{
  "best_metric": 3.9943079948425293,
  "best_model_checkpoint": "bert_base_lda_50_v1_book/checkpoint-580000",
  "epoch": 25.0,
  "eval_steps": 10000,
  "global_step": 593075,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.021076592336551025,
      "grad_norm": 2.334777593612671,
      "learning_rate": 5e-06,
      "loss": 12.8342,
      "step": 500
    },
    {
      "epoch": 0.04215318467310205,
      "grad_norm": 1.1738380193710327,
      "learning_rate": 1e-05,
      "loss": 10.5932,
      "step": 1000
    },
    {
      "epoch": 0.06322977700965308,
      "grad_norm": 1.2440646886825562,
      "learning_rate": 1.5e-05,
      "loss": 9.9427,
      "step": 1500
    },
    {
      "epoch": 0.0843063693462041,
      "grad_norm": 1.1406948566436768,
      "learning_rate": 2e-05,
      "loss": 9.8215,
      "step": 2000
    },
    {
      "epoch": 0.10538296168275513,
      "grad_norm": 1.0391911268234253,
      "learning_rate": 2.5e-05,
      "loss": 9.7621,
      "step": 2500
    },
    {
      "epoch": 0.12645955401930617,
      "grad_norm": 0.8712645769119263,
      "learning_rate": 3e-05,
      "loss": 9.7232,
      "step": 3000
    },
    {
      "epoch": 0.14753614635585718,
      "grad_norm": 0.9752113819122314,
      "learning_rate": 3.5e-05,
      "loss": 9.6935,
      "step": 3500
    },
    {
      "epoch": 0.1686127386924082,
      "grad_norm": 0.8634639382362366,
      "learning_rate": 4e-05,
      "loss": 9.6672,
      "step": 4000
    },
    {
      "epoch": 0.18968933102895924,
      "grad_norm": 1.0161821842193604,
      "learning_rate": 4.5e-05,
      "loss": 9.6432,
      "step": 4500
    },
    {
      "epoch": 0.21076592336551025,
      "grad_norm": 0.9344072341918945,
      "learning_rate": 5e-05,
      "loss": 9.6277,
      "step": 5000
    },
    {
      "epoch": 0.2318425157020613,
      "grad_norm": 0.961492657661438,
      "learning_rate": 5.500000000000001e-05,
      "loss": 9.6166,
      "step": 5500
    },
    {
      "epoch": 0.25291910803861234,
      "grad_norm": 0.8770144581794739,
      "learning_rate": 6e-05,
      "loss": 9.5981,
      "step": 6000
    },
    {
      "epoch": 0.27399570037516335,
      "grad_norm": 1.0077821016311646,
      "learning_rate": 6.500000000000001e-05,
      "loss": 9.5929,
      "step": 6500
    },
    {
      "epoch": 0.29507229271171437,
      "grad_norm": 0.8953953981399536,
      "learning_rate": 7e-05,
      "loss": 9.5665,
      "step": 7000
    },
    {
      "epoch": 0.3161488850482654,
      "grad_norm": 0.9563552141189575,
      "learning_rate": 7.500000000000001e-05,
      "loss": 9.5482,
      "step": 7500
    },
    {
      "epoch": 0.3372254773848164,
      "grad_norm": 0.9177157878875732,
      "learning_rate": 8e-05,
      "loss": 9.4884,
      "step": 8000
    },
    {
      "epoch": 0.35830206972136747,
      "grad_norm": 1.2003754377365112,
      "learning_rate": 8.5e-05,
      "loss": 9.3592,
      "step": 8500
    },
    {
      "epoch": 0.3793786620579185,
      "grad_norm": 1.1404178142547607,
      "learning_rate": 9e-05,
      "loss": 9.1702,
      "step": 9000
    },
    {
      "epoch": 0.4004552543944695,
      "grad_norm": 1.3061474561691284,
      "learning_rate": 9.5e-05,
      "loss": 9.0276,
      "step": 9500
    },
    {
      "epoch": 0.4215318467310205,
      "grad_norm": 1.116431713104248,
      "learning_rate": 0.0001,
      "loss": 8.955,
      "step": 10000
    },
    {
      "epoch": 0.4215318467310205,
      "eval_accuracy": 0.1644207676211004,
      "eval_loss": 8.740679740905762,
      "eval_runtime": 389.8479,
      "eval_samples_per_second": 309.623,
      "eval_steps_per_second": 3.227,
      "step": 10000
    },
    {
      "epoch": 0.4426084390675716,
      "grad_norm": 1.151736855506897,
      "learning_rate": 9.99142477382841e-05,
      "loss": 8.9091,
      "step": 10500
    },
    {
      "epoch": 0.4636850314041226,
      "grad_norm": 1.1560736894607544,
      "learning_rate": 9.98284954765682e-05,
      "loss": 8.8771,
      "step": 11000
    },
    {
      "epoch": 0.4847616237406736,
      "grad_norm": 1.2220590114593506,
      "learning_rate": 9.97427432148523e-05,
      "loss": 8.8587,
      "step": 11500
    },
    {
      "epoch": 0.5058382160772247,
      "grad_norm": 1.1817944049835205,
      "learning_rate": 9.96569909531364e-05,
      "loss": 8.8367,
      "step": 12000
    },
    {
      "epoch": 0.5269148084137757,
      "grad_norm": 1.1516121625900269,
      "learning_rate": 9.957123869142048e-05,
      "loss": 8.8238,
      "step": 12500
    },
    {
      "epoch": 0.5479914007503267,
      "grad_norm": 1.1616461277008057,
      "learning_rate": 9.94854864297046e-05,
      "loss": 8.8119,
      "step": 13000
    },
    {
      "epoch": 0.5690679930868777,
      "grad_norm": 1.2016310691833496,
      "learning_rate": 9.939973416798869e-05,
      "loss": 8.7984,
      "step": 13500
    },
    {
      "epoch": 0.5901445854234287,
      "grad_norm": 1.1256052255630493,
      "learning_rate": 9.931398190627278e-05,
      "loss": 8.7906,
      "step": 14000
    },
    {
      "epoch": 0.6112211777599798,
      "grad_norm": 1.1238527297973633,
      "learning_rate": 9.922822964455688e-05,
      "loss": 8.78,
      "step": 14500
    },
    {
      "epoch": 0.6322977700965308,
      "grad_norm": 0.9226935505867004,
      "learning_rate": 9.914247738284099e-05,
      "loss": 8.7724,
      "step": 15000
    },
    {
      "epoch": 0.6533743624330818,
      "grad_norm": 1.0073503255844116,
      "learning_rate": 9.905672512112507e-05,
      "loss": 8.7616,
      "step": 15500
    },
    {
      "epoch": 0.6744509547696328,
      "grad_norm": 1.0283187627792358,
      "learning_rate": 9.897097285940917e-05,
      "loss": 8.7571,
      "step": 16000
    },
    {
      "epoch": 0.6955275471061839,
      "grad_norm": 0.9642539024353027,
      "learning_rate": 9.888522059769327e-05,
      "loss": 8.7487,
      "step": 16500
    },
    {
      "epoch": 0.7166041394427349,
      "grad_norm": 0.9812933802604675,
      "learning_rate": 9.879946833597737e-05,
      "loss": 8.7435,
      "step": 17000
    },
    {
      "epoch": 0.737680731779286,
      "grad_norm": 0.9539075493812561,
      "learning_rate": 9.871371607426147e-05,
      "loss": 8.7311,
      "step": 17500
    },
    {
      "epoch": 0.758757324115837,
      "grad_norm": 0.9598174095153809,
      "learning_rate": 9.862796381254556e-05,
      "loss": 8.7355,
      "step": 18000
    },
    {
      "epoch": 0.779833916452388,
      "grad_norm": 0.9137150049209595,
      "learning_rate": 9.854221155082965e-05,
      "loss": 8.7256,
      "step": 18500
    },
    {
      "epoch": 0.800910508788939,
      "grad_norm": 0.8744050860404968,
      "learning_rate": 9.845645928911376e-05,
      "loss": 8.7258,
      "step": 19000
    },
    {
      "epoch": 0.82198710112549,
      "grad_norm": 0.8606169819831848,
      "learning_rate": 9.837070702739786e-05,
      "loss": 8.7187,
      "step": 19500
    },
    {
      "epoch": 0.843063693462041,
      "grad_norm": 0.8873212933540344,
      "learning_rate": 9.828495476568194e-05,
      "loss": 8.7137,
      "step": 20000
    },
    {
      "epoch": 0.843063693462041,
      "eval_accuracy": 0.1652503289982378,
      "eval_loss": 8.549050331115723,
      "eval_runtime": 398.205,
      "eval_samples_per_second": 303.125,
      "eval_steps_per_second": 3.159,
      "step": 20000
    },
    {
      "epoch": 0.864140285798592,
      "grad_norm": 0.8706851005554199,
      "learning_rate": 9.819920250396604e-05,
      "loss": 8.7064,
      "step": 20500
    },
    {
      "epoch": 0.8852168781351432,
      "grad_norm": 0.9130491614341736,
      "learning_rate": 9.811345024225015e-05,
      "loss": 8.6998,
      "step": 21000
    },
    {
      "epoch": 0.9062934704716942,
      "grad_norm": 0.8930593729019165,
      "learning_rate": 9.802769798053424e-05,
      "loss": 8.6986,
      "step": 21500
    },
    {
      "epoch": 0.9273700628082452,
      "grad_norm": 0.9613292813301086,
      "learning_rate": 9.794194571881834e-05,
      "loss": 8.6903,
      "step": 22000
    },
    {
      "epoch": 0.9484466551447962,
      "grad_norm": 0.8306567668914795,
      "learning_rate": 9.785619345710244e-05,
      "loss": 8.6885,
      "step": 22500
    },
    {
      "epoch": 0.9695232474813472,
      "grad_norm": 0.8821769952774048,
      "learning_rate": 9.777044119538653e-05,
      "loss": 8.6813,
      "step": 23000
    },
    {
      "epoch": 0.9905998398178982,
      "grad_norm": 0.8164405226707458,
      "learning_rate": 9.768468893367063e-05,
      "loss": 8.6746,
      "step": 23500
    },
    {
      "epoch": 1.0116764321544494,
      "grad_norm": 0.8719035387039185,
      "learning_rate": 9.759893667195473e-05,
      "loss": 8.6721,
      "step": 24000
    },
    {
      "epoch": 1.0327530244910004,
      "grad_norm": 0.8190293312072754,
      "learning_rate": 9.751318441023883e-05,
      "loss": 8.6645,
      "step": 24500
    },
    {
      "epoch": 1.0538296168275514,
      "grad_norm": 0.8539460301399231,
      "learning_rate": 9.742743214852293e-05,
      "loss": 8.6637,
      "step": 25000
    },
    {
      "epoch": 1.0749062091641024,
      "grad_norm": 0.8444121479988098,
      "learning_rate": 9.734167988680703e-05,
      "loss": 8.6537,
      "step": 25500
    },
    {
      "epoch": 1.0959828015006534,
      "grad_norm": 0.8428012728691101,
      "learning_rate": 9.725592762509111e-05,
      "loss": 8.6523,
      "step": 26000
    },
    {
      "epoch": 1.1170593938372044,
      "grad_norm": 0.8009039163589478,
      "learning_rate": 9.717017536337522e-05,
      "loss": 8.6465,
      "step": 26500
    },
    {
      "epoch": 1.1381359861737554,
      "grad_norm": 0.7723215222358704,
      "learning_rate": 9.708442310165932e-05,
      "loss": 8.6504,
      "step": 27000
    },
    {
      "epoch": 1.1592125785103065,
      "grad_norm": 0.7944744825363159,
      "learning_rate": 9.69986708399434e-05,
      "loss": 8.6431,
      "step": 27500
    },
    {
      "epoch": 1.1802891708468575,
      "grad_norm": 0.8164315223693848,
      "learning_rate": 9.69129185782275e-05,
      "loss": 8.645,
      "step": 28000
    },
    {
      "epoch": 1.2013657631834085,
      "grad_norm": 0.8102511763572693,
      "learning_rate": 9.68271663165116e-05,
      "loss": 8.6333,
      "step": 28500
    },
    {
      "epoch": 1.2224423555199595,
      "grad_norm": 0.8621619939804077,
      "learning_rate": 9.67414140547957e-05,
      "loss": 8.6301,
      "step": 29000
    },
    {
      "epoch": 1.2435189478565105,
      "grad_norm": 0.8856198787689209,
      "learning_rate": 9.66556617930798e-05,
      "loss": 8.6224,
      "step": 29500
    },
    {
      "epoch": 1.2645955401930615,
      "grad_norm": 0.7654852271080017,
      "learning_rate": 9.656990953136388e-05,
      "loss": 8.6265,
      "step": 30000
    },
    {
      "epoch": 1.2645955401930615,
      "eval_accuracy": 0.16629645683082692,
      "eval_loss": 8.478349685668945,
      "eval_runtime": 401.4225,
      "eval_samples_per_second": 300.696,
      "eval_steps_per_second": 3.134,
      "step": 30000
    },
    {
      "epoch": 1.2856721325296125,
      "grad_norm": 0.8410795331001282,
      "learning_rate": 9.6484157269648e-05,
      "loss": 8.6166,
      "step": 30500
    },
    {
      "epoch": 1.3067487248661636,
      "grad_norm": 0.788071870803833,
      "learning_rate": 9.639840500793209e-05,
      "loss": 8.622,
      "step": 31000
    },
    {
      "epoch": 1.3278253172027146,
      "grad_norm": 0.8022160530090332,
      "learning_rate": 9.631265274621618e-05,
      "loss": 8.6138,
      "step": 31500
    },
    {
      "epoch": 1.3489019095392658,
      "grad_norm": 0.7865484952926636,
      "learning_rate": 9.622690048450028e-05,
      "loss": 8.6163,
      "step": 32000
    },
    {
      "epoch": 1.3699785018758166,
      "grad_norm": 0.7818911671638489,
      "learning_rate": 9.614114822278439e-05,
      "loss": 8.6089,
      "step": 32500
    },
    {
      "epoch": 1.3910550942123678,
      "grad_norm": 0.7585428953170776,
      "learning_rate": 9.605539596106847e-05,
      "loss": 8.6108,
      "step": 33000
    },
    {
      "epoch": 1.4121316865489189,
      "grad_norm": 0.7433955073356628,
      "learning_rate": 9.596964369935257e-05,
      "loss": 8.6033,
      "step": 33500
    },
    {
      "epoch": 1.4332082788854699,
      "grad_norm": 0.8216428756713867,
      "learning_rate": 9.588389143763667e-05,
      "loss": 8.6,
      "step": 34000
    },
    {
      "epoch": 1.4542848712220209,
      "grad_norm": 0.7536180019378662,
      "learning_rate": 9.579813917592077e-05,
      "loss": 8.6011,
      "step": 34500
    },
    {
      "epoch": 1.475361463558572,
      "grad_norm": 0.7993325591087341,
      "learning_rate": 9.571238691420487e-05,
      "loss": 8.5973,
      "step": 35000
    },
    {
      "epoch": 1.496438055895123,
      "grad_norm": 0.7437774538993835,
      "learning_rate": 9.562663465248896e-05,
      "loss": 8.5931,
      "step": 35500
    },
    {
      "epoch": 1.517514648231674,
      "grad_norm": 0.7588856220245361,
      "learning_rate": 9.554088239077306e-05,
      "loss": 8.5853,
      "step": 36000
    },
    {
      "epoch": 1.538591240568225,
      "grad_norm": 0.7942984700202942,
      "learning_rate": 9.545513012905716e-05,
      "loss": 8.5874,
      "step": 36500
    },
    {
      "epoch": 1.559667832904776,
      "grad_norm": 0.7693947553634644,
      "learning_rate": 9.536937786734126e-05,
      "loss": 8.5873,
      "step": 37000
    },
    {
      "epoch": 1.580744425241327,
      "grad_norm": 0.8603740930557251,
      "learning_rate": 9.528362560562534e-05,
      "loss": 8.5566,
      "step": 37500
    },
    {
      "epoch": 1.601821017577878,
      "grad_norm": 1.08756685256958,
      "learning_rate": 9.519787334390946e-05,
      "loss": 8.0134,
      "step": 38000
    },
    {
      "epoch": 1.622897609914429,
      "grad_norm": 1.1883584260940552,
      "learning_rate": 9.511212108219355e-05,
      "loss": 7.5699,
      "step": 38500
    },
    {
      "epoch": 1.64397420225098,
      "grad_norm": 1.3286100625991821,
      "learning_rate": 9.502636882047764e-05,
      "loss": 7.3058,
      "step": 39000
    },
    {
      "epoch": 1.6650507945875312,
      "grad_norm": 1.3513630628585815,
      "learning_rate": 9.494061655876174e-05,
      "loss": 6.9949,
      "step": 39500
    },
    {
      "epoch": 1.686127386924082,
      "grad_norm": 1.385811448097229,
      "learning_rate": 9.485486429704585e-05,
      "loss": 6.7021,
      "step": 40000
    },
    {
      "epoch": 1.686127386924082,
      "eval_accuracy": 0.4554109790887892,
      "eval_loss": 6.098803520202637,
      "eval_runtime": 275.2373,
      "eval_samples_per_second": 438.552,
      "eval_steps_per_second": 4.571,
      "step": 40000
    },
    {
      "epoch": 1.7072039792606333,
      "grad_norm": 1.4035117626190186,
      "learning_rate": 9.476911203532993e-05,
      "loss": 6.4848,
      "step": 40500
    },
    {
      "epoch": 1.728280571597184,
      "grad_norm": 1.3117551803588867,
      "learning_rate": 9.468335977361403e-05,
      "loss": 6.3218,
      "step": 41000
    },
    {
      "epoch": 1.7493571639337353,
      "grad_norm": 1.516714096069336,
      "learning_rate": 9.459760751189813e-05,
      "loss": 6.188,
      "step": 41500
    },
    {
      "epoch": 1.770433756270286,
      "grad_norm": 1.2671970129013062,
      "learning_rate": 9.451185525018223e-05,
      "loss": 6.0549,
      "step": 42000
    },
    {
      "epoch": 1.7915103486068373,
      "grad_norm": 1.2253568172454834,
      "learning_rate": 9.442610298846633e-05,
      "loss": 5.9391,
      "step": 42500
    },
    {
      "epoch": 1.8125869409433881,
      "grad_norm": 1.2414438724517822,
      "learning_rate": 9.434035072675043e-05,
      "loss": 5.8476,
      "step": 43000
    },
    {
      "epoch": 1.8336635332799394,
      "grad_norm": 1.1751784086227417,
      "learning_rate": 9.425459846503451e-05,
      "loss": 5.7693,
      "step": 43500
    },
    {
      "epoch": 1.8547401256164904,
      "grad_norm": 1.1862704753875732,
      "learning_rate": 9.416884620331862e-05,
      "loss": 5.7098,
      "step": 44000
    },
    {
      "epoch": 1.8758167179530414,
      "grad_norm": 1.237450122833252,
      "learning_rate": 9.408309394160272e-05,
      "loss": 5.6637,
      "step": 44500
    },
    {
      "epoch": 1.8968933102895924,
      "grad_norm": 1.2492175102233887,
      "learning_rate": 9.39973416798868e-05,
      "loss": 5.6087,
      "step": 45000
    },
    {
      "epoch": 1.9179699026261434,
      "grad_norm": 1.1579716205596924,
      "learning_rate": 9.39115894181709e-05,
      "loss": 5.5749,
      "step": 45500
    },
    {
      "epoch": 1.9390464949626944,
      "grad_norm": 1.1826061010360718,
      "learning_rate": 9.382583715645502e-05,
      "loss": 5.5329,
      "step": 46000
    },
    {
      "epoch": 1.9601230872992454,
      "grad_norm": 1.1524370908737183,
      "learning_rate": 9.37400848947391e-05,
      "loss": 5.502,
      "step": 46500
    },
    {
      "epoch": 1.9811996796357965,
      "grad_norm": 1.1566799879074097,
      "learning_rate": 9.36543326330232e-05,
      "loss": 5.4771,
      "step": 47000
    },
    {
      "epoch": 2.0022762719723475,
      "grad_norm": 1.2325533628463745,
      "learning_rate": 9.35685803713073e-05,
      "loss": 5.453,
      "step": 47500
    },
    {
      "epoch": 2.0233528643088987,
      "grad_norm": 1.1723849773406982,
      "learning_rate": 9.34828281095914e-05,
      "loss": 5.4158,
      "step": 48000
    },
    {
      "epoch": 2.0444294566454495,
      "grad_norm": 1.179297924041748,
      "learning_rate": 9.339707584787549e-05,
      "loss": 5.3952,
      "step": 48500
    },
    {
      "epoch": 2.0655060489820007,
      "grad_norm": 1.1852002143859863,
      "learning_rate": 9.331132358615959e-05,
      "loss": 5.37,
      "step": 49000
    },
    {
      "epoch": 2.0865826413185515,
      "grad_norm": 1.1736326217651367,
      "learning_rate": 9.322557132444369e-05,
      "loss": 5.3491,
      "step": 49500
    },
    {
      "epoch": 2.1076592336551028,
      "grad_norm": 1.190647006034851,
      "learning_rate": 9.313981906272779e-05,
      "loss": 5.3349,
      "step": 50000
    },
    {
      "epoch": 2.1076592336551028,
      "eval_accuracy": 0.6050748638846798,
      "eval_loss": 5.010458469390869,
      "eval_runtime": 275.2836,
      "eval_samples_per_second": 438.479,
      "eval_steps_per_second": 4.57,
      "step": 50000
    },
    {
      "epoch": 2.1287358259916536,
      "grad_norm": 1.138453722000122,
      "learning_rate": 9.305406680101189e-05,
      "loss": 5.3189,
      "step": 50500
    },
    {
      "epoch": 2.149812418328205,
      "grad_norm": 1.1474530696868896,
      "learning_rate": 9.296831453929597e-05,
      "loss": 5.2974,
      "step": 51000
    },
    {
      "epoch": 2.1708890106647556,
      "grad_norm": 1.1708842515945435,
      "learning_rate": 9.288256227758008e-05,
      "loss": 5.2792,
      "step": 51500
    },
    {
      "epoch": 2.191965603001307,
      "grad_norm": 1.1333872079849243,
      "learning_rate": 9.279681001586418e-05,
      "loss": 5.265,
      "step": 52000
    },
    {
      "epoch": 2.2130421953378576,
      "grad_norm": 1.1572861671447754,
      "learning_rate": 9.271105775414827e-05,
      "loss": 5.2459,
      "step": 52500
    },
    {
      "epoch": 2.234118787674409,
      "grad_norm": 1.1170496940612793,
      "learning_rate": 9.262530549243236e-05,
      "loss": 5.2328,
      "step": 53000
    },
    {
      "epoch": 2.2551953800109596,
      "grad_norm": 1.1192835569381714,
      "learning_rate": 9.253955323071646e-05,
      "loss": 5.2218,
      "step": 53500
    },
    {
      "epoch": 2.276271972347511,
      "grad_norm": 1.1598303318023682,
      "learning_rate": 9.245380096900056e-05,
      "loss": 5.2123,
      "step": 54000
    },
    {
      "epoch": 2.2973485646840617,
      "grad_norm": 1.1191178560256958,
      "learning_rate": 9.236804870728466e-05,
      "loss": 5.1881,
      "step": 54500
    },
    {
      "epoch": 2.318425157020613,
      "grad_norm": 1.1477385759353638,
      "learning_rate": 9.228229644556876e-05,
      "loss": 5.1823,
      "step": 55000
    },
    {
      "epoch": 2.339501749357164,
      "grad_norm": 1.1701122522354126,
      "learning_rate": 9.219654418385286e-05,
      "loss": 5.1629,
      "step": 55500
    },
    {
      "epoch": 2.360578341693715,
      "grad_norm": 1.1012564897537231,
      "learning_rate": 9.211079192213695e-05,
      "loss": 5.1629,
      "step": 56000
    },
    {
      "epoch": 2.381654934030266,
      "grad_norm": 1.1162365674972534,
      "learning_rate": 9.202503966042105e-05,
      "loss": 5.1484,
      "step": 56500
    },
    {
      "epoch": 2.402731526366817,
      "grad_norm": 1.1296428442001343,
      "learning_rate": 9.193928739870514e-05,
      "loss": 5.1358,
      "step": 57000
    },
    {
      "epoch": 2.423808118703368,
      "grad_norm": 1.1225992441177368,
      "learning_rate": 9.185353513698925e-05,
      "loss": 5.124,
      "step": 57500
    },
    {
      "epoch": 2.444884711039919,
      "grad_norm": 1.1550874710083008,
      "learning_rate": 9.176778287527335e-05,
      "loss": 5.1168,
      "step": 58000
    },
    {
      "epoch": 2.4659613033764702,
      "grad_norm": 1.1405001878738403,
      "learning_rate": 9.168203061355743e-05,
      "loss": 5.108,
      "step": 58500
    },
    {
      "epoch": 2.487037895713021,
      "grad_norm": 1.1477960348129272,
      "learning_rate": 9.159627835184153e-05,
      "loss": 5.0962,
      "step": 59000
    },
    {
      "epoch": 2.5081144880495723,
      "grad_norm": 1.1296536922454834,
      "learning_rate": 9.151052609012564e-05,
      "loss": 5.0903,
      "step": 59500
    },
    {
      "epoch": 2.529191080386123,
      "grad_norm": 1.1792579889297485,
      "learning_rate": 9.142477382840973e-05,
      "loss": 5.0794,
      "step": 60000
    },
    {
      "epoch": 2.529191080386123,
      "eval_accuracy": 0.6389996141449213,
      "eval_loss": 4.775449275970459,
      "eval_runtime": 278.6426,
      "eval_samples_per_second": 433.193,
      "eval_steps_per_second": 4.515,
      "step": 60000
    },
    {
      "epoch": 2.5502676727226743,
      "grad_norm": 1.1807113885879517,
      "learning_rate": 9.133902156669383e-05,
      "loss": 5.0691,
      "step": 60500
    },
    {
      "epoch": 2.571344265059225,
      "grad_norm": 1.1462461948394775,
      "learning_rate": 9.125326930497792e-05,
      "loss": 5.0635,
      "step": 61000
    },
    {
      "epoch": 2.5924208573957763,
      "grad_norm": 1.1145563125610352,
      "learning_rate": 9.116751704326202e-05,
      "loss": 5.0518,
      "step": 61500
    },
    {
      "epoch": 2.613497449732327,
      "grad_norm": 1.1006208658218384,
      "learning_rate": 9.108176478154612e-05,
      "loss": 5.0401,
      "step": 62000
    },
    {
      "epoch": 2.6345740420688784,
      "grad_norm": 1.1260210275650024,
      "learning_rate": 9.09960125198302e-05,
      "loss": 5.0453,
      "step": 62500
    },
    {
      "epoch": 2.655650634405429,
      "grad_norm": 1.1253161430358887,
      "learning_rate": 9.091026025811432e-05,
      "loss": 5.0298,
      "step": 63000
    },
    {
      "epoch": 2.6767272267419804,
      "grad_norm": 1.153130054473877,
      "learning_rate": 9.082450799639842e-05,
      "loss": 5.0239,
      "step": 63500
    },
    {
      "epoch": 2.6978038190785316,
      "grad_norm": 1.1153396368026733,
      "learning_rate": 9.07387557346825e-05,
      "loss": 5.0189,
      "step": 64000
    },
    {
      "epoch": 2.7188804114150824,
      "grad_norm": 1.2268102169036865,
      "learning_rate": 9.06530034729666e-05,
      "loss": 5.01,
      "step": 64500
    },
    {
      "epoch": 2.739957003751633,
      "grad_norm": 1.1449357271194458,
      "learning_rate": 9.05672512112507e-05,
      "loss": 5.0024,
      "step": 65000
    },
    {
      "epoch": 2.7610335960881844,
      "grad_norm": 1.1376043558120728,
      "learning_rate": 9.04814989495348e-05,
      "loss": 4.9947,
      "step": 65500
    },
    {
      "epoch": 2.7821101884247357,
      "grad_norm": 1.1206307411193848,
      "learning_rate": 9.03957466878189e-05,
      "loss": 4.9933,
      "step": 66000
    },
    {
      "epoch": 2.8031867807612865,
      "grad_norm": 1.1202338933944702,
      "learning_rate": 9.030999442610299e-05,
      "loss": 4.9835,
      "step": 66500
    },
    {
      "epoch": 2.8242633730978377,
      "grad_norm": 1.0986719131469727,
      "learning_rate": 9.022424216438709e-05,
      "loss": 4.9729,
      "step": 67000
    },
    {
      "epoch": 2.8453399654343885,
      "grad_norm": 1.1213562488555908,
      "learning_rate": 9.013848990267119e-05,
      "loss": 4.9671,
      "step": 67500
    },
    {
      "epoch": 2.8664165577709397,
      "grad_norm": 1.0928248167037964,
      "learning_rate": 9.005273764095529e-05,
      "loss": 4.9612,
      "step": 68000
    },
    {
      "epoch": 2.8874931501074905,
      "grad_norm": 1.110120415687561,
      "learning_rate": 8.996698537923937e-05,
      "loss": 4.9615,
      "step": 68500
    },
    {
      "epoch": 2.9085697424440418,
      "grad_norm": 1.1316410303115845,
      "learning_rate": 8.988123311752348e-05,
      "loss": 4.951,
      "step": 69000
    },
    {
      "epoch": 2.9296463347805926,
      "grad_norm": 1.1325515508651733,
      "learning_rate": 8.979548085580758e-05,
      "loss": 4.9482,
      "step": 69500
    },
    {
      "epoch": 2.950722927117144,
      "grad_norm": 1.1129415035247803,
      "learning_rate": 8.970972859409167e-05,
      "loss": 4.9395,
      "step": 70000
    },
    {
      "epoch": 2.950722927117144,
      "eval_accuracy": 0.6570454378914437,
      "eval_loss": 4.650896072387695,
      "eval_runtime": 279.7651,
      "eval_samples_per_second": 431.455,
      "eval_steps_per_second": 4.497,
      "step": 70000
    },
    {
      "epoch": 2.9717995194536946,
      "grad_norm": 1.0997278690338135,
      "learning_rate": 8.962397633237576e-05,
      "loss": 4.9375,
      "step": 70500
    },
    {
      "epoch": 2.992876111790246,
      "grad_norm": 1.1157746315002441,
      "learning_rate": 8.953822407065988e-05,
      "loss": 4.9294,
      "step": 71000
    },
    {
      "epoch": 3.0139527041267966,
      "grad_norm": 1.1451656818389893,
      "learning_rate": 8.945247180894396e-05,
      "loss": 4.9225,
      "step": 71500
    },
    {
      "epoch": 3.035029296463348,
      "grad_norm": 1.1236497163772583,
      "learning_rate": 8.936671954722806e-05,
      "loss": 4.9161,
      "step": 72000
    },
    {
      "epoch": 3.0561058887998986,
      "grad_norm": 1.1361490488052368,
      "learning_rate": 8.928096728551216e-05,
      "loss": 4.9158,
      "step": 72500
    },
    {
      "epoch": 3.07718248113645,
      "grad_norm": 1.1723065376281738,
      "learning_rate": 8.919521502379626e-05,
      "loss": 4.9067,
      "step": 73000
    },
    {
      "epoch": 3.0982590734730007,
      "grad_norm": 1.1378494501113892,
      "learning_rate": 8.910946276208035e-05,
      "loss": 4.9086,
      "step": 73500
    },
    {
      "epoch": 3.119335665809552,
      "grad_norm": 1.1043180227279663,
      "learning_rate": 8.902371050036445e-05,
      "loss": 4.8953,
      "step": 74000
    },
    {
      "epoch": 3.140412258146103,
      "grad_norm": 1.1213690042495728,
      "learning_rate": 8.893795823864855e-05,
      "loss": 4.8972,
      "step": 74500
    },
    {
      "epoch": 3.161488850482654,
      "grad_norm": 1.0834482908248901,
      "learning_rate": 8.885220597693265e-05,
      "loss": 4.8904,
      "step": 75000
    },
    {
      "epoch": 3.182565442819205,
      "grad_norm": 1.1411031484603882,
      "learning_rate": 8.876645371521675e-05,
      "loss": 4.8842,
      "step": 75500
    },
    {
      "epoch": 3.203642035155756,
      "grad_norm": 1.076891303062439,
      "learning_rate": 8.868070145350083e-05,
      "loss": 4.8805,
      "step": 76000
    },
    {
      "epoch": 3.224718627492307,
      "grad_norm": 1.1186237335205078,
      "learning_rate": 8.859494919178493e-05,
      "loss": 4.881,
      "step": 76500
    },
    {
      "epoch": 3.245795219828858,
      "grad_norm": 1.1032005548477173,
      "learning_rate": 8.850919693006904e-05,
      "loss": 4.8707,
      "step": 77000
    },
    {
      "epoch": 3.2668718121654092,
      "grad_norm": 1.0974488258361816,
      "learning_rate": 8.842344466835313e-05,
      "loss": 4.8674,
      "step": 77500
    },
    {
      "epoch": 3.28794840450196,
      "grad_norm": 1.1435718536376953,
      "learning_rate": 8.833769240663723e-05,
      "loss": 4.8657,
      "step": 78000
    },
    {
      "epoch": 3.3090249968385113,
      "grad_norm": 1.1145579814910889,
      "learning_rate": 8.825194014492132e-05,
      "loss": 4.8623,
      "step": 78500
    },
    {
      "epoch": 3.330101589175062,
      "grad_norm": 1.146836280822754,
      "learning_rate": 8.816618788320542e-05,
      "loss": 4.8562,
      "step": 79000
    },
    {
      "epoch": 3.3511781815116133,
      "grad_norm": 1.130529522895813,
      "learning_rate": 8.808043562148952e-05,
      "loss": 4.8543,
      "step": 79500
    },
    {
      "epoch": 3.372254773848164,
      "grad_norm": 1.1255016326904297,
      "learning_rate": 8.799468335977362e-05,
      "loss": 4.849,
      "step": 80000
    },
    {
      "epoch": 3.372254773848164,
      "eval_accuracy": 0.6692387271231184,
      "eval_loss": 4.56860876083374,
      "eval_runtime": 279.6681,
      "eval_samples_per_second": 431.604,
      "eval_steps_per_second": 4.498,
      "step": 80000
    },
    {
      "epoch": 3.3933313661847153,
      "grad_norm": 1.1140973567962646,
      "learning_rate": 8.790893109805772e-05,
      "loss": 4.8465,
      "step": 80500
    },
    {
      "epoch": 3.414407958521266,
      "grad_norm": 1.1110116243362427,
      "learning_rate": 8.782317883634182e-05,
      "loss": 4.8435,
      "step": 81000
    },
    {
      "epoch": 3.4354845508578173,
      "grad_norm": 1.121904730796814,
      "learning_rate": 8.773742657462591e-05,
      "loss": 4.8433,
      "step": 81500
    },
    {
      "epoch": 3.456561143194368,
      "grad_norm": 1.1628352403640747,
      "learning_rate": 8.765167431291e-05,
      "loss": 4.8386,
      "step": 82000
    },
    {
      "epoch": 3.4776377355309194,
      "grad_norm": 1.1047242879867554,
      "learning_rate": 8.756592205119411e-05,
      "loss": 4.8336,
      "step": 82500
    },
    {
      "epoch": 3.4987143278674706,
      "grad_norm": 1.0981991291046143,
      "learning_rate": 8.748016978947821e-05,
      "loss": 4.831,
      "step": 83000
    },
    {
      "epoch": 3.5197909202040214,
      "grad_norm": 1.1057915687561035,
      "learning_rate": 8.73944175277623e-05,
      "loss": 4.8274,
      "step": 83500
    },
    {
      "epoch": 3.540867512540572,
      "grad_norm": 1.1470248699188232,
      "learning_rate": 8.730866526604639e-05,
      "loss": 4.8251,
      "step": 84000
    },
    {
      "epoch": 3.5619441048771234,
      "grad_norm": 1.1226195096969604,
      "learning_rate": 8.72229130043305e-05,
      "loss": 4.82,
      "step": 84500
    },
    {
      "epoch": 3.5830206972136747,
      "grad_norm": 1.0975100994110107,
      "learning_rate": 8.713716074261459e-05,
      "loss": 4.8182,
      "step": 85000
    },
    {
      "epoch": 3.6040972895502255,
      "grad_norm": 1.1730355024337769,
      "learning_rate": 8.705140848089869e-05,
      "loss": 4.8155,
      "step": 85500
    },
    {
      "epoch": 3.6251738818867767,
      "grad_norm": 1.1833267211914062,
      "learning_rate": 8.696565621918278e-05,
      "loss": 4.8061,
      "step": 86000
    },
    {
      "epoch": 3.6462504742233275,
      "grad_norm": 1.1283652782440186,
      "learning_rate": 8.687990395746688e-05,
      "loss": 4.8088,
      "step": 86500
    },
    {
      "epoch": 3.6673270665598787,
      "grad_norm": 1.1132993698120117,
      "learning_rate": 8.679415169575098e-05,
      "loss": 4.8043,
      "step": 87000
    },
    {
      "epoch": 3.6884036588964295,
      "grad_norm": 1.1020381450653076,
      "learning_rate": 8.670839943403508e-05,
      "loss": 4.7999,
      "step": 87500
    },
    {
      "epoch": 3.7094802512329808,
      "grad_norm": 1.1287403106689453,
      "learning_rate": 8.662264717231916e-05,
      "loss": 4.7989,
      "step": 88000
    },
    {
      "epoch": 3.7305568435695315,
      "grad_norm": 1.1254242658615112,
      "learning_rate": 8.653689491060328e-05,
      "loss": 4.7954,
      "step": 88500
    },
    {
      "epoch": 3.751633435906083,
      "grad_norm": 1.0821113586425781,
      "learning_rate": 8.645114264888737e-05,
      "loss": 4.7953,
      "step": 89000
    },
    {
      "epoch": 3.7727100282426336,
      "grad_norm": 1.1381176710128784,
      "learning_rate": 8.636539038717146e-05,
      "loss": 4.7853,
      "step": 89500
    },
    {
      "epoch": 3.793786620579185,
      "grad_norm": 1.0930678844451904,
      "learning_rate": 8.627963812545556e-05,
      "loss": 4.7838,
      "step": 90000
    },
    {
      "epoch": 3.793786620579185,
      "eval_accuracy": 0.678659641658063,
      "eval_loss": 4.505644798278809,
      "eval_runtime": 299.5019,
      "eval_samples_per_second": 403.023,
      "eval_steps_per_second": 4.2,
      "step": 90000
    },
    {
      "epoch": 3.8148632129157356,
      "grad_norm": 1.0843104124069214,
      "learning_rate": 8.619388586373967e-05,
      "loss": 4.7864,
      "step": 90500
    },
    {
      "epoch": 3.835939805252287,
      "grad_norm": 1.2175663709640503,
      "learning_rate": 8.610813360202375e-05,
      "loss": 4.788,
      "step": 91000
    },
    {
      "epoch": 3.857016397588838,
      "grad_norm": 1.1617919206619263,
      "learning_rate": 8.602238134030785e-05,
      "loss": 4.7822,
      "step": 91500
    },
    {
      "epoch": 3.878092989925389,
      "grad_norm": 1.0972070693969727,
      "learning_rate": 8.593662907859195e-05,
      "loss": 4.7707,
      "step": 92000
    },
    {
      "epoch": 3.8991695822619397,
      "grad_norm": 1.0707718133926392,
      "learning_rate": 8.585087681687605e-05,
      "loss": 4.774,
      "step": 92500
    },
    {
      "epoch": 3.920246174598491,
      "grad_norm": 2.845102071762085,
      "learning_rate": 8.576512455516015e-05,
      "loss": 4.7823,
      "step": 93000
    },
    {
      "epoch": 3.941322766935042,
      "grad_norm": 1.100677728652954,
      "learning_rate": 8.567937229344425e-05,
      "loss": 4.7693,
      "step": 93500
    },
    {
      "epoch": 3.962399359271593,
      "grad_norm": 1.089390754699707,
      "learning_rate": 8.559362003172834e-05,
      "loss": 4.7677,
      "step": 94000
    },
    {
      "epoch": 3.983475951608144,
      "grad_norm": 1.121633529663086,
      "learning_rate": 8.550786777001244e-05,
      "loss": 4.7601,
      "step": 94500
    },
    {
      "epoch": 4.004552543944695,
      "grad_norm": 1.11043119430542,
      "learning_rate": 8.542211550829654e-05,
      "loss": 4.7583,
      "step": 95000
    },
    {
      "epoch": 4.025629136281246,
      "grad_norm": 1.0853804349899292,
      "learning_rate": 8.533636324658063e-05,
      "loss": 4.746,
      "step": 95500
    },
    {
      "epoch": 4.046705728617797,
      "grad_norm": 1.0994473695755005,
      "learning_rate": 8.525061098486474e-05,
      "loss": 4.7478,
      "step": 96000
    },
    {
      "epoch": 4.067782320954348,
      "grad_norm": 1.1095446348190308,
      "learning_rate": 8.516485872314884e-05,
      "loss": 4.7401,
      "step": 96500
    },
    {
      "epoch": 4.088858913290899,
      "grad_norm": 1.1242592334747314,
      "learning_rate": 8.507910646143292e-05,
      "loss": 4.7408,
      "step": 97000
    },
    {
      "epoch": 4.10993550562745,
      "grad_norm": 1.0952211618423462,
      "learning_rate": 8.499335419971702e-05,
      "loss": 4.736,
      "step": 97500
    },
    {
      "epoch": 4.1310120979640015,
      "grad_norm": 1.1150703430175781,
      "learning_rate": 8.490760193800112e-05,
      "loss": 4.7333,
      "step": 98000
    },
    {
      "epoch": 4.152088690300552,
      "grad_norm": 1.104730248451233,
      "learning_rate": 8.482184967628522e-05,
      "loss": 4.7327,
      "step": 98500
    },
    {
      "epoch": 4.173165282637103,
      "grad_norm": 1.0930567979812622,
      "learning_rate": 8.473609741456931e-05,
      "loss": 4.7344,
      "step": 99000
    },
    {
      "epoch": 4.194241874973654,
      "grad_norm": 1.1428852081298828,
      "learning_rate": 8.46503451528534e-05,
      "loss": 4.7309,
      "step": 99500
    },
    {
      "epoch": 4.2153184673102055,
      "grad_norm": 1.1155760288238525,
      "learning_rate": 8.456459289113751e-05,
      "loss": 4.7252,
      "step": 100000
    },
    {
      "epoch": 4.2153184673102055,
      "eval_accuracy": 0.6854397407754294,
      "eval_loss": 4.458179950714111,
      "eval_runtime": 299.553,
      "eval_samples_per_second": 402.954,
      "eval_steps_per_second": 4.2,
      "step": 100000
    },
    {
      "epoch": 4.236395059646756,
      "grad_norm": 1.1301461458206177,
      "learning_rate": 8.447884062942161e-05,
      "loss": 4.7276,
      "step": 100500
    },
    {
      "epoch": 4.257471651983307,
      "grad_norm": 1.1180411577224731,
      "learning_rate": 8.43930883677057e-05,
      "loss": 4.7243,
      "step": 101000
    },
    {
      "epoch": 4.278548244319858,
      "grad_norm": 1.0940566062927246,
      "learning_rate": 8.430733610598979e-05,
      "loss": 4.7194,
      "step": 101500
    },
    {
      "epoch": 4.29962483665641,
      "grad_norm": 1.121488332748413,
      "learning_rate": 8.42215838442739e-05,
      "loss": 4.7157,
      "step": 102000
    },
    {
      "epoch": 4.32070142899296,
      "grad_norm": 1.0981614589691162,
      "learning_rate": 8.413583158255799e-05,
      "loss": 4.7151,
      "step": 102500
    },
    {
      "epoch": 4.341778021329511,
      "grad_norm": 1.1027588844299316,
      "learning_rate": 8.405007932084209e-05,
      "loss": 4.711,
      "step": 103000
    },
    {
      "epoch": 4.362854613666062,
      "grad_norm": 1.0875078439712524,
      "learning_rate": 8.396432705912619e-05,
      "loss": 4.7092,
      "step": 103500
    },
    {
      "epoch": 4.383931206002614,
      "grad_norm": 1.1062065362930298,
      "learning_rate": 8.387857479741028e-05,
      "loss": 4.7115,
      "step": 104000
    },
    {
      "epoch": 4.405007798339165,
      "grad_norm": 1.1069047451019287,
      "learning_rate": 8.379282253569438e-05,
      "loss": 4.7068,
      "step": 104500
    },
    {
      "epoch": 4.426084390675715,
      "grad_norm": 1.0898696184158325,
      "learning_rate": 8.370707027397848e-05,
      "loss": 4.7021,
      "step": 105000
    },
    {
      "epoch": 4.4471609830122665,
      "grad_norm": 1.0988343954086304,
      "learning_rate": 8.362131801226258e-05,
      "loss": 4.7005,
      "step": 105500
    },
    {
      "epoch": 4.468237575348818,
      "grad_norm": 1.0638262033462524,
      "learning_rate": 8.353556575054668e-05,
      "loss": 4.696,
      "step": 106000
    },
    {
      "epoch": 4.489314167685369,
      "grad_norm": 1.1274583339691162,
      "learning_rate": 8.344981348883077e-05,
      "loss": 4.6979,
      "step": 106500
    },
    {
      "epoch": 4.510390760021919,
      "grad_norm": 1.1261272430419922,
      "learning_rate": 8.336406122711486e-05,
      "loss": 4.6937,
      "step": 107000
    },
    {
      "epoch": 4.5314673523584705,
      "grad_norm": 1.0899850130081177,
      "learning_rate": 8.327830896539897e-05,
      "loss": 4.6945,
      "step": 107500
    },
    {
      "epoch": 4.552543944695022,
      "grad_norm": 1.131067156791687,
      "learning_rate": 8.319255670368307e-05,
      "loss": 4.6887,
      "step": 108000
    },
    {
      "epoch": 4.573620537031573,
      "grad_norm": 1.1259212493896484,
      "learning_rate": 8.310680444196715e-05,
      "loss": 4.6881,
      "step": 108500
    },
    {
      "epoch": 4.594697129368123,
      "grad_norm": 1.1104803085327148,
      "learning_rate": 8.302105218025125e-05,
      "loss": 4.6794,
      "step": 109000
    },
    {
      "epoch": 4.615773721704675,
      "grad_norm": 1.1038588285446167,
      "learning_rate": 8.293529991853536e-05,
      "loss": 4.6881,
      "step": 109500
    },
    {
      "epoch": 4.636850314041226,
      "grad_norm": 1.1372562646865845,
      "learning_rate": 8.284954765681945e-05,
      "loss": 4.6807,
      "step": 110000
    },
    {
      "epoch": 4.636850314041226,
      "eval_accuracy": 0.6913242149694058,
      "eval_loss": 4.413020610809326,
      "eval_runtime": 287.2551,
      "eval_samples_per_second": 420.205,
      "eval_steps_per_second": 4.379,
      "step": 110000
    },
    {
      "epoch": 4.657926906377777,
      "grad_norm": 1.1095455884933472,
      "learning_rate": 8.276379539510355e-05,
      "loss": 4.6793,
      "step": 110500
    },
    {
      "epoch": 4.679003498714328,
      "grad_norm": 1.1247427463531494,
      "learning_rate": 8.267804313338765e-05,
      "loss": 4.6794,
      "step": 111000
    },
    {
      "epoch": 4.700080091050879,
      "grad_norm": 1.1243239641189575,
      "learning_rate": 8.259229087167174e-05,
      "loss": 4.6744,
      "step": 111500
    },
    {
      "epoch": 4.72115668338743,
      "grad_norm": 1.13285493850708,
      "learning_rate": 8.250653860995584e-05,
      "loss": 4.6735,
      "step": 112000
    },
    {
      "epoch": 4.742233275723981,
      "grad_norm": 1.1231188774108887,
      "learning_rate": 8.242078634823994e-05,
      "loss": 4.6708,
      "step": 112500
    },
    {
      "epoch": 4.763309868060532,
      "grad_norm": 1.1044697761535645,
      "learning_rate": 8.233503408652403e-05,
      "loss": 4.6716,
      "step": 113000
    },
    {
      "epoch": 4.784386460397083,
      "grad_norm": 1.0879117250442505,
      "learning_rate": 8.224928182480814e-05,
      "loss": 4.6728,
      "step": 113500
    },
    {
      "epoch": 4.805463052733634,
      "grad_norm": 1.0777926445007324,
      "learning_rate": 8.216352956309224e-05,
      "loss": 4.6607,
      "step": 114000
    },
    {
      "epoch": 4.826539645070185,
      "grad_norm": 1.110925316810608,
      "learning_rate": 8.207777730137632e-05,
      "loss": 4.6615,
      "step": 114500
    },
    {
      "epoch": 4.847616237406736,
      "grad_norm": 1.0801339149475098,
      "learning_rate": 8.199202503966042e-05,
      "loss": 4.6584,
      "step": 115000
    },
    {
      "epoch": 4.868692829743287,
      "grad_norm": 1.1161648035049438,
      "learning_rate": 8.190627277794453e-05,
      "loss": 4.6604,
      "step": 115500
    },
    {
      "epoch": 4.889769422079838,
      "grad_norm": 1.1243332624435425,
      "learning_rate": 8.182052051622862e-05,
      "loss": 4.6559,
      "step": 116000
    },
    {
      "epoch": 4.910846014416389,
      "grad_norm": 1.1234240531921387,
      "learning_rate": 8.173476825451271e-05,
      "loss": 4.655,
      "step": 116500
    },
    {
      "epoch": 4.9319226067529405,
      "grad_norm": 1.1085236072540283,
      "learning_rate": 8.164901599279681e-05,
      "loss": 4.6553,
      "step": 117000
    },
    {
      "epoch": 4.952999199089491,
      "grad_norm": 1.1116122007369995,
      "learning_rate": 8.156326373108091e-05,
      "loss": 4.6557,
      "step": 117500
    },
    {
      "epoch": 4.974075791426042,
      "grad_norm": 1.0994398593902588,
      "learning_rate": 8.147751146936501e-05,
      "loss": 4.6521,
      "step": 118000
    },
    {
      "epoch": 4.995152383762593,
      "grad_norm": 1.0984135866165161,
      "learning_rate": 8.139175920764911e-05,
      "loss": 4.6453,
      "step": 118500
    },
    {
      "epoch": 5.0162289760991445,
      "grad_norm": 1.1176012754440308,
      "learning_rate": 8.13060069459332e-05,
      "loss": 4.6458,
      "step": 119000
    },
    {
      "epoch": 5.037305568435695,
      "grad_norm": 1.1008782386779785,
      "learning_rate": 8.12202546842173e-05,
      "loss": 4.6403,
      "step": 119500
    },
    {
      "epoch": 5.058382160772246,
      "grad_norm": 1.1210675239562988,
      "learning_rate": 8.11345024225014e-05,
      "loss": 4.6389,
      "step": 120000
    },
    {
      "epoch": 5.058382160772246,
      "eval_accuracy": 0.696627740007839,
      "eval_loss": 4.373435020446777,
      "eval_runtime": 284.7745,
      "eval_samples_per_second": 423.865,
      "eval_steps_per_second": 4.418,
      "step": 120000
    },
    {
      "epoch": 5.079458753108797,
      "grad_norm": 1.1253334283828735,
      "learning_rate": 8.104875016078549e-05,
      "loss": 4.6332,
      "step": 120500
    },
    {
      "epoch": 5.100535345445349,
      "grad_norm": 1.1218702793121338,
      "learning_rate": 8.09629978990696e-05,
      "loss": 4.6324,
      "step": 121000
    },
    {
      "epoch": 5.1216119377819,
      "grad_norm": 1.1146291494369507,
      "learning_rate": 8.08772456373537e-05,
      "loss": 4.6344,
      "step": 121500
    },
    {
      "epoch": 5.14268853011845,
      "grad_norm": 1.1463702917099,
      "learning_rate": 8.079149337563778e-05,
      "loss": 4.6326,
      "step": 122000
    },
    {
      "epoch": 5.163765122455001,
      "grad_norm": 1.1017736196517944,
      "learning_rate": 8.070574111392188e-05,
      "loss": 4.6286,
      "step": 122500
    },
    {
      "epoch": 5.184841714791553,
      "grad_norm": 1.1015379428863525,
      "learning_rate": 8.061998885220599e-05,
      "loss": 4.6278,
      "step": 123000
    },
    {
      "epoch": 5.205918307128104,
      "grad_norm": 1.1191444396972656,
      "learning_rate": 8.053423659049008e-05,
      "loss": 4.6284,
      "step": 123500
    },
    {
      "epoch": 5.226994899464654,
      "grad_norm": 1.1022967100143433,
      "learning_rate": 8.044848432877418e-05,
      "loss": 4.6286,
      "step": 124000
    },
    {
      "epoch": 5.2480714918012055,
      "grad_norm": 1.0950236320495605,
      "learning_rate": 8.036273206705827e-05,
      "loss": 4.6296,
      "step": 124500
    },
    {
      "epoch": 5.269148084137757,
      "grad_norm": 1.1511650085449219,
      "learning_rate": 8.027697980534237e-05,
      "loss": 4.6225,
      "step": 125000
    },
    {
      "epoch": 5.290224676474308,
      "grad_norm": 1.1365327835083008,
      "learning_rate": 8.019122754362647e-05,
      "loss": 4.6231,
      "step": 125500
    },
    {
      "epoch": 5.311301268810858,
      "grad_norm": 1.109666347503662,
      "learning_rate": 8.010547528191057e-05,
      "loss": 4.6229,
      "step": 126000
    },
    {
      "epoch": 5.3323778611474095,
      "grad_norm": 1.1099971532821655,
      "learning_rate": 8.001972302019465e-05,
      "loss": 4.6174,
      "step": 126500
    },
    {
      "epoch": 5.353454453483961,
      "grad_norm": 1.1028119325637817,
      "learning_rate": 7.993397075847876e-05,
      "loss": 4.6161,
      "step": 127000
    },
    {
      "epoch": 5.374531045820512,
      "grad_norm": 1.0946153402328491,
      "learning_rate": 7.984821849676286e-05,
      "loss": 4.6187,
      "step": 127500
    },
    {
      "epoch": 5.395607638157063,
      "grad_norm": 1.1361911296844482,
      "learning_rate": 7.976246623504695e-05,
      "loss": 4.6168,
      "step": 128000
    },
    {
      "epoch": 5.416684230493614,
      "grad_norm": 1.1638838052749634,
      "learning_rate": 7.967671397333105e-05,
      "loss": 4.611,
      "step": 128500
    },
    {
      "epoch": 5.437760822830165,
      "grad_norm": 1.1224141120910645,
      "learning_rate": 7.959096171161516e-05,
      "loss": 4.6104,
      "step": 129000
    },
    {
      "epoch": 5.458837415166716,
      "grad_norm": 1.0925666093826294,
      "learning_rate": 7.950520944989924e-05,
      "loss": 4.6073,
      "step": 129500
    },
    {
      "epoch": 5.479914007503266,
      "grad_norm": 1.1101325750350952,
      "learning_rate": 7.941945718818334e-05,
      "loss": 4.6076,
      "step": 130000
    },
    {
      "epoch": 5.479914007503266,
      "eval_accuracy": 0.7008235392680915,
      "eval_loss": 4.344308376312256,
      "eval_runtime": 293.9107,
      "eval_samples_per_second": 410.689,
      "eval_steps_per_second": 4.28,
      "step": 130000
    },
    {
      "epoch": 5.500990599839818,
      "grad_norm": 1.1048481464385986,
      "learning_rate": 7.933370492646744e-05,
      "loss": 4.6123,
      "step": 130500
    },
    {
      "epoch": 5.522067192176369,
      "grad_norm": 1.1081664562225342,
      "learning_rate": 7.924795266475154e-05,
      "loss": 4.6031,
      "step": 131000
    },
    {
      "epoch": 5.54314378451292,
      "grad_norm": 1.1265970468521118,
      "learning_rate": 7.916220040303564e-05,
      "loss": 4.6003,
      "step": 131500
    },
    {
      "epoch": 5.564220376849471,
      "grad_norm": 1.113221526145935,
      "learning_rate": 7.907644814131972e-05,
      "loss": 4.6071,
      "step": 132000
    },
    {
      "epoch": 5.585296969186022,
      "grad_norm": 1.0987931489944458,
      "learning_rate": 7.899069587960383e-05,
      "loss": 4.5975,
      "step": 132500
    },
    {
      "epoch": 5.606373561522573,
      "grad_norm": 1.0793430805206299,
      "learning_rate": 7.890494361788793e-05,
      "loss": 4.5992,
      "step": 133000
    },
    {
      "epoch": 5.627450153859124,
      "grad_norm": 1.1641173362731934,
      "learning_rate": 7.881919135617202e-05,
      "loss": 4.5969,
      "step": 133500
    },
    {
      "epoch": 5.648526746195675,
      "grad_norm": 1.133632779121399,
      "learning_rate": 7.873343909445611e-05,
      "loss": 4.5912,
      "step": 134000
    },
    {
      "epoch": 5.669603338532226,
      "grad_norm": 1.0917397737503052,
      "learning_rate": 7.864768683274023e-05,
      "loss": 4.5995,
      "step": 134500
    },
    {
      "epoch": 5.690679930868777,
      "grad_norm": 1.0795658826828003,
      "learning_rate": 7.856193457102431e-05,
      "loss": 4.5972,
      "step": 135000
    },
    {
      "epoch": 5.711756523205328,
      "grad_norm": 1.110172986984253,
      "learning_rate": 7.847618230930841e-05,
      "loss": 4.5959,
      "step": 135500
    },
    {
      "epoch": 5.7328331155418795,
      "grad_norm": 1.1031721830368042,
      "learning_rate": 7.839043004759251e-05,
      "loss": 4.5952,
      "step": 136000
    },
    {
      "epoch": 5.75390970787843,
      "grad_norm": 1.1092205047607422,
      "learning_rate": 7.83046777858766e-05,
      "loss": 4.5912,
      "step": 136500
    },
    {
      "epoch": 5.774986300214981,
      "grad_norm": 1.1314862966537476,
      "learning_rate": 7.82189255241607e-05,
      "loss": 4.591,
      "step": 137000
    },
    {
      "epoch": 5.796062892551532,
      "grad_norm": 1.149390459060669,
      "learning_rate": 7.81331732624448e-05,
      "loss": 4.5883,
      "step": 137500
    },
    {
      "epoch": 5.8171394848880835,
      "grad_norm": 1.1047414541244507,
      "learning_rate": 7.804742100072889e-05,
      "loss": 4.5867,
      "step": 138000
    },
    {
      "epoch": 5.838216077224635,
      "grad_norm": 1.1157714128494263,
      "learning_rate": 7.7961668739013e-05,
      "loss": 4.5891,
      "step": 138500
    },
    {
      "epoch": 5.859292669561185,
      "grad_norm": 1.0917019844055176,
      "learning_rate": 7.78759164772971e-05,
      "loss": 4.5827,
      "step": 139000
    },
    {
      "epoch": 5.880369261897736,
      "grad_norm": 1.115652084350586,
      "learning_rate": 7.779016421558118e-05,
      "loss": 4.5836,
      "step": 139500
    },
    {
      "epoch": 5.901445854234288,
      "grad_norm": 1.1525861024856567,
      "learning_rate": 7.770441195386528e-05,
      "loss": 4.5871,
      "step": 140000
    },
    {
      "epoch": 5.901445854234288,
      "eval_accuracy": 0.7045733673500344,
      "eval_loss": 4.317647457122803,
      "eval_runtime": 291.6997,
      "eval_samples_per_second": 413.802,
      "eval_steps_per_second": 4.313,
      "step": 140000
    },
    {
      "epoch": 5.922522446570838,
      "grad_norm": 1.105224609375,
      "learning_rate": 7.761865969214939e-05,
      "loss": 4.5788,
      "step": 140500
    },
    {
      "epoch": 5.943599038907389,
      "grad_norm": 1.0743074417114258,
      "learning_rate": 7.753290743043348e-05,
      "loss": 4.5812,
      "step": 141000
    },
    {
      "epoch": 5.96467563124394,
      "grad_norm": 1.1093602180480957,
      "learning_rate": 7.744715516871758e-05,
      "loss": 4.5805,
      "step": 141500
    },
    {
      "epoch": 5.985752223580492,
      "grad_norm": 1.1304460763931274,
      "learning_rate": 7.736140290700167e-05,
      "loss": 4.5757,
      "step": 142000
    },
    {
      "epoch": 6.006828815917043,
      "grad_norm": 5.820346355438232,
      "learning_rate": 7.727565064528577e-05,
      "loss": 4.6073,
      "step": 142500
    },
    {
      "epoch": 6.027905408253593,
      "grad_norm": 1.1930015087127686,
      "learning_rate": 7.718989838356987e-05,
      "loss": 4.5884,
      "step": 143000
    },
    {
      "epoch": 6.0489820005901445,
      "grad_norm": 1.2455768585205078,
      "learning_rate": 7.710414612185397e-05,
      "loss": 4.5853,
      "step": 143500
    },
    {
      "epoch": 6.070058592926696,
      "grad_norm": 1.1485270261764526,
      "learning_rate": 7.701839386013807e-05,
      "loss": 4.5805,
      "step": 144000
    },
    {
      "epoch": 6.091135185263247,
      "grad_norm": 35.46892547607422,
      "learning_rate": 7.693264159842217e-05,
      "loss": 4.5805,
      "step": 144500
    },
    {
      "epoch": 6.112211777599797,
      "grad_norm": 1.1956305503845215,
      "learning_rate": 7.684688933670626e-05,
      "loss": 4.5757,
      "step": 145000
    },
    {
      "epoch": 6.1332883699363485,
      "grad_norm": 1.2007899284362793,
      "learning_rate": 7.676113707499035e-05,
      "loss": 4.5774,
      "step": 145500
    },
    {
      "epoch": 6.1543649622729,
      "grad_norm": 1.3421683311462402,
      "learning_rate": 7.667538481327446e-05,
      "loss": 4.5773,
      "step": 146000
    },
    {
      "epoch": 6.175441554609451,
      "grad_norm": 1.1749438047409058,
      "learning_rate": 7.658963255155856e-05,
      "loss": 4.5793,
      "step": 146500
    },
    {
      "epoch": 6.196518146946001,
      "grad_norm": 1.1912133693695068,
      "learning_rate": 7.650388028984264e-05,
      "loss": 4.5748,
      "step": 147000
    },
    {
      "epoch": 6.217594739282553,
      "grad_norm": 1.116612195968628,
      "learning_rate": 7.641812802812674e-05,
      "loss": 4.5716,
      "step": 147500
    },
    {
      "epoch": 6.238671331619104,
      "grad_norm": 1.1055740118026733,
      "learning_rate": 7.633237576641084e-05,
      "loss": 4.5691,
      "step": 148000
    },
    {
      "epoch": 6.259747923955655,
      "grad_norm": 1.3414555788040161,
      "learning_rate": 7.624662350469494e-05,
      "loss": 4.5733,
      "step": 148500
    },
    {
      "epoch": 6.280824516292206,
      "grad_norm": 1.2480053901672363,
      "learning_rate": 7.616087124297904e-05,
      "loss": 4.5706,
      "step": 149000
    },
    {
      "epoch": 6.301901108628757,
      "grad_norm": 1.1070712804794312,
      "learning_rate": 7.607511898126313e-05,
      "loss": 4.5679,
      "step": 149500
    },
    {
      "epoch": 6.322977700965308,
      "grad_norm": 1.3548486232757568,
      "learning_rate": 7.598936671954723e-05,
      "loss": 4.5675,
      "step": 150000
    },
    {
      "epoch": 6.322977700965308,
      "eval_accuracy": 0.7069817570503899,
      "eval_loss": 4.308512210845947,
      "eval_runtime": 292.6228,
      "eval_samples_per_second": 412.497,
      "eval_steps_per_second": 4.299,
      "step": 150000
    },
    {
      "epoch": 6.344054293301859,
      "grad_norm": 1.2117482423782349,
      "learning_rate": 7.590361445783133e-05,
      "loss": 4.5684,
      "step": 150500
    },
    {
      "epoch": 6.36513088563841,
      "grad_norm": 1.1005157232284546,
      "learning_rate": 7.581786219611543e-05,
      "loss": 4.5653,
      "step": 151000
    },
    {
      "epoch": 6.386207477974961,
      "grad_norm": 3.3034961223602295,
      "learning_rate": 7.573210993439951e-05,
      "loss": 4.5646,
      "step": 151500
    },
    {
      "epoch": 6.407284070311512,
      "grad_norm": 1.1253522634506226,
      "learning_rate": 7.564635767268363e-05,
      "loss": 4.5625,
      "step": 152000
    },
    {
      "epoch": 6.428360662648063,
      "grad_norm": 1.228980302810669,
      "learning_rate": 7.556060541096772e-05,
      "loss": 4.5602,
      "step": 152500
    },
    {
      "epoch": 6.449437254984614,
      "grad_norm": 1.1707117557525635,
      "learning_rate": 7.547485314925181e-05,
      "loss": 4.5633,
      "step": 153000
    },
    {
      "epoch": 6.470513847321165,
      "grad_norm": 1.1680549383163452,
      "learning_rate": 7.538910088753591e-05,
      "loss": 4.5608,
      "step": 153500
    },
    {
      "epoch": 6.491590439657716,
      "grad_norm": 1.1928716897964478,
      "learning_rate": 7.530334862582002e-05,
      "loss": 4.5596,
      "step": 154000
    },
    {
      "epoch": 6.512667031994267,
      "grad_norm": 2.0265512466430664,
      "learning_rate": 7.52175963641041e-05,
      "loss": 4.5623,
      "step": 154500
    },
    {
      "epoch": 6.5337436243308185,
      "grad_norm": 1.2095561027526855,
      "learning_rate": 7.51318441023882e-05,
      "loss": 4.5632,
      "step": 155000
    },
    {
      "epoch": 6.55482021666737,
      "grad_norm": 2.1763975620269775,
      "learning_rate": 7.50460918406723e-05,
      "loss": 4.5621,
      "step": 155500
    },
    {
      "epoch": 6.57589680900392,
      "grad_norm": 1.1184228658676147,
      "learning_rate": 7.49603395789564e-05,
      "loss": 4.5566,
      "step": 156000
    },
    {
      "epoch": 6.596973401340471,
      "grad_norm": 1.1163439750671387,
      "learning_rate": 7.48745873172405e-05,
      "loss": 4.5513,
      "step": 156500
    },
    {
      "epoch": 6.6180499936770225,
      "grad_norm": 1.3480476140975952,
      "learning_rate": 7.47888350555246e-05,
      "loss": 4.5504,
      "step": 157000
    },
    {
      "epoch": 6.639126586013573,
      "grad_norm": 1.1463696956634521,
      "learning_rate": 7.47030827938087e-05,
      "loss": 4.553,
      "step": 157500
    },
    {
      "epoch": 6.660203178350124,
      "grad_norm": 1.2276846170425415,
      "learning_rate": 7.461733053209279e-05,
      "loss": 4.5528,
      "step": 158000
    },
    {
      "epoch": 6.681279770686675,
      "grad_norm": 1.137640118598938,
      "learning_rate": 7.453157827037689e-05,
      "loss": 4.5546,
      "step": 158500
    },
    {
      "epoch": 6.702356363023227,
      "grad_norm": 1.3521560430526733,
      "learning_rate": 7.444582600866098e-05,
      "loss": 4.5488,
      "step": 159000
    },
    {
      "epoch": 6.723432955359778,
      "grad_norm": 1.176430583000183,
      "learning_rate": 7.436007374694507e-05,
      "loss": 4.552,
      "step": 159500
    },
    {
      "epoch": 6.744509547696328,
      "grad_norm": 1.1381274461746216,
      "learning_rate": 7.427432148522919e-05,
      "loss": 4.5509,
      "step": 160000
    },
    {
      "epoch": 6.744509547696328,
      "eval_accuracy": 0.7097232830450314,
      "eval_loss": 4.286092281341553,
      "eval_runtime": 283.0842,
      "eval_samples_per_second": 426.396,
      "eval_steps_per_second": 4.444,
      "step": 160000
    },
    {
      "epoch": 6.765586140032879,
      "grad_norm": 1.1777690649032593,
      "learning_rate": 7.418856922351327e-05,
      "loss": 4.5477,
      "step": 160500
    },
    {
      "epoch": 6.786662732369431,
      "grad_norm": 1.3592698574066162,
      "learning_rate": 7.410281696179737e-05,
      "loss": 4.5442,
      "step": 161000
    },
    {
      "epoch": 6.807739324705982,
      "grad_norm": 1.1636422872543335,
      "learning_rate": 7.401706470008147e-05,
      "loss": 4.5409,
      "step": 161500
    },
    {
      "epoch": 6.828815917042532,
      "grad_norm": 1.1112773418426514,
      "learning_rate": 7.393131243836557e-05,
      "loss": 4.542,
      "step": 162000
    },
    {
      "epoch": 6.8498925093790834,
      "grad_norm": 1.8861095905303955,
      "learning_rate": 7.384556017664966e-05,
      "loss": 4.5393,
      "step": 162500
    },
    {
      "epoch": 6.870969101715635,
      "grad_norm": 1.1067795753479004,
      "learning_rate": 7.375980791493376e-05,
      "loss": 4.5375,
      "step": 163000
    },
    {
      "epoch": 6.892045694052186,
      "grad_norm": 1.1492574214935303,
      "learning_rate": 7.367405565321786e-05,
      "loss": 4.5345,
      "step": 163500
    },
    {
      "epoch": 6.913122286388736,
      "grad_norm": 1.1314256191253662,
      "learning_rate": 7.358830339150196e-05,
      "loss": 4.5364,
      "step": 164000
    },
    {
      "epoch": 6.9341988787252875,
      "grad_norm": 1.523006558418274,
      "learning_rate": 7.350255112978606e-05,
      "loss": 4.5397,
      "step": 164500
    },
    {
      "epoch": 6.955275471061839,
      "grad_norm": 1.1450444459915161,
      "learning_rate": 7.341679886807014e-05,
      "loss": 4.5368,
      "step": 165000
    },
    {
      "epoch": 6.97635206339839,
      "grad_norm": 1.1367568969726562,
      "learning_rate": 7.333104660635425e-05,
      "loss": 4.5381,
      "step": 165500
    },
    {
      "epoch": 6.997428655734941,
      "grad_norm": 1.087058663368225,
      "learning_rate": 7.324529434463835e-05,
      "loss": 4.5341,
      "step": 166000
    },
    {
      "epoch": 7.018505248071492,
      "grad_norm": 1.1436907052993774,
      "learning_rate": 7.315954208292244e-05,
      "loss": 4.5313,
      "step": 166500
    },
    {
      "epoch": 7.039581840408043,
      "grad_norm": 1.133208155632019,
      "learning_rate": 7.307378982120653e-05,
      "loss": 4.5249,
      "step": 167000
    },
    {
      "epoch": 7.060658432744594,
      "grad_norm": 1.283957839012146,
      "learning_rate": 7.298803755949063e-05,
      "loss": 4.5249,
      "step": 167500
    },
    {
      "epoch": 7.081735025081145,
      "grad_norm": 3.748706102371216,
      "learning_rate": 7.290228529777473e-05,
      "loss": 4.5267,
      "step": 168000
    },
    {
      "epoch": 7.102811617417696,
      "grad_norm": 1.2492198944091797,
      "learning_rate": 7.281653303605883e-05,
      "loss": 4.5285,
      "step": 168500
    },
    {
      "epoch": 7.123888209754247,
      "grad_norm": 1.2630292177200317,
      "learning_rate": 7.273078077434293e-05,
      "loss": 4.5317,
      "step": 169000
    },
    {
      "epoch": 7.144964802090798,
      "grad_norm": 1.1108098030090332,
      "learning_rate": 7.264502851262703e-05,
      "loss": 4.5255,
      "step": 169500
    },
    {
      "epoch": 7.166041394427349,
      "grad_norm": 1.1274200677871704,
      "learning_rate": 7.255927625091112e-05,
      "loss": 4.5201,
      "step": 170000
    },
    {
      "epoch": 7.166041394427349,
      "eval_accuracy": 0.7130893971544746,
      "eval_loss": 4.264368534088135,
      "eval_runtime": 279.3261,
      "eval_samples_per_second": 432.133,
      "eval_steps_per_second": 4.504,
      "step": 170000
    },
    {
      "epoch": 7.1871179867639,
      "grad_norm": 1.1209239959716797,
      "learning_rate": 7.247352398919521e-05,
      "loss": 4.5251,
      "step": 170500
    },
    {
      "epoch": 7.208194579100451,
      "grad_norm": 1.1141046285629272,
      "learning_rate": 7.238777172747931e-05,
      "loss": 4.5218,
      "step": 171000
    },
    {
      "epoch": 7.229271171437002,
      "grad_norm": 1.1222493648529053,
      "learning_rate": 7.230201946576342e-05,
      "loss": 4.5211,
      "step": 171500
    },
    {
      "epoch": 7.250347763773553,
      "grad_norm": 1.133284568786621,
      "learning_rate": 7.22162672040475e-05,
      "loss": 4.5201,
      "step": 172000
    },
    {
      "epoch": 7.271424356110104,
      "grad_norm": 1.1128318309783936,
      "learning_rate": 7.21305149423316e-05,
      "loss": 4.5187,
      "step": 172500
    },
    {
      "epoch": 7.292500948446655,
      "grad_norm": 1.116495966911316,
      "learning_rate": 7.20447626806157e-05,
      "loss": 4.511,
      "step": 173000
    },
    {
      "epoch": 7.313577540783206,
      "grad_norm": 1.1344844102859497,
      "learning_rate": 7.19590104188998e-05,
      "loss": 4.5121,
      "step": 173500
    },
    {
      "epoch": 7.3346541331197574,
      "grad_norm": 1.1052567958831787,
      "learning_rate": 7.18732581571839e-05,
      "loss": 4.5161,
      "step": 174000
    },
    {
      "epoch": 7.355730725456308,
      "grad_norm": 1.1152877807617188,
      "learning_rate": 7.1787505895468e-05,
      "loss": 4.5112,
      "step": 174500
    },
    {
      "epoch": 7.376807317792859,
      "grad_norm": 1.1226544380187988,
      "learning_rate": 7.17017536337521e-05,
      "loss": 4.5124,
      "step": 175000
    },
    {
      "epoch": 7.39788391012941,
      "grad_norm": 1.0775394439697266,
      "learning_rate": 7.161600137203619e-05,
      "loss": 4.5129,
      "step": 175500
    },
    {
      "epoch": 7.4189605024659615,
      "grad_norm": 1.134056568145752,
      "learning_rate": 7.153024911032029e-05,
      "loss": 4.5069,
      "step": 176000
    },
    {
      "epoch": 7.440037094802513,
      "grad_norm": 1.1618086099624634,
      "learning_rate": 7.144449684860438e-05,
      "loss": 4.5102,
      "step": 176500
    },
    {
      "epoch": 7.461113687139063,
      "grad_norm": 1.4960647821426392,
      "learning_rate": 7.135874458688849e-05,
      "loss": 4.5129,
      "step": 177000
    },
    {
      "epoch": 7.482190279475614,
      "grad_norm": 1.161183476448059,
      "learning_rate": 7.127299232517259e-05,
      "loss": 4.5109,
      "step": 177500
    },
    {
      "epoch": 7.503266871812166,
      "grad_norm": 1.131587028503418,
      "learning_rate": 7.118724006345667e-05,
      "loss": 4.5084,
      "step": 178000
    },
    {
      "epoch": 7.524343464148717,
      "grad_norm": 1.1481846570968628,
      "learning_rate": 7.110148780174077e-05,
      "loss": 4.5039,
      "step": 178500
    },
    {
      "epoch": 7.545420056485267,
      "grad_norm": 1.1456024646759033,
      "learning_rate": 7.101573554002488e-05,
      "loss": 4.5023,
      "step": 179000
    },
    {
      "epoch": 7.566496648821818,
      "grad_norm": 1.2499806880950928,
      "learning_rate": 7.092998327830897e-05,
      "loss": 4.5069,
      "step": 179500
    },
    {
      "epoch": 7.58757324115837,
      "grad_norm": 1.1300605535507202,
      "learning_rate": 7.084423101659306e-05,
      "loss": 4.5046,
      "step": 180000
    },
    {
      "epoch": 7.58757324115837,
      "eval_accuracy": 0.7151133975685496,
      "eval_loss": 4.252105236053467,
      "eval_runtime": 277.6222,
      "eval_samples_per_second": 434.785,
      "eval_steps_per_second": 4.531,
      "step": 180000
    },
    {
      "epoch": 7.608649833494921,
      "grad_norm": 1.1569077968597412,
      "learning_rate": 7.075847875487716e-05,
      "loss": 4.503,
      "step": 180500
    },
    {
      "epoch": 7.629726425831471,
      "grad_norm": 1.1098415851593018,
      "learning_rate": 7.067272649316126e-05,
      "loss": 4.5035,
      "step": 181000
    },
    {
      "epoch": 7.650803018168022,
      "grad_norm": 1.1164501905441284,
      "learning_rate": 7.058697423144536e-05,
      "loss": 4.4991,
      "step": 181500
    },
    {
      "epoch": 7.671879610504574,
      "grad_norm": 1.093335509300232,
      "learning_rate": 7.050122196972946e-05,
      "loss": 4.4987,
      "step": 182000
    },
    {
      "epoch": 7.692956202841125,
      "grad_norm": 1.1368944644927979,
      "learning_rate": 7.041546970801354e-05,
      "loss": 4.4993,
      "step": 182500
    },
    {
      "epoch": 7.714032795177676,
      "grad_norm": 1.1430537700653076,
      "learning_rate": 7.032971744629765e-05,
      "loss": 4.4987,
      "step": 183000
    },
    {
      "epoch": 7.7351093875142265,
      "grad_norm": 1.1317275762557983,
      "learning_rate": 7.024396518458175e-05,
      "loss": 4.5011,
      "step": 183500
    },
    {
      "epoch": 7.756185979850778,
      "grad_norm": 1.1010555028915405,
      "learning_rate": 7.015821292286584e-05,
      "loss": 4.4976,
      "step": 184000
    },
    {
      "epoch": 7.777262572187329,
      "grad_norm": 1.138485312461853,
      "learning_rate": 7.007246066114993e-05,
      "loss": 4.4972,
      "step": 184500
    },
    {
      "epoch": 7.798339164523879,
      "grad_norm": 1.1439129114151,
      "learning_rate": 6.998670839943405e-05,
      "loss": 4.4962,
      "step": 185000
    },
    {
      "epoch": 7.8194157568604306,
      "grad_norm": 1.1394860744476318,
      "learning_rate": 6.990095613771813e-05,
      "loss": 4.4934,
      "step": 185500
    },
    {
      "epoch": 7.840492349196982,
      "grad_norm": 1.151188611984253,
      "learning_rate": 6.981520387600223e-05,
      "loss": 4.4941,
      "step": 186000
    },
    {
      "epoch": 7.861568941533533,
      "grad_norm": 1.1130443811416626,
      "learning_rate": 6.972945161428633e-05,
      "loss": 4.4954,
      "step": 186500
    },
    {
      "epoch": 7.882645533870084,
      "grad_norm": 1.1395665407180786,
      "learning_rate": 6.964369935257043e-05,
      "loss": 4.494,
      "step": 187000
    },
    {
      "epoch": 7.903722126206635,
      "grad_norm": 1.1009217500686646,
      "learning_rate": 6.955794709085452e-05,
      "loss": 4.4939,
      "step": 187500
    },
    {
      "epoch": 7.924798718543186,
      "grad_norm": 1.132899522781372,
      "learning_rate": 6.947219482913862e-05,
      "loss": 4.4821,
      "step": 188000
    },
    {
      "epoch": 7.945875310879737,
      "grad_norm": 1.116179347038269,
      "learning_rate": 6.938644256742272e-05,
      "loss": 4.4946,
      "step": 188500
    },
    {
      "epoch": 7.966951903216288,
      "grad_norm": 1.143011212348938,
      "learning_rate": 6.930069030570682e-05,
      "loss": 4.4895,
      "step": 189000
    },
    {
      "epoch": 7.988028495552839,
      "grad_norm": 1.1341722011566162,
      "learning_rate": 6.921493804399092e-05,
      "loss": 4.4857,
      "step": 189500
    },
    {
      "epoch": 8.00910508788939,
      "grad_norm": 1.1401695013046265,
      "learning_rate": 6.9129185782275e-05,
      "loss": 4.4859,
      "step": 190000
    },
    {
      "epoch": 8.00910508788939,
      "eval_accuracy": 0.7175486553143412,
      "eval_loss": 4.23392391204834,
      "eval_runtime": 279.3543,
      "eval_samples_per_second": 432.089,
      "eval_steps_per_second": 4.503,
      "step": 190000
    },
    {
      "epoch": 8.03018168022594,
      "grad_norm": 1.1347689628601074,
      "learning_rate": 6.904343352055911e-05,
      "loss": 4.4804,
      "step": 190500
    },
    {
      "epoch": 8.051258272562492,
      "grad_norm": 1.1102697849273682,
      "learning_rate": 6.895768125884321e-05,
      "loss": 4.4831,
      "step": 191000
    },
    {
      "epoch": 8.072334864899043,
      "grad_norm": 1.1142332553863525,
      "learning_rate": 6.88719289971273e-05,
      "loss": 4.4794,
      "step": 191500
    },
    {
      "epoch": 8.093411457235595,
      "grad_norm": 1.1559245586395264,
      "learning_rate": 6.87861767354114e-05,
      "loss": 4.4746,
      "step": 192000
    },
    {
      "epoch": 8.114488049572145,
      "grad_norm": 1.1671228408813477,
      "learning_rate": 6.870042447369551e-05,
      "loss": 4.4813,
      "step": 192500
    },
    {
      "epoch": 8.135564641908696,
      "grad_norm": 1.1430891752243042,
      "learning_rate": 6.861467221197959e-05,
      "loss": 4.4798,
      "step": 193000
    },
    {
      "epoch": 8.156641234245248,
      "grad_norm": 1.1618921756744385,
      "learning_rate": 6.852891995026369e-05,
      "loss": 4.4812,
      "step": 193500
    },
    {
      "epoch": 8.177717826581798,
      "grad_norm": 1.1428983211517334,
      "learning_rate": 6.844316768854779e-05,
      "loss": 4.4825,
      "step": 194000
    },
    {
      "epoch": 8.19879441891835,
      "grad_norm": 1.1467968225479126,
      "learning_rate": 6.835741542683189e-05,
      "loss": 4.478,
      "step": 194500
    },
    {
      "epoch": 8.2198710112549,
      "grad_norm": 1.1417841911315918,
      "learning_rate": 6.827166316511599e-05,
      "loss": 4.4749,
      "step": 195000
    },
    {
      "epoch": 8.24094760359145,
      "grad_norm": 1.1041440963745117,
      "learning_rate": 6.818591090340008e-05,
      "loss": 4.4767,
      "step": 195500
    },
    {
      "epoch": 8.262024195928003,
      "grad_norm": 1.1128705739974976,
      "learning_rate": 6.810015864168417e-05,
      "loss": 4.4751,
      "step": 196000
    },
    {
      "epoch": 8.283100788264553,
      "grad_norm": 1.1191339492797852,
      "learning_rate": 6.801440637996828e-05,
      "loss": 4.4735,
      "step": 196500
    },
    {
      "epoch": 8.304177380601104,
      "grad_norm": 1.1993907690048218,
      "learning_rate": 6.792865411825238e-05,
      "loss": 4.4747,
      "step": 197000
    },
    {
      "epoch": 8.325253972937656,
      "grad_norm": 1.1371949911117554,
      "learning_rate": 6.784290185653646e-05,
      "loss": 4.474,
      "step": 197500
    },
    {
      "epoch": 8.346330565274206,
      "grad_norm": 1.2834117412567139,
      "learning_rate": 6.775714959482056e-05,
      "loss": 4.4752,
      "step": 198000
    },
    {
      "epoch": 8.367407157610758,
      "grad_norm": 1.1058961153030396,
      "learning_rate": 6.767139733310467e-05,
      "loss": 4.4775,
      "step": 198500
    },
    {
      "epoch": 8.388483749947309,
      "grad_norm": 1.142759084701538,
      "learning_rate": 6.758564507138876e-05,
      "loss": 4.4786,
      "step": 199000
    },
    {
      "epoch": 8.409560342283859,
      "grad_norm": 1.1553460359573364,
      "learning_rate": 6.749989280967286e-05,
      "loss": 4.4798,
      "step": 199500
    },
    {
      "epoch": 8.430636934620411,
      "grad_norm": 1.2074365615844727,
      "learning_rate": 6.741414054795696e-05,
      "loss": 4.4759,
      "step": 200000
    },
    {
      "epoch": 8.430636934620411,
      "eval_accuracy": 0.7194162508132406,
      "eval_loss": 4.223086357116699,
      "eval_runtime": 274.2842,
      "eval_samples_per_second": 440.076,
      "eval_steps_per_second": 4.586,
      "step": 200000
    },
    {
      "epoch": 8.451713526956961,
      "grad_norm": 1.1548733711242676,
      "learning_rate": 6.732838828624105e-05,
      "loss": 4.4737,
      "step": 200500
    },
    {
      "epoch": 8.472790119293512,
      "grad_norm": 1.178628921508789,
      "learning_rate": 6.724263602452515e-05,
      "loss": 4.4757,
      "step": 201000
    },
    {
      "epoch": 8.493866711630064,
      "grad_norm": 1.1705995798110962,
      "learning_rate": 6.715688376280924e-05,
      "loss": 4.4705,
      "step": 201500
    },
    {
      "epoch": 8.514943303966614,
      "grad_norm": 1.1712039709091187,
      "learning_rate": 6.707113150109335e-05,
      "loss": 4.4677,
      "step": 202000
    },
    {
      "epoch": 8.536019896303166,
      "grad_norm": 1.1746054887771606,
      "learning_rate": 6.698537923937745e-05,
      "loss": 4.4763,
      "step": 202500
    },
    {
      "epoch": 8.557096488639717,
      "grad_norm": 1.137654423713684,
      "learning_rate": 6.689962697766153e-05,
      "loss": 4.4685,
      "step": 203000
    },
    {
      "epoch": 8.578173080976267,
      "grad_norm": 1.1665832996368408,
      "learning_rate": 6.681387471594563e-05,
      "loss": 4.47,
      "step": 203500
    },
    {
      "epoch": 8.59924967331282,
      "grad_norm": 1.1618205308914185,
      "learning_rate": 6.672812245422974e-05,
      "loss": 4.4686,
      "step": 204000
    },
    {
      "epoch": 8.62032626564937,
      "grad_norm": 1.1608387231826782,
      "learning_rate": 6.664237019251383e-05,
      "loss": 4.4687,
      "step": 204500
    },
    {
      "epoch": 8.64140285798592,
      "grad_norm": 1.1802006959915161,
      "learning_rate": 6.655661793079792e-05,
      "loss": 4.4665,
      "step": 205000
    },
    {
      "epoch": 8.662479450322472,
      "grad_norm": 1.174422025680542,
      "learning_rate": 6.647086566908202e-05,
      "loss": 4.4624,
      "step": 205500
    },
    {
      "epoch": 8.683556042659022,
      "grad_norm": 1.1615204811096191,
      "learning_rate": 6.638511340736612e-05,
      "loss": 4.4624,
      "step": 206000
    },
    {
      "epoch": 8.704632634995574,
      "grad_norm": 1.1633237600326538,
      "learning_rate": 6.629936114565022e-05,
      "loss": 4.4615,
      "step": 206500
    },
    {
      "epoch": 8.725709227332125,
      "grad_norm": 1.1525323390960693,
      "learning_rate": 6.621360888393432e-05,
      "loss": 4.463,
      "step": 207000
    },
    {
      "epoch": 8.746785819668675,
      "grad_norm": 1.1440870761871338,
      "learning_rate": 6.61278566222184e-05,
      "loss": 4.4565,
      "step": 207500
    },
    {
      "epoch": 8.767862412005227,
      "grad_norm": 1.1216177940368652,
      "learning_rate": 6.604210436050251e-05,
      "loss": 4.4611,
      "step": 208000
    },
    {
      "epoch": 8.788939004341778,
      "grad_norm": 1.1415072679519653,
      "learning_rate": 6.595635209878661e-05,
      "loss": 4.4608,
      "step": 208500
    },
    {
      "epoch": 8.81001559667833,
      "grad_norm": 1.1483526229858398,
      "learning_rate": 6.58705998370707e-05,
      "loss": 4.4623,
      "step": 209000
    },
    {
      "epoch": 8.83109218901488,
      "grad_norm": 1.1670836210250854,
      "learning_rate": 6.57848475753548e-05,
      "loss": 4.4605,
      "step": 209500
    },
    {
      "epoch": 8.85216878135143,
      "grad_norm": 1.142865538597107,
      "learning_rate": 6.569909531363891e-05,
      "loss": 4.4563,
      "step": 210000
    },
    {
      "epoch": 8.85216878135143,
      "eval_accuracy": 0.7215297452157243,
      "eval_loss": 4.208909034729004,
      "eval_runtime": 277.8372,
      "eval_samples_per_second": 434.449,
      "eval_steps_per_second": 4.528,
      "step": 210000
    },
    {
      "epoch": 8.873245373687983,
      "grad_norm": 1.1463638544082642,
      "learning_rate": 6.561334305192299e-05,
      "loss": 4.4591,
      "step": 210500
    },
    {
      "epoch": 8.894321966024533,
      "grad_norm": 1.1487358808517456,
      "learning_rate": 6.552759079020709e-05,
      "loss": 4.4581,
      "step": 211000
    },
    {
      "epoch": 8.915398558361083,
      "grad_norm": 1.169592261314392,
      "learning_rate": 6.544183852849119e-05,
      "loss": 4.4571,
      "step": 211500
    },
    {
      "epoch": 8.936475150697635,
      "grad_norm": 1.1475026607513428,
      "learning_rate": 6.535608626677529e-05,
      "loss": 4.4623,
      "step": 212000
    },
    {
      "epoch": 8.957551743034186,
      "grad_norm": 1.110728144645691,
      "learning_rate": 6.527033400505939e-05,
      "loss": 4.4589,
      "step": 212500
    },
    {
      "epoch": 8.978628335370738,
      "grad_norm": 1.1302319765090942,
      "learning_rate": 6.518458174334348e-05,
      "loss": 4.4577,
      "step": 213000
    },
    {
      "epoch": 8.999704927707288,
      "grad_norm": 1.1736196279525757,
      "learning_rate": 6.509882948162758e-05,
      "loss": 4.4594,
      "step": 213500
    },
    {
      "epoch": 9.020781520043839,
      "grad_norm": 1.2273809909820557,
      "learning_rate": 6.501307721991168e-05,
      "loss": 4.4505,
      "step": 214000
    },
    {
      "epoch": 9.04185811238039,
      "grad_norm": 1.1520136594772339,
      "learning_rate": 6.492732495819578e-05,
      "loss": 4.4502,
      "step": 214500
    },
    {
      "epoch": 9.062934704716941,
      "grad_norm": 1.160274624824524,
      "learning_rate": 6.484157269647986e-05,
      "loss": 4.4508,
      "step": 215000
    },
    {
      "epoch": 9.084011297053493,
      "grad_norm": 1.1483266353607178,
      "learning_rate": 6.475582043476398e-05,
      "loss": 4.4498,
      "step": 215500
    },
    {
      "epoch": 9.105087889390044,
      "grad_norm": 1.1712725162506104,
      "learning_rate": 6.467006817304807e-05,
      "loss": 4.4469,
      "step": 216000
    },
    {
      "epoch": 9.126164481726594,
      "grad_norm": 1.1717654466629028,
      "learning_rate": 6.458431591133216e-05,
      "loss": 4.44,
      "step": 216500
    },
    {
      "epoch": 9.147241074063146,
      "grad_norm": 1.1783709526062012,
      "learning_rate": 6.449856364961626e-05,
      "loss": 4.4464,
      "step": 217000
    },
    {
      "epoch": 9.168317666399696,
      "grad_norm": 1.2069039344787598,
      "learning_rate": 6.441281138790037e-05,
      "loss": 4.4426,
      "step": 217500
    },
    {
      "epoch": 9.189394258736247,
      "grad_norm": 1.1284900903701782,
      "learning_rate": 6.432705912618445e-05,
      "loss": 4.4447,
      "step": 218000
    },
    {
      "epoch": 9.210470851072799,
      "grad_norm": 1.1834458112716675,
      "learning_rate": 6.424130686446855e-05,
      "loss": 4.4465,
      "step": 218500
    },
    {
      "epoch": 9.23154744340935,
      "grad_norm": 1.1460472345352173,
      "learning_rate": 6.415555460275265e-05,
      "loss": 4.4442,
      "step": 219000
    },
    {
      "epoch": 9.252624035745901,
      "grad_norm": 1.1449357271194458,
      "learning_rate": 6.406980234103675e-05,
      "loss": 4.4453,
      "step": 219500
    },
    {
      "epoch": 9.273700628082452,
      "grad_norm": 1.1727733612060547,
      "learning_rate": 6.398405007932085e-05,
      "loss": 4.4461,
      "step": 220000
    },
    {
      "epoch": 9.273700628082452,
      "eval_accuracy": 0.7233407623554237,
      "eval_loss": 4.19862174987793,
      "eval_runtime": 277.6895,
      "eval_samples_per_second": 434.68,
      "eval_steps_per_second": 4.53,
      "step": 220000
    },
    {
      "epoch": 9.294777220419002,
      "grad_norm": 1.1372312307357788,
      "learning_rate": 6.389829781760495e-05,
      "loss": 4.4432,
      "step": 220500
    },
    {
      "epoch": 9.315853812755554,
      "grad_norm": 1.196029782295227,
      "learning_rate": 6.381254555588903e-05,
      "loss": 4.4455,
      "step": 221000
    },
    {
      "epoch": 9.336930405092104,
      "grad_norm": 1.1588951349258423,
      "learning_rate": 6.372679329417314e-05,
      "loss": 4.4414,
      "step": 221500
    },
    {
      "epoch": 9.358006997428657,
      "grad_norm": 1.1560182571411133,
      "learning_rate": 6.364104103245724e-05,
      "loss": 4.4383,
      "step": 222000
    },
    {
      "epoch": 9.379083589765207,
      "grad_norm": 1.1393439769744873,
      "learning_rate": 6.355528877074132e-05,
      "loss": 4.444,
      "step": 222500
    },
    {
      "epoch": 9.400160182101757,
      "grad_norm": 1.1955194473266602,
      "learning_rate": 6.346953650902542e-05,
      "loss": 4.4416,
      "step": 223000
    },
    {
      "epoch": 9.42123677443831,
      "grad_norm": 1.1600620746612549,
      "learning_rate": 6.338378424730954e-05,
      "loss": 4.4416,
      "step": 223500
    },
    {
      "epoch": 9.44231336677486,
      "grad_norm": 1.1748638153076172,
      "learning_rate": 6.329803198559362e-05,
      "loss": 4.4376,
      "step": 224000
    },
    {
      "epoch": 9.46338995911141,
      "grad_norm": 1.1697100400924683,
      "learning_rate": 6.321227972387772e-05,
      "loss": 4.4365,
      "step": 224500
    },
    {
      "epoch": 9.484466551447962,
      "grad_norm": 1.1541308164596558,
      "learning_rate": 6.312652746216182e-05,
      "loss": 4.4346,
      "step": 225000
    },
    {
      "epoch": 9.505543143784513,
      "grad_norm": 1.143985390663147,
      "learning_rate": 6.304077520044591e-05,
      "loss": 4.4408,
      "step": 225500
    },
    {
      "epoch": 9.526619736121065,
      "grad_norm": 1.1697115898132324,
      "learning_rate": 6.295502293873001e-05,
      "loss": 4.4364,
      "step": 226000
    },
    {
      "epoch": 9.547696328457615,
      "grad_norm": 1.1900396347045898,
      "learning_rate": 6.286927067701411e-05,
      "loss": 4.4354,
      "step": 226500
    },
    {
      "epoch": 9.568772920794165,
      "grad_norm": 1.1745429039001465,
      "learning_rate": 6.278351841529821e-05,
      "loss": 4.4383,
      "step": 227000
    },
    {
      "epoch": 9.589849513130718,
      "grad_norm": 1.1616510152816772,
      "learning_rate": 6.269776615358231e-05,
      "loss": 4.4357,
      "step": 227500
    },
    {
      "epoch": 9.610926105467268,
      "grad_norm": 1.1585580110549927,
      "learning_rate": 6.26120138918664e-05,
      "loss": 4.4362,
      "step": 228000
    },
    {
      "epoch": 9.63200269780382,
      "grad_norm": 1.1407135725021362,
      "learning_rate": 6.252626163015049e-05,
      "loss": 4.434,
      "step": 228500
    },
    {
      "epoch": 9.65307929014037,
      "grad_norm": 1.182172417640686,
      "learning_rate": 6.24405093684346e-05,
      "loss": 4.4325,
      "step": 229000
    },
    {
      "epoch": 9.67415588247692,
      "grad_norm": 1.1857538223266602,
      "learning_rate": 6.23547571067187e-05,
      "loss": 4.4306,
      "step": 229500
    },
    {
      "epoch": 9.695232474813473,
      "grad_norm": 1.1536754369735718,
      "learning_rate": 6.226900484500279e-05,
      "loss": 4.4263,
      "step": 230000
    },
    {
      "epoch": 9.695232474813473,
      "eval_accuracy": 0.7251212070883123,
      "eval_loss": 4.1844635009765625,
      "eval_runtime": 271.2988,
      "eval_samples_per_second": 444.919,
      "eval_steps_per_second": 4.637,
      "step": 230000
    },
    {
      "epoch": 9.716309067150023,
      "grad_norm": 1.2050073146820068,
      "learning_rate": 6.218325258328688e-05,
      "loss": 4.435,
      "step": 230500
    },
    {
      "epoch": 9.737385659486574,
      "grad_norm": 1.1450605392456055,
      "learning_rate": 6.209750032157098e-05,
      "loss": 4.4332,
      "step": 231000
    },
    {
      "epoch": 9.758462251823126,
      "grad_norm": 1.1868361234664917,
      "learning_rate": 6.201174805985508e-05,
      "loss": 4.4303,
      "step": 231500
    },
    {
      "epoch": 9.779538844159676,
      "grad_norm": 1.2124894857406616,
      "learning_rate": 6.192599579813918e-05,
      "loss": 4.4311,
      "step": 232000
    },
    {
      "epoch": 9.800615436496226,
      "grad_norm": 1.1631382703781128,
      "learning_rate": 6.184024353642328e-05,
      "loss": 4.4283,
      "step": 232500
    },
    {
      "epoch": 9.821692028832778,
      "grad_norm": 1.1520304679870605,
      "learning_rate": 6.175449127470738e-05,
      "loss": 4.4241,
      "step": 233000
    },
    {
      "epoch": 9.842768621169329,
      "grad_norm": 1.150206208229065,
      "learning_rate": 6.166873901299147e-05,
      "loss": 4.4318,
      "step": 233500
    },
    {
      "epoch": 9.863845213505881,
      "grad_norm": 1.1402528285980225,
      "learning_rate": 6.158298675127557e-05,
      "loss": 4.4256,
      "step": 234000
    },
    {
      "epoch": 9.884921805842431,
      "grad_norm": 1.156522274017334,
      "learning_rate": 6.149723448955966e-05,
      "loss": 4.4298,
      "step": 234500
    },
    {
      "epoch": 9.905998398178982,
      "grad_norm": 1.2586175203323364,
      "learning_rate": 6.141148222784377e-05,
      "loss": 4.4342,
      "step": 235000
    },
    {
      "epoch": 9.927074990515534,
      "grad_norm": 1.1543622016906738,
      "learning_rate": 6.132572996612787e-05,
      "loss": 4.4242,
      "step": 235500
    },
    {
      "epoch": 9.948151582852084,
      "grad_norm": 1.1741142272949219,
      "learning_rate": 6.123997770441195e-05,
      "loss": 4.4272,
      "step": 236000
    },
    {
      "epoch": 9.969228175188636,
      "grad_norm": 1.1796863079071045,
      "learning_rate": 6.115422544269605e-05,
      "loss": 4.4296,
      "step": 236500
    },
    {
      "epoch": 9.990304767525187,
      "grad_norm": 1.2081612348556519,
      "learning_rate": 6.106847318098015e-05,
      "loss": 4.4275,
      "step": 237000
    },
    {
      "epoch": 10.011381359861737,
      "grad_norm": 1.1237479448318481,
      "learning_rate": 6.098272091926425e-05,
      "loss": 4.4254,
      "step": 237500
    },
    {
      "epoch": 10.032457952198289,
      "grad_norm": 1.153712511062622,
      "learning_rate": 6.0896968657548345e-05,
      "loss": 4.4219,
      "step": 238000
    },
    {
      "epoch": 10.05353454453484,
      "grad_norm": 1.1635050773620605,
      "learning_rate": 6.081121639583244e-05,
      "loss": 4.4184,
      "step": 238500
    },
    {
      "epoch": 10.07461113687139,
      "grad_norm": 1.195746660232544,
      "learning_rate": 6.072546413411654e-05,
      "loss": 4.4206,
      "step": 239000
    },
    {
      "epoch": 10.095687729207942,
      "grad_norm": 1.1460704803466797,
      "learning_rate": 6.063971187240064e-05,
      "loss": 4.4185,
      "step": 239500
    },
    {
      "epoch": 10.116764321544492,
      "grad_norm": 1.1387083530426025,
      "learning_rate": 6.055395961068473e-05,
      "loss": 4.4123,
      "step": 240000
    },
    {
      "epoch": 10.116764321544492,
      "eval_accuracy": 0.7269909345386374,
      "eval_loss": 4.17601203918457,
      "eval_runtime": 277.6291,
      "eval_samples_per_second": 434.774,
      "eval_steps_per_second": 4.531,
      "step": 240000
    },
    {
      "epoch": 10.137840913881044,
      "grad_norm": 1.1902437210083008,
      "learning_rate": 6.046820734896883e-05,
      "loss": 4.416,
      "step": 240500
    },
    {
      "epoch": 10.158917506217595,
      "grad_norm": 1.1988255977630615,
      "learning_rate": 6.038245508725293e-05,
      "loss": 4.4194,
      "step": 241000
    },
    {
      "epoch": 10.179994098554145,
      "grad_norm": 1.1322425603866577,
      "learning_rate": 6.029670282553702e-05,
      "loss": 4.4165,
      "step": 241500
    },
    {
      "epoch": 10.201070690890697,
      "grad_norm": 1.1622695922851562,
      "learning_rate": 6.0210950563821125e-05,
      "loss": 4.4194,
      "step": 242000
    },
    {
      "epoch": 10.222147283227248,
      "grad_norm": 1.2122379541397095,
      "learning_rate": 6.0125198302105223e-05,
      "loss": 4.4182,
      "step": 242500
    },
    {
      "epoch": 10.2432238755638,
      "grad_norm": 1.1540045738220215,
      "learning_rate": 6.0039446040389315e-05,
      "loss": 4.4156,
      "step": 243000
    },
    {
      "epoch": 10.26430046790035,
      "grad_norm": 1.171218752861023,
      "learning_rate": 5.995369377867341e-05,
      "loss": 4.4207,
      "step": 243500
    },
    {
      "epoch": 10.2853770602369,
      "grad_norm": 1.1413403749465942,
      "learning_rate": 5.986794151695752e-05,
      "loss": 4.4119,
      "step": 244000
    },
    {
      "epoch": 10.306453652573452,
      "grad_norm": 1.1586894989013672,
      "learning_rate": 5.97821892552416e-05,
      "loss": 4.4173,
      "step": 244500
    },
    {
      "epoch": 10.327530244910003,
      "grad_norm": 1.1535483598709106,
      "learning_rate": 5.969643699352571e-05,
      "loss": 4.4165,
      "step": 245000
    },
    {
      "epoch": 10.348606837246553,
      "grad_norm": 1.1423090696334839,
      "learning_rate": 5.9610684731809806e-05,
      "loss": 4.416,
      "step": 245500
    },
    {
      "epoch": 10.369683429583105,
      "grad_norm": 1.1479898691177368,
      "learning_rate": 5.95249324700939e-05,
      "loss": 4.4115,
      "step": 246000
    },
    {
      "epoch": 10.390760021919656,
      "grad_norm": 1.1805821657180786,
      "learning_rate": 5.9439180208377996e-05,
      "loss": 4.432,
      "step": 246500
    },
    {
      "epoch": 10.411836614256208,
      "grad_norm": 1.1339266300201416,
      "learning_rate": 5.93534279466621e-05,
      "loss": 4.4165,
      "step": 247000
    },
    {
      "epoch": 10.432913206592758,
      "grad_norm": 1.1768091917037964,
      "learning_rate": 5.9267675684946186e-05,
      "loss": 4.416,
      "step": 247500
    },
    {
      "epoch": 10.453989798929308,
      "grad_norm": 1.156909704208374,
      "learning_rate": 5.918192342323029e-05,
      "loss": 4.4102,
      "step": 248000
    },
    {
      "epoch": 10.47506639126586,
      "grad_norm": 1.1447136402130127,
      "learning_rate": 5.909617116151439e-05,
      "loss": 4.4103,
      "step": 248500
    },
    {
      "epoch": 10.496142983602411,
      "grad_norm": 1.2058275938034058,
      "learning_rate": 5.901041889979848e-05,
      "loss": 4.4088,
      "step": 249000
    },
    {
      "epoch": 10.517219575938963,
      "grad_norm": 1.1674219369888306,
      "learning_rate": 5.892466663808258e-05,
      "loss": 4.4114,
      "step": 249500
    },
    {
      "epoch": 10.538296168275513,
      "grad_norm": 1.1610088348388672,
      "learning_rate": 5.8838914376366684e-05,
      "loss": 4.4131,
      "step": 250000
    },
    {
      "epoch": 10.538296168275513,
      "eval_accuracy": 0.7284340516409187,
      "eval_loss": 4.164177417755127,
      "eval_runtime": 273.5846,
      "eval_samples_per_second": 441.202,
      "eval_steps_per_second": 4.598,
      "step": 250000
    },
    {
      "epoch": 10.559372760612064,
      "grad_norm": 1.1787890195846558,
      "learning_rate": 5.8753162114650776e-05,
      "loss": 4.4144,
      "step": 250500
    },
    {
      "epoch": 10.580449352948616,
      "grad_norm": 1.2173396348953247,
      "learning_rate": 5.8667409852934874e-05,
      "loss": 4.4081,
      "step": 251000
    },
    {
      "epoch": 10.601525945285166,
      "grad_norm": 1.159134030342102,
      "learning_rate": 5.858165759121897e-05,
      "loss": 4.4058,
      "step": 251500
    },
    {
      "epoch": 10.622602537621717,
      "grad_norm": 1.1657211780548096,
      "learning_rate": 5.8495905329503064e-05,
      "loss": 4.4088,
      "step": 252000
    },
    {
      "epoch": 10.643679129958269,
      "grad_norm": 1.1482386589050293,
      "learning_rate": 5.841015306778716e-05,
      "loss": 4.4074,
      "step": 252500
    },
    {
      "epoch": 10.664755722294819,
      "grad_norm": 1.179276704788208,
      "learning_rate": 5.832440080607127e-05,
      "loss": 4.4101,
      "step": 253000
    },
    {
      "epoch": 10.685832314631371,
      "grad_norm": 1.1603631973266602,
      "learning_rate": 5.823864854435536e-05,
      "loss": 4.4047,
      "step": 253500
    },
    {
      "epoch": 10.706908906967922,
      "grad_norm": 1.1711663007736206,
      "learning_rate": 5.815289628263946e-05,
      "loss": 4.4058,
      "step": 254000
    },
    {
      "epoch": 10.727985499304472,
      "grad_norm": 1.1791682243347168,
      "learning_rate": 5.8067144020923556e-05,
      "loss": 4.4107,
      "step": 254500
    },
    {
      "epoch": 10.749062091641024,
      "grad_norm": 1.1984466314315796,
      "learning_rate": 5.798139175920765e-05,
      "loss": 4.4069,
      "step": 255000
    },
    {
      "epoch": 10.770138683977574,
      "grad_norm": 1.1972206830978394,
      "learning_rate": 5.789563949749175e-05,
      "loss": 4.4071,
      "step": 255500
    },
    {
      "epoch": 10.791215276314126,
      "grad_norm": 1.1433113813400269,
      "learning_rate": 5.780988723577585e-05,
      "loss": 4.4027,
      "step": 256000
    },
    {
      "epoch": 10.812291868650677,
      "grad_norm": 1.1805214881896973,
      "learning_rate": 5.772413497405994e-05,
      "loss": 4.405,
      "step": 256500
    },
    {
      "epoch": 10.833368460987227,
      "grad_norm": 1.1352704763412476,
      "learning_rate": 5.763838271234404e-05,
      "loss": 4.4014,
      "step": 257000
    },
    {
      "epoch": 10.85444505332378,
      "grad_norm": 1.1817725896835327,
      "learning_rate": 5.7552630450628146e-05,
      "loss": 4.4049,
      "step": 257500
    },
    {
      "epoch": 10.87552164566033,
      "grad_norm": 1.1448544263839722,
      "learning_rate": 5.746687818891223e-05,
      "loss": 4.4017,
      "step": 258000
    },
    {
      "epoch": 10.89659823799688,
      "grad_norm": 1.1513675451278687,
      "learning_rate": 5.7381125927196335e-05,
      "loss": 4.4039,
      "step": 258500
    },
    {
      "epoch": 10.917674830333432,
      "grad_norm": 1.1631115674972534,
      "learning_rate": 5.7295373665480434e-05,
      "loss": 4.4006,
      "step": 259000
    },
    {
      "epoch": 10.938751422669982,
      "grad_norm": 1.2040818929672241,
      "learning_rate": 5.7209621403764525e-05,
      "loss": 4.4005,
      "step": 259500
    },
    {
      "epoch": 10.959828015006533,
      "grad_norm": 1.1675500869750977,
      "learning_rate": 5.7123869142048624e-05,
      "loss": 4.3987,
      "step": 260000
    },
    {
      "epoch": 10.959828015006533,
      "eval_accuracy": 0.7298374262404533,
      "eval_loss": 4.15518045425415,
      "eval_runtime": 275.9433,
      "eval_samples_per_second": 437.43,
      "eval_steps_per_second": 4.559,
      "step": 260000
    },
    {
      "epoch": 10.980904607343085,
      "grad_norm": 1.1762433052062988,
      "learning_rate": 5.703811688033273e-05,
      "loss": 4.3957,
      "step": 260500
    },
    {
      "epoch": 11.001981199679635,
      "grad_norm": 1.1535770893096924,
      "learning_rate": 5.695236461861681e-05,
      "loss": 4.3983,
      "step": 261000
    },
    {
      "epoch": 11.023057792016187,
      "grad_norm": 1.185641884803772,
      "learning_rate": 5.686661235690092e-05,
      "loss": 4.39,
      "step": 261500
    },
    {
      "epoch": 11.044134384352738,
      "grad_norm": 1.173852801322937,
      "learning_rate": 5.678086009518502e-05,
      "loss": 4.3979,
      "step": 262000
    },
    {
      "epoch": 11.065210976689288,
      "grad_norm": 1.2272343635559082,
      "learning_rate": 5.669510783346911e-05,
      "loss": 4.3994,
      "step": 262500
    },
    {
      "epoch": 11.08628756902584,
      "grad_norm": 1.4989439249038696,
      "learning_rate": 5.6609355571753207e-05,
      "loss": 4.3909,
      "step": 263000
    },
    {
      "epoch": 11.10736416136239,
      "grad_norm": 1.2014514207839966,
      "learning_rate": 5.652360331003731e-05,
      "loss": 4.3966,
      "step": 263500
    },
    {
      "epoch": 11.128440753698943,
      "grad_norm": 1.173733115196228,
      "learning_rate": 5.6437851048321396e-05,
      "loss": 4.3932,
      "step": 264000
    },
    {
      "epoch": 11.149517346035493,
      "grad_norm": 1.1928930282592773,
      "learning_rate": 5.63520987866055e-05,
      "loss": 4.3985,
      "step": 264500
    },
    {
      "epoch": 11.170593938372043,
      "grad_norm": 1.165486216545105,
      "learning_rate": 5.62663465248896e-05,
      "loss": 4.3952,
      "step": 265000
    },
    {
      "epoch": 11.191670530708596,
      "grad_norm": 1.1740100383758545,
      "learning_rate": 5.618059426317369e-05,
      "loss": 4.3928,
      "step": 265500
    },
    {
      "epoch": 11.212747123045146,
      "grad_norm": 1.1714136600494385,
      "learning_rate": 5.609484200145779e-05,
      "loss": 4.3972,
      "step": 266000
    },
    {
      "epoch": 11.233823715381696,
      "grad_norm": 1.1621508598327637,
      "learning_rate": 5.6009089739741895e-05,
      "loss": 4.3931,
      "step": 266500
    },
    {
      "epoch": 11.254900307718248,
      "grad_norm": 1.201778769493103,
      "learning_rate": 5.5923337478025986e-05,
      "loss": 4.3904,
      "step": 267000
    },
    {
      "epoch": 11.275976900054799,
      "grad_norm": 1.1961544752120972,
      "learning_rate": 5.5837585216310085e-05,
      "loss": 4.3944,
      "step": 267500
    },
    {
      "epoch": 11.29705349239135,
      "grad_norm": 1.2199435234069824,
      "learning_rate": 5.575183295459418e-05,
      "loss": 4.3885,
      "step": 268000
    },
    {
      "epoch": 11.318130084727901,
      "grad_norm": 1.2014232873916626,
      "learning_rate": 5.5666080692878274e-05,
      "loss": 4.3919,
      "step": 268500
    },
    {
      "epoch": 11.339206677064452,
      "grad_norm": 1.8504341840744019,
      "learning_rate": 5.558032843116238e-05,
      "loss": 4.3924,
      "step": 269000
    },
    {
      "epoch": 11.360283269401004,
      "grad_norm": 1.1582274436950684,
      "learning_rate": 5.549457616944648e-05,
      "loss": 4.3923,
      "step": 269500
    },
    {
      "epoch": 11.381359861737554,
      "grad_norm": 3.4087886810302734,
      "learning_rate": 5.540882390773057e-05,
      "loss": 4.3866,
      "step": 270000
    },
    {
      "epoch": 11.381359861737554,
      "eval_accuracy": 0.7306330101768203,
      "eval_loss": 4.150056838989258,
      "eval_runtime": 262.0832,
      "eval_samples_per_second": 460.564,
      "eval_steps_per_second": 4.8,
      "step": 270000
    },
    {
      "epoch": 11.402436454074106,
      "grad_norm": 1.1912403106689453,
      "learning_rate": 5.532307164601467e-05,
      "loss": 4.3889,
      "step": 270500
    },
    {
      "epoch": 11.423513046410656,
      "grad_norm": 1.1735081672668457,
      "learning_rate": 5.5237319384298766e-05,
      "loss": 4.3878,
      "step": 271000
    },
    {
      "epoch": 11.444589638747207,
      "grad_norm": 1.1555038690567017,
      "learning_rate": 5.515156712258286e-05,
      "loss": 4.3853,
      "step": 271500
    },
    {
      "epoch": 11.465666231083759,
      "grad_norm": 1.1812608242034912,
      "learning_rate": 5.506581486086696e-05,
      "loss": 4.3871,
      "step": 272000
    },
    {
      "epoch": 11.48674282342031,
      "grad_norm": 1.1786071062088013,
      "learning_rate": 5.498006259915105e-05,
      "loss": 4.3877,
      "step": 272500
    },
    {
      "epoch": 11.50781941575686,
      "grad_norm": 1.1883119344711304,
      "learning_rate": 5.489431033743515e-05,
      "loss": 4.3873,
      "step": 273000
    },
    {
      "epoch": 11.528896008093412,
      "grad_norm": 1.1601227521896362,
      "learning_rate": 5.480855807571925e-05,
      "loss": 4.3881,
      "step": 273500
    },
    {
      "epoch": 11.549972600429962,
      "grad_norm": 1.151808738708496,
      "learning_rate": 5.472280581400334e-05,
      "loss": 4.3848,
      "step": 274000
    },
    {
      "epoch": 11.571049192766514,
      "grad_norm": 1.2093164920806885,
      "learning_rate": 5.463705355228744e-05,
      "loss": 4.3845,
      "step": 274500
    },
    {
      "epoch": 11.592125785103065,
      "grad_norm": 1.2021443843841553,
      "learning_rate": 5.4551301290571546e-05,
      "loss": 4.3892,
      "step": 275000
    },
    {
      "epoch": 11.613202377439615,
      "grad_norm": 1.1772023439407349,
      "learning_rate": 5.446554902885563e-05,
      "loss": 4.3856,
      "step": 275500
    },
    {
      "epoch": 11.634278969776167,
      "grad_norm": 1.1881555318832397,
      "learning_rate": 5.4379796767139736e-05,
      "loss": 4.3835,
      "step": 276000
    },
    {
      "epoch": 11.655355562112717,
      "grad_norm": 1.188485860824585,
      "learning_rate": 5.4294044505423834e-05,
      "loss": 4.3837,
      "step": 276500
    },
    {
      "epoch": 11.67643215444927,
      "grad_norm": 1.1528960466384888,
      "learning_rate": 5.4208292243707925e-05,
      "loss": 4.3845,
      "step": 277000
    },
    {
      "epoch": 11.69750874678582,
      "grad_norm": 1.1723155975341797,
      "learning_rate": 5.4122539981992024e-05,
      "loss": 4.3833,
      "step": 277500
    },
    {
      "epoch": 11.71858533912237,
      "grad_norm": 1.1466975212097168,
      "learning_rate": 5.403678772027613e-05,
      "loss": 4.378,
      "step": 278000
    },
    {
      "epoch": 11.739661931458922,
      "grad_norm": 1.2107646465301514,
      "learning_rate": 5.395103545856022e-05,
      "loss": 4.3834,
      "step": 278500
    },
    {
      "epoch": 11.760738523795473,
      "grad_norm": 1.1653629541397095,
      "learning_rate": 5.386528319684432e-05,
      "loss": 4.3814,
      "step": 279000
    },
    {
      "epoch": 11.781815116132023,
      "grad_norm": 1.1622916460037231,
      "learning_rate": 5.377953093512842e-05,
      "loss": 4.3785,
      "step": 279500
    },
    {
      "epoch": 11.802891708468575,
      "grad_norm": 1.1830071210861206,
      "learning_rate": 5.369377867341251e-05,
      "loss": 4.3844,
      "step": 280000
    },
    {
      "epoch": 11.802891708468575,
      "eval_accuracy": 0.7325285642565248,
      "eval_loss": 4.138386249542236,
      "eval_runtime": 258.6339,
      "eval_samples_per_second": 466.706,
      "eval_steps_per_second": 4.864,
      "step": 280000
    },
    {
      "epoch": 11.823968300805126,
      "grad_norm": 1.1943190097808838,
      "learning_rate": 5.3608026411696614e-05,
      "loss": 4.3852,
      "step": 280500
    },
    {
      "epoch": 11.845044893141678,
      "grad_norm": 1.1835899353027344,
      "learning_rate": 5.352227414998071e-05,
      "loss": 4.3742,
      "step": 281000
    },
    {
      "epoch": 11.866121485478228,
      "grad_norm": 1.2238751649856567,
      "learning_rate": 5.34365218882648e-05,
      "loss": 4.3788,
      "step": 281500
    },
    {
      "epoch": 11.887198077814778,
      "grad_norm": 1.1654952764511108,
      "learning_rate": 5.33507696265489e-05,
      "loss": 4.3728,
      "step": 282000
    },
    {
      "epoch": 11.90827467015133,
      "grad_norm": 1.147831916809082,
      "learning_rate": 5.3265017364833e-05,
      "loss": 4.3798,
      "step": 282500
    },
    {
      "epoch": 11.92935126248788,
      "grad_norm": 1.1907230615615845,
      "learning_rate": 5.317926510311709e-05,
      "loss": 4.3778,
      "step": 283000
    },
    {
      "epoch": 11.950427854824433,
      "grad_norm": 1.230374813079834,
      "learning_rate": 5.3093512841401197e-05,
      "loss": 4.3769,
      "step": 283500
    },
    {
      "epoch": 11.971504447160983,
      "grad_norm": 1.1775585412979126,
      "learning_rate": 5.3007760579685295e-05,
      "loss": 4.3799,
      "step": 284000
    },
    {
      "epoch": 11.992581039497534,
      "grad_norm": 1.1675734519958496,
      "learning_rate": 5.2922008317969386e-05,
      "loss": 4.3803,
      "step": 284500
    },
    {
      "epoch": 12.013657631834086,
      "grad_norm": 1.213075876235962,
      "learning_rate": 5.2836256056253485e-05,
      "loss": 4.374,
      "step": 285000
    },
    {
      "epoch": 12.034734224170636,
      "grad_norm": 1.1833487749099731,
      "learning_rate": 5.275050379453759e-05,
      "loss": 4.3757,
      "step": 285500
    },
    {
      "epoch": 12.055810816507186,
      "grad_norm": 1.1432676315307617,
      "learning_rate": 5.2664751532821675e-05,
      "loss": 4.3716,
      "step": 286000
    },
    {
      "epoch": 12.076887408843739,
      "grad_norm": 1.153456449508667,
      "learning_rate": 5.257899927110578e-05,
      "loss": 4.3718,
      "step": 286500
    },
    {
      "epoch": 12.097964001180289,
      "grad_norm": 1.1587854623794556,
      "learning_rate": 5.249324700938988e-05,
      "loss": 4.3689,
      "step": 287000
    },
    {
      "epoch": 12.119040593516841,
      "grad_norm": 1.1728121042251587,
      "learning_rate": 5.240749474767397e-05,
      "loss": 4.3688,
      "step": 287500
    },
    {
      "epoch": 12.140117185853391,
      "grad_norm": 1.2219918966293335,
      "learning_rate": 5.232174248595807e-05,
      "loss": 4.3742,
      "step": 288000
    },
    {
      "epoch": 12.161193778189942,
      "grad_norm": 1.1592674255371094,
      "learning_rate": 5.223599022424217e-05,
      "loss": 4.3726,
      "step": 288500
    },
    {
      "epoch": 12.182270370526494,
      "grad_norm": 1.213034749031067,
      "learning_rate": 5.215023796252626e-05,
      "loss": 4.3759,
      "step": 289000
    },
    {
      "epoch": 12.203346962863044,
      "grad_norm": 1.1780245304107666,
      "learning_rate": 5.206448570081036e-05,
      "loss": 4.3669,
      "step": 289500
    },
    {
      "epoch": 12.224423555199595,
      "grad_norm": 1.18780517578125,
      "learning_rate": 5.197873343909446e-05,
      "loss": 4.3661,
      "step": 290000
    },
    {
      "epoch": 12.224423555199595,
      "eval_accuracy": 0.7336225990479237,
      "eval_loss": 4.1314215660095215,
      "eval_runtime": 270.0491,
      "eval_samples_per_second": 446.978,
      "eval_steps_per_second": 4.658,
      "step": 290000
    },
    {
      "epoch": 12.245500147536147,
      "grad_norm": 1.2138501405715942,
      "learning_rate": 5.189298117737855e-05,
      "loss": 4.368,
      "step": 290500
    },
    {
      "epoch": 12.266576739872697,
      "grad_norm": 1.1736468076705933,
      "learning_rate": 5.180722891566265e-05,
      "loss": 4.3681,
      "step": 291000
    },
    {
      "epoch": 12.28765333220925,
      "grad_norm": 1.1888612508773804,
      "learning_rate": 5.1721476653946756e-05,
      "loss": 4.3635,
      "step": 291500
    },
    {
      "epoch": 12.3087299245458,
      "grad_norm": 1.1900362968444824,
      "learning_rate": 5.163572439223085e-05,
      "loss": 4.3691,
      "step": 292000
    },
    {
      "epoch": 12.32980651688235,
      "grad_norm": 1.193382978439331,
      "learning_rate": 5.1549972130514946e-05,
      "loss": 4.3692,
      "step": 292500
    },
    {
      "epoch": 12.350883109218902,
      "grad_norm": 1.1995850801467896,
      "learning_rate": 5.1464219868799044e-05,
      "loss": 4.3681,
      "step": 293000
    },
    {
      "epoch": 12.371959701555452,
      "grad_norm": 1.1981472969055176,
      "learning_rate": 5.1378467607083136e-05,
      "loss": 4.364,
      "step": 293500
    },
    {
      "epoch": 12.393036293892003,
      "grad_norm": 1.1869224309921265,
      "learning_rate": 5.1292715345367234e-05,
      "loss": 4.3667,
      "step": 294000
    },
    {
      "epoch": 12.414112886228555,
      "grad_norm": 1.1817636489868164,
      "learning_rate": 5.120696308365134e-05,
      "loss": 4.3637,
      "step": 294500
    },
    {
      "epoch": 12.435189478565105,
      "grad_norm": 1.1855127811431885,
      "learning_rate": 5.112121082193543e-05,
      "loss": 4.3693,
      "step": 295000
    },
    {
      "epoch": 12.456266070901657,
      "grad_norm": 1.2304267883300781,
      "learning_rate": 5.103545856021953e-05,
      "loss": 4.3651,
      "step": 295500
    },
    {
      "epoch": 12.477342663238208,
      "grad_norm": 1.1854957342147827,
      "learning_rate": 5.094970629850363e-05,
      "loss": 4.3655,
      "step": 296000
    },
    {
      "epoch": 12.498419255574758,
      "grad_norm": 1.1764925718307495,
      "learning_rate": 5.086395403678772e-05,
      "loss": 4.3647,
      "step": 296500
    },
    {
      "epoch": 12.51949584791131,
      "grad_norm": 1.171266794204712,
      "learning_rate": 5.0778201775071824e-05,
      "loss": 4.3697,
      "step": 297000
    },
    {
      "epoch": 12.54057244024786,
      "grad_norm": 1.2083613872528076,
      "learning_rate": 5.069244951335592e-05,
      "loss": 4.3687,
      "step": 297500
    },
    {
      "epoch": 12.561649032584413,
      "grad_norm": 1.1801468133926392,
      "learning_rate": 5.0606697251640014e-05,
      "loss": 4.3624,
      "step": 298000
    },
    {
      "epoch": 12.582725624920963,
      "grad_norm": 1.2129511833190918,
      "learning_rate": 5.052094498992411e-05,
      "loss": 4.3674,
      "step": 298500
    },
    {
      "epoch": 12.603802217257513,
      "grad_norm": 1.2167295217514038,
      "learning_rate": 5.043519272820822e-05,
      "loss": 4.3656,
      "step": 299000
    },
    {
      "epoch": 12.624878809594065,
      "grad_norm": 1.167583703994751,
      "learning_rate": 5.03494404664923e-05,
      "loss": 4.3589,
      "step": 299500
    },
    {
      "epoch": 12.645955401930616,
      "grad_norm": 1.2089864015579224,
      "learning_rate": 5.026368820477641e-05,
      "loss": 4.3614,
      "step": 300000
    },
    {
      "epoch": 12.645955401930616,
      "eval_accuracy": 0.7351368309847621,
      "eval_loss": 4.120747089385986,
      "eval_runtime": 272.773,
      "eval_samples_per_second": 442.515,
      "eval_steps_per_second": 4.612,
      "step": 300000
    },
    {
      "epoch": 12.667031994267166,
      "grad_norm": 1.1771266460418701,
      "learning_rate": 5.0177935943060505e-05,
      "loss": 4.3588,
      "step": 300500
    },
    {
      "epoch": 12.688108586603718,
      "grad_norm": 1.1999653577804565,
      "learning_rate": 5.00921836813446e-05,
      "loss": 4.3642,
      "step": 301000
    },
    {
      "epoch": 12.709185178940269,
      "grad_norm": 1.174315094947815,
      "learning_rate": 5.0006431419628695e-05,
      "loss": 4.3626,
      "step": 301500
    },
    {
      "epoch": 12.73026177127682,
      "grad_norm": 1.2025353908538818,
      "learning_rate": 4.992067915791279e-05,
      "loss": 4.3615,
      "step": 302000
    },
    {
      "epoch": 12.751338363613371,
      "grad_norm": 1.221038818359375,
      "learning_rate": 4.9834926896196885e-05,
      "loss": 4.3572,
      "step": 302500
    },
    {
      "epoch": 12.772414955949921,
      "grad_norm": 1.2587604522705078,
      "learning_rate": 4.974917463448099e-05,
      "loss": 4.3625,
      "step": 303000
    },
    {
      "epoch": 12.793491548286474,
      "grad_norm": 1.1968122720718384,
      "learning_rate": 4.966342237276508e-05,
      "loss": 4.3618,
      "step": 303500
    },
    {
      "epoch": 12.814568140623024,
      "grad_norm": 1.1830565929412842,
      "learning_rate": 4.957767011104918e-05,
      "loss": 4.359,
      "step": 304000
    },
    {
      "epoch": 12.835644732959576,
      "grad_norm": 1.1904743909835815,
      "learning_rate": 4.949191784933328e-05,
      "loss": 4.3597,
      "step": 304500
    },
    {
      "epoch": 12.856721325296126,
      "grad_norm": 1.1873362064361572,
      "learning_rate": 4.9406165587617376e-05,
      "loss": 4.3577,
      "step": 305000
    },
    {
      "epoch": 12.877797917632677,
      "grad_norm": 1.1889967918395996,
      "learning_rate": 4.932041332590147e-05,
      "loss": 4.3609,
      "step": 305500
    },
    {
      "epoch": 12.898874509969229,
      "grad_norm": 1.2139995098114014,
      "learning_rate": 4.923466106418557e-05,
      "loss": 4.3562,
      "step": 306000
    },
    {
      "epoch": 12.91995110230578,
      "grad_norm": 1.1859562397003174,
      "learning_rate": 4.9148908802469665e-05,
      "loss": 4.3551,
      "step": 306500
    },
    {
      "epoch": 12.94102769464233,
      "grad_norm": 1.1756069660186768,
      "learning_rate": 4.906315654075376e-05,
      "loss": 4.3582,
      "step": 307000
    },
    {
      "epoch": 12.962104286978882,
      "grad_norm": 1.2132295370101929,
      "learning_rate": 4.897740427903786e-05,
      "loss": 4.3607,
      "step": 307500
    },
    {
      "epoch": 12.983180879315432,
      "grad_norm": 1.2050633430480957,
      "learning_rate": 4.889165201732196e-05,
      "loss": 4.3604,
      "step": 308000
    },
    {
      "epoch": 13.004257471651984,
      "grad_norm": 1.17202889919281,
      "learning_rate": 4.880589975560606e-05,
      "loss": 4.3558,
      "step": 308500
    },
    {
      "epoch": 13.025334063988534,
      "grad_norm": 1.1899088621139526,
      "learning_rate": 4.8720147493890156e-05,
      "loss": 4.3506,
      "step": 309000
    },
    {
      "epoch": 13.046410656325085,
      "grad_norm": 1.17721426486969,
      "learning_rate": 4.8634395232174254e-05,
      "loss": 4.354,
      "step": 309500
    },
    {
      "epoch": 13.067487248661637,
      "grad_norm": 1.1829088926315308,
      "learning_rate": 4.8548642970458346e-05,
      "loss": 4.3483,
      "step": 310000
    },
    {
      "epoch": 13.067487248661637,
      "eval_accuracy": 0.7364853900620065,
      "eval_loss": 4.1125874519348145,
      "eval_runtime": 258.1122,
      "eval_samples_per_second": 467.649,
      "eval_steps_per_second": 4.874,
      "step": 310000
    },
    {
      "epoch": 13.088563840998187,
      "grad_norm": 1.1930055618286133,
      "learning_rate": 4.846289070874245e-05,
      "loss": 4.3513,
      "step": 310500
    },
    {
      "epoch": 13.109640433334738,
      "grad_norm": 1.2451531887054443,
      "learning_rate": 4.837713844702654e-05,
      "loss": 4.352,
      "step": 311000
    },
    {
      "epoch": 13.13071702567129,
      "grad_norm": 1.1988741159439087,
      "learning_rate": 4.829138618531064e-05,
      "loss": 4.3495,
      "step": 311500
    },
    {
      "epoch": 13.15179361800784,
      "grad_norm": 1.1919677257537842,
      "learning_rate": 4.820563392359474e-05,
      "loss": 4.3519,
      "step": 312000
    },
    {
      "epoch": 13.172870210344392,
      "grad_norm": 1.2187660932540894,
      "learning_rate": 4.811988166187884e-05,
      "loss": 4.3504,
      "step": 312500
    },
    {
      "epoch": 13.193946802680943,
      "grad_norm": 1.2096967697143555,
      "learning_rate": 4.803412940016293e-05,
      "loss": 4.3492,
      "step": 313000
    },
    {
      "epoch": 13.215023395017493,
      "grad_norm": 1.206836223602295,
      "learning_rate": 4.7948377138447034e-05,
      "loss": 4.3512,
      "step": 313500
    },
    {
      "epoch": 13.236099987354045,
      "grad_norm": 1.1900012493133545,
      "learning_rate": 4.7862624876731126e-05,
      "loss": 4.3475,
      "step": 314000
    },
    {
      "epoch": 13.257176579690595,
      "grad_norm": 1.1741305589675903,
      "learning_rate": 4.7776872615015224e-05,
      "loss": 4.3488,
      "step": 314500
    },
    {
      "epoch": 13.278253172027147,
      "grad_norm": 1.2190009355545044,
      "learning_rate": 4.7691120353299315e-05,
      "loss": 4.3454,
      "step": 315000
    },
    {
      "epoch": 13.299329764363698,
      "grad_norm": 1.1416819095611572,
      "learning_rate": 4.760536809158342e-05,
      "loss": 4.3523,
      "step": 315500
    },
    {
      "epoch": 13.320406356700248,
      "grad_norm": 1.1666723489761353,
      "learning_rate": 4.751961582986751e-05,
      "loss": 4.3492,
      "step": 316000
    },
    {
      "epoch": 13.3414829490368,
      "grad_norm": 1.165536642074585,
      "learning_rate": 4.743386356815161e-05,
      "loss": 4.3499,
      "step": 316500
    },
    {
      "epoch": 13.36255954137335,
      "grad_norm": 1.1903507709503174,
      "learning_rate": 4.734811130643571e-05,
      "loss": 4.3473,
      "step": 317000
    },
    {
      "epoch": 13.383636133709901,
      "grad_norm": 1.1586719751358032,
      "learning_rate": 4.726235904471981e-05,
      "loss": 4.3441,
      "step": 317500
    },
    {
      "epoch": 13.404712726046453,
      "grad_norm": 1.2590084075927734,
      "learning_rate": 4.71766067830039e-05,
      "loss": 4.3487,
      "step": 318000
    },
    {
      "epoch": 13.425789318383003,
      "grad_norm": 1.1629725694656372,
      "learning_rate": 4.7090854521288004e-05,
      "loss": 4.351,
      "step": 318500
    },
    {
      "epoch": 13.446865910719556,
      "grad_norm": 1.2204878330230713,
      "learning_rate": 4.7005102259572095e-05,
      "loss": 4.3477,
      "step": 319000
    },
    {
      "epoch": 13.467942503056106,
      "grad_norm": 1.235795021057129,
      "learning_rate": 4.6919349997856193e-05,
      "loss": 4.3487,
      "step": 319500
    },
    {
      "epoch": 13.489019095392656,
      "grad_norm": 1.2118037939071655,
      "learning_rate": 4.683359773614029e-05,
      "loss": 4.3453,
      "step": 320000
    },
    {
      "epoch": 13.489019095392656,
      "eval_accuracy": 0.7371289007944797,
      "eval_loss": 4.1086931228637695,
      "eval_runtime": 256.1143,
      "eval_samples_per_second": 471.297,
      "eval_steps_per_second": 4.912,
      "step": 320000
    },
    {
      "epoch": 13.510095687729208,
      "grad_norm": 1.2356196641921997,
      "learning_rate": 4.674784547442439e-05,
      "loss": 4.3497,
      "step": 320500
    },
    {
      "epoch": 13.531172280065759,
      "grad_norm": 1.205570101737976,
      "learning_rate": 4.666209321270849e-05,
      "loss": 4.3466,
      "step": 321000
    },
    {
      "epoch": 13.55224887240231,
      "grad_norm": 1.2005232572555542,
      "learning_rate": 4.657634095099259e-05,
      "loss": 4.3454,
      "step": 321500
    },
    {
      "epoch": 13.573325464738861,
      "grad_norm": 1.2084245681762695,
      "learning_rate": 4.6490588689276685e-05,
      "loss": 4.3447,
      "step": 322000
    },
    {
      "epoch": 13.594402057075412,
      "grad_norm": 1.1685093641281128,
      "learning_rate": 4.6404836427560777e-05,
      "loss": 4.3471,
      "step": 322500
    },
    {
      "epoch": 13.615478649411964,
      "grad_norm": 1.2132033109664917,
      "learning_rate": 4.6319084165844875e-05,
      "loss": 4.3401,
      "step": 323000
    },
    {
      "epoch": 13.636555241748514,
      "grad_norm": 1.202457070350647,
      "learning_rate": 4.623333190412897e-05,
      "loss": 4.3463,
      "step": 323500
    },
    {
      "epoch": 13.657631834085064,
      "grad_norm": 1.2460639476776123,
      "learning_rate": 4.614757964241307e-05,
      "loss": 4.3423,
      "step": 324000
    },
    {
      "epoch": 13.678708426421617,
      "grad_norm": 1.2022384405136108,
      "learning_rate": 4.606182738069717e-05,
      "loss": 4.3438,
      "step": 324500
    },
    {
      "epoch": 13.699785018758167,
      "grad_norm": 1.1920162439346313,
      "learning_rate": 4.597607511898127e-05,
      "loss": 4.3419,
      "step": 325000
    },
    {
      "epoch": 13.720861611094719,
      "grad_norm": 1.3628007173538208,
      "learning_rate": 4.589032285726536e-05,
      "loss": 4.343,
      "step": 325500
    },
    {
      "epoch": 13.74193820343127,
      "grad_norm": 1.180465579032898,
      "learning_rate": 4.5804570595549465e-05,
      "loss": 4.3429,
      "step": 326000
    },
    {
      "epoch": 13.76301479576782,
      "grad_norm": 1.1669490337371826,
      "learning_rate": 4.5718818333833556e-05,
      "loss": 4.3444,
      "step": 326500
    },
    {
      "epoch": 13.784091388104372,
      "grad_norm": 1.1546744108200073,
      "learning_rate": 4.5633066072117655e-05,
      "loss": 4.3458,
      "step": 327000
    },
    {
      "epoch": 13.805167980440922,
      "grad_norm": 1.1786580085754395,
      "learning_rate": 4.554731381040175e-05,
      "loss": 4.337,
      "step": 327500
    },
    {
      "epoch": 13.826244572777473,
      "grad_norm": 1.22744882106781,
      "learning_rate": 4.546156154868585e-05,
      "loss": 4.3426,
      "step": 328000
    },
    {
      "epoch": 13.847321165114025,
      "grad_norm": 1.245670199394226,
      "learning_rate": 4.537580928696994e-05,
      "loss": 4.342,
      "step": 328500
    },
    {
      "epoch": 13.868397757450575,
      "grad_norm": 1.417982578277588,
      "learning_rate": 4.529005702525405e-05,
      "loss": 4.3407,
      "step": 329000
    },
    {
      "epoch": 13.889474349787127,
      "grad_norm": 1.2358019351959229,
      "learning_rate": 4.520430476353814e-05,
      "loss": 4.3388,
      "step": 329500
    },
    {
      "epoch": 13.910550942123677,
      "grad_norm": 1.237534523010254,
      "learning_rate": 4.511855250182224e-05,
      "loss": 4.3391,
      "step": 330000
    },
    {
      "epoch": 13.910550942123677,
      "eval_accuracy": 0.737929904791243,
      "eval_loss": 4.103092670440674,
      "eval_runtime": 254.7218,
      "eval_samples_per_second": 473.874,
      "eval_steps_per_second": 4.939,
      "step": 330000
    },
    {
      "epoch": 13.931627534460228,
      "grad_norm": 1.1705646514892578,
      "learning_rate": 4.5032800240106336e-05,
      "loss": 4.3408,
      "step": 330500
    },
    {
      "epoch": 13.95270412679678,
      "grad_norm": 1.1684753894805908,
      "learning_rate": 4.4947047978390434e-05,
      "loss": 4.3439,
      "step": 331000
    },
    {
      "epoch": 13.97378071913333,
      "grad_norm": 1.2012567520141602,
      "learning_rate": 4.4861295716674526e-05,
      "loss": 4.3415,
      "step": 331500
    },
    {
      "epoch": 13.994857311469882,
      "grad_norm": 1.2190338373184204,
      "learning_rate": 4.4775543454958624e-05,
      "loss": 4.3409,
      "step": 332000
    },
    {
      "epoch": 14.015933903806433,
      "grad_norm": 1.2103983163833618,
      "learning_rate": 4.468979119324272e-05,
      "loss": 4.3349,
      "step": 332500
    },
    {
      "epoch": 14.037010496142983,
      "grad_norm": 1.1910558938980103,
      "learning_rate": 4.460403893152682e-05,
      "loss": 4.3358,
      "step": 333000
    },
    {
      "epoch": 14.058087088479535,
      "grad_norm": 1.2209070920944214,
      "learning_rate": 4.451828666981092e-05,
      "loss": 4.3342,
      "step": 333500
    },
    {
      "epoch": 14.079163680816086,
      "grad_norm": 1.20394766330719,
      "learning_rate": 4.443253440809502e-05,
      "loss": 4.3353,
      "step": 334000
    },
    {
      "epoch": 14.100240273152636,
      "grad_norm": 1.2366174459457397,
      "learning_rate": 4.4346782146379116e-05,
      "loss": 4.3329,
      "step": 334500
    },
    {
      "epoch": 14.121316865489188,
      "grad_norm": 1.1978815793991089,
      "learning_rate": 4.426102988466321e-05,
      "loss": 4.3323,
      "step": 335000
    },
    {
      "epoch": 14.142393457825738,
      "grad_norm": 1.2050082683563232,
      "learning_rate": 4.4175277622947305e-05,
      "loss": 4.3256,
      "step": 335500
    },
    {
      "epoch": 14.16347005016229,
      "grad_norm": 1.2054814100265503,
      "learning_rate": 4.4089525361231404e-05,
      "loss": 4.333,
      "step": 336000
    },
    {
      "epoch": 14.184546642498841,
      "grad_norm": 1.169961929321289,
      "learning_rate": 4.40037730995155e-05,
      "loss": 4.3285,
      "step": 336500
    },
    {
      "epoch": 14.205623234835391,
      "grad_norm": 1.2171956300735474,
      "learning_rate": 4.39180208377996e-05,
      "loss": 4.3331,
      "step": 337000
    },
    {
      "epoch": 14.226699827171943,
      "grad_norm": 1.2457071542739868,
      "learning_rate": 4.38322685760837e-05,
      "loss": 4.3339,
      "step": 337500
    },
    {
      "epoch": 14.247776419508494,
      "grad_norm": 1.1889572143554688,
      "learning_rate": 4.374651631436779e-05,
      "loss": 4.3325,
      "step": 338000
    },
    {
      "epoch": 14.268853011845044,
      "grad_norm": 1.202056884765625,
      "learning_rate": 4.3660764052651895e-05,
      "loss": 4.3328,
      "step": 338500
    },
    {
      "epoch": 14.289929604181596,
      "grad_norm": 1.200290560722351,
      "learning_rate": 4.357501179093599e-05,
      "loss": 4.332,
      "step": 339000
    },
    {
      "epoch": 14.311006196518147,
      "grad_norm": 1.230680227279663,
      "learning_rate": 4.3489259529220085e-05,
      "loss": 4.3315,
      "step": 339500
    },
    {
      "epoch": 14.332082788854699,
      "grad_norm": 1.2586053609848022,
      "learning_rate": 4.3403507267504183e-05,
      "loss": 4.3372,
      "step": 340000
    },
    {
      "epoch": 14.332082788854699,
      "eval_accuracy": 0.739579108658673,
      "eval_loss": 4.094237327575684,
      "eval_runtime": 255.3715,
      "eval_samples_per_second": 472.668,
      "eval_steps_per_second": 4.926,
      "step": 340000
    },
    {
      "epoch": 14.353159381191249,
      "grad_norm": 1.1800273656845093,
      "learning_rate": 4.331775500578828e-05,
      "loss": 4.3296,
      "step": 340500
    },
    {
      "epoch": 14.3742359735278,
      "grad_norm": 1.2091583013534546,
      "learning_rate": 4.323200274407237e-05,
      "loss": 4.3298,
      "step": 341000
    },
    {
      "epoch": 14.395312565864351,
      "grad_norm": 1.2177671194076538,
      "learning_rate": 4.314625048235648e-05,
      "loss": 4.3299,
      "step": 341500
    },
    {
      "epoch": 14.416389158200902,
      "grad_norm": 1.2240327596664429,
      "learning_rate": 4.306049822064057e-05,
      "loss": 4.3342,
      "step": 342000
    },
    {
      "epoch": 14.437465750537454,
      "grad_norm": 1.2206676006317139,
      "learning_rate": 4.297474595892467e-05,
      "loss": 4.3295,
      "step": 342500
    },
    {
      "epoch": 14.458542342874004,
      "grad_norm": 1.1926277875900269,
      "learning_rate": 4.2888993697208767e-05,
      "loss": 4.3281,
      "step": 343000
    },
    {
      "epoch": 14.479618935210555,
      "grad_norm": 1.1899778842926025,
      "learning_rate": 4.2803241435492865e-05,
      "loss": 4.3296,
      "step": 343500
    },
    {
      "epoch": 14.500695527547107,
      "grad_norm": 1.1982485055923462,
      "learning_rate": 4.2717489173776956e-05,
      "loss": 4.3291,
      "step": 344000
    },
    {
      "epoch": 14.521772119883657,
      "grad_norm": 1.2076399326324463,
      "learning_rate": 4.263173691206106e-05,
      "loss": 4.3284,
      "step": 344500
    },
    {
      "epoch": 14.542848712220207,
      "grad_norm": 1.2251867055892944,
      "learning_rate": 4.254598465034515e-05,
      "loss": 4.3301,
      "step": 345000
    },
    {
      "epoch": 14.56392530455676,
      "grad_norm": 1.1994904279708862,
      "learning_rate": 4.246023238862925e-05,
      "loss": 4.3293,
      "step": 345500
    },
    {
      "epoch": 14.58500189689331,
      "grad_norm": 1.2246085405349731,
      "learning_rate": 4.237448012691335e-05,
      "loss": 4.3275,
      "step": 346000
    },
    {
      "epoch": 14.606078489229862,
      "grad_norm": 1.2302201986312866,
      "learning_rate": 4.228872786519745e-05,
      "loss": 4.3272,
      "step": 346500
    },
    {
      "epoch": 14.627155081566412,
      "grad_norm": 1.2145278453826904,
      "learning_rate": 4.220297560348154e-05,
      "loss": 4.3304,
      "step": 347000
    },
    {
      "epoch": 14.648231673902963,
      "grad_norm": 1.2226327657699585,
      "learning_rate": 4.2117223341765645e-05,
      "loss": 4.3255,
      "step": 347500
    },
    {
      "epoch": 14.669308266239515,
      "grad_norm": 1.2108298540115356,
      "learning_rate": 4.2031471080049736e-05,
      "loss": 4.325,
      "step": 348000
    },
    {
      "epoch": 14.690384858576065,
      "grad_norm": 1.2322731018066406,
      "learning_rate": 4.1945718818333834e-05,
      "loss": 4.3258,
      "step": 348500
    },
    {
      "epoch": 14.711461450912616,
      "grad_norm": 1.1907821893692017,
      "learning_rate": 4.185996655661793e-05,
      "loss": 4.3274,
      "step": 349000
    },
    {
      "epoch": 14.732538043249168,
      "grad_norm": 1.2267268896102905,
      "learning_rate": 4.177421429490203e-05,
      "loss": 4.3283,
      "step": 349500
    },
    {
      "epoch": 14.753614635585718,
      "grad_norm": 1.2292156219482422,
      "learning_rate": 4.168846203318613e-05,
      "loss": 4.3255,
      "step": 350000
    },
    {
      "epoch": 14.753614635585718,
      "eval_accuracy": 0.7402959535293206,
      "eval_loss": 4.089069843292236,
      "eval_runtime": 254.379,
      "eval_samples_per_second": 474.512,
      "eval_steps_per_second": 4.945,
      "step": 350000
    },
    {
      "epoch": 14.77469122792227,
      "grad_norm": 1.2224507331848145,
      "learning_rate": 4.160270977147022e-05,
      "loss": 4.3239,
      "step": 350500
    },
    {
      "epoch": 14.79576782025882,
      "grad_norm": 1.1992204189300537,
      "learning_rate": 4.1516957509754326e-05,
      "loss": 4.3276,
      "step": 351000
    },
    {
      "epoch": 14.81684441259537,
      "grad_norm": 1.218889594078064,
      "learning_rate": 4.143120524803842e-05,
      "loss": 4.3204,
      "step": 351500
    },
    {
      "epoch": 14.837921004931923,
      "grad_norm": 1.2430130243301392,
      "learning_rate": 4.1345452986322516e-05,
      "loss": 4.3233,
      "step": 352000
    },
    {
      "epoch": 14.858997597268473,
      "grad_norm": 1.2056000232696533,
      "learning_rate": 4.1259700724606614e-05,
      "loss": 4.3226,
      "step": 352500
    },
    {
      "epoch": 14.880074189605025,
      "grad_norm": 1.2264423370361328,
      "learning_rate": 4.117394846289071e-05,
      "loss": 4.3221,
      "step": 353000
    },
    {
      "epoch": 14.901150781941576,
      "grad_norm": 1.2099350690841675,
      "learning_rate": 4.1088196201174804e-05,
      "loss": 4.3257,
      "step": 353500
    },
    {
      "epoch": 14.922227374278126,
      "grad_norm": 1.2018463611602783,
      "learning_rate": 4.100244393945891e-05,
      "loss": 4.3212,
      "step": 354000
    },
    {
      "epoch": 14.943303966614678,
      "grad_norm": 1.2285740375518799,
      "learning_rate": 4.0916691677743e-05,
      "loss": 4.3277,
      "step": 354500
    },
    {
      "epoch": 14.964380558951229,
      "grad_norm": 1.1938539743423462,
      "learning_rate": 4.08309394160271e-05,
      "loss": 4.3238,
      "step": 355000
    },
    {
      "epoch": 14.985457151287779,
      "grad_norm": 1.190608263015747,
      "learning_rate": 4.07451871543112e-05,
      "loss": 4.3214,
      "step": 355500
    },
    {
      "epoch": 15.006533743624331,
      "grad_norm": 1.1888355016708374,
      "learning_rate": 4.0659434892595295e-05,
      "loss": 4.3179,
      "step": 356000
    },
    {
      "epoch": 15.027610335960881,
      "grad_norm": 1.1955257654190063,
      "learning_rate": 4.057368263087939e-05,
      "loss": 4.3115,
      "step": 356500
    },
    {
      "epoch": 15.048686928297434,
      "grad_norm": 1.1947569847106934,
      "learning_rate": 4.048793036916349e-05,
      "loss": 4.3176,
      "step": 357000
    },
    {
      "epoch": 15.069763520633984,
      "grad_norm": 1.2525466680526733,
      "learning_rate": 4.0402178107447584e-05,
      "loss": 4.3214,
      "step": 357500
    },
    {
      "epoch": 15.090840112970534,
      "grad_norm": 1.2551616430282593,
      "learning_rate": 4.031642584573168e-05,
      "loss": 4.3178,
      "step": 358000
    },
    {
      "epoch": 15.111916705307086,
      "grad_norm": 1.237062931060791,
      "learning_rate": 4.023067358401578e-05,
      "loss": 4.3173,
      "step": 358500
    },
    {
      "epoch": 15.132993297643637,
      "grad_norm": 1.2167061567306519,
      "learning_rate": 4.014492132229988e-05,
      "loss": 4.3179,
      "step": 359000
    },
    {
      "epoch": 15.154069889980189,
      "grad_norm": 1.1977704763412476,
      "learning_rate": 4.005916906058397e-05,
      "loss": 4.3161,
      "step": 359500
    },
    {
      "epoch": 15.17514648231674,
      "grad_norm": 1.1998213529586792,
      "learning_rate": 3.9973416798868075e-05,
      "loss": 4.3166,
      "step": 360000
    },
    {
      "epoch": 15.17514648231674,
      "eval_accuracy": 0.7409911198968023,
      "eval_loss": 4.0856523513793945,
      "eval_runtime": 254.443,
      "eval_samples_per_second": 474.393,
      "eval_steps_per_second": 4.944,
      "step": 360000
    },
    {
      "epoch": 15.19622307465329,
      "grad_norm": 1.2424161434173584,
      "learning_rate": 3.988766453715217e-05,
      "loss": 4.319,
      "step": 360500
    },
    {
      "epoch": 15.217299666989842,
      "grad_norm": 1.20900297164917,
      "learning_rate": 3.9801912275436265e-05,
      "loss": 4.3173,
      "step": 361000
    },
    {
      "epoch": 15.238376259326392,
      "grad_norm": 1.2076860666275024,
      "learning_rate": 3.971616001372036e-05,
      "loss": 4.3188,
      "step": 361500
    },
    {
      "epoch": 15.259452851662942,
      "grad_norm": 1.256195306777954,
      "learning_rate": 3.963040775200446e-05,
      "loss": 4.3159,
      "step": 362000
    },
    {
      "epoch": 15.280529443999495,
      "grad_norm": 1.2666856050491333,
      "learning_rate": 3.954465549028856e-05,
      "loss": 4.3116,
      "step": 362500
    },
    {
      "epoch": 15.301606036336045,
      "grad_norm": 1.221269130706787,
      "learning_rate": 3.945890322857266e-05,
      "loss": 4.3162,
      "step": 363000
    },
    {
      "epoch": 15.322682628672597,
      "grad_norm": 1.1990474462509155,
      "learning_rate": 3.9373150966856756e-05,
      "loss": 4.3128,
      "step": 363500
    },
    {
      "epoch": 15.343759221009147,
      "grad_norm": 1.2272567749023438,
      "learning_rate": 3.928739870514085e-05,
      "loss": 4.3142,
      "step": 364000
    },
    {
      "epoch": 15.364835813345698,
      "grad_norm": 1.2452703714370728,
      "learning_rate": 3.920164644342495e-05,
      "loss": 4.3149,
      "step": 364500
    },
    {
      "epoch": 15.38591240568225,
      "grad_norm": 1.2297862768173218,
      "learning_rate": 3.9115894181709045e-05,
      "loss": 4.3141,
      "step": 365000
    },
    {
      "epoch": 15.4069889980188,
      "grad_norm": 1.2087408304214478,
      "learning_rate": 3.903014191999314e-05,
      "loss": 4.3095,
      "step": 365500
    },
    {
      "epoch": 15.42806559035535,
      "grad_norm": 1.2771378755569458,
      "learning_rate": 3.894438965827724e-05,
      "loss": 4.3152,
      "step": 366000
    },
    {
      "epoch": 15.449142182691903,
      "grad_norm": 1.2283015251159668,
      "learning_rate": 3.885863739656134e-05,
      "loss": 4.3144,
      "step": 366500
    },
    {
      "epoch": 15.470218775028453,
      "grad_norm": 1.2135292291641235,
      "learning_rate": 3.877288513484543e-05,
      "loss": 4.3103,
      "step": 367000
    },
    {
      "epoch": 15.491295367365005,
      "grad_norm": 1.209444522857666,
      "learning_rate": 3.868713287312953e-05,
      "loss": 4.3133,
      "step": 367500
    },
    {
      "epoch": 15.512371959701555,
      "grad_norm": 1.2372729778289795,
      "learning_rate": 3.860138061141363e-05,
      "loss": 4.3118,
      "step": 368000
    },
    {
      "epoch": 15.533448552038106,
      "grad_norm": 1.2129958868026733,
      "learning_rate": 3.8515628349697726e-05,
      "loss": 4.314,
      "step": 368500
    },
    {
      "epoch": 15.554525144374658,
      "grad_norm": 1.2356773614883423,
      "learning_rate": 3.842987608798182e-05,
      "loss": 4.312,
      "step": 369000
    },
    {
      "epoch": 15.575601736711208,
      "grad_norm": 1.2493010759353638,
      "learning_rate": 3.834412382626592e-05,
      "loss": 4.3089,
      "step": 369500
    },
    {
      "epoch": 15.596678329047759,
      "grad_norm": 1.2475214004516602,
      "learning_rate": 3.8258371564550014e-05,
      "loss": 4.315,
      "step": 370000
    },
    {
      "epoch": 15.596678329047759,
      "eval_accuracy": 0.7420581365043087,
      "eval_loss": 4.075220108032227,
      "eval_runtime": 253.8593,
      "eval_samples_per_second": 475.484,
      "eval_steps_per_second": 4.956,
      "step": 370000
    },
    {
      "epoch": 15.61775492138431,
      "grad_norm": 1.2500431537628174,
      "learning_rate": 3.817261930283411e-05,
      "loss": 4.3105,
      "step": 370500
    },
    {
      "epoch": 15.638831513720861,
      "grad_norm": 1.2431563138961792,
      "learning_rate": 3.808686704111821e-05,
      "loss": 4.3109,
      "step": 371000
    },
    {
      "epoch": 15.659908106057413,
      "grad_norm": 1.2608145475387573,
      "learning_rate": 3.800111477940231e-05,
      "loss": 4.3118,
      "step": 371500
    },
    {
      "epoch": 15.680984698393964,
      "grad_norm": 1.2234452962875366,
      "learning_rate": 3.79153625176864e-05,
      "loss": 4.3105,
      "step": 372000
    },
    {
      "epoch": 15.702061290730514,
      "grad_norm": 1.2343026399612427,
      "learning_rate": 3.7829610255970506e-05,
      "loss": 4.3143,
      "step": 372500
    },
    {
      "epoch": 15.723137883067066,
      "grad_norm": 1.2104781866073608,
      "learning_rate": 3.77438579942546e-05,
      "loss": 4.3067,
      "step": 373000
    },
    {
      "epoch": 15.744214475403616,
      "grad_norm": 1.2287921905517578,
      "learning_rate": 3.7658105732538696e-05,
      "loss": 4.3046,
      "step": 373500
    },
    {
      "epoch": 15.765291067740169,
      "grad_norm": 1.1893996000289917,
      "learning_rate": 3.7572353470822794e-05,
      "loss": 4.3111,
      "step": 374000
    },
    {
      "epoch": 15.786367660076719,
      "grad_norm": 1.209667682647705,
      "learning_rate": 3.748660120910689e-05,
      "loss": 4.3069,
      "step": 374500
    },
    {
      "epoch": 15.80744425241327,
      "grad_norm": 1.2688055038452148,
      "learning_rate": 3.740084894739099e-05,
      "loss": 4.3036,
      "step": 375000
    },
    {
      "epoch": 15.828520844749821,
      "grad_norm": 1.2087920904159546,
      "learning_rate": 3.731509668567509e-05,
      "loss": 4.3064,
      "step": 375500
    },
    {
      "epoch": 15.849597437086372,
      "grad_norm": 1.1906718015670776,
      "learning_rate": 3.722934442395919e-05,
      "loss": 4.3094,
      "step": 376000
    },
    {
      "epoch": 15.870674029422922,
      "grad_norm": 1.1988335847854614,
      "learning_rate": 3.714359216224328e-05,
      "loss": 4.3084,
      "step": 376500
    },
    {
      "epoch": 15.891750621759474,
      "grad_norm": 1.219698190689087,
      "learning_rate": 3.705783990052738e-05,
      "loss": 4.3102,
      "step": 377000
    },
    {
      "epoch": 15.912827214096025,
      "grad_norm": 1.211488127708435,
      "learning_rate": 3.6972087638811475e-05,
      "loss": 4.3059,
      "step": 377500
    },
    {
      "epoch": 15.933903806432577,
      "grad_norm": 1.2031537294387817,
      "learning_rate": 3.6886335377095574e-05,
      "loss": 4.3075,
      "step": 378000
    },
    {
      "epoch": 15.954980398769127,
      "grad_norm": 1.1822209358215332,
      "learning_rate": 3.680058311537967e-05,
      "loss": 4.3061,
      "step": 378500
    },
    {
      "epoch": 15.976056991105677,
      "grad_norm": 1.2292230129241943,
      "learning_rate": 3.671483085366377e-05,
      "loss": 4.3088,
      "step": 379000
    },
    {
      "epoch": 15.99713358344223,
      "grad_norm": 1.2003979682922363,
      "learning_rate": 3.662907859194786e-05,
      "loss": 4.3025,
      "step": 379500
    },
    {
      "epoch": 16.01821017577878,
      "grad_norm": 1.2567667961120605,
      "learning_rate": 3.654332633023197e-05,
      "loss": 4.3041,
      "step": 380000
    },
    {
      "epoch": 16.01821017577878,
      "eval_accuracy": 0.743136153280742,
      "eval_loss": 4.070431232452393,
      "eval_runtime": 254.9724,
      "eval_samples_per_second": 473.408,
      "eval_steps_per_second": 4.934,
      "step": 380000
    },
    {
      "epoch": 16.039286768115332,
      "grad_norm": 1.2135577201843262,
      "learning_rate": 3.645757406851606e-05,
      "loss": 4.3009,
      "step": 380500
    },
    {
      "epoch": 16.06036336045188,
      "grad_norm": 1.2680631875991821,
      "learning_rate": 3.6371821806800157e-05,
      "loss": 4.3014,
      "step": 381000
    },
    {
      "epoch": 16.081439952788433,
      "grad_norm": 1.2161966562271118,
      "learning_rate": 3.6286069545084255e-05,
      "loss": 4.3067,
      "step": 381500
    },
    {
      "epoch": 16.102516545124985,
      "grad_norm": 1.227168321609497,
      "learning_rate": 3.620031728336835e-05,
      "loss": 4.3037,
      "step": 382000
    },
    {
      "epoch": 16.123593137461537,
      "grad_norm": 1.220467448234558,
      "learning_rate": 3.6114565021652445e-05,
      "loss": 4.2952,
      "step": 382500
    },
    {
      "epoch": 16.144669729798085,
      "grad_norm": 1.2277214527130127,
      "learning_rate": 3.602881275993655e-05,
      "loss": 4.3003,
      "step": 383000
    },
    {
      "epoch": 16.165746322134638,
      "grad_norm": 1.3101227283477783,
      "learning_rate": 3.594306049822064e-05,
      "loss": 4.3044,
      "step": 383500
    },
    {
      "epoch": 16.18682291447119,
      "grad_norm": 1.23399817943573,
      "learning_rate": 3.585730823650474e-05,
      "loss": 4.3009,
      "step": 384000
    },
    {
      "epoch": 16.20789950680774,
      "grad_norm": 1.21759033203125,
      "learning_rate": 3.577155597478883e-05,
      "loss": 4.3004,
      "step": 384500
    },
    {
      "epoch": 16.22897609914429,
      "grad_norm": 1.2085446119308472,
      "learning_rate": 3.5685803713072936e-05,
      "loss": 4.2969,
      "step": 385000
    },
    {
      "epoch": 16.250052691480843,
      "grad_norm": 1.220329999923706,
      "learning_rate": 3.560005145135703e-05,
      "loss": 4.2998,
      "step": 385500
    },
    {
      "epoch": 16.27112928381739,
      "grad_norm": 1.2465665340423584,
      "learning_rate": 3.5514299189641126e-05,
      "loss": 4.2979,
      "step": 386000
    },
    {
      "epoch": 16.292205876153943,
      "grad_norm": 1.2478259801864624,
      "learning_rate": 3.5428546927925224e-05,
      "loss": 4.3033,
      "step": 386500
    },
    {
      "epoch": 16.313282468490495,
      "grad_norm": 1.2247140407562256,
      "learning_rate": 3.534279466620932e-05,
      "loss": 4.2969,
      "step": 387000
    },
    {
      "epoch": 16.334359060827044,
      "grad_norm": 1.2085820436477661,
      "learning_rate": 3.525704240449342e-05,
      "loss": 4.2974,
      "step": 387500
    },
    {
      "epoch": 16.355435653163596,
      "grad_norm": 1.21865975856781,
      "learning_rate": 3.517129014277752e-05,
      "loss": 4.2965,
      "step": 388000
    },
    {
      "epoch": 16.376512245500148,
      "grad_norm": 1.1987487077713013,
      "learning_rate": 3.508553788106161e-05,
      "loss": 4.2973,
      "step": 388500
    },
    {
      "epoch": 16.3975888378367,
      "grad_norm": 1.2421199083328247,
      "learning_rate": 3.499978561934571e-05,
      "loss": 4.2956,
      "step": 389000
    },
    {
      "epoch": 16.41866543017325,
      "grad_norm": 1.1867400407791138,
      "learning_rate": 3.491403335762981e-05,
      "loss": 4.298,
      "step": 389500
    },
    {
      "epoch": 16.4397420225098,
      "grad_norm": 1.2186658382415771,
      "learning_rate": 3.4828281095913906e-05,
      "loss": 4.2986,
      "step": 390000
    },
    {
      "epoch": 16.4397420225098,
      "eval_accuracy": 0.7439678887414938,
      "eval_loss": 4.064914226531982,
      "eval_runtime": 255.1513,
      "eval_samples_per_second": 473.076,
      "eval_steps_per_second": 4.93,
      "step": 390000
    },
    {
      "epoch": 16.460818614846353,
      "grad_norm": 1.2529385089874268,
      "learning_rate": 3.4742528834198004e-05,
      "loss": 4.2953,
      "step": 390500
    },
    {
      "epoch": 16.4818952071829,
      "grad_norm": 1.2005215883255005,
      "learning_rate": 3.46567765724821e-05,
      "loss": 4.2992,
      "step": 391000
    },
    {
      "epoch": 16.502971799519454,
      "grad_norm": 1.2254198789596558,
      "learning_rate": 3.45710243107662e-05,
      "loss": 4.297,
      "step": 391500
    },
    {
      "epoch": 16.524048391856006,
      "grad_norm": 1.224564552307129,
      "learning_rate": 3.448527204905029e-05,
      "loss": 4.2999,
      "step": 392000
    },
    {
      "epoch": 16.545124984192555,
      "grad_norm": 1.221266746520996,
      "learning_rate": 3.43995197873344e-05,
      "loss": 4.2941,
      "step": 392500
    },
    {
      "epoch": 16.566201576529107,
      "grad_norm": 1.2223858833312988,
      "learning_rate": 3.431376752561849e-05,
      "loss": 4.2951,
      "step": 393000
    },
    {
      "epoch": 16.58727816886566,
      "grad_norm": 1.1899433135986328,
      "learning_rate": 3.422801526390259e-05,
      "loss": 4.2992,
      "step": 393500
    },
    {
      "epoch": 16.608354761202207,
      "grad_norm": 1.202358603477478,
      "learning_rate": 3.4142263002186686e-05,
      "loss": 4.2963,
      "step": 394000
    },
    {
      "epoch": 16.62943135353876,
      "grad_norm": 1.2531260251998901,
      "learning_rate": 3.4056510740470784e-05,
      "loss": 4.2996,
      "step": 394500
    },
    {
      "epoch": 16.65050794587531,
      "grad_norm": 1.228501558303833,
      "learning_rate": 3.3970758478754875e-05,
      "loss": 4.2979,
      "step": 395000
    },
    {
      "epoch": 16.67158453821186,
      "grad_norm": 1.220013976097107,
      "learning_rate": 3.388500621703898e-05,
      "loss": 4.2978,
      "step": 395500
    },
    {
      "epoch": 16.692661130548412,
      "grad_norm": 1.2601642608642578,
      "learning_rate": 3.379925395532307e-05,
      "loss": 4.2983,
      "step": 396000
    },
    {
      "epoch": 16.713737722884964,
      "grad_norm": 1.1838117837905884,
      "learning_rate": 3.371350169360717e-05,
      "loss": 4.2934,
      "step": 396500
    },
    {
      "epoch": 16.734814315221517,
      "grad_norm": 1.2267428636550903,
      "learning_rate": 3.362774943189127e-05,
      "loss": 4.2963,
      "step": 397000
    },
    {
      "epoch": 16.755890907558065,
      "grad_norm": 1.2209008932113647,
      "learning_rate": 3.354199717017537e-05,
      "loss": 4.2946,
      "step": 397500
    },
    {
      "epoch": 16.776967499894617,
      "grad_norm": 1.2603098154067993,
      "learning_rate": 3.345624490845946e-05,
      "loss": 4.2935,
      "step": 398000
    },
    {
      "epoch": 16.79804409223117,
      "grad_norm": 1.2326359748840332,
      "learning_rate": 3.3370492646743564e-05,
      "loss": 4.2926,
      "step": 398500
    },
    {
      "epoch": 16.819120684567718,
      "grad_norm": 1.2161887884140015,
      "learning_rate": 3.3284740385027655e-05,
      "loss": 4.2984,
      "step": 399000
    },
    {
      "epoch": 16.84019727690427,
      "grad_norm": 1.2129220962524414,
      "learning_rate": 3.319898812331175e-05,
      "loss": 4.2901,
      "step": 399500
    },
    {
      "epoch": 16.861273869240822,
      "grad_norm": 1.2572506666183472,
      "learning_rate": 3.311323586159585e-05,
      "loss": 4.293,
      "step": 400000
    },
    {
      "epoch": 16.861273869240822,
      "eval_accuracy": 0.7445877259208825,
      "eval_loss": 4.062014579772949,
      "eval_runtime": 254.7634,
      "eval_samples_per_second": 473.796,
      "eval_steps_per_second": 4.938,
      "step": 400000
    },
    {
      "epoch": 16.88235046157737,
      "grad_norm": 1.227662444114685,
      "learning_rate": 3.302748359987995e-05,
      "loss": 4.2885,
      "step": 400500
    },
    {
      "epoch": 16.903427053913923,
      "grad_norm": 1.2272685766220093,
      "learning_rate": 3.294173133816404e-05,
      "loss": 4.2937,
      "step": 401000
    },
    {
      "epoch": 16.924503646250475,
      "grad_norm": 1.2672008275985718,
      "learning_rate": 3.2855979076448147e-05,
      "loss": 4.2962,
      "step": 401500
    },
    {
      "epoch": 16.945580238587024,
      "grad_norm": 1.2498761415481567,
      "learning_rate": 3.277022681473224e-05,
      "loss": 4.2922,
      "step": 402000
    },
    {
      "epoch": 16.966656830923576,
      "grad_norm": 1.2548388242721558,
      "learning_rate": 3.2684474553016336e-05,
      "loss": 4.2933,
      "step": 402500
    },
    {
      "epoch": 16.987733423260128,
      "grad_norm": 1.2178502082824707,
      "learning_rate": 3.2598722291300435e-05,
      "loss": 4.2954,
      "step": 403000
    },
    {
      "epoch": 17.00881001559668,
      "grad_norm": 1.2302992343902588,
      "learning_rate": 3.251297002958453e-05,
      "loss": 4.2905,
      "step": 403500
    },
    {
      "epoch": 17.02988660793323,
      "grad_norm": 1.249819278717041,
      "learning_rate": 3.242721776786863e-05,
      "loss": 4.2901,
      "step": 404000
    },
    {
      "epoch": 17.05096320026978,
      "grad_norm": 1.2457138299942017,
      "learning_rate": 3.234146550615272e-05,
      "loss": 4.2847,
      "step": 404500
    },
    {
      "epoch": 17.072039792606333,
      "grad_norm": 1.232865333557129,
      "learning_rate": 3.225571324443683e-05,
      "loss": 4.2892,
      "step": 405000
    },
    {
      "epoch": 17.09311638494288,
      "grad_norm": 1.2603700160980225,
      "learning_rate": 3.216996098272092e-05,
      "loss": 4.2891,
      "step": 405500
    },
    {
      "epoch": 17.114192977279433,
      "grad_norm": 1.2291202545166016,
      "learning_rate": 3.208420872100502e-05,
      "loss": 4.2868,
      "step": 406000
    },
    {
      "epoch": 17.135269569615986,
      "grad_norm": 1.2202527523040771,
      "learning_rate": 3.1998456459289116e-05,
      "loss": 4.2849,
      "step": 406500
    },
    {
      "epoch": 17.156346161952534,
      "grad_norm": 1.234372854232788,
      "learning_rate": 3.1912704197573214e-05,
      "loss": 4.2858,
      "step": 407000
    },
    {
      "epoch": 17.177422754289086,
      "grad_norm": 1.3185986280441284,
      "learning_rate": 3.1826951935857306e-05,
      "loss": 4.2829,
      "step": 407500
    },
    {
      "epoch": 17.19849934662564,
      "grad_norm": 1.2391513586044312,
      "learning_rate": 3.174119967414141e-05,
      "loss": 4.289,
      "step": 408000
    },
    {
      "epoch": 17.219575938962187,
      "grad_norm": 1.2678101062774658,
      "learning_rate": 3.16554474124255e-05,
      "loss": 4.2867,
      "step": 408500
    },
    {
      "epoch": 17.24065253129874,
      "grad_norm": 1.24124276638031,
      "learning_rate": 3.15696951507096e-05,
      "loss": 4.2881,
      "step": 409000
    },
    {
      "epoch": 17.26172912363529,
      "grad_norm": 1.2661693096160889,
      "learning_rate": 3.14839428889937e-05,
      "loss": 4.2864,
      "step": 409500
    },
    {
      "epoch": 17.282805715971843,
      "grad_norm": 1.2680538892745972,
      "learning_rate": 3.13981906272778e-05,
      "loss": 4.2881,
      "step": 410000
    },
    {
      "epoch": 17.282805715971843,
      "eval_accuracy": 0.7456742294499613,
      "eval_loss": 4.053155899047852,
      "eval_runtime": 254.8765,
      "eval_samples_per_second": 473.586,
      "eval_steps_per_second": 4.936,
      "step": 410000
    },
    {
      "epoch": 17.303882308308392,
      "grad_norm": 1.2803905010223389,
      "learning_rate": 3.131243836556189e-05,
      "loss": 4.2908,
      "step": 410500
    },
    {
      "epoch": 17.324958900644944,
      "grad_norm": 1.2419826984405518,
      "learning_rate": 3.1226686103845994e-05,
      "loss": 4.2824,
      "step": 411000
    },
    {
      "epoch": 17.346035492981496,
      "grad_norm": 1.2859811782836914,
      "learning_rate": 3.1140933842130086e-05,
      "loss": 4.2849,
      "step": 411500
    },
    {
      "epoch": 17.367112085318045,
      "grad_norm": 1.25287926197052,
      "learning_rate": 3.1055181580414184e-05,
      "loss": 4.2872,
      "step": 412000
    },
    {
      "epoch": 17.388188677654597,
      "grad_norm": 1.2089923620224,
      "learning_rate": 3.096942931869828e-05,
      "loss": 4.2901,
      "step": 412500
    },
    {
      "epoch": 17.40926526999115,
      "grad_norm": 1.224830150604248,
      "learning_rate": 3.088367705698238e-05,
      "loss": 4.2814,
      "step": 413000
    },
    {
      "epoch": 17.430341862327698,
      "grad_norm": 1.2108339071273804,
      "learning_rate": 3.079792479526647e-05,
      "loss": 4.2863,
      "step": 413500
    },
    {
      "epoch": 17.45141845466425,
      "grad_norm": 1.2245113849639893,
      "learning_rate": 3.071217253355058e-05,
      "loss": 4.2836,
      "step": 414000
    },
    {
      "epoch": 17.472495047000802,
      "grad_norm": 1.2422621250152588,
      "learning_rate": 3.062642027183467e-05,
      "loss": 4.2829,
      "step": 414500
    },
    {
      "epoch": 17.49357163933735,
      "grad_norm": 1.2413445711135864,
      "learning_rate": 3.054066801011877e-05,
      "loss": 4.2839,
      "step": 415000
    },
    {
      "epoch": 17.514648231673903,
      "grad_norm": 1.276723861694336,
      "learning_rate": 3.045491574840287e-05,
      "loss": 4.286,
      "step": 415500
    },
    {
      "epoch": 17.535724824010455,
      "grad_norm": 1.255544900894165,
      "learning_rate": 3.0369163486686964e-05,
      "loss": 4.2824,
      "step": 416000
    },
    {
      "epoch": 17.556801416347007,
      "grad_norm": 1.2709625959396362,
      "learning_rate": 3.028341122497106e-05,
      "loss": 4.2813,
      "step": 416500
    },
    {
      "epoch": 17.577878008683555,
      "grad_norm": 1.2757691144943237,
      "learning_rate": 3.019765896325516e-05,
      "loss": 4.2866,
      "step": 417000
    },
    {
      "epoch": 17.598954601020107,
      "grad_norm": 1.2571974992752075,
      "learning_rate": 3.0111906701539255e-05,
      "loss": 4.2863,
      "step": 417500
    },
    {
      "epoch": 17.62003119335666,
      "grad_norm": 1.2620062828063965,
      "learning_rate": 3.002615443982335e-05,
      "loss": 4.2813,
      "step": 418000
    },
    {
      "epoch": 17.641107785693208,
      "grad_norm": 1.2279225587844849,
      "learning_rate": 2.9940402178107452e-05,
      "loss": 4.281,
      "step": 418500
    },
    {
      "epoch": 17.66218437802976,
      "grad_norm": 1.2517921924591064,
      "learning_rate": 2.9854649916391547e-05,
      "loss": 4.2811,
      "step": 419000
    },
    {
      "epoch": 17.683260970366312,
      "grad_norm": 1.290460467338562,
      "learning_rate": 2.976889765467564e-05,
      "loss": 4.2857,
      "step": 419500
    },
    {
      "epoch": 17.70433756270286,
      "grad_norm": 1.253234624862671,
      "learning_rate": 2.9683145392959737e-05,
      "loss": 4.282,
      "step": 420000
    },
    {
      "epoch": 17.70433756270286,
      "eval_accuracy": 0.7464532199681013,
      "eval_loss": 4.050754547119141,
      "eval_runtime": 253.8713,
      "eval_samples_per_second": 475.461,
      "eval_steps_per_second": 4.955,
      "step": 420000
    },
    {
      "epoch": 17.725414155039413,
      "grad_norm": 1.3073203563690186,
      "learning_rate": 2.9597393131243838e-05,
      "loss": 4.2825,
      "step": 420500
    },
    {
      "epoch": 17.746490747375965,
      "grad_norm": 1.2296487092971802,
      "learning_rate": 2.9511640869527933e-05,
      "loss": 4.2795,
      "step": 421000
    },
    {
      "epoch": 17.767567339712514,
      "grad_norm": 1.2437549829483032,
      "learning_rate": 2.942588860781203e-05,
      "loss": 4.2791,
      "step": 421500
    },
    {
      "epoch": 17.788643932049066,
      "grad_norm": 1.2276737689971924,
      "learning_rate": 2.934013634609613e-05,
      "loss": 4.2807,
      "step": 422000
    },
    {
      "epoch": 17.809720524385618,
      "grad_norm": 1.2491365671157837,
      "learning_rate": 2.9254384084380225e-05,
      "loss": 4.2812,
      "step": 422500
    },
    {
      "epoch": 17.830797116722167,
      "grad_norm": 1.2725005149841309,
      "learning_rate": 2.9168631822664323e-05,
      "loss": 4.2816,
      "step": 423000
    },
    {
      "epoch": 17.85187370905872,
      "grad_norm": 1.227061152458191,
      "learning_rate": 2.908287956094842e-05,
      "loss": 4.2795,
      "step": 423500
    },
    {
      "epoch": 17.87295030139527,
      "grad_norm": 1.2260409593582153,
      "learning_rate": 2.899712729923252e-05,
      "loss": 4.2816,
      "step": 424000
    },
    {
      "epoch": 17.894026893731823,
      "grad_norm": 1.2494770288467407,
      "learning_rate": 2.8911375037516615e-05,
      "loss": 4.2793,
      "step": 424500
    },
    {
      "epoch": 17.91510348606837,
      "grad_norm": 1.2261497974395752,
      "learning_rate": 2.8825622775800716e-05,
      "loss": 4.2823,
      "step": 425000
    },
    {
      "epoch": 17.936180078404924,
      "grad_norm": 1.2670890092849731,
      "learning_rate": 2.873987051408481e-05,
      "loss": 4.2791,
      "step": 425500
    },
    {
      "epoch": 17.957256670741476,
      "grad_norm": 1.2530848979949951,
      "learning_rate": 2.8654118252368906e-05,
      "loss": 4.2788,
      "step": 426000
    },
    {
      "epoch": 17.978333263078024,
      "grad_norm": 1.294309377670288,
      "learning_rate": 2.8568365990653008e-05,
      "loss": 4.2773,
      "step": 426500
    },
    {
      "epoch": 17.999409855414576,
      "grad_norm": 1.2321404218673706,
      "learning_rate": 2.8482613728937103e-05,
      "loss": 4.2815,
      "step": 427000
    },
    {
      "epoch": 18.02048644775113,
      "grad_norm": 1.231532335281372,
      "learning_rate": 2.8396861467221198e-05,
      "loss": 4.2758,
      "step": 427500
    },
    {
      "epoch": 18.041563040087677,
      "grad_norm": 1.2730671167373657,
      "learning_rate": 2.83111092055053e-05,
      "loss": 4.2746,
      "step": 428000
    },
    {
      "epoch": 18.06263963242423,
      "grad_norm": 1.2523207664489746,
      "learning_rate": 2.8225356943789394e-05,
      "loss": 4.2763,
      "step": 428500
    },
    {
      "epoch": 18.08371622476078,
      "grad_norm": 1.2297675609588623,
      "learning_rate": 2.813960468207349e-05,
      "loss": 4.2739,
      "step": 429000
    },
    {
      "epoch": 18.10479281709733,
      "grad_norm": 1.2207614183425903,
      "learning_rate": 2.805385242035759e-05,
      "loss": 4.2728,
      "step": 429500
    },
    {
      "epoch": 18.125869409433882,
      "grad_norm": 1.2939807176589966,
      "learning_rate": 2.7968100158641686e-05,
      "loss": 4.2738,
      "step": 430000
    },
    {
      "epoch": 18.125869409433882,
      "eval_accuracy": 0.7470560347268003,
      "eval_loss": 4.046945571899414,
      "eval_runtime": 254.7942,
      "eval_samples_per_second": 473.739,
      "eval_steps_per_second": 4.937,
      "step": 430000
    },
    {
      "epoch": 18.146946001770434,
      "grad_norm": 1.269366979598999,
      "learning_rate": 2.788234789692578e-05,
      "loss": 4.2735,
      "step": 430500
    },
    {
      "epoch": 18.168022594106986,
      "grad_norm": 1.2754448652267456,
      "learning_rate": 2.7796595635209882e-05,
      "loss": 4.2747,
      "step": 431000
    },
    {
      "epoch": 18.189099186443535,
      "grad_norm": 1.2659934759140015,
      "learning_rate": 2.7710843373493977e-05,
      "loss": 4.2776,
      "step": 431500
    },
    {
      "epoch": 18.210175778780087,
      "grad_norm": 1.2298548221588135,
      "learning_rate": 2.7625091111778072e-05,
      "loss": 4.2768,
      "step": 432000
    },
    {
      "epoch": 18.23125237111664,
      "grad_norm": 1.2684962749481201,
      "learning_rate": 2.7539338850062174e-05,
      "loss": 4.2727,
      "step": 432500
    },
    {
      "epoch": 18.252328963453188,
      "grad_norm": 1.2355148792266846,
      "learning_rate": 2.745358658834627e-05,
      "loss": 4.2759,
      "step": 433000
    },
    {
      "epoch": 18.27340555578974,
      "grad_norm": 1.2494068145751953,
      "learning_rate": 2.7367834326630364e-05,
      "loss": 4.2719,
      "step": 433500
    },
    {
      "epoch": 18.294482148126292,
      "grad_norm": 1.2445040941238403,
      "learning_rate": 2.7282082064914465e-05,
      "loss": 4.2741,
      "step": 434000
    },
    {
      "epoch": 18.31555874046284,
      "grad_norm": 1.2553945779800415,
      "learning_rate": 2.719632980319856e-05,
      "loss": 4.2721,
      "step": 434500
    },
    {
      "epoch": 18.336635332799393,
      "grad_norm": 1.2074700593948364,
      "learning_rate": 2.7110577541482655e-05,
      "loss": 4.2707,
      "step": 435000
    },
    {
      "epoch": 18.357711925135945,
      "grad_norm": 1.2521814107894897,
      "learning_rate": 2.7024825279766757e-05,
      "loss": 4.2752,
      "step": 435500
    },
    {
      "epoch": 18.378788517472493,
      "grad_norm": 1.2440481185913086,
      "learning_rate": 2.6939073018050852e-05,
      "loss": 4.2741,
      "step": 436000
    },
    {
      "epoch": 18.399865109809046,
      "grad_norm": 1.2552986145019531,
      "learning_rate": 2.685332075633495e-05,
      "loss": 4.2725,
      "step": 436500
    },
    {
      "epoch": 18.420941702145598,
      "grad_norm": 1.2366400957107544,
      "learning_rate": 2.6767568494619045e-05,
      "loss": 4.2732,
      "step": 437000
    },
    {
      "epoch": 18.44201829448215,
      "grad_norm": 1.272694706916809,
      "learning_rate": 2.6681816232903143e-05,
      "loss": 4.2715,
      "step": 437500
    },
    {
      "epoch": 18.4630948868187,
      "grad_norm": 1.2905014753341675,
      "learning_rate": 2.6596063971187242e-05,
      "loss": 4.2707,
      "step": 438000
    },
    {
      "epoch": 18.48417147915525,
      "grad_norm": 1.309250831604004,
      "learning_rate": 2.6510311709471337e-05,
      "loss": 4.2704,
      "step": 438500
    },
    {
      "epoch": 18.505248071491803,
      "grad_norm": 1.2340824604034424,
      "learning_rate": 2.642455944775544e-05,
      "loss": 4.2713,
      "step": 439000
    },
    {
      "epoch": 18.52632466382835,
      "grad_norm": 1.2254282236099243,
      "learning_rate": 2.6338807186039533e-05,
      "loss": 4.27,
      "step": 439500
    },
    {
      "epoch": 18.547401256164903,
      "grad_norm": 1.2640701532363892,
      "learning_rate": 2.6253054924323628e-05,
      "loss": 4.2676,
      "step": 440000
    },
    {
      "epoch": 18.547401256164903,
      "eval_accuracy": 0.747554139069791,
      "eval_loss": 4.042923927307129,
      "eval_runtime": 254.9336,
      "eval_samples_per_second": 473.48,
      "eval_steps_per_second": 4.935,
      "step": 440000
    },
    {
      "epoch": 18.568477848501455,
      "grad_norm": 1.2646886110305786,
      "learning_rate": 2.616730266260773e-05,
      "loss": 4.2683,
      "step": 440500
    },
    {
      "epoch": 18.589554440838004,
      "grad_norm": 1.221270203590393,
      "learning_rate": 2.6081550400891825e-05,
      "loss": 4.2704,
      "step": 441000
    },
    {
      "epoch": 18.610631033174556,
      "grad_norm": 1.257999300956726,
      "learning_rate": 2.599579813917592e-05,
      "loss": 4.2701,
      "step": 441500
    },
    {
      "epoch": 18.63170762551111,
      "grad_norm": 1.2518991231918335,
      "learning_rate": 2.591004587746002e-05,
      "loss": 4.2692,
      "step": 442000
    },
    {
      "epoch": 18.652784217847657,
      "grad_norm": 1.288578748703003,
      "learning_rate": 2.5824293615744116e-05,
      "loss": 4.2705,
      "step": 442500
    },
    {
      "epoch": 18.67386081018421,
      "grad_norm": 1.276168942451477,
      "learning_rate": 2.573854135402821e-05,
      "loss": 4.2713,
      "step": 443000
    },
    {
      "epoch": 18.69493740252076,
      "grad_norm": 1.2504286766052246,
      "learning_rate": 2.5652789092312313e-05,
      "loss": 4.2692,
      "step": 443500
    },
    {
      "epoch": 18.716013994857313,
      "grad_norm": 1.2744758129119873,
      "learning_rate": 2.5567036830596408e-05,
      "loss": 4.2693,
      "step": 444000
    },
    {
      "epoch": 18.737090587193862,
      "grad_norm": 1.2605938911437988,
      "learning_rate": 2.5481284568880503e-05,
      "loss": 4.2707,
      "step": 444500
    },
    {
      "epoch": 18.758167179530414,
      "grad_norm": 1.2607053518295288,
      "learning_rate": 2.5395532307164605e-05,
      "loss": 4.2655,
      "step": 445000
    },
    {
      "epoch": 18.779243771866966,
      "grad_norm": 1.5370845794677734,
      "learning_rate": 2.53097800454487e-05,
      "loss": 4.2718,
      "step": 445500
    },
    {
      "epoch": 18.800320364203515,
      "grad_norm": 1.2638561725616455,
      "learning_rate": 2.5224027783732794e-05,
      "loss": 4.2711,
      "step": 446000
    },
    {
      "epoch": 18.821396956540067,
      "grad_norm": 1.3151655197143555,
      "learning_rate": 2.5138275522016896e-05,
      "loss": 4.262,
      "step": 446500
    },
    {
      "epoch": 18.84247354887662,
      "grad_norm": 1.256486415863037,
      "learning_rate": 2.505252326030099e-05,
      "loss": 4.2646,
      "step": 447000
    },
    {
      "epoch": 18.863550141213167,
      "grad_norm": 1.261559247970581,
      "learning_rate": 2.496677099858509e-05,
      "loss": 4.2665,
      "step": 447500
    },
    {
      "epoch": 18.88462673354972,
      "grad_norm": 1.2537294626235962,
      "learning_rate": 2.4881018736869184e-05,
      "loss": 4.2696,
      "step": 448000
    },
    {
      "epoch": 18.90570332588627,
      "grad_norm": 1.2870759963989258,
      "learning_rate": 2.4795266475153283e-05,
      "loss": 4.266,
      "step": 448500
    },
    {
      "epoch": 18.92677991822282,
      "grad_norm": 1.2642253637313843,
      "learning_rate": 2.470951421343738e-05,
      "loss": 4.2714,
      "step": 449000
    },
    {
      "epoch": 18.947856510559372,
      "grad_norm": 1.2575712203979492,
      "learning_rate": 2.4623761951721476e-05,
      "loss": 4.2665,
      "step": 449500
    },
    {
      "epoch": 18.968933102895924,
      "grad_norm": 1.2555410861968994,
      "learning_rate": 2.4538009690005574e-05,
      "loss": 4.2666,
      "step": 450000
    },
    {
      "epoch": 18.968933102895924,
      "eval_accuracy": 0.7484695242394203,
      "eval_loss": 4.036364555358887,
      "eval_runtime": 256.4019,
      "eval_samples_per_second": 470.769,
      "eval_steps_per_second": 4.906,
      "step": 450000
    },
    {
      "epoch": 18.990009695232473,
      "grad_norm": 1.3054544925689697,
      "learning_rate": 2.4452257428289672e-05,
      "loss": 4.2628,
      "step": 450500
    },
    {
      "epoch": 19.011086287569025,
      "grad_norm": 1.2855356931686401,
      "learning_rate": 2.436650516657377e-05,
      "loss": 4.2641,
      "step": 451000
    },
    {
      "epoch": 19.032162879905577,
      "grad_norm": 1.2780327796936035,
      "learning_rate": 2.428075290485787e-05,
      "loss": 4.2626,
      "step": 451500
    },
    {
      "epoch": 19.05323947224213,
      "grad_norm": 1.2521767616271973,
      "learning_rate": 2.4195000643141964e-05,
      "loss": 4.2613,
      "step": 452000
    },
    {
      "epoch": 19.074316064578678,
      "grad_norm": 1.2825371026992798,
      "learning_rate": 2.4109248381426062e-05,
      "loss": 4.267,
      "step": 452500
    },
    {
      "epoch": 19.09539265691523,
      "grad_norm": 1.2793084383010864,
      "learning_rate": 2.402349611971016e-05,
      "loss": 4.2632,
      "step": 453000
    },
    {
      "epoch": 19.116469249251782,
      "grad_norm": 1.2280391454696655,
      "learning_rate": 2.3937743857994255e-05,
      "loss": 4.2594,
      "step": 453500
    },
    {
      "epoch": 19.13754584158833,
      "grad_norm": 1.3084862232208252,
      "learning_rate": 2.3851991596278354e-05,
      "loss": 4.2634,
      "step": 454000
    },
    {
      "epoch": 19.158622433924883,
      "grad_norm": 1.2982312440872192,
      "learning_rate": 2.3766239334562452e-05,
      "loss": 4.2621,
      "step": 454500
    },
    {
      "epoch": 19.179699026261435,
      "grad_norm": 1.280814528465271,
      "learning_rate": 2.3680487072846547e-05,
      "loss": 4.2587,
      "step": 455000
    },
    {
      "epoch": 19.200775618597984,
      "grad_norm": 1.2891268730163574,
      "learning_rate": 2.3594734811130645e-05,
      "loss": 4.2587,
      "step": 455500
    },
    {
      "epoch": 19.221852210934536,
      "grad_norm": 1.2391064167022705,
      "learning_rate": 2.3508982549414744e-05,
      "loss": 4.2593,
      "step": 456000
    },
    {
      "epoch": 19.242928803271088,
      "grad_norm": 1.2680084705352783,
      "learning_rate": 2.342323028769884e-05,
      "loss": 4.2628,
      "step": 456500
    },
    {
      "epoch": 19.264005395607636,
      "grad_norm": 1.2467697858810425,
      "learning_rate": 2.3337478025982937e-05,
      "loss": 4.2585,
      "step": 457000
    },
    {
      "epoch": 19.28508198794419,
      "grad_norm": 1.315354347229004,
      "learning_rate": 2.3251725764267035e-05,
      "loss": 4.2614,
      "step": 457500
    },
    {
      "epoch": 19.30615858028074,
      "grad_norm": 1.2915756702423096,
      "learning_rate": 2.316597350255113e-05,
      "loss": 4.2642,
      "step": 458000
    },
    {
      "epoch": 19.327235172617293,
      "grad_norm": 1.2535816431045532,
      "learning_rate": 2.308022124083523e-05,
      "loss": 4.2615,
      "step": 458500
    },
    {
      "epoch": 19.34831176495384,
      "grad_norm": 1.2980530261993408,
      "learning_rate": 2.2994468979119323e-05,
      "loss": 4.2634,
      "step": 459000
    },
    {
      "epoch": 19.369388357290394,
      "grad_norm": 1.303011178970337,
      "learning_rate": 2.290871671740342e-05,
      "loss": 4.2607,
      "step": 459500
    },
    {
      "epoch": 19.390464949626946,
      "grad_norm": 1.2506895065307617,
      "learning_rate": 2.282296445568752e-05,
      "loss": 4.2598,
      "step": 460000
    },
    {
      "epoch": 19.390464949626946,
      "eval_accuracy": 0.7493241066232046,
      "eval_loss": 4.031537055969238,
      "eval_runtime": 255.9208,
      "eval_samples_per_second": 471.654,
      "eval_steps_per_second": 4.916,
      "step": 460000
    },
    {
      "epoch": 19.411541541963494,
      "grad_norm": 1.2767223119735718,
      "learning_rate": 2.2737212193971615e-05,
      "loss": 4.2583,
      "step": 460500
    },
    {
      "epoch": 19.432618134300046,
      "grad_norm": 1.2857931852340698,
      "learning_rate": 2.2651459932255713e-05,
      "loss": 4.2642,
      "step": 461000
    },
    {
      "epoch": 19.4536947266366,
      "grad_norm": 1.255143165588379,
      "learning_rate": 2.256570767053981e-05,
      "loss": 4.2582,
      "step": 461500
    },
    {
      "epoch": 19.474771318973147,
      "grad_norm": 1.2223973274230957,
      "learning_rate": 2.2479955408823906e-05,
      "loss": 4.2569,
      "step": 462000
    },
    {
      "epoch": 19.4958479113097,
      "grad_norm": 1.2606490850448608,
      "learning_rate": 2.2394203147108005e-05,
      "loss": 4.2593,
      "step": 462500
    },
    {
      "epoch": 19.51692450364625,
      "grad_norm": 1.2714134454727173,
      "learning_rate": 2.2308450885392103e-05,
      "loss": 4.2562,
      "step": 463000
    },
    {
      "epoch": 19.5380010959828,
      "grad_norm": 1.2749547958374023,
      "learning_rate": 2.22226986236762e-05,
      "loss": 4.2562,
      "step": 463500
    },
    {
      "epoch": 19.559077688319352,
      "grad_norm": 1.2344410419464111,
      "learning_rate": 2.21369463619603e-05,
      "loss": 4.2592,
      "step": 464000
    },
    {
      "epoch": 19.580154280655904,
      "grad_norm": 1.3650463819503784,
      "learning_rate": 2.2051194100244394e-05,
      "loss": 4.2598,
      "step": 464500
    },
    {
      "epoch": 19.601230872992456,
      "grad_norm": 1.3392360210418701,
      "learning_rate": 2.1965441838528493e-05,
      "loss": 4.2586,
      "step": 465000
    },
    {
      "epoch": 19.622307465329005,
      "grad_norm": 1.2725521326065063,
      "learning_rate": 2.187968957681259e-05,
      "loss": 4.2575,
      "step": 465500
    },
    {
      "epoch": 19.643384057665557,
      "grad_norm": 1.235400915145874,
      "learning_rate": 2.179393731509669e-05,
      "loss": 4.2605,
      "step": 466000
    },
    {
      "epoch": 19.66446065000211,
      "grad_norm": 1.291383147239685,
      "learning_rate": 2.1708185053380784e-05,
      "loss": 4.2561,
      "step": 466500
    },
    {
      "epoch": 19.685537242338658,
      "grad_norm": 1.283194661140442,
      "learning_rate": 2.1622432791664883e-05,
      "loss": 4.2615,
      "step": 467000
    },
    {
      "epoch": 19.70661383467521,
      "grad_norm": 1.2486878633499146,
      "learning_rate": 2.1536680529948978e-05,
      "loss": 4.2576,
      "step": 467500
    },
    {
      "epoch": 19.727690427011762,
      "grad_norm": 1.2678519487380981,
      "learning_rate": 2.1450928268233076e-05,
      "loss": 4.2578,
      "step": 468000
    },
    {
      "epoch": 19.74876701934831,
      "grad_norm": 1.2635868787765503,
      "learning_rate": 2.1365176006517174e-05,
      "loss": 4.2565,
      "step": 468500
    },
    {
      "epoch": 19.769843611684863,
      "grad_norm": 1.3035540580749512,
      "learning_rate": 2.127942374480127e-05,
      "loss": 4.2562,
      "step": 469000
    },
    {
      "epoch": 19.790920204021415,
      "grad_norm": 1.2717002630233765,
      "learning_rate": 2.1193671483085367e-05,
      "loss": 4.2602,
      "step": 469500
    },
    {
      "epoch": 19.811996796357963,
      "grad_norm": 1.3174604177474976,
      "learning_rate": 2.1107919221369466e-05,
      "loss": 4.258,
      "step": 470000
    },
    {
      "epoch": 19.811996796357963,
      "eval_accuracy": 0.7499075332437094,
      "eval_loss": 4.028635025024414,
      "eval_runtime": 255.5233,
      "eval_samples_per_second": 472.387,
      "eval_steps_per_second": 4.923,
      "step": 470000
    },
    {
      "epoch": 19.833073388694515,
      "grad_norm": 1.2458850145339966,
      "learning_rate": 2.102216695965356e-05,
      "loss": 4.2574,
      "step": 470500
    },
    {
      "epoch": 19.854149981031068,
      "grad_norm": 1.239104151725769,
      "learning_rate": 2.093641469793766e-05,
      "loss": 4.254,
      "step": 471000
    },
    {
      "epoch": 19.87522657336762,
      "grad_norm": 1.2426058053970337,
      "learning_rate": 2.0850662436221757e-05,
      "loss": 4.2545,
      "step": 471500
    },
    {
      "epoch": 19.89630316570417,
      "grad_norm": 1.2764760255813599,
      "learning_rate": 2.0764910174505852e-05,
      "loss": 4.2554,
      "step": 472000
    },
    {
      "epoch": 19.91737975804072,
      "grad_norm": 1.26553213596344,
      "learning_rate": 2.067915791278995e-05,
      "loss": 4.2567,
      "step": 472500
    },
    {
      "epoch": 19.938456350377272,
      "grad_norm": 1.3121154308319092,
      "learning_rate": 2.059340565107405e-05,
      "loss": 4.253,
      "step": 473000
    },
    {
      "epoch": 19.95953294271382,
      "grad_norm": 1.2691972255706787,
      "learning_rate": 2.0507653389358144e-05,
      "loss": 4.2541,
      "step": 473500
    },
    {
      "epoch": 19.980609535050373,
      "grad_norm": 1.3131153583526611,
      "learning_rate": 2.0421901127642242e-05,
      "loss": 4.2524,
      "step": 474000
    },
    {
      "epoch": 20.001686127386925,
      "grad_norm": 1.3028374910354614,
      "learning_rate": 2.033614886592634e-05,
      "loss": 4.2612,
      "step": 474500
    },
    {
      "epoch": 20.022762719723474,
      "grad_norm": 1.2753231525421143,
      "learning_rate": 2.0250396604210435e-05,
      "loss": 4.253,
      "step": 475000
    },
    {
      "epoch": 20.043839312060026,
      "grad_norm": 1.3440321683883667,
      "learning_rate": 2.0164644342494534e-05,
      "loss": 4.2529,
      "step": 475500
    },
    {
      "epoch": 20.064915904396578,
      "grad_norm": 1.2624708414077759,
      "learning_rate": 2.007889208077863e-05,
      "loss": 4.247,
      "step": 476000
    },
    {
      "epoch": 20.085992496733127,
      "grad_norm": 1.2654080390930176,
      "learning_rate": 1.9993139819062727e-05,
      "loss": 4.2492,
      "step": 476500
    },
    {
      "epoch": 20.10706908906968,
      "grad_norm": 1.2830772399902344,
      "learning_rate": 1.9907387557346825e-05,
      "loss": 4.2533,
      "step": 477000
    },
    {
      "epoch": 20.12814568140623,
      "grad_norm": 1.2827842235565186,
      "learning_rate": 1.9821635295630923e-05,
      "loss": 4.2485,
      "step": 477500
    },
    {
      "epoch": 20.14922227374278,
      "grad_norm": 1.2869633436203003,
      "learning_rate": 1.9735883033915022e-05,
      "loss": 4.2475,
      "step": 478000
    },
    {
      "epoch": 20.17029886607933,
      "grad_norm": 1.2916399240493774,
      "learning_rate": 1.965013077219912e-05,
      "loss": 4.254,
      "step": 478500
    },
    {
      "epoch": 20.191375458415884,
      "grad_norm": 1.3249179124832153,
      "learning_rate": 1.9564378510483215e-05,
      "loss": 4.2489,
      "step": 479000
    },
    {
      "epoch": 20.212452050752436,
      "grad_norm": 1.292829155921936,
      "learning_rate": 1.9478626248767313e-05,
      "loss": 4.2531,
      "step": 479500
    },
    {
      "epoch": 20.233528643088984,
      "grad_norm": 1.3133717775344849,
      "learning_rate": 1.939287398705141e-05,
      "loss": 4.2503,
      "step": 480000
    },
    {
      "epoch": 20.233528643088984,
      "eval_accuracy": 0.75049086161305,
      "eval_loss": 4.024969100952148,
      "eval_runtime": 256.4627,
      "eval_samples_per_second": 470.657,
      "eval_steps_per_second": 4.905,
      "step": 480000
    },
    {
      "epoch": 20.254605235425537,
      "grad_norm": 1.2608989477157593,
      "learning_rate": 1.9307121725335506e-05,
      "loss": 4.247,
      "step": 480500
    },
    {
      "epoch": 20.27568182776209,
      "grad_norm": 1.3263756036758423,
      "learning_rate": 1.9221369463619605e-05,
      "loss": 4.247,
      "step": 481000
    },
    {
      "epoch": 20.296758420098637,
      "grad_norm": 1.3054836988449097,
      "learning_rate": 1.9135617201903703e-05,
      "loss": 4.2465,
      "step": 481500
    },
    {
      "epoch": 20.31783501243519,
      "grad_norm": 1.296824336051941,
      "learning_rate": 1.9049864940187798e-05,
      "loss": 4.2487,
      "step": 482000
    },
    {
      "epoch": 20.33891160477174,
      "grad_norm": 1.3231263160705566,
      "learning_rate": 1.8964112678471896e-05,
      "loss": 4.2508,
      "step": 482500
    },
    {
      "epoch": 20.35998819710829,
      "grad_norm": 1.2718220949172974,
      "learning_rate": 1.8878360416755995e-05,
      "loss": 4.2509,
      "step": 483000
    },
    {
      "epoch": 20.381064789444842,
      "grad_norm": 1.2969200611114502,
      "learning_rate": 1.879260815504009e-05,
      "loss": 4.2493,
      "step": 483500
    },
    {
      "epoch": 20.402141381781394,
      "grad_norm": 1.25778067111969,
      "learning_rate": 1.8706855893324188e-05,
      "loss": 4.2467,
      "step": 484000
    },
    {
      "epoch": 20.423217974117943,
      "grad_norm": 1.2875171899795532,
      "learning_rate": 1.8621103631608283e-05,
      "loss": 4.2489,
      "step": 484500
    },
    {
      "epoch": 20.444294566454495,
      "grad_norm": 1.5344430208206177,
      "learning_rate": 1.853535136989238e-05,
      "loss": 4.2463,
      "step": 485000
    },
    {
      "epoch": 20.465371158791047,
      "grad_norm": 1.2995513677597046,
      "learning_rate": 1.844959910817648e-05,
      "loss": 4.2456,
      "step": 485500
    },
    {
      "epoch": 20.4864477511276,
      "grad_norm": 1.3318088054656982,
      "learning_rate": 1.8363846846460574e-05,
      "loss": 4.2502,
      "step": 486000
    },
    {
      "epoch": 20.507524343464148,
      "grad_norm": 1.3028156757354736,
      "learning_rate": 1.8278094584744673e-05,
      "loss": 4.2447,
      "step": 486500
    },
    {
      "epoch": 20.5286009358007,
      "grad_norm": 1.318292498588562,
      "learning_rate": 1.819234232302877e-05,
      "loss": 4.2485,
      "step": 487000
    },
    {
      "epoch": 20.549677528137252,
      "grad_norm": 1.309216856956482,
      "learning_rate": 1.8106590061312866e-05,
      "loss": 4.2489,
      "step": 487500
    },
    {
      "epoch": 20.5707541204738,
      "grad_norm": 1.2924284934997559,
      "learning_rate": 1.8020837799596964e-05,
      "loss": 4.2479,
      "step": 488000
    },
    {
      "epoch": 20.591830712810353,
      "grad_norm": 1.2730611562728882,
      "learning_rate": 1.7935085537881062e-05,
      "loss": 4.2507,
      "step": 488500
    },
    {
      "epoch": 20.612907305146905,
      "grad_norm": 1.3122999668121338,
      "learning_rate": 1.7849333276165157e-05,
      "loss": 4.2462,
      "step": 489000
    },
    {
      "epoch": 20.633983897483454,
      "grad_norm": 1.3009910583496094,
      "learning_rate": 1.7763581014449256e-05,
      "loss": 4.2449,
      "step": 489500
    },
    {
      "epoch": 20.655060489820006,
      "grad_norm": 1.2579985857009888,
      "learning_rate": 1.7677828752733354e-05,
      "loss": 4.2446,
      "step": 490000
    },
    {
      "epoch": 20.655060489820006,
      "eval_accuracy": 0.7512521098802007,
      "eval_loss": 4.020163059234619,
      "eval_runtime": 255.0308,
      "eval_samples_per_second": 473.3,
      "eval_steps_per_second": 4.933,
      "step": 490000
    },
    {
      "epoch": 20.676137082156558,
      "grad_norm": 1.2867764234542847,
      "learning_rate": 1.7592076491017452e-05,
      "loss": 4.248,
      "step": 490500
    },
    {
      "epoch": 20.697213674493106,
      "grad_norm": 1.3010413646697998,
      "learning_rate": 1.7506324229301547e-05,
      "loss": 4.2501,
      "step": 491000
    },
    {
      "epoch": 20.71829026682966,
      "grad_norm": 1.2973271608352661,
      "learning_rate": 1.7420571967585646e-05,
      "loss": 4.2482,
      "step": 491500
    },
    {
      "epoch": 20.73936685916621,
      "grad_norm": 1.3121922016143799,
      "learning_rate": 1.7334819705869744e-05,
      "loss": 4.2441,
      "step": 492000
    },
    {
      "epoch": 20.760443451502763,
      "grad_norm": 1.2721258401870728,
      "learning_rate": 1.7249067444153842e-05,
      "loss": 4.2495,
      "step": 492500
    },
    {
      "epoch": 20.78152004383931,
      "grad_norm": 1.26395583152771,
      "learning_rate": 1.7163315182437937e-05,
      "loss": 4.2413,
      "step": 493000
    },
    {
      "epoch": 20.802596636175863,
      "grad_norm": 1.326419711112976,
      "learning_rate": 1.7077562920722035e-05,
      "loss": 4.2439,
      "step": 493500
    },
    {
      "epoch": 20.823673228512416,
      "grad_norm": 1.2681080102920532,
      "learning_rate": 1.6991810659006134e-05,
      "loss": 4.2457,
      "step": 494000
    },
    {
      "epoch": 20.844749820848964,
      "grad_norm": 1.2592014074325562,
      "learning_rate": 1.690605839729023e-05,
      "loss": 4.2463,
      "step": 494500
    },
    {
      "epoch": 20.865826413185516,
      "grad_norm": 1.2352417707443237,
      "learning_rate": 1.6820306135574327e-05,
      "loss": 4.2447,
      "step": 495000
    },
    {
      "epoch": 20.88690300552207,
      "grad_norm": 1.3348698616027832,
      "learning_rate": 1.6734553873858425e-05,
      "loss": 4.2483,
      "step": 495500
    },
    {
      "epoch": 20.907979597858617,
      "grad_norm": 1.3098927736282349,
      "learning_rate": 1.664880161214252e-05,
      "loss": 4.2447,
      "step": 496000
    },
    {
      "epoch": 20.92905619019517,
      "grad_norm": 1.2859501838684082,
      "learning_rate": 1.656304935042662e-05,
      "loss": 4.2483,
      "step": 496500
    },
    {
      "epoch": 20.95013278253172,
      "grad_norm": 1.284772515296936,
      "learning_rate": 1.6477297088710717e-05,
      "loss": 4.2442,
      "step": 497000
    },
    {
      "epoch": 20.97120937486827,
      "grad_norm": 1.2983043193817139,
      "learning_rate": 1.639154482699481e-05,
      "loss": 4.2458,
      "step": 497500
    },
    {
      "epoch": 20.992285967204822,
      "grad_norm": 1.2776328325271606,
      "learning_rate": 1.630579256527891e-05,
      "loss": 4.2434,
      "step": 498000
    },
    {
      "epoch": 21.013362559541374,
      "grad_norm": 1.2806689739227295,
      "learning_rate": 1.6220040303563008e-05,
      "loss": 4.2416,
      "step": 498500
    },
    {
      "epoch": 21.034439151877926,
      "grad_norm": 1.3134959936141968,
      "learning_rate": 1.6134288041847103e-05,
      "loss": 4.2452,
      "step": 499000
    },
    {
      "epoch": 21.055515744214475,
      "grad_norm": 1.2876100540161133,
      "learning_rate": 1.60485357801312e-05,
      "loss": 4.2395,
      "step": 499500
    },
    {
      "epoch": 21.076592336551027,
      "grad_norm": 1.3717353343963623,
      "learning_rate": 1.59627835184153e-05,
      "loss": 4.2401,
      "step": 500000
    },
    {
      "epoch": 21.076592336551027,
      "eval_accuracy": 0.7517408221839063,
      "eval_loss": 4.015739917755127,
      "eval_runtime": 255.9228,
      "eval_samples_per_second": 471.65,
      "eval_steps_per_second": 4.916,
      "step": 500000
    },
    {
      "epoch": 21.09766892888758,
      "grad_norm": 1.2869490385055542,
      "learning_rate": 1.5877031256699395e-05,
      "loss": 4.2401,
      "step": 500500
    },
    {
      "epoch": 21.118745521224128,
      "grad_norm": 1.2520095109939575,
      "learning_rate": 1.5791278994983493e-05,
      "loss": 4.2415,
      "step": 501000
    },
    {
      "epoch": 21.13982211356068,
      "grad_norm": 1.2873499393463135,
      "learning_rate": 1.570552673326759e-05,
      "loss": 4.2369,
      "step": 501500
    },
    {
      "epoch": 21.16089870589723,
      "grad_norm": 1.2714900970458984,
      "learning_rate": 1.5619774471551686e-05,
      "loss": 4.2423,
      "step": 502000
    },
    {
      "epoch": 21.18197529823378,
      "grad_norm": 1.2598556280136108,
      "learning_rate": 1.5534022209835785e-05,
      "loss": 4.2419,
      "step": 502500
    },
    {
      "epoch": 21.203051890570332,
      "grad_norm": 1.2671536207199097,
      "learning_rate": 1.544826994811988e-05,
      "loss": 4.2416,
      "step": 503000
    },
    {
      "epoch": 21.224128482906885,
      "grad_norm": 1.2995229959487915,
      "learning_rate": 1.5362517686403978e-05,
      "loss": 4.2405,
      "step": 503500
    },
    {
      "epoch": 21.245205075243433,
      "grad_norm": 1.2986161708831787,
      "learning_rate": 1.5276765424688076e-05,
      "loss": 4.2406,
      "step": 504000
    },
    {
      "epoch": 21.266281667579985,
      "grad_norm": 1.337524652481079,
      "learning_rate": 1.5191013162972173e-05,
      "loss": 4.2389,
      "step": 504500
    },
    {
      "epoch": 21.287358259916537,
      "grad_norm": 1.28804612159729,
      "learning_rate": 1.5105260901256271e-05,
      "loss": 4.2407,
      "step": 505000
    },
    {
      "epoch": 21.308434852253086,
      "grad_norm": 1.3123669624328613,
      "learning_rate": 1.501950863954037e-05,
      "loss": 4.2391,
      "step": 505500
    },
    {
      "epoch": 21.329511444589638,
      "grad_norm": 1.2873826026916504,
      "learning_rate": 1.4933756377824464e-05,
      "loss": 4.2373,
      "step": 506000
    },
    {
      "epoch": 21.35058803692619,
      "grad_norm": 1.2703092098236084,
      "learning_rate": 1.4848004116108563e-05,
      "loss": 4.2353,
      "step": 506500
    },
    {
      "epoch": 21.371664629262742,
      "grad_norm": 1.2705659866333008,
      "learning_rate": 1.4762251854392661e-05,
      "loss": 4.2425,
      "step": 507000
    },
    {
      "epoch": 21.39274122159929,
      "grad_norm": 1.2772016525268555,
      "learning_rate": 1.4676499592676758e-05,
      "loss": 4.2414,
      "step": 507500
    },
    {
      "epoch": 21.413817813935843,
      "grad_norm": 1.2497607469558716,
      "learning_rate": 1.4590747330960856e-05,
      "loss": 4.2387,
      "step": 508000
    },
    {
      "epoch": 21.434894406272395,
      "grad_norm": 1.2944022417068481,
      "learning_rate": 1.4504995069244952e-05,
      "loss": 4.2368,
      "step": 508500
    },
    {
      "epoch": 21.455970998608944,
      "grad_norm": 1.2324334383010864,
      "learning_rate": 1.4419242807529049e-05,
      "loss": 4.2395,
      "step": 509000
    },
    {
      "epoch": 21.477047590945496,
      "grad_norm": 1.2489604949951172,
      "learning_rate": 1.4333490545813147e-05,
      "loss": 4.2377,
      "step": 509500
    },
    {
      "epoch": 21.498124183282048,
      "grad_norm": 1.3092281818389893,
      "learning_rate": 1.4247738284097246e-05,
      "loss": 4.2359,
      "step": 510000
    },
    {
      "epoch": 21.498124183282048,
      "eval_accuracy": 0.7523945387384745,
      "eval_loss": 4.012452602386475,
      "eval_runtime": 254.746,
      "eval_samples_per_second": 473.829,
      "eval_steps_per_second": 4.938,
      "step": 510000
    },
    {
      "epoch": 21.519200775618597,
      "grad_norm": 1.2477611303329468,
      "learning_rate": 1.416198602238134e-05,
      "loss": 4.2383,
      "step": 510500
    },
    {
      "epoch": 21.54027736795515,
      "grad_norm": 1.2537931203842163,
      "learning_rate": 1.4076233760665439e-05,
      "loss": 4.2368,
      "step": 511000
    },
    {
      "epoch": 21.5613539602917,
      "grad_norm": 1.2709437608718872,
      "learning_rate": 1.3990481498949534e-05,
      "loss": 4.2347,
      "step": 511500
    },
    {
      "epoch": 21.58243055262825,
      "grad_norm": 1.2578589916229248,
      "learning_rate": 1.3904729237233632e-05,
      "loss": 4.2403,
      "step": 512000
    },
    {
      "epoch": 21.6035071449648,
      "grad_norm": 1.2628810405731201,
      "learning_rate": 1.381897697551773e-05,
      "loss": 4.2383,
      "step": 512500
    },
    {
      "epoch": 21.624583737301354,
      "grad_norm": 1.3196122646331787,
      "learning_rate": 1.3733224713801825e-05,
      "loss": 4.241,
      "step": 513000
    },
    {
      "epoch": 21.645660329637906,
      "grad_norm": 1.3461692333221436,
      "learning_rate": 1.3647472452085924e-05,
      "loss": 4.2413,
      "step": 513500
    },
    {
      "epoch": 21.666736921974454,
      "grad_norm": 1.2827414274215698,
      "learning_rate": 1.3561720190370022e-05,
      "loss": 4.2374,
      "step": 514000
    },
    {
      "epoch": 21.687813514311006,
      "grad_norm": 1.2843877077102661,
      "learning_rate": 1.3475967928654119e-05,
      "loss": 4.2351,
      "step": 514500
    },
    {
      "epoch": 21.70889010664756,
      "grad_norm": 1.2885445356369019,
      "learning_rate": 1.3390215666938217e-05,
      "loss": 4.2386,
      "step": 515000
    },
    {
      "epoch": 21.729966698984107,
      "grad_norm": 1.246342420578003,
      "learning_rate": 1.3304463405222315e-05,
      "loss": 4.2369,
      "step": 515500
    },
    {
      "epoch": 21.75104329132066,
      "grad_norm": 1.2893306016921997,
      "learning_rate": 1.321871114350641e-05,
      "loss": 4.238,
      "step": 516000
    },
    {
      "epoch": 21.77211988365721,
      "grad_norm": 2.3378725051879883,
      "learning_rate": 1.3132958881790508e-05,
      "loss": 4.2339,
      "step": 516500
    },
    {
      "epoch": 21.79319647599376,
      "grad_norm": 1.2766624689102173,
      "learning_rate": 1.3047206620074607e-05,
      "loss": 4.2327,
      "step": 517000
    },
    {
      "epoch": 21.814273068330312,
      "grad_norm": 1.3061493635177612,
      "learning_rate": 1.2961454358358702e-05,
      "loss": 4.2358,
      "step": 517500
    },
    {
      "epoch": 21.835349660666864,
      "grad_norm": 1.3490527868270874,
      "learning_rate": 1.28757020966428e-05,
      "loss": 4.2342,
      "step": 518000
    },
    {
      "epoch": 21.856426253003413,
      "grad_norm": 1.2776349782943726,
      "learning_rate": 1.2789949834926898e-05,
      "loss": 4.2324,
      "step": 518500
    },
    {
      "epoch": 21.877502845339965,
      "grad_norm": 1.2812455892562866,
      "learning_rate": 1.2704197573210993e-05,
      "loss": 4.2386,
      "step": 519000
    },
    {
      "epoch": 21.898579437676517,
      "grad_norm": 1.2636483907699585,
      "learning_rate": 1.2618445311495092e-05,
      "loss": 4.2326,
      "step": 519500
    },
    {
      "epoch": 21.91965603001307,
      "grad_norm": 5.026463508605957,
      "learning_rate": 1.2532693049779186e-05,
      "loss": 4.2301,
      "step": 520000
    },
    {
      "epoch": 21.91965603001307,
      "eval_accuracy": 0.7527027015459574,
      "eval_loss": 4.010770797729492,
      "eval_runtime": 254.8565,
      "eval_samples_per_second": 473.623,
      "eval_steps_per_second": 4.936,
      "step": 520000
    },
    {
      "epoch": 21.940732622349618,
      "grad_norm": 1.3325153589248657,
      "learning_rate": 1.2446940788063285e-05,
      "loss": 4.2336,
      "step": 520500
    },
    {
      "epoch": 21.96180921468617,
      "grad_norm": 1.329563856124878,
      "learning_rate": 1.2361188526347383e-05,
      "loss": 4.2359,
      "step": 521000
    },
    {
      "epoch": 21.982885807022722,
      "grad_norm": 1.284597635269165,
      "learning_rate": 1.227543626463148e-05,
      "loss": 4.236,
      "step": 521500
    },
    {
      "epoch": 22.00396239935927,
      "grad_norm": 1.3343936204910278,
      "learning_rate": 1.2189684002915578e-05,
      "loss": 4.2295,
      "step": 522000
    },
    {
      "epoch": 22.025038991695823,
      "grad_norm": 1.3150198459625244,
      "learning_rate": 1.2103931741199675e-05,
      "loss": 4.2292,
      "step": 522500
    },
    {
      "epoch": 22.046115584032375,
      "grad_norm": 1.3090901374816895,
      "learning_rate": 1.2018179479483773e-05,
      "loss": 4.2318,
      "step": 523000
    },
    {
      "epoch": 22.067192176368923,
      "grad_norm": 1.2930960655212402,
      "learning_rate": 1.193242721776787e-05,
      "loss": 4.233,
      "step": 523500
    },
    {
      "epoch": 22.088268768705476,
      "grad_norm": 1.3193879127502441,
      "learning_rate": 1.1846674956051966e-05,
      "loss": 4.234,
      "step": 524000
    },
    {
      "epoch": 22.109345361042028,
      "grad_norm": 1.3005449771881104,
      "learning_rate": 1.1760922694336064e-05,
      "loss": 4.2286,
      "step": 524500
    },
    {
      "epoch": 22.130421953378576,
      "grad_norm": 1.3268805742263794,
      "learning_rate": 1.1675170432620161e-05,
      "loss": 4.2339,
      "step": 525000
    },
    {
      "epoch": 22.15149854571513,
      "grad_norm": 1.2806557416915894,
      "learning_rate": 1.1589418170904258e-05,
      "loss": 4.2259,
      "step": 525500
    },
    {
      "epoch": 22.17257513805168,
      "grad_norm": 1.277265191078186,
      "learning_rate": 1.1503665909188354e-05,
      "loss": 4.2299,
      "step": 526000
    },
    {
      "epoch": 22.193651730388233,
      "grad_norm": 1.3337393999099731,
      "learning_rate": 1.1417913647472453e-05,
      "loss": 4.2319,
      "step": 526500
    },
    {
      "epoch": 22.21472832272478,
      "grad_norm": 1.2928199768066406,
      "learning_rate": 1.133216138575655e-05,
      "loss": 4.2296,
      "step": 527000
    },
    {
      "epoch": 22.235804915061333,
      "grad_norm": 1.3093153238296509,
      "learning_rate": 1.1246409124040646e-05,
      "loss": 4.2294,
      "step": 527500
    },
    {
      "epoch": 22.256881507397885,
      "grad_norm": 1.3133089542388916,
      "learning_rate": 1.1160656862324744e-05,
      "loss": 4.232,
      "step": 528000
    },
    {
      "epoch": 22.277958099734434,
      "grad_norm": 1.2961432933807373,
      "learning_rate": 1.1074904600608842e-05,
      "loss": 4.2302,
      "step": 528500
    },
    {
      "epoch": 22.299034692070986,
      "grad_norm": 1.300223708152771,
      "learning_rate": 1.0989152338892939e-05,
      "loss": 4.2263,
      "step": 529000
    },
    {
      "epoch": 22.320111284407538,
      "grad_norm": 1.309266448020935,
      "learning_rate": 1.0903400077177037e-05,
      "loss": 4.2311,
      "step": 529500
    },
    {
      "epoch": 22.341187876744087,
      "grad_norm": 1.2745083570480347,
      "learning_rate": 1.0817647815461134e-05,
      "loss": 4.2318,
      "step": 530000
    },
    {
      "epoch": 22.341187876744087,
      "eval_accuracy": 0.753187119415914,
      "eval_loss": 4.007522106170654,
      "eval_runtime": 254.9803,
      "eval_samples_per_second": 473.393,
      "eval_steps_per_second": 4.934,
      "step": 530000
    },
    {
      "epoch": 22.36226446908064,
      "grad_norm": 1.27109956741333,
      "learning_rate": 1.073189555374523e-05,
      "loss": 4.2313,
      "step": 530500
    },
    {
      "epoch": 22.38334106141719,
      "grad_norm": 1.3444362878799438,
      "learning_rate": 1.0646143292029327e-05,
      "loss": 4.2344,
      "step": 531000
    },
    {
      "epoch": 22.40441765375374,
      "grad_norm": 1.3122986555099487,
      "learning_rate": 1.0560391030313425e-05,
      "loss": 4.2248,
      "step": 531500
    },
    {
      "epoch": 22.42549424609029,
      "grad_norm": 1.2821848392486572,
      "learning_rate": 1.0474638768597522e-05,
      "loss": 4.2278,
      "step": 532000
    },
    {
      "epoch": 22.446570838426844,
      "grad_norm": 1.2990642786026,
      "learning_rate": 1.0388886506881619e-05,
      "loss": 4.2295,
      "step": 532500
    },
    {
      "epoch": 22.467647430763392,
      "grad_norm": 1.319882869720459,
      "learning_rate": 1.0303134245165717e-05,
      "loss": 4.2305,
      "step": 533000
    },
    {
      "epoch": 22.488724023099945,
      "grad_norm": 1.3165510892868042,
      "learning_rate": 1.0217381983449814e-05,
      "loss": 4.228,
      "step": 533500
    },
    {
      "epoch": 22.509800615436497,
      "grad_norm": 1.280920147895813,
      "learning_rate": 1.013162972173391e-05,
      "loss": 4.2293,
      "step": 534000
    },
    {
      "epoch": 22.53087720777305,
      "grad_norm": 1.2954820394515991,
      "learning_rate": 1.0045877460018009e-05,
      "loss": 4.2302,
      "step": 534500
    },
    {
      "epoch": 22.551953800109597,
      "grad_norm": 1.2912240028381348,
      "learning_rate": 9.960125198302105e-06,
      "loss": 4.2333,
      "step": 535000
    },
    {
      "epoch": 22.57303039244615,
      "grad_norm": 1.309599757194519,
      "learning_rate": 9.874372936586203e-06,
      "loss": 4.2292,
      "step": 535500
    },
    {
      "epoch": 22.5941069847827,
      "grad_norm": 1.2793676853179932,
      "learning_rate": 9.7886206748703e-06,
      "loss": 4.2274,
      "step": 536000
    },
    {
      "epoch": 22.61518357711925,
      "grad_norm": 1.3115402460098267,
      "learning_rate": 9.702868413154398e-06,
      "loss": 4.2268,
      "step": 536500
    },
    {
      "epoch": 22.636260169455802,
      "grad_norm": 1.3101269006729126,
      "learning_rate": 9.617116151438495e-06,
      "loss": 4.2267,
      "step": 537000
    },
    {
      "epoch": 22.657336761792354,
      "grad_norm": 1.302416205406189,
      "learning_rate": 9.531363889722592e-06,
      "loss": 4.2274,
      "step": 537500
    },
    {
      "epoch": 22.678413354128903,
      "grad_norm": 1.2654656171798706,
      "learning_rate": 9.44561162800669e-06,
      "loss": 4.2248,
      "step": 538000
    },
    {
      "epoch": 22.699489946465455,
      "grad_norm": 1.3007142543792725,
      "learning_rate": 9.359859366290787e-06,
      "loss": 4.2275,
      "step": 538500
    },
    {
      "epoch": 22.720566538802007,
      "grad_norm": 1.2624433040618896,
      "learning_rate": 9.274107104574883e-06,
      "loss": 4.2294,
      "step": 539000
    },
    {
      "epoch": 22.741643131138556,
      "grad_norm": 1.2674312591552734,
      "learning_rate": 9.18835484285898e-06,
      "loss": 4.2245,
      "step": 539500
    },
    {
      "epoch": 22.762719723475108,
      "grad_norm": 1.333701252937317,
      "learning_rate": 9.102602581143078e-06,
      "loss": 4.2247,
      "step": 540000
    },
    {
      "epoch": 22.762719723475108,
      "eval_accuracy": 0.7535502076113693,
      "eval_loss": 4.006025314331055,
      "eval_runtime": 254.9954,
      "eval_samples_per_second": 473.365,
      "eval_steps_per_second": 4.933,
      "step": 540000
    },
    {
      "epoch": 22.78379631581166,
      "grad_norm": 1.3177605867385864,
      "learning_rate": 9.016850319427175e-06,
      "loss": 4.2294,
      "step": 540500
    },
    {
      "epoch": 22.804872908148212,
      "grad_norm": 1.2842469215393066,
      "learning_rate": 8.931098057711271e-06,
      "loss": 4.2263,
      "step": 541000
    },
    {
      "epoch": 22.82594950048476,
      "grad_norm": 1.3305730819702148,
      "learning_rate": 8.84534579599537e-06,
      "loss": 4.2277,
      "step": 541500
    },
    {
      "epoch": 22.847026092821313,
      "grad_norm": 1.3207364082336426,
      "learning_rate": 8.759593534279468e-06,
      "loss": 4.227,
      "step": 542000
    },
    {
      "epoch": 22.868102685157865,
      "grad_norm": 1.2922980785369873,
      "learning_rate": 8.673841272563565e-06,
      "loss": 4.2254,
      "step": 542500
    },
    {
      "epoch": 22.889179277494414,
      "grad_norm": 1.3153431415557861,
      "learning_rate": 8.588089010847663e-06,
      "loss": 4.2269,
      "step": 543000
    },
    {
      "epoch": 22.910255869830966,
      "grad_norm": 1.3293088674545288,
      "learning_rate": 8.50233674913176e-06,
      "loss": 4.2265,
      "step": 543500
    },
    {
      "epoch": 22.931332462167518,
      "grad_norm": 1.3572715520858765,
      "learning_rate": 8.416584487415856e-06,
      "loss": 4.223,
      "step": 544000
    },
    {
      "epoch": 22.952409054504066,
      "grad_norm": 1.3862813711166382,
      "learning_rate": 8.330832225699953e-06,
      "loss": 4.221,
      "step": 544500
    },
    {
      "epoch": 22.97348564684062,
      "grad_norm": 1.3563883304595947,
      "learning_rate": 8.245079963984051e-06,
      "loss": 4.2216,
      "step": 545000
    },
    {
      "epoch": 22.99456223917717,
      "grad_norm": 1.3374483585357666,
      "learning_rate": 8.159327702268148e-06,
      "loss": 4.2285,
      "step": 545500
    },
    {
      "epoch": 23.01563883151372,
      "grad_norm": 1.313022494316101,
      "learning_rate": 8.073575440552244e-06,
      "loss": 4.2239,
      "step": 546000
    },
    {
      "epoch": 23.03671542385027,
      "grad_norm": 1.2984898090362549,
      "learning_rate": 7.987823178836343e-06,
      "loss": 4.2235,
      "step": 546500
    },
    {
      "epoch": 23.057792016186824,
      "grad_norm": 1.3074736595153809,
      "learning_rate": 7.90207091712044e-06,
      "loss": 4.2235,
      "step": 547000
    },
    {
      "epoch": 23.078868608523376,
      "grad_norm": 1.340393304824829,
      "learning_rate": 7.816318655404536e-06,
      "loss": 4.2232,
      "step": 547500
    },
    {
      "epoch": 23.099945200859924,
      "grad_norm": 1.315596580505371,
      "learning_rate": 7.730566393688634e-06,
      "loss": 4.2246,
      "step": 548000
    },
    {
      "epoch": 23.121021793196476,
      "grad_norm": 1.295782208442688,
      "learning_rate": 7.64481413197273e-06,
      "loss": 4.2235,
      "step": 548500
    },
    {
      "epoch": 23.14209838553303,
      "grad_norm": 1.2932542562484741,
      "learning_rate": 7.559061870256828e-06,
      "loss": 4.2225,
      "step": 549000
    },
    {
      "epoch": 23.163174977869577,
      "grad_norm": 1.3115209341049194,
      "learning_rate": 7.473309608540925e-06,
      "loss": 4.2231,
      "step": 549500
    },
    {
      "epoch": 23.18425157020613,
      "grad_norm": 1.3135550022125244,
      "learning_rate": 7.387557346825023e-06,
      "loss": 4.2241,
      "step": 550000
    },
    {
      "epoch": 23.18425157020613,
      "eval_accuracy": 0.7539634576294487,
      "eval_loss": 4.0021138191223145,
      "eval_runtime": 254.8811,
      "eval_samples_per_second": 473.578,
      "eval_steps_per_second": 4.936,
      "step": 550000
    },
    {
      "epoch": 23.20532816254268,
      "grad_norm": 1.3412859439849854,
      "learning_rate": 7.3018050851091205e-06,
      "loss": 4.2196,
      "step": 550500
    },
    {
      "epoch": 23.22640475487923,
      "grad_norm": 1.3183077573776245,
      "learning_rate": 7.216052823393217e-06,
      "loss": 4.2233,
      "step": 551000
    },
    {
      "epoch": 23.247481347215782,
      "grad_norm": 1.3424664735794067,
      "learning_rate": 7.1303005616773155e-06,
      "loss": 4.222,
      "step": 551500
    },
    {
      "epoch": 23.268557939552334,
      "grad_norm": 1.3505834341049194,
      "learning_rate": 7.044548299961412e-06,
      "loss": 4.2263,
      "step": 552000
    },
    {
      "epoch": 23.289634531888883,
      "grad_norm": 1.3483772277832031,
      "learning_rate": 6.958796038245509e-06,
      "loss": 4.2171,
      "step": 552500
    },
    {
      "epoch": 23.310711124225435,
      "grad_norm": 1.3984652757644653,
      "learning_rate": 6.873043776529606e-06,
      "loss": 4.2258,
      "step": 553000
    },
    {
      "epoch": 23.331787716561987,
      "grad_norm": 1.3465911149978638,
      "learning_rate": 6.787291514813704e-06,
      "loss": 4.2196,
      "step": 553500
    },
    {
      "epoch": 23.35286430889854,
      "grad_norm": 1.3455379009246826,
      "learning_rate": 6.701539253097801e-06,
      "loss": 4.2223,
      "step": 554000
    },
    {
      "epoch": 23.373940901235088,
      "grad_norm": 1.340900182723999,
      "learning_rate": 6.615786991381898e-06,
      "loss": 4.2216,
      "step": 554500
    },
    {
      "epoch": 23.39501749357164,
      "grad_norm": 1.292914867401123,
      "learning_rate": 6.530034729665996e-06,
      "loss": 4.2218,
      "step": 555000
    },
    {
      "epoch": 23.416094085908192,
      "grad_norm": 1.3063126802444458,
      "learning_rate": 6.444282467950093e-06,
      "loss": 4.2209,
      "step": 555500
    },
    {
      "epoch": 23.43717067824474,
      "grad_norm": 1.2974998950958252,
      "learning_rate": 6.358530206234189e-06,
      "loss": 4.2201,
      "step": 556000
    },
    {
      "epoch": 23.458247270581293,
      "grad_norm": 1.2745603322982788,
      "learning_rate": 6.272777944518287e-06,
      "loss": 4.2235,
      "step": 556500
    },
    {
      "epoch": 23.479323862917845,
      "grad_norm": 1.3094799518585205,
      "learning_rate": 6.187025682802384e-06,
      "loss": 4.2232,
      "step": 557000
    },
    {
      "epoch": 23.500400455254393,
      "grad_norm": 1.3310980796813965,
      "learning_rate": 6.101273421086482e-06,
      "loss": 4.2184,
      "step": 557500
    },
    {
      "epoch": 23.521477047590945,
      "grad_norm": 1.328223466873169,
      "learning_rate": 6.015521159370579e-06,
      "loss": 4.2218,
      "step": 558000
    },
    {
      "epoch": 23.542553639927497,
      "grad_norm": 1.3627145290374756,
      "learning_rate": 5.929768897654676e-06,
      "loss": 4.2198,
      "step": 558500
    },
    {
      "epoch": 23.563630232264046,
      "grad_norm": 1.3108727931976318,
      "learning_rate": 5.844016635938773e-06,
      "loss": 4.2204,
      "step": 559000
    },
    {
      "epoch": 23.584706824600598,
      "grad_norm": 1.2730662822723389,
      "learning_rate": 5.75826437422287e-06,
      "loss": 4.2194,
      "step": 559500
    },
    {
      "epoch": 23.60578341693715,
      "grad_norm": 1.369991421699524,
      "learning_rate": 5.672512112506967e-06,
      "loss": 4.2201,
      "step": 560000
    },
    {
      "epoch": 23.60578341693715,
      "eval_accuracy": 0.7545900168942385,
      "eval_loss": 3.9990134239196777,
      "eval_runtime": 253.7801,
      "eval_samples_per_second": 475.632,
      "eval_steps_per_second": 4.957,
      "step": 560000
    },
    {
      "epoch": 23.6268600092737,
      "grad_norm": 1.3139984607696533,
      "learning_rate": 5.5867598507910655e-06,
      "loss": 4.2211,
      "step": 560500
    },
    {
      "epoch": 23.64793660161025,
      "grad_norm": 1.349632740020752,
      "learning_rate": 5.501007589075162e-06,
      "loss": 4.2199,
      "step": 561000
    },
    {
      "epoch": 23.669013193946803,
      "grad_norm": 1.3889633417129517,
      "learning_rate": 5.41525532735926e-06,
      "loss": 4.22,
      "step": 561500
    },
    {
      "epoch": 23.690089786283355,
      "grad_norm": 1.3906526565551758,
      "learning_rate": 5.329503065643356e-06,
      "loss": 4.2167,
      "step": 562000
    },
    {
      "epoch": 23.711166378619904,
      "grad_norm": 1.3821567296981812,
      "learning_rate": 5.243750803927454e-06,
      "loss": 4.2169,
      "step": 562500
    },
    {
      "epoch": 23.732242970956456,
      "grad_norm": 1.3548955917358398,
      "learning_rate": 5.157998542211551e-06,
      "loss": 4.22,
      "step": 563000
    },
    {
      "epoch": 23.753319563293008,
      "grad_norm": 1.2988624572753906,
      "learning_rate": 5.072246280495649e-06,
      "loss": 4.2162,
      "step": 563500
    },
    {
      "epoch": 23.774396155629557,
      "grad_norm": 1.3006418943405151,
      "learning_rate": 4.986494018779746e-06,
      "loss": 4.2196,
      "step": 564000
    },
    {
      "epoch": 23.79547274796611,
      "grad_norm": 1.340950846672058,
      "learning_rate": 4.900741757063843e-06,
      "loss": 4.22,
      "step": 564500
    },
    {
      "epoch": 23.81654934030266,
      "grad_norm": 1.3255163431167603,
      "learning_rate": 4.81498949534794e-06,
      "loss": 4.2171,
      "step": 565000
    },
    {
      "epoch": 23.83762593263921,
      "grad_norm": 1.3288049697875977,
      "learning_rate": 4.729237233632038e-06,
      "loss": 4.2158,
      "step": 565500
    },
    {
      "epoch": 23.85870252497576,
      "grad_norm": 1.300318717956543,
      "learning_rate": 4.643484971916134e-06,
      "loss": 4.2207,
      "step": 566000
    },
    {
      "epoch": 23.879779117312314,
      "grad_norm": 1.33096182346344,
      "learning_rate": 4.557732710200232e-06,
      "loss": 4.2168,
      "step": 566500
    },
    {
      "epoch": 23.900855709648862,
      "grad_norm": 1.3673837184906006,
      "learning_rate": 4.471980448484329e-06,
      "loss": 4.2176,
      "step": 567000
    },
    {
      "epoch": 23.921932301985414,
      "grad_norm": 1.3524250984191895,
      "learning_rate": 4.386228186768427e-06,
      "loss": 4.218,
      "step": 567500
    },
    {
      "epoch": 23.943008894321967,
      "grad_norm": 1.3254584074020386,
      "learning_rate": 4.300475925052524e-06,
      "loss": 4.2184,
      "step": 568000
    },
    {
      "epoch": 23.96408548665852,
      "grad_norm": 1.315349817276001,
      "learning_rate": 4.214723663336621e-06,
      "loss": 4.2222,
      "step": 568500
    },
    {
      "epoch": 23.985162078995067,
      "grad_norm": 1.3442597389221191,
      "learning_rate": 4.128971401620718e-06,
      "loss": 4.2158,
      "step": 569000
    },
    {
      "epoch": 24.00623867133162,
      "grad_norm": 1.312483310699463,
      "learning_rate": 4.043219139904815e-06,
      "loss": 4.2167,
      "step": 569500
    },
    {
      "epoch": 24.02731526366817,
      "grad_norm": 1.273382544517517,
      "learning_rate": 3.957466878188912e-06,
      "loss": 4.2151,
      "step": 570000
    },
    {
      "epoch": 24.02731526366817,
      "eval_accuracy": 0.7548587067290734,
      "eval_loss": 3.998117685317993,
      "eval_runtime": 255.5731,
      "eval_samples_per_second": 472.295,
      "eval_steps_per_second": 4.922,
      "step": 570000
    },
    {
      "epoch": 24.04839185600472,
      "grad_norm": 1.349961757659912,
      "learning_rate": 3.87171461647301e-06,
      "loss": 4.2207,
      "step": 570500
    },
    {
      "epoch": 24.069468448341272,
      "grad_norm": 1.285981297492981,
      "learning_rate": 3.7859623547571067e-06,
      "loss": 4.2135,
      "step": 571000
    },
    {
      "epoch": 24.090545040677824,
      "grad_norm": 1.3049403429031372,
      "learning_rate": 3.7002100930412046e-06,
      "loss": 4.2144,
      "step": 571500
    },
    {
      "epoch": 24.111621633014373,
      "grad_norm": 1.2917020320892334,
      "learning_rate": 3.614457831325301e-06,
      "loss": 4.2202,
      "step": 572000
    },
    {
      "epoch": 24.132698225350925,
      "grad_norm": 1.353723406791687,
      "learning_rate": 3.5287055696093987e-06,
      "loss": 4.2155,
      "step": 572500
    },
    {
      "epoch": 24.153774817687477,
      "grad_norm": 1.3026580810546875,
      "learning_rate": 3.4429533078934957e-06,
      "loss": 4.2161,
      "step": 573000
    },
    {
      "epoch": 24.174851410024026,
      "grad_norm": 1.3186146020889282,
      "learning_rate": 3.357201046177593e-06,
      "loss": 4.2189,
      "step": 573500
    },
    {
      "epoch": 24.195928002360578,
      "grad_norm": 1.324676275253296,
      "learning_rate": 3.2714487844616906e-06,
      "loss": 4.2183,
      "step": 574000
    },
    {
      "epoch": 24.21700459469713,
      "grad_norm": 1.3474537134170532,
      "learning_rate": 3.1856965227457872e-06,
      "loss": 4.2171,
      "step": 574500
    },
    {
      "epoch": 24.238081187033682,
      "grad_norm": 1.3015869855880737,
      "learning_rate": 3.099944261029885e-06,
      "loss": 4.2201,
      "step": 575000
    },
    {
      "epoch": 24.25915777937023,
      "grad_norm": 1.350832223892212,
      "learning_rate": 3.014191999313982e-06,
      "loss": 4.2158,
      "step": 575500
    },
    {
      "epoch": 24.280234371706783,
      "grad_norm": 1.2905648946762085,
      "learning_rate": 2.928439737598079e-06,
      "loss": 4.2131,
      "step": 576000
    },
    {
      "epoch": 24.301310964043335,
      "grad_norm": 1.3341201543807983,
      "learning_rate": 2.8426874758821767e-06,
      "loss": 4.2174,
      "step": 576500
    },
    {
      "epoch": 24.322387556379883,
      "grad_norm": 1.3144711256027222,
      "learning_rate": 2.7569352141662737e-06,
      "loss": 4.211,
      "step": 577000
    },
    {
      "epoch": 24.343464148716436,
      "grad_norm": 1.322046160697937,
      "learning_rate": 2.6711829524503707e-06,
      "loss": 4.2133,
      "step": 577500
    },
    {
      "epoch": 24.364540741052988,
      "grad_norm": 1.3815555572509766,
      "learning_rate": 2.585430690734468e-06,
      "loss": 4.2127,
      "step": 578000
    },
    {
      "epoch": 24.385617333389536,
      "grad_norm": 1.335869312286377,
      "learning_rate": 2.4996784290185657e-06,
      "loss": 4.2162,
      "step": 578500
    },
    {
      "epoch": 24.40669392572609,
      "grad_norm": 1.3446495532989502,
      "learning_rate": 2.4139261673026627e-06,
      "loss": 4.211,
      "step": 579000
    },
    {
      "epoch": 24.42777051806264,
      "grad_norm": 1.3282661437988281,
      "learning_rate": 2.3281739055867597e-06,
      "loss": 4.2142,
      "step": 579500
    },
    {
      "epoch": 24.44884711039919,
      "grad_norm": 1.3273917436599731,
      "learning_rate": 2.242421643870857e-06,
      "loss": 4.2144,
      "step": 580000
    },
    {
      "epoch": 24.44884711039919,
      "eval_accuracy": 0.7553443904581703,
      "eval_loss": 3.9943079948425293,
      "eval_runtime": 254.7723,
      "eval_samples_per_second": 473.78,
      "eval_steps_per_second": 4.938,
      "step": 580000
    },
    {
      "epoch": 24.46992370273574,
      "grad_norm": 1.3435946702957153,
      "learning_rate": 2.1566693821549547e-06,
      "loss": 4.2135,
      "step": 580500
    },
    {
      "epoch": 24.491000295072293,
      "grad_norm": 1.344053030014038,
      "learning_rate": 2.0709171204390517e-06,
      "loss": 4.2171,
      "step": 581000
    },
    {
      "epoch": 24.512076887408845,
      "grad_norm": 1.342323899269104,
      "learning_rate": 1.985164858723149e-06,
      "loss": 4.2113,
      "step": 581500
    },
    {
      "epoch": 24.533153479745394,
      "grad_norm": 1.329810380935669,
      "learning_rate": 1.8994125970072462e-06,
      "loss": 4.218,
      "step": 582000
    },
    {
      "epoch": 24.554230072081946,
      "grad_norm": 1.3052133321762085,
      "learning_rate": 1.8136603352913432e-06,
      "loss": 4.2129,
      "step": 582500
    },
    {
      "epoch": 24.5753066644185,
      "grad_norm": 1.313083291053772,
      "learning_rate": 1.7279080735754405e-06,
      "loss": 4.2166,
      "step": 583000
    },
    {
      "epoch": 24.596383256755047,
      "grad_norm": 1.3530458211898804,
      "learning_rate": 1.642155811859538e-06,
      "loss": 4.2149,
      "step": 583500
    },
    {
      "epoch": 24.6174598490916,
      "grad_norm": 1.3114814758300781,
      "learning_rate": 1.5564035501436352e-06,
      "loss": 4.2143,
      "step": 584000
    },
    {
      "epoch": 24.63853644142815,
      "grad_norm": 1.3284308910369873,
      "learning_rate": 1.4706512884277324e-06,
      "loss": 4.2152,
      "step": 584500
    },
    {
      "epoch": 24.6596130337647,
      "grad_norm": 1.3248780965805054,
      "learning_rate": 1.3848990267118295e-06,
      "loss": 4.2128,
      "step": 585000
    },
    {
      "epoch": 24.680689626101252,
      "grad_norm": 1.3034669160842896,
      "learning_rate": 1.299146764995927e-06,
      "loss": 4.2143,
      "step": 585500
    },
    {
      "epoch": 24.701766218437804,
      "grad_norm": 1.3322868347167969,
      "learning_rate": 1.213394503280024e-06,
      "loss": 4.2125,
      "step": 586000
    },
    {
      "epoch": 24.722842810774353,
      "grad_norm": 1.3254417181015015,
      "learning_rate": 1.1276422415641212e-06,
      "loss": 4.2127,
      "step": 586500
    },
    {
      "epoch": 24.743919403110905,
      "grad_norm": 1.315095067024231,
      "learning_rate": 1.0418899798482187e-06,
      "loss": 4.2156,
      "step": 587000
    },
    {
      "epoch": 24.764995995447457,
      "grad_norm": 1.3563679456710815,
      "learning_rate": 9.561377181323157e-07,
      "loss": 4.2119,
      "step": 587500
    },
    {
      "epoch": 24.786072587784005,
      "grad_norm": 1.3376871347427368,
      "learning_rate": 8.70385456416413e-07,
      "loss": 4.2132,
      "step": 588000
    },
    {
      "epoch": 24.807149180120557,
      "grad_norm": 1.2854983806610107,
      "learning_rate": 7.846331947005103e-07,
      "loss": 4.2153,
      "step": 588500
    },
    {
      "epoch": 24.82822577245711,
      "grad_norm": 1.3570996522903442,
      "learning_rate": 6.988809329846075e-07,
      "loss": 4.2104,
      "step": 589000
    },
    {
      "epoch": 24.84930236479366,
      "grad_norm": 1.3520281314849854,
      "learning_rate": 6.131286712687047e-07,
      "loss": 4.214,
      "step": 589500
    },
    {
      "epoch": 24.87037895713021,
      "grad_norm": 1.298264741897583,
      "learning_rate": 5.27376409552802e-07,
      "loss": 4.2123,
      "step": 590000
    },
    {
      "epoch": 24.87037895713021,
      "eval_accuracy": 0.7552397811657311,
      "eval_loss": 3.9949495792388916,
      "eval_runtime": 254.5096,
      "eval_samples_per_second": 474.269,
      "eval_steps_per_second": 4.943,
      "step": 590000
    },
    {
      "epoch": 24.891455549466762,
      "grad_norm": 1.349381685256958,
      "learning_rate": 4.4162414783689917e-07,
      "loss": 4.212,
      "step": 590500
    },
    {
      "epoch": 24.912532141803315,
      "grad_norm": 1.2925467491149902,
      "learning_rate": 3.558718861209964e-07,
      "loss": 4.2122,
      "step": 591000
    },
    {
      "epoch": 24.933608734139863,
      "grad_norm": 1.4453474283218384,
      "learning_rate": 2.701196244050937e-07,
      "loss": 4.213,
      "step": 591500
    },
    {
      "epoch": 24.954685326476415,
      "grad_norm": 1.3745180368423462,
      "learning_rate": 1.8436736268919094e-07,
      "loss": 4.2131,
      "step": 592000
    },
    {
      "epoch": 24.975761918812967,
      "grad_norm": 1.3142321109771729,
      "learning_rate": 9.861510097328816e-08,
      "loss": 4.2113,
      "step": 592500
    },
    {
      "epoch": 24.996838511149516,
      "grad_norm": 1.3333407640457153,
      "learning_rate": 1.2862839257385414e-08,
      "loss": 4.2116,
      "step": 593000
    },
    {
      "epoch": 25.0,
      "step": 593075,
      "total_flos": 1.5095203731046195e+19,
      "train_loss": 4.736705173041704,
      "train_runtime": 270691.9239,
      "train_samples_per_second": 210.326,
      "train_steps_per_second": 2.191
    }
  ],
  "logging_steps": 500,
  "max_steps": 593075,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 25,
  "save_steps": 10000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.5095203731046195e+19,
  "train_batch_size": 96,
  "trial_name": null,
  "trial_params": null
}