diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8875 @@
+{
+  "best_metric": 4.645676612854004,
+  "best_model_checkpoint": "distilbert_lda_100_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": 2.101956605911255,
+      "learning_rate": 5e-06,
+      "loss": 13.6088,
+      "step": 500
+    },
+    {
+      "epoch": 0.04215318467310205,
+      "grad_norm": 0.9518340826034546,
+      "learning_rate": 1e-05,
+      "loss": 11.2759,
+      "step": 1000
+    },
+    {
+      "epoch": 0.06322977700965308,
+      "grad_norm": 0.9004790186882019,
+      "learning_rate": 1.5e-05,
+      "loss": 10.6487,
+      "step": 1500
+    },
+    {
+      "epoch": 0.0843063693462041,
+      "grad_norm": 1.1118971109390259,
+      "learning_rate": 2e-05,
+      "loss": 10.5165,
+      "step": 2000
+    },
+    {
+      "epoch": 0.10538296168275513,
+      "grad_norm": 0.8737998008728027,
+      "learning_rate": 2.5e-05,
+      "loss": 10.4509,
+      "step": 2500
+    },
+    {
+      "epoch": 0.12645955401930617,
+      "grad_norm": 0.8549145460128784,
+      "learning_rate": 3e-05,
+      "loss": 10.4085,
+      "step": 3000
+    },
+    {
+      "epoch": 0.14753614635585718,
+      "grad_norm": 0.863103985786438,
+      "learning_rate": 3.5e-05,
+      "loss": 10.3795,
+      "step": 3500
+    },
+    {
+      "epoch": 0.1686127386924082,
+      "grad_norm": 0.8508085012435913,
+      "learning_rate": 4e-05,
+      "loss": 10.3529,
+      "step": 4000
+    },
+    {
+      "epoch": 0.18968933102895924,
+      "grad_norm": 0.8150739669799805,
+      "learning_rate": 4.5e-05,
+      "loss": 10.3276,
+      "step": 4500
+    },
+    {
+      "epoch": 0.21076592336551025,
+      "grad_norm": 0.8614461421966553,
+      "learning_rate": 5e-05,
+      "loss": 10.3106,
+      "step": 5000
+    },
+    {
+      "epoch": 0.2318425157020613,
+      "grad_norm": 0.8280836343765259,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 10.3004,
+      "step": 5500
+    },
+    {
+      "epoch": 0.25291910803861234,
+      "grad_norm": 0.8030481934547424,
+      "learning_rate": 6e-05,
+      "loss": 10.282,
+      "step": 6000
+    },
+    {
+      "epoch": 0.27399570037516335,
+      "grad_norm": 0.7611499428749084,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 10.276,
+      "step": 6500
+    },
+    {
+      "epoch": 0.29507229271171437,
+      "grad_norm": 0.7952233552932739,
+      "learning_rate": 7e-05,
+      "loss": 10.2617,
+      "step": 7000
+    },
+    {
+      "epoch": 0.3161488850482654,
+      "grad_norm": 0.7696254849433899,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 10.2511,
+      "step": 7500
+    },
+    {
+      "epoch": 0.3372254773848164,
+      "grad_norm": 0.8051698207855225,
+      "learning_rate": 8e-05,
+      "loss": 10.2469,
+      "step": 8000
+    },
+    {
+      "epoch": 0.35830206972136747,
+      "grad_norm": 0.8747822046279907,
+      "learning_rate": 8.5e-05,
+      "loss": 10.2181,
+      "step": 8500
+    },
+    {
+      "epoch": 0.3793786620579185,
+      "grad_norm": 0.8448159098625183,
+      "learning_rate": 9e-05,
+      "loss": 10.1808,
+      "step": 9000
+    },
+    {
+      "epoch": 0.4004552543944695,
+      "grad_norm": 0.9002806544303894,
+      "learning_rate": 9.5e-05,
+      "loss": 10.0331,
+      "step": 9500
+    },
+    {
+      "epoch": 0.4215318467310205,
+      "grad_norm": 0.944429874420166,
+      "learning_rate": 0.0001,
+      "loss": 9.8243,
+      "step": 10000
+    },
+    {
+      "epoch": 0.4215318467310205,
+      "eval_accuracy": 0.16385110061035926,
+      "eval_loss": 9.549203872680664,
+      "eval_runtime": 341.6694,
+      "eval_samples_per_second": 353.283,
+      "eval_steps_per_second": 3.682,
+      "step": 10000
+    },
+    {
+      "epoch": 0.4426084390675716,
+      "grad_norm": 0.981827974319458,
+      "learning_rate": 9.99142477382841e-05,
+      "loss": 9.6709,
+      "step": 10500
+    },
+    {
+      "epoch": 0.4636850314041226,
+      "grad_norm": 1.0043810606002808,
+      "learning_rate": 9.98284954765682e-05,
+      "loss": 9.5706,
+      "step": 11000
+    },
+    {
+      "epoch": 0.4847616237406736,
+      "grad_norm": 1.2471486330032349,
+      "learning_rate": 9.97427432148523e-05,
+      "loss": 9.4562,
+      "step": 11500
+    },
+    {
+      "epoch": 0.5058382160772247,
+      "grad_norm": 1.2650797367095947,
+      "learning_rate": 9.96569909531364e-05,
+      "loss": 9.3304,
+      "step": 12000
+    },
+    {
+      "epoch": 0.5269148084137757,
+      "grad_norm": 1.5970815420150757,
+      "learning_rate": 9.957123869142048e-05,
+      "loss": 9.2181,
+      "step": 12500
+    },
+    {
+      "epoch": 0.5479914007503267,
+      "grad_norm": 1.9574729204177856,
+      "learning_rate": 9.94854864297046e-05,
+      "loss": 8.9941,
+      "step": 13000
+    },
+    {
+      "epoch": 0.5690679930868777,
+      "grad_norm": 1.838077187538147,
+      "learning_rate": 9.939973416798869e-05,
+      "loss": 8.3777,
+      "step": 13500
+    },
+    {
+      "epoch": 0.5901445854234287,
+      "grad_norm": 1.7170852422714233,
+      "learning_rate": 9.931398190627278e-05,
+      "loss": 7.9157,
+      "step": 14000
+    },
+    {
+      "epoch": 0.6112211777599798,
+      "grad_norm": 1.5859240293502808,
+      "learning_rate": 9.922822964455688e-05,
+      "loss": 7.5137,
+      "step": 14500
+    },
+    {
+      "epoch": 0.6322977700965308,
+      "grad_norm": 1.493490219116211,
+      "learning_rate": 9.914247738284099e-05,
+      "loss": 7.2651,
+      "step": 15000
+    },
+    {
+      "epoch": 0.6533743624330818,
+      "grad_norm": 1.4224799871444702,
+      "learning_rate": 9.905672512112507e-05,
+      "loss": 7.094,
+      "step": 15500
+    },
+    {
+      "epoch": 0.6744509547696328,
+      "grad_norm": 1.4037384986877441,
+      "learning_rate": 9.897097285940917e-05,
+      "loss": 6.961,
+      "step": 16000
+    },
+    {
+      "epoch": 0.6955275471061839,
+      "grad_norm": 1.419514775276184,
+      "learning_rate": 9.888522059769327e-05,
+      "loss": 6.8491,
+      "step": 16500
+    },
+    {
+      "epoch": 0.7166041394427349,
+      "grad_norm": 1.372401475906372,
+      "learning_rate": 9.879946833597737e-05,
+      "loss": 6.7638,
+      "step": 17000
+    },
+    {
+      "epoch": 0.737680731779286,
+      "grad_norm": 1.3918890953063965,
+      "learning_rate": 9.871371607426147e-05,
+      "loss": 6.6919,
+      "step": 17500
+    },
+    {
+      "epoch": 0.758757324115837,
+      "grad_norm": 1.35601007938385,
+      "learning_rate": 9.862796381254556e-05,
+      "loss": 6.6352,
+      "step": 18000
+    },
+    {
+      "epoch": 0.779833916452388,
+      "grad_norm": 1.33617103099823,
+      "learning_rate": 9.854221155082965e-05,
+      "loss": 6.5726,
+      "step": 18500
+    },
+    {
+      "epoch": 0.800910508788939,
+      "grad_norm": 1.3643242120742798,
+      "learning_rate": 9.845645928911376e-05,
+      "loss": 6.5236,
+      "step": 19000
+    },
+    {
+      "epoch": 0.82198710112549,
+      "grad_norm": 1.3792543411254883,
+      "learning_rate": 9.837070702739786e-05,
+      "loss": 6.4664,
+      "step": 19500
+    },
+    {
+      "epoch": 0.843063693462041,
+      "grad_norm": 1.3003557920455933,
+      "learning_rate": 9.828495476568194e-05,
+      "loss": 6.4155,
+      "step": 20000
+    },
+    {
+      "epoch": 0.843063693462041,
+      "eval_accuracy": 0.5574733179831384,
+      "eval_loss": 6.016814231872559,
+      "eval_runtime": 306.3364,
+      "eval_samples_per_second": 394.031,
+      "eval_steps_per_second": 4.107,
+      "step": 20000
+    },
+    {
+      "epoch": 0.864140285798592,
+      "grad_norm": 1.4122759103775024,
+      "learning_rate": 9.819920250396604e-05,
+      "loss": 6.3688,
+      "step": 20500
+    },
+    {
+      "epoch": 0.8852168781351432,
+      "grad_norm": 1.392291784286499,
+      "learning_rate": 9.811345024225015e-05,
+      "loss": 6.3288,
+      "step": 21000
+    },
+    {
+      "epoch": 0.9062934704716942,
+      "grad_norm": 1.3219914436340332,
+      "learning_rate": 9.802769798053424e-05,
+      "loss": 6.2974,
+      "step": 21500
+    },
+    {
+      "epoch": 0.9273700628082452,
+      "grad_norm": 1.3590742349624634,
+      "learning_rate": 9.794194571881834e-05,
+      "loss": 6.2544,
+      "step": 22000
+    },
+    {
+      "epoch": 0.9484466551447962,
+      "grad_norm": 1.3937479257583618,
+      "learning_rate": 9.785619345710244e-05,
+      "loss": 6.2266,
+      "step": 22500
+    },
+    {
+      "epoch": 0.9695232474813472,
+      "grad_norm": 1.3732026815414429,
+      "learning_rate": 9.777044119538653e-05,
+      "loss": 6.1947,
+      "step": 23000
+    },
+    {
+      "epoch": 0.9905998398178982,
+      "grad_norm": 1.4210565090179443,
+      "learning_rate": 9.768468893367063e-05,
+      "loss": 6.1715,
+      "step": 23500
+    },
+    {
+      "epoch": 1.0116764321544494,
+      "grad_norm": 1.4192605018615723,
+      "learning_rate": 9.759893667195473e-05,
+      "loss": 6.1459,
+      "step": 24000
+    },
+    {
+      "epoch": 1.0327530244910004,
+      "grad_norm": 1.3620891571044922,
+      "learning_rate": 9.751318441023883e-05,
+      "loss": 6.1221,
+      "step": 24500
+    },
+    {
+      "epoch": 1.0538296168275514,
+      "grad_norm": 1.3885345458984375,
+      "learning_rate": 9.742743214852293e-05,
+      "loss": 6.0997,
+      "step": 25000
+    },
+    {
+      "epoch": 1.0749062091641024,
+      "grad_norm": 1.2971861362457275,
+      "learning_rate": 9.734167988680703e-05,
+      "loss": 6.0774,
+      "step": 25500
+    },
+    {
+      "epoch": 1.0959828015006534,
+      "grad_norm": 1.320605993270874,
+      "learning_rate": 9.725592762509111e-05,
+      "loss": 6.056,
+      "step": 26000
+    },
+    {
+      "epoch": 1.1170593938372044,
+      "grad_norm": 1.330433964729309,
+      "learning_rate": 9.717017536337522e-05,
+      "loss": 6.0351,
+      "step": 26500
+    },
+    {
+      "epoch": 1.1381359861737554,
+      "grad_norm": 1.3301512002944946,
+      "learning_rate": 9.708442310165932e-05,
+      "loss": 6.0243,
+      "step": 27000
+    },
+    {
+      "epoch": 1.1592125785103065,
+      "grad_norm": 1.3334851264953613,
+      "learning_rate": 9.69986708399434e-05,
+      "loss": 6.0022,
+      "step": 27500
+    },
+    {
+      "epoch": 1.1802891708468575,
+      "grad_norm": 1.3306342363357544,
+      "learning_rate": 9.69129185782275e-05,
+      "loss": 5.991,
+      "step": 28000
+    },
+    {
+      "epoch": 1.2013657631834085,
+      "grad_norm": 1.3352630138397217,
+      "learning_rate": 9.68271663165116e-05,
+      "loss": 5.9685,
+      "step": 28500
+    },
+    {
+      "epoch": 1.2224423555199595,
+      "grad_norm": 1.3466356992721558,
+      "learning_rate": 9.67414140547957e-05,
+      "loss": 5.9552,
+      "step": 29000
+    },
+    {
+      "epoch": 1.2435189478565105,
+      "grad_norm": 1.335174798965454,
+      "learning_rate": 9.66556617930798e-05,
+      "loss": 5.9342,
+      "step": 29500
+    },
+    {
+      "epoch": 1.2645955401930615,
+      "grad_norm": 1.304030179977417,
+      "learning_rate": 9.656990953136388e-05,
+      "loss": 5.9278,
+      "step": 30000
+    },
+    {
+      "epoch": 1.2645955401930615,
+      "eval_accuracy": 0.6084669987082292,
+      "eval_loss": 5.591890811920166,
+      "eval_runtime": 363.6565,
+      "eval_samples_per_second": 331.923,
+      "eval_steps_per_second": 3.459,
+      "step": 30000
+    },
+    {
+      "epoch": 1.2856721325296125,
+      "grad_norm": 1.3268048763275146,
+      "learning_rate": 9.6484157269648e-05,
+      "loss": 5.9092,
+      "step": 30500
+    },
+    {
+      "epoch": 1.3067487248661636,
+      "grad_norm": 1.3600881099700928,
+      "learning_rate": 9.639840500793209e-05,
+      "loss": 5.9029,
+      "step": 31000
+    },
+    {
+      "epoch": 1.3278253172027146,
+      "grad_norm": 1.3397537469863892,
+      "learning_rate": 9.631265274621618e-05,
+      "loss": 5.8866,
+      "step": 31500
+    },
+    {
+      "epoch": 1.3489019095392658,
+      "grad_norm": 1.3418998718261719,
+      "learning_rate": 9.622690048450028e-05,
+      "loss": 5.8815,
+      "step": 32000
+    },
+    {
+      "epoch": 1.3699785018758166,
+      "grad_norm": 1.293783187866211,
+      "learning_rate": 9.614114822278439e-05,
+      "loss": 5.8671,
+      "step": 32500
+    },
+    {
+      "epoch": 1.3910550942123678,
+      "grad_norm": 1.3485147953033447,
+      "learning_rate": 9.605539596106847e-05,
+      "loss": 5.8589,
+      "step": 33000
+    },
+    {
+      "epoch": 1.4121316865489189,
+      "grad_norm": 1.2696123123168945,
+      "learning_rate": 9.596964369935257e-05,
+      "loss": 5.8442,
+      "step": 33500
+    },
+    {
+      "epoch": 1.4332082788854699,
+      "grad_norm": 1.3595514297485352,
+      "learning_rate": 9.588389143763667e-05,
+      "loss": 5.8378,
+      "step": 34000
+    },
+    {
+      "epoch": 1.4542848712220209,
+      "grad_norm": 1.3014330863952637,
+      "learning_rate": 9.579813917592077e-05,
+      "loss": 5.8254,
+      "step": 34500
+    },
+    {
+      "epoch": 1.475361463558572,
+      "grad_norm": 1.324857473373413,
+      "learning_rate": 9.571238691420487e-05,
+      "loss": 5.8171,
+      "step": 35000
+    },
+    {
+      "epoch": 1.496438055895123,
+      "grad_norm": 1.3223344087600708,
+      "learning_rate": 9.562663465248896e-05,
+      "loss": 5.8044,
+      "step": 35500
+    },
+    {
+      "epoch": 1.517514648231674,
+      "grad_norm": 1.321217656135559,
+      "learning_rate": 9.554088239077306e-05,
+      "loss": 5.7898,
+      "step": 36000
+    },
+    {
+      "epoch": 1.538591240568225,
+      "grad_norm": 1.2716072797775269,
+      "learning_rate": 9.545513012905716e-05,
+      "loss": 5.7843,
+      "step": 36500
+    },
+    {
+      "epoch": 1.559667832904776,
+      "grad_norm": 1.2983591556549072,
+      "learning_rate": 9.536937786734126e-05,
+      "loss": 5.7817,
+      "step": 37000
+    },
+    {
+      "epoch": 1.580744425241327,
+      "grad_norm": 1.3365631103515625,
+      "learning_rate": 9.528362560562534e-05,
+      "loss": 5.7682,
+      "step": 37500
+    },
+    {
+      "epoch": 1.601821017577878,
+      "grad_norm": 1.3221347332000732,
+      "learning_rate": 9.519787334390946e-05,
+      "loss": 5.7593,
+      "step": 38000
+    },
+    {
+      "epoch": 1.622897609914429,
+      "grad_norm": 1.3225332498550415,
+      "learning_rate": 9.511212108219355e-05,
+      "loss": 5.7504,
+      "step": 38500
+    },
+    {
+      "epoch": 1.64397420225098,
+      "grad_norm": 1.3258090019226074,
+      "learning_rate": 9.502636882047764e-05,
+      "loss": 5.7426,
+      "step": 39000
+    },
+    {
+      "epoch": 1.6650507945875312,
+      "grad_norm": 1.3410358428955078,
+      "learning_rate": 9.494061655876174e-05,
+      "loss": 5.7353,
+      "step": 39500
+    },
+    {
+      "epoch": 1.686127386924082,
+      "grad_norm": 1.2656409740447998,
+      "learning_rate": 9.485486429704585e-05,
+      "loss": 5.7301,
+      "step": 40000
+    },
+    {
+      "epoch": 1.686127386924082,
+      "eval_accuracy": 0.6304159620245271,
+      "eval_loss": 5.405135631561279,
+      "eval_runtime": 377.2394,
+      "eval_samples_per_second": 319.972,
+      "eval_steps_per_second": 3.335,
+      "step": 40000
+    },
+    {
+      "epoch": 1.7072039792606333,
+      "grad_norm": 1.3077281713485718,
+      "learning_rate": 9.476911203532993e-05,
+      "loss": 5.7199,
+      "step": 40500
+    },
+    {
+      "epoch": 1.728280571597184,
+      "grad_norm": 1.2842686176300049,
+      "learning_rate": 9.468335977361403e-05,
+      "loss": 5.7152,
+      "step": 41000
+    },
+    {
+      "epoch": 1.7493571639337353,
+      "grad_norm": 1.2855496406555176,
+      "learning_rate": 9.459760751189813e-05,
+      "loss": 5.7089,
+      "step": 41500
+    },
+    {
+      "epoch": 1.770433756270286,
+      "grad_norm": 1.3180679082870483,
+      "learning_rate": 9.451185525018223e-05,
+      "loss": 5.6991,
+      "step": 42000
+    },
+    {
+      "epoch": 1.7915103486068373,
+      "grad_norm": 1.33673894405365,
+      "learning_rate": 9.442610298846633e-05,
+      "loss": 5.6915,
+      "step": 42500
+    },
+    {
+      "epoch": 1.8125869409433881,
+      "grad_norm": 1.321456789970398,
+      "learning_rate": 9.434035072675043e-05,
+      "loss": 5.685,
+      "step": 43000
+    },
+    {
+      "epoch": 1.8336635332799394,
+      "grad_norm": 1.3071925640106201,
+      "learning_rate": 9.425459846503451e-05,
+      "loss": 5.6748,
+      "step": 43500
+    },
+    {
+      "epoch": 1.8547401256164904,
+      "grad_norm": 1.3366185426712036,
+      "learning_rate": 9.416884620331862e-05,
+      "loss": 5.6693,
+      "step": 44000
+    },
+    {
+      "epoch": 1.8758167179530414,
+      "grad_norm": 1.3000937700271606,
+      "learning_rate": 9.408309394160272e-05,
+      "loss": 5.6679,
+      "step": 44500
+    },
+    {
+      "epoch": 1.8968933102895924,
+      "grad_norm": 1.3357912302017212,
+      "learning_rate": 9.39973416798868e-05,
+      "loss": 5.6548,
+      "step": 45000
+    },
+    {
+      "epoch": 1.9179699026261434,
+      "grad_norm": 1.2905248403549194,
+      "learning_rate": 9.39115894181709e-05,
+      "loss": 5.6527,
+      "step": 45500
+    },
+    {
+      "epoch": 1.9390464949626944,
+      "grad_norm": 1.3013639450073242,
+      "learning_rate": 9.382583715645502e-05,
+      "loss": 5.6409,
+      "step": 46000
+    },
+    {
+      "epoch": 1.9601230872992454,
+      "grad_norm": 1.301288366317749,
+      "learning_rate": 9.37400848947391e-05,
+      "loss": 5.6346,
+      "step": 46500
+    },
+    {
+      "epoch": 1.9811996796357965,
+      "grad_norm": 1.355980396270752,
+      "learning_rate": 9.36543326330232e-05,
+      "loss": 5.6356,
+      "step": 47000
+    },
+    {
+      "epoch": 2.0022762719723475,
+      "grad_norm": 1.3759366273880005,
+      "learning_rate": 9.35685803713073e-05,
+      "loss": 5.6326,
+      "step": 47500
+    },
+    {
+      "epoch": 2.0233528643088987,
+      "grad_norm": 1.296671748161316,
+      "learning_rate": 9.34828281095914e-05,
+      "loss": 5.6141,
+      "step": 48000
+    },
+    {
+      "epoch": 2.0444294566454495,
+      "grad_norm": 1.284118413925171,
+      "learning_rate": 9.339707584787549e-05,
+      "loss": 5.611,
+      "step": 48500
+    },
+    {
+      "epoch": 2.0655060489820007,
+      "grad_norm": 1.3254456520080566,
+      "learning_rate": 9.331132358615959e-05,
+      "loss": 5.6007,
+      "step": 49000
+    },
+    {
+      "epoch": 2.0865826413185515,
+      "grad_norm": 1.2984586954116821,
+      "learning_rate": 9.322557132444369e-05,
+      "loss": 5.5939,
+      "step": 49500
+    },
+    {
+      "epoch": 2.1076592336551028,
+      "grad_norm": 1.3202879428863525,
+      "learning_rate": 9.313981906272779e-05,
+      "loss": 5.5965,
+      "step": 50000
+    },
+    {
+      "epoch": 2.1076592336551028,
+      "eval_accuracy": 0.6430722031501053,
+      "eval_loss": 5.280669689178467,
+      "eval_runtime": 392.3739,
+      "eval_samples_per_second": 307.63,
+      "eval_steps_per_second": 3.206,
+      "step": 50000
+    },
+    {
+      "epoch": 2.1287358259916536,
+      "grad_norm": 1.30453622341156,
+      "learning_rate": 9.305406680101189e-05,
+      "loss": 5.5921,
+      "step": 50500
+    },
+    {
+      "epoch": 2.149812418328205,
+      "grad_norm": 1.3594733476638794,
+      "learning_rate": 9.296831453929597e-05,
+      "loss": 5.5832,
+      "step": 51000
+    },
+    {
+      "epoch": 2.1708890106647556,
+      "grad_norm": 1.293358325958252,
+      "learning_rate": 9.288256227758008e-05,
+      "loss": 5.5769,
+      "step": 51500
+    },
+    {
+      "epoch": 2.191965603001307,
+      "grad_norm": 1.3217575550079346,
+      "learning_rate": 9.279681001586418e-05,
+      "loss": 5.5738,
+      "step": 52000
+    },
+    {
+      "epoch": 2.2130421953378576,
+      "grad_norm": 1.312394618988037,
+      "learning_rate": 9.271105775414827e-05,
+      "loss": 5.5638,
+      "step": 52500
+    },
+    {
+      "epoch": 2.234118787674409,
+      "grad_norm": 1.3103684186935425,
+      "learning_rate": 9.262530549243236e-05,
+      "loss": 5.5666,
+      "step": 53000
+    },
+    {
+      "epoch": 2.2551953800109596,
+      "grad_norm": 1.289625883102417,
+      "learning_rate": 9.253955323071646e-05,
+      "loss": 5.5611,
+      "step": 53500
+    },
+    {
+      "epoch": 2.276271972347511,
+      "grad_norm": 1.351354718208313,
+      "learning_rate": 9.245380096900056e-05,
+      "loss": 5.5569,
+      "step": 54000
+    },
+    {
+      "epoch": 2.2973485646840617,
+      "grad_norm": 1.2922601699829102,
+      "learning_rate": 9.236804870728466e-05,
+      "loss": 5.544,
+      "step": 54500
+    },
+    {
+      "epoch": 2.318425157020613,
+      "grad_norm": 1.3493670225143433,
+      "learning_rate": 9.228229644556876e-05,
+      "loss": 5.5467,
+      "step": 55000
+    },
+    {
+      "epoch": 2.339501749357164,
+      "grad_norm": 1.342743158340454,
+      "learning_rate": 9.219654418385286e-05,
+      "loss": 5.5343,
+      "step": 55500
+    },
+    {
+      "epoch": 2.360578341693715,
+      "grad_norm": 1.3256583213806152,
+      "learning_rate": 9.211079192213695e-05,
+      "loss": 5.5395,
+      "step": 56000
+    },
+    {
+      "epoch": 2.381654934030266,
+      "grad_norm": 1.3172714710235596,
+      "learning_rate": 9.202503966042105e-05,
+      "loss": 5.5322,
+      "step": 56500
+    },
+    {
+      "epoch": 2.402731526366817,
+      "grad_norm": 1.317274570465088,
+      "learning_rate": 9.193928739870514e-05,
+      "loss": 5.5242,
+      "step": 57000
+    },
+    {
+      "epoch": 2.423808118703368,
+      "grad_norm": 1.2736917734146118,
+      "learning_rate": 9.185353513698925e-05,
+      "loss": 5.5176,
+      "step": 57500
+    },
+    {
+      "epoch": 2.444884711039919,
+      "grad_norm": 1.3049134016036987,
+      "learning_rate": 9.176778287527335e-05,
+      "loss": 5.5168,
+      "step": 58000
+    },
+    {
+      "epoch": 2.4659613033764702,
+      "grad_norm": 1.3689264059066772,
+      "learning_rate": 9.168203061355743e-05,
+      "loss": 5.514,
+      "step": 58500
+    },
+    {
+      "epoch": 2.487037895713021,
+      "grad_norm": 1.3105288743972778,
+      "learning_rate": 9.159627835184153e-05,
+      "loss": 5.5088,
+      "step": 59000
+    },
+    {
+      "epoch": 2.5081144880495723,
+      "grad_norm": 1.2810420989990234,
+      "learning_rate": 9.151052609012564e-05,
+      "loss": 5.5059,
+      "step": 59500
+    },
+    {
+      "epoch": 2.529191080386123,
+      "grad_norm": 1.2867431640625,
+      "learning_rate": 9.142477382840973e-05,
+      "loss": 5.5012,
+      "step": 60000
+    },
+    {
+      "epoch": 2.529191080386123,
+      "eval_accuracy": 0.6540088263584881,
+      "eval_loss": 5.183504581451416,
+      "eval_runtime": 385.3369,
+      "eval_samples_per_second": 313.248,
+      "eval_steps_per_second": 3.265,
+      "step": 60000
+    },
+    {
+      "epoch": 2.5502676727226743,
+      "grad_norm": 1.279603362083435,
+      "learning_rate": 9.133902156669383e-05,
+      "loss": 5.496,
+      "step": 60500
+    },
+    {
+      "epoch": 2.571344265059225,
+      "grad_norm": 1.2801570892333984,
+      "learning_rate": 9.125326930497792e-05,
+      "loss": 5.4922,
+      "step": 61000
+    },
+    {
+      "epoch": 2.5924208573957763,
+      "grad_norm": 1.333313226699829,
+      "learning_rate": 9.116751704326202e-05,
+      "loss": 5.4867,
+      "step": 61500
+    },
+    {
+      "epoch": 2.613497449732327,
+      "grad_norm": 1.309173583984375,
+      "learning_rate": 9.108176478154612e-05,
+      "loss": 5.4798,
+      "step": 62000
+    },
+    {
+      "epoch": 2.6345740420688784,
+      "grad_norm": 1.3271788358688354,
+      "learning_rate": 9.09960125198302e-05,
+      "loss": 5.4864,
+      "step": 62500
+    },
+    {
+      "epoch": 2.655650634405429,
+      "grad_norm": 1.305768370628357,
+      "learning_rate": 9.091026025811432e-05,
+      "loss": 5.4754,
+      "step": 63000
+    },
+    {
+      "epoch": 2.6767272267419804,
+      "grad_norm": 1.3215919733047485,
+      "learning_rate": 9.082450799639842e-05,
+      "loss": 5.4733,
+      "step": 63500
+    },
+    {
+      "epoch": 2.6978038190785316,
+      "grad_norm": 1.2726967334747314,
+      "learning_rate": 9.07387557346825e-05,
+      "loss": 5.472,
+      "step": 64000
+    },
+    {
+      "epoch": 2.7188804114150824,
+      "grad_norm": 1.3071924448013306,
+      "learning_rate": 9.06530034729666e-05,
+      "loss": 5.4655,
+      "step": 64500
+    },
+    {
+      "epoch": 2.739957003751633,
+      "grad_norm": 1.2841559648513794,
+      "learning_rate": 9.05672512112507e-05,
+      "loss": 5.4641,
+      "step": 65000
+    },
+    {
+      "epoch": 2.7610335960881844,
+      "grad_norm": 1.3524171113967896,
+      "learning_rate": 9.04814989495348e-05,
+      "loss": 5.4599,
+      "step": 65500
+    },
+    {
+      "epoch": 2.7821101884247357,
+      "grad_norm": 1.3277137279510498,
+      "learning_rate": 9.03957466878189e-05,
+      "loss": 5.4602,
+      "step": 66000
+    },
+    {
+      "epoch": 2.8031867807612865,
+      "grad_norm": 1.3051809072494507,
+      "learning_rate": 9.030999442610299e-05,
+      "loss": 5.4522,
+      "step": 66500
+    },
+    {
+      "epoch": 2.8242633730978377,
+      "grad_norm": 1.2728397846221924,
+      "learning_rate": 9.022424216438709e-05,
+      "loss": 5.4467,
+      "step": 67000
+    },
+    {
+      "epoch": 2.8453399654343885,
+      "grad_norm": 1.2872259616851807,
+      "learning_rate": 9.013848990267119e-05,
+      "loss": 5.4437,
+      "step": 67500
+    },
+    {
+      "epoch": 2.8664165577709397,
+      "grad_norm": 1.2879129648208618,
+      "learning_rate": 9.005273764095529e-05,
+      "loss": 5.439,
+      "step": 68000
+    },
+    {
+      "epoch": 2.8874931501074905,
+      "grad_norm": 1.3541512489318848,
+      "learning_rate": 8.996698537923937e-05,
+      "loss": 5.442,
+      "step": 68500
+    },
+    {
+      "epoch": 2.9085697424440418,
+      "grad_norm": 1.3692233562469482,
+      "learning_rate": 8.988123311752348e-05,
+      "loss": 5.4336,
+      "step": 69000
+    },
+    {
+      "epoch": 2.9296463347805926,
+      "grad_norm": 1.3255224227905273,
+      "learning_rate": 8.979548085580758e-05,
+      "loss": 5.4349,
+      "step": 69500
+    },
+    {
+      "epoch": 2.950722927117144,
+      "grad_norm": 1.261643886566162,
+      "learning_rate": 8.970972859409167e-05,
+      "loss": 5.4266,
+      "step": 70000
+    },
+    {
+      "epoch": 2.950722927117144,
+      "eval_accuracy": 0.6610655548689939,
+      "eval_loss": 5.124541759490967,
+      "eval_runtime": 551.9042,
+      "eval_samples_per_second": 218.708,
+      "eval_steps_per_second": 2.279,
+      "step": 70000
+    },
+    {
+      "epoch": 2.9717995194536946,
+      "grad_norm": 1.3088123798370361,
+      "learning_rate": 8.962397633237576e-05,
+      "loss": 5.4293,
+      "step": 70500
+    },
+    {
+      "epoch": 2.992876111790246,
+      "grad_norm": 1.2815210819244385,
+      "learning_rate": 8.953822407065988e-05,
+      "loss": 5.4233,
+      "step": 71000
+    },
+    {
+      "epoch": 3.0139527041267966,
+      "grad_norm": 1.3185235261917114,
+      "learning_rate": 8.945247180894396e-05,
+      "loss": 5.4187,
+      "step": 71500
+    },
+    {
+      "epoch": 3.035029296463348,
+      "grad_norm": 1.2827225923538208,
+      "learning_rate": 8.936671954722806e-05,
+      "loss": 5.4156,
+      "step": 72000
+    },
+    {
+      "epoch": 3.0561058887998986,
+      "grad_norm": 1.2683727741241455,
+      "learning_rate": 8.928096728551216e-05,
+      "loss": 5.4142,
+      "step": 72500
+    },
+    {
+      "epoch": 3.07718248113645,
+      "grad_norm": 1.2771836519241333,
+      "learning_rate": 8.919521502379626e-05,
+      "loss": 5.4082,
+      "step": 73000
+    },
+    {
+      "epoch": 3.0982590734730007,
+      "grad_norm": 1.3184714317321777,
+      "learning_rate": 8.910946276208035e-05,
+      "loss": 5.411,
+      "step": 73500
+    },
+    {
+      "epoch": 3.119335665809552,
+      "grad_norm": 1.2990520000457764,
+      "learning_rate": 8.902371050036445e-05,
+      "loss": 5.3989,
+      "step": 74000
+    },
+    {
+      "epoch": 3.140412258146103,
+      "grad_norm": 1.2541097402572632,
+      "learning_rate": 8.893795823864855e-05,
+      "loss": 5.4036,
+      "step": 74500
+    },
+    {
+      "epoch": 3.161488850482654,
+      "grad_norm": 1.2509688138961792,
+      "learning_rate": 8.885220597693265e-05,
+      "loss": 5.401,
+      "step": 75000
+    },
+    {
+      "epoch": 3.182565442819205,
+      "grad_norm": 1.322439193725586,
+      "learning_rate": 8.876645371521675e-05,
+      "loss": 5.3962,
+      "step": 75500
+    },
+    {
+      "epoch": 3.203642035155756,
+      "grad_norm": 1.2941162586212158,
+      "learning_rate": 8.868070145350083e-05,
+      "loss": 5.395,
+      "step": 76000
+    },
+    {
+      "epoch": 3.224718627492307,
+      "grad_norm": 1.3247716426849365,
+      "learning_rate": 8.859494919178493e-05,
+      "loss": 5.3958,
+      "step": 76500
+    },
+    {
+      "epoch": 3.245795219828858,
+      "grad_norm": 1.315844178199768,
+      "learning_rate": 8.850919693006904e-05,
+      "loss": 5.3893,
+      "step": 77000
+    },
+    {
+      "epoch": 3.2668718121654092,
+      "grad_norm": 1.3192059993743896,
+      "learning_rate": 8.842344466835313e-05,
+      "loss": 5.3863,
+      "step": 77500
+    },
+    {
+      "epoch": 3.28794840450196,
+      "grad_norm": 1.2718379497528076,
+      "learning_rate": 8.833769240663723e-05,
+      "loss": 5.3823,
+      "step": 78000
+    },
+    {
+      "epoch": 3.3090249968385113,
+      "grad_norm": 1.3503544330596924,
+      "learning_rate": 8.825194014492132e-05,
+      "loss": 5.3824,
+      "step": 78500
+    },
+    {
+      "epoch": 3.330101589175062,
+      "grad_norm": 1.3278894424438477,
+      "learning_rate": 8.816618788320542e-05,
+      "loss": 5.3816,
+      "step": 79000
+    },
+    {
+      "epoch": 3.3511781815116133,
+      "grad_norm": 1.3233107328414917,
+      "learning_rate": 8.808043562148952e-05,
+      "loss": 5.3787,
+      "step": 79500
+    },
+    {
+      "epoch": 3.372254773848164,
+      "grad_norm": 1.2931227684020996,
+      "learning_rate": 8.799468335977362e-05,
+      "loss": 5.3773,
+      "step": 80000
+    },
+    {
+      "epoch": 3.372254773848164,
+      "eval_accuracy": 0.6675647649374604,
+      "eval_loss": 5.0741658210754395,
+      "eval_runtime": 383.9834,
+      "eval_samples_per_second": 314.352,
+      "eval_steps_per_second": 3.276,
+      "step": 80000
+    },
+    {
+      "epoch": 3.3933313661847153,
+      "grad_norm": 1.275254249572754,
+      "learning_rate": 8.790893109805772e-05,
+      "loss": 5.3757,
+      "step": 80500
+    },
+    {
+      "epoch": 3.414407958521266,
+      "grad_norm": 1.2292425632476807,
+      "learning_rate": 8.782317883634182e-05,
+      "loss": 5.3714,
+      "step": 81000
+    },
+    {
+      "epoch": 3.4354845508578173,
+      "grad_norm": 1.2697803974151611,
+      "learning_rate": 8.773742657462591e-05,
+      "loss": 5.3681,
+      "step": 81500
+    },
+    {
+      "epoch": 3.456561143194368,
+      "grad_norm": 1.3429304361343384,
+      "learning_rate": 8.765167431291e-05,
+      "loss": 5.3671,
+      "step": 82000
+    },
+    {
+      "epoch": 3.4776377355309194,
+      "grad_norm": 1.3328132629394531,
+      "learning_rate": 8.756592205119411e-05,
+      "loss": 5.3649,
+      "step": 82500
+    },
+    {
+      "epoch": 3.4987143278674706,
+      "grad_norm": 1.2718851566314697,
+      "learning_rate": 8.748016978947821e-05,
+      "loss": 5.3633,
+      "step": 83000
+    },
+    {
+      "epoch": 3.5197909202040214,
+      "grad_norm": 1.2901719808578491,
+      "learning_rate": 8.73944175277623e-05,
+      "loss": 5.3618,
+      "step": 83500
+    },
+    {
+      "epoch": 3.540867512540572,
+      "grad_norm": 1.2861982583999634,
+      "learning_rate": 8.730866526604639e-05,
+      "loss": 5.3627,
+      "step": 84000
+    },
+    {
+      "epoch": 3.5619441048771234,
+      "grad_norm": 1.2919082641601562,
+      "learning_rate": 8.72229130043305e-05,
+      "loss": 5.3601,
+      "step": 84500
+    },
+    {
+      "epoch": 3.5830206972136747,
+      "grad_norm": 1.290877342224121,
+      "learning_rate": 8.713716074261459e-05,
+      "loss": 5.3581,
+      "step": 85000
+    },
+    {
+      "epoch": 3.6040972895502255,
+      "grad_norm": 1.2808139324188232,
+      "learning_rate": 8.705140848089869e-05,
+      "loss": 5.3559,
+      "step": 85500
+    },
+    {
+      "epoch": 3.6251738818867767,
+      "grad_norm": 1.334856390953064,
+      "learning_rate": 8.696565621918278e-05,
+      "loss": 5.349,
+      "step": 86000
+    },
+    {
+      "epoch": 3.6462504742233275,
+      "grad_norm": 1.2683329582214355,
+      "learning_rate": 8.687990395746688e-05,
+      "loss": 5.3504,
+      "step": 86500
+    },
+    {
+      "epoch": 3.6673270665598787,
+      "grad_norm": 1.266661524772644,
+      "learning_rate": 8.679415169575098e-05,
+      "loss": 5.3464,
+      "step": 87000
+    },
+    {
+      "epoch": 3.6884036588964295,
+      "grad_norm": 1.304843544960022,
+      "learning_rate": 8.670839943403508e-05,
+      "loss": 5.3454,
+      "step": 87500
+    },
+    {
+      "epoch": 3.7094802512329808,
+      "grad_norm": 1.3002351522445679,
+      "learning_rate": 8.662264717231916e-05,
+      "loss": 5.3457,
+      "step": 88000
+    },
+    {
+      "epoch": 3.7305568435695315,
+      "grad_norm": 1.2976861000061035,
+      "learning_rate": 8.653689491060328e-05,
+      "loss": 5.3443,
+      "step": 88500
+    },
+    {
+      "epoch": 3.751633435906083,
+      "grad_norm": 1.2773562669754028,
+      "learning_rate": 8.645114264888737e-05,
+      "loss": 5.3429,
+      "step": 89000
+    },
+    {
+      "epoch": 3.7727100282426336,
+      "grad_norm": 1.3188040256500244,
+      "learning_rate": 8.636539038717146e-05,
+      "loss": 5.3368,
+      "step": 89500
+    },
+    {
+      "epoch": 3.793786620579185,
+      "grad_norm": 1.3156687021255493,
+      "learning_rate": 8.627963812545556e-05,
+      "loss": 5.3364,
+      "step": 90000
+    },
+    {
+      "epoch": 3.793786620579185,
+      "eval_accuracy": 0.6725612473684944,
+      "eval_loss": 5.032149791717529,
+      "eval_runtime": 385.3348,
+      "eval_samples_per_second": 313.25,
+      "eval_steps_per_second": 3.265,
+      "step": 90000
+    },
+    {
+      "epoch": 3.8148632129157356,
+      "grad_norm": 1.2458521127700806,
+      "learning_rate": 8.619388586373967e-05,
+      "loss": 5.3373,
+      "step": 90500
+    },
+    {
+      "epoch": 3.835939805252287,
+      "grad_norm": 1.2860074043273926,
+      "learning_rate": 8.610813360202375e-05,
+      "loss": 5.3303,
+      "step": 91000
+    },
+    {
+      "epoch": 3.857016397588838,
+      "grad_norm": 1.3438979387283325,
+      "learning_rate": 8.602238134030785e-05,
+      "loss": 5.333,
+      "step": 91500
+    },
+    {
+      "epoch": 3.878092989925389,
+      "grad_norm": 1.3054558038711548,
+      "learning_rate": 8.593662907859195e-05,
+      "loss": 5.3251,
+      "step": 92000
+    },
+    {
+      "epoch": 3.8991695822619397,
+      "grad_norm": 1.2332439422607422,
+      "learning_rate": 8.585087681687605e-05,
+      "loss": 5.3273,
+      "step": 92500
+    },
+    {
+      "epoch": 3.920246174598491,
+      "grad_norm": 1.2935154438018799,
+      "learning_rate": 8.576512455516015e-05,
+      "loss": 5.3255,
+      "step": 93000
+    },
+    {
+      "epoch": 3.941322766935042,
+      "grad_norm": 1.2742094993591309,
+      "learning_rate": 8.567937229344425e-05,
+      "loss": 5.3228,
+      "step": 93500
+    },
+    {
+      "epoch": 3.962399359271593,
+      "grad_norm": 1.308293342590332,
+      "learning_rate": 8.559362003172834e-05,
+      "loss": 5.3205,
+      "step": 94000
+    },
+    {
+      "epoch": 3.983475951608144,
+      "grad_norm": 1.296563982963562,
+      "learning_rate": 8.550786777001244e-05,
+      "loss": 5.3193,
+      "step": 94500
+    },
+    {
+      "epoch": 4.004552543944695,
+      "grad_norm": 1.313360571861267,
+      "learning_rate": 8.542211550829654e-05,
+      "loss": 5.3149,
+      "step": 95000
+    },
+    {
+      "epoch": 4.025629136281246,
+      "grad_norm": 1.2772917747497559,
+      "learning_rate": 8.533636324658063e-05,
+      "loss": 5.3052,
+      "step": 95500
+    },
+    {
+      "epoch": 4.046705728617797,
+      "grad_norm": 1.2967076301574707,
+      "learning_rate": 8.525061098486474e-05,
+      "loss": 5.3109,
+      "step": 96000
+    },
+    {
+      "epoch": 4.067782320954348,
+      "grad_norm": 1.3167554140090942,
+      "learning_rate": 8.516485872314884e-05,
+      "loss": 5.3054,
+      "step": 96500
+    },
+    {
+      "epoch": 4.088858913290899,
+      "grad_norm": 1.3260854482650757,
+      "learning_rate": 8.507910646143292e-05,
+      "loss": 5.3057,
+      "step": 97000
+    },
+    {
+      "epoch": 4.10993550562745,
+      "grad_norm": 1.30555260181427,
+      "learning_rate": 8.499335419971702e-05,
+      "loss": 5.3029,
+      "step": 97500
+    },
+    {
+      "epoch": 4.1310120979640015,
+      "grad_norm": 1.2971004247665405,
+      "learning_rate": 8.490760193800112e-05,
+      "loss": 5.3014,
+      "step": 98000
+    },
+    {
+      "epoch": 4.152088690300552,
+      "grad_norm": 1.310563325881958,
+      "learning_rate": 8.482184967628522e-05,
+      "loss": 5.3006,
+      "step": 98500
+    },
+    {
+      "epoch": 4.173165282637103,
+      "grad_norm": 1.2731282711029053,
+      "learning_rate": 8.473609741456931e-05,
+      "loss": 5.3027,
+      "step": 99000
+    },
+    {
+      "epoch": 4.194241874973654,
+      "grad_norm": 1.3269964456558228,
+      "learning_rate": 8.46503451528534e-05,
+      "loss": 5.3023,
+      "step": 99500
+    },
+    {
+      "epoch": 4.2153184673102055,
+      "grad_norm": 1.2968741655349731,
+      "learning_rate": 8.456459289113751e-05,
+      "loss": 5.2973,
+      "step": 100000
+    },
+    {
+      "epoch": 4.2153184673102055,
+      "eval_accuracy": 0.676740369351985,
+      "eval_loss": 5.004430770874023,
+      "eval_runtime": 388.0006,
+      "eval_samples_per_second": 311.097,
+      "eval_steps_per_second": 3.242,
+      "step": 100000
+    },
+    {
+      "epoch": 4.236395059646756,
+      "grad_norm": 1.3101304769515991,
+      "learning_rate": 8.447884062942161e-05,
+      "loss": 5.2995,
+      "step": 100500
+    },
+    {
+      "epoch": 4.257471651983307,
+      "grad_norm": 1.2825075387954712,
+      "learning_rate": 8.43930883677057e-05,
+      "loss": 5.2995,
+      "step": 101000
+    },
+    {
+      "epoch": 4.278548244319858,
+      "grad_norm": 1.321528673171997,
+      "learning_rate": 8.430733610598979e-05,
+      "loss": 5.2941,
+      "step": 101500
+    },
+    {
+      "epoch": 4.29962483665641,
+      "grad_norm": 1.2923473119735718,
+      "learning_rate": 8.42215838442739e-05,
+      "loss": 5.2922,
+      "step": 102000
+    },
+    {
+      "epoch": 4.32070142899296,
+      "grad_norm": 1.3102103471755981,
+      "learning_rate": 8.413583158255799e-05,
+      "loss": 5.2916,
+      "step": 102500
+    },
+    {
+      "epoch": 4.341778021329511,
+      "grad_norm": 1.260189175605774,
+      "learning_rate": 8.405007932084209e-05,
+      "loss": 5.2886,
+      "step": 103000
+    },
+    {
+      "epoch": 4.362854613666062,
+      "grad_norm": 1.2568172216415405,
+      "learning_rate": 8.396432705912619e-05,
+      "loss": 5.2883,
+      "step": 103500
+    },
+    {
+      "epoch": 4.383931206002614,
+      "grad_norm": 1.28819739818573,
+      "learning_rate": 8.387857479741028e-05,
+      "loss": 5.2918,
+      "step": 104000
+    },
+    {
+      "epoch": 4.405007798339165,
+      "grad_norm": 1.26441490650177,
+      "learning_rate": 8.379282253569438e-05,
+      "loss": 5.2877,
+      "step": 104500
+    },
+    {
+      "epoch": 4.426084390675715,
+      "grad_norm": 1.2674381732940674,
+      "learning_rate": 8.370707027397848e-05,
+      "loss": 5.2849,
+      "step": 105000
+    },
+    {
+      "epoch": 4.4471609830122665,
+      "grad_norm": 1.311007022857666,
+      "learning_rate": 8.362131801226258e-05,
+      "loss": 5.2843,
+      "step": 105500
+    },
+    {
+      "epoch": 4.468237575348818,
+      "grad_norm": 1.296278476715088,
+      "learning_rate": 8.353556575054668e-05,
+      "loss": 5.2807,
+      "step": 106000
+    },
+    {
+      "epoch": 4.489314167685369,
+      "grad_norm": 1.3236894607543945,
+      "learning_rate": 8.344981348883077e-05,
+      "loss": 5.2837,
+      "step": 106500
+    },
+    {
+      "epoch": 4.510390760021919,
+      "grad_norm": 1.3011701107025146,
+      "learning_rate": 8.336406122711486e-05,
+      "loss": 5.2795,
+      "step": 107000
+    },
+    {
+      "epoch": 4.5314673523584705,
+      "grad_norm": 1.2835659980773926,
+      "learning_rate": 8.327830896539897e-05,
+      "loss": 5.2829,
+      "step": 107500
+    },
+    {
+      "epoch": 4.552543944695022,
+      "grad_norm": 1.2994040250778198,
+      "learning_rate": 8.319255670368307e-05,
+      "loss": 5.2777,
+      "step": 108000
+    },
+    {
+      "epoch": 4.573620537031573,
+      "grad_norm": 1.2848464250564575,
+      "learning_rate": 8.310680444196715e-05,
+      "loss": 5.2768,
+      "step": 108500
+    },
+    {
+      "epoch": 4.594697129368123,
+      "grad_norm": 1.2826271057128906,
+      "learning_rate": 8.302105218025125e-05,
+      "loss": 5.2703,
+      "step": 109000
+    },
+    {
+      "epoch": 4.615773721704675,
+      "grad_norm": 1.316697359085083,
+      "learning_rate": 8.293529991853536e-05,
+      "loss": 5.2771,
+      "step": 109500
+    },
+    {
+      "epoch": 4.636850314041226,
+      "grad_norm": 1.3072599172592163,
+      "learning_rate": 8.284954765681945e-05,
+      "loss": 5.2724,
+      "step": 110000
+    },
+    {
+      "epoch": 4.636850314041226,
+      "eval_accuracy": 0.679932551029844,
+      "eval_loss": 4.977242946624756,
+      "eval_runtime": 411.6101,
+      "eval_samples_per_second": 293.253,
+      "eval_steps_per_second": 3.056,
+      "step": 110000
+    },
+    {
+      "epoch": 4.657926906377777,
+      "grad_norm": 1.3174201250076294,
+      "learning_rate": 8.276379539510355e-05,
+      "loss": 5.2715,
+      "step": 110500
+    },
+    {
+      "epoch": 4.679003498714328,
+      "grad_norm": 1.283056378364563,
+      "learning_rate": 8.267804313338765e-05,
+      "loss": 5.2718,
+      "step": 111000
+    },
+    {
+      "epoch": 4.700080091050879,
+      "grad_norm": 1.3456379175186157,
+      "learning_rate": 8.259229087167174e-05,
+      "loss": 5.2667,
+      "step": 111500
+    },
+    {
+      "epoch": 4.72115668338743,
+      "grad_norm": 1.3283311128616333,
+      "learning_rate": 8.250653860995584e-05,
+      "loss": 5.2652,
+      "step": 112000
+    },
+    {
+      "epoch": 4.742233275723981,
+      "grad_norm": 1.297393798828125,
+      "learning_rate": 8.242078634823994e-05,
+      "loss": 5.2648,
+      "step": 112500
+    },
+    {
+      "epoch": 4.763309868060532,
+      "grad_norm": 1.3147990703582764,
+      "learning_rate": 8.233503408652403e-05,
+      "loss": 5.2665,
+      "step": 113000
+    },
+    {
+      "epoch": 4.784386460397083,
+      "grad_norm": 1.2669214010238647,
+      "learning_rate": 8.224928182480814e-05,
+      "loss": 5.2684,
+      "step": 113500
+    },
+    {
+      "epoch": 4.805463052733634,
+      "grad_norm": 1.243544578552246,
+      "learning_rate": 8.216352956309224e-05,
+      "loss": 5.2582,
+      "step": 114000
+    },
+    {
+      "epoch": 4.826539645070185,
+      "grad_norm": 1.2834265232086182,
+      "learning_rate": 8.207777730137632e-05,
+      "loss": 5.2603,
+      "step": 114500
+    },
+    {
+      "epoch": 4.847616237406736,
+      "grad_norm": 1.344043254852295,
+      "learning_rate": 8.199202503966042e-05,
+      "loss": 5.2593,
+      "step": 115000
+    },
+    {
+      "epoch": 4.868692829743287,
+      "grad_norm": 1.2844358682632446,
+      "learning_rate": 8.190627277794453e-05,
+      "loss": 5.2608,
+      "step": 115500
+    },
+    {
+      "epoch": 4.889769422079838,
+      "grad_norm": 1.2777135372161865,
+      "learning_rate": 8.182052051622862e-05,
+      "loss": 5.2555,
+      "step": 116000
+    },
+    {
+      "epoch": 4.910846014416389,
+      "grad_norm": 1.2810074090957642,
+      "learning_rate": 8.173476825451271e-05,
+      "loss": 5.2562,
+      "step": 116500
+    },
+    {
+      "epoch": 4.9319226067529405,
+      "grad_norm": 1.2800641059875488,
+      "learning_rate": 8.164901599279681e-05,
+      "loss": 5.256,
+      "step": 117000
+    },
+    {
+      "epoch": 4.952999199089491,
+      "grad_norm": 1.3152170181274414,
+      "learning_rate": 8.156326373108091e-05,
+      "loss": 5.2565,
+      "step": 117500
+    },
+    {
+      "epoch": 4.974075791426042,
+      "grad_norm": 1.2668205499649048,
+      "learning_rate": 8.147751146936501e-05,
+      "loss": 5.254,
+      "step": 118000
+    },
+    {
+      "epoch": 4.995152383762593,
+      "grad_norm": 1.316371202468872,
+      "learning_rate": 8.139175920764911e-05,
+      "loss": 5.2468,
+      "step": 118500
+    },
+    {
+      "epoch": 5.0162289760991445,
+      "grad_norm": 1.3267393112182617,
+      "learning_rate": 8.13060069459332e-05,
+      "loss": 5.2501,
+      "step": 119000
+    },
+    {
+      "epoch": 5.037305568435695,
+      "grad_norm": 1.279232382774353,
+      "learning_rate": 8.12202546842173e-05,
+      "loss": 5.2435,
+      "step": 119500
+    },
+    {
+      "epoch": 5.058382160772246,
+      "grad_norm": 1.3551067113876343,
+      "learning_rate": 8.11345024225014e-05,
+      "loss": 5.2442,
+      "step": 120000
+    },
+    {
+      "epoch": 5.058382160772246,
+      "eval_accuracy": 0.6835674632269141,
+      "eval_loss": 4.9517340660095215,
+      "eval_runtime": 391.9328,
+      "eval_samples_per_second": 307.976,
+      "eval_steps_per_second": 3.21,
+      "step": 120000
+    },
+    {
+      "epoch": 5.079458753108797,
+      "grad_norm": 1.296622395515442,
+      "learning_rate": 8.104875016078549e-05,
+      "loss": 5.2379,
+      "step": 120500
+    },
+    {
+      "epoch": 5.100535345445349,
+      "grad_norm": 1.2998985052108765,
+      "learning_rate": 8.09629978990696e-05,
+      "loss": 5.2379,
+      "step": 121000
+    },
+    {
+      "epoch": 5.1216119377819,
+      "grad_norm": 1.3264979124069214,
+      "learning_rate": 8.08772456373537e-05,
+      "loss": 5.2396,
+      "step": 121500
+    },
+    {
+      "epoch": 5.14268853011845,
+      "grad_norm": 1.291715145111084,
+      "learning_rate": 8.079149337563778e-05,
+      "loss": 5.2385,
+      "step": 122000
+    },
+    {
+      "epoch": 5.163765122455001,
+      "grad_norm": 1.3055702447891235,
+      "learning_rate": 8.070574111392188e-05,
+      "loss": 5.2356,
+      "step": 122500
+    },
+    {
+      "epoch": 5.184841714791553,
+      "grad_norm": 1.2707175016403198,
+      "learning_rate": 8.061998885220599e-05,
+      "loss": 5.2359,
+      "step": 123000
+    },
+    {
+      "epoch": 5.205918307128104,
+      "grad_norm": 1.2703309059143066,
+      "learning_rate": 8.053423659049008e-05,
+      "loss": 5.2361,
+      "step": 123500
+    },
+    {
+      "epoch": 5.226994899464654,
+      "grad_norm": 1.2912262678146362,
+      "learning_rate": 8.044848432877418e-05,
+      "loss": 5.2361,
+      "step": 124000
+    },
+    {
+      "epoch": 5.2480714918012055,
+      "grad_norm": 1.2901428937911987,
+      "learning_rate": 8.036273206705827e-05,
+      "loss": 5.239,
+      "step": 124500
+    },
+    {
+      "epoch": 5.269148084137757,
+      "grad_norm": 1.3538832664489746,
+      "learning_rate": 8.027697980534237e-05,
+      "loss": 5.2318,
+      "step": 125000
+    },
+    {
+      "epoch": 5.290224676474308,
+      "grad_norm": 1.3132553100585938,
+      "learning_rate": 8.019122754362647e-05,
+      "loss": 5.2333,
+      "step": 125500
+    },
+    {
+      "epoch": 5.311301268810858,
+      "grad_norm": 1.2882691621780396,
+      "learning_rate": 8.010547528191057e-05,
+      "loss": 5.2333,
+      "step": 126000
+    },
+    {
+      "epoch": 5.3323778611474095,
+      "grad_norm": 1.2922494411468506,
+      "learning_rate": 8.001972302019465e-05,
+      "loss": 5.2279,
+      "step": 126500
+    },
+    {
+      "epoch": 5.353454453483961,
+      "grad_norm": 1.2675259113311768,
+      "learning_rate": 7.993397075847876e-05,
+      "loss": 5.2274,
+      "step": 127000
+    },
+    {
+      "epoch": 5.374531045820512,
+      "grad_norm": 1.3118537664413452,
+      "learning_rate": 7.984821849676286e-05,
+      "loss": 5.2313,
+      "step": 127500
+    },
+    {
+      "epoch": 5.395607638157063,
+      "grad_norm": 1.3166561126708984,
+      "learning_rate": 7.976246623504695e-05,
+      "loss": 5.2294,
+      "step": 128000
+    },
+    {
+      "epoch": 5.416684230493614,
+      "grad_norm": 1.3203997611999512,
+      "learning_rate": 7.967671397333105e-05,
+      "loss": 5.2232,
+      "step": 128500
+    },
+    {
+      "epoch": 5.437760822830165,
+      "grad_norm": 1.2758651971817017,
+      "learning_rate": 7.959096171161516e-05,
+      "loss": 5.2234,
+      "step": 129000
+    },
+    {
+      "epoch": 5.458837415166716,
+      "grad_norm": 1.2822166681289673,
+      "learning_rate": 7.950520944989924e-05,
+      "loss": 5.2235,
+      "step": 129500
+    },
+    {
+      "epoch": 5.479914007503266,
+      "grad_norm": 1.3105649948120117,
+      "learning_rate": 7.941945718818334e-05,
+      "loss": 5.2231,
+      "step": 130000
+    },
+    {
+      "epoch": 5.479914007503266,
+      "eval_accuracy": 0.6862588739765383,
+      "eval_loss": 4.929084777832031,
+      "eval_runtime": 426.4364,
+      "eval_samples_per_second": 283.057,
+      "eval_steps_per_second": 2.95,
+      "step": 130000
+    },
+    {
+      "epoch": 5.500990599839818,
+      "grad_norm": 1.295576572418213,
+      "learning_rate": 7.933370492646744e-05,
+      "loss": 5.226,
+      "step": 130500
+    },
+    {
+      "epoch": 5.522067192176369,
+      "grad_norm": 1.2673089504241943,
+      "learning_rate": 7.924795266475154e-05,
+      "loss": 5.2187,
+      "step": 131000
+    },
+    {
+      "epoch": 5.54314378451292,
+      "grad_norm": 1.3100231885910034,
+      "learning_rate": 7.916220040303564e-05,
+      "loss": 5.2153,
+      "step": 131500
+    },
+    {
+      "epoch": 5.564220376849471,
+      "grad_norm": 1.3027418851852417,
+      "learning_rate": 7.907644814131972e-05,
+      "loss": 5.2201,
+      "step": 132000
+    },
+    {
+      "epoch": 5.585296969186022,
+      "grad_norm": 1.2707483768463135,
+      "learning_rate": 7.899069587960383e-05,
+      "loss": 5.214,
+      "step": 132500
+    },
+    {
+      "epoch": 5.606373561522573,
+      "grad_norm": 1.3006540536880493,
+      "learning_rate": 7.890494361788793e-05,
+      "loss": 5.2163,
+      "step": 133000
+    },
+    {
+      "epoch": 5.627450153859124,
+      "grad_norm": 1.3388031721115112,
+      "learning_rate": 7.881919135617202e-05,
+      "loss": 5.2124,
+      "step": 133500
+    },
+    {
+      "epoch": 5.648526746195675,
+      "grad_norm": 1.2674751281738281,
+      "learning_rate": 7.873343909445611e-05,
+      "loss": 5.2108,
+      "step": 134000
+    },
+    {
+      "epoch": 5.669603338532226,
+      "grad_norm": 1.313044786453247,
+      "learning_rate": 7.864768683274023e-05,
+      "loss": 5.2186,
+      "step": 134500
+    },
+    {
+      "epoch": 5.690679930868777,
+      "grad_norm": 1.9930195808410645,
+      "learning_rate": 7.856193457102431e-05,
+      "loss": 5.2177,
+      "step": 135000
+    },
+    {
+      "epoch": 5.711756523205328,
+      "grad_norm": 1.2985007762908936,
+      "learning_rate": 7.847618230930841e-05,
+      "loss": 5.2129,
+      "step": 135500
+    },
+    {
+      "epoch": 5.7328331155418795,
+      "grad_norm": 1.3034703731536865,
+      "learning_rate": 7.839043004759251e-05,
+      "loss": 5.2125,
+      "step": 136000
+    },
+    {
+      "epoch": 5.75390970787843,
+      "grad_norm": 1.300583004951477,
+      "learning_rate": 7.83046777858766e-05,
+      "loss": 5.2098,
+      "step": 136500
+    },
+    {
+      "epoch": 5.774986300214981,
+      "grad_norm": 1.287124752998352,
+      "learning_rate": 7.82189255241607e-05,
+      "loss": 5.2084,
+      "step": 137000
+    },
+    {
+      "epoch": 5.796062892551532,
+      "grad_norm": 1.3143622875213623,
+      "learning_rate": 7.81331732624448e-05,
+      "loss": 5.2086,
+      "step": 137500
+    },
+    {
+      "epoch": 5.8171394848880835,
+      "grad_norm": 1.2771177291870117,
+      "learning_rate": 7.804742100072889e-05,
+      "loss": 5.2059,
+      "step": 138000
+    },
+    {
+      "epoch": 5.838216077224635,
+      "grad_norm": 1.3292794227600098,
+      "learning_rate": 7.7961668739013e-05,
+      "loss": 5.2109,
+      "step": 138500
+    },
+    {
+      "epoch": 5.859292669561185,
+      "grad_norm": 1.2895891666412354,
+      "learning_rate": 7.78759164772971e-05,
+      "loss": 5.2053,
+      "step": 139000
+    },
+    {
+      "epoch": 5.880369261897736,
+      "grad_norm": 1.330141305923462,
+      "learning_rate": 7.779016421558118e-05,
+      "loss": 5.203,
+      "step": 139500
+    },
+    {
+      "epoch": 5.901445854234288,
+      "grad_norm": 1.3091015815734863,
+      "learning_rate": 7.770441195386528e-05,
+      "loss": 5.2074,
+      "step": 140000
+    },
+    {
+      "epoch": 5.901445854234288,
+      "eval_accuracy": 0.6887733632398462,
+      "eval_loss": 4.910532474517822,
+      "eval_runtime": 440.6112,
+      "eval_samples_per_second": 273.951,
+      "eval_steps_per_second": 2.855,
+      "step": 140000
+    },
+    {
+      "epoch": 5.922522446570838,
+      "grad_norm": 1.2901860475540161,
+      "learning_rate": 7.761865969214939e-05,
+      "loss": 5.1986,
+      "step": 140500
+    },
+    {
+      "epoch": 5.943599038907389,
+      "grad_norm": 1.315026879310608,
+      "learning_rate": 7.753290743043348e-05,
+      "loss": 5.2022,
+      "step": 141000
+    },
+    {
+      "epoch": 5.96467563124394,
+      "grad_norm": 1.3133258819580078,
+      "learning_rate": 7.744715516871758e-05,
+      "loss": 5.2042,
+      "step": 141500
+    },
+    {
+      "epoch": 5.985752223580492,
+      "grad_norm": 1.3393021821975708,
+      "learning_rate": 7.736140290700167e-05,
+      "loss": 5.198,
+      "step": 142000
+    },
+    {
+      "epoch": 6.006828815917043,
+      "grad_norm": 1.3312336206436157,
+      "learning_rate": 7.727565064528577e-05,
+      "loss": 5.1978,
+      "step": 142500
+    },
+    {
+      "epoch": 6.027905408253593,
+      "grad_norm": 1.2925186157226562,
+      "learning_rate": 7.718989838356987e-05,
+      "loss": 5.1944,
+      "step": 143000
+    },
+    {
+      "epoch": 6.0489820005901445,
+      "grad_norm": 1.3233197927474976,
+      "learning_rate": 7.710414612185397e-05,
+      "loss": 5.1961,
+      "step": 143500
+    },
+    {
+      "epoch": 6.070058592926696,
+      "grad_norm": 1.2674156427383423,
+      "learning_rate": 7.701839386013807e-05,
+      "loss": 5.1915,
+      "step": 144000
+    },
+    {
+      "epoch": 6.091135185263247,
+      "grad_norm": 1.308335781097412,
+      "learning_rate": 7.693264159842217e-05,
+      "loss": 5.1902,
+      "step": 144500
+    },
+    {
+      "epoch": 6.112211777599797,
+      "grad_norm": 1.3076956272125244,
+      "learning_rate": 7.684688933670626e-05,
+      "loss": 5.191,
+      "step": 145000
+    },
+    {
+      "epoch": 6.1332883699363485,
+      "grad_norm": 1.3039543628692627,
+      "learning_rate": 7.676113707499035e-05,
+      "loss": 5.1926,
+      "step": 145500
+    },
+    {
+      "epoch": 6.1543649622729,
+      "grad_norm": 1.319203495979309,
+      "learning_rate": 7.667538481327446e-05,
+      "loss": 5.1908,
+      "step": 146000
+    },
+    {
+      "epoch": 6.175441554609451,
+      "grad_norm": 1.3273495435714722,
+      "learning_rate": 7.658963255155856e-05,
+      "loss": 5.1921,
+      "step": 146500
+    },
+    {
+      "epoch": 6.196518146946001,
+      "grad_norm": 1.312489628791809,
+      "learning_rate": 7.650388028984264e-05,
+      "loss": 5.1877,
+      "step": 147000
+    },
+    {
+      "epoch": 6.217594739282553,
+      "grad_norm": 1.291069507598877,
+      "learning_rate": 7.641812802812674e-05,
+      "loss": 5.1874,
+      "step": 147500
+    },
+    {
+      "epoch": 6.238671331619104,
+      "grad_norm": 1.2625854015350342,
+      "learning_rate": 7.633237576641084e-05,
+      "loss": 5.1851,
+      "step": 148000
+    },
+    {
+      "epoch": 6.259747923955655,
+      "grad_norm": 1.2892005443572998,
+      "learning_rate": 7.624662350469494e-05,
+      "loss": 5.188,
+      "step": 148500
+    },
+    {
+      "epoch": 6.280824516292206,
+      "grad_norm": 1.3193202018737793,
+      "learning_rate": 7.616087124297904e-05,
+      "loss": 5.1879,
+      "step": 149000
+    },
+    {
+      "epoch": 6.301901108628757,
+      "grad_norm": 1.3224939107894897,
+      "learning_rate": 7.607511898126313e-05,
+      "loss": 5.1865,
+      "step": 149500
+    },
+    {
+      "epoch": 6.322977700965308,
+      "grad_norm": 1.31862473487854,
+      "learning_rate": 7.598936671954723e-05,
+      "loss": 5.1812,
+      "step": 150000
+    },
+    {
+      "epoch": 6.322977700965308,
+      "eval_accuracy": 0.691114071840627,
+      "eval_loss": 4.8955888748168945,
+      "eval_runtime": 471.6163,
+      "eval_samples_per_second": 255.941,
+      "eval_steps_per_second": 2.667,
+      "step": 150000
+    },
+    {
+      "epoch": 6.344054293301859,
+      "grad_norm": 1.3102821111679077,
+      "learning_rate": 7.590361445783133e-05,
+      "loss": 5.1813,
+      "step": 150500
+    },
+    {
+      "epoch": 6.36513088563841,
+      "grad_norm": 1.2615158557891846,
+      "learning_rate": 7.581786219611543e-05,
+      "loss": 5.1821,
+      "step": 151000
+    },
+    {
+      "epoch": 6.386207477974961,
+      "grad_norm": 1.2552192211151123,
+      "learning_rate": 7.573210993439951e-05,
+      "loss": 5.1823,
+      "step": 151500
+    },
+    {
+      "epoch": 6.407284070311512,
+      "grad_norm": 1.277457594871521,
+      "learning_rate": 7.564635767268363e-05,
+      "loss": 5.1816,
+      "step": 152000
+    },
+    {
+      "epoch": 6.428360662648063,
+      "grad_norm": 1.288087248802185,
+      "learning_rate": 7.556060541096772e-05,
+      "loss": 5.1781,
+      "step": 152500
+    },
+    {
+      "epoch": 6.449437254984614,
+      "grad_norm": 1.3226521015167236,
+      "learning_rate": 7.547485314925181e-05,
+      "loss": 5.1803,
+      "step": 153000
+    },
+    {
+      "epoch": 6.470513847321165,
+      "grad_norm": 1.2988865375518799,
+      "learning_rate": 7.538910088753591e-05,
+      "loss": 5.1777,
+      "step": 153500
+    },
+    {
+      "epoch": 6.491590439657716,
+      "grad_norm": 1.354048728942871,
+      "learning_rate": 7.530334862582002e-05,
+      "loss": 5.1794,
+      "step": 154000
+    },
+    {
+      "epoch": 6.512667031994267,
+      "grad_norm": 1.2946815490722656,
+      "learning_rate": 7.52175963641041e-05,
+      "loss": 5.1793,
+      "step": 154500
+    },
+    {
+      "epoch": 6.5337436243308185,
+      "grad_norm": 1.2748323678970337,
+      "learning_rate": 7.51318441023882e-05,
+      "loss": 5.1807,
+      "step": 155000
+    },
+    {
+      "epoch": 6.55482021666737,
+      "grad_norm": 1.2864291667938232,
+      "learning_rate": 7.50460918406723e-05,
+      "loss": 5.1769,
+      "step": 155500
+    },
+    {
+      "epoch": 6.57589680900392,
+      "grad_norm": 1.3454313278198242,
+      "learning_rate": 7.49603395789564e-05,
+      "loss": 5.175,
+      "step": 156000
+    },
+    {
+      "epoch": 6.596973401340471,
+      "grad_norm": 1.2581576108932495,
+      "learning_rate": 7.48745873172405e-05,
+      "loss": 5.1713,
+      "step": 156500
+    },
+    {
+      "epoch": 6.6180499936770225,
+      "grad_norm": 1.324028730392456,
+      "learning_rate": 7.47888350555246e-05,
+      "loss": 5.1677,
+      "step": 157000
+    },
+    {
+      "epoch": 6.639126586013573,
+      "grad_norm": 1.282078504562378,
+      "learning_rate": 7.47030827938087e-05,
+      "loss": 5.1724,
+      "step": 157500
+    },
+    {
+      "epoch": 6.660203178350124,
+      "grad_norm": 1.265564203262329,
+      "learning_rate": 7.461733053209279e-05,
+      "loss": 5.1732,
+      "step": 158000
+    },
+    {
+      "epoch": 6.681279770686675,
+      "grad_norm": 1.2876176834106445,
+      "learning_rate": 7.453157827037689e-05,
+      "loss": 5.1749,
+      "step": 158500
+    },
+    {
+      "epoch": 6.702356363023227,
+      "grad_norm": 1.2822721004486084,
+      "learning_rate": 7.444582600866098e-05,
+      "loss": 5.1683,
+      "step": 159000
+    },
+    {
+      "epoch": 6.723432955359778,
+      "grad_norm": 1.3303354978561401,
+      "learning_rate": 7.436007374694507e-05,
+      "loss": 5.173,
+      "step": 159500
+    },
+    {
+      "epoch": 6.744509547696328,
+      "grad_norm": 1.302960753440857,
+      "learning_rate": 7.427432148522919e-05,
+      "loss": 5.1733,
+      "step": 160000
+    },
+    {
+      "epoch": 6.744509547696328,
+      "eval_accuracy": 0.693361929473727,
+      "eval_loss": 4.881340503692627,
+      "eval_runtime": 473.9763,
+      "eval_samples_per_second": 254.667,
+      "eval_steps_per_second": 2.654,
+      "step": 160000
+    },
+    {
+      "epoch": 6.765586140032879,
+      "grad_norm": 1.3261351585388184,
+      "learning_rate": 7.418856922351327e-05,
+      "loss": 5.1699,
+      "step": 160500
+    },
+    {
+      "epoch": 6.786662732369431,
+      "grad_norm": 1.2957472801208496,
+      "learning_rate": 7.410281696179737e-05,
+      "loss": 5.1671,
+      "step": 161000
+    },
+    {
+      "epoch": 6.807739324705982,
+      "grad_norm": 1.2983335256576538,
+      "learning_rate": 7.401706470008147e-05,
+      "loss": 5.1645,
+      "step": 161500
+    },
+    {
+      "epoch": 6.828815917042532,
+      "grad_norm": 1.2316908836364746,
+      "learning_rate": 7.393131243836557e-05,
+      "loss": 5.1628,
+      "step": 162000
+    },
+    {
+      "epoch": 6.8498925093790834,
+      "grad_norm": 1.3176705837249756,
+      "learning_rate": 7.384556017664966e-05,
+      "loss": 5.1622,
+      "step": 162500
+    },
+    {
+      "epoch": 6.870969101715635,
+      "grad_norm": 1.290685772895813,
+      "learning_rate": 7.375980791493376e-05,
+      "loss": 5.1602,
+      "step": 163000
+    },
+    {
+      "epoch": 6.892045694052186,
+      "grad_norm": 1.292539358139038,
+      "learning_rate": 7.367405565321786e-05,
+      "loss": 5.1606,
+      "step": 163500
+    },
+    {
+      "epoch": 6.913122286388736,
+      "grad_norm": 1.2720119953155518,
+      "learning_rate": 7.358830339150196e-05,
+      "loss": 5.162,
+      "step": 164000
+    },
+    {
+      "epoch": 6.9341988787252875,
+      "grad_norm": 1.356644630432129,
+      "learning_rate": 7.350255112978606e-05,
+      "loss": 5.1651,
+      "step": 164500
+    },
+    {
+      "epoch": 6.955275471061839,
+      "grad_norm": 1.3013664484024048,
+      "learning_rate": 7.341679886807014e-05,
+      "loss": 5.162,
+      "step": 165000
+    },
+    {
+      "epoch": 6.97635206339839,
+      "grad_norm": 1.3517738580703735,
+      "learning_rate": 7.333104660635425e-05,
+      "loss": 5.1585,
+      "step": 165500
+    },
+    {
+      "epoch": 6.997428655734941,
+      "grad_norm": 1.3021160364151,
+      "learning_rate": 7.324529434463835e-05,
+      "loss": 5.1611,
+      "step": 166000
+    },
+    {
+      "epoch": 7.018505248071492,
+      "grad_norm": 1.2942171096801758,
+      "learning_rate": 7.315954208292244e-05,
+      "loss": 5.1565,
+      "step": 166500
+    },
+    {
+      "epoch": 7.039581840408043,
+      "grad_norm": 1.2842696905136108,
+      "learning_rate": 7.307378982120653e-05,
+      "loss": 5.1514,
+      "step": 167000
+    },
+    {
+      "epoch": 7.060658432744594,
+      "grad_norm": 1.3094406127929688,
+      "learning_rate": 7.298803755949063e-05,
+      "loss": 5.152,
+      "step": 167500
+    },
+    {
+      "epoch": 7.081735025081145,
+      "grad_norm": 1.3262333869934082,
+      "learning_rate": 7.290228529777473e-05,
+      "loss": 5.1517,
+      "step": 168000
+    },
+    {
+      "epoch": 7.102811617417696,
+      "grad_norm": 1.3389296531677246,
+      "learning_rate": 7.281653303605883e-05,
+      "loss": 5.1539,
+      "step": 168500
+    },
+    {
+      "epoch": 7.123888209754247,
+      "grad_norm": 1.3371188640594482,
+      "learning_rate": 7.273078077434293e-05,
+      "loss": 5.1574,
+      "step": 169000
+    },
+    {
+      "epoch": 7.144964802090798,
+      "grad_norm": 1.3263627290725708,
+      "learning_rate": 7.264502851262703e-05,
+      "loss": 5.1492,
+      "step": 169500
+    },
+    {
+      "epoch": 7.166041394427349,
+      "grad_norm": 1.339030146598816,
+      "learning_rate": 7.255927625091112e-05,
+      "loss": 5.147,
+      "step": 170000
+    },
+    {
+      "epoch": 7.166041394427349,
+      "eval_accuracy": 0.6952644518173305,
+      "eval_loss": 4.866620063781738,
+      "eval_runtime": 434.1747,
+      "eval_samples_per_second": 278.013,
+      "eval_steps_per_second": 2.897,
+      "step": 170000
+    },
+    {
+      "epoch": 7.1871179867639,
+      "grad_norm": 1.303316593170166,
+      "learning_rate": 7.247352398919521e-05,
+      "loss": 5.1507,
+      "step": 170500
+    },
+    {
+      "epoch": 7.208194579100451,
+      "grad_norm": 1.292258858680725,
+      "learning_rate": 7.238777172747931e-05,
+      "loss": 5.1493,
+      "step": 171000
+    },
+    {
+      "epoch": 7.229271171437002,
+      "grad_norm": 1.2687932252883911,
+      "learning_rate": 7.230201946576342e-05,
+      "loss": 5.1526,
+      "step": 171500
+    },
+    {
+      "epoch": 7.250347763773553,
+      "grad_norm": 1.3280694484710693,
+      "learning_rate": 7.22162672040475e-05,
+      "loss": 5.1503,
+      "step": 172000
+    },
+    {
+      "epoch": 7.271424356110104,
+      "grad_norm": 1.2832854986190796,
+      "learning_rate": 7.21305149423316e-05,
+      "loss": 5.1488,
+      "step": 172500
+    },
+    {
+      "epoch": 7.292500948446655,
+      "grad_norm": 1.2606652975082397,
+      "learning_rate": 7.20447626806157e-05,
+      "loss": 5.1434,
+      "step": 173000
+    },
+    {
+      "epoch": 7.313577540783206,
+      "grad_norm": 1.324495553970337,
+      "learning_rate": 7.19590104188998e-05,
+      "loss": 5.1453,
+      "step": 173500
+    },
+    {
+      "epoch": 7.3346541331197574,
+      "grad_norm": 1.2649351358413696,
+      "learning_rate": 7.18732581571839e-05,
+      "loss": 5.1449,
+      "step": 174000
+    },
+    {
+      "epoch": 7.355730725456308,
+      "grad_norm": 1.3301286697387695,
+      "learning_rate": 7.1787505895468e-05,
+      "loss": 5.1445,
+      "step": 174500
+    },
+    {
+      "epoch": 7.376807317792859,
+      "grad_norm": 1.3117254972457886,
+      "learning_rate": 7.17017536337521e-05,
+      "loss": 5.1451,
+      "step": 175000
+    },
+    {
+      "epoch": 7.39788391012941,
+      "grad_norm": 1.2663692235946655,
+      "learning_rate": 7.161600137203619e-05,
+      "loss": 5.1482,
+      "step": 175500
+    },
+    {
+      "epoch": 7.4189605024659615,
+      "grad_norm": 1.299490213394165,
+      "learning_rate": 7.153024911032029e-05,
+      "loss": 5.1399,
+      "step": 176000
+    },
+    {
+      "epoch": 7.440037094802513,
+      "grad_norm": 1.394643783569336,
+      "learning_rate": 7.144449684860438e-05,
+      "loss": 5.145,
+      "step": 176500
+    },
+    {
+      "epoch": 7.461113687139063,
+      "grad_norm": 1.287297248840332,
+      "learning_rate": 7.135874458688849e-05,
+      "loss": 5.1459,
+      "step": 177000
+    },
+    {
+      "epoch": 7.482190279475614,
+      "grad_norm": 1.3449349403381348,
+      "learning_rate": 7.127299232517259e-05,
+      "loss": 5.1455,
+      "step": 177500
+    },
+    {
+      "epoch": 7.503266871812166,
+      "grad_norm": 1.3113905191421509,
+      "learning_rate": 7.118724006345667e-05,
+      "loss": 5.1414,
+      "step": 178000
+    },
+    {
+      "epoch": 7.524343464148717,
+      "grad_norm": 1.3419477939605713,
+      "learning_rate": 7.110148780174077e-05,
+      "loss": 5.1379,
+      "step": 178500
+    },
+    {
+      "epoch": 7.545420056485267,
+      "grad_norm": 1.2903634309768677,
+      "learning_rate": 7.101573554002488e-05,
+      "loss": 5.1389,
+      "step": 179000
+    },
+    {
+      "epoch": 7.566496648821818,
+      "grad_norm": 1.294148564338684,
+      "learning_rate": 7.092998327830897e-05,
+      "loss": 5.141,
+      "step": 179500
+    },
+    {
+      "epoch": 7.58757324115837,
+      "grad_norm": 1.3114814758300781,
+      "learning_rate": 7.084423101659306e-05,
+      "loss": 5.1368,
+      "step": 180000
+    },
+    {
+      "epoch": 7.58757324115837,
+      "eval_accuracy": 0.6966848336305617,
+      "eval_loss": 4.856717109680176,
+      "eval_runtime": 441.6956,
+      "eval_samples_per_second": 273.279,
+      "eval_steps_per_second": 2.848,
+      "step": 180000
+    },
+    {
+      "epoch": 7.608649833494921,
+      "grad_norm": 1.2910640239715576,
+      "learning_rate": 7.075847875487716e-05,
+      "loss": 5.1383,
+      "step": 180500
+    },
+    {
+      "epoch": 7.629726425831471,
+      "grad_norm": 1.3074181079864502,
+      "learning_rate": 7.067272649316126e-05,
+      "loss": 5.1396,
+      "step": 181000
+    },
+    {
+      "epoch": 7.650803018168022,
+      "grad_norm": 1.3171992301940918,
+      "learning_rate": 7.058697423144536e-05,
+      "loss": 5.1335,
+      "step": 181500
+    },
+    {
+      "epoch": 7.671879610504574,
+      "grad_norm": 1.3088700771331787,
+      "learning_rate": 7.050122196972946e-05,
+      "loss": 5.1347,
+      "step": 182000
+    },
+    {
+      "epoch": 7.692956202841125,
+      "grad_norm": 1.285516619682312,
+      "learning_rate": 7.041546970801354e-05,
+      "loss": 5.1355,
+      "step": 182500
+    },
+    {
+      "epoch": 7.714032795177676,
+      "grad_norm": 1.3399360179901123,
+      "learning_rate": 7.032971744629765e-05,
+      "loss": 5.1361,
+      "step": 183000
+    },
+    {
+      "epoch": 7.7351093875142265,
+      "grad_norm": 1.3115800619125366,
+      "learning_rate": 7.024396518458175e-05,
+      "loss": 5.1386,
+      "step": 183500
+    },
+    {
+      "epoch": 7.756185979850778,
+      "grad_norm": 1.3070324659347534,
+      "learning_rate": 7.015821292286584e-05,
+      "loss": 5.1345,
+      "step": 184000
+    },
+    {
+      "epoch": 7.777262572187329,
+      "grad_norm": 1.35909104347229,
+      "learning_rate": 7.007246066114993e-05,
+      "loss": 5.1345,
+      "step": 184500
+    },
+    {
+      "epoch": 7.798339164523879,
+      "grad_norm": 1.3207504749298096,
+      "learning_rate": 6.998670839943405e-05,
+      "loss": 5.1343,
+      "step": 185000
+    },
+    {
+      "epoch": 7.8194157568604306,
+      "grad_norm": 1.2968013286590576,
+      "learning_rate": 6.990095613771813e-05,
+      "loss": 5.132,
+      "step": 185500
+    },
+    {
+      "epoch": 7.840492349196982,
+      "grad_norm": 1.3535782098770142,
+      "learning_rate": 6.981520387600223e-05,
+      "loss": 5.1316,
+      "step": 186000
+    },
+    {
+      "epoch": 7.861568941533533,
+      "grad_norm": 1.2922688722610474,
+      "learning_rate": 6.972945161428633e-05,
+      "loss": 5.1315,
+      "step": 186500
+    },
+    {
+      "epoch": 7.882645533870084,
+      "grad_norm": 1.2952519655227661,
+      "learning_rate": 6.964369935257043e-05,
+      "loss": 5.1317,
+      "step": 187000
+    },
+    {
+      "epoch": 7.903722126206635,
+      "grad_norm": 1.3138635158538818,
+      "learning_rate": 6.955794709085452e-05,
+      "loss": 5.1349,
+      "step": 187500
+    },
+    {
+      "epoch": 7.924798718543186,
+      "grad_norm": 1.3534938097000122,
+      "learning_rate": 6.947219482913862e-05,
+      "loss": 5.1209,
+      "step": 188000
+    },
+    {
+      "epoch": 7.945875310879737,
+      "grad_norm": 1.2811720371246338,
+      "learning_rate": 6.938644256742272e-05,
+      "loss": 5.1308,
+      "step": 188500
+    },
+    {
+      "epoch": 7.966951903216288,
+      "grad_norm": 1.3534867763519287,
+      "learning_rate": 6.930069030570682e-05,
+      "loss": 5.128,
+      "step": 189000
+    },
+    {
+      "epoch": 7.988028495552839,
+      "grad_norm": 1.314897894859314,
+      "learning_rate": 6.921493804399092e-05,
+      "loss": 5.1244,
+      "step": 189500
+    },
+    {
+      "epoch": 8.00910508788939,
+      "grad_norm": 1.3463077545166016,
+      "learning_rate": 6.9129185782275e-05,
+      "loss": 5.1244,
+      "step": 190000
+    },
+    {
+      "epoch": 8.00910508788939,
+      "eval_accuracy": 0.698204601985256,
+      "eval_loss": 4.844047546386719,
+      "eval_runtime": 418.5124,
+      "eval_samples_per_second": 288.417,
+      "eval_steps_per_second": 3.006,
+      "step": 190000
+    },
+    {
+      "epoch": 8.03018168022594,
+      "grad_norm": 1.2777342796325684,
+      "learning_rate": 6.904343352055911e-05,
+      "loss": 5.1206,
+      "step": 190500
+    },
+    {
+      "epoch": 8.051258272562492,
+      "grad_norm": 1.3277195692062378,
+      "learning_rate": 6.895768125884321e-05,
+      "loss": 5.1236,
+      "step": 191000
+    },
+    {
+      "epoch": 8.072334864899043,
+      "grad_norm": 1.274364709854126,
+      "learning_rate": 6.88719289971273e-05,
+      "loss": 5.1216,
+      "step": 191500
+    },
+    {
+      "epoch": 8.093411457235595,
+      "grad_norm": 1.3108906745910645,
+      "learning_rate": 6.87861767354114e-05,
+      "loss": 5.1159,
+      "step": 192000
+    },
+    {
+      "epoch": 8.114488049572145,
+      "grad_norm": 1.3980183601379395,
+      "learning_rate": 6.870042447369551e-05,
+      "loss": 5.123,
+      "step": 192500
+    },
+    {
+      "epoch": 8.135564641908696,
+      "grad_norm": 1.30540931224823,
+      "learning_rate": 6.861467221197959e-05,
+      "loss": 5.1212,
+      "step": 193000
+    },
+    {
+      "epoch": 8.156641234245248,
+      "grad_norm": 1.3241055011749268,
+      "learning_rate": 6.852891995026369e-05,
+      "loss": 5.1232,
+      "step": 193500
+    },
+    {
+      "epoch": 8.177717826581798,
+      "grad_norm": 1.2977019548416138,
+      "learning_rate": 6.844316768854779e-05,
+      "loss": 5.123,
+      "step": 194000
+    },
+    {
+      "epoch": 8.19879441891835,
+      "grad_norm": 1.3086942434310913,
+      "learning_rate": 6.835741542683189e-05,
+      "loss": 5.1196,
+      "step": 194500
+    },
+    {
+      "epoch": 8.2198710112549,
+      "grad_norm": 1.3191075325012207,
+      "learning_rate": 6.827166316511599e-05,
+      "loss": 5.1151,
+      "step": 195000
+    },
+    {
+      "epoch": 8.24094760359145,
+      "grad_norm": 1.3092154264450073,
+      "learning_rate": 6.818591090340008e-05,
+      "loss": 5.1202,
+      "step": 195500
+    },
+    {
+      "epoch": 8.262024195928003,
+      "grad_norm": 1.268862247467041,
+      "learning_rate": 6.810015864168417e-05,
+      "loss": 5.1169,
+      "step": 196000
+    },
+    {
+      "epoch": 8.283100788264553,
+      "grad_norm": 1.3176894187927246,
+      "learning_rate": 6.801440637996828e-05,
+      "loss": 5.1137,
+      "step": 196500
+    },
+    {
+      "epoch": 8.304177380601104,
+      "grad_norm": 1.3100178241729736,
+      "learning_rate": 6.792865411825238e-05,
+      "loss": 5.1188,
+      "step": 197000
+    },
+    {
+      "epoch": 8.325253972937656,
+      "grad_norm": 1.3138573169708252,
+      "learning_rate": 6.784290185653646e-05,
+      "loss": 5.1147,
+      "step": 197500
+    },
+    {
+      "epoch": 8.346330565274206,
+      "grad_norm": 1.3463623523712158,
+      "learning_rate": 6.775714959482056e-05,
+      "loss": 5.1149,
+      "step": 198000
+    },
+    {
+      "epoch": 8.367407157610758,
+      "grad_norm": 1.3244996070861816,
+      "learning_rate": 6.767139733310467e-05,
+      "loss": 5.1176,
+      "step": 198500
+    },
+    {
+      "epoch": 8.388483749947309,
+      "grad_norm": 1.3197048902511597,
+      "learning_rate": 6.758564507138876e-05,
+      "loss": 5.1198,
+      "step": 199000
+    },
+    {
+      "epoch": 8.409560342283859,
+      "grad_norm": 1.2934545278549194,
+      "learning_rate": 6.749989280967286e-05,
+      "loss": 5.1182,
+      "step": 199500
+    },
+    {
+      "epoch": 8.430636934620411,
+      "grad_norm": 1.3267353773117065,
+      "learning_rate": 6.741414054795696e-05,
+      "loss": 5.1142,
+      "step": 200000
+    },
+    {
+      "epoch": 8.430636934620411,
+      "eval_accuracy": 0.6997812555712408,
+      "eval_loss": 4.831547737121582,
+      "eval_runtime": 448.7808,
+      "eval_samples_per_second": 268.964,
+      "eval_steps_per_second": 2.803,
+      "step": 200000
+    },
+    {
+      "epoch": 8.451713526956961,
+      "grad_norm": 1.3134899139404297,
+      "learning_rate": 6.732838828624105e-05,
+      "loss": 5.1136,
+      "step": 200500
+    },
+    {
+      "epoch": 8.472790119293512,
+      "grad_norm": 1.3624060153961182,
+      "learning_rate": 6.724263602452515e-05,
+      "loss": 5.1141,
+      "step": 201000
+    },
+    {
+      "epoch": 8.493866711630064,
+      "grad_norm": 1.3093907833099365,
+      "learning_rate": 6.715688376280924e-05,
+      "loss": 5.1125,
+      "step": 201500
+    },
+    {
+      "epoch": 8.514943303966614,
+      "grad_norm": 1.3175724744796753,
+      "learning_rate": 6.707113150109335e-05,
+      "loss": 5.1102,
+      "step": 202000
+    },
+    {
+      "epoch": 8.536019896303166,
+      "grad_norm": 1.3114393949508667,
+      "learning_rate": 6.698537923937745e-05,
+      "loss": 5.1163,
+      "step": 202500
+    },
+    {
+      "epoch": 8.557096488639717,
+      "grad_norm": 1.331745982170105,
+      "learning_rate": 6.689962697766153e-05,
+      "loss": 5.111,
+      "step": 203000
+    },
+    {
+      "epoch": 8.578173080976267,
+      "grad_norm": 1.332694172859192,
+      "learning_rate": 6.681387471594563e-05,
+      "loss": 5.1106,
+      "step": 203500
+    },
+    {
+      "epoch": 8.59924967331282,
+      "grad_norm": 1.3173582553863525,
+      "learning_rate": 6.672812245422974e-05,
+      "loss": 5.111,
+      "step": 204000
+    },
+    {
+      "epoch": 8.62032626564937,
+      "grad_norm": 1.3389126062393188,
+      "learning_rate": 6.664237019251383e-05,
+      "loss": 5.1114,
+      "step": 204500
+    },
+    {
+      "epoch": 8.64140285798592,
+      "grad_norm": 1.3265480995178223,
+      "learning_rate": 6.655661793079792e-05,
+      "loss": 5.1111,
+      "step": 205000
+    },
+    {
+      "epoch": 8.662479450322472,
+      "grad_norm": 1.3604921102523804,
+      "learning_rate": 6.647086566908202e-05,
+      "loss": 5.1044,
+      "step": 205500
+    },
+    {
+      "epoch": 8.683556042659022,
+      "grad_norm": 1.317467212677002,
+      "learning_rate": 6.638511340736612e-05,
+      "loss": 5.1065,
+      "step": 206000
+    },
+    {
+      "epoch": 8.704632634995574,
+      "grad_norm": 1.3411884307861328,
+      "learning_rate": 6.629936114565022e-05,
+      "loss": 5.1035,
+      "step": 206500
+    },
+    {
+      "epoch": 8.725709227332125,
+      "grad_norm": 1.3278048038482666,
+      "learning_rate": 6.621360888393432e-05,
+      "loss": 5.1054,
+      "step": 207000
+    },
+    {
+      "epoch": 8.746785819668675,
+      "grad_norm": 1.3197256326675415,
+      "learning_rate": 6.61278566222184e-05,
+      "loss": 5.1,
+      "step": 207500
+    },
+    {
+      "epoch": 8.767862412005227,
+      "grad_norm": 1.2880334854125977,
+      "learning_rate": 6.604210436050251e-05,
+      "loss": 5.1045,
+      "step": 208000
+    },
+    {
+      "epoch": 8.788939004341778,
+      "grad_norm": 1.3366769552230835,
+      "learning_rate": 6.595635209878661e-05,
+      "loss": 5.1064,
+      "step": 208500
+    },
+    {
+      "epoch": 8.81001559667833,
+      "grad_norm": 1.3714700937271118,
+      "learning_rate": 6.58705998370707e-05,
+      "loss": 5.1067,
+      "step": 209000
+    },
+    {
+      "epoch": 8.83109218901488,
+      "grad_norm": 1.3231408596038818,
+      "learning_rate": 6.57848475753548e-05,
+      "loss": 5.1052,
+      "step": 209500
+    },
+    {
+      "epoch": 8.85216878135143,
+      "grad_norm": 1.356917142868042,
+      "learning_rate": 6.569909531363891e-05,
+      "loss": 5.1017,
+      "step": 210000
+    },
+    {
+      "epoch": 8.85216878135143,
+      "eval_accuracy": 0.7013012251537175,
+      "eval_loss": 4.8245158195495605,
+      "eval_runtime": 450.9483,
+      "eval_samples_per_second": 267.671,
+      "eval_steps_per_second": 2.79,
+      "step": 210000
+    },
+    {
+      "epoch": 8.873245373687983,
+      "grad_norm": 1.3352161645889282,
+      "learning_rate": 6.561334305192299e-05,
+      "loss": 5.1026,
+      "step": 210500
+    },
+    {
+      "epoch": 8.894321966024533,
+      "grad_norm": 1.308169960975647,
+      "learning_rate": 6.552759079020709e-05,
+      "loss": 5.1041,
+      "step": 211000
+    },
+    {
+      "epoch": 8.915398558361083,
+      "grad_norm": 1.2983496189117432,
+      "learning_rate": 6.544183852849119e-05,
+      "loss": 5.0996,
+      "step": 211500
+    },
+    {
+      "epoch": 8.936475150697635,
+      "grad_norm": 1.30133056640625,
+      "learning_rate": 6.535608626677529e-05,
+      "loss": 5.1082,
+      "step": 212000
+    },
+    {
+      "epoch": 8.957551743034186,
+      "grad_norm": 1.3195756673812866,
+      "learning_rate": 6.527033400505939e-05,
+      "loss": 5.1059,
+      "step": 212500
+    },
+    {
+      "epoch": 8.978628335370738,
+      "grad_norm": 1.3136705160140991,
+      "learning_rate": 6.518458174334348e-05,
+      "loss": 5.1044,
+      "step": 213000
+    },
+    {
+      "epoch": 8.999704927707288,
+      "grad_norm": 1.3431062698364258,
+      "learning_rate": 6.509882948162758e-05,
+      "loss": 5.1034,
+      "step": 213500
+    },
+    {
+      "epoch": 9.020781520043839,
+      "grad_norm": 1.399232268333435,
+      "learning_rate": 6.501307721991168e-05,
+      "loss": 5.0967,
+      "step": 214000
+    },
+    {
+      "epoch": 9.04185811238039,
+      "grad_norm": 1.350939154624939,
+      "learning_rate": 6.492732495819578e-05,
+      "loss": 5.095,
+      "step": 214500
+    },
+    {
+      "epoch": 9.062934704716941,
+      "grad_norm": 1.3113495111465454,
+      "learning_rate": 6.484157269647986e-05,
+      "loss": 5.0973,
+      "step": 215000
+    },
+    {
+      "epoch": 9.084011297053493,
+      "grad_norm": 1.3239378929138184,
+      "learning_rate": 6.475582043476398e-05,
+      "loss": 5.0985,
+      "step": 215500
+    },
+    {
+      "epoch": 9.105087889390044,
+      "grad_norm": 1.337153673171997,
+      "learning_rate": 6.467006817304807e-05,
+      "loss": 5.0938,
+      "step": 216000
+    },
+    {
+      "epoch": 9.126164481726594,
+      "grad_norm": 1.3299312591552734,
+      "learning_rate": 6.458431591133216e-05,
+      "loss": 5.088,
+      "step": 216500
+    },
+    {
+      "epoch": 9.147241074063146,
+      "grad_norm": 1.3434714078903198,
+      "learning_rate": 6.449856364961626e-05,
+      "loss": 5.0934,
+      "step": 217000
+    },
+    {
+      "epoch": 9.168317666399696,
+      "grad_norm": 1.3038884401321411,
+      "learning_rate": 6.441281138790037e-05,
+      "loss": 5.0907,
+      "step": 217500
+    },
+    {
+      "epoch": 9.189394258736247,
+      "grad_norm": 1.303848385810852,
+      "learning_rate": 6.432705912618445e-05,
+      "loss": 5.0927,
+      "step": 218000
+    },
+    {
+      "epoch": 9.210470851072799,
+      "grad_norm": 1.3705435991287231,
+      "learning_rate": 6.424130686446855e-05,
+      "loss": 5.0952,
+      "step": 218500
+    },
+    {
+      "epoch": 9.23154744340935,
+      "grad_norm": 1.2962002754211426,
+      "learning_rate": 6.415555460275265e-05,
+      "loss": 5.0928,
+      "step": 219000
+    },
+    {
+      "epoch": 9.252624035745901,
+      "grad_norm": 1.2896671295166016,
+      "learning_rate": 6.406980234103675e-05,
+      "loss": 5.0938,
+      "step": 219500
+    },
+    {
+      "epoch": 9.273700628082452,
+      "grad_norm": 1.358551025390625,
+      "learning_rate": 6.398405007932085e-05,
+      "loss": 5.0955,
+      "step": 220000
+    },
+    {
+      "epoch": 9.273700628082452,
+      "eval_accuracy": 0.7025215569895956,
+      "eval_loss": 4.812854290008545,
+      "eval_runtime": 446.2867,
+      "eval_samples_per_second": 270.467,
+      "eval_steps_per_second": 2.819,
+      "step": 220000
+    },
+    {
+      "epoch": 9.294777220419002,
+      "grad_norm": 1.3137249946594238,
+      "learning_rate": 6.389829781760495e-05,
+      "loss": 5.0928,
+      "step": 220500
+    },
+    {
+      "epoch": 9.315853812755554,
+      "grad_norm": 1.3557196855545044,
+      "learning_rate": 6.381254555588903e-05,
+      "loss": 5.0952,
+      "step": 221000
+    },
+    {
+      "epoch": 9.336930405092104,
+      "grad_norm": 1.306570291519165,
+      "learning_rate": 6.372679329417314e-05,
+      "loss": 5.0892,
+      "step": 221500
+    },
+    {
+      "epoch": 9.358006997428657,
+      "grad_norm": 1.4003163576126099,
+      "learning_rate": 6.364104103245724e-05,
+      "loss": 5.0879,
+      "step": 222000
+    },
+    {
+      "epoch": 9.379083589765207,
+      "grad_norm": 1.353301763534546,
+      "learning_rate": 6.355528877074132e-05,
+      "loss": 5.0917,
+      "step": 222500
+    },
+    {
+      "epoch": 9.400160182101757,
+      "grad_norm": 1.3768064975738525,
+      "learning_rate": 6.346953650902542e-05,
+      "loss": 5.0905,
+      "step": 223000
+    },
+    {
+      "epoch": 9.42123677443831,
+      "grad_norm": 1.3337328433990479,
+      "learning_rate": 6.338378424730954e-05,
+      "loss": 5.0894,
+      "step": 223500
+    },
+    {
+      "epoch": 9.44231336677486,
+      "grad_norm": 1.3361189365386963,
+      "learning_rate": 6.329803198559362e-05,
+      "loss": 5.0874,
+      "step": 224000
+    },
+    {
+      "epoch": 9.46338995911141,
+      "grad_norm": 1.322868824005127,
+      "learning_rate": 6.321227972387772e-05,
+      "loss": 5.0858,
+      "step": 224500
+    },
+    {
+      "epoch": 9.484466551447962,
+      "grad_norm": 1.3310190439224243,
+      "learning_rate": 6.312652746216182e-05,
+      "loss": 5.0835,
+      "step": 225000
+    },
+    {
+      "epoch": 9.505543143784513,
+      "grad_norm": 1.3338582515716553,
+      "learning_rate": 6.304077520044591e-05,
+      "loss": 5.0939,
+      "step": 225500
+    },
+    {
+      "epoch": 9.526619736121065,
+      "grad_norm": 1.3622485399246216,
+      "learning_rate": 6.295502293873001e-05,
+      "loss": 5.0872,
+      "step": 226000
+    },
+    {
+      "epoch": 9.547696328457615,
+      "grad_norm": 1.3249783515930176,
+      "learning_rate": 6.286927067701411e-05,
+      "loss": 5.0837,
+      "step": 226500
+    },
+    {
+      "epoch": 9.568772920794165,
+      "grad_norm": 1.3224903345108032,
+      "learning_rate": 6.278351841529821e-05,
+      "loss": 5.0862,
+      "step": 227000
+    },
+    {
+      "epoch": 9.589849513130718,
+      "grad_norm": 1.3264579772949219,
+      "learning_rate": 6.269776615358231e-05,
+      "loss": 5.0852,
+      "step": 227500
+    },
+    {
+      "epoch": 9.610926105467268,
+      "grad_norm": 1.3152449131011963,
+      "learning_rate": 6.26120138918664e-05,
+      "loss": 5.0845,
+      "step": 228000
+    },
+    {
+      "epoch": 9.63200269780382,
+      "grad_norm": 1.3075388669967651,
+      "learning_rate": 6.252626163015049e-05,
+      "loss": 5.0839,
+      "step": 228500
+    },
+    {
+      "epoch": 9.65307929014037,
+      "grad_norm": 1.363019585609436,
+      "learning_rate": 6.24405093684346e-05,
+      "loss": 5.0832,
+      "step": 229000
+    },
+    {
+      "epoch": 9.67415588247692,
+      "grad_norm": 1.340134620666504,
+      "learning_rate": 6.23547571067187e-05,
+      "loss": 5.0807,
+      "step": 229500
+    },
+    {
+      "epoch": 9.695232474813473,
+      "grad_norm": 1.3023189306259155,
+      "learning_rate": 6.226900484500279e-05,
+      "loss": 5.0784,
+      "step": 230000
+    },
+    {
+      "epoch": 9.695232474813473,
+      "eval_accuracy": 0.703861382287927,
+      "eval_loss": 4.804189682006836,
+      "eval_runtime": 451.5155,
+      "eval_samples_per_second": 267.335,
+      "eval_steps_per_second": 2.786,
+      "step": 230000
+    },
+    {
+      "epoch": 9.716309067150023,
+      "grad_norm": 1.3299694061279297,
+      "learning_rate": 6.218325258328688e-05,
+      "loss": 5.0874,
+      "step": 230500
+    },
+    {
+      "epoch": 9.737385659486574,
+      "grad_norm": 1.302130937576294,
+      "learning_rate": 6.209750032157098e-05,
+      "loss": 5.0818,
+      "step": 231000
+    },
+    {
+      "epoch": 9.758462251823126,
+      "grad_norm": 1.3610036373138428,
+      "learning_rate": 6.201174805985508e-05,
+      "loss": 5.0797,
+      "step": 231500
+    },
+    {
+      "epoch": 9.779538844159676,
+      "grad_norm": 1.3892238140106201,
+      "learning_rate": 6.192599579813918e-05,
+      "loss": 5.082,
+      "step": 232000
+    },
+    {
+      "epoch": 9.800615436496226,
+      "grad_norm": 1.3064723014831543,
+      "learning_rate": 6.184024353642328e-05,
+      "loss": 5.0798,
+      "step": 232500
+    },
+    {
+      "epoch": 9.821692028832778,
+      "grad_norm": 1.3203063011169434,
+      "learning_rate": 6.175449127470738e-05,
+      "loss": 5.0772,
+      "step": 233000
+    },
+    {
+      "epoch": 9.842768621169329,
+      "grad_norm": 1.3449581861495972,
+      "learning_rate": 6.166873901299147e-05,
+      "loss": 5.0832,
+      "step": 233500
+    },
+    {
+      "epoch": 9.863845213505881,
+      "grad_norm": 1.3069466352462769,
+      "learning_rate": 6.158298675127557e-05,
+      "loss": 5.0773,
+      "step": 234000
+    },
+    {
+      "epoch": 9.884921805842431,
+      "grad_norm": 1.360684871673584,
+      "learning_rate": 6.149723448955966e-05,
+      "loss": 5.0825,
+      "step": 234500
+    },
+    {
+      "epoch": 9.905998398178982,
+      "grad_norm": 1.3119878768920898,
+      "learning_rate": 6.141148222784377e-05,
+      "loss": 5.0754,
+      "step": 235000
+    },
+    {
+      "epoch": 9.927074990515534,
+      "grad_norm": 1.29421067237854,
+      "learning_rate": 6.132572996612787e-05,
+      "loss": 5.0764,
+      "step": 235500
+    },
+    {
+      "epoch": 9.948151582852084,
+      "grad_norm": 1.3884732723236084,
+      "learning_rate": 6.123997770441195e-05,
+      "loss": 5.0794,
+      "step": 236000
+    },
+    {
+      "epoch": 9.969228175188636,
+      "grad_norm": 1.3366549015045166,
+      "learning_rate": 6.115422544269605e-05,
+      "loss": 5.0819,
+      "step": 236500
+    },
+    {
+      "epoch": 9.990304767525187,
+      "grad_norm": 1.3907825946807861,
+      "learning_rate": 6.106847318098015e-05,
+      "loss": 5.0791,
+      "step": 237000
+    },
+    {
+      "epoch": 10.011381359861737,
+      "grad_norm": 1.2739030122756958,
+      "learning_rate": 6.098272091926425e-05,
+      "loss": 5.0765,
+      "step": 237500
+    },
+    {
+      "epoch": 10.032457952198289,
+      "grad_norm": 1.3331949710845947,
+      "learning_rate": 6.0896968657548345e-05,
+      "loss": 5.0742,
+      "step": 238000
+    },
+    {
+      "epoch": 10.05353454453484,
+      "grad_norm": 1.363141417503357,
+      "learning_rate": 6.081121639583244e-05,
+      "loss": 5.0719,
+      "step": 238500
+    },
+    {
+      "epoch": 10.07461113687139,
+      "grad_norm": 1.3083430528640747,
+      "learning_rate": 6.072546413411654e-05,
+      "loss": 5.0725,
+      "step": 239000
+    },
+    {
+      "epoch": 10.095687729207942,
+      "grad_norm": 1.353929877281189,
+      "learning_rate": 6.063971187240064e-05,
+      "loss": 5.0735,
+      "step": 239500
+    },
+    {
+      "epoch": 10.116764321544492,
+      "grad_norm": 1.3693755865097046,
+      "learning_rate": 6.055395961068473e-05,
+      "loss": 5.0662,
+      "step": 240000
+    },
+    {
+      "epoch": 10.116764321544492,
+      "eval_accuracy": 0.7053069575245285,
+      "eval_loss": 4.797445774078369,
+      "eval_runtime": 470.9165,
+      "eval_samples_per_second": 256.321,
+      "eval_steps_per_second": 2.671,
+      "step": 240000
+    },
+    {
+      "epoch": 10.137840913881044,
+      "grad_norm": 1.3230807781219482,
+      "learning_rate": 6.046820734896883e-05,
+      "loss": 5.0701,
+      "step": 240500
+    },
+    {
+      "epoch": 10.158917506217595,
+      "grad_norm": 1.34675133228302,
+      "learning_rate": 6.038245508725293e-05,
+      "loss": 5.0721,
+      "step": 241000
+    },
+    {
+      "epoch": 10.179994098554145,
+      "grad_norm": 1.313423752784729,
+      "learning_rate": 6.029670282553702e-05,
+      "loss": 5.0702,
+      "step": 241500
+    },
+    {
+      "epoch": 10.201070690890697,
+      "grad_norm": 1.3679618835449219,
+      "learning_rate": 6.0210950563821125e-05,
+      "loss": 5.0729,
+      "step": 242000
+    },
+    {
+      "epoch": 10.222147283227248,
+      "grad_norm": 1.349377155303955,
+      "learning_rate": 6.0125198302105223e-05,
+      "loss": 5.0715,
+      "step": 242500
+    },
+    {
+      "epoch": 10.2432238755638,
+      "grad_norm": 1.3322434425354004,
+      "learning_rate": 6.0039446040389315e-05,
+      "loss": 5.0682,
+      "step": 243000
+    },
+    {
+      "epoch": 10.26430046790035,
+      "grad_norm": 1.3693747520446777,
+      "learning_rate": 5.995369377867341e-05,
+      "loss": 5.0756,
+      "step": 243500
+    },
+    {
+      "epoch": 10.2853770602369,
+      "grad_norm": 1.3259042501449585,
+      "learning_rate": 5.986794151695752e-05,
+      "loss": 5.065,
+      "step": 244000
+    },
+    {
+      "epoch": 10.306453652573452,
+      "grad_norm": 1.3329776525497437,
+      "learning_rate": 5.97821892552416e-05,
+      "loss": 5.0721,
+      "step": 244500
+    },
+    {
+      "epoch": 10.327530244910003,
+      "grad_norm": 1.345353364944458,
+      "learning_rate": 5.969643699352571e-05,
+      "loss": 5.0703,
+      "step": 245000
+    },
+    {
+      "epoch": 10.348606837246553,
+      "grad_norm": 1.3503364324569702,
+      "learning_rate": 5.9610684731809806e-05,
+      "loss": 5.0692,
+      "step": 245500
+    },
+    {
+      "epoch": 10.369683429583105,
+      "grad_norm": 1.287203073501587,
+      "learning_rate": 5.95249324700939e-05,
+      "loss": 5.0674,
+      "step": 246000
+    },
+    {
+      "epoch": 10.390760021919656,
+      "grad_norm": 1.3458172082901,
+      "learning_rate": 5.9439180208377996e-05,
+      "loss": 5.0684,
+      "step": 246500
+    },
+    {
+      "epoch": 10.411836614256208,
+      "grad_norm": 1.276747703552246,
+      "learning_rate": 5.93534279466621e-05,
+      "loss": 5.0668,
+      "step": 247000
+    },
+    {
+      "epoch": 10.432913206592758,
+      "grad_norm": 1.3450318574905396,
+      "learning_rate": 5.9267675684946186e-05,
+      "loss": 5.0695,
+      "step": 247500
+    },
+    {
+      "epoch": 10.453989798929308,
+      "grad_norm": 1.3293246030807495,
+      "learning_rate": 5.918192342323029e-05,
+      "loss": 5.0635,
+      "step": 248000
+    },
+    {
+      "epoch": 10.47506639126586,
+      "grad_norm": 1.352140188217163,
+      "learning_rate": 5.909617116151439e-05,
+      "loss": 5.065,
+      "step": 248500
+    },
+    {
+      "epoch": 10.496142983602411,
+      "grad_norm": 1.4271734952926636,
+      "learning_rate": 5.901041889979848e-05,
+      "loss": 5.0626,
+      "step": 249000
+    },
+    {
+      "epoch": 10.517219575938963,
+      "grad_norm": 1.3162047863006592,
+      "learning_rate": 5.892466663808258e-05,
+      "loss": 5.0638,
+      "step": 249500
+    },
+    {
+      "epoch": 10.538296168275513,
+      "grad_norm": 1.3522886037826538,
+      "learning_rate": 5.8838914376366684e-05,
+      "loss": 5.067,
+      "step": 250000
+    },
+    {
+      "epoch": 10.538296168275513,
+      "eval_accuracy": 0.7062448484961124,
+      "eval_loss": 4.787059783935547,
+      "eval_runtime": 475.218,
+      "eval_samples_per_second": 254.001,
+      "eval_steps_per_second": 2.647,
+      "step": 250000
+    },
+    {
+      "epoch": 10.559372760612064,
+      "grad_norm": 1.3683222532272339,
+      "learning_rate": 5.8753162114650776e-05,
+      "loss": 5.072,
+      "step": 250500
+    },
+    {
+      "epoch": 10.580449352948616,
+      "grad_norm": 1.3752944469451904,
+      "learning_rate": 5.8667409852934874e-05,
+      "loss": 5.0627,
+      "step": 251000
+    },
+    {
+      "epoch": 10.601525945285166,
+      "grad_norm": 1.326973795890808,
+      "learning_rate": 5.858165759121897e-05,
+      "loss": 5.0604,
+      "step": 251500
+    },
+    {
+      "epoch": 10.622602537621717,
+      "grad_norm": 1.3465813398361206,
+      "learning_rate": 5.8495905329503064e-05,
+      "loss": 5.0667,
+      "step": 252000
+    },
+    {
+      "epoch": 10.643679129958269,
+      "grad_norm": 1.3302936553955078,
+      "learning_rate": 5.841015306778716e-05,
+      "loss": 5.065,
+      "step": 252500
+    },
+    {
+      "epoch": 10.664755722294819,
+      "grad_norm": 1.3354437351226807,
+      "learning_rate": 5.832440080607127e-05,
+      "loss": 5.0647,
+      "step": 253000
+    },
+    {
+      "epoch": 10.685832314631371,
+      "grad_norm": 1.3643972873687744,
+      "learning_rate": 5.823864854435536e-05,
+      "loss": 5.0577,
+      "step": 253500
+    },
+    {
+      "epoch": 10.706908906967922,
+      "grad_norm": 1.3446961641311646,
+      "learning_rate": 5.815289628263946e-05,
+      "loss": 5.0617,
+      "step": 254000
+    },
+    {
+      "epoch": 10.727985499304472,
+      "grad_norm": 1.3321641683578491,
+      "learning_rate": 5.8067144020923556e-05,
+      "loss": 5.0665,
+      "step": 254500
+    },
+    {
+      "epoch": 10.749062091641024,
+      "grad_norm": 1.4166176319122314,
+      "learning_rate": 5.798139175920765e-05,
+      "loss": 5.0617,
+      "step": 255000
+    },
+    {
+      "epoch": 10.770138683977574,
+      "grad_norm": 1.3242825269699097,
+      "learning_rate": 5.789563949749175e-05,
+      "loss": 5.0641,
+      "step": 255500
+    },
+    {
+      "epoch": 10.791215276314126,
+      "grad_norm": 1.322081208229065,
+      "learning_rate": 5.780988723577585e-05,
+      "loss": 5.0614,
+      "step": 256000
+    },
+    {
+      "epoch": 10.812291868650677,
+      "grad_norm": 1.3559086322784424,
+      "learning_rate": 5.772413497405994e-05,
+      "loss": 5.0609,
+      "step": 256500
+    },
+    {
+      "epoch": 10.833368460987227,
+      "grad_norm": 1.3242689371109009,
+      "learning_rate": 5.763838271234404e-05,
+      "loss": 5.0547,
+      "step": 257000
+    },
+    {
+      "epoch": 10.85444505332378,
+      "grad_norm": 1.3701753616333008,
+      "learning_rate": 5.7552630450628146e-05,
+      "loss": 5.0604,
+      "step": 257500
+    },
+    {
+      "epoch": 10.87552164566033,
+      "grad_norm": 1.3465561866760254,
+      "learning_rate": 5.746687818891223e-05,
+      "loss": 5.0592,
+      "step": 258000
+    },
+    {
+      "epoch": 10.89659823799688,
+      "grad_norm": 1.361735463142395,
+      "learning_rate": 5.7381125927196335e-05,
+      "loss": 5.0585,
+      "step": 258500
+    },
+    {
+      "epoch": 10.917674830333432,
+      "grad_norm": 1.3532217741012573,
+      "learning_rate": 5.7295373665480434e-05,
+      "loss": 5.0599,
+      "step": 259000
+    },
+    {
+      "epoch": 10.938751422669982,
+      "grad_norm": 1.3646976947784424,
+      "learning_rate": 5.7209621403764525e-05,
+      "loss": 5.0579,
+      "step": 259500
+    },
+    {
+      "epoch": 10.959828015006533,
+      "grad_norm": 1.394838571548462,
+      "learning_rate": 5.7123869142048624e-05,
+      "loss": 5.0545,
+      "step": 260000
+    },
+    {
+      "epoch": 10.959828015006533,
+      "eval_accuracy": 0.7073674774681542,
+      "eval_loss": 4.779219627380371,
+      "eval_runtime": 461.8707,
+      "eval_samples_per_second": 261.342,
+      "eval_steps_per_second": 2.724,
+      "step": 260000
+    },
+    {
+      "epoch": 10.980904607343085,
+      "grad_norm": 1.3068954944610596,
+      "learning_rate": 5.703811688033273e-05,
+      "loss": 5.0539,
+      "step": 260500
+    },
+    {
+      "epoch": 11.001981199679635,
+      "grad_norm": 1.333781361579895,
+      "learning_rate": 5.695236461861681e-05,
+      "loss": 5.0544,
+      "step": 261000
+    },
+    {
+      "epoch": 11.023057792016187,
+      "grad_norm": 1.3785450458526611,
+      "learning_rate": 5.686661235690092e-05,
+      "loss": 5.0486,
+      "step": 261500
+    },
+    {
+      "epoch": 11.044134384352738,
+      "grad_norm": 1.3382635116577148,
+      "learning_rate": 5.678086009518502e-05,
+      "loss": 5.0549,
+      "step": 262000
+    },
+    {
+      "epoch": 11.065210976689288,
+      "grad_norm": 1.352453351020813,
+      "learning_rate": 5.669510783346911e-05,
+      "loss": 5.0574,
+      "step": 262500
+    },
+    {
+      "epoch": 11.08628756902584,
+      "grad_norm": 1.32887864112854,
+      "learning_rate": 5.6609355571753207e-05,
+      "loss": 5.0503,
+      "step": 263000
+    },
+    {
+      "epoch": 11.10736416136239,
+      "grad_norm": 1.3871301412582397,
+      "learning_rate": 5.652360331003731e-05,
+      "loss": 5.0535,
+      "step": 263500
+    },
+    {
+      "epoch": 11.128440753698943,
+      "grad_norm": 1.3213167190551758,
+      "learning_rate": 5.6437851048321396e-05,
+      "loss": 5.05,
+      "step": 264000
+    },
+    {
+      "epoch": 11.149517346035493,
+      "grad_norm": 1.3571499586105347,
+      "learning_rate": 5.63520987866055e-05,
+      "loss": 5.054,
+      "step": 264500
+    },
+    {
+      "epoch": 11.170593938372043,
+      "grad_norm": 1.342622995376587,
+      "learning_rate": 5.62663465248896e-05,
+      "loss": 5.0503,
+      "step": 265000
+    },
+    {
+      "epoch": 11.191670530708596,
+      "grad_norm": 1.3691377639770508,
+      "learning_rate": 5.618059426317369e-05,
+      "loss": 5.0499,
+      "step": 265500
+    },
+    {
+      "epoch": 11.212747123045146,
+      "grad_norm": 1.3681830167770386,
+      "learning_rate": 5.609484200145779e-05,
+      "loss": 5.0554,
+      "step": 266000
+    },
+    {
+      "epoch": 11.233823715381696,
+      "grad_norm": 1.342695951461792,
+      "learning_rate": 5.6009089739741895e-05,
+      "loss": 5.0528,
+      "step": 266500
+    },
+    {
+      "epoch": 11.254900307718248,
+      "grad_norm": 1.3733820915222168,
+      "learning_rate": 5.5923337478025986e-05,
+      "loss": 5.0496,
+      "step": 267000
+    },
+    {
+      "epoch": 11.275976900054799,
+      "grad_norm": 1.418822169303894,
+      "learning_rate": 5.5837585216310085e-05,
+      "loss": 5.0526,
+      "step": 267500
+    },
+    {
+      "epoch": 11.29705349239135,
+      "grad_norm": 1.3996915817260742,
+      "learning_rate": 5.575183295459418e-05,
+      "loss": 5.0477,
+      "step": 268000
+    },
+    {
+      "epoch": 11.318130084727901,
+      "grad_norm": 1.3652806282043457,
+      "learning_rate": 5.5666080692878274e-05,
+      "loss": 5.0503,
+      "step": 268500
+    },
+    {
+      "epoch": 11.339206677064452,
+      "grad_norm": 1.3066692352294922,
+      "learning_rate": 5.558032843116238e-05,
+      "loss": 5.0513,
+      "step": 269000
+    },
+    {
+      "epoch": 11.360283269401004,
+      "grad_norm": 1.3446789979934692,
+      "learning_rate": 5.549457616944648e-05,
+      "loss": 5.0513,
+      "step": 269500
+    },
+    {
+      "epoch": 11.381359861737554,
+      "grad_norm": 1.34601628780365,
+      "learning_rate": 5.540882390773057e-05,
+      "loss": 5.0461,
+      "step": 270000
+    },
+    {
+      "epoch": 11.381359861737554,
+      "eval_accuracy": 0.7081627386649615,
+      "eval_loss": 4.776198387145996,
+      "eval_runtime": 454.2427,
+      "eval_samples_per_second": 265.73,
+      "eval_steps_per_second": 2.769,
+      "step": 270000
+    },
+    {
+      "epoch": 11.402436454074106,
+      "grad_norm": 1.3677948713302612,
+      "learning_rate": 5.532307164601467e-05,
+      "loss": 5.0505,
+      "step": 270500
+    },
+    {
+      "epoch": 11.423513046410656,
+      "grad_norm": 1.3758268356323242,
+      "learning_rate": 5.5237319384298766e-05,
+      "loss": 5.0477,
+      "step": 271000
+    },
+    {
+      "epoch": 11.444589638747207,
+      "grad_norm": 1.3345792293548584,
+      "learning_rate": 5.515156712258286e-05,
+      "loss": 5.0447,
+      "step": 271500
+    },
+    {
+      "epoch": 11.465666231083759,
+      "grad_norm": 1.3229191303253174,
+      "learning_rate": 5.506581486086696e-05,
+      "loss": 5.0478,
+      "step": 272000
+    },
+    {
+      "epoch": 11.48674282342031,
+      "grad_norm": 1.3337035179138184,
+      "learning_rate": 5.498006259915105e-05,
+      "loss": 5.0465,
+      "step": 272500
+    },
+    {
+      "epoch": 11.50781941575686,
+      "grad_norm": 1.3180122375488281,
+      "learning_rate": 5.489431033743515e-05,
+      "loss": 5.0466,
+      "step": 273000
+    },
+    {
+      "epoch": 11.528896008093412,
+      "grad_norm": 1.3044873476028442,
+      "learning_rate": 5.480855807571925e-05,
+      "loss": 5.0493,
+      "step": 273500
+    },
+    {
+      "epoch": 11.549972600429962,
+      "grad_norm": 1.3265283107757568,
+      "learning_rate": 5.472280581400334e-05,
+      "loss": 5.0457,
+      "step": 274000
+    },
+    {
+      "epoch": 11.571049192766514,
+      "grad_norm": 1.3854825496673584,
+      "learning_rate": 5.463705355228744e-05,
+      "loss": 5.0454,
+      "step": 274500
+    },
+    {
+      "epoch": 11.592125785103065,
+      "grad_norm": 1.3951648473739624,
+      "learning_rate": 5.4551301290571546e-05,
+      "loss": 5.0488,
+      "step": 275000
+    },
+    {
+      "epoch": 11.613202377439615,
+      "grad_norm": 1.3733264207839966,
+      "learning_rate": 5.446554902885563e-05,
+      "loss": 5.0458,
+      "step": 275500
+    },
+    {
+      "epoch": 11.634278969776167,
+      "grad_norm": 1.3279320001602173,
+      "learning_rate": 5.4379796767139736e-05,
+      "loss": 5.0422,
+      "step": 276000
+    },
+    {
+      "epoch": 11.655355562112717,
+      "grad_norm": 1.3722243309020996,
+      "learning_rate": 5.4294044505423834e-05,
+      "loss": 5.042,
+      "step": 276500
+    },
+    {
+      "epoch": 11.67643215444927,
+      "grad_norm": 1.353024959564209,
+      "learning_rate": 5.4208292243707925e-05,
+      "loss": 5.0439,
+      "step": 277000
+    },
+    {
+      "epoch": 11.69750874678582,
+      "grad_norm": 1.3496346473693848,
+      "learning_rate": 5.4122539981992024e-05,
+      "loss": 5.0445,
+      "step": 277500
+    },
+    {
+      "epoch": 11.71858533912237,
+      "grad_norm": 1.3232852220535278,
+      "learning_rate": 5.403678772027613e-05,
+      "loss": 5.0387,
+      "step": 278000
+    },
+    {
+      "epoch": 11.739661931458922,
+      "grad_norm": 1.3865522146224976,
+      "learning_rate": 5.395103545856022e-05,
+      "loss": 5.043,
+      "step": 278500
+    },
+    {
+      "epoch": 11.760738523795473,
+      "grad_norm": 1.3305288553237915,
+      "learning_rate": 5.386528319684432e-05,
+      "loss": 5.0393,
+      "step": 279000
+    },
+    {
+      "epoch": 11.781815116132023,
+      "grad_norm": 1.3299384117126465,
+      "learning_rate": 5.377953093512842e-05,
+      "loss": 5.0371,
+      "step": 279500
+    },
+    {
+      "epoch": 11.802891708468575,
+      "grad_norm": 1.360175609588623,
+      "learning_rate": 5.369377867341251e-05,
+      "loss": 5.0456,
+      "step": 280000
+    },
+    {
+      "epoch": 11.802891708468575,
+      "eval_accuracy": 0.7092983645812267,
+      "eval_loss": 4.7662529945373535,
+      "eval_runtime": 454.4278,
+      "eval_samples_per_second": 265.622,
+      "eval_steps_per_second": 2.768,
+      "step": 280000
+    },
+    {
+      "epoch": 11.823968300805126,
+      "grad_norm": 1.3606709241867065,
+      "learning_rate": 5.3608026411696614e-05,
+      "loss": 5.046,
+      "step": 280500
+    },
+    {
+      "epoch": 11.845044893141678,
+      "grad_norm": 1.351615309715271,
+      "learning_rate": 5.352227414998071e-05,
+      "loss": 5.0353,
+      "step": 281000
+    },
+    {
+      "epoch": 11.866121485478228,
+      "grad_norm": 1.405328631401062,
+      "learning_rate": 5.34365218882648e-05,
+      "loss": 5.0387,
+      "step": 281500
+    },
+    {
+      "epoch": 11.887198077814778,
+      "grad_norm": 1.3197768926620483,
+      "learning_rate": 5.33507696265489e-05,
+      "loss": 5.0336,
+      "step": 282000
+    },
+    {
+      "epoch": 11.90827467015133,
+      "grad_norm": 1.3467516899108887,
+      "learning_rate": 5.3265017364833e-05,
+      "loss": 5.0428,
+      "step": 282500
+    },
+    {
+      "epoch": 11.92935126248788,
+      "grad_norm": 1.3832674026489258,
+      "learning_rate": 5.317926510311709e-05,
+      "loss": 5.0375,
+      "step": 283000
+    },
+    {
+      "epoch": 11.950427854824433,
+      "grad_norm": 1.3843154907226562,
+      "learning_rate": 5.3093512841401197e-05,
+      "loss": 5.0373,
+      "step": 283500
+    },
+    {
+      "epoch": 11.971504447160983,
+      "grad_norm": 1.3506910800933838,
+      "learning_rate": 5.3007760579685295e-05,
+      "loss": 5.0393,
+      "step": 284000
+    },
+    {
+      "epoch": 11.992581039497534,
+      "grad_norm": 1.3743553161621094,
+      "learning_rate": 5.2922008317969386e-05,
+      "loss": 5.0436,
+      "step": 284500
+    },
+    {
+      "epoch": 12.013657631834086,
+      "grad_norm": 1.36726713180542,
+      "learning_rate": 5.2836256056253485e-05,
+      "loss": 5.0357,
+      "step": 285000
+    },
+    {
+      "epoch": 12.034734224170636,
+      "grad_norm": 1.3825244903564453,
+      "learning_rate": 5.275050379453759e-05,
+      "loss": 5.0381,
+      "step": 285500
+    },
+    {
+      "epoch": 12.055810816507186,
+      "grad_norm": 1.3317618370056152,
+      "learning_rate": 5.2664751532821675e-05,
+      "loss": 5.0341,
+      "step": 286000
+    },
+    {
+      "epoch": 12.076887408843739,
+      "grad_norm": 1.362870454788208,
+      "learning_rate": 5.257899927110578e-05,
+      "loss": 5.0343,
+      "step": 286500
+    },
+    {
+      "epoch": 12.097964001180289,
+      "grad_norm": 1.3647220134735107,
+      "learning_rate": 5.249324700938988e-05,
+      "loss": 5.0316,
+      "step": 287000
+    },
+    {
+      "epoch": 12.119040593516841,
+      "grad_norm": 1.3661941289901733,
+      "learning_rate": 5.240749474767397e-05,
+      "loss": 5.0323,
+      "step": 287500
+    },
+    {
+      "epoch": 12.140117185853391,
+      "grad_norm": 1.3798184394836426,
+      "learning_rate": 5.232174248595807e-05,
+      "loss": 5.0363,
+      "step": 288000
+    },
+    {
+      "epoch": 12.161193778189942,
+      "grad_norm": 1.3126131296157837,
+      "learning_rate": 5.223599022424217e-05,
+      "loss": 5.0362,
+      "step": 288500
+    },
+    {
+      "epoch": 12.182270370526494,
+      "grad_norm": 1.3465566635131836,
+      "learning_rate": 5.215023796252626e-05,
+      "loss": 5.0398,
+      "step": 289000
+    },
+    {
+      "epoch": 12.203346962863044,
+      "grad_norm": 1.3684533834457397,
+      "learning_rate": 5.206448570081036e-05,
+      "loss": 5.0296,
+      "step": 289500
+    },
+    {
+      "epoch": 12.224423555199595,
+      "grad_norm": 1.387593388557434,
+      "learning_rate": 5.197873343909446e-05,
+      "loss": 5.0294,
+      "step": 290000
+    },
+    {
+      "epoch": 12.224423555199595,
+      "eval_accuracy": 0.7103016049101925,
+      "eval_loss": 4.759915828704834,
+      "eval_runtime": 428.0669,
+      "eval_samples_per_second": 281.979,
+      "eval_steps_per_second": 2.939,
+      "step": 290000
+    },
+    {
+      "epoch": 12.245500147536147,
+      "grad_norm": 1.4253774881362915,
+      "learning_rate": 5.189298117737855e-05,
+      "loss": 5.0316,
+      "step": 290500
+    },
+    {
+      "epoch": 12.266576739872697,
+      "grad_norm": 1.3434678316116333,
+      "learning_rate": 5.180722891566265e-05,
+      "loss": 5.0308,
+      "step": 291000
+    },
+    {
+      "epoch": 12.28765333220925,
+      "grad_norm": 1.3824933767318726,
+      "learning_rate": 5.1721476653946756e-05,
+      "loss": 5.0268,
+      "step": 291500
+    },
+    {
+      "epoch": 12.3087299245458,
+      "grad_norm": 1.33455491065979,
+      "learning_rate": 5.163572439223085e-05,
+      "loss": 5.0314,
+      "step": 292000
+    },
+    {
+      "epoch": 12.32980651688235,
+      "grad_norm": 1.3667352199554443,
+      "learning_rate": 5.1549972130514946e-05,
+      "loss": 5.0319,
+      "step": 292500
+    },
+    {
+      "epoch": 12.350883109218902,
+      "grad_norm": 1.392549991607666,
+      "learning_rate": 5.1464219868799044e-05,
+      "loss": 5.0319,
+      "step": 293000
+    },
+    {
+      "epoch": 12.371959701555452,
+      "grad_norm": 1.377046823501587,
+      "learning_rate": 5.1378467607083136e-05,
+      "loss": 5.0292,
+      "step": 293500
+    },
+    {
+      "epoch": 12.393036293892003,
+      "grad_norm": 1.3356010913848877,
+      "learning_rate": 5.1292715345367234e-05,
+      "loss": 5.0309,
+      "step": 294000
+    },
+    {
+      "epoch": 12.414112886228555,
+      "grad_norm": 1.4097566604614258,
+      "learning_rate": 5.120696308365134e-05,
+      "loss": 5.0282,
+      "step": 294500
+    },
+    {
+      "epoch": 12.435189478565105,
+      "grad_norm": 1.38156259059906,
+      "learning_rate": 5.112121082193543e-05,
+      "loss": 5.0312,
+      "step": 295000
+    },
+    {
+      "epoch": 12.456266070901657,
+      "grad_norm": 1.37826406955719,
+      "learning_rate": 5.103545856021953e-05,
+      "loss": 5.0283,
+      "step": 295500
+    },
+    {
+      "epoch": 12.477342663238208,
+      "grad_norm": 1.3887619972229004,
+      "learning_rate": 5.094970629850363e-05,
+      "loss": 5.0301,
+      "step": 296000
+    },
+    {
+      "epoch": 12.498419255574758,
+      "grad_norm": 1.3714841604232788,
+      "learning_rate": 5.086395403678772e-05,
+      "loss": 5.029,
+      "step": 296500
+    },
+    {
+      "epoch": 12.51949584791131,
+      "grad_norm": 1.3779914379119873,
+      "learning_rate": 5.0778201775071824e-05,
+      "loss": 5.0358,
+      "step": 297000
+    },
+    {
+      "epoch": 12.54057244024786,
+      "grad_norm": 1.383215069770813,
+      "learning_rate": 5.069244951335592e-05,
+      "loss": 5.0319,
+      "step": 297500
+    },
+    {
+      "epoch": 12.561649032584413,
+      "grad_norm": 1.3689595460891724,
+      "learning_rate": 5.0606697251640014e-05,
+      "loss": 5.0249,
+      "step": 298000
+    },
+    {
+      "epoch": 12.582725624920963,
+      "grad_norm": 1.4173258543014526,
+      "learning_rate": 5.052094498992411e-05,
+      "loss": 5.0311,
+      "step": 298500
+    },
+    {
+      "epoch": 12.603802217257513,
+      "grad_norm": 1.4003221988677979,
+      "learning_rate": 5.043519272820822e-05,
+      "loss": 5.0302,
+      "step": 299000
+    },
+    {
+      "epoch": 12.624878809594065,
+      "grad_norm": 1.331686019897461,
+      "learning_rate": 5.03494404664923e-05,
+      "loss": 5.0235,
+      "step": 299500
+    },
+    {
+      "epoch": 12.645955401930616,
+      "grad_norm": 1.431599736213684,
+      "learning_rate": 5.026368820477641e-05,
+      "loss": 5.0258,
+      "step": 300000
+    },
+    {
+      "epoch": 12.645955401930616,
+      "eval_accuracy": 0.7113456165124568,
+      "eval_loss": 4.752755165100098,
+      "eval_runtime": 443.4827,
+      "eval_samples_per_second": 272.177,
+      "eval_steps_per_second": 2.837,
+      "step": 300000
+    },
+    {
+      "epoch": 12.667031994267166,
+      "grad_norm": 1.3943068981170654,
+      "learning_rate": 5.0177935943060505e-05,
+      "loss": 5.0231,
+      "step": 300500
+    },
+    {
+      "epoch": 12.688108586603718,
+      "grad_norm": 1.3815979957580566,
+      "learning_rate": 5.00921836813446e-05,
+      "loss": 5.0278,
+      "step": 301000
+    },
+    {
+      "epoch": 12.709185178940269,
+      "grad_norm": 1.384202003479004,
+      "learning_rate": 5.0006431419628695e-05,
+      "loss": 5.0262,
+      "step": 301500
+    },
+    {
+      "epoch": 12.73026177127682,
+      "grad_norm": 1.3606541156768799,
+      "learning_rate": 4.992067915791279e-05,
+      "loss": 5.0266,
+      "step": 302000
+    },
+    {
+      "epoch": 12.751338363613371,
+      "grad_norm": 1.3602447509765625,
+      "learning_rate": 4.9834926896196885e-05,
+      "loss": 5.0202,
+      "step": 302500
+    },
+    {
+      "epoch": 12.772414955949921,
+      "grad_norm": 1.4164466857910156,
+      "learning_rate": 4.974917463448099e-05,
+      "loss": 5.0269,
+      "step": 303000
+    },
+    {
+      "epoch": 12.793491548286474,
+      "grad_norm": 1.3614903688430786,
+      "learning_rate": 4.966342237276508e-05,
+      "loss": 5.0279,
+      "step": 303500
+    },
+    {
+      "epoch": 12.814568140623024,
+      "grad_norm": 1.3349113464355469,
+      "learning_rate": 4.957767011104918e-05,
+      "loss": 5.0261,
+      "step": 304000
+    },
+    {
+      "epoch": 12.835644732959576,
+      "grad_norm": 1.3644523620605469,
+      "learning_rate": 4.949191784933328e-05,
+      "loss": 5.0254,
+      "step": 304500
+    },
+    {
+      "epoch": 12.856721325296126,
+      "grad_norm": 1.3246243000030518,
+      "learning_rate": 4.9406165587617376e-05,
+      "loss": 5.0232,
+      "step": 305000
+    },
+    {
+      "epoch": 12.877797917632677,
+      "grad_norm": 1.3864765167236328,
+      "learning_rate": 4.932041332590147e-05,
+      "loss": 5.0265,
+      "step": 305500
+    },
+    {
+      "epoch": 12.898874509969229,
+      "grad_norm": 1.3756834268569946,
+      "learning_rate": 4.923466106418557e-05,
+      "loss": 5.0218,
+      "step": 306000
+    },
+    {
+      "epoch": 12.91995110230578,
+      "grad_norm": 1.3684511184692383,
+      "learning_rate": 4.9148908802469665e-05,
+      "loss": 5.0204,
+      "step": 306500
+    },
+    {
+      "epoch": 12.94102769464233,
+      "grad_norm": 1.3413596153259277,
+      "learning_rate": 4.906315654075376e-05,
+      "loss": 5.022,
+      "step": 307000
+    },
+    {
+      "epoch": 12.962104286978882,
+      "grad_norm": 1.4269444942474365,
+      "learning_rate": 4.897740427903786e-05,
+      "loss": 5.0263,
+      "step": 307500
+    },
+    {
+      "epoch": 12.983180879315432,
+      "grad_norm": 1.398376226425171,
+      "learning_rate": 4.889165201732196e-05,
+      "loss": 5.0286,
+      "step": 308000
+    },
+    {
+      "epoch": 13.004257471651984,
+      "grad_norm": 1.3307123184204102,
+      "learning_rate": 4.880589975560606e-05,
+      "loss": 5.021,
+      "step": 308500
+    },
+    {
+      "epoch": 13.025334063988534,
+      "grad_norm": 1.403230905532837,
+      "learning_rate": 4.8720147493890156e-05,
+      "loss": 5.0142,
+      "step": 309000
+    },
+    {
+      "epoch": 13.046410656325085,
+      "grad_norm": 1.3675681352615356,
+      "learning_rate": 4.8634395232174254e-05,
+      "loss": 5.0209,
+      "step": 309500
+    },
+    {
+      "epoch": 13.067487248661637,
+      "grad_norm": 1.3374537229537964,
+      "learning_rate": 4.8548642970458346e-05,
+      "loss": 5.0149,
+      "step": 310000
+    },
+    {
+      "epoch": 13.067487248661637,
+      "eval_accuracy": 0.7122795235376458,
+      "eval_loss": 4.746424674987793,
+      "eval_runtime": 589.7418,
+      "eval_samples_per_second": 204.676,
+      "eval_steps_per_second": 2.133,
+      "step": 310000
+    },
+    {
+      "epoch": 13.088563840998187,
+      "grad_norm": 1.396132469177246,
+      "learning_rate": 4.846289070874245e-05,
+      "loss": 5.0175,
+      "step": 310500
+    },
+    {
+      "epoch": 13.109640433334738,
+      "grad_norm": 1.3938826322555542,
+      "learning_rate": 4.837713844702654e-05,
+      "loss": 5.0179,
+      "step": 311000
+    },
+    {
+      "epoch": 13.13071702567129,
+      "grad_norm": 1.3980175256729126,
+      "learning_rate": 4.829138618531064e-05,
+      "loss": 5.0147,
+      "step": 311500
+    },
+    {
+      "epoch": 13.15179361800784,
+      "grad_norm": 1.3269809484481812,
+      "learning_rate": 4.820563392359474e-05,
+      "loss": 5.0188,
+      "step": 312000
+    },
+    {
+      "epoch": 13.172870210344392,
+      "grad_norm": 1.3787254095077515,
+      "learning_rate": 4.811988166187884e-05,
+      "loss": 5.0173,
+      "step": 312500
+    },
+    {
+      "epoch": 13.193946802680943,
+      "grad_norm": 1.3363980054855347,
+      "learning_rate": 4.803412940016293e-05,
+      "loss": 5.0161,
+      "step": 313000
+    },
+    {
+      "epoch": 13.215023395017493,
+      "grad_norm": 1.3899754285812378,
+      "learning_rate": 4.7948377138447034e-05,
+      "loss": 5.0174,
+      "step": 313500
+    },
+    {
+      "epoch": 13.236099987354045,
+      "grad_norm": 1.3810607194900513,
+      "learning_rate": 4.7862624876731126e-05,
+      "loss": 5.0146,
+      "step": 314000
+    },
+    {
+      "epoch": 13.257176579690595,
+      "grad_norm": 1.3974484205245972,
+      "learning_rate": 4.7776872615015224e-05,
+      "loss": 5.0143,
+      "step": 314500
+    },
+    {
+      "epoch": 13.278253172027147,
+      "grad_norm": 1.4100830554962158,
+      "learning_rate": 4.7691120353299315e-05,
+      "loss": 5.0109,
+      "step": 315000
+    },
+    {
+      "epoch": 13.299329764363698,
+      "grad_norm": 1.3764352798461914,
+      "learning_rate": 4.760536809158342e-05,
+      "loss": 5.0195,
+      "step": 315500
+    },
+    {
+      "epoch": 13.320406356700248,
+      "grad_norm": 1.333310604095459,
+      "learning_rate": 4.751961582986751e-05,
+      "loss": 5.0177,
+      "step": 316000
+    },
+    {
+      "epoch": 13.3414829490368,
+      "grad_norm": 1.376098394393921,
+      "learning_rate": 4.743386356815161e-05,
+      "loss": 5.0175,
+      "step": 316500
+    },
+    {
+      "epoch": 13.36255954137335,
+      "grad_norm": 1.3715081214904785,
+      "learning_rate": 4.734811130643571e-05,
+      "loss": 5.015,
+      "step": 317000
+    },
+    {
+      "epoch": 13.383636133709901,
+      "grad_norm": 1.383402943611145,
+      "learning_rate": 4.726235904471981e-05,
+      "loss": 5.0131,
+      "step": 317500
+    },
+    {
+      "epoch": 13.404712726046453,
+      "grad_norm": 1.4268983602523804,
+      "learning_rate": 4.71766067830039e-05,
+      "loss": 5.014,
+      "step": 318000
+    },
+    {
+      "epoch": 13.425789318383003,
+      "grad_norm": 1.3433964252471924,
+      "learning_rate": 4.7090854521288004e-05,
+      "loss": 5.016,
+      "step": 318500
+    },
+    {
+      "epoch": 13.446865910719556,
+      "grad_norm": 1.3966625928878784,
+      "learning_rate": 4.7005102259572095e-05,
+      "loss": 5.0143,
+      "step": 319000
+    },
+    {
+      "epoch": 13.467942503056106,
+      "grad_norm": 1.4121098518371582,
+      "learning_rate": 4.6919349997856193e-05,
+      "loss": 5.0165,
+      "step": 319500
+    },
+    {
+      "epoch": 13.489019095392656,
+      "grad_norm": 1.4301269054412842,
+      "learning_rate": 4.683359773614029e-05,
+      "loss": 5.0114,
+      "step": 320000
+    },
+    {
+      "epoch": 13.489019095392656,
+      "eval_accuracy": 0.7130796372816967,
+      "eval_loss": 4.741966724395752,
+      "eval_runtime": 464.5693,
+      "eval_samples_per_second": 259.823,
+      "eval_steps_per_second": 2.708,
+      "step": 320000
+    },
+    {
+      "epoch": 13.510095687729208,
+      "grad_norm": 1.385047435760498,
+      "learning_rate": 4.674784547442439e-05,
+      "loss": 5.0177,
+      "step": 320500
+    },
+    {
+      "epoch": 13.531172280065759,
+      "grad_norm": 1.364404559135437,
+      "learning_rate": 4.666209321270849e-05,
+      "loss": 5.0147,
+      "step": 321000
+    },
+    {
+      "epoch": 13.55224887240231,
+      "grad_norm": 1.3469067811965942,
+      "learning_rate": 4.657634095099259e-05,
+      "loss": 5.0147,
+      "step": 321500
+    },
+    {
+      "epoch": 13.573325464738861,
+      "grad_norm": 1.4495978355407715,
+      "learning_rate": 4.6490588689276685e-05,
+      "loss": 5.0134,
+      "step": 322000
+    },
+    {
+      "epoch": 13.594402057075412,
+      "grad_norm": 1.3800907135009766,
+      "learning_rate": 4.6404836427560777e-05,
+      "loss": 5.0148,
+      "step": 322500
+    },
+    {
+      "epoch": 13.615478649411964,
+      "grad_norm": 1.4272035360336304,
+      "learning_rate": 4.6319084165844875e-05,
+      "loss": 5.0088,
+      "step": 323000
+    },
+    {
+      "epoch": 13.636555241748514,
+      "grad_norm": 1.3970749378204346,
+      "learning_rate": 4.623333190412897e-05,
+      "loss": 5.0148,
+      "step": 323500
+    },
+    {
+      "epoch": 13.657631834085064,
+      "grad_norm": 1.4643702507019043,
+      "learning_rate": 4.614757964241307e-05,
+      "loss": 5.0086,
+      "step": 324000
+    },
+    {
+      "epoch": 13.678708426421617,
+      "grad_norm": 1.3747425079345703,
+      "learning_rate": 4.606182738069717e-05,
+      "loss": 5.0122,
+      "step": 324500
+    },
+    {
+      "epoch": 13.699785018758167,
+      "grad_norm": 1.3796627521514893,
+      "learning_rate": 4.597607511898127e-05,
+      "loss": 5.0114,
+      "step": 325000
+    },
+    {
+      "epoch": 13.720861611094719,
+      "grad_norm": 1.381720781326294,
+      "learning_rate": 4.589032285726536e-05,
+      "loss": 5.0114,
+      "step": 325500
+    },
+    {
+      "epoch": 13.74193820343127,
+      "grad_norm": 1.3596237897872925,
+      "learning_rate": 4.5804570595549465e-05,
+      "loss": 5.0112,
+      "step": 326000
+    },
+    {
+      "epoch": 13.76301479576782,
+      "grad_norm": 1.392291784286499,
+      "learning_rate": 4.5718818333833556e-05,
+      "loss": 5.0141,
+      "step": 326500
+    },
+    {
+      "epoch": 13.784091388104372,
+      "grad_norm": 1.3536847829818726,
+      "learning_rate": 4.5633066072117655e-05,
+      "loss": 5.0142,
+      "step": 327000
+    },
+    {
+      "epoch": 13.805167980440922,
+      "grad_norm": 1.347533941268921,
+      "learning_rate": 4.554731381040175e-05,
+      "loss": 5.0073,
+      "step": 327500
+    },
+    {
+      "epoch": 13.826244572777473,
+      "grad_norm": 1.4066452980041504,
+      "learning_rate": 4.546156154868585e-05,
+      "loss": 5.01,
+      "step": 328000
+    },
+    {
+      "epoch": 13.847321165114025,
+      "grad_norm": 1.452540397644043,
+      "learning_rate": 4.537580928696994e-05,
+      "loss": 5.0099,
+      "step": 328500
+    },
+    {
+      "epoch": 13.868397757450575,
+      "grad_norm": 1.409185767173767,
+      "learning_rate": 4.529005702525405e-05,
+      "loss": 5.0084,
+      "step": 329000
+    },
+    {
+      "epoch": 13.889474349787127,
+      "grad_norm": 1.4291402101516724,
+      "learning_rate": 4.520430476353814e-05,
+      "loss": 5.0082,
+      "step": 329500
+    },
+    {
+      "epoch": 13.910550942123677,
+      "grad_norm": 1.4286351203918457,
+      "learning_rate": 4.511855250182224e-05,
+      "loss": 5.0086,
+      "step": 330000
+    },
+    {
+      "epoch": 13.910550942123677,
+      "eval_accuracy": 0.713665210144436,
+      "eval_loss": 4.737768650054932,
+      "eval_runtime": 418.1806,
+      "eval_samples_per_second": 288.646,
+      "eval_steps_per_second": 3.008,
+      "step": 330000
+    },
+    {
+      "epoch": 13.931627534460228,
+      "grad_norm": 1.3239811658859253,
+      "learning_rate": 4.5032800240106336e-05,
+      "loss": 5.0091,
+      "step": 330500
+    },
+    {
+      "epoch": 13.95270412679678,
+      "grad_norm": 1.3621129989624023,
+      "learning_rate": 4.4947047978390434e-05,
+      "loss": 5.0131,
+      "step": 331000
+    },
+    {
+      "epoch": 13.97378071913333,
+      "grad_norm": 1.4036415815353394,
+      "learning_rate": 4.4861295716674526e-05,
+      "loss": 5.0101,
+      "step": 331500
+    },
+    {
+      "epoch": 13.994857311469882,
+      "grad_norm": 1.3839472532272339,
+      "learning_rate": 4.4775543454958624e-05,
+      "loss": 5.0116,
+      "step": 332000
+    },
+    {
+      "epoch": 14.015933903806433,
+      "grad_norm": 1.375698208808899,
+      "learning_rate": 4.468979119324272e-05,
+      "loss": 5.0032,
+      "step": 332500
+    },
+    {
+      "epoch": 14.037010496142983,
+      "grad_norm": 1.376570701599121,
+      "learning_rate": 4.460403893152682e-05,
+      "loss": 5.0035,
+      "step": 333000
+    },
+    {
+      "epoch": 14.058087088479535,
+      "grad_norm": 1.4007188081741333,
+      "learning_rate": 4.451828666981092e-05,
+      "loss": 5.0045,
+      "step": 333500
+    },
+    {
+      "epoch": 14.079163680816086,
+      "grad_norm": 1.3995590209960938,
+      "learning_rate": 4.443253440809502e-05,
+      "loss": 5.004,
+      "step": 334000
+    },
+    {
+      "epoch": 14.100240273152636,
+      "grad_norm": 1.3386423587799072,
+      "learning_rate": 4.4346782146379116e-05,
+      "loss": 5.0021,
+      "step": 334500
+    },
+    {
+      "epoch": 14.121316865489188,
+      "grad_norm": 1.3641568422317505,
+      "learning_rate": 4.426102988466321e-05,
+      "loss": 5.0022,
+      "step": 335000
+    },
+    {
+      "epoch": 14.142393457825738,
+      "grad_norm": 1.3644893169403076,
+      "learning_rate": 4.4175277622947305e-05,
+      "loss": 4.9968,
+      "step": 335500
+    },
+    {
+      "epoch": 14.16347005016229,
+      "grad_norm": 1.4048681259155273,
+      "learning_rate": 4.4089525361231404e-05,
+      "loss": 5.0031,
+      "step": 336000
+    },
+    {
+      "epoch": 14.184546642498841,
+      "grad_norm": 1.3581401109695435,
+      "learning_rate": 4.40037730995155e-05,
+      "loss": 4.9981,
+      "step": 336500
+    },
+    {
+      "epoch": 14.205623234835391,
+      "grad_norm": 1.4001336097717285,
+      "learning_rate": 4.39180208377996e-05,
+      "loss": 5.0027,
+      "step": 337000
+    },
+    {
+      "epoch": 14.226699827171943,
+      "grad_norm": 1.4044684171676636,
+      "learning_rate": 4.38322685760837e-05,
+      "loss": 5.004,
+      "step": 337500
+    },
+    {
+      "epoch": 14.247776419508494,
+      "grad_norm": 1.3862544298171997,
+      "learning_rate": 4.374651631436779e-05,
+      "loss": 5.0017,
+      "step": 338000
+    },
+    {
+      "epoch": 14.268853011845044,
+      "grad_norm": 1.4099310636520386,
+      "learning_rate": 4.3660764052651895e-05,
+      "loss": 5.0034,
+      "step": 338500
+    },
+    {
+      "epoch": 14.289929604181596,
+      "grad_norm": 1.4048449993133545,
+      "learning_rate": 4.357501179093599e-05,
+      "loss": 5.0028,
+      "step": 339000
+    },
+    {
+      "epoch": 14.311006196518147,
+      "grad_norm": 1.3789886236190796,
+      "learning_rate": 4.3489259529220085e-05,
+      "loss": 5.0009,
+      "step": 339500
+    },
+    {
+      "epoch": 14.332082788854699,
+      "grad_norm": 1.4323118925094604,
+      "learning_rate": 4.3403507267504183e-05,
+      "loss": 5.004,
+      "step": 340000
+    },
+    {
+      "epoch": 14.332082788854699,
+      "eval_accuracy": 0.7147425470012128,
+      "eval_loss": 4.731046199798584,
+      "eval_runtime": 445.013,
+      "eval_samples_per_second": 271.242,
+      "eval_steps_per_second": 2.827,
+      "step": 340000
+    },
+    {
+      "epoch": 14.353159381191249,
+      "grad_norm": 1.4047952890396118,
+      "learning_rate": 4.331775500578828e-05,
+      "loss": 4.9998,
+      "step": 340500
+    },
+    {
+      "epoch": 14.3742359735278,
+      "grad_norm": 1.3912140130996704,
+      "learning_rate": 4.323200274407237e-05,
+      "loss": 5.0007,
+      "step": 341000
+    },
+    {
+      "epoch": 14.395312565864351,
+      "grad_norm": 1.4339492321014404,
+      "learning_rate": 4.314625048235648e-05,
+      "loss": 4.9979,
+      "step": 341500
+    },
+    {
+      "epoch": 14.416389158200902,
+      "grad_norm": 1.3824228048324585,
+      "learning_rate": 4.306049822064057e-05,
+      "loss": 5.0037,
+      "step": 342000
+    },
+    {
+      "epoch": 14.437465750537454,
+      "grad_norm": 1.3840168714523315,
+      "learning_rate": 4.297474595892467e-05,
+      "loss": 5.0021,
+      "step": 342500
+    },
+    {
+      "epoch": 14.458542342874004,
+      "grad_norm": 1.3867284059524536,
+      "learning_rate": 4.2888993697208767e-05,
+      "loss": 4.9979,
+      "step": 343000
+    },
+    {
+      "epoch": 14.479618935210555,
+      "grad_norm": 1.3405182361602783,
+      "learning_rate": 4.2803241435492865e-05,
+      "loss": 5.0001,
+      "step": 343500
+    },
+    {
+      "epoch": 14.500695527547107,
+      "grad_norm": 1.3907437324523926,
+      "learning_rate": 4.2717489173776956e-05,
+      "loss": 5.0001,
+      "step": 344000
+    },
+    {
+      "epoch": 14.521772119883657,
+      "grad_norm": 1.4162390232086182,
+      "learning_rate": 4.263173691206106e-05,
+      "loss": 4.9987,
+      "step": 344500
+    },
+    {
+      "epoch": 14.542848712220207,
+      "grad_norm": 1.4220322370529175,
+      "learning_rate": 4.254598465034515e-05,
+      "loss": 5.0002,
+      "step": 345000
+    },
+    {
+      "epoch": 14.56392530455676,
+      "grad_norm": 1.3817297220230103,
+      "learning_rate": 4.246023238862925e-05,
+      "loss": 4.9972,
+      "step": 345500
+    },
+    {
+      "epoch": 14.58500189689331,
+      "grad_norm": 1.4091525077819824,
+      "learning_rate": 4.237448012691335e-05,
+      "loss": 4.9974,
+      "step": 346000
+    },
+    {
+      "epoch": 14.606078489229862,
+      "grad_norm": 1.4002481698989868,
+      "learning_rate": 4.228872786519745e-05,
+      "loss": 4.9965,
+      "step": 346500
+    },
+    {
+      "epoch": 14.627155081566412,
+      "grad_norm": 1.4204384088516235,
+      "learning_rate": 4.220297560348154e-05,
+      "loss": 5.0002,
+      "step": 347000
+    },
+    {
+      "epoch": 14.648231673902963,
+      "grad_norm": 1.4157198667526245,
+      "learning_rate": 4.2117223341765645e-05,
+      "loss": 4.9958,
+      "step": 347500
+    },
+    {
+      "epoch": 14.669308266239515,
+      "grad_norm": 1.38182532787323,
+      "learning_rate": 4.2031471080049736e-05,
+      "loss": 4.996,
+      "step": 348000
+    },
+    {
+      "epoch": 14.690384858576065,
+      "grad_norm": 1.3887721300125122,
+      "learning_rate": 4.1945718818333834e-05,
+      "loss": 4.998,
+      "step": 348500
+    },
+    {
+      "epoch": 14.711461450912616,
+      "grad_norm": 1.390017032623291,
+      "learning_rate": 4.185996655661793e-05,
+      "loss": 4.9989,
+      "step": 349000
+    },
+    {
+      "epoch": 14.732538043249168,
+      "grad_norm": 1.3894298076629639,
+      "learning_rate": 4.177421429490203e-05,
+      "loss": 4.9986,
+      "step": 349500
+    },
+    {
+      "epoch": 14.753614635585718,
+      "grad_norm": 1.4439281225204468,
+      "learning_rate": 4.168846203318613e-05,
+      "loss": 4.9941,
+      "step": 350000
+    },
+    {
+      "epoch": 14.753614635585718,
+      "eval_accuracy": 0.715205538638921,
+      "eval_loss": 4.726293087005615,
+      "eval_runtime": 460.2074,
+      "eval_samples_per_second": 262.286,
+      "eval_steps_per_second": 2.734,
+      "step": 350000
+    },
+    {
+      "epoch": 14.77469122792227,
+      "grad_norm": 1.366763949394226,
+      "learning_rate": 4.160270977147022e-05,
+      "loss": 4.9947,
+      "step": 350500
+    },
+    {
+      "epoch": 14.79576782025882,
+      "grad_norm": 1.4316478967666626,
+      "learning_rate": 4.1516957509754326e-05,
+      "loss": 4.9992,
+      "step": 351000
+    },
+    {
+      "epoch": 14.81684441259537,
+      "grad_norm": 1.4261616468429565,
+      "learning_rate": 4.143120524803842e-05,
+      "loss": 4.9932,
+      "step": 351500
+    },
+    {
+      "epoch": 14.837921004931923,
+      "grad_norm": 1.390966773033142,
+      "learning_rate": 4.1345452986322516e-05,
+      "loss": 4.9943,
+      "step": 352000
+    },
+    {
+      "epoch": 14.858997597268473,
+      "grad_norm": 1.396024465560913,
+      "learning_rate": 4.1259700724606614e-05,
+      "loss": 4.9946,
+      "step": 352500
+    },
+    {
+      "epoch": 14.880074189605025,
+      "grad_norm": 1.4493980407714844,
+      "learning_rate": 4.117394846289071e-05,
+      "loss": 4.9947,
+      "step": 353000
+    },
+    {
+      "epoch": 14.901150781941576,
+      "grad_norm": 1.3898714780807495,
+      "learning_rate": 4.1088196201174804e-05,
+      "loss": 4.9973,
+      "step": 353500
+    },
+    {
+      "epoch": 14.922227374278126,
+      "grad_norm": 1.3662023544311523,
+      "learning_rate": 4.100244393945891e-05,
+      "loss": 4.9933,
+      "step": 354000
+    },
+    {
+      "epoch": 14.943303966614678,
+      "grad_norm": 1.4345927238464355,
+      "learning_rate": 4.0916691677743e-05,
+      "loss": 4.9988,
+      "step": 354500
+    },
+    {
+      "epoch": 14.964380558951229,
+      "grad_norm": 1.3640470504760742,
+      "learning_rate": 4.08309394160271e-05,
+      "loss": 4.9958,
+      "step": 355000
+    },
+    {
+      "epoch": 14.985457151287779,
+      "grad_norm": 1.43264639377594,
+      "learning_rate": 4.07451871543112e-05,
+      "loss": 4.9945,
+      "step": 355500
+    },
+    {
+      "epoch": 15.006533743624331,
+      "grad_norm": 1.371166706085205,
+      "learning_rate": 4.0659434892595295e-05,
+      "loss": 4.9896,
+      "step": 356000
+    },
+    {
+      "epoch": 15.027610335960881,
+      "grad_norm": 1.4170937538146973,
+      "learning_rate": 4.057368263087939e-05,
+      "loss": 4.9858,
+      "step": 356500
+    },
+    {
+      "epoch": 15.048686928297434,
+      "grad_norm": 1.3824968338012695,
+      "learning_rate": 4.048793036916349e-05,
+      "loss": 4.99,
+      "step": 357000
+    },
+    {
+      "epoch": 15.069763520633984,
+      "grad_norm": 1.4515454769134521,
+      "learning_rate": 4.0402178107447584e-05,
+      "loss": 4.9923,
+      "step": 357500
+    },
+    {
+      "epoch": 15.090840112970534,
+      "grad_norm": 1.4352400302886963,
+      "learning_rate": 4.031642584573168e-05,
+      "loss": 4.9911,
+      "step": 358000
+    },
+    {
+      "epoch": 15.111916705307086,
+      "grad_norm": 1.4122790098190308,
+      "learning_rate": 4.023067358401578e-05,
+      "loss": 4.9874,
+      "step": 358500
+    },
+    {
+      "epoch": 15.132993297643637,
+      "grad_norm": 1.4563071727752686,
+      "learning_rate": 4.014492132229988e-05,
+      "loss": 4.9911,
+      "step": 359000
+    },
+    {
+      "epoch": 15.154069889980189,
+      "grad_norm": 1.4318819046020508,
+      "learning_rate": 4.005916906058397e-05,
+      "loss": 4.989,
+      "step": 359500
+    },
+    {
+      "epoch": 15.17514648231674,
+      "grad_norm": 1.3806135654449463,
+      "learning_rate": 3.9973416798868075e-05,
+      "loss": 4.9902,
+      "step": 360000
+    },
+    {
+      "epoch": 15.17514648231674,
+      "eval_accuracy": 0.7157444870521119,
+      "eval_loss": 4.722222805023193,
+      "eval_runtime": 423.6561,
+      "eval_samples_per_second": 284.915,
+      "eval_steps_per_second": 2.969,
+      "step": 360000
+    },
+    {
+      "epoch": 15.19622307465329,
+      "grad_norm": 1.4326872825622559,
+      "learning_rate": 3.988766453715217e-05,
+      "loss": 4.9919,
+      "step": 360500
+    },
+    {
+      "epoch": 15.217299666989842,
+      "grad_norm": 1.4095557928085327,
+      "learning_rate": 3.9801912275436265e-05,
+      "loss": 4.9897,
+      "step": 361000
+    },
+    {
+      "epoch": 15.238376259326392,
+      "grad_norm": 1.4121787548065186,
+      "learning_rate": 3.971616001372036e-05,
+      "loss": 4.9925,
+      "step": 361500
+    },
+    {
+      "epoch": 15.259452851662942,
+      "grad_norm": 1.4262945652008057,
+      "learning_rate": 3.963040775200446e-05,
+      "loss": 4.989,
+      "step": 362000
+    },
+    {
+      "epoch": 15.280529443999495,
+      "grad_norm": 1.4193719625473022,
+      "learning_rate": 3.954465549028856e-05,
+      "loss": 4.9865,
+      "step": 362500
+    },
+    {
+      "epoch": 15.301606036336045,
+      "grad_norm": 1.367762804031372,
+      "learning_rate": 3.945890322857266e-05,
+      "loss": 4.9909,
+      "step": 363000
+    },
+    {
+      "epoch": 15.322682628672597,
+      "grad_norm": 1.371561050415039,
+      "learning_rate": 3.9373150966856756e-05,
+      "loss": 4.9859,
+      "step": 363500
+    },
+    {
+      "epoch": 15.343759221009147,
+      "grad_norm": 1.4042510986328125,
+      "learning_rate": 3.928739870514085e-05,
+      "loss": 4.9874,
+      "step": 364000
+    },
+    {
+      "epoch": 15.364835813345698,
+      "grad_norm": 1.4316272735595703,
+      "learning_rate": 3.920164644342495e-05,
+      "loss": 4.9865,
+      "step": 364500
+    },
+    {
+      "epoch": 15.38591240568225,
+      "grad_norm": 1.4387049674987793,
+      "learning_rate": 3.9115894181709045e-05,
+      "loss": 4.9868,
+      "step": 365000
+    },
+    {
+      "epoch": 15.4069889980188,
+      "grad_norm": 1.3991371393203735,
+      "learning_rate": 3.903014191999314e-05,
+      "loss": 4.9833,
+      "step": 365500
+    },
+    {
+      "epoch": 15.42806559035535,
+      "grad_norm": 1.4553977251052856,
+      "learning_rate": 3.894438965827724e-05,
+      "loss": 4.9902,
+      "step": 366000
+    },
+    {
+      "epoch": 15.449142182691903,
+      "grad_norm": 1.378907322883606,
+      "learning_rate": 3.885863739656134e-05,
+      "loss": 4.9869,
+      "step": 366500
+    },
+    {
+      "epoch": 15.470218775028453,
+      "grad_norm": 1.4654325246810913,
+      "learning_rate": 3.877288513484543e-05,
+      "loss": 4.9853,
+      "step": 367000
+    },
+    {
+      "epoch": 15.491295367365005,
+      "grad_norm": 1.3931434154510498,
+      "learning_rate": 3.868713287312953e-05,
+      "loss": 4.9878,
+      "step": 367500
+    },
+    {
+      "epoch": 15.512371959701555,
+      "grad_norm": 1.4192183017730713,
+      "learning_rate": 3.860138061141363e-05,
+      "loss": 4.9857,
+      "step": 368000
+    },
+    {
+      "epoch": 15.533448552038106,
+      "grad_norm": 1.4404408931732178,
+      "learning_rate": 3.8515628349697726e-05,
+      "loss": 4.9886,
+      "step": 368500
+    },
+    {
+      "epoch": 15.554525144374658,
+      "grad_norm": 1.414708137512207,
+      "learning_rate": 3.842987608798182e-05,
+      "loss": 4.9862,
+      "step": 369000
+    },
+    {
+      "epoch": 15.575601736711208,
+      "grad_norm": 1.4469889402389526,
+      "learning_rate": 3.834412382626592e-05,
+      "loss": 4.9827,
+      "step": 369500
+    },
+    {
+      "epoch": 15.596678329047759,
+      "grad_norm": 1.4104082584381104,
+      "learning_rate": 3.8258371564550014e-05,
+      "loss": 4.9867,
+      "step": 370000
+    },
+    {
+      "epoch": 15.596678329047759,
+      "eval_accuracy": 0.7168208439015398,
+      "eval_loss": 4.715757369995117,
+      "eval_runtime": 383.8701,
+      "eval_samples_per_second": 314.445,
+      "eval_steps_per_second": 3.277,
+      "step": 370000
+    },
+    {
+      "epoch": 15.61775492138431,
+      "grad_norm": 1.4229207038879395,
+      "learning_rate": 3.817261930283411e-05,
+      "loss": 4.9841,
+      "step": 370500
+    },
+    {
+      "epoch": 15.638831513720861,
+      "grad_norm": 1.524247646331787,
+      "learning_rate": 3.808686704111821e-05,
+      "loss": 4.9845,
+      "step": 371000
+    },
+    {
+      "epoch": 15.659908106057413,
+      "grad_norm": 1.4315611124038696,
+      "learning_rate": 3.800111477940231e-05,
+      "loss": 4.9842,
+      "step": 371500
+    },
+    {
+      "epoch": 15.680984698393964,
+      "grad_norm": 1.3967938423156738,
+      "learning_rate": 3.79153625176864e-05,
+      "loss": 4.9865,
+      "step": 372000
+    },
+    {
+      "epoch": 15.702061290730514,
+      "grad_norm": 1.4247682094573975,
+      "learning_rate": 3.7829610255970506e-05,
+      "loss": 4.988,
+      "step": 372500
+    },
+    {
+      "epoch": 15.723137883067066,
+      "grad_norm": 1.4636257886886597,
+      "learning_rate": 3.77438579942546e-05,
+      "loss": 4.9823,
+      "step": 373000
+    },
+    {
+      "epoch": 15.744214475403616,
+      "grad_norm": 1.3833144903182983,
+      "learning_rate": 3.7658105732538696e-05,
+      "loss": 4.9793,
+      "step": 373500
+    },
+    {
+      "epoch": 15.765291067740169,
+      "grad_norm": 1.3875837326049805,
+      "learning_rate": 3.7572353470822794e-05,
+      "loss": 4.9861,
+      "step": 374000
+    },
+    {
+      "epoch": 15.786367660076719,
+      "grad_norm": 1.4256742000579834,
+      "learning_rate": 3.748660120910689e-05,
+      "loss": 4.9802,
+      "step": 374500
+    },
+    {
+      "epoch": 15.80744425241327,
+      "grad_norm": 1.497991919517517,
+      "learning_rate": 3.740084894739099e-05,
+      "loss": 4.9773,
+      "step": 375000
+    },
+    {
+      "epoch": 15.828520844749821,
+      "grad_norm": 1.4145623445510864,
+      "learning_rate": 3.731509668567509e-05,
+      "loss": 4.9804,
+      "step": 375500
+    },
+    {
+      "epoch": 15.849597437086372,
+      "grad_norm": 1.4386074542999268,
+      "learning_rate": 3.722934442395919e-05,
+      "loss": 4.9835,
+      "step": 376000
+    },
+    {
+      "epoch": 15.870674029422922,
+      "grad_norm": 1.4090478420257568,
+      "learning_rate": 3.714359216224328e-05,
+      "loss": 4.9828,
+      "step": 376500
+    },
+    {
+      "epoch": 15.891750621759474,
+      "grad_norm": 1.4416399002075195,
+      "learning_rate": 3.705783990052738e-05,
+      "loss": 4.9861,
+      "step": 377000
+    },
+    {
+      "epoch": 15.912827214096025,
+      "grad_norm": 1.4128856658935547,
+      "learning_rate": 3.6972087638811475e-05,
+      "loss": 4.9802,
+      "step": 377500
+    },
+    {
+      "epoch": 15.933903806432577,
+      "grad_norm": 1.4148684740066528,
+      "learning_rate": 3.6886335377095574e-05,
+      "loss": 4.9827,
+      "step": 378000
+    },
+    {
+      "epoch": 15.954980398769127,
+      "grad_norm": 1.358132243156433,
+      "learning_rate": 3.680058311537967e-05,
+      "loss": 4.9819,
+      "step": 378500
+    },
+    {
+      "epoch": 15.976056991105677,
+      "grad_norm": 1.3970136642456055,
+      "learning_rate": 3.671483085366377e-05,
+      "loss": 4.9824,
+      "step": 379000
+    },
+    {
+      "epoch": 15.99713358344223,
+      "grad_norm": 1.422939419746399,
+      "learning_rate": 3.662907859194786e-05,
+      "loss": 4.9794,
+      "step": 379500
+    },
+    {
+      "epoch": 16.01821017577878,
+      "grad_norm": 1.4389123916625977,
+      "learning_rate": 3.654332633023197e-05,
+      "loss": 4.9796,
+      "step": 380000
+    },
+    {
+      "epoch": 16.01821017577878,
+      "eval_accuracy": 0.7174619029917247,
+      "eval_loss": 4.71160888671875,
+      "eval_runtime": 707.6178,
+      "eval_samples_per_second": 170.581,
+      "eval_steps_per_second": 1.778,
+      "step": 380000
+    },
+    {
+      "epoch": 16.039286768115332,
+      "grad_norm": 1.359557867050171,
+      "learning_rate": 3.645757406851606e-05,
+      "loss": 4.9773,
+      "step": 380500
+    },
+    {
+      "epoch": 16.06036336045188,
+      "grad_norm": 1.4432597160339355,
+      "learning_rate": 3.6371821806800157e-05,
+      "loss": 4.9756,
+      "step": 381000
+    },
+    {
+      "epoch": 16.081439952788433,
+      "grad_norm": 1.364013433456421,
+      "learning_rate": 3.6286069545084255e-05,
+      "loss": 4.9811,
+      "step": 381500
+    },
+    {
+      "epoch": 16.102516545124985,
+      "grad_norm": 1.4194997549057007,
+      "learning_rate": 3.620031728336835e-05,
+      "loss": 4.9768,
+      "step": 382000
+    },
+    {
+      "epoch": 16.123593137461537,
+      "grad_norm": 1.3927031755447388,
+      "learning_rate": 3.6114565021652445e-05,
+      "loss": 4.9719,
+      "step": 382500
+    },
+    {
+      "epoch": 16.144669729798085,
+      "grad_norm": 1.3861465454101562,
+      "learning_rate": 3.602881275993655e-05,
+      "loss": 4.9749,
+      "step": 383000
+    },
+    {
+      "epoch": 16.165746322134638,
+      "grad_norm": 1.443176031112671,
+      "learning_rate": 3.594306049822064e-05,
+      "loss": 4.9797,
+      "step": 383500
+    },
+    {
+      "epoch": 16.18682291447119,
+      "grad_norm": 1.443726897239685,
+      "learning_rate": 3.585730823650474e-05,
+      "loss": 4.9767,
+      "step": 384000
+    },
+    {
+      "epoch": 16.20789950680774,
+      "grad_norm": 1.403106927871704,
+      "learning_rate": 3.577155597478883e-05,
+      "loss": 4.9783,
+      "step": 384500
+    },
+    {
+      "epoch": 16.22897609914429,
+      "grad_norm": 1.4214428663253784,
+      "learning_rate": 3.5685803713072936e-05,
+      "loss": 4.9713,
+      "step": 385000
+    },
+    {
+      "epoch": 16.250052691480843,
+      "grad_norm": 1.436005711555481,
+      "learning_rate": 3.560005145135703e-05,
+      "loss": 4.9755,
+      "step": 385500
+    },
+    {
+      "epoch": 16.27112928381739,
+      "grad_norm": 1.4586254358291626,
+      "learning_rate": 3.5514299189641126e-05,
+      "loss": 4.9726,
+      "step": 386000
+    },
+    {
+      "epoch": 16.292205876153943,
+      "grad_norm": 1.4233094453811646,
+      "learning_rate": 3.5428546927925224e-05,
+      "loss": 4.9787,
+      "step": 386500
+    },
+    {
+      "epoch": 16.313282468490495,
+      "grad_norm": 1.3856542110443115,
+      "learning_rate": 3.534279466620932e-05,
+      "loss": 4.9744,
+      "step": 387000
+    },
+    {
+      "epoch": 16.334359060827044,
+      "grad_norm": 1.4372364282608032,
+      "learning_rate": 3.525704240449342e-05,
+      "loss": 4.9749,
+      "step": 387500
+    },
+    {
+      "epoch": 16.355435653163596,
+      "grad_norm": 1.4100600481033325,
+      "learning_rate": 3.517129014277752e-05,
+      "loss": 4.9715,
+      "step": 388000
+    },
+    {
+      "epoch": 16.376512245500148,
+      "grad_norm": 1.4449211359024048,
+      "learning_rate": 3.508553788106161e-05,
+      "loss": 4.9717,
+      "step": 388500
+    },
+    {
+      "epoch": 16.3975888378367,
+      "grad_norm": 1.4347094297409058,
+      "learning_rate": 3.499978561934571e-05,
+      "loss": 4.9723,
+      "step": 389000
+    },
+    {
+      "epoch": 16.41866543017325,
+      "grad_norm": 1.388414740562439,
+      "learning_rate": 3.491403335762981e-05,
+      "loss": 4.9735,
+      "step": 389500
+    },
+    {
+      "epoch": 16.4397420225098,
+      "grad_norm": 1.3812241554260254,
+      "learning_rate": 3.4828281095913906e-05,
+      "loss": 4.9751,
+      "step": 390000
+    },
+    {
+      "epoch": 16.4397420225098,
+      "eval_accuracy": 0.7179732409440068,
+      "eval_loss": 4.70511531829834,
+      "eval_runtime": 416.7313,
+      "eval_samples_per_second": 289.65,
+      "eval_steps_per_second": 3.019,
+      "step": 390000
+    },
+    {
+      "epoch": 16.460818614846353,
+      "grad_norm": 1.4588754177093506,
+      "learning_rate": 3.4742528834198004e-05,
+      "loss": 4.9712,
+      "step": 390500
+    },
+    {
+      "epoch": 16.4818952071829,
+      "grad_norm": 1.3964200019836426,
+      "learning_rate": 3.46567765724821e-05,
+      "loss": 4.975,
+      "step": 391000
+    },
+    {
+      "epoch": 16.502971799519454,
+      "grad_norm": 1.4611189365386963,
+      "learning_rate": 3.45710243107662e-05,
+      "loss": 4.972,
+      "step": 391500
+    },
+    {
+      "epoch": 16.524048391856006,
+      "grad_norm": 1.418857455253601,
+      "learning_rate": 3.448527204905029e-05,
+      "loss": 4.9792,
+      "step": 392000
+    },
+    {
+      "epoch": 16.545124984192555,
+      "grad_norm": 1.405012607574463,
+      "learning_rate": 3.43995197873344e-05,
+      "loss": 4.9705,
+      "step": 392500
+    },
+    {
+      "epoch": 16.566201576529107,
+      "grad_norm": 1.444247841835022,
+      "learning_rate": 3.431376752561849e-05,
+      "loss": 4.9745,
+      "step": 393000
+    },
+    {
+      "epoch": 16.58727816886566,
+      "grad_norm": 1.4209394454956055,
+      "learning_rate": 3.422801526390259e-05,
+      "loss": 4.9755,
+      "step": 393500
+    },
+    {
+      "epoch": 16.608354761202207,
+      "grad_norm": 1.4114818572998047,
+      "learning_rate": 3.4142263002186686e-05,
+      "loss": 4.9733,
+      "step": 394000
+    },
+    {
+      "epoch": 16.62943135353876,
+      "grad_norm": 1.4346266984939575,
+      "learning_rate": 3.4056510740470784e-05,
+      "loss": 4.9778,
+      "step": 394500
+    },
+    {
+      "epoch": 16.65050794587531,
+      "grad_norm": 1.4401416778564453,
+      "learning_rate": 3.3970758478754875e-05,
+      "loss": 4.9732,
+      "step": 395000
+    },
+    {
+      "epoch": 16.67158453821186,
+      "grad_norm": 1.4092142581939697,
+      "learning_rate": 3.388500621703898e-05,
+      "loss": 4.9723,
+      "step": 395500
+    },
+    {
+      "epoch": 16.692661130548412,
+      "grad_norm": 1.4517334699630737,
+      "learning_rate": 3.379925395532307e-05,
+      "loss": 4.9743,
+      "step": 396000
+    },
+    {
+      "epoch": 16.713737722884964,
+      "grad_norm": 1.3363827466964722,
+      "learning_rate": 3.371350169360717e-05,
+      "loss": 4.9704,
+      "step": 396500
+    },
+    {
+      "epoch": 16.734814315221517,
+      "grad_norm": 1.4226515293121338,
+      "learning_rate": 3.362774943189127e-05,
+      "loss": 4.9745,
+      "step": 397000
+    },
+    {
+      "epoch": 16.755890907558065,
+      "grad_norm": 1.3876314163208008,
+      "learning_rate": 3.354199717017537e-05,
+      "loss": 4.9702,
+      "step": 397500
+    },
+    {
+      "epoch": 16.776967499894617,
+      "grad_norm": 1.4402391910552979,
+      "learning_rate": 3.345624490845946e-05,
+      "loss": 4.9711,
+      "step": 398000
+    },
+    {
+      "epoch": 16.79804409223117,
+      "grad_norm": 1.4286390542984009,
+      "learning_rate": 3.3370492646743564e-05,
+      "loss": 4.9698,
+      "step": 398500
+    },
+    {
+      "epoch": 16.819120684567718,
+      "grad_norm": 1.395855188369751,
+      "learning_rate": 3.3284740385027655e-05,
+      "loss": 4.977,
+      "step": 399000
+    },
+    {
+      "epoch": 16.84019727690427,
+      "grad_norm": 1.3465309143066406,
+      "learning_rate": 3.319898812331175e-05,
+      "loss": 4.9682,
+      "step": 399500
+    },
+    {
+      "epoch": 16.861273869240822,
+      "grad_norm": 1.4175748825073242,
+      "learning_rate": 3.311323586159585e-05,
+      "loss": 4.9683,
+      "step": 400000
+    },
+    {
+      "epoch": 16.861273869240822,
+      "eval_accuracy": 0.7184489838949294,
+      "eval_loss": 4.703793525695801,
+      "eval_runtime": 424.4616,
+      "eval_samples_per_second": 284.374,
+      "eval_steps_per_second": 2.964,
+      "step": 400000
+    },
+    {
+      "epoch": 16.88235046157737,
+      "grad_norm": 1.459302306175232,
+      "learning_rate": 3.302748359987995e-05,
+      "loss": 4.9664,
+      "step": 400500
+    },
+    {
+      "epoch": 16.903427053913923,
+      "grad_norm": 1.3823376893997192,
+      "learning_rate": 3.294173133816404e-05,
+      "loss": 4.97,
+      "step": 401000
+    },
+    {
+      "epoch": 16.924503646250475,
+      "grad_norm": 1.4607036113739014,
+      "learning_rate": 3.2855979076448147e-05,
+      "loss": 4.9742,
+      "step": 401500
+    },
+    {
+      "epoch": 16.945580238587024,
+      "grad_norm": 1.4358510971069336,
+      "learning_rate": 3.277022681473224e-05,
+      "loss": 4.971,
+      "step": 402000
+    },
+    {
+      "epoch": 16.966656830923576,
+      "grad_norm": 1.451393485069275,
+      "learning_rate": 3.2684474553016336e-05,
+      "loss": 4.9695,
+      "step": 402500
+    },
+    {
+      "epoch": 16.987733423260128,
+      "grad_norm": 1.4473155736923218,
+      "learning_rate": 3.2598722291300435e-05,
+      "loss": 4.9707,
+      "step": 403000
+    },
+    {
+      "epoch": 17.00881001559668,
+      "grad_norm": 1.4609649181365967,
+      "learning_rate": 3.251297002958453e-05,
+      "loss": 4.9697,
+      "step": 403500
+    },
+    {
+      "epoch": 17.02988660793323,
+      "grad_norm": 1.390286922454834,
+      "learning_rate": 3.242721776786863e-05,
+      "loss": 4.9685,
+      "step": 404000
+    },
+    {
+      "epoch": 17.05096320026978,
+      "grad_norm": 1.4172338247299194,
+      "learning_rate": 3.234146550615272e-05,
+      "loss": 4.9661,
+      "step": 404500
+    },
+    {
+      "epoch": 17.072039792606333,
+      "grad_norm": 1.4194505214691162,
+      "learning_rate": 3.225571324443683e-05,
+      "loss": 4.9671,
+      "step": 405000
+    },
+    {
+      "epoch": 17.09311638494288,
+      "grad_norm": 1.4503978490829468,
+      "learning_rate": 3.216996098272092e-05,
+      "loss": 4.967,
+      "step": 405500
+    },
+    {
+      "epoch": 17.114192977279433,
+      "grad_norm": 1.4436626434326172,
+      "learning_rate": 3.208420872100502e-05,
+      "loss": 4.9656,
+      "step": 406000
+    },
+    {
+      "epoch": 17.135269569615986,
+      "grad_norm": 1.4530192613601685,
+      "learning_rate": 3.1998456459289116e-05,
+      "loss": 4.963,
+      "step": 406500
+    },
+    {
+      "epoch": 17.156346161952534,
+      "grad_norm": 1.38369882106781,
+      "learning_rate": 3.1912704197573214e-05,
+      "loss": 4.9617,
+      "step": 407000
+    },
+    {
+      "epoch": 17.177422754289086,
+      "grad_norm": 1.4719420671463013,
+      "learning_rate": 3.1826951935857306e-05,
+      "loss": 4.9617,
+      "step": 407500
+    },
+    {
+      "epoch": 17.19849934662564,
+      "grad_norm": 1.4470149278640747,
+      "learning_rate": 3.174119967414141e-05,
+      "loss": 4.9662,
+      "step": 408000
+    },
+    {
+      "epoch": 17.219575938962187,
+      "grad_norm": 1.4514477252960205,
+      "learning_rate": 3.16554474124255e-05,
+      "loss": 4.9652,
+      "step": 408500
+    },
+    {
+      "epoch": 17.24065253129874,
+      "grad_norm": 1.4137061834335327,
+      "learning_rate": 3.15696951507096e-05,
+      "loss": 4.9676,
+      "step": 409000
+    },
+    {
+      "epoch": 17.26172912363529,
+      "grad_norm": 1.4483462572097778,
+      "learning_rate": 3.14839428889937e-05,
+      "loss": 4.9642,
+      "step": 409500
+    },
+    {
+      "epoch": 17.282805715971843,
+      "grad_norm": 1.4757940769195557,
+      "learning_rate": 3.13981906272778e-05,
+      "loss": 4.967,
+      "step": 410000
+    },
+    {
+      "epoch": 17.282805715971843,
+      "eval_accuracy": 0.7196022374388582,
+      "eval_loss": 4.695497035980225,
+      "eval_runtime": 448.9603,
+      "eval_samples_per_second": 268.857,
+      "eval_steps_per_second": 2.802,
+      "step": 410000
+    },
+    {
+      "epoch": 17.303882308308392,
+      "grad_norm": 1.4248424768447876,
+      "learning_rate": 3.131243836556189e-05,
+      "loss": 4.9674,
+      "step": 410500
+    },
+    {
+      "epoch": 17.324958900644944,
+      "grad_norm": 1.423072099685669,
+      "learning_rate": 3.1226686103845994e-05,
+      "loss": 4.9617,
+      "step": 411000
+    },
+    {
+      "epoch": 17.346035492981496,
+      "grad_norm": 1.4605790376663208,
+      "learning_rate": 3.1140933842130086e-05,
+      "loss": 4.9617,
+      "step": 411500
+    },
+    {
+      "epoch": 17.367112085318045,
+      "grad_norm": 1.4501166343688965,
+      "learning_rate": 3.1055181580414184e-05,
+      "loss": 4.9662,
+      "step": 412000
+    },
+    {
+      "epoch": 17.388188677654597,
+      "grad_norm": 1.4112434387207031,
+      "learning_rate": 3.096942931869828e-05,
+      "loss": 4.9668,
+      "step": 412500
+    },
+    {
+      "epoch": 17.40926526999115,
+      "grad_norm": 1.4233763217926025,
+      "learning_rate": 3.088367705698238e-05,
+      "loss": 4.9605,
+      "step": 413000
+    },
+    {
+      "epoch": 17.430341862327698,
+      "grad_norm": 1.4048868417739868,
+      "learning_rate": 3.079792479526647e-05,
+      "loss": 4.9632,
+      "step": 413500
+    },
+    {
+      "epoch": 17.45141845466425,
+      "grad_norm": 1.406559944152832,
+      "learning_rate": 3.071217253355058e-05,
+      "loss": 4.9619,
+      "step": 414000
+    },
+    {
+      "epoch": 17.472495047000802,
+      "grad_norm": 1.4297690391540527,
+      "learning_rate": 3.062642027183467e-05,
+      "loss": 4.9611,
+      "step": 414500
+    },
+    {
+      "epoch": 17.49357163933735,
+      "grad_norm": 1.4182250499725342,
+      "learning_rate": 3.054066801011877e-05,
+      "loss": 4.9631,
+      "step": 415000
+    },
+    {
+      "epoch": 17.514648231673903,
+      "grad_norm": 1.4594875574111938,
+      "learning_rate": 3.045491574840287e-05,
+      "loss": 4.9656,
+      "step": 415500
+    },
+    {
+      "epoch": 17.535724824010455,
+      "grad_norm": 1.4270740747451782,
+      "learning_rate": 3.0369163486686964e-05,
+      "loss": 4.962,
+      "step": 416000
+    },
+    {
+      "epoch": 17.556801416347007,
+      "grad_norm": 1.4481427669525146,
+      "learning_rate": 3.028341122497106e-05,
+      "loss": 4.9596,
+      "step": 416500
+    },
+    {
+      "epoch": 17.577878008683555,
+      "grad_norm": 1.4569650888442993,
+      "learning_rate": 3.019765896325516e-05,
+      "loss": 4.9659,
+      "step": 417000
+    },
+    {
+      "epoch": 17.598954601020107,
+      "grad_norm": 1.4816147089004517,
+      "learning_rate": 3.0111906701539255e-05,
+      "loss": 4.9636,
+      "step": 417500
+    },
+    {
+      "epoch": 17.62003119335666,
+      "grad_norm": 1.4890145063400269,
+      "learning_rate": 3.002615443982335e-05,
+      "loss": 4.9615,
+      "step": 418000
+    },
+    {
+      "epoch": 17.641107785693208,
+      "grad_norm": 1.4744318723678589,
+      "learning_rate": 2.9940402178107452e-05,
+      "loss": 4.9609,
+      "step": 418500
+    },
+    {
+      "epoch": 17.66218437802976,
+      "grad_norm": 1.4595578908920288,
+      "learning_rate": 2.9854649916391547e-05,
+      "loss": 4.9587,
+      "step": 419000
+    },
+    {
+      "epoch": 17.683260970366312,
+      "grad_norm": 1.4918544292449951,
+      "learning_rate": 2.976889765467564e-05,
+      "loss": 4.9652,
+      "step": 419500
+    },
+    {
+      "epoch": 17.70433756270286,
+      "grad_norm": 1.4290977716445923,
+      "learning_rate": 2.9683145392959737e-05,
+      "loss": 4.961,
+      "step": 420000
+    },
+    {
+      "epoch": 17.70433756270286,
+      "eval_accuracy": 0.7199684681754533,
+      "eval_loss": 4.6946563720703125,
+      "eval_runtime": 440.2597,
+      "eval_samples_per_second": 274.17,
+      "eval_steps_per_second": 2.857,
+      "step": 420000
+    },
+    {
+      "epoch": 17.725414155039413,
+      "grad_norm": 1.4476240873336792,
+      "learning_rate": 2.9597393131243838e-05,
+      "loss": 4.961,
+      "step": 420500
+    },
+    {
+      "epoch": 17.746490747375965,
+      "grad_norm": 1.4003117084503174,
+      "learning_rate": 2.9511640869527933e-05,
+      "loss": 4.9567,
+      "step": 421000
+    },
+    {
+      "epoch": 17.767567339712514,
+      "grad_norm": 1.4295785427093506,
+      "learning_rate": 2.942588860781203e-05,
+      "loss": 4.9572,
+      "step": 421500
+    },
+    {
+      "epoch": 17.788643932049066,
+      "grad_norm": 1.4380590915679932,
+      "learning_rate": 2.934013634609613e-05,
+      "loss": 4.9617,
+      "step": 422000
+    },
+    {
+      "epoch": 17.809720524385618,
+      "grad_norm": 1.4086601734161377,
+      "learning_rate": 2.9254384084380225e-05,
+      "loss": 4.9612,
+      "step": 422500
+    },
+    {
+      "epoch": 17.830797116722167,
+      "grad_norm": 1.450409173965454,
+      "learning_rate": 2.9168631822664323e-05,
+      "loss": 4.9611,
+      "step": 423000
+    },
+    {
+      "epoch": 17.85187370905872,
+      "grad_norm": 1.4309135675430298,
+      "learning_rate": 2.908287956094842e-05,
+      "loss": 4.9621,
+      "step": 423500
+    },
+    {
+      "epoch": 17.87295030139527,
+      "grad_norm": 1.443009853363037,
+      "learning_rate": 2.899712729923252e-05,
+      "loss": 4.96,
+      "step": 424000
+    },
+    {
+      "epoch": 17.894026893731823,
+      "grad_norm": 1.4476872682571411,
+      "learning_rate": 2.8911375037516615e-05,
+      "loss": 4.9607,
+      "step": 424500
+    },
+    {
+      "epoch": 17.91510348606837,
+      "grad_norm": 1.4112868309020996,
+      "learning_rate": 2.8825622775800716e-05,
+      "loss": 4.9616,
+      "step": 425000
+    },
+    {
+      "epoch": 17.936180078404924,
+      "grad_norm": 1.5055495500564575,
+      "learning_rate": 2.873987051408481e-05,
+      "loss": 4.9591,
+      "step": 425500
+    },
+    {
+      "epoch": 17.957256670741476,
+      "grad_norm": 1.4419622421264648,
+      "learning_rate": 2.8654118252368906e-05,
+      "loss": 4.9563,
+      "step": 426000
+    },
+    {
+      "epoch": 17.978333263078024,
+      "grad_norm": 1.4547268152236938,
+      "learning_rate": 2.8568365990653008e-05,
+      "loss": 4.9577,
+      "step": 426500
+    },
+    {
+      "epoch": 17.999409855414576,
+      "grad_norm": 1.3986682891845703,
+      "learning_rate": 2.8482613728937103e-05,
+      "loss": 4.9611,
+      "step": 427000
+    },
+    {
+      "epoch": 18.02048644775113,
+      "grad_norm": 1.433235764503479,
+      "learning_rate": 2.8396861467221198e-05,
+      "loss": 4.9555,
+      "step": 427500
+    },
+    {
+      "epoch": 18.041563040087677,
+      "grad_norm": 1.4878346920013428,
+      "learning_rate": 2.83111092055053e-05,
+      "loss": 4.955,
+      "step": 428000
+    },
+    {
+      "epoch": 18.06263963242423,
+      "grad_norm": 1.4595096111297607,
+      "learning_rate": 2.8225356943789394e-05,
+      "loss": 4.9554,
+      "step": 428500
+    },
+    {
+      "epoch": 18.08371622476078,
+      "grad_norm": 1.4647470712661743,
+      "learning_rate": 2.813960468207349e-05,
+      "loss": 4.9548,
+      "step": 429000
+    },
+    {
+      "epoch": 18.10479281709733,
+      "grad_norm": 1.4277528524398804,
+      "learning_rate": 2.805385242035759e-05,
+      "loss": 4.9536,
+      "step": 429500
+    },
+    {
+      "epoch": 18.125869409433882,
+      "grad_norm": 1.4632010459899902,
+      "learning_rate": 2.7968100158641686e-05,
+      "loss": 4.953,
+      "step": 430000
+    },
+    {
+      "epoch": 18.125869409433882,
+      "eval_accuracy": 0.7204451248345951,
+      "eval_loss": 4.691041469573975,
+      "eval_runtime": 427.6532,
+      "eval_samples_per_second": 282.252,
+      "eval_steps_per_second": 2.942,
+      "step": 430000
+    },
+    {
+      "epoch": 18.146946001770434,
+      "grad_norm": 1.4899200201034546,
+      "learning_rate": 2.788234789692578e-05,
+      "loss": 4.9538,
+      "step": 430500
+    },
+    {
+      "epoch": 18.168022594106986,
+      "grad_norm": 1.4858582019805908,
+      "learning_rate": 2.7796595635209882e-05,
+      "loss": 4.9551,
+      "step": 431000
+    },
+    {
+      "epoch": 18.189099186443535,
+      "grad_norm": 1.4393603801727295,
+      "learning_rate": 2.7710843373493977e-05,
+      "loss": 4.9567,
+      "step": 431500
+    },
+    {
+      "epoch": 18.210175778780087,
+      "grad_norm": 1.434805154800415,
+      "learning_rate": 2.7625091111778072e-05,
+      "loss": 4.9583,
+      "step": 432000
+    },
+    {
+      "epoch": 18.23125237111664,
+      "grad_norm": 1.4924241304397583,
+      "learning_rate": 2.7539338850062174e-05,
+      "loss": 4.9549,
+      "step": 432500
+    },
+    {
+      "epoch": 18.252328963453188,
+      "grad_norm": 1.4193366765975952,
+      "learning_rate": 2.745358658834627e-05,
+      "loss": 4.9563,
+      "step": 433000
+    },
+    {
+      "epoch": 18.27340555578974,
+      "grad_norm": 1.4564720392227173,
+      "learning_rate": 2.7367834326630364e-05,
+      "loss": 4.952,
+      "step": 433500
+    },
+    {
+      "epoch": 18.294482148126292,
+      "grad_norm": 1.4424024820327759,
+      "learning_rate": 2.7282082064914465e-05,
+      "loss": 4.9525,
+      "step": 434000
+    },
+    {
+      "epoch": 18.31555874046284,
+      "grad_norm": 1.4633270502090454,
+      "learning_rate": 2.719632980319856e-05,
+      "loss": 4.951,
+      "step": 434500
+    },
+    {
+      "epoch": 18.336635332799393,
+      "grad_norm": 1.3960765600204468,
+      "learning_rate": 2.7110577541482655e-05,
+      "loss": 4.9522,
+      "step": 435000
+    },
+    {
+      "epoch": 18.357711925135945,
+      "grad_norm": 1.4497439861297607,
+      "learning_rate": 2.7024825279766757e-05,
+      "loss": 4.9551,
+      "step": 435500
+    },
+    {
+      "epoch": 18.378788517472493,
+      "grad_norm": 1.4604684114456177,
+      "learning_rate": 2.6939073018050852e-05,
+      "loss": 4.9547,
+      "step": 436000
+    },
+    {
+      "epoch": 18.399865109809046,
+      "grad_norm": 1.3897981643676758,
+      "learning_rate": 2.685332075633495e-05,
+      "loss": 4.9539,
+      "step": 436500
+    },
+    {
+      "epoch": 18.420941702145598,
+      "grad_norm": 1.4386805295944214,
+      "learning_rate": 2.6767568494619045e-05,
+      "loss": 4.9534,
+      "step": 437000
+    },
+    {
+      "epoch": 18.44201829448215,
+      "grad_norm": 1.470502495765686,
+      "learning_rate": 2.6681816232903143e-05,
+      "loss": 4.9507,
+      "step": 437500
+    },
+    {
+      "epoch": 18.4630948868187,
+      "grad_norm": 1.488389492034912,
+      "learning_rate": 2.6596063971187242e-05,
+      "loss": 4.9487,
+      "step": 438000
+    },
+    {
+      "epoch": 18.48417147915525,
+      "grad_norm": 1.4998657703399658,
+      "learning_rate": 2.6510311709471337e-05,
+      "loss": 4.95,
+      "step": 438500
+    },
+    {
+      "epoch": 18.505248071491803,
+      "grad_norm": 1.417507290840149,
+      "learning_rate": 2.642455944775544e-05,
+      "loss": 4.9521,
+      "step": 439000
+    },
+    {
+      "epoch": 18.52632466382835,
+      "grad_norm": 1.4461889266967773,
+      "learning_rate": 2.6338807186039533e-05,
+      "loss": 4.9513,
+      "step": 439500
+    },
+    {
+      "epoch": 18.547401256164903,
+      "grad_norm": 1.4456669092178345,
+      "learning_rate": 2.6253054924323628e-05,
+      "loss": 4.9491,
+      "step": 440000
+    },
+    {
+      "epoch": 18.547401256164903,
+      "eval_accuracy": 0.7207758821697517,
+      "eval_loss": 4.688427448272705,
+      "eval_runtime": 430.9181,
+      "eval_samples_per_second": 280.114,
+      "eval_steps_per_second": 2.919,
+      "step": 440000
+    },
+    {
+      "epoch": 18.568477848501455,
+      "grad_norm": 1.4177930355072021,
+      "learning_rate": 2.616730266260773e-05,
+      "loss": 4.9504,
+      "step": 440500
+    },
+    {
+      "epoch": 18.589554440838004,
+      "grad_norm": 1.4344291687011719,
+      "learning_rate": 2.6081550400891825e-05,
+      "loss": 4.9517,
+      "step": 441000
+    },
+    {
+      "epoch": 18.610631033174556,
+      "grad_norm": 1.4437711238861084,
+      "learning_rate": 2.599579813917592e-05,
+      "loss": 4.9523,
+      "step": 441500
+    },
+    {
+      "epoch": 18.63170762551111,
+      "grad_norm": 1.4459620714187622,
+      "learning_rate": 2.591004587746002e-05,
+      "loss": 4.9516,
+      "step": 442000
+    },
+    {
+      "epoch": 18.652784217847657,
+      "grad_norm": 1.4621723890304565,
+      "learning_rate": 2.5824293615744116e-05,
+      "loss": 4.95,
+      "step": 442500
+    },
+    {
+      "epoch": 18.67386081018421,
+      "grad_norm": 1.4987390041351318,
+      "learning_rate": 2.573854135402821e-05,
+      "loss": 4.9526,
+      "step": 443000
+    },
+    {
+      "epoch": 18.69493740252076,
+      "grad_norm": 1.435215950012207,
+      "learning_rate": 2.5652789092312313e-05,
+      "loss": 4.9506,
+      "step": 443500
+    },
+    {
+      "epoch": 18.716013994857313,
+      "grad_norm": 1.483940839767456,
+      "learning_rate": 2.5567036830596408e-05,
+      "loss": 4.9494,
+      "step": 444000
+    },
+    {
+      "epoch": 18.737090587193862,
+      "grad_norm": 1.477479338645935,
+      "learning_rate": 2.5481284568880503e-05,
+      "loss": 4.9522,
+      "step": 444500
+    },
+    {
+      "epoch": 18.758167179530414,
+      "grad_norm": 1.4568766355514526,
+      "learning_rate": 2.5395532307164605e-05,
+      "loss": 4.9459,
+      "step": 445000
+    },
+    {
+      "epoch": 18.779243771866966,
+      "grad_norm": 1.5580699443817139,
+      "learning_rate": 2.53097800454487e-05,
+      "loss": 4.9541,
+      "step": 445500
+    },
+    {
+      "epoch": 18.800320364203515,
+      "grad_norm": 1.4616845846176147,
+      "learning_rate": 2.5224027783732794e-05,
+      "loss": 4.9529,
+      "step": 446000
+    },
+    {
+      "epoch": 18.821396956540067,
+      "grad_norm": 1.5197380781173706,
+      "learning_rate": 2.5138275522016896e-05,
+      "loss": 4.9438,
+      "step": 446500
+    },
+    {
+      "epoch": 18.84247354887662,
+      "grad_norm": 1.4391001462936401,
+      "learning_rate": 2.505252326030099e-05,
+      "loss": 4.9473,
+      "step": 447000
+    },
+    {
+      "epoch": 18.863550141213167,
+      "grad_norm": 1.4701675176620483,
+      "learning_rate": 2.496677099858509e-05,
+      "loss": 4.9482,
+      "step": 447500
+    },
+    {
+      "epoch": 18.88462673354972,
+      "grad_norm": 1.436396837234497,
+      "learning_rate": 2.4881018736869184e-05,
+      "loss": 4.9512,
+      "step": 448000
+    },
+    {
+      "epoch": 18.90570332588627,
+      "grad_norm": 1.4701218605041504,
+      "learning_rate": 2.4795266475153283e-05,
+      "loss": 4.9464,
+      "step": 448500
+    },
+    {
+      "epoch": 18.92677991822282,
+      "grad_norm": 1.4442418813705444,
+      "learning_rate": 2.470951421343738e-05,
+      "loss": 4.9526,
+      "step": 449000
+    },
+    {
+      "epoch": 18.947856510559372,
+      "grad_norm": 1.4531573057174683,
+      "learning_rate": 2.4623761951721476e-05,
+      "loss": 4.9467,
+      "step": 449500
+    },
+    {
+      "epoch": 18.968933102895924,
+      "grad_norm": 1.4915395975112915,
+      "learning_rate": 2.4538009690005574e-05,
+      "loss": 4.9485,
+      "step": 450000
+    },
+    {
+      "epoch": 18.968933102895924,
+      "eval_accuracy": 0.7216983023676281,
+      "eval_loss": 4.682538986206055,
+      "eval_runtime": 420.9053,
+      "eval_samples_per_second": 286.777,
+      "eval_steps_per_second": 2.989,
+      "step": 450000
+    },
+    {
+      "epoch": 18.990009695232473,
+      "grad_norm": 1.460379958152771,
+      "learning_rate": 2.4452257428289672e-05,
+      "loss": 4.9454,
+      "step": 450500
+    },
+    {
+      "epoch": 19.011086287569025,
+      "grad_norm": 1.4832719564437866,
+      "learning_rate": 2.436650516657377e-05,
+      "loss": 4.9449,
+      "step": 451000
+    },
+    {
+      "epoch": 19.032162879905577,
+      "grad_norm": 1.4601773023605347,
+      "learning_rate": 2.428075290485787e-05,
+      "loss": 4.9428,
+      "step": 451500
+    },
+    {
+      "epoch": 19.05323947224213,
+      "grad_norm": 1.4579888582229614,
+      "learning_rate": 2.4195000643141964e-05,
+      "loss": 4.9446,
+      "step": 452000
+    },
+    {
+      "epoch": 19.074316064578678,
+      "grad_norm": 1.4399833679199219,
+      "learning_rate": 2.4109248381426062e-05,
+      "loss": 4.9479,
+      "step": 452500
+    },
+    {
+      "epoch": 19.09539265691523,
+      "grad_norm": 1.4426056146621704,
+      "learning_rate": 2.402349611971016e-05,
+      "loss": 4.9461,
+      "step": 453000
+    },
+    {
+      "epoch": 19.116469249251782,
+      "grad_norm": 1.4141095876693726,
+      "learning_rate": 2.3937743857994255e-05,
+      "loss": 4.9416,
+      "step": 453500
+    },
+    {
+      "epoch": 19.13754584158833,
+      "grad_norm": 1.45500648021698,
+      "learning_rate": 2.3851991596278354e-05,
+      "loss": 4.9467,
+      "step": 454000
+    },
+    {
+      "epoch": 19.158622433924883,
+      "grad_norm": 1.4800188541412354,
+      "learning_rate": 2.3766239334562452e-05,
+      "loss": 4.9436,
+      "step": 454500
+    },
+    {
+      "epoch": 19.179699026261435,
+      "grad_norm": 1.4849153757095337,
+      "learning_rate": 2.3680487072846547e-05,
+      "loss": 4.9413,
+      "step": 455000
+    },
+    {
+      "epoch": 19.200775618597984,
+      "grad_norm": 1.4700464010238647,
+      "learning_rate": 2.3594734811130645e-05,
+      "loss": 4.9395,
+      "step": 455500
+    },
+    {
+      "epoch": 19.221852210934536,
+      "grad_norm": 1.4519753456115723,
+      "learning_rate": 2.3508982549414744e-05,
+      "loss": 4.9421,
+      "step": 456000
+    },
+    {
+      "epoch": 19.242928803271088,
+      "grad_norm": 1.4615542888641357,
+      "learning_rate": 2.342323028769884e-05,
+      "loss": 4.9437,
+      "step": 456500
+    },
+    {
+      "epoch": 19.264005395607636,
+      "grad_norm": 1.4394351243972778,
+      "learning_rate": 2.3337478025982937e-05,
+      "loss": 4.9421,
+      "step": 457000
+    },
+    {
+      "epoch": 19.28508198794419,
+      "grad_norm": 1.476636290550232,
+      "learning_rate": 2.3251725764267035e-05,
+      "loss": 4.9438,
+      "step": 457500
+    },
+    {
+      "epoch": 19.30615858028074,
+      "grad_norm": 1.4645787477493286,
+      "learning_rate": 2.316597350255113e-05,
+      "loss": 4.9481,
+      "step": 458000
+    },
+    {
+      "epoch": 19.327235172617293,
+      "grad_norm": 1.4503345489501953,
+      "learning_rate": 2.308022124083523e-05,
+      "loss": 4.9438,
+      "step": 458500
+    },
+    {
+      "epoch": 19.34831176495384,
+      "grad_norm": 1.4796425104141235,
+      "learning_rate": 2.2994468979119323e-05,
+      "loss": 4.946,
+      "step": 459000
+    },
+    {
+      "epoch": 19.369388357290394,
+      "grad_norm": 1.4773235321044922,
+      "learning_rate": 2.290871671740342e-05,
+      "loss": 4.9425,
+      "step": 459500
+    },
+    {
+      "epoch": 19.390464949626946,
+      "grad_norm": 1.4007508754730225,
+      "learning_rate": 2.282296445568752e-05,
+      "loss": 4.9439,
+      "step": 460000
+    },
+    {
+      "epoch": 19.390464949626946,
+      "eval_accuracy": 0.7221639694947075,
+      "eval_loss": 4.679049491882324,
+      "eval_runtime": 366.7129,
+      "eval_samples_per_second": 329.157,
+      "eval_steps_per_second": 3.43,
+      "step": 460000
+    },
+    {
+      "epoch": 19.411541541963494,
+      "grad_norm": 1.4662177562713623,
+      "learning_rate": 2.2737212193971615e-05,
+      "loss": 4.9409,
+      "step": 460500
+    },
+    {
+      "epoch": 19.432618134300046,
+      "grad_norm": 1.4514259099960327,
+      "learning_rate": 2.2651459932255713e-05,
+      "loss": 4.9464,
+      "step": 461000
+    },
+    {
+      "epoch": 19.4536947266366,
+      "grad_norm": 1.4648381471633911,
+      "learning_rate": 2.256570767053981e-05,
+      "loss": 4.9399,
+      "step": 461500
+    },
+    {
+      "epoch": 19.474771318973147,
+      "grad_norm": 1.4359486103057861,
+      "learning_rate": 2.2479955408823906e-05,
+      "loss": 4.9397,
+      "step": 462000
+    },
+    {
+      "epoch": 19.4958479113097,
+      "grad_norm": 1.45102059841156,
+      "learning_rate": 2.2394203147108005e-05,
+      "loss": 4.9429,
+      "step": 462500
+    },
+    {
+      "epoch": 19.51692450364625,
+      "grad_norm": 1.461331844329834,
+      "learning_rate": 2.2308450885392103e-05,
+      "loss": 4.9389,
+      "step": 463000
+    },
+    {
+      "epoch": 19.5380010959828,
+      "grad_norm": 1.452243447303772,
+      "learning_rate": 2.22226986236762e-05,
+      "loss": 4.9403,
+      "step": 463500
+    },
+    {
+      "epoch": 19.559077688319352,
+      "grad_norm": 1.437991738319397,
+      "learning_rate": 2.21369463619603e-05,
+      "loss": 4.9416,
+      "step": 464000
+    },
+    {
+      "epoch": 19.580154280655904,
+      "grad_norm": 1.5059435367584229,
+      "learning_rate": 2.2051194100244394e-05,
+      "loss": 4.9405,
+      "step": 464500
+    },
+    {
+      "epoch": 19.601230872992456,
+      "grad_norm": 1.494391679763794,
+      "learning_rate": 2.1965441838528493e-05,
+      "loss": 4.9426,
+      "step": 465000
+    },
+    {
+      "epoch": 19.622307465329005,
+      "grad_norm": 1.4394044876098633,
+      "learning_rate": 2.187968957681259e-05,
+      "loss": 4.9406,
+      "step": 465500
+    },
+    {
+      "epoch": 19.643384057665557,
+      "grad_norm": 1.4123703241348267,
+      "learning_rate": 2.179393731509669e-05,
+      "loss": 4.9432,
+      "step": 466000
+    },
+    {
+      "epoch": 19.66446065000211,
+      "grad_norm": 1.4652270078659058,
+      "learning_rate": 2.1708185053380784e-05,
+      "loss": 4.9388,
+      "step": 466500
+    },
+    {
+      "epoch": 19.685537242338658,
+      "grad_norm": 1.4578845500946045,
+      "learning_rate": 2.1622432791664883e-05,
+      "loss": 4.9434,
+      "step": 467000
+    },
+    {
+      "epoch": 19.70661383467521,
+      "grad_norm": 1.5021026134490967,
+      "learning_rate": 2.1536680529948978e-05,
+      "loss": 4.9402,
+      "step": 467500
+    },
+    {
+      "epoch": 19.727690427011762,
+      "grad_norm": 1.455981969833374,
+      "learning_rate": 2.1450928268233076e-05,
+      "loss": 4.9417,
+      "step": 468000
+    },
+    {
+      "epoch": 19.74876701934831,
+      "grad_norm": 1.4234529733657837,
+      "learning_rate": 2.1365176006517174e-05,
+      "loss": 4.9384,
+      "step": 468500
+    },
+    {
+      "epoch": 19.769843611684863,
+      "grad_norm": 1.482045292854309,
+      "learning_rate": 2.127942374480127e-05,
+      "loss": 4.9394,
+      "step": 469000
+    },
+    {
+      "epoch": 19.790920204021415,
+      "grad_norm": 1.503816843032837,
+      "learning_rate": 2.1193671483085367e-05,
+      "loss": 4.9451,
+      "step": 469500
+    },
+    {
+      "epoch": 19.811996796357963,
+      "grad_norm": 1.4746991395950317,
+      "learning_rate": 2.1107919221369466e-05,
+      "loss": 4.9417,
+      "step": 470000
+    },
+    {
+      "epoch": 19.811996796357963,
+      "eval_accuracy": 0.7226267528181728,
+      "eval_loss": 4.675680160522461,
+      "eval_runtime": 394.3923,
+      "eval_samples_per_second": 306.056,
+      "eval_steps_per_second": 3.19,
+      "step": 470000
+    },
+    {
+      "epoch": 19.833073388694515,
+      "grad_norm": 1.4381850957870483,
+      "learning_rate": 2.102216695965356e-05,
+      "loss": 4.9423,
+      "step": 470500
+    },
+    {
+      "epoch": 19.854149981031068,
+      "grad_norm": 1.4429439306259155,
+      "learning_rate": 2.093641469793766e-05,
+      "loss": 4.9383,
+      "step": 471000
+    },
+    {
+      "epoch": 19.87522657336762,
+      "grad_norm": 1.4425567388534546,
+      "learning_rate": 2.0850662436221757e-05,
+      "loss": 4.9382,
+      "step": 471500
+    },
+    {
+      "epoch": 19.89630316570417,
+      "grad_norm": 1.4733551740646362,
+      "learning_rate": 2.0764910174505852e-05,
+      "loss": 4.9383,
+      "step": 472000
+    },
+    {
+      "epoch": 19.91737975804072,
+      "grad_norm": 1.47037935256958,
+      "learning_rate": 2.067915791278995e-05,
+      "loss": 4.9416,
+      "step": 472500
+    },
+    {
+      "epoch": 19.938456350377272,
+      "grad_norm": 1.4710288047790527,
+      "learning_rate": 2.059340565107405e-05,
+      "loss": 4.9362,
+      "step": 473000
+    },
+    {
+      "epoch": 19.95953294271382,
+      "grad_norm": 1.4038069248199463,
+      "learning_rate": 2.0507653389358144e-05,
+      "loss": 4.939,
+      "step": 473500
+    },
+    {
+      "epoch": 19.980609535050373,
+      "grad_norm": 1.4996339082717896,
+      "learning_rate": 2.0421901127642242e-05,
+      "loss": 4.9367,
+      "step": 474000
+    },
+    {
+      "epoch": 20.001686127386925,
+      "grad_norm": 1.5198642015457153,
+      "learning_rate": 2.033614886592634e-05,
+      "loss": 4.9447,
+      "step": 474500
+    },
+    {
+      "epoch": 20.022762719723474,
+      "grad_norm": 1.4490277767181396,
+      "learning_rate": 2.0250396604210435e-05,
+      "loss": 4.9374,
+      "step": 475000
+    },
+    {
+      "epoch": 20.043839312060026,
+      "grad_norm": 1.5348432064056396,
+      "learning_rate": 2.0164644342494534e-05,
+      "loss": 4.9374,
+      "step": 475500
+    },
+    {
+      "epoch": 20.064915904396578,
+      "grad_norm": 1.4712462425231934,
+      "learning_rate": 2.007889208077863e-05,
+      "loss": 4.9303,
+      "step": 476000
+    },
+    {
+      "epoch": 20.085992496733127,
+      "grad_norm": 1.4824450016021729,
+      "learning_rate": 1.9993139819062727e-05,
+      "loss": 4.9348,
+      "step": 476500
+    },
+    {
+      "epoch": 20.10706908906968,
+      "grad_norm": 1.455761194229126,
+      "learning_rate": 1.9907387557346825e-05,
+      "loss": 4.9356,
+      "step": 477000
+    },
+    {
+      "epoch": 20.12814568140623,
+      "grad_norm": 1.4821882247924805,
+      "learning_rate": 1.9821635295630923e-05,
+      "loss": 4.9362,
+      "step": 477500
+    },
+    {
+      "epoch": 20.14922227374278,
+      "grad_norm": 1.529541254043579,
+      "learning_rate": 1.9735883033915022e-05,
+      "loss": 4.9342,
+      "step": 478000
+    },
+    {
+      "epoch": 20.17029886607933,
+      "grad_norm": 1.485135555267334,
+      "learning_rate": 1.965013077219912e-05,
+      "loss": 4.9384,
+      "step": 478500
+    },
+    {
+      "epoch": 20.191375458415884,
+      "grad_norm": 1.4836335182189941,
+      "learning_rate": 1.9564378510483215e-05,
+      "loss": 4.9327,
+      "step": 479000
+    },
+    {
+      "epoch": 20.212452050752436,
+      "grad_norm": 1.496147632598877,
+      "learning_rate": 1.9478626248767313e-05,
+      "loss": 4.9385,
+      "step": 479500
+    },
+    {
+      "epoch": 20.233528643088984,
+      "grad_norm": 1.4936833381652832,
+      "learning_rate": 1.939287398705141e-05,
+      "loss": 4.9334,
+      "step": 480000
+    },
+    {
+      "epoch": 20.233528643088984,
+      "eval_accuracy": 0.7232947245541879,
+      "eval_loss": 4.671345233917236,
+      "eval_runtime": 381.2552,
+      "eval_samples_per_second": 316.602,
+      "eval_steps_per_second": 3.3,
+      "step": 480000
+    },
+    {
+      "epoch": 20.254605235425537,
+      "grad_norm": 1.4604382514953613,
+      "learning_rate": 1.9307121725335506e-05,
+      "loss": 4.9318,
+      "step": 480500
+    },
+    {
+      "epoch": 20.27568182776209,
+      "grad_norm": 1.5063279867172241,
+      "learning_rate": 1.9221369463619605e-05,
+      "loss": 4.9308,
+      "step": 481000
+    },
+    {
+      "epoch": 20.296758420098637,
+      "grad_norm": 1.4823752641677856,
+      "learning_rate": 1.9135617201903703e-05,
+      "loss": 4.9317,
+      "step": 481500
+    },
+    {
+      "epoch": 20.31783501243519,
+      "grad_norm": 1.5040032863616943,
+      "learning_rate": 1.9049864940187798e-05,
+      "loss": 4.9325,
+      "step": 482000
+    },
+    {
+      "epoch": 20.33891160477174,
+      "grad_norm": 1.4905809164047241,
+      "learning_rate": 1.8964112678471896e-05,
+      "loss": 4.9339,
+      "step": 482500
+    },
+    {
+      "epoch": 20.35998819710829,
+      "grad_norm": 1.5059491395950317,
+      "learning_rate": 1.8878360416755995e-05,
+      "loss": 4.9358,
+      "step": 483000
+    },
+    {
+      "epoch": 20.381064789444842,
+      "grad_norm": 1.4962173700332642,
+      "learning_rate": 1.879260815504009e-05,
+      "loss": 4.9336,
+      "step": 483500
+    },
+    {
+      "epoch": 20.402141381781394,
+      "grad_norm": 1.4654191732406616,
+      "learning_rate": 1.8706855893324188e-05,
+      "loss": 4.9296,
+      "step": 484000
+    },
+    {
+      "epoch": 20.423217974117943,
+      "grad_norm": 1.4832103252410889,
+      "learning_rate": 1.8621103631608283e-05,
+      "loss": 4.9344,
+      "step": 484500
+    },
+    {
+      "epoch": 20.444294566454495,
+      "grad_norm": 1.4715557098388672,
+      "learning_rate": 1.853535136989238e-05,
+      "loss": 4.931,
+      "step": 485000
+    },
+    {
+      "epoch": 20.465371158791047,
+      "grad_norm": 1.489740014076233,
+      "learning_rate": 1.844959910817648e-05,
+      "loss": 4.9296,
+      "step": 485500
+    },
+    {
+      "epoch": 20.4864477511276,
+      "grad_norm": 1.4953728914260864,
+      "learning_rate": 1.8363846846460574e-05,
+      "loss": 4.9363,
+      "step": 486000
+    },
+    {
+      "epoch": 20.507524343464148,
+      "grad_norm": 1.5110092163085938,
+      "learning_rate": 1.8278094584744673e-05,
+      "loss": 4.9284,
+      "step": 486500
+    },
+    {
+      "epoch": 20.5286009358007,
+      "grad_norm": 1.5139436721801758,
+      "learning_rate": 1.819234232302877e-05,
+      "loss": 4.9333,
+      "step": 487000
+    },
+    {
+      "epoch": 20.549677528137252,
+      "grad_norm": 1.4745526313781738,
+      "learning_rate": 1.8106590061312866e-05,
+      "loss": 4.9346,
+      "step": 487500
+    },
+    {
+      "epoch": 20.5707541204738,
+      "grad_norm": 1.4585838317871094,
+      "learning_rate": 1.8020837799596964e-05,
+      "loss": 4.9343,
+      "step": 488000
+    },
+    {
+      "epoch": 20.591830712810353,
+      "grad_norm": 1.5133434534072876,
+      "learning_rate": 1.7935085537881062e-05,
+      "loss": 4.9346,
+      "step": 488500
+    },
+    {
+      "epoch": 20.612907305146905,
+      "grad_norm": 1.5108935832977295,
+      "learning_rate": 1.7849333276165157e-05,
+      "loss": 4.9299,
+      "step": 489000
+    },
+    {
+      "epoch": 20.633983897483454,
+      "grad_norm": 1.4955495595932007,
+      "learning_rate": 1.7763581014449256e-05,
+      "loss": 4.9308,
+      "step": 489500
+    },
+    {
+      "epoch": 20.655060489820006,
+      "grad_norm": 1.4924123287200928,
+      "learning_rate": 1.7677828752733354e-05,
+      "loss": 4.929,
+      "step": 490000
+    },
+    {
+      "epoch": 20.655060489820006,
+      "eval_accuracy": 0.7238329303828497,
+      "eval_loss": 4.668636322021484,
+      "eval_runtime": 356.7638,
+      "eval_samples_per_second": 338.336,
+      "eval_steps_per_second": 3.526,
+      "step": 490000
+    },
+    {
+      "epoch": 20.676137082156558,
+      "grad_norm": 1.5233550071716309,
+      "learning_rate": 1.7592076491017452e-05,
+      "loss": 4.9324,
+      "step": 490500
+    },
+    {
+      "epoch": 20.697213674493106,
+      "grad_norm": 1.5019307136535645,
+      "learning_rate": 1.7506324229301547e-05,
+      "loss": 4.9354,
+      "step": 491000
+    },
+    {
+      "epoch": 20.71829026682966,
+      "grad_norm": 1.5046247243881226,
+      "learning_rate": 1.7420571967585646e-05,
+      "loss": 4.9329,
+      "step": 491500
+    },
+    {
+      "epoch": 20.73936685916621,
+      "grad_norm": 1.5094295740127563,
+      "learning_rate": 1.7334819705869744e-05,
+      "loss": 4.9304,
+      "step": 492000
+    },
+    {
+      "epoch": 20.760443451502763,
+      "grad_norm": 1.5048331022262573,
+      "learning_rate": 1.7249067444153842e-05,
+      "loss": 4.9347,
+      "step": 492500
+    },
+    {
+      "epoch": 20.78152004383931,
+      "grad_norm": 1.456424593925476,
+      "learning_rate": 1.7163315182437937e-05,
+      "loss": 4.9284,
+      "step": 493000
+    },
+    {
+      "epoch": 20.802596636175863,
+      "grad_norm": 1.5055689811706543,
+      "learning_rate": 1.7077562920722035e-05,
+      "loss": 4.9305,
+      "step": 493500
+    },
+    {
+      "epoch": 20.823673228512416,
+      "grad_norm": 1.4862622022628784,
+      "learning_rate": 1.6991810659006134e-05,
+      "loss": 4.9314,
+      "step": 494000
+    },
+    {
+      "epoch": 20.844749820848964,
+      "grad_norm": 1.4790401458740234,
+      "learning_rate": 1.690605839729023e-05,
+      "loss": 4.9303,
+      "step": 494500
+    },
+    {
+      "epoch": 20.865826413185516,
+      "grad_norm": 1.4665931463241577,
+      "learning_rate": 1.6820306135574327e-05,
+      "loss": 4.93,
+      "step": 495000
+    },
+    {
+      "epoch": 20.88690300552207,
+      "grad_norm": 1.5375653505325317,
+      "learning_rate": 1.6734553873858425e-05,
+      "loss": 4.9325,
+      "step": 495500
+    },
+    {
+      "epoch": 20.907979597858617,
+      "grad_norm": 1.4997636079788208,
+      "learning_rate": 1.664880161214252e-05,
+      "loss": 4.9295,
+      "step": 496000
+    },
+    {
+      "epoch": 20.92905619019517,
+      "grad_norm": 1.4691869020462036,
+      "learning_rate": 1.656304935042662e-05,
+      "loss": 4.9331,
+      "step": 496500
+    },
+    {
+      "epoch": 20.95013278253172,
+      "grad_norm": 1.4566730260849,
+      "learning_rate": 1.6477297088710717e-05,
+      "loss": 4.9307,
+      "step": 497000
+    },
+    {
+      "epoch": 20.97120937486827,
+      "grad_norm": 1.4915809631347656,
+      "learning_rate": 1.639154482699481e-05,
+      "loss": 4.9302,
+      "step": 497500
+    },
+    {
+      "epoch": 20.992285967204822,
+      "grad_norm": 1.4467136859893799,
+      "learning_rate": 1.630579256527891e-05,
+      "loss": 4.9288,
+      "step": 498000
+    },
+    {
+      "epoch": 21.013362559541374,
+      "grad_norm": 1.47294282913208,
+      "learning_rate": 1.6220040303563008e-05,
+      "loss": 4.9289,
+      "step": 498500
+    },
+    {
+      "epoch": 21.034439151877926,
+      "grad_norm": 1.5288922786712646,
+      "learning_rate": 1.6134288041847103e-05,
+      "loss": 4.9313,
+      "step": 499000
+    },
+    {
+      "epoch": 21.055515744214475,
+      "grad_norm": 1.485084891319275,
+      "learning_rate": 1.60485357801312e-05,
+      "loss": 4.9249,
+      "step": 499500
+    },
+    {
+      "epoch": 21.076592336551027,
+      "grad_norm": 1.4977340698242188,
+      "learning_rate": 1.59627835184153e-05,
+      "loss": 4.925,
+      "step": 500000
+    },
+    {
+      "epoch": 21.076592336551027,
+      "eval_accuracy": 0.7241926680687247,
+      "eval_loss": 4.6645097732543945,
+      "eval_runtime": 368.9645,
+      "eval_samples_per_second": 327.148,
+      "eval_steps_per_second": 3.41,
+      "step": 500000
+    },
+    {
+      "epoch": 21.09766892888758,
+      "grad_norm": 1.5033365488052368,
+      "learning_rate": 1.5877031256699395e-05,
+      "loss": 4.9252,
+      "step": 500500
+    },
+    {
+      "epoch": 21.118745521224128,
+      "grad_norm": 1.435158133506775,
+      "learning_rate": 1.5791278994983493e-05,
+      "loss": 4.9281,
+      "step": 501000
+    },
+    {
+      "epoch": 21.13982211356068,
+      "grad_norm": 1.4631190299987793,
+      "learning_rate": 1.570552673326759e-05,
+      "loss": 4.9236,
+      "step": 501500
+    },
+    {
+      "epoch": 21.16089870589723,
+      "grad_norm": 1.4686282873153687,
+      "learning_rate": 1.5619774471551686e-05,
+      "loss": 4.9282,
+      "step": 502000
+    },
+    {
+      "epoch": 21.18197529823378,
+      "grad_norm": 1.4500598907470703,
+      "learning_rate": 1.5534022209835785e-05,
+      "loss": 4.9276,
+      "step": 502500
+    },
+    {
+      "epoch": 21.203051890570332,
+      "grad_norm": 1.4320921897888184,
+      "learning_rate": 1.544826994811988e-05,
+      "loss": 4.9261,
+      "step": 503000
+    },
+    {
+      "epoch": 21.224128482906885,
+      "grad_norm": 1.4765352010726929,
+      "learning_rate": 1.5362517686403978e-05,
+      "loss": 4.9269,
+      "step": 503500
+    },
+    {
+      "epoch": 21.245205075243433,
+      "grad_norm": 1.5027906894683838,
+      "learning_rate": 1.5276765424688076e-05,
+      "loss": 4.9272,
+      "step": 504000
+    },
+    {
+      "epoch": 21.266281667579985,
+      "grad_norm": 1.555532455444336,
+      "learning_rate": 1.5191013162972173e-05,
+      "loss": 4.9256,
+      "step": 504500
+    },
+    {
+      "epoch": 21.287358259916537,
+      "grad_norm": 1.498854160308838,
+      "learning_rate": 1.5105260901256271e-05,
+      "loss": 4.9252,
+      "step": 505000
+    },
+    {
+      "epoch": 21.308434852253086,
+      "grad_norm": 1.4815778732299805,
+      "learning_rate": 1.501950863954037e-05,
+      "loss": 4.9258,
+      "step": 505500
+    },
+    {
+      "epoch": 21.329511444589638,
+      "grad_norm": 1.457384705543518,
+      "learning_rate": 1.4933756377824464e-05,
+      "loss": 4.9236,
+      "step": 506000
+    },
+    {
+      "epoch": 21.35058803692619,
+      "grad_norm": 1.4688811302185059,
+      "learning_rate": 1.4848004116108563e-05,
+      "loss": 4.9196,
+      "step": 506500
+    },
+    {
+      "epoch": 21.371664629262742,
+      "grad_norm": 1.4677354097366333,
+      "learning_rate": 1.4762251854392661e-05,
+      "loss": 4.9268,
+      "step": 507000
+    },
+    {
+      "epoch": 21.39274122159929,
+      "grad_norm": 1.5140643119812012,
+      "learning_rate": 1.4676499592676758e-05,
+      "loss": 4.9271,
+      "step": 507500
+    },
+    {
+      "epoch": 21.413817813935843,
+      "grad_norm": 1.4800934791564941,
+      "learning_rate": 1.4590747330960856e-05,
+      "loss": 4.926,
+      "step": 508000
+    },
+    {
+      "epoch": 21.434894406272395,
+      "grad_norm": 1.4839015007019043,
+      "learning_rate": 1.4504995069244952e-05,
+      "loss": 4.9225,
+      "step": 508500
+    },
+    {
+      "epoch": 21.455970998608944,
+      "grad_norm": 1.4639065265655518,
+      "learning_rate": 1.4419242807529049e-05,
+      "loss": 4.9242,
+      "step": 509000
+    },
+    {
+      "epoch": 21.477047590945496,
+      "grad_norm": 1.4416567087173462,
+      "learning_rate": 1.4333490545813147e-05,
+      "loss": 4.9249,
+      "step": 509500
+    },
+    {
+      "epoch": 21.498124183282048,
+      "grad_norm": 1.5041029453277588,
+      "learning_rate": 1.4247738284097246e-05,
+      "loss": 4.9207,
+      "step": 510000
+    },
+    {
+      "epoch": 21.498124183282048,
+      "eval_accuracy": 0.7245710912067751,
+      "eval_loss": 4.661778926849365,
+      "eval_runtime": 364.8703,
+      "eval_samples_per_second": 330.819,
+      "eval_steps_per_second": 3.448,
+      "step": 510000
+    },
+    {
+      "epoch": 21.519200775618597,
+      "grad_norm": 1.4870781898498535,
+      "learning_rate": 1.416198602238134e-05,
+      "loss": 4.9262,
+      "step": 510500
+    },
+    {
+      "epoch": 21.54027736795515,
+      "grad_norm": 1.4189505577087402,
+      "learning_rate": 1.4076233760665439e-05,
+      "loss": 4.9227,
+      "step": 511000
+    },
+    {
+      "epoch": 21.5613539602917,
+      "grad_norm": 1.4771302938461304,
+      "learning_rate": 1.3990481498949534e-05,
+      "loss": 4.9202,
+      "step": 511500
+    },
+    {
+      "epoch": 21.58243055262825,
+      "grad_norm": 1.487554669380188,
+      "learning_rate": 1.3904729237233632e-05,
+      "loss": 4.9254,
+      "step": 512000
+    },
+    {
+      "epoch": 21.6035071449648,
+      "grad_norm": 1.4512929916381836,
+      "learning_rate": 1.381897697551773e-05,
+      "loss": 4.925,
+      "step": 512500
+    },
+    {
+      "epoch": 21.624583737301354,
+      "grad_norm": 1.5105007886886597,
+      "learning_rate": 1.3733224713801825e-05,
+      "loss": 4.9273,
+      "step": 513000
+    },
+    {
+      "epoch": 21.645660329637906,
+      "grad_norm": 1.4862252473831177,
+      "learning_rate": 1.3647472452085924e-05,
+      "loss": 4.928,
+      "step": 513500
+    },
+    {
+      "epoch": 21.666736921974454,
+      "grad_norm": 1.4831256866455078,
+      "learning_rate": 1.3561720190370022e-05,
+      "loss": 4.9244,
+      "step": 514000
+    },
+    {
+      "epoch": 21.687813514311006,
+      "grad_norm": 1.4580435752868652,
+      "learning_rate": 1.3475967928654119e-05,
+      "loss": 4.9226,
+      "step": 514500
+    },
+    {
+      "epoch": 21.70889010664756,
+      "grad_norm": 1.4962964057922363,
+      "learning_rate": 1.3390215666938217e-05,
+      "loss": 4.9252,
+      "step": 515000
+    },
+    {
+      "epoch": 21.729966698984107,
+      "grad_norm": 1.480322241783142,
+      "learning_rate": 1.3304463405222315e-05,
+      "loss": 4.9253,
+      "step": 515500
+    },
+    {
+      "epoch": 21.75104329132066,
+      "grad_norm": 1.4653795957565308,
+      "learning_rate": 1.321871114350641e-05,
+      "loss": 4.9269,
+      "step": 516000
+    },
+    {
+      "epoch": 21.77211988365721,
+      "grad_norm": 1.532654881477356,
+      "learning_rate": 1.3132958881790508e-05,
+      "loss": 4.9199,
+      "step": 516500
+    },
+    {
+      "epoch": 21.79319647599376,
+      "grad_norm": 1.4565072059631348,
+      "learning_rate": 1.3047206620074607e-05,
+      "loss": 4.9206,
+      "step": 517000
+    },
+    {
+      "epoch": 21.814273068330312,
+      "grad_norm": 1.5093743801116943,
+      "learning_rate": 1.2961454358358702e-05,
+      "loss": 4.9226,
+      "step": 517500
+    },
+    {
+      "epoch": 21.835349660666864,
+      "grad_norm": 1.5220152139663696,
+      "learning_rate": 1.28757020966428e-05,
+      "loss": 4.9193,
+      "step": 518000
+    },
+    {
+      "epoch": 21.856426253003413,
+      "grad_norm": 1.4584753513336182,
+      "learning_rate": 1.2789949834926898e-05,
+      "loss": 4.9202,
+      "step": 518500
+    },
+    {
+      "epoch": 21.877502845339965,
+      "grad_norm": 1.4870339632034302,
+      "learning_rate": 1.2704197573210993e-05,
+      "loss": 4.9267,
+      "step": 519000
+    },
+    {
+      "epoch": 21.898579437676517,
+      "grad_norm": 1.4503830671310425,
+      "learning_rate": 1.2618445311495092e-05,
+      "loss": 4.9199,
+      "step": 519500
+    },
+    {
+      "epoch": 21.91965603001307,
+      "grad_norm": 1.5390998125076294,
+      "learning_rate": 1.2532693049779186e-05,
+      "loss": 4.9177,
+      "step": 520000
+    },
+    {
+      "epoch": 21.91965603001307,
+      "eval_accuracy": 0.7250243376734957,
+      "eval_loss": 4.6599016189575195,
+      "eval_runtime": 347.6788,
+      "eval_samples_per_second": 347.177,
+      "eval_steps_per_second": 3.618,
+      "step": 520000
+    },
+    {
+      "epoch": 21.940732622349618,
+      "grad_norm": 1.5402497053146362,
+      "learning_rate": 1.2446940788063285e-05,
+      "loss": 4.9213,
+      "step": 520500
+    },
+    {
+      "epoch": 21.96180921468617,
+      "grad_norm": 1.4884966611862183,
+      "learning_rate": 1.2361188526347383e-05,
+      "loss": 4.9216,
+      "step": 521000
+    },
+    {
+      "epoch": 21.982885807022722,
+      "grad_norm": 1.4971798658370972,
+      "learning_rate": 1.227543626463148e-05,
+      "loss": 4.9223,
+      "step": 521500
+    },
+    {
+      "epoch": 22.00396239935927,
+      "grad_norm": 1.5470681190490723,
+      "learning_rate": 1.2189684002915578e-05,
+      "loss": 4.9181,
+      "step": 522000
+    },
+    {
+      "epoch": 22.025038991695823,
+      "grad_norm": 1.508211374282837,
+      "learning_rate": 1.2103931741199675e-05,
+      "loss": 4.9176,
+      "step": 522500
+    },
+    {
+      "epoch": 22.046115584032375,
+      "grad_norm": 1.4989255666732788,
+      "learning_rate": 1.2018179479483773e-05,
+      "loss": 4.9199,
+      "step": 523000
+    },
+    {
+      "epoch": 22.067192176368923,
+      "grad_norm": 1.476397156715393,
+      "learning_rate": 1.193242721776787e-05,
+      "loss": 4.92,
+      "step": 523500
+    },
+    {
+      "epoch": 22.088268768705476,
+      "grad_norm": 1.49717378616333,
+      "learning_rate": 1.1846674956051966e-05,
+      "loss": 4.9233,
+      "step": 524000
+    },
+    {
+      "epoch": 22.109345361042028,
+      "grad_norm": 1.5120691061019897,
+      "learning_rate": 1.1760922694336064e-05,
+      "loss": 4.9162,
+      "step": 524500
+    },
+    {
+      "epoch": 22.130421953378576,
+      "grad_norm": 1.5115807056427002,
+      "learning_rate": 1.1675170432620161e-05,
+      "loss": 4.9233,
+      "step": 525000
+    },
+    {
+      "epoch": 22.15149854571513,
+      "grad_norm": 1.4722729921340942,
+      "learning_rate": 1.1589418170904258e-05,
+      "loss": 4.9157,
+      "step": 525500
+    },
+    {
+      "epoch": 22.17257513805168,
+      "grad_norm": 1.4607603549957275,
+      "learning_rate": 1.1503665909188354e-05,
+      "loss": 4.9161,
+      "step": 526000
+    },
+    {
+      "epoch": 22.193651730388233,
+      "grad_norm": 1.5532848834991455,
+      "learning_rate": 1.1417913647472453e-05,
+      "loss": 4.9167,
+      "step": 526500
+    },
+    {
+      "epoch": 22.21472832272478,
+      "grad_norm": 1.5053950548171997,
+      "learning_rate": 1.133216138575655e-05,
+      "loss": 4.9175,
+      "step": 527000
+    },
+    {
+      "epoch": 22.235804915061333,
+      "grad_norm": 1.5315022468566895,
+      "learning_rate": 1.1246409124040646e-05,
+      "loss": 4.9159,
+      "step": 527500
+    },
+    {
+      "epoch": 22.256881507397885,
+      "grad_norm": 1.4763540029525757,
+      "learning_rate": 1.1160656862324744e-05,
+      "loss": 4.9198,
+      "step": 528000
+    },
+    {
+      "epoch": 22.277958099734434,
+      "grad_norm": 1.4806019067764282,
+      "learning_rate": 1.1074904600608842e-05,
+      "loss": 4.9187,
+      "step": 528500
+    },
+    {
+      "epoch": 22.299034692070986,
+      "grad_norm": 1.5485577583312988,
+      "learning_rate": 1.0989152338892939e-05,
+      "loss": 4.9144,
+      "step": 529000
+    },
+    {
+      "epoch": 22.320111284407538,
+      "grad_norm": 1.596684217453003,
+      "learning_rate": 1.0903400077177037e-05,
+      "loss": 4.9193,
+      "step": 529500
+    },
+    {
+      "epoch": 22.341187876744087,
+      "grad_norm": 1.4631584882736206,
+      "learning_rate": 1.0817647815461134e-05,
+      "loss": 4.9191,
+      "step": 530000
+    },
+    {
+      "epoch": 22.341187876744087,
+      "eval_accuracy": 0.7252325834950624,
+      "eval_loss": 4.658360004425049,
+      "eval_runtime": 326.61,
+      "eval_samples_per_second": 369.572,
+      "eval_steps_per_second": 3.852,
+      "step": 530000
+    },
+    {
+      "epoch": 22.36226446908064,
+      "grad_norm": 1.4828606843948364,
+      "learning_rate": 1.073189555374523e-05,
+      "loss": 4.9194,
+      "step": 530500
+    },
+    {
+      "epoch": 22.38334106141719,
+      "grad_norm": 1.5242230892181396,
+      "learning_rate": 1.0646143292029327e-05,
+      "loss": 4.9218,
+      "step": 531000
+    },
+    {
+      "epoch": 22.40441765375374,
+      "grad_norm": 1.5240638256072998,
+      "learning_rate": 1.0560391030313425e-05,
+      "loss": 4.9145,
+      "step": 531500
+    },
+    {
+      "epoch": 22.42549424609029,
+      "grad_norm": 1.5059552192687988,
+      "learning_rate": 1.0474638768597522e-05,
+      "loss": 4.9155,
+      "step": 532000
+    },
+    {
+      "epoch": 22.446570838426844,
+      "grad_norm": 1.5134317874908447,
+      "learning_rate": 1.0388886506881619e-05,
+      "loss": 4.9177,
+      "step": 532500
+    },
+    {
+      "epoch": 22.467647430763392,
+      "grad_norm": 1.527584433555603,
+      "learning_rate": 1.0303134245165717e-05,
+      "loss": 4.9203,
+      "step": 533000
+    },
+    {
+      "epoch": 22.488724023099945,
+      "grad_norm": 1.486150860786438,
+      "learning_rate": 1.0217381983449814e-05,
+      "loss": 4.9162,
+      "step": 533500
+    },
+    {
+      "epoch": 22.509800615436497,
+      "grad_norm": 1.4658702611923218,
+      "learning_rate": 1.013162972173391e-05,
+      "loss": 4.9164,
+      "step": 534000
+    },
+    {
+      "epoch": 22.53087720777305,
+      "grad_norm": 1.5152534246444702,
+      "learning_rate": 1.0045877460018009e-05,
+      "loss": 4.9182,
+      "step": 534500
+    },
+    {
+      "epoch": 22.551953800109597,
+      "grad_norm": 1.4744986295700073,
+      "learning_rate": 9.960125198302105e-06,
+      "loss": 4.9201,
+      "step": 535000
+    },
+    {
+      "epoch": 22.57303039244615,
+      "grad_norm": 1.4833065271377563,
+      "learning_rate": 9.874372936586203e-06,
+      "loss": 4.9165,
+      "step": 535500
+    },
+    {
+      "epoch": 22.5941069847827,
+      "grad_norm": 1.4888352155685425,
+      "learning_rate": 9.7886206748703e-06,
+      "loss": 4.9159,
+      "step": 536000
+    },
+    {
+      "epoch": 22.61518357711925,
+      "grad_norm": 1.511367917060852,
+      "learning_rate": 9.702868413154398e-06,
+      "loss": 4.915,
+      "step": 536500
+    },
+    {
+      "epoch": 22.636260169455802,
+      "grad_norm": 1.5214749574661255,
+      "learning_rate": 9.617116151438495e-06,
+      "loss": 4.914,
+      "step": 537000
+    },
+    {
+      "epoch": 22.657336761792354,
+      "grad_norm": 1.5288934707641602,
+      "learning_rate": 9.531363889722592e-06,
+      "loss": 4.9153,
+      "step": 537500
+    },
+    {
+      "epoch": 22.678413354128903,
+      "grad_norm": 1.456832766532898,
+      "learning_rate": 9.44561162800669e-06,
+      "loss": 4.914,
+      "step": 538000
+    },
+    {
+      "epoch": 22.699489946465455,
+      "grad_norm": 1.5246154069900513,
+      "learning_rate": 9.359859366290787e-06,
+      "loss": 4.9163,
+      "step": 538500
+    },
+    {
+      "epoch": 22.720566538802007,
+      "grad_norm": 1.4495621919631958,
+      "learning_rate": 9.274107104574883e-06,
+      "loss": 4.9177,
+      "step": 539000
+    },
+    {
+      "epoch": 22.741643131138556,
+      "grad_norm": 1.4576432704925537,
+      "learning_rate": 9.18835484285898e-06,
+      "loss": 4.9131,
+      "step": 539500
+    },
+    {
+      "epoch": 22.762719723475108,
+      "grad_norm": 1.5276992321014404,
+      "learning_rate": 9.102602581143078e-06,
+      "loss": 4.9138,
+      "step": 540000
+    },
+    {
+      "epoch": 22.762719723475108,
+      "eval_accuracy": 0.7255022191969556,
+      "eval_loss": 4.657695293426514,
+      "eval_runtime": 295.476,
+      "eval_samples_per_second": 408.514,
+      "eval_steps_per_second": 4.258,
+      "step": 540000
+    },
+    {
+      "epoch": 22.78379631581166,
+      "grad_norm": 1.5450248718261719,
+      "learning_rate": 9.016850319427175e-06,
+      "loss": 4.9187,
+      "step": 540500
+    },
+    {
+      "epoch": 22.804872908148212,
+      "grad_norm": 1.479037880897522,
+      "learning_rate": 8.931098057711271e-06,
+      "loss": 4.915,
+      "step": 541000
+    },
+    {
+      "epoch": 22.82594950048476,
+      "grad_norm": 1.4942636489868164,
+      "learning_rate": 8.84534579599537e-06,
+      "loss": 4.9162,
+      "step": 541500
+    },
+    {
+      "epoch": 22.847026092821313,
+      "grad_norm": 1.533632516860962,
+      "learning_rate": 8.759593534279468e-06,
+      "loss": 4.9146,
+      "step": 542000
+    },
+    {
+      "epoch": 22.868102685157865,
+      "grad_norm": 1.484515905380249,
+      "learning_rate": 8.673841272563565e-06,
+      "loss": 4.9138,
+      "step": 542500
+    },
+    {
+      "epoch": 22.889179277494414,
+      "grad_norm": 1.5358173847198486,
+      "learning_rate": 8.588089010847663e-06,
+      "loss": 4.9162,
+      "step": 543000
+    },
+    {
+      "epoch": 22.910255869830966,
+      "grad_norm": 1.52066171169281,
+      "learning_rate": 8.50233674913176e-06,
+      "loss": 4.915,
+      "step": 543500
+    },
+    {
+      "epoch": 22.931332462167518,
+      "grad_norm": 1.5422933101654053,
+      "learning_rate": 8.416584487415856e-06,
+      "loss": 4.9104,
+      "step": 544000
+    },
+    {
+      "epoch": 22.952409054504066,
+      "grad_norm": 1.5424706935882568,
+      "learning_rate": 8.330832225699953e-06,
+      "loss": 4.9097,
+      "step": 544500
+    },
+    {
+      "epoch": 22.97348564684062,
+      "grad_norm": 1.5450992584228516,
+      "learning_rate": 8.245079963984051e-06,
+      "loss": 4.909,
+      "step": 545000
+    },
+    {
+      "epoch": 22.99456223917717,
+      "grad_norm": 1.5338884592056274,
+      "learning_rate": 8.159327702268148e-06,
+      "loss": 4.9179,
+      "step": 545500
+    },
+    {
+      "epoch": 23.01563883151372,
+      "grad_norm": 1.4697067737579346,
+      "learning_rate": 8.073575440552244e-06,
+      "loss": 4.913,
+      "step": 546000
+    },
+    {
+      "epoch": 23.03671542385027,
+      "grad_norm": 1.5171921253204346,
+      "learning_rate": 7.987823178836343e-06,
+      "loss": 4.9122,
+      "step": 546500
+    },
+    {
+      "epoch": 23.057792016186824,
+      "grad_norm": 1.5345385074615479,
+      "learning_rate": 7.90207091712044e-06,
+      "loss": 4.9111,
+      "step": 547000
+    },
+    {
+      "epoch": 23.078868608523376,
+      "grad_norm": 1.5153064727783203,
+      "learning_rate": 7.816318655404536e-06,
+      "loss": 4.9107,
+      "step": 547500
+    },
+    {
+      "epoch": 23.099945200859924,
+      "grad_norm": 1.5465337038040161,
+      "learning_rate": 7.730566393688634e-06,
+      "loss": 4.9131,
+      "step": 548000
+    },
+    {
+      "epoch": 23.121021793196476,
+      "grad_norm": 1.4974602460861206,
+      "learning_rate": 7.64481413197273e-06,
+      "loss": 4.9127,
+      "step": 548500
+    },
+    {
+      "epoch": 23.14209838553303,
+      "grad_norm": 1.4910885095596313,
+      "learning_rate": 7.559061870256828e-06,
+      "loss": 4.9127,
+      "step": 549000
+    },
+    {
+      "epoch": 23.163174977869577,
+      "grad_norm": 1.522118091583252,
+      "learning_rate": 7.473309608540925e-06,
+      "loss": 4.9116,
+      "step": 549500
+    },
+    {
+      "epoch": 23.18425157020613,
+      "grad_norm": 1.5086324214935303,
+      "learning_rate": 7.387557346825023e-06,
+      "loss": 4.9139,
+      "step": 550000
+    },
+    {
+      "epoch": 23.18425157020613,
+      "eval_accuracy": 0.725867029098645,
+      "eval_loss": 4.653336048126221,
+      "eval_runtime": 627.4631,
+      "eval_samples_per_second": 192.371,
+      "eval_steps_per_second": 2.005,
+      "step": 550000
+    },
+    {
+      "epoch": 23.20532816254268,
+      "grad_norm": 1.5392159223556519,
+      "learning_rate": 7.3018050851091205e-06,
+      "loss": 4.9082,
+      "step": 550500
+    },
+    {
+      "epoch": 23.22640475487923,
+      "grad_norm": 1.5129756927490234,
+      "learning_rate": 7.216052823393217e-06,
+      "loss": 4.9124,
+      "step": 551000
+    },
+    {
+      "epoch": 23.247481347215782,
+      "grad_norm": 1.5607037544250488,
+      "learning_rate": 7.1303005616773155e-06,
+      "loss": 4.9118,
+      "step": 551500
+    },
+    {
+      "epoch": 23.268557939552334,
+      "grad_norm": 1.5403523445129395,
+      "learning_rate": 7.044548299961412e-06,
+      "loss": 4.9155,
+      "step": 552000
+    },
+    {
+      "epoch": 23.289634531888883,
+      "grad_norm": 1.557913899421692,
+      "learning_rate": 6.958796038245509e-06,
+      "loss": 4.9062,
+      "step": 552500
+    },
+    {
+      "epoch": 23.310711124225435,
+      "grad_norm": 1.591451644897461,
+      "learning_rate": 6.873043776529606e-06,
+      "loss": 4.9157,
+      "step": 553000
+    },
+    {
+      "epoch": 23.331787716561987,
+      "grad_norm": 1.5444737672805786,
+      "learning_rate": 6.787291514813704e-06,
+      "loss": 4.9095,
+      "step": 553500
+    },
+    {
+      "epoch": 23.35286430889854,
+      "grad_norm": 1.5382282733917236,
+      "learning_rate": 6.701539253097801e-06,
+      "loss": 4.9101,
+      "step": 554000
+    },
+    {
+      "epoch": 23.373940901235088,
+      "grad_norm": 1.5260249376296997,
+      "learning_rate": 6.615786991381898e-06,
+      "loss": 4.9107,
+      "step": 554500
+    },
+    {
+      "epoch": 23.39501749357164,
+      "grad_norm": 1.4934182167053223,
+      "learning_rate": 6.530034729665996e-06,
+      "loss": 4.9105,
+      "step": 555000
+    },
+    {
+      "epoch": 23.416094085908192,
+      "grad_norm": 1.5079596042633057,
+      "learning_rate": 6.444282467950093e-06,
+      "loss": 4.9096,
+      "step": 555500
+    },
+    {
+      "epoch": 23.43717067824474,
+      "grad_norm": 1.53147292137146,
+      "learning_rate": 6.358530206234189e-06,
+      "loss": 4.9083,
+      "step": 556000
+    },
+    {
+      "epoch": 23.458247270581293,
+      "grad_norm": 1.522903323173523,
+      "learning_rate": 6.272777944518287e-06,
+      "loss": 4.9116,
+      "step": 556500
+    },
+    {
+      "epoch": 23.479323862917845,
+      "grad_norm": 1.4818412065505981,
+      "learning_rate": 6.187025682802384e-06,
+      "loss": 4.9121,
+      "step": 557000
+    },
+    {
+      "epoch": 23.500400455254393,
+      "grad_norm": 1.5318297147750854,
+      "learning_rate": 6.101273421086482e-06,
+      "loss": 4.909,
+      "step": 557500
+    },
+    {
+      "epoch": 23.521477047590945,
+      "grad_norm": 1.5218069553375244,
+      "learning_rate": 6.015521159370579e-06,
+      "loss": 4.9101,
+      "step": 558000
+    },
+    {
+      "epoch": 23.542553639927497,
+      "grad_norm": 1.5154309272766113,
+      "learning_rate": 5.929768897654676e-06,
+      "loss": 4.9087,
+      "step": 558500
+    },
+    {
+      "epoch": 23.563630232264046,
+      "grad_norm": 1.4918776750564575,
+      "learning_rate": 5.844016635938773e-06,
+      "loss": 4.9108,
+      "step": 559000
+    },
+    {
+      "epoch": 23.584706824600598,
+      "grad_norm": 1.5136938095092773,
+      "learning_rate": 5.75826437422287e-06,
+      "loss": 4.9092,
+      "step": 559500
+    },
+    {
+      "epoch": 23.60578341693715,
+      "grad_norm": 1.5496834516525269,
+      "learning_rate": 5.672512112506967e-06,
+      "loss": 4.9098,
+      "step": 560000
+    },
+    {
+      "epoch": 23.60578341693715,
+      "eval_accuracy": 0.726400085430329,
+      "eval_loss": 4.650839805603027,
+      "eval_runtime": 323.2639,
+      "eval_samples_per_second": 373.398,
+      "eval_steps_per_second": 3.892,
+      "step": 560000
+    },
+    {
+      "epoch": 23.6268600092737,
+      "grad_norm": 1.4951565265655518,
+      "learning_rate": 5.5867598507910655e-06,
+      "loss": 4.9089,
+      "step": 560500
+    },
+    {
+      "epoch": 23.64793660161025,
+      "grad_norm": 1.5149236917495728,
+      "learning_rate": 5.501007589075162e-06,
+      "loss": 4.9093,
+      "step": 561000
+    },
+    {
+      "epoch": 23.669013193946803,
+      "grad_norm": 1.5891391038894653,
+      "learning_rate": 5.41525532735926e-06,
+      "loss": 4.9101,
+      "step": 561500
+    },
+    {
+      "epoch": 23.690089786283355,
+      "grad_norm": 1.5792932510375977,
+      "learning_rate": 5.329503065643356e-06,
+      "loss": 4.9043,
+      "step": 562000
+    },
+    {
+      "epoch": 23.711166378619904,
+      "grad_norm": 1.545493721961975,
+      "learning_rate": 5.243750803927454e-06,
+      "loss": 4.9066,
+      "step": 562500
+    },
+    {
+      "epoch": 23.732242970956456,
+      "grad_norm": 1.5572460889816284,
+      "learning_rate": 5.157998542211551e-06,
+      "loss": 4.91,
+      "step": 563000
+    },
+    {
+      "epoch": 23.753319563293008,
+      "grad_norm": 1.4952164888381958,
+      "learning_rate": 5.072246280495649e-06,
+      "loss": 4.905,
+      "step": 563500
+    },
+    {
+      "epoch": 23.774396155629557,
+      "grad_norm": 1.518404245376587,
+      "learning_rate": 4.986494018779746e-06,
+      "loss": 4.9095,
+      "step": 564000
+    },
+    {
+      "epoch": 23.79547274796611,
+      "grad_norm": 1.5382519960403442,
+      "learning_rate": 4.900741757063843e-06,
+      "loss": 4.9093,
+      "step": 564500
+    },
+    {
+      "epoch": 23.81654934030266,
+      "grad_norm": 1.5015113353729248,
+      "learning_rate": 4.81498949534794e-06,
+      "loss": 4.906,
+      "step": 565000
+    },
+    {
+      "epoch": 23.83762593263921,
+      "grad_norm": 1.5013318061828613,
+      "learning_rate": 4.729237233632038e-06,
+      "loss": 4.9072,
+      "step": 565500
+    },
+    {
+      "epoch": 23.85870252497576,
+      "grad_norm": 1.5244736671447754,
+      "learning_rate": 4.643484971916134e-06,
+      "loss": 4.9088,
+      "step": 566000
+    },
+    {
+      "epoch": 23.879779117312314,
+      "grad_norm": 1.4956146478652954,
+      "learning_rate": 4.557732710200232e-06,
+      "loss": 4.9059,
+      "step": 566500
+    },
+    {
+      "epoch": 23.900855709648862,
+      "grad_norm": 1.533063530921936,
+      "learning_rate": 4.471980448484329e-06,
+      "loss": 4.9059,
+      "step": 567000
+    },
+    {
+      "epoch": 23.921932301985414,
+      "grad_norm": 1.5465497970581055,
+      "learning_rate": 4.386228186768427e-06,
+      "loss": 4.9072,
+      "step": 567500
+    },
+    {
+      "epoch": 23.943008894321967,
+      "grad_norm": 1.5235766172409058,
+      "learning_rate": 4.300475925052524e-06,
+      "loss": 4.9059,
+      "step": 568000
+    },
+    {
+      "epoch": 23.96408548665852,
+      "grad_norm": 1.5379122495651245,
+      "learning_rate": 4.214723663336621e-06,
+      "loss": 4.9109,
+      "step": 568500
+    },
+    {
+      "epoch": 23.985162078995067,
+      "grad_norm": 1.5281672477722168,
+      "learning_rate": 4.128971401620718e-06,
+      "loss": 4.9058,
+      "step": 569000
+    },
+    {
+      "epoch": 24.00623867133162,
+      "grad_norm": 1.5308150053024292,
+      "learning_rate": 4.043219139904815e-06,
+      "loss": 4.9069,
+      "step": 569500
+    },
+    {
+      "epoch": 24.02731526366817,
+      "grad_norm": 1.4486385583877563,
+      "learning_rate": 3.957466878188912e-06,
+      "loss": 4.9063,
+      "step": 570000
+    },
+    {
+      "epoch": 24.02731526366817,
+      "eval_accuracy": 0.7265118680868156,
+      "eval_loss": 4.649716377258301,
+      "eval_runtime": 306.752,
+      "eval_samples_per_second": 393.497,
+      "eval_steps_per_second": 4.101,
+      "step": 570000
+    },
+    {
+      "epoch": 24.04839185600472,
+      "grad_norm": 1.5405749082565308,
+      "learning_rate": 3.87171461647301e-06,
+      "loss": 4.9088,
+      "step": 570500
+    },
+    {
+      "epoch": 24.069468448341272,
+      "grad_norm": 1.4908668994903564,
+      "learning_rate": 3.7859623547571067e-06,
+      "loss": 4.9035,
+      "step": 571000
+    },
+    {
+      "epoch": 24.090545040677824,
+      "grad_norm": 1.490394115447998,
+      "learning_rate": 3.7002100930412046e-06,
+      "loss": 4.9036,
+      "step": 571500
+    },
+    {
+      "epoch": 24.111621633014373,
+      "grad_norm": 1.4899578094482422,
+      "learning_rate": 3.614457831325301e-06,
+      "loss": 4.9106,
+      "step": 572000
+    },
+    {
+      "epoch": 24.132698225350925,
+      "grad_norm": 1.565929651260376,
+      "learning_rate": 3.5287055696093987e-06,
+      "loss": 4.9046,
+      "step": 572500
+    },
+    {
+      "epoch": 24.153774817687477,
+      "grad_norm": 1.514243483543396,
+      "learning_rate": 3.4429533078934957e-06,
+      "loss": 4.9061,
+      "step": 573000
+    },
+    {
+      "epoch": 24.174851410024026,
+      "grad_norm": 1.4710720777511597,
+      "learning_rate": 3.357201046177593e-06,
+      "loss": 4.9101,
+      "step": 573500
+    },
+    {
+      "epoch": 24.195928002360578,
+      "grad_norm": 1.5020426511764526,
+      "learning_rate": 3.2714487844616906e-06,
+      "loss": 4.9096,
+      "step": 574000
+    },
+    {
+      "epoch": 24.21700459469713,
+      "grad_norm": 1.5218706130981445,
+      "learning_rate": 3.1856965227457872e-06,
+      "loss": 4.9086,
+      "step": 574500
+    },
+    {
+      "epoch": 24.238081187033682,
+      "grad_norm": 1.5299206972122192,
+      "learning_rate": 3.099944261029885e-06,
+      "loss": 4.9109,
+      "step": 575000
+    },
+    {
+      "epoch": 24.25915777937023,
+      "grad_norm": 1.5117130279541016,
+      "learning_rate": 3.014191999313982e-06,
+      "loss": 4.9057,
+      "step": 575500
+    },
+    {
+      "epoch": 24.280234371706783,
+      "grad_norm": 1.5358120203018188,
+      "learning_rate": 2.928439737598079e-06,
+      "loss": 4.9025,
+      "step": 576000
+    },
+    {
+      "epoch": 24.301310964043335,
+      "grad_norm": 1.5115569829940796,
+      "learning_rate": 2.8426874758821767e-06,
+      "loss": 4.9067,
+      "step": 576500
+    },
+    {
+      "epoch": 24.322387556379883,
+      "grad_norm": 1.5157814025878906,
+      "learning_rate": 2.7569352141662737e-06,
+      "loss": 4.9005,
+      "step": 577000
+    },
+    {
+      "epoch": 24.343464148716436,
+      "grad_norm": 1.5091580152511597,
+      "learning_rate": 2.6711829524503707e-06,
+      "loss": 4.9025,
+      "step": 577500
+    },
+    {
+      "epoch": 24.364540741052988,
+      "grad_norm": 1.5568325519561768,
+      "learning_rate": 2.585430690734468e-06,
+      "loss": 4.9026,
+      "step": 578000
+    },
+    {
+      "epoch": 24.385617333389536,
+      "grad_norm": 1.5562915802001953,
+      "learning_rate": 2.4996784290185657e-06,
+      "loss": 4.9037,
+      "step": 578500
+    },
+    {
+      "epoch": 24.40669392572609,
+      "grad_norm": 1.5374467372894287,
+      "learning_rate": 2.4139261673026627e-06,
+      "loss": 4.9003,
+      "step": 579000
+    },
+    {
+      "epoch": 24.42777051806264,
+      "grad_norm": 1.5111652612686157,
+      "learning_rate": 2.3281739055867597e-06,
+      "loss": 4.9049,
+      "step": 579500
+    },
+    {
+      "epoch": 24.44884711039919,
+      "grad_norm": 1.5010154247283936,
+      "learning_rate": 2.242421643870857e-06,
+      "loss": 4.9048,
+      "step": 580000
+    },
+    {
+      "epoch": 24.44884711039919,
+      "eval_accuracy": 0.7270754365006916,
+      "eval_loss": 4.645676612854004,
+      "eval_runtime": 296.7667,
+      "eval_samples_per_second": 406.737,
+      "eval_steps_per_second": 4.239,
+      "step": 580000
+    },
+    {
+      "epoch": 24.46992370273574,
+      "grad_norm": 1.5421115159988403,
+      "learning_rate": 2.1566693821549547e-06,
+      "loss": 4.905,
+      "step": 580500
+    },
+    {
+      "epoch": 24.491000295072293,
+      "grad_norm": 1.497231125831604,
+      "learning_rate": 2.0709171204390517e-06,
+      "loss": 4.907,
+      "step": 581000
+    },
+    {
+      "epoch": 24.512076887408845,
+      "grad_norm": 1.522965908050537,
+      "learning_rate": 1.985164858723149e-06,
+      "loss": 4.901,
+      "step": 581500
+    },
+    {
+      "epoch": 24.533153479745394,
+      "grad_norm": 1.5470595359802246,
+      "learning_rate": 1.8994125970072462e-06,
+      "loss": 4.9077,
+      "step": 582000
+    },
+    {
+      "epoch": 24.554230072081946,
+      "grad_norm": 1.562029242515564,
+      "learning_rate": 1.8136603352913432e-06,
+      "loss": 4.9025,
+      "step": 582500
+    },
+    {
+      "epoch": 24.5753066644185,
+      "grad_norm": 1.5170392990112305,
+      "learning_rate": 1.7279080735754405e-06,
+      "loss": 4.9079,
+      "step": 583000
+    },
+    {
+      "epoch": 24.596383256755047,
+      "grad_norm": 1.57255220413208,
+      "learning_rate": 1.642155811859538e-06,
+      "loss": 4.9058,
+      "step": 583500
+    },
+    {
+      "epoch": 24.6174598490916,
+      "grad_norm": 1.5237170457839966,
+      "learning_rate": 1.5564035501436352e-06,
+      "loss": 4.903,
+      "step": 584000
+    },
+    {
+      "epoch": 24.63853644142815,
+      "grad_norm": 1.50190269947052,
+      "learning_rate": 1.4706512884277324e-06,
+      "loss": 4.9035,
+      "step": 584500
+    },
+    {
+      "epoch": 24.6596130337647,
+      "grad_norm": 1.4841363430023193,
+      "learning_rate": 1.3848990267118295e-06,
+      "loss": 4.9013,
+      "step": 585000
+    },
+    {
+      "epoch": 24.680689626101252,
+      "grad_norm": 1.5443657636642456,
+      "learning_rate": 1.299146764995927e-06,
+      "loss": 4.9049,
+      "step": 585500
+    },
+    {
+      "epoch": 24.701766218437804,
+      "grad_norm": 1.569656252861023,
+      "learning_rate": 1.213394503280024e-06,
+      "loss": 4.9025,
+      "step": 586000
+    },
+    {
+      "epoch": 24.722842810774353,
+      "grad_norm": 1.52163827419281,
+      "learning_rate": 1.1276422415641212e-06,
+      "loss": 4.9028,
+      "step": 586500
+    },
+    {
+      "epoch": 24.743919403110905,
+      "grad_norm": 1.5137012004852295,
+      "learning_rate": 1.0418899798482187e-06,
+      "loss": 4.9047,
+      "step": 587000
+    },
+    {
+      "epoch": 24.764995995447457,
+      "grad_norm": 1.5274319648742676,
+      "learning_rate": 9.561377181323157e-07,
+      "loss": 4.9019,
+      "step": 587500
+    },
+    {
+      "epoch": 24.786072587784005,
+      "grad_norm": 1.523327112197876,
+      "learning_rate": 8.70385456416413e-07,
+      "loss": 4.9019,
+      "step": 588000
+    },
+    {
+      "epoch": 24.807149180120557,
+      "grad_norm": 1.4792492389678955,
+      "learning_rate": 7.846331947005103e-07,
+      "loss": 4.9061,
+      "step": 588500
+    },
+    {
+      "epoch": 24.82822577245711,
+      "grad_norm": 1.5337882041931152,
+      "learning_rate": 6.988809329846075e-07,
+      "loss": 4.8985,
+      "step": 589000
+    },
+    {
+      "epoch": 24.84930236479366,
+      "grad_norm": 1.5257762670516968,
+      "learning_rate": 6.131286712687047e-07,
+      "loss": 4.903,
+      "step": 589500
+    },
+    {
+      "epoch": 24.87037895713021,
+      "grad_norm": 1.5350905656814575,
+      "learning_rate": 5.27376409552802e-07,
+      "loss": 4.9011,
+      "step": 590000
+    },
+    {
+      "epoch": 24.87037895713021,
+      "eval_accuracy": 0.7269681651922136,
+      "eval_loss": 4.646290302276611,
+      "eval_runtime": 339.1504,
+      "eval_samples_per_second": 355.907,
+      "eval_steps_per_second": 3.709,
+      "step": 590000
+    },
+    {
+      "epoch": 24.891455549466762,
+      "grad_norm": 1.498400330543518,
+      "learning_rate": 4.4162414783689917e-07,
+      "loss": 4.9024,
+      "step": 590500
+    },
+    {
+      "epoch": 24.912532141803315,
+      "grad_norm": 1.4723578691482544,
+      "learning_rate": 3.558718861209964e-07,
+      "loss": 4.9019,
+      "step": 591000
+    },
+    {
+      "epoch": 24.933608734139863,
+      "grad_norm": 1.5054519176483154,
+      "learning_rate": 2.701196244050937e-07,
+      "loss": 4.9021,
+      "step": 591500
+    },
+    {
+      "epoch": 24.954685326476415,
+      "grad_norm": 1.5530868768692017,
+      "learning_rate": 1.8436736268919094e-07,
+      "loss": 4.902,
+      "step": 592000
+    },
+    {
+      "epoch": 24.975761918812967,
+      "grad_norm": 1.4906389713287354,
+      "learning_rate": 9.861510097328816e-08,
+      "loss": 4.9024,
+      "step": 592500
+    },
+    {
+      "epoch": 24.996838511149516,
+      "grad_norm": 1.5512412786483765,
+      "learning_rate": 1.2862839257385414e-08,
+      "loss": 4.9029,
+      "step": 593000
+    },
+    {
+      "epoch": 25.0,
+      "step": 593075,
+      "total_flos": 7.663921469738496e+18,
+      "train_loss": 5.237498263496676,
+      "train_runtime": 217964.2672,
+      "train_samples_per_second": 261.206,
+      "train_steps_per_second": 2.721
+    }
+  ],
+  "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": 7.663921469738496e+18,
+  "train_batch_size": 96,
+  "trial_name": null,
+  "trial_params": null
+}