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": 2.5841891765594482,
+  "best_model_checkpoint": "distilbert_lda_5_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": 2.088611364364624,
+      "learning_rate": 5e-06,
+      "loss": 10.626,
+      "step": 500
+    },
+    {
+      "epoch": 0.04215318467310205,
+      "grad_norm": 1.1770726442337036,
+      "learning_rate": 1e-05,
+      "loss": 8.32,
+      "step": 1000
+    },
+    {
+      "epoch": 0.06322977700965308,
+      "grad_norm": 1.1559813022613525,
+      "learning_rate": 1.5e-05,
+      "loss": 7.6892,
+      "step": 1500
+    },
+    {
+      "epoch": 0.0843063693462041,
+      "grad_norm": 1.335925817489624,
+      "learning_rate": 2e-05,
+      "loss": 7.5508,
+      "step": 2000
+    },
+    {
+      "epoch": 0.10538296168275513,
+      "grad_norm": 1.2206315994262695,
+      "learning_rate": 2.5e-05,
+      "loss": 7.4815,
+      "step": 2500
+    },
+    {
+      "epoch": 0.12645955401930617,
+      "grad_norm": 1.2919621467590332,
+      "learning_rate": 3e-05,
+      "loss": 7.4383,
+      "step": 3000
+    },
+    {
+      "epoch": 0.14753614635585718,
+      "grad_norm": 0.964166522026062,
+      "learning_rate": 3.5e-05,
+      "loss": 7.4084,
+      "step": 3500
+    },
+    {
+      "epoch": 0.1686127386924082,
+      "grad_norm": 1.0329304933547974,
+      "learning_rate": 4e-05,
+      "loss": 7.3805,
+      "step": 4000
+    },
+    {
+      "epoch": 0.18968933102895924,
+      "grad_norm": 0.8332110047340393,
+      "learning_rate": 4.5e-05,
+      "loss": 7.3536,
+      "step": 4500
+    },
+    {
+      "epoch": 0.21076592336551025,
+      "grad_norm": 1.052863359451294,
+      "learning_rate": 5e-05,
+      "loss": 7.3362,
+      "step": 5000
+    },
+    {
+      "epoch": 0.2318425157020613,
+      "grad_norm": 1.403682827949524,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 7.3256,
+      "step": 5500
+    },
+    {
+      "epoch": 0.25291910803861234,
+      "grad_norm": 1.0069406032562256,
+      "learning_rate": 6e-05,
+      "loss": 7.3064,
+      "step": 6000
+    },
+    {
+      "epoch": 0.27399570037516335,
+      "grad_norm": 1.0557713508605957,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 7.3006,
+      "step": 6500
+    },
+    {
+      "epoch": 0.29507229271171437,
+      "grad_norm": 0.7878584265708923,
+      "learning_rate": 7e-05,
+      "loss": 7.2855,
+      "step": 7000
+    },
+    {
+      "epoch": 0.3161488850482654,
+      "grad_norm": 1.0351243019104004,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 7.2778,
+      "step": 7500
+    },
+    {
+      "epoch": 0.3372254773848164,
+      "grad_norm": 0.9870970845222473,
+      "learning_rate": 8e-05,
+      "loss": 7.2757,
+      "step": 8000
+    },
+    {
+      "epoch": 0.35830206972136747,
+      "grad_norm": 0.9504342675209045,
+      "learning_rate": 8.5e-05,
+      "loss": 7.2593,
+      "step": 8500
+    },
+    {
+      "epoch": 0.3793786620579185,
+      "grad_norm": 1.0653961896896362,
+      "learning_rate": 9e-05,
+      "loss": 7.2101,
+      "step": 9000
+    },
+    {
+      "epoch": 0.4004552543944695,
+      "grad_norm": 1.8898038864135742,
+      "learning_rate": 9.5e-05,
+      "loss": 7.1694,
+      "step": 9500
+    },
+    {
+      "epoch": 0.4215318467310205,
+      "grad_norm": 1.3138624429702759,
+      "learning_rate": 0.0001,
+      "loss": 7.1144,
+      "step": 10000
+    },
+    {
+      "epoch": 0.4215318467310205,
+      "eval_accuracy": 0.16429861751512692,
+      "eval_loss": 6.937150955200195,
+      "eval_runtime": 499.4333,
+      "eval_samples_per_second": 241.686,
+      "eval_steps_per_second": 2.519,
+      "step": 10000
+    },
+    {
+      "epoch": 0.4426084390675716,
+      "grad_norm": 1.4901069402694702,
+      "learning_rate": 9.99142477382841e-05,
+      "loss": 7.0901,
+      "step": 10500
+    },
+    {
+      "epoch": 0.4636850314041226,
+      "grad_norm": 1.3905450105667114,
+      "learning_rate": 9.98284954765682e-05,
+      "loss": 7.0786,
+      "step": 11000
+    },
+    {
+      "epoch": 0.4847616237406736,
+      "grad_norm": 1.021634578704834,
+      "learning_rate": 9.97427432148523e-05,
+      "loss": 7.0767,
+      "step": 11500
+    },
+    {
+      "epoch": 0.5058382160772247,
+      "grad_norm": 1.4076811075210571,
+      "learning_rate": 9.96569909531364e-05,
+      "loss": 7.0642,
+      "step": 12000
+    },
+    {
+      "epoch": 0.5269148084137757,
+      "grad_norm": 1.1568232774734497,
+      "learning_rate": 9.957123869142048e-05,
+      "loss": 7.0564,
+      "step": 12500
+    },
+    {
+      "epoch": 0.5479914007503267,
+      "grad_norm": 1.184289574623108,
+      "learning_rate": 9.94854864297046e-05,
+      "loss": 7.054,
+      "step": 13000
+    },
+    {
+      "epoch": 0.5690679930868777,
+      "grad_norm": 1.2141892910003662,
+      "learning_rate": 9.939973416798869e-05,
+      "loss": 7.0437,
+      "step": 13500
+    },
+    {
+      "epoch": 0.5901445854234287,
+      "grad_norm": 0.6611568331718445,
+      "learning_rate": 9.931398190627278e-05,
+      "loss": 7.0393,
+      "step": 14000
+    },
+    {
+      "epoch": 0.6112211777599798,
+      "grad_norm": 0.8463046550750732,
+      "learning_rate": 9.922822964455688e-05,
+      "loss": 7.0321,
+      "step": 14500
+    },
+    {
+      "epoch": 0.6322977700965308,
+      "grad_norm": 0.950988233089447,
+      "learning_rate": 9.914247738284099e-05,
+      "loss": 7.0296,
+      "step": 15000
+    },
+    {
+      "epoch": 0.6533743624330818,
+      "grad_norm": 1.0940940380096436,
+      "learning_rate": 9.905672512112507e-05,
+      "loss": 7.021,
+      "step": 15500
+    },
+    {
+      "epoch": 0.6744509547696328,
+      "grad_norm": 0.749474287033081,
+      "learning_rate": 9.897097285940917e-05,
+      "loss": 7.0175,
+      "step": 16000
+    },
+    {
+      "epoch": 0.6955275471061839,
+      "grad_norm": 1.038729190826416,
+      "learning_rate": 9.888522059769327e-05,
+      "loss": 6.988,
+      "step": 16500
+    },
+    {
+      "epoch": 0.7166041394427349,
+      "grad_norm": 1.1662368774414062,
+      "learning_rate": 9.879946833597737e-05,
+      "loss": 6.9165,
+      "step": 17000
+    },
+    {
+      "epoch": 0.737680731779286,
+      "grad_norm": 1.6647366285324097,
+      "learning_rate": 9.871371607426147e-05,
+      "loss": 6.7961,
+      "step": 17500
+    },
+    {
+      "epoch": 0.758757324115837,
+      "grad_norm": 2.357746124267578,
+      "learning_rate": 9.862796381254556e-05,
+      "loss": 6.1585,
+      "step": 18000
+    },
+    {
+      "epoch": 0.779833916452388,
+      "grad_norm": 2.189093589782715,
+      "learning_rate": 9.854221155082965e-05,
+      "loss": 5.6351,
+      "step": 18500
+    },
+    {
+      "epoch": 0.800910508788939,
+      "grad_norm": 1.828485131263733,
+      "learning_rate": 9.845645928911376e-05,
+      "loss": 5.2763,
+      "step": 19000
+    },
+    {
+      "epoch": 0.82198710112549,
+      "grad_norm": 1.5406452417373657,
+      "learning_rate": 9.837070702739786e-05,
+      "loss": 4.9832,
+      "step": 19500
+    },
+    {
+      "epoch": 0.843063693462041,
+      "grad_norm": 1.6812318563461304,
+      "learning_rate": 9.828495476568194e-05,
+      "loss": 4.7967,
+      "step": 20000
+    },
+    {
+      "epoch": 0.843063693462041,
+      "eval_accuracy": 0.47917314323052124,
+      "eval_loss": 4.354434967041016,
+      "eval_runtime": 434.005,
+      "eval_samples_per_second": 278.121,
+      "eval_steps_per_second": 2.899,
+      "step": 20000
+    },
+    {
+      "epoch": 0.864140285798592,
+      "grad_norm": 1.574847936630249,
+      "learning_rate": 9.819920250396604e-05,
+      "loss": 4.6665,
+      "step": 20500
+    },
+    {
+      "epoch": 0.8852168781351432,
+      "grad_norm": 1.3453481197357178,
+      "learning_rate": 9.811345024225015e-05,
+      "loss": 4.5628,
+      "step": 21000
+    },
+    {
+      "epoch": 0.9062934704716942,
+      "grad_norm": 1.4394464492797852,
+      "learning_rate": 9.802769798053424e-05,
+      "loss": 4.4837,
+      "step": 21500
+    },
+    {
+      "epoch": 0.9273700628082452,
+      "grad_norm": 1.1597732305526733,
+      "learning_rate": 9.794194571881834e-05,
+      "loss": 4.3974,
+      "step": 22000
+    },
+    {
+      "epoch": 0.9484466551447962,
+      "grad_norm": 1.2579827308654785,
+      "learning_rate": 9.785619345710244e-05,
+      "loss": 4.3261,
+      "step": 22500
+    },
+    {
+      "epoch": 0.9695232474813472,
+      "grad_norm": 1.1924660205841064,
+      "learning_rate": 9.777044119538653e-05,
+      "loss": 4.2595,
+      "step": 23000
+    },
+    {
+      "epoch": 0.9905998398178982,
+      "grad_norm": 1.509472370147705,
+      "learning_rate": 9.768468893367063e-05,
+      "loss": 4.2041,
+      "step": 23500
+    },
+    {
+      "epoch": 1.0116764321544494,
+      "grad_norm": 1.235950231552124,
+      "learning_rate": 9.759893667195473e-05,
+      "loss": 4.1567,
+      "step": 24000
+    },
+    {
+      "epoch": 1.0327530244910004,
+      "grad_norm": 1.1240583658218384,
+      "learning_rate": 9.751318441023883e-05,
+      "loss": 4.1161,
+      "step": 24500
+    },
+    {
+      "epoch": 1.0538296168275514,
+      "grad_norm": 1.4128776788711548,
+      "learning_rate": 9.742743214852293e-05,
+      "loss": 4.0734,
+      "step": 25000
+    },
+    {
+      "epoch": 1.0749062091641024,
+      "grad_norm": 1.3451560735702515,
+      "learning_rate": 9.734167988680703e-05,
+      "loss": 4.0354,
+      "step": 25500
+    },
+    {
+      "epoch": 1.0959828015006534,
+      "grad_norm": 1.5378894805908203,
+      "learning_rate": 9.725592762509111e-05,
+      "loss": 4.0028,
+      "step": 26000
+    },
+    {
+      "epoch": 1.1170593938372044,
+      "grad_norm": 1.313926100730896,
+      "learning_rate": 9.717017536337522e-05,
+      "loss": 3.971,
+      "step": 26500
+    },
+    {
+      "epoch": 1.1381359861737554,
+      "grad_norm": 1.1868152618408203,
+      "learning_rate": 9.708442310165932e-05,
+      "loss": 3.9489,
+      "step": 27000
+    },
+    {
+      "epoch": 1.1592125785103065,
+      "grad_norm": 1.3833489418029785,
+      "learning_rate": 9.69986708399434e-05,
+      "loss": 3.9218,
+      "step": 27500
+    },
+    {
+      "epoch": 1.1802891708468575,
+      "grad_norm": 1.2326527833938599,
+      "learning_rate": 9.69129185782275e-05,
+      "loss": 3.902,
+      "step": 28000
+    },
+    {
+      "epoch": 1.2013657631834085,
+      "grad_norm": 1.1055015325546265,
+      "learning_rate": 9.68271663165116e-05,
+      "loss": 3.8721,
+      "step": 28500
+    },
+    {
+      "epoch": 1.2224423555199595,
+      "grad_norm": 1.3235180377960205,
+      "learning_rate": 9.67414140547957e-05,
+      "loss": 3.8519,
+      "step": 29000
+    },
+    {
+      "epoch": 1.2435189478565105,
+      "grad_norm": 1.1689666509628296,
+      "learning_rate": 9.66556617930798e-05,
+      "loss": 3.8258,
+      "step": 29500
+    },
+    {
+      "epoch": 1.2645955401930615,
+      "grad_norm": 1.2586160898208618,
+      "learning_rate": 9.656990953136388e-05,
+      "loss": 3.8152,
+      "step": 30000
+    },
+    {
+      "epoch": 1.2645955401930615,
+      "eval_accuracy": 0.5920967831665076,
+      "eval_loss": 3.506312847137451,
+      "eval_runtime": 500.9689,
+      "eval_samples_per_second": 240.945,
+      "eval_steps_per_second": 2.511,
+      "step": 30000
+    },
+    {
+      "epoch": 1.2856721325296125,
+      "grad_norm": 1.2559783458709717,
+      "learning_rate": 9.6484157269648e-05,
+      "loss": 3.7923,
+      "step": 30500
+    },
+    {
+      "epoch": 1.3067487248661636,
+      "grad_norm": 1.1722824573516846,
+      "learning_rate": 9.639840500793209e-05,
+      "loss": 3.7827,
+      "step": 31000
+    },
+    {
+      "epoch": 1.3278253172027146,
+      "grad_norm": 1.420372724533081,
+      "learning_rate": 9.631265274621618e-05,
+      "loss": 3.7632,
+      "step": 31500
+    },
+    {
+      "epoch": 1.3489019095392658,
+      "grad_norm": 1.1999328136444092,
+      "learning_rate": 9.622690048450028e-05,
+      "loss": 3.7544,
+      "step": 32000
+    },
+    {
+      "epoch": 1.3699785018758166,
+      "grad_norm": 1.6110503673553467,
+      "learning_rate": 9.614114822278439e-05,
+      "loss": 3.7371,
+      "step": 32500
+    },
+    {
+      "epoch": 1.3910550942123678,
+      "grad_norm": 1.1868494749069214,
+      "learning_rate": 9.605539596106847e-05,
+      "loss": 3.7281,
+      "step": 33000
+    },
+    {
+      "epoch": 1.4121316865489189,
+      "grad_norm": 1.2073482275009155,
+      "learning_rate": 9.596964369935257e-05,
+      "loss": 3.7099,
+      "step": 33500
+    },
+    {
+      "epoch": 1.4332082788854699,
+      "grad_norm": 1.125970721244812,
+      "learning_rate": 9.588389143763667e-05,
+      "loss": 3.7018,
+      "step": 34000
+    },
+    {
+      "epoch": 1.4542848712220209,
+      "grad_norm": 1.2882457971572876,
+      "learning_rate": 9.579813917592077e-05,
+      "loss": 3.687,
+      "step": 34500
+    },
+    {
+      "epoch": 1.475361463558572,
+      "grad_norm": 1.1866166591644287,
+      "learning_rate": 9.571238691420487e-05,
+      "loss": 3.6769,
+      "step": 35000
+    },
+    {
+      "epoch": 1.496438055895123,
+      "grad_norm": 1.2313462495803833,
+      "learning_rate": 9.562663465248896e-05,
+      "loss": 3.6631,
+      "step": 35500
+    },
+    {
+      "epoch": 1.517514648231674,
+      "grad_norm": 1.1852868795394897,
+      "learning_rate": 9.554088239077306e-05,
+      "loss": 3.6486,
+      "step": 36000
+    },
+    {
+      "epoch": 1.538591240568225,
+      "grad_norm": 1.1608946323394775,
+      "learning_rate": 9.545513012905716e-05,
+      "loss": 3.6405,
+      "step": 36500
+    },
+    {
+      "epoch": 1.559667832904776,
+      "grad_norm": 1.1895480155944824,
+      "learning_rate": 9.536937786734126e-05,
+      "loss": 3.6392,
+      "step": 37000
+    },
+    {
+      "epoch": 1.580744425241327,
+      "grad_norm": 1.260651707649231,
+      "learning_rate": 9.528362560562534e-05,
+      "loss": 3.6244,
+      "step": 37500
+    },
+    {
+      "epoch": 1.601821017577878,
+      "grad_norm": 1.265782356262207,
+      "learning_rate": 9.519787334390946e-05,
+      "loss": 3.6137,
+      "step": 38000
+    },
+    {
+      "epoch": 1.622897609914429,
+      "grad_norm": 1.1333792209625244,
+      "learning_rate": 9.511212108219355e-05,
+      "loss": 3.6032,
+      "step": 38500
+    },
+    {
+      "epoch": 1.64397420225098,
+      "grad_norm": 1.1849192380905151,
+      "learning_rate": 9.502636882047764e-05,
+      "loss": 3.5949,
+      "step": 39000
+    },
+    {
+      "epoch": 1.6650507945875312,
+      "grad_norm": 1.2153621912002563,
+      "learning_rate": 9.494061655876174e-05,
+      "loss": 3.5852,
+      "step": 39500
+    },
+    {
+      "epoch": 1.686127386924082,
+      "grad_norm": 1.1532171964645386,
+      "learning_rate": 9.485486429704585e-05,
+      "loss": 3.5801,
+      "step": 40000
+    },
+    {
+      "epoch": 1.686127386924082,
+      "eval_accuracy": 0.6240521111139643,
+      "eval_loss": 3.28659725189209,
+      "eval_runtime": 465.2746,
+      "eval_samples_per_second": 259.43,
+      "eval_steps_per_second": 2.704,
+      "step": 40000
+    },
+    {
+      "epoch": 1.7072039792606333,
+      "grad_norm": 1.5950621366500854,
+      "learning_rate": 9.476911203532993e-05,
+      "loss": 3.5682,
+      "step": 40500
+    },
+    {
+      "epoch": 1.728280571597184,
+      "grad_norm": 1.1409204006195068,
+      "learning_rate": 9.468335977361403e-05,
+      "loss": 3.5631,
+      "step": 41000
+    },
+    {
+      "epoch": 1.7493571639337353,
+      "grad_norm": 1.1387003660202026,
+      "learning_rate": 9.459760751189813e-05,
+      "loss": 3.5557,
+      "step": 41500
+    },
+    {
+      "epoch": 1.770433756270286,
+      "grad_norm": 1.511974573135376,
+      "learning_rate": 9.451185525018223e-05,
+      "loss": 3.5475,
+      "step": 42000
+    },
+    {
+      "epoch": 1.7915103486068373,
+      "grad_norm": 1.4156588315963745,
+      "learning_rate": 9.442610298846633e-05,
+      "loss": 3.5394,
+      "step": 42500
+    },
+    {
+      "epoch": 1.8125869409433881,
+      "grad_norm": 1.176735281944275,
+      "learning_rate": 9.434035072675043e-05,
+      "loss": 3.5302,
+      "step": 43000
+    },
+    {
+      "epoch": 1.8336635332799394,
+      "grad_norm": 1.2431520223617554,
+      "learning_rate": 9.425459846503451e-05,
+      "loss": 3.5218,
+      "step": 43500
+    },
+    {
+      "epoch": 1.8547401256164904,
+      "grad_norm": 1.1392661333084106,
+      "learning_rate": 9.416884620331862e-05,
+      "loss": 3.514,
+      "step": 44000
+    },
+    {
+      "epoch": 1.8758167179530414,
+      "grad_norm": 1.2419050931930542,
+      "learning_rate": 9.408309394160272e-05,
+      "loss": 3.5119,
+      "step": 44500
+    },
+    {
+      "epoch": 1.8968933102895924,
+      "grad_norm": 1.460762858390808,
+      "learning_rate": 9.39973416798868e-05,
+      "loss": 3.5014,
+      "step": 45000
+    },
+    {
+      "epoch": 1.9179699026261434,
+      "grad_norm": 1.280250072479248,
+      "learning_rate": 9.39115894181709e-05,
+      "loss": 3.498,
+      "step": 45500
+    },
+    {
+      "epoch": 1.9390464949626944,
+      "grad_norm": 1.256041407585144,
+      "learning_rate": 9.382583715645502e-05,
+      "loss": 3.4863,
+      "step": 46000
+    },
+    {
+      "epoch": 1.9601230872992454,
+      "grad_norm": 1.1554384231567383,
+      "learning_rate": 9.37400848947391e-05,
+      "loss": 3.478,
+      "step": 46500
+    },
+    {
+      "epoch": 1.9811996796357965,
+      "grad_norm": 1.2954728603363037,
+      "learning_rate": 9.36543326330232e-05,
+      "loss": 3.4807,
+      "step": 47000
+    },
+    {
+      "epoch": 2.0022762719723475,
+      "grad_norm": 1.396352767944336,
+      "learning_rate": 9.35685803713073e-05,
+      "loss": 3.4757,
+      "step": 47500
+    },
+    {
+      "epoch": 2.0233528643088987,
+      "grad_norm": 1.216383457183838,
+      "learning_rate": 9.34828281095914e-05,
+      "loss": 3.4628,
+      "step": 48000
+    },
+    {
+      "epoch": 2.0444294566454495,
+      "grad_norm": 1.2102289199829102,
+      "learning_rate": 9.339707584787549e-05,
+      "loss": 3.4574,
+      "step": 48500
+    },
+    {
+      "epoch": 2.0655060489820007,
+      "grad_norm": 1.1639467477798462,
+      "learning_rate": 9.331132358615959e-05,
+      "loss": 3.448,
+      "step": 49000
+    },
+    {
+      "epoch": 2.0865826413185515,
+      "grad_norm": 1.2535285949707031,
+      "learning_rate": 9.322557132444369e-05,
+      "loss": 3.4413,
+      "step": 49500
+    },
+    {
+      "epoch": 2.1076592336551028,
+      "grad_norm": 1.195374608039856,
+      "learning_rate": 9.313981906272779e-05,
+      "loss": 3.4416,
+      "step": 50000
+    },
+    {
+      "epoch": 2.1076592336551028,
+      "eval_accuracy": 0.6409360825715527,
+      "eval_loss": 3.1632423400878906,
+      "eval_runtime": 486.6585,
+      "eval_samples_per_second": 248.03,
+      "eval_steps_per_second": 2.585,
+      "step": 50000
+    },
+    {
+      "epoch": 2.1287358259916536,
+      "grad_norm": 1.098705530166626,
+      "learning_rate": 9.305406680101189e-05,
+      "loss": 3.4403,
+      "step": 50500
+    },
+    {
+      "epoch": 2.149812418328205,
+      "grad_norm": 1.360826849937439,
+      "learning_rate": 9.296831453929597e-05,
+      "loss": 3.4311,
+      "step": 51000
+    },
+    {
+      "epoch": 2.1708890106647556,
+      "grad_norm": 1.3063390254974365,
+      "learning_rate": 9.288256227758008e-05,
+      "loss": 3.4232,
+      "step": 51500
+    },
+    {
+      "epoch": 2.191965603001307,
+      "grad_norm": 1.5551719665527344,
+      "learning_rate": 9.279681001586418e-05,
+      "loss": 3.4206,
+      "step": 52000
+    },
+    {
+      "epoch": 2.2130421953378576,
+      "grad_norm": 1.2018306255340576,
+      "learning_rate": 9.271105775414827e-05,
+      "loss": 3.4135,
+      "step": 52500
+    },
+    {
+      "epoch": 2.234118787674409,
+      "grad_norm": 1.1199662685394287,
+      "learning_rate": 9.262530549243236e-05,
+      "loss": 3.4129,
+      "step": 53000
+    },
+    {
+      "epoch": 2.2551953800109596,
+      "grad_norm": 1.14421808719635,
+      "learning_rate": 9.253955323071646e-05,
+      "loss": 3.4083,
+      "step": 53500
+    },
+    {
+      "epoch": 2.276271972347511,
+      "grad_norm": 1.1502058506011963,
+      "learning_rate": 9.245380096900056e-05,
+      "loss": 3.4066,
+      "step": 54000
+    },
+    {
+      "epoch": 2.2973485646840617,
+      "grad_norm": 1.090627908706665,
+      "learning_rate": 9.236804870728466e-05,
+      "loss": 3.3949,
+      "step": 54500
+    },
+    {
+      "epoch": 2.318425157020613,
+      "grad_norm": 1.13905930519104,
+      "learning_rate": 9.228229644556876e-05,
+      "loss": 3.3959,
+      "step": 55000
+    },
+    {
+      "epoch": 2.339501749357164,
+      "grad_norm": 1.1141719818115234,
+      "learning_rate": 9.219654418385286e-05,
+      "loss": 3.385,
+      "step": 55500
+    },
+    {
+      "epoch": 2.360578341693715,
+      "grad_norm": 1.1627238988876343,
+      "learning_rate": 9.211079192213695e-05,
+      "loss": 3.3897,
+      "step": 56000
+    },
+    {
+      "epoch": 2.381654934030266,
+      "grad_norm": 1.1090179681777954,
+      "learning_rate": 9.202503966042105e-05,
+      "loss": 3.3811,
+      "step": 56500
+    },
+    {
+      "epoch": 2.402731526366817,
+      "grad_norm": 1.2078492641448975,
+      "learning_rate": 9.193928739870514e-05,
+      "loss": 3.3765,
+      "step": 57000
+    },
+    {
+      "epoch": 2.423808118703368,
+      "grad_norm": 1.094149112701416,
+      "learning_rate": 9.185353513698925e-05,
+      "loss": 3.3719,
+      "step": 57500
+    },
+    {
+      "epoch": 2.444884711039919,
+      "grad_norm": 1.136108160018921,
+      "learning_rate": 9.176778287527335e-05,
+      "loss": 3.3697,
+      "step": 58000
+    },
+    {
+      "epoch": 2.4659613033764702,
+      "grad_norm": 1.2385177612304688,
+      "learning_rate": 9.168203061355743e-05,
+      "loss": 3.3676,
+      "step": 58500
+    },
+    {
+      "epoch": 2.487037895713021,
+      "grad_norm": 1.084393858909607,
+      "learning_rate": 9.159627835184153e-05,
+      "loss": 3.3639,
+      "step": 59000
+    },
+    {
+      "epoch": 2.5081144880495723,
+      "grad_norm": 1.1269991397857666,
+      "learning_rate": 9.151052609012564e-05,
+      "loss": 3.3614,
+      "step": 59500
+    },
+    {
+      "epoch": 2.529191080386123,
+      "grad_norm": 1.133736252784729,
+      "learning_rate": 9.142477382840973e-05,
+      "loss": 3.3557,
+      "step": 60000
+    },
+    {
+      "epoch": 2.529191080386123,
+      "eval_accuracy": 0.65438991175476,
+      "eval_loss": 3.0762295722961426,
+      "eval_runtime": 463.2599,
+      "eval_samples_per_second": 260.558,
+      "eval_steps_per_second": 2.716,
+      "step": 60000
+    },
+    {
+      "epoch": 2.5502676727226743,
+      "grad_norm": 1.1133415699005127,
+      "learning_rate": 9.133902156669383e-05,
+      "loss": 3.3487,
+      "step": 60500
+    },
+    {
+      "epoch": 2.571344265059225,
+      "grad_norm": 1.1757781505584717,
+      "learning_rate": 9.125326930497792e-05,
+      "loss": 3.3496,
+      "step": 61000
+    },
+    {
+      "epoch": 2.5924208573957763,
+      "grad_norm": 1.1235966682434082,
+      "learning_rate": 9.116751704326202e-05,
+      "loss": 3.3429,
+      "step": 61500
+    },
+    {
+      "epoch": 2.613497449732327,
+      "grad_norm": 1.2611438035964966,
+      "learning_rate": 9.108176478154612e-05,
+      "loss": 3.3382,
+      "step": 62000
+    },
+    {
+      "epoch": 2.6345740420688784,
+      "grad_norm": 1.2102614641189575,
+      "learning_rate": 9.09960125198302e-05,
+      "loss": 3.3435,
+      "step": 62500
+    },
+    {
+      "epoch": 2.655650634405429,
+      "grad_norm": 1.210385799407959,
+      "learning_rate": 9.091026025811432e-05,
+      "loss": 3.3332,
+      "step": 63000
+    },
+    {
+      "epoch": 2.6767272267419804,
+      "grad_norm": 1.3112365007400513,
+      "learning_rate": 9.082450799639842e-05,
+      "loss": 3.3313,
+      "step": 63500
+    },
+    {
+      "epoch": 2.6978038190785316,
+      "grad_norm": 1.0592286586761475,
+      "learning_rate": 9.07387557346825e-05,
+      "loss": 3.331,
+      "step": 64000
+    },
+    {
+      "epoch": 2.7188804114150824,
+      "grad_norm": 1.069266438484192,
+      "learning_rate": 9.06530034729666e-05,
+      "loss": 3.3251,
+      "step": 64500
+    },
+    {
+      "epoch": 2.739957003751633,
+      "grad_norm": 1.2145230770111084,
+      "learning_rate": 9.05672512112507e-05,
+      "loss": 3.3215,
+      "step": 65000
+    },
+    {
+      "epoch": 2.7610335960881844,
+      "grad_norm": 1.34259831905365,
+      "learning_rate": 9.04814989495348e-05,
+      "loss": 3.319,
+      "step": 65500
+    },
+    {
+      "epoch": 2.7821101884247357,
+      "grad_norm": 1.160184621810913,
+      "learning_rate": 9.03957466878189e-05,
+      "loss": 3.3195,
+      "step": 66000
+    },
+    {
+      "epoch": 2.8031867807612865,
+      "grad_norm": 1.0743733644485474,
+      "learning_rate": 9.030999442610299e-05,
+      "loss": 3.3125,
+      "step": 66500
+    },
+    {
+      "epoch": 2.8242633730978377,
+      "grad_norm": 1.07710599899292,
+      "learning_rate": 9.022424216438709e-05,
+      "loss": 3.3059,
+      "step": 67000
+    },
+    {
+      "epoch": 2.8453399654343885,
+      "grad_norm": 1.2694309949874878,
+      "learning_rate": 9.013848990267119e-05,
+      "loss": 3.3052,
+      "step": 67500
+    },
+    {
+      "epoch": 2.8664165577709397,
+      "grad_norm": 1.0658957958221436,
+      "learning_rate": 9.005273764095529e-05,
+      "loss": 3.3015,
+      "step": 68000
+    },
+    {
+      "epoch": 2.8874931501074905,
+      "grad_norm": 1.4015356302261353,
+      "learning_rate": 8.996698537923937e-05,
+      "loss": 3.3045,
+      "step": 68500
+    },
+    {
+      "epoch": 2.9085697424440418,
+      "grad_norm": 1.1149303913116455,
+      "learning_rate": 8.988123311752348e-05,
+      "loss": 3.2951,
+      "step": 69000
+    },
+    {
+      "epoch": 2.9296463347805926,
+      "grad_norm": 1.1507779359817505,
+      "learning_rate": 8.979548085580758e-05,
+      "loss": 3.296,
+      "step": 69500
+    },
+    {
+      "epoch": 2.950722927117144,
+      "grad_norm": 1.1876226663589478,
+      "learning_rate": 8.970972859409167e-05,
+      "loss": 3.2898,
+      "step": 70000
+    },
+    {
+      "epoch": 2.950722927117144,
+      "eval_accuracy": 0.6625236454981528,
+      "eval_loss": 3.02008056640625,
+      "eval_runtime": 473.5217,
+      "eval_samples_per_second": 254.911,
+      "eval_steps_per_second": 2.657,
+      "step": 70000
+    },
+    {
+      "epoch": 2.9717995194536946,
+      "grad_norm": 1.0792324542999268,
+      "learning_rate": 8.962397633237576e-05,
+      "loss": 3.2916,
+      "step": 70500
+    },
+    {
+      "epoch": 2.992876111790246,
+      "grad_norm": 1.1097954511642456,
+      "learning_rate": 8.953822407065988e-05,
+      "loss": 3.2863,
+      "step": 71000
+    },
+    {
+      "epoch": 3.0139527041267966,
+      "grad_norm": 1.1607153415679932,
+      "learning_rate": 8.945247180894396e-05,
+      "loss": 3.2851,
+      "step": 71500
+    },
+    {
+      "epoch": 3.035029296463348,
+      "grad_norm": 1.1246107816696167,
+      "learning_rate": 8.936671954722806e-05,
+      "loss": 3.2792,
+      "step": 72000
+    },
+    {
+      "epoch": 3.0561058887998986,
+      "grad_norm": 1.0766249895095825,
+      "learning_rate": 8.928096728551216e-05,
+      "loss": 3.2805,
+      "step": 72500
+    },
+    {
+      "epoch": 3.07718248113645,
+      "grad_norm": 1.2137751579284668,
+      "learning_rate": 8.919521502379626e-05,
+      "loss": 3.2741,
+      "step": 73000
+    },
+    {
+      "epoch": 3.0982590734730007,
+      "grad_norm": 1.1317554712295532,
+      "learning_rate": 8.910946276208035e-05,
+      "loss": 3.2764,
+      "step": 73500
+    },
+    {
+      "epoch": 3.119335665809552,
+      "grad_norm": 1.223796010017395,
+      "learning_rate": 8.902371050036445e-05,
+      "loss": 3.2651,
+      "step": 74000
+    },
+    {
+      "epoch": 3.140412258146103,
+      "grad_norm": 1.0957058668136597,
+      "learning_rate": 8.893795823864855e-05,
+      "loss": 3.2708,
+      "step": 74500
+    },
+    {
+      "epoch": 3.161488850482654,
+      "grad_norm": 1.1239248514175415,
+      "learning_rate": 8.885220597693265e-05,
+      "loss": 3.2671,
+      "step": 75000
+    },
+    {
+      "epoch": 3.182565442819205,
+      "grad_norm": 1.2558660507202148,
+      "learning_rate": 8.876645371521675e-05,
+      "loss": 3.2611,
+      "step": 75500
+    },
+    {
+      "epoch": 3.203642035155756,
+      "grad_norm": 1.1479278802871704,
+      "learning_rate": 8.868070145350083e-05,
+      "loss": 3.2622,
+      "step": 76000
+    },
+    {
+      "epoch": 3.224718627492307,
+      "grad_norm": 1.2556383609771729,
+      "learning_rate": 8.859494919178493e-05,
+      "loss": 3.2611,
+      "step": 76500
+    },
+    {
+      "epoch": 3.245795219828858,
+      "grad_norm": 1.3326573371887207,
+      "learning_rate": 8.850919693006904e-05,
+      "loss": 3.2556,
+      "step": 77000
+    },
+    {
+      "epoch": 3.2668718121654092,
+      "grad_norm": 1.0682497024536133,
+      "learning_rate": 8.842344466835313e-05,
+      "loss": 3.2523,
+      "step": 77500
+    },
+    {
+      "epoch": 3.28794840450196,
+      "grad_norm": 1.1187690496444702,
+      "learning_rate": 8.833769240663723e-05,
+      "loss": 3.2501,
+      "step": 78000
+    },
+    {
+      "epoch": 3.3090249968385113,
+      "grad_norm": 1.389695644378662,
+      "learning_rate": 8.825194014492132e-05,
+      "loss": 3.2501,
+      "step": 78500
+    },
+    {
+      "epoch": 3.330101589175062,
+      "grad_norm": 1.2352409362792969,
+      "learning_rate": 8.816618788320542e-05,
+      "loss": 3.2485,
+      "step": 79000
+    },
+    {
+      "epoch": 3.3511781815116133,
+      "grad_norm": 1.2109867334365845,
+      "learning_rate": 8.808043562148952e-05,
+      "loss": 3.2461,
+      "step": 79500
+    },
+    {
+      "epoch": 3.372254773848164,
+      "grad_norm": 1.308921456336975,
+      "learning_rate": 8.799468335977362e-05,
+      "loss": 3.2439,
+      "step": 80000
+    },
+    {
+      "epoch": 3.372254773848164,
+      "eval_accuracy": 0.6696885103987911,
+      "eval_loss": 2.976672649383545,
+      "eval_runtime": 449.1384,
+      "eval_samples_per_second": 268.75,
+      "eval_steps_per_second": 2.801,
+      "step": 80000
+    },
+    {
+      "epoch": 3.3933313661847153,
+      "grad_norm": 1.1651296615600586,
+      "learning_rate": 8.790893109805772e-05,
+      "loss": 3.2436,
+      "step": 80500
+    },
+    {
+      "epoch": 3.414407958521266,
+      "grad_norm": 1.2016701698303223,
+      "learning_rate": 8.782317883634182e-05,
+      "loss": 3.2383,
+      "step": 81000
+    },
+    {
+      "epoch": 3.4354845508578173,
+      "grad_norm": 1.2344244718551636,
+      "learning_rate": 8.773742657462591e-05,
+      "loss": 3.2367,
+      "step": 81500
+    },
+    {
+      "epoch": 3.456561143194368,
+      "grad_norm": 1.1405915021896362,
+      "learning_rate": 8.765167431291e-05,
+      "loss": 3.2346,
+      "step": 82000
+    },
+    {
+      "epoch": 3.4776377355309194,
+      "grad_norm": 1.1531031131744385,
+      "learning_rate": 8.756592205119411e-05,
+      "loss": 3.2345,
+      "step": 82500
+    },
+    {
+      "epoch": 3.4987143278674706,
+      "grad_norm": 1.2486790418624878,
+      "learning_rate": 8.748016978947821e-05,
+      "loss": 3.2323,
+      "step": 83000
+    },
+    {
+      "epoch": 3.5197909202040214,
+      "grad_norm": 1.176477074623108,
+      "learning_rate": 8.73944175277623e-05,
+      "loss": 3.2303,
+      "step": 83500
+    },
+    {
+      "epoch": 3.540867512540572,
+      "grad_norm": 1.3879839181900024,
+      "learning_rate": 8.730866526604639e-05,
+      "loss": 3.2297,
+      "step": 84000
+    },
+    {
+      "epoch": 3.5619441048771234,
+      "grad_norm": 1.0695738792419434,
+      "learning_rate": 8.72229130043305e-05,
+      "loss": 3.2285,
+      "step": 84500
+    },
+    {
+      "epoch": 3.5830206972136747,
+      "grad_norm": 1.2656728029251099,
+      "learning_rate": 8.713716074261459e-05,
+      "loss": 3.2262,
+      "step": 85000
+    },
+    {
+      "epoch": 3.6040972895502255,
+      "grad_norm": 1.3721638917922974,
+      "learning_rate": 8.705140848089869e-05,
+      "loss": 3.2246,
+      "step": 85500
+    },
+    {
+      "epoch": 3.6251738818867767,
+      "grad_norm": 1.1258668899536133,
+      "learning_rate": 8.696565621918278e-05,
+      "loss": 3.2196,
+      "step": 86000
+    },
+    {
+      "epoch": 3.6462504742233275,
+      "grad_norm": 1.168200969696045,
+      "learning_rate": 8.687990395746688e-05,
+      "loss": 3.2193,
+      "step": 86500
+    },
+    {
+      "epoch": 3.6673270665598787,
+      "grad_norm": 1.1972341537475586,
+      "learning_rate": 8.679415169575098e-05,
+      "loss": 3.2189,
+      "step": 87000
+    },
+    {
+      "epoch": 3.6884036588964295,
+      "grad_norm": 1.1618013381958008,
+      "learning_rate": 8.670839943403508e-05,
+      "loss": 3.2175,
+      "step": 87500
+    },
+    {
+      "epoch": 3.7094802512329808,
+      "grad_norm": 1.2538747787475586,
+      "learning_rate": 8.662264717231916e-05,
+      "loss": 3.2162,
+      "step": 88000
+    },
+    {
+      "epoch": 3.7305568435695315,
+      "grad_norm": 1.1711055040359497,
+      "learning_rate": 8.653689491060328e-05,
+      "loss": 3.2136,
+      "step": 88500
+    },
+    {
+      "epoch": 3.751633435906083,
+      "grad_norm": 1.333916187286377,
+      "learning_rate": 8.645114264888737e-05,
+      "loss": 3.2117,
+      "step": 89000
+    },
+    {
+      "epoch": 3.7727100282426336,
+      "grad_norm": 1.1053155660629272,
+      "learning_rate": 8.636539038717146e-05,
+      "loss": 3.2062,
+      "step": 89500
+    },
+    {
+      "epoch": 3.793786620579185,
+      "grad_norm": 1.1377023458480835,
+      "learning_rate": 8.627963812545556e-05,
+      "loss": 3.2059,
+      "step": 90000
+    },
+    {
+      "epoch": 3.793786620579185,
+      "eval_accuracy": 0.6751935475274787,
+      "eval_loss": 2.9381518363952637,
+      "eval_runtime": 399.3733,
+      "eval_samples_per_second": 302.239,
+      "eval_steps_per_second": 3.15,
+      "step": 90000
+    },
+    {
+      "epoch": 3.8148632129157356,
+      "grad_norm": 1.1586827039718628,
+      "learning_rate": 8.619388586373967e-05,
+      "loss": 3.2095,
+      "step": 90500
+    },
+    {
+      "epoch": 3.835939805252287,
+      "grad_norm": 1.1912150382995605,
+      "learning_rate": 8.610813360202375e-05,
+      "loss": 3.2011,
+      "step": 91000
+    },
+    {
+      "epoch": 3.857016397588838,
+      "grad_norm": 1.1633180379867554,
+      "learning_rate": 8.602238134030785e-05,
+      "loss": 3.2029,
+      "step": 91500
+    },
+    {
+      "epoch": 3.878092989925389,
+      "grad_norm": 1.1038143634796143,
+      "learning_rate": 8.593662907859195e-05,
+      "loss": 3.1944,
+      "step": 92000
+    },
+    {
+      "epoch": 3.8991695822619397,
+      "grad_norm": 1.1121149063110352,
+      "learning_rate": 8.585087681687605e-05,
+      "loss": 3.1991,
+      "step": 92500
+    },
+    {
+      "epoch": 3.920246174598491,
+      "grad_norm": 1.1213479042053223,
+      "learning_rate": 8.576512455516015e-05,
+      "loss": 3.1993,
+      "step": 93000
+    },
+    {
+      "epoch": 3.941322766935042,
+      "grad_norm": 1.2445160150527954,
+      "learning_rate": 8.567937229344425e-05,
+      "loss": 3.1941,
+      "step": 93500
+    },
+    {
+      "epoch": 3.962399359271593,
+      "grad_norm": 1.1563202142715454,
+      "learning_rate": 8.559362003172834e-05,
+      "loss": 3.1941,
+      "step": 94000
+    },
+    {
+      "epoch": 3.983475951608144,
+      "grad_norm": 1.1611775159835815,
+      "learning_rate": 8.550786777001244e-05,
+      "loss": 3.1919,
+      "step": 94500
+    },
+    {
+      "epoch": 4.004552543944695,
+      "grad_norm": 1.2735614776611328,
+      "learning_rate": 8.542211550829654e-05,
+      "loss": 3.1893,
+      "step": 95000
+    },
+    {
+      "epoch": 4.025629136281246,
+      "grad_norm": 1.1780765056610107,
+      "learning_rate": 8.533636324658063e-05,
+      "loss": 3.1801,
+      "step": 95500
+    },
+    {
+      "epoch": 4.046705728617797,
+      "grad_norm": 1.2774953842163086,
+      "learning_rate": 8.525061098486474e-05,
+      "loss": 3.1856,
+      "step": 96000
+    },
+    {
+      "epoch": 4.067782320954348,
+      "grad_norm": 1.2011133432388306,
+      "learning_rate": 8.516485872314884e-05,
+      "loss": 3.1793,
+      "step": 96500
+    },
+    {
+      "epoch": 4.088858913290899,
+      "grad_norm": 1.2055039405822754,
+      "learning_rate": 8.507910646143292e-05,
+      "loss": 3.1798,
+      "step": 97000
+    },
+    {
+      "epoch": 4.10993550562745,
+      "grad_norm": 1.1159151792526245,
+      "learning_rate": 8.499335419971702e-05,
+      "loss": 3.1774,
+      "step": 97500
+    },
+    {
+      "epoch": 4.1310120979640015,
+      "grad_norm": 1.147810935974121,
+      "learning_rate": 8.490760193800112e-05,
+      "loss": 3.1754,
+      "step": 98000
+    },
+    {
+      "epoch": 4.152088690300552,
+      "grad_norm": 1.123781442642212,
+      "learning_rate": 8.482184967628522e-05,
+      "loss": 3.1738,
+      "step": 98500
+    },
+    {
+      "epoch": 4.173165282637103,
+      "grad_norm": 1.1476882696151733,
+      "learning_rate": 8.473609741456931e-05,
+      "loss": 3.178,
+      "step": 99000
+    },
+    {
+      "epoch": 4.194241874973654,
+      "grad_norm": 1.2432572841644287,
+      "learning_rate": 8.46503451528534e-05,
+      "loss": 3.1778,
+      "step": 99500
+    },
+    {
+      "epoch": 4.2153184673102055,
+      "grad_norm": 1.2331494092941284,
+      "learning_rate": 8.456459289113751e-05,
+      "loss": 3.1715,
+      "step": 100000
+    },
+    {
+      "epoch": 4.2153184673102055,
+      "eval_accuracy": 0.6798080752443616,
+      "eval_loss": 2.912792205810547,
+      "eval_runtime": 405.9307,
+      "eval_samples_per_second": 297.356,
+      "eval_steps_per_second": 3.099,
+      "step": 100000
+    },
+    {
+      "epoch": 4.236395059646756,
+      "grad_norm": 1.2560830116271973,
+      "learning_rate": 8.447884062942161e-05,
+      "loss": 3.1747,
+      "step": 100500
+    },
+    {
+      "epoch": 4.257471651983307,
+      "grad_norm": 1.174936056137085,
+      "learning_rate": 8.43930883677057e-05,
+      "loss": 3.1736,
+      "step": 101000
+    },
+    {
+      "epoch": 4.278548244319858,
+      "grad_norm": 1.1372392177581787,
+      "learning_rate": 8.430733610598979e-05,
+      "loss": 3.1677,
+      "step": 101500
+    },
+    {
+      "epoch": 4.29962483665641,
+      "grad_norm": 1.250074028968811,
+      "learning_rate": 8.42215838442739e-05,
+      "loss": 3.1663,
+      "step": 102000
+    },
+    {
+      "epoch": 4.32070142899296,
+      "grad_norm": 1.0547915697097778,
+      "learning_rate": 8.413583158255799e-05,
+      "loss": 3.1666,
+      "step": 102500
+    },
+    {
+      "epoch": 4.341778021329511,
+      "grad_norm": 1.1540709733963013,
+      "learning_rate": 8.405007932084209e-05,
+      "loss": 3.1633,
+      "step": 103000
+    },
+    {
+      "epoch": 4.362854613666062,
+      "grad_norm": 1.2346571683883667,
+      "learning_rate": 8.396432705912619e-05,
+      "loss": 3.1652,
+      "step": 103500
+    },
+    {
+      "epoch": 4.383931206002614,
+      "grad_norm": 1.1892918348312378,
+      "learning_rate": 8.387857479741028e-05,
+      "loss": 3.1669,
+      "step": 104000
+    },
+    {
+      "epoch": 4.405007798339165,
+      "grad_norm": 1.255914330482483,
+      "learning_rate": 8.379282253569438e-05,
+      "loss": 3.1634,
+      "step": 104500
+    },
+    {
+      "epoch": 4.426084390675715,
+      "grad_norm": 1.15084969997406,
+      "learning_rate": 8.370707027397848e-05,
+      "loss": 3.1599,
+      "step": 105000
+    },
+    {
+      "epoch": 4.4471609830122665,
+      "grad_norm": 1.141387701034546,
+      "learning_rate": 8.362131801226258e-05,
+      "loss": 3.1585,
+      "step": 105500
+    },
+    {
+      "epoch": 4.468237575348818,
+      "grad_norm": 1.0971401929855347,
+      "learning_rate": 8.353556575054668e-05,
+      "loss": 3.1552,
+      "step": 106000
+    },
+    {
+      "epoch": 4.489314167685369,
+      "grad_norm": 1.1627311706542969,
+      "learning_rate": 8.344981348883077e-05,
+      "loss": 3.158,
+      "step": 106500
+    },
+    {
+      "epoch": 4.510390760021919,
+      "grad_norm": 1.1975557804107666,
+      "learning_rate": 8.336406122711486e-05,
+      "loss": 3.1545,
+      "step": 107000
+    },
+    {
+      "epoch": 4.5314673523584705,
+      "grad_norm": 1.1922143697738647,
+      "learning_rate": 8.327830896539897e-05,
+      "loss": 3.1584,
+      "step": 107500
+    },
+    {
+      "epoch": 4.552543944695022,
+      "grad_norm": 1.2310205698013306,
+      "learning_rate": 8.319255670368307e-05,
+      "loss": 3.1544,
+      "step": 108000
+    },
+    {
+      "epoch": 4.573620537031573,
+      "grad_norm": 1.3194611072540283,
+      "learning_rate": 8.310680444196715e-05,
+      "loss": 3.1522,
+      "step": 108500
+    },
+    {
+      "epoch": 4.594697129368123,
+      "grad_norm": 1.0941317081451416,
+      "learning_rate": 8.302105218025125e-05,
+      "loss": 3.1456,
+      "step": 109000
+    },
+    {
+      "epoch": 4.615773721704675,
+      "grad_norm": 1.1610791683197021,
+      "learning_rate": 8.293529991853536e-05,
+      "loss": 3.1536,
+      "step": 109500
+    },
+    {
+      "epoch": 4.636850314041226,
+      "grad_norm": 1.1954604387283325,
+      "learning_rate": 8.284954765681945e-05,
+      "loss": 3.1509,
+      "step": 110000
+    },
+    {
+      "epoch": 4.636850314041226,
+      "eval_accuracy": 0.6833814684149944,
+      "eval_loss": 2.8884990215301514,
+      "eval_runtime": 416.2865,
+      "eval_samples_per_second": 289.959,
+      "eval_steps_per_second": 3.022,
+      "step": 110000
+    },
+    {
+      "epoch": 4.657926906377777,
+      "grad_norm": 1.1376802921295166,
+      "learning_rate": 8.276379539510355e-05,
+      "loss": 3.1497,
+      "step": 110500
+    },
+    {
+      "epoch": 4.679003498714328,
+      "grad_norm": 1.2534793615341187,
+      "learning_rate": 8.267804313338765e-05,
+      "loss": 3.1483,
+      "step": 111000
+    },
+    {
+      "epoch": 4.700080091050879,
+      "grad_norm": 1.1715461015701294,
+      "learning_rate": 8.259229087167174e-05,
+      "loss": 3.145,
+      "step": 111500
+    },
+    {
+      "epoch": 4.72115668338743,
+      "grad_norm": 1.2351901531219482,
+      "learning_rate": 8.250653860995584e-05,
+      "loss": 3.1428,
+      "step": 112000
+    },
+    {
+      "epoch": 4.742233275723981,
+      "grad_norm": 1.1293439865112305,
+      "learning_rate": 8.242078634823994e-05,
+      "loss": 3.1442,
+      "step": 112500
+    },
+    {
+      "epoch": 4.763309868060532,
+      "grad_norm": 1.1562830209732056,
+      "learning_rate": 8.233503408652403e-05,
+      "loss": 3.1431,
+      "step": 113000
+    },
+    {
+      "epoch": 4.784386460397083,
+      "grad_norm": 1.1397597789764404,
+      "learning_rate": 8.224928182480814e-05,
+      "loss": 3.1451,
+      "step": 113500
+    },
+    {
+      "epoch": 4.805463052733634,
+      "grad_norm": 1.2257388830184937,
+      "learning_rate": 8.216352956309224e-05,
+      "loss": 3.1359,
+      "step": 114000
+    },
+    {
+      "epoch": 4.826539645070185,
+      "grad_norm": 1.090147614479065,
+      "learning_rate": 8.207777730137632e-05,
+      "loss": 3.1374,
+      "step": 114500
+    },
+    {
+      "epoch": 4.847616237406736,
+      "grad_norm": 1.171140432357788,
+      "learning_rate": 8.199202503966042e-05,
+      "loss": 3.1377,
+      "step": 115000
+    },
+    {
+      "epoch": 4.868692829743287,
+      "grad_norm": 1.3769549131393433,
+      "learning_rate": 8.190627277794453e-05,
+      "loss": 3.1362,
+      "step": 115500
+    },
+    {
+      "epoch": 4.889769422079838,
+      "grad_norm": 1.086998701095581,
+      "learning_rate": 8.182052051622862e-05,
+      "loss": 3.1326,
+      "step": 116000
+    },
+    {
+      "epoch": 4.910846014416389,
+      "grad_norm": 1.294390082359314,
+      "learning_rate": 8.173476825451271e-05,
+      "loss": 3.1341,
+      "step": 116500
+    },
+    {
+      "epoch": 4.9319226067529405,
+      "grad_norm": 1.2771406173706055,
+      "learning_rate": 8.164901599279681e-05,
+      "loss": 3.1336,
+      "step": 117000
+    },
+    {
+      "epoch": 4.952999199089491,
+      "grad_norm": 1.1796345710754395,
+      "learning_rate": 8.156326373108091e-05,
+      "loss": 3.135,
+      "step": 117500
+    },
+    {
+      "epoch": 4.974075791426042,
+      "grad_norm": 1.3222382068634033,
+      "learning_rate": 8.147751146936501e-05,
+      "loss": 3.1332,
+      "step": 118000
+    },
+    {
+      "epoch": 4.995152383762593,
+      "grad_norm": 1.1622588634490967,
+      "learning_rate": 8.139175920764911e-05,
+      "loss": 3.1267,
+      "step": 118500
+    },
+    {
+      "epoch": 5.0162289760991445,
+      "grad_norm": 1.3050580024719238,
+      "learning_rate": 8.13060069459332e-05,
+      "loss": 3.1286,
+      "step": 119000
+    },
+    {
+      "epoch": 5.037305568435695,
+      "grad_norm": 1.0982577800750732,
+      "learning_rate": 8.12202546842173e-05,
+      "loss": 3.1232,
+      "step": 119500
+    },
+    {
+      "epoch": 5.058382160772246,
+      "grad_norm": 1.1449099779129028,
+      "learning_rate": 8.11345024225014e-05,
+      "loss": 3.1256,
+      "step": 120000
+    },
+    {
+      "epoch": 5.058382160772246,
+      "eval_accuracy": 0.6873757583438757,
+      "eval_loss": 2.860912561416626,
+      "eval_runtime": 384.7032,
+      "eval_samples_per_second": 313.764,
+      "eval_steps_per_second": 3.27,
+      "step": 120000
+    },
+    {
+      "epoch": 5.079458753108797,
+      "grad_norm": 1.2117756605148315,
+      "learning_rate": 8.104875016078549e-05,
+      "loss": 3.1178,
+      "step": 120500
+    },
+    {
+      "epoch": 5.100535345445349,
+      "grad_norm": 1.1897168159484863,
+      "learning_rate": 8.09629978990696e-05,
+      "loss": 3.1175,
+      "step": 121000
+    },
+    {
+      "epoch": 5.1216119377819,
+      "grad_norm": 1.1534850597381592,
+      "learning_rate": 8.08772456373537e-05,
+      "loss": 3.1217,
+      "step": 121500
+    },
+    {
+      "epoch": 5.14268853011845,
+      "grad_norm": 1.1904358863830566,
+      "learning_rate": 8.079149337563778e-05,
+      "loss": 3.1198,
+      "step": 122000
+    },
+    {
+      "epoch": 5.163765122455001,
+      "grad_norm": 1.1894731521606445,
+      "learning_rate": 8.070574111392188e-05,
+      "loss": 3.1166,
+      "step": 122500
+    },
+    {
+      "epoch": 5.184841714791553,
+      "grad_norm": 1.304951548576355,
+      "learning_rate": 8.061998885220599e-05,
+      "loss": 3.1162,
+      "step": 123000
+    },
+    {
+      "epoch": 5.205918307128104,
+      "grad_norm": 1.1531625986099243,
+      "learning_rate": 8.053423659049008e-05,
+      "loss": 3.1167,
+      "step": 123500
+    },
+    {
+      "epoch": 5.226994899464654,
+      "grad_norm": 1.3093647956848145,
+      "learning_rate": 8.044848432877418e-05,
+      "loss": 3.1163,
+      "step": 124000
+    },
+    {
+      "epoch": 5.2480714918012055,
+      "grad_norm": 1.1133514642715454,
+      "learning_rate": 8.036273206705827e-05,
+      "loss": 3.1187,
+      "step": 124500
+    },
+    {
+      "epoch": 5.269148084137757,
+      "grad_norm": 1.0890156030654907,
+      "learning_rate": 8.027697980534237e-05,
+      "loss": 3.112,
+      "step": 125000
+    },
+    {
+      "epoch": 5.290224676474308,
+      "grad_norm": 1.2314553260803223,
+      "learning_rate": 8.019122754362647e-05,
+      "loss": 3.1135,
+      "step": 125500
+    },
+    {
+      "epoch": 5.311301268810858,
+      "grad_norm": 1.145955204963684,
+      "learning_rate": 8.010547528191057e-05,
+      "loss": 3.1141,
+      "step": 126000
+    },
+    {
+      "epoch": 5.3323778611474095,
+      "grad_norm": 1.2175382375717163,
+      "learning_rate": 8.001972302019465e-05,
+      "loss": 3.1087,
+      "step": 126500
+    },
+    {
+      "epoch": 5.353454453483961,
+      "grad_norm": 1.2323992252349854,
+      "learning_rate": 7.993397075847876e-05,
+      "loss": 3.1087,
+      "step": 127000
+    },
+    {
+      "epoch": 5.374531045820512,
+      "grad_norm": 1.197698712348938,
+      "learning_rate": 7.984821849676286e-05,
+      "loss": 3.1124,
+      "step": 127500
+    },
+    {
+      "epoch": 5.395607638157063,
+      "grad_norm": 1.3759669065475464,
+      "learning_rate": 7.976246623504695e-05,
+      "loss": 3.1103,
+      "step": 128000
+    },
+    {
+      "epoch": 5.416684230493614,
+      "grad_norm": 1.2155548334121704,
+      "learning_rate": 7.967671397333105e-05,
+      "loss": 3.1041,
+      "step": 128500
+    },
+    {
+      "epoch": 5.437760822830165,
+      "grad_norm": 1.124679684638977,
+      "learning_rate": 7.959096171161516e-05,
+      "loss": 3.1042,
+      "step": 129000
+    },
+    {
+      "epoch": 5.458837415166716,
+      "grad_norm": 1.2262738943099976,
+      "learning_rate": 7.950520944989924e-05,
+      "loss": 3.1035,
+      "step": 129500
+    },
+    {
+      "epoch": 5.479914007503266,
+      "grad_norm": 1.1208912134170532,
+      "learning_rate": 7.941945718818334e-05,
+      "loss": 3.1038,
+      "step": 130000
+    },
+    {
+      "epoch": 5.479914007503266,
+      "eval_accuracy": 0.6901761739271883,
+      "eval_loss": 2.8451778888702393,
+      "eval_runtime": 386.9891,
+      "eval_samples_per_second": 311.911,
+      "eval_steps_per_second": 3.251,
+      "step": 130000
+    },
+    {
+      "epoch": 5.500990599839818,
+      "grad_norm": 1.1525845527648926,
+      "learning_rate": 7.933370492646744e-05,
+      "loss": 3.1089,
+      "step": 130500
+    },
+    {
+      "epoch": 5.522067192176369,
+      "grad_norm": 1.2495609521865845,
+      "learning_rate": 7.924795266475154e-05,
+      "loss": 3.1004,
+      "step": 131000
+    },
+    {
+      "epoch": 5.54314378451292,
+      "grad_norm": 1.144037127494812,
+      "learning_rate": 7.916220040303564e-05,
+      "loss": 3.0977,
+      "step": 131500
+    },
+    {
+      "epoch": 5.564220376849471,
+      "grad_norm": 1.2142318487167358,
+      "learning_rate": 7.907644814131972e-05,
+      "loss": 3.1016,
+      "step": 132000
+    },
+    {
+      "epoch": 5.585296969186022,
+      "grad_norm": 1.1252411603927612,
+      "learning_rate": 7.899069587960383e-05,
+      "loss": 3.0971,
+      "step": 132500
+    },
+    {
+      "epoch": 5.606373561522573,
+      "grad_norm": 1.133649468421936,
+      "learning_rate": 7.890494361788793e-05,
+      "loss": 3.098,
+      "step": 133000
+    },
+    {
+      "epoch": 5.627450153859124,
+      "grad_norm": 1.2604541778564453,
+      "learning_rate": 7.881919135617202e-05,
+      "loss": 3.0949,
+      "step": 133500
+    },
+    {
+      "epoch": 5.648526746195675,
+      "grad_norm": 1.1772538423538208,
+      "learning_rate": 7.873343909445611e-05,
+      "loss": 3.0922,
+      "step": 134000
+    },
+    {
+      "epoch": 5.669603338532226,
+      "grad_norm": 1.1320067644119263,
+      "learning_rate": 7.864768683274023e-05,
+      "loss": 3.0995,
+      "step": 134500
+    },
+    {
+      "epoch": 5.690679930868777,
+      "grad_norm": 1.4605534076690674,
+      "learning_rate": 7.856193457102431e-05,
+      "loss": 3.0981,
+      "step": 135000
+    },
+    {
+      "epoch": 5.711756523205328,
+      "grad_norm": 1.3180694580078125,
+      "learning_rate": 7.847618230930841e-05,
+      "loss": 3.0964,
+      "step": 135500
+    },
+    {
+      "epoch": 5.7328331155418795,
+      "grad_norm": 1.2434945106506348,
+      "learning_rate": 7.839043004759251e-05,
+      "loss": 3.0947,
+      "step": 136000
+    },
+    {
+      "epoch": 5.75390970787843,
+      "grad_norm": 1.1637346744537354,
+      "learning_rate": 7.83046777858766e-05,
+      "loss": 3.0936,
+      "step": 136500
+    },
+    {
+      "epoch": 5.774986300214981,
+      "grad_norm": 1.1840832233428955,
+      "learning_rate": 7.82189255241607e-05,
+      "loss": 3.0916,
+      "step": 137000
+    },
+    {
+      "epoch": 5.796062892551532,
+      "grad_norm": 1.1570559740066528,
+      "learning_rate": 7.81331732624448e-05,
+      "loss": 3.0913,
+      "step": 137500
+    },
+    {
+      "epoch": 5.8171394848880835,
+      "grad_norm": 1.229254961013794,
+      "learning_rate": 7.804742100072889e-05,
+      "loss": 3.0878,
+      "step": 138000
+    },
+    {
+      "epoch": 5.838216077224635,
+      "grad_norm": 1.177960753440857,
+      "learning_rate": 7.7961668739013e-05,
+      "loss": 3.0923,
+      "step": 138500
+    },
+    {
+      "epoch": 5.859292669561185,
+      "grad_norm": 1.1966673135757446,
+      "learning_rate": 7.78759164772971e-05,
+      "loss": 3.0893,
+      "step": 139000
+    },
+    {
+      "epoch": 5.880369261897736,
+      "grad_norm": 1.1800987720489502,
+      "learning_rate": 7.779016421558118e-05,
+      "loss": 3.0883,
+      "step": 139500
+    },
+    {
+      "epoch": 5.901445854234288,
+      "grad_norm": 1.244757056236267,
+      "learning_rate": 7.770441195386528e-05,
+      "loss": 3.0895,
+      "step": 140000
+    },
+    {
+      "epoch": 5.901445854234288,
+      "eval_accuracy": 0.6931685735004418,
+      "eval_loss": 2.8255019187927246,
+      "eval_runtime": 413.6784,
+      "eval_samples_per_second": 291.787,
+      "eval_steps_per_second": 3.041,
+      "step": 140000
+    },
+    {
+      "epoch": 5.922522446570838,
+      "grad_norm": 1.2023245096206665,
+      "learning_rate": 7.761865969214939e-05,
+      "loss": 3.0822,
+      "step": 140500
+    },
+    {
+      "epoch": 5.943599038907389,
+      "grad_norm": 1.2351374626159668,
+      "learning_rate": 7.753290743043348e-05,
+      "loss": 3.0864,
+      "step": 141000
+    },
+    {
+      "epoch": 5.96467563124394,
+      "grad_norm": 1.1610376834869385,
+      "learning_rate": 7.744715516871758e-05,
+      "loss": 3.0874,
+      "step": 141500
+    },
+    {
+      "epoch": 5.985752223580492,
+      "grad_norm": 1.2180143594741821,
+      "learning_rate": 7.736140290700167e-05,
+      "loss": 3.0831,
+      "step": 142000
+    },
+    {
+      "epoch": 6.006828815917043,
+      "grad_norm": 1.2271955013275146,
+      "learning_rate": 7.727565064528577e-05,
+      "loss": 3.0825,
+      "step": 142500
+    },
+    {
+      "epoch": 6.027905408253593,
+      "grad_norm": 1.1400116682052612,
+      "learning_rate": 7.718989838356987e-05,
+      "loss": 3.0802,
+      "step": 143000
+    },
+    {
+      "epoch": 6.0489820005901445,
+      "grad_norm": 1.140535593032837,
+      "learning_rate": 7.710414612185397e-05,
+      "loss": 3.0805,
+      "step": 143500
+    },
+    {
+      "epoch": 6.070058592926696,
+      "grad_norm": 1.1379704475402832,
+      "learning_rate": 7.701839386013807e-05,
+      "loss": 3.0781,
+      "step": 144000
+    },
+    {
+      "epoch": 6.091135185263247,
+      "grad_norm": 1.2076226472854614,
+      "learning_rate": 7.693264159842217e-05,
+      "loss": 3.0759,
+      "step": 144500
+    },
+    {
+      "epoch": 6.112211777599797,
+      "grad_norm": 1.1547982692718506,
+      "learning_rate": 7.684688933670626e-05,
+      "loss": 3.0759,
+      "step": 145000
+    },
+    {
+      "epoch": 6.1332883699363485,
+      "grad_norm": 1.1612470149993896,
+      "learning_rate": 7.676113707499035e-05,
+      "loss": 3.076,
+      "step": 145500
+    },
+    {
+      "epoch": 6.1543649622729,
+      "grad_norm": 1.1665313243865967,
+      "learning_rate": 7.667538481327446e-05,
+      "loss": 3.0757,
+      "step": 146000
+    },
+    {
+      "epoch": 6.175441554609451,
+      "grad_norm": 1.113387107849121,
+      "learning_rate": 7.658963255155856e-05,
+      "loss": 3.0772,
+      "step": 146500
+    },
+    {
+      "epoch": 6.196518146946001,
+      "grad_norm": 1.1938022375106812,
+      "learning_rate": 7.650388028984264e-05,
+      "loss": 3.0741,
+      "step": 147000
+    },
+    {
+      "epoch": 6.217594739282553,
+      "grad_norm": 1.3184177875518799,
+      "learning_rate": 7.641812802812674e-05,
+      "loss": 3.0726,
+      "step": 147500
+    },
+    {
+      "epoch": 6.238671331619104,
+      "grad_norm": 1.119075894355774,
+      "learning_rate": 7.633237576641084e-05,
+      "loss": 3.0706,
+      "step": 148000
+    },
+    {
+      "epoch": 6.259747923955655,
+      "grad_norm": 1.2483704090118408,
+      "learning_rate": 7.624662350469494e-05,
+      "loss": 3.0719,
+      "step": 148500
+    },
+    {
+      "epoch": 6.280824516292206,
+      "grad_norm": 1.2019047737121582,
+      "learning_rate": 7.616087124297904e-05,
+      "loss": 3.0705,
+      "step": 149000
+    },
+    {
+      "epoch": 6.301901108628757,
+      "grad_norm": 1.2333910465240479,
+      "learning_rate": 7.607511898126313e-05,
+      "loss": 3.0719,
+      "step": 149500
+    },
+    {
+      "epoch": 6.322977700965308,
+      "grad_norm": 1.1870752573013306,
+      "learning_rate": 7.598936671954723e-05,
+      "loss": 3.0671,
+      "step": 150000
+    },
+    {
+      "epoch": 6.322977700965308,
+      "eval_accuracy": 0.6954038707179614,
+      "eval_loss": 2.812063455581665,
+      "eval_runtime": 340.249,
+      "eval_samples_per_second": 354.758,
+      "eval_steps_per_second": 3.697,
+      "step": 150000
+    },
+    {
+      "epoch": 6.344054293301859,
+      "grad_norm": 1.1729393005371094,
+      "learning_rate": 7.590361445783133e-05,
+      "loss": 3.0694,
+      "step": 150500
+    },
+    {
+      "epoch": 6.36513088563841,
+      "grad_norm": 1.1118870973587036,
+      "learning_rate": 7.581786219611543e-05,
+      "loss": 3.0682,
+      "step": 151000
+    },
+    {
+      "epoch": 6.386207477974961,
+      "grad_norm": 1.3247240781784058,
+      "learning_rate": 7.573210993439951e-05,
+      "loss": 3.0678,
+      "step": 151500
+    },
+    {
+      "epoch": 6.407284070311512,
+      "grad_norm": 1.2429922819137573,
+      "learning_rate": 7.564635767268363e-05,
+      "loss": 3.0663,
+      "step": 152000
+    },
+    {
+      "epoch": 6.428360662648063,
+      "grad_norm": 1.1743700504302979,
+      "learning_rate": 7.556060541096772e-05,
+      "loss": 3.0636,
+      "step": 152500
+    },
+    {
+      "epoch": 6.449437254984614,
+      "grad_norm": 1.1639295816421509,
+      "learning_rate": 7.547485314925181e-05,
+      "loss": 3.0654,
+      "step": 153000
+    },
+    {
+      "epoch": 6.470513847321165,
+      "grad_norm": 1.186713695526123,
+      "learning_rate": 7.538910088753591e-05,
+      "loss": 3.0654,
+      "step": 153500
+    },
+    {
+      "epoch": 6.491590439657716,
+      "grad_norm": 1.1271131038665771,
+      "learning_rate": 7.530334862582002e-05,
+      "loss": 3.0615,
+      "step": 154000
+    },
+    {
+      "epoch": 6.512667031994267,
+      "grad_norm": 1.2945951223373413,
+      "learning_rate": 7.52175963641041e-05,
+      "loss": 3.065,
+      "step": 154500
+    },
+    {
+      "epoch": 6.5337436243308185,
+      "grad_norm": 1.1436232328414917,
+      "learning_rate": 7.51318441023882e-05,
+      "loss": 3.0658,
+      "step": 155000
+    },
+    {
+      "epoch": 6.55482021666737,
+      "grad_norm": 1.163512945175171,
+      "learning_rate": 7.50460918406723e-05,
+      "loss": 3.0632,
+      "step": 155500
+    },
+    {
+      "epoch": 6.57589680900392,
+      "grad_norm": 1.178025484085083,
+      "learning_rate": 7.49603395789564e-05,
+      "loss": 3.0626,
+      "step": 156000
+    },
+    {
+      "epoch": 6.596973401340471,
+      "grad_norm": 1.2258394956588745,
+      "learning_rate": 7.48745873172405e-05,
+      "loss": 3.0575,
+      "step": 156500
+    },
+    {
+      "epoch": 6.6180499936770225,
+      "grad_norm": 1.3446239233016968,
+      "learning_rate": 7.47888350555246e-05,
+      "loss": 3.0541,
+      "step": 157000
+    },
+    {
+      "epoch": 6.639126586013573,
+      "grad_norm": 1.212361216545105,
+      "learning_rate": 7.47030827938087e-05,
+      "loss": 3.0566,
+      "step": 157500
+    },
+    {
+      "epoch": 6.660203178350124,
+      "grad_norm": 1.1893709897994995,
+      "learning_rate": 7.461733053209279e-05,
+      "loss": 3.06,
+      "step": 158000
+    },
+    {
+      "epoch": 6.681279770686675,
+      "grad_norm": 1.174047589302063,
+      "learning_rate": 7.453157827037689e-05,
+      "loss": 3.064,
+      "step": 158500
+    },
+    {
+      "epoch": 6.702356363023227,
+      "grad_norm": 1.328777551651001,
+      "learning_rate": 7.444582600866098e-05,
+      "loss": 3.0551,
+      "step": 159000
+    },
+    {
+      "epoch": 6.723432955359778,
+      "grad_norm": 1.395482063293457,
+      "learning_rate": 7.436007374694507e-05,
+      "loss": 3.0581,
+      "step": 159500
+    },
+    {
+      "epoch": 6.744509547696328,
+      "grad_norm": 1.2295479774475098,
+      "learning_rate": 7.427432148522919e-05,
+      "loss": 3.0596,
+      "step": 160000
+    },
+    {
+      "epoch": 6.744509547696328,
+      "eval_accuracy": 0.6977985065680766,
+      "eval_loss": 2.7991831302642822,
+      "eval_runtime": 338.4692,
+      "eval_samples_per_second": 356.623,
+      "eval_steps_per_second": 3.717,
+      "step": 160000
+    },
+    {
+      "epoch": 6.765586140032879,
+      "grad_norm": 1.1636533737182617,
+      "learning_rate": 7.418856922351327e-05,
+      "loss": 3.0571,
+      "step": 160500
+    },
+    {
+      "epoch": 6.786662732369431,
+      "grad_norm": 1.2618932723999023,
+      "learning_rate": 7.410281696179737e-05,
+      "loss": 3.0536,
+      "step": 161000
+    },
+    {
+      "epoch": 6.807739324705982,
+      "grad_norm": 1.140405297279358,
+      "learning_rate": 7.401706470008147e-05,
+      "loss": 3.051,
+      "step": 161500
+    },
+    {
+      "epoch": 6.828815917042532,
+      "grad_norm": 1.2762137651443481,
+      "learning_rate": 7.393131243836557e-05,
+      "loss": 3.0518,
+      "step": 162000
+    },
+    {
+      "epoch": 6.8498925093790834,
+      "grad_norm": 1.1818777322769165,
+      "learning_rate": 7.384556017664966e-05,
+      "loss": 3.0507,
+      "step": 162500
+    },
+    {
+      "epoch": 6.870969101715635,
+      "grad_norm": 1.1954504251480103,
+      "learning_rate": 7.375980791493376e-05,
+      "loss": 3.0476,
+      "step": 163000
+    },
+    {
+      "epoch": 6.892045694052186,
+      "grad_norm": 1.32598876953125,
+      "learning_rate": 7.367405565321786e-05,
+      "loss": 3.0479,
+      "step": 163500
+    },
+    {
+      "epoch": 6.913122286388736,
+      "grad_norm": 1.1997956037521362,
+      "learning_rate": 7.358830339150196e-05,
+      "loss": 3.0499,
+      "step": 164000
+    },
+    {
+      "epoch": 6.9341988787252875,
+      "grad_norm": 1.1368622779846191,
+      "learning_rate": 7.350255112978606e-05,
+      "loss": 3.0529,
+      "step": 164500
+    },
+    {
+      "epoch": 6.955275471061839,
+      "grad_norm": 1.20081627368927,
+      "learning_rate": 7.341679886807014e-05,
+      "loss": 3.0497,
+      "step": 165000
+    },
+    {
+      "epoch": 6.97635206339839,
+      "grad_norm": 1.2233986854553223,
+      "learning_rate": 7.333104660635425e-05,
+      "loss": 3.0475,
+      "step": 165500
+    },
+    {
+      "epoch": 6.997428655734941,
+      "grad_norm": 1.167374610900879,
+      "learning_rate": 7.324529434463835e-05,
+      "loss": 3.0471,
+      "step": 166000
+    },
+    {
+      "epoch": 7.018505248071492,
+      "grad_norm": 1.3142369985580444,
+      "learning_rate": 7.315954208292244e-05,
+      "loss": 3.0474,
+      "step": 166500
+    },
+    {
+      "epoch": 7.039581840408043,
+      "grad_norm": 1.166033387184143,
+      "learning_rate": 7.307378982120653e-05,
+      "loss": 3.0405,
+      "step": 167000
+    },
+    {
+      "epoch": 7.060658432744594,
+      "grad_norm": 1.2062134742736816,
+      "learning_rate": 7.298803755949063e-05,
+      "loss": 3.042,
+      "step": 167500
+    },
+    {
+      "epoch": 7.081735025081145,
+      "grad_norm": 1.188899040222168,
+      "learning_rate": 7.290228529777473e-05,
+      "loss": 3.0407,
+      "step": 168000
+    },
+    {
+      "epoch": 7.102811617417696,
+      "grad_norm": 1.170951008796692,
+      "learning_rate": 7.281653303605883e-05,
+      "loss": 3.0425,
+      "step": 168500
+    },
+    {
+      "epoch": 7.123888209754247,
+      "grad_norm": 1.3054718971252441,
+      "learning_rate": 7.273078077434293e-05,
+      "loss": 3.0457,
+      "step": 169000
+    },
+    {
+      "epoch": 7.144964802090798,
+      "grad_norm": 1.152773141860962,
+      "learning_rate": 7.264502851262703e-05,
+      "loss": 3.0382,
+      "step": 169500
+    },
+    {
+      "epoch": 7.166041394427349,
+      "grad_norm": 1.2419204711914062,
+      "learning_rate": 7.255927625091112e-05,
+      "loss": 3.0371,
+      "step": 170000
+    },
+    {
+      "epoch": 7.166041394427349,
+      "eval_accuracy": 0.7001752439439229,
+      "eval_loss": 2.785972833633423,
+      "eval_runtime": 413.3181,
+      "eval_samples_per_second": 292.041,
+      "eval_steps_per_second": 3.044,
+      "step": 170000
+    },
+    {
+      "epoch": 7.1871179867639,
+      "grad_norm": 1.223521113395691,
+      "learning_rate": 7.247352398919521e-05,
+      "loss": 3.0413,
+      "step": 170500
+    },
+    {
+      "epoch": 7.208194579100451,
+      "grad_norm": 1.189825177192688,
+      "learning_rate": 7.238777172747931e-05,
+      "loss": 3.0377,
+      "step": 171000
+    },
+    {
+      "epoch": 7.229271171437002,
+      "grad_norm": 1.150272250175476,
+      "learning_rate": 7.230201946576342e-05,
+      "loss": 3.0411,
+      "step": 171500
+    },
+    {
+      "epoch": 7.250347763773553,
+      "grad_norm": 1.1277680397033691,
+      "learning_rate": 7.22162672040475e-05,
+      "loss": 3.0393,
+      "step": 172000
+    },
+    {
+      "epoch": 7.271424356110104,
+      "grad_norm": 1.2561289072036743,
+      "learning_rate": 7.21305149423316e-05,
+      "loss": 3.0362,
+      "step": 172500
+    },
+    {
+      "epoch": 7.292500948446655,
+      "grad_norm": 1.1211469173431396,
+      "learning_rate": 7.20447626806157e-05,
+      "loss": 3.0307,
+      "step": 173000
+    },
+    {
+      "epoch": 7.313577540783206,
+      "grad_norm": 1.2853935956954956,
+      "learning_rate": 7.19590104188998e-05,
+      "loss": 3.034,
+      "step": 173500
+    },
+    {
+      "epoch": 7.3346541331197574,
+      "grad_norm": 1.1723307371139526,
+      "learning_rate": 7.18732581571839e-05,
+      "loss": 3.0349,
+      "step": 174000
+    },
+    {
+      "epoch": 7.355730725456308,
+      "grad_norm": 1.2071915864944458,
+      "learning_rate": 7.1787505895468e-05,
+      "loss": 3.0339,
+      "step": 174500
+    },
+    {
+      "epoch": 7.376807317792859,
+      "grad_norm": 1.140781283378601,
+      "learning_rate": 7.17017536337521e-05,
+      "loss": 3.034,
+      "step": 175000
+    },
+    {
+      "epoch": 7.39788391012941,
+      "grad_norm": 1.162210464477539,
+      "learning_rate": 7.161600137203619e-05,
+      "loss": 3.0365,
+      "step": 175500
+    },
+    {
+      "epoch": 7.4189605024659615,
+      "grad_norm": 1.1570240259170532,
+      "learning_rate": 7.153024911032029e-05,
+      "loss": 3.0309,
+      "step": 176000
+    },
+    {
+      "epoch": 7.440037094802513,
+      "grad_norm": 1.2049775123596191,
+      "learning_rate": 7.144449684860438e-05,
+      "loss": 3.0329,
+      "step": 176500
+    },
+    {
+      "epoch": 7.461113687139063,
+      "grad_norm": 1.2218396663665771,
+      "learning_rate": 7.135874458688849e-05,
+      "loss": 3.0355,
+      "step": 177000
+    },
+    {
+      "epoch": 7.482190279475614,
+      "grad_norm": 1.222551703453064,
+      "learning_rate": 7.127299232517259e-05,
+      "loss": 3.0356,
+      "step": 177500
+    },
+    {
+      "epoch": 7.503266871812166,
+      "grad_norm": 1.1948316097259521,
+      "learning_rate": 7.118724006345667e-05,
+      "loss": 3.032,
+      "step": 178000
+    },
+    {
+      "epoch": 7.524343464148717,
+      "grad_norm": 1.2324901819229126,
+      "learning_rate": 7.110148780174077e-05,
+      "loss": 3.0276,
+      "step": 178500
+    },
+    {
+      "epoch": 7.545420056485267,
+      "grad_norm": 1.166361927986145,
+      "learning_rate": 7.101573554002488e-05,
+      "loss": 3.0285,
+      "step": 179000
+    },
+    {
+      "epoch": 7.566496648821818,
+      "grad_norm": 1.2692800760269165,
+      "learning_rate": 7.092998327830897e-05,
+      "loss": 3.0318,
+      "step": 179500
+    },
+    {
+      "epoch": 7.58757324115837,
+      "grad_norm": 1.163097620010376,
+      "learning_rate": 7.084423101659306e-05,
+      "loss": 3.0289,
+      "step": 180000
+    },
+    {
+      "epoch": 7.58757324115837,
+      "eval_accuracy": 0.7014439906811554,
+      "eval_loss": 2.7772982120513916,
+      "eval_runtime": 278.7662,
+      "eval_samples_per_second": 433.001,
+      "eval_steps_per_second": 4.513,
+      "step": 180000
+    },
+    {
+      "epoch": 7.608649833494921,
+      "grad_norm": 1.158246397972107,
+      "learning_rate": 7.075847875487716e-05,
+      "loss": 3.0268,
+      "step": 180500
+    },
+    {
+      "epoch": 7.629726425831471,
+      "grad_norm": 1.1630358695983887,
+      "learning_rate": 7.067272649316126e-05,
+      "loss": 3.0297,
+      "step": 181000
+    },
+    {
+      "epoch": 7.650803018168022,
+      "grad_norm": 1.2528800964355469,
+      "learning_rate": 7.058697423144536e-05,
+      "loss": 3.0235,
+      "step": 181500
+    },
+    {
+      "epoch": 7.671879610504574,
+      "grad_norm": 1.1320044994354248,
+      "learning_rate": 7.050122196972946e-05,
+      "loss": 3.0236,
+      "step": 182000
+    },
+    {
+      "epoch": 7.692956202841125,
+      "grad_norm": 1.16665780544281,
+      "learning_rate": 7.041546970801354e-05,
+      "loss": 3.0269,
+      "step": 182500
+    },
+    {
+      "epoch": 7.714032795177676,
+      "grad_norm": 1.2551246881484985,
+      "learning_rate": 7.032971744629765e-05,
+      "loss": 3.0282,
+      "step": 183000
+    },
+    {
+      "epoch": 7.7351093875142265,
+      "grad_norm": 1.2030282020568848,
+      "learning_rate": 7.024396518458175e-05,
+      "loss": 3.0289,
+      "step": 183500
+    },
+    {
+      "epoch": 7.756185979850778,
+      "grad_norm": 1.1422027349472046,
+      "learning_rate": 7.015821292286584e-05,
+      "loss": 3.0258,
+      "step": 184000
+    },
+    {
+      "epoch": 7.777262572187329,
+      "grad_norm": 1.173284888267517,
+      "learning_rate": 7.007246066114993e-05,
+      "loss": 3.0239,
+      "step": 184500
+    },
+    {
+      "epoch": 7.798339164523879,
+      "grad_norm": 1.3611693382263184,
+      "learning_rate": 6.998670839943405e-05,
+      "loss": 3.0246,
+      "step": 185000
+    },
+    {
+      "epoch": 7.8194157568604306,
+      "grad_norm": 1.1628236770629883,
+      "learning_rate": 6.990095613771813e-05,
+      "loss": 3.0211,
+      "step": 185500
+    },
+    {
+      "epoch": 7.840492349196982,
+      "grad_norm": 1.1959459781646729,
+      "learning_rate": 6.981520387600223e-05,
+      "loss": 3.0221,
+      "step": 186000
+    },
+    {
+      "epoch": 7.861568941533533,
+      "grad_norm": 1.1748566627502441,
+      "learning_rate": 6.972945161428633e-05,
+      "loss": 3.0229,
+      "step": 186500
+    },
+    {
+      "epoch": 7.882645533870084,
+      "grad_norm": 1.1584398746490479,
+      "learning_rate": 6.964369935257043e-05,
+      "loss": 3.0221,
+      "step": 187000
+    },
+    {
+      "epoch": 7.903722126206635,
+      "grad_norm": 1.1963914632797241,
+      "learning_rate": 6.955794709085452e-05,
+      "loss": 3.0226,
+      "step": 187500
+    },
+    {
+      "epoch": 7.924798718543186,
+      "grad_norm": 1.2454791069030762,
+      "learning_rate": 6.947219482913862e-05,
+      "loss": 3.0118,
+      "step": 188000
+    },
+    {
+      "epoch": 7.945875310879737,
+      "grad_norm": 1.2452200651168823,
+      "learning_rate": 6.938644256742272e-05,
+      "loss": 3.0226,
+      "step": 188500
+    },
+    {
+      "epoch": 7.966951903216288,
+      "grad_norm": 1.1953330039978027,
+      "learning_rate": 6.930069030570682e-05,
+      "loss": 3.0184,
+      "step": 189000
+    },
+    {
+      "epoch": 7.988028495552839,
+      "grad_norm": 1.1281824111938477,
+      "learning_rate": 6.921493804399092e-05,
+      "loss": 3.0159,
+      "step": 189500
+    },
+    {
+      "epoch": 8.00910508788939,
+      "grad_norm": 1.214613437652588,
+      "learning_rate": 6.9129185782275e-05,
+      "loss": 3.0178,
+      "step": 190000
+    },
+    {
+      "epoch": 8.00910508788939,
+      "eval_accuracy": 0.7028675444157196,
+      "eval_loss": 2.766869068145752,
+      "eval_runtime": 269.7943,
+      "eval_samples_per_second": 447.4,
+      "eval_steps_per_second": 4.663,
+      "step": 190000
+    },
+    {
+      "epoch": 8.03018168022594,
+      "grad_norm": 1.1709343194961548,
+      "learning_rate": 6.904343352055911e-05,
+      "loss": 3.0122,
+      "step": 190500
+    },
+    {
+      "epoch": 8.051258272562492,
+      "grad_norm": 1.222080111503601,
+      "learning_rate": 6.895768125884321e-05,
+      "loss": 3.0155,
+      "step": 191000
+    },
+    {
+      "epoch": 8.072334864899043,
+      "grad_norm": 1.350806474685669,
+      "learning_rate": 6.88719289971273e-05,
+      "loss": 3.0147,
+      "step": 191500
+    },
+    {
+      "epoch": 8.093411457235595,
+      "grad_norm": 1.1972124576568604,
+      "learning_rate": 6.87861767354114e-05,
+      "loss": 3.0087,
+      "step": 192000
+    },
+    {
+      "epoch": 8.114488049572145,
+      "grad_norm": 1.2101161479949951,
+      "learning_rate": 6.870042447369551e-05,
+      "loss": 3.0152,
+      "step": 192500
+    },
+    {
+      "epoch": 8.135564641908696,
+      "grad_norm": 1.1549127101898193,
+      "learning_rate": 6.861467221197959e-05,
+      "loss": 3.0145,
+      "step": 193000
+    },
+    {
+      "epoch": 8.156641234245248,
+      "grad_norm": 1.3437097072601318,
+      "learning_rate": 6.852891995026369e-05,
+      "loss": 3.0147,
+      "step": 193500
+    },
+    {
+      "epoch": 8.177717826581798,
+      "grad_norm": 1.2040529251098633,
+      "learning_rate": 6.844316768854779e-05,
+      "loss": 3.0141,
+      "step": 194000
+    },
+    {
+      "epoch": 8.19879441891835,
+      "grad_norm": 1.2064359188079834,
+      "learning_rate": 6.835741542683189e-05,
+      "loss": 3.0121,
+      "step": 194500
+    },
+    {
+      "epoch": 8.2198710112549,
+      "grad_norm": 1.1979795694351196,
+      "learning_rate": 6.827166316511599e-05,
+      "loss": 3.0083,
+      "step": 195000
+    },
+    {
+      "epoch": 8.24094760359145,
+      "grad_norm": 1.262338638305664,
+      "learning_rate": 6.818591090340008e-05,
+      "loss": 3.012,
+      "step": 195500
+    },
+    {
+      "epoch": 8.262024195928003,
+      "grad_norm": 1.2956135272979736,
+      "learning_rate": 6.810015864168417e-05,
+      "loss": 3.0072,
+      "step": 196000
+    },
+    {
+      "epoch": 8.283100788264553,
+      "grad_norm": 1.2011007070541382,
+      "learning_rate": 6.801440637996828e-05,
+      "loss": 3.0079,
+      "step": 196500
+    },
+    {
+      "epoch": 8.304177380601104,
+      "grad_norm": 1.2376735210418701,
+      "learning_rate": 6.792865411825238e-05,
+      "loss": 3.0111,
+      "step": 197000
+    },
+    {
+      "epoch": 8.325253972937656,
+      "grad_norm": 1.20741868019104,
+      "learning_rate": 6.784290185653646e-05,
+      "loss": 3.0081,
+      "step": 197500
+    },
+    {
+      "epoch": 8.346330565274206,
+      "grad_norm": 1.4059467315673828,
+      "learning_rate": 6.775714959482056e-05,
+      "loss": 3.0068,
+      "step": 198000
+    },
+    {
+      "epoch": 8.367407157610758,
+      "grad_norm": 1.1471384763717651,
+      "learning_rate": 6.767139733310467e-05,
+      "loss": 3.0086,
+      "step": 198500
+    },
+    {
+      "epoch": 8.388483749947309,
+      "grad_norm": 1.223591923713684,
+      "learning_rate": 6.758564507138876e-05,
+      "loss": 3.0106,
+      "step": 199000
+    },
+    {
+      "epoch": 8.409560342283859,
+      "grad_norm": 1.151410460472107,
+      "learning_rate": 6.749989280967286e-05,
+      "loss": 3.0107,
+      "step": 199500
+    },
+    {
+      "epoch": 8.430636934620411,
+      "grad_norm": 1.3132694959640503,
+      "learning_rate": 6.741414054795696e-05,
+      "loss": 3.0064,
+      "step": 200000
+    },
+    {
+      "epoch": 8.430636934620411,
+      "eval_accuracy": 0.705299483600653,
+      "eval_loss": 2.7545182704925537,
+      "eval_runtime": 294.7469,
+      "eval_samples_per_second": 409.524,
+      "eval_steps_per_second": 4.268,
+      "step": 200000
+    },
+    {
+      "epoch": 8.451713526956961,
+      "grad_norm": 1.3303207159042358,
+      "learning_rate": 6.732838828624105e-05,
+      "loss": 3.0061,
+      "step": 200500
+    },
+    {
+      "epoch": 8.472790119293512,
+      "grad_norm": 1.229698657989502,
+      "learning_rate": 6.724263602452515e-05,
+      "loss": 3.0071,
+      "step": 201000
+    },
+    {
+      "epoch": 8.493866711630064,
+      "grad_norm": 1.2676060199737549,
+      "learning_rate": 6.715688376280924e-05,
+      "loss": 3.0053,
+      "step": 201500
+    },
+    {
+      "epoch": 8.514943303966614,
+      "grad_norm": 1.2171728610992432,
+      "learning_rate": 6.707113150109335e-05,
+      "loss": 3.0027,
+      "step": 202000
+    },
+    {
+      "epoch": 8.536019896303166,
+      "grad_norm": 1.2572832107543945,
+      "learning_rate": 6.698537923937745e-05,
+      "loss": 3.0076,
+      "step": 202500
+    },
+    {
+      "epoch": 8.557096488639717,
+      "grad_norm": 1.1887730360031128,
+      "learning_rate": 6.689962697766153e-05,
+      "loss": 3.004,
+      "step": 203000
+    },
+    {
+      "epoch": 8.578173080976267,
+      "grad_norm": 1.2598347663879395,
+      "learning_rate": 6.681387471594563e-05,
+      "loss": 3.0031,
+      "step": 203500
+    },
+    {
+      "epoch": 8.59924967331282,
+      "grad_norm": 1.2408080101013184,
+      "learning_rate": 6.672812245422974e-05,
+      "loss": 3.0022,
+      "step": 204000
+    },
+    {
+      "epoch": 8.62032626564937,
+      "grad_norm": 1.2270398139953613,
+      "learning_rate": 6.664237019251383e-05,
+      "loss": 3.0047,
+      "step": 204500
+    },
+    {
+      "epoch": 8.64140285798592,
+      "grad_norm": 1.2013546228408813,
+      "learning_rate": 6.655661793079792e-05,
+      "loss": 3.0023,
+      "step": 205000
+    },
+    {
+      "epoch": 8.662479450322472,
+      "grad_norm": 1.224739670753479,
+      "learning_rate": 6.647086566908202e-05,
+      "loss": 2.998,
+      "step": 205500
+    },
+    {
+      "epoch": 8.683556042659022,
+      "grad_norm": 1.1850736141204834,
+      "learning_rate": 6.638511340736612e-05,
+      "loss": 2.9988,
+      "step": 206000
+    },
+    {
+      "epoch": 8.704632634995574,
+      "grad_norm": 1.216667652130127,
+      "learning_rate": 6.629936114565022e-05,
+      "loss": 2.9967,
+      "step": 206500
+    },
+    {
+      "epoch": 8.725709227332125,
+      "grad_norm": 1.2335216999053955,
+      "learning_rate": 6.621360888393432e-05,
+      "loss": 3.0008,
+      "step": 207000
+    },
+    {
+      "epoch": 8.746785819668675,
+      "grad_norm": 1.2645560503005981,
+      "learning_rate": 6.61278566222184e-05,
+      "loss": 2.9924,
+      "step": 207500
+    },
+    {
+      "epoch": 8.767862412005227,
+      "grad_norm": 1.2059534788131714,
+      "learning_rate": 6.604210436050251e-05,
+      "loss": 2.9987,
+      "step": 208000
+    },
+    {
+      "epoch": 8.788939004341778,
+      "grad_norm": 1.2235528230667114,
+      "learning_rate": 6.595635209878661e-05,
+      "loss": 2.9986,
+      "step": 208500
+    },
+    {
+      "epoch": 8.81001559667833,
+      "grad_norm": 1.2034275531768799,
+      "learning_rate": 6.58705998370707e-05,
+      "loss": 2.9999,
+      "step": 209000
+    },
+    {
+      "epoch": 8.83109218901488,
+      "grad_norm": 1.2129944562911987,
+      "learning_rate": 6.57848475753548e-05,
+      "loss": 2.9987,
+      "step": 209500
+    },
+    {
+      "epoch": 8.85216878135143,
+      "grad_norm": 1.3088895082473755,
+      "learning_rate": 6.569909531363891e-05,
+      "loss": 2.9931,
+      "step": 210000
+    },
+    {
+      "epoch": 8.85216878135143,
+      "eval_accuracy": 0.7063227960533356,
+      "eval_loss": 2.746647834777832,
+      "eval_runtime": 269.8529,
+      "eval_samples_per_second": 447.303,
+      "eval_steps_per_second": 4.662,
+      "step": 210000
+    },
+    {
+      "epoch": 8.873245373687983,
+      "grad_norm": 1.2259745597839355,
+      "learning_rate": 6.561334305192299e-05,
+      "loss": 2.9972,
+      "step": 210500
+    },
+    {
+      "epoch": 8.894321966024533,
+      "grad_norm": 1.2715039253234863,
+      "learning_rate": 6.552759079020709e-05,
+      "loss": 2.9976,
+      "step": 211000
+    },
+    {
+      "epoch": 8.915398558361083,
+      "grad_norm": 1.3087495565414429,
+      "learning_rate": 6.544183852849119e-05,
+      "loss": 2.9934,
+      "step": 211500
+    },
+    {
+      "epoch": 8.936475150697635,
+      "grad_norm": 1.2541286945343018,
+      "learning_rate": 6.535608626677529e-05,
+      "loss": 3.0003,
+      "step": 212000
+    },
+    {
+      "epoch": 8.957551743034186,
+      "grad_norm": 1.2257838249206543,
+      "learning_rate": 6.527033400505939e-05,
+      "loss": 2.9992,
+      "step": 212500
+    },
+    {
+      "epoch": 8.978628335370738,
+      "grad_norm": 1.2906502485275269,
+      "learning_rate": 6.518458174334348e-05,
+      "loss": 2.9992,
+      "step": 213000
+    },
+    {
+      "epoch": 8.999704927707288,
+      "grad_norm": 1.200895071029663,
+      "learning_rate": 6.509882948162758e-05,
+      "loss": 2.997,
+      "step": 213500
+    },
+    {
+      "epoch": 9.020781520043839,
+      "grad_norm": 1.2491098642349243,
+      "learning_rate": 6.501307721991168e-05,
+      "loss": 2.9921,
+      "step": 214000
+    },
+    {
+      "epoch": 9.04185811238039,
+      "grad_norm": 1.3455792665481567,
+      "learning_rate": 6.492732495819578e-05,
+      "loss": 2.9897,
+      "step": 214500
+    },
+    {
+      "epoch": 9.062934704716941,
+      "grad_norm": 1.2383557558059692,
+      "learning_rate": 6.484157269647986e-05,
+      "loss": 2.9918,
+      "step": 215000
+    },
+    {
+      "epoch": 9.084011297053493,
+      "grad_norm": 1.1951334476470947,
+      "learning_rate": 6.475582043476398e-05,
+      "loss": 2.9916,
+      "step": 215500
+    },
+    {
+      "epoch": 9.105087889390044,
+      "grad_norm": 1.1649657487869263,
+      "learning_rate": 6.467006817304807e-05,
+      "loss": 2.9891,
+      "step": 216000
+    },
+    {
+      "epoch": 9.126164481726594,
+      "grad_norm": 1.196176528930664,
+      "learning_rate": 6.458431591133216e-05,
+      "loss": 2.9852,
+      "step": 216500
+    },
+    {
+      "epoch": 9.147241074063146,
+      "grad_norm": 1.220537543296814,
+      "learning_rate": 6.449856364961626e-05,
+      "loss": 2.9886,
+      "step": 217000
+    },
+    {
+      "epoch": 9.168317666399696,
+      "grad_norm": 1.2328126430511475,
+      "learning_rate": 6.441281138790037e-05,
+      "loss": 2.9866,
+      "step": 217500
+    },
+    {
+      "epoch": 9.189394258736247,
+      "grad_norm": 1.1723554134368896,
+      "learning_rate": 6.432705912618445e-05,
+      "loss": 2.9877,
+      "step": 218000
+    },
+    {
+      "epoch": 9.210470851072799,
+      "grad_norm": 1.2665780782699585,
+      "learning_rate": 6.424130686446855e-05,
+      "loss": 2.9893,
+      "step": 218500
+    },
+    {
+      "epoch": 9.23154744340935,
+      "grad_norm": 1.245151400566101,
+      "learning_rate": 6.415555460275265e-05,
+      "loss": 2.9869,
+      "step": 219000
+    },
+    {
+      "epoch": 9.252624035745901,
+      "grad_norm": 1.210968255996704,
+      "learning_rate": 6.406980234103675e-05,
+      "loss": 2.9889,
+      "step": 219500
+    },
+    {
+      "epoch": 9.273700628082452,
+      "grad_norm": 1.2229198217391968,
+      "learning_rate": 6.398405007932085e-05,
+      "loss": 2.9905,
+      "step": 220000
+    },
+    {
+      "epoch": 9.273700628082452,
+      "eval_accuracy": 0.7076335006525117,
+      "eval_loss": 2.7371668815612793,
+      "eval_runtime": 285.1539,
+      "eval_samples_per_second": 423.301,
+      "eval_steps_per_second": 4.412,
+      "step": 220000
+    },
+    {
+      "epoch": 9.294777220419002,
+      "grad_norm": 1.198967695236206,
+      "learning_rate": 6.389829781760495e-05,
+      "loss": 2.985,
+      "step": 220500
+    },
+    {
+      "epoch": 9.315853812755554,
+      "grad_norm": 1.3251479864120483,
+      "learning_rate": 6.381254555588903e-05,
+      "loss": 2.9892,
+      "step": 221000
+    },
+    {
+      "epoch": 9.336930405092104,
+      "grad_norm": 1.1828927993774414,
+      "learning_rate": 6.372679329417314e-05,
+      "loss": 2.9837,
+      "step": 221500
+    },
+    {
+      "epoch": 9.358006997428657,
+      "grad_norm": 1.248824119567871,
+      "learning_rate": 6.364104103245724e-05,
+      "loss": 2.9831,
+      "step": 222000
+    },
+    {
+      "epoch": 9.379083589765207,
+      "grad_norm": 1.3417350053787231,
+      "learning_rate": 6.355528877074132e-05,
+      "loss": 2.9867,
+      "step": 222500
+    },
+    {
+      "epoch": 9.400160182101757,
+      "grad_norm": 1.2479565143585205,
+      "learning_rate": 6.346953650902542e-05,
+      "loss": 2.9838,
+      "step": 223000
+    },
+    {
+      "epoch": 9.42123677443831,
+      "grad_norm": 1.2634061574935913,
+      "learning_rate": 6.338378424730954e-05,
+      "loss": 2.9848,
+      "step": 223500
+    },
+    {
+      "epoch": 9.44231336677486,
+      "grad_norm": 1.3408063650131226,
+      "learning_rate": 6.329803198559362e-05,
+      "loss": 2.9825,
+      "step": 224000
+    },
+    {
+      "epoch": 9.46338995911141,
+      "grad_norm": 1.264139175415039,
+      "learning_rate": 6.321227972387772e-05,
+      "loss": 2.9838,
+      "step": 224500
+    },
+    {
+      "epoch": 9.484466551447962,
+      "grad_norm": 1.2032155990600586,
+      "learning_rate": 6.312652746216182e-05,
+      "loss": 2.9792,
+      "step": 225000
+    },
+    {
+      "epoch": 9.505543143784513,
+      "grad_norm": 1.1994582414627075,
+      "learning_rate": 6.304077520044591e-05,
+      "loss": 2.9872,
+      "step": 225500
+    },
+    {
+      "epoch": 9.526619736121065,
+      "grad_norm": 1.229493498802185,
+      "learning_rate": 6.295502293873001e-05,
+      "loss": 2.9808,
+      "step": 226000
+    },
+    {
+      "epoch": 9.547696328457615,
+      "grad_norm": 1.448414921760559,
+      "learning_rate": 6.286927067701411e-05,
+      "loss": 2.9792,
+      "step": 226500
+    },
+    {
+      "epoch": 9.568772920794165,
+      "grad_norm": 1.381797194480896,
+      "learning_rate": 6.278351841529821e-05,
+      "loss": 2.9836,
+      "step": 227000
+    },
+    {
+      "epoch": 9.589849513130718,
+      "grad_norm": 1.2295881509780884,
+      "learning_rate": 6.269776615358231e-05,
+      "loss": 2.9802,
+      "step": 227500
+    },
+    {
+      "epoch": 9.610926105467268,
+      "grad_norm": 1.258610486984253,
+      "learning_rate": 6.26120138918664e-05,
+      "loss": 2.9797,
+      "step": 228000
+    },
+    {
+      "epoch": 9.63200269780382,
+      "grad_norm": 1.1931253671646118,
+      "learning_rate": 6.252626163015049e-05,
+      "loss": 2.978,
+      "step": 228500
+    },
+    {
+      "epoch": 9.65307929014037,
+      "grad_norm": 1.4504826068878174,
+      "learning_rate": 6.24405093684346e-05,
+      "loss": 2.9777,
+      "step": 229000
+    },
+    {
+      "epoch": 9.67415588247692,
+      "grad_norm": 1.2486432790756226,
+      "learning_rate": 6.23547571067187e-05,
+      "loss": 2.9771,
+      "step": 229500
+    },
+    {
+      "epoch": 9.695232474813473,
+      "grad_norm": 1.3364046812057495,
+      "learning_rate": 6.226900484500279e-05,
+      "loss": 2.9751,
+      "step": 230000
+    },
+    {
+      "epoch": 9.695232474813473,
+      "eval_accuracy": 0.7091420022889984,
+      "eval_loss": 2.7285969257354736,
+      "eval_runtime": 296.0491,
+      "eval_samples_per_second": 407.723,
+      "eval_steps_per_second": 4.249,
+      "step": 230000
+    },
+    {
+      "epoch": 9.716309067150023,
+      "grad_norm": 1.3006479740142822,
+      "learning_rate": 6.218325258328688e-05,
+      "loss": 2.9821,
+      "step": 230500
+    },
+    {
+      "epoch": 9.737385659486574,
+      "grad_norm": 1.1916451454162598,
+      "learning_rate": 6.209750032157098e-05,
+      "loss": 2.9787,
+      "step": 231000
+    },
+    {
+      "epoch": 9.758462251823126,
+      "grad_norm": 1.2389692068099976,
+      "learning_rate": 6.201174805985508e-05,
+      "loss": 2.976,
+      "step": 231500
+    },
+    {
+      "epoch": 9.779538844159676,
+      "grad_norm": 1.265692114830017,
+      "learning_rate": 6.192599579813918e-05,
+      "loss": 2.9771,
+      "step": 232000
+    },
+    {
+      "epoch": 9.800615436496226,
+      "grad_norm": 1.1969783306121826,
+      "learning_rate": 6.184024353642328e-05,
+      "loss": 2.9727,
+      "step": 232500
+    },
+    {
+      "epoch": 9.821692028832778,
+      "grad_norm": 1.2388644218444824,
+      "learning_rate": 6.175449127470738e-05,
+      "loss": 2.9726,
+      "step": 233000
+    },
+    {
+      "epoch": 9.842768621169329,
+      "grad_norm": 1.231201410293579,
+      "learning_rate": 6.166873901299147e-05,
+      "loss": 2.9791,
+      "step": 233500
+    },
+    {
+      "epoch": 9.863845213505881,
+      "grad_norm": 1.241115927696228,
+      "learning_rate": 6.158298675127557e-05,
+      "loss": 2.9728,
+      "step": 234000
+    },
+    {
+      "epoch": 9.884921805842431,
+      "grad_norm": 1.2984199523925781,
+      "learning_rate": 6.149723448955966e-05,
+      "loss": 2.9765,
+      "step": 234500
+    },
+    {
+      "epoch": 9.905998398178982,
+      "grad_norm": 1.2083555459976196,
+      "learning_rate": 6.141148222784377e-05,
+      "loss": 2.9707,
+      "step": 235000
+    },
+    {
+      "epoch": 9.927074990515534,
+      "grad_norm": 1.2415496110916138,
+      "learning_rate": 6.132572996612787e-05,
+      "loss": 2.9714,
+      "step": 235500
+    },
+    {
+      "epoch": 9.948151582852084,
+      "grad_norm": 1.2677634954452515,
+      "learning_rate": 6.123997770441195e-05,
+      "loss": 2.976,
+      "step": 236000
+    },
+    {
+      "epoch": 9.969228175188636,
+      "grad_norm": 1.2245522737503052,
+      "learning_rate": 6.115422544269605e-05,
+      "loss": 2.9774,
+      "step": 236500
+    },
+    {
+      "epoch": 9.990304767525187,
+      "grad_norm": 1.2706023454666138,
+      "learning_rate": 6.106847318098015e-05,
+      "loss": 2.9724,
+      "step": 237000
+    },
+    {
+      "epoch": 10.011381359861737,
+      "grad_norm": 1.1838876008987427,
+      "learning_rate": 6.098272091926425e-05,
+      "loss": 2.9728,
+      "step": 237500
+    },
+    {
+      "epoch": 10.032457952198289,
+      "grad_norm": 1.2138760089874268,
+      "learning_rate": 6.0896968657548345e-05,
+      "loss": 2.9715,
+      "step": 238000
+    },
+    {
+      "epoch": 10.05353454453484,
+      "grad_norm": 1.2474576234817505,
+      "learning_rate": 6.081121639583244e-05,
+      "loss": 2.9682,
+      "step": 238500
+    },
+    {
+      "epoch": 10.07461113687139,
+      "grad_norm": 1.203450083732605,
+      "learning_rate": 6.072546413411654e-05,
+      "loss": 2.9688,
+      "step": 239000
+    },
+    {
+      "epoch": 10.095687729207942,
+      "grad_norm": 1.3365384340286255,
+      "learning_rate": 6.063971187240064e-05,
+      "loss": 2.9686,
+      "step": 239500
+    },
+    {
+      "epoch": 10.116764321544492,
+      "grad_norm": 1.2659412622451782,
+      "learning_rate": 6.055395961068473e-05,
+      "loss": 2.9645,
+      "step": 240000
+    },
+    {
+      "epoch": 10.116764321544492,
+      "eval_accuracy": 0.7107587080647508,
+      "eval_loss": 2.7234268188476562,
+      "eval_runtime": 381.6263,
+      "eval_samples_per_second": 316.294,
+      "eval_steps_per_second": 3.296,
+      "step": 240000
+    },
+    {
+      "epoch": 10.137840913881044,
+      "grad_norm": 1.2959920167922974,
+      "learning_rate": 6.046820734896883e-05,
+      "loss": 2.9668,
+      "step": 240500
+    },
+    {
+      "epoch": 10.158917506217595,
+      "grad_norm": 1.2372212409973145,
+      "learning_rate": 6.038245508725293e-05,
+      "loss": 2.9684,
+      "step": 241000
+    },
+    {
+      "epoch": 10.179994098554145,
+      "grad_norm": 1.2415674924850464,
+      "learning_rate": 6.029670282553702e-05,
+      "loss": 2.9664,
+      "step": 241500
+    },
+    {
+      "epoch": 10.201070690890697,
+      "grad_norm": 1.2827967405319214,
+      "learning_rate": 6.0210950563821125e-05,
+      "loss": 2.9705,
+      "step": 242000
+    },
+    {
+      "epoch": 10.222147283227248,
+      "grad_norm": 1.453996181488037,
+      "learning_rate": 6.0125198302105223e-05,
+      "loss": 2.9672,
+      "step": 242500
+    },
+    {
+      "epoch": 10.2432238755638,
+      "grad_norm": 1.2031841278076172,
+      "learning_rate": 6.0039446040389315e-05,
+      "loss": 2.9654,
+      "step": 243000
+    },
+    {
+      "epoch": 10.26430046790035,
+      "grad_norm": 1.2767603397369385,
+      "learning_rate": 5.995369377867341e-05,
+      "loss": 2.9706,
+      "step": 243500
+    },
+    {
+      "epoch": 10.2853770602369,
+      "grad_norm": 1.2217897176742554,
+      "learning_rate": 5.986794151695752e-05,
+      "loss": 2.9608,
+      "step": 244000
+    },
+    {
+      "epoch": 10.306453652573452,
+      "grad_norm": 1.2551250457763672,
+      "learning_rate": 5.97821892552416e-05,
+      "loss": 2.9685,
+      "step": 244500
+    },
+    {
+      "epoch": 10.327530244910003,
+      "grad_norm": 1.2330348491668701,
+      "learning_rate": 5.969643699352571e-05,
+      "loss": 2.9675,
+      "step": 245000
+    },
+    {
+      "epoch": 10.348606837246553,
+      "grad_norm": 1.2744108438491821,
+      "learning_rate": 5.9610684731809806e-05,
+      "loss": 2.9662,
+      "step": 245500
+    },
+    {
+      "epoch": 10.369683429583105,
+      "grad_norm": 1.2321125268936157,
+      "learning_rate": 5.95249324700939e-05,
+      "loss": 2.965,
+      "step": 246000
+    },
+    {
+      "epoch": 10.390760021919656,
+      "grad_norm": 1.3419785499572754,
+      "learning_rate": 5.9439180208377996e-05,
+      "loss": 2.9659,
+      "step": 246500
+    },
+    {
+      "epoch": 10.411836614256208,
+      "grad_norm": 1.187544822692871,
+      "learning_rate": 5.93534279466621e-05,
+      "loss": 2.9623,
+      "step": 247000
+    },
+    {
+      "epoch": 10.432913206592758,
+      "grad_norm": 1.229836106300354,
+      "learning_rate": 5.9267675684946186e-05,
+      "loss": 2.9666,
+      "step": 247500
+    },
+    {
+      "epoch": 10.453989798929308,
+      "grad_norm": 1.2494527101516724,
+      "learning_rate": 5.918192342323029e-05,
+      "loss": 2.9599,
+      "step": 248000
+    },
+    {
+      "epoch": 10.47506639126586,
+      "grad_norm": 1.1675976514816284,
+      "learning_rate": 5.909617116151439e-05,
+      "loss": 2.9619,
+      "step": 248500
+    },
+    {
+      "epoch": 10.496142983602411,
+      "grad_norm": 1.2633516788482666,
+      "learning_rate": 5.901041889979848e-05,
+      "loss": 2.9591,
+      "step": 249000
+    },
+    {
+      "epoch": 10.517219575938963,
+      "grad_norm": 1.1857633590698242,
+      "learning_rate": 5.892466663808258e-05,
+      "loss": 2.9638,
+      "step": 249500
+    },
+    {
+      "epoch": 10.538296168275513,
+      "grad_norm": 1.2132319211959839,
+      "learning_rate": 5.8838914376366684e-05,
+      "loss": 2.9627,
+      "step": 250000
+    },
+    {
+      "epoch": 10.538296168275513,
+      "eval_accuracy": 0.7115544204840119,
+      "eval_loss": 2.7142748832702637,
+      "eval_runtime": 284.4224,
+      "eval_samples_per_second": 424.39,
+      "eval_steps_per_second": 4.423,
+      "step": 250000
+    },
+    {
+      "epoch": 10.559372760612064,
+      "grad_norm": 1.2441118955612183,
+      "learning_rate": 5.8753162114650776e-05,
+      "loss": 2.9662,
+      "step": 250500
+    },
+    {
+      "epoch": 10.580449352948616,
+      "grad_norm": 1.2175041437149048,
+      "learning_rate": 5.8667409852934874e-05,
+      "loss": 2.9606,
+      "step": 251000
+    },
+    {
+      "epoch": 10.601525945285166,
+      "grad_norm": 1.2527360916137695,
+      "learning_rate": 5.858165759121897e-05,
+      "loss": 2.9565,
+      "step": 251500
+    },
+    {
+      "epoch": 10.622602537621717,
+      "grad_norm": 1.2384111881256104,
+      "learning_rate": 5.8495905329503064e-05,
+      "loss": 2.9608,
+      "step": 252000
+    },
+    {
+      "epoch": 10.643679129958269,
+      "grad_norm": 1.189744234085083,
+      "learning_rate": 5.841015306778716e-05,
+      "loss": 2.9599,
+      "step": 252500
+    },
+    {
+      "epoch": 10.664755722294819,
+      "grad_norm": 1.2324283123016357,
+      "learning_rate": 5.832440080607127e-05,
+      "loss": 2.9613,
+      "step": 253000
+    },
+    {
+      "epoch": 10.685832314631371,
+      "grad_norm": 1.2628949880599976,
+      "learning_rate": 5.823864854435536e-05,
+      "loss": 2.9549,
+      "step": 253500
+    },
+    {
+      "epoch": 10.706908906967922,
+      "grad_norm": 1.3511600494384766,
+      "learning_rate": 5.815289628263946e-05,
+      "loss": 2.9609,
+      "step": 254000
+    },
+    {
+      "epoch": 10.727985499304472,
+      "grad_norm": 1.3185107707977295,
+      "learning_rate": 5.8067144020923556e-05,
+      "loss": 2.9629,
+      "step": 254500
+    },
+    {
+      "epoch": 10.749062091641024,
+      "grad_norm": 1.2684773206710815,
+      "learning_rate": 5.798139175920765e-05,
+      "loss": 2.9582,
+      "step": 255000
+    },
+    {
+      "epoch": 10.770138683977574,
+      "grad_norm": 1.2667359113693237,
+      "learning_rate": 5.789563949749175e-05,
+      "loss": 2.96,
+      "step": 255500
+    },
+    {
+      "epoch": 10.791215276314126,
+      "grad_norm": 1.264464020729065,
+      "learning_rate": 5.780988723577585e-05,
+      "loss": 2.9575,
+      "step": 256000
+    },
+    {
+      "epoch": 10.812291868650677,
+      "grad_norm": 1.2119172811508179,
+      "learning_rate": 5.772413497405994e-05,
+      "loss": 2.9588,
+      "step": 256500
+    },
+    {
+      "epoch": 10.833368460987227,
+      "grad_norm": 1.2197701930999756,
+      "learning_rate": 5.763838271234404e-05,
+      "loss": 2.9538,
+      "step": 257000
+    },
+    {
+      "epoch": 10.85444505332378,
+      "grad_norm": 1.2129262685775757,
+      "learning_rate": 5.7552630450628146e-05,
+      "loss": 2.9582,
+      "step": 257500
+    },
+    {
+      "epoch": 10.87552164566033,
+      "grad_norm": 1.2931581735610962,
+      "learning_rate": 5.746687818891223e-05,
+      "loss": 2.9554,
+      "step": 258000
+    },
+    {
+      "epoch": 10.89659823799688,
+      "grad_norm": 1.286311388015747,
+      "learning_rate": 5.7381125927196335e-05,
+      "loss": 2.9568,
+      "step": 258500
+    },
+    {
+      "epoch": 10.917674830333432,
+      "grad_norm": 1.316627025604248,
+      "learning_rate": 5.7295373665480434e-05,
+      "loss": 2.9573,
+      "step": 259000
+    },
+    {
+      "epoch": 10.938751422669982,
+      "grad_norm": 1.2069400548934937,
+      "learning_rate": 5.7209621403764525e-05,
+      "loss": 2.9555,
+      "step": 259500
+    },
+    {
+      "epoch": 10.959828015006533,
+      "grad_norm": 1.2886022329330444,
+      "learning_rate": 5.7123869142048624e-05,
+      "loss": 2.9517,
+      "step": 260000
+    },
+    {
+      "epoch": 10.959828015006533,
+      "eval_accuracy": 0.7128025459474971,
+      "eval_loss": 2.7073309421539307,
+      "eval_runtime": 273.2375,
+      "eval_samples_per_second": 441.762,
+      "eval_steps_per_second": 4.604,
+      "step": 260000
+    },
+    {
+      "epoch": 10.980904607343085,
+      "grad_norm": 1.221501111984253,
+      "learning_rate": 5.703811688033273e-05,
+      "loss": 2.9509,
+      "step": 260500
+    },
+    {
+      "epoch": 11.001981199679635,
+      "grad_norm": 1.2333225011825562,
+      "learning_rate": 5.695236461861681e-05,
+      "loss": 2.9527,
+      "step": 261000
+    },
+    {
+      "epoch": 11.023057792016187,
+      "grad_norm": 1.2529879808425903,
+      "learning_rate": 5.686661235690092e-05,
+      "loss": 2.9474,
+      "step": 261500
+    },
+    {
+      "epoch": 11.044134384352738,
+      "grad_norm": 1.324324369430542,
+      "learning_rate": 5.678086009518502e-05,
+      "loss": 2.9542,
+      "step": 262000
+    },
+    {
+      "epoch": 11.065210976689288,
+      "grad_norm": 1.2776871919631958,
+      "learning_rate": 5.669510783346911e-05,
+      "loss": 2.9551,
+      "step": 262500
+    },
+    {
+      "epoch": 11.08628756902584,
+      "grad_norm": 1.3128008842468262,
+      "learning_rate": 5.6609355571753207e-05,
+      "loss": 2.9493,
+      "step": 263000
+    },
+    {
+      "epoch": 11.10736416136239,
+      "grad_norm": 1.357959270477295,
+      "learning_rate": 5.652360331003731e-05,
+      "loss": 2.9519,
+      "step": 263500
+    },
+    {
+      "epoch": 11.128440753698943,
+      "grad_norm": 1.3428312540054321,
+      "learning_rate": 5.6437851048321396e-05,
+      "loss": 2.9479,
+      "step": 264000
+    },
+    {
+      "epoch": 11.149517346035493,
+      "grad_norm": 1.2635782957077026,
+      "learning_rate": 5.63520987866055e-05,
+      "loss": 2.9526,
+      "step": 264500
+    },
+    {
+      "epoch": 11.170593938372043,
+      "grad_norm": 1.2766858339309692,
+      "learning_rate": 5.62663465248896e-05,
+      "loss": 2.9504,
+      "step": 265000
+    },
+    {
+      "epoch": 11.191670530708596,
+      "grad_norm": 1.2542147636413574,
+      "learning_rate": 5.618059426317369e-05,
+      "loss": 2.9466,
+      "step": 265500
+    },
+    {
+      "epoch": 11.212747123045146,
+      "grad_norm": 1.2337288856506348,
+      "learning_rate": 5.609484200145779e-05,
+      "loss": 2.9514,
+      "step": 266000
+    },
+    {
+      "epoch": 11.233823715381696,
+      "grad_norm": 1.2284198999404907,
+      "learning_rate": 5.6009089739741895e-05,
+      "loss": 2.9517,
+      "step": 266500
+    },
+    {
+      "epoch": 11.254900307718248,
+      "grad_norm": 1.2387008666992188,
+      "learning_rate": 5.5923337478025986e-05,
+      "loss": 2.9458,
+      "step": 267000
+    },
+    {
+      "epoch": 11.275976900054799,
+      "grad_norm": 1.2603894472122192,
+      "learning_rate": 5.5837585216310085e-05,
+      "loss": 2.9513,
+      "step": 267500
+    },
+    {
+      "epoch": 11.29705349239135,
+      "grad_norm": 1.3191286325454712,
+      "learning_rate": 5.575183295459418e-05,
+      "loss": 2.9454,
+      "step": 268000
+    },
+    {
+      "epoch": 11.318130084727901,
+      "grad_norm": 1.2168670892715454,
+      "learning_rate": 5.5666080692878274e-05,
+      "loss": 2.9494,
+      "step": 268500
+    },
+    {
+      "epoch": 11.339206677064452,
+      "grad_norm": 1.2221918106079102,
+      "learning_rate": 5.558032843116238e-05,
+      "loss": 2.949,
+      "step": 269000
+    },
+    {
+      "epoch": 11.360283269401004,
+      "grad_norm": 1.2680492401123047,
+      "learning_rate": 5.549457616944648e-05,
+      "loss": 2.9495,
+      "step": 269500
+    },
+    {
+      "epoch": 11.381359861737554,
+      "grad_norm": 1.2497395277023315,
+      "learning_rate": 5.540882390773057e-05,
+      "loss": 2.9439,
+      "step": 270000
+    },
+    {
+      "epoch": 11.381359861737554,
+      "eval_accuracy": 0.7135447144552998,
+      "eval_loss": 2.70328426361084,
+      "eval_runtime": 273.8781,
+      "eval_samples_per_second": 440.729,
+      "eval_steps_per_second": 4.593,
+      "step": 270000
+    },
+    {
+      "epoch": 11.402436454074106,
+      "grad_norm": 1.2360596656799316,
+      "learning_rate": 5.532307164601467e-05,
+      "loss": 2.9466,
+      "step": 270500
+    },
+    {
+      "epoch": 11.423513046410656,
+      "grad_norm": 1.3028604984283447,
+      "learning_rate": 5.5237319384298766e-05,
+      "loss": 2.9455,
+      "step": 271000
+    },
+    {
+      "epoch": 11.444589638747207,
+      "grad_norm": 1.1838159561157227,
+      "learning_rate": 5.515156712258286e-05,
+      "loss": 2.9416,
+      "step": 271500
+    },
+    {
+      "epoch": 11.465666231083759,
+      "grad_norm": 1.2555358409881592,
+      "learning_rate": 5.506581486086696e-05,
+      "loss": 2.9462,
+      "step": 272000
+    },
+    {
+      "epoch": 11.48674282342031,
+      "grad_norm": 1.2405608892440796,
+      "learning_rate": 5.498006259915105e-05,
+      "loss": 2.9456,
+      "step": 272500
+    },
+    {
+      "epoch": 11.50781941575686,
+      "grad_norm": 1.1961925029754639,
+      "learning_rate": 5.489431033743515e-05,
+      "loss": 2.945,
+      "step": 273000
+    },
+    {
+      "epoch": 11.528896008093412,
+      "grad_norm": 1.2516721487045288,
+      "learning_rate": 5.480855807571925e-05,
+      "loss": 2.9478,
+      "step": 273500
+    },
+    {
+      "epoch": 11.549972600429962,
+      "grad_norm": 1.2366604804992676,
+      "learning_rate": 5.472280581400334e-05,
+      "loss": 2.9428,
+      "step": 274000
+    },
+    {
+      "epoch": 11.571049192766514,
+      "grad_norm": 1.2716665267944336,
+      "learning_rate": 5.463705355228744e-05,
+      "loss": 2.9435,
+      "step": 274500
+    },
+    {
+      "epoch": 11.592125785103065,
+      "grad_norm": 1.2203969955444336,
+      "learning_rate": 5.4551301290571546e-05,
+      "loss": 2.9472,
+      "step": 275000
+    },
+    {
+      "epoch": 11.613202377439615,
+      "grad_norm": 1.2250478267669678,
+      "learning_rate": 5.446554902885563e-05,
+      "loss": 2.9447,
+      "step": 275500
+    },
+    {
+      "epoch": 11.634278969776167,
+      "grad_norm": 1.2054928541183472,
+      "learning_rate": 5.4379796767139736e-05,
+      "loss": 2.9411,
+      "step": 276000
+    },
+    {
+      "epoch": 11.655355562112717,
+      "grad_norm": 1.2898095846176147,
+      "learning_rate": 5.4294044505423834e-05,
+      "loss": 2.9418,
+      "step": 276500
+    },
+    {
+      "epoch": 11.67643215444927,
+      "grad_norm": 1.193176031112671,
+      "learning_rate": 5.4208292243707925e-05,
+      "loss": 2.9437,
+      "step": 277000
+    },
+    {
+      "epoch": 11.69750874678582,
+      "grad_norm": 1.277672529220581,
+      "learning_rate": 5.4122539981992024e-05,
+      "loss": 2.9424,
+      "step": 277500
+    },
+    {
+      "epoch": 11.71858533912237,
+      "grad_norm": 1.195682406425476,
+      "learning_rate": 5.403678772027613e-05,
+      "loss": 2.937,
+      "step": 278000
+    },
+    {
+      "epoch": 11.739661931458922,
+      "grad_norm": 1.214914321899414,
+      "learning_rate": 5.395103545856022e-05,
+      "loss": 2.9421,
+      "step": 278500
+    },
+    {
+      "epoch": 11.760738523795473,
+      "grad_norm": 1.3032463788986206,
+      "learning_rate": 5.386528319684432e-05,
+      "loss": 2.9377,
+      "step": 279000
+    },
+    {
+      "epoch": 11.781815116132023,
+      "grad_norm": 1.198391079902649,
+      "learning_rate": 5.377953093512842e-05,
+      "loss": 2.9364,
+      "step": 279500
+    },
+    {
+      "epoch": 11.802891708468575,
+      "grad_norm": 1.236789584159851,
+      "learning_rate": 5.369377867341251e-05,
+      "loss": 2.944,
+      "step": 280000
+    },
+    {
+      "epoch": 11.802891708468575,
+      "eval_accuracy": 0.7150541328822688,
+      "eval_loss": 2.69443678855896,
+      "eval_runtime": 248.1175,
+      "eval_samples_per_second": 486.487,
+      "eval_steps_per_second": 5.07,
+      "step": 280000
+    },
+    {
+      "epoch": 11.823968300805126,
+      "grad_norm": 1.2862184047698975,
+      "learning_rate": 5.3608026411696614e-05,
+      "loss": 2.9453,
+      "step": 280500
+    },
+    {
+      "epoch": 11.845044893141678,
+      "grad_norm": 1.3291096687316895,
+      "learning_rate": 5.352227414998071e-05,
+      "loss": 2.9345,
+      "step": 281000
+    },
+    {
+      "epoch": 11.866121485478228,
+      "grad_norm": 1.3151220083236694,
+      "learning_rate": 5.34365218882648e-05,
+      "loss": 2.9393,
+      "step": 281500
+    },
+    {
+      "epoch": 11.887198077814778,
+      "grad_norm": 1.193091630935669,
+      "learning_rate": 5.33507696265489e-05,
+      "loss": 2.9327,
+      "step": 282000
+    },
+    {
+      "epoch": 11.90827467015133,
+      "grad_norm": 1.2843496799468994,
+      "learning_rate": 5.3265017364833e-05,
+      "loss": 2.9428,
+      "step": 282500
+    },
+    {
+      "epoch": 11.92935126248788,
+      "grad_norm": 1.2695279121398926,
+      "learning_rate": 5.317926510311709e-05,
+      "loss": 2.9366,
+      "step": 283000
+    },
+    {
+      "epoch": 11.950427854824433,
+      "grad_norm": 1.2595186233520508,
+      "learning_rate": 5.3093512841401197e-05,
+      "loss": 2.9364,
+      "step": 283500
+    },
+    {
+      "epoch": 11.971504447160983,
+      "grad_norm": 1.260259747505188,
+      "learning_rate": 5.3007760579685295e-05,
+      "loss": 2.9393,
+      "step": 284000
+    },
+    {
+      "epoch": 11.992581039497534,
+      "grad_norm": 1.2344937324523926,
+      "learning_rate": 5.2922008317969386e-05,
+      "loss": 2.9422,
+      "step": 284500
+    },
+    {
+      "epoch": 12.013657631834086,
+      "grad_norm": 1.2368005514144897,
+      "learning_rate": 5.2836256056253485e-05,
+      "loss": 2.9358,
+      "step": 285000
+    },
+    {
+      "epoch": 12.034734224170636,
+      "grad_norm": 1.2921478748321533,
+      "learning_rate": 5.275050379453759e-05,
+      "loss": 2.9377,
+      "step": 285500
+    },
+    {
+      "epoch": 12.055810816507186,
+      "grad_norm": 1.2368264198303223,
+      "learning_rate": 5.2664751532821675e-05,
+      "loss": 2.934,
+      "step": 286000
+    },
+    {
+      "epoch": 12.076887408843739,
+      "grad_norm": 1.2401432991027832,
+      "learning_rate": 5.257899927110578e-05,
+      "loss": 2.9346,
+      "step": 286500
+    },
+    {
+      "epoch": 12.097964001180289,
+      "grad_norm": 1.226215124130249,
+      "learning_rate": 5.249324700938988e-05,
+      "loss": 2.9315,
+      "step": 287000
+    },
+    {
+      "epoch": 12.119040593516841,
+      "grad_norm": 1.2288109064102173,
+      "learning_rate": 5.240749474767397e-05,
+      "loss": 2.931,
+      "step": 287500
+    },
+    {
+      "epoch": 12.140117185853391,
+      "grad_norm": 1.3216756582260132,
+      "learning_rate": 5.232174248595807e-05,
+      "loss": 2.9357,
+      "step": 288000
+    },
+    {
+      "epoch": 12.161193778189942,
+      "grad_norm": 1.2462188005447388,
+      "learning_rate": 5.223599022424217e-05,
+      "loss": 2.9345,
+      "step": 288500
+    },
+    {
+      "epoch": 12.182270370526494,
+      "grad_norm": 1.3383574485778809,
+      "learning_rate": 5.215023796252626e-05,
+      "loss": 2.9395,
+      "step": 289000
+    },
+    {
+      "epoch": 12.203346962863044,
+      "grad_norm": 1.3061093091964722,
+      "learning_rate": 5.206448570081036e-05,
+      "loss": 2.9295,
+      "step": 289500
+    },
+    {
+      "epoch": 12.224423555199595,
+      "grad_norm": 1.2837328910827637,
+      "learning_rate": 5.197873343909446e-05,
+      "loss": 2.9295,
+      "step": 290000
+    },
+    {
+      "epoch": 12.224423555199595,
+      "eval_accuracy": 0.7156299179927667,
+      "eval_loss": 2.6887123584747314,
+      "eval_runtime": 255.324,
+      "eval_samples_per_second": 472.756,
+      "eval_steps_per_second": 4.927,
+      "step": 290000
+    },
+    {
+      "epoch": 12.245500147536147,
+      "grad_norm": 1.310732364654541,
+      "learning_rate": 5.189298117737855e-05,
+      "loss": 2.9313,
+      "step": 290500
+    },
+    {
+      "epoch": 12.266576739872697,
+      "grad_norm": 1.2365353107452393,
+      "learning_rate": 5.180722891566265e-05,
+      "loss": 2.9318,
+      "step": 291000
+    },
+    {
+      "epoch": 12.28765333220925,
+      "grad_norm": 1.2555338144302368,
+      "learning_rate": 5.1721476653946756e-05,
+      "loss": 2.9286,
+      "step": 291500
+    },
+    {
+      "epoch": 12.3087299245458,
+      "grad_norm": 1.258872151374817,
+      "learning_rate": 5.163572439223085e-05,
+      "loss": 2.9315,
+      "step": 292000
+    },
+    {
+      "epoch": 12.32980651688235,
+      "grad_norm": 1.2603719234466553,
+      "learning_rate": 5.1549972130514946e-05,
+      "loss": 2.9297,
+      "step": 292500
+    },
+    {
+      "epoch": 12.350883109218902,
+      "grad_norm": 1.263229489326477,
+      "learning_rate": 5.1464219868799044e-05,
+      "loss": 2.9307,
+      "step": 293000
+    },
+    {
+      "epoch": 12.371959701555452,
+      "grad_norm": 1.3087921142578125,
+      "learning_rate": 5.1378467607083136e-05,
+      "loss": 2.9293,
+      "step": 293500
+    },
+    {
+      "epoch": 12.393036293892003,
+      "grad_norm": 1.2571992874145508,
+      "learning_rate": 5.1292715345367234e-05,
+      "loss": 2.9307,
+      "step": 294000
+    },
+    {
+      "epoch": 12.414112886228555,
+      "grad_norm": 1.2401467561721802,
+      "learning_rate": 5.120696308365134e-05,
+      "loss": 2.9272,
+      "step": 294500
+    },
+    {
+      "epoch": 12.435189478565105,
+      "grad_norm": 1.250430941581726,
+      "learning_rate": 5.112121082193543e-05,
+      "loss": 2.9309,
+      "step": 295000
+    },
+    {
+      "epoch": 12.456266070901657,
+      "grad_norm": 1.2851345539093018,
+      "learning_rate": 5.103545856021953e-05,
+      "loss": 2.9295,
+      "step": 295500
+    },
+    {
+      "epoch": 12.477342663238208,
+      "grad_norm": 1.259631633758545,
+      "learning_rate": 5.094970629850363e-05,
+      "loss": 2.9285,
+      "step": 296000
+    },
+    {
+      "epoch": 12.498419255574758,
+      "grad_norm": 1.253143548965454,
+      "learning_rate": 5.086395403678772e-05,
+      "loss": 2.9305,
+      "step": 296500
+    },
+    {
+      "epoch": 12.51949584791131,
+      "grad_norm": 1.2258012294769287,
+      "learning_rate": 5.0778201775071824e-05,
+      "loss": 2.9334,
+      "step": 297000
+    },
+    {
+      "epoch": 12.54057244024786,
+      "grad_norm": 1.3721328973770142,
+      "learning_rate": 5.069244951335592e-05,
+      "loss": 2.9308,
+      "step": 297500
+    },
+    {
+      "epoch": 12.561649032584413,
+      "grad_norm": 1.236648440361023,
+      "learning_rate": 5.0606697251640014e-05,
+      "loss": 2.9246,
+      "step": 298000
+    },
+    {
+      "epoch": 12.582725624920963,
+      "grad_norm": 1.2675775289535522,
+      "learning_rate": 5.052094498992411e-05,
+      "loss": 2.931,
+      "step": 298500
+    },
+    {
+      "epoch": 12.603802217257513,
+      "grad_norm": 1.268498182296753,
+      "learning_rate": 5.043519272820822e-05,
+      "loss": 2.9316,
+      "step": 299000
+    },
+    {
+      "epoch": 12.624878809594065,
+      "grad_norm": 1.2424331903457642,
+      "learning_rate": 5.03494404664923e-05,
+      "loss": 2.9249,
+      "step": 299500
+    },
+    {
+      "epoch": 12.645955401930616,
+      "grad_norm": 1.2795358896255493,
+      "learning_rate": 5.026368820477641e-05,
+      "loss": 2.9263,
+      "step": 300000
+    },
+    {
+      "epoch": 12.645955401930616,
+      "eval_accuracy": 0.7171618224967606,
+      "eval_loss": 2.68229603767395,
+      "eval_runtime": 251.4166,
+      "eval_samples_per_second": 480.104,
+      "eval_steps_per_second": 5.004,
+      "step": 300000
+    },
+    {
+      "epoch": 12.667031994267166,
+      "grad_norm": 1.2829195261001587,
+      "learning_rate": 5.0177935943060505e-05,
+      "loss": 2.9225,
+      "step": 300500
+    },
+    {
+      "epoch": 12.688108586603718,
+      "grad_norm": 1.3196477890014648,
+      "learning_rate": 5.00921836813446e-05,
+      "loss": 2.9276,
+      "step": 301000
+    },
+    {
+      "epoch": 12.709185178940269,
+      "grad_norm": 1.241470456123352,
+      "learning_rate": 5.0006431419628695e-05,
+      "loss": 2.9275,
+      "step": 301500
+    },
+    {
+      "epoch": 12.73026177127682,
+      "grad_norm": 1.3103524446487427,
+      "learning_rate": 4.992067915791279e-05,
+      "loss": 2.9259,
+      "step": 302000
+    },
+    {
+      "epoch": 12.751338363613371,
+      "grad_norm": 1.2887961864471436,
+      "learning_rate": 4.9834926896196885e-05,
+      "loss": 2.9232,
+      "step": 302500
+    },
+    {
+      "epoch": 12.772414955949921,
+      "grad_norm": 1.3140320777893066,
+      "learning_rate": 4.974917463448099e-05,
+      "loss": 2.928,
+      "step": 303000
+    },
+    {
+      "epoch": 12.793491548286474,
+      "grad_norm": 1.234001874923706,
+      "learning_rate": 4.966342237276508e-05,
+      "loss": 2.9298,
+      "step": 303500
+    },
+    {
+      "epoch": 12.814568140623024,
+      "grad_norm": 1.262837290763855,
+      "learning_rate": 4.957767011104918e-05,
+      "loss": 2.9272,
+      "step": 304000
+    },
+    {
+      "epoch": 12.835644732959576,
+      "grad_norm": 1.2862893342971802,
+      "learning_rate": 4.949191784933328e-05,
+      "loss": 2.9255,
+      "step": 304500
+    },
+    {
+      "epoch": 12.856721325296126,
+      "grad_norm": 1.206764578819275,
+      "learning_rate": 4.9406165587617376e-05,
+      "loss": 2.9235,
+      "step": 305000
+    },
+    {
+      "epoch": 12.877797917632677,
+      "grad_norm": 1.258020281791687,
+      "learning_rate": 4.932041332590147e-05,
+      "loss": 2.9265,
+      "step": 305500
+    },
+    {
+      "epoch": 12.898874509969229,
+      "grad_norm": 1.2857751846313477,
+      "learning_rate": 4.923466106418557e-05,
+      "loss": 2.9226,
+      "step": 306000
+    },
+    {
+      "epoch": 12.91995110230578,
+      "grad_norm": 1.3123788833618164,
+      "learning_rate": 4.9148908802469665e-05,
+      "loss": 2.9207,
+      "step": 306500
+    },
+    {
+      "epoch": 12.94102769464233,
+      "grad_norm": 1.2305909395217896,
+      "learning_rate": 4.906315654075376e-05,
+      "loss": 2.9228,
+      "step": 307000
+    },
+    {
+      "epoch": 12.962104286978882,
+      "grad_norm": 1.2626748085021973,
+      "learning_rate": 4.897740427903786e-05,
+      "loss": 2.9269,
+      "step": 307500
+    },
+    {
+      "epoch": 12.983180879315432,
+      "grad_norm": 1.3005268573760986,
+      "learning_rate": 4.889165201732196e-05,
+      "loss": 2.9264,
+      "step": 308000
+    },
+    {
+      "epoch": 13.004257471651984,
+      "grad_norm": 1.2490612268447876,
+      "learning_rate": 4.880589975560606e-05,
+      "loss": 2.9228,
+      "step": 308500
+    },
+    {
+      "epoch": 13.025334063988534,
+      "grad_norm": 1.3518792390823364,
+      "learning_rate": 4.8720147493890156e-05,
+      "loss": 2.9159,
+      "step": 309000
+    },
+    {
+      "epoch": 13.046410656325085,
+      "grad_norm": 1.3126299381256104,
+      "learning_rate": 4.8634395232174254e-05,
+      "loss": 2.9211,
+      "step": 309500
+    },
+    {
+      "epoch": 13.067487248661637,
+      "grad_norm": 1.2986441850662231,
+      "learning_rate": 4.8548642970458346e-05,
+      "loss": 2.9172,
+      "step": 310000
+    },
+    {
+      "epoch": 13.067487248661637,
+      "eval_accuracy": 0.7179588800598506,
+      "eval_loss": 2.677154064178467,
+      "eval_runtime": 266.6075,
+      "eval_samples_per_second": 452.748,
+      "eval_steps_per_second": 4.719,
+      "step": 310000
+    },
+    {
+      "epoch": 13.088563840998187,
+      "grad_norm": 1.287224531173706,
+      "learning_rate": 4.846289070874245e-05,
+      "loss": 2.9206,
+      "step": 310500
+    },
+    {
+      "epoch": 13.109640433334738,
+      "grad_norm": 1.306684136390686,
+      "learning_rate": 4.837713844702654e-05,
+      "loss": 2.921,
+      "step": 311000
+    },
+    {
+      "epoch": 13.13071702567129,
+      "grad_norm": 1.336012601852417,
+      "learning_rate": 4.829138618531064e-05,
+      "loss": 2.9164,
+      "step": 311500
+    },
+    {
+      "epoch": 13.15179361800784,
+      "grad_norm": 1.2326065301895142,
+      "learning_rate": 4.820563392359474e-05,
+      "loss": 2.9192,
+      "step": 312000
+    },
+    {
+      "epoch": 13.172870210344392,
+      "grad_norm": 1.242527723312378,
+      "learning_rate": 4.811988166187884e-05,
+      "loss": 2.9161,
+      "step": 312500
+    },
+    {
+      "epoch": 13.193946802680943,
+      "grad_norm": 1.2790569067001343,
+      "learning_rate": 4.803412940016293e-05,
+      "loss": 2.9167,
+      "step": 313000
+    },
+    {
+      "epoch": 13.215023395017493,
+      "grad_norm": 1.3252450227737427,
+      "learning_rate": 4.7948377138447034e-05,
+      "loss": 2.9182,
+      "step": 313500
+    },
+    {
+      "epoch": 13.236099987354045,
+      "grad_norm": 1.2895890474319458,
+      "learning_rate": 4.7862624876731126e-05,
+      "loss": 2.9147,
+      "step": 314000
+    },
+    {
+      "epoch": 13.257176579690595,
+      "grad_norm": 1.2563567161560059,
+      "learning_rate": 4.7776872615015224e-05,
+      "loss": 2.9165,
+      "step": 314500
+    },
+    {
+      "epoch": 13.278253172027147,
+      "grad_norm": 1.2847484350204468,
+      "learning_rate": 4.7691120353299315e-05,
+      "loss": 2.914,
+      "step": 315000
+    },
+    {
+      "epoch": 13.299329764363698,
+      "grad_norm": 1.2387272119522095,
+      "learning_rate": 4.760536809158342e-05,
+      "loss": 2.9205,
+      "step": 315500
+    },
+    {
+      "epoch": 13.320406356700248,
+      "grad_norm": 1.2138867378234863,
+      "learning_rate": 4.751961582986751e-05,
+      "loss": 2.9202,
+      "step": 316000
+    },
+    {
+      "epoch": 13.3414829490368,
+      "grad_norm": 1.3116387128829956,
+      "learning_rate": 4.743386356815161e-05,
+      "loss": 2.9176,
+      "step": 316500
+    },
+    {
+      "epoch": 13.36255954137335,
+      "grad_norm": 1.2168372869491577,
+      "learning_rate": 4.734811130643571e-05,
+      "loss": 2.917,
+      "step": 317000
+    },
+    {
+      "epoch": 13.383636133709901,
+      "grad_norm": 1.3063400983810425,
+      "learning_rate": 4.726235904471981e-05,
+      "loss": 2.9128,
+      "step": 317500
+    },
+    {
+      "epoch": 13.404712726046453,
+      "grad_norm": 1.3164465427398682,
+      "learning_rate": 4.71766067830039e-05,
+      "loss": 2.916,
+      "step": 318000
+    },
+    {
+      "epoch": 13.425789318383003,
+      "grad_norm": 1.335506558418274,
+      "learning_rate": 4.7090854521288004e-05,
+      "loss": 2.9187,
+      "step": 318500
+    },
+    {
+      "epoch": 13.446865910719556,
+      "grad_norm": 1.3004755973815918,
+      "learning_rate": 4.7005102259572095e-05,
+      "loss": 2.9138,
+      "step": 319000
+    },
+    {
+      "epoch": 13.467942503056106,
+      "grad_norm": 1.4062693119049072,
+      "learning_rate": 4.6919349997856193e-05,
+      "loss": 2.9188,
+      "step": 319500
+    },
+    {
+      "epoch": 13.489019095392656,
+      "grad_norm": 1.5376149415969849,
+      "learning_rate": 4.683359773614029e-05,
+      "loss": 2.9126,
+      "step": 320000
+    },
+    {
+      "epoch": 13.489019095392656,
+      "eval_accuracy": 0.7187309066343285,
+      "eval_loss": 2.6722350120544434,
+      "eval_runtime": 277.0316,
+      "eval_samples_per_second": 435.712,
+      "eval_steps_per_second": 4.541,
+      "step": 320000
+    },
+    {
+      "epoch": 13.510095687729208,
+      "grad_norm": 1.3062412738800049,
+      "learning_rate": 4.674784547442439e-05,
+      "loss": 2.9201,
+      "step": 320500
+    },
+    {
+      "epoch": 13.531172280065759,
+      "grad_norm": 1.2982935905456543,
+      "learning_rate": 4.666209321270849e-05,
+      "loss": 2.9151,
+      "step": 321000
+    },
+    {
+      "epoch": 13.55224887240231,
+      "grad_norm": 1.2396240234375,
+      "learning_rate": 4.657634095099259e-05,
+      "loss": 2.9151,
+      "step": 321500
+    },
+    {
+      "epoch": 13.573325464738861,
+      "grad_norm": 1.336423635482788,
+      "learning_rate": 4.6490588689276685e-05,
+      "loss": 2.9157,
+      "step": 322000
+    },
+    {
+      "epoch": 13.594402057075412,
+      "grad_norm": 1.2347558736801147,
+      "learning_rate": 4.6404836427560777e-05,
+      "loss": 2.9164,
+      "step": 322500
+    },
+    {
+      "epoch": 13.615478649411964,
+      "grad_norm": 1.2569773197174072,
+      "learning_rate": 4.6319084165844875e-05,
+      "loss": 2.9102,
+      "step": 323000
+    },
+    {
+      "epoch": 13.636555241748514,
+      "grad_norm": 1.218084454536438,
+      "learning_rate": 4.623333190412897e-05,
+      "loss": 2.9158,
+      "step": 323500
+    },
+    {
+      "epoch": 13.657631834085064,
+      "grad_norm": 1.2909842729568481,
+      "learning_rate": 4.614757964241307e-05,
+      "loss": 2.9113,
+      "step": 324000
+    },
+    {
+      "epoch": 13.678708426421617,
+      "grad_norm": 1.2685658931732178,
+      "learning_rate": 4.606182738069717e-05,
+      "loss": 2.9148,
+      "step": 324500
+    },
+    {
+      "epoch": 13.699785018758167,
+      "grad_norm": 1.400235652923584,
+      "learning_rate": 4.597607511898127e-05,
+      "loss": 2.912,
+      "step": 325000
+    },
+    {
+      "epoch": 13.720861611094719,
+      "grad_norm": 1.2897617816925049,
+      "learning_rate": 4.589032285726536e-05,
+      "loss": 2.9121,
+      "step": 325500
+    },
+    {
+      "epoch": 13.74193820343127,
+      "grad_norm": 1.3247989416122437,
+      "learning_rate": 4.5804570595549465e-05,
+      "loss": 2.9118,
+      "step": 326000
+    },
+    {
+      "epoch": 13.76301479576782,
+      "grad_norm": 1.4035221338272095,
+      "learning_rate": 4.5718818333833556e-05,
+      "loss": 2.9146,
+      "step": 326500
+    },
+    {
+      "epoch": 13.784091388104372,
+      "grad_norm": 1.2971584796905518,
+      "learning_rate": 4.5633066072117655e-05,
+      "loss": 2.9139,
+      "step": 327000
+    },
+    {
+      "epoch": 13.805167980440922,
+      "grad_norm": 1.270384430885315,
+      "learning_rate": 4.554731381040175e-05,
+      "loss": 2.9084,
+      "step": 327500
+    },
+    {
+      "epoch": 13.826244572777473,
+      "grad_norm": 1.3161516189575195,
+      "learning_rate": 4.546156154868585e-05,
+      "loss": 2.9124,
+      "step": 328000
+    },
+    {
+      "epoch": 13.847321165114025,
+      "grad_norm": 1.3115137815475464,
+      "learning_rate": 4.537580928696994e-05,
+      "loss": 2.9115,
+      "step": 328500
+    },
+    {
+      "epoch": 13.868397757450575,
+      "grad_norm": 1.4432646036148071,
+      "learning_rate": 4.529005702525405e-05,
+      "loss": 2.9114,
+      "step": 329000
+    },
+    {
+      "epoch": 13.889474349787127,
+      "grad_norm": 1.3065561056137085,
+      "learning_rate": 4.520430476353814e-05,
+      "loss": 2.9094,
+      "step": 329500
+    },
+    {
+      "epoch": 13.910550942123677,
+      "grad_norm": 1.3969290256500244,
+      "learning_rate": 4.511855250182224e-05,
+      "loss": 2.9094,
+      "step": 330000
+    },
+    {
+      "epoch": 13.910550942123677,
+      "eval_accuracy": 0.7194153863267829,
+      "eval_loss": 2.668358564376831,
+      "eval_runtime": 275.0729,
+      "eval_samples_per_second": 438.815,
+      "eval_steps_per_second": 4.573,
+      "step": 330000
+    },
+    {
+      "epoch": 13.931627534460228,
+      "grad_norm": 1.2091257572174072,
+      "learning_rate": 4.5032800240106336e-05,
+      "loss": 2.9112,
+      "step": 330500
+    },
+    {
+      "epoch": 13.95270412679678,
+      "grad_norm": 1.2310550212860107,
+      "learning_rate": 4.4947047978390434e-05,
+      "loss": 2.9145,
+      "step": 331000
+    },
+    {
+      "epoch": 13.97378071913333,
+      "grad_norm": 1.399845838546753,
+      "learning_rate": 4.4861295716674526e-05,
+      "loss": 2.912,
+      "step": 331500
+    },
+    {
+      "epoch": 13.994857311469882,
+      "grad_norm": 1.2972668409347534,
+      "learning_rate": 4.4775543454958624e-05,
+      "loss": 2.912,
+      "step": 332000
+    },
+    {
+      "epoch": 14.015933903806433,
+      "grad_norm": 1.3075852394104004,
+      "learning_rate": 4.468979119324272e-05,
+      "loss": 2.9052,
+      "step": 332500
+    },
+    {
+      "epoch": 14.037010496142983,
+      "grad_norm": 1.2743669748306274,
+      "learning_rate": 4.460403893152682e-05,
+      "loss": 2.9059,
+      "step": 333000
+    },
+    {
+      "epoch": 14.058087088479535,
+      "grad_norm": 1.257858395576477,
+      "learning_rate": 4.451828666981092e-05,
+      "loss": 2.9054,
+      "step": 333500
+    },
+    {
+      "epoch": 14.079163680816086,
+      "grad_norm": 1.3593066930770874,
+      "learning_rate": 4.443253440809502e-05,
+      "loss": 2.9061,
+      "step": 334000
+    },
+    {
+      "epoch": 14.100240273152636,
+      "grad_norm": 1.1838126182556152,
+      "learning_rate": 4.4346782146379116e-05,
+      "loss": 2.9049,
+      "step": 334500
+    },
+    {
+      "epoch": 14.121316865489188,
+      "grad_norm": 1.3614015579223633,
+      "learning_rate": 4.426102988466321e-05,
+      "loss": 2.9037,
+      "step": 335000
+    },
+    {
+      "epoch": 14.142393457825738,
+      "grad_norm": 1.281829833984375,
+      "learning_rate": 4.4175277622947305e-05,
+      "loss": 2.8999,
+      "step": 335500
+    },
+    {
+      "epoch": 14.16347005016229,
+      "grad_norm": 1.2840781211853027,
+      "learning_rate": 4.4089525361231404e-05,
+      "loss": 2.9045,
+      "step": 336000
+    },
+    {
+      "epoch": 14.184546642498841,
+      "grad_norm": 1.2414348125457764,
+      "learning_rate": 4.40037730995155e-05,
+      "loss": 2.9021,
+      "step": 336500
+    },
+    {
+      "epoch": 14.205623234835391,
+      "grad_norm": 1.3014639616012573,
+      "learning_rate": 4.39180208377996e-05,
+      "loss": 2.9058,
+      "step": 337000
+    },
+    {
+      "epoch": 14.226699827171943,
+      "grad_norm": 1.321497917175293,
+      "learning_rate": 4.38322685760837e-05,
+      "loss": 2.9068,
+      "step": 337500
+    },
+    {
+      "epoch": 14.247776419508494,
+      "grad_norm": 1.2901164293289185,
+      "learning_rate": 4.374651631436779e-05,
+      "loss": 2.9038,
+      "step": 338000
+    },
+    {
+      "epoch": 14.268853011845044,
+      "grad_norm": 1.289092779159546,
+      "learning_rate": 4.3660764052651895e-05,
+      "loss": 2.9063,
+      "step": 338500
+    },
+    {
+      "epoch": 14.289929604181596,
+      "grad_norm": 1.3083350658416748,
+      "learning_rate": 4.357501179093599e-05,
+      "loss": 2.9057,
+      "step": 339000
+    },
+    {
+      "epoch": 14.311006196518147,
+      "grad_norm": 1.3186020851135254,
+      "learning_rate": 4.3489259529220085e-05,
+      "loss": 2.9014,
+      "step": 339500
+    },
+    {
+      "epoch": 14.332082788854699,
+      "grad_norm": 1.3914344310760498,
+      "learning_rate": 4.3403507267504183e-05,
+      "loss": 2.9054,
+      "step": 340000
+    },
+    {
+      "epoch": 14.332082788854699,
+      "eval_accuracy": 0.7204058329496982,
+      "eval_loss": 2.6613519191741943,
+      "eval_runtime": 430.7142,
+      "eval_samples_per_second": 280.246,
+      "eval_steps_per_second": 2.921,
+      "step": 340000
+    },
+    {
+      "epoch": 14.353159381191249,
+      "grad_norm": 1.351951003074646,
+      "learning_rate": 4.331775500578828e-05,
+      "loss": 2.9024,
+      "step": 340500
+    },
+    {
+      "epoch": 14.3742359735278,
+      "grad_norm": 1.2848197221755981,
+      "learning_rate": 4.323200274407237e-05,
+      "loss": 2.9032,
+      "step": 341000
+    },
+    {
+      "epoch": 14.395312565864351,
+      "grad_norm": 1.3082396984100342,
+      "learning_rate": 4.314625048235648e-05,
+      "loss": 2.9022,
+      "step": 341500
+    },
+    {
+      "epoch": 14.416389158200902,
+      "grad_norm": 1.3247700929641724,
+      "learning_rate": 4.306049822064057e-05,
+      "loss": 2.9054,
+      "step": 342000
+    },
+    {
+      "epoch": 14.437465750537454,
+      "grad_norm": 1.2992305755615234,
+      "learning_rate": 4.297474595892467e-05,
+      "loss": 2.9036,
+      "step": 342500
+    },
+    {
+      "epoch": 14.458542342874004,
+      "grad_norm": 1.2789113521575928,
+      "learning_rate": 4.2888993697208767e-05,
+      "loss": 2.8995,
+      "step": 343000
+    },
+    {
+      "epoch": 14.479618935210555,
+      "grad_norm": 1.2429088354110718,
+      "learning_rate": 4.2803241435492865e-05,
+      "loss": 2.9037,
+      "step": 343500
+    },
+    {
+      "epoch": 14.500695527547107,
+      "grad_norm": 1.2560932636260986,
+      "learning_rate": 4.2717489173776956e-05,
+      "loss": 2.9021,
+      "step": 344000
+    },
+    {
+      "epoch": 14.521772119883657,
+      "grad_norm": 1.2529054880142212,
+      "learning_rate": 4.263173691206106e-05,
+      "loss": 2.9004,
+      "step": 344500
+    },
+    {
+      "epoch": 14.542848712220207,
+      "grad_norm": 1.2551056146621704,
+      "learning_rate": 4.254598465034515e-05,
+      "loss": 2.903,
+      "step": 345000
+    },
+    {
+      "epoch": 14.56392530455676,
+      "grad_norm": 1.290547251701355,
+      "learning_rate": 4.246023238862925e-05,
+      "loss": 2.8996,
+      "step": 345500
+    },
+    {
+      "epoch": 14.58500189689331,
+      "grad_norm": 1.277643084526062,
+      "learning_rate": 4.237448012691335e-05,
+      "loss": 2.9019,
+      "step": 346000
+    },
+    {
+      "epoch": 14.606078489229862,
+      "grad_norm": 1.3447606563568115,
+      "learning_rate": 4.228872786519745e-05,
+      "loss": 2.8998,
+      "step": 346500
+    },
+    {
+      "epoch": 14.627155081566412,
+      "grad_norm": 1.3026471138000488,
+      "learning_rate": 4.220297560348154e-05,
+      "loss": 2.9024,
+      "step": 347000
+    },
+    {
+      "epoch": 14.648231673902963,
+      "grad_norm": 1.2938125133514404,
+      "learning_rate": 4.2117223341765645e-05,
+      "loss": 2.8987,
+      "step": 347500
+    },
+    {
+      "epoch": 14.669308266239515,
+      "grad_norm": 1.2731343507766724,
+      "learning_rate": 4.2031471080049736e-05,
+      "loss": 2.8997,
+      "step": 348000
+    },
+    {
+      "epoch": 14.690384858576065,
+      "grad_norm": 1.3010292053222656,
+      "learning_rate": 4.1945718818333834e-05,
+      "loss": 2.8995,
+      "step": 348500
+    },
+    {
+      "epoch": 14.711461450912616,
+      "grad_norm": 1.3210972547531128,
+      "learning_rate": 4.185996655661793e-05,
+      "loss": 2.9009,
+      "step": 349000
+    },
+    {
+      "epoch": 14.732538043249168,
+      "grad_norm": 1.2647473812103271,
+      "learning_rate": 4.177421429490203e-05,
+      "loss": 2.9017,
+      "step": 349500
+    },
+    {
+      "epoch": 14.753614635585718,
+      "grad_norm": 1.3216161727905273,
+      "learning_rate": 4.168846203318613e-05,
+      "loss": 2.8972,
+      "step": 350000
+    },
+    {
+      "epoch": 14.753614635585718,
+      "eval_accuracy": 0.7210088703388163,
+      "eval_loss": 2.6573455333709717,
+      "eval_runtime": 265.5325,
+      "eval_samples_per_second": 454.581,
+      "eval_steps_per_second": 4.738,
+      "step": 350000
+    },
+    {
+      "epoch": 14.77469122792227,
+      "grad_norm": 1.3596357107162476,
+      "learning_rate": 4.160270977147022e-05,
+      "loss": 2.8974,
+      "step": 350500
+    },
+    {
+      "epoch": 14.79576782025882,
+      "grad_norm": 1.288885474205017,
+      "learning_rate": 4.1516957509754326e-05,
+      "loss": 2.9029,
+      "step": 351000
+    },
+    {
+      "epoch": 14.81684441259537,
+      "grad_norm": 1.3190056085586548,
+      "learning_rate": 4.143120524803842e-05,
+      "loss": 2.8968,
+      "step": 351500
+    },
+    {
+      "epoch": 14.837921004931923,
+      "grad_norm": 1.3073593378067017,
+      "learning_rate": 4.1345452986322516e-05,
+      "loss": 2.8991,
+      "step": 352000
+    },
+    {
+      "epoch": 14.858997597268473,
+      "grad_norm": 1.2804540395736694,
+      "learning_rate": 4.1259700724606614e-05,
+      "loss": 2.8983,
+      "step": 352500
+    },
+    {
+      "epoch": 14.880074189605025,
+      "grad_norm": 1.3170182704925537,
+      "learning_rate": 4.117394846289071e-05,
+      "loss": 2.8983,
+      "step": 353000
+    },
+    {
+      "epoch": 14.901150781941576,
+      "grad_norm": 1.3471317291259766,
+      "learning_rate": 4.1088196201174804e-05,
+      "loss": 2.8987,
+      "step": 353500
+    },
+    {
+      "epoch": 14.922227374278126,
+      "grad_norm": 1.2909679412841797,
+      "learning_rate": 4.100244393945891e-05,
+      "loss": 2.8952,
+      "step": 354000
+    },
+    {
+      "epoch": 14.943303966614678,
+      "grad_norm": 1.2933170795440674,
+      "learning_rate": 4.0916691677743e-05,
+      "loss": 2.9024,
+      "step": 354500
+    },
+    {
+      "epoch": 14.964380558951229,
+      "grad_norm": 1.2709527015686035,
+      "learning_rate": 4.08309394160271e-05,
+      "loss": 2.8985,
+      "step": 355000
+    },
+    {
+      "epoch": 14.985457151287779,
+      "grad_norm": 1.2599327564239502,
+      "learning_rate": 4.07451871543112e-05,
+      "loss": 2.8981,
+      "step": 355500
+    },
+    {
+      "epoch": 15.006533743624331,
+      "grad_norm": 1.242160439491272,
+      "learning_rate": 4.0659434892595295e-05,
+      "loss": 2.8927,
+      "step": 356000
+    },
+    {
+      "epoch": 15.027610335960881,
+      "grad_norm": 1.2524168491363525,
+      "learning_rate": 4.057368263087939e-05,
+      "loss": 2.8896,
+      "step": 356500
+    },
+    {
+      "epoch": 15.048686928297434,
+      "grad_norm": 1.319894790649414,
+      "learning_rate": 4.048793036916349e-05,
+      "loss": 2.8941,
+      "step": 357000
+    },
+    {
+      "epoch": 15.069763520633984,
+      "grad_norm": 1.3710542917251587,
+      "learning_rate": 4.0402178107447584e-05,
+      "loss": 2.8957,
+      "step": 357500
+    },
+    {
+      "epoch": 15.090840112970534,
+      "grad_norm": 1.2753655910491943,
+      "learning_rate": 4.031642584573168e-05,
+      "loss": 2.8938,
+      "step": 358000
+    },
+    {
+      "epoch": 15.111916705307086,
+      "grad_norm": 1.369922161102295,
+      "learning_rate": 4.023067358401578e-05,
+      "loss": 2.8913,
+      "step": 358500
+    },
+    {
+      "epoch": 15.132993297643637,
+      "grad_norm": 1.3160110712051392,
+      "learning_rate": 4.014492132229988e-05,
+      "loss": 2.8948,
+      "step": 359000
+    },
+    {
+      "epoch": 15.154069889980189,
+      "grad_norm": 1.3168903589248657,
+      "learning_rate": 4.005916906058397e-05,
+      "loss": 2.8924,
+      "step": 359500
+    },
+    {
+      "epoch": 15.17514648231674,
+      "grad_norm": 1.2832598686218262,
+      "learning_rate": 3.9973416798868075e-05,
+      "loss": 2.8931,
+      "step": 360000
+    },
+    {
+      "epoch": 15.17514648231674,
+      "eval_accuracy": 0.7217353648043056,
+      "eval_loss": 2.654547691345215,
+      "eval_runtime": 266.7808,
+      "eval_samples_per_second": 452.454,
+      "eval_steps_per_second": 4.715,
+      "step": 360000
+    },
+    {
+      "epoch": 15.19622307465329,
+      "grad_norm": 1.3153085708618164,
+      "learning_rate": 3.988766453715217e-05,
+      "loss": 2.8947,
+      "step": 360500
+    },
+    {
+      "epoch": 15.217299666989842,
+      "grad_norm": 1.3812135457992554,
+      "learning_rate": 3.9801912275436265e-05,
+      "loss": 2.8946,
+      "step": 361000
+    },
+    {
+      "epoch": 15.238376259326392,
+      "grad_norm": 1.3311903476715088,
+      "learning_rate": 3.971616001372036e-05,
+      "loss": 2.8955,
+      "step": 361500
+    },
+    {
+      "epoch": 15.259452851662942,
+      "grad_norm": 1.2914338111877441,
+      "learning_rate": 3.963040775200446e-05,
+      "loss": 2.8935,
+      "step": 362000
+    },
+    {
+      "epoch": 15.280529443999495,
+      "grad_norm": 1.3398021459579468,
+      "learning_rate": 3.954465549028856e-05,
+      "loss": 2.8922,
+      "step": 362500
+    },
+    {
+      "epoch": 15.301606036336045,
+      "grad_norm": 1.2670217752456665,
+      "learning_rate": 3.945890322857266e-05,
+      "loss": 2.8932,
+      "step": 363000
+    },
+    {
+      "epoch": 15.322682628672597,
+      "grad_norm": 1.2349066734313965,
+      "learning_rate": 3.9373150966856756e-05,
+      "loss": 2.8893,
+      "step": 363500
+    },
+    {
+      "epoch": 15.343759221009147,
+      "grad_norm": 1.3394166231155396,
+      "learning_rate": 3.928739870514085e-05,
+      "loss": 2.8917,
+      "step": 364000
+    },
+    {
+      "epoch": 15.364835813345698,
+      "grad_norm": 1.287297248840332,
+      "learning_rate": 3.920164644342495e-05,
+      "loss": 2.8894,
+      "step": 364500
+    },
+    {
+      "epoch": 15.38591240568225,
+      "grad_norm": 1.3282909393310547,
+      "learning_rate": 3.9115894181709045e-05,
+      "loss": 2.8908,
+      "step": 365000
+    },
+    {
+      "epoch": 15.4069889980188,
+      "grad_norm": 1.3304091691970825,
+      "learning_rate": 3.903014191999314e-05,
+      "loss": 2.8867,
+      "step": 365500
+    },
+    {
+      "epoch": 15.42806559035535,
+      "grad_norm": 1.3689051866531372,
+      "learning_rate": 3.894438965827724e-05,
+      "loss": 2.8919,
+      "step": 366000
+    },
+    {
+      "epoch": 15.449142182691903,
+      "grad_norm": 1.2988474369049072,
+      "learning_rate": 3.885863739656134e-05,
+      "loss": 2.8913,
+      "step": 366500
+    },
+    {
+      "epoch": 15.470218775028453,
+      "grad_norm": 1.3081852197647095,
+      "learning_rate": 3.877288513484543e-05,
+      "loss": 2.8889,
+      "step": 367000
+    },
+    {
+      "epoch": 15.491295367365005,
+      "grad_norm": 1.2835369110107422,
+      "learning_rate": 3.868713287312953e-05,
+      "loss": 2.8912,
+      "step": 367500
+    },
+    {
+      "epoch": 15.512371959701555,
+      "grad_norm": 1.3159013986587524,
+      "learning_rate": 3.860138061141363e-05,
+      "loss": 2.8892,
+      "step": 368000
+    },
+    {
+      "epoch": 15.533448552038106,
+      "grad_norm": 1.3164845705032349,
+      "learning_rate": 3.8515628349697726e-05,
+      "loss": 2.8912,
+      "step": 368500
+    },
+    {
+      "epoch": 15.554525144374658,
+      "grad_norm": 1.2776457071304321,
+      "learning_rate": 3.842987608798182e-05,
+      "loss": 2.8897,
+      "step": 369000
+    },
+    {
+      "epoch": 15.575601736711208,
+      "grad_norm": 1.3433845043182373,
+      "learning_rate": 3.834412382626592e-05,
+      "loss": 2.8858,
+      "step": 369500
+    },
+    {
+      "epoch": 15.596678329047759,
+      "grad_norm": 1.2837656736373901,
+      "learning_rate": 3.8258371564550014e-05,
+      "loss": 2.8894,
+      "step": 370000
+    },
+    {
+      "epoch": 15.596678329047759,
+      "eval_accuracy": 0.7227235142043051,
+      "eval_loss": 2.6467647552490234,
+      "eval_runtime": 290.2025,
+      "eval_samples_per_second": 415.937,
+      "eval_steps_per_second": 4.335,
+      "step": 370000
+    },
+    {
+      "epoch": 15.61775492138431,
+      "grad_norm": 1.341591715812683,
+      "learning_rate": 3.817261930283411e-05,
+      "loss": 2.8867,
+      "step": 370500
+    },
+    {
+      "epoch": 15.638831513720861,
+      "grad_norm": 1.3581557273864746,
+      "learning_rate": 3.808686704111821e-05,
+      "loss": 2.8888,
+      "step": 371000
+    },
+    {
+      "epoch": 15.659908106057413,
+      "grad_norm": 1.2898932695388794,
+      "learning_rate": 3.800111477940231e-05,
+      "loss": 2.8895,
+      "step": 371500
+    },
+    {
+      "epoch": 15.680984698393964,
+      "grad_norm": 1.336448311805725,
+      "learning_rate": 3.79153625176864e-05,
+      "loss": 2.8914,
+      "step": 372000
+    },
+    {
+      "epoch": 15.702061290730514,
+      "grad_norm": 1.3165205717086792,
+      "learning_rate": 3.7829610255970506e-05,
+      "loss": 2.8913,
+      "step": 372500
+    },
+    {
+      "epoch": 15.723137883067066,
+      "grad_norm": 1.316488265991211,
+      "learning_rate": 3.77438579942546e-05,
+      "loss": 2.8864,
+      "step": 373000
+    },
+    {
+      "epoch": 15.744214475403616,
+      "grad_norm": 1.3213841915130615,
+      "learning_rate": 3.7658105732538696e-05,
+      "loss": 2.8839,
+      "step": 373500
+    },
+    {
+      "epoch": 15.765291067740169,
+      "grad_norm": 1.2756261825561523,
+      "learning_rate": 3.7572353470822794e-05,
+      "loss": 2.8887,
+      "step": 374000
+    },
+    {
+      "epoch": 15.786367660076719,
+      "grad_norm": 1.314042568206787,
+      "learning_rate": 3.748660120910689e-05,
+      "loss": 2.8846,
+      "step": 374500
+    },
+    {
+      "epoch": 15.80744425241327,
+      "grad_norm": 1.3297368288040161,
+      "learning_rate": 3.740084894739099e-05,
+      "loss": 2.8818,
+      "step": 375000
+    },
+    {
+      "epoch": 15.828520844749821,
+      "grad_norm": 1.270273208618164,
+      "learning_rate": 3.731509668567509e-05,
+      "loss": 2.8838,
+      "step": 375500
+    },
+    {
+      "epoch": 15.849597437086372,
+      "grad_norm": 1.3289670944213867,
+      "learning_rate": 3.722934442395919e-05,
+      "loss": 2.8885,
+      "step": 376000
+    },
+    {
+      "epoch": 15.870674029422922,
+      "grad_norm": 1.2824609279632568,
+      "learning_rate": 3.714359216224328e-05,
+      "loss": 2.8862,
+      "step": 376500
+    },
+    {
+      "epoch": 15.891750621759474,
+      "grad_norm": 1.3698564767837524,
+      "learning_rate": 3.705783990052738e-05,
+      "loss": 2.8875,
+      "step": 377000
+    },
+    {
+      "epoch": 15.912827214096025,
+      "grad_norm": 1.2759279012680054,
+      "learning_rate": 3.6972087638811475e-05,
+      "loss": 2.8826,
+      "step": 377500
+    },
+    {
+      "epoch": 15.933903806432577,
+      "grad_norm": 1.3354288339614868,
+      "learning_rate": 3.6886335377095574e-05,
+      "loss": 2.8885,
+      "step": 378000
+    },
+    {
+      "epoch": 15.954980398769127,
+      "grad_norm": 1.2605388164520264,
+      "learning_rate": 3.680058311537967e-05,
+      "loss": 2.885,
+      "step": 378500
+    },
+    {
+      "epoch": 15.976056991105677,
+      "grad_norm": 1.3370634317398071,
+      "learning_rate": 3.671483085366377e-05,
+      "loss": 2.887,
+      "step": 379000
+    },
+    {
+      "epoch": 15.99713358344223,
+      "grad_norm": 1.288521409034729,
+      "learning_rate": 3.662907859194786e-05,
+      "loss": 2.8817,
+      "step": 379500
+    },
+    {
+      "epoch": 16.01821017577878,
+      "grad_norm": 1.3859864473342896,
+      "learning_rate": 3.654332633023197e-05,
+      "loss": 2.8841,
+      "step": 380000
+    },
+    {
+      "epoch": 16.01821017577878,
+      "eval_accuracy": 0.7235369230801223,
+      "eval_loss": 2.6424920558929443,
+      "eval_runtime": 264.6356,
+      "eval_samples_per_second": 456.122,
+      "eval_steps_per_second": 4.754,
+      "step": 380000
+    },
+    {
+      "epoch": 16.039286768115332,
+      "grad_norm": 1.2224358320236206,
+      "learning_rate": 3.645757406851606e-05,
+      "loss": 2.8813,
+      "step": 380500
+    },
+    {
+      "epoch": 16.06036336045188,
+      "grad_norm": 1.2974895238876343,
+      "learning_rate": 3.6371821806800157e-05,
+      "loss": 2.88,
+      "step": 381000
+    },
+    {
+      "epoch": 16.081439952788433,
+      "grad_norm": 1.297402024269104,
+      "learning_rate": 3.6286069545084255e-05,
+      "loss": 2.8861,
+      "step": 381500
+    },
+    {
+      "epoch": 16.102516545124985,
+      "grad_norm": 1.3118854761123657,
+      "learning_rate": 3.620031728336835e-05,
+      "loss": 2.8819,
+      "step": 382000
+    },
+    {
+      "epoch": 16.123593137461537,
+      "grad_norm": 1.340611457824707,
+      "learning_rate": 3.6114565021652445e-05,
+      "loss": 2.8768,
+      "step": 382500
+    },
+    {
+      "epoch": 16.144669729798085,
+      "grad_norm": 1.3231691122055054,
+      "learning_rate": 3.602881275993655e-05,
+      "loss": 2.879,
+      "step": 383000
+    },
+    {
+      "epoch": 16.165746322134638,
+      "grad_norm": 1.3551805019378662,
+      "learning_rate": 3.594306049822064e-05,
+      "loss": 2.8844,
+      "step": 383500
+    },
+    {
+      "epoch": 16.18682291447119,
+      "grad_norm": 1.345295786857605,
+      "learning_rate": 3.585730823650474e-05,
+      "loss": 2.88,
+      "step": 384000
+    },
+    {
+      "epoch": 16.20789950680774,
+      "grad_norm": 1.2727906703948975,
+      "learning_rate": 3.577155597478883e-05,
+      "loss": 2.8816,
+      "step": 384500
+    },
+    {
+      "epoch": 16.22897609914429,
+      "grad_norm": 1.3263827562332153,
+      "learning_rate": 3.5685803713072936e-05,
+      "loss": 2.8757,
+      "step": 385000
+    },
+    {
+      "epoch": 16.250052691480843,
+      "grad_norm": 1.2888448238372803,
+      "learning_rate": 3.560005145135703e-05,
+      "loss": 2.8802,
+      "step": 385500
+    },
+    {
+      "epoch": 16.27112928381739,
+      "grad_norm": 1.3559516668319702,
+      "learning_rate": 3.5514299189641126e-05,
+      "loss": 2.8767,
+      "step": 386000
+    },
+    {
+      "epoch": 16.292205876153943,
+      "grad_norm": 1.3249772787094116,
+      "learning_rate": 3.5428546927925224e-05,
+      "loss": 2.8825,
+      "step": 386500
+    },
+    {
+      "epoch": 16.313282468490495,
+      "grad_norm": 1.3084138631820679,
+      "learning_rate": 3.534279466620932e-05,
+      "loss": 2.8784,
+      "step": 387000
+    },
+    {
+      "epoch": 16.334359060827044,
+      "grad_norm": 1.297060489654541,
+      "learning_rate": 3.525704240449342e-05,
+      "loss": 2.8793,
+      "step": 387500
+    },
+    {
+      "epoch": 16.355435653163596,
+      "grad_norm": 1.307222843170166,
+      "learning_rate": 3.517129014277752e-05,
+      "loss": 2.878,
+      "step": 388000
+    },
+    {
+      "epoch": 16.376512245500148,
+      "grad_norm": 1.322837233543396,
+      "learning_rate": 3.508553788106161e-05,
+      "loss": 2.8778,
+      "step": 388500
+    },
+    {
+      "epoch": 16.3975888378367,
+      "grad_norm": 1.3846923112869263,
+      "learning_rate": 3.499978561934571e-05,
+      "loss": 2.8765,
+      "step": 389000
+    },
+    {
+      "epoch": 16.41866543017325,
+      "grad_norm": 1.2573078870773315,
+      "learning_rate": 3.491403335762981e-05,
+      "loss": 2.8769,
+      "step": 389500
+    },
+    {
+      "epoch": 16.4397420225098,
+      "grad_norm": 1.366938591003418,
+      "learning_rate": 3.4828281095913906e-05,
+      "loss": 2.8799,
+      "step": 390000
+    },
+    {
+      "epoch": 16.4397420225098,
+      "eval_accuracy": 0.7241451248097234,
+      "eval_loss": 2.638646364212036,
+      "eval_runtime": 257.8962,
+      "eval_samples_per_second": 468.041,
+      "eval_steps_per_second": 4.878,
+      "step": 390000
+    },
+    {
+      "epoch": 16.460818614846353,
+      "grad_norm": 1.3838715553283691,
+      "learning_rate": 3.4742528834198004e-05,
+      "loss": 2.877,
+      "step": 390500
+    },
+    {
+      "epoch": 16.4818952071829,
+      "grad_norm": 1.3040562868118286,
+      "learning_rate": 3.46567765724821e-05,
+      "loss": 2.8794,
+      "step": 391000
+    },
+    {
+      "epoch": 16.502971799519454,
+      "grad_norm": 1.3105566501617432,
+      "learning_rate": 3.45710243107662e-05,
+      "loss": 2.8787,
+      "step": 391500
+    },
+    {
+      "epoch": 16.524048391856006,
+      "grad_norm": 1.3421549797058105,
+      "learning_rate": 3.448527204905029e-05,
+      "loss": 2.8816,
+      "step": 392000
+    },
+    {
+      "epoch": 16.545124984192555,
+      "grad_norm": 1.292428731918335,
+      "learning_rate": 3.43995197873344e-05,
+      "loss": 2.8747,
+      "step": 392500
+    },
+    {
+      "epoch": 16.566201576529107,
+      "grad_norm": 1.2973028421401978,
+      "learning_rate": 3.431376752561849e-05,
+      "loss": 2.8779,
+      "step": 393000
+    },
+    {
+      "epoch": 16.58727816886566,
+      "grad_norm": 1.3080816268920898,
+      "learning_rate": 3.422801526390259e-05,
+      "loss": 2.8792,
+      "step": 393500
+    },
+    {
+      "epoch": 16.608354761202207,
+      "grad_norm": 1.2986663579940796,
+      "learning_rate": 3.4142263002186686e-05,
+      "loss": 2.8786,
+      "step": 394000
+    },
+    {
+      "epoch": 16.62943135353876,
+      "grad_norm": 1.3106896877288818,
+      "learning_rate": 3.4056510740470784e-05,
+      "loss": 2.8827,
+      "step": 394500
+    },
+    {
+      "epoch": 16.65050794587531,
+      "grad_norm": 1.4124975204467773,
+      "learning_rate": 3.3970758478754875e-05,
+      "loss": 2.8797,
+      "step": 395000
+    },
+    {
+      "epoch": 16.67158453821186,
+      "grad_norm": 1.282852053642273,
+      "learning_rate": 3.388500621703898e-05,
+      "loss": 2.8782,
+      "step": 395500
+    },
+    {
+      "epoch": 16.692661130548412,
+      "grad_norm": 1.3497352600097656,
+      "learning_rate": 3.379925395532307e-05,
+      "loss": 2.8786,
+      "step": 396000
+    },
+    {
+      "epoch": 16.713737722884964,
+      "grad_norm": 1.2739773988723755,
+      "learning_rate": 3.371350169360717e-05,
+      "loss": 2.8747,
+      "step": 396500
+    },
+    {
+      "epoch": 16.734814315221517,
+      "grad_norm": 1.3375540971755981,
+      "learning_rate": 3.362774943189127e-05,
+      "loss": 2.8782,
+      "step": 397000
+    },
+    {
+      "epoch": 16.755890907558065,
+      "grad_norm": 1.240104079246521,
+      "learning_rate": 3.354199717017537e-05,
+      "loss": 2.8751,
+      "step": 397500
+    },
+    {
+      "epoch": 16.776967499894617,
+      "grad_norm": 1.334345817565918,
+      "learning_rate": 3.345624490845946e-05,
+      "loss": 2.8758,
+      "step": 398000
+    },
+    {
+      "epoch": 16.79804409223117,
+      "grad_norm": 1.2961493730545044,
+      "learning_rate": 3.3370492646743564e-05,
+      "loss": 2.8724,
+      "step": 398500
+    },
+    {
+      "epoch": 16.819120684567718,
+      "grad_norm": 1.2979934215545654,
+      "learning_rate": 3.3284740385027655e-05,
+      "loss": 2.8798,
+      "step": 399000
+    },
+    {
+      "epoch": 16.84019727690427,
+      "grad_norm": 1.2887928485870361,
+      "learning_rate": 3.319898812331175e-05,
+      "loss": 2.8719,
+      "step": 399500
+    },
+    {
+      "epoch": 16.861273869240822,
+      "grad_norm": 1.324894905090332,
+      "learning_rate": 3.311323586159585e-05,
+      "loss": 2.8742,
+      "step": 400000
+    },
+    {
+      "epoch": 16.861273869240822,
+      "eval_accuracy": 0.7245414193023823,
+      "eval_loss": 2.6369874477386475,
+      "eval_runtime": 259.0148,
+      "eval_samples_per_second": 466.02,
+      "eval_steps_per_second": 4.857,
+      "step": 400000
+    },
+    {
+      "epoch": 16.88235046157737,
+      "grad_norm": 1.3220235109329224,
+      "learning_rate": 3.302748359987995e-05,
+      "loss": 2.8711,
+      "step": 400500
+    },
+    {
+      "epoch": 16.903427053913923,
+      "grad_norm": 1.2922253608703613,
+      "learning_rate": 3.294173133816404e-05,
+      "loss": 2.8734,
+      "step": 401000
+    },
+    {
+      "epoch": 16.924503646250475,
+      "grad_norm": 1.3849409818649292,
+      "learning_rate": 3.2855979076448147e-05,
+      "loss": 2.8779,
+      "step": 401500
+    },
+    {
+      "epoch": 16.945580238587024,
+      "grad_norm": 1.3035324811935425,
+      "learning_rate": 3.277022681473224e-05,
+      "loss": 2.8759,
+      "step": 402000
+    },
+    {
+      "epoch": 16.966656830923576,
+      "grad_norm": 1.340030312538147,
+      "learning_rate": 3.2684474553016336e-05,
+      "loss": 2.8739,
+      "step": 402500
+    },
+    {
+      "epoch": 16.987733423260128,
+      "grad_norm": 1.3383108377456665,
+      "learning_rate": 3.2598722291300435e-05,
+      "loss": 2.8766,
+      "step": 403000
+    },
+    {
+      "epoch": 17.00881001559668,
+      "grad_norm": 1.3135006427764893,
+      "learning_rate": 3.251297002958453e-05,
+      "loss": 2.8746,
+      "step": 403500
+    },
+    {
+      "epoch": 17.02988660793323,
+      "grad_norm": 1.3420954942703247,
+      "learning_rate": 3.242721776786863e-05,
+      "loss": 2.8735,
+      "step": 404000
+    },
+    {
+      "epoch": 17.05096320026978,
+      "grad_norm": 1.3027710914611816,
+      "learning_rate": 3.234146550615272e-05,
+      "loss": 2.8695,
+      "step": 404500
+    },
+    {
+      "epoch": 17.072039792606333,
+      "grad_norm": 1.3091400861740112,
+      "learning_rate": 3.225571324443683e-05,
+      "loss": 2.872,
+      "step": 405000
+    },
+    {
+      "epoch": 17.09311638494288,
+      "grad_norm": 1.3795387744903564,
+      "learning_rate": 3.216996098272092e-05,
+      "loss": 2.8711,
+      "step": 405500
+    },
+    {
+      "epoch": 17.114192977279433,
+      "grad_norm": 1.2876386642456055,
+      "learning_rate": 3.208420872100502e-05,
+      "loss": 2.872,
+      "step": 406000
+    },
+    {
+      "epoch": 17.135269569615986,
+      "grad_norm": 1.2950193881988525,
+      "learning_rate": 3.1998456459289116e-05,
+      "loss": 2.8671,
+      "step": 406500
+    },
+    {
+      "epoch": 17.156346161952534,
+      "grad_norm": 1.2639837265014648,
+      "learning_rate": 3.1912704197573214e-05,
+      "loss": 2.8675,
+      "step": 407000
+    },
+    {
+      "epoch": 17.177422754289086,
+      "grad_norm": 1.324753761291504,
+      "learning_rate": 3.1826951935857306e-05,
+      "loss": 2.8674,
+      "step": 407500
+    },
+    {
+      "epoch": 17.19849934662564,
+      "grad_norm": 1.3375465869903564,
+      "learning_rate": 3.174119967414141e-05,
+      "loss": 2.8729,
+      "step": 408000
+    },
+    {
+      "epoch": 17.219575938962187,
+      "grad_norm": 1.3663063049316406,
+      "learning_rate": 3.16554474124255e-05,
+      "loss": 2.8714,
+      "step": 408500
+    },
+    {
+      "epoch": 17.24065253129874,
+      "grad_norm": 1.3356181383132935,
+      "learning_rate": 3.15696951507096e-05,
+      "loss": 2.8711,
+      "step": 409000
+    },
+    {
+      "epoch": 17.26172912363529,
+      "grad_norm": 1.3394851684570312,
+      "learning_rate": 3.14839428889937e-05,
+      "loss": 2.8696,
+      "step": 409500
+    },
+    {
+      "epoch": 17.282805715971843,
+      "grad_norm": 1.3374103307724,
+      "learning_rate": 3.13981906272778e-05,
+      "loss": 2.8716,
+      "step": 410000
+    },
+    {
+      "epoch": 17.282805715971843,
+      "eval_accuracy": 0.7255051500412257,
+      "eval_loss": 2.630141258239746,
+      "eval_runtime": 281.0333,
+      "eval_samples_per_second": 429.508,
+      "eval_steps_per_second": 4.476,
+      "step": 410000
+    },
+    {
+      "epoch": 17.303882308308392,
+      "grad_norm": 1.373040795326233,
+      "learning_rate": 3.131243836556189e-05,
+      "loss": 2.8726,
+      "step": 410500
+    },
+    {
+      "epoch": 17.324958900644944,
+      "grad_norm": 1.4642248153686523,
+      "learning_rate": 3.1226686103845994e-05,
+      "loss": 2.8659,
+      "step": 411000
+    },
+    {
+      "epoch": 17.346035492981496,
+      "grad_norm": 1.3126777410507202,
+      "learning_rate": 3.1140933842130086e-05,
+      "loss": 2.868,
+      "step": 411500
+    },
+    {
+      "epoch": 17.367112085318045,
+      "grad_norm": 1.34221351146698,
+      "learning_rate": 3.1055181580414184e-05,
+      "loss": 2.8716,
+      "step": 412000
+    },
+    {
+      "epoch": 17.388188677654597,
+      "grad_norm": 1.3335522413253784,
+      "learning_rate": 3.096942931869828e-05,
+      "loss": 2.8725,
+      "step": 412500
+    },
+    {
+      "epoch": 17.40926526999115,
+      "grad_norm": 1.323869228363037,
+      "learning_rate": 3.088367705698238e-05,
+      "loss": 2.8667,
+      "step": 413000
+    },
+    {
+      "epoch": 17.430341862327698,
+      "grad_norm": 1.3333135843276978,
+      "learning_rate": 3.079792479526647e-05,
+      "loss": 2.8686,
+      "step": 413500
+    },
+    {
+      "epoch": 17.45141845466425,
+      "grad_norm": 1.2785052061080933,
+      "learning_rate": 3.071217253355058e-05,
+      "loss": 2.8676,
+      "step": 414000
+    },
+    {
+      "epoch": 17.472495047000802,
+      "grad_norm": 1.3548232316970825,
+      "learning_rate": 3.062642027183467e-05,
+      "loss": 2.8671,
+      "step": 414500
+    },
+    {
+      "epoch": 17.49357163933735,
+      "grad_norm": 1.3674266338348389,
+      "learning_rate": 3.054066801011877e-05,
+      "loss": 2.868,
+      "step": 415000
+    },
+    {
+      "epoch": 17.514648231673903,
+      "grad_norm": 1.356676459312439,
+      "learning_rate": 3.045491574840287e-05,
+      "loss": 2.8686,
+      "step": 415500
+    },
+    {
+      "epoch": 17.535724824010455,
+      "grad_norm": 1.3202053308486938,
+      "learning_rate": 3.0369163486686964e-05,
+      "loss": 2.8675,
+      "step": 416000
+    },
+    {
+      "epoch": 17.556801416347007,
+      "grad_norm": 1.3285133838653564,
+      "learning_rate": 3.028341122497106e-05,
+      "loss": 2.8663,
+      "step": 416500
+    },
+    {
+      "epoch": 17.577878008683555,
+      "grad_norm": 1.3568185567855835,
+      "learning_rate": 3.019765896325516e-05,
+      "loss": 2.8701,
+      "step": 417000
+    },
+    {
+      "epoch": 17.598954601020107,
+      "grad_norm": 1.3564941883087158,
+      "learning_rate": 3.0111906701539255e-05,
+      "loss": 2.8696,
+      "step": 417500
+    },
+    {
+      "epoch": 17.62003119335666,
+      "grad_norm": 1.334836483001709,
+      "learning_rate": 3.002615443982335e-05,
+      "loss": 2.8666,
+      "step": 418000
+    },
+    {
+      "epoch": 17.641107785693208,
+      "grad_norm": 1.3547743558883667,
+      "learning_rate": 2.9940402178107452e-05,
+      "loss": 2.8667,
+      "step": 418500
+    },
+    {
+      "epoch": 17.66218437802976,
+      "grad_norm": 1.4004862308502197,
+      "learning_rate": 2.9854649916391547e-05,
+      "loss": 2.8641,
+      "step": 419000
+    },
+    {
+      "epoch": 17.683260970366312,
+      "grad_norm": 1.3627029657363892,
+      "learning_rate": 2.976889765467564e-05,
+      "loss": 2.8709,
+      "step": 419500
+    },
+    {
+      "epoch": 17.70433756270286,
+      "grad_norm": 1.3340140581130981,
+      "learning_rate": 2.9683145392959737e-05,
+      "loss": 2.8658,
+      "step": 420000
+    },
+    {
+      "epoch": 17.70433756270286,
+      "eval_accuracy": 0.7263263101211435,
+      "eval_loss": 2.626755714416504,
+      "eval_runtime": 268.9503,
+      "eval_samples_per_second": 448.804,
+      "eval_steps_per_second": 4.677,
+      "step": 420000
+    },
+    {
+      "epoch": 17.725414155039413,
+      "grad_norm": 1.338518738746643,
+      "learning_rate": 2.9597393131243838e-05,
+      "loss": 2.8682,
+      "step": 420500
+    },
+    {
+      "epoch": 17.746490747375965,
+      "grad_norm": 1.318394660949707,
+      "learning_rate": 2.9511640869527933e-05,
+      "loss": 2.8648,
+      "step": 421000
+    },
+    {
+      "epoch": 17.767567339712514,
+      "grad_norm": 1.4009883403778076,
+      "learning_rate": 2.942588860781203e-05,
+      "loss": 2.8647,
+      "step": 421500
+    },
+    {
+      "epoch": 17.788643932049066,
+      "grad_norm": 1.3165916204452515,
+      "learning_rate": 2.934013634609613e-05,
+      "loss": 2.8674,
+      "step": 422000
+    },
+    {
+      "epoch": 17.809720524385618,
+      "grad_norm": 1.2963992357254028,
+      "learning_rate": 2.9254384084380225e-05,
+      "loss": 2.867,
+      "step": 422500
+    },
+    {
+      "epoch": 17.830797116722167,
+      "grad_norm": 1.3340580463409424,
+      "learning_rate": 2.9168631822664323e-05,
+      "loss": 2.8668,
+      "step": 423000
+    },
+    {
+      "epoch": 17.85187370905872,
+      "grad_norm": 1.2974956035614014,
+      "learning_rate": 2.908287956094842e-05,
+      "loss": 2.8668,
+      "step": 423500
+    },
+    {
+      "epoch": 17.87295030139527,
+      "grad_norm": 1.364051342010498,
+      "learning_rate": 2.899712729923252e-05,
+      "loss": 2.8669,
+      "step": 424000
+    },
+    {
+      "epoch": 17.894026893731823,
+      "grad_norm": 1.3173972368240356,
+      "learning_rate": 2.8911375037516615e-05,
+      "loss": 2.8649,
+      "step": 424500
+    },
+    {
+      "epoch": 17.91510348606837,
+      "grad_norm": 1.3576480150222778,
+      "learning_rate": 2.8825622775800716e-05,
+      "loss": 2.8681,
+      "step": 425000
+    },
+    {
+      "epoch": 17.936180078404924,
+      "grad_norm": 1.3676549196243286,
+      "learning_rate": 2.873987051408481e-05,
+      "loss": 2.866,
+      "step": 425500
+    },
+    {
+      "epoch": 17.957256670741476,
+      "grad_norm": 1.2932003736495972,
+      "learning_rate": 2.8654118252368906e-05,
+      "loss": 2.8639,
+      "step": 426000
+    },
+    {
+      "epoch": 17.978333263078024,
+      "grad_norm": 1.3299518823623657,
+      "learning_rate": 2.8568365990653008e-05,
+      "loss": 2.8624,
+      "step": 426500
+    },
+    {
+      "epoch": 17.999409855414576,
+      "grad_norm": 1.2852784395217896,
+      "learning_rate": 2.8482613728937103e-05,
+      "loss": 2.867,
+      "step": 427000
+    },
+    {
+      "epoch": 18.02048644775113,
+      "grad_norm": 1.314225673675537,
+      "learning_rate": 2.8396861467221198e-05,
+      "loss": 2.863,
+      "step": 427500
+    },
+    {
+      "epoch": 18.041563040087677,
+      "grad_norm": 1.3626540899276733,
+      "learning_rate": 2.83111092055053e-05,
+      "loss": 2.8598,
+      "step": 428000
+    },
+    {
+      "epoch": 18.06263963242423,
+      "grad_norm": 1.3211498260498047,
+      "learning_rate": 2.8225356943789394e-05,
+      "loss": 2.8617,
+      "step": 428500
+    },
+    {
+      "epoch": 18.08371622476078,
+      "grad_norm": 1.4028791189193726,
+      "learning_rate": 2.813960468207349e-05,
+      "loss": 2.8603,
+      "step": 429000
+    },
+    {
+      "epoch": 18.10479281709733,
+      "grad_norm": 1.3236751556396484,
+      "learning_rate": 2.805385242035759e-05,
+      "loss": 2.8607,
+      "step": 429500
+    },
+    {
+      "epoch": 18.125869409433882,
+      "grad_norm": 1.3654799461364746,
+      "learning_rate": 2.7968100158641686e-05,
+      "loss": 2.8605,
+      "step": 430000
+    },
+    {
+      "epoch": 18.125869409433882,
+      "eval_accuracy": 0.7266077671842841,
+      "eval_loss": 2.6263086795806885,
+      "eval_runtime": 283.5706,
+      "eval_samples_per_second": 425.665,
+      "eval_steps_per_second": 4.436,
+      "step": 430000
+    },
+    {
+      "epoch": 18.146946001770434,
+      "grad_norm": 1.437875509262085,
+      "learning_rate": 2.788234789692578e-05,
+      "loss": 2.8617,
+      "step": 430500
+    },
+    {
+      "epoch": 18.168022594106986,
+      "grad_norm": 1.3722906112670898,
+      "learning_rate": 2.7796595635209882e-05,
+      "loss": 2.8609,
+      "step": 431000
+    },
+    {
+      "epoch": 18.189099186443535,
+      "grad_norm": 1.3127819299697876,
+      "learning_rate": 2.7710843373493977e-05,
+      "loss": 2.8631,
+      "step": 431500
+    },
+    {
+      "epoch": 18.210175778780087,
+      "grad_norm": 1.343760371208191,
+      "learning_rate": 2.7625091111778072e-05,
+      "loss": 2.8642,
+      "step": 432000
+    },
+    {
+      "epoch": 18.23125237111664,
+      "grad_norm": 1.3859531879425049,
+      "learning_rate": 2.7539338850062174e-05,
+      "loss": 2.8611,
+      "step": 432500
+    },
+    {
+      "epoch": 18.252328963453188,
+      "grad_norm": 1.3154531717300415,
+      "learning_rate": 2.745358658834627e-05,
+      "loss": 2.8618,
+      "step": 433000
+    },
+    {
+      "epoch": 18.27340555578974,
+      "grad_norm": 1.347521424293518,
+      "learning_rate": 2.7367834326630364e-05,
+      "loss": 2.8585,
+      "step": 433500
+    },
+    {
+      "epoch": 18.294482148126292,
+      "grad_norm": 1.3115208148956299,
+      "learning_rate": 2.7282082064914465e-05,
+      "loss": 2.8602,
+      "step": 434000
+    },
+    {
+      "epoch": 18.31555874046284,
+      "grad_norm": 1.3687646389007568,
+      "learning_rate": 2.719632980319856e-05,
+      "loss": 2.8596,
+      "step": 434500
+    },
+    {
+      "epoch": 18.336635332799393,
+      "grad_norm": 1.3091979026794434,
+      "learning_rate": 2.7110577541482655e-05,
+      "loss": 2.8589,
+      "step": 435000
+    },
+    {
+      "epoch": 18.357711925135945,
+      "grad_norm": 1.3261314630508423,
+      "learning_rate": 2.7024825279766757e-05,
+      "loss": 2.8623,
+      "step": 435500
+    },
+    {
+      "epoch": 18.378788517472493,
+      "grad_norm": 1.3855969905853271,
+      "learning_rate": 2.6939073018050852e-05,
+      "loss": 2.8593,
+      "step": 436000
+    },
+    {
+      "epoch": 18.399865109809046,
+      "grad_norm": 1.2912874221801758,
+      "learning_rate": 2.685332075633495e-05,
+      "loss": 2.8612,
+      "step": 436500
+    },
+    {
+      "epoch": 18.420941702145598,
+      "grad_norm": 1.3170056343078613,
+      "learning_rate": 2.6767568494619045e-05,
+      "loss": 2.8595,
+      "step": 437000
+    },
+    {
+      "epoch": 18.44201829448215,
+      "grad_norm": 1.3584718704223633,
+      "learning_rate": 2.6681816232903143e-05,
+      "loss": 2.8574,
+      "step": 437500
+    },
+    {
+      "epoch": 18.4630948868187,
+      "grad_norm": 1.3759140968322754,
+      "learning_rate": 2.6596063971187242e-05,
+      "loss": 2.8548,
+      "step": 438000
+    },
+    {
+      "epoch": 18.48417147915525,
+      "grad_norm": 1.4356215000152588,
+      "learning_rate": 2.6510311709471337e-05,
+      "loss": 2.8572,
+      "step": 438500
+    },
+    {
+      "epoch": 18.505248071491803,
+      "grad_norm": 1.3572279214859009,
+      "learning_rate": 2.642455944775544e-05,
+      "loss": 2.8587,
+      "step": 439000
+    },
+    {
+      "epoch": 18.52632466382835,
+      "grad_norm": 1.3160021305084229,
+      "learning_rate": 2.6338807186039533e-05,
+      "loss": 2.8569,
+      "step": 439500
+    },
+    {
+      "epoch": 18.547401256164903,
+      "grad_norm": 1.2881505489349365,
+      "learning_rate": 2.6253054924323628e-05,
+      "loss": 2.8549,
+      "step": 440000
+    },
+    {
+      "epoch": 18.547401256164903,
+      "eval_accuracy": 0.726844270538293,
+      "eval_loss": 2.623302459716797,
+      "eval_runtime": 334.434,
+      "eval_samples_per_second": 360.926,
+      "eval_steps_per_second": 3.762,
+      "step": 440000
+    },
+    {
+      "epoch": 18.568477848501455,
+      "grad_norm": 1.3418583869934082,
+      "learning_rate": 2.616730266260773e-05,
+      "loss": 2.8549,
+      "step": 440500
+    },
+    {
+      "epoch": 18.589554440838004,
+      "grad_norm": 1.3473259210586548,
+      "learning_rate": 2.6081550400891825e-05,
+      "loss": 2.8587,
+      "step": 441000
+    },
+    {
+      "epoch": 18.610631033174556,
+      "grad_norm": 1.313097596168518,
+      "learning_rate": 2.599579813917592e-05,
+      "loss": 2.8598,
+      "step": 441500
+    },
+    {
+      "epoch": 18.63170762551111,
+      "grad_norm": 1.3118226528167725,
+      "learning_rate": 2.591004587746002e-05,
+      "loss": 2.8576,
+      "step": 442000
+    },
+    {
+      "epoch": 18.652784217847657,
+      "grad_norm": 1.3428939580917358,
+      "learning_rate": 2.5824293615744116e-05,
+      "loss": 2.8562,
+      "step": 442500
+    },
+    {
+      "epoch": 18.67386081018421,
+      "grad_norm": 1.3617255687713623,
+      "learning_rate": 2.573854135402821e-05,
+      "loss": 2.858,
+      "step": 443000
+    },
+    {
+      "epoch": 18.69493740252076,
+      "grad_norm": 1.3715518712997437,
+      "learning_rate": 2.5652789092312313e-05,
+      "loss": 2.8568,
+      "step": 443500
+    },
+    {
+      "epoch": 18.716013994857313,
+      "grad_norm": 1.3365168571472168,
+      "learning_rate": 2.5567036830596408e-05,
+      "loss": 2.8571,
+      "step": 444000
+    },
+    {
+      "epoch": 18.737090587193862,
+      "grad_norm": 1.3530609607696533,
+      "learning_rate": 2.5481284568880503e-05,
+      "loss": 2.8581,
+      "step": 444500
+    },
+    {
+      "epoch": 18.758167179530414,
+      "grad_norm": 1.3307987451553345,
+      "learning_rate": 2.5395532307164605e-05,
+      "loss": 2.8535,
+      "step": 445000
+    },
+    {
+      "epoch": 18.779243771866966,
+      "grad_norm": 1.3699842691421509,
+      "learning_rate": 2.53097800454487e-05,
+      "loss": 2.8613,
+      "step": 445500
+    },
+    {
+      "epoch": 18.800320364203515,
+      "grad_norm": 1.3295459747314453,
+      "learning_rate": 2.5224027783732794e-05,
+      "loss": 2.8583,
+      "step": 446000
+    },
+    {
+      "epoch": 18.821396956540067,
+      "grad_norm": 1.3930593729019165,
+      "learning_rate": 2.5138275522016896e-05,
+      "loss": 2.849,
+      "step": 446500
+    },
+    {
+      "epoch": 18.84247354887662,
+      "grad_norm": 1.3761438131332397,
+      "learning_rate": 2.505252326030099e-05,
+      "loss": 2.8539,
+      "step": 447000
+    },
+    {
+      "epoch": 18.863550141213167,
+      "grad_norm": 1.351129412651062,
+      "learning_rate": 2.496677099858509e-05,
+      "loss": 2.8537,
+      "step": 447500
+    },
+    {
+      "epoch": 18.88462673354972,
+      "grad_norm": 1.3347859382629395,
+      "learning_rate": 2.4881018736869184e-05,
+      "loss": 2.8565,
+      "step": 448000
+    },
+    {
+      "epoch": 18.90570332588627,
+      "grad_norm": 1.3749622106552124,
+      "learning_rate": 2.4795266475153283e-05,
+      "loss": 2.8539,
+      "step": 448500
+    },
+    {
+      "epoch": 18.92677991822282,
+      "grad_norm": 1.362196922302246,
+      "learning_rate": 2.470951421343738e-05,
+      "loss": 2.8585,
+      "step": 449000
+    },
+    {
+      "epoch": 18.947856510559372,
+      "grad_norm": 1.3221112489700317,
+      "learning_rate": 2.4623761951721476e-05,
+      "loss": 2.8522,
+      "step": 449500
+    },
+    {
+      "epoch": 18.968933102895924,
+      "grad_norm": 1.3304688930511475,
+      "learning_rate": 2.4538009690005574e-05,
+      "loss": 2.8554,
+      "step": 450000
+    },
+    {
+      "epoch": 18.968933102895924,
+      "eval_accuracy": 0.7280531145267312,
+      "eval_loss": 2.6174681186676025,
+      "eval_runtime": 282.0473,
+      "eval_samples_per_second": 427.964,
+      "eval_steps_per_second": 4.46,
+      "step": 450000
+    },
+    {
+      "epoch": 18.990009695232473,
+      "grad_norm": 1.3827803134918213,
+      "learning_rate": 2.4452257428289672e-05,
+      "loss": 2.8516,
+      "step": 450500
+    },
+    {
+      "epoch": 19.011086287569025,
+      "grad_norm": 1.4003337621688843,
+      "learning_rate": 2.436650516657377e-05,
+      "loss": 2.8523,
+      "step": 451000
+    },
+    {
+      "epoch": 19.032162879905577,
+      "grad_norm": 1.386273741722107,
+      "learning_rate": 2.428075290485787e-05,
+      "loss": 2.8511,
+      "step": 451500
+    },
+    {
+      "epoch": 19.05323947224213,
+      "grad_norm": 1.3303699493408203,
+      "learning_rate": 2.4195000643141964e-05,
+      "loss": 2.8528,
+      "step": 452000
+    },
+    {
+      "epoch": 19.074316064578678,
+      "grad_norm": 1.3797391653060913,
+      "learning_rate": 2.4109248381426062e-05,
+      "loss": 2.8539,
+      "step": 452500
+    },
+    {
+      "epoch": 19.09539265691523,
+      "grad_norm": 1.3437832593917847,
+      "learning_rate": 2.402349611971016e-05,
+      "loss": 2.8533,
+      "step": 453000
+    },
+    {
+      "epoch": 19.116469249251782,
+      "grad_norm": 1.361084222793579,
+      "learning_rate": 2.3937743857994255e-05,
+      "loss": 2.8505,
+      "step": 453500
+    },
+    {
+      "epoch": 19.13754584158833,
+      "grad_norm": 1.3734490871429443,
+      "learning_rate": 2.3851991596278354e-05,
+      "loss": 2.8541,
+      "step": 454000
+    },
+    {
+      "epoch": 19.158622433924883,
+      "grad_norm": 1.3818408250808716,
+      "learning_rate": 2.3766239334562452e-05,
+      "loss": 2.849,
+      "step": 454500
+    },
+    {
+      "epoch": 19.179699026261435,
+      "grad_norm": 1.3607393503189087,
+      "learning_rate": 2.3680487072846547e-05,
+      "loss": 2.8486,
+      "step": 455000
+    },
+    {
+      "epoch": 19.200775618597984,
+      "grad_norm": 1.4176721572875977,
+      "learning_rate": 2.3594734811130645e-05,
+      "loss": 2.8481,
+      "step": 455500
+    },
+    {
+      "epoch": 19.221852210934536,
+      "grad_norm": 1.3234362602233887,
+      "learning_rate": 2.3508982549414744e-05,
+      "loss": 2.8484,
+      "step": 456000
+    },
+    {
+      "epoch": 19.242928803271088,
+      "grad_norm": 1.3334853649139404,
+      "learning_rate": 2.342323028769884e-05,
+      "loss": 2.8523,
+      "step": 456500
+    },
+    {
+      "epoch": 19.264005395607636,
+      "grad_norm": 1.3608765602111816,
+      "learning_rate": 2.3337478025982937e-05,
+      "loss": 2.8492,
+      "step": 457000
+    },
+    {
+      "epoch": 19.28508198794419,
+      "grad_norm": 1.4139153957366943,
+      "learning_rate": 2.3251725764267035e-05,
+      "loss": 2.8502,
+      "step": 457500
+    },
+    {
+      "epoch": 19.30615858028074,
+      "grad_norm": 1.3465907573699951,
+      "learning_rate": 2.316597350255113e-05,
+      "loss": 2.853,
+      "step": 458000
+    },
+    {
+      "epoch": 19.327235172617293,
+      "grad_norm": 1.3434823751449585,
+      "learning_rate": 2.308022124083523e-05,
+      "loss": 2.8501,
+      "step": 458500
+    },
+    {
+      "epoch": 19.34831176495384,
+      "grad_norm": 1.383257508277893,
+      "learning_rate": 2.2994468979119323e-05,
+      "loss": 2.853,
+      "step": 459000
+    },
+    {
+      "epoch": 19.369388357290394,
+      "grad_norm": 1.4407522678375244,
+      "learning_rate": 2.290871671740342e-05,
+      "loss": 2.85,
+      "step": 459500
+    },
+    {
+      "epoch": 19.390464949626946,
+      "grad_norm": 1.3973983526229858,
+      "learning_rate": 2.282296445568752e-05,
+      "loss": 2.8499,
+      "step": 460000
+    },
+    {
+      "epoch": 19.390464949626946,
+      "eval_accuracy": 0.7285532444472279,
+      "eval_loss": 2.6140990257263184,
+      "eval_runtime": 269.0556,
+      "eval_samples_per_second": 448.629,
+      "eval_steps_per_second": 4.676,
+      "step": 460000
+    },
+    {
+      "epoch": 19.411541541963494,
+      "grad_norm": 1.3488712310791016,
+      "learning_rate": 2.2737212193971615e-05,
+      "loss": 2.8485,
+      "step": 460500
+    },
+    {
+      "epoch": 19.432618134300046,
+      "grad_norm": 1.4187419414520264,
+      "learning_rate": 2.2651459932255713e-05,
+      "loss": 2.8535,
+      "step": 461000
+    },
+    {
+      "epoch": 19.4536947266366,
+      "grad_norm": 1.3795688152313232,
+      "learning_rate": 2.256570767053981e-05,
+      "loss": 2.8473,
+      "step": 461500
+    },
+    {
+      "epoch": 19.474771318973147,
+      "grad_norm": 1.3428224325180054,
+      "learning_rate": 2.2479955408823906e-05,
+      "loss": 2.8471,
+      "step": 462000
+    },
+    {
+      "epoch": 19.4958479113097,
+      "grad_norm": 1.343983769416809,
+      "learning_rate": 2.2394203147108005e-05,
+      "loss": 2.8498,
+      "step": 462500
+    },
+    {
+      "epoch": 19.51692450364625,
+      "grad_norm": 1.3629953861236572,
+      "learning_rate": 2.2308450885392103e-05,
+      "loss": 2.8482,
+      "step": 463000
+    },
+    {
+      "epoch": 19.5380010959828,
+      "grad_norm": 1.3485676050186157,
+      "learning_rate": 2.22226986236762e-05,
+      "loss": 2.8458,
+      "step": 463500
+    },
+    {
+      "epoch": 19.559077688319352,
+      "grad_norm": 1.3228490352630615,
+      "learning_rate": 2.21369463619603e-05,
+      "loss": 2.8478,
+      "step": 464000
+    },
+    {
+      "epoch": 19.580154280655904,
+      "grad_norm": 1.4440078735351562,
+      "learning_rate": 2.2051194100244394e-05,
+      "loss": 2.847,
+      "step": 464500
+    },
+    {
+      "epoch": 19.601230872992456,
+      "grad_norm": 1.3943135738372803,
+      "learning_rate": 2.1965441838528493e-05,
+      "loss": 2.8501,
+      "step": 465000
+    },
+    {
+      "epoch": 19.622307465329005,
+      "grad_norm": 1.3109320402145386,
+      "learning_rate": 2.187968957681259e-05,
+      "loss": 2.8468,
+      "step": 465500
+    },
+    {
+      "epoch": 19.643384057665557,
+      "grad_norm": 1.3717087507247925,
+      "learning_rate": 2.179393731509669e-05,
+      "loss": 2.8497,
+      "step": 466000
+    },
+    {
+      "epoch": 19.66446065000211,
+      "grad_norm": 1.3738688230514526,
+      "learning_rate": 2.1708185053380784e-05,
+      "loss": 2.8479,
+      "step": 466500
+    },
+    {
+      "epoch": 19.685537242338658,
+      "grad_norm": 1.350544810295105,
+      "learning_rate": 2.1622432791664883e-05,
+      "loss": 2.8501,
+      "step": 467000
+    },
+    {
+      "epoch": 19.70661383467521,
+      "grad_norm": 1.3468706607818604,
+      "learning_rate": 2.1536680529948978e-05,
+      "loss": 2.8478,
+      "step": 467500
+    },
+    {
+      "epoch": 19.727690427011762,
+      "grad_norm": 1.364469289779663,
+      "learning_rate": 2.1450928268233076e-05,
+      "loss": 2.848,
+      "step": 468000
+    },
+    {
+      "epoch": 19.74876701934831,
+      "grad_norm": 1.3141286373138428,
+      "learning_rate": 2.1365176006517174e-05,
+      "loss": 2.8471,
+      "step": 468500
+    },
+    {
+      "epoch": 19.769843611684863,
+      "grad_norm": 1.3615349531173706,
+      "learning_rate": 2.127942374480127e-05,
+      "loss": 2.8467,
+      "step": 469000
+    },
+    {
+      "epoch": 19.790920204021415,
+      "grad_norm": 1.3593207597732544,
+      "learning_rate": 2.1193671483085367e-05,
+      "loss": 2.8514,
+      "step": 469500
+    },
+    {
+      "epoch": 19.811996796357963,
+      "grad_norm": 1.3844629526138306,
+      "learning_rate": 2.1107919221369466e-05,
+      "loss": 2.8483,
+      "step": 470000
+    },
+    {
+      "epoch": 19.811996796357963,
+      "eval_accuracy": 0.728837095278357,
+      "eval_loss": 2.6110541820526123,
+      "eval_runtime": 267.4017,
+      "eval_samples_per_second": 451.403,
+      "eval_steps_per_second": 4.705,
+      "step": 470000
+    },
+    {
+      "epoch": 19.833073388694515,
+      "grad_norm": 1.3016059398651123,
+      "learning_rate": 2.102216695965356e-05,
+      "loss": 2.849,
+      "step": 470500
+    },
+    {
+      "epoch": 19.854149981031068,
+      "grad_norm": 1.3108514547348022,
+      "learning_rate": 2.093641469793766e-05,
+      "loss": 2.8469,
+      "step": 471000
+    },
+    {
+      "epoch": 19.87522657336762,
+      "grad_norm": 1.3011192083358765,
+      "learning_rate": 2.0850662436221757e-05,
+      "loss": 2.8463,
+      "step": 471500
+    },
+    {
+      "epoch": 19.89630316570417,
+      "grad_norm": 1.391687273979187,
+      "learning_rate": 2.0764910174505852e-05,
+      "loss": 2.8453,
+      "step": 472000
+    },
+    {
+      "epoch": 19.91737975804072,
+      "grad_norm": 1.3763200044631958,
+      "learning_rate": 2.067915791278995e-05,
+      "loss": 2.8486,
+      "step": 472500
+    },
+    {
+      "epoch": 19.938456350377272,
+      "grad_norm": 1.4159380197525024,
+      "learning_rate": 2.059340565107405e-05,
+      "loss": 2.8439,
+      "step": 473000
+    },
+    {
+      "epoch": 19.95953294271382,
+      "grad_norm": 1.3058205842971802,
+      "learning_rate": 2.0507653389358144e-05,
+      "loss": 2.8445,
+      "step": 473500
+    },
+    {
+      "epoch": 19.980609535050373,
+      "grad_norm": 1.5323214530944824,
+      "learning_rate": 2.0421901127642242e-05,
+      "loss": 2.8437,
+      "step": 474000
+    },
+    {
+      "epoch": 20.001686127386925,
+      "grad_norm": 1.4209623336791992,
+      "learning_rate": 2.033614886592634e-05,
+      "loss": 2.8515,
+      "step": 474500
+    },
+    {
+      "epoch": 20.022762719723474,
+      "grad_norm": 1.3624107837677002,
+      "learning_rate": 2.0250396604210435e-05,
+      "loss": 2.8457,
+      "step": 475000
+    },
+    {
+      "epoch": 20.043839312060026,
+      "grad_norm": 1.3771071434020996,
+      "learning_rate": 2.0164644342494534e-05,
+      "loss": 2.8454,
+      "step": 475500
+    },
+    {
+      "epoch": 20.064915904396578,
+      "grad_norm": 1.3530364036560059,
+      "learning_rate": 2.007889208077863e-05,
+      "loss": 2.839,
+      "step": 476000
+    },
+    {
+      "epoch": 20.085992496733127,
+      "grad_norm": 1.3525761365890503,
+      "learning_rate": 1.9993139819062727e-05,
+      "loss": 2.8419,
+      "step": 476500
+    },
+    {
+      "epoch": 20.10706908906968,
+      "grad_norm": 1.3513952493667603,
+      "learning_rate": 1.9907387557346825e-05,
+      "loss": 2.844,
+      "step": 477000
+    },
+    {
+      "epoch": 20.12814568140623,
+      "grad_norm": 1.4090635776519775,
+      "learning_rate": 1.9821635295630923e-05,
+      "loss": 2.8413,
+      "step": 477500
+    },
+    {
+      "epoch": 20.14922227374278,
+      "grad_norm": 1.3906279802322388,
+      "learning_rate": 1.9735883033915022e-05,
+      "loss": 2.8415,
+      "step": 478000
+    },
+    {
+      "epoch": 20.17029886607933,
+      "grad_norm": 1.3832578659057617,
+      "learning_rate": 1.965013077219912e-05,
+      "loss": 2.8462,
+      "step": 478500
+    },
+    {
+      "epoch": 20.191375458415884,
+      "grad_norm": 1.3939545154571533,
+      "learning_rate": 1.9564378510483215e-05,
+      "loss": 2.8427,
+      "step": 479000
+    },
+    {
+      "epoch": 20.212452050752436,
+      "grad_norm": 1.3964554071426392,
+      "learning_rate": 1.9478626248767313e-05,
+      "loss": 2.8454,
+      "step": 479500
+    },
+    {
+      "epoch": 20.233528643088984,
+      "grad_norm": 1.363648533821106,
+      "learning_rate": 1.939287398705141e-05,
+      "loss": 2.8417,
+      "step": 480000
+    },
+    {
+      "epoch": 20.233528643088984,
+      "eval_accuracy": 0.7296035394146798,
+      "eval_loss": 2.608185291290283,
+      "eval_runtime": 294.8198,
+      "eval_samples_per_second": 409.423,
+      "eval_steps_per_second": 4.267,
+      "step": 480000
+    },
+    {
+      "epoch": 20.254605235425537,
+      "grad_norm": 1.3936796188354492,
+      "learning_rate": 1.9307121725335506e-05,
+      "loss": 2.8384,
+      "step": 480500
+    },
+    {
+      "epoch": 20.27568182776209,
+      "grad_norm": 1.3856745958328247,
+      "learning_rate": 1.9221369463619605e-05,
+      "loss": 2.839,
+      "step": 481000
+    },
+    {
+      "epoch": 20.296758420098637,
+      "grad_norm": 1.343078851699829,
+      "learning_rate": 1.9135617201903703e-05,
+      "loss": 2.839,
+      "step": 481500
+    },
+    {
+      "epoch": 20.31783501243519,
+      "grad_norm": 1.3990718126296997,
+      "learning_rate": 1.9049864940187798e-05,
+      "loss": 2.8404,
+      "step": 482000
+    },
+    {
+      "epoch": 20.33891160477174,
+      "grad_norm": 1.4062094688415527,
+      "learning_rate": 1.8964112678471896e-05,
+      "loss": 2.8427,
+      "step": 482500
+    },
+    {
+      "epoch": 20.35998819710829,
+      "grad_norm": 1.4046804904937744,
+      "learning_rate": 1.8878360416755995e-05,
+      "loss": 2.8443,
+      "step": 483000
+    },
+    {
+      "epoch": 20.381064789444842,
+      "grad_norm": 1.3273723125457764,
+      "learning_rate": 1.879260815504009e-05,
+      "loss": 2.8399,
+      "step": 483500
+    },
+    {
+      "epoch": 20.402141381781394,
+      "grad_norm": 1.3577085733413696,
+      "learning_rate": 1.8706855893324188e-05,
+      "loss": 2.8388,
+      "step": 484000
+    },
+    {
+      "epoch": 20.423217974117943,
+      "grad_norm": 1.3747327327728271,
+      "learning_rate": 1.8621103631608283e-05,
+      "loss": 2.8402,
+      "step": 484500
+    },
+    {
+      "epoch": 20.444294566454495,
+      "grad_norm": 1.4414465427398682,
+      "learning_rate": 1.853535136989238e-05,
+      "loss": 2.8413,
+      "step": 485000
+    },
+    {
+      "epoch": 20.465371158791047,
+      "grad_norm": 1.334630012512207,
+      "learning_rate": 1.844959910817648e-05,
+      "loss": 2.8385,
+      "step": 485500
+    },
+    {
+      "epoch": 20.4864477511276,
+      "grad_norm": 1.3693829774856567,
+      "learning_rate": 1.8363846846460574e-05,
+      "loss": 2.8436,
+      "step": 486000
+    },
+    {
+      "epoch": 20.507524343464148,
+      "grad_norm": 1.415488362312317,
+      "learning_rate": 1.8278094584744673e-05,
+      "loss": 2.8379,
+      "step": 486500
+    },
+    {
+      "epoch": 20.5286009358007,
+      "grad_norm": 1.388505458831787,
+      "learning_rate": 1.819234232302877e-05,
+      "loss": 2.8418,
+      "step": 487000
+    },
+    {
+      "epoch": 20.549677528137252,
+      "grad_norm": 1.3294601440429688,
+      "learning_rate": 1.8106590061312866e-05,
+      "loss": 2.8413,
+      "step": 487500
+    },
+    {
+      "epoch": 20.5707541204738,
+      "grad_norm": 1.3442522287368774,
+      "learning_rate": 1.8020837799596964e-05,
+      "loss": 2.8426,
+      "step": 488000
+    },
+    {
+      "epoch": 20.591830712810353,
+      "grad_norm": 1.3539959192276,
+      "learning_rate": 1.7935085537881062e-05,
+      "loss": 2.8422,
+      "step": 488500
+    },
+    {
+      "epoch": 20.612907305146905,
+      "grad_norm": 1.4006271362304688,
+      "learning_rate": 1.7849333276165157e-05,
+      "loss": 2.8395,
+      "step": 489000
+    },
+    {
+      "epoch": 20.633983897483454,
+      "grad_norm": 1.4124704599380493,
+      "learning_rate": 1.7763581014449256e-05,
+      "loss": 2.8394,
+      "step": 489500
+    },
+    {
+      "epoch": 20.655060489820006,
+      "grad_norm": 1.361602544784546,
+      "learning_rate": 1.7677828752733354e-05,
+      "loss": 2.8365,
+      "step": 490000
+    },
+    {
+      "epoch": 20.655060489820006,
+      "eval_accuracy": 0.7302230716431735,
+      "eval_loss": 2.6040046215057373,
+      "eval_runtime": 256.822,
+      "eval_samples_per_second": 469.999,
+      "eval_steps_per_second": 4.898,
+      "step": 490000
+    },
+    {
+      "epoch": 20.676137082156558,
+      "grad_norm": 1.4504684209823608,
+      "learning_rate": 1.7592076491017452e-05,
+      "loss": 2.8396,
+      "step": 490500
+    },
+    {
+      "epoch": 20.697213674493106,
+      "grad_norm": 1.3938469886779785,
+      "learning_rate": 1.7506324229301547e-05,
+      "loss": 2.8419,
+      "step": 491000
+    },
+    {
+      "epoch": 20.71829026682966,
+      "grad_norm": 1.3424434661865234,
+      "learning_rate": 1.7420571967585646e-05,
+      "loss": 2.84,
+      "step": 491500
+    },
+    {
+      "epoch": 20.73936685916621,
+      "grad_norm": 1.4301432371139526,
+      "learning_rate": 1.7334819705869744e-05,
+      "loss": 2.8381,
+      "step": 492000
+    },
+    {
+      "epoch": 20.760443451502763,
+      "grad_norm": 1.3469057083129883,
+      "learning_rate": 1.7249067444153842e-05,
+      "loss": 2.8421,
+      "step": 492500
+    },
+    {
+      "epoch": 20.78152004383931,
+      "grad_norm": 1.3959057331085205,
+      "learning_rate": 1.7163315182437937e-05,
+      "loss": 2.8366,
+      "step": 493000
+    },
+    {
+      "epoch": 20.802596636175863,
+      "grad_norm": 1.3644378185272217,
+      "learning_rate": 1.7077562920722035e-05,
+      "loss": 2.8364,
+      "step": 493500
+    },
+    {
+      "epoch": 20.823673228512416,
+      "grad_norm": 1.3749688863754272,
+      "learning_rate": 1.6991810659006134e-05,
+      "loss": 2.8381,
+      "step": 494000
+    },
+    {
+      "epoch": 20.844749820848964,
+      "grad_norm": 1.3468384742736816,
+      "learning_rate": 1.690605839729023e-05,
+      "loss": 2.8393,
+      "step": 494500
+    },
+    {
+      "epoch": 20.865826413185516,
+      "grad_norm": 1.3564876317977905,
+      "learning_rate": 1.6820306135574327e-05,
+      "loss": 2.8371,
+      "step": 495000
+    },
+    {
+      "epoch": 20.88690300552207,
+      "grad_norm": 1.4270135164260864,
+      "learning_rate": 1.6734553873858425e-05,
+      "loss": 2.8399,
+      "step": 495500
+    },
+    {
+      "epoch": 20.907979597858617,
+      "grad_norm": 1.4028925895690918,
+      "learning_rate": 1.664880161214252e-05,
+      "loss": 2.8375,
+      "step": 496000
+    },
+    {
+      "epoch": 20.92905619019517,
+      "grad_norm": 1.371846318244934,
+      "learning_rate": 1.656304935042662e-05,
+      "loss": 2.8413,
+      "step": 496500
+    },
+    {
+      "epoch": 20.95013278253172,
+      "grad_norm": 1.3493578433990479,
+      "learning_rate": 1.6477297088710717e-05,
+      "loss": 2.8369,
+      "step": 497000
+    },
+    {
+      "epoch": 20.97120937486827,
+      "grad_norm": 1.3745207786560059,
+      "learning_rate": 1.639154482699481e-05,
+      "loss": 2.8389,
+      "step": 497500
+    },
+    {
+      "epoch": 20.992285967204822,
+      "grad_norm": 1.33699631690979,
+      "learning_rate": 1.630579256527891e-05,
+      "loss": 2.8369,
+      "step": 498000
+    },
+    {
+      "epoch": 21.013362559541374,
+      "grad_norm": 1.3996938467025757,
+      "learning_rate": 1.6220040303563008e-05,
+      "loss": 2.8364,
+      "step": 498500
+    },
+    {
+      "epoch": 21.034439151877926,
+      "grad_norm": 1.418601632118225,
+      "learning_rate": 1.6134288041847103e-05,
+      "loss": 2.8379,
+      "step": 499000
+    },
+    {
+      "epoch": 21.055515744214475,
+      "grad_norm": 1.3746311664581299,
+      "learning_rate": 1.60485357801312e-05,
+      "loss": 2.8327,
+      "step": 499500
+    },
+    {
+      "epoch": 21.076592336551027,
+      "grad_norm": 1.3654979467391968,
+      "learning_rate": 1.59627835184153e-05,
+      "loss": 2.8332,
+      "step": 500000
+    },
+    {
+      "epoch": 21.076592336551027,
+      "eval_accuracy": 0.7304450627349035,
+      "eval_loss": 2.6011037826538086,
+      "eval_runtime": 279.8126,
+      "eval_samples_per_second": 431.382,
+      "eval_steps_per_second": 4.496,
+      "step": 500000
+    },
+    {
+      "epoch": 21.09766892888758,
+      "grad_norm": 1.3871346712112427,
+      "learning_rate": 1.5877031256699395e-05,
+      "loss": 2.833,
+      "step": 500500
+    },
+    {
+      "epoch": 21.118745521224128,
+      "grad_norm": 1.3134562969207764,
+      "learning_rate": 1.5791278994983493e-05,
+      "loss": 2.8369,
+      "step": 501000
+    },
+    {
+      "epoch": 21.13982211356068,
+      "grad_norm": 1.3080514669418335,
+      "learning_rate": 1.570552673326759e-05,
+      "loss": 2.8318,
+      "step": 501500
+    },
+    {
+      "epoch": 21.16089870589723,
+      "grad_norm": 1.3946727514266968,
+      "learning_rate": 1.5619774471551686e-05,
+      "loss": 2.8368,
+      "step": 502000
+    },
+    {
+      "epoch": 21.18197529823378,
+      "grad_norm": 1.3150570392608643,
+      "learning_rate": 1.5534022209835785e-05,
+      "loss": 2.8364,
+      "step": 502500
+    },
+    {
+      "epoch": 21.203051890570332,
+      "grad_norm": 1.3574693202972412,
+      "learning_rate": 1.544826994811988e-05,
+      "loss": 2.8345,
+      "step": 503000
+    },
+    {
+      "epoch": 21.224128482906885,
+      "grad_norm": 1.3627499341964722,
+      "learning_rate": 1.5362517686403978e-05,
+      "loss": 2.835,
+      "step": 503500
+    },
+    {
+      "epoch": 21.245205075243433,
+      "grad_norm": 1.377774715423584,
+      "learning_rate": 1.5276765424688076e-05,
+      "loss": 2.8353,
+      "step": 504000
+    },
+    {
+      "epoch": 21.266281667579985,
+      "grad_norm": 1.4714667797088623,
+      "learning_rate": 1.5191013162972173e-05,
+      "loss": 2.8349,
+      "step": 504500
+    },
+    {
+      "epoch": 21.287358259916537,
+      "grad_norm": 1.3344879150390625,
+      "learning_rate": 1.5105260901256271e-05,
+      "loss": 2.8349,
+      "step": 505000
+    },
+    {
+      "epoch": 21.308434852253086,
+      "grad_norm": 1.4018884897232056,
+      "learning_rate": 1.501950863954037e-05,
+      "loss": 2.8351,
+      "step": 505500
+    },
+    {
+      "epoch": 21.329511444589638,
+      "grad_norm": 1.3674098253250122,
+      "learning_rate": 1.4933756377824464e-05,
+      "loss": 2.8327,
+      "step": 506000
+    },
+    {
+      "epoch": 21.35058803692619,
+      "grad_norm": 1.4040616750717163,
+      "learning_rate": 1.4848004116108563e-05,
+      "loss": 2.8293,
+      "step": 506500
+    },
+    {
+      "epoch": 21.371664629262742,
+      "grad_norm": 1.356368064880371,
+      "learning_rate": 1.4762251854392661e-05,
+      "loss": 2.8361,
+      "step": 507000
+    },
+    {
+      "epoch": 21.39274122159929,
+      "grad_norm": 1.360677719116211,
+      "learning_rate": 1.4676499592676758e-05,
+      "loss": 2.8359,
+      "step": 507500
+    },
+    {
+      "epoch": 21.413817813935843,
+      "grad_norm": 1.3671870231628418,
+      "learning_rate": 1.4590747330960856e-05,
+      "loss": 2.8334,
+      "step": 508000
+    },
+    {
+      "epoch": 21.434894406272395,
+      "grad_norm": 1.396283507347107,
+      "learning_rate": 1.4504995069244952e-05,
+      "loss": 2.8304,
+      "step": 508500
+    },
+    {
+      "epoch": 21.455970998608944,
+      "grad_norm": 1.3333899974822998,
+      "learning_rate": 1.4419242807529049e-05,
+      "loss": 2.8335,
+      "step": 509000
+    },
+    {
+      "epoch": 21.477047590945496,
+      "grad_norm": 1.3206731081008911,
+      "learning_rate": 1.4333490545813147e-05,
+      "loss": 2.8338,
+      "step": 509500
+    },
+    {
+      "epoch": 21.498124183282048,
+      "grad_norm": 1.4073832035064697,
+      "learning_rate": 1.4247738284097246e-05,
+      "loss": 2.8313,
+      "step": 510000
+    },
+    {
+      "epoch": 21.498124183282048,
+      "eval_accuracy": 0.7311059591416676,
+      "eval_loss": 2.59879994392395,
+      "eval_runtime": 302.201,
+      "eval_samples_per_second": 399.423,
+      "eval_steps_per_second": 4.163,
+      "step": 510000
+    },
+    {
+      "epoch": 21.519200775618597,
+      "grad_norm": 1.3738247156143188,
+      "learning_rate": 1.416198602238134e-05,
+      "loss": 2.8348,
+      "step": 510500
+    },
+    {
+      "epoch": 21.54027736795515,
+      "grad_norm": 1.3383632898330688,
+      "learning_rate": 1.4076233760665439e-05,
+      "loss": 2.8327,
+      "step": 511000
+    },
+    {
+      "epoch": 21.5613539602917,
+      "grad_norm": 1.3685204982757568,
+      "learning_rate": 1.3990481498949534e-05,
+      "loss": 2.831,
+      "step": 511500
+    },
+    {
+      "epoch": 21.58243055262825,
+      "grad_norm": 1.3857929706573486,
+      "learning_rate": 1.3904729237233632e-05,
+      "loss": 2.8341,
+      "step": 512000
+    },
+    {
+      "epoch": 21.6035071449648,
+      "grad_norm": 1.3626941442489624,
+      "learning_rate": 1.381897697551773e-05,
+      "loss": 2.8304,
+      "step": 512500
+    },
+    {
+      "epoch": 21.624583737301354,
+      "grad_norm": 1.3623549938201904,
+      "learning_rate": 1.3733224713801825e-05,
+      "loss": 2.8343,
+      "step": 513000
+    },
+    {
+      "epoch": 21.645660329637906,
+      "grad_norm": 1.3653994798660278,
+      "learning_rate": 1.3647472452085924e-05,
+      "loss": 2.8366,
+      "step": 513500
+    },
+    {
+      "epoch": 21.666736921974454,
+      "grad_norm": 1.4271252155303955,
+      "learning_rate": 1.3561720190370022e-05,
+      "loss": 2.8336,
+      "step": 514000
+    },
+    {
+      "epoch": 21.687813514311006,
+      "grad_norm": 1.348188042640686,
+      "learning_rate": 1.3475967928654119e-05,
+      "loss": 2.8313,
+      "step": 514500
+    },
+    {
+      "epoch": 21.70889010664756,
+      "grad_norm": 1.4128820896148682,
+      "learning_rate": 1.3390215666938217e-05,
+      "loss": 2.8331,
+      "step": 515000
+    },
+    {
+      "epoch": 21.729966698984107,
+      "grad_norm": 1.3467466831207275,
+      "learning_rate": 1.3304463405222315e-05,
+      "loss": 2.8335,
+      "step": 515500
+    },
+    {
+      "epoch": 21.75104329132066,
+      "grad_norm": 1.4332268238067627,
+      "learning_rate": 1.321871114350641e-05,
+      "loss": 2.8353,
+      "step": 516000
+    },
+    {
+      "epoch": 21.77211988365721,
+      "grad_norm": 1.367585301399231,
+      "learning_rate": 1.3132958881790508e-05,
+      "loss": 2.8292,
+      "step": 516500
+    },
+    {
+      "epoch": 21.79319647599376,
+      "grad_norm": 1.3276602029800415,
+      "learning_rate": 1.3047206620074607e-05,
+      "loss": 2.83,
+      "step": 517000
+    },
+    {
+      "epoch": 21.814273068330312,
+      "grad_norm": 1.4049382209777832,
+      "learning_rate": 1.2961454358358702e-05,
+      "loss": 2.8305,
+      "step": 517500
+    },
+    {
+      "epoch": 21.835349660666864,
+      "grad_norm": 1.3750227689743042,
+      "learning_rate": 1.28757020966428e-05,
+      "loss": 2.8286,
+      "step": 518000
+    },
+    {
+      "epoch": 21.856426253003413,
+      "grad_norm": 1.3645752668380737,
+      "learning_rate": 1.2789949834926898e-05,
+      "loss": 2.8277,
+      "step": 518500
+    },
+    {
+      "epoch": 21.877502845339965,
+      "grad_norm": 1.3580256700515747,
+      "learning_rate": 1.2704197573210993e-05,
+      "loss": 2.8353,
+      "step": 519000
+    },
+    {
+      "epoch": 21.898579437676517,
+      "grad_norm": 1.3635903596878052,
+      "learning_rate": 1.2618445311495092e-05,
+      "loss": 2.8277,
+      "step": 519500
+    },
+    {
+      "epoch": 21.91965603001307,
+      "grad_norm": 1.3960708379745483,
+      "learning_rate": 1.2532693049779186e-05,
+      "loss": 2.8267,
+      "step": 520000
+    },
+    {
+      "epoch": 21.91965603001307,
+      "eval_accuracy": 0.7314282548863971,
+      "eval_loss": 2.597130060195923,
+      "eval_runtime": 265.5547,
+      "eval_samples_per_second": 454.543,
+      "eval_steps_per_second": 4.737,
+      "step": 520000
+    },
+    {
+      "epoch": 21.940732622349618,
+      "grad_norm": 1.424583911895752,
+      "learning_rate": 1.2446940788063285e-05,
+      "loss": 2.8308,
+      "step": 520500
+    },
+    {
+      "epoch": 21.96180921468617,
+      "grad_norm": 1.4369407892227173,
+      "learning_rate": 1.2361188526347383e-05,
+      "loss": 2.8308,
+      "step": 521000
+    },
+    {
+      "epoch": 21.982885807022722,
+      "grad_norm": 1.4022798538208008,
+      "learning_rate": 1.227543626463148e-05,
+      "loss": 2.8308,
+      "step": 521500
+    },
+    {
+      "epoch": 22.00396239935927,
+      "grad_norm": 1.4078301191329956,
+      "learning_rate": 1.2189684002915578e-05,
+      "loss": 2.8268,
+      "step": 522000
+    },
+    {
+      "epoch": 22.025038991695823,
+      "grad_norm": 1.4006880521774292,
+      "learning_rate": 1.2103931741199675e-05,
+      "loss": 2.825,
+      "step": 522500
+    },
+    {
+      "epoch": 22.046115584032375,
+      "grad_norm": 1.3603681325912476,
+      "learning_rate": 1.2018179479483773e-05,
+      "loss": 2.8269,
+      "step": 523000
+    },
+    {
+      "epoch": 22.067192176368923,
+      "grad_norm": 1.378110647201538,
+      "learning_rate": 1.193242721776787e-05,
+      "loss": 2.8296,
+      "step": 523500
+    },
+    {
+      "epoch": 22.088268768705476,
+      "grad_norm": 1.4455020427703857,
+      "learning_rate": 1.1846674956051966e-05,
+      "loss": 2.8311,
+      "step": 524000
+    },
+    {
+      "epoch": 22.109345361042028,
+      "grad_norm": 1.4161103963851929,
+      "learning_rate": 1.1760922694336064e-05,
+      "loss": 2.8249,
+      "step": 524500
+    },
+    {
+      "epoch": 22.130421953378576,
+      "grad_norm": 1.3881396055221558,
+      "learning_rate": 1.1675170432620161e-05,
+      "loss": 2.8323,
+      "step": 525000
+    },
+    {
+      "epoch": 22.15149854571513,
+      "grad_norm": 1.3841625452041626,
+      "learning_rate": 1.1589418170904258e-05,
+      "loss": 2.8236,
+      "step": 525500
+    },
+    {
+      "epoch": 22.17257513805168,
+      "grad_norm": 1.363213062286377,
+      "learning_rate": 1.1503665909188354e-05,
+      "loss": 2.8255,
+      "step": 526000
+    },
+    {
+      "epoch": 22.193651730388233,
+      "grad_norm": 1.4149315357208252,
+      "learning_rate": 1.1417913647472453e-05,
+      "loss": 2.8275,
+      "step": 526500
+    },
+    {
+      "epoch": 22.21472832272478,
+      "grad_norm": 1.3616867065429688,
+      "learning_rate": 1.133216138575655e-05,
+      "loss": 2.8269,
+      "step": 527000
+    },
+    {
+      "epoch": 22.235804915061333,
+      "grad_norm": 1.4398738145828247,
+      "learning_rate": 1.1246409124040646e-05,
+      "loss": 2.8254,
+      "step": 527500
+    },
+    {
+      "epoch": 22.256881507397885,
+      "grad_norm": 1.3829830884933472,
+      "learning_rate": 1.1160656862324744e-05,
+      "loss": 2.8284,
+      "step": 528000
+    },
+    {
+      "epoch": 22.277958099734434,
+      "grad_norm": 1.3645559549331665,
+      "learning_rate": 1.1074904600608842e-05,
+      "loss": 2.8275,
+      "step": 528500
+    },
+    {
+      "epoch": 22.299034692070986,
+      "grad_norm": 1.4445343017578125,
+      "learning_rate": 1.0989152338892939e-05,
+      "loss": 2.8237,
+      "step": 529000
+    },
+    {
+      "epoch": 22.320111284407538,
+      "grad_norm": 1.477734088897705,
+      "learning_rate": 1.0903400077177037e-05,
+      "loss": 2.8289,
+      "step": 529500
+    },
+    {
+      "epoch": 22.341187876744087,
+      "grad_norm": 1.3276185989379883,
+      "learning_rate": 1.0817647815461134e-05,
+      "loss": 2.8285,
+      "step": 530000
+    },
+    {
+      "epoch": 22.341187876744087,
+      "eval_accuracy": 0.7316439398043793,
+      "eval_loss": 2.5954270362854004,
+      "eval_runtime": 272.9887,
+      "eval_samples_per_second": 442.165,
+      "eval_steps_per_second": 4.608,
+      "step": 530000
+    },
+    {
+      "epoch": 22.36226446908064,
+      "grad_norm": 1.3404492139816284,
+      "learning_rate": 1.073189555374523e-05,
+      "loss": 2.8288,
+      "step": 530500
+    },
+    {
+      "epoch": 22.38334106141719,
+      "grad_norm": 1.461935043334961,
+      "learning_rate": 1.0646143292029327e-05,
+      "loss": 2.8301,
+      "step": 531000
+    },
+    {
+      "epoch": 22.40441765375374,
+      "grad_norm": 1.4038807153701782,
+      "learning_rate": 1.0560391030313425e-05,
+      "loss": 2.8249,
+      "step": 531500
+    },
+    {
+      "epoch": 22.42549424609029,
+      "grad_norm": 1.3839021921157837,
+      "learning_rate": 1.0474638768597522e-05,
+      "loss": 2.8242,
+      "step": 532000
+    },
+    {
+      "epoch": 22.446570838426844,
+      "grad_norm": 1.376606822013855,
+      "learning_rate": 1.0388886506881619e-05,
+      "loss": 2.8252,
+      "step": 532500
+    },
+    {
+      "epoch": 22.467647430763392,
+      "grad_norm": 1.4007099866867065,
+      "learning_rate": 1.0303134245165717e-05,
+      "loss": 2.8282,
+      "step": 533000
+    },
+    {
+      "epoch": 22.488724023099945,
+      "grad_norm": 1.3907722234725952,
+      "learning_rate": 1.0217381983449814e-05,
+      "loss": 2.8247,
+      "step": 533500
+    },
+    {
+      "epoch": 22.509800615436497,
+      "grad_norm": 1.3220123052597046,
+      "learning_rate": 1.013162972173391e-05,
+      "loss": 2.8256,
+      "step": 534000
+    },
+    {
+      "epoch": 22.53087720777305,
+      "grad_norm": 1.3846632242202759,
+      "learning_rate": 1.0045877460018009e-05,
+      "loss": 2.828,
+      "step": 534500
+    },
+    {
+      "epoch": 22.551953800109597,
+      "grad_norm": 1.3760250806808472,
+      "learning_rate": 9.960125198302105e-06,
+      "loss": 2.8291,
+      "step": 535000
+    },
+    {
+      "epoch": 22.57303039244615,
+      "grad_norm": 1.383386492729187,
+      "learning_rate": 9.874372936586203e-06,
+      "loss": 2.825,
+      "step": 535500
+    },
+    {
+      "epoch": 22.5941069847827,
+      "grad_norm": 1.359326958656311,
+      "learning_rate": 9.7886206748703e-06,
+      "loss": 2.8246,
+      "step": 536000
+    },
+    {
+      "epoch": 22.61518357711925,
+      "grad_norm": 1.3919751644134521,
+      "learning_rate": 9.702868413154398e-06,
+      "loss": 2.8236,
+      "step": 536500
+    },
+    {
+      "epoch": 22.636260169455802,
+      "grad_norm": 1.3739207983016968,
+      "learning_rate": 9.617116151438495e-06,
+      "loss": 2.8236,
+      "step": 537000
+    },
+    {
+      "epoch": 22.657336761792354,
+      "grad_norm": 1.3899908065795898,
+      "learning_rate": 9.531363889722592e-06,
+      "loss": 2.8259,
+      "step": 537500
+    },
+    {
+      "epoch": 22.678413354128903,
+      "grad_norm": 1.375300645828247,
+      "learning_rate": 9.44561162800669e-06,
+      "loss": 2.8235,
+      "step": 538000
+    },
+    {
+      "epoch": 22.699489946465455,
+      "grad_norm": 1.41339910030365,
+      "learning_rate": 9.359859366290787e-06,
+      "loss": 2.825,
+      "step": 538500
+    },
+    {
+      "epoch": 22.720566538802007,
+      "grad_norm": 1.3590682744979858,
+      "learning_rate": 9.274107104574883e-06,
+      "loss": 2.8265,
+      "step": 539000
+    },
+    {
+      "epoch": 22.741643131138556,
+      "grad_norm": 1.3708281517028809,
+      "learning_rate": 9.18835484285898e-06,
+      "loss": 2.8226,
+      "step": 539500
+    },
+    {
+      "epoch": 22.762719723475108,
+      "grad_norm": 1.3774445056915283,
+      "learning_rate": 9.102602581143078e-06,
+      "loss": 2.8236,
+      "step": 540000
+    },
+    {
+      "epoch": 22.762719723475108,
+      "eval_accuracy": 0.7318702804688777,
+      "eval_loss": 2.5944252014160156,
+      "eval_runtime": 302.5475,
+      "eval_samples_per_second": 398.965,
+      "eval_steps_per_second": 4.158,
+      "step": 540000
+    },
+    {
+      "epoch": 22.78379631581166,
+      "grad_norm": 1.3897442817687988,
+      "learning_rate": 9.016850319427175e-06,
+      "loss": 2.8279,
+      "step": 540500
+    },
+    {
+      "epoch": 22.804872908148212,
+      "grad_norm": 1.349951982498169,
+      "learning_rate": 8.931098057711271e-06,
+      "loss": 2.8251,
+      "step": 541000
+    },
+    {
+      "epoch": 22.82594950048476,
+      "grad_norm": 1.3667008876800537,
+      "learning_rate": 8.84534579599537e-06,
+      "loss": 2.8248,
+      "step": 541500
+    },
+    {
+      "epoch": 22.847026092821313,
+      "grad_norm": 1.4389101266860962,
+      "learning_rate": 8.759593534279468e-06,
+      "loss": 2.8259,
+      "step": 542000
+    },
+    {
+      "epoch": 22.868102685157865,
+      "grad_norm": 1.3651082515716553,
+      "learning_rate": 8.673841272563565e-06,
+      "loss": 2.8223,
+      "step": 542500
+    },
+    {
+      "epoch": 22.889179277494414,
+      "grad_norm": 1.3748681545257568,
+      "learning_rate": 8.588089010847663e-06,
+      "loss": 2.8255,
+      "step": 543000
+    },
+    {
+      "epoch": 22.910255869830966,
+      "grad_norm": 1.3739687204360962,
+      "learning_rate": 8.50233674913176e-06,
+      "loss": 2.8246,
+      "step": 543500
+    },
+    {
+      "epoch": 22.931332462167518,
+      "grad_norm": 1.449851632118225,
+      "learning_rate": 8.416584487415856e-06,
+      "loss": 2.8204,
+      "step": 544000
+    },
+    {
+      "epoch": 22.952409054504066,
+      "grad_norm": 1.4528748989105225,
+      "learning_rate": 8.330832225699953e-06,
+      "loss": 2.82,
+      "step": 544500
+    },
+    {
+      "epoch": 22.97348564684062,
+      "grad_norm": 1.4438631534576416,
+      "learning_rate": 8.245079963984051e-06,
+      "loss": 2.8207,
+      "step": 545000
+    },
+    {
+      "epoch": 22.99456223917717,
+      "grad_norm": 1.465223789215088,
+      "learning_rate": 8.159327702268148e-06,
+      "loss": 2.826,
+      "step": 545500
+    },
+    {
+      "epoch": 23.01563883151372,
+      "grad_norm": 1.4519896507263184,
+      "learning_rate": 8.073575440552244e-06,
+      "loss": 2.8217,
+      "step": 546000
+    },
+    {
+      "epoch": 23.03671542385027,
+      "grad_norm": 1.3797649145126343,
+      "learning_rate": 7.987823178836343e-06,
+      "loss": 2.8236,
+      "step": 546500
+    },
+    {
+      "epoch": 23.057792016186824,
+      "grad_norm": 1.436012625694275,
+      "learning_rate": 7.90207091712044e-06,
+      "loss": 2.8206,
+      "step": 547000
+    },
+    {
+      "epoch": 23.078868608523376,
+      "grad_norm": 1.4045723676681519,
+      "learning_rate": 7.816318655404536e-06,
+      "loss": 2.8199,
+      "step": 547500
+    },
+    {
+      "epoch": 23.099945200859924,
+      "grad_norm": 1.3928680419921875,
+      "learning_rate": 7.730566393688634e-06,
+      "loss": 2.8225,
+      "step": 548000
+    },
+    {
+      "epoch": 23.121021793196476,
+      "grad_norm": 1.3844735622406006,
+      "learning_rate": 7.64481413197273e-06,
+      "loss": 2.8218,
+      "step": 548500
+    },
+    {
+      "epoch": 23.14209838553303,
+      "grad_norm": 1.357153296470642,
+      "learning_rate": 7.559061870256828e-06,
+      "loss": 2.821,
+      "step": 549000
+    },
+    {
+      "epoch": 23.163174977869577,
+      "grad_norm": 1.3910455703735352,
+      "learning_rate": 7.473309608540925e-06,
+      "loss": 2.8198,
+      "step": 549500
+    },
+    {
+      "epoch": 23.18425157020613,
+      "grad_norm": 1.4220564365386963,
+      "learning_rate": 7.387557346825023e-06,
+      "loss": 2.8246,
+      "step": 550000
+    },
+    {
+      "epoch": 23.18425157020613,
+      "eval_accuracy": 0.7323013219740654,
+      "eval_loss": 2.590684652328491,
+      "eval_runtime": 275.2193,
+      "eval_samples_per_second": 438.581,
+      "eval_steps_per_second": 4.571,
+      "step": 550000
+    },
+    {
+      "epoch": 23.20532816254268,
+      "grad_norm": 1.4060066938400269,
+      "learning_rate": 7.3018050851091205e-06,
+      "loss": 2.8176,
+      "step": 550500
+    },
+    {
+      "epoch": 23.22640475487923,
+      "grad_norm": 1.4198278188705444,
+      "learning_rate": 7.216052823393217e-06,
+      "loss": 2.8209,
+      "step": 551000
+    },
+    {
+      "epoch": 23.247481347215782,
+      "grad_norm": 1.4207963943481445,
+      "learning_rate": 7.1303005616773155e-06,
+      "loss": 2.8208,
+      "step": 551500
+    },
+    {
+      "epoch": 23.268557939552334,
+      "grad_norm": 1.426999568939209,
+      "learning_rate": 7.044548299961412e-06,
+      "loss": 2.8242,
+      "step": 552000
+    },
+    {
+      "epoch": 23.289634531888883,
+      "grad_norm": 1.4017759561538696,
+      "learning_rate": 6.958796038245509e-06,
+      "loss": 2.8154,
+      "step": 552500
+    },
+    {
+      "epoch": 23.310711124225435,
+      "grad_norm": 1.4759787321090698,
+      "learning_rate": 6.873043776529606e-06,
+      "loss": 2.8237,
+      "step": 553000
+    },
+    {
+      "epoch": 23.331787716561987,
+      "grad_norm": 1.4306811094284058,
+      "learning_rate": 6.787291514813704e-06,
+      "loss": 2.8181,
+      "step": 553500
+    },
+    {
+      "epoch": 23.35286430889854,
+      "grad_norm": 1.4093035459518433,
+      "learning_rate": 6.701539253097801e-06,
+      "loss": 2.8198,
+      "step": 554000
+    },
+    {
+      "epoch": 23.373940901235088,
+      "grad_norm": 1.4773316383361816,
+      "learning_rate": 6.615786991381898e-06,
+      "loss": 2.8203,
+      "step": 554500
+    },
+    {
+      "epoch": 23.39501749357164,
+      "grad_norm": 1.3714536428451538,
+      "learning_rate": 6.530034729665996e-06,
+      "loss": 2.8205,
+      "step": 555000
+    },
+    {
+      "epoch": 23.416094085908192,
+      "grad_norm": 1.3742094039916992,
+      "learning_rate": 6.444282467950093e-06,
+      "loss": 2.8194,
+      "step": 555500
+    },
+    {
+      "epoch": 23.43717067824474,
+      "grad_norm": 1.3805311918258667,
+      "learning_rate": 6.358530206234189e-06,
+      "loss": 2.8184,
+      "step": 556000
+    },
+    {
+      "epoch": 23.458247270581293,
+      "grad_norm": 1.4282187223434448,
+      "learning_rate": 6.272777944518287e-06,
+      "loss": 2.8221,
+      "step": 556500
+    },
+    {
+      "epoch": 23.479323862917845,
+      "grad_norm": 1.3903534412384033,
+      "learning_rate": 6.187025682802384e-06,
+      "loss": 2.8212,
+      "step": 557000
+    },
+    {
+      "epoch": 23.500400455254393,
+      "grad_norm": 1.377451777458191,
+      "learning_rate": 6.101273421086482e-06,
+      "loss": 2.8198,
+      "step": 557500
+    },
+    {
+      "epoch": 23.521477047590945,
+      "grad_norm": 1.380693793296814,
+      "learning_rate": 6.015521159370579e-06,
+      "loss": 2.8195,
+      "step": 558000
+    },
+    {
+      "epoch": 23.542553639927497,
+      "grad_norm": 1.4300856590270996,
+      "learning_rate": 5.929768897654676e-06,
+      "loss": 2.8167,
+      "step": 558500
+    },
+    {
+      "epoch": 23.563630232264046,
+      "grad_norm": 1.3954763412475586,
+      "learning_rate": 5.844016635938773e-06,
+      "loss": 2.8193,
+      "step": 559000
+    },
+    {
+      "epoch": 23.584706824600598,
+      "grad_norm": 1.3683284521102905,
+      "learning_rate": 5.75826437422287e-06,
+      "loss": 2.8195,
+      "step": 559500
+    },
+    {
+      "epoch": 23.60578341693715,
+      "grad_norm": 1.4127775430679321,
+      "learning_rate": 5.672512112506967e-06,
+      "loss": 2.8196,
+      "step": 560000
+    },
+    {
+      "epoch": 23.60578341693715,
+      "eval_accuracy": 0.7328530565351574,
+      "eval_loss": 2.5883095264434814,
+      "eval_runtime": 259.3069,
+      "eval_samples_per_second": 465.495,
+      "eval_steps_per_second": 4.851,
+      "step": 560000
+    },
+    {
+      "epoch": 23.6268600092737,
+      "grad_norm": 1.3992462158203125,
+      "learning_rate": 5.5867598507910655e-06,
+      "loss": 2.8194,
+      "step": 560500
+    },
+    {
+      "epoch": 23.64793660161025,
+      "grad_norm": 1.3987964391708374,
+      "learning_rate": 5.501007589075162e-06,
+      "loss": 2.8187,
+      "step": 561000
+    },
+    {
+      "epoch": 23.669013193946803,
+      "grad_norm": 1.4430257081985474,
+      "learning_rate": 5.41525532735926e-06,
+      "loss": 2.8196,
+      "step": 561500
+    },
+    {
+      "epoch": 23.690089786283355,
+      "grad_norm": 1.4419833421707153,
+      "learning_rate": 5.329503065643356e-06,
+      "loss": 2.8156,
+      "step": 562000
+    },
+    {
+      "epoch": 23.711166378619904,
+      "grad_norm": 1.4438389539718628,
+      "learning_rate": 5.243750803927454e-06,
+      "loss": 2.8177,
+      "step": 562500
+    },
+    {
+      "epoch": 23.732242970956456,
+      "grad_norm": 1.4502285718917847,
+      "learning_rate": 5.157998542211551e-06,
+      "loss": 2.8197,
+      "step": 563000
+    },
+    {
+      "epoch": 23.753319563293008,
+      "grad_norm": 1.4341175556182861,
+      "learning_rate": 5.072246280495649e-06,
+      "loss": 2.8168,
+      "step": 563500
+    },
+    {
+      "epoch": 23.774396155629557,
+      "grad_norm": 1.3912016153335571,
+      "learning_rate": 4.986494018779746e-06,
+      "loss": 2.8206,
+      "step": 564000
+    },
+    {
+      "epoch": 23.79547274796611,
+      "grad_norm": 1.4620940685272217,
+      "learning_rate": 4.900741757063843e-06,
+      "loss": 2.8171,
+      "step": 564500
+    },
+    {
+      "epoch": 23.81654934030266,
+      "grad_norm": 1.406883716583252,
+      "learning_rate": 4.81498949534794e-06,
+      "loss": 2.8167,
+      "step": 565000
+    },
+    {
+      "epoch": 23.83762593263921,
+      "grad_norm": 1.3698009252548218,
+      "learning_rate": 4.729237233632038e-06,
+      "loss": 2.8171,
+      "step": 565500
+    },
+    {
+      "epoch": 23.85870252497576,
+      "grad_norm": 1.3989384174346924,
+      "learning_rate": 4.643484971916134e-06,
+      "loss": 2.8203,
+      "step": 566000
+    },
+    {
+      "epoch": 23.879779117312314,
+      "grad_norm": 1.3887088298797607,
+      "learning_rate": 4.557732710200232e-06,
+      "loss": 2.8172,
+      "step": 566500
+    },
+    {
+      "epoch": 23.900855709648862,
+      "grad_norm": 1.423384666442871,
+      "learning_rate": 4.471980448484329e-06,
+      "loss": 2.8164,
+      "step": 567000
+    },
+    {
+      "epoch": 23.921932301985414,
+      "grad_norm": 1.5583797693252563,
+      "learning_rate": 4.386228186768427e-06,
+      "loss": 2.8175,
+      "step": 567500
+    },
+    {
+      "epoch": 23.943008894321967,
+      "grad_norm": 1.419384241104126,
+      "learning_rate": 4.300475925052524e-06,
+      "loss": 2.8166,
+      "step": 568000
+    },
+    {
+      "epoch": 23.96408548665852,
+      "grad_norm": 1.391713261604309,
+      "learning_rate": 4.214723663336621e-06,
+      "loss": 2.8206,
+      "step": 568500
+    },
+    {
+      "epoch": 23.985162078995067,
+      "grad_norm": 1.4320404529571533,
+      "learning_rate": 4.128971401620718e-06,
+      "loss": 2.8155,
+      "step": 569000
+    },
+    {
+      "epoch": 24.00623867133162,
+      "grad_norm": 1.422065258026123,
+      "learning_rate": 4.043219139904815e-06,
+      "loss": 2.8176,
+      "step": 569500
+    },
+    {
+      "epoch": 24.02731526366817,
+      "grad_norm": 1.381488561630249,
+      "learning_rate": 3.957466878188912e-06,
+      "loss": 2.8161,
+      "step": 570000
+    },
+    {
+      "epoch": 24.02731526366817,
+      "eval_accuracy": 0.7330296901533309,
+      "eval_loss": 2.587742567062378,
+      "eval_runtime": 281.2492,
+      "eval_samples_per_second": 429.178,
+      "eval_steps_per_second": 4.473,
+      "step": 570000
+    },
+    {
+      "epoch": 24.04839185600472,
+      "grad_norm": 1.4115533828735352,
+      "learning_rate": 3.87171461647301e-06,
+      "loss": 2.8187,
+      "step": 570500
+    },
+    {
+      "epoch": 24.069468448341272,
+      "grad_norm": 1.411168098449707,
+      "learning_rate": 3.7859623547571067e-06,
+      "loss": 2.8134,
+      "step": 571000
+    },
+    {
+      "epoch": 24.090545040677824,
+      "grad_norm": 1.4164671897888184,
+      "learning_rate": 3.7002100930412046e-06,
+      "loss": 2.812,
+      "step": 571500
+    },
+    {
+      "epoch": 24.111621633014373,
+      "grad_norm": 1.4191818237304688,
+      "learning_rate": 3.614457831325301e-06,
+      "loss": 2.8208,
+      "step": 572000
+    },
+    {
+      "epoch": 24.132698225350925,
+      "grad_norm": 1.4113930463790894,
+      "learning_rate": 3.5287055696093987e-06,
+      "loss": 2.8165,
+      "step": 572500
+    },
+    {
+      "epoch": 24.153774817687477,
+      "grad_norm": 1.3951871395111084,
+      "learning_rate": 3.4429533078934957e-06,
+      "loss": 2.8155,
+      "step": 573000
+    },
+    {
+      "epoch": 24.174851410024026,
+      "grad_norm": 1.4092600345611572,
+      "learning_rate": 3.357201046177593e-06,
+      "loss": 2.8192,
+      "step": 573500
+    },
+    {
+      "epoch": 24.195928002360578,
+      "grad_norm": 1.3970448970794678,
+      "learning_rate": 3.2714487844616906e-06,
+      "loss": 2.8179,
+      "step": 574000
+    },
+    {
+      "epoch": 24.21700459469713,
+      "grad_norm": 1.4322218894958496,
+      "learning_rate": 3.1856965227457872e-06,
+      "loss": 2.8195,
+      "step": 574500
+    },
+    {
+      "epoch": 24.238081187033682,
+      "grad_norm": 1.3599910736083984,
+      "learning_rate": 3.099944261029885e-06,
+      "loss": 2.8195,
+      "step": 575000
+    },
+    {
+      "epoch": 24.25915777937023,
+      "grad_norm": 1.3936123847961426,
+      "learning_rate": 3.014191999313982e-06,
+      "loss": 2.8161,
+      "step": 575500
+    },
+    {
+      "epoch": 24.280234371706783,
+      "grad_norm": 1.4189889430999756,
+      "learning_rate": 2.928439737598079e-06,
+      "loss": 2.8124,
+      "step": 576000
+    },
+    {
+      "epoch": 24.301310964043335,
+      "grad_norm": 1.401649832725525,
+      "learning_rate": 2.8426874758821767e-06,
+      "loss": 2.8168,
+      "step": 576500
+    },
+    {
+      "epoch": 24.322387556379883,
+      "grad_norm": 1.4115220308303833,
+      "learning_rate": 2.7569352141662737e-06,
+      "loss": 2.8125,
+      "step": 577000
+    },
+    {
+      "epoch": 24.343464148716436,
+      "grad_norm": 1.4133317470550537,
+      "learning_rate": 2.6711829524503707e-06,
+      "loss": 2.8137,
+      "step": 577500
+    },
+    {
+      "epoch": 24.364540741052988,
+      "grad_norm": 1.4301518201828003,
+      "learning_rate": 2.585430690734468e-06,
+      "loss": 2.8129,
+      "step": 578000
+    },
+    {
+      "epoch": 24.385617333389536,
+      "grad_norm": 1.401780605316162,
+      "learning_rate": 2.4996784290185657e-06,
+      "loss": 2.8159,
+      "step": 578500
+    },
+    {
+      "epoch": 24.40669392572609,
+      "grad_norm": 1.3922075033187866,
+      "learning_rate": 2.4139261673026627e-06,
+      "loss": 2.8117,
+      "step": 579000
+    },
+    {
+      "epoch": 24.42777051806264,
+      "grad_norm": 1.3856008052825928,
+      "learning_rate": 2.3281739055867597e-06,
+      "loss": 2.8143,
+      "step": 579500
+    },
+    {
+      "epoch": 24.44884711039919,
+      "grad_norm": 1.4096307754516602,
+      "learning_rate": 2.242421643870857e-06,
+      "loss": 2.8153,
+      "step": 580000
+    },
+    {
+      "epoch": 24.44884711039919,
+      "eval_accuracy": 0.7334683934502487,
+      "eval_loss": 2.5841891765594482,
+      "eval_runtime": 268.4793,
+      "eval_samples_per_second": 449.591,
+      "eval_steps_per_second": 4.686,
+      "step": 580000
+    },
+    {
+      "epoch": 24.46992370273574,
+      "grad_norm": 1.4350323677062988,
+      "learning_rate": 2.1566693821549547e-06,
+      "loss": 2.8122,
+      "step": 580500
+    },
+    {
+      "epoch": 24.491000295072293,
+      "grad_norm": 1.4005749225616455,
+      "learning_rate": 2.0709171204390517e-06,
+      "loss": 2.8173,
+      "step": 581000
+    },
+    {
+      "epoch": 24.512076887408845,
+      "grad_norm": 1.4624691009521484,
+      "learning_rate": 1.985164858723149e-06,
+      "loss": 2.8113,
+      "step": 581500
+    },
+    {
+      "epoch": 24.533153479745394,
+      "grad_norm": 1.4141383171081543,
+      "learning_rate": 1.8994125970072462e-06,
+      "loss": 2.8185,
+      "step": 582000
+    },
+    {
+      "epoch": 24.554230072081946,
+      "grad_norm": 1.4098459482192993,
+      "learning_rate": 1.8136603352913432e-06,
+      "loss": 2.8131,
+      "step": 582500
+    },
+    {
+      "epoch": 24.5753066644185,
+      "grad_norm": 1.3585373163223267,
+      "learning_rate": 1.7279080735754405e-06,
+      "loss": 2.8164,
+      "step": 583000
+    },
+    {
+      "epoch": 24.596383256755047,
+      "grad_norm": 1.4468103647232056,
+      "learning_rate": 1.642155811859538e-06,
+      "loss": 2.8165,
+      "step": 583500
+    },
+    {
+      "epoch": 24.6174598490916,
+      "grad_norm": 1.4108824729919434,
+      "learning_rate": 1.5564035501436352e-06,
+      "loss": 2.8147,
+      "step": 584000
+    },
+    {
+      "epoch": 24.63853644142815,
+      "grad_norm": 1.3923559188842773,
+      "learning_rate": 1.4706512884277324e-06,
+      "loss": 2.8152,
+      "step": 584500
+    },
+    {
+      "epoch": 24.6596130337647,
+      "grad_norm": 1.3736648559570312,
+      "learning_rate": 1.3848990267118295e-06,
+      "loss": 2.8121,
+      "step": 585000
+    },
+    {
+      "epoch": 24.680689626101252,
+      "grad_norm": 1.414063811302185,
+      "learning_rate": 1.299146764995927e-06,
+      "loss": 2.814,
+      "step": 585500
+    },
+    {
+      "epoch": 24.701766218437804,
+      "grad_norm": 1.4846856594085693,
+      "learning_rate": 1.213394503280024e-06,
+      "loss": 2.8121,
+      "step": 586000
+    },
+    {
+      "epoch": 24.722842810774353,
+      "grad_norm": 1.4140371084213257,
+      "learning_rate": 1.1276422415641212e-06,
+      "loss": 2.8132,
+      "step": 586500
+    },
+    {
+      "epoch": 24.743919403110905,
+      "grad_norm": 1.423340082168579,
+      "learning_rate": 1.0418899798482187e-06,
+      "loss": 2.8161,
+      "step": 587000
+    },
+    {
+      "epoch": 24.764995995447457,
+      "grad_norm": 1.419755220413208,
+      "learning_rate": 9.561377181323157e-07,
+      "loss": 2.8118,
+      "step": 587500
+    },
+    {
+      "epoch": 24.786072587784005,
+      "grad_norm": 1.4093141555786133,
+      "learning_rate": 8.70385456416413e-07,
+      "loss": 2.813,
+      "step": 588000
+    },
+    {
+      "epoch": 24.807149180120557,
+      "grad_norm": 1.352102518081665,
+      "learning_rate": 7.846331947005103e-07,
+      "loss": 2.8138,
+      "step": 588500
+    },
+    {
+      "epoch": 24.82822577245711,
+      "grad_norm": 1.375110149383545,
+      "learning_rate": 6.988809329846075e-07,
+      "loss": 2.8112,
+      "step": 589000
+    },
+    {
+      "epoch": 24.84930236479366,
+      "grad_norm": 1.4400821924209595,
+      "learning_rate": 6.131286712687047e-07,
+      "loss": 2.815,
+      "step": 589500
+    },
+    {
+      "epoch": 24.87037895713021,
+      "grad_norm": 1.4727100133895874,
+      "learning_rate": 5.27376409552802e-07,
+      "loss": 2.8117,
+      "step": 590000
+    },
+    {
+      "epoch": 24.87037895713021,
+      "eval_accuracy": 0.733197881502718,
+      "eval_loss": 2.584768056869507,
+      "eval_runtime": 279.7677,
+      "eval_samples_per_second": 431.451,
+      "eval_steps_per_second": 4.497,
+      "step": 590000
+    },
+    {
+      "epoch": 24.891455549466762,
+      "grad_norm": 1.3472633361816406,
+      "learning_rate": 4.4162414783689917e-07,
+      "loss": 2.8133,
+      "step": 590500
+    },
+    {
+      "epoch": 24.912532141803315,
+      "grad_norm": 1.3872286081314087,
+      "learning_rate": 3.558718861209964e-07,
+      "loss": 2.8135,
+      "step": 591000
+    },
+    {
+      "epoch": 24.933608734139863,
+      "grad_norm": 1.425608515739441,
+      "learning_rate": 2.701196244050937e-07,
+      "loss": 2.8143,
+      "step": 591500
+    },
+    {
+      "epoch": 24.954685326476415,
+      "grad_norm": 1.4403584003448486,
+      "learning_rate": 1.8436736268919094e-07,
+      "loss": 2.8131,
+      "step": 592000
+    },
+    {
+      "epoch": 24.975761918812967,
+      "grad_norm": 1.3558605909347534,
+      "learning_rate": 9.861510097328816e-08,
+      "loss": 2.8109,
+      "step": 592500
+    },
+    {
+      "epoch": 24.996838511149516,
+      "grad_norm": 1.4008831977844238,
+      "learning_rate": 1.2862839257385414e-08,
+      "loss": 2.8128,
+      "step": 593000
+    },
+    {
+      "epoch": 25.0,
+      "step": 593075,
+      "total_flos": 7.651144160057088e+18,
+      "train_loss": 3.137090846637658,
+      "train_runtime": 216876.098,
+      "train_samples_per_second": 262.516,
+      "train_steps_per_second": 2.735
+    }
+  ],
+  "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": 7.651144160057088e+18,
+  "train_batch_size": 96,
+  "trial_name": null,
+  "trial_params": null
+}