 1.0167536735534668, "learning_rate": 1.9778794387774334e-05, "loss": 0.20427911281585692, "step": 19140 }, { "epoch": 0.09511935865213633, "grad_norm": 1.1933777332305908, "learning_rate": 1.9778457767692245e-05, "loss": 0.18074101209640503, "step": 19150 }, { "epoch": 0.0951690293355056, "grad_norm": 2.628676652908325, "learning_rate": 1.9778120894547604e-05, "loss": 0.17875845432281495, "step": 19160 }, { "epoch": 0.09521870001887486, "grad_norm": 1.2789443731307983, "learning_rate": 1.9777783768349125e-05, "loss": 0.18466627597808838, "step": 19170 }, { "epoch": 0.09526837070224412, "grad_norm": 3.1872925758361816, "learning_rate": 1.9777446389105534e-05, "loss": 0.18436932563781738, "step": 19180 }, { "epoch": 0.09531804138561338, "grad_norm": 1.1945226192474365, "learning_rate": 1.9777108756825564e-05, "loss": 0.1840929627418518, "step": 19190 }, { "epoch": 0.09536771206898265, "grad_norm": 1.8323835134506226, "learning_rate": 1.9776770871517947e-05, "loss": 0.17921969890594483, "step": 19200 }, { "epoch": 0.09541738275235191, "grad_norm": 1.997169852256775, "learning_rate": 1.9776432733191435e-05, "loss": 0.19882025718688964, "step": 19210 }, { "epoch": 0.09546705343572116, "grad_norm": 2.62046480178833, "learning_rate": 1.977609434185477e-05, "loss": 0.20502448081970215, "step": 19220 }, { "epoch": 0.09551672411909043, "grad_norm": 0.8369928002357483, "learning_rate": 1.9775755697516716e-05, "loss": 0.1869359016418457, "step": 19230 }, { "epoch": 0.0955663948024597, "grad_norm": 1.817154884338379, "learning_rate": 1.977541680018604e-05, "loss": 0.1950300931930542, "step": 19240 }, { "epoch": 0.09561606548582896, "grad_norm": 0.6529495120048523, "learning_rate": 1.97750776498715e-05, "loss": 0.18741700649261475, "step": 19250 }, { "epoch": 0.09566573616919821, "grad_norm": 0.9125163555145264, "learning_rate": 1.9774738246581885e-05, "loss": 0.19726202487945557, "step": 19260 }, { "epoch": 0.09571540685256748, "grad_norm": 2.421940326690674, "learning_rate": 1.9774398590325975e-05, "loss": 0.1762165307998657, "step": 19270 }, { "epoch": 0.09576507753593674, "grad_norm": 1.0262486934661865, "learning_rate": 1.977405868111256e-05, "loss": 0.1781141996383667, "step": 19280 }, { "epoch": 0.095814748219306, "grad_norm": 0.5661802291870117, "learning_rate": 1.977371851895043e-05, "loss": 0.18841265439987182, "step": 19290 }, { "epoch": 0.09586441890267526, "grad_norm": 0.886685311794281, "learning_rate": 1.97733781038484e-05, "loss": 0.18341219425201416, "step": 19300 }, { "epoch": 0.09591408958604453, "grad_norm": 3.2188024520874023, "learning_rate": 1.9773037435815277e-05, "loss": 0.19093801975250244, "step": 19310 }, { "epoch": 0.09596376026941379, "grad_norm": 2.4672293663024902, "learning_rate": 1.9772696514859867e-05, "loss": 0.18160494565963745, "step": 19320 }, { "epoch": 0.09601343095278304, "grad_norm": 1.28509521484375, "learning_rate": 1.9772355340991007e-05, "loss": 0.19103221893310546, "step": 19330 }, { "epoch": 0.09606310163615231, "grad_norm": 1.6095067262649536, "learning_rate": 1.977201391421752e-05, "loss": 0.20090038776397706, "step": 19340 }, { "epoch": 0.09611277231952157, "grad_norm": 0.9185123443603516, "learning_rate": 1.9771672234548237e-05, "loss": 0.1907506227493286, "step": 19350 }, { "epoch": 0.09616244300289084, "grad_norm": 0.9315595626831055, "learning_rate": 1.977133030199201e-05, "loss": 0.17118711471557618, "step": 19360 }, { "epoch": 0.09621211368626009, "grad_norm": 3.328037977218628, "learning_rate": 1.9770988116557685e-05, "loss": 0.1947665810585022, "step": 19370 }, { "epoch": 0.09626178436962936, "grad_norm": 1.7281484603881836, "learning_rate": 1.977064567825411e-05, "loss": 0.18733204603195192, "step": 19380 }, { "epoch": 0.09631145505299862, "grad_norm": 1.161749243736267, "learning_rate": 1.977030298709016e-05, "loss": 0.1868559718132019, "step": 19390 }, { "epoch": 0.09636112573636788, "grad_norm": 1.6022812128067017, "learning_rate": 1.9769960043074694e-05, "loss": 0.1919901967048645, "step": 19400 }, { "epoch": 0.09641079641973714, "grad_norm": 2.642862319946289, "learning_rate": 1.976961684621659e-05, "loss": 0.18634743690490724, "step": 19410 }, { "epoch": 0.0964604671031064, "grad_norm": 2.1353840827941895, "learning_rate": 1.9769273396524732e-05, "loss": 0.19481492042541504, "step": 19420 }, { "epoch": 0.09651013778647567, "grad_norm": 1.2330269813537598, "learning_rate": 1.9768929694008005e-05, "loss": 0.17383575439453125, "step": 19430 }, { "epoch": 0.09655980846984492, "grad_norm": 0.7590880393981934, "learning_rate": 1.9768585738675307e-05, "loss": 0.1951375961303711, "step": 19440 }, { "epoch": 0.09660947915321419, "grad_norm": 1.5231029987335205, "learning_rate": 1.9768241530535536e-05, "loss": 0.19464125633239746, "step": 19450 }, { "epoch": 0.09665914983658346, "grad_norm": 1.0689072608947754, "learning_rate": 1.97678970695976e-05, "loss": 0.182060170173645, "step": 19460 }, { "epoch": 0.0967088205199527, "grad_norm": 2.9544100761413574, "learning_rate": 1.976755235587042e-05, "loss": 0.19441756010055541, "step": 19470 }, { "epoch": 0.09675849120332197, "grad_norm": 1.5036715269088745, "learning_rate": 1.976720738936291e-05, "loss": 0.18306634426116944, "step": 19480 }, { "epoch": 0.09680816188669124, "grad_norm": 1.6512783765792847, "learning_rate": 1.9766862170084004e-05, "loss": 0.19373733997344972, "step": 19490 }, { "epoch": 0.0968578325700605, "grad_norm": 1.8846774101257324, "learning_rate": 1.9766516698042625e-05, "loss": 0.19354803562164308, "step": 19500 }, { "epoch": 0.09690750325342976, "grad_norm": 1.370634913444519, "learning_rate": 1.9766170973247726e-05, "loss": 0.17907366752624512, "step": 19510 }, { "epoch": 0.09695717393679902, "grad_norm": 0.9261490106582642, "learning_rate": 1.9765824995708245e-05, "loss": 0.18747806549072266, "step": 19520 }, { "epoch": 0.09700684462016829, "grad_norm": 0.8346371054649353, "learning_rate": 1.9765478765433143e-05, "loss": 0.18256220817565919, "step": 19530 }, { "epoch": 0.09705651530353755, "grad_norm": 0.7687445282936096, "learning_rate": 1.9765132282431375e-05, "loss": 0.19002060890197753, "step": 19540 }, { "epoch": 0.0971061859869068, "grad_norm": 2.1587040424346924, "learning_rate": 1.9764785546711913e-05, "loss": 0.17909088134765624, "step": 19550 }, { "epoch": 0.09715585667027607, "grad_norm": 2.01690411567688, "learning_rate": 1.9764438558283726e-05, "loss": 0.19274888038635254, "step": 19560 }, { "epoch": 0.09720552735364534, "grad_norm": 1.8388800621032715, "learning_rate": 1.9764091317155795e-05, "loss": 0.20564939975738525, "step": 19570 }, { "epoch": 0.09725519803701459, "grad_norm": 1.6041693687438965, "learning_rate": 1.9763743823337105e-05, "loss": 0.1964427947998047, "step": 19580 }, { "epoch": 0.09730486872038385, "grad_norm": 1.320385456085205, "learning_rate": 1.9763396076836653e-05, "loss": 0.18158133029937745, "step": 19590 }, { "epoch": 0.09735453940375312, "grad_norm": 0.9162732362747192, "learning_rate": 1.9763048077663436e-05, "loss": 0.18282458782196045, "step": 19600 }, { "epoch": 0.09740421008712238, "grad_norm": 3.264610528945923, "learning_rate": 1.9762699825826458e-05, "loss": 0.181368088722229, "step": 19610 }, { "epoch": 0.09745388077049164, "grad_norm": 1.428941249847412, "learning_rate": 1.9762351321334737e-05, "loss": 0.184128475189209, "step": 19620 }, { "epoch": 0.0975035514538609, "grad_norm": 1.7932461500167847, "learning_rate": 1.9762002564197286e-05, "loss": 0.18608660697937013, "step": 19630 }, { "epoch": 0.09755322213723017, "grad_norm": 1.6730927228927612, "learning_rate": 1.9761653554423134e-05, "loss": 0.19474256038665771, "step": 19640 }, { "epoch": 0.09760289282059943, "grad_norm": 2.759842872619629, "learning_rate": 1.9761304292021318e-05, "loss": 0.19432138204574584, "step": 19650 }, { "epoch": 0.09765256350396868, "grad_norm": 1.7881044149398804, "learning_rate": 1.9760954777000866e-05, "loss": 0.20099382400512694, "step": 19660 }, { "epoch": 0.09770223418733795, "grad_norm": 1.181161880493164, "learning_rate": 1.9760605009370833e-05, "loss": 0.1832308292388916, "step": 19670 }, { "epoch": 0.09775190487070722, "grad_norm": 1.3742858171463013, "learning_rate": 1.9760254989140267e-05, "loss": 0.1878190517425537, "step": 19680 }, { "epoch": 0.09780157555407647, "grad_norm": 1.817973017692566, "learning_rate": 1.975990471631822e-05, "loss": 0.18514457941055298, "step": 19690 }, { "epoch": 0.09785124623744573, "grad_norm": 1.6977227926254272, "learning_rate": 1.975955419091377e-05, "loss": 0.1885904312133789, "step": 19700 }, { "epoch": 0.097900916920815, "grad_norm": 1.978018045425415, "learning_rate": 1.975920341293598e-05, "loss": 0.18450710773468018, "step": 19710 }, { "epoch": 0.09795058760418426, "grad_norm": 1.3691567182540894, "learning_rate": 1.9758852382393925e-05, "loss": 0.1829635739326477, "step": 19720 }, { "epoch": 0.09800025828755352, "grad_norm": 1.5753045082092285, "learning_rate": 1.9758501099296703e-05, "loss": 0.1870119333267212, "step": 19730 }, { "epoch": 0.09804992897092278, "grad_norm": 1.1398471593856812, "learning_rate": 1.975814956365339e-05, "loss": 0.1889537453651428, "step": 19740 }, { "epoch": 0.09809959965429205, "grad_norm": 2.562751293182373, "learning_rate": 1.975779777547309e-05, "loss": 0.19437285661697387, "step": 19750 }, { "epoch": 0.09814927033766131, "grad_norm": 1.937172532081604, "learning_rate": 1.975744573476491e-05, "loss": 0.18053886890411378, "step": 19760 }, { "epoch": 0.09819894102103056, "grad_norm": 1.773615837097168, "learning_rate": 1.975709344153796e-05, "loss": 0.20527911186218262, "step": 19770 }, { "epoch": 0.09824861170439983, "grad_norm": 2.0791711807250977, "learning_rate": 1.9756740895801346e-05, "loss": 0.20064306259155273, "step": 19780 }, { "epoch": 0.0982982823877691, "grad_norm": 2.7964396476745605, "learning_rate": 1.9756388097564206e-05, "loss": 0.1911684274673462, "step": 19790 }, { "epoch": 0.09834795307113835, "grad_norm": 1.8063387870788574, "learning_rate": 1.9756035046835668e-05, "loss": 0.1997925043106079, "step": 19800 }, { "epoch": 0.09839762375450761, "grad_norm": 2.7212376594543457, "learning_rate": 1.975568174362486e-05, "loss": 0.18304164409637452, "step": 19810 }, { "epoch": 0.09844729443787688, "grad_norm": 2.981203556060791, "learning_rate": 1.9755328187940934e-05, "loss": 0.1775810718536377, "step": 19820 }, { "epoch": 0.09849696512124614, "grad_norm": 0.7568292617797852, "learning_rate": 1.9754974379793038e-05, "loss": 0.19280383586883545, "step": 19830 }, { "epoch": 0.0985466358046154, "grad_norm": 1.4829875230789185, "learning_rate": 1.9754620319190326e-05, "loss": 0.18561828136444092, "step": 19840 }, { "epoch": 0.09859630648798466, "grad_norm": 1.1473517417907715, "learning_rate": 1.975426600614196e-05, "loss": 0.19673483371734618, "step": 19850 }, { "epoch": 0.09864597717135393, "grad_norm": 2.06392240524292, "learning_rate": 1.9753911440657118e-05, "loss": 0.1997312068939209, "step": 19860 }, { "epoch": 0.09869564785472318, "grad_norm": 2.163102388381958, "learning_rate": 1.9753556622744962e-05, "loss": 0.17977960109710694, "step": 19870 }, { "epoch": 0.09874531853809244, "grad_norm": 1.1687792539596558, "learning_rate": 1.9753201552414687e-05, "loss": 0.17740468978881835, "step": 19880 }, { "epoch": 0.09879498922146171, "grad_norm": 1.7211090326309204, "learning_rate": 1.9752846229675478e-05, "loss": 0.19222251176834107, "step": 19890 }, { "epoch": 0.09884465990483098, "grad_norm": 2.276435613632202, "learning_rate": 1.9752490654536528e-05, "loss": 0.18391003608703613, "step": 19900 }, { "epoch": 0.09889433058820023, "grad_norm": 1.7131953239440918, "learning_rate": 1.9752134827007042e-05, "loss": 0.18235260248184204, "step": 19910 }, { "epoch": 0.09894400127156949, "grad_norm": 1.081735372543335, "learning_rate": 1.9751778747096225e-05, "loss": 0.18552024364471437, "step": 19920 }, { "epoch": 0.09899367195493876, "grad_norm": 0.9773257970809937, "learning_rate": 1.97514224148133e-05, "loss": 0.16040712594985962, "step": 19930 }, { "epoch": 0.09904334263830802, "grad_norm": 0.661835789680481, "learning_rate": 1.9751065830167483e-05, "loss": 0.1980456829071045, "step": 19940 }, { "epoch": 0.09909301332167728, "grad_norm": 1.2824970483779907, "learning_rate": 1.9750708993168e-05, "loss": 0.1894351601600647, "step": 19950 }, { "epoch": 0.09914268400504654, "grad_norm": 1.626285433769226, "learning_rate": 1.975035190382409e-05, "loss": 0.2071549654006958, "step": 19960 }, { "epoch": 0.09919235468841581, "grad_norm": 2.4592480659484863, "learning_rate": 1.9749994562144992e-05, "loss": 0.18745050430297852, "step": 19970 }, { "epoch": 0.09924202537178506, "grad_norm": 3.3824567794799805, "learning_rate": 1.974963696813996e-05, "loss": 0.17915582656860352, "step": 19980 }, { "epoch": 0.09929169605515432, "grad_norm": 0.9695211052894592, "learning_rate": 1.9749279121818235e-05, "loss": 0.18287763595581055, "step": 19990 }, { "epoch": 0.09934136673852359, "grad_norm": 1.1946814060211182, "learning_rate": 1.9748921023189093e-05, "loss": 0.19654926061630248, "step": 20000 }, { "epoch": 0.09939103742189286, "grad_norm": 1.7435483932495117, "learning_rate": 1.9748562672261795e-05, "loss": 0.19160702228546142, "step": 20010 }, { "epoch": 0.09944070810526211, "grad_norm": 0.9113145470619202, "learning_rate": 1.9748204069045612e-05, "loss": 0.18778797388076782, "step": 20020 }, { "epoch": 0.09949037878863137, "grad_norm": 2.278079032897949, "learning_rate": 1.974784521354983e-05, "loss": 0.20234227180480957, "step": 20030 }, { "epoch": 0.09954004947200064, "grad_norm": 2.3759074211120605, "learning_rate": 1.9747486105783728e-05, "loss": 0.18130961656570435, "step": 20040 }, { "epoch": 0.0995897201553699, "grad_norm": 1.0586721897125244, "learning_rate": 1.974712674575661e-05, "loss": 0.20164496898651124, "step": 20050 }, { "epoch": 0.09963939083873916, "grad_norm": 3.6668758392333984, "learning_rate": 1.974676713347777e-05, "loss": 0.19700897932052613, "step": 20060 }, { "epoch": 0.09968906152210842, "grad_norm": 2.0854780673980713, "learning_rate": 1.9746407268956512e-05, "loss": 0.20137784481048585, "step": 20070 }, { "epoch": 0.09973873220547769, "grad_norm": 2.88450288772583, "learning_rate": 1.974604715220216e-05, "loss": 0.1905268430709839, "step": 20080 }, { "epoch": 0.09978840288884694, "grad_norm": 1.373984694480896, "learning_rate": 1.9745686783224022e-05, "loss": 0.18593084812164307, "step": 20090 }, { "epoch": 0.0998380735722162, "grad_norm": 1.9199843406677246, "learning_rate": 1.974532616203143e-05, "loss": 0.18778090476989745, "step": 20100 }, { "epoch": 0.09988774425558547, "grad_norm": 1.3786289691925049, "learning_rate": 1.9744965288633713e-05, "loss": 0.19184765815734864, "step": 20110 }, { "epoch": 0.09993741493895474, "grad_norm": 2.1933138370513916, "learning_rate": 1.9744604163040214e-05, "loss": 0.19377022981643677, "step": 20120 }, { "epoch": 0.09998708562232399, "grad_norm": 1.4372614622116089, "learning_rate": 1.9744242785260276e-05, "loss": 0.18441848754882811, "step": 20130 }, { "epoch": 0.10003675630569325, "grad_norm": 0.9868457913398743, "learning_rate": 1.9743881155303254e-05, "loss": 0.1827051043510437, "step": 20140 }, { "epoch": 0.10008642698906252, "grad_norm": 1.3400893211364746, "learning_rate": 1.9743519273178504e-05, "loss": 0.18704668283462525, "step": 20150 }, { "epoch": 0.10013609767243178, "grad_norm": 1.3853601217269897, "learning_rate": 1.9743157138895395e-05, "loss": 0.19186773300170898, "step": 20160 }, { "epoch": 0.10018576835580104, "grad_norm": 1.0786086320877075, "learning_rate": 1.9742794752463298e-05, "loss": 0.18902339935302734, "step": 20170 }, { "epoch": 0.1002354390391703, "grad_norm": 1.812394380569458, "learning_rate": 1.9742432113891586e-05, "loss": 0.19640438556671141, "step": 20180 }, { "epoch": 0.10028510972253957, "grad_norm": 1.7714587450027466, "learning_rate": 1.974206922318965e-05, "loss": 0.18915205001831054, "step": 20190 }, { "epoch": 0.10033478040590882, "grad_norm": 0.5636407732963562, "learning_rate": 1.9741706080366883e-05, "loss": 0.17649285793304442, "step": 20200 }, { "epoch": 0.10038445108927808, "grad_norm": 1.293167233467102, "learning_rate": 1.9741342685432676e-05, "loss": 0.19167367219924927, "step": 20210 }, { "epoch": 0.10043412177264735, "grad_norm": 1.958289623260498, "learning_rate": 1.9740979038396436e-05, "loss": 0.17875094413757325, "step": 20220 }, { "epoch": 0.10048379245601662, "grad_norm": 2.1273860931396484, "learning_rate": 1.9740615139267578e-05, "loss": 0.19011061191558837, "step": 20230 }, { "epoch": 0.10053346313938587, "grad_norm": 2.1491849422454834, "learning_rate": 1.9740250988055515e-05, "loss": 0.2033862829208374, "step": 20240 }, { "epoch": 0.10058313382275513, "grad_norm": 2.194629430770874, "learning_rate": 1.9739886584769675e-05, "loss": 0.18379493951797485, "step": 20250 }, { "epoch": 0.1006328045061244, "grad_norm": 0.681264340877533, "learning_rate": 1.9739521929419485e-05, "loss": 0.18255602121353148, "step": 20260 }, { "epoch": 0.10068247518949365, "grad_norm": 1.3303059339523315, "learning_rate": 1.9739157022014384e-05, "loss": 0.19611573219299316, "step": 20270 }, { "epoch": 0.10073214587286292, "grad_norm": 1.6702693700790405, "learning_rate": 1.973879186256381e-05, "loss": 0.18097097873687745, "step": 20280 }, { "epoch": 0.10078181655623218, "grad_norm": 1.181983470916748, "learning_rate": 1.9738426451077227e-05, "loss": 0.2054361581802368, "step": 20290 }, { "epoch": 0.10083148723960145, "grad_norm": 3.276829481124878, "learning_rate": 1.973806078756408e-05, "loss": 0.20561909675598145, "step": 20300 }, { "epoch": 0.1008811579229707, "grad_norm": 3.34610652923584, "learning_rate": 1.9737694872033836e-05, "loss": 0.19029467105865477, "step": 20310 }, { "epoch": 0.10093082860633996, "grad_norm": 0.3707225024700165, "learning_rate": 1.9737328704495962e-05, "loss": 0.1904165267944336, "step": 20320 }, { "epoch": 0.10098049928970923, "grad_norm": 2.0026257038116455, "learning_rate": 1.9736962284959936e-05, "loss": 0.19554293155670166, "step": 20330 }, { "epoch": 0.1010301699730785, "grad_norm": 1.2618412971496582, "learning_rate": 1.9736595613435243e-05, "loss": 0.19813809394836426, "step": 20340 }, { "epoch": 0.10107984065644775, "grad_norm": 1.6992802619934082, "learning_rate": 1.973622868993137e-05, "loss": 0.18611469268798828, "step": 20350 }, { "epoch": 0.10112951133981701, "grad_norm": 1.1513850688934326, "learning_rate": 1.973586151445781e-05, "loss": 0.18104767799377441, "step": 20360 }, { "epoch": 0.10117918202318628, "grad_norm": 0.8682419061660767, "learning_rate": 1.9735494087024075e-05, "loss": 0.18046858310699462, "step": 20370 }, { "epoch": 0.10122885270655553, "grad_norm": 1.3428466320037842, "learning_rate": 1.9735126407639666e-05, "loss": 0.19050730466842652, "step": 20380 }, { "epoch": 0.1012785233899248, "grad_norm": 4.508131504058838, "learning_rate": 1.9734758476314092e-05, "loss": 0.19464623928070068, "step": 20390 }, { "epoch": 0.10132819407329406, "grad_norm": 2.5376436710357666, "learning_rate": 1.973439029305689e-05, "loss": 0.1821347951889038, "step": 20400 }, { "epoch": 0.10137786475666333, "grad_norm": 1.7516357898712158, "learning_rate": 1.9734021857877583e-05, "loss": 0.18986332416534424, "step": 20410 }, { "epoch": 0.10142753544003258, "grad_norm": 1.1458927392959595, "learning_rate": 1.97336531707857e-05, "loss": 0.19260857105255128, "step": 20420 }, { "epoch": 0.10147720612340184, "grad_norm": 2.0832743644714355, "learning_rate": 1.973328423179079e-05, "loss": 0.1975390911102295, "step": 20430 }, { "epoch": 0.10152687680677111, "grad_norm": 1.5670983791351318, "learning_rate": 1.9732915040902394e-05, "loss": 0.19093101024627684, "step": 20440 }, { "epoch": 0.10157654749014038, "grad_norm": 1.8139464855194092, "learning_rate": 1.973254559813007e-05, "loss": 0.19051899909973144, "step": 20450 }, { "epoch": 0.10162621817350963, "grad_norm": 1.3660297393798828, "learning_rate": 1.9732175903483382e-05, "loss": 0.19966344833374022, "step": 20460 }, { "epoch": 0.1016758888568789, "grad_norm": 0.8395100235939026, "learning_rate": 1.9731805956971895e-05, "loss": 0.19892942905426025, "step": 20470 }, { "epoch": 0.10172555954024816, "grad_norm": 1.3053663969039917, "learning_rate": 1.973143575860518e-05, "loss": 0.17700296640396118, "step": 20480 }, { "epoch": 0.10177523022361741, "grad_norm": 1.0570971965789795, "learning_rate": 1.9731065308392823e-05, "loss": 0.17825813293457032, "step": 20490 }, { "epoch": 0.10182490090698668, "grad_norm": 1.4012994766235352, "learning_rate": 1.9730694606344406e-05, "loss": 0.19004402160644532, "step": 20500 }, { "epoch": 0.10187457159035594, "grad_norm": 1.1408450603485107, "learning_rate": 1.9730323652469526e-05, "loss": 0.1899944543838501, "step": 20510 }, { "epoch": 0.10192424227372521, "grad_norm": 0.9812561273574829, "learning_rate": 1.9729952446777785e-05, "loss": 0.19366016387939453, "step": 20520 }, { "epoch": 0.10197391295709446, "grad_norm": 0.8147026300430298, "learning_rate": 1.9729580989278785e-05, "loss": 0.19471890926361085, "step": 20530 }, { "epoch": 0.10202358364046372, "grad_norm": 1.1418259143829346, "learning_rate": 1.9729209279982142e-05, "loss": 0.18745429515838624, "step": 20540 }, { "epoch": 0.10207325432383299, "grad_norm": 2.9754645824432373, "learning_rate": 1.9728837318897474e-05, "loss": 0.21314501762390137, "step": 20550 }, { "epoch": 0.10212292500720226, "grad_norm": 1.9918133020401, "learning_rate": 1.9728465106034407e-05, "loss": 0.18483915328979492, "step": 20560 }, { "epoch": 0.10217259569057151, "grad_norm": 1.3668265342712402, "learning_rate": 1.9728092641402577e-05, "loss": 0.20562803745269775, "step": 20570 }, { "epoch": 0.10222226637394077, "grad_norm": 1.2269270420074463, "learning_rate": 1.9727719925011616e-05, "loss": 0.18205761909484863, "step": 20580 }, { "epoch": 0.10227193705731004, "grad_norm": 2.212918996810913, "learning_rate": 1.9727346956871178e-05, "loss": 0.1759930968284607, "step": 20590 }, { "epoch": 0.10232160774067929, "grad_norm": 0.945939838886261, "learning_rate": 1.9726973736990915e-05, "loss": 0.17719619274139403, "step": 20600 }, { "epoch": 0.10237127842404856, "grad_norm": 1.458344578742981, "learning_rate": 1.972660026538048e-05, "loss": 0.1887521982192993, "step": 20610 }, { "epoch": 0.10242094910741782, "grad_norm": 0.8731765747070312, "learning_rate": 1.972622654204954e-05, "loss": 0.17890427112579346, "step": 20620 }, { "epoch": 0.10247061979078709, "grad_norm": 1.1190141439437866, "learning_rate": 1.9725852567007767e-05, "loss": 0.1886986494064331, "step": 20630 }, { "epoch": 0.10252029047415634, "grad_norm": 1.292938232421875, "learning_rate": 1.9725478340264844e-05, "loss": 0.19804461002349855, "step": 20640 }, { "epoch": 0.1025699611575256, "grad_norm": 1.6631923913955688, "learning_rate": 1.972510386183045e-05, "loss": 0.17798563241958618, "step": 20650 }, { "epoch": 0.10261963184089487, "grad_norm": 0.5051987767219543, "learning_rate": 1.9724729131714278e-05, "loss": 0.19571347236633302, "step": 20660 }, { "epoch": 0.10266930252426412, "grad_norm": 1.8837792873382568, "learning_rate": 1.9724354149926026e-05, "loss": 0.1901463270187378, "step": 20670 }, { "epoch": 0.10271897320763339, "grad_norm": 2.04057240486145, "learning_rate": 1.9723978916475398e-05, "loss": 0.19090421199798585, "step": 20680 }, { "epoch": 0.10276864389100265, "grad_norm": 1.776492953300476, "learning_rate": 1.9723603431372108e-05, "loss": 0.20296187400817872, "step": 20690 }, { "epoch": 0.10281831457437192, "grad_norm": 2.7882113456726074, "learning_rate": 1.9723227694625867e-05, "loss": 0.18246290683746338, "step": 20700 }, { "epoch": 0.10286798525774117, "grad_norm": 2.3289036750793457, "learning_rate": 1.9722851706246405e-05, "loss": 0.19792792797088624, "step": 20710 }, { "epoch": 0.10291765594111044, "grad_norm": 1.0452295541763306, "learning_rate": 1.9722475466243453e-05, "loss": 0.19207653999328614, "step": 20720 }, { "epoch": 0.1029673266244797, "grad_norm": 1.4285130500793457, "learning_rate": 1.972209897462674e-05, "loss": 0.18535079956054687, "step": 20730 }, { "epoch": 0.10301699730784897, "grad_norm": 2.7013304233551025, "learning_rate": 1.972172223140602e-05, "loss": 0.19111499786376954, "step": 20740 }, { "epoch": 0.10306666799121822, "grad_norm": 2.3089888095855713, "learning_rate": 1.9721345236591034e-05, "loss": 0.1846242666244507, "step": 20750 }, { "epoch": 0.10311633867458748, "grad_norm": 0.5467361807823181, "learning_rate": 1.9720967990191544e-05, "loss": 0.2035130023956299, "step": 20760 }, { "epoch": 0.10316600935795675, "grad_norm": 2.1514527797698975, "learning_rate": 1.9720590492217312e-05, "loss": 0.20068836212158203, "step": 20770 }, { "epoch": 0.103215680041326, "grad_norm": 2.3933815956115723, "learning_rate": 1.9720212742678103e-05, "loss": 0.19643218517303468, "step": 20780 }, { "epoch": 0.10326535072469527, "grad_norm": 1.8236894607543945, "learning_rate": 1.9719834741583698e-05, "loss": 0.19222676753997803, "step": 20790 }, { "epoch": 0.10331502140806453, "grad_norm": 2.837571382522583, "learning_rate": 1.9719456488943876e-05, "loss": 0.19075365066528321, "step": 20800 }, { "epoch": 0.1033646920914338, "grad_norm": 2.254652500152588, "learning_rate": 1.9719077984768432e-05, "loss": 0.17787702083587648, "step": 20810 }, { "epoch": 0.10341436277480305, "grad_norm": 2.1041111946105957, "learning_rate": 1.9718699229067154e-05, "loss": 0.1924427032470703, "step": 20820 }, { "epoch": 0.10346403345817232, "grad_norm": 1.5424124002456665, "learning_rate": 1.971832022184985e-05, "loss": 0.17322452068328859, "step": 20830 }, { "epoch": 0.10351370414154158, "grad_norm": 0.9710997939109802, "learning_rate": 1.9717940963126327e-05, "loss": 0.20422706604003907, "step": 20840 }, { "epoch": 0.10356337482491085, "grad_norm": 2.3751230239868164, "learning_rate": 1.97175614529064e-05, "loss": 0.18309757709503174, "step": 20850 }, { "epoch": 0.1036130455082801, "grad_norm": 0.5722277760505676, "learning_rate": 1.9717181691199888e-05, "loss": 0.17744734287261962, "step": 20860 }, { "epoch": 0.10366271619164937, "grad_norm": 1.6193853616714478, "learning_rate": 1.971680167801662e-05, "loss": 0.20157370567321778, "step": 20870 }, { "epoch": 0.10371238687501863, "grad_norm": 3.3205490112304688, "learning_rate": 1.971642141336643e-05, "loss": 0.19515151977539064, "step": 20880 }, { "epoch": 0.10376205755838788, "grad_norm": 1.7337119579315186, "learning_rate": 1.9716040897259167e-05, "loss": 0.19323713779449464, "step": 20890 }, { "epoch": 0.10381172824175715, "grad_norm": 1.3503538370132446, "learning_rate": 1.9715660129704667e-05, "loss": 0.18965845108032225, "step": 20900 }, { "epoch": 0.10386139892512641, "grad_norm": 1.5429202318191528, "learning_rate": 1.971527911071279e-05, "loss": 0.19444701671600342, "step": 20910 }, { "epoch": 0.10391106960849568, "grad_norm": 1.3913322687149048, "learning_rate": 1.97148978402934e-05, "loss": 0.19179702997207643, "step": 20920 }, { "epoch": 0.10396074029186493, "grad_norm": 3.920799732208252, "learning_rate": 1.9714516318456354e-05, "loss": 0.19704285860061646, "step": 20930 }, { "epoch": 0.1040104109752342, "grad_norm": 2.352574110031128, "learning_rate": 1.9714134545211535e-05, "loss": 0.19291439056396484, "step": 20940 }, { "epoch": 0.10406008165860346, "grad_norm": 0.6888703107833862, "learning_rate": 1.9713752520568817e-05, "loss": 0.19794158935546874, "step": 20950 }, { "epoch": 0.10410975234197273, "grad_norm": 0.767350435256958, "learning_rate": 1.9713370244538095e-05, "loss": 0.16707131862640381, "step": 20960 }, { "epoch": 0.10415942302534198, "grad_norm": 0.6122194528579712, "learning_rate": 1.9712987717129254e-05, "loss": 0.19008681774139405, "step": 20970 }, { "epoch": 0.10420909370871125, "grad_norm": 1.467066764831543, "learning_rate": 1.9712604938352195e-05, "loss": 0.179802405834198, "step": 20980 }, { "epoch": 0.10425876439208051, "grad_norm": 1.313117265701294, "learning_rate": 1.9712221908216825e-05, "loss": 0.1969053030014038, "step": 20990 }, { "epoch": 0.10430843507544976, "grad_norm": 2.4437179565429688, "learning_rate": 1.9711838626733058e-05, "loss": 0.17738527059555054, "step": 21000 }, { "epoch": 0.10435810575881903, "grad_norm": 1.5444942712783813, "learning_rate": 1.9711455093910813e-05, "loss": 0.19781557321548462, "step": 21010 }, { "epoch": 0.1044077764421883, "grad_norm": 0.9970778226852417, "learning_rate": 1.9711071309760016e-05, "loss": 0.1993518590927124, "step": 21020 }, { "epoch": 0.10445744712555756, "grad_norm": 1.5285866260528564, "learning_rate": 1.9710687274290595e-05, "loss": 0.1932240605354309, "step": 21030 }, { "epoch": 0.10450711780892681, "grad_norm": 1.966134786605835, "learning_rate": 1.9710302987512496e-05, "loss": 0.1866538405418396, "step": 21040 }, { "epoch": 0.10455678849229608, "grad_norm": 1.5302435159683228, "learning_rate": 1.9709918449435656e-05, "loss": 0.18854212760925293, "step": 21050 }, { "epoch": 0.10460645917566534, "grad_norm": 1.3373475074768066, "learning_rate": 1.9709533660070033e-05, "loss": 0.19351673126220703, "step": 21060 }, { "epoch": 0.1046561298590346, "grad_norm": 1.6334677934646606, "learning_rate": 1.9709148619425584e-05, "loss": 0.19484304189682006, "step": 21070 }, { "epoch": 0.10470580054240386, "grad_norm": 1.5814218521118164, "learning_rate": 1.970876332751227e-05, "loss": 0.19020285606384277, "step": 21080 }, { "epoch": 0.10475547122577313, "grad_norm": 1.7168203592300415, "learning_rate": 1.9708377784340068e-05, "loss": 0.18527443408966066, "step": 21090 }, { "epoch": 0.10480514190914239, "grad_norm": 1.470018744468689, "learning_rate": 1.970799198991895e-05, "loss": 0.1951265811920166, "step": 21100 }, { "epoch": 0.10485481259251164, "grad_norm": 1.5113571882247925, "learning_rate": 1.9707605944258904e-05, "loss": 0.19615435600280762, "step": 21110 }, { "epoch": 0.10490448327588091, "grad_norm": 3.5331764221191406, "learning_rate": 1.9707219647369916e-05, "loss": 0.186106276512146, "step": 21120 }, { "epoch": 0.10495415395925017, "grad_norm": 1.4097319841384888, "learning_rate": 1.9706833099261988e-05, "loss": 0.1792589783668518, "step": 21130 }, { "epoch": 0.10500382464261944, "grad_norm": 2.6105496883392334, "learning_rate": 1.9706446299945125e-05, "loss": 0.177968168258667, "step": 21140 }, { "epoch": 0.10505349532598869, "grad_norm": 1.3676344156265259, "learning_rate": 1.970605924942933e-05, "loss": 0.197504723072052, "step": 21150 }, { "epoch": 0.10510316600935796, "grad_norm": 2.7845356464385986, "learning_rate": 1.970567194772463e-05, "loss": 0.20091958045959474, "step": 21160 }, { "epoch": 0.10515283669272722, "grad_norm": 1.867488980293274, "learning_rate": 1.970528439484104e-05, "loss": 0.18247202634811402, "step": 21170 }, { "epoch": 0.10520250737609647, "grad_norm": 1.6642398834228516, "learning_rate": 1.970489659078859e-05, "loss": 0.1818922996520996, "step": 21180 }, { "epoch": 0.10525217805946574, "grad_norm": 1.2169251441955566, "learning_rate": 1.9704508535577323e-05, "loss": 0.1915330171585083, "step": 21190 }, { "epoch": 0.105301848742835, "grad_norm": 1.8293477296829224, "learning_rate": 1.9704120229217274e-05, "loss": 0.18414343595504762, "step": 21200 }, { "epoch": 0.10535151942620427, "grad_norm": 2.583134174346924, "learning_rate": 1.9703731671718497e-05, "loss": 0.1889459252357483, "step": 21210 }, { "epoch": 0.10540119010957352, "grad_norm": 2.2135603427886963, "learning_rate": 1.9703342863091045e-05, "loss": 0.169852614402771, "step": 21220 }, { "epoch": 0.10545086079294279, "grad_norm": 0.9355728626251221, "learning_rate": 1.970295380334498e-05, "loss": 0.20729150772094726, "step": 21230 }, { "epoch": 0.10550053147631205, "grad_norm": 1.4860392808914185, "learning_rate": 1.970256449249037e-05, "loss": 0.18840707540512086, "step": 21240 }, { "epoch": 0.10555020215968132, "grad_norm": 1.2425932884216309, "learning_rate": 1.97021749305373e-05, "loss": 0.19907011985778808, "step": 21250 }, { "epoch": 0.10559987284305057, "grad_norm": 1.211074948310852, "learning_rate": 1.9701785117495837e-05, "loss": 0.18583171367645263, "step": 21260 }, { "epoch": 0.10564954352641984, "grad_norm": 0.5951156616210938, "learning_rate": 1.970139505337608e-05, "loss": 0.19483118057250975, "step": 21270 }, { "epoch": 0.1056992142097891, "grad_norm": 1.0482953786849976, "learning_rate": 1.9701004738188117e-05, "loss": 0.1737664222717285, "step": 21280 }, { "epoch": 0.10574888489315835, "grad_norm": 1.4593695402145386, "learning_rate": 1.9700614171942055e-05, "loss": 0.18190398216247558, "step": 21290 }, { "epoch": 0.10579855557652762, "grad_norm": 1.2570455074310303, "learning_rate": 1.9700223354647993e-05, "loss": 0.19038262367248535, "step": 21300 }, { "epoch": 0.10584822625989689, "grad_norm": 1.2461779117584229, "learning_rate": 1.9699832286316057e-05, "loss": 0.18756964206695556, "step": 21310 }, { "epoch": 0.10589789694326615, "grad_norm": 1.3450074195861816, "learning_rate": 1.969944096695636e-05, "loss": 0.2084531307220459, "step": 21320 }, { "epoch": 0.1059475676266354, "grad_norm": 3.0260162353515625, "learning_rate": 1.969904939657903e-05, "loss": 0.1925520896911621, "step": 21330 }, { "epoch": 0.10599723831000467, "grad_norm": 1.8341543674468994, "learning_rate": 1.96986575751942e-05, "loss": 0.1827033519744873, "step": 21340 }, { "epoch": 0.10604690899337393, "grad_norm": 2.1434996128082275, "learning_rate": 1.9698265502812014e-05, "loss": 0.17879147529602052, "step": 21350 }, { "epoch": 0.1060965796767432, "grad_norm": 0.8859435319900513, "learning_rate": 1.9697873179442615e-05, "loss": 0.17810559272766113, "step": 21360 }, { "epoch": 0.10614625036011245, "grad_norm": 2.09977126121521, "learning_rate": 1.9697480605096157e-05, "loss": 0.1862131714820862, "step": 21370 }, { "epoch": 0.10619592104348172, "grad_norm": 0.749733030796051, "learning_rate": 1.96970877797828e-05, "loss": 0.17333089113235473, "step": 21380 }, { "epoch": 0.10624559172685098, "grad_norm": 0.8822858333587646, "learning_rate": 1.9696694703512712e-05, "loss": 0.19252963066101075, "step": 21390 }, { "epoch": 0.10629526241022023, "grad_norm": 2.310910701751709, "learning_rate": 1.9696301376296063e-05, "loss": 0.18286193609237672, "step": 21400 }, { "epoch": 0.1063449330935895, "grad_norm": 2.8431625366210938, "learning_rate": 1.9695907798143034e-05, "loss": 0.1765836238861084, "step": 21410 }, { "epoch": 0.10639460377695877, "grad_norm": 1.7226052284240723, "learning_rate": 1.969551396906381e-05, "loss": 0.19532427787780762, "step": 21420 }, { "epoch": 0.10644427446032803, "grad_norm": 1.9562091827392578, "learning_rate": 1.9695119889068583e-05, "loss": 0.19219212532043456, "step": 21430 }, { "epoch": 0.10649394514369728, "grad_norm": 0.8219513893127441, "learning_rate": 1.9694725558167548e-05, "loss": 0.1991608738899231, "step": 21440 }, { "epoch": 0.10654361582706655, "grad_norm": 1.332373023033142, "learning_rate": 1.9694330976370915e-05, "loss": 0.19115016460418702, "step": 21450 }, { "epoch": 0.10659328651043581, "grad_norm": 1.9082741737365723, "learning_rate": 1.9693936143688897e-05, "loss": 0.19382312297821044, "step": 21460 }, { "epoch": 0.10664295719380507, "grad_norm": 2.7759952545166016, "learning_rate": 1.969354106013171e-05, "loss": 0.19482705593109131, "step": 21470 }, { "epoch": 0.10669262787717433, "grad_norm": 2.6765878200531006, "learning_rate": 1.969314572570957e-05, "loss": 0.1925346851348877, "step": 21480 }, { "epoch": 0.1067422985605436, "grad_norm": 3.68070650100708, "learning_rate": 1.9692750140432725e-05, "loss": 0.1902238130569458, "step": 21490 }, { "epoch": 0.10679196924391286, "grad_norm": 2.1414318084716797, "learning_rate": 1.96923543043114e-05, "loss": 0.18795948028564452, "step": 21500 }, { "epoch": 0.10684163992728211, "grad_norm": 2.002960443496704, "learning_rate": 1.9691958217355843e-05, "loss": 0.18753548860549926, "step": 21510 }, { "epoch": 0.10689131061065138, "grad_norm": 1.4481006860733032, "learning_rate": 1.9691561879576305e-05, "loss": 0.1943539023399353, "step": 21520 }, { "epoch": 0.10694098129402065, "grad_norm": 3.553715705871582, "learning_rate": 1.969116529098304e-05, "loss": 0.193135666847229, "step": 21530 }, { "epoch": 0.10699065197738991, "grad_norm": 2.298039197921753, "learning_rate": 1.9690768451586316e-05, "loss": 0.18761450052261353, "step": 21540 }, { "epoch": 0.10704032266075916, "grad_norm": 2.5385711193084717, "learning_rate": 1.9690371361396403e-05, "loss": 0.1883173704147339, "step": 21550 }, { "epoch": 0.10708999334412843, "grad_norm": 1.4687098264694214, "learning_rate": 1.9689974020423572e-05, "loss": 0.18611428737640381, "step": 21560 }, { "epoch": 0.1071396640274977, "grad_norm": 1.3811582326889038, "learning_rate": 1.9689576428678113e-05, "loss": 0.18721649646759034, "step": 21570 }, { "epoch": 0.10718933471086695, "grad_norm": 1.3432244062423706, "learning_rate": 1.9689178586170307e-05, "loss": 0.17335314750671388, "step": 21580 }, { "epoch": 0.10723900539423621, "grad_norm": 1.2900810241699219, "learning_rate": 1.9688780492910456e-05, "loss": 0.16551692485809327, "step": 21590 }, { "epoch": 0.10728867607760548, "grad_norm": 1.367341160774231, "learning_rate": 1.9688382148908863e-05, "loss": 0.18150770664215088, "step": 21600 }, { "epoch": 0.10733834676097474, "grad_norm": 1.3497684001922607, "learning_rate": 1.968798355417584e-05, "loss": 0.1920513391494751, "step": 21610 }, { "epoch": 0.107388017444344, "grad_norm": 1.4791990518569946, "learning_rate": 1.9687584708721693e-05, "loss": 0.20089972019195557, "step": 21620 }, { "epoch": 0.10743768812771326, "grad_norm": 2.2771432399749756, "learning_rate": 1.968718561255675e-05, "loss": 0.18744547367095948, "step": 21630 }, { "epoch": 0.10748735881108253, "grad_norm": 1.295927882194519, "learning_rate": 1.9686786265691336e-05, "loss": 0.17839587926864625, "step": 21640 }, { "epoch": 0.10753702949445179, "grad_norm": 0.8480555415153503, "learning_rate": 1.9686386668135794e-05, "loss": 0.17163450717926027, "step": 21650 }, { "epoch": 0.10758670017782104, "grad_norm": 3.198270320892334, "learning_rate": 1.9685986819900455e-05, "loss": 0.19303289651870728, "step": 21660 }, { "epoch": 0.10763637086119031, "grad_norm": 1.5305180549621582, "learning_rate": 1.9685586720995673e-05, "loss": 0.180466628074646, "step": 21670 }, { "epoch": 0.10768604154455957, "grad_norm": 1.363990306854248, "learning_rate": 1.9685186371431802e-05, "loss": 0.18183956146240235, "step": 21680 }, { "epoch": 0.10773571222792883, "grad_norm": 1.9291765689849854, "learning_rate": 1.96847857712192e-05, "loss": 0.17797415256500243, "step": 21690 }, { "epoch": 0.10778538291129809, "grad_norm": 0.8094537854194641, "learning_rate": 1.9684384920368237e-05, "loss": 0.19200608730316163, "step": 21700 }, { "epoch": 0.10783505359466736, "grad_norm": 3.1359548568725586, "learning_rate": 1.968398381888929e-05, "loss": 0.18731296062469482, "step": 21710 }, { "epoch": 0.10788472427803662, "grad_norm": 1.0000557899475098, "learning_rate": 1.968358246679273e-05, "loss": 0.18074607849121094, "step": 21720 }, { "epoch": 0.10793439496140587, "grad_norm": 2.7831473350524902, "learning_rate": 1.9683180864088952e-05, "loss": 0.18360965251922606, "step": 21730 }, { "epoch": 0.10798406564477514, "grad_norm": 1.0444321632385254, "learning_rate": 1.9682779010788343e-05, "loss": 0.1842895746231079, "step": 21740 }, { "epoch": 0.1080337363281444, "grad_norm": 2.021507978439331, "learning_rate": 1.968237690690131e-05, "loss": 0.19479280710220337, "step": 21750 }, { "epoch": 0.10808340701151366, "grad_norm": 1.1650875806808472, "learning_rate": 1.9681974552438257e-05, "loss": 0.17180030345916747, "step": 21760 }, { "epoch": 0.10813307769488292, "grad_norm": 1.7040657997131348, "learning_rate": 1.9681571947409596e-05, "loss": 0.19869732856750488, "step": 21770 }, { "epoch": 0.10818274837825219, "grad_norm": 2.859083414077759, "learning_rate": 1.9681169091825743e-05, "loss": 0.18069689273834227, "step": 21780 }, { "epoch": 0.10823241906162145, "grad_norm": 1.168311357498169, "learning_rate": 1.9680765985697127e-05, "loss": 0.1846521496772766, "step": 21790 }, { "epoch": 0.1082820897449907, "grad_norm": 0.7012618184089661, "learning_rate": 1.968036262903418e-05, "loss": 0.17547259330749512, "step": 21800 }, { "epoch": 0.10833176042835997, "grad_norm": 1.8730727434158325, "learning_rate": 1.967995902184734e-05, "loss": 0.1917588233947754, "step": 21810 }, { "epoch": 0.10838143111172924, "grad_norm": 1.26583993434906, "learning_rate": 1.9679555164147054e-05, "loss": 0.1937042474746704, "step": 21820 }, { "epoch": 0.1084311017950985, "grad_norm": 2.2151906490325928, "learning_rate": 1.967915105594377e-05, "loss": 0.19405808448791503, "step": 21830 }, { "epoch": 0.10848077247846775, "grad_norm": 0.7818794250488281, "learning_rate": 1.967874669724795e-05, "loss": 0.17968736886978148, "step": 21840 }, { "epoch": 0.10853044316183702, "grad_norm": 2.338989734649658, "learning_rate": 1.967834208807006e-05, "loss": 0.18936402797698976, "step": 21850 }, { "epoch": 0.10858011384520629, "grad_norm": 1.1350619792938232, "learning_rate": 1.9677937228420565e-05, "loss": 0.16917016506195068, "step": 21860 }, { "epoch": 0.10862978452857554, "grad_norm": 2.3788623809814453, "learning_rate": 1.9677532118309944e-05, "loss": 0.1958279013633728, "step": 21870 }, { "epoch": 0.1086794552119448, "grad_norm": 1.817977786064148, "learning_rate": 1.9677126757748684e-05, "loss": 0.18051500320434571, "step": 21880 }, { "epoch": 0.10872912589531407, "grad_norm": 3.344275712966919, "learning_rate": 1.967672114674728e-05, "loss": 0.1822829008102417, "step": 21890 }, { "epoch": 0.10877879657868333, "grad_norm": 1.029825210571289, "learning_rate": 1.9676315285316215e-05, "loss": 0.18538451194763184, "step": 21900 }, { "epoch": 0.10882846726205259, "grad_norm": 1.7317609786987305, "learning_rate": 1.967590917346601e-05, "loss": 0.1763991594314575, "step": 21910 }, { "epoch": 0.10887813794542185, "grad_norm": 2.763420581817627, "learning_rate": 1.9675502811207157e-05, "loss": 0.18294380903244017, "step": 21920 }, { "epoch": 0.10892780862879112, "grad_norm": 3.2699787616729736, "learning_rate": 1.9675096198550188e-05, "loss": 0.17936787605285645, "step": 21930 }, { "epoch": 0.10897747931216038, "grad_norm": 2.8615338802337646, "learning_rate": 1.9674689335505617e-05, "loss": 0.17680906057357787, "step": 21940 }, { "epoch": 0.10902714999552963, "grad_norm": 1.2338558435440063, "learning_rate": 1.9674282222083978e-05, "loss": 0.18743281364440917, "step": 21950 }, { "epoch": 0.1090768206788989, "grad_norm": 2.1781105995178223, "learning_rate": 1.9673874858295802e-05, "loss": 0.18164719343185426, "step": 21960 }, { "epoch": 0.10912649136226817, "grad_norm": 1.0732320547103882, "learning_rate": 1.967346724415164e-05, "loss": 0.1948173999786377, "step": 21970 }, { "epoch": 0.10917616204563742, "grad_norm": 2.5104875564575195, "learning_rate": 1.9673059379662028e-05, "loss": 0.17942943572998046, "step": 21980 }, { "epoch": 0.10922583272900668, "grad_norm": 1.3257603645324707, "learning_rate": 1.967265126483753e-05, "loss": 0.20489249229431153, "step": 21990 }, { "epoch": 0.10927550341237595, "grad_norm": 0.8918167352676392, "learning_rate": 1.9672242899688712e-05, "loss": 0.19078903198242186, "step": 22000 }, { "epoch": 0.10932517409574521, "grad_norm": 2.054697036743164, "learning_rate": 1.9671834284226134e-05, "loss": 0.18560856580734253, "step": 22010 }, { "epoch": 0.10937484477911447, "grad_norm": 1.1738728284835815, "learning_rate": 1.9671425418460372e-05, "loss": 0.18870972394943236, "step": 22020 }, { "epoch": 0.10942451546248373, "grad_norm": 1.3093591928482056, "learning_rate": 1.967101630240201e-05, "loss": 0.18335425853729248, "step": 22030 }, { "epoch": 0.109474186145853, "grad_norm": 0.8829242587089539, "learning_rate": 1.9670606936061633e-05, "loss": 0.19412174224853515, "step": 22040 }, { "epoch": 0.10952385682922226, "grad_norm": 3.1518635749816895, "learning_rate": 1.9670197319449838e-05, "loss": 0.18315482139587402, "step": 22050 }, { "epoch": 0.10957352751259151, "grad_norm": 1.6648133993148804, "learning_rate": 1.9669787452577226e-05, "loss": 0.18727535009384155, "step": 22060 }, { "epoch": 0.10962319819596078, "grad_norm": 0.991146445274353, "learning_rate": 1.9669377335454404e-05, "loss": 0.17659451961517333, "step": 22070 }, { "epoch": 0.10967286887933005, "grad_norm": 1.0364251136779785, "learning_rate": 1.966896696809198e-05, "loss": 0.19849834442138672, "step": 22080 }, { "epoch": 0.1097225395626993, "grad_norm": 1.47868812084198, "learning_rate": 1.9668556350500583e-05, "loss": 0.19327327013015747, "step": 22090 }, { "epoch": 0.10977221024606856, "grad_norm": 2.388547420501709, "learning_rate": 1.966814548269083e-05, "loss": 0.1747068405151367, "step": 22100 }, { "epoch": 0.10982188092943783, "grad_norm": 1.123557448387146, "learning_rate": 1.9667734364673364e-05, "loss": 0.19441444873809816, "step": 22110 }, { "epoch": 0.1098715516128071, "grad_norm": 0.9159895777702332, "learning_rate": 1.9667322996458816e-05, "loss": 0.18390003442764283, "step": 22120 }, { "epoch": 0.10992122229617635, "grad_norm": 1.3841928243637085, "learning_rate": 1.966691137805784e-05, "loss": 0.18828638792037963, "step": 22130 }, { "epoch": 0.10997089297954561, "grad_norm": 0.6831275820732117, "learning_rate": 1.9666499509481085e-05, "loss": 0.17669867277145385, "step": 22140 }, { "epoch": 0.11002056366291488, "grad_norm": 1.5411027669906616, "learning_rate": 1.9666087390739205e-05, "loss": 0.18593189716339112, "step": 22150 }, { "epoch": 0.11007023434628413, "grad_norm": 2.046295166015625, "learning_rate": 1.966567502184287e-05, "loss": 0.183856201171875, "step": 22160 }, { "epoch": 0.1101199050296534, "grad_norm": 1.5778437852859497, "learning_rate": 1.9665262402802753e-05, "loss": 0.18939414024353027, "step": 22170 }, { "epoch": 0.11016957571302266, "grad_norm": 1.3208165168762207, "learning_rate": 1.9664849533629536e-05, "loss": 0.1839495301246643, "step": 22180 }, { "epoch": 0.11021924639639193, "grad_norm": 1.2157710790634155, "learning_rate": 1.966443641433389e-05, "loss": 0.18404061794281007, "step": 22190 }, { "epoch": 0.11026891707976118, "grad_norm": 1.137845516204834, "learning_rate": 1.9664023044926522e-05, "loss": 0.17569512128829956, "step": 22200 }, { "epoch": 0.11031858776313044, "grad_norm": 2.9862587451934814, "learning_rate": 1.966360942541812e-05, "loss": 0.1883625030517578, "step": 22210 }, { "epoch": 0.11036825844649971, "grad_norm": 1.4960967302322388, "learning_rate": 1.9663195555819396e-05, "loss": 0.1886997938156128, "step": 22220 }, { "epoch": 0.11041792912986897, "grad_norm": 1.2706326246261597, "learning_rate": 1.9662781436141055e-05, "loss": 0.1675952196121216, "step": 22230 }, { "epoch": 0.11046759981323823, "grad_norm": 2.700948476791382, "learning_rate": 1.9662367066393813e-05, "loss": 0.18023704290390014, "step": 22240 }, { "epoch": 0.11051727049660749, "grad_norm": 2.149937391281128, "learning_rate": 1.96619524465884e-05, "loss": 0.1952844500541687, "step": 22250 }, { "epoch": 0.11056694117997676, "grad_norm": 2.7322514057159424, "learning_rate": 1.966153757673554e-05, "loss": 0.17764906883239745, "step": 22260 }, { "epoch": 0.11061661186334601, "grad_norm": 1.3575249910354614, "learning_rate": 1.9661122456845974e-05, "loss": 0.1785430431365967, "step": 22270 }, { "epoch": 0.11066628254671527, "grad_norm": 2.1253488063812256, "learning_rate": 1.966070708693044e-05, "loss": 0.1911381721496582, "step": 22280 }, { "epoch": 0.11071595323008454, "grad_norm": 2.402343511581421, "learning_rate": 1.9660291466999693e-05, "loss": 0.18301595449447633, "step": 22290 }, { "epoch": 0.1107656239134538, "grad_norm": 1.7138646841049194, "learning_rate": 1.965987559706449e-05, "loss": 0.16922662258148194, "step": 22300 }, { "epoch": 0.11081529459682306, "grad_norm": 0.8548092842102051, "learning_rate": 1.965945947713559e-05, "loss": 0.18446478843688965, "step": 22310 }, { "epoch": 0.11086496528019232, "grad_norm": 3.377815008163452, "learning_rate": 1.965904310722376e-05, "loss": 0.18266420364379882, "step": 22320 }, { "epoch": 0.11091463596356159, "grad_norm": 1.4907112121582031, "learning_rate": 1.965862648733978e-05, "loss": 0.18832659721374512, "step": 22330 }, { "epoch": 0.11096430664693085, "grad_norm": 2.5438225269317627, "learning_rate": 1.9658209617494427e-05, "loss": 0.1939724087715149, "step": 22340 }, { "epoch": 0.1110139773303001, "grad_norm": 1.4873930215835571, "learning_rate": 1.9657792497698492e-05, "loss": 0.18536102771759033, "step": 22350 }, { "epoch": 0.11106364801366937, "grad_norm": 4.187832832336426, "learning_rate": 1.9657375127962776e-05, "loss": 0.18015031814575194, "step": 22360 }, { "epoch": 0.11111331869703864, "grad_norm": 2.864245891571045, "learning_rate": 1.9656957508298072e-05, "loss": 0.19197965860366822, "step": 22370 }, { "epoch": 0.11116298938040789, "grad_norm": 1.6996181011199951, "learning_rate": 1.965653963871519e-05, "loss": 0.18777337074279785, "step": 22380 }, { "epoch": 0.11121266006377716, "grad_norm": 2.2612903118133545, "learning_rate": 1.9656121519224943e-05, "loss": 0.18904783725738525, "step": 22390 }, { "epoch": 0.11126233074714642, "grad_norm": 1.7079170942306519, "learning_rate": 1.9655703149838157e-05, "loss": 0.19267714023590088, "step": 22400 }, { "epoch": 0.11131200143051569, "grad_norm": 2.7860262393951416, "learning_rate": 1.9655284530565654e-05, "loss": 0.16890547275543213, "step": 22410 }, { "epoch": 0.11136167211388494, "grad_norm": 1.4987156391143799, "learning_rate": 1.965486566141827e-05, "loss": 0.18344271183013916, "step": 22420 }, { "epoch": 0.1114113427972542, "grad_norm": 1.9315344095230103, "learning_rate": 1.9654446542406843e-05, "loss": 0.1897791624069214, "step": 22430 }, { "epoch": 0.11146101348062347, "grad_norm": 0.9130749106407166, "learning_rate": 1.9654027173542223e-05, "loss": 0.1813925862312317, "step": 22440 }, { "epoch": 0.11151068416399273, "grad_norm": 1.8078868389129639, "learning_rate": 1.9653607554835265e-05, "loss": 0.1870405673980713, "step": 22450 }, { "epoch": 0.11156035484736199, "grad_norm": 1.015340805053711, "learning_rate": 1.965318768629682e-05, "loss": 0.18544883728027345, "step": 22460 }, { "epoch": 0.11161002553073125, "grad_norm": 3.029771566390991, "learning_rate": 1.9652767567937762e-05, "loss": 0.18226361274719238, "step": 22470 }, { "epoch": 0.11165969621410052, "grad_norm": 1.372821569442749, "learning_rate": 1.9652347199768957e-05, "loss": 0.18888192176818847, "step": 22480 }, { "epoch": 0.11170936689746977, "grad_norm": 1.821023941040039, "learning_rate": 1.9651926581801292e-05, "loss": 0.18902232646942138, "step": 22490 }, { "epoch": 0.11175903758083904, "grad_norm": 1.867110252380371, "learning_rate": 1.9651505714045644e-05, "loss": 0.18972172737121581, "step": 22500 }, { "epoch": 0.1118087082642083, "grad_norm": 1.729196310043335, "learning_rate": 1.9651084596512913e-05, "loss": 0.1848900318145752, "step": 22510 }, { "epoch": 0.11185837894757757, "grad_norm": 1.3592371940612793, "learning_rate": 1.965066322921399e-05, "loss": 0.19511740207672118, "step": 22520 }, { "epoch": 0.11190804963094682, "grad_norm": 2.76279354095459, "learning_rate": 1.965024161215978e-05, "loss": 0.20081801414489747, "step": 22530 }, { "epoch": 0.11195772031431608, "grad_norm": 1.1499569416046143, "learning_rate": 1.9649819745361202e-05, "loss": 0.1895670175552368, "step": 22540 }, { "epoch": 0.11200739099768535, "grad_norm": 1.4770753383636475, "learning_rate": 1.964939762882917e-05, "loss": 0.1893952012062073, "step": 22550 }, { "epoch": 0.1120570616810546, "grad_norm": 1.4846926927566528, "learning_rate": 1.9648975262574606e-05, "loss": 0.1702141284942627, "step": 22560 }, { "epoch": 0.11210673236442387, "grad_norm": 2.462369203567505, "learning_rate": 1.964855264660844e-05, "loss": 0.19001039266586303, "step": 22570 }, { "epoch": 0.11215640304779313, "grad_norm": 1.7335201501846313, "learning_rate": 1.964812978094161e-05, "loss": 0.1880026340484619, "step": 22580 }, { "epoch": 0.1122060737311624, "grad_norm": 1.7997575998306274, "learning_rate": 1.9647706665585064e-05, "loss": 0.1938660502433777, "step": 22590 }, { "epoch": 0.11225574441453165, "grad_norm": 1.4491398334503174, "learning_rate": 1.964728330054975e-05, "loss": 0.19499192237854004, "step": 22600 }, { "epoch": 0.11230541509790092, "grad_norm": 1.354954481124878, "learning_rate": 1.964685968584662e-05, "loss": 0.18612470626831054, "step": 22610 }, { "epoch": 0.11235508578127018, "grad_norm": 1.939731478691101, "learning_rate": 1.964643582148664e-05, "loss": 0.1719478964805603, "step": 22620 }, { "epoch": 0.11240475646463945, "grad_norm": 1.4577544927597046, "learning_rate": 1.964601170748078e-05, "loss": 0.19459092617034912, "step": 22630 }, { "epoch": 0.1124544271480087, "grad_norm": 0.9118837714195251, "learning_rate": 1.964558734384002e-05, "loss": 0.17830010652542114, "step": 22640 }, { "epoch": 0.11250409783137796, "grad_norm": 2.9067187309265137, "learning_rate": 1.9645162730575335e-05, "loss": 0.18221726417541503, "step": 22650 }, { "epoch": 0.11255376851474723, "grad_norm": 1.1617958545684814, "learning_rate": 1.964473786769771e-05, "loss": 0.1681925654411316, "step": 22660 }, { "epoch": 0.11260343919811648, "grad_norm": 2.6477015018463135, "learning_rate": 1.9644312755218154e-05, "loss": 0.17613484859466552, "step": 22670 }, { "epoch": 0.11265310988148575, "grad_norm": 1.4886071681976318, "learning_rate": 1.964388739314766e-05, "loss": 0.17583433389663697, "step": 22680 }, { "epoch": 0.11270278056485501, "grad_norm": 2.444638252258301, "learning_rate": 1.9643461781497238e-05, "loss": 0.1985781669616699, "step": 22690 }, { "epoch": 0.11275245124822428, "grad_norm": 1.7382787466049194, "learning_rate": 1.9643035920277903e-05, "loss": 0.1821505308151245, "step": 22700 }, { "epoch": 0.11280212193159353, "grad_norm": 1.7792073488235474, "learning_rate": 1.9642609809500675e-05, "loss": 0.18604251146316528, "step": 22710 }, { "epoch": 0.1128517926149628, "grad_norm": 3.70404052734375, "learning_rate": 1.9642183449176585e-05, "loss": 0.19327166080474853, "step": 22720 }, { "epoch": 0.11290146329833206, "grad_norm": 1.5793368816375732, "learning_rate": 1.9641756839316662e-05, "loss": 0.18806263208389282, "step": 22730 }, { "epoch": 0.11295113398170133, "grad_norm": 2.112532615661621, "learning_rate": 1.964132997993195e-05, "loss": 0.18661506175994874, "step": 22740 }, { "epoch": 0.11300080466507058, "grad_norm": 1.6241191625595093, "learning_rate": 1.9640902871033493e-05, "loss": 0.16786432266235352, "step": 22750 }, { "epoch": 0.11305047534843984, "grad_norm": 1.4092446565628052, "learning_rate": 1.9640475512632347e-05, "loss": 0.20922188758850097, "step": 22760 }, { "epoch": 0.11310014603180911, "grad_norm": 1.299938678741455, "learning_rate": 1.964004790473957e-05, "loss": 0.18323277235031127, "step": 22770 }, { "epoch": 0.11314981671517836, "grad_norm": 1.4192273616790771, "learning_rate": 1.963962004736623e-05, "loss": 0.19195976257324218, "step": 22780 }, { "epoch": 0.11319948739854763, "grad_norm": 1.3672752380371094, "learning_rate": 1.9639191940523398e-05, "loss": 0.1769030809402466, "step": 22790 }, { "epoch": 0.11324915808191689, "grad_norm": 2.249074935913086, "learning_rate": 1.9638763584222156e-05, "loss": 0.1809845209121704, "step": 22800 }, { "epoch": 0.11329882876528616, "grad_norm": 1.2965421676635742, "learning_rate": 1.9638334978473586e-05, "loss": 0.1839402437210083, "step": 22810 }, { "epoch": 0.11334849944865541, "grad_norm": 0.5122362375259399, "learning_rate": 1.9637906123288786e-05, "loss": 0.1801236629486084, "step": 22820 }, { "epoch": 0.11339817013202468, "grad_norm": 3.482090950012207, "learning_rate": 1.963747701867885e-05, "loss": 0.18839387893676757, "step": 22830 }, { "epoch": 0.11344784081539394, "grad_norm": 1.0720736980438232, "learning_rate": 1.963704766465488e-05, "loss": 0.18018529415130616, "step": 22840 }, { "epoch": 0.1134975114987632, "grad_norm": 1.2108513116836548, "learning_rate": 1.9636618061227994e-05, "loss": 0.18243671655654908, "step": 22850 }, { "epoch": 0.11354718218213246, "grad_norm": 1.9769290685653687, "learning_rate": 1.963618820840931e-05, "loss": 0.18201104402542115, "step": 22860 }, { "epoch": 0.11359685286550172, "grad_norm": 2.0594253540039062, "learning_rate": 1.9635758106209943e-05, "loss": 0.19190491437911988, "step": 22870 }, { "epoch": 0.11364652354887099, "grad_norm": 1.9202160835266113, "learning_rate": 1.9635327754641035e-05, "loss": 0.18252625465393066, "step": 22880 }, { "epoch": 0.11369619423224024, "grad_norm": 0.9285123348236084, "learning_rate": 1.9634897153713718e-05, "loss": 0.1810271620750427, "step": 22890 }, { "epoch": 0.11374586491560951, "grad_norm": 0.8702754974365234, "learning_rate": 1.963446630343914e-05, "loss": 0.1896612048149109, "step": 22900 }, { "epoch": 0.11379553559897877, "grad_norm": 1.596323847770691, "learning_rate": 1.963403520382844e-05, "loss": 0.2000883102416992, "step": 22910 }, { "epoch": 0.11384520628234804, "grad_norm": 1.0986125469207764, "learning_rate": 1.963360385489279e-05, "loss": 0.18984869718551636, "step": 22920 }, { "epoch": 0.11389487696571729, "grad_norm": 1.3420519828796387, "learning_rate": 1.9633172256643343e-05, "loss": 0.17718846797943116, "step": 22930 }, { "epoch": 0.11394454764908656, "grad_norm": 1.1586573123931885, "learning_rate": 1.963274040909127e-05, "loss": 0.1849939465522766, "step": 22940 }, { "epoch": 0.11399421833245582, "grad_norm": 2.019632577896118, "learning_rate": 1.9632308312247753e-05, "loss": 0.18635210990905762, "step": 22950 }, { "epoch": 0.11404388901582507, "grad_norm": 1.5103329420089722, "learning_rate": 1.9631875966123966e-05, "loss": 0.170641028881073, "step": 22960 }, { "epoch": 0.11409355969919434, "grad_norm": 1.131539225578308, "learning_rate": 1.96314433707311e-05, "loss": 0.17762820720672606, "step": 22970 }, { "epoch": 0.1141432303825636, "grad_norm": 0.8273907899856567, "learning_rate": 1.9631010526080356e-05, "loss": 0.18705960512161254, "step": 22980 }, { "epoch": 0.11419290106593287, "grad_norm": 2.0237271785736084, "learning_rate": 1.963057743218293e-05, "loss": 0.19399800300598144, "step": 22990 }, { "epoch": 0.11424257174930212, "grad_norm": 0.977602481842041, "learning_rate": 1.9630144089050035e-05, "loss": 0.17028437852859496, "step": 23000 }, { "epoch": 0.11429224243267139, "grad_norm": 2.012540102005005, "learning_rate": 1.962971049669288e-05, "loss": 0.19038960933685303, "step": 23010 }, { "epoch": 0.11434191311604065, "grad_norm": 1.0703097581863403, "learning_rate": 1.962927665512269e-05, "loss": 0.19132132530212403, "step": 23020 }, { "epoch": 0.11439158379940992, "grad_norm": 1.2220356464385986, "learning_rate": 1.9628842564350693e-05, "loss": 0.1887289524078369, "step": 23030 }, { "epoch": 0.11444125448277917, "grad_norm": 1.352774977684021, "learning_rate": 1.9628408224388123e-05, "loss": 0.1970276117324829, "step": 23040 }, { "epoch": 0.11449092516614844, "grad_norm": 1.0458873510360718, "learning_rate": 1.9627973635246215e-05, "loss": 0.19810642004013063, "step": 23050 }, { "epoch": 0.1145405958495177, "grad_norm": 1.5203297138214111, "learning_rate": 1.9627538796936226e-05, "loss": 0.1650092363357544, "step": 23060 }, { "epoch": 0.11459026653288695, "grad_norm": 1.3623496294021606, "learning_rate": 1.9627103709469398e-05, "loss": 0.17920498847961425, "step": 23070 }, { "epoch": 0.11463993721625622, "grad_norm": 2.4093334674835205, "learning_rate": 1.9626668372857004e-05, "loss": 0.18428491353988646, "step": 23080 }, { "epoch": 0.11468960789962548, "grad_norm": 1.2944340705871582, "learning_rate": 1.96262327871103e-05, "loss": 0.19758046865463258, "step": 23090 }, { "epoch": 0.11473927858299475, "grad_norm": 2.1789443492889404, "learning_rate": 1.9625796952240562e-05, "loss": 0.19201343059539794, "step": 23100 }, { "epoch": 0.114788949266364, "grad_norm": 1.179179310798645, "learning_rate": 1.962536086825907e-05, "loss": 0.17897615432739258, "step": 23110 }, { "epoch": 0.11483861994973327, "grad_norm": 1.2601840496063232, "learning_rate": 1.9624924535177105e-05, "loss": 0.1990141749382019, "step": 23120 }, { "epoch": 0.11488829063310253, "grad_norm": 1.3385006189346313, "learning_rate": 1.9624487953005967e-05, "loss": 0.17325563430786134, "step": 23130 }, { "epoch": 0.1149379613164718, "grad_norm": 1.4047143459320068, "learning_rate": 1.962405112175695e-05, "loss": 0.16773048639297486, "step": 23140 }, { "epoch": 0.11498763199984105, "grad_norm": 0.8872002959251404, "learning_rate": 1.9623614041441356e-05, "loss": 0.17842414379119872, "step": 23150 }, { "epoch": 0.11503730268321032, "grad_norm": 0.9800001382827759, "learning_rate": 1.9623176712070505e-05, "loss": 0.18429213762283325, "step": 23160 }, { "epoch": 0.11508697336657958, "grad_norm": 1.9243866205215454, "learning_rate": 1.962273913365571e-05, "loss": 0.18555111885070802, "step": 23170 }, { "epoch": 0.11513664404994883, "grad_norm": 0.7524684071540833, "learning_rate": 1.9622301306208292e-05, "loss": 0.17576323747634887, "step": 23180 }, { "epoch": 0.1151863147333181, "grad_norm": 1.2176194190979004, "learning_rate": 1.962186322973959e-05, "loss": 0.15980428457260132, "step": 23190 }, { "epoch": 0.11523598541668736, "grad_norm": 1.505883812904358, "learning_rate": 1.9621424904260933e-05, "loss": 0.18242232799530028, "step": 23200 }, { "epoch": 0.11528565610005663, "grad_norm": 1.6730881929397583, "learning_rate": 1.9620986329783668e-05, "loss": 0.19518251419067384, "step": 23210 }, { "epoch": 0.11533532678342588, "grad_norm": 0.6993176937103271, "learning_rate": 1.9620547506319143e-05, "loss": 0.19082634449005126, "step": 23220 }, { "epoch": 0.11538499746679515, "grad_norm": 0.8031846880912781, "learning_rate": 1.962010843387872e-05, "loss": 0.1963270902633667, "step": 23230 }, { "epoch": 0.11543466815016441, "grad_norm": 2.093787670135498, "learning_rate": 1.9619669112473758e-05, "loss": 0.18583459854125978, "step": 23240 }, { "epoch": 0.11548433883353368, "grad_norm": 0.9834160208702087, "learning_rate": 1.9619229542115628e-05, "loss": 0.19499425888061522, "step": 23250 }, { "epoch": 0.11553400951690293, "grad_norm": 2.407125234603882, "learning_rate": 1.9618789722815707e-05, "loss": 0.18141722679138184, "step": 23260 }, { "epoch": 0.1155836802002722, "grad_norm": 1.8623874187469482, "learning_rate": 1.961834965458537e-05, "loss": 0.1896566390991211, "step": 23270 }, { "epoch": 0.11563335088364146, "grad_norm": 2.2734336853027344, "learning_rate": 1.9617909337436017e-05, "loss": 0.1897403836250305, "step": 23280 }, { "epoch": 0.11568302156701071, "grad_norm": 0.9669637084007263, "learning_rate": 1.9617468771379036e-05, "loss": 0.1759522557258606, "step": 23290 }, { "epoch": 0.11573269225037998, "grad_norm": 2.9684298038482666, "learning_rate": 1.961702795642583e-05, "loss": 0.20004596710205078, "step": 23300 }, { "epoch": 0.11578236293374924, "grad_norm": 1.2119146585464478, "learning_rate": 1.9616586892587802e-05, "loss": 0.17097549438476561, "step": 23310 }, { "epoch": 0.11583203361711851, "grad_norm": 1.34110426902771, "learning_rate": 1.961614557987638e-05, "loss": 0.1937251329421997, "step": 23320 }, { "epoch": 0.11588170430048776, "grad_norm": 1.6051141023635864, "learning_rate": 1.9615704018302973e-05, "loss": 0.17588330507278443, "step": 23330 }, { "epoch": 0.11593137498385703, "grad_norm": 1.572957158088684, "learning_rate": 1.961526220787901e-05, "loss": 0.1847791314125061, "step": 23340 }, { "epoch": 0.11598104566722629, "grad_norm": 2.3686792850494385, "learning_rate": 1.961482014861593e-05, "loss": 0.1830286979675293, "step": 23350 }, { "epoch": 0.11603071635059554, "grad_norm": 1.2409238815307617, "learning_rate": 1.9614377840525173e-05, "loss": 0.19205632209777831, "step": 23360 }, { "epoch": 0.11608038703396481, "grad_norm": 1.1826274394989014, "learning_rate": 1.961393528361818e-05, "loss": 0.18298695087432862, "step": 23370 }, { "epoch": 0.11613005771733408, "grad_norm": 2.5714943408966064, "learning_rate": 1.961349247790641e-05, "loss": 0.18863344192504883, "step": 23380 }, { "epoch": 0.11617972840070334, "grad_norm": 1.8308814764022827, "learning_rate": 1.9613049423401315e-05, "loss": 0.18549985885620118, "step": 23390 }, { "epoch": 0.1162293990840726, "grad_norm": 1.2350963354110718, "learning_rate": 1.961260612011437e-05, "loss": 0.1674963116645813, "step": 23400 }, { "epoch": 0.11627906976744186, "grad_norm": 2.691056251525879, "learning_rate": 1.9612162568057043e-05, "loss": 0.19202930927276612, "step": 23410 }, { "epoch": 0.11632874045081112, "grad_norm": 0.9732889533042908, "learning_rate": 1.9611718767240814e-05, "loss": 0.18968193531036376, "step": 23420 }, { "epoch": 0.11637841113418039, "grad_norm": 3.5488173961639404, "learning_rate": 1.9611274717677168e-05, "loss": 0.17797653675079345, "step": 23430 }, { "epoch": 0.11642808181754964, "grad_norm": 1.0477932691574097, "learning_rate": 1.9610830419377597e-05, "loss": 0.19053514003753663, "step": 23440 }, { "epoch": 0.11647775250091891, "grad_norm": 1.3459051847457886, "learning_rate": 1.9610385872353598e-05, "loss": 0.18304245471954345, "step": 23450 }, { "epoch": 0.11652742318428817, "grad_norm": 1.0305612087249756, "learning_rate": 1.960994107661668e-05, "loss": 0.17905845642089843, "step": 23460 }, { "epoch": 0.11657709386765742, "grad_norm": 1.452683687210083, "learning_rate": 1.9609496032178346e-05, "loss": 0.17358568906784058, "step": 23470 }, { "epoch": 0.11662676455102669, "grad_norm": 2.9232611656188965, "learning_rate": 1.9609050739050122e-05, "loss": 0.1867268204689026, "step": 23480 }, { "epoch": 0.11667643523439596, "grad_norm": 2.737826108932495, "learning_rate": 1.960860519724353e-05, "loss": 0.1795680284500122, "step": 23490 }, { "epoch": 0.11672610591776522, "grad_norm": 1.0105900764465332, "learning_rate": 1.9608159406770094e-05, "loss": 0.1908029317855835, "step": 23500 }, { "epoch": 0.11677577660113447, "grad_norm": 1.8465301990509033, "learning_rate": 1.9607713367641357e-05, "loss": 0.19466772079467773, "step": 23510 }, { "epoch": 0.11682544728450374, "grad_norm": 1.9298861026763916, "learning_rate": 1.9607267079868865e-05, "loss": 0.16337211132049562, "step": 23520 }, { "epoch": 0.116875117967873, "grad_norm": 0.7471567988395691, "learning_rate": 1.960682054346416e-05, "loss": 0.17505186796188354, "step": 23530 }, { "epoch": 0.11692478865124227, "grad_norm": 3.0138893127441406, "learning_rate": 1.9606373758438805e-05, "loss": 0.19839210510253907, "step": 23540 }, { "epoch": 0.11697445933461152, "grad_norm": 1.971508264541626, "learning_rate": 1.9605926724804357e-05, "loss": 0.17779452800750734, "step": 23550 }, { "epoch": 0.11702413001798079, "grad_norm": 1.7051564455032349, "learning_rate": 1.9605479442572387e-05, "loss": 0.1851255178451538, "step": 23560 }, { "epoch": 0.11707380070135005, "grad_norm": 1.1464122533798218, "learning_rate": 1.960503191175447e-05, "loss": 0.18208413124084472, "step": 23570 }, { "epoch": 0.1171234713847193, "grad_norm": 3.0904624462127686, "learning_rate": 1.9604584132362198e-05, "loss": 0.18646812438964844, "step": 23580 }, { "epoch": 0.11717314206808857, "grad_norm": 2.9931955337524414, "learning_rate": 1.9604136104407144e-05, "loss": 0.18233709335327147, "step": 23590 }, { "epoch": 0.11722281275145784, "grad_norm": 0.8829938769340515, "learning_rate": 1.960368782790091e-05, "loss": 0.18623114824295045, "step": 23600 }, { "epoch": 0.1172724834348271, "grad_norm": 1.5608748197555542, "learning_rate": 1.9603239302855094e-05, "loss": 0.17425296306610108, "step": 23610 }, { "epoch": 0.11732215411819635, "grad_norm": 1.3776805400848389, "learning_rate": 1.960279052928131e-05, "loss": 0.1978244423866272, "step": 23620 }, { "epoch": 0.11737182480156562, "grad_norm": 2.2178568840026855, "learning_rate": 1.9602341507191168e-05, "loss": 0.18582342863082885, "step": 23630 }, { "epoch": 0.11742149548493488, "grad_norm": 2.142784595489502, "learning_rate": 1.9601892236596286e-05, "loss": 0.18029472827911378, "step": 23640 }, { "epoch": 0.11747116616830415, "grad_norm": 0.7257629632949829, "learning_rate": 1.9601442717508298e-05, "loss": 0.18721368312835693, "step": 23650 }, { "epoch": 0.1175208368516734, "grad_norm": 1.4668519496917725, "learning_rate": 1.960099294993883e-05, "loss": 0.1822287917137146, "step": 23660 }, { "epoch": 0.11757050753504267, "grad_norm": 1.4419876337051392, "learning_rate": 1.9600542933899525e-05, "loss": 0.17089035511016845, "step": 23670 }, { "epoch": 0.11762017821841193, "grad_norm": 0.7067856192588806, "learning_rate": 1.9600092669402027e-05, "loss": 0.1911992311477661, "step": 23680 }, { "epoch": 0.11766984890178118, "grad_norm": 1.6759475469589233, "learning_rate": 1.9599642156457993e-05, "loss": 0.17848708629608154, "step": 23690 }, { "epoch": 0.11771951958515045, "grad_norm": 0.923112154006958, "learning_rate": 1.959919139507908e-05, "loss": 0.2037869930267334, "step": 23700 }, { "epoch": 0.11776919026851972, "grad_norm": 2.4790406227111816, "learning_rate": 1.9598740385276956e-05, "loss": 0.1857994556427002, "step": 23710 }, { "epoch": 0.11781886095188898, "grad_norm": 2.8503918647766113, "learning_rate": 1.9598289127063282e-05, "loss": 0.1962348222732544, "step": 23720 }, { "epoch": 0.11786853163525823, "grad_norm": 2.2943806648254395, "learning_rate": 1.9597837620449752e-05, "loss": 0.2019651412963867, "step": 23730 }, { "epoch": 0.1179182023186275, "grad_norm": 1.7184178829193115, "learning_rate": 1.9597385865448042e-05, "loss": 0.18210729360580444, "step": 23740 }, { "epoch": 0.11796787300199676, "grad_norm": 1.315447211265564, "learning_rate": 1.9596933862069844e-05, "loss": 0.1829358696937561, "step": 23750 }, { "epoch": 0.11801754368536602, "grad_norm": 1.6193840503692627, "learning_rate": 1.959648161032686e-05, "loss": 0.1885669469833374, "step": 23760 }, { "epoch": 0.11806721436873528, "grad_norm": 1.5927666425704956, "learning_rate": 1.9596029110230787e-05, "loss": 0.18349366188049315, "step": 23770 }, { "epoch": 0.11811688505210455, "grad_norm": 0.9572808146476746, "learning_rate": 1.959557636179334e-05, "loss": 0.17965657711029054, "step": 23780 }, { "epoch": 0.11816655573547381, "grad_norm": 0.9633793234825134, "learning_rate": 1.9595123365026234e-05, "loss": 0.19739093780517578, "step": 23790 }, { "epoch": 0.11821622641884307, "grad_norm": 1.5237866640090942, "learning_rate": 1.9594670119941192e-05, "loss": 0.2050025463104248, "step": 23800 }, { "epoch": 0.11826589710221233, "grad_norm": 1.700057864189148, "learning_rate": 1.959421662654995e-05, "loss": 0.186869215965271, "step": 23810 }, { "epoch": 0.1183155677855816, "grad_norm": 2.6643552780151367, "learning_rate": 1.9593762884864235e-05, "loss": 0.17842614650726318, "step": 23820 }, { "epoch": 0.11836523846895086, "grad_norm": 1.9663904905319214, "learning_rate": 1.9593308894895794e-05, "loss": 0.1953609585762024, "step": 23830 }, { "epoch": 0.11841490915232011, "grad_norm": 1.6616030931472778, "learning_rate": 1.9592854656656378e-05, "loss": 0.18624413013458252, "step": 23840 }, { "epoch": 0.11846457983568938, "grad_norm": 1.0829110145568848, "learning_rate": 1.959240017015774e-05, "loss": 0.1851031541824341, "step": 23850 }, { "epoch": 0.11851425051905864, "grad_norm": 1.5332762002944946, "learning_rate": 1.9591945435411643e-05, "loss": 0.18615301847457885, "step": 23860 }, { "epoch": 0.1185639212024279, "grad_norm": 1.7003296613693237, "learning_rate": 1.9591490452429855e-05, "loss": 0.1861961603164673, "step": 23870 }, { "epoch": 0.11861359188579716, "grad_norm": 1.2499213218688965, "learning_rate": 1.959103522122415e-05, "loss": 0.2000974178314209, "step": 23880 }, { "epoch": 0.11866326256916643, "grad_norm": 3.2369110584259033, "learning_rate": 1.959057974180631e-05, "loss": 0.18636926412582397, "step": 23890 }, { "epoch": 0.1187129332525357, "grad_norm": 1.5148711204528809, "learning_rate": 1.959012401418812e-05, "loss": 0.18871874809265138, "step": 23900 }, { "epoch": 0.11876260393590495, "grad_norm": 1.0342508554458618, "learning_rate": 1.9589668038381383e-05, "loss": 0.18099184036254884, "step": 23910 }, { "epoch": 0.11881227461927421, "grad_norm": 2.0318822860717773, "learning_rate": 1.9589211814397888e-05, "loss": 0.1779738187789917, "step": 23920 }, { "epoch": 0.11886194530264348, "grad_norm": 1.1215002536773682, "learning_rate": 1.9588755342249447e-05, "loss": 0.19663784503936768, "step": 23930 }, { "epoch": 0.11891161598601274, "grad_norm": 1.951589822769165, "learning_rate": 1.9588298621947875e-05, "loss": 0.19811060428619384, "step": 23940 }, { "epoch": 0.118961286669382, "grad_norm": 1.0897016525268555, "learning_rate": 1.958784165350499e-05, "loss": 0.1720031976699829, "step": 23950 }, { "epoch": 0.11901095735275126, "grad_norm": 2.1146740913391113, "learning_rate": 1.9587384436932616e-05, "loss": 0.18810521364212035, "step": 23960 }, { "epoch": 0.11906062803612052, "grad_norm": 0.9177601933479309, "learning_rate": 1.958692697224259e-05, "loss": 0.18181511163711547, "step": 23970 }, { "epoch": 0.11911029871948978, "grad_norm": 1.468580722808838, "learning_rate": 1.958646925944675e-05, "loss": 0.1799434781074524, "step": 23980 }, { "epoch": 0.11915996940285904, "grad_norm": 1.8358899354934692, "learning_rate": 1.9586011298556934e-05, "loss": 0.18944259881973266, "step": 23990 }, { "epoch": 0.11920964008622831, "grad_norm": 1.0459588766098022, "learning_rate": 1.9585553089585002e-05, "loss": 0.1956284761428833, "step": 24000 }, { "epoch": 0.11925931076959757, "grad_norm": 1.853079915046692, "learning_rate": 1.9585094632542816e-05, "loss": 0.1958961606025696, "step": 24010 }, { "epoch": 0.11930898145296683, "grad_norm": 1.1647491455078125, "learning_rate": 1.9584635927442227e-05, "loss": 0.18938653469085692, "step": 24020 }, { "epoch": 0.11935865213633609, "grad_norm": 1.4364266395568848, "learning_rate": 1.958417697429512e-05, "loss": 0.18505784273147582, "step": 24030 }, { "epoch": 0.11940832281970536, "grad_norm": 2.6527342796325684, "learning_rate": 1.9583717773113365e-05, "loss": 0.19813594818115235, "step": 24040 }, { "epoch": 0.11945799350307462, "grad_norm": 1.7969967126846313, "learning_rate": 1.9583258323908846e-05, "loss": 0.19321739673614502, "step": 24050 }, { "epoch": 0.11950766418644387, "grad_norm": 3.6425070762634277, "learning_rate": 1.9582798626693456e-05, "loss": 0.18511006832122803, "step": 24060 }, { "epoch": 0.11955733486981314, "grad_norm": 1.523680567741394, "learning_rate": 1.9582338681479092e-05, "loss": 0.17408833503723145, "step": 24070 }, { "epoch": 0.1196070055531824, "grad_norm": 1.6542237997055054, "learning_rate": 1.958187848827765e-05, "loss": 0.1938474416732788, "step": 24080 }, { "epoch": 0.11965667623655166, "grad_norm": 1.5643503665924072, "learning_rate": 1.9581418047101054e-05, "loss": 0.18933849334716796, "step": 24090 }, { "epoch": 0.11970634691992092, "grad_norm": 1.449139952659607, "learning_rate": 1.958095735796121e-05, "loss": 0.1938150405883789, "step": 24100 }, { "epoch": 0.11975601760329019, "grad_norm": 1.2162423133850098, "learning_rate": 1.9580496420870037e-05, "loss": 0.1704463005065918, "step": 24110 }, { "epoch": 0.11980568828665945, "grad_norm": 1.4516212940216064, "learning_rate": 1.958003523583947e-05, "loss": 0.18859364986419677, "step": 24120 }, { "epoch": 0.1198553589700287, "grad_norm": 2.200923442840576, "learning_rate": 1.9579573802881444e-05, "loss": 0.1820320725440979, "step": 24130 }, { "epoch": 0.11990502965339797, "grad_norm": 1.6310079097747803, "learning_rate": 1.95791121220079e-05, "loss": 0.18868690729141235, "step": 24140 }, { "epoch": 0.11995470033676724, "grad_norm": 1.3960374593734741, "learning_rate": 1.9578650193230783e-05, "loss": 0.18405681848526, "step": 24150 }, { "epoch": 0.12000437102013649, "grad_norm": 0.8515486121177673, "learning_rate": 1.9578188016562058e-05, "loss": 0.1917540907859802, "step": 24160 }, { "epoch": 0.12005404170350575, "grad_norm": 2.1857335567474365, "learning_rate": 1.957772559201367e-05, "loss": 0.1935955286026001, "step": 24170 }, { "epoch": 0.12010371238687502, "grad_norm": 1.5948773622512817, "learning_rate": 1.9577262919597598e-05, "loss": 0.18032684326171874, "step": 24180 }, { "epoch": 0.12015338307024429, "grad_norm": 0.8321118950843811, "learning_rate": 1.957679999932581e-05, "loss": 0.18211874961853028, "step": 24190 }, { "epoch": 0.12020305375361354, "grad_norm": 1.5668935775756836, "learning_rate": 1.9576336831210293e-05, "loss": 0.18200173377990722, "step": 24200 }, { "epoch": 0.1202527244369828, "grad_norm": 1.3326034545898438, "learning_rate": 1.9575873415263028e-05, "loss": 0.19127027988433837, "step": 24210 }, { "epoch": 0.12030239512035207, "grad_norm": 1.3024123907089233, "learning_rate": 1.957540975149601e-05, "loss": 0.19514364004135132, "step": 24220 }, { "epoch": 0.12035206580372133, "grad_norm": 1.194207787513733, "learning_rate": 1.957494583992124e-05, "loss": 0.17741596698760986, "step": 24230 }, { "epoch": 0.12040173648709059, "grad_norm": 1.975841760635376, "learning_rate": 1.9574481680550717e-05, "loss": 0.1978473663330078, "step": 24240 }, { "epoch": 0.12045140717045985, "grad_norm": 1.7648584842681885, "learning_rate": 1.9574017273396458e-05, "loss": 0.1685880422592163, "step": 24250 }, { "epoch": 0.12050107785382912, "grad_norm": 2.191663980484009, "learning_rate": 1.9573552618470482e-05, "loss": 0.19047435522079467, "step": 24260 }, { "epoch": 0.12055074853719837, "grad_norm": 1.5767990350723267, "learning_rate": 1.9573087715784814e-05, "loss": 0.18093640804290773, "step": 24270 }, { "epoch": 0.12060041922056763, "grad_norm": 1.6422754526138306, "learning_rate": 1.9572622565351483e-05, "loss": 0.17796944379806517, "step": 24280 }, { "epoch": 0.1206500899039369, "grad_norm": 1.064571499824524, "learning_rate": 1.9572157167182536e-05, "loss": 0.18723645210266113, "step": 24290 }, { "epoch": 0.12069976058730617, "grad_norm": 1.1518495082855225, "learning_rate": 1.9571691521290005e-05, "loss": 0.17929599285125733, "step": 24300 }, { "epoch": 0.12074943127067542, "grad_norm": 1.6318079233169556, "learning_rate": 1.9571225627685947e-05, "loss": 0.1961800217628479, "step": 24310 }, { "epoch": 0.12079910195404468, "grad_norm": 0.8207446336746216, "learning_rate": 1.9570759486382416e-05, "loss": 0.1796501874923706, "step": 24320 }, { "epoch": 0.12084877263741395, "grad_norm": 2.6082520484924316, "learning_rate": 1.957029309739148e-05, "loss": 0.19774868488311767, "step": 24330 }, { "epoch": 0.12089844332078321, "grad_norm": 2.163830280303955, "learning_rate": 1.9569826460725206e-05, "loss": 0.1956491231918335, "step": 24340 }, { "epoch": 0.12094811400415247, "grad_norm": 1.8348870277404785, "learning_rate": 1.956935957639567e-05, "loss": 0.16771278381347657, "step": 24350 }, { "epoch": 0.12099778468752173, "grad_norm": 2.6912291049957275, "learning_rate": 1.9568892444414958e-05, "loss": 0.20311133861541747, "step": 24360 }, { "epoch": 0.121047455370891, "grad_norm": 1.5177767276763916, "learning_rate": 1.9568425064795156e-05, "loss": 0.20398664474487305, "step": 24370 }, { "epoch": 0.12109712605426025, "grad_norm": 1.2379416227340698, "learning_rate": 1.9567957437548363e-05, "loss": 0.20369381904602052, "step": 24380 }, { "epoch": 0.12114679673762951, "grad_norm": 3.1321065425872803, "learning_rate": 1.9567489562686676e-05, "loss": 0.19977829456329346, "step": 24390 }, { "epoch": 0.12119646742099878, "grad_norm": 1.9865819215774536, "learning_rate": 1.9567021440222205e-05, "loss": 0.1970905303955078, "step": 24400 }, { "epoch": 0.12124613810436805, "grad_norm": 0.9788123965263367, "learning_rate": 1.956655307016707e-05, "loss": 0.18570265769958497, "step": 24410 }, { "epoch": 0.1212958087877373, "grad_norm": 0.6209356784820557, "learning_rate": 1.9566084452533385e-05, "loss": 0.18836369514465331, "step": 24420 }, { "epoch": 0.12134547947110656, "grad_norm": 2.269742012023926, "learning_rate": 1.956561558733328e-05, "loss": 0.19275649785995483, "step": 24430 }, { "epoch": 0.12139515015447583, "grad_norm": 0.8769368529319763, "learning_rate": 1.9565146474578892e-05, "loss": 0.2044405460357666, "step": 24440 }, { "epoch": 0.1214448208378451, "grad_norm": 1.0304168462753296, "learning_rate": 1.9564677114282356e-05, "loss": 0.20600299835205077, "step": 24450 }, { "epoch": 0.12149449152121435, "grad_norm": 0.8141943216323853, "learning_rate": 1.9564207506455826e-05, "loss": 0.198772394657135, "step": 24460 }, { "epoch": 0.12154416220458361, "grad_norm": 1.5762808322906494, "learning_rate": 1.956373765111145e-05, "loss": 0.18297485113143921, "step": 24470 }, { "epoch": 0.12159383288795288, "grad_norm": 3.404141664505005, "learning_rate": 1.956326754826139e-05, "loss": 0.1880058765411377, "step": 24480 }, { "epoch": 0.12164350357132213, "grad_norm": 1.4367761611938477, "learning_rate": 1.9562797197917807e-05, "loss": 0.1842369556427002, "step": 24490 }, { "epoch": 0.1216931742546914, "grad_norm": 1.5341520309448242, "learning_rate": 1.956232660009288e-05, "loss": 0.1883604645729065, "step": 24500 }, { "epoch": 0.12174284493806066, "grad_norm": 3.608433961868286, "learning_rate": 1.9561855754798788e-05, "loss": 0.2107468605041504, "step": 24510 }, { "epoch": 0.12179251562142993, "grad_norm": 1.2251427173614502, "learning_rate": 1.9561384662047713e-05, "loss": 0.17674767971038818, "step": 24520 }, { "epoch": 0.12184218630479918, "grad_norm": 1.5233210325241089, "learning_rate": 1.9560913321851842e-05, "loss": 0.16712779998779298, "step": 24530 }, { "epoch": 0.12189185698816844, "grad_norm": 1.1125683784484863, "learning_rate": 1.9560441734223384e-05, "loss": 0.19830708503723143, "step": 24540 }, { "epoch": 0.12194152767153771, "grad_norm": 2.3794262409210205, "learning_rate": 1.9559969899174535e-05, "loss": 0.1756657123565674, "step": 24550 }, { "epoch": 0.12199119835490696, "grad_norm": 0.8258947730064392, "learning_rate": 1.955949781671751e-05, "loss": 0.1822911262512207, "step": 24560 }, { "epoch": 0.12204086903827623, "grad_norm": 0.8388886451721191, "learning_rate": 1.9559025486864528e-05, "loss": 0.19452788829803466, "step": 24570 }, { "epoch": 0.12209053972164549, "grad_norm": 0.8582034707069397, "learning_rate": 1.9558552909627806e-05, "loss": 0.1848160743713379, "step": 24580 }, { "epoch": 0.12214021040501476, "grad_norm": 1.674533724784851, "learning_rate": 1.955808008501958e-05, "loss": 0.17570300102233888, "step": 24590 }, { "epoch": 0.12218988108838401, "grad_norm": 0.8520254492759705, "learning_rate": 1.9557607013052083e-05, "loss": 0.19293713569641113, "step": 24600 }, { "epoch": 0.12223955177175327, "grad_norm": 0.9752630591392517, "learning_rate": 1.955713369373756e-05, "loss": 0.19870003461837768, "step": 24610 }, { "epoch": 0.12228922245512254, "grad_norm": 1.2119956016540527, "learning_rate": 1.955666012708826e-05, "loss": 0.19078969955444336, "step": 24620 }, { "epoch": 0.1223388931384918, "grad_norm": 1.4463611841201782, "learning_rate": 1.9556186313116438e-05, "loss": 0.1880630612373352, "step": 24630 }, { "epoch": 0.12238856382186106, "grad_norm": 1.2636672258377075, "learning_rate": 1.955571225183436e-05, "loss": 0.17415027618408202, "step": 24640 }, { "epoch": 0.12243823450523032, "grad_norm": 1.8700631856918335, "learning_rate": 1.9555237943254287e-05, "loss": 0.17871477603912353, "step": 24650 }, { "epoch": 0.12248790518859959, "grad_norm": 1.5602753162384033, "learning_rate": 1.95547633873885e-05, "loss": 0.18316450119018554, "step": 24660 }, { "epoch": 0.12253757587196884, "grad_norm": 1.438347339630127, "learning_rate": 1.9554288584249277e-05, "loss": 0.1712714910507202, "step": 24670 }, { "epoch": 0.1225872465553381, "grad_norm": 1.7354774475097656, "learning_rate": 1.955381353384891e-05, "loss": 0.19079415798187255, "step": 24680 }, { "epoch": 0.12263691723870737, "grad_norm": 1.0032504796981812, "learning_rate": 1.955333823619969e-05, "loss": 0.1811740517616272, "step": 24690 }, { "epoch": 0.12268658792207664, "grad_norm": 2.140146493911743, "learning_rate": 1.9552862691313912e-05, "loss": 0.18911888599395751, "step": 24700 }, { "epoch": 0.12273625860544589, "grad_norm": 0.5298312902450562, "learning_rate": 1.9552386899203893e-05, "loss": 0.19161882400512695, "step": 24710 }, { "epoch": 0.12278592928881515, "grad_norm": 1.086661696434021, "learning_rate": 1.955191085988194e-05, "loss": 0.18915946483612062, "step": 24720 }, { "epoch": 0.12283559997218442, "grad_norm": 1.5677491426467896, "learning_rate": 1.9551434573360377e-05, "loss": 0.16923964023590088, "step": 24730 }, { "epoch": 0.12288527065555369, "grad_norm": 1.2547255754470825, "learning_rate": 1.9550958039651528e-05, "loss": 0.1835004448890686, "step": 24740 }, { "epoch": 0.12293494133892294, "grad_norm": 2.292509078979492, "learning_rate": 1.9550481258767725e-05, "loss": 0.2084214210510254, "step": 24750 }, { "epoch": 0.1229846120222922, "grad_norm": 0.8368115425109863, "learning_rate": 1.9550004230721304e-05, "loss": 0.16781712770462037, "step": 24760 }, { "epoch": 0.12303428270566147, "grad_norm": 2.9055721759796143, "learning_rate": 1.9549526955524614e-05, "loss": 0.1891080141067505, "step": 24770 }, { "epoch": 0.12308395338903072, "grad_norm": 1.3496136665344238, "learning_rate": 1.954904943319001e-05, "loss": 0.18286962509155275, "step": 24780 }, { "epoch": 0.12313362407239999, "grad_norm": 1.9701900482177734, "learning_rate": 1.9548571663729842e-05, "loss": 0.1736815333366394, "step": 24790 }, { "epoch": 0.12318329475576925, "grad_norm": 1.8333592414855957, "learning_rate": 1.954809364715648e-05, "loss": 0.17102304697036744, "step": 24800 }, { "epoch": 0.12323296543913852, "grad_norm": 3.938786268234253, "learning_rate": 1.9547615383482295e-05, "loss": 0.19700700044631958, "step": 24810 }, { "epoch": 0.12328263612250777, "grad_norm": 2.0009872913360596, "learning_rate": 1.954713687271966e-05, "loss": 0.17840280532836914, "step": 24820 }, { "epoch": 0.12333230680587703, "grad_norm": 1.4747549295425415, "learning_rate": 1.9546658114880963e-05, "loss": 0.18118603229522706, "step": 24830 }, { "epoch": 0.1233819774892463, "grad_norm": 1.4576209783554077, "learning_rate": 1.9546179109978594e-05, "loss": 0.18195006847381592, "step": 24840 }, { "epoch": 0.12343164817261557, "grad_norm": 0.9785471558570862, "learning_rate": 1.9545699858024947e-05, "loss": 0.18913910388946534, "step": 24850 }, { "epoch": 0.12348131885598482, "grad_norm": 0.9425318837165833, "learning_rate": 1.9545220359032426e-05, "loss": 0.17916051149368287, "step": 24860 }, { "epoch": 0.12353098953935408, "grad_norm": 0.7382171750068665, "learning_rate": 1.954474061301344e-05, "loss": 0.1757920503616333, "step": 24870 }, { "epoch": 0.12358066022272335, "grad_norm": 1.1995320320129395, "learning_rate": 1.9544260619980405e-05, "loss": 0.18234061002731322, "step": 24880 }, { "epoch": 0.1236303309060926, "grad_norm": 2.1375062465667725, "learning_rate": 1.9543780379945742e-05, "loss": 0.18652987480163574, "step": 24890 }, { "epoch": 0.12368000158946187, "grad_norm": 0.9377221465110779, "learning_rate": 1.954329989292188e-05, "loss": 0.18693358898162843, "step": 24900 }, { "epoch": 0.12372967227283113, "grad_norm": 1.009663462638855, "learning_rate": 1.9542819158921254e-05, "loss": 0.18861489295959472, "step": 24910 }, { "epoch": 0.1237793429562004, "grad_norm": 0.6353587508201599, "learning_rate": 1.9542338177956304e-05, "loss": 0.17526423931121826, "step": 24920 }, { "epoch": 0.12382901363956965, "grad_norm": 1.622686505317688, "learning_rate": 1.954185695003948e-05, "loss": 0.1930422306060791, "step": 24930 }, { "epoch": 0.12387868432293891, "grad_norm": 0.5600825548171997, "learning_rate": 1.9541375475183234e-05, "loss": 0.20340354442596437, "step": 24940 }, { "epoch": 0.12392835500630818, "grad_norm": 1.4393987655639648, "learning_rate": 1.954089375340003e-05, "loss": 0.1720402479171753, "step": 24950 }, { "epoch": 0.12397802568967743, "grad_norm": 1.1496905088424683, "learning_rate": 1.9540411784702327e-05, "loss": 0.18636738061904906, "step": 24960 }, { "epoch": 0.1240276963730467, "grad_norm": 3.294520378112793, "learning_rate": 1.9539929569102608e-05, "loss": 0.17509300708770753, "step": 24970 }, { "epoch": 0.12407736705641596, "grad_norm": 2.4679625034332275, "learning_rate": 1.9539447106613347e-05, "loss": 0.1798289895057678, "step": 24980 }, { "epoch": 0.12412703773978523, "grad_norm": 1.0135732889175415, "learning_rate": 1.9538964397247026e-05, "loss": 0.1813367486000061, "step": 24990 }, { "epoch": 0.12417670842315448, "grad_norm": 2.4640707969665527, "learning_rate": 1.9538481441016145e-05, "loss": 0.171522057056427, "step": 25000 }, { "epoch": 0.12422637910652375, "grad_norm": 2.6457679271698, "learning_rate": 1.95379982379332e-05, "loss": 0.18341193199157715, "step": 25010 }, { "epoch": 0.12427604978989301, "grad_norm": 0.9336788058280945, "learning_rate": 1.9537514788010698e-05, "loss": 0.18451881408691406, "step": 25020 }, { "epoch": 0.12432572047326228, "grad_norm": 0.9764851331710815, "learning_rate": 1.9537031091261143e-05, "loss": 0.1997664213180542, "step": 25030 }, { "epoch": 0.12437539115663153, "grad_norm": 2.030061960220337, "learning_rate": 1.953654714769706e-05, "loss": 0.18549470901489257, "step": 25040 }, { "epoch": 0.1244250618400008, "grad_norm": 0.8543990850448608, "learning_rate": 1.9536062957330972e-05, "loss": 0.19239048957824706, "step": 25050 }, { "epoch": 0.12447473252337006, "grad_norm": 1.5907236337661743, "learning_rate": 1.9535578520175408e-05, "loss": 0.17349159717559814, "step": 25060 }, { "epoch": 0.12452440320673931, "grad_norm": 1.24094820022583, "learning_rate": 1.9535093836242906e-05, "loss": 0.1893258810043335, "step": 25070 }, { "epoch": 0.12457407389010858, "grad_norm": 1.0864202976226807, "learning_rate": 1.953460890554601e-05, "loss": 0.1868030309677124, "step": 25080 }, { "epoch": 0.12462374457347784, "grad_norm": 1.491695761680603, "learning_rate": 1.953412372809727e-05, "loss": 0.18290945291519164, "step": 25090 }, { "epoch": 0.12467341525684711, "grad_norm": 2.588937282562256, "learning_rate": 1.953363830390924e-05, "loss": 0.18879176378250123, "step": 25100 }, { "epoch": 0.12472308594021636, "grad_norm": 2.2654714584350586, "learning_rate": 1.9533152632994484e-05, "loss": 0.18568880558013917, "step": 25110 }, { "epoch": 0.12477275662358563, "grad_norm": 1.7858169078826904, "learning_rate": 1.9532666715365572e-05, "loss": 0.18565300703048707, "step": 25120 }, { "epoch": 0.12482242730695489, "grad_norm": 1.0824743509292603, "learning_rate": 1.953218055103508e-05, "loss": 0.18573291301727296, "step": 25130 }, { "epoch": 0.12487209799032416, "grad_norm": 2.387946605682373, "learning_rate": 1.9531694140015585e-05, "loss": 0.1967448592185974, "step": 25140 }, { "epoch": 0.12492176867369341, "grad_norm": 2.5556588172912598, "learning_rate": 1.9531207482319682e-05, "loss": 0.19199196100234986, "step": 25150 }, { "epoch": 0.12497143935706267, "grad_norm": 0.668637752532959, "learning_rate": 1.953072057795996e-05, "loss": 0.174690318107605, "step": 25160 }, { "epoch": 0.12502111004043193, "grad_norm": 2.2791383266448975, "learning_rate": 1.953023342694902e-05, "loss": 0.18993587493896485, "step": 25170 }, { "epoch": 0.1250707807238012, "grad_norm": 2.332188129425049, "learning_rate": 1.952974602929947e-05, "loss": 0.20028886795043946, "step": 25180 }, { "epoch": 0.12512045140717046, "grad_norm": 1.1580696105957031, "learning_rate": 1.9529258385023927e-05, "loss": 0.17487542629241942, "step": 25190 }, { "epoch": 0.1251701220905397, "grad_norm": 1.178222894668579, "learning_rate": 1.9528770494135006e-05, "loss": 0.18783512115478515, "step": 25200 }, { "epoch": 0.125219792773909, "grad_norm": 1.0008350610733032, "learning_rate": 1.9528282356645335e-05, "loss": 0.18894436359405517, "step": 25210 }, { "epoch": 0.12526946345727824, "grad_norm": 1.845837950706482, "learning_rate": 1.9527793972567554e-05, "loss": 0.18888394832611083, "step": 25220 }, { "epoch": 0.12531913414064752, "grad_norm": 1.5726747512817383, "learning_rate": 1.952730534191429e-05, "loss": 0.18343656063079833, "step": 25230 }, { "epoch": 0.12536880482401677, "grad_norm": 1.89307701587677, "learning_rate": 1.9526816464698197e-05, "loss": 0.1903966784477234, "step": 25240 }, { "epoch": 0.12541847550738602, "grad_norm": 1.6045113801956177, "learning_rate": 1.9526327340931927e-05, "loss": 0.17767175436019897, "step": 25250 }, { "epoch": 0.1254681461907553, "grad_norm": 2.67881178855896, "learning_rate": 1.952583797062813e-05, "loss": 0.1804807424545288, "step": 25260 }, { "epoch": 0.12551781687412455, "grad_norm": 0.8073554635047913, "learning_rate": 1.952534835379948e-05, "loss": 0.18936911821365357, "step": 25270 }, { "epoch": 0.1255674875574938, "grad_norm": 1.3931061029434204, "learning_rate": 1.9524858490458645e-05, "loss": 0.19412082433700562, "step": 25280 }, { "epoch": 0.12561715824086309, "grad_norm": 1.166000485420227, "learning_rate": 1.95243683806183e-05, "loss": 0.17175556421279908, "step": 25290 }, { "epoch": 0.12566682892423234, "grad_norm": 0.8547326326370239, "learning_rate": 1.952387802429113e-05, "loss": 0.18300895690917968, "step": 25300 }, { "epoch": 0.1257164996076016, "grad_norm": 2.2754721641540527, "learning_rate": 1.952338742148983e-05, "loss": 0.17656872272491456, "step": 25310 }, { "epoch": 0.12576617029097087, "grad_norm": 1.6367229223251343, "learning_rate": 1.9522896572227092e-05, "loss": 0.185172438621521, "step": 25320 }, { "epoch": 0.12581584097434012, "grad_norm": 2.2856202125549316, "learning_rate": 1.9522405476515618e-05, "loss": 0.1973961114883423, "step": 25330 }, { "epoch": 0.1258655116577094, "grad_norm": 1.2023004293441772, "learning_rate": 1.952191413436812e-05, "loss": 0.19744863510131835, "step": 25340 }, { "epoch": 0.12591518234107865, "grad_norm": 2.3684399127960205, "learning_rate": 1.9521422545797314e-05, "loss": 0.18731679916381835, "step": 25350 }, { "epoch": 0.1259648530244479, "grad_norm": 0.7990080118179321, "learning_rate": 1.9520930710815918e-05, "loss": 0.17192540168762208, "step": 25360 }, { "epoch": 0.12601452370781718, "grad_norm": 1.1099478006362915, "learning_rate": 1.9520438629436666e-05, "loss": 0.17725423574447632, "step": 25370 }, { "epoch": 0.12606419439118643, "grad_norm": 0.8108840584754944, "learning_rate": 1.9519946301672287e-05, "loss": 0.19604002237319945, "step": 25380 }, { "epoch": 0.1261138650745557, "grad_norm": 0.8882187008857727, "learning_rate": 1.9519453727535528e-05, "loss": 0.1813154935836792, "step": 25390 }, { "epoch": 0.12616353575792497, "grad_norm": 0.6524117588996887, "learning_rate": 1.9518960907039134e-05, "loss": 0.1851473331451416, "step": 25400 }, { "epoch": 0.12621320644129422, "grad_norm": 1.2740418910980225, "learning_rate": 1.951846784019586e-05, "loss": 0.1799706220626831, "step": 25410 }, { "epoch": 0.12626287712466347, "grad_norm": 1.308714747428894, "learning_rate": 1.9517974527018463e-05, "loss": 0.17280218601226807, "step": 25420 }, { "epoch": 0.12631254780803275, "grad_norm": 2.4389350414276123, "learning_rate": 1.9517480967519714e-05, "loss": 0.19020360708236694, "step": 25430 }, { "epoch": 0.126362218491402, "grad_norm": 1.472381591796875, "learning_rate": 1.9516987161712383e-05, "loss": 0.19797301292419434, "step": 25440 }, { "epoch": 0.12641188917477128, "grad_norm": 0.6900410056114197, "learning_rate": 1.9516493109609254e-05, "loss": 0.17459309101104736, "step": 25450 }, { "epoch": 0.12646155985814053, "grad_norm": 1.0121641159057617, "learning_rate": 1.9515998811223107e-05, "loss": 0.19567179679870605, "step": 25460 }, { "epoch": 0.12651123054150978, "grad_norm": 1.6396782398223877, "learning_rate": 1.9515504266566738e-05, "loss": 0.18849350214004518, "step": 25470 }, { "epoch": 0.12656090122487906, "grad_norm": 1.2819368839263916, "learning_rate": 1.951500947565294e-05, "loss": 0.1976207137107849, "step": 25480 }, { "epoch": 0.12661057190824831, "grad_norm": 0.8514979481697083, "learning_rate": 1.951451443849453e-05, "loss": 0.19498856067657472, "step": 25490 }, { "epoch": 0.12666024259161757, "grad_norm": 0.5379432439804077, "learning_rate": 1.9514019155104308e-05, "loss": 0.17494715452194215, "step": 25500 }, { "epoch": 0.12670991327498685, "grad_norm": 1.2283424139022827, "learning_rate": 1.9513523625495097e-05, "loss": 0.19759546518325805, "step": 25510 }, { "epoch": 0.1267595839583561, "grad_norm": 1.1326913833618164, "learning_rate": 1.951302784967972e-05, "loss": 0.1658935785293579, "step": 25520 }, { "epoch": 0.12680925464172535, "grad_norm": 1.3716425895690918, "learning_rate": 1.9512531827671005e-05, "loss": 0.19421976804733276, "step": 25530 }, { "epoch": 0.12685892532509463, "grad_norm": 1.4356731176376343, "learning_rate": 1.9512035559481794e-05, "loss": 0.17486640214920043, "step": 25540 }, { "epoch": 0.12690859600846388, "grad_norm": 2.8113160133361816, "learning_rate": 1.951153904512493e-05, "loss": 0.1694880247116089, "step": 25550 }, { "epoch": 0.12695826669183313, "grad_norm": 1.1121917963027954, "learning_rate": 1.9511042284613253e-05, "loss": 0.17749483585357667, "step": 25560 }, { "epoch": 0.1270079373752024, "grad_norm": 2.6327056884765625, "learning_rate": 1.9510545277959627e-05, "loss": 0.1816692590713501, "step": 25570 }, { "epoch": 0.12705760805857166, "grad_norm": 1.1260406970977783, "learning_rate": 1.951004802517692e-05, "loss": 0.17650550603866577, "step": 25580 }, { "epoch": 0.12710727874194094, "grad_norm": 1.945999026298523, "learning_rate": 1.9509550526277986e-05, "loss": 0.19156792163848876, "step": 25590 }, { "epoch": 0.1271569494253102, "grad_norm": 1.961822271347046, "learning_rate": 1.950905278127571e-05, "loss": 0.18816424608230592, "step": 25600 }, { "epoch": 0.12720662010867945, "grad_norm": 0.8168145418167114, "learning_rate": 1.9508554790182975e-05, "loss": 0.19784563779830933, "step": 25610 }, { "epoch": 0.12725629079204873, "grad_norm": 0.624488115310669, "learning_rate": 1.9508056553012663e-05, "loss": 0.1635555386543274, "step": 25620 }, { "epoch": 0.12730596147541798, "grad_norm": 1.1493313312530518, "learning_rate": 1.9507558069777667e-05, "loss": 0.185857892036438, "step": 25630 }, { "epoch": 0.12735563215878723, "grad_norm": 2.4039509296417236, "learning_rate": 1.9507059340490893e-05, "loss": 0.19184470176696777, "step": 25640 }, { "epoch": 0.1274053028421565, "grad_norm": 1.50118887424469, "learning_rate": 1.950656036516525e-05, "loss": 0.1927577018737793, "step": 25650 }, { "epoch": 0.12745497352552576, "grad_norm": 1.0755677223205566, "learning_rate": 1.9506061143813644e-05, "loss": 0.1840598464012146, "step": 25660 }, { "epoch": 0.127504644208895, "grad_norm": 1.1641929149627686, "learning_rate": 1.9505561676448997e-05, "loss": 0.18165152072906493, "step": 25670 }, { "epoch": 0.1275543148922643, "grad_norm": 1.0274910926818848, "learning_rate": 1.9505061963084237e-05, "loss": 0.17764295339584352, "step": 25680 }, { "epoch": 0.12760398557563354, "grad_norm": 0.7247384786605835, "learning_rate": 1.9504562003732297e-05, "loss": 0.19043749570846558, "step": 25690 }, { "epoch": 0.12765365625900282, "grad_norm": 1.898916244506836, "learning_rate": 1.9504061798406112e-05, "loss": 0.1758289098739624, "step": 25700 }, { "epoch": 0.12770332694237208, "grad_norm": 1.7488815784454346, "learning_rate": 1.950356134711863e-05, "loss": 0.18619803190231324, "step": 25710 }, { "epoch": 0.12775299762574133, "grad_norm": 0.9969421625137329, "learning_rate": 1.95030606498828e-05, "loss": 0.19529296159744264, "step": 25720 }, { "epoch": 0.1278026683091106, "grad_norm": 1.9753841161727905, "learning_rate": 1.9502559706711582e-05, "loss": 0.18590245246887208, "step": 25730 }, { "epoch": 0.12785233899247986, "grad_norm": 1.521745204925537, "learning_rate": 1.950205851761794e-05, "loss": 0.17649370431900024, "step": 25740 }, { "epoch": 0.1279020096758491, "grad_norm": 1.483282446861267, "learning_rate": 1.9501557082614842e-05, "loss": 0.1884901523590088, "step": 25750 }, { "epoch": 0.1279516803592184, "grad_norm": 1.2398581504821777, "learning_rate": 1.950105540171527e-05, "loss": 0.1839911460876465, "step": 25760 }, { "epoch": 0.12800135104258764, "grad_norm": 2.9560859203338623, "learning_rate": 1.9500553474932206e-05, "loss": 0.1891420602798462, "step": 25770 }, { "epoch": 0.1280510217259569, "grad_norm": 2.3769826889038086, "learning_rate": 1.9500051302278634e-05, "loss": 0.1759091019630432, "step": 25780 }, { "epoch": 0.12810069240932617, "grad_norm": 0.6679971814155579, "learning_rate": 1.9499548883767556e-05, "loss": 0.18008124828338623, "step": 25790 }, { "epoch": 0.12815036309269542, "grad_norm": 2.2082507610321045, "learning_rate": 1.9499046219411972e-05, "loss": 0.17669832706451416, "step": 25800 }, { "epoch": 0.1282000337760647, "grad_norm": 1.7688103914260864, "learning_rate": 1.9498543309224892e-05, "loss": 0.17496926784515382, "step": 25810 }, { "epoch": 0.12824970445943396, "grad_norm": 1.794503927230835, "learning_rate": 1.949804015321933e-05, "loss": 0.17258858680725098, "step": 25820 }, { "epoch": 0.1282993751428032, "grad_norm": 1.53024160861969, "learning_rate": 1.9497536751408307e-05, "loss": 0.19259827136993407, "step": 25830 }, { "epoch": 0.1283490458261725, "grad_norm": 1.5024571418762207, "learning_rate": 1.9497033103804853e-05, "loss": 0.20563864707946777, "step": 25840 }, { "epoch": 0.12839871650954174, "grad_norm": 2.467923879623413, "learning_rate": 1.9496529210422e-05, "loss": 0.19541873931884765, "step": 25850 }, { "epoch": 0.128448387192911, "grad_norm": 2.303701162338257, "learning_rate": 1.949602507127279e-05, "loss": 0.1810162305831909, "step": 25860 }, { "epoch": 0.12849805787628027, "grad_norm": 1.6474153995513916, "learning_rate": 1.9495520686370267e-05, "loss": 0.17819219827651978, "step": 25870 }, { "epoch": 0.12854772855964952, "grad_norm": 1.1944217681884766, "learning_rate": 1.9495016055727493e-05, "loss": 0.18093361854553222, "step": 25880 }, { "epoch": 0.12859739924301877, "grad_norm": 1.451909065246582, "learning_rate": 1.9494511179357513e-05, "loss": 0.20087723731994628, "step": 25890 }, { "epoch": 0.12864706992638805, "grad_norm": 1.2004928588867188, "learning_rate": 1.9494006057273406e-05, "loss": 0.17527263164520263, "step": 25900 }, { "epoch": 0.1286967406097573, "grad_norm": 3.8468878269195557, "learning_rate": 1.949350068948824e-05, "loss": 0.20037999153137206, "step": 25910 }, { "epoch": 0.12874641129312658, "grad_norm": 1.720215082168579, "learning_rate": 1.9492995076015092e-05, "loss": 0.19442256689071655, "step": 25920 }, { "epoch": 0.12879608197649584, "grad_norm": 0.921241044998169, "learning_rate": 1.949248921686705e-05, "loss": 0.17064764499664306, "step": 25930 }, { "epoch": 0.1288457526598651, "grad_norm": 1.3597865104675293, "learning_rate": 1.94919831120572e-05, "loss": 0.19277026653289794, "step": 25940 }, { "epoch": 0.12889542334323437, "grad_norm": 2.05363130569458, "learning_rate": 1.9491476761598645e-05, "loss": 0.1757202625274658, "step": 25950 }, { "epoch": 0.12894509402660362, "grad_norm": 0.7132987976074219, "learning_rate": 1.949097016550449e-05, "loss": 0.17037789821624755, "step": 25960 }, { "epoch": 0.12899476470997287, "grad_norm": 1.0597401857376099, "learning_rate": 1.9490463323787842e-05, "loss": 0.18681492805480956, "step": 25970 }, { "epoch": 0.12904443539334215, "grad_norm": 1.9050014019012451, "learning_rate": 1.948995623646182e-05, "loss": 0.19436774253845215, "step": 25980 }, { "epoch": 0.1290941060767114, "grad_norm": 1.6064008474349976, "learning_rate": 1.9489448903539547e-05, "loss": 0.19692869186401368, "step": 25990 }, { "epoch": 0.12914377676008065, "grad_norm": 2.9602887630462646, "learning_rate": 1.9488941325034146e-05, "loss": 0.18151957988739015, "step": 26000 }, { "epoch": 0.12919344744344993, "grad_norm": 1.3965485095977783, "learning_rate": 1.9488433500958766e-05, "loss": 0.18471647500991822, "step": 26010 }, { "epoch": 0.12924311812681918, "grad_norm": 1.438934564590454, "learning_rate": 1.948792543132654e-05, "loss": 0.1935628056526184, "step": 26020 }, { "epoch": 0.12929278881018846, "grad_norm": 1.4498337507247925, "learning_rate": 1.948741711615062e-05, "loss": 0.18100643157958984, "step": 26030 }, { "epoch": 0.12934245949355772, "grad_norm": 1.326333999633789, "learning_rate": 1.9486908555444156e-05, "loss": 0.17353312969207763, "step": 26040 }, { "epoch": 0.12939213017692697, "grad_norm": 1.5699498653411865, "learning_rate": 1.9486399749220314e-05, "loss": 0.17643091678619385, "step": 26050 }, { "epoch": 0.12944180086029625, "grad_norm": 0.685492992401123, "learning_rate": 1.9485890697492264e-05, "loss": 0.18453700542449952, "step": 26060 }, { "epoch": 0.1294914715436655, "grad_norm": 1.565277099609375, "learning_rate": 1.9485381400273174e-05, "loss": 0.18718069791793823, "step": 26070 }, { "epoch": 0.12954114222703475, "grad_norm": 1.1998103857040405, "learning_rate": 1.948487185757623e-05, "loss": 0.19197913408279418, "step": 26080 }, { "epoch": 0.12959081291040403, "grad_norm": 3.5168392658233643, "learning_rate": 1.948436206941461e-05, "loss": 0.18829047679901123, "step": 26090 }, { "epoch": 0.12964048359377328, "grad_norm": 2.8872714042663574, "learning_rate": 1.948385203580152e-05, "loss": 0.19171457290649413, "step": 26100 }, { "epoch": 0.12969015427714253, "grad_norm": 2.657898426055908, "learning_rate": 1.948334175675015e-05, "loss": 0.17886769771575928, "step": 26110 }, { "epoch": 0.1297398249605118, "grad_norm": 1.7119404077529907, "learning_rate": 1.9482831232273708e-05, "loss": 0.18566470146179198, "step": 26120 }, { "epoch": 0.12978949564388106, "grad_norm": 1.3128217458724976, "learning_rate": 1.9482320462385406e-05, "loss": 0.1832722783088684, "step": 26130 }, { "epoch": 0.12983916632725034, "grad_norm": 1.7583997249603271, "learning_rate": 1.9481809447098462e-05, "loss": 0.19145675897598266, "step": 26140 }, { "epoch": 0.1298888370106196, "grad_norm": 2.0482122898101807, "learning_rate": 1.9481298186426104e-05, "loss": 0.1807037591934204, "step": 26150 }, { "epoch": 0.12993850769398885, "grad_norm": 1.6174839735031128, "learning_rate": 1.9480786680381563e-05, "loss": 0.19618659019470214, "step": 26160 }, { "epoch": 0.12998817837735813, "grad_norm": 1.683370590209961, "learning_rate": 1.9480274928978075e-05, "loss": 0.17468523979187012, "step": 26170 }, { "epoch": 0.13003784906072738, "grad_norm": 1.7345644235610962, "learning_rate": 1.9479762932228882e-05, "loss": 0.19266024827957154, "step": 26180 }, { "epoch": 0.13008751974409663, "grad_norm": 2.523955821990967, "learning_rate": 1.9479250690147234e-05, "loss": 0.1872931480407715, "step": 26190 }, { "epoch": 0.1301371904274659, "grad_norm": 0.9234303832054138, "learning_rate": 1.947873820274639e-05, "loss": 0.16667643785476685, "step": 26200 }, { "epoch": 0.13018686111083516, "grad_norm": 2.02524995803833, "learning_rate": 1.9478225470039617e-05, "loss": 0.19696085453033446, "step": 26210 }, { "epoch": 0.1302365317942044, "grad_norm": 2.708911657333374, "learning_rate": 1.9477712492040176e-05, "loss": 0.1809716820716858, "step": 26220 }, { "epoch": 0.1302862024775737, "grad_norm": 0.6321983933448792, "learning_rate": 1.9477199268761346e-05, "loss": 0.1794722318649292, "step": 26230 }, { "epoch": 0.13033587316094294, "grad_norm": 0.7431910634040833, "learning_rate": 1.947668580021641e-05, "loss": 0.18345675468444825, "step": 26240 }, { "epoch": 0.13038554384431222, "grad_norm": 0.8079444169998169, "learning_rate": 1.947617208641866e-05, "loss": 0.17594695091247559, "step": 26250 }, { "epoch": 0.13043521452768148, "grad_norm": 1.7786582708358765, "learning_rate": 1.947565812738138e-05, "loss": 0.1881725788116455, "step": 26260 }, { "epoch": 0.13048488521105073, "grad_norm": 0.7957118153572083, "learning_rate": 1.9475143923117882e-05, "loss": 0.190306556224823, "step": 26270 }, { "epoch": 0.13053455589442, "grad_norm": 0.9919576048851013, "learning_rate": 1.9474629473641465e-05, "loss": 0.1809662938117981, "step": 26280 }, { "epoch": 0.13058422657778926, "grad_norm": 1.6745150089263916, "learning_rate": 1.9474114778965453e-05, "loss": 0.18087844848632811, "step": 26290 }, { "epoch": 0.1306338972611585, "grad_norm": 2.3069939613342285, "learning_rate": 1.9473599839103157e-05, "loss": 0.19171500205993652, "step": 26300 }, { "epoch": 0.1306835679445278, "grad_norm": 2.2740907669067383, "learning_rate": 1.9473084654067906e-05, "loss": 0.1930774927139282, "step": 26310 }, { "epoch": 0.13073323862789704, "grad_norm": 1.9022527933120728, "learning_rate": 1.9472569223873033e-05, "loss": 0.18585677146911622, "step": 26320 }, { "epoch": 0.1307829093112663, "grad_norm": 1.552340030670166, "learning_rate": 1.9472053548531876e-05, "loss": 0.17796804904937744, "step": 26330 }, { "epoch": 0.13083257999463557, "grad_norm": 2.701409101486206, "learning_rate": 1.9471537628057782e-05, "loss": 0.17731316089630128, "step": 26340 }, { "epoch": 0.13088225067800482, "grad_norm": 1.3213268518447876, "learning_rate": 1.9471021462464107e-05, "loss": 0.2019355297088623, "step": 26350 }, { "epoch": 0.13093192136137408, "grad_norm": 3.0024101734161377, "learning_rate": 1.94705050517642e-05, "loss": 0.16947929859161376, "step": 26360 }, { "epoch": 0.13098159204474336, "grad_norm": 0.8466993570327759, "learning_rate": 1.9469988395971432e-05, "loss": 0.2012331008911133, "step": 26370 }, { "epoch": 0.1310312627281126, "grad_norm": 1.219797134399414, "learning_rate": 1.946947149509917e-05, "loss": 0.18334053754806517, "step": 26380 }, { "epoch": 0.1310809334114819, "grad_norm": 2.1429429054260254, "learning_rate": 1.9468954349160798e-05, "loss": 0.1886416792869568, "step": 26390 }, { "epoch": 0.13113060409485114, "grad_norm": 0.7721337080001831, "learning_rate": 1.946843695816969e-05, "loss": 0.18134772777557373, "step": 26400 }, { "epoch": 0.1311802747782204, "grad_norm": 1.7877737283706665, "learning_rate": 1.9467919322139243e-05, "loss": 0.19802916049957275, "step": 26410 }, { "epoch": 0.13122994546158967, "grad_norm": 1.785975456237793, "learning_rate": 1.9467401441082848e-05, "loss": 0.1782430648803711, "step": 26420 }, { "epoch": 0.13127961614495892, "grad_norm": 1.0670865774154663, "learning_rate": 1.9466883315013913e-05, "loss": 0.1902111291885376, "step": 26430 }, { "epoch": 0.13132928682832817, "grad_norm": 1.6673636436462402, "learning_rate": 1.9466364943945845e-05, "loss": 0.1853633403778076, "step": 26440 }, { "epoch": 0.13137895751169745, "grad_norm": 2.1194827556610107, "learning_rate": 1.9465846327892056e-05, "loss": 0.19418785572052003, "step": 26450 }, { "epoch": 0.1314286281950667, "grad_norm": 0.6838322281837463, "learning_rate": 1.9465327466865972e-05, "loss": 0.19901182651519775, "step": 26460 }, { "epoch": 0.13147829887843596, "grad_norm": 1.6528830528259277, "learning_rate": 1.946480836088102e-05, "loss": 0.18127880096435547, "step": 26470 }, { "epoch": 0.13152796956180524, "grad_norm": 2.2023017406463623, "learning_rate": 1.946428900995063e-05, "loss": 0.18760335445404053, "step": 26480 }, { "epoch": 0.1315776402451745, "grad_norm": 1.6911451816558838, "learning_rate": 1.9463769414088248e-05, "loss": 0.18640742301940919, "step": 26490 }, { "epoch": 0.13162731092854377, "grad_norm": 2.4942145347595215, "learning_rate": 1.946324957330732e-05, "loss": 0.18357563018798828, "step": 26500 }, { "epoch": 0.13167698161191302, "grad_norm": 0.9400508403778076, "learning_rate": 1.9462729487621294e-05, "loss": 0.17600784301757813, "step": 26510 }, { "epoch": 0.13172665229528227, "grad_norm": 1.773156762123108, "learning_rate": 1.9462209157043638e-05, "loss": 0.18682587146759033, "step": 26520 }, { "epoch": 0.13177632297865155, "grad_norm": 0.9558724164962769, "learning_rate": 1.9461688581587808e-05, "loss": 0.18591320514678955, "step": 26530 }, { "epoch": 0.1318259936620208, "grad_norm": 1.0593382120132446, "learning_rate": 1.946116776126729e-05, "loss": 0.19278466701507568, "step": 26540 }, { "epoch": 0.13187566434539005, "grad_norm": 1.3216347694396973, "learning_rate": 1.946064669609555e-05, "loss": 0.1771111249923706, "step": 26550 }, { "epoch": 0.13192533502875933, "grad_norm": 2.7970330715179443, "learning_rate": 1.9460125386086076e-05, "loss": 0.1932147741317749, "step": 26560 }, { "epoch": 0.13197500571212858, "grad_norm": 1.178720474243164, "learning_rate": 1.945960383125236e-05, "loss": 0.18497810363769532, "step": 26570 }, { "epoch": 0.13202467639549784, "grad_norm": 3.4255616664886475, "learning_rate": 1.9459082031607904e-05, "loss": 0.1817347764968872, "step": 26580 }, { "epoch": 0.13207434707886712, "grad_norm": 1.2157433032989502, "learning_rate": 1.9458559987166205e-05, "loss": 0.1843521237373352, "step": 26590 }, { "epoch": 0.13212401776223637, "grad_norm": 2.0518391132354736, "learning_rate": 1.9458037697940783e-05, "loss": 0.1981769323348999, "step": 26600 }, { "epoch": 0.13217368844560565, "grad_norm": 1.562806487083435, "learning_rate": 1.945751516394514e-05, "loss": 0.18654519319534302, "step": 26610 }, { "epoch": 0.1322233591289749, "grad_norm": 3.278066635131836, "learning_rate": 1.9456992385192812e-05, "loss": 0.1889805316925049, "step": 26620 }, { "epoch": 0.13227302981234415, "grad_norm": 3.5085980892181396, "learning_rate": 1.9456469361697323e-05, "loss": 0.1945718050003052, "step": 26630 }, { "epoch": 0.13232270049571343, "grad_norm": 2.037654399871826, "learning_rate": 1.945594609347221e-05, "loss": 0.17262959480285645, "step": 26640 }, { "epoch": 0.13237237117908268, "grad_norm": 3.8233909606933594, "learning_rate": 1.9455422580531015e-05, "loss": 0.16230626106262208, "step": 26650 }, { "epoch": 0.13242204186245193, "grad_norm": 2.338029384613037, "learning_rate": 1.9454898822887284e-05, "loss": 0.1870488405227661, "step": 26660 }, { "epoch": 0.1324717125458212, "grad_norm": 1.1677855253219604, "learning_rate": 1.9454374820554572e-05, "loss": 0.17539416551589965, "step": 26670 }, { "epoch": 0.13252138322919046, "grad_norm": 1.1020822525024414, "learning_rate": 1.945385057354644e-05, "loss": 0.17700071334838868, "step": 26680 }, { "epoch": 0.13257105391255972, "grad_norm": 1.0553165674209595, "learning_rate": 1.9453326081876458e-05, "loss": 0.18422119617462157, "step": 26690 }, { "epoch": 0.132620724595929, "grad_norm": 0.7286959290504456, "learning_rate": 1.9452801345558198e-05, "loss": 0.17306292057037354, "step": 26700 }, { "epoch": 0.13267039527929825, "grad_norm": 0.8689184188842773, "learning_rate": 1.9452276364605242e-05, "loss": 0.17650561332702636, "step": 26710 }, { "epoch": 0.13272006596266753, "grad_norm": 2.611288547515869, "learning_rate": 1.9451751139031173e-05, "loss": 0.18196983337402345, "step": 26720 }, { "epoch": 0.13276973664603678, "grad_norm": 0.9884799122810364, "learning_rate": 1.945122566884958e-05, "loss": 0.18342846632003784, "step": 26730 }, { "epoch": 0.13281940732940603, "grad_norm": 1.642757534980774, "learning_rate": 1.9450699954074073e-05, "loss": 0.18712198734283447, "step": 26740 }, { "epoch": 0.1328690780127753, "grad_norm": 0.5611499547958374, "learning_rate": 1.9450173994718253e-05, "loss": 0.17698569297790528, "step": 26750 }, { "epoch": 0.13291874869614456, "grad_norm": 0.8611632585525513, "learning_rate": 1.9449647790795722e-05, "loss": 0.18541648387908935, "step": 26760 }, { "epoch": 0.1329684193795138, "grad_norm": 2.3904836177825928, "learning_rate": 1.9449121342320113e-05, "loss": 0.19593396186828613, "step": 26770 }, { "epoch": 0.1330180900628831, "grad_norm": 0.8498552441596985, "learning_rate": 1.9448594649305038e-05, "loss": 0.1899458646774292, "step": 26780 }, { "epoch": 0.13306776074625234, "grad_norm": 1.0489935874938965, "learning_rate": 1.9448067711764134e-05, "loss": 0.18347680568695068, "step": 26790 }, { "epoch": 0.1331174314296216, "grad_norm": 0.693487286567688, "learning_rate": 1.9447540529711036e-05, "loss": 0.18085932731628418, "step": 26800 }, { "epoch": 0.13316710211299088, "grad_norm": 2.4967563152313232, "learning_rate": 1.944701310315939e-05, "loss": 0.18132482767105101, "step": 26810 }, { "epoch": 0.13321677279636013, "grad_norm": 1.5586553812026978, "learning_rate": 1.9446485432122847e-05, "loss": 0.17418284416198732, "step": 26820 }, { "epoch": 0.1332664434797294, "grad_norm": 1.8493847846984863, "learning_rate": 1.9445957516615053e-05, "loss": 0.1799140214920044, "step": 26830 }, { "epoch": 0.13331611416309866, "grad_norm": 0.950076162815094, "learning_rate": 1.9445429356649678e-05, "loss": 0.1891400098800659, "step": 26840 }, { "epoch": 0.1333657848464679, "grad_norm": 1.774200439453125, "learning_rate": 1.944490095224039e-05, "loss": 0.17658636569976807, "step": 26850 }, { "epoch": 0.1334154555298372, "grad_norm": 1.719868779182434, "learning_rate": 1.944437230340086e-05, "loss": 0.19455924034118652, "step": 26860 }, { "epoch": 0.13346512621320644, "grad_norm": 1.4839411973953247, "learning_rate": 1.9443843410144777e-05, "loss": 0.17157161235809326, "step": 26870 }, { "epoch": 0.1335147968965757, "grad_norm": 1.1769843101501465, "learning_rate": 1.944331427248582e-05, "loss": 0.16112654209136962, "step": 26880 }, { "epoch": 0.13356446757994497, "grad_norm": 3.2189033031463623, "learning_rate": 1.9442784890437686e-05, "loss": 0.18028135299682618, "step": 26890 }, { "epoch": 0.13361413826331422, "grad_norm": 2.279512405395508, "learning_rate": 1.944225526401408e-05, "loss": 0.1931905508041382, "step": 26900 }, { "epoch": 0.13366380894668348, "grad_norm": 1.3100935220718384, "learning_rate": 1.9441725393228702e-05, "loss": 0.1893458127975464, "step": 26910 }, { "epoch": 0.13371347963005276, "grad_norm": 0.78221595287323, "learning_rate": 1.9441195278095268e-05, "loss": 0.19884846210479737, "step": 26920 }, { "epoch": 0.133763150313422, "grad_norm": 1.6212657690048218, "learning_rate": 1.9440664918627492e-05, "loss": 0.17109472751617433, "step": 26930 }, { "epoch": 0.1338128209967913, "grad_norm": 2.074383497238159, "learning_rate": 1.944013431483911e-05, "loss": 0.18343720436096192, "step": 26940 }, { "epoch": 0.13386249168016054, "grad_norm": 2.073859930038452, "learning_rate": 1.9439603466743842e-05, "loss": 0.20452516078948973, "step": 26950 }, { "epoch": 0.1339121623635298, "grad_norm": 1.8028769493103027, "learning_rate": 1.9439072374355435e-05, "loss": 0.1923826217651367, "step": 26960 }, { "epoch": 0.13396183304689907, "grad_norm": 0.9696648120880127, "learning_rate": 1.943854103768763e-05, "loss": 0.18748083114624023, "step": 26970 }, { "epoch": 0.13401150373026832, "grad_norm": 1.2956091165542603, "learning_rate": 1.9438009456754183e-05, "loss": 0.18301765918731688, "step": 26980 }, { "epoch": 0.13406117441363757, "grad_norm": 0.9560751914978027, "learning_rate": 1.9437477631568838e-05, "loss": 0.18750765323638915, "step": 26990 }, { "epoch": 0.13411084509700685, "grad_norm": 1.782896876335144, "learning_rate": 1.9436945562145372e-05, "loss": 0.18825361728668213, "step": 27000 }, { "epoch": 0.1341605157803761, "grad_norm": 1.2873575687408447, "learning_rate": 1.9436413248497543e-05, "loss": 0.18538697957992553, "step": 27010 }, { "epoch": 0.13421018646374536, "grad_norm": 1.5932412147521973, "learning_rate": 1.943588069063914e-05, "loss": 0.18977785110473633, "step": 27020 }, { "epoch": 0.13425985714711464, "grad_norm": 0.7986469864845276, "learning_rate": 1.9435347888583934e-05, "loss": 0.18758203983306884, "step": 27030 }, { "epoch": 0.1343095278304839, "grad_norm": 1.8905911445617676, "learning_rate": 1.943481484234572e-05, "loss": 0.17515945434570312, "step": 27040 }, { "epoch": 0.13435919851385314, "grad_norm": 1.4547089338302612, "learning_rate": 1.9434281551938296e-05, "loss": 0.17982900142669678, "step": 27050 }, { "epoch": 0.13440886919722242, "grad_norm": 2.2560689449310303, "learning_rate": 1.9433748017375452e-05, "loss": 0.17626574039459228, "step": 27060 }, { "epoch": 0.13445853988059167, "grad_norm": 0.8934620022773743, "learning_rate": 1.9433214238671005e-05, "loss": 0.1771489143371582, "step": 27070 }, { "epoch": 0.13450821056396095, "grad_norm": 2.04897403717041, "learning_rate": 1.9432680215838767e-05, "loss": 0.18187811374664306, "step": 27080 }, { "epoch": 0.1345578812473302, "grad_norm": 1.670207142829895, "learning_rate": 1.943214594889256e-05, "loss": 0.17331428527832032, "step": 27090 }, { "epoch": 0.13460755193069945, "grad_norm": 1.2819759845733643, "learning_rate": 1.9431611437846203e-05, "loss": 0.19066352844238282, "step": 27100 }, { "epoch": 0.13465722261406873, "grad_norm": 1.654479742050171, "learning_rate": 1.9431076682713538e-05, "loss": 0.18689072132110596, "step": 27110 }, { "epoch": 0.13470689329743799, "grad_norm": 1.275480031967163, "learning_rate": 1.94305416835084e-05, "loss": 0.2006300210952759, "step": 27120 }, { "epoch": 0.13475656398080724, "grad_norm": 0.8691011071205139, "learning_rate": 1.9430006440244634e-05, "loss": 0.17484605312347412, "step": 27130 }, { "epoch": 0.13480623466417652, "grad_norm": 3.0644452571868896, "learning_rate": 1.9429470952936093e-05, "loss": 0.1859147310256958, "step": 27140 }, { "epoch": 0.13485590534754577, "grad_norm": 2.16788649559021, "learning_rate": 1.9428935221596635e-05, "loss": 0.1838898777961731, "step": 27150 }, { "epoch": 0.13490557603091502, "grad_norm": 1.2484973669052124, "learning_rate": 1.9428399246240127e-05, "loss": 0.18992867469787597, "step": 27160 }, { "epoch": 0.1349552467142843, "grad_norm": 1.2576338052749634, "learning_rate": 1.9427863026880434e-05, "loss": 0.17363903522491456, "step": 27170 }, { "epoch": 0.13500491739765355, "grad_norm": 2.215710401535034, "learning_rate": 1.942732656353144e-05, "loss": 0.16910572052001954, "step": 27180 }, { "epoch": 0.13505458808102283, "grad_norm": 0.7921093106269836, "learning_rate": 1.9426789856207024e-05, "loss": 0.18723185062408448, "step": 27190 }, { "epoch": 0.13510425876439208, "grad_norm": 3.203923463821411, "learning_rate": 1.942625290492108e-05, "loss": 0.1900247097015381, "step": 27200 }, { "epoch": 0.13515392944776133, "grad_norm": 2.5093817710876465, "learning_rate": 1.9425715709687494e-05, "loss": 0.18670753240585328, "step": 27210 }, { "epoch": 0.1352036001311306, "grad_norm": 2.17218279838562, "learning_rate": 1.9425178270520183e-05, "loss": 0.18064141273498535, "step": 27220 }, { "epoch": 0.13525327081449987, "grad_norm": 1.6363533735275269, "learning_rate": 1.9424640587433042e-05, "loss": 0.17235856056213378, "step": 27230 }, { "epoch": 0.13530294149786912, "grad_norm": 1.9918444156646729, "learning_rate": 1.9424102660439993e-05, "loss": 0.1834190607070923, "step": 27240 }, { "epoch": 0.1353526121812384, "grad_norm": 1.345116376876831, "learning_rate": 1.9423564489554957e-05, "loss": 0.18486206531524657, "step": 27250 }, { "epoch": 0.13540228286460765, "grad_norm": 1.144951343536377, "learning_rate": 1.9423026074791865e-05, "loss": 0.18516988754272462, "step": 27260 }, { "epoch": 0.1354519535479769, "grad_norm": 1.7203402519226074, "learning_rate": 1.942248741616464e-05, "loss": 0.18655956983566285, "step": 27270 }, { "epoch": 0.13550162423134618, "grad_norm": 1.987969160079956, "learning_rate": 1.9421948513687233e-05, "loss": 0.18229583501815796, "step": 27280 }, { "epoch": 0.13555129491471543, "grad_norm": 2.0007541179656982, "learning_rate": 1.9421409367373586e-05, "loss": 0.1946274757385254, "step": 27290 }, { "epoch": 0.1356009655980847, "grad_norm": 0.9004433751106262, "learning_rate": 1.9420869977237657e-05, "loss": 0.18384565114974977, "step": 27300 }, { "epoch": 0.13565063628145396, "grad_norm": 3.339570999145508, "learning_rate": 1.9420330343293397e-05, "loss": 0.19517343044281005, "step": 27310 }, { "epoch": 0.13570030696482321, "grad_norm": 2.0705935955047607, "learning_rate": 1.9419790465554776e-05, "loss": 0.18380370140075683, "step": 27320 }, { "epoch": 0.1357499776481925, "grad_norm": 1.6899807453155518, "learning_rate": 1.9419250344035766e-05, "loss": 0.18229868412017822, "step": 27330 }, { "epoch": 0.13579964833156175, "grad_norm": 1.7177743911743164, "learning_rate": 1.9418709978750344e-05, "loss": 0.17641925811767578, "step": 27340 }, { "epoch": 0.135849319014931, "grad_norm": 1.2353113889694214, "learning_rate": 1.9418169369712495e-05, "loss": 0.17322196960449218, "step": 27350 }, { "epoch": 0.13589898969830028, "grad_norm": 1.4871606826782227, "learning_rate": 1.941762851693621e-05, "loss": 0.18714528083801268, "step": 27360 }, { "epoch": 0.13594866038166953, "grad_norm": 0.6233254671096802, "learning_rate": 1.9417087420435485e-05, "loss": 0.18814642429351808, "step": 27370 }, { "epoch": 0.13599833106503878, "grad_norm": 1.4607185125350952, "learning_rate": 1.9416546080224322e-05, "loss": 0.17880680561065673, "step": 27380 }, { "epoch": 0.13604800174840806, "grad_norm": 1.3398548364639282, "learning_rate": 1.9416004496316732e-05, "loss": 0.18199429512023926, "step": 27390 }, { "epoch": 0.1360976724317773, "grad_norm": 1.5293458700180054, "learning_rate": 1.9415462668726735e-05, "loss": 0.18732826709747313, "step": 27400 }, { "epoch": 0.1361473431151466, "grad_norm": 1.2968515157699585, "learning_rate": 1.941492059746835e-05, "loss": 0.17738056182861328, "step": 27410 }, { "epoch": 0.13619701379851584, "grad_norm": 1.0411409139633179, "learning_rate": 1.9414378282555603e-05, "loss": 0.19937161207199097, "step": 27420 }, { "epoch": 0.1362466844818851, "grad_norm": 1.5628459453582764, "learning_rate": 1.9413835724002536e-05, "loss": 0.17681028842926025, "step": 27430 }, { "epoch": 0.13629635516525437, "grad_norm": 1.9662352800369263, "learning_rate": 1.9413292921823182e-05, "loss": 0.1785546660423279, "step": 27440 }, { "epoch": 0.13634602584862363, "grad_norm": 1.5431309938430786, "learning_rate": 1.9412749876031593e-05, "loss": 0.19562441110610962, "step": 27450 }, { "epoch": 0.13639569653199288, "grad_norm": 1.537250280380249, "learning_rate": 1.941220658664182e-05, "loss": 0.18998925685882567, "step": 27460 }, { "epoch": 0.13644536721536216, "grad_norm": 1.4299372434616089, "learning_rate": 1.9411663053667928e-05, "loss": 0.20803778171539306, "step": 27470 }, { "epoch": 0.1364950378987314, "grad_norm": 0.895234227180481, "learning_rate": 1.941111927712398e-05, "loss": 0.1821634888648987, "step": 27480 }, { "epoch": 0.13654470858210066, "grad_norm": 1.257057785987854, "learning_rate": 1.9410575257024045e-05, "loss": 0.19531241655349732, "step": 27490 }, { "epoch": 0.13659437926546994, "grad_norm": 2.346421957015991, "learning_rate": 1.9410030993382213e-05, "loss": 0.19237897396087647, "step": 27500 }, { "epoch": 0.1366440499488392, "grad_norm": 0.6799458861351013, "learning_rate": 1.9409486486212558e-05, "loss": 0.17376748323440552, "step": 27510 }, { "epoch": 0.13669372063220847, "grad_norm": 1.149394154548645, "learning_rate": 1.9408941735529176e-05, "loss": 0.17602219581604003, "step": 27520 }, { "epoch": 0.13674339131557772, "grad_norm": 2.1301379203796387, "learning_rate": 1.9408396741346167e-05, "loss": 0.2017347812652588, "step": 27530 }, { "epoch": 0.13679306199894697, "grad_norm": 1.388420581817627, "learning_rate": 1.9407851503677632e-05, "loss": 0.17190988063812257, "step": 27540 }, { "epoch": 0.13684273268231625, "grad_norm": 1.7811837196350098, "learning_rate": 1.9407306022537683e-05, "loss": 0.17750765085220338, "step": 27550 }, { "epoch": 0.1368924033656855, "grad_norm": 0.8428352475166321, "learning_rate": 1.940676029794044e-05, "loss": 0.21212413311004638, "step": 27560 }, { "epoch": 0.13694207404905476, "grad_norm": 0.9363886713981628, "learning_rate": 1.940621432990002e-05, "loss": 0.16706817150115966, "step": 27570 }, { "epoch": 0.13699174473242404, "grad_norm": 1.7648849487304688, "learning_rate": 1.9405668118430553e-05, "loss": 0.17834644317626952, "step": 27580 }, { "epoch": 0.1370414154157933, "grad_norm": 1.6605874300003052, "learning_rate": 1.940512166354618e-05, "loss": 0.17546961307525635, "step": 27590 }, { "epoch": 0.13709108609916254, "grad_norm": 0.6277685761451721, "learning_rate": 1.940457496526104e-05, "loss": 0.17400498390197755, "step": 27600 }, { "epoch": 0.13714075678253182, "grad_norm": 2.031015396118164, "learning_rate": 1.9404028023589277e-05, "loss": 0.18826963901519775, "step": 27610 }, { "epoch": 0.13719042746590107, "grad_norm": 1.196386456489563, "learning_rate": 1.9403480838545054e-05, "loss": 0.18069729804992676, "step": 27620 }, { "epoch": 0.13724009814927035, "grad_norm": 2.832772731781006, "learning_rate": 1.940293341014253e-05, "loss": 0.19486795663833617, "step": 27630 }, { "epoch": 0.1372897688326396, "grad_norm": 1.2626588344573975, "learning_rate": 1.9402385738395867e-05, "loss": 0.18113181591033936, "step": 27640 }, { "epoch": 0.13733943951600885, "grad_norm": 1.161792516708374, "learning_rate": 1.940183782331924e-05, "loss": 0.18141601085662842, "step": 27650 }, { "epoch": 0.13738911019937813, "grad_norm": 1.225027084350586, "learning_rate": 1.9401289664926833e-05, "loss": 0.20036585330963136, "step": 27660 }, { "epoch": 0.13743878088274739, "grad_norm": 1.22544264793396, "learning_rate": 1.9400741263232826e-05, "loss": 0.17902827262878418, "step": 27670 }, { "epoch": 0.13748845156611664, "grad_norm": 0.9979649186134338, "learning_rate": 1.940019261825142e-05, "loss": 0.181141459941864, "step": 27680 }, { "epoch": 0.13753812224948592, "grad_norm": 3.0104055404663086, "learning_rate": 1.939964372999681e-05, "loss": 0.1767732620239258, "step": 27690 }, { "epoch": 0.13758779293285517, "grad_norm": 1.135705590248108, "learning_rate": 1.9399094598483193e-05, "loss": 0.17232565879821776, "step": 27700 }, { "epoch": 0.13763746361622442, "grad_norm": 1.5466220378875732, "learning_rate": 1.939854522372479e-05, "loss": 0.2041219711303711, "step": 27710 }, { "epoch": 0.1376871342995937, "grad_norm": 0.5634123086929321, "learning_rate": 1.9397995605735818e-05, "loss": 0.17748785018920898, "step": 27720 }, { "epoch": 0.13773680498296295, "grad_norm": 1.0254367589950562, "learning_rate": 1.9397445744530493e-05, "loss": 0.19888033866882324, "step": 27730 }, { "epoch": 0.13778647566633223, "grad_norm": 1.5506035089492798, "learning_rate": 1.9396895640123056e-05, "loss": 0.19625771045684814, "step": 27740 }, { "epoch": 0.13783614634970148, "grad_norm": 1.4668302536010742, "learning_rate": 1.9396345292527737e-05, "loss": 0.18711384534835815, "step": 27750 }, { "epoch": 0.13788581703307073, "grad_norm": 1.0416755676269531, "learning_rate": 1.939579470175878e-05, "loss": 0.18482794761657714, "step": 27760 }, { "epoch": 0.13793548771644001, "grad_norm": 1.4449671506881714, "learning_rate": 1.939524386783043e-05, "loss": 0.17224090099334716, "step": 27770 }, { "epoch": 0.13798515839980927, "grad_norm": 1.406320571899414, "learning_rate": 1.939469279075695e-05, "loss": 0.1848979949951172, "step": 27780 }, { "epoch": 0.13803482908317852, "grad_norm": 1.7166367769241333, "learning_rate": 1.93941414705526e-05, "loss": 0.1845099687576294, "step": 27790 }, { "epoch": 0.1380844997665478, "grad_norm": 1.0841820240020752, "learning_rate": 1.9393589907231643e-05, "loss": 0.18090469837188722, "step": 27800 }, { "epoch": 0.13813417044991705, "grad_norm": 2.383244514465332, "learning_rate": 1.939303810080836e-05, "loss": 0.17642457485198976, "step": 27810 }, { "epoch": 0.1381838411332863, "grad_norm": 2.299051523208618, "learning_rate": 1.9392486051297023e-05, "loss": 0.17042255401611328, "step": 27820 }, { "epoch": 0.13823351181665558, "grad_norm": 1.9850053787231445, "learning_rate": 1.9391933758711924e-05, "loss": 0.18235760927200317, "step": 27830 }, { "epoch": 0.13828318250002483, "grad_norm": 2.20318341255188, "learning_rate": 1.9391381223067357e-05, "loss": 0.1785784363746643, "step": 27840 }, { "epoch": 0.13833285318339408, "grad_norm": 0.7579206228256226, "learning_rate": 1.939082844437762e-05, "loss": 0.1828683137893677, "step": 27850 }, { "epoch": 0.13838252386676336, "grad_norm": 2.441469669342041, "learning_rate": 1.9390275422657018e-05, "loss": 0.1881819486618042, "step": 27860 }, { "epoch": 0.13843219455013261, "grad_norm": 2.6647260189056396, "learning_rate": 1.9389722157919862e-05, "loss": 0.17818273305892945, "step": 27870 }, { "epoch": 0.1384818652335019, "grad_norm": 1.4848214387893677, "learning_rate": 1.9389168650180478e-05, "loss": 0.20144662857055665, "step": 27880 }, { "epoch": 0.13853153591687115, "grad_norm": 1.4542549848556519, "learning_rate": 1.938861489945318e-05, "loss": 0.16094990968704223, "step": 27890 }, { "epoch": 0.1385812066002404, "grad_norm": 1.4721425771713257, "learning_rate": 1.9388060905752302e-05, "loss": 0.17210901975631715, "step": 27900 }, { "epoch": 0.13863087728360968, "grad_norm": 0.6883972883224487, "learning_rate": 1.9387506669092186e-05, "loss": 0.17505226135253907, "step": 27910 }, { "epoch": 0.13868054796697893, "grad_norm": 1.6376712322235107, "learning_rate": 1.938695218948717e-05, "loss": 0.19256938695907594, "step": 27920 }, { "epoch": 0.13873021865034818, "grad_norm": 2.5876517295837402, "learning_rate": 1.9386397466951608e-05, "loss": 0.1877586841583252, "step": 27930 }, { "epoch": 0.13877988933371746, "grad_norm": 1.2803144454956055, "learning_rate": 1.938584250149985e-05, "loss": 0.18159151077270508, "step": 27940 }, { "epoch": 0.1388295600170867, "grad_norm": 2.258728265762329, "learning_rate": 1.9385287293146265e-05, "loss": 0.1706477403640747, "step": 27950 }, { "epoch": 0.13887923070045596, "grad_norm": 1.9156562089920044, "learning_rate": 1.938473184190522e-05, "loss": 0.18765323162078856, "step": 27960 }, { "epoch": 0.13892890138382524, "grad_norm": 1.695452332496643, "learning_rate": 1.9384176147791085e-05, "loss": 0.17388068437576293, "step": 27970 }, { "epoch": 0.1389785720671945, "grad_norm": 3.9161758422851562, "learning_rate": 1.9383620210818247e-05, "loss": 0.18432981967926027, "step": 27980 }, { "epoch": 0.13902824275056377, "grad_norm": 1.093643307685852, "learning_rate": 1.938306403100109e-05, "loss": 0.17998063564300537, "step": 27990 }, { "epoch": 0.13907791343393303, "grad_norm": 2.4151668548583984, "learning_rate": 1.9382507608354006e-05, "loss": 0.17080304622650147, "step": 28000 }, { "epoch": 0.13912758411730228, "grad_norm": 1.1491363048553467, "learning_rate": 1.93819509428914e-05, "loss": 0.18001322746276854, "step": 28010 }, { "epoch": 0.13917725480067156, "grad_norm": 0.8652763366699219, "learning_rate": 1.9381394034627673e-05, "loss": 0.16442826986312867, "step": 28020 }, { "epoch": 0.1392269254840408, "grad_norm": 0.7787810564041138, "learning_rate": 1.9380836883577246e-05, "loss": 0.17970545291900636, "step": 28030 }, { "epoch": 0.13927659616741006, "grad_norm": 0.9232896566390991, "learning_rate": 1.938027948975453e-05, "loss": 0.1856459617614746, "step": 28040 }, { "epoch": 0.13932626685077934, "grad_norm": 1.3157522678375244, "learning_rate": 1.937972185317395e-05, "loss": 0.17501145601272583, "step": 28050 }, { "epoch": 0.1393759375341486, "grad_norm": 1.9728962182998657, "learning_rate": 1.9379163973849945e-05, "loss": 0.19780964851379396, "step": 28060 }, { "epoch": 0.13942560821751784, "grad_norm": 1.5998927354812622, "learning_rate": 1.9378605851796943e-05, "loss": 0.18249719142913817, "step": 28070 }, { "epoch": 0.13947527890088712, "grad_norm": 0.997861921787262, "learning_rate": 1.937804748702939e-05, "loss": 0.1783275604248047, "step": 28080 }, { "epoch": 0.13952494958425637, "grad_norm": 1.6891186237335205, "learning_rate": 1.9377488879561746e-05, "loss": 0.18355371952056884, "step": 28090 }, { "epoch": 0.13957462026762565, "grad_norm": 0.8453723788261414, "learning_rate": 1.9376930029408455e-05, "loss": 0.18064275979995728, "step": 28100 }, { "epoch": 0.1396242909509949, "grad_norm": 2.5764646530151367, "learning_rate": 1.937637093658399e-05, "loss": 0.17986072301864625, "step": 28110 }, { "epoch": 0.13967396163436416, "grad_norm": 1.3719772100448608, "learning_rate": 1.937581160110281e-05, "loss": 0.18531723022460939, "step": 28120 }, { "epoch": 0.13972363231773344, "grad_norm": 0.911641538143158, "learning_rate": 1.9375252022979397e-05, "loss": 0.17571882009506226, "step": 28130 }, { "epoch": 0.1397733030011027, "grad_norm": 1.8641198873519897, "learning_rate": 1.9374692202228233e-05, "loss": 0.17313592433929442, "step": 28140 }, { "epoch": 0.13982297368447194, "grad_norm": 1.4022141695022583, "learning_rate": 1.9374132138863807e-05, "loss": 0.18455779552459717, "step": 28150 }, { "epoch": 0.13987264436784122, "grad_norm": 2.6161482334136963, "learning_rate": 1.9373571832900607e-05, "loss": 0.18854868412017822, "step": 28160 }, { "epoch": 0.13992231505121047, "grad_norm": 1.1349596977233887, "learning_rate": 1.9373011284353134e-05, "loss": 0.17497563362121582, "step": 28170 }, { "epoch": 0.13997198573457972, "grad_norm": 1.2819736003875732, "learning_rate": 1.9372450493235902e-05, "loss": 0.19250715970993043, "step": 28180 }, { "epoch": 0.140021656417949, "grad_norm": 1.0400869846343994, "learning_rate": 1.9371889459563417e-05, "loss": 0.1980425238609314, "step": 28190 }, { "epoch": 0.14007132710131825, "grad_norm": 0.6998359560966492, "learning_rate": 1.9371328183350204e-05, "loss": 0.1708619475364685, "step": 28200 }, { "epoch": 0.14012099778468753, "grad_norm": 0.8153715133666992, "learning_rate": 1.9370766664610783e-05, "loss": 0.18034932613372803, "step": 28210 }, { "epoch": 0.14017066846805679, "grad_norm": 1.0048856735229492, "learning_rate": 1.9370204903359687e-05, "loss": 0.1855098247528076, "step": 28220 }, { "epoch": 0.14022033915142604, "grad_norm": 0.7371640801429749, "learning_rate": 1.936964289961146e-05, "loss": 0.18312883377075195, "step": 28230 }, { "epoch": 0.14027000983479532, "grad_norm": 1.4592914581298828, "learning_rate": 1.9369080653380642e-05, "loss": 0.18062112331390381, "step": 28240 }, { "epoch": 0.14031968051816457, "grad_norm": 0.9758469462394714, "learning_rate": 1.936851816468178e-05, "loss": 0.1820411801338196, "step": 28250 }, { "epoch": 0.14036935120153382, "grad_norm": 2.6356427669525146, "learning_rate": 1.9367955433529438e-05, "loss": 0.17902712821960448, "step": 28260 }, { "epoch": 0.1404190218849031, "grad_norm": 0.9981349110603333, "learning_rate": 1.9367392459938177e-05, "loss": 0.18797098398208617, "step": 28270 }, { "epoch": 0.14046869256827235, "grad_norm": 2.047706127166748, "learning_rate": 1.9366829243922566e-05, "loss": 0.1891371488571167, "step": 28280 }, { "epoch": 0.1405183632516416, "grad_norm": 1.0445057153701782, "learning_rate": 1.9366265785497175e-05, "loss": 0.18246766328811645, "step": 28290 }, { "epoch": 0.14056803393501088, "grad_norm": 0.8312987089157104, "learning_rate": 1.9365702084676595e-05, "loss": 0.18047113418579103, "step": 28300 }, { "epoch": 0.14061770461838013, "grad_norm": 2.5201315879821777, "learning_rate": 1.936513814147541e-05, "loss": 0.18634822368621826, "step": 28310 }, { "epoch": 0.14066737530174941, "grad_norm": 2.0631301403045654, "learning_rate": 1.9364573955908218e-05, "loss": 0.18062552213668823, "step": 28320 }, { "epoch": 0.14071704598511867, "grad_norm": 0.9373211860656738, "learning_rate": 1.9364009527989616e-05, "loss": 0.176885724067688, "step": 28330 }, { "epoch": 0.14076671666848792, "grad_norm": 0.7401484251022339, "learning_rate": 1.9363444857734208e-05, "loss": 0.1873692750930786, "step": 28340 }, { "epoch": 0.1408163873518572, "grad_norm": 1.3729935884475708, "learning_rate": 1.9362879945156615e-05, "loss": 0.18925553560256958, "step": 28350 }, { "epoch": 0.14086605803522645, "grad_norm": 1.3115904331207275, "learning_rate": 1.9362314790271458e-05, "loss": 0.17184755802154542, "step": 28360 }, { "epoch": 0.1409157287185957, "grad_norm": 2.1411244869232178, "learning_rate": 1.9361749393093353e-05, "loss": 0.17698172330856324, "step": 28370 }, { "epoch": 0.14096539940196498, "grad_norm": 1.706365704536438, "learning_rate": 1.936118375363694e-05, "loss": 0.18419041633605956, "step": 28380 }, { "epoch": 0.14101507008533423, "grad_norm": 2.8893754482269287, "learning_rate": 1.936061787191686e-05, "loss": 0.1717706561088562, "step": 28390 }, { "epoch": 0.14106474076870348, "grad_norm": 0.5440892577171326, "learning_rate": 1.9360051747947745e-05, "loss": 0.19059280157089234, "step": 28400 }, { "epoch": 0.14111441145207276, "grad_norm": 1.275819182395935, "learning_rate": 1.935948538174426e-05, "loss": 0.18288586139678956, "step": 28410 }, { "epoch": 0.14116408213544201, "grad_norm": 0.7729244828224182, "learning_rate": 1.9358918773321055e-05, "loss": 0.18123239278793335, "step": 28420 }, { "epoch": 0.1412137528188113, "grad_norm": 2.5858511924743652, "learning_rate": 1.9358351922692793e-05, "loss": 0.18683178424835206, "step": 28430 }, { "epoch": 0.14126342350218055, "grad_norm": 2.247861385345459, "learning_rate": 1.935778482987415e-05, "loss": 0.18427131175994874, "step": 28440 }, { "epoch": 0.1413130941855498, "grad_norm": 1.709581971168518, "learning_rate": 1.9357217494879794e-05, "loss": 0.19036293029785156, "step": 28450 }, { "epoch": 0.14136276486891908, "grad_norm": 1.1562325954437256, "learning_rate": 1.9356649917724414e-05, "loss": 0.1802137851715088, "step": 28460 }, { "epoch": 0.14141243555228833, "grad_norm": 1.2186468839645386, "learning_rate": 1.9356082098422698e-05, "loss": 0.18728781938552858, "step": 28470 }, { "epoch": 0.14146210623565758, "grad_norm": 1.2684814929962158, "learning_rate": 1.9355514036989335e-05, "loss": 0.1700136423110962, "step": 28480 }, { "epoch": 0.14151177691902686, "grad_norm": 1.773095726966858, "learning_rate": 1.9354945733439036e-05, "loss": 0.18094979524612426, "step": 28490 }, { "epoch": 0.1415614476023961, "grad_norm": 2.799283504486084, "learning_rate": 1.9354377187786496e-05, "loss": 0.19017765522003174, "step": 28500 }, { "epoch": 0.14161111828576536, "grad_norm": 1.722339153289795, "learning_rate": 1.935380840004644e-05, "loss": 0.1813354969024658, "step": 28510 }, { "epoch": 0.14166078896913464, "grad_norm": 1.6914318799972534, "learning_rate": 1.9353239370233585e-05, "loss": 0.17007958889007568, "step": 28520 }, { "epoch": 0.1417104596525039, "grad_norm": 1.2139475345611572, "learning_rate": 1.9352670098362653e-05, "loss": 0.19361892938613892, "step": 28530 }, { "epoch": 0.14176013033587317, "grad_norm": 1.0571861267089844, "learning_rate": 1.935210058444838e-05, "loss": 0.1799007773399353, "step": 28540 }, { "epoch": 0.14180980101924243, "grad_norm": 1.1955476999282837, "learning_rate": 1.9351530828505502e-05, "loss": 0.19770748615264894, "step": 28550 }, { "epoch": 0.14185947170261168, "grad_norm": 1.5438495874404907, "learning_rate": 1.935096083054877e-05, "loss": 0.18946969509124756, "step": 28560 }, { "epoch": 0.14190914238598096, "grad_norm": 1.4864099025726318, "learning_rate": 1.935039059059293e-05, "loss": 0.18076289892196656, "step": 28570 }, { "epoch": 0.1419588130693502, "grad_norm": 0.8153520822525024, "learning_rate": 1.9349820108652742e-05, "loss": 0.17336665391921996, "step": 28580 }, { "epoch": 0.14200848375271946, "grad_norm": 2.2559876441955566, "learning_rate": 1.9349249384742966e-05, "loss": 0.19284721612930297, "step": 28590 }, { "epoch": 0.14205815443608874, "grad_norm": 1.8419215679168701, "learning_rate": 1.934867841887838e-05, "loss": 0.16896607875823974, "step": 28600 }, { "epoch": 0.142107825119458, "grad_norm": 2.2123939990997314, "learning_rate": 1.934810721107375e-05, "loss": 0.1794438600540161, "step": 28610 }, { "epoch": 0.14215749580282724, "grad_norm": 1.8134233951568604, "learning_rate": 1.934753576134387e-05, "loss": 0.19902713298797609, "step": 28620 }, { "epoch": 0.14220716648619652, "grad_norm": 1.2930028438568115, "learning_rate": 1.934696406970352e-05, "loss": 0.1775752067565918, "step": 28630 }, { "epoch": 0.14225683716956578, "grad_norm": 3.357447624206543, "learning_rate": 1.9346392136167495e-05, "loss": 0.1838373899459839, "step": 28640 }, { "epoch": 0.14230650785293503, "grad_norm": 1.77732515335083, "learning_rate": 1.9345819960750606e-05, "loss": 0.17888498306274414, "step": 28650 }, { "epoch": 0.1423561785363043, "grad_norm": 2.4471206665039062, "learning_rate": 1.934524754346765e-05, "loss": 0.1793160319328308, "step": 28660 }, { "epoch": 0.14240584921967356, "grad_norm": 2.0731148719787598, "learning_rate": 1.9344674884333446e-05, "loss": 0.17625963687896729, "step": 28670 }, { "epoch": 0.14245551990304284, "grad_norm": 2.846801996231079, "learning_rate": 1.934410198336281e-05, "loss": 0.18701382875442504, "step": 28680 }, { "epoch": 0.1425051905864121, "grad_norm": 2.1465647220611572, "learning_rate": 1.934352884057058e-05, "loss": 0.19575817584991456, "step": 28690 }, { "epoch": 0.14255486126978134, "grad_norm": 2.954359769821167, "learning_rate": 1.9342955455971576e-05, "loss": 0.17976226806640624, "step": 28700 }, { "epoch": 0.14260453195315062, "grad_norm": 1.9553420543670654, "learning_rate": 1.9342381829580637e-05, "loss": 0.18851003646850586, "step": 28710 }, { "epoch": 0.14265420263651987, "grad_norm": 2.908048629760742, "learning_rate": 1.9341807961412618e-05, "loss": 0.19988151788711547, "step": 28720 }, { "epoch": 0.14270387331988912, "grad_norm": 0.9918566346168518, "learning_rate": 1.9341233851482365e-05, "loss": 0.19125266075134278, "step": 28730 }, { "epoch": 0.1427535440032584, "grad_norm": 1.925794005393982, "learning_rate": 1.9340659499804734e-05, "loss": 0.18565940856933594, "step": 28740 }, { "epoch": 0.14280321468662766, "grad_norm": 1.1173303127288818, "learning_rate": 1.934008490639459e-05, "loss": 0.18583799600601197, "step": 28750 }, { "epoch": 0.1428528853699969, "grad_norm": 2.085951089859009, "learning_rate": 1.933951007126681e-05, "loss": 0.17982807159423828, "step": 28760 }, { "epoch": 0.1429025560533662, "grad_norm": 1.451005458831787, "learning_rate": 1.9338934994436255e-05, "loss": 0.1785667657852173, "step": 28770 }, { "epoch": 0.14295222673673544, "grad_norm": 2.6911284923553467, "learning_rate": 1.9338359675917827e-05, "loss": 0.17979331016540528, "step": 28780 }, { "epoch": 0.14300189742010472, "grad_norm": 1.8220802545547485, "learning_rate": 1.9337784115726397e-05, "loss": 0.17458879947662354, "step": 28790 }, { "epoch": 0.14305156810347397, "grad_norm": 1.2928552627563477, "learning_rate": 1.9337208313876875e-05, "loss": 0.19029631614685058, "step": 28800 }, { "epoch": 0.14310123878684322, "grad_norm": 1.4202969074249268, "learning_rate": 1.9336632270384153e-05, "loss": 0.18155989646911622, "step": 28810 }, { "epoch": 0.1431509094702125, "grad_norm": 1.6994954347610474, "learning_rate": 1.9336055985263144e-05, "loss": 0.190349018573761, "step": 28820 }, { "epoch": 0.14320058015358175, "grad_norm": 1.4415899515151978, "learning_rate": 1.9335479458528757e-05, "loss": 0.1853480577468872, "step": 28830 }, { "epoch": 0.143250250836951, "grad_norm": 1.0401725769042969, "learning_rate": 1.933490269019592e-05, "loss": 0.20166049003601075, "step": 28840 }, { "epoch": 0.14329992152032028, "grad_norm": 1.3311632871627808, "learning_rate": 1.933432568027955e-05, "loss": 0.18956055641174316, "step": 28850 }, { "epoch": 0.14334959220368954, "grad_norm": 2.2841906547546387, "learning_rate": 1.9333748428794585e-05, "loss": 0.17804508209228515, "step": 28860 }, { "epoch": 0.1433992628870588, "grad_norm": 1.8411954641342163, "learning_rate": 1.9333170935755963e-05, "loss": 0.1839818239212036, "step": 28870 }, { "epoch": 0.14344893357042807, "grad_norm": 1.541093349456787, "learning_rate": 1.9332593201178635e-05, "loss": 0.1853471040725708, "step": 28880 }, { "epoch": 0.14349860425379732, "grad_norm": 1.548852562904358, "learning_rate": 1.9332015225077544e-05, "loss": 0.17046090364456176, "step": 28890 }, { "epoch": 0.1435482749371666, "grad_norm": 0.698384702205658, "learning_rate": 1.9331437007467653e-05, "loss": 0.17730567455291749, "step": 28900 }, { "epoch": 0.14359794562053585, "grad_norm": 0.9767554402351379, "learning_rate": 1.9330858548363922e-05, "loss": 0.16911444664001465, "step": 28910 }, { "epoch": 0.1436476163039051, "grad_norm": 0.986283540725708, "learning_rate": 1.9330279847781327e-05, "loss": 0.1673969507217407, "step": 28920 }, { "epoch": 0.14369728698727438, "grad_norm": 1.7302603721618652, "learning_rate": 1.9329700905734837e-05, "loss": 0.18267284631729125, "step": 28930 }, { "epoch": 0.14374695767064363, "grad_norm": 2.2575697898864746, "learning_rate": 1.932912172223944e-05, "loss": 0.1731148838996887, "step": 28940 }, { "epoch": 0.14379662835401288, "grad_norm": 1.5393433570861816, "learning_rate": 1.9328542297310124e-05, "loss": 0.19910914897918702, "step": 28950 }, { "epoch": 0.14384629903738216, "grad_norm": 1.6112326383590698, "learning_rate": 1.9327962630961884e-05, "loss": 0.18556251525878906, "step": 28960 }, { "epoch": 0.14389596972075142, "grad_norm": 1.307153582572937, "learning_rate": 1.9327382723209722e-05, "loss": 0.1850416660308838, "step": 28970 }, { "epoch": 0.14394564040412067, "grad_norm": 2.237255811691284, "learning_rate": 1.9326802574068647e-05, "loss": 0.18509942293167114, "step": 28980 }, { "epoch": 0.14399531108748995, "grad_norm": 1.4495912790298462, "learning_rate": 1.9326222183553668e-05, "loss": 0.19244444370269775, "step": 28990 }, { "epoch": 0.1440449817708592, "grad_norm": 1.1583750247955322, "learning_rate": 1.932564155167981e-05, "loss": 0.1832229733467102, "step": 29000 }, { "epoch": 0.14409465245422848, "grad_norm": 1.3773789405822754, "learning_rate": 1.93250606784621e-05, "loss": 0.17793498039245606, "step": 29010 }, { "epoch": 0.14414432313759773, "grad_norm": 2.0345349311828613, "learning_rate": 1.932447956391557e-05, "loss": 0.18964990377426147, "step": 29020 }, { "epoch": 0.14419399382096698, "grad_norm": 1.338908314704895, "learning_rate": 1.9323898208055257e-05, "loss": 0.18268222808837892, "step": 29030 }, { "epoch": 0.14424366450433626, "grad_norm": 1.7539982795715332, "learning_rate": 1.9323316610896207e-05, "loss": 0.17999157905578614, "step": 29040 }, { "epoch": 0.1442933351877055, "grad_norm": 1.2268160581588745, "learning_rate": 1.9322734772453472e-05, "loss": 0.18924182653427124, "step": 29050 }, { "epoch": 0.14434300587107476, "grad_norm": 1.2969330549240112, "learning_rate": 1.932215269274211e-05, "loss": 0.17778468132019043, "step": 29060 }, { "epoch": 0.14439267655444404, "grad_norm": 1.615728497505188, "learning_rate": 1.9321570371777183e-05, "loss": 0.20082285404205322, "step": 29070 }, { "epoch": 0.1444423472378133, "grad_norm": 1.1194851398468018, "learning_rate": 1.9320987809573763e-05, "loss": 0.18956727981567384, "step": 29080 }, { "epoch": 0.14449201792118255, "grad_norm": 2.884938955307007, "learning_rate": 1.9320405006146927e-05, "loss": 0.19526045322418212, "step": 29090 }, { "epoch": 0.14454168860455183, "grad_norm": 0.8876255750656128, "learning_rate": 1.931982196151176e-05, "loss": 0.17104653120040894, "step": 29100 }, { "epoch": 0.14459135928792108, "grad_norm": 1.2391972541809082, "learning_rate": 1.9319238675683342e-05, "loss": 0.18988338708877564, "step": 29110 }, { "epoch": 0.14464102997129036, "grad_norm": 2.35544753074646, "learning_rate": 1.9318655148676778e-05, "loss": 0.18227343559265136, "step": 29120 }, { "epoch": 0.1446907006546596, "grad_norm": 1.0596942901611328, "learning_rate": 1.9318071380507164e-05, "loss": 0.1747299075126648, "step": 29130 }, { "epoch": 0.14474037133802886, "grad_norm": 1.1315048933029175, "learning_rate": 1.931748737118961e-05, "loss": 0.18515772819519044, "step": 29140 }, { "epoch": 0.14479004202139814, "grad_norm": 0.981863260269165, "learning_rate": 1.931690312073923e-05, "loss": 0.19572832584381103, "step": 29150 }, { "epoch": 0.1448397127047674, "grad_norm": 0.9060422778129578, "learning_rate": 1.9316318629171143e-05, "loss": 0.1761557340621948, "step": 29160 }, { "epoch": 0.14488938338813664, "grad_norm": 1.889999508857727, "learning_rate": 1.9315733896500477e-05, "loss": 0.18459537029266357, "step": 29170 }, { "epoch": 0.14493905407150592, "grad_norm": 1.3964438438415527, "learning_rate": 1.931514892274236e-05, "loss": 0.18893032073974608, "step": 29180 }, { "epoch": 0.14498872475487518, "grad_norm": 2.2080883979797363, "learning_rate": 1.9314563707911938e-05, "loss": 0.19499680995941163, "step": 29190 }, { "epoch": 0.14503839543824443, "grad_norm": 1.3011888265609741, "learning_rate": 1.9313978252024352e-05, "loss": 0.17904789447784425, "step": 29200 }, { "epoch": 0.1450880661216137, "grad_norm": 0.6557995676994324, "learning_rate": 1.9313392555094753e-05, "loss": 0.1688518524169922, "step": 29210 }, { "epoch": 0.14513773680498296, "grad_norm": 1.907175064086914, "learning_rate": 1.9312806617138297e-05, "loss": 0.1927654981613159, "step": 29220 }, { "epoch": 0.14518740748835224, "grad_norm": 1.9758015871047974, "learning_rate": 1.9312220438170156e-05, "loss": 0.18675507307052613, "step": 29230 }, { "epoch": 0.1452370781717215, "grad_norm": 1.369927167892456, "learning_rate": 1.931163401820549e-05, "loss": 0.18084192276000977, "step": 29240 }, { "epoch": 0.14528674885509074, "grad_norm": 1.0520716905593872, "learning_rate": 1.9311047357259477e-05, "loss": 0.17437922954559326, "step": 29250 }, { "epoch": 0.14533641953846002, "grad_norm": 1.7777634859085083, "learning_rate": 1.9310460455347306e-05, "loss": 0.18790305852890016, "step": 29260 }, { "epoch": 0.14538609022182927, "grad_norm": 1.327437400817871, "learning_rate": 1.9309873312484165e-05, "loss": 0.17229560613632203, "step": 29270 }, { "epoch": 0.14543576090519852, "grad_norm": 2.12414288520813, "learning_rate": 1.930928592868524e-05, "loss": 0.18708721399307252, "step": 29280 }, { "epoch": 0.1454854315885678, "grad_norm": 2.191420078277588, "learning_rate": 1.9308698303965744e-05, "loss": 0.1879962682723999, "step": 29290 }, { "epoch": 0.14553510227193706, "grad_norm": 1.3325568437576294, "learning_rate": 1.9308110438340875e-05, "loss": 0.19324190616607667, "step": 29300 }, { "epoch": 0.1455847729553063, "grad_norm": 1.4279762506484985, "learning_rate": 1.930752233182585e-05, "loss": 0.17497193813323975, "step": 29310 }, { "epoch": 0.1456344436386756, "grad_norm": 1.5385887622833252, "learning_rate": 1.930693398443589e-05, "loss": 0.1796709418296814, "step": 29320 }, { "epoch": 0.14568411432204484, "grad_norm": 1.7359999418258667, "learning_rate": 1.9306345396186222e-05, "loss": 0.19430186748504638, "step": 29330 }, { "epoch": 0.14573378500541412, "grad_norm": 1.7825512886047363, "learning_rate": 1.930575656709208e-05, "loss": 0.1809403896331787, "step": 29340 }, { "epoch": 0.14578345568878337, "grad_norm": 1.5629665851593018, "learning_rate": 1.9305167497168695e-05, "loss": 0.18952524662017822, "step": 29350 }, { "epoch": 0.14583312637215262, "grad_norm": 1.7028093338012695, "learning_rate": 1.9304578186431316e-05, "loss": 0.19061830043792724, "step": 29360 }, { "epoch": 0.1458827970555219, "grad_norm": 1.074440598487854, "learning_rate": 1.9303988634895196e-05, "loss": 0.18470299243927002, "step": 29370 }, { "epoch": 0.14593246773889115, "grad_norm": 1.1845790147781372, "learning_rate": 1.9303398842575592e-05, "loss": 0.17787160873413085, "step": 29380 }, { "epoch": 0.1459821384222604, "grad_norm": 1.1241306066513062, "learning_rate": 1.9302808809487764e-05, "loss": 0.1828018307685852, "step": 29390 }, { "epoch": 0.14603180910562968, "grad_norm": 1.1200752258300781, "learning_rate": 1.930221853564699e-05, "loss": 0.1657138466835022, "step": 29400 }, { "epoch": 0.14608147978899894, "grad_norm": 1.9138740301132202, "learning_rate": 1.9301628021068537e-05, "loss": 0.18255300521850587, "step": 29410 }, { "epoch": 0.1461311504723682, "grad_norm": 0.8716686367988586, "learning_rate": 1.9301037265767692e-05, "loss": 0.1966461420059204, "step": 29420 }, { "epoch": 0.14618082115573747, "grad_norm": 2.160550832748413, "learning_rate": 1.930044626975974e-05, "loss": 0.19492702484130858, "step": 29430 }, { "epoch": 0.14623049183910672, "grad_norm": 2.148151159286499, "learning_rate": 1.9299855033059982e-05, "loss": 0.17994744777679444, "step": 29440 }, { "epoch": 0.14628016252247597, "grad_norm": 1.873039722442627, "learning_rate": 1.9299263555683712e-05, "loss": 0.18670332431793213, "step": 29450 }, { "epoch": 0.14632983320584525, "grad_norm": 2.786990165710449, "learning_rate": 1.929867183764624e-05, "loss": 0.181727135181427, "step": 29460 }, { "epoch": 0.1463795038892145, "grad_norm": 1.3511732816696167, "learning_rate": 1.929807987896288e-05, "loss": 0.1833944320678711, "step": 29470 }, { "epoch": 0.14642917457258378, "grad_norm": 1.1667389869689941, "learning_rate": 1.9297487679648954e-05, "loss": 0.18261516094207764, "step": 29480 }, { "epoch": 0.14647884525595303, "grad_norm": 1.9436372518539429, "learning_rate": 1.929689523971978e-05, "loss": 0.17012211084365844, "step": 29490 }, { "epoch": 0.14652851593932228, "grad_norm": 2.3849213123321533, "learning_rate": 1.92963025591907e-05, "loss": 0.1752729058265686, "step": 29500 }, { "epoch": 0.14657818662269156, "grad_norm": 1.6298162937164307, "learning_rate": 1.9295709638077048e-05, "loss": 0.19332149028778076, "step": 29510 }, { "epoch": 0.14662785730606082, "grad_norm": 1.8877736330032349, "learning_rate": 1.9295116476394167e-05, "loss": 0.19032487869262696, "step": 29520 }, { "epoch": 0.14667752798943007, "grad_norm": 1.398093819618225, "learning_rate": 1.929452307415741e-05, "loss": 0.17042847871780395, "step": 29530 }, { "epoch": 0.14672719867279935, "grad_norm": 2.060904026031494, "learning_rate": 1.929392943138213e-05, "loss": 0.17109676599502563, "step": 29540 }, { "epoch": 0.1467768693561686, "grad_norm": 0.726324200630188, "learning_rate": 1.9293335548083693e-05, "loss": 0.17671566009521483, "step": 29550 }, { "epoch": 0.14682654003953785, "grad_norm": 1.7301427125930786, "learning_rate": 1.929274142427747e-05, "loss": 0.19357858896255492, "step": 29560 }, { "epoch": 0.14687621072290713, "grad_norm": 1.7097257375717163, "learning_rate": 1.929214705997884e-05, "loss": 0.18524059057235717, "step": 29570 }, { "epoch": 0.14692588140627638, "grad_norm": 1.7197787761688232, "learning_rate": 1.9291552455203172e-05, "loss": 0.1854040265083313, "step": 29580 }, { "epoch": 0.14697555208964566, "grad_norm": 0.8400035500526428, "learning_rate": 1.929095760996587e-05, "loss": 0.18335254192352296, "step": 29590 }, { "epoch": 0.1470252227730149, "grad_norm": 1.2759971618652344, "learning_rate": 1.9290362524282315e-05, "loss": 0.18034346103668214, "step": 29600 }, { "epoch": 0.14707489345638416, "grad_norm": 1.8098490238189697, "learning_rate": 1.9289767198167918e-05, "loss": 0.18314723968505858, "step": 29610 }, { "epoch": 0.14712456413975344, "grad_norm": 2.5886149406433105, "learning_rate": 1.928917163163808e-05, "loss": 0.19724024534225465, "step": 29620 }, { "epoch": 0.1471742348231227, "grad_norm": 2.9265522956848145, "learning_rate": 1.928857582470821e-05, "loss": 0.17134276628494263, "step": 29630 }, { "epoch": 0.14722390550649195, "grad_norm": 1.1468251943588257, "learning_rate": 1.928797977739374e-05, "loss": 0.18441956043243407, "step": 29640 }, { "epoch": 0.14727357618986123, "grad_norm": 0.7389082312583923, "learning_rate": 1.9287383489710084e-05, "loss": 0.17819828987121583, "step": 29650 }, { "epoch": 0.14732324687323048, "grad_norm": 1.520451545715332, "learning_rate": 1.9286786961672678e-05, "loss": 0.18742666244506836, "step": 29660 }, { "epoch": 0.14737291755659973, "grad_norm": 1.5890462398529053, "learning_rate": 1.9286190193296955e-05, "loss": 0.1767444610595703, "step": 29670 }, { "epoch": 0.147422588239969, "grad_norm": 1.1536139249801636, "learning_rate": 1.9285593184598374e-05, "loss": 0.17258827686309813, "step": 29680 }, { "epoch": 0.14747225892333826, "grad_norm": 0.8128504157066345, "learning_rate": 1.9284995935592365e-05, "loss": 0.18556243181228638, "step": 29690 }, { "epoch": 0.14752192960670754, "grad_norm": 1.3724454641342163, "learning_rate": 1.9284398446294398e-05, "loss": 0.18312243223190308, "step": 29700 }, { "epoch": 0.1475716002900768, "grad_norm": 1.9268337488174438, "learning_rate": 1.9283800716719937e-05, "loss": 0.18461605310440063, "step": 29710 }, { "epoch": 0.14762127097344604, "grad_norm": 1.1074241399765015, "learning_rate": 1.928320274688444e-05, "loss": 0.19270824193954467, "step": 29720 }, { "epoch": 0.14767094165681532, "grad_norm": 1.6391133069992065, "learning_rate": 1.928260453680339e-05, "loss": 0.182741117477417, "step": 29730 }, { "epoch": 0.14772061234018458, "grad_norm": 1.9864397048950195, "learning_rate": 1.9282006086492268e-05, "loss": 0.1873424768447876, "step": 29740 }, { "epoch": 0.14777028302355383, "grad_norm": 2.4027299880981445, "learning_rate": 1.928140739596656e-05, "loss": 0.19198501110076904, "step": 29750 }, { "epoch": 0.1478199537069231, "grad_norm": 1.4057915210723877, "learning_rate": 1.9280808465241762e-05, "loss": 0.1963893413543701, "step": 29760 }, { "epoch": 0.14786962439029236, "grad_norm": 3.1051459312438965, "learning_rate": 1.928020929433337e-05, "loss": 0.18507782220840455, "step": 29770 }, { "epoch": 0.1479192950736616, "grad_norm": 1.382337212562561, "learning_rate": 1.9279609883256896e-05, "loss": 0.164992094039917, "step": 29780 }, { "epoch": 0.1479689657570309, "grad_norm": 1.477189064025879, "learning_rate": 1.9279010232027847e-05, "loss": 0.17413326501846313, "step": 29790 }, { "epoch": 0.14801863644040014, "grad_norm": 1.6018192768096924, "learning_rate": 1.9278410340661743e-05, "loss": 0.18055200576782227, "step": 29800 }, { "epoch": 0.14806830712376942, "grad_norm": 0.99693363904953, "learning_rate": 1.9277810209174116e-05, "loss": 0.17238065004348754, "step": 29810 }, { "epoch": 0.14811797780713867, "grad_norm": 1.0002212524414062, "learning_rate": 1.9277209837580487e-05, "loss": 0.18479981422424316, "step": 29820 }, { "epoch": 0.14816764849050792, "grad_norm": 1.892257809638977, "learning_rate": 1.9276609225896398e-05, "loss": 0.17458524703979492, "step": 29830 }, { "epoch": 0.1482173191738772, "grad_norm": 1.173675298690796, "learning_rate": 1.927600837413739e-05, "loss": 0.18411662578582763, "step": 29840 }, { "epoch": 0.14826698985724646, "grad_norm": 1.551701545715332, "learning_rate": 1.9275407282319016e-05, "loss": 0.17669614553451538, "step": 29850 }, { "epoch": 0.1483166605406157, "grad_norm": 1.627732515335083, "learning_rate": 1.9274805950456833e-05, "loss": 0.19068632125854493, "step": 29860 }, { "epoch": 0.148366331223985, "grad_norm": 1.597590684890747, "learning_rate": 1.92742043785664e-05, "loss": 0.1836825728416443, "step": 29870 }, { "epoch": 0.14841600190735424, "grad_norm": 2.485156774520874, "learning_rate": 1.9273602566663287e-05, "loss": 0.18504034280776976, "step": 29880 }, { "epoch": 0.1484656725907235, "grad_norm": 1.3613239526748657, "learning_rate": 1.927300051476307e-05, "loss": 0.17489969730377197, "step": 29890 }, { "epoch": 0.14851534327409277, "grad_norm": 0.9666286706924438, "learning_rate": 1.9272398222881324e-05, "loss": 0.17671154737472533, "step": 29900 }, { "epoch": 0.14856501395746202, "grad_norm": 0.525879979133606, "learning_rate": 1.9271795691033643e-05, "loss": 0.17941316366195678, "step": 29910 }, { "epoch": 0.1486146846408313, "grad_norm": 1.297761082649231, "learning_rate": 1.927119291923562e-05, "loss": 0.1798780918121338, "step": 29920 }, { "epoch": 0.14866435532420055, "grad_norm": 1.9139798879623413, "learning_rate": 1.9270589907502848e-05, "loss": 0.18332889080047607, "step": 29930 }, { "epoch": 0.1487140260075698, "grad_norm": 1.0058412551879883, "learning_rate": 1.926998665585094e-05, "loss": 0.19065327644348146, "step": 29940 }, { "epoch": 0.14876369669093908, "grad_norm": 0.6264312863349915, "learning_rate": 1.92693831642955e-05, "loss": 0.1758502960205078, "step": 29950 }, { "epoch": 0.14881336737430834, "grad_norm": 0.5712208151817322, "learning_rate": 1.9268779432852156e-05, "loss": 0.1844334602355957, "step": 29960 }, { "epoch": 0.1488630380576776, "grad_norm": 1.4589961767196655, "learning_rate": 1.9268175461536524e-05, "loss": 0.20264363288879395, "step": 29970 }, { "epoch": 0.14891270874104687, "grad_norm": 2.742901563644409, "learning_rate": 1.926757125036424e-05, "loss": 0.1838843822479248, "step": 29980 }, { "epoch": 0.14896237942441612, "grad_norm": 1.6964538097381592, "learning_rate": 1.9266966799350936e-05, "loss": 0.18324694633483887, "step": 29990 }, { "epoch": 0.14901205010778537, "grad_norm": 0.7259454727172852, "learning_rate": 1.9266362108512258e-05, "loss": 0.17711384296417237, "step": 30000 }, { "epoch": 0.14906172079115465, "grad_norm": 1.6816494464874268, "learning_rate": 1.9265757177863854e-05, "loss": 0.18098419904708862, "step": 30010 }, { "epoch": 0.1491113914745239, "grad_norm": 1.0215961933135986, "learning_rate": 1.926515200742138e-05, "loss": 0.19583287239074706, "step": 30020 }, { "epoch": 0.14916106215789318, "grad_norm": 2.1277987957000732, "learning_rate": 1.92645465972005e-05, "loss": 0.18167026042938234, "step": 30030 }, { "epoch": 0.14921073284126243, "grad_norm": 0.9365331530570984, "learning_rate": 1.9263940947216873e-05, "loss": 0.1822674512863159, "step": 30040 }, { "epoch": 0.14926040352463169, "grad_norm": 1.4504727125167847, "learning_rate": 1.9263335057486183e-05, "loss": 0.1737520694732666, "step": 30050 }, { "epoch": 0.14931007420800096, "grad_norm": 0.9405813217163086, "learning_rate": 1.9262728928024108e-05, "loss": 0.18638514280319213, "step": 30060 }, { "epoch": 0.14935974489137022, "grad_norm": 1.8880785703659058, "learning_rate": 1.926212255884633e-05, "loss": 0.17443599700927734, "step": 30070 }, { "epoch": 0.14940941557473947, "grad_norm": 2.1399123668670654, "learning_rate": 1.9261515949968544e-05, "loss": 0.169667387008667, "step": 30080 }, { "epoch": 0.14945908625810875, "grad_norm": 1.0178227424621582, "learning_rate": 1.9260909101406454e-05, "loss": 0.1773904085159302, "step": 30090 }, { "epoch": 0.149508756941478, "grad_norm": 0.9951404333114624, "learning_rate": 1.9260302013175755e-05, "loss": 0.17239397764205933, "step": 30100 }, { "epoch": 0.14955842762484725, "grad_norm": 0.6393226981163025, "learning_rate": 1.9259694685292167e-05, "loss": 0.18313859701156615, "step": 30110 }, { "epoch": 0.14960809830821653, "grad_norm": 1.0953320264816284, "learning_rate": 1.92590871177714e-05, "loss": 0.1821296215057373, "step": 30120 }, { "epoch": 0.14965776899158578, "grad_norm": 3.0834107398986816, "learning_rate": 1.9258479310629182e-05, "loss": 0.17111811637878419, "step": 30130 }, { "epoch": 0.14970743967495506, "grad_norm": 1.3319495916366577, "learning_rate": 1.9257871263881244e-05, "loss": 0.1763042092323303, "step": 30140 }, { "epoch": 0.1497571103583243, "grad_norm": 2.003737449645996, "learning_rate": 1.9257262977543322e-05, "loss": 0.17362687587738038, "step": 30150 }, { "epoch": 0.14980678104169357, "grad_norm": 2.0020980834960938, "learning_rate": 1.9256654451631153e-05, "loss": 0.1779194474220276, "step": 30160 }, { "epoch": 0.14985645172506284, "grad_norm": 1.0081062316894531, "learning_rate": 1.925604568616049e-05, "loss": 0.1774513006210327, "step": 30170 }, { "epoch": 0.1499061224084321, "grad_norm": 1.5278505086898804, "learning_rate": 1.9255436681147085e-05, "loss": 0.17313181161880492, "step": 30180 }, { "epoch": 0.14995579309180135, "grad_norm": 0.938944935798645, "learning_rate": 1.9254827436606698e-05, "loss": 0.18073031902313233, "step": 30190 }, { "epoch": 0.15000546377517063, "grad_norm": 1.6846219301223755, "learning_rate": 1.92542179525551e-05, "loss": 0.17660655975341796, "step": 30200 }, { "epoch": 0.15005513445853988, "grad_norm": 1.222695231437683, "learning_rate": 1.9253608229008066e-05, "loss": 0.17508383989334106, "step": 30210 }, { "epoch": 0.15010480514190913, "grad_norm": 1.3527007102966309, "learning_rate": 1.9252998265981367e-05, "loss": 0.17298078536987305, "step": 30220 }, { "epoch": 0.1501544758252784, "grad_norm": 0.8823666572570801, "learning_rate": 1.9252388063490797e-05, "loss": 0.17066879272460939, "step": 30230 }, { "epoch": 0.15020414650864766, "grad_norm": 1.88309645652771, "learning_rate": 1.9251777621552144e-05, "loss": 0.19140713214874266, "step": 30240 }, { "epoch": 0.15025381719201691, "grad_norm": 0.9409096240997314, "learning_rate": 1.9251166940181204e-05, "loss": 0.17455973625183105, "step": 30250 }, { "epoch": 0.1503034878753862, "grad_norm": 1.2533543109893799, "learning_rate": 1.9250556019393786e-05, "loss": 0.18845200538635254, "step": 30260 }, { "epoch": 0.15035315855875545, "grad_norm": 1.42643141746521, "learning_rate": 1.9249944859205692e-05, "loss": 0.17497649192810058, "step": 30270 }, { "epoch": 0.15040282924212472, "grad_norm": 1.363965392112732, "learning_rate": 1.9249333459632748e-05, "loss": 0.1911354184150696, "step": 30280 }, { "epoch": 0.15045249992549398, "grad_norm": 1.0770291090011597, "learning_rate": 1.9248721820690773e-05, "loss": 0.17422525882720946, "step": 30290 }, { "epoch": 0.15050217060886323, "grad_norm": 1.2682362794876099, "learning_rate": 1.9248109942395597e-05, "loss": 0.20069868564605714, "step": 30300 }, { "epoch": 0.1505518412922325, "grad_norm": 1.137071132659912, "learning_rate": 1.9247497824763055e-05, "loss": 0.17358014583587647, "step": 30310 }, { "epoch": 0.15060151197560176, "grad_norm": 3.1216273307800293, "learning_rate": 1.9246885467808988e-05, "loss": 0.18457597494125366, "step": 30320 }, { "epoch": 0.150651182658971, "grad_norm": 1.677504539489746, "learning_rate": 1.9246272871549238e-05, "loss": 0.19335274696350097, "step": 30330 }, { "epoch": 0.1507008533423403, "grad_norm": 1.9146424531936646, "learning_rate": 1.9245660035999668e-05, "loss": 0.18262960910797119, "step": 30340 }, { "epoch": 0.15075052402570954, "grad_norm": 1.6708693504333496, "learning_rate": 1.924504696117613e-05, "loss": 0.17737715244293212, "step": 30350 }, { "epoch": 0.1508001947090788, "grad_norm": 1.3415822982788086, "learning_rate": 1.92444336470945e-05, "loss": 0.17859878540039062, "step": 30360 }, { "epoch": 0.15084986539244807, "grad_norm": 1.1048039197921753, "learning_rate": 1.924382009377064e-05, "loss": 0.18422672748565674, "step": 30370 }, { "epoch": 0.15089953607581733, "grad_norm": 2.4735755920410156, "learning_rate": 1.9243206301220432e-05, "loss": 0.1905667304992676, "step": 30380 }, { "epoch": 0.1509492067591866, "grad_norm": 1.117459774017334, "learning_rate": 1.9242592269459767e-05, "loss": 0.16653261184692383, "step": 30390 }, { "epoch": 0.15099887744255586, "grad_norm": 2.922992706298828, "learning_rate": 1.9241977998504526e-05, "loss": 0.18810546398162842, "step": 30400 }, { "epoch": 0.1510485481259251, "grad_norm": 1.4133481979370117, "learning_rate": 1.924136348837061e-05, "loss": 0.17612862586975098, "step": 30410 }, { "epoch": 0.1510982188092944, "grad_norm": 1.7284700870513916, "learning_rate": 1.924074873907392e-05, "loss": 0.1773705840110779, "step": 30420 }, { "epoch": 0.15114788949266364, "grad_norm": 1.1952792406082153, "learning_rate": 1.9240133750630375e-05, "loss": 0.19439353942871093, "step": 30430 }, { "epoch": 0.1511975601760329, "grad_norm": 1.0497078895568848, "learning_rate": 1.923951852305588e-05, "loss": 0.18763574361801147, "step": 30440 }, { "epoch": 0.15124723085940217, "grad_norm": 1.5083017349243164, "learning_rate": 1.923890305636636e-05, "loss": 0.18892643451690674, "step": 30450 }, { "epoch": 0.15129690154277142, "grad_norm": 1.3791660070419312, "learning_rate": 1.9238287350577748e-05, "loss": 0.173024582862854, "step": 30460 }, { "epoch": 0.15134657222614067, "grad_norm": 1.7377500534057617, "learning_rate": 1.9237671405705967e-05, "loss": 0.19368101358413697, "step": 30470 }, { "epoch": 0.15139624290950995, "grad_norm": 0.9820003509521484, "learning_rate": 1.923705522176697e-05, "loss": 0.21241025924682616, "step": 30480 }, { "epoch": 0.1514459135928792, "grad_norm": 0.9413429498672485, "learning_rate": 1.92364387987767e-05, "loss": 0.17119388580322265, "step": 30490 }, { "epoch": 0.15149558427624849, "grad_norm": 1.8599913120269775, "learning_rate": 1.92358221367511e-05, "loss": 0.18377573490142823, "step": 30500 }, { "epoch": 0.15154525495961774, "grad_norm": 1.4672582149505615, "learning_rate": 1.923520523570614e-05, "loss": 0.18045129776000976, "step": 30510 }, { "epoch": 0.151594925642987, "grad_norm": 1.187127709388733, "learning_rate": 1.9234588095657783e-05, "loss": 0.1902700185775757, "step": 30520 }, { "epoch": 0.15164459632635627, "grad_norm": 0.8362231254577637, "learning_rate": 1.9233970716622e-05, "loss": 0.18174395561218262, "step": 30530 }, { "epoch": 0.15169426700972552, "grad_norm": 1.785758137702942, "learning_rate": 1.9233353098614763e-05, "loss": 0.17710726261138915, "step": 30540 }, { "epoch": 0.15174393769309477, "grad_norm": 1.5369702577590942, "learning_rate": 1.9232735241652065e-05, "loss": 0.19000710248947145, "step": 30550 }, { "epoch": 0.15179360837646405, "grad_norm": 0.7465130686759949, "learning_rate": 1.9232117145749893e-05, "loss": 0.1753772258758545, "step": 30560 }, { "epoch": 0.1518432790598333, "grad_norm": 1.5811328887939453, "learning_rate": 1.9231498810924236e-05, "loss": 0.18271673917770387, "step": 30570 }, { "epoch": 0.15189294974320255, "grad_norm": 1.3455289602279663, "learning_rate": 1.9230880237191104e-05, "loss": 0.17908705472946168, "step": 30580 }, { "epoch": 0.15194262042657183, "grad_norm": 1.8022024631500244, "learning_rate": 1.9230261424566504e-05, "loss": 0.17100372314453124, "step": 30590 }, { "epoch": 0.15199229110994109, "grad_norm": 1.3060041666030884, "learning_rate": 1.9229642373066447e-05, "loss": 0.18380398750305177, "step": 30600 }, { "epoch": 0.15204196179331037, "grad_norm": 0.9077751636505127, "learning_rate": 1.9229023082706958e-05, "loss": 0.18049012422561644, "step": 30610 }, { "epoch": 0.15209163247667962, "grad_norm": 1.0538480281829834, "learning_rate": 1.9228403553504066e-05, "loss": 0.1788811445236206, "step": 30620 }, { "epoch": 0.15214130316004887, "grad_norm": 0.9301947951316833, "learning_rate": 1.9227783785473794e-05, "loss": 0.1763114333152771, "step": 30630 }, { "epoch": 0.15219097384341815, "grad_norm": 1.395597219467163, "learning_rate": 1.922716377863219e-05, "loss": 0.1841686487197876, "step": 30640 }, { "epoch": 0.1522406445267874, "grad_norm": 1.8401192426681519, "learning_rate": 1.92265435329953e-05, "loss": 0.1877527952194214, "step": 30650 }, { "epoch": 0.15229031521015665, "grad_norm": 1.798919916152954, "learning_rate": 1.922592304857917e-05, "loss": 0.18733577728271483, "step": 30660 }, { "epoch": 0.15233998589352593, "grad_norm": 1.737995982170105, "learning_rate": 1.9225302325399866e-05, "loss": 0.17102425098419188, "step": 30670 }, { "epoch": 0.15238965657689518, "grad_norm": 1.0287480354309082, "learning_rate": 1.9224681363473442e-05, "loss": 0.19127682447433472, "step": 30680 }, { "epoch": 0.15243932726026443, "grad_norm": 1.7818373441696167, "learning_rate": 1.9224060162815977e-05, "loss": 0.17494261264801025, "step": 30690 }, { "epoch": 0.15248899794363371, "grad_norm": 1.6713565587997437, "learning_rate": 1.9223438723443542e-05, "loss": 0.20550334453582764, "step": 30700 }, { "epoch": 0.15253866862700297, "grad_norm": 1.0139029026031494, "learning_rate": 1.9222817045372223e-05, "loss": 0.18878278732299805, "step": 30710 }, { "epoch": 0.15258833931037225, "grad_norm": 1.350873351097107, "learning_rate": 1.9222195128618108e-05, "loss": 0.1779549837112427, "step": 30720 }, { "epoch": 0.1526380099937415, "grad_norm": 1.7045657634735107, "learning_rate": 1.9221572973197284e-05, "loss": 0.16414854526519776, "step": 30730 }, { "epoch": 0.15268768067711075, "grad_norm": 1.2878644466400146, "learning_rate": 1.9220950579125866e-05, "loss": 0.16813648939132692, "step": 30740 }, { "epoch": 0.15273735136048003, "grad_norm": 0.8476606607437134, "learning_rate": 1.922032794641995e-05, "loss": 0.18429145812988282, "step": 30750 }, { "epoch": 0.15278702204384928, "grad_norm": 3.4657914638519287, "learning_rate": 1.921970507509566e-05, "loss": 0.20293951034545898, "step": 30760 }, { "epoch": 0.15283669272721853, "grad_norm": 1.661055326461792, "learning_rate": 1.9219081965169105e-05, "loss": 0.1755720853805542, "step": 30770 }, { "epoch": 0.1528863634105878, "grad_norm": 1.7207813262939453, "learning_rate": 1.921845861665642e-05, "loss": 0.1799335479736328, "step": 30780 }, { "epoch": 0.15293603409395706, "grad_norm": 0.9905157089233398, "learning_rate": 1.921783502957373e-05, "loss": 0.20032691955566406, "step": 30790 }, { "epoch": 0.15298570477732631, "grad_norm": 1.2957117557525635, "learning_rate": 1.9217211203937175e-05, "loss": 0.169264018535614, "step": 30800 }, { "epoch": 0.1530353754606956, "grad_norm": 0.7385703325271606, "learning_rate": 1.9216587139762902e-05, "loss": 0.1729485034942627, "step": 30810 }, { "epoch": 0.15308504614406485, "grad_norm": 0.554045557975769, "learning_rate": 1.921596283706706e-05, "loss": 0.18433015346527098, "step": 30820 }, { "epoch": 0.15313471682743413, "grad_norm": 1.748764991760254, "learning_rate": 1.9215338295865803e-05, "loss": 0.17049145698547363, "step": 30830 }, { "epoch": 0.15318438751080338, "grad_norm": 0.982075035572052, "learning_rate": 1.9214713516175295e-05, "loss": 0.17030551433563232, "step": 30840 }, { "epoch": 0.15323405819417263, "grad_norm": 0.9516870975494385, "learning_rate": 1.921408849801171e-05, "loss": 0.17035231590270997, "step": 30850 }, { "epoch": 0.1532837288775419, "grad_norm": 2.04606556892395, "learning_rate": 1.9213463241391217e-05, "loss": 0.1858925700187683, "step": 30860 }, { "epoch": 0.15333339956091116, "grad_norm": 2.3653109073638916, "learning_rate": 1.921283774633e-05, "loss": 0.18659306764602662, "step": 30870 }, { "epoch": 0.1533830702442804, "grad_norm": 2.078486442565918, "learning_rate": 1.921221201284425e-05, "loss": 0.1938789129257202, "step": 30880 }, { "epoch": 0.1534327409276497, "grad_norm": 1.9104245901107788, "learning_rate": 1.9211586040950153e-05, "loss": 0.18881731033325194, "step": 30890 }, { "epoch": 0.15348241161101894, "grad_norm": 3.274637222290039, "learning_rate": 1.9210959830663914e-05, "loss": 0.18151581287384033, "step": 30900 }, { "epoch": 0.1535320822943882, "grad_norm": 1.4747066497802734, "learning_rate": 1.9210333382001737e-05, "loss": 0.1840034008026123, "step": 30910 }, { "epoch": 0.15358175297775747, "grad_norm": 1.8020182847976685, "learning_rate": 1.920970669497984e-05, "loss": 0.18083311319351197, "step": 30920 }, { "epoch": 0.15363142366112673, "grad_norm": 1.7022502422332764, "learning_rate": 1.9209079769614432e-05, "loss": 0.18025503158569336, "step": 30930 }, { "epoch": 0.153681094344496, "grad_norm": 3.5108425617218018, "learning_rate": 1.9208452605921745e-05, "loss": 0.18585768938064576, "step": 30940 }, { "epoch": 0.15373076502786526, "grad_norm": 1.3883860111236572, "learning_rate": 1.9207825203918006e-05, "loss": 0.18264875411987305, "step": 30950 }, { "epoch": 0.1537804357112345, "grad_norm": 0.9051364064216614, "learning_rate": 1.9207197563619453e-05, "loss": 0.17123143672943114, "step": 30960 }, { "epoch": 0.1538301063946038, "grad_norm": 0.7188379168510437, "learning_rate": 1.920656968504233e-05, "loss": 0.17519741058349608, "step": 30970 }, { "epoch": 0.15387977707797304, "grad_norm": 0.9967820644378662, "learning_rate": 1.9205941568202885e-05, "loss": 0.1790795087814331, "step": 30980 }, { "epoch": 0.1539294477613423, "grad_norm": 1.4432291984558105, "learning_rate": 1.9205313213117375e-05, "loss": 0.18736035823822023, "step": 30990 }, { "epoch": 0.15397911844471157, "grad_norm": 0.7114903926849365, "learning_rate": 1.920468461980206e-05, "loss": 0.17285934686660767, "step": 31000 }, { "epoch": 0.15402878912808082, "grad_norm": 3.8073573112487793, "learning_rate": 1.9204055788273206e-05, "loss": 0.1655641794204712, "step": 31010 }, { "epoch": 0.15407845981145007, "grad_norm": 1.3914271593093872, "learning_rate": 1.920342671854709e-05, "loss": 0.16825408935546876, "step": 31020 }, { "epoch": 0.15412813049481935, "grad_norm": 1.0580376386642456, "learning_rate": 1.920279741063999e-05, "loss": 0.17130987644195556, "step": 31030 }, { "epoch": 0.1541778011781886, "grad_norm": 1.0446377992630005, "learning_rate": 1.9202167864568197e-05, "loss": 0.1902665376663208, "step": 31040 }, { "epoch": 0.15422747186155786, "grad_norm": 1.3625648021697998, "learning_rate": 1.9201538080347997e-05, "loss": 0.18166687488555908, "step": 31050 }, { "epoch": 0.15427714254492714, "grad_norm": 2.4037466049194336, "learning_rate": 1.920090805799569e-05, "loss": 0.19982825517654418, "step": 31060 }, { "epoch": 0.1543268132282964, "grad_norm": 1.0571315288543701, "learning_rate": 1.9200277797527585e-05, "loss": 0.17774534225463867, "step": 31070 }, { "epoch": 0.15437648391166567, "grad_norm": 0.6833314895629883, "learning_rate": 1.9199647298959988e-05, "loss": 0.18059628009796141, "step": 31080 }, { "epoch": 0.15442615459503492, "grad_norm": 2.148792266845703, "learning_rate": 1.9199016562309216e-05, "loss": 0.17333029508590697, "step": 31090 }, { "epoch": 0.15447582527840417, "grad_norm": 1.3110660314559937, "learning_rate": 1.91983855875916e-05, "loss": 0.18903459310531617, "step": 31100 }, { "epoch": 0.15452549596177345, "grad_norm": 1.4669591188430786, "learning_rate": 1.9197754374823455e-05, "loss": 0.1830852746963501, "step": 31110 }, { "epoch": 0.1545751666451427, "grad_norm": 1.1610482931137085, "learning_rate": 1.9197122924021134e-05, "loss": 0.18200154304504396, "step": 31120 }, { "epoch": 0.15462483732851195, "grad_norm": 1.231818675994873, "learning_rate": 1.9196491235200965e-05, "loss": 0.17128503322601318, "step": 31130 }, { "epoch": 0.15467450801188123, "grad_norm": 1.155195951461792, "learning_rate": 1.91958593083793e-05, "loss": 0.16955525875091554, "step": 31140 }, { "epoch": 0.15472417869525049, "grad_norm": 1.8624603748321533, "learning_rate": 1.9195227143572497e-05, "loss": 0.16803102493286132, "step": 31150 }, { "epoch": 0.15477384937861974, "grad_norm": 1.0621349811553955, "learning_rate": 1.919459474079691e-05, "loss": 0.18395031690597535, "step": 31160 }, { "epoch": 0.15482352006198902, "grad_norm": 1.7397234439849854, "learning_rate": 1.919396210006891e-05, "loss": 0.16711647510528566, "step": 31170 }, { "epoch": 0.15487319074535827, "grad_norm": 1.0650732517242432, "learning_rate": 1.9193329221404865e-05, "loss": 0.17562283277511598, "step": 31180 }, { "epoch": 0.15492286142872755, "grad_norm": 1.6780188083648682, "learning_rate": 1.9192696104821156e-05, "loss": 0.1917824625968933, "step": 31190 }, { "epoch": 0.1549725321120968, "grad_norm": 1.6653393507003784, "learning_rate": 1.919206275033417e-05, "loss": 0.19755531549453736, "step": 31200 }, { "epoch": 0.15502220279546605, "grad_norm": 1.1360599994659424, "learning_rate": 1.9191429157960298e-05, "loss": 0.17622662782669068, "step": 31210 }, { "epoch": 0.15507187347883533, "grad_norm": 1.8088346719741821, "learning_rate": 1.919079532771593e-05, "loss": 0.18105952739715575, "step": 31220 }, { "epoch": 0.15512154416220458, "grad_norm": 1.412184715270996, "learning_rate": 1.919016125961748e-05, "loss": 0.1672229290008545, "step": 31230 }, { "epoch": 0.15517121484557383, "grad_norm": 1.933139443397522, "learning_rate": 1.918952695368135e-05, "loss": 0.17887916564941406, "step": 31240 }, { "epoch": 0.15522088552894311, "grad_norm": 1.2164552211761475, "learning_rate": 1.9188892409923953e-05, "loss": 0.18119606971740723, "step": 31250 }, { "epoch": 0.15527055621231237, "grad_norm": 1.508927583694458, "learning_rate": 1.918825762836172e-05, "loss": 0.16054131984710693, "step": 31260 }, { "epoch": 0.15532022689568162, "grad_norm": 2.0459532737731934, "learning_rate": 1.918762260901107e-05, "loss": 0.19396342039108277, "step": 31270 }, { "epoch": 0.1553698975790509, "grad_norm": 1.896201491355896, "learning_rate": 1.9186987351888443e-05, "loss": 0.179390025138855, "step": 31280 }, { "epoch": 0.15541956826242015, "grad_norm": 1.3165534734725952, "learning_rate": 1.9186351857010283e-05, "loss": 0.16650114059448243, "step": 31290 }, { "epoch": 0.15546923894578943, "grad_norm": 1.1413347721099854, "learning_rate": 1.918571612439302e-05, "loss": 0.1822476863861084, "step": 31300 }, { "epoch": 0.15551890962915868, "grad_norm": 2.1700356006622314, "learning_rate": 1.9185080154053124e-05, "loss": 0.18521342277526856, "step": 31310 }, { "epoch": 0.15556858031252793, "grad_norm": 3.756653070449829, "learning_rate": 1.9184443946007046e-05, "loss": 0.19456934928894043, "step": 31320 }, { "epoch": 0.1556182509958972, "grad_norm": 0.7917903661727905, "learning_rate": 1.918380750027125e-05, "loss": 0.19854917526245117, "step": 31330 }, { "epoch": 0.15566792167926646, "grad_norm": 0.7600002288818359, "learning_rate": 1.918317081686221e-05, "loss": 0.1811656355857849, "step": 31340 }, { "epoch": 0.15571759236263572, "grad_norm": 1.8711321353912354, "learning_rate": 1.9182533895796402e-05, "loss": 0.18932151794433594, "step": 31350 }, { "epoch": 0.155767263046005, "grad_norm": 0.8390267491340637, "learning_rate": 1.918189673709031e-05, "loss": 0.19100207090377808, "step": 31360 }, { "epoch": 0.15581693372937425, "grad_norm": 1.665611982345581, "learning_rate": 1.918125934076042e-05, "loss": 0.17652209997177123, "step": 31370 }, { "epoch": 0.1558666044127435, "grad_norm": 0.9708449244499207, "learning_rate": 1.918062170682323e-05, "loss": 0.19409162998199464, "step": 31380 }, { "epoch": 0.15591627509611278, "grad_norm": 2.5586647987365723, "learning_rate": 1.9179983835295242e-05, "loss": 0.18056762218475342, "step": 31390 }, { "epoch": 0.15596594577948203, "grad_norm": 2.2065510749816895, "learning_rate": 1.917934572619296e-05, "loss": 0.1791752338409424, "step": 31400 }, { "epoch": 0.1560156164628513, "grad_norm": 1.1623430252075195, "learning_rate": 1.9178707379532904e-05, "loss": 0.17436981201171875, "step": 31410 }, { "epoch": 0.15606528714622056, "grad_norm": 1.6489211320877075, "learning_rate": 1.9178068795331595e-05, "loss": 0.18170504570007323, "step": 31420 }, { "epoch": 0.1561149578295898, "grad_norm": 0.8251368999481201, "learning_rate": 1.917742997360555e-05, "loss": 0.1811108946800232, "step": 31430 }, { "epoch": 0.1561646285129591, "grad_norm": 1.9250181913375854, "learning_rate": 1.9176790914371316e-05, "loss": 0.19029222726821898, "step": 31440 }, { "epoch": 0.15621429919632834, "grad_norm": 1.2230770587921143, "learning_rate": 1.9176151617645414e-05, "loss": 0.17161778211593628, "step": 31450 }, { "epoch": 0.1562639698796976, "grad_norm": 2.2743842601776123, "learning_rate": 1.9175512083444403e-05, "loss": 0.187885582447052, "step": 31460 }, { "epoch": 0.15631364056306687, "grad_norm": 2.4008615016937256, "learning_rate": 1.9174872311784827e-05, "loss": 0.19083185195922853, "step": 31470 }, { "epoch": 0.15636331124643613, "grad_norm": 1.261854648590088, "learning_rate": 1.9174232302683244e-05, "loss": 0.18331468105316162, "step": 31480 }, { "epoch": 0.15641298192980538, "grad_norm": 1.4333752393722534, "learning_rate": 1.9173592056156218e-05, "loss": 0.17831627130508423, "step": 31490 }, { "epoch": 0.15646265261317466, "grad_norm": 2.10986590385437, "learning_rate": 1.917295157222032e-05, "loss": 0.20501739978790284, "step": 31500 }, { "epoch": 0.1565123232965439, "grad_norm": 1.1112494468688965, "learning_rate": 1.917231085089212e-05, "loss": 0.18068301677703857, "step": 31510 }, { "epoch": 0.1565619939799132, "grad_norm": 1.060528039932251, "learning_rate": 1.9171669892188206e-05, "loss": 0.1850225806236267, "step": 31520 }, { "epoch": 0.15661166466328244, "grad_norm": 1.566674828529358, "learning_rate": 1.917102869612516e-05, "loss": 0.18044484853744508, "step": 31530 }, { "epoch": 0.1566613353466517, "grad_norm": 2.728663206100464, "learning_rate": 1.9170387262719584e-05, "loss": 0.19007432460784912, "step": 31540 }, { "epoch": 0.15671100603002097, "grad_norm": 1.059122920036316, "learning_rate": 1.916974559198807e-05, "loss": 0.17995609045028688, "step": 31550 }, { "epoch": 0.15676067671339022, "grad_norm": 1.351833462715149, "learning_rate": 1.916910368394723e-05, "loss": 0.1720120668411255, "step": 31560 }, { "epoch": 0.15681034739675948, "grad_norm": 1.186326503753662, "learning_rate": 1.916846153861367e-05, "loss": 0.1815586805343628, "step": 31570 }, { "epoch": 0.15686001808012875, "grad_norm": 0.8508137464523315, "learning_rate": 1.9167819156004013e-05, "loss": 0.17947397232055665, "step": 31580 }, { "epoch": 0.156909688763498, "grad_norm": 1.263238549232483, "learning_rate": 1.916717653613488e-05, "loss": 0.1776526927947998, "step": 31590 }, { "epoch": 0.15695935944686726, "grad_norm": 0.9647364020347595, "learning_rate": 1.916653367902291e-05, "loss": 0.18886065483093262, "step": 31600 }, { "epoch": 0.15700903013023654, "grad_norm": 0.8768201470375061, "learning_rate": 1.9165890584684727e-05, "loss": 0.1726974368095398, "step": 31610 }, { "epoch": 0.1570587008136058, "grad_norm": 1.1525914669036865, "learning_rate": 1.9165247253136985e-05, "loss": 0.18612315654754638, "step": 31620 }, { "epoch": 0.15710837149697507, "grad_norm": 1.3895018100738525, "learning_rate": 1.9164603684396327e-05, "loss": 0.17919036149978637, "step": 31630 }, { "epoch": 0.15715804218034432, "grad_norm": 1.4674108028411865, "learning_rate": 1.9163959878479414e-05, "loss": 0.18211793899536133, "step": 31640 }, { "epoch": 0.15720771286371357, "grad_norm": 2.0311989784240723, "learning_rate": 1.91633158354029e-05, "loss": 0.18060901165008544, "step": 31650 }, { "epoch": 0.15725738354708285, "grad_norm": 2.5636157989501953, "learning_rate": 1.9162671555183462e-05, "loss": 0.1909271001815796, "step": 31660 }, { "epoch": 0.1573070542304521, "grad_norm": 1.0860294103622437, "learning_rate": 1.9162027037837766e-05, "loss": 0.1787118673324585, "step": 31670 }, { "epoch": 0.15735672491382136, "grad_norm": 3.0715529918670654, "learning_rate": 1.916138228338249e-05, "loss": 0.17777199745178224, "step": 31680 }, { "epoch": 0.15740639559719063, "grad_norm": 0.969161331653595, "learning_rate": 1.916073729183433e-05, "loss": 0.1671215534210205, "step": 31690 }, { "epoch": 0.1574560662805599, "grad_norm": 1.340391755104065, "learning_rate": 1.9160092063209967e-05, "loss": 0.18400416374206544, "step": 31700 }, { "epoch": 0.15750573696392914, "grad_norm": 1.6325910091400146, "learning_rate": 1.915944659752611e-05, "loss": 0.19063174724578857, "step": 31710 }, { "epoch": 0.15755540764729842, "grad_norm": 1.1106430292129517, "learning_rate": 1.9158800894799453e-05, "loss": 0.17943472862243653, "step": 31720 }, { "epoch": 0.15760507833066767, "grad_norm": 0.9361312985420227, "learning_rate": 1.9158154955046714e-05, "loss": 0.19051365852355956, "step": 31730 }, { "epoch": 0.15765474901403695, "grad_norm": 0.8423734903335571, "learning_rate": 1.915750877828461e-05, "loss": 0.17509266138076782, "step": 31740 }, { "epoch": 0.1577044196974062, "grad_norm": 0.9306275844573975, "learning_rate": 1.9156862364529854e-05, "loss": 0.1971269130706787, "step": 31750 }, { "epoch": 0.15775409038077545, "grad_norm": 1.2219033241271973, "learning_rate": 1.9156215713799187e-05, "loss": 0.1809842586517334, "step": 31760 }, { "epoch": 0.15780376106414473, "grad_norm": 1.625565528869629, "learning_rate": 1.9155568826109335e-05, "loss": 0.18109641075134278, "step": 31770 }, { "epoch": 0.15785343174751398, "grad_norm": 2.001190185546875, "learning_rate": 1.9154921701477046e-05, "loss": 0.18524680137634278, "step": 31780 }, { "epoch": 0.15790310243088324, "grad_norm": 4.427062511444092, "learning_rate": 1.9154274339919065e-05, "loss": 0.16424754858016968, "step": 31790 }, { "epoch": 0.15795277311425251, "grad_norm": 2.1764891147613525, "learning_rate": 1.9153626741452144e-05, "loss": 0.1905003309249878, "step": 31800 }, { "epoch": 0.15800244379762177, "grad_norm": 1.391195297241211, "learning_rate": 1.9152978906093042e-05, "loss": 0.18495044708251954, "step": 31810 }, { "epoch": 0.15805211448099102, "grad_norm": 2.9447710514068604, "learning_rate": 1.9152330833858527e-05, "loss": 0.18467413187026976, "step": 31820 }, { "epoch": 0.1581017851643603, "grad_norm": 0.6556002497673035, "learning_rate": 1.915168252476537e-05, "loss": 0.17802197933197023, "step": 31830 }, { "epoch": 0.15815145584772955, "grad_norm": 2.0109307765960693, "learning_rate": 1.9151033978830353e-05, "loss": 0.18424841165542602, "step": 31840 }, { "epoch": 0.1582011265310988, "grad_norm": 1.3274353742599487, "learning_rate": 1.9150385196070253e-05, "loss": 0.18574817180633546, "step": 31850 }, { "epoch": 0.15825079721446808, "grad_norm": 1.717194676399231, "learning_rate": 1.914973617650186e-05, "loss": 0.1724240303039551, "step": 31860 }, { "epoch": 0.15830046789783733, "grad_norm": 1.1512242555618286, "learning_rate": 1.9149086920141978e-05, "loss": 0.18579654693603515, "step": 31870 }, { "epoch": 0.1583501385812066, "grad_norm": 1.1771050691604614, "learning_rate": 1.9148437427007404e-05, "loss": 0.17654727697372435, "step": 31880 }, { "epoch": 0.15839980926457586, "grad_norm": 3.7786693572998047, "learning_rate": 1.9147787697114947e-05, "loss": 0.1913233995437622, "step": 31890 }, { "epoch": 0.15844947994794512, "grad_norm": 1.4095184803009033, "learning_rate": 1.9147137730481423e-05, "loss": 0.17932530641555786, "step": 31900 }, { "epoch": 0.1584991506313144, "grad_norm": 1.861075520515442, "learning_rate": 1.914648752712365e-05, "loss": 0.18632793426513672, "step": 31910 }, { "epoch": 0.15854882131468365, "grad_norm": 2.1545066833496094, "learning_rate": 1.914583708705846e-05, "loss": 0.17825798988342284, "step": 31920 }, { "epoch": 0.1585984919980529, "grad_norm": 0.994408369064331, "learning_rate": 1.9145186410302684e-05, "loss": 0.19468388557434083, "step": 31930 }, { "epoch": 0.15864816268142218, "grad_norm": 1.1122175455093384, "learning_rate": 1.914453549687316e-05, "loss": 0.190224552154541, "step": 31940 }, { "epoch": 0.15869783336479143, "grad_norm": 0.7607638239860535, "learning_rate": 1.9143884346786735e-05, "loss": 0.18486371040344238, "step": 31950 }, { "epoch": 0.15874750404816068, "grad_norm": 1.2561782598495483, "learning_rate": 1.9143232960060262e-05, "loss": 0.18175833225250243, "step": 31960 }, { "epoch": 0.15879717473152996, "grad_norm": 1.3710933923721313, "learning_rate": 1.9142581336710592e-05, "loss": 0.1694579839706421, "step": 31970 }, { "epoch": 0.1588468454148992, "grad_norm": 1.8608254194259644, "learning_rate": 1.9141929476754595e-05, "loss": 0.18412580490112304, "step": 31980 }, { "epoch": 0.1588965160982685, "grad_norm": 0.8631646037101746, "learning_rate": 1.914127738020914e-05, "loss": 0.1771206736564636, "step": 31990 }, { "epoch": 0.15894618678163774, "grad_norm": 2.421048879623413, "learning_rate": 1.9140625047091098e-05, "loss": 0.1745534658432007, "step": 32000 }, { "epoch": 0.158995857465007, "grad_norm": 2.070504903793335, "learning_rate": 1.9139972477417357e-05, "loss": 0.18823899030685426, "step": 32010 }, { "epoch": 0.15904552814837628, "grad_norm": 1.0706405639648438, "learning_rate": 1.9139319671204805e-05, "loss": 0.17847247123718263, "step": 32020 }, { "epoch": 0.15909519883174553, "grad_norm": 1.3528776168823242, "learning_rate": 1.9138666628470336e-05, "loss": 0.19043418169021606, "step": 32030 }, { "epoch": 0.15914486951511478, "grad_norm": 0.9744521379470825, "learning_rate": 1.9138013349230845e-05, "loss": 0.18586376905441285, "step": 32040 }, { "epoch": 0.15919454019848406, "grad_norm": 1.5026330947875977, "learning_rate": 1.9137359833503245e-05, "loss": 0.18349807262420653, "step": 32050 }, { "epoch": 0.1592442108818533, "grad_norm": 1.2603358030319214, "learning_rate": 1.9136706081304445e-05, "loss": 0.18299068212509156, "step": 32060 }, { "epoch": 0.15929388156522256, "grad_norm": 3.1955177783966064, "learning_rate": 1.913605209265137e-05, "loss": 0.18095796108245848, "step": 32070 }, { "epoch": 0.15934355224859184, "grad_norm": 1.5214195251464844, "learning_rate": 1.9135397867560935e-05, "loss": 0.17610390186309816, "step": 32080 }, { "epoch": 0.1593932229319611, "grad_norm": 1.9952763319015503, "learning_rate": 1.9134743406050077e-05, "loss": 0.1932172656059265, "step": 32090 }, { "epoch": 0.15944289361533037, "grad_norm": 1.3251841068267822, "learning_rate": 1.9134088708135733e-05, "loss": 0.18833695650100707, "step": 32100 }, { "epoch": 0.15949256429869962, "grad_norm": 1.0848745107650757, "learning_rate": 1.913343377383485e-05, "loss": 0.1890195608139038, "step": 32110 }, { "epoch": 0.15954223498206888, "grad_norm": 0.7570167183876038, "learning_rate": 1.913277860316437e-05, "loss": 0.17356401681900024, "step": 32120 }, { "epoch": 0.15959190566543816, "grad_norm": 1.7299515008926392, "learning_rate": 1.9132123196141252e-05, "loss": 0.18564200401306152, "step": 32130 }, { "epoch": 0.1596415763488074, "grad_norm": 1.1402881145477295, "learning_rate": 1.9131467552782457e-05, "loss": 0.1896491527557373, "step": 32140 }, { "epoch": 0.15969124703217666, "grad_norm": 1.6371217966079712, "learning_rate": 1.9130811673104953e-05, "loss": 0.16480910778045654, "step": 32150 }, { "epoch": 0.15974091771554594, "grad_norm": 1.287414789199829, "learning_rate": 1.9130155557125715e-05, "loss": 0.18328554630279542, "step": 32160 }, { "epoch": 0.1597905883989152, "grad_norm": 1.3982394933700562, "learning_rate": 1.9129499204861723e-05, "loss": 0.18430463075637818, "step": 32170 }, { "epoch": 0.15984025908228444, "grad_norm": 2.018392562866211, "learning_rate": 1.9128842616329962e-05, "loss": 0.18434813022613525, "step": 32180 }, { "epoch": 0.15988992976565372, "grad_norm": 1.1275593042373657, "learning_rate": 1.9128185791547424e-05, "loss": 0.18732293844223022, "step": 32190 }, { "epoch": 0.15993960044902297, "grad_norm": 1.3643618822097778, "learning_rate": 1.9127528730531107e-05, "loss": 0.1872736930847168, "step": 32200 }, { "epoch": 0.15998927113239225, "grad_norm": 1.337855339050293, "learning_rate": 1.912687143329802e-05, "loss": 0.17955923080444336, "step": 32210 }, { "epoch": 0.1600389418157615, "grad_norm": 1.6132017374038696, "learning_rate": 1.9126213899865166e-05, "loss": 0.2000173568725586, "step": 32220 }, { "epoch": 0.16008861249913076, "grad_norm": 1.5213998556137085, "learning_rate": 1.9125556130249566e-05, "loss": 0.18242554664611815, "step": 32230 }, { "epoch": 0.16013828318250004, "grad_norm": 0.9829002022743225, "learning_rate": 1.9124898124468244e-05, "loss": 0.1790795922279358, "step": 32240 }, { "epoch": 0.1601879538658693, "grad_norm": 1.7745343446731567, "learning_rate": 1.912423988253823e-05, "loss": 0.16791833639144899, "step": 32250 }, { "epoch": 0.16023762454923854, "grad_norm": 1.315213918685913, "learning_rate": 1.9123581404476554e-05, "loss": 0.18160055875778197, "step": 32260 }, { "epoch": 0.16028729523260782, "grad_norm": 1.243782639503479, "learning_rate": 1.9122922690300257e-05, "loss": 0.18018083572387694, "step": 32270 }, { "epoch": 0.16033696591597707, "grad_norm": 1.2521171569824219, "learning_rate": 1.9122263740026394e-05, "loss": 0.1765294313430786, "step": 32280 }, { "epoch": 0.16038663659934632, "grad_norm": 0.9073197841644287, "learning_rate": 1.912160455367201e-05, "loss": 0.17344331741333008, "step": 32290 }, { "epoch": 0.1604363072827156, "grad_norm": 1.7462186813354492, "learning_rate": 1.912094513125417e-05, "loss": 0.179267954826355, "step": 32300 }, { "epoch": 0.16048597796608485, "grad_norm": 0.9867767691612244, "learning_rate": 1.9120285472789938e-05, "loss": 0.17767801284790039, "step": 32310 }, { "epoch": 0.16053564864945413, "grad_norm": 1.564353585243225, "learning_rate": 1.9119625578296384e-05, "loss": 0.19089958667755128, "step": 32320 }, { "epoch": 0.16058531933282338, "grad_norm": 1.4107789993286133, "learning_rate": 1.9118965447790587e-05, "loss": 0.18324530124664307, "step": 32330 }, { "epoch": 0.16063499001619264, "grad_norm": 2.4909753799438477, "learning_rate": 1.9118305081289626e-05, "loss": 0.17411831617355347, "step": 32340 }, { "epoch": 0.16068466069956192, "grad_norm": 0.6721743941307068, "learning_rate": 1.9117644478810603e-05, "loss": 0.17268500328063965, "step": 32350 }, { "epoch": 0.16073433138293117, "grad_norm": 1.4926255941390991, "learning_rate": 1.9116983640370603e-05, "loss": 0.17718254327774047, "step": 32360 }, { "epoch": 0.16078400206630042, "grad_norm": 2.4041240215301514, "learning_rate": 1.9116322565986733e-05, "loss": 0.18333780765533447, "step": 32370 }, { "epoch": 0.1608336727496697, "grad_norm": 2.6168606281280518, "learning_rate": 1.91156612556761e-05, "loss": 0.208166241645813, "step": 32380 }, { "epoch": 0.16088334343303895, "grad_norm": 0.7215338349342346, "learning_rate": 1.9114999709455823e-05, "loss": 0.19334754943847657, "step": 32390 }, { "epoch": 0.1609330141164082, "grad_norm": 1.1788028478622437, "learning_rate": 1.9114337927343013e-05, "loss": 0.19183727502822875, "step": 32400 }, { "epoch": 0.16098268479977748, "grad_norm": 1.029953122138977, "learning_rate": 1.9113675909354804e-05, "loss": 0.17959063053131102, "step": 32410 }, { "epoch": 0.16103235548314673, "grad_norm": 0.873068630695343, "learning_rate": 1.911301365550833e-05, "loss": 0.17481932640075684, "step": 32420 }, { "epoch": 0.161082026166516, "grad_norm": 1.5792126655578613, "learning_rate": 1.9112351165820722e-05, "loss": 0.1866380453109741, "step": 32430 }, { "epoch": 0.16113169684988526, "grad_norm": 1.2981438636779785, "learning_rate": 1.911168844030913e-05, "loss": 0.17260953187942504, "step": 32440 }, { "epoch": 0.16118136753325452, "grad_norm": 1.4389088153839111, "learning_rate": 1.9111025478990707e-05, "loss": 0.17541782855987548, "step": 32450 }, { "epoch": 0.1612310382166238, "grad_norm": 2.408656597137451, "learning_rate": 1.911036228188261e-05, "loss": 0.1840360641479492, "step": 32460 }, { "epoch": 0.16128070889999305, "grad_norm": 0.94684898853302, "learning_rate": 1.9109698849001996e-05, "loss": 0.18356525897979736, "step": 32470 }, { "epoch": 0.1613303795833623, "grad_norm": 1.2630621194839478, "learning_rate": 1.910903518036604e-05, "loss": 0.18249092102050782, "step": 32480 }, { "epoch": 0.16138005026673158, "grad_norm": 2.6655142307281494, "learning_rate": 1.910837127599192e-05, "loss": 0.17629904747009278, "step": 32490 }, { "epoch": 0.16142972095010083, "grad_norm": 1.9463804960250854, "learning_rate": 1.9107707135896806e-05, "loss": 0.195759916305542, "step": 32500 }, { "epoch": 0.16147939163347008, "grad_norm": 3.1290769577026367, "learning_rate": 1.9107042760097897e-05, "loss": 0.18346905708312988, "step": 32510 }, { "epoch": 0.16152906231683936, "grad_norm": 1.1328850984573364, "learning_rate": 1.9106378148612385e-05, "loss": 0.18135240077972412, "step": 32520 }, { "epoch": 0.1615787330002086, "grad_norm": 0.9788316488265991, "learning_rate": 1.910571330145747e-05, "loss": 0.1772789716720581, "step": 32530 }, { "epoch": 0.16162840368357786, "grad_norm": 1.4620991945266724, "learning_rate": 1.910504821865035e-05, "loss": 0.18245975971221923, "step": 32540 }, { "epoch": 0.16167807436694714, "grad_norm": 1.2762444019317627, "learning_rate": 1.9104382900208246e-05, "loss": 0.16907562017440797, "step": 32550 }, { "epoch": 0.1617277450503164, "grad_norm": 0.7306430339813232, "learning_rate": 1.9103717346148374e-05, "loss": 0.17029206752777098, "step": 32560 }, { "epoch": 0.16177741573368568, "grad_norm": 0.592319667339325, "learning_rate": 1.9103051556487954e-05, "loss": 0.2007871150970459, "step": 32570 }, { "epoch": 0.16182708641705493, "grad_norm": 0.7369861602783203, "learning_rate": 1.9102385531244225e-05, "loss": 0.17005374431610107, "step": 32580 }, { "epoch": 0.16187675710042418, "grad_norm": 0.9693955183029175, "learning_rate": 1.9101719270434417e-05, "loss": 0.17123088836669922, "step": 32590 }, { "epoch": 0.16192642778379346, "grad_norm": 1.1414945125579834, "learning_rate": 1.910105277407577e-05, "loss": 0.19457589387893676, "step": 32600 }, { "epoch": 0.1619760984671627, "grad_norm": 2.214867353439331, "learning_rate": 1.910038604218554e-05, "loss": 0.19702755212783812, "step": 32610 }, { "epoch": 0.16202576915053196, "grad_norm": 0.8078622817993164, "learning_rate": 1.909971907478098e-05, "loss": 0.17765730619430542, "step": 32620 }, { "epoch": 0.16207543983390124, "grad_norm": 1.6461310386657715, "learning_rate": 1.9099051871879345e-05, "loss": 0.19088401794433593, "step": 32630 }, { "epoch": 0.1621251105172705, "grad_norm": 0.6399253010749817, "learning_rate": 1.909838443349791e-05, "loss": 0.21048803329467775, "step": 32640 }, { "epoch": 0.16217478120063974, "grad_norm": 2.0029361248016357, "learning_rate": 1.9097716759653942e-05, "loss": 0.18217051029205322, "step": 32650 }, { "epoch": 0.16222445188400902, "grad_norm": 1.195875883102417, "learning_rate": 1.9097048850364722e-05, "loss": 0.19710235595703124, "step": 32660 }, { "epoch": 0.16227412256737828, "grad_norm": 0.6718167066574097, "learning_rate": 1.9096380705647537e-05, "loss": 0.18220378160476686, "step": 32670 }, { "epoch": 0.16232379325074756, "grad_norm": 2.258962392807007, "learning_rate": 1.9095712325519678e-05, "loss": 0.1841710090637207, "step": 32680 }, { "epoch": 0.1623734639341168, "grad_norm": 2.9062082767486572, "learning_rate": 1.9095043709998436e-05, "loss": 0.1881579875946045, "step": 32690 }, { "epoch": 0.16242313461748606, "grad_norm": 1.505967140197754, "learning_rate": 1.9094374859101124e-05, "loss": 0.19269335269927979, "step": 32700 }, { "epoch": 0.16247280530085534, "grad_norm": 3.1176977157592773, "learning_rate": 1.909370577284505e-05, "loss": 0.18223366737365723, "step": 32710 }, { "epoch": 0.1625224759842246, "grad_norm": 0.8032265901565552, "learning_rate": 1.909303645124752e-05, "loss": 0.19027546644210816, "step": 32720 }, { "epoch": 0.16257214666759384, "grad_norm": 2.30832839012146, "learning_rate": 1.9092366894325867e-05, "loss": 0.1734304904937744, "step": 32730 }, { "epoch": 0.16262181735096312, "grad_norm": 2.044260263442993, "learning_rate": 1.9091697102097412e-05, "loss": 0.18231260776519775, "step": 32740 }, { "epoch": 0.16267148803433237, "grad_norm": 1.0802713632583618, "learning_rate": 1.9091027074579497e-05, "loss": 0.18983503580093383, "step": 32750 }, { "epoch": 0.16272115871770162, "grad_norm": 2.7352261543273926, "learning_rate": 1.9090356811789454e-05, "loss": 0.178509521484375, "step": 32760 }, { "epoch": 0.1627708294010709, "grad_norm": 2.367982864379883, "learning_rate": 1.908968631374463e-05, "loss": 0.1856890082359314, "step": 32770 }, { "epoch": 0.16282050008444016, "grad_norm": 1.3223010301589966, "learning_rate": 1.908901558046238e-05, "loss": 0.16341097354888917, "step": 32780 }, { "epoch": 0.16287017076780944, "grad_norm": 1.2377525568008423, "learning_rate": 1.908834461196006e-05, "loss": 0.18457541465759278, "step": 32790 }, { "epoch": 0.1629198414511787, "grad_norm": 1.0251649618148804, "learning_rate": 1.9087673408255037e-05, "loss": 0.175705885887146, "step": 32800 }, { "epoch": 0.16296951213454794, "grad_norm": 2.534345865249634, "learning_rate": 1.908700196936468e-05, "loss": 0.18257902860641478, "step": 32810 }, { "epoch": 0.16301918281791722, "grad_norm": 1.255789041519165, "learning_rate": 1.9086330295306364e-05, "loss": 0.1823716163635254, "step": 32820 }, { "epoch": 0.16306885350128647, "grad_norm": 1.938042402267456, "learning_rate": 1.9085658386097476e-05, "loss": 0.1900726556777954, "step": 32830 }, { "epoch": 0.16311852418465572, "grad_norm": 2.1313533782958984, "learning_rate": 1.90849862417554e-05, "loss": 0.17830219268798828, "step": 32840 }, { "epoch": 0.163168194868025, "grad_norm": 1.316112756729126, "learning_rate": 1.9084313862297536e-05, "loss": 0.1708993911743164, "step": 32850 }, { "epoch": 0.16321786555139425, "grad_norm": 0.8508999347686768, "learning_rate": 1.908364124774128e-05, "loss": 0.171645188331604, "step": 32860 }, { "epoch": 0.1632675362347635, "grad_norm": 1.4324042797088623, "learning_rate": 1.908296839810404e-05, "loss": 0.16055545806884766, "step": 32870 }, { "epoch": 0.16331720691813278, "grad_norm": 1.0201969146728516, "learning_rate": 1.9082295313403236e-05, "loss": 0.19040210247039796, "step": 32880 }, { "epoch": 0.16336687760150204, "grad_norm": 0.9307816624641418, "learning_rate": 1.9081621993656273e-05, "loss": 0.17852094173431396, "step": 32890 }, { "epoch": 0.16341654828487132, "grad_norm": 1.7887381315231323, "learning_rate": 1.9080948438880588e-05, "loss": 0.1733228325843811, "step": 32900 }, { "epoch": 0.16346621896824057, "grad_norm": 1.4402868747711182, "learning_rate": 1.9080274649093612e-05, "loss": 0.18932775259017945, "step": 32910 }, { "epoch": 0.16351588965160982, "grad_norm": 1.017142653465271, "learning_rate": 1.9079600624312772e-05, "loss": 0.1796209216117859, "step": 32920 }, { "epoch": 0.1635655603349791, "grad_norm": 2.2431352138519287, "learning_rate": 1.9078926364555522e-05, "loss": 0.19006990194320678, "step": 32930 }, { "epoch": 0.16361523101834835, "grad_norm": 0.9891034364700317, "learning_rate": 1.9078251869839312e-05, "loss": 0.16627001762390137, "step": 32940 }, { "epoch": 0.1636649017017176, "grad_norm": 1.6827566623687744, "learning_rate": 1.9077577140181587e-05, "loss": 0.1694380521774292, "step": 32950 }, { "epoch": 0.16371457238508688, "grad_norm": 1.31000816822052, "learning_rate": 1.907690217559982e-05, "loss": 0.17573536634445192, "step": 32960 }, { "epoch": 0.16376424306845613, "grad_norm": 1.2246804237365723, "learning_rate": 1.907622697611147e-05, "loss": 0.18847274780273438, "step": 32970 }, { "epoch": 0.16381391375182539, "grad_norm": 1.0512179136276245, "learning_rate": 1.907555154173402e-05, "loss": 0.17347862720489501, "step": 32980 }, { "epoch": 0.16386358443519466, "grad_norm": 0.9607326984405518, "learning_rate": 1.9074875872484947e-05, "loss": 0.1899867534637451, "step": 32990 }, { "epoch": 0.16391325511856392, "grad_norm": 1.23508882522583, "learning_rate": 1.907419996838173e-05, "loss": 0.1853569746017456, "step": 33000 }, { "epoch": 0.1639629258019332, "grad_norm": 0.8501999378204346, "learning_rate": 1.9073523829441868e-05, "loss": 0.17662534713745118, "step": 33010 }, { "epoch": 0.16401259648530245, "grad_norm": 0.8968721628189087, "learning_rate": 1.9072847455682858e-05, "loss": 0.18383915424346925, "step": 33020 }, { "epoch": 0.1640622671686717, "grad_norm": 0.7168241143226624, "learning_rate": 1.9072170847122202e-05, "loss": 0.17795770168304442, "step": 33030 }, { "epoch": 0.16411193785204098, "grad_norm": 2.266188859939575, "learning_rate": 1.9071494003777413e-05, "loss": 0.17597289085388185, "step": 33040 }, { "epoch": 0.16416160853541023, "grad_norm": 1.94230055809021, "learning_rate": 1.907081692566601e-05, "loss": 0.16707478761672973, "step": 33050 }, { "epoch": 0.16421127921877948, "grad_norm": 2.4918291568756104, "learning_rate": 1.907013961280551e-05, "loss": 0.17653114795684816, "step": 33060 }, { "epoch": 0.16426094990214876, "grad_norm": 1.2955940961837769, "learning_rate": 1.9069462065213443e-05, "loss": 0.18179748058319092, "step": 33070 }, { "epoch": 0.164310620585518, "grad_norm": 2.3157098293304443, "learning_rate": 1.9068784282907343e-05, "loss": 0.188083279132843, "step": 33080 }, { "epoch": 0.16436029126888727, "grad_norm": 2.4361798763275146, "learning_rate": 1.9068106265904758e-05, "loss": 0.1737133502960205, "step": 33090 }, { "epoch": 0.16440996195225654, "grad_norm": 1.7553373575210571, "learning_rate": 1.9067428014223224e-05, "loss": 0.17201377153396608, "step": 33100 }, { "epoch": 0.1644596326356258, "grad_norm": 1.0522470474243164, "learning_rate": 1.90667495278803e-05, "loss": 0.16857634782791137, "step": 33110 }, { "epoch": 0.16450930331899508, "grad_norm": 1.8390012979507446, "learning_rate": 1.9066070806893545e-05, "loss": 0.19158439636230468, "step": 33120 }, { "epoch": 0.16455897400236433, "grad_norm": 0.7849538922309875, "learning_rate": 1.906539185128052e-05, "loss": 0.17257940769195557, "step": 33130 }, { "epoch": 0.16460864468573358, "grad_norm": 2.7125086784362793, "learning_rate": 1.90647126610588e-05, "loss": 0.1896693468093872, "step": 33140 }, { "epoch": 0.16465831536910286, "grad_norm": 1.1514577865600586, "learning_rate": 1.9064033236245962e-05, "loss": 0.1849459171295166, "step": 33150 }, { "epoch": 0.1647079860524721, "grad_norm": 0.891545295715332, "learning_rate": 1.906335357685959e-05, "loss": 0.1742410659790039, "step": 33160 }, { "epoch": 0.16475765673584136, "grad_norm": 1.8146113157272339, "learning_rate": 1.9062673682917267e-05, "loss": 0.17812471389770507, "step": 33170 }, { "epoch": 0.16480732741921064, "grad_norm": 1.0150163173675537, "learning_rate": 1.90619935544366e-05, "loss": 0.17735276222229004, "step": 33180 }, { "epoch": 0.1648569981025799, "grad_norm": 0.7695809006690979, "learning_rate": 1.9061313191435178e-05, "loss": 0.18078250885009767, "step": 33190 }, { "epoch": 0.16490666878594915, "grad_norm": 1.566793441772461, "learning_rate": 1.9060632593930615e-05, "loss": 0.18012003898620604, "step": 33200 }, { "epoch": 0.16495633946931842, "grad_norm": 0.8949476480484009, "learning_rate": 1.9059951761940522e-05, "loss": 0.17451329231262208, "step": 33210 }, { "epoch": 0.16500601015268768, "grad_norm": 1.0146112442016602, "learning_rate": 1.9059270695482524e-05, "loss": 0.19956955909729004, "step": 33220 }, { "epoch": 0.16505568083605696, "grad_norm": 2.8594002723693848, "learning_rate": 1.905858939457424e-05, "loss": 0.1885228395462036, "step": 33230 }, { "epoch": 0.1651053515194262, "grad_norm": 1.1492598056793213, "learning_rate": 1.9057907859233307e-05, "loss": 0.18085722923278807, "step": 33240 }, { "epoch": 0.16515502220279546, "grad_norm": 1.6974525451660156, "learning_rate": 1.9057226089477358e-05, "loss": 0.18479878902435304, "step": 33250 }, { "epoch": 0.16520469288616474, "grad_norm": 1.1182469129562378, "learning_rate": 1.905654408532404e-05, "loss": 0.17970647811889648, "step": 33260 }, { "epoch": 0.165254363569534, "grad_norm": 1.9696614742279053, "learning_rate": 1.9055861846791006e-05, "loss": 0.17384278774261475, "step": 33270 }, { "epoch": 0.16530403425290324, "grad_norm": 1.9963246583938599, "learning_rate": 1.9055179373895904e-05, "loss": 0.1879738450050354, "step": 33280 }, { "epoch": 0.16535370493627252, "grad_norm": 1.14406418800354, "learning_rate": 1.9054496666656405e-05, "loss": 0.1782778024673462, "step": 33290 }, { "epoch": 0.16540337561964177, "grad_norm": 1.1992567777633667, "learning_rate": 1.905381372509017e-05, "loss": 0.17978063821792603, "step": 33300 }, { "epoch": 0.16545304630301103, "grad_norm": 1.0919177532196045, "learning_rate": 1.9053130549214877e-05, "loss": 0.18542044162750243, "step": 33310 }, { "epoch": 0.1655027169863803, "grad_norm": 1.3032926321029663, "learning_rate": 1.9052447139048206e-05, "loss": 0.1999053716659546, "step": 33320 }, { "epoch": 0.16555238766974956, "grad_norm": 1.2327489852905273, "learning_rate": 1.9051763494607842e-05, "loss": 0.18501534461975097, "step": 33330 }, { "epoch": 0.1656020583531188, "grad_norm": 1.0440104007720947, "learning_rate": 1.905107961591148e-05, "loss": 0.18653790950775145, "step": 33340 }, { "epoch": 0.1656517290364881, "grad_norm": 1.6630761623382568, "learning_rate": 1.9050395502976813e-05, "loss": 0.17887768745422364, "step": 33350 }, { "epoch": 0.16570139971985734, "grad_norm": 0.8886459469795227, "learning_rate": 1.904971115582155e-05, "loss": 0.1834873914718628, "step": 33360 }, { "epoch": 0.16575107040322662, "grad_norm": 1.4769558906555176, "learning_rate": 1.9049026574463404e-05, "loss": 0.18949153423309326, "step": 33370 }, { "epoch": 0.16580074108659587, "grad_norm": 2.1729562282562256, "learning_rate": 1.904834175892009e-05, "loss": 0.1817837119102478, "step": 33380 }, { "epoch": 0.16585041176996512, "grad_norm": 1.7681794166564941, "learning_rate": 1.9047656709209327e-05, "loss": 0.19462738037109376, "step": 33390 }, { "epoch": 0.1659000824533344, "grad_norm": 1.2565921545028687, "learning_rate": 1.904697142534885e-05, "loss": 0.17598717212677, "step": 33400 }, { "epoch": 0.16594975313670365, "grad_norm": 0.8567456603050232, "learning_rate": 1.9046285907356384e-05, "loss": 0.19254813194274903, "step": 33410 }, { "epoch": 0.1659994238200729, "grad_norm": 2.808298110961914, "learning_rate": 1.9045600155249677e-05, "loss": 0.1727636456489563, "step": 33420 }, { "epoch": 0.16604909450344219, "grad_norm": 1.318018913269043, "learning_rate": 1.904491416904648e-05, "loss": 0.17730114459991456, "step": 33430 }, { "epoch": 0.16609876518681144, "grad_norm": 1.0095425844192505, "learning_rate": 1.9044227948764535e-05, "loss": 0.17892647981643678, "step": 33440 }, { "epoch": 0.1661484358701807, "grad_norm": 0.6420156955718994, "learning_rate": 1.9043541494421612e-05, "loss": 0.17793238162994385, "step": 33450 }, { "epoch": 0.16619810655354997, "grad_norm": 2.072403907775879, "learning_rate": 1.904285480603547e-05, "loss": 0.161728572845459, "step": 33460 }, { "epoch": 0.16624777723691922, "grad_norm": 0.6675548553466797, "learning_rate": 1.9042167883623878e-05, "loss": 0.17031797170639038, "step": 33470 }, { "epoch": 0.1662974479202885, "grad_norm": 1.446399450302124, "learning_rate": 1.9041480727204622e-05, "loss": 0.169395112991333, "step": 33480 }, { "epoch": 0.16634711860365775, "grad_norm": 1.4333319664001465, "learning_rate": 1.904079333679548e-05, "loss": 0.1971200942993164, "step": 33490 }, { "epoch": 0.166396789287027, "grad_norm": 2.307032585144043, "learning_rate": 1.904010571241424e-05, "loss": 0.1877053737640381, "step": 33500 }, { "epoch": 0.16644645997039628, "grad_norm": 1.1319358348846436, "learning_rate": 1.9039417854078695e-05, "loss": 0.19893549680709838, "step": 33510 }, { "epoch": 0.16649613065376553, "grad_norm": 0.6387315392494202, "learning_rate": 1.9038729761806654e-05, "loss": 0.17411518096923828, "step": 33520 }, { "epoch": 0.16654580133713479, "grad_norm": 1.773881435394287, "learning_rate": 1.9038041435615924e-05, "loss": 0.1875016689300537, "step": 33530 }, { "epoch": 0.16659547202050407, "grad_norm": 0.9640386700630188, "learning_rate": 1.9037352875524314e-05, "loss": 0.18004376888275148, "step": 33540 }, { "epoch": 0.16664514270387332, "grad_norm": 2.014544725418091, "learning_rate": 1.9036664081549644e-05, "loss": 0.17065370082855225, "step": 33550 }, { "epoch": 0.16669481338724257, "grad_norm": 2.4282867908477783, "learning_rate": 1.903597505370974e-05, "loss": 0.1881498098373413, "step": 33560 }, { "epoch": 0.16674448407061185, "grad_norm": 1.2669605016708374, "learning_rate": 1.9035285792022436e-05, "loss": 0.18065862655639647, "step": 33570 }, { "epoch": 0.1667941547539811, "grad_norm": 0.9532135128974915, "learning_rate": 1.903459629650557e-05, "loss": 0.17753357887268068, "step": 33580 }, { "epoch": 0.16684382543735038, "grad_norm": 0.6866584420204163, "learning_rate": 1.9033906567176986e-05, "loss": 0.190571928024292, "step": 33590 }, { "epoch": 0.16689349612071963, "grad_norm": 1.155208706855774, "learning_rate": 1.9033216604054526e-05, "loss": 0.16788429021835327, "step": 33600 }, { "epoch": 0.16694316680408888, "grad_norm": 1.2107182741165161, "learning_rate": 1.9032526407156058e-05, "loss": 0.18264319896697997, "step": 33610 }, { "epoch": 0.16699283748745816, "grad_norm": 1.085337519645691, "learning_rate": 1.9031835976499438e-05, "loss": 0.1743844270706177, "step": 33620 }, { "epoch": 0.16704250817082741, "grad_norm": 1.7046306133270264, "learning_rate": 1.903114531210253e-05, "loss": 0.18265917301177978, "step": 33630 }, { "epoch": 0.16709217885419667, "grad_norm": 0.8586347699165344, "learning_rate": 1.9030454413983217e-05, "loss": 0.17680597305297852, "step": 33640 }, { "epoch": 0.16714184953756595, "grad_norm": 2.7816550731658936, "learning_rate": 1.9029763282159374e-05, "loss": 0.18061970472335814, "step": 33650 }, { "epoch": 0.1671915202209352, "grad_norm": 1.7817013263702393, "learning_rate": 1.902907191664889e-05, "loss": 0.17870794534683226, "step": 33660 }, { "epoch": 0.16724119090430445, "grad_norm": 1.0619405508041382, "learning_rate": 1.9028380317469654e-05, "loss": 0.1779077649116516, "step": 33670 }, { "epoch": 0.16729086158767373, "grad_norm": 1.0704506635665894, "learning_rate": 1.902768848463956e-05, "loss": 0.1650640845298767, "step": 33680 }, { "epoch": 0.16734053227104298, "grad_norm": 0.8311164975166321, "learning_rate": 1.9026996418176526e-05, "loss": 0.18613102436065673, "step": 33690 }, { "epoch": 0.16739020295441226, "grad_norm": 1.6756525039672852, "learning_rate": 1.9026304118098448e-05, "loss": 0.17710840702056885, "step": 33700 }, { "epoch": 0.1674398736377815, "grad_norm": 0.7134966850280762, "learning_rate": 1.902561158442325e-05, "loss": 0.19118572473526002, "step": 33710 }, { "epoch": 0.16748954432115076, "grad_norm": 0.9261496067047119, "learning_rate": 1.9024918817168857e-05, "loss": 0.17224971055984498, "step": 33720 }, { "epoch": 0.16753921500452004, "grad_norm": 1.1457771062850952, "learning_rate": 1.902422581635319e-05, "loss": 0.16647055149078369, "step": 33730 }, { "epoch": 0.1675888856878893, "grad_norm": 0.7081064581871033, "learning_rate": 1.902353258199419e-05, "loss": 0.1778966188430786, "step": 33740 }, { "epoch": 0.16763855637125855, "grad_norm": 2.0045759677886963, "learning_rate": 1.9022839114109795e-05, "loss": 0.1821434259414673, "step": 33750 }, { "epoch": 0.16768822705462783, "grad_norm": 1.1168283224105835, "learning_rate": 1.902214541271795e-05, "loss": 0.18122777938842774, "step": 33760 }, { "epoch": 0.16773789773799708, "grad_norm": 1.221444845199585, "learning_rate": 1.902145147783661e-05, "loss": 0.1758838415145874, "step": 33770 }, { "epoch": 0.16778756842136633, "grad_norm": 1.490026831626892, "learning_rate": 1.902075730948373e-05, "loss": 0.18491523265838622, "step": 33780 }, { "epoch": 0.1678372391047356, "grad_norm": 1.8754584789276123, "learning_rate": 1.902006290767728e-05, "loss": 0.1735628366470337, "step": 33790 }, { "epoch": 0.16788690978810486, "grad_norm": 0.9634385704994202, "learning_rate": 1.9019368272435234e-05, "loss": 0.17565633058547975, "step": 33800 }, { "epoch": 0.16793658047147414, "grad_norm": 0.9952893853187561, "learning_rate": 1.9018673403775555e-05, "loss": 0.19302525520324706, "step": 33810 }, { "epoch": 0.1679862511548434, "grad_norm": 1.752785325050354, "learning_rate": 1.9017978301716242e-05, "loss": 0.17434327602386473, "step": 33820 }, { "epoch": 0.16803592183821264, "grad_norm": 0.9130649566650391, "learning_rate": 1.9017282966275273e-05, "loss": 0.19197349548339843, "step": 33830 }, { "epoch": 0.16808559252158192, "grad_norm": 1.4156163930892944, "learning_rate": 1.9016587397470645e-05, "loss": 0.18237555027008057, "step": 33840 }, { "epoch": 0.16813526320495117, "grad_norm": 1.155158519744873, "learning_rate": 1.901589159532036e-05, "loss": 0.17802178859710693, "step": 33850 }, { "epoch": 0.16818493388832043, "grad_norm": 1.0733364820480347, "learning_rate": 1.9015195559842432e-05, "loss": 0.19088573455810548, "step": 33860 }, { "epoch": 0.1682346045716897, "grad_norm": 2.108819007873535, "learning_rate": 1.9014499291054866e-05, "loss": 0.18355586528778076, "step": 33870 }, { "epoch": 0.16828427525505896, "grad_norm": 1.5863794088363647, "learning_rate": 1.9013802788975676e-05, "loss": 0.18635122776031493, "step": 33880 }, { "epoch": 0.1683339459384282, "grad_norm": 2.3432579040527344, "learning_rate": 1.90131060536229e-05, "loss": 0.18593368530273438, "step": 33890 }, { "epoch": 0.1683836166217975, "grad_norm": 1.9654985666275024, "learning_rate": 1.901240908501456e-05, "loss": 0.17104051113128663, "step": 33900 }, { "epoch": 0.16843328730516674, "grad_norm": 0.9692022204399109, "learning_rate": 1.90117118831687e-05, "loss": 0.20303130149841309, "step": 33910 }, { "epoch": 0.16848295798853602, "grad_norm": 0.8765313625335693, "learning_rate": 1.901101444810336e-05, "loss": 0.18519996404647826, "step": 33920 }, { "epoch": 0.16853262867190527, "grad_norm": 1.7644473314285278, "learning_rate": 1.9010316779836587e-05, "loss": 0.17769205570220947, "step": 33930 }, { "epoch": 0.16858229935527452, "grad_norm": 2.547290325164795, "learning_rate": 1.9009618878386442e-05, "loss": 0.1810052752494812, "step": 33940 }, { "epoch": 0.1686319700386438, "grad_norm": 2.086578607559204, "learning_rate": 1.900892074377098e-05, "loss": 0.17695369720458984, "step": 33950 }, { "epoch": 0.16868164072201305, "grad_norm": 1.1530561447143555, "learning_rate": 1.900822237600827e-05, "loss": 0.18409372568130494, "step": 33960 }, { "epoch": 0.1687313114053823, "grad_norm": 1.8349964618682861, "learning_rate": 1.900752377511639e-05, "loss": 0.19029459953308106, "step": 33970 }, { "epoch": 0.16878098208875159, "grad_norm": 1.6836973428726196, "learning_rate": 1.9006824941113416e-05, "loss": 0.18619630336761475, "step": 33980 }, { "epoch": 0.16883065277212084, "grad_norm": 0.807698667049408, "learning_rate": 1.9006125874017438e-05, "loss": 0.1876781940460205, "step": 33990 }, { "epoch": 0.1688803234554901, "grad_norm": 0.8594428300857544, "learning_rate": 1.9005426573846536e-05, "loss": 0.18257973194122315, "step": 34000 }, { "epoch": 0.16892999413885937, "grad_norm": 1.6194781064987183, "learning_rate": 1.9004727040618816e-05, "loss": 0.17721154689788818, "step": 34010 }, { "epoch": 0.16897966482222862, "grad_norm": 1.8152555227279663, "learning_rate": 1.9004027274352386e-05, "loss": 0.18430085182189943, "step": 34020 }, { "epoch": 0.1690293355055979, "grad_norm": 1.6284418106079102, "learning_rate": 1.900332727506535e-05, "loss": 0.18112668991088868, "step": 34030 }, { "epoch": 0.16907900618896715, "grad_norm": 2.0954010486602783, "learning_rate": 1.9002627042775818e-05, "loss": 0.18032339811325074, "step": 34040 }, { "epoch": 0.1691286768723364, "grad_norm": 1.347826600074768, "learning_rate": 1.9001926577501923e-05, "loss": 0.18193861246109008, "step": 34050 }, { "epoch": 0.16917834755570568, "grad_norm": 1.8874658346176147, "learning_rate": 1.9001225879261787e-05, "loss": 0.18932068347930908, "step": 34060 }, { "epoch": 0.16922801823907493, "grad_norm": 1.6433717012405396, "learning_rate": 1.9000524948073544e-05, "loss": 0.192240309715271, "step": 34070 }, { "epoch": 0.16927768892244419, "grad_norm": 2.0860373973846436, "learning_rate": 1.8999823783955332e-05, "loss": 0.19129146337509156, "step": 34080 }, { "epoch": 0.16932735960581347, "grad_norm": 1.7402009963989258, "learning_rate": 1.8999122386925303e-05, "loss": 0.18383469581604003, "step": 34090 }, { "epoch": 0.16937703028918272, "grad_norm": 1.4196362495422363, "learning_rate": 1.8998420757001604e-05, "loss": 0.17138469219207764, "step": 34100 }, { "epoch": 0.16942670097255197, "grad_norm": 1.3756006956100464, "learning_rate": 1.899771889420239e-05, "loss": 0.17877911329269408, "step": 34110 }, { "epoch": 0.16947637165592125, "grad_norm": 1.049758791923523, "learning_rate": 1.899701679854584e-05, "loss": 0.1679982900619507, "step": 34120 }, { "epoch": 0.1695260423392905, "grad_norm": 2.2567298412323, "learning_rate": 1.89963144700501e-05, "loss": 0.17832984924316406, "step": 34130 }, { "epoch": 0.16957571302265975, "grad_norm": 1.2946343421936035, "learning_rate": 1.8995611908733366e-05, "loss": 0.1710496425628662, "step": 34140 }, { "epoch": 0.16962538370602903, "grad_norm": 1.317099928855896, "learning_rate": 1.8994909114613808e-05, "loss": 0.171470046043396, "step": 34150 }, { "epoch": 0.16967505438939828, "grad_norm": 1.3670014142990112, "learning_rate": 1.8994206087709623e-05, "loss": 0.18340531587600709, "step": 34160 }, { "epoch": 0.16972472507276756, "grad_norm": 1.7631504535675049, "learning_rate": 1.8993502828039e-05, "loss": 0.19135262966156005, "step": 34170 }, { "epoch": 0.16977439575613681, "grad_norm": 1.505370020866394, "learning_rate": 1.899279933562014e-05, "loss": 0.17927588224411012, "step": 34180 }, { "epoch": 0.16982406643950607, "grad_norm": 1.2197084426879883, "learning_rate": 1.8992095610471248e-05, "loss": 0.17532702684402465, "step": 34190 }, { "epoch": 0.16987373712287535, "grad_norm": 1.6948256492614746, "learning_rate": 1.899139165261054e-05, "loss": 0.19846220016479493, "step": 34200 }, { "epoch": 0.1699234078062446, "grad_norm": 2.10545015335083, "learning_rate": 1.8990687462056224e-05, "loss": 0.16622252464294435, "step": 34210 }, { "epoch": 0.16997307848961385, "grad_norm": 0.707639753818512, "learning_rate": 1.898998303882654e-05, "loss": 0.19071224927902222, "step": 34220 }, { "epoch": 0.17002274917298313, "grad_norm": 1.0956140756607056, "learning_rate": 1.8989278382939704e-05, "loss": 0.17910865545272828, "step": 34230 }, { "epoch": 0.17007241985635238, "grad_norm": 0.6572604179382324, "learning_rate": 1.898857349441396e-05, "loss": 0.1831458330154419, "step": 34240 }, { "epoch": 0.17012209053972163, "grad_norm": 1.04008150100708, "learning_rate": 1.898786837326755e-05, "loss": 0.17792677879333496, "step": 34250 }, { "epoch": 0.1701717612230909, "grad_norm": 1.9508543014526367, "learning_rate": 1.898716301951872e-05, "loss": 0.17491958141326905, "step": 34260 }, { "epoch": 0.17022143190646016, "grad_norm": 0.9374940395355225, "learning_rate": 1.898645743318572e-05, "loss": 0.1868882179260254, "step": 34270 }, { "epoch": 0.17027110258982944, "grad_norm": 1.5963571071624756, "learning_rate": 1.898575161428682e-05, "loss": 0.1839831829071045, "step": 34280 }, { "epoch": 0.1703207732731987, "grad_norm": 1.6119431257247925, "learning_rate": 1.898504556284028e-05, "loss": 0.16883440017700196, "step": 34290 }, { "epoch": 0.17037044395656795, "grad_norm": 1.9972350597381592, "learning_rate": 1.898433927886437e-05, "loss": 0.18458073139190673, "step": 34300 }, { "epoch": 0.17042011463993723, "grad_norm": 2.0598223209381104, "learning_rate": 1.8983632762377374e-05, "loss": 0.17879377603530883, "step": 34310 }, { "epoch": 0.17046978532330648, "grad_norm": 1.1468639373779297, "learning_rate": 1.8982926013397576e-05, "loss": 0.1796002984046936, "step": 34320 }, { "epoch": 0.17051945600667573, "grad_norm": 1.1582378149032593, "learning_rate": 1.8982219031943264e-05, "loss": 0.173142671585083, "step": 34330 }, { "epoch": 0.170569126690045, "grad_norm": 2.6290037631988525, "learning_rate": 1.8981511818032736e-05, "loss": 0.19932379722595214, "step": 34340 }, { "epoch": 0.17061879737341426, "grad_norm": 2.7340855598449707, "learning_rate": 1.8980804371684292e-05, "loss": 0.1891249656677246, "step": 34350 }, { "epoch": 0.1706684680567835, "grad_norm": 0.939043402671814, "learning_rate": 1.898009669291624e-05, "loss": 0.17926528453826904, "step": 34360 }, { "epoch": 0.1707181387401528, "grad_norm": 0.9133495092391968, "learning_rate": 1.89793887817469e-05, "loss": 0.18993173837661742, "step": 34370 }, { "epoch": 0.17076780942352204, "grad_norm": 0.969574511051178, "learning_rate": 1.8978680638194586e-05, "loss": 0.1881170392036438, "step": 34380 }, { "epoch": 0.17081748010689132, "grad_norm": 1.353670597076416, "learning_rate": 1.897797226227763e-05, "loss": 0.1708160161972046, "step": 34390 }, { "epoch": 0.17086715079026057, "grad_norm": 0.5890130400657654, "learning_rate": 1.897726365401436e-05, "loss": 0.17391772270202638, "step": 34400 }, { "epoch": 0.17091682147362983, "grad_norm": 0.877470076084137, "learning_rate": 1.897655481342312e-05, "loss": 0.18663592338562013, "step": 34410 }, { "epoch": 0.1709664921569991, "grad_norm": 2.1431219577789307, "learning_rate": 1.8975845740522246e-05, "loss": 0.18825159072875977, "step": 34420 }, { "epoch": 0.17101616284036836, "grad_norm": 1.8706151247024536, "learning_rate": 1.8975136435330092e-05, "loss": 0.19818251132965087, "step": 34430 }, { "epoch": 0.1710658335237376, "grad_norm": 1.0527544021606445, "learning_rate": 1.8974426897865018e-05, "loss": 0.1934076428413391, "step": 34440 }, { "epoch": 0.1711155042071069, "grad_norm": 1.8853883743286133, "learning_rate": 1.8973717128145386e-05, "loss": 0.18706551790237427, "step": 34450 }, { "epoch": 0.17116517489047614, "grad_norm": 0.9271144270896912, "learning_rate": 1.897300712618956e-05, "loss": 0.1990717649459839, "step": 34460 }, { "epoch": 0.1712148455738454, "grad_norm": 1.148164987564087, "learning_rate": 1.897229689201592e-05, "loss": 0.18160536289215087, "step": 34470 }, { "epoch": 0.17126451625721467, "grad_norm": 1.349009394645691, "learning_rate": 1.8971586425642843e-05, "loss": 0.18555200099945068, "step": 34480 }, { "epoch": 0.17131418694058392, "grad_norm": 1.566652774810791, "learning_rate": 1.8970875727088718e-05, "loss": 0.17926127910614015, "step": 34490 }, { "epoch": 0.1713638576239532, "grad_norm": 1.0583207607269287, "learning_rate": 1.8970164796371935e-05, "loss": 0.1849893569946289, "step": 34500 }, { "epoch": 0.17141352830732245, "grad_norm": 2.582102060317993, "learning_rate": 1.8969453633510893e-05, "loss": 0.17836194038391112, "step": 34510 }, { "epoch": 0.1714631989906917, "grad_norm": 0.7671673893928528, "learning_rate": 1.8968742238524e-05, "loss": 0.17112596035003663, "step": 34520 }, { "epoch": 0.17151286967406099, "grad_norm": 0.944871187210083, "learning_rate": 1.8968030611429663e-05, "loss": 0.17279529571533203, "step": 34530 }, { "epoch": 0.17156254035743024, "grad_norm": 0.8883454203605652, "learning_rate": 1.8967318752246298e-05, "loss": 0.1848830461502075, "step": 34540 }, { "epoch": 0.1716122110407995, "grad_norm": 0.996708333492279, "learning_rate": 1.896660666099233e-05, "loss": 0.17784459590911866, "step": 34550 }, { "epoch": 0.17166188172416877, "grad_norm": 1.6714398860931396, "learning_rate": 1.8965894337686186e-05, "loss": 0.17591278553009032, "step": 34560 }, { "epoch": 0.17171155240753802, "grad_norm": 1.3738205432891846, "learning_rate": 1.8965181782346302e-05, "loss": 0.18083959817886353, "step": 34570 }, { "epoch": 0.17176122309090727, "grad_norm": 2.162656545639038, "learning_rate": 1.8964468994991118e-05, "loss": 0.17424250841140748, "step": 34580 }, { "epoch": 0.17181089377427655, "grad_norm": 0.8286508917808533, "learning_rate": 1.8963755975639084e-05, "loss": 0.16600972414016724, "step": 34590 }, { "epoch": 0.1718605644576458, "grad_norm": 0.9963124394416809, "learning_rate": 1.8963042724308647e-05, "loss": 0.16600391864776612, "step": 34600 }, { "epoch": 0.17191023514101508, "grad_norm": 0.762381374835968, "learning_rate": 1.896232924101827e-05, "loss": 0.17282778024673462, "step": 34610 }, { "epoch": 0.17195990582438433, "grad_norm": 0.9796485304832458, "learning_rate": 1.8961615525786414e-05, "loss": 0.1696650743484497, "step": 34620 }, { "epoch": 0.1720095765077536, "grad_norm": 2.060171604156494, "learning_rate": 1.8960901578631553e-05, "loss": 0.18259067535400392, "step": 34630 }, { "epoch": 0.17205924719112287, "grad_norm": 1.491829752922058, "learning_rate": 1.896018739957216e-05, "loss": 0.16189918518066407, "step": 34640 }, { "epoch": 0.17210891787449212, "grad_norm": 0.7694727182388306, "learning_rate": 1.8959472988626722e-05, "loss": 0.17464566230773926, "step": 34650 }, { "epoch": 0.17215858855786137, "grad_norm": 1.5988227128982544, "learning_rate": 1.8958758345813724e-05, "loss": 0.17953078746795653, "step": 34660 }, { "epoch": 0.17220825924123065, "grad_norm": 1.5676031112670898, "learning_rate": 1.895804347115166e-05, "loss": 0.17704741954803466, "step": 34670 }, { "epoch": 0.1722579299245999, "grad_norm": 1.5040652751922607, "learning_rate": 1.8957328364659036e-05, "loss": 0.19152908325195311, "step": 34680 }, { "epoch": 0.17230760060796915, "grad_norm": 1.0902903079986572, "learning_rate": 1.8956613026354356e-05, "loss": 0.18252651691436766, "step": 34690 }, { "epoch": 0.17235727129133843, "grad_norm": 2.38277268409729, "learning_rate": 1.895589745625613e-05, "loss": 0.18627514839172363, "step": 34700 }, { "epoch": 0.17240694197470768, "grad_norm": 1.9976774454116821, "learning_rate": 1.8955181654382882e-05, "loss": 0.18792636394500734, "step": 34710 }, { "epoch": 0.17245661265807696, "grad_norm": 0.9767537117004395, "learning_rate": 1.895446562075313e-05, "loss": 0.1787933588027954, "step": 34720 }, { "epoch": 0.17250628334144621, "grad_norm": 0.9538509249687195, "learning_rate": 1.8953749355385408e-05, "loss": 0.16761648654937744, "step": 34730 }, { "epoch": 0.17255595402481547, "grad_norm": 0.7643747329711914, "learning_rate": 1.8953032858298256e-05, "loss": 0.16627801656723024, "step": 34740 }, { "epoch": 0.17260562470818475, "grad_norm": 0.9768049120903015, "learning_rate": 1.8952316129510215e-05, "loss": 0.1973428726196289, "step": 34750 }, { "epoch": 0.172655295391554, "grad_norm": 1.2540143728256226, "learning_rate": 1.8951599169039828e-05, "loss": 0.1737417221069336, "step": 34760 }, { "epoch": 0.17270496607492325, "grad_norm": 1.0585193634033203, "learning_rate": 1.8950881976905653e-05, "loss": 0.20242662429809571, "step": 34770 }, { "epoch": 0.17275463675829253, "grad_norm": 0.7784952521324158, "learning_rate": 1.8950164553126255e-05, "loss": 0.1822869896888733, "step": 34780 }, { "epoch": 0.17280430744166178, "grad_norm": 1.2906720638275146, "learning_rate": 1.8949446897720193e-05, "loss": 0.19161999225616455, "step": 34790 }, { "epoch": 0.17285397812503103, "grad_norm": 0.8285309672355652, "learning_rate": 1.894872901070605e-05, "loss": 0.18748791217803956, "step": 34800 }, { "epoch": 0.1729036488084003, "grad_norm": 1.1378830671310425, "learning_rate": 1.8948010892102394e-05, "loss": 0.17772823572158813, "step": 34810 }, { "epoch": 0.17295331949176956, "grad_norm": 0.8547798991203308, "learning_rate": 1.8947292541927813e-05, "loss": 0.18198685646057128, "step": 34820 }, { "epoch": 0.17300299017513884, "grad_norm": 3.724065065383911, "learning_rate": 1.89465739602009e-05, "loss": 0.18523191213607787, "step": 34830 }, { "epoch": 0.1730526608585081, "grad_norm": 0.7118996977806091, "learning_rate": 1.894585514694025e-05, "loss": 0.1716998338699341, "step": 34840 }, { "epoch": 0.17310233154187735, "grad_norm": 0.9989844560623169, "learning_rate": 1.8945136102164467e-05, "loss": 0.17126201391220092, "step": 34850 }, { "epoch": 0.17315200222524663, "grad_norm": 1.7651869058609009, "learning_rate": 1.8944416825892155e-05, "loss": 0.18635120391845703, "step": 34860 }, { "epoch": 0.17320167290861588, "grad_norm": 1.8986350297927856, "learning_rate": 1.8943697318141938e-05, "loss": 0.1780700922012329, "step": 34870 }, { "epoch": 0.17325134359198513, "grad_norm": 0.7351781129837036, "learning_rate": 1.8942977578932425e-05, "loss": 0.17998247146606444, "step": 34880 }, { "epoch": 0.1733010142753544, "grad_norm": 2.0292868614196777, "learning_rate": 1.894225760828225e-05, "loss": 0.16586147546768187, "step": 34890 }, { "epoch": 0.17335068495872366, "grad_norm": 3.05594539642334, "learning_rate": 1.8941537406210045e-05, "loss": 0.16571896076202391, "step": 34900 }, { "epoch": 0.1734003556420929, "grad_norm": 2.122222661972046, "learning_rate": 1.8940816972734443e-05, "loss": 0.1818070888519287, "step": 34910 }, { "epoch": 0.1734500263254622, "grad_norm": 1.7790510654449463, "learning_rate": 1.8940096307874095e-05, "loss": 0.20901365280151368, "step": 34920 }, { "epoch": 0.17349969700883144, "grad_norm": 1.0993537902832031, "learning_rate": 1.893937541164765e-05, "loss": 0.1770765781402588, "step": 34930 }, { "epoch": 0.1735493676922007, "grad_norm": 2.741579055786133, "learning_rate": 1.8938654284073766e-05, "loss": 0.18384943008422852, "step": 34940 }, { "epoch": 0.17359903837556998, "grad_norm": 2.0182228088378906, "learning_rate": 1.8937932925171098e-05, "loss": 0.17683374881744385, "step": 34950 }, { "epoch": 0.17364870905893923, "grad_norm": 1.1094307899475098, "learning_rate": 1.8937211334958324e-05, "loss": 0.1736844301223755, "step": 34960 }, { "epoch": 0.1736983797423085, "grad_norm": 1.204461932182312, "learning_rate": 1.8936489513454114e-05, "loss": 0.19470704793930055, "step": 34970 }, { "epoch": 0.17374805042567776, "grad_norm": 1.3821980953216553, "learning_rate": 1.8935767460677146e-05, "loss": 0.17737095355987548, "step": 34980 }, { "epoch": 0.173797721109047, "grad_norm": 1.0045208930969238, "learning_rate": 1.893504517664611e-05, "loss": 0.1856090784072876, "step": 34990 }, { "epoch": 0.1738473917924163, "grad_norm": 1.2343451976776123, "learning_rate": 1.89343226613797e-05, "loss": 0.183837628364563, "step": 35000 }, { "epoch": 0.17389706247578554, "grad_norm": 1.370786190032959, "learning_rate": 1.8933599914896605e-05, "loss": 0.17622087001800538, "step": 35010 }, { "epoch": 0.1739467331591548, "grad_norm": 2.38185715675354, "learning_rate": 1.893287693721554e-05, "loss": 0.16606900691986085, "step": 35020 }, { "epoch": 0.17399640384252407, "grad_norm": 0.8987035155296326, "learning_rate": 1.8932153728355215e-05, "loss": 0.17060824632644653, "step": 35030 }, { "epoch": 0.17404607452589332, "grad_norm": 2.151982307434082, "learning_rate": 1.8931430288334337e-05, "loss": 0.18873447179794312, "step": 35040 }, { "epoch": 0.17409574520926258, "grad_norm": 0.9649510979652405, "learning_rate": 1.8930706617171642e-05, "loss": 0.19530515670776366, "step": 35050 }, { "epoch": 0.17414541589263186, "grad_norm": 1.2414885759353638, "learning_rate": 1.8929982714885844e-05, "loss": 0.1859411597251892, "step": 35060 }, { "epoch": 0.1741950865760011, "grad_norm": 1.189197063446045, "learning_rate": 1.8929258581495688e-05, "loss": 0.1834282398223877, "step": 35070 }, { "epoch": 0.1742447572593704, "grad_norm": 1.065100908279419, "learning_rate": 1.8928534217019907e-05, "loss": 0.17293155193328857, "step": 35080 }, { "epoch": 0.17429442794273964, "grad_norm": 1.6268606185913086, "learning_rate": 1.8927809621477252e-05, "loss": 0.1776728391647339, "step": 35090 }, { "epoch": 0.1743440986261089, "grad_norm": 1.0255473852157593, "learning_rate": 1.8927084794886475e-05, "loss": 0.1629708528518677, "step": 35100 }, { "epoch": 0.17439376930947817, "grad_norm": 1.6838847398757935, "learning_rate": 1.892635973726633e-05, "loss": 0.1912182927131653, "step": 35110 }, { "epoch": 0.17444343999284742, "grad_norm": 2.4658188819885254, "learning_rate": 1.8925634448635585e-05, "loss": 0.18123306035995485, "step": 35120 }, { "epoch": 0.17449311067621667, "grad_norm": 2.3538482189178467, "learning_rate": 1.892490892901301e-05, "loss": 0.17104098796844483, "step": 35130 }, { "epoch": 0.17454278135958595, "grad_norm": 1.389017939567566, "learning_rate": 1.892418317841738e-05, "loss": 0.1835213541984558, "step": 35140 }, { "epoch": 0.1745924520429552, "grad_norm": 0.8704129457473755, "learning_rate": 1.8923457196867477e-05, "loss": 0.1826444983482361, "step": 35150 }, { "epoch": 0.17464212272632446, "grad_norm": 0.9501023292541504, "learning_rate": 1.892273098438209e-05, "loss": 0.17144763469696045, "step": 35160 }, { "epoch": 0.17469179340969374, "grad_norm": 1.318678379058838, "learning_rate": 1.8922004540980015e-05, "loss": 0.1692659378051758, "step": 35170 }, { "epoch": 0.174741464093063, "grad_norm": 1.6157870292663574, "learning_rate": 1.892127786668005e-05, "loss": 0.18210151195526122, "step": 35180 }, { "epoch": 0.17479113477643227, "grad_norm": 2.919564723968506, "learning_rate": 1.8920550961501e-05, "loss": 0.19387590885162354, "step": 35190 }, { "epoch": 0.17484080545980152, "grad_norm": 4.079762935638428, "learning_rate": 1.8919823825461675e-05, "loss": 0.1860353946685791, "step": 35200 }, { "epoch": 0.17489047614317077, "grad_norm": 1.2606004476547241, "learning_rate": 1.89190964585809e-05, "loss": 0.18590091466903685, "step": 35210 }, { "epoch": 0.17494014682654005, "grad_norm": 0.7381290197372437, "learning_rate": 1.891836886087749e-05, "loss": 0.19377374649047852, "step": 35220 }, { "epoch": 0.1749898175099093, "grad_norm": 1.1715376377105713, "learning_rate": 1.8917641032370288e-05, "loss": 0.17729690074920654, "step": 35230 }, { "epoch": 0.17503948819327855, "grad_norm": 1.820998191833496, "learning_rate": 1.8916912973078116e-05, "loss": 0.17557345628738402, "step": 35240 }, { "epoch": 0.17508915887664783, "grad_norm": 2.2725729942321777, "learning_rate": 1.891618468301982e-05, "loss": 0.18457980155944825, "step": 35250 }, { "epoch": 0.17513882956001708, "grad_norm": 1.6575363874435425, "learning_rate": 1.891545616221425e-05, "loss": 0.18719215393066407, "step": 35260 }, { "epoch": 0.17518850024338634, "grad_norm": 1.608733057975769, "learning_rate": 1.8914727410680263e-05, "loss": 0.17070350646972657, "step": 35270 }, { "epoch": 0.17523817092675562, "grad_norm": 1.0355069637298584, "learning_rate": 1.8913998428436712e-05, "loss": 0.1716547966003418, "step": 35280 }, { "epoch": 0.17528784161012487, "grad_norm": 1.3228453397750854, "learning_rate": 1.8913269215502467e-05, "loss": 0.18054800033569335, "step": 35290 }, { "epoch": 0.17533751229349415, "grad_norm": 1.3295520544052124, "learning_rate": 1.8912539771896397e-05, "loss": 0.1580216407775879, "step": 35300 }, { "epoch": 0.1753871829768634, "grad_norm": 1.758687973022461, "learning_rate": 1.8911810097637384e-05, "loss": 0.18680026531219482, "step": 35310 }, { "epoch": 0.17543685366023265, "grad_norm": 1.534551978111267, "learning_rate": 1.891108019274431e-05, "loss": 0.18118323087692262, "step": 35320 }, { "epoch": 0.17548652434360193, "grad_norm": 1.0753172636032104, "learning_rate": 1.891035005723606e-05, "loss": 0.16997523307800294, "step": 35330 }, { "epoch": 0.17553619502697118, "grad_norm": 1.3823891878128052, "learning_rate": 1.8909619691131534e-05, "loss": 0.17681899070739746, "step": 35340 }, { "epoch": 0.17558586571034043, "grad_norm": 0.568622887134552, "learning_rate": 1.8908889094449633e-05, "loss": 0.1916884183883667, "step": 35350 }, { "epoch": 0.1756355363937097, "grad_norm": 1.2768603563308716, "learning_rate": 1.8908158267209264e-05, "loss": 0.19508795738220214, "step": 35360 }, { "epoch": 0.17568520707707896, "grad_norm": 1.0256640911102295, "learning_rate": 1.8907427209429342e-05, "loss": 0.1824345588684082, "step": 35370 }, { "epoch": 0.17573487776044822, "grad_norm": 1.0798207521438599, "learning_rate": 1.8906695921128787e-05, "loss": 0.194787335395813, "step": 35380 }, { "epoch": 0.1757845484438175, "grad_norm": 1.5275802612304688, "learning_rate": 1.8905964402326518e-05, "loss": 0.19148291349411012, "step": 35390 }, { "epoch": 0.17583421912718675, "grad_norm": 1.740694522857666, "learning_rate": 1.8905232653041478e-05, "loss": 0.1866361379623413, "step": 35400 }, { "epoch": 0.17588388981055603, "grad_norm": 1.054269552230835, "learning_rate": 1.8904500673292593e-05, "loss": 0.17081602811813354, "step": 35410 }, { "epoch": 0.17593356049392528, "grad_norm": 1.5506017208099365, "learning_rate": 1.8903768463098807e-05, "loss": 0.17020952701568604, "step": 35420 }, { "epoch": 0.17598323117729453, "grad_norm": 0.9107095003128052, "learning_rate": 1.890303602247908e-05, "loss": 0.17815940380096434, "step": 35430 }, { "epoch": 0.1760329018606638, "grad_norm": 1.557369351387024, "learning_rate": 1.890230335145236e-05, "loss": 0.18195284605026246, "step": 35440 }, { "epoch": 0.17608257254403306, "grad_norm": 1.771647572517395, "learning_rate": 1.8901570450037604e-05, "loss": 0.17581326961517335, "step": 35450 }, { "epoch": 0.1761322432274023, "grad_norm": 0.8703497052192688, "learning_rate": 1.890083731825379e-05, "loss": 0.17320735454559327, "step": 35460 }, { "epoch": 0.1761819139107716, "grad_norm": 1.3613430261611938, "learning_rate": 1.8900103956119878e-05, "loss": 0.1735947847366333, "step": 35470 }, { "epoch": 0.17623158459414084, "grad_norm": 1.2132887840270996, "learning_rate": 1.8899370363654858e-05, "loss": 0.20335025787353517, "step": 35480 }, { "epoch": 0.1762812552775101, "grad_norm": 1.668117642402649, "learning_rate": 1.8898636540877714e-05, "loss": 0.185865581035614, "step": 35490 }, { "epoch": 0.17633092596087938, "grad_norm": 1.2461713552474976, "learning_rate": 1.889790248780743e-05, "loss": 0.1684117555618286, "step": 35500 }, { "epoch": 0.17638059664424863, "grad_norm": 1.6358513832092285, "learning_rate": 1.8897168204463008e-05, "loss": 0.19219117164611815, "step": 35510 }, { "epoch": 0.1764302673276179, "grad_norm": 0.8981404900550842, "learning_rate": 1.889643369086345e-05, "loss": 0.1833944797515869, "step": 35520 }, { "epoch": 0.17647993801098716, "grad_norm": 1.2405548095703125, "learning_rate": 1.889569894702777e-05, "loss": 0.17862271070480346, "step": 35530 }, { "epoch": 0.1765296086943564, "grad_norm": 2.276370048522949, "learning_rate": 1.8894963972974973e-05, "loss": 0.19325942993164064, "step": 35540 }, { "epoch": 0.1765792793777257, "grad_norm": 1.1769187450408936, "learning_rate": 1.8894228768724087e-05, "loss": 0.1818796157836914, "step": 35550 }, { "epoch": 0.17662895006109494, "grad_norm": 2.8813207149505615, "learning_rate": 1.8893493334294135e-05, "loss": 0.1791390895843506, "step": 35560 }, { "epoch": 0.1766786207444642, "grad_norm": 0.7910242676734924, "learning_rate": 1.8892757669704152e-05, "loss": 0.1687463641166687, "step": 35570 }, { "epoch": 0.17672829142783347, "grad_norm": 0.8393366932868958, "learning_rate": 1.8892021774973177e-05, "loss": 0.19682013988494873, "step": 35580 }, { "epoch": 0.17677796211120272, "grad_norm": 1.365224838256836, "learning_rate": 1.8891285650120253e-05, "loss": 0.19344673156738282, "step": 35590 }, { "epoch": 0.17682763279457198, "grad_norm": 1.0934778451919556, "learning_rate": 1.889054929516443e-05, "loss": 0.1764983892440796, "step": 35600 }, { "epoch": 0.17687730347794126, "grad_norm": 1.8481723070144653, "learning_rate": 1.8889812710124772e-05, "loss": 0.16869924068450928, "step": 35610 }, { "epoch": 0.1769269741613105, "grad_norm": 1.3929975032806396, "learning_rate": 1.8889075895020332e-05, "loss": 0.18309985399246215, "step": 35620 }, { "epoch": 0.1769766448446798, "grad_norm": 1.8407247066497803, "learning_rate": 1.888833884987018e-05, "loss": 0.1697329044342041, "step": 35630 }, { "epoch": 0.17702631552804904, "grad_norm": 2.3964014053344727, "learning_rate": 1.8887601574693394e-05, "loss": 0.17720048427581786, "step": 35640 }, { "epoch": 0.1770759862114183, "grad_norm": 0.7118605375289917, "learning_rate": 1.888686406950905e-05, "loss": 0.17588655948638915, "step": 35650 }, { "epoch": 0.17712565689478757, "grad_norm": 1.907262921333313, "learning_rate": 1.888612633433624e-05, "loss": 0.18708195686340331, "step": 35660 }, { "epoch": 0.17717532757815682, "grad_norm": 1.655534029006958, "learning_rate": 1.8885388369194056e-05, "loss": 0.16703364849090577, "step": 35670 }, { "epoch": 0.17722499826152607, "grad_norm": 1.4899576902389526, "learning_rate": 1.888465017410159e-05, "loss": 0.18131799697875978, "step": 35680 }, { "epoch": 0.17727466894489535, "grad_norm": 1.5972028970718384, "learning_rate": 1.888391174907795e-05, "loss": 0.19432318210601807, "step": 35690 }, { "epoch": 0.1773243396282646, "grad_norm": 1.4805068969726562, "learning_rate": 1.8883173094142246e-05, "loss": 0.1716972827911377, "step": 35700 }, { "epoch": 0.17737401031163386, "grad_norm": 1.389668583869934, "learning_rate": 1.8882434209313595e-05, "loss": 0.18905702829360962, "step": 35710 }, { "epoch": 0.17742368099500314, "grad_norm": 1.7285170555114746, "learning_rate": 1.8881695094611116e-05, "loss": 0.1705556631088257, "step": 35720 }, { "epoch": 0.1774733516783724, "grad_norm": 1.2751867771148682, "learning_rate": 1.8880955750053944e-05, "loss": 0.1668403148651123, "step": 35730 }, { "epoch": 0.17752302236174164, "grad_norm": 3.56069016456604, "learning_rate": 1.8880216175661202e-05, "loss": 0.18036458492279053, "step": 35740 }, { "epoch": 0.17757269304511092, "grad_norm": 1.929584264755249, "learning_rate": 1.887947637145204e-05, "loss": 0.17909997701644897, "step": 35750 }, { "epoch": 0.17762236372848017, "grad_norm": 1.4823416471481323, "learning_rate": 1.88787363374456e-05, "loss": 0.17853388786315919, "step": 35760 }, { "epoch": 0.17767203441184945, "grad_norm": 2.2031331062316895, "learning_rate": 1.887799607366103e-05, "loss": 0.19002692699432372, "step": 35770 }, { "epoch": 0.1777217050952187, "grad_norm": 1.703863263130188, "learning_rate": 1.8877255580117494e-05, "loss": 0.20134520530700684, "step": 35780 }, { "epoch": 0.17777137577858795, "grad_norm": 1.4600811004638672, "learning_rate": 1.8876514856834155e-05, "loss": 0.17453341484069823, "step": 35790 }, { "epoch": 0.17782104646195723, "grad_norm": 1.1952189207077026, "learning_rate": 1.887577390383018e-05, "loss": 0.19464170932769775, "step": 35800 }, { "epoch": 0.17787071714532648, "grad_norm": 1.1977910995483398, "learning_rate": 1.887503272112474e-05, "loss": 0.18539756536483765, "step": 35810 }, { "epoch": 0.17792038782869574, "grad_norm": 1.0657163858413696, "learning_rate": 1.8874291308737026e-05, "loss": 0.18849787712097169, "step": 35820 }, { "epoch": 0.17797005851206502, "grad_norm": 1.4102598428726196, "learning_rate": 1.887354966668622e-05, "loss": 0.19565047025680543, "step": 35830 }, { "epoch": 0.17801972919543427, "grad_norm": 1.0376001596450806, "learning_rate": 1.8872807794991517e-05, "loss": 0.19434995651245118, "step": 35840 }, { "epoch": 0.17806939987880352, "grad_norm": 1.3859342336654663, "learning_rate": 1.8872065693672116e-05, "loss": 0.20419743061065673, "step": 35850 }, { "epoch": 0.1781190705621728, "grad_norm": 0.8496487140655518, "learning_rate": 1.8871323362747223e-05, "loss": 0.19570720195770264, "step": 35860 }, { "epoch": 0.17816874124554205, "grad_norm": 0.8337380290031433, "learning_rate": 1.887058080223605e-05, "loss": 0.17522082328796387, "step": 35870 }, { "epoch": 0.17821841192891133, "grad_norm": 1.2291587591171265, "learning_rate": 1.8869838012157807e-05, "loss": 0.177306866645813, "step": 35880 }, { "epoch": 0.17826808261228058, "grad_norm": 3.2324166297912598, "learning_rate": 1.8869094992531726e-05, "loss": 0.18027955293655396, "step": 35890 }, { "epoch": 0.17831775329564983, "grad_norm": 1.0448033809661865, "learning_rate": 1.8868351743377033e-05, "loss": 0.17210195064544678, "step": 35900 }, { "epoch": 0.1783674239790191, "grad_norm": 0.6741039156913757, "learning_rate": 1.8867608264712962e-05, "loss": 0.18255822658538817, "step": 35910 }, { "epoch": 0.17841709466238836, "grad_norm": 0.6392578482627869, "learning_rate": 1.886686455655875e-05, "loss": 0.17890805006027222, "step": 35920 }, { "epoch": 0.17846676534575762, "grad_norm": 0.7436628937721252, "learning_rate": 1.8866120618933656e-05, "loss": 0.1760188579559326, "step": 35930 }, { "epoch": 0.1785164360291269, "grad_norm": 1.5101196765899658, "learning_rate": 1.8865376451856924e-05, "loss": 0.18232626914978028, "step": 35940 }, { "epoch": 0.17856610671249615, "grad_norm": 0.7145525217056274, "learning_rate": 1.8864632055347813e-05, "loss": 0.17325637340545655, "step": 35950 }, { "epoch": 0.1786157773958654, "grad_norm": 1.063891887664795, "learning_rate": 1.8863887429425584e-05, "loss": 0.1687441349029541, "step": 35960 }, { "epoch": 0.17866544807923468, "grad_norm": 1.421128749847412, "learning_rate": 1.886314257410952e-05, "loss": 0.1940904140472412, "step": 35970 }, { "epoch": 0.17871511876260393, "grad_norm": 1.4734783172607422, "learning_rate": 1.8862397489418885e-05, "loss": 0.17858433723449707, "step": 35980 }, { "epoch": 0.1787647894459732, "grad_norm": 0.701244592666626, "learning_rate": 1.886165217537297e-05, "loss": 0.18453463315963745, "step": 35990 }, { "epoch": 0.17881446012934246, "grad_norm": 1.363862156867981, "learning_rate": 1.8860906631991052e-05, "loss": 0.16537675857543946, "step": 36000 }, { "epoch": 0.1788641308127117, "grad_norm": 0.8344661593437195, "learning_rate": 1.886016085929244e-05, "loss": 0.19082188606262207, "step": 36010 }, { "epoch": 0.178913801496081, "grad_norm": 1.28384268283844, "learning_rate": 1.8859414857296428e-05, "loss": 0.17960689067840577, "step": 36020 }, { "epoch": 0.17896347217945024, "grad_norm": 1.600264549255371, "learning_rate": 1.885866862602232e-05, "loss": 0.16798145771026612, "step": 36030 }, { "epoch": 0.1790131428628195, "grad_norm": 1.2194355726242065, "learning_rate": 1.885792216548943e-05, "loss": 0.17275041341781616, "step": 36040 }, { "epoch": 0.17906281354618878, "grad_norm": 1.6972812414169312, "learning_rate": 1.8857175475717076e-05, "loss": 0.19063175916671754, "step": 36050 }, { "epoch": 0.17911248422955803, "grad_norm": 1.3141496181488037, "learning_rate": 1.8856428556724583e-05, "loss": 0.17057209014892577, "step": 36060 }, { "epoch": 0.17916215491292728, "grad_norm": 0.8239927887916565, "learning_rate": 1.8855681408531276e-05, "loss": 0.17937763929367065, "step": 36070 }, { "epoch": 0.17921182559629656, "grad_norm": 1.9251974821090698, "learning_rate": 1.88549340311565e-05, "loss": 0.17996364831924438, "step": 36080 }, { "epoch": 0.1792614962796658, "grad_norm": 0.6058910489082336, "learning_rate": 1.885418642461959e-05, "loss": 0.17088634967803956, "step": 36090 }, { "epoch": 0.1793111669630351, "grad_norm": 2.1374013423919678, "learning_rate": 1.8853438588939893e-05, "loss": 0.17820184230804442, "step": 36100 }, { "epoch": 0.17936083764640434, "grad_norm": 1.8035188913345337, "learning_rate": 1.8852690524136767e-05, "loss": 0.18334832191467285, "step": 36110 }, { "epoch": 0.1794105083297736, "grad_norm": 2.197736978530884, "learning_rate": 1.885194223022957e-05, "loss": 0.17567745447158814, "step": 36120 }, { "epoch": 0.17946017901314287, "grad_norm": 1.0091907978057861, "learning_rate": 1.8851193707237665e-05, "loss": 0.18018718957901, "step": 36130 }, { "epoch": 0.17950984969651212, "grad_norm": 1.3757370710372925, "learning_rate": 1.885044495518043e-05, "loss": 0.1752352833747864, "step": 36140 }, { "epoch": 0.17955952037988138, "grad_norm": 0.7886584401130676, "learning_rate": 1.8849695974077233e-05, "loss": 0.1744784712791443, "step": 36150 }, { "epoch": 0.17960919106325066, "grad_norm": 2.875382423400879, "learning_rate": 1.8848946763947466e-05, "loss": 0.16002097129821777, "step": 36160 }, { "epoch": 0.1796588617466199, "grad_norm": 1.8186159133911133, "learning_rate": 1.8848197324810517e-05, "loss": 0.17848647832870485, "step": 36170 }, { "epoch": 0.17970853242998916, "grad_norm": 1.0814557075500488, "learning_rate": 1.8847447656685776e-05, "loss": 0.1864159107208252, "step": 36180 }, { "epoch": 0.17975820311335844, "grad_norm": 1.6484566926956177, "learning_rate": 1.8846697759592647e-05, "loss": 0.18531275987625123, "step": 36190 }, { "epoch": 0.1798078737967277, "grad_norm": 3.0890047550201416, "learning_rate": 1.8845947633550538e-05, "loss": 0.1881515383720398, "step": 36200 }, { "epoch": 0.17985754448009697, "grad_norm": 2.0546863079071045, "learning_rate": 1.8845197278578856e-05, "loss": 0.17733228206634521, "step": 36210 }, { "epoch": 0.17990721516346622, "grad_norm": 0.9992073774337769, "learning_rate": 1.884444669469703e-05, "loss": 0.1684321641921997, "step": 36220 }, { "epoch": 0.17995688584683547, "grad_norm": 1.5848270654678345, "learning_rate": 1.884369588192448e-05, "loss": 0.17660621404647828, "step": 36230 }, { "epoch": 0.18000655653020475, "grad_norm": 1.2310541868209839, "learning_rate": 1.8842944840280637e-05, "loss": 0.18723998069763184, "step": 36240 }, { "epoch": 0.180056227213574, "grad_norm": 0.986245334148407, "learning_rate": 1.8842193569784933e-05, "loss": 0.18165888786315917, "step": 36250 }, { "epoch": 0.18010589789694326, "grad_norm": 1.6554912328720093, "learning_rate": 1.8841442070456818e-05, "loss": 0.18174984455108642, "step": 36260 }, { "epoch": 0.18015556858031254, "grad_norm": 0.7494587302207947, "learning_rate": 1.8840690342315738e-05, "loss": 0.18772944211959838, "step": 36270 }, { "epoch": 0.1802052392636818, "grad_norm": 1.7054051160812378, "learning_rate": 1.8839938385381146e-05, "loss": 0.17964645624160766, "step": 36280 }, { "epoch": 0.18025490994705104, "grad_norm": 1.325388789176941, "learning_rate": 1.8839186199672502e-05, "loss": 0.1842864751815796, "step": 36290 }, { "epoch": 0.18030458063042032, "grad_norm": 1.1494978666305542, "learning_rate": 1.8838433785209272e-05, "loss": 0.17750146389007568, "step": 36300 }, { "epoch": 0.18035425131378957, "grad_norm": 1.09866201877594, "learning_rate": 1.883768114201093e-05, "loss": 0.1844830632209778, "step": 36310 }, { "epoch": 0.18040392199715885, "grad_norm": 2.960066556930542, "learning_rate": 1.8836928270096956e-05, "loss": 0.19497405290603637, "step": 36320 }, { "epoch": 0.1804535926805281, "grad_norm": 1.6388630867004395, "learning_rate": 1.883617516948683e-05, "loss": 0.17855061292648317, "step": 36330 }, { "epoch": 0.18050326336389735, "grad_norm": 0.8948831558227539, "learning_rate": 1.8835421840200044e-05, "loss": 0.18332881927490235, "step": 36340 }, { "epoch": 0.18055293404726663, "grad_norm": 1.5318830013275146, "learning_rate": 1.883466828225609e-05, "loss": 0.17112668752670288, "step": 36350 }, { "epoch": 0.18060260473063589, "grad_norm": 2.304702043533325, "learning_rate": 1.883391449567448e-05, "loss": 0.19880001544952391, "step": 36360 }, { "epoch": 0.18065227541400514, "grad_norm": 0.45442691445350647, "learning_rate": 1.8833160480474708e-05, "loss": 0.18272080421447753, "step": 36370 }, { "epoch": 0.18070194609737442, "grad_norm": 2.4776086807250977, "learning_rate": 1.88324062366763e-05, "loss": 0.1805893063545227, "step": 36380 }, { "epoch": 0.18075161678074367, "grad_norm": 0.8771950006484985, "learning_rate": 1.8831651764298768e-05, "loss": 0.1801395058631897, "step": 36390 }, { "epoch": 0.18080128746411292, "grad_norm": 2.1961140632629395, "learning_rate": 1.8830897063361634e-05, "loss": 0.17963995933532714, "step": 36400 }, { "epoch": 0.1808509581474822, "grad_norm": 1.1216623783111572, "learning_rate": 1.8830142133884443e-05, "loss": 0.18307008743286132, "step": 36410 }, { "epoch": 0.18090062883085145, "grad_norm": 1.9761310815811157, "learning_rate": 1.8829386975886718e-05, "loss": 0.17864456176757812, "step": 36420 }, { "epoch": 0.18095029951422073, "grad_norm": 1.4931166172027588, "learning_rate": 1.882863158938801e-05, "loss": 0.18419665098190308, "step": 36430 }, { "epoch": 0.18099997019758998, "grad_norm": 0.9960756301879883, "learning_rate": 1.882787597440787e-05, "loss": 0.18893421888351442, "step": 36440 }, { "epoch": 0.18104964088095923, "grad_norm": 2.1388463973999023, "learning_rate": 1.8827120130965845e-05, "loss": 0.1819040894508362, "step": 36450 }, { "epoch": 0.1810993115643285, "grad_norm": 0.9573988914489746, "learning_rate": 1.88263640590815e-05, "loss": 0.167249596118927, "step": 36460 }, { "epoch": 0.18114898224769777, "grad_norm": 0.9819737076759338, "learning_rate": 1.8825607758774405e-05, "loss": 0.19352347850799562, "step": 36470 }, { "epoch": 0.18119865293106702, "grad_norm": 0.7116605043411255, "learning_rate": 1.8824851230064126e-05, "loss": 0.18003575801849364, "step": 36480 }, { "epoch": 0.1812483236144363, "grad_norm": 0.7482122778892517, "learning_rate": 1.882409447297025e-05, "loss": 0.18412106037139891, "step": 36490 }, { "epoch": 0.18129799429780555, "grad_norm": 1.6048318147659302, "learning_rate": 1.8823337487512355e-05, "loss": 0.17223091125488282, "step": 36500 }, { "epoch": 0.1813476649811748, "grad_norm": 1.3734573125839233, "learning_rate": 1.882258027371003e-05, "loss": 0.17383017539978027, "step": 36510 }, { "epoch": 0.18139733566454408, "grad_norm": 1.1039360761642456, "learning_rate": 1.882182283158288e-05, "loss": 0.17930324077606202, "step": 36520 }, { "epoch": 0.18144700634791333, "grad_norm": 1.0049852132797241, "learning_rate": 1.88210651611505e-05, "loss": 0.17668187618255615, "step": 36530 }, { "epoch": 0.18149667703128258, "grad_norm": 3.1952080726623535, "learning_rate": 1.8820307262432504e-05, "loss": 0.17068309783935548, "step": 36540 }, { "epoch": 0.18154634771465186, "grad_norm": 3.33642578125, "learning_rate": 1.88195491354485e-05, "loss": 0.18327651023864747, "step": 36550 }, { "epoch": 0.18159601839802111, "grad_norm": 2.1343555450439453, "learning_rate": 1.8818790780218107e-05, "loss": 0.18235397338867188, "step": 36560 }, { "epoch": 0.1816456890813904, "grad_norm": 1.6086689233779907, "learning_rate": 1.881803219676096e-05, "loss": 0.17534842491149902, "step": 36570 }, { "epoch": 0.18169535976475965, "grad_norm": 1.0734561681747437, "learning_rate": 1.8817273385096682e-05, "loss": 0.18526138067245485, "step": 36580 }, { "epoch": 0.1817450304481289, "grad_norm": 1.8098371028900146, "learning_rate": 1.8816514345244914e-05, "loss": 0.17410757541656494, "step": 36590 }, { "epoch": 0.18179470113149818, "grad_norm": 0.7642523050308228, "learning_rate": 1.8815755077225302e-05, "loss": 0.18618718385696412, "step": 36600 }, { "epoch": 0.18184437181486743, "grad_norm": 1.0699443817138672, "learning_rate": 1.8814995581057492e-05, "loss": 0.16884956359863282, "step": 36610 }, { "epoch": 0.18189404249823668, "grad_norm": 1.1374226808547974, "learning_rate": 1.881423585676114e-05, "loss": 0.1772097945213318, "step": 36620 }, { "epoch": 0.18194371318160596, "grad_norm": 1.776816487312317, "learning_rate": 1.881347590435591e-05, "loss": 0.19395730495452881, "step": 36630 }, { "epoch": 0.1819933838649752, "grad_norm": 1.456715703010559, "learning_rate": 1.8812715723861466e-05, "loss": 0.17907118797302246, "step": 36640 }, { "epoch": 0.18204305454834446, "grad_norm": 1.5473002195358276, "learning_rate": 1.881195531529748e-05, "loss": 0.1791650891304016, "step": 36650 }, { "epoch": 0.18209272523171374, "grad_norm": 2.4210503101348877, "learning_rate": 1.881119467868364e-05, "loss": 0.1705081582069397, "step": 36660 }, { "epoch": 0.182142395915083, "grad_norm": 1.177736520767212, "learning_rate": 1.881043381403962e-05, "loss": 0.1882112979888916, "step": 36670 }, { "epoch": 0.18219206659845227, "grad_norm": 0.9718822836875916, "learning_rate": 1.8809672721385114e-05, "loss": 0.19045736789703369, "step": 36680 }, { "epoch": 0.18224173728182153, "grad_norm": 2.6107890605926514, "learning_rate": 1.8808911400739824e-05, "loss": 0.1771141767501831, "step": 36690 }, { "epoch": 0.18229140796519078, "grad_norm": 1.1729626655578613, "learning_rate": 1.8808149852123447e-05, "loss": 0.17046515941619872, "step": 36700 }, { "epoch": 0.18234107864856006, "grad_norm": 1.8846267461776733, "learning_rate": 1.8807388075555694e-05, "loss": 0.19500305652618408, "step": 36710 }, { "epoch": 0.1823907493319293, "grad_norm": 1.2593129873275757, "learning_rate": 1.8806626071056278e-05, "loss": 0.183945631980896, "step": 36720 }, { "epoch": 0.18244042001529856, "grad_norm": 0.5830979347229004, "learning_rate": 1.880586383864492e-05, "loss": 0.17752113342285156, "step": 36730 }, { "epoch": 0.18249009069866784, "grad_norm": 1.9024441242218018, "learning_rate": 1.8805101378341347e-05, "loss": 0.17310600280761718, "step": 36740 }, { "epoch": 0.1825397613820371, "grad_norm": 2.5734236240386963, "learning_rate": 1.880433869016529e-05, "loss": 0.17732832431793213, "step": 36750 }, { "epoch": 0.18258943206540634, "grad_norm": 1.3234034776687622, "learning_rate": 1.880357577413649e-05, "loss": 0.1835865020751953, "step": 36760 }, { "epoch": 0.18263910274877562, "grad_norm": 1.3231569528579712, "learning_rate": 1.880281263027469e-05, "loss": 0.18397635221481323, "step": 36770 }, { "epoch": 0.18268877343214487, "grad_norm": 2.339338541030884, "learning_rate": 1.8802049258599634e-05, "loss": 0.1825416326522827, "step": 36780 }, { "epoch": 0.18273844411551415, "grad_norm": 1.5211808681488037, "learning_rate": 1.8801285659131086e-05, "loss": 0.16847816705703736, "step": 36790 }, { "epoch": 0.1827881147988834, "grad_norm": 2.653339385986328, "learning_rate": 1.8800521831888803e-05, "loss": 0.17039134502410888, "step": 36800 }, { "epoch": 0.18283778548225266, "grad_norm": 1.4564169645309448, "learning_rate": 1.8799757776892553e-05, "loss": 0.18969677686691283, "step": 36810 }, { "epoch": 0.18288745616562194, "grad_norm": 1.4361461400985718, "learning_rate": 1.879899349416211e-05, "loss": 0.1894544005393982, "step": 36820 }, { "epoch": 0.1829371268489912, "grad_norm": 0.7336744666099548, "learning_rate": 1.8798228983717254e-05, "loss": 0.19303042888641359, "step": 36830 }, { "epoch": 0.18298679753236044, "grad_norm": 1.9592881202697754, "learning_rate": 1.8797464245577768e-05, "loss": 0.16717493534088135, "step": 36840 }, { "epoch": 0.18303646821572972, "grad_norm": 1.1744410991668701, "learning_rate": 1.8796699279763446e-05, "loss": 0.17768921852111816, "step": 36850 }, { "epoch": 0.18308613889909897, "grad_norm": 1.069328784942627, "learning_rate": 1.8795934086294084e-05, "loss": 0.1898723602294922, "step": 36860 }, { "epoch": 0.18313580958246822, "grad_norm": 0.8662187457084656, "learning_rate": 1.8795168665189483e-05, "loss": 0.1886443614959717, "step": 36870 }, { "epoch": 0.1831854802658375, "grad_norm": 0.7006816267967224, "learning_rate": 1.8794403016469457e-05, "loss": 0.1824103832244873, "step": 36880 }, { "epoch": 0.18323515094920675, "grad_norm": 1.3606728315353394, "learning_rate": 1.8793637140153812e-05, "loss": 0.17887575626373292, "step": 36890 }, { "epoch": 0.18328482163257603, "grad_norm": 1.928236484527588, "learning_rate": 1.8792871036262373e-05, "loss": 0.184162700176239, "step": 36900 }, { "epoch": 0.18333449231594529, "grad_norm": 0.8566878437995911, "learning_rate": 1.879210470481497e-05, "loss": 0.17628670930862428, "step": 36910 }, { "epoch": 0.18338416299931454, "grad_norm": 1.3436086177825928, "learning_rate": 1.879133814583143e-05, "loss": 0.17749873399734498, "step": 36920 }, { "epoch": 0.18343383368268382, "grad_norm": 1.0545907020568848, "learning_rate": 1.87905713593316e-05, "loss": 0.18550699949264526, "step": 36930 }, { "epoch": 0.18348350436605307, "grad_norm": 1.6316273212432861, "learning_rate": 1.8789804345335314e-05, "loss": 0.17627009153366088, "step": 36940 }, { "epoch": 0.18353317504942232, "grad_norm": 1.8837119340896606, "learning_rate": 1.8789037103862425e-05, "loss": 0.172643780708313, "step": 36950 }, { "epoch": 0.1835828457327916, "grad_norm": 0.8430865406990051, "learning_rate": 1.878826963493279e-05, "loss": 0.18991127014160156, "step": 36960 }, { "epoch": 0.18363251641616085, "grad_norm": 1.317911982536316, "learning_rate": 1.878750193856627e-05, "loss": 0.20132710933685302, "step": 36970 }, { "epoch": 0.1836821870995301, "grad_norm": 1.1827094554901123, "learning_rate": 1.8786734014782735e-05, "loss": 0.18432358503341675, "step": 36980 }, { "epoch": 0.18373185778289938, "grad_norm": 2.066990375518799, "learning_rate": 1.8785965863602054e-05, "loss": 0.17909176349639894, "step": 36990 }, { "epoch": 0.18378152846626863, "grad_norm": 1.1930298805236816, "learning_rate": 1.878519748504411e-05, "loss": 0.17499607801437378, "step": 37000 }, { "epoch": 0.1838311991496379, "grad_norm": 2.798884630203247, "learning_rate": 1.8784428879128787e-05, "loss": 0.18181432485580445, "step": 37010 }, { "epoch": 0.18388086983300717, "grad_norm": 1.0359208583831787, "learning_rate": 1.8783660045875977e-05, "loss": 0.17491432428359985, "step": 37020 }, { "epoch": 0.18393054051637642, "grad_norm": 2.676177501678467, "learning_rate": 1.878289098530558e-05, "loss": 0.1861126661300659, "step": 37030 }, { "epoch": 0.1839802111997457, "grad_norm": 2.0686261653900146, "learning_rate": 1.878212169743749e-05, "loss": 0.16971900463104247, "step": 37040 }, { "epoch": 0.18402988188311495, "grad_norm": 0.9268680214881897, "learning_rate": 1.878135218229163e-05, "loss": 0.18748964071273805, "step": 37050 }, { "epoch": 0.1840795525664842, "grad_norm": 1.0856797695159912, "learning_rate": 1.8780582439887898e-05, "loss": 0.1743904709815979, "step": 37060 }, { "epoch": 0.18412922324985348, "grad_norm": 1.3510642051696777, "learning_rate": 1.8779812470246226e-05, "loss": 0.1829311490058899, "step": 37070 }, { "epoch": 0.18417889393322273, "grad_norm": 1.0284013748168945, "learning_rate": 1.8779042273386537e-05, "loss": 0.18077781200408935, "step": 37080 }, { "epoch": 0.18422856461659198, "grad_norm": 1.6788101196289062, "learning_rate": 1.8778271849328766e-05, "loss": 0.17879347801208495, "step": 37090 }, { "epoch": 0.18427823529996126, "grad_norm": 2.0745902061462402, "learning_rate": 1.8777501198092848e-05, "loss": 0.17259875535964966, "step": 37100 }, { "epoch": 0.18432790598333051, "grad_norm": 2.637819290161133, "learning_rate": 1.8776730319698726e-05, "loss": 0.1777377247810364, "step": 37110 }, { "epoch": 0.1843775766666998, "grad_norm": 1.6624058485031128, "learning_rate": 1.8775959214166354e-05, "loss": 0.16635695695877076, "step": 37120 }, { "epoch": 0.18442724735006905, "grad_norm": 0.7786219120025635, "learning_rate": 1.877518788151569e-05, "loss": 0.19644110202789306, "step": 37130 }, { "epoch": 0.1844769180334383, "grad_norm": 1.572422742843628, "learning_rate": 1.8774416321766686e-05, "loss": 0.1867427945137024, "step": 37140 }, { "epoch": 0.18452658871680758, "grad_norm": 0.5989679098129272, "learning_rate": 1.8773644534939317e-05, "loss": 0.18445465564727784, "step": 37150 }, { "epoch": 0.18457625940017683, "grad_norm": 1.274158000946045, "learning_rate": 1.877287252105356e-05, "loss": 0.18848437070846558, "step": 37160 }, { "epoch": 0.18462593008354608, "grad_norm": 0.8754662275314331, "learning_rate": 1.877210028012938e-05, "loss": 0.17035428285598755, "step": 37170 }, { "epoch": 0.18467560076691536, "grad_norm": 3.1995363235473633, "learning_rate": 1.877132781218678e-05, "loss": 0.19477491378784179, "step": 37180 }, { "epoch": 0.1847252714502846, "grad_norm": 1.3116003274917603, "learning_rate": 1.8770555117245738e-05, "loss": 0.17320923805236815, "step": 37190 }, { "epoch": 0.18477494213365386, "grad_norm": 3.7293953895568848, "learning_rate": 1.8769782195326264e-05, "loss": 0.1800260901451111, "step": 37200 }, { "epoch": 0.18482461281702314, "grad_norm": 1.2519543170928955, "learning_rate": 1.8769009046448347e-05, "loss": 0.17413864135742188, "step": 37210 }, { "epoch": 0.1848742835003924, "grad_norm": 1.424731969833374, "learning_rate": 1.8768235670632e-05, "loss": 0.18112514019012452, "step": 37220 }, { "epoch": 0.18492395418376167, "grad_norm": 2.6424381732940674, "learning_rate": 1.8767462067897242e-05, "loss": 0.17885241508483887, "step": 37230 }, { "epoch": 0.18497362486713093, "grad_norm": 0.7174888253211975, "learning_rate": 1.876668823826409e-05, "loss": 0.1810334324836731, "step": 37240 }, { "epoch": 0.18502329555050018, "grad_norm": 1.0787971019744873, "learning_rate": 1.876591418175257e-05, "loss": 0.18202943801879884, "step": 37250 }, { "epoch": 0.18507296623386946, "grad_norm": 1.905060052871704, "learning_rate": 1.876513989838272e-05, "loss": 0.18256266117095948, "step": 37260 }, { "epoch": 0.1851226369172387, "grad_norm": 2.377474546432495, "learning_rate": 1.8764365388174573e-05, "loss": 0.18520629405975342, "step": 37270 }, { "epoch": 0.18517230760060796, "grad_norm": 1.2700711488723755, "learning_rate": 1.8763590651148173e-05, "loss": 0.18756078481674193, "step": 37280 }, { "epoch": 0.18522197828397724, "grad_norm": 1.4056814908981323, "learning_rate": 1.876281568732357e-05, "loss": 0.17154067754745483, "step": 37290 }, { "epoch": 0.1852716489673465, "grad_norm": 1.066100835800171, "learning_rate": 1.876204049672082e-05, "loss": 0.16670241355895996, "step": 37300 }, { "epoch": 0.18532131965071574, "grad_norm": 2.4661507606506348, "learning_rate": 1.8761265079359984e-05, "loss": 0.15951905250549317, "step": 37310 }, { "epoch": 0.18537099033408502, "grad_norm": 3.790508985519409, "learning_rate": 1.8760489435261132e-05, "loss": 0.1782156229019165, "step": 37320 }, { "epoch": 0.18542066101745427, "grad_norm": 3.3489737510681152, "learning_rate": 1.8759713564444336e-05, "loss": 0.17103878259658814, "step": 37330 }, { "epoch": 0.18547033170082353, "grad_norm": 0.9035129547119141, "learning_rate": 1.8758937466929676e-05, "loss": 0.16303493976593017, "step": 37340 }, { "epoch": 0.1855200023841928, "grad_norm": 1.511707067489624, "learning_rate": 1.875816114273723e-05, "loss": 0.16729395389556884, "step": 37350 }, { "epoch": 0.18556967306756206, "grad_norm": 2.0373740196228027, "learning_rate": 1.8757384591887102e-05, "loss": 0.1707122802734375, "step": 37360 }, { "epoch": 0.18561934375093134, "grad_norm": 1.6683990955352783, "learning_rate": 1.8756607814399382e-05, "loss": 0.1887498140335083, "step": 37370 }, { "epoch": 0.1856690144343006, "grad_norm": 1.3887238502502441, "learning_rate": 1.875583081029417e-05, "loss": 0.18037655353546142, "step": 37380 }, { "epoch": 0.18571868511766984, "grad_norm": 1.093437671661377, "learning_rate": 1.8755053579591575e-05, "loss": 0.17418091297149657, "step": 37390 }, { "epoch": 0.18576835580103912, "grad_norm": 0.7381601929664612, "learning_rate": 1.875427612231172e-05, "loss": 0.18138202428817748, "step": 37400 }, { "epoch": 0.18581802648440837, "grad_norm": 1.278130292892456, "learning_rate": 1.8753498438474717e-05, "loss": 0.1933995008468628, "step": 37410 }, { "epoch": 0.18586769716777762, "grad_norm": 0.871819019317627, "learning_rate": 1.8752720528100688e-05, "loss": 0.17214165925979613, "step": 37420 }, { "epoch": 0.1859173678511469, "grad_norm": 1.561442255973816, "learning_rate": 1.8751942391209777e-05, "loss": 0.17432830333709717, "step": 37430 }, { "epoch": 0.18596703853451615, "grad_norm": 1.3921124935150146, "learning_rate": 1.8751164027822114e-05, "loss": 0.18752744197845458, "step": 37440 }, { "epoch": 0.1860167092178854, "grad_norm": 1.4075865745544434, "learning_rate": 1.8750385437957845e-05, "loss": 0.17853702306747438, "step": 37450 }, { "epoch": 0.18606637990125469, "grad_norm": 2.4073328971862793, "learning_rate": 1.8749606621637118e-05, "loss": 0.17328233718872071, "step": 37460 }, { "epoch": 0.18611605058462394, "grad_norm": 1.8246726989746094, "learning_rate": 1.8748827578880093e-05, "loss": 0.17819883823394775, "step": 37470 }, { "epoch": 0.18616572126799322, "grad_norm": 2.1026763916015625, "learning_rate": 1.8748048309706927e-05, "loss": 0.18116545677185059, "step": 37480 }, { "epoch": 0.18621539195136247, "grad_norm": 1.352339506149292, "learning_rate": 1.8747268814137784e-05, "loss": 0.1729945421218872, "step": 37490 }, { "epoch": 0.18626506263473172, "grad_norm": 1.4978747367858887, "learning_rate": 1.8746489092192846e-05, "loss": 0.16474249362945556, "step": 37500 }, { "epoch": 0.186314733318101, "grad_norm": 1.3182145357131958, "learning_rate": 1.8745709143892287e-05, "loss": 0.1704614281654358, "step": 37510 }, { "epoch": 0.18636440400147025, "grad_norm": 2.0051324367523193, "learning_rate": 1.874492896925629e-05, "loss": 0.18648396730422973, "step": 37520 }, { "epoch": 0.1864140746848395, "grad_norm": 0.9712943434715271, "learning_rate": 1.8744148568305047e-05, "loss": 0.18724868297576905, "step": 37530 }, { "epoch": 0.18646374536820878, "grad_norm": 1.5755279064178467, "learning_rate": 1.8743367941058755e-05, "loss": 0.17979235649108888, "step": 37540 }, { "epoch": 0.18651341605157803, "grad_norm": 0.791750431060791, "learning_rate": 1.874258708753762e-05, "loss": 0.17704286575317382, "step": 37550 }, { "epoch": 0.1865630867349473, "grad_norm": 0.8490264415740967, "learning_rate": 1.874180600776184e-05, "loss": 0.16935899257659912, "step": 37560 }, { "epoch": 0.18661275741831657, "grad_norm": 1.0180169343948364, "learning_rate": 1.874102470175164e-05, "loss": 0.17067861557006836, "step": 37570 }, { "epoch": 0.18666242810168582, "grad_norm": 1.9140617847442627, "learning_rate": 1.8740243169527232e-05, "loss": 0.1834949493408203, "step": 37580 }, { "epoch": 0.1867120987850551, "grad_norm": 1.8545929193496704, "learning_rate": 1.873946141110885e-05, "loss": 0.17978053092956542, "step": 37590 }, { "epoch": 0.18676176946842435, "grad_norm": 0.9870556592941284, "learning_rate": 1.873867942651672e-05, "loss": 0.17989835739135743, "step": 37600 }, { "epoch": 0.1868114401517936, "grad_norm": 1.5228863954544067, "learning_rate": 1.8737897215771073e-05, "loss": 0.17230448722839356, "step": 37610 }, { "epoch": 0.18686111083516288, "grad_norm": 1.8305400609970093, "learning_rate": 1.8737114778892168e-05, "loss": 0.17876605987548827, "step": 37620 }, { "epoch": 0.18691078151853213, "grad_norm": 2.193824291229248, "learning_rate": 1.8736332115900243e-05, "loss": 0.1795274019241333, "step": 37630 }, { "epoch": 0.18696045220190138, "grad_norm": 0.8436744213104248, "learning_rate": 1.8735549226815553e-05, "loss": 0.17422540187835694, "step": 37640 }, { "epoch": 0.18701012288527066, "grad_norm": 1.0742045640945435, "learning_rate": 1.8734766111658366e-05, "loss": 0.16865649223327636, "step": 37650 }, { "epoch": 0.18705979356863991, "grad_norm": 1.4335448741912842, "learning_rate": 1.873398277044894e-05, "loss": 0.1641589879989624, "step": 37660 }, { "epoch": 0.18710946425200917, "grad_norm": 1.3199037313461304, "learning_rate": 1.8733199203207554e-05, "loss": 0.18171238899230957, "step": 37670 }, { "epoch": 0.18715913493537845, "grad_norm": 3.0079727172851562, "learning_rate": 1.873241540995448e-05, "loss": 0.1789188027381897, "step": 37680 }, { "epoch": 0.1872088056187477, "grad_norm": 1.3475762605667114, "learning_rate": 1.873163139071001e-05, "loss": 0.18322227001190186, "step": 37690 }, { "epoch": 0.18725847630211698, "grad_norm": 1.3797566890716553, "learning_rate": 1.8730847145494428e-05, "loss": 0.18674538135528565, "step": 37700 }, { "epoch": 0.18730814698548623, "grad_norm": 0.8918802738189697, "learning_rate": 1.8730062674328037e-05, "loss": 0.18964229822158812, "step": 37710 }, { "epoch": 0.18735781766885548, "grad_norm": 1.6227778196334839, "learning_rate": 1.872927797723113e-05, "loss": 0.18445122241973877, "step": 37720 }, { "epoch": 0.18740748835222476, "grad_norm": 1.3050395250320435, "learning_rate": 1.8728493054224023e-05, "loss": 0.16984806060791016, "step": 37730 }, { "epoch": 0.187457159035594, "grad_norm": 1.4422271251678467, "learning_rate": 1.872770790532702e-05, "loss": 0.17231560945510865, "step": 37740 }, { "epoch": 0.18750682971896326, "grad_norm": 1.7190967798233032, "learning_rate": 1.872692253056045e-05, "loss": 0.18406393527984619, "step": 37750 }, { "epoch": 0.18755650040233254, "grad_norm": 1.7894716262817383, "learning_rate": 1.872613692994463e-05, "loss": 0.17865877151489257, "step": 37760 }, { "epoch": 0.1876061710857018, "grad_norm": 2.180281162261963, "learning_rate": 1.87253511034999e-05, "loss": 0.1867828130722046, "step": 37770 }, { "epoch": 0.18765584176907105, "grad_norm": 2.2231132984161377, "learning_rate": 1.8724565051246585e-05, "loss": 0.18198130130767823, "step": 37780 }, { "epoch": 0.18770551245244033, "grad_norm": 0.8170567154884338, "learning_rate": 1.872377877320504e-05, "loss": 0.17348486185073853, "step": 37790 }, { "epoch": 0.18775518313580958, "grad_norm": 1.129071831703186, "learning_rate": 1.8722992269395605e-05, "loss": 0.1736762762069702, "step": 37800 }, { "epoch": 0.18780485381917886, "grad_norm": 1.6966371536254883, "learning_rate": 1.872220553983864e-05, "loss": 0.18446288108825684, "step": 37810 }, { "epoch": 0.1878545245025481, "grad_norm": 1.2138105630874634, "learning_rate": 1.87214185845545e-05, "loss": 0.18566665649414063, "step": 37820 }, { "epoch": 0.18790419518591736, "grad_norm": 0.7483623623847961, "learning_rate": 1.8720631403563555e-05, "loss": 0.17535946369171143, "step": 37830 }, { "epoch": 0.18795386586928664, "grad_norm": 1.3073862791061401, "learning_rate": 1.871984399688618e-05, "loss": 0.18625638484954835, "step": 37840 }, { "epoch": 0.1880035365526559, "grad_norm": 1.4024543762207031, "learning_rate": 1.871905636454274e-05, "loss": 0.16905828714370727, "step": 37850 }, { "epoch": 0.18805320723602514, "grad_norm": 0.7942233085632324, "learning_rate": 1.8718268506553633e-05, "loss": 0.18499472141265869, "step": 37860 }, { "epoch": 0.18810287791939442, "grad_norm": 1.2220075130462646, "learning_rate": 1.871748042293924e-05, "loss": 0.18471468687057496, "step": 37870 }, { "epoch": 0.18815254860276368, "grad_norm": 1.2571109533309937, "learning_rate": 1.871669211371996e-05, "loss": 0.18295613527297974, "step": 37880 }, { "epoch": 0.18820221928613293, "grad_norm": 1.506795883178711, "learning_rate": 1.8715903578916194e-05, "loss": 0.18100361824035643, "step": 37890 }, { "epoch": 0.1882518899695022, "grad_norm": 1.1125457286834717, "learning_rate": 1.871511481854835e-05, "loss": 0.18251826763153076, "step": 37900 }, { "epoch": 0.18830156065287146, "grad_norm": 1.6708343029022217, "learning_rate": 1.8714325832636833e-05, "loss": 0.18152756690979005, "step": 37910 }, { "epoch": 0.18835123133624074, "grad_norm": 1.833621859550476, "learning_rate": 1.871353662120207e-05, "loss": 0.16543762683868407, "step": 37920 }, { "epoch": 0.18840090201961, "grad_norm": 1.6319444179534912, "learning_rate": 1.871274718426448e-05, "loss": 0.1973814010620117, "step": 37930 }, { "epoch": 0.18845057270297924, "grad_norm": 0.8106419444084167, "learning_rate": 1.87119575218445e-05, "loss": 0.17856817245483397, "step": 37940 }, { "epoch": 0.18850024338634852, "grad_norm": 1.3048570156097412, "learning_rate": 1.871116763396256e-05, "loss": 0.1736508846282959, "step": 37950 }, { "epoch": 0.18854991406971777, "grad_norm": 1.9486230611801147, "learning_rate": 1.8710377520639104e-05, "loss": 0.17023943662643432, "step": 37960 }, { "epoch": 0.18859958475308702, "grad_norm": 1.050123691558838, "learning_rate": 1.870958718189458e-05, "loss": 0.17086830139160156, "step": 37970 }, { "epoch": 0.1886492554364563, "grad_norm": 1.7772974967956543, "learning_rate": 1.8708796617749445e-05, "loss": 0.16682114601135253, "step": 37980 }, { "epoch": 0.18869892611982556, "grad_norm": 2.4424281120300293, "learning_rate": 1.8708005828224155e-05, "loss": 0.1803043842315674, "step": 37990 }, { "epoch": 0.1887485968031948, "grad_norm": 1.012581467628479, "learning_rate": 1.870721481333917e-05, "loss": 0.16361095905303955, "step": 38000 }, { "epoch": 0.1887982674865641, "grad_norm": 0.9271417856216431, "learning_rate": 1.870642357311497e-05, "loss": 0.18650901317596436, "step": 38010 }, { "epoch": 0.18884793816993334, "grad_norm": 0.9404870867729187, "learning_rate": 1.8705632107572032e-05, "loss": 0.17872579097747804, "step": 38020 }, { "epoch": 0.1888976088533026, "grad_norm": 0.9682477116584778, "learning_rate": 1.8704840416730832e-05, "loss": 0.19286996126174927, "step": 38030 }, { "epoch": 0.18894727953667187, "grad_norm": 1.0015851259231567, "learning_rate": 1.870404850061186e-05, "loss": 0.1820387363433838, "step": 38040 }, { "epoch": 0.18899695022004112, "grad_norm": 1.935573697090149, "learning_rate": 1.8703256359235618e-05, "loss": 0.18049249649047852, "step": 38050 }, { "epoch": 0.1890466209034104, "grad_norm": 1.4900822639465332, "learning_rate": 1.8702463992622598e-05, "loss": 0.17204432487487792, "step": 38060 }, { "epoch": 0.18909629158677965, "grad_norm": 1.0390958786010742, "learning_rate": 1.8701671400793305e-05, "loss": 0.17452597618103027, "step": 38070 }, { "epoch": 0.1891459622701489, "grad_norm": 1.6572726964950562, "learning_rate": 1.8700878583768262e-05, "loss": 0.19125022888183593, "step": 38080 }, { "epoch": 0.18919563295351818, "grad_norm": 2.146113634109497, "learning_rate": 1.8700085541567974e-05, "loss": 0.17792110443115233, "step": 38090 }, { "epoch": 0.18924530363688744, "grad_norm": 1.0091755390167236, "learning_rate": 1.8699292274212975e-05, "loss": 0.18431200981140136, "step": 38100 }, { "epoch": 0.1892949743202567, "grad_norm": 2.011106491088867, "learning_rate": 1.8698498781723785e-05, "loss": 0.1798728346824646, "step": 38110 }, { "epoch": 0.18934464500362597, "grad_norm": 2.184486150741577, "learning_rate": 1.869770506412095e-05, "loss": 0.18481374979019166, "step": 38120 }, { "epoch": 0.18939431568699522, "grad_norm": 1.2706055641174316, "learning_rate": 1.8696911121425e-05, "loss": 0.18576617240905763, "step": 38130 }, { "epoch": 0.18944398637036447, "grad_norm": 2.0986955165863037, "learning_rate": 1.8696116953656488e-05, "loss": 0.16278131008148194, "step": 38140 }, { "epoch": 0.18949365705373375, "grad_norm": 1.4893081188201904, "learning_rate": 1.8695322560835968e-05, "loss": 0.16244661808013916, "step": 38150 }, { "epoch": 0.189543327737103, "grad_norm": 1.0011167526245117, "learning_rate": 1.8694527942983995e-05, "loss": 0.1735379934310913, "step": 38160 }, { "epoch": 0.18959299842047228, "grad_norm": 1.2934067249298096, "learning_rate": 1.8693733100121135e-05, "loss": 0.1688291311264038, "step": 38170 }, { "epoch": 0.18964266910384153, "grad_norm": 1.078650951385498, "learning_rate": 1.8692938032267956e-05, "loss": 0.18221731185913087, "step": 38180 }, { "epoch": 0.18969233978721078, "grad_norm": 1.7177481651306152, "learning_rate": 1.869214273944504e-05, "loss": 0.1726933717727661, "step": 38190 }, { "epoch": 0.18974201047058006, "grad_norm": 1.130437970161438, "learning_rate": 1.869134722167296e-05, "loss": 0.18447004556655883, "step": 38200 }, { "epoch": 0.18979168115394932, "grad_norm": 1.3365963697433472, "learning_rate": 1.8690551478972313e-05, "loss": 0.17645875215530396, "step": 38210 }, { "epoch": 0.18984135183731857, "grad_norm": 0.710261881351471, "learning_rate": 1.8689755511363686e-05, "loss": 0.1898356318473816, "step": 38220 }, { "epoch": 0.18989102252068785, "grad_norm": 1.6732492446899414, "learning_rate": 1.8688959318867682e-05, "loss": 0.18594348430633545, "step": 38230 }, { "epoch": 0.1899406932040571, "grad_norm": 1.6074317693710327, "learning_rate": 1.8688162901504902e-05, "loss": 0.1862884283065796, "step": 38240 }, { "epoch": 0.18999036388742635, "grad_norm": 4.730265140533447, "learning_rate": 1.868736625929596e-05, "loss": 0.1874624967575073, "step": 38250 }, { "epoch": 0.19004003457079563, "grad_norm": 3.026644468307495, "learning_rate": 1.8686569392261473e-05, "loss": 0.194040048122406, "step": 38260 }, { "epoch": 0.19008970525416488, "grad_norm": 1.1902450323104858, "learning_rate": 1.868577230042206e-05, "loss": 0.17590758800506592, "step": 38270 }, { "epoch": 0.19013937593753416, "grad_norm": 2.1131973266601562, "learning_rate": 1.868497498379836e-05, "loss": 0.16622960567474365, "step": 38280 }, { "epoch": 0.1901890466209034, "grad_norm": 1.1111385822296143, "learning_rate": 1.868417744241099e-05, "loss": 0.1810054063796997, "step": 38290 }, { "epoch": 0.19023871730427266, "grad_norm": 0.9359015822410583, "learning_rate": 1.8683379676280606e-05, "loss": 0.17571613788604737, "step": 38300 }, { "epoch": 0.19028838798764194, "grad_norm": 0.5351343154907227, "learning_rate": 1.8682581685427844e-05, "loss": 0.1842260479927063, "step": 38310 }, { "epoch": 0.1903380586710112, "grad_norm": 1.6838698387145996, "learning_rate": 1.8681783469873362e-05, "loss": 0.17528924942016602, "step": 38320 }, { "epoch": 0.19038772935438045, "grad_norm": 1.1876798868179321, "learning_rate": 1.8680985029637813e-05, "loss": 0.18949700593948365, "step": 38330 }, { "epoch": 0.19043740003774973, "grad_norm": 0.812335193157196, "learning_rate": 1.868018636474186e-05, "loss": 0.18154850006103515, "step": 38340 }, { "epoch": 0.19048707072111898, "grad_norm": 1.613603949546814, "learning_rate": 1.867938747520618e-05, "loss": 0.18844325542449952, "step": 38350 }, { "epoch": 0.19053674140448823, "grad_norm": 1.0313960313796997, "learning_rate": 1.8678588361051434e-05, "loss": 0.17608873844146727, "step": 38360 }, { "epoch": 0.1905864120878575, "grad_norm": 1.5484453439712524, "learning_rate": 1.8677789022298318e-05, "loss": 0.18610291481018065, "step": 38370 }, { "epoch": 0.19063608277122676, "grad_norm": 1.2846686840057373, "learning_rate": 1.8676989458967507e-05, "loss": 0.17929574251174926, "step": 38380 }, { "epoch": 0.19068575345459604, "grad_norm": 0.8219882249832153, "learning_rate": 1.86761896710797e-05, "loss": 0.18188157081604003, "step": 38390 }, { "epoch": 0.1907354241379653, "grad_norm": 2.3520913124084473, "learning_rate": 1.867538965865559e-05, "loss": 0.17463375329971315, "step": 38400 }, { "epoch": 0.19078509482133454, "grad_norm": 0.5423552989959717, "learning_rate": 1.8674589421715884e-05, "loss": 0.18856314420700074, "step": 38410 }, { "epoch": 0.19083476550470382, "grad_norm": 2.7612991333007812, "learning_rate": 1.8673788960281298e-05, "loss": 0.186631178855896, "step": 38420 }, { "epoch": 0.19088443618807308, "grad_norm": 1.643868088722229, "learning_rate": 1.8672988274372535e-05, "loss": 0.18551366329193114, "step": 38430 }, { "epoch": 0.19093410687144233, "grad_norm": 2.6065406799316406, "learning_rate": 1.867218736401032e-05, "loss": 0.18727625608444215, "step": 38440 }, { "epoch": 0.1909837775548116, "grad_norm": 1.8617889881134033, "learning_rate": 1.867138622921539e-05, "loss": 0.19216982126235962, "step": 38450 }, { "epoch": 0.19103344823818086, "grad_norm": 0.7034515142440796, "learning_rate": 1.867058487000847e-05, "loss": 0.1782684087753296, "step": 38460 }, { "epoch": 0.1910831189215501, "grad_norm": 0.997970700263977, "learning_rate": 1.8669783286410295e-05, "loss": 0.1726158857345581, "step": 38470 }, { "epoch": 0.1911327896049194, "grad_norm": 2.6640002727508545, "learning_rate": 1.866898147844162e-05, "loss": 0.17638672590255738, "step": 38480 }, { "epoch": 0.19118246028828864, "grad_norm": 1.4055919647216797, "learning_rate": 1.8668179446123184e-05, "loss": 0.18436790704727174, "step": 38490 }, { "epoch": 0.19123213097165792, "grad_norm": 1.0491138696670532, "learning_rate": 1.8667377189475755e-05, "loss": 0.19150176048278808, "step": 38500 }, { "epoch": 0.19128180165502717, "grad_norm": 1.7382398843765259, "learning_rate": 1.8666574708520086e-05, "loss": 0.17846802473068238, "step": 38510 }, { "epoch": 0.19133147233839642, "grad_norm": 1.577857494354248, "learning_rate": 1.866577200327695e-05, "loss": 0.19238643646240233, "step": 38520 }, { "epoch": 0.1913811430217657, "grad_norm": 1.3933314085006714, "learning_rate": 1.8664969073767117e-05, "loss": 0.1847004175186157, "step": 38530 }, { "epoch": 0.19143081370513496, "grad_norm": 0.8159397840499878, "learning_rate": 1.8664165920011372e-05, "loss": 0.16521012783050537, "step": 38540 }, { "epoch": 0.1914804843885042, "grad_norm": 0.6258159279823303, "learning_rate": 1.8663362542030495e-05, "loss": 0.16812268495559693, "step": 38550 }, { "epoch": 0.1915301550718735, "grad_norm": 2.467127561569214, "learning_rate": 1.8662558939845276e-05, "loss": 0.18833824396133422, "step": 38560 }, { "epoch": 0.19157982575524274, "grad_norm": 1.6073609590530396, "learning_rate": 1.8661755113476518e-05, "loss": 0.17183480262756348, "step": 38570 }, { "epoch": 0.191629496438612, "grad_norm": 2.3738291263580322, "learning_rate": 1.866095106294502e-05, "loss": 0.19281392097473143, "step": 38580 }, { "epoch": 0.19167916712198127, "grad_norm": 1.018750548362732, "learning_rate": 1.866014678827159e-05, "loss": 0.17992167472839354, "step": 38590 }, { "epoch": 0.19172883780535052, "grad_norm": 1.6992249488830566, "learning_rate": 1.865934228947704e-05, "loss": 0.18533003330230713, "step": 38600 }, { "epoch": 0.1917785084887198, "grad_norm": 1.156376838684082, "learning_rate": 1.86585375665822e-05, "loss": 0.1676161289215088, "step": 38610 }, { "epoch": 0.19182817917208905, "grad_norm": 1.178236722946167, "learning_rate": 1.8657732619607888e-05, "loss": 0.16798142194747925, "step": 38620 }, { "epoch": 0.1918778498554583, "grad_norm": 2.0079050064086914, "learning_rate": 1.8656927448574932e-05, "loss": 0.17618608474731445, "step": 38630 }, { "epoch": 0.19192752053882758, "grad_norm": 1.1972405910491943, "learning_rate": 1.8656122053504177e-05, "loss": 0.18977411985397338, "step": 38640 }, { "epoch": 0.19197719122219684, "grad_norm": 1.7689827680587769, "learning_rate": 1.8655316434416468e-05, "loss": 0.15580363273620607, "step": 38650 }, { "epoch": 0.1920268619055661, "grad_norm": 1.093643069267273, "learning_rate": 1.8654510591332643e-05, "loss": 0.17898330688476563, "step": 38660 }, { "epoch": 0.19207653258893537, "grad_norm": 1.4460232257843018, "learning_rate": 1.8653704524273565e-05, "loss": 0.18804564476013183, "step": 38670 }, { "epoch": 0.19212620327230462, "grad_norm": 1.898729681968689, "learning_rate": 1.8652898233260097e-05, "loss": 0.17720646858215333, "step": 38680 }, { "epoch": 0.19217587395567387, "grad_norm": 1.727307677268982, "learning_rate": 1.86520917183131e-05, "loss": 0.1970409035682678, "step": 38690 }, { "epoch": 0.19222554463904315, "grad_norm": 2.4428234100341797, "learning_rate": 1.865128497945345e-05, "loss": 0.1651940941810608, "step": 38700 }, { "epoch": 0.1922752153224124, "grad_norm": 1.22299063205719, "learning_rate": 1.865047801670202e-05, "loss": 0.17361791133880616, "step": 38710 }, { "epoch": 0.19232488600578168, "grad_norm": 1.5248994827270508, "learning_rate": 1.86496708300797e-05, "loss": 0.19933961629867553, "step": 38720 }, { "epoch": 0.19237455668915093, "grad_norm": 1.9147433042526245, "learning_rate": 1.8648863419607375e-05, "loss": 0.1928071141242981, "step": 38730 }, { "epoch": 0.19242422737252018, "grad_norm": 2.1929855346679688, "learning_rate": 1.8648055785305942e-05, "loss": 0.17568765878677367, "step": 38740 }, { "epoch": 0.19247389805588946, "grad_norm": 1.3468581438064575, "learning_rate": 1.8647247927196303e-05, "loss": 0.17601139545440675, "step": 38750 }, { "epoch": 0.19252356873925872, "grad_norm": 1.3590110540390015, "learning_rate": 1.8646439845299364e-05, "loss": 0.17908501625061035, "step": 38760 }, { "epoch": 0.19257323942262797, "grad_norm": 1.1358588933944702, "learning_rate": 1.864563153963604e-05, "loss": 0.17409226894378663, "step": 38770 }, { "epoch": 0.19262291010599725, "grad_norm": 1.110158920288086, "learning_rate": 1.864482301022725e-05, "loss": 0.16512880325317383, "step": 38780 }, { "epoch": 0.1926725807893665, "grad_norm": 0.802667498588562, "learning_rate": 1.864401425709391e-05, "loss": 0.17961573600769043, "step": 38790 }, { "epoch": 0.19272225147273575, "grad_norm": 0.6629295349121094, "learning_rate": 1.864320528025696e-05, "loss": 0.1771853446960449, "step": 38800 }, { "epoch": 0.19277192215610503, "grad_norm": 0.8758081793785095, "learning_rate": 1.8642396079737332e-05, "loss": 0.1781765937805176, "step": 38810 }, { "epoch": 0.19282159283947428, "grad_norm": 1.3652560710906982, "learning_rate": 1.8641586655555972e-05, "loss": 0.19444105625152588, "step": 38820 }, { "epoch": 0.19287126352284353, "grad_norm": 0.9659101963043213, "learning_rate": 1.864077700773382e-05, "loss": 0.175160014629364, "step": 38830 }, { "epoch": 0.1929209342062128, "grad_norm": 1.43899405002594, "learning_rate": 1.8639967136291837e-05, "loss": 0.18549070358276368, "step": 38840 }, { "epoch": 0.19297060488958206, "grad_norm": 1.3243862390518188, "learning_rate": 1.8639157041250974e-05, "loss": 0.17504613399505614, "step": 38850 }, { "epoch": 0.19302027557295134, "grad_norm": 1.934040904045105, "learning_rate": 1.86383467226322e-05, "loss": 0.17007238864898683, "step": 38860 }, { "epoch": 0.1930699462563206, "grad_norm": 0.6809346079826355, "learning_rate": 1.863753618045649e-05, "loss": 0.19217171669006347, "step": 38870 }, { "epoch": 0.19311961693968985, "grad_norm": 1.3788762092590332, "learning_rate": 1.8636725414744816e-05, "loss": 0.18222764730453492, "step": 38880 }, { "epoch": 0.19316928762305913, "grad_norm": 0.8267526626586914, "learning_rate": 1.8635914425518162e-05, "loss": 0.18584585189819336, "step": 38890 }, { "epoch": 0.19321895830642838, "grad_norm": 0.7923805713653564, "learning_rate": 1.863510321279751e-05, "loss": 0.18182945251464844, "step": 38900 }, { "epoch": 0.19326862898979763, "grad_norm": 0.7363661527633667, "learning_rate": 1.8634291776603866e-05, "loss": 0.17276334762573242, "step": 38910 }, { "epoch": 0.1933182996731669, "grad_norm": 0.9202699661254883, "learning_rate": 1.8633480116958217e-05, "loss": 0.18864012956619264, "step": 38920 }, { "epoch": 0.19336797035653616, "grad_norm": 1.0809016227722168, "learning_rate": 1.8632668233881576e-05, "loss": 0.17820700407028198, "step": 38930 }, { "epoch": 0.1934176410399054, "grad_norm": 0.9242062568664551, "learning_rate": 1.8631856127394955e-05, "loss": 0.17928698062896728, "step": 38940 }, { "epoch": 0.1934673117232747, "grad_norm": 1.3488765954971313, "learning_rate": 1.863104379751936e-05, "loss": 0.17909189462661743, "step": 38950 }, { "epoch": 0.19351698240664394, "grad_norm": 1.9917632341384888, "learning_rate": 1.863023124427583e-05, "loss": 0.16786402463912964, "step": 38960 }, { "epoch": 0.19356665309001322, "grad_norm": 1.042535662651062, "learning_rate": 1.8629418467685384e-05, "loss": 0.17258975505828858, "step": 38970 }, { "epoch": 0.19361632377338248, "grad_norm": 0.9366686940193176, "learning_rate": 1.8628605467769053e-05, "loss": 0.16822288036346436, "step": 38980 }, { "epoch": 0.19366599445675173, "grad_norm": 1.4617183208465576, "learning_rate": 1.8627792244547888e-05, "loss": 0.17963261604309083, "step": 38990 }, { "epoch": 0.193715665140121, "grad_norm": 1.5564991235733032, "learning_rate": 1.8626978798042926e-05, "loss": 0.18197587728500367, "step": 39000 }, { "epoch": 0.19376533582349026, "grad_norm": 1.323460578918457, "learning_rate": 1.862616512827522e-05, "loss": 0.1746896505355835, "step": 39010 }, { "epoch": 0.1938150065068595, "grad_norm": 1.5756561756134033, "learning_rate": 1.8625351235265833e-05, "loss": 0.17794806957244874, "step": 39020 }, { "epoch": 0.1938646771902288, "grad_norm": 1.7445210218429565, "learning_rate": 1.862453711903582e-05, "loss": 0.1923678398132324, "step": 39030 }, { "epoch": 0.19391434787359804, "grad_norm": 1.0626583099365234, "learning_rate": 1.8623722779606254e-05, "loss": 0.18321595191955567, "step": 39040 }, { "epoch": 0.1939640185569673, "grad_norm": 1.597138524055481, "learning_rate": 1.8622908216998212e-05, "loss": 0.16621365547180175, "step": 39050 }, { "epoch": 0.19401368924033657, "grad_norm": 1.792453646659851, "learning_rate": 1.862209343123277e-05, "loss": 0.17303931713104248, "step": 39060 }, { "epoch": 0.19406335992370582, "grad_norm": 1.3144177198410034, "learning_rate": 1.862127842233102e-05, "loss": 0.19051917791366577, "step": 39070 }, { "epoch": 0.1941130306070751, "grad_norm": 1.4786555767059326, "learning_rate": 1.862046319031405e-05, "loss": 0.17865368127822875, "step": 39080 }, { "epoch": 0.19416270129044436, "grad_norm": 2.159324884414673, "learning_rate": 1.861964773520296e-05, "loss": 0.17935643196105958, "step": 39090 }, { "epoch": 0.1942123719738136, "grad_norm": 1.5722805261611938, "learning_rate": 1.861883205701885e-05, "loss": 0.17930912971496582, "step": 39100 }, { "epoch": 0.1942620426571829, "grad_norm": 1.3969379663467407, "learning_rate": 1.8618016155782834e-05, "loss": 0.16732921600341796, "step": 39110 }, { "epoch": 0.19431171334055214, "grad_norm": 1.2864207029342651, "learning_rate": 1.8617200031516024e-05, "loss": 0.18851754665374756, "step": 39120 }, { "epoch": 0.1943613840239214, "grad_norm": 2.1219024658203125, "learning_rate": 1.8616383684239537e-05, "loss": 0.17732967138290406, "step": 39130 }, { "epoch": 0.19441105470729067, "grad_norm": 1.1147314310073853, "learning_rate": 1.8615567113974512e-05, "loss": 0.18593425750732423, "step": 39140 }, { "epoch": 0.19446072539065992, "grad_norm": 1.8630468845367432, "learning_rate": 1.8614750320742073e-05, "loss": 0.19968218803405763, "step": 39150 }, { "epoch": 0.19451039607402917, "grad_norm": 2.3885388374328613, "learning_rate": 1.861393330456336e-05, "loss": 0.1630539894104004, "step": 39160 }, { "epoch": 0.19456006675739845, "grad_norm": 1.2204773426055908, "learning_rate": 1.8613116065459513e-05, "loss": 0.19047985076904297, "step": 39170 }, { "epoch": 0.1946097374407677, "grad_norm": 2.696282386779785, "learning_rate": 1.861229860345169e-05, "loss": 0.18290505409240723, "step": 39180 }, { "epoch": 0.19465940812413698, "grad_norm": 0.9056835174560547, "learning_rate": 1.861148091856104e-05, "loss": 0.17384709119796754, "step": 39190 }, { "epoch": 0.19470907880750624, "grad_norm": 0.9376941919326782, "learning_rate": 1.8610663010808723e-05, "loss": 0.18107810020446777, "step": 39200 }, { "epoch": 0.1947587494908755, "grad_norm": 1.5191863775253296, "learning_rate": 1.8609844880215913e-05, "loss": 0.1709025979042053, "step": 39210 }, { "epoch": 0.19480842017424477, "grad_norm": 1.06672203540802, "learning_rate": 1.8609026526803778e-05, "loss": 0.17481759786605836, "step": 39220 }, { "epoch": 0.19485809085761402, "grad_norm": 1.1244347095489502, "learning_rate": 1.86082079505935e-05, "loss": 0.17933034896850586, "step": 39230 }, { "epoch": 0.19490776154098327, "grad_norm": 0.6715375781059265, "learning_rate": 1.8607389151606257e-05, "loss": 0.1909756302833557, "step": 39240 }, { "epoch": 0.19495743222435255, "grad_norm": 1.2173714637756348, "learning_rate": 1.860657012986325e-05, "loss": 0.1950102210044861, "step": 39250 }, { "epoch": 0.1950071029077218, "grad_norm": 1.058335304260254, "learning_rate": 1.860575088538566e-05, "loss": 0.16947581768035888, "step": 39260 }, { "epoch": 0.19505677359109105, "grad_norm": 0.9422357678413391, "learning_rate": 1.8604931418194708e-05, "loss": 0.17821776866912842, "step": 39270 }, { "epoch": 0.19510644427446033, "grad_norm": 1.0416463613510132, "learning_rate": 1.8604111728311583e-05, "loss": 0.17582309246063232, "step": 39280 }, { "epoch": 0.19515611495782959, "grad_norm": 1.5960872173309326, "learning_rate": 1.860329181575751e-05, "loss": 0.18544461727142333, "step": 39290 }, { "epoch": 0.19520578564119886, "grad_norm": 2.2159128189086914, "learning_rate": 1.86024716805537e-05, "loss": 0.197456955909729, "step": 39300 }, { "epoch": 0.19525545632456812, "grad_norm": 2.406137466430664, "learning_rate": 1.8601651322721387e-05, "loss": 0.18376197814941406, "step": 39310 }, { "epoch": 0.19530512700793737, "grad_norm": 1.218458652496338, "learning_rate": 1.8600830742281794e-05, "loss": 0.17925424575805665, "step": 39320 }, { "epoch": 0.19535479769130665, "grad_norm": 1.2364925146102905, "learning_rate": 1.860000993925616e-05, "loss": 0.18438801765441895, "step": 39330 }, { "epoch": 0.1954044683746759, "grad_norm": 1.6697009801864624, "learning_rate": 1.8599188913665726e-05, "loss": 0.1847520351409912, "step": 39340 }, { "epoch": 0.19545413905804515, "grad_norm": 2.362028121948242, "learning_rate": 1.859836766553174e-05, "loss": 0.1915820598602295, "step": 39350 }, { "epoch": 0.19550380974141443, "grad_norm": 0.7708528637886047, "learning_rate": 1.8597546194875456e-05, "loss": 0.17352880239486695, "step": 39360 }, { "epoch": 0.19555348042478368, "grad_norm": 1.5240494012832642, "learning_rate": 1.8596724501718137e-05, "loss": 0.1740235447883606, "step": 39370 }, { "epoch": 0.19560315110815293, "grad_norm": 1.650280237197876, "learning_rate": 1.8595902586081037e-05, "loss": 0.19043556451797486, "step": 39380 }, { "epoch": 0.1956528217915222, "grad_norm": 2.2454311847686768, "learning_rate": 1.8595080447985442e-05, "loss": 0.17973185777664186, "step": 39390 }, { "epoch": 0.19570249247489147, "grad_norm": 0.776053786277771, "learning_rate": 1.8594258087452616e-05, "loss": 0.17496249675750733, "step": 39400 }, { "epoch": 0.19575216315826074, "grad_norm": 1.6983003616333008, "learning_rate": 1.8593435504503848e-05, "loss": 0.1941349983215332, "step": 39410 }, { "epoch": 0.19580183384163, "grad_norm": 1.7339437007904053, "learning_rate": 1.8592612699160427e-05, "loss": 0.18453899621963502, "step": 39420 }, { "epoch": 0.19585150452499925, "grad_norm": 2.1249287128448486, "learning_rate": 1.859178967144364e-05, "loss": 0.19274365901947021, "step": 39430 }, { "epoch": 0.19590117520836853, "grad_norm": 1.2474859952926636, "learning_rate": 1.8590966421374794e-05, "loss": 0.17336231470108032, "step": 39440 }, { "epoch": 0.19595084589173778, "grad_norm": 1.470319151878357, "learning_rate": 1.859014294897519e-05, "loss": 0.1832853674888611, "step": 39450 }, { "epoch": 0.19600051657510703, "grad_norm": 1.5344669818878174, "learning_rate": 1.8589319254266142e-05, "loss": 0.17953813076019287, "step": 39460 }, { "epoch": 0.1960501872584763, "grad_norm": 0.7357122302055359, "learning_rate": 1.8588495337268963e-05, "loss": 0.1780301332473755, "step": 39470 }, { "epoch": 0.19609985794184556, "grad_norm": 2.0509188175201416, "learning_rate": 1.858767119800498e-05, "loss": 0.18210453987121583, "step": 39480 }, { "epoch": 0.19614952862521481, "grad_norm": 2.112523317337036, "learning_rate": 1.8586846836495517e-05, "loss": 0.18772757053375244, "step": 39490 }, { "epoch": 0.1961991993085841, "grad_norm": 1.2619032859802246, "learning_rate": 1.858602225276191e-05, "loss": 0.17666594982147216, "step": 39500 }, { "epoch": 0.19624886999195335, "grad_norm": 1.3844170570373535, "learning_rate": 1.8585197446825504e-05, "loss": 0.17594966888427735, "step": 39510 }, { "epoch": 0.19629854067532262, "grad_norm": 1.0014400482177734, "learning_rate": 1.8584372418707635e-05, "loss": 0.19119242429733277, "step": 39520 }, { "epoch": 0.19634821135869188, "grad_norm": 1.0243446826934814, "learning_rate": 1.858354716842966e-05, "loss": 0.18562111854553223, "step": 39530 }, { "epoch": 0.19639788204206113, "grad_norm": 1.8818325996398926, "learning_rate": 1.8582721696012937e-05, "loss": 0.17052760124206542, "step": 39540 }, { "epoch": 0.1964475527254304, "grad_norm": 0.8409714102745056, "learning_rate": 1.858189600147883e-05, "loss": 0.17868123054504395, "step": 39550 }, { "epoch": 0.19649722340879966, "grad_norm": 1.7017366886138916, "learning_rate": 1.8581070084848702e-05, "loss": 0.16174330711364746, "step": 39560 }, { "epoch": 0.1965468940921689, "grad_norm": 1.384421706199646, "learning_rate": 1.8580243946143933e-05, "loss": 0.1768345594406128, "step": 39570 }, { "epoch": 0.1965965647755382, "grad_norm": 1.5705482959747314, "learning_rate": 1.8579417585385896e-05, "loss": 0.17879621982574462, "step": 39580 }, { "epoch": 0.19664623545890744, "grad_norm": 1.431547999382019, "learning_rate": 1.8578591002595984e-05, "loss": 0.17359616756439208, "step": 39590 }, { "epoch": 0.1966959061422767, "grad_norm": 1.8758163452148438, "learning_rate": 1.8577764197795588e-05, "loss": 0.1648013710975647, "step": 39600 }, { "epoch": 0.19674557682564597, "grad_norm": 1.2707023620605469, "learning_rate": 1.8576937171006102e-05, "loss": 0.18007439374923706, "step": 39610 }, { "epoch": 0.19679524750901523, "grad_norm": 0.9628942012786865, "learning_rate": 1.8576109922248932e-05, "loss": 0.18049993515014648, "step": 39620 }, { "epoch": 0.19684491819238448, "grad_norm": 1.3924983739852905, "learning_rate": 1.857528245154548e-05, "loss": 0.17395738363265992, "step": 39630 }, { "epoch": 0.19689458887575376, "grad_norm": 0.9118995666503906, "learning_rate": 1.8574454758917174e-05, "loss": 0.17123225927352906, "step": 39640 }, { "epoch": 0.196944259559123, "grad_norm": 1.6276137828826904, "learning_rate": 1.8573626844385423e-05, "loss": 0.18640066385269166, "step": 39650 }, { "epoch": 0.1969939302424923, "grad_norm": 0.8109365701675415, "learning_rate": 1.8572798707971656e-05, "loss": 0.187141752243042, "step": 39660 }, { "epoch": 0.19704360092586154, "grad_norm": 1.2026898860931396, "learning_rate": 1.8571970349697307e-05, "loss": 0.18410584926605225, "step": 39670 }, { "epoch": 0.1970932716092308, "grad_norm": 2.3884289264678955, "learning_rate": 1.857114176958381e-05, "loss": 0.178104031085968, "step": 39680 }, { "epoch": 0.19714294229260007, "grad_norm": 0.8456233143806458, "learning_rate": 1.857031296765261e-05, "loss": 0.1866849660873413, "step": 39690 }, { "epoch": 0.19719261297596932, "grad_norm": 1.299726963043213, "learning_rate": 1.8569483943925157e-05, "loss": 0.17522741556167604, "step": 39700 }, { "epoch": 0.19724228365933857, "grad_norm": 2.658709764480591, "learning_rate": 1.8568654698422905e-05, "loss": 0.18547736406326293, "step": 39710 }, { "epoch": 0.19729195434270785, "grad_norm": 0.9163419604301453, "learning_rate": 1.856782523116732e-05, "loss": 0.18183764219284057, "step": 39720 }, { "epoch": 0.1973416250260771, "grad_norm": 1.50336754322052, "learning_rate": 1.8566995542179857e-05, "loss": 0.1753014326095581, "step": 39730 }, { "epoch": 0.19739129570944636, "grad_norm": 2.662952423095703, "learning_rate": 1.8566165631481995e-05, "loss": 0.17952065467834472, "step": 39740 }, { "epoch": 0.19744096639281564, "grad_norm": 1.3415470123291016, "learning_rate": 1.856533549909521e-05, "loss": 0.17486828565597534, "step": 39750 }, { "epoch": 0.1974906370761849, "grad_norm": 0.8819271326065063, "learning_rate": 1.8564505145040987e-05, "loss": 0.17362427711486816, "step": 39760 }, { "epoch": 0.19754030775955417, "grad_norm": 0.6570131778717041, "learning_rate": 1.8563674569340814e-05, "loss": 0.18199510574340821, "step": 39770 }, { "epoch": 0.19758997844292342, "grad_norm": 1.4914116859436035, "learning_rate": 1.856284377201619e-05, "loss": 0.1930412769317627, "step": 39780 }, { "epoch": 0.19763964912629267, "grad_norm": 0.8844903707504272, "learning_rate": 1.856201275308861e-05, "loss": 0.18358538150787354, "step": 39790 }, { "epoch": 0.19768931980966195, "grad_norm": 1.374893307685852, "learning_rate": 1.856118151257958e-05, "loss": 0.18170076608657837, "step": 39800 }, { "epoch": 0.1977389904930312, "grad_norm": 1.4390336275100708, "learning_rate": 1.8560350050510617e-05, "loss": 0.1852221369743347, "step": 39810 }, { "epoch": 0.19778866117640045, "grad_norm": 1.0230294466018677, "learning_rate": 1.855951836690324e-05, "loss": 0.18025342226028443, "step": 39820 }, { "epoch": 0.19783833185976973, "grad_norm": 1.5041141510009766, "learning_rate": 1.8558686461778962e-05, "loss": 0.1679948091506958, "step": 39830 }, { "epoch": 0.19788800254313899, "grad_norm": 1.0729384422302246, "learning_rate": 1.8557854335159328e-05, "loss": 0.17651221752166749, "step": 39840 }, { "epoch": 0.19793767322650824, "grad_norm": 1.5059484243392944, "learning_rate": 1.855702198706586e-05, "loss": 0.17067121267318724, "step": 39850 }, { "epoch": 0.19798734390987752, "grad_norm": 1.6627528667449951, "learning_rate": 1.8556189417520104e-05, "loss": 0.17903831005096435, "step": 39860 }, { "epoch": 0.19803701459324677, "grad_norm": 1.8437399864196777, "learning_rate": 1.8555356626543605e-05, "loss": 0.1588253617286682, "step": 39870 }, { "epoch": 0.19808668527661605, "grad_norm": 0.9914293885231018, "learning_rate": 1.8554523614157918e-05, "loss": 0.17198805809020995, "step": 39880 }, { "epoch": 0.1981363559599853, "grad_norm": 2.4652178287506104, "learning_rate": 1.8553690380384598e-05, "loss": 0.18299469947814942, "step": 39890 }, { "epoch": 0.19818602664335455, "grad_norm": 1.0270423889160156, "learning_rate": 1.8552856925245215e-05, "loss": 0.1813413143157959, "step": 39900 }, { "epoch": 0.19823569732672383, "grad_norm": 2.582366704940796, "learning_rate": 1.8552023248761326e-05, "loss": 0.18957092761993408, "step": 39910 }, { "epoch": 0.19828536801009308, "grad_norm": 0.6896459460258484, "learning_rate": 1.8551189350954518e-05, "loss": 0.16730375289916993, "step": 39920 }, { "epoch": 0.19833503869346233, "grad_norm": 1.087126612663269, "learning_rate": 1.855035523184637e-05, "loss": 0.16725174188613892, "step": 39930 }, { "epoch": 0.19838470937683161, "grad_norm": 1.8102850914001465, "learning_rate": 1.8549520891458464e-05, "loss": 0.17860658168792726, "step": 39940 }, { "epoch": 0.19843438006020087, "grad_norm": 1.4461387395858765, "learning_rate": 1.8548686329812396e-05, "loss": 0.18936328887939452, "step": 39950 }, { "epoch": 0.19848405074357012, "grad_norm": 1.7767473459243774, "learning_rate": 1.8547851546929762e-05, "loss": 0.1847731351852417, "step": 39960 }, { "epoch": 0.1985337214269394, "grad_norm": 0.9126027822494507, "learning_rate": 1.8547016542832165e-05, "loss": 0.1740025043487549, "step": 39970 }, { "epoch": 0.19858339211030865, "grad_norm": 0.9361405372619629, "learning_rate": 1.854618131754122e-05, "loss": 0.18077371120452881, "step": 39980 }, { "epoch": 0.19863306279367793, "grad_norm": 1.3683184385299683, "learning_rate": 1.854534587107854e-05, "loss": 0.1760735034942627, "step": 39990 }, { "epoch": 0.19868273347704718, "grad_norm": 0.7518851161003113, "learning_rate": 1.854451020346574e-05, "loss": 0.1796000123023987, "step": 40000 }, { "epoch": 0.19873240416041643, "grad_norm": 1.1322237253189087, "learning_rate": 1.8543674314724455e-05, "loss": 0.17391691207885743, "step": 40010 }, { "epoch": 0.1987820748437857, "grad_norm": 0.7160820364952087, "learning_rate": 1.8542838204876313e-05, "loss": 0.17371094226837158, "step": 40020 }, { "epoch": 0.19883174552715496, "grad_norm": 2.891066312789917, "learning_rate": 1.8542001873942953e-05, "loss": 0.17491626739501953, "step": 40030 }, { "epoch": 0.19888141621052421, "grad_norm": 0.9064094424247742, "learning_rate": 1.854116532194602e-05, "loss": 0.1888594150543213, "step": 40040 }, { "epoch": 0.1989310868938935, "grad_norm": 1.142494559288025, "learning_rate": 1.854032854890716e-05, "loss": 0.18825154304504393, "step": 40050 }, { "epoch": 0.19898075757726275, "grad_norm": 1.2359058856964111, "learning_rate": 1.8539491554848034e-05, "loss": 0.17464182376861573, "step": 40060 }, { "epoch": 0.199030428260632, "grad_norm": 2.3350565433502197, "learning_rate": 1.85386543397903e-05, "loss": 0.17020391225814818, "step": 40070 }, { "epoch": 0.19908009894400128, "grad_norm": 0.7672534584999084, "learning_rate": 1.8537816903755624e-05, "loss": 0.17403043508529664, "step": 40080 }, { "epoch": 0.19912976962737053, "grad_norm": 0.7988475561141968, "learning_rate": 1.853697924676568e-05, "loss": 0.17936959266662597, "step": 40090 }, { "epoch": 0.1991794403107398, "grad_norm": 1.1490538120269775, "learning_rate": 1.8536141368842146e-05, "loss": 0.17504463195800782, "step": 40100 }, { "epoch": 0.19922911099410906, "grad_norm": 1.0701899528503418, "learning_rate": 1.8535303270006704e-05, "loss": 0.16047143936157227, "step": 40110 }, { "epoch": 0.1992787816774783, "grad_norm": 1.8971892595291138, "learning_rate": 1.8534464950281048e-05, "loss": 0.18157594203948973, "step": 40120 }, { "epoch": 0.1993284523608476, "grad_norm": 1.6077592372894287, "learning_rate": 1.8533626409686868e-05, "loss": 0.18201173543930055, "step": 40130 }, { "epoch": 0.19937812304421684, "grad_norm": 0.9635781049728394, "learning_rate": 1.853278764824587e-05, "loss": 0.1914052724838257, "step": 40140 }, { "epoch": 0.1994277937275861, "grad_norm": 2.937114953994751, "learning_rate": 1.8531948665979756e-05, "loss": 0.17911627292633056, "step": 40150 }, { "epoch": 0.19947746441095537, "grad_norm": 1.4771161079406738, "learning_rate": 1.8531109462910245e-05, "loss": 0.18075575828552246, "step": 40160 }, { "epoch": 0.19952713509432463, "grad_norm": 1.839655876159668, "learning_rate": 1.8530270039059046e-05, "loss": 0.17509249448776246, "step": 40170 }, { "epoch": 0.19957680577769388, "grad_norm": 1.6982648372650146, "learning_rate": 1.8529430394447893e-05, "loss": 0.17025375366210938, "step": 40180 }, { "epoch": 0.19962647646106316, "grad_norm": 1.1351842880249023, "learning_rate": 1.8528590529098507e-05, "loss": 0.18210138082504274, "step": 40190 }, { "epoch": 0.1996761471444324, "grad_norm": 1.1731953620910645, "learning_rate": 1.8527750443032627e-05, "loss": 0.1766441822052002, "step": 40200 }, { "epoch": 0.1997258178278017, "grad_norm": 1.0257818698883057, "learning_rate": 1.8526910136271997e-05, "loss": 0.1799386739730835, "step": 40210 }, { "epoch": 0.19977548851117094, "grad_norm": 1.1679829359054565, "learning_rate": 1.8526069608838362e-05, "loss": 0.17971991300582885, "step": 40220 }, { "epoch": 0.1998251591945402, "grad_norm": 1.5678526163101196, "learning_rate": 1.852522886075347e-05, "loss": 0.18040900230407714, "step": 40230 }, { "epoch": 0.19987482987790947, "grad_norm": 1.941779613494873, "learning_rate": 1.8524387892039087e-05, "loss": 0.1689131736755371, "step": 40240 }, { "epoch": 0.19992450056127872, "grad_norm": 1.8796108961105347, "learning_rate": 1.8523546702716974e-05, "loss": 0.1729958772659302, "step": 40250 }, { "epoch": 0.19997417124464797, "grad_norm": 1.4637386798858643, "learning_rate": 1.8522705292808897e-05, "loss": 0.17317866086959838, "step": 40260 }, { "epoch": 0.20002384192801725, "grad_norm": 1.3167093992233276, "learning_rate": 1.8521863662336632e-05, "loss": 0.17594869136810304, "step": 40270 }, { "epoch": 0.2000735126113865, "grad_norm": 2.5712532997131348, "learning_rate": 1.8521021811321966e-05, "loss": 0.17147531509399414, "step": 40280 }, { "epoch": 0.20012318329475576, "grad_norm": 2.6399965286254883, "learning_rate": 1.8520179739786676e-05, "loss": 0.17415705919265748, "step": 40290 }, { "epoch": 0.20017285397812504, "grad_norm": 0.811718761920929, "learning_rate": 1.8519337447752565e-05, "loss": 0.17774875164031984, "step": 40300 }, { "epoch": 0.2002225246614943, "grad_norm": 0.7537497282028198, "learning_rate": 1.8518494935241423e-05, "loss": 0.18971703052520753, "step": 40310 }, { "epoch": 0.20027219534486357, "grad_norm": 1.1095192432403564, "learning_rate": 1.8517652202275062e-05, "loss": 0.17318789958953856, "step": 40320 }, { "epoch": 0.20032186602823282, "grad_norm": 0.9402281045913696, "learning_rate": 1.851680924887528e-05, "loss": 0.18151336908340454, "step": 40330 }, { "epoch": 0.20037153671160207, "grad_norm": 1.233045220375061, "learning_rate": 1.8515966075063905e-05, "loss": 0.19165295362472534, "step": 40340 }, { "epoch": 0.20042120739497135, "grad_norm": 0.9891203045845032, "learning_rate": 1.851512268086275e-05, "loss": 0.18167901039123535, "step": 40350 }, { "epoch": 0.2004708780783406, "grad_norm": 2.1011040210723877, "learning_rate": 1.8514279066293643e-05, "loss": 0.19521937370300294, "step": 40360 }, { "epoch": 0.20052054876170985, "grad_norm": 0.831429660320282, "learning_rate": 1.8513435231378417e-05, "loss": 0.16784276962280273, "step": 40370 }, { "epoch": 0.20057021944507913, "grad_norm": 0.7791756391525269, "learning_rate": 1.851259117613891e-05, "loss": 0.16108205318450927, "step": 40380 }, { "epoch": 0.20061989012844839, "grad_norm": 1.0355286598205566, "learning_rate": 1.851174690059697e-05, "loss": 0.1802559494972229, "step": 40390 }, { "epoch": 0.20066956081181764, "grad_norm": 1.625105381011963, "learning_rate": 1.851090240477444e-05, "loss": 0.17749783992767335, "step": 40400 }, { "epoch": 0.20071923149518692, "grad_norm": 1.9921671152114868, "learning_rate": 1.851005768869318e-05, "loss": 0.18465278148651124, "step": 40410 }, { "epoch": 0.20076890217855617, "grad_norm": 1.131306529045105, "learning_rate": 1.8509212752375045e-05, "loss": 0.17694867849349977, "step": 40420 }, { "epoch": 0.20081857286192542, "grad_norm": 1.8865559101104736, "learning_rate": 1.8508367595841905e-05, "loss": 0.1844928741455078, "step": 40430 }, { "epoch": 0.2008682435452947, "grad_norm": 1.8870490789413452, "learning_rate": 1.8507522219115635e-05, "loss": 0.17842514514923097, "step": 40440 }, { "epoch": 0.20091791422866395, "grad_norm": 3.355896472930908, "learning_rate": 1.8506676622218112e-05, "loss": 0.1759122848510742, "step": 40450 }, { "epoch": 0.20096758491203323, "grad_norm": 1.7652918100357056, "learning_rate": 1.8505830805171217e-05, "loss": 0.1937575101852417, "step": 40460 }, { "epoch": 0.20101725559540248, "grad_norm": 2.356700897216797, "learning_rate": 1.850498476799684e-05, "loss": 0.18994879722595215, "step": 40470 }, { "epoch": 0.20106692627877173, "grad_norm": 1.1658748388290405, "learning_rate": 1.8504138510716874e-05, "loss": 0.17674360275268555, "step": 40480 }, { "epoch": 0.20111659696214101, "grad_norm": 1.8978041410446167, "learning_rate": 1.8503292033353228e-05, "loss": 0.1748736619949341, "step": 40490 }, { "epoch": 0.20116626764551027, "grad_norm": 1.5669456720352173, "learning_rate": 1.8502445335927803e-05, "loss": 0.1767621159553528, "step": 40500 }, { "epoch": 0.20121593832887952, "grad_norm": 1.3599692583084106, "learning_rate": 1.8501598418462513e-05, "loss": 0.18425681591033935, "step": 40510 }, { "epoch": 0.2012656090122488, "grad_norm": 1.243119239807129, "learning_rate": 1.850075128097927e-05, "loss": 0.19864609241485595, "step": 40520 }, { "epoch": 0.20131527969561805, "grad_norm": 2.4407901763916016, "learning_rate": 1.8499903923500002e-05, "loss": 0.1781378746032715, "step": 40530 }, { "epoch": 0.2013649503789873, "grad_norm": 1.056107759475708, "learning_rate": 1.8499056346046637e-05, "loss": 0.17528313398361206, "step": 40540 }, { "epoch": 0.20141462106235658, "grad_norm": 0.9252955317497253, "learning_rate": 1.8498208548641116e-05, "loss": 0.19048864841461183, "step": 40550 }, { "epoch": 0.20146429174572583, "grad_norm": 1.110190510749817, "learning_rate": 1.849736053130537e-05, "loss": 0.16637907028198243, "step": 40560 }, { "epoch": 0.2015139624290951, "grad_norm": 1.2735447883605957, "learning_rate": 1.8496512294061353e-05, "loss": 0.18126130104064941, "step": 40570 }, { "epoch": 0.20156363311246436, "grad_norm": 0.9269357323646545, "learning_rate": 1.8495663836931014e-05, "loss": 0.18915715217590331, "step": 40580 }, { "epoch": 0.20161330379583361, "grad_norm": 2.676279306411743, "learning_rate": 1.849481515993631e-05, "loss": 0.18179244995117189, "step": 40590 }, { "epoch": 0.2016629744792029, "grad_norm": 2.1707568168640137, "learning_rate": 1.8493966263099204e-05, "loss": 0.18927181959152223, "step": 40600 }, { "epoch": 0.20171264516257215, "grad_norm": 0.8543811440467834, "learning_rate": 1.8493117146441667e-05, "loss": 0.1806497812271118, "step": 40610 }, { "epoch": 0.2017623158459414, "grad_norm": 0.9825629591941833, "learning_rate": 1.8492267809985677e-05, "loss": 0.17693480253219604, "step": 40620 }, { "epoch": 0.20181198652931068, "grad_norm": 1.2979568243026733, "learning_rate": 1.8491418253753205e-05, "loss": 0.17013225555419922, "step": 40630 }, { "epoch": 0.20186165721267993, "grad_norm": 0.9801089763641357, "learning_rate": 1.8490568477766245e-05, "loss": 0.19000403881072997, "step": 40640 }, { "epoch": 0.20191132789604918, "grad_norm": 0.9807243347167969, "learning_rate": 1.848971848204679e-05, "loss": 0.16931526660919188, "step": 40650 }, { "epoch": 0.20196099857941846, "grad_norm": 1.3420323133468628, "learning_rate": 1.848886826661683e-05, "loss": 0.17331559658050538, "step": 40660 }, { "epoch": 0.2020106692627877, "grad_norm": 0.9484589099884033, "learning_rate": 1.8488017831498376e-05, "loss": 0.19896798133850097, "step": 40670 }, { "epoch": 0.202060339946157, "grad_norm": 1.469497799873352, "learning_rate": 1.848716717671343e-05, "loss": 0.181953501701355, "step": 40680 }, { "epoch": 0.20211001062952624, "grad_norm": 1.1659315824508667, "learning_rate": 1.8486316302284012e-05, "loss": 0.18227946758270264, "step": 40690 }, { "epoch": 0.2021596813128955, "grad_norm": 0.6751305460929871, "learning_rate": 1.848546520823214e-05, "loss": 0.17897529602050782, "step": 40700 }, { "epoch": 0.20220935199626477, "grad_norm": 1.501394271850586, "learning_rate": 1.8484613894579842e-05, "loss": 0.17407758235931398, "step": 40710 }, { "epoch": 0.20225902267963403, "grad_norm": 0.8295407891273499, "learning_rate": 1.8483762361349145e-05, "loss": 0.1899496793746948, "step": 40720 }, { "epoch": 0.20230869336300328, "grad_norm": 2.2687535285949707, "learning_rate": 1.848291060856209e-05, "loss": 0.1841797113418579, "step": 40730 }, { "epoch": 0.20235836404637256, "grad_norm": 0.610578179359436, "learning_rate": 1.848205863624072e-05, "loss": 0.17076382637023926, "step": 40740 }, { "epoch": 0.2024080347297418, "grad_norm": 0.7611512541770935, "learning_rate": 1.8481206444407083e-05, "loss": 0.16928036212921144, "step": 40750 }, { "epoch": 0.20245770541311106, "grad_norm": 0.8105504512786865, "learning_rate": 1.848035403308323e-05, "loss": 0.18822076320648193, "step": 40760 }, { "epoch": 0.20250737609648034, "grad_norm": 1.474419355392456, "learning_rate": 1.8479501402291228e-05, "loss": 0.1825042724609375, "step": 40770 }, { "epoch": 0.2025570467798496, "grad_norm": 1.5385836362838745, "learning_rate": 1.8478648552053138e-05, "loss": 0.17234292030334472, "step": 40780 }, { "epoch": 0.20260671746321887, "grad_norm": 1.2417289018630981, "learning_rate": 1.8477795482391034e-05, "loss": 0.18126100301742554, "step": 40790 }, { "epoch": 0.20265638814658812, "grad_norm": 0.7828376889228821, "learning_rate": 1.847694219332699e-05, "loss": 0.17018730640411378, "step": 40800 }, { "epoch": 0.20270605882995738, "grad_norm": 2.065051794052124, "learning_rate": 1.847608868488309e-05, "loss": 0.15622942447662352, "step": 40810 }, { "epoch": 0.20275572951332665, "grad_norm": 2.1958205699920654, "learning_rate": 1.8475234957081425e-05, "loss": 0.17022169828414918, "step": 40820 }, { "epoch": 0.2028054001966959, "grad_norm": 1.1595172882080078, "learning_rate": 1.8474381009944087e-05, "loss": 0.1620426058769226, "step": 40830 }, { "epoch": 0.20285507088006516, "grad_norm": 1.9445445537567139, "learning_rate": 1.8473526843493176e-05, "loss": 0.1919904589653015, "step": 40840 }, { "epoch": 0.20290474156343444, "grad_norm": 1.6586997509002686, "learning_rate": 1.8472672457750794e-05, "loss": 0.1813117504119873, "step": 40850 }, { "epoch": 0.2029544122468037, "grad_norm": 2.83526873588562, "learning_rate": 1.847181785273906e-05, "loss": 0.1759174346923828, "step": 40860 }, { "epoch": 0.20300408293017294, "grad_norm": 1.5339020490646362, "learning_rate": 1.8470963028480086e-05, "loss": 0.17293074131011962, "step": 40870 }, { "epoch": 0.20305375361354222, "grad_norm": 1.4640631675720215, "learning_rate": 1.847010798499599e-05, "loss": 0.174880051612854, "step": 40880 }, { "epoch": 0.20310342429691147, "grad_norm": 1.0860728025436401, "learning_rate": 1.846925272230891e-05, "loss": 0.17936325073242188, "step": 40890 }, { "epoch": 0.20315309498028075, "grad_norm": 0.8938053846359253, "learning_rate": 1.8468397240440974e-05, "loss": 0.17270684242248535, "step": 40900 }, { "epoch": 0.20320276566365, "grad_norm": 1.2495944499969482, "learning_rate": 1.846754153941432e-05, "loss": 0.17860926389694215, "step": 40910 }, { "epoch": 0.20325243634701926, "grad_norm": 1.2909668684005737, "learning_rate": 1.84666856192511e-05, "loss": 0.17564744949340821, "step": 40920 }, { "epoch": 0.20330210703038853, "grad_norm": 0.8076837658882141, "learning_rate": 1.8465829479973454e-05, "loss": 0.17668231725692748, "step": 40930 }, { "epoch": 0.2033517777137578, "grad_norm": 2.3552346229553223, "learning_rate": 1.846497312160355e-05, "loss": 0.1757823944091797, "step": 40940 }, { "epoch": 0.20340144839712704, "grad_norm": 0.8273899555206299, "learning_rate": 1.8464116544163542e-05, "loss": 0.17113125324249268, "step": 40950 }, { "epoch": 0.20345111908049632, "grad_norm": 0.930590808391571, "learning_rate": 1.8463259747675604e-05, "loss": 0.177942156791687, "step": 40960 }, { "epoch": 0.20350078976386557, "grad_norm": 1.087542176246643, "learning_rate": 1.8462402732161907e-05, "loss": 0.18607146739959718, "step": 40970 }, { "epoch": 0.20355046044723482, "grad_norm": 1.0194038152694702, "learning_rate": 1.846154549764463e-05, "loss": 0.16473218202590942, "step": 40980 }, { "epoch": 0.2036001311306041, "grad_norm": 1.8615803718566895, "learning_rate": 1.8460688044145953e-05, "loss": 0.16273654699325563, "step": 40990 }, { "epoch": 0.20364980181397335, "grad_norm": 1.4242078065872192, "learning_rate": 1.8459830371688077e-05, "loss": 0.1791501522064209, "step": 41000 }, { "epoch": 0.20369947249734263, "grad_norm": 1.0966193675994873, "learning_rate": 1.8458972480293188e-05, "loss": 0.19713698625564574, "step": 41010 }, { "epoch": 0.20374914318071188, "grad_norm": 0.768273651599884, "learning_rate": 1.8458114369983494e-05, "loss": 0.16711701154708863, "step": 41020 }, { "epoch": 0.20379881386408114, "grad_norm": 1.4930437803268433, "learning_rate": 1.8457256040781204e-05, "loss": 0.18077024221420288, "step": 41030 }, { "epoch": 0.20384848454745041, "grad_norm": 1.7926762104034424, "learning_rate": 1.8456397492708525e-05, "loss": 0.19927374124526978, "step": 41040 }, { "epoch": 0.20389815523081967, "grad_norm": 0.6929062604904175, "learning_rate": 1.845553872578768e-05, "loss": 0.17991844415664673, "step": 41050 }, { "epoch": 0.20394782591418892, "grad_norm": 1.8381131887435913, "learning_rate": 1.8454679740040893e-05, "loss": 0.18385121822357178, "step": 41060 }, { "epoch": 0.2039974965975582, "grad_norm": 0.8769130110740662, "learning_rate": 1.845382053549039e-05, "loss": 0.1669348120689392, "step": 41070 }, { "epoch": 0.20404716728092745, "grad_norm": 0.5939922332763672, "learning_rate": 1.8452961112158415e-05, "loss": 0.18627142906188965, "step": 41080 }, { "epoch": 0.2040968379642967, "grad_norm": 1.1495591402053833, "learning_rate": 1.8452101470067204e-05, "loss": 0.1648662805557251, "step": 41090 }, { "epoch": 0.20414650864766598, "grad_norm": 1.5855861902236938, "learning_rate": 1.8451241609239007e-05, "loss": 0.1884296417236328, "step": 41100 }, { "epoch": 0.20419617933103523, "grad_norm": 1.5329986810684204, "learning_rate": 1.8450381529696072e-05, "loss": 0.17468626499176027, "step": 41110 }, { "epoch": 0.2042458500144045, "grad_norm": 0.7392380237579346, "learning_rate": 1.8449521231460665e-05, "loss": 0.17490408420562745, "step": 41120 }, { "epoch": 0.20429552069777376, "grad_norm": 1.0802828073501587, "learning_rate": 1.8448660714555045e-05, "loss": 0.16939759254455566, "step": 41130 }, { "epoch": 0.20434519138114302, "grad_norm": 1.7417998313903809, "learning_rate": 1.844779997900148e-05, "loss": 0.17299926280975342, "step": 41140 }, { "epoch": 0.2043948620645123, "grad_norm": 3.3214776515960693, "learning_rate": 1.8446939024822248e-05, "loss": 0.17844678163528443, "step": 41150 }, { "epoch": 0.20444453274788155, "grad_norm": 0.7002577185630798, "learning_rate": 1.8446077852039634e-05, "loss": 0.16946611404418946, "step": 41160 }, { "epoch": 0.2044942034312508, "grad_norm": 1.602837324142456, "learning_rate": 1.844521646067592e-05, "loss": 0.17409865856170653, "step": 41170 }, { "epoch": 0.20454387411462008, "grad_norm": 1.4317177534103394, "learning_rate": 1.8444354850753397e-05, "loss": 0.18058488368988038, "step": 41180 }, { "epoch": 0.20459354479798933, "grad_norm": 2.1008195877075195, "learning_rate": 1.8443493022294366e-05, "loss": 0.1705712080001831, "step": 41190 }, { "epoch": 0.20464321548135858, "grad_norm": 1.3530902862548828, "learning_rate": 1.8442630975321133e-05, "loss": 0.18450214862823486, "step": 41200 }, { "epoch": 0.20469288616472786, "grad_norm": 1.4080798625946045, "learning_rate": 1.8441768709856e-05, "loss": 0.18016984462738037, "step": 41210 }, { "epoch": 0.2047425568480971, "grad_norm": 1.1923383474349976, "learning_rate": 1.8440906225921297e-05, "loss": 0.1781625747680664, "step": 41220 }, { "epoch": 0.20479222753146636, "grad_norm": 1.6562323570251465, "learning_rate": 1.844004352353933e-05, "loss": 0.17819461822509766, "step": 41230 }, { "epoch": 0.20484189821483564, "grad_norm": 1.7660691738128662, "learning_rate": 1.8439180602732426e-05, "loss": 0.2019479751586914, "step": 41240 }, { "epoch": 0.2048915688982049, "grad_norm": 2.131502866744995, "learning_rate": 1.8438317463522926e-05, "loss": 0.18383826017379762, "step": 41250 }, { "epoch": 0.20494123958157417, "grad_norm": 1.086193323135376, "learning_rate": 1.843745410593316e-05, "loss": 0.17410523891448976, "step": 41260 }, { "epoch": 0.20499091026494343, "grad_norm": 0.796544075012207, "learning_rate": 1.8436590529985478e-05, "loss": 0.17398951053619385, "step": 41270 }, { "epoch": 0.20504058094831268, "grad_norm": 1.191626787185669, "learning_rate": 1.8435726735702225e-05, "loss": 0.17402150630950927, "step": 41280 }, { "epoch": 0.20509025163168196, "grad_norm": 1.5436476469039917, "learning_rate": 1.843486272310575e-05, "loss": 0.18182940483093263, "step": 41290 }, { "epoch": 0.2051399223150512, "grad_norm": 1.2073463201522827, "learning_rate": 1.8433998492218426e-05, "loss": 0.16446166038513182, "step": 41300 }, { "epoch": 0.20518959299842046, "grad_norm": 1.4224122762680054, "learning_rate": 1.8433134043062607e-05, "loss": 0.17955081462860106, "step": 41310 }, { "epoch": 0.20523926368178974, "grad_norm": 1.0929851531982422, "learning_rate": 1.8432269375660672e-05, "loss": 0.17898573875427246, "step": 41320 }, { "epoch": 0.205288934365159, "grad_norm": 0.9422641396522522, "learning_rate": 1.8431404490035e-05, "loss": 0.16983840465545655, "step": 41330 }, { "epoch": 0.20533860504852824, "grad_norm": 1.0311027765274048, "learning_rate": 1.8430539386207965e-05, "loss": 0.1892026662826538, "step": 41340 }, { "epoch": 0.20538827573189752, "grad_norm": 1.3789163827896118, "learning_rate": 1.842967406420196e-05, "loss": 0.18724914789199829, "step": 41350 }, { "epoch": 0.20543794641526678, "grad_norm": 1.1426023244857788, "learning_rate": 1.8428808524039383e-05, "loss": 0.18709156513214112, "step": 41360 }, { "epoch": 0.20548761709863606, "grad_norm": 0.8458142876625061, "learning_rate": 1.8427942765742628e-05, "loss": 0.1795055389404297, "step": 41370 }, { "epoch": 0.2055372877820053, "grad_norm": 1.3307501077651978, "learning_rate": 1.8427076789334103e-05, "loss": 0.18250679969787598, "step": 41380 }, { "epoch": 0.20558695846537456, "grad_norm": 0.8980101346969604, "learning_rate": 1.842621059483622e-05, "loss": 0.17208776473999024, "step": 41390 }, { "epoch": 0.20563662914874384, "grad_norm": 1.2128608226776123, "learning_rate": 1.8425344182271395e-05, "loss": 0.1724928140640259, "step": 41400 }, { "epoch": 0.2056862998321131, "grad_norm": 1.0552115440368652, "learning_rate": 1.8424477551662052e-05, "loss": 0.17210710048675537, "step": 41410 }, { "epoch": 0.20573597051548234, "grad_norm": 1.3772978782653809, "learning_rate": 1.8423610703030612e-05, "loss": 0.18117728233337402, "step": 41420 }, { "epoch": 0.20578564119885162, "grad_norm": 1.8715521097183228, "learning_rate": 1.842274363639952e-05, "loss": 0.17824068069458007, "step": 41430 }, { "epoch": 0.20583531188222087, "grad_norm": 0.9828046560287476, "learning_rate": 1.8421876351791206e-05, "loss": 0.16759440898895264, "step": 41440 }, { "epoch": 0.20588498256559012, "grad_norm": 0.8549792766571045, "learning_rate": 1.842100884922812e-05, "loss": 0.18403635025024415, "step": 41450 }, { "epoch": 0.2059346532489594, "grad_norm": 1.307658076286316, "learning_rate": 1.842014112873271e-05, "loss": 0.1659249782562256, "step": 41460 }, { "epoch": 0.20598432393232866, "grad_norm": 1.319100022315979, "learning_rate": 1.841927319032743e-05, "loss": 0.17653768062591552, "step": 41470 }, { "epoch": 0.20603399461569794, "grad_norm": 0.6634377837181091, "learning_rate": 1.841840503403475e-05, "loss": 0.17640256881713867, "step": 41480 }, { "epoch": 0.2060836652990672, "grad_norm": 1.5073317289352417, "learning_rate": 1.8417536659877128e-05, "loss": 0.19120149612426757, "step": 41490 }, { "epoch": 0.20613333598243644, "grad_norm": 1.337093710899353, "learning_rate": 1.8416668067877048e-05, "loss": 0.17995719909667968, "step": 41500 }, { "epoch": 0.20618300666580572, "grad_norm": 1.6722564697265625, "learning_rate": 1.8415799258056978e-05, "loss": 0.17938790321350098, "step": 41510 }, { "epoch": 0.20623267734917497, "grad_norm": 1.047174096107483, "learning_rate": 1.841493023043941e-05, "loss": 0.19079298973083497, "step": 41520 }, { "epoch": 0.20628234803254422, "grad_norm": 0.842549741268158, "learning_rate": 1.841406098504683e-05, "loss": 0.18867032527923583, "step": 41530 }, { "epoch": 0.2063320187159135, "grad_norm": 1.708946704864502, "learning_rate": 1.8413191521901733e-05, "loss": 0.15944627523422242, "step": 41540 }, { "epoch": 0.20638168939928275, "grad_norm": 2.162966012954712, "learning_rate": 1.8412321841026624e-05, "loss": 0.1838834524154663, "step": 41550 }, { "epoch": 0.206431360082652, "grad_norm": 1.9313472509384155, "learning_rate": 1.8411451942444006e-05, "loss": 0.17418360710144043, "step": 41560 }, { "epoch": 0.20648103076602128, "grad_norm": 1.0643171072006226, "learning_rate": 1.84105818261764e-05, "loss": 0.17438859939575196, "step": 41570 }, { "epoch": 0.20653070144939054, "grad_norm": 1.4171255826950073, "learning_rate": 1.840971149224631e-05, "loss": 0.17435271739959718, "step": 41580 }, { "epoch": 0.20658037213275982, "grad_norm": 3.4005439281463623, "learning_rate": 1.840884094067627e-05, "loss": 0.18290836811065675, "step": 41590 }, { "epoch": 0.20663004281612907, "grad_norm": 1.3282699584960938, "learning_rate": 1.840797017148881e-05, "loss": 0.18762674331665039, "step": 41600 }, { "epoch": 0.20667971349949832, "grad_norm": 1.0608417987823486, "learning_rate": 1.8407099184706464e-05, "loss": 0.18606353998184205, "step": 41610 }, { "epoch": 0.2067293841828676, "grad_norm": 0.8518176674842834, "learning_rate": 1.8406227980351766e-05, "loss": 0.16700183153152465, "step": 41620 }, { "epoch": 0.20677905486623685, "grad_norm": 1.0295511484146118, "learning_rate": 1.840535655844727e-05, "loss": 0.1783432960510254, "step": 41630 }, { "epoch": 0.2068287255496061, "grad_norm": 1.1577837467193604, "learning_rate": 1.8404484919015528e-05, "loss": 0.16688554286956786, "step": 41640 }, { "epoch": 0.20687839623297538, "grad_norm": 1.4193012714385986, "learning_rate": 1.8403613062079096e-05, "loss": 0.18933503627777098, "step": 41650 }, { "epoch": 0.20692806691634463, "grad_norm": 2.0319271087646484, "learning_rate": 1.8402740987660534e-05, "loss": 0.18018044233322145, "step": 41660 }, { "epoch": 0.20697773759971388, "grad_norm": 0.8478510975837708, "learning_rate": 1.8401868695782417e-05, "loss": 0.16935535669326782, "step": 41670 }, { "epoch": 0.20702740828308316, "grad_norm": 1.339753270149231, "learning_rate": 1.840099618646731e-05, "loss": 0.16785292625427245, "step": 41680 }, { "epoch": 0.20707707896645242, "grad_norm": 1.2122136354446411, "learning_rate": 1.8400123459737808e-05, "loss": 0.17548830509185792, "step": 41690 }, { "epoch": 0.2071267496498217, "grad_norm": 1.0787314176559448, "learning_rate": 1.8399250515616485e-05, "loss": 0.18459277153015136, "step": 41700 }, { "epoch": 0.20717642033319095, "grad_norm": 1.1073554754257202, "learning_rate": 1.8398377354125936e-05, "loss": 0.1731577157974243, "step": 41710 }, { "epoch": 0.2072260910165602, "grad_norm": 1.0943949222564697, "learning_rate": 1.8397503975288758e-05, "loss": 0.17928884029388428, "step": 41720 }, { "epoch": 0.20727576169992948, "grad_norm": 2.9050705432891846, "learning_rate": 1.839663037912755e-05, "loss": 0.18707653284072875, "step": 41730 }, { "epoch": 0.20732543238329873, "grad_norm": 0.7247934937477112, "learning_rate": 1.8395756565664927e-05, "loss": 0.17666250467300415, "step": 41740 }, { "epoch": 0.20737510306666798, "grad_norm": 1.2943960428237915, "learning_rate": 1.83948825349235e-05, "loss": 0.1657942056655884, "step": 41750 }, { "epoch": 0.20742477375003726, "grad_norm": 1.690767765045166, "learning_rate": 1.839400828692589e-05, "loss": 0.1746798038482666, "step": 41760 }, { "epoch": 0.2074744444334065, "grad_norm": 1.5343587398529053, "learning_rate": 1.839313382169472e-05, "loss": 0.17918362617492675, "step": 41770 }, { "epoch": 0.20752411511677576, "grad_norm": 2.874192714691162, "learning_rate": 1.839225913925262e-05, "loss": 0.18472766876220703, "step": 41780 }, { "epoch": 0.20757378580014504, "grad_norm": 0.8801800012588501, "learning_rate": 1.839138423962223e-05, "loss": 0.1907962441444397, "step": 41790 }, { "epoch": 0.2076234564835143, "grad_norm": 1.5315754413604736, "learning_rate": 1.839050912282619e-05, "loss": 0.17489488124847413, "step": 41800 }, { "epoch": 0.20767312716688358, "grad_norm": 0.6428095698356628, "learning_rate": 1.8389633788887142e-05, "loss": 0.1696308970451355, "step": 41810 }, { "epoch": 0.20772279785025283, "grad_norm": 1.666544795036316, "learning_rate": 1.8388758237827747e-05, "loss": 0.1752593159675598, "step": 41820 }, { "epoch": 0.20777246853362208, "grad_norm": 1.029070258140564, "learning_rate": 1.8387882469670667e-05, "loss": 0.18683894872665405, "step": 41830 }, { "epoch": 0.20782213921699136, "grad_norm": 1.2112044095993042, "learning_rate": 1.8387006484438558e-05, "loss": 0.1731393814086914, "step": 41840 }, { "epoch": 0.2078718099003606, "grad_norm": 1.009082555770874, "learning_rate": 1.838613028215409e-05, "loss": 0.18296530246734619, "step": 41850 }, { "epoch": 0.20792148058372986, "grad_norm": 2.1234779357910156, "learning_rate": 1.8385253862839947e-05, "loss": 0.17148785591125487, "step": 41860 }, { "epoch": 0.20797115126709914, "grad_norm": 2.333340883255005, "learning_rate": 1.8384377226518806e-05, "loss": 0.18966279029846192, "step": 41870 }, { "epoch": 0.2080208219504684, "grad_norm": 0.6489870548248291, "learning_rate": 1.838350037321335e-05, "loss": 0.16543910503387452, "step": 41880 }, { "epoch": 0.20807049263383764, "grad_norm": 2.5815558433532715, "learning_rate": 1.8382623302946276e-05, "loss": 0.18092552423477173, "step": 41890 }, { "epoch": 0.20812016331720692, "grad_norm": 2.058331251144409, "learning_rate": 1.8381746015740284e-05, "loss": 0.18686161041259766, "step": 41900 }, { "epoch": 0.20816983400057618, "grad_norm": 0.7260206341743469, "learning_rate": 1.8380868511618073e-05, "loss": 0.17768304347991942, "step": 41910 }, { "epoch": 0.20821950468394546, "grad_norm": 1.6723339557647705, "learning_rate": 1.8379990790602355e-05, "loss": 0.17697713375091553, "step": 41920 }, { "epoch": 0.2082691753673147, "grad_norm": 1.1815543174743652, "learning_rate": 1.8379112852715845e-05, "loss": 0.16370460987091065, "step": 41930 }, { "epoch": 0.20831884605068396, "grad_norm": 1.9116450548171997, "learning_rate": 1.8378234697981267e-05, "loss": 0.17835159301757814, "step": 41940 }, { "epoch": 0.20836851673405324, "grad_norm": 1.1539992094039917, "learning_rate": 1.8377356326421335e-05, "loss": 0.18132467269897462, "step": 41950 }, { "epoch": 0.2084181874174225, "grad_norm": 2.3683650493621826, "learning_rate": 1.8376477738058794e-05, "loss": 0.1909611701965332, "step": 41960 }, { "epoch": 0.20846785810079174, "grad_norm": 0.8419846296310425, "learning_rate": 1.837559893291638e-05, "loss": 0.16859524250030516, "step": 41970 }, { "epoch": 0.20851752878416102, "grad_norm": 2.0915157794952393, "learning_rate": 1.8374719911016834e-05, "loss": 0.17496572732925414, "step": 41980 }, { "epoch": 0.20856719946753027, "grad_norm": 1.9168729782104492, "learning_rate": 1.8373840672382898e-05, "loss": 0.16467586755752564, "step": 41990 }, { "epoch": 0.20861687015089952, "grad_norm": 1.1301429271697998, "learning_rate": 1.8372961217037338e-05, "loss": 0.18001923561096192, "step": 42000 }, { "epoch": 0.2086665408342688, "grad_norm": 2.1105456352233887, "learning_rate": 1.8372081545002904e-05, "loss": 0.18046512603759765, "step": 42010 }, { "epoch": 0.20871621151763806, "grad_norm": 0.9720484018325806, "learning_rate": 1.837120165630237e-05, "loss": 0.1709896445274353, "step": 42020 }, { "epoch": 0.2087658822010073, "grad_norm": 0.9535382390022278, "learning_rate": 1.83703215509585e-05, "loss": 0.17898952960968018, "step": 42030 }, { "epoch": 0.2088155528843766, "grad_norm": 0.9995770454406738, "learning_rate": 1.8369441228994076e-05, "loss": 0.16407649517059325, "step": 42040 }, { "epoch": 0.20886522356774584, "grad_norm": 1.662035584449768, "learning_rate": 1.836856069043188e-05, "loss": 0.18472365140914918, "step": 42050 }, { "epoch": 0.20891489425111512, "grad_norm": 2.4742672443389893, "learning_rate": 1.836767993529469e-05, "loss": 0.18173426389694214, "step": 42060 }, { "epoch": 0.20896456493448437, "grad_norm": 0.8347698450088501, "learning_rate": 1.8366798963605314e-05, "loss": 0.18921403884887694, "step": 42070 }, { "epoch": 0.20901423561785362, "grad_norm": 0.6882219910621643, "learning_rate": 1.8365917775386546e-05, "loss": 0.18103785514831544, "step": 42080 }, { "epoch": 0.2090639063012229, "grad_norm": 1.8202234506607056, "learning_rate": 1.8365036370661187e-05, "loss": 0.16665372848510743, "step": 42090 }, { "epoch": 0.20911357698459215, "grad_norm": 1.6522705554962158, "learning_rate": 1.836415474945205e-05, "loss": 0.1679581880569458, "step": 42100 }, { "epoch": 0.2091632476679614, "grad_norm": 0.8331157565116882, "learning_rate": 1.8363272911781956e-05, "loss": 0.18621509075164794, "step": 42110 }, { "epoch": 0.20921291835133068, "grad_norm": 0.7817906141281128, "learning_rate": 1.8362390857673716e-05, "loss": 0.19210387468338014, "step": 42120 }, { "epoch": 0.20926258903469994, "grad_norm": 2.3174989223480225, "learning_rate": 1.8361508587150166e-05, "loss": 0.17703967094421386, "step": 42130 }, { "epoch": 0.2093122597180692, "grad_norm": 0.9670224785804749, "learning_rate": 1.8360626100234135e-05, "loss": 0.17669552564620972, "step": 42140 }, { "epoch": 0.20936193040143847, "grad_norm": 1.7422187328338623, "learning_rate": 1.835974339694846e-05, "loss": 0.17108561992645263, "step": 42150 }, { "epoch": 0.20941160108480772, "grad_norm": 0.9742275476455688, "learning_rate": 1.835886047731599e-05, "loss": 0.17867555618286132, "step": 42160 }, { "epoch": 0.209461271768177, "grad_norm": 1.3048582077026367, "learning_rate": 1.835797734135957e-05, "loss": 0.16571913957595824, "step": 42170 }, { "epoch": 0.20951094245154625, "grad_norm": 0.49532198905944824, "learning_rate": 1.8357093989102058e-05, "loss": 0.1807028293609619, "step": 42180 }, { "epoch": 0.2095606131349155, "grad_norm": 2.176690101623535, "learning_rate": 1.8356210420566315e-05, "loss": 0.16748781204223634, "step": 42190 }, { "epoch": 0.20961028381828478, "grad_norm": 0.7207711935043335, "learning_rate": 1.8355326635775207e-05, "loss": 0.18283431529998778, "step": 42200 }, { "epoch": 0.20965995450165403, "grad_norm": 1.371471881866455, "learning_rate": 1.83544426347516e-05, "loss": 0.18376026153564454, "step": 42210 }, { "epoch": 0.20970962518502329, "grad_norm": 1.0990875959396362, "learning_rate": 1.835355841751838e-05, "loss": 0.18437695503234863, "step": 42220 }, { "epoch": 0.20975929586839256, "grad_norm": 1.2934755086898804, "learning_rate": 1.8352673984098427e-05, "loss": 0.17691490650177003, "step": 42230 }, { "epoch": 0.20980896655176182, "grad_norm": 1.620864748954773, "learning_rate": 1.835178933451463e-05, "loss": 0.1853821396827698, "step": 42240 }, { "epoch": 0.20985863723513107, "grad_norm": 2.4946906566619873, "learning_rate": 1.835090446878988e-05, "loss": 0.17731356620788574, "step": 42250 }, { "epoch": 0.20990830791850035, "grad_norm": 0.7074885964393616, "learning_rate": 1.8350019386947087e-05, "loss": 0.17202587127685548, "step": 42260 }, { "epoch": 0.2099579786018696, "grad_norm": 2.75338077545166, "learning_rate": 1.8349134089009143e-05, "loss": 0.19356645345687867, "step": 42270 }, { "epoch": 0.21000764928523888, "grad_norm": 0.8143137097358704, "learning_rate": 1.834824857499897e-05, "loss": 0.16767003536224365, "step": 42280 }, { "epoch": 0.21005731996860813, "grad_norm": 1.8795784711837769, "learning_rate": 1.834736284493948e-05, "loss": 0.17171094417572022, "step": 42290 }, { "epoch": 0.21010699065197738, "grad_norm": 1.350199818611145, "learning_rate": 1.8346476898853595e-05, "loss": 0.18650224208831787, "step": 42300 }, { "epoch": 0.21015666133534666, "grad_norm": 0.9150961637496948, "learning_rate": 1.8345590736764242e-05, "loss": 0.17660562992095946, "step": 42310 }, { "epoch": 0.2102063320187159, "grad_norm": 1.3468486070632935, "learning_rate": 1.834470435869436e-05, "loss": 0.19676451683044432, "step": 42320 }, { "epoch": 0.21025600270208517, "grad_norm": 1.3200517892837524, "learning_rate": 1.8343817764666883e-05, "loss": 0.1949912428855896, "step": 42330 }, { "epoch": 0.21030567338545444, "grad_norm": 2.009359359741211, "learning_rate": 1.8342930954704756e-05, "loss": 0.16066489219665528, "step": 42340 }, { "epoch": 0.2103553440688237, "grad_norm": 1.1409964561462402, "learning_rate": 1.8342043928830932e-05, "loss": 0.18348164558410646, "step": 42350 }, { "epoch": 0.21040501475219295, "grad_norm": 1.4311941862106323, "learning_rate": 1.8341156687068366e-05, "loss": 0.1781921625137329, "step": 42360 }, { "epoch": 0.21045468543556223, "grad_norm": 0.7418750524520874, "learning_rate": 1.8340269229440015e-05, "loss": 0.1966611385345459, "step": 42370 }, { "epoch": 0.21050435611893148, "grad_norm": 0.9912852644920349, "learning_rate": 1.8339381555968853e-05, "loss": 0.18346407413482665, "step": 42380 }, { "epoch": 0.21055402680230076, "grad_norm": 0.5346701741218567, "learning_rate": 1.833849366667785e-05, "loss": 0.16764791011810304, "step": 42390 }, { "epoch": 0.21060369748567, "grad_norm": 1.8729599714279175, "learning_rate": 1.833760556158998e-05, "loss": 0.18483588695526124, "step": 42400 }, { "epoch": 0.21065336816903926, "grad_norm": 1.8341914415359497, "learning_rate": 1.8336717240728234e-05, "loss": 0.16065044403076173, "step": 42410 }, { "epoch": 0.21070303885240854, "grad_norm": 1.5626342296600342, "learning_rate": 1.8335828704115597e-05, "loss": 0.16270849704742432, "step": 42420 }, { "epoch": 0.2107527095357778, "grad_norm": 0.8924530744552612, "learning_rate": 1.8334939951775062e-05, "loss": 0.17461748123168946, "step": 42430 }, { "epoch": 0.21080238021914705, "grad_norm": 2.370220422744751, "learning_rate": 1.8334050983729636e-05, "loss": 0.18115190267562867, "step": 42440 }, { "epoch": 0.21085205090251632, "grad_norm": 2.0691378116607666, "learning_rate": 1.8333161800002323e-05, "loss": 0.19107623100280763, "step": 42450 }, { "epoch": 0.21090172158588558, "grad_norm": 1.4095253944396973, "learning_rate": 1.833227240061613e-05, "loss": 0.18297231197357178, "step": 42460 }, { "epoch": 0.21095139226925483, "grad_norm": 0.6578679084777832, "learning_rate": 1.8331382785594076e-05, "loss": 0.17851170301437377, "step": 42470 }, { "epoch": 0.2110010629526241, "grad_norm": 0.8920248746871948, "learning_rate": 1.833049295495919e-05, "loss": 0.16782511472702027, "step": 42480 }, { "epoch": 0.21105073363599336, "grad_norm": 1.8000094890594482, "learning_rate": 1.8329602908734493e-05, "loss": 0.19169563055038452, "step": 42490 }, { "epoch": 0.21110040431936264, "grad_norm": 1.8634628057479858, "learning_rate": 1.8328712646943022e-05, "loss": 0.17656726837158204, "step": 42500 }, { "epoch": 0.2111500750027319, "grad_norm": 0.582120418548584, "learning_rate": 1.8327822169607815e-05, "loss": 0.1640254259109497, "step": 42510 }, { "epoch": 0.21119974568610114, "grad_norm": 1.2803000211715698, "learning_rate": 1.8326931476751922e-05, "loss": 0.1808115005493164, "step": 42520 }, { "epoch": 0.21124941636947042, "grad_norm": 1.87867271900177, "learning_rate": 1.832604056839839e-05, "loss": 0.1896107316017151, "step": 42530 }, { "epoch": 0.21129908705283967, "grad_norm": 1.0167951583862305, "learning_rate": 1.8325149444570276e-05, "loss": 0.17266173362731935, "step": 42540 }, { "epoch": 0.21134875773620893, "grad_norm": 1.1847714185714722, "learning_rate": 1.832425810529064e-05, "loss": 0.18030107021331787, "step": 42550 }, { "epoch": 0.2113984284195782, "grad_norm": 2.193678379058838, "learning_rate": 1.832336655058255e-05, "loss": 0.17351894378662108, "step": 42560 }, { "epoch": 0.21144809910294746, "grad_norm": 1.397065281867981, "learning_rate": 1.8322474780469085e-05, "loss": 0.1931218147277832, "step": 42570 }, { "epoch": 0.2114977697863167, "grad_norm": 1.8874222040176392, "learning_rate": 1.8321582794973315e-05, "loss": 0.17604238986968995, "step": 42580 }, { "epoch": 0.211547440469686, "grad_norm": 0.6872298121452332, "learning_rate": 1.832069059411833e-05, "loss": 0.16939467191696167, "step": 42590 }, { "epoch": 0.21159711115305524, "grad_norm": 1.7890679836273193, "learning_rate": 1.8319798177927217e-05, "loss": 0.1742169141769409, "step": 42600 }, { "epoch": 0.21164678183642452, "grad_norm": 0.9234161376953125, "learning_rate": 1.8318905546423074e-05, "loss": 0.16971750259399415, "step": 42610 }, { "epoch": 0.21169645251979377, "grad_norm": 1.1872584819793701, "learning_rate": 1.8318012699628997e-05, "loss": 0.20053513050079347, "step": 42620 }, { "epoch": 0.21174612320316302, "grad_norm": 1.8703583478927612, "learning_rate": 1.83171196375681e-05, "loss": 0.1784693717956543, "step": 42630 }, { "epoch": 0.2117957938865323, "grad_norm": 0.9529250860214233, "learning_rate": 1.831622636026349e-05, "loss": 0.17394555807113649, "step": 42640 }, { "epoch": 0.21184546456990155, "grad_norm": 0.9022469520568848, "learning_rate": 1.8315332867738286e-05, "loss": 0.17763935327529906, "step": 42650 }, { "epoch": 0.2118951352532708, "grad_norm": 0.7163489460945129, "learning_rate": 1.831443916001561e-05, "loss": 0.17230842113494874, "step": 42660 }, { "epoch": 0.21194480593664008, "grad_norm": 1.2792092561721802, "learning_rate": 1.831354523711859e-05, "loss": 0.1613885283470154, "step": 42670 }, { "epoch": 0.21199447662000934, "grad_norm": 1.1097514629364014, "learning_rate": 1.8312651099070364e-05, "loss": 0.19517018795013427, "step": 42680 }, { "epoch": 0.2120441473033786, "grad_norm": 1.883280634880066, "learning_rate": 1.831175674589407e-05, "loss": 0.1916282892227173, "step": 42690 }, { "epoch": 0.21209381798674787, "grad_norm": 1.4432721138000488, "learning_rate": 1.8310862177612852e-05, "loss": 0.16996067762374878, "step": 42700 }, { "epoch": 0.21214348867011712, "grad_norm": 1.169266700744629, "learning_rate": 1.830996739424986e-05, "loss": 0.18139252662658692, "step": 42710 }, { "epoch": 0.2121931593534864, "grad_norm": 0.785390317440033, "learning_rate": 1.830907239582826e-05, "loss": 0.18617337942123413, "step": 42720 }, { "epoch": 0.21224283003685565, "grad_norm": 1.183029294013977, "learning_rate": 1.8308177182371204e-05, "loss": 0.16761096715927123, "step": 42730 }, { "epoch": 0.2122925007202249, "grad_norm": 1.6238547563552856, "learning_rate": 1.8307281753901862e-05, "loss": 0.19486762285232545, "step": 42740 }, { "epoch": 0.21234217140359418, "grad_norm": 0.8996672034263611, "learning_rate": 1.830638611044341e-05, "loss": 0.1716387987136841, "step": 42750 }, { "epoch": 0.21239184208696343, "grad_norm": 0.7116263508796692, "learning_rate": 1.8305490252019023e-05, "loss": 0.1880906343460083, "step": 42760 }, { "epoch": 0.21244151277033269, "grad_norm": 2.6737048625946045, "learning_rate": 1.8304594178651887e-05, "loss": 0.18880609273910523, "step": 42770 }, { "epoch": 0.21249118345370197, "grad_norm": 1.3020238876342773, "learning_rate": 1.83036978903652e-05, "loss": 0.17540992498397828, "step": 42780 }, { "epoch": 0.21254085413707122, "grad_norm": 2.4084367752075195, "learning_rate": 1.8302801387182145e-05, "loss": 0.185654878616333, "step": 42790 }, { "epoch": 0.21259052482044047, "grad_norm": 1.4137651920318604, "learning_rate": 1.8301904669125927e-05, "loss": 0.16783463954925537, "step": 42800 }, { "epoch": 0.21264019550380975, "grad_norm": 1.6426292657852173, "learning_rate": 1.830100773621976e-05, "loss": 0.18948973417282106, "step": 42810 }, { "epoch": 0.212689866187179, "grad_norm": 1.137368083000183, "learning_rate": 1.8300110588486845e-05, "loss": 0.17798399925231934, "step": 42820 }, { "epoch": 0.21273953687054825, "grad_norm": 1.0357261896133423, "learning_rate": 1.8299213225950404e-05, "loss": 0.1948992371559143, "step": 42830 }, { "epoch": 0.21278920755391753, "grad_norm": 1.6748428344726562, "learning_rate": 1.8298315648633665e-05, "loss": 0.16491296291351318, "step": 42840 }, { "epoch": 0.21283887823728678, "grad_norm": 1.0327222347259521, "learning_rate": 1.8297417856559854e-05, "loss": 0.18320055007934571, "step": 42850 }, { "epoch": 0.21288854892065606, "grad_norm": 0.8585812449455261, "learning_rate": 1.8296519849752203e-05, "loss": 0.18326358795166015, "step": 42860 }, { "epoch": 0.21293821960402531, "grad_norm": 0.9329996705055237, "learning_rate": 1.8295621628233958e-05, "loss": 0.177998149394989, "step": 42870 }, { "epoch": 0.21298789028739457, "grad_norm": 1.3959438800811768, "learning_rate": 1.829472319202836e-05, "loss": 0.1730794668197632, "step": 42880 }, { "epoch": 0.21303756097076385, "grad_norm": 1.2086808681488037, "learning_rate": 1.8293824541158657e-05, "loss": 0.17036445140838624, "step": 42890 }, { "epoch": 0.2130872316541331, "grad_norm": 1.89687180519104, "learning_rate": 1.8292925675648113e-05, "loss": 0.16727819442749023, "step": 42900 }, { "epoch": 0.21313690233750235, "grad_norm": 0.7409428358078003, "learning_rate": 1.8292026595519986e-05, "loss": 0.17623109817504884, "step": 42910 }, { "epoch": 0.21318657302087163, "grad_norm": 0.8207670450210571, "learning_rate": 1.8291127300797543e-05, "loss": 0.18380982875823976, "step": 42920 }, { "epoch": 0.21323624370424088, "grad_norm": 1.4345614910125732, "learning_rate": 1.8290227791504063e-05, "loss": 0.17173253297805785, "step": 42930 }, { "epoch": 0.21328591438761013, "grad_norm": 2.013883113861084, "learning_rate": 1.828932806766282e-05, "loss": 0.1873713731765747, "step": 42940 }, { "epoch": 0.2133355850709794, "grad_norm": 0.767317533493042, "learning_rate": 1.82884281292971e-05, "loss": 0.17161306142807006, "step": 42950 }, { "epoch": 0.21338525575434866, "grad_norm": 1.1461632251739502, "learning_rate": 1.8287527976430187e-05, "loss": 0.17825534343719482, "step": 42960 }, { "epoch": 0.21343492643771794, "grad_norm": 1.5619075298309326, "learning_rate": 1.828662760908539e-05, "loss": 0.16981351375579834, "step": 42970 }, { "epoch": 0.2134845971210872, "grad_norm": 1.233657956123352, "learning_rate": 1.8285727027285998e-05, "loss": 0.1891181468963623, "step": 42980 }, { "epoch": 0.21353426780445645, "grad_norm": 1.0298128128051758, "learning_rate": 1.8284826231055324e-05, "loss": 0.1716867208480835, "step": 42990 }, { "epoch": 0.21358393848782573, "grad_norm": 2.789463758468628, "learning_rate": 1.8283925220416677e-05, "loss": 0.17117955684661865, "step": 43000 }, { "epoch": 0.21363360917119498, "grad_norm": 2.5432310104370117, "learning_rate": 1.8283023995393377e-05, "loss": 0.1809387445449829, "step": 43010 }, { "epoch": 0.21368327985456423, "grad_norm": 0.8946362137794495, "learning_rate": 1.8282122556008746e-05, "loss": 0.16888864040374757, "step": 43020 }, { "epoch": 0.2137329505379335, "grad_norm": 0.9271112680435181, "learning_rate": 1.828122090228611e-05, "loss": 0.16931533813476562, "step": 43030 }, { "epoch": 0.21378262122130276, "grad_norm": 2.3065524101257324, "learning_rate": 1.8280319034248812e-05, "loss": 0.18137267827987671, "step": 43040 }, { "epoch": 0.213832291904672, "grad_norm": 0.8488814234733582, "learning_rate": 1.8279416951920187e-05, "loss": 0.17052721977233887, "step": 43050 }, { "epoch": 0.2138819625880413, "grad_norm": 1.0799264907836914, "learning_rate": 1.8278514655323577e-05, "loss": 0.18110051155090331, "step": 43060 }, { "epoch": 0.21393163327141054, "grad_norm": 1.2044508457183838, "learning_rate": 1.827761214448234e-05, "loss": 0.17091550827026367, "step": 43070 }, { "epoch": 0.21398130395477982, "grad_norm": 0.7969244718551636, "learning_rate": 1.827670941941982e-05, "loss": 0.18929389715194703, "step": 43080 }, { "epoch": 0.21403097463814907, "grad_norm": 1.2953178882598877, "learning_rate": 1.8275806480159398e-05, "loss": 0.18066785335540772, "step": 43090 }, { "epoch": 0.21408064532151833, "grad_norm": 1.06233549118042, "learning_rate": 1.8274903326724428e-05, "loss": 0.18058571815490723, "step": 43100 }, { "epoch": 0.2141303160048876, "grad_norm": 1.0027652978897095, "learning_rate": 1.827399995913829e-05, "loss": 0.17461695671081542, "step": 43110 }, { "epoch": 0.21417998668825686, "grad_norm": 2.943328857421875, "learning_rate": 1.8273096377424355e-05, "loss": 0.16957414150238037, "step": 43120 }, { "epoch": 0.2142296573716261, "grad_norm": 1.9914767742156982, "learning_rate": 1.8272192581606018e-05, "loss": 0.19079720973968506, "step": 43130 }, { "epoch": 0.2142793280549954, "grad_norm": 1.2083349227905273, "learning_rate": 1.827128857170666e-05, "loss": 0.19270228147506713, "step": 43140 }, { "epoch": 0.21432899873836464, "grad_norm": 1.9578027725219727, "learning_rate": 1.8270384347749678e-05, "loss": 0.18121469020843506, "step": 43150 }, { "epoch": 0.2143786694217339, "grad_norm": 0.6374087333679199, "learning_rate": 1.8269479909758476e-05, "loss": 0.17471325397491455, "step": 43160 }, { "epoch": 0.21442834010510317, "grad_norm": 0.8745777010917664, "learning_rate": 1.8268575257756457e-05, "loss": 0.16416847705841064, "step": 43170 }, { "epoch": 0.21447801078847242, "grad_norm": 1.6043609380722046, "learning_rate": 1.826767039176704e-05, "loss": 0.18154165744781495, "step": 43180 }, { "epoch": 0.2145276814718417, "grad_norm": 2.0652449131011963, "learning_rate": 1.8266765311813635e-05, "loss": 0.1887660503387451, "step": 43190 }, { "epoch": 0.21457735215521095, "grad_norm": 1.9586713314056396, "learning_rate": 1.8265860017919664e-05, "loss": 0.18228040933609008, "step": 43200 }, { "epoch": 0.2146270228385802, "grad_norm": 2.779256820678711, "learning_rate": 1.8264954510108563e-05, "loss": 0.18879858255386353, "step": 43210 }, { "epoch": 0.21467669352194949, "grad_norm": 1.151615858078003, "learning_rate": 1.8264048788403764e-05, "loss": 0.17197608947753906, "step": 43220 }, { "epoch": 0.21472636420531874, "grad_norm": 2.6020216941833496, "learning_rate": 1.8263142852828702e-05, "loss": 0.1773401141166687, "step": 43230 }, { "epoch": 0.214776034888688, "grad_norm": 1.4413213729858398, "learning_rate": 1.8262236703406827e-05, "loss": 0.16110455989837646, "step": 43240 }, { "epoch": 0.21482570557205727, "grad_norm": 1.4634400606155396, "learning_rate": 1.8261330340161587e-05, "loss": 0.1794452428817749, "step": 43250 }, { "epoch": 0.21487537625542652, "grad_norm": 1.923323154449463, "learning_rate": 1.826042376311644e-05, "loss": 0.17760472297668456, "step": 43260 }, { "epoch": 0.21492504693879577, "grad_norm": 1.054344892501831, "learning_rate": 1.825951697229485e-05, "loss": 0.18006389141082763, "step": 43270 }, { "epoch": 0.21497471762216505, "grad_norm": 1.1933692693710327, "learning_rate": 1.8258609967720278e-05, "loss": 0.17436208724975585, "step": 43280 }, { "epoch": 0.2150243883055343, "grad_norm": 2.0339303016662598, "learning_rate": 1.8257702749416203e-05, "loss": 0.16765742301940917, "step": 43290 }, { "epoch": 0.21507405898890358, "grad_norm": 1.3779574632644653, "learning_rate": 1.82567953174061e-05, "loss": 0.1658569574356079, "step": 43300 }, { "epoch": 0.21512372967227283, "grad_norm": 1.4225823879241943, "learning_rate": 1.8255887671713455e-05, "loss": 0.18554461002349854, "step": 43310 }, { "epoch": 0.21517340035564209, "grad_norm": 0.8654471039772034, "learning_rate": 1.8254979812361754e-05, "loss": 0.1765562415122986, "step": 43320 }, { "epoch": 0.21522307103901137, "grad_norm": 3.5807178020477295, "learning_rate": 1.8254071739374496e-05, "loss": 0.17889147996902466, "step": 43330 }, { "epoch": 0.21527274172238062, "grad_norm": 1.8315505981445312, "learning_rate": 1.8253163452775183e-05, "loss": 0.18091931343078613, "step": 43340 }, { "epoch": 0.21532241240574987, "grad_norm": 0.9757227897644043, "learning_rate": 1.8252254952587315e-05, "loss": 0.17462207078933717, "step": 43350 }, { "epoch": 0.21537208308911915, "grad_norm": 1.3663320541381836, "learning_rate": 1.825134623883441e-05, "loss": 0.17943294048309327, "step": 43360 }, { "epoch": 0.2154217537724884, "grad_norm": 1.2781144380569458, "learning_rate": 1.825043731153998e-05, "loss": 0.1837010622024536, "step": 43370 }, { "epoch": 0.21547142445585765, "grad_norm": 1.3037099838256836, "learning_rate": 1.824952817072755e-05, "loss": 0.17119531631469725, "step": 43380 }, { "epoch": 0.21552109513922693, "grad_norm": 0.7802542448043823, "learning_rate": 1.8248618816420647e-05, "loss": 0.17928457260131836, "step": 43390 }, { "epoch": 0.21557076582259618, "grad_norm": 1.713441252708435, "learning_rate": 1.8247709248642804e-05, "loss": 0.16206789016723633, "step": 43400 }, { "epoch": 0.21562043650596546, "grad_norm": 1.7761372327804565, "learning_rate": 1.8246799467417565e-05, "loss": 0.17965881824493407, "step": 43410 }, { "epoch": 0.21567010718933471, "grad_norm": 0.938966691493988, "learning_rate": 1.8245889472768472e-05, "loss": 0.19017041921615602, "step": 43420 }, { "epoch": 0.21571977787270397, "grad_norm": 2.9778759479522705, "learning_rate": 1.8244979264719075e-05, "loss": 0.16775970458984374, "step": 43430 }, { "epoch": 0.21576944855607325, "grad_norm": 1.011293649673462, "learning_rate": 1.8244068843292927e-05, "loss": 0.1811526298522949, "step": 43440 }, { "epoch": 0.2158191192394425, "grad_norm": 0.5841280817985535, "learning_rate": 1.8243158208513593e-05, "loss": 0.17301815748214722, "step": 43450 }, { "epoch": 0.21586878992281175, "grad_norm": 2.3164570331573486, "learning_rate": 1.824224736040464e-05, "loss": 0.18696731328964233, "step": 43460 }, { "epoch": 0.21591846060618103, "grad_norm": 1.1484216451644897, "learning_rate": 1.824133629898964e-05, "loss": 0.16657124757766723, "step": 43470 }, { "epoch": 0.21596813128955028, "grad_norm": 1.374761700630188, "learning_rate": 1.824042502429217e-05, "loss": 0.15428059101104735, "step": 43480 }, { "epoch": 0.21601780197291953, "grad_norm": 1.1972800493240356, "learning_rate": 1.8239513536335813e-05, "loss": 0.18364338874816893, "step": 43490 }, { "epoch": 0.2160674726562888, "grad_norm": 1.3958326578140259, "learning_rate": 1.823860183514416e-05, "loss": 0.19651172161102295, "step": 43500 }, { "epoch": 0.21611714333965806, "grad_norm": 1.304083228111267, "learning_rate": 1.8237689920740802e-05, "loss": 0.1786391019821167, "step": 43510 }, { "epoch": 0.21616681402302731, "grad_norm": 1.8915369510650635, "learning_rate": 1.8236777793149344e-05, "loss": 0.1755110502243042, "step": 43520 }, { "epoch": 0.2162164847063966, "grad_norm": 1.1782965660095215, "learning_rate": 1.8235865452393386e-05, "loss": 0.18108081817626953, "step": 43530 }, { "epoch": 0.21626615538976585, "grad_norm": 1.6634132862091064, "learning_rate": 1.8234952898496544e-05, "loss": 0.1894398331642151, "step": 43540 }, { "epoch": 0.21631582607313513, "grad_norm": 1.00772225856781, "learning_rate": 1.823404013148243e-05, "loss": 0.17815362215042113, "step": 43550 }, { "epoch": 0.21636549675650438, "grad_norm": 1.0397287607192993, "learning_rate": 1.8233127151374668e-05, "loss": 0.17071572542190552, "step": 43560 }, { "epoch": 0.21641516743987363, "grad_norm": 0.7959228754043579, "learning_rate": 1.8232213958196884e-05, "loss": 0.17765026092529296, "step": 43570 }, { "epoch": 0.2164648381232429, "grad_norm": 2.430192470550537, "learning_rate": 1.8231300551972716e-05, "loss": 0.17850204706192016, "step": 43580 }, { "epoch": 0.21651450880661216, "grad_norm": 1.458977460861206, "learning_rate": 1.82303869327258e-05, "loss": 0.169344162940979, "step": 43590 }, { "epoch": 0.2165641794899814, "grad_norm": 1.4081441164016724, "learning_rate": 1.8229473100479776e-05, "loss": 0.19215576648712157, "step": 43600 }, { "epoch": 0.2166138501733507, "grad_norm": 1.06410551071167, "learning_rate": 1.82285590552583e-05, "loss": 0.17878175973892213, "step": 43610 }, { "epoch": 0.21666352085671994, "grad_norm": 1.0327658653259277, "learning_rate": 1.8227644797085026e-05, "loss": 0.16996269226074218, "step": 43620 }, { "epoch": 0.2167131915400892, "grad_norm": 0.5765419006347656, "learning_rate": 1.822673032598361e-05, "loss": 0.17935292720794677, "step": 43630 }, { "epoch": 0.21676286222345847, "grad_norm": 0.9817699790000916, "learning_rate": 1.822581564197772e-05, "loss": 0.1726663589477539, "step": 43640 }, { "epoch": 0.21681253290682773, "grad_norm": 1.0069749355316162, "learning_rate": 1.822490074509103e-05, "loss": 0.17612338066101074, "step": 43650 }, { "epoch": 0.216862203590197, "grad_norm": 1.4192417860031128, "learning_rate": 1.8223985635347216e-05, "loss": 0.17826220989227295, "step": 43660 }, { "epoch": 0.21691187427356626, "grad_norm": 1.2334787845611572, "learning_rate": 1.8223070312769958e-05, "loss": 0.1887392520904541, "step": 43670 }, { "epoch": 0.2169615449569355, "grad_norm": 1.0266331434249878, "learning_rate": 1.822215477738295e-05, "loss": 0.19262717962265014, "step": 43680 }, { "epoch": 0.2170112156403048, "grad_norm": 0.943475067615509, "learning_rate": 1.822123902920988e-05, "loss": 0.16480761766433716, "step": 43690 }, { "epoch": 0.21706088632367404, "grad_norm": 2.807643175125122, "learning_rate": 1.822032306827445e-05, "loss": 0.17883284091949464, "step": 43700 }, { "epoch": 0.2171105570070433, "grad_norm": 1.3396408557891846, "learning_rate": 1.821940689460037e-05, "loss": 0.18310048580169677, "step": 43710 }, { "epoch": 0.21716022769041257, "grad_norm": 1.0369246006011963, "learning_rate": 1.8218490508211336e-05, "loss": 0.18877246379852294, "step": 43720 }, { "epoch": 0.21720989837378182, "grad_norm": 2.103079080581665, "learning_rate": 1.8217573909131077e-05, "loss": 0.18081293106079102, "step": 43730 }, { "epoch": 0.21725956905715108, "grad_norm": 0.9138498902320862, "learning_rate": 1.8216657097383307e-05, "loss": 0.17434377670288087, "step": 43740 }, { "epoch": 0.21730923974052035, "grad_norm": 0.9680373668670654, "learning_rate": 1.8215740072991758e-05, "loss": 0.18327414989471436, "step": 43750 }, { "epoch": 0.2173589104238896, "grad_norm": 0.9418266415596008, "learning_rate": 1.8214822835980154e-05, "loss": 0.17862887382507325, "step": 43760 }, { "epoch": 0.21740858110725889, "grad_norm": 2.25211763381958, "learning_rate": 1.8213905386372245e-05, "loss": 0.1948711633682251, "step": 43770 }, { "epoch": 0.21745825179062814, "grad_norm": 0.7665742039680481, "learning_rate": 1.821298772419176e-05, "loss": 0.1821230173110962, "step": 43780 }, { "epoch": 0.2175079224739974, "grad_norm": 0.9710050821304321, "learning_rate": 1.821206984946246e-05, "loss": 0.18256866931915283, "step": 43790 }, { "epoch": 0.21755759315736667, "grad_norm": 0.7096509337425232, "learning_rate": 1.8211151762208095e-05, "loss": 0.17748945951461792, "step": 43800 }, { "epoch": 0.21760726384073592, "grad_norm": 1.736314058303833, "learning_rate": 1.821023346245242e-05, "loss": 0.182502818107605, "step": 43810 }, { "epoch": 0.21765693452410517, "grad_norm": 2.196918487548828, "learning_rate": 1.820931495021921e-05, "loss": 0.17773680686950682, "step": 43820 }, { "epoch": 0.21770660520747445, "grad_norm": 1.4944807291030884, "learning_rate": 1.8208396225532225e-05, "loss": 0.17957206964492797, "step": 43830 }, { "epoch": 0.2177562758908437, "grad_norm": 1.1320643424987793, "learning_rate": 1.8207477288415246e-05, "loss": 0.17182257175445556, "step": 43840 }, { "epoch": 0.21780594657421296, "grad_norm": 1.9068090915679932, "learning_rate": 1.8206558138892053e-05, "loss": 0.1778636693954468, "step": 43850 }, { "epoch": 0.21785561725758223, "grad_norm": 0.9734763503074646, "learning_rate": 1.820563877698644e-05, "loss": 0.17925797700881957, "step": 43860 }, { "epoch": 0.2179052879409515, "grad_norm": 1.4550193548202515, "learning_rate": 1.820471920272219e-05, "loss": 0.181421160697937, "step": 43870 }, { "epoch": 0.21795495862432077, "grad_norm": 2.670267343521118, "learning_rate": 1.8203799416123113e-05, "loss": 0.16638263463973998, "step": 43880 }, { "epoch": 0.21800462930769002, "grad_norm": 0.9433534145355225, "learning_rate": 1.8202879417213e-05, "loss": 0.1839904546737671, "step": 43890 }, { "epoch": 0.21805429999105927, "grad_norm": 1.1449739933013916, "learning_rate": 1.8201959206015666e-05, "loss": 0.18752100467681884, "step": 43900 }, { "epoch": 0.21810397067442855, "grad_norm": 2.14884614944458, "learning_rate": 1.820103878255493e-05, "loss": 0.16502187252044678, "step": 43910 }, { "epoch": 0.2181536413577978, "grad_norm": 0.622697651386261, "learning_rate": 1.8200118146854603e-05, "loss": 0.17343772649765016, "step": 43920 }, { "epoch": 0.21820331204116705, "grad_norm": 1.2716838121414185, "learning_rate": 1.819919729893852e-05, "loss": 0.17390744686126708, "step": 43930 }, { "epoch": 0.21825298272453633, "grad_norm": 0.9247456789016724, "learning_rate": 1.8198276238830503e-05, "loss": 0.16695454120635986, "step": 43940 }, { "epoch": 0.21830265340790558, "grad_norm": 0.8119117617607117, "learning_rate": 1.8197354966554396e-05, "loss": 0.18367879390716552, "step": 43950 }, { "epoch": 0.21835232409127484, "grad_norm": 1.948621392250061, "learning_rate": 1.8196433482134038e-05, "loss": 0.17428238391876222, "step": 43960 }, { "epoch": 0.21840199477464411, "grad_norm": 0.9675350785255432, "learning_rate": 1.8195511785593276e-05, "loss": 0.1789056181907654, "step": 43970 }, { "epoch": 0.21845166545801337, "grad_norm": 1.595225214958191, "learning_rate": 1.8194589876955965e-05, "loss": 0.16643953323364258, "step": 43980 }, { "epoch": 0.21850133614138265, "grad_norm": 1.5577895641326904, "learning_rate": 1.8193667756245965e-05, "loss": 0.166787052154541, "step": 43990 }, { "epoch": 0.2185510068247519, "grad_norm": 0.9600861072540283, "learning_rate": 1.819274542348714e-05, "loss": 0.19413583278656005, "step": 44000 }, { "epoch": 0.21860067750812115, "grad_norm": 0.7688597440719604, "learning_rate": 1.8191822878703348e-05, "loss": 0.18869454860687257, "step": 44010 }, { "epoch": 0.21865034819149043, "grad_norm": 1.4093042612075806, "learning_rate": 1.8190900121918482e-05, "loss": 0.19448997974395751, "step": 44020 }, { "epoch": 0.21870001887485968, "grad_norm": 1.2755041122436523, "learning_rate": 1.8189977153156412e-05, "loss": 0.17297227382659913, "step": 44030 }, { "epoch": 0.21874968955822893, "grad_norm": 1.1410799026489258, "learning_rate": 1.8189053972441027e-05, "loss": 0.169465970993042, "step": 44040 }, { "epoch": 0.2187993602415982, "grad_norm": 1.4167438745498657, "learning_rate": 1.8188130579796216e-05, "loss": 0.17115072011947632, "step": 44050 }, { "epoch": 0.21884903092496746, "grad_norm": 2.2290172576904297, "learning_rate": 1.818720697524588e-05, "loss": 0.19955517053604127, "step": 44060 }, { "epoch": 0.21889870160833672, "grad_norm": 3.089829206466675, "learning_rate": 1.8186283158813916e-05, "loss": 0.191914963722229, "step": 44070 }, { "epoch": 0.218948372291706, "grad_norm": 1.4261374473571777, "learning_rate": 1.8185359130524238e-05, "loss": 0.17678241729736327, "step": 44080 }, { "epoch": 0.21899804297507525, "grad_norm": 2.17974853515625, "learning_rate": 1.818443489040076e-05, "loss": 0.17704898118972778, "step": 44090 }, { "epoch": 0.21904771365844453, "grad_norm": 1.8880126476287842, "learning_rate": 1.818351043846739e-05, "loss": 0.1839378833770752, "step": 44100 }, { "epoch": 0.21909738434181378, "grad_norm": 1.2342456579208374, "learning_rate": 1.8182585774748067e-05, "loss": 0.17164769172668456, "step": 44110 }, { "epoch": 0.21914705502518303, "grad_norm": 1.2939462661743164, "learning_rate": 1.8181660899266714e-05, "loss": 0.18066904544830323, "step": 44120 }, { "epoch": 0.2191967257085523, "grad_norm": 0.9684925079345703, "learning_rate": 1.8180735812047264e-05, "loss": 0.168770432472229, "step": 44130 }, { "epoch": 0.21924639639192156, "grad_norm": 0.920224666595459, "learning_rate": 1.817981051311366e-05, "loss": 0.18066103458404542, "step": 44140 }, { "epoch": 0.2192960670752908, "grad_norm": 2.0487403869628906, "learning_rate": 1.817888500248985e-05, "loss": 0.1639464259147644, "step": 44150 }, { "epoch": 0.2193457377586601, "grad_norm": 1.5728998184204102, "learning_rate": 1.8177959280199785e-05, "loss": 0.18384177684783937, "step": 44160 }, { "epoch": 0.21939540844202934, "grad_norm": 1.2978999614715576, "learning_rate": 1.8177033346267424e-05, "loss": 0.16757732629776, "step": 44170 }, { "epoch": 0.2194450791253986, "grad_norm": 1.7701785564422607, "learning_rate": 1.8176107200716725e-05, "loss": 0.18139419555664063, "step": 44180 }, { "epoch": 0.21949474980876787, "grad_norm": 1.41416597366333, "learning_rate": 1.817518084357166e-05, "loss": 0.1804504871368408, "step": 44190 }, { "epoch": 0.21954442049213713, "grad_norm": 0.6948330402374268, "learning_rate": 1.81742542748562e-05, "loss": 0.17316304445266723, "step": 44200 }, { "epoch": 0.2195940911755064, "grad_norm": 1.6912720203399658, "learning_rate": 1.8173327494594327e-05, "loss": 0.17654705047607422, "step": 44210 }, { "epoch": 0.21964376185887566, "grad_norm": 0.9233183264732361, "learning_rate": 1.817240050281003e-05, "loss": 0.15922682285308837, "step": 44220 }, { "epoch": 0.2196934325422449, "grad_norm": 1.05474054813385, "learning_rate": 1.8171473299527286e-05, "loss": 0.17069995403289795, "step": 44230 }, { "epoch": 0.2197431032256142, "grad_norm": 2.309657096862793, "learning_rate": 1.8170545884770103e-05, "loss": 0.17253199815750123, "step": 44240 }, { "epoch": 0.21979277390898344, "grad_norm": 0.8621155023574829, "learning_rate": 1.8169618258562474e-05, "loss": 0.18249763250350953, "step": 44250 }, { "epoch": 0.2198424445923527, "grad_norm": 2.03460955619812, "learning_rate": 1.8168690420928416e-05, "loss": 0.18300682306289673, "step": 44260 }, { "epoch": 0.21989211527572197, "grad_norm": 0.8966912627220154, "learning_rate": 1.816776237189193e-05, "loss": 0.17774864435195922, "step": 44270 }, { "epoch": 0.21994178595909122, "grad_norm": 1.5638116598129272, "learning_rate": 1.8166834111477037e-05, "loss": 0.19046064615249633, "step": 44280 }, { "epoch": 0.21999145664246048, "grad_norm": 0.6977726817131042, "learning_rate": 1.8165905639707764e-05, "loss": 0.17111752033233643, "step": 44290 }, { "epoch": 0.22004112732582976, "grad_norm": 0.894006609916687, "learning_rate": 1.8164976956608135e-05, "loss": 0.1658334493637085, "step": 44300 }, { "epoch": 0.220090798009199, "grad_norm": 1.1173653602600098, "learning_rate": 1.8164048062202185e-05, "loss": 0.1729738712310791, "step": 44310 }, { "epoch": 0.22014046869256826, "grad_norm": 1.634909749031067, "learning_rate": 1.8163118956513956e-05, "loss": 0.18435213565826417, "step": 44320 }, { "epoch": 0.22019013937593754, "grad_norm": 1.2008987665176392, "learning_rate": 1.816218963956749e-05, "loss": 0.16697767972946168, "step": 44330 }, { "epoch": 0.2202398100593068, "grad_norm": 1.0166964530944824, "learning_rate": 1.8161260111386837e-05, "loss": 0.17696447372436525, "step": 44340 }, { "epoch": 0.22028948074267607, "grad_norm": 0.8098322749137878, "learning_rate": 1.8160330371996052e-05, "loss": 0.1716388702392578, "step": 44350 }, { "epoch": 0.22033915142604532, "grad_norm": 1.261542797088623, "learning_rate": 1.81594004214192e-05, "loss": 0.17905274629592896, "step": 44360 }, { "epoch": 0.22038882210941457, "grad_norm": 1.736459493637085, "learning_rate": 1.8158470259680342e-05, "loss": 0.17619850635528564, "step": 44370 }, { "epoch": 0.22043849279278385, "grad_norm": 2.0000715255737305, "learning_rate": 1.815753988680356e-05, "loss": 0.17116761207580566, "step": 44380 }, { "epoch": 0.2204881634761531, "grad_norm": 1.5750099420547485, "learning_rate": 1.8156609302812923e-05, "loss": 0.18814831972122192, "step": 44390 }, { "epoch": 0.22053783415952236, "grad_norm": 2.012059450149536, "learning_rate": 1.8155678507732517e-05, "loss": 0.1858989715576172, "step": 44400 }, { "epoch": 0.22058750484289164, "grad_norm": 1.4509984254837036, "learning_rate": 1.8154747501586433e-05, "loss": 0.17613861560821534, "step": 44410 }, { "epoch": 0.2206371755262609, "grad_norm": 0.7973770499229431, "learning_rate": 1.815381628439876e-05, "loss": 0.1871653437614441, "step": 44420 }, { "epoch": 0.22068684620963014, "grad_norm": 0.8410167694091797, "learning_rate": 1.81528848561936e-05, "loss": 0.16428093910217284, "step": 44430 }, { "epoch": 0.22073651689299942, "grad_norm": 1.7064366340637207, "learning_rate": 1.815195321699506e-05, "loss": 0.1864755392074585, "step": 44440 }, { "epoch": 0.22078618757636867, "grad_norm": 1.3158960342407227, "learning_rate": 1.8151021366827246e-05, "loss": 0.17793110609054566, "step": 44450 }, { "epoch": 0.22083585825973795, "grad_norm": 0.719445526599884, "learning_rate": 1.8150089305714276e-05, "loss": 0.1753467798233032, "step": 44460 }, { "epoch": 0.2208855289431072, "grad_norm": 1.3254640102386475, "learning_rate": 1.8149157033680274e-05, "loss": 0.17487049102783203, "step": 44470 }, { "epoch": 0.22093519962647645, "grad_norm": 0.7361233234405518, "learning_rate": 1.8148224550749366e-05, "loss": 0.1824814796447754, "step": 44480 }, { "epoch": 0.22098487030984573, "grad_norm": 1.192740797996521, "learning_rate": 1.814729185694568e-05, "loss": 0.16304335594177247, "step": 44490 }, { "epoch": 0.22103454099321498, "grad_norm": 1.9348348379135132, "learning_rate": 1.8146358952293357e-05, "loss": 0.16082563400268554, "step": 44500 }, { "epoch": 0.22108421167658424, "grad_norm": 1.3685319423675537, "learning_rate": 1.814542583681654e-05, "loss": 0.1712012529373169, "step": 44510 }, { "epoch": 0.22113388235995352, "grad_norm": 0.6641894578933716, "learning_rate": 1.8144492510539375e-05, "loss": 0.17359228134155275, "step": 44520 }, { "epoch": 0.22118355304332277, "grad_norm": 0.7352303862571716, "learning_rate": 1.814355897348602e-05, "loss": 0.16812061071395873, "step": 44530 }, { "epoch": 0.22123322372669202, "grad_norm": 1.6065576076507568, "learning_rate": 1.8142625225680636e-05, "loss": 0.1724057674407959, "step": 44540 }, { "epoch": 0.2212828944100613, "grad_norm": 0.6675021648406982, "learning_rate": 1.8141691267147386e-05, "loss": 0.1934834122657776, "step": 44550 }, { "epoch": 0.22133256509343055, "grad_norm": 1.1308023929595947, "learning_rate": 1.814075709791044e-05, "loss": 0.17113213539123534, "step": 44560 }, { "epoch": 0.22138223577679983, "grad_norm": 2.438974618911743, "learning_rate": 1.8139822717993968e-05, "loss": 0.1805511474609375, "step": 44570 }, { "epoch": 0.22143190646016908, "grad_norm": 2.0590553283691406, "learning_rate": 1.813888812742216e-05, "loss": 0.1947835326194763, "step": 44580 }, { "epoch": 0.22148157714353833, "grad_norm": 1.362770915031433, "learning_rate": 1.8137953326219198e-05, "loss": 0.18558886051177978, "step": 44590 }, { "epoch": 0.2215312478269076, "grad_norm": 0.7666119337081909, "learning_rate": 1.8137018314409277e-05, "loss": 0.1792216420173645, "step": 44600 }, { "epoch": 0.22158091851027686, "grad_norm": 1.3433159589767456, "learning_rate": 1.8136083092016595e-05, "loss": 0.17612154483795167, "step": 44610 }, { "epoch": 0.22163058919364612, "grad_norm": 2.198615550994873, "learning_rate": 1.8135147659065352e-05, "loss": 0.17437257766723632, "step": 44620 }, { "epoch": 0.2216802598770154, "grad_norm": 1.3168890476226807, "learning_rate": 1.813421201557976e-05, "loss": 0.17699352502822877, "step": 44630 }, { "epoch": 0.22172993056038465, "grad_norm": 1.6016789674758911, "learning_rate": 1.8133276161584033e-05, "loss": 0.1767566680908203, "step": 44640 }, { "epoch": 0.2217796012437539, "grad_norm": 2.4605560302734375, "learning_rate": 1.813234009710239e-05, "loss": 0.17146050930023193, "step": 44650 }, { "epoch": 0.22182927192712318, "grad_norm": 0.9206135869026184, "learning_rate": 1.813140382215905e-05, "loss": 0.16822028160095215, "step": 44660 }, { "epoch": 0.22187894261049243, "grad_norm": 1.116752028465271, "learning_rate": 1.8130467336778252e-05, "loss": 0.16463127136230468, "step": 44670 }, { "epoch": 0.2219286132938617, "grad_norm": 1.7944567203521729, "learning_rate": 1.8129530640984226e-05, "loss": 0.17634344100952148, "step": 44680 }, { "epoch": 0.22197828397723096, "grad_norm": 0.7351122498512268, "learning_rate": 1.812859373480122e-05, "loss": 0.1668018579483032, "step": 44690 }, { "epoch": 0.2220279546606002, "grad_norm": 1.567082166671753, "learning_rate": 1.8127656618253474e-05, "loss": 0.17709016799926758, "step": 44700 }, { "epoch": 0.2220776253439695, "grad_norm": 1.0952662229537964, "learning_rate": 1.812671929136524e-05, "loss": 0.18004899024963378, "step": 44710 }, { "epoch": 0.22212729602733874, "grad_norm": 1.7471438646316528, "learning_rate": 1.8125781754160782e-05, "loss": 0.17856736183166505, "step": 44720 }, { "epoch": 0.222176966710708, "grad_norm": 2.4537975788116455, "learning_rate": 1.812484400666436e-05, "loss": 0.17340390682220458, "step": 44730 }, { "epoch": 0.22222663739407728, "grad_norm": 0.9730885624885559, "learning_rate": 1.8123906048900243e-05, "loss": 0.17046515941619872, "step": 44740 }, { "epoch": 0.22227630807744653, "grad_norm": 1.216908574104309, "learning_rate": 1.81229678808927e-05, "loss": 0.16642658710479735, "step": 44750 }, { "epoch": 0.22232597876081578, "grad_norm": 1.7594473361968994, "learning_rate": 1.8122029502666017e-05, "loss": 0.168868088722229, "step": 44760 }, { "epoch": 0.22237564944418506, "grad_norm": 2.241130828857422, "learning_rate": 1.8121090914244474e-05, "loss": 0.17376528978347777, "step": 44770 }, { "epoch": 0.2224253201275543, "grad_norm": 1.2806328535079956, "learning_rate": 1.8120152115652363e-05, "loss": 0.169089412689209, "step": 44780 }, { "epoch": 0.2224749908109236, "grad_norm": 1.3916265964508057, "learning_rate": 1.8119213106913986e-05, "loss": 0.18034226894378663, "step": 44790 }, { "epoch": 0.22252466149429284, "grad_norm": 0.7626283764839172, "learning_rate": 1.8118273888053633e-05, "loss": 0.18538849353790282, "step": 44800 }, { "epoch": 0.2225743321776621, "grad_norm": 1.4045157432556152, "learning_rate": 1.8117334459095618e-05, "loss": 0.19212342500686647, "step": 44810 }, { "epoch": 0.22262400286103137, "grad_norm": 2.433894157409668, "learning_rate": 1.8116394820064247e-05, "loss": 0.17895543575286865, "step": 44820 }, { "epoch": 0.22267367354440062, "grad_norm": 0.7566271424293518, "learning_rate": 1.8115454970983846e-05, "loss": 0.1817193627357483, "step": 44830 }, { "epoch": 0.22272334422776988, "grad_norm": 1.3436498641967773, "learning_rate": 1.8114514911878733e-05, "loss": 0.1692429542541504, "step": 44840 }, { "epoch": 0.22277301491113916, "grad_norm": 1.2765183448791504, "learning_rate": 1.8113574642773237e-05, "loss": 0.18126262426376344, "step": 44850 }, { "epoch": 0.2228226855945084, "grad_norm": 1.1579561233520508, "learning_rate": 1.811263416369169e-05, "loss": 0.18805313110351562, "step": 44860 }, { "epoch": 0.22287235627787766, "grad_norm": 1.3858404159545898, "learning_rate": 1.8111693474658435e-05, "loss": 0.1760702610015869, "step": 44870 }, { "epoch": 0.22292202696124694, "grad_norm": 0.949099063873291, "learning_rate": 1.811075257569781e-05, "loss": 0.15897047519683838, "step": 44880 }, { "epoch": 0.2229716976446162, "grad_norm": 0.8298464417457581, "learning_rate": 1.8109811466834173e-05, "loss": 0.19333606958389282, "step": 44890 }, { "epoch": 0.22302136832798547, "grad_norm": 1.4753369092941284, "learning_rate": 1.8108870148091876e-05, "loss": 0.18511179685592652, "step": 44900 }, { "epoch": 0.22307103901135472, "grad_norm": 1.0470751523971558, "learning_rate": 1.810792861949528e-05, "loss": 0.18452048301696777, "step": 44910 }, { "epoch": 0.22312070969472397, "grad_norm": 1.4742681980133057, "learning_rate": 1.810698688106875e-05, "loss": 0.1813073992729187, "step": 44920 }, { "epoch": 0.22317038037809325, "grad_norm": 1.167404294013977, "learning_rate": 1.8106044932836663e-05, "loss": 0.1721484661102295, "step": 44930 }, { "epoch": 0.2232200510614625, "grad_norm": 0.7013806700706482, "learning_rate": 1.8105102774823388e-05, "loss": 0.17658988237380982, "step": 44940 }, { "epoch": 0.22326972174483176, "grad_norm": 1.482055902481079, "learning_rate": 1.8104160407053313e-05, "loss": 0.18630211353302, "step": 44950 }, { "epoch": 0.22331939242820104, "grad_norm": 1.1019765138626099, "learning_rate": 1.8103217829550824e-05, "loss": 0.17909910678863525, "step": 44960 }, { "epoch": 0.2233690631115703, "grad_norm": 1.5029178857803345, "learning_rate": 1.8102275042340317e-05, "loss": 0.18439575433731079, "step": 44970 }, { "epoch": 0.22341873379493954, "grad_norm": 1.3736859560012817, "learning_rate": 1.810133204544619e-05, "loss": 0.16930261850357056, "step": 44980 }, { "epoch": 0.22346840447830882, "grad_norm": 0.8626578450202942, "learning_rate": 1.810038883889285e-05, "loss": 0.16276776790618896, "step": 44990 }, { "epoch": 0.22351807516167807, "grad_norm": 2.047229528427124, "learning_rate": 1.8099445422704702e-05, "loss": 0.1841593027114868, "step": 45000 }, { "epoch": 0.22356774584504735, "grad_norm": 2.398369789123535, "learning_rate": 1.8098501796906163e-05, "loss": 0.18667864799499512, "step": 45010 }, { "epoch": 0.2236174165284166, "grad_norm": 0.7501425743103027, "learning_rate": 1.809755796152165e-05, "loss": 0.17024630308151245, "step": 45020 }, { "epoch": 0.22366708721178585, "grad_norm": 1.0599297285079956, "learning_rate": 1.8096613916575597e-05, "loss": 0.16623930931091307, "step": 45030 }, { "epoch": 0.22371675789515513, "grad_norm": 1.0015034675598145, "learning_rate": 1.8095669662092433e-05, "loss": 0.18228411674499512, "step": 45040 }, { "epoch": 0.22376642857852438, "grad_norm": 3.2383389472961426, "learning_rate": 1.809472519809659e-05, "loss": 0.1745958685874939, "step": 45050 }, { "epoch": 0.22381609926189364, "grad_norm": 0.8821287155151367, "learning_rate": 1.8093780524612514e-05, "loss": 0.16824076175689698, "step": 45060 }, { "epoch": 0.22386576994526292, "grad_norm": 1.4910789728164673, "learning_rate": 1.809283564166465e-05, "loss": 0.1813760757446289, "step": 45070 }, { "epoch": 0.22391544062863217, "grad_norm": 1.239099383354187, "learning_rate": 1.809189054927746e-05, "loss": 0.17277806997299194, "step": 45080 }, { "epoch": 0.22396511131200142, "grad_norm": 1.2771220207214355, "learning_rate": 1.809094524747539e-05, "loss": 0.16072959899902345, "step": 45090 }, { "epoch": 0.2240147819953707, "grad_norm": 0.7154965400695801, "learning_rate": 1.8089999736282914e-05, "loss": 0.19445786476135254, "step": 45100 }, { "epoch": 0.22406445267873995, "grad_norm": 1.6158157587051392, "learning_rate": 1.8089054015724493e-05, "loss": 0.16096653938293456, "step": 45110 }, { "epoch": 0.2241141233621092, "grad_norm": 1.0680153369903564, "learning_rate": 1.8088108085824613e-05, "loss": 0.18729413747787477, "step": 45120 }, { "epoch": 0.22416379404547848, "grad_norm": 1.916952133178711, "learning_rate": 1.8087161946607744e-05, "loss": 0.17622295618057252, "step": 45130 }, { "epoch": 0.22421346472884773, "grad_norm": 0.9404963850975037, "learning_rate": 1.8086215598098378e-05, "loss": 0.16849994659423828, "step": 45140 }, { "epoch": 0.224263135412217, "grad_norm": 1.8257009983062744, "learning_rate": 1.8085269040321e-05, "loss": 0.1711806058883667, "step": 45150 }, { "epoch": 0.22431280609558626, "grad_norm": 1.5171376466751099, "learning_rate": 1.8084322273300113e-05, "loss": 0.17958931922912597, "step": 45160 }, { "epoch": 0.22436247677895552, "grad_norm": 0.9212065935134888, "learning_rate": 1.8083375297060216e-05, "loss": 0.18413375616073607, "step": 45170 }, { "epoch": 0.2244121474623248, "grad_norm": 0.7616311311721802, "learning_rate": 1.8082428111625815e-05, "loss": 0.16426446437835693, "step": 45180 }, { "epoch": 0.22446181814569405, "grad_norm": 2.296527147293091, "learning_rate": 1.8081480717021425e-05, "loss": 0.17511502504348755, "step": 45190 }, { "epoch": 0.2245114888290633, "grad_norm": 0.8181788921356201, "learning_rate": 1.8080533113271562e-05, "loss": 0.18460795879364014, "step": 45200 }, { "epoch": 0.22456115951243258, "grad_norm": 1.7115161418914795, "learning_rate": 1.8079585300400754e-05, "loss": 0.1628572463989258, "step": 45210 }, { "epoch": 0.22461083019580183, "grad_norm": 1.465470314025879, "learning_rate": 1.8078637278433527e-05, "loss": 0.1654948115348816, "step": 45220 }, { "epoch": 0.22466050087917108, "grad_norm": 1.9642785787582397, "learning_rate": 1.8077689047394412e-05, "loss": 0.16830815076828004, "step": 45230 }, { "epoch": 0.22471017156254036, "grad_norm": 0.9669272899627686, "learning_rate": 1.8076740607307958e-05, "loss": 0.18467801809310913, "step": 45240 }, { "epoch": 0.2247598422459096, "grad_norm": 0.9785714149475098, "learning_rate": 1.80757919581987e-05, "loss": 0.16905684471130372, "step": 45250 }, { "epoch": 0.2248095129292789, "grad_norm": 1.4408597946166992, "learning_rate": 1.8074843100091193e-05, "loss": 0.16925381422042846, "step": 45260 }, { "epoch": 0.22485918361264814, "grad_norm": 0.8682587742805481, "learning_rate": 1.8073894033009996e-05, "loss": 0.19532623291015624, "step": 45270 }, { "epoch": 0.2249088542960174, "grad_norm": 0.5412379503250122, "learning_rate": 1.8072944756979665e-05, "loss": 0.17355914115905763, "step": 45280 }, { "epoch": 0.22495852497938668, "grad_norm": 0.8459766507148743, "learning_rate": 1.8071995272024772e-05, "loss": 0.1651397943496704, "step": 45290 }, { "epoch": 0.22500819566275593, "grad_norm": 2.092053174972534, "learning_rate": 1.8071045578169887e-05, "loss": 0.16587669849395753, "step": 45300 }, { "epoch": 0.22505786634612518, "grad_norm": 3.5760128498077393, "learning_rate": 1.8070095675439584e-05, "loss": 0.17420508861541747, "step": 45310 }, { "epoch": 0.22510753702949446, "grad_norm": 0.8261352777481079, "learning_rate": 1.8069145563858455e-05, "loss": 0.19103522300720216, "step": 45320 }, { "epoch": 0.2251572077128637, "grad_norm": 2.106276750564575, "learning_rate": 1.806819524345108e-05, "loss": 0.18737168312072755, "step": 45330 }, { "epoch": 0.22520687839623296, "grad_norm": 1.9557167291641235, "learning_rate": 1.8067244714242056e-05, "loss": 0.1613476276397705, "step": 45340 }, { "epoch": 0.22525654907960224, "grad_norm": 0.8095406293869019, "learning_rate": 1.8066293976255983e-05, "loss": 0.1870936393737793, "step": 45350 }, { "epoch": 0.2253062197629715, "grad_norm": 0.676324725151062, "learning_rate": 1.8065343029517467e-05, "loss": 0.17941100597381593, "step": 45360 }, { "epoch": 0.22535589044634077, "grad_norm": 1.2679095268249512, "learning_rate": 1.8064391874051118e-05, "loss": 0.18084897994995117, "step": 45370 }, { "epoch": 0.22540556112971002, "grad_norm": 2.1370651721954346, "learning_rate": 1.8063440509881545e-05, "loss": 0.184702730178833, "step": 45380 }, { "epoch": 0.22545523181307928, "grad_norm": 1.1483739614486694, "learning_rate": 1.8062488937033375e-05, "loss": 0.17142157554626464, "step": 45390 }, { "epoch": 0.22550490249644856, "grad_norm": 1.1464449167251587, "learning_rate": 1.8061537155531235e-05, "loss": 0.1796618938446045, "step": 45400 }, { "epoch": 0.2255545731798178, "grad_norm": 2.402362823486328, "learning_rate": 1.8060585165399754e-05, "loss": 0.18694820404052734, "step": 45410 }, { "epoch": 0.22560424386318706, "grad_norm": 0.751714289188385, "learning_rate": 1.805963296666357e-05, "loss": 0.1688701868057251, "step": 45420 }, { "epoch": 0.22565391454655634, "grad_norm": 0.9630565047264099, "learning_rate": 1.8058680559347323e-05, "loss": 0.18143672943115235, "step": 45430 }, { "epoch": 0.2257035852299256, "grad_norm": 3.10953688621521, "learning_rate": 1.8057727943475667e-05, "loss": 0.17981833219528198, "step": 45440 }, { "epoch": 0.22575325591329484, "grad_norm": 0.7985247373580933, "learning_rate": 1.805677511907325e-05, "loss": 0.17838503122329713, "step": 45450 }, { "epoch": 0.22580292659666412, "grad_norm": 1.694964051246643, "learning_rate": 1.8055822086164733e-05, "loss": 0.1762787103652954, "step": 45460 }, { "epoch": 0.22585259728003337, "grad_norm": 1.2999205589294434, "learning_rate": 1.805486884477478e-05, "loss": 0.17764732837677003, "step": 45470 }, { "epoch": 0.22590226796340265, "grad_norm": 1.6437369585037231, "learning_rate": 1.805391539492806e-05, "loss": 0.18290868997573853, "step": 45480 }, { "epoch": 0.2259519386467719, "grad_norm": 1.5602000951766968, "learning_rate": 1.8052961736649245e-05, "loss": 0.17595887184143066, "step": 45490 }, { "epoch": 0.22600160933014116, "grad_norm": 0.8428716659545898, "learning_rate": 1.8052007869963022e-05, "loss": 0.17666544914245605, "step": 45500 }, { "epoch": 0.22605128001351044, "grad_norm": 1.0803923606872559, "learning_rate": 1.805105379489407e-05, "loss": 0.18111205101013184, "step": 45510 }, { "epoch": 0.2261009506968797, "grad_norm": 1.2315876483917236, "learning_rate": 1.8050099511467082e-05, "loss": 0.18355653285980225, "step": 45520 }, { "epoch": 0.22615062138024894, "grad_norm": 0.7452492713928223, "learning_rate": 1.804914501970676e-05, "loss": 0.16280152797698974, "step": 45530 }, { "epoch": 0.22620029206361822, "grad_norm": 1.6531469821929932, "learning_rate": 1.80481903196378e-05, "loss": 0.17496466636657715, "step": 45540 }, { "epoch": 0.22624996274698747, "grad_norm": 1.4942070245742798, "learning_rate": 1.8047235411284906e-05, "loss": 0.1774375319480896, "step": 45550 }, { "epoch": 0.22629963343035672, "grad_norm": 1.676307201385498, "learning_rate": 1.80462802946728e-05, "loss": 0.18480594158172609, "step": 45560 }, { "epoch": 0.226349304113726, "grad_norm": 3.367332696914673, "learning_rate": 1.8045324969826196e-05, "loss": 0.17693808078765869, "step": 45570 }, { "epoch": 0.22639897479709525, "grad_norm": 1.4373410940170288, "learning_rate": 1.8044369436769817e-05, "loss": 0.17909047603607178, "step": 45580 }, { "epoch": 0.22644864548046453, "grad_norm": 0.6707233786582947, "learning_rate": 1.8043413695528385e-05, "loss": 0.16429851055145264, "step": 45590 }, { "epoch": 0.22649831616383378, "grad_norm": 1.1711570024490356, "learning_rate": 1.804245774612665e-05, "loss": 0.18154162168502808, "step": 45600 }, { "epoch": 0.22654798684720304, "grad_norm": 1.4541794061660767, "learning_rate": 1.804150158858933e-05, "loss": 0.17376872301101684, "step": 45610 }, { "epoch": 0.22659765753057232, "grad_norm": 1.5520386695861816, "learning_rate": 1.8040545222941194e-05, "loss": 0.19550702571868897, "step": 45620 }, { "epoch": 0.22664732821394157, "grad_norm": 0.8329172730445862, "learning_rate": 1.8039588649206976e-05, "loss": 0.16857903003692626, "step": 45630 }, { "epoch": 0.22669699889731082, "grad_norm": 1.7359368801116943, "learning_rate": 1.8038631867411434e-05, "loss": 0.1717156410217285, "step": 45640 }, { "epoch": 0.2267466695806801, "grad_norm": 0.8545939326286316, "learning_rate": 1.8037674877579334e-05, "loss": 0.17710186243057252, "step": 45650 }, { "epoch": 0.22679634026404935, "grad_norm": 2.0617198944091797, "learning_rate": 1.803671767973544e-05, "loss": 0.16993362903594972, "step": 45660 }, { "epoch": 0.2268460109474186, "grad_norm": 2.534472703933716, "learning_rate": 1.8035760273904524e-05, "loss": 0.17917721271514891, "step": 45670 }, { "epoch": 0.22689568163078788, "grad_norm": 2.2788493633270264, "learning_rate": 1.8034802660111364e-05, "loss": 0.18285393714904785, "step": 45680 }, { "epoch": 0.22694535231415713, "grad_norm": 1.1153026819229126, "learning_rate": 1.803384483838074e-05, "loss": 0.17535843849182128, "step": 45690 }, { "epoch": 0.2269950229975264, "grad_norm": 1.6194325685501099, "learning_rate": 1.8032886808737442e-05, "loss": 0.16355714797973633, "step": 45700 }, { "epoch": 0.22704469368089567, "grad_norm": 1.741919755935669, "learning_rate": 1.8031928571206262e-05, "loss": 0.17403943538665773, "step": 45710 }, { "epoch": 0.22709436436426492, "grad_norm": 0.8102206587791443, "learning_rate": 1.8030970125812005e-05, "loss": 0.17530938386917114, "step": 45720 }, { "epoch": 0.2271440350476342, "grad_norm": 0.8810812830924988, "learning_rate": 1.8030011472579467e-05, "loss": 0.17302651405334474, "step": 45730 }, { "epoch": 0.22719370573100345, "grad_norm": 0.6691656708717346, "learning_rate": 1.802905261153346e-05, "loss": 0.15840048789978028, "step": 45740 }, { "epoch": 0.2272433764143727, "grad_norm": 1.9008917808532715, "learning_rate": 1.8028093542698802e-05, "loss": 0.15940375328063966, "step": 45750 }, { "epoch": 0.22729304709774198, "grad_norm": 1.532155990600586, "learning_rate": 1.802713426610031e-05, "loss": 0.1715121865272522, "step": 45760 }, { "epoch": 0.22734271778111123, "grad_norm": 1.414204716682434, "learning_rate": 1.802617478176281e-05, "loss": 0.18542683124542236, "step": 45770 }, { "epoch": 0.22739238846448048, "grad_norm": 0.8363146185874939, "learning_rate": 1.8025215089711136e-05, "loss": 0.16782244443893432, "step": 45780 }, { "epoch": 0.22744205914784976, "grad_norm": 0.87109375, "learning_rate": 1.8024255189970117e-05, "loss": 0.17319626808166505, "step": 45790 }, { "epoch": 0.22749172983121901, "grad_norm": 0.9631379246711731, "learning_rate": 1.8023295082564605e-05, "loss": 0.17617390155792237, "step": 45800 }, { "epoch": 0.2275414005145883, "grad_norm": 1.0262690782546997, "learning_rate": 1.8022334767519442e-05, "loss": 0.1681060314178467, "step": 45810 }, { "epoch": 0.22759107119795755, "grad_norm": 1.1313139200210571, "learning_rate": 1.802137424485948e-05, "loss": 0.18390130996704102, "step": 45820 }, { "epoch": 0.2276407418813268, "grad_norm": 2.313735008239746, "learning_rate": 1.8020413514609578e-05, "loss": 0.18711036443710327, "step": 45830 }, { "epoch": 0.22769041256469608, "grad_norm": 1.0300240516662598, "learning_rate": 1.8019452576794598e-05, "loss": 0.16662380695343018, "step": 45840 }, { "epoch": 0.22774008324806533, "grad_norm": 1.3180015087127686, "learning_rate": 1.8018491431439412e-05, "loss": 0.1719815969467163, "step": 45850 }, { "epoch": 0.22778975393143458, "grad_norm": 1.160674810409546, "learning_rate": 1.801753007856889e-05, "loss": 0.17994608879089355, "step": 45860 }, { "epoch": 0.22783942461480386, "grad_norm": 1.9327776432037354, "learning_rate": 1.8016568518207912e-05, "loss": 0.1731091022491455, "step": 45870 }, { "epoch": 0.2278890952981731, "grad_norm": 0.6104832887649536, "learning_rate": 1.8015606750381366e-05, "loss": 0.18870062828063966, "step": 45880 }, { "epoch": 0.22793876598154236, "grad_norm": 1.1605106592178345, "learning_rate": 1.801464477511414e-05, "loss": 0.19423744678497315, "step": 45890 }, { "epoch": 0.22798843666491164, "grad_norm": 2.6739742755889893, "learning_rate": 1.801368259243113e-05, "loss": 0.1788040041923523, "step": 45900 }, { "epoch": 0.2280381073482809, "grad_norm": 2.0734593868255615, "learning_rate": 1.8012720202357238e-05, "loss": 0.17757203578948974, "step": 45910 }, { "epoch": 0.22808777803165015, "grad_norm": 0.9009135961532593, "learning_rate": 1.8011757604917366e-05, "loss": 0.17099525928497314, "step": 45920 }, { "epoch": 0.22813744871501943, "grad_norm": 0.9566878080368042, "learning_rate": 1.801079480013643e-05, "loss": 0.18856161832809448, "step": 45930 }, { "epoch": 0.22818711939838868, "grad_norm": 2.229151487350464, "learning_rate": 1.800983178803935e-05, "loss": 0.17786726951599122, "step": 45940 }, { "epoch": 0.22823679008175796, "grad_norm": 2.0273361206054688, "learning_rate": 1.8008868568651038e-05, "loss": 0.1714189052581787, "step": 45950 }, { "epoch": 0.2282864607651272, "grad_norm": 0.8266705274581909, "learning_rate": 1.8007905141996428e-05, "loss": 0.17882542610168456, "step": 45960 }, { "epoch": 0.22833613144849646, "grad_norm": 1.083309531211853, "learning_rate": 1.8006941508100456e-05, "loss": 0.18710827827453613, "step": 45970 }, { "epoch": 0.22838580213186574, "grad_norm": 1.085270643234253, "learning_rate": 1.800597766698805e-05, "loss": 0.1797069787979126, "step": 45980 }, { "epoch": 0.228435472815235, "grad_norm": 1.4910825490951538, "learning_rate": 1.8005013618684168e-05, "loss": 0.1857602596282959, "step": 45990 }, { "epoch": 0.22848514349860424, "grad_norm": 1.3456002473831177, "learning_rate": 1.800404936321375e-05, "loss": 0.17220956087112427, "step": 46000 }, { "epoch": 0.22853481418197352, "grad_norm": 1.214491367340088, "learning_rate": 1.8003084900601754e-05, "loss": 0.1744004726409912, "step": 46010 }, { "epoch": 0.22858448486534277, "grad_norm": 1.9166663885116577, "learning_rate": 1.8002120230873138e-05, "loss": 0.1835716485977173, "step": 46020 }, { "epoch": 0.22863415554871203, "grad_norm": 1.166061282157898, "learning_rate": 1.8001155354052867e-05, "loss": 0.19577094316482543, "step": 46030 }, { "epoch": 0.2286838262320813, "grad_norm": 2.5363237857818604, "learning_rate": 1.8000190270165916e-05, "loss": 0.18042855262756347, "step": 46040 }, { "epoch": 0.22873349691545056, "grad_norm": 2.5361151695251465, "learning_rate": 1.7999224979237254e-05, "loss": 0.19006584882736205, "step": 46050 }, { "epoch": 0.22878316759881984, "grad_norm": 0.9129019975662231, "learning_rate": 1.7998259481291867e-05, "loss": 0.167291259765625, "step": 46060 }, { "epoch": 0.2288328382821891, "grad_norm": 1.30681574344635, "learning_rate": 1.799729377635474e-05, "loss": 0.16993579864501954, "step": 46070 }, { "epoch": 0.22888250896555834, "grad_norm": 0.718174397945404, "learning_rate": 1.7996327864450868e-05, "loss": 0.19263107776641847, "step": 46080 }, { "epoch": 0.22893217964892762, "grad_norm": 0.684973955154419, "learning_rate": 1.7995361745605245e-05, "loss": 0.17776567935943605, "step": 46090 }, { "epoch": 0.22898185033229687, "grad_norm": 0.8624271750450134, "learning_rate": 1.7994395419842873e-05, "loss": 0.16177899837493898, "step": 46100 }, { "epoch": 0.22903152101566612, "grad_norm": 1.2062207460403442, "learning_rate": 1.7993428887188763e-05, "loss": 0.16908340454101561, "step": 46110 }, { "epoch": 0.2290811916990354, "grad_norm": 1.0487486124038696, "learning_rate": 1.7992462147667926e-05, "loss": 0.18105180263519288, "step": 46120 }, { "epoch": 0.22913086238240465, "grad_norm": 1.2860926389694214, "learning_rate": 1.7991495201305384e-05, "loss": 0.1764431118965149, "step": 46130 }, { "epoch": 0.2291805330657739, "grad_norm": 0.5657998919487, "learning_rate": 1.799052804812616e-05, "loss": 0.16646348237991332, "step": 46140 }, { "epoch": 0.22923020374914319, "grad_norm": 2.090883255004883, "learning_rate": 1.798956068815528e-05, "loss": 0.1904401659965515, "step": 46150 }, { "epoch": 0.22927987443251244, "grad_norm": 1.208886981010437, "learning_rate": 1.7988593121417785e-05, "loss": 0.1678167462348938, "step": 46160 }, { "epoch": 0.22932954511588172, "grad_norm": 1.0139069557189941, "learning_rate": 1.7987625347938714e-05, "loss": 0.1692621350288391, "step": 46170 }, { "epoch": 0.22937921579925097, "grad_norm": 0.8528492450714111, "learning_rate": 1.798665736774311e-05, "loss": 0.17914990186691285, "step": 46180 }, { "epoch": 0.22942888648262022, "grad_norm": 0.7095669507980347, "learning_rate": 1.7985689180856018e-05, "loss": 0.17777769565582274, "step": 46190 }, { "epoch": 0.2294785571659895, "grad_norm": 2.0939083099365234, "learning_rate": 1.7984720787302507e-05, "loss": 0.1814065933227539, "step": 46200 }, { "epoch": 0.22952822784935875, "grad_norm": 0.6451483964920044, "learning_rate": 1.798375218710763e-05, "loss": 0.17565008401870727, "step": 46210 }, { "epoch": 0.229577898532728, "grad_norm": 1.2827861309051514, "learning_rate": 1.7982783380296455e-05, "loss": 0.17504225969314574, "step": 46220 }, { "epoch": 0.22962756921609728, "grad_norm": 2.3159024715423584, "learning_rate": 1.7981814366894062e-05, "loss": 0.17882733345031737, "step": 46230 }, { "epoch": 0.22967723989946653, "grad_norm": 0.9800800085067749, "learning_rate": 1.7980845146925517e-05, "loss": 0.18481845855712892, "step": 46240 }, { "epoch": 0.22972691058283579, "grad_norm": 0.781511664390564, "learning_rate": 1.797987572041591e-05, "loss": 0.16983214616775513, "step": 46250 }, { "epoch": 0.22977658126620507, "grad_norm": 0.896323561668396, "learning_rate": 1.797890608739033e-05, "loss": 0.1742356538772583, "step": 46260 }, { "epoch": 0.22982625194957432, "grad_norm": 1.0150567293167114, "learning_rate": 1.7977936247873866e-05, "loss": 0.18756954669952391, "step": 46270 }, { "epoch": 0.2298759226329436, "grad_norm": 0.9287708401679993, "learning_rate": 1.7976966201891622e-05, "loss": 0.15565621852874756, "step": 46280 }, { "epoch": 0.22992559331631285, "grad_norm": 1.1722749471664429, "learning_rate": 1.7975995949468695e-05, "loss": 0.17874091863632202, "step": 46290 }, { "epoch": 0.2299752639996821, "grad_norm": 0.819537878036499, "learning_rate": 1.7975025490630206e-05, "loss": 0.16187784671783448, "step": 46300 }, { "epoch": 0.23002493468305138, "grad_norm": 1.590973973274231, "learning_rate": 1.7974054825401263e-05, "loss": 0.18935725688934327, "step": 46310 }, { "epoch": 0.23007460536642063, "grad_norm": 1.9721336364746094, "learning_rate": 1.7973083953806982e-05, "loss": 0.1682231307029724, "step": 46320 }, { "epoch": 0.23012427604978988, "grad_norm": 1.7964489459991455, "learning_rate": 1.79721128758725e-05, "loss": 0.17747340202331544, "step": 46330 }, { "epoch": 0.23017394673315916, "grad_norm": 1.036177396774292, "learning_rate": 1.7971141591622942e-05, "loss": 0.17020481824874878, "step": 46340 }, { "epoch": 0.23022361741652841, "grad_norm": 1.0393197536468506, "learning_rate": 1.7970170101083442e-05, "loss": 0.16938787698745728, "step": 46350 }, { "epoch": 0.23027328809989767, "grad_norm": 1.9074808359146118, "learning_rate": 1.7969198404279144e-05, "loss": 0.1660427451133728, "step": 46360 }, { "epoch": 0.23032295878326695, "grad_norm": 0.6680958867073059, "learning_rate": 1.79682265012352e-05, "loss": 0.16858389377593994, "step": 46370 }, { "epoch": 0.2303726294666362, "grad_norm": 1.7240490913391113, "learning_rate": 1.7967254391976752e-05, "loss": 0.1594149112701416, "step": 46380 }, { "epoch": 0.23042230015000548, "grad_norm": 1.378413438796997, "learning_rate": 1.7966282076528968e-05, "loss": 0.1890308141708374, "step": 46390 }, { "epoch": 0.23047197083337473, "grad_norm": 1.4322106838226318, "learning_rate": 1.7965309554917007e-05, "loss": 0.1844362735748291, "step": 46400 }, { "epoch": 0.23052164151674398, "grad_norm": 0.8735024929046631, "learning_rate": 1.7964336827166032e-05, "loss": 0.18328235149383545, "step": 46410 }, { "epoch": 0.23057131220011326, "grad_norm": 1.365983486175537, "learning_rate": 1.7963363893301228e-05, "loss": 0.16296584606170655, "step": 46420 }, { "epoch": 0.2306209828834825, "grad_norm": 1.5923173427581787, "learning_rate": 1.7962390753347768e-05, "loss": 0.17268548011779786, "step": 46430 }, { "epoch": 0.23067065356685176, "grad_norm": 1.8149105310440063, "learning_rate": 1.7961417407330833e-05, "loss": 0.17358347177505493, "step": 46440 }, { "epoch": 0.23072032425022104, "grad_norm": 1.3450801372528076, "learning_rate": 1.796044385527562e-05, "loss": 0.17543063163757325, "step": 46450 }, { "epoch": 0.2307699949335903, "grad_norm": 1.439153790473938, "learning_rate": 1.7959470097207325e-05, "loss": 0.18216147422790527, "step": 46460 }, { "epoch": 0.23081966561695955, "grad_norm": 1.7611896991729736, "learning_rate": 1.7958496133151136e-05, "loss": 0.16856104135513306, "step": 46470 }, { "epoch": 0.23086933630032883, "grad_norm": 2.0410242080688477, "learning_rate": 1.795752196313227e-05, "loss": 0.17077443599700928, "step": 46480 }, { "epoch": 0.23091900698369808, "grad_norm": 0.9419223070144653, "learning_rate": 1.7956547587175936e-05, "loss": 0.1748589515686035, "step": 46490 }, { "epoch": 0.23096867766706736, "grad_norm": 0.5796515941619873, "learning_rate": 1.795557300530735e-05, "loss": 0.1712270736694336, "step": 46500 }, { "epoch": 0.2310183483504366, "grad_norm": 1.6759992837905884, "learning_rate": 1.7954598217551732e-05, "loss": 0.164522123336792, "step": 46510 }, { "epoch": 0.23106801903380586, "grad_norm": 1.2143642902374268, "learning_rate": 1.795362322393431e-05, "loss": 0.16857235431671141, "step": 46520 }, { "epoch": 0.23111768971717514, "grad_norm": 1.0033785104751587, "learning_rate": 1.795264802448032e-05, "loss": 0.15624830722808838, "step": 46530 }, { "epoch": 0.2311673604005444, "grad_norm": 1.1400954723358154, "learning_rate": 1.7951672619214996e-05, "loss": 0.19538533687591553, "step": 46540 }, { "epoch": 0.23121703108391364, "grad_norm": 1.2142592668533325, "learning_rate": 1.7950697008163577e-05, "loss": 0.16571383476257323, "step": 46550 }, { "epoch": 0.23126670176728292, "grad_norm": 1.486439824104309, "learning_rate": 1.7949721191351322e-05, "loss": 0.17839901447296141, "step": 46560 }, { "epoch": 0.23131637245065217, "grad_norm": 1.9402046203613281, "learning_rate": 1.7948745168803478e-05, "loss": 0.17768651247024536, "step": 46570 }, { "epoch": 0.23136604313402143, "grad_norm": 0.9128273129463196, "learning_rate": 1.7947768940545304e-05, "loss": 0.18083710670471193, "step": 46580 }, { "epoch": 0.2314157138173907, "grad_norm": 1.456140398979187, "learning_rate": 1.7946792506602065e-05, "loss": 0.16855136156082154, "step": 46590 }, { "epoch": 0.23146538450075996, "grad_norm": 1.2449190616607666, "learning_rate": 1.7945815866999033e-05, "loss": 0.16539229154586793, "step": 46600 }, { "epoch": 0.23151505518412924, "grad_norm": 0.9194140434265137, "learning_rate": 1.794483902176148e-05, "loss": 0.1893003463745117, "step": 46610 }, { "epoch": 0.2315647258674985, "grad_norm": 0.9560869932174683, "learning_rate": 1.7943861970914686e-05, "loss": 0.1742119312286377, "step": 46620 }, { "epoch": 0.23161439655086774, "grad_norm": 1.238116979598999, "learning_rate": 1.7942884714483943e-05, "loss": 0.1765766739845276, "step": 46630 }, { "epoch": 0.23166406723423702, "grad_norm": 2.4878523349761963, "learning_rate": 1.794190725249453e-05, "loss": 0.1625959038734436, "step": 46640 }, { "epoch": 0.23171373791760627, "grad_norm": 1.2093623876571655, "learning_rate": 1.7940929584971757e-05, "loss": 0.1661696434020996, "step": 46650 }, { "epoch": 0.23176340860097552, "grad_norm": 1.9042247533798218, "learning_rate": 1.7939951711940916e-05, "loss": 0.18120434284210205, "step": 46660 }, { "epoch": 0.2318130792843448, "grad_norm": 1.088602066040039, "learning_rate": 1.793897363342732e-05, "loss": 0.17480747699737548, "step": 46670 }, { "epoch": 0.23186274996771405, "grad_norm": 0.8205082416534424, "learning_rate": 1.7937995349456276e-05, "loss": 0.16835812330245972, "step": 46680 }, { "epoch": 0.2319124206510833, "grad_norm": 1.0298832654953003, "learning_rate": 1.7937016860053106e-05, "loss": 0.17384289503097533, "step": 46690 }, { "epoch": 0.23196209133445259, "grad_norm": 1.1758270263671875, "learning_rate": 1.7936038165243127e-05, "loss": 0.1650652289390564, "step": 46700 }, { "epoch": 0.23201176201782184, "grad_norm": 0.9705408811569214, "learning_rate": 1.7935059265051674e-05, "loss": 0.18431200981140136, "step": 46710 }, { "epoch": 0.2320614327011911, "grad_norm": 1.0665600299835205, "learning_rate": 1.793408015950408e-05, "loss": 0.17478082180023194, "step": 46720 }, { "epoch": 0.23211110338456037, "grad_norm": 2.0495827198028564, "learning_rate": 1.793310084862568e-05, "loss": 0.18980268239974976, "step": 46730 }, { "epoch": 0.23216077406792962, "grad_norm": 0.7699131965637207, "learning_rate": 1.793212133244182e-05, "loss": 0.17133536338806152, "step": 46740 }, { "epoch": 0.2322104447512989, "grad_norm": 1.13470458984375, "learning_rate": 1.7931141610977845e-05, "loss": 0.18115785121917724, "step": 46750 }, { "epoch": 0.23226011543466815, "grad_norm": 1.212618350982666, "learning_rate": 1.793016168425912e-05, "loss": 0.18112024068832397, "step": 46760 }, { "epoch": 0.2323097861180374, "grad_norm": 0.9788942337036133, "learning_rate": 1.7929181552311e-05, "loss": 0.1792904257774353, "step": 46770 }, { "epoch": 0.23235945680140668, "grad_norm": 1.5485055446624756, "learning_rate": 1.7928201215158845e-05, "loss": 0.18298654556274413, "step": 46780 }, { "epoch": 0.23240912748477593, "grad_norm": 1.031451940536499, "learning_rate": 1.792722067282803e-05, "loss": 0.19296576976776122, "step": 46790 }, { "epoch": 0.2324587981681452, "grad_norm": 1.2265492677688599, "learning_rate": 1.7926239925343933e-05, "loss": 0.179259991645813, "step": 46800 }, { "epoch": 0.23250846885151447, "grad_norm": 0.8498410582542419, "learning_rate": 1.7925258972731936e-05, "loss": 0.17287187576293944, "step": 46810 }, { "epoch": 0.23255813953488372, "grad_norm": 0.809431791305542, "learning_rate": 1.792427781501742e-05, "loss": 0.18463587760925293, "step": 46820 }, { "epoch": 0.23260781021825297, "grad_norm": 1.3647489547729492, "learning_rate": 1.7923296452225782e-05, "loss": 0.1696542024612427, "step": 46830 }, { "epoch": 0.23265748090162225, "grad_norm": 1.05886971950531, "learning_rate": 1.792231488438242e-05, "loss": 0.17444865703582763, "step": 46840 }, { "epoch": 0.2327071515849915, "grad_norm": 1.11488938331604, "learning_rate": 1.7921333111512732e-05, "loss": 0.17460163831710815, "step": 46850 }, { "epoch": 0.23275682226836078, "grad_norm": 0.6317813396453857, "learning_rate": 1.7920351133642126e-05, "loss": 0.18250768184661864, "step": 46860 }, { "epoch": 0.23280649295173003, "grad_norm": 0.873408317565918, "learning_rate": 1.791936895079602e-05, "loss": 0.16741272211074829, "step": 46870 }, { "epoch": 0.23285616363509928, "grad_norm": 0.8656021356582642, "learning_rate": 1.7918386562999828e-05, "loss": 0.17483389377593994, "step": 46880 }, { "epoch": 0.23290583431846856, "grad_norm": 2.0990099906921387, "learning_rate": 1.7917403970278976e-05, "loss": 0.1809914827346802, "step": 46890 }, { "epoch": 0.23295550500183781, "grad_norm": 1.2483049631118774, "learning_rate": 1.7916421172658892e-05, "loss": 0.17510956525802612, "step": 46900 }, { "epoch": 0.23300517568520707, "grad_norm": 1.7236264944076538, "learning_rate": 1.7915438170165015e-05, "loss": 0.1827960729598999, "step": 46910 }, { "epoch": 0.23305484636857635, "grad_norm": 1.5407509803771973, "learning_rate": 1.7914454962822775e-05, "loss": 0.1861095666885376, "step": 46920 }, { "epoch": 0.2331045170519456, "grad_norm": 0.8245788812637329, "learning_rate": 1.7913471550657628e-05, "loss": 0.18426318168640138, "step": 46930 }, { "epoch": 0.23315418773531485, "grad_norm": 0.49995478987693787, "learning_rate": 1.791248793369502e-05, "loss": 0.18122767210006713, "step": 46940 }, { "epoch": 0.23320385841868413, "grad_norm": 2.2191059589385986, "learning_rate": 1.7911504111960402e-05, "loss": 0.1777145743370056, "step": 46950 }, { "epoch": 0.23325352910205338, "grad_norm": 2.4760677814483643, "learning_rate": 1.7910520085479237e-05, "loss": 0.1746022582054138, "step": 46960 }, { "epoch": 0.23330319978542266, "grad_norm": 1.6524629592895508, "learning_rate": 1.7909535854276998e-05, "loss": 0.1627427816390991, "step": 46970 }, { "epoch": 0.2333528704687919, "grad_norm": 2.307849407196045, "learning_rate": 1.7908551418379147e-05, "loss": 0.17625052928924562, "step": 46980 }, { "epoch": 0.23340254115216116, "grad_norm": 1.1760574579238892, "learning_rate": 1.7907566777811167e-05, "loss": 0.18831502199172973, "step": 46990 }, { "epoch": 0.23345221183553044, "grad_norm": 2.006655216217041, "learning_rate": 1.7906581932598535e-05, "loss": 0.17479214668273926, "step": 47000 }, { "epoch": 0.2335018825188997, "grad_norm": 1.191232681274414, "learning_rate": 1.790559688276675e-05, "loss": 0.1711815118789673, "step": 47010 }, { "epoch": 0.23355155320226895, "grad_norm": 1.4821032285690308, "learning_rate": 1.790461162834129e-05, "loss": 0.16927156448364258, "step": 47020 }, { "epoch": 0.23360122388563823, "grad_norm": 1.4798749685287476, "learning_rate": 1.790362616934766e-05, "loss": 0.1744221568107605, "step": 47030 }, { "epoch": 0.23365089456900748, "grad_norm": 0.9781846404075623, "learning_rate": 1.7902640505811365e-05, "loss": 0.17867069244384765, "step": 47040 }, { "epoch": 0.23370056525237673, "grad_norm": 0.9760748147964478, "learning_rate": 1.790165463775791e-05, "loss": 0.16081690788269043, "step": 47050 }, { "epoch": 0.233750235935746, "grad_norm": 1.1094119548797607, "learning_rate": 1.7900668565212806e-05, "loss": 0.1782330632209778, "step": 47060 }, { "epoch": 0.23379990661911526, "grad_norm": 1.270683765411377, "learning_rate": 1.789968228820158e-05, "loss": 0.18343799114227294, "step": 47070 }, { "epoch": 0.23384957730248454, "grad_norm": 0.8668015599250793, "learning_rate": 1.7898695806749756e-05, "loss": 0.17616046667099, "step": 47080 }, { "epoch": 0.2338992479858538, "grad_norm": 0.613165020942688, "learning_rate": 1.7897709120882855e-05, "loss": 0.16471935510635377, "step": 47090 }, { "epoch": 0.23394891866922304, "grad_norm": 0.7236626148223877, "learning_rate": 1.7896722230626422e-05, "loss": 0.17439591884613037, "step": 47100 }, { "epoch": 0.23399858935259232, "grad_norm": 0.7445188164710999, "learning_rate": 1.789573513600599e-05, "loss": 0.160310959815979, "step": 47110 }, { "epoch": 0.23404826003596158, "grad_norm": 0.7394803166389465, "learning_rate": 1.789474783704711e-05, "loss": 0.17927172183990478, "step": 47120 }, { "epoch": 0.23409793071933083, "grad_norm": 1.6845601797103882, "learning_rate": 1.7893760333775324e-05, "loss": 0.17052870988845825, "step": 47130 }, { "epoch": 0.2341476014027001, "grad_norm": 1.6130071878433228, "learning_rate": 1.78927726262162e-05, "loss": 0.1816693902015686, "step": 47140 }, { "epoch": 0.23419727208606936, "grad_norm": 1.3236901760101318, "learning_rate": 1.7891784714395292e-05, "loss": 0.17774081230163574, "step": 47150 }, { "epoch": 0.2342469427694386, "grad_norm": 1.1980597972869873, "learning_rate": 1.789079659833817e-05, "loss": 0.17439991235733032, "step": 47160 }, { "epoch": 0.2342966134528079, "grad_norm": 0.6858444809913635, "learning_rate": 1.7889808278070405e-05, "loss": 0.16853642463684082, "step": 47170 }, { "epoch": 0.23434628413617714, "grad_norm": 0.962577760219574, "learning_rate": 1.7888819753617572e-05, "loss": 0.17328732013702391, "step": 47180 }, { "epoch": 0.23439595481954642, "grad_norm": 1.2729393243789673, "learning_rate": 1.7887831025005256e-05, "loss": 0.16440937519073487, "step": 47190 }, { "epoch": 0.23444562550291567, "grad_norm": 1.4712461233139038, "learning_rate": 1.7886842092259047e-05, "loss": 0.17302243709564208, "step": 47200 }, { "epoch": 0.23449529618628492, "grad_norm": 2.2213940620422363, "learning_rate": 1.7885852955404534e-05, "loss": 0.1934983253479004, "step": 47210 }, { "epoch": 0.2345449668696542, "grad_norm": 1.6842689514160156, "learning_rate": 1.788486361446732e-05, "loss": 0.17096318006515504, "step": 47220 }, { "epoch": 0.23459463755302346, "grad_norm": 1.9764348268508911, "learning_rate": 1.7883874069473e-05, "loss": 0.18084825277328492, "step": 47230 }, { "epoch": 0.2346443082363927, "grad_norm": 1.4281944036483765, "learning_rate": 1.7882884320447195e-05, "loss": 0.1744060754776001, "step": 47240 }, { "epoch": 0.234693978919762, "grad_norm": 0.6613242030143738, "learning_rate": 1.7881894367415512e-05, "loss": 0.16490858793258667, "step": 47250 }, { "epoch": 0.23474364960313124, "grad_norm": 1.8106707334518433, "learning_rate": 1.788090421040357e-05, "loss": 0.17870724201202393, "step": 47260 }, { "epoch": 0.2347933202865005, "grad_norm": 1.307255744934082, "learning_rate": 1.7879913849437e-05, "loss": 0.16979764699935912, "step": 47270 }, { "epoch": 0.23484299096986977, "grad_norm": 2.1119372844696045, "learning_rate": 1.7878923284541425e-05, "loss": 0.17693554162979125, "step": 47280 }, { "epoch": 0.23489266165323902, "grad_norm": 1.3155699968338013, "learning_rate": 1.787793251574248e-05, "loss": 0.18594008684158325, "step": 47290 }, { "epoch": 0.2349423323366083, "grad_norm": 1.9266554117202759, "learning_rate": 1.7876941543065816e-05, "loss": 0.1735159635543823, "step": 47300 }, { "epoch": 0.23499200301997755, "grad_norm": 2.110393524169922, "learning_rate": 1.7875950366537068e-05, "loss": 0.17919843196868895, "step": 47310 }, { "epoch": 0.2350416737033468, "grad_norm": 0.6936786770820618, "learning_rate": 1.7874958986181893e-05, "loss": 0.17503731250762938, "step": 47320 }, { "epoch": 0.23509134438671608, "grad_norm": 1.672410488128662, "learning_rate": 1.787396740202594e-05, "loss": 0.1785458207130432, "step": 47330 }, { "epoch": 0.23514101507008534, "grad_norm": 1.533932089805603, "learning_rate": 1.7872975614094883e-05, "loss": 0.1741115927696228, "step": 47340 }, { "epoch": 0.2351906857534546, "grad_norm": 1.2556664943695068, "learning_rate": 1.7871983622414383e-05, "loss": 0.18437020778656005, "step": 47350 }, { "epoch": 0.23524035643682387, "grad_norm": 1.5037307739257812, "learning_rate": 1.7870991427010106e-05, "loss": 0.16461843252182007, "step": 47360 }, { "epoch": 0.23529002712019312, "grad_norm": 1.085447907447815, "learning_rate": 1.786999902790774e-05, "loss": 0.17290565967559815, "step": 47370 }, { "epoch": 0.23533969780356237, "grad_norm": 1.358238935470581, "learning_rate": 1.7869006425132966e-05, "loss": 0.18654706478118896, "step": 47380 }, { "epoch": 0.23538936848693165, "grad_norm": 0.9037073850631714, "learning_rate": 1.7868013618711467e-05, "loss": 0.1774266004562378, "step": 47390 }, { "epoch": 0.2354390391703009, "grad_norm": 0.6695727705955505, "learning_rate": 1.786702060866894e-05, "loss": 0.16969306468963624, "step": 47400 }, { "epoch": 0.23548870985367018, "grad_norm": 1.4189646244049072, "learning_rate": 1.786602739503108e-05, "loss": 0.17455915212631226, "step": 47410 }, { "epoch": 0.23553838053703943, "grad_norm": 2.619307518005371, "learning_rate": 1.7865033977823596e-05, "loss": 0.1716625690460205, "step": 47420 }, { "epoch": 0.23558805122040868, "grad_norm": 1.6037174463272095, "learning_rate": 1.7864040357072197e-05, "loss": 0.1827281355857849, "step": 47430 }, { "epoch": 0.23563772190377796, "grad_norm": 1.4686585664749146, "learning_rate": 1.7863046532802593e-05, "loss": 0.1787865400314331, "step": 47440 }, { "epoch": 0.23568739258714722, "grad_norm": 2.3122055530548096, "learning_rate": 1.7862052505040506e-05, "loss": 0.17676937580108643, "step": 47450 }, { "epoch": 0.23573706327051647, "grad_norm": 1.4454941749572754, "learning_rate": 1.786105827381166e-05, "loss": 0.1781732201576233, "step": 47460 }, { "epoch": 0.23578673395388575, "grad_norm": 1.5988703966140747, "learning_rate": 1.786006383914179e-05, "loss": 0.17408665418624877, "step": 47470 }, { "epoch": 0.235836404637255, "grad_norm": 0.7245978116989136, "learning_rate": 1.7859069201056628e-05, "loss": 0.17371101379394532, "step": 47480 }, { "epoch": 0.23588607532062425, "grad_norm": 1.9690170288085938, "learning_rate": 1.785807435958191e-05, "loss": 0.1735974907875061, "step": 47490 }, { "epoch": 0.23593574600399353, "grad_norm": 1.503501057624817, "learning_rate": 1.7857079314743394e-05, "loss": 0.17397701740264893, "step": 47500 }, { "epoch": 0.23598541668736278, "grad_norm": 1.6776621341705322, "learning_rate": 1.785608406656682e-05, "loss": 0.17389163970947266, "step": 47510 }, { "epoch": 0.23603508737073203, "grad_norm": 0.7241677045822144, "learning_rate": 1.7855088615077954e-05, "loss": 0.17555826902389526, "step": 47520 }, { "epoch": 0.2360847580541013, "grad_norm": 1.3759653568267822, "learning_rate": 1.785409296030255e-05, "loss": 0.17837233543395997, "step": 47530 }, { "epoch": 0.23613442873747056, "grad_norm": 1.3003406524658203, "learning_rate": 1.7853097102266376e-05, "loss": 0.1771146297454834, "step": 47540 }, { "epoch": 0.23618409942083984, "grad_norm": 0.6698868870735168, "learning_rate": 1.785210104099521e-05, "loss": 0.17256160974502563, "step": 47550 }, { "epoch": 0.2362337701042091, "grad_norm": 0.9395743012428284, "learning_rate": 1.7851104776514825e-05, "loss": 0.19247939586639404, "step": 47560 }, { "epoch": 0.23628344078757835, "grad_norm": 0.9924072027206421, "learning_rate": 1.7850108308851007e-05, "loss": 0.17103879451751708, "step": 47570 }, { "epoch": 0.23633311147094763, "grad_norm": 1.3688777685165405, "learning_rate": 1.7849111638029538e-05, "loss": 0.18017840385437012, "step": 47580 }, { "epoch": 0.23638278215431688, "grad_norm": 1.210590124130249, "learning_rate": 1.784811476407622e-05, "loss": 0.17788393497467042, "step": 47590 }, { "epoch": 0.23643245283768613, "grad_norm": 0.8897838592529297, "learning_rate": 1.7847117687016843e-05, "loss": 0.18032773733139038, "step": 47600 }, { "epoch": 0.2364821235210554, "grad_norm": 2.024158239364624, "learning_rate": 1.784612040687722e-05, "loss": 0.17797247171401978, "step": 47610 }, { "epoch": 0.23653179420442466, "grad_norm": 0.8740531206130981, "learning_rate": 1.7845122923683154e-05, "loss": 0.17223283052444457, "step": 47620 }, { "epoch": 0.2365814648877939, "grad_norm": 1.532396912574768, "learning_rate": 1.784412523746046e-05, "loss": 0.20095155239105225, "step": 47630 }, { "epoch": 0.2366311355711632, "grad_norm": 1.1004693508148193, "learning_rate": 1.784312734823496e-05, "loss": 0.18319820165634154, "step": 47640 }, { "epoch": 0.23668080625453244, "grad_norm": 1.7128376960754395, "learning_rate": 1.784212925603248e-05, "loss": 0.16829450130462648, "step": 47650 }, { "epoch": 0.23673047693790172, "grad_norm": 0.9799556732177734, "learning_rate": 1.7841130960878844e-05, "loss": 0.1846153974533081, "step": 47660 }, { "epoch": 0.23678014762127098, "grad_norm": 1.2675917148590088, "learning_rate": 1.7840132462799893e-05, "loss": 0.1793203592300415, "step": 47670 }, { "epoch": 0.23682981830464023, "grad_norm": 1.2577844858169556, "learning_rate": 1.7839133761821467e-05, "loss": 0.16430811882019042, "step": 47680 }, { "epoch": 0.2368794889880095, "grad_norm": 1.8866126537322998, "learning_rate": 1.783813485796941e-05, "loss": 0.18431805372238158, "step": 47690 }, { "epoch": 0.23692915967137876, "grad_norm": 0.8510133028030396, "learning_rate": 1.7837135751269575e-05, "loss": 0.17507407665252686, "step": 47700 }, { "epoch": 0.236978830354748, "grad_norm": 0.9325466752052307, "learning_rate": 1.783613644174782e-05, "loss": 0.16433247327804565, "step": 47710 }, { "epoch": 0.2370285010381173, "grad_norm": 1.2279211282730103, "learning_rate": 1.7835136929430005e-05, "loss": 0.1667623281478882, "step": 47720 }, { "epoch": 0.23707817172148654, "grad_norm": 1.672283411026001, "learning_rate": 1.7834137214341993e-05, "loss": 0.1762028455734253, "step": 47730 }, { "epoch": 0.2371278424048558, "grad_norm": 0.8426225185394287, "learning_rate": 1.783313729650966e-05, "loss": 0.1911322832107544, "step": 47740 }, { "epoch": 0.23717751308822507, "grad_norm": 1.6660728454589844, "learning_rate": 1.7832137175958887e-05, "loss": 0.17892401218414306, "step": 47750 }, { "epoch": 0.23722718377159432, "grad_norm": 0.9717153906822205, "learning_rate": 1.783113685271555e-05, "loss": 0.17794172763824462, "step": 47760 }, { "epoch": 0.2372768544549636, "grad_norm": 1.1911540031433105, "learning_rate": 1.783013632680554e-05, "loss": 0.16956491470336915, "step": 47770 }, { "epoch": 0.23732652513833286, "grad_norm": 1.4587275981903076, "learning_rate": 1.7829135598254753e-05, "loss": 0.18040895462036133, "step": 47780 }, { "epoch": 0.2373761958217021, "grad_norm": 1.0404138565063477, "learning_rate": 1.782813466708908e-05, "loss": 0.18206146955490113, "step": 47790 }, { "epoch": 0.2374258665050714, "grad_norm": 1.4776772260665894, "learning_rate": 1.7827133533334432e-05, "loss": 0.1799011707305908, "step": 47800 }, { "epoch": 0.23747553718844064, "grad_norm": 1.8288627862930298, "learning_rate": 1.7826132197016715e-05, "loss": 0.17297425270080566, "step": 47810 }, { "epoch": 0.2375252078718099, "grad_norm": 2.3831331729888916, "learning_rate": 1.7825130658161844e-05, "loss": 0.18539180755615234, "step": 47820 }, { "epoch": 0.23757487855517917, "grad_norm": 2.108140707015991, "learning_rate": 1.7824128916795737e-05, "loss": 0.18460912704467775, "step": 47830 }, { "epoch": 0.23762454923854842, "grad_norm": 1.780348777770996, "learning_rate": 1.7823126972944324e-05, "loss": 0.1852652311325073, "step": 47840 }, { "epoch": 0.23767421992191767, "grad_norm": 0.654855489730835, "learning_rate": 1.7822124826633527e-05, "loss": 0.18119707107543945, "step": 47850 }, { "epoch": 0.23772389060528695, "grad_norm": 2.2478878498077393, "learning_rate": 1.7821122477889282e-05, "loss": 0.18589602708816527, "step": 47860 }, { "epoch": 0.2377735612886562, "grad_norm": 0.7843212485313416, "learning_rate": 1.7820119926737536e-05, "loss": 0.17119460105895995, "step": 47870 }, { "epoch": 0.23782323197202548, "grad_norm": 1.8143436908721924, "learning_rate": 1.7819117173204228e-05, "loss": 0.17103344202041626, "step": 47880 }, { "epoch": 0.23787290265539474, "grad_norm": 0.9035488963127136, "learning_rate": 1.7818114217315315e-05, "loss": 0.1725285530090332, "step": 47890 }, { "epoch": 0.237922573338764, "grad_norm": 1.3265986442565918, "learning_rate": 1.7817111059096746e-05, "loss": 0.18468393087387086, "step": 47900 }, { "epoch": 0.23797224402213327, "grad_norm": 1.7189923524856567, "learning_rate": 1.781610769857449e-05, "loss": 0.1695570707321167, "step": 47910 }, { "epoch": 0.23802191470550252, "grad_norm": 1.725034236907959, "learning_rate": 1.7815104135774505e-05, "loss": 0.16789753437042237, "step": 47920 }, { "epoch": 0.23807158538887177, "grad_norm": 1.397137999534607, "learning_rate": 1.7814100370722766e-05, "loss": 0.16231656074523926, "step": 47930 }, { "epoch": 0.23812125607224105, "grad_norm": 2.223043918609619, "learning_rate": 1.781309640344526e-05, "loss": 0.18144736289978028, "step": 47940 }, { "epoch": 0.2381709267556103, "grad_norm": 0.5602142810821533, "learning_rate": 1.781209223396795e-05, "loss": 0.18185112476348878, "step": 47950 }, { "epoch": 0.23822059743897955, "grad_norm": 1.4144859313964844, "learning_rate": 1.781108786231684e-05, "loss": 0.16270190477371216, "step": 47960 }, { "epoch": 0.23827026812234883, "grad_norm": 1.1047545671463013, "learning_rate": 1.7810083288517915e-05, "loss": 0.18612773418426515, "step": 47970 }, { "epoch": 0.23831993880571808, "grad_norm": 1.5181472301483154, "learning_rate": 1.7809078512597175e-05, "loss": 0.17931315898895264, "step": 47980 }, { "epoch": 0.23836960948908736, "grad_norm": 1.02468740940094, "learning_rate": 1.7808073534580626e-05, "loss": 0.18087031841278076, "step": 47990 }, { "epoch": 0.23841928017245662, "grad_norm": 0.9814710021018982, "learning_rate": 1.780706835449427e-05, "loss": 0.17508358955383302, "step": 48000 }, { "epoch": 0.23846895085582587, "grad_norm": 1.8699185848236084, "learning_rate": 1.7806062972364128e-05, "loss": 0.17440894842147828, "step": 48010 }, { "epoch": 0.23851862153919515, "grad_norm": 1.2264173030853271, "learning_rate": 1.7805057388216212e-05, "loss": 0.16855651140213013, "step": 48020 }, { "epoch": 0.2385682922225644, "grad_norm": 1.174498438835144, "learning_rate": 1.7804051602076545e-05, "loss": 0.17060682773590088, "step": 48030 }, { "epoch": 0.23861796290593365, "grad_norm": 1.2153517007827759, "learning_rate": 1.7803045613971166e-05, "loss": 0.17922779321670532, "step": 48040 }, { "epoch": 0.23866763358930293, "grad_norm": 1.0080029964447021, "learning_rate": 1.7802039423926102e-05, "loss": 0.18709259033203124, "step": 48050 }, { "epoch": 0.23871730427267218, "grad_norm": 1.7915207147598267, "learning_rate": 1.7801033031967398e-05, "loss": 0.1741676688194275, "step": 48060 }, { "epoch": 0.23876697495604143, "grad_norm": 1.7011574506759644, "learning_rate": 1.780002643812109e-05, "loss": 0.178487229347229, "step": 48070 }, { "epoch": 0.2388166456394107, "grad_norm": 1.4584875106811523, "learning_rate": 1.779901964241324e-05, "loss": 0.17481845617294312, "step": 48080 }, { "epoch": 0.23886631632277996, "grad_norm": 1.5526902675628662, "learning_rate": 1.7798012644869894e-05, "loss": 0.18683032989501952, "step": 48090 }, { "epoch": 0.23891598700614924, "grad_norm": 1.896713376045227, "learning_rate": 1.7797005445517114e-05, "loss": 0.17314008474349976, "step": 48100 }, { "epoch": 0.2389656576895185, "grad_norm": 0.8627037405967712, "learning_rate": 1.7795998044380974e-05, "loss": 0.17858222723007203, "step": 48110 }, { "epoch": 0.23901532837288775, "grad_norm": 1.0392893552780151, "learning_rate": 1.7794990441487535e-05, "loss": 0.16875689029693602, "step": 48120 }, { "epoch": 0.23906499905625703, "grad_norm": 1.7747679948806763, "learning_rate": 1.779398263686288e-05, "loss": 0.17650277614593507, "step": 48130 }, { "epoch": 0.23911466973962628, "grad_norm": 1.466400384902954, "learning_rate": 1.7792974630533087e-05, "loss": 0.18105227947235109, "step": 48140 }, { "epoch": 0.23916434042299553, "grad_norm": 0.5765471458435059, "learning_rate": 1.7791966422524244e-05, "loss": 0.17628281116485595, "step": 48150 }, { "epoch": 0.2392140111063648, "grad_norm": 1.1785564422607422, "learning_rate": 1.7790958012862444e-05, "loss": 0.18395960330963135, "step": 48160 }, { "epoch": 0.23926368178973406, "grad_norm": 0.8795657753944397, "learning_rate": 1.778994940157378e-05, "loss": 0.17109527587890624, "step": 48170 }, { "epoch": 0.2393133524731033, "grad_norm": 0.6760841012001038, "learning_rate": 1.778894058868436e-05, "loss": 0.16962206363677979, "step": 48180 }, { "epoch": 0.2393630231564726, "grad_norm": 0.7446267604827881, "learning_rate": 1.7787931574220292e-05, "loss": 0.17845333814620973, "step": 48190 }, { "epoch": 0.23941269383984184, "grad_norm": 1.5542683601379395, "learning_rate": 1.7786922358207683e-05, "loss": 0.17692445516586303, "step": 48200 }, { "epoch": 0.23946236452321112, "grad_norm": 1.2805362939834595, "learning_rate": 1.7785912940672654e-05, "loss": 0.18695240020751952, "step": 48210 }, { "epoch": 0.23951203520658038, "grad_norm": 2.2399978637695312, "learning_rate": 1.778490332164133e-05, "loss": 0.1598205804824829, "step": 48220 }, { "epoch": 0.23956170588994963, "grad_norm": 1.3876107931137085, "learning_rate": 1.778389350113984e-05, "loss": 0.17901936769485474, "step": 48230 }, { "epoch": 0.2396113765733189, "grad_norm": 1.1620014905929565, "learning_rate": 1.778288347919431e-05, "loss": 0.18059589862823486, "step": 48240 }, { "epoch": 0.23966104725668816, "grad_norm": 1.5162588357925415, "learning_rate": 1.7781873255830892e-05, "loss": 0.16345973014831544, "step": 48250 }, { "epoch": 0.2397107179400574, "grad_norm": 1.3965415954589844, "learning_rate": 1.7780862831075718e-05, "loss": 0.19132637977600098, "step": 48260 }, { "epoch": 0.2397603886234267, "grad_norm": 0.8569199442863464, "learning_rate": 1.7779852204954944e-05, "loss": 0.18713048696517945, "step": 48270 }, { "epoch": 0.23981005930679594, "grad_norm": 1.923646330833435, "learning_rate": 1.7778841377494723e-05, "loss": 0.17399172782897948, "step": 48280 }, { "epoch": 0.2398597299901652, "grad_norm": 2.0699737071990967, "learning_rate": 1.7777830348721218e-05, "loss": 0.17820212841033936, "step": 48290 }, { "epoch": 0.23990940067353447, "grad_norm": 1.9917268753051758, "learning_rate": 1.7776819118660586e-05, "loss": 0.16966922283172609, "step": 48300 }, { "epoch": 0.23995907135690372, "grad_norm": 1.2907532453536987, "learning_rate": 1.7775807687339006e-05, "loss": 0.1718686819076538, "step": 48310 }, { "epoch": 0.24000874204027298, "grad_norm": 0.9227108955383301, "learning_rate": 1.7774796054782646e-05, "loss": 0.17460041046142577, "step": 48320 }, { "epoch": 0.24005841272364226, "grad_norm": 1.8241791725158691, "learning_rate": 1.7773784221017695e-05, "loss": 0.17636513710021973, "step": 48330 }, { "epoch": 0.2401080834070115, "grad_norm": 1.264638066291809, "learning_rate": 1.777277218607033e-05, "loss": 0.16399636268615722, "step": 48340 }, { "epoch": 0.2401577540903808, "grad_norm": 0.896763265132904, "learning_rate": 1.777175994996675e-05, "loss": 0.17582302093505858, "step": 48350 }, { "epoch": 0.24020742477375004, "grad_norm": 0.9779313802719116, "learning_rate": 1.777074751273314e-05, "loss": 0.1766696810722351, "step": 48360 }, { "epoch": 0.2402570954571193, "grad_norm": 1.1199345588684082, "learning_rate": 1.7769734874395713e-05, "loss": 0.19907552003860474, "step": 48370 }, { "epoch": 0.24030676614048857, "grad_norm": 1.6287065744400024, "learning_rate": 1.7768722034980673e-05, "loss": 0.1616360664367676, "step": 48380 }, { "epoch": 0.24035643682385782, "grad_norm": 1.2508540153503418, "learning_rate": 1.7767708994514232e-05, "loss": 0.18044664859771728, "step": 48390 }, { "epoch": 0.24040610750722707, "grad_norm": 2.475720167160034, "learning_rate": 1.7766695753022604e-05, "loss": 0.17810959815979005, "step": 48400 }, { "epoch": 0.24045577819059635, "grad_norm": 0.9572844505310059, "learning_rate": 1.7765682310532008e-05, "loss": 0.1815878391265869, "step": 48410 }, { "epoch": 0.2405054488739656, "grad_norm": 1.2005406618118286, "learning_rate": 1.7764668667068684e-05, "loss": 0.17198572158813477, "step": 48420 }, { "epoch": 0.24055511955733486, "grad_norm": 1.369799256324768, "learning_rate": 1.776365482265885e-05, "loss": 0.1679742693901062, "step": 48430 }, { "epoch": 0.24060479024070414, "grad_norm": 1.2682186365127563, "learning_rate": 1.7762640777328754e-05, "loss": 0.18503694534301757, "step": 48440 }, { "epoch": 0.2406544609240734, "grad_norm": 1.1880561113357544, "learning_rate": 1.7761626531104637e-05, "loss": 0.17676715850830077, "step": 48450 }, { "epoch": 0.24070413160744267, "grad_norm": 1.0115656852722168, "learning_rate": 1.7760612084012746e-05, "loss": 0.1802980422973633, "step": 48460 }, { "epoch": 0.24075380229081192, "grad_norm": 0.6948521733283997, "learning_rate": 1.775959743607933e-05, "loss": 0.17339744567871093, "step": 48470 }, { "epoch": 0.24080347297418117, "grad_norm": 2.7248120307922363, "learning_rate": 1.7758582587330657e-05, "loss": 0.17896888256072999, "step": 48480 }, { "epoch": 0.24085314365755045, "grad_norm": 2.753190517425537, "learning_rate": 1.7757567537792986e-05, "loss": 0.16771464347839354, "step": 48490 }, { "epoch": 0.2409028143409197, "grad_norm": 1.181265115737915, "learning_rate": 1.775655228749258e-05, "loss": 0.1679179549217224, "step": 48500 }, { "epoch": 0.24095248502428895, "grad_norm": 0.9641768932342529, "learning_rate": 1.775553683645573e-05, "loss": 0.16915762424468994, "step": 48510 }, { "epoch": 0.24100215570765823, "grad_norm": 2.010080337524414, "learning_rate": 1.7754521184708702e-05, "loss": 0.17529993057250975, "step": 48520 }, { "epoch": 0.24105182639102748, "grad_norm": 0.793286144733429, "learning_rate": 1.775350533227778e-05, "loss": 0.18067264556884766, "step": 48530 }, { "epoch": 0.24110149707439674, "grad_norm": 0.6747002601623535, "learning_rate": 1.775248927918926e-05, "loss": 0.17044199705123902, "step": 48540 }, { "epoch": 0.24115116775776602, "grad_norm": 1.5193427801132202, "learning_rate": 1.7751473025469434e-05, "loss": 0.1674313426017761, "step": 48550 }, { "epoch": 0.24120083844113527, "grad_norm": 1.7424707412719727, "learning_rate": 1.7750456571144604e-05, "loss": 0.17288603782653808, "step": 48560 }, { "epoch": 0.24125050912450455, "grad_norm": 1.2804406881332397, "learning_rate": 1.7749439916241072e-05, "loss": 0.17499930858612062, "step": 48570 }, { "epoch": 0.2413001798078738, "grad_norm": 1.1480607986450195, "learning_rate": 1.774842306078515e-05, "loss": 0.17852045297622682, "step": 48580 }, { "epoch": 0.24134985049124305, "grad_norm": 0.8550298810005188, "learning_rate": 1.774740600480316e-05, "loss": 0.17272849082946778, "step": 48590 }, { "epoch": 0.24139952117461233, "grad_norm": 0.7777416706085205, "learning_rate": 1.7746388748321415e-05, "loss": 0.17616758346557618, "step": 48600 }, { "epoch": 0.24144919185798158, "grad_norm": 2.207674026489258, "learning_rate": 1.7745371291366244e-05, "loss": 0.1760564923286438, "step": 48610 }, { "epoch": 0.24149886254135083, "grad_norm": 1.1871633529663086, "learning_rate": 1.7744353633963975e-05, "loss": 0.1724345088005066, "step": 48620 }, { "epoch": 0.2415485332247201, "grad_norm": 1.0416947603225708, "learning_rate": 1.7743335776140947e-05, "loss": 0.16566176414489747, "step": 48630 }, { "epoch": 0.24159820390808937, "grad_norm": 2.5108649730682373, "learning_rate": 1.774231771792351e-05, "loss": 0.17902344465255737, "step": 48640 }, { "epoch": 0.24164787459145862, "grad_norm": 0.937437117099762, "learning_rate": 1.7741299459337998e-05, "loss": 0.1843884229660034, "step": 48650 }, { "epoch": 0.2416975452748279, "grad_norm": 1.2415798902511597, "learning_rate": 1.7740281000410772e-05, "loss": 0.17869865894317627, "step": 48660 }, { "epoch": 0.24174721595819715, "grad_norm": 1.2455294132232666, "learning_rate": 1.7739262341168185e-05, "loss": 0.16358635425567628, "step": 48670 }, { "epoch": 0.24179688664156643, "grad_norm": 0.7292351126670837, "learning_rate": 1.7738243481636598e-05, "loss": 0.1887240767478943, "step": 48680 }, { "epoch": 0.24184655732493568, "grad_norm": 0.9279763698577881, "learning_rate": 1.773722442184238e-05, "loss": 0.1700849413871765, "step": 48690 }, { "epoch": 0.24189622800830493, "grad_norm": 0.5740818977355957, "learning_rate": 1.773620516181191e-05, "loss": 0.16871739625930787, "step": 48700 }, { "epoch": 0.2419458986916742, "grad_norm": 1.336087942123413, "learning_rate": 1.773518570157156e-05, "loss": 0.17238149642944336, "step": 48710 }, { "epoch": 0.24199556937504346, "grad_norm": 1.7434306144714355, "learning_rate": 1.7734166041147713e-05, "loss": 0.17370153665542604, "step": 48720 }, { "epoch": 0.24204524005841271, "grad_norm": 1.6581852436065674, "learning_rate": 1.7733146180566757e-05, "loss": 0.1823193311691284, "step": 48730 }, { "epoch": 0.242094910741782, "grad_norm": 1.6292133331298828, "learning_rate": 1.7732126119855088e-05, "loss": 0.1734668493270874, "step": 48740 }, { "epoch": 0.24214458142515125, "grad_norm": 1.429353952407837, "learning_rate": 1.7731105859039105e-05, "loss": 0.17925947904586792, "step": 48750 }, { "epoch": 0.2421942521085205, "grad_norm": 1.877315878868103, "learning_rate": 1.773008539814521e-05, "loss": 0.19171861410140992, "step": 48760 }, { "epoch": 0.24224392279188978, "grad_norm": 2.102640151977539, "learning_rate": 1.7729064737199816e-05, "loss": 0.16847124099731445, "step": 48770 }, { "epoch": 0.24229359347525903, "grad_norm": 1.3739110231399536, "learning_rate": 1.772804387622933e-05, "loss": 0.19427101612091063, "step": 48780 }, { "epoch": 0.2423432641586283, "grad_norm": 0.5692094564437866, "learning_rate": 1.772702281526018e-05, "loss": 0.1717463731765747, "step": 48790 }, { "epoch": 0.24239293484199756, "grad_norm": 2.103849172592163, "learning_rate": 1.7726001554318783e-05, "loss": 0.1879819869995117, "step": 48800 }, { "epoch": 0.2424426055253668, "grad_norm": 2.445363998413086, "learning_rate": 1.7724980093431572e-05, "loss": 0.1790684938430786, "step": 48810 }, { "epoch": 0.2424922762087361, "grad_norm": 1.1064246892929077, "learning_rate": 1.7723958432624985e-05, "loss": 0.17929083108901978, "step": 48820 }, { "epoch": 0.24254194689210534, "grad_norm": 1.415570855140686, "learning_rate": 1.7722936571925456e-05, "loss": 0.17379767894744874, "step": 48830 }, { "epoch": 0.2425916175754746, "grad_norm": 0.7629646062850952, "learning_rate": 1.7721914511359437e-05, "loss": 0.17134501934051513, "step": 48840 }, { "epoch": 0.24264128825884387, "grad_norm": 1.2742782831192017, "learning_rate": 1.7720892250953375e-05, "loss": 0.17944591045379638, "step": 48850 }, { "epoch": 0.24269095894221313, "grad_norm": 2.1631650924682617, "learning_rate": 1.7719869790733725e-05, "loss": 0.17723467350006103, "step": 48860 }, { "epoch": 0.24274062962558238, "grad_norm": 0.8716935515403748, "learning_rate": 1.7718847130726948e-05, "loss": 0.19073951244354248, "step": 48870 }, { "epoch": 0.24279030030895166, "grad_norm": 2.247040033340454, "learning_rate": 1.771782427095951e-05, "loss": 0.18430693149566652, "step": 48880 }, { "epoch": 0.2428399709923209, "grad_norm": 1.891047716140747, "learning_rate": 1.7716801211457888e-05, "loss": 0.1727602243423462, "step": 48890 }, { "epoch": 0.2428896416756902, "grad_norm": 0.980623722076416, "learning_rate": 1.7715777952248552e-05, "loss": 0.17139801979064942, "step": 48900 }, { "epoch": 0.24293931235905944, "grad_norm": 2.29045033454895, "learning_rate": 1.7714754493357982e-05, "loss": 0.16613852977752686, "step": 48910 }, { "epoch": 0.2429889830424287, "grad_norm": 1.4500702619552612, "learning_rate": 1.771373083481267e-05, "loss": 0.1731210470199585, "step": 48920 }, { "epoch": 0.24303865372579797, "grad_norm": 1.2040036916732788, "learning_rate": 1.7712706976639106e-05, "loss": 0.169790780544281, "step": 48930 }, { "epoch": 0.24308832440916722, "grad_norm": 1.5899038314819336, "learning_rate": 1.7711682918863783e-05, "loss": 0.16245661973953246, "step": 48940 }, { "epoch": 0.24313799509253647, "grad_norm": 1.2386646270751953, "learning_rate": 1.771065866151321e-05, "loss": 0.18808770179748535, "step": 48950 }, { "epoch": 0.24318766577590575, "grad_norm": 1.5293805599212646, "learning_rate": 1.770963420461389e-05, "loss": 0.16612663269042968, "step": 48960 }, { "epoch": 0.243237336459275, "grad_norm": 0.8625050783157349, "learning_rate": 1.7708609548192337e-05, "loss": 0.1816478490829468, "step": 48970 }, { "epoch": 0.24328700714264426, "grad_norm": 2.2618143558502197, "learning_rate": 1.7707584692275067e-05, "loss": 0.17828258275985717, "step": 48980 }, { "epoch": 0.24333667782601354, "grad_norm": 1.4699453115463257, "learning_rate": 1.770655963688861e-05, "loss": 0.1579385757446289, "step": 48990 }, { "epoch": 0.2433863485093828, "grad_norm": 1.347562313079834, "learning_rate": 1.770553438205948e-05, "loss": 0.16691040992736816, "step": 49000 }, { "epoch": 0.24343601919275204, "grad_norm": 0.8984689116477966, "learning_rate": 1.770450892781422e-05, "loss": 0.1645340085029602, "step": 49010 }, { "epoch": 0.24348568987612132, "grad_norm": 1.2701646089553833, "learning_rate": 1.7703483274179372e-05, "loss": 0.18614412546157838, "step": 49020 }, { "epoch": 0.24353536055949057, "grad_norm": 0.5642551183700562, "learning_rate": 1.7702457421181473e-05, "loss": 0.17996424436569214, "step": 49030 }, { "epoch": 0.24358503124285985, "grad_norm": 0.9458100199699402, "learning_rate": 1.770143136884707e-05, "loss": 0.17218599319458008, "step": 49040 }, { "epoch": 0.2436347019262291, "grad_norm": 0.8329864144325256, "learning_rate": 1.770040511720272e-05, "loss": 0.1709383487701416, "step": 49050 }, { "epoch": 0.24368437260959835, "grad_norm": 0.8623117804527283, "learning_rate": 1.7699378666274983e-05, "loss": 0.1810229778289795, "step": 49060 }, { "epoch": 0.24373404329296763, "grad_norm": 1.1910231113433838, "learning_rate": 1.7698352016090423e-05, "loss": 0.18491580486297607, "step": 49070 }, { "epoch": 0.24378371397633689, "grad_norm": 2.9127416610717773, "learning_rate": 1.769732516667561e-05, "loss": 0.18014297485351563, "step": 49080 }, { "epoch": 0.24383338465970614, "grad_norm": 1.1698644161224365, "learning_rate": 1.769629811805711e-05, "loss": 0.17842352390289307, "step": 49090 }, { "epoch": 0.24388305534307542, "grad_norm": 1.6768401861190796, "learning_rate": 1.7695270870261512e-05, "loss": 0.17605890035629274, "step": 49100 }, { "epoch": 0.24393272602644467, "grad_norm": 1.110048532485962, "learning_rate": 1.76942434233154e-05, "loss": 0.17206062078475953, "step": 49110 }, { "epoch": 0.24398239670981392, "grad_norm": 1.445678472518921, "learning_rate": 1.7693215777245364e-05, "loss": 0.17545416355133056, "step": 49120 }, { "epoch": 0.2440320673931832, "grad_norm": 1.1030197143554688, "learning_rate": 1.7692187932077994e-05, "loss": 0.16984617710113525, "step": 49130 }, { "epoch": 0.24408173807655245, "grad_norm": 1.7627842426300049, "learning_rate": 1.7691159887839893e-05, "loss": 0.1747964382171631, "step": 49140 }, { "epoch": 0.24413140875992173, "grad_norm": 0.9893549680709839, "learning_rate": 1.7690131644557666e-05, "loss": 0.16868119239807128, "step": 49150 }, { "epoch": 0.24418107944329098, "grad_norm": 2.959904670715332, "learning_rate": 1.7689103202257923e-05, "loss": 0.18020150661468506, "step": 49160 }, { "epoch": 0.24423075012666023, "grad_norm": 0.9129132032394409, "learning_rate": 1.7688074560967285e-05, "loss": 0.1735006093978882, "step": 49170 }, { "epoch": 0.2442804208100295, "grad_norm": 1.3926798105239868, "learning_rate": 1.7687045720712364e-05, "loss": 0.17933237552642822, "step": 49180 }, { "epoch": 0.24433009149339877, "grad_norm": 1.1257630586624146, "learning_rate": 1.7686016681519796e-05, "loss": 0.18086600303649902, "step": 49190 }, { "epoch": 0.24437976217676802, "grad_norm": 1.225740909576416, "learning_rate": 1.76849874434162e-05, "loss": 0.17397828102111818, "step": 49200 }, { "epoch": 0.2444294328601373, "grad_norm": 2.1844568252563477, "learning_rate": 1.7683958006428225e-05, "loss": 0.18291165828704833, "step": 49210 }, { "epoch": 0.24447910354350655, "grad_norm": 1.4003318548202515, "learning_rate": 1.76829283705825e-05, "loss": 0.18392839431762695, "step": 49220 }, { "epoch": 0.2445287742268758, "grad_norm": 0.6833970546722412, "learning_rate": 1.7681898535905686e-05, "loss": 0.17852294445037842, "step": 49230 }, { "epoch": 0.24457844491024508, "grad_norm": 1.2930560111999512, "learning_rate": 1.768086850242442e-05, "loss": 0.1783290386199951, "step": 49240 }, { "epoch": 0.24462811559361433, "grad_norm": 2.316624402999878, "learning_rate": 1.7679838270165367e-05, "loss": 0.16155714988708497, "step": 49250 }, { "epoch": 0.2446777862769836, "grad_norm": 1.8399741649627686, "learning_rate": 1.7678807839155192e-05, "loss": 0.1800441026687622, "step": 49260 }, { "epoch": 0.24472745696035286, "grad_norm": 1.5479929447174072, "learning_rate": 1.7677777209420554e-05, "loss": 0.167833411693573, "step": 49270 }, { "epoch": 0.24477712764372211, "grad_norm": 1.9898604154586792, "learning_rate": 1.7676746380988127e-05, "loss": 0.17370529174804689, "step": 49280 }, { "epoch": 0.2448267983270914, "grad_norm": 2.4309403896331787, "learning_rate": 1.7675715353884594e-05, "loss": 0.1658182144165039, "step": 49290 }, { "epoch": 0.24487646901046065, "grad_norm": 1.1003690958023071, "learning_rate": 1.7674684128136632e-05, "loss": 0.18323757648468017, "step": 49300 }, { "epoch": 0.2449261396938299, "grad_norm": 1.343860149383545, "learning_rate": 1.7673652703770933e-05, "loss": 0.16260088682174684, "step": 49310 }, { "epoch": 0.24497581037719918, "grad_norm": 0.9478362202644348, "learning_rate": 1.7672621080814182e-05, "loss": 0.1862495183944702, "step": 49320 }, { "epoch": 0.24502548106056843, "grad_norm": 2.3724138736724854, "learning_rate": 1.7671589259293087e-05, "loss": 0.17967758178710938, "step": 49330 }, { "epoch": 0.24507515174393768, "grad_norm": 1.596358299255371, "learning_rate": 1.7670557239234348e-05, "loss": 0.19139010906219484, "step": 49340 }, { "epoch": 0.24512482242730696, "grad_norm": 0.706757664680481, "learning_rate": 1.7669525020664666e-05, "loss": 0.16052308082580566, "step": 49350 }, { "epoch": 0.2451744931106762, "grad_norm": 0.9338663220405579, "learning_rate": 1.7668492603610766e-05, "loss": 0.18248369693756103, "step": 49360 }, { "epoch": 0.2452241637940455, "grad_norm": 0.8147158026695251, "learning_rate": 1.7667459988099357e-05, "loss": 0.16408385038375856, "step": 49370 }, { "epoch": 0.24527383447741474, "grad_norm": 1.064975619316101, "learning_rate": 1.7666427174157168e-05, "loss": 0.17472782135009765, "step": 49380 }, { "epoch": 0.245323505160784, "grad_norm": 1.668770432472229, "learning_rate": 1.7665394161810925e-05, "loss": 0.16580357551574706, "step": 49390 }, { "epoch": 0.24537317584415327, "grad_norm": 0.883210301399231, "learning_rate": 1.7664360951087367e-05, "loss": 0.18523701429367065, "step": 49400 }, { "epoch": 0.24542284652752253, "grad_norm": 0.856276273727417, "learning_rate": 1.7663327542013227e-05, "loss": 0.18227095603942872, "step": 49410 }, { "epoch": 0.24547251721089178, "grad_norm": 1.189239501953125, "learning_rate": 1.766229393461525e-05, "loss": 0.18360064029693604, "step": 49420 }, { "epoch": 0.24552218789426106, "grad_norm": 1.0846723318099976, "learning_rate": 1.7661260128920187e-05, "loss": 0.18620750904083253, "step": 49430 }, { "epoch": 0.2455718585776303, "grad_norm": 1.8865597248077393, "learning_rate": 1.7660226124954794e-05, "loss": 0.17809033393859863, "step": 49440 }, { "epoch": 0.24562152926099956, "grad_norm": 0.7662704586982727, "learning_rate": 1.7659191922745828e-05, "loss": 0.1790260672569275, "step": 49450 }, { "epoch": 0.24567119994436884, "grad_norm": 1.8341701030731201, "learning_rate": 1.7658157522320057e-05, "loss": 0.1665790557861328, "step": 49460 }, { "epoch": 0.2457208706277381, "grad_norm": 1.4558477401733398, "learning_rate": 1.7657122923704246e-05, "loss": 0.17579987049102783, "step": 49470 }, { "epoch": 0.24577054131110737, "grad_norm": 1.042259931564331, "learning_rate": 1.7656088126925172e-05, "loss": 0.16908233165740966, "step": 49480 }, { "epoch": 0.24582021199447662, "grad_norm": 0.6903946399688721, "learning_rate": 1.765505313200962e-05, "loss": 0.16946299076080323, "step": 49490 }, { "epoch": 0.24586988267784587, "grad_norm": 2.005406141281128, "learning_rate": 1.7654017938984364e-05, "loss": 0.18680591583251954, "step": 49500 }, { "epoch": 0.24591955336121515, "grad_norm": 1.037939190864563, "learning_rate": 1.7652982547876204e-05, "loss": 0.17720458507537842, "step": 49510 }, { "epoch": 0.2459692240445844, "grad_norm": 0.794931173324585, "learning_rate": 1.7651946958711933e-05, "loss": 0.16913472414016723, "step": 49520 }, { "epoch": 0.24601889472795366, "grad_norm": 1.633133888244629, "learning_rate": 1.7650911171518355e-05, "loss": 0.1706732153892517, "step": 49530 }, { "epoch": 0.24606856541132294, "grad_norm": 1.4240535497665405, "learning_rate": 1.7649875186322268e-05, "loss": 0.17421603202819824, "step": 49540 }, { "epoch": 0.2461182360946922, "grad_norm": 1.071794033050537, "learning_rate": 1.764883900315049e-05, "loss": 0.17537071704864501, "step": 49550 }, { "epoch": 0.24616790677806144, "grad_norm": 3.3488082885742188, "learning_rate": 1.7647802622029832e-05, "loss": 0.17341139316558837, "step": 49560 }, { "epoch": 0.24621757746143072, "grad_norm": 2.125980854034424, "learning_rate": 1.7646766042987117e-05, "loss": 0.16544814109802247, "step": 49570 }, { "epoch": 0.24626724814479997, "grad_norm": 0.8083221912384033, "learning_rate": 1.764572926604917e-05, "loss": 0.17547707557678222, "step": 49580 }, { "epoch": 0.24631691882816925, "grad_norm": 0.5520002841949463, "learning_rate": 1.7644692291242828e-05, "loss": 0.17056496143341066, "step": 49590 }, { "epoch": 0.2463665895115385, "grad_norm": 1.2613683938980103, "learning_rate": 1.764365511859492e-05, "loss": 0.1724246025085449, "step": 49600 }, { "epoch": 0.24641626019490775, "grad_norm": 1.432342767715454, "learning_rate": 1.7642617748132292e-05, "loss": 0.17506290674209596, "step": 49610 }, { "epoch": 0.24646593087827703, "grad_norm": 1.3968864679336548, "learning_rate": 1.764158017988179e-05, "loss": 0.180646812915802, "step": 49620 }, { "epoch": 0.24651560156164629, "grad_norm": 2.7820489406585693, "learning_rate": 1.764054241387027e-05, "loss": 0.17722556591033936, "step": 49630 }, { "epoch": 0.24656527224501554, "grad_norm": 1.7254332304000854, "learning_rate": 1.7639504450124577e-05, "loss": 0.1775486946105957, "step": 49640 }, { "epoch": 0.24661494292838482, "grad_norm": 1.2159315347671509, "learning_rate": 1.7638466288671584e-05, "loss": 0.17690527439117432, "step": 49650 }, { "epoch": 0.24666461361175407, "grad_norm": 1.2130143642425537, "learning_rate": 1.7637427929538153e-05, "loss": 0.1764456510543823, "step": 49660 }, { "epoch": 0.24671428429512332, "grad_norm": 0.7068045735359192, "learning_rate": 1.763638937275116e-05, "loss": 0.1700157880783081, "step": 49670 }, { "epoch": 0.2467639549784926, "grad_norm": 0.5920352935791016, "learning_rate": 1.7635350618337477e-05, "loss": 0.16252737045288085, "step": 49680 }, { "epoch": 0.24681362566186185, "grad_norm": 0.8512648940086365, "learning_rate": 1.7634311666323992e-05, "loss": 0.17622977495193481, "step": 49690 }, { "epoch": 0.24686329634523113, "grad_norm": 0.9744777679443359, "learning_rate": 1.7633272516737595e-05, "loss": 0.17714033126831055, "step": 49700 }, { "epoch": 0.24691296702860038, "grad_norm": 1.0148793458938599, "learning_rate": 1.7632233169605172e-05, "loss": 0.17744081020355223, "step": 49710 }, { "epoch": 0.24696263771196963, "grad_norm": 1.8172980546951294, "learning_rate": 1.763119362495362e-05, "loss": 0.16452515125274658, "step": 49720 }, { "epoch": 0.24701230839533891, "grad_norm": 0.7671515941619873, "learning_rate": 1.7630153882809848e-05, "loss": 0.17106550931930542, "step": 49730 }, { "epoch": 0.24706197907870817, "grad_norm": 0.9949701428413391, "learning_rate": 1.762911394320076e-05, "loss": 0.18680623769760132, "step": 49740 }, { "epoch": 0.24711164976207742, "grad_norm": 1.311637282371521, "learning_rate": 1.762807380615327e-05, "loss": 0.1636354923248291, "step": 49750 }, { "epoch": 0.2471613204454467, "grad_norm": 1.0363259315490723, "learning_rate": 1.76270334716943e-05, "loss": 0.1811394453048706, "step": 49760 }, { "epoch": 0.24721099112881595, "grad_norm": 1.088267207145691, "learning_rate": 1.7625992939850772e-05, "loss": 0.19372355937957764, "step": 49770 }, { "epoch": 0.2472606618121852, "grad_norm": 1.6225504875183105, "learning_rate": 1.7624952210649606e-05, "loss": 0.17817788124084472, "step": 49780 }, { "epoch": 0.24731033249555448, "grad_norm": 1.5049686431884766, "learning_rate": 1.7623911284117747e-05, "loss": 0.1694354772567749, "step": 49790 }, { "epoch": 0.24736000317892373, "grad_norm": 1.9183231592178345, "learning_rate": 1.7622870160282132e-05, "loss": 0.18496835231781006, "step": 49800 }, { "epoch": 0.24740967386229298, "grad_norm": 1.7550849914550781, "learning_rate": 1.7621828839169698e-05, "loss": 0.17860174179077148, "step": 49810 }, { "epoch": 0.24745934454566226, "grad_norm": 1.002449631690979, "learning_rate": 1.76207873208074e-05, "loss": 0.19028446674346924, "step": 49820 }, { "epoch": 0.24750901522903151, "grad_norm": 1.0580389499664307, "learning_rate": 1.761974560522219e-05, "loss": 0.18072307109832764, "step": 49830 }, { "epoch": 0.2475586859124008, "grad_norm": 1.4862558841705322, "learning_rate": 1.761870369244103e-05, "loss": 0.19109463691711426, "step": 49840 }, { "epoch": 0.24760835659577005, "grad_norm": 0.7258725166320801, "learning_rate": 1.761766158249088e-05, "loss": 0.179237961769104, "step": 49850 }, { "epoch": 0.2476580272791393, "grad_norm": 2.6548216342926025, "learning_rate": 1.761661927539871e-05, "loss": 0.16723709106445311, "step": 49860 }, { "epoch": 0.24770769796250858, "grad_norm": 1.5586034059524536, "learning_rate": 1.7615576771191496e-05, "loss": 0.1713970422744751, "step": 49870 }, { "epoch": 0.24775736864587783, "grad_norm": 0.7808062434196472, "learning_rate": 1.7614534069896215e-05, "loss": 0.1912083625793457, "step": 49880 }, { "epoch": 0.24780703932924708, "grad_norm": 1.1633729934692383, "learning_rate": 1.7613491171539858e-05, "loss": 0.1711850047111511, "step": 49890 }, { "epoch": 0.24785671001261636, "grad_norm": 1.745976448059082, "learning_rate": 1.761244807614941e-05, "loss": 0.16974482536315919, "step": 49900 }, { "epoch": 0.2479063806959856, "grad_norm": 0.48783519864082336, "learning_rate": 1.7611404783751864e-05, "loss": 0.17002902030944825, "step": 49910 }, { "epoch": 0.24795605137935486, "grad_norm": 0.9954911470413208, "learning_rate": 1.7610361294374225e-05, "loss": 0.1699911594390869, "step": 49920 }, { "epoch": 0.24800572206272414, "grad_norm": 1.0160744190216064, "learning_rate": 1.7609317608043496e-05, "loss": 0.16737087965011596, "step": 49930 }, { "epoch": 0.2480553927460934, "grad_norm": 0.7896813154220581, "learning_rate": 1.7608273724786687e-05, "loss": 0.17296454906463624, "step": 49940 }, { "epoch": 0.24810506342946267, "grad_norm": 1.4041334390640259, "learning_rate": 1.760722964463081e-05, "loss": 0.1707387685775757, "step": 49950 }, { "epoch": 0.24815473411283193, "grad_norm": 0.8453493118286133, "learning_rate": 1.7606185367602894e-05, "loss": 0.1875759482383728, "step": 49960 }, { "epoch": 0.24820440479620118, "grad_norm": 1.1158989667892456, "learning_rate": 1.7605140893729956e-05, "loss": 0.17972851991653443, "step": 49970 }, { "epoch": 0.24825407547957046, "grad_norm": 1.671797752380371, "learning_rate": 1.7604096223039027e-05, "loss": 0.16366006135940553, "step": 49980 }, { "epoch": 0.2483037461629397, "grad_norm": 1.7760387659072876, "learning_rate": 1.7603051355557147e-05, "loss": 0.16646370887756348, "step": 49990 }, { "epoch": 0.24835341684630896, "grad_norm": 1.2249410152435303, "learning_rate": 1.7602006291311356e-05, "loss": 0.18089457750320434, "step": 50000 }, { "epoch": 0.24840308752967824, "grad_norm": 1.8238478899002075, "learning_rate": 1.76009610303287e-05, "loss": 0.18856868743896485, "step": 50010 }, { "epoch": 0.2484527582130475, "grad_norm": 1.4486666917800903, "learning_rate": 1.7599915572636226e-05, "loss": 0.1786940336227417, "step": 50020 }, { "epoch": 0.24850242889641674, "grad_norm": 1.0810004472732544, "learning_rate": 1.7598869918260995e-05, "loss": 0.16385438442230224, "step": 50030 }, { "epoch": 0.24855209957978602, "grad_norm": 1.2898318767547607, "learning_rate": 1.7597824067230064e-05, "loss": 0.19341326951980592, "step": 50040 }, { "epoch": 0.24860177026315528, "grad_norm": 1.4113383293151855, "learning_rate": 1.75967780195705e-05, "loss": 0.18865711688995362, "step": 50050 }, { "epoch": 0.24865144094652455, "grad_norm": 1.0070712566375732, "learning_rate": 1.759573177530938e-05, "loss": 0.1784184455871582, "step": 50060 }, { "epoch": 0.2487011116298938, "grad_norm": 1.303836464881897, "learning_rate": 1.759468533447377e-05, "loss": 0.17458442449569703, "step": 50070 }, { "epoch": 0.24875078231326306, "grad_norm": 1.3156132698059082, "learning_rate": 1.759363869709076e-05, "loss": 0.17299742698669435, "step": 50080 }, { "epoch": 0.24880045299663234, "grad_norm": 1.7183516025543213, "learning_rate": 1.7592591863187435e-05, "loss": 0.1674654960632324, "step": 50090 }, { "epoch": 0.2488501236800016, "grad_norm": 1.4103394746780396, "learning_rate": 1.7591544832790884e-05, "loss": 0.1814943313598633, "step": 50100 }, { "epoch": 0.24889979436337084, "grad_norm": 1.1211113929748535, "learning_rate": 1.7590497605928206e-05, "loss": 0.18741487264633178, "step": 50110 }, { "epoch": 0.24894946504674012, "grad_norm": 1.1168538331985474, "learning_rate": 1.7589450182626505e-05, "loss": 0.16515300273895264, "step": 50120 }, { "epoch": 0.24899913573010937, "grad_norm": 2.1560871601104736, "learning_rate": 1.758840256291288e-05, "loss": 0.1687355399131775, "step": 50130 }, { "epoch": 0.24904880641347862, "grad_norm": 1.551570177078247, "learning_rate": 1.758735474681445e-05, "loss": 0.17273905277252197, "step": 50140 }, { "epoch": 0.2490984770968479, "grad_norm": 2.4148097038269043, "learning_rate": 1.758630673435833e-05, "loss": 0.16178224086761475, "step": 50150 }, { "epoch": 0.24914814778021716, "grad_norm": 1.3502469062805176, "learning_rate": 1.758525852557164e-05, "loss": 0.17257590293884278, "step": 50160 }, { "epoch": 0.24919781846358643, "grad_norm": 0.8653885126113892, "learning_rate": 1.758421012048151e-05, "loss": 0.16648313999176026, "step": 50170 }, { "epoch": 0.2492474891469557, "grad_norm": 1.4784352779388428, "learning_rate": 1.7583161519115073e-05, "loss": 0.18641355037689208, "step": 50180 }, { "epoch": 0.24929715983032494, "grad_norm": 2.244035005569458, "learning_rate": 1.7582112721499464e-05, "loss": 0.17433438301086426, "step": 50190 }, { "epoch": 0.24934683051369422, "grad_norm": 1.16177499294281, "learning_rate": 1.7581063727661826e-05, "loss": 0.1750638723373413, "step": 50200 }, { "epoch": 0.24939650119706347, "grad_norm": 1.499389886856079, "learning_rate": 1.7580014537629307e-05, "loss": 0.18760466575622559, "step": 50210 }, { "epoch": 0.24944617188043272, "grad_norm": 1.2473783493041992, "learning_rate": 1.757896515142906e-05, "loss": 0.17077045440673827, "step": 50220 }, { "epoch": 0.249495842563802, "grad_norm": 0.7922122478485107, "learning_rate": 1.757791556908824e-05, "loss": 0.18066871166229248, "step": 50230 }, { "epoch": 0.24954551324717125, "grad_norm": 1.0677696466445923, "learning_rate": 1.7576865790634016e-05, "loss": 0.18554556369781494, "step": 50240 }, { "epoch": 0.2495951839305405, "grad_norm": 0.832114040851593, "learning_rate": 1.757581581609355e-05, "loss": 0.17626606225967406, "step": 50250 }, { "epoch": 0.24964485461390978, "grad_norm": 2.211254835128784, "learning_rate": 1.7574765645494015e-05, "loss": 0.17603477239608764, "step": 50260 }, { "epoch": 0.24969452529727904, "grad_norm": 3.1308867931365967, "learning_rate": 1.7573715278862594e-05, "loss": 0.1674717903137207, "step": 50270 }, { "epoch": 0.24974419598064831, "grad_norm": 1.5242151021957397, "learning_rate": 1.7572664716226466e-05, "loss": 0.17809247970581055, "step": 50280 }, { "epoch": 0.24979386666401757, "grad_norm": 1.5569653511047363, "learning_rate": 1.757161395761282e-05, "loss": 0.1779848098754883, "step": 50290 }, { "epoch": 0.24984353734738682, "grad_norm": 1.337874412536621, "learning_rate": 1.757056300304885e-05, "loss": 0.17638089656829833, "step": 50300 }, { "epoch": 0.2498932080307561, "grad_norm": 1.3719888925552368, "learning_rate": 1.756951185256175e-05, "loss": 0.1650542974472046, "step": 50310 }, { "epoch": 0.24994287871412535, "grad_norm": 0.666706919670105, "learning_rate": 1.756846050617873e-05, "loss": 0.16249217987060546, "step": 50320 }, { "epoch": 0.2499925493974946, "grad_norm": 0.8197651505470276, "learning_rate": 1.7567408963926996e-05, "loss": 0.17615897655487062, "step": 50330 }, { "epoch": 0.25004222008086385, "grad_norm": 2.455716848373413, "learning_rate": 1.756635722583376e-05, "loss": 0.1862836956977844, "step": 50340 }, { "epoch": 0.25009189076423316, "grad_norm": 1.05819833278656, "learning_rate": 1.756530529192624e-05, "loss": 0.16555893421173096, "step": 50350 }, { "epoch": 0.2501415614476024, "grad_norm": 1.6953763961791992, "learning_rate": 1.756425316223166e-05, "loss": 0.181396222114563, "step": 50360 }, { "epoch": 0.25019123213097166, "grad_norm": 0.7406819462776184, "learning_rate": 1.756320083677725e-05, "loss": 0.17555274963378906, "step": 50370 }, { "epoch": 0.2502409028143409, "grad_norm": 1.0399689674377441, "learning_rate": 1.7562148315590243e-05, "loss": 0.17808732986450196, "step": 50380 }, { "epoch": 0.25029057349771017, "grad_norm": 1.9838024377822876, "learning_rate": 1.7561095598697883e-05, "loss": 0.15450278520584107, "step": 50390 }, { "epoch": 0.2503402441810794, "grad_norm": 0.6567367315292358, "learning_rate": 1.7560042686127407e-05, "loss": 0.18600718975067138, "step": 50400 }, { "epoch": 0.2503899148644487, "grad_norm": 1.491387963294983, "learning_rate": 1.7558989577906062e-05, "loss": 0.16312109231948851, "step": 50410 }, { "epoch": 0.250439585547818, "grad_norm": 2.1347405910491943, "learning_rate": 1.755793627406111e-05, "loss": 0.1876645565032959, "step": 50420 }, { "epoch": 0.25048925623118723, "grad_norm": 1.2200510501861572, "learning_rate": 1.7556882774619804e-05, "loss": 0.17899361848831177, "step": 50430 }, { "epoch": 0.2505389269145565, "grad_norm": 1.2209587097167969, "learning_rate": 1.7555829079609413e-05, "loss": 0.1779162049293518, "step": 50440 }, { "epoch": 0.25058859759792573, "grad_norm": 0.8957834243774414, "learning_rate": 1.75547751890572e-05, "loss": 0.18314459323883056, "step": 50450 }, { "epoch": 0.25063826828129504, "grad_norm": 1.6932581663131714, "learning_rate": 1.7553721102990443e-05, "loss": 0.17403671741485596, "step": 50460 }, { "epoch": 0.2506879389646643, "grad_norm": 0.9220479130744934, "learning_rate": 1.7552666821436424e-05, "loss": 0.17404394149780272, "step": 50470 }, { "epoch": 0.25073760964803354, "grad_norm": 2.253812551498413, "learning_rate": 1.755161234442242e-05, "loss": 0.1726135015487671, "step": 50480 }, { "epoch": 0.2507872803314028, "grad_norm": 1.0867778062820435, "learning_rate": 1.755055767197573e-05, "loss": 0.1864282727241516, "step": 50490 }, { "epoch": 0.25083695101477205, "grad_norm": 1.501929521560669, "learning_rate": 1.754950280412364e-05, "loss": 0.17023825645446777, "step": 50500 }, { "epoch": 0.2508866216981413, "grad_norm": 1.6515071392059326, "learning_rate": 1.7548447740893456e-05, "loss": 0.1629666805267334, "step": 50510 }, { "epoch": 0.2509362923815106, "grad_norm": 0.9030997157096863, "learning_rate": 1.754739248231248e-05, "loss": 0.1657063603401184, "step": 50520 }, { "epoch": 0.25098596306487986, "grad_norm": 0.8010819554328918, "learning_rate": 1.7546337028408018e-05, "loss": 0.16141922473907472, "step": 50530 }, { "epoch": 0.2510356337482491, "grad_norm": 0.5913136005401611, "learning_rate": 1.7545281379207385e-05, "loss": 0.18776657581329345, "step": 50540 }, { "epoch": 0.25108530443161836, "grad_norm": 2.1476712226867676, "learning_rate": 1.754422553473791e-05, "loss": 0.16068419218063354, "step": 50550 }, { "epoch": 0.2511349751149876, "grad_norm": 1.0821539163589478, "learning_rate": 1.7543169495026912e-05, "loss": 0.16598260402679443, "step": 50560 }, { "epoch": 0.2511846457983569, "grad_norm": 1.4014135599136353, "learning_rate": 1.7542113260101715e-05, "loss": 0.17858781814575195, "step": 50570 }, { "epoch": 0.25123431648172617, "grad_norm": 0.7186217308044434, "learning_rate": 1.7541056829989662e-05, "loss": 0.1691925048828125, "step": 50580 }, { "epoch": 0.2512839871650954, "grad_norm": 1.6575360298156738, "learning_rate": 1.754000020471809e-05, "loss": 0.1826760172843933, "step": 50590 }, { "epoch": 0.2513336578484647, "grad_norm": 1.0781738758087158, "learning_rate": 1.7538943384314346e-05, "loss": 0.18067811727523803, "step": 50600 }, { "epoch": 0.2513833285318339, "grad_norm": 0.8416363596916199, "learning_rate": 1.7537886368805777e-05, "loss": 0.16938817501068115, "step": 50610 }, { "epoch": 0.2514329992152032, "grad_norm": 1.0587549209594727, "learning_rate": 1.7536829158219738e-05, "loss": 0.16702990531921386, "step": 50620 }, { "epoch": 0.2514826698985725, "grad_norm": 1.9158145189285278, "learning_rate": 1.7535771752583596e-05, "loss": 0.17323877811431884, "step": 50630 }, { "epoch": 0.25153234058194174, "grad_norm": 1.1992892026901245, "learning_rate": 1.753471415192471e-05, "loss": 0.1723261833190918, "step": 50640 }, { "epoch": 0.251582011265311, "grad_norm": 1.4049466848373413, "learning_rate": 1.7533656356270444e-05, "loss": 0.17160255908966066, "step": 50650 }, { "epoch": 0.25163168194868024, "grad_norm": 1.246248483657837, "learning_rate": 1.753259836564819e-05, "loss": 0.17201752662658693, "step": 50660 }, { "epoch": 0.2516813526320495, "grad_norm": 3.263498067855835, "learning_rate": 1.7531540180085313e-05, "loss": 0.16853880882263184, "step": 50670 }, { "epoch": 0.2517310233154188, "grad_norm": 1.6741738319396973, "learning_rate": 1.7530481799609208e-05, "loss": 0.18576231002807617, "step": 50680 }, { "epoch": 0.25178069399878805, "grad_norm": 1.4269038438796997, "learning_rate": 1.7529423224247258e-05, "loss": 0.17927267551422119, "step": 50690 }, { "epoch": 0.2518303646821573, "grad_norm": 1.0877962112426758, "learning_rate": 1.7528364454026866e-05, "loss": 0.16601970195770263, "step": 50700 }, { "epoch": 0.25188003536552656, "grad_norm": 1.0987492799758911, "learning_rate": 1.7527305488975428e-05, "loss": 0.16340793371200563, "step": 50710 }, { "epoch": 0.2519297060488958, "grad_norm": 0.7845184206962585, "learning_rate": 1.7526246329120348e-05, "loss": 0.16821272373199464, "step": 50720 }, { "epoch": 0.25197937673226506, "grad_norm": 1.2947579622268677, "learning_rate": 1.7525186974489042e-05, "loss": 0.17278215885162354, "step": 50730 }, { "epoch": 0.25202904741563437, "grad_norm": 1.0581077337265015, "learning_rate": 1.7524127425108924e-05, "loss": 0.1638315200805664, "step": 50740 }, { "epoch": 0.2520787180990036, "grad_norm": 1.2162765264511108, "learning_rate": 1.752306768100741e-05, "loss": 0.16244513988494874, "step": 50750 }, { "epoch": 0.25212838878237287, "grad_norm": 0.6822167634963989, "learning_rate": 1.7522007742211934e-05, "loss": 0.1623812198638916, "step": 50760 }, { "epoch": 0.2521780594657421, "grad_norm": 2.415268898010254, "learning_rate": 1.7520947608749917e-05, "loss": 0.178950035572052, "step": 50770 }, { "epoch": 0.2522277301491114, "grad_norm": 0.811752438545227, "learning_rate": 1.7519887280648805e-05, "loss": 0.1602293610572815, "step": 50780 }, { "epoch": 0.2522774008324807, "grad_norm": 1.4067765474319458, "learning_rate": 1.751882675793603e-05, "loss": 0.17360100746154786, "step": 50790 }, { "epoch": 0.25232707151584993, "grad_norm": 0.9162693619728088, "learning_rate": 1.751776604063904e-05, "loss": 0.1709376096725464, "step": 50800 }, { "epoch": 0.2523767421992192, "grad_norm": 1.5335499048233032, "learning_rate": 1.7516705128785293e-05, "loss": 0.17718029022216797, "step": 50810 }, { "epoch": 0.25242641288258844, "grad_norm": 1.1887785196304321, "learning_rate": 1.7515644022402235e-05, "loss": 0.17190637588500976, "step": 50820 }, { "epoch": 0.2524760835659577, "grad_norm": 1.5542434453964233, "learning_rate": 1.7514582721517332e-05, "loss": 0.18312952518463135, "step": 50830 }, { "epoch": 0.25252575424932694, "grad_norm": 1.5922967195510864, "learning_rate": 1.7513521226158056e-05, "loss": 0.17416067123413087, "step": 50840 }, { "epoch": 0.25257542493269625, "grad_norm": 1.0950883626937866, "learning_rate": 1.7512459536351864e-05, "loss": 0.15928280353546143, "step": 50850 }, { "epoch": 0.2526250956160655, "grad_norm": 0.8497464060783386, "learning_rate": 1.7511397652126243e-05, "loss": 0.18212575912475587, "step": 50860 }, { "epoch": 0.25267476629943475, "grad_norm": 2.1789040565490723, "learning_rate": 1.751033557350867e-05, "loss": 0.1755215644836426, "step": 50870 }, { "epoch": 0.252724436982804, "grad_norm": 1.09894597530365, "learning_rate": 1.7509273300526634e-05, "loss": 0.1851439118385315, "step": 50880 }, { "epoch": 0.25277410766617325, "grad_norm": 0.5074933767318726, "learning_rate": 1.750821083320762e-05, "loss": 0.16651694774627684, "step": 50890 }, { "epoch": 0.25282377834954256, "grad_norm": 1.319087266921997, "learning_rate": 1.7507148171579133e-05, "loss": 0.18382532596588136, "step": 50900 }, { "epoch": 0.2528734490329118, "grad_norm": 1.9010151624679565, "learning_rate": 1.7506085315668672e-05, "loss": 0.17893033027648925, "step": 50910 }, { "epoch": 0.25292311971628106, "grad_norm": 1.3557935953140259, "learning_rate": 1.7505022265503734e-05, "loss": 0.1777127742767334, "step": 50920 }, { "epoch": 0.2529727903996503, "grad_norm": 0.6584174036979675, "learning_rate": 1.7503959021111844e-05, "loss": 0.17314667701721193, "step": 50930 }, { "epoch": 0.25302246108301957, "grad_norm": 1.893239974975586, "learning_rate": 1.7502895582520507e-05, "loss": 0.19634509086608887, "step": 50940 }, { "epoch": 0.2530721317663888, "grad_norm": 1.637439250946045, "learning_rate": 1.750183194975725e-05, "loss": 0.1745431661605835, "step": 50950 }, { "epoch": 0.2531218024497581, "grad_norm": 1.1348246335983276, "learning_rate": 1.75007681228496e-05, "loss": 0.18648154735565187, "step": 50960 }, { "epoch": 0.2531714731331274, "grad_norm": 1.1520180702209473, "learning_rate": 1.749970410182509e-05, "loss": 0.18512465953826904, "step": 50970 }, { "epoch": 0.25322114381649663, "grad_norm": 1.2252243757247925, "learning_rate": 1.7498639886711246e-05, "loss": 0.1859443187713623, "step": 50980 }, { "epoch": 0.2532708144998659, "grad_norm": 1.8353465795516968, "learning_rate": 1.749757547753562e-05, "loss": 0.17268190383911133, "step": 50990 }, { "epoch": 0.25332048518323513, "grad_norm": 0.8699620366096497, "learning_rate": 1.7496510874325757e-05, "loss": 0.17498830556869507, "step": 51000 }, { "epoch": 0.2533701558666044, "grad_norm": 1.085565209388733, "learning_rate": 1.7495446077109205e-05, "loss": 0.17905272245407106, "step": 51010 }, { "epoch": 0.2534198265499737, "grad_norm": 0.8298930525779724, "learning_rate": 1.7494381085913523e-05, "loss": 0.17271764278411866, "step": 51020 }, { "epoch": 0.25346949723334294, "grad_norm": 0.6777648329734802, "learning_rate": 1.7493315900766273e-05, "loss": 0.1766733169555664, "step": 51030 }, { "epoch": 0.2535191679167122, "grad_norm": 1.0515336990356445, "learning_rate": 1.7492250521695017e-05, "loss": 0.1761109709739685, "step": 51040 }, { "epoch": 0.25356883860008145, "grad_norm": 1.7204054594039917, "learning_rate": 1.7491184948727332e-05, "loss": 0.18073469400405884, "step": 51050 }, { "epoch": 0.2536185092834507, "grad_norm": 0.6779294610023499, "learning_rate": 1.7490119181890794e-05, "loss": 0.18014264106750488, "step": 51060 }, { "epoch": 0.25366817996682, "grad_norm": 1.024440884590149, "learning_rate": 1.7489053221212984e-05, "loss": 0.17070815563201905, "step": 51070 }, { "epoch": 0.25371785065018926, "grad_norm": 1.0702532529830933, "learning_rate": 1.7487987066721484e-05, "loss": 0.17368054389953613, "step": 51080 }, { "epoch": 0.2537675213335585, "grad_norm": 0.8089594841003418, "learning_rate": 1.748692071844389e-05, "loss": 0.16022982597351074, "step": 51090 }, { "epoch": 0.25381719201692776, "grad_norm": 0.6320927143096924, "learning_rate": 1.74858541764078e-05, "loss": 0.1811543345451355, "step": 51100 }, { "epoch": 0.253866862700297, "grad_norm": 1.8701647520065308, "learning_rate": 1.7484787440640817e-05, "loss": 0.16607427597045898, "step": 51110 }, { "epoch": 0.25391653338366627, "grad_norm": 1.0048191547393799, "learning_rate": 1.7483720511170538e-05, "loss": 0.1849263548851013, "step": 51120 }, { "epoch": 0.2539662040670356, "grad_norm": 1.0679174661636353, "learning_rate": 1.7482653388024586e-05, "loss": 0.16996045112609864, "step": 51130 }, { "epoch": 0.2540158747504048, "grad_norm": 0.6939079165458679, "learning_rate": 1.748158607123057e-05, "loss": 0.16355682611465455, "step": 51140 }, { "epoch": 0.2540655454337741, "grad_norm": 1.5798081159591675, "learning_rate": 1.748051856081612e-05, "loss": 0.17526538372039796, "step": 51150 }, { "epoch": 0.2541152161171433, "grad_norm": 1.1983426809310913, "learning_rate": 1.7479450856808847e-05, "loss": 0.18290255069732667, "step": 51160 }, { "epoch": 0.2541648868005126, "grad_norm": 0.9262390732765198, "learning_rate": 1.7478382959236403e-05, "loss": 0.18310940265655518, "step": 51170 }, { "epoch": 0.2542145574838819, "grad_norm": 2.1082746982574463, "learning_rate": 1.747731486812641e-05, "loss": 0.17603952884674073, "step": 51180 }, { "epoch": 0.25426422816725114, "grad_norm": 1.5498138666152954, "learning_rate": 1.7476246583506516e-05, "loss": 0.17035608291625975, "step": 51190 }, { "epoch": 0.2543138988506204, "grad_norm": 1.0023741722106934, "learning_rate": 1.7475178105404368e-05, "loss": 0.15792648792266845, "step": 51200 }, { "epoch": 0.25436356953398964, "grad_norm": 1.5127712488174438, "learning_rate": 1.7474109433847613e-05, "loss": 0.17693287134170532, "step": 51210 }, { "epoch": 0.2544132402173589, "grad_norm": 1.3846217393875122, "learning_rate": 1.7473040568863915e-05, "loss": 0.16045916080474854, "step": 51220 }, { "epoch": 0.25446291090072815, "grad_norm": 0.8533744215965271, "learning_rate": 1.747197151048093e-05, "loss": 0.17505718469619752, "step": 51230 }, { "epoch": 0.25451258158409745, "grad_norm": 1.0788207054138184, "learning_rate": 1.7470902258726327e-05, "loss": 0.1816399097442627, "step": 51240 }, { "epoch": 0.2545622522674667, "grad_norm": 0.8738555312156677, "learning_rate": 1.7469832813627775e-05, "loss": 0.18587825298309327, "step": 51250 }, { "epoch": 0.25461192295083596, "grad_norm": 0.87406986951828, "learning_rate": 1.7468763175212956e-05, "loss": 0.1711321711540222, "step": 51260 }, { "epoch": 0.2546615936342052, "grad_norm": 1.2124319076538086, "learning_rate": 1.746769334350955e-05, "loss": 0.1779651641845703, "step": 51270 }, { "epoch": 0.25471126431757446, "grad_norm": 0.7060151100158691, "learning_rate": 1.7466623318545242e-05, "loss": 0.1681140184402466, "step": 51280 }, { "epoch": 0.25476093500094377, "grad_norm": 1.5329599380493164, "learning_rate": 1.746555310034772e-05, "loss": 0.18427950143814087, "step": 51290 }, { "epoch": 0.254810605684313, "grad_norm": 1.9678305387496948, "learning_rate": 1.746448268894469e-05, "loss": 0.17816249132156373, "step": 51300 }, { "epoch": 0.25486027636768227, "grad_norm": 0.9116772413253784, "learning_rate": 1.746341208436385e-05, "loss": 0.16238940954208375, "step": 51310 }, { "epoch": 0.2549099470510515, "grad_norm": 1.3806967735290527, "learning_rate": 1.7462341286632906e-05, "loss": 0.17726287841796876, "step": 51320 }, { "epoch": 0.2549596177344208, "grad_norm": 2.1876749992370605, "learning_rate": 1.7461270295779568e-05, "loss": 0.168428635597229, "step": 51330 }, { "epoch": 0.25500928841779, "grad_norm": 1.2889320850372314, "learning_rate": 1.7460199111831558e-05, "loss": 0.1682222843170166, "step": 51340 }, { "epoch": 0.25505895910115933, "grad_norm": 2.107103109359741, "learning_rate": 1.745912773481659e-05, "loss": 0.16589560508728027, "step": 51350 }, { "epoch": 0.2551086297845286, "grad_norm": 1.1746461391448975, "learning_rate": 1.7458056164762397e-05, "loss": 0.18014466762542725, "step": 51360 }, { "epoch": 0.25515830046789784, "grad_norm": 2.3700172901153564, "learning_rate": 1.745698440169671e-05, "loss": 0.18052183389663695, "step": 51370 }, { "epoch": 0.2552079711512671, "grad_norm": 1.3566633462905884, "learning_rate": 1.745591244564727e-05, "loss": 0.19024674892425536, "step": 51380 }, { "epoch": 0.25525764183463634, "grad_norm": 0.967118501663208, "learning_rate": 1.7454840296641804e-05, "loss": 0.1767976999282837, "step": 51390 }, { "epoch": 0.25530731251800565, "grad_norm": 1.3555498123168945, "learning_rate": 1.7453767954708076e-05, "loss": 0.18383996486663817, "step": 51400 }, { "epoch": 0.2553569832013749, "grad_norm": 1.0602325201034546, "learning_rate": 1.7452695419873828e-05, "loss": 0.1619335412979126, "step": 51410 }, { "epoch": 0.25540665388474415, "grad_norm": 1.3673505783081055, "learning_rate": 1.7451622692166818e-05, "loss": 0.16478981971740722, "step": 51420 }, { "epoch": 0.2554563245681134, "grad_norm": 0.994497537612915, "learning_rate": 1.745054977161481e-05, "loss": 0.1821151614189148, "step": 51430 }, { "epoch": 0.25550599525148265, "grad_norm": 1.6605185270309448, "learning_rate": 1.744947665824557e-05, "loss": 0.16492304801940919, "step": 51440 }, { "epoch": 0.2555556659348519, "grad_norm": 1.4222697019577026, "learning_rate": 1.7448403352086868e-05, "loss": 0.17462118864059448, "step": 51450 }, { "epoch": 0.2556053366182212, "grad_norm": 1.6275734901428223, "learning_rate": 1.7447329853166483e-05, "loss": 0.19043012857437133, "step": 51460 }, { "epoch": 0.25565500730159046, "grad_norm": 2.253718614578247, "learning_rate": 1.7446256161512197e-05, "loss": 0.16158533096313477, "step": 51470 }, { "epoch": 0.2557046779849597, "grad_norm": 4.081194877624512, "learning_rate": 1.7445182277151797e-05, "loss": 0.18407077789306642, "step": 51480 }, { "epoch": 0.25575434866832897, "grad_norm": 0.792512059211731, "learning_rate": 1.744410820011307e-05, "loss": 0.16382911205291747, "step": 51490 }, { "epoch": 0.2558040193516982, "grad_norm": 1.7068967819213867, "learning_rate": 1.7443033930423814e-05, "loss": 0.17052432298660278, "step": 51500 }, { "epoch": 0.2558536900350675, "grad_norm": 2.246422052383423, "learning_rate": 1.7441959468111838e-05, "loss": 0.17280640602111816, "step": 51510 }, { "epoch": 0.2559033607184368, "grad_norm": 0.8047261834144592, "learning_rate": 1.744088481320494e-05, "loss": 0.1739686131477356, "step": 51520 }, { "epoch": 0.25595303140180603, "grad_norm": 0.6099956631660461, "learning_rate": 1.7439809965730933e-05, "loss": 0.16080780029296876, "step": 51530 }, { "epoch": 0.2560027020851753, "grad_norm": 1.594722032546997, "learning_rate": 1.743873492571764e-05, "loss": 0.1786554455757141, "step": 51540 }, { "epoch": 0.25605237276854453, "grad_norm": 1.31941819190979, "learning_rate": 1.7437659693192874e-05, "loss": 0.19205131530761718, "step": 51550 }, { "epoch": 0.2561020434519138, "grad_norm": 0.9120762944221497, "learning_rate": 1.743658426818447e-05, "loss": 0.17706060409545898, "step": 51560 }, { "epoch": 0.2561517141352831, "grad_norm": 1.5850085020065308, "learning_rate": 1.7435508650720252e-05, "loss": 0.17268009185791017, "step": 51570 }, { "epoch": 0.25620138481865234, "grad_norm": 1.4104089736938477, "learning_rate": 1.7434432840828063e-05, "loss": 0.16338049173355101, "step": 51580 }, { "epoch": 0.2562510555020216, "grad_norm": 1.500163197517395, "learning_rate": 1.7433356838535734e-05, "loss": 0.18007819652557372, "step": 51590 }, { "epoch": 0.25630072618539085, "grad_norm": 0.9081949591636658, "learning_rate": 1.7432280643871127e-05, "loss": 0.17918119430541993, "step": 51600 }, { "epoch": 0.2563503968687601, "grad_norm": 0.8440309166908264, "learning_rate": 1.743120425686208e-05, "loss": 0.17887389659881592, "step": 51610 }, { "epoch": 0.2564000675521294, "grad_norm": 0.9014836549758911, "learning_rate": 1.743012767753646e-05, "loss": 0.17330336570739746, "step": 51620 }, { "epoch": 0.25644973823549866, "grad_norm": 1.6459510326385498, "learning_rate": 1.742905090592212e-05, "loss": 0.1774150848388672, "step": 51630 }, { "epoch": 0.2564994089188679, "grad_norm": 0.9396331310272217, "learning_rate": 1.742797394204693e-05, "loss": 0.14875996112823486, "step": 51640 }, { "epoch": 0.25654907960223716, "grad_norm": 1.1578625440597534, "learning_rate": 1.742689678593876e-05, "loss": 0.16459468603134156, "step": 51650 }, { "epoch": 0.2565987502856064, "grad_norm": 1.5051681995391846, "learning_rate": 1.742581943762549e-05, "loss": 0.1871103048324585, "step": 51660 }, { "epoch": 0.25664842096897567, "grad_norm": 0.807856023311615, "learning_rate": 1.7424741897135e-05, "loss": 0.1845066785812378, "step": 51670 }, { "epoch": 0.256698091652345, "grad_norm": 0.9231998324394226, "learning_rate": 1.742366416449517e-05, "loss": 0.18307920694351196, "step": 51680 }, { "epoch": 0.2567477623357142, "grad_norm": 1.8554108142852783, "learning_rate": 1.7422586239733895e-05, "loss": 0.1796162247657776, "step": 51690 }, { "epoch": 0.2567974330190835, "grad_norm": 1.4507237672805786, "learning_rate": 1.7421508122879077e-05, "loss": 0.17402796745300292, "step": 51700 }, { "epoch": 0.25684710370245273, "grad_norm": 1.7581442594528198, "learning_rate": 1.7420429813958614e-05, "loss": 0.16727544069290162, "step": 51710 }, { "epoch": 0.256896774385822, "grad_norm": 0.7830767631530762, "learning_rate": 1.741935131300041e-05, "loss": 0.15968490839004518, "step": 51720 }, { "epoch": 0.2569464450691913, "grad_norm": 1.7154425382614136, "learning_rate": 1.7418272620032375e-05, "loss": 0.16640983819961547, "step": 51730 }, { "epoch": 0.25699611575256054, "grad_norm": 1.0552232265472412, "learning_rate": 1.741719373508243e-05, "loss": 0.17525575160980225, "step": 51740 }, { "epoch": 0.2570457864359298, "grad_norm": 0.9750630855560303, "learning_rate": 1.741611465817849e-05, "loss": 0.1538846015930176, "step": 51750 }, { "epoch": 0.25709545711929904, "grad_norm": 1.3527086973190308, "learning_rate": 1.7415035389348487e-05, "loss": 0.18389838933944702, "step": 51760 }, { "epoch": 0.2571451278026683, "grad_norm": 1.8105264902114868, "learning_rate": 1.741395592862035e-05, "loss": 0.19356467723846435, "step": 51770 }, { "epoch": 0.25719479848603755, "grad_norm": 1.4983603954315186, "learning_rate": 1.7412876276022015e-05, "loss": 0.17851451635360718, "step": 51780 }, { "epoch": 0.25724446916940685, "grad_norm": 3.142768621444702, "learning_rate": 1.741179643158142e-05, "loss": 0.1753080368041992, "step": 51790 }, { "epoch": 0.2572941398527761, "grad_norm": 1.5140331983566284, "learning_rate": 1.7410716395326515e-05, "loss": 0.17970407009124756, "step": 51800 }, { "epoch": 0.25734381053614536, "grad_norm": 1.383447289466858, "learning_rate": 1.740963616728525e-05, "loss": 0.17813239097595215, "step": 51810 }, { "epoch": 0.2573934812195146, "grad_norm": 1.395013451576233, "learning_rate": 1.740855574748558e-05, "loss": 0.16408355236053468, "step": 51820 }, { "epoch": 0.25744315190288386, "grad_norm": 1.0393390655517578, "learning_rate": 1.7407475135955467e-05, "loss": 0.17461962699890138, "step": 51830 }, { "epoch": 0.25749282258625317, "grad_norm": 1.0021495819091797, "learning_rate": 1.7406394332722873e-05, "loss": 0.1678410291671753, "step": 51840 }, { "epoch": 0.2575424932696224, "grad_norm": 1.3257020711898804, "learning_rate": 1.7405313337815774e-05, "loss": 0.17365461587905884, "step": 51850 }, { "epoch": 0.25759216395299167, "grad_norm": 3.250505208969116, "learning_rate": 1.740423215126214e-05, "loss": 0.19384114742279052, "step": 51860 }, { "epoch": 0.2576418346363609, "grad_norm": 1.376752257347107, "learning_rate": 1.7403150773089962e-05, "loss": 0.1699511766433716, "step": 51870 }, { "epoch": 0.2576915053197302, "grad_norm": 0.8827232122421265, "learning_rate": 1.7402069203327216e-05, "loss": 0.15357991456985473, "step": 51880 }, { "epoch": 0.2577411760030994, "grad_norm": 0.7925800085067749, "learning_rate": 1.740098744200189e-05, "loss": 0.183049476146698, "step": 51890 }, { "epoch": 0.25779084668646873, "grad_norm": 1.756395697593689, "learning_rate": 1.7399905489141987e-05, "loss": 0.18386409282684327, "step": 51900 }, { "epoch": 0.257840517369838, "grad_norm": 0.9531145095825195, "learning_rate": 1.7398823344775508e-05, "loss": 0.1654682159423828, "step": 51910 }, { "epoch": 0.25789018805320724, "grad_norm": 0.9162717461585999, "learning_rate": 1.7397741008930458e-05, "loss": 0.1850151777267456, "step": 51920 }, { "epoch": 0.2579398587365765, "grad_norm": 0.7350218296051025, "learning_rate": 1.739665848163484e-05, "loss": 0.17063558101654053, "step": 51930 }, { "epoch": 0.25798952941994574, "grad_norm": 1.2531810998916626, "learning_rate": 1.7395575762916675e-05, "loss": 0.17281322479248046, "step": 51940 }, { "epoch": 0.25803920010331505, "grad_norm": 1.3630367517471313, "learning_rate": 1.7394492852803985e-05, "loss": 0.16659460067749024, "step": 51950 }, { "epoch": 0.2580888707866843, "grad_norm": 0.5224915146827698, "learning_rate": 1.7393409751324794e-05, "loss": 0.1660557508468628, "step": 51960 }, { "epoch": 0.25813854147005355, "grad_norm": 0.9327402710914612, "learning_rate": 1.739232645850713e-05, "loss": 0.1683824062347412, "step": 51970 }, { "epoch": 0.2581882121534228, "grad_norm": 1.5091981887817383, "learning_rate": 1.7391242974379027e-05, "loss": 0.17486822605133057, "step": 51980 }, { "epoch": 0.25823788283679205, "grad_norm": 2.4006237983703613, "learning_rate": 1.739015929896853e-05, "loss": 0.16729896068572997, "step": 51990 }, { "epoch": 0.2582875535201613, "grad_norm": 1.760961651802063, "learning_rate": 1.7389075432303682e-05, "loss": 0.17607059478759765, "step": 52000 }, { "epoch": 0.2583372242035306, "grad_norm": 1.662132978439331, "learning_rate": 1.7387991374412533e-05, "loss": 0.18786358833312988, "step": 52010 }, { "epoch": 0.25838689488689986, "grad_norm": 1.214689016342163, "learning_rate": 1.738690712532314e-05, "loss": 0.18215398788452147, "step": 52020 }, { "epoch": 0.2584365655702691, "grad_norm": 2.4850988388061523, "learning_rate": 1.7385822685063557e-05, "loss": 0.17911510467529296, "step": 52030 }, { "epoch": 0.25848623625363837, "grad_norm": 1.1813127994537354, "learning_rate": 1.7384738053661855e-05, "loss": 0.1698759078979492, "step": 52040 }, { "epoch": 0.2585359069370076, "grad_norm": 0.9567163586616516, "learning_rate": 1.7383653231146102e-05, "loss": 0.17380253076553345, "step": 52050 }, { "epoch": 0.2585855776203769, "grad_norm": 1.0079234838485718, "learning_rate": 1.7382568217544367e-05, "loss": 0.1746651530265808, "step": 52060 }, { "epoch": 0.2586352483037462, "grad_norm": 0.9724189043045044, "learning_rate": 1.738148301288474e-05, "loss": 0.17157210111618043, "step": 52070 }, { "epoch": 0.25868491898711543, "grad_norm": 1.0526208877563477, "learning_rate": 1.73803976171953e-05, "loss": 0.16690397262573242, "step": 52080 }, { "epoch": 0.2587345896704847, "grad_norm": 1.6511783599853516, "learning_rate": 1.7379312030504137e-05, "loss": 0.16682449579238892, "step": 52090 }, { "epoch": 0.25878426035385393, "grad_norm": 2.3727362155914307, "learning_rate": 1.7378226252839345e-05, "loss": 0.17316789627075196, "step": 52100 }, { "epoch": 0.2588339310372232, "grad_norm": 1.5830025672912598, "learning_rate": 1.7377140284229023e-05, "loss": 0.18026657104492189, "step": 52110 }, { "epoch": 0.2588836017205925, "grad_norm": 1.4588323831558228, "learning_rate": 1.7376054124701277e-05, "loss": 0.16909419298171996, "step": 52120 }, { "epoch": 0.25893327240396175, "grad_norm": 1.6656914949417114, "learning_rate": 1.737496777428422e-05, "loss": 0.1715498685836792, "step": 52130 }, { "epoch": 0.258982943087331, "grad_norm": 0.8276429772377014, "learning_rate": 1.7373881233005958e-05, "loss": 0.16564221382141114, "step": 52140 }, { "epoch": 0.25903261377070025, "grad_norm": 0.6842420101165771, "learning_rate": 1.7372794500894615e-05, "loss": 0.17408058643341065, "step": 52150 }, { "epoch": 0.2590822844540695, "grad_norm": 0.7033465504646301, "learning_rate": 1.7371707577978316e-05, "loss": 0.15452935695648193, "step": 52160 }, { "epoch": 0.2591319551374388, "grad_norm": 2.0587310791015625, "learning_rate": 1.737062046428519e-05, "loss": 0.15813124179840088, "step": 52170 }, { "epoch": 0.25918162582080806, "grad_norm": 0.772442102432251, "learning_rate": 1.7369533159843368e-05, "loss": 0.15302019119262694, "step": 52180 }, { "epoch": 0.2592312965041773, "grad_norm": 1.0497170686721802, "learning_rate": 1.736844566468099e-05, "loss": 0.17682113647460937, "step": 52190 }, { "epoch": 0.25928096718754656, "grad_norm": 0.90831059217453, "learning_rate": 1.7367357978826203e-05, "loss": 0.173171067237854, "step": 52200 }, { "epoch": 0.2593306378709158, "grad_norm": 1.1520417928695679, "learning_rate": 1.7366270102307158e-05, "loss": 0.1627523422241211, "step": 52210 }, { "epoch": 0.25938030855428507, "grad_norm": 1.248394250869751, "learning_rate": 1.7365182035151998e-05, "loss": 0.16814143657684327, "step": 52220 }, { "epoch": 0.2594299792376544, "grad_norm": 1.5416423082351685, "learning_rate": 1.736409377738889e-05, "loss": 0.18011070489883424, "step": 52230 }, { "epoch": 0.2594796499210236, "grad_norm": 1.4294081926345825, "learning_rate": 1.7363005329045998e-05, "loss": 0.1715996026992798, "step": 52240 }, { "epoch": 0.2595293206043929, "grad_norm": 0.7428419589996338, "learning_rate": 1.7361916690151486e-05, "loss": 0.16705694198608398, "step": 52250 }, { "epoch": 0.25957899128776213, "grad_norm": 1.160494327545166, "learning_rate": 1.736082786073353e-05, "loss": 0.1756596565246582, "step": 52260 }, { "epoch": 0.2596286619711314, "grad_norm": 1.9977177381515503, "learning_rate": 1.7359738840820313e-05, "loss": 0.17861435413360596, "step": 52270 }, { "epoch": 0.2596783326545007, "grad_norm": 0.8599940538406372, "learning_rate": 1.735864963044001e-05, "loss": 0.16174854040145875, "step": 52280 }, { "epoch": 0.25972800333786994, "grad_norm": 2.1423237323760986, "learning_rate": 1.7357560229620815e-05, "loss": 0.18393423557281494, "step": 52290 }, { "epoch": 0.2597776740212392, "grad_norm": 1.6653692722320557, "learning_rate": 1.7356470638390918e-05, "loss": 0.17426155805587767, "step": 52300 }, { "epoch": 0.25982734470460844, "grad_norm": 1.1329649686813354, "learning_rate": 1.7355380856778517e-05, "loss": 0.19194457530975342, "step": 52310 }, { "epoch": 0.2598770153879777, "grad_norm": 1.2627947330474854, "learning_rate": 1.7354290884811823e-05, "loss": 0.17729077339172364, "step": 52320 }, { "epoch": 0.25992668607134695, "grad_norm": 1.140236496925354, "learning_rate": 1.7353200722519035e-05, "loss": 0.1783225417137146, "step": 52330 }, { "epoch": 0.25997635675471625, "grad_norm": 0.7011925578117371, "learning_rate": 1.735211036992837e-05, "loss": 0.18028159141540528, "step": 52340 }, { "epoch": 0.2600260274380855, "grad_norm": 2.1411399841308594, "learning_rate": 1.735101982706804e-05, "loss": 0.16173384189605713, "step": 52350 }, { "epoch": 0.26007569812145476, "grad_norm": 1.7518818378448486, "learning_rate": 1.7349929093966275e-05, "loss": 0.18860589265823363, "step": 52360 }, { "epoch": 0.260125368804824, "grad_norm": 0.7176386713981628, "learning_rate": 1.7348838170651304e-05, "loss": 0.177806556224823, "step": 52370 }, { "epoch": 0.26017503948819326, "grad_norm": 1.1731021404266357, "learning_rate": 1.734774705715135e-05, "loss": 0.17881826162338257, "step": 52380 }, { "epoch": 0.26022471017156257, "grad_norm": 0.6510473489761353, "learning_rate": 1.734665575349466e-05, "loss": 0.18082802295684813, "step": 52390 }, { "epoch": 0.2602743808549318, "grad_norm": 0.8750069737434387, "learning_rate": 1.7345564259709474e-05, "loss": 0.1757786512374878, "step": 52400 }, { "epoch": 0.26032405153830107, "grad_norm": 0.8873647451400757, "learning_rate": 1.7344472575824036e-05, "loss": 0.16125073432922363, "step": 52410 }, { "epoch": 0.2603737222216703, "grad_norm": 1.1560149192810059, "learning_rate": 1.7343380701866604e-05, "loss": 0.18093887567520142, "step": 52420 }, { "epoch": 0.2604233929050396, "grad_norm": 1.517429232597351, "learning_rate": 1.7342288637865427e-05, "loss": 0.18346513509750367, "step": 52430 }, { "epoch": 0.2604730635884088, "grad_norm": 0.8112846612930298, "learning_rate": 1.7341196383848777e-05, "loss": 0.15480778217315674, "step": 52440 }, { "epoch": 0.26052273427177813, "grad_norm": 1.0489503145217896, "learning_rate": 1.7340103939844914e-05, "loss": 0.1667338490486145, "step": 52450 }, { "epoch": 0.2605724049551474, "grad_norm": 1.8103872537612915, "learning_rate": 1.7339011305882115e-05, "loss": 0.20171551704406737, "step": 52460 }, { "epoch": 0.26062207563851664, "grad_norm": 1.1660643815994263, "learning_rate": 1.733791848198865e-05, "loss": 0.18737947940826416, "step": 52470 }, { "epoch": 0.2606717463218859, "grad_norm": 0.8726718425750732, "learning_rate": 1.733682546819281e-05, "loss": 0.1526463508605957, "step": 52480 }, { "epoch": 0.26072141700525514, "grad_norm": 0.9492872953414917, "learning_rate": 1.7335732264522875e-05, "loss": 0.17573070526123047, "step": 52490 }, { "epoch": 0.26077108768862445, "grad_norm": 2.2232136726379395, "learning_rate": 1.733463887100714e-05, "loss": 0.1794811248779297, "step": 52500 }, { "epoch": 0.2608207583719937, "grad_norm": 1.6068246364593506, "learning_rate": 1.7333545287673896e-05, "loss": 0.16739850044250487, "step": 52510 }, { "epoch": 0.26087042905536295, "grad_norm": 2.338195323944092, "learning_rate": 1.733245151455145e-05, "loss": 0.16902732849121094, "step": 52520 }, { "epoch": 0.2609200997387322, "grad_norm": 2.2185983657836914, "learning_rate": 1.733135755166811e-05, "loss": 0.16767821311950684, "step": 52530 }, { "epoch": 0.26096977042210145, "grad_norm": 0.6899316906929016, "learning_rate": 1.733026339905218e-05, "loss": 0.18152575492858886, "step": 52540 }, { "epoch": 0.2610194411054707, "grad_norm": 1.0524730682373047, "learning_rate": 1.7329169056731982e-05, "loss": 0.17562201023101806, "step": 52550 }, { "epoch": 0.26106911178884, "grad_norm": 1.1457551717758179, "learning_rate": 1.732807452473584e-05, "loss": 0.17461137771606444, "step": 52560 }, { "epoch": 0.26111878247220927, "grad_norm": 0.8033813238143921, "learning_rate": 1.732697980309207e-05, "loss": 0.18007335662841797, "step": 52570 }, { "epoch": 0.2611684531555785, "grad_norm": 0.9439001679420471, "learning_rate": 1.7325884891829012e-05, "loss": 0.17547858953475953, "step": 52580 }, { "epoch": 0.26121812383894777, "grad_norm": 0.897524893283844, "learning_rate": 1.7324789790975e-05, "loss": 0.18502936363220215, "step": 52590 }, { "epoch": 0.261267794522317, "grad_norm": 1.4478013515472412, "learning_rate": 1.732369450055837e-05, "loss": 0.1887621283531189, "step": 52600 }, { "epoch": 0.2613174652056863, "grad_norm": 2.3901140689849854, "learning_rate": 1.732259902060747e-05, "loss": 0.1937548041343689, "step": 52610 }, { "epoch": 0.2613671358890556, "grad_norm": 1.7200394868850708, "learning_rate": 1.7321503351150656e-05, "loss": 0.19860444068908692, "step": 52620 }, { "epoch": 0.26141680657242483, "grad_norm": 1.0082608461380005, "learning_rate": 1.7320407492216277e-05, "loss": 0.16731094121932982, "step": 52630 }, { "epoch": 0.2614664772557941, "grad_norm": 1.1416453123092651, "learning_rate": 1.7319311443832693e-05, "loss": 0.16491949558258057, "step": 52640 }, { "epoch": 0.26151614793916333, "grad_norm": 1.4384584426879883, "learning_rate": 1.7318215206028276e-05, "loss": 0.19083477258682252, "step": 52650 }, { "epoch": 0.2615658186225326, "grad_norm": 1.036371111869812, "learning_rate": 1.731711877883139e-05, "loss": 0.17047032117843627, "step": 52660 }, { "epoch": 0.2616154893059019, "grad_norm": 0.99226313829422, "learning_rate": 1.7316022162270413e-05, "loss": 0.17935739755630492, "step": 52670 }, { "epoch": 0.26166515998927115, "grad_norm": 1.2467888593673706, "learning_rate": 1.7314925356373725e-05, "loss": 0.1872525691986084, "step": 52680 }, { "epoch": 0.2617148306726404, "grad_norm": 1.2602708339691162, "learning_rate": 1.731382836116971e-05, "loss": 0.1865242600440979, "step": 52690 }, { "epoch": 0.26176450135600965, "grad_norm": 0.7425592541694641, "learning_rate": 1.7312731176686752e-05, "loss": 0.1662658452987671, "step": 52700 }, { "epoch": 0.2618141720393789, "grad_norm": 0.7490596175193787, "learning_rate": 1.7311633802953257e-05, "loss": 0.1798640727996826, "step": 52710 }, { "epoch": 0.26186384272274815, "grad_norm": 1.5780333280563354, "learning_rate": 1.731053623999762e-05, "loss": 0.1928853750228882, "step": 52720 }, { "epoch": 0.26191351340611746, "grad_norm": 1.5187418460845947, "learning_rate": 1.730943848784824e-05, "loss": 0.173250150680542, "step": 52730 }, { "epoch": 0.2619631840894867, "grad_norm": 1.1623554229736328, "learning_rate": 1.7308340546533533e-05, "loss": 0.16458505392074585, "step": 52740 }, { "epoch": 0.26201285477285596, "grad_norm": 0.7930631041526794, "learning_rate": 1.7307242416081914e-05, "loss": 0.1760850191116333, "step": 52750 }, { "epoch": 0.2620625254562252, "grad_norm": 1.6163054704666138, "learning_rate": 1.7306144096521797e-05, "loss": 0.17292280197143556, "step": 52760 }, { "epoch": 0.26211219613959447, "grad_norm": 2.328458070755005, "learning_rate": 1.7305045587881605e-05, "loss": 0.1949997663497925, "step": 52770 }, { "epoch": 0.2621618668229638, "grad_norm": 0.8358714580535889, "learning_rate": 1.7303946890189775e-05, "loss": 0.19627634286880494, "step": 52780 }, { "epoch": 0.262211537506333, "grad_norm": 1.588448405265808, "learning_rate": 1.7302848003474737e-05, "loss": 0.16826765537261962, "step": 52790 }, { "epoch": 0.2622612081897023, "grad_norm": 1.6889899969100952, "learning_rate": 1.7301748927764927e-05, "loss": 0.17243742942810059, "step": 52800 }, { "epoch": 0.26231087887307153, "grad_norm": 0.8522118330001831, "learning_rate": 1.7300649663088788e-05, "loss": 0.1630040407180786, "step": 52810 }, { "epoch": 0.2623605495564408, "grad_norm": 2.121258497238159, "learning_rate": 1.7299550209474774e-05, "loss": 0.1764759063720703, "step": 52820 }, { "epoch": 0.26241022023981003, "grad_norm": 1.1237764358520508, "learning_rate": 1.7298450566951336e-05, "loss": 0.16228629350662233, "step": 52830 }, { "epoch": 0.26245989092317934, "grad_norm": 0.6181142926216125, "learning_rate": 1.7297350735546933e-05, "loss": 0.18760255575180054, "step": 52840 }, { "epoch": 0.2625095616065486, "grad_norm": 1.613465428352356, "learning_rate": 1.7296250715290026e-05, "loss": 0.17728344202041627, "step": 52850 }, { "epoch": 0.26255923228991784, "grad_norm": 1.2743347883224487, "learning_rate": 1.7295150506209085e-05, "loss": 0.19008238315582277, "step": 52860 }, { "epoch": 0.2626089029732871, "grad_norm": 1.2657135725021362, "learning_rate": 1.7294050108332577e-05, "loss": 0.16797966957092286, "step": 52870 }, { "epoch": 0.26265857365665635, "grad_norm": 1.8111088275909424, "learning_rate": 1.7292949521688994e-05, "loss": 0.17162188291549682, "step": 52880 }, { "epoch": 0.26270824434002565, "grad_norm": 1.1397374868392944, "learning_rate": 1.7291848746306802e-05, "loss": 0.17416033744812012, "step": 52890 }, { "epoch": 0.2627579150233949, "grad_norm": 1.9322794675827026, "learning_rate": 1.72907477822145e-05, "loss": 0.18928127288818358, "step": 52900 }, { "epoch": 0.26280758570676416, "grad_norm": 1.073604702949524, "learning_rate": 1.7289646629440575e-05, "loss": 0.1806863069534302, "step": 52910 }, { "epoch": 0.2628572563901334, "grad_norm": 0.941963255405426, "learning_rate": 1.728854528801353e-05, "loss": 0.17097082138061523, "step": 52920 }, { "epoch": 0.26290692707350266, "grad_norm": 0.8425620198249817, "learning_rate": 1.728744375796186e-05, "loss": 0.1618281364440918, "step": 52930 }, { "epoch": 0.2629565977568719, "grad_norm": 2.310474157333374, "learning_rate": 1.7286342039314074e-05, "loss": 0.18298968076705932, "step": 52940 }, { "epoch": 0.2630062684402412, "grad_norm": 0.9692542552947998, "learning_rate": 1.728524013209869e-05, "loss": 0.16418148279190065, "step": 52950 }, { "epoch": 0.26305593912361047, "grad_norm": 0.9323384165763855, "learning_rate": 1.728413803634422e-05, "loss": 0.17212268114089965, "step": 52960 }, { "epoch": 0.2631056098069797, "grad_norm": 1.114420771598816, "learning_rate": 1.728303575207918e-05, "loss": 0.16759516000747682, "step": 52970 }, { "epoch": 0.263155280490349, "grad_norm": 1.2103374004364014, "learning_rate": 1.728193327933211e-05, "loss": 0.1660060167312622, "step": 52980 }, { "epoch": 0.2632049511737182, "grad_norm": 1.2126046419143677, "learning_rate": 1.7280830618131537e-05, "loss": 0.18617761135101318, "step": 52990 }, { "epoch": 0.26325462185708753, "grad_norm": 0.8160831332206726, "learning_rate": 1.727972776850599e-05, "loss": 0.1915677785873413, "step": 53000 }, { "epoch": 0.2633042925404568, "grad_norm": 1.6022968292236328, "learning_rate": 1.7278624730484016e-05, "loss": 0.1723534107208252, "step": 53010 }, { "epoch": 0.26335396322382604, "grad_norm": 0.7452788949012756, "learning_rate": 1.7277521504094165e-05, "loss": 0.1826815128326416, "step": 53020 }, { "epoch": 0.2634036339071953, "grad_norm": 2.259714365005493, "learning_rate": 1.7276418089364977e-05, "loss": 0.1780511736869812, "step": 53030 }, { "epoch": 0.26345330459056454, "grad_norm": 1.6821637153625488, "learning_rate": 1.727531448632502e-05, "loss": 0.17130850553512572, "step": 53040 }, { "epoch": 0.2635029752739338, "grad_norm": 1.1538337469100952, "learning_rate": 1.7274210695002844e-05, "loss": 0.17154008150100708, "step": 53050 }, { "epoch": 0.2635526459573031, "grad_norm": 2.6603922843933105, "learning_rate": 1.7273106715427025e-05, "loss": 0.16100537776947021, "step": 53060 }, { "epoch": 0.26360231664067235, "grad_norm": 0.86137855052948, "learning_rate": 1.727200254762612e-05, "loss": 0.18367501497268676, "step": 53070 }, { "epoch": 0.2636519873240416, "grad_norm": 0.9975249767303467, "learning_rate": 1.7270898191628725e-05, "loss": 0.18229187726974488, "step": 53080 }, { "epoch": 0.26370165800741086, "grad_norm": 1.918545126914978, "learning_rate": 1.7269793647463406e-05, "loss": 0.18188624382019042, "step": 53090 }, { "epoch": 0.2637513286907801, "grad_norm": 1.5997107028961182, "learning_rate": 1.7268688915158743e-05, "loss": 0.17874966859817504, "step": 53100 }, { "epoch": 0.2638009993741494, "grad_norm": 1.1590020656585693, "learning_rate": 1.726758399474334e-05, "loss": 0.1697966933250427, "step": 53110 }, { "epoch": 0.26385067005751867, "grad_norm": 2.5240156650543213, "learning_rate": 1.7266478886245783e-05, "loss": 0.16955626010894775, "step": 53120 }, { "epoch": 0.2639003407408879, "grad_norm": 1.8304188251495361, "learning_rate": 1.726537358969467e-05, "loss": 0.1681215763092041, "step": 53130 }, { "epoch": 0.26395001142425717, "grad_norm": 1.3584814071655273, "learning_rate": 1.7264268105118617e-05, "loss": 0.15891661643981933, "step": 53140 }, { "epoch": 0.2639996821076264, "grad_norm": 1.04225492477417, "learning_rate": 1.7263162432546225e-05, "loss": 0.1712325096130371, "step": 53150 }, { "epoch": 0.2640493527909957, "grad_norm": 1.0602377653121948, "learning_rate": 1.7262056572006108e-05, "loss": 0.18587054014205934, "step": 53160 }, { "epoch": 0.264099023474365, "grad_norm": 0.6588356494903564, "learning_rate": 1.7260950523526886e-05, "loss": 0.18154449462890626, "step": 53170 }, { "epoch": 0.26414869415773423, "grad_norm": 1.4370523691177368, "learning_rate": 1.7259844287137183e-05, "loss": 0.18427984714508056, "step": 53180 }, { "epoch": 0.2641983648411035, "grad_norm": 0.8211039304733276, "learning_rate": 1.7258737862865632e-05, "loss": 0.16911323070526124, "step": 53190 }, { "epoch": 0.26424803552447274, "grad_norm": 0.9176629781723022, "learning_rate": 1.7257631250740862e-05, "loss": 0.18252137899398804, "step": 53200 }, { "epoch": 0.264297706207842, "grad_norm": 1.5462870597839355, "learning_rate": 1.7256524450791513e-05, "loss": 0.16332333087921141, "step": 53210 }, { "epoch": 0.2643473768912113, "grad_norm": 0.5975703001022339, "learning_rate": 1.725541746304623e-05, "loss": 0.15944103002548218, "step": 53220 }, { "epoch": 0.26439704757458055, "grad_norm": 0.7242865562438965, "learning_rate": 1.725431028753366e-05, "loss": 0.1713798761367798, "step": 53230 }, { "epoch": 0.2644467182579498, "grad_norm": 1.6307661533355713, "learning_rate": 1.725320292428246e-05, "loss": 0.17923927307128906, "step": 53240 }, { "epoch": 0.26449638894131905, "grad_norm": 0.658058762550354, "learning_rate": 1.7252095373321277e-05, "loss": 0.18420082330703735, "step": 53250 }, { "epoch": 0.2645460596246883, "grad_norm": 1.6319732666015625, "learning_rate": 1.7250987634678788e-05, "loss": 0.19161605834960938, "step": 53260 }, { "epoch": 0.26459573030805755, "grad_norm": 2.056824207305908, "learning_rate": 1.7249879708383654e-05, "loss": 0.18104496002197265, "step": 53270 }, { "epoch": 0.26464540099142686, "grad_norm": 1.8655256032943726, "learning_rate": 1.7248771594464545e-05, "loss": 0.18201466798782348, "step": 53280 }, { "epoch": 0.2646950716747961, "grad_norm": 1.1289578676223755, "learning_rate": 1.7247663292950146e-05, "loss": 0.18420807123184205, "step": 53290 }, { "epoch": 0.26474474235816536, "grad_norm": 1.7396740913391113, "learning_rate": 1.7246554803869134e-05, "loss": 0.1815027952194214, "step": 53300 }, { "epoch": 0.2647944130415346, "grad_norm": 1.320320725440979, "learning_rate": 1.7245446127250195e-05, "loss": 0.18419361114501953, "step": 53310 }, { "epoch": 0.26484408372490387, "grad_norm": 1.7181800603866577, "learning_rate": 1.7244337263122025e-05, "loss": 0.1825173616409302, "step": 53320 }, { "epoch": 0.2648937544082732, "grad_norm": 1.6830025911331177, "learning_rate": 1.724322821151332e-05, "loss": 0.1749272346496582, "step": 53330 }, { "epoch": 0.2649434250916424, "grad_norm": 1.963024616241455, "learning_rate": 1.724211897245278e-05, "loss": 0.17053890228271484, "step": 53340 }, { "epoch": 0.2649930957750117, "grad_norm": 1.1539602279663086, "learning_rate": 1.7241009545969114e-05, "loss": 0.17879948616027833, "step": 53350 }, { "epoch": 0.26504276645838093, "grad_norm": 0.9239482879638672, "learning_rate": 1.723989993209103e-05, "loss": 0.17093701362609864, "step": 53360 }, { "epoch": 0.2650924371417502, "grad_norm": 1.5094434022903442, "learning_rate": 1.723879013084725e-05, "loss": 0.18143625259399415, "step": 53370 }, { "epoch": 0.26514210782511943, "grad_norm": 0.9495876431465149, "learning_rate": 1.723768014226649e-05, "loss": 0.18336913585662842, "step": 53380 }, { "epoch": 0.26519177850848874, "grad_norm": 1.635617733001709, "learning_rate": 1.7236569966377476e-05, "loss": 0.1714223265647888, "step": 53390 }, { "epoch": 0.265241449191858, "grad_norm": 2.204902172088623, "learning_rate": 1.7235459603208946e-05, "loss": 0.1681402325630188, "step": 53400 }, { "epoch": 0.26529111987522724, "grad_norm": 0.8750771880149841, "learning_rate": 1.7234349052789627e-05, "loss": 0.1746676445007324, "step": 53410 }, { "epoch": 0.2653407905585965, "grad_norm": 1.0277900695800781, "learning_rate": 1.7233238315148268e-05, "loss": 0.1718266487121582, "step": 53420 }, { "epoch": 0.26539046124196575, "grad_norm": 2.740875482559204, "learning_rate": 1.7232127390313603e-05, "loss": 0.18102191686630248, "step": 53430 }, { "epoch": 0.26544013192533505, "grad_norm": 1.479170799255371, "learning_rate": 1.7231016278314395e-05, "loss": 0.17344536781311035, "step": 53440 }, { "epoch": 0.2654898026087043, "grad_norm": 1.5786430835723877, "learning_rate": 1.7229904979179388e-05, "loss": 0.16274690628051758, "step": 53450 }, { "epoch": 0.26553947329207356, "grad_norm": 0.6803237795829773, "learning_rate": 1.7228793492937354e-05, "loss": 0.170514714717865, "step": 53460 }, { "epoch": 0.2655891439754428, "grad_norm": 2.4660263061523438, "learning_rate": 1.7227681819617044e-05, "loss": 0.17971343994140626, "step": 53470 }, { "epoch": 0.26563881465881206, "grad_norm": 1.1994105577468872, "learning_rate": 1.722656995924724e-05, "loss": 0.1699626088142395, "step": 53480 }, { "epoch": 0.2656884853421813, "grad_norm": 0.856867253780365, "learning_rate": 1.7225457911856706e-05, "loss": 0.17460236549377442, "step": 53490 }, { "epoch": 0.2657381560255506, "grad_norm": 1.638921856880188, "learning_rate": 1.7224345677474227e-05, "loss": 0.18415088653564454, "step": 53500 }, { "epoch": 0.26578782670891987, "grad_norm": 1.0860798358917236, "learning_rate": 1.7223233256128592e-05, "loss": 0.16994423866271974, "step": 53510 }, { "epoch": 0.2658374973922891, "grad_norm": 1.2022253274917603, "learning_rate": 1.7222120647848583e-05, "loss": 0.1590356707572937, "step": 53520 }, { "epoch": 0.2658871680756584, "grad_norm": 1.6894549131393433, "learning_rate": 1.722100785266299e-05, "loss": 0.17079622745513917, "step": 53530 }, { "epoch": 0.2659368387590276, "grad_norm": 1.0204551219940186, "learning_rate": 1.7219894870600624e-05, "loss": 0.1777796268463135, "step": 53540 }, { "epoch": 0.26598650944239693, "grad_norm": 0.9572424292564392, "learning_rate": 1.7218781701690276e-05, "loss": 0.17015383243560792, "step": 53550 }, { "epoch": 0.2660361801257662, "grad_norm": 1.1177316904067993, "learning_rate": 1.7217668345960767e-05, "loss": 0.18108412027359008, "step": 53560 }, { "epoch": 0.26608585080913544, "grad_norm": 1.2692387104034424, "learning_rate": 1.7216554803440896e-05, "loss": 0.17653837203979492, "step": 53570 }, { "epoch": 0.2661355214925047, "grad_norm": 1.0756862163543701, "learning_rate": 1.7215441074159493e-05, "loss": 0.17023079395294188, "step": 53580 }, { "epoch": 0.26618519217587394, "grad_norm": 1.0465190410614014, "learning_rate": 1.7214327158145374e-05, "loss": 0.17327163219451905, "step": 53590 }, { "epoch": 0.2662348628592432, "grad_norm": 1.2714606523513794, "learning_rate": 1.721321305542737e-05, "loss": 0.1797341823577881, "step": 53600 }, { "epoch": 0.2662845335426125, "grad_norm": 0.8292521834373474, "learning_rate": 1.7212098766034312e-05, "loss": 0.16899985074996948, "step": 53610 }, { "epoch": 0.26633420422598175, "grad_norm": 2.0885512828826904, "learning_rate": 1.721098428999504e-05, "loss": 0.16878597736358641, "step": 53620 }, { "epoch": 0.266383874909351, "grad_norm": 1.0278695821762085, "learning_rate": 1.7209869627338392e-05, "loss": 0.18351035118103026, "step": 53630 }, { "epoch": 0.26643354559272026, "grad_norm": 0.7068997025489807, "learning_rate": 1.7208754778093215e-05, "loss": 0.17088804244995118, "step": 53640 }, { "epoch": 0.2664832162760895, "grad_norm": 1.0957424640655518, "learning_rate": 1.7207639742288366e-05, "loss": 0.1883960485458374, "step": 53650 }, { "epoch": 0.2665328869594588, "grad_norm": 1.2014179229736328, "learning_rate": 1.7206524519952697e-05, "loss": 0.17805640697479247, "step": 53660 }, { "epoch": 0.26658255764282807, "grad_norm": 1.6317360401153564, "learning_rate": 1.7205409111115074e-05, "loss": 0.17794570922851563, "step": 53670 }, { "epoch": 0.2666322283261973, "grad_norm": 1.1795517206192017, "learning_rate": 1.7204293515804357e-05, "loss": 0.18842897415161133, "step": 53680 }, { "epoch": 0.26668189900956657, "grad_norm": 0.9766896367073059, "learning_rate": 1.720317773404942e-05, "loss": 0.1803992509841919, "step": 53690 }, { "epoch": 0.2667315696929358, "grad_norm": 1.3513020277023315, "learning_rate": 1.7202061765879143e-05, "loss": 0.17606449127197266, "step": 53700 }, { "epoch": 0.2667812403763051, "grad_norm": 1.142295241355896, "learning_rate": 1.72009456113224e-05, "loss": 0.17737457752227784, "step": 53710 }, { "epoch": 0.2668309110596744, "grad_norm": 1.2038558721542358, "learning_rate": 1.7199829270408082e-05, "loss": 0.17854664325714112, "step": 53720 }, { "epoch": 0.26688058174304363, "grad_norm": 2.248079299926758, "learning_rate": 1.7198712743165076e-05, "loss": 0.17064390182495118, "step": 53730 }, { "epoch": 0.2669302524264129, "grad_norm": 1.4276388883590698, "learning_rate": 1.719759602962228e-05, "loss": 0.18057916164398194, "step": 53740 }, { "epoch": 0.26697992310978214, "grad_norm": 1.4622727632522583, "learning_rate": 1.719647912980859e-05, "loss": 0.1752075433731079, "step": 53750 }, { "epoch": 0.2670295937931514, "grad_norm": 0.696829617023468, "learning_rate": 1.7195362043752915e-05, "loss": 0.16714277267456054, "step": 53760 }, { "epoch": 0.2670792644765207, "grad_norm": 1.3362783193588257, "learning_rate": 1.7194244771484167e-05, "loss": 0.1639033317565918, "step": 53770 }, { "epoch": 0.26712893515988995, "grad_norm": 1.1920561790466309, "learning_rate": 1.719312731303125e-05, "loss": 0.16440922021865845, "step": 53780 }, { "epoch": 0.2671786058432592, "grad_norm": 0.7748225331306458, "learning_rate": 1.7192009668423096e-05, "loss": 0.16084165573120118, "step": 53790 }, { "epoch": 0.26722827652662845, "grad_norm": 3.1783273220062256, "learning_rate": 1.7190891837688622e-05, "loss": 0.17378010749816894, "step": 53800 }, { "epoch": 0.2672779472099977, "grad_norm": 1.1866308450698853, "learning_rate": 1.718977382085676e-05, "loss": 0.17785636186599732, "step": 53810 }, { "epoch": 0.26732761789336695, "grad_norm": 2.255587339401245, "learning_rate": 1.718865561795644e-05, "loss": 0.17491350173950196, "step": 53820 }, { "epoch": 0.26737728857673626, "grad_norm": 1.559523582458496, "learning_rate": 1.7187537229016607e-05, "loss": 0.17207133769989014, "step": 53830 }, { "epoch": 0.2674269592601055, "grad_norm": 2.5532193183898926, "learning_rate": 1.71864186540662e-05, "loss": 0.17410348653793334, "step": 53840 }, { "epoch": 0.26747662994347476, "grad_norm": 1.8680604696273804, "learning_rate": 1.7185299893134166e-05, "loss": 0.1534122109413147, "step": 53850 }, { "epoch": 0.267526300626844, "grad_norm": 0.6341091394424438, "learning_rate": 1.7184180946249456e-05, "loss": 0.1668158531188965, "step": 53860 }, { "epoch": 0.26757597131021327, "grad_norm": 0.8056943416595459, "learning_rate": 1.7183061813441035e-05, "loss": 0.1643507480621338, "step": 53870 }, { "epoch": 0.2676256419935826, "grad_norm": 1.4897552728652954, "learning_rate": 1.718194249473786e-05, "loss": 0.17605527639389038, "step": 53880 }, { "epoch": 0.2676753126769518, "grad_norm": 2.204970359802246, "learning_rate": 1.7180822990168908e-05, "loss": 0.16753756999969482, "step": 53890 }, { "epoch": 0.2677249833603211, "grad_norm": 1.9593240022659302, "learning_rate": 1.7179703299763136e-05, "loss": 0.15838621854782103, "step": 53900 }, { "epoch": 0.26777465404369033, "grad_norm": 1.1314072608947754, "learning_rate": 1.7178583423549534e-05, "loss": 0.1794879674911499, "step": 53910 }, { "epoch": 0.2678243247270596, "grad_norm": 1.347601294517517, "learning_rate": 1.7177463361557082e-05, "loss": 0.17487025260925293, "step": 53920 }, { "epoch": 0.26787399541042883, "grad_norm": 0.8076694011688232, "learning_rate": 1.717634311381476e-05, "loss": 0.16566730737686158, "step": 53930 }, { "epoch": 0.26792366609379814, "grad_norm": 1.2808446884155273, "learning_rate": 1.7175222680351566e-05, "loss": 0.17587536573410034, "step": 53940 }, { "epoch": 0.2679733367771674, "grad_norm": 1.4008532762527466, "learning_rate": 1.717410206119649e-05, "loss": 0.17062443494796753, "step": 53950 }, { "epoch": 0.26802300746053664, "grad_norm": 1.4441242218017578, "learning_rate": 1.717298125637854e-05, "loss": 0.18097999095916747, "step": 53960 }, { "epoch": 0.2680726781439059, "grad_norm": 2.3701791763305664, "learning_rate": 1.717186026592672e-05, "loss": 0.1722898006439209, "step": 53970 }, { "epoch": 0.26812234882727515, "grad_norm": 1.292741060256958, "learning_rate": 1.7170739089870038e-05, "loss": 0.17458395957946776, "step": 53980 }, { "epoch": 0.26817201951064445, "grad_norm": 1.303012728691101, "learning_rate": 1.7169617728237514e-05, "loss": 0.1783048391342163, "step": 53990 }, { "epoch": 0.2682216901940137, "grad_norm": 1.7677456140518188, "learning_rate": 1.7168496181058164e-05, "loss": 0.19224562644958496, "step": 54000 }, { "epoch": 0.26827136087738296, "grad_norm": 1.3384623527526855, "learning_rate": 1.7167374448361017e-05, "loss": 0.17087196111679076, "step": 54010 }, { "epoch": 0.2683210315607522, "grad_norm": 1.9661058187484741, "learning_rate": 1.7166252530175097e-05, "loss": 0.1817354679107666, "step": 54020 }, { "epoch": 0.26837070224412146, "grad_norm": 0.88618403673172, "learning_rate": 1.716513042652945e-05, "loss": 0.1732199788093567, "step": 54030 }, { "epoch": 0.2684203729274907, "grad_norm": 1.1287782192230225, "learning_rate": 1.71640081374531e-05, "loss": 0.17341501712799073, "step": 54040 }, { "epoch": 0.26847004361086, "grad_norm": 2.2104551792144775, "learning_rate": 1.7162885662975107e-05, "loss": 0.18015815019607545, "step": 54050 }, { "epoch": 0.2685197142942293, "grad_norm": 1.3279571533203125, "learning_rate": 1.716176300312451e-05, "loss": 0.16410841941833496, "step": 54060 }, { "epoch": 0.2685693849775985, "grad_norm": 1.7254050970077515, "learning_rate": 1.7160640157930366e-05, "loss": 0.19486724138259887, "step": 54070 }, { "epoch": 0.2686190556609678, "grad_norm": 1.1715525388717651, "learning_rate": 1.7159517127421735e-05, "loss": 0.18190664052963257, "step": 54080 }, { "epoch": 0.268668726344337, "grad_norm": 0.9120535850524902, "learning_rate": 1.715839391162768e-05, "loss": 0.1970674753189087, "step": 54090 }, { "epoch": 0.2687183970277063, "grad_norm": 0.55275958776474, "learning_rate": 1.7157270510577265e-05, "loss": 0.1726043701171875, "step": 54100 }, { "epoch": 0.2687680677110756, "grad_norm": 1.3296937942504883, "learning_rate": 1.715614692429957e-05, "loss": 0.15709137916564941, "step": 54110 }, { "epoch": 0.26881773839444484, "grad_norm": 2.3546717166900635, "learning_rate": 1.7155023152823674e-05, "loss": 0.1770092487335205, "step": 54120 }, { "epoch": 0.2688674090778141, "grad_norm": 0.9455515742301941, "learning_rate": 1.715389919617865e-05, "loss": 0.16651042699813842, "step": 54130 }, { "epoch": 0.26891707976118334, "grad_norm": 0.8957271575927734, "learning_rate": 1.715277505439359e-05, "loss": 0.16517479419708253, "step": 54140 }, { "epoch": 0.2689667504445526, "grad_norm": 1.2592378854751587, "learning_rate": 1.715165072749759e-05, "loss": 0.1762521743774414, "step": 54150 }, { "epoch": 0.2690164211279219, "grad_norm": 1.592621922492981, "learning_rate": 1.7150526215519746e-05, "loss": 0.18624380826950074, "step": 54160 }, { "epoch": 0.26906609181129115, "grad_norm": 0.7953356504440308, "learning_rate": 1.714940151848916e-05, "loss": 0.16151853799819946, "step": 54170 }, { "epoch": 0.2691157624946604, "grad_norm": 1.5950770378112793, "learning_rate": 1.7148276636434933e-05, "loss": 0.16793596744537354, "step": 54180 }, { "epoch": 0.26916543317802966, "grad_norm": 2.2617263793945312, "learning_rate": 1.7147151569386186e-05, "loss": 0.16884148120880127, "step": 54190 }, { "epoch": 0.2692151038613989, "grad_norm": 1.5162056684494019, "learning_rate": 1.7146026317372023e-05, "loss": 0.17792530059814454, "step": 54200 }, { "epoch": 0.26926477454476816, "grad_norm": 1.3137874603271484, "learning_rate": 1.714490088042158e-05, "loss": 0.17583647966384888, "step": 54210 }, { "epoch": 0.26931444522813747, "grad_norm": 1.2767517566680908, "learning_rate": 1.714377525856397e-05, "loss": 0.17277944087982178, "step": 54220 }, { "epoch": 0.2693641159115067, "grad_norm": 2.3009073734283447, "learning_rate": 1.714264945182833e-05, "loss": 0.17591633796691894, "step": 54230 }, { "epoch": 0.26941378659487597, "grad_norm": 0.9438679814338684, "learning_rate": 1.7141523460243793e-05, "loss": 0.1742849588394165, "step": 54240 }, { "epoch": 0.2694634572782452, "grad_norm": 2.0401227474212646, "learning_rate": 1.7140397283839502e-05, "loss": 0.17774858474731445, "step": 54250 }, { "epoch": 0.2695131279616145, "grad_norm": 1.6767653226852417, "learning_rate": 1.7139270922644602e-05, "loss": 0.17846983671188354, "step": 54260 }, { "epoch": 0.2695627986449838, "grad_norm": 1.2974538803100586, "learning_rate": 1.713814437668824e-05, "loss": 0.17031512260437012, "step": 54270 }, { "epoch": 0.26961246932835303, "grad_norm": 1.3956875801086426, "learning_rate": 1.7137017645999566e-05, "loss": 0.17471860647201537, "step": 54280 }, { "epoch": 0.2696621400117223, "grad_norm": 0.9031305313110352, "learning_rate": 1.7135890730607753e-05, "loss": 0.16936087608337402, "step": 54290 }, { "epoch": 0.26971181069509154, "grad_norm": 2.33648419380188, "learning_rate": 1.7134763630541954e-05, "loss": 0.16920793056488037, "step": 54300 }, { "epoch": 0.2697614813784608, "grad_norm": 1.08368718624115, "learning_rate": 1.7133636345831338e-05, "loss": 0.17162609100341797, "step": 54310 }, { "epoch": 0.26981115206183004, "grad_norm": 0.7161290645599365, "learning_rate": 1.7132508876505084e-05, "loss": 0.16216087341308594, "step": 54320 }, { "epoch": 0.26986082274519935, "grad_norm": 1.3074445724487305, "learning_rate": 1.713138122259237e-05, "loss": 0.1800453782081604, "step": 54330 }, { "epoch": 0.2699104934285686, "grad_norm": 1.1141225099563599, "learning_rate": 1.7130253384122376e-05, "loss": 0.1743839979171753, "step": 54340 }, { "epoch": 0.26996016411193785, "grad_norm": 0.6691555380821228, "learning_rate": 1.712912536112429e-05, "loss": 0.16657763719558716, "step": 54350 }, { "epoch": 0.2700098347953071, "grad_norm": 1.0198692083358765, "learning_rate": 1.712799715362731e-05, "loss": 0.17064037322998046, "step": 54360 }, { "epoch": 0.27005950547867635, "grad_norm": 1.4281821250915527, "learning_rate": 1.7126868761660628e-05, "loss": 0.17285451889038086, "step": 54370 }, { "epoch": 0.27010917616204566, "grad_norm": 2.0416181087493896, "learning_rate": 1.7125740185253445e-05, "loss": 0.169089674949646, "step": 54380 }, { "epoch": 0.2701588468454149, "grad_norm": 1.0188909769058228, "learning_rate": 1.7124611424434975e-05, "loss": 0.16787145137786866, "step": 54390 }, { "epoch": 0.27020851752878416, "grad_norm": 0.8987677097320557, "learning_rate": 1.7123482479234423e-05, "loss": 0.1809459686279297, "step": 54400 }, { "epoch": 0.2702581882121534, "grad_norm": 1.320034146308899, "learning_rate": 1.712235334968101e-05, "loss": 0.18139078617095947, "step": 54410 }, { "epoch": 0.27030785889552267, "grad_norm": 1.499661922454834, "learning_rate": 1.7121224035803958e-05, "loss": 0.17358356714248657, "step": 54420 }, { "epoch": 0.2703575295788919, "grad_norm": 1.0347342491149902, "learning_rate": 1.712009453763249e-05, "loss": 0.16713377237319946, "step": 54430 }, { "epoch": 0.2704072002622612, "grad_norm": 2.1018779277801514, "learning_rate": 1.7118964855195836e-05, "loss": 0.17609307765960694, "step": 54440 }, { "epoch": 0.2704568709456305, "grad_norm": 0.9937911629676819, "learning_rate": 1.7117834988523235e-05, "loss": 0.17094610929489135, "step": 54450 }, { "epoch": 0.27050654162899973, "grad_norm": 1.1302978992462158, "learning_rate": 1.711670493764393e-05, "loss": 0.16857727766036987, "step": 54460 }, { "epoch": 0.270556212312369, "grad_norm": 1.8968969583511353, "learning_rate": 1.7115574702587156e-05, "loss": 0.1687731385231018, "step": 54470 }, { "epoch": 0.27060588299573823, "grad_norm": 0.971251368522644, "learning_rate": 1.7114444283382174e-05, "loss": 0.18444499969482422, "step": 54480 }, { "epoch": 0.27065555367910754, "grad_norm": 0.8395819067955017, "learning_rate": 1.7113313680058233e-05, "loss": 0.176447331905365, "step": 54490 }, { "epoch": 0.2707052243624768, "grad_norm": 1.5153390169143677, "learning_rate": 1.7112182892644597e-05, "loss": 0.1787981867790222, "step": 54500 }, { "epoch": 0.27075489504584604, "grad_norm": 1.075696587562561, "learning_rate": 1.7111051921170525e-05, "loss": 0.16882641315460206, "step": 54510 }, { "epoch": 0.2708045657292153, "grad_norm": 0.6436941623687744, "learning_rate": 1.710992076566529e-05, "loss": 0.17638533115386962, "step": 54520 }, { "epoch": 0.27085423641258455, "grad_norm": 0.6007490158081055, "learning_rate": 1.7108789426158164e-05, "loss": 0.16131293773651123, "step": 54530 }, { "epoch": 0.2709039070959538, "grad_norm": 1.3369144201278687, "learning_rate": 1.7107657902678426e-05, "loss": 0.17926430702209473, "step": 54540 }, { "epoch": 0.2709535777793231, "grad_norm": 0.7130588889122009, "learning_rate": 1.7106526195255356e-05, "loss": 0.16757769584655763, "step": 54550 }, { "epoch": 0.27100324846269236, "grad_norm": 1.8429147005081177, "learning_rate": 1.710539430391825e-05, "loss": 0.1666066527366638, "step": 54560 }, { "epoch": 0.2710529191460616, "grad_norm": 1.3279889822006226, "learning_rate": 1.7104262228696393e-05, "loss": 0.1700491189956665, "step": 54570 }, { "epoch": 0.27110258982943086, "grad_norm": 0.7925785183906555, "learning_rate": 1.7103129969619087e-05, "loss": 0.1644596576690674, "step": 54580 }, { "epoch": 0.2711522605128001, "grad_norm": 0.7080703377723694, "learning_rate": 1.7101997526715636e-05, "loss": 0.16981172561645508, "step": 54590 }, { "epoch": 0.2712019311961694, "grad_norm": 2.5388920307159424, "learning_rate": 1.7100864900015345e-05, "loss": 0.19962880611419678, "step": 54600 }, { "epoch": 0.2712516018795387, "grad_norm": 1.7847990989685059, "learning_rate": 1.709973208954752e-05, "loss": 0.17336666584014893, "step": 54610 }, { "epoch": 0.2713012725629079, "grad_norm": 1.3941075801849365, "learning_rate": 1.709859909534149e-05, "loss": 0.17775697708129884, "step": 54620 }, { "epoch": 0.2713509432462772, "grad_norm": 0.9937261939048767, "learning_rate": 1.7097465917426563e-05, "loss": 0.16975226402282714, "step": 54630 }, { "epoch": 0.27140061392964643, "grad_norm": 1.3754205703735352, "learning_rate": 1.709633255583208e-05, "loss": 0.17966227531433104, "step": 54640 }, { "epoch": 0.2714502846130157, "grad_norm": 1.3179564476013184, "learning_rate": 1.7095199010587355e-05, "loss": 0.16766541004180907, "step": 54650 }, { "epoch": 0.271499955296385, "grad_norm": 1.0854277610778809, "learning_rate": 1.7094065281721736e-05, "loss": 0.1752028465270996, "step": 54660 }, { "epoch": 0.27154962597975424, "grad_norm": 0.972032904624939, "learning_rate": 1.709293136926456e-05, "loss": 0.16800482273101808, "step": 54670 }, { "epoch": 0.2715992966631235, "grad_norm": 2.088458299636841, "learning_rate": 1.7091797273245175e-05, "loss": 0.18386540412902833, "step": 54680 }, { "epoch": 0.27164896734649274, "grad_norm": 1.0043418407440186, "learning_rate": 1.7090662993692923e-05, "loss": 0.17212934494018556, "step": 54690 }, { "epoch": 0.271698638029862, "grad_norm": 1.4405324459075928, "learning_rate": 1.7089528530637166e-05, "loss": 0.18154873847961425, "step": 54700 }, { "epoch": 0.2717483087132313, "grad_norm": 0.8798256516456604, "learning_rate": 1.7088393884107262e-05, "loss": 0.18039950132369995, "step": 54710 }, { "epoch": 0.27179797939660055, "grad_norm": 2.6069350242614746, "learning_rate": 1.708725905413257e-05, "loss": 0.18372318744659424, "step": 54720 }, { "epoch": 0.2718476500799698, "grad_norm": 1.6845016479492188, "learning_rate": 1.7086124040742472e-05, "loss": 0.17575632333755492, "step": 54730 }, { "epoch": 0.27189732076333906, "grad_norm": 2.4092347621917725, "learning_rate": 1.7084988843966326e-05, "loss": 0.16960394382476807, "step": 54740 }, { "epoch": 0.2719469914467083, "grad_norm": 0.8348323702812195, "learning_rate": 1.708385346383352e-05, "loss": 0.1749920964241028, "step": 54750 }, { "epoch": 0.27199666213007756, "grad_norm": 0.6571294069290161, "learning_rate": 1.7082717900373432e-05, "loss": 0.16949038505554198, "step": 54760 }, { "epoch": 0.27204633281344687, "grad_norm": 1.4875651597976685, "learning_rate": 1.7081582153615456e-05, "loss": 0.1802459716796875, "step": 54770 }, { "epoch": 0.2720960034968161, "grad_norm": 1.789332389831543, "learning_rate": 1.708044622358898e-05, "loss": 0.16182855367660523, "step": 54780 }, { "epoch": 0.27214567418018537, "grad_norm": 1.0281633138656616, "learning_rate": 1.7079310110323404e-05, "loss": 0.17274340391159057, "step": 54790 }, { "epoch": 0.2721953448635546, "grad_norm": 0.9463455080986023, "learning_rate": 1.7078173813848126e-05, "loss": 0.17014656066894532, "step": 54800 }, { "epoch": 0.2722450155469239, "grad_norm": 1.3641334772109985, "learning_rate": 1.707703733419256e-05, "loss": 0.18067286014556885, "step": 54810 }, { "epoch": 0.2722946862302932, "grad_norm": 0.7205398678779602, "learning_rate": 1.707590067138611e-05, "loss": 0.1804060935974121, "step": 54820 }, { "epoch": 0.27234435691366243, "grad_norm": 0.6650072932243347, "learning_rate": 1.7074763825458195e-05, "loss": 0.17186278104782104, "step": 54830 }, { "epoch": 0.2723940275970317, "grad_norm": 1.8815598487854004, "learning_rate": 1.707362679643824e-05, "loss": 0.17771995067596436, "step": 54840 }, { "epoch": 0.27244369828040094, "grad_norm": 0.734085202217102, "learning_rate": 1.7072489584355667e-05, "loss": 0.18069517612457275, "step": 54850 }, { "epoch": 0.2724933689637702, "grad_norm": 0.6390389204025269, "learning_rate": 1.7071352189239902e-05, "loss": 0.18790184259414672, "step": 54860 }, { "epoch": 0.27254303964713944, "grad_norm": 0.9217582941055298, "learning_rate": 1.707021461112039e-05, "loss": 0.1659625291824341, "step": 54870 }, { "epoch": 0.27259271033050875, "grad_norm": 1.883098840713501, "learning_rate": 1.706907685002657e-05, "loss": 0.19827769994735717, "step": 54880 }, { "epoch": 0.272642381013878, "grad_norm": 1.002563714981079, "learning_rate": 1.7067938905987877e-05, "loss": 0.17036184072494506, "step": 54890 }, { "epoch": 0.27269205169724725, "grad_norm": 1.124850869178772, "learning_rate": 1.706680077903377e-05, "loss": 0.18529220819473266, "step": 54900 }, { "epoch": 0.2727417223806165, "grad_norm": 1.4783951044082642, "learning_rate": 1.70656624691937e-05, "loss": 0.18039766550064087, "step": 54910 }, { "epoch": 0.27279139306398575, "grad_norm": 1.2299740314483643, "learning_rate": 1.7064523976497124e-05, "loss": 0.17719388008117676, "step": 54920 }, { "epoch": 0.27284106374735506, "grad_norm": 2.393028497695923, "learning_rate": 1.7063385300973512e-05, "loss": 0.1635982871055603, "step": 54930 }, { "epoch": 0.2728907344307243, "grad_norm": 1.7884562015533447, "learning_rate": 1.7062246442652327e-05, "loss": 0.17251718044281006, "step": 54940 }, { "epoch": 0.27294040511409356, "grad_norm": 0.9056192636489868, "learning_rate": 1.7061107401563047e-05, "loss": 0.18154582977294922, "step": 54950 }, { "epoch": 0.2729900757974628, "grad_norm": 2.7663674354553223, "learning_rate": 1.705996817773514e-05, "loss": 0.17629458904266357, "step": 54960 }, { "epoch": 0.27303974648083207, "grad_norm": 1.3019071817398071, "learning_rate": 1.70588287711981e-05, "loss": 0.16969728469848633, "step": 54970 }, { "epoch": 0.2730894171642013, "grad_norm": 1.8969697952270508, "learning_rate": 1.7057689181981408e-05, "loss": 0.1534052848815918, "step": 54980 }, { "epoch": 0.2731390878475706, "grad_norm": 1.0128940343856812, "learning_rate": 1.7056549410114558e-05, "loss": 0.1750732183456421, "step": 54990 }, { "epoch": 0.2731887585309399, "grad_norm": 2.656045436859131, "learning_rate": 1.7055409455627046e-05, "loss": 0.18723076581954956, "step": 55000 }, { "epoch": 0.27323842921430913, "grad_norm": 0.6133389472961426, "learning_rate": 1.7054269318548375e-05, "loss": 0.1838507890701294, "step": 55010 }, { "epoch": 0.2732880998976784, "grad_norm": 1.0834593772888184, "learning_rate": 1.705312899890805e-05, "loss": 0.19099705219268798, "step": 55020 }, { "epoch": 0.27333777058104763, "grad_norm": 1.7203443050384521, "learning_rate": 1.7051988496735583e-05, "loss": 0.15856258869171141, "step": 55030 }, { "epoch": 0.27338744126441694, "grad_norm": 1.1800646781921387, "learning_rate": 1.705084781206049e-05, "loss": 0.19089661836624144, "step": 55040 }, { "epoch": 0.2734371119477862, "grad_norm": 1.312077283859253, "learning_rate": 1.7049706944912287e-05, "loss": 0.17834095954895018, "step": 55050 }, { "epoch": 0.27348678263115545, "grad_norm": 1.3656760454177856, "learning_rate": 1.7048565895320506e-05, "loss": 0.18071576356887817, "step": 55060 }, { "epoch": 0.2735364533145247, "grad_norm": 1.3521037101745605, "learning_rate": 1.7047424663314673e-05, "loss": 0.18335611820220948, "step": 55070 }, { "epoch": 0.27358612399789395, "grad_norm": 0.903531551361084, "learning_rate": 1.704628324892432e-05, "loss": 0.17058181762695312, "step": 55080 }, { "epoch": 0.2736357946812632, "grad_norm": 1.5294992923736572, "learning_rate": 1.7045141652178997e-05, "loss": 0.17715742588043212, "step": 55090 }, { "epoch": 0.2736854653646325, "grad_norm": 1.7633143663406372, "learning_rate": 1.7043999873108235e-05, "loss": 0.1835399389266968, "step": 55100 }, { "epoch": 0.27373513604800176, "grad_norm": 4.054622173309326, "learning_rate": 1.7042857911741594e-05, "loss": 0.16986576318740845, "step": 55110 }, { "epoch": 0.273784806731371, "grad_norm": 1.9897578954696655, "learning_rate": 1.7041715768108617e-05, "loss": 0.18777917623519896, "step": 55120 }, { "epoch": 0.27383447741474026, "grad_norm": 1.9551756381988525, "learning_rate": 1.7040573442238867e-05, "loss": 0.16489043235778808, "step": 55130 }, { "epoch": 0.2738841480981095, "grad_norm": 1.6769919395446777, "learning_rate": 1.7039430934161907e-05, "loss": 0.16699265241622924, "step": 55140 }, { "epoch": 0.2739338187814788, "grad_norm": 1.2344337701797485, "learning_rate": 1.7038288243907308e-05, "loss": 0.15367354154586793, "step": 55150 }, { "epoch": 0.2739834894648481, "grad_norm": 1.5985963344573975, "learning_rate": 1.703714537150464e-05, "loss": 0.17938625812530518, "step": 55160 }, { "epoch": 0.2740331601482173, "grad_norm": 1.055085301399231, "learning_rate": 1.7036002316983476e-05, "loss": 0.17142435312271118, "step": 55170 }, { "epoch": 0.2740828308315866, "grad_norm": 1.2741228342056274, "learning_rate": 1.7034859080373404e-05, "loss": 0.17912100553512572, "step": 55180 }, { "epoch": 0.27413250151495583, "grad_norm": 1.570339560508728, "learning_rate": 1.7033715661704006e-05, "loss": 0.17311040163040162, "step": 55190 }, { "epoch": 0.2741821721983251, "grad_norm": 1.4372577667236328, "learning_rate": 1.703257206100487e-05, "loss": 0.1830010175704956, "step": 55200 }, { "epoch": 0.2742318428816944, "grad_norm": 0.8760338425636292, "learning_rate": 1.7031428278305605e-05, "loss": 0.18415768146514894, "step": 55210 }, { "epoch": 0.27428151356506364, "grad_norm": 0.8402316570281982, "learning_rate": 1.70302843136358e-05, "loss": 0.17475805282592774, "step": 55220 }, { "epoch": 0.2743311842484329, "grad_norm": 1.0163487195968628, "learning_rate": 1.7029140167025064e-05, "loss": 0.17554888725280762, "step": 55230 }, { "epoch": 0.27438085493180214, "grad_norm": 1.793207049369812, "learning_rate": 1.7027995838503007e-05, "loss": 0.17374794483184813, "step": 55240 }, { "epoch": 0.2744305256151714, "grad_norm": 0.6635333895683289, "learning_rate": 1.7026851328099243e-05, "loss": 0.17968302965164185, "step": 55250 }, { "epoch": 0.2744801962985407, "grad_norm": 1.5233404636383057, "learning_rate": 1.702570663584339e-05, "loss": 0.17557911872863768, "step": 55260 }, { "epoch": 0.27452986698190995, "grad_norm": 1.1121498346328735, "learning_rate": 1.7024561761765078e-05, "loss": 0.181510066986084, "step": 55270 }, { "epoch": 0.2745795376652792, "grad_norm": 0.7071771025657654, "learning_rate": 1.702341670589393e-05, "loss": 0.19440686702728271, "step": 55280 }, { "epoch": 0.27462920834864846, "grad_norm": 1.2223403453826904, "learning_rate": 1.7022271468259578e-05, "loss": 0.1734483242034912, "step": 55290 }, { "epoch": 0.2746788790320177, "grad_norm": 1.5888125896453857, "learning_rate": 1.702112604889167e-05, "loss": 0.18083693981170654, "step": 55300 }, { "epoch": 0.27472854971538696, "grad_norm": 0.9463608264923096, "learning_rate": 1.7019980447819843e-05, "loss": 0.1779264211654663, "step": 55310 }, { "epoch": 0.27477822039875627, "grad_norm": 1.857856035232544, "learning_rate": 1.7018834665073743e-05, "loss": 0.16269617080688475, "step": 55320 }, { "epoch": 0.2748278910821255, "grad_norm": 2.1280908584594727, "learning_rate": 1.7017688700683023e-05, "loss": 0.17383320331573487, "step": 55330 }, { "epoch": 0.27487756176549477, "grad_norm": 0.9018408060073853, "learning_rate": 1.701654255467734e-05, "loss": 0.17034506797790527, "step": 55340 }, { "epoch": 0.274927232448864, "grad_norm": 1.3296924829483032, "learning_rate": 1.7015396227086357e-05, "loss": 0.17833659648895264, "step": 55350 }, { "epoch": 0.2749769031322333, "grad_norm": 0.6199924945831299, "learning_rate": 1.7014249717939744e-05, "loss": 0.17798244953155518, "step": 55360 }, { "epoch": 0.2750265738156026, "grad_norm": 2.4925038814544678, "learning_rate": 1.7013103027267166e-05, "loss": 0.16353380680084229, "step": 55370 }, { "epoch": 0.27507624449897183, "grad_norm": 1.3688496351242065, "learning_rate": 1.70119561550983e-05, "loss": 0.18374792337417603, "step": 55380 }, { "epoch": 0.2751259151823411, "grad_norm": 1.5460455417633057, "learning_rate": 1.7010809101462832e-05, "loss": 0.18291476964950562, "step": 55390 }, { "epoch": 0.27517558586571034, "grad_norm": 0.7778146266937256, "learning_rate": 1.7009661866390443e-05, "loss": 0.16439604759216309, "step": 55400 }, { "epoch": 0.2752252565490796, "grad_norm": 1.2100319862365723, "learning_rate": 1.700851444991082e-05, "loss": 0.16251412630081177, "step": 55410 }, { "epoch": 0.27527492723244884, "grad_norm": 1.1644287109375, "learning_rate": 1.7007366852053663e-05, "loss": 0.18240901231765747, "step": 55420 }, { "epoch": 0.27532459791581815, "grad_norm": 1.2071539163589478, "learning_rate": 1.7006219072848667e-05, "loss": 0.17593839168548583, "step": 55430 }, { "epoch": 0.2753742685991874, "grad_norm": 1.6871280670166016, "learning_rate": 1.7005071112325546e-05, "loss": 0.16662461757659913, "step": 55440 }, { "epoch": 0.27542393928255665, "grad_norm": 2.4561007022857666, "learning_rate": 1.700392297051399e-05, "loss": 0.18279836177825928, "step": 55450 }, { "epoch": 0.2754736099659259, "grad_norm": 0.9527942538261414, "learning_rate": 1.7002774647443733e-05, "loss": 0.17135026454925537, "step": 55460 }, { "epoch": 0.27552328064929515, "grad_norm": 1.0259088277816772, "learning_rate": 1.700162614314448e-05, "loss": 0.18597068786621093, "step": 55470 }, { "epoch": 0.27557295133266446, "grad_norm": 1.1852983236312866, "learning_rate": 1.700047745764596e-05, "loss": 0.16868542432785033, "step": 55480 }, { "epoch": 0.2756226220160337, "grad_norm": 0.9914664626121521, "learning_rate": 1.6999328590977894e-05, "loss": 0.171506667137146, "step": 55490 }, { "epoch": 0.27567229269940297, "grad_norm": 0.9211398959159851, "learning_rate": 1.6998179543170023e-05, "loss": 0.1717148780822754, "step": 55500 }, { "epoch": 0.2757219633827722, "grad_norm": 1.1635686159133911, "learning_rate": 1.6997030314252075e-05, "loss": 0.1647059440612793, "step": 55510 }, { "epoch": 0.27577163406614147, "grad_norm": 1.0159939527511597, "learning_rate": 1.69958809042538e-05, "loss": 0.17457664012908936, "step": 55520 }, { "epoch": 0.2758213047495107, "grad_norm": 1.164919137954712, "learning_rate": 1.6994731313204937e-05, "loss": 0.181569504737854, "step": 55530 }, { "epoch": 0.27587097543288003, "grad_norm": 0.858870804309845, "learning_rate": 1.699358154113524e-05, "loss": 0.16677005290985109, "step": 55540 }, { "epoch": 0.2759206461162493, "grad_norm": 0.8711562752723694, "learning_rate": 1.6992431588074465e-05, "loss": 0.17425448894500734, "step": 55550 }, { "epoch": 0.27597031679961853, "grad_norm": 1.276249885559082, "learning_rate": 1.6991281454052373e-05, "loss": 0.18026294708251953, "step": 55560 }, { "epoch": 0.2760199874829878, "grad_norm": 1.186941385269165, "learning_rate": 1.6990131139098727e-05, "loss": 0.16573126316070558, "step": 55570 }, { "epoch": 0.27606965816635703, "grad_norm": 1.2754087448120117, "learning_rate": 1.69889806432433e-05, "loss": 0.1718233346939087, "step": 55580 }, { "epoch": 0.27611932884972634, "grad_norm": 0.9199329018592834, "learning_rate": 1.698782996651586e-05, "loss": 0.17299528121948243, "step": 55590 }, { "epoch": 0.2761689995330956, "grad_norm": 0.7726691961288452, "learning_rate": 1.6986679108946195e-05, "loss": 0.17372647523880005, "step": 55600 }, { "epoch": 0.27621867021646485, "grad_norm": 0.6364781856536865, "learning_rate": 1.6985528070564084e-05, "loss": 0.16478946208953857, "step": 55610 }, { "epoch": 0.2762683408998341, "grad_norm": 1.4887664318084717, "learning_rate": 1.698437685139931e-05, "loss": 0.18369681835174562, "step": 55620 }, { "epoch": 0.27631801158320335, "grad_norm": 1.2516697645187378, "learning_rate": 1.6983225451481675e-05, "loss": 0.1803782343864441, "step": 55630 }, { "epoch": 0.2763676822665726, "grad_norm": 0.9138617515563965, "learning_rate": 1.6982073870840974e-05, "loss": 0.16971689462661743, "step": 55640 }, { "epoch": 0.2764173529499419, "grad_norm": 0.7674553990364075, "learning_rate": 1.6980922109507007e-05, "loss": 0.17017340660095215, "step": 55650 }, { "epoch": 0.27646702363331116, "grad_norm": 1.71095871925354, "learning_rate": 1.6979770167509582e-05, "loss": 0.1661001443862915, "step": 55660 }, { "epoch": 0.2765166943166804, "grad_norm": 1.7310680150985718, "learning_rate": 1.6978618044878516e-05, "loss": 0.16838685274124146, "step": 55670 }, { "epoch": 0.27656636500004966, "grad_norm": 0.940657913684845, "learning_rate": 1.6977465741643616e-05, "loss": 0.18094303607940673, "step": 55680 }, { "epoch": 0.2766160356834189, "grad_norm": 1.1777610778808594, "learning_rate": 1.697631325783471e-05, "loss": 0.18544681072235109, "step": 55690 }, { "epoch": 0.27666570636678817, "grad_norm": 0.7335392832756042, "learning_rate": 1.6975160593481617e-05, "loss": 0.18151406049728394, "step": 55700 }, { "epoch": 0.2767153770501575, "grad_norm": 1.2026612758636475, "learning_rate": 1.697400774861418e-05, "loss": 0.1722390055656433, "step": 55710 }, { "epoch": 0.2767650477335267, "grad_norm": 1.4748475551605225, "learning_rate": 1.6972854723262222e-05, "loss": 0.1719593048095703, "step": 55720 }, { "epoch": 0.276814718416896, "grad_norm": 1.9649739265441895, "learning_rate": 1.697170151745559e-05, "loss": 0.17044928073883056, "step": 55730 }, { "epoch": 0.27686438910026523, "grad_norm": 0.9481295347213745, "learning_rate": 1.6970548131224124e-05, "loss": 0.16784021854400635, "step": 55740 }, { "epoch": 0.2769140597836345, "grad_norm": 0.896636426448822, "learning_rate": 1.6969394564597676e-05, "loss": 0.17592408657073974, "step": 55750 }, { "epoch": 0.2769637304670038, "grad_norm": 3.670653820037842, "learning_rate": 1.69682408176061e-05, "loss": 0.17501237392425537, "step": 55760 }, { "epoch": 0.27701340115037304, "grad_norm": 1.7033288478851318, "learning_rate": 1.6967086890279252e-05, "loss": 0.1919538974761963, "step": 55770 }, { "epoch": 0.2770630718337423, "grad_norm": 0.8992712497711182, "learning_rate": 1.6965932782646995e-05, "loss": 0.18869688510894775, "step": 55780 }, { "epoch": 0.27711274251711154, "grad_norm": 1.0301316976547241, "learning_rate": 1.69647784947392e-05, "loss": 0.17123135328292846, "step": 55790 }, { "epoch": 0.2771624132004808, "grad_norm": 1.9087700843811035, "learning_rate": 1.696362402658574e-05, "loss": 0.17855004072189332, "step": 55800 }, { "epoch": 0.27721208388385005, "grad_norm": 1.815842628479004, "learning_rate": 1.696246937821649e-05, "loss": 0.1735105037689209, "step": 55810 }, { "epoch": 0.27726175456721935, "grad_norm": 1.2776907682418823, "learning_rate": 1.696131454966133e-05, "loss": 0.18017693758010864, "step": 55820 }, { "epoch": 0.2773114252505886, "grad_norm": 1.4585341215133667, "learning_rate": 1.6960159540950147e-05, "loss": 0.17408356666564942, "step": 55830 }, { "epoch": 0.27736109593395786, "grad_norm": 1.0540611743927002, "learning_rate": 1.695900435211284e-05, "loss": 0.16624597311019898, "step": 55840 }, { "epoch": 0.2774107666173271, "grad_norm": 1.4044630527496338, "learning_rate": 1.6957848983179294e-05, "loss": 0.18639841079711914, "step": 55850 }, { "epoch": 0.27746043730069636, "grad_norm": 0.9129371643066406, "learning_rate": 1.6956693434179413e-05, "loss": 0.15627361536026002, "step": 55860 }, { "epoch": 0.27751010798406567, "grad_norm": 1.6233608722686768, "learning_rate": 1.6955537705143102e-05, "loss": 0.16047890186309816, "step": 55870 }, { "epoch": 0.2775597786674349, "grad_norm": 2.7360095977783203, "learning_rate": 1.6954381796100276e-05, "loss": 0.16693826913833618, "step": 55880 }, { "epoch": 0.27760944935080417, "grad_norm": 0.6611164212226868, "learning_rate": 1.695322570708084e-05, "loss": 0.17092164754867553, "step": 55890 }, { "epoch": 0.2776591200341734, "grad_norm": 1.3068140745162964, "learning_rate": 1.6952069438114726e-05, "loss": 0.16670756340026854, "step": 55900 }, { "epoch": 0.2777087907175427, "grad_norm": 0.7800928950309753, "learning_rate": 1.6950912989231846e-05, "loss": 0.18064496517181397, "step": 55910 }, { "epoch": 0.2777584614009119, "grad_norm": 1.0280836820602417, "learning_rate": 1.694975636046213e-05, "loss": 0.1676027774810791, "step": 55920 }, { "epoch": 0.27780813208428123, "grad_norm": 1.5783299207687378, "learning_rate": 1.6948599551835517e-05, "loss": 0.17202943563461304, "step": 55930 }, { "epoch": 0.2778578027676505, "grad_norm": 1.6011825799942017, "learning_rate": 1.6947442563381943e-05, "loss": 0.17079895734786987, "step": 55940 }, { "epoch": 0.27790747345101974, "grad_norm": 1.391162395477295, "learning_rate": 1.6946285395131346e-05, "loss": 0.17893131971359252, "step": 55950 }, { "epoch": 0.277957144134389, "grad_norm": 1.0941804647445679, "learning_rate": 1.6945128047113678e-05, "loss": 0.173021399974823, "step": 55960 }, { "epoch": 0.27800681481775824, "grad_norm": 3.2144765853881836, "learning_rate": 1.6943970519358886e-05, "loss": 0.17156667709350587, "step": 55970 }, { "epoch": 0.27805648550112755, "grad_norm": 0.8714061379432678, "learning_rate": 1.6942812811896932e-05, "loss": 0.17262780666351318, "step": 55980 }, { "epoch": 0.2781061561844968, "grad_norm": 2.363492488861084, "learning_rate": 1.6941654924757773e-05, "loss": 0.16673848628997803, "step": 55990 }, { "epoch": 0.27815582686786605, "grad_norm": 0.7669265270233154, "learning_rate": 1.6940496857971375e-05, "loss": 0.17726995944976806, "step": 56000 }, { "epoch": 0.2782054975512353, "grad_norm": 1.252234935760498, "learning_rate": 1.693933861156771e-05, "loss": 0.1694735288619995, "step": 56010 }, { "epoch": 0.27825516823460456, "grad_norm": 0.8996654152870178, "learning_rate": 1.6938180185576752e-05, "loss": 0.1626623272895813, "step": 56020 }, { "epoch": 0.2783048389179738, "grad_norm": 2.1577184200286865, "learning_rate": 1.693702158002848e-05, "loss": 0.17133508920669555, "step": 56030 }, { "epoch": 0.2783545096013431, "grad_norm": 1.5229113101959229, "learning_rate": 1.6935862794952877e-05, "loss": 0.17713522911071777, "step": 56040 }, { "epoch": 0.27840418028471237, "grad_norm": 2.1016926765441895, "learning_rate": 1.6934703830379934e-05, "loss": 0.18100630044937133, "step": 56050 }, { "epoch": 0.2784538509680816, "grad_norm": 1.127016544342041, "learning_rate": 1.6933544686339647e-05, "loss": 0.16869536638259888, "step": 56060 }, { "epoch": 0.27850352165145087, "grad_norm": 1.700614333152771, "learning_rate": 1.693238536286201e-05, "loss": 0.15749107599258422, "step": 56070 }, { "epoch": 0.2785531923348201, "grad_norm": 2.636526346206665, "learning_rate": 1.6931225859977027e-05, "loss": 0.18179683685302733, "step": 56080 }, { "epoch": 0.27860286301818943, "grad_norm": 1.9084270000457764, "learning_rate": 1.6930066177714705e-05, "loss": 0.1761051058769226, "step": 56090 }, { "epoch": 0.2786525337015587, "grad_norm": 0.8647151589393616, "learning_rate": 1.692890631610506e-05, "loss": 0.1772756338119507, "step": 56100 }, { "epoch": 0.27870220438492793, "grad_norm": 1.2020902633666992, "learning_rate": 1.6927746275178104e-05, "loss": 0.16979113817214966, "step": 56110 }, { "epoch": 0.2787518750682972, "grad_norm": 1.1090904474258423, "learning_rate": 1.6926586054963856e-05, "loss": 0.16656843423843384, "step": 56120 }, { "epoch": 0.27880154575166644, "grad_norm": 1.6617143154144287, "learning_rate": 1.6925425655492347e-05, "loss": 0.18498263359069825, "step": 56130 }, { "epoch": 0.2788512164350357, "grad_norm": 1.9152911901474, "learning_rate": 1.692426507679361e-05, "loss": 0.18457014560699464, "step": 56140 }, { "epoch": 0.278900887118405, "grad_norm": 1.5350552797317505, "learning_rate": 1.6923104318897677e-05, "loss": 0.17813836336135863, "step": 56150 }, { "epoch": 0.27895055780177425, "grad_norm": 1.4755433797836304, "learning_rate": 1.6921943381834584e-05, "loss": 0.17765414714813232, "step": 56160 }, { "epoch": 0.2790002284851435, "grad_norm": 1.66800057888031, "learning_rate": 1.6920782265634384e-05, "loss": 0.17112619876861573, "step": 56170 }, { "epoch": 0.27904989916851275, "grad_norm": 1.0153969526290894, "learning_rate": 1.691962097032712e-05, "loss": 0.17423613071441652, "step": 56180 }, { "epoch": 0.279099569851882, "grad_norm": 1.417473316192627, "learning_rate": 1.6918459495942848e-05, "loss": 0.17167999744415283, "step": 56190 }, { "epoch": 0.2791492405352513, "grad_norm": 0.8934184908866882, "learning_rate": 1.6917297842511624e-05, "loss": 0.17570430040359497, "step": 56200 }, { "epoch": 0.27919891121862056, "grad_norm": 1.3178768157958984, "learning_rate": 1.6916136010063514e-05, "loss": 0.1834249973297119, "step": 56210 }, { "epoch": 0.2792485819019898, "grad_norm": 1.2699155807495117, "learning_rate": 1.6914973998628585e-05, "loss": 0.17765393257141113, "step": 56220 }, { "epoch": 0.27929825258535906, "grad_norm": 1.899036169052124, "learning_rate": 1.691381180823691e-05, "loss": 0.1917507529258728, "step": 56230 }, { "epoch": 0.2793479232687283, "grad_norm": 1.7149176597595215, "learning_rate": 1.6912649438918564e-05, "loss": 0.17793605327606202, "step": 56240 }, { "epoch": 0.27939759395209757, "grad_norm": 0.6966951489448547, "learning_rate": 1.6911486890703632e-05, "loss": 0.1635734796524048, "step": 56250 }, { "epoch": 0.2794472646354669, "grad_norm": 1.5376760959625244, "learning_rate": 1.6910324163622197e-05, "loss": 0.16521875858306884, "step": 56260 }, { "epoch": 0.2794969353188361, "grad_norm": 1.308850884437561, "learning_rate": 1.690916125770435e-05, "loss": 0.17501319646835328, "step": 56270 }, { "epoch": 0.2795466060022054, "grad_norm": 2.837517023086548, "learning_rate": 1.690799817298019e-05, "loss": 0.18034473657608033, "step": 56280 }, { "epoch": 0.27959627668557463, "grad_norm": 1.1394342184066772, "learning_rate": 1.6906834909479816e-05, "loss": 0.17010717391967772, "step": 56290 }, { "epoch": 0.2796459473689439, "grad_norm": 1.1027532815933228, "learning_rate": 1.6905671467233325e-05, "loss": 0.17814738750457765, "step": 56300 }, { "epoch": 0.2796956180523132, "grad_norm": 1.0698258876800537, "learning_rate": 1.690450784627084e-05, "loss": 0.1866987705230713, "step": 56310 }, { "epoch": 0.27974528873568244, "grad_norm": 1.0914037227630615, "learning_rate": 1.6903344046622464e-05, "loss": 0.1687839984893799, "step": 56320 }, { "epoch": 0.2797949594190517, "grad_norm": 1.4640007019042969, "learning_rate": 1.6902180068318324e-05, "loss": 0.17250179052352904, "step": 56330 }, { "epoch": 0.27984463010242094, "grad_norm": 1.6871715784072876, "learning_rate": 1.6901015911388533e-05, "loss": 0.17171106338500977, "step": 56340 }, { "epoch": 0.2798943007857902, "grad_norm": 0.9256941080093384, "learning_rate": 1.6899851575863233e-05, "loss": 0.15502591133117677, "step": 56350 }, { "epoch": 0.27994397146915945, "grad_norm": 0.8795332908630371, "learning_rate": 1.6898687061772543e-05, "loss": 0.16879844665527344, "step": 56360 }, { "epoch": 0.27999364215252875, "grad_norm": 2.3679728507995605, "learning_rate": 1.6897522369146607e-05, "loss": 0.182686710357666, "step": 56370 }, { "epoch": 0.280043312835898, "grad_norm": 1.798974633216858, "learning_rate": 1.6896357498015565e-05, "loss": 0.19962879419326782, "step": 56380 }, { "epoch": 0.28009298351926726, "grad_norm": 1.0471575260162354, "learning_rate": 1.6895192448409564e-05, "loss": 0.1646182894706726, "step": 56390 }, { "epoch": 0.2801426542026365, "grad_norm": 1.3388382196426392, "learning_rate": 1.6894027220358754e-05, "loss": 0.17612698078155517, "step": 56400 }, { "epoch": 0.28019232488600576, "grad_norm": 0.8690395951271057, "learning_rate": 1.6892861813893297e-05, "loss": 0.16934800148010254, "step": 56410 }, { "epoch": 0.28024199556937507, "grad_norm": 1.898114800453186, "learning_rate": 1.6891696229043342e-05, "loss": 0.16614885330200196, "step": 56420 }, { "epoch": 0.2802916662527443, "grad_norm": 2.4707489013671875, "learning_rate": 1.689053046583906e-05, "loss": 0.1634178042411804, "step": 56430 }, { "epoch": 0.28034133693611357, "grad_norm": 0.5869463682174683, "learning_rate": 1.6889364524310622e-05, "loss": 0.17007309198379517, "step": 56440 }, { "epoch": 0.2803910076194828, "grad_norm": 0.9234416484832764, "learning_rate": 1.6888198404488197e-05, "loss": 0.18135772943496703, "step": 56450 }, { "epoch": 0.2804406783028521, "grad_norm": 1.531147837638855, "learning_rate": 1.6887032106401972e-05, "loss": 0.1782996416091919, "step": 56460 }, { "epoch": 0.2804903489862213, "grad_norm": 2.061969518661499, "learning_rate": 1.6885865630082122e-05, "loss": 0.17054238319396972, "step": 56470 }, { "epoch": 0.28054001966959063, "grad_norm": 1.484455943107605, "learning_rate": 1.6884698975558836e-05, "loss": 0.17044130563735962, "step": 56480 }, { "epoch": 0.2805896903529599, "grad_norm": 3.134032726287842, "learning_rate": 1.6883532142862314e-05, "loss": 0.18045881986618043, "step": 56490 }, { "epoch": 0.28063936103632914, "grad_norm": 1.2942575216293335, "learning_rate": 1.6882365132022743e-05, "loss": 0.1752526044845581, "step": 56500 }, { "epoch": 0.2806890317196984, "grad_norm": 0.8358426094055176, "learning_rate": 1.688119794307033e-05, "loss": 0.1686464309692383, "step": 56510 }, { "epoch": 0.28073870240306764, "grad_norm": 1.1048400402069092, "learning_rate": 1.6880030576035286e-05, "loss": 0.1654738187789917, "step": 56520 }, { "epoch": 0.28078837308643695, "grad_norm": 0.7377939224243164, "learning_rate": 1.6878863030947813e-05, "loss": 0.1762380838394165, "step": 56530 }, { "epoch": 0.2808380437698062, "grad_norm": 0.6599865555763245, "learning_rate": 1.6877695307838132e-05, "loss": 0.1638335704803467, "step": 56540 }, { "epoch": 0.28088771445317545, "grad_norm": 1.9667524099349976, "learning_rate": 1.6876527406736464e-05, "loss": 0.1732640743255615, "step": 56550 }, { "epoch": 0.2809373851365447, "grad_norm": 1.3225078582763672, "learning_rate": 1.6875359327673028e-05, "loss": 0.16298131942749022, "step": 56560 }, { "epoch": 0.28098705581991396, "grad_norm": 0.8485682010650635, "learning_rate": 1.6874191070678056e-05, "loss": 0.18909769058227538, "step": 56570 }, { "epoch": 0.2810367265032832, "grad_norm": 4.078521251678467, "learning_rate": 1.6873022635781788e-05, "loss": 0.1821345090866089, "step": 56580 }, { "epoch": 0.2810863971866525, "grad_norm": 0.8255689144134521, "learning_rate": 1.6871854023014455e-05, "loss": 0.1776721239089966, "step": 56590 }, { "epoch": 0.28113606787002177, "grad_norm": 1.118821620941162, "learning_rate": 1.68706852324063e-05, "loss": 0.17320449352264405, "step": 56600 }, { "epoch": 0.281185738553391, "grad_norm": 1.1737762689590454, "learning_rate": 1.686951626398758e-05, "loss": 0.17696336507797242, "step": 56610 }, { "epoch": 0.28123540923676027, "grad_norm": 0.9708634614944458, "learning_rate": 1.6868347117788535e-05, "loss": 0.15999696254730225, "step": 56620 }, { "epoch": 0.2812850799201295, "grad_norm": 1.4668841361999512, "learning_rate": 1.686717779383943e-05, "loss": 0.17271575927734376, "step": 56630 }, { "epoch": 0.28133475060349883, "grad_norm": 0.9420873522758484, "learning_rate": 1.686600829217053e-05, "loss": 0.16756350994110109, "step": 56640 }, { "epoch": 0.2813844212868681, "grad_norm": 0.9253408312797546, "learning_rate": 1.686483861281209e-05, "loss": 0.1918921709060669, "step": 56650 }, { "epoch": 0.28143409197023733, "grad_norm": 1.4731550216674805, "learning_rate": 1.686366875579439e-05, "loss": 0.17534008026123046, "step": 56660 }, { "epoch": 0.2814837626536066, "grad_norm": 0.8570688366889954, "learning_rate": 1.68624987211477e-05, "loss": 0.16562073230743407, "step": 56670 }, { "epoch": 0.28153343333697584, "grad_norm": 2.1814868450164795, "learning_rate": 1.6861328508902305e-05, "loss": 0.18075337409973144, "step": 56680 }, { "epoch": 0.2815831040203451, "grad_norm": 1.1513890027999878, "learning_rate": 1.6860158119088485e-05, "loss": 0.17344672679901124, "step": 56690 }, { "epoch": 0.2816327747037144, "grad_norm": 0.8817263245582581, "learning_rate": 1.6858987551736533e-05, "loss": 0.17556267976760864, "step": 56700 }, { "epoch": 0.28168244538708365, "grad_norm": 1.1110594272613525, "learning_rate": 1.6857816806876742e-05, "loss": 0.18058849573135377, "step": 56710 }, { "epoch": 0.2817321160704529, "grad_norm": 1.8654080629348755, "learning_rate": 1.6856645884539407e-05, "loss": 0.1734888195991516, "step": 56720 }, { "epoch": 0.28178178675382215, "grad_norm": 2.9411795139312744, "learning_rate": 1.6855474784754833e-05, "loss": 0.17904890775680543, "step": 56730 }, { "epoch": 0.2818314574371914, "grad_norm": 1.3835855722427368, "learning_rate": 1.685430350755333e-05, "loss": 0.15283024311065674, "step": 56740 }, { "epoch": 0.2818811281205607, "grad_norm": 1.1894354820251465, "learning_rate": 1.6853132052965208e-05, "loss": 0.16716251373291016, "step": 56750 }, { "epoch": 0.28193079880392996, "grad_norm": 1.9718750715255737, "learning_rate": 1.6851960421020784e-05, "loss": 0.17445635795593262, "step": 56760 }, { "epoch": 0.2819804694872992, "grad_norm": 2.065272808074951, "learning_rate": 1.6850788611750375e-05, "loss": 0.1808839797973633, "step": 56770 }, { "epoch": 0.28203014017066846, "grad_norm": 1.3152148723602295, "learning_rate": 1.6849616625184318e-05, "loss": 0.17882753610610963, "step": 56780 }, { "epoch": 0.2820798108540377, "grad_norm": 0.7961500883102417, "learning_rate": 1.6848444461352934e-05, "loss": 0.164428973197937, "step": 56790 }, { "epoch": 0.28212948153740697, "grad_norm": 1.1058906316757202, "learning_rate": 1.6847272120286556e-05, "loss": 0.19001193046569825, "step": 56800 }, { "epoch": 0.2821791522207763, "grad_norm": 1.102166771888733, "learning_rate": 1.6846099602015537e-05, "loss": 0.17580368518829345, "step": 56810 }, { "epoch": 0.2822288229041455, "grad_norm": 1.8160849809646606, "learning_rate": 1.684492690657021e-05, "loss": 0.17312943935394287, "step": 56820 }, { "epoch": 0.2822784935875148, "grad_norm": 0.9806413054466248, "learning_rate": 1.6843754033980926e-05, "loss": 0.16085197925567626, "step": 56830 }, { "epoch": 0.28232816427088403, "grad_norm": 1.1211663484573364, "learning_rate": 1.684258098427804e-05, "loss": 0.17882928848266602, "step": 56840 }, { "epoch": 0.2823778349542533, "grad_norm": 2.007093667984009, "learning_rate": 1.684140775749191e-05, "loss": 0.16575567722320556, "step": 56850 }, { "epoch": 0.2824275056376226, "grad_norm": 1.448224663734436, "learning_rate": 1.6840234353652896e-05, "loss": 0.17532286643981934, "step": 56860 }, { "epoch": 0.28247717632099184, "grad_norm": 1.285637617111206, "learning_rate": 1.683906077279137e-05, "loss": 0.17342753410339357, "step": 56870 }, { "epoch": 0.2825268470043611, "grad_norm": 0.9440377950668335, "learning_rate": 1.68378870149377e-05, "loss": 0.17062268257141114, "step": 56880 }, { "epoch": 0.28257651768773034, "grad_norm": 1.385119080543518, "learning_rate": 1.6836713080122263e-05, "loss": 0.17206622362136842, "step": 56890 }, { "epoch": 0.2826261883710996, "grad_norm": 0.7035889029502869, "learning_rate": 1.6835538968375437e-05, "loss": 0.1843106508255005, "step": 56900 }, { "epoch": 0.28267585905446885, "grad_norm": 1.062738060951233, "learning_rate": 1.6834364679727614e-05, "loss": 0.16957248449325563, "step": 56910 }, { "epoch": 0.28272552973783815, "grad_norm": 2.2647736072540283, "learning_rate": 1.6833190214209184e-05, "loss": 0.16144936084747313, "step": 56920 }, { "epoch": 0.2827752004212074, "grad_norm": 0.8292577266693115, "learning_rate": 1.683201557185054e-05, "loss": 0.16124987602233887, "step": 56930 }, { "epoch": 0.28282487110457666, "grad_norm": 1.0979963541030884, "learning_rate": 1.6830840752682074e-05, "loss": 0.1774155616760254, "step": 56940 }, { "epoch": 0.2828745417879459, "grad_norm": 1.4740486145019531, "learning_rate": 1.6829665756734197e-05, "loss": 0.17581377029418946, "step": 56950 }, { "epoch": 0.28292421247131516, "grad_norm": 1.1624131202697754, "learning_rate": 1.6828490584037317e-05, "loss": 0.16597695350646974, "step": 56960 }, { "epoch": 0.28297388315468447, "grad_norm": 1.2560521364212036, "learning_rate": 1.6827315234621846e-05, "loss": 0.1685910701751709, "step": 56970 }, { "epoch": 0.2830235538380537, "grad_norm": 1.0836042165756226, "learning_rate": 1.6826139708518203e-05, "loss": 0.17864307165145873, "step": 56980 }, { "epoch": 0.283073224521423, "grad_norm": 1.3156938552856445, "learning_rate": 1.682496400575681e-05, "loss": 0.17866779565811158, "step": 56990 }, { "epoch": 0.2831228952047922, "grad_norm": 0.9371159672737122, "learning_rate": 1.682378812636809e-05, "loss": 0.18048359155654908, "step": 57000 }, { "epoch": 0.2831725658881615, "grad_norm": 1.1730852127075195, "learning_rate": 1.682261207038248e-05, "loss": 0.18406476974487304, "step": 57010 }, { "epoch": 0.2832222365715307, "grad_norm": 1.7859904766082764, "learning_rate": 1.6821435837830412e-05, "loss": 0.17470266819000244, "step": 57020 }, { "epoch": 0.28327190725490003, "grad_norm": 0.7567705512046814, "learning_rate": 1.6820259428742325e-05, "loss": 0.17903082370758056, "step": 57030 }, { "epoch": 0.2833215779382693, "grad_norm": 1.3752127885818481, "learning_rate": 1.6819082843148668e-05, "loss": 0.17081546783447266, "step": 57040 }, { "epoch": 0.28337124862163854, "grad_norm": 0.729459285736084, "learning_rate": 1.6817906081079888e-05, "loss": 0.1738970994949341, "step": 57050 }, { "epoch": 0.2834209193050078, "grad_norm": 1.9208556413650513, "learning_rate": 1.681672914256644e-05, "loss": 0.173712158203125, "step": 57060 }, { "epoch": 0.28347058998837704, "grad_norm": 1.529726266860962, "learning_rate": 1.6815552027638787e-05, "loss": 0.17522165775299073, "step": 57070 }, { "epoch": 0.28352026067174635, "grad_norm": 1.2289888858795166, "learning_rate": 1.6814374736327383e-05, "loss": 0.1779584527015686, "step": 57080 }, { "epoch": 0.2835699313551156, "grad_norm": 1.496201753616333, "learning_rate": 1.6813197268662702e-05, "loss": 0.15999751091003417, "step": 57090 }, { "epoch": 0.28361960203848485, "grad_norm": 1.532771348953247, "learning_rate": 1.6812019624675214e-05, "loss": 0.17276248931884766, "step": 57100 }, { "epoch": 0.2836692727218541, "grad_norm": 0.9048123359680176, "learning_rate": 1.68108418043954e-05, "loss": 0.17585439682006837, "step": 57110 }, { "epoch": 0.28371894340522336, "grad_norm": 1.6007556915283203, "learning_rate": 1.6809663807853738e-05, "loss": 0.1700369596481323, "step": 57120 }, { "epoch": 0.2837686140885926, "grad_norm": 1.3240394592285156, "learning_rate": 1.6808485635080712e-05, "loss": 0.18725935220718384, "step": 57130 }, { "epoch": 0.2838182847719619, "grad_norm": 1.7886979579925537, "learning_rate": 1.6807307286106818e-05, "loss": 0.17216041088104247, "step": 57140 }, { "epoch": 0.28386795545533117, "grad_norm": 1.6963249444961548, "learning_rate": 1.6806128760962546e-05, "loss": 0.177534818649292, "step": 57150 }, { "epoch": 0.2839176261387004, "grad_norm": 1.684546947479248, "learning_rate": 1.6804950059678403e-05, "loss": 0.16774680614471435, "step": 57160 }, { "epoch": 0.28396729682206967, "grad_norm": 0.7421315908432007, "learning_rate": 1.6803771182284886e-05, "loss": 0.17536754608154298, "step": 57170 }, { "epoch": 0.2840169675054389, "grad_norm": 0.9120832681655884, "learning_rate": 1.6802592128812504e-05, "loss": 0.17614326477050782, "step": 57180 }, { "epoch": 0.28406663818880823, "grad_norm": 1.5368252992630005, "learning_rate": 1.6801412899291774e-05, "loss": 0.17612113952636718, "step": 57190 }, { "epoch": 0.2841163088721775, "grad_norm": 1.7236443758010864, "learning_rate": 1.6800233493753214e-05, "loss": 0.18266024589538574, "step": 57200 }, { "epoch": 0.28416597955554673, "grad_norm": 1.5452921390533447, "learning_rate": 1.6799053912227347e-05, "loss": 0.172776460647583, "step": 57210 }, { "epoch": 0.284215650238916, "grad_norm": 1.4829413890838623, "learning_rate": 1.67978741547447e-05, "loss": 0.1753209114074707, "step": 57220 }, { "epoch": 0.28426532092228524, "grad_norm": 1.380418300628662, "learning_rate": 1.6796694221335796e-05, "loss": 0.17681689262390138, "step": 57230 }, { "epoch": 0.2843149916056545, "grad_norm": 0.9227482080459595, "learning_rate": 1.6795514112031183e-05, "loss": 0.17643072605133056, "step": 57240 }, { "epoch": 0.2843646622890238, "grad_norm": 0.654449462890625, "learning_rate": 1.6794333826861396e-05, "loss": 0.15276792049407958, "step": 57250 }, { "epoch": 0.28441433297239305, "grad_norm": 2.321470022201538, "learning_rate": 1.6793153365856984e-05, "loss": 0.178421688079834, "step": 57260 }, { "epoch": 0.2844640036557623, "grad_norm": 4.042707443237305, "learning_rate": 1.6791972729048493e-05, "loss": 0.17035553455352784, "step": 57270 }, { "epoch": 0.28451367433913155, "grad_norm": 3.6148064136505127, "learning_rate": 1.6790791916466475e-05, "loss": 0.18473348617553711, "step": 57280 }, { "epoch": 0.2845633450225008, "grad_norm": 0.9473696947097778, "learning_rate": 1.6789610928141495e-05, "loss": 0.1689607620239258, "step": 57290 }, { "epoch": 0.28461301570587005, "grad_norm": 2.2371292114257812, "learning_rate": 1.6788429764104118e-05, "loss": 0.17052100896835326, "step": 57300 }, { "epoch": 0.28466268638923936, "grad_norm": 1.1011550426483154, "learning_rate": 1.6787248424384905e-05, "loss": 0.15497968196868897, "step": 57310 }, { "epoch": 0.2847123570726086, "grad_norm": 1.6419488191604614, "learning_rate": 1.6786066909014433e-05, "loss": 0.16998844146728515, "step": 57320 }, { "epoch": 0.28476202775597786, "grad_norm": 1.6270060539245605, "learning_rate": 1.6784885218023273e-05, "loss": 0.18569643497467042, "step": 57330 }, { "epoch": 0.2848116984393471, "grad_norm": 1.475311040878296, "learning_rate": 1.6783703351442015e-05, "loss": 0.16212397813796997, "step": 57340 }, { "epoch": 0.28486136912271637, "grad_norm": 0.9905627965927124, "learning_rate": 1.678252130930124e-05, "loss": 0.16157057285308837, "step": 57350 }, { "epoch": 0.2849110398060857, "grad_norm": 1.7922298908233643, "learning_rate": 1.6781339091631548e-05, "loss": 0.16621726751327515, "step": 57360 }, { "epoch": 0.2849607104894549, "grad_norm": 1.021733283996582, "learning_rate": 1.6780156698463523e-05, "loss": 0.15483735799789428, "step": 57370 }, { "epoch": 0.2850103811728242, "grad_norm": 0.6390948295593262, "learning_rate": 1.6778974129827767e-05, "loss": 0.16879534721374512, "step": 57380 }, { "epoch": 0.28506005185619343, "grad_norm": 0.897460401058197, "learning_rate": 1.6777791385754887e-05, "loss": 0.17085849046707152, "step": 57390 }, { "epoch": 0.2851097225395627, "grad_norm": 1.2782763242721558, "learning_rate": 1.677660846627549e-05, "loss": 0.17609720230102538, "step": 57400 }, { "epoch": 0.28515939322293193, "grad_norm": 0.6627143025398254, "learning_rate": 1.6775425371420194e-05, "loss": 0.17855257987976075, "step": 57410 }, { "epoch": 0.28520906390630124, "grad_norm": 1.7765002250671387, "learning_rate": 1.677424210121961e-05, "loss": 0.1732587695121765, "step": 57420 }, { "epoch": 0.2852587345896705, "grad_norm": 1.899682641029358, "learning_rate": 1.6773058655704368e-05, "loss": 0.16801817417144777, "step": 57430 }, { "epoch": 0.28530840527303974, "grad_norm": 1.6103949546813965, "learning_rate": 1.677187503490509e-05, "loss": 0.17075035572052003, "step": 57440 }, { "epoch": 0.285358075956409, "grad_norm": 1.2135405540466309, "learning_rate": 1.6770691238852406e-05, "loss": 0.1831794023513794, "step": 57450 }, { "epoch": 0.28540774663977825, "grad_norm": 1.2676621675491333, "learning_rate": 1.6769507267576958e-05, "loss": 0.15845940113067628, "step": 57460 }, { "epoch": 0.28545741732314756, "grad_norm": 2.2043418884277344, "learning_rate": 1.6768323121109384e-05, "loss": 0.16909513473510743, "step": 57470 }, { "epoch": 0.2855070880065168, "grad_norm": 0.7309322953224182, "learning_rate": 1.6767138799480326e-05, "loss": 0.17694102525711058, "step": 57480 }, { "epoch": 0.28555675868988606, "grad_norm": 1.0908938646316528, "learning_rate": 1.6765954302720437e-05, "loss": 0.17991188764572144, "step": 57490 }, { "epoch": 0.2856064293732553, "grad_norm": 1.24165678024292, "learning_rate": 1.6764769630860376e-05, "loss": 0.16762540340423585, "step": 57500 }, { "epoch": 0.28565610005662456, "grad_norm": 1.1836901903152466, "learning_rate": 1.6763584783930795e-05, "loss": 0.1751046895980835, "step": 57510 }, { "epoch": 0.2857057707399938, "grad_norm": 1.3807048797607422, "learning_rate": 1.6762399761962357e-05, "loss": 0.1802881360054016, "step": 57520 }, { "epoch": 0.2857554414233631, "grad_norm": 1.4038478136062622, "learning_rate": 1.6761214564985734e-05, "loss": 0.17670135498046874, "step": 57530 }, { "epoch": 0.2858051121067324, "grad_norm": 1.870008111000061, "learning_rate": 1.6760029193031593e-05, "loss": 0.18528033494949342, "step": 57540 }, { "epoch": 0.2858547827901016, "grad_norm": 2.211815357208252, "learning_rate": 1.675884364613062e-05, "loss": 0.17435016632080078, "step": 57550 }, { "epoch": 0.2859044534734709, "grad_norm": 1.0514436960220337, "learning_rate": 1.675765792431349e-05, "loss": 0.1719052314758301, "step": 57560 }, { "epoch": 0.28595412415684013, "grad_norm": 1.3130996227264404, "learning_rate": 1.6756472027610887e-05, "loss": 0.1778361678123474, "step": 57570 }, { "epoch": 0.28600379484020944, "grad_norm": 2.2171506881713867, "learning_rate": 1.675528595605351e-05, "loss": 0.1732628107070923, "step": 57580 }, { "epoch": 0.2860534655235787, "grad_norm": 1.1724315881729126, "learning_rate": 1.6754099709672043e-05, "loss": 0.1638727903366089, "step": 57590 }, { "epoch": 0.28610313620694794, "grad_norm": 3.27933406829834, "learning_rate": 1.675291328849719e-05, "loss": 0.1591670274734497, "step": 57600 }, { "epoch": 0.2861528068903172, "grad_norm": 0.748789370059967, "learning_rate": 1.6751726692559663e-05, "loss": 0.16928354501724244, "step": 57610 }, { "epoch": 0.28620247757368644, "grad_norm": 1.1821634769439697, "learning_rate": 1.675053992189016e-05, "loss": 0.18260517120361328, "step": 57620 }, { "epoch": 0.2862521482570557, "grad_norm": 1.431321620941162, "learning_rate": 1.6749352976519397e-05, "loss": 0.18156297206878663, "step": 57630 }, { "epoch": 0.286301818940425, "grad_norm": 1.4758068323135376, "learning_rate": 1.6748165856478097e-05, "loss": 0.18932509422302246, "step": 57640 }, { "epoch": 0.28635148962379425, "grad_norm": 1.1326513290405273, "learning_rate": 1.6746978561796975e-05, "loss": 0.16174890995025634, "step": 57650 }, { "epoch": 0.2864011603071635, "grad_norm": 0.6227664947509766, "learning_rate": 1.674579109250676e-05, "loss": 0.17639362812042236, "step": 57660 }, { "epoch": 0.28645083099053276, "grad_norm": 1.0537294149398804, "learning_rate": 1.6744603448638187e-05, "loss": 0.18538215160369872, "step": 57670 }, { "epoch": 0.286500501673902, "grad_norm": 0.7796006798744202, "learning_rate": 1.6743415630221983e-05, "loss": 0.17566368579864503, "step": 57680 }, { "epoch": 0.2865501723572713, "grad_norm": 1.0897653102874756, "learning_rate": 1.6742227637288898e-05, "loss": 0.1637897491455078, "step": 57690 }, { "epoch": 0.28659984304064057, "grad_norm": 1.5731555223464966, "learning_rate": 1.6741039469869672e-05, "loss": 0.1685211181640625, "step": 57700 }, { "epoch": 0.2866495137240098, "grad_norm": 1.8263590335845947, "learning_rate": 1.673985112799506e-05, "loss": 0.18230838775634767, "step": 57710 }, { "epoch": 0.28669918440737907, "grad_norm": 0.9137693047523499, "learning_rate": 1.6738662611695802e-05, "loss": 0.18548290729522704, "step": 57720 }, { "epoch": 0.2867488550907483, "grad_norm": 1.674472689628601, "learning_rate": 1.673747392100267e-05, "loss": 0.16206755638122558, "step": 57730 }, { "epoch": 0.2867985257741176, "grad_norm": 1.8719542026519775, "learning_rate": 1.673628505594642e-05, "loss": 0.17493021488189697, "step": 57740 }, { "epoch": 0.2868481964574869, "grad_norm": 0.9193127155303955, "learning_rate": 1.6735096016557825e-05, "loss": 0.17935256958007811, "step": 57750 }, { "epoch": 0.28689786714085613, "grad_norm": 0.7286001443862915, "learning_rate": 1.673390680286765e-05, "loss": 0.1845882296562195, "step": 57760 }, { "epoch": 0.2869475378242254, "grad_norm": 1.2433730363845825, "learning_rate": 1.6732717414906675e-05, "loss": 0.17575585842132568, "step": 57770 }, { "epoch": 0.28699720850759464, "grad_norm": 1.5402536392211914, "learning_rate": 1.673152785270568e-05, "loss": 0.16144813299179078, "step": 57780 }, { "epoch": 0.2870468791909639, "grad_norm": 1.4877640008926392, "learning_rate": 1.6730338116295454e-05, "loss": 0.16714184284210204, "step": 57790 }, { "epoch": 0.2870965498743332, "grad_norm": 0.5492892265319824, "learning_rate": 1.672914820570678e-05, "loss": 0.1723373770713806, "step": 57800 }, { "epoch": 0.28714622055770245, "grad_norm": 0.5513433218002319, "learning_rate": 1.6727958120970455e-05, "loss": 0.16851069927215576, "step": 57810 }, { "epoch": 0.2871958912410717, "grad_norm": 1.530755639076233, "learning_rate": 1.6726767862117283e-05, "loss": 0.18399560451507568, "step": 57820 }, { "epoch": 0.28724556192444095, "grad_norm": 0.9937992095947266, "learning_rate": 1.672557742917806e-05, "loss": 0.17956315279006957, "step": 57830 }, { "epoch": 0.2872952326078102, "grad_norm": 0.9740555882453918, "learning_rate": 1.67243868221836e-05, "loss": 0.1743282675743103, "step": 57840 }, { "epoch": 0.28734490329117945, "grad_norm": 0.7994943857192993, "learning_rate": 1.6723196041164712e-05, "loss": 0.14932048320770264, "step": 57850 }, { "epoch": 0.28739457397454876, "grad_norm": 1.1611769199371338, "learning_rate": 1.6722005086152213e-05, "loss": 0.1884608030319214, "step": 57860 }, { "epoch": 0.287444244657918, "grad_norm": 2.202622175216675, "learning_rate": 1.6720813957176924e-05, "loss": 0.1656068444252014, "step": 57870 }, { "epoch": 0.28749391534128726, "grad_norm": 1.6730109453201294, "learning_rate": 1.6719622654269677e-05, "loss": 0.1825266122817993, "step": 57880 }, { "epoch": 0.2875435860246565, "grad_norm": 0.8708582520484924, "learning_rate": 1.6718431177461294e-05, "loss": 0.1664170503616333, "step": 57890 }, { "epoch": 0.28759325670802577, "grad_norm": 1.092253565788269, "learning_rate": 1.6717239526782614e-05, "loss": 0.16849316358566285, "step": 57900 }, { "epoch": 0.2876429273913951, "grad_norm": 1.1461971998214722, "learning_rate": 1.6716047702264475e-05, "loss": 0.17753214836120607, "step": 57910 }, { "epoch": 0.2876925980747643, "grad_norm": 0.6544420719146729, "learning_rate": 1.6714855703937724e-05, "loss": 0.16854145526885986, "step": 57920 }, { "epoch": 0.2877422687581336, "grad_norm": 0.7681586742401123, "learning_rate": 1.6713663531833205e-05, "loss": 0.178552508354187, "step": 57930 }, { "epoch": 0.28779193944150283, "grad_norm": 0.9317878484725952, "learning_rate": 1.6712471185981774e-05, "loss": 0.17641925811767578, "step": 57940 }, { "epoch": 0.2878416101248721, "grad_norm": 1.8213444948196411, "learning_rate": 1.6711278666414288e-05, "loss": 0.1728972911834717, "step": 57950 }, { "epoch": 0.28789128080824133, "grad_norm": 1.2733694314956665, "learning_rate": 1.6710085973161608e-05, "loss": 0.17367029190063477, "step": 57960 }, { "epoch": 0.28794095149161064, "grad_norm": 1.0748385190963745, "learning_rate": 1.67088931062546e-05, "loss": 0.1821570873260498, "step": 57970 }, { "epoch": 0.2879906221749799, "grad_norm": 1.6942477226257324, "learning_rate": 1.670770006572414e-05, "loss": 0.17145419120788574, "step": 57980 }, { "epoch": 0.28804029285834915, "grad_norm": 0.8464547991752625, "learning_rate": 1.6706506851601094e-05, "loss": 0.1827038884162903, "step": 57990 }, { "epoch": 0.2880899635417184, "grad_norm": 1.2513842582702637, "learning_rate": 1.670531346391635e-05, "loss": 0.16802611351013183, "step": 58000 }, { "epoch": 0.28813963422508765, "grad_norm": 0.7166813015937805, "learning_rate": 1.670411990270079e-05, "loss": 0.17985477447509765, "step": 58010 }, { "epoch": 0.28818930490845696, "grad_norm": 0.9101690053939819, "learning_rate": 1.6702926167985302e-05, "loss": 0.1673330545425415, "step": 58020 }, { "epoch": 0.2882389755918262, "grad_norm": 1.2317110300064087, "learning_rate": 1.670173225980078e-05, "loss": 0.16459826231002808, "step": 58030 }, { "epoch": 0.28828864627519546, "grad_norm": 1.2739331722259521, "learning_rate": 1.6700538178178125e-05, "loss": 0.17771307229995728, "step": 58040 }, { "epoch": 0.2883383169585647, "grad_norm": 0.8159629702568054, "learning_rate": 1.669934392314823e-05, "loss": 0.15268447399139404, "step": 58050 }, { "epoch": 0.28838798764193396, "grad_norm": 1.5898613929748535, "learning_rate": 1.6698149494742013e-05, "loss": 0.17905433177948, "step": 58060 }, { "epoch": 0.2884376583253032, "grad_norm": 1.4133648872375488, "learning_rate": 1.6696954892990377e-05, "loss": 0.17215778827667236, "step": 58070 }, { "epoch": 0.2884873290086725, "grad_norm": 1.0109601020812988, "learning_rate": 1.6695760117924243e-05, "loss": 0.1809984803199768, "step": 58080 }, { "epoch": 0.2885369996920418, "grad_norm": 0.8106544613838196, "learning_rate": 1.669456516957453e-05, "loss": 0.16399246454238892, "step": 58090 }, { "epoch": 0.288586670375411, "grad_norm": 1.398273229598999, "learning_rate": 1.669337004797216e-05, "loss": 0.19689745903015138, "step": 58100 }, { "epoch": 0.2886363410587803, "grad_norm": 0.5730389356613159, "learning_rate": 1.6692174753148064e-05, "loss": 0.1719098687171936, "step": 58110 }, { "epoch": 0.28868601174214953, "grad_norm": 1.304212212562561, "learning_rate": 1.669097928513318e-05, "loss": 0.15770938396453857, "step": 58120 }, { "epoch": 0.28873568242551884, "grad_norm": 0.674691379070282, "learning_rate": 1.668978364395844e-05, "loss": 0.1769064784049988, "step": 58130 }, { "epoch": 0.2887853531088881, "grad_norm": 1.578779697418213, "learning_rate": 1.6688587829654788e-05, "loss": 0.16066217422485352, "step": 58140 }, { "epoch": 0.28883502379225734, "grad_norm": 0.6851184368133545, "learning_rate": 1.6687391842253172e-05, "loss": 0.16480038166046143, "step": 58150 }, { "epoch": 0.2888846944756266, "grad_norm": 1.6297053098678589, "learning_rate": 1.6686195681784547e-05, "loss": 0.17963874340057373, "step": 58160 }, { "epoch": 0.28893436515899584, "grad_norm": 1.0271846055984497, "learning_rate": 1.6684999348279864e-05, "loss": 0.16981104612350464, "step": 58170 }, { "epoch": 0.2889840358423651, "grad_norm": 1.1143494844436646, "learning_rate": 1.6683802841770087e-05, "loss": 0.16211572885513306, "step": 58180 }, { "epoch": 0.2890337065257344, "grad_norm": 1.5161688327789307, "learning_rate": 1.668260616228618e-05, "loss": 0.16234915256500243, "step": 58190 }, { "epoch": 0.28908337720910365, "grad_norm": 1.5248146057128906, "learning_rate": 1.6681409309859113e-05, "loss": 0.18077595233917237, "step": 58200 }, { "epoch": 0.2891330478924729, "grad_norm": 1.3235032558441162, "learning_rate": 1.6680212284519857e-05, "loss": 0.1851675033569336, "step": 58210 }, { "epoch": 0.28918271857584216, "grad_norm": 1.472610354423523, "learning_rate": 1.6679015086299398e-05, "loss": 0.18540362119674683, "step": 58220 }, { "epoch": 0.2892323892592114, "grad_norm": 2.362051010131836, "learning_rate": 1.667781771522871e-05, "loss": 0.17941219806671144, "step": 58230 }, { "epoch": 0.2892820599425807, "grad_norm": 1.3547441959381104, "learning_rate": 1.6676620171338784e-05, "loss": 0.18188271522521973, "step": 58240 }, { "epoch": 0.28933173062594997, "grad_norm": 0.8376471400260925, "learning_rate": 1.6675422454660615e-05, "loss": 0.17364294528961183, "step": 58250 }, { "epoch": 0.2893814013093192, "grad_norm": 2.1573400497436523, "learning_rate": 1.6674224565225197e-05, "loss": 0.166717004776001, "step": 58260 }, { "epoch": 0.28943107199268847, "grad_norm": 1.0405457019805908, "learning_rate": 1.667302650306353e-05, "loss": 0.16336935758590698, "step": 58270 }, { "epoch": 0.2894807426760577, "grad_norm": 1.167864203453064, "learning_rate": 1.6671828268206618e-05, "loss": 0.17628107070922852, "step": 58280 }, { "epoch": 0.289530413359427, "grad_norm": 1.8629518747329712, "learning_rate": 1.6670629860685474e-05, "loss": 0.17477192878723144, "step": 58290 }, { "epoch": 0.2895800840427963, "grad_norm": 0.9792229533195496, "learning_rate": 1.6669431280531113e-05, "loss": 0.15950956344604492, "step": 58300 }, { "epoch": 0.28962975472616553, "grad_norm": 1.320659875869751, "learning_rate": 1.666823252777455e-05, "loss": 0.18044626712799072, "step": 58310 }, { "epoch": 0.2896794254095348, "grad_norm": 0.8272233009338379, "learning_rate": 1.6667033602446812e-05, "loss": 0.19048352241516114, "step": 58320 }, { "epoch": 0.28972909609290404, "grad_norm": 1.6253271102905273, "learning_rate": 1.6665834504578925e-05, "loss": 0.1634970188140869, "step": 58330 }, { "epoch": 0.2897787667762733, "grad_norm": 0.8368366360664368, "learning_rate": 1.6664635234201922e-05, "loss": 0.16997084617614747, "step": 58340 }, { "epoch": 0.2898284374596426, "grad_norm": 0.9189881682395935, "learning_rate": 1.6663435791346836e-05, "loss": 0.17103230953216553, "step": 58350 }, { "epoch": 0.28987810814301185, "grad_norm": 0.7016109824180603, "learning_rate": 1.6662236176044712e-05, "loss": 0.1756758451461792, "step": 58360 }, { "epoch": 0.2899277788263811, "grad_norm": 1.1112791299819946, "learning_rate": 1.6661036388326595e-05, "loss": 0.1794920563697815, "step": 58370 }, { "epoch": 0.28997744950975035, "grad_norm": 1.8836520910263062, "learning_rate": 1.665983642822354e-05, "loss": 0.15592503547668457, "step": 58380 }, { "epoch": 0.2900271201931196, "grad_norm": 2.3492772579193115, "learning_rate": 1.6658636295766587e-05, "loss": 0.16830230951309205, "step": 58390 }, { "epoch": 0.29007679087648885, "grad_norm": 0.8939653038978577, "learning_rate": 1.665743599098681e-05, "loss": 0.1781094789505005, "step": 58400 }, { "epoch": 0.29012646155985816, "grad_norm": 1.6058554649353027, "learning_rate": 1.6656235513915266e-05, "loss": 0.16349165439605712, "step": 58410 }, { "epoch": 0.2901761322432274, "grad_norm": 0.7588523626327515, "learning_rate": 1.6655034864583018e-05, "loss": 0.16233739852905274, "step": 58420 }, { "epoch": 0.29022580292659667, "grad_norm": 1.0295997858047485, "learning_rate": 1.665383404302115e-05, "loss": 0.1794008731842041, "step": 58430 }, { "epoch": 0.2902754736099659, "grad_norm": 1.5083805322647095, "learning_rate": 1.6652633049260726e-05, "loss": 0.17444770336151122, "step": 58440 }, { "epoch": 0.29032514429333517, "grad_norm": 1.0203924179077148, "learning_rate": 1.6651431883332838e-05, "loss": 0.18200507164001464, "step": 58450 }, { "epoch": 0.2903748149767045, "grad_norm": 0.7903437614440918, "learning_rate": 1.6650230545268567e-05, "loss": 0.1689828634262085, "step": 58460 }, { "epoch": 0.29042448566007373, "grad_norm": 0.9122129082679749, "learning_rate": 1.6649029035099e-05, "loss": 0.17742371559143066, "step": 58470 }, { "epoch": 0.290474156343443, "grad_norm": 2.11604380607605, "learning_rate": 1.664782735285524e-05, "loss": 0.17882127761840821, "step": 58480 }, { "epoch": 0.29052382702681223, "grad_norm": 1.2644118070602417, "learning_rate": 1.664662549856838e-05, "loss": 0.18345521688461303, "step": 58490 }, { "epoch": 0.2905734977101815, "grad_norm": 1.4311226606369019, "learning_rate": 1.6645423472269523e-05, "loss": 0.16659266948699952, "step": 58500 }, { "epoch": 0.29062316839355073, "grad_norm": 1.1536861658096313, "learning_rate": 1.6644221273989778e-05, "loss": 0.18378803730010987, "step": 58510 }, { "epoch": 0.29067283907692004, "grad_norm": 0.666755735874176, "learning_rate": 1.664301890376026e-05, "loss": 0.1673293113708496, "step": 58520 }, { "epoch": 0.2907225097602893, "grad_norm": 0.6691044569015503, "learning_rate": 1.6641816361612085e-05, "loss": 0.19373029470443726, "step": 58530 }, { "epoch": 0.29077218044365855, "grad_norm": 0.7227533459663391, "learning_rate": 1.664061364757637e-05, "loss": 0.1727329134941101, "step": 58540 }, { "epoch": 0.2908218511270278, "grad_norm": 2.1589736938476562, "learning_rate": 1.6639410761684243e-05, "loss": 0.16743451356887817, "step": 58550 }, { "epoch": 0.29087152181039705, "grad_norm": 0.9610415697097778, "learning_rate": 1.6638207703966838e-05, "loss": 0.17595999240875243, "step": 58560 }, { "epoch": 0.29092119249376636, "grad_norm": 1.6365103721618652, "learning_rate": 1.6637004474455288e-05, "loss": 0.1818188190460205, "step": 58570 }, { "epoch": 0.2909708631771356, "grad_norm": 1.2538578510284424, "learning_rate": 1.663580107318073e-05, "loss": 0.17882063388824462, "step": 58580 }, { "epoch": 0.29102053386050486, "grad_norm": 0.9539650678634644, "learning_rate": 1.6634597500174304e-05, "loss": 0.16986281871795655, "step": 58590 }, { "epoch": 0.2910702045438741, "grad_norm": 1.047972321510315, "learning_rate": 1.6633393755467168e-05, "loss": 0.17563817501068116, "step": 58600 }, { "epoch": 0.29111987522724336, "grad_norm": 0.6348991394042969, "learning_rate": 1.6632189839090465e-05, "loss": 0.16590774059295654, "step": 58610 }, { "epoch": 0.2911695459106126, "grad_norm": 1.658834457397461, "learning_rate": 1.6630985751075356e-05, "loss": 0.1811084747314453, "step": 58620 }, { "epoch": 0.2912192165939819, "grad_norm": 0.9630634784698486, "learning_rate": 1.6629781491453e-05, "loss": 0.17252447605133056, "step": 58630 }, { "epoch": 0.2912688872773512, "grad_norm": 1.4806407690048218, "learning_rate": 1.662857706025457e-05, "loss": 0.17607110738754272, "step": 58640 }, { "epoch": 0.2913185579607204, "grad_norm": 0.9684041738510132, "learning_rate": 1.6627372457511226e-05, "loss": 0.17136712074279786, "step": 58650 }, { "epoch": 0.2913682286440897, "grad_norm": 1.966618537902832, "learning_rate": 1.6626167683254153e-05, "loss": 0.16972291469573975, "step": 58660 }, { "epoch": 0.29141789932745893, "grad_norm": 0.8350143432617188, "learning_rate": 1.662496273751452e-05, "loss": 0.15871033668518067, "step": 58670 }, { "epoch": 0.29146757001082824, "grad_norm": 1.9776408672332764, "learning_rate": 1.6623757620323516e-05, "loss": 0.17675421237945557, "step": 58680 }, { "epoch": 0.2915172406941975, "grad_norm": 2.0233612060546875, "learning_rate": 1.662255233171233e-05, "loss": 0.17828506231307983, "step": 58690 }, { "epoch": 0.29156691137756674, "grad_norm": 1.9450010061264038, "learning_rate": 1.6621346871712152e-05, "loss": 0.1854712963104248, "step": 58700 }, { "epoch": 0.291616582060936, "grad_norm": 0.668189287185669, "learning_rate": 1.6620141240354177e-05, "loss": 0.18094779253005983, "step": 58710 }, { "epoch": 0.29166625274430524, "grad_norm": 1.2782690525054932, "learning_rate": 1.6618935437669608e-05, "loss": 0.16485333442687988, "step": 58720 }, { "epoch": 0.2917159234276745, "grad_norm": 1.042404294013977, "learning_rate": 1.6617729463689655e-05, "loss": 0.16323521137237548, "step": 58730 }, { "epoch": 0.2917655941110438, "grad_norm": 2.1256606578826904, "learning_rate": 1.6616523318445525e-05, "loss": 0.1642817735671997, "step": 58740 }, { "epoch": 0.29181526479441305, "grad_norm": 0.9336919188499451, "learning_rate": 1.661531700196843e-05, "loss": 0.18899685144424438, "step": 58750 }, { "epoch": 0.2918649354777823, "grad_norm": 2.1013166904449463, "learning_rate": 1.661411051428959e-05, "loss": 0.16925060749053955, "step": 58760 }, { "epoch": 0.29191460616115156, "grad_norm": 1.4467432498931885, "learning_rate": 1.6612903855440227e-05, "loss": 0.1702810764312744, "step": 58770 }, { "epoch": 0.2919642768445208, "grad_norm": 2.2637851238250732, "learning_rate": 1.6611697025451576e-05, "loss": 0.16675761938095093, "step": 58780 }, { "epoch": 0.2920139475278901, "grad_norm": 1.0762356519699097, "learning_rate": 1.6610490024354863e-05, "loss": 0.17743637561798095, "step": 58790 }, { "epoch": 0.29206361821125937, "grad_norm": 2.2465171813964844, "learning_rate": 1.6609282852181326e-05, "loss": 0.17449780702590942, "step": 58800 }, { "epoch": 0.2921132888946286, "grad_norm": 1.2569050788879395, "learning_rate": 1.6608075508962206e-05, "loss": 0.17813981771469117, "step": 58810 }, { "epoch": 0.29216295957799787, "grad_norm": 1.3756498098373413, "learning_rate": 1.660686799472875e-05, "loss": 0.16737202405929566, "step": 58820 }, { "epoch": 0.2922126302613671, "grad_norm": 1.7150295972824097, "learning_rate": 1.66056603095122e-05, "loss": 0.17964862585067748, "step": 58830 }, { "epoch": 0.2922623009447364, "grad_norm": 0.9909020066261292, "learning_rate": 1.6604452453343826e-05, "loss": 0.16719207763671876, "step": 58840 }, { "epoch": 0.2923119716281057, "grad_norm": 0.9194408655166626, "learning_rate": 1.660324442625487e-05, "loss": 0.15659226179122926, "step": 58850 }, { "epoch": 0.29236164231147493, "grad_norm": 1.1498769521713257, "learning_rate": 1.6602036228276612e-05, "loss": 0.1896116018295288, "step": 58860 }, { "epoch": 0.2924113129948442, "grad_norm": 1.5779403448104858, "learning_rate": 1.660082785944031e-05, "loss": 0.16660077571868898, "step": 58870 }, { "epoch": 0.29246098367821344, "grad_norm": 0.9278914928436279, "learning_rate": 1.6599619319777233e-05, "loss": 0.16119579076766968, "step": 58880 }, { "epoch": 0.2925106543615827, "grad_norm": 0.6929521560668945, "learning_rate": 1.659841060931866e-05, "loss": 0.16431131362915039, "step": 58890 }, { "epoch": 0.29256032504495194, "grad_norm": 1.1308597326278687, "learning_rate": 1.6597201728095878e-05, "loss": 0.16387637853622436, "step": 58900 }, { "epoch": 0.29260999572832125, "grad_norm": 1.9817806482315063, "learning_rate": 1.6595992676140164e-05, "loss": 0.16478822231292725, "step": 58910 }, { "epoch": 0.2926596664116905, "grad_norm": 0.9755784273147583, "learning_rate": 1.6594783453482818e-05, "loss": 0.15813418626785278, "step": 58920 }, { "epoch": 0.29270933709505975, "grad_norm": 0.6756320595741272, "learning_rate": 1.659357406015512e-05, "loss": 0.1702764391899109, "step": 58930 }, { "epoch": 0.292759007778429, "grad_norm": 0.7075243592262268, "learning_rate": 1.6592364496188382e-05, "loss": 0.16418535709381105, "step": 58940 }, { "epoch": 0.29280867846179826, "grad_norm": 1.0220558643341064, "learning_rate": 1.65911547616139e-05, "loss": 0.17670769691467286, "step": 58950 }, { "epoch": 0.29285834914516756, "grad_norm": 1.0718576908111572, "learning_rate": 1.6589944856462984e-05, "loss": 0.17877590656280518, "step": 58960 }, { "epoch": 0.2929080198285368, "grad_norm": 2.2296934127807617, "learning_rate": 1.6588734780766943e-05, "loss": 0.1751536250114441, "step": 58970 }, { "epoch": 0.29295769051190607, "grad_norm": 1.111032485961914, "learning_rate": 1.6587524534557095e-05, "loss": 0.16729238033294677, "step": 58980 }, { "epoch": 0.2930073611952753, "grad_norm": 2.1325936317443848, "learning_rate": 1.658631411786476e-05, "loss": 0.17167586088180542, "step": 58990 }, { "epoch": 0.29305703187864457, "grad_norm": 1.3418688774108887, "learning_rate": 1.6585103530721266e-05, "loss": 0.16949687004089356, "step": 59000 }, { "epoch": 0.2931067025620138, "grad_norm": 1.6329139471054077, "learning_rate": 1.658389277315794e-05, "loss": 0.15475521087646485, "step": 59010 }, { "epoch": 0.29315637324538313, "grad_norm": 0.9684159159660339, "learning_rate": 1.6582681845206115e-05, "loss": 0.17473785877227782, "step": 59020 }, { "epoch": 0.2932060439287524, "grad_norm": 0.8073574900627136, "learning_rate": 1.658147074689713e-05, "loss": 0.14739735126495362, "step": 59030 }, { "epoch": 0.29325571461212163, "grad_norm": 1.1171537637710571, "learning_rate": 1.658025947826233e-05, "loss": 0.16493487358093262, "step": 59040 }, { "epoch": 0.2933053852954909, "grad_norm": 2.1451592445373535, "learning_rate": 1.6579048039333062e-05, "loss": 0.16116610765457154, "step": 59050 }, { "epoch": 0.29335505597886014, "grad_norm": 1.446009635925293, "learning_rate": 1.6577836430140675e-05, "loss": 0.1765398859977722, "step": 59060 }, { "epoch": 0.29340472666222944, "grad_norm": 1.5223292112350464, "learning_rate": 1.6576624650716526e-05, "loss": 0.17615411281585694, "step": 59070 }, { "epoch": 0.2934543973455987, "grad_norm": 1.1928751468658447, "learning_rate": 1.6575412701091976e-05, "loss": 0.17295439243316652, "step": 59080 }, { "epoch": 0.29350406802896795, "grad_norm": 1.543602705001831, "learning_rate": 1.6574200581298384e-05, "loss": 0.17068812847137452, "step": 59090 }, { "epoch": 0.2935537387123372, "grad_norm": 1.2736505270004272, "learning_rate": 1.6572988291367132e-05, "loss": 0.1753499150276184, "step": 59100 }, { "epoch": 0.29360340939570645, "grad_norm": 0.8850905299186707, "learning_rate": 1.657177583132958e-05, "loss": 0.1688237428665161, "step": 59110 }, { "epoch": 0.2936530800790757, "grad_norm": 1.3773161172866821, "learning_rate": 1.6570563201217114e-05, "loss": 0.16836230754852294, "step": 59120 }, { "epoch": 0.293702750762445, "grad_norm": 1.050779104232788, "learning_rate": 1.6569350401061116e-05, "loss": 0.15643699169158937, "step": 59130 }, { "epoch": 0.29375242144581426, "grad_norm": 0.9438339471817017, "learning_rate": 1.656813743089297e-05, "loss": 0.1720650911331177, "step": 59140 }, { "epoch": 0.2938020921291835, "grad_norm": 2.270204782485962, "learning_rate": 1.6566924290744067e-05, "loss": 0.17424533367156983, "step": 59150 }, { "epoch": 0.29385176281255276, "grad_norm": 2.692187547683716, "learning_rate": 1.6565710980645805e-05, "loss": 0.18354958295822144, "step": 59160 }, { "epoch": 0.293901433495922, "grad_norm": 0.8151272535324097, "learning_rate": 1.656449750062958e-05, "loss": 0.1664425253868103, "step": 59170 }, { "epoch": 0.2939511041792913, "grad_norm": 1.4775668382644653, "learning_rate": 1.6563283850726805e-05, "loss": 0.17400022745132446, "step": 59180 }, { "epoch": 0.2940007748626606, "grad_norm": 1.7317997217178345, "learning_rate": 1.6562070030968876e-05, "loss": 0.1759510040283203, "step": 59190 }, { "epoch": 0.2940504455460298, "grad_norm": 0.9891210198402405, "learning_rate": 1.6560856041387218e-05, "loss": 0.180908465385437, "step": 59200 }, { "epoch": 0.2941001162293991, "grad_norm": 2.2274930477142334, "learning_rate": 1.6559641882013238e-05, "loss": 0.17869086265563966, "step": 59210 }, { "epoch": 0.29414978691276833, "grad_norm": 1.0759915113449097, "learning_rate": 1.6558427552878372e-05, "loss": 0.19017300605773926, "step": 59220 }, { "epoch": 0.2941994575961376, "grad_norm": 1.1033865213394165, "learning_rate": 1.6557213054014034e-05, "loss": 0.17242281436920165, "step": 59230 }, { "epoch": 0.2942491282795069, "grad_norm": 1.0249648094177246, "learning_rate": 1.655599838545166e-05, "loss": 0.1588583469390869, "step": 59240 }, { "epoch": 0.29429879896287614, "grad_norm": 0.6653825044631958, "learning_rate": 1.655478354722268e-05, "loss": 0.17551698684692382, "step": 59250 }, { "epoch": 0.2943484696462454, "grad_norm": 0.9059562683105469, "learning_rate": 1.655356853935854e-05, "loss": 0.17399823665618896, "step": 59260 }, { "epoch": 0.29439814032961464, "grad_norm": 0.9946908354759216, "learning_rate": 1.6552353361890682e-05, "loss": 0.16320266723632812, "step": 59270 }, { "epoch": 0.2944478110129839, "grad_norm": 0.9784996509552002, "learning_rate": 1.6551138014850553e-05, "loss": 0.17400752305984496, "step": 59280 }, { "epoch": 0.2944974816963532, "grad_norm": 1.1430915594100952, "learning_rate": 1.6549922498269604e-05, "loss": 0.16121556758880615, "step": 59290 }, { "epoch": 0.29454715237972245, "grad_norm": 0.9329724907875061, "learning_rate": 1.6548706812179296e-05, "loss": 0.16825628280639648, "step": 59300 }, { "epoch": 0.2945968230630917, "grad_norm": 1.3318382501602173, "learning_rate": 1.654749095661109e-05, "loss": 0.18379542827606202, "step": 59310 }, { "epoch": 0.29464649374646096, "grad_norm": 0.8638004064559937, "learning_rate": 1.6546274931596448e-05, "loss": 0.16838057041168214, "step": 59320 }, { "epoch": 0.2946961644298302, "grad_norm": 0.77080899477005, "learning_rate": 1.6545058737166845e-05, "loss": 0.15981775522232056, "step": 59330 }, { "epoch": 0.29474583511319946, "grad_norm": 1.430753469467163, "learning_rate": 1.6543842373353753e-05, "loss": 0.172583270072937, "step": 59340 }, { "epoch": 0.29479550579656877, "grad_norm": 1.6747581958770752, "learning_rate": 1.654262584018865e-05, "loss": 0.1668141722679138, "step": 59350 }, { "epoch": 0.294845176479938, "grad_norm": 1.3592227697372437, "learning_rate": 1.6541409137703026e-05, "loss": 0.17134013175964355, "step": 59360 }, { "epoch": 0.29489484716330727, "grad_norm": 1.2228038311004639, "learning_rate": 1.654019226592836e-05, "loss": 0.16304916143417358, "step": 59370 }, { "epoch": 0.2949445178466765, "grad_norm": 0.74290531873703, "learning_rate": 1.653897522489615e-05, "loss": 0.17197008132934571, "step": 59380 }, { "epoch": 0.2949941885300458, "grad_norm": 1.0674725770950317, "learning_rate": 1.653775801463789e-05, "loss": 0.1753947377204895, "step": 59390 }, { "epoch": 0.2950438592134151, "grad_norm": 0.5712097883224487, "learning_rate": 1.653654063518508e-05, "loss": 0.17261340618133544, "step": 59400 }, { "epoch": 0.29509352989678433, "grad_norm": 1.327953577041626, "learning_rate": 1.653532308656922e-05, "loss": 0.17563358545303345, "step": 59410 }, { "epoch": 0.2951432005801536, "grad_norm": 1.1367557048797607, "learning_rate": 1.6534105368821837e-05, "loss": 0.1715155601501465, "step": 59420 }, { "epoch": 0.29519287126352284, "grad_norm": 1.049736738204956, "learning_rate": 1.653288748197443e-05, "loss": 0.18082748651504515, "step": 59430 }, { "epoch": 0.2952425419468921, "grad_norm": 1.5809534788131714, "learning_rate": 1.653166942605852e-05, "loss": 0.16884543895721435, "step": 59440 }, { "epoch": 0.29529221263026134, "grad_norm": 1.5274335145950317, "learning_rate": 1.6530451201105635e-05, "loss": 0.17764830589294434, "step": 59450 }, { "epoch": 0.29534188331363065, "grad_norm": 1.7139180898666382, "learning_rate": 1.6529232807147296e-05, "loss": 0.16741635799407958, "step": 59460 }, { "epoch": 0.2953915539969999, "grad_norm": 1.2107676267623901, "learning_rate": 1.652801424421504e-05, "loss": 0.17027831077575684, "step": 59470 }, { "epoch": 0.29544122468036915, "grad_norm": 1.2928639650344849, "learning_rate": 1.6526795512340396e-05, "loss": 0.17412883043289185, "step": 59480 }, { "epoch": 0.2954908953637384, "grad_norm": 1.1521906852722168, "learning_rate": 1.6525576611554912e-05, "loss": 0.18736600875854492, "step": 59490 }, { "epoch": 0.29554056604710766, "grad_norm": 1.0722335577011108, "learning_rate": 1.6524357541890127e-05, "loss": 0.16659278869628907, "step": 59500 }, { "epoch": 0.29559023673047696, "grad_norm": 1.8142215013504028, "learning_rate": 1.6523138303377595e-05, "loss": 0.18173785209655763, "step": 59510 }, { "epoch": 0.2956399074138462, "grad_norm": 1.2598228454589844, "learning_rate": 1.6521918896048865e-05, "loss": 0.1708935022354126, "step": 59520 }, { "epoch": 0.29568957809721547, "grad_norm": 0.9645238518714905, "learning_rate": 1.6520699319935494e-05, "loss": 0.17480835914611817, "step": 59530 }, { "epoch": 0.2957392487805847, "grad_norm": 2.2183609008789062, "learning_rate": 1.6519479575069048e-05, "loss": 0.1875010371208191, "step": 59540 }, { "epoch": 0.29578891946395397, "grad_norm": 1.5298068523406982, "learning_rate": 1.6518259661481092e-05, "loss": 0.16869041919708253, "step": 59550 }, { "epoch": 0.2958385901473232, "grad_norm": 1.4894073009490967, "learning_rate": 1.65170395792032e-05, "loss": 0.17471896409988402, "step": 59560 }, { "epoch": 0.29588826083069253, "grad_norm": 0.7639135122299194, "learning_rate": 1.651581932826694e-05, "loss": 0.15300607681274414, "step": 59570 }, { "epoch": 0.2959379315140618, "grad_norm": 1.180318832397461, "learning_rate": 1.65145989087039e-05, "loss": 0.16655983924865722, "step": 59580 }, { "epoch": 0.29598760219743103, "grad_norm": 0.6345869898796082, "learning_rate": 1.6513378320545653e-05, "loss": 0.17041007280349732, "step": 59590 }, { "epoch": 0.2960372728808003, "grad_norm": 1.5860962867736816, "learning_rate": 1.6512157563823802e-05, "loss": 0.18281195163726807, "step": 59600 }, { "epoch": 0.29608694356416954, "grad_norm": 0.7666813135147095, "learning_rate": 1.6510936638569926e-05, "loss": 0.17614271640777587, "step": 59610 }, { "epoch": 0.29613661424753884, "grad_norm": 0.9219846725463867, "learning_rate": 1.6509715544815627e-05, "loss": 0.184479284286499, "step": 59620 }, { "epoch": 0.2961862849309081, "grad_norm": 1.1740790605545044, "learning_rate": 1.650849428259251e-05, "loss": 0.16960630416870118, "step": 59630 }, { "epoch": 0.29623595561427735, "grad_norm": 1.048003911972046, "learning_rate": 1.650727285193218e-05, "loss": 0.1767014741897583, "step": 59640 }, { "epoch": 0.2962856262976466, "grad_norm": 0.6339986324310303, "learning_rate": 1.650605125286624e-05, "loss": 0.18526227474212648, "step": 59650 }, { "epoch": 0.29633529698101585, "grad_norm": 1.4171193838119507, "learning_rate": 1.650482948542631e-05, "loss": 0.1738142728805542, "step": 59660 }, { "epoch": 0.2963849676643851, "grad_norm": 1.064347267150879, "learning_rate": 1.650360754964401e-05, "loss": 0.17439185380935668, "step": 59670 }, { "epoch": 0.2964346383477544, "grad_norm": 1.6577955484390259, "learning_rate": 1.6502385445550964e-05, "loss": 0.18081326484680177, "step": 59680 }, { "epoch": 0.29648430903112366, "grad_norm": 0.9965952038764954, "learning_rate": 1.6501163173178792e-05, "loss": 0.16901637315750123, "step": 59690 }, { "epoch": 0.2965339797144929, "grad_norm": 1.4436737298965454, "learning_rate": 1.6499940732559135e-05, "loss": 0.17103757858276367, "step": 59700 }, { "epoch": 0.29658365039786216, "grad_norm": 1.2323555946350098, "learning_rate": 1.6498718123723624e-05, "loss": 0.1705808162689209, "step": 59710 }, { "epoch": 0.2966333210812314, "grad_norm": 1.2912418842315674, "learning_rate": 1.6497495346703903e-05, "loss": 0.17761390209197997, "step": 59720 }, { "epoch": 0.2966829917646007, "grad_norm": 1.6991174221038818, "learning_rate": 1.6496272401531615e-05, "loss": 0.1651278018951416, "step": 59730 }, { "epoch": 0.29673266244797, "grad_norm": 1.7325366735458374, "learning_rate": 1.6495049288238408e-05, "loss": 0.18790220022201537, "step": 59740 }, { "epoch": 0.2967823331313392, "grad_norm": 1.5907392501831055, "learning_rate": 1.6493826006855934e-05, "loss": 0.18570311069488527, "step": 59750 }, { "epoch": 0.2968320038147085, "grad_norm": 1.399511694908142, "learning_rate": 1.6492602557415856e-05, "loss": 0.1774768352508545, "step": 59760 }, { "epoch": 0.29688167449807773, "grad_norm": 1.1194405555725098, "learning_rate": 1.649137893994984e-05, "loss": 0.18050739765167237, "step": 59770 }, { "epoch": 0.296931345181447, "grad_norm": 1.617240309715271, "learning_rate": 1.649015515448954e-05, "loss": 0.163765549659729, "step": 59780 }, { "epoch": 0.2969810158648163, "grad_norm": 1.4449331760406494, "learning_rate": 1.6488931201066637e-05, "loss": 0.1582504391670227, "step": 59790 }, { "epoch": 0.29703068654818554, "grad_norm": 1.2378956079483032, "learning_rate": 1.6487707079712804e-05, "loss": 0.17542500495910646, "step": 59800 }, { "epoch": 0.2970803572315548, "grad_norm": 2.102816343307495, "learning_rate": 1.6486482790459718e-05, "loss": 0.167777681350708, "step": 59810 }, { "epoch": 0.29713002791492404, "grad_norm": 0.7315004467964172, "learning_rate": 1.6485258333339067e-05, "loss": 0.17951469421386718, "step": 59820 }, { "epoch": 0.2971796985982933, "grad_norm": 0.8450497984886169, "learning_rate": 1.6484033708382538e-05, "loss": 0.17826683521270753, "step": 59830 }, { "epoch": 0.2972293692816626, "grad_norm": 1.246472716331482, "learning_rate": 1.6482808915621822e-05, "loss": 0.16428855657577515, "step": 59840 }, { "epoch": 0.29727903996503185, "grad_norm": 0.9043911695480347, "learning_rate": 1.6481583955088617e-05, "loss": 0.16425669193267822, "step": 59850 }, { "epoch": 0.2973287106484011, "grad_norm": 0.8939652442932129, "learning_rate": 1.6480358826814627e-05, "loss": 0.18256396055221558, "step": 59860 }, { "epoch": 0.29737838133177036, "grad_norm": 1.4292593002319336, "learning_rate": 1.647913353083156e-05, "loss": 0.17773224115371705, "step": 59870 }, { "epoch": 0.2974280520151396, "grad_norm": 1.5161205530166626, "learning_rate": 1.647790806717112e-05, "loss": 0.18023393154144288, "step": 59880 }, { "epoch": 0.29747772269850886, "grad_norm": 0.8475948572158813, "learning_rate": 1.647668243586502e-05, "loss": 0.16887078285217286, "step": 59890 }, { "epoch": 0.29752739338187817, "grad_norm": 1.2953495979309082, "learning_rate": 1.647545663694499e-05, "loss": 0.1851442813873291, "step": 59900 }, { "epoch": 0.2975770640652474, "grad_norm": 1.6575201749801636, "learning_rate": 1.647423067044274e-05, "loss": 0.16940410137176515, "step": 59910 }, { "epoch": 0.2976267347486167, "grad_norm": 0.9619349241256714, "learning_rate": 1.6473004536390008e-05, "loss": 0.16529273986816406, "step": 59920 }, { "epoch": 0.2976764054319859, "grad_norm": 0.7818090915679932, "learning_rate": 1.647177823481852e-05, "loss": 0.1642702579498291, "step": 59930 }, { "epoch": 0.2977260761153552, "grad_norm": 2.3094143867492676, "learning_rate": 1.647055176576001e-05, "loss": 0.1659029722213745, "step": 59940 }, { "epoch": 0.2977757467987245, "grad_norm": 1.1562341451644897, "learning_rate": 1.6469325129246223e-05, "loss": 0.16622424125671387, "step": 59950 }, { "epoch": 0.29782541748209373, "grad_norm": 1.3453000783920288, "learning_rate": 1.6468098325308902e-05, "loss": 0.1667580246925354, "step": 59960 }, { "epoch": 0.297875088165463, "grad_norm": 0.6754657030105591, "learning_rate": 1.6466871353979802e-05, "loss": 0.17285507917404175, "step": 59970 }, { "epoch": 0.29792475884883224, "grad_norm": 0.6747556924819946, "learning_rate": 1.6465644215290666e-05, "loss": 0.16861430406570435, "step": 59980 }, { "epoch": 0.2979744295322015, "grad_norm": 1.4428728818893433, "learning_rate": 1.646441690927326e-05, "loss": 0.167693829536438, "step": 59990 }, { "epoch": 0.29802410021557074, "grad_norm": 2.0329742431640625, "learning_rate": 1.6463189435959342e-05, "loss": 0.16269627809524537, "step": 60000 }, { "epoch": 0.29807377089894005, "grad_norm": 0.6750900149345398, "learning_rate": 1.6461961795380684e-05, "loss": 0.1846480131149292, "step": 60010 }, { "epoch": 0.2981234415823093, "grad_norm": 1.5163474082946777, "learning_rate": 1.6460733987569052e-05, "loss": 0.16136674880981444, "step": 60020 }, { "epoch": 0.29817311226567855, "grad_norm": 1.4278401136398315, "learning_rate": 1.645950601255622e-05, "loss": 0.17479488849639893, "step": 60030 }, { "epoch": 0.2982227829490478, "grad_norm": 0.7178955674171448, "learning_rate": 1.6458277870373972e-05, "loss": 0.1690973997116089, "step": 60040 }, { "epoch": 0.29827245363241706, "grad_norm": 0.9531404972076416, "learning_rate": 1.6457049561054085e-05, "loss": 0.1842888593673706, "step": 60050 }, { "epoch": 0.29832212431578636, "grad_norm": 1.6578220129013062, "learning_rate": 1.645582108462836e-05, "loss": 0.16663031578063964, "step": 60060 }, { "epoch": 0.2983717949991556, "grad_norm": 2.3680098056793213, "learning_rate": 1.6454592441128574e-05, "loss": 0.16684238910675048, "step": 60070 }, { "epoch": 0.29842146568252487, "grad_norm": 2.124659776687622, "learning_rate": 1.6453363630586535e-05, "loss": 0.1641631007194519, "step": 60080 }, { "epoch": 0.2984711363658941, "grad_norm": 0.873148500919342, "learning_rate": 1.6452134653034034e-05, "loss": 0.15754330158233643, "step": 60090 }, { "epoch": 0.29852080704926337, "grad_norm": 1.6646007299423218, "learning_rate": 1.6450905508502886e-05, "loss": 0.17597520351409912, "step": 60100 }, { "epoch": 0.2985704777326326, "grad_norm": 0.7411253452301025, "learning_rate": 1.64496761970249e-05, "loss": 0.17623207569122315, "step": 60110 }, { "epoch": 0.29862014841600193, "grad_norm": 0.8605940937995911, "learning_rate": 1.6448446718631883e-05, "loss": 0.17760387659072877, "step": 60120 }, { "epoch": 0.2986698190993712, "grad_norm": 1.164232611656189, "learning_rate": 1.644721707335566e-05, "loss": 0.17411720752716064, "step": 60130 }, { "epoch": 0.29871948978274043, "grad_norm": 1.005445122718811, "learning_rate": 1.644598726122805e-05, "loss": 0.1707903504371643, "step": 60140 }, { "epoch": 0.2987691604661097, "grad_norm": 1.392463207244873, "learning_rate": 1.6444757282280883e-05, "loss": 0.16865415573120118, "step": 60150 }, { "epoch": 0.29881883114947894, "grad_norm": 1.1122130155563354, "learning_rate": 1.6443527136545987e-05, "loss": 0.17352802753448487, "step": 60160 }, { "epoch": 0.29886850183284824, "grad_norm": 0.8612738251686096, "learning_rate": 1.6442296824055202e-05, "loss": 0.16093003749847412, "step": 60170 }, { "epoch": 0.2989181725162175, "grad_norm": 1.2501857280731201, "learning_rate": 1.6441066344840365e-05, "loss": 0.16583195924758912, "step": 60180 }, { "epoch": 0.29896784319958675, "grad_norm": 0.9782553911209106, "learning_rate": 1.643983569893332e-05, "loss": 0.18087774515151978, "step": 60190 }, { "epoch": 0.299017513882956, "grad_norm": 3.448664903640747, "learning_rate": 1.6438604886365915e-05, "loss": 0.1761146903038025, "step": 60200 }, { "epoch": 0.29906718456632525, "grad_norm": 1.159038782119751, "learning_rate": 1.6437373907170005e-05, "loss": 0.16299057006835938, "step": 60210 }, { "epoch": 0.2991168552496945, "grad_norm": 1.2722200155258179, "learning_rate": 1.643614276137745e-05, "loss": 0.1719723105430603, "step": 60220 }, { "epoch": 0.2991665259330638, "grad_norm": 1.0302982330322266, "learning_rate": 1.6434911449020105e-05, "loss": 0.16708844900131226, "step": 60230 }, { "epoch": 0.29921619661643306, "grad_norm": 2.4472126960754395, "learning_rate": 1.6433679970129844e-05, "loss": 0.1739404559135437, "step": 60240 }, { "epoch": 0.2992658672998023, "grad_norm": 0.7917999029159546, "learning_rate": 1.6432448324738527e-05, "loss": 0.16819028854370116, "step": 60250 }, { "epoch": 0.29931553798317156, "grad_norm": 1.3034337759017944, "learning_rate": 1.6431216512878033e-05, "loss": 0.16843051910400392, "step": 60260 }, { "epoch": 0.2993652086665408, "grad_norm": 1.2883200645446777, "learning_rate": 1.6429984534580247e-05, "loss": 0.16340290307998656, "step": 60270 }, { "epoch": 0.2994148793499101, "grad_norm": 1.8211394548416138, "learning_rate": 1.642875238987704e-05, "loss": 0.17841808795928954, "step": 60280 }, { "epoch": 0.2994645500332794, "grad_norm": 1.0405126810073853, "learning_rate": 1.6427520078800308e-05, "loss": 0.172041916847229, "step": 60290 }, { "epoch": 0.2995142207166486, "grad_norm": 0.8067581653594971, "learning_rate": 1.6426287601381944e-05, "loss": 0.1831273317337036, "step": 60300 }, { "epoch": 0.2995638914000179, "grad_norm": 0.9606837034225464, "learning_rate": 1.6425054957653838e-05, "loss": 0.18702261447906493, "step": 60310 }, { "epoch": 0.29961356208338713, "grad_norm": 2.0086164474487305, "learning_rate": 1.6423822147647896e-05, "loss": 0.17500274181365966, "step": 60320 }, { "epoch": 0.2996632327667564, "grad_norm": 1.98073410987854, "learning_rate": 1.6422589171396017e-05, "loss": 0.17275452613830566, "step": 60330 }, { "epoch": 0.2997129034501257, "grad_norm": 1.561119556427002, "learning_rate": 1.6421356028930114e-05, "loss": 0.16794391870498657, "step": 60340 }, { "epoch": 0.29976257413349494, "grad_norm": 1.0355428457260132, "learning_rate": 1.6420122720282096e-05, "loss": 0.19436274766921996, "step": 60350 }, { "epoch": 0.2998122448168642, "grad_norm": 0.8490691184997559, "learning_rate": 1.6418889245483883e-05, "loss": 0.18315656185150148, "step": 60360 }, { "epoch": 0.29986191550023344, "grad_norm": 1.4322128295898438, "learning_rate": 1.6417655604567402e-05, "loss": 0.15565783977508546, "step": 60370 }, { "epoch": 0.2999115861836027, "grad_norm": 0.8717592358589172, "learning_rate": 1.641642179756457e-05, "loss": 0.17458415031433105, "step": 60380 }, { "epoch": 0.29996125686697195, "grad_norm": 1.6020692586898804, "learning_rate": 1.641518782450733e-05, "loss": 0.16994383335113525, "step": 60390 }, { "epoch": 0.30001092755034126, "grad_norm": 0.5791435837745667, "learning_rate": 1.6413953685427603e-05, "loss": 0.16260929107666017, "step": 60400 }, { "epoch": 0.3000605982337105, "grad_norm": 2.1037776470184326, "learning_rate": 1.6412719380357328e-05, "loss": 0.1731326699256897, "step": 60410 }, { "epoch": 0.30011026891707976, "grad_norm": 1.0108935832977295, "learning_rate": 1.641148490932846e-05, "loss": 0.1552670955657959, "step": 60420 }, { "epoch": 0.300159939600449, "grad_norm": 1.2794945240020752, "learning_rate": 1.6410250272372942e-05, "loss": 0.18616068363189697, "step": 60430 }, { "epoch": 0.30020961028381826, "grad_norm": 0.6659987568855286, "learning_rate": 1.640901546952272e-05, "loss": 0.1636103868484497, "step": 60440 }, { "epoch": 0.30025928096718757, "grad_norm": 1.4896788597106934, "learning_rate": 1.6407780500809756e-05, "loss": 0.17516417503356935, "step": 60450 }, { "epoch": 0.3003089516505568, "grad_norm": 2.3439340591430664, "learning_rate": 1.640654536626601e-05, "loss": 0.17054864168167114, "step": 60460 }, { "epoch": 0.3003586223339261, "grad_norm": 0.9363262057304382, "learning_rate": 1.6405310065923447e-05, "loss": 0.1681732416152954, "step": 60470 }, { "epoch": 0.3004082930172953, "grad_norm": 1.254144549369812, "learning_rate": 1.6404074599814032e-05, "loss": 0.17642483711242676, "step": 60480 }, { "epoch": 0.3004579637006646, "grad_norm": 0.6234614253044128, "learning_rate": 1.6402838967969742e-05, "loss": 0.17156518697738649, "step": 60490 }, { "epoch": 0.30050763438403383, "grad_norm": 1.1725809574127197, "learning_rate": 1.6401603170422555e-05, "loss": 0.18394442796707153, "step": 60500 }, { "epoch": 0.30055730506740314, "grad_norm": 0.7867509722709656, "learning_rate": 1.640036720720445e-05, "loss": 0.16017683744430541, "step": 60510 }, { "epoch": 0.3006069757507724, "grad_norm": 2.555642604827881, "learning_rate": 1.639913107834742e-05, "loss": 0.1631207823753357, "step": 60520 }, { "epoch": 0.30065664643414164, "grad_norm": 1.3558095693588257, "learning_rate": 1.6397894783883447e-05, "loss": 0.18031187057495118, "step": 60530 }, { "epoch": 0.3007063171175109, "grad_norm": 0.7142014503479004, "learning_rate": 1.639665832384453e-05, "loss": 0.1823631763458252, "step": 60540 }, { "epoch": 0.30075598780088014, "grad_norm": 1.3617678880691528, "learning_rate": 1.6395421698262665e-05, "loss": 0.17866308689117433, "step": 60550 }, { "epoch": 0.30080565848424945, "grad_norm": 1.4523752927780151, "learning_rate": 1.6394184907169863e-05, "loss": 0.1781502366065979, "step": 60560 }, { "epoch": 0.3008553291676187, "grad_norm": 1.0449808835983276, "learning_rate": 1.6392947950598123e-05, "loss": 0.16374523639678956, "step": 60570 }, { "epoch": 0.30090499985098795, "grad_norm": 2.517214298248291, "learning_rate": 1.6391710828579462e-05, "loss": 0.17701282501220703, "step": 60580 }, { "epoch": 0.3009546705343572, "grad_norm": 1.7223330736160278, "learning_rate": 1.6390473541145895e-05, "loss": 0.16996195316314697, "step": 60590 }, { "epoch": 0.30100434121772646, "grad_norm": 0.9622312784194946, "learning_rate": 1.6389236088329443e-05, "loss": 0.16070469617843627, "step": 60600 }, { "epoch": 0.3010540119010957, "grad_norm": 1.1740453243255615, "learning_rate": 1.6387998470162128e-05, "loss": 0.1618627667427063, "step": 60610 }, { "epoch": 0.301103682584465, "grad_norm": 1.467687726020813, "learning_rate": 1.638676068667598e-05, "loss": 0.16778078079223632, "step": 60620 }, { "epoch": 0.30115335326783427, "grad_norm": 1.3758188486099243, "learning_rate": 1.6385522737903036e-05, "loss": 0.1863969087600708, "step": 60630 }, { "epoch": 0.3012030239512035, "grad_norm": 0.8032158613204956, "learning_rate": 1.6384284623875332e-05, "loss": 0.16456496715545654, "step": 60640 }, { "epoch": 0.30125269463457277, "grad_norm": 1.9478647708892822, "learning_rate": 1.6383046344624907e-05, "loss": 0.16281468868255616, "step": 60650 }, { "epoch": 0.301302365317942, "grad_norm": 1.4610127210617065, "learning_rate": 1.638180790018381e-05, "loss": 0.17079579830169678, "step": 60660 }, { "epoch": 0.30135203600131133, "grad_norm": 2.6040050983428955, "learning_rate": 1.6380569290584087e-05, "loss": 0.1899965524673462, "step": 60670 }, { "epoch": 0.3014017066846806, "grad_norm": 2.1067402362823486, "learning_rate": 1.63793305158578e-05, "loss": 0.16292598247528076, "step": 60680 }, { "epoch": 0.30145137736804983, "grad_norm": 0.927477240562439, "learning_rate": 1.6378091576037004e-05, "loss": 0.17408050298690797, "step": 60690 }, { "epoch": 0.3015010480514191, "grad_norm": 1.6931877136230469, "learning_rate": 1.637685247115376e-05, "loss": 0.16915223598480225, "step": 60700 }, { "epoch": 0.30155071873478834, "grad_norm": 1.9552425146102905, "learning_rate": 1.6375613201240143e-05, "loss": 0.1722196578979492, "step": 60710 }, { "epoch": 0.3016003894181576, "grad_norm": 1.778409481048584, "learning_rate": 1.6374373766328217e-05, "loss": 0.17265806198120118, "step": 60720 }, { "epoch": 0.3016500601015269, "grad_norm": 1.1020822525024414, "learning_rate": 1.637313416645006e-05, "loss": 0.16986401081085206, "step": 60730 }, { "epoch": 0.30169973078489615, "grad_norm": 2.132417678833008, "learning_rate": 1.6371894401637757e-05, "loss": 0.16250357627868653, "step": 60740 }, { "epoch": 0.3017494014682654, "grad_norm": 2.0649912357330322, "learning_rate": 1.6370654471923387e-05, "loss": 0.1766197681427002, "step": 60750 }, { "epoch": 0.30179907215163465, "grad_norm": 0.8311225175857544, "learning_rate": 1.6369414377339037e-05, "loss": 0.1616879940032959, "step": 60760 }, { "epoch": 0.3018487428350039, "grad_norm": 1.6231926679611206, "learning_rate": 1.6368174117916808e-05, "loss": 0.17746390104293824, "step": 60770 }, { "epoch": 0.3018984135183732, "grad_norm": 1.2457000017166138, "learning_rate": 1.6366933693688794e-05, "loss": 0.16877381801605223, "step": 60780 }, { "epoch": 0.30194808420174246, "grad_norm": 0.8108894228935242, "learning_rate": 1.636569310468709e-05, "loss": 0.18087714910507202, "step": 60790 }, { "epoch": 0.3019977548851117, "grad_norm": 1.0696338415145874, "learning_rate": 1.6364452350943813e-05, "loss": 0.16740859746932985, "step": 60800 }, { "epoch": 0.30204742556848097, "grad_norm": 1.2080323696136475, "learning_rate": 1.636321143249107e-05, "loss": 0.1727271556854248, "step": 60810 }, { "epoch": 0.3020970962518502, "grad_norm": 1.104111909866333, "learning_rate": 1.636197034936097e-05, "loss": 0.16301450729370118, "step": 60820 }, { "epoch": 0.30214676693521947, "grad_norm": 0.845012366771698, "learning_rate": 1.6360729101585637e-05, "loss": 0.16838352680206298, "step": 60830 }, { "epoch": 0.3021964376185888, "grad_norm": 1.3323949575424194, "learning_rate": 1.635948768919719e-05, "loss": 0.1779366612434387, "step": 60840 }, { "epoch": 0.302246108301958, "grad_norm": 2.3675599098205566, "learning_rate": 1.6358246112227764e-05, "loss": 0.16272168159484862, "step": 60850 }, { "epoch": 0.3022957789853273, "grad_norm": 0.9438437819480896, "learning_rate": 1.635700437070948e-05, "loss": 0.19047685861587524, "step": 60860 }, { "epoch": 0.30234544966869653, "grad_norm": 1.3110477924346924, "learning_rate": 1.635576246467448e-05, "loss": 0.16888389587402344, "step": 60870 }, { "epoch": 0.3023951203520658, "grad_norm": 1.2034944295883179, "learning_rate": 1.6354520394154905e-05, "loss": 0.16133333444595338, "step": 60880 }, { "epoch": 0.3024447910354351, "grad_norm": 2.0717573165893555, "learning_rate": 1.6353278159182896e-05, "loss": 0.1889817714691162, "step": 60890 }, { "epoch": 0.30249446171880434, "grad_norm": 1.209302544593811, "learning_rate": 1.63520357597906e-05, "loss": 0.1761444091796875, "step": 60900 }, { "epoch": 0.3025441324021736, "grad_norm": 1.5516260862350464, "learning_rate": 1.635079319601018e-05, "loss": 0.17716660499572753, "step": 60910 }, { "epoch": 0.30259380308554285, "grad_norm": 2.47037410736084, "learning_rate": 1.634955046787378e-05, "loss": 0.18052802085876465, "step": 60920 }, { "epoch": 0.3026434737689121, "grad_norm": 0.9873165488243103, "learning_rate": 1.6348307575413567e-05, "loss": 0.17516086101531983, "step": 60930 }, { "epoch": 0.30269314445228135, "grad_norm": 1.9286189079284668, "learning_rate": 1.634706451866171e-05, "loss": 0.16627061367034912, "step": 60940 }, { "epoch": 0.30274281513565066, "grad_norm": 0.9651185274124146, "learning_rate": 1.6345821297650372e-05, "loss": 0.16908979415893555, "step": 60950 }, { "epoch": 0.3027924858190199, "grad_norm": 1.3876937627792358, "learning_rate": 1.6344577912411727e-05, "loss": 0.17120132446289063, "step": 60960 }, { "epoch": 0.30284215650238916, "grad_norm": 1.2610363960266113, "learning_rate": 1.634333436297796e-05, "loss": 0.17678241729736327, "step": 60970 }, { "epoch": 0.3028918271857584, "grad_norm": 0.760261058807373, "learning_rate": 1.634209064938125e-05, "loss": 0.17221111059188843, "step": 60980 }, { "epoch": 0.30294149786912766, "grad_norm": 1.1235312223434448, "learning_rate": 1.6340846771653788e-05, "loss": 0.180245304107666, "step": 60990 }, { "epoch": 0.30299116855249697, "grad_norm": 2.1476023197174072, "learning_rate": 1.6339602729827756e-05, "loss": 0.18568545579910278, "step": 61000 }, { "epoch": 0.3030408392358662, "grad_norm": 1.6192978620529175, "learning_rate": 1.6338358523935358e-05, "loss": 0.17356187105178833, "step": 61010 }, { "epoch": 0.3030905099192355, "grad_norm": 1.288127064704895, "learning_rate": 1.633711415400879e-05, "loss": 0.16255048513412476, "step": 61020 }, { "epoch": 0.3031401806026047, "grad_norm": 0.9634940028190613, "learning_rate": 1.633586962008025e-05, "loss": 0.16792404651641846, "step": 61030 }, { "epoch": 0.303189851285974, "grad_norm": 0.8920181393623352, "learning_rate": 1.6334624922181958e-05, "loss": 0.1606973648071289, "step": 61040 }, { "epoch": 0.30323952196934323, "grad_norm": 0.8232312798500061, "learning_rate": 1.6333380060346117e-05, "loss": 0.16980574131011963, "step": 61050 }, { "epoch": 0.30328919265271254, "grad_norm": 0.6451213359832764, "learning_rate": 1.6332135034604946e-05, "loss": 0.17326072454452515, "step": 61060 }, { "epoch": 0.3033388633360818, "grad_norm": 1.440589427947998, "learning_rate": 1.6330889844990673e-05, "loss": 0.18036468029022218, "step": 61070 }, { "epoch": 0.30338853401945104, "grad_norm": 1.1150074005126953, "learning_rate": 1.6329644491535514e-05, "loss": 0.18125404119491578, "step": 61080 }, { "epoch": 0.3034382047028203, "grad_norm": 0.8898089528083801, "learning_rate": 1.6328398974271694e-05, "loss": 0.18278319835662843, "step": 61090 }, { "epoch": 0.30348787538618954, "grad_norm": 1.1473714113235474, "learning_rate": 1.6327153293231456e-05, "loss": 0.16771309375762938, "step": 61100 }, { "epoch": 0.30353754606955885, "grad_norm": 1.4704053401947021, "learning_rate": 1.6325907448447037e-05, "loss": 0.18004562854766845, "step": 61110 }, { "epoch": 0.3035872167529281, "grad_norm": 2.1282339096069336, "learning_rate": 1.6324661439950674e-05, "loss": 0.16332184076309203, "step": 61120 }, { "epoch": 0.30363688743629735, "grad_norm": 1.4563908576965332, "learning_rate": 1.632341526777462e-05, "loss": 0.17924630641937256, "step": 61130 }, { "epoch": 0.3036865581196666, "grad_norm": 0.8295780420303345, "learning_rate": 1.6322168931951117e-05, "loss": 0.16191062927246094, "step": 61140 }, { "epoch": 0.30373622880303586, "grad_norm": 0.7504797577857971, "learning_rate": 1.6320922432512426e-05, "loss": 0.165809166431427, "step": 61150 }, { "epoch": 0.3037858994864051, "grad_norm": 0.7254090309143066, "learning_rate": 1.6319675769490805e-05, "loss": 0.15959885120391845, "step": 61160 }, { "epoch": 0.3038355701697744, "grad_norm": 1.0481337308883667, "learning_rate": 1.6318428942918516e-05, "loss": 0.16879037618637086, "step": 61170 }, { "epoch": 0.30388524085314367, "grad_norm": 1.2514716386795044, "learning_rate": 1.6317181952827825e-05, "loss": 0.16739832162857055, "step": 61180 }, { "epoch": 0.3039349115365129, "grad_norm": 1.0143481492996216, "learning_rate": 1.6315934799251006e-05, "loss": 0.1742410898208618, "step": 61190 }, { "epoch": 0.30398458221988217, "grad_norm": 1.859026312828064, "learning_rate": 1.6314687482220332e-05, "loss": 0.16415200233459473, "step": 61200 }, { "epoch": 0.3040342529032514, "grad_norm": 2.384157180786133, "learning_rate": 1.6313440001768088e-05, "loss": 0.1821521997451782, "step": 61210 }, { "epoch": 0.30408392358662073, "grad_norm": 1.3853265047073364, "learning_rate": 1.6312192357926556e-05, "loss": 0.1636345148086548, "step": 61220 }, { "epoch": 0.30413359426999, "grad_norm": 1.0434907674789429, "learning_rate": 1.631094455072802e-05, "loss": 0.16786162853240966, "step": 61230 }, { "epoch": 0.30418326495335923, "grad_norm": 1.3026694059371948, "learning_rate": 1.630969658020478e-05, "loss": 0.17500474452972412, "step": 61240 }, { "epoch": 0.3042329356367285, "grad_norm": 1.3452093601226807, "learning_rate": 1.6308448446389127e-05, "loss": 0.15521326065063476, "step": 61250 }, { "epoch": 0.30428260632009774, "grad_norm": 1.874751091003418, "learning_rate": 1.6307200149313364e-05, "loss": 0.15853769779205323, "step": 61260 }, { "epoch": 0.304332277003467, "grad_norm": 1.2469820976257324, "learning_rate": 1.63059516890098e-05, "loss": 0.17208685874938964, "step": 61270 }, { "epoch": 0.3043819476868363, "grad_norm": 0.9264102578163147, "learning_rate": 1.630470306551074e-05, "loss": 0.1731812834739685, "step": 61280 }, { "epoch": 0.30443161837020555, "grad_norm": 1.6757185459136963, "learning_rate": 1.63034542788485e-05, "loss": 0.1630859851837158, "step": 61290 }, { "epoch": 0.3044812890535748, "grad_norm": 0.7599923014640808, "learning_rate": 1.6302205329055398e-05, "loss": 0.184361732006073, "step": 61300 }, { "epoch": 0.30453095973694405, "grad_norm": 1.657616138458252, "learning_rate": 1.630095621616376e-05, "loss": 0.16677682399749755, "step": 61310 }, { "epoch": 0.3045806304203133, "grad_norm": 0.6093842387199402, "learning_rate": 1.6299706940205903e-05, "loss": 0.17017624378204346, "step": 61320 }, { "epoch": 0.3046303011036826, "grad_norm": 0.8163010478019714, "learning_rate": 1.629845750121416e-05, "loss": 0.16697431802749635, "step": 61330 }, { "epoch": 0.30467997178705186, "grad_norm": 1.5008617639541626, "learning_rate": 1.6297207899220876e-05, "loss": 0.1738840937614441, "step": 61340 }, { "epoch": 0.3047296424704211, "grad_norm": 1.3805402517318726, "learning_rate": 1.6295958134258384e-05, "loss": 0.17954509258270263, "step": 61350 }, { "epoch": 0.30477931315379037, "grad_norm": 0.977345883846283, "learning_rate": 1.6294708206359025e-05, "loss": 0.17214794158935548, "step": 61360 }, { "epoch": 0.3048289838371596, "grad_norm": 0.7367059588432312, "learning_rate": 1.6293458115555146e-05, "loss": 0.17323161363601686, "step": 61370 }, { "epoch": 0.30487865452052887, "grad_norm": 0.973973274230957, "learning_rate": 1.6292207861879106e-05, "loss": 0.18324685096740723, "step": 61380 }, { "epoch": 0.3049283252038982, "grad_norm": 1.0090852975845337, "learning_rate": 1.6290957445363255e-05, "loss": 0.1696143388748169, "step": 61390 }, { "epoch": 0.30497799588726743, "grad_norm": 1.5009983777999878, "learning_rate": 1.6289706866039954e-05, "loss": 0.16940473318099974, "step": 61400 }, { "epoch": 0.3050276665706367, "grad_norm": 1.1683498620986938, "learning_rate": 1.628845612394157e-05, "loss": 0.16862263679504394, "step": 61410 }, { "epoch": 0.30507733725400593, "grad_norm": 0.7310979962348938, "learning_rate": 1.6287205219100467e-05, "loss": 0.1719442367553711, "step": 61420 }, { "epoch": 0.3051270079373752, "grad_norm": 1.601101279258728, "learning_rate": 1.6285954151549025e-05, "loss": 0.18592236042022706, "step": 61430 }, { "epoch": 0.3051766786207445, "grad_norm": 1.01914644241333, "learning_rate": 1.6284702921319616e-05, "loss": 0.17548491954803466, "step": 61440 }, { "epoch": 0.30522634930411374, "grad_norm": 1.331972599029541, "learning_rate": 1.628345152844462e-05, "loss": 0.16980714797973634, "step": 61450 }, { "epoch": 0.305276019987483, "grad_norm": 2.2940728664398193, "learning_rate": 1.6282199972956425e-05, "loss": 0.1794853091239929, "step": 61460 }, { "epoch": 0.30532569067085225, "grad_norm": 1.3256553411483765, "learning_rate": 1.628094825488742e-05, "loss": 0.17782416343688964, "step": 61470 }, { "epoch": 0.3053753613542215, "grad_norm": 1.544443964958191, "learning_rate": 1.6279696374270004e-05, "loss": 0.16469618082046508, "step": 61480 }, { "epoch": 0.30542503203759075, "grad_norm": 1.5197077989578247, "learning_rate": 1.6278444331136567e-05, "loss": 0.17000010013580322, "step": 61490 }, { "epoch": 0.30547470272096006, "grad_norm": 1.084526777267456, "learning_rate": 1.6277192125519517e-05, "loss": 0.17092180252075195, "step": 61500 }, { "epoch": 0.3055243734043293, "grad_norm": 1.4143329858779907, "learning_rate": 1.627593975745126e-05, "loss": 0.19185256958007812, "step": 61510 }, { "epoch": 0.30557404408769856, "grad_norm": 1.362013578414917, "learning_rate": 1.6274687226964202e-05, "loss": 0.17533267736434938, "step": 61520 }, { "epoch": 0.3056237147710678, "grad_norm": 0.8808651566505432, "learning_rate": 1.6273434534090762e-05, "loss": 0.1621704578399658, "step": 61530 }, { "epoch": 0.30567338545443706, "grad_norm": 1.2762166261672974, "learning_rate": 1.627218167886336e-05, "loss": 0.16388936042785646, "step": 61540 }, { "epoch": 0.30572305613780637, "grad_norm": 1.1586991548538208, "learning_rate": 1.6270928661314414e-05, "loss": 0.1688992977142334, "step": 61550 }, { "epoch": 0.3057727268211756, "grad_norm": 1.192765712738037, "learning_rate": 1.6269675481476363e-05, "loss": 0.1681300640106201, "step": 61560 }, { "epoch": 0.3058223975045449, "grad_norm": 1.1358768939971924, "learning_rate": 1.6268422139381628e-05, "loss": 0.17018401622772217, "step": 61570 }, { "epoch": 0.3058720681879141, "grad_norm": 0.7516933679580688, "learning_rate": 1.6267168635062645e-05, "loss": 0.1603764772415161, "step": 61580 }, { "epoch": 0.3059217388712834, "grad_norm": 1.432914137840271, "learning_rate": 1.6265914968551863e-05, "loss": 0.1609020233154297, "step": 61590 }, { "epoch": 0.30597140955465263, "grad_norm": 1.7437766790390015, "learning_rate": 1.626466113988172e-05, "loss": 0.1930919408798218, "step": 61600 }, { "epoch": 0.30602108023802194, "grad_norm": 1.2340360879898071, "learning_rate": 1.6263407149084666e-05, "loss": 0.1721426248550415, "step": 61610 }, { "epoch": 0.3060707509213912, "grad_norm": 1.0260725021362305, "learning_rate": 1.6262152996193152e-05, "loss": 0.1911190152168274, "step": 61620 }, { "epoch": 0.30612042160476044, "grad_norm": 0.7166907787322998, "learning_rate": 1.6260898681239634e-05, "loss": 0.16059211492538453, "step": 61630 }, { "epoch": 0.3061700922881297, "grad_norm": 0.7846518158912659, "learning_rate": 1.625964420425658e-05, "loss": 0.1668774127960205, "step": 61640 }, { "epoch": 0.30621976297149894, "grad_norm": 1.1345479488372803, "learning_rate": 1.625838956527645e-05, "loss": 0.17728511095046998, "step": 61650 }, { "epoch": 0.30626943365486825, "grad_norm": 1.419558048248291, "learning_rate": 1.6257134764331713e-05, "loss": 0.1667366147041321, "step": 61660 }, { "epoch": 0.3063191043382375, "grad_norm": 1.2198762893676758, "learning_rate": 1.6255879801454844e-05, "loss": 0.17921912670135498, "step": 61670 }, { "epoch": 0.30636877502160675, "grad_norm": 0.908272922039032, "learning_rate": 1.6254624676678322e-05, "loss": 0.16584303379058837, "step": 61680 }, { "epoch": 0.306418445704976, "grad_norm": 1.4054858684539795, "learning_rate": 1.6253369390034624e-05, "loss": 0.1787710666656494, "step": 61690 }, { "epoch": 0.30646811638834526, "grad_norm": 0.6133210062980652, "learning_rate": 1.625211394155625e-05, "loss": 0.1684882164001465, "step": 61700 }, { "epoch": 0.3065177870717145, "grad_norm": 1.0377497673034668, "learning_rate": 1.6250858331275676e-05, "loss": 0.16432491540908814, "step": 61710 }, { "epoch": 0.3065674577550838, "grad_norm": 0.8457319140434265, "learning_rate": 1.62496025592254e-05, "loss": 0.15285273790359497, "step": 61720 }, { "epoch": 0.30661712843845307, "grad_norm": 1.6914087533950806, "learning_rate": 1.6248346625437924e-05, "loss": 0.1583171844482422, "step": 61730 }, { "epoch": 0.3066667991218223, "grad_norm": 1.3668162822723389, "learning_rate": 1.624709052994575e-05, "loss": 0.1715508818626404, "step": 61740 }, { "epoch": 0.30671646980519157, "grad_norm": 1.2238088846206665, "learning_rate": 1.6245834272781383e-05, "loss": 0.16096065044403077, "step": 61750 }, { "epoch": 0.3067661404885608, "grad_norm": 1.717444896697998, "learning_rate": 1.6244577853977337e-05, "loss": 0.17837623357772828, "step": 61760 }, { "epoch": 0.30681581117193013, "grad_norm": 0.743750274181366, "learning_rate": 1.6243321273566128e-05, "loss": 0.16527237892150878, "step": 61770 }, { "epoch": 0.3068654818552994, "grad_norm": 1.3468469381332397, "learning_rate": 1.6242064531580276e-05, "loss": 0.1663733720779419, "step": 61780 }, { "epoch": 0.30691515253866863, "grad_norm": 1.3964555263519287, "learning_rate": 1.6240807628052302e-05, "loss": 0.18168349266052247, "step": 61790 }, { "epoch": 0.3069648232220379, "grad_norm": 0.927828311920166, "learning_rate": 1.6239550563014736e-05, "loss": 0.15648421049118041, "step": 61800 }, { "epoch": 0.30701449390540714, "grad_norm": 1.570577621459961, "learning_rate": 1.623829333650011e-05, "loss": 0.16849212646484374, "step": 61810 }, { "epoch": 0.3070641645887764, "grad_norm": 1.233046054840088, "learning_rate": 1.6237035948540962e-05, "loss": 0.1790909767150879, "step": 61820 }, { "epoch": 0.3071138352721457, "grad_norm": 1.0001678466796875, "learning_rate": 1.6235778399169834e-05, "loss": 0.1768466591835022, "step": 61830 }, { "epoch": 0.30716350595551495, "grad_norm": 0.7561017870903015, "learning_rate": 1.623452068841926e-05, "loss": 0.16177639961242676, "step": 61840 }, { "epoch": 0.3072131766388842, "grad_norm": 2.3480136394500732, "learning_rate": 1.6233262816321805e-05, "loss": 0.1720563769340515, "step": 61850 }, { "epoch": 0.30726284732225345, "grad_norm": 1.6568636894226074, "learning_rate": 1.6232004782910013e-05, "loss": 0.17207416296005248, "step": 61860 }, { "epoch": 0.3073125180056227, "grad_norm": 1.4042547941207886, "learning_rate": 1.623074658821644e-05, "loss": 0.1806645393371582, "step": 61870 }, { "epoch": 0.307362188688992, "grad_norm": 0.7819239497184753, "learning_rate": 1.622948823227365e-05, "loss": 0.16574409008026122, "step": 61880 }, { "epoch": 0.30741185937236126, "grad_norm": 0.7886955738067627, "learning_rate": 1.6228229715114214e-05, "loss": 0.1712036371231079, "step": 61890 }, { "epoch": 0.3074615300557305, "grad_norm": 0.7103290557861328, "learning_rate": 1.62269710367707e-05, "loss": 0.1656942844390869, "step": 61900 }, { "epoch": 0.30751120073909977, "grad_norm": 1.2611517906188965, "learning_rate": 1.622571219727567e-05, "loss": 0.17783732414245607, "step": 61910 }, { "epoch": 0.307560871422469, "grad_norm": 1.1828886270523071, "learning_rate": 1.622445319666172e-05, "loss": 0.16289095878601073, "step": 61920 }, { "epoch": 0.30761054210583827, "grad_norm": 2.1702237129211426, "learning_rate": 1.6223194034961418e-05, "loss": 0.16655865907669068, "step": 61930 }, { "epoch": 0.3076602127892076, "grad_norm": 1.7153810262680054, "learning_rate": 1.622193471220736e-05, "loss": 0.17725818157196044, "step": 61940 }, { "epoch": 0.30770988347257683, "grad_norm": 2.63193941116333, "learning_rate": 1.622067522843213e-05, "loss": 0.1612270474433899, "step": 61950 }, { "epoch": 0.3077595541559461, "grad_norm": 1.187412977218628, "learning_rate": 1.621941558366833e-05, "loss": 0.17127366065979005, "step": 61960 }, { "epoch": 0.30780922483931533, "grad_norm": 0.7940352559089661, "learning_rate": 1.6218155777948554e-05, "loss": 0.15535662174224854, "step": 61970 }, { "epoch": 0.3078588955226846, "grad_norm": 0.5519760251045227, "learning_rate": 1.6216895811305406e-05, "loss": 0.19210015535354613, "step": 61980 }, { "epoch": 0.30790856620605384, "grad_norm": 0.7818849682807922, "learning_rate": 1.6215635683771497e-05, "loss": 0.17805255651474, "step": 61990 }, { "epoch": 0.30795823688942314, "grad_norm": 1.457290530204773, "learning_rate": 1.6214375395379432e-05, "loss": 0.1832338333129883, "step": 62000 }, { "epoch": 0.3080079075727924, "grad_norm": 1.0502997636795044, "learning_rate": 1.621311494616183e-05, "loss": 0.1578348159790039, "step": 62010 }, { "epoch": 0.30805757825616165, "grad_norm": 1.3237513303756714, "learning_rate": 1.6211854336151317e-05, "loss": 0.1628049373626709, "step": 62020 }, { "epoch": 0.3081072489395309, "grad_norm": 1.4728244543075562, "learning_rate": 1.6210593565380507e-05, "loss": 0.16334681510925292, "step": 62030 }, { "epoch": 0.30815691962290015, "grad_norm": 1.230915904045105, "learning_rate": 1.620933263388203e-05, "loss": 0.1738392949104309, "step": 62040 }, { "epoch": 0.30820659030626946, "grad_norm": 1.1830791234970093, "learning_rate": 1.6208071541688526e-05, "loss": 0.1575117826461792, "step": 62050 }, { "epoch": 0.3082562609896387, "grad_norm": 1.3956776857376099, "learning_rate": 1.6206810288832624e-05, "loss": 0.16514043807983397, "step": 62060 }, { "epoch": 0.30830593167300796, "grad_norm": 1.0230114459991455, "learning_rate": 1.620554887534697e-05, "loss": 0.16726064682006836, "step": 62070 }, { "epoch": 0.3083556023563772, "grad_norm": 1.2787423133850098, "learning_rate": 1.6204287301264203e-05, "loss": 0.1659359574317932, "step": 62080 }, { "epoch": 0.30840527303974646, "grad_norm": 1.9096144437789917, "learning_rate": 1.620302556661698e-05, "loss": 0.17553738355636597, "step": 62090 }, { "epoch": 0.3084549437231157, "grad_norm": 0.8330322504043579, "learning_rate": 1.6201763671437945e-05, "loss": 0.1697315216064453, "step": 62100 }, { "epoch": 0.308504614406485, "grad_norm": 1.262605905532837, "learning_rate": 1.6200501615759764e-05, "loss": 0.18543585538864135, "step": 62110 }, { "epoch": 0.3085542850898543, "grad_norm": 1.5852572917938232, "learning_rate": 1.6199239399615093e-05, "loss": 0.16150214672088622, "step": 62120 }, { "epoch": 0.3086039557732235, "grad_norm": 1.2764568328857422, "learning_rate": 1.6197977023036596e-05, "loss": 0.1622401714324951, "step": 62130 }, { "epoch": 0.3086536264565928, "grad_norm": 2.0808331966400146, "learning_rate": 1.6196714486056947e-05, "loss": 0.18983596563339233, "step": 62140 }, { "epoch": 0.30870329713996203, "grad_norm": 2.501516342163086, "learning_rate": 1.6195451788708817e-05, "loss": 0.17678570747375488, "step": 62150 }, { "epoch": 0.30875296782333134, "grad_norm": 1.186908483505249, "learning_rate": 1.619418893102489e-05, "loss": 0.17644970417022704, "step": 62160 }, { "epoch": 0.3088026385067006, "grad_norm": 1.1699585914611816, "learning_rate": 1.619292591303784e-05, "loss": 0.1651772141456604, "step": 62170 }, { "epoch": 0.30885230919006984, "grad_norm": 1.9277478456497192, "learning_rate": 1.619166273478036e-05, "loss": 0.17786798477172852, "step": 62180 }, { "epoch": 0.3089019798734391, "grad_norm": 1.1701465845108032, "learning_rate": 1.6190399396285138e-05, "loss": 0.1825575351715088, "step": 62190 }, { "epoch": 0.30895165055680834, "grad_norm": 1.0768548250198364, "learning_rate": 1.6189135897584866e-05, "loss": 0.1757098913192749, "step": 62200 }, { "epoch": 0.3090013212401776, "grad_norm": 1.507041573524475, "learning_rate": 1.618787223871225e-05, "loss": 0.17617162466049194, "step": 62210 }, { "epoch": 0.3090509919235469, "grad_norm": 1.0343953371047974, "learning_rate": 1.6186608419699984e-05, "loss": 0.16209487915039061, "step": 62220 }, { "epoch": 0.30910066260691615, "grad_norm": 0.8997039198875427, "learning_rate": 1.6185344440580785e-05, "loss": 0.16657500267028807, "step": 62230 }, { "epoch": 0.3091503332902854, "grad_norm": 0.8188945055007935, "learning_rate": 1.6184080301387352e-05, "loss": 0.1725115180015564, "step": 62240 }, { "epoch": 0.30920000397365466, "grad_norm": 1.939687728881836, "learning_rate": 1.6182816002152413e-05, "loss": 0.1807248830795288, "step": 62250 }, { "epoch": 0.3092496746570239, "grad_norm": 1.4908839464187622, "learning_rate": 1.618155154290868e-05, "loss": 0.17767354249954223, "step": 62260 }, { "epoch": 0.3092993453403932, "grad_norm": 0.994386613368988, "learning_rate": 1.618028692368888e-05, "loss": 0.18270959854125976, "step": 62270 }, { "epoch": 0.30934901602376247, "grad_norm": 0.8946754336357117, "learning_rate": 1.6179022144525734e-05, "loss": 0.17038342952728272, "step": 62280 }, { "epoch": 0.3093986867071317, "grad_norm": 1.1596375703811646, "learning_rate": 1.6177757205451987e-05, "loss": 0.17866799831390381, "step": 62290 }, { "epoch": 0.30944835739050097, "grad_norm": 1.1710078716278076, "learning_rate": 1.6176492106500364e-05, "loss": 0.16881582736968995, "step": 62300 }, { "epoch": 0.3094980280738702, "grad_norm": 0.987777054309845, "learning_rate": 1.6175226847703606e-05, "loss": 0.1748831033706665, "step": 62310 }, { "epoch": 0.3095476987572395, "grad_norm": 1.6787360906600952, "learning_rate": 1.6173961429094465e-05, "loss": 0.17294405698776244, "step": 62320 }, { "epoch": 0.3095973694406088, "grad_norm": 1.4885528087615967, "learning_rate": 1.617269585070568e-05, "loss": 0.17864763736724854, "step": 62330 }, { "epoch": 0.30964704012397803, "grad_norm": 1.1021931171417236, "learning_rate": 1.617143011257001e-05, "loss": 0.17753525972366332, "step": 62340 }, { "epoch": 0.3096967108073473, "grad_norm": 1.7042874097824097, "learning_rate": 1.617016421472021e-05, "loss": 0.18170799016952516, "step": 62350 }, { "epoch": 0.30974638149071654, "grad_norm": 1.7382553815841675, "learning_rate": 1.6168898157189042e-05, "loss": 0.17956233024597168, "step": 62360 }, { "epoch": 0.3097960521740858, "grad_norm": 1.6460490226745605, "learning_rate": 1.6167631940009273e-05, "loss": 0.16960400342941284, "step": 62370 }, { "epoch": 0.3098457228574551, "grad_norm": 0.7039844393730164, "learning_rate": 1.6166365563213663e-05, "loss": 0.16311193704605104, "step": 62380 }, { "epoch": 0.30989539354082435, "grad_norm": 0.8420670628547668, "learning_rate": 1.6165099026834992e-05, "loss": 0.19046508073806762, "step": 62390 }, { "epoch": 0.3099450642241936, "grad_norm": 1.1307802200317383, "learning_rate": 1.6163832330906042e-05, "loss": 0.1816786289215088, "step": 62400 }, { "epoch": 0.30999473490756285, "grad_norm": 1.1760313510894775, "learning_rate": 1.6162565475459588e-05, "loss": 0.18274972438812256, "step": 62410 }, { "epoch": 0.3100444055909321, "grad_norm": 1.7396053075790405, "learning_rate": 1.6161298460528417e-05, "loss": 0.16017708778381348, "step": 62420 }, { "epoch": 0.31009407627430136, "grad_norm": 1.8139996528625488, "learning_rate": 1.616003128614532e-05, "loss": 0.16700661182403564, "step": 62430 }, { "epoch": 0.31014374695767066, "grad_norm": 1.2794100046157837, "learning_rate": 1.615876395234309e-05, "loss": 0.16125710010528566, "step": 62440 }, { "epoch": 0.3101934176410399, "grad_norm": 0.9552228450775146, "learning_rate": 1.6157496459154526e-05, "loss": 0.16926753520965576, "step": 62450 }, { "epoch": 0.31024308832440917, "grad_norm": 0.6186746954917908, "learning_rate": 1.6156228806612426e-05, "loss": 0.1634754180908203, "step": 62460 }, { "epoch": 0.3102927590077784, "grad_norm": 1.546824336051941, "learning_rate": 1.6154960994749602e-05, "loss": 0.17645751237869262, "step": 62470 }, { "epoch": 0.31034242969114767, "grad_norm": 0.6156466007232666, "learning_rate": 1.615369302359886e-05, "loss": 0.16048426628112794, "step": 62480 }, { "epoch": 0.310392100374517, "grad_norm": 1.127901315689087, "learning_rate": 1.6152424893193018e-05, "loss": 0.17678866386413575, "step": 62490 }, { "epoch": 0.31044177105788623, "grad_norm": 1.8684002161026, "learning_rate": 1.6151156603564895e-05, "loss": 0.17051544189453124, "step": 62500 }, { "epoch": 0.3104914417412555, "grad_norm": 1.3074653148651123, "learning_rate": 1.6149888154747312e-05, "loss": 0.16476378440856934, "step": 62510 }, { "epoch": 0.31054111242462473, "grad_norm": 0.9038940668106079, "learning_rate": 1.6148619546773098e-05, "loss": 0.1601574659347534, "step": 62520 }, { "epoch": 0.310590783107994, "grad_norm": 1.1297792196273804, "learning_rate": 1.614735077967508e-05, "loss": 0.16904088258743286, "step": 62530 }, { "epoch": 0.31064045379136324, "grad_norm": 1.4315012693405151, "learning_rate": 1.6146081853486094e-05, "loss": 0.18182191848754883, "step": 62540 }, { "epoch": 0.31069012447473254, "grad_norm": 2.4190399646759033, "learning_rate": 1.6144812768238982e-05, "loss": 0.1624687671661377, "step": 62550 }, { "epoch": 0.3107397951581018, "grad_norm": 1.727953314781189, "learning_rate": 1.614354352396659e-05, "loss": 0.16729508638381957, "step": 62560 }, { "epoch": 0.31078946584147105, "grad_norm": 1.3501721620559692, "learning_rate": 1.6142274120701756e-05, "loss": 0.16471967697143555, "step": 62570 }, { "epoch": 0.3108391365248403, "grad_norm": 1.5727742910385132, "learning_rate": 1.614100455847734e-05, "loss": 0.1775004506111145, "step": 62580 }, { "epoch": 0.31088880720820955, "grad_norm": 1.5948694944381714, "learning_rate": 1.6139734837326197e-05, "loss": 0.18366943597793578, "step": 62590 }, { "epoch": 0.31093847789157886, "grad_norm": 1.4703094959259033, "learning_rate": 1.6138464957281183e-05, "loss": 0.15928289890289307, "step": 62600 }, { "epoch": 0.3109881485749481, "grad_norm": 0.7870783805847168, "learning_rate": 1.6137194918375167e-05, "loss": 0.16889605522155762, "step": 62610 }, { "epoch": 0.31103781925831736, "grad_norm": 1.6282410621643066, "learning_rate": 1.6135924720641013e-05, "loss": 0.15439579486846924, "step": 62620 }, { "epoch": 0.3110874899416866, "grad_norm": 0.876442551612854, "learning_rate": 1.6134654364111594e-05, "loss": 0.1582297921180725, "step": 62630 }, { "epoch": 0.31113716062505586, "grad_norm": 1.082389235496521, "learning_rate": 1.6133383848819784e-05, "loss": 0.171200168132782, "step": 62640 }, { "epoch": 0.3111868313084251, "grad_norm": 1.2695350646972656, "learning_rate": 1.6132113174798466e-05, "loss": 0.17263383865356446, "step": 62650 }, { "epoch": 0.3112365019917944, "grad_norm": 1.6108020544052124, "learning_rate": 1.6130842342080527e-05, "loss": 0.18729751110076903, "step": 62660 }, { "epoch": 0.3112861726751637, "grad_norm": 3.0403285026550293, "learning_rate": 1.6129571350698857e-05, "loss": 0.1675009846687317, "step": 62670 }, { "epoch": 0.3113358433585329, "grad_norm": 1.3469595909118652, "learning_rate": 1.612830020068634e-05, "loss": 0.1857454776763916, "step": 62680 }, { "epoch": 0.3113855140419022, "grad_norm": 1.581005334854126, "learning_rate": 1.612702889207588e-05, "loss": 0.16328738927841185, "step": 62690 }, { "epoch": 0.31143518472527143, "grad_norm": 1.3965158462524414, "learning_rate": 1.6125757424900373e-05, "loss": 0.1750526547431946, "step": 62700 }, { "epoch": 0.31148485540864074, "grad_norm": 0.9495474100112915, "learning_rate": 1.612448579919273e-05, "loss": 0.17552857398986815, "step": 62710 }, { "epoch": 0.31153452609201, "grad_norm": 1.4992319345474243, "learning_rate": 1.6123214014985856e-05, "loss": 0.17074645757675172, "step": 62720 }, { "epoch": 0.31158419677537924, "grad_norm": 1.5808237791061401, "learning_rate": 1.6121942072312663e-05, "loss": 0.18377481698989867, "step": 62730 }, { "epoch": 0.3116338674587485, "grad_norm": 0.9689716696739197, "learning_rate": 1.6120669971206075e-05, "loss": 0.18667840957641602, "step": 62740 }, { "epoch": 0.31168353814211774, "grad_norm": 1.5393917560577393, "learning_rate": 1.6119397711699006e-05, "loss": 0.19287495613098143, "step": 62750 }, { "epoch": 0.311733208825487, "grad_norm": 1.2277789115905762, "learning_rate": 1.6118125293824385e-05, "loss": 0.16303739547729493, "step": 62760 }, { "epoch": 0.3117828795088563, "grad_norm": 1.4695243835449219, "learning_rate": 1.6116852717615143e-05, "loss": 0.16223688125610353, "step": 62770 }, { "epoch": 0.31183255019222555, "grad_norm": 1.093534231185913, "learning_rate": 1.6115579983104214e-05, "loss": 0.1737079620361328, "step": 62780 }, { "epoch": 0.3118822208755948, "grad_norm": 0.9924380779266357, "learning_rate": 1.611430709032453e-05, "loss": 0.1735583186149597, "step": 62790 }, { "epoch": 0.31193189155896406, "grad_norm": 0.7064553499221802, "learning_rate": 1.611303403930904e-05, "loss": 0.17929477691650392, "step": 62800 }, { "epoch": 0.3119815622423333, "grad_norm": 1.5948233604431152, "learning_rate": 1.6111760830090683e-05, "loss": 0.18503081798553467, "step": 62810 }, { "epoch": 0.3120312329257026, "grad_norm": 0.6448482275009155, "learning_rate": 1.6110487462702415e-05, "loss": 0.1566542148590088, "step": 62820 }, { "epoch": 0.31208090360907187, "grad_norm": 0.7627049088478088, "learning_rate": 1.610921393717719e-05, "loss": 0.15397846698760986, "step": 62830 }, { "epoch": 0.3121305742924411, "grad_norm": 0.8520624041557312, "learning_rate": 1.6107940253547964e-05, "loss": 0.16910231113433838, "step": 62840 }, { "epoch": 0.3121802449758104, "grad_norm": 0.9995978474617004, "learning_rate": 1.61066664118477e-05, "loss": 0.17181775569915772, "step": 62850 }, { "epoch": 0.3122299156591796, "grad_norm": 0.878445565700531, "learning_rate": 1.6105392412109364e-05, "loss": 0.16882559061050414, "step": 62860 }, { "epoch": 0.3122795863425489, "grad_norm": 1.3788994550704956, "learning_rate": 1.6104118254365925e-05, "loss": 0.1817052721977234, "step": 62870 }, { "epoch": 0.3123292570259182, "grad_norm": 0.9411935806274414, "learning_rate": 1.6102843938650364e-05, "loss": 0.16092214584350586, "step": 62880 }, { "epoch": 0.31237892770928744, "grad_norm": 1.0471564531326294, "learning_rate": 1.6101569464995654e-05, "loss": 0.17637407779693604, "step": 62890 }, { "epoch": 0.3124285983926567, "grad_norm": 1.6331865787506104, "learning_rate": 1.610029483343478e-05, "loss": 0.17010891437530518, "step": 62900 }, { "epoch": 0.31247826907602594, "grad_norm": 1.2103103399276733, "learning_rate": 1.609902004400073e-05, "loss": 0.1686938762664795, "step": 62910 }, { "epoch": 0.3125279397593952, "grad_norm": 1.5196834802627563, "learning_rate": 1.609774509672649e-05, "loss": 0.18495954275131227, "step": 62920 }, { "epoch": 0.3125776104427645, "grad_norm": 1.227096438407898, "learning_rate": 1.6096469991645062e-05, "loss": 0.16915979385375976, "step": 62930 }, { "epoch": 0.31262728112613375, "grad_norm": 1.6800868511199951, "learning_rate": 1.6095194728789443e-05, "loss": 0.16731863021850585, "step": 62940 }, { "epoch": 0.312676951809503, "grad_norm": 1.772470235824585, "learning_rate": 1.609391930819263e-05, "loss": 0.18868234157562255, "step": 62950 }, { "epoch": 0.31272662249287225, "grad_norm": 1.5626914501190186, "learning_rate": 1.609264372988764e-05, "loss": 0.17184829711914062, "step": 62960 }, { "epoch": 0.3127762931762415, "grad_norm": 0.926871657371521, "learning_rate": 1.6091367993907477e-05, "loss": 0.1628340721130371, "step": 62970 }, { "epoch": 0.31282596385961076, "grad_norm": 1.0837483406066895, "learning_rate": 1.6090092100285162e-05, "loss": 0.1581876039505005, "step": 62980 }, { "epoch": 0.31287563454298006, "grad_norm": 1.1137292385101318, "learning_rate": 1.608881604905371e-05, "loss": 0.17439429759979247, "step": 62990 }, { "epoch": 0.3129253052263493, "grad_norm": 1.308701753616333, "learning_rate": 1.6087539840246148e-05, "loss": 0.1731213092803955, "step": 63000 }, { "epoch": 0.31297497590971857, "grad_norm": 1.726144552230835, "learning_rate": 1.6086263473895504e-05, "loss": 0.16875252723693848, "step": 63010 }, { "epoch": 0.3130246465930878, "grad_norm": 1.115851640701294, "learning_rate": 1.6084986950034808e-05, "loss": 0.17194398641586303, "step": 63020 }, { "epoch": 0.31307431727645707, "grad_norm": 0.9707150459289551, "learning_rate": 1.6083710268697093e-05, "loss": 0.18339061737060547, "step": 63030 }, { "epoch": 0.3131239879598264, "grad_norm": 1.0055975914001465, "learning_rate": 1.6082433429915405e-05, "loss": 0.16543874740600586, "step": 63040 }, { "epoch": 0.31317365864319563, "grad_norm": 0.6706790328025818, "learning_rate": 1.6081156433722788e-05, "loss": 0.17787039279937744, "step": 63050 }, { "epoch": 0.3132233293265649, "grad_norm": 1.3752503395080566, "learning_rate": 1.6079879280152286e-05, "loss": 0.17739707231521606, "step": 63060 }, { "epoch": 0.31327300000993413, "grad_norm": 0.7627235054969788, "learning_rate": 1.607860196923695e-05, "loss": 0.1678652286529541, "step": 63070 }, { "epoch": 0.3133226706933034, "grad_norm": 1.162643313407898, "learning_rate": 1.6077324501009843e-05, "loss": 0.16677929162979127, "step": 63080 }, { "epoch": 0.31337234137667264, "grad_norm": 1.6730576753616333, "learning_rate": 1.6076046875504017e-05, "loss": 0.16019082069396973, "step": 63090 }, { "epoch": 0.31342201206004194, "grad_norm": 1.8845261335372925, "learning_rate": 1.6074769092752543e-05, "loss": 0.17414438724517822, "step": 63100 }, { "epoch": 0.3134716827434112, "grad_norm": 1.33834707736969, "learning_rate": 1.607349115278849e-05, "loss": 0.18962600231170654, "step": 63110 }, { "epoch": 0.31352135342678045, "grad_norm": 0.9652643799781799, "learning_rate": 1.6072213055644925e-05, "loss": 0.18015897274017334, "step": 63120 }, { "epoch": 0.3135710241101497, "grad_norm": 1.9821467399597168, "learning_rate": 1.6070934801354926e-05, "loss": 0.17550282478332518, "step": 63130 }, { "epoch": 0.31362069479351895, "grad_norm": 0.8080114126205444, "learning_rate": 1.606965638995158e-05, "loss": 0.1750950336456299, "step": 63140 }, { "epoch": 0.31367036547688826, "grad_norm": 1.2693839073181152, "learning_rate": 1.6068377821467965e-05, "loss": 0.1660178780555725, "step": 63150 }, { "epoch": 0.3137200361602575, "grad_norm": 1.340797781944275, "learning_rate": 1.6067099095937172e-05, "loss": 0.17835478782653807, "step": 63160 }, { "epoch": 0.31376970684362676, "grad_norm": 2.186384677886963, "learning_rate": 1.6065820213392295e-05, "loss": 0.16725140810012817, "step": 63170 }, { "epoch": 0.313819377526996, "grad_norm": 0.9658874273300171, "learning_rate": 1.606454117386643e-05, "loss": 0.1934924840927124, "step": 63180 }, { "epoch": 0.31386904821036526, "grad_norm": 1.4193003177642822, "learning_rate": 1.6063261977392676e-05, "loss": 0.16734795570373534, "step": 63190 }, { "epoch": 0.3139187188937345, "grad_norm": 1.5804634094238281, "learning_rate": 1.606198262400414e-05, "loss": 0.16418676376342772, "step": 63200 }, { "epoch": 0.3139683895771038, "grad_norm": 1.708805799484253, "learning_rate": 1.606070311373393e-05, "loss": 0.1719364643096924, "step": 63210 }, { "epoch": 0.3140180602604731, "grad_norm": 0.8162305951118469, "learning_rate": 1.605942344661516e-05, "loss": 0.18116953372955322, "step": 63220 }, { "epoch": 0.3140677309438423, "grad_norm": 1.2461979389190674, "learning_rate": 1.6058143622680952e-05, "loss": 0.18734776973724365, "step": 63230 }, { "epoch": 0.3141174016272116, "grad_norm": 0.7110218405723572, "learning_rate": 1.6056863641964418e-05, "loss": 0.17067296504974366, "step": 63240 }, { "epoch": 0.31416707231058083, "grad_norm": 1.1785694360733032, "learning_rate": 1.6055583504498694e-05, "loss": 0.15931740999221802, "step": 63250 }, { "epoch": 0.31421674299395014, "grad_norm": 2.1222493648529053, "learning_rate": 1.60543032103169e-05, "loss": 0.1711192846298218, "step": 63260 }, { "epoch": 0.3142664136773194, "grad_norm": 1.3107460737228394, "learning_rate": 1.6053022759452174e-05, "loss": 0.1672094702720642, "step": 63270 }, { "epoch": 0.31431608436068864, "grad_norm": 0.6827269792556763, "learning_rate": 1.6051742151937655e-05, "loss": 0.1713330030441284, "step": 63280 }, { "epoch": 0.3143657550440579, "grad_norm": 1.6885042190551758, "learning_rate": 1.6050461387806478e-05, "loss": 0.17920631170272827, "step": 63290 }, { "epoch": 0.31441542572742714, "grad_norm": 1.5399373769760132, "learning_rate": 1.6049180467091796e-05, "loss": 0.16387739181518554, "step": 63300 }, { "epoch": 0.3144650964107964, "grad_norm": 1.5253621339797974, "learning_rate": 1.6047899389826754e-05, "loss": 0.17167099714279174, "step": 63310 }, { "epoch": 0.3145147670941657, "grad_norm": 0.5808646082878113, "learning_rate": 1.6046618156044508e-05, "loss": 0.15751934051513672, "step": 63320 }, { "epoch": 0.31456443777753496, "grad_norm": 1.5538647174835205, "learning_rate": 1.6045336765778217e-05, "loss": 0.16281070709228515, "step": 63330 }, { "epoch": 0.3146141084609042, "grad_norm": 1.1485321521759033, "learning_rate": 1.604405521906104e-05, "loss": 0.19196829795837403, "step": 63340 }, { "epoch": 0.31466377914427346, "grad_norm": 1.7583304643630981, "learning_rate": 1.6042773515926143e-05, "loss": 0.17277244329452515, "step": 63350 }, { "epoch": 0.3147134498276427, "grad_norm": 1.4675500392913818, "learning_rate": 1.60414916564067e-05, "loss": 0.18228390216827392, "step": 63360 }, { "epoch": 0.314763120511012, "grad_norm": 0.9600600004196167, "learning_rate": 1.604020964053588e-05, "loss": 0.1893411636352539, "step": 63370 }, { "epoch": 0.31481279119438127, "grad_norm": 1.1114976406097412, "learning_rate": 1.603892746834686e-05, "loss": 0.17503211498260499, "step": 63380 }, { "epoch": 0.3148624618777505, "grad_norm": 1.4645932912826538, "learning_rate": 1.603764513987283e-05, "loss": 0.17297435998916627, "step": 63390 }, { "epoch": 0.3149121325611198, "grad_norm": 1.68517005443573, "learning_rate": 1.6036362655146975e-05, "loss": 0.1662417769432068, "step": 63400 }, { "epoch": 0.314961803244489, "grad_norm": 1.4855777025222778, "learning_rate": 1.6035080014202473e-05, "loss": 0.17138359546661378, "step": 63410 }, { "epoch": 0.3150114739278583, "grad_norm": 0.7960230708122253, "learning_rate": 1.6033797217072532e-05, "loss": 0.17004599571228027, "step": 63420 }, { "epoch": 0.3150611446112276, "grad_norm": 1.7189117670059204, "learning_rate": 1.6032514263790344e-05, "loss": 0.17334952354431152, "step": 63430 }, { "epoch": 0.31511081529459684, "grad_norm": 0.7652947902679443, "learning_rate": 1.603123115438911e-05, "loss": 0.16869051456451417, "step": 63440 }, { "epoch": 0.3151604859779661, "grad_norm": 1.386993169784546, "learning_rate": 1.6029947888902046e-05, "loss": 0.16684794425964355, "step": 63450 }, { "epoch": 0.31521015666133534, "grad_norm": 0.8411228656768799, "learning_rate": 1.602866446736235e-05, "loss": 0.1723383665084839, "step": 63460 }, { "epoch": 0.3152598273447046, "grad_norm": 0.9480718374252319, "learning_rate": 1.6027380889803243e-05, "loss": 0.17353286743164062, "step": 63470 }, { "epoch": 0.3153094980280739, "grad_norm": 1.0754294395446777, "learning_rate": 1.602609715625794e-05, "loss": 0.15638997554779052, "step": 63480 }, { "epoch": 0.31535916871144315, "grad_norm": 0.7186952829360962, "learning_rate": 1.6024813266759666e-05, "loss": 0.16506987810134888, "step": 63490 }, { "epoch": 0.3154088393948124, "grad_norm": 1.7085943222045898, "learning_rate": 1.6023529221341648e-05, "loss": 0.17180349826812744, "step": 63500 }, { "epoch": 0.31545851007818165, "grad_norm": 2.1409964561462402, "learning_rate": 1.6022245020037114e-05, "loss": 0.16560349464416504, "step": 63510 }, { "epoch": 0.3155081807615509, "grad_norm": 1.2429068088531494, "learning_rate": 1.6020960662879303e-05, "loss": 0.17805148363113404, "step": 63520 }, { "epoch": 0.31555785144492016, "grad_norm": 1.2360368967056274, "learning_rate": 1.601967614990145e-05, "loss": 0.17980910539627076, "step": 63530 }, { "epoch": 0.31560752212828946, "grad_norm": 2.8105692863464355, "learning_rate": 1.6018391481136796e-05, "loss": 0.17330098152160645, "step": 63540 }, { "epoch": 0.3156571928116587, "grad_norm": 0.8411208391189575, "learning_rate": 1.601710665661859e-05, "loss": 0.16896474361419678, "step": 63550 }, { "epoch": 0.31570686349502797, "grad_norm": 1.2782806158065796, "learning_rate": 1.601582167638009e-05, "loss": 0.17153518199920653, "step": 63560 }, { "epoch": 0.3157565341783972, "grad_norm": 0.8002018928527832, "learning_rate": 1.6014536540454538e-05, "loss": 0.17067451477050782, "step": 63570 }, { "epoch": 0.31580620486176647, "grad_norm": 2.3488497734069824, "learning_rate": 1.60132512488752e-05, "loss": 0.16399199962615968, "step": 63580 }, { "epoch": 0.3158558755451357, "grad_norm": 1.3876949548721313, "learning_rate": 1.601196580167534e-05, "loss": 0.1470305800437927, "step": 63590 }, { "epoch": 0.31590554622850503, "grad_norm": 1.5867751836776733, "learning_rate": 1.601068019888822e-05, "loss": 0.16783612966537476, "step": 63600 }, { "epoch": 0.3159552169118743, "grad_norm": 1.2878313064575195, "learning_rate": 1.600939444054711e-05, "loss": 0.17190215587615967, "step": 63610 }, { "epoch": 0.31600488759524353, "grad_norm": 1.5761692523956299, "learning_rate": 1.6008108526685295e-05, "loss": 0.165567946434021, "step": 63620 }, { "epoch": 0.3160545582786128, "grad_norm": 1.2018144130706787, "learning_rate": 1.6006822457336044e-05, "loss": 0.15846859216690062, "step": 63630 }, { "epoch": 0.31610422896198204, "grad_norm": 0.5598787069320679, "learning_rate": 1.6005536232532644e-05, "loss": 0.16876096725463868, "step": 63640 }, { "epoch": 0.31615389964535134, "grad_norm": 0.7911145091056824, "learning_rate": 1.600424985230838e-05, "loss": 0.16775935888290405, "step": 63650 }, { "epoch": 0.3162035703287206, "grad_norm": 2.162630796432495, "learning_rate": 1.6002963316696543e-05, "loss": 0.1607032299041748, "step": 63660 }, { "epoch": 0.31625324101208985, "grad_norm": 2.421062469482422, "learning_rate": 1.600167662573043e-05, "loss": 0.17459913492202758, "step": 63670 }, { "epoch": 0.3163029116954591, "grad_norm": 0.9559186697006226, "learning_rate": 1.6000389779443337e-05, "loss": 0.16865407228469848, "step": 63680 }, { "epoch": 0.31635258237882835, "grad_norm": 1.4893397092819214, "learning_rate": 1.5999102777868573e-05, "loss": 0.16347261667251586, "step": 63690 }, { "epoch": 0.3164022530621976, "grad_norm": 2.3788557052612305, "learning_rate": 1.599781562103944e-05, "loss": 0.1582493782043457, "step": 63700 }, { "epoch": 0.3164519237455669, "grad_norm": 1.6105151176452637, "learning_rate": 1.599652830898925e-05, "loss": 0.1719451665878296, "step": 63710 }, { "epoch": 0.31650159442893616, "grad_norm": 1.4486949443817139, "learning_rate": 1.5995240841751316e-05, "loss": 0.1847351908683777, "step": 63720 }, { "epoch": 0.3165512651123054, "grad_norm": 1.1140193939208984, "learning_rate": 1.5993953219358963e-05, "loss": 0.1727679967880249, "step": 63730 }, { "epoch": 0.31660093579567467, "grad_norm": 1.0313658714294434, "learning_rate": 1.599266544184551e-05, "loss": 0.178460955619812, "step": 63740 }, { "epoch": 0.3166506064790439, "grad_norm": 0.9378364086151123, "learning_rate": 1.5991377509244283e-05, "loss": 0.17476344108581543, "step": 63750 }, { "epoch": 0.3167002771624132, "grad_norm": 1.317746639251709, "learning_rate": 1.5990089421588618e-05, "loss": 0.17006299495697022, "step": 63760 }, { "epoch": 0.3167499478457825, "grad_norm": 1.272301197052002, "learning_rate": 1.5988801178911842e-05, "loss": 0.15872602462768554, "step": 63770 }, { "epoch": 0.3167996185291517, "grad_norm": 2.838566541671753, "learning_rate": 1.5987512781247305e-05, "loss": 0.17846373319625855, "step": 63780 }, { "epoch": 0.316849289212521, "grad_norm": 1.2829442024230957, "learning_rate": 1.5986224228628343e-05, "loss": 0.17458024024963378, "step": 63790 }, { "epoch": 0.31689895989589023, "grad_norm": 2.8737101554870605, "learning_rate": 1.598493552108831e-05, "loss": 0.17163583040237426, "step": 63800 }, { "epoch": 0.3169486305792595, "grad_norm": 1.1420294046401978, "learning_rate": 1.5983646658660544e-05, "loss": 0.17736390829086304, "step": 63810 }, { "epoch": 0.3169983012626288, "grad_norm": 1.0060043334960938, "learning_rate": 1.598235764137841e-05, "loss": 0.1692527413368225, "step": 63820 }, { "epoch": 0.31704797194599804, "grad_norm": 1.1309772729873657, "learning_rate": 1.5981068469275268e-05, "loss": 0.16868555545806885, "step": 63830 }, { "epoch": 0.3170976426293673, "grad_norm": 1.8868215084075928, "learning_rate": 1.5979779142384478e-05, "loss": 0.18133583068847656, "step": 63840 }, { "epoch": 0.31714731331273655, "grad_norm": 1.6555441617965698, "learning_rate": 1.597848966073941e-05, "loss": 0.18114243745803832, "step": 63850 }, { "epoch": 0.3171969839961058, "grad_norm": 1.2309532165527344, "learning_rate": 1.597720002437343e-05, "loss": 0.17748491764068602, "step": 63860 }, { "epoch": 0.3172466546794751, "grad_norm": 1.1411373615264893, "learning_rate": 1.597591023331992e-05, "loss": 0.17910208702087402, "step": 63870 }, { "epoch": 0.31729632536284436, "grad_norm": 1.679912805557251, "learning_rate": 1.5974620287612252e-05, "loss": 0.17998931407928467, "step": 63880 }, { "epoch": 0.3173459960462136, "grad_norm": 0.9279202818870544, "learning_rate": 1.5973330187283817e-05, "loss": 0.1709596633911133, "step": 63890 }, { "epoch": 0.31739566672958286, "grad_norm": 1.2878278493881226, "learning_rate": 1.5972039932367995e-05, "loss": 0.17525941133499146, "step": 63900 }, { "epoch": 0.3174453374129521, "grad_norm": 1.9942702054977417, "learning_rate": 1.5970749522898182e-05, "loss": 0.16709399223327637, "step": 63910 }, { "epoch": 0.31749500809632136, "grad_norm": 1.6953799724578857, "learning_rate": 1.5969458958907777e-05, "loss": 0.16689277887344361, "step": 63920 }, { "epoch": 0.31754467877969067, "grad_norm": 1.072813630104065, "learning_rate": 1.5968168240430166e-05, "loss": 0.16535052061080932, "step": 63930 }, { "epoch": 0.3175943494630599, "grad_norm": 1.5093910694122314, "learning_rate": 1.5966877367498768e-05, "loss": 0.16900348663330078, "step": 63940 }, { "epoch": 0.3176440201464292, "grad_norm": 1.156205654144287, "learning_rate": 1.596558634014698e-05, "loss": 0.17543166875839233, "step": 63950 }, { "epoch": 0.3176936908297984, "grad_norm": 0.8921290636062622, "learning_rate": 1.5964295158408218e-05, "loss": 0.15614287853240966, "step": 63960 }, { "epoch": 0.3177433615131677, "grad_norm": 1.3158212900161743, "learning_rate": 1.5963003822315893e-05, "loss": 0.16293773651123047, "step": 63970 }, { "epoch": 0.317793032196537, "grad_norm": 1.1333129405975342, "learning_rate": 1.5961712331903426e-05, "loss": 0.1765455722808838, "step": 63980 }, { "epoch": 0.31784270287990624, "grad_norm": 1.2308096885681152, "learning_rate": 1.5960420687204242e-05, "loss": 0.18469727039337158, "step": 63990 }, { "epoch": 0.3178923735632755, "grad_norm": 1.4094219207763672, "learning_rate": 1.595912888825177e-05, "loss": 0.17306549549102784, "step": 64000 }, { "epoch": 0.31794204424664474, "grad_norm": 2.4479308128356934, "learning_rate": 1.5957836935079434e-05, "loss": 0.17580795288085938, "step": 64010 }, { "epoch": 0.317991714930014, "grad_norm": 0.8487840294837952, "learning_rate": 1.5956544827720673e-05, "loss": 0.16490107774734497, "step": 64020 }, { "epoch": 0.31804138561338324, "grad_norm": 0.930911660194397, "learning_rate": 1.595525256620893e-05, "loss": 0.1728609323501587, "step": 64030 }, { "epoch": 0.31809105629675255, "grad_norm": 0.9712290167808533, "learning_rate": 1.5953960150577644e-05, "loss": 0.1640864133834839, "step": 64040 }, { "epoch": 0.3181407269801218, "grad_norm": 1.7093757390975952, "learning_rate": 1.5952667580860267e-05, "loss": 0.1673877000808716, "step": 64050 }, { "epoch": 0.31819039766349105, "grad_norm": 0.8913703560829163, "learning_rate": 1.595137485709024e-05, "loss": 0.1703310489654541, "step": 64060 }, { "epoch": 0.3182400683468603, "grad_norm": 1.1323734521865845, "learning_rate": 1.595008197930103e-05, "loss": 0.15741209983825682, "step": 64070 }, { "epoch": 0.31828973903022956, "grad_norm": 0.767180323600769, "learning_rate": 1.5948788947526087e-05, "loss": 0.16279990673065187, "step": 64080 }, { "epoch": 0.31833940971359886, "grad_norm": 1.1955291032791138, "learning_rate": 1.594749576179888e-05, "loss": 0.1672611951828003, "step": 64090 }, { "epoch": 0.3183890803969681, "grad_norm": 1.4771168231964111, "learning_rate": 1.5946202422152875e-05, "loss": 0.16879897117614745, "step": 64100 }, { "epoch": 0.31843875108033737, "grad_norm": 1.481773018836975, "learning_rate": 1.594490892862154e-05, "loss": 0.165325129032135, "step": 64110 }, { "epoch": 0.3184884217637066, "grad_norm": 1.2853084802627563, "learning_rate": 1.5943615281238356e-05, "loss": 0.16499128341674804, "step": 64120 }, { "epoch": 0.31853809244707587, "grad_norm": 0.8271659016609192, "learning_rate": 1.5942321480036796e-05, "loss": 0.17572579383850098, "step": 64130 }, { "epoch": 0.3185877631304451, "grad_norm": 1.3440048694610596, "learning_rate": 1.5941027525050343e-05, "loss": 0.1752177357673645, "step": 64140 }, { "epoch": 0.31863743381381443, "grad_norm": 1.8107616901397705, "learning_rate": 1.593973341631249e-05, "loss": 0.17542985677719117, "step": 64150 }, { "epoch": 0.3186871044971837, "grad_norm": 1.8392351865768433, "learning_rate": 1.593843915385672e-05, "loss": 0.18133728504180907, "step": 64160 }, { "epoch": 0.31873677518055293, "grad_norm": 1.7562845945358276, "learning_rate": 1.5937144737716536e-05, "loss": 0.17809491157531737, "step": 64170 }, { "epoch": 0.3187864458639222, "grad_norm": 1.0923148393630981, "learning_rate": 1.5935850167925435e-05, "loss": 0.1539873957633972, "step": 64180 }, { "epoch": 0.31883611654729144, "grad_norm": 2.613447904586792, "learning_rate": 1.5934555444516916e-05, "loss": 0.17447466850280763, "step": 64190 }, { "epoch": 0.31888578723066074, "grad_norm": 2.65619158744812, "learning_rate": 1.5933260567524488e-05, "loss": 0.18318709135055541, "step": 64200 }, { "epoch": 0.31893545791403, "grad_norm": 0.876918613910675, "learning_rate": 1.5931965536981663e-05, "loss": 0.15645501613616944, "step": 64210 }, { "epoch": 0.31898512859739925, "grad_norm": 1.7479852437973022, "learning_rate": 1.593067035292195e-05, "loss": 0.1721707820892334, "step": 64220 }, { "epoch": 0.3190347992807685, "grad_norm": 0.9803881645202637, "learning_rate": 1.5929375015378878e-05, "loss": 0.17908457517623902, "step": 64230 }, { "epoch": 0.31908446996413775, "grad_norm": 0.7920376658439636, "learning_rate": 1.5928079524385965e-05, "loss": 0.1544266700744629, "step": 64240 }, { "epoch": 0.319134140647507, "grad_norm": 0.9164698719978333, "learning_rate": 1.592678387997674e-05, "loss": 0.17709748744964598, "step": 64250 }, { "epoch": 0.3191838113308763, "grad_norm": 1.8225173950195312, "learning_rate": 1.5925488082184726e-05, "loss": 0.17099461555480958, "step": 64260 }, { "epoch": 0.31923348201424556, "grad_norm": 1.266603708267212, "learning_rate": 1.5924192131043465e-05, "loss": 0.17614600658416749, "step": 64270 }, { "epoch": 0.3192831526976148, "grad_norm": 2.274294853210449, "learning_rate": 1.5922896026586493e-05, "loss": 0.1833578586578369, "step": 64280 }, { "epoch": 0.31933282338098407, "grad_norm": 1.3692940473556519, "learning_rate": 1.5921599768847355e-05, "loss": 0.17814309597015382, "step": 64290 }, { "epoch": 0.3193824940643533, "grad_norm": 1.1931673288345337, "learning_rate": 1.5920303357859595e-05, "loss": 0.16291455030441285, "step": 64300 }, { "epoch": 0.3194321647477226, "grad_norm": 2.029043436050415, "learning_rate": 1.5919006793656767e-05, "loss": 0.17171868085861205, "step": 64310 }, { "epoch": 0.3194818354310919, "grad_norm": 0.814907431602478, "learning_rate": 1.5917710076272416e-05, "loss": 0.18030593395233155, "step": 64320 }, { "epoch": 0.31953150611446113, "grad_norm": 1.137528419494629, "learning_rate": 1.5916413205740113e-05, "loss": 0.1815575122833252, "step": 64330 }, { "epoch": 0.3195811767978304, "grad_norm": 1.198154330253601, "learning_rate": 1.5915116182093415e-05, "loss": 0.17180063724517822, "step": 64340 }, { "epoch": 0.31963084748119963, "grad_norm": 0.5902708172798157, "learning_rate": 1.5913819005365888e-05, "loss": 0.1649146318435669, "step": 64350 }, { "epoch": 0.3196805181645689, "grad_norm": 1.7032930850982666, "learning_rate": 1.59125216755911e-05, "loss": 0.1882108211517334, "step": 64360 }, { "epoch": 0.3197301888479382, "grad_norm": 1.9099586009979248, "learning_rate": 1.5911224192802636e-05, "loss": 0.17360785007476806, "step": 64370 }, { "epoch": 0.31977985953130744, "grad_norm": 1.2542883157730103, "learning_rate": 1.590992655703406e-05, "loss": 0.1757072925567627, "step": 64380 }, { "epoch": 0.3198295302146767, "grad_norm": 2.0312156677246094, "learning_rate": 1.590862876831896e-05, "loss": 0.17039864063262938, "step": 64390 }, { "epoch": 0.31987920089804595, "grad_norm": 1.2696279287338257, "learning_rate": 1.5907330826690927e-05, "loss": 0.17632031440734863, "step": 64400 }, { "epoch": 0.3199288715814152, "grad_norm": 1.1761999130249023, "learning_rate": 1.590603273218355e-05, "loss": 0.1736292600631714, "step": 64410 }, { "epoch": 0.3199785422647845, "grad_norm": 1.2846834659576416, "learning_rate": 1.5904734484830414e-05, "loss": 0.17445590496063232, "step": 64420 }, { "epoch": 0.32002821294815376, "grad_norm": 1.4510174989700317, "learning_rate": 1.5903436084665127e-05, "loss": 0.1775691270828247, "step": 64430 }, { "epoch": 0.320077883631523, "grad_norm": 1.5429614782333374, "learning_rate": 1.590213753172129e-05, "loss": 0.17801976203918457, "step": 64440 }, { "epoch": 0.32012755431489226, "grad_norm": 1.3845019340515137, "learning_rate": 1.5900838826032503e-05, "loss": 0.16302769184112548, "step": 64450 }, { "epoch": 0.3201772249982615, "grad_norm": 0.7334738373756409, "learning_rate": 1.5899539967632383e-05, "loss": 0.17506790161132812, "step": 64460 }, { "epoch": 0.32022689568163076, "grad_norm": 2.507796049118042, "learning_rate": 1.589824095655454e-05, "loss": 0.17383451461791993, "step": 64470 }, { "epoch": 0.32027656636500007, "grad_norm": 1.8042927980422974, "learning_rate": 1.5896941792832592e-05, "loss": 0.17032291889190673, "step": 64480 }, { "epoch": 0.3203262370483693, "grad_norm": 0.7156003713607788, "learning_rate": 1.589564247650016e-05, "loss": 0.17269418239593506, "step": 64490 }, { "epoch": 0.3203759077317386, "grad_norm": 1.3815079927444458, "learning_rate": 1.5894343007590874e-05, "loss": 0.18108034133911133, "step": 64500 }, { "epoch": 0.3204255784151078, "grad_norm": 0.7328810095787048, "learning_rate": 1.5893043386138356e-05, "loss": 0.1753413438796997, "step": 64510 }, { "epoch": 0.3204752490984771, "grad_norm": 0.7812858819961548, "learning_rate": 1.589174361217625e-05, "loss": 0.16270575523376465, "step": 64520 }, { "epoch": 0.3205249197818464, "grad_norm": 1.0826945304870605, "learning_rate": 1.5890443685738186e-05, "loss": 0.18086198568344117, "step": 64530 }, { "epoch": 0.32057459046521564, "grad_norm": 0.6501607298851013, "learning_rate": 1.588914360685781e-05, "loss": 0.1846766948699951, "step": 64540 }, { "epoch": 0.3206242611485849, "grad_norm": 2.1205551624298096, "learning_rate": 1.5887843375568762e-05, "loss": 0.16778280735015869, "step": 64550 }, { "epoch": 0.32067393183195414, "grad_norm": 0.673047661781311, "learning_rate": 1.5886542991904695e-05, "loss": 0.17129027843475342, "step": 64560 }, { "epoch": 0.3207236025153234, "grad_norm": 1.4879084825515747, "learning_rate": 1.5885242455899265e-05, "loss": 0.1740964651107788, "step": 64570 }, { "epoch": 0.32077327319869264, "grad_norm": 1.2874138355255127, "learning_rate": 1.5883941767586123e-05, "loss": 0.17552709579467773, "step": 64580 }, { "epoch": 0.32082294388206195, "grad_norm": 1.5928572416305542, "learning_rate": 1.5882640926998938e-05, "loss": 0.16817810535430908, "step": 64590 }, { "epoch": 0.3208726145654312, "grad_norm": 1.18178391456604, "learning_rate": 1.588133993417137e-05, "loss": 0.1715809464454651, "step": 64600 }, { "epoch": 0.32092228524880045, "grad_norm": 2.254385471343994, "learning_rate": 1.5880038789137088e-05, "loss": 0.20072433948516846, "step": 64610 }, { "epoch": 0.3209719559321697, "grad_norm": 2.3411452770233154, "learning_rate": 1.587873749192977e-05, "loss": 0.19048171043395995, "step": 64620 }, { "epoch": 0.32102162661553896, "grad_norm": 1.2241843938827515, "learning_rate": 1.5877436042583084e-05, "loss": 0.17331422567367555, "step": 64630 }, { "epoch": 0.32107129729890826, "grad_norm": 0.713678240776062, "learning_rate": 1.587613444113072e-05, "loss": 0.1653868079185486, "step": 64640 }, { "epoch": 0.3211209679822775, "grad_norm": 1.6100809574127197, "learning_rate": 1.5874832687606358e-05, "loss": 0.18340954780578614, "step": 64650 }, { "epoch": 0.32117063866564677, "grad_norm": 1.4347169399261475, "learning_rate": 1.5873530782043686e-05, "loss": 0.19129977226257325, "step": 64660 }, { "epoch": 0.321220309349016, "grad_norm": 1.3131959438323975, "learning_rate": 1.58722287244764e-05, "loss": 0.1712787628173828, "step": 64670 }, { "epoch": 0.32126998003238527, "grad_norm": 0.8799113035202026, "learning_rate": 1.5870926514938202e-05, "loss": 0.16603506803512574, "step": 64680 }, { "epoch": 0.3213196507157545, "grad_norm": 0.6955410838127136, "learning_rate": 1.586962415346278e-05, "loss": 0.166497802734375, "step": 64690 }, { "epoch": 0.32136932139912383, "grad_norm": 0.8674681782722473, "learning_rate": 1.5868321640083844e-05, "loss": 0.1725364089012146, "step": 64700 }, { "epoch": 0.3214189920824931, "grad_norm": 0.604194164276123, "learning_rate": 1.586701897483511e-05, "loss": 0.17177274227142333, "step": 64710 }, { "epoch": 0.32146866276586233, "grad_norm": 1.1726515293121338, "learning_rate": 1.5865716157750278e-05, "loss": 0.17460978031158447, "step": 64720 }, { "epoch": 0.3215183334492316, "grad_norm": 0.47981134057044983, "learning_rate": 1.5864413188863074e-05, "loss": 0.16049911975860595, "step": 64730 }, { "epoch": 0.32156800413260084, "grad_norm": 1.606879711151123, "learning_rate": 1.5863110068207214e-05, "loss": 0.17462546825408937, "step": 64740 }, { "epoch": 0.32161767481597014, "grad_norm": 2.039597749710083, "learning_rate": 1.586180679581642e-05, "loss": 0.1771007776260376, "step": 64750 }, { "epoch": 0.3216673454993394, "grad_norm": 1.5715690851211548, "learning_rate": 1.5860503371724424e-05, "loss": 0.17005631923675538, "step": 64760 }, { "epoch": 0.32171701618270865, "grad_norm": 1.1981958150863647, "learning_rate": 1.5859199795964958e-05, "loss": 0.16946895122528077, "step": 64770 }, { "epoch": 0.3217666868660779, "grad_norm": 0.8150897026062012, "learning_rate": 1.5857896068571757e-05, "loss": 0.1885124683380127, "step": 64780 }, { "epoch": 0.32181635754944715, "grad_norm": 1.033281922340393, "learning_rate": 1.5856592189578562e-05, "loss": 0.17560312747955323, "step": 64790 }, { "epoch": 0.3218660282328164, "grad_norm": 0.9553619623184204, "learning_rate": 1.5855288159019117e-05, "loss": 0.184673273563385, "step": 64800 }, { "epoch": 0.3219156989161857, "grad_norm": 2.0484673976898193, "learning_rate": 1.585398397692717e-05, "loss": 0.16775013208389283, "step": 64810 }, { "epoch": 0.32196536959955496, "grad_norm": 1.1772993803024292, "learning_rate": 1.5852679643336462e-05, "loss": 0.17965471744537354, "step": 64820 }, { "epoch": 0.3220150402829242, "grad_norm": 0.8784593343734741, "learning_rate": 1.5851375158280768e-05, "loss": 0.16200919151306153, "step": 64830 }, { "epoch": 0.32206471096629347, "grad_norm": 1.5756494998931885, "learning_rate": 1.5850070521793834e-05, "loss": 0.17160160541534425, "step": 64840 }, { "epoch": 0.3221143816496627, "grad_norm": 1.706175684928894, "learning_rate": 1.5848765733909428e-05, "loss": 0.16278154850006105, "step": 64850 }, { "epoch": 0.322164052333032, "grad_norm": 1.9061825275421143, "learning_rate": 1.5847460794661314e-05, "loss": 0.15606825351715087, "step": 64860 }, { "epoch": 0.3222137230164013, "grad_norm": 1.2176371812820435, "learning_rate": 1.584615570408327e-05, "loss": 0.17279090881347656, "step": 64870 }, { "epoch": 0.32226339369977053, "grad_norm": 1.1329731941223145, "learning_rate": 1.5844850462209065e-05, "loss": 0.16757798194885254, "step": 64880 }, { "epoch": 0.3223130643831398, "grad_norm": 1.7581374645233154, "learning_rate": 1.5843545069072482e-05, "loss": 0.17915689945220947, "step": 64890 }, { "epoch": 0.32236273506650903, "grad_norm": 0.9489224553108215, "learning_rate": 1.5842239524707297e-05, "loss": 0.16431808471679688, "step": 64900 }, { "epoch": 0.3224124057498783, "grad_norm": 1.8019413948059082, "learning_rate": 1.5840933829147304e-05, "loss": 0.17419812679290772, "step": 64910 }, { "epoch": 0.3224620764332476, "grad_norm": 1.0153698921203613, "learning_rate": 1.5839627982426296e-05, "loss": 0.17575559616088868, "step": 64920 }, { "epoch": 0.32251174711661684, "grad_norm": 1.2985478639602661, "learning_rate": 1.583832198457806e-05, "loss": 0.16333310604095458, "step": 64930 }, { "epoch": 0.3225614177999861, "grad_norm": 1.417250156402588, "learning_rate": 1.58370158356364e-05, "loss": 0.17623428106307984, "step": 64940 }, { "epoch": 0.32261108848335535, "grad_norm": 1.7198917865753174, "learning_rate": 1.5835709535635117e-05, "loss": 0.178011155128479, "step": 64950 }, { "epoch": 0.3226607591667246, "grad_norm": 0.9059072732925415, "learning_rate": 1.5834403084608017e-05, "loss": 0.1706581711769104, "step": 64960 }, { "epoch": 0.3227104298500939, "grad_norm": 1.023642897605896, "learning_rate": 1.5833096482588916e-05, "loss": 0.17634956836700438, "step": 64970 }, { "epoch": 0.32276010053346316, "grad_norm": 1.442086935043335, "learning_rate": 1.583178972961162e-05, "loss": 0.16608915328979493, "step": 64980 }, { "epoch": 0.3228097712168324, "grad_norm": 1.2562326192855835, "learning_rate": 1.5830482825709944e-05, "loss": 0.16592519283294677, "step": 64990 }, { "epoch": 0.32285944190020166, "grad_norm": 0.8270801901817322, "learning_rate": 1.5829175770917725e-05, "loss": 0.16794636249542236, "step": 65000 }, { "epoch": 0.3229091125835709, "grad_norm": 1.0646611452102661, "learning_rate": 1.582786856526878e-05, "loss": 0.16672898530960084, "step": 65010 }, { "epoch": 0.32295878326694016, "grad_norm": 2.0463850498199463, "learning_rate": 1.5826561208796937e-05, "loss": 0.1634867548942566, "step": 65020 }, { "epoch": 0.32300845395030947, "grad_norm": 1.5216701030731201, "learning_rate": 1.5825253701536035e-05, "loss": 0.17410253286361693, "step": 65030 }, { "epoch": 0.3230581246336787, "grad_norm": 2.1007416248321533, "learning_rate": 1.5823946043519908e-05, "loss": 0.19505250453948975, "step": 65040 }, { "epoch": 0.323107795317048, "grad_norm": 1.514306902885437, "learning_rate": 1.5822638234782395e-05, "loss": 0.1757647156715393, "step": 65050 }, { "epoch": 0.3231574660004172, "grad_norm": 0.7449834942817688, "learning_rate": 1.5821330275357352e-05, "loss": 0.17652268409729005, "step": 65060 }, { "epoch": 0.3232071366837865, "grad_norm": 1.5616704225540161, "learning_rate": 1.582002216527862e-05, "loss": 0.1826424479484558, "step": 65070 }, { "epoch": 0.32325680736715573, "grad_norm": 1.2174930572509766, "learning_rate": 1.5818713904580058e-05, "loss": 0.18071600198745727, "step": 65080 }, { "epoch": 0.32330647805052504, "grad_norm": 1.9100269079208374, "learning_rate": 1.5817405493295514e-05, "loss": 0.1737912654876709, "step": 65090 }, { "epoch": 0.3233561487338943, "grad_norm": 1.2637494802474976, "learning_rate": 1.5816096931458854e-05, "loss": 0.1739754557609558, "step": 65100 }, { "epoch": 0.32340581941726354, "grad_norm": 0.7446720004081726, "learning_rate": 1.581478821910395e-05, "loss": 0.17161591053009034, "step": 65110 }, { "epoch": 0.3234554901006328, "grad_norm": 0.9362534880638123, "learning_rate": 1.5813479356264663e-05, "loss": 0.1720202684402466, "step": 65120 }, { "epoch": 0.32350516078400204, "grad_norm": 3.119786500930786, "learning_rate": 1.5812170342974867e-05, "loss": 0.17685456275939943, "step": 65130 }, { "epoch": 0.32355483146737135, "grad_norm": 0.9142961502075195, "learning_rate": 1.581086117926844e-05, "loss": 0.18299753665924073, "step": 65140 }, { "epoch": 0.3236045021507406, "grad_norm": 1.1655144691467285, "learning_rate": 1.5809551865179264e-05, "loss": 0.17575110197067262, "step": 65150 }, { "epoch": 0.32365417283410985, "grad_norm": 1.136161208152771, "learning_rate": 1.580824240074122e-05, "loss": 0.17488101720809937, "step": 65160 }, { "epoch": 0.3237038435174791, "grad_norm": 1.9538954496383667, "learning_rate": 1.5806932785988194e-05, "loss": 0.17776573896408082, "step": 65170 }, { "epoch": 0.32375351420084836, "grad_norm": 1.0122231245040894, "learning_rate": 1.5805623020954087e-05, "loss": 0.18208248615264894, "step": 65180 }, { "epoch": 0.3238031848842176, "grad_norm": 0.9653313755989075, "learning_rate": 1.5804313105672787e-05, "loss": 0.16827080249786378, "step": 65190 }, { "epoch": 0.3238528555675869, "grad_norm": 1.1443636417388916, "learning_rate": 1.58030030401782e-05, "loss": 0.18273670673370362, "step": 65200 }, { "epoch": 0.32390252625095617, "grad_norm": 0.8049322962760925, "learning_rate": 1.580169282450423e-05, "loss": 0.16803224086761476, "step": 65210 }, { "epoch": 0.3239521969343254, "grad_norm": 0.6094328165054321, "learning_rate": 1.580038245868478e-05, "loss": 0.16530883312225342, "step": 65220 }, { "epoch": 0.32400186761769467, "grad_norm": 1.0785502195358276, "learning_rate": 1.5799071942753762e-05, "loss": 0.16588332653045654, "step": 65230 }, { "epoch": 0.3240515383010639, "grad_norm": 1.390322208404541, "learning_rate": 1.5797761276745095e-05, "loss": 0.15365631580352784, "step": 65240 }, { "epoch": 0.32410120898443323, "grad_norm": 2.32075572013855, "learning_rate": 1.57964504606927e-05, "loss": 0.17542582750320435, "step": 65250 }, { "epoch": 0.3241508796678025, "grad_norm": 1.16114342212677, "learning_rate": 1.5795139494630498e-05, "loss": 0.18385028839111328, "step": 65260 }, { "epoch": 0.32420055035117173, "grad_norm": 1.1959547996520996, "learning_rate": 1.5793828378592412e-05, "loss": 0.17172141075134278, "step": 65270 }, { "epoch": 0.324250221034541, "grad_norm": 0.8989747166633606, "learning_rate": 1.579251711261238e-05, "loss": 0.1751793622970581, "step": 65280 }, { "epoch": 0.32429989171791024, "grad_norm": 0.5747246146202087, "learning_rate": 1.5791205696724334e-05, "loss": 0.16910865306854247, "step": 65290 }, { "epoch": 0.3243495624012795, "grad_norm": 1.9047266244888306, "learning_rate": 1.5789894130962213e-05, "loss": 0.19028794765472412, "step": 65300 }, { "epoch": 0.3243992330846488, "grad_norm": 0.9193642735481262, "learning_rate": 1.578858241535996e-05, "loss": 0.17264997959136963, "step": 65310 }, { "epoch": 0.32444890376801805, "grad_norm": 0.7400586605072021, "learning_rate": 1.5787270549951525e-05, "loss": 0.16426774263381957, "step": 65320 }, { "epoch": 0.3244985744513873, "grad_norm": 0.6935449838638306, "learning_rate": 1.5785958534770846e-05, "loss": 0.1757894515991211, "step": 65330 }, { "epoch": 0.32454824513475655, "grad_norm": 1.2324590682983398, "learning_rate": 1.5784646369851895e-05, "loss": 0.1727684736251831, "step": 65340 }, { "epoch": 0.3245979158181258, "grad_norm": 0.9259801506996155, "learning_rate": 1.578333405522862e-05, "loss": 0.16438617706298828, "step": 65350 }, { "epoch": 0.3246475865014951, "grad_norm": 0.9883133172988892, "learning_rate": 1.578202159093498e-05, "loss": 0.16242382526397706, "step": 65360 }, { "epoch": 0.32469725718486436, "grad_norm": 1.865114450454712, "learning_rate": 1.578070897700495e-05, "loss": 0.17259669303894043, "step": 65370 }, { "epoch": 0.3247469278682336, "grad_norm": 0.6878585815429688, "learning_rate": 1.5779396213472495e-05, "loss": 0.17121124267578125, "step": 65380 }, { "epoch": 0.32479659855160287, "grad_norm": 1.7851442098617554, "learning_rate": 1.5778083300371592e-05, "loss": 0.1643339991569519, "step": 65390 }, { "epoch": 0.3248462692349721, "grad_norm": 2.785057306289673, "learning_rate": 1.5776770237736215e-05, "loss": 0.1721837878227234, "step": 65400 }, { "epoch": 0.32489593991834137, "grad_norm": 0.9299442768096924, "learning_rate": 1.577545702560035e-05, "loss": 0.17556722164154054, "step": 65410 }, { "epoch": 0.3249456106017107, "grad_norm": 1.1356121301651, "learning_rate": 1.5774143663997973e-05, "loss": 0.16913943290710448, "step": 65420 }, { "epoch": 0.32499528128507993, "grad_norm": 1.5400549173355103, "learning_rate": 1.5772830152963082e-05, "loss": 0.1683030605316162, "step": 65430 }, { "epoch": 0.3250449519684492, "grad_norm": 0.7994211912155151, "learning_rate": 1.577151649252967e-05, "loss": 0.17022356986999512, "step": 65440 }, { "epoch": 0.32509462265181843, "grad_norm": 0.9478556513786316, "learning_rate": 1.5770202682731733e-05, "loss": 0.17367641925811766, "step": 65450 }, { "epoch": 0.3251442933351877, "grad_norm": 1.3636338710784912, "learning_rate": 1.5768888723603267e-05, "loss": 0.17411189079284667, "step": 65460 }, { "epoch": 0.325193964018557, "grad_norm": 1.2078566551208496, "learning_rate": 1.576757461517828e-05, "loss": 0.16781203746795653, "step": 65470 }, { "epoch": 0.32524363470192624, "grad_norm": 0.9297191500663757, "learning_rate": 1.5766260357490786e-05, "loss": 0.16704897880554198, "step": 65480 }, { "epoch": 0.3252933053852955, "grad_norm": 1.0344862937927246, "learning_rate": 1.576494595057479e-05, "loss": 0.16758289337158203, "step": 65490 }, { "epoch": 0.32534297606866475, "grad_norm": 0.570398211479187, "learning_rate": 1.5763631394464308e-05, "loss": 0.14961813688278197, "step": 65500 }, { "epoch": 0.325392646752034, "grad_norm": 1.0176749229431152, "learning_rate": 1.5762316689193365e-05, "loss": 0.15546964406967162, "step": 65510 }, { "epoch": 0.32544231743540325, "grad_norm": 1.110871434211731, "learning_rate": 1.5761001834795984e-05, "loss": 0.19191246032714843, "step": 65520 }, { "epoch": 0.32549198811877256, "grad_norm": 1.3153581619262695, "learning_rate": 1.5759686831306187e-05, "loss": 0.1736067771911621, "step": 65530 }, { "epoch": 0.3255416588021418, "grad_norm": 1.8295783996582031, "learning_rate": 1.5758371678758016e-05, "loss": 0.17972015142440795, "step": 65540 }, { "epoch": 0.32559132948551106, "grad_norm": 1.6758075952529907, "learning_rate": 1.57570563771855e-05, "loss": 0.17559945583343506, "step": 65550 }, { "epoch": 0.3256410001688803, "grad_norm": 0.764062762260437, "learning_rate": 1.5755740926622682e-05, "loss": 0.1681585431098938, "step": 65560 }, { "epoch": 0.32569067085224956, "grad_norm": 1.3273036479949951, "learning_rate": 1.57544253271036e-05, "loss": 0.15989246368408203, "step": 65570 }, { "epoch": 0.32574034153561887, "grad_norm": 1.771315336227417, "learning_rate": 1.5753109578662306e-05, "loss": 0.16303155422210694, "step": 65580 }, { "epoch": 0.3257900122189881, "grad_norm": 2.145327091217041, "learning_rate": 1.5751793681332847e-05, "loss": 0.1805997371673584, "step": 65590 }, { "epoch": 0.3258396829023574, "grad_norm": 1.364104986190796, "learning_rate": 1.5750477635149283e-05, "loss": 0.17652559280395508, "step": 65600 }, { "epoch": 0.3258893535857266, "grad_norm": 0.5368257164955139, "learning_rate": 1.5749161440145667e-05, "loss": 0.16272940635681152, "step": 65610 }, { "epoch": 0.3259390242690959, "grad_norm": 0.8300871253013611, "learning_rate": 1.5747845096356068e-05, "loss": 0.16496858596801758, "step": 65620 }, { "epoch": 0.32598869495246513, "grad_norm": 1.7495228052139282, "learning_rate": 1.5746528603814547e-05, "loss": 0.17054448127746583, "step": 65630 }, { "epoch": 0.32603836563583444, "grad_norm": 0.8535418510437012, "learning_rate": 1.5745211962555176e-05, "loss": 0.1783911943435669, "step": 65640 }, { "epoch": 0.3260880363192037, "grad_norm": 1.1965636014938354, "learning_rate": 1.5743895172612033e-05, "loss": 0.1700376272201538, "step": 65650 }, { "epoch": 0.32613770700257294, "grad_norm": 0.8307692408561707, "learning_rate": 1.5742578234019185e-05, "loss": 0.1786676526069641, "step": 65660 }, { "epoch": 0.3261873776859422, "grad_norm": 1.010481834411621, "learning_rate": 1.574126114681073e-05, "loss": 0.16754568815231324, "step": 65670 }, { "epoch": 0.32623704836931144, "grad_norm": 1.6060582399368286, "learning_rate": 1.573994391102074e-05, "loss": 0.16931431293487548, "step": 65680 }, { "epoch": 0.32628671905268075, "grad_norm": 1.222503662109375, "learning_rate": 1.573862652668331e-05, "loss": 0.16098151206970215, "step": 65690 }, { "epoch": 0.32633638973605, "grad_norm": 0.9329134821891785, "learning_rate": 1.5737308993832533e-05, "loss": 0.18082573413848876, "step": 65700 }, { "epoch": 0.32638606041941925, "grad_norm": 0.9369011521339417, "learning_rate": 1.5735991312502504e-05, "loss": 0.17547038793563843, "step": 65710 }, { "epoch": 0.3264357311027885, "grad_norm": 1.7236918210983276, "learning_rate": 1.573467348272733e-05, "loss": 0.17051515579223633, "step": 65720 }, { "epoch": 0.32648540178615776, "grad_norm": 0.836394727230072, "learning_rate": 1.573335550454111e-05, "loss": 0.16360232830047608, "step": 65730 }, { "epoch": 0.326535072469527, "grad_norm": 1.7811793088912964, "learning_rate": 1.5732037377977955e-05, "loss": 0.1845768928527832, "step": 65740 }, { "epoch": 0.3265847431528963, "grad_norm": 0.7883219718933105, "learning_rate": 1.5730719103071977e-05, "loss": 0.18221626281738282, "step": 65750 }, { "epoch": 0.32663441383626557, "grad_norm": 0.6298045516014099, "learning_rate": 1.5729400679857296e-05, "loss": 0.16821757555007935, "step": 65760 }, { "epoch": 0.3266840845196348, "grad_norm": 1.4660652875900269, "learning_rate": 1.5728082108368027e-05, "loss": 0.1720731258392334, "step": 65770 }, { "epoch": 0.3267337552030041, "grad_norm": 1.2549400329589844, "learning_rate": 1.5726763388638295e-05, "loss": 0.16961052417755126, "step": 65780 }, { "epoch": 0.3267834258863733, "grad_norm": 0.6023831367492676, "learning_rate": 1.572544452070223e-05, "loss": 0.17671303749084472, "step": 65790 }, { "epoch": 0.32683309656974263, "grad_norm": 1.30884850025177, "learning_rate": 1.5724125504593965e-05, "loss": 0.14526631832122802, "step": 65800 }, { "epoch": 0.3268827672531119, "grad_norm": 1.600999355316162, "learning_rate": 1.572280634034763e-05, "loss": 0.1761171579360962, "step": 65810 }, { "epoch": 0.32693243793648114, "grad_norm": 0.7721785306930542, "learning_rate": 1.572148702799737e-05, "loss": 0.1724332809448242, "step": 65820 }, { "epoch": 0.3269821086198504, "grad_norm": 0.9156476259231567, "learning_rate": 1.5720167567577325e-05, "loss": 0.1640923261642456, "step": 65830 }, { "epoch": 0.32703177930321964, "grad_norm": 2.299116849899292, "learning_rate": 1.5718847959121645e-05, "loss": 0.1619175672531128, "step": 65840 }, { "epoch": 0.3270814499865889, "grad_norm": 1.2739542722702026, "learning_rate": 1.5717528202664477e-05, "loss": 0.16007575988769532, "step": 65850 }, { "epoch": 0.3271311206699582, "grad_norm": 0.8736029863357544, "learning_rate": 1.5716208298239978e-05, "loss": 0.15808234214782715, "step": 65860 }, { "epoch": 0.32718079135332745, "grad_norm": 1.1622562408447266, "learning_rate": 1.571488824588231e-05, "loss": 0.1693752646446228, "step": 65870 }, { "epoch": 0.3272304620366967, "grad_norm": 1.7666741609573364, "learning_rate": 1.571356804562563e-05, "loss": 0.18710885047912598, "step": 65880 }, { "epoch": 0.32728013272006595, "grad_norm": 1.5611220598220825, "learning_rate": 1.5712247697504107e-05, "loss": 0.1675663948059082, "step": 65890 }, { "epoch": 0.3273298034034352, "grad_norm": 0.7251526713371277, "learning_rate": 1.571092720155191e-05, "loss": 0.18674249649047853, "step": 65900 }, { "epoch": 0.3273794740868045, "grad_norm": 1.4121263027191162, "learning_rate": 1.5709606557803213e-05, "loss": 0.18746442794799806, "step": 65910 }, { "epoch": 0.32742914477017376, "grad_norm": 1.2737163305282593, "learning_rate": 1.5708285766292194e-05, "loss": 0.1665891170501709, "step": 65920 }, { "epoch": 0.327478815453543, "grad_norm": 0.8675220012664795, "learning_rate": 1.570696482705303e-05, "loss": 0.17673158645629883, "step": 65930 }, { "epoch": 0.32752848613691227, "grad_norm": 2.109532356262207, "learning_rate": 1.570564374011992e-05, "loss": 0.17963119745254516, "step": 65940 }, { "epoch": 0.3275781568202815, "grad_norm": 1.5800226926803589, "learning_rate": 1.570432250552704e-05, "loss": 0.16702656745910643, "step": 65950 }, { "epoch": 0.32762782750365077, "grad_norm": 0.8515076041221619, "learning_rate": 1.5703001123308584e-05, "loss": 0.16729671955108644, "step": 65960 }, { "epoch": 0.3276774981870201, "grad_norm": 1.0221240520477295, "learning_rate": 1.570167959349876e-05, "loss": 0.17789094448089598, "step": 65970 }, { "epoch": 0.32772716887038933, "grad_norm": 1.4952576160430908, "learning_rate": 1.5700357916131756e-05, "loss": 0.1589599370956421, "step": 65980 }, { "epoch": 0.3277768395537586, "grad_norm": 0.8240554928779602, "learning_rate": 1.569903609124178e-05, "loss": 0.16337149143218993, "step": 65990 }, { "epoch": 0.32782651023712783, "grad_norm": 0.8152599930763245, "learning_rate": 1.5697714118863038e-05, "loss": 0.16846954822540283, "step": 66000 }, { "epoch": 0.3278761809204971, "grad_norm": 1.1813421249389648, "learning_rate": 1.5696391999029753e-05, "loss": 0.1730731964111328, "step": 66010 }, { "epoch": 0.3279258516038664, "grad_norm": 0.7741640210151672, "learning_rate": 1.569506973177613e-05, "loss": 0.17509636878967286, "step": 66020 }, { "epoch": 0.32797552228723564, "grad_norm": 1.6132599115371704, "learning_rate": 1.569374731713639e-05, "loss": 0.1687857389450073, "step": 66030 }, { "epoch": 0.3280251929706049, "grad_norm": 2.733952522277832, "learning_rate": 1.569242475514476e-05, "loss": 0.17831904888153077, "step": 66040 }, { "epoch": 0.32807486365397415, "grad_norm": 1.1858686208724976, "learning_rate": 1.569110204583547e-05, "loss": 0.1771238088607788, "step": 66050 }, { "epoch": 0.3281245343373434, "grad_norm": 1.3428844213485718, "learning_rate": 1.5689779189242745e-05, "loss": 0.1691303014755249, "step": 66060 }, { "epoch": 0.32817420502071265, "grad_norm": 0.8632615208625793, "learning_rate": 1.568845618540082e-05, "loss": 0.1609526515007019, "step": 66070 }, { "epoch": 0.32822387570408196, "grad_norm": 1.9498074054718018, "learning_rate": 1.568713303434394e-05, "loss": 0.18725016117095947, "step": 66080 }, { "epoch": 0.3282735463874512, "grad_norm": 1.8118257522583008, "learning_rate": 1.5685809736106342e-05, "loss": 0.18417633771896363, "step": 66090 }, { "epoch": 0.32832321707082046, "grad_norm": 1.3406000137329102, "learning_rate": 1.5684486290722274e-05, "loss": 0.1907656669616699, "step": 66100 }, { "epoch": 0.3283728877541897, "grad_norm": 0.7131048440933228, "learning_rate": 1.5683162698225984e-05, "loss": 0.17144775390625, "step": 66110 }, { "epoch": 0.32842255843755896, "grad_norm": 1.680040717124939, "learning_rate": 1.568183895865173e-05, "loss": 0.1750854253768921, "step": 66120 }, { "epoch": 0.32847222912092827, "grad_norm": 0.9683616757392883, "learning_rate": 1.5680515072033773e-05, "loss": 0.17096508741378785, "step": 66130 }, { "epoch": 0.3285218998042975, "grad_norm": 0.8932437896728516, "learning_rate": 1.5679191038406364e-05, "loss": 0.16100237369537354, "step": 66140 }, { "epoch": 0.3285715704876668, "grad_norm": 1.863629698753357, "learning_rate": 1.5677866857803782e-05, "loss": 0.16734046936035157, "step": 66150 }, { "epoch": 0.328621241171036, "grad_norm": 1.4783921241760254, "learning_rate": 1.5676542530260284e-05, "loss": 0.16976675987243653, "step": 66160 }, { "epoch": 0.3286709118544053, "grad_norm": 1.3430770635604858, "learning_rate": 1.5675218055810145e-05, "loss": 0.15480068922042847, "step": 66170 }, { "epoch": 0.32872058253777453, "grad_norm": 0.7486820816993713, "learning_rate": 1.5673893434487648e-05, "loss": 0.1670823574066162, "step": 66180 }, { "epoch": 0.32877025322114384, "grad_norm": 1.272134780883789, "learning_rate": 1.567256866632707e-05, "loss": 0.16365406513214112, "step": 66190 }, { "epoch": 0.3288199239045131, "grad_norm": 0.9177366495132446, "learning_rate": 1.567124375136269e-05, "loss": 0.16918661594390869, "step": 66200 }, { "epoch": 0.32886959458788234, "grad_norm": 0.7528315186500549, "learning_rate": 1.566991868962881e-05, "loss": 0.17738854885101318, "step": 66210 }, { "epoch": 0.3289192652712516, "grad_norm": 1.3179763555526733, "learning_rate": 1.566859348115971e-05, "loss": 0.1744173288345337, "step": 66220 }, { "epoch": 0.32896893595462084, "grad_norm": 0.859894871711731, "learning_rate": 1.566726812598969e-05, "loss": 0.1877891778945923, "step": 66230 }, { "epoch": 0.32901860663799015, "grad_norm": 1.2698289155960083, "learning_rate": 1.5665942624153053e-05, "loss": 0.1652449607849121, "step": 66240 }, { "epoch": 0.3290682773213594, "grad_norm": 1.4730726480484009, "learning_rate": 1.56646169756841e-05, "loss": 0.18342410326004027, "step": 66250 }, { "epoch": 0.32911794800472866, "grad_norm": 0.9786694645881653, "learning_rate": 1.566329118061713e-05, "loss": 0.1719071388244629, "step": 66260 }, { "epoch": 0.3291676186880979, "grad_norm": 1.224453091621399, "learning_rate": 1.5661965238986464e-05, "loss": 0.16726338863372803, "step": 66270 }, { "epoch": 0.32921728937146716, "grad_norm": 2.2322280406951904, "learning_rate": 1.5660639150826417e-05, "loss": 0.1681395173072815, "step": 66280 }, { "epoch": 0.3292669600548364, "grad_norm": 1.347266435623169, "learning_rate": 1.5659312916171304e-05, "loss": 0.1707735538482666, "step": 66290 }, { "epoch": 0.3293166307382057, "grad_norm": 1.0741558074951172, "learning_rate": 1.5657986535055446e-05, "loss": 0.16928763389587403, "step": 66300 }, { "epoch": 0.32936630142157497, "grad_norm": 1.236619472503662, "learning_rate": 1.5656660007513175e-05, "loss": 0.17119379043579103, "step": 66310 }, { "epoch": 0.3294159721049442, "grad_norm": 0.6477698087692261, "learning_rate": 1.5655333333578816e-05, "loss": 0.17566543817520142, "step": 66320 }, { "epoch": 0.3294656427883135, "grad_norm": 1.4721366167068481, "learning_rate": 1.5654006513286705e-05, "loss": 0.1788334369659424, "step": 66330 }, { "epoch": 0.3295153134716827, "grad_norm": 0.5896267890930176, "learning_rate": 1.5652679546671177e-05, "loss": 0.15270792245864867, "step": 66340 }, { "epoch": 0.32956498415505203, "grad_norm": 1.3599984645843506, "learning_rate": 1.5651352433766574e-05, "loss": 0.16489338874816895, "step": 66350 }, { "epoch": 0.3296146548384213, "grad_norm": 1.1180564165115356, "learning_rate": 1.5650025174607246e-05, "loss": 0.16269123554229736, "step": 66360 }, { "epoch": 0.32966432552179054, "grad_norm": 0.9525423049926758, "learning_rate": 1.5648697769227536e-05, "loss": 0.1807015657424927, "step": 66370 }, { "epoch": 0.3297139962051598, "grad_norm": 1.0451385974884033, "learning_rate": 1.56473702176618e-05, "loss": 0.17140822410583495, "step": 66380 }, { "epoch": 0.32976366688852904, "grad_norm": 1.0174291133880615, "learning_rate": 1.5646042519944392e-05, "loss": 0.17741727828979492, "step": 66390 }, { "epoch": 0.3298133375718983, "grad_norm": 1.468590497970581, "learning_rate": 1.5644714676109675e-05, "loss": 0.19341133832931517, "step": 66400 }, { "epoch": 0.3298630082552676, "grad_norm": 1.8296420574188232, "learning_rate": 1.5643386686192014e-05, "loss": 0.18144071102142334, "step": 66410 }, { "epoch": 0.32991267893863685, "grad_norm": 0.9877969622612, "learning_rate": 1.564205855022577e-05, "loss": 0.1630777597427368, "step": 66420 }, { "epoch": 0.3299623496220061, "grad_norm": 1.3745867013931274, "learning_rate": 1.5640730268245322e-05, "loss": 0.17862935066223146, "step": 66430 }, { "epoch": 0.33001202030537535, "grad_norm": 1.9453896284103394, "learning_rate": 1.5639401840285043e-05, "loss": 0.17165566682815553, "step": 66440 }, { "epoch": 0.3300616909887446, "grad_norm": 1.3723821640014648, "learning_rate": 1.5638073266379307e-05, "loss": 0.1701735496520996, "step": 66450 }, { "epoch": 0.3301113616721139, "grad_norm": 0.5437566637992859, "learning_rate": 1.5636744546562506e-05, "loss": 0.1870563507080078, "step": 66460 }, { "epoch": 0.33016103235548316, "grad_norm": 0.9809777140617371, "learning_rate": 1.5635415680869024e-05, "loss": 0.1683150053024292, "step": 66470 }, { "epoch": 0.3302107030388524, "grad_norm": 0.8779525756835938, "learning_rate": 1.563408666933325e-05, "loss": 0.16348034143447876, "step": 66480 }, { "epoch": 0.33026037372222167, "grad_norm": 2.5638322830200195, "learning_rate": 1.563275751198958e-05, "loss": 0.17243525981903077, "step": 66490 }, { "epoch": 0.3303100444055909, "grad_norm": 1.1377501487731934, "learning_rate": 1.5631428208872406e-05, "loss": 0.1715885281562805, "step": 66500 }, { "epoch": 0.33035971508896017, "grad_norm": 1.0107756853103638, "learning_rate": 1.5630098760016134e-05, "loss": 0.16360883712768554, "step": 66510 }, { "epoch": 0.3304093857723295, "grad_norm": 1.9099607467651367, "learning_rate": 1.562876916545517e-05, "loss": 0.1707763671875, "step": 66520 }, { "epoch": 0.33045905645569873, "grad_norm": 1.3929142951965332, "learning_rate": 1.5627439425223928e-05, "loss": 0.17081038951873778, "step": 66530 }, { "epoch": 0.330508727139068, "grad_norm": 2.0827226638793945, "learning_rate": 1.5626109539356812e-05, "loss": 0.16741411685943602, "step": 66540 }, { "epoch": 0.33055839782243723, "grad_norm": 1.377451777458191, "learning_rate": 1.5624779507888242e-05, "loss": 0.1548544406890869, "step": 66550 }, { "epoch": 0.3306080685058065, "grad_norm": 0.9321688413619995, "learning_rate": 1.5623449330852642e-05, "loss": 0.15296380519866942, "step": 66560 }, { "epoch": 0.3306577391891758, "grad_norm": 2.631502151489258, "learning_rate": 1.5622119008284436e-05, "loss": 0.16462491750717162, "step": 66570 }, { "epoch": 0.33070740987254504, "grad_norm": 1.091463327407837, "learning_rate": 1.562078854021805e-05, "loss": 0.15572607517242432, "step": 66580 }, { "epoch": 0.3307570805559143, "grad_norm": 1.4781193733215332, "learning_rate": 1.5619457926687916e-05, "loss": 0.16382722854614257, "step": 66590 }, { "epoch": 0.33080675123928355, "grad_norm": 1.014365315437317, "learning_rate": 1.561812716772847e-05, "loss": 0.1594368577003479, "step": 66600 }, { "epoch": 0.3308564219226528, "grad_norm": 1.0489304065704346, "learning_rate": 1.5616796263374153e-05, "loss": 0.17027980089187622, "step": 66610 }, { "epoch": 0.33090609260602205, "grad_norm": 1.2092702388763428, "learning_rate": 1.5615465213659406e-05, "loss": 0.18565971851348878, "step": 66620 }, { "epoch": 0.33095576328939136, "grad_norm": 0.6572652459144592, "learning_rate": 1.561413401861868e-05, "loss": 0.1625051736831665, "step": 66630 }, { "epoch": 0.3310054339727606, "grad_norm": 1.4958443641662598, "learning_rate": 1.5612802678286416e-05, "loss": 0.17602710723876952, "step": 66640 }, { "epoch": 0.33105510465612986, "grad_norm": 0.795074999332428, "learning_rate": 1.561147119269708e-05, "loss": 0.1714427351951599, "step": 66650 }, { "epoch": 0.3311047753394991, "grad_norm": 2.1056644916534424, "learning_rate": 1.5610139561885128e-05, "loss": 0.16927707195281982, "step": 66660 }, { "epoch": 0.33115444602286837, "grad_norm": 1.0803532600402832, "learning_rate": 1.560880778588502e-05, "loss": 0.17172071933746338, "step": 66670 }, { "epoch": 0.3312041167062376, "grad_norm": 0.6656628251075745, "learning_rate": 1.5607475864731217e-05, "loss": 0.1572053074836731, "step": 66680 }, { "epoch": 0.3312537873896069, "grad_norm": 0.856049656867981, "learning_rate": 1.5606143798458197e-05, "loss": 0.17272505760192872, "step": 66690 }, { "epoch": 0.3313034580729762, "grad_norm": 1.199220061302185, "learning_rate": 1.5604811587100427e-05, "loss": 0.16885621547698976, "step": 66700 }, { "epoch": 0.3313531287563454, "grad_norm": 2.123598098754883, "learning_rate": 1.5603479230692388e-05, "loss": 0.16408917903900147, "step": 66710 }, { "epoch": 0.3314027994397147, "grad_norm": 1.1905031204223633, "learning_rate": 1.560214672926856e-05, "loss": 0.16036710739135743, "step": 66720 }, { "epoch": 0.33145247012308393, "grad_norm": 0.9886015057563782, "learning_rate": 1.5600814082863424e-05, "loss": 0.19553377628326415, "step": 66730 }, { "epoch": 0.33150214080645324, "grad_norm": 0.7284523248672485, "learning_rate": 1.5599481291511474e-05, "loss": 0.16042958498001098, "step": 66740 }, { "epoch": 0.3315518114898225, "grad_norm": 0.5670669674873352, "learning_rate": 1.55981483552472e-05, "loss": 0.15985050201416015, "step": 66750 }, { "epoch": 0.33160148217319174, "grad_norm": 0.9664447903633118, "learning_rate": 1.55968152741051e-05, "loss": 0.17151224613189697, "step": 66760 }, { "epoch": 0.331651152856561, "grad_norm": 0.7982685565948486, "learning_rate": 1.5595482048119668e-05, "loss": 0.18100621700286865, "step": 66770 }, { "epoch": 0.33170082353993025, "grad_norm": 1.0829122066497803, "learning_rate": 1.559414867732541e-05, "loss": 0.1629918098449707, "step": 66780 }, { "epoch": 0.3317504942232995, "grad_norm": 1.1291911602020264, "learning_rate": 1.5592815161756837e-05, "loss": 0.17082884311676025, "step": 66790 }, { "epoch": 0.3318001649066688, "grad_norm": 0.8438541889190674, "learning_rate": 1.559148150144845e-05, "loss": 0.18296821117401124, "step": 66800 }, { "epoch": 0.33184983559003806, "grad_norm": 0.7984480261802673, "learning_rate": 1.5590147696434775e-05, "loss": 0.17120299339294434, "step": 66810 }, { "epoch": 0.3318995062734073, "grad_norm": 1.146133303642273, "learning_rate": 1.5588813746750326e-05, "loss": 0.1677496314048767, "step": 66820 }, { "epoch": 0.33194917695677656, "grad_norm": 0.9791122078895569, "learning_rate": 1.5587479652429623e-05, "loss": 0.1609712839126587, "step": 66830 }, { "epoch": 0.3319988476401458, "grad_norm": 0.9615665078163147, "learning_rate": 1.558614541350719e-05, "loss": 0.15560327768325805, "step": 66840 }, { "epoch": 0.3320485183235151, "grad_norm": 1.2104880809783936, "learning_rate": 1.5584811030017567e-05, "loss": 0.1808272957801819, "step": 66850 }, { "epoch": 0.33209818900688437, "grad_norm": 1.1225810050964355, "learning_rate": 1.558347650199527e-05, "loss": 0.18370745182037354, "step": 66860 }, { "epoch": 0.3321478596902536, "grad_norm": 1.6321877241134644, "learning_rate": 1.5582141829474852e-05, "loss": 0.16623296737670898, "step": 66870 }, { "epoch": 0.3321975303736229, "grad_norm": 1.0125455856323242, "learning_rate": 1.558080701249085e-05, "loss": 0.17727296352386473, "step": 66880 }, { "epoch": 0.3322472010569921, "grad_norm": 1.161019206047058, "learning_rate": 1.5579472051077803e-05, "loss": 0.16511156558990478, "step": 66890 }, { "epoch": 0.3322968717403614, "grad_norm": 1.1968224048614502, "learning_rate": 1.5578136945270264e-05, "loss": 0.1694599747657776, "step": 66900 }, { "epoch": 0.3323465424237307, "grad_norm": 0.8881900906562805, "learning_rate": 1.5576801695102784e-05, "loss": 0.16671178340911866, "step": 66910 }, { "epoch": 0.33239621310709994, "grad_norm": 2.4353785514831543, "learning_rate": 1.5575466300609917e-05, "loss": 0.19094231128692626, "step": 66920 }, { "epoch": 0.3324458837904692, "grad_norm": 1.3297333717346191, "learning_rate": 1.5574130761826226e-05, "loss": 0.16381332874298096, "step": 66930 }, { "epoch": 0.33249555447383844, "grad_norm": 0.9539176225662231, "learning_rate": 1.5572795078786272e-05, "loss": 0.1615041136741638, "step": 66940 }, { "epoch": 0.3325452251572077, "grad_norm": 1.0692850351333618, "learning_rate": 1.557145925152462e-05, "loss": 0.1663776159286499, "step": 66950 }, { "epoch": 0.332594895840577, "grad_norm": 0.9836120009422302, "learning_rate": 1.5570123280075847e-05, "loss": 0.17246155738830565, "step": 66960 }, { "epoch": 0.33264456652394625, "grad_norm": 0.8711956739425659, "learning_rate": 1.556878716447452e-05, "loss": 0.16903338432312012, "step": 66970 }, { "epoch": 0.3326942372073155, "grad_norm": 0.8776695132255554, "learning_rate": 1.556745090475522e-05, "loss": 0.1840456485748291, "step": 66980 }, { "epoch": 0.33274390789068475, "grad_norm": 2.18656849861145, "learning_rate": 1.556611450095253e-05, "loss": 0.17603569030761718, "step": 66990 }, { "epoch": 0.332793578574054, "grad_norm": 1.054124355316162, "learning_rate": 1.5564777953101033e-05, "loss": 0.17475947141647338, "step": 67000 }, { "epoch": 0.33284324925742326, "grad_norm": 1.5735474824905396, "learning_rate": 1.5563441261235325e-05, "loss": 0.1711829423904419, "step": 67010 }, { "epoch": 0.33289291994079256, "grad_norm": 1.5212798118591309, "learning_rate": 1.556210442538999e-05, "loss": 0.16138889789581298, "step": 67020 }, { "epoch": 0.3329425906241618, "grad_norm": 1.6338908672332764, "learning_rate": 1.556076744559963e-05, "loss": 0.1908062219619751, "step": 67030 }, { "epoch": 0.33299226130753107, "grad_norm": 1.0454245805740356, "learning_rate": 1.5559430321898843e-05, "loss": 0.17794525623321533, "step": 67040 }, { "epoch": 0.3330419319909003, "grad_norm": 1.8727586269378662, "learning_rate": 1.5558093054322236e-05, "loss": 0.18026434183120726, "step": 67050 }, { "epoch": 0.33309160267426957, "grad_norm": 1.31748628616333, "learning_rate": 1.5556755642904414e-05, "loss": 0.16259175539016724, "step": 67060 }, { "epoch": 0.3331412733576389, "grad_norm": 1.1712440252304077, "learning_rate": 1.555541808767999e-05, "loss": 0.17285361289978027, "step": 67070 }, { "epoch": 0.33319094404100813, "grad_norm": 0.7475131154060364, "learning_rate": 1.5554080388683585e-05, "loss": 0.1555778741836548, "step": 67080 }, { "epoch": 0.3332406147243774, "grad_norm": 1.1524072885513306, "learning_rate": 1.5552742545949807e-05, "loss": 0.1644442319869995, "step": 67090 }, { "epoch": 0.33329028540774663, "grad_norm": 1.305001974105835, "learning_rate": 1.5551404559513284e-05, "loss": 0.1807658553123474, "step": 67100 }, { "epoch": 0.3333399560911159, "grad_norm": 1.265568494796753, "learning_rate": 1.5550066429408647e-05, "loss": 0.1624300718307495, "step": 67110 }, { "epoch": 0.33338962677448514, "grad_norm": 2.6084349155426025, "learning_rate": 1.554872815567052e-05, "loss": 0.1809173345565796, "step": 67120 }, { "epoch": 0.33343929745785444, "grad_norm": 0.6942108869552612, "learning_rate": 1.554738973833354e-05, "loss": 0.16630058288574218, "step": 67130 }, { "epoch": 0.3334889681412237, "grad_norm": 1.3953425884246826, "learning_rate": 1.5546051177432348e-05, "loss": 0.1764308214187622, "step": 67140 }, { "epoch": 0.33353863882459295, "grad_norm": 0.9200425744056702, "learning_rate": 1.5544712473001575e-05, "loss": 0.16343491077423095, "step": 67150 }, { "epoch": 0.3335883095079622, "grad_norm": 2.17537260055542, "learning_rate": 1.554337362507588e-05, "loss": 0.17684174776077272, "step": 67160 }, { "epoch": 0.33363798019133145, "grad_norm": 1.3044902086257935, "learning_rate": 1.55420346336899e-05, "loss": 0.16227940320968628, "step": 67170 }, { "epoch": 0.33368765087470076, "grad_norm": 2.1807360649108887, "learning_rate": 1.5540695498878287e-05, "loss": 0.17845864295959474, "step": 67180 }, { "epoch": 0.33373732155807, "grad_norm": 0.7369138598442078, "learning_rate": 1.553935622067571e-05, "loss": 0.16218109130859376, "step": 67190 }, { "epoch": 0.33378699224143926, "grad_norm": 2.1553616523742676, "learning_rate": 1.553801679911682e-05, "loss": 0.1735447883605957, "step": 67200 }, { "epoch": 0.3338366629248085, "grad_norm": 1.532259464263916, "learning_rate": 1.5536677234236276e-05, "loss": 0.18198106288909913, "step": 67210 }, { "epoch": 0.33388633360817777, "grad_norm": 2.204871892929077, "learning_rate": 1.553533752606876e-05, "loss": 0.1723780393600464, "step": 67220 }, { "epoch": 0.333936004291547, "grad_norm": 1.0355031490325928, "learning_rate": 1.5533997674648928e-05, "loss": 0.17559938430786132, "step": 67230 }, { "epoch": 0.3339856749749163, "grad_norm": 1.5072721242904663, "learning_rate": 1.5532657680011464e-05, "loss": 0.1702793598175049, "step": 67240 }, { "epoch": 0.3340353456582856, "grad_norm": 2.093156099319458, "learning_rate": 1.5531317542191046e-05, "loss": 0.1656000018119812, "step": 67250 }, { "epoch": 0.33408501634165483, "grad_norm": 0.9815784692764282, "learning_rate": 1.5529977261222348e-05, "loss": 0.17862706184387206, "step": 67260 }, { "epoch": 0.3341346870250241, "grad_norm": 1.3347676992416382, "learning_rate": 1.5528636837140067e-05, "loss": 0.17706407308578492, "step": 67270 }, { "epoch": 0.33418435770839333, "grad_norm": 1.5644656419754028, "learning_rate": 1.5527296269978886e-05, "loss": 0.16940052509307862, "step": 67280 }, { "epoch": 0.33423402839176264, "grad_norm": 0.8002362251281738, "learning_rate": 1.55259555597735e-05, "loss": 0.17639241218566895, "step": 67290 }, { "epoch": 0.3342836990751319, "grad_norm": 0.9605807662010193, "learning_rate": 1.5524614706558607e-05, "loss": 0.16529515981674195, "step": 67300 }, { "epoch": 0.33433336975850114, "grad_norm": 1.2314568758010864, "learning_rate": 1.5523273710368908e-05, "loss": 0.1577660322189331, "step": 67310 }, { "epoch": 0.3343830404418704, "grad_norm": 0.8827168941497803, "learning_rate": 1.5521932571239103e-05, "loss": 0.18803532123565675, "step": 67320 }, { "epoch": 0.33443271112523965, "grad_norm": 1.2456061840057373, "learning_rate": 1.5520591289203906e-05, "loss": 0.16586658954620362, "step": 67330 }, { "epoch": 0.3344823818086089, "grad_norm": 1.5933536291122437, "learning_rate": 1.5519249864298026e-05, "loss": 0.15913492441177368, "step": 67340 }, { "epoch": 0.3345320524919782, "grad_norm": 0.8518229722976685, "learning_rate": 1.5517908296556177e-05, "loss": 0.1589155077934265, "step": 67350 }, { "epoch": 0.33458172317534746, "grad_norm": 2.5191938877105713, "learning_rate": 1.5516566586013077e-05, "loss": 0.17980403900146485, "step": 67360 }, { "epoch": 0.3346313938587167, "grad_norm": 1.769090175628662, "learning_rate": 1.5515224732703458e-05, "loss": 0.17734110355377197, "step": 67370 }, { "epoch": 0.33468106454208596, "grad_norm": 1.7782145738601685, "learning_rate": 1.5513882736662035e-05, "loss": 0.17631633281707765, "step": 67380 }, { "epoch": 0.3347307352254552, "grad_norm": 1.0961453914642334, "learning_rate": 1.551254059792355e-05, "loss": 0.170823872089386, "step": 67390 }, { "epoch": 0.3347804059088245, "grad_norm": 0.572031557559967, "learning_rate": 1.5511198316522724e-05, "loss": 0.16734130382537843, "step": 67400 }, { "epoch": 0.33483007659219377, "grad_norm": 1.092560887336731, "learning_rate": 1.5509855892494306e-05, "loss": 0.17950268983840942, "step": 67410 }, { "epoch": 0.334879747275563, "grad_norm": 0.917008101940155, "learning_rate": 1.5508513325873034e-05, "loss": 0.16598153114318848, "step": 67420 }, { "epoch": 0.3349294179589323, "grad_norm": 1.2042577266693115, "learning_rate": 1.550717061669365e-05, "loss": 0.17077652215957642, "step": 67430 }, { "epoch": 0.3349790886423015, "grad_norm": 1.2920634746551514, "learning_rate": 1.5505827764990905e-05, "loss": 0.17956197261810303, "step": 67440 }, { "epoch": 0.3350287593256708, "grad_norm": 1.6516098976135254, "learning_rate": 1.5504484770799548e-05, "loss": 0.17584826946258544, "step": 67450 }, { "epoch": 0.3350784300090401, "grad_norm": 0.8546099066734314, "learning_rate": 1.5503141634154343e-05, "loss": 0.16311891078948976, "step": 67460 }, { "epoch": 0.33512810069240934, "grad_norm": 1.9021986722946167, "learning_rate": 1.5501798355090044e-05, "loss": 0.16706645488739014, "step": 67470 }, { "epoch": 0.3351777713757786, "grad_norm": 0.9327000379562378, "learning_rate": 1.5500454933641415e-05, "loss": 0.16745043992996217, "step": 67480 }, { "epoch": 0.33522744205914784, "grad_norm": 0.6180543899536133, "learning_rate": 1.5499111369843223e-05, "loss": 0.17527391910552978, "step": 67490 }, { "epoch": 0.3352771127425171, "grad_norm": 0.6552025079727173, "learning_rate": 1.5497767663730243e-05, "loss": 0.16496388912200927, "step": 67500 }, { "epoch": 0.3353267834258864, "grad_norm": 1.0718252658843994, "learning_rate": 1.549642381533724e-05, "loss": 0.17354379892349242, "step": 67510 }, { "epoch": 0.33537645410925565, "grad_norm": 1.6046425104141235, "learning_rate": 1.5495079824699005e-05, "loss": 0.17862489223480224, "step": 67520 }, { "epoch": 0.3354261247926249, "grad_norm": 2.927957773208618, "learning_rate": 1.5493735691850314e-05, "loss": 0.1869995713233948, "step": 67530 }, { "epoch": 0.33547579547599415, "grad_norm": 0.8152427673339844, "learning_rate": 1.549239141682595e-05, "loss": 0.17362509965896605, "step": 67540 }, { "epoch": 0.3355254661593634, "grad_norm": 0.7897310256958008, "learning_rate": 1.5491046999660702e-05, "loss": 0.1650907278060913, "step": 67550 }, { "epoch": 0.33557513684273266, "grad_norm": 1.039400339126587, "learning_rate": 1.548970244038937e-05, "loss": 0.17111607789993286, "step": 67560 }, { "epoch": 0.33562480752610196, "grad_norm": 2.296158790588379, "learning_rate": 1.5488357739046744e-05, "loss": 0.17072318792343139, "step": 67570 }, { "epoch": 0.3356744782094712, "grad_norm": 1.0133064985275269, "learning_rate": 1.5487012895667626e-05, "loss": 0.17533668279647827, "step": 67580 }, { "epoch": 0.33572414889284047, "grad_norm": 2.4370267391204834, "learning_rate": 1.5485667910286823e-05, "loss": 0.17191951274871825, "step": 67590 }, { "epoch": 0.3357738195762097, "grad_norm": 1.2391520738601685, "learning_rate": 1.548432278293914e-05, "loss": 0.1753375291824341, "step": 67600 }, { "epoch": 0.33582349025957897, "grad_norm": 0.9135033488273621, "learning_rate": 1.5482977513659385e-05, "loss": 0.1744300127029419, "step": 67610 }, { "epoch": 0.3358731609429483, "grad_norm": 1.6459712982177734, "learning_rate": 1.548163210248238e-05, "loss": 0.17448052167892455, "step": 67620 }, { "epoch": 0.33592283162631753, "grad_norm": 1.2946879863739014, "learning_rate": 1.5480286549442934e-05, "loss": 0.15914852619171144, "step": 67630 }, { "epoch": 0.3359725023096868, "grad_norm": 0.7911171317100525, "learning_rate": 1.5478940854575882e-05, "loss": 0.17688865661621095, "step": 67640 }, { "epoch": 0.33602217299305603, "grad_norm": 1.356058120727539, "learning_rate": 1.547759501791604e-05, "loss": 0.16914403438568115, "step": 67650 }, { "epoch": 0.3360718436764253, "grad_norm": 1.12788724899292, "learning_rate": 1.5476249039498244e-05, "loss": 0.1685702085494995, "step": 67660 }, { "epoch": 0.33612151435979454, "grad_norm": 0.543023407459259, "learning_rate": 1.547490291935733e-05, "loss": 0.1593549966812134, "step": 67670 }, { "epoch": 0.33617118504316384, "grad_norm": 0.684115469455719, "learning_rate": 1.5473556657528123e-05, "loss": 0.16964845657348632, "step": 67680 }, { "epoch": 0.3362208557265331, "grad_norm": 1.0783830881118774, "learning_rate": 1.547221025404547e-05, "loss": 0.1769333600997925, "step": 67690 }, { "epoch": 0.33627052640990235, "grad_norm": 0.9404957890510559, "learning_rate": 1.547086370894422e-05, "loss": 0.177970027923584, "step": 67700 }, { "epoch": 0.3363201970932716, "grad_norm": 1.0499670505523682, "learning_rate": 1.546951702225921e-05, "loss": 0.17785243988037108, "step": 67710 }, { "epoch": 0.33636986777664085, "grad_norm": 0.8055151104927063, "learning_rate": 1.5468170194025304e-05, "loss": 0.1627551555633545, "step": 67720 }, { "epoch": 0.33641953846001016, "grad_norm": 1.6952601671218872, "learning_rate": 1.5466823224277352e-05, "loss": 0.18198113441467284, "step": 67730 }, { "epoch": 0.3364692091433794, "grad_norm": 1.911497950553894, "learning_rate": 1.5465476113050212e-05, "loss": 0.18303724527359008, "step": 67740 }, { "epoch": 0.33651887982674866, "grad_norm": 1.92339026927948, "learning_rate": 1.5464128860378747e-05, "loss": 0.184332275390625, "step": 67750 }, { "epoch": 0.3365685505101179, "grad_norm": 1.0442427396774292, "learning_rate": 1.5462781466297826e-05, "loss": 0.16196435689926147, "step": 67760 }, { "epoch": 0.33661822119348717, "grad_norm": 1.267810344696045, "learning_rate": 1.5461433930842315e-05, "loss": 0.16677360534667968, "step": 67770 }, { "epoch": 0.3366678918768564, "grad_norm": 1.1047667264938354, "learning_rate": 1.546008625404709e-05, "loss": 0.16508629322052001, "step": 67780 }, { "epoch": 0.3367175625602257, "grad_norm": 1.5817469358444214, "learning_rate": 1.545873843594703e-05, "loss": 0.15241515636444092, "step": 67790 }, { "epoch": 0.336767233243595, "grad_norm": 1.440253496170044, "learning_rate": 1.5457390476577014e-05, "loss": 0.1903329849243164, "step": 67800 }, { "epoch": 0.33681690392696423, "grad_norm": 1.808556079864502, "learning_rate": 1.5456042375971925e-05, "loss": 0.16586660146713256, "step": 67810 }, { "epoch": 0.3368665746103335, "grad_norm": 1.699390172958374, "learning_rate": 1.5454694134166654e-05, "loss": 0.18650710582733154, "step": 67820 }, { "epoch": 0.33691624529370273, "grad_norm": 0.8002347946166992, "learning_rate": 1.5453345751196095e-05, "loss": 0.16550344228744507, "step": 67830 }, { "epoch": 0.33696591597707204, "grad_norm": 1.453012466430664, "learning_rate": 1.545199722709514e-05, "loss": 0.1701462149620056, "step": 67840 }, { "epoch": 0.3370155866604413, "grad_norm": 1.4941045045852661, "learning_rate": 1.5450648561898684e-05, "loss": 0.16513029336929322, "step": 67850 }, { "epoch": 0.33706525734381054, "grad_norm": 0.8104692101478577, "learning_rate": 1.5449299755641638e-05, "loss": 0.16121581792831421, "step": 67860 }, { "epoch": 0.3371149280271798, "grad_norm": 2.4420342445373535, "learning_rate": 1.5447950808358905e-05, "loss": 0.17310707569122313, "step": 67870 }, { "epoch": 0.33716459871054905, "grad_norm": 1.5394090414047241, "learning_rate": 1.5446601720085394e-05, "loss": 0.18724865913391114, "step": 67880 }, { "epoch": 0.3372142693939183, "grad_norm": 1.6752917766571045, "learning_rate": 1.5445252490856025e-05, "loss": 0.16261075735092162, "step": 67890 }, { "epoch": 0.3372639400772876, "grad_norm": 1.225274682044983, "learning_rate": 1.5443903120705706e-05, "loss": 0.16292064189910888, "step": 67900 }, { "epoch": 0.33731361076065686, "grad_norm": 1.3905130624771118, "learning_rate": 1.5442553609669367e-05, "loss": 0.17524524927139282, "step": 67910 }, { "epoch": 0.3373632814440261, "grad_norm": 0.8575513362884521, "learning_rate": 1.5441203957781924e-05, "loss": 0.1582878589630127, "step": 67920 }, { "epoch": 0.33741295212739536, "grad_norm": 1.617523193359375, "learning_rate": 1.5439854165078317e-05, "loss": 0.15990583896636962, "step": 67930 }, { "epoch": 0.3374626228107646, "grad_norm": 1.3547228574752808, "learning_rate": 1.5438504231593467e-05, "loss": 0.16481428146362304, "step": 67940 }, { "epoch": 0.3375122934941339, "grad_norm": 1.419689416885376, "learning_rate": 1.5437154157362316e-05, "loss": 0.16755356788635253, "step": 67950 }, { "epoch": 0.33756196417750317, "grad_norm": 0.8333219885826111, "learning_rate": 1.54358039424198e-05, "loss": 0.17261773347854614, "step": 67960 }, { "epoch": 0.3376116348608724, "grad_norm": 1.372201919555664, "learning_rate": 1.5434453586800862e-05, "loss": 0.17848103046417235, "step": 67970 }, { "epoch": 0.3376613055442417, "grad_norm": 0.8556905388832092, "learning_rate": 1.5433103090540454e-05, "loss": 0.15309144258499147, "step": 67980 }, { "epoch": 0.3377109762276109, "grad_norm": 0.7718077301979065, "learning_rate": 1.5431752453673522e-05, "loss": 0.15938732624053956, "step": 67990 }, { "epoch": 0.3377606469109802, "grad_norm": 1.2699394226074219, "learning_rate": 1.543040167623502e-05, "loss": 0.17046575546264647, "step": 68000 }, { "epoch": 0.3378103175943495, "grad_norm": 1.7246406078338623, "learning_rate": 1.5429050758259908e-05, "loss": 0.16319553852081298, "step": 68010 }, { "epoch": 0.33785998827771874, "grad_norm": 1.619484305381775, "learning_rate": 1.5427699699783142e-05, "loss": 0.16082067489624025, "step": 68020 }, { "epoch": 0.337909658961088, "grad_norm": 0.792341411113739, "learning_rate": 1.5426348500839692e-05, "loss": 0.18622112274169922, "step": 68030 }, { "epoch": 0.33795932964445724, "grad_norm": 3.1944572925567627, "learning_rate": 1.5424997161464525e-05, "loss": 0.17538833618164062, "step": 68040 }, { "epoch": 0.3380090003278265, "grad_norm": 1.123335361480713, "learning_rate": 1.5423645681692613e-05, "loss": 0.16660276651382447, "step": 68050 }, { "epoch": 0.3380586710111958, "grad_norm": 1.1072652339935303, "learning_rate": 1.542229406155893e-05, "loss": 0.16757969856262206, "step": 68060 }, { "epoch": 0.33810834169456505, "grad_norm": 1.3834738731384277, "learning_rate": 1.5420942301098456e-05, "loss": 0.16674277782440186, "step": 68070 }, { "epoch": 0.3381580123779343, "grad_norm": 0.8375718593597412, "learning_rate": 1.5419590400346175e-05, "loss": 0.17968924045562745, "step": 68080 }, { "epoch": 0.33820768306130355, "grad_norm": 0.6546170115470886, "learning_rate": 1.5418238359337077e-05, "loss": 0.15667707920074464, "step": 68090 }, { "epoch": 0.3382573537446728, "grad_norm": 0.9425382614135742, "learning_rate": 1.5416886178106146e-05, "loss": 0.1792985677719116, "step": 68100 }, { "epoch": 0.33830702442804206, "grad_norm": 1.053494930267334, "learning_rate": 1.541553385668838e-05, "loss": 0.16732821464538575, "step": 68110 }, { "epoch": 0.33835669511141137, "grad_norm": 1.3408770561218262, "learning_rate": 1.5414181395118776e-05, "loss": 0.17140953540802, "step": 68120 }, { "epoch": 0.3384063657947806, "grad_norm": 2.0502264499664307, "learning_rate": 1.5412828793432332e-05, "loss": 0.15906479358673095, "step": 68130 }, { "epoch": 0.33845603647814987, "grad_norm": 1.1656379699707031, "learning_rate": 1.5411476051664057e-05, "loss": 0.16534852981567383, "step": 68140 }, { "epoch": 0.3385057071615191, "grad_norm": 1.4664275646209717, "learning_rate": 1.5410123169848957e-05, "loss": 0.15578992366790773, "step": 68150 }, { "epoch": 0.33855537784488837, "grad_norm": 0.9123203158378601, "learning_rate": 1.5408770148022042e-05, "loss": 0.18012728691101074, "step": 68160 }, { "epoch": 0.3386050485282577, "grad_norm": 1.4503741264343262, "learning_rate": 1.5407416986218334e-05, "loss": 0.1634439468383789, "step": 68170 }, { "epoch": 0.33865471921162693, "grad_norm": 0.8277409672737122, "learning_rate": 1.5406063684472847e-05, "loss": 0.16779298782348634, "step": 68180 }, { "epoch": 0.3387043898949962, "grad_norm": 1.8041162490844727, "learning_rate": 1.5404710242820608e-05, "loss": 0.16523528099060059, "step": 68190 }, { "epoch": 0.33875406057836543, "grad_norm": 1.611696481704712, "learning_rate": 1.540335666129664e-05, "loss": 0.17160823345184326, "step": 68200 }, { "epoch": 0.3388037312617347, "grad_norm": 1.2108205556869507, "learning_rate": 1.5402002939935973e-05, "loss": 0.16901865005493164, "step": 68210 }, { "epoch": 0.33885340194510394, "grad_norm": 1.928866982460022, "learning_rate": 1.540064907877364e-05, "loss": 0.17989858388900756, "step": 68220 }, { "epoch": 0.33890307262847325, "grad_norm": 2.1397104263305664, "learning_rate": 1.5399295077844682e-05, "loss": 0.17868375778198242, "step": 68230 }, { "epoch": 0.3389527433118425, "grad_norm": 2.176663637161255, "learning_rate": 1.539794093718414e-05, "loss": 0.18982744216918945, "step": 68240 }, { "epoch": 0.33900241399521175, "grad_norm": 0.599455714225769, "learning_rate": 1.5396586656827057e-05, "loss": 0.16744914054870605, "step": 68250 }, { "epoch": 0.339052084678581, "grad_norm": 1.8431123495101929, "learning_rate": 1.5395232236808483e-05, "loss": 0.1741083264350891, "step": 68260 }, { "epoch": 0.33910175536195025, "grad_norm": 1.1509027481079102, "learning_rate": 1.5393877677163463e-05, "loss": 0.1737603187561035, "step": 68270 }, { "epoch": 0.3391514260453195, "grad_norm": 2.058096170425415, "learning_rate": 1.5392522977927062e-05, "loss": 0.16754299402236938, "step": 68280 }, { "epoch": 0.3392010967286888, "grad_norm": 2.1151349544525146, "learning_rate": 1.539116813913433e-05, "loss": 0.17481545209884644, "step": 68290 }, { "epoch": 0.33925076741205806, "grad_norm": 1.4622080326080322, "learning_rate": 1.5389813160820337e-05, "loss": 0.17531657218933105, "step": 68300 }, { "epoch": 0.3393004380954273, "grad_norm": 1.304013729095459, "learning_rate": 1.5388458043020145e-05, "loss": 0.17255160808563233, "step": 68310 }, { "epoch": 0.33935010877879657, "grad_norm": 1.2649941444396973, "learning_rate": 1.5387102785768825e-05, "loss": 0.16095458269119262, "step": 68320 }, { "epoch": 0.3393997794621658, "grad_norm": 1.4177601337432861, "learning_rate": 1.5385747389101457e-05, "loss": 0.16892145872116088, "step": 68330 }, { "epoch": 0.3394494501455351, "grad_norm": 0.7761494517326355, "learning_rate": 1.5384391853053105e-05, "loss": 0.17827210426330567, "step": 68340 }, { "epoch": 0.3394991208289044, "grad_norm": 0.8233837485313416, "learning_rate": 1.5383036177658862e-05, "loss": 0.16246461868286133, "step": 68350 }, { "epoch": 0.33954879151227363, "grad_norm": 1.373209834098816, "learning_rate": 1.5381680362953805e-05, "loss": 0.16843039989471437, "step": 68360 }, { "epoch": 0.3395984621956429, "grad_norm": 0.9146872162818909, "learning_rate": 1.538032440897302e-05, "loss": 0.1656368374824524, "step": 68370 }, { "epoch": 0.33964813287901213, "grad_norm": 1.6825851202011108, "learning_rate": 1.5378968315751608e-05, "loss": 0.16524732112884521, "step": 68380 }, { "epoch": 0.3396978035623814, "grad_norm": 0.919169008731842, "learning_rate": 1.5377612083324657e-05, "loss": 0.16460232734680175, "step": 68390 }, { "epoch": 0.3397474742457507, "grad_norm": 1.3852709531784058, "learning_rate": 1.5376255711727268e-05, "loss": 0.17040042877197265, "step": 68400 }, { "epoch": 0.33979714492911994, "grad_norm": 2.653866767883301, "learning_rate": 1.5374899200994543e-05, "loss": 0.18189419507980348, "step": 68410 }, { "epoch": 0.3398468156124892, "grad_norm": 1.1069802045822144, "learning_rate": 1.537354255116159e-05, "loss": 0.17861422300338745, "step": 68420 }, { "epoch": 0.33989648629585845, "grad_norm": 0.8371444344520569, "learning_rate": 1.5372185762263513e-05, "loss": 0.17236608266830444, "step": 68430 }, { "epoch": 0.3399461569792277, "grad_norm": 0.9685602188110352, "learning_rate": 1.5370828834335427e-05, "loss": 0.17393941879272462, "step": 68440 }, { "epoch": 0.339995827662597, "grad_norm": 1.3261421918869019, "learning_rate": 1.5369471767412453e-05, "loss": 0.16549643278121948, "step": 68450 }, { "epoch": 0.34004549834596626, "grad_norm": 1.3886381387710571, "learning_rate": 1.536811456152971e-05, "loss": 0.1788393020629883, "step": 68460 }, { "epoch": 0.3400951690293355, "grad_norm": 0.9412185549736023, "learning_rate": 1.5366757216722316e-05, "loss": 0.16573342084884643, "step": 68470 }, { "epoch": 0.34014483971270476, "grad_norm": 2.18678617477417, "learning_rate": 1.5365399733025407e-05, "loss": 0.1670656681060791, "step": 68480 }, { "epoch": 0.340194510396074, "grad_norm": 1.6714040040969849, "learning_rate": 1.536404211047411e-05, "loss": 0.16901834011077882, "step": 68490 }, { "epoch": 0.34024418107944326, "grad_norm": 0.714405357837677, "learning_rate": 1.536268434910356e-05, "loss": 0.16380717754364013, "step": 68500 }, { "epoch": 0.34029385176281257, "grad_norm": 1.517745018005371, "learning_rate": 1.536132644894889e-05, "loss": 0.17590029239654542, "step": 68510 }, { "epoch": 0.3403435224461818, "grad_norm": 0.6977014541625977, "learning_rate": 1.5359968410045256e-05, "loss": 0.1760404109954834, "step": 68520 }, { "epoch": 0.3403931931295511, "grad_norm": 1.3139138221740723, "learning_rate": 1.5358610232427792e-05, "loss": 0.16201616525650026, "step": 68530 }, { "epoch": 0.3404428638129203, "grad_norm": 0.7666942477226257, "learning_rate": 1.5357251916131648e-05, "loss": 0.15602197647094726, "step": 68540 }, { "epoch": 0.3404925344962896, "grad_norm": 0.952971875667572, "learning_rate": 1.5355893461191978e-05, "loss": 0.15440144538879394, "step": 68550 }, { "epoch": 0.3405422051796589, "grad_norm": 1.021562933921814, "learning_rate": 1.535453486764394e-05, "loss": 0.1783376693725586, "step": 68560 }, { "epoch": 0.34059187586302814, "grad_norm": 1.657357096672058, "learning_rate": 1.535317613552269e-05, "loss": 0.1704526424407959, "step": 68570 }, { "epoch": 0.3406415465463974, "grad_norm": 0.8969232439994812, "learning_rate": 1.53518172648634e-05, "loss": 0.1698026180267334, "step": 68580 }, { "epoch": 0.34069121722976664, "grad_norm": 1.2617740631103516, "learning_rate": 1.5350458255701225e-05, "loss": 0.15365972518920898, "step": 68590 }, { "epoch": 0.3407408879131359, "grad_norm": 3.3353939056396484, "learning_rate": 1.534909910807134e-05, "loss": 0.16940815448760987, "step": 68600 }, { "epoch": 0.34079055859650514, "grad_norm": 0.9941749572753906, "learning_rate": 1.5347739822008926e-05, "loss": 0.16665234565734863, "step": 68610 }, { "epoch": 0.34084022927987445, "grad_norm": 0.8748244643211365, "learning_rate": 1.5346380397549155e-05, "loss": 0.16533970832824707, "step": 68620 }, { "epoch": 0.3408898999632437, "grad_norm": 0.7959000468254089, "learning_rate": 1.534502083472721e-05, "loss": 0.17803809642791749, "step": 68630 }, { "epoch": 0.34093957064661295, "grad_norm": 1.6029866933822632, "learning_rate": 1.534366113357827e-05, "loss": 0.1739845871925354, "step": 68640 }, { "epoch": 0.3409892413299822, "grad_norm": 3.002040386199951, "learning_rate": 1.534230129413753e-05, "loss": 0.16956989765167235, "step": 68650 }, { "epoch": 0.34103891201335146, "grad_norm": 0.8380971550941467, "learning_rate": 1.5340941316440186e-05, "loss": 0.16356366872787476, "step": 68660 }, { "epoch": 0.34108858269672077, "grad_norm": 1.7454767227172852, "learning_rate": 1.5339581200521423e-05, "loss": 0.16029003858566285, "step": 68670 }, { "epoch": 0.34113825338009, "grad_norm": 0.8005034327507019, "learning_rate": 1.533822094641645e-05, "loss": 0.16831809282302856, "step": 68680 }, { "epoch": 0.34118792406345927, "grad_norm": 1.0100493431091309, "learning_rate": 1.5336860554160463e-05, "loss": 0.1699275016784668, "step": 68690 }, { "epoch": 0.3412375947468285, "grad_norm": 1.8792012929916382, "learning_rate": 1.533550002378867e-05, "loss": 0.18750253915786744, "step": 68700 }, { "epoch": 0.3412872654301978, "grad_norm": 0.9258570671081543, "learning_rate": 1.533413935533628e-05, "loss": 0.16313775777816772, "step": 68710 }, { "epoch": 0.341336936113567, "grad_norm": 1.096318244934082, "learning_rate": 1.5332778548838508e-05, "loss": 0.16499464511871337, "step": 68720 }, { "epoch": 0.34138660679693633, "grad_norm": 2.370313882827759, "learning_rate": 1.5331417604330578e-05, "loss": 0.15903770923614502, "step": 68730 }, { "epoch": 0.3414362774803056, "grad_norm": 1.5646306276321411, "learning_rate": 1.5330056521847695e-05, "loss": 0.1564194679260254, "step": 68740 }, { "epoch": 0.34148594816367484, "grad_norm": 1.6581417322158813, "learning_rate": 1.5328695301425098e-05, "loss": 0.17811830043792726, "step": 68750 }, { "epoch": 0.3415356188470441, "grad_norm": 1.2356066703796387, "learning_rate": 1.5327333943098008e-05, "loss": 0.1612704038619995, "step": 68760 }, { "epoch": 0.34158528953041334, "grad_norm": 2.012160062789917, "learning_rate": 1.5325972446901656e-05, "loss": 0.16848959922790527, "step": 68770 }, { "epoch": 0.34163496021378265, "grad_norm": 1.3108171224594116, "learning_rate": 1.532461081287128e-05, "loss": 0.1564894437789917, "step": 68780 }, { "epoch": 0.3416846308971519, "grad_norm": 1.0123798847198486, "learning_rate": 1.5323249041042117e-05, "loss": 0.16117513179779053, "step": 68790 }, { "epoch": 0.34173430158052115, "grad_norm": 0.9747087955474854, "learning_rate": 1.532188713144941e-05, "loss": 0.15691503286361694, "step": 68800 }, { "epoch": 0.3417839722638904, "grad_norm": 1.759521484375, "learning_rate": 1.5320525084128405e-05, "loss": 0.16737515926361085, "step": 68810 }, { "epoch": 0.34183364294725965, "grad_norm": 1.1145520210266113, "learning_rate": 1.5319162899114347e-05, "loss": 0.17210302352905274, "step": 68820 }, { "epoch": 0.3418833136306289, "grad_norm": 2.3294079303741455, "learning_rate": 1.5317800576442493e-05, "loss": 0.18121068477630614, "step": 68830 }, { "epoch": 0.3419329843139982, "grad_norm": 1.257663369178772, "learning_rate": 1.53164381161481e-05, "loss": 0.16584210395812987, "step": 68840 }, { "epoch": 0.34198265499736746, "grad_norm": 1.1208772659301758, "learning_rate": 1.5315075518266426e-05, "loss": 0.17131242752075196, "step": 68850 }, { "epoch": 0.3420323256807367, "grad_norm": 1.7958928346633911, "learning_rate": 1.5313712782832734e-05, "loss": 0.1823744773864746, "step": 68860 }, { "epoch": 0.34208199636410597, "grad_norm": 0.8049215078353882, "learning_rate": 1.531234990988229e-05, "loss": 0.17331790924072266, "step": 68870 }, { "epoch": 0.3421316670474752, "grad_norm": 0.6975892782211304, "learning_rate": 1.531098689945037e-05, "loss": 0.15916931629180908, "step": 68880 }, { "epoch": 0.3421813377308445, "grad_norm": 1.2469764947891235, "learning_rate": 1.5309623751572242e-05, "loss": 0.17202479839324952, "step": 68890 }, { "epoch": 0.3422310084142138, "grad_norm": 0.9741374254226685, "learning_rate": 1.5308260466283185e-05, "loss": 0.16680158376693727, "step": 68900 }, { "epoch": 0.34228067909758303, "grad_norm": 1.0376946926116943, "learning_rate": 1.5306897043618483e-05, "loss": 0.18628368377685547, "step": 68910 }, { "epoch": 0.3423303497809523, "grad_norm": 1.4586273431777954, "learning_rate": 1.530553348361342e-05, "loss": 0.16646690368652345, "step": 68920 }, { "epoch": 0.34238002046432153, "grad_norm": 0.8086117506027222, "learning_rate": 1.5304169786303284e-05, "loss": 0.17019014358520507, "step": 68930 }, { "epoch": 0.3424296911476908, "grad_norm": 1.0816171169281006, "learning_rate": 1.5302805951723362e-05, "loss": 0.16239277124404908, "step": 68940 }, { "epoch": 0.3424793618310601, "grad_norm": 0.9791879653930664, "learning_rate": 1.530144197990896e-05, "loss": 0.1709580898284912, "step": 68950 }, { "epoch": 0.34252903251442934, "grad_norm": 1.3591605424880981, "learning_rate": 1.5300077870895366e-05, "loss": 0.17458754777908325, "step": 68960 }, { "epoch": 0.3425787031977986, "grad_norm": 0.847284197807312, "learning_rate": 1.529871362471789e-05, "loss": 0.17069858312606812, "step": 68970 }, { "epoch": 0.34262837388116785, "grad_norm": 1.0935723781585693, "learning_rate": 1.5297349241411838e-05, "loss": 0.1588205099105835, "step": 68980 }, { "epoch": 0.3426780445645371, "grad_norm": 0.6008445620536804, "learning_rate": 1.5295984721012517e-05, "loss": 0.16946816444396973, "step": 68990 }, { "epoch": 0.3427277152479064, "grad_norm": 1.25385582447052, "learning_rate": 1.529462006355524e-05, "loss": 0.16343533992767334, "step": 69000 }, { "epoch": 0.34277738593127566, "grad_norm": 1.4327185153961182, "learning_rate": 1.5293255269075325e-05, "loss": 0.16754504442214965, "step": 69010 }, { "epoch": 0.3428270566146449, "grad_norm": 1.970189094543457, "learning_rate": 1.529189033760809e-05, "loss": 0.17864153385162354, "step": 69020 }, { "epoch": 0.34287672729801416, "grad_norm": 1.7489523887634277, "learning_rate": 1.5290525269188863e-05, "loss": 0.15898120403289795, "step": 69030 }, { "epoch": 0.3429263979813834, "grad_norm": 0.9086447954177856, "learning_rate": 1.528916006385297e-05, "loss": 0.1592472553253174, "step": 69040 }, { "epoch": 0.34297606866475266, "grad_norm": 1.735050082206726, "learning_rate": 1.5287794721635737e-05, "loss": 0.17624328136444092, "step": 69050 }, { "epoch": 0.34302573934812197, "grad_norm": 0.7953007221221924, "learning_rate": 1.528642924257251e-05, "loss": 0.17763843536376953, "step": 69060 }, { "epoch": 0.3430754100314912, "grad_norm": 1.4169023036956787, "learning_rate": 1.528506362669861e-05, "loss": 0.17340223789215087, "step": 69070 }, { "epoch": 0.3431250807148605, "grad_norm": 1.0073484182357788, "learning_rate": 1.5283697874049398e-05, "loss": 0.169738507270813, "step": 69080 }, { "epoch": 0.3431747513982297, "grad_norm": 1.4923505783081055, "learning_rate": 1.5282331984660205e-05, "loss": 0.18209927082061766, "step": 69090 }, { "epoch": 0.343224422081599, "grad_norm": 1.3732051849365234, "learning_rate": 1.5280965958566386e-05, "loss": 0.17235685586929322, "step": 69100 }, { "epoch": 0.3432740927649683, "grad_norm": 0.6667981743812561, "learning_rate": 1.5279599795803286e-05, "loss": 0.16319226026535033, "step": 69110 }, { "epoch": 0.34332376344833754, "grad_norm": 1.3302088975906372, "learning_rate": 1.527823349640627e-05, "loss": 0.15742472410202027, "step": 69120 }, { "epoch": 0.3433734341317068, "grad_norm": 1.5498114824295044, "learning_rate": 1.5276867060410694e-05, "loss": 0.16646168231964112, "step": 69130 }, { "epoch": 0.34342310481507604, "grad_norm": 1.4792126417160034, "learning_rate": 1.527550048785192e-05, "loss": 0.17486135959625243, "step": 69140 }, { "epoch": 0.3434727754984453, "grad_norm": 1.0883543491363525, "learning_rate": 1.527413377876531e-05, "loss": 0.16811519861221313, "step": 69150 }, { "epoch": 0.34352244618181454, "grad_norm": 0.7082576155662537, "learning_rate": 1.5272766933186245e-05, "loss": 0.15540895462036133, "step": 69160 }, { "epoch": 0.34357211686518385, "grad_norm": 1.4016902446746826, "learning_rate": 1.5271399951150093e-05, "loss": 0.18436133861541748, "step": 69170 }, { "epoch": 0.3436217875485531, "grad_norm": 1.1694202423095703, "learning_rate": 1.527003283269222e-05, "loss": 0.1615062952041626, "step": 69180 }, { "epoch": 0.34367145823192236, "grad_norm": 1.7546920776367188, "learning_rate": 1.5268665577848024e-05, "loss": 0.17586882114410402, "step": 69190 }, { "epoch": 0.3437211289152916, "grad_norm": 2.0733346939086914, "learning_rate": 1.526729818665288e-05, "loss": 0.1645609974861145, "step": 69200 }, { "epoch": 0.34377079959866086, "grad_norm": 0.6302093267440796, "learning_rate": 1.526593065914218e-05, "loss": 0.17261626720428466, "step": 69210 }, { "epoch": 0.34382047028203017, "grad_norm": 1.292657732963562, "learning_rate": 1.526456299535131e-05, "loss": 0.17870125770568848, "step": 69220 }, { "epoch": 0.3438701409653994, "grad_norm": 0.9383437633514404, "learning_rate": 1.5263195195315663e-05, "loss": 0.16254501342773436, "step": 69230 }, { "epoch": 0.34391981164876867, "grad_norm": 2.468928575515747, "learning_rate": 1.5261827259070642e-05, "loss": 0.17323098182678223, "step": 69240 }, { "epoch": 0.3439694823321379, "grad_norm": 2.4411418437957764, "learning_rate": 1.5260459186651647e-05, "loss": 0.17596421241760254, "step": 69250 }, { "epoch": 0.3440191530155072, "grad_norm": 2.8483364582061768, "learning_rate": 1.5259090978094088e-05, "loss": 0.17906930446624755, "step": 69260 }, { "epoch": 0.3440688236988764, "grad_norm": 1.301134467124939, "learning_rate": 1.5257722633433362e-05, "loss": 0.16871769428253175, "step": 69270 }, { "epoch": 0.34411849438224573, "grad_norm": 0.981099009513855, "learning_rate": 1.5256354152704891e-05, "loss": 0.16205909252166747, "step": 69280 }, { "epoch": 0.344168165065615, "grad_norm": 0.8798121809959412, "learning_rate": 1.525498553594409e-05, "loss": 0.16260457038879395, "step": 69290 }, { "epoch": 0.34421783574898424, "grad_norm": 1.3551627397537231, "learning_rate": 1.5253616783186376e-05, "loss": 0.16907932758331298, "step": 69300 }, { "epoch": 0.3442675064323535, "grad_norm": 1.0320016145706177, "learning_rate": 1.5252247894467171e-05, "loss": 0.1589113235473633, "step": 69310 }, { "epoch": 0.34431717711572274, "grad_norm": 0.6373689770698547, "learning_rate": 1.5250878869821901e-05, "loss": 0.17300870418548583, "step": 69320 }, { "epoch": 0.34436684779909205, "grad_norm": 0.6910044550895691, "learning_rate": 1.5249509709286e-05, "loss": 0.16138300895690919, "step": 69330 }, { "epoch": 0.3444165184824613, "grad_norm": 1.4605095386505127, "learning_rate": 1.5248140412894894e-05, "loss": 0.16687388420104982, "step": 69340 }, { "epoch": 0.34446618916583055, "grad_norm": 0.8007649779319763, "learning_rate": 1.5246770980684029e-05, "loss": 0.16178615093231202, "step": 69350 }, { "epoch": 0.3445158598491998, "grad_norm": 1.5059058666229248, "learning_rate": 1.5245401412688839e-05, "loss": 0.16405136585235597, "step": 69360 }, { "epoch": 0.34456553053256905, "grad_norm": 1.014189600944519, "learning_rate": 1.5244031708944769e-05, "loss": 0.1716130018234253, "step": 69370 }, { "epoch": 0.3446152012159383, "grad_norm": 1.767537236213684, "learning_rate": 1.5242661869487268e-05, "loss": 0.17297835350036622, "step": 69380 }, { "epoch": 0.3446648718993076, "grad_norm": 1.2946749925613403, "learning_rate": 1.5241291894351783e-05, "loss": 0.17370835542678834, "step": 69390 }, { "epoch": 0.34471454258267686, "grad_norm": 1.4615710973739624, "learning_rate": 1.5239921783573775e-05, "loss": 0.17550418376922608, "step": 69400 }, { "epoch": 0.3447642132660461, "grad_norm": 1.0254945755004883, "learning_rate": 1.5238551537188694e-05, "loss": 0.16236613988876342, "step": 69410 }, { "epoch": 0.34481388394941537, "grad_norm": 1.0687898397445679, "learning_rate": 1.523718115523201e-05, "loss": 0.18072116374969482, "step": 69420 }, { "epoch": 0.3448635546327846, "grad_norm": 1.0847022533416748, "learning_rate": 1.5235810637739183e-05, "loss": 0.1746422290802002, "step": 69430 }, { "epoch": 0.3449132253161539, "grad_norm": 0.6567370295524597, "learning_rate": 1.5234439984745679e-05, "loss": 0.1558220624923706, "step": 69440 }, { "epoch": 0.3449628959995232, "grad_norm": 1.0391169786453247, "learning_rate": 1.5233069196286972e-05, "loss": 0.15950353145599366, "step": 69450 }, { "epoch": 0.34501256668289243, "grad_norm": 0.8824277520179749, "learning_rate": 1.5231698272398543e-05, "loss": 0.16945335865020753, "step": 69460 }, { "epoch": 0.3450622373662617, "grad_norm": 1.865982174873352, "learning_rate": 1.5230327213115862e-05, "loss": 0.18121315240859986, "step": 69470 }, { "epoch": 0.34511190804963093, "grad_norm": 1.4203482866287231, "learning_rate": 1.522895601847442e-05, "loss": 0.16918809413909913, "step": 69480 }, { "epoch": 0.3451615787330002, "grad_norm": 2.738604784011841, "learning_rate": 1.5227584688509694e-05, "loss": 0.1815786600112915, "step": 69490 }, { "epoch": 0.3452112494163695, "grad_norm": 2.554661512374878, "learning_rate": 1.5226213223257182e-05, "loss": 0.17725813388824463, "step": 69500 }, { "epoch": 0.34526092009973874, "grad_norm": 1.0160356760025024, "learning_rate": 1.5224841622752373e-05, "loss": 0.16747721433639526, "step": 69510 }, { "epoch": 0.345310590783108, "grad_norm": 0.9205919504165649, "learning_rate": 1.5223469887030759e-05, "loss": 0.16291193962097167, "step": 69520 }, { "epoch": 0.34536026146647725, "grad_norm": 0.9520564079284668, "learning_rate": 1.5222098016127849e-05, "loss": 0.16680376529693602, "step": 69530 }, { "epoch": 0.3454099321498465, "grad_norm": 0.94159334897995, "learning_rate": 1.5220726010079142e-05, "loss": 0.16631782054901123, "step": 69540 }, { "epoch": 0.3454596028332158, "grad_norm": 2.2211248874664307, "learning_rate": 1.5219353868920147e-05, "loss": 0.17355563640594482, "step": 69550 }, { "epoch": 0.34550927351658506, "grad_norm": 0.8285861611366272, "learning_rate": 1.5217981592686367e-05, "loss": 0.17120468616485596, "step": 69560 }, { "epoch": 0.3455589441999543, "grad_norm": 0.9463038444519043, "learning_rate": 1.5216609181413324e-05, "loss": 0.16634334325790406, "step": 69570 }, { "epoch": 0.34560861488332356, "grad_norm": 1.7845720052719116, "learning_rate": 1.5215236635136532e-05, "loss": 0.17937085628509522, "step": 69580 }, { "epoch": 0.3456582855666928, "grad_norm": 1.683125615119934, "learning_rate": 1.5213863953891514e-05, "loss": 0.17803831100463868, "step": 69590 }, { "epoch": 0.34570795625006207, "grad_norm": 1.1581034660339355, "learning_rate": 1.5212491137713792e-05, "loss": 0.17508503198623657, "step": 69600 }, { "epoch": 0.3457576269334314, "grad_norm": 1.380510926246643, "learning_rate": 1.5211118186638895e-05, "loss": 0.1774577021598816, "step": 69610 }, { "epoch": 0.3458072976168006, "grad_norm": 1.4058845043182373, "learning_rate": 1.5209745100702357e-05, "loss": 0.15628571510314943, "step": 69620 }, { "epoch": 0.3458569683001699, "grad_norm": 1.1364959478378296, "learning_rate": 1.520837187993971e-05, "loss": 0.16326498985290527, "step": 69630 }, { "epoch": 0.3459066389835391, "grad_norm": 0.7717613577842712, "learning_rate": 1.5206998524386491e-05, "loss": 0.16620091199874878, "step": 69640 }, { "epoch": 0.3459563096669084, "grad_norm": 0.9694701433181763, "learning_rate": 1.5205625034078244e-05, "loss": 0.16375837326049805, "step": 69650 }, { "epoch": 0.3460059803502777, "grad_norm": 0.5921523571014404, "learning_rate": 1.5204251409050517e-05, "loss": 0.16661736965179444, "step": 69660 }, { "epoch": 0.34605565103364694, "grad_norm": 1.633870244026184, "learning_rate": 1.5202877649338854e-05, "loss": 0.1454139232635498, "step": 69670 }, { "epoch": 0.3461053217170162, "grad_norm": 0.7530401349067688, "learning_rate": 1.5201503754978811e-05, "loss": 0.16391389369964598, "step": 69680 }, { "epoch": 0.34615499240038544, "grad_norm": 1.0131402015686035, "learning_rate": 1.5200129726005939e-05, "loss": 0.16481297016143798, "step": 69690 }, { "epoch": 0.3462046630837547, "grad_norm": 1.1766488552093506, "learning_rate": 1.5198755562455808e-05, "loss": 0.17222242355346679, "step": 69700 }, { "epoch": 0.34625433376712395, "grad_norm": 1.0967931747436523, "learning_rate": 1.5197381264363964e-05, "loss": 0.1722961187362671, "step": 69710 }, { "epoch": 0.34630400445049325, "grad_norm": 1.0507019758224487, "learning_rate": 1.5196006831765986e-05, "loss": 0.15742478370666504, "step": 69720 }, { "epoch": 0.3463536751338625, "grad_norm": 1.179140567779541, "learning_rate": 1.5194632264697441e-05, "loss": 0.1698890209197998, "step": 69730 }, { "epoch": 0.34640334581723176, "grad_norm": 1.1267226934432983, "learning_rate": 1.5193257563193904e-05, "loss": 0.16329030990600585, "step": 69740 }, { "epoch": 0.346453016500601, "grad_norm": 0.7973208427429199, "learning_rate": 1.5191882727290947e-05, "loss": 0.17328544855117797, "step": 69750 }, { "epoch": 0.34650268718397026, "grad_norm": 1.1248935461044312, "learning_rate": 1.5190507757024154e-05, "loss": 0.18284273147583008, "step": 69760 }, { "epoch": 0.34655235786733957, "grad_norm": 1.1252247095108032, "learning_rate": 1.5189132652429104e-05, "loss": 0.16514256000518798, "step": 69770 }, { "epoch": 0.3466020285507088, "grad_norm": 1.8780138492584229, "learning_rate": 1.5187757413541388e-05, "loss": 0.16802178621292113, "step": 69780 }, { "epoch": 0.34665169923407807, "grad_norm": 1.0663917064666748, "learning_rate": 1.5186382040396599e-05, "loss": 0.1835172414779663, "step": 69790 }, { "epoch": 0.3467013699174473, "grad_norm": 1.2830110788345337, "learning_rate": 1.5185006533030326e-05, "loss": 0.1683788537979126, "step": 69800 }, { "epoch": 0.3467510406008166, "grad_norm": 0.7088060975074768, "learning_rate": 1.5183630891478169e-05, "loss": 0.1670926332473755, "step": 69810 }, { "epoch": 0.3468007112841858, "grad_norm": 0.969247579574585, "learning_rate": 1.518225511577573e-05, "loss": 0.17025046348571776, "step": 69820 }, { "epoch": 0.34685038196755513, "grad_norm": 1.8426759243011475, "learning_rate": 1.5180879205958612e-05, "loss": 0.17117090225219728, "step": 69830 }, { "epoch": 0.3469000526509244, "grad_norm": 0.8857021331787109, "learning_rate": 1.5179503162062422e-05, "loss": 0.16947662830352783, "step": 69840 }, { "epoch": 0.34694972333429364, "grad_norm": 1.3228251934051514, "learning_rate": 1.5178126984122774e-05, "loss": 0.18676941394805907, "step": 69850 }, { "epoch": 0.3469993940176629, "grad_norm": 1.5342446565628052, "learning_rate": 1.517675067217528e-05, "loss": 0.18317801952362062, "step": 69860 }, { "epoch": 0.34704906470103214, "grad_norm": 1.1231701374053955, "learning_rate": 1.5175374226255564e-05, "loss": 0.157059383392334, "step": 69870 }, { "epoch": 0.3470987353844014, "grad_norm": 1.0601685047149658, "learning_rate": 1.517399764639924e-05, "loss": 0.1695953607559204, "step": 69880 }, { "epoch": 0.3471484060677707, "grad_norm": 1.1946574449539185, "learning_rate": 1.517262093264194e-05, "loss": 0.17236801385879516, "step": 69890 }, { "epoch": 0.34719807675113995, "grad_norm": 1.1380523443222046, "learning_rate": 1.5171244085019287e-05, "loss": 0.1734423279762268, "step": 69900 }, { "epoch": 0.3472477474345092, "grad_norm": 2.230768918991089, "learning_rate": 1.5169867103566918e-05, "loss": 0.17076724767684937, "step": 69910 }, { "epoch": 0.34729741811787845, "grad_norm": 1.0312038660049438, "learning_rate": 1.5168489988320466e-05, "loss": 0.1731829285621643, "step": 69920 }, { "epoch": 0.3473470888012477, "grad_norm": 1.7675180435180664, "learning_rate": 1.5167112739315573e-05, "loss": 0.17560741901397706, "step": 69930 }, { "epoch": 0.347396759484617, "grad_norm": 1.3315263986587524, "learning_rate": 1.516573535658788e-05, "loss": 0.16254088878631592, "step": 69940 }, { "epoch": 0.34744643016798626, "grad_norm": 1.0791243314743042, "learning_rate": 1.5164357840173031e-05, "loss": 0.157757568359375, "step": 69950 }, { "epoch": 0.3474961008513555, "grad_norm": 1.526437759399414, "learning_rate": 1.516298019010668e-05, "loss": 0.16253243684768676, "step": 69960 }, { "epoch": 0.34754577153472477, "grad_norm": 0.922974705696106, "learning_rate": 1.5161602406424476e-05, "loss": 0.1740018129348755, "step": 69970 }, { "epoch": 0.347595442218094, "grad_norm": 1.2313926219940186, "learning_rate": 1.516022448916208e-05, "loss": 0.169249963760376, "step": 69980 }, { "epoch": 0.34764511290146327, "grad_norm": 0.8327062129974365, "learning_rate": 1.5158846438355145e-05, "loss": 0.166298246383667, "step": 69990 }, { "epoch": 0.3476947835848326, "grad_norm": 1.3959921598434448, "learning_rate": 1.5157468254039337e-05, "loss": 0.1831942081451416, "step": 70000 }, { "epoch": 0.34774445426820183, "grad_norm": 0.6325491070747375, "learning_rate": 1.5156089936250328e-05, "loss": 0.16052629947662353, "step": 70010 }, { "epoch": 0.3477941249515711, "grad_norm": 0.9490272998809814, "learning_rate": 1.5154711485023786e-05, "loss": 0.17420680522918702, "step": 70020 }, { "epoch": 0.34784379563494033, "grad_norm": 1.7751538753509521, "learning_rate": 1.5153332900395381e-05, "loss": 0.1691875457763672, "step": 70030 }, { "epoch": 0.3478934663183096, "grad_norm": 1.34457528591156, "learning_rate": 1.5151954182400794e-05, "loss": 0.15834827423095704, "step": 70040 }, { "epoch": 0.3479431370016789, "grad_norm": 2.4837329387664795, "learning_rate": 1.5150575331075703e-05, "loss": 0.18312041759490966, "step": 70050 }, { "epoch": 0.34799280768504814, "grad_norm": 1.5808666944503784, "learning_rate": 1.5149196346455792e-05, "loss": 0.1735671043395996, "step": 70060 }, { "epoch": 0.3480424783684174, "grad_norm": 1.6021203994750977, "learning_rate": 1.5147817228576753e-05, "loss": 0.17913777828216554, "step": 70070 }, { "epoch": 0.34809214905178665, "grad_norm": 0.8159796595573425, "learning_rate": 1.5146437977474274e-05, "loss": 0.17241249084472657, "step": 70080 }, { "epoch": 0.3481418197351559, "grad_norm": 1.1050316095352173, "learning_rate": 1.5145058593184048e-05, "loss": 0.16663228273391723, "step": 70090 }, { "epoch": 0.34819149041852515, "grad_norm": 1.379302978515625, "learning_rate": 1.5143679075741776e-05, "loss": 0.16964321136474608, "step": 70100 }, { "epoch": 0.34824116110189446, "grad_norm": 1.1866172552108765, "learning_rate": 1.5142299425183153e-05, "loss": 0.17462280988693238, "step": 70110 }, { "epoch": 0.3482908317852637, "grad_norm": 1.5238432884216309, "learning_rate": 1.5140919641543892e-05, "loss": 0.1813868284225464, "step": 70120 }, { "epoch": 0.34834050246863296, "grad_norm": 1.228241205215454, "learning_rate": 1.5139539724859698e-05, "loss": 0.1691705822944641, "step": 70130 }, { "epoch": 0.3483901731520022, "grad_norm": 0.8798518180847168, "learning_rate": 1.5138159675166282e-05, "loss": 0.16958777904510497, "step": 70140 }, { "epoch": 0.34843984383537147, "grad_norm": 0.5918145775794983, "learning_rate": 1.513677949249936e-05, "loss": 0.159599769115448, "step": 70150 }, { "epoch": 0.3484895145187408, "grad_norm": 0.8498128056526184, "learning_rate": 1.5135399176894647e-05, "loss": 0.16686947345733644, "step": 70160 }, { "epoch": 0.34853918520211, "grad_norm": 0.8897320628166199, "learning_rate": 1.5134018728387872e-05, "loss": 0.17289026975631713, "step": 70170 }, { "epoch": 0.3485888558854793, "grad_norm": 1.5702672004699707, "learning_rate": 1.5132638147014755e-05, "loss": 0.17412563562393188, "step": 70180 }, { "epoch": 0.34863852656884853, "grad_norm": 0.8067978024482727, "learning_rate": 1.5131257432811026e-05, "loss": 0.17335883378982545, "step": 70190 }, { "epoch": 0.3486881972522178, "grad_norm": 1.534760594367981, "learning_rate": 1.512987658581242e-05, "loss": 0.18269566297531128, "step": 70200 }, { "epoch": 0.34873786793558703, "grad_norm": 1.0010967254638672, "learning_rate": 1.5128495606054671e-05, "loss": 0.1830296516418457, "step": 70210 }, { "epoch": 0.34878753861895634, "grad_norm": 0.9859007000923157, "learning_rate": 1.5127114493573516e-05, "loss": 0.17415846586227418, "step": 70220 }, { "epoch": 0.3488372093023256, "grad_norm": 0.9162255525588989, "learning_rate": 1.5125733248404699e-05, "loss": 0.16854453086853027, "step": 70230 }, { "epoch": 0.34888687998569484, "grad_norm": 0.7287999391555786, "learning_rate": 1.5124351870583968e-05, "loss": 0.17064734697341918, "step": 70240 }, { "epoch": 0.3489365506690641, "grad_norm": 0.8744284510612488, "learning_rate": 1.5122970360147066e-05, "loss": 0.17565934658050536, "step": 70250 }, { "epoch": 0.34898622135243335, "grad_norm": 0.9973095059394836, "learning_rate": 1.5121588717129757e-05, "loss": 0.16721231937408448, "step": 70260 }, { "epoch": 0.34903589203580265, "grad_norm": 1.8387749195098877, "learning_rate": 1.512020694156779e-05, "loss": 0.1676633596420288, "step": 70270 }, { "epoch": 0.3490855627191719, "grad_norm": 1.5716005563735962, "learning_rate": 1.5118825033496926e-05, "loss": 0.17475109100341796, "step": 70280 }, { "epoch": 0.34913523340254116, "grad_norm": 1.071172833442688, "learning_rate": 1.511744299295293e-05, "loss": 0.15924296379089356, "step": 70290 }, { "epoch": 0.3491849040859104, "grad_norm": 1.6562308073043823, "learning_rate": 1.5116060819971562e-05, "loss": 0.14962904453277587, "step": 70300 }, { "epoch": 0.34923457476927966, "grad_norm": 2.001084089279175, "learning_rate": 1.5114678514588602e-05, "loss": 0.1666126251220703, "step": 70310 }, { "epoch": 0.3492842454526489, "grad_norm": 0.6471066474914551, "learning_rate": 1.5113296076839814e-05, "loss": 0.17272813320159913, "step": 70320 }, { "epoch": 0.3493339161360182, "grad_norm": 1.5584315061569214, "learning_rate": 1.5111913506760982e-05, "loss": 0.1654014229774475, "step": 70330 }, { "epoch": 0.34938358681938747, "grad_norm": 1.8441908359527588, "learning_rate": 1.5110530804387882e-05, "loss": 0.17773523330688476, "step": 70340 }, { "epoch": 0.3494332575027567, "grad_norm": 1.3454769849777222, "learning_rate": 1.5109147969756301e-05, "loss": 0.16853184700012208, "step": 70350 }, { "epoch": 0.349482928186126, "grad_norm": 1.545322060585022, "learning_rate": 1.5107765002902027e-05, "loss": 0.1446807265281677, "step": 70360 }, { "epoch": 0.3495325988694952, "grad_norm": 1.0998996496200562, "learning_rate": 1.5106381903860845e-05, "loss": 0.16409623622894287, "step": 70370 }, { "epoch": 0.34958226955286453, "grad_norm": 1.2774215936660767, "learning_rate": 1.5104998672668557e-05, "loss": 0.17495590448379517, "step": 70380 }, { "epoch": 0.3496319402362338, "grad_norm": 1.2854077816009521, "learning_rate": 1.5103615309360952e-05, "loss": 0.17430338859558106, "step": 70390 }, { "epoch": 0.34968161091960304, "grad_norm": 0.8438593745231628, "learning_rate": 1.5102231813973838e-05, "loss": 0.1681424856185913, "step": 70400 }, { "epoch": 0.3497312816029723, "grad_norm": 1.1960108280181885, "learning_rate": 1.5100848186543014e-05, "loss": 0.15921190977096558, "step": 70410 }, { "epoch": 0.34978095228634154, "grad_norm": 2.631366729736328, "learning_rate": 1.5099464427104288e-05, "loss": 0.1675082802772522, "step": 70420 }, { "epoch": 0.3498306229697108, "grad_norm": 1.304260492324829, "learning_rate": 1.5098080535693477e-05, "loss": 0.15583224296569825, "step": 70430 }, { "epoch": 0.3498802936530801, "grad_norm": 0.8527277708053589, "learning_rate": 1.5096696512346389e-05, "loss": 0.1706540584564209, "step": 70440 }, { "epoch": 0.34992996433644935, "grad_norm": 0.8356271982192993, "learning_rate": 1.5095312357098844e-05, "loss": 0.18043434619903564, "step": 70450 }, { "epoch": 0.3499796350198186, "grad_norm": 1.0466777086257935, "learning_rate": 1.5093928069986666e-05, "loss": 0.17566450834274291, "step": 70460 }, { "epoch": 0.35002930570318785, "grad_norm": 1.0615979433059692, "learning_rate": 1.5092543651045677e-05, "loss": 0.15977296829223633, "step": 70470 }, { "epoch": 0.3500789763865571, "grad_norm": 1.472532033920288, "learning_rate": 1.5091159100311705e-05, "loss": 0.1778690218925476, "step": 70480 }, { "epoch": 0.3501286470699264, "grad_norm": 1.5056238174438477, "learning_rate": 1.5089774417820583e-05, "loss": 0.16013331413269044, "step": 70490 }, { "epoch": 0.35017831775329566, "grad_norm": 2.914032459259033, "learning_rate": 1.508838960360815e-05, "loss": 0.1611289381980896, "step": 70500 }, { "epoch": 0.3502279884366649, "grad_norm": 1.335373044013977, "learning_rate": 1.5087004657710234e-05, "loss": 0.176423180103302, "step": 70510 }, { "epoch": 0.35027765912003417, "grad_norm": 0.995075523853302, "learning_rate": 1.5085619580162685e-05, "loss": 0.17414864301681518, "step": 70520 }, { "epoch": 0.3503273298034034, "grad_norm": 0.7033314108848572, "learning_rate": 1.5084234371001346e-05, "loss": 0.1735943913459778, "step": 70530 }, { "epoch": 0.35037700048677267, "grad_norm": 0.9908255934715271, "learning_rate": 1.5082849030262066e-05, "loss": 0.1723616600036621, "step": 70540 }, { "epoch": 0.350426671170142, "grad_norm": 1.0462818145751953, "learning_rate": 1.5081463557980697e-05, "loss": 0.1656050205230713, "step": 70550 }, { "epoch": 0.35047634185351123, "grad_norm": 1.1501572132110596, "learning_rate": 1.5080077954193093e-05, "loss": 0.16293517351150513, "step": 70560 }, { "epoch": 0.3505260125368805, "grad_norm": 0.9783922433853149, "learning_rate": 1.5078692218935116e-05, "loss": 0.17467646598815917, "step": 70570 }, { "epoch": 0.35057568322024973, "grad_norm": 1.023925542831421, "learning_rate": 1.5077306352242622e-05, "loss": 0.16176466941833495, "step": 70580 }, { "epoch": 0.350625353903619, "grad_norm": 0.6134678721427917, "learning_rate": 1.5075920354151486e-05, "loss": 0.16626192331314088, "step": 70590 }, { "epoch": 0.3506750245869883, "grad_norm": 0.8814889192581177, "learning_rate": 1.507453422469757e-05, "loss": 0.1727758526802063, "step": 70600 }, { "epoch": 0.35072469527035754, "grad_norm": 1.1964244842529297, "learning_rate": 1.507314796391675e-05, "loss": 0.15946781635284424, "step": 70610 }, { "epoch": 0.3507743659537268, "grad_norm": 2.121628522872925, "learning_rate": 1.5071761571844903e-05, "loss": 0.1635727882385254, "step": 70620 }, { "epoch": 0.35082403663709605, "grad_norm": 0.8282020688056946, "learning_rate": 1.5070375048517902e-05, "loss": 0.17113337516784669, "step": 70630 }, { "epoch": 0.3508737073204653, "grad_norm": 1.7642500400543213, "learning_rate": 1.5068988393971634e-05, "loss": 0.16833133697509767, "step": 70640 }, { "epoch": 0.35092337800383455, "grad_norm": 1.3798127174377441, "learning_rate": 1.5067601608241983e-05, "loss": 0.15697935819625855, "step": 70650 }, { "epoch": 0.35097304868720386, "grad_norm": 0.7274913787841797, "learning_rate": 1.506621469136484e-05, "loss": 0.16965296268463134, "step": 70660 }, { "epoch": 0.3510227193705731, "grad_norm": 1.2250795364379883, "learning_rate": 1.50648276433761e-05, "loss": 0.1624486804008484, "step": 70670 }, { "epoch": 0.35107239005394236, "grad_norm": 0.7339266538619995, "learning_rate": 1.5063440464311659e-05, "loss": 0.16699690818786622, "step": 70680 }, { "epoch": 0.3511220607373116, "grad_norm": 1.2627243995666504, "learning_rate": 1.5062053154207412e-05, "loss": 0.17674834728240968, "step": 70690 }, { "epoch": 0.35117173142068087, "grad_norm": 1.3680379390716553, "learning_rate": 1.5060665713099263e-05, "loss": 0.16509060859680175, "step": 70700 }, { "epoch": 0.3512214021040502, "grad_norm": 1.049541711807251, "learning_rate": 1.5059278141023127e-05, "loss": 0.18013885021209716, "step": 70710 }, { "epoch": 0.3512710727874194, "grad_norm": 1.0589168071746826, "learning_rate": 1.5057890438014899e-05, "loss": 0.16554830074310303, "step": 70720 }, { "epoch": 0.3513207434707887, "grad_norm": 0.931084930896759, "learning_rate": 1.5056502604110503e-05, "loss": 0.17766307592391967, "step": 70730 }, { "epoch": 0.35137041415415793, "grad_norm": 0.9455238580703735, "learning_rate": 1.5055114639345853e-05, "loss": 0.16405942440032958, "step": 70740 }, { "epoch": 0.3514200848375272, "grad_norm": 0.710112452507019, "learning_rate": 1.505372654375687e-05, "loss": 0.18528808355331422, "step": 70750 }, { "epoch": 0.35146975552089643, "grad_norm": 0.7190626263618469, "learning_rate": 1.5052338317379477e-05, "loss": 0.17872027158737183, "step": 70760 }, { "epoch": 0.35151942620426574, "grad_norm": 1.7059986591339111, "learning_rate": 1.5050949960249596e-05, "loss": 0.17262635231018067, "step": 70770 }, { "epoch": 0.351569096887635, "grad_norm": 0.9083621501922607, "learning_rate": 1.5049561472403162e-05, "loss": 0.17105904817581177, "step": 70780 }, { "epoch": 0.35161876757100424, "grad_norm": 1.8027451038360596, "learning_rate": 1.5048172853876106e-05, "loss": 0.17233622074127197, "step": 70790 }, { "epoch": 0.3516684382543735, "grad_norm": 0.8256030678749084, "learning_rate": 1.5046784104704371e-05, "loss": 0.15949716567993164, "step": 70800 }, { "epoch": 0.35171810893774275, "grad_norm": 1.6759694814682007, "learning_rate": 1.5045395224923888e-05, "loss": 0.18126243352890015, "step": 70810 }, { "epoch": 0.35176777962111205, "grad_norm": 1.2786484956741333, "learning_rate": 1.5044006214570608e-05, "loss": 0.18525679111480714, "step": 70820 }, { "epoch": 0.3518174503044813, "grad_norm": 0.537099301815033, "learning_rate": 1.5042617073680474e-05, "loss": 0.17283415794372559, "step": 70830 }, { "epoch": 0.35186712098785056, "grad_norm": 1.4855471849441528, "learning_rate": 1.504122780228944e-05, "loss": 0.15299360752105712, "step": 70840 }, { "epoch": 0.3519167916712198, "grad_norm": 1.4369322061538696, "learning_rate": 1.5039838400433455e-05, "loss": 0.16862750053405762, "step": 70850 }, { "epoch": 0.35196646235458906, "grad_norm": 1.28817880153656, "learning_rate": 1.5038448868148479e-05, "loss": 0.17295668125152588, "step": 70860 }, { "epoch": 0.3520161330379583, "grad_norm": 1.458571434020996, "learning_rate": 1.503705920547047e-05, "loss": 0.17523951530456544, "step": 70870 }, { "epoch": 0.3520658037213276, "grad_norm": 0.7530366778373718, "learning_rate": 1.5035669412435398e-05, "loss": 0.16377592086791992, "step": 70880 }, { "epoch": 0.35211547440469687, "grad_norm": 1.6674573421478271, "learning_rate": 1.5034279489079226e-05, "loss": 0.17735767364501953, "step": 70890 }, { "epoch": 0.3521651450880661, "grad_norm": 0.8554290533065796, "learning_rate": 1.5032889435437925e-05, "loss": 0.1699673652648926, "step": 70900 }, { "epoch": 0.3522148157714354, "grad_norm": 0.6177035570144653, "learning_rate": 1.5031499251547468e-05, "loss": 0.16128973960876464, "step": 70910 }, { "epoch": 0.3522644864548046, "grad_norm": 0.5485839247703552, "learning_rate": 1.5030108937443832e-05, "loss": 0.1768815517425537, "step": 70920 }, { "epoch": 0.35231415713817393, "grad_norm": 0.8148598074913025, "learning_rate": 1.5028718493162999e-05, "loss": 0.17217864990234374, "step": 70930 }, { "epoch": 0.3523638278215432, "grad_norm": 1.8550471067428589, "learning_rate": 1.5027327918740957e-05, "loss": 0.1846766233444214, "step": 70940 }, { "epoch": 0.35241349850491244, "grad_norm": 1.842164397239685, "learning_rate": 1.5025937214213686e-05, "loss": 0.18264436721801758, "step": 70950 }, { "epoch": 0.3524631691882817, "grad_norm": 1.4170639514923096, "learning_rate": 1.5024546379617184e-05, "loss": 0.17162595987319945, "step": 70960 }, { "epoch": 0.35251283987165094, "grad_norm": 2.2639124393463135, "learning_rate": 1.502315541498744e-05, "loss": 0.1740108013153076, "step": 70970 }, { "epoch": 0.3525625105550202, "grad_norm": 1.392120122909546, "learning_rate": 1.5021764320360455e-05, "loss": 0.172682785987854, "step": 70980 }, { "epoch": 0.3526121812383895, "grad_norm": 0.8021690249443054, "learning_rate": 1.5020373095772227e-05, "loss": 0.1693817377090454, "step": 70990 }, { "epoch": 0.35266185192175875, "grad_norm": 0.9474785923957825, "learning_rate": 1.5018981741258762e-05, "loss": 0.15865564346313477, "step": 71000 }, { "epoch": 0.352711522605128, "grad_norm": 1.4611537456512451, "learning_rate": 1.501759025685607e-05, "loss": 0.16360881328582763, "step": 71010 }, { "epoch": 0.35276119328849725, "grad_norm": 1.7439649105072021, "learning_rate": 1.5016198642600154e-05, "loss": 0.17204301357269286, "step": 71020 }, { "epoch": 0.3528108639718665, "grad_norm": 1.3585329055786133, "learning_rate": 1.5014806898527038e-05, "loss": 0.15832279920578002, "step": 71030 }, { "epoch": 0.3528605346552358, "grad_norm": 0.9057320952415466, "learning_rate": 1.5013415024672737e-05, "loss": 0.16483895778656005, "step": 71040 }, { "epoch": 0.35291020533860507, "grad_norm": 0.8859281539916992, "learning_rate": 1.501202302107327e-05, "loss": 0.16516544818878173, "step": 71050 }, { "epoch": 0.3529598760219743, "grad_norm": 1.6430697441101074, "learning_rate": 1.5010630887764663e-05, "loss": 0.172646164894104, "step": 71060 }, { "epoch": 0.35300954670534357, "grad_norm": 1.9926435947418213, "learning_rate": 1.5009238624782943e-05, "loss": 0.1608796238899231, "step": 71070 }, { "epoch": 0.3530592173887128, "grad_norm": 1.507681131362915, "learning_rate": 1.5007846232164141e-05, "loss": 0.15719325542449952, "step": 71080 }, { "epoch": 0.35310888807208207, "grad_norm": 1.4057815074920654, "learning_rate": 1.5006453709944291e-05, "loss": 0.17967658042907714, "step": 71090 }, { "epoch": 0.3531585587554514, "grad_norm": 1.7165213823318481, "learning_rate": 1.5005061058159433e-05, "loss": 0.15380992889404296, "step": 71100 }, { "epoch": 0.35320822943882063, "grad_norm": 0.6571804285049438, "learning_rate": 1.5003668276845608e-05, "loss": 0.16123925447463988, "step": 71110 }, { "epoch": 0.3532579001221899, "grad_norm": 1.2985695600509644, "learning_rate": 1.500227536603886e-05, "loss": 0.16023035049438478, "step": 71120 }, { "epoch": 0.35330757080555913, "grad_norm": 1.8943276405334473, "learning_rate": 1.5000882325775234e-05, "loss": 0.16903401613235475, "step": 71130 }, { "epoch": 0.3533572414889284, "grad_norm": 0.7768806219100952, "learning_rate": 1.4999489156090787e-05, "loss": 0.17088034152984619, "step": 71140 }, { "epoch": 0.3534069121722977, "grad_norm": 1.2918291091918945, "learning_rate": 1.499809585702157e-05, "loss": 0.17945196628570556, "step": 71150 }, { "epoch": 0.35345658285566695, "grad_norm": 1.942971110343933, "learning_rate": 1.4996702428603642e-05, "loss": 0.16403791904449463, "step": 71160 }, { "epoch": 0.3535062535390362, "grad_norm": 1.035979986190796, "learning_rate": 1.4995308870873064e-05, "loss": 0.17750468254089355, "step": 71170 }, { "epoch": 0.35355592422240545, "grad_norm": 1.3548507690429688, "learning_rate": 1.4993915183865904e-05, "loss": 0.17359637022018432, "step": 71180 }, { "epoch": 0.3536055949057747, "grad_norm": 1.1808414459228516, "learning_rate": 1.4992521367618223e-05, "loss": 0.18088793754577637, "step": 71190 }, { "epoch": 0.35365526558914395, "grad_norm": 0.8983572721481323, "learning_rate": 1.4991127422166096e-05, "loss": 0.16369043588638305, "step": 71200 }, { "epoch": 0.35370493627251326, "grad_norm": 1.3078292608261108, "learning_rate": 1.49897333475456e-05, "loss": 0.1670583963394165, "step": 71210 }, { "epoch": 0.3537546069558825, "grad_norm": 1.8318570852279663, "learning_rate": 1.4988339143792811e-05, "loss": 0.16963522434234618, "step": 71220 }, { "epoch": 0.35380427763925176, "grad_norm": 1.1205501556396484, "learning_rate": 1.4986944810943813e-05, "loss": 0.1602788209915161, "step": 71230 }, { "epoch": 0.353853948322621, "grad_norm": 0.5359695553779602, "learning_rate": 1.4985550349034686e-05, "loss": 0.16765533685684203, "step": 71240 }, { "epoch": 0.35390361900599027, "grad_norm": 1.2296185493469238, "learning_rate": 1.4984155758101521e-05, "loss": 0.16767704486846924, "step": 71250 }, { "epoch": 0.3539532896893596, "grad_norm": 1.5784859657287598, "learning_rate": 1.4982761038180406e-05, "loss": 0.16777124404907226, "step": 71260 }, { "epoch": 0.3540029603727288, "grad_norm": 0.9016743898391724, "learning_rate": 1.4981366189307443e-05, "loss": 0.15614769458770753, "step": 71270 }, { "epoch": 0.3540526310560981, "grad_norm": 1.3425781726837158, "learning_rate": 1.4979971211518724e-05, "loss": 0.1661447286605835, "step": 71280 }, { "epoch": 0.35410230173946733, "grad_norm": 1.7762876749038696, "learning_rate": 1.4978576104850353e-05, "loss": 0.16022274494171143, "step": 71290 }, { "epoch": 0.3541519724228366, "grad_norm": 1.331717848777771, "learning_rate": 1.4977180869338435e-05, "loss": 0.16469560861587523, "step": 71300 }, { "epoch": 0.35420164310620583, "grad_norm": 1.8274716138839722, "learning_rate": 1.4975785505019075e-05, "loss": 0.15609086751937867, "step": 71310 }, { "epoch": 0.35425131378957514, "grad_norm": 0.956035315990448, "learning_rate": 1.497439001192839e-05, "loss": 0.15448712110519408, "step": 71320 }, { "epoch": 0.3543009844729444, "grad_norm": 1.387323021888733, "learning_rate": 1.4972994390102488e-05, "loss": 0.19390913248062133, "step": 71330 }, { "epoch": 0.35435065515631364, "grad_norm": 1.568483591079712, "learning_rate": 1.4971598639577494e-05, "loss": 0.16497716903686524, "step": 71340 }, { "epoch": 0.3544003258396829, "grad_norm": 0.7548929452896118, "learning_rate": 1.4970202760389523e-05, "loss": 0.18470594882965088, "step": 71350 }, { "epoch": 0.35444999652305215, "grad_norm": 0.7489092946052551, "learning_rate": 1.4968806752574706e-05, "loss": 0.18035485744476318, "step": 71360 }, { "epoch": 0.3544996672064214, "grad_norm": 1.368313193321228, "learning_rate": 1.4967410616169166e-05, "loss": 0.170514976978302, "step": 71370 }, { "epoch": 0.3545493378897907, "grad_norm": 0.9894872903823853, "learning_rate": 1.4966014351209039e-05, "loss": 0.16766687631607055, "step": 71380 }, { "epoch": 0.35459900857315996, "grad_norm": 1.1564886569976807, "learning_rate": 1.4964617957730455e-05, "loss": 0.1693565845489502, "step": 71390 }, { "epoch": 0.3546486792565292, "grad_norm": 0.8521565198898315, "learning_rate": 1.4963221435769557e-05, "loss": 0.16140034198760986, "step": 71400 }, { "epoch": 0.35469834993989846, "grad_norm": 1.0560219287872314, "learning_rate": 1.4961824785362483e-05, "loss": 0.1655733823776245, "step": 71410 }, { "epoch": 0.3547480206232677, "grad_norm": 1.9632600545883179, "learning_rate": 1.4960428006545376e-05, "loss": 0.16314163208007812, "step": 71420 }, { "epoch": 0.354797691306637, "grad_norm": 3.228994607925415, "learning_rate": 1.4959031099354388e-05, "loss": 0.1761014461517334, "step": 71430 }, { "epoch": 0.35484736199000627, "grad_norm": 1.1136415004730225, "learning_rate": 1.495763406382567e-05, "loss": 0.1690513849258423, "step": 71440 }, { "epoch": 0.3548970326733755, "grad_norm": 2.1894338130950928, "learning_rate": 1.4956236899995377e-05, "loss": 0.1718724012374878, "step": 71450 }, { "epoch": 0.3549467033567448, "grad_norm": 2.5560386180877686, "learning_rate": 1.4954839607899662e-05, "loss": 0.15837713479995727, "step": 71460 }, { "epoch": 0.354996374040114, "grad_norm": 0.9945138692855835, "learning_rate": 1.4953442187574694e-05, "loss": 0.1774226665496826, "step": 71470 }, { "epoch": 0.3550460447234833, "grad_norm": 0.7981659770011902, "learning_rate": 1.4952044639056632e-05, "loss": 0.1664330244064331, "step": 71480 }, { "epoch": 0.3550957154068526, "grad_norm": 1.4017521142959595, "learning_rate": 1.4950646962381645e-05, "loss": 0.1716353178024292, "step": 71490 }, { "epoch": 0.35514538609022184, "grad_norm": 0.633972704410553, "learning_rate": 1.4949249157585907e-05, "loss": 0.1713874340057373, "step": 71500 }, { "epoch": 0.3551950567735911, "grad_norm": 0.7063437104225159, "learning_rate": 1.4947851224705592e-05, "loss": 0.18945021629333497, "step": 71510 }, { "epoch": 0.35524472745696034, "grad_norm": 1.0569915771484375, "learning_rate": 1.4946453163776874e-05, "loss": 0.16782662868499756, "step": 71520 }, { "epoch": 0.3552943981403296, "grad_norm": 1.1355427503585815, "learning_rate": 1.4945054974835937e-05, "loss": 0.1682717204093933, "step": 71530 }, { "epoch": 0.3553440688236989, "grad_norm": 1.0825239419937134, "learning_rate": 1.4943656657918966e-05, "loss": 0.16219031810760498, "step": 71540 }, { "epoch": 0.35539373950706815, "grad_norm": 1.7323271036148071, "learning_rate": 1.4942258213062149e-05, "loss": 0.18539981842041015, "step": 71550 }, { "epoch": 0.3554434101904374, "grad_norm": 1.035521388053894, "learning_rate": 1.4940859640301679e-05, "loss": 0.179433012008667, "step": 71560 }, { "epoch": 0.35549308087380665, "grad_norm": 0.9660362005233765, "learning_rate": 1.4939460939673746e-05, "loss": 0.17294830083847046, "step": 71570 }, { "epoch": 0.3555427515571759, "grad_norm": 1.0971225500106812, "learning_rate": 1.4938062111214551e-05, "loss": 0.17257047891616822, "step": 71580 }, { "epoch": 0.35559242224054516, "grad_norm": 0.9439839720726013, "learning_rate": 1.4936663154960291e-05, "loss": 0.16344897747039794, "step": 71590 }, { "epoch": 0.35564209292391447, "grad_norm": 1.7357282638549805, "learning_rate": 1.4935264070947175e-05, "loss": 0.17139718532562256, "step": 71600 }, { "epoch": 0.3556917636072837, "grad_norm": 2.8260107040405273, "learning_rate": 1.4933864859211408e-05, "loss": 0.18593318462371827, "step": 71610 }, { "epoch": 0.35574143429065297, "grad_norm": 1.081504225730896, "learning_rate": 1.4932465519789206e-05, "loss": 0.17008135318756104, "step": 71620 }, { "epoch": 0.3557911049740222, "grad_norm": 0.6434152126312256, "learning_rate": 1.4931066052716778e-05, "loss": 0.1753965735435486, "step": 71630 }, { "epoch": 0.3558407756573915, "grad_norm": 1.9987404346466064, "learning_rate": 1.4929666458030343e-05, "loss": 0.16521316766738892, "step": 71640 }, { "epoch": 0.3558904463407608, "grad_norm": 1.418762445449829, "learning_rate": 1.4928266735766121e-05, "loss": 0.15136677026748657, "step": 71650 }, { "epoch": 0.35594011702413003, "grad_norm": 1.42717707157135, "learning_rate": 1.4926866885960338e-05, "loss": 0.16138710975646972, "step": 71660 }, { "epoch": 0.3559897877074993, "grad_norm": 1.236850619316101, "learning_rate": 1.492546690864922e-05, "loss": 0.1808554410934448, "step": 71670 }, { "epoch": 0.35603945839086854, "grad_norm": 1.495137333869934, "learning_rate": 1.4924066803869002e-05, "loss": 0.18057101964950562, "step": 71680 }, { "epoch": 0.3560891290742378, "grad_norm": 0.5769758224487305, "learning_rate": 1.4922666571655913e-05, "loss": 0.16169122457504273, "step": 71690 }, { "epoch": 0.35613879975760704, "grad_norm": 1.9226796627044678, "learning_rate": 1.4921266212046193e-05, "loss": 0.1765780806541443, "step": 71700 }, { "epoch": 0.35618847044097635, "grad_norm": 0.9102556705474854, "learning_rate": 1.4919865725076082e-05, "loss": 0.17622747421264648, "step": 71710 }, { "epoch": 0.3562381411243456, "grad_norm": 0.7125791311264038, "learning_rate": 1.4918465110781824e-05, "loss": 0.18439366817474365, "step": 71720 }, { "epoch": 0.35628781180771485, "grad_norm": 1.4780685901641846, "learning_rate": 1.4917064369199667e-05, "loss": 0.17342839241027833, "step": 71730 }, { "epoch": 0.3563374824910841, "grad_norm": 0.7431456446647644, "learning_rate": 1.491566350036586e-05, "loss": 0.1722193479537964, "step": 71740 }, { "epoch": 0.35638715317445335, "grad_norm": 0.884067177772522, "learning_rate": 1.4914262504316659e-05, "loss": 0.15851075649261476, "step": 71750 }, { "epoch": 0.35643682385782266, "grad_norm": 0.7635511755943298, "learning_rate": 1.4912861381088321e-05, "loss": 0.17522268295288085, "step": 71760 }, { "epoch": 0.3564864945411919, "grad_norm": 1.143364667892456, "learning_rate": 1.4911460130717106e-05, "loss": 0.16263830661773682, "step": 71770 }, { "epoch": 0.35653616522456116, "grad_norm": 1.0155222415924072, "learning_rate": 1.4910058753239276e-05, "loss": 0.18211660385131836, "step": 71780 }, { "epoch": 0.3565858359079304, "grad_norm": 1.3176348209381104, "learning_rate": 1.49086572486911e-05, "loss": 0.18056826591491698, "step": 71790 }, { "epoch": 0.35663550659129967, "grad_norm": 1.4365664720535278, "learning_rate": 1.490725561710885e-05, "loss": 0.16062376499176026, "step": 71800 }, { "epoch": 0.3566851772746689, "grad_norm": 0.8363361954689026, "learning_rate": 1.4905853858528794e-05, "loss": 0.17085518836975097, "step": 71810 }, { "epoch": 0.3567348479580382, "grad_norm": 1.319254755973816, "learning_rate": 1.4904451972987214e-05, "loss": 0.1605192542076111, "step": 71820 }, { "epoch": 0.3567845186414075, "grad_norm": 1.1056574583053589, "learning_rate": 1.4903049960520392e-05, "loss": 0.1758564829826355, "step": 71830 }, { "epoch": 0.35683418932477673, "grad_norm": 0.8606093525886536, "learning_rate": 1.4901647821164607e-05, "loss": 0.1543344259262085, "step": 71840 }, { "epoch": 0.356883860008146, "grad_norm": 0.9174781441688538, "learning_rate": 1.4900245554956146e-05, "loss": 0.17080005407333373, "step": 71850 }, { "epoch": 0.35693353069151523, "grad_norm": 2.0026133060455322, "learning_rate": 1.48988431619313e-05, "loss": 0.1712231755256653, "step": 71860 }, { "epoch": 0.35698320137488454, "grad_norm": 1.1139558553695679, "learning_rate": 1.4897440642126363e-05, "loss": 0.1797313690185547, "step": 71870 }, { "epoch": 0.3570328720582538, "grad_norm": 1.0128880739212036, "learning_rate": 1.4896037995577632e-05, "loss": 0.17092361450195312, "step": 71880 }, { "epoch": 0.35708254274162304, "grad_norm": 0.9860548377037048, "learning_rate": 1.4894635222321407e-05, "loss": 0.1704203963279724, "step": 71890 }, { "epoch": 0.3571322134249923, "grad_norm": 1.0140098333358765, "learning_rate": 1.4893232322393992e-05, "loss": 0.16891717910766602, "step": 71900 }, { "epoch": 0.35718188410836155, "grad_norm": 1.6405510902404785, "learning_rate": 1.489182929583169e-05, "loss": 0.18239645957946776, "step": 71910 }, { "epoch": 0.3572315547917308, "grad_norm": 1.3328787088394165, "learning_rate": 1.4890426142670813e-05, "loss": 0.1637364387512207, "step": 71920 }, { "epoch": 0.3572812254751001, "grad_norm": 1.0082857608795166, "learning_rate": 1.4889022862947671e-05, "loss": 0.17131974697113037, "step": 71930 }, { "epoch": 0.35733089615846936, "grad_norm": 0.850705623626709, "learning_rate": 1.4887619456698588e-05, "loss": 0.16887582540512086, "step": 71940 }, { "epoch": 0.3573805668418386, "grad_norm": 2.0103023052215576, "learning_rate": 1.4886215923959874e-05, "loss": 0.17335416078567506, "step": 71950 }, { "epoch": 0.35743023752520786, "grad_norm": 1.0116080045700073, "learning_rate": 1.488481226476786e-05, "loss": 0.17217748165130614, "step": 71960 }, { "epoch": 0.3574799082085771, "grad_norm": 1.3378275632858276, "learning_rate": 1.4883408479158867e-05, "loss": 0.1629809021949768, "step": 71970 }, { "epoch": 0.3575295788919464, "grad_norm": 1.6875139474868774, "learning_rate": 1.4882004567169226e-05, "loss": 0.1661773920059204, "step": 71980 }, { "epoch": 0.35757924957531567, "grad_norm": 1.084470272064209, "learning_rate": 1.4880600528835272e-05, "loss": 0.1672644019126892, "step": 71990 }, { "epoch": 0.3576289202586849, "grad_norm": 1.0251426696777344, "learning_rate": 1.4879196364193336e-05, "loss": 0.16903220415115355, "step": 72000 }, { "epoch": 0.3576785909420542, "grad_norm": 2.3264408111572266, "learning_rate": 1.487779207327976e-05, "loss": 0.16675229072570802, "step": 72010 }, { "epoch": 0.3577282616254234, "grad_norm": 1.2418127059936523, "learning_rate": 1.4876387656130889e-05, "loss": 0.1822273015975952, "step": 72020 }, { "epoch": 0.3577779323087927, "grad_norm": 1.5275585651397705, "learning_rate": 1.4874983112783063e-05, "loss": 0.16176888942718506, "step": 72030 }, { "epoch": 0.357827602992162, "grad_norm": 0.7558735013008118, "learning_rate": 1.4873578443272635e-05, "loss": 0.1651578426361084, "step": 72040 }, { "epoch": 0.35787727367553124, "grad_norm": 1.117080569267273, "learning_rate": 1.487217364763596e-05, "loss": 0.16236648559570313, "step": 72050 }, { "epoch": 0.3579269443589005, "grad_norm": 0.8353835344314575, "learning_rate": 1.4870768725909386e-05, "loss": 0.17450966835021972, "step": 72060 }, { "epoch": 0.35797661504226974, "grad_norm": 1.0015345811843872, "learning_rate": 1.4869363678129278e-05, "loss": 0.17480130195617677, "step": 72070 }, { "epoch": 0.358026285725639, "grad_norm": 1.3024909496307373, "learning_rate": 1.4867958504331995e-05, "loss": 0.16008667945861815, "step": 72080 }, { "epoch": 0.3580759564090083, "grad_norm": 2.253432035446167, "learning_rate": 1.4866553204553902e-05, "loss": 0.1634196400642395, "step": 72090 }, { "epoch": 0.35812562709237755, "grad_norm": 0.9169324040412903, "learning_rate": 1.4865147778831372e-05, "loss": 0.1605406880378723, "step": 72100 }, { "epoch": 0.3581752977757468, "grad_norm": 1.250985860824585, "learning_rate": 1.4863742227200772e-05, "loss": 0.1745868444442749, "step": 72110 }, { "epoch": 0.35822496845911606, "grad_norm": 1.310278296470642, "learning_rate": 1.4862336549698477e-05, "loss": 0.17623395919799806, "step": 72120 }, { "epoch": 0.3582746391424853, "grad_norm": 0.9175194501876831, "learning_rate": 1.4860930746360866e-05, "loss": 0.16678415536880492, "step": 72130 }, { "epoch": 0.35832430982585456, "grad_norm": 1.0552606582641602, "learning_rate": 1.4859524817224326e-05, "loss": 0.16883846521377563, "step": 72140 }, { "epoch": 0.35837398050922387, "grad_norm": 0.7520641088485718, "learning_rate": 1.4858118762325236e-05, "loss": 0.1634143590927124, "step": 72150 }, { "epoch": 0.3584236511925931, "grad_norm": 1.4739733934402466, "learning_rate": 1.4856712581699987e-05, "loss": 0.18161027431488036, "step": 72160 }, { "epoch": 0.35847332187596237, "grad_norm": 1.7679284811019897, "learning_rate": 1.4855306275384968e-05, "loss": 0.16815675497055055, "step": 72170 }, { "epoch": 0.3585229925593316, "grad_norm": 2.0300257205963135, "learning_rate": 1.4853899843416575e-05, "loss": 0.16809195280075073, "step": 72180 }, { "epoch": 0.3585726632427009, "grad_norm": 1.0422916412353516, "learning_rate": 1.4852493285831207e-05, "loss": 0.1640254020690918, "step": 72190 }, { "epoch": 0.3586223339260702, "grad_norm": 1.4584826231002808, "learning_rate": 1.485108660266526e-05, "loss": 0.18015495538711548, "step": 72200 }, { "epoch": 0.35867200460943943, "grad_norm": 1.4537163972854614, "learning_rate": 1.4849679793955146e-05, "loss": 0.1600005507469177, "step": 72210 }, { "epoch": 0.3587216752928087, "grad_norm": 1.0943034887313843, "learning_rate": 1.4848272859737267e-05, "loss": 0.16419434547424316, "step": 72220 }, { "epoch": 0.35877134597617794, "grad_norm": 1.4017181396484375, "learning_rate": 1.4846865800048037e-05, "loss": 0.16894814968109131, "step": 72230 }, { "epoch": 0.3588210166595472, "grad_norm": 2.764080286026001, "learning_rate": 1.484545861492387e-05, "loss": 0.159345281124115, "step": 72240 }, { "epoch": 0.35887068734291644, "grad_norm": 1.012247920036316, "learning_rate": 1.4844051304401181e-05, "loss": 0.17774641513824463, "step": 72250 }, { "epoch": 0.35892035802628575, "grad_norm": 1.8661997318267822, "learning_rate": 1.4842643868516388e-05, "loss": 0.1655946135520935, "step": 72260 }, { "epoch": 0.358970028709655, "grad_norm": 2.809420108795166, "learning_rate": 1.4841236307305925e-05, "loss": 0.16783876419067384, "step": 72270 }, { "epoch": 0.35901969939302425, "grad_norm": 0.719044029712677, "learning_rate": 1.483982862080621e-05, "loss": 0.16271507740020752, "step": 72280 }, { "epoch": 0.3590693700763935, "grad_norm": 1.0956395864486694, "learning_rate": 1.4838420809053677e-05, "loss": 0.15713629722595215, "step": 72290 }, { "epoch": 0.35911904075976275, "grad_norm": 1.3115977048873901, "learning_rate": 1.483701287208476e-05, "loss": 0.16628462076187134, "step": 72300 }, { "epoch": 0.35916871144313206, "grad_norm": 0.9246386885643005, "learning_rate": 1.4835604809935892e-05, "loss": 0.17663221359252929, "step": 72310 }, { "epoch": 0.3592183821265013, "grad_norm": 2.1194987297058105, "learning_rate": 1.483419662264352e-05, "loss": 0.16833778619766235, "step": 72320 }, { "epoch": 0.35926805280987056, "grad_norm": 1.292898416519165, "learning_rate": 1.483278831024408e-05, "loss": 0.16172664165496825, "step": 72330 }, { "epoch": 0.3593177234932398, "grad_norm": 1.02449631690979, "learning_rate": 1.4831379872774019e-05, "loss": 0.17858986854553222, "step": 72340 }, { "epoch": 0.35936739417660907, "grad_norm": 0.7940526008605957, "learning_rate": 1.4829971310269792e-05, "loss": 0.16427083015441896, "step": 72350 }, { "epoch": 0.3594170648599783, "grad_norm": 1.3589450120925903, "learning_rate": 1.4828562622767852e-05, "loss": 0.16755242347717286, "step": 72360 }, { "epoch": 0.3594667355433476, "grad_norm": 0.8534717559814453, "learning_rate": 1.482715381030465e-05, "loss": 0.1702576160430908, "step": 72370 }, { "epoch": 0.3595164062267169, "grad_norm": 1.5313879251480103, "learning_rate": 1.482574487291665e-05, "loss": 0.16422176361083984, "step": 72380 }, { "epoch": 0.35956607691008613, "grad_norm": 1.4672000408172607, "learning_rate": 1.4824335810640313e-05, "loss": 0.17079586982727052, "step": 72390 }, { "epoch": 0.3596157475934554, "grad_norm": 1.656030297279358, "learning_rate": 1.4822926623512104e-05, "loss": 0.1661424994468689, "step": 72400 }, { "epoch": 0.35966541827682463, "grad_norm": 0.9012424349784851, "learning_rate": 1.4821517311568495e-05, "loss": 0.19841268062591552, "step": 72410 }, { "epoch": 0.35971508896019394, "grad_norm": 1.3949060440063477, "learning_rate": 1.4820107874845955e-05, "loss": 0.1800084352493286, "step": 72420 }, { "epoch": 0.3597647596435632, "grad_norm": 1.2837815284729004, "learning_rate": 1.4818698313380961e-05, "loss": 0.1790919542312622, "step": 72430 }, { "epoch": 0.35981443032693244, "grad_norm": 1.8407913446426392, "learning_rate": 1.4817288627209993e-05, "loss": 0.16882919073104857, "step": 72440 }, { "epoch": 0.3598641010103017, "grad_norm": 1.040183424949646, "learning_rate": 1.4815878816369533e-05, "loss": 0.16769230365753174, "step": 72450 }, { "epoch": 0.35991377169367095, "grad_norm": 1.4640603065490723, "learning_rate": 1.4814468880896065e-05, "loss": 0.15885415077209472, "step": 72460 }, { "epoch": 0.3599634423770402, "grad_norm": 1.4552991390228271, "learning_rate": 1.4813058820826074e-05, "loss": 0.17163512706756592, "step": 72470 }, { "epoch": 0.3600131130604095, "grad_norm": 0.9494889378547668, "learning_rate": 1.481164863619606e-05, "loss": 0.1674177646636963, "step": 72480 }, { "epoch": 0.36006278374377876, "grad_norm": 1.0451158285140991, "learning_rate": 1.4810238327042514e-05, "loss": 0.14852298498153688, "step": 72490 }, { "epoch": 0.360112454427148, "grad_norm": 1.7996107339859009, "learning_rate": 1.4808827893401936e-05, "loss": 0.17515366077423095, "step": 72500 }, { "epoch": 0.36016212511051726, "grad_norm": 1.0840659141540527, "learning_rate": 1.4807417335310823e-05, "loss": 0.17075523138046264, "step": 72510 }, { "epoch": 0.3602117957938865, "grad_norm": 1.82577383518219, "learning_rate": 1.4806006652805682e-05, "loss": 0.17289746999740602, "step": 72520 }, { "epoch": 0.3602614664772558, "grad_norm": 1.660578966140747, "learning_rate": 1.4804595845923024e-05, "loss": 0.1665034055709839, "step": 72530 }, { "epoch": 0.3603111371606251, "grad_norm": 1.1362864971160889, "learning_rate": 1.4803184914699354e-05, "loss": 0.1655665874481201, "step": 72540 }, { "epoch": 0.3603608078439943, "grad_norm": 2.243671417236328, "learning_rate": 1.4801773859171191e-05, "loss": 0.18025113344192506, "step": 72550 }, { "epoch": 0.3604104785273636, "grad_norm": 1.5153656005859375, "learning_rate": 1.4800362679375051e-05, "loss": 0.16273045539855957, "step": 72560 }, { "epoch": 0.3604601492107328, "grad_norm": 1.3624788522720337, "learning_rate": 1.479895137534746e-05, "loss": 0.15832631587982177, "step": 72570 }, { "epoch": 0.3605098198941021, "grad_norm": 1.9939165115356445, "learning_rate": 1.479753994712493e-05, "loss": 0.1508535385131836, "step": 72580 }, { "epoch": 0.3605594905774714, "grad_norm": 0.998665452003479, "learning_rate": 1.4796128394743998e-05, "loss": 0.16695809364318848, "step": 72590 }, { "epoch": 0.36060916126084064, "grad_norm": 0.7624292969703674, "learning_rate": 1.4794716718241189e-05, "loss": 0.16112065315246582, "step": 72600 }, { "epoch": 0.3606588319442099, "grad_norm": 1.615967035293579, "learning_rate": 1.4793304917653042e-05, "loss": 0.17438007593154908, "step": 72610 }, { "epoch": 0.36070850262757914, "grad_norm": 1.5169477462768555, "learning_rate": 1.479189299301609e-05, "loss": 0.16101632118225098, "step": 72620 }, { "epoch": 0.3607581733109484, "grad_norm": 1.5437604188919067, "learning_rate": 1.4790480944366876e-05, "loss": 0.1902850866317749, "step": 72630 }, { "epoch": 0.3608078439943177, "grad_norm": 0.9284133911132812, "learning_rate": 1.478906877174194e-05, "loss": 0.15799038410186766, "step": 72640 }, { "epoch": 0.36085751467768695, "grad_norm": 0.9275867938995361, "learning_rate": 1.478765647517783e-05, "loss": 0.1660111665725708, "step": 72650 }, { "epoch": 0.3609071853610562, "grad_norm": 2.2159314155578613, "learning_rate": 1.4786244054711094e-05, "loss": 0.1786855936050415, "step": 72660 }, { "epoch": 0.36095685604442546, "grad_norm": 1.6487681865692139, "learning_rate": 1.4784831510378285e-05, "loss": 0.17242558002471925, "step": 72670 }, { "epoch": 0.3610065267277947, "grad_norm": 1.0955984592437744, "learning_rate": 1.4783418842215965e-05, "loss": 0.17834068536758424, "step": 72680 }, { "epoch": 0.36105619741116396, "grad_norm": 0.9852871894836426, "learning_rate": 1.4782006050260687e-05, "loss": 0.16382551193237305, "step": 72690 }, { "epoch": 0.36110586809453327, "grad_norm": 1.5661004781723022, "learning_rate": 1.4780593134549013e-05, "loss": 0.17449188232421875, "step": 72700 }, { "epoch": 0.3611555387779025, "grad_norm": 1.4997694492340088, "learning_rate": 1.4779180095117513e-05, "loss": 0.1628351092338562, "step": 72710 }, { "epoch": 0.36120520946127177, "grad_norm": 1.0441093444824219, "learning_rate": 1.4777766932002756e-05, "loss": 0.15977671146392822, "step": 72720 }, { "epoch": 0.361254880144641, "grad_norm": 1.1892380714416504, "learning_rate": 1.4776353645241308e-05, "loss": 0.1786017656326294, "step": 72730 }, { "epoch": 0.3613045508280103, "grad_norm": 1.0060338973999023, "learning_rate": 1.4774940234869748e-05, "loss": 0.16878082752227783, "step": 72740 }, { "epoch": 0.3613542215113796, "grad_norm": 0.9117385149002075, "learning_rate": 1.477352670092466e-05, "loss": 0.172129225730896, "step": 72750 }, { "epoch": 0.36140389219474883, "grad_norm": 1.0183172225952148, "learning_rate": 1.4772113043442615e-05, "loss": 0.16028552055358886, "step": 72760 }, { "epoch": 0.3614535628781181, "grad_norm": 1.711144208908081, "learning_rate": 1.4770699262460204e-05, "loss": 0.17619500160217286, "step": 72770 }, { "epoch": 0.36150323356148734, "grad_norm": 1.4729371070861816, "learning_rate": 1.4769285358014017e-05, "loss": 0.15224145650863646, "step": 72780 }, { "epoch": 0.3615529042448566, "grad_norm": 0.923591136932373, "learning_rate": 1.4767871330140638e-05, "loss": 0.15420703887939452, "step": 72790 }, { "epoch": 0.36160257492822584, "grad_norm": 2.1061534881591797, "learning_rate": 1.4766457178876668e-05, "loss": 0.17118773460388184, "step": 72800 }, { "epoch": 0.36165224561159515, "grad_norm": 1.7660341262817383, "learning_rate": 1.4765042904258703e-05, "loss": 0.16837966442108154, "step": 72810 }, { "epoch": 0.3617019162949644, "grad_norm": 1.1142116785049438, "learning_rate": 1.4763628506323345e-05, "loss": 0.17943713665008545, "step": 72820 }, { "epoch": 0.36175158697833365, "grad_norm": 1.759700059890747, "learning_rate": 1.4762213985107195e-05, "loss": 0.17455850839614867, "step": 72830 }, { "epoch": 0.3618012576617029, "grad_norm": 2.2838516235351562, "learning_rate": 1.4760799340646861e-05, "loss": 0.19120794534683228, "step": 72840 }, { "epoch": 0.36185092834507215, "grad_norm": 1.022098422050476, "learning_rate": 1.4759384572978957e-05, "loss": 0.16804285049438478, "step": 72850 }, { "epoch": 0.36190059902844146, "grad_norm": 1.0645346641540527, "learning_rate": 1.4757969682140092e-05, "loss": 0.17400752305984496, "step": 72860 }, { "epoch": 0.3619502697118107, "grad_norm": 1.712842583656311, "learning_rate": 1.4756554668166886e-05, "loss": 0.16738544702529906, "step": 72870 }, { "epoch": 0.36199994039517996, "grad_norm": 1.3392845392227173, "learning_rate": 1.4755139531095957e-05, "loss": 0.15680283308029175, "step": 72880 }, { "epoch": 0.3620496110785492, "grad_norm": 1.2928072214126587, "learning_rate": 1.4753724270963927e-05, "loss": 0.16888070106506348, "step": 72890 }, { "epoch": 0.36209928176191847, "grad_norm": 0.647149384021759, "learning_rate": 1.4752308887807427e-05, "loss": 0.16780202388763427, "step": 72900 }, { "epoch": 0.3621489524452877, "grad_norm": 1.0365568399429321, "learning_rate": 1.4750893381663085e-05, "loss": 0.17762646675109864, "step": 72910 }, { "epoch": 0.362198623128657, "grad_norm": 0.8073503375053406, "learning_rate": 1.4749477752567532e-05, "loss": 0.17854208946228028, "step": 72920 }, { "epoch": 0.3622482938120263, "grad_norm": 0.7698516845703125, "learning_rate": 1.47480620005574e-05, "loss": 0.16911779642105101, "step": 72930 }, { "epoch": 0.36229796449539553, "grad_norm": 1.0419059991836548, "learning_rate": 1.4746646125669335e-05, "loss": 0.15446323156356812, "step": 72940 }, { "epoch": 0.3623476351787648, "grad_norm": 1.1112221479415894, "learning_rate": 1.4745230127939977e-05, "loss": 0.17528946399688722, "step": 72950 }, { "epoch": 0.36239730586213403, "grad_norm": 1.509164810180664, "learning_rate": 1.4743814007405973e-05, "loss": 0.1789878487586975, "step": 72960 }, { "epoch": 0.3624469765455033, "grad_norm": 0.6405599117279053, "learning_rate": 1.4742397764103968e-05, "loss": 0.1671499252319336, "step": 72970 }, { "epoch": 0.3624966472288726, "grad_norm": 0.9414650201797485, "learning_rate": 1.4740981398070618e-05, "loss": 0.17026233673095703, "step": 72980 }, { "epoch": 0.36254631791224184, "grad_norm": 1.3367382287979126, "learning_rate": 1.4739564909342571e-05, "loss": 0.17276239395141602, "step": 72990 }, { "epoch": 0.3625959885956111, "grad_norm": 1.5637624263763428, "learning_rate": 1.4738148297956496e-05, "loss": 0.16848306655883788, "step": 73000 }, { "epoch": 0.36264565927898035, "grad_norm": 0.7568362355232239, "learning_rate": 1.4736731563949043e-05, "loss": 0.1763239860534668, "step": 73010 }, { "epoch": 0.3626953299623496, "grad_norm": 1.156854271888733, "learning_rate": 1.4735314707356883e-05, "loss": 0.15936708450317383, "step": 73020 }, { "epoch": 0.3627450006457189, "grad_norm": 1.3480180501937866, "learning_rate": 1.4733897728216681e-05, "loss": 0.16596791744232178, "step": 73030 }, { "epoch": 0.36279467132908816, "grad_norm": 1.2767633199691772, "learning_rate": 1.4732480626565113e-05, "loss": 0.16472854614257812, "step": 73040 }, { "epoch": 0.3628443420124574, "grad_norm": 1.1938050985336304, "learning_rate": 1.4731063402438847e-05, "loss": 0.16066381931304932, "step": 73050 }, { "epoch": 0.36289401269582666, "grad_norm": 1.0039325952529907, "learning_rate": 1.4729646055874565e-05, "loss": 0.1636607050895691, "step": 73060 }, { "epoch": 0.3629436833791959, "grad_norm": 2.0049586296081543, "learning_rate": 1.4728228586908942e-05, "loss": 0.16992721557617188, "step": 73070 }, { "epoch": 0.36299335406256517, "grad_norm": 2.178492784500122, "learning_rate": 1.4726810995578664e-05, "loss": 0.1780916213989258, "step": 73080 }, { "epoch": 0.3630430247459345, "grad_norm": 0.753254771232605, "learning_rate": 1.472539328192042e-05, "loss": 0.15541203022003175, "step": 73090 }, { "epoch": 0.3630926954293037, "grad_norm": 1.4689828157424927, "learning_rate": 1.4723975445970897e-05, "loss": 0.16832151412963867, "step": 73100 }, { "epoch": 0.363142366112673, "grad_norm": 1.2700536251068115, "learning_rate": 1.4722557487766788e-05, "loss": 0.1685873031616211, "step": 73110 }, { "epoch": 0.36319203679604223, "grad_norm": 0.9141748547554016, "learning_rate": 1.472113940734479e-05, "loss": 0.17161648273468016, "step": 73120 }, { "epoch": 0.3632417074794115, "grad_norm": 1.3508087396621704, "learning_rate": 1.4719721204741605e-05, "loss": 0.17080498933792115, "step": 73130 }, { "epoch": 0.3632913781627808, "grad_norm": 1.7616840600967407, "learning_rate": 1.4718302879993931e-05, "loss": 0.1668792724609375, "step": 73140 }, { "epoch": 0.36334104884615004, "grad_norm": 1.4396491050720215, "learning_rate": 1.4716884433138474e-05, "loss": 0.16402156352996827, "step": 73150 }, { "epoch": 0.3633907195295193, "grad_norm": 1.2101099491119385, "learning_rate": 1.4715465864211947e-05, "loss": 0.16692206859588624, "step": 73160 }, { "epoch": 0.36344039021288854, "grad_norm": 0.790846049785614, "learning_rate": 1.4714047173251059e-05, "loss": 0.16595954895019532, "step": 73170 }, { "epoch": 0.3634900608962578, "grad_norm": 1.371256947517395, "learning_rate": 1.4712628360292525e-05, "loss": 0.17808709144592286, "step": 73180 }, { "epoch": 0.36353973157962705, "grad_norm": 0.6975802183151245, "learning_rate": 1.4711209425373065e-05, "loss": 0.1689786434173584, "step": 73190 }, { "epoch": 0.36358940226299635, "grad_norm": 0.8364284038543701, "learning_rate": 1.4709790368529397e-05, "loss": 0.17862355709075928, "step": 73200 }, { "epoch": 0.3636390729463656, "grad_norm": 1.7664982080459595, "learning_rate": 1.4708371189798249e-05, "loss": 0.15630624294281006, "step": 73210 }, { "epoch": 0.36368874362973486, "grad_norm": 1.6921433210372925, "learning_rate": 1.470695188921635e-05, "loss": 0.16878056526184082, "step": 73220 }, { "epoch": 0.3637384143131041, "grad_norm": 0.9409610033035278, "learning_rate": 1.4705532466820428e-05, "loss": 0.17075824737548828, "step": 73230 }, { "epoch": 0.36378808499647336, "grad_norm": 1.7382323741912842, "learning_rate": 1.4704112922647214e-05, "loss": 0.17144474983215333, "step": 73240 }, { "epoch": 0.36383775567984267, "grad_norm": 1.0932650566101074, "learning_rate": 1.4702693256733453e-05, "loss": 0.15306297540664673, "step": 73250 }, { "epoch": 0.3638874263632119, "grad_norm": 1.036928653717041, "learning_rate": 1.4701273469115881e-05, "loss": 0.1759178400039673, "step": 73260 }, { "epoch": 0.36393709704658117, "grad_norm": 1.2735965251922607, "learning_rate": 1.469985355983124e-05, "loss": 0.17265475988388063, "step": 73270 }, { "epoch": 0.3639867677299504, "grad_norm": 1.2396196126937866, "learning_rate": 1.469843352891628e-05, "loss": 0.1720314860343933, "step": 73280 }, { "epoch": 0.3640364384133197, "grad_norm": 2.316110610961914, "learning_rate": 1.469701337640775e-05, "loss": 0.16611621379852295, "step": 73290 }, { "epoch": 0.3640861090966889, "grad_norm": 1.2203335762023926, "learning_rate": 1.46955931023424e-05, "loss": 0.17399611473083496, "step": 73300 }, { "epoch": 0.36413577978005823, "grad_norm": 0.6597286462783813, "learning_rate": 1.4694172706756992e-05, "loss": 0.1753218173980713, "step": 73310 }, { "epoch": 0.3641854504634275, "grad_norm": 0.937681257724762, "learning_rate": 1.469275218968828e-05, "loss": 0.17986973524093627, "step": 73320 }, { "epoch": 0.36423512114679674, "grad_norm": 0.8984641432762146, "learning_rate": 1.4691331551173027e-05, "loss": 0.1796309232711792, "step": 73330 }, { "epoch": 0.364284791830166, "grad_norm": 1.3945003747940063, "learning_rate": 1.4689910791248002e-05, "loss": 0.18379554748535157, "step": 73340 }, { "epoch": 0.36433446251353524, "grad_norm": 2.3498001098632812, "learning_rate": 1.4688489909949968e-05, "loss": 0.16320817470550536, "step": 73350 }, { "epoch": 0.36438413319690455, "grad_norm": 0.7451592683792114, "learning_rate": 1.4687068907315702e-05, "loss": 0.16904065608978272, "step": 73360 }, { "epoch": 0.3644338038802738, "grad_norm": 0.8635459542274475, "learning_rate": 1.4685647783381978e-05, "loss": 0.17660984992980958, "step": 73370 }, { "epoch": 0.36448347456364305, "grad_norm": 2.0017898082733154, "learning_rate": 1.4684226538185575e-05, "loss": 0.1723255157470703, "step": 73380 }, { "epoch": 0.3645331452470123, "grad_norm": 2.896869659423828, "learning_rate": 1.468280517176327e-05, "loss": 0.17161592245101928, "step": 73390 }, { "epoch": 0.36458281593038155, "grad_norm": 1.1990010738372803, "learning_rate": 1.468138368415185e-05, "loss": 0.15726279020309447, "step": 73400 }, { "epoch": 0.3646324866137508, "grad_norm": 1.3347450494766235, "learning_rate": 1.4679962075388105e-05, "loss": 0.16684131622314452, "step": 73410 }, { "epoch": 0.3646821572971201, "grad_norm": 1.578263759613037, "learning_rate": 1.4678540345508823e-05, "loss": 0.160447895526886, "step": 73420 }, { "epoch": 0.36473182798048936, "grad_norm": 1.0550448894500732, "learning_rate": 1.4677118494550798e-05, "loss": 0.16723990440368652, "step": 73430 }, { "epoch": 0.3647814986638586, "grad_norm": 1.1719274520874023, "learning_rate": 1.4675696522550824e-05, "loss": 0.1804363489151001, "step": 73440 }, { "epoch": 0.36483116934722787, "grad_norm": 0.708132266998291, "learning_rate": 1.4674274429545705e-05, "loss": 0.15967557430267335, "step": 73450 }, { "epoch": 0.3648808400305971, "grad_norm": 2.482896327972412, "learning_rate": 1.4672852215572243e-05, "loss": 0.1760185718536377, "step": 73460 }, { "epoch": 0.3649305107139664, "grad_norm": 2.559933662414551, "learning_rate": 1.4671429880667246e-05, "loss": 0.1705625295639038, "step": 73470 }, { "epoch": 0.3649801813973357, "grad_norm": 1.2648212909698486, "learning_rate": 1.4670007424867521e-05, "loss": 0.16434286832809447, "step": 73480 }, { "epoch": 0.36502985208070493, "grad_norm": 1.0393034219741821, "learning_rate": 1.4668584848209881e-05, "loss": 0.17671819925308227, "step": 73490 }, { "epoch": 0.3650795227640742, "grad_norm": 2.2545552253723145, "learning_rate": 1.4667162150731145e-05, "loss": 0.16813163757324218, "step": 73500 }, { "epoch": 0.36512919344744343, "grad_norm": 1.3626378774642944, "learning_rate": 1.4665739332468127e-05, "loss": 0.17042083740234376, "step": 73510 }, { "epoch": 0.3651788641308127, "grad_norm": 0.7446532249450684, "learning_rate": 1.4664316393457651e-05, "loss": 0.16942529678344725, "step": 73520 }, { "epoch": 0.365228534814182, "grad_norm": 0.8280197381973267, "learning_rate": 1.4662893333736544e-05, "loss": 0.17233588695526122, "step": 73530 }, { "epoch": 0.36527820549755124, "grad_norm": 0.47047120332717896, "learning_rate": 1.4661470153341632e-05, "loss": 0.17281794548034668, "step": 73540 }, { "epoch": 0.3653278761809205, "grad_norm": 2.147341728210449, "learning_rate": 1.4660046852309743e-05, "loss": 0.1793808102607727, "step": 73550 }, { "epoch": 0.36537754686428975, "grad_norm": 2.102541923522949, "learning_rate": 1.4658623430677719e-05, "loss": 0.1777801036834717, "step": 73560 }, { "epoch": 0.365427217547659, "grad_norm": 2.4550702571868896, "learning_rate": 1.4657199888482393e-05, "loss": 0.16850258111953736, "step": 73570 }, { "epoch": 0.3654768882310283, "grad_norm": 0.7920353412628174, "learning_rate": 1.4655776225760605e-05, "loss": 0.16568317413330078, "step": 73580 }, { "epoch": 0.36552655891439756, "grad_norm": 0.8514564037322998, "learning_rate": 1.4654352442549204e-05, "loss": 0.16698423624038697, "step": 73590 }, { "epoch": 0.3655762295977668, "grad_norm": 0.8695592880249023, "learning_rate": 1.4652928538885032e-05, "loss": 0.16833842992782594, "step": 73600 }, { "epoch": 0.36562590028113606, "grad_norm": 1.087281584739685, "learning_rate": 1.4651504514804937e-05, "loss": 0.1634742259979248, "step": 73610 }, { "epoch": 0.3656755709645053, "grad_norm": 1.1097629070281982, "learning_rate": 1.4650080370345779e-05, "loss": 0.17164504528045654, "step": 73620 }, { "epoch": 0.36572524164787457, "grad_norm": 1.0912879705429077, "learning_rate": 1.4648656105544407e-05, "loss": 0.17785688638687133, "step": 73630 }, { "epoch": 0.3657749123312439, "grad_norm": 1.6567336320877075, "learning_rate": 1.4647231720437687e-05, "loss": 0.1639650583267212, "step": 73640 }, { "epoch": 0.3658245830146131, "grad_norm": 0.957809567451477, "learning_rate": 1.4645807215062479e-05, "loss": 0.1707205057144165, "step": 73650 }, { "epoch": 0.3658742536979824, "grad_norm": 0.8893496990203857, "learning_rate": 1.4644382589455646e-05, "loss": 0.18669395446777343, "step": 73660 }, { "epoch": 0.36592392438135163, "grad_norm": 1.883196473121643, "learning_rate": 1.4642957843654063e-05, "loss": 0.17170411348342896, "step": 73670 }, { "epoch": 0.3659735950647209, "grad_norm": 2.0563480854034424, "learning_rate": 1.4641532977694592e-05, "loss": 0.16749954223632812, "step": 73680 }, { "epoch": 0.3660232657480902, "grad_norm": 1.1337872743606567, "learning_rate": 1.4640107991614117e-05, "loss": 0.18218681812286378, "step": 73690 }, { "epoch": 0.36607293643145944, "grad_norm": 1.2896052598953247, "learning_rate": 1.4638682885449513e-05, "loss": 0.16915273666381836, "step": 73700 }, { "epoch": 0.3661226071148287, "grad_norm": 0.8196660876274109, "learning_rate": 1.463725765923766e-05, "loss": 0.175700044631958, "step": 73710 }, { "epoch": 0.36617227779819794, "grad_norm": 1.4859216213226318, "learning_rate": 1.4635832313015446e-05, "loss": 0.1714865207672119, "step": 73720 }, { "epoch": 0.3662219484815672, "grad_norm": 1.3503789901733398, "learning_rate": 1.4634406846819754e-05, "loss": 0.18484551906585694, "step": 73730 }, { "epoch": 0.36627161916493645, "grad_norm": 1.8797659873962402, "learning_rate": 1.4632981260687476e-05, "loss": 0.17526445388793946, "step": 73740 }, { "epoch": 0.36632128984830575, "grad_norm": 1.3177690505981445, "learning_rate": 1.4631555554655506e-05, "loss": 0.16882126331329345, "step": 73750 }, { "epoch": 0.366370960531675, "grad_norm": 0.9858363270759583, "learning_rate": 1.463012972876074e-05, "loss": 0.17249213457107543, "step": 73760 }, { "epoch": 0.36642063121504426, "grad_norm": 0.7269831299781799, "learning_rate": 1.4628703783040078e-05, "loss": 0.18393810987472534, "step": 73770 }, { "epoch": 0.3664703018984135, "grad_norm": 2.183638334274292, "learning_rate": 1.4627277717530426e-05, "loss": 0.18171319961547852, "step": 73780 }, { "epoch": 0.36651997258178276, "grad_norm": 1.542719841003418, "learning_rate": 1.4625851532268686e-05, "loss": 0.174568510055542, "step": 73790 }, { "epoch": 0.36656964326515207, "grad_norm": 1.824997067451477, "learning_rate": 1.4624425227291768e-05, "loss": 0.17469229698181152, "step": 73800 }, { "epoch": 0.3666193139485213, "grad_norm": 0.9647706151008606, "learning_rate": 1.4622998802636583e-05, "loss": 0.16994121074676513, "step": 73810 }, { "epoch": 0.36666898463189057, "grad_norm": 1.5383788347244263, "learning_rate": 1.4621572258340049e-05, "loss": 0.1782918691635132, "step": 73820 }, { "epoch": 0.3667186553152598, "grad_norm": 0.9043261408805847, "learning_rate": 1.4620145594439082e-05, "loss": 0.16701287031173706, "step": 73830 }, { "epoch": 0.3667683259986291, "grad_norm": 1.9274595975875854, "learning_rate": 1.4618718810970606e-05, "loss": 0.16562697887420655, "step": 73840 }, { "epoch": 0.3668179966819983, "grad_norm": 0.5633271336555481, "learning_rate": 1.4617291907971545e-05, "loss": 0.1718538761138916, "step": 73850 }, { "epoch": 0.36686766736536763, "grad_norm": 1.1116163730621338, "learning_rate": 1.4615864885478826e-05, "loss": 0.168021559715271, "step": 73860 }, { "epoch": 0.3669173380487369, "grad_norm": 1.8455885648727417, "learning_rate": 1.461443774352938e-05, "loss": 0.17862884998321532, "step": 73870 }, { "epoch": 0.36696700873210614, "grad_norm": 2.0972037315368652, "learning_rate": 1.4613010482160137e-05, "loss": 0.1601511836051941, "step": 73880 }, { "epoch": 0.3670166794154754, "grad_norm": 0.9126210808753967, "learning_rate": 1.461158310140804e-05, "loss": 0.18274304866790772, "step": 73890 }, { "epoch": 0.36706635009884464, "grad_norm": 1.49635910987854, "learning_rate": 1.4610155601310026e-05, "loss": 0.1822413206100464, "step": 73900 }, { "epoch": 0.36711602078221395, "grad_norm": 0.9381679892539978, "learning_rate": 1.4608727981903039e-05, "loss": 0.18358060121536254, "step": 73910 }, { "epoch": 0.3671656914655832, "grad_norm": 1.0269831418991089, "learning_rate": 1.4607300243224027e-05, "loss": 0.15300395488739013, "step": 73920 }, { "epoch": 0.36721536214895245, "grad_norm": 1.2409273386001587, "learning_rate": 1.4605872385309936e-05, "loss": 0.17057263851165771, "step": 73930 }, { "epoch": 0.3672650328323217, "grad_norm": 1.3344812393188477, "learning_rate": 1.4604444408197717e-05, "loss": 0.17419250011444093, "step": 73940 }, { "epoch": 0.36731470351569095, "grad_norm": 2.1235198974609375, "learning_rate": 1.4603016311924329e-05, "loss": 0.18721387386322022, "step": 73950 }, { "epoch": 0.3673643741990602, "grad_norm": 0.9630357027053833, "learning_rate": 1.4601588096526733e-05, "loss": 0.15794198513031005, "step": 73960 }, { "epoch": 0.3674140448824295, "grad_norm": 0.8750341534614563, "learning_rate": 1.4600159762041886e-05, "loss": 0.17803955078125, "step": 73970 }, { "epoch": 0.36746371556579877, "grad_norm": 0.9971085786819458, "learning_rate": 1.4598731308506753e-05, "loss": 0.1709439277648926, "step": 73980 }, { "epoch": 0.367513386249168, "grad_norm": 1.4001033306121826, "learning_rate": 1.4597302735958304e-05, "loss": 0.16369140148162842, "step": 73990 }, { "epoch": 0.36756305693253727, "grad_norm": 1.2418135404586792, "learning_rate": 1.4595874044433506e-05, "loss": 0.16101608276367188, "step": 74000 }, { "epoch": 0.3676127276159065, "grad_norm": 1.4102511405944824, "learning_rate": 1.4594445233969338e-05, "loss": 0.18190832138061525, "step": 74010 }, { "epoch": 0.3676623982992758, "grad_norm": 1.3951205015182495, "learning_rate": 1.4593016304602771e-05, "loss": 0.17459595203399658, "step": 74020 }, { "epoch": 0.3677120689826451, "grad_norm": 1.4567370414733887, "learning_rate": 1.4591587256370792e-05, "loss": 0.16966105699539186, "step": 74030 }, { "epoch": 0.36776173966601433, "grad_norm": 1.628405213356018, "learning_rate": 1.4590158089310383e-05, "loss": 0.17144205570220947, "step": 74040 }, { "epoch": 0.3678114103493836, "grad_norm": 2.016413688659668, "learning_rate": 1.4588728803458525e-05, "loss": 0.1788345456123352, "step": 74050 }, { "epoch": 0.36786108103275283, "grad_norm": 1.1448966264724731, "learning_rate": 1.458729939885221e-05, "loss": 0.16633026599884032, "step": 74060 }, { "epoch": 0.3679107517161221, "grad_norm": 1.2526582479476929, "learning_rate": 1.4585869875528434e-05, "loss": 0.16884911060333252, "step": 74070 }, { "epoch": 0.3679604223994914, "grad_norm": 0.7874431610107422, "learning_rate": 1.4584440233524188e-05, "loss": 0.177399206161499, "step": 74080 }, { "epoch": 0.36801009308286065, "grad_norm": 0.6791880130767822, "learning_rate": 1.4583010472876473e-05, "loss": 0.16767829656600952, "step": 74090 }, { "epoch": 0.3680597637662299, "grad_norm": 1.4165681600570679, "learning_rate": 1.4581580593622289e-05, "loss": 0.17060608863830568, "step": 74100 }, { "epoch": 0.36810943444959915, "grad_norm": 1.327356219291687, "learning_rate": 1.4580150595798641e-05, "loss": 0.16517434120178223, "step": 74110 }, { "epoch": 0.3681591051329684, "grad_norm": 1.4698377847671509, "learning_rate": 1.4578720479442538e-05, "loss": 0.17430354356765748, "step": 74120 }, { "epoch": 0.3682087758163377, "grad_norm": 1.046094536781311, "learning_rate": 1.4577290244590988e-05, "loss": 0.16827046871185303, "step": 74130 }, { "epoch": 0.36825844649970696, "grad_norm": 2.2382662296295166, "learning_rate": 1.457585989128101e-05, "loss": 0.1772404432296753, "step": 74140 }, { "epoch": 0.3683081171830762, "grad_norm": 1.776218295097351, "learning_rate": 1.4574429419549617e-05, "loss": 0.16456949710845947, "step": 74150 }, { "epoch": 0.36835778786644546, "grad_norm": 0.5167977809906006, "learning_rate": 1.4572998829433829e-05, "loss": 0.17103886604309082, "step": 74160 }, { "epoch": 0.3684074585498147, "grad_norm": 1.417842149734497, "learning_rate": 1.457156812097067e-05, "loss": 0.16130359172821046, "step": 74170 }, { "epoch": 0.36845712923318397, "grad_norm": 1.5960103273391724, "learning_rate": 1.4570137294197166e-05, "loss": 0.18093181848526002, "step": 74180 }, { "epoch": 0.3685067999165533, "grad_norm": 0.9506289958953857, "learning_rate": 1.4568706349150348e-05, "loss": 0.16917977333068848, "step": 74190 }, { "epoch": 0.3685564705999225, "grad_norm": 1.316416621208191, "learning_rate": 1.4567275285867243e-05, "loss": 0.16155812740325928, "step": 74200 }, { "epoch": 0.3686061412832918, "grad_norm": 0.7125769853591919, "learning_rate": 1.4565844104384894e-05, "loss": 0.1583313226699829, "step": 74210 }, { "epoch": 0.36865581196666103, "grad_norm": 1.507411003112793, "learning_rate": 1.4564412804740333e-05, "loss": 0.1715974450111389, "step": 74220 }, { "epoch": 0.3687054826500303, "grad_norm": 2.0927891731262207, "learning_rate": 1.4562981386970604e-05, "loss": 0.17260947227478027, "step": 74230 }, { "epoch": 0.3687551533333996, "grad_norm": 1.9401668310165405, "learning_rate": 1.4561549851112753e-05, "loss": 0.17386929988861083, "step": 74240 }, { "epoch": 0.36880482401676884, "grad_norm": 1.0159236192703247, "learning_rate": 1.4560118197203824e-05, "loss": 0.1772047758102417, "step": 74250 }, { "epoch": 0.3688544947001381, "grad_norm": 1.7273980379104614, "learning_rate": 1.4558686425280874e-05, "loss": 0.17064647674560546, "step": 74260 }, { "epoch": 0.36890416538350734, "grad_norm": 1.1847301721572876, "learning_rate": 1.4557254535380948e-05, "loss": 0.1539454221725464, "step": 74270 }, { "epoch": 0.3689538360668766, "grad_norm": 1.158725380897522, "learning_rate": 1.4555822527541107e-05, "loss": 0.15892348289489747, "step": 74280 }, { "epoch": 0.36900350675024585, "grad_norm": 1.0054645538330078, "learning_rate": 1.4554390401798415e-05, "loss": 0.16863090991973878, "step": 74290 }, { "epoch": 0.36905317743361515, "grad_norm": 1.3466309309005737, "learning_rate": 1.4552958158189926e-05, "loss": 0.18307719230651856, "step": 74300 }, { "epoch": 0.3691028481169844, "grad_norm": 2.255340337753296, "learning_rate": 1.4551525796752715e-05, "loss": 0.16443290710449218, "step": 74310 }, { "epoch": 0.36915251880035366, "grad_norm": 1.5278892517089844, "learning_rate": 1.4550093317523843e-05, "loss": 0.15546700954437256, "step": 74320 }, { "epoch": 0.3692021894837229, "grad_norm": 0.849463701248169, "learning_rate": 1.4548660720540387e-05, "loss": 0.15958930253982545, "step": 74330 }, { "epoch": 0.36925186016709216, "grad_norm": 0.6855009198188782, "learning_rate": 1.4547228005839418e-05, "loss": 0.16271984577178955, "step": 74340 }, { "epoch": 0.36930153085046147, "grad_norm": 0.6741185784339905, "learning_rate": 1.4545795173458018e-05, "loss": 0.16043342351913453, "step": 74350 }, { "epoch": 0.3693512015338307, "grad_norm": 1.0763168334960938, "learning_rate": 1.4544362223433267e-05, "loss": 0.16482772827148437, "step": 74360 }, { "epoch": 0.36940087221719997, "grad_norm": 0.8293505311012268, "learning_rate": 1.4542929155802248e-05, "loss": 0.1795216679573059, "step": 74370 }, { "epoch": 0.3694505429005692, "grad_norm": 0.6454290747642517, "learning_rate": 1.4541495970602052e-05, "loss": 0.1622399091720581, "step": 74380 }, { "epoch": 0.3695002135839385, "grad_norm": 1.2466895580291748, "learning_rate": 1.4540062667869762e-05, "loss": 0.1644091486930847, "step": 74390 }, { "epoch": 0.3695498842673077, "grad_norm": 1.3765149116516113, "learning_rate": 1.453862924764248e-05, "loss": 0.16733245849609374, "step": 74400 }, { "epoch": 0.36959955495067703, "grad_norm": 0.900059700012207, "learning_rate": 1.4537195709957295e-05, "loss": 0.16347475051879884, "step": 74410 }, { "epoch": 0.3696492256340463, "grad_norm": 1.78762948513031, "learning_rate": 1.4535762054851308e-05, "loss": 0.16371325254440308, "step": 74420 }, { "epoch": 0.36969889631741554, "grad_norm": 1.0941168069839478, "learning_rate": 1.4534328282361627e-05, "loss": 0.17325923442840577, "step": 74430 }, { "epoch": 0.3697485670007848, "grad_norm": 1.8618956804275513, "learning_rate": 1.4532894392525352e-05, "loss": 0.16641533374786377, "step": 74440 }, { "epoch": 0.36979823768415404, "grad_norm": 3.177262306213379, "learning_rate": 1.453146038537959e-05, "loss": 0.17753299474716186, "step": 74450 }, { "epoch": 0.36984790836752335, "grad_norm": 1.0681328773498535, "learning_rate": 1.4530026260961458e-05, "loss": 0.18033962249755858, "step": 74460 }, { "epoch": 0.3698975790508926, "grad_norm": 1.5417308807373047, "learning_rate": 1.4528592019308067e-05, "loss": 0.16721056699752807, "step": 74470 }, { "epoch": 0.36994724973426185, "grad_norm": 1.2401450872421265, "learning_rate": 1.4527157660456535e-05, "loss": 0.16896772384643555, "step": 74480 }, { "epoch": 0.3699969204176311, "grad_norm": 1.005486249923706, "learning_rate": 1.4525723184443982e-05, "loss": 0.16962209939956666, "step": 74490 }, { "epoch": 0.37004659110100036, "grad_norm": 1.5290547609329224, "learning_rate": 1.4524288591307535e-05, "loss": 0.17769393920898438, "step": 74500 }, { "epoch": 0.3700962617843696, "grad_norm": 1.45984947681427, "learning_rate": 1.4522853881084318e-05, "loss": 0.15651811361312867, "step": 74510 }, { "epoch": 0.3701459324677389, "grad_norm": 1.4438613653182983, "learning_rate": 1.4521419053811461e-05, "loss": 0.17373225688934327, "step": 74520 }, { "epoch": 0.37019560315110817, "grad_norm": 1.1844016313552856, "learning_rate": 1.4519984109526098e-05, "loss": 0.16668221950531006, "step": 74530 }, { "epoch": 0.3702452738344774, "grad_norm": 1.1653518676757812, "learning_rate": 1.451854904826536e-05, "loss": 0.17827169895172118, "step": 74540 }, { "epoch": 0.37029494451784667, "grad_norm": 0.8267505168914795, "learning_rate": 1.4517113870066393e-05, "loss": 0.16774418354034423, "step": 74550 }, { "epoch": 0.3703446152012159, "grad_norm": 0.9829911589622498, "learning_rate": 1.4515678574966333e-05, "loss": 0.16919941902160646, "step": 74560 }, { "epoch": 0.3703942858845852, "grad_norm": 1.2494803667068481, "learning_rate": 1.4514243163002326e-05, "loss": 0.1629960536956787, "step": 74570 }, { "epoch": 0.3704439565679545, "grad_norm": 2.0661544799804688, "learning_rate": 1.4512807634211523e-05, "loss": 0.1539594292640686, "step": 74580 }, { "epoch": 0.37049362725132373, "grad_norm": 1.7066224813461304, "learning_rate": 1.4511371988631076e-05, "loss": 0.17828826904296874, "step": 74590 }, { "epoch": 0.370543297934693, "grad_norm": 0.8964051604270935, "learning_rate": 1.4509936226298133e-05, "loss": 0.15625003576278687, "step": 74600 }, { "epoch": 0.37059296861806224, "grad_norm": 0.6721506714820862, "learning_rate": 1.4508500347249852e-05, "loss": 0.15865607261657716, "step": 74610 }, { "epoch": 0.3706426393014315, "grad_norm": 0.8740583658218384, "learning_rate": 1.4507064351523398e-05, "loss": 0.15783405303955078, "step": 74620 }, { "epoch": 0.3706923099848008, "grad_norm": 1.0921427011489868, "learning_rate": 1.450562823915593e-05, "loss": 0.1661941647529602, "step": 74630 }, { "epoch": 0.37074198066817005, "grad_norm": 1.3917325735092163, "learning_rate": 1.4504192010184615e-05, "loss": 0.1619810104370117, "step": 74640 }, { "epoch": 0.3707916513515393, "grad_norm": 1.234269380569458, "learning_rate": 1.450275566464662e-05, "loss": 0.17224496603012085, "step": 74650 }, { "epoch": 0.37084132203490855, "grad_norm": 1.0848548412322998, "learning_rate": 1.450131920257912e-05, "loss": 0.17179106473922728, "step": 74660 }, { "epoch": 0.3708909927182778, "grad_norm": 1.7989013195037842, "learning_rate": 1.449988262401929e-05, "loss": 0.17936358451843262, "step": 74670 }, { "epoch": 0.37094066340164705, "grad_norm": 0.9200602173805237, "learning_rate": 1.4498445929004306e-05, "loss": 0.17773702144622802, "step": 74680 }, { "epoch": 0.37099033408501636, "grad_norm": 0.806104302406311, "learning_rate": 1.4497009117571348e-05, "loss": 0.1746842861175537, "step": 74690 }, { "epoch": 0.3710400047683856, "grad_norm": 0.8950415849685669, "learning_rate": 1.4495572189757604e-05, "loss": 0.16169631481170654, "step": 74700 }, { "epoch": 0.37108967545175486, "grad_norm": 0.5513341426849365, "learning_rate": 1.449413514560026e-05, "loss": 0.16010520458221436, "step": 74710 }, { "epoch": 0.3711393461351241, "grad_norm": 1.1108176708221436, "learning_rate": 1.4492697985136505e-05, "loss": 0.16177523136138916, "step": 74720 }, { "epoch": 0.37118901681849337, "grad_norm": 0.7659095525741577, "learning_rate": 1.4491260708403534e-05, "loss": 0.16381288766860963, "step": 74730 }, { "epoch": 0.3712386875018627, "grad_norm": 1.320866346359253, "learning_rate": 1.4489823315438538e-05, "loss": 0.1837838411331177, "step": 74740 }, { "epoch": 0.3712883581852319, "grad_norm": 0.812942624092102, "learning_rate": 1.4488385806278723e-05, "loss": 0.16409757137298583, "step": 74750 }, { "epoch": 0.3713380288686012, "grad_norm": 0.9567228555679321, "learning_rate": 1.4486948180961286e-05, "loss": 0.16819345951080322, "step": 74760 }, { "epoch": 0.37138769955197043, "grad_norm": 1.6931434869766235, "learning_rate": 1.4485510439523437e-05, "loss": 0.15533698797225953, "step": 74770 }, { "epoch": 0.3714373702353397, "grad_norm": 0.8549951314926147, "learning_rate": 1.448407258200238e-05, "loss": 0.15369616746902465, "step": 74780 }, { "epoch": 0.37148704091870893, "grad_norm": 0.6699059009552002, "learning_rate": 1.4482634608435329e-05, "loss": 0.1702527403831482, "step": 74790 }, { "epoch": 0.37153671160207824, "grad_norm": 0.5908893942832947, "learning_rate": 1.4481196518859495e-05, "loss": 0.15558176040649413, "step": 74800 }, { "epoch": 0.3715863822854475, "grad_norm": 1.3090033531188965, "learning_rate": 1.4479758313312098e-05, "loss": 0.18308045864105224, "step": 74810 }, { "epoch": 0.37163605296881674, "grad_norm": 1.514904260635376, "learning_rate": 1.4478319991830353e-05, "loss": 0.16741042137145995, "step": 74820 }, { "epoch": 0.371685723652186, "grad_norm": 0.8357824683189392, "learning_rate": 1.4476881554451489e-05, "loss": 0.17909595966339112, "step": 74830 }, { "epoch": 0.37173539433555525, "grad_norm": 0.7354533076286316, "learning_rate": 1.4475443001212734e-05, "loss": 0.16840543746948242, "step": 74840 }, { "epoch": 0.37178506501892455, "grad_norm": 0.7936018705368042, "learning_rate": 1.447400433215131e-05, "loss": 0.16842540502548217, "step": 74850 }, { "epoch": 0.3718347357022938, "grad_norm": 1.2283810377120972, "learning_rate": 1.4472565547304456e-05, "loss": 0.16423530578613282, "step": 74860 }, { "epoch": 0.37188440638566306, "grad_norm": 1.2857660055160522, "learning_rate": 1.4471126646709402e-05, "loss": 0.18040405511856078, "step": 74870 }, { "epoch": 0.3719340770690323, "grad_norm": 0.6305510997772217, "learning_rate": 1.446968763040339e-05, "loss": 0.18489290475845338, "step": 74880 }, { "epoch": 0.37198374775240156, "grad_norm": 0.9609971046447754, "learning_rate": 1.4468248498423654e-05, "loss": 0.16055572032928467, "step": 74890 }, { "epoch": 0.3720334184357708, "grad_norm": 0.9385831356048584, "learning_rate": 1.4466809250807448e-05, "loss": 0.1675405979156494, "step": 74900 }, { "epoch": 0.3720830891191401, "grad_norm": 0.9175843596458435, "learning_rate": 1.4465369887592013e-05, "loss": 0.16393494606018066, "step": 74910 }, { "epoch": 0.37213275980250937, "grad_norm": 1.0777790546417236, "learning_rate": 1.4463930408814602e-05, "loss": 0.17362754344940184, "step": 74920 }, { "epoch": 0.3721824304858786, "grad_norm": 1.1047636270523071, "learning_rate": 1.4462490814512466e-05, "loss": 0.1817263960838318, "step": 74930 }, { "epoch": 0.3722321011692479, "grad_norm": 0.6216298341751099, "learning_rate": 1.4461051104722865e-05, "loss": 0.17307047843933104, "step": 74940 }, { "epoch": 0.3722817718526171, "grad_norm": 0.8626562356948853, "learning_rate": 1.445961127948305e-05, "loss": 0.17301928997039795, "step": 74950 }, { "epoch": 0.37233144253598643, "grad_norm": 0.962264358997345, "learning_rate": 1.4458171338830291e-05, "loss": 0.16947062015533448, "step": 74960 }, { "epoch": 0.3723811132193557, "grad_norm": 1.1029982566833496, "learning_rate": 1.445673128280185e-05, "loss": 0.16530461311340333, "step": 74970 }, { "epoch": 0.37243078390272494, "grad_norm": 1.260571837425232, "learning_rate": 1.4455291111434996e-05, "loss": 0.18569343090057372, "step": 74980 }, { "epoch": 0.3724804545860942, "grad_norm": 1.0266213417053223, "learning_rate": 1.4453850824767e-05, "loss": 0.15911664962768554, "step": 74990 }, { "epoch": 0.37253012526946344, "grad_norm": 1.1759744882583618, "learning_rate": 1.4452410422835135e-05, "loss": 0.18255991935729982, "step": 75000 }, { "epoch": 0.3725797959528327, "grad_norm": 1.2935415506362915, "learning_rate": 1.4450969905676677e-05, "loss": 0.16478352546691893, "step": 75010 }, { "epoch": 0.372629466636202, "grad_norm": 1.2941405773162842, "learning_rate": 1.4449529273328908e-05, "loss": 0.16589289903640747, "step": 75020 }, { "epoch": 0.37267913731957125, "grad_norm": 0.7668851613998413, "learning_rate": 1.4448088525829111e-05, "loss": 0.16138911247253418, "step": 75030 }, { "epoch": 0.3727288080029405, "grad_norm": 0.7574955224990845, "learning_rate": 1.444664766321457e-05, "loss": 0.1688835382461548, "step": 75040 }, { "epoch": 0.37277847868630976, "grad_norm": 2.143310070037842, "learning_rate": 1.4445206685522577e-05, "loss": 0.1607607126235962, "step": 75050 }, { "epoch": 0.372828149369679, "grad_norm": 1.1717844009399414, "learning_rate": 1.4443765592790423e-05, "loss": 0.1831686019897461, "step": 75060 }, { "epoch": 0.3728778200530483, "grad_norm": 1.219472050666809, "learning_rate": 1.4442324385055402e-05, "loss": 0.17119206190109254, "step": 75070 }, { "epoch": 0.37292749073641757, "grad_norm": 1.5466642379760742, "learning_rate": 1.4440883062354811e-05, "loss": 0.1671862006187439, "step": 75080 }, { "epoch": 0.3729771614197868, "grad_norm": 1.1560322046279907, "learning_rate": 1.4439441624725952e-05, "loss": 0.1784761667251587, "step": 75090 }, { "epoch": 0.37302683210315607, "grad_norm": 0.8259589076042175, "learning_rate": 1.4438000072206126e-05, "loss": 0.16094645261764526, "step": 75100 }, { "epoch": 0.3730765027865253, "grad_norm": 0.599139392375946, "learning_rate": 1.443655840483265e-05, "loss": 0.17549715042114258, "step": 75110 }, { "epoch": 0.3731261734698946, "grad_norm": 0.6784117817878723, "learning_rate": 1.4435116622642822e-05, "loss": 0.1648523211479187, "step": 75120 }, { "epoch": 0.3731758441532639, "grad_norm": 1.1188608407974243, "learning_rate": 1.4433674725673959e-05, "loss": 0.16643006801605226, "step": 75130 }, { "epoch": 0.37322551483663313, "grad_norm": 0.7656654715538025, "learning_rate": 1.4432232713963376e-05, "loss": 0.15945985317230224, "step": 75140 }, { "epoch": 0.3732751855200024, "grad_norm": 0.7946177124977112, "learning_rate": 1.4430790587548394e-05, "loss": 0.17221711874008178, "step": 75150 }, { "epoch": 0.37332485620337164, "grad_norm": 0.76509690284729, "learning_rate": 1.442934834646633e-05, "loss": 0.17971384525299072, "step": 75160 }, { "epoch": 0.3733745268867409, "grad_norm": 1.1841447353363037, "learning_rate": 1.4427905990754516e-05, "loss": 0.16860554218292237, "step": 75170 }, { "epoch": 0.3734241975701102, "grad_norm": 1.139727234840393, "learning_rate": 1.4426463520450274e-05, "loss": 0.16832976341247557, "step": 75180 }, { "epoch": 0.37347386825347945, "grad_norm": 0.8981103301048279, "learning_rate": 1.4425020935590934e-05, "loss": 0.17623252868652345, "step": 75190 }, { "epoch": 0.3735235389368487, "grad_norm": 0.9589015245437622, "learning_rate": 1.4423578236213835e-05, "loss": 0.15563780069351196, "step": 75200 }, { "epoch": 0.37357320962021795, "grad_norm": 1.0254175662994385, "learning_rate": 1.4422135422356307e-05, "loss": 0.16071705818176268, "step": 75210 }, { "epoch": 0.3736228803035872, "grad_norm": 1.3677092790603638, "learning_rate": 1.4420692494055692e-05, "loss": 0.16428419351577758, "step": 75220 }, { "epoch": 0.37367255098695645, "grad_norm": 0.9179742336273193, "learning_rate": 1.4419249451349332e-05, "loss": 0.17972222566604615, "step": 75230 }, { "epoch": 0.37372222167032576, "grad_norm": 3.1044232845306396, "learning_rate": 1.4417806294274576e-05, "loss": 0.17202503681182862, "step": 75240 }, { "epoch": 0.373771892353695, "grad_norm": 1.377592921257019, "learning_rate": 1.4416363022868768e-05, "loss": 0.16985301971435546, "step": 75250 }, { "epoch": 0.37382156303706426, "grad_norm": 0.8598089814186096, "learning_rate": 1.4414919637169262e-05, "loss": 0.17725775241851807, "step": 75260 }, { "epoch": 0.3738712337204335, "grad_norm": 2.0625572204589844, "learning_rate": 1.4413476137213411e-05, "loss": 0.17777835130691527, "step": 75270 }, { "epoch": 0.37392090440380277, "grad_norm": 0.6421163082122803, "learning_rate": 1.4412032523038572e-05, "loss": 0.15544459819793702, "step": 75280 }, { "epoch": 0.3739705750871721, "grad_norm": 1.0390945672988892, "learning_rate": 1.4410588794682102e-05, "loss": 0.15925544500350952, "step": 75290 }, { "epoch": 0.3740202457705413, "grad_norm": 1.209417462348938, "learning_rate": 1.4409144952181374e-05, "loss": 0.172524356842041, "step": 75300 }, { "epoch": 0.3740699164539106, "grad_norm": 1.67681884765625, "learning_rate": 1.4407700995573742e-05, "loss": 0.17298319339752197, "step": 75310 }, { "epoch": 0.37411958713727983, "grad_norm": 0.6399607062339783, "learning_rate": 1.4406256924896582e-05, "loss": 0.1697615146636963, "step": 75320 }, { "epoch": 0.3741692578206491, "grad_norm": 0.7546241879463196, "learning_rate": 1.4404812740187267e-05, "loss": 0.1787626028060913, "step": 75330 }, { "epoch": 0.37421892850401833, "grad_norm": 0.7583991885185242, "learning_rate": 1.4403368441483165e-05, "loss": 0.1778745174407959, "step": 75340 }, { "epoch": 0.37426859918738764, "grad_norm": 1.9636576175689697, "learning_rate": 1.440192402882166e-05, "loss": 0.17577526569366456, "step": 75350 }, { "epoch": 0.3743182698707569, "grad_norm": 1.4230185747146606, "learning_rate": 1.440047950224013e-05, "loss": 0.16817610263824462, "step": 75360 }, { "epoch": 0.37436794055412614, "grad_norm": 1.3871251344680786, "learning_rate": 1.4399034861775962e-05, "loss": 0.1688016414642334, "step": 75370 }, { "epoch": 0.3744176112374954, "grad_norm": 1.158767819404602, "learning_rate": 1.439759010746654e-05, "loss": 0.1722901701927185, "step": 75380 }, { "epoch": 0.37446728192086465, "grad_norm": 1.4856261014938354, "learning_rate": 1.4396145239349253e-05, "loss": 0.16269266605377197, "step": 75390 }, { "epoch": 0.37451695260423395, "grad_norm": 0.8453526496887207, "learning_rate": 1.4394700257461497e-05, "loss": 0.16417953968048096, "step": 75400 }, { "epoch": 0.3745666232876032, "grad_norm": 1.1064664125442505, "learning_rate": 1.4393255161840663e-05, "loss": 0.16616456508636473, "step": 75410 }, { "epoch": 0.37461629397097246, "grad_norm": 0.8021753430366516, "learning_rate": 1.4391809952524151e-05, "loss": 0.15857241153717042, "step": 75420 }, { "epoch": 0.3746659646543417, "grad_norm": 0.7974324226379395, "learning_rate": 1.4390364629549364e-05, "loss": 0.16670560836791992, "step": 75430 }, { "epoch": 0.37471563533771096, "grad_norm": 1.576871395111084, "learning_rate": 1.4388919192953708e-05, "loss": 0.15572690963745117, "step": 75440 }, { "epoch": 0.3747653060210802, "grad_norm": 0.8157987594604492, "learning_rate": 1.438747364277459e-05, "loss": 0.17271502017974855, "step": 75450 }, { "epoch": 0.3748149767044495, "grad_norm": 1.8427590131759644, "learning_rate": 1.4386027979049415e-05, "loss": 0.16855475902557374, "step": 75460 }, { "epoch": 0.3748646473878188, "grad_norm": 0.766968846321106, "learning_rate": 1.43845822018156e-05, "loss": 0.17071020603179932, "step": 75470 }, { "epoch": 0.374914318071188, "grad_norm": 1.497080683708191, "learning_rate": 1.438313631111056e-05, "loss": 0.17287728786468506, "step": 75480 }, { "epoch": 0.3749639887545573, "grad_norm": 0.9369868040084839, "learning_rate": 1.4381690306971712e-05, "loss": 0.17483876943588256, "step": 75490 }, { "epoch": 0.3750136594379265, "grad_norm": 0.8739784955978394, "learning_rate": 1.4380244189436483e-05, "loss": 0.15885238647460936, "step": 75500 }, { "epoch": 0.37506333012129583, "grad_norm": 1.2762268781661987, "learning_rate": 1.4378797958542296e-05, "loss": 0.17093977928161622, "step": 75510 }, { "epoch": 0.3751130008046651, "grad_norm": 1.0647475719451904, "learning_rate": 1.437735161432658e-05, "loss": 0.16550135612487793, "step": 75520 }, { "epoch": 0.37516267148803434, "grad_norm": 0.8601415753364563, "learning_rate": 1.4375905156826764e-05, "loss": 0.16260433197021484, "step": 75530 }, { "epoch": 0.3752123421714036, "grad_norm": 1.72654390335083, "learning_rate": 1.437445858608028e-05, "loss": 0.177948522567749, "step": 75540 }, { "epoch": 0.37526201285477284, "grad_norm": 1.1557120084762573, "learning_rate": 1.4373011902124568e-05, "loss": 0.17681918144226075, "step": 75550 }, { "epoch": 0.3753116835381421, "grad_norm": 2.0479695796966553, "learning_rate": 1.4371565104997063e-05, "loss": 0.1659458637237549, "step": 75560 }, { "epoch": 0.3753613542215114, "grad_norm": 0.9218238592147827, "learning_rate": 1.4370118194735215e-05, "loss": 0.17200397253036498, "step": 75570 }, { "epoch": 0.37541102490488065, "grad_norm": 1.2919642925262451, "learning_rate": 1.4368671171376462e-05, "loss": 0.16920762062072753, "step": 75580 }, { "epoch": 0.3754606955882499, "grad_norm": 1.8215395212173462, "learning_rate": 1.4367224034958259e-05, "loss": 0.16082794666290284, "step": 75590 }, { "epoch": 0.37551036627161916, "grad_norm": 0.7275006175041199, "learning_rate": 1.4365776785518054e-05, "loss": 0.17432185411453247, "step": 75600 }, { "epoch": 0.3755600369549884, "grad_norm": 0.8040982484817505, "learning_rate": 1.4364329423093301e-05, "loss": 0.16530191898345947, "step": 75610 }, { "epoch": 0.3756097076383577, "grad_norm": 0.7958574891090393, "learning_rate": 1.4362881947721454e-05, "loss": 0.1562395453453064, "step": 75620 }, { "epoch": 0.37565937832172697, "grad_norm": 1.3041983842849731, "learning_rate": 1.4361434359439979e-05, "loss": 0.1646030306816101, "step": 75630 }, { "epoch": 0.3757090490050962, "grad_norm": 1.1615347862243652, "learning_rate": 1.4359986658286336e-05, "loss": 0.179801082611084, "step": 75640 }, { "epoch": 0.37575871968846547, "grad_norm": 0.8790993094444275, "learning_rate": 1.4358538844297992e-05, "loss": 0.15963549613952638, "step": 75650 }, { "epoch": 0.3758083903718347, "grad_norm": 1.581809639930725, "learning_rate": 1.4357090917512414e-05, "loss": 0.17953672409057617, "step": 75660 }, { "epoch": 0.375858061055204, "grad_norm": 1.8408129215240479, "learning_rate": 1.4355642877967075e-05, "loss": 0.16651246547698975, "step": 75670 }, { "epoch": 0.3759077317385733, "grad_norm": 1.0993860960006714, "learning_rate": 1.4354194725699451e-05, "loss": 0.17315553426742553, "step": 75680 }, { "epoch": 0.37595740242194253, "grad_norm": 1.3709431886672974, "learning_rate": 1.4352746460747016e-05, "loss": 0.1770409107208252, "step": 75690 }, { "epoch": 0.3760070731053118, "grad_norm": 0.8724188208580017, "learning_rate": 1.4351298083147253e-05, "loss": 0.1687494158744812, "step": 75700 }, { "epoch": 0.37605674378868104, "grad_norm": 1.1568704843521118, "learning_rate": 1.4349849592937646e-05, "loss": 0.16983535289764404, "step": 75710 }, { "epoch": 0.3761064144720503, "grad_norm": 1.3302000761032104, "learning_rate": 1.4348400990155682e-05, "loss": 0.17233753204345703, "step": 75720 }, { "epoch": 0.3761560851554196, "grad_norm": 0.754006564617157, "learning_rate": 1.4346952274838848e-05, "loss": 0.17068151235580445, "step": 75730 }, { "epoch": 0.37620575583878885, "grad_norm": 0.6455565094947815, "learning_rate": 1.4345503447024638e-05, "loss": 0.16024231910705566, "step": 75740 }, { "epoch": 0.3762554265221581, "grad_norm": 1.3784962892532349, "learning_rate": 1.4344054506750542e-05, "loss": 0.16758142709732055, "step": 75750 }, { "epoch": 0.37630509720552735, "grad_norm": 1.0420444011688232, "learning_rate": 1.4342605454054065e-05, "loss": 0.1540437698364258, "step": 75760 }, { "epoch": 0.3763547678888966, "grad_norm": 1.1518337726593018, "learning_rate": 1.4341156288972704e-05, "loss": 0.18115242719650268, "step": 75770 }, { "epoch": 0.37640443857226585, "grad_norm": 2.058445930480957, "learning_rate": 1.4339707011543966e-05, "loss": 0.15536030530929565, "step": 75780 }, { "epoch": 0.37645410925563516, "grad_norm": 1.0241400003433228, "learning_rate": 1.4338257621805354e-05, "loss": 0.17849972248077392, "step": 75790 }, { "epoch": 0.3765037799390044, "grad_norm": 0.9377481937408447, "learning_rate": 1.4336808119794379e-05, "loss": 0.1735407829284668, "step": 75800 }, { "epoch": 0.37655345062237366, "grad_norm": 0.7282743453979492, "learning_rate": 1.4335358505548555e-05, "loss": 0.16647506952285768, "step": 75810 }, { "epoch": 0.3766031213057429, "grad_norm": 1.7697008848190308, "learning_rate": 1.4333908779105393e-05, "loss": 0.16310317516326905, "step": 75820 }, { "epoch": 0.37665279198911217, "grad_norm": 0.6984759569168091, "learning_rate": 1.4332458940502417e-05, "loss": 0.1791078567504883, "step": 75830 }, { "epoch": 0.3767024626724815, "grad_norm": 0.8973516225814819, "learning_rate": 1.4331008989777143e-05, "loss": 0.16872498989105225, "step": 75840 }, { "epoch": 0.3767521333558507, "grad_norm": 2.008140802383423, "learning_rate": 1.43295589269671e-05, "loss": 0.17782057523727418, "step": 75850 }, { "epoch": 0.37680180403922, "grad_norm": 0.8692159056663513, "learning_rate": 1.4328108752109813e-05, "loss": 0.17740199565887452, "step": 75860 }, { "epoch": 0.37685147472258923, "grad_norm": 0.9475188851356506, "learning_rate": 1.432665846524281e-05, "loss": 0.18072561025619507, "step": 75870 }, { "epoch": 0.3769011454059585, "grad_norm": 1.1708065271377563, "learning_rate": 1.4325208066403627e-05, "loss": 0.17273738384246826, "step": 75880 }, { "epoch": 0.37695081608932773, "grad_norm": 1.755223274230957, "learning_rate": 1.4323757555629799e-05, "loss": 0.15927139520645142, "step": 75890 }, { "epoch": 0.37700048677269704, "grad_norm": 1.01521897315979, "learning_rate": 1.4322306932958861e-05, "loss": 0.16381899118423462, "step": 75900 }, { "epoch": 0.3770501574560663, "grad_norm": 0.6240790486335754, "learning_rate": 1.4320856198428359e-05, "loss": 0.16838046312332153, "step": 75910 }, { "epoch": 0.37709982813943554, "grad_norm": 0.6328084468841553, "learning_rate": 1.4319405352075837e-05, "loss": 0.1747328519821167, "step": 75920 }, { "epoch": 0.3771494988228048, "grad_norm": 0.9432356357574463, "learning_rate": 1.431795439393884e-05, "loss": 0.16988577842712402, "step": 75930 }, { "epoch": 0.37719916950617405, "grad_norm": 0.8659235835075378, "learning_rate": 1.431650332405492e-05, "loss": 0.17365856170654298, "step": 75940 }, { "epoch": 0.37724884018954336, "grad_norm": 1.0022073984146118, "learning_rate": 1.4315052142461631e-05, "loss": 0.17993495464324952, "step": 75950 }, { "epoch": 0.3772985108729126, "grad_norm": 1.1447936296463013, "learning_rate": 1.4313600849196525e-05, "loss": 0.17100344896316527, "step": 75960 }, { "epoch": 0.37734818155628186, "grad_norm": 1.6006317138671875, "learning_rate": 1.4312149444297165e-05, "loss": 0.18930174112319947, "step": 75970 }, { "epoch": 0.3773978522396511, "grad_norm": 2.2385096549987793, "learning_rate": 1.4310697927801112e-05, "loss": 0.17960822582244873, "step": 75980 }, { "epoch": 0.37744752292302036, "grad_norm": 1.0329583883285522, "learning_rate": 1.4309246299745927e-05, "loss": 0.18167693614959718, "step": 75990 }, { "epoch": 0.3774971936063896, "grad_norm": 1.5759168863296509, "learning_rate": 1.4307794560169184e-05, "loss": 0.1735243797302246, "step": 76000 }, { "epoch": 0.3775468642897589, "grad_norm": 1.2107348442077637, "learning_rate": 1.4306342709108447e-05, "loss": 0.16207290887832643, "step": 76010 }, { "epoch": 0.3775965349731282, "grad_norm": 1.239504337310791, "learning_rate": 1.4304890746601294e-05, "loss": 0.17496882677078246, "step": 76020 }, { "epoch": 0.3776462056564974, "grad_norm": 1.7660578489303589, "learning_rate": 1.4303438672685298e-05, "loss": 0.15690834522247316, "step": 76030 }, { "epoch": 0.3776958763398667, "grad_norm": 3.327631711959839, "learning_rate": 1.4301986487398038e-05, "loss": 0.15680913925170897, "step": 76040 }, { "epoch": 0.37774554702323593, "grad_norm": 0.7529646158218384, "learning_rate": 1.43005341907771e-05, "loss": 0.1648906111717224, "step": 76050 }, { "epoch": 0.3777952177066052, "grad_norm": 1.6888562440872192, "learning_rate": 1.4299081782860067e-05, "loss": 0.16752889156341552, "step": 76060 }, { "epoch": 0.3778448883899745, "grad_norm": 0.6421963572502136, "learning_rate": 1.4297629263684526e-05, "loss": 0.1700900673866272, "step": 76070 }, { "epoch": 0.37789455907334374, "grad_norm": 1.4939544200897217, "learning_rate": 1.4296176633288066e-05, "loss": 0.1646356463432312, "step": 76080 }, { "epoch": 0.377944229756713, "grad_norm": 1.330397367477417, "learning_rate": 1.4294723891708282e-05, "loss": 0.17722685337066652, "step": 76090 }, { "epoch": 0.37799390044008224, "grad_norm": 1.7763782739639282, "learning_rate": 1.429327103898277e-05, "loss": 0.17564948797225952, "step": 76100 }, { "epoch": 0.3780435711234515, "grad_norm": 0.8587262034416199, "learning_rate": 1.4291818075149131e-05, "loss": 0.16207228899002074, "step": 76110 }, { "epoch": 0.3780932418068208, "grad_norm": 0.7325595021247864, "learning_rate": 1.4290365000244968e-05, "loss": 0.16747066974639893, "step": 76120 }, { "epoch": 0.37814291249019005, "grad_norm": 1.3737159967422485, "learning_rate": 1.4288911814307882e-05, "loss": 0.180175256729126, "step": 76130 }, { "epoch": 0.3781925831735593, "grad_norm": 0.9472572803497314, "learning_rate": 1.4287458517375483e-05, "loss": 0.170163893699646, "step": 76140 }, { "epoch": 0.37824225385692856, "grad_norm": 1.2104310989379883, "learning_rate": 1.4286005109485382e-05, "loss": 0.17818150520324708, "step": 76150 }, { "epoch": 0.3782919245402978, "grad_norm": 1.3004348278045654, "learning_rate": 1.4284551590675188e-05, "loss": 0.1611921787261963, "step": 76160 }, { "epoch": 0.37834159522366706, "grad_norm": 1.7129385471343994, "learning_rate": 1.4283097960982523e-05, "loss": 0.1688016176223755, "step": 76170 }, { "epoch": 0.37839126590703637, "grad_norm": 0.9096989631652832, "learning_rate": 1.428164422044501e-05, "loss": 0.15471329689025878, "step": 76180 }, { "epoch": 0.3784409365904056, "grad_norm": 1.8268065452575684, "learning_rate": 1.4280190369100262e-05, "loss": 0.16988224983215333, "step": 76190 }, { "epoch": 0.37849060727377487, "grad_norm": 0.7068082094192505, "learning_rate": 1.4278736406985909e-05, "loss": 0.14683828353881836, "step": 76200 }, { "epoch": 0.3785402779571441, "grad_norm": 2.193356513977051, "learning_rate": 1.4277282334139577e-05, "loss": 0.17986323833465576, "step": 76210 }, { "epoch": 0.3785899486405134, "grad_norm": 1.5165926218032837, "learning_rate": 1.42758281505989e-05, "loss": 0.19031039476394654, "step": 76220 }, { "epoch": 0.3786396193238827, "grad_norm": 1.3105628490447998, "learning_rate": 1.4274373856401509e-05, "loss": 0.17110711336135864, "step": 76230 }, { "epoch": 0.37868929000725193, "grad_norm": 1.3096442222595215, "learning_rate": 1.427291945158504e-05, "loss": 0.1814390778541565, "step": 76240 }, { "epoch": 0.3787389606906212, "grad_norm": 1.621473789215088, "learning_rate": 1.4271464936187136e-05, "loss": 0.17473843097686767, "step": 76250 }, { "epoch": 0.37878863137399044, "grad_norm": 0.9597707986831665, "learning_rate": 1.4270010310245434e-05, "loss": 0.16226835250854493, "step": 76260 }, { "epoch": 0.3788383020573597, "grad_norm": 0.9922694563865662, "learning_rate": 1.4268555573797584e-05, "loss": 0.16811513900756836, "step": 76270 }, { "epoch": 0.37888797274072894, "grad_norm": 1.231320858001709, "learning_rate": 1.4267100726881233e-05, "loss": 0.16783418655395507, "step": 76280 }, { "epoch": 0.37893764342409825, "grad_norm": 1.2512285709381104, "learning_rate": 1.4265645769534031e-05, "loss": 0.1617764949798584, "step": 76290 }, { "epoch": 0.3789873141074675, "grad_norm": 0.6866055727005005, "learning_rate": 1.426419070179363e-05, "loss": 0.16278645992279053, "step": 76300 }, { "epoch": 0.37903698479083675, "grad_norm": 1.4164278507232666, "learning_rate": 1.4262735523697688e-05, "loss": 0.19372531175613403, "step": 76310 }, { "epoch": 0.379086655474206, "grad_norm": 1.4784457683563232, "learning_rate": 1.4261280235283868e-05, "loss": 0.16428468227386475, "step": 76320 }, { "epoch": 0.37913632615757525, "grad_norm": 0.9430868625640869, "learning_rate": 1.4259824836589825e-05, "loss": 0.17590619325637818, "step": 76330 }, { "epoch": 0.37918599684094456, "grad_norm": 0.7261987924575806, "learning_rate": 1.425836932765323e-05, "loss": 0.15502814054489136, "step": 76340 }, { "epoch": 0.3792356675243138, "grad_norm": 0.7850475311279297, "learning_rate": 1.4256913708511749e-05, "loss": 0.1643740177154541, "step": 76350 }, { "epoch": 0.37928533820768306, "grad_norm": 1.4206230640411377, "learning_rate": 1.4255457979203052e-05, "loss": 0.1618252396583557, "step": 76360 }, { "epoch": 0.3793350088910523, "grad_norm": 1.500556230545044, "learning_rate": 1.425400213976481e-05, "loss": 0.15669382810592652, "step": 76370 }, { "epoch": 0.37938467957442157, "grad_norm": 0.9734668731689453, "learning_rate": 1.4252546190234708e-05, "loss": 0.17003055810928344, "step": 76380 }, { "epoch": 0.3794343502577908, "grad_norm": 1.3727672100067139, "learning_rate": 1.425109013065042e-05, "loss": 0.17078955173492433, "step": 76390 }, { "epoch": 0.3794840209411601, "grad_norm": 1.1061421632766724, "learning_rate": 1.4249633961049627e-05, "loss": 0.17688090801239015, "step": 76400 }, { "epoch": 0.3795336916245294, "grad_norm": 1.541678547859192, "learning_rate": 1.4248177681470016e-05, "loss": 0.1619655132293701, "step": 76410 }, { "epoch": 0.37958336230789863, "grad_norm": 1.6989586353302002, "learning_rate": 1.4246721291949276e-05, "loss": 0.16324191093444823, "step": 76420 }, { "epoch": 0.3796330329912679, "grad_norm": 1.4863972663879395, "learning_rate": 1.4245264792525097e-05, "loss": 0.17089576721191407, "step": 76430 }, { "epoch": 0.37968270367463713, "grad_norm": 0.8934770822525024, "learning_rate": 1.4243808183235169e-05, "loss": 0.15785400867462157, "step": 76440 }, { "epoch": 0.37973237435800644, "grad_norm": 1.3099305629730225, "learning_rate": 1.4242351464117194e-05, "loss": 0.157939875125885, "step": 76450 }, { "epoch": 0.3797820450413757, "grad_norm": 0.8797751665115356, "learning_rate": 1.424089463520887e-05, "loss": 0.16478737592697143, "step": 76460 }, { "epoch": 0.37983171572474494, "grad_norm": 1.9231318235397339, "learning_rate": 1.4239437696547897e-05, "loss": 0.18333230018615723, "step": 76470 }, { "epoch": 0.3798813864081142, "grad_norm": 0.8457016944885254, "learning_rate": 1.423798064817198e-05, "loss": 0.16863182783126832, "step": 76480 }, { "epoch": 0.37993105709148345, "grad_norm": 1.5804283618927002, "learning_rate": 1.423652349011883e-05, "loss": 0.17023295164108276, "step": 76490 }, { "epoch": 0.3799807277748527, "grad_norm": 0.8841096758842468, "learning_rate": 1.423506622242615e-05, "loss": 0.1747041702270508, "step": 76500 }, { "epoch": 0.380030398458222, "grad_norm": 0.7217008471488953, "learning_rate": 1.4233608845131666e-05, "loss": 0.16076452732086183, "step": 76510 }, { "epoch": 0.38008006914159126, "grad_norm": 0.8696116209030151, "learning_rate": 1.4232151358273083e-05, "loss": 0.16326965093612672, "step": 76520 }, { "epoch": 0.3801297398249605, "grad_norm": 2.0185372829437256, "learning_rate": 1.4230693761888126e-05, "loss": 0.16549227237701417, "step": 76530 }, { "epoch": 0.38017941050832976, "grad_norm": 1.1098475456237793, "learning_rate": 1.4229236056014517e-05, "loss": 0.18384871482849122, "step": 76540 }, { "epoch": 0.380229081191699, "grad_norm": 1.113569974899292, "learning_rate": 1.4227778240689976e-05, "loss": 0.1698479175567627, "step": 76550 }, { "epoch": 0.3802787518750683, "grad_norm": 1.0576399564743042, "learning_rate": 1.4226320315952235e-05, "loss": 0.1721799850463867, "step": 76560 }, { "epoch": 0.3803284225584376, "grad_norm": 1.5763468742370605, "learning_rate": 1.4224862281839025e-05, "loss": 0.17433103322982788, "step": 76570 }, { "epoch": 0.3803780932418068, "grad_norm": 1.6384785175323486, "learning_rate": 1.4223404138388078e-05, "loss": 0.1666215181350708, "step": 76580 }, { "epoch": 0.3804277639251761, "grad_norm": 1.777134656906128, "learning_rate": 1.4221945885637128e-05, "loss": 0.1614635705947876, "step": 76590 }, { "epoch": 0.38047743460854533, "grad_norm": 0.8690303564071655, "learning_rate": 1.4220487523623918e-05, "loss": 0.155311119556427, "step": 76600 }, { "epoch": 0.3805271052919146, "grad_norm": 0.9257383346557617, "learning_rate": 1.4219029052386186e-05, "loss": 0.15943868160247804, "step": 76610 }, { "epoch": 0.3805767759752839, "grad_norm": 1.3037528991699219, "learning_rate": 1.4217570471961682e-05, "loss": 0.15793194770812988, "step": 76620 }, { "epoch": 0.38062644665865314, "grad_norm": 0.6571459174156189, "learning_rate": 1.4216111782388146e-05, "loss": 0.18441240787506102, "step": 76630 }, { "epoch": 0.3806761173420224, "grad_norm": 0.673604428768158, "learning_rate": 1.4214652983703334e-05, "loss": 0.15814210176467897, "step": 76640 }, { "epoch": 0.38072578802539164, "grad_norm": 1.6017259359359741, "learning_rate": 1.4213194075944998e-05, "loss": 0.1716365933418274, "step": 76650 }, { "epoch": 0.3807754587087609, "grad_norm": 1.43612539768219, "learning_rate": 1.4211735059150895e-05, "loss": 0.1709444046020508, "step": 76660 }, { "epoch": 0.3808251293921302, "grad_norm": 0.7084667086601257, "learning_rate": 1.421027593335878e-05, "loss": 0.16690218448638916, "step": 76670 }, { "epoch": 0.38087480007549945, "grad_norm": 2.2427899837493896, "learning_rate": 1.4208816698606416e-05, "loss": 0.1734842538833618, "step": 76680 }, { "epoch": 0.3809244707588687, "grad_norm": 0.788405179977417, "learning_rate": 1.4207357354931568e-05, "loss": 0.1637183904647827, "step": 76690 }, { "epoch": 0.38097414144223796, "grad_norm": 1.7003204822540283, "learning_rate": 1.4205897902372004e-05, "loss": 0.17413864135742188, "step": 76700 }, { "epoch": 0.3810238121256072, "grad_norm": 1.2568211555480957, "learning_rate": 1.4204438340965494e-05, "loss": 0.177816641330719, "step": 76710 }, { "epoch": 0.38107348280897646, "grad_norm": 1.4733314514160156, "learning_rate": 1.420297867074981e-05, "loss": 0.17635905742645264, "step": 76720 }, { "epoch": 0.38112315349234577, "grad_norm": 1.641836404800415, "learning_rate": 1.4201518891762726e-05, "loss": 0.16940373182296753, "step": 76730 }, { "epoch": 0.381172824175715, "grad_norm": 0.9195664525032043, "learning_rate": 1.4200059004042025e-05, "loss": 0.1610671877861023, "step": 76740 }, { "epoch": 0.38122249485908427, "grad_norm": 0.7516357898712158, "learning_rate": 1.4198599007625486e-05, "loss": 0.16758023500442504, "step": 76750 }, { "epoch": 0.3812721655424535, "grad_norm": 1.0542562007904053, "learning_rate": 1.419713890255089e-05, "loss": 0.16326446533203126, "step": 76760 }, { "epoch": 0.3813218362258228, "grad_norm": 0.6666421890258789, "learning_rate": 1.419567868885603e-05, "loss": 0.17192473411560058, "step": 76770 }, { "epoch": 0.3813715069091921, "grad_norm": 1.0180734395980835, "learning_rate": 1.4194218366578692e-05, "loss": 0.16930854320526123, "step": 76780 }, { "epoch": 0.38142117759256133, "grad_norm": 0.658486008644104, "learning_rate": 1.4192757935756667e-05, "loss": 0.15918197631835937, "step": 76790 }, { "epoch": 0.3814708482759306, "grad_norm": 1.4995530843734741, "learning_rate": 1.4191297396427754e-05, "loss": 0.1644543170928955, "step": 76800 }, { "epoch": 0.38152051895929984, "grad_norm": 1.1654713153839111, "learning_rate": 1.4189836748629753e-05, "loss": 0.16734604835510253, "step": 76810 }, { "epoch": 0.3815701896426691, "grad_norm": 0.9900249242782593, "learning_rate": 1.4188375992400457e-05, "loss": 0.17321990728378295, "step": 76820 }, { "epoch": 0.38161986032603834, "grad_norm": 1.5086183547973633, "learning_rate": 1.4186915127777674e-05, "loss": 0.17112951278686522, "step": 76830 }, { "epoch": 0.38166953100940765, "grad_norm": 1.7783600091934204, "learning_rate": 1.4185454154799212e-05, "loss": 0.1824342727661133, "step": 76840 }, { "epoch": 0.3817192016927769, "grad_norm": 0.7636600732803345, "learning_rate": 1.4183993073502879e-05, "loss": 0.18242379426956176, "step": 76850 }, { "epoch": 0.38176887237614615, "grad_norm": 1.1587556600570679, "learning_rate": 1.4182531883926489e-05, "loss": 0.1807547092437744, "step": 76860 }, { "epoch": 0.3818185430595154, "grad_norm": 1.1856107711791992, "learning_rate": 1.4181070586107856e-05, "loss": 0.17488563060760498, "step": 76870 }, { "epoch": 0.38186821374288465, "grad_norm": 1.0683906078338623, "learning_rate": 1.4179609180084797e-05, "loss": 0.1599973440170288, "step": 76880 }, { "epoch": 0.38191788442625396, "grad_norm": 1.417697548866272, "learning_rate": 1.4178147665895132e-05, "loss": 0.16297731399536133, "step": 76890 }, { "epoch": 0.3819675551096232, "grad_norm": 1.086251974105835, "learning_rate": 1.4176686043576683e-05, "loss": 0.1753093957901001, "step": 76900 }, { "epoch": 0.38201722579299247, "grad_norm": 1.3603781461715698, "learning_rate": 1.4175224313167281e-05, "loss": 0.16013147830963134, "step": 76910 }, { "epoch": 0.3820668964763617, "grad_norm": 2.201063632965088, "learning_rate": 1.4173762474704752e-05, "loss": 0.16917738914489747, "step": 76920 }, { "epoch": 0.38211656715973097, "grad_norm": 1.2258107662200928, "learning_rate": 1.417230052822693e-05, "loss": 0.17623260021209716, "step": 76930 }, { "epoch": 0.3821662378431002, "grad_norm": 2.011699676513672, "learning_rate": 1.4170838473771649e-05, "loss": 0.17321977615356446, "step": 76940 }, { "epoch": 0.3822159085264695, "grad_norm": 1.6103507280349731, "learning_rate": 1.4169376311376743e-05, "loss": 0.17862597703933716, "step": 76950 }, { "epoch": 0.3822655792098388, "grad_norm": 1.1196190118789673, "learning_rate": 1.4167914041080055e-05, "loss": 0.17796716690063477, "step": 76960 }, { "epoch": 0.38231524989320803, "grad_norm": 0.9184053540229797, "learning_rate": 1.4166451662919426e-05, "loss": 0.1729533553123474, "step": 76970 }, { "epoch": 0.3823649205765773, "grad_norm": 1.1880003213882446, "learning_rate": 1.4164989176932707e-05, "loss": 0.16823406219482423, "step": 76980 }, { "epoch": 0.38241459125994653, "grad_norm": 1.7526705265045166, "learning_rate": 1.4163526583157739e-05, "loss": 0.1667715787887573, "step": 76990 }, { "epoch": 0.38246426194331584, "grad_norm": 1.4325200319290161, "learning_rate": 1.416206388163238e-05, "loss": 0.16548993587493896, "step": 77000 }, { "epoch": 0.3825139326266851, "grad_norm": 0.6783562302589417, "learning_rate": 1.416060107239448e-05, "loss": 0.16439497470855713, "step": 77010 }, { "epoch": 0.38256360331005435, "grad_norm": 1.3558117151260376, "learning_rate": 1.4159138155481897e-05, "loss": 0.18965481519699096, "step": 77020 }, { "epoch": 0.3826132739934236, "grad_norm": 1.4591478109359741, "learning_rate": 1.4157675130932491e-05, "loss": 0.15417163372039794, "step": 77030 }, { "epoch": 0.38266294467679285, "grad_norm": 0.7137195467948914, "learning_rate": 1.4156211998784121e-05, "loss": 0.16690632104873657, "step": 77040 }, { "epoch": 0.3827126153601621, "grad_norm": 1.4263437986373901, "learning_rate": 1.4154748759074659e-05, "loss": 0.17878403663635253, "step": 77050 }, { "epoch": 0.3827622860435314, "grad_norm": 0.8520589470863342, "learning_rate": 1.415328541184197e-05, "loss": 0.17190244197845458, "step": 77060 }, { "epoch": 0.38281195672690066, "grad_norm": 1.493889331817627, "learning_rate": 1.4151821957123923e-05, "loss": 0.16551640033721923, "step": 77070 }, { "epoch": 0.3828616274102699, "grad_norm": 1.141082525253296, "learning_rate": 1.4150358394958395e-05, "loss": 0.16351912021636963, "step": 77080 }, { "epoch": 0.38291129809363916, "grad_norm": 2.2413432598114014, "learning_rate": 1.4148894725383259e-05, "loss": 0.1671057939529419, "step": 77090 }, { "epoch": 0.3829609687770084, "grad_norm": 1.4270308017730713, "learning_rate": 1.4147430948436398e-05, "loss": 0.17093515396118164, "step": 77100 }, { "epoch": 0.3830106394603777, "grad_norm": 0.9712519645690918, "learning_rate": 1.4145967064155688e-05, "loss": 0.16282670497894286, "step": 77110 }, { "epoch": 0.383060310143747, "grad_norm": 0.8317397236824036, "learning_rate": 1.4144503072579019e-05, "loss": 0.16762754917144776, "step": 77120 }, { "epoch": 0.3831099808271162, "grad_norm": 1.7775115966796875, "learning_rate": 1.4143038973744276e-05, "loss": 0.18542113304138183, "step": 77130 }, { "epoch": 0.3831596515104855, "grad_norm": 1.8832758665084839, "learning_rate": 1.4141574767689354e-05, "loss": 0.1769067406654358, "step": 77140 }, { "epoch": 0.38320932219385473, "grad_norm": 1.042873740196228, "learning_rate": 1.414011045445214e-05, "loss": 0.16685287952423095, "step": 77150 }, { "epoch": 0.383258992877224, "grad_norm": 1.0699468851089478, "learning_rate": 1.4138646034070533e-05, "loss": 0.16116852760314943, "step": 77160 }, { "epoch": 0.3833086635605933, "grad_norm": 0.7833046913146973, "learning_rate": 1.4137181506582427e-05, "loss": 0.1804700016975403, "step": 77170 }, { "epoch": 0.38335833424396254, "grad_norm": 1.422054409980774, "learning_rate": 1.4135716872025728e-05, "loss": 0.1591806411743164, "step": 77180 }, { "epoch": 0.3834080049273318, "grad_norm": 0.7697708606719971, "learning_rate": 1.4134252130438344e-05, "loss": 0.1545642375946045, "step": 77190 }, { "epoch": 0.38345767561070104, "grad_norm": 1.3016414642333984, "learning_rate": 1.4132787281858174e-05, "loss": 0.16919516324996947, "step": 77200 }, { "epoch": 0.3835073462940703, "grad_norm": 0.8926017880439758, "learning_rate": 1.413132232632313e-05, "loss": 0.16933796405792237, "step": 77210 }, { "epoch": 0.3835570169774396, "grad_norm": 1.0151164531707764, "learning_rate": 1.4129857263871126e-05, "loss": 0.1793671131134033, "step": 77220 }, { "epoch": 0.38360668766080885, "grad_norm": 1.9335970878601074, "learning_rate": 1.4128392094540078e-05, "loss": 0.15650779008865356, "step": 77230 }, { "epoch": 0.3836563583441781, "grad_norm": 0.9148843288421631, "learning_rate": 1.4126926818367898e-05, "loss": 0.16620166301727296, "step": 77240 }, { "epoch": 0.38370602902754736, "grad_norm": 3.141615152359009, "learning_rate": 1.4125461435392516e-05, "loss": 0.16120336055755616, "step": 77250 }, { "epoch": 0.3837556997109166, "grad_norm": 0.9732618927955627, "learning_rate": 1.412399594565185e-05, "loss": 0.1674007773399353, "step": 77260 }, { "epoch": 0.38380537039428586, "grad_norm": 1.058489441871643, "learning_rate": 1.4122530349183828e-05, "loss": 0.1756507158279419, "step": 77270 }, { "epoch": 0.38385504107765517, "grad_norm": 1.666746973991394, "learning_rate": 1.4121064646026375e-05, "loss": 0.1850329279899597, "step": 77280 }, { "epoch": 0.3839047117610244, "grad_norm": 1.3848153352737427, "learning_rate": 1.411959883621743e-05, "loss": 0.16512668132781982, "step": 77290 }, { "epoch": 0.38395438244439367, "grad_norm": 1.0024430751800537, "learning_rate": 1.4118132919794921e-05, "loss": 0.15431654453277588, "step": 77300 }, { "epoch": 0.3840040531277629, "grad_norm": 1.087773323059082, "learning_rate": 1.4116666896796787e-05, "loss": 0.17478663921356202, "step": 77310 }, { "epoch": 0.3840537238111322, "grad_norm": 0.6582472920417786, "learning_rate": 1.411520076726097e-05, "loss": 0.16545554399490356, "step": 77320 }, { "epoch": 0.3841033944945015, "grad_norm": 0.7555594444274902, "learning_rate": 1.4113734531225412e-05, "loss": 0.17330509424209595, "step": 77330 }, { "epoch": 0.38415306517787073, "grad_norm": 1.1439034938812256, "learning_rate": 1.411226818872806e-05, "loss": 0.185597026348114, "step": 77340 }, { "epoch": 0.38420273586124, "grad_norm": 1.7365847826004028, "learning_rate": 1.4110801739806858e-05, "loss": 0.16372547149658204, "step": 77350 }, { "epoch": 0.38425240654460924, "grad_norm": 0.7574490904808044, "learning_rate": 1.4109335184499762e-05, "loss": 0.17039235830307006, "step": 77360 }, { "epoch": 0.3843020772279785, "grad_norm": 1.2135488986968994, "learning_rate": 1.410786852284472e-05, "loss": 0.157196569442749, "step": 77370 }, { "epoch": 0.38435174791134774, "grad_norm": 1.0563901662826538, "learning_rate": 1.4106401754879695e-05, "loss": 0.17192296981811522, "step": 77380 }, { "epoch": 0.38440141859471705, "grad_norm": 1.776594877243042, "learning_rate": 1.4104934880642644e-05, "loss": 0.1718371868133545, "step": 77390 }, { "epoch": 0.3844510892780863, "grad_norm": 1.233604907989502, "learning_rate": 1.4103467900171528e-05, "loss": 0.17254936695098877, "step": 77400 }, { "epoch": 0.38450075996145555, "grad_norm": 2.3475558757781982, "learning_rate": 1.4102000813504315e-05, "loss": 0.1760622024536133, "step": 77410 }, { "epoch": 0.3845504306448248, "grad_norm": 1.7184149026870728, "learning_rate": 1.410053362067897e-05, "loss": 0.1672133445739746, "step": 77420 }, { "epoch": 0.38460010132819406, "grad_norm": 1.438051462173462, "learning_rate": 1.4099066321733461e-05, "loss": 0.17013825178146363, "step": 77430 }, { "epoch": 0.38464977201156336, "grad_norm": 1.1188884973526, "learning_rate": 1.4097598916705764e-05, "loss": 0.1765339970588684, "step": 77440 }, { "epoch": 0.3846994426949326, "grad_norm": 0.9295049905776978, "learning_rate": 1.4096131405633855e-05, "loss": 0.16437151432037353, "step": 77450 }, { "epoch": 0.38474911337830187, "grad_norm": 0.949530303478241, "learning_rate": 1.4094663788555711e-05, "loss": 0.18522074222564697, "step": 77460 }, { "epoch": 0.3847987840616711, "grad_norm": 0.9582059383392334, "learning_rate": 1.409319606550932e-05, "loss": 0.1678473949432373, "step": 77470 }, { "epoch": 0.38484845474504037, "grad_norm": 0.9206587672233582, "learning_rate": 1.4091728236532655e-05, "loss": 0.16457027196884155, "step": 77480 }, { "epoch": 0.3848981254284096, "grad_norm": 1.0697625875473022, "learning_rate": 1.4090260301663712e-05, "loss": 0.16245698928833008, "step": 77490 }, { "epoch": 0.38494779611177893, "grad_norm": 2.2604315280914307, "learning_rate": 1.4088792260940474e-05, "loss": 0.16888031959533692, "step": 77500 }, { "epoch": 0.3849974667951482, "grad_norm": 1.5988739728927612, "learning_rate": 1.4087324114400938e-05, "loss": 0.16799936294555665, "step": 77510 }, { "epoch": 0.38504713747851743, "grad_norm": 1.2025383710861206, "learning_rate": 1.4085855862083095e-05, "loss": 0.1673396944999695, "step": 77520 }, { "epoch": 0.3850968081618867, "grad_norm": 0.9728186726570129, "learning_rate": 1.4084387504024947e-05, "loss": 0.16854935884475708, "step": 77530 }, { "epoch": 0.38514647884525594, "grad_norm": 0.7156975865364075, "learning_rate": 1.4082919040264496e-05, "loss": 0.1499868631362915, "step": 77540 }, { "epoch": 0.38519614952862524, "grad_norm": 1.2396140098571777, "learning_rate": 1.4081450470839739e-05, "loss": 0.17197425365448, "step": 77550 }, { "epoch": 0.3852458202119945, "grad_norm": 0.5471721291542053, "learning_rate": 1.4079981795788684e-05, "loss": 0.16315736770629882, "step": 77560 }, { "epoch": 0.38529549089536375, "grad_norm": 0.5942683219909668, "learning_rate": 1.407851301514934e-05, "loss": 0.16194547414779664, "step": 77570 }, { "epoch": 0.385345161578733, "grad_norm": 1.7619929313659668, "learning_rate": 1.407704412895972e-05, "loss": 0.1672149419784546, "step": 77580 }, { "epoch": 0.38539483226210225, "grad_norm": 0.7633818984031677, "learning_rate": 1.4075575137257838e-05, "loss": 0.1693530797958374, "step": 77590 }, { "epoch": 0.3854445029454715, "grad_norm": 1.0937377214431763, "learning_rate": 1.407410604008171e-05, "loss": 0.15951080322265626, "step": 77600 }, { "epoch": 0.3854941736288408, "grad_norm": 0.648076057434082, "learning_rate": 1.4072636837469356e-05, "loss": 0.15322766304016114, "step": 77610 }, { "epoch": 0.38554384431221006, "grad_norm": 1.6289578676223755, "learning_rate": 1.4071167529458797e-05, "loss": 0.17262098789215088, "step": 77620 }, { "epoch": 0.3855935149955793, "grad_norm": 0.6105325222015381, "learning_rate": 1.4069698116088059e-05, "loss": 0.15631933212280275, "step": 77630 }, { "epoch": 0.38564318567894856, "grad_norm": 0.5804464221000671, "learning_rate": 1.4068228597395172e-05, "loss": 0.17039473056793214, "step": 77640 }, { "epoch": 0.3856928563623178, "grad_norm": 1.3155304193496704, "learning_rate": 1.4066758973418165e-05, "loss": 0.17204666137695312, "step": 77650 }, { "epoch": 0.38574252704568707, "grad_norm": 1.4502050876617432, "learning_rate": 1.4065289244195071e-05, "loss": 0.17278101444244384, "step": 77660 }, { "epoch": 0.3857921977290564, "grad_norm": 1.2585630416870117, "learning_rate": 1.4063819409763924e-05, "loss": 0.18262221813201904, "step": 77670 }, { "epoch": 0.3858418684124256, "grad_norm": 1.4659117460250854, "learning_rate": 1.4062349470162766e-05, "loss": 0.16751525402069092, "step": 77680 }, { "epoch": 0.3858915390957949, "grad_norm": 1.1682260036468506, "learning_rate": 1.4060879425429637e-05, "loss": 0.16842840909957885, "step": 77690 }, { "epoch": 0.38594120977916413, "grad_norm": 0.7649617791175842, "learning_rate": 1.405940927560258e-05, "loss": 0.16357256174087526, "step": 77700 }, { "epoch": 0.3859908804625334, "grad_norm": 0.697730302810669, "learning_rate": 1.4057939020719642e-05, "loss": 0.15245232582092286, "step": 77710 }, { "epoch": 0.3860405511459027, "grad_norm": 1.4294025897979736, "learning_rate": 1.4056468660818875e-05, "loss": 0.16420882940292358, "step": 77720 }, { "epoch": 0.38609022182927194, "grad_norm": 2.273298501968384, "learning_rate": 1.4054998195938333e-05, "loss": 0.1739555835723877, "step": 77730 }, { "epoch": 0.3861398925126412, "grad_norm": 1.7578095197677612, "learning_rate": 1.4053527626116066e-05, "loss": 0.184621262550354, "step": 77740 }, { "epoch": 0.38618956319601044, "grad_norm": 0.810145914554596, "learning_rate": 1.4052056951390132e-05, "loss": 0.17037309408187867, "step": 77750 }, { "epoch": 0.3862392338793797, "grad_norm": 1.4800559282302856, "learning_rate": 1.4050586171798598e-05, "loss": 0.16664761304855347, "step": 77760 }, { "epoch": 0.38628890456274895, "grad_norm": 2.041226625442505, "learning_rate": 1.4049115287379517e-05, "loss": 0.17825616598129274, "step": 77770 }, { "epoch": 0.38633857524611825, "grad_norm": 2.6880674362182617, "learning_rate": 1.4047644298170964e-05, "loss": 0.1548234701156616, "step": 77780 }, { "epoch": 0.3863882459294875, "grad_norm": 0.7374491095542908, "learning_rate": 1.4046173204211003e-05, "loss": 0.17641130685806275, "step": 77790 }, { "epoch": 0.38643791661285676, "grad_norm": 1.8893802165985107, "learning_rate": 1.4044702005537705e-05, "loss": 0.1711979866027832, "step": 77800 }, { "epoch": 0.386487587296226, "grad_norm": 0.9151430130004883, "learning_rate": 1.404323070218915e-05, "loss": 0.18421001434326173, "step": 77810 }, { "epoch": 0.38653725797959526, "grad_norm": 1.704331874847412, "learning_rate": 1.4041759294203407e-05, "loss": 0.16591553688049315, "step": 77820 }, { "epoch": 0.38658692866296457, "grad_norm": 0.5744929313659668, "learning_rate": 1.4040287781618557e-05, "loss": 0.16223409175872802, "step": 77830 }, { "epoch": 0.3866365993463338, "grad_norm": 1.8888953924179077, "learning_rate": 1.4038816164472686e-05, "loss": 0.15080174207687377, "step": 77840 }, { "epoch": 0.38668627002970307, "grad_norm": 1.9703843593597412, "learning_rate": 1.4037344442803877e-05, "loss": 0.17283488512039186, "step": 77850 }, { "epoch": 0.3867359407130723, "grad_norm": 0.9454822540283203, "learning_rate": 1.4035872616650217e-05, "loss": 0.16485321521759033, "step": 77860 }, { "epoch": 0.3867856113964416, "grad_norm": 1.1885713338851929, "learning_rate": 1.4034400686049795e-05, "loss": 0.16209492683410645, "step": 77870 }, { "epoch": 0.3868352820798108, "grad_norm": 2.0213236808776855, "learning_rate": 1.4032928651040706e-05, "loss": 0.159285306930542, "step": 77880 }, { "epoch": 0.38688495276318013, "grad_norm": 1.6016457080841064, "learning_rate": 1.4031456511661045e-05, "loss": 0.1571174144744873, "step": 77890 }, { "epoch": 0.3869346234465494, "grad_norm": 0.8020121455192566, "learning_rate": 1.402998426794891e-05, "loss": 0.16020698547363282, "step": 77900 }, { "epoch": 0.38698429412991864, "grad_norm": 1.4612915515899658, "learning_rate": 1.4028511919942402e-05, "loss": 0.17266340255737306, "step": 77910 }, { "epoch": 0.3870339648132879, "grad_norm": 1.6116182804107666, "learning_rate": 1.4027039467679626e-05, "loss": 0.14847004413604736, "step": 77920 }, { "epoch": 0.38708363549665714, "grad_norm": 0.6479580998420715, "learning_rate": 1.4025566911198688e-05, "loss": 0.15983210802078246, "step": 77930 }, { "epoch": 0.38713330618002645, "grad_norm": 2.549252986907959, "learning_rate": 1.40240942505377e-05, "loss": 0.16768181324005127, "step": 77940 }, { "epoch": 0.3871829768633957, "grad_norm": 1.5873697996139526, "learning_rate": 1.4022621485734766e-05, "loss": 0.15131778717041017, "step": 77950 }, { "epoch": 0.38723264754676495, "grad_norm": 1.2278608083724976, "learning_rate": 1.4021148616828008e-05, "loss": 0.1611499309539795, "step": 77960 }, { "epoch": 0.3872823182301342, "grad_norm": 1.2357676029205322, "learning_rate": 1.4019675643855541e-05, "loss": 0.17098543643951417, "step": 77970 }, { "epoch": 0.38733198891350346, "grad_norm": 1.387707233428955, "learning_rate": 1.4018202566855483e-05, "loss": 0.17054438591003418, "step": 77980 }, { "epoch": 0.3873816595968727, "grad_norm": 1.3805948495864868, "learning_rate": 1.4016729385865961e-05, "loss": 0.16850996017456055, "step": 77990 }, { "epoch": 0.387431330280242, "grad_norm": 1.383164882659912, "learning_rate": 1.4015256100925095e-05, "loss": 0.15624866485595704, "step": 78000 }, { "epoch": 0.38748100096361127, "grad_norm": 1.8913081884384155, "learning_rate": 1.4013782712071016e-05, "loss": 0.17652130126953125, "step": 78010 }, { "epoch": 0.3875306716469805, "grad_norm": 1.9204559326171875, "learning_rate": 1.4012309219341856e-05, "loss": 0.1683825135231018, "step": 78020 }, { "epoch": 0.38758034233034977, "grad_norm": 2.6707146167755127, "learning_rate": 1.4010835622775742e-05, "loss": 0.16968467235565185, "step": 78030 }, { "epoch": 0.387630013013719, "grad_norm": 1.8891061544418335, "learning_rate": 1.4009361922410818e-05, "loss": 0.16680707931518554, "step": 78040 }, { "epoch": 0.38767968369708833, "grad_norm": 0.7377228736877441, "learning_rate": 1.4007888118285218e-05, "loss": 0.16675320863723755, "step": 78050 }, { "epoch": 0.3877293543804576, "grad_norm": 1.4331194162368774, "learning_rate": 1.4006414210437088e-05, "loss": 0.1602133631706238, "step": 78060 }, { "epoch": 0.38777902506382683, "grad_norm": 2.009841203689575, "learning_rate": 1.4004940198904566e-05, "loss": 0.17026997804641725, "step": 78070 }, { "epoch": 0.3878286957471961, "grad_norm": 1.0304667949676514, "learning_rate": 1.4003466083725804e-05, "loss": 0.17074289321899414, "step": 78080 }, { "epoch": 0.38787836643056534, "grad_norm": 0.8387106657028198, "learning_rate": 1.4001991864938949e-05, "loss": 0.161776340007782, "step": 78090 }, { "epoch": 0.3879280371139346, "grad_norm": 0.7768232822418213, "learning_rate": 1.4000517542582154e-05, "loss": 0.18691425323486327, "step": 78100 }, { "epoch": 0.3879777077973039, "grad_norm": 1.0633680820465088, "learning_rate": 1.3999043116693573e-05, "loss": 0.17314579486846923, "step": 78110 }, { "epoch": 0.38802737848067315, "grad_norm": 1.1763463020324707, "learning_rate": 1.3997568587311363e-05, "loss": 0.1644505739212036, "step": 78120 }, { "epoch": 0.3880770491640424, "grad_norm": 0.9585666656494141, "learning_rate": 1.3996093954473687e-05, "loss": 0.17782151699066162, "step": 78130 }, { "epoch": 0.38812671984741165, "grad_norm": 0.6250803470611572, "learning_rate": 1.3994619218218706e-05, "loss": 0.1714242458343506, "step": 78140 }, { "epoch": 0.3881763905307809, "grad_norm": 1.20268976688385, "learning_rate": 1.399314437858459e-05, "loss": 0.17025312185287475, "step": 78150 }, { "epoch": 0.3882260612141502, "grad_norm": 1.189337968826294, "learning_rate": 1.3991669435609496e-05, "loss": 0.18248814344406128, "step": 78160 }, { "epoch": 0.38827573189751946, "grad_norm": 1.0741175413131714, "learning_rate": 1.3990194389331603e-05, "loss": 0.17237429618835448, "step": 78170 }, { "epoch": 0.3883254025808887, "grad_norm": 1.2820552587509155, "learning_rate": 1.3988719239789085e-05, "loss": 0.1761174201965332, "step": 78180 }, { "epoch": 0.38837507326425796, "grad_norm": 0.7205581068992615, "learning_rate": 1.3987243987020119e-05, "loss": 0.1606868863105774, "step": 78190 }, { "epoch": 0.3884247439476272, "grad_norm": 0.549599826335907, "learning_rate": 1.398576863106288e-05, "loss": 0.16072289943695067, "step": 78200 }, { "epoch": 0.38847441463099647, "grad_norm": 0.7814284563064575, "learning_rate": 1.3984293171955549e-05, "loss": 0.17927200794219972, "step": 78210 }, { "epoch": 0.3885240853143658, "grad_norm": 2.881622552871704, "learning_rate": 1.3982817609736317e-05, "loss": 0.17219102382659912, "step": 78220 }, { "epoch": 0.388573755997735, "grad_norm": 0.7955719828605652, "learning_rate": 1.3981341944443366e-05, "loss": 0.1785123586654663, "step": 78230 }, { "epoch": 0.3886234266811043, "grad_norm": 1.5204014778137207, "learning_rate": 1.3979866176114882e-05, "loss": 0.16860058307647705, "step": 78240 }, { "epoch": 0.38867309736447353, "grad_norm": 0.7841554284095764, "learning_rate": 1.3978390304789064e-05, "loss": 0.16774389743804932, "step": 78250 }, { "epoch": 0.3887227680478428, "grad_norm": 1.4114736318588257, "learning_rate": 1.3976914330504106e-05, "loss": 0.1633389949798584, "step": 78260 }, { "epoch": 0.3887724387312121, "grad_norm": 0.7998437881469727, "learning_rate": 1.3975438253298204e-05, "loss": 0.1525172233581543, "step": 78270 }, { "epoch": 0.38882210941458134, "grad_norm": 1.1188963651657104, "learning_rate": 1.3973962073209557e-05, "loss": 0.1690903902053833, "step": 78280 }, { "epoch": 0.3888717800979506, "grad_norm": 1.0201517343521118, "learning_rate": 1.397248579027637e-05, "loss": 0.17233266830444335, "step": 78290 }, { "epoch": 0.38892145078131984, "grad_norm": 1.733574390411377, "learning_rate": 1.3971009404536849e-05, "loss": 0.16057811975479125, "step": 78300 }, { "epoch": 0.3889711214646891, "grad_norm": 0.9688180685043335, "learning_rate": 1.39695329160292e-05, "loss": 0.16055086851119996, "step": 78310 }, { "epoch": 0.38902079214805835, "grad_norm": 1.268783450126648, "learning_rate": 1.3968056324791637e-05, "loss": 0.1758996605873108, "step": 78320 }, { "epoch": 0.38907046283142765, "grad_norm": 1.0950990915298462, "learning_rate": 1.3966579630862372e-05, "loss": 0.1701873779296875, "step": 78330 }, { "epoch": 0.3891201335147969, "grad_norm": 1.6764112710952759, "learning_rate": 1.3965102834279618e-05, "loss": 0.179001784324646, "step": 78340 }, { "epoch": 0.38916980419816616, "grad_norm": 0.8664872646331787, "learning_rate": 1.39636259350816e-05, "loss": 0.18116452693939208, "step": 78350 }, { "epoch": 0.3892194748815354, "grad_norm": 1.389587163925171, "learning_rate": 1.3962148933306534e-05, "loss": 0.16675171852111817, "step": 78360 }, { "epoch": 0.38926914556490466, "grad_norm": 1.2848385572433472, "learning_rate": 1.3960671828992649e-05, "loss": 0.17363567352294923, "step": 78370 }, { "epoch": 0.38931881624827397, "grad_norm": 1.8272377252578735, "learning_rate": 1.3959194622178167e-05, "loss": 0.16777369976043702, "step": 78380 }, { "epoch": 0.3893684869316432, "grad_norm": 0.9147531390190125, "learning_rate": 1.395771731290132e-05, "loss": 0.1626970052719116, "step": 78390 }, { "epoch": 0.3894181576150125, "grad_norm": 0.6893924474716187, "learning_rate": 1.3956239901200342e-05, "loss": 0.17819395065307617, "step": 78400 }, { "epoch": 0.3894678282983817, "grad_norm": 1.0772701501846313, "learning_rate": 1.3954762387113468e-05, "loss": 0.15886216163635253, "step": 78410 }, { "epoch": 0.389517498981751, "grad_norm": 0.6842103004455566, "learning_rate": 1.395328477067893e-05, "loss": 0.17748281955718995, "step": 78420 }, { "epoch": 0.3895671696651202, "grad_norm": 1.4507538080215454, "learning_rate": 1.3951807051934973e-05, "loss": 0.1750504970550537, "step": 78430 }, { "epoch": 0.38961684034848953, "grad_norm": 0.8983054161071777, "learning_rate": 1.3950329230919838e-05, "loss": 0.1681800365447998, "step": 78440 }, { "epoch": 0.3896665110318588, "grad_norm": 1.308739423751831, "learning_rate": 1.3948851307671771e-05, "loss": 0.17118968963623046, "step": 78450 }, { "epoch": 0.38971618171522804, "grad_norm": 1.2217353582382202, "learning_rate": 1.3947373282229019e-05, "loss": 0.17037608623504638, "step": 78460 }, { "epoch": 0.3897658523985973, "grad_norm": 1.3758738040924072, "learning_rate": 1.3945895154629837e-05, "loss": 0.16459438800811768, "step": 78470 }, { "epoch": 0.38981552308196654, "grad_norm": 0.9483239650726318, "learning_rate": 1.3944416924912472e-05, "loss": 0.15985454320907594, "step": 78480 }, { "epoch": 0.38986519376533585, "grad_norm": 0.8660454750061035, "learning_rate": 1.3942938593115188e-05, "loss": 0.16955019235610963, "step": 78490 }, { "epoch": 0.3899148644487051, "grad_norm": 0.9619429707527161, "learning_rate": 1.3941460159276234e-05, "loss": 0.16300718784332274, "step": 78500 }, { "epoch": 0.38996453513207435, "grad_norm": 0.8623200058937073, "learning_rate": 1.3939981623433874e-05, "loss": 0.1674729585647583, "step": 78510 }, { "epoch": 0.3900142058154436, "grad_norm": 1.7457129955291748, "learning_rate": 1.3938502985626376e-05, "loss": 0.1770164966583252, "step": 78520 }, { "epoch": 0.39006387649881286, "grad_norm": 0.8634381294250488, "learning_rate": 1.3937024245892006e-05, "loss": 0.16571953296661376, "step": 78530 }, { "epoch": 0.3901135471821821, "grad_norm": 2.1758766174316406, "learning_rate": 1.393554540426903e-05, "loss": 0.1726811408996582, "step": 78540 }, { "epoch": 0.3901632178655514, "grad_norm": 2.1657462120056152, "learning_rate": 1.3934066460795721e-05, "loss": 0.17147834300994874, "step": 78550 }, { "epoch": 0.39021288854892067, "grad_norm": 2.4853992462158203, "learning_rate": 1.3932587415510354e-05, "loss": 0.1717128038406372, "step": 78560 }, { "epoch": 0.3902625592322899, "grad_norm": 0.8508375287055969, "learning_rate": 1.3931108268451205e-05, "loss": 0.1668329954147339, "step": 78570 }, { "epoch": 0.39031222991565917, "grad_norm": 0.6162323355674744, "learning_rate": 1.3929629019656554e-05, "loss": 0.1518558979034424, "step": 78580 }, { "epoch": 0.3903619005990284, "grad_norm": 1.7127869129180908, "learning_rate": 1.3928149669164687e-05, "loss": 0.16789160966873168, "step": 78590 }, { "epoch": 0.39041157128239773, "grad_norm": 2.183490753173828, "learning_rate": 1.3926670217013886e-05, "loss": 0.17545731067657472, "step": 78600 }, { "epoch": 0.390461241965767, "grad_norm": 1.6667298078536987, "learning_rate": 1.3925190663242437e-05, "loss": 0.1677847385406494, "step": 78610 }, { "epoch": 0.39051091264913623, "grad_norm": 1.0741064548492432, "learning_rate": 1.3923711007888633e-05, "loss": 0.16567840576171874, "step": 78620 }, { "epoch": 0.3905605833325055, "grad_norm": 1.0487475395202637, "learning_rate": 1.3922231250990765e-05, "loss": 0.16925549507141113, "step": 78630 }, { "epoch": 0.39061025401587474, "grad_norm": 0.9520536065101624, "learning_rate": 1.3920751392587128e-05, "loss": 0.1510087490081787, "step": 78640 }, { "epoch": 0.390659924699244, "grad_norm": 0.7245051860809326, "learning_rate": 1.3919271432716024e-05, "loss": 0.17891411781311034, "step": 78650 }, { "epoch": 0.3907095953826133, "grad_norm": 0.9959081411361694, "learning_rate": 1.3917791371415753e-05, "loss": 0.1675774097442627, "step": 78660 }, { "epoch": 0.39075926606598255, "grad_norm": 0.9875747561454773, "learning_rate": 1.3916311208724611e-05, "loss": 0.16575207710266113, "step": 78670 }, { "epoch": 0.3908089367493518, "grad_norm": 1.5370274782180786, "learning_rate": 1.3914830944680913e-05, "loss": 0.17346768379211425, "step": 78680 }, { "epoch": 0.39085860743272105, "grad_norm": 1.6181896924972534, "learning_rate": 1.3913350579322964e-05, "loss": 0.16626060009002686, "step": 78690 }, { "epoch": 0.3909082781160903, "grad_norm": 1.3473834991455078, "learning_rate": 1.3911870112689075e-05, "loss": 0.1665278196334839, "step": 78700 }, { "epoch": 0.3909579487994596, "grad_norm": 1.6384425163269043, "learning_rate": 1.391038954481756e-05, "loss": 0.1623653292655945, "step": 78710 }, { "epoch": 0.39100761948282886, "grad_norm": 1.632016658782959, "learning_rate": 1.3908908875746734e-05, "loss": 0.15943797826766967, "step": 78720 }, { "epoch": 0.3910572901661981, "grad_norm": 0.9127589464187622, "learning_rate": 1.3907428105514921e-05, "loss": 0.16559511423110962, "step": 78730 }, { "epoch": 0.39110696084956736, "grad_norm": 0.7954705357551575, "learning_rate": 1.390594723416044e-05, "loss": 0.16134343147277833, "step": 78740 }, { "epoch": 0.3911566315329366, "grad_norm": 1.2824593782424927, "learning_rate": 1.3904466261721613e-05, "loss": 0.16705689430236817, "step": 78750 }, { "epoch": 0.39120630221630587, "grad_norm": 0.9989429116249084, "learning_rate": 1.390298518823677e-05, "loss": 0.16836168766021728, "step": 78760 }, { "epoch": 0.3912559728996752, "grad_norm": 0.8482696413993835, "learning_rate": 1.3901504013744241e-05, "loss": 0.1668321132659912, "step": 78770 }, { "epoch": 0.3913056435830444, "grad_norm": 1.6486536264419556, "learning_rate": 1.3900022738282354e-05, "loss": 0.17442243099212645, "step": 78780 }, { "epoch": 0.3913553142664137, "grad_norm": 1.2022074460983276, "learning_rate": 1.3898541361889448e-05, "loss": 0.16443458795547486, "step": 78790 }, { "epoch": 0.39140498494978293, "grad_norm": 1.2752207517623901, "learning_rate": 1.3897059884603858e-05, "loss": 0.16690741777420043, "step": 78800 }, { "epoch": 0.3914546556331522, "grad_norm": 1.5593379735946655, "learning_rate": 1.3895578306463928e-05, "loss": 0.17151731252670288, "step": 78810 }, { "epoch": 0.3915043263165215, "grad_norm": 1.5365209579467773, "learning_rate": 1.3894096627507996e-05, "loss": 0.16797059774398804, "step": 78820 }, { "epoch": 0.39155399699989074, "grad_norm": 1.2469793558120728, "learning_rate": 1.3892614847774413e-05, "loss": 0.16260242462158203, "step": 78830 }, { "epoch": 0.39160366768326, "grad_norm": 1.6363357305526733, "learning_rate": 1.3891132967301516e-05, "loss": 0.16656250953674318, "step": 78840 }, { "epoch": 0.39165333836662924, "grad_norm": 1.6442278623580933, "learning_rate": 1.3889650986127665e-05, "loss": 0.17914271354675293, "step": 78850 }, { "epoch": 0.3917030090499985, "grad_norm": 1.064529538154602, "learning_rate": 1.3888168904291213e-05, "loss": 0.1783968210220337, "step": 78860 }, { "epoch": 0.39175267973336775, "grad_norm": 0.5806207656860352, "learning_rate": 1.388668672183051e-05, "loss": 0.16850342750549316, "step": 78870 }, { "epoch": 0.39180235041673706, "grad_norm": 1.3108227252960205, "learning_rate": 1.3885204438783919e-05, "loss": 0.17543264627456664, "step": 78880 }, { "epoch": 0.3918520211001063, "grad_norm": 0.7518987655639648, "learning_rate": 1.38837220551898e-05, "loss": 0.1662961483001709, "step": 78890 }, { "epoch": 0.39190169178347556, "grad_norm": 0.8375481367111206, "learning_rate": 1.3882239571086514e-05, "loss": 0.1739502191543579, "step": 78900 }, { "epoch": 0.3919513624668448, "grad_norm": 0.9526516795158386, "learning_rate": 1.3880756986512428e-05, "loss": 0.16939003467559816, "step": 78910 }, { "epoch": 0.39200103315021406, "grad_norm": 1.4680399894714355, "learning_rate": 1.3879274301505913e-05, "loss": 0.1676870346069336, "step": 78920 }, { "epoch": 0.39205070383358337, "grad_norm": 0.7454482316970825, "learning_rate": 1.3877791516105338e-05, "loss": 0.17511227130889892, "step": 78930 }, { "epoch": 0.3921003745169526, "grad_norm": 1.3037474155426025, "learning_rate": 1.387630863034908e-05, "loss": 0.15265016555786132, "step": 78940 }, { "epoch": 0.3921500452003219, "grad_norm": 0.86154705286026, "learning_rate": 1.3874825644275511e-05, "loss": 0.16367827653884887, "step": 78950 }, { "epoch": 0.3921997158836911, "grad_norm": 1.947461485862732, "learning_rate": 1.3873342557923016e-05, "loss": 0.16513402462005616, "step": 78960 }, { "epoch": 0.3922493865670604, "grad_norm": 1.8349764347076416, "learning_rate": 1.387185937132997e-05, "loss": 0.1633285403251648, "step": 78970 }, { "epoch": 0.39229905725042963, "grad_norm": 1.9091131687164307, "learning_rate": 1.387037608453476e-05, "loss": 0.1648266077041626, "step": 78980 }, { "epoch": 0.39234872793379894, "grad_norm": 1.3440299034118652, "learning_rate": 1.3868892697575771e-05, "loss": 0.17825491428375245, "step": 78990 }, { "epoch": 0.3923983986171682, "grad_norm": 1.0797619819641113, "learning_rate": 1.38674092104914e-05, "loss": 0.17169723510742188, "step": 79000 }, { "epoch": 0.39244806930053744, "grad_norm": 1.024506688117981, "learning_rate": 1.3865925623320032e-05, "loss": 0.16619551181793213, "step": 79010 }, { "epoch": 0.3924977399839067, "grad_norm": 0.6555032134056091, "learning_rate": 1.3864441936100062e-05, "loss": 0.1594007968902588, "step": 79020 }, { "epoch": 0.39254741066727594, "grad_norm": 1.695942997932434, "learning_rate": 1.3862958148869889e-05, "loss": 0.1807084321975708, "step": 79030 }, { "epoch": 0.39259708135064525, "grad_norm": 0.8371043801307678, "learning_rate": 1.3861474261667912e-05, "loss": 0.16926493644714355, "step": 79040 }, { "epoch": 0.3926467520340145, "grad_norm": 0.8108519911766052, "learning_rate": 1.3859990274532532e-05, "loss": 0.15671682357788086, "step": 79050 }, { "epoch": 0.39269642271738375, "grad_norm": 0.7889569401741028, "learning_rate": 1.3858506187502158e-05, "loss": 0.1629460096359253, "step": 79060 }, { "epoch": 0.392746093400753, "grad_norm": 1.6454240083694458, "learning_rate": 1.3857022000615195e-05, "loss": 0.16310153007507325, "step": 79070 }, { "epoch": 0.39279576408412226, "grad_norm": 1.2579152584075928, "learning_rate": 1.3855537713910055e-05, "loss": 0.1868818759918213, "step": 79080 }, { "epoch": 0.3928454347674915, "grad_norm": 0.89825040102005, "learning_rate": 1.3854053327425146e-05, "loss": 0.1567061185836792, "step": 79090 }, { "epoch": 0.3928951054508608, "grad_norm": 0.619158923625946, "learning_rate": 1.385256884119889e-05, "loss": 0.1503405213356018, "step": 79100 }, { "epoch": 0.39294477613423007, "grad_norm": 1.703798532485962, "learning_rate": 1.38510842552697e-05, "loss": 0.1832939863204956, "step": 79110 }, { "epoch": 0.3929944468175993, "grad_norm": 0.9905452132225037, "learning_rate": 1.3849599569675994e-05, "loss": 0.1734554409980774, "step": 79120 }, { "epoch": 0.39304411750096857, "grad_norm": 1.3984979391098022, "learning_rate": 1.3848114784456203e-05, "loss": 0.163105309009552, "step": 79130 }, { "epoch": 0.3930937881843378, "grad_norm": 1.4995065927505493, "learning_rate": 1.3846629899648748e-05, "loss": 0.16679818630218507, "step": 79140 }, { "epoch": 0.39314345886770713, "grad_norm": 2.178457736968994, "learning_rate": 1.3845144915292059e-05, "loss": 0.16834657192230223, "step": 79150 }, { "epoch": 0.3931931295510764, "grad_norm": 1.1587039232254028, "learning_rate": 1.3843659831424567e-05, "loss": 0.17163989543914795, "step": 79160 }, { "epoch": 0.39324280023444563, "grad_norm": 0.952453076839447, "learning_rate": 1.3842174648084701e-05, "loss": 0.1676340103149414, "step": 79170 }, { "epoch": 0.3932924709178149, "grad_norm": 0.8179432153701782, "learning_rate": 1.38406893653109e-05, "loss": 0.15538549423217773, "step": 79180 }, { "epoch": 0.39334214160118414, "grad_norm": 0.5595888495445251, "learning_rate": 1.3839203983141602e-05, "loss": 0.18054678440093994, "step": 79190 }, { "epoch": 0.3933918122845534, "grad_norm": 0.9635975956916809, "learning_rate": 1.3837718501615251e-05, "loss": 0.16505155563354493, "step": 79200 }, { "epoch": 0.3934414829679227, "grad_norm": 1.3167805671691895, "learning_rate": 1.3836232920770286e-05, "loss": 0.16268784999847413, "step": 79210 }, { "epoch": 0.39349115365129195, "grad_norm": 1.7034053802490234, "learning_rate": 1.3834747240645157e-05, "loss": 0.175360369682312, "step": 79220 }, { "epoch": 0.3935408243346612, "grad_norm": 1.4925733804702759, "learning_rate": 1.383326146127831e-05, "loss": 0.16060346364974976, "step": 79230 }, { "epoch": 0.39359049501803045, "grad_norm": 1.3993102312088013, "learning_rate": 1.38317755827082e-05, "loss": 0.1571551203727722, "step": 79240 }, { "epoch": 0.3936401657013997, "grad_norm": 0.9642373323440552, "learning_rate": 1.3830289604973278e-05, "loss": 0.16709233522415162, "step": 79250 }, { "epoch": 0.39368983638476895, "grad_norm": 0.8005136847496033, "learning_rate": 1.3828803528111999e-05, "loss": 0.15757585763931276, "step": 79260 }, { "epoch": 0.39373950706813826, "grad_norm": 2.3219375610351562, "learning_rate": 1.3827317352162823e-05, "loss": 0.1681918501853943, "step": 79270 }, { "epoch": 0.3937891777515075, "grad_norm": 1.7303391695022583, "learning_rate": 1.3825831077164216e-05, "loss": 0.17072362899780275, "step": 79280 }, { "epoch": 0.39383884843487676, "grad_norm": 1.3194079399108887, "learning_rate": 1.382434470315464e-05, "loss": 0.17414058446884156, "step": 79290 }, { "epoch": 0.393888519118246, "grad_norm": 1.153359293937683, "learning_rate": 1.3822858230172558e-05, "loss": 0.1618185043334961, "step": 79300 }, { "epoch": 0.39393818980161527, "grad_norm": 2.7473323345184326, "learning_rate": 1.3821371658256444e-05, "loss": 0.16482995748519896, "step": 79310 }, { "epoch": 0.3939878604849846, "grad_norm": 0.8358714580535889, "learning_rate": 1.3819884987444768e-05, "loss": 0.16976089477539064, "step": 79320 }, { "epoch": 0.3940375311683538, "grad_norm": 0.8732185959815979, "learning_rate": 1.3818398217776004e-05, "loss": 0.15401067733764648, "step": 79330 }, { "epoch": 0.3940872018517231, "grad_norm": 1.8315199613571167, "learning_rate": 1.3816911349288628e-05, "loss": 0.16071720123291017, "step": 79340 }, { "epoch": 0.39413687253509233, "grad_norm": 1.1050869226455688, "learning_rate": 1.3815424382021123e-05, "loss": 0.16199026107788086, "step": 79350 }, { "epoch": 0.3941865432184616, "grad_norm": 1.5132306814193726, "learning_rate": 1.3813937316011968e-05, "loss": 0.17559733390808105, "step": 79360 }, { "epoch": 0.39423621390183083, "grad_norm": 2.07878041267395, "learning_rate": 1.381245015129965e-05, "loss": 0.1586705446243286, "step": 79370 }, { "epoch": 0.39428588458520014, "grad_norm": 0.9220645427703857, "learning_rate": 1.3810962887922655e-05, "loss": 0.1669074058532715, "step": 79380 }, { "epoch": 0.3943355552685694, "grad_norm": 1.3648964166641235, "learning_rate": 1.380947552591947e-05, "loss": 0.16697518825531005, "step": 79390 }, { "epoch": 0.39438522595193864, "grad_norm": 1.1464557647705078, "learning_rate": 1.380798806532859e-05, "loss": 0.1598736047744751, "step": 79400 }, { "epoch": 0.3944348966353079, "grad_norm": 0.9142909646034241, "learning_rate": 1.3806500506188511e-05, "loss": 0.1764460325241089, "step": 79410 }, { "epoch": 0.39448456731867715, "grad_norm": 0.8547201156616211, "learning_rate": 1.3805012848537734e-05, "loss": 0.16938146352767944, "step": 79420 }, { "epoch": 0.39453423800204646, "grad_norm": 1.12579345703125, "learning_rate": 1.380352509241475e-05, "loss": 0.18405365943908691, "step": 79430 }, { "epoch": 0.3945839086854157, "grad_norm": 1.4528999328613281, "learning_rate": 1.3802037237858065e-05, "loss": 0.1673956036567688, "step": 79440 }, { "epoch": 0.39463357936878496, "grad_norm": 0.7250106930732727, "learning_rate": 1.3800549284906188e-05, "loss": 0.16408157348632812, "step": 79450 }, { "epoch": 0.3946832500521542, "grad_norm": 1.7023043632507324, "learning_rate": 1.3799061233597619e-05, "loss": 0.17782127857208252, "step": 79460 }, { "epoch": 0.39473292073552346, "grad_norm": 1.0399478673934937, "learning_rate": 1.3797573083970876e-05, "loss": 0.15301724672317504, "step": 79470 }, { "epoch": 0.3947825914188927, "grad_norm": 1.0034962892532349, "learning_rate": 1.379608483606447e-05, "loss": 0.17305420637130736, "step": 79480 }, { "epoch": 0.394832262102262, "grad_norm": 1.8832887411117554, "learning_rate": 1.3794596489916913e-05, "loss": 0.158243989944458, "step": 79490 }, { "epoch": 0.3948819327856313, "grad_norm": 1.8102954626083374, "learning_rate": 1.3793108045566727e-05, "loss": 0.1546283483505249, "step": 79500 }, { "epoch": 0.3949316034690005, "grad_norm": 2.3148751258850098, "learning_rate": 1.3791619503052429e-05, "loss": 0.16759226322174073, "step": 79510 }, { "epoch": 0.3949812741523698, "grad_norm": 0.7342858910560608, "learning_rate": 1.379013086241254e-05, "loss": 0.1691190004348755, "step": 79520 }, { "epoch": 0.39503094483573903, "grad_norm": 0.808967113494873, "learning_rate": 1.378864212368559e-05, "loss": 0.17764766216278077, "step": 79530 }, { "epoch": 0.39508061551910834, "grad_norm": 1.4803824424743652, "learning_rate": 1.3787153286910107e-05, "loss": 0.1734799861907959, "step": 79540 }, { "epoch": 0.3951302862024776, "grad_norm": 0.7935383319854736, "learning_rate": 1.3785664352124617e-05, "loss": 0.17369158267974855, "step": 79550 }, { "epoch": 0.39517995688584684, "grad_norm": 1.0430303812026978, "learning_rate": 1.3784175319367656e-05, "loss": 0.16421716213226317, "step": 79560 }, { "epoch": 0.3952296275692161, "grad_norm": 0.6048547625541687, "learning_rate": 1.378268618867776e-05, "loss": 0.16619348526000977, "step": 79570 }, { "epoch": 0.39527929825258534, "grad_norm": 0.9314641952514648, "learning_rate": 1.3781196960093465e-05, "loss": 0.18610421419143677, "step": 79580 }, { "epoch": 0.3953289689359546, "grad_norm": 1.4852763414382935, "learning_rate": 1.3779707633653314e-05, "loss": 0.17337610721588134, "step": 79590 }, { "epoch": 0.3953786396193239, "grad_norm": 1.0999420881271362, "learning_rate": 1.377821820939585e-05, "loss": 0.15592833757400512, "step": 79600 }, { "epoch": 0.39542831030269315, "grad_norm": 1.07339608669281, "learning_rate": 1.3776728687359618e-05, "loss": 0.16348624229431152, "step": 79610 }, { "epoch": 0.3954779809860624, "grad_norm": 0.8159897923469543, "learning_rate": 1.3775239067583166e-05, "loss": 0.16579442024230956, "step": 79620 }, { "epoch": 0.39552765166943166, "grad_norm": 1.0563533306121826, "learning_rate": 1.3773749350105044e-05, "loss": 0.1654505252838135, "step": 79630 }, { "epoch": 0.3955773223528009, "grad_norm": 0.8217092156410217, "learning_rate": 1.3772259534963809e-05, "loss": 0.16621960401535035, "step": 79640 }, { "epoch": 0.3956269930361702, "grad_norm": 0.9198184013366699, "learning_rate": 1.377076962219801e-05, "loss": 0.16105626821517943, "step": 79650 }, { "epoch": 0.39567666371953947, "grad_norm": 1.0584992170333862, "learning_rate": 1.3769279611846209e-05, "loss": 0.1817626714706421, "step": 79660 }, { "epoch": 0.3957263344029087, "grad_norm": 1.4899020195007324, "learning_rate": 1.376778950394697e-05, "loss": 0.1631646752357483, "step": 79670 }, { "epoch": 0.39577600508627797, "grad_norm": 1.3819645643234253, "learning_rate": 1.3766299298538854e-05, "loss": 0.16645171642303466, "step": 79680 }, { "epoch": 0.3958256757696472, "grad_norm": 1.6279603242874146, "learning_rate": 1.3764808995660425e-05, "loss": 0.17133194208145142, "step": 79690 }, { "epoch": 0.3958753464530165, "grad_norm": 1.4175177812576294, "learning_rate": 1.3763318595350255e-05, "loss": 0.17162295579910278, "step": 79700 }, { "epoch": 0.3959250171363858, "grad_norm": 1.697371482849121, "learning_rate": 1.376182809764691e-05, "loss": 0.16097896099090575, "step": 79710 }, { "epoch": 0.39597468781975503, "grad_norm": 1.1228688955307007, "learning_rate": 1.3760337502588965e-05, "loss": 0.16137363910675048, "step": 79720 }, { "epoch": 0.3960243585031243, "grad_norm": 1.524415135383606, "learning_rate": 1.3758846810215e-05, "loss": 0.15351725816726686, "step": 79730 }, { "epoch": 0.39607402918649354, "grad_norm": 2.240551471710205, "learning_rate": 1.3757356020563593e-05, "loss": 0.17147451639175415, "step": 79740 }, { "epoch": 0.3961236998698628, "grad_norm": 1.4635064601898193, "learning_rate": 1.3755865133673321e-05, "loss": 0.16902120113372804, "step": 79750 }, { "epoch": 0.3961733705532321, "grad_norm": 2.201223850250244, "learning_rate": 1.3754374149582768e-05, "loss": 0.17853796482086182, "step": 79760 }, { "epoch": 0.39622304123660135, "grad_norm": 1.5959362983703613, "learning_rate": 1.3752883068330522e-05, "loss": 0.16962268352508544, "step": 79770 }, { "epoch": 0.3962727119199706, "grad_norm": 0.7665773034095764, "learning_rate": 1.3751391889955175e-05, "loss": 0.1776970148086548, "step": 79780 }, { "epoch": 0.39632238260333985, "grad_norm": 1.7211490869522095, "learning_rate": 1.374990061449531e-05, "loss": 0.16011143922805787, "step": 79790 }, { "epoch": 0.3963720532867091, "grad_norm": 0.8783499002456665, "learning_rate": 1.3748409241989524e-05, "loss": 0.17105481624603272, "step": 79800 }, { "epoch": 0.39642172397007835, "grad_norm": 1.0626120567321777, "learning_rate": 1.3746917772476418e-05, "loss": 0.16338748931884767, "step": 79810 }, { "epoch": 0.39647139465344766, "grad_norm": 1.6344083547592163, "learning_rate": 1.3745426205994586e-05, "loss": 0.1653111457824707, "step": 79820 }, { "epoch": 0.3965210653368169, "grad_norm": 1.1058353185653687, "learning_rate": 1.3743934542582627e-05, "loss": 0.20603566169738768, "step": 79830 }, { "epoch": 0.39657073602018617, "grad_norm": 0.9170107841491699, "learning_rate": 1.3742442782279151e-05, "loss": 0.15874955654144288, "step": 79840 }, { "epoch": 0.3966204067035554, "grad_norm": 1.5151606798171997, "learning_rate": 1.3740950925122759e-05, "loss": 0.1649917483329773, "step": 79850 }, { "epoch": 0.39667007738692467, "grad_norm": 1.670624017715454, "learning_rate": 1.3739458971152058e-05, "loss": 0.18580167293548583, "step": 79860 }, { "epoch": 0.396719748070294, "grad_norm": 1.3265029191970825, "learning_rate": 1.3737966920405665e-05, "loss": 0.16415555477142335, "step": 79870 }, { "epoch": 0.39676941875366323, "grad_norm": 2.715240478515625, "learning_rate": 1.3736474772922193e-05, "loss": 0.16168928146362305, "step": 79880 }, { "epoch": 0.3968190894370325, "grad_norm": 1.6560903787612915, "learning_rate": 1.3734982528740255e-05, "loss": 0.16505672931671142, "step": 79890 }, { "epoch": 0.39686876012040173, "grad_norm": 1.4155604839324951, "learning_rate": 1.373349018789847e-05, "loss": 0.18613076210021973, "step": 79900 }, { "epoch": 0.396918430803771, "grad_norm": 1.0992614030838013, "learning_rate": 1.373199775043546e-05, "loss": 0.172646164894104, "step": 79910 }, { "epoch": 0.39696810148714023, "grad_norm": 1.158167839050293, "learning_rate": 1.373050521638985e-05, "loss": 0.1648378849029541, "step": 79920 }, { "epoch": 0.39701777217050954, "grad_norm": 0.8652383089065552, "learning_rate": 1.3729012585800263e-05, "loss": 0.1761999249458313, "step": 79930 }, { "epoch": 0.3970674428538788, "grad_norm": 0.8183004260063171, "learning_rate": 1.3727519858705331e-05, "loss": 0.16420162916183473, "step": 79940 }, { "epoch": 0.39711711353724805, "grad_norm": 1.3150110244750977, "learning_rate": 1.3726027035143682e-05, "loss": 0.15979342460632323, "step": 79950 }, { "epoch": 0.3971667842206173, "grad_norm": 1.5806138515472412, "learning_rate": 1.3724534115153957e-05, "loss": 0.1600364089012146, "step": 79960 }, { "epoch": 0.39721645490398655, "grad_norm": 0.9577076435089111, "learning_rate": 1.3723041098774781e-05, "loss": 0.16764278411865235, "step": 79970 }, { "epoch": 0.39726612558735586, "grad_norm": 0.6468181610107422, "learning_rate": 1.37215479860448e-05, "loss": 0.1728533387184143, "step": 79980 }, { "epoch": 0.3973157962707251, "grad_norm": 1.2934318780899048, "learning_rate": 1.3720054777002654e-05, "loss": 0.16695685386657716, "step": 79990 }, { "epoch": 0.39736546695409436, "grad_norm": 1.1894193887710571, "learning_rate": 1.3718561471686984e-05, "loss": 0.16301019191741944, "step": 80000 }, { "epoch": 0.3974151376374636, "grad_norm": 1.5388213396072388, "learning_rate": 1.3717068070136443e-05, "loss": 0.157186758518219, "step": 80010 }, { "epoch": 0.39746480832083286, "grad_norm": 0.8101372718811035, "learning_rate": 1.3715574572389673e-05, "loss": 0.1601102828979492, "step": 80020 }, { "epoch": 0.3975144790042021, "grad_norm": 1.179990291595459, "learning_rate": 1.3714080978485327e-05, "loss": 0.1555282950401306, "step": 80030 }, { "epoch": 0.3975641496875714, "grad_norm": 1.5131200551986694, "learning_rate": 1.3712587288462058e-05, "loss": 0.17129597663879395, "step": 80040 }, { "epoch": 0.3976138203709407, "grad_norm": 2.060227155685425, "learning_rate": 1.3711093502358521e-05, "loss": 0.16331286430358888, "step": 80050 }, { "epoch": 0.3976634910543099, "grad_norm": 0.7497544884681702, "learning_rate": 1.370959962021338e-05, "loss": 0.17142544984817504, "step": 80060 }, { "epoch": 0.3977131617376792, "grad_norm": 1.7780156135559082, "learning_rate": 1.3708105642065288e-05, "loss": 0.16660693883895875, "step": 80070 }, { "epoch": 0.39776283242104843, "grad_norm": 0.9837016463279724, "learning_rate": 1.3706611567952915e-05, "loss": 0.15706244707107545, "step": 80080 }, { "epoch": 0.39781250310441774, "grad_norm": 0.8005642294883728, "learning_rate": 1.3705117397914927e-05, "loss": 0.15385934114456176, "step": 80090 }, { "epoch": 0.397862173787787, "grad_norm": 1.018249273300171, "learning_rate": 1.3703623131989986e-05, "loss": 0.16834895610809325, "step": 80100 }, { "epoch": 0.39791184447115624, "grad_norm": 1.3173871040344238, "learning_rate": 1.3702128770216771e-05, "loss": 0.1689502239227295, "step": 80110 }, { "epoch": 0.3979615151545255, "grad_norm": 1.4809764623641968, "learning_rate": 1.3700634312633954e-05, "loss": 0.1612562656402588, "step": 80120 }, { "epoch": 0.39801118583789474, "grad_norm": 0.8985336422920227, "learning_rate": 1.3699139759280202e-05, "loss": 0.1659323453903198, "step": 80130 }, { "epoch": 0.398060856521264, "grad_norm": 2.3818039894104004, "learning_rate": 1.3697645110194206e-05, "loss": 0.16839954853057862, "step": 80140 }, { "epoch": 0.3981105272046333, "grad_norm": 0.9537913799285889, "learning_rate": 1.369615036541464e-05, "loss": 0.18088775873184204, "step": 80150 }, { "epoch": 0.39816019788800255, "grad_norm": 1.0037158727645874, "learning_rate": 1.3694655524980189e-05, "loss": 0.15626084804534912, "step": 80160 }, { "epoch": 0.3982098685713718, "grad_norm": 1.3448525667190552, "learning_rate": 1.3693160588929537e-05, "loss": 0.1659692645072937, "step": 80170 }, { "epoch": 0.39825953925474106, "grad_norm": 1.8436352014541626, "learning_rate": 1.369166555730138e-05, "loss": 0.15378544330596924, "step": 80180 }, { "epoch": 0.3983092099381103, "grad_norm": 1.4881103038787842, "learning_rate": 1.3690170430134396e-05, "loss": 0.18088974952697753, "step": 80190 }, { "epoch": 0.3983588806214796, "grad_norm": 0.6646865010261536, "learning_rate": 1.3688675207467286e-05, "loss": 0.1653029203414917, "step": 80200 }, { "epoch": 0.39840855130484887, "grad_norm": 1.107025384902954, "learning_rate": 1.3687179889338746e-05, "loss": 0.17411117553710936, "step": 80210 }, { "epoch": 0.3984582219882181, "grad_norm": 1.089293360710144, "learning_rate": 1.3685684475787473e-05, "loss": 0.1633596420288086, "step": 80220 }, { "epoch": 0.39850789267158737, "grad_norm": 1.823307991027832, "learning_rate": 1.3684188966852165e-05, "loss": 0.17435718774795533, "step": 80230 }, { "epoch": 0.3985575633549566, "grad_norm": 0.48894640803337097, "learning_rate": 1.3682693362571531e-05, "loss": 0.1565699338912964, "step": 80240 }, { "epoch": 0.3986072340383259, "grad_norm": 0.7984921932220459, "learning_rate": 1.3681197662984274e-05, "loss": 0.16709043979644775, "step": 80250 }, { "epoch": 0.3986569047216952, "grad_norm": 0.8436833024024963, "learning_rate": 1.3679701868129099e-05, "loss": 0.15855784416198732, "step": 80260 }, { "epoch": 0.39870657540506443, "grad_norm": 2.0735814571380615, "learning_rate": 1.367820597804472e-05, "loss": 0.15039960145950318, "step": 80270 }, { "epoch": 0.3987562460884337, "grad_norm": 1.4122554063796997, "learning_rate": 1.367670999276985e-05, "loss": 0.17412782907485963, "step": 80280 }, { "epoch": 0.39880591677180294, "grad_norm": 1.0978811979293823, "learning_rate": 1.3675213912343204e-05, "loss": 0.16088565587997436, "step": 80290 }, { "epoch": 0.3988555874551722, "grad_norm": 2.7235264778137207, "learning_rate": 1.36737177368035e-05, "loss": 0.16906721591949464, "step": 80300 }, { "epoch": 0.3989052581385415, "grad_norm": 1.248876929283142, "learning_rate": 1.3672221466189457e-05, "loss": 0.1673839807510376, "step": 80310 }, { "epoch": 0.39895492882191075, "grad_norm": 1.1994730234146118, "learning_rate": 1.3670725100539798e-05, "loss": 0.16247057914733887, "step": 80320 }, { "epoch": 0.39900459950528, "grad_norm": 1.57273268699646, "learning_rate": 1.3669228639893254e-05, "loss": 0.16737390756607057, "step": 80330 }, { "epoch": 0.39905427018864925, "grad_norm": 0.8553153276443481, "learning_rate": 1.366773208428854e-05, "loss": 0.16016591787338258, "step": 80340 }, { "epoch": 0.3991039408720185, "grad_norm": 1.2395819425582886, "learning_rate": 1.3666235433764404e-05, "loss": 0.17123763561248778, "step": 80350 }, { "epoch": 0.39915361155538776, "grad_norm": 1.6585899591445923, "learning_rate": 1.3664738688359565e-05, "loss": 0.17072761058807373, "step": 80360 }, { "epoch": 0.39920328223875706, "grad_norm": 1.4271292686462402, "learning_rate": 1.366324184811276e-05, "loss": 0.17075235843658448, "step": 80370 }, { "epoch": 0.3992529529221263, "grad_norm": 1.225533366203308, "learning_rate": 1.3661744913062732e-05, "loss": 0.17100398540496825, "step": 80380 }, { "epoch": 0.39930262360549557, "grad_norm": 0.6012064814567566, "learning_rate": 1.3660247883248214e-05, "loss": 0.1673094153404236, "step": 80390 }, { "epoch": 0.3993522942888648, "grad_norm": 1.7258082628250122, "learning_rate": 1.3658750758707955e-05, "loss": 0.1661123275756836, "step": 80400 }, { "epoch": 0.39940196497223407, "grad_norm": 0.7536632418632507, "learning_rate": 1.3657253539480699e-05, "loss": 0.1683700919151306, "step": 80410 }, { "epoch": 0.3994516356556034, "grad_norm": 1.7835969924926758, "learning_rate": 1.365575622560519e-05, "loss": 0.16695613861083985, "step": 80420 }, { "epoch": 0.39950130633897263, "grad_norm": 0.7538939118385315, "learning_rate": 1.3654258817120181e-05, "loss": 0.17342408895492553, "step": 80430 }, { "epoch": 0.3995509770223419, "grad_norm": 0.9662572145462036, "learning_rate": 1.3652761314064422e-05, "loss": 0.1650362491607666, "step": 80440 }, { "epoch": 0.39960064770571113, "grad_norm": 1.205307126045227, "learning_rate": 1.365126371647667e-05, "loss": 0.17377936840057373, "step": 80450 }, { "epoch": 0.3996503183890804, "grad_norm": 1.54489004611969, "learning_rate": 1.364976602439568e-05, "loss": 0.1705469608306885, "step": 80460 }, { "epoch": 0.39969998907244964, "grad_norm": 1.9691449403762817, "learning_rate": 1.3648268237860211e-05, "loss": 0.1677065372467041, "step": 80470 }, { "epoch": 0.39974965975581894, "grad_norm": 1.6179475784301758, "learning_rate": 1.3646770356909029e-05, "loss": 0.16005151271820067, "step": 80480 }, { "epoch": 0.3997993304391882, "grad_norm": 0.797478437423706, "learning_rate": 1.3645272381580897e-05, "loss": 0.18043155670166017, "step": 80490 }, { "epoch": 0.39984900112255745, "grad_norm": 0.806597888469696, "learning_rate": 1.3643774311914582e-05, "loss": 0.16320374011993408, "step": 80500 }, { "epoch": 0.3998986718059267, "grad_norm": 0.680730938911438, "learning_rate": 1.3642276147948852e-05, "loss": 0.16176886558532716, "step": 80510 }, { "epoch": 0.39994834248929595, "grad_norm": 0.7113097310066223, "learning_rate": 1.3640777889722481e-05, "loss": 0.174938428401947, "step": 80520 }, { "epoch": 0.39999801317266526, "grad_norm": 0.9861984848976135, "learning_rate": 1.363927953727424e-05, "loss": 0.1549067497253418, "step": 80530 }, { "epoch": 0.4000476838560345, "grad_norm": 0.7250463366508484, "learning_rate": 1.3637781090642908e-05, "loss": 0.16981136798858643, "step": 80540 }, { "epoch": 0.40009735453940376, "grad_norm": 1.6810457706451416, "learning_rate": 1.3636282549867266e-05, "loss": 0.17026679515838622, "step": 80550 }, { "epoch": 0.400147025222773, "grad_norm": 0.8119445443153381, "learning_rate": 1.3634783914986091e-05, "loss": 0.16170122623443603, "step": 80560 }, { "epoch": 0.40019669590614226, "grad_norm": 2.073044538497925, "learning_rate": 1.3633285186038172e-05, "loss": 0.1731007933616638, "step": 80570 }, { "epoch": 0.4002463665895115, "grad_norm": 0.7002148628234863, "learning_rate": 1.3631786363062294e-05, "loss": 0.16914042234420776, "step": 80580 }, { "epoch": 0.4002960372728808, "grad_norm": 1.310775637626648, "learning_rate": 1.3630287446097244e-05, "loss": 0.15775460004806519, "step": 80590 }, { "epoch": 0.4003457079562501, "grad_norm": 0.9913158416748047, "learning_rate": 1.3628788435181813e-05, "loss": 0.1563870906829834, "step": 80600 }, { "epoch": 0.4003953786396193, "grad_norm": 0.8652465343475342, "learning_rate": 1.3627289330354797e-05, "loss": 0.1848980188369751, "step": 80610 }, { "epoch": 0.4004450493229886, "grad_norm": 1.7368420362472534, "learning_rate": 1.3625790131654992e-05, "loss": 0.17506704330444336, "step": 80620 }, { "epoch": 0.40049472000635783, "grad_norm": 1.650375485420227, "learning_rate": 1.3624290839121197e-05, "loss": 0.16990624666213988, "step": 80630 }, { "epoch": 0.40054439068972714, "grad_norm": 0.9771175384521484, "learning_rate": 1.3622791452792212e-05, "loss": 0.1761780261993408, "step": 80640 }, { "epoch": 0.4005940613730964, "grad_norm": 1.0428266525268555, "learning_rate": 1.3621291972706841e-05, "loss": 0.165861177444458, "step": 80650 }, { "epoch": 0.40064373205646564, "grad_norm": 1.0286864042282104, "learning_rate": 1.3619792398903888e-05, "loss": 0.16528728008270263, "step": 80660 }, { "epoch": 0.4006934027398349, "grad_norm": 1.210970401763916, "learning_rate": 1.3618292731422165e-05, "loss": 0.1820974111557007, "step": 80670 }, { "epoch": 0.40074307342320414, "grad_norm": 1.0383388996124268, "learning_rate": 1.361679297030048e-05, "loss": 0.16683245897293092, "step": 80680 }, { "epoch": 0.4007927441065734, "grad_norm": 1.413360357284546, "learning_rate": 1.361529311557765e-05, "loss": 0.16965291500091553, "step": 80690 }, { "epoch": 0.4008424147899427, "grad_norm": 1.2431282997131348, "learning_rate": 1.3613793167292486e-05, "loss": 0.16025354862213134, "step": 80700 }, { "epoch": 0.40089208547331195, "grad_norm": 1.3654133081436157, "learning_rate": 1.3612293125483809e-05, "loss": 0.167389440536499, "step": 80710 }, { "epoch": 0.4009417561566812, "grad_norm": 1.198569416999817, "learning_rate": 1.3610792990190436e-05, "loss": 0.17138899564743043, "step": 80720 }, { "epoch": 0.40099142684005046, "grad_norm": 1.2906196117401123, "learning_rate": 1.3609292761451191e-05, "loss": 0.15897225141525267, "step": 80730 }, { "epoch": 0.4010410975234197, "grad_norm": 2.46097469329834, "learning_rate": 1.3607792439304904e-05, "loss": 0.16590166091918945, "step": 80740 }, { "epoch": 0.40109076820678896, "grad_norm": 1.029516577720642, "learning_rate": 1.3606292023790398e-05, "loss": 0.17858035564422609, "step": 80750 }, { "epoch": 0.40114043889015827, "grad_norm": 0.9706183671951294, "learning_rate": 1.3604791514946505e-05, "loss": 0.1617579698562622, "step": 80760 }, { "epoch": 0.4011901095735275, "grad_norm": 0.9310874938964844, "learning_rate": 1.3603290912812055e-05, "loss": 0.1718173384666443, "step": 80770 }, { "epoch": 0.40123978025689677, "grad_norm": 2.0439817905426025, "learning_rate": 1.3601790217425887e-05, "loss": 0.17737807035446168, "step": 80780 }, { "epoch": 0.401289450940266, "grad_norm": 1.1066471338272095, "learning_rate": 1.3600289428826836e-05, "loss": 0.17874958515167236, "step": 80790 }, { "epoch": 0.4013391216236353, "grad_norm": 1.3278263807296753, "learning_rate": 1.359878854705374e-05, "loss": 0.17401882410049438, "step": 80800 }, { "epoch": 0.4013887923070046, "grad_norm": 0.7097739577293396, "learning_rate": 1.3597287572145445e-05, "loss": 0.17058953046798705, "step": 80810 }, { "epoch": 0.40143846299037383, "grad_norm": 2.2050726413726807, "learning_rate": 1.3595786504140796e-05, "loss": 0.17504067420959474, "step": 80820 }, { "epoch": 0.4014881336737431, "grad_norm": 1.1604206562042236, "learning_rate": 1.3594285343078637e-05, "loss": 0.1661970853805542, "step": 80830 }, { "epoch": 0.40153780435711234, "grad_norm": 0.9440744519233704, "learning_rate": 1.3592784088997819e-05, "loss": 0.16739048957824706, "step": 80840 }, { "epoch": 0.4015874750404816, "grad_norm": 1.6139235496520996, "learning_rate": 1.3591282741937194e-05, "loss": 0.15467761754989623, "step": 80850 }, { "epoch": 0.40163714572385084, "grad_norm": 2.1809115409851074, "learning_rate": 1.3589781301935614e-05, "loss": 0.17624542713165284, "step": 80860 }, { "epoch": 0.40168681640722015, "grad_norm": 2.2468788623809814, "learning_rate": 1.3588279769031935e-05, "loss": 0.16010687351226807, "step": 80870 }, { "epoch": 0.4017364870905894, "grad_norm": 1.0289440155029297, "learning_rate": 1.3586778143265023e-05, "loss": 0.15249006748199462, "step": 80880 }, { "epoch": 0.40178615777395865, "grad_norm": 0.7583642601966858, "learning_rate": 1.3585276424673732e-05, "loss": 0.16702967882156372, "step": 80890 }, { "epoch": 0.4018358284573279, "grad_norm": 1.8108347654342651, "learning_rate": 1.3583774613296929e-05, "loss": 0.17223060131072998, "step": 80900 }, { "epoch": 0.40188549914069716, "grad_norm": 1.4316625595092773, "learning_rate": 1.358227270917348e-05, "loss": 0.17043623924255372, "step": 80910 }, { "epoch": 0.40193516982406646, "grad_norm": 0.9236723184585571, "learning_rate": 1.3580770712342256e-05, "loss": 0.15577086210250854, "step": 80920 }, { "epoch": 0.4019848405074357, "grad_norm": 1.5128391981124878, "learning_rate": 1.3579268622842121e-05, "loss": 0.16510728597640992, "step": 80930 }, { "epoch": 0.40203451119080497, "grad_norm": 1.210387110710144, "learning_rate": 1.3577766440711957e-05, "loss": 0.15952665805816652, "step": 80940 }, { "epoch": 0.4020841818741742, "grad_norm": 1.2045341730117798, "learning_rate": 1.3576264165990635e-05, "loss": 0.1737773060798645, "step": 80950 }, { "epoch": 0.40213385255754347, "grad_norm": 2.0702462196350098, "learning_rate": 1.3574761798717035e-05, "loss": 0.17304807901382446, "step": 80960 }, { "epoch": 0.4021835232409127, "grad_norm": 1.8304991722106934, "learning_rate": 1.3573259338930036e-05, "loss": 0.17260894775390626, "step": 80970 }, { "epoch": 0.40223319392428203, "grad_norm": 1.6007777452468872, "learning_rate": 1.3571756786668523e-05, "loss": 0.1661899209022522, "step": 80980 }, { "epoch": 0.4022828646076513, "grad_norm": 1.4541645050048828, "learning_rate": 1.3570254141971379e-05, "loss": 0.16862605810165404, "step": 80990 }, { "epoch": 0.40233253529102053, "grad_norm": 1.754076361656189, "learning_rate": 1.3568751404877496e-05, "loss": 0.17984330654144287, "step": 81000 }, { "epoch": 0.4023822059743898, "grad_norm": 2.4134650230407715, "learning_rate": 1.3567248575425757e-05, "loss": 0.16752761602401733, "step": 81010 }, { "epoch": 0.40243187665775904, "grad_norm": 1.4201165437698364, "learning_rate": 1.3565745653655066e-05, "loss": 0.16646066904067994, "step": 81020 }, { "epoch": 0.40248154734112834, "grad_norm": 1.3384920358657837, "learning_rate": 1.3564242639604305e-05, "loss": 0.15424933433532714, "step": 81030 }, { "epoch": 0.4025312180244976, "grad_norm": 1.1167553663253784, "learning_rate": 1.356273953331238e-05, "loss": 0.16903719902038575, "step": 81040 }, { "epoch": 0.40258088870786685, "grad_norm": 0.8341071009635925, "learning_rate": 1.3561236334818189e-05, "loss": 0.17167272567749023, "step": 81050 }, { "epoch": 0.4026305593912361, "grad_norm": 1.7610771656036377, "learning_rate": 1.3559733044160634e-05, "loss": 0.17119014263153076, "step": 81060 }, { "epoch": 0.40268023007460535, "grad_norm": 1.4753483533859253, "learning_rate": 1.3558229661378617e-05, "loss": 0.16298387050628663, "step": 81070 }, { "epoch": 0.4027299007579746, "grad_norm": 0.9969508051872253, "learning_rate": 1.3556726186511047e-05, "loss": 0.15580308437347412, "step": 81080 }, { "epoch": 0.4027795714413439, "grad_norm": 0.7796233892440796, "learning_rate": 1.3555222619596836e-05, "loss": 0.1550530433654785, "step": 81090 }, { "epoch": 0.40282924212471316, "grad_norm": 2.1446752548217773, "learning_rate": 1.355371896067489e-05, "loss": 0.17221388816833497, "step": 81100 }, { "epoch": 0.4028789128080824, "grad_norm": 1.0318219661712646, "learning_rate": 1.3552215209784128e-05, "loss": 0.16518762111663818, "step": 81110 }, { "epoch": 0.40292858349145166, "grad_norm": 1.850282907485962, "learning_rate": 1.3550711366963466e-05, "loss": 0.1594729542732239, "step": 81120 }, { "epoch": 0.4029782541748209, "grad_norm": 0.9765130877494812, "learning_rate": 1.354920743225182e-05, "loss": 0.16972259283065796, "step": 81130 }, { "epoch": 0.4030279248581902, "grad_norm": 1.9051398038864136, "learning_rate": 1.3547703405688111e-05, "loss": 0.15974726676940917, "step": 81140 }, { "epoch": 0.4030775955415595, "grad_norm": 1.308536171913147, "learning_rate": 1.3546199287311266e-05, "loss": 0.16937049627304077, "step": 81150 }, { "epoch": 0.4031272662249287, "grad_norm": 2.335200786590576, "learning_rate": 1.3544695077160208e-05, "loss": 0.17918734550476073, "step": 81160 }, { "epoch": 0.403176936908298, "grad_norm": 1.7147729396820068, "learning_rate": 1.354319077527387e-05, "loss": 0.17847583293914795, "step": 81170 }, { "epoch": 0.40322660759166723, "grad_norm": 0.8459293842315674, "learning_rate": 1.3541686381691177e-05, "loss": 0.1678842306137085, "step": 81180 }, { "epoch": 0.4032762782750365, "grad_norm": 0.9941094517707825, "learning_rate": 1.3540181896451068e-05, "loss": 0.15801507234573364, "step": 81190 }, { "epoch": 0.4033259489584058, "grad_norm": 1.2661538124084473, "learning_rate": 1.3538677319592468e-05, "loss": 0.16583397388458251, "step": 81200 }, { "epoch": 0.40337561964177504, "grad_norm": 0.8427417874336243, "learning_rate": 1.3537172651154327e-05, "loss": 0.1647179365158081, "step": 81210 }, { "epoch": 0.4034252903251443, "grad_norm": 1.5130358934402466, "learning_rate": 1.353566789117558e-05, "loss": 0.16634445190429686, "step": 81220 }, { "epoch": 0.40347496100851354, "grad_norm": 0.888232946395874, "learning_rate": 1.3534163039695164e-05, "loss": 0.16480512619018556, "step": 81230 }, { "epoch": 0.4035246316918828, "grad_norm": 1.6065813302993774, "learning_rate": 1.3532658096752035e-05, "loss": 0.1688974380493164, "step": 81240 }, { "epoch": 0.4035743023752521, "grad_norm": 1.5054279565811157, "learning_rate": 1.3531153062385137e-05, "loss": 0.16958601474761964, "step": 81250 }, { "epoch": 0.40362397305862135, "grad_norm": 1.341915249824524, "learning_rate": 1.3529647936633412e-05, "loss": 0.17305750846862794, "step": 81260 }, { "epoch": 0.4036736437419906, "grad_norm": 1.06436026096344, "learning_rate": 1.352814271953582e-05, "loss": 0.17865197658538817, "step": 81270 }, { "epoch": 0.40372331442535986, "grad_norm": 1.6244292259216309, "learning_rate": 1.3526637411131312e-05, "loss": 0.1718080997467041, "step": 81280 }, { "epoch": 0.4037729851087291, "grad_norm": 0.9963334798812866, "learning_rate": 1.3525132011458846e-05, "loss": 0.1644853711128235, "step": 81290 }, { "epoch": 0.40382265579209836, "grad_norm": 1.2356740236282349, "learning_rate": 1.3523626520557382e-05, "loss": 0.16591296195983887, "step": 81300 }, { "epoch": 0.40387232647546767, "grad_norm": 1.2853761911392212, "learning_rate": 1.3522120938465882e-05, "loss": 0.17058825492858887, "step": 81310 }, { "epoch": 0.4039219971588369, "grad_norm": 0.9890192747116089, "learning_rate": 1.3520615265223305e-05, "loss": 0.17288577556610107, "step": 81320 }, { "epoch": 0.4039716678422062, "grad_norm": 1.1524014472961426, "learning_rate": 1.3519109500868624e-05, "loss": 0.16044120788574218, "step": 81330 }, { "epoch": 0.4040213385255754, "grad_norm": 1.2080788612365723, "learning_rate": 1.3517603645440797e-05, "loss": 0.18085412979125975, "step": 81340 }, { "epoch": 0.4040710092089447, "grad_norm": 1.9678220748901367, "learning_rate": 1.351609769897881e-05, "loss": 0.17379777431488036, "step": 81350 }, { "epoch": 0.404120679892314, "grad_norm": 1.426156759262085, "learning_rate": 1.3514591661521625e-05, "loss": 0.1630722999572754, "step": 81360 }, { "epoch": 0.40417035057568323, "grad_norm": 1.1514345407485962, "learning_rate": 1.3513085533108224e-05, "loss": 0.16694159507751466, "step": 81370 }, { "epoch": 0.4042200212590525, "grad_norm": 1.2462571859359741, "learning_rate": 1.3511579313777579e-05, "loss": 0.16007821559906005, "step": 81380 }, { "epoch": 0.40426969194242174, "grad_norm": 0.9940521121025085, "learning_rate": 1.3510073003568673e-05, "loss": 0.17109763622283936, "step": 81390 }, { "epoch": 0.404319362625791, "grad_norm": 2.0857341289520264, "learning_rate": 1.3508566602520487e-05, "loss": 0.16880409717559813, "step": 81400 }, { "epoch": 0.40436903330916024, "grad_norm": 1.0735000371932983, "learning_rate": 1.3507060110672012e-05, "loss": 0.18005300760269166, "step": 81410 }, { "epoch": 0.40441870399252955, "grad_norm": 2.244842052459717, "learning_rate": 1.350555352806223e-05, "loss": 0.17092372179031373, "step": 81420 }, { "epoch": 0.4044683746758988, "grad_norm": 0.6770842671394348, "learning_rate": 1.3504046854730133e-05, "loss": 0.15307135581970216, "step": 81430 }, { "epoch": 0.40451804535926805, "grad_norm": 1.407084584236145, "learning_rate": 1.350254009071471e-05, "loss": 0.16903150081634521, "step": 81440 }, { "epoch": 0.4045677160426373, "grad_norm": 1.4699411392211914, "learning_rate": 1.350103323605496e-05, "loss": 0.16979488134384155, "step": 81450 }, { "epoch": 0.40461738672600656, "grad_norm": 0.7050635814666748, "learning_rate": 1.3499526290789876e-05, "loss": 0.17660911083221437, "step": 81460 }, { "epoch": 0.40466705740937586, "grad_norm": 0.6875209808349609, "learning_rate": 1.349801925495846e-05, "loss": 0.16570115089416504, "step": 81470 }, { "epoch": 0.4047167280927451, "grad_norm": 1.2748280763626099, "learning_rate": 1.3496512128599713e-05, "loss": 0.160525643825531, "step": 81480 }, { "epoch": 0.40476639877611437, "grad_norm": 1.6511479616165161, "learning_rate": 1.3495004911752635e-05, "loss": 0.18511816263198852, "step": 81490 }, { "epoch": 0.4048160694594836, "grad_norm": 1.7033793926239014, "learning_rate": 1.3493497604456239e-05, "loss": 0.15189203023910522, "step": 81500 }, { "epoch": 0.40486574014285287, "grad_norm": 1.363853096961975, "learning_rate": 1.3491990206749526e-05, "loss": 0.16444773674011232, "step": 81510 }, { "epoch": 0.4049154108262221, "grad_norm": 0.7850697040557861, "learning_rate": 1.3490482718671514e-05, "loss": 0.18502569198608398, "step": 81520 }, { "epoch": 0.40496508150959143, "grad_norm": 1.3981918096542358, "learning_rate": 1.348897514026121e-05, "loss": 0.171664822101593, "step": 81530 }, { "epoch": 0.4050147521929607, "grad_norm": 0.6555158495903015, "learning_rate": 1.3487467471557633e-05, "loss": 0.16076405048370362, "step": 81540 }, { "epoch": 0.40506442287632993, "grad_norm": 0.7589845657348633, "learning_rate": 1.3485959712599802e-05, "loss": 0.16501415967941285, "step": 81550 }, { "epoch": 0.4051140935596992, "grad_norm": 0.977645993232727, "learning_rate": 1.3484451863426733e-05, "loss": 0.16355679035186768, "step": 81560 }, { "epoch": 0.40516376424306844, "grad_norm": 0.9711083173751831, "learning_rate": 1.3482943924077452e-05, "loss": 0.16780505180358887, "step": 81570 }, { "epoch": 0.40521343492643774, "grad_norm": 1.3604785203933716, "learning_rate": 1.3481435894590981e-05, "loss": 0.1836814045906067, "step": 81580 }, { "epoch": 0.405263105609807, "grad_norm": 1.1623153686523438, "learning_rate": 1.347992777500635e-05, "loss": 0.17015767097473145, "step": 81590 }, { "epoch": 0.40531277629317625, "grad_norm": 1.5324925184249878, "learning_rate": 1.3478419565362588e-05, "loss": 0.1618481159210205, "step": 81600 }, { "epoch": 0.4053624469765455, "grad_norm": 0.7197867035865784, "learning_rate": 1.3476911265698723e-05, "loss": 0.1628710389137268, "step": 81610 }, { "epoch": 0.40541211765991475, "grad_norm": 0.891009509563446, "learning_rate": 1.3475402876053796e-05, "loss": 0.15135424137115477, "step": 81620 }, { "epoch": 0.405461788343284, "grad_norm": 1.0079929828643799, "learning_rate": 1.3473894396466837e-05, "loss": 0.163216233253479, "step": 81630 }, { "epoch": 0.4055114590266533, "grad_norm": 0.7875277996063232, "learning_rate": 1.3472385826976893e-05, "loss": 0.1679783582687378, "step": 81640 }, { "epoch": 0.40556112971002256, "grad_norm": 1.1022404432296753, "learning_rate": 1.3470877167622996e-05, "loss": 0.1654643416404724, "step": 81650 }, { "epoch": 0.4056108003933918, "grad_norm": 4.600081443786621, "learning_rate": 1.3469368418444192e-05, "loss": 0.16435234546661376, "step": 81660 }, { "epoch": 0.40566047107676106, "grad_norm": 1.65735924243927, "learning_rate": 1.3467859579479532e-05, "loss": 0.15500872135162352, "step": 81670 }, { "epoch": 0.4057101417601303, "grad_norm": 0.9724518060684204, "learning_rate": 1.3466350650768058e-05, "loss": 0.172591495513916, "step": 81680 }, { "epoch": 0.4057598124434996, "grad_norm": 0.9174874424934387, "learning_rate": 1.3464841632348824e-05, "loss": 0.16823580265045165, "step": 81690 }, { "epoch": 0.4058094831268689, "grad_norm": 0.7882300615310669, "learning_rate": 1.3463332524260883e-05, "loss": 0.1707080602645874, "step": 81700 }, { "epoch": 0.4058591538102381, "grad_norm": 0.7476103901863098, "learning_rate": 1.3461823326543288e-05, "loss": 0.169083309173584, "step": 81710 }, { "epoch": 0.4059088244936074, "grad_norm": 2.319887161254883, "learning_rate": 1.3460314039235097e-05, "loss": 0.1693270206451416, "step": 81720 }, { "epoch": 0.40595849517697663, "grad_norm": 0.7227938175201416, "learning_rate": 1.3458804662375369e-05, "loss": 0.17845678329467773, "step": 81730 }, { "epoch": 0.4060081658603459, "grad_norm": 1.6908529996871948, "learning_rate": 1.3457295196003167e-05, "loss": 0.15288732051849366, "step": 81740 }, { "epoch": 0.4060578365437152, "grad_norm": 0.9152007102966309, "learning_rate": 1.3455785640157557e-05, "loss": 0.15375144481658937, "step": 81750 }, { "epoch": 0.40610750722708444, "grad_norm": 0.879769504070282, "learning_rate": 1.3454275994877604e-05, "loss": 0.16387016773223878, "step": 81760 }, { "epoch": 0.4061571779104537, "grad_norm": 0.9499111175537109, "learning_rate": 1.3452766260202375e-05, "loss": 0.16114968061447144, "step": 81770 }, { "epoch": 0.40620684859382294, "grad_norm": 0.8343245983123779, "learning_rate": 1.3451256436170945e-05, "loss": 0.1758420467376709, "step": 81780 }, { "epoch": 0.4062565192771922, "grad_norm": 1.2230368852615356, "learning_rate": 1.3449746522822386e-05, "loss": 0.16625566482543946, "step": 81790 }, { "epoch": 0.4063061899605615, "grad_norm": 1.4021897315979004, "learning_rate": 1.3448236520195774e-05, "loss": 0.15937163829803466, "step": 81800 }, { "epoch": 0.40635586064393076, "grad_norm": 0.7612698674201965, "learning_rate": 1.3446726428330186e-05, "loss": 0.1554018497467041, "step": 81810 }, { "epoch": 0.4064055313273, "grad_norm": 0.902716338634491, "learning_rate": 1.3445216247264704e-05, "loss": 0.16092779636383056, "step": 81820 }, { "epoch": 0.40645520201066926, "grad_norm": 1.2892546653747559, "learning_rate": 1.3443705977038415e-05, "loss": 0.1623722195625305, "step": 81830 }, { "epoch": 0.4065048726940385, "grad_norm": 0.941444993019104, "learning_rate": 1.3442195617690397e-05, "loss": 0.1549154281616211, "step": 81840 }, { "epoch": 0.40655454337740776, "grad_norm": 1.2737257480621338, "learning_rate": 1.3440685169259742e-05, "loss": 0.19604976177215577, "step": 81850 }, { "epoch": 0.40660421406077707, "grad_norm": 1.1822293996810913, "learning_rate": 1.3439174631785537e-05, "loss": 0.1580902338027954, "step": 81860 }, { "epoch": 0.4066538847441463, "grad_norm": 1.8324666023254395, "learning_rate": 1.3437664005306875e-05, "loss": 0.16934468746185302, "step": 81870 }, { "epoch": 0.4067035554275156, "grad_norm": 3.0846784114837646, "learning_rate": 1.3436153289862852e-05, "loss": 0.1739504337310791, "step": 81880 }, { "epoch": 0.4067532261108848, "grad_norm": 0.8732671141624451, "learning_rate": 1.3434642485492561e-05, "loss": 0.17117791175842284, "step": 81890 }, { "epoch": 0.4068028967942541, "grad_norm": 0.9316058158874512, "learning_rate": 1.3433131592235106e-05, "loss": 0.1617289423942566, "step": 81900 }, { "epoch": 0.4068525674776234, "grad_norm": 2.1228530406951904, "learning_rate": 1.3431620610129585e-05, "loss": 0.16259011030197143, "step": 81910 }, { "epoch": 0.40690223816099264, "grad_norm": 1.4248511791229248, "learning_rate": 1.3430109539215101e-05, "loss": 0.1809898853302002, "step": 81920 }, { "epoch": 0.4069519088443619, "grad_norm": 0.8020604252815247, "learning_rate": 1.3428598379530764e-05, "loss": 0.16308683156967163, "step": 81930 }, { "epoch": 0.40700157952773114, "grad_norm": 0.8191528916358948, "learning_rate": 1.3427087131115677e-05, "loss": 0.17107901573181153, "step": 81940 }, { "epoch": 0.4070512502111004, "grad_norm": 0.979036808013916, "learning_rate": 1.3425575794008953e-05, "loss": 0.17366154193878175, "step": 81950 }, { "epoch": 0.40710092089446964, "grad_norm": 1.1101915836334229, "learning_rate": 1.3424064368249706e-05, "loss": 0.17811472415924073, "step": 81960 }, { "epoch": 0.40715059157783895, "grad_norm": 1.769246220588684, "learning_rate": 1.3422552853877051e-05, "loss": 0.17603880167007446, "step": 81970 }, { "epoch": 0.4072002622612082, "grad_norm": 0.8336567878723145, "learning_rate": 1.3421041250930102e-05, "loss": 0.16967518329620362, "step": 81980 }, { "epoch": 0.40724993294457745, "grad_norm": 1.558431625366211, "learning_rate": 1.3419529559447984e-05, "loss": 0.16496520042419432, "step": 81990 }, { "epoch": 0.4072996036279467, "grad_norm": 1.8376227617263794, "learning_rate": 1.3418017779469814e-05, "loss": 0.16005392074584962, "step": 82000 }, { "epoch": 0.40734927431131596, "grad_norm": 0.8159908056259155, "learning_rate": 1.3416505911034716e-05, "loss": 0.16970884799957275, "step": 82010 }, { "epoch": 0.40739894499468526, "grad_norm": 0.9951298236846924, "learning_rate": 1.3414993954181821e-05, "loss": 0.1754663109779358, "step": 82020 }, { "epoch": 0.4074486156780545, "grad_norm": 2.162539482116699, "learning_rate": 1.3413481908950257e-05, "loss": 0.1717858076095581, "step": 82030 }, { "epoch": 0.40749828636142377, "grad_norm": 1.3963137865066528, "learning_rate": 1.3411969775379152e-05, "loss": 0.16744849681854249, "step": 82040 }, { "epoch": 0.407547957044793, "grad_norm": 1.7968722581863403, "learning_rate": 1.341045755350764e-05, "loss": 0.16052359342575073, "step": 82050 }, { "epoch": 0.40759762772816227, "grad_norm": 0.8005324602127075, "learning_rate": 1.3408945243374862e-05, "loss": 0.17774254083633423, "step": 82060 }, { "epoch": 0.4076472984115315, "grad_norm": 1.056878924369812, "learning_rate": 1.3407432845019947e-05, "loss": 0.17140910625457764, "step": 82070 }, { "epoch": 0.40769696909490083, "grad_norm": 0.8300901651382446, "learning_rate": 1.340592035848204e-05, "loss": 0.17378907203674315, "step": 82080 }, { "epoch": 0.4077466397782701, "grad_norm": 0.8068904280662537, "learning_rate": 1.3404407783800287e-05, "loss": 0.1596415877342224, "step": 82090 }, { "epoch": 0.40779631046163933, "grad_norm": 1.424230694770813, "learning_rate": 1.3402895121013828e-05, "loss": 0.16197556257247925, "step": 82100 }, { "epoch": 0.4078459811450086, "grad_norm": 2.3620457649230957, "learning_rate": 1.3401382370161813e-05, "loss": 0.16730140447616576, "step": 82110 }, { "epoch": 0.40789565182837784, "grad_norm": 0.6680508852005005, "learning_rate": 1.3399869531283387e-05, "loss": 0.16631096601486206, "step": 82120 }, { "epoch": 0.40794532251174714, "grad_norm": 1.6753123998641968, "learning_rate": 1.3398356604417707e-05, "loss": 0.18300809860229492, "step": 82130 }, { "epoch": 0.4079949931951164, "grad_norm": 1.8114302158355713, "learning_rate": 1.3396843589603923e-05, "loss": 0.14887701272964476, "step": 82140 }, { "epoch": 0.40804466387848565, "grad_norm": 1.4115028381347656, "learning_rate": 1.3395330486881193e-05, "loss": 0.167035710811615, "step": 82150 }, { "epoch": 0.4080943345618549, "grad_norm": 1.0374877452850342, "learning_rate": 1.3393817296288674e-05, "loss": 0.1631094694137573, "step": 82160 }, { "epoch": 0.40814400524522415, "grad_norm": 1.8618358373641968, "learning_rate": 1.3392304017865529e-05, "loss": 0.16382029056549072, "step": 82170 }, { "epoch": 0.4081936759285934, "grad_norm": 1.1627538204193115, "learning_rate": 1.339079065165092e-05, "loss": 0.15391579866409302, "step": 82180 }, { "epoch": 0.4082433466119627, "grad_norm": 0.7446494102478027, "learning_rate": 1.3389277197684015e-05, "loss": 0.15329819917678833, "step": 82190 }, { "epoch": 0.40829301729533196, "grad_norm": 0.9370869398117065, "learning_rate": 1.3387763656003976e-05, "loss": 0.17437362670898438, "step": 82200 }, { "epoch": 0.4083426879787012, "grad_norm": 0.5942883491516113, "learning_rate": 1.3386250026649976e-05, "loss": 0.1817166805267334, "step": 82210 }, { "epoch": 0.40839235866207046, "grad_norm": 1.1433058977127075, "learning_rate": 1.3384736309661187e-05, "loss": 0.15451717376708984, "step": 82220 }, { "epoch": 0.4084420293454397, "grad_norm": 1.0220003128051758, "learning_rate": 1.3383222505076784e-05, "loss": 0.16457753181457518, "step": 82230 }, { "epoch": 0.408491700028809, "grad_norm": 0.5815691947937012, "learning_rate": 1.3381708612935942e-05, "loss": 0.17470378875732423, "step": 82240 }, { "epoch": 0.4085413707121783, "grad_norm": 0.9754931926727295, "learning_rate": 1.3380194633277843e-05, "loss": 0.16407408714294433, "step": 82250 }, { "epoch": 0.4085910413955475, "grad_norm": 0.8056095838546753, "learning_rate": 1.3378680566141665e-05, "loss": 0.16921441555023192, "step": 82260 }, { "epoch": 0.4086407120789168, "grad_norm": 1.5372998714447021, "learning_rate": 1.3377166411566589e-05, "loss": 0.1681426763534546, "step": 82270 }, { "epoch": 0.40869038276228603, "grad_norm": 1.1612954139709473, "learning_rate": 1.3375652169591807e-05, "loss": 0.17887927293777467, "step": 82280 }, { "epoch": 0.4087400534456553, "grad_norm": 1.0359413623809814, "learning_rate": 1.3374137840256503e-05, "loss": 0.1618059277534485, "step": 82290 }, { "epoch": 0.4087897241290246, "grad_norm": 1.0404760837554932, "learning_rate": 1.337262342359987e-05, "loss": 0.17107913494110108, "step": 82300 }, { "epoch": 0.40883939481239384, "grad_norm": 1.441543698310852, "learning_rate": 1.3371108919661098e-05, "loss": 0.1709665298461914, "step": 82310 }, { "epoch": 0.4088890654957631, "grad_norm": 0.9585732817649841, "learning_rate": 1.3369594328479382e-05, "loss": 0.16564009189605713, "step": 82320 }, { "epoch": 0.40893873617913234, "grad_norm": 0.9795976877212524, "learning_rate": 1.3368079650093922e-05, "loss": 0.1664275646209717, "step": 82330 }, { "epoch": 0.4089884068625016, "grad_norm": 0.7813146710395813, "learning_rate": 1.3366564884543913e-05, "loss": 0.16288180351257325, "step": 82340 }, { "epoch": 0.40903807754587085, "grad_norm": 1.9347513914108276, "learning_rate": 1.3365050031868556e-05, "loss": 0.17532800436019896, "step": 82350 }, { "epoch": 0.40908774822924016, "grad_norm": 0.9982765316963196, "learning_rate": 1.336353509210706e-05, "loss": 0.16185483932495118, "step": 82360 }, { "epoch": 0.4091374189126094, "grad_norm": 0.8545798063278198, "learning_rate": 1.336202006529863e-05, "loss": 0.1644674301147461, "step": 82370 }, { "epoch": 0.40918708959597866, "grad_norm": 1.05491042137146, "learning_rate": 1.3360504951482472e-05, "loss": 0.16942133903503417, "step": 82380 }, { "epoch": 0.4092367602793479, "grad_norm": 1.5760844945907593, "learning_rate": 1.3358989750697797e-05, "loss": 0.16309943199157714, "step": 82390 }, { "epoch": 0.40928643096271716, "grad_norm": 1.3496085405349731, "learning_rate": 1.3357474462983817e-05, "loss": 0.17004547119140626, "step": 82400 }, { "epoch": 0.40933610164608647, "grad_norm": 0.737912118434906, "learning_rate": 1.3355959088379745e-05, "loss": 0.16981658935546876, "step": 82410 }, { "epoch": 0.4093857723294557, "grad_norm": 0.9003021121025085, "learning_rate": 1.3354443626924805e-05, "loss": 0.16861554384231567, "step": 82420 }, { "epoch": 0.409435443012825, "grad_norm": 2.709749460220337, "learning_rate": 1.3352928078658209e-05, "loss": 0.1669975996017456, "step": 82430 }, { "epoch": 0.4094851136961942, "grad_norm": 1.497795581817627, "learning_rate": 1.3351412443619188e-05, "loss": 0.1645967960357666, "step": 82440 }, { "epoch": 0.4095347843795635, "grad_norm": 0.9448518753051758, "learning_rate": 1.3349896721846956e-05, "loss": 0.15965750217437744, "step": 82450 }, { "epoch": 0.40958445506293273, "grad_norm": 1.4846049547195435, "learning_rate": 1.3348380913380744e-05, "loss": 0.16667702198028564, "step": 82460 }, { "epoch": 0.40963412574630204, "grad_norm": 1.1524748802185059, "learning_rate": 1.3346865018259782e-05, "loss": 0.1610150456428528, "step": 82470 }, { "epoch": 0.4096837964296713, "grad_norm": 1.0340386629104614, "learning_rate": 1.3345349036523298e-05, "loss": 0.163567054271698, "step": 82480 }, { "epoch": 0.40973346711304054, "grad_norm": 1.0567667484283447, "learning_rate": 1.3343832968210526e-05, "loss": 0.16860209703445433, "step": 82490 }, { "epoch": 0.4097831377964098, "grad_norm": 0.7541680335998535, "learning_rate": 1.3342316813360702e-05, "loss": 0.1562468409538269, "step": 82500 }, { "epoch": 0.40983280847977904, "grad_norm": 0.8913496732711792, "learning_rate": 1.3340800572013065e-05, "loss": 0.18302063941955565, "step": 82510 }, { "epoch": 0.40988247916314835, "grad_norm": 1.2923811674118042, "learning_rate": 1.3339284244206847e-05, "loss": 0.17341746091842652, "step": 82520 }, { "epoch": 0.4099321498465176, "grad_norm": 2.2742044925689697, "learning_rate": 1.3337767829981301e-05, "loss": 0.17206161022186278, "step": 82530 }, { "epoch": 0.40998182052988685, "grad_norm": 1.1794086694717407, "learning_rate": 1.3336251329375662e-05, "loss": 0.17624189853668212, "step": 82540 }, { "epoch": 0.4100314912132561, "grad_norm": 0.6149118542671204, "learning_rate": 1.333473474242918e-05, "loss": 0.15174856185913085, "step": 82550 }, { "epoch": 0.41008116189662536, "grad_norm": 1.869109869003296, "learning_rate": 1.3333218069181107e-05, "loss": 0.16939029693603516, "step": 82560 }, { "epoch": 0.4101308325799946, "grad_norm": 2.015796184539795, "learning_rate": 1.3331701309670688e-05, "loss": 0.1658486843109131, "step": 82570 }, { "epoch": 0.4101805032633639, "grad_norm": 1.4898384809494019, "learning_rate": 1.3330184463937178e-05, "loss": 0.16479687690734862, "step": 82580 }, { "epoch": 0.41023017394673317, "grad_norm": 1.9439635276794434, "learning_rate": 1.3328667532019837e-05, "loss": 0.17146142721176147, "step": 82590 }, { "epoch": 0.4102798446301024, "grad_norm": 1.4485934972763062, "learning_rate": 1.3327150513957917e-05, "loss": 0.15927942991256713, "step": 82600 }, { "epoch": 0.41032951531347167, "grad_norm": 2.316500425338745, "learning_rate": 1.3325633409790674e-05, "loss": 0.16588441133499146, "step": 82610 }, { "epoch": 0.4103791859968409, "grad_norm": 1.0694366693496704, "learning_rate": 1.3324116219557381e-05, "loss": 0.16224884986877441, "step": 82620 }, { "epoch": 0.41042885668021023, "grad_norm": 1.5826901197433472, "learning_rate": 1.3322598943297296e-05, "loss": 0.16783833503723145, "step": 82630 }, { "epoch": 0.4104785273635795, "grad_norm": 0.8066660165786743, "learning_rate": 1.3321081581049685e-05, "loss": 0.16555341482162475, "step": 82640 }, { "epoch": 0.41052819804694873, "grad_norm": 0.9404478073120117, "learning_rate": 1.331956413285382e-05, "loss": 0.1663972854614258, "step": 82650 }, { "epoch": 0.410577868730318, "grad_norm": 1.8325412273406982, "learning_rate": 1.331804659874897e-05, "loss": 0.1783261775970459, "step": 82660 }, { "epoch": 0.41062753941368724, "grad_norm": 0.8828571438789368, "learning_rate": 1.3316528978774408e-05, "loss": 0.1686178207397461, "step": 82670 }, { "epoch": 0.4106772100970565, "grad_norm": 1.076451301574707, "learning_rate": 1.3315011272969409e-05, "loss": 0.1694550633430481, "step": 82680 }, { "epoch": 0.4107268807804258, "grad_norm": 0.9453479051589966, "learning_rate": 1.3313493481373247e-05, "loss": 0.1835099458694458, "step": 82690 }, { "epoch": 0.41077655146379505, "grad_norm": 1.3541007041931152, "learning_rate": 1.3311975604025211e-05, "loss": 0.1558782696723938, "step": 82700 }, { "epoch": 0.4108262221471643, "grad_norm": 1.998454213142395, "learning_rate": 1.331045764096458e-05, "loss": 0.15333590507507325, "step": 82710 }, { "epoch": 0.41087589283053355, "grad_norm": 1.5533734560012817, "learning_rate": 1.3308939592230635e-05, "loss": 0.16558995246887206, "step": 82720 }, { "epoch": 0.4109255635139028, "grad_norm": 1.2617617845535278, "learning_rate": 1.3307421457862662e-05, "loss": 0.1630603313446045, "step": 82730 }, { "epoch": 0.4109752341972721, "grad_norm": 1.2845878601074219, "learning_rate": 1.330590323789995e-05, "loss": 0.1594540596008301, "step": 82740 }, { "epoch": 0.41102490488064136, "grad_norm": 1.957904577255249, "learning_rate": 1.3304384932381793e-05, "loss": 0.159721302986145, "step": 82750 }, { "epoch": 0.4110745755640106, "grad_norm": 0.7731145024299622, "learning_rate": 1.330286654134748e-05, "loss": 0.16877353191375732, "step": 82760 }, { "epoch": 0.41112424624737987, "grad_norm": 1.3956096172332764, "learning_rate": 1.3301348064836312e-05, "loss": 0.1500009536743164, "step": 82770 }, { "epoch": 0.4111739169307491, "grad_norm": 1.2594943046569824, "learning_rate": 1.3299829502887584e-05, "loss": 0.17597537040710448, "step": 82780 }, { "epoch": 0.41122358761411837, "grad_norm": 1.4383667707443237, "learning_rate": 1.3298310855540595e-05, "loss": 0.18124079704284668, "step": 82790 }, { "epoch": 0.4112732582974877, "grad_norm": 0.9653117656707764, "learning_rate": 1.3296792122834647e-05, "loss": 0.1722014307975769, "step": 82800 }, { "epoch": 0.41132292898085693, "grad_norm": 1.7215089797973633, "learning_rate": 1.3295273304809044e-05, "loss": 0.15837219953536988, "step": 82810 }, { "epoch": 0.4113725996642262, "grad_norm": 0.8949782252311707, "learning_rate": 1.3293754401503092e-05, "loss": 0.17783823013305664, "step": 82820 }, { "epoch": 0.41142227034759543, "grad_norm": 0.9218804240226746, "learning_rate": 1.3292235412956103e-05, "loss": 0.16795201301574708, "step": 82830 }, { "epoch": 0.4114719410309647, "grad_norm": 1.7037273645401, "learning_rate": 1.3290716339207383e-05, "loss": 0.1522819995880127, "step": 82840 }, { "epoch": 0.411521611714334, "grad_norm": 1.5578783750534058, "learning_rate": 1.3289197180296246e-05, "loss": 0.17656536102294923, "step": 82850 }, { "epoch": 0.41157128239770324, "grad_norm": 0.8755620718002319, "learning_rate": 1.3287677936262014e-05, "loss": 0.16775647401809693, "step": 82860 }, { "epoch": 0.4116209530810725, "grad_norm": 1.2588227987289429, "learning_rate": 1.3286158607143996e-05, "loss": 0.18643271923065186, "step": 82870 }, { "epoch": 0.41167062376444175, "grad_norm": 0.8860414028167725, "learning_rate": 1.3284639192981512e-05, "loss": 0.16514053344726562, "step": 82880 }, { "epoch": 0.411720294447811, "grad_norm": 1.0551971197128296, "learning_rate": 1.3283119693813888e-05, "loss": 0.15972750186920165, "step": 82890 }, { "epoch": 0.41176996513118025, "grad_norm": 0.9502556324005127, "learning_rate": 1.3281600109680448e-05, "loss": 0.15957080125808715, "step": 82900 }, { "epoch": 0.41181963581454956, "grad_norm": 0.680407702922821, "learning_rate": 1.3280080440620515e-05, "loss": 0.16325095891952515, "step": 82910 }, { "epoch": 0.4118693064979188, "grad_norm": 1.543246865272522, "learning_rate": 1.3278560686673421e-05, "loss": 0.15928889513015748, "step": 82920 }, { "epoch": 0.41191897718128806, "grad_norm": 1.2257161140441895, "learning_rate": 1.3277040847878492e-05, "loss": 0.16374150514602662, "step": 82930 }, { "epoch": 0.4119686478646573, "grad_norm": 1.5438249111175537, "learning_rate": 1.3275520924275064e-05, "loss": 0.1765129566192627, "step": 82940 }, { "epoch": 0.41201831854802656, "grad_norm": 1.3448512554168701, "learning_rate": 1.3274000915902472e-05, "loss": 0.16974303722381592, "step": 82950 }, { "epoch": 0.41206798923139587, "grad_norm": 1.6558276414871216, "learning_rate": 1.3272480822800054e-05, "loss": 0.17027266025543214, "step": 82960 }, { "epoch": 0.4121176599147651, "grad_norm": 1.45026695728302, "learning_rate": 1.3270960645007147e-05, "loss": 0.16842570304870605, "step": 82970 }, { "epoch": 0.4121673305981344, "grad_norm": 1.0008379220962524, "learning_rate": 1.3269440382563094e-05, "loss": 0.16164373159408568, "step": 82980 }, { "epoch": 0.4122170012815036, "grad_norm": 0.8335357904434204, "learning_rate": 1.3267920035507238e-05, "loss": 0.17030941247940062, "step": 82990 }, { "epoch": 0.4122666719648729, "grad_norm": 2.5269627571105957, "learning_rate": 1.3266399603878926e-05, "loss": 0.17612555027008056, "step": 83000 }, { "epoch": 0.41231634264824213, "grad_norm": 0.6755022406578064, "learning_rate": 1.3264879087717505e-05, "loss": 0.1683830976486206, "step": 83010 }, { "epoch": 0.41236601333161144, "grad_norm": 0.978015661239624, "learning_rate": 1.3263358487062323e-05, "loss": 0.16747784614562988, "step": 83020 }, { "epoch": 0.4124156840149807, "grad_norm": 0.8671082854270935, "learning_rate": 1.3261837801952741e-05, "loss": 0.16075410842895507, "step": 83030 }, { "epoch": 0.41246535469834994, "grad_norm": 1.0042383670806885, "learning_rate": 1.3260317032428105e-05, "loss": 0.16834821701049804, "step": 83040 }, { "epoch": 0.4125150253817192, "grad_norm": 1.628786563873291, "learning_rate": 1.3258796178527779e-05, "loss": 0.16826972961425782, "step": 83050 }, { "epoch": 0.41256469606508844, "grad_norm": 1.005385398864746, "learning_rate": 1.3257275240291113e-05, "loss": 0.1855364680290222, "step": 83060 }, { "epoch": 0.41261436674845775, "grad_norm": 0.9209679961204529, "learning_rate": 1.325575421775748e-05, "loss": 0.16692161560058594, "step": 83070 }, { "epoch": 0.412664037431827, "grad_norm": 2.289055585861206, "learning_rate": 1.325423311096623e-05, "loss": 0.16594429016113282, "step": 83080 }, { "epoch": 0.41271370811519625, "grad_norm": 0.9786763787269592, "learning_rate": 1.3252711919956737e-05, "loss": 0.1658931016921997, "step": 83090 }, { "epoch": 0.4127633787985655, "grad_norm": 1.3735483884811401, "learning_rate": 1.325119064476837e-05, "loss": 0.17810287475585937, "step": 83100 }, { "epoch": 0.41281304948193476, "grad_norm": 1.2700530290603638, "learning_rate": 1.3249669285440494e-05, "loss": 0.16908376216888427, "step": 83110 }, { "epoch": 0.412862720165304, "grad_norm": 2.3893072605133057, "learning_rate": 1.3248147842012484e-05, "loss": 0.16454758644104003, "step": 83120 }, { "epoch": 0.4129123908486733, "grad_norm": 0.7288709282875061, "learning_rate": 1.3246626314523717e-05, "loss": 0.17008981704711915, "step": 83130 }, { "epoch": 0.41296206153204257, "grad_norm": 2.439831018447876, "learning_rate": 1.3245104703013562e-05, "loss": 0.17598962783813477, "step": 83140 }, { "epoch": 0.4130117322154118, "grad_norm": 2.023024082183838, "learning_rate": 1.3243583007521403e-05, "loss": 0.16518542766571045, "step": 83150 }, { "epoch": 0.41306140289878107, "grad_norm": 1.2548059225082397, "learning_rate": 1.3242061228086623e-05, "loss": 0.16198596954345704, "step": 83160 }, { "epoch": 0.4131110735821503, "grad_norm": 1.066834807395935, "learning_rate": 1.3240539364748597e-05, "loss": 0.16638996601104736, "step": 83170 }, { "epoch": 0.41316074426551963, "grad_norm": 2.0354762077331543, "learning_rate": 1.323901741754672e-05, "loss": 0.16037423610687257, "step": 83180 }, { "epoch": 0.4132104149488889, "grad_norm": 1.3773273229599, "learning_rate": 1.3237495386520371e-05, "loss": 0.1660891056060791, "step": 83190 }, { "epoch": 0.41326008563225813, "grad_norm": 1.6364861726760864, "learning_rate": 1.3235973271708946e-05, "loss": 0.1634663462638855, "step": 83200 }, { "epoch": 0.4133097563156274, "grad_norm": 0.9212201237678528, "learning_rate": 1.3234451073151833e-05, "loss": 0.15636823177337647, "step": 83210 }, { "epoch": 0.41335942699899664, "grad_norm": 1.3392380475997925, "learning_rate": 1.3232928790888427e-05, "loss": 0.1657848596572876, "step": 83220 }, { "epoch": 0.4134090976823659, "grad_norm": 1.319464921951294, "learning_rate": 1.3231406424958122e-05, "loss": 0.16465189456939697, "step": 83230 }, { "epoch": 0.4134587683657352, "grad_norm": 0.5593464374542236, "learning_rate": 1.3229883975400321e-05, "loss": 0.15894320011138915, "step": 83240 }, { "epoch": 0.41350843904910445, "grad_norm": 1.4974473714828491, "learning_rate": 1.322836144225442e-05, "loss": 0.1653801679611206, "step": 83250 }, { "epoch": 0.4135581097324737, "grad_norm": 1.739545464515686, "learning_rate": 1.3226838825559823e-05, "loss": 0.17111437320709227, "step": 83260 }, { "epoch": 0.41360778041584295, "grad_norm": 0.9532568454742432, "learning_rate": 1.3225316125355934e-05, "loss": 0.1834789991378784, "step": 83270 }, { "epoch": 0.4136574510992122, "grad_norm": 1.4702520370483398, "learning_rate": 1.3223793341682159e-05, "loss": 0.17420847415924073, "step": 83280 }, { "epoch": 0.4137071217825815, "grad_norm": 1.1215876340866089, "learning_rate": 1.322227047457791e-05, "loss": 0.16801862716674804, "step": 83290 }, { "epoch": 0.41375679246595076, "grad_norm": 1.6060999631881714, "learning_rate": 1.3220747524082595e-05, "loss": 0.16575987339019777, "step": 83300 }, { "epoch": 0.41380646314932, "grad_norm": 0.769130527973175, "learning_rate": 1.3219224490235634e-05, "loss": 0.15850729942321778, "step": 83310 }, { "epoch": 0.41385613383268927, "grad_norm": 3.026252269744873, "learning_rate": 1.3217701373076434e-05, "loss": 0.17270994186401367, "step": 83320 }, { "epoch": 0.4139058045160585, "grad_norm": 0.7321016788482666, "learning_rate": 1.3216178172644419e-05, "loss": 0.18013471364974976, "step": 83330 }, { "epoch": 0.41395547519942777, "grad_norm": 1.2965844869613647, "learning_rate": 1.3214654888979001e-05, "loss": 0.17074551582336425, "step": 83340 }, { "epoch": 0.4140051458827971, "grad_norm": 1.1100207567214966, "learning_rate": 1.3213131522119612e-05, "loss": 0.17756955623626708, "step": 83350 }, { "epoch": 0.41405481656616633, "grad_norm": 1.328871250152588, "learning_rate": 1.3211608072105666e-05, "loss": 0.16679885387420654, "step": 83360 }, { "epoch": 0.4141044872495356, "grad_norm": 1.1481945514678955, "learning_rate": 1.3210084538976599e-05, "loss": 0.17882035970687865, "step": 83370 }, { "epoch": 0.41415415793290483, "grad_norm": 0.5991736650466919, "learning_rate": 1.3208560922771833e-05, "loss": 0.17145644426345824, "step": 83380 }, { "epoch": 0.4142038286162741, "grad_norm": 0.7366726398468018, "learning_rate": 1.3207037223530802e-05, "loss": 0.16207261085510255, "step": 83390 }, { "epoch": 0.4142534992996434, "grad_norm": 2.6640589237213135, "learning_rate": 1.3205513441292937e-05, "loss": 0.19679412841796876, "step": 83400 }, { "epoch": 0.41430316998301264, "grad_norm": 1.1297580003738403, "learning_rate": 1.3203989576097673e-05, "loss": 0.16929632425308228, "step": 83410 }, { "epoch": 0.4143528406663819, "grad_norm": 0.6971924901008606, "learning_rate": 1.3202465627984444e-05, "loss": 0.16810400485992433, "step": 83420 }, { "epoch": 0.41440251134975115, "grad_norm": 1.811053991317749, "learning_rate": 1.3200941596992695e-05, "loss": 0.17701318264007568, "step": 83430 }, { "epoch": 0.4144521820331204, "grad_norm": 1.0574406385421753, "learning_rate": 1.3199417483161865e-05, "loss": 0.16166143417358397, "step": 83440 }, { "epoch": 0.41450185271648965, "grad_norm": 0.8927680253982544, "learning_rate": 1.3197893286531395e-05, "loss": 0.16599514484405517, "step": 83450 }, { "epoch": 0.41455152339985896, "grad_norm": 1.260908842086792, "learning_rate": 1.3196369007140733e-05, "loss": 0.1726003408432007, "step": 83460 }, { "epoch": 0.4146011940832282, "grad_norm": 0.8132991790771484, "learning_rate": 1.3194844645029328e-05, "loss": 0.16690694093704223, "step": 83470 }, { "epoch": 0.41465086476659746, "grad_norm": 1.3839868307113647, "learning_rate": 1.3193320200236626e-05, "loss": 0.1641218066215515, "step": 83480 }, { "epoch": 0.4147005354499667, "grad_norm": 0.9542976021766663, "learning_rate": 1.319179567280208e-05, "loss": 0.1686945915222168, "step": 83490 }, { "epoch": 0.41475020613333596, "grad_norm": 1.6371819972991943, "learning_rate": 1.3190271062765147e-05, "loss": 0.16968425512313842, "step": 83500 }, { "epoch": 0.41479987681670527, "grad_norm": 1.5200139284133911, "learning_rate": 1.3188746370165282e-05, "loss": 0.16914435625076293, "step": 83510 }, { "epoch": 0.4148495475000745, "grad_norm": 1.795369267463684, "learning_rate": 1.3187221595041944e-05, "loss": 0.15930542945861817, "step": 83520 }, { "epoch": 0.4148992181834438, "grad_norm": 0.6713704466819763, "learning_rate": 1.3185696737434592e-05, "loss": 0.16616345643997193, "step": 83530 }, { "epoch": 0.414948888866813, "grad_norm": 1.515660047531128, "learning_rate": 1.318417179738269e-05, "loss": 0.168145751953125, "step": 83540 }, { "epoch": 0.4149985595501823, "grad_norm": 1.3093531131744385, "learning_rate": 1.31826467749257e-05, "loss": 0.17420032024383544, "step": 83550 }, { "epoch": 0.41504823023355153, "grad_norm": 1.5107791423797607, "learning_rate": 1.3181121670103095e-05, "loss": 0.1711880683898926, "step": 83560 }, { "epoch": 0.41509790091692084, "grad_norm": 1.980991244316101, "learning_rate": 1.3179596482954335e-05, "loss": 0.1677820563316345, "step": 83570 }, { "epoch": 0.4151475716002901, "grad_norm": 1.446668028831482, "learning_rate": 1.3178071213518902e-05, "loss": 0.16813143491744995, "step": 83580 }, { "epoch": 0.41519724228365934, "grad_norm": 1.7388677597045898, "learning_rate": 1.317654586183626e-05, "loss": 0.17053046226501464, "step": 83590 }, { "epoch": 0.4152469129670286, "grad_norm": 2.0816051959991455, "learning_rate": 1.317502042794589e-05, "loss": 0.1708444118499756, "step": 83600 }, { "epoch": 0.41529658365039784, "grad_norm": 1.4576791524887085, "learning_rate": 1.3173494911887269e-05, "loss": 0.1648649215698242, "step": 83610 }, { "epoch": 0.41534625433376715, "grad_norm": 1.5980972051620483, "learning_rate": 1.317196931369987e-05, "loss": 0.15347753763198851, "step": 83620 }, { "epoch": 0.4153959250171364, "grad_norm": 1.8990460634231567, "learning_rate": 1.3170443633423187e-05, "loss": 0.17104381322860718, "step": 83630 }, { "epoch": 0.41544559570050565, "grad_norm": 1.2598646879196167, "learning_rate": 1.3168917871096694e-05, "loss": 0.16329847574234008, "step": 83640 }, { "epoch": 0.4154952663838749, "grad_norm": 0.9984232187271118, "learning_rate": 1.3167392026759882e-05, "loss": 0.16515774726867677, "step": 83650 }, { "epoch": 0.41554493706724416, "grad_norm": 1.6257542371749878, "learning_rate": 1.3165866100452238e-05, "loss": 0.17216718196868896, "step": 83660 }, { "epoch": 0.4155946077506134, "grad_norm": 1.0302132368087769, "learning_rate": 1.3164340092213253e-05, "loss": 0.17125272750854492, "step": 83670 }, { "epoch": 0.4156442784339827, "grad_norm": 1.232433557510376, "learning_rate": 1.3162814002082416e-05, "loss": 0.16890087127685546, "step": 83680 }, { "epoch": 0.41569394911735197, "grad_norm": 0.9396535158157349, "learning_rate": 1.3161287830099225e-05, "loss": 0.1642477035522461, "step": 83690 }, { "epoch": 0.4157436198007212, "grad_norm": 1.4600833654403687, "learning_rate": 1.3159761576303178e-05, "loss": 0.1629119873046875, "step": 83700 }, { "epoch": 0.41579329048409047, "grad_norm": 1.9648617506027222, "learning_rate": 1.315823524073377e-05, "loss": 0.16689417362213135, "step": 83710 }, { "epoch": 0.4158429611674597, "grad_norm": 1.1206926107406616, "learning_rate": 1.3156708823430505e-05, "loss": 0.16388860940933228, "step": 83720 }, { "epoch": 0.41589263185082903, "grad_norm": 2.369929313659668, "learning_rate": 1.3155182324432887e-05, "loss": 0.1636215090751648, "step": 83730 }, { "epoch": 0.4159423025341983, "grad_norm": 0.9016634225845337, "learning_rate": 1.3153655743780418e-05, "loss": 0.16208169460296631, "step": 83740 }, { "epoch": 0.41599197321756753, "grad_norm": 1.5510234832763672, "learning_rate": 1.3152129081512603e-05, "loss": 0.17016191482543946, "step": 83750 }, { "epoch": 0.4160416439009368, "grad_norm": 1.674833059310913, "learning_rate": 1.3150602337668956e-05, "loss": 0.18693732023239135, "step": 83760 }, { "epoch": 0.41609131458430604, "grad_norm": 1.9209917783737183, "learning_rate": 1.3149075512288988e-05, "loss": 0.17340736389160155, "step": 83770 }, { "epoch": 0.4161409852676753, "grad_norm": 0.7688372731208801, "learning_rate": 1.3147548605412212e-05, "loss": 0.167241632938385, "step": 83780 }, { "epoch": 0.4161906559510446, "grad_norm": 1.0831258296966553, "learning_rate": 1.3146021617078146e-05, "loss": 0.175657320022583, "step": 83790 }, { "epoch": 0.41624032663441385, "grad_norm": 0.7100947499275208, "learning_rate": 1.3144494547326303e-05, "loss": 0.16435585021972657, "step": 83800 }, { "epoch": 0.4162899973177831, "grad_norm": 1.001883864402771, "learning_rate": 1.3142967396196206e-05, "loss": 0.16592087745666503, "step": 83810 }, { "epoch": 0.41633966800115235, "grad_norm": 0.7462778091430664, "learning_rate": 1.3141440163727374e-05, "loss": 0.1766801118850708, "step": 83820 }, { "epoch": 0.4163893386845216, "grad_norm": 1.8191068172454834, "learning_rate": 1.3139912849959338e-05, "loss": 0.17379904985427858, "step": 83830 }, { "epoch": 0.4164390093678909, "grad_norm": 1.5418635606765747, "learning_rate": 1.3138385454931616e-05, "loss": 0.16960542201995848, "step": 83840 }, { "epoch": 0.41648868005126016, "grad_norm": 0.9467248916625977, "learning_rate": 1.3136857978683743e-05, "loss": 0.17763081789016724, "step": 83850 }, { "epoch": 0.4165383507346294, "grad_norm": 0.8010486960411072, "learning_rate": 1.3135330421255248e-05, "loss": 0.1689859986305237, "step": 83860 }, { "epoch": 0.41658802141799867, "grad_norm": 0.8428260087966919, "learning_rate": 1.313380278268566e-05, "loss": 0.1663431167602539, "step": 83870 }, { "epoch": 0.4166376921013679, "grad_norm": 1.5585274696350098, "learning_rate": 1.3132275063014518e-05, "loss": 0.167530620098114, "step": 83880 }, { "epoch": 0.41668736278473717, "grad_norm": 1.457604169845581, "learning_rate": 1.3130747262281353e-05, "loss": 0.16979103088378905, "step": 83890 }, { "epoch": 0.4167370334681065, "grad_norm": 2.3476858139038086, "learning_rate": 1.3129219380525711e-05, "loss": 0.1643131732940674, "step": 83900 }, { "epoch": 0.41678670415147573, "grad_norm": 1.365119218826294, "learning_rate": 1.312769141778713e-05, "loss": 0.1644132614135742, "step": 83910 }, { "epoch": 0.416836374834845, "grad_norm": 0.63211989402771, "learning_rate": 1.3126163374105149e-05, "loss": 0.15072190761566162, "step": 83920 }, { "epoch": 0.41688604551821423, "grad_norm": 1.3994576930999756, "learning_rate": 1.3124635249519322e-05, "loss": 0.1656876564025879, "step": 83930 }, { "epoch": 0.4169357162015835, "grad_norm": 1.4063866138458252, "learning_rate": 1.3123107044069187e-05, "loss": 0.16485568284988403, "step": 83940 }, { "epoch": 0.41698538688495274, "grad_norm": 1.1360491514205933, "learning_rate": 1.31215787577943e-05, "loss": 0.16913166046142578, "step": 83950 }, { "epoch": 0.41703505756832204, "grad_norm": 1.137751817703247, "learning_rate": 1.3120050390734206e-05, "loss": 0.16636180877685547, "step": 83960 }, { "epoch": 0.4170847282516913, "grad_norm": 1.212170958518982, "learning_rate": 1.3118521942928467e-05, "loss": 0.16054645776748658, "step": 83970 }, { "epoch": 0.41713439893506055, "grad_norm": 1.3275362253189087, "learning_rate": 1.311699341441663e-05, "loss": 0.172841477394104, "step": 83980 }, { "epoch": 0.4171840696184298, "grad_norm": 1.556141972541809, "learning_rate": 1.3115464805238261e-05, "loss": 0.16556978225708008, "step": 83990 }, { "epoch": 0.41723374030179905, "grad_norm": 1.3146119117736816, "learning_rate": 1.3113936115432913e-05, "loss": 0.18468408584594725, "step": 84000 }, { "epoch": 0.41728341098516836, "grad_norm": 0.7971798777580261, "learning_rate": 1.3112407345040151e-05, "loss": 0.1657571792602539, "step": 84010 }, { "epoch": 0.4173330816685376, "grad_norm": 2.1761562824249268, "learning_rate": 1.3110878494099538e-05, "loss": 0.16631933450698852, "step": 84020 }, { "epoch": 0.41738275235190686, "grad_norm": 0.9271419644355774, "learning_rate": 1.3109349562650637e-05, "loss": 0.17779102325439453, "step": 84030 }, { "epoch": 0.4174324230352761, "grad_norm": 1.9117616415023804, "learning_rate": 1.3107820550733026e-05, "loss": 0.1724607229232788, "step": 84040 }, { "epoch": 0.41748209371864536, "grad_norm": 1.0076391696929932, "learning_rate": 1.3106291458386264e-05, "loss": 0.1765296697616577, "step": 84050 }, { "epoch": 0.4175317644020146, "grad_norm": 0.8709198236465454, "learning_rate": 1.310476228564993e-05, "loss": 0.17222929000854492, "step": 84060 }, { "epoch": 0.4175814350853839, "grad_norm": 1.58665132522583, "learning_rate": 1.3103233032563601e-05, "loss": 0.17101904153823852, "step": 84070 }, { "epoch": 0.4176311057687532, "grad_norm": 1.9382755756378174, "learning_rate": 1.3101703699166843e-05, "loss": 0.16240849494934081, "step": 84080 }, { "epoch": 0.4176807764521224, "grad_norm": 1.5625165700912476, "learning_rate": 1.3100174285499239e-05, "loss": 0.1610654592514038, "step": 84090 }, { "epoch": 0.4177304471354917, "grad_norm": 1.3397550582885742, "learning_rate": 1.3098644791600374e-05, "loss": 0.17453935146331787, "step": 84100 }, { "epoch": 0.41778011781886093, "grad_norm": 0.6701281070709229, "learning_rate": 1.3097115217509827e-05, "loss": 0.17406611442565917, "step": 84110 }, { "epoch": 0.41782978850223024, "grad_norm": 1.049941897392273, "learning_rate": 1.3095585563267184e-05, "loss": 0.15781359672546386, "step": 84120 }, { "epoch": 0.4178794591855995, "grad_norm": 2.0696589946746826, "learning_rate": 1.3094055828912032e-05, "loss": 0.1702536463737488, "step": 84130 }, { "epoch": 0.41792912986896874, "grad_norm": 0.6844988465309143, "learning_rate": 1.3092526014483957e-05, "loss": 0.16020066738128663, "step": 84140 }, { "epoch": 0.417978800552338, "grad_norm": 2.0087904930114746, "learning_rate": 1.3090996120022554e-05, "loss": 0.17906320095062256, "step": 84150 }, { "epoch": 0.41802847123570724, "grad_norm": 1.0709450244903564, "learning_rate": 1.3089466145567414e-05, "loss": 0.1682788372039795, "step": 84160 }, { "epoch": 0.4180781419190765, "grad_norm": 1.126190423965454, "learning_rate": 1.308793609115813e-05, "loss": 0.1788037657737732, "step": 84170 }, { "epoch": 0.4181278126024458, "grad_norm": 1.1367310285568237, "learning_rate": 1.3086405956834305e-05, "loss": 0.18438115119934081, "step": 84180 }, { "epoch": 0.41817748328581505, "grad_norm": 1.0702646970748901, "learning_rate": 1.3084875742635532e-05, "loss": 0.17743208408355712, "step": 84190 }, { "epoch": 0.4182271539691843, "grad_norm": 1.2912611961364746, "learning_rate": 1.3083345448601419e-05, "loss": 0.15891288518905639, "step": 84200 }, { "epoch": 0.41827682465255356, "grad_norm": 0.5690413117408752, "learning_rate": 1.3081815074771562e-05, "loss": 0.17345635890960692, "step": 84210 }, { "epoch": 0.4183264953359228, "grad_norm": 1.0780961513519287, "learning_rate": 1.3080284621185571e-05, "loss": 0.15644479990005494, "step": 84220 }, { "epoch": 0.4183761660192921, "grad_norm": 0.9573687314987183, "learning_rate": 1.3078754087883055e-05, "loss": 0.16759442090988158, "step": 84230 }, { "epoch": 0.41842583670266137, "grad_norm": 1.4110721349716187, "learning_rate": 1.3077223474903618e-05, "loss": 0.1587422251701355, "step": 84240 }, { "epoch": 0.4184755073860306, "grad_norm": 1.371006727218628, "learning_rate": 1.3075692782286874e-05, "loss": 0.18695701360702516, "step": 84250 }, { "epoch": 0.4185251780693999, "grad_norm": 0.6458331346511841, "learning_rate": 1.3074162010072442e-05, "loss": 0.15845397710800171, "step": 84260 }, { "epoch": 0.4185748487527691, "grad_norm": 1.8302253484725952, "learning_rate": 1.3072631158299928e-05, "loss": 0.1714712858200073, "step": 84270 }, { "epoch": 0.4186245194361384, "grad_norm": 1.0755164623260498, "learning_rate": 1.307110022700896e-05, "loss": 0.17842605113983154, "step": 84280 }, { "epoch": 0.4186741901195077, "grad_norm": 0.9703105688095093, "learning_rate": 1.3069569216239146e-05, "loss": 0.17092976570129395, "step": 84290 }, { "epoch": 0.41872386080287693, "grad_norm": 1.3500093221664429, "learning_rate": 1.306803812603012e-05, "loss": 0.15458801984786988, "step": 84300 }, { "epoch": 0.4187735314862462, "grad_norm": 1.2666608095169067, "learning_rate": 1.30665069564215e-05, "loss": 0.17732901573181153, "step": 84310 }, { "epoch": 0.41882320216961544, "grad_norm": 1.5529810190200806, "learning_rate": 1.3064975707452913e-05, "loss": 0.17015433311462402, "step": 84320 }, { "epoch": 0.4188728728529847, "grad_norm": 1.180863618850708, "learning_rate": 1.306344437916399e-05, "loss": 0.17500536441802977, "step": 84330 }, { "epoch": 0.418922543536354, "grad_norm": 1.0468342304229736, "learning_rate": 1.3061912971594355e-05, "loss": 0.16978399753570556, "step": 84340 }, { "epoch": 0.41897221421972325, "grad_norm": 0.8935122489929199, "learning_rate": 1.3060381484783645e-05, "loss": 0.16414426565170287, "step": 84350 }, { "epoch": 0.4190218849030925, "grad_norm": 0.7199994325637817, "learning_rate": 1.3058849918771488e-05, "loss": 0.16845803260803222, "step": 84360 }, { "epoch": 0.41907155558646175, "grad_norm": 1.234505295753479, "learning_rate": 1.3057318273597531e-05, "loss": 0.15770130157470702, "step": 84370 }, { "epoch": 0.419121226269831, "grad_norm": 2.3946518898010254, "learning_rate": 1.3055786549301401e-05, "loss": 0.16501858234405517, "step": 84380 }, { "epoch": 0.41917089695320026, "grad_norm": 1.1564139127731323, "learning_rate": 1.305425474592275e-05, "loss": 0.1680340886116028, "step": 84390 }, { "epoch": 0.41922056763656956, "grad_norm": 1.9528818130493164, "learning_rate": 1.305272286350121e-05, "loss": 0.15669736862182618, "step": 84400 }, { "epoch": 0.4192702383199388, "grad_norm": 1.0883604288101196, "learning_rate": 1.3051190902076432e-05, "loss": 0.16276540756225585, "step": 84410 }, { "epoch": 0.41931990900330807, "grad_norm": 1.116721272468567, "learning_rate": 1.3049658861688057e-05, "loss": 0.1588514566421509, "step": 84420 }, { "epoch": 0.4193695796866773, "grad_norm": 0.9967860579490662, "learning_rate": 1.3048126742375739e-05, "loss": 0.16422197818756104, "step": 84430 }, { "epoch": 0.41941925037004657, "grad_norm": 2.0641446113586426, "learning_rate": 1.3046594544179127e-05, "loss": 0.15919873714447022, "step": 84440 }, { "epoch": 0.4194689210534159, "grad_norm": 1.2871851921081543, "learning_rate": 1.3045062267137874e-05, "loss": 0.17332131862640382, "step": 84450 }, { "epoch": 0.41951859173678513, "grad_norm": 0.8283445239067078, "learning_rate": 1.3043529911291632e-05, "loss": 0.16981861591339112, "step": 84460 }, { "epoch": 0.4195682624201544, "grad_norm": 1.604177713394165, "learning_rate": 1.304199747668006e-05, "loss": 0.17257697582244874, "step": 84470 }, { "epoch": 0.41961793310352363, "grad_norm": 0.9516016840934753, "learning_rate": 1.3040464963342819e-05, "loss": 0.15597684383392335, "step": 84480 }, { "epoch": 0.4196676037868929, "grad_norm": 0.9196249842643738, "learning_rate": 1.3038932371319563e-05, "loss": 0.17404043674468994, "step": 84490 }, { "epoch": 0.41971727447026214, "grad_norm": 1.0982863903045654, "learning_rate": 1.3037399700649959e-05, "loss": 0.16113849878311157, "step": 84500 }, { "epoch": 0.41976694515363144, "grad_norm": 0.6602425575256348, "learning_rate": 1.3035866951373675e-05, "loss": 0.17821104526519777, "step": 84510 }, { "epoch": 0.4198166158370007, "grad_norm": 1.5402976274490356, "learning_rate": 1.3034334123530375e-05, "loss": 0.16474857330322265, "step": 84520 }, { "epoch": 0.41986628652036995, "grad_norm": 1.501046061515808, "learning_rate": 1.3032801217159725e-05, "loss": 0.1710644006729126, "step": 84530 }, { "epoch": 0.4199159572037392, "grad_norm": 1.2125359773635864, "learning_rate": 1.3031268232301402e-05, "loss": 0.18453787565231322, "step": 84540 }, { "epoch": 0.41996562788710845, "grad_norm": 1.1041301488876343, "learning_rate": 1.3029735168995074e-05, "loss": 0.1635521411895752, "step": 84550 }, { "epoch": 0.42001529857047776, "grad_norm": 1.0519217252731323, "learning_rate": 1.3028202027280421e-05, "loss": 0.150518536567688, "step": 84560 }, { "epoch": 0.420064969253847, "grad_norm": 1.2797834873199463, "learning_rate": 1.3026668807197114e-05, "loss": 0.1719169497489929, "step": 84570 }, { "epoch": 0.42011463993721626, "grad_norm": 1.3659812211990356, "learning_rate": 1.3025135508784837e-05, "loss": 0.1813731908798218, "step": 84580 }, { "epoch": 0.4201643106205855, "grad_norm": 1.8812366724014282, "learning_rate": 1.3023602132083269e-05, "loss": 0.18005354404449464, "step": 84590 }, { "epoch": 0.42021398130395476, "grad_norm": 1.0856280326843262, "learning_rate": 1.3022068677132093e-05, "loss": 0.16326029300689698, "step": 84600 }, { "epoch": 0.420263651987324, "grad_norm": 1.6263517141342163, "learning_rate": 1.3020535143970996e-05, "loss": 0.17387151718139648, "step": 84610 }, { "epoch": 0.4203133226706933, "grad_norm": 0.8866721987724304, "learning_rate": 1.3019001532639663e-05, "loss": 0.16254739761352538, "step": 84620 }, { "epoch": 0.4203629933540626, "grad_norm": 1.11581289768219, "learning_rate": 1.301746784317778e-05, "loss": 0.1670796036720276, "step": 84630 }, { "epoch": 0.4204126640374318, "grad_norm": 0.6271365284919739, "learning_rate": 1.3015934075625048e-05, "loss": 0.15606920719146727, "step": 84640 }, { "epoch": 0.4204623347208011, "grad_norm": 2.0025739669799805, "learning_rate": 1.3014400230021153e-05, "loss": 0.15905534029006957, "step": 84650 }, { "epoch": 0.42051200540417033, "grad_norm": 1.469677448272705, "learning_rate": 1.3012866306405794e-05, "loss": 0.1753571391105652, "step": 84660 }, { "epoch": 0.42056167608753964, "grad_norm": 0.7146973013877869, "learning_rate": 1.3011332304818665e-05, "loss": 0.16065330505371095, "step": 84670 }, { "epoch": 0.4206113467709089, "grad_norm": 1.670721173286438, "learning_rate": 1.3009798225299466e-05, "loss": 0.17035336494445802, "step": 84680 }, { "epoch": 0.42066101745427814, "grad_norm": 1.1884562969207764, "learning_rate": 1.3008264067887902e-05, "loss": 0.17630140781402587, "step": 84690 }, { "epoch": 0.4207106881376474, "grad_norm": 1.7159764766693115, "learning_rate": 1.3006729832623668e-05, "loss": 0.1659805417060852, "step": 84700 }, { "epoch": 0.42076035882101664, "grad_norm": 1.5343143939971924, "learning_rate": 1.300519551954648e-05, "loss": 0.16074345111846924, "step": 84710 }, { "epoch": 0.4208100295043859, "grad_norm": 0.6914028525352478, "learning_rate": 1.3003661128696037e-05, "loss": 0.16105499267578124, "step": 84720 }, { "epoch": 0.4208597001877552, "grad_norm": 1.3388426303863525, "learning_rate": 1.3002126660112052e-05, "loss": 0.174229633808136, "step": 84730 }, { "epoch": 0.42090937087112446, "grad_norm": 1.1704596281051636, "learning_rate": 1.300059211383424e-05, "loss": 0.17432143688201904, "step": 84740 }, { "epoch": 0.4209590415544937, "grad_norm": 0.901565432548523, "learning_rate": 1.299905748990231e-05, "loss": 0.16696243286132811, "step": 84750 }, { "epoch": 0.42100871223786296, "grad_norm": 0.8006930947303772, "learning_rate": 1.2997522788355972e-05, "loss": 0.16918848752975463, "step": 84760 }, { "epoch": 0.4210583829212322, "grad_norm": 1.556147575378418, "learning_rate": 1.2995988009234954e-05, "loss": 0.15544309616088867, "step": 84770 }, { "epoch": 0.4211080536046015, "grad_norm": 0.7784901261329651, "learning_rate": 1.2994453152578973e-05, "loss": 0.18009454011917114, "step": 84780 }, { "epoch": 0.42115772428797077, "grad_norm": 0.853658139705658, "learning_rate": 1.2992918218427744e-05, "loss": 0.16151008605957032, "step": 84790 }, { "epoch": 0.42120739497134, "grad_norm": 1.4718948602676392, "learning_rate": 1.2991383206820998e-05, "loss": 0.16816962957382203, "step": 84800 }, { "epoch": 0.4212570656547093, "grad_norm": 0.9322205781936646, "learning_rate": 1.2989848117798456e-05, "loss": 0.17686998844146729, "step": 84810 }, { "epoch": 0.4213067363380785, "grad_norm": 1.3254520893096924, "learning_rate": 1.2988312951399846e-05, "loss": 0.16410439014434813, "step": 84820 }, { "epoch": 0.4213564070214478, "grad_norm": 1.2970497608184814, "learning_rate": 1.29867777076649e-05, "loss": 0.17135744094848632, "step": 84830 }, { "epoch": 0.4214060777048171, "grad_norm": 2.3676955699920654, "learning_rate": 1.2985242386633346e-05, "loss": 0.16202085018157958, "step": 84840 }, { "epoch": 0.42145574838818634, "grad_norm": 1.9727120399475098, "learning_rate": 1.298370698834492e-05, "loss": 0.1548928737640381, "step": 84850 }, { "epoch": 0.4215054190715556, "grad_norm": 0.9044303894042969, "learning_rate": 1.2982171512839359e-05, "loss": 0.1542547345161438, "step": 84860 }, { "epoch": 0.42155508975492484, "grad_norm": 1.0705140829086304, "learning_rate": 1.2980635960156396e-05, "loss": 0.16249847412109375, "step": 84870 }, { "epoch": 0.4216047604382941, "grad_norm": 1.4719549417495728, "learning_rate": 1.2979100330335774e-05, "loss": 0.16632649898529053, "step": 84880 }, { "epoch": 0.4216544311216634, "grad_norm": 0.7553927898406982, "learning_rate": 1.2977564623417234e-05, "loss": 0.17040048837661742, "step": 84890 }, { "epoch": 0.42170410180503265, "grad_norm": 2.649634838104248, "learning_rate": 1.2976028839440516e-05, "loss": 0.17419581413269042, "step": 84900 }, { "epoch": 0.4217537724884019, "grad_norm": 1.3137625455856323, "learning_rate": 1.297449297844537e-05, "loss": 0.17985069751739502, "step": 84910 }, { "epoch": 0.42180344317177115, "grad_norm": 1.5850019454956055, "learning_rate": 1.2972957040471543e-05, "loss": 0.17018613815307618, "step": 84920 }, { "epoch": 0.4218531138551404, "grad_norm": 1.2049421072006226, "learning_rate": 1.2971421025558782e-05, "loss": 0.16339802742004395, "step": 84930 }, { "epoch": 0.42190278453850966, "grad_norm": 1.337104320526123, "learning_rate": 1.296988493374684e-05, "loss": 0.16473610401153566, "step": 84940 }, { "epoch": 0.42195245522187896, "grad_norm": 1.199038028717041, "learning_rate": 1.296834876507547e-05, "loss": 0.1719701647758484, "step": 84950 }, { "epoch": 0.4220021259052482, "grad_norm": 1.2882390022277832, "learning_rate": 1.2966812519584423e-05, "loss": 0.1638707160949707, "step": 84960 }, { "epoch": 0.42205179658861747, "grad_norm": 2.295433521270752, "learning_rate": 1.2965276197313466e-05, "loss": 0.1589186191558838, "step": 84970 }, { "epoch": 0.4221014672719867, "grad_norm": 1.0752809047698975, "learning_rate": 1.2963739798302353e-05, "loss": 0.16786613464355468, "step": 84980 }, { "epoch": 0.42215113795535597, "grad_norm": 1.3853185176849365, "learning_rate": 1.2962203322590845e-05, "loss": 0.1715665340423584, "step": 84990 }, { "epoch": 0.4222008086387253, "grad_norm": 1.5717382431030273, "learning_rate": 1.2960666770218705e-05, "loss": 0.16597689390182496, "step": 85000 }, { "epoch": 0.42225047932209453, "grad_norm": 1.1962069272994995, "learning_rate": 1.29591301412257e-05, "loss": 0.17927103042602538, "step": 85010 }, { "epoch": 0.4223001500054638, "grad_norm": 1.3738009929656982, "learning_rate": 1.2957593435651598e-05, "loss": 0.1749461531639099, "step": 85020 }, { "epoch": 0.42234982068883303, "grad_norm": 1.0288541316986084, "learning_rate": 1.2956056653536167e-05, "loss": 0.1759092092514038, "step": 85030 }, { "epoch": 0.4223994913722023, "grad_norm": 0.9592280983924866, "learning_rate": 1.2954519794919175e-05, "loss": 0.16895697116851807, "step": 85040 }, { "epoch": 0.42244916205557154, "grad_norm": 0.9413324594497681, "learning_rate": 1.2952982859840402e-05, "loss": 0.1587531328201294, "step": 85050 }, { "epoch": 0.42249883273894084, "grad_norm": 0.6895803809165955, "learning_rate": 1.2951445848339618e-05, "loss": 0.16565299034118652, "step": 85060 }, { "epoch": 0.4225485034223101, "grad_norm": 1.836613416671753, "learning_rate": 1.2949908760456605e-05, "loss": 0.16931159496307374, "step": 85070 }, { "epoch": 0.42259817410567935, "grad_norm": 1.8948078155517578, "learning_rate": 1.294837159623114e-05, "loss": 0.17223799228668213, "step": 85080 }, { "epoch": 0.4226478447890486, "grad_norm": 0.7174872756004333, "learning_rate": 1.2946834355703e-05, "loss": 0.16275272369384766, "step": 85090 }, { "epoch": 0.42269751547241785, "grad_norm": 0.5963274836540222, "learning_rate": 1.2945297038911973e-05, "loss": 0.15380873680114746, "step": 85100 }, { "epoch": 0.42274718615578716, "grad_norm": 0.925713300704956, "learning_rate": 1.2943759645897843e-05, "loss": 0.16294691562652588, "step": 85110 }, { "epoch": 0.4227968568391564, "grad_norm": 1.3547357320785522, "learning_rate": 1.2942222176700397e-05, "loss": 0.14960894584655762, "step": 85120 }, { "epoch": 0.42284652752252566, "grad_norm": 0.8601296544075012, "learning_rate": 1.2940684631359423e-05, "loss": 0.17513681650161744, "step": 85130 }, { "epoch": 0.4228961982058949, "grad_norm": 3.023238182067871, "learning_rate": 1.2939147009914714e-05, "loss": 0.15986487865447999, "step": 85140 }, { "epoch": 0.42294586888926416, "grad_norm": 1.1289153099060059, "learning_rate": 1.2937609312406064e-05, "loss": 0.16635020971298217, "step": 85150 }, { "epoch": 0.4229955395726334, "grad_norm": 0.7932754755020142, "learning_rate": 1.2936071538873263e-05, "loss": 0.17165234088897705, "step": 85160 }, { "epoch": 0.4230452102560027, "grad_norm": 1.6382046937942505, "learning_rate": 1.2934533689356112e-05, "loss": 0.17131673097610473, "step": 85170 }, { "epoch": 0.423094880939372, "grad_norm": 1.1123772859573364, "learning_rate": 1.293299576389441e-05, "loss": 0.1562531113624573, "step": 85180 }, { "epoch": 0.4231445516227412, "grad_norm": 0.9295961856842041, "learning_rate": 1.2931457762527957e-05, "loss": 0.17750804424285888, "step": 85190 }, { "epoch": 0.4231942223061105, "grad_norm": 0.7075653672218323, "learning_rate": 1.2929919685296555e-05, "loss": 0.150117826461792, "step": 85200 }, { "epoch": 0.42324389298947973, "grad_norm": 1.5375269651412964, "learning_rate": 1.2928381532240011e-05, "loss": 0.1596813201904297, "step": 85210 }, { "epoch": 0.42329356367284904, "grad_norm": 1.146906852722168, "learning_rate": 1.2926843303398129e-05, "loss": 0.17891783714294435, "step": 85220 }, { "epoch": 0.4233432343562183, "grad_norm": 0.8720142841339111, "learning_rate": 1.292530499881072e-05, "loss": 0.1580110788345337, "step": 85230 }, { "epoch": 0.42339290503958754, "grad_norm": 0.9026685357093811, "learning_rate": 1.292376661851759e-05, "loss": 0.15528091192245483, "step": 85240 }, { "epoch": 0.4234425757229568, "grad_norm": 0.8624204993247986, "learning_rate": 1.2922228162558561e-05, "loss": 0.17335054874420167, "step": 85250 }, { "epoch": 0.42349224640632604, "grad_norm": 0.6820574402809143, "learning_rate": 1.2920689630973441e-05, "loss": 0.1687122702598572, "step": 85260 }, { "epoch": 0.4235419170896953, "grad_norm": 0.9379026889801025, "learning_rate": 1.2919151023802049e-05, "loss": 0.15921216011047362, "step": 85270 }, { "epoch": 0.4235915877730646, "grad_norm": 1.0024452209472656, "learning_rate": 1.2917612341084199e-05, "loss": 0.16567139625549315, "step": 85280 }, { "epoch": 0.42364125845643386, "grad_norm": 0.6829321384429932, "learning_rate": 1.2916073582859716e-05, "loss": 0.16342651844024658, "step": 85290 }, { "epoch": 0.4236909291398031, "grad_norm": 1.296078085899353, "learning_rate": 1.2914534749168417e-05, "loss": 0.16890296936035157, "step": 85300 }, { "epoch": 0.42374059982317236, "grad_norm": 0.9773350358009338, "learning_rate": 1.2912995840050136e-05, "loss": 0.17301748991012572, "step": 85310 }, { "epoch": 0.4237902705065416, "grad_norm": 1.535239338874817, "learning_rate": 1.2911456855544693e-05, "loss": 0.1653600215911865, "step": 85320 }, { "epoch": 0.4238399411899109, "grad_norm": 1.7923831939697266, "learning_rate": 1.2909917795691915e-05, "loss": 0.16586333513259888, "step": 85330 }, { "epoch": 0.42388961187328017, "grad_norm": 1.064186692237854, "learning_rate": 1.2908378660531637e-05, "loss": 0.15550171136856078, "step": 85340 }, { "epoch": 0.4239392825566494, "grad_norm": 1.3811155557632446, "learning_rate": 1.2906839450103687e-05, "loss": 0.18736619949340821, "step": 85350 }, { "epoch": 0.4239889532400187, "grad_norm": 2.569413185119629, "learning_rate": 1.2905300164447903e-05, "loss": 0.17328896522521972, "step": 85360 }, { "epoch": 0.4240386239233879, "grad_norm": 0.9773740768432617, "learning_rate": 1.2903760803604115e-05, "loss": 0.16885465383529663, "step": 85370 }, { "epoch": 0.4240882946067572, "grad_norm": 1.080436110496521, "learning_rate": 1.290222136761217e-05, "loss": 0.1631213903427124, "step": 85380 }, { "epoch": 0.4241379652901265, "grad_norm": 2.0397019386291504, "learning_rate": 1.2900681856511898e-05, "loss": 0.1692365050315857, "step": 85390 }, { "epoch": 0.42418763597349574, "grad_norm": 1.1151106357574463, "learning_rate": 1.289914227034315e-05, "loss": 0.1570284843444824, "step": 85400 }, { "epoch": 0.424237306656865, "grad_norm": 2.1487979888916016, "learning_rate": 1.2897602609145765e-05, "loss": 0.16939446926116944, "step": 85410 }, { "epoch": 0.42428697734023424, "grad_norm": 0.8400064706802368, "learning_rate": 1.289606287295959e-05, "loss": 0.16166987419128417, "step": 85420 }, { "epoch": 0.4243366480236035, "grad_norm": 0.9369289875030518, "learning_rate": 1.2894523061824468e-05, "loss": 0.15438787937164306, "step": 85430 }, { "epoch": 0.4243863187069728, "grad_norm": 0.8165996670722961, "learning_rate": 1.2892983175780258e-05, "loss": 0.17705570459365844, "step": 85440 }, { "epoch": 0.42443598939034205, "grad_norm": 1.6249653100967407, "learning_rate": 1.2891443214866803e-05, "loss": 0.17808246612548828, "step": 85450 }, { "epoch": 0.4244856600737113, "grad_norm": 1.5567703247070312, "learning_rate": 1.2889903179123961e-05, "loss": 0.17840360403060912, "step": 85460 }, { "epoch": 0.42453533075708055, "grad_norm": 1.5322121381759644, "learning_rate": 1.288836306859159e-05, "loss": 0.15941787958145143, "step": 85470 }, { "epoch": 0.4245850014404498, "grad_norm": 0.9843775033950806, "learning_rate": 1.288682288330954e-05, "loss": 0.17497825622558594, "step": 85480 }, { "epoch": 0.42463467212381906, "grad_norm": 0.8881663084030151, "learning_rate": 1.2885282623317676e-05, "loss": 0.15877792835235596, "step": 85490 }, { "epoch": 0.42468434280718836, "grad_norm": 0.7817594408988953, "learning_rate": 1.2883742288655859e-05, "loss": 0.15790145397186278, "step": 85500 }, { "epoch": 0.4247340134905576, "grad_norm": 1.5894789695739746, "learning_rate": 1.2882201879363949e-05, "loss": 0.16911559104919432, "step": 85510 }, { "epoch": 0.42478368417392687, "grad_norm": 0.7354007363319397, "learning_rate": 1.2880661395481813e-05, "loss": 0.17340505123138428, "step": 85520 }, { "epoch": 0.4248333548572961, "grad_norm": 1.4048151969909668, "learning_rate": 1.2879120837049319e-05, "loss": 0.1571420192718506, "step": 85530 }, { "epoch": 0.42488302554066537, "grad_norm": 1.2121267318725586, "learning_rate": 1.287758020410634e-05, "loss": 0.15426998138427733, "step": 85540 }, { "epoch": 0.4249326962240346, "grad_norm": 0.9729377627372742, "learning_rate": 1.2876039496692736e-05, "loss": 0.16675763130187987, "step": 85550 }, { "epoch": 0.42498236690740393, "grad_norm": 0.8391351103782654, "learning_rate": 1.2874498714848389e-05, "loss": 0.17347314357757568, "step": 85560 }, { "epoch": 0.4250320375907732, "grad_norm": 1.2876522541046143, "learning_rate": 1.2872957858613171e-05, "loss": 0.18352851867675782, "step": 85570 }, { "epoch": 0.42508170827414243, "grad_norm": 1.2528783082962036, "learning_rate": 1.2871416928026957e-05, "loss": 0.16303155422210694, "step": 85580 }, { "epoch": 0.4251313789575117, "grad_norm": 0.9321761131286621, "learning_rate": 1.2869875923129632e-05, "loss": 0.1590900182723999, "step": 85590 }, { "epoch": 0.42518104964088094, "grad_norm": 1.2871687412261963, "learning_rate": 1.2868334843961069e-05, "loss": 0.17082347869873046, "step": 85600 }, { "epoch": 0.42523072032425024, "grad_norm": 1.9115475416183472, "learning_rate": 1.2866793690561152e-05, "loss": 0.16781747341156006, "step": 85610 }, { "epoch": 0.4252803910076195, "grad_norm": 1.0975245237350464, "learning_rate": 1.2865252462969768e-05, "loss": 0.15582504272460937, "step": 85620 }, { "epoch": 0.42533006169098875, "grad_norm": 0.9660632610321045, "learning_rate": 1.2863711161226801e-05, "loss": 0.1715923309326172, "step": 85630 }, { "epoch": 0.425379732374358, "grad_norm": 1.5834076404571533, "learning_rate": 1.2862169785372142e-05, "loss": 0.1654037356376648, "step": 85640 }, { "epoch": 0.42542940305772725, "grad_norm": 1.2399297952651978, "learning_rate": 1.2860628335445683e-05, "loss": 0.18337662220001222, "step": 85650 }, { "epoch": 0.4254790737410965, "grad_norm": 1.3772460222244263, "learning_rate": 1.285908681148731e-05, "loss": 0.17060593366622925, "step": 85660 }, { "epoch": 0.4255287444244658, "grad_norm": 1.8288147449493408, "learning_rate": 1.285754521353692e-05, "loss": 0.1798194408416748, "step": 85670 }, { "epoch": 0.42557841510783506, "grad_norm": 0.972436785697937, "learning_rate": 1.2856003541634407e-05, "loss": 0.16933009624481202, "step": 85680 }, { "epoch": 0.4256280857912043, "grad_norm": 1.3557792901992798, "learning_rate": 1.2854461795819672e-05, "loss": 0.16980340480804443, "step": 85690 }, { "epoch": 0.42567775647457357, "grad_norm": 1.6687902212142944, "learning_rate": 1.2852919976132616e-05, "loss": 0.1640097141265869, "step": 85700 }, { "epoch": 0.4257274271579428, "grad_norm": 1.7480857372283936, "learning_rate": 1.2851378082613133e-05, "loss": 0.16154788732528685, "step": 85710 }, { "epoch": 0.4257770978413121, "grad_norm": 0.9811912178993225, "learning_rate": 1.2849836115301135e-05, "loss": 0.165887713432312, "step": 85720 }, { "epoch": 0.4258267685246814, "grad_norm": 1.575282335281372, "learning_rate": 1.2848294074236524e-05, "loss": 0.1562061309814453, "step": 85730 }, { "epoch": 0.42587643920805063, "grad_norm": 1.3021749258041382, "learning_rate": 1.284675195945921e-05, "loss": 0.17353293895721436, "step": 85740 }, { "epoch": 0.4259261098914199, "grad_norm": 0.8221110105514526, "learning_rate": 1.2845209771009097e-05, "loss": 0.1674646258354187, "step": 85750 }, { "epoch": 0.42597578057478913, "grad_norm": 1.5021240711212158, "learning_rate": 1.2843667508926099e-05, "loss": 0.1581561326980591, "step": 85760 }, { "epoch": 0.4260254512581584, "grad_norm": 0.9788241386413574, "learning_rate": 1.284212517325013e-05, "loss": 0.16374778747558594, "step": 85770 }, { "epoch": 0.4260751219415277, "grad_norm": 1.680566668510437, "learning_rate": 1.2840582764021106e-05, "loss": 0.1624821662902832, "step": 85780 }, { "epoch": 0.42612479262489694, "grad_norm": 1.1362221240997314, "learning_rate": 1.283904028127894e-05, "loss": 0.1650702953338623, "step": 85790 }, { "epoch": 0.4261744633082662, "grad_norm": 2.5044665336608887, "learning_rate": 1.2837497725063552e-05, "loss": 0.18183016777038574, "step": 85800 }, { "epoch": 0.42622413399163545, "grad_norm": 2.144519805908203, "learning_rate": 1.2835955095414867e-05, "loss": 0.16261558532714843, "step": 85810 }, { "epoch": 0.4262738046750047, "grad_norm": 1.10389244556427, "learning_rate": 1.2834412392372802e-05, "loss": 0.15948960781097413, "step": 85820 }, { "epoch": 0.426323475358374, "grad_norm": 0.7293723821640015, "learning_rate": 1.2832869615977286e-05, "loss": 0.15729715824127197, "step": 85830 }, { "epoch": 0.42637314604174326, "grad_norm": 1.1490892171859741, "learning_rate": 1.283132676626824e-05, "loss": 0.16934425830841066, "step": 85840 }, { "epoch": 0.4264228167251125, "grad_norm": 1.398982286453247, "learning_rate": 1.2829783843285598e-05, "loss": 0.1743498921394348, "step": 85850 }, { "epoch": 0.42647248740848176, "grad_norm": 1.2991050481796265, "learning_rate": 1.2828240847069286e-05, "loss": 0.1672147750854492, "step": 85860 }, { "epoch": 0.426522158091851, "grad_norm": 1.4035634994506836, "learning_rate": 1.282669777765924e-05, "loss": 0.15263220071792602, "step": 85870 }, { "epoch": 0.42657182877522026, "grad_norm": 1.8203073740005493, "learning_rate": 1.2825154635095393e-05, "loss": 0.16431574821472167, "step": 85880 }, { "epoch": 0.42662149945858957, "grad_norm": 1.5576260089874268, "learning_rate": 1.2823611419417678e-05, "loss": 0.1553809881210327, "step": 85890 }, { "epoch": 0.4266711701419588, "grad_norm": 1.3796592950820923, "learning_rate": 1.2822068130666036e-05, "loss": 0.16824078559875488, "step": 85900 }, { "epoch": 0.4267208408253281, "grad_norm": 2.127215623855591, "learning_rate": 1.2820524768880405e-05, "loss": 0.1636001229286194, "step": 85910 }, { "epoch": 0.4267705115086973, "grad_norm": 1.3032971620559692, "learning_rate": 1.2818981334100728e-05, "loss": 0.16866660118103027, "step": 85920 }, { "epoch": 0.4268201821920666, "grad_norm": 1.396614670753479, "learning_rate": 1.2817437826366947e-05, "loss": 0.1773458242416382, "step": 85930 }, { "epoch": 0.4268698528754359, "grad_norm": 1.6363799571990967, "learning_rate": 1.2815894245719007e-05, "loss": 0.16372413635253907, "step": 85940 }, { "epoch": 0.42691952355880514, "grad_norm": 1.0663189888000488, "learning_rate": 1.2814350592196858e-05, "loss": 0.16526271104812623, "step": 85950 }, { "epoch": 0.4269691942421744, "grad_norm": 1.8968764543533325, "learning_rate": 1.2812806865840447e-05, "loss": 0.17096840143203734, "step": 85960 }, { "epoch": 0.42701886492554364, "grad_norm": 1.6127753257751465, "learning_rate": 1.281126306668972e-05, "loss": 0.16329704523086547, "step": 85970 }, { "epoch": 0.4270685356089129, "grad_norm": 0.8410516381263733, "learning_rate": 1.2809719194784643e-05, "loss": 0.1703396201133728, "step": 85980 }, { "epoch": 0.42711820629228214, "grad_norm": 1.4892053604125977, "learning_rate": 1.2808175250165157e-05, "loss": 0.17148253917694092, "step": 85990 }, { "epoch": 0.42716787697565145, "grad_norm": 0.5761504173278809, "learning_rate": 1.280663123287123e-05, "loss": 0.16911520957946777, "step": 86000 }, { "epoch": 0.4272175476590207, "grad_norm": 2.0781188011169434, "learning_rate": 1.2805087142942812e-05, "loss": 0.16256844997406006, "step": 86010 }, { "epoch": 0.42726721834238995, "grad_norm": 0.7678692936897278, "learning_rate": 1.2803542980419866e-05, "loss": 0.1546323537826538, "step": 86020 }, { "epoch": 0.4273168890257592, "grad_norm": 2.760719060897827, "learning_rate": 1.2801998745342354e-05, "loss": 0.17105188369750976, "step": 86030 }, { "epoch": 0.42736655970912846, "grad_norm": 0.5747849941253662, "learning_rate": 1.2800454437750242e-05, "loss": 0.1591139554977417, "step": 86040 }, { "epoch": 0.42741623039249776, "grad_norm": 0.9040775299072266, "learning_rate": 1.2798910057683499e-05, "loss": 0.16635406017303467, "step": 86050 }, { "epoch": 0.427465901075867, "grad_norm": 0.755922794342041, "learning_rate": 1.2797365605182084e-05, "loss": 0.1630903124809265, "step": 86060 }, { "epoch": 0.42751557175923627, "grad_norm": 1.5304025411605835, "learning_rate": 1.2795821080285975e-05, "loss": 0.16652312278747558, "step": 86070 }, { "epoch": 0.4275652424426055, "grad_norm": 0.8412111401557922, "learning_rate": 1.2794276483035139e-05, "loss": 0.1605151414871216, "step": 86080 }, { "epoch": 0.42761491312597477, "grad_norm": 1.0927386283874512, "learning_rate": 1.2792731813469555e-05, "loss": 0.1733992576599121, "step": 86090 }, { "epoch": 0.427664583809344, "grad_norm": 1.492759108543396, "learning_rate": 1.279118707162919e-05, "loss": 0.1626041293144226, "step": 86100 }, { "epoch": 0.42771425449271333, "grad_norm": 1.9477559328079224, "learning_rate": 1.2789642257554027e-05, "loss": 0.16924779415130614, "step": 86110 }, { "epoch": 0.4277639251760826, "grad_norm": 2.2430200576782227, "learning_rate": 1.2788097371284045e-05, "loss": 0.1668605923652649, "step": 86120 }, { "epoch": 0.42781359585945183, "grad_norm": 1.1339184045791626, "learning_rate": 1.2786552412859223e-05, "loss": 0.16846283674240112, "step": 86130 }, { "epoch": 0.4278632665428211, "grad_norm": 1.5145153999328613, "learning_rate": 1.2785007382319545e-05, "loss": 0.16004045009613038, "step": 86140 }, { "epoch": 0.42791293722619034, "grad_norm": 1.6635686159133911, "learning_rate": 1.2783462279704999e-05, "loss": 0.16065701246261596, "step": 86150 }, { "epoch": 0.42796260790955964, "grad_norm": 0.813837468624115, "learning_rate": 1.2781917105055564e-05, "loss": 0.16254796981811523, "step": 86160 }, { "epoch": 0.4280122785929289, "grad_norm": 1.1946661472320557, "learning_rate": 1.2780371858411233e-05, "loss": 0.17098982334136964, "step": 86170 }, { "epoch": 0.42806194927629815, "grad_norm": 2.152056932449341, "learning_rate": 1.2778826539811996e-05, "loss": 0.1699947953224182, "step": 86180 }, { "epoch": 0.4281116199596674, "grad_norm": 1.7155883312225342, "learning_rate": 1.277728114929785e-05, "loss": 0.17448246479034424, "step": 86190 }, { "epoch": 0.42816129064303665, "grad_norm": 0.9576699733734131, "learning_rate": 1.2775735686908781e-05, "loss": 0.1664726734161377, "step": 86200 }, { "epoch": 0.4282109613264059, "grad_norm": 0.9310975074768066, "learning_rate": 1.2774190152684788e-05, "loss": 0.17364239692687988, "step": 86210 }, { "epoch": 0.4282606320097752, "grad_norm": 0.7375388741493225, "learning_rate": 1.2772644546665873e-05, "loss": 0.1606875777244568, "step": 86220 }, { "epoch": 0.42831030269314446, "grad_norm": 0.7838871479034424, "learning_rate": 1.2771098868892028e-05, "loss": 0.16151115894317628, "step": 86230 }, { "epoch": 0.4283599733765137, "grad_norm": 1.3986132144927979, "learning_rate": 1.2769553119403262e-05, "loss": 0.16627020835876466, "step": 86240 }, { "epoch": 0.42840964405988297, "grad_norm": 0.7240331768989563, "learning_rate": 1.276800729823957e-05, "loss": 0.15701594352722167, "step": 86250 }, { "epoch": 0.4284593147432522, "grad_norm": 0.9092841744422913, "learning_rate": 1.2766461405440969e-05, "loss": 0.15779945850372315, "step": 86260 }, { "epoch": 0.4285089854266215, "grad_norm": 1.4813048839569092, "learning_rate": 1.2764915441047453e-05, "loss": 0.15667152404785156, "step": 86270 }, { "epoch": 0.4285586561099908, "grad_norm": 1.322688102722168, "learning_rate": 1.276336940509904e-05, "loss": 0.1606661319732666, "step": 86280 }, { "epoch": 0.42860832679336003, "grad_norm": 2.2531793117523193, "learning_rate": 1.2761823297635738e-05, "loss": 0.1516035318374634, "step": 86290 }, { "epoch": 0.4286579974767293, "grad_norm": 0.9192180633544922, "learning_rate": 1.2760277118697557e-05, "loss": 0.14905209541320802, "step": 86300 }, { "epoch": 0.42870766816009853, "grad_norm": 0.957710862159729, "learning_rate": 1.2758730868324515e-05, "loss": 0.161243999004364, "step": 86310 }, { "epoch": 0.4287573388434678, "grad_norm": 1.6950175762176514, "learning_rate": 1.2757184546556629e-05, "loss": 0.1622737765312195, "step": 86320 }, { "epoch": 0.4288070095268371, "grad_norm": 1.846603512763977, "learning_rate": 1.2755638153433916e-05, "loss": 0.1583520531654358, "step": 86330 }, { "epoch": 0.42885668021020634, "grad_norm": 1.0338212251663208, "learning_rate": 1.2754091688996394e-05, "loss": 0.16101595163345336, "step": 86340 }, { "epoch": 0.4289063508935756, "grad_norm": 1.4601752758026123, "learning_rate": 1.2752545153284086e-05, "loss": 0.16364922523498535, "step": 86350 }, { "epoch": 0.42895602157694485, "grad_norm": 1.4392201900482178, "learning_rate": 1.2750998546337017e-05, "loss": 0.1649634599685669, "step": 86360 }, { "epoch": 0.4290056922603141, "grad_norm": 1.3514710664749146, "learning_rate": 1.2749451868195211e-05, "loss": 0.16263160705566407, "step": 86370 }, { "epoch": 0.4290553629436834, "grad_norm": 1.2066823244094849, "learning_rate": 1.2747905118898697e-05, "loss": 0.17046494483947755, "step": 86380 }, { "epoch": 0.42910503362705266, "grad_norm": 0.870020866394043, "learning_rate": 1.2746358298487503e-05, "loss": 0.16848326921463014, "step": 86390 }, { "epoch": 0.4291547043104219, "grad_norm": 0.8383588790893555, "learning_rate": 1.2744811407001661e-05, "loss": 0.1527864456176758, "step": 86400 }, { "epoch": 0.42920437499379116, "grad_norm": 0.8279457092285156, "learning_rate": 1.2743264444481202e-05, "loss": 0.16476659774780272, "step": 86410 }, { "epoch": 0.4292540456771604, "grad_norm": 1.0836902856826782, "learning_rate": 1.2741717410966163e-05, "loss": 0.160833477973938, "step": 86420 }, { "epoch": 0.42930371636052966, "grad_norm": 0.8881202340126038, "learning_rate": 1.2740170306496582e-05, "loss": 0.1794309139251709, "step": 86430 }, { "epoch": 0.42935338704389897, "grad_norm": 2.0699174404144287, "learning_rate": 1.2738623131112492e-05, "loss": 0.182423996925354, "step": 86440 }, { "epoch": 0.4294030577272682, "grad_norm": 0.901861310005188, "learning_rate": 1.2737075884853938e-05, "loss": 0.16661497354507446, "step": 86450 }, { "epoch": 0.4294527284106375, "grad_norm": 1.6136934757232666, "learning_rate": 1.2735528567760962e-05, "loss": 0.1702096939086914, "step": 86460 }, { "epoch": 0.4295023990940067, "grad_norm": 0.8500539064407349, "learning_rate": 1.2733981179873604e-05, "loss": 0.1750784397125244, "step": 86470 }, { "epoch": 0.429552069777376, "grad_norm": 1.0332270860671997, "learning_rate": 1.2732433721231914e-05, "loss": 0.17502286434173583, "step": 86480 }, { "epoch": 0.4296017404607453, "grad_norm": 0.9367872476577759, "learning_rate": 1.2730886191875938e-05, "loss": 0.1658119559288025, "step": 86490 }, { "epoch": 0.42965141114411454, "grad_norm": 1.2856906652450562, "learning_rate": 1.2729338591845726e-05, "loss": 0.16709475517272948, "step": 86500 }, { "epoch": 0.4297010818274838, "grad_norm": 0.8049591183662415, "learning_rate": 1.2727790921181325e-05, "loss": 0.17750555276870728, "step": 86510 }, { "epoch": 0.42975075251085304, "grad_norm": 0.959609866142273, "learning_rate": 1.2726243179922793e-05, "loss": 0.16205873489379882, "step": 86520 }, { "epoch": 0.4298004231942223, "grad_norm": 2.0888283252716064, "learning_rate": 1.2724695368110185e-05, "loss": 0.17721062898635864, "step": 86530 }, { "epoch": 0.42985009387759154, "grad_norm": 0.9659224152565002, "learning_rate": 1.2723147485783558e-05, "loss": 0.16237711906433105, "step": 86540 }, { "epoch": 0.42989976456096085, "grad_norm": 0.8365532755851746, "learning_rate": 1.2721599532982967e-05, "loss": 0.15764107704162597, "step": 86550 }, { "epoch": 0.4299494352443301, "grad_norm": 1.1063742637634277, "learning_rate": 1.2720051509748474e-05, "loss": 0.15759899616241455, "step": 86560 }, { "epoch": 0.42999910592769935, "grad_norm": 1.267602801322937, "learning_rate": 1.2718503416120143e-05, "loss": 0.1731116533279419, "step": 86570 }, { "epoch": 0.4300487766110686, "grad_norm": 1.3722888231277466, "learning_rate": 1.2716955252138034e-05, "loss": 0.17088708877563477, "step": 86580 }, { "epoch": 0.43009844729443786, "grad_norm": 1.1396478414535522, "learning_rate": 1.2715407017842218e-05, "loss": 0.15729581117630004, "step": 86590 }, { "epoch": 0.43014811797780717, "grad_norm": 0.8193960785865784, "learning_rate": 1.2713858713272762e-05, "loss": 0.16613459587097168, "step": 86600 }, { "epoch": 0.4301977886611764, "grad_norm": 0.8172623515129089, "learning_rate": 1.2712310338469732e-05, "loss": 0.16299207210540773, "step": 86610 }, { "epoch": 0.43024745934454567, "grad_norm": 0.9484906792640686, "learning_rate": 1.2710761893473201e-05, "loss": 0.1552128791809082, "step": 86620 }, { "epoch": 0.4302971300279149, "grad_norm": 0.8831996917724609, "learning_rate": 1.2709213378323242e-05, "loss": 0.17460103034973146, "step": 86630 }, { "epoch": 0.43034680071128417, "grad_norm": 1.3779958486557007, "learning_rate": 1.2707664793059928e-05, "loss": 0.1769250750541687, "step": 86640 }, { "epoch": 0.4303964713946534, "grad_norm": 0.6756758689880371, "learning_rate": 1.270611613772334e-05, "loss": 0.17231149673461915, "step": 86650 }, { "epoch": 0.43044614207802273, "grad_norm": 1.3220878839492798, "learning_rate": 1.2704567412353552e-05, "loss": 0.16721915006637572, "step": 86660 }, { "epoch": 0.430495812761392, "grad_norm": 0.8124064803123474, "learning_rate": 1.270301861699065e-05, "loss": 0.1608142375946045, "step": 86670 }, { "epoch": 0.43054548344476123, "grad_norm": 1.5032035112380981, "learning_rate": 1.2701469751674714e-05, "loss": 0.1675994873046875, "step": 86680 }, { "epoch": 0.4305951541281305, "grad_norm": 1.1213805675506592, "learning_rate": 1.2699920816445823e-05, "loss": 0.15611695051193236, "step": 86690 }, { "epoch": 0.43064482481149974, "grad_norm": 1.0231815576553345, "learning_rate": 1.269837181134407e-05, "loss": 0.1632075071334839, "step": 86700 }, { "epoch": 0.43069449549486905, "grad_norm": 0.9219984412193298, "learning_rate": 1.2696822736409535e-05, "loss": 0.1707409381866455, "step": 86710 }, { "epoch": 0.4307441661782383, "grad_norm": 0.6978250741958618, "learning_rate": 1.2695273591682315e-05, "loss": 0.17346696853637694, "step": 86720 }, { "epoch": 0.43079383686160755, "grad_norm": 0.8601102828979492, "learning_rate": 1.2693724377202498e-05, "loss": 0.16386916637420654, "step": 86730 }, { "epoch": 0.4308435075449768, "grad_norm": 0.8767032623291016, "learning_rate": 1.269217509301018e-05, "loss": 0.17743183374404908, "step": 86740 }, { "epoch": 0.43089317822834605, "grad_norm": 1.4575715065002441, "learning_rate": 1.269062573914545e-05, "loss": 0.16345808506011963, "step": 86750 }, { "epoch": 0.4309428489117153, "grad_norm": 2.501798629760742, "learning_rate": 1.268907631564841e-05, "loss": 0.1697974681854248, "step": 86760 }, { "epoch": 0.4309925195950846, "grad_norm": 1.282841444015503, "learning_rate": 1.2687526822559153e-05, "loss": 0.16856911182403564, "step": 86770 }, { "epoch": 0.43104219027845386, "grad_norm": 0.7263182997703552, "learning_rate": 1.2685977259917785e-05, "loss": 0.16161376237869263, "step": 86780 }, { "epoch": 0.4310918609618231, "grad_norm": 0.7256495356559753, "learning_rate": 1.2684427627764404e-05, "loss": 0.17916855812072754, "step": 86790 }, { "epoch": 0.43114153164519237, "grad_norm": 1.0597556829452515, "learning_rate": 1.2682877926139117e-05, "loss": 0.17238887548446655, "step": 86800 }, { "epoch": 0.4311912023285616, "grad_norm": 0.5521795749664307, "learning_rate": 1.2681328155082027e-05, "loss": 0.17341537475585939, "step": 86810 }, { "epoch": 0.4312408730119309, "grad_norm": 2.2712783813476562, "learning_rate": 1.2679778314633243e-05, "loss": 0.15086143016815184, "step": 86820 }, { "epoch": 0.4312905436953002, "grad_norm": 0.8318865895271301, "learning_rate": 1.2678228404832873e-05, "loss": 0.1708857536315918, "step": 86830 }, { "epoch": 0.43134021437866943, "grad_norm": 0.7439736127853394, "learning_rate": 1.2676678425721028e-05, "loss": 0.1739768624305725, "step": 86840 }, { "epoch": 0.4313898850620387, "grad_norm": 1.3959189653396606, "learning_rate": 1.267512837733782e-05, "loss": 0.17201461791992187, "step": 86850 }, { "epoch": 0.43143955574540793, "grad_norm": 1.1758553981781006, "learning_rate": 1.267357825972337e-05, "loss": 0.15761585235595704, "step": 86860 }, { "epoch": 0.4314892264287772, "grad_norm": 1.116860032081604, "learning_rate": 1.2672028072917787e-05, "loss": 0.17603003978729248, "step": 86870 }, { "epoch": 0.4315388971121465, "grad_norm": 0.6977918744087219, "learning_rate": 1.2670477816961191e-05, "loss": 0.17415313720703124, "step": 86880 }, { "epoch": 0.43158856779551574, "grad_norm": 1.3495444059371948, "learning_rate": 1.2668927491893703e-05, "loss": 0.16253650188446045, "step": 86890 }, { "epoch": 0.431638238478885, "grad_norm": 1.3477345705032349, "learning_rate": 1.2667377097755446e-05, "loss": 0.16189332008361818, "step": 86900 }, { "epoch": 0.43168790916225425, "grad_norm": 0.7498250603675842, "learning_rate": 1.2665826634586543e-05, "loss": 0.16498314142227172, "step": 86910 }, { "epoch": 0.4317375798456235, "grad_norm": 0.989766001701355, "learning_rate": 1.2664276102427116e-05, "loss": 0.17655038833618164, "step": 86920 }, { "epoch": 0.4317872505289928, "grad_norm": 1.5800000429153442, "learning_rate": 1.2662725501317295e-05, "loss": 0.15724592208862304, "step": 86930 }, { "epoch": 0.43183692121236206, "grad_norm": 0.7775421142578125, "learning_rate": 1.2661174831297212e-05, "loss": 0.1663118362426758, "step": 86940 }, { "epoch": 0.4318865918957313, "grad_norm": 0.8740673065185547, "learning_rate": 1.2659624092406992e-05, "loss": 0.1646839737892151, "step": 86950 }, { "epoch": 0.43193626257910056, "grad_norm": 0.9225558638572693, "learning_rate": 1.2658073284686772e-05, "loss": 0.16293140649795532, "step": 86960 }, { "epoch": 0.4319859332624698, "grad_norm": 0.9587259292602539, "learning_rate": 1.2656522408176682e-05, "loss": 0.1824970245361328, "step": 86970 }, { "epoch": 0.43203560394583906, "grad_norm": 0.6483235955238342, "learning_rate": 1.2654971462916859e-05, "loss": 0.159745454788208, "step": 86980 }, { "epoch": 0.43208527462920837, "grad_norm": 0.8177329301834106, "learning_rate": 1.2653420448947441e-05, "loss": 0.1677854061126709, "step": 86990 }, { "epoch": 0.4321349453125776, "grad_norm": 1.1803607940673828, "learning_rate": 1.2651869366308573e-05, "loss": 0.16462488174438478, "step": 87000 }, { "epoch": 0.4321846159959469, "grad_norm": 1.1897146701812744, "learning_rate": 1.2650318215040392e-05, "loss": 0.1654098391532898, "step": 87010 }, { "epoch": 0.4322342866793161, "grad_norm": 0.8594350218772888, "learning_rate": 1.264876699518304e-05, "loss": 0.16855565309524537, "step": 87020 }, { "epoch": 0.4322839573626854, "grad_norm": 1.066051959991455, "learning_rate": 1.264721570677666e-05, "loss": 0.17313820123672485, "step": 87030 }, { "epoch": 0.43233362804605463, "grad_norm": 2.9063191413879395, "learning_rate": 1.2645664349861405e-05, "loss": 0.17614305019378662, "step": 87040 }, { "epoch": 0.43238329872942394, "grad_norm": 0.8623315691947937, "learning_rate": 1.2644112924477418e-05, "loss": 0.17445180416107178, "step": 87050 }, { "epoch": 0.4324329694127932, "grad_norm": 1.0115361213684082, "learning_rate": 1.2642561430664856e-05, "loss": 0.16000884771347046, "step": 87060 }, { "epoch": 0.43248264009616244, "grad_norm": 1.0542711019515991, "learning_rate": 1.2641009868463863e-05, "loss": 0.16684260368347167, "step": 87070 }, { "epoch": 0.4325323107795317, "grad_norm": 0.9891615509986877, "learning_rate": 1.2639458237914597e-05, "loss": 0.16286616325378417, "step": 87080 }, { "epoch": 0.43258198146290094, "grad_norm": 1.488620400428772, "learning_rate": 1.2637906539057214e-05, "loss": 0.15841164588928222, "step": 87090 }, { "epoch": 0.43263165214627025, "grad_norm": 1.7948806285858154, "learning_rate": 1.2636354771931871e-05, "loss": 0.1748773217201233, "step": 87100 }, { "epoch": 0.4326813228296395, "grad_norm": 0.9931033849716187, "learning_rate": 1.2634802936578728e-05, "loss": 0.16868300437927247, "step": 87110 }, { "epoch": 0.43273099351300875, "grad_norm": 1.5999835729599, "learning_rate": 1.2633251033037942e-05, "loss": 0.15348856449127196, "step": 87120 }, { "epoch": 0.432780664196378, "grad_norm": 0.9853799939155579, "learning_rate": 1.2631699061349678e-05, "loss": 0.17549524307250977, "step": 87130 }, { "epoch": 0.43283033487974726, "grad_norm": 1.4466732740402222, "learning_rate": 1.26301470215541e-05, "loss": 0.16878869533538818, "step": 87140 }, { "epoch": 0.4328800055631165, "grad_norm": 1.1434906721115112, "learning_rate": 1.2628594913691377e-05, "loss": 0.16535191535949706, "step": 87150 }, { "epoch": 0.4329296762464858, "grad_norm": 1.0112451314926147, "learning_rate": 1.2627042737801673e-05, "loss": 0.15649499893188476, "step": 87160 }, { "epoch": 0.43297934692985507, "grad_norm": 1.05720853805542, "learning_rate": 1.2625490493925158e-05, "loss": 0.16152050495147705, "step": 87170 }, { "epoch": 0.4330290176132243, "grad_norm": 1.4338023662567139, "learning_rate": 1.2623938182102004e-05, "loss": 0.16000932455062866, "step": 87180 }, { "epoch": 0.4330786882965936, "grad_norm": 1.0361554622650146, "learning_rate": 1.2622385802372388e-05, "loss": 0.17784881591796875, "step": 87190 }, { "epoch": 0.4331283589799628, "grad_norm": 1.076040267944336, "learning_rate": 1.262083335477648e-05, "loss": 0.16817270517349242, "step": 87200 }, { "epoch": 0.43317802966333213, "grad_norm": 1.7923725843429565, "learning_rate": 1.261928083935446e-05, "loss": 0.15360589027404786, "step": 87210 }, { "epoch": 0.4332277003467014, "grad_norm": 1.558898687362671, "learning_rate": 1.2617728256146507e-05, "loss": 0.16883816719055175, "step": 87220 }, { "epoch": 0.43327737103007063, "grad_norm": 1.5990906953811646, "learning_rate": 1.2616175605192796e-05, "loss": 0.16413991451263427, "step": 87230 }, { "epoch": 0.4333270417134399, "grad_norm": 1.085421085357666, "learning_rate": 1.2614622886533512e-05, "loss": 0.15964257717132568, "step": 87240 }, { "epoch": 0.43337671239680914, "grad_norm": 0.9679504632949829, "learning_rate": 1.2613070100208837e-05, "loss": 0.15974268913269044, "step": 87250 }, { "epoch": 0.4334263830801784, "grad_norm": 0.9166457653045654, "learning_rate": 1.2611517246258962e-05, "loss": 0.1757015109062195, "step": 87260 }, { "epoch": 0.4334760537635477, "grad_norm": 0.6281803846359253, "learning_rate": 1.2609964324724071e-05, "loss": 0.1633694887161255, "step": 87270 }, { "epoch": 0.43352572444691695, "grad_norm": 1.7971253395080566, "learning_rate": 1.2608411335644352e-05, "loss": 0.1671047568321228, "step": 87280 }, { "epoch": 0.4335753951302862, "grad_norm": 1.460990071296692, "learning_rate": 1.2606858279059997e-05, "loss": 0.17140071392059325, "step": 87290 }, { "epoch": 0.43362506581365545, "grad_norm": 1.3421745300292969, "learning_rate": 1.2605305155011198e-05, "loss": 0.1674057960510254, "step": 87300 }, { "epoch": 0.4336747364970247, "grad_norm": 1.0796208381652832, "learning_rate": 1.2603751963538147e-05, "loss": 0.17816669940948487, "step": 87310 }, { "epoch": 0.433724407180394, "grad_norm": 0.9657697081565857, "learning_rate": 1.2602198704681041e-05, "loss": 0.16284401416778566, "step": 87320 }, { "epoch": 0.43377407786376326, "grad_norm": 0.6907281875610352, "learning_rate": 1.2600645378480083e-05, "loss": 0.1697464942932129, "step": 87330 }, { "epoch": 0.4338237485471325, "grad_norm": 0.9173533320426941, "learning_rate": 1.2599091984975466e-05, "loss": 0.15667619705200195, "step": 87340 }, { "epoch": 0.43387341923050177, "grad_norm": 1.686653733253479, "learning_rate": 1.2597538524207393e-05, "loss": 0.16244809627532958, "step": 87350 }, { "epoch": 0.433923089913871, "grad_norm": 0.8230006098747253, "learning_rate": 1.259598499621607e-05, "loss": 0.1550676107406616, "step": 87360 }, { "epoch": 0.43397276059724027, "grad_norm": 0.7715627551078796, "learning_rate": 1.2594431401041699e-05, "loss": 0.15737133026123046, "step": 87370 }, { "epoch": 0.4340224312806096, "grad_norm": 1.439944863319397, "learning_rate": 1.2592877738724486e-05, "loss": 0.1514916777610779, "step": 87380 }, { "epoch": 0.43407210196397883, "grad_norm": 2.2518718242645264, "learning_rate": 1.2591324009304633e-05, "loss": 0.17075579166412352, "step": 87390 }, { "epoch": 0.4341217726473481, "grad_norm": 1.06404709815979, "learning_rate": 1.2589770212822363e-05, "loss": 0.16704797744750977, "step": 87400 }, { "epoch": 0.43417144333071733, "grad_norm": 1.2767400741577148, "learning_rate": 1.2588216349317881e-05, "loss": 0.18437464237213136, "step": 87410 }, { "epoch": 0.4342211140140866, "grad_norm": 2.079275131225586, "learning_rate": 1.2586662418831401e-05, "loss": 0.1535440444946289, "step": 87420 }, { "epoch": 0.4342707846974559, "grad_norm": 1.7343559265136719, "learning_rate": 1.2585108421403136e-05, "loss": 0.1710248589515686, "step": 87430 }, { "epoch": 0.43432045538082514, "grad_norm": 1.0973442792892456, "learning_rate": 1.2583554357073307e-05, "loss": 0.1664925456047058, "step": 87440 }, { "epoch": 0.4343701260641944, "grad_norm": 0.8038709759712219, "learning_rate": 1.2582000225882127e-05, "loss": 0.16235883235931398, "step": 87450 }, { "epoch": 0.43441979674756365, "grad_norm": 0.870613694190979, "learning_rate": 1.2580446027869822e-05, "loss": 0.17223241329193115, "step": 87460 }, { "epoch": 0.4344694674309329, "grad_norm": 1.239069938659668, "learning_rate": 1.2578891763076608e-05, "loss": 0.15537605285644532, "step": 87470 }, { "epoch": 0.43451913811430215, "grad_norm": 1.0616731643676758, "learning_rate": 1.2577337431542716e-05, "loss": 0.14818253517150878, "step": 87480 }, { "epoch": 0.43456880879767146, "grad_norm": 0.7884083390235901, "learning_rate": 1.2575783033308364e-05, "loss": 0.1533258557319641, "step": 87490 }, { "epoch": 0.4346184794810407, "grad_norm": 1.28818941116333, "learning_rate": 1.2574228568413784e-05, "loss": 0.1685653567314148, "step": 87500 }, { "epoch": 0.43466815016440996, "grad_norm": 1.4190701246261597, "learning_rate": 1.2572674036899204e-05, "loss": 0.15336003303527831, "step": 87510 }, { "epoch": 0.4347178208477792, "grad_norm": 0.9515408277511597, "learning_rate": 1.2571119438804853e-05, "loss": 0.1663593053817749, "step": 87520 }, { "epoch": 0.43476749153114846, "grad_norm": 1.6905254125595093, "learning_rate": 1.2569564774170965e-05, "loss": 0.18025095462799073, "step": 87530 }, { "epoch": 0.43481716221451777, "grad_norm": 1.3851618766784668, "learning_rate": 1.2568010043037778e-05, "loss": 0.16329212188720704, "step": 87540 }, { "epoch": 0.434866832897887, "grad_norm": 1.2694939374923706, "learning_rate": 1.2566455245445519e-05, "loss": 0.1647486209869385, "step": 87550 }, { "epoch": 0.4349165035812563, "grad_norm": 1.7040348052978516, "learning_rate": 1.256490038143443e-05, "loss": 0.1729472279548645, "step": 87560 }, { "epoch": 0.4349661742646255, "grad_norm": 3.1391663551330566, "learning_rate": 1.2563345451044753e-05, "loss": 0.16142752170562744, "step": 87570 }, { "epoch": 0.4350158449479948, "grad_norm": 2.053574323654175, "learning_rate": 1.2561790454316726e-05, "loss": 0.1592962384223938, "step": 87580 }, { "epoch": 0.43506551563136403, "grad_norm": 1.0156488418579102, "learning_rate": 1.256023539129059e-05, "loss": 0.1725975751876831, "step": 87590 }, { "epoch": 0.43511518631473334, "grad_norm": 0.6842926144599915, "learning_rate": 1.2558680262006594e-05, "loss": 0.1541095495223999, "step": 87600 }, { "epoch": 0.4351648569981026, "grad_norm": 1.3794305324554443, "learning_rate": 1.255712506650498e-05, "loss": 0.1799790859222412, "step": 87610 }, { "epoch": 0.43521452768147184, "grad_norm": 1.5145105123519897, "learning_rate": 1.2555569804826e-05, "loss": 0.1751146674156189, "step": 87620 }, { "epoch": 0.4352641983648411, "grad_norm": 0.6706549525260925, "learning_rate": 1.25540144770099e-05, "loss": 0.16300954818725585, "step": 87630 }, { "epoch": 0.43531386904821034, "grad_norm": 1.3172458410263062, "learning_rate": 1.2552459083096932e-05, "loss": 0.16733750104904174, "step": 87640 }, { "epoch": 0.43536353973157965, "grad_norm": 1.512681245803833, "learning_rate": 1.2550903623127347e-05, "loss": 0.1756464958190918, "step": 87650 }, { "epoch": 0.4354132104149489, "grad_norm": 0.9993164539337158, "learning_rate": 1.2549348097141403e-05, "loss": 0.1637190818786621, "step": 87660 }, { "epoch": 0.43546288109831816, "grad_norm": 1.16487717628479, "learning_rate": 1.2547792505179356e-05, "loss": 0.1581324338912964, "step": 87670 }, { "epoch": 0.4355125517816874, "grad_norm": 1.1341030597686768, "learning_rate": 1.2546236847281464e-05, "loss": 0.1618643283843994, "step": 87680 }, { "epoch": 0.43556222246505666, "grad_norm": 1.2300407886505127, "learning_rate": 1.2544681123487985e-05, "loss": 0.18169660568237306, "step": 87690 }, { "epoch": 0.4356118931484259, "grad_norm": 1.1584131717681885, "learning_rate": 1.2543125333839181e-05, "loss": 0.1585907220840454, "step": 87700 }, { "epoch": 0.4356615638317952, "grad_norm": 0.6684743762016296, "learning_rate": 1.254156947837532e-05, "loss": 0.15171746015548707, "step": 87710 }, { "epoch": 0.43571123451516447, "grad_norm": 1.4459282159805298, "learning_rate": 1.2540013557136657e-05, "loss": 0.15546952486038207, "step": 87720 }, { "epoch": 0.4357609051985337, "grad_norm": 1.2239933013916016, "learning_rate": 1.2538457570163467e-05, "loss": 0.16836742162704468, "step": 87730 }, { "epoch": 0.435810575881903, "grad_norm": 1.6044520139694214, "learning_rate": 1.2536901517496017e-05, "loss": 0.1686995506286621, "step": 87740 }, { "epoch": 0.4358602465652722, "grad_norm": 2.2725136280059814, "learning_rate": 1.2535345399174574e-05, "loss": 0.17817111015319825, "step": 87750 }, { "epoch": 0.43590991724864153, "grad_norm": 0.6610761284828186, "learning_rate": 1.2533789215239414e-05, "loss": 0.16368522644042968, "step": 87760 }, { "epoch": 0.4359595879320108, "grad_norm": 1.0683555603027344, "learning_rate": 1.2532232965730806e-05, "loss": 0.16748275756835937, "step": 87770 }, { "epoch": 0.43600925861538004, "grad_norm": 0.6000364422798157, "learning_rate": 1.2530676650689026e-05, "loss": 0.1675087571144104, "step": 87780 }, { "epoch": 0.4360589292987493, "grad_norm": 1.778615951538086, "learning_rate": 1.2529120270154353e-05, "loss": 0.17865166664123536, "step": 87790 }, { "epoch": 0.43610859998211854, "grad_norm": 0.987834632396698, "learning_rate": 1.2527563824167066e-05, "loss": 0.1636784315109253, "step": 87800 }, { "epoch": 0.4361582706654878, "grad_norm": 1.7410794496536255, "learning_rate": 1.2526007312767443e-05, "loss": 0.15804189443588257, "step": 87810 }, { "epoch": 0.4362079413488571, "grad_norm": 1.7612760066986084, "learning_rate": 1.2524450735995766e-05, "loss": 0.16306988000869752, "step": 87820 }, { "epoch": 0.43625761203222635, "grad_norm": 1.1799763441085815, "learning_rate": 1.2522894093892317e-05, "loss": 0.16133639812469483, "step": 87830 }, { "epoch": 0.4363072827155956, "grad_norm": 1.103027582168579, "learning_rate": 1.2521337386497386e-05, "loss": 0.17590618133544922, "step": 87840 }, { "epoch": 0.43635695339896485, "grad_norm": 1.089741587638855, "learning_rate": 1.2519780613851254e-05, "loss": 0.15244364738464355, "step": 87850 }, { "epoch": 0.4364066240823341, "grad_norm": 1.9071345329284668, "learning_rate": 1.2518223775994214e-05, "loss": 0.1583620071411133, "step": 87860 }, { "epoch": 0.4364562947657034, "grad_norm": 0.7190669775009155, "learning_rate": 1.2516666872966557e-05, "loss": 0.16625722646713256, "step": 87870 }, { "epoch": 0.43650596544907266, "grad_norm": 0.9362053275108337, "learning_rate": 1.2515109904808571e-05, "loss": 0.16520588397979735, "step": 87880 }, { "epoch": 0.4365556361324419, "grad_norm": 1.7997242212295532, "learning_rate": 1.2513552871560552e-05, "loss": 0.17865962982177735, "step": 87890 }, { "epoch": 0.43660530681581117, "grad_norm": 1.478065848350525, "learning_rate": 1.2511995773262798e-05, "loss": 0.1707468271255493, "step": 87900 }, { "epoch": 0.4366549774991804, "grad_norm": 0.8566610217094421, "learning_rate": 1.2510438609955602e-05, "loss": 0.17670793533325196, "step": 87910 }, { "epoch": 0.43670464818254967, "grad_norm": 1.8873473405838013, "learning_rate": 1.2508881381679263e-05, "loss": 0.16885944604873657, "step": 87920 }, { "epoch": 0.436754318865919, "grad_norm": 1.1237223148345947, "learning_rate": 1.2507324088474082e-05, "loss": 0.1511697292327881, "step": 87930 }, { "epoch": 0.43680398954928823, "grad_norm": 0.6941211223602295, "learning_rate": 1.2505766730380364e-05, "loss": 0.15838420391082764, "step": 87940 }, { "epoch": 0.4368536602326575, "grad_norm": 1.132338285446167, "learning_rate": 1.2504209307438414e-05, "loss": 0.17223737239837647, "step": 87950 }, { "epoch": 0.43690333091602673, "grad_norm": 1.6238337755203247, "learning_rate": 1.2502651819688529e-05, "loss": 0.15346217155456543, "step": 87960 }, { "epoch": 0.436953001599396, "grad_norm": 1.2814395427703857, "learning_rate": 1.2501094267171024e-05, "loss": 0.16803539991378785, "step": 87970 }, { "epoch": 0.4370026722827653, "grad_norm": 1.1252154111862183, "learning_rate": 1.2499536649926198e-05, "loss": 0.16467792987823487, "step": 87980 }, { "epoch": 0.43705234296613454, "grad_norm": 1.4474064111709595, "learning_rate": 1.2497978967994376e-05, "loss": 0.1690801501274109, "step": 87990 }, { "epoch": 0.4371020136495038, "grad_norm": 1.1215898990631104, "learning_rate": 1.249642122141586e-05, "loss": 0.15557665824890138, "step": 88000 }, { "epoch": 0.43715168433287305, "grad_norm": 0.7211962342262268, "learning_rate": 1.2494863410230967e-05, "loss": 0.15714881420135499, "step": 88010 }, { "epoch": 0.4372013550162423, "grad_norm": 1.5157356262207031, "learning_rate": 1.2493305534480012e-05, "loss": 0.15963436365127565, "step": 88020 }, { "epoch": 0.43725102569961155, "grad_norm": 0.6444419622421265, "learning_rate": 1.2491747594203312e-05, "loss": 0.17234667539596557, "step": 88030 }, { "epoch": 0.43730069638298086, "grad_norm": 0.5915625691413879, "learning_rate": 1.2490189589441188e-05, "loss": 0.16145906448364258, "step": 88040 }, { "epoch": 0.4373503670663501, "grad_norm": 1.2931891679763794, "learning_rate": 1.2488631520233956e-05, "loss": 0.1698094129562378, "step": 88050 }, { "epoch": 0.43740003774971936, "grad_norm": 1.5742368698120117, "learning_rate": 1.248707338662194e-05, "loss": 0.16856579780578612, "step": 88060 }, { "epoch": 0.4374497084330886, "grad_norm": 0.9286395907402039, "learning_rate": 1.2485515188645468e-05, "loss": 0.1605035185813904, "step": 88070 }, { "epoch": 0.43749937911645786, "grad_norm": 1.1837433576583862, "learning_rate": 1.248395692634486e-05, "loss": 0.16732563972473144, "step": 88080 }, { "epoch": 0.43754904979982717, "grad_norm": 2.6283271312713623, "learning_rate": 1.2482398599760447e-05, "loss": 0.15755281448364258, "step": 88090 }, { "epoch": 0.4375987204831964, "grad_norm": 1.7501009702682495, "learning_rate": 1.2480840208932557e-05, "loss": 0.16930992603302003, "step": 88100 }, { "epoch": 0.4376483911665657, "grad_norm": 1.3788002729415894, "learning_rate": 1.247928175390152e-05, "loss": 0.15538593530654907, "step": 88110 }, { "epoch": 0.4376980618499349, "grad_norm": 0.8017485737800598, "learning_rate": 1.2477723234707667e-05, "loss": 0.1674445629119873, "step": 88120 }, { "epoch": 0.4377477325333042, "grad_norm": 0.8809719085693359, "learning_rate": 1.247616465139133e-05, "loss": 0.16448110342025757, "step": 88130 }, { "epoch": 0.43779740321667343, "grad_norm": 0.8467004299163818, "learning_rate": 1.2474606003992853e-05, "loss": 0.17004749774932862, "step": 88140 }, { "epoch": 0.43784707390004274, "grad_norm": 1.2162235975265503, "learning_rate": 1.2473047292552565e-05, "loss": 0.1827920436859131, "step": 88150 }, { "epoch": 0.437896744583412, "grad_norm": 1.8562352657318115, "learning_rate": 1.2471488517110809e-05, "loss": 0.16515800952911378, "step": 88160 }, { "epoch": 0.43794641526678124, "grad_norm": 1.2016092538833618, "learning_rate": 1.2469929677707922e-05, "loss": 0.17134199142456055, "step": 88170 }, { "epoch": 0.4379960859501505, "grad_norm": 0.6394097805023193, "learning_rate": 1.2468370774384248e-05, "loss": 0.160817551612854, "step": 88180 }, { "epoch": 0.43804575663351975, "grad_norm": 0.8902721405029297, "learning_rate": 1.2466811807180127e-05, "loss": 0.16646435260772705, "step": 88190 }, { "epoch": 0.43809542731688905, "grad_norm": 1.4093369245529175, "learning_rate": 1.2465252776135915e-05, "loss": 0.16530357599258422, "step": 88200 }, { "epoch": 0.4381450980002583, "grad_norm": 1.498801827430725, "learning_rate": 1.246369368129195e-05, "loss": 0.16522655487060547, "step": 88210 }, { "epoch": 0.43819476868362756, "grad_norm": 1.521807312965393, "learning_rate": 1.2462134522688582e-05, "loss": 0.15178093910217286, "step": 88220 }, { "epoch": 0.4382444393669968, "grad_norm": 0.9866085052490234, "learning_rate": 1.2460575300366164e-05, "loss": 0.17445952892303468, "step": 88230 }, { "epoch": 0.43829411005036606, "grad_norm": 1.2381483316421509, "learning_rate": 1.2459016014365046e-05, "loss": 0.1674232840538025, "step": 88240 }, { "epoch": 0.4383437807337353, "grad_norm": 1.5402253866195679, "learning_rate": 1.2457456664725584e-05, "loss": 0.15244086980819702, "step": 88250 }, { "epoch": 0.4383934514171046, "grad_norm": 2.393488645553589, "learning_rate": 1.245589725148813e-05, "loss": 0.1576091766357422, "step": 88260 }, { "epoch": 0.43844312210047387, "grad_norm": 1.1212800741195679, "learning_rate": 1.2454337774693042e-05, "loss": 0.1559206485748291, "step": 88270 }, { "epoch": 0.4384927927838431, "grad_norm": 2.14715838432312, "learning_rate": 1.2452778234380678e-05, "loss": 0.14951660633087158, "step": 88280 }, { "epoch": 0.4385424634672124, "grad_norm": 1.0155009031295776, "learning_rate": 1.2451218630591404e-05, "loss": 0.18058788776397705, "step": 88290 }, { "epoch": 0.4385921341505816, "grad_norm": 2.0525665283203125, "learning_rate": 1.2449658963365576e-05, "loss": 0.16222236156463624, "step": 88300 }, { "epoch": 0.43864180483395093, "grad_norm": 1.1363190412521362, "learning_rate": 1.2448099232743559e-05, "loss": 0.1588526725769043, "step": 88310 }, { "epoch": 0.4386914755173202, "grad_norm": 1.9596837759017944, "learning_rate": 1.2446539438765715e-05, "loss": 0.1623828411102295, "step": 88320 }, { "epoch": 0.43874114620068944, "grad_norm": 0.9143638014793396, "learning_rate": 1.2444979581472415e-05, "loss": 0.1618569850921631, "step": 88330 }, { "epoch": 0.4387908168840587, "grad_norm": 0.8331720232963562, "learning_rate": 1.2443419660904029e-05, "loss": 0.17053338289260864, "step": 88340 }, { "epoch": 0.43884048756742794, "grad_norm": 0.9969601035118103, "learning_rate": 1.2441859677100923e-05, "loss": 0.16459590196609497, "step": 88350 }, { "epoch": 0.4388901582507972, "grad_norm": 1.0366345643997192, "learning_rate": 1.244029963010347e-05, "loss": 0.17367955446243286, "step": 88360 }, { "epoch": 0.4389398289341665, "grad_norm": 2.2080492973327637, "learning_rate": 1.2438739519952045e-05, "loss": 0.17151904106140137, "step": 88370 }, { "epoch": 0.43898949961753575, "grad_norm": 1.064358115196228, "learning_rate": 1.2437179346687022e-05, "loss": 0.1688361406326294, "step": 88380 }, { "epoch": 0.439039170300905, "grad_norm": 0.8946011662483215, "learning_rate": 1.2435619110348773e-05, "loss": 0.15872015953063964, "step": 88390 }, { "epoch": 0.43908884098427425, "grad_norm": 0.7776949405670166, "learning_rate": 1.2434058810977686e-05, "loss": 0.16428059339523315, "step": 88400 }, { "epoch": 0.4391385116676435, "grad_norm": 1.2090933322906494, "learning_rate": 1.2432498448614134e-05, "loss": 0.15279632806777954, "step": 88410 }, { "epoch": 0.4391881823510128, "grad_norm": 1.1238675117492676, "learning_rate": 1.24309380232985e-05, "loss": 0.15464293956756592, "step": 88420 }, { "epoch": 0.43923785303438206, "grad_norm": 1.272033452987671, "learning_rate": 1.2429377535071167e-05, "loss": 0.16283000707626344, "step": 88430 }, { "epoch": 0.4392875237177513, "grad_norm": 2.024425506591797, "learning_rate": 1.2427816983972522e-05, "loss": 0.1746154546737671, "step": 88440 }, { "epoch": 0.43933719440112057, "grad_norm": 1.611321210861206, "learning_rate": 1.2426256370042951e-05, "loss": 0.171760630607605, "step": 88450 }, { "epoch": 0.4393868650844898, "grad_norm": 1.2730792760849, "learning_rate": 1.2424695693322837e-05, "loss": 0.17532973289489745, "step": 88460 }, { "epoch": 0.43943653576785907, "grad_norm": 0.9356038570404053, "learning_rate": 1.2423134953852577e-05, "loss": 0.15647851228713988, "step": 88470 }, { "epoch": 0.4394862064512284, "grad_norm": 0.8419961929321289, "learning_rate": 1.2421574151672556e-05, "loss": 0.1560274362564087, "step": 88480 }, { "epoch": 0.43953587713459763, "grad_norm": 1.0965913534164429, "learning_rate": 1.242001328682317e-05, "loss": 0.17428503036499024, "step": 88490 }, { "epoch": 0.4395855478179669, "grad_norm": 1.0205259323120117, "learning_rate": 1.2418452359344812e-05, "loss": 0.16626100540161132, "step": 88500 }, { "epoch": 0.43963521850133613, "grad_norm": 1.4779820442199707, "learning_rate": 1.241689136927788e-05, "loss": 0.169296932220459, "step": 88510 }, { "epoch": 0.4396848891847054, "grad_norm": 0.9754734635353088, "learning_rate": 1.2415330316662771e-05, "loss": 0.1657892107963562, "step": 88520 }, { "epoch": 0.4397345598680747, "grad_norm": 1.752211570739746, "learning_rate": 1.2413769201539885e-05, "loss": 0.1616539716720581, "step": 88530 }, { "epoch": 0.43978423055144394, "grad_norm": 1.0800479650497437, "learning_rate": 1.2412208023949624e-05, "loss": 0.18540263175964355, "step": 88540 }, { "epoch": 0.4398339012348132, "grad_norm": 0.9534083008766174, "learning_rate": 1.2410646783932388e-05, "loss": 0.1588301181793213, "step": 88550 }, { "epoch": 0.43988357191818245, "grad_norm": 1.159723162651062, "learning_rate": 1.2409085481528583e-05, "loss": 0.15157251358032225, "step": 88560 }, { "epoch": 0.4399332426015517, "grad_norm": 1.3011317253112793, "learning_rate": 1.2407524116778613e-05, "loss": 0.16441683769226073, "step": 88570 }, { "epoch": 0.43998291328492095, "grad_norm": 0.9445092082023621, "learning_rate": 1.2405962689722888e-05, "loss": 0.17111316919326783, "step": 88580 }, { "epoch": 0.44003258396829026, "grad_norm": 1.73910653591156, "learning_rate": 1.2404401200401816e-05, "loss": 0.18160403966903688, "step": 88590 }, { "epoch": 0.4400822546516595, "grad_norm": 1.1887214183807373, "learning_rate": 1.2402839648855807e-05, "loss": 0.15934596061706544, "step": 88600 }, { "epoch": 0.44013192533502876, "grad_norm": 0.765005886554718, "learning_rate": 1.2401278035125273e-05, "loss": 0.16796776056289672, "step": 88610 }, { "epoch": 0.440181596018398, "grad_norm": 1.3876134157180786, "learning_rate": 1.239971635925063e-05, "loss": 0.16119828224182128, "step": 88620 }, { "epoch": 0.44023126670176727, "grad_norm": 0.971722424030304, "learning_rate": 1.2398154621272295e-05, "loss": 0.1578136444091797, "step": 88630 }, { "epoch": 0.4402809373851365, "grad_norm": 1.5687445402145386, "learning_rate": 1.2396592821230681e-05, "loss": 0.16684234142303467, "step": 88640 }, { "epoch": 0.4403306080685058, "grad_norm": 3.151185989379883, "learning_rate": 1.2395030959166209e-05, "loss": 0.1564855694770813, "step": 88650 }, { "epoch": 0.4403802787518751, "grad_norm": 0.7061538696289062, "learning_rate": 1.2393469035119297e-05, "loss": 0.1620997190475464, "step": 88660 }, { "epoch": 0.44042994943524433, "grad_norm": 1.0321041345596313, "learning_rate": 1.2391907049130369e-05, "loss": 0.168113112449646, "step": 88670 }, { "epoch": 0.4404796201186136, "grad_norm": 2.190269947052002, "learning_rate": 1.239034500123985e-05, "loss": 0.16625213623046875, "step": 88680 }, { "epoch": 0.44052929080198283, "grad_norm": 0.638314425945282, "learning_rate": 1.2388782891488162e-05, "loss": 0.17046544551849366, "step": 88690 }, { "epoch": 0.44057896148535214, "grad_norm": 0.671109676361084, "learning_rate": 1.2387220719915733e-05, "loss": 0.16595244407653809, "step": 88700 }, { "epoch": 0.4406286321687214, "grad_norm": 1.2050498723983765, "learning_rate": 1.2385658486562995e-05, "loss": 0.1676732897758484, "step": 88710 }, { "epoch": 0.44067830285209064, "grad_norm": 2.0145013332366943, "learning_rate": 1.2384096191470371e-05, "loss": 0.1635391116142273, "step": 88720 }, { "epoch": 0.4407279735354599, "grad_norm": 0.907046914100647, "learning_rate": 1.2382533834678295e-05, "loss": 0.16133178472518922, "step": 88730 }, { "epoch": 0.44077764421882915, "grad_norm": 0.94202721118927, "learning_rate": 1.2380971416227202e-05, "loss": 0.1601993203163147, "step": 88740 }, { "epoch": 0.4408273149021984, "grad_norm": 2.320469617843628, "learning_rate": 1.237940893615753e-05, "loss": 0.17368724346160888, "step": 88750 }, { "epoch": 0.4408769855855677, "grad_norm": 1.234792947769165, "learning_rate": 1.237784639450971e-05, "loss": 0.1861906409263611, "step": 88760 }, { "epoch": 0.44092665626893696, "grad_norm": 0.9961288571357727, "learning_rate": 1.2376283791324179e-05, "loss": 0.16171069145202638, "step": 88770 }, { "epoch": 0.4409763269523062, "grad_norm": 0.550373911857605, "learning_rate": 1.2374721126641381e-05, "loss": 0.15173726081848143, "step": 88780 }, { "epoch": 0.44102599763567546, "grad_norm": 0.7520491480827332, "learning_rate": 1.2373158400501755e-05, "loss": 0.16020219326019286, "step": 88790 }, { "epoch": 0.4410756683190447, "grad_norm": 1.09280526638031, "learning_rate": 1.2371595612945741e-05, "loss": 0.1568335175514221, "step": 88800 }, { "epoch": 0.441125339002414, "grad_norm": 0.5929107666015625, "learning_rate": 1.237003276401379e-05, "loss": 0.16138312816619874, "step": 88810 }, { "epoch": 0.44117500968578327, "grad_norm": 1.825688123703003, "learning_rate": 1.2368469853746343e-05, "loss": 0.17342573404312134, "step": 88820 }, { "epoch": 0.4412246803691525, "grad_norm": 1.0788747072219849, "learning_rate": 1.2366906882183849e-05, "loss": 0.1585613965988159, "step": 88830 }, { "epoch": 0.4412743510525218, "grad_norm": 0.5537697076797485, "learning_rate": 1.2365343849366756e-05, "loss": 0.1763153076171875, "step": 88840 }, { "epoch": 0.441324021735891, "grad_norm": 1.3417961597442627, "learning_rate": 1.2363780755335512e-05, "loss": 0.16552948951721191, "step": 88850 }, { "epoch": 0.4413736924192603, "grad_norm": 0.9783541560173035, "learning_rate": 1.2362217600130575e-05, "loss": 0.16803216934204102, "step": 88860 }, { "epoch": 0.4414233631026296, "grad_norm": 1.001283049583435, "learning_rate": 1.2360654383792396e-05, "loss": 0.1591695189476013, "step": 88870 }, { "epoch": 0.44147303378599884, "grad_norm": 0.7682027220726013, "learning_rate": 1.2359091106361431e-05, "loss": 0.1724885106086731, "step": 88880 }, { "epoch": 0.4415227044693681, "grad_norm": 1.4238423109054565, "learning_rate": 1.2357527767878139e-05, "loss": 0.16218690872192382, "step": 88890 }, { "epoch": 0.44157237515273734, "grad_norm": 0.9023235440254211, "learning_rate": 1.2355964368382974e-05, "loss": 0.15496401786804198, "step": 88900 }, { "epoch": 0.4416220458361066, "grad_norm": 0.9965882301330566, "learning_rate": 1.2354400907916398e-05, "loss": 0.16573591232299806, "step": 88910 }, { "epoch": 0.4416717165194759, "grad_norm": 1.747316837310791, "learning_rate": 1.2352837386518875e-05, "loss": 0.16941194534301757, "step": 88920 }, { "epoch": 0.44172138720284515, "grad_norm": 1.1567386388778687, "learning_rate": 1.2351273804230864e-05, "loss": 0.1682162404060364, "step": 88930 }, { "epoch": 0.4417710578862144, "grad_norm": 1.4560039043426514, "learning_rate": 1.2349710161092831e-05, "loss": 0.16329824924468994, "step": 88940 }, { "epoch": 0.44182072856958365, "grad_norm": 0.8296149969100952, "learning_rate": 1.2348146457145249e-05, "loss": 0.15859298706054686, "step": 88950 }, { "epoch": 0.4418703992529529, "grad_norm": 2.076724052429199, "learning_rate": 1.2346582692428577e-05, "loss": 0.1640216588973999, "step": 88960 }, { "epoch": 0.44192006993632216, "grad_norm": 2.035184383392334, "learning_rate": 1.2345018866983294e-05, "loss": 0.16900436878204345, "step": 88970 }, { "epoch": 0.44196974061969146, "grad_norm": 0.5892922282218933, "learning_rate": 1.234345498084986e-05, "loss": 0.16035068035125732, "step": 88980 }, { "epoch": 0.4420194113030607, "grad_norm": 1.1411247253417969, "learning_rate": 1.2341891034068754e-05, "loss": 0.153268563747406, "step": 88990 }, { "epoch": 0.44206908198642997, "grad_norm": 1.0683188438415527, "learning_rate": 1.234032702668045e-05, "loss": 0.16203827857971193, "step": 89000 }, { "epoch": 0.4421187526697992, "grad_norm": 0.9169057607650757, "learning_rate": 1.2338762958725423e-05, "loss": 0.1648927927017212, "step": 89010 }, { "epoch": 0.44216842335316847, "grad_norm": 1.851110816001892, "learning_rate": 1.2337198830244152e-05, "loss": 0.1743492841720581, "step": 89020 }, { "epoch": 0.4422180940365378, "grad_norm": 0.9908380508422852, "learning_rate": 1.2335634641277117e-05, "loss": 0.153617787361145, "step": 89030 }, { "epoch": 0.44226776471990703, "grad_norm": 2.1071245670318604, "learning_rate": 1.2334070391864794e-05, "loss": 0.17126498222351075, "step": 89040 }, { "epoch": 0.4423174354032763, "grad_norm": 1.0766710042953491, "learning_rate": 1.2332506082047668e-05, "loss": 0.17394530773162842, "step": 89050 }, { "epoch": 0.44236710608664553, "grad_norm": 0.6995921730995178, "learning_rate": 1.2330941711866223e-05, "loss": 0.15944209098815917, "step": 89060 }, { "epoch": 0.4424167767700148, "grad_norm": 1.3863143920898438, "learning_rate": 1.2329377281360941e-05, "loss": 0.16385107040405272, "step": 89070 }, { "epoch": 0.44246644745338404, "grad_norm": 1.331241488456726, "learning_rate": 1.2327812790572315e-05, "loss": 0.16668124198913575, "step": 89080 }, { "epoch": 0.44251611813675334, "grad_norm": 1.355183482170105, "learning_rate": 1.232624823954083e-05, "loss": 0.16640552282333373, "step": 89090 }, { "epoch": 0.4425657888201226, "grad_norm": 1.0077762603759766, "learning_rate": 1.2324683628306974e-05, "loss": 0.1802377223968506, "step": 89100 }, { "epoch": 0.44261545950349185, "grad_norm": 1.485638976097107, "learning_rate": 1.2323118956911242e-05, "loss": 0.1663991093635559, "step": 89110 }, { "epoch": 0.4426651301868611, "grad_norm": 1.5666942596435547, "learning_rate": 1.2321554225394125e-05, "loss": 0.17325783967971803, "step": 89120 }, { "epoch": 0.44271480087023035, "grad_norm": 1.0431814193725586, "learning_rate": 1.2319989433796119e-05, "loss": 0.16954710483551025, "step": 89130 }, { "epoch": 0.44276447155359966, "grad_norm": 0.7179310917854309, "learning_rate": 1.2318424582157718e-05, "loss": 0.1735679864883423, "step": 89140 }, { "epoch": 0.4428141422369689, "grad_norm": 2.1056582927703857, "learning_rate": 1.2316859670519421e-05, "loss": 0.16079033613204957, "step": 89150 }, { "epoch": 0.44286381292033816, "grad_norm": 1.4784795045852661, "learning_rate": 1.2315294698921729e-05, "loss": 0.15389968156814576, "step": 89160 }, { "epoch": 0.4429134836037074, "grad_norm": 1.9013822078704834, "learning_rate": 1.2313729667405138e-05, "loss": 0.1561036229133606, "step": 89170 }, { "epoch": 0.44296315428707667, "grad_norm": 1.2447346448898315, "learning_rate": 1.2312164576010157e-05, "loss": 0.15562775135040283, "step": 89180 }, { "epoch": 0.4430128249704459, "grad_norm": 1.9644312858581543, "learning_rate": 1.2310599424777285e-05, "loss": 0.16751594543457032, "step": 89190 }, { "epoch": 0.4430624956538152, "grad_norm": 1.0180608034133911, "learning_rate": 1.2309034213747026e-05, "loss": 0.1756695866584778, "step": 89200 }, { "epoch": 0.4431121663371845, "grad_norm": 0.6850646734237671, "learning_rate": 1.2307468942959892e-05, "loss": 0.15420398712158204, "step": 89210 }, { "epoch": 0.44316183702055373, "grad_norm": 1.7352104187011719, "learning_rate": 1.2305903612456392e-05, "loss": 0.16360007524490355, "step": 89220 }, { "epoch": 0.443211507703923, "grad_norm": 0.9726281762123108, "learning_rate": 1.230433822227703e-05, "loss": 0.16189086437225342, "step": 89230 }, { "epoch": 0.44326117838729223, "grad_norm": 0.8398205041885376, "learning_rate": 1.2302772772462323e-05, "loss": 0.16334669589996337, "step": 89240 }, { "epoch": 0.44331084907066154, "grad_norm": 1.6125214099884033, "learning_rate": 1.2301207263052786e-05, "loss": 0.16310758590698243, "step": 89250 }, { "epoch": 0.4433605197540308, "grad_norm": 1.4334641695022583, "learning_rate": 1.2299641694088926e-05, "loss": 0.17721824645996093, "step": 89260 }, { "epoch": 0.44341019043740004, "grad_norm": 1.9059455394744873, "learning_rate": 1.2298076065611261e-05, "loss": 0.16578028202056885, "step": 89270 }, { "epoch": 0.4434598611207693, "grad_norm": 0.6871940493583679, "learning_rate": 1.2296510377660318e-05, "loss": 0.15125684738159179, "step": 89280 }, { "epoch": 0.44350953180413855, "grad_norm": 1.4087327718734741, "learning_rate": 1.2294944630276606e-05, "loss": 0.16913011074066162, "step": 89290 }, { "epoch": 0.4435592024875078, "grad_norm": 1.9722087383270264, "learning_rate": 1.229337882350065e-05, "loss": 0.15634311437606813, "step": 89300 }, { "epoch": 0.4436088731708771, "grad_norm": 2.5535542964935303, "learning_rate": 1.2291812957372976e-05, "loss": 0.16734431982040404, "step": 89310 }, { "epoch": 0.44365854385424636, "grad_norm": 1.4626309871673584, "learning_rate": 1.2290247031934101e-05, "loss": 0.1581650495529175, "step": 89320 }, { "epoch": 0.4437082145376156, "grad_norm": 1.7588276863098145, "learning_rate": 1.2288681047224551e-05, "loss": 0.1803330659866333, "step": 89330 }, { "epoch": 0.44375788522098486, "grad_norm": 0.9680184721946716, "learning_rate": 1.2287115003284859e-05, "loss": 0.15702316761016846, "step": 89340 }, { "epoch": 0.4438075559043541, "grad_norm": 1.2008882761001587, "learning_rate": 1.2285548900155551e-05, "loss": 0.16982908248901368, "step": 89350 }, { "epoch": 0.4438572265877234, "grad_norm": 1.2939467430114746, "learning_rate": 1.2283982737877153e-05, "loss": 0.17504656314849854, "step": 89360 }, { "epoch": 0.44390689727109267, "grad_norm": 1.1513671875, "learning_rate": 1.2282416516490203e-05, "loss": 0.16883634328842162, "step": 89370 }, { "epoch": 0.4439565679544619, "grad_norm": 1.3981105089187622, "learning_rate": 1.228085023603523e-05, "loss": 0.1771239757537842, "step": 89380 }, { "epoch": 0.4440062386378312, "grad_norm": 1.4330111742019653, "learning_rate": 1.2279283896552772e-05, "loss": 0.1761365056037903, "step": 89390 }, { "epoch": 0.4440559093212004, "grad_norm": 0.9748993515968323, "learning_rate": 1.227771749808336e-05, "loss": 0.1672877311706543, "step": 89400 }, { "epoch": 0.4441055800045697, "grad_norm": 1.1094129085540771, "learning_rate": 1.2276151040667536e-05, "loss": 0.14881613254547119, "step": 89410 }, { "epoch": 0.444155250687939, "grad_norm": 1.6043753623962402, "learning_rate": 1.2274584524345841e-05, "loss": 0.17467018365859985, "step": 89420 }, { "epoch": 0.44420492137130824, "grad_norm": 0.9265424609184265, "learning_rate": 1.227301794915881e-05, "loss": 0.16666929721832274, "step": 89430 }, { "epoch": 0.4442545920546775, "grad_norm": 0.7504160404205322, "learning_rate": 1.2271451315146989e-05, "loss": 0.15773181915283202, "step": 89440 }, { "epoch": 0.44430426273804674, "grad_norm": 1.3924087285995483, "learning_rate": 1.2269884622350923e-05, "loss": 0.15186028480529784, "step": 89450 }, { "epoch": 0.444353933421416, "grad_norm": 1.6797568798065186, "learning_rate": 1.2268317870811155e-05, "loss": 0.163388454914093, "step": 89460 }, { "epoch": 0.4444036041047853, "grad_norm": 1.4719077348709106, "learning_rate": 1.2266751060568231e-05, "loss": 0.16017110347747804, "step": 89470 }, { "epoch": 0.44445327478815455, "grad_norm": 1.1356141567230225, "learning_rate": 1.2265184191662702e-05, "loss": 0.16293997764587403, "step": 89480 }, { "epoch": 0.4445029454715238, "grad_norm": 1.1248666048049927, "learning_rate": 1.2263617264135118e-05, "loss": 0.17711172103881836, "step": 89490 }, { "epoch": 0.44455261615489305, "grad_norm": 0.6995410919189453, "learning_rate": 1.2262050278026028e-05, "loss": 0.17810494899749757, "step": 89500 }, { "epoch": 0.4446022868382623, "grad_norm": 1.692017674446106, "learning_rate": 1.2260483233375985e-05, "loss": 0.17531251907348633, "step": 89510 }, { "epoch": 0.44465195752163156, "grad_norm": 0.979611873626709, "learning_rate": 1.2258916130225548e-05, "loss": 0.16010491847991942, "step": 89520 }, { "epoch": 0.44470162820500087, "grad_norm": 1.2367571592330933, "learning_rate": 1.2257348968615265e-05, "loss": 0.1659505009651184, "step": 89530 }, { "epoch": 0.4447512988883701, "grad_norm": 1.6470518112182617, "learning_rate": 1.2255781748585705e-05, "loss": 0.16426639556884765, "step": 89540 }, { "epoch": 0.44480096957173937, "grad_norm": 0.7079128623008728, "learning_rate": 1.2254214470177415e-05, "loss": 0.17355687618255616, "step": 89550 }, { "epoch": 0.4448506402551086, "grad_norm": 0.9428579211235046, "learning_rate": 1.2252647133430964e-05, "loss": 0.15822170972824096, "step": 89560 }, { "epoch": 0.44490031093847787, "grad_norm": 1.2145408391952515, "learning_rate": 1.2251079738386909e-05, "loss": 0.17516323328018188, "step": 89570 }, { "epoch": 0.4449499816218472, "grad_norm": 1.7168316841125488, "learning_rate": 1.2249512285085816e-05, "loss": 0.18037917613983154, "step": 89580 }, { "epoch": 0.44499965230521643, "grad_norm": 1.158643364906311, "learning_rate": 1.224794477356825e-05, "loss": 0.1639082908630371, "step": 89590 }, { "epoch": 0.4450493229885857, "grad_norm": 1.4199028015136719, "learning_rate": 1.2246377203874775e-05, "loss": 0.1751335620880127, "step": 89600 }, { "epoch": 0.44509899367195493, "grad_norm": 1.074761986732483, "learning_rate": 1.2244809576045963e-05, "loss": 0.174118709564209, "step": 89610 }, { "epoch": 0.4451486643553242, "grad_norm": 0.7944638133049011, "learning_rate": 1.2243241890122384e-05, "loss": 0.16300537586212158, "step": 89620 }, { "epoch": 0.44519833503869344, "grad_norm": 1.565683364868164, "learning_rate": 1.2241674146144604e-05, "loss": 0.18430815935134887, "step": 89630 }, { "epoch": 0.44524800572206275, "grad_norm": 1.0859904289245605, "learning_rate": 1.2240106344153201e-05, "loss": 0.17490768432617188, "step": 89640 }, { "epoch": 0.445297676405432, "grad_norm": 1.2648431062698364, "learning_rate": 1.2238538484188744e-05, "loss": 0.16553189754486083, "step": 89650 }, { "epoch": 0.44534734708880125, "grad_norm": 0.8089122176170349, "learning_rate": 1.223697056629181e-05, "loss": 0.16774249076843262, "step": 89660 }, { "epoch": 0.4453970177721705, "grad_norm": 0.9850976467132568, "learning_rate": 1.223540259050298e-05, "loss": 0.16156959533691406, "step": 89670 }, { "epoch": 0.44544668845553975, "grad_norm": 1.0935614109039307, "learning_rate": 1.2233834556862828e-05, "loss": 0.1673784375190735, "step": 89680 }, { "epoch": 0.44549635913890906, "grad_norm": 1.9261268377304077, "learning_rate": 1.2232266465411938e-05, "loss": 0.17299890518188477, "step": 89690 }, { "epoch": 0.4455460298222783, "grad_norm": 0.5673452615737915, "learning_rate": 1.2230698316190888e-05, "loss": 0.15702120065689087, "step": 89700 }, { "epoch": 0.44559570050564756, "grad_norm": 2.0022575855255127, "learning_rate": 1.2229130109240264e-05, "loss": 0.16766564846038817, "step": 89710 }, { "epoch": 0.4456453711890168, "grad_norm": 0.6990725994110107, "learning_rate": 1.2227561844600646e-05, "loss": 0.16308305263519288, "step": 89720 }, { "epoch": 0.44569504187238607, "grad_norm": 1.6080437898635864, "learning_rate": 1.2225993522312625e-05, "loss": 0.16374553442001344, "step": 89730 }, { "epoch": 0.4457447125557553, "grad_norm": 0.9912058711051941, "learning_rate": 1.2224425142416785e-05, "loss": 0.15538861751556396, "step": 89740 }, { "epoch": 0.4457943832391246, "grad_norm": 0.9877636432647705, "learning_rate": 1.2222856704953718e-05, "loss": 0.1751528263092041, "step": 89750 }, { "epoch": 0.4458440539224939, "grad_norm": 0.8892027139663696, "learning_rate": 1.2221288209964013e-05, "loss": 0.15842776298522948, "step": 89760 }, { "epoch": 0.44589372460586313, "grad_norm": 1.608956217765808, "learning_rate": 1.2219719657488263e-05, "loss": 0.15757303237915038, "step": 89770 }, { "epoch": 0.4459433952892324, "grad_norm": 1.5464131832122803, "learning_rate": 1.2218151047567062e-05, "loss": 0.1566240072250366, "step": 89780 }, { "epoch": 0.44599306597260163, "grad_norm": 1.395365834236145, "learning_rate": 1.2216582380241003e-05, "loss": 0.1545366168022156, "step": 89790 }, { "epoch": 0.44604273665597094, "grad_norm": 1.4042221307754517, "learning_rate": 1.221501365555068e-05, "loss": 0.16809258460998536, "step": 89800 }, { "epoch": 0.4460924073393402, "grad_norm": 2.2653138637542725, "learning_rate": 1.2213444873536697e-05, "loss": 0.15865378379821776, "step": 89810 }, { "epoch": 0.44614207802270944, "grad_norm": 1.3264542818069458, "learning_rate": 1.2211876034239652e-05, "loss": 0.17503395080566406, "step": 89820 }, { "epoch": 0.4461917487060787, "grad_norm": 1.7135462760925293, "learning_rate": 1.2210307137700144e-05, "loss": 0.15954099893569945, "step": 89830 }, { "epoch": 0.44624141938944795, "grad_norm": 1.0638421773910522, "learning_rate": 1.2208738183958774e-05, "loss": 0.17760028839111328, "step": 89840 }, { "epoch": 0.4462910900728172, "grad_norm": 0.7927803993225098, "learning_rate": 1.2207169173056149e-05, "loss": 0.16512629985809327, "step": 89850 }, { "epoch": 0.4463407607561865, "grad_norm": 0.8763209581375122, "learning_rate": 1.2205600105032874e-05, "loss": 0.1706179141998291, "step": 89860 }, { "epoch": 0.44639043143955576, "grad_norm": 1.6118406057357788, "learning_rate": 1.2204030979929551e-05, "loss": 0.1717020034790039, "step": 89870 }, { "epoch": 0.446440102122925, "grad_norm": 0.7840917110443115, "learning_rate": 1.2202461797786796e-05, "loss": 0.17602720260620117, "step": 89880 }, { "epoch": 0.44648977280629426, "grad_norm": 1.7907463312149048, "learning_rate": 1.2200892558645214e-05, "loss": 0.16427161693572997, "step": 89890 }, { "epoch": 0.4465394434896635, "grad_norm": 1.843049168586731, "learning_rate": 1.2199323262545419e-05, "loss": 0.17340948581695556, "step": 89900 }, { "epoch": 0.4465891141730328, "grad_norm": 1.1513463258743286, "learning_rate": 1.2197753909528022e-05, "loss": 0.16824865341186523, "step": 89910 }, { "epoch": 0.44663878485640207, "grad_norm": 1.345067024230957, "learning_rate": 1.2196184499633639e-05, "loss": 0.1747083306312561, "step": 89920 }, { "epoch": 0.4466884555397713, "grad_norm": 1.7038415670394897, "learning_rate": 1.219461503290288e-05, "loss": 0.16388449668884278, "step": 89930 }, { "epoch": 0.4467381262231406, "grad_norm": 1.5294421911239624, "learning_rate": 1.219304550937637e-05, "loss": 0.17717593908309937, "step": 89940 }, { "epoch": 0.4467877969065098, "grad_norm": 0.6620285511016846, "learning_rate": 1.219147592909472e-05, "loss": 0.16910597085952758, "step": 89950 }, { "epoch": 0.4468374675898791, "grad_norm": 0.7763649225234985, "learning_rate": 1.218990629209856e-05, "loss": 0.16691042184829713, "step": 89960 }, { "epoch": 0.4468871382732484, "grad_norm": 1.2364963293075562, "learning_rate": 1.2188336598428501e-05, "loss": 0.1534580707550049, "step": 89970 }, { "epoch": 0.44693680895661764, "grad_norm": 0.9419384598731995, "learning_rate": 1.2186766848125176e-05, "loss": 0.16505808830261232, "step": 89980 }, { "epoch": 0.4469864796399869, "grad_norm": 1.3167310953140259, "learning_rate": 1.21851970412292e-05, "loss": 0.1717296361923218, "step": 89990 }, { "epoch": 0.44703615032335614, "grad_norm": 1.3263797760009766, "learning_rate": 1.2183627177781201e-05, "loss": 0.16834059953689576, "step": 90000 }, { "epoch": 0.4470858210067254, "grad_norm": 1.0557185411453247, "learning_rate": 1.2182057257821812e-05, "loss": 0.17169904708862305, "step": 90010 }, { "epoch": 0.4471354916900947, "grad_norm": 1.2818068265914917, "learning_rate": 1.2180487281391657e-05, "loss": 0.16320536136627198, "step": 90020 }, { "epoch": 0.44718516237346395, "grad_norm": 1.0911929607391357, "learning_rate": 1.217891724853137e-05, "loss": 0.16228370666503905, "step": 90030 }, { "epoch": 0.4472348330568332, "grad_norm": 1.9423623085021973, "learning_rate": 1.2177347159281579e-05, "loss": 0.1694866895675659, "step": 90040 }, { "epoch": 0.44728450374020245, "grad_norm": 1.04288911819458, "learning_rate": 1.2175777013682921e-05, "loss": 0.15905408859252929, "step": 90050 }, { "epoch": 0.4473341744235717, "grad_norm": 1.8485138416290283, "learning_rate": 1.2174206811776027e-05, "loss": 0.1611022710800171, "step": 90060 }, { "epoch": 0.44738384510694096, "grad_norm": 1.1164981126785278, "learning_rate": 1.2172636553601533e-05, "loss": 0.16349810361862183, "step": 90070 }, { "epoch": 0.44743351579031027, "grad_norm": 2.247239828109741, "learning_rate": 1.217106623920008e-05, "loss": 0.1641390323638916, "step": 90080 }, { "epoch": 0.4474831864736795, "grad_norm": 1.0794190168380737, "learning_rate": 1.2169495868612307e-05, "loss": 0.1651850700378418, "step": 90090 }, { "epoch": 0.44753285715704877, "grad_norm": 1.0798639059066772, "learning_rate": 1.2167925441878855e-05, "loss": 0.16219696998596192, "step": 90100 }, { "epoch": 0.447582527840418, "grad_norm": 1.689247727394104, "learning_rate": 1.2166354959040361e-05, "loss": 0.16681114435195923, "step": 90110 }, { "epoch": 0.4476321985237873, "grad_norm": 0.7510598301887512, "learning_rate": 1.2164784420137472e-05, "loss": 0.1680748701095581, "step": 90120 }, { "epoch": 0.4476818692071566, "grad_norm": 1.1140819787979126, "learning_rate": 1.2163213825210834e-05, "loss": 0.16687326431274413, "step": 90130 }, { "epoch": 0.44773153989052583, "grad_norm": 0.781226634979248, "learning_rate": 1.216164317430109e-05, "loss": 0.16091406345367432, "step": 90140 }, { "epoch": 0.4477812105738951, "grad_norm": 0.8256127238273621, "learning_rate": 1.2160072467448892e-05, "loss": 0.16795048713684083, "step": 90150 }, { "epoch": 0.44783088125726433, "grad_norm": 1.4124395847320557, "learning_rate": 1.2158501704694888e-05, "loss": 0.16371266841888427, "step": 90160 }, { "epoch": 0.4478805519406336, "grad_norm": 1.605431318283081, "learning_rate": 1.2156930886079725e-05, "loss": 0.164797580242157, "step": 90170 }, { "epoch": 0.44793022262400284, "grad_norm": 1.1643085479736328, "learning_rate": 1.215536001164406e-05, "loss": 0.1765834093093872, "step": 90180 }, { "epoch": 0.44797989330737215, "grad_norm": 1.744529366493225, "learning_rate": 1.2153789081428542e-05, "loss": 0.1686071276664734, "step": 90190 }, { "epoch": 0.4480295639907414, "grad_norm": 1.2388361692428589, "learning_rate": 1.2152218095473828e-05, "loss": 0.15428574085235597, "step": 90200 }, { "epoch": 0.44807923467411065, "grad_norm": 1.1882500648498535, "learning_rate": 1.2150647053820577e-05, "loss": 0.16133506298065187, "step": 90210 }, { "epoch": 0.4481289053574799, "grad_norm": 0.9485244154930115, "learning_rate": 1.2149075956509447e-05, "loss": 0.16200947761535645, "step": 90220 }, { "epoch": 0.44817857604084915, "grad_norm": 1.7344660758972168, "learning_rate": 1.2147504803581092e-05, "loss": 0.1646106481552124, "step": 90230 }, { "epoch": 0.4482282467242184, "grad_norm": 0.7564727663993835, "learning_rate": 1.2145933595076179e-05, "loss": 0.1605301856994629, "step": 90240 }, { "epoch": 0.4482779174075877, "grad_norm": 1.4765516519546509, "learning_rate": 1.2144362331035364e-05, "loss": 0.1665749430656433, "step": 90250 }, { "epoch": 0.44832758809095696, "grad_norm": 1.2482138872146606, "learning_rate": 1.2142791011499318e-05, "loss": 0.15861890316009522, "step": 90260 }, { "epoch": 0.4483772587743262, "grad_norm": 1.9393589496612549, "learning_rate": 1.2141219636508699e-05, "loss": 0.1674405336380005, "step": 90270 }, { "epoch": 0.44842692945769547, "grad_norm": 1.061968445777893, "learning_rate": 1.2139648206104176e-05, "loss": 0.15607771873474122, "step": 90280 }, { "epoch": 0.4484766001410647, "grad_norm": 1.2277262210845947, "learning_rate": 1.213807672032642e-05, "loss": 0.1655795931816101, "step": 90290 }, { "epoch": 0.448526270824434, "grad_norm": 1.9964568614959717, "learning_rate": 1.2136505179216098e-05, "loss": 0.18207569122314454, "step": 90300 }, { "epoch": 0.4485759415078033, "grad_norm": 0.8538001179695129, "learning_rate": 1.213493358281388e-05, "loss": 0.16473451852798462, "step": 90310 }, { "epoch": 0.44862561219117253, "grad_norm": 1.9699007272720337, "learning_rate": 1.2133361931160444e-05, "loss": 0.16892664432525634, "step": 90320 }, { "epoch": 0.4486752828745418, "grad_norm": 1.189497947692871, "learning_rate": 1.2131790224296453e-05, "loss": 0.17242848873138428, "step": 90330 }, { "epoch": 0.44872495355791103, "grad_norm": 1.2954691648483276, "learning_rate": 1.213021846226259e-05, "loss": 0.1675681471824646, "step": 90340 }, { "epoch": 0.4487746242412803, "grad_norm": 1.7840932607650757, "learning_rate": 1.212864664509953e-05, "loss": 0.16112208366394043, "step": 90350 }, { "epoch": 0.4488242949246496, "grad_norm": 0.7816578149795532, "learning_rate": 1.212707477284795e-05, "loss": 0.16395261287689208, "step": 90360 }, { "epoch": 0.44887396560801884, "grad_norm": 1.7047309875488281, "learning_rate": 1.2125502845548534e-05, "loss": 0.15914347171783447, "step": 90370 }, { "epoch": 0.4489236362913881, "grad_norm": 0.95008784532547, "learning_rate": 1.2123930863241956e-05, "loss": 0.17473807334899902, "step": 90380 }, { "epoch": 0.44897330697475735, "grad_norm": 0.6098822951316833, "learning_rate": 1.2122358825968901e-05, "loss": 0.15823535919189452, "step": 90390 }, { "epoch": 0.4490229776581266, "grad_norm": 0.953822135925293, "learning_rate": 1.2120786733770054e-05, "loss": 0.15916503667831422, "step": 90400 }, { "epoch": 0.4490726483414959, "grad_norm": 1.0739742517471313, "learning_rate": 1.2119214586686094e-05, "loss": 0.16970545053482056, "step": 90410 }, { "epoch": 0.44912231902486516, "grad_norm": 1.0741456747055054, "learning_rate": 1.2117642384757717e-05, "loss": 0.15494600534439087, "step": 90420 }, { "epoch": 0.4491719897082344, "grad_norm": 1.2959028482437134, "learning_rate": 1.2116070128025608e-05, "loss": 0.1756061792373657, "step": 90430 }, { "epoch": 0.44922166039160366, "grad_norm": 1.1249858140945435, "learning_rate": 1.2114497816530453e-05, "loss": 0.16610931158065795, "step": 90440 }, { "epoch": 0.4492713310749729, "grad_norm": 0.6814362406730652, "learning_rate": 1.2112925450312944e-05, "loss": 0.14542899131774903, "step": 90450 }, { "epoch": 0.44932100175834216, "grad_norm": 0.6999148726463318, "learning_rate": 1.2111353029413775e-05, "loss": 0.1644808053970337, "step": 90460 }, { "epoch": 0.44937067244171147, "grad_norm": 1.1261340379714966, "learning_rate": 1.210978055387364e-05, "loss": 0.15732405185699463, "step": 90470 }, { "epoch": 0.4494203431250807, "grad_norm": 0.9110631346702576, "learning_rate": 1.2108208023733229e-05, "loss": 0.17040095329284669, "step": 90480 }, { "epoch": 0.44947001380845, "grad_norm": 1.457964301109314, "learning_rate": 1.2106635439033246e-05, "loss": 0.16312036514282227, "step": 90490 }, { "epoch": 0.4495196844918192, "grad_norm": 1.148569107055664, "learning_rate": 1.2105062799814382e-05, "loss": 0.16258891820907592, "step": 90500 }, { "epoch": 0.4495693551751885, "grad_norm": 2.307170867919922, "learning_rate": 1.2103490106117341e-05, "loss": 0.15962731838226318, "step": 90510 }, { "epoch": 0.4496190258585578, "grad_norm": 1.9636574983596802, "learning_rate": 1.2101917357982822e-05, "loss": 0.15479688644409179, "step": 90520 }, { "epoch": 0.44966869654192704, "grad_norm": 2.1464407444000244, "learning_rate": 1.2100344555451524e-05, "loss": 0.15400590896606445, "step": 90530 }, { "epoch": 0.4497183672252963, "grad_norm": 0.7546982765197754, "learning_rate": 1.2098771698564154e-05, "loss": 0.17040667533874512, "step": 90540 }, { "epoch": 0.44976803790866554, "grad_norm": 0.892403244972229, "learning_rate": 1.2097198787361417e-05, "loss": 0.16761283874511718, "step": 90550 }, { "epoch": 0.4498177085920348, "grad_norm": 1.3709274530410767, "learning_rate": 1.209562582188402e-05, "loss": 0.1712332248687744, "step": 90560 }, { "epoch": 0.44986737927540404, "grad_norm": 1.1981436014175415, "learning_rate": 1.2094052802172668e-05, "loss": 0.1686212658882141, "step": 90570 }, { "epoch": 0.44991704995877335, "grad_norm": 2.329545497894287, "learning_rate": 1.2092479728268073e-05, "loss": 0.15968821048736573, "step": 90580 }, { "epoch": 0.4499667206421426, "grad_norm": 2.0632505416870117, "learning_rate": 1.2090906600210942e-05, "loss": 0.16206140518188478, "step": 90590 }, { "epoch": 0.45001639132551186, "grad_norm": 1.8821724653244019, "learning_rate": 1.2089333418041991e-05, "loss": 0.16434839963912964, "step": 90600 }, { "epoch": 0.4500660620088811, "grad_norm": 0.6728906035423279, "learning_rate": 1.208776018180193e-05, "loss": 0.16942098140716552, "step": 90610 }, { "epoch": 0.45011573269225036, "grad_norm": 1.7606159448623657, "learning_rate": 1.2086186891531472e-05, "loss": 0.1693657636642456, "step": 90620 }, { "epoch": 0.45016540337561967, "grad_norm": 0.6401776671409607, "learning_rate": 1.2084613547271342e-05, "loss": 0.16354509592056274, "step": 90630 }, { "epoch": 0.4502150740589889, "grad_norm": 0.937309980392456, "learning_rate": 1.2083040149062248e-05, "loss": 0.17092132568359375, "step": 90640 }, { "epoch": 0.45026474474235817, "grad_norm": 1.4461982250213623, "learning_rate": 1.2081466696944912e-05, "loss": 0.16045749187469482, "step": 90650 }, { "epoch": 0.4503144154257274, "grad_norm": 0.9450016021728516, "learning_rate": 1.207989319096006e-05, "loss": 0.1742257833480835, "step": 90660 }, { "epoch": 0.4503640861090967, "grad_norm": 1.527881383895874, "learning_rate": 1.2078319631148403e-05, "loss": 0.1703449010848999, "step": 90670 }, { "epoch": 0.4504137567924659, "grad_norm": 1.3117729425430298, "learning_rate": 1.2076746017550672e-05, "loss": 0.16020410060882567, "step": 90680 }, { "epoch": 0.45046342747583523, "grad_norm": 1.180750846862793, "learning_rate": 1.2075172350207587e-05, "loss": 0.17848504781723024, "step": 90690 }, { "epoch": 0.4505130981592045, "grad_norm": 0.9988741278648376, "learning_rate": 1.2073598629159877e-05, "loss": 0.16887027025222778, "step": 90700 }, { "epoch": 0.45056276884257374, "grad_norm": 1.4475175142288208, "learning_rate": 1.2072024854448268e-05, "loss": 0.15652352571487427, "step": 90710 }, { "epoch": 0.450612439525943, "grad_norm": 1.3420383930206299, "learning_rate": 1.207045102611349e-05, "loss": 0.15676323175430298, "step": 90720 }, { "epoch": 0.45066211020931224, "grad_norm": 1.2657768726348877, "learning_rate": 1.2068877144196271e-05, "loss": 0.15743472576141357, "step": 90730 }, { "epoch": 0.45071178089268155, "grad_norm": 0.9500768780708313, "learning_rate": 1.2067303208737342e-05, "loss": 0.15598543882369995, "step": 90740 }, { "epoch": 0.4507614515760508, "grad_norm": 0.8535325527191162, "learning_rate": 1.206572921977744e-05, "loss": 0.17739812135696412, "step": 90750 }, { "epoch": 0.45081112225942005, "grad_norm": 1.562631368637085, "learning_rate": 1.2064155177357295e-05, "loss": 0.17488983869552613, "step": 90760 }, { "epoch": 0.4508607929427893, "grad_norm": 0.7658548355102539, "learning_rate": 1.2062581081517643e-05, "loss": 0.1621290922164917, "step": 90770 }, { "epoch": 0.45091046362615855, "grad_norm": 1.0647863149642944, "learning_rate": 1.2061006932299224e-05, "loss": 0.15972049236297609, "step": 90780 }, { "epoch": 0.4509601343095278, "grad_norm": 0.7987766265869141, "learning_rate": 1.2059432729742775e-05, "loss": 0.15455002784729005, "step": 90790 }, { "epoch": 0.4510098049928971, "grad_norm": 1.3297500610351562, "learning_rate": 1.2057858473889032e-05, "loss": 0.17545044422149658, "step": 90800 }, { "epoch": 0.45105947567626636, "grad_norm": 0.8686362504959106, "learning_rate": 1.205628416477874e-05, "loss": 0.15634551048278808, "step": 90810 }, { "epoch": 0.4511091463596356, "grad_norm": 1.312024712562561, "learning_rate": 1.205470980245264e-05, "loss": 0.17662508487701417, "step": 90820 }, { "epoch": 0.45115881704300487, "grad_norm": 1.639502763748169, "learning_rate": 1.205313538695148e-05, "loss": 0.1687960982322693, "step": 90830 }, { "epoch": 0.4512084877263741, "grad_norm": 1.5236297845840454, "learning_rate": 1.2051560918315998e-05, "loss": 0.16951415538787842, "step": 90840 }, { "epoch": 0.4512581584097434, "grad_norm": 1.772942066192627, "learning_rate": 1.2049986396586948e-05, "loss": 0.15250333547592163, "step": 90850 }, { "epoch": 0.4513078290931127, "grad_norm": 2.047346353530884, "learning_rate": 1.2048411821805072e-05, "loss": 0.16851714849472046, "step": 90860 }, { "epoch": 0.45135749977648193, "grad_norm": 2.9743146896362305, "learning_rate": 1.2046837194011119e-05, "loss": 0.1644010066986084, "step": 90870 }, { "epoch": 0.4514071704598512, "grad_norm": 1.003775715827942, "learning_rate": 1.2045262513245846e-05, "loss": 0.16518800258636473, "step": 90880 }, { "epoch": 0.45145684114322043, "grad_norm": 0.920792818069458, "learning_rate": 1.2043687779549999e-05, "loss": 0.17534525394439698, "step": 90890 }, { "epoch": 0.4515065118265897, "grad_norm": 2.9236981868743896, "learning_rate": 1.2042112992964336e-05, "loss": 0.16620175838470458, "step": 90900 }, { "epoch": 0.451556182509959, "grad_norm": 1.458928108215332, "learning_rate": 1.204053815352961e-05, "loss": 0.16235036849975587, "step": 90910 }, { "epoch": 0.45160585319332824, "grad_norm": 1.8968946933746338, "learning_rate": 1.2038963261286577e-05, "loss": 0.1689905047416687, "step": 90920 }, { "epoch": 0.4516555238766975, "grad_norm": 1.476111888885498, "learning_rate": 1.2037388316275995e-05, "loss": 0.15945980548858643, "step": 90930 }, { "epoch": 0.45170519456006675, "grad_norm": 1.2859894037246704, "learning_rate": 1.203581331853862e-05, "loss": 0.17064779996871948, "step": 90940 }, { "epoch": 0.451754865243436, "grad_norm": 1.350769281387329, "learning_rate": 1.2034238268115216e-05, "loss": 0.16761492490768432, "step": 90950 }, { "epoch": 0.4518045359268053, "grad_norm": 1.0393565893173218, "learning_rate": 1.2032663165046545e-05, "loss": 0.14961426258087157, "step": 90960 }, { "epoch": 0.45185420661017456, "grad_norm": 1.8030738830566406, "learning_rate": 1.2031088009373368e-05, "loss": 0.16233887672424316, "step": 90970 }, { "epoch": 0.4519038772935438, "grad_norm": 0.8118807077407837, "learning_rate": 1.2029512801136448e-05, "loss": 0.14424803256988525, "step": 90980 }, { "epoch": 0.45195354797691306, "grad_norm": 2.346454620361328, "learning_rate": 1.2027937540376556e-05, "loss": 0.16180837154388428, "step": 90990 }, { "epoch": 0.4520032186602823, "grad_norm": 2.0235087871551514, "learning_rate": 1.2026362227134456e-05, "loss": 0.17306472063064576, "step": 91000 }, { "epoch": 0.45205288934365156, "grad_norm": 0.9972463250160217, "learning_rate": 1.2024786861450912e-05, "loss": 0.16582744121551513, "step": 91010 }, { "epoch": 0.45210256002702087, "grad_norm": 1.4297900199890137, "learning_rate": 1.2023211443366703e-05, "loss": 0.17681679725646973, "step": 91020 }, { "epoch": 0.4521522307103901, "grad_norm": 1.5084837675094604, "learning_rate": 1.2021635972922593e-05, "loss": 0.15953654050827026, "step": 91030 }, { "epoch": 0.4522019013937594, "grad_norm": 1.38228178024292, "learning_rate": 1.2020060450159357e-05, "loss": 0.16072095632553102, "step": 91040 }, { "epoch": 0.4522515720771286, "grad_norm": 1.2650448083877563, "learning_rate": 1.2018484875117768e-05, "loss": 0.1570263147354126, "step": 91050 }, { "epoch": 0.4523012427604979, "grad_norm": 0.8573407530784607, "learning_rate": 1.2016909247838604e-05, "loss": 0.17835278511047364, "step": 91060 }, { "epoch": 0.4523509134438672, "grad_norm": 0.8758424520492554, "learning_rate": 1.201533356836264e-05, "loss": 0.15823442935943605, "step": 91070 }, { "epoch": 0.45240058412723644, "grad_norm": 0.8068158626556396, "learning_rate": 1.2013757836730649e-05, "loss": 0.17588725090026855, "step": 91080 }, { "epoch": 0.4524502548106057, "grad_norm": 1.351071834564209, "learning_rate": 1.201218205298342e-05, "loss": 0.17926473617553712, "step": 91090 }, { "epoch": 0.45249992549397494, "grad_norm": 1.2410856485366821, "learning_rate": 1.2010606217161724e-05, "loss": 0.15323706865310668, "step": 91100 }, { "epoch": 0.4525495961773442, "grad_norm": 1.3489104509353638, "learning_rate": 1.200903032930635e-05, "loss": 0.15861719846725464, "step": 91110 }, { "epoch": 0.45259926686071345, "grad_norm": 1.3837568759918213, "learning_rate": 1.2007454389458079e-05, "loss": 0.16986026763916015, "step": 91120 }, { "epoch": 0.45264893754408275, "grad_norm": 0.912748396396637, "learning_rate": 1.2005878397657696e-05, "loss": 0.1595984697341919, "step": 91130 }, { "epoch": 0.452698608227452, "grad_norm": 1.5568774938583374, "learning_rate": 1.2004302353945987e-05, "loss": 0.1447008728981018, "step": 91140 }, { "epoch": 0.45274827891082126, "grad_norm": 0.9254654049873352, "learning_rate": 1.2002726258363735e-05, "loss": 0.18518502712249757, "step": 91150 }, { "epoch": 0.4527979495941905, "grad_norm": 1.0250885486602783, "learning_rate": 1.2001150110951737e-05, "loss": 0.16944851875305175, "step": 91160 }, { "epoch": 0.45284762027755976, "grad_norm": 1.200616478919983, "learning_rate": 1.199957391175078e-05, "loss": 0.15567550659179688, "step": 91170 }, { "epoch": 0.45289729096092907, "grad_norm": 2.241908311843872, "learning_rate": 1.1997997660801652e-05, "loss": 0.15580534934997559, "step": 91180 }, { "epoch": 0.4529469616442983, "grad_norm": 1.4806418418884277, "learning_rate": 1.1996421358145147e-05, "loss": 0.17982326745986937, "step": 91190 }, { "epoch": 0.45299663232766757, "grad_norm": 1.3048486709594727, "learning_rate": 1.199484500382206e-05, "loss": 0.16882119178771973, "step": 91200 }, { "epoch": 0.4530463030110368, "grad_norm": 0.6789721846580505, "learning_rate": 1.1993268597873185e-05, "loss": 0.1617209196090698, "step": 91210 }, { "epoch": 0.4530959736944061, "grad_norm": 1.8702024221420288, "learning_rate": 1.1991692140339325e-05, "loss": 0.14752414226531982, "step": 91220 }, { "epoch": 0.4531456443777753, "grad_norm": 1.2469350099563599, "learning_rate": 1.1990115631261268e-05, "loss": 0.15865477323532104, "step": 91230 }, { "epoch": 0.45319531506114463, "grad_norm": 0.8113486766815186, "learning_rate": 1.1988539070679821e-05, "loss": 0.17012360095977783, "step": 91240 }, { "epoch": 0.4532449857445139, "grad_norm": 0.8260590434074402, "learning_rate": 1.1986962458635783e-05, "loss": 0.16343319416046143, "step": 91250 }, { "epoch": 0.45329465642788314, "grad_norm": 0.7648178935050964, "learning_rate": 1.1985385795169954e-05, "loss": 0.15285048484802247, "step": 91260 }, { "epoch": 0.4533443271112524, "grad_norm": 1.117785930633545, "learning_rate": 1.1983809080323142e-05, "loss": 0.1738719940185547, "step": 91270 }, { "epoch": 0.45339399779462164, "grad_norm": 0.7843330502510071, "learning_rate": 1.1982232314136144e-05, "loss": 0.16750513315200805, "step": 91280 }, { "epoch": 0.45344366847799095, "grad_norm": 1.3118114471435547, "learning_rate": 1.1980655496649773e-05, "loss": 0.1521574378013611, "step": 91290 }, { "epoch": 0.4534933391613602, "grad_norm": 1.0056716203689575, "learning_rate": 1.1979078627904834e-05, "loss": 0.16091341972351075, "step": 91300 }, { "epoch": 0.45354300984472945, "grad_norm": 1.0210673809051514, "learning_rate": 1.1977501707942136e-05, "loss": 0.16738660335540773, "step": 91310 }, { "epoch": 0.4535926805280987, "grad_norm": 1.4876770973205566, "learning_rate": 1.1975924736802489e-05, "loss": 0.16308047771453857, "step": 91320 }, { "epoch": 0.45364235121146795, "grad_norm": 0.826450526714325, "learning_rate": 1.1974347714526708e-05, "loss": 0.16794135570526122, "step": 91330 }, { "epoch": 0.4536920218948372, "grad_norm": 1.597608208656311, "learning_rate": 1.1972770641155595e-05, "loss": 0.1491934299468994, "step": 91340 }, { "epoch": 0.4537416925782065, "grad_norm": 0.8871995210647583, "learning_rate": 1.1971193516729975e-05, "loss": 0.15353355407714844, "step": 91350 }, { "epoch": 0.45379136326157576, "grad_norm": 1.44602370262146, "learning_rate": 1.1969616341290659e-05, "loss": 0.16414891481399535, "step": 91360 }, { "epoch": 0.453841033944945, "grad_norm": 0.7597837448120117, "learning_rate": 1.1968039114878465e-05, "loss": 0.16536340713500977, "step": 91370 }, { "epoch": 0.45389070462831427, "grad_norm": 1.1357200145721436, "learning_rate": 1.196646183753421e-05, "loss": 0.1723341941833496, "step": 91380 }, { "epoch": 0.4539403753116835, "grad_norm": 1.1990077495574951, "learning_rate": 1.1964884509298711e-05, "loss": 0.15791447162628175, "step": 91390 }, { "epoch": 0.4539900459950528, "grad_norm": 1.2725285291671753, "learning_rate": 1.1963307130212791e-05, "loss": 0.16247448921203614, "step": 91400 }, { "epoch": 0.4540397166784221, "grad_norm": 1.4661157131195068, "learning_rate": 1.1961729700317273e-05, "loss": 0.15890979766845703, "step": 91410 }, { "epoch": 0.45408938736179133, "grad_norm": 1.4898179769515991, "learning_rate": 1.1960152219652978e-05, "loss": 0.16578915119171142, "step": 91420 }, { "epoch": 0.4541390580451606, "grad_norm": 0.9262014627456665, "learning_rate": 1.1958574688260733e-05, "loss": 0.15477709770202636, "step": 91430 }, { "epoch": 0.45418872872852983, "grad_norm": 1.240031361579895, "learning_rate": 1.1956997106181363e-05, "loss": 0.15813286304473878, "step": 91440 }, { "epoch": 0.4542383994118991, "grad_norm": 1.2723149061203003, "learning_rate": 1.1955419473455694e-05, "loss": 0.16094653606414794, "step": 91450 }, { "epoch": 0.4542880700952684, "grad_norm": 1.1378728151321411, "learning_rate": 1.1953841790124557e-05, "loss": 0.15889179706573486, "step": 91460 }, { "epoch": 0.45433774077863764, "grad_norm": 1.4551708698272705, "learning_rate": 1.1952264056228777e-05, "loss": 0.16938717365264894, "step": 91470 }, { "epoch": 0.4543874114620069, "grad_norm": 2.2473065853118896, "learning_rate": 1.195068627180919e-05, "loss": 0.16693905591964722, "step": 91480 }, { "epoch": 0.45443708214537615, "grad_norm": 1.3710715770721436, "learning_rate": 1.1949108436906625e-05, "loss": 0.17059171199798584, "step": 91490 }, { "epoch": 0.4544867528287454, "grad_norm": 0.9399511218070984, "learning_rate": 1.1947530551561922e-05, "loss": 0.16127386093139648, "step": 91500 }, { "epoch": 0.4545364235121147, "grad_norm": 1.3545138835906982, "learning_rate": 1.1945952615815908e-05, "loss": 0.15822110176086426, "step": 91510 }, { "epoch": 0.45458609419548396, "grad_norm": 1.3322800397872925, "learning_rate": 1.1944374629709424e-05, "loss": 0.16241649389266968, "step": 91520 }, { "epoch": 0.4546357648788532, "grad_norm": 1.4176801443099976, "learning_rate": 1.1942796593283304e-05, "loss": 0.17073556184768676, "step": 91530 }, { "epoch": 0.45468543556222246, "grad_norm": 1.5086668729782104, "learning_rate": 1.1941218506578389e-05, "loss": 0.16729142665863037, "step": 91540 }, { "epoch": 0.4547351062455917, "grad_norm": 0.9242820143699646, "learning_rate": 1.1939640369635521e-05, "loss": 0.17141590118408204, "step": 91550 }, { "epoch": 0.45478477692896097, "grad_norm": 1.0739836692810059, "learning_rate": 1.193806218249554e-05, "loss": 0.1644287586212158, "step": 91560 }, { "epoch": 0.4548344476123303, "grad_norm": 0.6649619340896606, "learning_rate": 1.1936483945199289e-05, "loss": 0.16183282136917115, "step": 91570 }, { "epoch": 0.4548841182956995, "grad_norm": 0.9825329184532166, "learning_rate": 1.1934905657787611e-05, "loss": 0.16829559803009034, "step": 91580 }, { "epoch": 0.4549337889790688, "grad_norm": 1.5977373123168945, "learning_rate": 1.1933327320301355e-05, "loss": 0.17141540050506593, "step": 91590 }, { "epoch": 0.45498345966243803, "grad_norm": 0.7910296320915222, "learning_rate": 1.1931748932781365e-05, "loss": 0.15001894235610963, "step": 91600 }, { "epoch": 0.4550331303458073, "grad_norm": 1.2099111080169678, "learning_rate": 1.193017049526849e-05, "loss": 0.16741797924041749, "step": 91610 }, { "epoch": 0.4550828010291766, "grad_norm": 1.5803394317626953, "learning_rate": 1.1928592007803575e-05, "loss": 0.1565070629119873, "step": 91620 }, { "epoch": 0.45513247171254584, "grad_norm": 0.7086181044578552, "learning_rate": 1.1927013470427476e-05, "loss": 0.16463930606842042, "step": 91630 }, { "epoch": 0.4551821423959151, "grad_norm": 1.2591947317123413, "learning_rate": 1.1925434883181045e-05, "loss": 0.15138388872146608, "step": 91640 }, { "epoch": 0.45523181307928434, "grad_norm": 1.2308977842330933, "learning_rate": 1.1923856246105132e-05, "loss": 0.150721275806427, "step": 91650 }, { "epoch": 0.4552814837626536, "grad_norm": 0.9196967482566833, "learning_rate": 1.1922277559240594e-05, "loss": 0.1679150104522705, "step": 91660 }, { "epoch": 0.45533115444602285, "grad_norm": 0.7229776978492737, "learning_rate": 1.1920698822628285e-05, "loss": 0.16422946453094484, "step": 91670 }, { "epoch": 0.45538082512939215, "grad_norm": 1.9447429180145264, "learning_rate": 1.1919120036309062e-05, "loss": 0.16562280654907227, "step": 91680 }, { "epoch": 0.4554304958127614, "grad_norm": 1.1440293788909912, "learning_rate": 1.1917541200323783e-05, "loss": 0.17902686595916747, "step": 91690 }, { "epoch": 0.45548016649613066, "grad_norm": 0.8894818425178528, "learning_rate": 1.1915962314713311e-05, "loss": 0.1608037233352661, "step": 91700 }, { "epoch": 0.4555298371794999, "grad_norm": 1.0377639532089233, "learning_rate": 1.1914383379518503e-05, "loss": 0.15942094326019288, "step": 91710 }, { "epoch": 0.45557950786286916, "grad_norm": 1.5559247732162476, "learning_rate": 1.1912804394780227e-05, "loss": 0.17834999561309814, "step": 91720 }, { "epoch": 0.4556291785462384, "grad_norm": 1.4801868200302124, "learning_rate": 1.1911225360539337e-05, "loss": 0.14774754047393798, "step": 91730 }, { "epoch": 0.4556788492296077, "grad_norm": 1.6169971227645874, "learning_rate": 1.1909646276836706e-05, "loss": 0.17597423791885375, "step": 91740 }, { "epoch": 0.45572851991297697, "grad_norm": 1.401840090751648, "learning_rate": 1.1908067143713195e-05, "loss": 0.16064774990081787, "step": 91750 }, { "epoch": 0.4557781905963462, "grad_norm": 1.170473337173462, "learning_rate": 1.1906487961209676e-05, "loss": 0.18697667121887207, "step": 91760 }, { "epoch": 0.4558278612797155, "grad_norm": 1.1099584102630615, "learning_rate": 1.1904908729367013e-05, "loss": 0.15259649753570556, "step": 91770 }, { "epoch": 0.4558775319630847, "grad_norm": 1.053623080253601, "learning_rate": 1.1903329448226079e-05, "loss": 0.16550198793411255, "step": 91780 }, { "epoch": 0.45592720264645403, "grad_norm": 1.1936426162719727, "learning_rate": 1.1901750117827746e-05, "loss": 0.1612485647201538, "step": 91790 }, { "epoch": 0.4559768733298233, "grad_norm": 0.8835188746452332, "learning_rate": 1.1900170738212884e-05, "loss": 0.16572561264038085, "step": 91800 }, { "epoch": 0.45602654401319254, "grad_norm": 1.001993179321289, "learning_rate": 1.1898591309422364e-05, "loss": 0.15828851461410523, "step": 91810 }, { "epoch": 0.4560762146965618, "grad_norm": 1.0985544919967651, "learning_rate": 1.1897011831497067e-05, "loss": 0.16857196092605592, "step": 91820 }, { "epoch": 0.45612588537993104, "grad_norm": 1.2143646478652954, "learning_rate": 1.1895432304477865e-05, "loss": 0.1611107349395752, "step": 91830 }, { "epoch": 0.4561755560633003, "grad_norm": 0.7866697907447815, "learning_rate": 1.1893852728405641e-05, "loss": 0.17057251930236816, "step": 91840 }, { "epoch": 0.4562252267466696, "grad_norm": 1.640649676322937, "learning_rate": 1.1892273103321268e-05, "loss": 0.15521538257598877, "step": 91850 }, { "epoch": 0.45627489743003885, "grad_norm": 1.3303916454315186, "learning_rate": 1.1890693429265627e-05, "loss": 0.16627421379089355, "step": 91860 }, { "epoch": 0.4563245681134081, "grad_norm": 3.719351291656494, "learning_rate": 1.18891137062796e-05, "loss": 0.170617938041687, "step": 91870 }, { "epoch": 0.45637423879677735, "grad_norm": 1.1454417705535889, "learning_rate": 1.188753393440407e-05, "loss": 0.15629210472106933, "step": 91880 }, { "epoch": 0.4564239094801466, "grad_norm": 0.8008795976638794, "learning_rate": 1.1885954113679921e-05, "loss": 0.17091732025146483, "step": 91890 }, { "epoch": 0.4564735801635159, "grad_norm": 1.1230732202529907, "learning_rate": 1.1884374244148038e-05, "loss": 0.15427145957946778, "step": 91900 }, { "epoch": 0.45652325084688516, "grad_norm": 0.6501005291938782, "learning_rate": 1.1882794325849307e-05, "loss": 0.17638187408447265, "step": 91910 }, { "epoch": 0.4565729215302544, "grad_norm": 1.5724385976791382, "learning_rate": 1.1881214358824616e-05, "loss": 0.1573173999786377, "step": 91920 }, { "epoch": 0.45662259221362367, "grad_norm": 1.5540579557418823, "learning_rate": 1.1879634343114855e-05, "loss": 0.15440608263015748, "step": 91930 }, { "epoch": 0.4566722628969929, "grad_norm": 1.039074420928955, "learning_rate": 1.1878054278760911e-05, "loss": 0.1780555486679077, "step": 91940 }, { "epoch": 0.45672193358036217, "grad_norm": 1.004051923751831, "learning_rate": 1.1876474165803675e-05, "loss": 0.16191656589508058, "step": 91950 }, { "epoch": 0.4567716042637315, "grad_norm": 1.648556113243103, "learning_rate": 1.1874894004284046e-05, "loss": 0.15165228843688966, "step": 91960 }, { "epoch": 0.45682127494710073, "grad_norm": 1.2605140209197998, "learning_rate": 1.1873313794242914e-05, "loss": 0.18157399892807008, "step": 91970 }, { "epoch": 0.45687094563047, "grad_norm": 1.364087462425232, "learning_rate": 1.1871733535721172e-05, "loss": 0.17128840684890748, "step": 91980 }, { "epoch": 0.45692061631383923, "grad_norm": 1.3557854890823364, "learning_rate": 1.1870153228759721e-05, "loss": 0.17202680110931395, "step": 91990 }, { "epoch": 0.4569702869972085, "grad_norm": 1.2441474199295044, "learning_rate": 1.1868572873399455e-05, "loss": 0.15402395725250245, "step": 92000 }, { "epoch": 0.4570199576805778, "grad_norm": 1.0795092582702637, "learning_rate": 1.1866992469681278e-05, "loss": 0.17963716983795167, "step": 92010 }, { "epoch": 0.45706962836394704, "grad_norm": 1.7599838972091675, "learning_rate": 1.1865412017646083e-05, "loss": 0.15948305130004883, "step": 92020 }, { "epoch": 0.4571192990473163, "grad_norm": 1.3556547164916992, "learning_rate": 1.1863831517334776e-05, "loss": 0.16012189388275147, "step": 92030 }, { "epoch": 0.45716896973068555, "grad_norm": 1.983087182044983, "learning_rate": 1.1862250968788258e-05, "loss": 0.1560386061668396, "step": 92040 }, { "epoch": 0.4572186404140548, "grad_norm": 1.1544228792190552, "learning_rate": 1.1860670372047434e-05, "loss": 0.1706146478652954, "step": 92050 }, { "epoch": 0.45726831109742405, "grad_norm": 0.86232990026474, "learning_rate": 1.1859089727153207e-05, "loss": 0.15572150945663452, "step": 92060 }, { "epoch": 0.45731798178079336, "grad_norm": 1.1561176776885986, "learning_rate": 1.1857509034146491e-05, "loss": 0.17345646619796753, "step": 92070 }, { "epoch": 0.4573676524641626, "grad_norm": 0.9146047234535217, "learning_rate": 1.1855928293068181e-05, "loss": 0.16456393003463746, "step": 92080 }, { "epoch": 0.45741732314753186, "grad_norm": 1.100392460823059, "learning_rate": 1.1854347503959196e-05, "loss": 0.1730044364929199, "step": 92090 }, { "epoch": 0.4574669938309011, "grad_norm": 1.983296513557434, "learning_rate": 1.1852766666860445e-05, "loss": 0.17131956815719604, "step": 92100 }, { "epoch": 0.45751666451427037, "grad_norm": 1.7788915634155273, "learning_rate": 1.1851185781812835e-05, "loss": 0.16151515245437623, "step": 92110 }, { "epoch": 0.4575663351976397, "grad_norm": 1.9296938180923462, "learning_rate": 1.1849604848857284e-05, "loss": 0.16623660326004028, "step": 92120 }, { "epoch": 0.4576160058810089, "grad_norm": 1.1530734300613403, "learning_rate": 1.1848023868034701e-05, "loss": 0.15611345767974855, "step": 92130 }, { "epoch": 0.4576656765643782, "grad_norm": 0.9582672715187073, "learning_rate": 1.1846442839386004e-05, "loss": 0.15827631950378418, "step": 92140 }, { "epoch": 0.45771534724774743, "grad_norm": 1.6237263679504395, "learning_rate": 1.1844861762952109e-05, "loss": 0.16692831516265869, "step": 92150 }, { "epoch": 0.4577650179311167, "grad_norm": 1.4554435014724731, "learning_rate": 1.1843280638773931e-05, "loss": 0.16729891300201416, "step": 92160 }, { "epoch": 0.45781468861448593, "grad_norm": 0.7723685503005981, "learning_rate": 1.1841699466892393e-05, "loss": 0.15794651508331298, "step": 92170 }, { "epoch": 0.45786435929785524, "grad_norm": 1.8357818126678467, "learning_rate": 1.1840118247348418e-05, "loss": 0.1635746717453003, "step": 92180 }, { "epoch": 0.4579140299812245, "grad_norm": 1.6041595935821533, "learning_rate": 1.1838536980182917e-05, "loss": 0.16407625675201415, "step": 92190 }, { "epoch": 0.45796370066459374, "grad_norm": 1.1882052421569824, "learning_rate": 1.183695566543682e-05, "loss": 0.1635773301124573, "step": 92200 }, { "epoch": 0.458013371347963, "grad_norm": 1.7584253549575806, "learning_rate": 1.1835374303151048e-05, "loss": 0.1503279685974121, "step": 92210 }, { "epoch": 0.45806304203133225, "grad_norm": 1.2649128437042236, "learning_rate": 1.1833792893366527e-05, "loss": 0.15368301868438722, "step": 92220 }, { "epoch": 0.45811271271470155, "grad_norm": 1.5363247394561768, "learning_rate": 1.1832211436124183e-05, "loss": 0.16268070936203002, "step": 92230 }, { "epoch": 0.4581623833980708, "grad_norm": 1.2394241094589233, "learning_rate": 1.1830629931464945e-05, "loss": 0.16337933540344238, "step": 92240 }, { "epoch": 0.45821205408144006, "grad_norm": 1.2470996379852295, "learning_rate": 1.182904837942974e-05, "loss": 0.16530978679656982, "step": 92250 }, { "epoch": 0.4582617247648093, "grad_norm": 0.896537184715271, "learning_rate": 1.1827466780059497e-05, "loss": 0.16909217834472656, "step": 92260 }, { "epoch": 0.45831139544817856, "grad_norm": 1.3453495502471924, "learning_rate": 1.1825885133395152e-05, "loss": 0.1574156641960144, "step": 92270 }, { "epoch": 0.4583610661315478, "grad_norm": 0.8284351229667664, "learning_rate": 1.1824303439477632e-05, "loss": 0.15918073654174805, "step": 92280 }, { "epoch": 0.4584107368149171, "grad_norm": 0.9415052533149719, "learning_rate": 1.182272169834787e-05, "loss": 0.1581141710281372, "step": 92290 }, { "epoch": 0.45846040749828637, "grad_norm": 0.6805582642555237, "learning_rate": 1.1821139910046806e-05, "loss": 0.1622767925262451, "step": 92300 }, { "epoch": 0.4585100781816556, "grad_norm": 1.0321977138519287, "learning_rate": 1.1819558074615374e-05, "loss": 0.16190414428710936, "step": 92310 }, { "epoch": 0.4585597488650249, "grad_norm": 1.5624661445617676, "learning_rate": 1.181797619209451e-05, "loss": 0.15944066047668456, "step": 92320 }, { "epoch": 0.4586094195483941, "grad_norm": 0.952174961566925, "learning_rate": 1.1816394262525153e-05, "loss": 0.15453397035598754, "step": 92330 }, { "epoch": 0.45865909023176343, "grad_norm": 1.621055245399475, "learning_rate": 1.181481228594824e-05, "loss": 0.16259756088256835, "step": 92340 }, { "epoch": 0.4587087609151327, "grad_norm": 1.4809346199035645, "learning_rate": 1.1813230262404718e-05, "loss": 0.1686167001724243, "step": 92350 }, { "epoch": 0.45875843159850194, "grad_norm": 1.9569085836410522, "learning_rate": 1.1811648191935525e-05, "loss": 0.16992684602737426, "step": 92360 }, { "epoch": 0.4588081022818712, "grad_norm": 0.7575348615646362, "learning_rate": 1.1810066074581608e-05, "loss": 0.15963075160980225, "step": 92370 }, { "epoch": 0.45885777296524044, "grad_norm": 0.9979701638221741, "learning_rate": 1.1808483910383906e-05, "loss": 0.16549232006072997, "step": 92380 }, { "epoch": 0.4589074436486097, "grad_norm": 1.1292617321014404, "learning_rate": 1.1806901699383366e-05, "loss": 0.16108686923980714, "step": 92390 }, { "epoch": 0.458957114331979, "grad_norm": 1.8110853433609009, "learning_rate": 1.180531944162094e-05, "loss": 0.16353363990783693, "step": 92400 }, { "epoch": 0.45900678501534825, "grad_norm": 0.9243302941322327, "learning_rate": 1.180373713713757e-05, "loss": 0.17374352216720582, "step": 92410 }, { "epoch": 0.4590564556987175, "grad_norm": 1.6998388767242432, "learning_rate": 1.1802154785974209e-05, "loss": 0.17133746147155762, "step": 92420 }, { "epoch": 0.45910612638208675, "grad_norm": 0.6940227746963501, "learning_rate": 1.1800572388171807e-05, "loss": 0.16618233919143677, "step": 92430 }, { "epoch": 0.459155797065456, "grad_norm": 1.5484347343444824, "learning_rate": 1.1798989943771315e-05, "loss": 0.17838587760925292, "step": 92440 }, { "epoch": 0.4592054677488253, "grad_norm": 1.647365689277649, "learning_rate": 1.1797407452813687e-05, "loss": 0.16227321624755858, "step": 92450 }, { "epoch": 0.45925513843219457, "grad_norm": 1.6299086809158325, "learning_rate": 1.1795824915339877e-05, "loss": 0.1646937608718872, "step": 92460 }, { "epoch": 0.4593048091155638, "grad_norm": 0.5993409156799316, "learning_rate": 1.1794242331390842e-05, "loss": 0.170721173286438, "step": 92470 }, { "epoch": 0.45935447979893307, "grad_norm": 0.8540219664573669, "learning_rate": 1.1792659701007534e-05, "loss": 0.16576656103134155, "step": 92480 }, { "epoch": 0.4594041504823023, "grad_norm": 1.381230354309082, "learning_rate": 1.1791077024230914e-05, "loss": 0.16651310920715331, "step": 92490 }, { "epoch": 0.45945382116567157, "grad_norm": 0.9256967306137085, "learning_rate": 1.178949430110194e-05, "loss": 0.15500280857086182, "step": 92500 }, { "epoch": 0.4595034918490409, "grad_norm": 1.3322415351867676, "learning_rate": 1.1787911531661576e-05, "loss": 0.15310869216918946, "step": 92510 }, { "epoch": 0.45955316253241013, "grad_norm": 1.4412399530410767, "learning_rate": 1.1786328715950781e-05, "loss": 0.15400524139404298, "step": 92520 }, { "epoch": 0.4596028332157794, "grad_norm": 1.2270337343215942, "learning_rate": 1.1784745854010516e-05, "loss": 0.15657061338424683, "step": 92530 }, { "epoch": 0.45965250389914863, "grad_norm": 1.2326730489730835, "learning_rate": 1.1783162945881745e-05, "loss": 0.17078551054000854, "step": 92540 }, { "epoch": 0.4597021745825179, "grad_norm": 1.5665923357009888, "learning_rate": 1.1781579991605432e-05, "loss": 0.16105151176452637, "step": 92550 }, { "epoch": 0.4597518452658872, "grad_norm": 0.9480689167976379, "learning_rate": 1.1779996991222546e-05, "loss": 0.17364835739135742, "step": 92560 }, { "epoch": 0.45980151594925645, "grad_norm": 1.2688556909561157, "learning_rate": 1.1778413944774057e-05, "loss": 0.16824990510940552, "step": 92570 }, { "epoch": 0.4598511866326257, "grad_norm": 1.0555304288864136, "learning_rate": 1.1776830852300927e-05, "loss": 0.16739914417266846, "step": 92580 }, { "epoch": 0.45990085731599495, "grad_norm": 0.9525240659713745, "learning_rate": 1.1775247713844129e-05, "loss": 0.1673598051071167, "step": 92590 }, { "epoch": 0.4599505279993642, "grad_norm": 0.8197140097618103, "learning_rate": 1.1773664529444636e-05, "loss": 0.16144249439239503, "step": 92600 }, { "epoch": 0.46000019868273345, "grad_norm": 0.802850067615509, "learning_rate": 1.1772081299143419e-05, "loss": 0.15565381050109864, "step": 92610 }, { "epoch": 0.46004986936610276, "grad_norm": 1.0898635387420654, "learning_rate": 1.1770498022981447e-05, "loss": 0.17084064483642578, "step": 92620 }, { "epoch": 0.460099540049472, "grad_norm": 1.31430983543396, "learning_rate": 1.1768914700999698e-05, "loss": 0.17175331115722656, "step": 92630 }, { "epoch": 0.46014921073284126, "grad_norm": 1.7882802486419678, "learning_rate": 1.176733133323915e-05, "loss": 0.14893494844436644, "step": 92640 }, { "epoch": 0.4601988814162105, "grad_norm": 1.2594170570373535, "learning_rate": 1.1765747919740775e-05, "loss": 0.16221213340759277, "step": 92650 }, { "epoch": 0.46024855209957977, "grad_norm": 1.9373213052749634, "learning_rate": 1.1764164460545557e-05, "loss": 0.1582359790802002, "step": 92660 }, { "epoch": 0.4602982227829491, "grad_norm": 0.9671627879142761, "learning_rate": 1.176258095569447e-05, "loss": 0.16048035621643067, "step": 92670 }, { "epoch": 0.4603478934663183, "grad_norm": 0.8253730535507202, "learning_rate": 1.17609974052285e-05, "loss": 0.15272166728973388, "step": 92680 }, { "epoch": 0.4603975641496876, "grad_norm": 0.9161152243614197, "learning_rate": 1.1759413809188622e-05, "loss": 0.1649368166923523, "step": 92690 }, { "epoch": 0.46044723483305683, "grad_norm": 0.8561058044433594, "learning_rate": 1.1757830167615824e-05, "loss": 0.16484827995300294, "step": 92700 }, { "epoch": 0.4604969055164261, "grad_norm": 1.4819992780685425, "learning_rate": 1.1756246480551085e-05, "loss": 0.15983188152313232, "step": 92710 }, { "epoch": 0.46054657619979533, "grad_norm": 1.6778100728988647, "learning_rate": 1.1754662748035398e-05, "loss": 0.16505913734436034, "step": 92720 }, { "epoch": 0.46059624688316464, "grad_norm": 1.3540383577346802, "learning_rate": 1.175307897010974e-05, "loss": 0.16419610977172852, "step": 92730 }, { "epoch": 0.4606459175665339, "grad_norm": 1.9224762916564941, "learning_rate": 1.1751495146815107e-05, "loss": 0.16637701988220216, "step": 92740 }, { "epoch": 0.46069558824990314, "grad_norm": 1.8548349142074585, "learning_rate": 1.174991127819248e-05, "loss": 0.17756277322769165, "step": 92750 }, { "epoch": 0.4607452589332724, "grad_norm": 1.0868052244186401, "learning_rate": 1.1748327364282854e-05, "loss": 0.17379791736602784, "step": 92760 }, { "epoch": 0.46079492961664165, "grad_norm": 0.9978454113006592, "learning_rate": 1.1746743405127219e-05, "loss": 0.1746151089668274, "step": 92770 }, { "epoch": 0.46084460030001095, "grad_norm": 1.079611897468567, "learning_rate": 1.1745159400766566e-05, "loss": 0.15994046926498412, "step": 92780 }, { "epoch": 0.4608942709833802, "grad_norm": 0.6128813624382019, "learning_rate": 1.1743575351241892e-05, "loss": 0.17183852195739746, "step": 92790 }, { "epoch": 0.46094394166674946, "grad_norm": 1.5188723802566528, "learning_rate": 1.1741991256594187e-05, "loss": 0.17461401224136353, "step": 92800 }, { "epoch": 0.4609936123501187, "grad_norm": 1.3289427757263184, "learning_rate": 1.1740407116864449e-05, "loss": 0.16767668724060059, "step": 92810 }, { "epoch": 0.46104328303348796, "grad_norm": 1.4109405279159546, "learning_rate": 1.1738822932093677e-05, "loss": 0.16462444067001342, "step": 92820 }, { "epoch": 0.4610929537168572, "grad_norm": 1.1898990869522095, "learning_rate": 1.1737238702322861e-05, "loss": 0.17677013874053954, "step": 92830 }, { "epoch": 0.4611426244002265, "grad_norm": 0.7542949914932251, "learning_rate": 1.173565442759301e-05, "loss": 0.1589712142944336, "step": 92840 }, { "epoch": 0.46119229508359577, "grad_norm": 2.283087730407715, "learning_rate": 1.1734070107945123e-05, "loss": 0.17799854278564453, "step": 92850 }, { "epoch": 0.461241965766965, "grad_norm": 2.198861598968506, "learning_rate": 1.1732485743420195e-05, "loss": 0.16249244213104247, "step": 92860 }, { "epoch": 0.4612916364503343, "grad_norm": 2.0337374210357666, "learning_rate": 1.1730901334059234e-05, "loss": 0.16437896490097045, "step": 92870 }, { "epoch": 0.4613413071337035, "grad_norm": 0.8856660723686218, "learning_rate": 1.1729316879903242e-05, "loss": 0.17883902788162231, "step": 92880 }, { "epoch": 0.46139097781707283, "grad_norm": 0.7672285437583923, "learning_rate": 1.1727732380993221e-05, "loss": 0.15975866317749024, "step": 92890 }, { "epoch": 0.4614406485004421, "grad_norm": 1.0810950994491577, "learning_rate": 1.1726147837370185e-05, "loss": 0.17131236791610718, "step": 92900 }, { "epoch": 0.46149031918381134, "grad_norm": 2.6887753009796143, "learning_rate": 1.1724563249075136e-05, "loss": 0.18134527206420897, "step": 92910 }, { "epoch": 0.4615399898671806, "grad_norm": 1.5009965896606445, "learning_rate": 1.1722978616149083e-05, "loss": 0.16560752391815187, "step": 92920 }, { "epoch": 0.46158966055054984, "grad_norm": 1.559911847114563, "learning_rate": 1.1721393938633037e-05, "loss": 0.16863759756088256, "step": 92930 }, { "epoch": 0.4616393312339191, "grad_norm": 1.4652347564697266, "learning_rate": 1.1719809216568007e-05, "loss": 0.16667401790618896, "step": 92940 }, { "epoch": 0.4616890019172884, "grad_norm": 1.263857126235962, "learning_rate": 1.1718224449995008e-05, "loss": 0.15786371231079102, "step": 92950 }, { "epoch": 0.46173867260065765, "grad_norm": 0.9320747256278992, "learning_rate": 1.1716639638955047e-05, "loss": 0.15423250198364258, "step": 92960 }, { "epoch": 0.4617883432840269, "grad_norm": 0.8429988622665405, "learning_rate": 1.1715054783489145e-05, "loss": 0.17572295665740967, "step": 92970 }, { "epoch": 0.46183801396739615, "grad_norm": 1.9303585290908813, "learning_rate": 1.1713469883638315e-05, "loss": 0.16460872888565065, "step": 92980 }, { "epoch": 0.4618876846507654, "grad_norm": 1.6952853202819824, "learning_rate": 1.1711884939443574e-05, "loss": 0.1572197198867798, "step": 92990 }, { "epoch": 0.4619373553341347, "grad_norm": 1.2963277101516724, "learning_rate": 1.1710299950945939e-05, "loss": 0.1670478105545044, "step": 93000 }, { "epoch": 0.46198702601750397, "grad_norm": 1.5030068159103394, "learning_rate": 1.1708714918186427e-05, "loss": 0.15871355533599854, "step": 93010 }, { "epoch": 0.4620366967008732, "grad_norm": 2.285431385040283, "learning_rate": 1.1707129841206061e-05, "loss": 0.15414979457855224, "step": 93020 }, { "epoch": 0.46208636738424247, "grad_norm": 1.2617141008377075, "learning_rate": 1.1705544720045863e-05, "loss": 0.17082600593566893, "step": 93030 }, { "epoch": 0.4621360380676117, "grad_norm": 0.8440439701080322, "learning_rate": 1.170395955474685e-05, "loss": 0.16475765705108641, "step": 93040 }, { "epoch": 0.462185708750981, "grad_norm": 1.2171615362167358, "learning_rate": 1.1702374345350051e-05, "loss": 0.16783865690231323, "step": 93050 }, { "epoch": 0.4622353794343503, "grad_norm": 0.7673376202583313, "learning_rate": 1.170078909189649e-05, "loss": 0.1671091914176941, "step": 93060 }, { "epoch": 0.46228505011771953, "grad_norm": 1.655257225036621, "learning_rate": 1.169920379442719e-05, "loss": 0.17559466361999512, "step": 93070 }, { "epoch": 0.4623347208010888, "grad_norm": 0.6578161716461182, "learning_rate": 1.1697618452983177e-05, "loss": 0.1561190128326416, "step": 93080 }, { "epoch": 0.46238439148445803, "grad_norm": 0.9973415732383728, "learning_rate": 1.169603306760548e-05, "loss": 0.1716272234916687, "step": 93090 }, { "epoch": 0.4624340621678273, "grad_norm": 0.809522271156311, "learning_rate": 1.1694447638335132e-05, "loss": 0.16148922443389893, "step": 93100 }, { "epoch": 0.4624837328511966, "grad_norm": 1.2781741619110107, "learning_rate": 1.1692862165213162e-05, "loss": 0.1595378637313843, "step": 93110 }, { "epoch": 0.46253340353456585, "grad_norm": 1.1954766511917114, "learning_rate": 1.1691276648280597e-05, "loss": 0.16793723106384278, "step": 93120 }, { "epoch": 0.4625830742179351, "grad_norm": 0.7297605872154236, "learning_rate": 1.1689691087578471e-05, "loss": 0.17356657981872559, "step": 93130 }, { "epoch": 0.46263274490130435, "grad_norm": 1.6688885688781738, "learning_rate": 1.1688105483147823e-05, "loss": 0.16099193096160888, "step": 93140 }, { "epoch": 0.4626824155846736, "grad_norm": 0.7531397342681885, "learning_rate": 1.1686519835029682e-05, "loss": 0.16005489826202393, "step": 93150 }, { "epoch": 0.46273208626804285, "grad_norm": 0.7658840417861938, "learning_rate": 1.1684934143265086e-05, "loss": 0.1730471134185791, "step": 93160 }, { "epoch": 0.46278175695141216, "grad_norm": 0.8667909502983093, "learning_rate": 1.168334840789507e-05, "loss": 0.17142913341522217, "step": 93170 }, { "epoch": 0.4628314276347814, "grad_norm": 1.8097680807113647, "learning_rate": 1.1681762628960676e-05, "loss": 0.18444222211837769, "step": 93180 }, { "epoch": 0.46288109831815066, "grad_norm": 1.5678253173828125, "learning_rate": 1.1680176806502942e-05, "loss": 0.1564786434173584, "step": 93190 }, { "epoch": 0.4629307690015199, "grad_norm": 1.3009955883026123, "learning_rate": 1.1678590940562908e-05, "loss": 0.16697486639022827, "step": 93200 }, { "epoch": 0.46298043968488917, "grad_norm": 0.9550512433052063, "learning_rate": 1.1677005031181614e-05, "loss": 0.17078245878219606, "step": 93210 }, { "epoch": 0.4630301103682585, "grad_norm": 1.5922857522964478, "learning_rate": 1.1675419078400101e-05, "loss": 0.16668202877044677, "step": 93220 }, { "epoch": 0.4630797810516277, "grad_norm": 0.7111814022064209, "learning_rate": 1.167383308225942e-05, "loss": 0.16827534437179564, "step": 93230 }, { "epoch": 0.463129451734997, "grad_norm": 1.0837256908416748, "learning_rate": 1.167224704280061e-05, "loss": 0.17386300563812257, "step": 93240 }, { "epoch": 0.46317912241836623, "grad_norm": 0.9504644870758057, "learning_rate": 1.167066096006472e-05, "loss": 0.15996254682540895, "step": 93250 }, { "epoch": 0.4632287931017355, "grad_norm": 1.8896231651306152, "learning_rate": 1.1669074834092796e-05, "loss": 0.18422415256500244, "step": 93260 }, { "epoch": 0.46327846378510473, "grad_norm": 1.0044447183609009, "learning_rate": 1.1667488664925888e-05, "loss": 0.15740185976028442, "step": 93270 }, { "epoch": 0.46332813446847404, "grad_norm": 0.8828732371330261, "learning_rate": 1.1665902452605042e-05, "loss": 0.16351476907730103, "step": 93280 }, { "epoch": 0.4633778051518433, "grad_norm": 1.1263960599899292, "learning_rate": 1.1664316197171313e-05, "loss": 0.15887928009033203, "step": 93290 }, { "epoch": 0.46342747583521254, "grad_norm": 0.9199638366699219, "learning_rate": 1.1662729898665743e-05, "loss": 0.17941195964813234, "step": 93300 }, { "epoch": 0.4634771465185818, "grad_norm": 1.0507092475891113, "learning_rate": 1.1661143557129397e-05, "loss": 0.15520002841949462, "step": 93310 }, { "epoch": 0.46352681720195105, "grad_norm": 0.9645573496818542, "learning_rate": 1.1659557172603324e-05, "loss": 0.17292194366455077, "step": 93320 }, { "epoch": 0.4635764878853203, "grad_norm": 1.1493699550628662, "learning_rate": 1.1657970745128577e-05, "loss": 0.17517514228820802, "step": 93330 }, { "epoch": 0.4636261585686896, "grad_norm": 1.0206495523452759, "learning_rate": 1.1656384274746216e-05, "loss": 0.17134439945220947, "step": 93340 }, { "epoch": 0.46367582925205886, "grad_norm": 2.3423008918762207, "learning_rate": 1.1654797761497294e-05, "loss": 0.1705751895904541, "step": 93350 }, { "epoch": 0.4637254999354281, "grad_norm": 0.8311551213264465, "learning_rate": 1.1653211205422871e-05, "loss": 0.16050403118133544, "step": 93360 }, { "epoch": 0.46377517061879736, "grad_norm": 1.4207377433776855, "learning_rate": 1.165162460656401e-05, "loss": 0.16145861148834229, "step": 93370 }, { "epoch": 0.4638248413021666, "grad_norm": 0.808433473110199, "learning_rate": 1.1650037964961765e-05, "loss": 0.15962307453155516, "step": 93380 }, { "epoch": 0.4638745119855359, "grad_norm": 1.0237855911254883, "learning_rate": 1.1648451280657203e-05, "loss": 0.17473143339157104, "step": 93390 }, { "epoch": 0.46392418266890517, "grad_norm": 1.429242730140686, "learning_rate": 1.1646864553691384e-05, "loss": 0.14414347410202027, "step": 93400 }, { "epoch": 0.4639738533522744, "grad_norm": 0.8851934671401978, "learning_rate": 1.1645277784105371e-05, "loss": 0.1715775728225708, "step": 93410 }, { "epoch": 0.4640235240356437, "grad_norm": 0.8769072890281677, "learning_rate": 1.1643690971940232e-05, "loss": 0.15718717575073243, "step": 93420 }, { "epoch": 0.4640731947190129, "grad_norm": 1.6430585384368896, "learning_rate": 1.164210411723703e-05, "loss": 0.1658865213394165, "step": 93430 }, { "epoch": 0.4641228654023822, "grad_norm": 2.4435853958129883, "learning_rate": 1.1640517220036833e-05, "loss": 0.15564178228378295, "step": 93440 }, { "epoch": 0.4641725360857515, "grad_norm": 0.8123533129692078, "learning_rate": 1.1638930280380711e-05, "loss": 0.1697877526283264, "step": 93450 }, { "epoch": 0.46422220676912074, "grad_norm": 0.9680718183517456, "learning_rate": 1.1637343298309734e-05, "loss": 0.16664557456970214, "step": 93460 }, { "epoch": 0.46427187745249, "grad_norm": 1.1317262649536133, "learning_rate": 1.163575627386497e-05, "loss": 0.16601240634918213, "step": 93470 }, { "epoch": 0.46432154813585924, "grad_norm": 1.0385795831680298, "learning_rate": 1.1634169207087492e-05, "loss": 0.15987744331359863, "step": 93480 }, { "epoch": 0.4643712188192285, "grad_norm": 0.8327942490577698, "learning_rate": 1.163258209801837e-05, "loss": 0.15606944561004638, "step": 93490 }, { "epoch": 0.4644208895025978, "grad_norm": 0.7360208034515381, "learning_rate": 1.1630994946698679e-05, "loss": 0.16337802410125732, "step": 93500 }, { "epoch": 0.46447056018596705, "grad_norm": 1.3541781902313232, "learning_rate": 1.1629407753169498e-05, "loss": 0.1496181845664978, "step": 93510 }, { "epoch": 0.4645202308693363, "grad_norm": 1.6451469659805298, "learning_rate": 1.1627820517471898e-05, "loss": 0.16513671875, "step": 93520 }, { "epoch": 0.46456990155270556, "grad_norm": 1.8537654876708984, "learning_rate": 1.162623323964696e-05, "loss": 0.16734888553619384, "step": 93530 }, { "epoch": 0.4646195722360748, "grad_norm": 0.7546082139015198, "learning_rate": 1.1624645919735756e-05, "loss": 0.16784982681274413, "step": 93540 }, { "epoch": 0.46466924291944406, "grad_norm": 1.072683572769165, "learning_rate": 1.162305855777937e-05, "loss": 0.16023322343826293, "step": 93550 }, { "epoch": 0.46471891360281337, "grad_norm": 0.8787374496459961, "learning_rate": 1.1621471153818879e-05, "loss": 0.1659226894378662, "step": 93560 }, { "epoch": 0.4647685842861826, "grad_norm": 0.8221102356910706, "learning_rate": 1.161988370789537e-05, "loss": 0.16186947822570802, "step": 93570 }, { "epoch": 0.46481825496955187, "grad_norm": 1.9293299913406372, "learning_rate": 1.1618296220049921e-05, "loss": 0.1702728509902954, "step": 93580 }, { "epoch": 0.4648679256529211, "grad_norm": 1.1750038862228394, "learning_rate": 1.1616708690323614e-05, "loss": 0.15403751134872437, "step": 93590 }, { "epoch": 0.4649175963362904, "grad_norm": 0.8297878503799438, "learning_rate": 1.161512111875754e-05, "loss": 0.17119169235229492, "step": 93600 }, { "epoch": 0.4649672670196597, "grad_norm": 1.2538697719573975, "learning_rate": 1.1613533505392777e-05, "loss": 0.1649700880050659, "step": 93610 }, { "epoch": 0.46501693770302893, "grad_norm": 0.9390158653259277, "learning_rate": 1.1611945850270418e-05, "loss": 0.1601642370223999, "step": 93620 }, { "epoch": 0.4650666083863982, "grad_norm": 1.7430627346038818, "learning_rate": 1.1610358153431546e-05, "loss": 0.1586564302444458, "step": 93630 }, { "epoch": 0.46511627906976744, "grad_norm": 1.1860575675964355, "learning_rate": 1.1608770414917257e-05, "loss": 0.1565619945526123, "step": 93640 }, { "epoch": 0.4651659497531367, "grad_norm": 0.9192205667495728, "learning_rate": 1.1607182634768634e-05, "loss": 0.15237858295440673, "step": 93650 }, { "epoch": 0.46521562043650594, "grad_norm": 1.0073665380477905, "learning_rate": 1.160559481302677e-05, "loss": 0.15896222591400147, "step": 93660 }, { "epoch": 0.46526529111987525, "grad_norm": 1.0357481241226196, "learning_rate": 1.160400694973276e-05, "loss": 0.16043517589569092, "step": 93670 }, { "epoch": 0.4653149618032445, "grad_norm": 0.8549625873565674, "learning_rate": 1.1602419044927693e-05, "loss": 0.1654687523841858, "step": 93680 }, { "epoch": 0.46536463248661375, "grad_norm": 1.1408131122589111, "learning_rate": 1.1600831098652668e-05, "loss": 0.15786395072937012, "step": 93690 }, { "epoch": 0.465414303169983, "grad_norm": 1.5871055126190186, "learning_rate": 1.1599243110948775e-05, "loss": 0.175519859790802, "step": 93700 }, { "epoch": 0.46546397385335225, "grad_norm": 1.2222851514816284, "learning_rate": 1.1597655081857116e-05, "loss": 0.17497878074645995, "step": 93710 }, { "epoch": 0.46551364453672156, "grad_norm": 1.338827133178711, "learning_rate": 1.1596067011418784e-05, "loss": 0.15772929191589355, "step": 93720 }, { "epoch": 0.4655633152200908, "grad_norm": 2.1213431358337402, "learning_rate": 1.159447889967488e-05, "loss": 0.16157552003860473, "step": 93730 }, { "epoch": 0.46561298590346006, "grad_norm": 1.757469892501831, "learning_rate": 1.1592890746666504e-05, "loss": 0.1793049693107605, "step": 93740 }, { "epoch": 0.4656626565868293, "grad_norm": 2.0736706256866455, "learning_rate": 1.1591302552434753e-05, "loss": 0.16003454923629762, "step": 93750 }, { "epoch": 0.46571232727019857, "grad_norm": 0.9430372714996338, "learning_rate": 1.1589714317020732e-05, "loss": 0.15280603170394896, "step": 93760 }, { "epoch": 0.4657619979535678, "grad_norm": 1.187311053276062, "learning_rate": 1.1588126040465547e-05, "loss": 0.16205382347106934, "step": 93770 }, { "epoch": 0.4658116686369371, "grad_norm": 1.5813891887664795, "learning_rate": 1.1586537722810296e-05, "loss": 0.16585180759429932, "step": 93780 }, { "epoch": 0.4658613393203064, "grad_norm": 0.7582706212997437, "learning_rate": 1.1584949364096088e-05, "loss": 0.1467670679092407, "step": 93790 }, { "epoch": 0.46591101000367563, "grad_norm": 2.0246739387512207, "learning_rate": 1.1583360964364028e-05, "loss": 0.1520063281059265, "step": 93800 }, { "epoch": 0.4659606806870449, "grad_norm": 1.4163998365402222, "learning_rate": 1.158177252365522e-05, "loss": 0.17321527004241943, "step": 93810 }, { "epoch": 0.46601035137041413, "grad_norm": 0.9151715636253357, "learning_rate": 1.1580184042010779e-05, "loss": 0.1630298137664795, "step": 93820 }, { "epoch": 0.46606002205378344, "grad_norm": 1.6003586053848267, "learning_rate": 1.1578595519471809e-05, "loss": 0.16876702308654784, "step": 93830 }, { "epoch": 0.4661096927371527, "grad_norm": 0.9946461915969849, "learning_rate": 1.157700695607942e-05, "loss": 0.16204428672790527, "step": 93840 }, { "epoch": 0.46615936342052194, "grad_norm": 2.118164300918579, "learning_rate": 1.1575418351874725e-05, "loss": 0.15885781049728392, "step": 93850 }, { "epoch": 0.4662090341038912, "grad_norm": 1.6727573871612549, "learning_rate": 1.1573829706898838e-05, "loss": 0.16404352188110352, "step": 93860 }, { "epoch": 0.46625870478726045, "grad_norm": 1.4834518432617188, "learning_rate": 1.1572241021192874e-05, "loss": 0.1568983793258667, "step": 93870 }, { "epoch": 0.4663083754706297, "grad_norm": 0.6820208430290222, "learning_rate": 1.1570652294797941e-05, "loss": 0.16315652132034303, "step": 93880 }, { "epoch": 0.466358046153999, "grad_norm": 0.8179580569267273, "learning_rate": 1.156906352775516e-05, "loss": 0.16526322364807128, "step": 93890 }, { "epoch": 0.46640771683736826, "grad_norm": 2.290433406829834, "learning_rate": 1.1567474720105642e-05, "loss": 0.15180002450942992, "step": 93900 }, { "epoch": 0.4664573875207375, "grad_norm": 0.9569130539894104, "learning_rate": 1.156588587189051e-05, "loss": 0.17519772052764893, "step": 93910 }, { "epoch": 0.46650705820410676, "grad_norm": 0.910770833492279, "learning_rate": 1.1564296983150882e-05, "loss": 0.1658316969871521, "step": 93920 }, { "epoch": 0.466556728887476, "grad_norm": 1.2689337730407715, "learning_rate": 1.1562708053927878e-05, "loss": 0.16111714839935304, "step": 93930 }, { "epoch": 0.4666063995708453, "grad_norm": 0.8047231435775757, "learning_rate": 1.1561119084262618e-05, "loss": 0.17880666255950928, "step": 93940 }, { "epoch": 0.46665607025421457, "grad_norm": 1.437883734703064, "learning_rate": 1.1559530074196224e-05, "loss": 0.16878408193588257, "step": 93950 }, { "epoch": 0.4667057409375838, "grad_norm": 1.5175485610961914, "learning_rate": 1.1557941023769817e-05, "loss": 0.16923787593841552, "step": 93960 }, { "epoch": 0.4667554116209531, "grad_norm": 1.3932666778564453, "learning_rate": 1.155635193302452e-05, "loss": 0.17025580406188964, "step": 93970 }, { "epoch": 0.4668050823043223, "grad_norm": 0.7978792190551758, "learning_rate": 1.1554762802001465e-05, "loss": 0.15543782711029053, "step": 93980 }, { "epoch": 0.4668547529876916, "grad_norm": 1.5863842964172363, "learning_rate": 1.1553173630741774e-05, "loss": 0.17179601192474364, "step": 93990 }, { "epoch": 0.4669044236710609, "grad_norm": 1.1495709419250488, "learning_rate": 1.1551584419286572e-05, "loss": 0.1625544786453247, "step": 94000 }, { "epoch": 0.46695409435443014, "grad_norm": 0.8542205095291138, "learning_rate": 1.1549995167676991e-05, "loss": 0.16282327175140382, "step": 94010 }, { "epoch": 0.4670037650377994, "grad_norm": 1.2819057703018188, "learning_rate": 1.1548405875954157e-05, "loss": 0.15952872037887572, "step": 94020 }, { "epoch": 0.46705343572116864, "grad_norm": 1.0434602499008179, "learning_rate": 1.1546816544159203e-05, "loss": 0.18024749755859376, "step": 94030 }, { "epoch": 0.4671031064045379, "grad_norm": 1.178360939025879, "learning_rate": 1.1545227172333257e-05, "loss": 0.161895751953125, "step": 94040 }, { "epoch": 0.4671527770879072, "grad_norm": 1.0357273817062378, "learning_rate": 1.1543637760517455e-05, "loss": 0.15705251693725586, "step": 94050 }, { "epoch": 0.46720244777127645, "grad_norm": 1.4358195066452026, "learning_rate": 1.1542048308752927e-05, "loss": 0.1585228681564331, "step": 94060 }, { "epoch": 0.4672521184546457, "grad_norm": 0.9261965155601501, "learning_rate": 1.1540458817080808e-05, "loss": 0.1645024061203003, "step": 94070 }, { "epoch": 0.46730178913801496, "grad_norm": 1.5608539581298828, "learning_rate": 1.1538869285542236e-05, "loss": 0.15894209146499633, "step": 94080 }, { "epoch": 0.4673514598213842, "grad_norm": 1.8035725355148315, "learning_rate": 1.1537279714178345e-05, "loss": 0.1726778507232666, "step": 94090 }, { "epoch": 0.46740113050475346, "grad_norm": 1.45462965965271, "learning_rate": 1.153569010303027e-05, "loss": 0.16810827255249022, "step": 94100 }, { "epoch": 0.46745080118812277, "grad_norm": 0.9380812644958496, "learning_rate": 1.1534100452139157e-05, "loss": 0.16453222036361695, "step": 94110 }, { "epoch": 0.467500471871492, "grad_norm": 1.0758763551712036, "learning_rate": 1.1532510761546138e-05, "loss": 0.16691060066223146, "step": 94120 }, { "epoch": 0.46755014255486127, "grad_norm": 1.3240983486175537, "learning_rate": 1.1530921031292359e-05, "loss": 0.16171225309371948, "step": 94130 }, { "epoch": 0.4675998132382305, "grad_norm": 1.1055694818496704, "learning_rate": 1.1529331261418958e-05, "loss": 0.15248048305511475, "step": 94140 }, { "epoch": 0.4676494839215998, "grad_norm": 1.5659360885620117, "learning_rate": 1.152774145196708e-05, "loss": 0.17734622955322266, "step": 94150 }, { "epoch": 0.4676991546049691, "grad_norm": 1.1421900987625122, "learning_rate": 1.1526151602977867e-05, "loss": 0.15979816913604736, "step": 94160 }, { "epoch": 0.46774882528833833, "grad_norm": 0.8856784105300903, "learning_rate": 1.1524561714492462e-05, "loss": 0.17024881839752198, "step": 94170 }, { "epoch": 0.4677984959717076, "grad_norm": 1.955544114112854, "learning_rate": 1.1522971786552015e-05, "loss": 0.1653200149536133, "step": 94180 }, { "epoch": 0.46784816665507684, "grad_norm": 0.9885273575782776, "learning_rate": 1.1521381819197672e-05, "loss": 0.1645490050315857, "step": 94190 }, { "epoch": 0.4678978373384461, "grad_norm": 0.9941492080688477, "learning_rate": 1.1519791812470577e-05, "loss": 0.1523462176322937, "step": 94200 }, { "epoch": 0.46794750802181534, "grad_norm": 0.8202993273735046, "learning_rate": 1.1518201766411886e-05, "loss": 0.1803170323371887, "step": 94210 }, { "epoch": 0.46799717870518465, "grad_norm": 1.2819782495498657, "learning_rate": 1.1516611681062738e-05, "loss": 0.17489951848983765, "step": 94220 }, { "epoch": 0.4680468493885539, "grad_norm": 1.2640917301177979, "learning_rate": 1.151502155646429e-05, "loss": 0.16191765069961547, "step": 94230 }, { "epoch": 0.46809652007192315, "grad_norm": 1.0189424753189087, "learning_rate": 1.1513431392657692e-05, "loss": 0.16938996315002441, "step": 94240 }, { "epoch": 0.4681461907552924, "grad_norm": 0.7200022339820862, "learning_rate": 1.15118411896841e-05, "loss": 0.1589566469192505, "step": 94250 }, { "epoch": 0.46819586143866165, "grad_norm": 2.23616886138916, "learning_rate": 1.1510250947584668e-05, "loss": 0.15371599197387695, "step": 94260 }, { "epoch": 0.46824553212203096, "grad_norm": 1.692657709121704, "learning_rate": 1.1508660666400546e-05, "loss": 0.15162277221679688, "step": 94270 }, { "epoch": 0.4682952028054002, "grad_norm": 1.5855056047439575, "learning_rate": 1.1507070346172896e-05, "loss": 0.17943031787872316, "step": 94280 }, { "epoch": 0.46834487348876946, "grad_norm": 1.2039501667022705, "learning_rate": 1.1505479986942867e-05, "loss": 0.17330420017242432, "step": 94290 }, { "epoch": 0.4683945441721387, "grad_norm": 1.1291375160217285, "learning_rate": 1.1503889588751622e-05, "loss": 0.1585087537765503, "step": 94300 }, { "epoch": 0.46844421485550797, "grad_norm": 1.1096508502960205, "learning_rate": 1.150229915164032e-05, "loss": 0.16702697277069092, "step": 94310 }, { "epoch": 0.4684938855388772, "grad_norm": 1.172021746635437, "learning_rate": 1.150070867565012e-05, "loss": 0.1688190221786499, "step": 94320 }, { "epoch": 0.4685435562222465, "grad_norm": 0.7116147875785828, "learning_rate": 1.1499118160822184e-05, "loss": 0.16533994674682617, "step": 94330 }, { "epoch": 0.4685932269056158, "grad_norm": 0.7865654826164246, "learning_rate": 1.1497527607197672e-05, "loss": 0.15411558151245117, "step": 94340 }, { "epoch": 0.46864289758898503, "grad_norm": 1.00325608253479, "learning_rate": 1.1495937014817748e-05, "loss": 0.1662790894508362, "step": 94350 }, { "epoch": 0.4686925682723543, "grad_norm": 1.5275020599365234, "learning_rate": 1.1494346383723576e-05, "loss": 0.17070525884628296, "step": 94360 }, { "epoch": 0.46874223895572353, "grad_norm": 1.0072729587554932, "learning_rate": 1.1492755713956319e-05, "loss": 0.15104445219039916, "step": 94370 }, { "epoch": 0.46879190963909284, "grad_norm": 0.8869972825050354, "learning_rate": 1.1491165005557146e-05, "loss": 0.16497067213058472, "step": 94380 }, { "epoch": 0.4688415803224621, "grad_norm": 1.0064150094985962, "learning_rate": 1.148957425856722e-05, "loss": 0.1659320116043091, "step": 94390 }, { "epoch": 0.46889125100583134, "grad_norm": 0.772742509841919, "learning_rate": 1.1487983473027714e-05, "loss": 0.16160666942596436, "step": 94400 }, { "epoch": 0.4689409216892006, "grad_norm": 0.9244619011878967, "learning_rate": 1.1486392648979792e-05, "loss": 0.16345291137695311, "step": 94410 }, { "epoch": 0.46899059237256985, "grad_norm": 1.194827914237976, "learning_rate": 1.1484801786464625e-05, "loss": 0.14235795736312867, "step": 94420 }, { "epoch": 0.4690402630559391, "grad_norm": 0.6675119400024414, "learning_rate": 1.1483210885523386e-05, "loss": 0.15664193630218506, "step": 94430 }, { "epoch": 0.4690899337393084, "grad_norm": 1.3828303813934326, "learning_rate": 1.1481619946197245e-05, "loss": 0.17034623622894288, "step": 94440 }, { "epoch": 0.46913960442267766, "grad_norm": 2.1713006496429443, "learning_rate": 1.1480028968527374e-05, "loss": 0.16735148429870605, "step": 94450 }, { "epoch": 0.4691892751060469, "grad_norm": 1.5569199323654175, "learning_rate": 1.1478437952554953e-05, "loss": 0.17989267110824586, "step": 94460 }, { "epoch": 0.46923894578941616, "grad_norm": 1.7293471097946167, "learning_rate": 1.1476846898321147e-05, "loss": 0.16353410482406616, "step": 94470 }, { "epoch": 0.4692886164727854, "grad_norm": 1.105796217918396, "learning_rate": 1.1475255805867143e-05, "loss": 0.1685415506362915, "step": 94480 }, { "epoch": 0.4693382871561547, "grad_norm": 1.0035814046859741, "learning_rate": 1.1473664675234108e-05, "loss": 0.170112144947052, "step": 94490 }, { "epoch": 0.469387957839524, "grad_norm": 0.9428405165672302, "learning_rate": 1.1472073506463225e-05, "loss": 0.15807199478149414, "step": 94500 }, { "epoch": 0.4694376285228932, "grad_norm": 2.37400484085083, "learning_rate": 1.1470482299595669e-05, "loss": 0.1709949254989624, "step": 94510 }, { "epoch": 0.4694872992062625, "grad_norm": 1.2745449542999268, "learning_rate": 1.1468891054672624e-05, "loss": 0.15718954801559448, "step": 94520 }, { "epoch": 0.46953696988963173, "grad_norm": 0.7606973052024841, "learning_rate": 1.1467299771735269e-05, "loss": 0.16038124561309813, "step": 94530 }, { "epoch": 0.469586640573001, "grad_norm": 2.256495714187622, "learning_rate": 1.1465708450824788e-05, "loss": 0.17310441732406617, "step": 94540 }, { "epoch": 0.4696363112563703, "grad_norm": 1.4289296865463257, "learning_rate": 1.1464117091982361e-05, "loss": 0.17157325744628907, "step": 94550 }, { "epoch": 0.46968598193973954, "grad_norm": 1.4911473989486694, "learning_rate": 1.1462525695249172e-05, "loss": 0.165684974193573, "step": 94560 }, { "epoch": 0.4697356526231088, "grad_norm": 0.7782481908798218, "learning_rate": 1.1460934260666402e-05, "loss": 0.15602536201477052, "step": 94570 }, { "epoch": 0.46978532330647804, "grad_norm": 0.925532341003418, "learning_rate": 1.1459342788275245e-05, "loss": 0.15590839385986327, "step": 94580 }, { "epoch": 0.4698349939898473, "grad_norm": 1.3708406686782837, "learning_rate": 1.1457751278116885e-05, "loss": 0.1755968451499939, "step": 94590 }, { "epoch": 0.4698846646732166, "grad_norm": 1.8935112953186035, "learning_rate": 1.1456159730232503e-05, "loss": 0.16167317628860473, "step": 94600 }, { "epoch": 0.46993433535658585, "grad_norm": 1.5545357465744019, "learning_rate": 1.1454568144663298e-05, "loss": 0.16912583112716675, "step": 94610 }, { "epoch": 0.4699840060399551, "grad_norm": 1.2345926761627197, "learning_rate": 1.1452976521450451e-05, "loss": 0.16354985237121583, "step": 94620 }, { "epoch": 0.47003367672332436, "grad_norm": 1.5755594968795776, "learning_rate": 1.1451384860635158e-05, "loss": 0.173426353931427, "step": 94630 }, { "epoch": 0.4700833474066936, "grad_norm": 0.9184392094612122, "learning_rate": 1.1449793162258605e-05, "loss": 0.17226430177688598, "step": 94640 }, { "epoch": 0.47013301809006286, "grad_norm": 1.4290170669555664, "learning_rate": 1.1448201426361989e-05, "loss": 0.16186531782150268, "step": 94650 }, { "epoch": 0.47018268877343217, "grad_norm": 0.830230712890625, "learning_rate": 1.1446609652986503e-05, "loss": 0.15953004360198975, "step": 94660 }, { "epoch": 0.4702323594568014, "grad_norm": 0.9717537760734558, "learning_rate": 1.1445017842173341e-05, "loss": 0.15097131729125976, "step": 94670 }, { "epoch": 0.47028203014017067, "grad_norm": 1.182918667793274, "learning_rate": 1.14434259939637e-05, "loss": 0.16038684844970702, "step": 94680 }, { "epoch": 0.4703317008235399, "grad_norm": 1.317575216293335, "learning_rate": 1.1441834108398771e-05, "loss": 0.1560905933380127, "step": 94690 }, { "epoch": 0.4703813715069092, "grad_norm": 0.6152819991111755, "learning_rate": 1.1440242185519756e-05, "loss": 0.15366752147674562, "step": 94700 }, { "epoch": 0.4704310421902785, "grad_norm": 1.2323169708251953, "learning_rate": 1.1438650225367855e-05, "loss": 0.17029521465301514, "step": 94710 }, { "epoch": 0.47048071287364773, "grad_norm": 1.542266607284546, "learning_rate": 1.1437058227984264e-05, "loss": 0.15817515850067138, "step": 94720 }, { "epoch": 0.470530383557017, "grad_norm": 0.9111343622207642, "learning_rate": 1.1435466193410183e-05, "loss": 0.16815929412841796, "step": 94730 }, { "epoch": 0.47058005424038624, "grad_norm": 0.9831379652023315, "learning_rate": 1.1433874121686813e-05, "loss": 0.16355998516082765, "step": 94740 }, { "epoch": 0.4706297249237555, "grad_norm": 0.7906298637390137, "learning_rate": 1.1432282012855358e-05, "loss": 0.16595182418823243, "step": 94750 }, { "epoch": 0.47067939560712474, "grad_norm": 1.2385159730911255, "learning_rate": 1.1430689866957021e-05, "loss": 0.1621890187263489, "step": 94760 }, { "epoch": 0.47072906629049405, "grad_norm": 0.7243406176567078, "learning_rate": 1.1429097684033002e-05, "loss": 0.15595970153808594, "step": 94770 }, { "epoch": 0.4707787369738633, "grad_norm": 1.131820797920227, "learning_rate": 1.1427505464124513e-05, "loss": 0.16910279989242555, "step": 94780 }, { "epoch": 0.47082840765723255, "grad_norm": 0.801660418510437, "learning_rate": 1.1425913207272756e-05, "loss": 0.16560393571853638, "step": 94790 }, { "epoch": 0.4708780783406018, "grad_norm": 2.086031675338745, "learning_rate": 1.142432091351894e-05, "loss": 0.17905845642089843, "step": 94800 }, { "epoch": 0.47092774902397105, "grad_norm": 1.2876136302947998, "learning_rate": 1.142272858290427e-05, "loss": 0.16016788482666017, "step": 94810 }, { "epoch": 0.47097741970734036, "grad_norm": 1.4199042320251465, "learning_rate": 1.1421136215469957e-05, "loss": 0.16309831142425538, "step": 94820 }, { "epoch": 0.4710270903907096, "grad_norm": 1.0462002754211426, "learning_rate": 1.141954381125721e-05, "loss": 0.1592634916305542, "step": 94830 }, { "epoch": 0.47107676107407886, "grad_norm": 0.6543075442314148, "learning_rate": 1.1417951370307239e-05, "loss": 0.14926576614379883, "step": 94840 }, { "epoch": 0.4711264317574481, "grad_norm": 0.8021411895751953, "learning_rate": 1.1416358892661255e-05, "loss": 0.15359799861907958, "step": 94850 }, { "epoch": 0.47117610244081737, "grad_norm": 0.928278386592865, "learning_rate": 1.1414766378360476e-05, "loss": 0.15072028636932372, "step": 94860 }, { "epoch": 0.4712257731241866, "grad_norm": 1.5298882722854614, "learning_rate": 1.1413173827446109e-05, "loss": 0.16623952388763427, "step": 94870 }, { "epoch": 0.4712754438075559, "grad_norm": 1.5529683828353882, "learning_rate": 1.1411581239959374e-05, "loss": 0.16594231128692627, "step": 94880 }, { "epoch": 0.4713251144909252, "grad_norm": 1.0886223316192627, "learning_rate": 1.1409988615941485e-05, "loss": 0.16239864826202394, "step": 94890 }, { "epoch": 0.47137478517429443, "grad_norm": 1.2564936876296997, "learning_rate": 1.1408395955433653e-05, "loss": 0.18212666511535644, "step": 94900 }, { "epoch": 0.4714244558576637, "grad_norm": 1.8833001852035522, "learning_rate": 1.1406803258477104e-05, "loss": 0.1721283197402954, "step": 94910 }, { "epoch": 0.47147412654103293, "grad_norm": 1.2673559188842773, "learning_rate": 1.140521052511305e-05, "loss": 0.16044580936431885, "step": 94920 }, { "epoch": 0.4715237972244022, "grad_norm": 1.6021637916564941, "learning_rate": 1.1403617755382714e-05, "loss": 0.1616516351699829, "step": 94930 }, { "epoch": 0.4715734679077715, "grad_norm": 0.9388792514801025, "learning_rate": 1.1402024949327315e-05, "loss": 0.17325778007507325, "step": 94940 }, { "epoch": 0.47162313859114074, "grad_norm": 0.8617663383483887, "learning_rate": 1.1400432106988075e-05, "loss": 0.17110579013824462, "step": 94950 }, { "epoch": 0.47167280927451, "grad_norm": 1.5900894403457642, "learning_rate": 1.1398839228406214e-05, "loss": 0.16500134468078614, "step": 94960 }, { "epoch": 0.47172247995787925, "grad_norm": 1.0880485773086548, "learning_rate": 1.1397246313622957e-05, "loss": 0.17162652015686036, "step": 94970 }, { "epoch": 0.4717721506412485, "grad_norm": 1.291138768196106, "learning_rate": 1.1395653362679521e-05, "loss": 0.1681518316268921, "step": 94980 }, { "epoch": 0.4718218213246178, "grad_norm": 1.9550237655639648, "learning_rate": 1.1394060375617144e-05, "loss": 0.160502290725708, "step": 94990 }, { "epoch": 0.47187149200798706, "grad_norm": 1.3254221677780151, "learning_rate": 1.1392467352477043e-05, "loss": 0.17206289768218994, "step": 95000 }, { "epoch": 0.4719211626913563, "grad_norm": 0.6418502330780029, "learning_rate": 1.1390874293300447e-05, "loss": 0.16672587394714355, "step": 95010 }, { "epoch": 0.47197083337472556, "grad_norm": 0.8817602396011353, "learning_rate": 1.1389281198128583e-05, "loss": 0.1641181230545044, "step": 95020 }, { "epoch": 0.4720205040580948, "grad_norm": 1.713549256324768, "learning_rate": 1.1387688067002683e-05, "loss": 0.16924281120300294, "step": 95030 }, { "epoch": 0.47207017474146407, "grad_norm": 2.2679283618927, "learning_rate": 1.1386094899963971e-05, "loss": 0.1687614679336548, "step": 95040 }, { "epoch": 0.4721198454248334, "grad_norm": 1.8102587461471558, "learning_rate": 1.1384501697053682e-05, "loss": 0.16411253213882446, "step": 95050 }, { "epoch": 0.4721695161082026, "grad_norm": 1.2531969547271729, "learning_rate": 1.1382908458313044e-05, "loss": 0.15871071815490723, "step": 95060 }, { "epoch": 0.4722191867915719, "grad_norm": 1.664661169052124, "learning_rate": 1.1381315183783293e-05, "loss": 0.169492769241333, "step": 95070 }, { "epoch": 0.47226885747494113, "grad_norm": 2.076935291290283, "learning_rate": 1.1379721873505659e-05, "loss": 0.16995432376861572, "step": 95080 }, { "epoch": 0.4723185281583104, "grad_norm": 1.7264310121536255, "learning_rate": 1.1378128527521378e-05, "loss": 0.1536358118057251, "step": 95090 }, { "epoch": 0.4723681988416797, "grad_norm": 1.2358982563018799, "learning_rate": 1.1376535145871685e-05, "loss": 0.1694063663482666, "step": 95100 }, { "epoch": 0.47241786952504894, "grad_norm": 1.8507096767425537, "learning_rate": 1.1374941728597815e-05, "loss": 0.17297505140304564, "step": 95110 }, { "epoch": 0.4724675402084182, "grad_norm": 1.3979555368423462, "learning_rate": 1.1373348275741005e-05, "loss": 0.16969037055969238, "step": 95120 }, { "epoch": 0.47251721089178744, "grad_norm": 0.739786684513092, "learning_rate": 1.1371754787342497e-05, "loss": 0.17093199491500854, "step": 95130 }, { "epoch": 0.4725668815751567, "grad_norm": 0.8671039342880249, "learning_rate": 1.1370161263443525e-05, "loss": 0.16642253398895263, "step": 95140 }, { "epoch": 0.47261655225852595, "grad_norm": 1.6418631076812744, "learning_rate": 1.1368567704085329e-05, "loss": 0.16095983982086182, "step": 95150 }, { "epoch": 0.47266622294189525, "grad_norm": 1.0339998006820679, "learning_rate": 1.1366974109309155e-05, "loss": 0.16431397199630737, "step": 95160 }, { "epoch": 0.4727158936252645, "grad_norm": 0.5660015940666199, "learning_rate": 1.1365380479156238e-05, "loss": 0.17332481145858764, "step": 95170 }, { "epoch": 0.47276556430863376, "grad_norm": 1.0009739398956299, "learning_rate": 1.1363786813667821e-05, "loss": 0.16788907051086427, "step": 95180 }, { "epoch": 0.472815234992003, "grad_norm": 0.9087463617324829, "learning_rate": 1.1362193112885152e-05, "loss": 0.18748188018798828, "step": 95190 }, { "epoch": 0.47286490567537226, "grad_norm": 1.5190966129302979, "learning_rate": 1.1360599376849473e-05, "loss": 0.16310226917266846, "step": 95200 }, { "epoch": 0.47291457635874157, "grad_norm": 0.7519693374633789, "learning_rate": 1.135900560560203e-05, "loss": 0.1413649559020996, "step": 95210 }, { "epoch": 0.4729642470421108, "grad_norm": 0.8555396199226379, "learning_rate": 1.135741179918407e-05, "loss": 0.16869477033615113, "step": 95220 }, { "epoch": 0.47301391772548007, "grad_norm": 0.8165891766548157, "learning_rate": 1.1355817957636836e-05, "loss": 0.1652796149253845, "step": 95230 }, { "epoch": 0.4730635884088493, "grad_norm": 1.3755117654800415, "learning_rate": 1.1354224081001577e-05, "loss": 0.1667543411254883, "step": 95240 }, { "epoch": 0.4731132590922186, "grad_norm": 1.8033500909805298, "learning_rate": 1.1352630169319546e-05, "loss": 0.1687110424041748, "step": 95250 }, { "epoch": 0.4731629297755878, "grad_norm": 0.7813699245452881, "learning_rate": 1.1351036222631989e-05, "loss": 0.1638386368751526, "step": 95260 }, { "epoch": 0.47321260045895713, "grad_norm": 1.5505380630493164, "learning_rate": 1.1349442240980157e-05, "loss": 0.1732025146484375, "step": 95270 }, { "epoch": 0.4732622711423264, "grad_norm": 0.6753590703010559, "learning_rate": 1.1347848224405302e-05, "loss": 0.17020970582962036, "step": 95280 }, { "epoch": 0.47331194182569564, "grad_norm": 1.0364816188812256, "learning_rate": 1.1346254172948679e-05, "loss": 0.15961320400238038, "step": 95290 }, { "epoch": 0.4733616125090649, "grad_norm": 1.1024110317230225, "learning_rate": 1.1344660086651537e-05, "loss": 0.15589214563369752, "step": 95300 }, { "epoch": 0.47341128319243414, "grad_norm": 1.3240545988082886, "learning_rate": 1.1343065965555133e-05, "loss": 0.1692942976951599, "step": 95310 }, { "epoch": 0.47346095387580345, "grad_norm": 0.9663083553314209, "learning_rate": 1.1341471809700724e-05, "loss": 0.1661142349243164, "step": 95320 }, { "epoch": 0.4735106245591727, "grad_norm": 1.1671268939971924, "learning_rate": 1.1339877619129563e-05, "loss": 0.15470913648605347, "step": 95330 }, { "epoch": 0.47356029524254195, "grad_norm": 0.7841514348983765, "learning_rate": 1.1338283393882907e-05, "loss": 0.15219354629516602, "step": 95340 }, { "epoch": 0.4736099659259112, "grad_norm": 1.1841531991958618, "learning_rate": 1.1336689134002014e-05, "loss": 0.16912469863891602, "step": 95350 }, { "epoch": 0.47365963660928045, "grad_norm": 0.8733235597610474, "learning_rate": 1.1335094839528147e-05, "loss": 0.16289448738098145, "step": 95360 }, { "epoch": 0.4737093072926497, "grad_norm": 1.968621015548706, "learning_rate": 1.133350051050256e-05, "loss": 0.16381628513336183, "step": 95370 }, { "epoch": 0.473758977976019, "grad_norm": 1.7769718170166016, "learning_rate": 1.1331906146966519e-05, "loss": 0.17343387603759766, "step": 95380 }, { "epoch": 0.47380864865938827, "grad_norm": 1.2710572481155396, "learning_rate": 1.1330311748961278e-05, "loss": 0.16361792087554933, "step": 95390 }, { "epoch": 0.4738583193427575, "grad_norm": 0.8120166063308716, "learning_rate": 1.1328717316528107e-05, "loss": 0.1638808012008667, "step": 95400 }, { "epoch": 0.47390799002612677, "grad_norm": 1.5604300498962402, "learning_rate": 1.1327122849708262e-05, "loss": 0.1657109260559082, "step": 95410 }, { "epoch": 0.473957660709496, "grad_norm": 1.6716805696487427, "learning_rate": 1.1325528348543016e-05, "loss": 0.1671138048171997, "step": 95420 }, { "epoch": 0.4740073313928653, "grad_norm": 0.9952954649925232, "learning_rate": 1.1323933813073628e-05, "loss": 0.15217899084091185, "step": 95430 }, { "epoch": 0.4740570020762346, "grad_norm": 1.2241982221603394, "learning_rate": 1.1322339243341363e-05, "loss": 0.16932063102722167, "step": 95440 }, { "epoch": 0.47410667275960383, "grad_norm": 1.467512607574463, "learning_rate": 1.1320744639387491e-05, "loss": 0.18290939331054687, "step": 95450 }, { "epoch": 0.4741563434429731, "grad_norm": 1.6691243648529053, "learning_rate": 1.131915000125328e-05, "loss": 0.1575498104095459, "step": 95460 }, { "epoch": 0.47420601412634233, "grad_norm": 0.8361870050430298, "learning_rate": 1.1317555328979997e-05, "loss": 0.16182686090469361, "step": 95470 }, { "epoch": 0.4742556848097116, "grad_norm": 1.026774525642395, "learning_rate": 1.1315960622608912e-05, "loss": 0.14729876518249513, "step": 95480 }, { "epoch": 0.4743053554930809, "grad_norm": 2.158663034439087, "learning_rate": 1.1314365882181295e-05, "loss": 0.17085481882095338, "step": 95490 }, { "epoch": 0.47435502617645015, "grad_norm": 0.7467711567878723, "learning_rate": 1.1312771107738419e-05, "loss": 0.16886615753173828, "step": 95500 }, { "epoch": 0.4744046968598194, "grad_norm": 1.2375019788742065, "learning_rate": 1.1311176299321554e-05, "loss": 0.16877562999725343, "step": 95510 }, { "epoch": 0.47445436754318865, "grad_norm": 1.3627365827560425, "learning_rate": 1.130958145697197e-05, "loss": 0.16206753253936768, "step": 95520 }, { "epoch": 0.4745040382265579, "grad_norm": 0.61945641040802, "learning_rate": 1.1307986580730948e-05, "loss": 0.15265434980392456, "step": 95530 }, { "epoch": 0.4745537089099272, "grad_norm": 0.6948167085647583, "learning_rate": 1.1306391670639755e-05, "loss": 0.15407226085662842, "step": 95540 }, { "epoch": 0.47460337959329646, "grad_norm": 0.9255023002624512, "learning_rate": 1.1304796726739676e-05, "loss": 0.16197171211242675, "step": 95550 }, { "epoch": 0.4746530502766657, "grad_norm": 1.467297911643982, "learning_rate": 1.1303201749071984e-05, "loss": 0.16615021228790283, "step": 95560 }, { "epoch": 0.47470272096003496, "grad_norm": 1.4732636213302612, "learning_rate": 1.1301606737677948e-05, "loss": 0.1758764624595642, "step": 95570 }, { "epoch": 0.4747523916434042, "grad_norm": 1.2029528617858887, "learning_rate": 1.1300011692598856e-05, "loss": 0.18672516345977783, "step": 95580 }, { "epoch": 0.47480206232677347, "grad_norm": 1.2024344205856323, "learning_rate": 1.1298416613875984e-05, "loss": 0.1545945644378662, "step": 95590 }, { "epoch": 0.4748517330101428, "grad_norm": 0.8735709190368652, "learning_rate": 1.1296821501550615e-05, "loss": 0.16973307132720947, "step": 95600 }, { "epoch": 0.474901403693512, "grad_norm": 1.1041285991668701, "learning_rate": 1.1295226355664025e-05, "loss": 0.17268872261047363, "step": 95610 }, { "epoch": 0.4749510743768813, "grad_norm": 1.311704158782959, "learning_rate": 1.1293631176257497e-05, "loss": 0.15202126502990723, "step": 95620 }, { "epoch": 0.47500074506025053, "grad_norm": 1.7873350381851196, "learning_rate": 1.1292035963372311e-05, "loss": 0.1496502161026001, "step": 95630 }, { "epoch": 0.4750504157436198, "grad_norm": 1.413007140159607, "learning_rate": 1.129044071704976e-05, "loss": 0.17504334449768066, "step": 95640 }, { "epoch": 0.4751000864269891, "grad_norm": 0.8036723136901855, "learning_rate": 1.1288845437331118e-05, "loss": 0.16647982597351074, "step": 95650 }, { "epoch": 0.47514975711035834, "grad_norm": 2.362990617752075, "learning_rate": 1.1287250124257674e-05, "loss": 0.1632075786590576, "step": 95660 }, { "epoch": 0.4751994277937276, "grad_norm": 0.7912246584892273, "learning_rate": 1.1285654777870715e-05, "loss": 0.16513872146606445, "step": 95670 }, { "epoch": 0.47524909847709684, "grad_norm": 1.6979050636291504, "learning_rate": 1.1284059398211529e-05, "loss": 0.15368467569351196, "step": 95680 }, { "epoch": 0.4752987691604661, "grad_norm": 1.295801043510437, "learning_rate": 1.12824639853214e-05, "loss": 0.1549829363822937, "step": 95690 }, { "epoch": 0.47534843984383535, "grad_norm": 0.8699283599853516, "learning_rate": 1.1280868539241623e-05, "loss": 0.16764096021652222, "step": 95700 }, { "epoch": 0.47539811052720465, "grad_norm": 1.2616246938705444, "learning_rate": 1.127927306001348e-05, "loss": 0.15910452604293823, "step": 95710 }, { "epoch": 0.4754477812105739, "grad_norm": 0.9062687754631042, "learning_rate": 1.1277677547678264e-05, "loss": 0.15201408863067628, "step": 95720 }, { "epoch": 0.47549745189394316, "grad_norm": 1.3828792572021484, "learning_rate": 1.127608200227727e-05, "loss": 0.17341303825378418, "step": 95730 }, { "epoch": 0.4755471225773124, "grad_norm": 0.821662187576294, "learning_rate": 1.1274486423851782e-05, "loss": 0.17174043655395507, "step": 95740 }, { "epoch": 0.47559679326068166, "grad_norm": 1.1374711990356445, "learning_rate": 1.1272890812443103e-05, "loss": 0.1636246085166931, "step": 95750 }, { "epoch": 0.47564646394405097, "grad_norm": 0.7053349018096924, "learning_rate": 1.1271295168092517e-05, "loss": 0.16824856996536255, "step": 95760 }, { "epoch": 0.4756961346274202, "grad_norm": 2.7728326320648193, "learning_rate": 1.1269699490841326e-05, "loss": 0.17044591903686523, "step": 95770 }, { "epoch": 0.47574580531078947, "grad_norm": 1.0283602476119995, "learning_rate": 1.126810378073082e-05, "loss": 0.17498596906661987, "step": 95780 }, { "epoch": 0.4757954759941587, "grad_norm": 0.705198347568512, "learning_rate": 1.12665080378023e-05, "loss": 0.16265125274658204, "step": 95790 }, { "epoch": 0.475845146677528, "grad_norm": 1.5311596393585205, "learning_rate": 1.126491226209706e-05, "loss": 0.1565704107284546, "step": 95800 }, { "epoch": 0.4758948173608972, "grad_norm": 0.8204113245010376, "learning_rate": 1.1263316453656401e-05, "loss": 0.16874375343322753, "step": 95810 }, { "epoch": 0.47594448804426653, "grad_norm": 1.1265791654586792, "learning_rate": 1.126172061252162e-05, "loss": 0.1698807954788208, "step": 95820 }, { "epoch": 0.4759941587276358, "grad_norm": 1.4725978374481201, "learning_rate": 1.1260124738734018e-05, "loss": 0.148321270942688, "step": 95830 }, { "epoch": 0.47604382941100504, "grad_norm": 1.4938818216323853, "learning_rate": 1.1258528832334892e-05, "loss": 0.16059322357177735, "step": 95840 }, { "epoch": 0.4760935000943743, "grad_norm": 2.3769445419311523, "learning_rate": 1.1256932893365544e-05, "loss": 0.16533176898956298, "step": 95850 }, { "epoch": 0.47614317077774354, "grad_norm": 1.4329333305358887, "learning_rate": 1.1255336921867282e-05, "loss": 0.15045225620269775, "step": 95860 }, { "epoch": 0.47619284146111285, "grad_norm": 1.2385276556015015, "learning_rate": 1.1253740917881403e-05, "loss": 0.17001992464065552, "step": 95870 }, { "epoch": 0.4762425121444821, "grad_norm": 0.8738772869110107, "learning_rate": 1.1252144881449214e-05, "loss": 0.15473148822784424, "step": 95880 }, { "epoch": 0.47629218282785135, "grad_norm": 1.489889144897461, "learning_rate": 1.1250548812612018e-05, "loss": 0.1663602352142334, "step": 95890 }, { "epoch": 0.4763418535112206, "grad_norm": 0.9770123958587646, "learning_rate": 1.1248952711411128e-05, "loss": 0.16441988945007324, "step": 95900 }, { "epoch": 0.47639152419458985, "grad_norm": 0.9102811217308044, "learning_rate": 1.1247356577887835e-05, "loss": 0.17363348007202148, "step": 95910 }, { "epoch": 0.4764411948779591, "grad_norm": 1.1168888807296753, "learning_rate": 1.1245760412083459e-05, "loss": 0.1603529930114746, "step": 95920 }, { "epoch": 0.4764908655613284, "grad_norm": 1.2185872793197632, "learning_rate": 1.1244164214039303e-05, "loss": 0.16468193531036376, "step": 95930 }, { "epoch": 0.47654053624469767, "grad_norm": 1.5407460927963257, "learning_rate": 1.1242567983796679e-05, "loss": 0.1548243522644043, "step": 95940 }, { "epoch": 0.4765902069280669, "grad_norm": 1.1940735578536987, "learning_rate": 1.1240971721396896e-05, "loss": 0.16066042184829712, "step": 95950 }, { "epoch": 0.47663987761143617, "grad_norm": 1.9736436605453491, "learning_rate": 1.1239375426881262e-05, "loss": 0.16166090965270996, "step": 95960 }, { "epoch": 0.4766895482948054, "grad_norm": 0.6677075028419495, "learning_rate": 1.1237779100291093e-05, "loss": 0.16636359691619873, "step": 95970 }, { "epoch": 0.47673921897817473, "grad_norm": 1.3474340438842773, "learning_rate": 1.1236182741667693e-05, "loss": 0.1448122024536133, "step": 95980 }, { "epoch": 0.476788889661544, "grad_norm": 2.23699951171875, "learning_rate": 1.1234586351052386e-05, "loss": 0.18008273839950562, "step": 95990 }, { "epoch": 0.47683856034491323, "grad_norm": 1.3481042385101318, "learning_rate": 1.123298992848648e-05, "loss": 0.16874284744262696, "step": 96000 }, { "epoch": 0.4768882310282825, "grad_norm": 1.6562634706497192, "learning_rate": 1.1231393474011287e-05, "loss": 0.16195615530014038, "step": 96010 }, { "epoch": 0.47693790171165173, "grad_norm": 1.8679906129837036, "learning_rate": 1.122979698766813e-05, "loss": 0.1580662965774536, "step": 96020 }, { "epoch": 0.476987572395021, "grad_norm": 1.8237816095352173, "learning_rate": 1.122820046949832e-05, "loss": 0.16496140956878663, "step": 96030 }, { "epoch": 0.4770372430783903, "grad_norm": 1.0363446474075317, "learning_rate": 1.1226603919543176e-05, "loss": 0.16572086811065673, "step": 96040 }, { "epoch": 0.47708691376175955, "grad_norm": 0.8648551106452942, "learning_rate": 1.1225007337844013e-05, "loss": 0.18313820362091066, "step": 96050 }, { "epoch": 0.4771365844451288, "grad_norm": 1.6587411165237427, "learning_rate": 1.1223410724442154e-05, "loss": 0.16273119449615478, "step": 96060 }, { "epoch": 0.47718625512849805, "grad_norm": 0.6848745346069336, "learning_rate": 1.1221814079378922e-05, "loss": 0.1620081663131714, "step": 96070 }, { "epoch": 0.4772359258118673, "grad_norm": 0.6632774472236633, "learning_rate": 1.1220217402695627e-05, "loss": 0.1790026307106018, "step": 96080 }, { "epoch": 0.4772855964952366, "grad_norm": 1.9363826513290405, "learning_rate": 1.12186206944336e-05, "loss": 0.165167236328125, "step": 96090 }, { "epoch": 0.47733526717860586, "grad_norm": 1.2045546770095825, "learning_rate": 1.1217023954634159e-05, "loss": 0.15412187576293945, "step": 96100 }, { "epoch": 0.4773849378619751, "grad_norm": 1.2530157566070557, "learning_rate": 1.1215427183338623e-05, "loss": 0.15249464511871338, "step": 96110 }, { "epoch": 0.47743460854534436, "grad_norm": 1.3201884031295776, "learning_rate": 1.1213830380588323e-05, "loss": 0.18516275882720948, "step": 96120 }, { "epoch": 0.4774842792287136, "grad_norm": 0.9795552492141724, "learning_rate": 1.1212233546424582e-05, "loss": 0.16917948722839354, "step": 96130 }, { "epoch": 0.47753394991208287, "grad_norm": 3.856565475463867, "learning_rate": 1.1210636680888725e-05, "loss": 0.15771734714508057, "step": 96140 }, { "epoch": 0.4775836205954522, "grad_norm": 1.240073323249817, "learning_rate": 1.1209039784022075e-05, "loss": 0.16803276538848877, "step": 96150 }, { "epoch": 0.4776332912788214, "grad_norm": 1.5606333017349243, "learning_rate": 1.1207442855865961e-05, "loss": 0.1610252857208252, "step": 96160 }, { "epoch": 0.4776829619621907, "grad_norm": 0.7708411812782288, "learning_rate": 1.1205845896461716e-05, "loss": 0.15553567409515381, "step": 96170 }, { "epoch": 0.47773263264555993, "grad_norm": 1.7426997423171997, "learning_rate": 1.1204248905850661e-05, "loss": 0.17659549713134765, "step": 96180 }, { "epoch": 0.4777823033289292, "grad_norm": 2.394280433654785, "learning_rate": 1.1202651884074125e-05, "loss": 0.14884356260299683, "step": 96190 }, { "epoch": 0.4778319740122985, "grad_norm": 0.7886313796043396, "learning_rate": 1.1201054831173446e-05, "loss": 0.16166828870773314, "step": 96200 }, { "epoch": 0.47788164469566774, "grad_norm": 1.1435959339141846, "learning_rate": 1.119945774718995e-05, "loss": 0.1557483434677124, "step": 96210 }, { "epoch": 0.477931315379037, "grad_norm": 1.2697292566299438, "learning_rate": 1.119786063216497e-05, "loss": 0.14366018772125244, "step": 96220 }, { "epoch": 0.47798098606240624, "grad_norm": 1.7422679662704468, "learning_rate": 1.1196263486139837e-05, "loss": 0.15966668128967285, "step": 96230 }, { "epoch": 0.4780306567457755, "grad_norm": 0.5830254554748535, "learning_rate": 1.119466630915589e-05, "loss": 0.15333178043365478, "step": 96240 }, { "epoch": 0.47808032742914475, "grad_norm": 1.0710104703903198, "learning_rate": 1.1193069101254457e-05, "loss": 0.1581049919128418, "step": 96250 }, { "epoch": 0.47812999811251405, "grad_norm": 1.2157264947891235, "learning_rate": 1.1191471862476875e-05, "loss": 0.18444734811782837, "step": 96260 }, { "epoch": 0.4781796687958833, "grad_norm": 1.9494903087615967, "learning_rate": 1.118987459286448e-05, "loss": 0.16583607196807862, "step": 96270 }, { "epoch": 0.47822933947925256, "grad_norm": 1.335414171218872, "learning_rate": 1.118827729245861e-05, "loss": 0.17066783905029298, "step": 96280 }, { "epoch": 0.4782790101626218, "grad_norm": 1.5326917171478271, "learning_rate": 1.1186679961300599e-05, "loss": 0.162258780002594, "step": 96290 }, { "epoch": 0.47832868084599106, "grad_norm": 0.9798955917358398, "learning_rate": 1.1185082599431791e-05, "loss": 0.15229995250701905, "step": 96300 }, { "epoch": 0.47837835152936037, "grad_norm": 1.4785069227218628, "learning_rate": 1.1183485206893521e-05, "loss": 0.17542992830276488, "step": 96310 }, { "epoch": 0.4784280222127296, "grad_norm": 1.4484323263168335, "learning_rate": 1.1181887783727126e-05, "loss": 0.16831822395324708, "step": 96320 }, { "epoch": 0.47847769289609887, "grad_norm": 1.260995864868164, "learning_rate": 1.1180290329973953e-05, "loss": 0.16474376916885375, "step": 96330 }, { "epoch": 0.4785273635794681, "grad_norm": 0.6919202208518982, "learning_rate": 1.1178692845675343e-05, "loss": 0.16048661470413209, "step": 96340 }, { "epoch": 0.4785770342628374, "grad_norm": 1.4996323585510254, "learning_rate": 1.1177095330872634e-05, "loss": 0.16326863765716554, "step": 96350 }, { "epoch": 0.4786267049462066, "grad_norm": 1.0944921970367432, "learning_rate": 1.117549778560717e-05, "loss": 0.16098175048828126, "step": 96360 }, { "epoch": 0.47867637562957593, "grad_norm": 3.701634407043457, "learning_rate": 1.1173900209920299e-05, "loss": 0.15884824991226196, "step": 96370 }, { "epoch": 0.4787260463129452, "grad_norm": 1.045318841934204, "learning_rate": 1.117230260385336e-05, "loss": 0.16215165853500366, "step": 96380 }, { "epoch": 0.47877571699631444, "grad_norm": 1.0710595846176147, "learning_rate": 1.11707049674477e-05, "loss": 0.17478455305099488, "step": 96390 }, { "epoch": 0.4788253876796837, "grad_norm": 0.8868114352226257, "learning_rate": 1.1169107300744671e-05, "loss": 0.15907564163208007, "step": 96400 }, { "epoch": 0.47887505836305294, "grad_norm": 1.4072134494781494, "learning_rate": 1.116750960378561e-05, "loss": 0.1557000160217285, "step": 96410 }, { "epoch": 0.47892472904642225, "grad_norm": 2.0614802837371826, "learning_rate": 1.1165911876611874e-05, "loss": 0.17321927547454835, "step": 96420 }, { "epoch": 0.4789743997297915, "grad_norm": 0.749349057674408, "learning_rate": 1.1164314119264805e-05, "loss": 0.16892045736312866, "step": 96430 }, { "epoch": 0.47902407041316075, "grad_norm": 0.805099368095398, "learning_rate": 1.1162716331785756e-05, "loss": 0.17743544578552245, "step": 96440 }, { "epoch": 0.47907374109653, "grad_norm": 2.4435110092163086, "learning_rate": 1.1161118514216074e-05, "loss": 0.1597992181777954, "step": 96450 }, { "epoch": 0.47912341177989926, "grad_norm": 2.0996575355529785, "learning_rate": 1.1159520666597111e-05, "loss": 0.15846543312072753, "step": 96460 }, { "epoch": 0.4791730824632685, "grad_norm": 1.1963526010513306, "learning_rate": 1.1157922788970223e-05, "loss": 0.16441919803619384, "step": 96470 }, { "epoch": 0.4792227531466378, "grad_norm": 1.0740975141525269, "learning_rate": 1.1156324881376756e-05, "loss": 0.15567760467529296, "step": 96480 }, { "epoch": 0.47927242383000707, "grad_norm": 0.9338625073432922, "learning_rate": 1.1154726943858067e-05, "loss": 0.14958611726760865, "step": 96490 }, { "epoch": 0.4793220945133763, "grad_norm": 1.0463703870773315, "learning_rate": 1.1153128976455512e-05, "loss": 0.15438072681427, "step": 96500 }, { "epoch": 0.47937176519674557, "grad_norm": 1.6557248830795288, "learning_rate": 1.1151530979210438e-05, "loss": 0.16662344932556153, "step": 96510 }, { "epoch": 0.4794214358801148, "grad_norm": 1.3346866369247437, "learning_rate": 1.1149932952164206e-05, "loss": 0.17817389965057373, "step": 96520 }, { "epoch": 0.4794711065634841, "grad_norm": 1.2262910604476929, "learning_rate": 1.1148334895358173e-05, "loss": 0.15243997573852539, "step": 96530 }, { "epoch": 0.4795207772468534, "grad_norm": 2.8144853115081787, "learning_rate": 1.1146736808833694e-05, "loss": 0.17394356727600097, "step": 96540 }, { "epoch": 0.47957044793022263, "grad_norm": 0.8592624068260193, "learning_rate": 1.1145138692632128e-05, "loss": 0.15593039989471436, "step": 96550 }, { "epoch": 0.4796201186135919, "grad_norm": 1.6940464973449707, "learning_rate": 1.1143540546794833e-05, "loss": 0.16741518974304198, "step": 96560 }, { "epoch": 0.47966978929696114, "grad_norm": 0.7971652150154114, "learning_rate": 1.114194237136317e-05, "loss": 0.15599453449249268, "step": 96570 }, { "epoch": 0.4797194599803304, "grad_norm": 1.542952060699463, "learning_rate": 1.1140344166378493e-05, "loss": 0.168130624294281, "step": 96580 }, { "epoch": 0.4797691306636997, "grad_norm": 0.9727439284324646, "learning_rate": 1.1138745931882171e-05, "loss": 0.16730976104736328, "step": 96590 }, { "epoch": 0.47981880134706895, "grad_norm": 1.6314891576766968, "learning_rate": 1.1137147667915563e-05, "loss": 0.16315345764160155, "step": 96600 }, { "epoch": 0.4798684720304382, "grad_norm": 1.7789711952209473, "learning_rate": 1.113554937452003e-05, "loss": 0.17821661233901978, "step": 96610 }, { "epoch": 0.47991814271380745, "grad_norm": 1.4233818054199219, "learning_rate": 1.1133951051736933e-05, "loss": 0.17375084161758422, "step": 96620 }, { "epoch": 0.4799678133971767, "grad_norm": 1.447485327720642, "learning_rate": 1.1132352699607641e-05, "loss": 0.15473453998565673, "step": 96630 }, { "epoch": 0.48001748408054595, "grad_norm": 0.9211647510528564, "learning_rate": 1.1130754318173516e-05, "loss": 0.1601698637008667, "step": 96640 }, { "epoch": 0.48006715476391526, "grad_norm": 0.7942110896110535, "learning_rate": 1.1129155907475921e-05, "loss": 0.1718374013900757, "step": 96650 }, { "epoch": 0.4801168254472845, "grad_norm": 1.3611665964126587, "learning_rate": 1.112755746755623e-05, "loss": 0.16317070722579957, "step": 96660 }, { "epoch": 0.48016649613065376, "grad_norm": 2.92338490486145, "learning_rate": 1.11259589984558e-05, "loss": 0.15679869651794434, "step": 96670 }, { "epoch": 0.480216166814023, "grad_norm": 0.708638072013855, "learning_rate": 1.1124360500216009e-05, "loss": 0.15718717575073243, "step": 96680 }, { "epoch": 0.48026583749739227, "grad_norm": 1.1555473804473877, "learning_rate": 1.112276197287822e-05, "loss": 0.1567077398300171, "step": 96690 }, { "epoch": 0.4803155081807616, "grad_norm": 1.414621114730835, "learning_rate": 1.11211634164838e-05, "loss": 0.17125576734542847, "step": 96700 }, { "epoch": 0.4803651788641308, "grad_norm": 2.081125020980835, "learning_rate": 1.1119564831074124e-05, "loss": 0.15764399766921997, "step": 96710 }, { "epoch": 0.4804148495475001, "grad_norm": 0.9441980123519897, "learning_rate": 1.1117966216690557e-05, "loss": 0.16520376205444337, "step": 96720 }, { "epoch": 0.48046452023086933, "grad_norm": 1.5265294313430786, "learning_rate": 1.1116367573374473e-05, "loss": 0.1665586233139038, "step": 96730 }, { "epoch": 0.4805141909142386, "grad_norm": 1.2670906782150269, "learning_rate": 1.1114768901167251e-05, "loss": 0.1731898546218872, "step": 96740 }, { "epoch": 0.48056386159760783, "grad_norm": 0.7094134092330933, "learning_rate": 1.1113170200110255e-05, "loss": 0.1563262939453125, "step": 96750 }, { "epoch": 0.48061353228097714, "grad_norm": 1.8627225160598755, "learning_rate": 1.1111571470244862e-05, "loss": 0.16160695552825927, "step": 96760 }, { "epoch": 0.4806632029643464, "grad_norm": 2.0306618213653564, "learning_rate": 1.1109972711612447e-05, "loss": 0.18342037200927735, "step": 96770 }, { "epoch": 0.48071287364771564, "grad_norm": 1.0884050130844116, "learning_rate": 1.1108373924254383e-05, "loss": 0.16291322708129882, "step": 96780 }, { "epoch": 0.4807625443310849, "grad_norm": 0.7608242034912109, "learning_rate": 1.1106775108212044e-05, "loss": 0.1705823063850403, "step": 96790 }, { "epoch": 0.48081221501445415, "grad_norm": 1.0951735973358154, "learning_rate": 1.1105176263526816e-05, "loss": 0.15949230194091796, "step": 96800 }, { "epoch": 0.48086188569782345, "grad_norm": 1.1430047750473022, "learning_rate": 1.1103577390240068e-05, "loss": 0.16419119834899903, "step": 96810 }, { "epoch": 0.4809115563811927, "grad_norm": 1.3224797248840332, "learning_rate": 1.1101978488393181e-05, "loss": 0.15851705074310302, "step": 96820 }, { "epoch": 0.48096122706456196, "grad_norm": 1.144363522529602, "learning_rate": 1.1100379558027536e-05, "loss": 0.15091179609298705, "step": 96830 }, { "epoch": 0.4810108977479312, "grad_norm": 1.1366511583328247, "learning_rate": 1.1098780599184508e-05, "loss": 0.16410524845123292, "step": 96840 }, { "epoch": 0.48106056843130046, "grad_norm": 1.6174771785736084, "learning_rate": 1.109718161190548e-05, "loss": 0.1739431142807007, "step": 96850 }, { "epoch": 0.4811102391146697, "grad_norm": 1.0513849258422852, "learning_rate": 1.1095582596231833e-05, "loss": 0.15945279598236084, "step": 96860 }, { "epoch": 0.481159909798039, "grad_norm": 0.7493961453437805, "learning_rate": 1.109398355220495e-05, "loss": 0.15592012405395508, "step": 96870 }, { "epoch": 0.48120958048140827, "grad_norm": 1.2879959344863892, "learning_rate": 1.1092384479866213e-05, "loss": 0.18498680591583253, "step": 96880 }, { "epoch": 0.4812592511647775, "grad_norm": 1.2173222303390503, "learning_rate": 1.1090785379257006e-05, "loss": 0.173089337348938, "step": 96890 }, { "epoch": 0.4813089218481468, "grad_norm": 2.078183889389038, "learning_rate": 1.1089186250418708e-05, "loss": 0.17754327058792113, "step": 96900 }, { "epoch": 0.481358592531516, "grad_norm": 0.9748377203941345, "learning_rate": 1.1087587093392712e-05, "loss": 0.16903618574142457, "step": 96910 }, { "epoch": 0.48140826321488533, "grad_norm": 1.0689923763275146, "learning_rate": 1.1085987908220396e-05, "loss": 0.16028838157653807, "step": 96920 }, { "epoch": 0.4814579338982546, "grad_norm": 2.1803414821624756, "learning_rate": 1.108438869494315e-05, "loss": 0.1753879427909851, "step": 96930 }, { "epoch": 0.48150760458162384, "grad_norm": 0.8149197697639465, "learning_rate": 1.1082789453602362e-05, "loss": 0.16069066524505615, "step": 96940 }, { "epoch": 0.4815572752649931, "grad_norm": 1.3735896348953247, "learning_rate": 1.1081190184239418e-05, "loss": 0.15967981815338134, "step": 96950 }, { "epoch": 0.48160694594836234, "grad_norm": 0.7880153059959412, "learning_rate": 1.1079590886895708e-05, "loss": 0.1743168592453003, "step": 96960 }, { "epoch": 0.4816566166317316, "grad_norm": 1.8048055171966553, "learning_rate": 1.107799156161262e-05, "loss": 0.16958277225494384, "step": 96970 }, { "epoch": 0.4817062873151009, "grad_norm": 1.3949897289276123, "learning_rate": 1.1076392208431541e-05, "loss": 0.17317943572998046, "step": 96980 }, { "epoch": 0.48175595799847015, "grad_norm": 2.163228988647461, "learning_rate": 1.1074792827393862e-05, "loss": 0.17449941635131835, "step": 96990 }, { "epoch": 0.4818056286818394, "grad_norm": 0.8385003805160522, "learning_rate": 1.1073193418540984e-05, "loss": 0.1616714358329773, "step": 97000 }, { "epoch": 0.48185529936520866, "grad_norm": 0.634880781173706, "learning_rate": 1.1071593981914287e-05, "loss": 0.16219041347503663, "step": 97010 }, { "epoch": 0.4819049700485779, "grad_norm": 1.0847851037979126, "learning_rate": 1.1069994517555172e-05, "loss": 0.1559392213821411, "step": 97020 }, { "epoch": 0.4819546407319472, "grad_norm": 1.2024074792861938, "learning_rate": 1.1068395025505025e-05, "loss": 0.1585959792137146, "step": 97030 }, { "epoch": 0.48200431141531647, "grad_norm": 0.7102449536323547, "learning_rate": 1.1066795505805249e-05, "loss": 0.1633187413215637, "step": 97040 }, { "epoch": 0.4820539820986857, "grad_norm": 1.0118860006332397, "learning_rate": 1.106519595849723e-05, "loss": 0.16820948123931884, "step": 97050 }, { "epoch": 0.48210365278205497, "grad_norm": 2.7092716693878174, "learning_rate": 1.1063596383622368e-05, "loss": 0.15170927047729493, "step": 97060 }, { "epoch": 0.4821533234654242, "grad_norm": 0.7415677905082703, "learning_rate": 1.1061996781222063e-05, "loss": 0.1685918927192688, "step": 97070 }, { "epoch": 0.4822029941487935, "grad_norm": 1.8643722534179688, "learning_rate": 1.1060397151337706e-05, "loss": 0.15617694854736328, "step": 97080 }, { "epoch": 0.4822526648321628, "grad_norm": 0.8129472136497498, "learning_rate": 1.1058797494010698e-05, "loss": 0.17196096181869508, "step": 97090 }, { "epoch": 0.48230233551553203, "grad_norm": 1.427894949913025, "learning_rate": 1.1057197809282437e-05, "loss": 0.15641870498657226, "step": 97100 }, { "epoch": 0.4823520061989013, "grad_norm": 1.7680442333221436, "learning_rate": 1.105559809719432e-05, "loss": 0.17582812309265136, "step": 97110 }, { "epoch": 0.48240167688227054, "grad_norm": 1.0383093357086182, "learning_rate": 1.1053998357787748e-05, "loss": 0.16780465841293335, "step": 97120 }, { "epoch": 0.4824513475656398, "grad_norm": 0.8449713587760925, "learning_rate": 1.1052398591104123e-05, "loss": 0.1778266668319702, "step": 97130 }, { "epoch": 0.4825010182490091, "grad_norm": 0.9979696273803711, "learning_rate": 1.1050798797184845e-05, "loss": 0.1572441816329956, "step": 97140 }, { "epoch": 0.48255068893237835, "grad_norm": 0.9012149572372437, "learning_rate": 1.1049198976071317e-05, "loss": 0.1632799744606018, "step": 97150 }, { "epoch": 0.4826003596157476, "grad_norm": 0.9230563640594482, "learning_rate": 1.1047599127804944e-05, "loss": 0.1549208641052246, "step": 97160 }, { "epoch": 0.48265003029911685, "grad_norm": 0.7538444995880127, "learning_rate": 1.1045999252427126e-05, "loss": 0.1691794753074646, "step": 97170 }, { "epoch": 0.4826997009824861, "grad_norm": 1.347135305404663, "learning_rate": 1.1044399349979266e-05, "loss": 0.1540836811065674, "step": 97180 }, { "epoch": 0.48274937166585535, "grad_norm": 0.8981205821037292, "learning_rate": 1.1042799420502772e-05, "loss": 0.15047478675842285, "step": 97190 }, { "epoch": 0.48279904234922466, "grad_norm": 1.1800599098205566, "learning_rate": 1.1041199464039047e-05, "loss": 0.16653884649276735, "step": 97200 }, { "epoch": 0.4828487130325939, "grad_norm": 0.9846938252449036, "learning_rate": 1.1039599480629499e-05, "loss": 0.16245641708374023, "step": 97210 }, { "epoch": 0.48289838371596316, "grad_norm": 0.6327434778213501, "learning_rate": 1.1037999470315534e-05, "loss": 0.1556382417678833, "step": 97220 }, { "epoch": 0.4829480543993324, "grad_norm": 1.1198840141296387, "learning_rate": 1.1036399433138563e-05, "loss": 0.16843740940093993, "step": 97230 }, { "epoch": 0.48299772508270167, "grad_norm": 1.5380192995071411, "learning_rate": 1.1034799369139988e-05, "loss": 0.1734388589859009, "step": 97240 }, { "epoch": 0.483047395766071, "grad_norm": 1.0244735479354858, "learning_rate": 1.1033199278361224e-05, "loss": 0.16671922206878662, "step": 97250 }, { "epoch": 0.4830970664494402, "grad_norm": 0.8556439876556396, "learning_rate": 1.1031599160843678e-05, "loss": 0.1531727910041809, "step": 97260 }, { "epoch": 0.4831467371328095, "grad_norm": 0.7255716323852539, "learning_rate": 1.1029999016628758e-05, "loss": 0.16311390399932862, "step": 97270 }, { "epoch": 0.48319640781617873, "grad_norm": 1.563310146331787, "learning_rate": 1.102839884575788e-05, "loss": 0.173147714138031, "step": 97280 }, { "epoch": 0.483246078499548, "grad_norm": 0.7952736616134644, "learning_rate": 1.1026798648272452e-05, "loss": 0.1558382987976074, "step": 97290 }, { "epoch": 0.48329574918291723, "grad_norm": 1.4047034978866577, "learning_rate": 1.1025198424213887e-05, "loss": 0.1678142547607422, "step": 97300 }, { "epoch": 0.48334541986628654, "grad_norm": 0.9923181533813477, "learning_rate": 1.1023598173623602e-05, "loss": 0.16034036874771118, "step": 97310 }, { "epoch": 0.4833950905496558, "grad_norm": 0.7087533473968506, "learning_rate": 1.1021997896543006e-05, "loss": 0.16362178325653076, "step": 97320 }, { "epoch": 0.48344476123302504, "grad_norm": 0.9655755162239075, "learning_rate": 1.1020397593013515e-05, "loss": 0.17469983100891112, "step": 97330 }, { "epoch": 0.4834944319163943, "grad_norm": 1.0028905868530273, "learning_rate": 1.1018797263076543e-05, "loss": 0.161641526222229, "step": 97340 }, { "epoch": 0.48354410259976355, "grad_norm": 1.072987675666809, "learning_rate": 1.1017196906773511e-05, "loss": 0.17636520862579347, "step": 97350 }, { "epoch": 0.48359377328313286, "grad_norm": 1.1554826498031616, "learning_rate": 1.101559652414583e-05, "loss": 0.16224281787872313, "step": 97360 }, { "epoch": 0.4836434439665021, "grad_norm": 1.1197664737701416, "learning_rate": 1.101399611523492e-05, "loss": 0.16626369953155518, "step": 97370 }, { "epoch": 0.48369311464987136, "grad_norm": 0.9796545505523682, "learning_rate": 1.1012395680082197e-05, "loss": 0.1602482795715332, "step": 97380 }, { "epoch": 0.4837427853332406, "grad_norm": 1.2200729846954346, "learning_rate": 1.1010795218729085e-05, "loss": 0.16272227764129638, "step": 97390 }, { "epoch": 0.48379245601660986, "grad_norm": 0.9952968955039978, "learning_rate": 1.1009194731216992e-05, "loss": 0.16789206266403198, "step": 97400 }, { "epoch": 0.4838421266999791, "grad_norm": 1.159982681274414, "learning_rate": 1.100759421758735e-05, "loss": 0.1637096881866455, "step": 97410 }, { "epoch": 0.4838917973833484, "grad_norm": 1.1591800451278687, "learning_rate": 1.1005993677881577e-05, "loss": 0.16408495903015136, "step": 97420 }, { "epoch": 0.4839414680667177, "grad_norm": 1.592842698097229, "learning_rate": 1.1004393112141089e-05, "loss": 0.16735486984252929, "step": 97430 }, { "epoch": 0.4839911387500869, "grad_norm": 1.490280270576477, "learning_rate": 1.1002792520407311e-05, "loss": 0.16987240314483643, "step": 97440 }, { "epoch": 0.4840408094334562, "grad_norm": 1.2249865531921387, "learning_rate": 1.1001191902721666e-05, "loss": 0.16209374666213988, "step": 97450 }, { "epoch": 0.48409048011682543, "grad_norm": 0.7575695514678955, "learning_rate": 1.0999591259125575e-05, "loss": 0.17281414270401002, "step": 97460 }, { "epoch": 0.48414015080019474, "grad_norm": 0.6243166923522949, "learning_rate": 1.0997990589660468e-05, "loss": 0.16428996324539186, "step": 97470 }, { "epoch": 0.484189821483564, "grad_norm": 1.2540979385375977, "learning_rate": 1.0996389894367762e-05, "loss": 0.16541604995727538, "step": 97480 }, { "epoch": 0.48423949216693324, "grad_norm": 0.6254497170448303, "learning_rate": 1.0994789173288889e-05, "loss": 0.15636024475097657, "step": 97490 }, { "epoch": 0.4842891628503025, "grad_norm": 1.06781005859375, "learning_rate": 1.099318842646527e-05, "loss": 0.16699789762496947, "step": 97500 }, { "epoch": 0.48433883353367174, "grad_norm": 0.7715321183204651, "learning_rate": 1.0991587653938339e-05, "loss": 0.1611255884170532, "step": 97510 }, { "epoch": 0.484388504217041, "grad_norm": 1.8505018949508667, "learning_rate": 1.0989986855749512e-05, "loss": 0.14999696016311645, "step": 97520 }, { "epoch": 0.4844381749004103, "grad_norm": 0.901852548122406, "learning_rate": 1.0988386031940226e-05, "loss": 0.19865179061889648, "step": 97530 }, { "epoch": 0.48448784558377955, "grad_norm": 0.9026694893836975, "learning_rate": 1.0986785182551907e-05, "loss": 0.16968812942504882, "step": 97540 }, { "epoch": 0.4845375162671488, "grad_norm": 1.6243999004364014, "learning_rate": 1.0985184307625982e-05, "loss": 0.1624928593635559, "step": 97550 }, { "epoch": 0.48458718695051806, "grad_norm": 0.5752531290054321, "learning_rate": 1.0983583407203886e-05, "loss": 0.16656873226165772, "step": 97560 }, { "epoch": 0.4846368576338873, "grad_norm": 0.8865007162094116, "learning_rate": 1.0981982481327047e-05, "loss": 0.1665518879890442, "step": 97570 }, { "epoch": 0.4846865283172566, "grad_norm": 1.1621887683868408, "learning_rate": 1.0980381530036895e-05, "loss": 0.1830647826194763, "step": 97580 }, { "epoch": 0.48473619900062587, "grad_norm": 1.8177865743637085, "learning_rate": 1.0978780553374864e-05, "loss": 0.16433731317520142, "step": 97590 }, { "epoch": 0.4847858696839951, "grad_norm": 1.854438304901123, "learning_rate": 1.0977179551382383e-05, "loss": 0.17335630655288697, "step": 97600 }, { "epoch": 0.48483554036736437, "grad_norm": 1.1474061012268066, "learning_rate": 1.0975578524100894e-05, "loss": 0.1519020676612854, "step": 97610 }, { "epoch": 0.4848852110507336, "grad_norm": 0.9206736087799072, "learning_rate": 1.097397747157182e-05, "loss": 0.16258853673934937, "step": 97620 }, { "epoch": 0.4849348817341029, "grad_norm": 0.9094113707542419, "learning_rate": 1.0972376393836603e-05, "loss": 0.16668205261230468, "step": 97630 }, { "epoch": 0.4849845524174722, "grad_norm": 1.5254490375518799, "learning_rate": 1.0970775290936674e-05, "loss": 0.1668906807899475, "step": 97640 }, { "epoch": 0.48503422310084143, "grad_norm": 2.0430667400360107, "learning_rate": 1.0969174162913473e-05, "loss": 0.16877390146255494, "step": 97650 }, { "epoch": 0.4850838937842107, "grad_norm": 1.1862683296203613, "learning_rate": 1.0967573009808432e-05, "loss": 0.17454798221588136, "step": 97660 }, { "epoch": 0.48513356446757994, "grad_norm": 0.8466509580612183, "learning_rate": 1.0965971831662991e-05, "loss": 0.16251344680786134, "step": 97670 }, { "epoch": 0.4851832351509492, "grad_norm": 1.0535281896591187, "learning_rate": 1.0964370628518588e-05, "loss": 0.1855900526046753, "step": 97680 }, { "epoch": 0.4852329058343185, "grad_norm": 0.7816407084465027, "learning_rate": 1.0962769400416662e-05, "loss": 0.1655228853225708, "step": 97690 }, { "epoch": 0.48528257651768775, "grad_norm": 1.0423413515090942, "learning_rate": 1.0961168147398651e-05, "loss": 0.16889688968658448, "step": 97700 }, { "epoch": 0.485332247201057, "grad_norm": 1.0794368982315063, "learning_rate": 1.0959566869505993e-05, "loss": 0.15858241319656372, "step": 97710 }, { "epoch": 0.48538191788442625, "grad_norm": 0.7272047400474548, "learning_rate": 1.0957965566780132e-05, "loss": 0.15064723491668702, "step": 97720 }, { "epoch": 0.4854315885677955, "grad_norm": 0.667765736579895, "learning_rate": 1.0956364239262505e-05, "loss": 0.16814608573913575, "step": 97730 }, { "epoch": 0.48548125925116475, "grad_norm": 0.7169165015220642, "learning_rate": 1.0954762886994557e-05, "loss": 0.16130048036575317, "step": 97740 }, { "epoch": 0.48553092993453406, "grad_norm": 0.9408273696899414, "learning_rate": 1.0953161510017728e-05, "loss": 0.16067588329315186, "step": 97750 }, { "epoch": 0.4855806006179033, "grad_norm": 1.4202544689178467, "learning_rate": 1.0951560108373469e-05, "loss": 0.16589808464050293, "step": 97760 }, { "epoch": 0.48563027130127256, "grad_norm": 2.1843767166137695, "learning_rate": 1.0949958682103212e-05, "loss": 0.1529792070388794, "step": 97770 }, { "epoch": 0.4856799419846418, "grad_norm": 2.274641513824463, "learning_rate": 1.0948357231248405e-05, "loss": 0.16400582790374757, "step": 97780 }, { "epoch": 0.48572961266801107, "grad_norm": 1.8227627277374268, "learning_rate": 1.0946755755850496e-05, "loss": 0.15564234256744386, "step": 97790 }, { "epoch": 0.4857792833513804, "grad_norm": 0.9065441489219666, "learning_rate": 1.0945154255950928e-05, "loss": 0.15243486166000367, "step": 97800 }, { "epoch": 0.4858289540347496, "grad_norm": 1.4455262422561646, "learning_rate": 1.0943552731591147e-05, "loss": 0.16607222557067872, "step": 97810 }, { "epoch": 0.4858786247181189, "grad_norm": 1.6021217107772827, "learning_rate": 1.0941951182812602e-05, "loss": 0.1628196120262146, "step": 97820 }, { "epoch": 0.48592829540148813, "grad_norm": 2.0810635089874268, "learning_rate": 1.094034960965674e-05, "loss": 0.16050422191619873, "step": 97830 }, { "epoch": 0.4859779660848574, "grad_norm": 1.03205144405365, "learning_rate": 1.0938748012165008e-05, "loss": 0.15994898080825806, "step": 97840 }, { "epoch": 0.48602763676822663, "grad_norm": 0.794253408908844, "learning_rate": 1.0937146390378854e-05, "loss": 0.1731892704963684, "step": 97850 }, { "epoch": 0.48607730745159594, "grad_norm": 0.9071798920631409, "learning_rate": 1.0935544744339727e-05, "loss": 0.1509371280670166, "step": 97860 }, { "epoch": 0.4861269781349652, "grad_norm": 0.7754942178726196, "learning_rate": 1.0933943074089079e-05, "loss": 0.15506160259246826, "step": 97870 }, { "epoch": 0.48617664881833444, "grad_norm": 2.5330967903137207, "learning_rate": 1.093234137966836e-05, "loss": 0.1624239444732666, "step": 97880 }, { "epoch": 0.4862263195017037, "grad_norm": 0.6367204189300537, "learning_rate": 1.093073966111902e-05, "loss": 0.160580050945282, "step": 97890 }, { "epoch": 0.48627599018507295, "grad_norm": 0.929655134677887, "learning_rate": 1.0929137918482516e-05, "loss": 0.15556540489196777, "step": 97900 }, { "epoch": 0.48632566086844226, "grad_norm": 1.2086424827575684, "learning_rate": 1.0927536151800293e-05, "loss": 0.1625605821609497, "step": 97910 }, { "epoch": 0.4863753315518115, "grad_norm": 0.7804058194160461, "learning_rate": 1.0925934361113808e-05, "loss": 0.16684017181396485, "step": 97920 }, { "epoch": 0.48642500223518076, "grad_norm": 1.0090081691741943, "learning_rate": 1.0924332546464513e-05, "loss": 0.15343432426452636, "step": 97930 }, { "epoch": 0.48647467291855, "grad_norm": 0.7696762681007385, "learning_rate": 1.0922730707893866e-05, "loss": 0.16596808433532714, "step": 97940 }, { "epoch": 0.48652434360191926, "grad_norm": 2.3748230934143066, "learning_rate": 1.0921128845443317e-05, "loss": 0.1641615867614746, "step": 97950 }, { "epoch": 0.4865740142852885, "grad_norm": 1.258543848991394, "learning_rate": 1.0919526959154325e-05, "loss": 0.16437410116195678, "step": 97960 }, { "epoch": 0.4866236849686578, "grad_norm": 1.573598861694336, "learning_rate": 1.0917925049068344e-05, "loss": 0.16906049251556396, "step": 97970 }, { "epoch": 0.4866733556520271, "grad_norm": 1.2143974304199219, "learning_rate": 1.0916323115226832e-05, "loss": 0.15079305171966553, "step": 97980 }, { "epoch": 0.4867230263353963, "grad_norm": 1.723987340927124, "learning_rate": 1.0914721157671246e-05, "loss": 0.16593728065490723, "step": 97990 }, { "epoch": 0.4867726970187656, "grad_norm": 1.7133249044418335, "learning_rate": 1.0913119176443044e-05, "loss": 0.17475666999816894, "step": 98000 }, { "epoch": 0.48682236770213483, "grad_norm": 0.8811006546020508, "learning_rate": 1.0911517171583685e-05, "loss": 0.16538304090499878, "step": 98010 }, { "epoch": 0.4868720383855041, "grad_norm": 2.2332308292388916, "learning_rate": 1.0909915143134629e-05, "loss": 0.16221487522125244, "step": 98020 }, { "epoch": 0.4869217090688734, "grad_norm": 1.2660044431686401, "learning_rate": 1.0908313091137334e-05, "loss": 0.15623693466186522, "step": 98030 }, { "epoch": 0.48697137975224264, "grad_norm": 1.1087077856063843, "learning_rate": 1.0906711015633262e-05, "loss": 0.16254204511642456, "step": 98040 }, { "epoch": 0.4870210504356119, "grad_norm": 1.3360284566879272, "learning_rate": 1.0905108916663874e-05, "loss": 0.1728522300720215, "step": 98050 }, { "epoch": 0.48707072111898114, "grad_norm": 1.7555063962936401, "learning_rate": 1.0903506794270631e-05, "loss": 0.1629873752593994, "step": 98060 }, { "epoch": 0.4871203918023504, "grad_norm": 1.002339482307434, "learning_rate": 1.0901904648494991e-05, "loss": 0.17143850326538085, "step": 98070 }, { "epoch": 0.4871700624857197, "grad_norm": 0.8411861062049866, "learning_rate": 1.0900302479378427e-05, "loss": 0.17297934293746947, "step": 98080 }, { "epoch": 0.48721973316908895, "grad_norm": 1.2939813137054443, "learning_rate": 1.0898700286962399e-05, "loss": 0.1648135304450989, "step": 98090 }, { "epoch": 0.4872694038524582, "grad_norm": 1.8332905769348145, "learning_rate": 1.0897098071288364e-05, "loss": 0.1665476679801941, "step": 98100 }, { "epoch": 0.48731907453582746, "grad_norm": 1.2779325246810913, "learning_rate": 1.0895495832397794e-05, "loss": 0.15637404918670655, "step": 98110 }, { "epoch": 0.4873687452191967, "grad_norm": 0.7710250616073608, "learning_rate": 1.089389357033215e-05, "loss": 0.17387109994888306, "step": 98120 }, { "epoch": 0.48741841590256596, "grad_norm": 0.6833867430686951, "learning_rate": 1.0892291285132899e-05, "loss": 0.16827174425125122, "step": 98130 }, { "epoch": 0.48746808658593527, "grad_norm": 2.346261978149414, "learning_rate": 1.0890688976841508e-05, "loss": 0.14444918632507325, "step": 98140 }, { "epoch": 0.4875177572693045, "grad_norm": 1.3228414058685303, "learning_rate": 1.0889086645499447e-05, "loss": 0.16663179397583008, "step": 98150 }, { "epoch": 0.48756742795267377, "grad_norm": 1.1117738485336304, "learning_rate": 1.088748429114818e-05, "loss": 0.15604076385498047, "step": 98160 }, { "epoch": 0.487617098636043, "grad_norm": 0.968887448310852, "learning_rate": 1.0885881913829178e-05, "loss": 0.1585124611854553, "step": 98170 }, { "epoch": 0.4876667693194123, "grad_norm": 0.7446886301040649, "learning_rate": 1.0884279513583907e-05, "loss": 0.17516074180603028, "step": 98180 }, { "epoch": 0.4877164400027816, "grad_norm": 1.2332221269607544, "learning_rate": 1.0882677090453837e-05, "loss": 0.1650160789489746, "step": 98190 }, { "epoch": 0.48776611068615083, "grad_norm": 1.546111822128296, "learning_rate": 1.0881074644480438e-05, "loss": 0.16689825057983398, "step": 98200 }, { "epoch": 0.4878157813695201, "grad_norm": 1.481576681137085, "learning_rate": 1.0879472175705184e-05, "loss": 0.15478169918060303, "step": 98210 }, { "epoch": 0.48786545205288934, "grad_norm": 0.812466561794281, "learning_rate": 1.0877869684169541e-05, "loss": 0.16851348876953126, "step": 98220 }, { "epoch": 0.4879151227362586, "grad_norm": 2.5193188190460205, "learning_rate": 1.0876267169914986e-05, "loss": 0.15773229598999022, "step": 98230 }, { "epoch": 0.48796479341962784, "grad_norm": 2.010892868041992, "learning_rate": 1.087466463298299e-05, "loss": 0.172732675075531, "step": 98240 }, { "epoch": 0.48801446410299715, "grad_norm": 1.201335072517395, "learning_rate": 1.087306207341502e-05, "loss": 0.16501343250274658, "step": 98250 }, { "epoch": 0.4880641347863664, "grad_norm": 0.8294926881790161, "learning_rate": 1.087145949125256e-05, "loss": 0.15924793481826782, "step": 98260 }, { "epoch": 0.48811380546973565, "grad_norm": 1.3136717081069946, "learning_rate": 1.0869856886537079e-05, "loss": 0.16984773874282838, "step": 98270 }, { "epoch": 0.4881634761531049, "grad_norm": 1.760278582572937, "learning_rate": 1.0868254259310048e-05, "loss": 0.16309586763381959, "step": 98280 }, { "epoch": 0.48821314683647415, "grad_norm": 1.0481339693069458, "learning_rate": 1.0866651609612946e-05, "loss": 0.16803224086761476, "step": 98290 }, { "epoch": 0.48826281751984346, "grad_norm": 1.310957908630371, "learning_rate": 1.086504893748725e-05, "loss": 0.16588430404663085, "step": 98300 }, { "epoch": 0.4883124882032127, "grad_norm": 0.9666761159896851, "learning_rate": 1.0863446242974436e-05, "loss": 0.16218011379241942, "step": 98310 }, { "epoch": 0.48836215888658197, "grad_norm": 1.616204023361206, "learning_rate": 1.0861843526115978e-05, "loss": 0.1625354766845703, "step": 98320 }, { "epoch": 0.4884118295699512, "grad_norm": 1.2848825454711914, "learning_rate": 1.0860240786953357e-05, "loss": 0.17419180870056153, "step": 98330 }, { "epoch": 0.48846150025332047, "grad_norm": 1.0208148956298828, "learning_rate": 1.085863802552805e-05, "loss": 0.14840319156646728, "step": 98340 }, { "epoch": 0.4885111709366897, "grad_norm": 0.6306906342506409, "learning_rate": 1.0857035241881537e-05, "loss": 0.16031486988067628, "step": 98350 }, { "epoch": 0.488560841620059, "grad_norm": 0.7601355314254761, "learning_rate": 1.0855432436055297e-05, "loss": 0.15440351963043214, "step": 98360 }, { "epoch": 0.4886105123034283, "grad_norm": 1.0193153619766235, "learning_rate": 1.085382960809081e-05, "loss": 0.16224406957626342, "step": 98370 }, { "epoch": 0.48866018298679753, "grad_norm": 0.7457737922668457, "learning_rate": 1.0852226758029556e-05, "loss": 0.15650722980499268, "step": 98380 }, { "epoch": 0.4887098536701668, "grad_norm": 1.960699200630188, "learning_rate": 1.0850623885913013e-05, "loss": 0.1491692543029785, "step": 98390 }, { "epoch": 0.48875952435353603, "grad_norm": 2.023604393005371, "learning_rate": 1.084902099178267e-05, "loss": 0.17522156238555908, "step": 98400 }, { "epoch": 0.48880919503690534, "grad_norm": 1.4253864288330078, "learning_rate": 1.084741807568e-05, "loss": 0.15294508934020995, "step": 98410 }, { "epoch": 0.4888588657202746, "grad_norm": 0.7044122219085693, "learning_rate": 1.0845815137646496e-05, "loss": 0.1790279269218445, "step": 98420 }, { "epoch": 0.48890853640364385, "grad_norm": 1.5910136699676514, "learning_rate": 1.0844212177723636e-05, "loss": 0.15471878051757812, "step": 98430 }, { "epoch": 0.4889582070870131, "grad_norm": 1.5596572160720825, "learning_rate": 1.0842609195952905e-05, "loss": 0.16491992473602296, "step": 98440 }, { "epoch": 0.48900787777038235, "grad_norm": 0.8706914186477661, "learning_rate": 1.0841006192375783e-05, "loss": 0.158797287940979, "step": 98450 }, { "epoch": 0.4890575484537516, "grad_norm": 0.7349331974983215, "learning_rate": 1.083940316703376e-05, "loss": 0.15921833515167236, "step": 98460 }, { "epoch": 0.4891072191371209, "grad_norm": 1.6499364376068115, "learning_rate": 1.083780011996832e-05, "loss": 0.16685402393341064, "step": 98470 }, { "epoch": 0.48915688982049016, "grad_norm": 1.2054718732833862, "learning_rate": 1.0836197051220951e-05, "loss": 0.1502208113670349, "step": 98480 }, { "epoch": 0.4892065605038594, "grad_norm": 1.1490705013275146, "learning_rate": 1.083459396083314e-05, "loss": 0.15097843408584594, "step": 98490 }, { "epoch": 0.48925623118722866, "grad_norm": 1.1906607151031494, "learning_rate": 1.0832990848846369e-05, "loss": 0.15815427303314208, "step": 98500 }, { "epoch": 0.4893059018705979, "grad_norm": 0.711931049823761, "learning_rate": 1.0831387715302133e-05, "loss": 0.15349507331848145, "step": 98510 }, { "epoch": 0.4893555725539672, "grad_norm": 0.8842736482620239, "learning_rate": 1.0829784560241917e-05, "loss": 0.17810311317443847, "step": 98520 }, { "epoch": 0.4894052432373365, "grad_norm": 1.4081565141677856, "learning_rate": 1.082818138370721e-05, "loss": 0.15946383476257325, "step": 98530 }, { "epoch": 0.4894549139207057, "grad_norm": 1.294588565826416, "learning_rate": 1.0826578185739498e-05, "loss": 0.16951289176940917, "step": 98540 }, { "epoch": 0.489504584604075, "grad_norm": 0.8567485809326172, "learning_rate": 1.082497496638028e-05, "loss": 0.16185649633407592, "step": 98550 }, { "epoch": 0.48955425528744423, "grad_norm": 1.9468193054199219, "learning_rate": 1.0823371725671038e-05, "loss": 0.17772603034973145, "step": 98560 }, { "epoch": 0.4896039259708135, "grad_norm": 1.011975884437561, "learning_rate": 1.082176846365327e-05, "loss": 0.17015480995178223, "step": 98570 }, { "epoch": 0.4896535966541828, "grad_norm": 0.8218734860420227, "learning_rate": 1.0820165180368462e-05, "loss": 0.16709762811660767, "step": 98580 }, { "epoch": 0.48970326733755204, "grad_norm": 0.7740026116371155, "learning_rate": 1.0818561875858112e-05, "loss": 0.17539801597595214, "step": 98590 }, { "epoch": 0.4897529380209213, "grad_norm": 1.1270437240600586, "learning_rate": 1.0816958550163708e-05, "loss": 0.1628039836883545, "step": 98600 }, { "epoch": 0.48980260870429054, "grad_norm": 1.4915462732315063, "learning_rate": 1.0815355203326746e-05, "loss": 0.15819911956787108, "step": 98610 }, { "epoch": 0.4898522793876598, "grad_norm": 1.4679787158966064, "learning_rate": 1.0813751835388717e-05, "loss": 0.16190026998519896, "step": 98620 }, { "epoch": 0.4899019500710291, "grad_norm": 1.437031626701355, "learning_rate": 1.0812148446391117e-05, "loss": 0.17857731580734254, "step": 98630 }, { "epoch": 0.48995162075439835, "grad_norm": 1.2655242681503296, "learning_rate": 1.0810545036375446e-05, "loss": 0.16278226375579835, "step": 98640 }, { "epoch": 0.4900012914377676, "grad_norm": 1.7870675325393677, "learning_rate": 1.0808941605383192e-05, "loss": 0.1655022144317627, "step": 98650 }, { "epoch": 0.49005096212113686, "grad_norm": 1.3070752620697021, "learning_rate": 1.0807338153455856e-05, "loss": 0.17102441787719727, "step": 98660 }, { "epoch": 0.4901006328045061, "grad_norm": 0.7635286450386047, "learning_rate": 1.0805734680634931e-05, "loss": 0.1536371350288391, "step": 98670 }, { "epoch": 0.49015030348787536, "grad_norm": 0.9353456497192383, "learning_rate": 1.0804131186961917e-05, "loss": 0.1644596576690674, "step": 98680 }, { "epoch": 0.49019997417124467, "grad_norm": 0.9702944755554199, "learning_rate": 1.0802527672478313e-05, "loss": 0.16921026706695558, "step": 98690 }, { "epoch": 0.4902496448546139, "grad_norm": 0.9809480309486389, "learning_rate": 1.0800924137225617e-05, "loss": 0.1579911708831787, "step": 98700 }, { "epoch": 0.49029931553798317, "grad_norm": 2.1194872856140137, "learning_rate": 1.0799320581245323e-05, "loss": 0.1563178539276123, "step": 98710 }, { "epoch": 0.4903489862213524, "grad_norm": 1.1363410949707031, "learning_rate": 1.0797717004578936e-05, "loss": 0.16421728134155272, "step": 98720 }, { "epoch": 0.4903986569047217, "grad_norm": 1.1949881315231323, "learning_rate": 1.0796113407267951e-05, "loss": 0.16544339656829835, "step": 98730 }, { "epoch": 0.490448327588091, "grad_norm": 0.7267498970031738, "learning_rate": 1.079450978935387e-05, "loss": 0.15856354236602782, "step": 98740 }, { "epoch": 0.49049799827146023, "grad_norm": 1.0429579019546509, "learning_rate": 1.0792906150878199e-05, "loss": 0.1635247588157654, "step": 98750 }, { "epoch": 0.4905476689548295, "grad_norm": 1.0357236862182617, "learning_rate": 1.0791302491882436e-05, "loss": 0.16382015943527223, "step": 98760 }, { "epoch": 0.49059733963819874, "grad_norm": 0.7217621207237244, "learning_rate": 1.0789698812408082e-05, "loss": 0.17232677936553956, "step": 98770 }, { "epoch": 0.490647010321568, "grad_norm": 1.1851677894592285, "learning_rate": 1.0788095112496642e-05, "loss": 0.1634766697883606, "step": 98780 }, { "epoch": 0.49069668100493724, "grad_norm": 1.6711286306381226, "learning_rate": 1.0786491392189614e-05, "loss": 0.158986496925354, "step": 98790 }, { "epoch": 0.49074635168830655, "grad_norm": 1.1325901746749878, "learning_rate": 1.0784887651528504e-05, "loss": 0.16386690139770507, "step": 98800 }, { "epoch": 0.4907960223716758, "grad_norm": 0.6844885945320129, "learning_rate": 1.0783283890554819e-05, "loss": 0.16647348403930665, "step": 98810 }, { "epoch": 0.49084569305504505, "grad_norm": 1.3227254152297974, "learning_rate": 1.0781680109310062e-05, "loss": 0.15599484443664552, "step": 98820 }, { "epoch": 0.4908953637384143, "grad_norm": 1.2500028610229492, "learning_rate": 1.0780076307835739e-05, "loss": 0.15849952697753905, "step": 98830 }, { "epoch": 0.49094503442178355, "grad_norm": 0.9301629662513733, "learning_rate": 1.0778472486173353e-05, "loss": 0.16763887405395508, "step": 98840 }, { "epoch": 0.49099470510515286, "grad_norm": 1.0869171619415283, "learning_rate": 1.0776868644364413e-05, "loss": 0.15598709583282472, "step": 98850 }, { "epoch": 0.4910443757885221, "grad_norm": 1.0549285411834717, "learning_rate": 1.0775264782450425e-05, "loss": 0.16308633089065552, "step": 98860 }, { "epoch": 0.49109404647189137, "grad_norm": 0.9053800702095032, "learning_rate": 1.0773660900472893e-05, "loss": 0.16877052783966065, "step": 98870 }, { "epoch": 0.4911437171552606, "grad_norm": 1.4103562831878662, "learning_rate": 1.0772056998473332e-05, "loss": 0.16290659904479982, "step": 98880 }, { "epoch": 0.49119338783862987, "grad_norm": 0.966844916343689, "learning_rate": 1.0770453076493246e-05, "loss": 0.1615970492362976, "step": 98890 }, { "epoch": 0.4912430585219991, "grad_norm": 1.0748075246810913, "learning_rate": 1.0768849134574142e-05, "loss": 0.1618030548095703, "step": 98900 }, { "epoch": 0.49129272920536843, "grad_norm": 0.751953661441803, "learning_rate": 1.0767245172757534e-05, "loss": 0.15258004665374755, "step": 98910 }, { "epoch": 0.4913423998887377, "grad_norm": 1.4649772644042969, "learning_rate": 1.0765641191084929e-05, "loss": 0.1771398067474365, "step": 98920 }, { "epoch": 0.49139207057210693, "grad_norm": 1.8498997688293457, "learning_rate": 1.0764037189597836e-05, "loss": 0.16783480644226073, "step": 98930 }, { "epoch": 0.4914417412554762, "grad_norm": 1.6333307027816772, "learning_rate": 1.0762433168337768e-05, "loss": 0.153632652759552, "step": 98940 }, { "epoch": 0.49149141193884543, "grad_norm": 1.4534838199615479, "learning_rate": 1.0760829127346238e-05, "loss": 0.183011257648468, "step": 98950 }, { "epoch": 0.49154108262221474, "grad_norm": 1.2260057926177979, "learning_rate": 1.0759225066664756e-05, "loss": 0.15917448997497557, "step": 98960 }, { "epoch": 0.491590753305584, "grad_norm": 1.1068061590194702, "learning_rate": 1.0757620986334834e-05, "loss": 0.1575433850288391, "step": 98970 }, { "epoch": 0.49164042398895325, "grad_norm": 0.7927934527397156, "learning_rate": 1.0756016886397984e-05, "loss": 0.1612175703048706, "step": 98980 }, { "epoch": 0.4916900946723225, "grad_norm": 3.111990213394165, "learning_rate": 1.0754412766895724e-05, "loss": 0.17114450931549072, "step": 98990 }, { "epoch": 0.49173976535569175, "grad_norm": 1.6614865064620972, "learning_rate": 1.075280862786956e-05, "loss": 0.15560026168823243, "step": 99000 } ], "logging_steps": 10, "max_steps": 201326, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 1000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 5.8202259958296425e+20, "train_batch_size": 1, "trial_name": null, "trial_params": null }