diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,2237 +1,8132 @@
 {
-  "best_metric": 0.9781976744186046,
-  "best_model_checkpoint": "mobilenet_v2_1.0_224-finetuned-plantdisease/checkpoint-2614",
-  "epoch": 19.965576592082616,
+  "best_metric": 0.9777191259513872,
+  "best_model_checkpoint": "mobilenet_v2_1.0_224-finetuned-plantdisease/checkpoint-9164",
+  "epoch": 9.995635093845483,
   "eval_steps": 500,
-  "global_step": 2900,
+  "global_step": 11450,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.06884681583476764,
-      "grad_norm": 5.361344337463379,
-      "learning_rate": 1.724137931034483e-06,
-      "loss": 2.7576,
+      "epoch": 0.008729812309035356,
+      "grad_norm": 5.370448589324951,
+      "learning_rate": 4.366812227074236e-07,
+      "loss": 3.7534,
       "step": 10
     },
     {
-      "epoch": 0.13769363166953527,
-      "grad_norm": 5.03896427154541,
-      "learning_rate": 3.448275862068966e-06,
-      "loss": 2.7085,
+      "epoch": 0.017459624618070713,
+      "grad_norm": 5.382869720458984,
+      "learning_rate": 8.733624454148472e-07,
+      "loss": 3.7447,
       "step": 20
     },
     {
-      "epoch": 0.20654044750430292,
-      "grad_norm": 5.099004745483398,
-      "learning_rate": 5.172413793103448e-06,
-      "loss": 2.7048,
+      "epoch": 0.026189436927106065,
+      "grad_norm": 5.5538249015808105,
+      "learning_rate": 1.3100436681222706e-06,
+      "loss": 3.7409,
       "step": 30
     },
     {
-      "epoch": 0.27538726333907054,
-      "grad_norm": 5.294215202331543,
-      "learning_rate": 6.896551724137932e-06,
-      "loss": 2.6637,
+      "epoch": 0.034919249236141425,
+      "grad_norm": 5.5582685470581055,
+      "learning_rate": 1.7467248908296944e-06,
+      "loss": 3.7325,
       "step": 40
     },
     {
-      "epoch": 0.3442340791738382,
-      "grad_norm": 5.142210483551025,
-      "learning_rate": 8.620689655172414e-06,
-      "loss": 2.6127,
+      "epoch": 0.04364906154517678,
+      "grad_norm": 5.354773998260498,
+      "learning_rate": 2.1834061135371177e-06,
+      "loss": 3.7135,
       "step": 50
     },
     {
-      "epoch": 0.41308089500860584,
-      "grad_norm": 5.095492839813232,
-      "learning_rate": 1.0344827586206897e-05,
-      "loss": 2.5458,
+      "epoch": 0.05237887385421213,
+      "grad_norm": 5.103704929351807,
+      "learning_rate": 2.6200873362445413e-06,
+      "loss": 3.714,
       "step": 60
     },
     {
-      "epoch": 0.4819277108433735,
-      "grad_norm": 4.894885063171387,
-      "learning_rate": 1.206896551724138e-05,
-      "loss": 2.4968,
+      "epoch": 0.06110868616324749,
+      "grad_norm": 5.359903812408447,
+      "learning_rate": 3.056768558951965e-06,
+      "loss": 3.6754,
       "step": 70
     },
     {
-      "epoch": 0.5507745266781411,
-      "grad_norm": 4.9816365242004395,
-      "learning_rate": 1.3793103448275863e-05,
-      "loss": 2.4015,
+      "epoch": 0.06983849847228285,
+      "grad_norm": 5.380093097686768,
+      "learning_rate": 3.493449781659389e-06,
+      "loss": 3.6767,
       "step": 80
     },
     {
-      "epoch": 0.6196213425129088,
-      "grad_norm": 4.671796798706055,
-      "learning_rate": 1.5517241379310346e-05,
-      "loss": 2.3308,
+      "epoch": 0.0785683107813182,
+      "grad_norm": 5.455582141876221,
+      "learning_rate": 3.930131004366813e-06,
+      "loss": 3.6431,
       "step": 90
     },
     {
-      "epoch": 0.6884681583476764,
-      "grad_norm": 4.635168552398682,
-      "learning_rate": 1.7241379310344828e-05,
-      "loss": 2.2212,
+      "epoch": 0.08729812309035356,
+      "grad_norm": 5.4371490478515625,
+      "learning_rate": 4.3668122270742355e-06,
+      "loss": 3.6285,
       "step": 100
     },
     {
-      "epoch": 0.7573149741824441,
-      "grad_norm": 4.575173854827881,
-      "learning_rate": 1.896551724137931e-05,
-      "loss": 2.1272,
+      "epoch": 0.09602793539938892,
+      "grad_norm": 5.358748435974121,
+      "learning_rate": 4.80349344978166e-06,
+      "loss": 3.594,
       "step": 110
     },
     {
-      "epoch": 0.8261617900172117,
-      "grad_norm": 4.4043660163879395,
-      "learning_rate": 2.0689655172413793e-05,
-      "loss": 2.0085,
+      "epoch": 0.10475774770842426,
+      "grad_norm": 5.466850757598877,
+      "learning_rate": 5.240174672489083e-06,
+      "loss": 3.5717,
       "step": 120
     },
     {
-      "epoch": 0.8950086058519794,
-      "grad_norm": 4.405401229858398,
-      "learning_rate": 2.2413793103448276e-05,
-      "loss": 1.9442,
+      "epoch": 0.11348756001745962,
+      "grad_norm": 5.143874168395996,
+      "learning_rate": 5.676855895196507e-06,
+      "loss": 3.5456,
       "step": 130
     },
     {
-      "epoch": 0.963855421686747,
-      "grad_norm": 4.29350471496582,
-      "learning_rate": 2.413793103448276e-05,
-      "loss": 1.7982,
+      "epoch": 0.12221737232649497,
+      "grad_norm": 5.1142683029174805,
+      "learning_rate": 6.11353711790393e-06,
+      "loss": 3.5045,
       "step": 140
     },
     {
-      "epoch": 0.9982788296041308,
-      "eval_accuracy": 0.40358527131782945,
-      "eval_loss": 1.9824707508087158,
-      "eval_runtime": 8.0421,
-      "eval_samples_per_second": 256.65,
-      "eval_steps_per_second": 8.082,
-      "step": 145
-    },
-    {
-      "epoch": 1.0327022375215147,
-      "grad_norm": 4.268815040588379,
-      "learning_rate": 2.5862068965517244e-05,
-      "loss": 1.7064,
+      "epoch": 0.13094718463553034,
+      "grad_norm": 5.174585819244385,
+      "learning_rate": 6.550218340611354e-06,
+      "loss": 3.452,
       "step": 150
     },
     {
-      "epoch": 1.1015490533562822,
-      "grad_norm": 4.16851806640625,
-      "learning_rate": 2.7586206896551727e-05,
-      "loss": 1.5684,
+      "epoch": 0.1396769969445657,
+      "grad_norm": 5.233867168426514,
+      "learning_rate": 6.986899563318778e-06,
+      "loss": 3.4269,
       "step": 160
     },
     {
-      "epoch": 1.1703958691910499,
-      "grad_norm": 4.322150230407715,
-      "learning_rate": 2.9310344827586206e-05,
-      "loss": 1.4844,
+      "epoch": 0.14840680925360106,
+      "grad_norm": 4.996443271636963,
+      "learning_rate": 7.423580786026202e-06,
+      "loss": 3.3866,
       "step": 170
     },
     {
-      "epoch": 1.2392426850258176,
-      "grad_norm": 3.958456039428711,
-      "learning_rate": 3.103448275862069e-05,
-      "loss": 1.3968,
+      "epoch": 0.1571366215626364,
+      "grad_norm": 5.243433475494385,
+      "learning_rate": 7.860262008733626e-06,
+      "loss": 3.3293,
       "step": 180
     },
     {
-      "epoch": 1.3080895008605853,
-      "grad_norm": 3.7061522006988525,
-      "learning_rate": 3.275862068965517e-05,
-      "loss": 1.2972,
+      "epoch": 0.16586643387167177,
+      "grad_norm": 5.2202911376953125,
+      "learning_rate": 8.296943231441049e-06,
+      "loss": 3.2767,
       "step": 190
     },
     {
-      "epoch": 1.3769363166953528,
-      "grad_norm": 3.8323745727539062,
-      "learning_rate": 3.4482758620689657e-05,
-      "loss": 1.1544,
+      "epoch": 0.17459624618070713,
+      "grad_norm": 4.891767501831055,
+      "learning_rate": 8.733624454148471e-06,
+      "loss": 3.2698,
       "step": 200
     },
     {
-      "epoch": 1.4457831325301205,
-      "grad_norm": 3.52496600151062,
-      "learning_rate": 3.620689655172414e-05,
-      "loss": 1.0645,
+      "epoch": 0.18332605848974248,
+      "grad_norm": 4.960419654846191,
+      "learning_rate": 9.170305676855896e-06,
+      "loss": 3.2333,
       "step": 210
     },
     {
-      "epoch": 1.5146299483648882,
-      "grad_norm": 3.744525194168091,
-      "learning_rate": 3.793103448275862e-05,
-      "loss": 1.0238,
+      "epoch": 0.19205587079877784,
+      "grad_norm": 4.834344387054443,
+      "learning_rate": 9.60698689956332e-06,
+      "loss": 3.1378,
       "step": 220
     },
     {
-      "epoch": 1.5834767641996557,
-      "grad_norm": 3.494678497314453,
-      "learning_rate": 3.965517241379311e-05,
-      "loss": 0.9237,
+      "epoch": 0.2007856831078132,
+      "grad_norm": 5.046371936798096,
+      "learning_rate": 1.0043668122270743e-05,
+      "loss": 3.0687,
       "step": 230
     },
     {
-      "epoch": 1.6523235800344234,
-      "grad_norm": 3.6277663707733154,
-      "learning_rate": 4.1379310344827587e-05,
-      "loss": 0.881,
+      "epoch": 0.20951549541684852,
+      "grad_norm": 5.039073467254639,
+      "learning_rate": 1.0480349344978165e-05,
+      "loss": 3.0194,
       "step": 240
     },
     {
-      "epoch": 1.721170395869191,
-      "grad_norm": 2.9900975227355957,
-      "learning_rate": 4.3103448275862066e-05,
-      "loss": 0.7883,
+      "epoch": 0.21824530772588388,
+      "grad_norm": 5.1495771408081055,
+      "learning_rate": 1.091703056768559e-05,
+      "loss": 2.9677,
       "step": 250
     },
     {
-      "epoch": 1.7900172117039586,
-      "grad_norm": 3.553596258163452,
-      "learning_rate": 4.482758620689655e-05,
-      "loss": 0.7713,
+      "epoch": 0.22697512003491924,
+      "grad_norm": 4.857588768005371,
+      "learning_rate": 1.1353711790393014e-05,
+      "loss": 2.9015,
       "step": 260
     },
     {
-      "epoch": 1.8588640275387265,
-      "grad_norm": 3.206496477127075,
-      "learning_rate": 4.655172413793104e-05,
-      "loss": 0.7063,
+      "epoch": 0.2357049323439546,
+      "grad_norm": 4.682951927185059,
+      "learning_rate": 1.1790393013100438e-05,
+      "loss": 2.8179,
       "step": 270
     },
     {
-      "epoch": 1.927710843373494,
-      "grad_norm": 3.446793556213379,
-      "learning_rate": 4.827586206896552e-05,
-      "loss": 0.6517,
+      "epoch": 0.24443474465298995,
+      "grad_norm": 4.9485602378845215,
+      "learning_rate": 1.222707423580786e-05,
+      "loss": 2.8279,
       "step": 280
     },
     {
-      "epoch": 1.9965576592082617,
-      "grad_norm": 3.7154784202575684,
-      "learning_rate": 5e-05,
-      "loss": 0.6137,
-      "step": 290
-    },
-    {
-      "epoch": 1.9965576592082617,
-      "eval_accuracy": 0.6414728682170543,
-      "eval_loss": 1.1130067110061646,
-      "eval_runtime": 7.6055,
-      "eval_samples_per_second": 271.384,
-      "eval_steps_per_second": 8.546,
+      "epoch": 0.25316455696202533,
+      "grad_norm": 4.801344871520996,
+      "learning_rate": 1.2663755458515283e-05,
+      "loss": 2.7402,
       "step": 290
     },
     {
-      "epoch": 2.0654044750430294,
-      "grad_norm": 3.542126417160034,
-      "learning_rate": 4.980842911877395e-05,
-      "loss": 0.6058,
+      "epoch": 0.2618943692710607,
+      "grad_norm": 4.586411952972412,
+      "learning_rate": 1.3100436681222708e-05,
+      "loss": 2.7108,
       "step": 300
     },
     {
-      "epoch": 2.134251290877797,
-      "grad_norm": 3.6798830032348633,
-      "learning_rate": 4.96168582375479e-05,
-      "loss": 0.5967,
+      "epoch": 0.27062418158009605,
+      "grad_norm": 4.808084487915039,
+      "learning_rate": 1.3537117903930133e-05,
+      "loss": 2.585,
       "step": 310
     },
     {
-      "epoch": 2.2030981067125643,
-      "grad_norm": 3.2142186164855957,
-      "learning_rate": 4.9425287356321845e-05,
-      "loss": 0.5376,
+      "epoch": 0.2793539938891314,
+      "grad_norm": 4.550442695617676,
+      "learning_rate": 1.3973799126637555e-05,
+      "loss": 2.6376,
       "step": 320
     },
     {
-      "epoch": 2.2719449225473323,
-      "grad_norm": 3.1692137718200684,
-      "learning_rate": 4.9233716475095786e-05,
-      "loss": 0.5135,
+      "epoch": 0.28808380619816676,
+      "grad_norm": 4.389446258544922,
+      "learning_rate": 1.4410480349344979e-05,
+      "loss": 2.511,
       "step": 330
     },
     {
-      "epoch": 2.3407917383820998,
-      "grad_norm": 2.785217523574829,
-      "learning_rate": 4.904214559386973e-05,
-      "loss": 0.4794,
+      "epoch": 0.2968136185072021,
+      "grad_norm": 4.278360366821289,
+      "learning_rate": 1.4847161572052404e-05,
+      "loss": 2.4379,
       "step": 340
     },
     {
-      "epoch": 2.4096385542168672,
-      "grad_norm": 3.252037763595581,
-      "learning_rate": 4.885057471264368e-05,
-      "loss": 0.5012,
+      "epoch": 0.30554343081623747,
+      "grad_norm": 4.738358497619629,
+      "learning_rate": 1.5283842794759826e-05,
+      "loss": 2.387,
       "step": 350
     },
     {
-      "epoch": 2.478485370051635,
-      "grad_norm": 2.9605207443237305,
-      "learning_rate": 4.865900383141763e-05,
-      "loss": 0.4613,
+      "epoch": 0.3142732431252728,
+      "grad_norm": 4.373560428619385,
+      "learning_rate": 1.572052401746725e-05,
+      "loss": 2.3534,
       "step": 360
     },
     {
-      "epoch": 2.5473321858864026,
-      "grad_norm": 3.263587474822998,
-      "learning_rate": 4.846743295019157e-05,
-      "loss": 0.4191,
+      "epoch": 0.3230030554343082,
+      "grad_norm": 4.679076671600342,
+      "learning_rate": 1.6157205240174673e-05,
+      "loss": 2.321,
       "step": 370
     },
     {
-      "epoch": 2.6161790017211706,
-      "grad_norm": 3.321648120880127,
-      "learning_rate": 4.827586206896552e-05,
-      "loss": 0.4206,
+      "epoch": 0.33173286774334354,
+      "grad_norm": 4.373522758483887,
+      "learning_rate": 1.6593886462882098e-05,
+      "loss": 2.2783,
       "step": 380
     },
     {
-      "epoch": 2.685025817555938,
-      "grad_norm": 2.7976036071777344,
-      "learning_rate": 4.8084291187739464e-05,
-      "loss": 0.462,
+      "epoch": 0.3404626800523789,
+      "grad_norm": 4.47551155090332,
+      "learning_rate": 1.703056768558952e-05,
+      "loss": 2.1886,
       "step": 390
     },
     {
-      "epoch": 2.7538726333907055,
-      "grad_norm": 3.0656702518463135,
-      "learning_rate": 4.789272030651341e-05,
-      "loss": 0.41,
+      "epoch": 0.34919249236141425,
+      "grad_norm": 4.3290181159973145,
+      "learning_rate": 1.7467248908296942e-05,
+      "loss": 2.1986,
       "step": 400
     },
     {
-      "epoch": 2.8227194492254735,
-      "grad_norm": 3.332609176635742,
-      "learning_rate": 4.770114942528736e-05,
-      "loss": 0.4012,
+      "epoch": 0.3579223046704496,
+      "grad_norm": 4.084609031677246,
+      "learning_rate": 1.7903930131004367e-05,
+      "loss": 2.0736,
       "step": 410
     },
     {
-      "epoch": 2.891566265060241,
-      "grad_norm": 3.5820395946502686,
-      "learning_rate": 4.7509578544061307e-05,
-      "loss": 0.3905,
+      "epoch": 0.36665211697948497,
+      "grad_norm": 4.143791198730469,
+      "learning_rate": 1.8340611353711792e-05,
+      "loss": 2.045,
       "step": 420
     },
     {
-      "epoch": 2.960413080895009,
-      "grad_norm": 3.3419365882873535,
-      "learning_rate": 4.7318007662835254e-05,
-      "loss": 0.4176,
+      "epoch": 0.3753819292885203,
+      "grad_norm": 4.159194469451904,
+      "learning_rate": 1.8777292576419214e-05,
+      "loss": 1.9542,
       "step": 430
     },
     {
-      "epoch": 2.9948364888123926,
-      "eval_accuracy": 0.8468992248062015,
-      "eval_loss": 0.48874419927597046,
-      "eval_runtime": 7.9016,
-      "eval_samples_per_second": 261.213,
-      "eval_steps_per_second": 8.226,
-      "step": 435
-    },
-    {
-      "epoch": 3.0292598967297764,
-      "grad_norm": 3.0542681217193604,
-      "learning_rate": 4.7126436781609195e-05,
-      "loss": 0.4024,
+      "epoch": 0.3841117415975557,
+      "grad_norm": 4.298882007598877,
+      "learning_rate": 1.921397379912664e-05,
+      "loss": 1.9863,
       "step": 440
     },
     {
-      "epoch": 3.098106712564544,
-      "grad_norm": 2.760667562484741,
-      "learning_rate": 4.693486590038315e-05,
-      "loss": 0.3706,
+      "epoch": 0.39284155390659103,
+      "grad_norm": 4.460660934448242,
+      "learning_rate": 1.965065502183406e-05,
+      "loss": 1.9066,
       "step": 450
     },
     {
-      "epoch": 3.1669535283993113,
-      "grad_norm": 3.2174575328826904,
-      "learning_rate": 4.674329501915709e-05,
-      "loss": 0.3509,
+      "epoch": 0.4015713662156264,
+      "grad_norm": 4.154500484466553,
+      "learning_rate": 2.0087336244541487e-05,
+      "loss": 1.7901,
       "step": 460
     },
     {
-      "epoch": 3.2358003442340793,
-      "grad_norm": 3.6685619354248047,
-      "learning_rate": 4.655172413793104e-05,
-      "loss": 0.3613,
+      "epoch": 0.41030117852466175,
+      "grad_norm": 4.019309043884277,
+      "learning_rate": 2.052401746724891e-05,
+      "loss": 1.8314,
       "step": 470
     },
     {
-      "epoch": 3.3046471600688467,
-      "grad_norm": 3.4590651988983154,
-      "learning_rate": 4.6360153256704985e-05,
-      "loss": 0.346,
+      "epoch": 0.41903099083369705,
+      "grad_norm": 4.08663272857666,
+      "learning_rate": 2.096069868995633e-05,
+      "loss": 1.7671,
       "step": 480
     },
     {
-      "epoch": 3.3734939759036147,
-      "grad_norm": 3.0743446350097656,
-      "learning_rate": 4.616858237547893e-05,
-      "loss": 0.3356,
+      "epoch": 0.4277608031427324,
+      "grad_norm": 4.213662624359131,
+      "learning_rate": 2.1397379912663756e-05,
+      "loss": 1.7467,
       "step": 490
     },
     {
-      "epoch": 3.442340791738382,
-      "grad_norm": 3.5646955966949463,
-      "learning_rate": 4.597701149425287e-05,
-      "loss": 0.3404,
+      "epoch": 0.43649061545176776,
+      "grad_norm": 3.9979076385498047,
+      "learning_rate": 2.183406113537118e-05,
+      "loss": 1.6146,
       "step": 500
     },
     {
-      "epoch": 3.5111876075731496,
-      "grad_norm": 3.2671332359313965,
-      "learning_rate": 4.578544061302682e-05,
-      "loss": 0.3419,
+      "epoch": 0.4452204277608031,
+      "grad_norm": 3.973045825958252,
+      "learning_rate": 2.2270742358078603e-05,
+      "loss": 1.6054,
       "step": 510
     },
     {
-      "epoch": 3.580034423407917,
-      "grad_norm": 2.7561419010162354,
-      "learning_rate": 4.559386973180077e-05,
-      "loss": 0.3306,
+      "epoch": 0.4539502400698385,
+      "grad_norm": 4.335777282714844,
+      "learning_rate": 2.2707423580786028e-05,
+      "loss": 1.6053,
       "step": 520
     },
     {
-      "epoch": 3.648881239242685,
-      "grad_norm": 3.1499881744384766,
-      "learning_rate": 4.5402298850574716e-05,
-      "loss": 0.3575,
+      "epoch": 0.46268005237887383,
+      "grad_norm": 4.511986255645752,
+      "learning_rate": 2.3144104803493453e-05,
+      "loss": 1.5458,
       "step": 530
     },
     {
-      "epoch": 3.7177280550774525,
-      "grad_norm": 2.576195240020752,
-      "learning_rate": 4.5210727969348656e-05,
-      "loss": 0.2954,
+      "epoch": 0.4714098646879092,
+      "grad_norm": 3.8047099113464355,
+      "learning_rate": 2.3580786026200875e-05,
+      "loss": 1.5328,
       "step": 540
     },
     {
-      "epoch": 3.7865748709122204,
-      "grad_norm": 2.685713529586792,
-      "learning_rate": 4.501915708812261e-05,
-      "loss": 0.2971,
+      "epoch": 0.48013967699694454,
+      "grad_norm": 3.791404962539673,
+      "learning_rate": 2.4017467248908297e-05,
+      "loss": 1.4993,
       "step": 550
     },
     {
-      "epoch": 3.855421686746988,
-      "grad_norm": 3.1608219146728516,
-      "learning_rate": 4.482758620689655e-05,
-      "loss": 0.3057,
+      "epoch": 0.4888694893059799,
+      "grad_norm": 4.048672199249268,
+      "learning_rate": 2.445414847161572e-05,
+      "loss": 1.4274,
       "step": 560
     },
     {
-      "epoch": 3.9242685025817554,
-      "grad_norm": 3.3374826908111572,
-      "learning_rate": 4.46360153256705e-05,
-      "loss": 0.2815,
+      "epoch": 0.49759930161501525,
+      "grad_norm": 4.073417663574219,
+      "learning_rate": 2.4890829694323144e-05,
+      "loss": 1.3728,
       "step": 570
     },
     {
-      "epoch": 3.9931153184165233,
-      "grad_norm": 3.0209603309631348,
-      "learning_rate": 4.4444444444444447e-05,
-      "loss": 0.3107,
+      "epoch": 0.5063291139240507,
+      "grad_norm": 4.089795112609863,
+      "learning_rate": 2.5327510917030566e-05,
+      "loss": 1.3612,
       "step": 580
     },
     {
-      "epoch": 4.0,
-      "eval_accuracy": 0.8943798449612403,
-      "eval_loss": 0.34142178297042847,
-      "eval_runtime": 7.9804,
-      "eval_samples_per_second": 258.634,
-      "eval_steps_per_second": 8.145,
-      "step": 581
-    },
-    {
-      "epoch": 4.061962134251291,
-      "grad_norm": 2.7664377689361572,
-      "learning_rate": 4.4252873563218394e-05,
-      "loss": 0.2689,
+      "epoch": 0.515058926233086,
+      "grad_norm": 3.6688883304595947,
+      "learning_rate": 2.576419213973799e-05,
+      "loss": 1.2784,
       "step": 590
     },
     {
-      "epoch": 4.130808950086059,
-      "grad_norm": 2.80204439163208,
-      "learning_rate": 4.406130268199234e-05,
-      "loss": 0.3003,
+      "epoch": 0.5237887385421214,
+      "grad_norm": 3.9120841026306152,
+      "learning_rate": 2.6200873362445416e-05,
+      "loss": 1.2878,
       "step": 600
     },
     {
-      "epoch": 4.199655765920826,
-      "grad_norm": 3.0250327587127686,
-      "learning_rate": 4.386973180076628e-05,
-      "loss": 0.2768,
+      "epoch": 0.5325185508511567,
+      "grad_norm": 3.715569257736206,
+      "learning_rate": 2.663755458515284e-05,
+      "loss": 1.1857,
       "step": 610
     },
     {
-      "epoch": 4.268502581755594,
-      "grad_norm": 3.8367512226104736,
-      "learning_rate": 4.367816091954024e-05,
-      "loss": 0.2582,
+      "epoch": 0.5412483631601921,
+      "grad_norm": 3.895285129547119,
+      "learning_rate": 2.7074235807860267e-05,
+      "loss": 1.1727,
       "step": 620
     },
     {
-      "epoch": 4.337349397590361,
-      "grad_norm": 3.3538951873779297,
-      "learning_rate": 4.348659003831418e-05,
-      "loss": 0.2849,
+      "epoch": 0.5499781754692274,
+      "grad_norm": 3.671576738357544,
+      "learning_rate": 2.7510917030567685e-05,
+      "loss": 1.1431,
       "step": 630
     },
     {
-      "epoch": 4.406196213425129,
-      "grad_norm": 3.2220115661621094,
-      "learning_rate": 4.3295019157088125e-05,
-      "loss": 0.2744,
+      "epoch": 0.5587079877782628,
+      "grad_norm": 4.0363993644714355,
+      "learning_rate": 2.794759825327511e-05,
+      "loss": 1.1228,
       "step": 640
     },
     {
-      "epoch": 4.475043029259897,
-      "grad_norm": 3.196542263031006,
-      "learning_rate": 4.3103448275862066e-05,
-      "loss": 0.2744,
+      "epoch": 0.5674378000872982,
+      "grad_norm": 4.259347915649414,
+      "learning_rate": 2.8384279475982532e-05,
+      "loss": 1.1352,
       "step": 650
     },
     {
-      "epoch": 4.5438898450946645,
-      "grad_norm": 3.07588791847229,
-      "learning_rate": 4.291187739463602e-05,
-      "loss": 0.2766,
+      "epoch": 0.5761676123963335,
+      "grad_norm": 3.2646217346191406,
+      "learning_rate": 2.8820960698689958e-05,
+      "loss": 1.0213,
       "step": 660
     },
     {
-      "epoch": 4.612736660929432,
-      "grad_norm": 3.722421169281006,
-      "learning_rate": 4.272030651340996e-05,
-      "loss": 0.2699,
+      "epoch": 0.5848974247053689,
+      "grad_norm": 3.9578239917755127,
+      "learning_rate": 2.9257641921397383e-05,
+      "loss": 1.0741,
       "step": 670
     },
     {
-      "epoch": 4.6815834767641995,
-      "grad_norm": 2.5164051055908203,
-      "learning_rate": 4.252873563218391e-05,
-      "loss": 0.2401,
+      "epoch": 0.5936272370144042,
+      "grad_norm": 3.894042730331421,
+      "learning_rate": 2.9694323144104808e-05,
+      "loss": 1.0305,
       "step": 680
     },
     {
-      "epoch": 4.750430292598967,
-      "grad_norm": 3.1823108196258545,
-      "learning_rate": 4.2337164750957856e-05,
-      "loss": 0.2293,
+      "epoch": 0.6023570493234396,
+      "grad_norm": 4.07789945602417,
+      "learning_rate": 3.0131004366812227e-05,
+      "loss": 1.0185,
       "step": 690
     },
     {
-      "epoch": 4.8192771084337345,
-      "grad_norm": 3.4586544036865234,
-      "learning_rate": 4.21455938697318e-05,
-      "loss": 0.27,
+      "epoch": 0.6110868616324749,
+      "grad_norm": 4.043015003204346,
+      "learning_rate": 3.056768558951965e-05,
+      "loss": 1.0215,
       "step": 700
     },
     {
-      "epoch": 4.888123924268503,
-      "grad_norm": 3.251978635787964,
-      "learning_rate": 4.195402298850575e-05,
-      "loss": 0.2551,
+      "epoch": 0.6198166739415103,
+      "grad_norm": 3.8258657455444336,
+      "learning_rate": 3.1004366812227074e-05,
+      "loss": 0.9917,
       "step": 710
     },
     {
-      "epoch": 4.95697074010327,
-      "grad_norm": 2.606138229370117,
-      "learning_rate": 4.17624521072797e-05,
-      "loss": 0.2255,
+      "epoch": 0.6285464862505457,
+      "grad_norm": 3.633079767227173,
+      "learning_rate": 3.14410480349345e-05,
+      "loss": 0.9212,
       "step": 720
     },
     {
-      "epoch": 4.998278829604131,
-      "eval_accuracy": 0.9123062015503876,
-      "eval_loss": 0.27319762110710144,
-      "eval_runtime": 7.8624,
-      "eval_samples_per_second": 262.515,
-      "eval_steps_per_second": 8.267,
-      "step": 726
-    },
-    {
-      "epoch": 5.025817555938038,
-      "grad_norm": 2.684196949005127,
-      "learning_rate": 4.1570881226053646e-05,
-      "loss": 0.2487,
+      "epoch": 0.637276298559581,
+      "grad_norm": 4.321951866149902,
+      "learning_rate": 3.1877729257641924e-05,
+      "loss": 0.9052,
       "step": 730
     },
     {
-      "epoch": 5.094664371772805,
-      "grad_norm": 2.931312084197998,
-      "learning_rate": 4.1379310344827587e-05,
-      "loss": 0.2314,
+      "epoch": 0.6460061108686164,
+      "grad_norm": 3.5607919692993164,
+      "learning_rate": 3.2314410480349346e-05,
+      "loss": 0.894,
       "step": 740
     },
     {
-      "epoch": 5.163511187607573,
-      "grad_norm": 3.0064966678619385,
-      "learning_rate": 4.1187739463601534e-05,
-      "loss": 0.2293,
+      "epoch": 0.6547359231776517,
+      "grad_norm": 3.680312156677246,
+      "learning_rate": 3.275109170305677e-05,
+      "loss": 0.8355,
       "step": 750
     },
     {
-      "epoch": 5.232358003442341,
-      "grad_norm": 3.3251922130584717,
-      "learning_rate": 4.099616858237548e-05,
-      "loss": 0.2388,
+      "epoch": 0.6634657354866871,
+      "grad_norm": 4.108597755432129,
+      "learning_rate": 3.3187772925764197e-05,
+      "loss": 0.7819,
       "step": 760
     },
     {
-      "epoch": 5.301204819277109,
-      "grad_norm": 1.7967256307601929,
-      "learning_rate": 4.080459770114943e-05,
-      "loss": 0.2169,
+      "epoch": 0.6721955477957224,
+      "grad_norm": 4.298781871795654,
+      "learning_rate": 3.362445414847162e-05,
+      "loss": 0.823,
       "step": 770
     },
     {
-      "epoch": 5.370051635111876,
-      "grad_norm": 3.74770450592041,
-      "learning_rate": 4.061302681992337e-05,
-      "loss": 0.2096,
+      "epoch": 0.6809253601047578,
+      "grad_norm": 3.7209835052490234,
+      "learning_rate": 3.406113537117904e-05,
+      "loss": 0.8231,
       "step": 780
     },
     {
-      "epoch": 5.438898450946644,
-      "grad_norm": 3.072739601135254,
-      "learning_rate": 4.0421455938697324e-05,
-      "loss": 0.2047,
+      "epoch": 0.6896551724137931,
+      "grad_norm": 3.6912119388580322,
+      "learning_rate": 3.449781659388647e-05,
+      "loss": 0.7911,
       "step": 790
     },
     {
-      "epoch": 5.507745266781411,
-      "grad_norm": 2.96258807182312,
-      "learning_rate": 4.0229885057471265e-05,
-      "loss": 0.2313,
+      "epoch": 0.6983849847228285,
+      "grad_norm": 3.601679563522339,
+      "learning_rate": 3.4934497816593884e-05,
+      "loss": 0.7401,
       "step": 800
     },
     {
-      "epoch": 5.576592082616179,
-      "grad_norm": 2.5839080810546875,
-      "learning_rate": 4.003831417624521e-05,
-      "loss": 0.2112,
+      "epoch": 0.7071147970318639,
+      "grad_norm": 3.856973171234131,
+      "learning_rate": 3.537117903930131e-05,
+      "loss": 0.7565,
       "step": 810
     },
     {
-      "epoch": 5.645438898450947,
-      "grad_norm": 3.1343889236450195,
-      "learning_rate": 3.984674329501916e-05,
-      "loss": 0.2482,
+      "epoch": 0.7158446093408992,
+      "grad_norm": 4.325893878936768,
+      "learning_rate": 3.5807860262008734e-05,
+      "loss": 0.7557,
       "step": 820
     },
     {
-      "epoch": 5.714285714285714,
-      "grad_norm": 2.240661382675171,
-      "learning_rate": 3.965517241379311e-05,
-      "loss": 0.2146,
+      "epoch": 0.7245744216499346,
+      "grad_norm": 4.065762996673584,
+      "learning_rate": 3.624454148471616e-05,
+      "loss": 0.7348,
       "step": 830
     },
     {
-      "epoch": 5.783132530120482,
-      "grad_norm": 3.1076242923736572,
-      "learning_rate": 3.9463601532567055e-05,
-      "loss": 0.2234,
+      "epoch": 0.7333042339589699,
+      "grad_norm": 3.6065175533294678,
+      "learning_rate": 3.6681222707423585e-05,
+      "loss": 0.7162,
       "step": 840
     },
     {
-      "epoch": 5.851979345955249,
-      "grad_norm": 2.9466423988342285,
-      "learning_rate": 3.9272030651340996e-05,
-      "loss": 0.2301,
+      "epoch": 0.7420340462680053,
+      "grad_norm": 3.7237186431884766,
+      "learning_rate": 3.711790393013101e-05,
+      "loss": 0.6657,
       "step": 850
     },
     {
-      "epoch": 5.920826161790017,
-      "grad_norm": 3.1047613620758057,
-      "learning_rate": 3.908045977011495e-05,
-      "loss": 0.2202,
+      "epoch": 0.7507638585770406,
+      "grad_norm": 3.3277764320373535,
+      "learning_rate": 3.755458515283843e-05,
+      "loss": 0.6956,
       "step": 860
     },
     {
-      "epoch": 5.989672977624785,
-      "grad_norm": 2.182670831680298,
-      "learning_rate": 3.888888888888889e-05,
-      "loss": 0.1833,
+      "epoch": 0.759493670886076,
+      "grad_norm": 4.210085868835449,
+      "learning_rate": 3.799126637554585e-05,
+      "loss": 0.6844,
       "step": 870
     },
     {
-      "epoch": 5.9965576592082614,
-      "eval_accuracy": 0.7582364341085271,
-      "eval_loss": 0.7461967468261719,
-      "eval_runtime": 7.912,
-      "eval_samples_per_second": 260.87,
-      "eval_steps_per_second": 8.215,
-      "step": 871
-    },
-    {
-      "epoch": 6.058519793459553,
-      "grad_norm": 2.8595638275146484,
-      "learning_rate": 3.869731800766284e-05,
-      "loss": 0.198,
+      "epoch": 0.7682234831951114,
+      "grad_norm": 3.41595458984375,
+      "learning_rate": 3.842794759825328e-05,
+      "loss": 0.6489,
       "step": 880
     },
     {
-      "epoch": 6.12736660929432,
-      "grad_norm": 1.812150478363037,
-      "learning_rate": 3.850574712643678e-05,
-      "loss": 0.17,
+      "epoch": 0.7769532955041467,
+      "grad_norm": 4.264623641967773,
+      "learning_rate": 3.88646288209607e-05,
+      "loss": 0.6868,
       "step": 890
     },
     {
-      "epoch": 6.196213425129088,
-      "grad_norm": 2.8384931087493896,
-      "learning_rate": 3.831417624521073e-05,
-      "loss": 0.2387,
+      "epoch": 0.7856831078131821,
+      "grad_norm": 3.9044318199157715,
+      "learning_rate": 3.930131004366812e-05,
+      "loss": 0.6235,
       "step": 900
     },
     {
-      "epoch": 6.265060240963855,
-      "grad_norm": 4.172893047332764,
-      "learning_rate": 3.8122605363984674e-05,
-      "loss": 0.2202,
+      "epoch": 0.7944129201222174,
+      "grad_norm": 3.767618179321289,
+      "learning_rate": 3.9737991266375545e-05,
+      "loss": 0.6105,
       "step": 910
     },
     {
-      "epoch": 6.333907056798623,
-      "grad_norm": 4.461463928222656,
-      "learning_rate": 3.793103448275862e-05,
-      "loss": 0.2501,
+      "epoch": 0.8031427324312528,
+      "grad_norm": 3.6234493255615234,
+      "learning_rate": 4.017467248908297e-05,
+      "loss": 0.6195,
       "step": 920
     },
     {
-      "epoch": 6.402753872633391,
-      "grad_norm": 3.430365562438965,
-      "learning_rate": 3.773946360153257e-05,
-      "loss": 0.185,
+      "epoch": 0.8118725447402881,
+      "grad_norm": 3.8668875694274902,
+      "learning_rate": 4.0611353711790395e-05,
+      "loss": 0.6341,
       "step": 930
     },
     {
-      "epoch": 6.4716006884681585,
-      "grad_norm": 3.3742434978485107,
-      "learning_rate": 3.7547892720306517e-05,
-      "loss": 0.2166,
+      "epoch": 0.8206023570493235,
+      "grad_norm": 3.911647081375122,
+      "learning_rate": 4.104803493449782e-05,
+      "loss": 0.6803,
       "step": 940
     },
     {
-      "epoch": 6.540447504302926,
-      "grad_norm": 3.7096402645111084,
-      "learning_rate": 3.735632183908046e-05,
-      "loss": 0.234,
+      "epoch": 0.8293321693583589,
+      "grad_norm": 4.5981950759887695,
+      "learning_rate": 4.1484716157205246e-05,
+      "loss": 0.5992,
       "step": 950
     },
     {
-      "epoch": 6.6092943201376935,
-      "grad_norm": 2.182164192199707,
-      "learning_rate": 3.716475095785441e-05,
-      "loss": 0.1852,
+      "epoch": 0.8380619816673941,
+      "grad_norm": 3.940392017364502,
+      "learning_rate": 4.192139737991266e-05,
+      "loss": 0.5896,
       "step": 960
     },
     {
-      "epoch": 6.678141135972461,
-      "grad_norm": 2.585125684738159,
-      "learning_rate": 3.697318007662835e-05,
-      "loss": 0.1947,
+      "epoch": 0.8467917939764295,
+      "grad_norm": 3.4633290767669678,
+      "learning_rate": 4.235807860262009e-05,
+      "loss": 0.5743,
       "step": 970
     },
     {
-      "epoch": 6.746987951807229,
-      "grad_norm": 2.5588345527648926,
-      "learning_rate": 3.67816091954023e-05,
-      "loss": 0.211,
+      "epoch": 0.8555216062854648,
+      "grad_norm": 3.471381187438965,
+      "learning_rate": 4.279475982532751e-05,
+      "loss": 0.5745,
       "step": 980
     },
     {
-      "epoch": 6.815834767641997,
-      "grad_norm": 3.0795257091522217,
-      "learning_rate": 3.659003831417625e-05,
-      "loss": 0.1967,
+      "epoch": 0.8642514185945002,
+      "grad_norm": 3.305868625640869,
+      "learning_rate": 4.323144104803494e-05,
+      "loss": 0.6056,
       "step": 990
     },
     {
-      "epoch": 6.884681583476764,
-      "grad_norm": 3.7196614742279053,
-      "learning_rate": 3.6398467432950195e-05,
-      "loss": 0.1836,
+      "epoch": 0.8729812309035355,
+      "grad_norm": 3.885556221008301,
+      "learning_rate": 4.366812227074236e-05,
+      "loss": 0.5638,
       "step": 1000
     },
     {
-      "epoch": 6.953528399311532,
-      "grad_norm": 2.3857243061065674,
-      "learning_rate": 3.620689655172414e-05,
-      "loss": 0.2062,
+      "epoch": 0.8817110432125709,
+      "grad_norm": 3.944361448287964,
+      "learning_rate": 4.4104803493449784e-05,
+      "loss": 0.524,
       "step": 1010
     },
     {
-      "epoch": 6.994836488812393,
-      "eval_accuracy": 0.8803294573643411,
-      "eval_loss": 0.3770907521247864,
-      "eval_runtime": 7.9957,
-      "eval_samples_per_second": 258.14,
-      "eval_steps_per_second": 8.129,
-      "step": 1016
-    },
-    {
-      "epoch": 7.022375215146299,
-      "grad_norm": 2.5683236122131348,
-      "learning_rate": 3.601532567049808e-05,
-      "loss": 0.1754,
+      "epoch": 0.8904408555216062,
+      "grad_norm": 3.539358377456665,
+      "learning_rate": 4.4541484716157205e-05,
+      "loss": 0.5493,
       "step": 1020
     },
     {
-      "epoch": 7.091222030981067,
-      "grad_norm": 2.679636001586914,
-      "learning_rate": 3.582375478927204e-05,
-      "loss": 0.1675,
+      "epoch": 0.8991706678306416,
+      "grad_norm": 3.6482980251312256,
+      "learning_rate": 4.497816593886463e-05,
+      "loss": 0.5493,
       "step": 1030
     },
     {
-      "epoch": 7.160068846815835,
-      "grad_norm": 2.5351173877716064,
-      "learning_rate": 3.563218390804598e-05,
-      "loss": 0.1565,
+      "epoch": 0.907900480139677,
+      "grad_norm": 3.9914190769195557,
+      "learning_rate": 4.5414847161572056e-05,
+      "loss": 0.4909,
       "step": 1040
     },
     {
-      "epoch": 7.228915662650603,
-      "grad_norm": 2.9072484970092773,
-      "learning_rate": 3.5440613026819926e-05,
-      "loss": 0.1813,
+      "epoch": 0.9166302924487123,
+      "grad_norm": 4.449773788452148,
+      "learning_rate": 4.585152838427948e-05,
+      "loss": 0.5089,
       "step": 1050
     },
     {
-      "epoch": 7.29776247848537,
-      "grad_norm": 2.5954692363739014,
-      "learning_rate": 3.5249042145593867e-05,
-      "loss": 0.2118,
+      "epoch": 0.9253601047577477,
+      "grad_norm": 4.087109565734863,
+      "learning_rate": 4.6288209606986906e-05,
+      "loss": 0.5272,
       "step": 1060
     },
     {
-      "epoch": 7.366609294320138,
-      "grad_norm": 2.7799787521362305,
-      "learning_rate": 3.505747126436782e-05,
-      "loss": 0.1744,
+      "epoch": 0.934089917066783,
+      "grad_norm": 4.070830345153809,
+      "learning_rate": 4.672489082969432e-05,
+      "loss": 0.5054,
       "step": 1070
     },
     {
-      "epoch": 7.435456110154905,
-      "grad_norm": 3.0674917697906494,
-      "learning_rate": 3.486590038314176e-05,
-      "loss": 0.1922,
+      "epoch": 0.9428197293758184,
+      "grad_norm": 3.926940679550171,
+      "learning_rate": 4.716157205240175e-05,
+      "loss": 0.5005,
       "step": 1080
     },
     {
-      "epoch": 7.504302925989673,
-      "grad_norm": 3.5969247817993164,
-      "learning_rate": 3.467432950191571e-05,
-      "loss": 0.1794,
+      "epoch": 0.9515495416848537,
+      "grad_norm": 3.5378997325897217,
+      "learning_rate": 4.759825327510917e-05,
+      "loss": 0.5203,
       "step": 1090
     },
     {
-      "epoch": 7.573149741824441,
-      "grad_norm": 2.6146047115325928,
-      "learning_rate": 3.4482758620689657e-05,
-      "loss": 0.1682,
+      "epoch": 0.9602793539938891,
+      "grad_norm": 3.6640408039093018,
+      "learning_rate": 4.8034934497816594e-05,
+      "loss": 0.4391,
       "step": 1100
     },
     {
-      "epoch": 7.641996557659208,
-      "grad_norm": 2.785433053970337,
-      "learning_rate": 3.4291187739463604e-05,
-      "loss": 0.1901,
+      "epoch": 0.9690091663029244,
+      "grad_norm": 3.9486780166625977,
+      "learning_rate": 4.847161572052402e-05,
+      "loss": 0.4608,
       "step": 1110
     },
     {
-      "epoch": 7.710843373493976,
-      "grad_norm": 2.3648314476013184,
-      "learning_rate": 3.409961685823755e-05,
-      "loss": 0.1521,
+      "epoch": 0.9777389786119598,
+      "grad_norm": 3.081714153289795,
+      "learning_rate": 4.890829694323144e-05,
+      "loss": 0.4607,
       "step": 1120
     },
     {
-      "epoch": 7.779690189328743,
-      "grad_norm": 2.3298792839050293,
-      "learning_rate": 3.390804597701149e-05,
-      "loss": 0.1901,
+      "epoch": 0.9864687909209952,
+      "grad_norm": 4.005092620849609,
+      "learning_rate": 4.9344978165938866e-05,
+      "loss": 0.4745,
       "step": 1130
     },
     {
-      "epoch": 7.848537005163511,
-      "grad_norm": 2.2863216400146484,
-      "learning_rate": 3.371647509578545e-05,
-      "loss": 0.1706,
+      "epoch": 0.9951986032300305,
+      "grad_norm": 3.1498849391937256,
+      "learning_rate": 4.978165938864629e-05,
+      "loss": 0.3974,
       "step": 1140
     },
     {
-      "epoch": 7.917383820998279,
-      "grad_norm": 3.0322484970092773,
-      "learning_rate": 3.352490421455939e-05,
-      "loss": 0.1612,
+      "epoch": 0.9995635093845482,
+      "eval_accuracy": 0.8978639823226123,
+      "eval_loss": 0.3598543405532837,
+      "eval_runtime": 62.6632,
+      "eval_samples_per_second": 259.993,
+      "eval_steps_per_second": 8.139,
+      "step": 1145
+    },
+    {
+      "epoch": 1.003928415539066,
+      "grad_norm": 3.5931997299194336,
+      "learning_rate": 4.997573993207181e-05,
+      "loss": 0.4407,
       "step": 1150
     },
     {
-      "epoch": 7.986230636833047,
-      "grad_norm": 2.785752534866333,
-      "learning_rate": 3.3333333333333335e-05,
-      "loss": 0.1657,
+      "epoch": 1.0126582278481013,
+      "grad_norm": 2.662714719772339,
+      "learning_rate": 4.9927219796215426e-05,
+      "loss": 0.4129,
       "step": 1160
     },
     {
-      "epoch": 8.0,
-      "eval_accuracy": 0.8541666666666666,
-      "eval_loss": 0.47179850935935974,
-      "eval_runtime": 7.8842,
-      "eval_samples_per_second": 261.788,
-      "eval_steps_per_second": 8.244,
-      "step": 1162
-    },
-    {
-      "epoch": 8.055077452667813,
-      "grad_norm": 3.0953948497772217,
-      "learning_rate": 3.314176245210728e-05,
-      "loss": 0.1848,
+      "epoch": 1.0213880401571367,
+      "grad_norm": 3.9502806663513184,
+      "learning_rate": 4.9878699660359054e-05,
+      "loss": 0.4622,
       "step": 1170
     },
     {
-      "epoch": 8.123924268502583,
-      "grad_norm": 2.7531402111053467,
-      "learning_rate": 3.295019157088123e-05,
-      "loss": 0.1451,
+      "epoch": 1.030117852466172,
+      "grad_norm": 3.2790305614471436,
+      "learning_rate": 4.9830179524502674e-05,
+      "loss": 0.4446,
       "step": 1180
     },
     {
-      "epoch": 8.19277108433735,
-      "grad_norm": 3.849405288696289,
-      "learning_rate": 3.275862068965517e-05,
-      "loss": 0.1682,
+      "epoch": 1.0388476647752074,
+      "grad_norm": 3.6010549068450928,
+      "learning_rate": 4.978165938864629e-05,
+      "loss": 0.4411,
       "step": 1190
     },
     {
-      "epoch": 8.261617900172118,
-      "grad_norm": 2.8468093872070312,
-      "learning_rate": 3.256704980842912e-05,
-      "loss": 0.1868,
+      "epoch": 1.0475774770842428,
+      "grad_norm": 3.6778202056884766,
+      "learning_rate": 4.9733139252789915e-05,
+      "loss": 0.4662,
       "step": 1200
     },
     {
-      "epoch": 8.330464716006885,
-      "grad_norm": 3.0676023960113525,
-      "learning_rate": 3.2375478927203066e-05,
-      "loss": 0.1745,
+      "epoch": 1.0563072893932781,
+      "grad_norm": 3.7837071418762207,
+      "learning_rate": 4.968461911693353e-05,
+      "loss": 0.4305,
       "step": 1210
     },
     {
-      "epoch": 8.399311531841652,
-      "grad_norm": 2.8244659900665283,
-      "learning_rate": 3.218390804597701e-05,
-      "loss": 0.1783,
+      "epoch": 1.0650371017023135,
+      "grad_norm": 3.306307554244995,
+      "learning_rate": 4.963609898107715e-05,
+      "loss": 0.4111,
       "step": 1220
     },
     {
-      "epoch": 8.46815834767642,
-      "grad_norm": 3.3981621265411377,
-      "learning_rate": 3.1992337164750954e-05,
-      "loss": 0.1509,
+      "epoch": 1.0737669140113488,
+      "grad_norm": 3.2483878135681152,
+      "learning_rate": 4.9587578845220763e-05,
+      "loss": 0.3983,
       "step": 1230
     },
     {
-      "epoch": 8.537005163511187,
-      "grad_norm": 3.2597496509552,
-      "learning_rate": 3.180076628352491e-05,
-      "loss": 0.1463,
+      "epoch": 1.0824967263203842,
+      "grad_norm": 3.236823558807373,
+      "learning_rate": 4.953905870936439e-05,
+      "loss": 0.386,
       "step": 1240
     },
     {
-      "epoch": 8.605851979345955,
-      "grad_norm": 2.98665452003479,
-      "learning_rate": 3.160919540229885e-05,
-      "loss": 0.1854,
+      "epoch": 1.0912265386294195,
+      "grad_norm": 3.8218398094177246,
+      "learning_rate": 4.949053857350801e-05,
+      "loss": 0.4355,
       "step": 1250
     },
     {
-      "epoch": 8.674698795180722,
-      "grad_norm": 3.56426739692688,
-      "learning_rate": 3.1417624521072797e-05,
-      "loss": 0.1801,
+      "epoch": 1.099956350938455,
+      "grad_norm": 2.8009932041168213,
+      "learning_rate": 4.9442018437651625e-05,
+      "loss": 0.4203,
       "step": 1260
     },
     {
-      "epoch": 8.74354561101549,
-      "grad_norm": 3.350994110107422,
-      "learning_rate": 3.1226053639846744e-05,
-      "loss": 0.146,
+      "epoch": 1.1086861632474903,
+      "grad_norm": 3.2365312576293945,
+      "learning_rate": 4.9393498301795246e-05,
+      "loss": 0.3648,
       "step": 1270
     },
     {
-      "epoch": 8.812392426850257,
-      "grad_norm": 2.637911081314087,
-      "learning_rate": 3.103448275862069e-05,
-      "loss": 0.1759,
+      "epoch": 1.1174159755565256,
+      "grad_norm": 3.3708131313323975,
+      "learning_rate": 4.9344978165938866e-05,
+      "loss": 0.4026,
       "step": 1280
     },
     {
-      "epoch": 8.881239242685027,
-      "grad_norm": 3.2409827709198,
-      "learning_rate": 3.084291187739464e-05,
-      "loss": 0.1897,
+      "epoch": 1.126145787865561,
+      "grad_norm": 3.124727487564087,
+      "learning_rate": 4.929645803008249e-05,
+      "loss": 0.3873,
       "step": 1290
     },
     {
-      "epoch": 8.950086058519794,
-      "grad_norm": 2.5662424564361572,
-      "learning_rate": 3.065134099616858e-05,
-      "loss": 0.1427,
+      "epoch": 1.1348756001745963,
+      "grad_norm": 3.1519768238067627,
+      "learning_rate": 4.92479378942261e-05,
+      "loss": 0.3864,
       "step": 1300
     },
     {
-      "epoch": 8.99827882960413,
-      "eval_accuracy": 0.8473837209302325,
-      "eval_loss": 0.4902307689189911,
-      "eval_runtime": 7.7511,
-      "eval_samples_per_second": 266.284,
-      "eval_steps_per_second": 8.386,
-      "step": 1307
-    },
-    {
-      "epoch": 9.018932874354562,
-      "grad_norm": 2.6389100551605225,
-      "learning_rate": 3.045977011494253e-05,
-      "loss": 0.1577,
+      "epoch": 1.1436054124836317,
+      "grad_norm": 3.662675619125366,
+      "learning_rate": 4.919941775836973e-05,
+      "loss": 0.4071,
       "step": 1310
     },
     {
-      "epoch": 9.087779690189329,
-      "grad_norm": 2.561455249786377,
-      "learning_rate": 3.0268199233716475e-05,
-      "loss": 0.1686,
+      "epoch": 1.152335224792667,
+      "grad_norm": 3.7715182304382324,
+      "learning_rate": 4.915089762251335e-05,
+      "loss": 0.37,
       "step": 1320
     },
     {
-      "epoch": 9.156626506024097,
-      "grad_norm": 2.796520233154297,
-      "learning_rate": 3.0076628352490422e-05,
-      "loss": 0.1623,
+      "epoch": 1.1610650371017024,
+      "grad_norm": 3.4320363998413086,
+      "learning_rate": 4.910237748665696e-05,
+      "loss": 0.4143,
       "step": 1330
     },
     {
-      "epoch": 9.225473321858864,
-      "grad_norm": 3.3344945907592773,
-      "learning_rate": 2.988505747126437e-05,
-      "loss": 0.1684,
+      "epoch": 1.1697948494107377,
+      "grad_norm": 2.953200101852417,
+      "learning_rate": 4.905385735080058e-05,
+      "loss": 0.3522,
       "step": 1340
     },
     {
-      "epoch": 9.294320137693632,
-      "grad_norm": 2.494370698928833,
-      "learning_rate": 2.9693486590038317e-05,
-      "loss": 0.1679,
+      "epoch": 1.178524661719773,
+      "grad_norm": 3.239183187484741,
+      "learning_rate": 4.90053372149442e-05,
+      "loss": 0.3704,
       "step": 1350
     },
     {
-      "epoch": 9.363166953528399,
-      "grad_norm": 2.9866812229156494,
-      "learning_rate": 2.950191570881226e-05,
-      "loss": 0.168,
+      "epoch": 1.1872544740288085,
+      "grad_norm": 3.6990933418273926,
+      "learning_rate": 4.8956817079087824e-05,
+      "loss": 0.392,
       "step": 1360
     },
     {
-      "epoch": 9.432013769363166,
-      "grad_norm": 2.8670761585235596,
-      "learning_rate": 2.9310344827586206e-05,
-      "loss": 0.1587,
+      "epoch": 1.1959842863378438,
+      "grad_norm": 3.497069835662842,
+      "learning_rate": 4.890829694323144e-05,
+      "loss": 0.3628,
       "step": 1370
     },
     {
-      "epoch": 9.500860585197934,
-      "grad_norm": 2.6785264015197754,
-      "learning_rate": 2.9118773946360157e-05,
-      "loss": 0.1487,
+      "epoch": 1.2047140986468792,
+      "grad_norm": 4.022006034851074,
+      "learning_rate": 4.8859776807375065e-05,
+      "loss": 0.3842,
       "step": 1380
     },
     {
-      "epoch": 9.569707401032701,
-      "grad_norm": 3.9908454418182373,
-      "learning_rate": 2.89272030651341e-05,
-      "loss": 0.1913,
+      "epoch": 1.2134439109559145,
+      "grad_norm": 3.68331241607666,
+      "learning_rate": 4.8811256671518685e-05,
+      "loss": 0.3386,
       "step": 1390
     },
     {
-      "epoch": 9.638554216867469,
-      "grad_norm": 3.4217581748962402,
-      "learning_rate": 2.8735632183908045e-05,
-      "loss": 0.1568,
+      "epoch": 1.2221737232649499,
+      "grad_norm": 2.9863877296447754,
+      "learning_rate": 4.87627365356623e-05,
+      "loss": 0.3398,
       "step": 1400
     },
     {
-      "epoch": 9.707401032702238,
-      "grad_norm": 3.4280409812927246,
-      "learning_rate": 2.8544061302681996e-05,
-      "loss": 0.1335,
+      "epoch": 1.2309035355739852,
+      "grad_norm": 3.9146687984466553,
+      "learning_rate": 4.871421639980592e-05,
+      "loss": 0.3543,
       "step": 1410
     },
     {
-      "epoch": 9.776247848537006,
-      "grad_norm": 3.57127046585083,
-      "learning_rate": 2.835249042145594e-05,
-      "loss": 0.1508,
+      "epoch": 1.2396333478830206,
+      "grad_norm": 3.9005401134490967,
+      "learning_rate": 4.866569626394954e-05,
+      "loss": 0.4041,
       "step": 1420
     },
     {
-      "epoch": 9.845094664371773,
-      "grad_norm": 2.238633394241333,
-      "learning_rate": 2.8160919540229884e-05,
-      "loss": 0.1878,
+      "epoch": 1.248363160192056,
+      "grad_norm": 2.9291443824768066,
+      "learning_rate": 4.861717612809316e-05,
+      "loss": 0.3363,
       "step": 1430
     },
     {
-      "epoch": 9.91394148020654,
-      "grad_norm": 2.7804625034332275,
-      "learning_rate": 2.796934865900383e-05,
-      "loss": 0.1497,
+      "epoch": 1.2570929725010913,
+      "grad_norm": 3.2885146141052246,
+      "learning_rate": 4.8568655992236775e-05,
+      "loss": 0.375,
       "step": 1440
     },
     {
-      "epoch": 9.982788296041308,
-      "grad_norm": 3.1482419967651367,
-      "learning_rate": 2.777777777777778e-05,
-      "loss": 0.1598,
+      "epoch": 1.2658227848101267,
+      "grad_norm": 3.188255548477173,
+      "learning_rate": 4.85201358563804e-05,
+      "loss": 0.3467,
       "step": 1450
     },
     {
-      "epoch": 9.996557659208262,
-      "eval_accuracy": 0.9273255813953488,
-      "eval_loss": 0.22291144728660583,
-      "eval_runtime": 7.6884,
-      "eval_samples_per_second": 268.456,
-      "eval_steps_per_second": 8.454,
-      "step": 1452
-    },
-    {
-      "epoch": 10.051635111876076,
-      "grad_norm": 2.5692484378814697,
-      "learning_rate": 2.7586206896551727e-05,
-      "loss": 0.1397,
+      "epoch": 1.274552597119162,
+      "grad_norm": 3.08927059173584,
+      "learning_rate": 4.847161572052402e-05,
+      "loss": 0.3863,
       "step": 1460
     },
     {
-      "epoch": 10.120481927710843,
-      "grad_norm": 1.686840295791626,
-      "learning_rate": 2.739463601532567e-05,
-      "loss": 0.1246,
+      "epoch": 1.2832824094281974,
+      "grad_norm": 2.3255248069763184,
+      "learning_rate": 4.8423095584667636e-05,
+      "loss": 0.3498,
       "step": 1470
     },
     {
-      "epoch": 10.18932874354561,
-      "grad_norm": 2.7757647037506104,
-      "learning_rate": 2.720306513409962e-05,
-      "loss": 0.14,
+      "epoch": 1.2920122217372327,
+      "grad_norm": 3.268420696258545,
+      "learning_rate": 4.837457544881126e-05,
+      "loss": 0.3092,
       "step": 1480
     },
     {
-      "epoch": 10.258175559380378,
-      "grad_norm": 2.688798189163208,
-      "learning_rate": 2.7011494252873566e-05,
-      "loss": 0.1386,
+      "epoch": 1.300742034046268,
+      "grad_norm": 3.7126426696777344,
+      "learning_rate": 4.832605531295488e-05,
+      "loss": 0.3402,
       "step": 1490
     },
     {
-      "epoch": 10.327022375215146,
-      "grad_norm": 3.0354042053222656,
-      "learning_rate": 2.681992337164751e-05,
-      "loss": 0.1585,
+      "epoch": 1.3094718463553034,
+      "grad_norm": 2.976686954498291,
+      "learning_rate": 4.82775351770985e-05,
+      "loss": 0.3038,
       "step": 1500
     },
     {
-      "epoch": 10.395869191049915,
-      "grad_norm": 3.36676287651062,
-      "learning_rate": 2.662835249042146e-05,
-      "loss": 0.1368,
+      "epoch": 1.3182016586643388,
+      "grad_norm": 3.393733501434326,
+      "learning_rate": 4.822901504124212e-05,
+      "loss": 0.3353,
       "step": 1510
     },
     {
-      "epoch": 10.464716006884682,
-      "grad_norm": 2.3574771881103516,
-      "learning_rate": 2.6436781609195405e-05,
-      "loss": 0.1559,
+      "epoch": 1.3269314709733742,
+      "grad_norm": 3.2993807792663574,
+      "learning_rate": 4.818049490538574e-05,
+      "loss": 0.3427,
       "step": 1520
     },
     {
-      "epoch": 10.53356282271945,
-      "grad_norm": 2.6529464721679688,
-      "learning_rate": 2.624521072796935e-05,
-      "loss": 0.154,
+      "epoch": 1.3356612832824095,
+      "grad_norm": 3.6997506618499756,
+      "learning_rate": 4.813197476952936e-05,
+      "loss": 0.3535,
       "step": 1530
     },
     {
-      "epoch": 10.602409638554217,
-      "grad_norm": 3.449251174926758,
-      "learning_rate": 2.6053639846743293e-05,
-      "loss": 0.1401,
+      "epoch": 1.3443910955914449,
+      "grad_norm": 3.4463653564453125,
+      "learning_rate": 4.808345463367297e-05,
+      "loss": 0.3431,
       "step": 1540
     },
     {
-      "epoch": 10.671256454388985,
-      "grad_norm": 2.449934482574463,
-      "learning_rate": 2.5862068965517244e-05,
-      "loss": 0.1408,
+      "epoch": 1.3531209079004802,
+      "grad_norm": 3.227196455001831,
+      "learning_rate": 4.8034934497816594e-05,
+      "loss": 0.3114,
       "step": 1550
     },
     {
-      "epoch": 10.740103270223752,
-      "grad_norm": 2.8734378814697266,
-      "learning_rate": 2.5670498084291188e-05,
-      "loss": 0.1445,
+      "epoch": 1.3618507202095156,
+      "grad_norm": 3.5765860080718994,
+      "learning_rate": 4.7986414361960214e-05,
+      "loss": 0.3203,
       "step": 1560
     },
     {
-      "epoch": 10.80895008605852,
-      "grad_norm": 2.4971940517425537,
-      "learning_rate": 2.5478927203065132e-05,
-      "loss": 0.1643,
+      "epoch": 1.370580532518551,
+      "grad_norm": 4.020304203033447,
+      "learning_rate": 4.7937894226103835e-05,
+      "loss": 0.3496,
       "step": 1570
     },
     {
-      "epoch": 10.877796901893287,
-      "grad_norm": 3.7385921478271484,
-      "learning_rate": 2.5287356321839083e-05,
-      "loss": 0.1521,
+      "epoch": 1.3793103448275863,
+      "grad_norm": 3.5590262413024902,
+      "learning_rate": 4.7889374090247456e-05,
+      "loss": 0.3294,
       "step": 1580
     },
     {
-      "epoch": 10.946643717728055,
-      "grad_norm": 2.7152345180511475,
-      "learning_rate": 2.5095785440613027e-05,
-      "loss": 0.1504,
+      "epoch": 1.3880401571366217,
+      "grad_norm": 3.3192408084869385,
+      "learning_rate": 4.7840853954391076e-05,
+      "loss": 0.3334,
       "step": 1590
     },
     {
-      "epoch": 10.994836488812393,
-      "eval_accuracy": 0.8972868217054264,
-      "eval_loss": 0.3021294176578522,
-      "eval_runtime": 7.4857,
-      "eval_samples_per_second": 275.726,
-      "eval_steps_per_second": 8.683,
-      "step": 1597
-    },
-    {
-      "epoch": 11.015490533562822,
-      "grad_norm": 3.032806158065796,
-      "learning_rate": 2.4904214559386975e-05,
-      "loss": 0.1366,
+      "epoch": 1.396769969445657,
+      "grad_norm": 3.537508726119995,
+      "learning_rate": 4.77923338185347e-05,
+      "loss": 0.3349,
       "step": 1600
     },
     {
-      "epoch": 11.08433734939759,
-      "grad_norm": 2.85953950881958,
-      "learning_rate": 2.4712643678160922e-05,
-      "loss": 0.1433,
+      "epoch": 1.4054997817546924,
+      "grad_norm": 4.00840425491333,
+      "learning_rate": 4.774381368267831e-05,
+      "loss": 0.334,
       "step": 1610
     },
     {
-      "epoch": 11.153184165232357,
-      "grad_norm": 2.1525700092315674,
-      "learning_rate": 2.4521072796934867e-05,
-      "loss": 0.1455,
+      "epoch": 1.4142295940637277,
+      "grad_norm": 3.3885114192962646,
+      "learning_rate": 4.769529354682193e-05,
+      "loss": 0.2888,
       "step": 1620
     },
     {
-      "epoch": 11.222030981067126,
-      "grad_norm": 2.248845100402832,
-      "learning_rate": 2.4329501915708814e-05,
-      "loss": 0.1269,
+      "epoch": 1.422959406372763,
+      "grad_norm": 3.376528739929199,
+      "learning_rate": 4.764677341096555e-05,
+      "loss": 0.3122,
       "step": 1630
     },
     {
-      "epoch": 11.290877796901894,
-      "grad_norm": 2.333122968673706,
-      "learning_rate": 2.413793103448276e-05,
-      "loss": 0.156,
+      "epoch": 1.4316892186817984,
+      "grad_norm": 3.8087868690490723,
+      "learning_rate": 4.759825327510917e-05,
+      "loss": 0.3433,
       "step": 1640
     },
     {
-      "epoch": 11.359724612736661,
-      "grad_norm": 2.1817004680633545,
-      "learning_rate": 2.3946360153256706e-05,
-      "loss": 0.1423,
+      "epoch": 1.4404190309908338,
+      "grad_norm": 2.8861641883850098,
+      "learning_rate": 4.754973313925279e-05,
+      "loss": 0.2913,
       "step": 1650
     },
     {
-      "epoch": 11.428571428571429,
-      "grad_norm": 2.1537561416625977,
-      "learning_rate": 2.3754789272030653e-05,
-      "loss": 0.1327,
+      "epoch": 1.4491488432998691,
+      "grad_norm": 3.975708246231079,
+      "learning_rate": 4.750121300339641e-05,
+      "loss": 0.3294,
       "step": 1660
     },
     {
-      "epoch": 11.497418244406196,
-      "grad_norm": 1.6727231740951538,
-      "learning_rate": 2.3563218390804597e-05,
-      "loss": 0.1272,
+      "epoch": 1.4578786556089045,
+      "grad_norm": 4.433221817016602,
+      "learning_rate": 4.7452692867540034e-05,
+      "loss": 0.2937,
       "step": 1670
     },
     {
-      "epoch": 11.566265060240964,
-      "grad_norm": 3.15240216255188,
-      "learning_rate": 2.3371647509578545e-05,
-      "loss": 0.1474,
+      "epoch": 1.4666084679179399,
+      "grad_norm": 3.9927902221679688,
+      "learning_rate": 4.740417273168365e-05,
+      "loss": 0.2853,
       "step": 1680
     },
     {
-      "epoch": 11.635111876075731,
-      "grad_norm": 2.031428813934326,
-      "learning_rate": 2.3180076628352492e-05,
-      "loss": 0.1341,
+      "epoch": 1.4753382802269752,
+      "grad_norm": 3.353848695755005,
+      "learning_rate": 4.735565259582727e-05,
+      "loss": 0.3477,
       "step": 1690
     },
     {
-      "epoch": 11.703958691910499,
-      "grad_norm": 2.5997140407562256,
-      "learning_rate": 2.2988505747126437e-05,
-      "loss": 0.153,
+      "epoch": 1.4840680925360106,
+      "grad_norm": 3.72670841217041,
+      "learning_rate": 4.730713245997089e-05,
+      "loss": 0.3342,
       "step": 1700
     },
     {
-      "epoch": 11.772805507745266,
-      "grad_norm": 2.405869483947754,
-      "learning_rate": 2.2796934865900384e-05,
-      "loss": 0.1434,
+      "epoch": 1.492797904845046,
+      "grad_norm": 3.5278804302215576,
+      "learning_rate": 4.725861232411451e-05,
+      "loss": 0.2832,
       "step": 1710
     },
     {
-      "epoch": 11.841652323580034,
-      "grad_norm": 2.8463237285614014,
-      "learning_rate": 2.2605363984674328e-05,
-      "loss": 0.1545,
+      "epoch": 1.5015277171540813,
+      "grad_norm": 3.5316507816314697,
+      "learning_rate": 4.721009218825813e-05,
+      "loss": 0.2962,
       "step": 1720
     },
     {
-      "epoch": 11.910499139414803,
-      "grad_norm": 2.8656628131866455,
-      "learning_rate": 2.2413793103448276e-05,
-      "loss": 0.1412,
+      "epoch": 1.5102575294631166,
+      "grad_norm": 3.522723913192749,
+      "learning_rate": 4.716157205240175e-05,
+      "loss": 0.3165,
       "step": 1730
     },
     {
-      "epoch": 11.97934595524957,
-      "grad_norm": 2.9610302448272705,
-      "learning_rate": 2.2222222222222223e-05,
-      "loss": 0.1456,
+      "epoch": 1.518987341772152,
+      "grad_norm": 3.0365447998046875,
+      "learning_rate": 4.711305191654537e-05,
+      "loss": 0.2921,
       "step": 1740
     },
     {
-      "epoch": 12.0,
-      "eval_accuracy": 0.9224806201550387,
-      "eval_loss": 0.24217405915260315,
-      "eval_runtime": 7.2188,
-      "eval_samples_per_second": 285.921,
-      "eval_steps_per_second": 9.004,
-      "step": 1743
-    },
-    {
-      "epoch": 12.048192771084338,
-      "grad_norm": 2.278578042984009,
-      "learning_rate": 2.203065134099617e-05,
-      "loss": 0.1079,
+      "epoch": 1.5277171540811874,
+      "grad_norm": 3.355435848236084,
+      "learning_rate": 4.7064531780688984e-05,
+      "loss": 0.2887,
       "step": 1750
     },
     {
-      "epoch": 12.117039586919105,
-      "grad_norm": 2.1894547939300537,
-      "learning_rate": 2.183908045977012e-05,
-      "loss": 0.1298,
+      "epoch": 1.5364469663902227,
+      "grad_norm": 4.480086326599121,
+      "learning_rate": 4.7016011644832605e-05,
+      "loss": 0.28,
       "step": 1760
     },
     {
-      "epoch": 12.185886402753873,
-      "grad_norm": 2.7255043983459473,
-      "learning_rate": 2.1647509578544062e-05,
-      "loss": 0.1502,
+      "epoch": 1.545176778699258,
+      "grad_norm": 3.315585136413574,
+      "learning_rate": 4.6967491508976226e-05,
+      "loss": 0.3122,
       "step": 1770
     },
     {
-      "epoch": 12.25473321858864,
-      "grad_norm": 1.9726896286010742,
-      "learning_rate": 2.145593869731801e-05,
-      "loss": 0.1589,
+      "epoch": 1.5539065910082934,
+      "grad_norm": 3.043123483657837,
+      "learning_rate": 4.6918971373119846e-05,
+      "loss": 0.2679,
       "step": 1780
     },
     {
-      "epoch": 12.323580034423408,
-      "grad_norm": 2.095195770263672,
-      "learning_rate": 2.1264367816091954e-05,
-      "loss": 0.1345,
+      "epoch": 1.5626364033173288,
+      "grad_norm": 2.6386468410491943,
+      "learning_rate": 4.687045123726347e-05,
+      "loss": 0.2796,
       "step": 1790
     },
     {
-      "epoch": 12.392426850258175,
-      "grad_norm": 2.9772017002105713,
-      "learning_rate": 2.10727969348659e-05,
-      "loss": 0.1537,
+      "epoch": 1.5713662156263641,
+      "grad_norm": 3.1339259147644043,
+      "learning_rate": 4.682193110140709e-05,
+      "loss": 0.2942,
       "step": 1800
     },
     {
-      "epoch": 12.461273666092943,
-      "grad_norm": 1.4328383207321167,
-      "learning_rate": 2.088122605363985e-05,
-      "loss": 0.1274,
+      "epoch": 1.5800960279353995,
+      "grad_norm": 3.921851873397827,
+      "learning_rate": 4.677341096555071e-05,
+      "loss": 0.3087,
       "step": 1810
     },
     {
-      "epoch": 12.53012048192771,
-      "grad_norm": 2.386387586593628,
-      "learning_rate": 2.0689655172413793e-05,
-      "loss": 0.1333,
+      "epoch": 1.5888258402444349,
+      "grad_norm": 3.262922525405884,
+      "learning_rate": 4.672489082969432e-05,
+      "loss": 0.2879,
       "step": 1820
     },
     {
-      "epoch": 12.598967297762478,
-      "grad_norm": 1.9341486692428589,
-      "learning_rate": 2.049808429118774e-05,
-      "loss": 0.1397,
+      "epoch": 1.5975556525534702,
+      "grad_norm": 3.206650733947754,
+      "learning_rate": 4.667637069383794e-05,
+      "loss": 0.278,
       "step": 1830
     },
     {
-      "epoch": 12.667814113597245,
-      "grad_norm": 2.633357048034668,
-      "learning_rate": 2.0306513409961685e-05,
-      "loss": 0.1151,
+      "epoch": 1.6062854648625056,
+      "grad_norm": 3.8127830028533936,
+      "learning_rate": 4.662785055798157e-05,
+      "loss": 0.2656,
       "step": 1840
     },
     {
-      "epoch": 12.736660929432015,
-      "grad_norm": 3.059253215789795,
-      "learning_rate": 2.0114942528735632e-05,
-      "loss": 0.119,
+      "epoch": 1.615015277171541,
+      "grad_norm": 3.552639961242676,
+      "learning_rate": 4.657933042212518e-05,
+      "loss": 0.2858,
       "step": 1850
     },
     {
-      "epoch": 12.805507745266782,
-      "grad_norm": 1.629744291305542,
-      "learning_rate": 1.992337164750958e-05,
-      "loss": 0.1194,
+      "epoch": 1.6237450894805763,
+      "grad_norm": 3.771047353744507,
+      "learning_rate": 4.6530810286268804e-05,
+      "loss": 0.3176,
       "step": 1860
     },
     {
-      "epoch": 12.87435456110155,
-      "grad_norm": 3.2468857765197754,
-      "learning_rate": 1.9731800766283527e-05,
-      "loss": 0.1567,
+      "epoch": 1.6324749017896116,
+      "grad_norm": 3.8466198444366455,
+      "learning_rate": 4.6482290150412424e-05,
+      "loss": 0.2594,
       "step": 1870
     },
     {
-      "epoch": 12.943201376936317,
-      "grad_norm": 2.7284107208251953,
-      "learning_rate": 1.9540229885057475e-05,
-      "loss": 0.119,
+      "epoch": 1.641204714098647,
+      "grad_norm": 3.0491561889648438,
+      "learning_rate": 4.6433770014556045e-05,
+      "loss": 0.3074,
       "step": 1880
     },
     {
-      "epoch": 12.99827882960413,
-      "eval_accuracy": 0.9021317829457365,
-      "eval_loss": 0.2836342751979828,
-      "eval_runtime": 7.2838,
-      "eval_samples_per_second": 283.368,
-      "eval_steps_per_second": 8.924,
-      "step": 1888
-    },
-    {
-      "epoch": 13.012048192771084,
-      "grad_norm": 3.377997875213623,
-      "learning_rate": 1.934865900383142e-05,
-      "loss": 0.119,
+      "epoch": 1.6499345264076823,
+      "grad_norm": 3.4488463401794434,
+      "learning_rate": 4.638524987869966e-05,
+      "loss": 0.2625,
       "step": 1890
     },
     {
-      "epoch": 13.080895008605852,
-      "grad_norm": 2.6370134353637695,
-      "learning_rate": 1.9157088122605367e-05,
-      "loss": 0.1394,
+      "epoch": 1.6586643387167177,
+      "grad_norm": 3.1591031551361084,
+      "learning_rate": 4.633672974284328e-05,
+      "loss": 0.2824,
       "step": 1900
     },
     {
-      "epoch": 13.14974182444062,
-      "grad_norm": 2.5573413372039795,
-      "learning_rate": 1.896551724137931e-05,
-      "loss": 0.1475,
+      "epoch": 1.667394151025753,
+      "grad_norm": 3.56533145904541,
+      "learning_rate": 4.6288209606986906e-05,
+      "loss": 0.2855,
       "step": 1910
     },
     {
-      "epoch": 13.218588640275387,
-      "grad_norm": 2.2910749912261963,
-      "learning_rate": 1.8773946360153258e-05,
-      "loss": 0.1332,
+      "epoch": 1.6761239633347884,
+      "grad_norm": 3.567021131515503,
+      "learning_rate": 4.623968947113052e-05,
+      "loss": 0.2823,
       "step": 1920
     },
     {
-      "epoch": 13.287435456110154,
-      "grad_norm": 2.3592233657836914,
-      "learning_rate": 1.8582375478927206e-05,
-      "loss": 0.1373,
+      "epoch": 1.6848537756438238,
+      "grad_norm": 4.153507709503174,
+      "learning_rate": 4.619116933527414e-05,
+      "loss": 0.2907,
       "step": 1930
     },
     {
-      "epoch": 13.356282271944922,
-      "grad_norm": 2.7366604804992676,
-      "learning_rate": 1.839080459770115e-05,
-      "loss": 0.129,
+      "epoch": 1.6935835879528591,
+      "grad_norm": 4.215022087097168,
+      "learning_rate": 4.614264919941776e-05,
+      "loss": 0.2699,
       "step": 1940
     },
     {
-      "epoch": 13.42512908777969,
-      "grad_norm": 2.452705144882202,
-      "learning_rate": 1.8199233716475097e-05,
-      "loss": 0.1183,
+      "epoch": 1.7023134002618945,
+      "grad_norm": 3.388324499130249,
+      "learning_rate": 4.609412906356138e-05,
+      "loss": 0.2915,
       "step": 1950
     },
     {
-      "epoch": 13.493975903614459,
-      "grad_norm": 3.3416621685028076,
-      "learning_rate": 1.800766283524904e-05,
-      "loss": 0.154,
+      "epoch": 1.7110432125709298,
+      "grad_norm": 3.654491901397705,
+      "learning_rate": 4.6045608927704996e-05,
+      "loss": 0.2804,
       "step": 1960
     },
     {
-      "epoch": 13.562822719449226,
-      "grad_norm": 3.04327654838562,
-      "learning_rate": 1.781609195402299e-05,
-      "loss": 0.1264,
+      "epoch": 1.7197730248799652,
+      "grad_norm": 3.2709882259368896,
+      "learning_rate": 4.5997088791848616e-05,
+      "loss": 0.271,
       "step": 1970
     },
     {
-      "epoch": 13.631669535283994,
-      "grad_norm": 2.6300888061523438,
-      "learning_rate": 1.7624521072796933e-05,
-      "loss": 0.1646,
+      "epoch": 1.7285028371890006,
+      "grad_norm": 3.5081441402435303,
+      "learning_rate": 4.5948568655992244e-05,
+      "loss": 0.2807,
       "step": 1980
     },
     {
-      "epoch": 13.700516351118761,
-      "grad_norm": 3.809067487716675,
-      "learning_rate": 1.743295019157088e-05,
-      "loss": 0.1363,
+      "epoch": 1.737232649498036,
+      "grad_norm": 2.9276814460754395,
+      "learning_rate": 4.590004852013586e-05,
+      "loss": 0.2756,
       "step": 1990
     },
     {
-      "epoch": 13.769363166953529,
-      "grad_norm": 3.2865700721740723,
-      "learning_rate": 1.7241379310344828e-05,
-      "loss": 0.1145,
+      "epoch": 1.7459624618070713,
+      "grad_norm": 3.9031786918640137,
+      "learning_rate": 4.585152838427948e-05,
+      "loss": 0.2964,
       "step": 2000
     },
     {
-      "epoch": 13.838209982788296,
-      "grad_norm": 2.2611865997314453,
-      "learning_rate": 1.7049808429118776e-05,
-      "loss": 0.116,
+      "epoch": 1.7546922741161066,
+      "grad_norm": 3.184328079223633,
+      "learning_rate": 4.58030082484231e-05,
+      "loss": 0.2366,
       "step": 2010
     },
     {
-      "epoch": 13.907056798623064,
-      "grad_norm": 2.3190841674804688,
-      "learning_rate": 1.6858237547892723e-05,
-      "loss": 0.1105,
+      "epoch": 1.763422086425142,
+      "grad_norm": 3.6933434009552,
+      "learning_rate": 4.575448811256672e-05,
+      "loss": 0.2574,
       "step": 2020
     },
     {
-      "epoch": 13.975903614457831,
-      "grad_norm": 3.105708360671997,
-      "learning_rate": 1.6666666666666667e-05,
-      "loss": 0.114,
+      "epoch": 1.7721518987341773,
+      "grad_norm": 3.177960157394409,
+      "learning_rate": 4.570596797671033e-05,
+      "loss": 0.2602,
       "step": 2030
     },
     {
-      "epoch": 13.996557659208262,
-      "eval_accuracy": 0.9292635658914729,
-      "eval_loss": 0.20378853380680084,
-      "eval_runtime": 7.3214,
-      "eval_samples_per_second": 281.914,
-      "eval_steps_per_second": 8.878,
-      "step": 2033
-    },
-    {
-      "epoch": 14.044750430292599,
-      "grad_norm": 2.9859328269958496,
-      "learning_rate": 1.6475095785440615e-05,
-      "loss": 0.1319,
+      "epoch": 1.7808817110432127,
+      "grad_norm": 4.310092926025391,
+      "learning_rate": 4.565744784085395e-05,
+      "loss": 0.2821,
       "step": 2040
     },
     {
-      "epoch": 14.113597246127366,
-      "grad_norm": 4.717568874359131,
-      "learning_rate": 1.628352490421456e-05,
-      "loss": 0.1426,
+      "epoch": 1.789611523352248,
+      "grad_norm": 2.7888848781585693,
+      "learning_rate": 4.560892770499758e-05,
+      "loss": 0.2409,
       "step": 2050
     },
     {
-      "epoch": 14.182444061962133,
-      "grad_norm": 3.3271358013153076,
-      "learning_rate": 1.6091954022988507e-05,
-      "loss": 0.1458,
+      "epoch": 1.7983413356612834,
+      "grad_norm": 3.6411709785461426,
+      "learning_rate": 4.5560407569141194e-05,
+      "loss": 0.2675,
       "step": 2060
     },
     {
-      "epoch": 14.251290877796903,
-      "grad_norm": 3.017382860183716,
-      "learning_rate": 1.5900383141762454e-05,
-      "loss": 0.1576,
+      "epoch": 1.8070711479703188,
+      "grad_norm": 2.790893077850342,
+      "learning_rate": 4.5511887433284815e-05,
+      "loss": 0.3142,
       "step": 2070
     },
     {
-      "epoch": 14.32013769363167,
-      "grad_norm": 3.2462213039398193,
-      "learning_rate": 1.5708812260536398e-05,
-      "loss": 0.1438,
+      "epoch": 1.8158009602793541,
+      "grad_norm": 4.2732768058776855,
+      "learning_rate": 4.5463367297428435e-05,
+      "loss": 0.2925,
       "step": 2080
     },
     {
-      "epoch": 14.388984509466438,
-      "grad_norm": 3.2721612453460693,
-      "learning_rate": 1.5517241379310346e-05,
-      "loss": 0.1632,
+      "epoch": 1.8245307725883895,
+      "grad_norm": 3.220660448074341,
+      "learning_rate": 4.5414847161572056e-05,
+      "loss": 0.2705,
       "step": 2090
     },
     {
-      "epoch": 14.457831325301205,
-      "grad_norm": 2.183432102203369,
-      "learning_rate": 1.532567049808429e-05,
-      "loss": 0.1071,
+      "epoch": 1.8332605848974248,
+      "grad_norm": 4.008994102478027,
+      "learning_rate": 4.5366327025715677e-05,
+      "loss": 0.256,
       "step": 2100
     },
     {
-      "epoch": 14.526678141135973,
-      "grad_norm": 2.744126319885254,
-      "learning_rate": 1.5134099616858237e-05,
-      "loss": 0.1234,
+      "epoch": 1.8419903972064602,
+      "grad_norm": 3.519592761993408,
+      "learning_rate": 4.531780688985929e-05,
+      "loss": 0.2503,
       "step": 2110
     },
     {
-      "epoch": 14.59552495697074,
-      "grad_norm": 1.959697961807251,
-      "learning_rate": 1.4942528735632185e-05,
-      "loss": 0.1224,
+      "epoch": 1.8507202095154955,
+      "grad_norm": 3.621957540512085,
+      "learning_rate": 4.526928675400292e-05,
+      "loss": 0.2641,
       "step": 2120
     },
     {
-      "epoch": 14.664371772805508,
-      "grad_norm": 3.4430899620056152,
-      "learning_rate": 1.475095785440613e-05,
-      "loss": 0.1174,
+      "epoch": 1.859450021824531,
+      "grad_norm": 3.7663767337799072,
+      "learning_rate": 4.522076661814653e-05,
+      "loss": 0.2256,
       "step": 2130
     },
     {
-      "epoch": 14.733218588640275,
-      "grad_norm": 4.045992374420166,
-      "learning_rate": 1.4559386973180078e-05,
-      "loss": 0.1315,
+      "epoch": 1.8681798341335663,
+      "grad_norm": 2.5823781490325928,
+      "learning_rate": 4.517224648229015e-05,
+      "loss": 0.267,
       "step": 2140
     },
     {
-      "epoch": 14.802065404475043,
-      "grad_norm": 2.5135693550109863,
-      "learning_rate": 1.4367816091954022e-05,
-      "loss": 0.111,
+      "epoch": 1.8769096464426016,
+      "grad_norm": 2.787856340408325,
+      "learning_rate": 4.512372634643377e-05,
+      "loss": 0.2841,
       "step": 2150
     },
     {
-      "epoch": 14.87091222030981,
-      "grad_norm": 3.5096664428710938,
-      "learning_rate": 1.417624521072797e-05,
-      "loss": 0.1429,
+      "epoch": 1.885639458751637,
+      "grad_norm": 3.87716007232666,
+      "learning_rate": 4.507520621057739e-05,
+      "loss": 0.2001,
       "step": 2160
     },
     {
-      "epoch": 14.939759036144578,
-      "grad_norm": 2.6633498668670654,
-      "learning_rate": 1.3984674329501916e-05,
-      "loss": 0.1378,
+      "epoch": 1.8943692710606723,
+      "grad_norm": 3.641904830932617,
+      "learning_rate": 4.5026686074721014e-05,
+      "loss": 0.2586,
       "step": 2170
     },
     {
-      "epoch": 14.994836488812393,
-      "eval_accuracy": 0.9239341085271318,
-      "eval_loss": 0.2173481285572052,
-      "eval_runtime": 7.5899,
-      "eval_samples_per_second": 271.94,
-      "eval_steps_per_second": 8.564,
-      "step": 2178
-    },
-    {
-      "epoch": 15.008605851979347,
-      "grad_norm": 2.4078171253204346,
-      "learning_rate": 1.3793103448275863e-05,
-      "loss": 0.1133,
+      "epoch": 1.9030990833697077,
+      "grad_norm": 3.6821677684783936,
+      "learning_rate": 4.497816593886463e-05,
+      "loss": 0.2698,
       "step": 2180
     },
     {
-      "epoch": 15.077452667814114,
-      "grad_norm": 3.3953912258148193,
-      "learning_rate": 1.360153256704981e-05,
-      "loss": 0.1486,
+      "epoch": 1.911828895678743,
+      "grad_norm": 2.7858848571777344,
+      "learning_rate": 4.4929645803008255e-05,
+      "loss": 0.2734,
       "step": 2190
     },
     {
-      "epoch": 15.146299483648882,
-      "grad_norm": 2.9477121829986572,
-      "learning_rate": 1.3409961685823755e-05,
-      "loss": 0.1097,
+      "epoch": 1.9205587079877784,
+      "grad_norm": 3.4211864471435547,
+      "learning_rate": 4.488112566715187e-05,
+      "loss": 0.2387,
       "step": 2200
     },
     {
-      "epoch": 15.21514629948365,
-      "grad_norm": 2.8032114505767822,
-      "learning_rate": 1.3218390804597702e-05,
-      "loss": 0.136,
+      "epoch": 1.9292885202968137,
+      "grad_norm": 2.879937171936035,
+      "learning_rate": 4.483260553129549e-05,
+      "loss": 0.2375,
       "step": 2210
     },
     {
-      "epoch": 15.283993115318417,
-      "grad_norm": 3.5413339138031006,
-      "learning_rate": 1.3026819923371647e-05,
-      "loss": 0.1434,
+      "epoch": 1.938018332605849,
+      "grad_norm": 3.930103063583374,
+      "learning_rate": 4.478408539543911e-05,
+      "loss": 0.2767,
       "step": 2220
     },
     {
-      "epoch": 15.352839931153184,
-      "grad_norm": 2.4817466735839844,
-      "learning_rate": 1.2835249042145594e-05,
-      "loss": 0.1178,
+      "epoch": 1.9467481449148845,
+      "grad_norm": 3.938791275024414,
+      "learning_rate": 4.473556525958273e-05,
+      "loss": 0.2536,
       "step": 2230
     },
     {
-      "epoch": 15.421686746987952,
-      "grad_norm": 2.3475193977355957,
-      "learning_rate": 1.2643678160919542e-05,
-      "loss": 0.0938,
+      "epoch": 1.9554779572239198,
+      "grad_norm": 2.574296236038208,
+      "learning_rate": 4.468704512372635e-05,
+      "loss": 0.2188,
       "step": 2240
     },
     {
-      "epoch": 15.49053356282272,
-      "grad_norm": 3.2757270336151123,
-      "learning_rate": 1.2452107279693487e-05,
-      "loss": 0.159,
+      "epoch": 1.9642077695329552,
+      "grad_norm": 4.026519298553467,
+      "learning_rate": 4.4638524987869964e-05,
+      "loss": 0.2557,
       "step": 2250
     },
     {
-      "epoch": 15.559380378657487,
-      "grad_norm": 2.7539010047912598,
-      "learning_rate": 1.2260536398467433e-05,
-      "loss": 0.1351,
+      "epoch": 1.9729375818419905,
+      "grad_norm": 3.5013089179992676,
+      "learning_rate": 4.459000485201359e-05,
+      "loss": 0.2438,
       "step": 2260
     },
     {
-      "epoch": 15.628227194492254,
-      "grad_norm": 2.8831663131713867,
-      "learning_rate": 1.206896551724138e-05,
-      "loss": 0.1341,
+      "epoch": 1.9816673941510259,
+      "grad_norm": 2.124563694000244,
+      "learning_rate": 4.4541484716157205e-05,
+      "loss": 0.2489,
       "step": 2270
     },
     {
-      "epoch": 15.697074010327022,
-      "grad_norm": 3.276233434677124,
-      "learning_rate": 1.1877394636015327e-05,
-      "loss": 0.1348,
+      "epoch": 1.9903972064600612,
+      "grad_norm": 2.5535762310028076,
+      "learning_rate": 4.4492964580300826e-05,
+      "loss": 0.2468,
       "step": 2280
     },
     {
-      "epoch": 15.76592082616179,
-      "grad_norm": 2.2141618728637695,
-      "learning_rate": 1.1685823754789272e-05,
-      "loss": 0.1166,
+      "epoch": 1.9991270187690966,
+      "grad_norm": 3.7411351203918457,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 0.2155,
       "step": 2290
     },
     {
-      "epoch": 15.834767641996558,
-      "grad_norm": 2.0663790702819824,
-      "learning_rate": 1.1494252873563218e-05,
-      "loss": 0.0998,
+      "epoch": 2.0,
+      "eval_accuracy": 0.9602872575497177,
+      "eval_loss": 0.15250827372074127,
+      "eval_runtime": 62.7478,
+      "eval_samples_per_second": 259.643,
+      "eval_steps_per_second": 8.128,
+      "step": 2291
+    },
+    {
+      "epoch": 2.007856831078132,
+      "grad_norm": 3.3374812602996826,
+      "learning_rate": 4.439592430858807e-05,
+      "loss": 0.2331,
       "step": 2300
     },
     {
-      "epoch": 15.903614457831326,
-      "grad_norm": 2.9078729152679443,
-      "learning_rate": 1.1302681992337164e-05,
-      "loss": 0.1267,
+      "epoch": 2.0165866433871673,
+      "grad_norm": 2.8195152282714844,
+      "learning_rate": 4.434740417273169e-05,
+      "loss": 0.2219,
       "step": 2310
     },
     {
-      "epoch": 15.972461273666093,
-      "grad_norm": 2.513744831085205,
-      "learning_rate": 1.1111111111111112e-05,
-      "loss": 0.1249,
+      "epoch": 2.0253164556962027,
+      "grad_norm": 3.702488899230957,
+      "learning_rate": 4.42988840368753e-05,
+      "loss": 0.2212,
       "step": 2320
     },
     {
-      "epoch": 16.0,
-      "eval_accuracy": 0.9186046511627907,
-      "eval_loss": 0.24669910967350006,
-      "eval_runtime": 7.8261,
-      "eval_samples_per_second": 263.732,
-      "eval_steps_per_second": 8.306,
-      "step": 2324
-    },
-    {
-      "epoch": 16.04130808950086,
-      "grad_norm": 2.510911226272583,
-      "learning_rate": 1.091954022988506e-05,
-      "loss": 0.1052,
+      "epoch": 2.034046268005238,
+      "grad_norm": 2.593886613845825,
+      "learning_rate": 4.425036390101893e-05,
+      "loss": 0.2278,
       "step": 2330
     },
     {
-      "epoch": 16.110154905335627,
-      "grad_norm": 2.5557456016540527,
-      "learning_rate": 1.0727969348659005e-05,
-      "loss": 0.1293,
+      "epoch": 2.0427760803142734,
+      "grad_norm": 3.5325887203216553,
+      "learning_rate": 4.420184376516254e-05,
+      "loss": 0.2197,
       "step": 2340
     },
     {
-      "epoch": 16.179001721170398,
-      "grad_norm": 2.11968994140625,
-      "learning_rate": 1.053639846743295e-05,
-      "loss": 0.1139,
+      "epoch": 2.0515058926233087,
+      "grad_norm": 4.355841159820557,
+      "learning_rate": 4.415332362930616e-05,
+      "loss": 0.2384,
       "step": 2350
     },
     {
-      "epoch": 16.247848537005165,
-      "grad_norm": 2.4394736289978027,
-      "learning_rate": 1.0344827586206897e-05,
-      "loss": 0.1206,
+      "epoch": 2.060235704932344,
+      "grad_norm": 3.8640925884246826,
+      "learning_rate": 4.4104803493449784e-05,
+      "loss": 0.2102,
       "step": 2360
     },
     {
-      "epoch": 16.316695352839933,
-      "grad_norm": 2.758610963821411,
-      "learning_rate": 1.0153256704980842e-05,
-      "loss": 0.1098,
+      "epoch": 2.0689655172413794,
+      "grad_norm": 3.6872246265411377,
+      "learning_rate": 4.4056283357593404e-05,
+      "loss": 0.2536,
       "step": 2370
     },
     {
-      "epoch": 16.3855421686747,
-      "grad_norm": 2.410712480545044,
-      "learning_rate": 9.96168582375479e-06,
-      "loss": 0.1052,
+      "epoch": 2.077695329550415,
+      "grad_norm": 3.1011269092559814,
+      "learning_rate": 4.4007763221737025e-05,
+      "loss": 0.2231,
       "step": 2380
     },
     {
-      "epoch": 16.454388984509468,
-      "grad_norm": 2.830049991607666,
-      "learning_rate": 9.770114942528738e-06,
-      "loss": 0.11,
+      "epoch": 2.08642514185945,
+      "grad_norm": 3.202810525894165,
+      "learning_rate": 4.395924308588064e-05,
+      "loss": 0.2609,
       "step": 2390
     },
     {
-      "epoch": 16.523235800344235,
-      "grad_norm": 2.921481132507324,
-      "learning_rate": 9.578544061302683e-06,
-      "loss": 0.1038,
+      "epoch": 2.0951549541684855,
+      "grad_norm": 3.2603683471679688,
+      "learning_rate": 4.3910722950024266e-05,
+      "loss": 0.2367,
       "step": 2400
     },
     {
-      "epoch": 16.592082616179002,
-      "grad_norm": 3.5350148677825928,
-      "learning_rate": 9.386973180076629e-06,
-      "loss": 0.1447,
+      "epoch": 2.103884766477521,
+      "grad_norm": 2.8735013008117676,
+      "learning_rate": 4.386220281416788e-05,
+      "loss": 0.2528,
       "step": 2410
     },
     {
-      "epoch": 16.66092943201377,
-      "grad_norm": 3.2687947750091553,
-      "learning_rate": 9.195402298850575e-06,
-      "loss": 0.119,
+      "epoch": 2.1126145787865562,
+      "grad_norm": 2.745392322540283,
+      "learning_rate": 4.38136826783115e-05,
+      "loss": 0.2259,
       "step": 2420
     },
     {
-      "epoch": 16.729776247848537,
-      "grad_norm": 2.843630790710449,
-      "learning_rate": 9.00383141762452e-06,
-      "loss": 0.1361,
+      "epoch": 2.1213443910955916,
+      "grad_norm": 2.783156633377075,
+      "learning_rate": 4.376516254245512e-05,
+      "loss": 0.2285,
       "step": 2430
     },
     {
-      "epoch": 16.798623063683305,
-      "grad_norm": 2.6418073177337646,
-      "learning_rate": 8.812260536398467e-06,
-      "loss": 0.1125,
+      "epoch": 2.130074203404627,
+      "grad_norm": 3.6078295707702637,
+      "learning_rate": 4.371664240659874e-05,
+      "loss": 0.2222,
       "step": 2440
     },
     {
-      "epoch": 16.867469879518072,
-      "grad_norm": 2.408013343811035,
-      "learning_rate": 8.620689655172414e-06,
-      "loss": 0.1354,
+      "epoch": 2.1388040157136623,
+      "grad_norm": 3.379075527191162,
+      "learning_rate": 4.366812227074236e-05,
+      "loss": 0.2574,
       "step": 2450
     },
     {
-      "epoch": 16.93631669535284,
-      "grad_norm": 3.4269073009490967,
-      "learning_rate": 8.429118773946362e-06,
-      "loss": 0.1504,
+      "epoch": 2.1475338280226977,
+      "grad_norm": 2.965963840484619,
+      "learning_rate": 4.3619602134885976e-05,
+      "loss": 0.2305,
       "step": 2460
     },
     {
-      "epoch": 16.998278829604132,
-      "eval_accuracy": 0.9253875968992248,
-      "eval_loss": 0.232245072722435,
-      "eval_runtime": 7.9776,
-      "eval_samples_per_second": 258.725,
-      "eval_steps_per_second": 8.148,
-      "step": 2469
-    },
-    {
-      "epoch": 17.005163511187607,
-      "grad_norm": 3.9189724922180176,
-      "learning_rate": 8.237547892720307e-06,
-      "loss": 0.1346,
+      "epoch": 2.156263640331733,
+      "grad_norm": 3.110180616378784,
+      "learning_rate": 4.35710819990296e-05,
+      "loss": 0.2127,
       "step": 2470
     },
     {
-      "epoch": 17.074010327022375,
-      "grad_norm": 3.447214126586914,
-      "learning_rate": 8.045977011494253e-06,
-      "loss": 0.1254,
+      "epoch": 2.1649934526407684,
+      "grad_norm": 2.2609825134277344,
+      "learning_rate": 4.352256186317322e-05,
+      "loss": 0.1914,
       "step": 2480
     },
     {
-      "epoch": 17.142857142857142,
-      "grad_norm": 1.5260127782821655,
-      "learning_rate": 7.854406130268199e-06,
-      "loss": 0.1264,
+      "epoch": 2.1737232649498037,
+      "grad_norm": 3.6740341186523438,
+      "learning_rate": 4.347404172731684e-05,
+      "loss": 0.2116,
       "step": 2490
     },
     {
-      "epoch": 17.21170395869191,
-      "grad_norm": 2.7160017490386963,
-      "learning_rate": 7.662835249042145e-06,
-      "loss": 0.1124,
+      "epoch": 2.182453077258839,
+      "grad_norm": 4.600437641143799,
+      "learning_rate": 4.342552159146046e-05,
+      "loss": 0.2694,
       "step": 2500
     },
     {
-      "epoch": 17.280550774526677,
-      "grad_norm": 2.530102491378784,
-      "learning_rate": 7.4712643678160925e-06,
-      "loss": 0.1116,
+      "epoch": 2.1911828895678744,
+      "grad_norm": 2.320915460586548,
+      "learning_rate": 4.337700145560408e-05,
+      "loss": 0.2063,
       "step": 2510
     },
     {
-      "epoch": 17.349397590361445,
-      "grad_norm": 2.009098768234253,
-      "learning_rate": 7.279693486590039e-06,
-      "loss": 0.0977,
+      "epoch": 2.19991270187691,
+      "grad_norm": 3.467024326324463,
+      "learning_rate": 4.33284813197477e-05,
+      "loss": 0.2474,
       "step": 2520
     },
     {
-      "epoch": 17.418244406196212,
-      "grad_norm": 2.3196418285369873,
-      "learning_rate": 7.088122605363985e-06,
-      "loss": 0.1527,
+      "epoch": 2.208642514185945,
+      "grad_norm": 2.8249664306640625,
+      "learning_rate": 4.327996118389131e-05,
+      "loss": 0.2376,
       "step": 2530
     },
     {
-      "epoch": 17.48709122203098,
-      "grad_norm": 2.2051141262054443,
-      "learning_rate": 6.896551724137932e-06,
-      "loss": 0.1285,
+      "epoch": 2.2173723264949805,
+      "grad_norm": 3.371345281600952,
+      "learning_rate": 4.323144104803494e-05,
+      "loss": 0.2286,
       "step": 2540
     },
     {
-      "epoch": 17.555938037865747,
-      "grad_norm": 2.9338431358337402,
-      "learning_rate": 6.7049808429118775e-06,
-      "loss": 0.1182,
+      "epoch": 2.226102138804016,
+      "grad_norm": 2.801051139831543,
+      "learning_rate": 4.3182920912178554e-05,
+      "loss": 0.1987,
       "step": 2550
     },
     {
-      "epoch": 17.624784853700515,
-      "grad_norm": 2.8327407836914062,
-      "learning_rate": 6.513409961685823e-06,
-      "loss": 0.0985,
+      "epoch": 2.234831951113051,
+      "grad_norm": 2.635571002960205,
+      "learning_rate": 4.3134400776322174e-05,
+      "loss": 0.2364,
       "step": 2560
     },
     {
-      "epoch": 17.693631669535286,
-      "grad_norm": 2.8627865314483643,
-      "learning_rate": 6.321839080459771e-06,
-      "loss": 0.1185,
+      "epoch": 2.2435617634220866,
+      "grad_norm": 3.7617597579956055,
+      "learning_rate": 4.3085880640465795e-05,
+      "loss": 0.2607,
       "step": 2570
     },
     {
-      "epoch": 17.762478485370053,
-      "grad_norm": 2.3446600437164307,
-      "learning_rate": 6.130268199233717e-06,
-      "loss": 0.1024,
+      "epoch": 2.252291575731122,
+      "grad_norm": 2.7652156352996826,
+      "learning_rate": 4.3037360504609415e-05,
+      "loss": 0.2227,
       "step": 2580
     },
     {
-      "epoch": 17.83132530120482,
-      "grad_norm": 3.2858073711395264,
-      "learning_rate": 5.938697318007663e-06,
-      "loss": 0.1213,
+      "epoch": 2.2610213880401573,
+      "grad_norm": 2.7807669639587402,
+      "learning_rate": 4.2988840368753036e-05,
+      "loss": 0.2403,
       "step": 2590
     },
     {
-      "epoch": 17.900172117039588,
-      "grad_norm": 2.4796440601348877,
-      "learning_rate": 5.747126436781609e-06,
-      "loss": 0.0974,
+      "epoch": 2.2697512003491926,
+      "grad_norm": 3.653930425643921,
+      "learning_rate": 4.294032023289665e-05,
+      "loss": 0.226,
       "step": 2600
     },
     {
-      "epoch": 17.969018932874356,
-      "grad_norm": 1.4947516918182373,
-      "learning_rate": 5.555555555555556e-06,
-      "loss": 0.0972,
+      "epoch": 2.278481012658228,
+      "grad_norm": 2.039379358291626,
+      "learning_rate": 4.289180009704028e-05,
+      "loss": 0.2299,
       "step": 2610
     },
     {
-      "epoch": 17.99655765920826,
-      "eval_accuracy": 0.9781976744186046,
-      "eval_loss": 0.08412329852581024,
-      "eval_runtime": 8.8996,
-      "eval_samples_per_second": 231.921,
-      "eval_steps_per_second": 7.304,
-      "step": 2614
-    },
-    {
-      "epoch": 18.037865748709123,
-      "grad_norm": 2.7059268951416016,
-      "learning_rate": 5.3639846743295025e-06,
-      "loss": 0.1074,
+      "epoch": 2.2872108249672634,
+      "grad_norm": 2.568441390991211,
+      "learning_rate": 4.284327996118389e-05,
+      "loss": 0.2354,
       "step": 2620
     },
     {
-      "epoch": 18.10671256454389,
-      "grad_norm": 2.0837604999542236,
-      "learning_rate": 5.172413793103448e-06,
-      "loss": 0.106,
+      "epoch": 2.2959406372762987,
+      "grad_norm": 3.425060987472534,
+      "learning_rate": 4.279475982532751e-05,
+      "loss": 0.2334,
       "step": 2630
     },
     {
-      "epoch": 18.175559380378658,
-      "grad_norm": 3.52249813079834,
-      "learning_rate": 4.980842911877395e-06,
-      "loss": 0.1475,
+      "epoch": 2.304670449585334,
+      "grad_norm": 3.790520429611206,
+      "learning_rate": 4.274623968947113e-05,
+      "loss": 0.2126,
       "step": 2640
     },
     {
-      "epoch": 18.244406196213426,
-      "grad_norm": 2.8017690181732178,
-      "learning_rate": 4.789272030651342e-06,
-      "loss": 0.128,
+      "epoch": 2.3134002618943694,
+      "grad_norm": 3.2851438522338867,
+      "learning_rate": 4.269771955361475e-05,
+      "loss": 0.202,
       "step": 2650
     },
     {
-      "epoch": 18.313253012048193,
-      "grad_norm": 1.7369544506072998,
-      "learning_rate": 4.5977011494252875e-06,
-      "loss": 0.0977,
+      "epoch": 2.322130074203405,
+      "grad_norm": 2.7088074684143066,
+      "learning_rate": 4.264919941775837e-05,
+      "loss": 0.2,
       "step": 2660
     },
     {
-      "epoch": 18.38209982788296,
-      "grad_norm": 2.63714599609375,
-      "learning_rate": 4.406130268199233e-06,
-      "loss": 0.1067,
+      "epoch": 2.33085988651244,
+      "grad_norm": 2.6394031047821045,
+      "learning_rate": 4.260067928190199e-05,
+      "loss": 0.2323,
       "step": 2670
     },
     {
-      "epoch": 18.450946643717728,
-      "grad_norm": 2.7819557189941406,
-      "learning_rate": 4.214559386973181e-06,
-      "loss": 0.1364,
+      "epoch": 2.3395896988214755,
+      "grad_norm": 3.6911752223968506,
+      "learning_rate": 4.2552159146045614e-05,
+      "loss": 0.2426,
       "step": 2680
     },
     {
-      "epoch": 18.519793459552496,
-      "grad_norm": 2.259345769882202,
-      "learning_rate": 4.022988505747127e-06,
-      "loss": 0.0957,
+      "epoch": 2.348319511130511,
+      "grad_norm": 2.896843671798706,
+      "learning_rate": 4.250363901018923e-05,
+      "loss": 0.205,
       "step": 2690
     },
     {
-      "epoch": 18.588640275387263,
-      "grad_norm": 2.4775497913360596,
-      "learning_rate": 3.8314176245210725e-06,
-      "loss": 0.1273,
+      "epoch": 2.357049323439546,
+      "grad_norm": 3.258150339126587,
+      "learning_rate": 4.245511887433285e-05,
+      "loss": 0.2154,
       "step": 2700
     },
     {
-      "epoch": 18.65748709122203,
-      "grad_norm": 2.889547824859619,
-      "learning_rate": 3.6398467432950196e-06,
-      "loss": 0.1293,
+      "epoch": 2.3657791357485816,
+      "grad_norm": 2.3189845085144043,
+      "learning_rate": 4.240659873847647e-05,
+      "loss": 0.1969,
       "step": 2710
     },
     {
-      "epoch": 18.726333907056798,
-      "grad_norm": 2.8670730590820312,
-      "learning_rate": 3.448275862068966e-06,
-      "loss": 0.1279,
+      "epoch": 2.374508948057617,
+      "grad_norm": 3.0700485706329346,
+      "learning_rate": 4.235807860262009e-05,
+      "loss": 0.2196,
       "step": 2720
     },
     {
-      "epoch": 18.795180722891565,
-      "grad_norm": 1.9103004932403564,
-      "learning_rate": 3.2567049808429117e-06,
-      "loss": 0.1292,
+      "epoch": 2.3832387603666523,
+      "grad_norm": 2.3715126514434814,
+      "learning_rate": 4.230955846676371e-05,
+      "loss": 0.2214,
       "step": 2730
     },
     {
-      "epoch": 18.864027538726333,
-      "grad_norm": 1.6978858709335327,
-      "learning_rate": 3.0651340996168583e-06,
-      "loss": 0.1182,
+      "epoch": 2.3919685726756876,
+      "grad_norm": 3.8778791427612305,
+      "learning_rate": 4.2261038330907324e-05,
+      "loss": 0.2229,
       "step": 2740
     },
     {
-      "epoch": 18.9328743545611,
-      "grad_norm": 3.6564149856567383,
-      "learning_rate": 2.8735632183908046e-06,
-      "loss": 0.1293,
+      "epoch": 2.400698384984723,
+      "grad_norm": 2.7153546810150146,
+      "learning_rate": 4.221251819505095e-05,
+      "loss": 0.1865,
       "step": 2750
     },
     {
-      "epoch": 18.994836488812393,
-      "eval_accuracy": 0.9467054263565892,
-      "eval_loss": 0.15116389095783234,
-      "eval_runtime": 7.8213,
-      "eval_samples_per_second": 263.893,
-      "eval_steps_per_second": 8.311,
-      "step": 2759
-    },
-    {
-      "epoch": 19.001721170395868,
-      "grad_norm": 3.7127535343170166,
-      "learning_rate": 2.6819923371647512e-06,
-      "loss": 0.1027,
+      "epoch": 2.4094281972937583,
+      "grad_norm": 2.5187482833862305,
+      "learning_rate": 4.216399805919457e-05,
+      "loss": 0.2027,
       "step": 2760
     },
     {
-      "epoch": 19.070567986230635,
-      "grad_norm": 2.170724868774414,
-      "learning_rate": 2.4904214559386975e-06,
-      "loss": 0.0993,
+      "epoch": 2.4181580096027937,
+      "grad_norm": 3.387876510620117,
+      "learning_rate": 4.2115477923338185e-05,
+      "loss": 0.2097,
       "step": 2770
     },
     {
-      "epoch": 19.139414802065403,
-      "grad_norm": 3.131479501724243,
-      "learning_rate": 2.2988505747126437e-06,
-      "loss": 0.1378,
+      "epoch": 2.426887821911829,
+      "grad_norm": 2.9012184143066406,
+      "learning_rate": 4.2066957787481806e-05,
+      "loss": 0.1936,
       "step": 2780
     },
     {
-      "epoch": 19.20826161790017,
-      "grad_norm": 1.942002534866333,
-      "learning_rate": 2.1072796934865904e-06,
-      "loss": 0.1163,
+      "epoch": 2.4356176342208644,
+      "grad_norm": 3.1651363372802734,
+      "learning_rate": 4.2018437651625426e-05,
+      "loss": 0.2453,
       "step": 2790
     },
     {
-      "epoch": 19.27710843373494,
-      "grad_norm": 2.8317081928253174,
-      "learning_rate": 1.9157088122605362e-06,
-      "loss": 0.1467,
+      "epoch": 2.4443474465298998,
+      "grad_norm": 3.4698972702026367,
+      "learning_rate": 4.196991751576905e-05,
+      "loss": 0.2172,
       "step": 2800
     },
     {
-      "epoch": 19.34595524956971,
-      "grad_norm": 1.791185736656189,
-      "learning_rate": 1.724137931034483e-06,
-      "loss": 0.1139,
+      "epoch": 2.453077258838935,
+      "grad_norm": 3.2747743129730225,
+      "learning_rate": 4.192139737991266e-05,
+      "loss": 0.244,
       "step": 2810
     },
     {
-      "epoch": 19.414802065404476,
-      "grad_norm": 2.639554023742676,
-      "learning_rate": 1.5325670498084292e-06,
-      "loss": 0.1139,
+      "epoch": 2.4618070711479705,
+      "grad_norm": 4.365453720092773,
+      "learning_rate": 4.187287724405629e-05,
+      "loss": 0.2208,
       "step": 2820
     },
     {
-      "epoch": 19.483648881239244,
-      "grad_norm": 1.9929583072662354,
-      "learning_rate": 1.3409961685823756e-06,
-      "loss": 0.1093,
+      "epoch": 2.470536883457006,
+      "grad_norm": 3.2108471393585205,
+      "learning_rate": 4.182435710819991e-05,
+      "loss": 0.2253,
       "step": 2830
     },
     {
-      "epoch": 19.55249569707401,
-      "grad_norm": 2.7852931022644043,
-      "learning_rate": 1.1494252873563219e-06,
-      "loss": 0.1305,
+      "epoch": 2.479266695766041,
+      "grad_norm": 3.678192377090454,
+      "learning_rate": 4.177583697234352e-05,
+      "loss": 0.231,
       "step": 2840
     },
     {
-      "epoch": 19.62134251290878,
-      "grad_norm": 3.228508710861206,
-      "learning_rate": 9.578544061302681e-07,
-      "loss": 0.1236,
+      "epoch": 2.4879965080750766,
+      "grad_norm": 3.295539617538452,
+      "learning_rate": 4.172731683648714e-05,
+      "loss": 0.24,
       "step": 2850
     },
     {
-      "epoch": 19.690189328743546,
-      "grad_norm": 3.5496957302093506,
-      "learning_rate": 7.662835249042146e-07,
-      "loss": 0.1357,
+      "epoch": 2.496726320384112,
+      "grad_norm": 3.0643677711486816,
+      "learning_rate": 4.1678796700630764e-05,
+      "loss": 0.2142,
       "step": 2860
     },
     {
-      "epoch": 19.759036144578314,
-      "grad_norm": 1.8435531854629517,
-      "learning_rate": 5.747126436781609e-07,
-      "loss": 0.1134,
+      "epoch": 2.505456132693147,
+      "grad_norm": 3.3195557594299316,
+      "learning_rate": 4.1630276564774384e-05,
+      "loss": 0.2218,
       "step": 2870
     },
     {
-      "epoch": 19.82788296041308,
-      "grad_norm": 3.3433897495269775,
-      "learning_rate": 3.831417624521073e-07,
-      "loss": 0.1472,
+      "epoch": 2.5141859450021826,
+      "grad_norm": 2.9350225925445557,
+      "learning_rate": 4.1581756428918e-05,
+      "loss": 0.1869,
       "step": 2880
     },
     {
-      "epoch": 19.89672977624785,
-      "grad_norm": 2.541207790374756,
-      "learning_rate": 1.9157088122605365e-07,
-      "loss": 0.1057,
+      "epoch": 2.5229157573112175,
+      "grad_norm": 3.483301877975464,
+      "learning_rate": 4.1533236293061625e-05,
+      "loss": 0.1986,
       "step": 2890
     },
     {
-      "epoch": 19.965576592082616,
-      "grad_norm": 3.152078151702881,
-      "learning_rate": 0.0,
-      "loss": 0.1072,
+      "epoch": 2.5316455696202533,
+      "grad_norm": 3.9944510459899902,
+      "learning_rate": 4.1484716157205246e-05,
+      "loss": 0.2112,
       "step": 2900
     },
     {
-      "epoch": 19.965576592082616,
-      "eval_accuracy": 0.9447674418604651,
-      "eval_loss": 0.16629785299301147,
-      "eval_runtime": 7.3483,
-      "eval_samples_per_second": 280.88,
-      "eval_steps_per_second": 8.846,
-      "step": 2900
+      "epoch": 2.5403753819292882,
+      "grad_norm": 3.825562000274658,
+      "learning_rate": 4.143619602134886e-05,
+      "loss": 0.2102,
+      "step": 2910
+    },
+    {
+      "epoch": 2.549105194238324,
+      "grad_norm": 3.575993776321411,
+      "learning_rate": 4.138767588549248e-05,
+      "loss": 0.2303,
+      "step": 2920
+    },
+    {
+      "epoch": 2.557835006547359,
+      "grad_norm": 2.6293749809265137,
+      "learning_rate": 4.13391557496361e-05,
+      "loss": 0.1759,
+      "step": 2930
+    },
+    {
+      "epoch": 2.5665648188563948,
+      "grad_norm": 3.9990651607513428,
+      "learning_rate": 4.129063561377972e-05,
+      "loss": 0.247,
+      "step": 2940
+    },
+    {
+      "epoch": 2.5752946311654297,
+      "grad_norm": 3.152682065963745,
+      "learning_rate": 4.1242115477923335e-05,
+      "loss": 0.22,
+      "step": 2950
+    },
+    {
+      "epoch": 2.5840244434744655,
+      "grad_norm": 2.9134416580200195,
+      "learning_rate": 4.119359534206696e-05,
+      "loss": 0.1966,
+      "step": 2960
+    },
+    {
+      "epoch": 2.5927542557835004,
+      "grad_norm": 4.1252288818359375,
+      "learning_rate": 4.114507520621058e-05,
+      "loss": 0.1766,
+      "step": 2970
+    },
+    {
+      "epoch": 2.601484068092536,
+      "grad_norm": 3.3701207637786865,
+      "learning_rate": 4.1096555070354197e-05,
+      "loss": 0.2062,
+      "step": 2980
+    },
+    {
+      "epoch": 2.610213880401571,
+      "grad_norm": 3.7040228843688965,
+      "learning_rate": 4.104803493449782e-05,
+      "loss": 0.1997,
+      "step": 2990
+    },
+    {
+      "epoch": 2.618943692710607,
+      "grad_norm": 3.8628756999969482,
+      "learning_rate": 4.099951479864144e-05,
+      "loss": 0.2126,
+      "step": 3000
+    },
+    {
+      "epoch": 2.627673505019642,
+      "grad_norm": 3.3776793479919434,
+      "learning_rate": 4.095099466278506e-05,
+      "loss": 0.2277,
+      "step": 3010
+    },
+    {
+      "epoch": 2.6364033173286776,
+      "grad_norm": 1.8289759159088135,
+      "learning_rate": 4.090247452692867e-05,
+      "loss": 0.2033,
+      "step": 3020
+    },
+    {
+      "epoch": 2.6451331296377125,
+      "grad_norm": 3.379987955093384,
+      "learning_rate": 4.08539543910723e-05,
+      "loss": 0.216,
+      "step": 3030
+    },
+    {
+      "epoch": 2.6538629419467483,
+      "grad_norm": 3.1933844089508057,
+      "learning_rate": 4.080543425521592e-05,
+      "loss": 0.175,
+      "step": 3040
+    },
+    {
+      "epoch": 2.6625927542557832,
+      "grad_norm": 4.522855758666992,
+      "learning_rate": 4.0756914119359534e-05,
+      "loss": 0.2276,
+      "step": 3050
+    },
+    {
+      "epoch": 2.671322566564819,
+      "grad_norm": 3.2988367080688477,
+      "learning_rate": 4.0708393983503154e-05,
+      "loss": 0.2112,
+      "step": 3060
+    },
+    {
+      "epoch": 2.680052378873854,
+      "grad_norm": 3.88380765914917,
+      "learning_rate": 4.0659873847646775e-05,
+      "loss": 0.1831,
+      "step": 3070
+    },
+    {
+      "epoch": 2.6887821911828897,
+      "grad_norm": 2.2652664184570312,
+      "learning_rate": 4.0611353711790395e-05,
+      "loss": 0.2062,
+      "step": 3080
+    },
+    {
+      "epoch": 2.6975120034919247,
+      "grad_norm": 2.7714035511016846,
+      "learning_rate": 4.0562833575934016e-05,
+      "loss": 0.2016,
+      "step": 3090
+    },
+    {
+      "epoch": 2.7062418158009605,
+      "grad_norm": 3.18550181388855,
+      "learning_rate": 4.0514313440077636e-05,
+      "loss": 0.2043,
+      "step": 3100
+    },
+    {
+      "epoch": 2.7149716281099954,
+      "grad_norm": 4.249231815338135,
+      "learning_rate": 4.046579330422126e-05,
+      "loss": 0.1868,
+      "step": 3110
+    },
+    {
+      "epoch": 2.723701440419031,
+      "grad_norm": 3.2999160289764404,
+      "learning_rate": 4.041727316836487e-05,
+      "loss": 0.1849,
+      "step": 3120
+    },
+    {
+      "epoch": 2.732431252728066,
+      "grad_norm": 3.1966731548309326,
+      "learning_rate": 4.036875303250849e-05,
+      "loss": 0.2143,
+      "step": 3130
+    },
+    {
+      "epoch": 2.741161065037102,
+      "grad_norm": 3.071214199066162,
+      "learning_rate": 4.032023289665211e-05,
+      "loss": 0.1541,
+      "step": 3140
+    },
+    {
+      "epoch": 2.749890877346137,
+      "grad_norm": 2.8251476287841797,
+      "learning_rate": 4.027171276079573e-05,
+      "loss": 0.1818,
+      "step": 3150
+    },
+    {
+      "epoch": 2.7586206896551726,
+      "grad_norm": 3.3325555324554443,
+      "learning_rate": 4.022319262493935e-05,
+      "loss": 0.2099,
+      "step": 3160
+    },
+    {
+      "epoch": 2.7673505019642075,
+      "grad_norm": 3.3313956260681152,
+      "learning_rate": 4.017467248908297e-05,
+      "loss": 0.2064,
+      "step": 3170
+    },
+    {
+      "epoch": 2.7760803142732433,
+      "grad_norm": 3.0525896549224854,
+      "learning_rate": 4.0126152353226594e-05,
+      "loss": 0.1882,
+      "step": 3180
+    },
+    {
+      "epoch": 2.7848101265822782,
+      "grad_norm": 2.8672525882720947,
+      "learning_rate": 4.007763221737021e-05,
+      "loss": 0.1722,
+      "step": 3190
+    },
+    {
+      "epoch": 2.793539938891314,
+      "grad_norm": 3.050518751144409,
+      "learning_rate": 4.002911208151383e-05,
+      "loss": 0.1933,
+      "step": 3200
     },
     {
-      "epoch": 19.965576592082616,
-      "step": 2900,
-      "total_flos": 9.813099581032366e+17,
-      "train_loss": 0.3346556580683281,
-      "train_runtime": 3422.7405,
-      "train_samples_per_second": 108.533,
-      "train_steps_per_second": 0.847
+      "epoch": 2.802269751200349,
+      "grad_norm": 3.527492046356201,
+      "learning_rate": 3.998059194565745e-05,
+      "loss": 0.1614,
+      "step": 3210
+    },
+    {
+      "epoch": 2.8109995635093847,
+      "grad_norm": 4.246466159820557,
+      "learning_rate": 3.993207180980107e-05,
+      "loss": 0.2099,
+      "step": 3220
+    },
+    {
+      "epoch": 2.8197293758184196,
+      "grad_norm": 3.365166425704956,
+      "learning_rate": 3.988355167394469e-05,
+      "loss": 0.2046,
+      "step": 3230
+    },
+    {
+      "epoch": 2.8284591881274554,
+      "grad_norm": 2.6562325954437256,
+      "learning_rate": 3.983503153808831e-05,
+      "loss": 0.1593,
+      "step": 3240
+    },
+    {
+      "epoch": 2.8371890004364904,
+      "grad_norm": 3.081038236618042,
+      "learning_rate": 3.978651140223193e-05,
+      "loss": 0.1784,
+      "step": 3250
+    },
+    {
+      "epoch": 2.845918812745526,
+      "grad_norm": 3.1560895442962646,
+      "learning_rate": 3.9737991266375545e-05,
+      "loss": 0.2313,
+      "step": 3260
+    },
+    {
+      "epoch": 2.854648625054561,
+      "grad_norm": 3.1035749912261963,
+      "learning_rate": 3.9689471130519165e-05,
+      "loss": 0.1988,
+      "step": 3270
+    },
+    {
+      "epoch": 2.863378437363597,
+      "grad_norm": 3.2834436893463135,
+      "learning_rate": 3.9640950994662786e-05,
+      "loss": 0.2095,
+      "step": 3280
+    },
+    {
+      "epoch": 2.872108249672632,
+      "grad_norm": 2.712871789932251,
+      "learning_rate": 3.9592430858806406e-05,
+      "loss": 0.1723,
+      "step": 3290
+    },
+    {
+      "epoch": 2.8808380619816676,
+      "grad_norm": 2.6281795501708984,
+      "learning_rate": 3.954391072295003e-05,
+      "loss": 0.1921,
+      "step": 3300
+    },
+    {
+      "epoch": 2.8895678742907025,
+      "grad_norm": 2.3966925144195557,
+      "learning_rate": 3.949539058709365e-05,
+      "loss": 0.1927,
+      "step": 3310
+    },
+    {
+      "epoch": 2.8982976865997383,
+      "grad_norm": 3.192667245864868,
+      "learning_rate": 3.944687045123727e-05,
+      "loss": 0.2165,
+      "step": 3320
+    },
+    {
+      "epoch": 2.907027498908773,
+      "grad_norm": 3.994009256362915,
+      "learning_rate": 3.939835031538088e-05,
+      "loss": 0.2054,
+      "step": 3330
+    },
+    {
+      "epoch": 2.915757311217809,
+      "grad_norm": 3.6502673625946045,
+      "learning_rate": 3.93498301795245e-05,
+      "loss": 0.2229,
+      "step": 3340
+    },
+    {
+      "epoch": 2.924487123526844,
+      "grad_norm": 3.1017322540283203,
+      "learning_rate": 3.930131004366812e-05,
+      "loss": 0.1659,
+      "step": 3350
+    },
+    {
+      "epoch": 2.9332169358358797,
+      "grad_norm": 3.0115444660186768,
+      "learning_rate": 3.9252789907811743e-05,
+      "loss": 0.182,
+      "step": 3360
+    },
+    {
+      "epoch": 2.9419467481449146,
+      "grad_norm": 3.564761161804199,
+      "learning_rate": 3.9204269771955364e-05,
+      "loss": 0.2281,
+      "step": 3370
+    },
+    {
+      "epoch": 2.9506765604539504,
+      "grad_norm": 2.8959414958953857,
+      "learning_rate": 3.9155749636098985e-05,
+      "loss": 0.1896,
+      "step": 3380
+    },
+    {
+      "epoch": 2.9594063727629853,
+      "grad_norm": 2.9973928928375244,
+      "learning_rate": 3.9107229500242605e-05,
+      "loss": 0.2058,
+      "step": 3390
+    },
+    {
+      "epoch": 2.968136185072021,
+      "grad_norm": 3.4878735542297363,
+      "learning_rate": 3.905870936438622e-05,
+      "loss": 0.1839,
+      "step": 3400
+    },
+    {
+      "epoch": 2.976865997381056,
+      "grad_norm": 2.474815607070923,
+      "learning_rate": 3.901018922852984e-05,
+      "loss": 0.2162,
+      "step": 3410
+    },
+    {
+      "epoch": 2.985595809690092,
+      "grad_norm": 3.7291135787963867,
+      "learning_rate": 3.896166909267347e-05,
+      "loss": 0.1895,
+      "step": 3420
+    },
+    {
+      "epoch": 2.9943256219991268,
+      "grad_norm": 3.1307461261749268,
+      "learning_rate": 3.891314895681708e-05,
+      "loss": 0.2058,
+      "step": 3430
+    },
+    {
+      "epoch": 2.999563509384548,
+      "eval_accuracy": 0.9558679106309845,
+      "eval_loss": 0.14920948445796967,
+      "eval_runtime": 61.5134,
+      "eval_samples_per_second": 264.853,
+      "eval_steps_per_second": 8.291,
+      "step": 3436
+    },
+    {
+      "epoch": 3.0030554343081626,
+      "grad_norm": 3.64821457862854,
+      "learning_rate": 3.88646288209607e-05,
+      "loss": 0.1772,
+      "step": 3440
+    },
+    {
+      "epoch": 3.011785246617198,
+      "grad_norm": 2.404139280319214,
+      "learning_rate": 3.881610868510432e-05,
+      "loss": 0.1475,
+      "step": 3450
+    },
+    {
+      "epoch": 3.0205150589262333,
+      "grad_norm": 2.3943281173706055,
+      "learning_rate": 3.876758854924794e-05,
+      "loss": 0.1839,
+      "step": 3460
+    },
+    {
+      "epoch": 3.0292448712352686,
+      "grad_norm": 2.9818949699401855,
+      "learning_rate": 3.8719068413391556e-05,
+      "loss": 0.1981,
+      "step": 3470
+    },
+    {
+      "epoch": 3.037974683544304,
+      "grad_norm": 4.090831279754639,
+      "learning_rate": 3.8670548277535176e-05,
+      "loss": 0.1597,
+      "step": 3480
+    },
+    {
+      "epoch": 3.0467044958533394,
+      "grad_norm": 3.2076456546783447,
+      "learning_rate": 3.8622028141678804e-05,
+      "loss": 0.2012,
+      "step": 3490
+    },
+    {
+      "epoch": 3.0554343081623747,
+      "grad_norm": 3.2840895652770996,
+      "learning_rate": 3.857350800582242e-05,
+      "loss": 0.1901,
+      "step": 3500
+    },
+    {
+      "epoch": 3.06416412047141,
+      "grad_norm": 2.6999125480651855,
+      "learning_rate": 3.852498786996604e-05,
+      "loss": 0.1808,
+      "step": 3510
+    },
+    {
+      "epoch": 3.0728939327804454,
+      "grad_norm": 2.939896821975708,
+      "learning_rate": 3.847646773410966e-05,
+      "loss": 0.185,
+      "step": 3520
+    },
+    {
+      "epoch": 3.081623745089481,
+      "grad_norm": 3.7036166191101074,
+      "learning_rate": 3.842794759825328e-05,
+      "loss": 0.2125,
+      "step": 3530
+    },
+    {
+      "epoch": 3.090353557398516,
+      "grad_norm": 3.631962299346924,
+      "learning_rate": 3.837942746239689e-05,
+      "loss": 0.1828,
+      "step": 3540
+    },
+    {
+      "epoch": 3.0990833697075515,
+      "grad_norm": 2.847594976425171,
+      "learning_rate": 3.8330907326540513e-05,
+      "loss": 0.1442,
+      "step": 3550
+    },
+    {
+      "epoch": 3.107813182016587,
+      "grad_norm": 2.8544461727142334,
+      "learning_rate": 3.828238719068414e-05,
+      "loss": 0.1772,
+      "step": 3560
+    },
+    {
+      "epoch": 3.116542994325622,
+      "grad_norm": 2.833056688308716,
+      "learning_rate": 3.8233867054827755e-05,
+      "loss": 0.1847,
+      "step": 3570
+    },
+    {
+      "epoch": 3.1252728066346576,
+      "grad_norm": 3.2772698402404785,
+      "learning_rate": 3.8185346918971375e-05,
+      "loss": 0.1872,
+      "step": 3580
+    },
+    {
+      "epoch": 3.1340026189436925,
+      "grad_norm": 2.8444337844848633,
+      "learning_rate": 3.8136826783114996e-05,
+      "loss": 0.1829,
+      "step": 3590
+    },
+    {
+      "epoch": 3.1427324312527283,
+      "grad_norm": 4.018392562866211,
+      "learning_rate": 3.8088306647258616e-05,
+      "loss": 0.1709,
+      "step": 3600
+    },
+    {
+      "epoch": 3.151462243561763,
+      "grad_norm": 3.0900518894195557,
+      "learning_rate": 3.803978651140223e-05,
+      "loss": 0.1551,
+      "step": 3610
+    },
+    {
+      "epoch": 3.160192055870799,
+      "grad_norm": 2.1413562297821045,
+      "learning_rate": 3.799126637554585e-05,
+      "loss": 0.1867,
+      "step": 3620
+    },
+    {
+      "epoch": 3.168921868179834,
+      "grad_norm": 3.050671100616455,
+      "learning_rate": 3.794274623968948e-05,
+      "loss": 0.1884,
+      "step": 3630
+    },
+    {
+      "epoch": 3.1776516804888697,
+      "grad_norm": 3.41404390335083,
+      "learning_rate": 3.789422610383309e-05,
+      "loss": 0.1864,
+      "step": 3640
+    },
+    {
+      "epoch": 3.1863814927979046,
+      "grad_norm": 3.2197773456573486,
+      "learning_rate": 3.784570596797671e-05,
+      "loss": 0.1954,
+      "step": 3650
+    },
+    {
+      "epoch": 3.1951113051069404,
+      "grad_norm": 4.097252368927002,
+      "learning_rate": 3.779718583212033e-05,
+      "loss": 0.1938,
+      "step": 3660
+    },
+    {
+      "epoch": 3.2038411174159753,
+      "grad_norm": 3.3386523723602295,
+      "learning_rate": 3.774866569626395e-05,
+      "loss": 0.1995,
+      "step": 3670
+    },
+    {
+      "epoch": 3.212570929725011,
+      "grad_norm": 2.2313122749328613,
+      "learning_rate": 3.770014556040757e-05,
+      "loss": 0.1557,
+      "step": 3680
+    },
+    {
+      "epoch": 3.221300742034046,
+      "grad_norm": 2.797321081161499,
+      "learning_rate": 3.765162542455119e-05,
+      "loss": 0.1771,
+      "step": 3690
+    },
+    {
+      "epoch": 3.230030554343082,
+      "grad_norm": 3.4761507511138916,
+      "learning_rate": 3.7603105288694815e-05,
+      "loss": 0.1603,
+      "step": 3700
+    },
+    {
+      "epoch": 3.2387603666521168,
+      "grad_norm": 3.6318702697753906,
+      "learning_rate": 3.755458515283843e-05,
+      "loss": 0.1596,
+      "step": 3710
+    },
+    {
+      "epoch": 3.2474901789611526,
+      "grad_norm": 2.6632442474365234,
+      "learning_rate": 3.750606501698205e-05,
+      "loss": 0.1559,
+      "step": 3720
+    },
+    {
+      "epoch": 3.2562199912701875,
+      "grad_norm": 3.2897541522979736,
+      "learning_rate": 3.745754488112567e-05,
+      "loss": 0.149,
+      "step": 3730
+    },
+    {
+      "epoch": 3.2649498035792233,
+      "grad_norm": 2.372366189956665,
+      "learning_rate": 3.740902474526929e-05,
+      "loss": 0.1909,
+      "step": 3740
+    },
+    {
+      "epoch": 3.273679615888258,
+      "grad_norm": 4.001101493835449,
+      "learning_rate": 3.736050460941291e-05,
+      "loss": 0.1891,
+      "step": 3750
+    },
+    {
+      "epoch": 3.282409428197294,
+      "grad_norm": 3.442195177078247,
+      "learning_rate": 3.7311984473556525e-05,
+      "loss": 0.1883,
+      "step": 3760
+    },
+    {
+      "epoch": 3.291139240506329,
+      "grad_norm": 3.3341877460479736,
+      "learning_rate": 3.726346433770015e-05,
+      "loss": 0.183,
+      "step": 3770
+    },
+    {
+      "epoch": 3.2998690528153647,
+      "grad_norm": 2.7835216522216797,
+      "learning_rate": 3.7214944201843766e-05,
+      "loss": 0.1593,
+      "step": 3780
+    },
+    {
+      "epoch": 3.3085988651243996,
+      "grad_norm": 4.420855522155762,
+      "learning_rate": 3.7166424065987386e-05,
+      "loss": 0.1795,
+      "step": 3790
+    },
+    {
+      "epoch": 3.3173286774334354,
+      "grad_norm": 2.7410755157470703,
+      "learning_rate": 3.711790393013101e-05,
+      "loss": 0.1519,
+      "step": 3800
+    },
+    {
+      "epoch": 3.3260584897424703,
+      "grad_norm": 3.839470863342285,
+      "learning_rate": 3.706938379427463e-05,
+      "loss": 0.2107,
+      "step": 3810
+    },
+    {
+      "epoch": 3.334788302051506,
+      "grad_norm": 2.829495668411255,
+      "learning_rate": 3.702086365841825e-05,
+      "loss": 0.1765,
+      "step": 3820
+    },
+    {
+      "epoch": 3.343518114360541,
+      "grad_norm": 3.0382578372955322,
+      "learning_rate": 3.697234352256186e-05,
+      "loss": 0.1899,
+      "step": 3830
+    },
+    {
+      "epoch": 3.352247926669577,
+      "grad_norm": 2.4491844177246094,
+      "learning_rate": 3.692382338670549e-05,
+      "loss": 0.1659,
+      "step": 3840
+    },
+    {
+      "epoch": 3.3609777389786117,
+      "grad_norm": 3.5624337196350098,
+      "learning_rate": 3.68753032508491e-05,
+      "loss": 0.1681,
+      "step": 3850
+    },
+    {
+      "epoch": 3.3697075512876475,
+      "grad_norm": 3.9547486305236816,
+      "learning_rate": 3.682678311499272e-05,
+      "loss": 0.2041,
+      "step": 3860
+    },
+    {
+      "epoch": 3.3784373635966825,
+      "grad_norm": 3.0073511600494385,
+      "learning_rate": 3.6778262979136344e-05,
+      "loss": 0.1754,
+      "step": 3870
+    },
+    {
+      "epoch": 3.3871671759057183,
+      "grad_norm": 3.345168352127075,
+      "learning_rate": 3.6729742843279964e-05,
+      "loss": 0.1714,
+      "step": 3880
+    },
+    {
+      "epoch": 3.395896988214753,
+      "grad_norm": 2.464186668395996,
+      "learning_rate": 3.6681222707423585e-05,
+      "loss": 0.1861,
+      "step": 3890
+    },
+    {
+      "epoch": 3.404626800523789,
+      "grad_norm": 2.6841201782226562,
+      "learning_rate": 3.66327025715672e-05,
+      "loss": 0.1676,
+      "step": 3900
+    },
+    {
+      "epoch": 3.413356612832824,
+      "grad_norm": 2.905978202819824,
+      "learning_rate": 3.6584182435710826e-05,
+      "loss": 0.1952,
+      "step": 3910
+    },
+    {
+      "epoch": 3.4220864251418597,
+      "grad_norm": 2.9638357162475586,
+      "learning_rate": 3.653566229985444e-05,
+      "loss": 0.1617,
+      "step": 3920
+    },
+    {
+      "epoch": 3.4308162374508946,
+      "grad_norm": 3.0130560398101807,
+      "learning_rate": 3.648714216399806e-05,
+      "loss": 0.1554,
+      "step": 3930
+    },
+    {
+      "epoch": 3.4395460497599304,
+      "grad_norm": 3.267518997192383,
+      "learning_rate": 3.643862202814168e-05,
+      "loss": 0.1896,
+      "step": 3940
+    },
+    {
+      "epoch": 3.4482758620689653,
+      "grad_norm": 2.6104979515075684,
+      "learning_rate": 3.63901018922853e-05,
+      "loss": 0.1693,
+      "step": 3950
+    },
+    {
+      "epoch": 3.457005674378001,
+      "grad_norm": 2.9736759662628174,
+      "learning_rate": 3.634158175642892e-05,
+      "loss": 0.1654,
+      "step": 3960
+    },
+    {
+      "epoch": 3.465735486687036,
+      "grad_norm": 3.04263973236084,
+      "learning_rate": 3.6293061620572536e-05,
+      "loss": 0.1655,
+      "step": 3970
+    },
+    {
+      "epoch": 3.474465298996072,
+      "grad_norm": 3.322094202041626,
+      "learning_rate": 3.624454148471616e-05,
+      "loss": 0.1718,
+      "step": 3980
+    },
+    {
+      "epoch": 3.4831951113051067,
+      "grad_norm": 2.8771190643310547,
+      "learning_rate": 3.619602134885978e-05,
+      "loss": 0.188,
+      "step": 3990
+    },
+    {
+      "epoch": 3.4919249236141425,
+      "grad_norm": 3.3343777656555176,
+      "learning_rate": 3.61475012130034e-05,
+      "loss": 0.1765,
+      "step": 4000
+    },
+    {
+      "epoch": 3.5006547359231774,
+      "grad_norm": 3.087955951690674,
+      "learning_rate": 3.609898107714702e-05,
+      "loss": 0.1557,
+      "step": 4010
+    },
+    {
+      "epoch": 3.5093845482322132,
+      "grad_norm": 2.9191036224365234,
+      "learning_rate": 3.605046094129064e-05,
+      "loss": 0.1657,
+      "step": 4020
+    },
+    {
+      "epoch": 3.518114360541248,
+      "grad_norm": 3.2453222274780273,
+      "learning_rate": 3.600194080543426e-05,
+      "loss": 0.1921,
+      "step": 4030
+    },
+    {
+      "epoch": 3.526844172850284,
+      "grad_norm": 3.4277548789978027,
+      "learning_rate": 3.595342066957787e-05,
+      "loss": 0.1583,
+      "step": 4040
+    },
+    {
+      "epoch": 3.535573985159319,
+      "grad_norm": 2.132359504699707,
+      "learning_rate": 3.59049005337215e-05,
+      "loss": 0.168,
+      "step": 4050
+    },
+    {
+      "epoch": 3.5443037974683547,
+      "grad_norm": 3.881998062133789,
+      "learning_rate": 3.5856380397865114e-05,
+      "loss": 0.1992,
+      "step": 4060
+    },
+    {
+      "epoch": 3.5530336097773896,
+      "grad_norm": 3.376692771911621,
+      "learning_rate": 3.5807860262008734e-05,
+      "loss": 0.1804,
+      "step": 4070
+    },
+    {
+      "epoch": 3.5617634220864254,
+      "grad_norm": 2.9599671363830566,
+      "learning_rate": 3.5759340126152355e-05,
+      "loss": 0.1561,
+      "step": 4080
+    },
+    {
+      "epoch": 3.5704932343954603,
+      "grad_norm": 3.439408302307129,
+      "learning_rate": 3.5710819990295976e-05,
+      "loss": 0.1471,
+      "step": 4090
+    },
+    {
+      "epoch": 3.579223046704496,
+      "grad_norm": 3.4653496742248535,
+      "learning_rate": 3.5662299854439596e-05,
+      "loss": 0.189,
+      "step": 4100
+    },
+    {
+      "epoch": 3.587952859013531,
+      "grad_norm": 3.479707717895508,
+      "learning_rate": 3.561377971858321e-05,
+      "loss": 0.1771,
+      "step": 4110
+    },
+    {
+      "epoch": 3.596682671322567,
+      "grad_norm": 3.077345848083496,
+      "learning_rate": 3.556525958272684e-05,
+      "loss": 0.1764,
+      "step": 4120
+    },
+    {
+      "epoch": 3.6054124836316017,
+      "grad_norm": 2.612633228302002,
+      "learning_rate": 3.551673944687045e-05,
+      "loss": 0.1336,
+      "step": 4130
+    },
+    {
+      "epoch": 3.6141422959406375,
+      "grad_norm": 2.649402379989624,
+      "learning_rate": 3.546821931101407e-05,
+      "loss": 0.1609,
+      "step": 4140
+    },
+    {
+      "epoch": 3.6228721082496724,
+      "grad_norm": 3.7441020011901855,
+      "learning_rate": 3.541969917515769e-05,
+      "loss": 0.1758,
+      "step": 4150
+    },
+    {
+      "epoch": 3.6316019205587082,
+      "grad_norm": 2.341038465499878,
+      "learning_rate": 3.537117903930131e-05,
+      "loss": 0.1431,
+      "step": 4160
+    },
+    {
+      "epoch": 3.640331732867743,
+      "grad_norm": 2.990847587585449,
+      "learning_rate": 3.532265890344493e-05,
+      "loss": 0.1621,
+      "step": 4170
+    },
+    {
+      "epoch": 3.649061545176779,
+      "grad_norm": 3.005911111831665,
+      "learning_rate": 3.527413876758855e-05,
+      "loss": 0.1481,
+      "step": 4180
+    },
+    {
+      "epoch": 3.657791357485814,
+      "grad_norm": 3.05849027633667,
+      "learning_rate": 3.5225618631732174e-05,
+      "loss": 0.1781,
+      "step": 4190
+    },
+    {
+      "epoch": 3.6665211697948497,
+      "grad_norm": 2.643735647201538,
+      "learning_rate": 3.517709849587579e-05,
+      "loss": 0.1688,
+      "step": 4200
+    },
+    {
+      "epoch": 3.6752509821038846,
+      "grad_norm": 3.1337473392486572,
+      "learning_rate": 3.512857836001941e-05,
+      "loss": 0.1463,
+      "step": 4210
+    },
+    {
+      "epoch": 3.6839807944129204,
+      "grad_norm": 2.945814609527588,
+      "learning_rate": 3.508005822416303e-05,
+      "loss": 0.1545,
+      "step": 4220
+    },
+    {
+      "epoch": 3.6927106067219553,
+      "grad_norm": 2.5990426540374756,
+      "learning_rate": 3.503153808830665e-05,
+      "loss": 0.1525,
+      "step": 4230
+    },
+    {
+      "epoch": 3.701440419030991,
+      "grad_norm": 2.7641971111297607,
+      "learning_rate": 3.498301795245027e-05,
+      "loss": 0.1661,
+      "step": 4240
+    },
+    {
+      "epoch": 3.710170231340026,
+      "grad_norm": 2.606818437576294,
+      "learning_rate": 3.4934497816593884e-05,
+      "loss": 0.1906,
+      "step": 4250
+    },
+    {
+      "epoch": 3.718900043649062,
+      "grad_norm": 3.2500600814819336,
+      "learning_rate": 3.488597768073751e-05,
+      "loss": 0.1404,
+      "step": 4260
+    },
+    {
+      "epoch": 3.7276298559580967,
+      "grad_norm": 3.6567978858947754,
+      "learning_rate": 3.4837457544881125e-05,
+      "loss": 0.1715,
+      "step": 4270
+    },
+    {
+      "epoch": 3.7363596682671325,
+      "grad_norm": 2.9746952056884766,
+      "learning_rate": 3.4788937409024746e-05,
+      "loss": 0.1633,
+      "step": 4280
+    },
+    {
+      "epoch": 3.7450894805761674,
+      "grad_norm": 2.986154079437256,
+      "learning_rate": 3.4740417273168366e-05,
+      "loss": 0.141,
+      "step": 4290
+    },
+    {
+      "epoch": 3.753819292885203,
+      "grad_norm": 2.4456489086151123,
+      "learning_rate": 3.469189713731199e-05,
+      "loss": 0.1576,
+      "step": 4300
+    },
+    {
+      "epoch": 3.762549105194238,
+      "grad_norm": 2.2719054222106934,
+      "learning_rate": 3.464337700145561e-05,
+      "loss": 0.158,
+      "step": 4310
+    },
+    {
+      "epoch": 3.771278917503274,
+      "grad_norm": 2.6917176246643066,
+      "learning_rate": 3.459485686559922e-05,
+      "loss": 0.1624,
+      "step": 4320
+    },
+    {
+      "epoch": 3.780008729812309,
+      "grad_norm": 3.041710615158081,
+      "learning_rate": 3.454633672974285e-05,
+      "loss": 0.1573,
+      "step": 4330
+    },
+    {
+      "epoch": 3.7887385421213446,
+      "grad_norm": 2.517232894897461,
+      "learning_rate": 3.449781659388647e-05,
+      "loss": 0.1256,
+      "step": 4340
+    },
+    {
+      "epoch": 3.7974683544303796,
+      "grad_norm": 3.19498348236084,
+      "learning_rate": 3.444929645803008e-05,
+      "loss": 0.1399,
+      "step": 4350
+    },
+    {
+      "epoch": 3.8061981667394154,
+      "grad_norm": 3.773149013519287,
+      "learning_rate": 3.44007763221737e-05,
+      "loss": 0.1627,
+      "step": 4360
+    },
+    {
+      "epoch": 3.8149279790484503,
+      "grad_norm": 3.985891819000244,
+      "learning_rate": 3.4352256186317324e-05,
+      "loss": 0.161,
+      "step": 4370
+    },
+    {
+      "epoch": 3.823657791357486,
+      "grad_norm": 2.994173049926758,
+      "learning_rate": 3.4303736050460944e-05,
+      "loss": 0.1604,
+      "step": 4380
+    },
+    {
+      "epoch": 3.832387603666521,
+      "grad_norm": 3.8228750228881836,
+      "learning_rate": 3.425521591460456e-05,
+      "loss": 0.1609,
+      "step": 4390
+    },
+    {
+      "epoch": 3.841117415975557,
+      "grad_norm": 3.412660598754883,
+      "learning_rate": 3.4206695778748185e-05,
+      "loss": 0.1591,
+      "step": 4400
+    },
+    {
+      "epoch": 3.8498472282845917,
+      "grad_norm": 2.460543632507324,
+      "learning_rate": 3.4158175642891806e-05,
+      "loss": 0.171,
+      "step": 4410
+    },
+    {
+      "epoch": 3.8585770405936275,
+      "grad_norm": 2.8546485900878906,
+      "learning_rate": 3.410965550703542e-05,
+      "loss": 0.1662,
+      "step": 4420
+    },
+    {
+      "epoch": 3.8673068529026624,
+      "grad_norm": 3.296644926071167,
+      "learning_rate": 3.406113537117904e-05,
+      "loss": 0.1669,
+      "step": 4430
+    },
+    {
+      "epoch": 3.876036665211698,
+      "grad_norm": 3.5040674209594727,
+      "learning_rate": 3.401261523532266e-05,
+      "loss": 0.1694,
+      "step": 4440
+    },
+    {
+      "epoch": 3.884766477520733,
+      "grad_norm": 3.1331686973571777,
+      "learning_rate": 3.396409509946628e-05,
+      "loss": 0.1507,
+      "step": 4450
+    },
+    {
+      "epoch": 3.893496289829769,
+      "grad_norm": 3.2440221309661865,
+      "learning_rate": 3.3915574963609895e-05,
+      "loss": 0.1727,
+      "step": 4460
+    },
+    {
+      "epoch": 3.902226102138804,
+      "grad_norm": 2.514347553253174,
+      "learning_rate": 3.386705482775352e-05,
+      "loss": 0.1401,
+      "step": 4470
+    },
+    {
+      "epoch": 3.9109559144478396,
+      "grad_norm": 3.354827404022217,
+      "learning_rate": 3.381853469189714e-05,
+      "loss": 0.1472,
+      "step": 4480
+    },
+    {
+      "epoch": 3.9196857267568745,
+      "grad_norm": 2.985978126525879,
+      "learning_rate": 3.377001455604076e-05,
+      "loss": 0.1783,
+      "step": 4490
+    },
+    {
+      "epoch": 3.9284155390659103,
+      "grad_norm": 2.6409835815429688,
+      "learning_rate": 3.372149442018438e-05,
+      "loss": 0.1468,
+      "step": 4500
+    },
+    {
+      "epoch": 3.9371453513749453,
+      "grad_norm": 2.718919515609741,
+      "learning_rate": 3.3672974284328e-05,
+      "loss": 0.1672,
+      "step": 4510
+    },
+    {
+      "epoch": 3.945875163683981,
+      "grad_norm": 3.5522565841674805,
+      "learning_rate": 3.362445414847162e-05,
+      "loss": 0.1666,
+      "step": 4520
+    },
+    {
+      "epoch": 3.954604975993016,
+      "grad_norm": 3.1858441829681396,
+      "learning_rate": 3.357593401261523e-05,
+      "loss": 0.181,
+      "step": 4530
+    },
+    {
+      "epoch": 3.9633347883020518,
+      "grad_norm": 3.0110580921173096,
+      "learning_rate": 3.352741387675886e-05,
+      "loss": 0.1729,
+      "step": 4540
+    },
+    {
+      "epoch": 3.9720646006110867,
+      "grad_norm": 3.0811493396759033,
+      "learning_rate": 3.347889374090248e-05,
+      "loss": 0.1651,
+      "step": 4550
+    },
+    {
+      "epoch": 3.9807944129201225,
+      "grad_norm": 2.8543362617492676,
+      "learning_rate": 3.3430373605046094e-05,
+      "loss": 0.1338,
+      "step": 4560
+    },
+    {
+      "epoch": 3.9895242252291574,
+      "grad_norm": 3.2236084938049316,
+      "learning_rate": 3.3381853469189714e-05,
+      "loss": 0.1349,
+      "step": 4570
+    },
+    {
+      "epoch": 3.998254037538193,
+      "grad_norm": 3.425377368927002,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.1524,
+      "step": 4580
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.9694328504787626,
+      "eval_loss": 0.10254349559545517,
+      "eval_runtime": 60.9682,
+      "eval_samples_per_second": 267.221,
+      "eval_steps_per_second": 8.365,
+      "step": 4582
+    },
+    {
+      "epoch": 4.006983849847228,
+      "grad_norm": 3.464984655380249,
+      "learning_rate": 3.3284813197476955e-05,
+      "loss": 0.1749,
+      "step": 4590
+    },
+    {
+      "epoch": 4.015713662156264,
+      "grad_norm": 3.4568910598754883,
+      "learning_rate": 3.323629306162057e-05,
+      "loss": 0.1637,
+      "step": 4600
+    },
+    {
+      "epoch": 4.024443474465299,
+      "grad_norm": 4.279798984527588,
+      "learning_rate": 3.3187772925764197e-05,
+      "loss": 0.1673,
+      "step": 4610
+    },
+    {
+      "epoch": 4.033173286774335,
+      "grad_norm": 2.7573869228363037,
+      "learning_rate": 3.313925278990782e-05,
+      "loss": 0.1473,
+      "step": 4620
+    },
+    {
+      "epoch": 4.0419030990833695,
+      "grad_norm": 1.6597820520401,
+      "learning_rate": 3.309073265405143e-05,
+      "loss": 0.1903,
+      "step": 4630
+    },
+    {
+      "epoch": 4.050632911392405,
+      "grad_norm": 2.1218221187591553,
+      "learning_rate": 3.304221251819505e-05,
+      "loss": 0.136,
+      "step": 4640
+    },
+    {
+      "epoch": 4.05936272370144,
+      "grad_norm": 2.941210985183716,
+      "learning_rate": 3.299369238233867e-05,
+      "loss": 0.1613,
+      "step": 4650
+    },
+    {
+      "epoch": 4.068092536010476,
+      "grad_norm": 2.4769389629364014,
+      "learning_rate": 3.294517224648229e-05,
+      "loss": 0.1299,
+      "step": 4660
+    },
+    {
+      "epoch": 4.076822348319511,
+      "grad_norm": 2.917901039123535,
+      "learning_rate": 3.289665211062591e-05,
+      "loss": 0.1522,
+      "step": 4670
+    },
+    {
+      "epoch": 4.085552160628547,
+      "grad_norm": 2.216571092605591,
+      "learning_rate": 3.2848131974769534e-05,
+      "loss": 0.1137,
+      "step": 4680
+    },
+    {
+      "epoch": 4.094281972937582,
+      "grad_norm": 3.3516063690185547,
+      "learning_rate": 3.2799611838913154e-05,
+      "loss": 0.1479,
+      "step": 4690
+    },
+    {
+      "epoch": 4.1030117852466175,
+      "grad_norm": 2.5216290950775146,
+      "learning_rate": 3.275109170305677e-05,
+      "loss": 0.1792,
+      "step": 4700
+    },
+    {
+      "epoch": 4.111741597555652,
+      "grad_norm": 3.8577330112457275,
+      "learning_rate": 3.270257156720039e-05,
+      "loss": 0.1647,
+      "step": 4710
+    },
+    {
+      "epoch": 4.120471409864688,
+      "grad_norm": 3.3492794036865234,
+      "learning_rate": 3.265405143134401e-05,
+      "loss": 0.1646,
+      "step": 4720
+    },
+    {
+      "epoch": 4.129201222173723,
+      "grad_norm": 2.7044191360473633,
+      "learning_rate": 3.260553129548763e-05,
+      "loss": 0.1373,
+      "step": 4730
+    },
+    {
+      "epoch": 4.137931034482759,
+      "grad_norm": 2.3833065032958984,
+      "learning_rate": 3.255701115963125e-05,
+      "loss": 0.1606,
+      "step": 4740
+    },
+    {
+      "epoch": 4.146660846791794,
+      "grad_norm": 2.541414260864258,
+      "learning_rate": 3.250849102377487e-05,
+      "loss": 0.1249,
+      "step": 4750
+    },
+    {
+      "epoch": 4.15539065910083,
+      "grad_norm": 3.2906863689422607,
+      "learning_rate": 3.245997088791849e-05,
+      "loss": 0.1458,
+      "step": 4760
+    },
+    {
+      "epoch": 4.1641204714098645,
+      "grad_norm": 2.537529706954956,
+      "learning_rate": 3.2411450752062105e-05,
+      "loss": 0.1559,
+      "step": 4770
+    },
+    {
+      "epoch": 4.1728502837189,
+      "grad_norm": 3.309135675430298,
+      "learning_rate": 3.2362930616205726e-05,
+      "loss": 0.1478,
+      "step": 4780
+    },
+    {
+      "epoch": 4.181580096027935,
+      "grad_norm": 3.2824246883392334,
+      "learning_rate": 3.2314410480349346e-05,
+      "loss": 0.1532,
+      "step": 4790
+    },
+    {
+      "epoch": 4.190309908336971,
+      "grad_norm": 2.130497694015503,
+      "learning_rate": 3.226589034449297e-05,
+      "loss": 0.1552,
+      "step": 4800
+    },
+    {
+      "epoch": 4.199039720646006,
+      "grad_norm": 2.358793020248413,
+      "learning_rate": 3.221737020863659e-05,
+      "loss": 0.1633,
+      "step": 4810
+    },
+    {
+      "epoch": 4.207769532955042,
+      "grad_norm": 2.5943782329559326,
+      "learning_rate": 3.216885007278021e-05,
+      "loss": 0.1277,
+      "step": 4820
+    },
+    {
+      "epoch": 4.216499345264077,
+      "grad_norm": 2.494443416595459,
+      "learning_rate": 3.212032993692383e-05,
+      "loss": 0.1644,
+      "step": 4830
+    },
+    {
+      "epoch": 4.2252291575731125,
+      "grad_norm": 2.339279890060425,
+      "learning_rate": 3.207180980106744e-05,
+      "loss": 0.1607,
+      "step": 4840
+    },
+    {
+      "epoch": 4.233958969882147,
+      "grad_norm": 2.0483193397521973,
+      "learning_rate": 3.202328966521106e-05,
+      "loss": 0.1386,
+      "step": 4850
+    },
+    {
+      "epoch": 4.242688782191183,
+      "grad_norm": 3.7117726802825928,
+      "learning_rate": 3.197476952935468e-05,
+      "loss": 0.1412,
+      "step": 4860
+    },
+    {
+      "epoch": 4.251418594500218,
+      "grad_norm": 1.738502860069275,
+      "learning_rate": 3.1926249393498304e-05,
+      "loss": 0.141,
+      "step": 4870
+    },
+    {
+      "epoch": 4.260148406809254,
+      "grad_norm": 3.4214649200439453,
+      "learning_rate": 3.1877729257641924e-05,
+      "loss": 0.164,
+      "step": 4880
+    },
+    {
+      "epoch": 4.268878219118289,
+      "grad_norm": 3.121246576309204,
+      "learning_rate": 3.1829209121785545e-05,
+      "loss": 0.1476,
+      "step": 4890
+    },
+    {
+      "epoch": 4.277608031427325,
+      "grad_norm": 2.7231833934783936,
+      "learning_rate": 3.1780688985929165e-05,
+      "loss": 0.1419,
+      "step": 4900
+    },
+    {
+      "epoch": 4.2863378437363595,
+      "grad_norm": 2.7895748615264893,
+      "learning_rate": 3.173216885007278e-05,
+      "loss": 0.161,
+      "step": 4910
+    },
+    {
+      "epoch": 4.295067656045395,
+      "grad_norm": 2.6406960487365723,
+      "learning_rate": 3.16836487142164e-05,
+      "loss": 0.1358,
+      "step": 4920
+    },
+    {
+      "epoch": 4.30379746835443,
+      "grad_norm": 2.4051973819732666,
+      "learning_rate": 3.163512857836002e-05,
+      "loss": 0.1544,
+      "step": 4930
+    },
+    {
+      "epoch": 4.312527280663466,
+      "grad_norm": 2.49039626121521,
+      "learning_rate": 3.158660844250364e-05,
+      "loss": 0.1551,
+      "step": 4940
+    },
+    {
+      "epoch": 4.321257092972501,
+      "grad_norm": 2.592521905899048,
+      "learning_rate": 3.153808830664726e-05,
+      "loss": 0.1836,
+      "step": 4950
+    },
+    {
+      "epoch": 4.329986905281537,
+      "grad_norm": 3.2178955078125,
+      "learning_rate": 3.148956817079088e-05,
+      "loss": 0.1562,
+      "step": 4960
+    },
+    {
+      "epoch": 4.338716717590572,
+      "grad_norm": 2.5670714378356934,
+      "learning_rate": 3.14410480349345e-05,
+      "loss": 0.1412,
+      "step": 4970
+    },
+    {
+      "epoch": 4.3474465298996074,
+      "grad_norm": 4.081165790557861,
+      "learning_rate": 3.1392527899078116e-05,
+      "loss": 0.166,
+      "step": 4980
+    },
+    {
+      "epoch": 4.356176342208642,
+      "grad_norm": 2.3925557136535645,
+      "learning_rate": 3.134400776322174e-05,
+      "loss": 0.1351,
+      "step": 4990
+    },
+    {
+      "epoch": 4.364906154517678,
+      "grad_norm": 2.4587152004241943,
+      "learning_rate": 3.1295487627365364e-05,
+      "loss": 0.1389,
+      "step": 5000
+    },
+    {
+      "epoch": 4.373635966826713,
+      "grad_norm": 3.3576841354370117,
+      "learning_rate": 3.124696749150898e-05,
+      "loss": 0.1431,
+      "step": 5010
+    },
+    {
+      "epoch": 4.382365779135749,
+      "grad_norm": 3.1087608337402344,
+      "learning_rate": 3.11984473556526e-05,
+      "loss": 0.157,
+      "step": 5020
+    },
+    {
+      "epoch": 4.391095591444784,
+      "grad_norm": 3.4390625953674316,
+      "learning_rate": 3.114992721979622e-05,
+      "loss": 0.1437,
+      "step": 5030
+    },
+    {
+      "epoch": 4.39982540375382,
+      "grad_norm": 2.3669025897979736,
+      "learning_rate": 3.110140708393984e-05,
+      "loss": 0.1229,
+      "step": 5040
+    },
+    {
+      "epoch": 4.4085552160628545,
+      "grad_norm": 3.4715614318847656,
+      "learning_rate": 3.105288694808345e-05,
+      "loss": 0.1436,
+      "step": 5050
+    },
+    {
+      "epoch": 4.41728502837189,
+      "grad_norm": 3.359426975250244,
+      "learning_rate": 3.1004366812227074e-05,
+      "loss": 0.1563,
+      "step": 5060
+    },
+    {
+      "epoch": 4.426014840680925,
+      "grad_norm": 3.365325450897217,
+      "learning_rate": 3.09558466763707e-05,
+      "loss": 0.1411,
+      "step": 5070
+    },
+    {
+      "epoch": 4.434744652989961,
+      "grad_norm": 2.716036319732666,
+      "learning_rate": 3.0907326540514315e-05,
+      "loss": 0.1535,
+      "step": 5080
+    },
+    {
+      "epoch": 4.443474465298996,
+      "grad_norm": 2.797657012939453,
+      "learning_rate": 3.0858806404657935e-05,
+      "loss": 0.1314,
+      "step": 5090
+    },
+    {
+      "epoch": 4.452204277608032,
+      "grad_norm": 2.7117514610290527,
+      "learning_rate": 3.0810286268801556e-05,
+      "loss": 0.1436,
+      "step": 5100
+    },
+    {
+      "epoch": 4.460934089917067,
+      "grad_norm": 2.1483347415924072,
+      "learning_rate": 3.0761766132945176e-05,
+      "loss": 0.1591,
+      "step": 5110
+    },
+    {
+      "epoch": 4.469663902226102,
+      "grad_norm": 3.0980610847473145,
+      "learning_rate": 3.071324599708879e-05,
+      "loss": 0.1443,
+      "step": 5120
+    },
+    {
+      "epoch": 4.478393714535137,
+      "grad_norm": 3.3128209114074707,
+      "learning_rate": 3.066472586123241e-05,
+      "loss": 0.1313,
+      "step": 5130
+    },
+    {
+      "epoch": 4.487123526844173,
+      "grad_norm": 3.275357484817505,
+      "learning_rate": 3.061620572537604e-05,
+      "loss": 0.1744,
+      "step": 5140
+    },
+    {
+      "epoch": 4.495853339153208,
+      "grad_norm": 3.222581148147583,
+      "learning_rate": 3.056768558951965e-05,
+      "loss": 0.1584,
+      "step": 5150
+    },
+    {
+      "epoch": 4.504583151462244,
+      "grad_norm": 3.389233350753784,
+      "learning_rate": 3.051916545366327e-05,
+      "loss": 0.1477,
+      "step": 5160
+    },
+    {
+      "epoch": 4.513312963771279,
+      "grad_norm": 3.1281654834747314,
+      "learning_rate": 3.047064531780689e-05,
+      "loss": 0.1283,
+      "step": 5170
+    },
+    {
+      "epoch": 4.522042776080315,
+      "grad_norm": 3.123300075531006,
+      "learning_rate": 3.042212518195051e-05,
+      "loss": 0.1496,
+      "step": 5180
+    },
+    {
+      "epoch": 4.5307725883893495,
+      "grad_norm": 2.7296836376190186,
+      "learning_rate": 3.0373605046094127e-05,
+      "loss": 0.133,
+      "step": 5190
+    },
+    {
+      "epoch": 4.539502400698385,
+      "grad_norm": 2.5812196731567383,
+      "learning_rate": 3.032508491023775e-05,
+      "loss": 0.1388,
+      "step": 5200
+    },
+    {
+      "epoch": 4.54823221300742,
+      "grad_norm": 3.827601671218872,
+      "learning_rate": 3.0276564774381372e-05,
+      "loss": 0.145,
+      "step": 5210
+    },
+    {
+      "epoch": 4.556962025316456,
+      "grad_norm": 2.3603134155273438,
+      "learning_rate": 3.022804463852499e-05,
+      "loss": 0.1126,
+      "step": 5220
+    },
+    {
+      "epoch": 4.565691837625491,
+      "grad_norm": 2.6947975158691406,
+      "learning_rate": 3.017952450266861e-05,
+      "loss": 0.1282,
+      "step": 5230
+    },
+    {
+      "epoch": 4.574421649934527,
+      "grad_norm": 2.6288444995880127,
+      "learning_rate": 3.0131004366812227e-05,
+      "loss": 0.1371,
+      "step": 5240
+    },
+    {
+      "epoch": 4.583151462243562,
+      "grad_norm": 3.6533212661743164,
+      "learning_rate": 3.0082484230955847e-05,
+      "loss": 0.1555,
+      "step": 5250
+    },
+    {
+      "epoch": 4.591881274552597,
+      "grad_norm": 2.927175760269165,
+      "learning_rate": 3.0033964095099464e-05,
+      "loss": 0.1396,
+      "step": 5260
+    },
+    {
+      "epoch": 4.600611086861632,
+      "grad_norm": 3.4521915912628174,
+      "learning_rate": 2.9985443959243088e-05,
+      "loss": 0.1399,
+      "step": 5270
+    },
+    {
+      "epoch": 4.609340899170668,
+      "grad_norm": 3.1614108085632324,
+      "learning_rate": 2.993692382338671e-05,
+      "loss": 0.139,
+      "step": 5280
+    },
+    {
+      "epoch": 4.618070711479703,
+      "grad_norm": 2.77795672416687,
+      "learning_rate": 2.9888403687530326e-05,
+      "loss": 0.1623,
+      "step": 5290
+    },
+    {
+      "epoch": 4.626800523788739,
+      "grad_norm": 3.930523633956909,
+      "learning_rate": 2.9839883551673947e-05,
+      "loss": 0.1534,
+      "step": 5300
+    },
+    {
+      "epoch": 4.635530336097774,
+      "grad_norm": 2.8020386695861816,
+      "learning_rate": 2.9791363415817564e-05,
+      "loss": 0.1309,
+      "step": 5310
+    },
+    {
+      "epoch": 4.64426014840681,
+      "grad_norm": 2.3614084720611572,
+      "learning_rate": 2.9742843279961184e-05,
+      "loss": 0.1422,
+      "step": 5320
+    },
+    {
+      "epoch": 4.6529899607158445,
+      "grad_norm": 3.849187135696411,
+      "learning_rate": 2.9694323144104808e-05,
+      "loss": 0.129,
+      "step": 5330
+    },
+    {
+      "epoch": 4.66171977302488,
+      "grad_norm": 3.5703964233398438,
+      "learning_rate": 2.9645803008248425e-05,
+      "loss": 0.1403,
+      "step": 5340
+    },
+    {
+      "epoch": 4.670449585333915,
+      "grad_norm": 2.742767810821533,
+      "learning_rate": 2.9597282872392046e-05,
+      "loss": 0.1234,
+      "step": 5350
+    },
+    {
+      "epoch": 4.679179397642951,
+      "grad_norm": 2.6478323936462402,
+      "learning_rate": 2.9548762736535663e-05,
+      "loss": 0.1395,
+      "step": 5360
+    },
+    {
+      "epoch": 4.687909209951986,
+      "grad_norm": 2.8150362968444824,
+      "learning_rate": 2.9500242600679284e-05,
+      "loss": 0.1378,
+      "step": 5370
+    },
+    {
+      "epoch": 4.696639022261022,
+      "grad_norm": 3.0545525550842285,
+      "learning_rate": 2.94517224648229e-05,
+      "loss": 0.1244,
+      "step": 5380
+    },
+    {
+      "epoch": 4.705368834570057,
+      "grad_norm": 3.3390815258026123,
+      "learning_rate": 2.940320232896652e-05,
+      "loss": 0.1433,
+      "step": 5390
+    },
+    {
+      "epoch": 4.714098646879092,
+      "grad_norm": 2.237645149230957,
+      "learning_rate": 2.9354682193110145e-05,
+      "loss": 0.1386,
+      "step": 5400
+    },
+    {
+      "epoch": 4.722828459188127,
+      "grad_norm": 2.8226351737976074,
+      "learning_rate": 2.9306162057253762e-05,
+      "loss": 0.1468,
+      "step": 5410
+    },
+    {
+      "epoch": 4.731558271497163,
+      "grad_norm": 2.2140517234802246,
+      "learning_rate": 2.9257641921397383e-05,
+      "loss": 0.1271,
+      "step": 5420
+    },
+    {
+      "epoch": 4.740288083806198,
+      "grad_norm": 2.853294610977173,
+      "learning_rate": 2.9209121785541e-05,
+      "loss": 0.1357,
+      "step": 5430
+    },
+    {
+      "epoch": 4.749017896115234,
+      "grad_norm": 3.399142265319824,
+      "learning_rate": 2.916060164968462e-05,
+      "loss": 0.1517,
+      "step": 5440
+    },
+    {
+      "epoch": 4.757747708424269,
+      "grad_norm": 1.853452205657959,
+      "learning_rate": 2.9112081513828238e-05,
+      "loss": 0.1323,
+      "step": 5450
+    },
+    {
+      "epoch": 4.7664775207333046,
+      "grad_norm": 2.7379791736602783,
+      "learning_rate": 2.906356137797186e-05,
+      "loss": 0.1471,
+      "step": 5460
+    },
+    {
+      "epoch": 4.7752073330423395,
+      "grad_norm": 3.7588677406311035,
+      "learning_rate": 2.9015041242115482e-05,
+      "loss": 0.1315,
+      "step": 5470
+    },
+    {
+      "epoch": 4.783937145351375,
+      "grad_norm": 2.9723212718963623,
+      "learning_rate": 2.89665211062591e-05,
+      "loss": 0.131,
+      "step": 5480
+    },
+    {
+      "epoch": 4.79266695766041,
+      "grad_norm": 2.8574113845825195,
+      "learning_rate": 2.891800097040272e-05,
+      "loss": 0.1495,
+      "step": 5490
+    },
+    {
+      "epoch": 4.801396769969446,
+      "grad_norm": 2.9763901233673096,
+      "learning_rate": 2.8869480834546337e-05,
+      "loss": 0.1388,
+      "step": 5500
+    },
+    {
+      "epoch": 4.810126582278481,
+      "grad_norm": 3.1920862197875977,
+      "learning_rate": 2.8820960698689958e-05,
+      "loss": 0.1523,
+      "step": 5510
+    },
+    {
+      "epoch": 4.818856394587517,
+      "grad_norm": 2.83996844291687,
+      "learning_rate": 2.8772440562833575e-05,
+      "loss": 0.1126,
+      "step": 5520
+    },
+    {
+      "epoch": 4.827586206896552,
+      "grad_norm": 2.066861391067505,
+      "learning_rate": 2.8723920426977195e-05,
+      "loss": 0.1337,
+      "step": 5530
+    },
+    {
+      "epoch": 4.836316019205587,
+      "grad_norm": 3.4259138107299805,
+      "learning_rate": 2.867540029112082e-05,
+      "loss": 0.1575,
+      "step": 5540
+    },
+    {
+      "epoch": 4.845045831514622,
+      "grad_norm": 3.806020975112915,
+      "learning_rate": 2.8626880155264436e-05,
+      "loss": 0.148,
+      "step": 5550
+    },
+    {
+      "epoch": 4.853775643823658,
+      "grad_norm": 4.127275466918945,
+      "learning_rate": 2.8578360019408057e-05,
+      "loss": 0.1487,
+      "step": 5560
+    },
+    {
+      "epoch": 4.862505456132693,
+      "grad_norm": 3.0096826553344727,
+      "learning_rate": 2.8529839883551674e-05,
+      "loss": 0.1627,
+      "step": 5570
+    },
+    {
+      "epoch": 4.871235268441729,
+      "grad_norm": 3.183000087738037,
+      "learning_rate": 2.8481319747695295e-05,
+      "loss": 0.1241,
+      "step": 5580
+    },
+    {
+      "epoch": 4.879965080750764,
+      "grad_norm": 2.7466609477996826,
+      "learning_rate": 2.843279961183892e-05,
+      "loss": 0.1518,
+      "step": 5590
+    },
+    {
+      "epoch": 4.8886948930597995,
+      "grad_norm": 2.0740559101104736,
+      "learning_rate": 2.8384279475982532e-05,
+      "loss": 0.143,
+      "step": 5600
+    },
+    {
+      "epoch": 4.8974247053688345,
+      "grad_norm": 2.2618136405944824,
+      "learning_rate": 2.8335759340126156e-05,
+      "loss": 0.145,
+      "step": 5610
+    },
+    {
+      "epoch": 4.90615451767787,
+      "grad_norm": 3.275090217590332,
+      "learning_rate": 2.8287239204269774e-05,
+      "loss": 0.141,
+      "step": 5620
+    },
+    {
+      "epoch": 4.914884329986905,
+      "grad_norm": 2.9142794609069824,
+      "learning_rate": 2.8238719068413394e-05,
+      "loss": 0.1263,
+      "step": 5630
+    },
+    {
+      "epoch": 4.923614142295941,
+      "grad_norm": 3.941188097000122,
+      "learning_rate": 2.819019893255701e-05,
+      "loss": 0.167,
+      "step": 5640
+    },
+    {
+      "epoch": 4.932343954604976,
+      "grad_norm": 3.0273237228393555,
+      "learning_rate": 2.8141678796700632e-05,
+      "loss": 0.1444,
+      "step": 5650
+    },
+    {
+      "epoch": 4.941073766914012,
+      "grad_norm": 1.5899831056594849,
+      "learning_rate": 2.8093158660844256e-05,
+      "loss": 0.1089,
+      "step": 5660
+    },
+    {
+      "epoch": 4.949803579223047,
+      "grad_norm": 4.153631687164307,
+      "learning_rate": 2.804463852498787e-05,
+      "loss": 0.1728,
+      "step": 5670
+    },
+    {
+      "epoch": 4.958533391532082,
+      "grad_norm": 2.7368574142456055,
+      "learning_rate": 2.7996118389131493e-05,
+      "loss": 0.1371,
+      "step": 5680
+    },
+    {
+      "epoch": 4.967263203841117,
+      "grad_norm": 3.6088647842407227,
+      "learning_rate": 2.794759825327511e-05,
+      "loss": 0.1402,
+      "step": 5690
+    },
+    {
+      "epoch": 4.975993016150153,
+      "grad_norm": 2.830106496810913,
+      "learning_rate": 2.789907811741873e-05,
+      "loss": 0.1483,
+      "step": 5700
+    },
+    {
+      "epoch": 4.984722828459188,
+      "grad_norm": 2.925632953643799,
+      "learning_rate": 2.7850557981562348e-05,
+      "loss": 0.1268,
+      "step": 5710
+    },
+    {
+      "epoch": 4.993452640768224,
+      "grad_norm": 2.303786277770996,
+      "learning_rate": 2.780203784570597e-05,
+      "loss": 0.1274,
+      "step": 5720
+    },
+    {
+      "epoch": 4.999563509384548,
+      "eval_accuracy": 0.9705990670267616,
+      "eval_loss": 0.09282852709293365,
+      "eval_runtime": 61.4449,
+      "eval_samples_per_second": 265.148,
+      "eval_steps_per_second": 8.3,
+      "step": 5727
+    },
+    {
+      "epoch": 5.002182453077259,
+      "grad_norm": 3.6829633712768555,
+      "learning_rate": 2.7753517709849593e-05,
+      "loss": 0.1331,
+      "step": 5730
+    },
+    {
+      "epoch": 5.0109122653862945,
+      "grad_norm": 2.4418623447418213,
+      "learning_rate": 2.7704997573993207e-05,
+      "loss": 0.1386,
+      "step": 5740
+    },
+    {
+      "epoch": 5.019642077695329,
+      "grad_norm": 2.420471429824829,
+      "learning_rate": 2.765647743813683e-05,
+      "loss": 0.1287,
+      "step": 5750
+    },
+    {
+      "epoch": 5.028371890004365,
+      "grad_norm": 1.8955364227294922,
+      "learning_rate": 2.7607957302280448e-05,
+      "loss": 0.1353,
+      "step": 5760
+    },
+    {
+      "epoch": 5.0371017023134,
+      "grad_norm": 3.353316307067871,
+      "learning_rate": 2.7559437166424068e-05,
+      "loss": 0.1445,
+      "step": 5770
+    },
+    {
+      "epoch": 5.045831514622436,
+      "grad_norm": 2.9570837020874023,
+      "learning_rate": 2.7510917030567685e-05,
+      "loss": 0.1374,
+      "step": 5780
+    },
+    {
+      "epoch": 5.054561326931471,
+      "grad_norm": 3.622004747390747,
+      "learning_rate": 2.7462396894711306e-05,
+      "loss": 0.1298,
+      "step": 5790
+    },
+    {
+      "epoch": 5.063291139240507,
+      "grad_norm": 2.643129587173462,
+      "learning_rate": 2.741387675885493e-05,
+      "loss": 0.146,
+      "step": 5800
+    },
+    {
+      "epoch": 5.072020951549542,
+      "grad_norm": 2.9872100353240967,
+      "learning_rate": 2.7365356622998544e-05,
+      "loss": 0.17,
+      "step": 5810
+    },
+    {
+      "epoch": 5.080750763858577,
+      "grad_norm": 3.114686965942383,
+      "learning_rate": 2.7316836487142168e-05,
+      "loss": 0.1391,
+      "step": 5820
+    },
+    {
+      "epoch": 5.089480576167612,
+      "grad_norm": 2.8469395637512207,
+      "learning_rate": 2.7268316351285785e-05,
+      "loss": 0.1388,
+      "step": 5830
+    },
+    {
+      "epoch": 5.098210388476648,
+      "grad_norm": 3.2871453762054443,
+      "learning_rate": 2.7219796215429405e-05,
+      "loss": 0.1221,
+      "step": 5840
+    },
+    {
+      "epoch": 5.106940200785683,
+      "grad_norm": 4.159573078155518,
+      "learning_rate": 2.7171276079573022e-05,
+      "loss": 0.1357,
+      "step": 5850
+    },
+    {
+      "epoch": 5.115670013094719,
+      "grad_norm": 2.4209814071655273,
+      "learning_rate": 2.7122755943716643e-05,
+      "loss": 0.1315,
+      "step": 5860
+    },
+    {
+      "epoch": 5.124399825403754,
+      "grad_norm": 3.1330792903900146,
+      "learning_rate": 2.7074235807860267e-05,
+      "loss": 0.144,
+      "step": 5870
+    },
+    {
+      "epoch": 5.1331296377127895,
+      "grad_norm": 3.4734888076782227,
+      "learning_rate": 2.702571567200388e-05,
+      "loss": 0.1255,
+      "step": 5880
+    },
+    {
+      "epoch": 5.141859450021824,
+      "grad_norm": 2.6538310050964355,
+      "learning_rate": 2.6977195536147505e-05,
+      "loss": 0.1309,
+      "step": 5890
+    },
+    {
+      "epoch": 5.15058926233086,
+      "grad_norm": 2.8028576374053955,
+      "learning_rate": 2.6928675400291122e-05,
+      "loss": 0.128,
+      "step": 5900
+    },
+    {
+      "epoch": 5.159319074639895,
+      "grad_norm": 2.747344970703125,
+      "learning_rate": 2.6880155264434742e-05,
+      "loss": 0.1287,
+      "step": 5910
+    },
+    {
+      "epoch": 5.168048886948931,
+      "grad_norm": 2.56372332572937,
+      "learning_rate": 2.6831635128578363e-05,
+      "loss": 0.141,
+      "step": 5920
+    },
+    {
+      "epoch": 5.176778699257966,
+      "grad_norm": 2.5955655574798584,
+      "learning_rate": 2.678311499272198e-05,
+      "loss": 0.1405,
+      "step": 5930
+    },
+    {
+      "epoch": 5.185508511567002,
+      "grad_norm": 2.351151943206787,
+      "learning_rate": 2.6734594856865604e-05,
+      "loss": 0.1145,
+      "step": 5940
+    },
+    {
+      "epoch": 5.194238323876037,
+      "grad_norm": 3.593594551086426,
+      "learning_rate": 2.6686074721009218e-05,
+      "loss": 0.1321,
+      "step": 5950
+    },
+    {
+      "epoch": 5.202968136185072,
+      "grad_norm": 2.935622453689575,
+      "learning_rate": 2.663755458515284e-05,
+      "loss": 0.1233,
+      "step": 5960
+    },
+    {
+      "epoch": 5.211697948494107,
+      "grad_norm": 2.3054494857788086,
+      "learning_rate": 2.658903444929646e-05,
+      "loss": 0.1238,
+      "step": 5970
+    },
+    {
+      "epoch": 5.220427760803143,
+      "grad_norm": 3.240325927734375,
+      "learning_rate": 2.654051431344008e-05,
+      "loss": 0.1613,
+      "step": 5980
+    },
+    {
+      "epoch": 5.229157573112178,
+      "grad_norm": 2.3457300662994385,
+      "learning_rate": 2.64919941775837e-05,
+      "loss": 0.1498,
+      "step": 5990
+    },
+    {
+      "epoch": 5.237887385421214,
+      "grad_norm": 3.284043550491333,
+      "learning_rate": 2.6443474041727317e-05,
+      "loss": 0.1328,
+      "step": 6000
+    },
+    {
+      "epoch": 5.246617197730249,
+      "grad_norm": 2.991929531097412,
+      "learning_rate": 2.639495390587094e-05,
+      "loss": 0.1334,
+      "step": 6010
+    },
+    {
+      "epoch": 5.2553470100392845,
+      "grad_norm": 3.4685487747192383,
+      "learning_rate": 2.6346433770014555e-05,
+      "loss": 0.1463,
+      "step": 6020
+    },
+    {
+      "epoch": 5.264076822348319,
+      "grad_norm": 3.8685977458953857,
+      "learning_rate": 2.629791363415818e-05,
+      "loss": 0.1219,
+      "step": 6030
+    },
+    {
+      "epoch": 5.272806634657355,
+      "grad_norm": 2.8441107273101807,
+      "learning_rate": 2.6249393498301796e-05,
+      "loss": 0.1243,
+      "step": 6040
+    },
+    {
+      "epoch": 5.28153644696639,
+      "grad_norm": 3.628505229949951,
+      "learning_rate": 2.6200873362445416e-05,
+      "loss": 0.1465,
+      "step": 6050
+    },
+    {
+      "epoch": 5.290266259275426,
+      "grad_norm": 3.1612138748168945,
+      "learning_rate": 2.6152353226589037e-05,
+      "loss": 0.124,
+      "step": 6060
+    },
+    {
+      "epoch": 5.298996071584461,
+      "grad_norm": 2.9345169067382812,
+      "learning_rate": 2.6103833090732654e-05,
+      "loss": 0.153,
+      "step": 6070
+    },
+    {
+      "epoch": 5.307725883893497,
+      "grad_norm": 2.2323479652404785,
+      "learning_rate": 2.6055312954876278e-05,
+      "loss": 0.1307,
+      "step": 6080
+    },
+    {
+      "epoch": 5.3164556962025316,
+      "grad_norm": 2.986766815185547,
+      "learning_rate": 2.6006792819019892e-05,
+      "loss": 0.1137,
+      "step": 6090
+    },
+    {
+      "epoch": 5.325185508511567,
+      "grad_norm": 3.07307767868042,
+      "learning_rate": 2.5958272683163516e-05,
+      "loss": 0.1372,
+      "step": 6100
+    },
+    {
+      "epoch": 5.333915320820602,
+      "grad_norm": 2.370492935180664,
+      "learning_rate": 2.5909752547307133e-05,
+      "loss": 0.1549,
+      "step": 6110
+    },
+    {
+      "epoch": 5.342645133129638,
+      "grad_norm": 1.7776038646697998,
+      "learning_rate": 2.5861232411450753e-05,
+      "loss": 0.1218,
+      "step": 6120
+    },
+    {
+      "epoch": 5.351374945438673,
+      "grad_norm": 3.1490137577056885,
+      "learning_rate": 2.5812712275594374e-05,
+      "loss": 0.1556,
+      "step": 6130
+    },
+    {
+      "epoch": 5.360104757747709,
+      "grad_norm": 2.9647440910339355,
+      "learning_rate": 2.576419213973799e-05,
+      "loss": 0.126,
+      "step": 6140
+    },
+    {
+      "epoch": 5.368834570056744,
+      "grad_norm": 3.315322160720825,
+      "learning_rate": 2.5715672003881615e-05,
+      "loss": 0.133,
+      "step": 6150
+    },
+    {
+      "epoch": 5.3775643823657795,
+      "grad_norm": 2.5604379177093506,
+      "learning_rate": 2.566715186802523e-05,
+      "loss": 0.1437,
+      "step": 6160
+    },
+    {
+      "epoch": 5.386294194674814,
+      "grad_norm": 1.5733325481414795,
+      "learning_rate": 2.5618631732168853e-05,
+      "loss": 0.1374,
+      "step": 6170
+    },
+    {
+      "epoch": 5.39502400698385,
+      "grad_norm": 2.330185890197754,
+      "learning_rate": 2.5570111596312467e-05,
+      "loss": 0.1158,
+      "step": 6180
+    },
+    {
+      "epoch": 5.403753819292885,
+      "grad_norm": 3.2574543952941895,
+      "learning_rate": 2.552159146045609e-05,
+      "loss": 0.1374,
+      "step": 6190
+    },
+    {
+      "epoch": 5.412483631601921,
+      "grad_norm": 3.3817057609558105,
+      "learning_rate": 2.547307132459971e-05,
+      "loss": 0.1272,
+      "step": 6200
+    },
+    {
+      "epoch": 5.421213443910956,
+      "grad_norm": 3.8969085216522217,
+      "learning_rate": 2.5424551188743328e-05,
+      "loss": 0.1297,
+      "step": 6210
+    },
+    {
+      "epoch": 5.429943256219992,
+      "grad_norm": 2.5738420486450195,
+      "learning_rate": 2.5376031052886952e-05,
+      "loss": 0.121,
+      "step": 6220
+    },
+    {
+      "epoch": 5.4386730685290265,
+      "grad_norm": 3.007840633392334,
+      "learning_rate": 2.5327510917030566e-05,
+      "loss": 0.1561,
+      "step": 6230
+    },
+    {
+      "epoch": 5.447402880838062,
+      "grad_norm": 2.9330294132232666,
+      "learning_rate": 2.527899078117419e-05,
+      "loss": 0.1304,
+      "step": 6240
+    },
+    {
+      "epoch": 5.456132693147097,
+      "grad_norm": 2.3257663249969482,
+      "learning_rate": 2.523047064531781e-05,
+      "loss": 0.1273,
+      "step": 6250
+    },
+    {
+      "epoch": 5.464862505456133,
+      "grad_norm": 2.6171205043792725,
+      "learning_rate": 2.5181950509461428e-05,
+      "loss": 0.1135,
+      "step": 6260
+    },
+    {
+      "epoch": 5.473592317765168,
+      "grad_norm": 3.440798759460449,
+      "learning_rate": 2.5133430373605048e-05,
+      "loss": 0.1246,
+      "step": 6270
+    },
+    {
+      "epoch": 5.482322130074204,
+      "grad_norm": 3.418937921524048,
+      "learning_rate": 2.5084910237748665e-05,
+      "loss": 0.1354,
+      "step": 6280
+    },
+    {
+      "epoch": 5.491051942383239,
+      "grad_norm": 2.9420177936553955,
+      "learning_rate": 2.503639010189229e-05,
+      "loss": 0.1405,
+      "step": 6290
+    },
+    {
+      "epoch": 5.4997817546922745,
+      "grad_norm": 3.042564630508423,
+      "learning_rate": 2.4987869966035906e-05,
+      "loss": 0.1357,
+      "step": 6300
+    },
+    {
+      "epoch": 5.508511567001309,
+      "grad_norm": 2.1227240562438965,
+      "learning_rate": 2.4939349830179527e-05,
+      "loss": 0.1417,
+      "step": 6310
+    },
+    {
+      "epoch": 5.517241379310345,
+      "grad_norm": 3.1265482902526855,
+      "learning_rate": 2.4890829694323144e-05,
+      "loss": 0.1382,
+      "step": 6320
+    },
+    {
+      "epoch": 5.52597119161938,
+      "grad_norm": 2.782801389694214,
+      "learning_rate": 2.4842309558466765e-05,
+      "loss": 0.1635,
+      "step": 6330
+    },
+    {
+      "epoch": 5.534701003928416,
+      "grad_norm": 3.488227605819702,
+      "learning_rate": 2.4793789422610382e-05,
+      "loss": 0.1303,
+      "step": 6340
+    },
+    {
+      "epoch": 5.543430816237451,
+      "grad_norm": 2.8877530097961426,
+      "learning_rate": 2.4745269286754006e-05,
+      "loss": 0.1676,
+      "step": 6350
+    },
+    {
+      "epoch": 5.552160628546487,
+      "grad_norm": 2.2188923358917236,
+      "learning_rate": 2.4696749150897623e-05,
+      "loss": 0.1118,
+      "step": 6360
+    },
+    {
+      "epoch": 5.5608904408555215,
+      "grad_norm": 2.9819607734680176,
+      "learning_rate": 2.4648229015041243e-05,
+      "loss": 0.1603,
+      "step": 6370
+    },
+    {
+      "epoch": 5.569620253164557,
+      "grad_norm": 3.2080116271972656,
+      "learning_rate": 2.4599708879184864e-05,
+      "loss": 0.1304,
+      "step": 6380
+    },
+    {
+      "epoch": 5.578350065473592,
+      "grad_norm": 1.6882622241973877,
+      "learning_rate": 2.455118874332848e-05,
+      "loss": 0.1171,
+      "step": 6390
+    },
+    {
+      "epoch": 5.587079877782628,
+      "grad_norm": 2.5788047313690186,
+      "learning_rate": 2.45026686074721e-05,
+      "loss": 0.1193,
+      "step": 6400
+    },
+    {
+      "epoch": 5.595809690091663,
+      "grad_norm": 3.859628915786743,
+      "learning_rate": 2.445414847161572e-05,
+      "loss": 0.124,
+      "step": 6410
+    },
+    {
+      "epoch": 5.604539502400699,
+      "grad_norm": 1.5977929830551147,
+      "learning_rate": 2.4405628335759343e-05,
+      "loss": 0.1194,
+      "step": 6420
+    },
+    {
+      "epoch": 5.613269314709734,
+      "grad_norm": 3.073011875152588,
+      "learning_rate": 2.435710819990296e-05,
+      "loss": 0.1344,
+      "step": 6430
+    },
+    {
+      "epoch": 5.6219991270187695,
+      "grad_norm": 2.9999372959136963,
+      "learning_rate": 2.430858806404658e-05,
+      "loss": 0.1487,
+      "step": 6440
+    },
+    {
+      "epoch": 5.630728939327804,
+      "grad_norm": 2.077570676803589,
+      "learning_rate": 2.42600679281902e-05,
+      "loss": 0.118,
+      "step": 6450
+    },
+    {
+      "epoch": 5.63945875163684,
+      "grad_norm": 3.1440892219543457,
+      "learning_rate": 2.4211547792333818e-05,
+      "loss": 0.1394,
+      "step": 6460
+    },
+    {
+      "epoch": 5.648188563945875,
+      "grad_norm": 2.836007833480835,
+      "learning_rate": 2.416302765647744e-05,
+      "loss": 0.1308,
+      "step": 6470
+    },
+    {
+      "epoch": 5.656918376254911,
+      "grad_norm": 2.675652027130127,
+      "learning_rate": 2.411450752062106e-05,
+      "loss": 0.1208,
+      "step": 6480
+    },
+    {
+      "epoch": 5.665648188563946,
+      "grad_norm": 2.833317756652832,
+      "learning_rate": 2.406598738476468e-05,
+      "loss": 0.1244,
+      "step": 6490
+    },
+    {
+      "epoch": 5.674378000872982,
+      "grad_norm": 2.444568157196045,
+      "learning_rate": 2.4017467248908297e-05,
+      "loss": 0.1153,
+      "step": 6500
+    },
+    {
+      "epoch": 5.6831078131820165,
+      "grad_norm": 3.581678628921509,
+      "learning_rate": 2.3968947113051917e-05,
+      "loss": 0.1431,
+      "step": 6510
+    },
+    {
+      "epoch": 5.691837625491052,
+      "grad_norm": 1.7067352533340454,
+      "learning_rate": 2.3920426977195538e-05,
+      "loss": 0.1322,
+      "step": 6520
+    },
+    {
+      "epoch": 5.700567437800087,
+      "grad_norm": 4.281068325042725,
+      "learning_rate": 2.3871906841339155e-05,
+      "loss": 0.1349,
+      "step": 6530
+    },
+    {
+      "epoch": 5.709297250109123,
+      "grad_norm": 2.958136558532715,
+      "learning_rate": 2.3823386705482776e-05,
+      "loss": 0.1319,
+      "step": 6540
+    },
+    {
+      "epoch": 5.718027062418158,
+      "grad_norm": 2.9312613010406494,
+      "learning_rate": 2.3774866569626396e-05,
+      "loss": 0.103,
+      "step": 6550
+    },
+    {
+      "epoch": 5.726756874727194,
+      "grad_norm": 2.7693583965301514,
+      "learning_rate": 2.3726346433770017e-05,
+      "loss": 0.1283,
+      "step": 6560
+    },
+    {
+      "epoch": 5.735486687036229,
+      "grad_norm": 2.7267799377441406,
+      "learning_rate": 2.3677826297913634e-05,
+      "loss": 0.1131,
+      "step": 6570
+    },
+    {
+      "epoch": 5.7442164993452645,
+      "grad_norm": 2.184208631515503,
+      "learning_rate": 2.3629306162057255e-05,
+      "loss": 0.1289,
+      "step": 6580
+    },
+    {
+      "epoch": 5.752946311654299,
+      "grad_norm": 2.8971517086029053,
+      "learning_rate": 2.3580786026200875e-05,
+      "loss": 0.1185,
+      "step": 6590
+    },
+    {
+      "epoch": 5.761676123963335,
+      "grad_norm": 3.3640193939208984,
+      "learning_rate": 2.3532265890344492e-05,
+      "loss": 0.139,
+      "step": 6600
+    },
+    {
+      "epoch": 5.77040593627237,
+      "grad_norm": 3.1321167945861816,
+      "learning_rate": 2.3483745754488113e-05,
+      "loss": 0.1594,
+      "step": 6610
+    },
+    {
+      "epoch": 5.779135748581406,
+      "grad_norm": 2.4703168869018555,
+      "learning_rate": 2.3435225618631733e-05,
+      "loss": 0.1528,
+      "step": 6620
+    },
+    {
+      "epoch": 5.787865560890441,
+      "grad_norm": 2.4260854721069336,
+      "learning_rate": 2.3386705482775354e-05,
+      "loss": 0.1272,
+      "step": 6630
+    },
+    {
+      "epoch": 5.796595373199477,
+      "grad_norm": 2.809253692626953,
+      "learning_rate": 2.333818534691897e-05,
+      "loss": 0.1421,
+      "step": 6640
+    },
+    {
+      "epoch": 5.8053251855085115,
+      "grad_norm": 2.53581166267395,
+      "learning_rate": 2.328966521106259e-05,
+      "loss": 0.1395,
+      "step": 6650
+    },
+    {
+      "epoch": 5.814054997817547,
+      "grad_norm": 2.912879467010498,
+      "learning_rate": 2.3241145075206212e-05,
+      "loss": 0.1269,
+      "step": 6660
+    },
+    {
+      "epoch": 5.822784810126582,
+      "grad_norm": 2.0264980792999268,
+      "learning_rate": 2.319262493934983e-05,
+      "loss": 0.1335,
+      "step": 6670
+    },
+    {
+      "epoch": 5.831514622435618,
+      "grad_norm": 2.8756725788116455,
+      "learning_rate": 2.3144104803493453e-05,
+      "loss": 0.1246,
+      "step": 6680
+    },
+    {
+      "epoch": 5.840244434744653,
+      "grad_norm": 2.49039363861084,
+      "learning_rate": 2.309558466763707e-05,
+      "loss": 0.1292,
+      "step": 6690
+    },
+    {
+      "epoch": 5.848974247053688,
+      "grad_norm": 2.6733226776123047,
+      "learning_rate": 2.304706453178069e-05,
+      "loss": 0.1064,
+      "step": 6700
+    },
+    {
+      "epoch": 5.857704059362724,
+      "grad_norm": 2.1536784172058105,
+      "learning_rate": 2.2998544395924308e-05,
+      "loss": 0.138,
+      "step": 6710
+    },
+    {
+      "epoch": 5.8664338716717594,
+      "grad_norm": 3.428746461868286,
+      "learning_rate": 2.295002426006793e-05,
+      "loss": 0.1362,
+      "step": 6720
+    },
+    {
+      "epoch": 5.875163683980794,
+      "grad_norm": 3.4653851985931396,
+      "learning_rate": 2.290150412421155e-05,
+      "loss": 0.1235,
+      "step": 6730
+    },
+    {
+      "epoch": 5.883893496289829,
+      "grad_norm": 2.283843755722046,
+      "learning_rate": 2.2852983988355166e-05,
+      "loss": 0.1081,
+      "step": 6740
+    },
+    {
+      "epoch": 5.892623308598865,
+      "grad_norm": 3.812995433807373,
+      "learning_rate": 2.280446385249879e-05,
+      "loss": 0.1305,
+      "step": 6750
+    },
+    {
+      "epoch": 5.901353120907901,
+      "grad_norm": 2.6548011302948,
+      "learning_rate": 2.2755943716642407e-05,
+      "loss": 0.135,
+      "step": 6760
+    },
+    {
+      "epoch": 5.910082933216936,
+      "grad_norm": 2.09759783744812,
+      "learning_rate": 2.2707423580786028e-05,
+      "loss": 0.1066,
+      "step": 6770
+    },
+    {
+      "epoch": 5.918812745525971,
+      "grad_norm": 2.5961296558380127,
+      "learning_rate": 2.2658903444929645e-05,
+      "loss": 0.1312,
+      "step": 6780
+    },
+    {
+      "epoch": 5.9275425578350065,
+      "grad_norm": 2.6198313236236572,
+      "learning_rate": 2.2610383309073266e-05,
+      "loss": 0.1213,
+      "step": 6790
+    },
+    {
+      "epoch": 5.936272370144042,
+      "grad_norm": 2.5672667026519775,
+      "learning_rate": 2.2561863173216886e-05,
+      "loss": 0.1097,
+      "step": 6800
+    },
+    {
+      "epoch": 5.945002182453077,
+      "grad_norm": 1.9707518815994263,
+      "learning_rate": 2.2513343037360507e-05,
+      "loss": 0.1015,
+      "step": 6810
+    },
+    {
+      "epoch": 5.953731994762112,
+      "grad_norm": 2.9768049716949463,
+      "learning_rate": 2.2464822901504127e-05,
+      "loss": 0.1246,
+      "step": 6820
+    },
+    {
+      "epoch": 5.962461807071148,
+      "grad_norm": 3.3005049228668213,
+      "learning_rate": 2.2416302765647744e-05,
+      "loss": 0.1286,
+      "step": 6830
+    },
+    {
+      "epoch": 5.971191619380184,
+      "grad_norm": 4.172717094421387,
+      "learning_rate": 2.2367782629791365e-05,
+      "loss": 0.1346,
+      "step": 6840
+    },
+    {
+      "epoch": 5.979921431689219,
+      "grad_norm": 3.4869165420532227,
+      "learning_rate": 2.2319262493934982e-05,
+      "loss": 0.1205,
+      "step": 6850
+    },
+    {
+      "epoch": 5.9886512439982535,
+      "grad_norm": 3.327515125274658,
+      "learning_rate": 2.2270742358078603e-05,
+      "loss": 0.1267,
+      "step": 6860
+    },
+    {
+      "epoch": 5.997381056307289,
+      "grad_norm": 3.0895607471466064,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 0.1141,
+      "step": 6870
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.9722563221212865,
+      "eval_loss": 0.08740793168544769,
+      "eval_runtime": 61.534,
+      "eval_samples_per_second": 264.764,
+      "eval_steps_per_second": 8.288,
+      "step": 6873
+    },
+    {
+      "epoch": 6.006110868616325,
+      "grad_norm": 2.24489688873291,
+      "learning_rate": 2.2173702086365844e-05,
+      "loss": 0.1132,
+      "step": 6880
+    },
+    {
+      "epoch": 6.01484068092536,
+      "grad_norm": 3.8057761192321777,
+      "learning_rate": 2.2125181950509464e-05,
+      "loss": 0.123,
+      "step": 6890
+    },
+    {
+      "epoch": 6.023570493234396,
+      "grad_norm": 2.2050459384918213,
+      "learning_rate": 2.207666181465308e-05,
+      "loss": 0.1365,
+      "step": 6900
+    },
+    {
+      "epoch": 6.032300305543431,
+      "grad_norm": 3.0642194747924805,
+      "learning_rate": 2.2028141678796702e-05,
+      "loss": 0.1423,
+      "step": 6910
+    },
+    {
+      "epoch": 6.041030117852467,
+      "grad_norm": 2.7323355674743652,
+      "learning_rate": 2.197962154294032e-05,
+      "loss": 0.1451,
+      "step": 6920
+    },
+    {
+      "epoch": 6.0497599301615015,
+      "grad_norm": 3.7342848777770996,
+      "learning_rate": 2.193110140708394e-05,
+      "loss": 0.1157,
+      "step": 6930
+    },
+    {
+      "epoch": 6.058489742470537,
+      "grad_norm": 1.9272093772888184,
+      "learning_rate": 2.188258127122756e-05,
+      "loss": 0.1019,
+      "step": 6940
+    },
+    {
+      "epoch": 6.067219554779572,
+      "grad_norm": 2.0700807571411133,
+      "learning_rate": 2.183406113537118e-05,
+      "loss": 0.1223,
+      "step": 6950
+    },
+    {
+      "epoch": 6.075949367088608,
+      "grad_norm": 2.6618826389312744,
+      "learning_rate": 2.17855409995148e-05,
+      "loss": 0.1181,
+      "step": 6960
+    },
+    {
+      "epoch": 6.084679179397643,
+      "grad_norm": 4.103969573974609,
+      "learning_rate": 2.173702086365842e-05,
+      "loss": 0.1357,
+      "step": 6970
+    },
+    {
+      "epoch": 6.093408991706679,
+      "grad_norm": 2.845344066619873,
+      "learning_rate": 2.168850072780204e-05,
+      "loss": 0.1183,
+      "step": 6980
+    },
+    {
+      "epoch": 6.102138804015714,
+      "grad_norm": 2.7917563915252686,
+      "learning_rate": 2.1639980591945656e-05,
+      "loss": 0.1287,
+      "step": 6990
+    },
+    {
+      "epoch": 6.110868616324749,
+      "grad_norm": 2.6179635524749756,
+      "learning_rate": 2.1591460456089277e-05,
+      "loss": 0.1346,
+      "step": 7000
+    },
+    {
+      "epoch": 6.119598428633784,
+      "grad_norm": 1.979193091392517,
+      "learning_rate": 2.1542940320232897e-05,
+      "loss": 0.125,
+      "step": 7010
+    },
+    {
+      "epoch": 6.12832824094282,
+      "grad_norm": 2.7642877101898193,
+      "learning_rate": 2.1494420184376518e-05,
+      "loss": 0.1275,
+      "step": 7020
+    },
+    {
+      "epoch": 6.137058053251855,
+      "grad_norm": 2.8622918128967285,
+      "learning_rate": 2.144590004852014e-05,
+      "loss": 0.1109,
+      "step": 7030
+    },
+    {
+      "epoch": 6.145787865560891,
+      "grad_norm": 2.6740987300872803,
+      "learning_rate": 2.1397379912663756e-05,
+      "loss": 0.1144,
+      "step": 7040
+    },
+    {
+      "epoch": 6.154517677869926,
+      "grad_norm": 2.9361984729766846,
+      "learning_rate": 2.1348859776807376e-05,
+      "loss": 0.1058,
+      "step": 7050
+    },
+    {
+      "epoch": 6.163247490178962,
+      "grad_norm": 2.7780208587646484,
+      "learning_rate": 2.1300339640950993e-05,
+      "loss": 0.1391,
+      "step": 7060
+    },
+    {
+      "epoch": 6.1719773024879965,
+      "grad_norm": 3.0886313915252686,
+      "learning_rate": 2.1251819505094614e-05,
+      "loss": 0.134,
+      "step": 7070
+    },
+    {
+      "epoch": 6.180707114797032,
+      "grad_norm": 2.369558334350586,
+      "learning_rate": 2.1203299369238234e-05,
+      "loss": 0.1589,
+      "step": 7080
+    },
+    {
+      "epoch": 6.189436927106067,
+      "grad_norm": 2.826566219329834,
+      "learning_rate": 2.1154779233381855e-05,
+      "loss": 0.1125,
+      "step": 7090
+    },
+    {
+      "epoch": 6.198166739415103,
+      "grad_norm": 3.705179452896118,
+      "learning_rate": 2.1106259097525476e-05,
+      "loss": 0.1422,
+      "step": 7100
+    },
+    {
+      "epoch": 6.206896551724138,
+      "grad_norm": 2.0699045658111572,
+      "learning_rate": 2.1057738961669093e-05,
+      "loss": 0.1174,
+      "step": 7110
+    },
+    {
+      "epoch": 6.215626364033174,
+      "grad_norm": 1.7472636699676514,
+      "learning_rate": 2.1009218825812713e-05,
+      "loss": 0.1,
+      "step": 7120
+    },
+    {
+      "epoch": 6.224356176342209,
+      "grad_norm": 3.3158297538757324,
+      "learning_rate": 2.096069868995633e-05,
+      "loss": 0.1296,
+      "step": 7130
+    },
+    {
+      "epoch": 6.233085988651244,
+      "grad_norm": 3.7028303146362305,
+      "learning_rate": 2.0912178554099954e-05,
+      "loss": 0.1189,
+      "step": 7140
+    },
+    {
+      "epoch": 6.241815800960279,
+      "grad_norm": 3.309446334838867,
+      "learning_rate": 2.086365841824357e-05,
+      "loss": 0.1197,
+      "step": 7150
+    },
+    {
+      "epoch": 6.250545613269315,
+      "grad_norm": 2.997817277908325,
+      "learning_rate": 2.0815138282387192e-05,
+      "loss": 0.095,
+      "step": 7160
+    },
+    {
+      "epoch": 6.25927542557835,
+      "grad_norm": 3.9455857276916504,
+      "learning_rate": 2.0766618146530813e-05,
+      "loss": 0.1141,
+      "step": 7170
+    },
+    {
+      "epoch": 6.268005237887385,
+      "grad_norm": 2.8850817680358887,
+      "learning_rate": 2.071809801067443e-05,
+      "loss": 0.1241,
+      "step": 7180
+    },
+    {
+      "epoch": 6.276735050196421,
+      "grad_norm": 3.4822731018066406,
+      "learning_rate": 2.066957787481805e-05,
+      "loss": 0.1447,
+      "step": 7190
+    },
+    {
+      "epoch": 6.2854648625054566,
+      "grad_norm": 1.834179401397705,
+      "learning_rate": 2.0621057738961667e-05,
+      "loss": 0.113,
+      "step": 7200
+    },
+    {
+      "epoch": 6.2941946748144915,
+      "grad_norm": 2.8471169471740723,
+      "learning_rate": 2.057253760310529e-05,
+      "loss": 0.1095,
+      "step": 7210
+    },
+    {
+      "epoch": 6.302924487123526,
+      "grad_norm": 2.841275930404663,
+      "learning_rate": 2.052401746724891e-05,
+      "loss": 0.1174,
+      "step": 7220
+    },
+    {
+      "epoch": 6.311654299432562,
+      "grad_norm": 3.3309431076049805,
+      "learning_rate": 2.047549733139253e-05,
+      "loss": 0.1182,
+      "step": 7230
+    },
+    {
+      "epoch": 6.320384111741598,
+      "grad_norm": 2.214808225631714,
+      "learning_rate": 2.042697719553615e-05,
+      "loss": 0.1345,
+      "step": 7240
+    },
+    {
+      "epoch": 6.329113924050633,
+      "grad_norm": 2.0826220512390137,
+      "learning_rate": 2.0378457059679767e-05,
+      "loss": 0.1114,
+      "step": 7250
+    },
+    {
+      "epoch": 6.337843736359668,
+      "grad_norm": 2.8715829849243164,
+      "learning_rate": 2.0329936923823387e-05,
+      "loss": 0.1335,
+      "step": 7260
+    },
+    {
+      "epoch": 6.346573548668704,
+      "grad_norm": 2.2511582374572754,
+      "learning_rate": 2.0281416787967008e-05,
+      "loss": 0.1558,
+      "step": 7270
+    },
+    {
+      "epoch": 6.355303360977739,
+      "grad_norm": 3.4435458183288574,
+      "learning_rate": 2.023289665211063e-05,
+      "loss": 0.1433,
+      "step": 7280
+    },
+    {
+      "epoch": 6.364033173286774,
+      "grad_norm": 2.411850929260254,
+      "learning_rate": 2.0184376516254246e-05,
+      "loss": 0.1224,
+      "step": 7290
+    },
+    {
+      "epoch": 6.372762985595809,
+      "grad_norm": 2.7871601581573486,
+      "learning_rate": 2.0135856380397866e-05,
+      "loss": 0.1206,
+      "step": 7300
+    },
+    {
+      "epoch": 6.381492797904845,
+      "grad_norm": 2.9544012546539307,
+      "learning_rate": 2.0087336244541487e-05,
+      "loss": 0.133,
+      "step": 7310
+    },
+    {
+      "epoch": 6.390222610213881,
+      "grad_norm": 3.283802032470703,
+      "learning_rate": 2.0038816108685104e-05,
+      "loss": 0.1259,
+      "step": 7320
+    },
+    {
+      "epoch": 6.398952422522916,
+      "grad_norm": 2.1609585285186768,
+      "learning_rate": 1.9990295972828724e-05,
+      "loss": 0.1499,
+      "step": 7330
+    },
+    {
+      "epoch": 6.407682234831951,
+      "grad_norm": 2.2527894973754883,
+      "learning_rate": 1.9941775836972345e-05,
+      "loss": 0.1206,
+      "step": 7340
+    },
+    {
+      "epoch": 6.4164120471409865,
+      "grad_norm": 3.909329891204834,
+      "learning_rate": 1.9893255701115965e-05,
+      "loss": 0.099,
+      "step": 7350
+    },
+    {
+      "epoch": 6.425141859450022,
+      "grad_norm": 2.997509002685547,
+      "learning_rate": 1.9844735565259583e-05,
+      "loss": 0.1399,
+      "step": 7360
+    },
+    {
+      "epoch": 6.433871671759057,
+      "grad_norm": 2.639704465866089,
+      "learning_rate": 1.9796215429403203e-05,
+      "loss": 0.1224,
+      "step": 7370
+    },
+    {
+      "epoch": 6.442601484068092,
+      "grad_norm": 3.1105806827545166,
+      "learning_rate": 1.9747695293546824e-05,
+      "loss": 0.1106,
+      "step": 7380
+    },
+    {
+      "epoch": 6.451331296377128,
+      "grad_norm": 2.972954273223877,
+      "learning_rate": 1.969917515769044e-05,
+      "loss": 0.1216,
+      "step": 7390
+    },
+    {
+      "epoch": 6.460061108686164,
+      "grad_norm": 3.996022939682007,
+      "learning_rate": 1.965065502183406e-05,
+      "loss": 0.1279,
+      "step": 7400
+    },
+    {
+      "epoch": 6.468790920995199,
+      "grad_norm": 3.0898971557617188,
+      "learning_rate": 1.9602134885977682e-05,
+      "loss": 0.1247,
+      "step": 7410
+    },
+    {
+      "epoch": 6.4775207333042335,
+      "grad_norm": 2.5472044944763184,
+      "learning_rate": 1.9553614750121303e-05,
+      "loss": 0.1411,
+      "step": 7420
+    },
+    {
+      "epoch": 6.486250545613269,
+      "grad_norm": 3.311650514602661,
+      "learning_rate": 1.950509461426492e-05,
+      "loss": 0.1053,
+      "step": 7430
+    },
+    {
+      "epoch": 6.494980357922305,
+      "grad_norm": 2.9363017082214355,
+      "learning_rate": 1.945657447840854e-05,
+      "loss": 0.1191,
+      "step": 7440
+    },
+    {
+      "epoch": 6.50371017023134,
+      "grad_norm": 3.120633602142334,
+      "learning_rate": 1.940805434255216e-05,
+      "loss": 0.1202,
+      "step": 7450
+    },
+    {
+      "epoch": 6.512439982540375,
+      "grad_norm": 3.49874210357666,
+      "learning_rate": 1.9359534206695778e-05,
+      "loss": 0.1128,
+      "step": 7460
+    },
+    {
+      "epoch": 6.521169794849411,
+      "grad_norm": 2.98953914642334,
+      "learning_rate": 1.9311014070839402e-05,
+      "loss": 0.1191,
+      "step": 7470
+    },
+    {
+      "epoch": 6.5298996071584465,
+      "grad_norm": 2.440988779067993,
+      "learning_rate": 1.926249393498302e-05,
+      "loss": 0.12,
+      "step": 7480
+    },
+    {
+      "epoch": 6.538629419467481,
+      "grad_norm": 2.9546585083007812,
+      "learning_rate": 1.921397379912664e-05,
+      "loss": 0.143,
+      "step": 7490
+    },
+    {
+      "epoch": 6.547359231776516,
+      "grad_norm": 1.7305012941360474,
+      "learning_rate": 1.9165453663270257e-05,
+      "loss": 0.1083,
+      "step": 7500
+    },
+    {
+      "epoch": 6.556089044085552,
+      "grad_norm": 2.084421396255493,
+      "learning_rate": 1.9116933527413877e-05,
+      "loss": 0.1228,
+      "step": 7510
+    },
+    {
+      "epoch": 6.564818856394588,
+      "grad_norm": 3.237086057662964,
+      "learning_rate": 1.9068413391557498e-05,
+      "loss": 0.1265,
+      "step": 7520
+    },
+    {
+      "epoch": 6.573548668703623,
+      "grad_norm": 2.24908185005188,
+      "learning_rate": 1.9019893255701115e-05,
+      "loss": 0.0891,
+      "step": 7530
+    },
+    {
+      "epoch": 6.582278481012658,
+      "grad_norm": 3.3673548698425293,
+      "learning_rate": 1.897137311984474e-05,
+      "loss": 0.1377,
+      "step": 7540
+    },
+    {
+      "epoch": 6.591008293321694,
+      "grad_norm": 2.560291051864624,
+      "learning_rate": 1.8922852983988356e-05,
+      "loss": 0.1114,
+      "step": 7550
+    },
+    {
+      "epoch": 6.599738105630729,
+      "grad_norm": 1.9948253631591797,
+      "learning_rate": 1.8874332848131977e-05,
+      "loss": 0.1261,
+      "step": 7560
+    },
+    {
+      "epoch": 6.608467917939764,
+      "grad_norm": 3.7410624027252197,
+      "learning_rate": 1.8825812712275594e-05,
+      "loss": 0.1111,
+      "step": 7570
+    },
+    {
+      "epoch": 6.617197730248799,
+      "grad_norm": 3.520691156387329,
+      "learning_rate": 1.8777292576419214e-05,
+      "loss": 0.1396,
+      "step": 7580
+    },
+    {
+      "epoch": 6.625927542557835,
+      "grad_norm": 2.289947986602783,
+      "learning_rate": 1.8728772440562835e-05,
+      "loss": 0.1211,
+      "step": 7590
+    },
+    {
+      "epoch": 6.634657354866871,
+      "grad_norm": 2.078162431716919,
+      "learning_rate": 1.8680252304706455e-05,
+      "loss": 0.1258,
+      "step": 7600
+    },
+    {
+      "epoch": 6.643387167175906,
+      "grad_norm": 2.3327958583831787,
+      "learning_rate": 1.8631732168850076e-05,
+      "loss": 0.1177,
+      "step": 7610
+    },
+    {
+      "epoch": 6.652116979484941,
+      "grad_norm": 2.689931631088257,
+      "learning_rate": 1.8583212032993693e-05,
+      "loss": 0.1144,
+      "step": 7620
+    },
+    {
+      "epoch": 6.660846791793976,
+      "grad_norm": 2.395716428756714,
+      "learning_rate": 1.8534691897137314e-05,
+      "loss": 0.1202,
+      "step": 7630
+    },
+    {
+      "epoch": 6.669576604103012,
+      "grad_norm": 2.892514705657959,
+      "learning_rate": 1.848617176128093e-05,
+      "loss": 0.1227,
+      "step": 7640
+    },
+    {
+      "epoch": 6.678306416412047,
+      "grad_norm": 2.741924285888672,
+      "learning_rate": 1.843765162542455e-05,
+      "loss": 0.0994,
+      "step": 7650
+    },
+    {
+      "epoch": 6.687036228721082,
+      "grad_norm": 3.1253838539123535,
+      "learning_rate": 1.8389131489568172e-05,
+      "loss": 0.1212,
+      "step": 7660
+    },
+    {
+      "epoch": 6.695766041030118,
+      "grad_norm": 3.520061492919922,
+      "learning_rate": 1.8340611353711792e-05,
+      "loss": 0.1113,
+      "step": 7670
+    },
+    {
+      "epoch": 6.704495853339154,
+      "grad_norm": 1.9933674335479736,
+      "learning_rate": 1.8292091217855413e-05,
+      "loss": 0.0858,
+      "step": 7680
+    },
+    {
+      "epoch": 6.713225665648189,
+      "grad_norm": 2.0083909034729004,
+      "learning_rate": 1.824357108199903e-05,
+      "loss": 0.1199,
+      "step": 7690
+    },
+    {
+      "epoch": 6.7219554779572235,
+      "grad_norm": 3.1717681884765625,
+      "learning_rate": 1.819505094614265e-05,
+      "loss": 0.1207,
+      "step": 7700
+    },
+    {
+      "epoch": 6.730685290266259,
+      "grad_norm": 1.9426745176315308,
+      "learning_rate": 1.8146530810286268e-05,
+      "loss": 0.1157,
+      "step": 7710
+    },
+    {
+      "epoch": 6.739415102575295,
+      "grad_norm": 2.9203710556030273,
+      "learning_rate": 1.809801067442989e-05,
+      "loss": 0.1251,
+      "step": 7720
+    },
+    {
+      "epoch": 6.74814491488433,
+      "grad_norm": 3.200591564178467,
+      "learning_rate": 1.804949053857351e-05,
+      "loss": 0.1481,
+      "step": 7730
+    },
+    {
+      "epoch": 6.756874727193365,
+      "grad_norm": 4.032431125640869,
+      "learning_rate": 1.800097040271713e-05,
+      "loss": 0.135,
+      "step": 7740
+    },
+    {
+      "epoch": 6.765604539502401,
+      "grad_norm": 2.4872477054595947,
+      "learning_rate": 1.795245026686075e-05,
+      "loss": 0.0975,
+      "step": 7750
+    },
+    {
+      "epoch": 6.7743343518114365,
+      "grad_norm": 2.5336356163024902,
+      "learning_rate": 1.7903930131004367e-05,
+      "loss": 0.1311,
+      "step": 7760
+    },
+    {
+      "epoch": 6.783064164120471,
+      "grad_norm": 2.6946463584899902,
+      "learning_rate": 1.7855409995147988e-05,
+      "loss": 0.1073,
+      "step": 7770
+    },
+    {
+      "epoch": 6.791793976429506,
+      "grad_norm": 3.247837543487549,
+      "learning_rate": 1.7806889859291605e-05,
+      "loss": 0.1126,
+      "step": 7780
+    },
+    {
+      "epoch": 6.800523788738542,
+      "grad_norm": 2.301456928253174,
+      "learning_rate": 1.7758369723435225e-05,
+      "loss": 0.1314,
+      "step": 7790
+    },
+    {
+      "epoch": 6.809253601047578,
+      "grad_norm": 1.9753727912902832,
+      "learning_rate": 1.7709849587578846e-05,
+      "loss": 0.1,
+      "step": 7800
+    },
+    {
+      "epoch": 6.817983413356613,
+      "grad_norm": 2.194286584854126,
+      "learning_rate": 1.7661329451722467e-05,
+      "loss": 0.1114,
+      "step": 7810
+    },
+    {
+      "epoch": 6.826713225665648,
+      "grad_norm": 2.973609447479248,
+      "learning_rate": 1.7612809315866087e-05,
+      "loss": 0.1065,
+      "step": 7820
+    },
+    {
+      "epoch": 6.8354430379746836,
+      "grad_norm": 2.5955142974853516,
+      "learning_rate": 1.7564289180009704e-05,
+      "loss": 0.1326,
+      "step": 7830
+    },
+    {
+      "epoch": 6.844172850283719,
+      "grad_norm": 2.9520103931427,
+      "learning_rate": 1.7515769044153325e-05,
+      "loss": 0.1285,
+      "step": 7840
+    },
+    {
+      "epoch": 6.852902662592754,
+      "grad_norm": 3.955249071121216,
+      "learning_rate": 1.7467248908296942e-05,
+      "loss": 0.1029,
+      "step": 7850
+    },
+    {
+      "epoch": 6.861632474901789,
+      "grad_norm": 2.9063422679901123,
+      "learning_rate": 1.7418728772440563e-05,
+      "loss": 0.0971,
+      "step": 7860
+    },
+    {
+      "epoch": 6.870362287210825,
+      "grad_norm": 2.6652753353118896,
+      "learning_rate": 1.7370208636584183e-05,
+      "loss": 0.1306,
+      "step": 7870
+    },
+    {
+      "epoch": 6.879092099519861,
+      "grad_norm": 2.687307834625244,
+      "learning_rate": 1.7321688500727804e-05,
+      "loss": 0.1265,
+      "step": 7880
+    },
+    {
+      "epoch": 6.887821911828896,
+      "grad_norm": 2.639251708984375,
+      "learning_rate": 1.7273168364871424e-05,
+      "loss": 0.1162,
+      "step": 7890
+    },
+    {
+      "epoch": 6.896551724137931,
+      "grad_norm": 3.628904104232788,
+      "learning_rate": 1.722464822901504e-05,
+      "loss": 0.1368,
+      "step": 7900
+    },
+    {
+      "epoch": 6.905281536446966,
+      "grad_norm": 2.818596124649048,
+      "learning_rate": 1.7176128093158662e-05,
+      "loss": 0.1304,
+      "step": 7910
+    },
+    {
+      "epoch": 6.914011348756002,
+      "grad_norm": 2.2759499549865723,
+      "learning_rate": 1.712760795730228e-05,
+      "loss": 0.1264,
+      "step": 7920
+    },
+    {
+      "epoch": 6.922741161065037,
+      "grad_norm": 3.426044225692749,
+      "learning_rate": 1.7079087821445903e-05,
+      "loss": 0.1131,
+      "step": 7930
+    },
+    {
+      "epoch": 6.931470973374072,
+      "grad_norm": 1.9681727886199951,
+      "learning_rate": 1.703056768558952e-05,
+      "loss": 0.1231,
+      "step": 7940
+    },
+    {
+      "epoch": 6.940200785683108,
+      "grad_norm": 1.9072273969650269,
+      "learning_rate": 1.698204754973314e-05,
+      "loss": 0.096,
+      "step": 7950
+    },
+    {
+      "epoch": 6.948930597992144,
+      "grad_norm": 2.6058807373046875,
+      "learning_rate": 1.693352741387676e-05,
+      "loss": 0.1137,
+      "step": 7960
+    },
+    {
+      "epoch": 6.9576604103011785,
+      "grad_norm": 1.815313696861267,
+      "learning_rate": 1.688500727802038e-05,
+      "loss": 0.0869,
+      "step": 7970
+    },
+    {
+      "epoch": 6.9663902226102135,
+      "grad_norm": 3.1374306678771973,
+      "learning_rate": 1.6836487142164e-05,
+      "loss": 0.1256,
+      "step": 7980
+    },
+    {
+      "epoch": 6.975120034919249,
+      "grad_norm": 1.8264007568359375,
+      "learning_rate": 1.6787967006307616e-05,
+      "loss": 0.0917,
+      "step": 7990
+    },
+    {
+      "epoch": 6.983849847228285,
+      "grad_norm": 2.1605708599090576,
+      "learning_rate": 1.673944687045124e-05,
+      "loss": 0.134,
+      "step": 8000
+    },
+    {
+      "epoch": 6.99257965953732,
+      "grad_norm": 2.8631725311279297,
+      "learning_rate": 1.6690926734594857e-05,
+      "loss": 0.1275,
+      "step": 8010
+    },
+    {
+      "epoch": 6.999563509384548,
+      "eval_accuracy": 0.9620058924625583,
+      "eval_loss": 0.12257199734449387,
+      "eval_runtime": 61.6953,
+      "eval_samples_per_second": 264.072,
+      "eval_steps_per_second": 8.266,
+      "step": 8018
+    },
+    {
+      "epoch": 7.001309471846355,
+      "grad_norm": 2.7233829498291016,
+      "learning_rate": 1.6642406598738478e-05,
+      "loss": 0.1128,
+      "step": 8020
+    },
+    {
+      "epoch": 7.010039284155391,
+      "grad_norm": 2.6346898078918457,
+      "learning_rate": 1.6593886462882098e-05,
+      "loss": 0.107,
+      "step": 8030
+    },
+    {
+      "epoch": 7.018769096464426,
+      "grad_norm": 3.0289077758789062,
+      "learning_rate": 1.6545366327025715e-05,
+      "loss": 0.1282,
+      "step": 8040
+    },
+    {
+      "epoch": 7.027498908773461,
+      "grad_norm": 3.5590922832489014,
+      "learning_rate": 1.6496846191169336e-05,
+      "loss": 0.1248,
+      "step": 8050
+    },
+    {
+      "epoch": 7.036228721082496,
+      "grad_norm": 2.1778759956359863,
+      "learning_rate": 1.6448326055312957e-05,
+      "loss": 0.1206,
+      "step": 8060
+    },
+    {
+      "epoch": 7.044958533391532,
+      "grad_norm": 3.13328218460083,
+      "learning_rate": 1.6399805919456577e-05,
+      "loss": 0.1064,
+      "step": 8070
+    },
+    {
+      "epoch": 7.053688345700567,
+      "grad_norm": 2.862576484680176,
+      "learning_rate": 1.6351285783600194e-05,
+      "loss": 0.1133,
+      "step": 8080
+    },
+    {
+      "epoch": 7.062418158009603,
+      "grad_norm": 2.9792587757110596,
+      "learning_rate": 1.6302765647743815e-05,
+      "loss": 0.1542,
+      "step": 8090
+    },
+    {
+      "epoch": 7.071147970318638,
+      "grad_norm": 3.1806788444519043,
+      "learning_rate": 1.6254245511887435e-05,
+      "loss": 0.1284,
+      "step": 8100
+    },
+    {
+      "epoch": 7.0798777826276735,
+      "grad_norm": 2.1008007526397705,
+      "learning_rate": 1.6205725376031052e-05,
+      "loss": 0.0993,
+      "step": 8110
+    },
+    {
+      "epoch": 7.0886075949367084,
+      "grad_norm": 3.443948984146118,
+      "learning_rate": 1.6157205240174673e-05,
+      "loss": 0.1321,
+      "step": 8120
+    },
+    {
+      "epoch": 7.097337407245744,
+      "grad_norm": 3.075568675994873,
+      "learning_rate": 1.6108685104318294e-05,
+      "loss": 0.1135,
+      "step": 8130
+    },
+    {
+      "epoch": 7.106067219554779,
+      "grad_norm": 2.7205452919006348,
+      "learning_rate": 1.6060164968461914e-05,
+      "loss": 0.1045,
+      "step": 8140
+    },
+    {
+      "epoch": 7.114797031863815,
+      "grad_norm": 1.9306424856185913,
+      "learning_rate": 1.601164483260553e-05,
+      "loss": 0.1155,
+      "step": 8150
+    },
+    {
+      "epoch": 7.12352684417285,
+      "grad_norm": 1.848753571510315,
+      "learning_rate": 1.5963124696749152e-05,
+      "loss": 0.121,
+      "step": 8160
+    },
+    {
+      "epoch": 7.132256656481886,
+      "grad_norm": 2.816012144088745,
+      "learning_rate": 1.5914604560892772e-05,
+      "loss": 0.1195,
+      "step": 8170
+    },
+    {
+      "epoch": 7.140986468790921,
+      "grad_norm": 1.6042299270629883,
+      "learning_rate": 1.586608442503639e-05,
+      "loss": 0.109,
+      "step": 8180
+    },
+    {
+      "epoch": 7.149716281099956,
+      "grad_norm": 1.9858665466308594,
+      "learning_rate": 1.581756428918001e-05,
+      "loss": 0.1218,
+      "step": 8190
+    },
+    {
+      "epoch": 7.158446093408991,
+      "grad_norm": 3.661896228790283,
+      "learning_rate": 1.576904415332363e-05,
+      "loss": 0.1162,
+      "step": 8200
+    },
+    {
+      "epoch": 7.167175905718027,
+      "grad_norm": 3.4283740520477295,
+      "learning_rate": 1.572052401746725e-05,
+      "loss": 0.1192,
+      "step": 8210
+    },
+    {
+      "epoch": 7.175905718027062,
+      "grad_norm": 2.0922162532806396,
+      "learning_rate": 1.567200388161087e-05,
+      "loss": 0.1228,
+      "step": 8220
+    },
+    {
+      "epoch": 7.184635530336098,
+      "grad_norm": 3.018186330795288,
+      "learning_rate": 1.562348374575449e-05,
+      "loss": 0.1239,
+      "step": 8230
+    },
+    {
+      "epoch": 7.193365342645133,
+      "grad_norm": 3.431612491607666,
+      "learning_rate": 1.557496360989811e-05,
+      "loss": 0.1199,
+      "step": 8240
+    },
+    {
+      "epoch": 7.2020951549541685,
+      "grad_norm": 1.7211098670959473,
+      "learning_rate": 1.5526443474041727e-05,
+      "loss": 0.1241,
+      "step": 8250
+    },
+    {
+      "epoch": 7.210824967263203,
+      "grad_norm": 3.201613664627075,
+      "learning_rate": 1.547792333818535e-05,
+      "loss": 0.1528,
+      "step": 8260
+    },
+    {
+      "epoch": 7.219554779572239,
+      "grad_norm": 2.8337209224700928,
+      "learning_rate": 1.5429403202328968e-05,
+      "loss": 0.1037,
+      "step": 8270
+    },
+    {
+      "epoch": 7.228284591881274,
+      "grad_norm": 2.8004138469696045,
+      "learning_rate": 1.5380883066472588e-05,
+      "loss": 0.1221,
+      "step": 8280
+    },
+    {
+      "epoch": 7.23701440419031,
+      "grad_norm": 4.067490577697754,
+      "learning_rate": 1.5332362930616205e-05,
+      "loss": 0.1399,
+      "step": 8290
+    },
+    {
+      "epoch": 7.245744216499345,
+      "grad_norm": 3.5075736045837402,
+      "learning_rate": 1.5283842794759826e-05,
+      "loss": 0.1412,
+      "step": 8300
+    },
+    {
+      "epoch": 7.254474028808381,
+      "grad_norm": 3.4626917839050293,
+      "learning_rate": 1.5235322658903445e-05,
+      "loss": 0.1311,
+      "step": 8310
+    },
+    {
+      "epoch": 7.263203841117416,
+      "grad_norm": 3.102483034133911,
+      "learning_rate": 1.5186802523047064e-05,
+      "loss": 0.1054,
+      "step": 8320
+    },
+    {
+      "epoch": 7.271933653426451,
+      "grad_norm": 3.181889533996582,
+      "learning_rate": 1.5138282387190686e-05,
+      "loss": 0.1469,
+      "step": 8330
+    },
+    {
+      "epoch": 7.280663465735486,
+      "grad_norm": 2.642778158187866,
+      "learning_rate": 1.5089762251334305e-05,
+      "loss": 0.1359,
+      "step": 8340
+    },
+    {
+      "epoch": 7.289393278044522,
+      "grad_norm": 2.339479446411133,
+      "learning_rate": 1.5041242115477924e-05,
+      "loss": 0.0924,
+      "step": 8350
+    },
+    {
+      "epoch": 7.298123090353557,
+      "grad_norm": 3.010129690170288,
+      "learning_rate": 1.4992721979621544e-05,
+      "loss": 0.1211,
+      "step": 8360
+    },
+    {
+      "epoch": 7.306852902662593,
+      "grad_norm": 2.169443130493164,
+      "learning_rate": 1.4944201843765163e-05,
+      "loss": 0.0962,
+      "step": 8370
+    },
+    {
+      "epoch": 7.315582714971628,
+      "grad_norm": 2.3417961597442627,
+      "learning_rate": 1.4895681707908782e-05,
+      "loss": 0.1263,
+      "step": 8380
+    },
+    {
+      "epoch": 7.3243125272806635,
+      "grad_norm": 2.7159245014190674,
+      "learning_rate": 1.4847161572052404e-05,
+      "loss": 0.0773,
+      "step": 8390
+    },
+    {
+      "epoch": 7.333042339589698,
+      "grad_norm": 1.89850652217865,
+      "learning_rate": 1.4798641436196023e-05,
+      "loss": 0.1134,
+      "step": 8400
+    },
+    {
+      "epoch": 7.341772151898734,
+      "grad_norm": 2.856013536453247,
+      "learning_rate": 1.4750121300339642e-05,
+      "loss": 0.1278,
+      "step": 8410
+    },
+    {
+      "epoch": 7.350501964207769,
+      "grad_norm": 2.531646728515625,
+      "learning_rate": 1.470160116448326e-05,
+      "loss": 0.1234,
+      "step": 8420
+    },
+    {
+      "epoch": 7.359231776516805,
+      "grad_norm": 3.2163310050964355,
+      "learning_rate": 1.4653081028626881e-05,
+      "loss": 0.1112,
+      "step": 8430
+    },
+    {
+      "epoch": 7.36796158882584,
+      "grad_norm": 2.611832618713379,
+      "learning_rate": 1.46045608927705e-05,
+      "loss": 0.1146,
+      "step": 8440
+    },
+    {
+      "epoch": 7.376691401134876,
+      "grad_norm": 3.106451988220215,
+      "learning_rate": 1.4556040756914119e-05,
+      "loss": 0.1088,
+      "step": 8450
+    },
+    {
+      "epoch": 7.385421213443911,
+      "grad_norm": 2.7591700553894043,
+      "learning_rate": 1.4507520621057741e-05,
+      "loss": 0.1313,
+      "step": 8460
+    },
+    {
+      "epoch": 7.394151025752946,
+      "grad_norm": 2.3061065673828125,
+      "learning_rate": 1.445900048520136e-05,
+      "loss": 0.1187,
+      "step": 8470
+    },
+    {
+      "epoch": 7.402880838061981,
+      "grad_norm": 2.7407071590423584,
+      "learning_rate": 1.4410480349344979e-05,
+      "loss": 0.1076,
+      "step": 8480
+    },
+    {
+      "epoch": 7.411610650371017,
+      "grad_norm": 3.5780117511749268,
+      "learning_rate": 1.4361960213488598e-05,
+      "loss": 0.1261,
+      "step": 8490
+    },
+    {
+      "epoch": 7.420340462680052,
+      "grad_norm": 2.2158961296081543,
+      "learning_rate": 1.4313440077632218e-05,
+      "loss": 0.1042,
+      "step": 8500
+    },
+    {
+      "epoch": 7.429070274989088,
+      "grad_norm": 3.758617639541626,
+      "learning_rate": 1.4264919941775837e-05,
+      "loss": 0.1146,
+      "step": 8510
+    },
+    {
+      "epoch": 7.437800087298123,
+      "grad_norm": 3.035114049911499,
+      "learning_rate": 1.421639980591946e-05,
+      "loss": 0.1149,
+      "step": 8520
+    },
+    {
+      "epoch": 7.4465298996071585,
+      "grad_norm": 3.016139030456543,
+      "learning_rate": 1.4167879670063078e-05,
+      "loss": 0.0983,
+      "step": 8530
+    },
+    {
+      "epoch": 7.455259711916193,
+      "grad_norm": 1.1190143823623657,
+      "learning_rate": 1.4119359534206697e-05,
+      "loss": 0.1016,
+      "step": 8540
+    },
+    {
+      "epoch": 7.463989524225229,
+      "grad_norm": 1.6610057353973389,
+      "learning_rate": 1.4070839398350316e-05,
+      "loss": 0.1098,
+      "step": 8550
+    },
+    {
+      "epoch": 7.472719336534264,
+      "grad_norm": 2.246140956878662,
+      "learning_rate": 1.4022319262493935e-05,
+      "loss": 0.1226,
+      "step": 8560
+    },
+    {
+      "epoch": 7.4814491488433,
+      "grad_norm": 3.7668988704681396,
+      "learning_rate": 1.3973799126637555e-05,
+      "loss": 0.0909,
+      "step": 8570
+    },
+    {
+      "epoch": 7.490178961152335,
+      "grad_norm": 1.9712340831756592,
+      "learning_rate": 1.3925278990781174e-05,
+      "loss": 0.1065,
+      "step": 8580
+    },
+    {
+      "epoch": 7.498908773461371,
+      "grad_norm": 2.9757723808288574,
+      "learning_rate": 1.3876758854924796e-05,
+      "loss": 0.1222,
+      "step": 8590
+    },
+    {
+      "epoch": 7.5076385857704055,
+      "grad_norm": 3.4910058975219727,
+      "learning_rate": 1.3828238719068415e-05,
+      "loss": 0.1156,
+      "step": 8600
+    },
+    {
+      "epoch": 7.516368398079441,
+      "grad_norm": 2.6508007049560547,
+      "learning_rate": 1.3779718583212034e-05,
+      "loss": 0.099,
+      "step": 8610
+    },
+    {
+      "epoch": 7.525098210388476,
+      "grad_norm": 1.8022457361221313,
+      "learning_rate": 1.3731198447355653e-05,
+      "loss": 0.0829,
+      "step": 8620
+    },
+    {
+      "epoch": 7.533828022697512,
+      "grad_norm": 2.3336634635925293,
+      "learning_rate": 1.3682678311499272e-05,
+      "loss": 0.1409,
+      "step": 8630
+    },
+    {
+      "epoch": 7.542557835006547,
+      "grad_norm": 3.1852900981903076,
+      "learning_rate": 1.3634158175642892e-05,
+      "loss": 0.125,
+      "step": 8640
+    },
+    {
+      "epoch": 7.551287647315583,
+      "grad_norm": 2.8601016998291016,
+      "learning_rate": 1.3585638039786511e-05,
+      "loss": 0.1035,
+      "step": 8650
+    },
+    {
+      "epoch": 7.560017459624618,
+      "grad_norm": 2.5289573669433594,
+      "learning_rate": 1.3537117903930133e-05,
+      "loss": 0.1093,
+      "step": 8660
+    },
+    {
+      "epoch": 7.5687472719336535,
+      "grad_norm": 2.4447848796844482,
+      "learning_rate": 1.3488597768073752e-05,
+      "loss": 0.1076,
+      "step": 8670
+    },
+    {
+      "epoch": 7.577477084242688,
+      "grad_norm": 3.016014337539673,
+      "learning_rate": 1.3440077632217371e-05,
+      "loss": 0.1189,
+      "step": 8680
+    },
+    {
+      "epoch": 7.586206896551724,
+      "grad_norm": 2.6992433071136475,
+      "learning_rate": 1.339155749636099e-05,
+      "loss": 0.1088,
+      "step": 8690
+    },
+    {
+      "epoch": 7.594936708860759,
+      "grad_norm": 2.2244362831115723,
+      "learning_rate": 1.3343037360504609e-05,
+      "loss": 0.0965,
+      "step": 8700
+    },
+    {
+      "epoch": 7.603666521169795,
+      "grad_norm": 2.8653414249420166,
+      "learning_rate": 1.329451722464823e-05,
+      "loss": 0.1022,
+      "step": 8710
+    },
+    {
+      "epoch": 7.61239633347883,
+      "grad_norm": 3.810765027999878,
+      "learning_rate": 1.324599708879185e-05,
+      "loss": 0.1208,
+      "step": 8720
+    },
+    {
+      "epoch": 7.621126145787866,
+      "grad_norm": 2.089237689971924,
+      "learning_rate": 1.319747695293547e-05,
+      "loss": 0.0989,
+      "step": 8730
+    },
+    {
+      "epoch": 7.6298559580969005,
+      "grad_norm": 1.9120042324066162,
+      "learning_rate": 1.314895681707909e-05,
+      "loss": 0.112,
+      "step": 8740
+    },
+    {
+      "epoch": 7.638585770405936,
+      "grad_norm": 2.164149284362793,
+      "learning_rate": 1.3100436681222708e-05,
+      "loss": 0.1183,
+      "step": 8750
+    },
+    {
+      "epoch": 7.647315582714971,
+      "grad_norm": 1.7820502519607544,
+      "learning_rate": 1.3051916545366327e-05,
+      "loss": 0.1187,
+      "step": 8760
+    },
+    {
+      "epoch": 7.656045395024007,
+      "grad_norm": 3.0987606048583984,
+      "learning_rate": 1.3003396409509946e-05,
+      "loss": 0.1192,
+      "step": 8770
+    },
+    {
+      "epoch": 7.664775207333042,
+      "grad_norm": 2.332767963409424,
+      "learning_rate": 1.2954876273653566e-05,
+      "loss": 0.1219,
+      "step": 8780
+    },
+    {
+      "epoch": 7.673505019642078,
+      "grad_norm": 2.8338541984558105,
+      "learning_rate": 1.2906356137797187e-05,
+      "loss": 0.0953,
+      "step": 8790
+    },
+    {
+      "epoch": 7.682234831951113,
+      "grad_norm": 2.000577926635742,
+      "learning_rate": 1.2857836001940808e-05,
+      "loss": 0.0969,
+      "step": 8800
+    },
+    {
+      "epoch": 7.6909646442601485,
+      "grad_norm": 3.364076614379883,
+      "learning_rate": 1.2809315866084426e-05,
+      "loss": 0.1117,
+      "step": 8810
+    },
+    {
+      "epoch": 7.699694456569183,
+      "grad_norm": 4.16958475112915,
+      "learning_rate": 1.2760795730228045e-05,
+      "loss": 0.1216,
+      "step": 8820
+    },
+    {
+      "epoch": 7.708424268878219,
+      "grad_norm": 2.2611634731292725,
+      "learning_rate": 1.2712275594371664e-05,
+      "loss": 0.0929,
+      "step": 8830
+    },
+    {
+      "epoch": 7.717154081187254,
+      "grad_norm": 2.592312812805176,
+      "learning_rate": 1.2663755458515283e-05,
+      "loss": 0.1147,
+      "step": 8840
+    },
+    {
+      "epoch": 7.72588389349629,
+      "grad_norm": 3.680349588394165,
+      "learning_rate": 1.2615235322658905e-05,
+      "loss": 0.1184,
+      "step": 8850
+    },
+    {
+      "epoch": 7.734613705805325,
+      "grad_norm": 3.146328926086426,
+      "learning_rate": 1.2566715186802524e-05,
+      "loss": 0.1083,
+      "step": 8860
+    },
+    {
+      "epoch": 7.743343518114361,
+      "grad_norm": 3.165249824523926,
+      "learning_rate": 1.2518195050946145e-05,
+      "loss": 0.1303,
+      "step": 8870
+    },
+    {
+      "epoch": 7.7520733304233955,
+      "grad_norm": 3.2401604652404785,
+      "learning_rate": 1.2469674915089763e-05,
+      "loss": 0.1032,
+      "step": 8880
+    },
+    {
+      "epoch": 7.760803142732431,
+      "grad_norm": 1.9936988353729248,
+      "learning_rate": 1.2421154779233382e-05,
+      "loss": 0.1187,
+      "step": 8890
+    },
+    {
+      "epoch": 7.769532955041466,
+      "grad_norm": 2.625159502029419,
+      "learning_rate": 1.2372634643377003e-05,
+      "loss": 0.1062,
+      "step": 8900
+    },
+    {
+      "epoch": 7.778262767350502,
+      "grad_norm": 1.672642707824707,
+      "learning_rate": 1.2324114507520622e-05,
+      "loss": 0.1183,
+      "step": 8910
+    },
+    {
+      "epoch": 7.786992579659537,
+      "grad_norm": 2.378697156906128,
+      "learning_rate": 1.227559437166424e-05,
+      "loss": 0.1197,
+      "step": 8920
+    },
+    {
+      "epoch": 7.795722391968573,
+      "grad_norm": 2.9251904487609863,
+      "learning_rate": 1.222707423580786e-05,
+      "loss": 0.1141,
+      "step": 8930
+    },
+    {
+      "epoch": 7.804452204277608,
+      "grad_norm": 1.902925968170166,
+      "learning_rate": 1.217855409995148e-05,
+      "loss": 0.1034,
+      "step": 8940
+    },
+    {
+      "epoch": 7.8131820165866435,
+      "grad_norm": 2.213522434234619,
+      "learning_rate": 1.21300339640951e-05,
+      "loss": 0.1124,
+      "step": 8950
+    },
+    {
+      "epoch": 7.821911828895678,
+      "grad_norm": 3.5406954288482666,
+      "learning_rate": 1.208151382823872e-05,
+      "loss": 0.0988,
+      "step": 8960
+    },
+    {
+      "epoch": 7.830641641204714,
+      "grad_norm": 2.070265531539917,
+      "learning_rate": 1.203299369238234e-05,
+      "loss": 0.1182,
+      "step": 8970
+    },
+    {
+      "epoch": 7.839371453513749,
+      "grad_norm": 3.4907076358795166,
+      "learning_rate": 1.1984473556525959e-05,
+      "loss": 0.1009,
+      "step": 8980
+    },
+    {
+      "epoch": 7.848101265822785,
+      "grad_norm": 2.2645416259765625,
+      "learning_rate": 1.1935953420669578e-05,
+      "loss": 0.1021,
+      "step": 8990
+    },
+    {
+      "epoch": 7.85683107813182,
+      "grad_norm": 2.2879273891448975,
+      "learning_rate": 1.1887433284813198e-05,
+      "loss": 0.1166,
+      "step": 9000
+    },
+    {
+      "epoch": 7.865560890440856,
+      "grad_norm": 2.280569553375244,
+      "learning_rate": 1.1838913148956817e-05,
+      "loss": 0.1145,
+      "step": 9010
+    },
+    {
+      "epoch": 7.8742907027498905,
+      "grad_norm": 2.2898128032684326,
+      "learning_rate": 1.1790393013100438e-05,
+      "loss": 0.095,
+      "step": 9020
+    },
+    {
+      "epoch": 7.883020515058926,
+      "grad_norm": 2.112298011779785,
+      "learning_rate": 1.1741872877244056e-05,
+      "loss": 0.0834,
+      "step": 9030
+    },
+    {
+      "epoch": 7.891750327367961,
+      "grad_norm": 3.349541425704956,
+      "learning_rate": 1.1693352741387677e-05,
+      "loss": 0.1125,
+      "step": 9040
+    },
+    {
+      "epoch": 7.900480139676997,
+      "grad_norm": 2.482618570327759,
+      "learning_rate": 1.1644832605531296e-05,
+      "loss": 0.1059,
+      "step": 9050
+    },
+    {
+      "epoch": 7.909209951986032,
+      "grad_norm": 2.8729076385498047,
+      "learning_rate": 1.1596312469674915e-05,
+      "loss": 0.1312,
+      "step": 9060
+    },
+    {
+      "epoch": 7.917939764295068,
+      "grad_norm": 2.159461498260498,
+      "learning_rate": 1.1547792333818535e-05,
+      "loss": 0.1032,
+      "step": 9070
+    },
+    {
+      "epoch": 7.926669576604103,
+      "grad_norm": 1.993208646774292,
+      "learning_rate": 1.1499272197962154e-05,
+      "loss": 0.1139,
+      "step": 9080
+    },
+    {
+      "epoch": 7.9353993889131385,
+      "grad_norm": 2.4738166332244873,
+      "learning_rate": 1.1450752062105775e-05,
+      "loss": 0.1021,
+      "step": 9090
+    },
+    {
+      "epoch": 7.944129201222173,
+      "grad_norm": 2.5491795539855957,
+      "learning_rate": 1.1402231926249395e-05,
+      "loss": 0.1195,
+      "step": 9100
+    },
+    {
+      "epoch": 7.952859013531209,
+      "grad_norm": 2.6320948600769043,
+      "learning_rate": 1.1353711790393014e-05,
+      "loss": 0.0882,
+      "step": 9110
+    },
+    {
+      "epoch": 7.961588825840244,
+      "grad_norm": 2.642817497253418,
+      "learning_rate": 1.1305191654536633e-05,
+      "loss": 0.1278,
+      "step": 9120
+    },
+    {
+      "epoch": 7.97031863814928,
+      "grad_norm": 1.8427996635437012,
+      "learning_rate": 1.1256671518680253e-05,
+      "loss": 0.1091,
+      "step": 9130
+    },
+    {
+      "epoch": 7.979048450458315,
+      "grad_norm": 3.3067104816436768,
+      "learning_rate": 1.1208151382823872e-05,
+      "loss": 0.1177,
+      "step": 9140
+    },
+    {
+      "epoch": 7.987778262767351,
+      "grad_norm": 1.9407857656478882,
+      "learning_rate": 1.1159631246967491e-05,
+      "loss": 0.1146,
+      "step": 9150
+    },
+    {
+      "epoch": 7.9965080750763855,
+      "grad_norm": 2.7859206199645996,
+      "learning_rate": 1.1111111111111112e-05,
+      "loss": 0.1323,
+      "step": 9160
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.9777191259513872,
+      "eval_loss": 0.07020638883113861,
+      "eval_runtime": 60.5088,
+      "eval_samples_per_second": 269.25,
+      "eval_steps_per_second": 8.429,
+      "step": 9164
+    },
+    {
+      "epoch": 8.005237887385421,
+      "grad_norm": 3.0209567546844482,
+      "learning_rate": 1.1062590975254732e-05,
+      "loss": 0.1044,
+      "step": 9170
+    },
+    {
+      "epoch": 8.013967699694456,
+      "grad_norm": 3.3612194061279297,
+      "learning_rate": 1.1014070839398351e-05,
+      "loss": 0.1367,
+      "step": 9180
+    },
+    {
+      "epoch": 8.022697512003491,
+      "grad_norm": 3.807859182357788,
+      "learning_rate": 1.096555070354197e-05,
+      "loss": 0.0977,
+      "step": 9190
+    },
+    {
+      "epoch": 8.031427324312528,
+      "grad_norm": 1.820942759513855,
+      "learning_rate": 1.091703056768559e-05,
+      "loss": 0.1027,
+      "step": 9200
+    },
+    {
+      "epoch": 8.040157136621563,
+      "grad_norm": 1.7831873893737793,
+      "learning_rate": 1.086851043182921e-05,
+      "loss": 0.114,
+      "step": 9210
+    },
+    {
+      "epoch": 8.048886948930598,
+      "grad_norm": 2.5594382286071777,
+      "learning_rate": 1.0819990295972828e-05,
+      "loss": 0.1036,
+      "step": 9220
+    },
+    {
+      "epoch": 8.057616761239633,
+      "grad_norm": 2.1451315879821777,
+      "learning_rate": 1.0771470160116449e-05,
+      "loss": 0.1002,
+      "step": 9230
+    },
+    {
+      "epoch": 8.06634657354867,
+      "grad_norm": 2.5839290618896484,
+      "learning_rate": 1.072295002426007e-05,
+      "loss": 0.1026,
+      "step": 9240
+    },
+    {
+      "epoch": 8.075076385857704,
+      "grad_norm": 2.6166088581085205,
+      "learning_rate": 1.0674429888403688e-05,
+      "loss": 0.1,
+      "step": 9250
+    },
+    {
+      "epoch": 8.083806198166739,
+      "grad_norm": 3.226916790008545,
+      "learning_rate": 1.0625909752547307e-05,
+      "loss": 0.0857,
+      "step": 9260
+    },
+    {
+      "epoch": 8.092536010475774,
+      "grad_norm": 2.4332773685455322,
+      "learning_rate": 1.0577389616690927e-05,
+      "loss": 0.0862,
+      "step": 9270
+    },
+    {
+      "epoch": 8.10126582278481,
+      "grad_norm": 2.966545820236206,
+      "learning_rate": 1.0528869480834546e-05,
+      "loss": 0.1025,
+      "step": 9280
+    },
+    {
+      "epoch": 8.109995635093846,
+      "grad_norm": 2.112661123275757,
+      "learning_rate": 1.0480349344978165e-05,
+      "loss": 0.0996,
+      "step": 9290
+    },
+    {
+      "epoch": 8.11872544740288,
+      "grad_norm": 3.0900607109069824,
+      "learning_rate": 1.0431829209121786e-05,
+      "loss": 0.1052,
+      "step": 9300
+    },
+    {
+      "epoch": 8.127455259711915,
+      "grad_norm": 2.729537010192871,
+      "learning_rate": 1.0383309073265406e-05,
+      "loss": 0.1012,
+      "step": 9310
+    },
+    {
+      "epoch": 8.136185072020952,
+      "grad_norm": 2.780118227005005,
+      "learning_rate": 1.0334788937409025e-05,
+      "loss": 0.1136,
+      "step": 9320
+    },
+    {
+      "epoch": 8.144914884329987,
+      "grad_norm": 1.7321745157241821,
+      "learning_rate": 1.0286268801552646e-05,
+      "loss": 0.1115,
+      "step": 9330
+    },
+    {
+      "epoch": 8.153644696639022,
+      "grad_norm": 2.258513927459717,
+      "learning_rate": 1.0237748665696265e-05,
+      "loss": 0.1353,
+      "step": 9340
+    },
+    {
+      "epoch": 8.162374508948057,
+      "grad_norm": 3.355522632598877,
+      "learning_rate": 1.0189228529839883e-05,
+      "loss": 0.0798,
+      "step": 9350
+    },
+    {
+      "epoch": 8.171104321257094,
+      "grad_norm": 1.8760606050491333,
+      "learning_rate": 1.0140708393983504e-05,
+      "loss": 0.1136,
+      "step": 9360
+    },
+    {
+      "epoch": 8.179834133566128,
+      "grad_norm": 2.638148546218872,
+      "learning_rate": 1.0092188258127123e-05,
+      "loss": 0.1096,
+      "step": 9370
+    },
+    {
+      "epoch": 8.188563945875163,
+      "grad_norm": 3.003227472305298,
+      "learning_rate": 1.0043668122270743e-05,
+      "loss": 0.1027,
+      "step": 9380
+    },
+    {
+      "epoch": 8.197293758184198,
+      "grad_norm": 2.2695999145507812,
+      "learning_rate": 9.995147986414362e-06,
+      "loss": 0.1027,
+      "step": 9390
+    },
+    {
+      "epoch": 8.206023570493235,
+      "grad_norm": 2.753089666366577,
+      "learning_rate": 9.946627850557983e-06,
+      "loss": 0.1131,
+      "step": 9400
+    },
+    {
+      "epoch": 8.21475338280227,
+      "grad_norm": 2.087954521179199,
+      "learning_rate": 9.898107714701602e-06,
+      "loss": 0.1096,
+      "step": 9410
+    },
+    {
+      "epoch": 8.223483195111305,
+      "grad_norm": 2.1234874725341797,
+      "learning_rate": 9.84958757884522e-06,
+      "loss": 0.1281,
+      "step": 9420
+    },
+    {
+      "epoch": 8.23221300742034,
+      "grad_norm": 2.573425531387329,
+      "learning_rate": 9.801067442988841e-06,
+      "loss": 0.1221,
+      "step": 9430
+    },
+    {
+      "epoch": 8.240942819729376,
+      "grad_norm": 2.972698211669922,
+      "learning_rate": 9.75254730713246e-06,
+      "loss": 0.1063,
+      "step": 9440
+    },
+    {
+      "epoch": 8.249672632038411,
+      "grad_norm": 3.052992105484009,
+      "learning_rate": 9.70402717127608e-06,
+      "loss": 0.1117,
+      "step": 9450
+    },
+    {
+      "epoch": 8.258402444347446,
+      "grad_norm": 2.1158103942871094,
+      "learning_rate": 9.655507035419701e-06,
+      "loss": 0.092,
+      "step": 9460
+    },
+    {
+      "epoch": 8.267132256656481,
+      "grad_norm": 2.4370791912078857,
+      "learning_rate": 9.60698689956332e-06,
+      "loss": 0.1092,
+      "step": 9470
+    },
+    {
+      "epoch": 8.275862068965518,
+      "grad_norm": 3.0041496753692627,
+      "learning_rate": 9.558466763706939e-06,
+      "loss": 0.0869,
+      "step": 9480
+    },
+    {
+      "epoch": 8.284591881274553,
+      "grad_norm": 1.6758767366409302,
+      "learning_rate": 9.509946627850557e-06,
+      "loss": 0.1012,
+      "step": 9490
+    },
+    {
+      "epoch": 8.293321693583588,
+      "grad_norm": 2.937406301498413,
+      "learning_rate": 9.461426491994178e-06,
+      "loss": 0.1152,
+      "step": 9500
+    },
+    {
+      "epoch": 8.302051505892623,
+      "grad_norm": 2.93967866897583,
+      "learning_rate": 9.412906356137797e-06,
+      "loss": 0.099,
+      "step": 9510
+    },
+    {
+      "epoch": 8.31078131820166,
+      "grad_norm": 2.0390257835388184,
+      "learning_rate": 9.364386220281417e-06,
+      "loss": 0.0926,
+      "step": 9520
+    },
+    {
+      "epoch": 8.319511130510694,
+      "grad_norm": 2.6134767532348633,
+      "learning_rate": 9.315866084425038e-06,
+      "loss": 0.1139,
+      "step": 9530
+    },
+    {
+      "epoch": 8.328240942819729,
+      "grad_norm": 3.398548126220703,
+      "learning_rate": 9.267345948568657e-06,
+      "loss": 0.1051,
+      "step": 9540
+    },
+    {
+      "epoch": 8.336970755128764,
+      "grad_norm": 3.847663164138794,
+      "learning_rate": 9.218825812712276e-06,
+      "loss": 0.119,
+      "step": 9550
+    },
+    {
+      "epoch": 8.3457005674378,
+      "grad_norm": 3.1890709400177,
+      "learning_rate": 9.170305676855896e-06,
+      "loss": 0.1459,
+      "step": 9560
+    },
+    {
+      "epoch": 8.354430379746836,
+      "grad_norm": 1.5987610816955566,
+      "learning_rate": 9.121785540999515e-06,
+      "loss": 0.0985,
+      "step": 9570
+    },
+    {
+      "epoch": 8.36316019205587,
+      "grad_norm": 2.2405405044555664,
+      "learning_rate": 9.073265405143134e-06,
+      "loss": 0.1044,
+      "step": 9580
+    },
+    {
+      "epoch": 8.371890004364905,
+      "grad_norm": 2.4690017700195312,
+      "learning_rate": 9.024745269286754e-06,
+      "loss": 0.0976,
+      "step": 9590
+    },
+    {
+      "epoch": 8.380619816673942,
+      "grad_norm": 2.168797254562378,
+      "learning_rate": 8.976225133430375e-06,
+      "loss": 0.0933,
+      "step": 9600
+    },
+    {
+      "epoch": 8.389349628982977,
+      "grad_norm": 2.74342942237854,
+      "learning_rate": 8.927704997573994e-06,
+      "loss": 0.099,
+      "step": 9610
+    },
+    {
+      "epoch": 8.398079441292012,
+      "grad_norm": 3.6532704830169678,
+      "learning_rate": 8.879184861717613e-06,
+      "loss": 0.1303,
+      "step": 9620
+    },
+    {
+      "epoch": 8.406809253601047,
+      "grad_norm": 2.5002212524414062,
+      "learning_rate": 8.830664725861233e-06,
+      "loss": 0.1139,
+      "step": 9630
+    },
+    {
+      "epoch": 8.415539065910083,
+      "grad_norm": 2.5484557151794434,
+      "learning_rate": 8.782144590004852e-06,
+      "loss": 0.1261,
+      "step": 9640
+    },
+    {
+      "epoch": 8.424268878219118,
+      "grad_norm": 2.8932912349700928,
+      "learning_rate": 8.733624454148471e-06,
+      "loss": 0.1126,
+      "step": 9650
+    },
+    {
+      "epoch": 8.432998690528153,
+      "grad_norm": 2.334230899810791,
+      "learning_rate": 8.685104318292092e-06,
+      "loss": 0.1021,
+      "step": 9660
+    },
+    {
+      "epoch": 8.441728502837188,
+      "grad_norm": 2.884859561920166,
+      "learning_rate": 8.636584182435712e-06,
+      "loss": 0.0958,
+      "step": 9670
+    },
+    {
+      "epoch": 8.450458315146225,
+      "grad_norm": 1.8755512237548828,
+      "learning_rate": 8.588064046579331e-06,
+      "loss": 0.0928,
+      "step": 9680
+    },
+    {
+      "epoch": 8.45918812745526,
+      "grad_norm": 2.3897783756256104,
+      "learning_rate": 8.539543910722951e-06,
+      "loss": 0.0897,
+      "step": 9690
+    },
+    {
+      "epoch": 8.467917939764295,
+      "grad_norm": 3.4187004566192627,
+      "learning_rate": 8.49102377486657e-06,
+      "loss": 0.0978,
+      "step": 9700
+    },
+    {
+      "epoch": 8.47664775207333,
+      "grad_norm": 3.359574317932129,
+      "learning_rate": 8.44250363901019e-06,
+      "loss": 0.1029,
+      "step": 9710
+    },
+    {
+      "epoch": 8.485377564382366,
+      "grad_norm": 2.7510204315185547,
+      "learning_rate": 8.393983503153808e-06,
+      "loss": 0.1115,
+      "step": 9720
+    },
+    {
+      "epoch": 8.494107376691401,
+      "grad_norm": 2.2325894832611084,
+      "learning_rate": 8.345463367297429e-06,
+      "loss": 0.1237,
+      "step": 9730
+    },
+    {
+      "epoch": 8.502837189000436,
+      "grad_norm": 2.400143623352051,
+      "learning_rate": 8.296943231441049e-06,
+      "loss": 0.1202,
+      "step": 9740
+    },
+    {
+      "epoch": 8.511567001309471,
+      "grad_norm": 3.032205581665039,
+      "learning_rate": 8.248423095584668e-06,
+      "loss": 0.092,
+      "step": 9750
+    },
+    {
+      "epoch": 8.520296813618508,
+      "grad_norm": 3.307790517807007,
+      "learning_rate": 8.199902959728289e-06,
+      "loss": 0.1069,
+      "step": 9760
+    },
+    {
+      "epoch": 8.529026625927543,
+      "grad_norm": 3.3414857387542725,
+      "learning_rate": 8.151382823871907e-06,
+      "loss": 0.1315,
+      "step": 9770
+    },
+    {
+      "epoch": 8.537756438236578,
+      "grad_norm": 2.243468999862671,
+      "learning_rate": 8.102862688015526e-06,
+      "loss": 0.0986,
+      "step": 9780
+    },
+    {
+      "epoch": 8.546486250545613,
+      "grad_norm": 2.635031223297119,
+      "learning_rate": 8.054342552159147e-06,
+      "loss": 0.1067,
+      "step": 9790
+    },
+    {
+      "epoch": 8.55521606285465,
+      "grad_norm": 2.037966728210449,
+      "learning_rate": 8.005822416302766e-06,
+      "loss": 0.1053,
+      "step": 9800
+    },
+    {
+      "epoch": 8.563945875163684,
+      "grad_norm": 3.361685276031494,
+      "learning_rate": 7.957302280446386e-06,
+      "loss": 0.1166,
+      "step": 9810
+    },
+    {
+      "epoch": 8.572675687472719,
+      "grad_norm": 1.941724419593811,
+      "learning_rate": 7.908782144590005e-06,
+      "loss": 0.0954,
+      "step": 9820
+    },
+    {
+      "epoch": 8.581405499781754,
+      "grad_norm": 2.7277681827545166,
+      "learning_rate": 7.860262008733626e-06,
+      "loss": 0.1212,
+      "step": 9830
+    },
+    {
+      "epoch": 8.59013531209079,
+      "grad_norm": 2.5248844623565674,
+      "learning_rate": 7.811741872877244e-06,
+      "loss": 0.1009,
+      "step": 9840
+    },
+    {
+      "epoch": 8.598865124399826,
+      "grad_norm": 0.636101245880127,
+      "learning_rate": 7.763221737020863e-06,
+      "loss": 0.0992,
+      "step": 9850
+    },
+    {
+      "epoch": 8.60759493670886,
+      "grad_norm": 2.8368330001831055,
+      "learning_rate": 7.714701601164484e-06,
+      "loss": 0.0886,
+      "step": 9860
+    },
+    {
+      "epoch": 8.616324749017895,
+      "grad_norm": 2.3569271564483643,
+      "learning_rate": 7.666181465308103e-06,
+      "loss": 0.0835,
+      "step": 9870
+    },
+    {
+      "epoch": 8.625054561326932,
+      "grad_norm": 2.118471622467041,
+      "learning_rate": 7.617661329451722e-06,
+      "loss": 0.1155,
+      "step": 9880
+    },
+    {
+      "epoch": 8.633784373635967,
+      "grad_norm": 1.9520134925842285,
+      "learning_rate": 7.569141193595343e-06,
+      "loss": 0.0935,
+      "step": 9890
+    },
+    {
+      "epoch": 8.642514185945002,
+      "grad_norm": 2.778207540512085,
+      "learning_rate": 7.520621057738962e-06,
+      "loss": 0.1167,
+      "step": 9900
+    },
+    {
+      "epoch": 8.651243998254037,
+      "grad_norm": 3.1950080394744873,
+      "learning_rate": 7.4721009218825815e-06,
+      "loss": 0.124,
+      "step": 9910
+    },
+    {
+      "epoch": 8.659973810563073,
+      "grad_norm": 1.913509726524353,
+      "learning_rate": 7.423580786026202e-06,
+      "loss": 0.0922,
+      "step": 9920
+    },
+    {
+      "epoch": 8.668703622872108,
+      "grad_norm": 2.9603874683380127,
+      "learning_rate": 7.375060650169821e-06,
+      "loss": 0.1225,
+      "step": 9930
+    },
+    {
+      "epoch": 8.677433435181143,
+      "grad_norm": 2.5559980869293213,
+      "learning_rate": 7.326540514313441e-06,
+      "loss": 0.1159,
+      "step": 9940
+    },
+    {
+      "epoch": 8.686163247490178,
+      "grad_norm": 2.817579507827759,
+      "learning_rate": 7.2780203784570594e-06,
+      "loss": 0.069,
+      "step": 9950
+    },
+    {
+      "epoch": 8.694893059799215,
+      "grad_norm": 2.3405838012695312,
+      "learning_rate": 7.22950024260068e-06,
+      "loss": 0.083,
+      "step": 9960
+    },
+    {
+      "epoch": 8.70362287210825,
+      "grad_norm": 4.057965278625488,
+      "learning_rate": 7.180980106744299e-06,
+      "loss": 0.1346,
+      "step": 9970
+    },
+    {
+      "epoch": 8.712352684417285,
+      "grad_norm": 3.50754976272583,
+      "learning_rate": 7.1324599708879185e-06,
+      "loss": 0.113,
+      "step": 9980
+    },
+    {
+      "epoch": 8.72108249672632,
+      "grad_norm": 2.4381942749023438,
+      "learning_rate": 7.083939835031539e-06,
+      "loss": 0.0959,
+      "step": 9990
+    },
+    {
+      "epoch": 8.729812309035356,
+      "grad_norm": 1.3491463661193848,
+      "learning_rate": 7.035419699175158e-06,
+      "loss": 0.1141,
+      "step": 10000
+    },
+    {
+      "epoch": 8.738542121344391,
+      "grad_norm": 3.835019588470459,
+      "learning_rate": 6.986899563318778e-06,
+      "loss": 0.0997,
+      "step": 10010
+    },
+    {
+      "epoch": 8.747271933653426,
+      "grad_norm": 1.8673381805419922,
+      "learning_rate": 6.938379427462398e-06,
+      "loss": 0.1044,
+      "step": 10020
+    },
+    {
+      "epoch": 8.756001745962461,
+      "grad_norm": 3.303786039352417,
+      "learning_rate": 6.889859291606017e-06,
+      "loss": 0.1368,
+      "step": 10030
+    },
+    {
+      "epoch": 8.764731558271498,
+      "grad_norm": 2.6152758598327637,
+      "learning_rate": 6.841339155749636e-06,
+      "loss": 0.1186,
+      "step": 10040
+    },
+    {
+      "epoch": 8.773461370580533,
+      "grad_norm": 3.460881471633911,
+      "learning_rate": 6.792819019893256e-06,
+      "loss": 0.1263,
+      "step": 10050
+    },
+    {
+      "epoch": 8.782191182889568,
+      "grad_norm": 2.3491194248199463,
+      "learning_rate": 6.744298884036876e-06,
+      "loss": 0.1147,
+      "step": 10060
+    },
+    {
+      "epoch": 8.790920995198602,
+      "grad_norm": 2.6336238384246826,
+      "learning_rate": 6.695778748180495e-06,
+      "loss": 0.1044,
+      "step": 10070
+    },
+    {
+      "epoch": 8.79965080750764,
+      "grad_norm": 2.4493720531463623,
+      "learning_rate": 6.647258612324115e-06,
+      "loss": 0.0914,
+      "step": 10080
+    },
+    {
+      "epoch": 8.808380619816674,
+      "grad_norm": 2.0005953311920166,
+      "learning_rate": 6.598738476467735e-06,
+      "loss": 0.137,
+      "step": 10090
+    },
+    {
+      "epoch": 8.817110432125709,
+      "grad_norm": 2.4096193313598633,
+      "learning_rate": 6.550218340611354e-06,
+      "loss": 0.1235,
+      "step": 10100
+    },
+    {
+      "epoch": 8.825840244434744,
+      "grad_norm": 2.0157690048217773,
+      "learning_rate": 6.501698204754973e-06,
+      "loss": 0.0981,
+      "step": 10110
+    },
+    {
+      "epoch": 8.83457005674378,
+      "grad_norm": 1.8452261686325073,
+      "learning_rate": 6.4531780688985935e-06,
+      "loss": 0.0971,
+      "step": 10120
+    },
+    {
+      "epoch": 8.843299869052816,
+      "grad_norm": 2.7903761863708496,
+      "learning_rate": 6.404657933042213e-06,
+      "loss": 0.0935,
+      "step": 10130
+    },
+    {
+      "epoch": 8.85202968136185,
+      "grad_norm": 2.508514165878296,
+      "learning_rate": 6.356137797185832e-06,
+      "loss": 0.0991,
+      "step": 10140
+    },
+    {
+      "epoch": 8.860759493670885,
+      "grad_norm": 2.214094638824463,
+      "learning_rate": 6.307617661329453e-06,
+      "loss": 0.11,
+      "step": 10150
+    },
+    {
+      "epoch": 8.869489305979922,
+      "grad_norm": 2.2677242755889893,
+      "learning_rate": 6.259097525473072e-06,
+      "loss": 0.1491,
+      "step": 10160
+    },
+    {
+      "epoch": 8.878219118288957,
+      "grad_norm": 1.6787909269332886,
+      "learning_rate": 6.210577389616691e-06,
+      "loss": 0.1153,
+      "step": 10170
+    },
+    {
+      "epoch": 8.886948930597992,
+      "grad_norm": 2.0860350131988525,
+      "learning_rate": 6.162057253760311e-06,
+      "loss": 0.0906,
+      "step": 10180
+    },
+    {
+      "epoch": 8.895678742907027,
+      "grad_norm": 2.6599624156951904,
+      "learning_rate": 6.11353711790393e-06,
+      "loss": 0.1011,
+      "step": 10190
+    },
+    {
+      "epoch": 8.904408555216063,
+      "grad_norm": 2.9594058990478516,
+      "learning_rate": 6.06501698204755e-06,
+      "loss": 0.1179,
+      "step": 10200
+    },
+    {
+      "epoch": 8.913138367525098,
+      "grad_norm": 2.69802188873291,
+      "learning_rate": 6.01649684619117e-06,
+      "loss": 0.1086,
+      "step": 10210
+    },
+    {
+      "epoch": 8.921868179834133,
+      "grad_norm": 2.7719838619232178,
+      "learning_rate": 5.967976710334789e-06,
+      "loss": 0.1064,
+      "step": 10220
+    },
+    {
+      "epoch": 8.930597992143168,
+      "grad_norm": 2.729365587234497,
+      "learning_rate": 5.9194565744784085e-06,
+      "loss": 0.108,
+      "step": 10230
+    },
+    {
+      "epoch": 8.939327804452205,
+      "grad_norm": 3.4504830837249756,
+      "learning_rate": 5.870936438622028e-06,
+      "loss": 0.095,
+      "step": 10240
+    },
+    {
+      "epoch": 8.94805761676124,
+      "grad_norm": 2.7254908084869385,
+      "learning_rate": 5.822416302765648e-06,
+      "loss": 0.0973,
+      "step": 10250
+    },
+    {
+      "epoch": 8.956787429070275,
+      "grad_norm": 2.1603481769561768,
+      "learning_rate": 5.773896166909268e-06,
+      "loss": 0.0989,
+      "step": 10260
+    },
+    {
+      "epoch": 8.96551724137931,
+      "grad_norm": 3.3005552291870117,
+      "learning_rate": 5.725376031052887e-06,
+      "loss": 0.0931,
+      "step": 10270
+    },
+    {
+      "epoch": 8.974247053688346,
+      "grad_norm": 2.790048122406006,
+      "learning_rate": 5.676855895196507e-06,
+      "loss": 0.1102,
+      "step": 10280
+    },
+    {
+      "epoch": 8.982976865997381,
+      "grad_norm": 2.8936121463775635,
+      "learning_rate": 5.628335759340127e-06,
+      "loss": 0.1204,
+      "step": 10290
+    },
+    {
+      "epoch": 8.991706678306416,
+      "grad_norm": 3.3234341144561768,
+      "learning_rate": 5.5798156234837455e-06,
+      "loss": 0.1212,
+      "step": 10300
+    },
+    {
+      "epoch": 8.999563509384549,
+      "eval_accuracy": 0.9607169162779278,
+      "eval_loss": 0.12570950388908386,
+      "eval_runtime": 61.7961,
+      "eval_samples_per_second": 263.641,
+      "eval_steps_per_second": 8.253,
+      "step": 10309
+    },
+    {
+      "epoch": 9.000436490615451,
+      "grad_norm": 4.348520755767822,
+      "learning_rate": 5.531295487627366e-06,
+      "loss": 0.132,
+      "step": 10310
+    },
+    {
+      "epoch": 9.009166302924488,
+      "grad_norm": 2.5040526390075684,
+      "learning_rate": 5.482775351770985e-06,
+      "loss": 0.116,
+      "step": 10320
+    },
+    {
+      "epoch": 9.017896115233523,
+      "grad_norm": 2.754953622817993,
+      "learning_rate": 5.434255215914605e-06,
+      "loss": 0.1035,
+      "step": 10330
+    },
+    {
+      "epoch": 9.026625927542558,
+      "grad_norm": 3.0102930068969727,
+      "learning_rate": 5.385735080058224e-06,
+      "loss": 0.1104,
+      "step": 10340
+    },
+    {
+      "epoch": 9.035355739851592,
+      "grad_norm": 2.0556371212005615,
+      "learning_rate": 5.337214944201844e-06,
+      "loss": 0.1387,
+      "step": 10350
+    },
+    {
+      "epoch": 9.04408555216063,
+      "grad_norm": 2.688657760620117,
+      "learning_rate": 5.288694808345464e-06,
+      "loss": 0.1082,
+      "step": 10360
+    },
+    {
+      "epoch": 9.052815364469664,
+      "grad_norm": 3.8439743518829346,
+      "learning_rate": 5.240174672489083e-06,
+      "loss": 0.1144,
+      "step": 10370
+    },
+    {
+      "epoch": 9.061545176778699,
+      "grad_norm": 2.683326482772827,
+      "learning_rate": 5.191654536632703e-06,
+      "loss": 0.1075,
+      "step": 10380
+    },
+    {
+      "epoch": 9.070274989087734,
+      "grad_norm": 2.6530981063842773,
+      "learning_rate": 5.143134400776323e-06,
+      "loss": 0.0948,
+      "step": 10390
+    },
+    {
+      "epoch": 9.07900480139677,
+      "grad_norm": 3.5326459407806396,
+      "learning_rate": 5.094614264919942e-06,
+      "loss": 0.1101,
+      "step": 10400
+    },
+    {
+      "epoch": 9.087734613705805,
+      "grad_norm": 3.0565762519836426,
+      "learning_rate": 5.046094129063561e-06,
+      "loss": 0.103,
+      "step": 10410
+    },
+    {
+      "epoch": 9.09646442601484,
+      "grad_norm": 2.949897050857544,
+      "learning_rate": 4.997573993207181e-06,
+      "loss": 0.136,
+      "step": 10420
+    },
+    {
+      "epoch": 9.105194238323875,
+      "grad_norm": 2.1031014919281006,
+      "learning_rate": 4.949053857350801e-06,
+      "loss": 0.117,
+      "step": 10430
+    },
+    {
+      "epoch": 9.113924050632912,
+      "grad_norm": 2.6953930854797363,
+      "learning_rate": 4.9005337214944205e-06,
+      "loss": 0.0944,
+      "step": 10440
+    },
+    {
+      "epoch": 9.122653862941947,
+      "grad_norm": 2.513293981552124,
+      "learning_rate": 4.85201358563804e-06,
+      "loss": 0.1121,
+      "step": 10450
+    },
+    {
+      "epoch": 9.131383675250982,
+      "grad_norm": 2.8910365104675293,
+      "learning_rate": 4.80349344978166e-06,
+      "loss": 0.0821,
+      "step": 10460
+    },
+    {
+      "epoch": 9.140113487560017,
+      "grad_norm": 2.991837978363037,
+      "learning_rate": 4.754973313925279e-06,
+      "loss": 0.115,
+      "step": 10470
+    },
+    {
+      "epoch": 9.148843299869053,
+      "grad_norm": 3.1621274948120117,
+      "learning_rate": 4.7064531780688984e-06,
+      "loss": 0.0987,
+      "step": 10480
+    },
+    {
+      "epoch": 9.157573112178088,
+      "grad_norm": 2.6590192317962646,
+      "learning_rate": 4.657933042212519e-06,
+      "loss": 0.0924,
+      "step": 10490
+    },
+    {
+      "epoch": 9.166302924487123,
+      "grad_norm": 1.3833907842636108,
+      "learning_rate": 4.609412906356138e-06,
+      "loss": 0.0772,
+      "step": 10500
+    },
+    {
+      "epoch": 9.175032736796158,
+      "grad_norm": 2.954089879989624,
+      "learning_rate": 4.5608927704997575e-06,
+      "loss": 0.1372,
+      "step": 10510
+    },
+    {
+      "epoch": 9.183762549105195,
+      "grad_norm": 2.8369104862213135,
+      "learning_rate": 4.512372634643377e-06,
+      "loss": 0.1054,
+      "step": 10520
+    },
+    {
+      "epoch": 9.19249236141423,
+      "grad_norm": 1.4422260522842407,
+      "learning_rate": 4.463852498786997e-06,
+      "loss": 0.1025,
+      "step": 10530
+    },
+    {
+      "epoch": 9.201222173723265,
+      "grad_norm": 1.9895925521850586,
+      "learning_rate": 4.415332362930617e-06,
+      "loss": 0.0992,
+      "step": 10540
+    },
+    {
+      "epoch": 9.2099519860323,
+      "grad_norm": 2.9992966651916504,
+      "learning_rate": 4.3668122270742355e-06,
+      "loss": 0.0927,
+      "step": 10550
+    },
+    {
+      "epoch": 9.218681798341336,
+      "grad_norm": 2.2719290256500244,
+      "learning_rate": 4.318292091217856e-06,
+      "loss": 0.0987,
+      "step": 10560
+    },
+    {
+      "epoch": 9.227411610650371,
+      "grad_norm": 3.9194726943969727,
+      "learning_rate": 4.269771955361476e-06,
+      "loss": 0.0936,
+      "step": 10570
+    },
+    {
+      "epoch": 9.236141422959406,
+      "grad_norm": 2.4642674922943115,
+      "learning_rate": 4.221251819505095e-06,
+      "loss": 0.1074,
+      "step": 10580
+    },
+    {
+      "epoch": 9.244871235268441,
+      "grad_norm": 3.479707956314087,
+      "learning_rate": 4.172731683648714e-06,
+      "loss": 0.1051,
+      "step": 10590
+    },
+    {
+      "epoch": 9.253601047577478,
+      "grad_norm": 3.2341325283050537,
+      "learning_rate": 4.124211547792334e-06,
+      "loss": 0.0878,
+      "step": 10600
+    },
+    {
+      "epoch": 9.262330859886513,
+      "grad_norm": 2.302276849746704,
+      "learning_rate": 4.075691411935954e-06,
+      "loss": 0.1036,
+      "step": 10610
+    },
+    {
+      "epoch": 9.271060672195548,
+      "grad_norm": 2.9073214530944824,
+      "learning_rate": 4.027171276079573e-06,
+      "loss": 0.1181,
+      "step": 10620
+    },
+    {
+      "epoch": 9.279790484504582,
+      "grad_norm": 2.0145514011383057,
+      "learning_rate": 3.978651140223193e-06,
+      "loss": 0.0956,
+      "step": 10630
+    },
+    {
+      "epoch": 9.28852029681362,
+      "grad_norm": 4.683056831359863,
+      "learning_rate": 3.930131004366813e-06,
+      "loss": 0.138,
+      "step": 10640
+    },
+    {
+      "epoch": 9.297250109122654,
+      "grad_norm": 2.681694507598877,
+      "learning_rate": 3.881610868510432e-06,
+      "loss": 0.08,
+      "step": 10650
+    },
+    {
+      "epoch": 9.305979921431689,
+      "grad_norm": 2.887195348739624,
+      "learning_rate": 3.833090732654051e-06,
+      "loss": 0.1113,
+      "step": 10660
+    },
+    {
+      "epoch": 9.314709733740724,
+      "grad_norm": 2.9222280979156494,
+      "learning_rate": 3.7845705967976715e-06,
+      "loss": 0.1074,
+      "step": 10670
+    },
+    {
+      "epoch": 9.32343954604976,
+      "grad_norm": 2.944333076477051,
+      "learning_rate": 3.7360504609412907e-06,
+      "loss": 0.106,
+      "step": 10680
+    },
+    {
+      "epoch": 9.332169358358795,
+      "grad_norm": 2.468632459640503,
+      "learning_rate": 3.6875303250849104e-06,
+      "loss": 0.1049,
+      "step": 10690
+    },
+    {
+      "epoch": 9.34089917066783,
+      "grad_norm": 0.9164177775382996,
+      "learning_rate": 3.6390101892285297e-06,
+      "loss": 0.0718,
+      "step": 10700
+    },
+    {
+      "epoch": 9.349628982976865,
+      "grad_norm": 2.6750760078430176,
+      "learning_rate": 3.5904900533721494e-06,
+      "loss": 0.1086,
+      "step": 10710
+    },
+    {
+      "epoch": 9.358358795285902,
+      "grad_norm": 1.8694539070129395,
+      "learning_rate": 3.5419699175157695e-06,
+      "loss": 0.1099,
+      "step": 10720
+    },
+    {
+      "epoch": 9.367088607594937,
+      "grad_norm": 2.571378707885742,
+      "learning_rate": 3.493449781659389e-06,
+      "loss": 0.0856,
+      "step": 10730
+    },
+    {
+      "epoch": 9.375818419903972,
+      "grad_norm": 1.7155669927597046,
+      "learning_rate": 3.4449296458030085e-06,
+      "loss": 0.1044,
+      "step": 10740
+    },
+    {
+      "epoch": 9.384548232213007,
+      "grad_norm": 2.1449294090270996,
+      "learning_rate": 3.396409509946628e-06,
+      "loss": 0.096,
+      "step": 10750
+    },
+    {
+      "epoch": 9.393278044522043,
+      "grad_norm": 2.918898344039917,
+      "learning_rate": 3.3478893740902475e-06,
+      "loss": 0.0983,
+      "step": 10760
+    },
+    {
+      "epoch": 9.402007856831078,
+      "grad_norm": 2.2519733905792236,
+      "learning_rate": 3.2993692382338676e-06,
+      "loss": 0.1027,
+      "step": 10770
+    },
+    {
+      "epoch": 9.410737669140113,
+      "grad_norm": 4.343425750732422,
+      "learning_rate": 3.2508491023774865e-06,
+      "loss": 0.1118,
+      "step": 10780
+    },
+    {
+      "epoch": 9.419467481449148,
+      "grad_norm": 2.5257129669189453,
+      "learning_rate": 3.2023289665211066e-06,
+      "loss": 0.1093,
+      "step": 10790
+    },
+    {
+      "epoch": 9.428197293758185,
+      "grad_norm": 1.197124719619751,
+      "learning_rate": 3.1538088306647263e-06,
+      "loss": 0.1083,
+      "step": 10800
+    },
+    {
+      "epoch": 9.43692710606722,
+      "grad_norm": 3.033644437789917,
+      "learning_rate": 3.1052886948083456e-06,
+      "loss": 0.1076,
+      "step": 10810
+    },
+    {
+      "epoch": 9.445656918376255,
+      "grad_norm": 2.3595004081726074,
+      "learning_rate": 3.056768558951965e-06,
+      "loss": 0.0772,
+      "step": 10820
+    },
+    {
+      "epoch": 9.45438673068529,
+      "grad_norm": 1.9762500524520874,
+      "learning_rate": 3.008248423095585e-06,
+      "loss": 0.1033,
+      "step": 10830
+    },
+    {
+      "epoch": 9.463116542994326,
+      "grad_norm": 3.2624242305755615,
+      "learning_rate": 2.9597282872392042e-06,
+      "loss": 0.1018,
+      "step": 10840
+    },
+    {
+      "epoch": 9.471846355303361,
+      "grad_norm": 2.5619966983795166,
+      "learning_rate": 2.911208151382824e-06,
+      "loss": 0.1084,
+      "step": 10850
+    },
+    {
+      "epoch": 9.480576167612396,
+      "grad_norm": 2.26334547996521,
+      "learning_rate": 2.8626880155264436e-06,
+      "loss": 0.099,
+      "step": 10860
+    },
+    {
+      "epoch": 9.489305979921431,
+      "grad_norm": 2.397780656814575,
+      "learning_rate": 2.8141678796700633e-06,
+      "loss": 0.1217,
+      "step": 10870
+    },
+    {
+      "epoch": 9.498035792230468,
+      "grad_norm": 3.3402929306030273,
+      "learning_rate": 2.765647743813683e-06,
+      "loss": 0.1199,
+      "step": 10880
+    },
+    {
+      "epoch": 9.506765604539503,
+      "grad_norm": 3.3935964107513428,
+      "learning_rate": 2.7171276079573023e-06,
+      "loss": 0.1158,
+      "step": 10890
+    },
+    {
+      "epoch": 9.515495416848538,
+      "grad_norm": 2.4392571449279785,
+      "learning_rate": 2.668607472100922e-06,
+      "loss": 0.1306,
+      "step": 10900
+    },
+    {
+      "epoch": 9.524225229157572,
+      "grad_norm": 2.4739038944244385,
+      "learning_rate": 2.6200873362445413e-06,
+      "loss": 0.0886,
+      "step": 10910
+    },
+    {
+      "epoch": 9.532955041466609,
+      "grad_norm": 2.2183120250701904,
+      "learning_rate": 2.5715672003881614e-06,
+      "loss": 0.1015,
+      "step": 10920
+    },
+    {
+      "epoch": 9.541684853775644,
+      "grad_norm": 3.256857395172119,
+      "learning_rate": 2.5230470645317807e-06,
+      "loss": 0.1328,
+      "step": 10930
+    },
+    {
+      "epoch": 9.550414666084679,
+      "grad_norm": 2.740285634994507,
+      "learning_rate": 2.4745269286754004e-06,
+      "loss": 0.0996,
+      "step": 10940
+    },
+    {
+      "epoch": 9.559144478393714,
+      "grad_norm": 2.5595703125,
+      "learning_rate": 2.42600679281902e-06,
+      "loss": 0.075,
+      "step": 10950
+    },
+    {
+      "epoch": 9.56787429070275,
+      "grad_norm": 3.089890480041504,
+      "learning_rate": 2.3774866569626394e-06,
+      "loss": 0.1229,
+      "step": 10960
+    },
+    {
+      "epoch": 9.576604103011785,
+      "grad_norm": 3.369868755340576,
+      "learning_rate": 2.3289665211062595e-06,
+      "loss": 0.1194,
+      "step": 10970
+    },
+    {
+      "epoch": 9.58533391532082,
+      "grad_norm": 2.359818458557129,
+      "learning_rate": 2.2804463852498788e-06,
+      "loss": 0.107,
+      "step": 10980
+    },
+    {
+      "epoch": 9.594063727629855,
+      "grad_norm": 2.449138641357422,
+      "learning_rate": 2.2319262493934985e-06,
+      "loss": 0.1078,
+      "step": 10990
+    },
+    {
+      "epoch": 9.602793539938892,
+      "grad_norm": 1.8725260496139526,
+      "learning_rate": 2.1834061135371177e-06,
+      "loss": 0.0882,
+      "step": 11000
+    },
+    {
+      "epoch": 9.611523352247927,
+      "grad_norm": 1.452510952949524,
+      "learning_rate": 2.134885977680738e-06,
+      "loss": 0.1029,
+      "step": 11010
+    },
+    {
+      "epoch": 9.620253164556962,
+      "grad_norm": 4.510222911834717,
+      "learning_rate": 2.086365841824357e-06,
+      "loss": 0.1253,
+      "step": 11020
+    },
+    {
+      "epoch": 9.628982976865997,
+      "grad_norm": 2.815927267074585,
+      "learning_rate": 2.037845705967977e-06,
+      "loss": 0.1132,
+      "step": 11030
+    },
+    {
+      "epoch": 9.637712789175033,
+      "grad_norm": 3.04374098777771,
+      "learning_rate": 1.9893255701115965e-06,
+      "loss": 0.1043,
+      "step": 11040
+    },
+    {
+      "epoch": 9.646442601484068,
+      "grad_norm": 3.407210111618042,
+      "learning_rate": 1.940805434255216e-06,
+      "loss": 0.1207,
+      "step": 11050
+    },
+    {
+      "epoch": 9.655172413793103,
+      "grad_norm": 3.3913309574127197,
+      "learning_rate": 1.8922852983988357e-06,
+      "loss": 0.1153,
+      "step": 11060
+    },
+    {
+      "epoch": 9.663902226102138,
+      "grad_norm": 1.3927017450332642,
+      "learning_rate": 1.8437651625424552e-06,
+      "loss": 0.0865,
+      "step": 11070
+    },
+    {
+      "epoch": 9.672632038411175,
+      "grad_norm": 1.9914878606796265,
+      "learning_rate": 1.7952450266860747e-06,
+      "loss": 0.1071,
+      "step": 11080
+    },
+    {
+      "epoch": 9.68136185072021,
+      "grad_norm": 2.445969820022583,
+      "learning_rate": 1.7467248908296944e-06,
+      "loss": 0.0921,
+      "step": 11090
+    },
+    {
+      "epoch": 9.690091663029245,
+      "grad_norm": 3.2452139854431152,
+      "learning_rate": 1.698204754973314e-06,
+      "loss": 0.1383,
+      "step": 11100
+    },
+    {
+      "epoch": 9.69882147533828,
+      "grad_norm": 1.981126070022583,
+      "learning_rate": 1.6496846191169338e-06,
+      "loss": 0.1139,
+      "step": 11110
+    },
+    {
+      "epoch": 9.707551287647316,
+      "grad_norm": 1.5634502172470093,
+      "learning_rate": 1.6011644832605533e-06,
+      "loss": 0.0922,
+      "step": 11120
+    },
+    {
+      "epoch": 9.716281099956351,
+      "grad_norm": 3.633537769317627,
+      "learning_rate": 1.5526443474041728e-06,
+      "loss": 0.107,
+      "step": 11130
+    },
+    {
+      "epoch": 9.725010912265386,
+      "grad_norm": 2.735826015472412,
+      "learning_rate": 1.5041242115477925e-06,
+      "loss": 0.1027,
+      "step": 11140
+    },
+    {
+      "epoch": 9.733740724574421,
+      "grad_norm": 2.9639649391174316,
+      "learning_rate": 1.455604075691412e-06,
+      "loss": 0.0849,
+      "step": 11150
+    },
+    {
+      "epoch": 9.742470536883458,
+      "grad_norm": 2.332066297531128,
+      "learning_rate": 1.4070839398350317e-06,
+      "loss": 0.1055,
+      "step": 11160
+    },
+    {
+      "epoch": 9.751200349192493,
+      "grad_norm": 2.8028910160064697,
+      "learning_rate": 1.3585638039786512e-06,
+      "loss": 0.1135,
+      "step": 11170
+    },
+    {
+      "epoch": 9.759930161501527,
+      "grad_norm": 3.3625102043151855,
+      "learning_rate": 1.3100436681222706e-06,
+      "loss": 0.1219,
+      "step": 11180
+    },
+    {
+      "epoch": 9.768659973810562,
+      "grad_norm": 3.788644790649414,
+      "learning_rate": 1.2615235322658903e-06,
+      "loss": 0.1035,
+      "step": 11190
+    },
+    {
+      "epoch": 9.777389786119599,
+      "grad_norm": 3.0981266498565674,
+      "learning_rate": 1.21300339640951e-06,
+      "loss": 0.0964,
+      "step": 11200
+    },
+    {
+      "epoch": 9.786119598428634,
+      "grad_norm": 2.1601572036743164,
+      "learning_rate": 1.1644832605531297e-06,
+      "loss": 0.09,
+      "step": 11210
+    },
+    {
+      "epoch": 9.794849410737669,
+      "grad_norm": 2.264108419418335,
+      "learning_rate": 1.1159631246967492e-06,
+      "loss": 0.0925,
+      "step": 11220
+    },
+    {
+      "epoch": 9.803579223046704,
+      "grad_norm": 4.96955680847168,
+      "learning_rate": 1.067442988840369e-06,
+      "loss": 0.1429,
+      "step": 11230
+    },
+    {
+      "epoch": 9.81230903535574,
+      "grad_norm": 2.7126574516296387,
+      "learning_rate": 1.0189228529839884e-06,
+      "loss": 0.1139,
+      "step": 11240
+    },
+    {
+      "epoch": 9.821038847664775,
+      "grad_norm": 2.6963400840759277,
+      "learning_rate": 9.70402717127608e-07,
+      "loss": 0.1106,
+      "step": 11250
+    },
+    {
+      "epoch": 9.82976865997381,
+      "grad_norm": 2.730841636657715,
+      "learning_rate": 9.218825812712276e-07,
+      "loss": 0.1031,
+      "step": 11260
+    },
+    {
+      "epoch": 9.838498472282845,
+      "grad_norm": 3.486825704574585,
+      "learning_rate": 8.733624454148472e-07,
+      "loss": 0.1023,
+      "step": 11270
+    },
+    {
+      "epoch": 9.847228284591882,
+      "grad_norm": 3.1530649662017822,
+      "learning_rate": 8.248423095584669e-07,
+      "loss": 0.084,
+      "step": 11280
+    },
+    {
+      "epoch": 9.855958096900917,
+      "grad_norm": 1.514291524887085,
+      "learning_rate": 7.763221737020864e-07,
+      "loss": 0.1088,
+      "step": 11290
+    },
+    {
+      "epoch": 9.864687909209952,
+      "grad_norm": 1.9659836292266846,
+      "learning_rate": 7.27802037845706e-07,
+      "loss": 0.0722,
+      "step": 11300
+    },
+    {
+      "epoch": 9.873417721518987,
+      "grad_norm": 2.1583216190338135,
+      "learning_rate": 6.792819019893256e-07,
+      "loss": 0.1086,
+      "step": 11310
+    },
+    {
+      "epoch": 9.882147533828023,
+      "grad_norm": 3.034625291824341,
+      "learning_rate": 6.307617661329452e-07,
+      "loss": 0.1046,
+      "step": 11320
+    },
+    {
+      "epoch": 9.890877346137058,
+      "grad_norm": 2.7435011863708496,
+      "learning_rate": 5.822416302765649e-07,
+      "loss": 0.1083,
+      "step": 11330
+    },
+    {
+      "epoch": 9.899607158446093,
+      "grad_norm": 2.3298757076263428,
+      "learning_rate": 5.337214944201845e-07,
+      "loss": 0.11,
+      "step": 11340
+    },
+    {
+      "epoch": 9.908336970755128,
+      "grad_norm": 3.238316774368286,
+      "learning_rate": 4.85201358563804e-07,
+      "loss": 0.1281,
+      "step": 11350
+    },
+    {
+      "epoch": 9.917066783064165,
+      "grad_norm": 2.9456119537353516,
+      "learning_rate": 4.366812227074236e-07,
+      "loss": 0.0932,
+      "step": 11360
+    },
+    {
+      "epoch": 9.9257965953732,
+      "grad_norm": 1.8803859949111938,
+      "learning_rate": 3.881610868510432e-07,
+      "loss": 0.1022,
+      "step": 11370
+    },
+    {
+      "epoch": 9.934526407682235,
+      "grad_norm": 2.5518546104431152,
+      "learning_rate": 3.396409509946628e-07,
+      "loss": 0.0876,
+      "step": 11380
+    },
+    {
+      "epoch": 9.94325621999127,
+      "grad_norm": 2.5071511268615723,
+      "learning_rate": 2.9112081513828244e-07,
+      "loss": 0.1075,
+      "step": 11390
+    },
+    {
+      "epoch": 9.951986032300306,
+      "grad_norm": 4.196099758148193,
+      "learning_rate": 2.42600679281902e-07,
+      "loss": 0.1197,
+      "step": 11400
+    },
+    {
+      "epoch": 9.960715844609341,
+      "grad_norm": 2.428870916366577,
+      "learning_rate": 1.940805434255216e-07,
+      "loss": 0.1262,
+      "step": 11410
+    },
+    {
+      "epoch": 9.969445656918376,
+      "grad_norm": 2.2001659870147705,
+      "learning_rate": 1.4556040756914122e-07,
+      "loss": 0.1013,
+      "step": 11420
+    },
+    {
+      "epoch": 9.978175469227411,
+      "grad_norm": 3.1071674823760986,
+      "learning_rate": 9.70402717127608e-08,
+      "loss": 0.0883,
+      "step": 11430
+    },
+    {
+      "epoch": 9.986905281536448,
+      "grad_norm": 2.7770655155181885,
+      "learning_rate": 4.85201358563804e-08,
+      "loss": 0.1187,
+      "step": 11440
+    },
+    {
+      "epoch": 9.995635093845483,
+      "grad_norm": 2.909649133682251,
+      "learning_rate": 0.0,
+      "loss": 0.0981,
+      "step": 11450
+    },
+    {
+      "epoch": 9.995635093845483,
+      "eval_accuracy": 0.9750797937638105,
+      "eval_loss": 0.07497124373912811,
+      "eval_runtime": 61.2653,
+      "eval_samples_per_second": 265.925,
+      "eval_steps_per_second": 8.324,
+      "step": 11450
+    },
+    {
+      "epoch": 9.995635093845483,
+      "step": 11450,
+      "total_flos": 3.92899376185344e+18,
+      "train_loss": 0.3148357209992721,
+      "train_runtime": 13104.719,
+      "train_samples_per_second": 111.886,
+      "train_steps_per_second": 0.874
     }
   ],
   "logging_steps": 10,
-  "max_steps": 2900,
+  "max_steps": 11450,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 20,
+  "num_train_epochs": 10,
   "save_steps": 500,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -2245,7 +8140,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 9.813099581032366e+17,
+  "total_flos": 3.92899376185344e+18,
   "train_batch_size": 32,
   "trial_name": null,
   "trial_params": null