diff --git "a/checkpoint-60000/trainer_state.json" "b/checkpoint-60000/trainer_state.json"
--- "a/checkpoint-60000/trainer_state.json"
+++ "b/checkpoint-60000/trainer_state.json"
@@ -1,7 +1,7 @@
 {
-  "best_metric": 3.3706517219543457,
-  "best_model_checkpoint": "/scratch/cl5625/exceptions/models/100M_495/checkpoint-60000",
-  "epoch": 6.457862447529868,
+  "best_metric": 3.3771378993988037,
+  "best_model_checkpoint": "/scratch/cl5625/exceptions/models/100M_high_10_495/checkpoint-60000",
+  "epoch": 6.46900269541779,
   "eval_steps": 1000,
   "global_step": 60000,
   "is_hyper_param_search": false,
@@ -9,8948 +9,8948 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.005381552039608223,
-      "grad_norm": 3.403714418411255,
-      "learning_rate": 0.00028799999999999995,
-      "loss": 8.9829,
+      "epoch": 0.005390835579514825,
+      "grad_norm": 0.9467681646347046,
+      "learning_rate": 0.000276,
+      "loss": 8.7571,
       "step": 50
     },
     {
-      "epoch": 0.010763104079216447,
-      "grad_norm": 3.788738965988159,
-      "learning_rate": 0.000588,
-      "loss": 6.9704,
+      "epoch": 0.01078167115902965,
+      "grad_norm": 3.062793254852295,
+      "learning_rate": 0.0005759999999999999,
+      "loss": 6.9857,
       "step": 100
     },
     {
-      "epoch": 0.01614465611882467,
-      "grad_norm": 1.4487857818603516,
-      "learning_rate": 0.000599689688611141,
-      "loss": 6.4875,
+      "epoch": 0.016172506738544475,
+      "grad_norm": 2.382413148880005,
+      "learning_rate": 0.000599702104695089,
+      "loss": 6.597,
       "step": 150
     },
     {
-      "epoch": 0.021526208158432893,
-      "grad_norm": 2.067326068878174,
-      "learning_rate": 0.0005993664475810796,
-      "loss": 6.2749,
+      "epoch": 0.0215633423180593,
+      "grad_norm": 0.9547237753868103,
+      "learning_rate": 0.0005993783054506205,
+      "loss": 6.3206,
       "step": 200
     },
     {
-      "epoch": 0.026907760198041114,
-      "grad_norm": 1.0104901790618896,
-      "learning_rate": 0.0005990432065510182,
-      "loss": 6.0719,
+      "epoch": 0.026954177897574125,
+      "grad_norm": 1.4377750158309937,
+      "learning_rate": 0.0005990545062061521,
+      "loss": 6.1209,
       "step": 250
     },
     {
-      "epoch": 0.03228931223764934,
-      "grad_norm": 1.5355337858200073,
-      "learning_rate": 0.0005987199655209567,
-      "loss": 5.9689,
+      "epoch": 0.03234501347708895,
+      "grad_norm": 1.6764158010482788,
+      "learning_rate": 0.0005987307069616836,
+      "loss": 6.003,
       "step": 300
     },
     {
-      "epoch": 0.03767086427725756,
-      "grad_norm": 1.7764770984649658,
-      "learning_rate": 0.0005983967244908953,
-      "loss": 5.8883,
+      "epoch": 0.03773584905660377,
+      "grad_norm": 0.94362473487854,
+      "learning_rate": 0.0005984069077172153,
+      "loss": 5.9015,
       "step": 350
     },
     {
-      "epoch": 0.04305241631686579,
-      "grad_norm": 1.896673321723938,
-      "learning_rate": 0.0005980734834608338,
-      "loss": 5.7905,
+      "epoch": 0.0431266846361186,
+      "grad_norm": 1.4955400228500366,
+      "learning_rate": 0.0005980831084727469,
+      "loss": 5.8417,
       "step": 400
     },
     {
-      "epoch": 0.048433968356474004,
-      "grad_norm": 1.7309101819992065,
-      "learning_rate": 0.0005977502424307725,
-      "loss": 5.7165,
+      "epoch": 0.04851752021563342,
+      "grad_norm": 0.8737039566040039,
+      "learning_rate": 0.0005977593092282784,
+      "loss": 5.7544,
       "step": 450
     },
     {
-      "epoch": 0.05381552039608223,
-      "grad_norm": 1.5152183771133423,
-      "learning_rate": 0.0005974270014007111,
-      "loss": 5.6356,
+      "epoch": 0.05390835579514825,
+      "grad_norm": 1.6916964054107666,
+      "learning_rate": 0.00059743550998381,
+      "loss": 5.7004,
       "step": 500
     },
     {
-      "epoch": 0.05919707243569045,
-      "grad_norm": 1.2986925840377808,
-      "learning_rate": 0.0005971037603706497,
-      "loss": 5.5617,
+      "epoch": 0.05929919137466307,
+      "grad_norm": 1.3075671195983887,
+      "learning_rate": 0.0005971117107393416,
+      "loss": 5.5902,
       "step": 550
     },
     {
-      "epoch": 0.06457862447529868,
-      "grad_norm": 2.1971282958984375,
-      "learning_rate": 0.0005967805193405882,
-      "loss": 5.516,
+      "epoch": 0.0646900269541779,
+      "grad_norm": 1.5872267484664917,
+      "learning_rate": 0.0005967879114948732,
+      "loss": 5.5606,
       "step": 600
     },
     {
-      "epoch": 0.0699601765149069,
-      "grad_norm": 1.4911054372787476,
-      "learning_rate": 0.0005964572783105269,
-      "loss": 5.4241,
+      "epoch": 0.07008086253369272,
+      "grad_norm": 1.3364683389663696,
+      "learning_rate": 0.0005964641122504047,
+      "loss": 5.4624,
       "step": 650
     },
     {
-      "epoch": 0.07534172855451512,
-      "grad_norm": 1.4151604175567627,
-      "learning_rate": 0.0005961340372804654,
-      "loss": 5.362,
+      "epoch": 0.07547169811320754,
+      "grad_norm": 1.1322250366210938,
+      "learning_rate": 0.0005961403130059363,
+      "loss": 5.4412,
       "step": 700
     },
     {
-      "epoch": 0.08072328059412334,
-      "grad_norm": 1.035212755203247,
-      "learning_rate": 0.000595810796250404,
-      "loss": 5.3373,
+      "epoch": 0.08086253369272237,
+      "grad_norm": 1.526710867881775,
+      "learning_rate": 0.0005958165137614678,
+      "loss": 5.3674,
       "step": 750
     },
     {
-      "epoch": 0.08610483263373157,
-      "grad_norm": 1.0760645866394043,
-      "learning_rate": 0.0005954875552203426,
-      "loss": 5.2561,
+      "epoch": 0.0862533692722372,
+      "grad_norm": 1.0809352397918701,
+      "learning_rate": 0.0005954927145169995,
+      "loss": 5.3087,
       "step": 800
     },
     {
-      "epoch": 0.09148638467333979,
-      "grad_norm": 1.1470861434936523,
-      "learning_rate": 0.0005951643141902811,
-      "loss": 5.2337,
+      "epoch": 0.09164420485175202,
+      "grad_norm": 1.260084867477417,
+      "learning_rate": 0.0005951689152725309,
+      "loss": 5.2775,
       "step": 850
     },
     {
-      "epoch": 0.09686793671294801,
-      "grad_norm": 1.1675933599472046,
-      "learning_rate": 0.0005948410731602198,
-      "loss": 5.1697,
+      "epoch": 0.09703504043126684,
+      "grad_norm": 1.0708142518997192,
+      "learning_rate": 0.0005948451160280626,
+      "loss": 5.1994,
       "step": 900
     },
     {
-      "epoch": 0.10224948875255624,
-      "grad_norm": 0.6720972061157227,
-      "learning_rate": 0.0005945178321301583,
-      "loss": 5.1131,
+      "epoch": 0.10242587601078167,
+      "grad_norm": 0.828127920627594,
+      "learning_rate": 0.0005945213167835941,
+      "loss": 5.1905,
       "step": 950
     },
     {
-      "epoch": 0.10763104079216446,
-      "grad_norm": 0.8663607835769653,
-      "learning_rate": 0.000594194591100097,
-      "loss": 5.08,
+      "epoch": 0.1078167115902965,
+      "grad_norm": 1.1112550497055054,
+      "learning_rate": 0.0005941975175391257,
+      "loss": 5.134,
       "step": 1000
     },
     {
-      "epoch": 0.10763104079216446,
-      "eval_accuracy": 0.22722828330284353,
-      "eval_loss": 5.01658296585083,
-      "eval_runtime": 185.3477,
-      "eval_samples_per_second": 97.174,
-      "eval_steps_per_second": 6.075,
+      "epoch": 0.1078167115902965,
+      "eval_accuracy": 0.22364556298004884,
+      "eval_loss": 5.057981014251709,
+      "eval_runtime": 155.8033,
+      "eval_samples_per_second": 115.601,
+      "eval_steps_per_second": 7.227,
       "step": 1000
     },
     {
-      "epoch": 0.11301259283177269,
-      "grad_norm": 1.171659231185913,
-      "learning_rate": 0.0005938713500700355,
-      "loss": 5.0623,
+      "epoch": 0.11320754716981132,
+      "grad_norm": 1.0472087860107422,
+      "learning_rate": 0.0005938737182946572,
+      "loss": 5.0945,
       "step": 1050
     },
     {
-      "epoch": 0.1183941448713809,
-      "grad_norm": 1.2680262327194214,
-      "learning_rate": 0.000593548109039974,
-      "loss": 5.0281,
+      "epoch": 0.11859838274932614,
+      "grad_norm": 1.4178184270858765,
+      "learning_rate": 0.0005935499190501888,
+      "loss": 5.0532,
       "step": 1100
     },
     {
-      "epoch": 0.12377569691098914,
-      "grad_norm": 0.9914045929908752,
-      "learning_rate": 0.0005932248680099127,
-      "loss": 4.9721,
+      "epoch": 0.12398921832884097,
+      "grad_norm": 0.9259712100028992,
+      "learning_rate": 0.0005932261198057204,
+      "loss": 5.0331,
       "step": 1150
     },
     {
-      "epoch": 0.12915724895059735,
-      "grad_norm": 1.0515373945236206,
-      "learning_rate": 0.0005929016269798512,
-      "loss": 4.9805,
+      "epoch": 0.1293800539083558,
+      "grad_norm": 0.8550077080726624,
+      "learning_rate": 0.000592902320561252,
+      "loss": 5.0119,
       "step": 1200
     },
     {
-      "epoch": 0.13453880099020557,
-      "grad_norm": 1.0933362245559692,
-      "learning_rate": 0.0005925783859497898,
-      "loss": 4.8982,
+      "epoch": 0.1347708894878706,
+      "grad_norm": 0.8048805594444275,
+      "learning_rate": 0.0005925785213167835,
+      "loss": 4.9853,
       "step": 1250
     },
     {
-      "epoch": 0.1399203530298138,
-      "grad_norm": 1.5771769285202026,
-      "learning_rate": 0.0005922551449197284,
-      "loss": 4.9081,
+      "epoch": 0.14016172506738545,
+      "grad_norm": 1.0447702407836914,
+      "learning_rate": 0.0005922547220723151,
+      "loss": 4.9313,
       "step": 1300
     },
     {
-      "epoch": 0.14530190506942203,
-      "grad_norm": 1.2906471490859985,
-      "learning_rate": 0.0005919319038896671,
-      "loss": 4.8856,
+      "epoch": 0.14555256064690028,
+      "grad_norm": 1.220167636871338,
+      "learning_rate": 0.0005919309228278468,
+      "loss": 4.9171,
       "step": 1350
     },
     {
-      "epoch": 0.15068345710903025,
-      "grad_norm": 1.1471878290176392,
-      "learning_rate": 0.0005916086628596056,
-      "loss": 4.8496,
+      "epoch": 0.1509433962264151,
+      "grad_norm": 0.8236722350120544,
+      "learning_rate": 0.0005916071235833783,
+      "loss": 4.8912,
       "step": 1400
     },
     {
-      "epoch": 0.15606500914863847,
-      "grad_norm": 0.8490638732910156,
-      "learning_rate": 0.0005912854218295442,
-      "loss": 4.8223,
+      "epoch": 0.15633423180592992,
+      "grad_norm": 0.9686391949653625,
+      "learning_rate": 0.0005912833243389097,
+      "loss": 4.8686,
       "step": 1450
     },
     {
-      "epoch": 0.16144656118824668,
-      "grad_norm": 1.2815980911254883,
-      "learning_rate": 0.0005909621807994827,
-      "loss": 4.8234,
+      "epoch": 0.16172506738544473,
+      "grad_norm": 0.928552508354187,
+      "learning_rate": 0.0005909595250944414,
+      "loss": 4.8514,
       "step": 1500
     },
     {
-      "epoch": 0.1668281132278549,
-      "grad_norm": 1.162052035331726,
-      "learning_rate": 0.0005906389397694213,
-      "loss": 4.7904,
+      "epoch": 0.16711590296495957,
+      "grad_norm": 1.1130082607269287,
+      "learning_rate": 0.000590635725849973,
+      "loss": 4.8119,
       "step": 1550
     },
     {
-      "epoch": 0.17220966526746315,
-      "grad_norm": 1.0659350156784058,
-      "learning_rate": 0.00059031569873936,
-      "loss": 4.7667,
+      "epoch": 0.1725067385444744,
+      "grad_norm": 1.0583394765853882,
+      "learning_rate": 0.0005903119266055045,
+      "loss": 4.8314,
       "step": 1600
     },
     {
-      "epoch": 0.17759121730707136,
-      "grad_norm": 0.935391366481781,
-      "learning_rate": 0.0005899924577092985,
-      "loss": 4.734,
+      "epoch": 0.1778975741239892,
+      "grad_norm": 0.912996768951416,
+      "learning_rate": 0.0005899881273610361,
+      "loss": 4.809,
       "step": 1650
     },
     {
-      "epoch": 0.18297276934667958,
-      "grad_norm": 1.0256898403167725,
-      "learning_rate": 0.0005896692166792371,
-      "loss": 4.738,
+      "epoch": 0.18328840970350405,
+      "grad_norm": 0.7502346038818359,
+      "learning_rate": 0.0005896643281165677,
+      "loss": 4.7298,
       "step": 1700
     },
     {
-      "epoch": 0.1883543213862878,
-      "grad_norm": 0.8851535320281982,
-      "learning_rate": 0.0005893459756491757,
-      "loss": 4.6881,
+      "epoch": 0.18867924528301888,
+      "grad_norm": 1.1314946413040161,
+      "learning_rate": 0.0005893405288720993,
+      "loss": 4.7486,
       "step": 1750
     },
     {
-      "epoch": 0.19373587342589602,
-      "grad_norm": 1.0578432083129883,
-      "learning_rate": 0.0005890227346191143,
-      "loss": 4.6711,
+      "epoch": 0.1940700808625337,
+      "grad_norm": 1.0135972499847412,
+      "learning_rate": 0.0005890167296276308,
+      "loss": 4.7355,
       "step": 1800
     },
     {
-      "epoch": 0.19911742546550426,
-      "grad_norm": 1.0169049501419067,
-      "learning_rate": 0.0005886994935890529,
-      "loss": 4.6803,
+      "epoch": 0.19946091644204852,
+      "grad_norm": 0.8325548768043518,
+      "learning_rate": 0.0005886929303831624,
+      "loss": 4.6924,
       "step": 1850
     },
     {
-      "epoch": 0.20449897750511248,
-      "grad_norm": 0.8385932445526123,
-      "learning_rate": 0.0005883762525589915,
-      "loss": 4.6329,
+      "epoch": 0.20485175202156333,
+      "grad_norm": 0.8735172152519226,
+      "learning_rate": 0.0005883691311386939,
+      "loss": 4.6884,
       "step": 1900
     },
     {
-      "epoch": 0.2098805295447207,
-      "grad_norm": 1.224091649055481,
-      "learning_rate": 0.00058805301152893,
-      "loss": 4.6126,
+      "epoch": 0.21024258760107817,
+      "grad_norm": 1.009144902229309,
+      "learning_rate": 0.0005880453318942256,
+      "loss": 4.6563,
       "step": 1950
     },
     {
-      "epoch": 0.2152620815843289,
-      "grad_norm": 1.068516492843628,
-      "learning_rate": 0.0005877297704988686,
-      "loss": 4.6063,
+      "epoch": 0.215633423180593,
+      "grad_norm": 1.0438272953033447,
+      "learning_rate": 0.0005877215326497571,
+      "loss": 4.6368,
       "step": 2000
     },
     {
-      "epoch": 0.2152620815843289,
-      "eval_accuracy": 0.26832732464754905,
-      "eval_loss": 4.5254340171813965,
-      "eval_runtime": 184.072,
-      "eval_samples_per_second": 97.848,
-      "eval_steps_per_second": 6.117,
+      "epoch": 0.215633423180593,
+      "eval_accuracy": 0.26113689383776234,
+      "eval_loss": 4.590283393859863,
+      "eval_runtime": 153.3564,
+      "eval_samples_per_second": 117.445,
+      "eval_steps_per_second": 7.342,
       "step": 2000
     },
     {
-      "epoch": 0.22064363362393713,
-      "grad_norm": 1.1607660055160522,
-      "learning_rate": 0.0005874065294688072,
-      "loss": 4.5794,
+      "epoch": 0.2210242587601078,
+      "grad_norm": 0.8841946721076965,
+      "learning_rate": 0.0005873977334052887,
+      "loss": 4.6585,
       "step": 2050
     },
     {
-      "epoch": 0.22602518566354537,
-      "grad_norm": 0.8364855051040649,
-      "learning_rate": 0.0005870832884387457,
-      "loss": 4.5403,
+      "epoch": 0.22641509433962265,
+      "grad_norm": 0.9768223762512207,
+      "learning_rate": 0.0005870739341608202,
+      "loss": 4.6248,
       "step": 2100
     },
     {
-      "epoch": 0.2314067377031536,
-      "grad_norm": 1.0584214925765991,
-      "learning_rate": 0.0005867600474086844,
-      "loss": 4.5344,
+      "epoch": 0.23180592991913745,
+      "grad_norm": 0.88718181848526,
+      "learning_rate": 0.0005867501349163519,
+      "loss": 4.5645,
       "step": 2150
     },
     {
-      "epoch": 0.2367882897427618,
-      "grad_norm": 0.8201826214790344,
-      "learning_rate": 0.0005864368063786229,
-      "loss": 4.5198,
+      "epoch": 0.2371967654986523,
+      "grad_norm": 0.928945779800415,
+      "learning_rate": 0.0005864263356718833,
+      "loss": 4.6031,
       "step": 2200
     },
     {
-      "epoch": 0.24216984178237003,
-      "grad_norm": 0.8647080063819885,
-      "learning_rate": 0.0005861135653485616,
-      "loss": 4.5024,
+      "epoch": 0.24258760107816713,
+      "grad_norm": 0.9557152986526489,
+      "learning_rate": 0.000586102536427415,
+      "loss": 4.5394,
       "step": 2250
     },
     {
-      "epoch": 0.24755139382197827,
-      "grad_norm": 1.2521132230758667,
-      "learning_rate": 0.0005857903243185001,
-      "loss": 4.4975,
+      "epoch": 0.24797843665768193,
+      "grad_norm": 0.938102126121521,
+      "learning_rate": 0.0005857787371829465,
+      "loss": 4.5208,
       "step": 2300
     },
     {
-      "epoch": 0.2529329458615865,
-      "grad_norm": 0.8756551742553711,
-      "learning_rate": 0.0005854670832884386,
-      "loss": 4.4975,
+      "epoch": 0.25336927223719674,
+      "grad_norm": 0.8805590271949768,
+      "learning_rate": 0.0005854549379384781,
+      "loss": 4.5136,
       "step": 2350
     },
     {
-      "epoch": 0.2583144979011947,
-      "grad_norm": 0.9284951090812683,
-      "learning_rate": 0.0005851438422583773,
-      "loss": 4.4694,
+      "epoch": 0.2587601078167116,
+      "grad_norm": 1.2024765014648438,
+      "learning_rate": 0.0005851311386940096,
+      "loss": 4.5026,
       "step": 2400
     },
     {
-      "epoch": 0.2636960499408029,
-      "grad_norm": 0.8526713252067566,
-      "learning_rate": 0.0005848206012283159,
-      "loss": 4.4355,
+      "epoch": 0.2641509433962264,
+      "grad_norm": 0.9804381132125854,
+      "learning_rate": 0.0005848073394495412,
+      "loss": 4.5039,
       "step": 2450
     },
     {
-      "epoch": 0.26907760198041114,
-      "grad_norm": 0.8831415176391602,
-      "learning_rate": 0.0005844973601982545,
-      "loss": 4.4085,
+      "epoch": 0.2695417789757412,
+      "grad_norm": 0.9215933084487915,
+      "learning_rate": 0.0005844835402050728,
+      "loss": 4.4967,
       "step": 2500
     },
     {
-      "epoch": 0.27445915402001936,
-      "grad_norm": 0.9462632536888123,
-      "learning_rate": 0.000584174119168193,
-      "loss": 4.4267,
+      "epoch": 0.2749326145552561,
+      "grad_norm": 0.9548609852790833,
+      "learning_rate": 0.0005841597409606044,
+      "loss": 4.4551,
       "step": 2550
     },
     {
-      "epoch": 0.2798407060596276,
-      "grad_norm": 0.9993759393692017,
-      "learning_rate": 0.0005838508781381316,
-      "loss": 4.4065,
+      "epoch": 0.2803234501347709,
+      "grad_norm": 0.8881437182426453,
+      "learning_rate": 0.000583835941716136,
+      "loss": 4.4451,
       "step": 2600
     },
     {
-      "epoch": 0.2852222580992358,
-      "grad_norm": 0.8348021507263184,
-      "learning_rate": 0.0005835276371080702,
-      "loss": 4.3843,
+      "epoch": 0.2857142857142857,
+      "grad_norm": 1.0776073932647705,
+      "learning_rate": 0.0005835121424716675,
+      "loss": 4.4423,
       "step": 2650
     },
     {
-      "epoch": 0.29060381013884407,
-      "grad_norm": 0.779861330986023,
-      "learning_rate": 0.0005832043960780088,
-      "loss": 4.389,
+      "epoch": 0.29110512129380056,
+      "grad_norm": 0.9139967560768127,
+      "learning_rate": 0.0005831883432271992,
+      "loss": 4.4323,
       "step": 2700
     },
     {
-      "epoch": 0.2959853621784523,
-      "grad_norm": 0.863554835319519,
-      "learning_rate": 0.0005828811550479474,
-      "loss": 4.3843,
+      "epoch": 0.29649595687331537,
+      "grad_norm": 0.8609856367111206,
+      "learning_rate": 0.0005828645439827307,
+      "loss": 4.3929,
       "step": 2750
     },
     {
-      "epoch": 0.3013669142180605,
-      "grad_norm": 0.8065552115440369,
-      "learning_rate": 0.0005825579140178859,
-      "loss": 4.3862,
+      "epoch": 0.3018867924528302,
+      "grad_norm": 0.8271277546882629,
+      "learning_rate": 0.0005825407447382622,
+      "loss": 4.4317,
       "step": 2800
     },
     {
-      "epoch": 0.3067484662576687,
-      "grad_norm": 0.7346043586730957,
-      "learning_rate": 0.0005822346729878246,
-      "loss": 4.3553,
+      "epoch": 0.30727762803234504,
+      "grad_norm": 0.7900182604789734,
+      "learning_rate": 0.0005822169454937938,
+      "loss": 4.3984,
       "step": 2850
     },
     {
-      "epoch": 0.31213001829727693,
-      "grad_norm": 0.9188678860664368,
-      "learning_rate": 0.0005819114319577631,
-      "loss": 4.3256,
+      "epoch": 0.31266846361185985,
+      "grad_norm": 0.7303901314735413,
+      "learning_rate": 0.0005818931462493254,
+      "loss": 4.3783,
       "step": 2900
     },
     {
-      "epoch": 0.31751157033688515,
-      "grad_norm": 0.705977201461792,
-      "learning_rate": 0.0005815881909277017,
-      "loss": 4.3318,
+      "epoch": 0.31805929919137466,
+      "grad_norm": 0.6963273286819458,
+      "learning_rate": 0.0005815693470048569,
+      "loss": 4.3785,
       "step": 2950
     },
     {
-      "epoch": 0.32289312237649337,
-      "grad_norm": 0.9044803380966187,
-      "learning_rate": 0.0005812649498976403,
-      "loss": 4.3318,
+      "epoch": 0.32345013477088946,
+      "grad_norm": 0.7128702998161316,
+      "learning_rate": 0.0005812455477603885,
+      "loss": 4.3722,
       "step": 3000
     },
     {
-      "epoch": 0.32289312237649337,
-      "eval_accuracy": 0.2972615886725453,
-      "eval_loss": 4.248094081878662,
-      "eval_runtime": 184.1083,
-      "eval_samples_per_second": 97.828,
-      "eval_steps_per_second": 6.116,
+      "epoch": 0.32345013477088946,
+      "eval_accuracy": 0.2934933978702513,
+      "eval_loss": 4.272831916809082,
+      "eval_runtime": 153.1756,
+      "eval_samples_per_second": 117.584,
+      "eval_steps_per_second": 7.351,
       "step": 3000
     },
     {
-      "epoch": 0.3282746744161016,
-      "grad_norm": 0.7348528504371643,
-      "learning_rate": 0.0005809417088675789,
-      "loss": 4.3203,
+      "epoch": 0.3288409703504043,
+      "grad_norm": 0.8196587562561035,
+      "learning_rate": 0.0005809217485159201,
+      "loss": 4.3483,
       "step": 3050
     },
     {
-      "epoch": 0.3336562264557098,
-      "grad_norm": 0.8613377213478088,
-      "learning_rate": 0.0005806184678375175,
-      "loss": 4.3203,
+      "epoch": 0.33423180592991913,
+      "grad_norm": 0.730968177318573,
+      "learning_rate": 0.0005805979492714517,
+      "loss": 4.3439,
       "step": 3100
     },
     {
-      "epoch": 0.3390377784953181,
-      "grad_norm": 0.6706194281578064,
-      "learning_rate": 0.000580295226807456,
-      "loss": 4.2918,
+      "epoch": 0.33962264150943394,
+      "grad_norm": 0.7348554134368896,
+      "learning_rate": 0.0005802741500269832,
+      "loss": 4.307,
       "step": 3150
     },
     {
-      "epoch": 0.3444193305349263,
-      "grad_norm": 0.8804624080657959,
-      "learning_rate": 0.0005799719857773946,
-      "loss": 4.2862,
+      "epoch": 0.3450134770889488,
+      "grad_norm": 0.7481687068939209,
+      "learning_rate": 0.0005799503507825148,
+      "loss": 4.3136,
       "step": 3200
     },
     {
-      "epoch": 0.3498008825745345,
-      "grad_norm": 0.6948880553245544,
-      "learning_rate": 0.0005796487447473331,
-      "loss": 4.2839,
+      "epoch": 0.3504043126684636,
+      "grad_norm": 0.7450636029243469,
+      "learning_rate": 0.0005796265515380463,
+      "loss": 4.2863,
       "step": 3250
     },
     {
-      "epoch": 0.35518243461414273,
-      "grad_norm": 0.768258810043335,
-      "learning_rate": 0.0005793255037172718,
-      "loss": 4.2639,
+      "epoch": 0.3557951482479784,
+      "grad_norm": 0.86126309633255,
+      "learning_rate": 0.000579302752293578,
+      "loss": 4.3108,
       "step": 3300
     },
     {
-      "epoch": 0.36056398665375095,
-      "grad_norm": 0.7073678970336914,
-      "learning_rate": 0.0005790022626872104,
-      "loss": 4.2469,
+      "epoch": 0.3611859838274933,
+      "grad_norm": 0.7103971838951111,
+      "learning_rate": 0.0005789789530491095,
+      "loss": 4.287,
       "step": 3350
     },
     {
-      "epoch": 0.36594553869335916,
-      "grad_norm": 0.8866555690765381,
-      "learning_rate": 0.000578679021657149,
-      "loss": 4.2393,
+      "epoch": 0.3665768194070081,
+      "grad_norm": 0.8686874508857727,
+      "learning_rate": 0.0005786551538046411,
+      "loss": 4.2802,
       "step": 3400
     },
     {
-      "epoch": 0.3713270907329674,
-      "grad_norm": 0.629542887210846,
-      "learning_rate": 0.0005783557806270875,
-      "loss": 4.2315,
+      "epoch": 0.3719676549865229,
+      "grad_norm": 0.6263065934181213,
+      "learning_rate": 0.0005783313545601726,
+      "loss": 4.2675,
       "step": 3450
     },
     {
-      "epoch": 0.3767086427725756,
-      "grad_norm": 0.8919417858123779,
-      "learning_rate": 0.0005780325395970262,
-      "loss": 4.2348,
+      "epoch": 0.37735849056603776,
+      "grad_norm": 0.7077925205230713,
+      "learning_rate": 0.0005780075553157043,
+      "loss": 4.2435,
       "step": 3500
     },
     {
-      "epoch": 0.3820901948121838,
-      "grad_norm": 0.7498871088027954,
-      "learning_rate": 0.0005777092985669647,
-      "loss": 4.2246,
+      "epoch": 0.38274932614555257,
+      "grad_norm": 0.8216277360916138,
+      "learning_rate": 0.0005776837560712357,
+      "loss": 4.2474,
       "step": 3550
     },
     {
-      "epoch": 0.38747174685179203,
-      "grad_norm": 0.7134843468666077,
-      "learning_rate": 0.0005773860575369033,
-      "loss": 4.2223,
+      "epoch": 0.3881401617250674,
+      "grad_norm": 0.8052435517311096,
+      "learning_rate": 0.0005773599568267673,
+      "loss": 4.2508,
       "step": 3600
     },
     {
-      "epoch": 0.3928532988914003,
-      "grad_norm": 0.64930260181427,
-      "learning_rate": 0.0005770628165068419,
-      "loss": 4.1985,
+      "epoch": 0.3935309973045822,
+      "grad_norm": 1.003618836402893,
+      "learning_rate": 0.0005770361575822989,
+      "loss": 4.2544,
       "step": 3650
     },
     {
-      "epoch": 0.3982348509310085,
-      "grad_norm": 0.6372562646865845,
-      "learning_rate": 0.0005767395754767804,
-      "loss": 4.2213,
+      "epoch": 0.39892183288409705,
+      "grad_norm": 0.8883004784584045,
+      "learning_rate": 0.0005767123583378305,
+      "loss": 4.228,
       "step": 3700
     },
     {
-      "epoch": 0.40361640297061674,
-      "grad_norm": 0.8313030004501343,
-      "learning_rate": 0.0005764163344467191,
-      "loss": 4.1942,
+      "epoch": 0.40431266846361186,
+      "grad_norm": 0.8125618696212769,
+      "learning_rate": 0.000576388559093362,
+      "loss": 4.2213,
       "step": 3750
     },
     {
-      "epoch": 0.40899795501022496,
-      "grad_norm": 0.7237107753753662,
-      "learning_rate": 0.0005760930934166576,
-      "loss": 4.1753,
+      "epoch": 0.40970350404312667,
+      "grad_norm": 0.7294287085533142,
+      "learning_rate": 0.0005760647598488936,
+      "loss": 4.2398,
       "step": 3800
     },
     {
-      "epoch": 0.4143795070498332,
-      "grad_norm": 0.602759063243866,
-      "learning_rate": 0.0005757698523865963,
-      "loss": 4.1893,
+      "epoch": 0.41509433962264153,
+      "grad_norm": 0.7409259080886841,
+      "learning_rate": 0.0005757409606044253,
+      "loss": 4.2228,
       "step": 3850
     },
     {
-      "epoch": 0.4197610590894414,
-      "grad_norm": 0.6982997059822083,
-      "learning_rate": 0.0005754466113565348,
-      "loss": 4.1906,
+      "epoch": 0.42048517520215634,
+      "grad_norm": 0.9595204591751099,
+      "learning_rate": 0.0005754171613599568,
+      "loss": 4.189,
       "step": 3900
     },
     {
-      "epoch": 0.4251426111290496,
-      "grad_norm": 0.8566523790359497,
-      "learning_rate": 0.0005751233703264733,
-      "loss": 4.1769,
+      "epoch": 0.42587601078167114,
+      "grad_norm": 0.6629497408866882,
+      "learning_rate": 0.0005750933621154884,
+      "loss": 4.1884,
       "step": 3950
     },
     {
-      "epoch": 0.4305241631686578,
-      "grad_norm": 0.7572330236434937,
-      "learning_rate": 0.000574800129296412,
-      "loss": 4.1659,
+      "epoch": 0.431266846361186,
+      "grad_norm": 0.8883629441261292,
+      "learning_rate": 0.0005747695628710199,
+      "loss": 4.1956,
       "step": 4000
     },
     {
-      "epoch": 0.4305241631686578,
-      "eval_accuracy": 0.31151891195867365,
-      "eval_loss": 4.096553802490234,
-      "eval_runtime": 183.6747,
-      "eval_samples_per_second": 98.059,
-      "eval_steps_per_second": 6.13,
+      "epoch": 0.431266846361186,
+      "eval_accuracy": 0.30888592652826535,
+      "eval_loss": 4.123743534088135,
+      "eval_runtime": 153.2705,
+      "eval_samples_per_second": 117.511,
+      "eval_steps_per_second": 7.346,
       "step": 4000
     },
     {
-      "epoch": 0.43590571520826604,
-      "grad_norm": 0.7648617029190063,
-      "learning_rate": 0.0005744768882663505,
-      "loss": 4.1546,
+      "epoch": 0.4366576819407008,
+      "grad_norm": 0.6576101779937744,
+      "learning_rate": 0.0005744457636265515,
+      "loss": 4.1943,
       "step": 4050
     },
     {
-      "epoch": 0.44128726724787426,
-      "grad_norm": 0.8174023032188416,
-      "learning_rate": 0.0005741536472362891,
-      "loss": 4.1598,
+      "epoch": 0.4420485175202156,
+      "grad_norm": 0.6346908211708069,
+      "learning_rate": 0.0005741219643820831,
+      "loss": 4.1902,
       "step": 4100
     },
     {
-      "epoch": 0.44666881928748253,
-      "grad_norm": 0.7902569770812988,
-      "learning_rate": 0.0005738304062062277,
-      "loss": 4.1417,
+      "epoch": 0.4474393530997305,
+      "grad_norm": 0.6723673343658447,
+      "learning_rate": 0.0005737981651376146,
+      "loss": 4.1837,
       "step": 4150
     },
     {
-      "epoch": 0.45205037132709075,
-      "grad_norm": 0.775563657283783,
-      "learning_rate": 0.0005735071651761664,
-      "loss": 4.1588,
+      "epoch": 0.4528301886792453,
+      "grad_norm": 0.8088533282279968,
+      "learning_rate": 0.0005734743658931462,
+      "loss": 4.1711,
       "step": 4200
     },
     {
-      "epoch": 0.45743192336669897,
-      "grad_norm": 0.7469772100448608,
-      "learning_rate": 0.0005731839241461049,
-      "loss": 4.129,
+      "epoch": 0.4582210242587601,
+      "grad_norm": 0.5922421813011169,
+      "learning_rate": 0.0005731505666486778,
+      "loss": 4.1633,
       "step": 4250
     },
     {
-      "epoch": 0.4628134754063072,
-      "grad_norm": 0.7498493194580078,
-      "learning_rate": 0.0005728606831160435,
-      "loss": 4.1349,
+      "epoch": 0.4636118598382749,
+      "grad_norm": 0.7532811164855957,
+      "learning_rate": 0.0005728267674042093,
+      "loss": 4.1571,
       "step": 4300
     },
     {
-      "epoch": 0.4681950274459154,
-      "grad_norm": 0.777646541595459,
-      "learning_rate": 0.000572537442085982,
-      "loss": 4.1185,
+      "epoch": 0.46900269541778977,
+      "grad_norm": 0.7632968425750732,
+      "learning_rate": 0.0005725029681597409,
+      "loss": 4.1624,
       "step": 4350
     },
     {
-      "epoch": 0.4735765794855236,
-      "grad_norm": 0.7824798822402954,
-      "learning_rate": 0.0005722142010559206,
-      "loss": 4.1398,
+      "epoch": 0.4743935309973046,
+      "grad_norm": 0.683935284614563,
+      "learning_rate": 0.0005721791689152725,
+      "loss": 4.1474,
       "step": 4400
     },
     {
-      "epoch": 0.47895813152513184,
-      "grad_norm": 0.7924827337265015,
-      "learning_rate": 0.0005718909600258593,
-      "loss": 4.12,
+      "epoch": 0.4797843665768194,
+      "grad_norm": 0.828084409236908,
+      "learning_rate": 0.0005718553696708041,
+      "loss": 4.1485,
       "step": 4450
     },
     {
-      "epoch": 0.48433968356474005,
-      "grad_norm": 0.6577082872390747,
-      "learning_rate": 0.0005715677189957978,
-      "loss": 4.1047,
+      "epoch": 0.48517520215633425,
+      "grad_norm": 0.6277663707733154,
+      "learning_rate": 0.0005715315704263356,
+      "loss": 4.1318,
       "step": 4500
     },
     {
-      "epoch": 0.48972123560434827,
-      "grad_norm": 0.7151188254356384,
-      "learning_rate": 0.0005712444779657364,
-      "loss": 4.1206,
+      "epoch": 0.49056603773584906,
+      "grad_norm": 0.7237159013748169,
+      "learning_rate": 0.0005712077711818672,
+      "loss": 4.1216,
       "step": 4550
     },
     {
-      "epoch": 0.49510278764395654,
-      "grad_norm": 0.64276123046875,
-      "learning_rate": 0.000570921236935675,
-      "loss": 4.1035,
+      "epoch": 0.49595687331536387,
+      "grad_norm": 0.7076147794723511,
+      "learning_rate": 0.0005708839719373987,
+      "loss": 4.1303,
       "step": 4600
     },
     {
-      "epoch": 0.5004843396835648,
-      "grad_norm": 0.6700813174247742,
-      "learning_rate": 0.0005705979959056136,
-      "loss": 4.0922,
+      "epoch": 0.5013477088948787,
+      "grad_norm": 0.6837750673294067,
+      "learning_rate": 0.0005705601726929304,
+      "loss": 4.1089,
       "step": 4650
     },
     {
-      "epoch": 0.505865891723173,
-      "grad_norm": 0.6062188744544983,
-      "learning_rate": 0.0005702747548755522,
-      "loss": 4.0885,
+      "epoch": 0.5067385444743935,
+      "grad_norm": 0.8128517270088196,
+      "learning_rate": 0.0005702363734484619,
+      "loss": 4.1158,
       "step": 4700
     },
     {
-      "epoch": 0.5112474437627812,
-      "grad_norm": 0.6247645616531372,
-      "learning_rate": 0.0005699515138454908,
-      "loss": 4.0825,
+      "epoch": 0.5121293800539084,
+      "grad_norm": 0.5411143898963928,
+      "learning_rate": 0.0005699125742039935,
+      "loss": 4.0971,
       "step": 4750
     },
     {
-      "epoch": 0.5166289958023894,
-      "grad_norm": 0.6468704342842102,
-      "learning_rate": 0.0005696282728154293,
-      "loss": 4.1031,
+      "epoch": 0.5175202156334232,
+      "grad_norm": 0.6508839130401611,
+      "learning_rate": 0.000569588774959525,
+      "loss": 4.1118,
       "step": 4800
     },
     {
-      "epoch": 0.5220105478419976,
-      "grad_norm": 0.5434349179267883,
-      "learning_rate": 0.0005693050317853679,
-      "loss": 4.0938,
+      "epoch": 0.522911051212938,
+      "grad_norm": 0.6948415040969849,
+      "learning_rate": 0.0005692649757150567,
+      "loss": 4.1042,
       "step": 4850
     },
     {
-      "epoch": 0.5273920998816058,
-      "grad_norm": 0.7494503855705261,
-      "learning_rate": 0.0005689817907553065,
-      "loss": 4.0747,
+      "epoch": 0.5283018867924528,
+      "grad_norm": 0.6959540247917175,
+      "learning_rate": 0.0005689411764705881,
+      "loss": 4.1124,
       "step": 4900
     },
     {
-      "epoch": 0.5327736519212141,
-      "grad_norm": 0.5512547492980957,
-      "learning_rate": 0.000568658549725245,
-      "loss": 4.0845,
+      "epoch": 0.5336927223719676,
+      "grad_norm": 0.7044671177864075,
+      "learning_rate": 0.0005686173772261197,
+      "loss": 4.0879,
       "step": 4950
     },
     {
-      "epoch": 0.5381552039608223,
-      "grad_norm": 0.6949486136436462,
-      "learning_rate": 0.0005683353086951837,
-      "loss": 4.0724,
+      "epoch": 0.5390835579514824,
+      "grad_norm": 0.633337676525116,
+      "learning_rate": 0.0005682935779816514,
+      "loss": 4.076,
       "step": 5000
     },
     {
-      "epoch": 0.5381552039608223,
-      "eval_accuracy": 0.3202588416015827,
-      "eval_loss": 3.9990289211273193,
-      "eval_runtime": 184.0076,
-      "eval_samples_per_second": 97.882,
-      "eval_steps_per_second": 6.119,
+      "epoch": 0.5390835579514824,
+      "eval_accuracy": 0.31857580836933636,
+      "eval_loss": 4.020145893096924,
+      "eval_runtime": 153.5029,
+      "eval_samples_per_second": 117.333,
+      "eval_steps_per_second": 7.335,
       "step": 5000
     },
     {
-      "epoch": 0.5435367560004305,
-      "grad_norm": 0.6374182105064392,
-      "learning_rate": 0.0005680120676651222,
-      "loss": 4.0595,
+      "epoch": 0.5444743935309974,
+      "grad_norm": 0.5795180201530457,
+      "learning_rate": 0.0005679697787371829,
+      "loss": 4.0695,
       "step": 5050
     },
     {
-      "epoch": 0.5489183080400387,
-      "grad_norm": 0.6074301600456238,
-      "learning_rate": 0.0005676888266350609,
-      "loss": 4.0512,
+      "epoch": 0.5498652291105122,
+      "grad_norm": 0.7057806849479675,
+      "learning_rate": 0.0005676459794927145,
+      "loss": 4.0772,
       "step": 5100
     },
     {
-      "epoch": 0.5542998600796469,
-      "grad_norm": 0.5869015455245972,
-      "learning_rate": 0.0005673655856049994,
-      "loss": 4.0554,
+      "epoch": 0.555256064690027,
+      "grad_norm": 0.6123654246330261,
+      "learning_rate": 0.000567322180248246,
+      "loss": 4.0789,
       "step": 5150
     },
     {
-      "epoch": 0.5596814121192552,
-      "grad_norm": 0.636626660823822,
-      "learning_rate": 0.0005670423445749379,
-      "loss": 4.0444,
+      "epoch": 0.5606469002695418,
+      "grad_norm": 0.6368526220321655,
+      "learning_rate": 0.0005669983810037777,
+      "loss": 4.0569,
       "step": 5200
     },
     {
-      "epoch": 0.5650629641588634,
-      "grad_norm": 0.6491479277610779,
-      "learning_rate": 0.0005667191035448766,
-      "loss": 4.0441,
+      "epoch": 0.5660377358490566,
+      "grad_norm": 0.7483193278312683,
+      "learning_rate": 0.0005666745817593092,
+      "loss": 4.0595,
       "step": 5250
     },
     {
-      "epoch": 0.5704445161984716,
-      "grad_norm": 0.6757907271385193,
-      "learning_rate": 0.0005663958625148152,
-      "loss": 4.0463,
+      "epoch": 0.5714285714285714,
+      "grad_norm": 0.6111021041870117,
+      "learning_rate": 0.0005663507825148408,
+      "loss": 4.042,
       "step": 5300
     },
     {
-      "epoch": 0.5758260682380799,
-      "grad_norm": 0.6085935831069946,
-      "learning_rate": 0.0005660726214847538,
-      "loss": 4.0338,
+      "epoch": 0.5768194070080862,
+      "grad_norm": 0.7408255934715271,
+      "learning_rate": 0.0005660269832703723,
+      "loss": 4.0586,
       "step": 5350
     },
     {
-      "epoch": 0.5812076202776881,
-      "grad_norm": 0.5898264050483704,
-      "learning_rate": 0.0005657493804546923,
-      "loss": 4.045,
+      "epoch": 0.5822102425876011,
+      "grad_norm": 0.717922031879425,
+      "learning_rate": 0.0005657031840259039,
+      "loss": 4.0458,
       "step": 5400
     },
     {
-      "epoch": 0.5865891723172963,
-      "grad_norm": 0.7703672051429749,
-      "learning_rate": 0.0005654261394246309,
-      "loss": 4.0247,
+      "epoch": 0.5876010781671159,
+      "grad_norm": 0.6948340535163879,
+      "learning_rate": 0.0005653793847814355,
+      "loss": 4.0578,
       "step": 5450
     },
     {
-      "epoch": 0.5919707243569046,
-      "grad_norm": 0.6306314468383789,
-      "learning_rate": 0.0005651028983945695,
-      "loss": 4.0191,
+      "epoch": 0.5929919137466307,
+      "grad_norm": 0.6431124210357666,
+      "learning_rate": 0.000565055585536967,
+      "loss": 4.0547,
       "step": 5500
     },
     {
-      "epoch": 0.5973522763965128,
-      "grad_norm": 0.6411070227622986,
-      "learning_rate": 0.000564779657364508,
-      "loss": 4.0399,
+      "epoch": 0.5983827493261455,
+      "grad_norm": 0.6256029605865479,
+      "learning_rate": 0.0005647317862924986,
+      "loss": 4.0613,
       "step": 5550
     },
     {
-      "epoch": 0.602733828436121,
-      "grad_norm": 0.7495064735412598,
-      "learning_rate": 0.0005644564163344467,
-      "loss": 4.0159,
+      "epoch": 0.6037735849056604,
+      "grad_norm": 0.6387584209442139,
+      "learning_rate": 0.0005644079870480302,
+      "loss": 4.0393,
       "step": 5600
     },
     {
-      "epoch": 0.6081153804757292,
-      "grad_norm": 0.6548343896865845,
-      "learning_rate": 0.0005641331753043852,
-      "loss": 4.0313,
+      "epoch": 0.6091644204851752,
+      "grad_norm": 0.5745953321456909,
+      "learning_rate": 0.0005640841878035617,
+      "loss": 4.0271,
       "step": 5650
     },
     {
-      "epoch": 0.6134969325153374,
-      "grad_norm": 0.6325730085372925,
-      "learning_rate": 0.0005638099342743239,
-      "loss": 4.0135,
+      "epoch": 0.6145552560646901,
+      "grad_norm": 0.6845305562019348,
+      "learning_rate": 0.0005637603885590933,
+      "loss": 4.0351,
       "step": 5700
     },
     {
-      "epoch": 0.6188784845549457,
-      "grad_norm": 0.5813512802124023,
-      "learning_rate": 0.0005634866932442624,
-      "loss": 4.0193,
+      "epoch": 0.6199460916442049,
+      "grad_norm": 0.6254762411117554,
+      "learning_rate": 0.0005634365893146248,
+      "loss": 4.0458,
       "step": 5750
     },
     {
-      "epoch": 0.6242600365945539,
-      "grad_norm": 0.7371019124984741,
-      "learning_rate": 0.000563163452214201,
-      "loss": 3.9911,
+      "epoch": 0.6253369272237197,
+      "grad_norm": 0.5808873772621155,
+      "learning_rate": 0.0005631127900701565,
+      "loss": 4.0354,
       "step": 5800
     },
     {
-      "epoch": 0.6296415886341621,
-      "grad_norm": 0.5817407369613647,
-      "learning_rate": 0.0005628402111841396,
-      "loss": 4.0193,
+      "epoch": 0.6307277628032345,
+      "grad_norm": 0.650372326374054,
+      "learning_rate": 0.000562788990825688,
+      "loss": 4.0449,
       "step": 5850
     },
     {
-      "epoch": 0.6350231406737703,
-      "grad_norm": 0.6328310966491699,
-      "learning_rate": 0.0005625169701540782,
-      "loss": 3.9958,
+      "epoch": 0.6361185983827493,
+      "grad_norm": 0.6067761182785034,
+      "learning_rate": 0.0005624651915812196,
+      "loss": 4.014,
       "step": 5900
     },
     {
-      "epoch": 0.6404046927133785,
-      "grad_norm": 0.7258861064910889,
-      "learning_rate": 0.0005621937291240168,
-      "loss": 3.9997,
+      "epoch": 0.6415094339622641,
+      "grad_norm": 0.6532148718833923,
+      "learning_rate": 0.0005621413923367511,
+      "loss": 4.0074,
       "step": 5950
     },
     {
-      "epoch": 0.6457862447529867,
-      "grad_norm": 0.6138096451759338,
-      "learning_rate": 0.0005618704880939553,
-      "loss": 3.9907,
+      "epoch": 0.6469002695417789,
+      "grad_norm": 0.6020557284355164,
+      "learning_rate": 0.0005618175930922828,
+      "loss": 4.0178,
       "step": 6000
     },
     {
-      "epoch": 0.6457862447529867,
-      "eval_accuracy": 0.32727434126198807,
-      "eval_loss": 3.926227331161499,
-      "eval_runtime": 183.661,
-      "eval_samples_per_second": 98.067,
-      "eval_steps_per_second": 6.131,
+      "epoch": 0.6469002695417789,
+      "eval_accuracy": 0.3249853508744004,
+      "eval_loss": 3.9412128925323486,
+      "eval_runtime": 153.3779,
+      "eval_samples_per_second": 117.429,
+      "eval_steps_per_second": 7.341,
       "step": 6000
     },
     {
-      "epoch": 0.651167796792595,
-      "grad_norm": 0.6955623626708984,
-      "learning_rate": 0.0005615472470638939,
-      "loss": 3.9931,
+      "epoch": 0.6522911051212938,
+      "grad_norm": 0.6567849516868591,
+      "learning_rate": 0.0005614937938478143,
+      "loss": 3.9889,
       "step": 6050
     },
     {
-      "epoch": 0.6565493488322032,
-      "grad_norm": 0.5675811767578125,
-      "learning_rate": 0.0005612240060338325,
-      "loss": 3.9823,
+      "epoch": 0.6576819407008087,
+      "grad_norm": 0.7067267894744873,
+      "learning_rate": 0.0005611699946033459,
+      "loss": 4.0251,
       "step": 6100
     },
     {
-      "epoch": 0.6619309008718114,
-      "grad_norm": 0.6101326942443848,
-      "learning_rate": 0.0005609007650037711,
-      "loss": 3.9581,
+      "epoch": 0.6630727762803235,
+      "grad_norm": 0.5447336435317993,
+      "learning_rate": 0.0005608461953588774,
+      "loss": 4.0062,
       "step": 6150
     },
     {
-      "epoch": 0.6673124529114196,
-      "grad_norm": 0.5811055302619934,
-      "learning_rate": 0.0005605775239737097,
-      "loss": 3.9611,
+      "epoch": 0.6684636118598383,
+      "grad_norm": 0.6051011681556702,
+      "learning_rate": 0.000560522396114409,
+      "loss": 3.9816,
       "step": 6200
     },
     {
-      "epoch": 0.6726940049510278,
-      "grad_norm": 0.6250592470169067,
-      "learning_rate": 0.0005602542829436483,
-      "loss": 3.9918,
+      "epoch": 0.6738544474393531,
+      "grad_norm": 0.6470229625701904,
+      "learning_rate": 0.0005601985968699405,
+      "loss": 3.9996,
       "step": 6250
     },
     {
-      "epoch": 0.6780755569906362,
-      "grad_norm": 0.6698481440544128,
-      "learning_rate": 0.0005599310419135868,
-      "loss": 3.9753,
+      "epoch": 0.6792452830188679,
+      "grad_norm": 0.699021577835083,
+      "learning_rate": 0.0005598747976254721,
+      "loss": 4.0176,
       "step": 6300
     },
     {
-      "epoch": 0.6834571090302444,
-      "grad_norm": 0.682300865650177,
-      "learning_rate": 0.0005596078008835255,
-      "loss": 3.9756,
+      "epoch": 0.6846361185983828,
+      "grad_norm": 0.6702553629875183,
+      "learning_rate": 0.0005595509983810038,
+      "loss": 3.9787,
       "step": 6350
     },
     {
-      "epoch": 0.6888386610698526,
-      "grad_norm": 0.5564926266670227,
-      "learning_rate": 0.000559284559853464,
-      "loss": 3.9725,
+      "epoch": 0.6900269541778976,
+      "grad_norm": 0.5683832764625549,
+      "learning_rate": 0.0005592271991365353,
+      "loss": 3.9837,
       "step": 6400
     },
     {
-      "epoch": 0.6942202131094608,
-      "grad_norm": 0.5958801507949829,
-      "learning_rate": 0.0005589613188234026,
-      "loss": 3.9805,
+      "epoch": 0.6954177897574124,
+      "grad_norm": 0.6393706202507019,
+      "learning_rate": 0.0005589033998920669,
+      "loss": 3.9689,
       "step": 6450
     },
     {
-      "epoch": 0.699601765149069,
-      "grad_norm": 0.6558675169944763,
-      "learning_rate": 0.0005586380777933412,
-      "loss": 3.9501,
+      "epoch": 0.7008086253369272,
+      "grad_norm": 0.6676239967346191,
+      "learning_rate": 0.0005585796006475984,
+      "loss": 3.983,
       "step": 6500
     },
     {
-      "epoch": 0.7049833171886772,
-      "grad_norm": 0.5374082922935486,
-      "learning_rate": 0.0005583148367632797,
-      "loss": 3.9658,
+      "epoch": 0.706199460916442,
+      "grad_norm": 0.5636298060417175,
+      "learning_rate": 0.0005582558014031301,
+      "loss": 4.0066,
       "step": 6550
     },
     {
-      "epoch": 0.7103648692282855,
-      "grad_norm": 0.5923078656196594,
-      "learning_rate": 0.0005579915957332184,
-      "loss": 3.9615,
+      "epoch": 0.7115902964959568,
+      "grad_norm": 0.618071973323822,
+      "learning_rate": 0.0005579320021586616,
+      "loss": 3.9622,
       "step": 6600
     },
     {
-      "epoch": 0.7157464212678937,
-      "grad_norm": 0.635322093963623,
-      "learning_rate": 0.0005576683547031569,
-      "loss": 3.964,
+      "epoch": 0.7169811320754716,
+      "grad_norm": 0.5887283086776733,
+      "learning_rate": 0.0005576082029141932,
+      "loss": 3.9926,
       "step": 6650
     },
     {
-      "epoch": 0.7211279733075019,
-      "grad_norm": 0.6079649329185486,
-      "learning_rate": 0.0005573451136730956,
-      "loss": 3.9573,
+      "epoch": 0.7223719676549866,
+      "grad_norm": 0.6095805168151855,
+      "learning_rate": 0.0005572844036697247,
+      "loss": 3.9656,
       "step": 6700
     },
     {
-      "epoch": 0.7265095253471101,
-      "grad_norm": 0.6069414615631104,
-      "learning_rate": 0.0005570218726430341,
-      "loss": 3.918,
+      "epoch": 0.7277628032345014,
+      "grad_norm": 0.5963723063468933,
+      "learning_rate": 0.0005569606044252563,
+      "loss": 3.9725,
       "step": 6750
     },
     {
-      "epoch": 0.7318910773867183,
-      "grad_norm": 0.6097965240478516,
-      "learning_rate": 0.0005566986316129728,
-      "loss": 3.9562,
+      "epoch": 0.7331536388140162,
+      "grad_norm": 0.6220936179161072,
+      "learning_rate": 0.0005566368051807879,
+      "loss": 3.9604,
       "step": 6800
     },
     {
-      "epoch": 0.7372726294263265,
-      "grad_norm": 0.6329127550125122,
-      "learning_rate": 0.0005563753905829113,
-      "loss": 3.9635,
+      "epoch": 0.738544474393531,
+      "grad_norm": 0.7677363753318787,
+      "learning_rate": 0.0005563130059363194,
+      "loss": 3.9399,
       "step": 6850
     },
     {
-      "epoch": 0.7426541814659348,
-      "grad_norm": 0.5949454307556152,
-      "learning_rate": 0.0005560521495528498,
-      "loss": 3.9418,
+      "epoch": 0.7439353099730458,
+      "grad_norm": 0.7177647352218628,
+      "learning_rate": 0.000555989206691851,
+      "loss": 3.9652,
       "step": 6900
     },
     {
-      "epoch": 0.748035733505543,
-      "grad_norm": 0.6039018630981445,
-      "learning_rate": 0.0005557289085227884,
-      "loss": 3.9432,
+      "epoch": 0.7493261455525606,
+      "grad_norm": 0.5666942596435547,
+      "learning_rate": 0.0005556654074473826,
+      "loss": 3.9621,
       "step": 6950
     },
     {
-      "epoch": 0.7534172855451512,
-      "grad_norm": 0.6565973162651062,
-      "learning_rate": 0.000555405667492727,
-      "loss": 3.9245,
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.6877643465995789,
+      "learning_rate": 0.0005553416082029141,
+      "loss": 3.9497,
       "step": 7000
     },
     {
-      "epoch": 0.7534172855451512,
-      "eval_accuracy": 0.33284975554730034,
-      "eval_loss": 3.869795799255371,
-      "eval_runtime": 183.8335,
-      "eval_samples_per_second": 97.975,
-      "eval_steps_per_second": 6.125,
+      "epoch": 0.7547169811320755,
+      "eval_accuracy": 0.33131622868868676,
+      "eval_loss": 3.8834235668182373,
+      "eval_runtime": 153.3456,
+      "eval_samples_per_second": 117.454,
+      "eval_steps_per_second": 7.343,
       "step": 7000
     },
     {
-      "epoch": 0.7587988375847594,
-      "grad_norm": 0.5873769521713257,
-      "learning_rate": 0.0005550824264626657,
-      "loss": 3.9419,
+      "epoch": 0.7601078167115903,
+      "grad_norm": 0.6757393479347229,
+      "learning_rate": 0.0005550178089584457,
+      "loss": 3.9406,
       "step": 7050
     },
     {
-      "epoch": 0.7641803896243676,
-      "grad_norm": 0.5975030660629272,
-      "learning_rate": 0.0005547591854326042,
-      "loss": 3.9205,
+      "epoch": 0.7654986522911051,
+      "grad_norm": 0.6207107901573181,
+      "learning_rate": 0.0005546940097139772,
+      "loss": 3.9529,
       "step": 7100
     },
     {
-      "epoch": 0.7695619416639758,
-      "grad_norm": 0.7010316252708435,
-      "learning_rate": 0.0005544359444025428,
-      "loss": 3.921,
+      "epoch": 0.77088948787062,
+      "grad_norm": 0.5236271619796753,
+      "learning_rate": 0.0005543702104695089,
+      "loss": 3.9499,
       "step": 7150
     },
     {
-      "epoch": 0.7749434937035841,
-      "grad_norm": 0.6028709411621094,
-      "learning_rate": 0.0005541127033724813,
-      "loss": 3.9054,
+      "epoch": 0.7762803234501348,
+      "grad_norm": 0.6513909101486206,
+      "learning_rate": 0.0005540464112250404,
+      "loss": 3.9531,
       "step": 7200
     },
     {
-      "epoch": 0.7803250457431924,
-      "grad_norm": 0.6167606711387634,
-      "learning_rate": 0.0005537894623424199,
-      "loss": 3.9206,
+      "epoch": 0.7816711590296496,
+      "grad_norm": 0.5805194973945618,
+      "learning_rate": 0.000553722611980572,
+      "loss": 3.91,
       "step": 7250
     },
     {
-      "epoch": 0.7857065977828006,
-      "grad_norm": 0.6583150029182434,
-      "learning_rate": 0.0005534662213123586,
-      "loss": 3.9263,
+      "epoch": 0.7870619946091644,
+      "grad_norm": 0.6637202501296997,
+      "learning_rate": 0.0005533988127361035,
+      "loss": 3.948,
       "step": 7300
     },
     {
-      "epoch": 0.7910881498224088,
-      "grad_norm": 0.6138869524002075,
-      "learning_rate": 0.0005531429802822971,
-      "loss": 3.9296,
+      "epoch": 0.7924528301886793,
+      "grad_norm": 0.558056116104126,
+      "learning_rate": 0.0005530750134916352,
+      "loss": 3.9333,
       "step": 7350
     },
     {
-      "epoch": 0.796469701862017,
-      "grad_norm": 0.7079772353172302,
-      "learning_rate": 0.0005528197392522357,
-      "loss": 3.9301,
+      "epoch": 0.7978436657681941,
+      "grad_norm": 0.7991946339607239,
+      "learning_rate": 0.0005527512142471668,
+      "loss": 3.9324,
       "step": 7400
     },
     {
-      "epoch": 0.8018512539016253,
-      "grad_norm": 0.6088504195213318,
-      "learning_rate": 0.0005524964982221743,
-      "loss": 3.9284,
+      "epoch": 0.8032345013477089,
+      "grad_norm": 0.5782480835914612,
+      "learning_rate": 0.0005524274150026982,
+      "loss": 3.9235,
       "step": 7450
     },
     {
-      "epoch": 0.8072328059412335,
-      "grad_norm": 0.5483489632606506,
-      "learning_rate": 0.0005521732571921129,
-      "loss": 3.8974,
+      "epoch": 0.8086253369272237,
+      "grad_norm": 0.661090612411499,
+      "learning_rate": 0.0005521036157582299,
+      "loss": 3.9098,
       "step": 7500
     },
     {
-      "epoch": 0.8126143579808417,
-      "grad_norm": 0.6508141160011292,
-      "learning_rate": 0.0005518500161620514,
-      "loss": 3.916,
+      "epoch": 0.8140161725067385,
+      "grad_norm": 0.6254606246948242,
+      "learning_rate": 0.0005517798165137614,
+      "loss": 3.936,
       "step": 7550
     },
     {
-      "epoch": 0.8179959100204499,
-      "grad_norm": 0.5982174277305603,
-      "learning_rate": 0.00055152677513199,
-      "loss": 3.9016,
+      "epoch": 0.8194070080862533,
+      "grad_norm": 0.5371659398078918,
+      "learning_rate": 0.000551456017269293,
+      "loss": 3.938,
       "step": 7600
     },
     {
-      "epoch": 0.8233774620600581,
-      "grad_norm": 0.5615628361701965,
-      "learning_rate": 0.0005512035341019286,
-      "loss": 3.9112,
+      "epoch": 0.8247978436657682,
+      "grad_norm": 0.644661009311676,
+      "learning_rate": 0.0005511322180248245,
+      "loss": 3.9149,
       "step": 7650
     },
     {
-      "epoch": 0.8287590140996663,
-      "grad_norm": 0.6235767006874084,
-      "learning_rate": 0.0005508802930718672,
-      "loss": 3.9002,
+      "epoch": 0.8301886792452831,
+      "grad_norm": 0.6013936996459961,
+      "learning_rate": 0.0005508084187803562,
+      "loss": 3.9182,
       "step": 7700
     },
     {
-      "epoch": 0.8341405661392746,
-      "grad_norm": 0.6372842788696289,
-      "learning_rate": 0.0005505570520418058,
-      "loss": 3.9002,
+      "epoch": 0.8355795148247979,
+      "grad_norm": 0.5374178886413574,
+      "learning_rate": 0.0005504846195358877,
+      "loss": 3.9007,
       "step": 7750
     },
     {
-      "epoch": 0.8395221181788828,
-      "grad_norm": 0.5565422773361206,
-      "learning_rate": 0.0005502338110117443,
-      "loss": 3.886,
+      "epoch": 0.8409703504043127,
+      "grad_norm": 0.6176519393920898,
+      "learning_rate": 0.0005501608202914193,
+      "loss": 3.9278,
       "step": 7800
     },
     {
-      "epoch": 0.844903670218491,
-      "grad_norm": 0.5980644822120667,
-      "learning_rate": 0.000549910569981683,
-      "loss": 3.8933,
+      "epoch": 0.8463611859838275,
+      "grad_norm": 0.6597527861595154,
+      "learning_rate": 0.0005498370210469508,
+      "loss": 3.9062,
       "step": 7850
     },
     {
-      "epoch": 0.8502852222580992,
-      "grad_norm": 0.5744678378105164,
-      "learning_rate": 0.0005495873289516215,
-      "loss": 3.8747,
+      "epoch": 0.8517520215633423,
+      "grad_norm": 0.5706176161766052,
+      "learning_rate": 0.0005495132218024824,
+      "loss": 3.9084,
       "step": 7900
     },
     {
-      "epoch": 0.8556667742977074,
-      "grad_norm": 0.5711344480514526,
-      "learning_rate": 0.0005492640879215602,
-      "loss": 3.8882,
+      "epoch": 0.8571428571428571,
+      "grad_norm": 0.6241531372070312,
+      "learning_rate": 0.000549189422558014,
+      "loss": 3.9047,
       "step": 7950
     },
     {
-      "epoch": 0.8610483263373157,
-      "grad_norm": 0.559182345867157,
-      "learning_rate": 0.0005489408468914987,
-      "loss": 3.8694,
+      "epoch": 0.862533692722372,
+      "grad_norm": 0.7237799763679504,
+      "learning_rate": 0.0005488656233135456,
+      "loss": 3.8934,
       "step": 8000
     },
     {
-      "epoch": 0.8610483263373157,
-      "eval_accuracy": 0.3374283882398026,
-      "eval_loss": 3.820354461669922,
-      "eval_runtime": 183.958,
-      "eval_samples_per_second": 97.908,
-      "eval_steps_per_second": 6.121,
+      "epoch": 0.862533692722372,
+      "eval_accuracy": 0.33549306300205106,
+      "eval_loss": 3.8358263969421387,
+      "eval_runtime": 153.0976,
+      "eval_samples_per_second": 117.644,
+      "eval_steps_per_second": 7.355,
       "step": 8000
     },
     {
-      "epoch": 0.8664298783769239,
-      "grad_norm": 0.5815455317497253,
-      "learning_rate": 0.0005486176058614372,
-      "loss": 3.8719,
+      "epoch": 0.8679245283018868,
+      "grad_norm": 0.5674304366111755,
+      "learning_rate": 0.0005485418240690771,
+      "loss": 3.9022,
       "step": 8050
     },
     {
-      "epoch": 0.8718114304165321,
-      "grad_norm": 0.556144654750824,
-      "learning_rate": 0.0005482943648313759,
-      "loss": 3.8804,
+      "epoch": 0.8733153638814016,
+      "grad_norm": 0.6205889582633972,
+      "learning_rate": 0.0005482180248246087,
+      "loss": 3.8871,
       "step": 8100
     },
     {
-      "epoch": 0.8771929824561403,
-      "grad_norm": 0.5465092658996582,
-      "learning_rate": 0.0005479711238013145,
-      "loss": 3.8732,
+      "epoch": 0.8787061994609164,
+      "grad_norm": 0.5285912156105042,
+      "learning_rate": 0.0005478942255801403,
+      "loss": 3.8994,
       "step": 8150
     },
     {
-      "epoch": 0.8825745344957485,
-      "grad_norm": 0.6253562569618225,
-      "learning_rate": 0.0005476478827712531,
-      "loss": 3.8984,
+      "epoch": 0.8840970350404312,
+      "grad_norm": 0.5494291186332703,
+      "learning_rate": 0.0005475704263356718,
+      "loss": 3.8885,
       "step": 8200
     },
     {
-      "epoch": 0.8879560865353568,
-      "grad_norm": 0.5278693437576294,
-      "learning_rate": 0.0005473246417411916,
-      "loss": 3.8743,
+      "epoch": 0.889487870619946,
+      "grad_norm": 0.6058967709541321,
+      "learning_rate": 0.0005472466270912034,
+      "loss": 3.8736,
       "step": 8250
     },
     {
-      "epoch": 0.8933376385749651,
-      "grad_norm": 0.6014917492866516,
-      "learning_rate": 0.0005470014007111302,
-      "loss": 3.8721,
+      "epoch": 0.894878706199461,
+      "grad_norm": 0.6890183687210083,
+      "learning_rate": 0.000546922827846735,
+      "loss": 3.8986,
       "step": 8300
     },
     {
-      "epoch": 0.8987191906145733,
-      "grad_norm": 0.5801762342453003,
-      "learning_rate": 0.0005466781596810688,
-      "loss": 3.891,
+      "epoch": 0.9002695417789758,
+      "grad_norm": 0.625860333442688,
+      "learning_rate": 0.0005465990286022665,
+      "loss": 3.8816,
       "step": 8350
     },
     {
-      "epoch": 0.9041007426541815,
-      "grad_norm": 0.559445858001709,
-      "learning_rate": 0.0005463549186510073,
-      "loss": 3.8685,
+      "epoch": 0.9056603773584906,
+      "grad_norm": 0.6296173334121704,
+      "learning_rate": 0.0005462752293577981,
+      "loss": 3.8932,
       "step": 8400
     },
     {
-      "epoch": 0.9094822946937897,
-      "grad_norm": 0.5420745015144348,
-      "learning_rate": 0.000546031677620946,
-      "loss": 3.8733,
+      "epoch": 0.9110512129380054,
+      "grad_norm": 0.5746651291847229,
+      "learning_rate": 0.0005459514301133296,
+      "loss": 3.9057,
       "step": 8450
     },
     {
-      "epoch": 0.9148638467333979,
-      "grad_norm": 0.5471014380455017,
-      "learning_rate": 0.0005457084365908845,
-      "loss": 3.8714,
+      "epoch": 0.9164420485175202,
+      "grad_norm": 0.6228104829788208,
+      "learning_rate": 0.0005456276308688613,
+      "loss": 3.8985,
       "step": 8500
     },
     {
-      "epoch": 0.9202453987730062,
-      "grad_norm": 0.5171864032745361,
-      "learning_rate": 0.0005453851955608232,
-      "loss": 3.8572,
+      "epoch": 0.921832884097035,
+      "grad_norm": 0.5756477117538452,
+      "learning_rate": 0.0005453038316243929,
+      "loss": 3.8781,
       "step": 8550
     },
     {
-      "epoch": 0.9256269508126144,
-      "grad_norm": 0.5859182476997375,
-      "learning_rate": 0.0005450619545307617,
-      "loss": 3.8806,
+      "epoch": 0.9272237196765498,
+      "grad_norm": 0.5988864302635193,
+      "learning_rate": 0.0005449800323799244,
+      "loss": 3.8724,
       "step": 8600
     },
     {
-      "epoch": 0.9310085028522226,
-      "grad_norm": 0.5607949495315552,
-      "learning_rate": 0.0005447387135007003,
-      "loss": 3.8568,
+      "epoch": 0.9326145552560647,
+      "grad_norm": 0.5490165948867798,
+      "learning_rate": 0.000544656233135456,
+      "loss": 3.8638,
       "step": 8650
     },
     {
-      "epoch": 0.9363900548918308,
-      "grad_norm": 0.5764897465705872,
-      "learning_rate": 0.0005444154724706389,
-      "loss": 3.8628,
+      "epoch": 0.9380053908355795,
+      "grad_norm": 0.5722247362136841,
+      "learning_rate": 0.0005443324338909875,
+      "loss": 3.8698,
       "step": 8700
     },
     {
-      "epoch": 0.941771606931439,
-      "grad_norm": 0.5329459309577942,
-      "learning_rate": 0.0005440922314405775,
-      "loss": 3.8604,
+      "epoch": 0.9433962264150944,
+      "grad_norm": 0.592191755771637,
+      "learning_rate": 0.0005440086346465192,
+      "loss": 3.8706,
       "step": 8750
     },
     {
-      "epoch": 0.9471531589710472,
-      "grad_norm": 0.712491512298584,
-      "learning_rate": 0.0005437689904105161,
-      "loss": 3.8376,
+      "epoch": 0.9487870619946092,
+      "grad_norm": 0.5509844422340393,
+      "learning_rate": 0.0005436848354020506,
+      "loss": 3.8615,
       "step": 8800
     },
     {
-      "epoch": 0.9525347110106555,
-      "grad_norm": 0.6513589024543762,
-      "learning_rate": 0.0005434457493804546,
-      "loss": 3.8705,
+      "epoch": 0.954177897574124,
+      "grad_norm": 0.6250285506248474,
+      "learning_rate": 0.0005433610361575823,
+      "loss": 3.8504,
       "step": 8850
     },
     {
-      "epoch": 0.9579162630502637,
-      "grad_norm": 0.5159483551979065,
-      "learning_rate": 0.0005431225083503932,
-      "loss": 3.8302,
+      "epoch": 0.9595687331536388,
+      "grad_norm": 0.5853751301765442,
+      "learning_rate": 0.0005430372369131138,
+      "loss": 3.8685,
       "step": 8900
     },
     {
-      "epoch": 0.9632978150898719,
-      "grad_norm": 0.619147539138794,
-      "learning_rate": 0.0005427992673203318,
-      "loss": 3.8394,
+      "epoch": 0.9649595687331537,
+      "grad_norm": 0.6114296913146973,
+      "learning_rate": 0.0005427134376686454,
+      "loss": 3.8746,
       "step": 8950
     },
     {
-      "epoch": 0.9686793671294801,
-      "grad_norm": 0.5892804861068726,
-      "learning_rate": 0.0005424760262902704,
-      "loss": 3.8479,
+      "epoch": 0.9703504043126685,
+      "grad_norm": 0.5549982190132141,
+      "learning_rate": 0.0005423896384241769,
+      "loss": 3.8623,
       "step": 9000
     },
     {
-      "epoch": 0.9686793671294801,
-      "eval_accuracy": 0.34035245475666587,
-      "eval_loss": 3.7828423976898193,
-      "eval_runtime": 183.4819,
-      "eval_samples_per_second": 98.162,
-      "eval_steps_per_second": 6.137,
+      "epoch": 0.9703504043126685,
+      "eval_accuracy": 0.33976007921230134,
+      "eval_loss": 3.796833038330078,
+      "eval_runtime": 153.5005,
+      "eval_samples_per_second": 117.335,
+      "eval_steps_per_second": 7.335,
       "step": 9000
     },
     {
-      "epoch": 0.9740609191690883,
-      "grad_norm": 0.5918485522270203,
-      "learning_rate": 0.000542152785260209,
-      "loss": 3.8558,
+      "epoch": 0.9757412398921833,
+      "grad_norm": 0.5819941759109497,
+      "learning_rate": 0.0005420658391797086,
+      "loss": 3.8561,
       "step": 9050
     },
     {
-      "epoch": 0.9794424712086965,
-      "grad_norm": 0.48340779542922974,
-      "learning_rate": 0.0005418295442301476,
-      "loss": 3.8433,
+      "epoch": 0.9811320754716981,
+      "grad_norm": 0.6158163547515869,
+      "learning_rate": 0.0005417420399352401,
+      "loss": 3.8681,
       "step": 9100
     },
     {
-      "epoch": 0.9848240232483048,
-      "grad_norm": 0.559395968914032,
-      "learning_rate": 0.0005415063032000861,
-      "loss": 3.847,
+      "epoch": 0.9865229110512129,
+      "grad_norm": 0.5715709924697876,
+      "learning_rate": 0.0005414182406907717,
+      "loss": 3.8422,
       "step": 9150
     },
     {
-      "epoch": 0.9902055752879131,
-      "grad_norm": 0.5653789043426514,
-      "learning_rate": 0.0005411830621700248,
-      "loss": 3.8355,
+      "epoch": 0.9919137466307277,
+      "grad_norm": 0.5587881207466125,
+      "learning_rate": 0.0005410944414463032,
+      "loss": 3.8401,
       "step": 9200
     },
     {
-      "epoch": 0.9955871273275213,
-      "grad_norm": 0.5224032402038574,
-      "learning_rate": 0.0005408598211399633,
-      "loss": 3.8249,
+      "epoch": 0.9973045822102425,
+      "grad_norm": 0.5460543036460876,
+      "learning_rate": 0.0005407706422018348,
+      "loss": 3.8508,
       "step": 9250
     },
     {
-      "epoch": 1.0009686793671295,
-      "grad_norm": 0.572032630443573,
-      "learning_rate": 0.0005405365801099019,
-      "loss": 3.828,
+      "epoch": 1.0026954177897573,
+      "grad_norm": 0.6066132187843323,
+      "learning_rate": 0.0005404468429573664,
+      "loss": 3.8255,
       "step": 9300
     },
     {
-      "epoch": 1.0063502314067376,
-      "grad_norm": 0.5785325169563293,
-      "learning_rate": 0.0005402133390798405,
-      "loss": 3.7596,
+      "epoch": 1.0080862533692723,
+      "grad_norm": 0.5967060327529907,
+      "learning_rate": 0.000540123043712898,
+      "loss": 3.7934,
       "step": 9350
     },
     {
-      "epoch": 1.011731783446346,
-      "grad_norm": 0.5417508482933044,
-      "learning_rate": 0.000539890098049779,
-      "loss": 3.7565,
+      "epoch": 1.013477088948787,
+      "grad_norm": 0.5728771686553955,
+      "learning_rate": 0.0005397992444684295,
+      "loss": 3.7864,
       "step": 9400
     },
     {
-      "epoch": 1.017113335485954,
-      "grad_norm": 0.6021242141723633,
-      "learning_rate": 0.0005395668570197177,
-      "loss": 3.7638,
+      "epoch": 1.0188679245283019,
+      "grad_norm": 0.6757526993751526,
+      "learning_rate": 0.0005394754452239611,
+      "loss": 3.7864,
       "step": 9450
     },
     {
-      "epoch": 1.0224948875255624,
-      "grad_norm": 0.5668533444404602,
-      "learning_rate": 0.0005392436159896562,
-      "loss": 3.7673,
+      "epoch": 1.0242587601078168,
+      "grad_norm": 0.6057505011558533,
+      "learning_rate": 0.0005391516459794927,
+      "loss": 3.7886,
       "step": 9500
     },
     {
-      "epoch": 1.0278764395651705,
-      "grad_norm": 0.5705947279930115,
-      "learning_rate": 0.0005389203749595948,
-      "loss": 3.7738,
+      "epoch": 1.0296495956873315,
+      "grad_norm": 0.5644155740737915,
+      "learning_rate": 0.0005388278467350242,
+      "loss": 3.787,
       "step": 9550
     },
     {
-      "epoch": 1.0332579916047788,
-      "grad_norm": 0.5880433320999146,
-      "learning_rate": 0.0005385971339295334,
-      "loss": 3.7634,
+      "epoch": 1.0350404312668464,
+      "grad_norm": 0.6167004108428955,
+      "learning_rate": 0.0005385040474905557,
+      "loss": 3.7731,
       "step": 9600
     },
     {
-      "epoch": 1.0386395436443872,
-      "grad_norm": 0.629132866859436,
-      "learning_rate": 0.000538273892899472,
-      "loss": 3.7783,
+      "epoch": 1.0404312668463611,
+      "grad_norm": 0.6160714626312256,
+      "learning_rate": 0.0005381802482460874,
+      "loss": 3.8133,
       "step": 9650
     },
     {
-      "epoch": 1.0440210956839953,
-      "grad_norm": 0.6227521300315857,
-      "learning_rate": 0.0005379506518694106,
-      "loss": 3.7852,
+      "epoch": 1.045822102425876,
+      "grad_norm": 0.6024379134178162,
+      "learning_rate": 0.000537856449001619,
+      "loss": 3.8009,
       "step": 9700
     },
     {
-      "epoch": 1.0494026477236036,
-      "grad_norm": 0.5902397632598877,
-      "learning_rate": 0.0005376274108393491,
-      "loss": 3.7773,
+      "epoch": 1.0512129380053907,
+      "grad_norm": 0.5305362343788147,
+      "learning_rate": 0.0005375326497571505,
+      "loss": 3.7983,
       "step": 9750
     },
     {
-      "epoch": 1.0547841997632117,
-      "grad_norm": 0.5385565161705017,
-      "learning_rate": 0.0005373041698092877,
-      "loss": 3.7692,
+      "epoch": 1.0566037735849056,
+      "grad_norm": 0.6367751955986023,
+      "learning_rate": 0.000537208850512682,
+      "loss": 3.8077,
       "step": 9800
     },
     {
-      "epoch": 1.06016575180282,
-      "grad_norm": 0.5663689374923706,
-      "learning_rate": 0.0005369809287792263,
-      "loss": 3.7817,
+      "epoch": 1.0619946091644206,
+      "grad_norm": 0.53994220495224,
+      "learning_rate": 0.0005368850512682137,
+      "loss": 3.8023,
       "step": 9850
     },
     {
-      "epoch": 1.0655473038424281,
-      "grad_norm": 0.5260084867477417,
-      "learning_rate": 0.000536657687749165,
-      "loss": 3.7742,
+      "epoch": 1.0673854447439353,
+      "grad_norm": 0.5434533953666687,
+      "learning_rate": 0.0005365612520237453,
+      "loss": 3.7914,
       "step": 9900
     },
     {
-      "epoch": 1.0709288558820365,
-      "grad_norm": 0.5762213468551636,
-      "learning_rate": 0.0005363344467191035,
-      "loss": 3.7833,
+      "epoch": 1.0727762803234502,
+      "grad_norm": 0.6101065278053284,
+      "learning_rate": 0.0005362374527792768,
+      "loss": 3.788,
       "step": 9950
     },
     {
-      "epoch": 1.0763104079216446,
-      "grad_norm": 0.6788592338562012,
-      "learning_rate": 0.000536011205689042,
-      "loss": 3.7598,
+      "epoch": 1.0781671159029649,
+      "grad_norm": 0.6409618258476257,
+      "learning_rate": 0.0005359136535348084,
+      "loss": 3.7815,
       "step": 10000
     },
     {
-      "epoch": 1.0763104079216446,
-      "eval_accuracy": 0.34371200204788965,
-      "eval_loss": 3.7531447410583496,
-      "eval_runtime": 183.7635,
-      "eval_samples_per_second": 98.012,
-      "eval_steps_per_second": 6.127,
+      "epoch": 1.0781671159029649,
+      "eval_accuracy": 0.3427394500490622,
+      "eval_loss": 3.7646424770355225,
+      "eval_runtime": 153.0954,
+      "eval_samples_per_second": 117.646,
+      "eval_steps_per_second": 7.355,
       "step": 10000
     },
     {
-      "epoch": 1.081691959961253,
-      "grad_norm": 0.6287451386451721,
-      "learning_rate": 0.0005356879646589807,
-      "loss": 3.7918,
+      "epoch": 1.0835579514824798,
+      "grad_norm": 0.5196843147277832,
+      "learning_rate": 0.0005355898542903399,
+      "loss": 3.7823,
       "step": 10050
     },
     {
-      "epoch": 1.087073512000861,
-      "grad_norm": 0.5594741702079773,
-      "learning_rate": 0.0005353647236289192,
-      "loss": 3.77,
+      "epoch": 1.0889487870619945,
+      "grad_norm": 0.5755773782730103,
+      "learning_rate": 0.0005352660550458716,
+      "loss": 3.7917,
       "step": 10100
     },
     {
-      "epoch": 1.0924550640404693,
-      "grad_norm": 0.6059030294418335,
-      "learning_rate": 0.000535047947419459,
-      "loss": 3.748,
+      "epoch": 1.0943396226415094,
+      "grad_norm": 0.5776042938232422,
+      "learning_rate": 0.000534942255801403,
+      "loss": 3.7844,
       "step": 10150
     },
     {
-      "epoch": 1.0978366160800774,
-      "grad_norm": 0.5481324791908264,
-      "learning_rate": 0.0005347247063893976,
-      "loss": 3.768,
+      "epoch": 1.0997304582210243,
+      "grad_norm": 0.5790849924087524,
+      "learning_rate": 0.0005346184565569347,
+      "loss": 3.7854,
       "step": 10200
     },
     {
-      "epoch": 1.1032181681196858,
-      "grad_norm": 0.6304662227630615,
-      "learning_rate": 0.0005344014653593362,
-      "loss": 3.7647,
+      "epoch": 1.105121293800539,
+      "grad_norm": 0.5127938389778137,
+      "learning_rate": 0.0005342946573124662,
+      "loss": 3.7945,
       "step": 10250
     },
     {
-      "epoch": 1.1085997201592939,
-      "grad_norm": 0.6041542291641235,
-      "learning_rate": 0.0005340782243292748,
-      "loss": 3.7727,
+      "epoch": 1.110512129380054,
+      "grad_norm": 0.5364027619361877,
+      "learning_rate": 0.0005339708580679978,
+      "loss": 3.7794,
       "step": 10300
     },
     {
-      "epoch": 1.1139812721989022,
-      "grad_norm": 0.525950014591217,
-      "learning_rate": 0.0005337549832992134,
-      "loss": 3.7709,
+      "epoch": 1.1159029649595686,
+      "grad_norm": 0.6965321898460388,
+      "learning_rate": 0.0005336470588235293,
+      "loss": 3.767,
       "step": 10350
     },
     {
-      "epoch": 1.1193628242385103,
-      "grad_norm": 0.5421223640441895,
-      "learning_rate": 0.000533431742269152,
-      "loss": 3.7719,
+      "epoch": 1.1212938005390836,
+      "grad_norm": 0.5849443674087524,
+      "learning_rate": 0.000533323259579061,
+      "loss": 3.7827,
       "step": 10400
     },
     {
-      "epoch": 1.1247443762781186,
-      "grad_norm": 0.6486043334007263,
-      "learning_rate": 0.0005331085012390905,
-      "loss": 3.7697,
+      "epoch": 1.1266846361185983,
+      "grad_norm": 0.5768404006958008,
+      "learning_rate": 0.0005329994603345925,
+      "loss": 3.7695,
       "step": 10450
     },
     {
-      "epoch": 1.1301259283177267,
-      "grad_norm": 0.5491141676902771,
-      "learning_rate": 0.0005327852602090292,
-      "loss": 3.7562,
+      "epoch": 1.1320754716981132,
+      "grad_norm": 0.6388025879859924,
+      "learning_rate": 0.0005326756610901241,
+      "loss": 3.7617,
       "step": 10500
     },
     {
-      "epoch": 1.135507480357335,
-      "grad_norm": 0.6132428050041199,
-      "learning_rate": 0.0005324620191789678,
-      "loss": 3.7747,
+      "epoch": 1.137466307277628,
+      "grad_norm": 0.5754392743110657,
+      "learning_rate": 0.0005323518618456556,
+      "loss": 3.7825,
       "step": 10550
     },
     {
-      "epoch": 1.1408890323969434,
-      "grad_norm": 0.5425587892532349,
-      "learning_rate": 0.0005321387781489063,
-      "loss": 3.7812,
+      "epoch": 1.1428571428571428,
+      "grad_norm": 0.5683891177177429,
+      "learning_rate": 0.0005320280626011872,
+      "loss": 3.7924,
       "step": 10600
     },
     {
-      "epoch": 1.1462705844365515,
-      "grad_norm": 0.584142804145813,
-      "learning_rate": 0.0005318155371188449,
-      "loss": 3.7623,
+      "epoch": 1.1482479784366577,
+      "grad_norm": 0.5924477577209473,
+      "learning_rate": 0.0005317042633567188,
+      "loss": 3.7689,
       "step": 10650
     },
     {
-      "epoch": 1.1516521364761596,
-      "grad_norm": 0.5481099486351013,
-      "learning_rate": 0.0005314922960887834,
-      "loss": 3.7707,
+      "epoch": 1.1536388140161726,
+      "grad_norm": 0.5483968257904053,
+      "learning_rate": 0.0005313804641122504,
+      "loss": 3.7661,
       "step": 10700
     },
     {
-      "epoch": 1.157033688515768,
-      "grad_norm": 0.5392444133758545,
-      "learning_rate": 0.0005311690550587221,
-      "loss": 3.7551,
+      "epoch": 1.1590296495956873,
+      "grad_norm": 0.5982570052146912,
+      "learning_rate": 0.0005310566648677819,
+      "loss": 3.7554,
       "step": 10750
     },
     {
-      "epoch": 1.1624152405553763,
-      "grad_norm": 0.9533267617225647,
-      "learning_rate": 0.0005308458140286607,
-      "loss": 3.7787,
+      "epoch": 1.1644204851752022,
+      "grad_norm": 0.5551779270172119,
+      "learning_rate": 0.0005307328656233135,
+      "loss": 3.7443,
       "step": 10800
     },
     {
-      "epoch": 1.1677967925949844,
-      "grad_norm": 0.5618041157722473,
-      "learning_rate": 0.0005305225729985993,
-      "loss": 3.765,
+      "epoch": 1.169811320754717,
+      "grad_norm": 0.5809611678123474,
+      "learning_rate": 0.000530409066378845,
+      "loss": 3.7638,
       "step": 10850
     },
     {
-      "epoch": 1.1731783446345927,
-      "grad_norm": 0.5430598855018616,
-      "learning_rate": 0.0005301993319685378,
-      "loss": 3.7584,
+      "epoch": 1.1752021563342319,
+      "grad_norm": 0.5487034916877747,
+      "learning_rate": 0.0005300852671343766,
+      "loss": 3.7647,
       "step": 10900
     },
     {
-      "epoch": 1.1785598966742008,
-      "grad_norm": 0.5783632397651672,
-      "learning_rate": 0.0005298760909384765,
-      "loss": 3.7525,
+      "epoch": 1.1805929919137466,
+      "grad_norm": 0.5652864575386047,
+      "learning_rate": 0.0005297614678899081,
+      "loss": 3.7763,
       "step": 10950
     },
     {
-      "epoch": 1.1839414487138091,
-      "grad_norm": 0.49477651715278625,
-      "learning_rate": 0.000529552849908415,
-      "loss": 3.7743,
+      "epoch": 1.1859838274932615,
+      "grad_norm": 0.5819478631019592,
+      "learning_rate": 0.0005294376686454398,
+      "loss": 3.792,
       "step": 11000
     },
     {
-      "epoch": 1.1839414487138091,
-      "eval_accuracy": 0.34631119643018765,
-      "eval_loss": 3.727846384048462,
-      "eval_runtime": 183.6058,
-      "eval_samples_per_second": 98.096,
-      "eval_steps_per_second": 6.133,
+      "epoch": 1.1859838274932615,
+      "eval_accuracy": 0.3450138809496827,
+      "eval_loss": 3.7365806102752686,
+      "eval_runtime": 153.0144,
+      "eval_samples_per_second": 117.708,
+      "eval_steps_per_second": 7.359,
       "step": 11000
     },
     {
-      "epoch": 1.1893230007534172,
-      "grad_norm": 0.569889485836029,
-      "learning_rate": 0.0005292296088783535,
-      "loss": 3.7646,
+      "epoch": 1.1913746630727764,
+      "grad_norm": 0.6075570583343506,
+      "learning_rate": 0.0005291138694009714,
+      "loss": 3.7794,
       "step": 11050
     },
     {
-      "epoch": 1.1947045527930256,
-      "grad_norm": 0.6403191685676575,
-      "learning_rate": 0.0005289063678482922,
-      "loss": 3.7454,
+      "epoch": 1.196765498652291,
+      "grad_norm": 0.6090914011001587,
+      "learning_rate": 0.0005287900701565029,
+      "loss": 3.7724,
       "step": 11100
     },
     {
-      "epoch": 1.2000861048326337,
-      "grad_norm": 0.5591626763343811,
-      "learning_rate": 0.0005285831268182307,
-      "loss": 3.7508,
+      "epoch": 1.202156334231806,
+      "grad_norm": 0.5532864332199097,
+      "learning_rate": 0.0005284662709120345,
+      "loss": 3.7699,
       "step": 11150
     },
     {
-      "epoch": 1.205467656872242,
-      "grad_norm": 0.5478915572166443,
-      "learning_rate": 0.0005282598857881694,
-      "loss": 3.7397,
+      "epoch": 1.2075471698113207,
+      "grad_norm": 0.553909957408905,
+      "learning_rate": 0.0005281424716675661,
+      "loss": 3.7621,
       "step": 11200
     },
     {
-      "epoch": 1.21084920891185,
-      "grad_norm": 0.5612436532974243,
-      "learning_rate": 0.0005279366447581079,
-      "loss": 3.7477,
+      "epoch": 1.2129380053908356,
+      "grad_norm": 0.5670080780982971,
+      "learning_rate": 0.0005278186724230977,
+      "loss": 3.7532,
       "step": 11250
     },
     {
-      "epoch": 1.2162307609514584,
-      "grad_norm": 0.5590159296989441,
-      "learning_rate": 0.0005276134037280465,
-      "loss": 3.7391,
+      "epoch": 1.2183288409703503,
+      "grad_norm": 0.5886201858520508,
+      "learning_rate": 0.0005274948731786292,
+      "loss": 3.7683,
       "step": 11300
     },
     {
-      "epoch": 1.2216123129910665,
-      "grad_norm": 0.5453316569328308,
-      "learning_rate": 0.0005272901626979851,
-      "loss": 3.7417,
+      "epoch": 1.2237196765498652,
+      "grad_norm": 0.614900529384613,
+      "learning_rate": 0.0005271710739341608,
+      "loss": 3.7548,
       "step": 11350
     },
     {
-      "epoch": 1.2269938650306749,
-      "grad_norm": 0.5481910705566406,
-      "learning_rate": 0.0005269669216679236,
-      "loss": 3.7437,
+      "epoch": 1.2291105121293802,
+      "grad_norm": 0.5293157696723938,
+      "learning_rate": 0.0005268472746896923,
+      "loss": 3.7698,
       "step": 11400
     },
     {
-      "epoch": 1.232375417070283,
-      "grad_norm": 0.5396573543548584,
-      "learning_rate": 0.0005266436806378623,
-      "loss": 3.7468,
+      "epoch": 1.2345013477088949,
+      "grad_norm": 0.5872692465782166,
+      "learning_rate": 0.000526523475445224,
+      "loss": 3.7603,
       "step": 11450
     },
     {
-      "epoch": 1.2377569691098913,
-      "grad_norm": 0.5333154797554016,
-      "learning_rate": 0.0005263204396078008,
-      "loss": 3.7417,
+      "epoch": 1.2398921832884098,
+      "grad_norm": 0.6059370636940002,
+      "learning_rate": 0.0005261996762007554,
+      "loss": 3.7521,
       "step": 11500
     },
     {
-      "epoch": 1.2431385211494996,
-      "grad_norm": 0.5501976013183594,
-      "learning_rate": 0.0005259971985777394,
-      "loss": 3.7369,
+      "epoch": 1.2452830188679245,
+      "grad_norm": 0.5155454874038696,
+      "learning_rate": 0.0005258758769562871,
+      "loss": 3.7615,
       "step": 11550
     },
     {
-      "epoch": 1.2485200731891077,
-      "grad_norm": 0.62269127368927,
-      "learning_rate": 0.000525673957547678,
-      "loss": 3.736,
+      "epoch": 1.2506738544474394,
+      "grad_norm": 0.548378050327301,
+      "learning_rate": 0.0005255520777118186,
+      "loss": 3.7379,
       "step": 11600
     },
     {
-      "epoch": 1.2539016252287158,
-      "grad_norm": 0.6184050440788269,
-      "learning_rate": 0.0005253507165176167,
-      "loss": 3.741,
+      "epoch": 1.256064690026954,
+      "grad_norm": 0.5714156627655029,
+      "learning_rate": 0.0005252282784673502,
+      "loss": 3.7706,
       "step": 11650
     },
     {
-      "epoch": 1.2592831772683242,
-      "grad_norm": 0.5125299692153931,
-      "learning_rate": 0.0005250274754875552,
-      "loss": 3.7523,
+      "epoch": 1.261455525606469,
+      "grad_norm": 0.6832541227340698,
+      "learning_rate": 0.0005249044792228817,
+      "loss": 3.7662,
       "step": 11700
     },
     {
-      "epoch": 1.2646647293079325,
-      "grad_norm": 0.5639045238494873,
-      "learning_rate": 0.0005247042344574938,
-      "loss": 3.7253,
+      "epoch": 1.266846361185984,
+      "grad_norm": 0.5656735897064209,
+      "learning_rate": 0.0005245806799784133,
+      "loss": 3.7565,
       "step": 11750
     },
     {
-      "epoch": 1.2700462813475406,
-      "grad_norm": 0.5985804200172424,
-      "learning_rate": 0.0005243809934274323,
-      "loss": 3.7521,
+      "epoch": 1.2722371967654986,
+      "grad_norm": 0.5994908213615417,
+      "learning_rate": 0.0005242568807339449,
+      "loss": 3.7508,
       "step": 11800
     },
     {
-      "epoch": 1.275427833387149,
-      "grad_norm": 0.6571457386016846,
-      "learning_rate": 0.0005240577523973709,
-      "loss": 3.7349,
+      "epoch": 1.2776280323450135,
+      "grad_norm": 0.5809037685394287,
+      "learning_rate": 0.0005239330814894765,
+      "loss": 3.7605,
       "step": 11850
     },
     {
-      "epoch": 1.280809385426757,
-      "grad_norm": 0.5685673952102661,
-      "learning_rate": 0.0005237345113673095,
-      "loss": 3.7483,
+      "epoch": 1.2830188679245282,
+      "grad_norm": 0.6419063806533813,
+      "learning_rate": 0.000523609282245008,
+      "loss": 3.7474,
       "step": 11900
     },
     {
-      "epoch": 1.2861909374663654,
-      "grad_norm": 0.5823640823364258,
-      "learning_rate": 0.0005234112703372481,
-      "loss": 3.7362,
+      "epoch": 1.2884097035040432,
+      "grad_norm": 0.5255048871040344,
+      "learning_rate": 0.0005232854830005396,
+      "loss": 3.7593,
       "step": 11950
     },
     {
-      "epoch": 1.2915724895059735,
-      "grad_norm": 0.5654664039611816,
-      "learning_rate": 0.0005230880293071867,
-      "loss": 3.7448,
+      "epoch": 1.2938005390835579,
+      "grad_norm": 0.6410238146781921,
+      "learning_rate": 0.0005229616837560712,
+      "loss": 3.7548,
       "step": 12000
     },
     {
-      "epoch": 1.2915724895059735,
-      "eval_accuracy": 0.34880749652772536,
-      "eval_loss": 3.703580856323242,
-      "eval_runtime": 183.5771,
-      "eval_samples_per_second": 98.111,
-      "eval_steps_per_second": 6.134,
+      "epoch": 1.2938005390835579,
+      "eval_accuracy": 0.3474446633558683,
+      "eval_loss": 3.7138984203338623,
+      "eval_runtime": 153.1227,
+      "eval_samples_per_second": 117.625,
+      "eval_steps_per_second": 7.354,
       "step": 12000
     },
     {
-      "epoch": 1.2969540415455818,
-      "grad_norm": 0.5596755146980286,
-      "learning_rate": 0.0005227647882771253,
-      "loss": 3.7173,
+      "epoch": 1.2991913746630728,
+      "grad_norm": 0.547203004360199,
+      "learning_rate": 0.0005226378845116028,
+      "loss": 3.7502,
       "step": 12050
     },
     {
-      "epoch": 1.30233559358519,
-      "grad_norm": 0.6434191465377808,
-      "learning_rate": 0.0005224415472470639,
-      "loss": 3.7368,
+      "epoch": 1.3045822102425877,
+      "grad_norm": 0.5448917150497437,
+      "learning_rate": 0.0005223140852671344,
+      "loss": 3.7572,
       "step": 12100
     },
     {
-      "epoch": 1.3077171456247982,
-      "grad_norm": 0.6069260835647583,
-      "learning_rate": 0.0005221183062170024,
-      "loss": 3.7265,
+      "epoch": 1.3099730458221024,
+      "grad_norm": 0.6201884150505066,
+      "learning_rate": 0.0005219902860226659,
+      "loss": 3.774,
       "step": 12150
     },
     {
-      "epoch": 1.3130986976644063,
-      "grad_norm": 0.6183269023895264,
-      "learning_rate": 0.0005217950651869409,
-      "loss": 3.7278,
+      "epoch": 1.3153638814016173,
+      "grad_norm": 0.589424729347229,
+      "learning_rate": 0.0005216664867781975,
+      "loss": 3.744,
       "step": 12200
     },
     {
-      "epoch": 1.3184802497040147,
-      "grad_norm": 0.5558634996414185,
-      "learning_rate": 0.0005214718241568796,
-      "loss": 3.7341,
+      "epoch": 1.320754716981132,
+      "grad_norm": 0.5179701447486877,
+      "learning_rate": 0.000521342687533729,
+      "loss": 3.7497,
       "step": 12250
     },
     {
-      "epoch": 1.3238618017436228,
-      "grad_norm": 0.6996021270751953,
-      "learning_rate": 0.0005211485831268182,
-      "loss": 3.7268,
+      "epoch": 1.326145552560647,
+      "grad_norm": 0.5681805610656738,
+      "learning_rate": 0.0005210188882892606,
+      "loss": 3.7606,
       "step": 12300
     },
     {
-      "epoch": 1.329243353783231,
-      "grad_norm": 0.5907831788063049,
-      "learning_rate": 0.0005208253420967568,
-      "loss": 3.7281,
+      "epoch": 1.3315363881401616,
+      "grad_norm": 0.6115100383758545,
+      "learning_rate": 0.0005206950890447922,
+      "loss": 3.7477,
       "step": 12350
     },
     {
-      "epoch": 1.3346249058228392,
-      "grad_norm": 0.5960204601287842,
-      "learning_rate": 0.0005205021010666953,
-      "loss": 3.7284,
+      "epoch": 1.3369272237196765,
+      "grad_norm": 0.6114750504493713,
+      "learning_rate": 0.0005203712898003238,
+      "loss": 3.738,
       "step": 12400
     },
     {
-      "epoch": 1.3400064578624475,
-      "grad_norm": 0.6140725612640381,
-      "learning_rate": 0.0005201788600366339,
-      "loss": 3.7348,
+      "epoch": 1.3423180592991915,
+      "grad_norm": 0.5691312551498413,
+      "learning_rate": 0.0005200474905558553,
+      "loss": 3.7491,
       "step": 12450
     },
     {
-      "epoch": 1.3453880099020559,
-      "grad_norm": 0.6222954988479614,
-      "learning_rate": 0.0005198556190065725,
-      "loss": 3.7337,
+      "epoch": 1.3477088948787062,
+      "grad_norm": 0.8015425801277161,
+      "learning_rate": 0.0005197236913113869,
+      "loss": 3.7322,
       "step": 12500
     },
     {
-      "epoch": 1.350769561941664,
-      "grad_norm": 0.5547105669975281,
-      "learning_rate": 0.0005195323779765112,
-      "loss": 3.7219,
+      "epoch": 1.353099730458221,
+      "grad_norm": 0.5946375131607056,
+      "learning_rate": 0.0005193998920669184,
+      "loss": 3.7453,
       "step": 12550
     },
     {
-      "epoch": 1.356151113981272,
-      "grad_norm": Infinity,
-      "learning_rate": 0.000519215601767051,
-      "loss": 3.746,
+      "epoch": 1.3584905660377358,
+      "grad_norm": 0.5824279189109802,
+      "learning_rate": 0.0005190760928224501,
+      "loss": 3.7389,
       "step": 12600
     },
     {
-      "epoch": 1.3615326660208804,
-      "grad_norm": 0.5531308054924011,
-      "learning_rate": 0.0005188923607369895,
-      "loss": 3.7283,
+      "epoch": 1.3638814016172507,
+      "grad_norm": 0.5873216986656189,
+      "learning_rate": 0.0005187522935779816,
+      "loss": 3.7272,
       "step": 12650
     },
     {
-      "epoch": 1.3669142180604887,
-      "grad_norm": 0.5557667016983032,
-      "learning_rate": 0.000518569119706928,
-      "loss": 3.7145,
+      "epoch": 1.3692722371967654,
+      "grad_norm": 0.5877545475959778,
+      "learning_rate": 0.0005184284943335132,
+      "loss": 3.728,
       "step": 12700
     },
     {
-      "epoch": 1.3722957701000968,
-      "grad_norm": 0.5849653482437134,
-      "learning_rate": 0.0005182458786768667,
-      "loss": 3.7303,
+      "epoch": 1.3746630727762803,
+      "grad_norm": 0.644077479839325,
+      "learning_rate": 0.0005181046950890447,
+      "loss": 3.745,
       "step": 12750
     },
     {
-      "epoch": 1.3776773221397052,
-      "grad_norm": 0.5382705926895142,
-      "learning_rate": 0.0005179226376468052,
-      "loss": 3.7238,
+      "epoch": 1.3800539083557952,
+      "grad_norm": 0.5267413258552551,
+      "learning_rate": 0.0005177808958445764,
+      "loss": 3.7234,
       "step": 12800
     },
     {
-      "epoch": 1.3830588741793133,
-      "grad_norm": 0.5681452751159668,
-      "learning_rate": 0.0005175993966167438,
-      "loss": 3.7338,
+      "epoch": 1.38544474393531,
+      "grad_norm": 0.6268362998962402,
+      "learning_rate": 0.0005174570966001078,
+      "loss": 3.7362,
       "step": 12850
     },
     {
-      "epoch": 1.3884404262189216,
-      "grad_norm": 0.5677098631858826,
-      "learning_rate": 0.0005172761555866824,
-      "loss": 3.7381,
+      "epoch": 1.3908355795148248,
+      "grad_norm": 0.5079869031906128,
+      "learning_rate": 0.0005171332973556395,
+      "loss": 3.7359,
       "step": 12900
     },
     {
-      "epoch": 1.3938219782585297,
-      "grad_norm": 0.5775512456893921,
-      "learning_rate": 0.0005169529145566211,
-      "loss": 3.713,
+      "epoch": 1.3962264150943398,
+      "grad_norm": 0.5891706943511963,
+      "learning_rate": 0.000516809498111171,
+      "loss": 3.7284,
       "step": 12950
     },
     {
-      "epoch": 1.399203530298138,
-      "grad_norm": 0.6128884553909302,
-      "learning_rate": 0.0005166296735265596,
-      "loss": 3.7293,
+      "epoch": 1.4016172506738545,
+      "grad_norm": 0.5170325040817261,
+      "learning_rate": 0.0005164856988667026,
+      "loss": 3.7231,
       "step": 13000
     },
     {
-      "epoch": 1.399203530298138,
-      "eval_accuracy": 0.35082072588603985,
-      "eval_loss": 3.6829423904418945,
-      "eval_runtime": 183.7457,
-      "eval_samples_per_second": 98.021,
-      "eval_steps_per_second": 6.128,
+      "epoch": 1.4016172506738545,
+      "eval_accuracy": 0.3502823508269191,
+      "eval_loss": 3.687718152999878,
+      "eval_runtime": 153.5069,
+      "eval_samples_per_second": 117.33,
+      "eval_steps_per_second": 7.335,
       "step": 13000
     },
     {
-      "epoch": 1.4045850823377461,
-      "grad_norm": 0.5482479333877563,
-      "learning_rate": 0.0005163064324964982,
-      "loss": 3.7239,
+      "epoch": 1.4070080862533692,
+      "grad_norm": 0.5408639907836914,
+      "learning_rate": 0.0005161618996222341,
+      "loss": 3.7026,
       "step": 13050
     },
     {
-      "epoch": 1.4099666343773545,
-      "grad_norm": 0.5541461706161499,
-      "learning_rate": 0.0005159831914664367,
-      "loss": 3.711,
+      "epoch": 1.412398921832884,
+      "grad_norm": 0.5476001501083374,
+      "learning_rate": 0.0005158381003777657,
+      "loss": 3.7237,
       "step": 13100
     },
     {
-      "epoch": 1.4153481864169626,
-      "grad_norm": 0.6503528952598572,
-      "learning_rate": 0.0005156599504363753,
-      "loss": 3.7191,
+      "epoch": 1.417789757412399,
+      "grad_norm": 0.6275716423988342,
+      "learning_rate": 0.0005155143011332973,
+      "loss": 3.7267,
       "step": 13150
     },
     {
-      "epoch": 1.420729738456571,
-      "grad_norm": 0.5830383896827698,
-      "learning_rate": 0.000515336709406314,
-      "loss": 3.7161,
+      "epoch": 1.4231805929919137,
+      "grad_norm": 0.5737962126731873,
+      "learning_rate": 0.0005151905018888289,
+      "loss": 3.7082,
       "step": 13200
     },
     {
-      "epoch": 1.426111290496179,
-      "grad_norm": 0.5402637124061584,
-      "learning_rate": 0.0005150134683762525,
-      "loss": 3.7054,
+      "epoch": 1.4285714285714286,
+      "grad_norm": 0.5455244183540344,
+      "learning_rate": 0.0005148667026443604,
+      "loss": 3.7278,
       "step": 13250
     },
     {
-      "epoch": 1.4314928425357873,
-      "grad_norm": 0.5502687096595764,
-      "learning_rate": 0.0005146902273461911,
-      "loss": 3.7027,
+      "epoch": 1.4339622641509435,
+      "grad_norm": 0.617461085319519,
+      "learning_rate": 0.000514542903399892,
+      "loss": 3.722,
       "step": 13300
     },
     {
-      "epoch": 1.4368743945753955,
-      "grad_norm": 0.7305072546005249,
-      "learning_rate": 0.0005143669863161297,
-      "loss": 3.7138,
+      "epoch": 1.4393530997304582,
+      "grad_norm": 0.6026687026023865,
+      "learning_rate": 0.0005142191041554237,
+      "loss": 3.7293,
       "step": 13350
     },
     {
-      "epoch": 1.4422559466150038,
-      "grad_norm": 0.554663896560669,
-      "learning_rate": 0.0005140437452860683,
-      "loss": 3.705,
+      "epoch": 1.444743935309973,
+      "grad_norm": 0.5842117071151733,
+      "learning_rate": 0.0005138953049109552,
+      "loss": 3.7174,
       "step": 13400
     },
     {
-      "epoch": 1.447637498654612,
-      "grad_norm": 0.5779620409011841,
-      "learning_rate": 0.0005137205042560069,
-      "loss": 3.7073,
+      "epoch": 1.4501347708894878,
+      "grad_norm": 0.5006920099258423,
+      "learning_rate": 0.0005135715056664868,
+      "loss": 3.7323,
       "step": 13450
     },
     {
-      "epoch": 1.4530190506942202,
-      "grad_norm": 0.5610778331756592,
-      "learning_rate": 0.0005133972632259455,
-      "loss": 3.7048,
+      "epoch": 1.4555256064690028,
+      "grad_norm": 0.6773222088813782,
+      "learning_rate": 0.0005132477064220183,
+      "loss": 3.7261,
       "step": 13500
     },
     {
-      "epoch": 1.4584006027338283,
-      "grad_norm": 0.6052946448326111,
-      "learning_rate": 0.000513074022195884,
-      "loss": 3.7203,
+      "epoch": 1.4609164420485174,
+      "grad_norm": 0.5614179968833923,
+      "learning_rate": 0.0005129239071775499,
+      "loss": 3.703,
       "step": 13550
     },
     {
-      "epoch": 1.4637821547734367,
-      "grad_norm": 0.6141031384468079,
-      "learning_rate": 0.0005127507811658226,
-      "loss": 3.6992,
+      "epoch": 1.4663072776280324,
+      "grad_norm": 0.5550870299339294,
+      "learning_rate": 0.0005126001079330814,
+      "loss": 3.7336,
       "step": 13600
     },
     {
-      "epoch": 1.469163706813045,
-      "grad_norm": 0.5198776125907898,
-      "learning_rate": 0.0005124275401357612,
-      "loss": 3.7027,
+      "epoch": 1.4716981132075473,
+      "grad_norm": 0.5358152389526367,
+      "learning_rate": 0.000512276308688613,
+      "loss": 3.7126,
       "step": 13650
     },
     {
-      "epoch": 1.474545258852653,
-      "grad_norm": 0.5567837953567505,
-      "learning_rate": 0.0005121042991056997,
-      "loss": 3.7215,
+      "epoch": 1.477088948787062,
+      "grad_norm": 0.6187070608139038,
+      "learning_rate": 0.0005119525094441446,
+      "loss": 3.731,
       "step": 13700
     },
     {
-      "epoch": 1.4799268108922612,
-      "grad_norm": 0.5448922514915466,
-      "learning_rate": 0.0005117810580756384,
-      "loss": 3.6968,
+      "epoch": 1.482479784366577,
+      "grad_norm": 0.5459868907928467,
+      "learning_rate": 0.0005116287101996762,
+      "loss": 3.7033,
       "step": 13750
     },
     {
-      "epoch": 1.4853083629318695,
-      "grad_norm": 0.5946159958839417,
-      "learning_rate": 0.0005114578170455769,
-      "loss": 3.7088,
+      "epoch": 1.4878706199460916,
+      "grad_norm": 0.5670978426933289,
+      "learning_rate": 0.0005113049109552077,
+      "loss": 3.7285,
       "step": 13800
     },
     {
-      "epoch": 1.4906899149714778,
-      "grad_norm": 0.5692389011383057,
-      "learning_rate": 0.0005111345760155156,
-      "loss": 3.7093,
+      "epoch": 1.4932614555256065,
+      "grad_norm": 0.6740704774856567,
+      "learning_rate": 0.0005109811117107393,
+      "loss": 3.7142,
       "step": 13850
     },
     {
-      "epoch": 1.496071467011086,
-      "grad_norm": 0.5627337694168091,
-      "learning_rate": 0.0005108113349854541,
-      "loss": 3.7001,
+      "epoch": 1.4986522911051212,
+      "grad_norm": 0.572716236114502,
+      "learning_rate": 0.0005106573124662708,
+      "loss": 3.7217,
       "step": 13900
     },
     {
-      "epoch": 1.501453019050694,
-      "grad_norm": 0.6168450713157654,
-      "learning_rate": 0.0005104880939553926,
-      "loss": 3.7138,
+      "epoch": 1.5040431266846361,
+      "grad_norm": 0.6068322062492371,
+      "learning_rate": 0.0005103335132218025,
+      "loss": 3.7116,
       "step": 13950
     },
     {
-      "epoch": 1.5068345710903024,
-      "grad_norm": 0.5749646425247192,
-      "learning_rate": 0.0005101648529253313,
-      "loss": 3.705,
+      "epoch": 1.509433962264151,
+      "grad_norm": 0.5378690361976624,
+      "learning_rate": 0.000510009713977334,
+      "loss": 3.7126,
       "step": 14000
     },
     {
-      "epoch": 1.5068345710903024,
-      "eval_accuracy": 0.35293130823183616,
-      "eval_loss": 3.662102222442627,
-      "eval_runtime": 183.8694,
-      "eval_samples_per_second": 97.955,
-      "eval_steps_per_second": 6.124,
+      "epoch": 1.509433962264151,
+      "eval_accuracy": 0.35229797054876555,
+      "eval_loss": 3.667318820953369,
+      "eval_runtime": 153.4116,
+      "eval_samples_per_second": 117.403,
+      "eval_steps_per_second": 7.34,
       "step": 14000
     },
     {
-      "epoch": 1.5122161231299107,
-      "grad_norm": 0.5757107734680176,
-      "learning_rate": 0.0005098416118952699,
-      "loss": 3.6945,
+      "epoch": 1.5148247978436657,
+      "grad_norm": 0.5783511996269226,
+      "learning_rate": 0.0005096859147328656,
+      "loss": 3.7083,
       "step": 14050
     },
     {
-      "epoch": 1.5175976751695188,
-      "grad_norm": 0.5482128262519836,
-      "learning_rate": 0.0005095183708652085,
-      "loss": 3.686,
+      "epoch": 1.5202156334231804,
+      "grad_norm": 0.589012861251831,
+      "learning_rate": 0.0005093621154883971,
+      "loss": 3.7083,
       "step": 14100
     },
     {
-      "epoch": 1.5229792272091272,
-      "grad_norm": 0.7440946102142334,
-      "learning_rate": 0.000509195129835147,
-      "loss": 3.7081,
+      "epoch": 1.5256064690026954,
+      "grad_norm": 0.5106600522994995,
+      "learning_rate": 0.0005090383162439288,
+      "loss": 3.7009,
       "step": 14150
     },
     {
-      "epoch": 1.5283607792487355,
-      "grad_norm": 0.5398544669151306,
-      "learning_rate": 0.0005088718888050856,
-      "loss": 3.7163,
+      "epoch": 1.5309973045822103,
+      "grad_norm": 0.6712096333503723,
+      "learning_rate": 0.0005087145169994602,
+      "loss": 3.7006,
       "step": 14200
     },
     {
-      "epoch": 1.5337423312883436,
-      "grad_norm": 0.6044565439224243,
-      "learning_rate": 0.0005085486477750242,
-      "loss": 3.6954,
+      "epoch": 1.536388140161725,
+      "grad_norm": 0.5968721508979797,
+      "learning_rate": 0.0005083907177549918,
+      "loss": 3.716,
       "step": 14250
     },
     {
-      "epoch": 1.5391238833279517,
-      "grad_norm": 0.5324616432189941,
-      "learning_rate": 0.0005082254067449629,
-      "loss": 3.6999,
+      "epoch": 1.54177897574124,
+      "grad_norm": 0.5645915865898132,
+      "learning_rate": 0.0005080669185105234,
+      "loss": 3.7179,
       "step": 14300
     },
     {
-      "epoch": 1.54450543536756,
-      "grad_norm": 0.6088085174560547,
-      "learning_rate": 0.0005079021657149014,
-      "loss": 3.7096,
+      "epoch": 1.5471698113207548,
+      "grad_norm": 0.5808230638504028,
+      "learning_rate": 0.000507743119266055,
+      "loss": 3.7097,
       "step": 14350
     },
     {
-      "epoch": 1.5498869874071683,
-      "grad_norm": 0.5295168161392212,
-      "learning_rate": 0.0005075789246848399,
-      "loss": 3.6895,
+      "epoch": 1.5525606469002695,
+      "grad_norm": 0.5534523129463196,
+      "learning_rate": 0.0005074193200215865,
+      "loss": 3.724,
       "step": 14400
     },
     {
-      "epoch": 1.5552685394467765,
-      "grad_norm": 0.614923894405365,
-      "learning_rate": 0.0005072556836547785,
-      "loss": 3.7028,
+      "epoch": 1.5579514824797842,
+      "grad_norm": 0.5474613308906555,
+      "learning_rate": 0.0005070955207771181,
+      "loss": 3.7154,
       "step": 14450
     },
     {
-      "epoch": 1.5606500914863846,
-      "grad_norm": 0.5911874175071716,
-      "learning_rate": 0.0005069324426247171,
-      "loss": 3.6933,
+      "epoch": 1.5633423180592994,
+      "grad_norm": 0.5958494544029236,
+      "learning_rate": 0.0005067717215326498,
+      "loss": 3.6964,
       "step": 14500
     },
     {
-      "epoch": 1.566031643525993,
-      "grad_norm": 0.5018694400787354,
-      "learning_rate": 0.0005066092015946557,
-      "loss": 3.6953,
+      "epoch": 1.568733153638814,
+      "grad_norm": 0.5638560652732849,
+      "learning_rate": 0.0005064479222881813,
+      "loss": 3.7124,
       "step": 14550
     },
     {
-      "epoch": 1.5714131955656012,
-      "grad_norm": 0.5696578621864319,
-      "learning_rate": 0.0005062859605645943,
-      "loss": 3.6944,
+      "epoch": 1.5741239892183287,
+      "grad_norm": 0.5805913209915161,
+      "learning_rate": 0.0005061241230437129,
+      "loss": 3.6964,
       "step": 14600
     },
     {
-      "epoch": 1.5767947476052093,
-      "grad_norm": 0.5895256996154785,
-      "learning_rate": 0.0005059691843551341,
-      "loss": 3.677,
+      "epoch": 1.5795148247978437,
+      "grad_norm": 0.5940707325935364,
+      "learning_rate": 0.0005058003237992444,
+      "loss": 3.6956,
       "step": 14650
     },
     {
-      "epoch": 1.5821762996448174,
-      "grad_norm": 0.5158836245536804,
-      "learning_rate": 0.0005056459433250727,
-      "loss": 3.6883,
+      "epoch": 1.5849056603773586,
+      "grad_norm": 0.5153263211250305,
+      "learning_rate": 0.0005054830005396654,
+      "loss": 3.7099,
       "step": 14700
     },
     {
-      "epoch": 1.5875578516844258,
-      "grad_norm": 0.5363539457321167,
-      "learning_rate": 0.0005053227022950113,
-      "loss": 3.6828,
+      "epoch": 1.5902964959568733,
+      "grad_norm": 0.5778594017028809,
+      "learning_rate": 0.0005051592012951969,
+      "loss": 3.7112,
       "step": 14750
     },
     {
-      "epoch": 1.592939403724034,
-      "grad_norm": 0.5499960780143738,
-      "learning_rate": 0.0005049994612649499,
-      "loss": 3.6859,
+      "epoch": 1.595687331536388,
+      "grad_norm": 0.532878041267395,
+      "learning_rate": 0.0005048354020507286,
+      "loss": 3.6816,
       "step": 14800
     },
     {
-      "epoch": 1.5983209557636422,
-      "grad_norm": 0.5901286005973816,
-      "learning_rate": 0.0005046762202348884,
-      "loss": 3.6829,
+      "epoch": 1.6010781671159031,
+      "grad_norm": 0.5925537943840027,
+      "learning_rate": 0.00050451160280626,
+      "loss": 3.7016,
       "step": 14850
     },
     {
-      "epoch": 1.6037025078032503,
-      "grad_norm": 0.5542553663253784,
-      "learning_rate": 0.000504352979204827,
-      "loss": 3.6978,
+      "epoch": 1.6064690026954178,
+      "grad_norm": 0.5424179434776306,
+      "learning_rate": 0.0005041878035617917,
+      "loss": 3.7079,
       "step": 14900
     },
     {
-      "epoch": 1.6090840598428586,
-      "grad_norm": 0.5885649919509888,
-      "learning_rate": 0.0005040297381747656,
-      "loss": 3.6685,
+      "epoch": 1.6118598382749325,
+      "grad_norm": 0.5615659356117249,
+      "learning_rate": 0.0005038640043173232,
+      "loss": 3.6983,
       "step": 14950
     },
     {
-      "epoch": 1.614465611882467,
-      "grad_norm": 0.6301882266998291,
-      "learning_rate": 0.0005037064971447042,
-      "loss": 3.6789,
+      "epoch": 1.6172506738544474,
+      "grad_norm": 0.5864365696907043,
+      "learning_rate": 0.0005035402050728548,
+      "loss": 3.6994,
       "step": 15000
     },
     {
-      "epoch": 1.614465611882467,
-      "eval_accuracy": 0.3545692505156394,
-      "eval_loss": 3.641286849975586,
-      "eval_runtime": 183.7066,
-      "eval_samples_per_second": 98.042,
-      "eval_steps_per_second": 6.129,
+      "epoch": 1.6172506738544474,
+      "eval_accuracy": 0.3544227864228655,
+      "eval_loss": 3.6493024826049805,
+      "eval_runtime": 153.1831,
+      "eval_samples_per_second": 117.578,
+      "eval_steps_per_second": 7.351,
       "step": 15000
     },
     {
-      "epoch": 1.619847163922075,
-      "grad_norm": 0.5616388916969299,
-      "learning_rate": 0.0005033832561146428,
-      "loss": 3.6855,
+      "epoch": 1.6226415094339623,
+      "grad_norm": 0.5703364610671997,
+      "learning_rate": 0.0005032164058283863,
+      "loss": 3.6888,
       "step": 15050
     },
     {
-      "epoch": 1.6252287159616834,
-      "grad_norm": 0.5974994897842407,
-      "learning_rate": 0.0005030600150845813,
-      "loss": 3.6761,
+      "epoch": 1.628032345013477,
+      "grad_norm": 0.539567232131958,
+      "learning_rate": 0.0005028926065839179,
+      "loss": 3.6832,
       "step": 15100
     },
     {
-      "epoch": 1.6306102680012917,
-      "grad_norm": 0.5747628211975098,
-      "learning_rate": 0.00050273677405452,
-      "loss": 3.6876,
+      "epoch": 1.633423180592992,
+      "grad_norm": 0.5440678000450134,
+      "learning_rate": 0.0005025688073394495,
+      "loss": 3.6811,
       "step": 15150
     },
     {
-      "epoch": 1.6359918200408998,
-      "grad_norm": 0.5737857222557068,
-      "learning_rate": 0.0005024135330244585,
-      "loss": 3.673,
+      "epoch": 1.6388140161725069,
+      "grad_norm": 0.5952890515327454,
+      "learning_rate": 0.0005022450080949811,
+      "loss": 3.6861,
       "step": 15200
     },
     {
-      "epoch": 1.641373372080508,
-      "grad_norm": 0.5631479620933533,
-      "learning_rate": 0.0005020902919943972,
-      "loss": 3.6691,
+      "epoch": 1.6442048517520216,
+      "grad_norm": 0.5702643990516663,
+      "learning_rate": 0.0005019212088505126,
+      "loss": 3.681,
       "step": 15250
     },
     {
-      "epoch": 1.6467549241201163,
-      "grad_norm": 0.5611960887908936,
-      "learning_rate": 0.0005017670509643357,
-      "loss": 3.6705,
+      "epoch": 1.6495956873315363,
+      "grad_norm": 0.586983323097229,
+      "learning_rate": 0.0005015974096060442,
+      "loss": 3.6872,
       "step": 15300
     },
     {
-      "epoch": 1.6521364761597246,
-      "grad_norm": 0.6078739166259766,
-      "learning_rate": 0.0005014438099342743,
-      "loss": 3.6812,
+      "epoch": 1.6549865229110512,
+      "grad_norm": 0.5341205596923828,
+      "learning_rate": 0.0005012736103615758,
+      "loss": 3.7083,
       "step": 15350
     },
     {
-      "epoch": 1.6575180281993327,
-      "grad_norm": 0.5565915107727051,
-      "learning_rate": 0.0005011205689042129,
-      "loss": 3.6809,
+      "epoch": 1.6603773584905661,
+      "grad_norm": 0.6124740242958069,
+      "learning_rate": 0.0005009498111171074,
+      "loss": 3.6987,
       "step": 15400
     },
     {
-      "epoch": 1.6628995802389408,
-      "grad_norm": 0.5558174848556519,
-      "learning_rate": 0.0005007973278741514,
-      "loss": 3.6696,
+      "epoch": 1.6657681940700808,
+      "grad_norm": 0.592282235622406,
+      "learning_rate": 0.0005006260118726389,
+      "loss": 3.6669,
       "step": 15450
     },
     {
-      "epoch": 1.6682811322785491,
-      "grad_norm": 0.5639570355415344,
-      "learning_rate": 0.00050047408684409,
-      "loss": 3.6985,
+      "epoch": 1.6711590296495957,
+      "grad_norm": 0.6049423217773438,
+      "learning_rate": 0.0005003022126281705,
+      "loss": 3.6851,
       "step": 15500
     },
     {
-      "epoch": 1.6736626843181575,
-      "grad_norm": 0.5616956949234009,
-      "learning_rate": 0.0005001508458140286,
-      "loss": 3.6601,
+      "epoch": 1.6765498652291106,
+      "grad_norm": 0.5736304521560669,
+      "learning_rate": 0.000499978413383702,
+      "loss": 3.6793,
       "step": 15550
     },
     {
-      "epoch": 1.6790442363577656,
-      "grad_norm": 0.5630506873130798,
-      "learning_rate": 0.0004998276047839673,
-      "loss": 3.6584,
+      "epoch": 1.6819407008086253,
+      "grad_norm": 0.5263177752494812,
+      "learning_rate": 0.0004996546141392336,
+      "loss": 3.6688,
       "step": 15600
     },
     {
-      "epoch": 1.6844257883973737,
-      "grad_norm": 0.5845687985420227,
-      "learning_rate": 0.0004995043637539058,
-      "loss": 3.6828,
+      "epoch": 1.68733153638814,
+      "grad_norm": 0.5575107336044312,
+      "learning_rate": 0.0004993308148947651,
+      "loss": 3.6864,
       "step": 15650
     },
     {
-      "epoch": 1.689807340436982,
-      "grad_norm": 0.5357277989387512,
-      "learning_rate": 0.0004991811227238443,
-      "loss": 3.6791,
+      "epoch": 1.692722371967655,
+      "grad_norm": 0.5515274405479431,
+      "learning_rate": 0.0004990070156502968,
+      "loss": 3.6753,
       "step": 15700
     },
     {
-      "epoch": 1.6951888924765903,
-      "grad_norm": 0.574770987033844,
-      "learning_rate": 0.0004988578816937829,
-      "loss": 3.6924,
+      "epoch": 1.6981132075471699,
+      "grad_norm": 0.5717898011207581,
+      "learning_rate": 0.0004986832164058284,
+      "loss": 3.7009,
       "step": 15750
     },
     {
-      "epoch": 1.7005704445161984,
-      "grad_norm": 0.5270922183990479,
-      "learning_rate": 0.0004985346406637215,
-      "loss": 3.6946,
+      "epoch": 1.7035040431266846,
+      "grad_norm": 0.5158862471580505,
+      "learning_rate": 0.0004983594171613599,
+      "loss": 3.6948,
       "step": 15800
     },
     {
-      "epoch": 1.7059519965558065,
-      "grad_norm": 0.5539059638977051,
-      "learning_rate": 0.0004982113996336602,
-      "loss": 3.6825,
+      "epoch": 1.7088948787061995,
+      "grad_norm": 0.5187628865242004,
+      "learning_rate": 0.0004980356179168915,
+      "loss": 3.6904,
       "step": 15850
     },
     {
-      "epoch": 1.7113335485954149,
-      "grad_norm": 0.5775881409645081,
-      "learning_rate": 0.0004978881586035987,
-      "loss": 3.6685,
+      "epoch": 1.7142857142857144,
+      "grad_norm": 0.5600886940956116,
+      "learning_rate": 0.000497711818672423,
+      "loss": 3.6883,
       "step": 15900
     },
     {
-      "epoch": 1.7167151006350232,
-      "grad_norm": 0.5449931025505066,
-      "learning_rate": 0.0004975649175735373,
-      "loss": 3.6627,
+      "epoch": 1.719676549865229,
+      "grad_norm": 0.55384361743927,
+      "learning_rate": 0.0004973880194279547,
+      "loss": 3.6751,
       "step": 15950
     },
     {
-      "epoch": 1.7220966526746313,
-      "grad_norm": 0.5795072317123413,
-      "learning_rate": 0.0004972416765434759,
-      "loss": 3.6831,
+      "epoch": 1.7250673854447438,
+      "grad_norm": 0.6270291805267334,
+      "learning_rate": 0.0004970642201834862,
+      "loss": 3.6737,
       "step": 16000
     },
     {
-      "epoch": 1.7220966526746313,
-      "eval_accuracy": 0.35655455608178566,
-      "eval_loss": 3.624427318572998,
-      "eval_runtime": 183.5278,
-      "eval_samples_per_second": 98.138,
-      "eval_steps_per_second": 6.135,
+      "epoch": 1.7250673854447438,
+      "eval_accuracy": 0.3562606500202475,
+      "eval_loss": 3.6298611164093018,
+      "eval_runtime": 153.4578,
+      "eval_samples_per_second": 117.368,
+      "eval_steps_per_second": 7.338,
       "step": 16000
     },
     {
-      "epoch": 1.7274782047142396,
-      "grad_norm": 0.6113314628601074,
-      "learning_rate": 0.0004969184355134145,
-      "loss": 3.656,
+      "epoch": 1.7304582210242587,
+      "grad_norm": 0.5840327739715576,
+      "learning_rate": 0.0004967404209390178,
+      "loss": 3.6739,
       "step": 16050
     },
     {
-      "epoch": 1.732859756753848,
-      "grad_norm": 0.5706090331077576,
-      "learning_rate": 0.0004965951944833531,
-      "loss": 3.6796,
+      "epoch": 1.7358490566037736,
+      "grad_norm": 0.5414550304412842,
+      "learning_rate": 0.0004964166216945493,
+      "loss": 3.6603,
       "step": 16100
     },
     {
-      "epoch": 1.738241308793456,
-      "grad_norm": 0.5592207908630371,
-      "learning_rate": 0.0004962719534532916,
-      "loss": 3.6769,
+      "epoch": 1.7412398921832883,
+      "grad_norm": 0.5441046953201294,
+      "learning_rate": 0.000496092822450081,
+      "loss": 3.6829,
       "step": 16150
     },
     {
-      "epoch": 1.7436228608330642,
-      "grad_norm": 0.5829014182090759,
-      "learning_rate": 0.0004959487124232302,
-      "loss": 3.6614,
+      "epoch": 1.7466307277628033,
+      "grad_norm": 0.5752876400947571,
+      "learning_rate": 0.0004957690232056125,
+      "loss": 3.694,
       "step": 16200
     },
     {
-      "epoch": 1.7490044128726725,
-      "grad_norm": 0.5765231847763062,
-      "learning_rate": 0.0004956254713931688,
-      "loss": 3.6814,
+      "epoch": 1.7520215633423182,
+      "grad_norm": 0.5660447478294373,
+      "learning_rate": 0.0004954452239611441,
+      "loss": 3.6711,
       "step": 16250
     },
     {
-      "epoch": 1.7543859649122808,
-      "grad_norm": 0.5730772614479065,
-      "learning_rate": 0.0004953022303631074,
-      "loss": 3.6816,
+      "epoch": 1.7574123989218329,
+      "grad_norm": 0.5629813075065613,
+      "learning_rate": 0.0004951214247166756,
+      "loss": 3.6921,
       "step": 16300
     },
     {
-      "epoch": 1.759767516951889,
-      "grad_norm": 0.5396572947502136,
-      "learning_rate": 0.0004949789893330459,
-      "loss": 3.6554,
+      "epoch": 1.7628032345013476,
+      "grad_norm": 0.5693957209587097,
+      "learning_rate": 0.0004947976254722072,
+      "loss": 3.6744,
       "step": 16350
     },
     {
-      "epoch": 1.765149068991497,
-      "grad_norm": 0.6133454442024231,
-      "learning_rate": 0.0004946557483029846,
-      "loss": 3.6513,
+      "epoch": 1.7681940700808625,
+      "grad_norm": 0.5386444330215454,
+      "learning_rate": 0.0004944738262277387,
+      "loss": 3.6686,
       "step": 16400
     },
     {
-      "epoch": 1.7705306210311054,
-      "grad_norm": 0.6089504957199097,
-      "learning_rate": 0.0004943325072729231,
-      "loss": 3.6547,
+      "epoch": 1.7735849056603774,
+      "grad_norm": 0.5628581047058105,
+      "learning_rate": 0.0004941500269832703,
+      "loss": 3.6724,
       "step": 16450
     },
     {
-      "epoch": 1.7759121730707137,
-      "grad_norm": 0.5463808178901672,
-      "learning_rate": 0.0004940092662428617,
-      "loss": 3.6564,
+      "epoch": 1.778975741239892,
+      "grad_norm": 0.6008003354072571,
+      "learning_rate": 0.0004938262277388019,
+      "loss": 3.6623,
       "step": 16500
     },
     {
-      "epoch": 1.7812937251103218,
-      "grad_norm": 0.5989441275596619,
-      "learning_rate": 0.0004936860252128003,
-      "loss": 3.666,
+      "epoch": 1.784366576819407,
+      "grad_norm": 0.6104196906089783,
+      "learning_rate": 0.0004935024284943335,
+      "loss": 3.6869,
       "step": 16550
     },
     {
-      "epoch": 1.78667527714993,
-      "grad_norm": 0.567065954208374,
-      "learning_rate": 0.0004933627841827388,
-      "loss": 3.6631,
+      "epoch": 1.789757412398922,
+      "grad_norm": 0.7222172617912292,
+      "learning_rate": 0.000493178629249865,
+      "loss": 3.6729,
       "step": 16600
     },
     {
-      "epoch": 1.7920568291895382,
-      "grad_norm": 0.5451539158821106,
-      "learning_rate": 0.0004930460079732786,
-      "loss": 3.6649,
+      "epoch": 1.7951482479784366,
+      "grad_norm": 0.5604591369628906,
+      "learning_rate": 0.0004928548300053966,
+      "loss": 3.6748,
       "step": 16650
     },
     {
-      "epoch": 1.7974383812291466,
-      "grad_norm": 0.5379535555839539,
-      "learning_rate": 0.0004927227669432173,
-      "loss": 3.6464,
+      "epoch": 1.8005390835579513,
+      "grad_norm": 0.5300571322441101,
+      "learning_rate": 0.0004925310307609282,
+      "loss": 3.6781,
       "step": 16700
     },
     {
-      "epoch": 1.8028199332687547,
-      "grad_norm": 0.6250214576721191,
-      "learning_rate": 0.0004923995259131558,
-      "loss": 3.6656,
+      "epoch": 1.8059299191374663,
+      "grad_norm": 0.5437871813774109,
+      "learning_rate": 0.0004922072315164598,
+      "loss": 3.666,
       "step": 16750
     },
     {
-      "epoch": 1.8082014853083628,
-      "grad_norm": 0.5901435017585754,
-      "learning_rate": 0.0004920762848830945,
-      "loss": 3.6638,
+      "epoch": 1.8113207547169812,
+      "grad_norm": 0.5072501301765442,
+      "learning_rate": 0.0004918834322719913,
+      "loss": 3.6649,
       "step": 16800
     },
     {
-      "epoch": 1.813583037347971,
-      "grad_norm": 0.5645632147789001,
-      "learning_rate": 0.000491753043853033,
-      "loss": 3.6667,
+      "epoch": 1.8167115902964959,
+      "grad_norm": 0.616094708442688,
+      "learning_rate": 0.0004915596330275229,
+      "loss": 3.6713,
       "step": 16850
     },
     {
-      "epoch": 1.8189645893875794,
-      "grad_norm": 0.5927627086639404,
-      "learning_rate": 0.0004914298028229717,
-      "loss": 3.6623,
+      "epoch": 1.8221024258760108,
+      "grad_norm": 0.5298705697059631,
+      "learning_rate": 0.0004912358337830544,
+      "loss": 3.6558,
       "step": 16900
     },
     {
-      "epoch": 1.8243461414271875,
-      "grad_norm": 0.5503095388412476,
-      "learning_rate": 0.0004911065617929102,
-      "loss": 3.6626,
+      "epoch": 1.8274932614555257,
+      "grad_norm": 0.556859016418457,
+      "learning_rate": 0.000490912034538586,
+      "loss": 3.6878,
       "step": 16950
     },
     {
-      "epoch": 1.8297276934667959,
-      "grad_norm": 0.5205283164978027,
-      "learning_rate": 0.0004907833207628487,
-      "loss": 3.6394,
+      "epoch": 1.8328840970350404,
+      "grad_norm": 0.5950144529342651,
+      "learning_rate": 0.0004905882352941175,
+      "loss": 3.6732,
       "step": 17000
     },
     {
-      "epoch": 1.8297276934667959,
-      "eval_accuracy": 0.3581703331764748,
-      "eval_loss": 3.607296943664551,
-      "eval_runtime": 183.4429,
-      "eval_samples_per_second": 98.183,
-      "eval_steps_per_second": 6.138,
+      "epoch": 1.8328840970350404,
+      "eval_accuracy": 0.3574870151650095,
+      "eval_loss": 3.6185386180877686,
+      "eval_runtime": 153.1468,
+      "eval_samples_per_second": 117.606,
+      "eval_steps_per_second": 7.352,
       "step": 17000
     },
     {
-      "epoch": 1.8351092455064042,
-      "grad_norm": 0.5636503100395203,
-      "learning_rate": 0.0004904600797327874,
-      "loss": 3.6651,
+      "epoch": 1.838274932614555,
+      "grad_norm": 0.5314426422119141,
+      "learning_rate": 0.0004902644360496492,
+      "loss": 3.6536,
       "step": 17050
     },
     {
-      "epoch": 1.8404907975460123,
-      "grad_norm": 0.588615357875824,
-      "learning_rate": 0.0004901368387027259,
-      "loss": 3.6401,
+      "epoch": 1.8436657681940702,
+      "grad_norm": 0.5572214126586914,
+      "learning_rate": 0.0004899406368051808,
+      "loss": 3.6686,
       "step": 17100
     },
     {
-      "epoch": 1.8458723495856204,
-      "grad_norm": 0.5810367465019226,
-      "learning_rate": 0.0004898135976726646,
-      "loss": 3.6499,
+      "epoch": 1.849056603773585,
+      "grad_norm": 0.5513522624969482,
+      "learning_rate": 0.0004896168375607123,
+      "loss": 3.6638,
       "step": 17150
     },
     {
-      "epoch": 1.8512539016252287,
-      "grad_norm": 0.5747438073158264,
-      "learning_rate": 0.0004894903566426031,
-      "loss": 3.6557,
+      "epoch": 1.8544474393530996,
+      "grad_norm": 0.8657709956169128,
+      "learning_rate": 0.0004892930383162439,
+      "loss": 3.6495,
       "step": 17200
     },
     {
-      "epoch": 1.856635453664837,
-      "grad_norm": 0.6331074237823486,
-      "learning_rate": 0.0004891671156125417,
-      "loss": 3.6344,
+      "epoch": 1.8598382749326146,
+      "grad_norm": 0.5733378529548645,
+      "learning_rate": 0.0004889692390717754,
+      "loss": 3.6719,
       "step": 17250
     },
     {
-      "epoch": 1.8620170057044452,
-      "grad_norm": 0.61235511302948,
-      "learning_rate": 0.0004888438745824803,
-      "loss": 3.642,
+      "epoch": 1.8652291105121295,
+      "grad_norm": 0.5828373432159424,
+      "learning_rate": 0.0004886454398273071,
+      "loss": 3.6609,
       "step": 17300
     },
     {
-      "epoch": 1.8673985577440533,
-      "grad_norm": 0.5766220092773438,
-      "learning_rate": 0.0004885206335524189,
-      "loss": 3.6547,
+      "epoch": 1.8706199460916442,
+      "grad_norm": 0.7197691202163696,
+      "learning_rate": 0.0004883216405828386,
+      "loss": 3.6626,
       "step": 17350
     },
     {
-      "epoch": 1.8727801097836616,
-      "grad_norm": 0.5888968706130981,
-      "learning_rate": 0.0004881973925223575,
-      "loss": 3.6395,
+      "epoch": 1.8760107816711589,
+      "grad_norm": 0.6195437908172607,
+      "learning_rate": 0.00048799784133837017,
+      "loss": 3.6596,
       "step": 17400
     },
     {
-      "epoch": 1.87816166182327,
-      "grad_norm": 0.542854368686676,
-      "learning_rate": 0.00048787415149229604,
-      "loss": 3.6483,
+      "epoch": 1.881401617250674,
+      "grad_norm": 0.5613642334938049,
+      "learning_rate": 0.0004876740420939017,
+      "loss": 3.6769,
       "step": 17450
     },
     {
-      "epoch": 1.883543213862878,
-      "grad_norm": 0.5601978898048401,
-      "learning_rate": 0.00048755091046223464,
-      "loss": 3.6658,
+      "epoch": 1.8867924528301887,
+      "grad_norm": 0.562290608882904,
+      "learning_rate": 0.0004873502428494333,
+      "loss": 3.67,
       "step": 17500
     },
     {
-      "epoch": 1.8889247659024861,
-      "grad_norm": 0.5671666264533997,
-      "learning_rate": 0.0004872276694321732,
-      "loss": 3.6378,
+      "epoch": 1.8921832884097034,
+      "grad_norm": 0.5696184635162354,
+      "learning_rate": 0.0004870329195898542,
+      "loss": 3.6503,
       "step": 17550
     },
     {
-      "epoch": 1.8943063179420945,
-      "grad_norm": 0.671059787273407,
-      "learning_rate": 0.00048690442840211177,
-      "loss": 3.633,
+      "epoch": 1.8975741239892183,
+      "grad_norm": 0.5617765784263611,
+      "learning_rate": 0.00048670912034538583,
+      "loss": 3.6591,
       "step": 17600
     },
     {
-      "epoch": 1.8996878699817028,
-      "grad_norm": 0.5720502138137817,
-      "learning_rate": 0.0004865811873720504,
-      "loss": 3.6441,
+      "epoch": 1.9029649595687332,
+      "grad_norm": 0.5480373501777649,
+      "learning_rate": 0.0004863853211009174,
+      "loss": 3.6353,
       "step": 17650
     },
     {
-      "epoch": 1.905069422021311,
-      "grad_norm": 0.6033637523651123,
-      "learning_rate": 0.00048625794634198896,
-      "loss": 3.6437,
+      "epoch": 1.908355795148248,
+      "grad_norm": 0.5804269909858704,
+      "learning_rate": 0.000486061521856449,
+      "loss": 3.6462,
       "step": 17700
     },
     {
-      "epoch": 1.910450974060919,
-      "grad_norm": 0.5671701431274414,
-      "learning_rate": 0.00048593470531192756,
-      "loss": 3.6434,
+      "epoch": 1.9137466307277629,
+      "grad_norm": 0.5778653621673584,
+      "learning_rate": 0.00048573772261198054,
+      "loss": 3.6497,
       "step": 17750
     },
     {
-      "epoch": 1.9158325261005273,
-      "grad_norm": 0.5615334510803223,
-      "learning_rate": 0.00048561146428186615,
-      "loss": 3.6495,
+      "epoch": 1.9191374663072778,
+      "grad_norm": 0.5419909358024597,
+      "learning_rate": 0.00048541392336751214,
+      "loss": 3.669,
       "step": 17800
     },
     {
-      "epoch": 1.9212140781401357,
-      "grad_norm": 0.5892298817634583,
-      "learning_rate": 0.0004852882232518047,
-      "loss": 3.6486,
+      "epoch": 1.9245283018867925,
+      "grad_norm": 0.5353772640228271,
+      "learning_rate": 0.0004850901241230437,
+      "loss": 3.6661,
       "step": 17850
     },
     {
-      "epoch": 1.9265956301797438,
-      "grad_norm": 0.5318509936332703,
-      "learning_rate": 0.00048496498222174334,
-      "loss": 3.6422,
+      "epoch": 1.9299191374663072,
+      "grad_norm": 0.5765721797943115,
+      "learning_rate": 0.0004847663248785753,
+      "loss": 3.6727,
       "step": 17900
     },
     {
-      "epoch": 1.931977182219352,
-      "grad_norm": 0.5301926136016846,
-      "learning_rate": 0.00048464174119168193,
-      "loss": 3.6389,
+      "epoch": 1.935309973045822,
+      "grad_norm": 0.550597071647644,
+      "learning_rate": 0.0004844425256341068,
+      "loss": 3.6618,
       "step": 17950
     },
     {
-      "epoch": 1.9373587342589604,
-      "grad_norm": 0.5343649387359619,
-      "learning_rate": 0.0004843185001616205,
-      "loss": 3.6421,
+      "epoch": 1.940700808625337,
+      "grad_norm": 0.5706955790519714,
+      "learning_rate": 0.00048411872638963834,
+      "loss": 3.65,
       "step": 18000
     },
     {
-      "epoch": 1.9373587342589604,
-      "eval_accuracy": 0.3601151112154662,
-      "eval_loss": 3.593770742416382,
-      "eval_runtime": 183.3778,
-      "eval_samples_per_second": 98.218,
-      "eval_steps_per_second": 6.14,
+      "epoch": 1.940700808625337,
+      "eval_accuracy": 0.3590631339556464,
+      "eval_loss": 3.6006405353546143,
+      "eval_runtime": 153.5359,
+      "eval_samples_per_second": 117.308,
+      "eval_steps_per_second": 7.334,
       "step": 18000
     },
     {
-      "epoch": 1.9427402862985685,
-      "grad_norm": 0.544508159160614,
-      "learning_rate": 0.00048399525913155907,
-      "loss": 3.6457,
+      "epoch": 1.9460916442048517,
+      "grad_norm": 0.6120831966400146,
+      "learning_rate": 0.00048379492714516995,
+      "loss": 3.6376,
       "step": 18050
     },
     {
-      "epoch": 1.9481218383381766,
-      "grad_norm": 0.5659684538841248,
-      "learning_rate": 0.0004836720181014976,
-      "loss": 3.6461,
+      "epoch": 1.9514824797843666,
+      "grad_norm": 0.5756568312644958,
+      "learning_rate": 0.0004834711279007015,
+      "loss": 3.6454,
       "step": 18100
     },
     {
-      "epoch": 1.953503390377785,
-      "grad_norm": 0.5567793846130371,
-      "learning_rate": 0.0004833487770714362,
-      "loss": 3.6475,
+      "epoch": 1.9568733153638815,
+      "grad_norm": 0.5653838515281677,
+      "learning_rate": 0.0004831473286562331,
+      "loss": 3.6521,
       "step": 18150
     },
     {
-      "epoch": 1.9588849424173933,
-      "grad_norm": 0.6041224002838135,
-      "learning_rate": 0.00048302553604137485,
-      "loss": 3.6456,
+      "epoch": 1.9622641509433962,
+      "grad_norm": 0.5330579280853271,
+      "learning_rate": 0.00048282352941176465,
+      "loss": 3.6531,
       "step": 18200
     },
     {
-      "epoch": 1.9642664944570014,
-      "grad_norm": 0.6013017296791077,
-      "learning_rate": 0.0004827022950113134,
-      "loss": 3.6393,
+      "epoch": 1.967654986522911,
+      "grad_norm": 0.5068782567977905,
+      "learning_rate": 0.00048249973016729626,
+      "loss": 3.6443,
       "step": 18250
     },
     {
-      "epoch": 1.9696480464966095,
-      "grad_norm": 0.6019763946533203,
-      "learning_rate": 0.000482379053981252,
-      "loss": 3.6352,
+      "epoch": 1.9730458221024259,
+      "grad_norm": 0.509175181388855,
+      "learning_rate": 0.0004821759309228278,
+      "loss": 3.6688,
       "step": 18300
     },
     {
-      "epoch": 1.9750295985362178,
-      "grad_norm": 0.5311346054077148,
-      "learning_rate": 0.0004820558129511906,
-      "loss": 3.6312,
+      "epoch": 1.9784366576819408,
+      "grad_norm": 0.5572383403778076,
+      "learning_rate": 0.00048185213167835936,
+      "loss": 3.6484,
       "step": 18350
     },
     {
-      "epoch": 1.9804111505758262,
-      "grad_norm": 0.5672343373298645,
-      "learning_rate": 0.0004817325719211291,
-      "loss": 3.6395,
+      "epoch": 1.9838274932614555,
+      "grad_norm": 0.6003437638282776,
+      "learning_rate": 0.00048152833243389096,
+      "loss": 3.6516,
       "step": 18400
     },
     {
-      "epoch": 1.9857927026154343,
-      "grad_norm": 0.570595383644104,
-      "learning_rate": 0.0004814093308910677,
-      "loss": 3.6178,
+      "epoch": 1.9892183288409704,
+      "grad_norm": 0.5463119149208069,
+      "learning_rate": 0.0004812045331894225,
+      "loss": 3.66,
       "step": 18450
     },
     {
-      "epoch": 1.9911742546550424,
-      "grad_norm": 0.5667010545730591,
-      "learning_rate": 0.00048108608986100637,
-      "loss": 3.6229,
+      "epoch": 1.9946091644204853,
+      "grad_norm": 0.6415205597877502,
+      "learning_rate": 0.0004808807339449541,
+      "loss": 3.6402,
       "step": 18500
     },
     {
-      "epoch": 1.9965558066946507,
-      "grad_norm": 0.581763744354248,
-      "learning_rate": 0.0004807628488309449,
-      "loss": 3.632,
+      "epoch": 2.0,
+      "grad_norm": 1.1502970457077026,
+      "learning_rate": 0.0004805569347004856,
+      "loss": 3.6421,
       "step": 18550
     },
     {
-      "epoch": 2.001937358734259,
-      "grad_norm": 0.5959236025810242,
-      "learning_rate": 0.0004804396078008835,
-      "loss": 3.6083,
+      "epoch": 2.0053908355795147,
+      "grad_norm": 0.5730555653572083,
+      "learning_rate": 0.0004802331354560173,
+      "loss": 3.5402,
       "step": 18600
     },
     {
-      "epoch": 2.007318910773867,
-      "grad_norm": 0.5947738885879517,
-      "learning_rate": 0.0004801228315914233,
-      "loss": 3.5659,
+      "epoch": 2.01078167115903,
+      "grad_norm": 0.5260567665100098,
+      "learning_rate": 0.00047990933621154877,
+      "loss": 3.5514,
       "step": 18650
     },
     {
-      "epoch": 2.0127004628134753,
-      "grad_norm": 0.6648337841033936,
-      "learning_rate": 0.0004797995905613619,
-      "loss": 3.546,
+      "epoch": 2.0161725067385445,
+      "grad_norm": 0.5512271523475647,
+      "learning_rate": 0.0004795855369670804,
+      "loss": 3.5484,
       "step": 18700
     },
     {
-      "epoch": 2.018082014853084,
-      "grad_norm": 0.5789828300476074,
-      "learning_rate": 0.00047947634953130044,
-      "loss": 3.5452,
+      "epoch": 2.0215633423180592,
+      "grad_norm": 0.5696899890899658,
+      "learning_rate": 0.0004792617377226119,
+      "loss": 3.5563,
       "step": 18750
     },
     {
-      "epoch": 2.023463566892692,
-      "grad_norm": 0.5471194386482239,
-      "learning_rate": 0.00047915310850123904,
-      "loss": 3.5378,
+      "epoch": 2.026954177897574,
+      "grad_norm": 0.557259738445282,
+      "learning_rate": 0.0004789379384781435,
+      "loss": 3.5788,
       "step": 18800
     },
     {
-      "epoch": 2.0288451189323,
-      "grad_norm": 0.5744550824165344,
-      "learning_rate": 0.0004788298674711777,
-      "loss": 3.5403,
+      "epoch": 2.032345013477089,
+      "grad_norm": 0.5228050351142883,
+      "learning_rate": 0.0004786141392336751,
+      "loss": 3.5763,
       "step": 18850
     },
     {
-      "epoch": 2.034226670971908,
-      "grad_norm": 0.609020471572876,
-      "learning_rate": 0.00047850662644111623,
-      "loss": 3.5548,
+      "epoch": 2.0377358490566038,
+      "grad_norm": 0.5340172648429871,
+      "learning_rate": 0.00047829033998920663,
+      "loss": 3.5583,
       "step": 18900
     },
     {
-      "epoch": 2.0396082230115167,
-      "grad_norm": 0.5844295620918274,
-      "learning_rate": 0.0004781833854110548,
-      "loss": 3.5391,
+      "epoch": 2.0431266846361185,
+      "grad_norm": 0.5462618470191956,
+      "learning_rate": 0.00047796654074473824,
+      "loss": 3.5624,
       "step": 18950
     },
     {
-      "epoch": 2.044989775051125,
-      "grad_norm": 0.568282961845398,
-      "learning_rate": 0.00047786014438099336,
-      "loss": 3.5593,
+      "epoch": 2.0485175202156336,
+      "grad_norm": 0.5697525143623352,
+      "learning_rate": 0.0004776427415002698,
+      "loss": 3.5935,
       "step": 19000
     },
     {
-      "epoch": 2.044989775051125,
-      "eval_accuracy": 0.36145284557023805,
-      "eval_loss": 3.5825202465057373,
-      "eval_runtime": 183.3433,
-      "eval_samples_per_second": 98.236,
-      "eval_steps_per_second": 6.141,
+      "epoch": 2.0485175202156336,
+      "eval_accuracy": 0.3605037625951786,
+      "eval_loss": 3.590785264968872,
+      "eval_runtime": 152.7274,
+      "eval_samples_per_second": 117.929,
+      "eval_steps_per_second": 7.373,
       "step": 19000
     },
     {
-      "epoch": 2.050371327090733,
-      "grad_norm": 0.6129640936851501,
-      "learning_rate": 0.00047753690335093196,
-      "loss": 3.5447,
+      "epoch": 2.0539083557951483,
+      "grad_norm": 0.5625264644622803,
+      "learning_rate": 0.0004773189422558014,
+      "loss": 3.5717,
       "step": 19050
     },
     {
-      "epoch": 2.055752879130341,
-      "grad_norm": 0.5720137357711792,
-      "learning_rate": 0.00047721366232087055,
-      "loss": 3.5577,
+      "epoch": 2.059299191374663,
+      "grad_norm": 0.6776948571205139,
+      "learning_rate": 0.00047699514301133294,
+      "loss": 3.5631,
       "step": 19100
     },
     {
-      "epoch": 2.0611344311699495,
-      "grad_norm": 0.6474032402038574,
-      "learning_rate": 0.00047689042129080915,
-      "loss": 3.5638,
+      "epoch": 2.0646900269541777,
+      "grad_norm": 0.551921010017395,
+      "learning_rate": 0.00047667134376686455,
+      "loss": 3.5767,
       "step": 19150
     },
     {
-      "epoch": 2.0665159832095576,
-      "grad_norm": 0.5881885886192322,
-      "learning_rate": 0.00047656718026074774,
-      "loss": 3.5615,
+      "epoch": 2.070080862533693,
+      "grad_norm": 0.6170827150344849,
+      "learning_rate": 0.0004763475445223961,
+      "loss": 3.59,
       "step": 19200
     },
     {
-      "epoch": 2.0718975352491658,
-      "grad_norm": 0.7327039241790771,
-      "learning_rate": 0.00047624393923068634,
-      "loss": 3.5571,
+      "epoch": 2.0754716981132075,
+      "grad_norm": 0.5683097839355469,
+      "learning_rate": 0.0004760237452779276,
+      "loss": 3.5723,
       "step": 19250
     },
     {
-      "epoch": 2.0772790872887743,
-      "grad_norm": 0.5858220458030701,
-      "learning_rate": 0.0004759206982006249,
-      "loss": 3.5505,
+      "epoch": 2.0808625336927222,
+      "grad_norm": 0.6205763220787048,
+      "learning_rate": 0.0004756999460334592,
+      "loss": 3.5611,
       "step": 19300
     },
     {
-      "epoch": 2.0826606393283824,
-      "grad_norm": 0.6055868864059448,
-      "learning_rate": 0.00047559745717056347,
-      "loss": 3.5574,
+      "epoch": 2.0862533692722374,
+      "grad_norm": 0.5944758653640747,
+      "learning_rate": 0.00047537614678899075,
+      "loss": 3.5774,
       "step": 19350
     },
     {
-      "epoch": 2.0880421913679905,
-      "grad_norm": 0.5690827369689941,
-      "learning_rate": 0.000475274216140502,
-      "loss": 3.5673,
+      "epoch": 2.091644204851752,
+      "grad_norm": 0.5868034362792969,
+      "learning_rate": 0.00047505234754452235,
+      "loss": 3.5547,
       "step": 19400
     },
     {
-      "epoch": 2.0934237434075986,
-      "grad_norm": 0.5776065587997437,
-      "learning_rate": 0.00047495097511044066,
-      "loss": 3.5479,
+      "epoch": 2.0970350404312668,
+      "grad_norm": 0.5427871346473694,
+      "learning_rate": 0.0004747285483000539,
+      "loss": 3.5713,
       "step": 19450
     },
     {
-      "epoch": 2.098805295447207,
-      "grad_norm": 0.6034450531005859,
-      "learning_rate": 0.00047462773408037925,
-      "loss": 3.5725,
+      "epoch": 2.1024258760107815,
+      "grad_norm": 0.5842890739440918,
+      "learning_rate": 0.0004744047490555855,
+      "loss": 3.5719,
       "step": 19500
     },
     {
-      "epoch": 2.1041868474868153,
-      "grad_norm": 0.5865857005119324,
-      "learning_rate": 0.0004743044930503178,
-      "loss": 3.5505,
+      "epoch": 2.1078167115902966,
+      "grad_norm": 0.6317399144172668,
+      "learning_rate": 0.00047408094981111706,
+      "loss": 3.5627,
       "step": 19550
     },
     {
-      "epoch": 2.1095683995264234,
-      "grad_norm": 0.639350950717926,
-      "learning_rate": 0.0004739812520202564,
-      "loss": 3.5456,
+      "epoch": 2.1132075471698113,
+      "grad_norm": 0.5856570601463318,
+      "learning_rate": 0.00047375715056664866,
+      "loss": 3.5652,
       "step": 19600
     },
     {
-      "epoch": 2.1149499515660315,
-      "grad_norm": 0.586966872215271,
-      "learning_rate": 0.000473658010990195,
-      "loss": 3.551,
+      "epoch": 2.118598382749326,
+      "grad_norm": 0.6095489263534546,
+      "learning_rate": 0.0004734333513221802,
+      "loss": 3.5702,
       "step": 19650
     },
     {
-      "epoch": 2.12033150360564,
-      "grad_norm": 0.5592098236083984,
-      "learning_rate": 0.0004733347699601336,
-      "loss": 3.5438,
+      "epoch": 2.123989218328841,
+      "grad_norm": 0.5798874497413635,
+      "learning_rate": 0.00047310955207771177,
+      "loss": 3.5752,
       "step": 19700
     },
     {
-      "epoch": 2.125713055645248,
-      "grad_norm": 0.6772919297218323,
-      "learning_rate": 0.0004730115289300722,
-      "loss": 3.5532,
+      "epoch": 2.129380053908356,
+      "grad_norm": 0.524194061756134,
+      "learning_rate": 0.00047278575283324337,
+      "loss": 3.5807,
       "step": 19750
     },
     {
-      "epoch": 2.1310946076848563,
-      "grad_norm": 0.5720005631446838,
-      "learning_rate": 0.00047268828790001077,
-      "loss": 3.5744,
+      "epoch": 2.1347708894878705,
+      "grad_norm": 0.5766165852546692,
+      "learning_rate": 0.0004724619535887749,
+      "loss": 3.5624,
       "step": 19800
     },
     {
-      "epoch": 2.1364761597244644,
-      "grad_norm": 0.6096287965774536,
-      "learning_rate": 0.0004723650468699493,
-      "loss": 3.545,
+      "epoch": 2.1401617250673857,
+      "grad_norm": 0.5298904776573181,
+      "learning_rate": 0.0004721381543443065,
+      "loss": 3.5894,
       "step": 19850
     },
     {
-      "epoch": 2.141857711764073,
-      "grad_norm": 0.5602501630783081,
-      "learning_rate": 0.0004720418058398879,
-      "loss": 3.5619,
+      "epoch": 2.1455525606469004,
+      "grad_norm": 0.6056088209152222,
+      "learning_rate": 0.000471814355099838,
+      "loss": 3.5903,
       "step": 19900
     },
     {
-      "epoch": 2.147239263803681,
-      "grad_norm": 0.5784476399421692,
-      "learning_rate": 0.00047171856480982644,
-      "loss": 3.57,
+      "epoch": 2.150943396226415,
+      "grad_norm": 0.5861086845397949,
+      "learning_rate": 0.0004714905558553697,
+      "loss": 3.5874,
       "step": 19950
     },
     {
-      "epoch": 2.152620815843289,
-      "grad_norm": 0.5974809527397156,
-      "learning_rate": 0.0004713953237797651,
-      "loss": 3.5573,
+      "epoch": 2.1563342318059298,
+      "grad_norm": 0.5253286957740784,
+      "learning_rate": 0.0004711667566109012,
+      "loss": 3.5703,
       "step": 20000
     },
     {
-      "epoch": 2.152620815843289,
-      "eval_accuracy": 0.3621622402747788,
-      "eval_loss": 3.5724599361419678,
-      "eval_runtime": 183.7401,
-      "eval_samples_per_second": 98.024,
-      "eval_steps_per_second": 6.128,
+      "epoch": 2.1563342318059298,
+      "eval_accuracy": 0.36150814989013563,
+      "eval_loss": 3.583982467651367,
+      "eval_runtime": 153.1948,
+      "eval_samples_per_second": 117.569,
+      "eval_steps_per_second": 7.35,
       "step": 20000
     },
     {
-      "epoch": 2.1580023678828972,
-      "grad_norm": 0.625355064868927,
-      "learning_rate": 0.0004710720827497037,
-      "loss": 3.5736,
+      "epoch": 2.161725067385445,
+      "grad_norm": 0.5482057332992554,
+      "learning_rate": 0.00047084295736643273,
+      "loss": 3.571,
       "step": 20050
     },
     {
-      "epoch": 2.163383919922506,
-      "grad_norm": 0.5633546113967896,
-      "learning_rate": 0.00047074884171964223,
-      "loss": 3.5605,
+      "epoch": 2.1671159029649596,
+      "grad_norm": 0.6015800833702087,
+      "learning_rate": 0.00047051915812196433,
+      "loss": 3.577,
       "step": 20100
     },
     {
-      "epoch": 2.168765471962114,
-      "grad_norm": 0.5668460726737976,
-      "learning_rate": 0.0004704256006895808,
-      "loss": 3.575,
+      "epoch": 2.1725067385444743,
+      "grad_norm": 0.5562338829040527,
+      "learning_rate": 0.0004701953588774959,
+      "loss": 3.57,
       "step": 20150
     },
     {
-      "epoch": 2.174147024001722,
-      "grad_norm": 0.6517674326896667,
-      "learning_rate": 0.00047010235965951936,
-      "loss": 3.5651,
+      "epoch": 2.177897574123989,
+      "grad_norm": 0.5639262795448303,
+      "learning_rate": 0.0004698715596330275,
+      "loss": 3.5881,
       "step": 20200
     },
     {
-      "epoch": 2.1795285760413305,
-      "grad_norm": 0.6065477132797241,
-      "learning_rate": 0.00046977911862945796,
-      "loss": 3.5512,
+      "epoch": 2.183288409703504,
+      "grad_norm": 0.5844449996948242,
+      "learning_rate": 0.00046954776038855904,
+      "loss": 3.5739,
       "step": 20250
     },
     {
-      "epoch": 2.1849101280809387,
-      "grad_norm": 0.5984815955162048,
-      "learning_rate": 0.0004694558775993966,
-      "loss": 3.5643,
+      "epoch": 2.188679245283019,
+      "grad_norm": 0.595220685005188,
+      "learning_rate": 0.00046922396114409064,
+      "loss": 3.5793,
       "step": 20300
     },
     {
-      "epoch": 2.1902916801205468,
-      "grad_norm": 0.6116432547569275,
-      "learning_rate": 0.00046913263656933515,
-      "loss": 3.5655,
+      "epoch": 2.1940700808625335,
+      "grad_norm": 0.6522249579429626,
+      "learning_rate": 0.0004689066378845116,
+      "loss": 3.5624,
       "step": 20350
     },
     {
-      "epoch": 2.195673232160155,
-      "grad_norm": 0.6076990962028503,
-      "learning_rate": 0.00046880939553927374,
-      "loss": 3.5665,
+      "epoch": 2.1994609164420487,
+      "grad_norm": 0.5735236406326294,
+      "learning_rate": 0.00046858283864004315,
+      "loss": 3.5681,
       "step": 20400
     },
     {
-      "epoch": 2.2010547841997634,
-      "grad_norm": 0.5770449638366699,
-      "learning_rate": 0.00046848615450921234,
-      "loss": 3.571,
+      "epoch": 2.2048517520215634,
+      "grad_norm": 0.577090859413147,
+      "learning_rate": 0.0004682590393955747,
+      "loss": 3.5699,
       "step": 20450
     },
     {
-      "epoch": 2.2064363362393715,
-      "grad_norm": 0.5903902649879456,
-      "learning_rate": 0.0004681629134791509,
-      "loss": 3.5448,
+      "epoch": 2.210242587601078,
+      "grad_norm": 0.5954312086105347,
+      "learning_rate": 0.0004679352401511063,
+      "loss": 3.5807,
       "step": 20500
     },
     {
-      "epoch": 2.2118178882789796,
-      "grad_norm": 0.5791151523590088,
-      "learning_rate": 0.00046783967244908947,
-      "loss": 3.5764,
+      "epoch": 2.215633423180593,
+      "grad_norm": 0.5523508191108704,
+      "learning_rate": 0.0004676114409066378,
+      "loss": 3.5774,
       "step": 20550
     },
     {
-      "epoch": 2.2171994403185877,
-      "grad_norm": 0.5685939788818359,
-      "learning_rate": 0.0004675164314190281,
-      "loss": 3.57,
+      "epoch": 2.221024258760108,
+      "grad_norm": 0.5406270027160645,
+      "learning_rate": 0.00046728764166216946,
+      "loss": 3.5725,
       "step": 20600
     },
     {
-      "epoch": 2.2225809923581963,
-      "grad_norm": 0.6150746941566467,
-      "learning_rate": 0.00046719319038896666,
-      "loss": 3.545,
+      "epoch": 2.2264150943396226,
+      "grad_norm": 0.5979112982749939,
+      "learning_rate": 0.00046696384241770095,
+      "loss": 3.5733,
       "step": 20650
     },
     {
-      "epoch": 2.2279625443978044,
-      "grad_norm": 0.5683572888374329,
-      "learning_rate": 0.0004668764141795065,
-      "loss": 3.5403,
+      "epoch": 2.2318059299191373,
+      "grad_norm": 0.5970372557640076,
+      "learning_rate": 0.00046664004317323256,
+      "loss": 3.567,
       "step": 20700
     },
     {
-      "epoch": 2.2333440964374125,
-      "grad_norm": 0.5446680784225464,
-      "learning_rate": 0.00046655317314944506,
-      "loss": 3.5548,
+      "epoch": 2.2371967654986524,
+      "grad_norm": 0.5633722543716431,
+      "learning_rate": 0.0004663162439287641,
+      "loss": 3.5687,
       "step": 20750
     },
     {
-      "epoch": 2.2387256484770206,
-      "grad_norm": 0.5595022439956665,
-      "learning_rate": 0.00046622993211938366,
-      "loss": 3.555,
+      "epoch": 2.242587601078167,
+      "grad_norm": 0.6036709547042847,
+      "learning_rate": 0.00046599244468429566,
+      "loss": 3.5761,
       "step": 20800
     },
     {
-      "epoch": 2.244107200516629,
-      "grad_norm": 0.6505151391029358,
-      "learning_rate": 0.0004659066910893222,
-      "loss": 3.5721,
+      "epoch": 2.247978436657682,
+      "grad_norm": 0.5964614152908325,
+      "learning_rate": 0.00046566864543982726,
+      "loss": 3.5715,
       "step": 20850
     },
     {
-      "epoch": 2.2494887525562373,
-      "grad_norm": 0.6140245199203491,
-      "learning_rate": 0.0004655834500592608,
-      "loss": 3.5549,
+      "epoch": 2.2533692722371965,
+      "grad_norm": 0.563269853591919,
+      "learning_rate": 0.0004653448461953588,
+      "loss": 3.5657,
       "step": 20900
     },
     {
-      "epoch": 2.2548703045958454,
-      "grad_norm": 0.5573344230651855,
-      "learning_rate": 0.00046526020902919944,
-      "loss": 3.5707,
+      "epoch": 2.2587601078167117,
+      "grad_norm": 0.5765753984451294,
+      "learning_rate": 0.0004650210469508904,
+      "loss": 3.5741,
       "step": 20950
     },
     {
-      "epoch": 2.2602518566354535,
-      "grad_norm": 0.5895587205886841,
-      "learning_rate": 0.000464936967999138,
-      "loss": 3.543,
+      "epoch": 2.2641509433962264,
+      "grad_norm": 0.6818150877952576,
+      "learning_rate": 0.00046469724770642197,
+      "loss": 3.5583,
       "step": 21000
     },
     {
-      "epoch": 2.2602518566354535,
-      "eval_accuracy": 0.3634812863328466,
-      "eval_loss": 3.5625159740448,
-      "eval_runtime": 183.0115,
-      "eval_samples_per_second": 98.415,
-      "eval_steps_per_second": 6.153,
+      "epoch": 2.2641509433962264,
+      "eval_accuracy": 0.36253926579549506,
+      "eval_loss": 3.5719754695892334,
+      "eval_runtime": 153.1185,
+      "eval_samples_per_second": 117.628,
+      "eval_steps_per_second": 7.354,
       "step": 21000
     },
     {
-      "epoch": 2.265633408675062,
-      "grad_norm": 0.5756750702857971,
-      "learning_rate": 0.0004646137269690766,
-      "loss": 3.5756,
+      "epoch": 2.269541778975741,
+      "grad_norm": 0.5509704351425171,
+      "learning_rate": 0.0004643734484619536,
+      "loss": 3.5731,
       "step": 21050
     },
     {
-      "epoch": 2.27101496071467,
-      "grad_norm": 0.6374192833900452,
-      "learning_rate": 0.00046429048593901517,
-      "loss": 3.5658,
+      "epoch": 2.274932614555256,
+      "grad_norm": 0.5755780339241028,
+      "learning_rate": 0.0004640496492174851,
+      "loss": 3.5767,
       "step": 21100
     },
     {
-      "epoch": 2.2763965127542782,
-      "grad_norm": 0.6448386907577515,
-      "learning_rate": 0.0004639672449089537,
-      "loss": 3.5562,
+      "epoch": 2.280323450134771,
+      "grad_norm": 0.5620628595352173,
+      "learning_rate": 0.00046372584997301673,
+      "loss": 3.5879,
       "step": 21150
     },
     {
-      "epoch": 2.281778064793887,
-      "grad_norm": 0.5764261484146118,
-      "learning_rate": 0.0004636440038788923,
-      "loss": 3.5522,
+      "epoch": 2.2857142857142856,
+      "grad_norm": 0.5774373412132263,
+      "learning_rate": 0.0004634020507285483,
+      "loss": 3.5399,
       "step": 21200
     },
     {
-      "epoch": 2.287159616833495,
-      "grad_norm": 0.571958601474762,
-      "learning_rate": 0.00046332076284883095,
-      "loss": 3.5489,
+      "epoch": 2.2911051212938007,
+      "grad_norm": 0.576257586479187,
+      "learning_rate": 0.00046308472746896923,
+      "loss": 3.5681,
       "step": 21250
     },
     {
-      "epoch": 2.292541168873103,
-      "grad_norm": 0.606433629989624,
-      "learning_rate": 0.0004629975218187695,
-      "loss": 3.5598,
+      "epoch": 2.2964959568733154,
+      "grad_norm": 0.527144730091095,
+      "learning_rate": 0.00046276092822450073,
+      "loss": 3.5693,
       "step": 21300
     },
     {
-      "epoch": 2.297922720912711,
-      "grad_norm": 0.5860156416893005,
-      "learning_rate": 0.0004626742807887081,
-      "loss": 3.5557,
+      "epoch": 2.30188679245283,
+      "grad_norm": 0.5843461751937866,
+      "learning_rate": 0.00046243712898003233,
+      "loss": 3.5636,
       "step": 21350
     },
     {
-      "epoch": 2.303304272952319,
-      "grad_norm": 0.646503210067749,
-      "learning_rate": 0.00046235103975864663,
-      "loss": 3.5757,
+      "epoch": 2.3072776280323453,
+      "grad_norm": 0.5927402377128601,
+      "learning_rate": 0.0004621133297355639,
+      "loss": 3.5765,
       "step": 21400
     },
     {
-      "epoch": 2.3086858249919278,
-      "grad_norm": 0.5846951007843018,
-      "learning_rate": 0.0004620277987285852,
-      "loss": 3.5529,
+      "epoch": 2.31266846361186,
+      "grad_norm": 0.5736572742462158,
+      "learning_rate": 0.0004617895304910955,
+      "loss": 3.5753,
       "step": 21450
     },
     {
-      "epoch": 2.314067377031536,
-      "grad_norm": 0.5809066891670227,
-      "learning_rate": 0.0004617045576985239,
-      "loss": 3.5635,
+      "epoch": 2.3180592991913747,
+      "grad_norm": 0.5711485743522644,
+      "learning_rate": 0.00046146573124662704,
+      "loss": 3.5782,
       "step": 21500
     },
     {
-      "epoch": 2.319448929071144,
-      "grad_norm": 0.5956180691719055,
-      "learning_rate": 0.0004613813166684624,
-      "loss": 3.5751,
+      "epoch": 2.3234501347708894,
+      "grad_norm": 0.5785765647888184,
+      "learning_rate": 0.00046114193200215864,
+      "loss": 3.5589,
       "step": 21550
     },
     {
-      "epoch": 2.3248304811107525,
-      "grad_norm": 0.5681479573249817,
-      "learning_rate": 0.000461058075638401,
-      "loss": 3.5771,
+      "epoch": 2.3288409703504045,
+      "grad_norm": 0.5806024670600891,
+      "learning_rate": 0.0004608181327576902,
+      "loss": 3.5782,
       "step": 21600
     },
     {
-      "epoch": 2.3302120331503606,
-      "grad_norm": 0.5650503635406494,
-      "learning_rate": 0.00046073483460833955,
-      "loss": 3.5631,
+      "epoch": 2.334231805929919,
+      "grad_norm": 0.5791244506835938,
+      "learning_rate": 0.00046049433351322175,
+      "loss": 3.5756,
       "step": 21650
     },
     {
-      "epoch": 2.3355935851899687,
-      "grad_norm": 0.5988578796386719,
-      "learning_rate": 0.00046041159357827814,
-      "loss": 3.5487,
+      "epoch": 2.339622641509434,
+      "grad_norm": 0.55106520652771,
+      "learning_rate": 0.00046017053426875335,
+      "loss": 3.5554,
       "step": 21700
     },
     {
-      "epoch": 2.340975137229577,
-      "grad_norm": 0.5927512645721436,
-      "learning_rate": 0.00046008835254821674,
-      "loss": 3.5482,
+      "epoch": 2.3450134770889486,
+      "grad_norm": 0.5803852081298828,
+      "learning_rate": 0.0004598467350242849,
+      "loss": 3.5713,
       "step": 21750
     },
     {
-      "epoch": 2.3463566892691854,
-      "grad_norm": 0.5896438956260681,
-      "learning_rate": 0.0004597651115181554,
-      "loss": 3.5726,
+      "epoch": 2.3504043126684637,
+      "grad_norm": 0.5649197697639465,
+      "learning_rate": 0.0004595229357798165,
+      "loss": 3.5931,
       "step": 21800
     },
     {
-      "epoch": 2.3517382413087935,
-      "grad_norm": 0.6439616084098816,
-      "learning_rate": 0.00045944187048809393,
-      "loss": 3.5549,
+      "epoch": 2.3557951482479784,
+      "grad_norm": 0.5821561217308044,
+      "learning_rate": 0.00045919913653534806,
+      "loss": 3.5783,
       "step": 21850
     },
     {
-      "epoch": 2.3571197933484016,
-      "grad_norm": 0.6187168955802917,
-      "learning_rate": 0.0004591186294580325,
-      "loss": 3.5549,
+      "epoch": 2.361185983827493,
+      "grad_norm": 0.6483994126319885,
+      "learning_rate": 0.00045887533729087966,
+      "loss": 3.5553,
       "step": 21900
     },
     {
-      "epoch": 2.3625013453880097,
-      "grad_norm": 0.5949316024780273,
-      "learning_rate": 0.00045879538842797106,
-      "loss": 3.5587,
+      "epoch": 2.3665768194070083,
+      "grad_norm": 0.6019250154495239,
+      "learning_rate": 0.00045855153804641116,
+      "loss": 3.5687,
       "step": 21950
     },
     {
-      "epoch": 2.3678828974276183,
-      "grad_norm": 0.5802180171012878,
-      "learning_rate": 0.00045847214739790966,
-      "loss": 3.5358,
+      "epoch": 2.371967654986523,
+      "grad_norm": 0.5473122596740723,
+      "learning_rate": 0.0004582277388019427,
+      "loss": 3.5614,
       "step": 22000
     },
     {
-      "epoch": 2.3678828974276183,
-      "eval_accuracy": 0.3646419164804809,
-      "eval_loss": 3.551939010620117,
-      "eval_runtime": 183.8949,
-      "eval_samples_per_second": 97.942,
-      "eval_steps_per_second": 6.123,
+      "epoch": 2.371967654986523,
+      "eval_accuracy": 0.3640629052413175,
+      "eval_loss": 3.5617072582244873,
+      "eval_runtime": 152.9872,
+      "eval_samples_per_second": 117.729,
+      "eval_steps_per_second": 7.36,
       "step": 22000
     },
     {
-      "epoch": 2.3732644494672264,
-      "grad_norm": 0.6083611249923706,
-      "learning_rate": 0.0004581489063678482,
-      "loss": 3.5409,
+      "epoch": 2.3773584905660377,
+      "grad_norm": 0.5987036824226379,
+      "learning_rate": 0.0004579039395574743,
+      "loss": 3.5639,
       "step": 22050
     },
     {
-      "epoch": 2.3786460015068345,
-      "grad_norm": 0.6172609329223633,
-      "learning_rate": 0.00045782566533778685,
-      "loss": 3.5533,
+      "epoch": 2.382749326145553,
+      "grad_norm": 0.5624399185180664,
+      "learning_rate": 0.00045758014031300586,
+      "loss": 3.5773,
       "step": 22100
     },
     {
-      "epoch": 2.384027553546443,
-      "grad_norm": 0.6144202351570129,
-      "learning_rate": 0.00045750242430772544,
-      "loss": 3.5594,
+      "epoch": 2.3881401617250675,
+      "grad_norm": 0.5358259677886963,
+      "learning_rate": 0.00045725634106853747,
+      "loss": 3.5788,
       "step": 22150
     },
     {
-      "epoch": 2.389409105586051,
-      "grad_norm": 0.5577259063720703,
-      "learning_rate": 0.000457179183277664,
-      "loss": 3.5359,
+      "epoch": 2.393530997304582,
+      "grad_norm": 0.5895576477050781,
+      "learning_rate": 0.000456932541824069,
+      "loss": 3.5632,
       "step": 22200
     },
     {
-      "epoch": 2.3947906576256592,
-      "grad_norm": 0.6218588352203369,
-      "learning_rate": 0.0004568559422476026,
-      "loss": 3.5437,
+      "epoch": 2.398921832884097,
+      "grad_norm": 0.5324219465255737,
+      "learning_rate": 0.0004566087425796006,
+      "loss": 3.5707,
       "step": 22250
     },
     {
-      "epoch": 2.4001722096652673,
-      "grad_norm": 0.5700936317443848,
-      "learning_rate": 0.00045653270121754117,
-      "loss": 3.5465,
+      "epoch": 2.404312668463612,
+      "grad_norm": 0.5704677700996399,
+      "learning_rate": 0.0004562849433351322,
+      "loss": 3.5595,
       "step": 22300
     },
     {
-      "epoch": 2.4055537617048754,
-      "grad_norm": 0.5605334639549255,
-      "learning_rate": 0.0004562094601874797,
-      "loss": 3.5429,
+      "epoch": 2.4097035040431267,
+      "grad_norm": 0.560779869556427,
+      "learning_rate": 0.0004559611440906638,
+      "loss": 3.5823,
       "step": 22350
     },
     {
-      "epoch": 2.410935313744484,
-      "grad_norm": 0.5979263186454773,
-      "learning_rate": 0.00045588621915741836,
-      "loss": 3.5515,
+      "epoch": 2.4150943396226414,
+      "grad_norm": 0.5572506189346313,
+      "learning_rate": 0.00045563734484619533,
+      "loss": 3.5646,
       "step": 22400
     },
     {
-      "epoch": 2.416316865784092,
-      "grad_norm": 0.6374608874320984,
-      "learning_rate": 0.00045556297812735696,
-      "loss": 3.5474,
+      "epoch": 2.420485175202156,
+      "grad_norm": 0.5772961378097534,
+      "learning_rate": 0.0004553135456017269,
+      "loss": 3.5674,
       "step": 22450
     },
     {
-      "epoch": 2.4216984178237,
-      "grad_norm": 0.6078911423683167,
-      "learning_rate": 0.0004552397370972955,
-      "loss": 3.5454,
+      "epoch": 2.4258760107816713,
+      "grad_norm": 0.598270833492279,
+      "learning_rate": 0.0004549897463572585,
+      "loss": 3.5746,
       "step": 22500
     },
     {
-      "epoch": 2.4270799698633088,
-      "grad_norm": 0.6158359050750732,
-      "learning_rate": 0.0004549164960672341,
-      "loss": 3.5487,
+      "epoch": 2.431266846361186,
+      "grad_norm": 0.5786691308021545,
+      "learning_rate": 0.00045466594711279,
+      "loss": 3.5661,
       "step": 22550
     },
     {
-      "epoch": 2.432461521902917,
-      "grad_norm": 0.6169496178627014,
-      "learning_rate": 0.00045459325503717263,
-      "loss": 3.5437,
+      "epoch": 2.4366576819407006,
+      "grad_norm": 0.6232877969741821,
+      "learning_rate": 0.00045434214786832164,
+      "loss": 3.5967,
       "step": 22600
     },
     {
-      "epoch": 2.437843073942525,
-      "grad_norm": 0.5959857702255249,
-      "learning_rate": 0.0004542700140071113,
-      "loss": 3.5595,
+      "epoch": 2.442048517520216,
+      "grad_norm": 0.5886116623878479,
+      "learning_rate": 0.00045401834862385314,
+      "loss": 3.5578,
       "step": 22650
     },
     {
-      "epoch": 2.443224625982133,
-      "grad_norm": 0.6245175004005432,
-      "learning_rate": 0.00045395323779765103,
-      "loss": 3.5371,
+      "epoch": 2.4474393530997305,
+      "grad_norm": 0.5678022503852844,
+      "learning_rate": 0.00045369454937938474,
+      "loss": 3.5477,
       "step": 22700
     },
     {
-      "epoch": 2.4486061780217416,
-      "grad_norm": 0.6321040987968445,
-      "learning_rate": 0.0004536299967675897,
-      "loss": 3.5392,
+      "epoch": 2.452830188679245,
+      "grad_norm": 0.6078322529792786,
+      "learning_rate": 0.0004533707501349163,
+      "loss": 3.552,
       "step": 22750
     },
     {
-      "epoch": 2.4539877300613497,
-      "grad_norm": 0.6347010135650635,
-      "learning_rate": 0.0004533067557375283,
-      "loss": 3.5722,
+      "epoch": 2.4582210242587603,
+      "grad_norm": 0.5677856206893921,
+      "learning_rate": 0.0004530469508904479,
+      "loss": 3.5631,
       "step": 22800
     },
     {
-      "epoch": 2.459369282100958,
-      "grad_norm": 0.5940034985542297,
-      "learning_rate": 0.0004529835147074668,
-      "loss": 3.5691,
+      "epoch": 2.463611859838275,
+      "grad_norm": 0.5610161423683167,
+      "learning_rate": 0.00045272315164597945,
+      "loss": 3.5505,
       "step": 22850
     },
     {
-      "epoch": 2.464750834140566,
-      "grad_norm": 0.5675280690193176,
-      "learning_rate": 0.0004526602736774054,
-      "loss": 3.5472,
+      "epoch": 2.4690026954177897,
+      "grad_norm": 0.6065378785133362,
+      "learning_rate": 0.000452399352401511,
+      "loss": 3.5652,
       "step": 22900
     },
     {
-      "epoch": 2.4701323861801745,
-      "grad_norm": 0.7855820059776306,
-      "learning_rate": 0.00045233703264734395,
-      "loss": 3.5494,
+      "epoch": 2.4743935309973044,
+      "grad_norm": 0.6066365242004395,
+      "learning_rate": 0.0004520755531570426,
+      "loss": 3.5621,
       "step": 22950
     },
     {
-      "epoch": 2.4755139382197826,
-      "grad_norm": 0.5842916369438171,
-      "learning_rate": 0.00045201379161728255,
-      "loss": 3.5404,
+      "epoch": 2.4797843665768196,
+      "grad_norm": 0.5956305265426636,
+      "learning_rate": 0.00045175175391257415,
+      "loss": 3.5545,
       "step": 23000
     },
     {
-      "epoch": 2.4755139382197826,
-      "eval_accuracy": 0.36584807218810944,
-      "eval_loss": 3.5398452281951904,
-      "eval_runtime": 183.3001,
-      "eval_samples_per_second": 98.26,
-      "eval_steps_per_second": 6.143,
+      "epoch": 2.4797843665768196,
+      "eval_accuracy": 0.36502068044740216,
+      "eval_loss": 3.548067092895508,
+      "eval_runtime": 153.1534,
+      "eval_samples_per_second": 117.601,
+      "eval_steps_per_second": 7.352,
       "step": 23000
     },
     {
-      "epoch": 2.4808954902593907,
-      "grad_norm": 0.6048821210861206,
-      "learning_rate": 0.0004516905505872212,
-      "loss": 3.5668,
+      "epoch": 2.4851752021563343,
+      "grad_norm": 0.5613536834716797,
+      "learning_rate": 0.00045142795466810576,
+      "loss": 3.5645,
       "step": 23050
     },
     {
-      "epoch": 2.4862770422989993,
-      "grad_norm": 0.6530661582946777,
-      "learning_rate": 0.00045136730955715973,
-      "loss": 3.5559,
+      "epoch": 2.490566037735849,
+      "grad_norm": 0.6019158959388733,
+      "learning_rate": 0.0004511041554236373,
+      "loss": 3.5693,
       "step": 23100
     },
     {
-      "epoch": 2.4916585943386074,
-      "grad_norm": 0.6091562509536743,
-      "learning_rate": 0.00045104406852709833,
-      "loss": 3.5529,
+      "epoch": 2.4959568733153636,
+      "grad_norm": 0.5869699716567993,
+      "learning_rate": 0.0004507803561791689,
+      "loss": 3.5652,
       "step": 23150
     },
     {
-      "epoch": 2.4970401463782155,
-      "grad_norm": 0.5845630168914795,
-      "learning_rate": 0.0004507208274970369,
-      "loss": 3.5491,
+      "epoch": 2.501347708894879,
+      "grad_norm": 0.5562755465507507,
+      "learning_rate": 0.00045045655693470046,
+      "loss": 3.5607,
       "step": 23200
     },
     {
-      "epoch": 2.5024216984178236,
-      "grad_norm": 0.6327889561653137,
-      "learning_rate": 0.00045039758646697546,
-      "loss": 3.5468,
+      "epoch": 2.5067385444743935,
+      "grad_norm": 0.5713745951652527,
+      "learning_rate": 0.00045013275769023207,
+      "loss": 3.5615,
       "step": 23250
     },
     {
-      "epoch": 2.5078032504574317,
-      "grad_norm": 0.6545748710632324,
-      "learning_rate": 0.0004500743454369141,
-      "loss": 3.5479,
+      "epoch": 2.512129380053908,
+      "grad_norm": 0.5539861917495728,
+      "learning_rate": 0.00044980895844576356,
+      "loss": 3.5461,
       "step": 23300
     },
     {
-      "epoch": 2.5131848024970402,
-      "grad_norm": 0.5493546724319458,
-      "learning_rate": 0.0004497511044068527,
-      "loss": 3.5505,
+      "epoch": 2.5175202156334233,
+      "grad_norm": 0.5920975208282471,
+      "learning_rate": 0.0004494851592012951,
+      "loss": 3.5544,
       "step": 23350
     },
     {
-      "epoch": 2.5185663545366483,
-      "grad_norm": 0.5882018804550171,
-      "learning_rate": 0.00044942786337679125,
-      "loss": 3.5564,
+      "epoch": 2.522911051212938,
+      "grad_norm": 0.6020824313163757,
+      "learning_rate": 0.0004491613599568267,
+      "loss": 3.5497,
       "step": 23400
     },
     {
-      "epoch": 2.5239479065762565,
-      "grad_norm": 0.5862922072410583,
-      "learning_rate": 0.00044910462234672984,
-      "loss": 3.5399,
+      "epoch": 2.5283018867924527,
+      "grad_norm": 0.5737643241882324,
+      "learning_rate": 0.00044883756071235827,
+      "loss": 3.5615,
       "step": 23450
     },
     {
-      "epoch": 2.529329458615865,
-      "grad_norm": 0.6278120279312134,
-      "learning_rate": 0.0004487813813166684,
-      "loss": 3.5556,
+      "epoch": 2.533692722371968,
+      "grad_norm": 0.5423343181610107,
+      "learning_rate": 0.0004485137614678899,
+      "loss": 3.562,
       "step": 23500
     },
     {
-      "epoch": 2.534711010655473,
-      "grad_norm": 0.5472739338874817,
-      "learning_rate": 0.000448458140286607,
-      "loss": 3.5509,
+      "epoch": 2.5390835579514826,
+      "grad_norm": 0.563010036945343,
+      "learning_rate": 0.0004481899622234214,
+      "loss": 3.566,
       "step": 23550
     },
     {
-      "epoch": 2.540092562695081,
-      "grad_norm": 0.5767682194709778,
-      "learning_rate": 0.00044813489925654563,
-      "loss": 3.5568,
+      "epoch": 2.5444743935309972,
+      "grad_norm": 0.5833520293235779,
+      "learning_rate": 0.00044786616297895303,
+      "loss": 3.5685,
       "step": 23600
     },
     {
-      "epoch": 2.5454741147346893,
-      "grad_norm": 0.5870820879936218,
-      "learning_rate": 0.00044781165822648417,
-      "loss": 3.5482,
+      "epoch": 2.5498652291105124,
+      "grad_norm": 0.5853322148323059,
+      "learning_rate": 0.0004475423637344846,
+      "loss": 3.5609,
       "step": 23650
     },
     {
-      "epoch": 2.550855666774298,
-      "grad_norm": 0.5661634802818298,
-      "learning_rate": 0.00044748841719642276,
-      "loss": 3.5405,
+      "epoch": 2.555256064690027,
+      "grad_norm": 0.5766289234161377,
+      "learning_rate": 0.00044721856449001613,
+      "loss": 3.566,
       "step": 23700
     },
     {
-      "epoch": 2.556237218813906,
-      "grad_norm": 0.5709953904151917,
-      "learning_rate": 0.00044716517616636136,
-      "loss": 3.5596,
+      "epoch": 2.560646900269542,
+      "grad_norm": 0.5451152920722961,
+      "learning_rate": 0.00044689476524554774,
+      "loss": 3.5586,
       "step": 23750
     },
     {
-      "epoch": 2.561618770853514,
-      "grad_norm": 0.6531484127044678,
-      "learning_rate": 0.0004468419351362999,
-      "loss": 3.5527,
+      "epoch": 2.5660377358490565,
+      "grad_norm": 0.6018720865249634,
+      "learning_rate": 0.0004465709660010793,
+      "loss": 3.5716,
       "step": 23800
     },
     {
-      "epoch": 2.567000322893122,
-      "grad_norm": 0.6157983541488647,
-      "learning_rate": 0.0004465186941062385,
-      "loss": 3.5533,
+      "epoch": 2.571428571428571,
+      "grad_norm": 0.6361149549484253,
+      "learning_rate": 0.0004462471667566109,
+      "loss": 3.5777,
       "step": 23850
     },
     {
-      "epoch": 2.5723818749327307,
-      "grad_norm": 0.5923236012458801,
-      "learning_rate": 0.00044619545307617714,
-      "loss": 3.5506,
+      "epoch": 2.5768194070080863,
+      "grad_norm": 0.5743505358695984,
+      "learning_rate": 0.00044592336751214244,
+      "loss": 3.5758,
       "step": 23900
     },
     {
-      "epoch": 2.577763426972339,
-      "grad_norm": 0.5403733253479004,
-      "learning_rate": 0.0004458722120461157,
-      "loss": 3.5499,
+      "epoch": 2.582210242587601,
+      "grad_norm": 0.6800460815429688,
+      "learning_rate": 0.00044559956826767405,
+      "loss": 3.5519,
       "step": 23950
     },
     {
-      "epoch": 2.583144979011947,
-      "grad_norm": 0.5778699517250061,
-      "learning_rate": 0.0004455489710160543,
-      "loss": 3.5278,
+      "epoch": 2.5876010781671157,
+      "grad_norm": 0.558197557926178,
+      "learning_rate": 0.00044527576902320554,
+      "loss": 3.5468,
       "step": 24000
     },
     {
-      "epoch": 2.583144979011947,
-      "eval_accuracy": 0.3667318547775924,
-      "eval_loss": 3.5326128005981445,
-      "eval_runtime": 184.0926,
-      "eval_samples_per_second": 97.837,
-      "eval_steps_per_second": 6.116,
+      "epoch": 2.5876010781671157,
+      "eval_accuracy": 0.3658648047328329,
+      "eval_loss": 3.541566848754883,
+      "eval_runtime": 153.333,
+      "eval_samples_per_second": 117.463,
+      "eval_steps_per_second": 7.343,
       "step": 24000
     },
     {
-      "epoch": 2.5885265310515555,
-      "grad_norm": 0.5626922845840454,
-      "learning_rate": 0.0004452257299859928,
-      "loss": 3.5634,
+      "epoch": 2.592991913746631,
+      "grad_norm": 0.6217593550682068,
+      "learning_rate": 0.00044495196977873715,
+      "loss": 3.5476,
       "step": 24050
     },
     {
-      "epoch": 2.5939080830911636,
-      "grad_norm": 0.6572050452232361,
-      "learning_rate": 0.0004449024889559314,
-      "loss": 3.5348,
+      "epoch": 2.5983827493261455,
+      "grad_norm": 0.5886304974555969,
+      "learning_rate": 0.0004446281705342687,
+      "loss": 3.5491,
       "step": 24100
     },
     {
-      "epoch": 2.5992896351307717,
-      "grad_norm": 0.5784235000610352,
-      "learning_rate": 0.00044457924792587,
-      "loss": 3.5436,
+      "epoch": 2.6037735849056602,
+      "grad_norm": 0.5797514319419861,
+      "learning_rate": 0.00044430437128980025,
+      "loss": 3.5519,
       "step": 24150
     },
     {
-      "epoch": 2.60467118717038,
-      "grad_norm": 0.6261576414108276,
-      "learning_rate": 0.0004442560068958086,
-      "loss": 3.5527,
+      "epoch": 2.6091644204851754,
+      "grad_norm": 0.6183805465698242,
+      "learning_rate": 0.00044398057204533185,
+      "loss": 3.5687,
       "step": 24200
     },
     {
-      "epoch": 2.610052739209988,
-      "grad_norm": 0.5937851071357727,
-      "learning_rate": 0.0004439327658657472,
-      "loss": 3.5537,
+      "epoch": 2.61455525606469,
+      "grad_norm": 0.5732604265213013,
+      "learning_rate": 0.0004436567728008634,
+      "loss": 3.5712,
       "step": 24250
     },
     {
-      "epoch": 2.6154342912495965,
-      "grad_norm": 0.6657639741897583,
-      "learning_rate": 0.0004436095248356858,
-      "loss": 3.5508,
+      "epoch": 2.6199460916442048,
+      "grad_norm": 0.619526207447052,
+      "learning_rate": 0.000443332973556395,
+      "loss": 3.5583,
       "step": 24300
     },
     {
-      "epoch": 2.6208158432892046,
-      "grad_norm": 0.5803267359733582,
-      "learning_rate": 0.00044328628380562433,
-      "loss": 3.5441,
+      "epoch": 2.62533692722372,
+      "grad_norm": 0.5381686687469482,
+      "learning_rate": 0.00044300917431192656,
+      "loss": 3.5493,
       "step": 24350
     },
     {
-      "epoch": 2.6261973953288127,
-      "grad_norm": 0.5679405927658081,
-      "learning_rate": 0.0004429630427755629,
-      "loss": 3.5489,
+      "epoch": 2.6307277628032346,
+      "grad_norm": 0.5444945693016052,
+      "learning_rate": 0.00044268537506745816,
+      "loss": 3.5612,
       "step": 24400
     },
     {
-      "epoch": 2.6315789473684212,
-      "grad_norm": 0.6222752928733826,
-      "learning_rate": 0.0004426398017455016,
-      "loss": 3.5365,
+      "epoch": 2.6361185983827493,
+      "grad_norm": 0.5936858057975769,
+      "learning_rate": 0.0004423615758229897,
+      "loss": 3.5355,
       "step": 24450
     },
     {
-      "epoch": 2.6369604994080293,
-      "grad_norm": 0.5961997509002686,
-      "learning_rate": 0.0004423165607154401,
-      "loss": 3.5326,
+      "epoch": 2.641509433962264,
+      "grad_norm": 0.5836228728294373,
+      "learning_rate": 0.0004420377765785213,
+      "loss": 3.5424,
       "step": 24500
     },
     {
-      "epoch": 2.6423420514476375,
-      "grad_norm": 0.5750383138656616,
-      "learning_rate": 0.0004419933196853787,
-      "loss": 3.5383,
+      "epoch": 2.6469002695417787,
+      "grad_norm": 0.8124279975891113,
+      "learning_rate": 0.00044171397733405287,
+      "loss": 3.5597,
       "step": 24550
     },
     {
-      "epoch": 2.6477236034872456,
-      "grad_norm": 0.5970505475997925,
-      "learning_rate": 0.00044167007865531725,
-      "loss": 3.5428,
+      "epoch": 2.652291105121294,
+      "grad_norm": 0.5548441410064697,
+      "learning_rate": 0.00044139017808958437,
+      "loss": 3.5389,
       "step": 24600
     },
     {
-      "epoch": 2.653105155526854,
-      "grad_norm": 0.5977314710617065,
-      "learning_rate": 0.00044134683762525584,
-      "loss": 3.5382,
+      "epoch": 2.6576819407008085,
+      "grad_norm": 0.5583662390708923,
+      "learning_rate": 0.00044106637884511597,
+      "loss": 3.5528,
       "step": 24650
     },
     {
-      "epoch": 2.658486707566462,
-      "grad_norm": 0.6956198811531067,
-      "learning_rate": 0.00044102359659519444,
-      "loss": 3.5359,
+      "epoch": 2.6630727762803232,
+      "grad_norm": 0.582872748374939,
+      "learning_rate": 0.0004407425796006475,
+      "loss": 3.5542,
       "step": 24700
     },
     {
-      "epoch": 2.6638682596060703,
-      "grad_norm": 0.6126677989959717,
-      "learning_rate": 0.00044070682038573425,
-      "loss": 3.5476,
+      "epoch": 2.6684636118598384,
+      "grad_norm": 0.6061405539512634,
+      "learning_rate": 0.00044041878035617913,
+      "loss": 3.558,
       "step": 24750
     },
     {
-      "epoch": 2.6692498116456784,
-      "grad_norm": 0.5887094140052795,
-      "learning_rate": 0.0004403835793556728,
-      "loss": 3.543,
+      "epoch": 2.673854447439353,
+      "grad_norm": 0.5676862001419067,
+      "learning_rate": 0.0004400949811117107,
+      "loss": 3.5549,
       "step": 24800
     },
     {
-      "epoch": 2.674631363685287,
-      "grad_norm": 0.6262331604957581,
-      "learning_rate": 0.00044006033832561143,
-      "loss": 3.5405,
+      "epoch": 2.6792452830188678,
+      "grad_norm": 0.5791465640068054,
+      "learning_rate": 0.0004397711818672423,
+      "loss": 3.5568,
       "step": 24850
     },
     {
-      "epoch": 2.680012915724895,
-      "grad_norm": 0.5854912996292114,
-      "learning_rate": 0.00043973709729555003,
-      "loss": 3.5519,
+      "epoch": 2.684636118598383,
+      "grad_norm": 0.6262496113777161,
+      "learning_rate": 0.00043944738262277383,
+      "loss": 3.5727,
       "step": 24900
     },
     {
-      "epoch": 2.685394467764503,
-      "grad_norm": 0.62160724401474,
-      "learning_rate": 0.00043941385626548857,
-      "loss": 3.5514,
+      "epoch": 2.6900269541778976,
+      "grad_norm": 0.5959411263465881,
+      "learning_rate": 0.00043912358337830544,
+      "loss": 3.5615,
       "step": 24950
     },
     {
-      "epoch": 2.6907760198041117,
-      "grad_norm": 0.57602459192276,
-      "learning_rate": 0.00043909061523542716,
-      "loss": 3.5471,
+      "epoch": 2.6954177897574123,
+      "grad_norm": 0.577245831489563,
+      "learning_rate": 0.000438799784133837,
+      "loss": 3.5558,
       "step": 25000
     },
     {
-      "epoch": 2.6907760198041117,
-      "eval_accuracy": 0.36766931189365576,
-      "eval_loss": 3.522217273712158,
-      "eval_runtime": 183.3693,
-      "eval_samples_per_second": 98.223,
-      "eval_steps_per_second": 6.141,
+      "epoch": 2.6954177897574123,
+      "eval_accuracy": 0.3667249009927723,
+      "eval_loss": 3.530350685119629,
+      "eval_runtime": 153.0546,
+      "eval_samples_per_second": 117.677,
+      "eval_steps_per_second": 7.357,
       "step": 25000
     },
     {
-      "epoch": 2.69615757184372,
-      "grad_norm": 0.5819248557090759,
-      "learning_rate": 0.00043876737420536576,
-      "loss": 3.5299,
+      "epoch": 2.7008086253369274,
+      "grad_norm": 0.594308078289032,
+      "learning_rate": 0.00043847598488936854,
+      "loss": 3.5478,
       "step": 25050
     },
     {
-      "epoch": 2.701539123883328,
-      "grad_norm": 0.6219841837882996,
-      "learning_rate": 0.00043844413317530435,
-      "loss": 3.5267,
+      "epoch": 2.706199460916442,
+      "grad_norm": 0.5546954274177551,
+      "learning_rate": 0.00043815218564490014,
+      "loss": 3.5571,
       "step": 25100
     },
     {
-      "epoch": 2.706920675922936,
-      "grad_norm": 0.6201439499855042,
-      "learning_rate": 0.00043812089214524295,
-      "loss": 3.5541,
+      "epoch": 2.711590296495957,
+      "grad_norm": 0.6364242434501648,
+      "learning_rate": 0.0004378283864004317,
+      "loss": 3.5332,
       "step": 25150
     },
     {
-      "epoch": 2.712302227962544,
-      "grad_norm": 0.6550151705741882,
-      "learning_rate": 0.00043779765111518154,
-      "loss": 3.5478,
+      "epoch": 2.7169811320754715,
+      "grad_norm": 0.5977651476860046,
+      "learning_rate": 0.0004375045871559633,
+      "loss": 3.5564,
       "step": 25200
     },
     {
-      "epoch": 2.7176837800021527,
-      "grad_norm": 0.6322425603866577,
-      "learning_rate": 0.0004374744100851201,
-      "loss": 3.55,
+      "epoch": 2.7223719676549867,
+      "grad_norm": 0.6769561171531677,
+      "learning_rate": 0.00043718726389638425,
+      "loss": 3.5599,
       "step": 25250
     },
     {
-      "epoch": 2.723065332041761,
-      "grad_norm": 0.6911249756813049,
-      "learning_rate": 0.0004371511690550587,
-      "loss": 3.553,
+      "epoch": 2.7277628032345014,
+      "grad_norm": 0.5432511568069458,
+      "learning_rate": 0.00043686346465191575,
+      "loss": 3.5624,
       "step": 25300
     },
     {
-      "epoch": 2.728446884081369,
-      "grad_norm": 0.5741549730300903,
-      "learning_rate": 0.0004368279280249972,
-      "loss": 3.5331,
+      "epoch": 2.733153638814016,
+      "grad_norm": 0.5543899536132812,
+      "learning_rate": 0.0004365396654074473,
+      "loss": 3.5528,
       "step": 25350
     },
     {
-      "epoch": 2.7338284361209775,
-      "grad_norm": 0.6233206987380981,
-      "learning_rate": 0.00043650468699493587,
-      "loss": 3.5434,
+      "epoch": 2.7385444743935308,
+      "grad_norm": 0.645545244216919,
+      "learning_rate": 0.0004362158661629789,
+      "loss": 3.5424,
       "step": 25400
     },
     {
-      "epoch": 2.7392099881605856,
-      "grad_norm": 0.6208220720291138,
-      "learning_rate": 0.00043618144596487446,
-      "loss": 3.5101,
+      "epoch": 2.743935309973046,
+      "grad_norm": 0.5830022692680359,
+      "learning_rate": 0.00043589206691851045,
+      "loss": 3.5438,
       "step": 25450
     },
     {
-      "epoch": 2.7445915402001937,
-      "grad_norm": 0.5892960429191589,
-      "learning_rate": 0.000435858204934813,
-      "loss": 3.5372,
+      "epoch": 2.7493261455525606,
+      "grad_norm": 0.6203137636184692,
+      "learning_rate": 0.00043556826767404206,
+      "loss": 3.5594,
       "step": 25500
     },
     {
-      "epoch": 2.749973092239802,
-      "grad_norm": 0.6114374399185181,
-      "learning_rate": 0.0004355349639047516,
-      "loss": 3.529,
+      "epoch": 2.7547169811320753,
+      "grad_norm": 0.6280099153518677,
+      "learning_rate": 0.0004352444684295736,
+      "loss": 3.5518,
       "step": 25550
     },
     {
-      "epoch": 2.7553546442794103,
-      "grad_norm": 0.6182452440261841,
-      "learning_rate": 0.0004352117228746902,
-      "loss": 3.5209,
+      "epoch": 2.7601078167115904,
+      "grad_norm": 0.6008601188659668,
+      "learning_rate": 0.0004349206691851052,
+      "loss": 3.5539,
       "step": 25600
     },
     {
-      "epoch": 2.7607361963190185,
-      "grad_norm": 0.5845827460289001,
-      "learning_rate": 0.00043488848184462873,
-      "loss": 3.5392,
+      "epoch": 2.765498652291105,
+      "grad_norm": 0.546109139919281,
+      "learning_rate": 0.00043459686994063676,
+      "loss": 3.5449,
       "step": 25650
     },
     {
-      "epoch": 2.7661177483586266,
-      "grad_norm": 0.6704187989234924,
-      "learning_rate": 0.0004345652408145674,
-      "loss": 3.5356,
+      "epoch": 2.77088948787062,
+      "grad_norm": 0.6347863078117371,
+      "learning_rate": 0.00043427307069616837,
+      "loss": 3.5517,
       "step": 25700
     },
     {
-      "epoch": 2.7714993003982347,
-      "grad_norm": 0.5977811217308044,
-      "learning_rate": 0.000434241999784506,
-      "loss": 3.5524,
+      "epoch": 2.776280323450135,
+      "grad_norm": 0.5688282251358032,
+      "learning_rate": 0.0004339492714516999,
+      "loss": 3.5418,
       "step": 25750
     },
     {
-      "epoch": 2.776880852437843,
-      "grad_norm": 0.6204485893249512,
-      "learning_rate": 0.0004339187587544445,
-      "loss": 3.5414,
+      "epoch": 2.7816711590296497,
+      "grad_norm": 0.5507652759552002,
+      "learning_rate": 0.00043362547220723147,
+      "loss": 3.5658,
       "step": 25800
     },
     {
-      "epoch": 2.7822624044774513,
-      "grad_norm": 0.6410983800888062,
-      "learning_rate": 0.0004335955177243831,
-      "loss": 3.5416,
+      "epoch": 2.7870619946091644,
+      "grad_norm": 0.5676924586296082,
+      "learning_rate": 0.0004333016729627631,
+      "loss": 3.5359,
       "step": 25850
     },
     {
-      "epoch": 2.7876439565170594,
-      "grad_norm": 0.632500946521759,
-      "learning_rate": 0.00043327227669432165,
-      "loss": 3.5271,
+      "epoch": 2.7924528301886795,
+      "grad_norm": 0.6018034219741821,
+      "learning_rate": 0.0004329778737182946,
+      "loss": 3.5475,
       "step": 25900
     },
     {
-      "epoch": 2.793025508556668,
-      "grad_norm": 0.6137693524360657,
-      "learning_rate": 0.00043294903566426025,
-      "loss": 3.5518,
+      "epoch": 2.797843665768194,
+      "grad_norm": 0.5850654244422913,
+      "learning_rate": 0.00043265407447382623,
+      "loss": 3.5441,
       "step": 25950
     },
     {
-      "epoch": 2.798407060596276,
-      "grad_norm": 0.6099450588226318,
-      "learning_rate": 0.00043263225945480005,
-      "loss": 3.5396,
+      "epoch": 2.803234501347709,
+      "grad_norm": 0.5349710583686829,
+      "learning_rate": 0.0004323302752293577,
+      "loss": 3.5606,
       "step": 26000
     },
     {
-      "epoch": 2.798407060596276,
-      "eval_accuracy": 0.36848779409756227,
-      "eval_loss": 3.513767719268799,
-      "eval_runtime": 184.121,
-      "eval_samples_per_second": 97.822,
-      "eval_steps_per_second": 6.116,
+      "epoch": 2.803234501347709,
+      "eval_accuracy": 0.3676626840674991,
+      "eval_loss": 3.520634651184082,
+      "eval_runtime": 153.4664,
+      "eval_samples_per_second": 117.361,
+      "eval_steps_per_second": 7.337,
       "step": 26000
     },
     {
-      "epoch": 2.803788612635884,
-      "grad_norm": 0.6080275774002075,
-      "learning_rate": 0.0004323090184247387,
-      "loss": 3.532,
+      "epoch": 2.8086253369272236,
+      "grad_norm": 0.6177300810813904,
+      "learning_rate": 0.00043200647598488933,
+      "loss": 3.564,
       "step": 26050
     },
     {
-      "epoch": 2.8091701646754923,
-      "grad_norm": 0.6676295399665833,
-      "learning_rate": 0.0004319857773946773,
-      "loss": 3.5422,
+      "epoch": 2.8140161725067383,
+      "grad_norm": 0.5958245396614075,
+      "learning_rate": 0.0004316826767404209,
+      "loss": 3.5497,
       "step": 26100
     },
     {
-      "epoch": 2.8145517167151004,
-      "grad_norm": 0.6267027854919434,
-      "learning_rate": 0.00043166253636461584,
-      "loss": 3.5366,
+      "epoch": 2.8194070080862534,
+      "grad_norm": 0.5535616278648376,
+      "learning_rate": 0.0004313588774959525,
+      "loss": 3.5466,
       "step": 26150
     },
     {
-      "epoch": 2.819933268754709,
-      "grad_norm": 0.6126961708068848,
-      "learning_rate": 0.00043133929533455443,
-      "loss": 3.5592,
+      "epoch": 2.824797843665768,
+      "grad_norm": 0.5271294116973877,
+      "learning_rate": 0.00043103507825148404,
+      "loss": 3.5569,
       "step": 26200
     },
     {
-      "epoch": 2.825314820794317,
-      "grad_norm": 0.6002199053764343,
-      "learning_rate": 0.00043101605430449297,
-      "loss": 3.5227,
+      "epoch": 2.830188679245283,
+      "grad_norm": 0.5427255034446716,
+      "learning_rate": 0.0004307112790070156,
+      "loss": 3.5379,
       "step": 26250
     },
     {
-      "epoch": 2.830696372833925,
-      "grad_norm": 0.6439182758331299,
-      "learning_rate": 0.00043069281327443157,
-      "loss": 3.5576,
+      "epoch": 2.835579514824798,
+      "grad_norm": 0.6394116282463074,
+      "learning_rate": 0.0004303874797625472,
+      "loss": 3.5387,
       "step": 26300
     },
     {
-      "epoch": 2.8360779248735337,
-      "grad_norm": 0.6624460816383362,
-      "learning_rate": 0.0004303695722443702,
-      "loss": 3.5325,
+      "epoch": 2.8409703504043127,
+      "grad_norm": 0.612621009349823,
+      "learning_rate": 0.00043006368051807874,
+      "loss": 3.5596,
       "step": 26350
     },
     {
-      "epoch": 2.841459476913142,
-      "grad_norm": 0.5866040587425232,
-      "learning_rate": 0.00043004633121430876,
-      "loss": 3.513,
+      "epoch": 2.8463611859838274,
+      "grad_norm": 0.5625531673431396,
+      "learning_rate": 0.00042973988127361035,
+      "loss": 3.5332,
       "step": 26400
     },
     {
-      "epoch": 2.84684102895275,
-      "grad_norm": 0.5632692575454712,
-      "learning_rate": 0.00042972309018424735,
-      "loss": 3.542,
+      "epoch": 2.8517520215633425,
+      "grad_norm": 0.612446665763855,
+      "learning_rate": 0.0004294160820291419,
+      "loss": 3.5558,
       "step": 26450
     },
     {
-      "epoch": 2.852222580992358,
-      "grad_norm": 0.6331861615180969,
-      "learning_rate": 0.00042939984915418594,
-      "loss": 3.5396,
+      "epoch": 2.857142857142857,
+      "grad_norm": 0.5687634348869324,
+      "learning_rate": 0.0004290922827846735,
+      "loss": 3.5627,
       "step": 26500
     },
     {
-      "epoch": 2.857604133031966,
-      "grad_norm": 0.6353737711906433,
-      "learning_rate": 0.0004290766081241245,
-      "loss": 3.5167,
+      "epoch": 2.862533692722372,
+      "grad_norm": 0.6355977654457092,
+      "learning_rate": 0.00042876848354020505,
+      "loss": 3.5437,
       "step": 26550
     },
     {
-      "epoch": 2.8629856850715747,
-      "grad_norm": 0.5899486541748047,
-      "learning_rate": 0.0004287533670940631,
-      "loss": 3.5364,
+      "epoch": 2.867924528301887,
+      "grad_norm": 0.5741527080535889,
+      "learning_rate": 0.00042844468429573655,
+      "loss": 3.5444,
       "step": 26600
     },
     {
-      "epoch": 2.868367237111183,
-      "grad_norm": 0.6116732954978943,
-      "learning_rate": 0.00042843012606400173,
-      "loss": 3.5369,
+      "epoch": 2.8733153638814017,
+      "grad_norm": 0.5699136853218079,
+      "learning_rate": 0.00042812088505126815,
+      "loss": 3.5299,
       "step": 26650
     },
     {
-      "epoch": 2.873748789150791,
-      "grad_norm": 0.6197775602340698,
-      "learning_rate": 0.00042810688503394027,
-      "loss": 3.53,
+      "epoch": 2.8787061994609164,
+      "grad_norm": 0.5550360083580017,
+      "learning_rate": 0.0004277970858067997,
+      "loss": 3.5686,
       "step": 26700
     },
     {
-      "epoch": 2.8791303411903995,
-      "grad_norm": 0.626814067363739,
-      "learning_rate": 0.00042778364400387886,
-      "loss": 3.5297,
+      "epoch": 2.884097035040431,
+      "grad_norm": 0.5849996209144592,
+      "learning_rate": 0.0004274732865623313,
+      "loss": 3.5375,
       "step": 26750
     },
     {
-      "epoch": 2.8845118932300076,
-      "grad_norm": 0.6053553223609924,
-      "learning_rate": 0.0004274604029738174,
-      "loss": 3.533,
+      "epoch": 2.889487870619946,
+      "grad_norm": 0.5445963740348816,
+      "learning_rate": 0.00042714948731786286,
+      "loss": 3.5205,
       "step": 26800
     },
     {
-      "epoch": 2.8898934452696157,
-      "grad_norm": 0.611613929271698,
-      "learning_rate": 0.000427137161943756,
-      "loss": 3.529,
+      "epoch": 2.894878706199461,
+      "grad_norm": 0.5899405479431152,
+      "learning_rate": 0.00042682568807339447,
+      "loss": 3.5233,
       "step": 26850
     },
     {
-      "epoch": 2.895274997309224,
-      "grad_norm": 0.6430195569992065,
-      "learning_rate": 0.00042681392091369465,
-      "loss": 3.5079,
+      "epoch": 2.9002695417789757,
+      "grad_norm": 0.5955359935760498,
+      "learning_rate": 0.000426501888828926,
+      "loss": 3.5429,
       "step": 26900
     },
     {
-      "epoch": 2.9006565493488323,
-      "grad_norm": 0.5988154411315918,
-      "learning_rate": 0.0004264906798836332,
-      "loss": 3.5295,
+      "epoch": 2.9056603773584904,
+      "grad_norm": 0.6034966111183167,
+      "learning_rate": 0.0004261780895844576,
+      "loss": 3.5632,
       "step": 26950
     },
     {
-      "epoch": 2.9060381013884404,
-      "grad_norm": 0.6235425472259521,
-      "learning_rate": 0.0004261674388535718,
-      "loss": 3.5357,
+      "epoch": 2.9110512129380055,
+      "grad_norm": 0.7126505970954895,
+      "learning_rate": 0.00042585429033998917,
+      "loss": 3.5451,
       "step": 27000
     },
     {
-      "epoch": 2.9060381013884404,
-      "eval_accuracy": 0.3692725939062462,
-      "eval_loss": 3.503479480743408,
-      "eval_runtime": 183.7528,
-      "eval_samples_per_second": 98.018,
-      "eval_steps_per_second": 6.128,
+      "epoch": 2.9110512129380055,
+      "eval_accuracy": 0.36876551087881604,
+      "eval_loss": 3.5126547813415527,
+      "eval_runtime": 153.3905,
+      "eval_samples_per_second": 117.419,
+      "eval_steps_per_second": 7.341,
       "step": 27000
     },
     {
-      "epoch": 2.9114196534280485,
-      "grad_norm": 0.6861902475357056,
-      "learning_rate": 0.0004258441978235104,
-      "loss": 3.5353,
+      "epoch": 2.91644204851752,
+      "grad_norm": 0.6429764628410339,
+      "learning_rate": 0.0004255304910955207,
+      "loss": 3.534,
       "step": 27050
     },
     {
-      "epoch": 2.9168012054676566,
-      "grad_norm": 0.5947350859642029,
-      "learning_rate": 0.0004255209567934489,
-      "loss": 3.5347,
+      "epoch": 2.921832884097035,
+      "grad_norm": 0.6181586980819702,
+      "learning_rate": 0.0004252066918510523,
+      "loss": 3.5407,
       "step": 27100
     },
     {
-      "epoch": 2.922182757507265,
-      "grad_norm": 0.6242401003837585,
-      "learning_rate": 0.0004251977157633875,
-      "loss": 3.5268,
+      "epoch": 2.92722371967655,
+      "grad_norm": 0.628360390663147,
+      "learning_rate": 0.0004248828926065839,
+      "loss": 3.5374,
       "step": 27150
     },
     {
-      "epoch": 2.9275643095468733,
-      "grad_norm": 0.627075731754303,
-      "learning_rate": 0.00042487447473332616,
-      "loss": 3.5198,
+      "epoch": 2.9326145552560647,
+      "grad_norm": 0.5996028780937195,
+      "learning_rate": 0.0004245590933621155,
+      "loss": 3.5454,
       "step": 27200
     },
     {
-      "epoch": 2.9329458615864814,
-      "grad_norm": 0.6009472608566284,
-      "learning_rate": 0.0004245512337032647,
-      "loss": 3.5408,
+      "epoch": 2.9380053908355794,
+      "grad_norm": 0.5633025765419006,
+      "learning_rate": 0.00042423529411764703,
+      "loss": 3.5514,
       "step": 27250
     },
     {
-      "epoch": 2.93832741362609,
-      "grad_norm": 0.5950798988342285,
-      "learning_rate": 0.0004242279926732033,
-      "loss": 3.5415,
+      "epoch": 2.9433962264150946,
+      "grad_norm": 0.6133574843406677,
+      "learning_rate": 0.00042391149487317864,
+      "loss": 3.5166,
       "step": 27300
     },
     {
-      "epoch": 2.943708965665698,
-      "grad_norm": 0.6193643808364868,
-      "learning_rate": 0.00042390475164314184,
-      "loss": 3.5347,
+      "epoch": 2.9487870619946093,
+      "grad_norm": 0.5703015327453613,
+      "learning_rate": 0.0004235941716135995,
+      "loss": 3.5365,
       "step": 27350
     },
     {
-      "epoch": 2.949090517705306,
-      "grad_norm": 0.5791000723838806,
-      "learning_rate": 0.00042358151061308043,
-      "loss": 3.5357,
+      "epoch": 2.954177897574124,
+      "grad_norm": 0.5899073481559753,
+      "learning_rate": 0.0004232703723691311,
+      "loss": 3.5314,
       "step": 27400
     },
     {
-      "epoch": 2.9544720697449143,
-      "grad_norm": 0.6314085721969604,
-      "learning_rate": 0.000423258269583019,
-      "loss": 3.5148,
+      "epoch": 2.9595687331536387,
+      "grad_norm": 0.5997247695922852,
+      "learning_rate": 0.00042294657312466264,
+      "loss": 3.5231,
       "step": 27450
     },
     {
-      "epoch": 2.9598536217845224,
-      "grad_norm": 0.5699372291564941,
-      "learning_rate": 0.0004229350285529576,
-      "loss": 3.5322,
+      "epoch": 2.964959568733154,
+      "grad_norm": 0.558027982711792,
+      "learning_rate": 0.00042262277388019424,
+      "loss": 3.5498,
       "step": 27500
     },
     {
-      "epoch": 2.965235173824131,
-      "grad_norm": 0.6208819150924683,
-      "learning_rate": 0.0004226117875228962,
-      "loss": 3.5076,
+      "epoch": 2.9703504043126685,
+      "grad_norm": 0.5814929008483887,
+      "learning_rate": 0.0004222989746357258,
+      "loss": 3.5511,
       "step": 27550
     },
     {
-      "epoch": 2.970616725863739,
-      "grad_norm": 0.6337210536003113,
-      "learning_rate": 0.0004222885464928348,
-      "loss": 3.5351,
+      "epoch": 2.975741239892183,
+      "grad_norm": 0.5788719654083252,
+      "learning_rate": 0.0004219751753912574,
+      "loss": 3.5327,
       "step": 27600
     },
     {
-      "epoch": 2.975998277903347,
-      "grad_norm": 0.5993634462356567,
-      "learning_rate": 0.00042196530546277335,
-      "loss": 3.5361,
+      "epoch": 2.981132075471698,
+      "grad_norm": 0.5898560881614685,
+      "learning_rate": 0.00042165137614678895,
+      "loss": 3.538,
       "step": 27650
     },
     {
-      "epoch": 2.9813798299429557,
-      "grad_norm": 0.617313802242279,
-      "learning_rate": 0.00042164206443271195,
-      "loss": 3.5267,
+      "epoch": 2.986522911051213,
+      "grad_norm": 0.5724667906761169,
+      "learning_rate": 0.00042132757690232055,
+      "loss": 3.5281,
       "step": 27700
     },
     {
-      "epoch": 2.986761381982564,
-      "grad_norm": 0.6192960739135742,
-      "learning_rate": 0.0004213188234026505,
-      "loss": 3.523,
+      "epoch": 2.9919137466307277,
+      "grad_norm": 0.6660099625587463,
+      "learning_rate": 0.0004210037776578521,
+      "loss": 3.5262,
       "step": 27750
     },
     {
-      "epoch": 2.992142934022172,
-      "grad_norm": 0.6196896433830261,
-      "learning_rate": 0.00042099558237258914,
-      "loss": 3.539,
+      "epoch": 2.9973045822102424,
+      "grad_norm": 0.6116336584091187,
+      "learning_rate": 0.00042067997841338365,
+      "loss": 3.5368,
       "step": 27800
     },
     {
-      "epoch": 2.9975244860617805,
-      "grad_norm": 0.6242361664772034,
-      "learning_rate": 0.00042067234134252773,
-      "loss": 3.5244,
+      "epoch": 3.0026954177897576,
+      "grad_norm": 0.5541266202926636,
+      "learning_rate": 0.00042035617916891526,
+      "loss": 3.4973,
       "step": 27850
     },
     {
-      "epoch": 3.0029060381013886,
-      "grad_norm": 0.5909894108772278,
-      "learning_rate": 0.00042034910031246627,
-      "loss": 3.4789,
+      "epoch": 3.0080862533692723,
+      "grad_norm": 0.6395186185836792,
+      "learning_rate": 0.0004200323799244468,
+      "loss": 3.4449,
       "step": 27900
     },
     {
-      "epoch": 3.0082875901409967,
-      "grad_norm": 0.6177986264228821,
-      "learning_rate": 0.00042002585928240486,
-      "loss": 3.4404,
+      "epoch": 3.013477088948787,
+      "grad_norm": 0.5770643949508667,
+      "learning_rate": 0.0004197085806799784,
+      "loss": 3.4453,
       "step": 27950
     },
     {
-      "epoch": 3.0136691421806048,
-      "grad_norm": 0.6032999753952026,
-      "learning_rate": 0.00041970261825234346,
-      "loss": 3.4295,
+      "epoch": 3.018867924528302,
+      "grad_norm": 0.5565028190612793,
+      "learning_rate": 0.0004193847814355099,
+      "loss": 3.4453,
       "step": 28000
     },
     {
-      "epoch": 3.0136691421806048,
-      "eval_accuracy": 0.3703062088280254,
-      "eval_loss": 3.500861406326294,
-      "eval_runtime": 183.788,
-      "eval_samples_per_second": 97.999,
-      "eval_steps_per_second": 6.127,
+      "epoch": 3.018867924528302,
+      "eval_accuracy": 0.36929378121937007,
+      "eval_loss": 3.5084283351898193,
+      "eval_runtime": 153.4117,
+      "eval_samples_per_second": 117.403,
+      "eval_steps_per_second": 7.34,
       "step": 28000
     },
     {
-      "epoch": 3.0190506942202133,
-      "grad_norm": 0.591021716594696,
-      "learning_rate": 0.00041938584204288327,
-      "loss": 3.4332,
+      "epoch": 3.024258760107817,
+      "grad_norm": 0.5812212824821472,
+      "learning_rate": 0.0004190609821910415,
+      "loss": 3.4429,
       "step": 28050
     },
     {
-      "epoch": 3.0244322462598214,
-      "grad_norm": 0.6264454126358032,
-      "learning_rate": 0.0004190626010128218,
-      "loss": 3.4296,
+      "epoch": 3.0296495956873315,
+      "grad_norm": 0.5854341983795166,
+      "learning_rate": 0.00041873718294657306,
+      "loss": 3.4504,
       "step": 28100
     },
     {
-      "epoch": 3.0298137982994295,
-      "grad_norm": 0.6042491793632507,
-      "learning_rate": 0.00041873935998276046,
-      "loss": 3.4508,
+      "epoch": 3.035040431266846,
+      "grad_norm": 0.6041473150253296,
+      "learning_rate": 0.00041841338370210467,
+      "loss": 3.4338,
       "step": 28150
     },
     {
-      "epoch": 3.0351953503390376,
-      "grad_norm": 0.6125722527503967,
-      "learning_rate": 0.00041841611895269905,
-      "loss": 3.4274,
+      "epoch": 3.0404312668463613,
+      "grad_norm": 0.6009054183959961,
+      "learning_rate": 0.0004180895844576362,
+      "loss": 3.4518,
       "step": 28200
     },
     {
-      "epoch": 3.040576902378646,
-      "grad_norm": 0.6271188855171204,
-      "learning_rate": 0.0004180928779226376,
-      "loss": 3.4471,
+      "epoch": 3.045822102425876,
+      "grad_norm": 0.5958666801452637,
+      "learning_rate": 0.00041776578521316777,
+      "loss": 3.4422,
       "step": 28250
     },
     {
-      "epoch": 3.0459584544182543,
-      "grad_norm": 0.6595613956451416,
-      "learning_rate": 0.0004177696368925762,
-      "loss": 3.4556,
+      "epoch": 3.0512129380053907,
+      "grad_norm": 0.574678897857666,
+      "learning_rate": 0.0004174419859686994,
+      "loss": 3.4661,
       "step": 28300
     },
     {
-      "epoch": 3.0513400064578624,
-      "grad_norm": 0.6097109317779541,
-      "learning_rate": 0.0004174463958625148,
-      "loss": 3.4446,
+      "epoch": 3.056603773584906,
+      "grad_norm": 0.614147961139679,
+      "learning_rate": 0.0004171181867242309,
+      "loss": 3.449,
       "step": 28350
     },
     {
-      "epoch": 3.0567215584974705,
-      "grad_norm": 0.6225860714912415,
-      "learning_rate": 0.0004171231548324533,
-      "loss": 3.4311,
+      "epoch": 3.0619946091644206,
+      "grad_norm": 0.61612868309021,
+      "learning_rate": 0.00041679438747976253,
+      "loss": 3.4703,
       "step": 28400
     },
     {
-      "epoch": 3.062103110537079,
-      "grad_norm": 0.589008629322052,
-      "learning_rate": 0.00041679991380239197,
-      "loss": 3.4261,
+      "epoch": 3.0673854447439353,
+      "grad_norm": 0.5863500237464905,
+      "learning_rate": 0.0004164705882352941,
+      "loss": 3.4495,
       "step": 28450
     },
     {
-      "epoch": 3.067484662576687,
-      "grad_norm": 0.6507358551025391,
-      "learning_rate": 0.00041647667277233056,
-      "loss": 3.4413,
+      "epoch": 3.07277628032345,
+      "grad_norm": 0.6393812298774719,
+      "learning_rate": 0.0004161467889908257,
+      "loss": 3.4427,
       "step": 28500
     },
     {
-      "epoch": 3.0728662146162953,
-      "grad_norm": 0.6316334009170532,
-      "learning_rate": 0.0004161534317422691,
-      "loss": 3.4423,
+      "epoch": 3.078167115902965,
+      "grad_norm": 0.6059166789054871,
+      "learning_rate": 0.00041582298974635724,
+      "loss": 3.4466,
       "step": 28550
     },
     {
-      "epoch": 3.0782477666559034,
-      "grad_norm": 0.6280677318572998,
-      "learning_rate": 0.0004158301907122077,
-      "loss": 3.4597,
+      "epoch": 3.08355795148248,
+      "grad_norm": 0.5664580464363098,
+      "learning_rate": 0.00041549919050188884,
+      "loss": 3.4649,
       "step": 28600
     },
     {
-      "epoch": 3.083629318695512,
-      "grad_norm": 0.6200211048126221,
-      "learning_rate": 0.00041550694968214624,
-      "loss": 3.4607,
+      "epoch": 3.0889487870619945,
+      "grad_norm": 0.7083315849304199,
+      "learning_rate": 0.00041517539125742034,
+      "loss": 3.4521,
       "step": 28650
     },
     {
-      "epoch": 3.08901087073512,
-      "grad_norm": 0.6185154318809509,
-      "learning_rate": 0.0004151837086520849,
-      "loss": 3.4432,
+      "epoch": 3.0943396226415096,
+      "grad_norm": 0.6239797472953796,
+      "learning_rate": 0.0004148515920129519,
+      "loss": 3.4709,
       "step": 28700
     },
     {
-      "epoch": 3.094392422774728,
-      "grad_norm": 0.6128413081169128,
-      "learning_rate": 0.0004148604676220235,
-      "loss": 3.4402,
+      "epoch": 3.0997304582210243,
+      "grad_norm": 0.6037058234214783,
+      "learning_rate": 0.0004145277927684835,
+      "loss": 3.4368,
       "step": 28750
     },
     {
-      "epoch": 3.0997739748143363,
-      "grad_norm": 0.589779257774353,
-      "learning_rate": 0.000414537226591962,
-      "loss": 3.4371,
+      "epoch": 3.105121293800539,
+      "grad_norm": 0.5654220581054688,
+      "learning_rate": 0.00041420399352401504,
+      "loss": 3.4649,
       "step": 28800
     },
     {
-      "epoch": 3.105155526853945,
-      "grad_norm": 0.6613255143165588,
-      "learning_rate": 0.0004142139855619006,
-      "loss": 3.4356,
+      "epoch": 3.1105121293800537,
+      "grad_norm": 0.5870050191879272,
+      "learning_rate": 0.00041388019427954665,
+      "loss": 3.4538,
       "step": 28850
     },
     {
-      "epoch": 3.110537078893553,
-      "grad_norm": 0.6076911687850952,
-      "learning_rate": 0.0004138907445318392,
-      "loss": 3.4368,
+      "epoch": 3.115902964959569,
+      "grad_norm": 0.6253658533096313,
+      "learning_rate": 0.0004135563950350782,
+      "loss": 3.4686,
       "step": 28900
     },
     {
-      "epoch": 3.115918630933161,
-      "grad_norm": 0.720397412776947,
-      "learning_rate": 0.00041356750350177775,
-      "loss": 3.4421,
+      "epoch": 3.1212938005390836,
+      "grad_norm": 0.6019054651260376,
+      "learning_rate": 0.0004132325957906098,
+      "loss": 3.4656,
       "step": 28950
     },
     {
-      "epoch": 3.121300182972769,
-      "grad_norm": 0.6323873996734619,
-      "learning_rate": 0.0004132442624717164,
-      "loss": 3.4282,
+      "epoch": 3.1266846361185983,
+      "grad_norm": 0.6578657627105713,
+      "learning_rate": 0.00041290879654614135,
+      "loss": 3.4629,
       "step": 29000
     },
     {
-      "epoch": 3.121300182972769,
-      "eval_accuracy": 0.3708154649132119,
-      "eval_loss": 3.4972314834594727,
-      "eval_runtime": 183.1453,
-      "eval_samples_per_second": 98.343,
-      "eval_steps_per_second": 6.148,
+      "epoch": 3.1266846361185983,
+      "eval_accuracy": 0.3699133199856882,
+      "eval_loss": 3.5049118995666504,
+      "eval_runtime": 152.826,
+      "eval_samples_per_second": 117.853,
+      "eval_steps_per_second": 7.368,
       "step": 29000
     },
     {
-      "epoch": 3.1266817350123777,
-      "grad_norm": 0.6112760305404663,
-      "learning_rate": 0.000412921021441655,
-      "loss": 3.4391,
+      "epoch": 3.1320754716981134,
+      "grad_norm": 0.6159405708312988,
+      "learning_rate": 0.00041258499730167296,
+      "loss": 3.4682,
       "step": 29050
     },
     {
-      "epoch": 3.132063287051986,
-      "grad_norm": 0.6278811693191528,
-      "learning_rate": 0.00041259778041159354,
-      "loss": 3.4484,
+      "epoch": 3.137466307277628,
+      "grad_norm": 0.5930494070053101,
+      "learning_rate": 0.0004122611980572045,
+      "loss": 3.4657,
       "step": 29100
     },
     {
-      "epoch": 3.137444839091594,
-      "grad_norm": 0.6156930327415466,
-      "learning_rate": 0.00041227453938153213,
-      "loss": 3.4411,
+      "epoch": 3.142857142857143,
+      "grad_norm": 0.6473372578620911,
+      "learning_rate": 0.00041193739881273606,
+      "loss": 3.4697,
       "step": 29150
     },
     {
-      "epoch": 3.1428263911312024,
-      "grad_norm": 0.6341788172721863,
-      "learning_rate": 0.00041195129835147067,
-      "loss": 3.4691,
+      "epoch": 3.1482479784366575,
+      "grad_norm": 0.5993855595588684,
+      "learning_rate": 0.00041161359956826766,
+      "loss": 3.4661,
       "step": 29200
     },
     {
-      "epoch": 3.1482079431708105,
-      "grad_norm": 0.5902044177055359,
-      "learning_rate": 0.00041162805732140927,
-      "loss": 3.4541,
+      "epoch": 3.1536388140161726,
+      "grad_norm": 0.6028909683227539,
+      "learning_rate": 0.0004112898003237992,
+      "loss": 3.4822,
       "step": 29250
     },
     {
-      "epoch": 3.1535894952104186,
-      "grad_norm": 0.6705465316772461,
-      "learning_rate": 0.0004113048162913479,
-      "loss": 3.4468,
+      "epoch": 3.1590296495956873,
+      "grad_norm": 0.6628546714782715,
+      "learning_rate": 0.0004109660010793308,
+      "loss": 3.4721,
       "step": 29300
     },
     {
-      "epoch": 3.1589710472500268,
-      "grad_norm": 0.6405426263809204,
-      "learning_rate": 0.00041098157526128646,
-      "loss": 3.4516,
+      "epoch": 3.164420485175202,
+      "grad_norm": 0.5944086909294128,
+      "learning_rate": 0.0004106422018348623,
+      "loss": 3.4643,
       "step": 29350
     },
     {
-      "epoch": 3.1643525992896353,
-      "grad_norm": 0.5979316830635071,
-      "learning_rate": 0.00041065833423122505,
-      "loss": 3.4547,
+      "epoch": 3.169811320754717,
+      "grad_norm": 0.5732219815254211,
+      "learning_rate": 0.0004103184025903939,
+      "loss": 3.4571,
       "step": 29400
     },
     {
-      "epoch": 3.1697341513292434,
-      "grad_norm": 0.6459177136421204,
-      "learning_rate": 0.00041033509320116365,
-      "loss": 3.4378,
+      "epoch": 3.175202156334232,
+      "grad_norm": 0.5670364499092102,
+      "learning_rate": 0.0004100010793308148,
+      "loss": 3.4699,
       "step": 29450
     },
     {
-      "epoch": 3.1751157033688515,
-      "grad_norm": 0.6655340194702148,
-      "learning_rate": 0.0004100118521711022,
-      "loss": 3.4489,
+      "epoch": 3.1805929919137466,
+      "grad_norm": 0.5993989706039429,
+      "learning_rate": 0.0004096772800863464,
+      "loss": 3.4667,
       "step": 29500
     },
     {
-      "epoch": 3.1804972554084596,
-      "grad_norm": 0.5788933634757996,
-      "learning_rate": 0.0004096886111410408,
-      "loss": 3.4622,
+      "epoch": 3.1859838274932613,
+      "grad_norm": 0.6201438903808594,
+      "learning_rate": 0.000409353480841878,
+      "loss": 3.4756,
       "step": 29550
     },
     {
-      "epoch": 3.185878807448068,
-      "grad_norm": 0.6267481446266174,
-      "learning_rate": 0.00040936537011097943,
-      "loss": 3.4591,
+      "epoch": 3.1913746630727764,
+      "grad_norm": 0.6027490496635437,
+      "learning_rate": 0.0004090296815974096,
+      "loss": 3.4726,
       "step": 29600
     },
     {
-      "epoch": 3.1912603594876763,
-      "grad_norm": 0.6072609424591064,
-      "learning_rate": 0.00040904212908091797,
-      "loss": 3.4379,
+      "epoch": 3.196765498652291,
+      "grad_norm": 0.6361322999000549,
+      "learning_rate": 0.00040870588235294113,
+      "loss": 3.4611,
       "step": 29650
     },
     {
-      "epoch": 3.1966419115272844,
-      "grad_norm": 0.6465627551078796,
-      "learning_rate": 0.00040871888805085656,
-      "loss": 3.4535,
+      "epoch": 3.202156334231806,
+      "grad_norm": 0.5988063812255859,
+      "learning_rate": 0.00040838208310847273,
+      "loss": 3.4926,
       "step": 29700
     },
     {
-      "epoch": 3.2020234635668925,
-      "grad_norm": 0.5927116274833679,
-      "learning_rate": 0.0004083956470207951,
-      "loss": 3.4669,
+      "epoch": 3.207547169811321,
+      "grad_norm": 0.6073200702667236,
+      "learning_rate": 0.0004080582838640043,
+      "loss": 3.4618,
       "step": 29750
     },
     {
-      "epoch": 3.207405015606501,
-      "grad_norm": 0.6772713661193848,
-      "learning_rate": 0.0004080724059907337,
-      "loss": 3.4538,
+      "epoch": 3.2129380053908356,
+      "grad_norm": 0.6566458940505981,
+      "learning_rate": 0.0004077344846195359,
+      "loss": 3.4734,
       "step": 29800
     },
     {
-      "epoch": 3.212786567646109,
-      "grad_norm": 0.9021631479263306,
-      "learning_rate": 0.00040774916496067235,
-      "loss": 3.459,
+      "epoch": 3.2183288409703503,
+      "grad_norm": 0.6257891058921814,
+      "learning_rate": 0.00040741068537506744,
+      "loss": 3.448,
       "step": 29850
     },
     {
-      "epoch": 3.2181681196857173,
-      "grad_norm": 0.5905815958976746,
-      "learning_rate": 0.0004074259239306109,
-      "loss": 3.4825,
+      "epoch": 3.223719676549865,
+      "grad_norm": 0.5836365818977356,
+      "learning_rate": 0.000407086886130599,
+      "loss": 3.4603,
       "step": 29900
     },
     {
-      "epoch": 3.2235496717253254,
-      "grad_norm": 0.6410558819770813,
-      "learning_rate": 0.0004071026829005495,
-      "loss": 3.4499,
+      "epoch": 3.22911051212938,
+      "grad_norm": 0.6139360070228577,
+      "learning_rate": 0.0004067630868861306,
+      "loss": 3.4748,
       "step": 29950
     },
     {
-      "epoch": 3.228931223764934,
-      "grad_norm": 0.621482253074646,
-      "learning_rate": 0.0004067794418704881,
-      "loss": 3.4715,
+      "epoch": 3.234501347708895,
+      "grad_norm": 0.5842255353927612,
+      "learning_rate": 0.0004064392876416621,
+      "loss": 3.4768,
       "step": 30000
     },
     {
-      "epoch": 3.228931223764934,
-      "eval_accuracy": 0.3717197828984918,
-      "eval_loss": 3.49033522605896,
-      "eval_runtime": 183.6311,
-      "eval_samples_per_second": 98.082,
-      "eval_steps_per_second": 6.132,
+      "epoch": 3.234501347708895,
+      "eval_accuracy": 0.370657266308554,
+      "eval_loss": 3.497044801712036,
+      "eval_runtime": 153.4383,
+      "eval_samples_per_second": 117.383,
+      "eval_steps_per_second": 7.338,
       "step": 30000
     },
     {
-      "epoch": 3.234312775804542,
-      "grad_norm": 0.6337369680404663,
-      "learning_rate": 0.0004064562008404266,
-      "loss": 3.4548,
+      "epoch": 3.2398921832884096,
+      "grad_norm": 0.6446197628974915,
+      "learning_rate": 0.0004061154883971937,
+      "loss": 3.4815,
       "step": 30050
     },
     {
-      "epoch": 3.23969432784415,
-      "grad_norm": 0.6360739469528198,
-      "learning_rate": 0.0004061329598103652,
-      "loss": 3.4506,
+      "epoch": 3.2452830188679247,
+      "grad_norm": 0.567846953868866,
+      "learning_rate": 0.00040579168915272525,
+      "loss": 3.461,
       "step": 30100
     },
     {
-      "epoch": 3.2450758798837587,
-      "grad_norm": 0.6237755417823792,
-      "learning_rate": 0.00040580971878030386,
-      "loss": 3.4415,
+      "epoch": 3.2506738544474394,
+      "grad_norm": 0.5950538516044617,
+      "learning_rate": 0.00040546788990825685,
+      "loss": 3.4662,
       "step": 30150
     },
     {
-      "epoch": 3.250457431923367,
-      "grad_norm": 0.6507006287574768,
-      "learning_rate": 0.0004054864777502424,
-      "loss": 3.4434,
+      "epoch": 3.256064690026954,
+      "grad_norm": 0.5918688774108887,
+      "learning_rate": 0.0004051440906637884,
+      "loss": 3.4689,
       "step": 30200
     },
     {
-      "epoch": 3.255838983962975,
-      "grad_norm": 0.617294192314148,
-      "learning_rate": 0.000405163236720181,
-      "loss": 3.4477,
+      "epoch": 3.2614555256064692,
+      "grad_norm": 0.5920317769050598,
+      "learning_rate": 0.00040482029141931995,
+      "loss": 3.45,
       "step": 30250
     },
     {
-      "epoch": 3.261220536002583,
-      "grad_norm": 0.5685617923736572,
-      "learning_rate": 0.00040483999569011954,
-      "loss": 3.4413,
+      "epoch": 3.266846361185984,
+      "grad_norm": 0.6167895197868347,
+      "learning_rate": 0.00040449649217485156,
+      "loss": 3.4825,
       "step": 30300
     },
     {
-      "epoch": 3.2666020880421915,
-      "grad_norm": 0.6729376316070557,
-      "learning_rate": 0.00040451675466005813,
-      "loss": 3.4493,
+      "epoch": 3.2722371967654986,
+      "grad_norm": 0.5782353281974792,
+      "learning_rate": 0.0004041726929303831,
+      "loss": 3.4646,
       "step": 30350
     },
     {
-      "epoch": 3.2719836400817996,
-      "grad_norm": 0.6817924380302429,
-      "learning_rate": 0.0004041935136299967,
-      "loss": 3.4626,
+      "epoch": 3.2776280323450133,
+      "grad_norm": 0.6258628964424133,
+      "learning_rate": 0.0004038488936859147,
+      "loss": 3.4673,
       "step": 30400
     },
     {
-      "epoch": 3.2773651921214078,
-      "grad_norm": 0.6592165231704712,
-      "learning_rate": 0.0004038702725999353,
-      "loss": 3.4708,
+      "epoch": 3.2830188679245285,
+      "grad_norm": 0.6296204924583435,
+      "learning_rate": 0.00040352509444144626,
+      "loss": 3.4669,
       "step": 30450
     },
     {
-      "epoch": 3.282746744161016,
-      "grad_norm": 0.6141457557678223,
-      "learning_rate": 0.0004035470315698739,
-      "loss": 3.4626,
+      "epoch": 3.288409703504043,
+      "grad_norm": 0.6166867017745972,
+      "learning_rate": 0.00040320129519697787,
+      "loss": 3.4964,
       "step": 30500
     },
     {
-      "epoch": 3.2881282962006244,
-      "grad_norm": 0.6246821284294128,
-      "learning_rate": 0.0004032237905398125,
-      "loss": 3.4635,
+      "epoch": 3.293800539083558,
+      "grad_norm": 0.6180795431137085,
+      "learning_rate": 0.0004028774959525094,
+      "loss": 3.4671,
       "step": 30550
     },
     {
-      "epoch": 3.2935098482402325,
-      "grad_norm": 0.6646206378936768,
-      "learning_rate": 0.00040290054950975105,
-      "loss": 3.435,
+      "epoch": 3.2991913746630726,
+      "grad_norm": 0.6313896179199219,
+      "learning_rate": 0.000402553696708041,
+      "loss": 3.4733,
       "step": 30600
     },
     {
-      "epoch": 3.2988914002798406,
-      "grad_norm": 0.6625818610191345,
-      "learning_rate": 0.00040257730847968965,
-      "loss": 3.4564,
+      "epoch": 3.3045822102425877,
+      "grad_norm": 0.6259616613388062,
+      "learning_rate": 0.0004022298974635726,
+      "loss": 3.4639,
       "step": 30650
     },
     {
-      "epoch": 3.304272952319449,
-      "grad_norm": 0.6363469362258911,
-      "learning_rate": 0.0004022540674496283,
-      "loss": 3.4357,
+      "epoch": 3.3099730458221024,
+      "grad_norm": 0.5843161940574646,
+      "learning_rate": 0.00040190609821910407,
+      "loss": 3.4654,
       "step": 30700
     },
     {
-      "epoch": 3.3096545043590573,
-      "grad_norm": 0.6316894888877869,
-      "learning_rate": 0.00040193082641956684,
-      "loss": 3.4715,
+      "epoch": 3.315363881401617,
+      "grad_norm": 0.6164865493774414,
+      "learning_rate": 0.0004015822989746357,
+      "loss": 3.4562,
       "step": 30750
     },
     {
-      "epoch": 3.3150360563986654,
-      "grad_norm": 0.6063272953033447,
-      "learning_rate": 0.00040160758538950543,
-      "loss": 3.4526,
+      "epoch": 3.3207547169811322,
+      "grad_norm": 0.6043446063995361,
+      "learning_rate": 0.0004012584997301672,
+      "loss": 3.4698,
       "step": 30800
     },
     {
-      "epoch": 3.3204176084382735,
-      "grad_norm": 0.6325346827507019,
-      "learning_rate": 0.00040128434435944397,
-      "loss": 3.4518,
+      "epoch": 3.326145552560647,
+      "grad_norm": 0.7261855006217957,
+      "learning_rate": 0.00040093470048569883,
+      "loss": 3.4807,
       "step": 30850
     },
     {
-      "epoch": 3.3257991604778816,
-      "grad_norm": 0.5992056727409363,
-      "learning_rate": 0.00040096110332938257,
-      "loss": 3.4489,
+      "epoch": 3.3315363881401616,
+      "grad_norm": 0.6464900374412537,
+      "learning_rate": 0.0004006109012412304,
+      "loss": 3.4702,
       "step": 30900
     },
     {
-      "epoch": 3.33118071251749,
-      "grad_norm": 0.7044916749000549,
-      "learning_rate": 0.00040063786229932116,
-      "loss": 3.4561,
+      "epoch": 3.3369272237196768,
+      "grad_norm": 0.6383877396583557,
+      "learning_rate": 0.000400287101996762,
+      "loss": 3.4766,
       "step": 30950
     },
     {
-      "epoch": 3.3365622645570983,
-      "grad_norm": 0.6150743365287781,
-      "learning_rate": 0.00040031462126925975,
-      "loss": 3.4431,
+      "epoch": 3.3423180592991915,
+      "grad_norm": 0.591076672077179,
+      "learning_rate": 0.00039996330275229354,
+      "loss": 3.465,
       "step": 31000
     },
     {
-      "epoch": 3.3365622645570983,
-      "eval_accuracy": 0.3726238835779961,
-      "eval_loss": 3.4839208126068115,
-      "eval_runtime": 183.3101,
-      "eval_samples_per_second": 98.254,
-      "eval_steps_per_second": 6.143,
+      "epoch": 3.3423180592991915,
+      "eval_accuracy": 0.37180866096072407,
+      "eval_loss": 3.492544174194336,
+      "eval_runtime": 153.1036,
+      "eval_samples_per_second": 117.639,
+      "eval_steps_per_second": 7.354,
       "step": 31000
     },
     {
-      "epoch": 3.3419438165967064,
-      "grad_norm": 0.6841480135917664,
-      "learning_rate": 0.00039999138023919835,
-      "loss": 3.4383,
+      "epoch": 3.347708894878706,
+      "grad_norm": 0.622185230255127,
+      "learning_rate": 0.00039963950350782514,
+      "loss": 3.4748,
       "step": 31050
     },
     {
-      "epoch": 3.347325368636315,
-      "grad_norm": 0.6460050344467163,
-      "learning_rate": 0.00039966813920913694,
-      "loss": 3.445,
+      "epoch": 3.353099730458221,
+      "grad_norm": 0.6088114976882935,
+      "learning_rate": 0.0003993157042633567,
+      "loss": 3.4634,
       "step": 31100
     },
     {
-      "epoch": 3.352706920675923,
-      "grad_norm": 0.6154130697250366,
-      "learning_rate": 0.0003993448981790755,
-      "loss": 3.4549,
+      "epoch": 3.358490566037736,
+      "grad_norm": 0.6120857000350952,
+      "learning_rate": 0.00039899190501888824,
+      "loss": 3.4684,
       "step": 31150
     },
     {
-      "epoch": 3.358088472715531,
-      "grad_norm": 0.6503008604049683,
-      "learning_rate": 0.0003990216571490141,
-      "loss": 3.4634,
+      "epoch": 3.3638814016172507,
+      "grad_norm": 0.5797421932220459,
+      "learning_rate": 0.00039866810577441985,
+      "loss": 3.4749,
       "step": 31200
     },
     {
-      "epoch": 3.3634700247551392,
-      "grad_norm": 0.6205646991729736,
-      "learning_rate": 0.0003986984161189526,
-      "loss": 3.4688,
+      "epoch": 3.3692722371967654,
+      "grad_norm": 0.5956501364707947,
+      "learning_rate": 0.0003983443065299514,
+      "loss": 3.4623,
       "step": 31250
     },
     {
-      "epoch": 3.368851576794748,
-      "grad_norm": 0.6051877737045288,
-      "learning_rate": 0.00039837517508889127,
-      "loss": 3.4712,
+      "epoch": 3.37466307277628,
+      "grad_norm": 0.6451794505119324,
+      "learning_rate": 0.000398020507285483,
+      "loss": 3.4714,
       "step": 31300
     },
     {
-      "epoch": 3.374233128834356,
-      "grad_norm": 0.6055548787117004,
-      "learning_rate": 0.00039805193405882986,
-      "loss": 3.443,
+      "epoch": 3.3800539083557952,
+      "grad_norm": 0.6833682060241699,
+      "learning_rate": 0.0003976967080410145,
+      "loss": 3.4763,
       "step": 31350
     },
     {
-      "epoch": 3.379614680873964,
-      "grad_norm": 0.5936395525932312,
-      "learning_rate": 0.0003977286930287684,
-      "loss": 3.4544,
+      "epoch": 3.38544474393531,
+      "grad_norm": 0.5953586101531982,
+      "learning_rate": 0.0003973729087965461,
+      "loss": 3.4845,
       "step": 31400
     },
     {
-      "epoch": 3.384996232913572,
-      "grad_norm": 0.6387148499488831,
-      "learning_rate": 0.00039741191681930826,
-      "loss": 3.4498,
+      "epoch": 3.3908355795148246,
+      "grad_norm": 0.6166749596595764,
+      "learning_rate": 0.000397055585536967,
+      "loss": 3.4655,
       "step": 31450
     },
     {
-      "epoch": 3.3903777849531807,
-      "grad_norm": 0.6734790802001953,
-      "learning_rate": 0.0003970886757892468,
-      "loss": 3.4803,
+      "epoch": 3.3962264150943398,
+      "grad_norm": 0.6007559895515442,
+      "learning_rate": 0.0003967317862924986,
+      "loss": 3.4639,
       "step": 31500
     },
     {
-      "epoch": 3.3957593369927888,
-      "grad_norm": 0.6716505289077759,
-      "learning_rate": 0.0003967654347591854,
-      "loss": 3.4491,
+      "epoch": 3.4016172506738545,
+      "grad_norm": 0.5914653539657593,
+      "learning_rate": 0.00039640798704803016,
+      "loss": 3.4516,
       "step": 31550
     },
     {
-      "epoch": 3.401140889032397,
-      "grad_norm": 0.6382240653038025,
-      "learning_rate": 0.00039644219372912394,
-      "loss": 3.4813,
+      "epoch": 3.407008086253369,
+      "grad_norm": 0.6352680325508118,
+      "learning_rate": 0.00039608418780356176,
+      "loss": 3.4614,
       "step": 31600
     },
     {
-      "epoch": 3.4065224410720054,
-      "grad_norm": 0.673555314540863,
-      "learning_rate": 0.0003961189526990626,
-      "loss": 3.4517,
+      "epoch": 3.4123989218328843,
+      "grad_norm": 0.6051030158996582,
+      "learning_rate": 0.0003957603885590933,
+      "loss": 3.4784,
       "step": 31650
     },
     {
-      "epoch": 3.4119039931116135,
-      "grad_norm": 0.6426491737365723,
-      "learning_rate": 0.0003957957116690012,
-      "loss": 3.4569,
+      "epoch": 3.417789757412399,
+      "grad_norm": 0.6388689279556274,
+      "learning_rate": 0.0003954365893146249,
+      "loss": 3.463,
       "step": 31700
     },
     {
-      "epoch": 3.4172855451512216,
-      "grad_norm": 0.5995083451271057,
-      "learning_rate": 0.0003954724706389397,
-      "loss": 3.4387,
+      "epoch": 3.4231805929919137,
+      "grad_norm": 0.6489126086235046,
+      "learning_rate": 0.00039511279007015647,
+      "loss": 3.4724,
       "step": 31750
     },
     {
-      "epoch": 3.4226670971908297,
-      "grad_norm": 0.6536892056465149,
-      "learning_rate": 0.0003951492296088783,
-      "loss": 3.4568,
+      "epoch": 3.4285714285714284,
+      "grad_norm": 0.6948974132537842,
+      "learning_rate": 0.00039478899082568807,
+      "loss": 3.4706,
       "step": 31800
     },
     {
-      "epoch": 3.428048649230438,
-      "grad_norm": 0.6076087951660156,
-      "learning_rate": 0.0003948259885788169,
-      "loss": 3.4747,
+      "epoch": 3.4339622641509435,
+      "grad_norm": 0.604185163974762,
+      "learning_rate": 0.0003944651915812196,
+      "loss": 3.4593,
       "step": 31850
     },
     {
-      "epoch": 3.4334302012700464,
-      "grad_norm": 0.830601155757904,
-      "learning_rate": 0.00039450274754875545,
-      "loss": 3.4685,
+      "epoch": 3.439353099730458,
+      "grad_norm": 0.6367453336715698,
+      "learning_rate": 0.0003941413923367512,
+      "loss": 3.461,
       "step": 31900
     },
     {
-      "epoch": 3.4388117533096545,
-      "grad_norm": 0.5986692309379578,
-      "learning_rate": 0.0003941795065186941,
-      "loss": 3.4513,
+      "epoch": 3.444743935309973,
+      "grad_norm": 0.6053892374038696,
+      "learning_rate": 0.0003938175930922828,
+      "loss": 3.4706,
       "step": 31950
     },
     {
-      "epoch": 3.4441933053492626,
-      "grad_norm": 0.6253620386123657,
-      "learning_rate": 0.0003938562654886327,
-      "loss": 3.4488,
+      "epoch": 3.450134770889488,
+      "grad_norm": 0.6216090321540833,
+      "learning_rate": 0.0003934937938478143,
+      "loss": 3.473,
       "step": 32000
     },
     {
-      "epoch": 3.4441933053492626,
-      "eval_accuracy": 0.37277338995162884,
-      "eval_loss": 3.4776968955993652,
-      "eval_runtime": 183.9112,
-      "eval_samples_per_second": 97.933,
-      "eval_steps_per_second": 6.123,
+      "epoch": 3.450134770889488,
+      "eval_accuracy": 0.37205323861119444,
+      "eval_loss": 3.48478627204895,
+      "eval_runtime": 153.8313,
+      "eval_samples_per_second": 117.083,
+      "eval_steps_per_second": 7.32,
       "step": 32000
     },
     {
-      "epoch": 3.449574857388871,
-      "grad_norm": 0.6315608620643616,
-      "learning_rate": 0.00039353302445857124,
-      "loss": 3.4636,
+      "epoch": 3.4555256064690028,
+      "grad_norm": 0.5917905569076538,
+      "learning_rate": 0.0003931699946033459,
+      "loss": 3.4692,
       "step": 32050
     },
     {
-      "epoch": 3.4549564094284793,
-      "grad_norm": 0.6577159762382507,
-      "learning_rate": 0.00039320978342850983,
-      "loss": 3.4492,
+      "epoch": 3.4609164420485174,
+      "grad_norm": 0.6017573475837708,
+      "learning_rate": 0.00039284619535887743,
+      "loss": 3.4692,
       "step": 32100
     },
     {
-      "epoch": 3.4603379614680874,
-      "grad_norm": 0.626023530960083,
-      "learning_rate": 0.00039288654239844837,
-      "loss": 3.4626,
+      "epoch": 3.466307277628032,
+      "grad_norm": 0.5919556617736816,
+      "learning_rate": 0.00039252239611440904,
+      "loss": 3.4683,
       "step": 32150
     },
     {
-      "epoch": 3.4657195135076955,
-      "grad_norm": 0.6065675616264343,
-      "learning_rate": 0.00039256330136838697,
-      "loss": 3.4728,
+      "epoch": 3.4716981132075473,
+      "grad_norm": 0.6309771537780762,
+      "learning_rate": 0.0003921985968699406,
+      "loss": 3.467,
       "step": 32200
     },
     {
-      "epoch": 3.471101065547304,
-      "grad_norm": 0.6406152248382568,
-      "learning_rate": 0.0003922400603383256,
-      "loss": 3.4686,
+      "epoch": 3.477088948787062,
+      "grad_norm": 0.6524568796157837,
+      "learning_rate": 0.0003918747976254722,
+      "loss": 3.4758,
       "step": 32250
     },
     {
-      "epoch": 3.476482617586912,
-      "grad_norm": 0.6500027179718018,
-      "learning_rate": 0.00039191681930826416,
-      "loss": 3.4588,
+      "epoch": 3.4824797843665767,
+      "grad_norm": 0.627716600894928,
+      "learning_rate": 0.00039155099838100374,
+      "loss": 3.4856,
       "step": 32300
     },
     {
-      "epoch": 3.4818641696265202,
-      "grad_norm": 0.6565770506858826,
-      "learning_rate": 0.00039159357827820275,
-      "loss": 3.4696,
+      "epoch": 3.487870619946092,
+      "grad_norm": 0.6386584043502808,
+      "learning_rate": 0.0003912271991365353,
+      "loss": 3.4726,
       "step": 32350
     },
     {
-      "epoch": 3.4872457216661283,
-      "grad_norm": 0.6239467263221741,
-      "learning_rate": 0.00039127033724814135,
-      "loss": 3.463,
+      "epoch": 3.4932614555256065,
+      "grad_norm": 0.6196281909942627,
+      "learning_rate": 0.0003909033998920669,
+      "loss": 3.4715,
       "step": 32400
     },
     {
-      "epoch": 3.492627273705737,
-      "grad_norm": 0.617245078086853,
-      "learning_rate": 0.0003909470962180799,
-      "loss": 3.4473,
+      "epoch": 3.498652291105121,
+      "grad_norm": 0.6428321599960327,
+      "learning_rate": 0.00039057960064759845,
+      "loss": 3.4555,
       "step": 32450
     },
     {
-      "epoch": 3.498008825745345,
-      "grad_norm": 0.6251423954963684,
-      "learning_rate": 0.00039062385518801854,
-      "loss": 3.4511,
+      "epoch": 3.5040431266846364,
+      "grad_norm": 0.624134361743927,
+      "learning_rate": 0.00039025580140313005,
+      "loss": 3.479,
       "step": 32500
     },
     {
-      "epoch": 3.503390377784953,
-      "grad_norm": 0.6421258449554443,
-      "learning_rate": 0.00039030061415795713,
-      "loss": 3.471,
+      "epoch": 3.509433962264151,
+      "grad_norm": 0.6787989735603333,
+      "learning_rate": 0.0003899320021586616,
+      "loss": 3.455,
       "step": 32550
     },
     {
-      "epoch": 3.5087719298245617,
-      "grad_norm": 0.6087847352027893,
-      "learning_rate": 0.00038997737312789567,
-      "loss": 3.4537,
+      "epoch": 3.5148247978436657,
+      "grad_norm": 0.6143307089805603,
+      "learning_rate": 0.0003896082029141932,
+      "loss": 3.4767,
       "step": 32600
     },
     {
-      "epoch": 3.5141534818641698,
-      "grad_norm": 0.6602423191070557,
-      "learning_rate": 0.00038965413209783426,
-      "loss": 3.4552,
+      "epoch": 3.5202156334231804,
+      "grad_norm": 0.6654953956604004,
+      "learning_rate": 0.00038928440366972476,
+      "loss": 3.4612,
       "step": 32650
     },
     {
-      "epoch": 3.519535033903778,
-      "grad_norm": 0.6009964346885681,
-      "learning_rate": 0.0003893308910677728,
-      "loss": 3.4633,
+      "epoch": 3.525606469002695,
+      "grad_norm": 0.5875223278999329,
+      "learning_rate": 0.00038896060442525636,
+      "loss": 3.4585,
       "step": 32700
     },
     {
-      "epoch": 3.524916585943386,
-      "grad_norm": 0.6000418066978455,
-      "learning_rate": 0.0003890076500377114,
-      "loss": 3.4547,
+      "epoch": 3.5309973045822103,
+      "grad_norm": 0.6045464277267456,
+      "learning_rate": 0.00038863680518078786,
+      "loss": 3.4686,
       "step": 32750
     },
     {
-      "epoch": 3.530298137982994,
-      "grad_norm": 0.6483384966850281,
-      "learning_rate": 0.00038868440900765005,
-      "loss": 3.4461,
+      "epoch": 3.536388140161725,
+      "grad_norm": 0.627065896987915,
+      "learning_rate": 0.0003883130059363194,
+      "loss": 3.4725,
       "step": 32800
     },
     {
-      "epoch": 3.5356796900226026,
-      "grad_norm": 0.6097680926322937,
-      "learning_rate": 0.0003883611679775886,
-      "loss": 3.4698,
+      "epoch": 3.5417789757412397,
+      "grad_norm": 0.6074482202529907,
+      "learning_rate": 0.000387989206691851,
+      "loss": 3.46,
       "step": 32850
     },
     {
-      "epoch": 3.5410612420622107,
-      "grad_norm": 0.6209619641304016,
-      "learning_rate": 0.0003880379269475272,
-      "loss": 3.44,
+      "epoch": 3.547169811320755,
+      "grad_norm": 0.5814282298088074,
+      "learning_rate": 0.00038766540744738256,
+      "loss": 3.4597,
       "step": 32900
     },
     {
-      "epoch": 3.546442794101819,
-      "grad_norm": 0.5964656472206116,
-      "learning_rate": 0.0003877146859174657,
-      "loss": 3.4756,
+      "epoch": 3.5525606469002695,
+      "grad_norm": 0.5899258255958557,
+      "learning_rate": 0.00038734160820291417,
+      "loss": 3.4831,
       "step": 32950
     },
     {
-      "epoch": 3.5518243461414274,
-      "grad_norm": 0.64708411693573,
-      "learning_rate": 0.0003873914448874043,
-      "loss": 3.4677,
+      "epoch": 3.557951482479784,
+      "grad_norm": 0.648349940776825,
+      "learning_rate": 0.0003870178089584457,
+      "loss": 3.4612,
       "step": 33000
     },
     {
-      "epoch": 3.5518243461414274,
-      "eval_accuracy": 0.37304708657603347,
-      "eval_loss": 3.472172737121582,
-      "eval_runtime": 183.9035,
-      "eval_samples_per_second": 97.937,
-      "eval_steps_per_second": 6.123,
+      "epoch": 3.557951482479784,
+      "eval_accuracy": 0.3726940733435242,
+      "eval_loss": 3.4788687229156494,
+      "eval_runtime": 153.1782,
+      "eval_samples_per_second": 117.582,
+      "eval_steps_per_second": 7.351,
       "step": 33000
     },
     {
-      "epoch": 3.5572058981810355,
-      "grad_norm": 0.6064699292182922,
-      "learning_rate": 0.0003870682038573429,
-      "loss": 3.4523,
+      "epoch": 3.5633423180592994,
+      "grad_norm": 0.614753246307373,
+      "learning_rate": 0.0003866940097139773,
+      "loss": 3.4633,
       "step": 33050
     },
     {
-      "epoch": 3.5625874502206436,
-      "grad_norm": 0.6033612489700317,
-      "learning_rate": 0.0003867449628272815,
-      "loss": 3.4576,
+      "epoch": 3.568733153638814,
+      "grad_norm": 0.6309217214584351,
+      "learning_rate": 0.0003863702104695089,
+      "loss": 3.4844,
       "step": 33100
     },
     {
-      "epoch": 3.5679690022602517,
-      "grad_norm": 0.6540218591690063,
-      "learning_rate": 0.0003864217217972201,
-      "loss": 3.4472,
+      "epoch": 3.5741239892183287,
+      "grad_norm": 0.6211805939674377,
+      "learning_rate": 0.0003860464112250404,
+      "loss": 3.4601,
       "step": 33150
     },
     {
-      "epoch": 3.57335055429986,
-      "grad_norm": 0.7801140546798706,
-      "learning_rate": 0.0003860984807671587,
-      "loss": 3.4502,
+      "epoch": 3.579514824797844,
+      "grad_norm": 0.6448453664779663,
+      "learning_rate": 0.00038572261198057203,
+      "loss": 3.4795,
       "step": 33200
     },
     {
-      "epoch": 3.5787321063394684,
-      "grad_norm": 0.6421427726745605,
-      "learning_rate": 0.00038577523973709724,
-      "loss": 3.4756,
+      "epoch": 3.5849056603773586,
+      "grad_norm": 0.5850814580917358,
+      "learning_rate": 0.0003853988127361036,
+      "loss": 3.4752,
       "step": 33250
     },
     {
-      "epoch": 3.5841136583790765,
-      "grad_norm": 0.6703349351882935,
-      "learning_rate": 0.00038545199870703583,
-      "loss": 3.4665,
+      "epoch": 3.5902964959568733,
+      "grad_norm": 0.7022829055786133,
+      "learning_rate": 0.0003850750134916352,
+      "loss": 3.4982,
       "step": 33300
     },
     {
-      "epoch": 3.5894952104186846,
-      "grad_norm": 0.6108054518699646,
-      "learning_rate": 0.0003851287576769744,
-      "loss": 3.4745,
+      "epoch": 3.595687331536388,
+      "grad_norm": 0.6456696391105652,
+      "learning_rate": 0.0003847512142471667,
+      "loss": 3.4641,
       "step": 33350
     },
     {
-      "epoch": 3.594876762458293,
-      "grad_norm": 0.6100577712059021,
-      "learning_rate": 0.000384805516646913,
-      "loss": 3.4816,
+      "epoch": 3.601078167115903,
+      "grad_norm": 0.6414846181869507,
+      "learning_rate": 0.0003844274150026983,
+      "loss": 3.4752,
       "step": 33400
     },
     {
-      "epoch": 3.6002583144979012,
-      "grad_norm": 0.6833879947662354,
-      "learning_rate": 0.0003844822756168516,
-      "loss": 3.443,
+      "epoch": 3.606469002695418,
+      "grad_norm": 0.6075315475463867,
+      "learning_rate": 0.00038410361575822984,
+      "loss": 3.4849,
       "step": 33450
     },
     {
-      "epoch": 3.6056398665375093,
-      "grad_norm": 0.6495818495750427,
-      "learning_rate": 0.00038415903458679016,
-      "loss": 3.4332,
+      "epoch": 3.6118598382749325,
+      "grad_norm": 0.6397665739059448,
+      "learning_rate": 0.00038377981651376144,
+      "loss": 3.4782,
       "step": 33500
     },
     {
-      "epoch": 3.611021418577118,
-      "grad_norm": 0.6183668971061707,
-      "learning_rate": 0.00038383579355672875,
-      "loss": 3.4451,
+      "epoch": 3.617250673854447,
+      "grad_norm": 0.6640121936798096,
+      "learning_rate": 0.00038346249325418234,
+      "loss": 3.466,
       "step": 33550
     },
     {
-      "epoch": 3.616402970616726,
-      "grad_norm": 0.697557806968689,
-      "learning_rate": 0.00038351255252666735,
-      "loss": 3.4699,
+      "epoch": 3.6226415094339623,
+      "grad_norm": 0.6269997954368591,
+      "learning_rate": 0.00038313869400971395,
+      "loss": 3.4645,
       "step": 33600
     },
     {
-      "epoch": 3.621784522656334,
-      "grad_norm": 0.6472216248512268,
-      "learning_rate": 0.00038318931149660594,
-      "loss": 3.4375,
+      "epoch": 3.628032345013477,
+      "grad_norm": 0.6265262961387634,
+      "learning_rate": 0.0003828148947652455,
+      "loss": 3.453,
       "step": 33650
     },
     {
-      "epoch": 3.627166074695942,
-      "grad_norm": 0.6490259170532227,
-      "learning_rate": 0.00038286607046654454,
-      "loss": 3.4793,
+      "epoch": 3.6334231805929917,
+      "grad_norm": 0.6128087043762207,
+      "learning_rate": 0.0003824910955207771,
+      "loss": 3.4544,
       "step": 33700
     },
     {
-      "epoch": 3.6325476267355503,
-      "grad_norm": 0.6329680681228638,
-      "learning_rate": 0.00038254282943648313,
-      "loss": 3.457,
+      "epoch": 3.638814016172507,
+      "grad_norm": 0.6193901300430298,
+      "learning_rate": 0.00038216729627630865,
+      "loss": 3.474,
       "step": 33750
     },
     {
-      "epoch": 3.637929178775159,
-      "grad_norm": 0.6038073301315308,
-      "learning_rate": 0.00038221958840642167,
-      "loss": 3.4531,
+      "epoch": 3.6442048517520216,
+      "grad_norm": 0.6126763224601746,
+      "learning_rate": 0.00038184349703184026,
+      "loss": 3.4554,
       "step": 33800
     },
     {
-      "epoch": 3.643310730814767,
-      "grad_norm": 0.6116501688957214,
-      "learning_rate": 0.00038189634737636027,
-      "loss": 3.4593,
+      "epoch": 3.6495956873315363,
+      "grad_norm": 0.6538126468658447,
+      "learning_rate": 0.0003815196977873718,
+      "loss": 3.4665,
       "step": 33850
     },
     {
-      "epoch": 3.648692282854375,
-      "grad_norm": 0.6598987579345703,
-      "learning_rate": 0.0003815731063462988,
-      "loss": 3.4461,
+      "epoch": 3.6549865229110514,
+      "grad_norm": 0.6257369518280029,
+      "learning_rate": 0.00038120237452779276,
+      "loss": 3.4673,
       "step": 33900
     },
     {
-      "epoch": 3.6540738348939836,
-      "grad_norm": 0.6475114822387695,
-      "learning_rate": 0.00038124986531623745,
-      "loss": 3.4672,
+      "epoch": 3.660377358490566,
+      "grad_norm": 0.661566972732544,
+      "learning_rate": 0.0003808785752833243,
+      "loss": 3.4626,
       "step": 33950
     },
     {
-      "epoch": 3.6594553869335917,
-      "grad_norm": 0.5857601761817932,
-      "learning_rate": 0.00038092662428617605,
-      "loss": 3.4431,
+      "epoch": 3.665768194070081,
+      "grad_norm": 0.6305378675460815,
+      "learning_rate": 0.0003805547760388559,
+      "loss": 3.4835,
       "step": 34000
     },
     {
-      "epoch": 3.6594553869335917,
-      "eval_accuracy": 0.37437558543534116,
-      "eval_loss": 3.4629297256469727,
-      "eval_runtime": 183.5567,
-      "eval_samples_per_second": 98.122,
-      "eval_steps_per_second": 6.134,
+      "epoch": 3.665768194070081,
+      "eval_accuracy": 0.3734493195667227,
+      "eval_loss": 3.472283363342285,
+      "eval_runtime": 153.3014,
+      "eval_samples_per_second": 117.488,
+      "eval_steps_per_second": 7.345,
       "step": 34000
     },
     {
-      "epoch": 3.6648369389732,
-      "grad_norm": 0.6019673943519592,
-      "learning_rate": 0.0003806033832561146,
-      "loss": 3.475,
+      "epoch": 3.671159029649596,
+      "grad_norm": 0.6474854350090027,
+      "learning_rate": 0.0003802309767943874,
+      "loss": 3.4627,
       "step": 34050
     },
     {
-      "epoch": 3.670218491012808,
-      "grad_norm": 0.6328139901161194,
-      "learning_rate": 0.0003802801422260532,
-      "loss": 3.4465,
+      "epoch": 3.6765498652291106,
+      "grad_norm": 0.6843244433403015,
+      "learning_rate": 0.000379907177549919,
+      "loss": 3.4723,
       "step": 34100
     },
     {
-      "epoch": 3.675600043052416,
-      "grad_norm": 0.6695095896720886,
-      "learning_rate": 0.0003799569011959918,
-      "loss": 3.4646,
+      "epoch": 3.6819407008086253,
+      "grad_norm": 0.6456762552261353,
+      "learning_rate": 0.00037958337830545057,
+      "loss": 3.4786,
       "step": 34150
     },
     {
-      "epoch": 3.6809815950920246,
-      "grad_norm": 0.658932626247406,
-      "learning_rate": 0.0003796336601659303,
-      "loss": 3.4557,
+      "epoch": 3.68733153638814,
+      "grad_norm": 0.6572555899620056,
+      "learning_rate": 0.00037925957906098217,
+      "loss": 3.4764,
       "step": 34200
     },
     {
-      "epoch": 3.6863631471316327,
-      "grad_norm": 0.6852182149887085,
-      "learning_rate": 0.00037931041913586897,
-      "loss": 3.4613,
+      "epoch": 3.6927223719676547,
+      "grad_norm": 0.6187539100646973,
+      "learning_rate": 0.0003789357798165137,
+      "loss": 3.4703,
       "step": 34250
     },
     {
-      "epoch": 3.691744699171241,
-      "grad_norm": 0.6308695077896118,
-      "learning_rate": 0.00037898717810580756,
-      "loss": 3.4739,
+      "epoch": 3.69811320754717,
+      "grad_norm": 0.6492563486099243,
+      "learning_rate": 0.00037861198057204527,
+      "loss": 3.4791,
       "step": 34300
     },
     {
-      "epoch": 3.6971262512108494,
-      "grad_norm": 0.6065436601638794,
-      "learning_rate": 0.0003786639370757461,
-      "loss": 3.4611,
+      "epoch": 3.7035040431266846,
+      "grad_norm": 0.7034260034561157,
+      "learning_rate": 0.0003782881813275769,
+      "loss": 3.4888,
       "step": 34350
     },
     {
-      "epoch": 3.7025078032504575,
-      "grad_norm": 0.6547542214393616,
-      "learning_rate": 0.0003783406960456847,
-      "loss": 3.4448,
+      "epoch": 3.7088948787061993,
+      "grad_norm": 0.6355736255645752,
+      "learning_rate": 0.0003779643820831084,
+      "loss": 3.4612,
       "step": 34400
     },
     {
-      "epoch": 3.7078893552900656,
-      "grad_norm": 0.5994100570678711,
-      "learning_rate": 0.00037801745501562324,
-      "loss": 3.4588,
+      "epoch": 3.7142857142857144,
+      "grad_norm": 0.636001467704773,
+      "learning_rate": 0.00037764058283864003,
+      "loss": 3.4468,
       "step": 34450
     },
     {
-      "epoch": 3.713270907329674,
-      "grad_norm": 0.6268024444580078,
-      "learning_rate": 0.0003776942139855619,
-      "loss": 3.447,
+      "epoch": 3.719676549865229,
+      "grad_norm": 0.6464974880218506,
+      "learning_rate": 0.0003773167835941716,
+      "loss": 3.465,
       "step": 34500
     },
     {
-      "epoch": 3.7186524593692822,
-      "grad_norm": 0.606737494468689,
-      "learning_rate": 0.0003773709729555005,
-      "loss": 3.4744,
+      "epoch": 3.725067385444744,
+      "grad_norm": 0.6081432700157166,
+      "learning_rate": 0.0003769929843497032,
+      "loss": 3.4829,
       "step": 34550
     },
     {
-      "epoch": 3.7240340114088903,
-      "grad_norm": 0.62318354845047,
-      "learning_rate": 0.000377047731925439,
-      "loss": 3.4464,
+      "epoch": 3.730458221024259,
+      "grad_norm": 0.6616902947425842,
+      "learning_rate": 0.00037666918510523474,
+      "loss": 3.4744,
       "step": 34600
     },
     {
-      "epoch": 3.7294155634484984,
-      "grad_norm": 0.6048056483268738,
-      "learning_rate": 0.0003767244908953776,
-      "loss": 3.4394,
+      "epoch": 3.7358490566037736,
+      "grad_norm": 0.6474754214286804,
+      "learning_rate": 0.00037634538586076634,
+      "loss": 3.4509,
       "step": 34650
     },
     {
-      "epoch": 3.7347971154881066,
-      "grad_norm": 0.6488344669342041,
-      "learning_rate": 0.0003764012498653162,
-      "loss": 3.4665,
+      "epoch": 3.7412398921832883,
+      "grad_norm": 0.6622031927108765,
+      "learning_rate": 0.00037602158661629784,
+      "loss": 3.4899,
       "step": 34700
     },
     {
-      "epoch": 3.740178667527715,
-      "grad_norm": 0.6304295063018799,
-      "learning_rate": 0.00037607800883525475,
-      "loss": 3.4658,
+      "epoch": 3.7466307277628035,
+      "grad_norm": 0.6893497705459595,
+      "learning_rate": 0.0003756977873718294,
+      "loss": 3.4528,
       "step": 34750
     },
     {
-      "epoch": 3.745560219567323,
-      "grad_norm": 0.6360478401184082,
-      "learning_rate": 0.0003757547678051934,
-      "loss": 3.4437,
+      "epoch": 3.752021563342318,
+      "grad_norm": 0.659303605556488,
+      "learning_rate": 0.000375373988127361,
+      "loss": 3.4855,
       "step": 34800
     },
     {
-      "epoch": 3.7509417716069313,
-      "grad_norm": 0.6223590970039368,
-      "learning_rate": 0.000375431526775132,
-      "loss": 3.4584,
+      "epoch": 3.757412398921833,
+      "grad_norm": 0.6131672859191895,
+      "learning_rate": 0.00037505018888289254,
+      "loss": 3.498,
       "step": 34850
     },
     {
-      "epoch": 3.75632332364654,
-      "grad_norm": 0.6473628878593445,
-      "learning_rate": 0.00037510828574507054,
-      "loss": 3.45,
+      "epoch": 3.7628032345013476,
+      "grad_norm": 0.6271697878837585,
+      "learning_rate": 0.00037472638963842415,
+      "loss": 3.4823,
       "step": 34900
     },
     {
-      "epoch": 3.761704875686148,
-      "grad_norm": 0.6152955293655396,
-      "learning_rate": 0.00037478504471500913,
-      "loss": 3.46,
+      "epoch": 3.7681940700808623,
+      "grad_norm": 0.5825786590576172,
+      "learning_rate": 0.0003744025903939557,
+      "loss": 3.4611,
       "step": 34950
     },
     {
-      "epoch": 3.767086427725756,
-      "grad_norm": 0.6649402976036072,
-      "learning_rate": 0.00037446180368494767,
-      "loss": 3.4643,
+      "epoch": 3.7735849056603774,
+      "grad_norm": 0.6334702968597412,
+      "learning_rate": 0.0003740787911494873,
+      "loss": 3.4624,
       "step": 35000
     },
     {
-      "epoch": 3.767086427725756,
-      "eval_accuracy": 0.37429007561263117,
-      "eval_loss": 3.458296775817871,
-      "eval_runtime": 183.8021,
-      "eval_samples_per_second": 97.991,
-      "eval_steps_per_second": 6.126,
+      "epoch": 3.7735849056603774,
+      "eval_accuracy": 0.37404745371414144,
+      "eval_loss": 3.4667880535125732,
+      "eval_runtime": 153.676,
+      "eval_samples_per_second": 117.201,
+      "eval_steps_per_second": 7.327,
       "step": 35000
     },
     {
-      "epoch": 3.772467979765364,
-      "grad_norm": 0.6364124417304993,
-      "learning_rate": 0.00037413856265488627,
-      "loss": 3.4658,
+      "epoch": 3.778975741239892,
+      "grad_norm": 0.6846307516098022,
+      "learning_rate": 0.00037375499190501885,
+      "loss": 3.4849,
       "step": 35050
     },
     {
-      "epoch": 3.7778495318049723,
-      "grad_norm": 0.6472469568252563,
-      "learning_rate": 0.0003738153216248249,
-      "loss": 3.4518,
+      "epoch": 3.784366576819407,
+      "grad_norm": 0.6677894592285156,
+      "learning_rate": 0.0003734311926605504,
+      "loss": 3.45,
       "step": 35100
     },
     {
-      "epoch": 3.783231083844581,
-      "grad_norm": 0.6233261823654175,
-      "learning_rate": 0.00037349208059476346,
-      "loss": 3.4637,
+      "epoch": 3.789757412398922,
+      "grad_norm": 0.6437786817550659,
+      "learning_rate": 0.000373107393416082,
+      "loss": 3.4753,
       "step": 35150
     },
     {
-      "epoch": 3.788612635884189,
-      "grad_norm": 0.6633452773094177,
-      "learning_rate": 0.00037316883956470205,
-      "loss": 3.4705,
+      "epoch": 3.7951482479784366,
+      "grad_norm": 0.6343787312507629,
+      "learning_rate": 0.00037278359417161356,
+      "loss": 3.4557,
       "step": 35200
     },
     {
-      "epoch": 3.793994187923797,
-      "grad_norm": 0.6813004016876221,
-      "learning_rate": 0.00037284559853464064,
-      "loss": 3.4368,
+      "epoch": 3.8005390835579513,
+      "grad_norm": 0.6103297472000122,
+      "learning_rate": 0.00037245979492714517,
+      "loss": 3.4806,
       "step": 35250
     },
     {
-      "epoch": 3.7993757399634056,
-      "grad_norm": 0.6277920603752136,
-      "learning_rate": 0.0003725223575045792,
-      "loss": 3.4613,
+      "epoch": 3.8059299191374665,
+      "grad_norm": 0.6892771124839783,
+      "learning_rate": 0.0003721359956826767,
+      "loss": 3.4708,
       "step": 35300
     },
     {
-      "epoch": 3.8047572920030137,
-      "grad_norm": 0.630336344242096,
-      "learning_rate": 0.0003721991164745178,
-      "loss": 3.4495,
+      "epoch": 3.811320754716981,
+      "grad_norm": 0.6053640246391296,
+      "learning_rate": 0.0003718121964382083,
+      "loss": 3.4722,
       "step": 35350
     },
     {
-      "epoch": 3.810138844042622,
-      "grad_norm": 0.6519911885261536,
-      "learning_rate": 0.00037187587544445643,
-      "loss": 3.4691,
+      "epoch": 3.816711590296496,
+      "grad_norm": 0.696236252784729,
+      "learning_rate": 0.0003714883971937398,
+      "loss": 3.4902,
       "step": 35400
     },
     {
-      "epoch": 3.8155203960822304,
-      "grad_norm": 0.6086752414703369,
-      "learning_rate": 0.00037155909923499624,
-      "loss": 3.4539,
+      "epoch": 3.822102425876011,
+      "grad_norm": 0.6670373678207397,
+      "learning_rate": 0.0003711645979492714,
+      "loss": 3.4476,
       "step": 35450
     },
     {
-      "epoch": 3.8209019481218385,
-      "grad_norm": 0.6103160381317139,
-      "learning_rate": 0.0003712358582049348,
-      "loss": 3.469,
+      "epoch": 3.8274932614555257,
+      "grad_norm": 0.592532217502594,
+      "learning_rate": 0.00037084079870480297,
+      "loss": 3.4821,
       "step": 35500
     },
     {
-      "epoch": 3.8262835001614466,
-      "grad_norm": 0.6309521794319153,
-      "learning_rate": 0.00037091261717487337,
-      "loss": 3.4514,
+      "epoch": 3.8328840970350404,
+      "grad_norm": 0.6642966866493225,
+      "learning_rate": 0.0003705169994603345,
+      "loss": 3.4645,
       "step": 35550
     },
     {
-      "epoch": 3.8316650522010547,
-      "grad_norm": 0.662923276424408,
-      "learning_rate": 0.00037058937614481197,
-      "loss": 3.4601,
+      "epoch": 3.838274932614555,
+      "grad_norm": 0.616497278213501,
+      "learning_rate": 0.00037019320021586613,
+      "loss": 3.4545,
       "step": 35600
     },
     {
-      "epoch": 3.837046604240663,
-      "grad_norm": 0.6447102427482605,
-      "learning_rate": 0.0003702661351147505,
-      "loss": 3.4478,
+      "epoch": 3.8436657681940702,
+      "grad_norm": 0.5879899263381958,
+      "learning_rate": 0.0003698694009713977,
+      "loss": 3.4667,
       "step": 35650
     },
     {
-      "epoch": 3.8424281562802713,
-      "grad_norm": 0.6346889138221741,
-      "learning_rate": 0.0003699428940846891,
-      "loss": 3.4508,
+      "epoch": 3.849056603773585,
+      "grad_norm": 0.6230347752571106,
+      "learning_rate": 0.0003695456017269293,
+      "loss": 3.4643,
       "step": 35700
     },
     {
-      "epoch": 3.8478097083198795,
-      "grad_norm": 0.6585954427719116,
-      "learning_rate": 0.00036961965305462775,
-      "loss": 3.4388,
+      "epoch": 3.8544474393530996,
+      "grad_norm": 0.6210671067237854,
+      "learning_rate": 0.00036922180248246083,
+      "loss": 3.4449,
       "step": 35750
     },
     {
-      "epoch": 3.8531912603594876,
-      "grad_norm": 0.6209655404090881,
-      "learning_rate": 0.0003692964120245663,
-      "loss": 3.4514,
+      "epoch": 3.8598382749326143,
+      "grad_norm": 0.6543208360671997,
+      "learning_rate": 0.00036889800323799244,
+      "loss": 3.4654,
       "step": 35800
     },
     {
-      "epoch": 3.858572812399096,
-      "grad_norm": 0.6361556053161621,
-      "learning_rate": 0.0003689731709945049,
-      "loss": 3.4455,
+      "epoch": 3.8652291105121295,
+      "grad_norm": 0.6200073957443237,
+      "learning_rate": 0.000368574203993524,
+      "loss": 3.4697,
       "step": 35850
     },
     {
-      "epoch": 3.863954364438704,
-      "grad_norm": 0.6221081018447876,
-      "learning_rate": 0.0003686499299644434,
-      "loss": 3.4473,
+      "epoch": 3.870619946091644,
+      "grad_norm": 0.6162194013595581,
+      "learning_rate": 0.0003682504047490556,
+      "loss": 3.4612,
       "step": 35900
     },
     {
-      "epoch": 3.8693359164783123,
-      "grad_norm": 0.6172398328781128,
-      "learning_rate": 0.000368326688934382,
-      "loss": 3.4484,
+      "epoch": 3.876010781671159,
+      "grad_norm": 0.5803920030593872,
+      "learning_rate": 0.00036792660550458714,
+      "loss": 3.4654,
       "step": 35950
     },
     {
-      "epoch": 3.8747174685179204,
-      "grad_norm": 0.6353816390037537,
-      "learning_rate": 0.0003680034479043206,
-      "loss": 3.4524,
+      "epoch": 3.881401617250674,
+      "grad_norm": 0.6236528754234314,
+      "learning_rate": 0.00036760280626011864,
+      "loss": 3.4481,
       "step": 36000
     },
     {
-      "epoch": 3.8747174685179204,
-      "eval_accuracy": 0.3758289264627476,
-      "eval_loss": 3.450455904006958,
-      "eval_runtime": 183.7947,
-      "eval_samples_per_second": 97.995,
-      "eval_steps_per_second": 6.126,
+      "epoch": 3.881401617250674,
+      "eval_accuracy": 0.37474978598097425,
+      "eval_loss": 3.459172010421753,
+      "eval_runtime": 153.0192,
+      "eval_samples_per_second": 117.704,
+      "eval_steps_per_second": 7.359,
       "step": 36000
     },
     {
-      "epoch": 3.8800990205575285,
-      "grad_norm": 0.658366858959198,
-      "learning_rate": 0.0003676802068742592,
-      "loss": 3.4512,
+      "epoch": 3.8867924528301887,
+      "grad_norm": 0.6089425683021545,
+      "learning_rate": 0.0003672790070156503,
+      "loss": 3.4504,
       "step": 36050
     },
     {
-      "epoch": 3.885480572597137,
-      "grad_norm": 0.6353536248207092,
-      "learning_rate": 0.0003673569658441978,
-      "loss": 3.4343,
+      "epoch": 3.8921832884097034,
+      "grad_norm": 0.656832754611969,
+      "learning_rate": 0.0003669552077711818,
+      "loss": 3.4747,
       "step": 36100
     },
     {
-      "epoch": 3.890862124636745,
-      "grad_norm": 0.6400610208511353,
-      "learning_rate": 0.0003670337248141364,
-      "loss": 3.4476,
+      "epoch": 3.8975741239892185,
+      "grad_norm": 0.659642219543457,
+      "learning_rate": 0.0003666314085267134,
+      "loss": 3.4648,
       "step": 36150
     },
     {
-      "epoch": 3.8962436766763533,
-      "grad_norm": 0.6574820876121521,
-      "learning_rate": 0.00036671048378407494,
-      "loss": 3.4501,
+      "epoch": 3.9029649595687332,
+      "grad_norm": 0.5899969935417175,
+      "learning_rate": 0.00036630760928224495,
+      "loss": 3.4442,
       "step": 36200
     },
     {
-      "epoch": 3.901625228715962,
-      "grad_norm": 0.6868521571159363,
-      "learning_rate": 0.00036638724275401353,
-      "loss": 3.4451,
+      "epoch": 3.908355795148248,
+      "grad_norm": 0.6331989765167236,
+      "learning_rate": 0.00036598381003777656,
+      "loss": 3.4502,
       "step": 36250
     },
     {
-      "epoch": 3.90700678075557,
-      "grad_norm": 0.6565641164779663,
-      "learning_rate": 0.00036607046654455334,
-      "loss": 3.4302,
+      "epoch": 3.913746630727763,
+      "grad_norm": 0.643619179725647,
+      "learning_rate": 0.0003656600107933081,
+      "loss": 3.4626,
       "step": 36300
     },
     {
-      "epoch": 3.912388332795178,
-      "grad_norm": 0.6692764163017273,
-      "learning_rate": 0.00036574722551449193,
-      "loss": 3.4704,
+      "epoch": 3.9191374663072778,
+      "grad_norm": 0.6313002705574036,
+      "learning_rate": 0.0003653362115488397,
+      "loss": 3.4566,
       "step": 36350
     },
     {
-      "epoch": 3.9177698848347866,
-      "grad_norm": 0.6343313455581665,
-      "learning_rate": 0.00036542398448443053,
-      "loss": 3.4322,
+      "epoch": 3.9245283018867925,
+      "grad_norm": 0.6360496282577515,
+      "learning_rate": 0.00036501241230437126,
+      "loss": 3.4548,
       "step": 36400
     },
     {
-      "epoch": 3.9231514368743947,
-      "grad_norm": 0.5992373824119568,
-      "learning_rate": 0.0003651007434543691,
-      "loss": 3.4494,
+      "epoch": 3.929919137466307,
+      "grad_norm": 0.6425337195396423,
+      "learning_rate": 0.0003646886130599028,
+      "loss": 3.4638,
       "step": 36450
     },
     {
-      "epoch": 3.928532988914003,
-      "grad_norm": 0.601618230342865,
-      "learning_rate": 0.0003647775024243077,
-      "loss": 3.4482,
+      "epoch": 3.935309973045822,
+      "grad_norm": 0.6314306855201721,
+      "learning_rate": 0.0003643648138154344,
+      "loss": 3.4694,
       "step": 36500
     },
     {
-      "epoch": 3.933914540953611,
-      "grad_norm": 0.7365906238555908,
-      "learning_rate": 0.00036445426139424626,
-      "loss": 3.4362,
+      "epoch": 3.940700808625337,
+      "grad_norm": 0.6240665316581726,
+      "learning_rate": 0.00036404101457096597,
+      "loss": 3.4566,
       "step": 36550
     },
     {
-      "epoch": 3.939296092993219,
-      "grad_norm": 0.6756502389907837,
-      "learning_rate": 0.00036413102036418485,
-      "loss": 3.4433,
+      "epoch": 3.9460916442048517,
+      "grad_norm": 0.6572285294532776,
+      "learning_rate": 0.00036371721532649757,
+      "loss": 3.4689,
       "step": 36600
     },
     {
-      "epoch": 3.9446776450328276,
-      "grad_norm": 0.6230269074440002,
-      "learning_rate": 0.0003638077793341234,
-      "loss": 3.4581,
+      "epoch": 3.9514824797843664,
+      "grad_norm": 0.6254251003265381,
+      "learning_rate": 0.0003633934160820291,
+      "loss": 3.463,
       "step": 36650
     },
     {
-      "epoch": 3.9500591970724357,
-      "grad_norm": 0.7034335732460022,
-      "learning_rate": 0.00036348453830406204,
-      "loss": 3.437,
+      "epoch": 3.9568733153638815,
+      "grad_norm": 0.6269444823265076,
+      "learning_rate": 0.0003630760928224501,
+      "loss": 3.461,
       "step": 36700
     },
     {
-      "epoch": 3.955440749112044,
-      "grad_norm": 0.644279420375824,
-      "learning_rate": 0.00036316129727400064,
-      "loss": 3.45,
+      "epoch": 3.9622641509433962,
+      "grad_norm": 0.6157881021499634,
+      "learning_rate": 0.00036275229357798157,
+      "loss": 3.4509,
       "step": 36750
     },
     {
-      "epoch": 3.9608223011516523,
-      "grad_norm": 0.6632562875747681,
-      "learning_rate": 0.0003628380562439392,
-      "loss": 3.4279,
+      "epoch": 3.967654986522911,
+      "grad_norm": 0.7093312740325928,
+      "learning_rate": 0.0003624284943335132,
+      "loss": 3.4493,
       "step": 36800
     },
     {
-      "epoch": 3.9662038531912605,
-      "grad_norm": 0.6237409114837646,
-      "learning_rate": 0.00036251481521387777,
-      "loss": 3.4434,
+      "epoch": 3.973045822102426,
+      "grad_norm": 0.6717362999916077,
+      "learning_rate": 0.00036210469508904473,
+      "loss": 3.4477,
       "step": 36850
     },
     {
-      "epoch": 3.9715854052308686,
-      "grad_norm": 0.6951926946640015,
-      "learning_rate": 0.00036219157418381637,
-      "loss": 3.4464,
+      "epoch": 3.9784366576819408,
+      "grad_norm": 0.6778683066368103,
+      "learning_rate": 0.00036178089584457633,
+      "loss": 3.4636,
       "step": 36900
     },
     {
-      "epoch": 3.9769669572704767,
-      "grad_norm": 0.6288528442382812,
-      "learning_rate": 0.0003618683331537549,
-      "loss": 3.4348,
+      "epoch": 3.9838274932614555,
+      "grad_norm": 0.6660913825035095,
+      "learning_rate": 0.0003614570966001079,
+      "loss": 3.4728,
       "step": 36950
     },
     {
-      "epoch": 3.9823485093100848,
-      "grad_norm": 0.6089127063751221,
-      "learning_rate": 0.00036154509212369356,
-      "loss": 3.4473,
+      "epoch": 3.9892183288409706,
+      "grad_norm": 0.6355459690093994,
+      "learning_rate": 0.0003611332973556395,
+      "loss": 3.4675,
       "step": 37000
     },
     {
-      "epoch": 3.9823485093100848,
-      "eval_accuracy": 0.37642988558524954,
-      "eval_loss": 3.4435272216796875,
-      "eval_runtime": 184.0558,
-      "eval_samples_per_second": 97.856,
-      "eval_steps_per_second": 6.118,
+      "epoch": 3.9892183288409706,
+      "eval_accuracy": 0.37527197175981064,
+      "eval_loss": 3.4561896324157715,
+      "eval_runtime": 153.322,
+      "eval_samples_per_second": 117.472,
+      "eval_steps_per_second": 7.344,
       "step": 37000
     },
     {
-      "epoch": 3.9877300613496933,
-      "grad_norm": 0.6476089358329773,
-      "learning_rate": 0.00036122185109363215,
-      "loss": 3.4363,
+      "epoch": 3.9946091644204853,
+      "grad_norm": 0.6289277672767639,
+      "learning_rate": 0.00036080949811117104,
+      "loss": 3.465,
       "step": 37050
     },
     {
-      "epoch": 3.9931116133893014,
-      "grad_norm": 0.6080595850944519,
-      "learning_rate": 0.0003608986100635707,
-      "loss": 3.4389,
+      "epoch": 4.0,
+      "grad_norm": 1.2047271728515625,
+      "learning_rate": 0.00036048569886670264,
+      "loss": 3.4564,
       "step": 37100
     },
     {
-      "epoch": 3.9984931654289095,
-      "grad_norm": 0.6331565976142883,
-      "learning_rate": 0.0003605753690335093,
-      "loss": 3.4649,
+      "epoch": 4.005390835579515,
+      "grad_norm": 0.6129406094551086,
+      "learning_rate": 0.0003601618996222342,
+      "loss": 3.3705,
       "step": 37150
     },
     {
-      "epoch": 4.003874717468518,
-      "grad_norm": 0.6497759819030762,
-      "learning_rate": 0.0003602521280034478,
-      "loss": 3.3734,
+      "epoch": 4.010781671159029,
+      "grad_norm": 0.6324290037155151,
+      "learning_rate": 0.00035983810037776574,
+      "loss": 3.37,
       "step": 37200
     },
     {
-      "epoch": 4.009256269508126,
-      "grad_norm": 0.6251653432846069,
-      "learning_rate": 0.0003599288869733865,
-      "loss": 3.3612,
+      "epoch": 4.0161725067385445,
+      "grad_norm": 0.6524683237075806,
+      "learning_rate": 0.00035951430113329735,
+      "loss": 3.3674,
       "step": 37250
     },
     {
-      "epoch": 4.014637821547734,
-      "grad_norm": 0.6418185830116272,
-      "learning_rate": 0.00035960564594332507,
-      "loss": 3.3622,
+      "epoch": 4.02156334231806,
+      "grad_norm": 0.6591086387634277,
+      "learning_rate": 0.0003591905018888289,
+      "loss": 3.3481,
       "step": 37300
     },
     {
-      "epoch": 4.020019373587343,
-      "grad_norm": 0.6301612854003906,
-      "learning_rate": 0.0003592824049132636,
-      "loss": 3.3671,
+      "epoch": 4.026954177897574,
+      "grad_norm": 0.7124868035316467,
+      "learning_rate": 0.0003588667026443605,
+      "loss": 3.3765,
       "step": 37350
     },
     {
-      "epoch": 4.0254009256269505,
-      "grad_norm": 0.6887567639350891,
-      "learning_rate": 0.0003589591638832022,
-      "loss": 3.3564,
+      "epoch": 4.032345013477089,
+      "grad_norm": 0.6549927592277527,
+      "learning_rate": 0.000358542903399892,
+      "loss": 3.3823,
       "step": 37400
     },
     {
-      "epoch": 4.030782477666559,
-      "grad_norm": 0.6322470903396606,
-      "learning_rate": 0.0003586359228531408,
-      "loss": 3.3546,
+      "epoch": 4.037735849056604,
+      "grad_norm": 0.6362309455871582,
+      "learning_rate": 0.0003582191041554236,
+      "loss": 3.3673,
       "step": 37450
     },
     {
-      "epoch": 4.036164029706168,
-      "grad_norm": 0.6689240336418152,
-      "learning_rate": 0.00035831268182307934,
-      "loss": 3.3426,
+      "epoch": 4.0431266846361185,
+      "grad_norm": 0.6904882192611694,
+      "learning_rate": 0.00035789530491095516,
+      "loss": 3.3846,
       "step": 37500
     },
     {
-      "epoch": 4.041545581745775,
-      "grad_norm": 0.7104955911636353,
-      "learning_rate": 0.000357989440793018,
-      "loss": 3.3496,
+      "epoch": 4.048517520215634,
+      "grad_norm": 0.6836847066879272,
+      "learning_rate": 0.00035757150566648676,
+      "loss": 3.3753,
       "step": 37550
     },
     {
-      "epoch": 4.046927133785384,
-      "grad_norm": 0.6500760912895203,
-      "learning_rate": 0.0003576661997629566,
-      "loss": 3.3582,
+      "epoch": 4.053908355795148,
+      "grad_norm": 0.645194411277771,
+      "learning_rate": 0.0003572477064220183,
+      "loss": 3.3688,
       "step": 37600
     },
     {
-      "epoch": 4.0523086858249915,
-      "grad_norm": 0.6422422528266907,
-      "learning_rate": 0.0003573429587328951,
-      "loss": 3.3661,
+      "epoch": 4.059299191374663,
+      "grad_norm": 0.6430479884147644,
+      "learning_rate": 0.00035692390717754986,
+      "loss": 3.3831,
       "step": 37650
     },
     {
-      "epoch": 4.0576902378646,
-      "grad_norm": 0.633980393409729,
-      "learning_rate": 0.0003570197177028337,
-      "loss": 3.3624,
+      "epoch": 4.064690026954178,
+      "grad_norm": 0.6787160634994507,
+      "learning_rate": 0.00035660010793308147,
+      "loss": 3.3883,
       "step": 37700
     },
     {
-      "epoch": 4.063071789904209,
-      "grad_norm": 0.6458887457847595,
-      "learning_rate": 0.00035669647667277226,
-      "loss": 3.3799,
+      "epoch": 4.070080862533692,
+      "grad_norm": 0.6342318058013916,
+      "learning_rate": 0.000356276308688613,
+      "loss": 3.3643,
       "step": 37750
     },
     {
-      "epoch": 4.068453341943816,
-      "grad_norm": 0.6776472330093384,
-      "learning_rate": 0.00035637323564271085,
-      "loss": 3.3828,
+      "epoch": 4.0754716981132075,
+      "grad_norm": 0.644733190536499,
+      "learning_rate": 0.0003559525094441446,
+      "loss": 3.3656,
       "step": 37800
     },
     {
-      "epoch": 4.073834893983425,
-      "grad_norm": 0.6132575273513794,
-      "learning_rate": 0.0003560499946126495,
-      "loss": 3.3563,
+      "epoch": 4.080862533692723,
+      "grad_norm": 0.6571251749992371,
+      "learning_rate": 0.00035562871019967617,
+      "loss": 3.3761,
       "step": 37850
     },
     {
-      "epoch": 4.079216446023033,
-      "grad_norm": 0.6325465440750122,
-      "learning_rate": 0.00035572675358258804,
-      "loss": 3.3695,
+      "epoch": 4.086253369272237,
+      "grad_norm": 0.6402137875556946,
+      "learning_rate": 0.0003553049109552078,
+      "loss": 3.3858,
       "step": 37900
     },
     {
-      "epoch": 4.084597998062641,
-      "grad_norm": 0.6543712019920349,
-      "learning_rate": 0.00035540351255252664,
-      "loss": 3.3814,
+      "epoch": 4.091644204851752,
+      "grad_norm": 0.6880391836166382,
+      "learning_rate": 0.00035498111171073933,
+      "loss": 3.3956,
       "step": 37950
     },
     {
-      "epoch": 4.08997955010225,
-      "grad_norm": 0.6246477365493774,
-      "learning_rate": 0.00035508027152246523,
-      "loss": 3.3674,
+      "epoch": 4.097035040431267,
+      "grad_norm": 0.7105101346969604,
+      "learning_rate": 0.0003546573124662708,
+      "loss": 3.3835,
       "step": 38000
     },
     {
-      "epoch": 4.08997955010225,
-      "eval_accuracy": 0.37657982657043354,
-      "eval_loss": 3.448155641555786,
-      "eval_runtime": 183.9068,
-      "eval_samples_per_second": 97.935,
-      "eval_steps_per_second": 6.123,
+      "epoch": 4.097035040431267,
+      "eval_accuracy": 0.37587032321300495,
+      "eval_loss": 3.4573237895965576,
+      "eval_runtime": 153.0609,
+      "eval_samples_per_second": 117.672,
+      "eval_steps_per_second": 7.357,
       "step": 38000
     },
     {
-      "epoch": 4.095361102141858,
-      "grad_norm": 0.6695476174354553,
-      "learning_rate": 0.0003547570304924038,
-      "loss": 3.3798,
+      "epoch": 4.1024258760107815,
+      "grad_norm": 0.6255121231079102,
+      "learning_rate": 0.0003543335132218025,
+      "loss": 3.3775,
       "step": 38050
     },
     {
-      "epoch": 4.100742654181466,
-      "grad_norm": 0.6408191919326782,
-      "learning_rate": 0.0003544337894623424,
-      "loss": 3.3558,
+      "epoch": 4.107816711590297,
+      "grad_norm": 0.6274397969245911,
+      "learning_rate": 0.000354009713977334,
+      "loss": 3.3866,
       "step": 38100
     },
     {
-      "epoch": 4.106124206221074,
-      "grad_norm": 0.6668918132781982,
-      "learning_rate": 0.000354110548432281,
-      "loss": 3.3707,
+      "epoch": 4.113207547169812,
+      "grad_norm": 0.6001650094985962,
+      "learning_rate": 0.0003536859147328656,
+      "loss": 3.3779,
       "step": 38150
     },
     {
-      "epoch": 4.111505758260682,
-      "grad_norm": 0.6790832281112671,
-      "learning_rate": 0.00035378730740221956,
-      "loss": 3.3478,
+      "epoch": 4.118598382749326,
+      "grad_norm": 0.6584101915359497,
+      "learning_rate": 0.00035336211548839713,
+      "loss": 3.4019,
       "step": 38200
     },
     {
-      "epoch": 4.1168873103002905,
-      "grad_norm": 0.6532649397850037,
-      "learning_rate": 0.00035346406637215815,
-      "loss": 3.3548,
+      "epoch": 4.123989218328841,
+      "grad_norm": 0.6487120985984802,
+      "learning_rate": 0.00035303831624392874,
+      "loss": 3.3931,
       "step": 38250
     },
     {
-      "epoch": 4.122268862339899,
-      "grad_norm": 0.6393909454345703,
-      "learning_rate": 0.0003531408253420967,
-      "loss": 3.3676,
+      "epoch": 4.129380053908355,
+      "grad_norm": 0.6466823220252991,
+      "learning_rate": 0.0003527145169994603,
+      "loss": 3.3971,
       "step": 38300
     },
     {
-      "epoch": 4.127650414379507,
-      "grad_norm": 0.6585549116134644,
-      "learning_rate": 0.0003528175843120353,
-      "loss": 3.3672,
+      "epoch": 4.1347708894878705,
+      "grad_norm": 0.6093391180038452,
+      "learning_rate": 0.0003523907177549919,
+      "loss": 3.3865,
       "step": 38350
     },
     {
-      "epoch": 4.133031966419115,
-      "grad_norm": 0.6466474533081055,
-      "learning_rate": 0.00035249434328197394,
-      "loss": 3.3504,
+      "epoch": 4.140161725067386,
+      "grad_norm": 0.672467052936554,
+      "learning_rate": 0.00035206691851052345,
+      "loss": 3.3805,
       "step": 38400
     },
     {
-      "epoch": 4.138413518458724,
-      "grad_norm": 0.6520695090293884,
-      "learning_rate": 0.0003521711022519125,
-      "loss": 3.3844,
+      "epoch": 4.1455525606469,
+      "grad_norm": 0.6343290209770203,
+      "learning_rate": 0.000351743119266055,
+      "loss": 3.3897,
       "step": 38450
     },
     {
-      "epoch": 4.1437950704983315,
-      "grad_norm": 0.6609041094779968,
-      "learning_rate": 0.00035184786122185107,
-      "loss": 3.3961,
+      "epoch": 4.150943396226415,
+      "grad_norm": 0.6955000162124634,
+      "learning_rate": 0.0003514193200215866,
+      "loss": 3.3954,
       "step": 38500
     },
     {
-      "epoch": 4.14917662253794,
-      "grad_norm": 0.6652035713195801,
-      "learning_rate": 0.00035152462019178967,
-      "loss": 3.374,
+      "epoch": 4.15633423180593,
+      "grad_norm": 0.6665741801261902,
+      "learning_rate": 0.00035109552077711815,
+      "loss": 3.3845,
       "step": 38550
     },
     {
-      "epoch": 4.154558174577549,
-      "grad_norm": 0.709632933139801,
-      "learning_rate": 0.0003512013791617282,
-      "loss": 3.3889,
+      "epoch": 4.1617250673854445,
+      "grad_norm": 0.6940324306488037,
+      "learning_rate": 0.00035077172153264976,
+      "loss": 3.3819,
       "step": 38600
     },
     {
-      "epoch": 4.159939726617156,
-      "grad_norm": 0.6614471673965454,
-      "learning_rate": 0.0003508781381316668,
-      "loss": 3.3652,
+      "epoch": 4.16711590296496,
+      "grad_norm": 0.6948425769805908,
+      "learning_rate": 0.0003504479222881813,
+      "loss": 3.4078,
       "step": 38650
     },
     {
-      "epoch": 4.165321278656765,
-      "grad_norm": 0.7235234379768372,
-      "learning_rate": 0.00035055489710160545,
-      "loss": 3.3865,
+      "epoch": 4.172506738544475,
+      "grad_norm": 0.6753456592559814,
+      "learning_rate": 0.0003501241230437129,
+      "loss": 3.3875,
       "step": 38700
     },
     {
-      "epoch": 4.1707028306963725,
-      "grad_norm": 0.6870402693748474,
-      "learning_rate": 0.000350231656071544,
-      "loss": 3.3727,
+      "epoch": 4.177897574123989,
+      "grad_norm": 0.6313568949699402,
+      "learning_rate": 0.0003498003237992444,
+      "loss": 3.3786,
       "step": 38750
     },
     {
-      "epoch": 4.176084382735981,
-      "grad_norm": 0.6684983968734741,
-      "learning_rate": 0.0003499084150414826,
-      "loss": 3.3698,
+      "epoch": 4.183288409703504,
+      "grad_norm": 0.7218206524848938,
+      "learning_rate": 0.000349476524554776,
+      "loss": 3.4027,
       "step": 38800
     },
     {
-      "epoch": 4.18146593477559,
-      "grad_norm": 0.6721704602241516,
-      "learning_rate": 0.0003495851740114211,
-      "loss": 3.3873,
+      "epoch": 4.188679245283019,
+      "grad_norm": 0.6363169550895691,
+      "learning_rate": 0.00034915272531030756,
+      "loss": 3.4058,
       "step": 38850
     },
     {
-      "epoch": 4.186847486815197,
-      "grad_norm": 0.7069846391677856,
-      "learning_rate": 0.0003492619329813597,
-      "loss": 3.3699,
+      "epoch": 4.1940700808625335,
+      "grad_norm": 0.6768284440040588,
+      "learning_rate": 0.0003488289260658391,
+      "loss": 3.4039,
       "step": 38900
     },
     {
-      "epoch": 4.192229038854806,
-      "grad_norm": 0.6892578601837158,
-      "learning_rate": 0.0003489386919512983,
-      "loss": 3.3793,
+      "epoch": 4.199460916442049,
+      "grad_norm": 0.6650631427764893,
+      "learning_rate": 0.0003485051268213707,
+      "loss": 3.3959,
       "step": 38950
     },
     {
-      "epoch": 4.197610590894414,
-      "grad_norm": 0.6782596111297607,
-      "learning_rate": 0.0003486154509212369,
-      "loss": 3.3835,
+      "epoch": 4.204851752021563,
+      "grad_norm": 0.6531771421432495,
+      "learning_rate": 0.00034818132757690227,
+      "loss": 3.3926,
       "step": 39000
     },
     {
-      "epoch": 4.197610590894414,
-      "eval_accuracy": 0.37674291455504305,
-      "eval_loss": 3.443652629852295,
-      "eval_runtime": 183.6957,
-      "eval_samples_per_second": 98.048,
-      "eval_steps_per_second": 6.13,
+      "epoch": 4.204851752021563,
+      "eval_accuracy": 0.37624245935376954,
+      "eval_loss": 3.4502270221710205,
+      "eval_runtime": 153.1558,
+      "eval_samples_per_second": 117.599,
+      "eval_steps_per_second": 7.352,
       "step": 39000
     },
     {
-      "epoch": 4.202992142934022,
-      "grad_norm": 0.6513041853904724,
-      "learning_rate": 0.0003482922098911755,
-      "loss": 3.3898,
+      "epoch": 4.210242587601078,
+      "grad_norm": 0.6340200901031494,
+      "learning_rate": 0.0003478575283324339,
+      "loss": 3.3914,
       "step": 39050
     },
     {
-      "epoch": 4.208373694973631,
-      "grad_norm": 0.6344335675239563,
-      "learning_rate": 0.0003479689688611141,
-      "loss": 3.4027,
+      "epoch": 4.215633423180593,
+      "grad_norm": 0.6285250186920166,
+      "learning_rate": 0.0003475337290879654,
+      "loss": 3.3935,
       "step": 39100
     },
     {
-      "epoch": 4.213755247013238,
-      "grad_norm": 0.6895449757575989,
-      "learning_rate": 0.00034764572783105264,
-      "loss": 3.374,
+      "epoch": 4.2210242587601075,
+      "grad_norm": 0.6567074656486511,
+      "learning_rate": 0.00034720992984349703,
+      "loss": 3.3887,
       "step": 39150
     },
     {
-      "epoch": 4.219136799052847,
-      "grad_norm": 0.6263428330421448,
-      "learning_rate": 0.00034732248680099123,
-      "loss": 3.3678,
+      "epoch": 4.226415094339623,
+      "grad_norm": 0.615469753742218,
+      "learning_rate": 0.0003468861305990286,
+      "loss": 3.3863,
       "step": 39200
     },
     {
-      "epoch": 4.224518351092455,
-      "grad_norm": 0.6432785987854004,
-      "learning_rate": 0.0003469992457709299,
-      "loss": 3.4044,
+      "epoch": 4.231805929919138,
+      "grad_norm": 0.679693341255188,
+      "learning_rate": 0.0003465623313545602,
+      "loss": 3.3851,
       "step": 39250
     },
     {
-      "epoch": 4.229899903132063,
-      "grad_norm": 0.6940696835517883,
-      "learning_rate": 0.0003466760047408684,
-      "loss": 3.3697,
+      "epoch": 4.237196765498652,
+      "grad_norm": 0.6851875185966492,
+      "learning_rate": 0.00034623853211009173,
+      "loss": 3.397,
       "step": 39300
     },
     {
-      "epoch": 4.2352814551716715,
-      "grad_norm": 0.6519486308097839,
-      "learning_rate": 0.000346352763710807,
-      "loss": 3.3843,
+      "epoch": 4.242587601078167,
+      "grad_norm": 0.6618112325668335,
+      "learning_rate": 0.00034591473286562323,
+      "loss": 3.3952,
       "step": 39350
     },
     {
-      "epoch": 4.24066300721128,
-      "grad_norm": 0.6791576147079468,
-      "learning_rate": 0.00034602952268074556,
-      "loss": 3.3808,
+      "epoch": 4.247978436657682,
+      "grad_norm": 0.6320863962173462,
+      "learning_rate": 0.0003455909336211549,
+      "loss": 3.3999,
       "step": 39400
     },
     {
-      "epoch": 4.246044559250888,
-      "grad_norm": 0.6561434268951416,
-      "learning_rate": 0.00034570628165068415,
-      "loss": 3.372,
+      "epoch": 4.2533692722371965,
+      "grad_norm": 0.6585174202919006,
+      "learning_rate": 0.0003452671343766864,
+      "loss": 3.3906,
       "step": 39450
     },
     {
-      "epoch": 4.251426111290496,
-      "grad_norm": 0.664330005645752,
-      "learning_rate": 0.0003453830406206227,
-      "loss": 3.3777,
+      "epoch": 4.258760107816712,
+      "grad_norm": 0.6757383942604065,
+      "learning_rate": 0.000344943335132218,
+      "loss": 3.3821,
       "step": 39500
     },
     {
-      "epoch": 4.256807663330104,
-      "grad_norm": 0.665077805519104,
-      "learning_rate": 0.00034505979959056134,
-      "loss": 3.3798,
+      "epoch": 4.264150943396227,
+      "grad_norm": 0.7111407518386841,
+      "learning_rate": 0.00034461953588774954,
+      "loss": 3.3879,
       "step": 39550
     },
     {
-      "epoch": 4.2621892153697125,
-      "grad_norm": 0.6542532444000244,
-      "learning_rate": 0.00034473655856049994,
-      "loss": 3.3848,
+      "epoch": 4.269541778975741,
+      "grad_norm": 0.646880567073822,
+      "learning_rate": 0.0003443022126281705,
+      "loss": 3.3862,
       "step": 39600
     },
     {
-      "epoch": 4.267570767409321,
-      "grad_norm": 0.6718242764472961,
-      "learning_rate": 0.0003444133175304385,
-      "loss": 3.3723,
+      "epoch": 4.274932614555256,
+      "grad_norm": 0.690201461315155,
+      "learning_rate": 0.00034397841338370204,
+      "loss": 3.4138,
       "step": 39650
     },
     {
-      "epoch": 4.272952319448929,
-      "grad_norm": 0.6607034802436829,
-      "learning_rate": 0.00034409007650037707,
-      "loss": 3.3751,
+      "epoch": 4.280323450134771,
+      "grad_norm": 0.688504159450531,
+      "learning_rate": 0.00034365461413923365,
+      "loss": 3.3974,
       "step": 39700
     },
     {
-      "epoch": 4.278333871488537,
-      "grad_norm": 0.6631885170936584,
-      "learning_rate": 0.00034376683547031567,
-      "loss": 3.379,
+      "epoch": 4.285714285714286,
+      "grad_norm": 0.6902996897697449,
+      "learning_rate": 0.0003433308148947652,
+      "loss": 3.3922,
       "step": 39750
     },
     {
-      "epoch": 4.283715423528146,
-      "grad_norm": 0.6903967261314392,
-      "learning_rate": 0.0003434435944402542,
-      "loss": 3.3807,
+      "epoch": 4.291105121293801,
+      "grad_norm": 0.617358922958374,
+      "learning_rate": 0.0003430070156502968,
+      "loss": 3.407,
       "step": 39800
     },
     {
-      "epoch": 4.2890969755677535,
-      "grad_norm": 0.6927313804626465,
-      "learning_rate": 0.00034312035341019286,
-      "loss": 3.3925,
+      "epoch": 4.296495956873315,
+      "grad_norm": 0.6996208429336548,
+      "learning_rate": 0.00034268321640582835,
+      "loss": 3.4111,
       "step": 39850
     },
     {
-      "epoch": 4.294478527607362,
-      "grad_norm": 0.6660029888153076,
-      "learning_rate": 0.00034279711238013145,
-      "loss": 3.3945,
+      "epoch": 4.30188679245283,
+      "grad_norm": 0.6344699263572693,
+      "learning_rate": 0.00034235941716135996,
+      "loss": 3.3991,
       "step": 39900
     },
     {
-      "epoch": 4.299860079646971,
-      "grad_norm": 0.6319341063499451,
-      "learning_rate": 0.00034247387135007,
-      "loss": 3.3878,
+      "epoch": 4.307277628032345,
+      "grad_norm": 0.6900128126144409,
+      "learning_rate": 0.0003420356179168915,
+      "loss": 3.4037,
       "step": 39950
     },
     {
-      "epoch": 4.305241631686578,
-      "grad_norm": 0.7009266018867493,
-      "learning_rate": 0.0003421506303200086,
-      "loss": 3.3741,
+      "epoch": 4.3126684636118595,
+      "grad_norm": 0.666391134262085,
+      "learning_rate": 0.0003417118186724231,
+      "loss": 3.397,
       "step": 40000
     },
     {
-      "epoch": 4.305241631686578,
-      "eval_accuracy": 0.3771503628843474,
-      "eval_loss": 3.4397826194763184,
-      "eval_runtime": 183.7941,
-      "eval_samples_per_second": 97.996,
-      "eval_steps_per_second": 6.126,
+      "epoch": 4.3126684636118595,
+      "eval_accuracy": 0.37674943372831193,
+      "eval_loss": 3.447213888168335,
+      "eval_runtime": 153.0175,
+      "eval_samples_per_second": 117.705,
+      "eval_steps_per_second": 7.359,
       "step": 40000
     },
     {
-      "epoch": 4.310623183726187,
-      "grad_norm": 0.6515907645225525,
-      "learning_rate": 0.0003418273892899471,
-      "loss": 3.3745,
+      "epoch": 4.318059299191375,
+      "grad_norm": 0.64959716796875,
+      "learning_rate": 0.00034138801942795467,
+      "loss": 3.3886,
       "step": 40050
     },
     {
-      "epoch": 4.3160047357657945,
-      "grad_norm": 0.6438180804252625,
-      "learning_rate": 0.0003415041482598858,
-      "loss": 3.3736,
+      "epoch": 4.32345013477089,
+      "grad_norm": 0.6654952764511108,
+      "learning_rate": 0.00034106422018348616,
+      "loss": 3.4146,
       "step": 40100
     },
     {
-      "epoch": 4.321386287805403,
-      "grad_norm": 0.6765071153640747,
-      "learning_rate": 0.00034118090722982437,
-      "loss": 3.3899,
+      "epoch": 4.328840970350404,
+      "grad_norm": 0.6415016055107117,
+      "learning_rate": 0.00034074042093901777,
+      "loss": 3.424,
       "step": 40150
     },
     {
-      "epoch": 4.326767839845012,
-      "grad_norm": 0.6893318891525269,
-      "learning_rate": 0.0003408576661997629,
-      "loss": 3.3995,
+      "epoch": 4.334231805929919,
+      "grad_norm": 0.7035545706748962,
+      "learning_rate": 0.0003404166216945493,
+      "loss": 3.3916,
       "step": 40200
     },
     {
-      "epoch": 4.332149391884619,
-      "grad_norm": 0.6063430905342102,
-      "learning_rate": 0.00034054088999030277,
-      "loss": 3.3855,
+      "epoch": 4.339622641509434,
+      "grad_norm": 0.6636276245117188,
+      "learning_rate": 0.0003400928224500809,
+      "loss": 3.3994,
       "step": 40250
     },
     {
-      "epoch": 4.337530943924228,
-      "grad_norm": 0.6637575030326843,
-      "learning_rate": 0.0003402176489602413,
-      "loss": 3.3877,
+      "epoch": 4.345013477088949,
+      "grad_norm": 0.656891942024231,
+      "learning_rate": 0.00033976902320561247,
+      "loss": 3.4045,
       "step": 40300
     },
     {
-      "epoch": 4.342912495963836,
-      "grad_norm": 0.7113229632377625,
-      "learning_rate": 0.0003398944079301799,
-      "loss": 3.3709,
+      "epoch": 4.350404312668464,
+      "grad_norm": 0.6576977372169495,
+      "learning_rate": 0.0003394452239611441,
+      "loss": 3.3972,
       "step": 40350
     },
     {
-      "epoch": 4.348294048003444,
-      "grad_norm": 0.7056190371513367,
-      "learning_rate": 0.0003395711669001185,
-      "loss": 3.3923,
+      "epoch": 4.355795148247978,
+      "grad_norm": 0.6619424223899841,
+      "learning_rate": 0.00033912142471667563,
+      "loss": 3.4106,
       "step": 40400
     },
     {
-      "epoch": 4.3536756000430525,
-      "grad_norm": 0.6472856998443604,
-      "learning_rate": 0.00033924792587005704,
-      "loss": 3.3856,
+      "epoch": 4.361185983827493,
+      "grad_norm": 0.6731002926826477,
+      "learning_rate": 0.00033879762547220723,
+      "loss": 3.381,
       "step": 40450
     },
     {
-      "epoch": 4.359057152082661,
-      "grad_norm": 0.6156180500984192,
-      "learning_rate": 0.0003389246848399957,
-      "loss": 3.3742,
+      "epoch": 4.366576819407008,
+      "grad_norm": 0.7423571944236755,
+      "learning_rate": 0.0003384738262277388,
+      "loss": 3.3891,
       "step": 40500
     },
     {
-      "epoch": 4.364438704122269,
-      "grad_norm": 0.6725302934646606,
-      "learning_rate": 0.0003386014438099343,
-      "loss": 3.3823,
+      "epoch": 4.3719676549865225,
+      "grad_norm": 0.6814403533935547,
+      "learning_rate": 0.00033815002698327033,
+      "loss": 3.4113,
       "step": 40550
     },
     {
-      "epoch": 4.369820256161877,
-      "grad_norm": 0.6599554419517517,
-      "learning_rate": 0.0003382782027798728,
-      "loss": 3.386,
+      "epoch": 4.377358490566038,
+      "grad_norm": 0.6880382895469666,
+      "learning_rate": 0.00033782622773880194,
+      "loss": 3.3949,
       "step": 40600
     },
     {
-      "epoch": 4.375201808201485,
-      "grad_norm": 0.6923736929893494,
-      "learning_rate": 0.0003379549617498114,
-      "loss": 3.4055,
+      "epoch": 4.382749326145553,
+      "grad_norm": 0.6599591374397278,
+      "learning_rate": 0.0003375024284943335,
+      "loss": 3.3867,
       "step": 40650
     },
     {
-      "epoch": 4.3805833602410935,
-      "grad_norm": 0.6569280028343201,
-      "learning_rate": 0.00033763172071974996,
-      "loss": 3.3698,
+      "epoch": 4.388140161725067,
+      "grad_norm": 0.6115888953208923,
+      "learning_rate": 0.0003371786292498651,
+      "loss": 3.4154,
       "step": 40700
     },
     {
-      "epoch": 4.385964912280702,
-      "grad_norm": 0.6321343183517456,
-      "learning_rate": 0.00033730847968968855,
-      "loss": 3.3821,
+      "epoch": 4.393530997304582,
+      "grad_norm": 0.7015092968940735,
+      "learning_rate": 0.0003368548300053966,
+      "loss": 3.4113,
       "step": 40750
     },
     {
-      "epoch": 4.39134646432031,
-      "grad_norm": 0.6720413565635681,
-      "learning_rate": 0.0003369852386596272,
-      "loss": 3.3865,
+      "epoch": 4.398921832884097,
+      "grad_norm": 0.6767927408218384,
+      "learning_rate": 0.0003365310307609282,
+      "loss": 3.3836,
       "step": 40800
     },
     {
-      "epoch": 4.396728016359918,
-      "grad_norm": 0.758339524269104,
-      "learning_rate": 0.00033666199762956574,
-      "loss": 3.3976,
+      "epoch": 4.404312668463612,
+      "grad_norm": 0.6451999545097351,
+      "learning_rate": 0.00033620723151645975,
+      "loss": 3.4047,
       "step": 40850
     },
     {
-      "epoch": 4.402109568399527,
-      "grad_norm": 0.6490445733070374,
-      "learning_rate": 0.00033633875659950434,
-      "loss": 3.3682,
+      "epoch": 4.409703504043127,
+      "grad_norm": 0.6515558362007141,
+      "learning_rate": 0.0003358834322719913,
+      "loss": 3.3929,
       "step": 40900
     },
     {
-      "epoch": 4.4074911204391345,
-      "grad_norm": 0.668504536151886,
-      "learning_rate": 0.0003360155155694429,
-      "loss": 3.3829,
+      "epoch": 4.415094339622642,
+      "grad_norm": 0.6988462805747986,
+      "learning_rate": 0.0003355596330275229,
+      "loss": 3.4009,
       "step": 40950
     },
     {
-      "epoch": 4.412872672478743,
-      "grad_norm": 0.7141649723052979,
-      "learning_rate": 0.0003356922745393815,
-      "loss": 3.3964,
+      "epoch": 4.420485175202156,
+      "grad_norm": 0.6392936110496521,
+      "learning_rate": 0.00033523583378305445,
+      "loss": 3.4106,
       "step": 41000
     },
     {
-      "epoch": 4.412872672478743,
-      "eval_accuracy": 0.37797264793932733,
-      "eval_loss": 3.4327268600463867,
-      "eval_runtime": 183.6039,
-      "eval_samples_per_second": 98.097,
-      "eval_steps_per_second": 6.133,
+      "epoch": 4.420485175202156,
+      "eval_accuracy": 0.37732724978570936,
+      "eval_loss": 3.439938545227051,
+      "eval_runtime": 153.1949,
+      "eval_samples_per_second": 117.569,
+      "eval_steps_per_second": 7.35,
       "step": 41000
     },
     {
-      "epoch": 4.418254224518351,
-      "grad_norm": 0.6447309255599976,
-      "learning_rate": 0.0003353690335093201,
-      "loss": 3.3818,
+      "epoch": 4.425876010781671,
+      "grad_norm": 0.651278018951416,
+      "learning_rate": 0.00033491203453858606,
+      "loss": 3.4121,
       "step": 41050
     },
     {
-      "epoch": 4.423635776557959,
-      "grad_norm": 0.6799582242965698,
-      "learning_rate": 0.0003350457924792587,
-      "loss": 3.3834,
+      "epoch": 4.431266846361186,
+      "grad_norm": 0.7106104493141174,
+      "learning_rate": 0.0003345882352941176,
+      "loss": 3.3939,
       "step": 41100
     },
     {
-      "epoch": 4.429017328597568,
-      "grad_norm": 0.7001302242279053,
-      "learning_rate": 0.00033472255144919726,
-      "loss": 3.3851,
+      "epoch": 4.436657681940701,
+      "grad_norm": 0.6541004180908203,
+      "learning_rate": 0.0003342644360496492,
+      "loss": 3.3725,
       "step": 41150
     },
     {
-      "epoch": 4.4343988806371755,
-      "grad_norm": 0.6865254044532776,
-      "learning_rate": 0.00033439931041913585,
-      "loss": 3.3843,
+      "epoch": 4.442048517520216,
+      "grad_norm": 0.6633983850479126,
+      "learning_rate": 0.00033394063680518076,
+      "loss": 3.4092,
       "step": 41200
     },
     {
-      "epoch": 4.439780432676784,
-      "grad_norm": 0.6883783340454102,
-      "learning_rate": 0.0003340760693890744,
-      "loss": 3.397,
+      "epoch": 4.44743935309973,
+      "grad_norm": 0.7057939767837524,
+      "learning_rate": 0.00033361683756071237,
+      "loss": 3.3972,
       "step": 41250
     },
     {
-      "epoch": 4.445161984716393,
-      "grad_norm": 0.6731565594673157,
-      "learning_rate": 0.000333752828359013,
-      "loss": 3.3859,
+      "epoch": 4.452830188679245,
+      "grad_norm": 0.6378117203712463,
+      "learning_rate": 0.0003332930383162439,
+      "loss": 3.3939,
       "step": 41300
     },
     {
-      "epoch": 4.450543536756,
-      "grad_norm": 0.6746081709861755,
-      "learning_rate": 0.00033342958732895164,
-      "loss": 3.3994,
+      "epoch": 4.45822102425876,
+      "grad_norm": 0.6508496999740601,
+      "learning_rate": 0.0003329692390717754,
+      "loss": 3.386,
       "step": 41350
     },
     {
-      "epoch": 4.455925088795609,
-      "grad_norm": 0.6333791017532349,
-      "learning_rate": 0.0003331063462988902,
-      "loss": 3.3857,
+      "epoch": 4.463611859838275,
+      "grad_norm": 0.6816674470901489,
+      "learning_rate": 0.0003326454398273071,
+      "loss": 3.3986,
       "step": 41400
     },
     {
-      "epoch": 4.461306640835216,
-      "grad_norm": 0.6702046394348145,
-      "learning_rate": 0.00033278310526882877,
-      "loss": 3.387,
+      "epoch": 4.46900269541779,
+      "grad_norm": 0.7071017622947693,
+      "learning_rate": 0.00033232164058283857,
+      "loss": 3.3836,
       "step": 41450
     },
     {
-      "epoch": 4.466688192874825,
-      "grad_norm": 0.6647279858589172,
-      "learning_rate": 0.0003324598642387673,
-      "loss": 3.3865,
+      "epoch": 4.474393530997305,
+      "grad_norm": 0.615455687046051,
+      "learning_rate": 0.0003319978413383702,
+      "loss": 3.3863,
       "step": 41500
     },
     {
-      "epoch": 4.4720697449144335,
-      "grad_norm": 0.63032466173172,
-      "learning_rate": 0.0003321366232087059,
-      "loss": 3.3858,
+      "epoch": 4.479784366576819,
+      "grad_norm": 0.6691483855247498,
+      "learning_rate": 0.0003316740420939017,
+      "loss": 3.4171,
       "step": 41550
     },
     {
-      "epoch": 4.477451296954041,
-      "grad_norm": 0.6740289330482483,
-      "learning_rate": 0.0003318133821786445,
-      "loss": 3.3869,
+      "epoch": 4.485175202156334,
+      "grad_norm": 0.7083690166473389,
+      "learning_rate": 0.00033135024284943333,
+      "loss": 3.4086,
       "step": 41600
     },
     {
-      "epoch": 4.48283284899365,
-      "grad_norm": 0.6252920627593994,
-      "learning_rate": 0.0003314901411485831,
-      "loss": 3.4067,
+      "epoch": 4.490566037735849,
+      "grad_norm": 0.6974473595619202,
+      "learning_rate": 0.0003310264436049649,
+      "loss": 3.4116,
       "step": 41650
     },
     {
-      "epoch": 4.488214401033258,
-      "grad_norm": 0.6515995860099792,
-      "learning_rate": 0.0003311669001185217,
-      "loss": 3.383,
+      "epoch": 4.495956873315364,
+      "grad_norm": 0.7207027673721313,
+      "learning_rate": 0.0003307026443604965,
+      "loss": 3.4138,
       "step": 41700
     },
     {
-      "epoch": 4.493595953072866,
-      "grad_norm": 0.6926605701446533,
-      "learning_rate": 0.0003308436590884603,
-      "loss": 3.371,
+      "epoch": 4.501347708894879,
+      "grad_norm": 0.7258405685424805,
+      "learning_rate": 0.00033037884511602804,
+      "loss": 3.3978,
       "step": 41750
     },
     {
-      "epoch": 4.4989775051124745,
-      "grad_norm": 0.6879797577857971,
-      "learning_rate": 0.0003305204180583988,
-      "loss": 3.3882,
+      "epoch": 4.506738544474393,
+      "grad_norm": 0.6333111524581909,
+      "learning_rate": 0.0003300550458715596,
+      "loss": 3.4105,
       "step": 41800
     },
     {
-      "epoch": 4.504359057152083,
-      "grad_norm": 0.6967585682868958,
-      "learning_rate": 0.0003301971770283374,
-      "loss": 3.403,
+      "epoch": 4.512129380053908,
+      "grad_norm": 0.673058271408081,
+      "learning_rate": 0.0003297312466270912,
+      "loss": 3.3864,
       "step": 41850
     },
     {
-      "epoch": 4.509740609191691,
-      "grad_norm": 0.6855674386024475,
-      "learning_rate": 0.00032987393599827607,
-      "loss": 3.3982,
+      "epoch": 4.517520215633423,
+      "grad_norm": 0.639115571975708,
+      "learning_rate": 0.00032941392336751214,
+      "loss": 3.4088,
       "step": 41900
     },
     {
-      "epoch": 4.515122161231299,
-      "grad_norm": 0.7353305816650391,
-      "learning_rate": 0.0003295571597888158,
-      "loss": 3.3927,
+      "epoch": 4.5229110512129385,
+      "grad_norm": 0.698583722114563,
+      "learning_rate": 0.0003290901241230437,
+      "loss": 3.4115,
       "step": 41950
     },
     {
-      "epoch": 4.520503713270907,
-      "grad_norm": 0.6033384203910828,
-      "learning_rate": 0.00032923391875875447,
-      "loss": 3.3886,
+      "epoch": 4.528301886792453,
+      "grad_norm": 0.673023521900177,
+      "learning_rate": 0.0003287663248785753,
+      "loss": 3.4111,
       "step": 42000
     },
     {
-      "epoch": 4.520503713270907,
-      "eval_accuracy": 0.3779805796001378,
-      "eval_loss": 3.4293768405914307,
-      "eval_runtime": 184.2706,
-      "eval_samples_per_second": 97.742,
-      "eval_steps_per_second": 6.111,
+      "epoch": 4.528301886792453,
+      "eval_accuracy": 0.37781347145867916,
+      "eval_loss": 3.4361653327941895,
+      "eval_runtime": 153.4909,
+      "eval_samples_per_second": 117.342,
+      "eval_steps_per_second": 7.336,
       "step": 42000
     },
     {
-      "epoch": 4.5258852653105155,
-      "grad_norm": 0.6615278124809265,
-      "learning_rate": 0.000328910677728693,
-      "loss": 3.3826,
+      "epoch": 4.533692722371968,
+      "grad_norm": 0.6819300055503845,
+      "learning_rate": 0.00032844252563410685,
+      "loss": 3.3866,
       "step": 42050
     },
     {
-      "epoch": 4.531266817350124,
-      "grad_norm": 1.198451280593872,
-      "learning_rate": 0.0003285874366986316,
-      "loss": 3.3808,
+      "epoch": 4.539083557951482,
+      "grad_norm": 0.6776103973388672,
+      "learning_rate": 0.00032811872638963834,
+      "loss": 3.3973,
       "step": 42100
     },
     {
-      "epoch": 4.536648369389732,
-      "grad_norm": 0.6494317054748535,
-      "learning_rate": 0.00032826419566857015,
-      "loss": 3.3677,
+      "epoch": 4.544474393530997,
+      "grad_norm": 0.666259765625,
+      "learning_rate": 0.00032779492714516995,
+      "loss": 3.402,
       "step": 42150
     },
     {
-      "epoch": 4.54202992142934,
-      "grad_norm": 0.6694767475128174,
-      "learning_rate": 0.00032794095463850874,
-      "loss": 3.3983,
+      "epoch": 4.549865229110512,
+      "grad_norm": 0.6919211149215698,
+      "learning_rate": 0.0003274711279007015,
+      "loss": 3.412,
       "step": 42200
     },
     {
-      "epoch": 4.547411473468949,
-      "grad_norm": 0.6695189476013184,
-      "learning_rate": 0.0003276177136084473,
-      "loss": 3.3945,
+      "epoch": 4.555256064690027,
+      "grad_norm": 0.6883066892623901,
+      "learning_rate": 0.0003271473286562331,
+      "loss": 3.407,
       "step": 42250
     },
     {
-      "epoch": 4.5527930255085565,
-      "grad_norm": 0.6397526860237122,
-      "learning_rate": 0.00032729447257838593,
-      "loss": 3.3999,
+      "epoch": 4.560646900269542,
+      "grad_norm": 0.6889147162437439,
+      "learning_rate": 0.00032682352941176466,
+      "loss": 3.3982,
       "step": 42300
     },
     {
-      "epoch": 4.558174577548165,
-      "grad_norm": 0.6829812526702881,
-      "learning_rate": 0.0003269712315483245,
-      "loss": 3.3843,
+      "epoch": 4.566037735849057,
+      "grad_norm": 0.6600341200828552,
+      "learning_rate": 0.00032649973016729626,
+      "loss": 3.421,
       "step": 42350
     },
     {
-      "epoch": 4.563556129587774,
-      "grad_norm": 0.6462880373001099,
-      "learning_rate": 0.00032664799051826306,
-      "loss": 3.3858,
+      "epoch": 4.571428571428571,
+      "grad_norm": 0.6524129509925842,
+      "learning_rate": 0.0003261759309228278,
+      "loss": 3.3978,
       "step": 42400
     },
     {
-      "epoch": 4.568937681627381,
-      "grad_norm": 0.6427911520004272,
-      "learning_rate": 0.00032632474948820166,
-      "loss": 3.385,
+      "epoch": 4.576819407008086,
+      "grad_norm": 0.6513863801956177,
+      "learning_rate": 0.0003258521316783594,
+      "loss": 3.4099,
       "step": 42450
     },
     {
-      "epoch": 4.57431923366699,
-      "grad_norm": 0.7188562750816345,
-      "learning_rate": 0.00032600150845814025,
-      "loss": 3.4039,
+      "epoch": 4.5822102425876015,
+      "grad_norm": 0.6730241775512695,
+      "learning_rate": 0.00032552833243389097,
+      "loss": 3.3932,
       "step": 42500
     },
     {
-      "epoch": 4.579700785706597,
-      "grad_norm": 0.6983649730682373,
-      "learning_rate": 0.0003256782674280788,
-      "loss": 3.3888,
+      "epoch": 4.587601078167116,
+      "grad_norm": 0.6982329487800598,
+      "learning_rate": 0.0003252045331894225,
+      "loss": 3.4039,
       "step": 42550
     },
     {
-      "epoch": 4.585082337746206,
-      "grad_norm": 0.6858247518539429,
-      "learning_rate": 0.00032535502639801744,
-      "loss": 3.3953,
+      "epoch": 4.592991913746631,
+      "grad_norm": 0.6633615493774414,
+      "learning_rate": 0.0003248807339449541,
+      "loss": 3.3863,
       "step": 42600
     },
     {
-      "epoch": 4.5904638897858145,
-      "grad_norm": 0.6501060724258423,
-      "learning_rate": 0.00032503178536795604,
-      "loss": 3.4001,
+      "epoch": 4.598382749326145,
+      "grad_norm": 0.6624379754066467,
+      "learning_rate": 0.00032455693470048567,
+      "loss": 3.4121,
       "step": 42650
     },
     {
-      "epoch": 4.595845441825422,
-      "grad_norm": 0.6415064334869385,
-      "learning_rate": 0.0003247085443378946,
-      "loss": 3.3793,
+      "epoch": 4.60377358490566,
+      "grad_norm": 0.6563555598258972,
+      "learning_rate": 0.0003242331354560173,
+      "loss": 3.4058,
       "step": 42700
     },
     {
-      "epoch": 4.601226993865031,
-      "grad_norm": 0.6613024473190308,
-      "learning_rate": 0.0003243853033078332,
-      "loss": 3.3915,
+      "epoch": 4.609164420485175,
+      "grad_norm": 0.6863491535186768,
+      "learning_rate": 0.0003239093362115488,
+      "loss": 3.3958,
       "step": 42750
     },
     {
-      "epoch": 4.606608545904638,
-      "grad_norm": 0.6808081865310669,
-      "learning_rate": 0.0003240620622777717,
-      "loss": 3.3959,
+      "epoch": 4.6145552560646905,
+      "grad_norm": 0.6450868844985962,
+      "learning_rate": 0.00032358553696708043,
+      "loss": 3.396,
       "step": 42800
     },
     {
-      "epoch": 4.611990097944247,
-      "grad_norm": 0.6326537728309631,
-      "learning_rate": 0.00032373882124771036,
-      "loss": 3.4062,
+      "epoch": 4.619946091644205,
+      "grad_norm": 0.6849892139434814,
+      "learning_rate": 0.00032326173772261193,
+      "loss": 3.4051,
       "step": 42850
     },
     {
-      "epoch": 4.6173716499838555,
-      "grad_norm": 0.6862735152244568,
-      "learning_rate": 0.00032341558021764896,
-      "loss": 3.3776,
+      "epoch": 4.62533692722372,
+      "grad_norm": 0.6607340574264526,
+      "learning_rate": 0.00032293793847814353,
+      "loss": 3.4162,
       "step": 42900
     },
     {
-      "epoch": 4.622753202023463,
-      "grad_norm": 0.6448619365692139,
-      "learning_rate": 0.0003230923391875875,
-      "loss": 3.3933,
+      "epoch": 4.630727762803234,
+      "grad_norm": 0.6358455419540405,
+      "learning_rate": 0.0003226141392336751,
+      "loss": 3.4262,
       "step": 42950
     },
     {
-      "epoch": 4.628134754063072,
-      "grad_norm": 0.6452128887176514,
-      "learning_rate": 0.0003227690981575261,
-      "loss": 3.3795,
+      "epoch": 4.636118598382749,
+      "grad_norm": 0.6990907788276672,
+      "learning_rate": 0.00032229033998920663,
+      "loss": 3.4011,
       "step": 43000
     },
     {
-      "epoch": 4.628134754063072,
-      "eval_accuracy": 0.37889619748575043,
-      "eval_loss": 3.4241116046905518,
-      "eval_runtime": 184.037,
-      "eval_samples_per_second": 97.866,
-      "eval_steps_per_second": 6.118,
+      "epoch": 4.636118598382749,
+      "eval_accuracy": 0.3783593435670591,
+      "eval_loss": 3.4296069145202637,
+      "eval_runtime": 153.266,
+      "eval_samples_per_second": 117.515,
+      "eval_steps_per_second": 7.347,
       "step": 43000
     },
     {
-      "epoch": 4.63351630610268,
-      "grad_norm": 0.6427309513092041,
-      "learning_rate": 0.0003224458571274647,
-      "loss": 3.3793,
+      "epoch": 4.6415094339622645,
+      "grad_norm": 0.6236608028411865,
+      "learning_rate": 0.00032196654074473824,
+      "loss": 3.403,
       "step": 43050
     },
     {
-      "epoch": 4.638897858142288,
-      "grad_norm": 0.7281280755996704,
-      "learning_rate": 0.00032212261609740323,
-      "loss": 3.385,
+      "epoch": 4.646900269541779,
+      "grad_norm": 0.6639171838760376,
+      "learning_rate": 0.0003216427415002698,
+      "loss": 3.3982,
       "step": 43100
     },
     {
-      "epoch": 4.6442794101818965,
-      "grad_norm": 0.6569000482559204,
-      "learning_rate": 0.0003217993750673419,
-      "loss": 3.4093,
+      "epoch": 4.652291105121294,
+      "grad_norm": 0.7040235996246338,
+      "learning_rate": 0.0003213189422558014,
+      "loss": 3.4073,
       "step": 43150
     },
     {
-      "epoch": 4.649660962221505,
-      "grad_norm": 0.6708672046661377,
-      "learning_rate": 0.00032147613403728047,
-      "loss": 3.4015,
+      "epoch": 4.657681940700809,
+      "grad_norm": 0.7440651655197144,
+      "learning_rate": 0.00032099514301133295,
+      "loss": 3.4032,
       "step": 43200
     },
     {
-      "epoch": 4.655042514261113,
-      "grad_norm": 0.6629785299301147,
-      "learning_rate": 0.000321152893007219,
-      "loss": 3.3755,
+      "epoch": 4.663072776280323,
+      "grad_norm": 0.6286876201629639,
+      "learning_rate": 0.00032067134376686455,
+      "loss": 3.3933,
       "step": 43250
     },
     {
-      "epoch": 4.660424066300721,
-      "grad_norm": 0.6651688814163208,
-      "learning_rate": 0.0003208296519771576,
-      "loss": 3.3926,
+      "epoch": 4.668463611859838,
+      "grad_norm": 0.6850646734237671,
+      "learning_rate": 0.0003203475445223961,
+      "loss": 3.3945,
       "step": 43300
     },
     {
-      "epoch": 4.665805618340329,
-      "grad_norm": 0.6142232418060303,
-      "learning_rate": 0.00032050641094709615,
-      "loss": 3.3973,
+      "epoch": 4.6738544474393535,
+      "grad_norm": 0.6414585113525391,
+      "learning_rate": 0.0003200237452779277,
+      "loss": 3.3923,
       "step": 43350
     },
     {
-      "epoch": 4.6711871703799375,
-      "grad_norm": 0.6867098212242126,
-      "learning_rate": 0.00032018316991703474,
-      "loss": 3.3816,
+      "epoch": 4.679245283018868,
+      "grad_norm": 0.6378012299537659,
+      "learning_rate": 0.00031969994603345926,
+      "loss": 3.4163,
       "step": 43400
     },
     {
-      "epoch": 4.676568722419546,
-      "grad_norm": 0.6545022130012512,
-      "learning_rate": 0.0003198599288869734,
-      "loss": 3.3913,
+      "epoch": 4.684636118598383,
+      "grad_norm": 0.6793572306632996,
+      "learning_rate": 0.00031937614678899075,
+      "loss": 3.4093,
       "step": 43450
     },
     {
-      "epoch": 4.681950274459154,
-      "grad_norm": 0.7143167853355408,
-      "learning_rate": 0.00031953668785691193,
-      "loss": 3.3938,
+      "epoch": 4.690026954177897,
+      "grad_norm": 0.705450713634491,
+      "learning_rate": 0.00031905234754452236,
+      "loss": 3.4201,
       "step": 43500
     },
     {
-      "epoch": 4.687331826498762,
-      "grad_norm": 0.6519163846969604,
-      "learning_rate": 0.0003192134468268505,
-      "loss": 3.3991,
+      "epoch": 4.695417789757412,
+      "grad_norm": 0.6721307039260864,
+      "learning_rate": 0.0003187285483000539,
+      "loss": 3.3845,
       "step": 43550
     },
     {
-      "epoch": 4.692713378538371,
-      "grad_norm": 0.6904447078704834,
-      "learning_rate": 0.0003188902057967891,
-      "loss": 3.406,
+      "epoch": 4.7008086253369274,
+      "grad_norm": 0.6674100756645203,
+      "learning_rate": 0.0003184047490555855,
+      "loss": 3.398,
       "step": 43600
     },
     {
-      "epoch": 4.6980949305779784,
-      "grad_norm": 0.699619710445404,
-      "learning_rate": 0.00031856696476672766,
-      "loss": 3.3848,
+      "epoch": 4.706199460916442,
+      "grad_norm": 0.6818583607673645,
+      "learning_rate": 0.00031808094981111706,
+      "loss": 3.4174,
       "step": 43650
     },
     {
-      "epoch": 4.703476482617587,
-      "grad_norm": 0.6185043454170227,
-      "learning_rate": 0.0003182437237366663,
-      "loss": 3.3858,
+      "epoch": 4.711590296495957,
+      "grad_norm": 0.6525809168815613,
+      "learning_rate": 0.00031775715056664867,
+      "loss": 3.4055,
       "step": 43700
     },
     {
-      "epoch": 4.7088580346571955,
-      "grad_norm": 0.6501342058181763,
-      "learning_rate": 0.0003179204827066049,
-      "loss": 3.3658,
+      "epoch": 4.716981132075472,
+      "grad_norm": 0.6633160710334778,
+      "learning_rate": 0.0003174333513221802,
+      "loss": 3.3887,
       "step": 43750
     },
     {
-      "epoch": 4.714239586696803,
-      "grad_norm": 0.6937869787216187,
-      "learning_rate": 0.00031759724167654344,
-      "loss": 3.3866,
+      "epoch": 4.722371967654986,
+      "grad_norm": 0.6640220880508423,
+      "learning_rate": 0.00031710955207771177,
+      "loss": 3.3965,
       "step": 43800
     },
     {
-      "epoch": 4.719621138736412,
-      "grad_norm": 0.6628052592277527,
-      "learning_rate": 0.00031727400064648204,
-      "loss": 3.4006,
+      "epoch": 4.727762803234501,
+      "grad_norm": 0.678041398525238,
+      "learning_rate": 0.0003167857528332434,
+      "loss": 3.4104,
       "step": 43850
     },
     {
-      "epoch": 4.725002690776019,
-      "grad_norm": 0.7351566553115845,
-      "learning_rate": 0.0003169507596164206,
-      "loss": 3.3806,
+      "epoch": 4.7331536388140165,
+      "grad_norm": 0.7227162718772888,
+      "learning_rate": 0.0003164619535887749,
+      "loss": 3.3917,
       "step": 43900
     },
     {
-      "epoch": 4.730384242815628,
-      "grad_norm": 0.6588840484619141,
-      "learning_rate": 0.0003166275185863592,
-      "loss": 3.3738,
+      "epoch": 4.738544474393531,
+      "grad_norm": 0.700898289680481,
+      "learning_rate": 0.00031613815434430653,
+      "loss": 3.402,
       "step": 43950
     },
     {
-      "epoch": 4.7357657948552365,
-      "grad_norm": 0.6494613289833069,
-      "learning_rate": 0.0003163042775562978,
-      "loss": 3.3819,
+      "epoch": 4.743935309973046,
+      "grad_norm": 0.6322097182273865,
+      "learning_rate": 0.0003158143550998381,
+      "loss": 3.4046,
       "step": 44000
     },
     {
-      "epoch": 4.7357657948552365,
-      "eval_accuracy": 0.37949628738514984,
-      "eval_loss": 3.4194657802581787,
-      "eval_runtime": 183.9093,
-      "eval_samples_per_second": 97.934,
-      "eval_steps_per_second": 6.123,
+      "epoch": 4.743935309973046,
+      "eval_accuracy": 0.37918054209316093,
+      "eval_loss": 3.4268388748168945,
+      "eval_runtime": 153.4759,
+      "eval_samples_per_second": 117.354,
+      "eval_steps_per_second": 7.337,
       "step": 44000
     },
     {
-      "epoch": 4.741147346894844,
-      "grad_norm": 0.6664300560951233,
-      "learning_rate": 0.00031598103652623636,
-      "loss": 3.3901,
+      "epoch": 4.74932614555256,
+      "grad_norm": 0.6361574530601501,
+      "learning_rate": 0.0003154905558553697,
+      "loss": 3.4213,
       "step": 44050
     },
     {
-      "epoch": 4.746528898934453,
-      "grad_norm": 0.6783460974693298,
-      "learning_rate": 0.00031565779549617496,
-      "loss": 3.3859,
+      "epoch": 4.754716981132075,
+      "grad_norm": 0.73552006483078,
+      "learning_rate": 0.0003151667566109012,
+      "loss": 3.3915,
       "step": 44100
     },
     {
-      "epoch": 4.751910450974061,
-      "grad_norm": 0.7820475101470947,
-      "learning_rate": 0.00031533455446611355,
-      "loss": 3.3796,
+      "epoch": 4.7601078167115904,
+      "grad_norm": 0.6872468590736389,
+      "learning_rate": 0.00031484295736643284,
+      "loss": 3.3987,
       "step": 44150
     },
     {
-      "epoch": 4.757292003013669,
-      "grad_norm": 0.6787198781967163,
-      "learning_rate": 0.0003150113134360521,
-      "loss": 3.3913,
+      "epoch": 4.765498652291106,
+      "grad_norm": 0.6601293683052063,
+      "learning_rate": 0.00031451915812196434,
+      "loss": 3.3916,
       "step": 44200
     },
     {
-      "epoch": 4.7626735550532775,
-      "grad_norm": 0.7018619775772095,
-      "learning_rate": 0.0003146880724059907,
-      "loss": 3.3918,
+      "epoch": 4.77088948787062,
+      "grad_norm": 0.6650718450546265,
+      "learning_rate": 0.0003142018348623853,
+      "loss": 3.4115,
       "step": 44250
     },
     {
-      "epoch": 4.768055107092886,
-      "grad_norm": 0.6731393337249756,
-      "learning_rate": 0.00031436483137592934,
-      "loss": 3.388,
+      "epoch": 4.776280323450135,
+      "grad_norm": 0.6799063086509705,
+      "learning_rate": 0.00031387803561791684,
+      "loss": 3.3941,
       "step": 44300
     },
     {
-      "epoch": 4.773436659132494,
-      "grad_norm": 0.6883779764175415,
-      "learning_rate": 0.0003140415903458679,
-      "loss": 3.3897,
+      "epoch": 4.781671159029649,
+      "grad_norm": 0.7390704154968262,
+      "learning_rate": 0.00031355423637344844,
+      "loss": 3.3922,
       "step": 44350
     },
     {
-      "epoch": 4.778818211172102,
-      "grad_norm": 0.6423301100730896,
-      "learning_rate": 0.00031371834931580647,
-      "loss": 3.3903,
+      "epoch": 4.787061994609164,
+      "grad_norm": 0.6746432781219482,
+      "learning_rate": 0.00031323043712898,
+      "loss": 3.4003,
       "step": 44400
     },
     {
-      "epoch": 4.78419976321171,
-      "grad_norm": 0.6493849158287048,
-      "learning_rate": 0.000313395108285745,
-      "loss": 3.3851,
+      "epoch": 4.7924528301886795,
+      "grad_norm": 0.6579575538635254,
+      "learning_rate": 0.0003129066378845116,
+      "loss": 3.4131,
       "step": 44450
     },
     {
-      "epoch": 4.7895813152513185,
-      "grad_norm": 0.6674280762672424,
-      "learning_rate": 0.0003130718672556836,
-      "loss": 3.3867,
+      "epoch": 4.797843665768194,
+      "grad_norm": 0.6642597317695618,
+      "learning_rate": 0.00031258283864004315,
+      "loss": 3.4096,
       "step": 44500
     },
     {
-      "epoch": 4.794962867290927,
-      "grad_norm": 0.7001754641532898,
-      "learning_rate": 0.0003127486262256222,
-      "loss": 3.3732,
+      "epoch": 4.803234501347709,
+      "grad_norm": 0.6496700644493103,
+      "learning_rate": 0.0003122590393955747,
+      "loss": 3.3996,
       "step": 44550
     },
     {
-      "epoch": 4.800344419330535,
-      "grad_norm": 0.6968476176261902,
-      "learning_rate": 0.0003124253851955608,
-      "loss": 3.3838,
+      "epoch": 4.808625336927224,
+      "grad_norm": 0.6549248695373535,
+      "learning_rate": 0.0003119352401511063,
+      "loss": 3.4105,
       "step": 44600
     },
     {
-      "epoch": 4.805725971370143,
-      "grad_norm": 0.6995323300361633,
-      "learning_rate": 0.00031210860898610066,
-      "loss": 3.3765,
+      "epoch": 4.814016172506738,
+      "grad_norm": 0.6920045018196106,
+      "learning_rate": 0.00031161144090663786,
+      "loss": 3.4052,
       "step": 44650
     },
     {
-      "epoch": 4.811107523409751,
-      "grad_norm": 0.7006618976593018,
-      "learning_rate": 0.0003117853679560392,
-      "loss": 3.399,
+      "epoch": 4.819407008086253,
+      "grad_norm": 0.6692620515823364,
+      "learning_rate": 0.00031128764166216946,
+      "loss": 3.3948,
       "step": 44700
     },
     {
-      "epoch": 4.8164890754493594,
-      "grad_norm": 0.707897961139679,
-      "learning_rate": 0.0003114621269259778,
-      "loss": 3.3948,
+      "epoch": 4.824797843665769,
+      "grad_norm": 0.6563695669174194,
+      "learning_rate": 0.00031096384241770096,
+      "loss": 3.412,
       "step": 44750
     },
     {
-      "epoch": 4.821870627488968,
-      "grad_norm": 0.6923506259918213,
-      "learning_rate": 0.00031113888589591633,
-      "loss": 3.385,
+      "epoch": 4.830188679245283,
+      "grad_norm": 0.6337430477142334,
+      "learning_rate": 0.0003106400431732326,
+      "loss": 3.4227,
       "step": 44800
     },
     {
-      "epoch": 4.827252179528576,
-      "grad_norm": 0.7260332703590393,
-      "learning_rate": 0.00031081564486585493,
-      "loss": 3.4115,
+      "epoch": 4.835579514824798,
+      "grad_norm": 0.7310241460800171,
+      "learning_rate": 0.0003103162439287641,
+      "loss": 3.4088,
       "step": 44850
     },
     {
-      "epoch": 4.832633731568184,
-      "grad_norm": 0.652719259262085,
-      "learning_rate": 0.0003104924038357935,
-      "loss": 3.3948,
+      "epoch": 4.840970350404312,
+      "grad_norm": 0.6920210123062134,
+      "learning_rate": 0.0003099924446842957,
+      "loss": 3.4162,
       "step": 44900
     },
     {
-      "epoch": 4.838015283607793,
-      "grad_norm": 0.7057687640190125,
-      "learning_rate": 0.0003101691628057321,
-      "loss": 3.374,
+      "epoch": 4.846361185983827,
+      "grad_norm": 0.6893260478973389,
+      "learning_rate": 0.00030966864543982727,
+      "loss": 3.4167,
       "step": 44950
     },
     {
-      "epoch": 4.8433968356474,
-      "grad_norm": 0.7490352392196655,
-      "learning_rate": 0.0003098459217756707,
-      "loss": 3.3833,
+      "epoch": 4.8517520215633425,
+      "grad_norm": 0.6642902493476868,
+      "learning_rate": 0.0003093448461953588,
+      "loss": 3.4087,
       "step": 45000
     },
     {
-      "epoch": 4.8433968356474,
-      "eval_accuracy": 0.3801891668507428,
-      "eval_loss": 3.414224624633789,
-      "eval_runtime": 183.8774,
-      "eval_samples_per_second": 97.951,
-      "eval_steps_per_second": 6.124,
+      "epoch": 4.8517520215633425,
+      "eval_accuracy": 0.37931331592207024,
+      "eval_loss": 3.4195821285247803,
+      "eval_runtime": 153.6635,
+      "eval_samples_per_second": 117.211,
+      "eval_steps_per_second": 7.328,
       "step": 45000
     },
     {
-      "epoch": 4.848778387687009,
-      "grad_norm": 0.6444492936134338,
-      "learning_rate": 0.0003095226807456093,
-      "loss": 3.3875,
+      "epoch": 4.857142857142857,
+      "grad_norm": 0.7197234034538269,
+      "learning_rate": 0.0003090210469508904,
+      "loss": 3.3988,
       "step": 45050
     },
     {
-      "epoch": 4.8541599397266175,
-      "grad_norm": 0.641381561756134,
-      "learning_rate": 0.00030919943971554785,
-      "loss": 3.4057,
+      "epoch": 4.862533692722372,
+      "grad_norm": 0.6641247272491455,
+      "learning_rate": 0.00030869724770642197,
+      "loss": 3.4039,
       "step": 45100
     },
     {
-      "epoch": 4.859541491766225,
-      "grad_norm": 0.7594972848892212,
-      "learning_rate": 0.00030887619868548644,
-      "loss": 3.3809,
+      "epoch": 4.867924528301887,
+      "grad_norm": 0.6879065632820129,
+      "learning_rate": 0.0003083734484619536,
+      "loss": 3.4085,
       "step": 45150
     },
     {
-      "epoch": 4.864923043805834,
-      "grad_norm": 0.7299668788909912,
-      "learning_rate": 0.000308552957655425,
-      "loss": 3.396,
+      "epoch": 4.873315363881401,
+      "grad_norm": 0.6977842450141907,
+      "learning_rate": 0.00030804964921748513,
+      "loss": 3.3884,
       "step": 45200
     },
     {
-      "epoch": 4.870304595845441,
-      "grad_norm": 0.6922218203544617,
-      "learning_rate": 0.00030822971662536363,
-      "loss": 3.3895,
+      "epoch": 4.878706199460916,
+      "grad_norm": 0.6804500222206116,
+      "learning_rate": 0.00030772584997301673,
+      "loss": 3.4001,
       "step": 45250
     },
     {
-      "epoch": 4.87568614788505,
-      "grad_norm": 0.7033401727676392,
-      "learning_rate": 0.0003079064755953022,
-      "loss": 3.3758,
+      "epoch": 4.884097035040432,
+      "grad_norm": 0.6629409790039062,
+      "learning_rate": 0.0003074020507285483,
+      "loss": 3.3857,
       "step": 45300
     },
     {
-      "epoch": 4.8810676999246585,
-      "grad_norm": 0.6653620004653931,
-      "learning_rate": 0.00030758323456524077,
-      "loss": 3.3823,
+      "epoch": 4.889487870619946,
+      "grad_norm": 0.7181710004806519,
+      "learning_rate": 0.0003070782514840799,
+      "loss": 3.3993,
       "step": 45350
     },
     {
-      "epoch": 4.886449251964266,
-      "grad_norm": 0.6495395302772522,
-      "learning_rate": 0.00030725999353517936,
-      "loss": 3.4005,
+      "epoch": 4.894878706199461,
+      "grad_norm": 0.6759134531021118,
+      "learning_rate": 0.00030675445223961144,
+      "loss": 3.4032,
       "step": 45400
     },
     {
-      "epoch": 4.891830804003875,
-      "grad_norm": 0.6559237837791443,
-      "learning_rate": 0.00030693675250511795,
-      "loss": 3.3799,
+      "epoch": 4.900269541778976,
+      "grad_norm": 0.6469165682792664,
+      "learning_rate": 0.00030643065299514294,
+      "loss": 3.403,
       "step": 45450
     },
     {
-      "epoch": 4.897212356043483,
-      "grad_norm": 0.729101300239563,
-      "learning_rate": 0.00030661351147505655,
-      "loss": 3.3926,
+      "epoch": 4.90566037735849,
+      "grad_norm": 0.7133024334907532,
+      "learning_rate": 0.00030610685375067454,
+      "loss": 3.3953,
       "step": 45500
     },
     {
-      "epoch": 4.902593908083091,
-      "grad_norm": 0.6655796766281128,
-      "learning_rate": 0.00030629027044499514,
-      "loss": 3.4006,
+      "epoch": 4.9110512129380055,
+      "grad_norm": 0.6878347396850586,
+      "learning_rate": 0.0003057830545062061,
+      "loss": 3.3815,
       "step": 45550
     },
     {
-      "epoch": 4.9079754601226995,
-      "grad_norm": 0.6939736604690552,
-      "learning_rate": 0.00030596702941493374,
-      "loss": 3.39,
+      "epoch": 4.916442048517521,
+      "grad_norm": 0.6375924944877625,
+      "learning_rate": 0.0003054592552617377,
+      "loss": 3.3909,
       "step": 45600
     },
     {
-      "epoch": 4.913357012162308,
-      "grad_norm": 0.700905978679657,
-      "learning_rate": 0.0003056437883848723,
-      "loss": 3.3907,
+      "epoch": 4.921832884097035,
+      "grad_norm": 0.687326192855835,
+      "learning_rate": 0.00030513545601726925,
+      "loss": 3.3934,
       "step": 45650
     },
     {
-      "epoch": 4.918738564201916,
-      "grad_norm": 0.6739131212234497,
-      "learning_rate": 0.0003053205473548109,
-      "loss": 3.3904,
+      "epoch": 4.92722371967655,
+      "grad_norm": 0.653589129447937,
+      "learning_rate": 0.00030481165677280085,
+      "loss": 3.4051,
       "step": 45700
     },
     {
-      "epoch": 4.924120116241524,
-      "grad_norm": 0.7265480756759644,
-      "learning_rate": 0.0003049973063247494,
-      "loss": 3.3989,
+      "epoch": 4.932614555256064,
+      "grad_norm": 0.6509602069854736,
+      "learning_rate": 0.0003044878575283324,
+      "loss": 3.4027,
       "step": 45750
     },
     {
-      "epoch": 4.929501668281132,
-      "grad_norm": 0.7020339965820312,
-      "learning_rate": 0.00030467406529468806,
-      "loss": 3.3908,
+      "epoch": 4.938005390835579,
+      "grad_norm": 0.7300403118133545,
+      "learning_rate": 0.000304164058283864,
+      "loss": 3.3909,
       "step": 45800
     },
     {
-      "epoch": 4.9348832203207404,
-      "grad_norm": 0.6936817765235901,
-      "learning_rate": 0.00030435082426462666,
-      "loss": 3.3611,
+      "epoch": 4.943396226415095,
+      "grad_norm": 0.7137998938560486,
+      "learning_rate": 0.00030384025903939556,
+      "loss": 3.3971,
       "step": 45850
     },
     {
-      "epoch": 4.940264772360349,
-      "grad_norm": 0.734406590461731,
-      "learning_rate": 0.0003040275832345652,
-      "loss": 3.3782,
+      "epoch": 4.948787061994609,
+      "grad_norm": 0.6470627188682556,
+      "learning_rate": 0.0003035164597949271,
+      "loss": 3.399,
       "step": 45900
     },
     {
-      "epoch": 4.945646324399957,
-      "grad_norm": 0.6767195463180542,
-      "learning_rate": 0.0003037043422045038,
-      "loss": 3.3924,
+      "epoch": 4.954177897574124,
+      "grad_norm": 0.7308759093284607,
+      "learning_rate": 0.0003031926605504587,
+      "loss": 3.3996,
       "step": 45950
     },
     {
-      "epoch": 4.951027876439565,
-      "grad_norm": 0.7091426849365234,
-      "learning_rate": 0.0003033811011744424,
-      "loss": 3.3862,
+      "epoch": 4.959568733153639,
+      "grad_norm": 0.6627731323242188,
+      "learning_rate": 0.00030286886130599026,
+      "loss": 3.3922,
       "step": 46000
     },
     {
-      "epoch": 4.951027876439565,
-      "eval_accuracy": 0.38058357683350935,
-      "eval_loss": 3.4084718227386475,
-      "eval_runtime": 183.6409,
-      "eval_samples_per_second": 98.077,
-      "eval_steps_per_second": 6.132,
+      "epoch": 4.959568733153639,
+      "eval_accuracy": 0.37991644810232844,
+      "eval_loss": 3.415886878967285,
+      "eval_runtime": 152.9186,
+      "eval_samples_per_second": 117.782,
+      "eval_steps_per_second": 7.363,
       "step": 46000
     },
     {
-      "epoch": 4.956409428479174,
-      "grad_norm": 0.7199757695198059,
-      "learning_rate": 0.00030305786014438093,
-      "loss": 3.3799,
+      "epoch": 4.964959568733153,
+      "grad_norm": 0.6494779586791992,
+      "learning_rate": 0.00030254506206152187,
+      "loss": 3.4084,
       "step": 46050
     },
     {
-      "epoch": 4.961790980518781,
-      "grad_norm": 0.6601982116699219,
-      "learning_rate": 0.0003027346191143196,
-      "loss": 3.3868,
+      "epoch": 4.9703504043126685,
+      "grad_norm": 0.6564545035362244,
+      "learning_rate": 0.00030222126281705336,
+      "loss": 3.3899,
       "step": 46100
     },
     {
-      "epoch": 4.96717253255839,
-      "grad_norm": 0.7033658027648926,
-      "learning_rate": 0.00030241137808425817,
-      "loss": 3.3959,
+      "epoch": 4.975741239892184,
+      "grad_norm": 0.702162504196167,
+      "learning_rate": 0.000301897463572585,
+      "loss": 3.4078,
       "step": 46150
     },
     {
-      "epoch": 4.9725540845979985,
-      "grad_norm": 0.7092815637588501,
-      "learning_rate": 0.0003020881370541967,
-      "loss": 3.3805,
+      "epoch": 4.981132075471698,
+      "grad_norm": 0.6896324753761292,
+      "learning_rate": 0.0003015736643281165,
+      "loss": 3.3967,
       "step": 46200
     },
     {
-      "epoch": 4.977935636637606,
-      "grad_norm": 0.6839442849159241,
-      "learning_rate": 0.0003017648960241353,
-      "loss": 3.3813,
+      "epoch": 4.986522911051213,
+      "grad_norm": 0.6490485668182373,
+      "learning_rate": 0.00030124986508364807,
+      "loss": 3.4053,
       "step": 46250
     },
     {
-      "epoch": 4.983317188677215,
-      "grad_norm": 0.6668979525566101,
-      "learning_rate": 0.00030144165499407385,
-      "loss": 3.3951,
+      "epoch": 4.991913746630727,
+      "grad_norm": 0.6790404319763184,
+      "learning_rate": 0.0003009260658391797,
+      "loss": 3.4055,
       "step": 46300
     },
     {
-      "epoch": 4.988698740716822,
-      "grad_norm": 0.6921564936637878,
-      "learning_rate": 0.00030111841396401244,
-      "loss": 3.3757,
+      "epoch": 4.997304582210242,
+      "grad_norm": 0.6946545243263245,
+      "learning_rate": 0.0003006022665947112,
+      "loss": 3.3907,
       "step": 46350
     },
     {
-      "epoch": 4.994080292756431,
-      "grad_norm": 0.7131747007369995,
-      "learning_rate": 0.0003007951729339511,
-      "loss": 3.3993,
+      "epoch": 5.002695417789758,
+      "grad_norm": 0.679398775100708,
+      "learning_rate": 0.00030027846735024283,
+      "loss": 3.3613,
       "step": 46400
     },
     {
-      "epoch": 4.9994618447960395,
-      "grad_norm": 0.6464070677757263,
-      "learning_rate": 0.00030047193190388963,
-      "loss": 3.3913,
+      "epoch": 5.008086253369272,
+      "grad_norm": 0.6553539037704468,
+      "learning_rate": 0.00029996114409066373,
+      "loss": 3.2982,
       "step": 46450
     },
     {
-      "epoch": 5.004843396835647,
-      "grad_norm": 0.7530785799026489,
-      "learning_rate": 0.0003001486908738282,
-      "loss": 3.2879,
+      "epoch": 5.013477088948787,
+      "grad_norm": 0.6750492453575134,
+      "learning_rate": 0.00029963734484619533,
+      "loss": 3.3002,
       "step": 46500
     },
     {
-      "epoch": 5.010224948875256,
-      "grad_norm": 0.6731668710708618,
-      "learning_rate": 0.0002998254498437668,
-      "loss": 3.2943,
+      "epoch": 5.018867924528302,
+      "grad_norm": 0.7658734321594238,
+      "learning_rate": 0.0002993135456017269,
+      "loss": 3.2964,
       "step": 46550
     },
     {
-      "epoch": 5.015606500914864,
-      "grad_norm": 0.7087476849555969,
-      "learning_rate": 0.0002995022088137054,
-      "loss": 3.3026,
+      "epoch": 5.024258760107816,
+      "grad_norm": 0.7131168842315674,
+      "learning_rate": 0.0002989897463572585,
+      "loss": 3.3083,
       "step": 46600
     },
     {
-      "epoch": 5.020988052954472,
-      "grad_norm": 0.6851585507392883,
-      "learning_rate": 0.00029917896778364396,
-      "loss": 3.2938,
+      "epoch": 5.0296495956873315,
+      "grad_norm": 0.6732473969459534,
+      "learning_rate": 0.00029866594711279004,
+      "loss": 3.3067,
       "step": 46650
     },
     {
-      "epoch": 5.0263696049940805,
-      "grad_norm": 0.7182973027229309,
-      "learning_rate": 0.00029885572675358255,
-      "loss": 3.292,
+      "epoch": 5.035040431266847,
+      "grad_norm": 0.6674792766571045,
+      "learning_rate": 0.00029834214786832164,
+      "loss": 3.3222,
       "step": 46700
     },
     {
-      "epoch": 5.031751157033688,
-      "grad_norm": 0.7623227834701538,
-      "learning_rate": 0.00029853248572352114,
-      "loss": 3.3104,
+      "epoch": 5.040431266846361,
+      "grad_norm": 0.7131854295730591,
+      "learning_rate": 0.0002980183486238532,
+      "loss": 3.3109,
       "step": 46750
     },
     {
-      "epoch": 5.037132709073297,
-      "grad_norm": 0.6905185580253601,
-      "learning_rate": 0.00029820924469345974,
-      "loss": 3.2992,
+      "epoch": 5.045822102425876,
+      "grad_norm": 0.6808410286903381,
+      "learning_rate": 0.00029769454937938474,
+      "loss": 3.3282,
       "step": 46800
     },
     {
-      "epoch": 5.042514261112905,
-      "grad_norm": 0.6601005792617798,
-      "learning_rate": 0.0002978860036633983,
-      "loss": 3.3106,
+      "epoch": 5.051212938005391,
+      "grad_norm": 0.7066081166267395,
+      "learning_rate": 0.0002973707501349163,
+      "loss": 3.3079,
       "step": 46850
     },
     {
-      "epoch": 5.047895813152513,
-      "grad_norm": 0.6945598721504211,
-      "learning_rate": 0.00029756922745393814,
-      "loss": 3.315,
+      "epoch": 5.056603773584905,
+      "grad_norm": 0.7031406760215759,
+      "learning_rate": 0.0002970469508904479,
+      "loss": 3.3192,
       "step": 46900
     },
     {
-      "epoch": 5.0532773651921215,
-      "grad_norm": 0.6779568791389465,
-      "learning_rate": 0.00029724598642387674,
-      "loss": 3.3124,
+      "epoch": 5.061994609164421,
+      "grad_norm": 0.6843595504760742,
+      "learning_rate": 0.00029672315164597945,
+      "loss": 3.3268,
       "step": 46950
     },
     {
-      "epoch": 5.05865891723173,
-      "grad_norm": 0.6947134137153625,
-      "learning_rate": 0.0002969227453938153,
-      "loss": 3.3131,
+      "epoch": 5.067385444743936,
+      "grad_norm": 0.7550929188728333,
+      "learning_rate": 0.00029639935240151105,
+      "loss": 3.3178,
       "step": 47000
     },
     {
-      "epoch": 5.05865891723173,
-      "eval_accuracy": 0.38070374692743214,
-      "eval_loss": 3.413435459136963,
-      "eval_runtime": 184.1595,
-      "eval_samples_per_second": 97.801,
-      "eval_steps_per_second": 6.114,
+      "epoch": 5.067385444743936,
+      "eval_accuracy": 0.3798980857642878,
+      "eval_loss": 3.421375274658203,
+      "eval_runtime": 153.4481,
+      "eval_samples_per_second": 117.375,
+      "eval_steps_per_second": 7.338,
       "step": 47000
     },
     {
-      "epoch": 5.064040469271338,
-      "grad_norm": 0.7506875395774841,
-      "learning_rate": 0.00029659950436375387,
-      "loss": 3.3094,
+      "epoch": 5.07277628032345,
+      "grad_norm": 0.6526950597763062,
+      "learning_rate": 0.0002960755531570426,
+      "loss": 3.3205,
       "step": 47050
     },
     {
-      "epoch": 5.069422021310946,
-      "grad_norm": 0.6973223686218262,
-      "learning_rate": 0.00029627626333369246,
-      "loss": 3.3049,
+      "epoch": 5.078167115902965,
+      "grad_norm": 0.7146009802818298,
+      "learning_rate": 0.0002957517539125742,
+      "loss": 3.3223,
       "step": 47100
     },
     {
-      "epoch": 5.074803573350554,
-      "grad_norm": 0.7483325004577637,
-      "learning_rate": 0.00029595302230363106,
-      "loss": 3.3156,
+      "epoch": 5.083557951482479,
+      "grad_norm": 0.7246925830841064,
+      "learning_rate": 0.00029542795466810576,
+      "loss": 3.3116,
       "step": 47150
     },
     {
-      "epoch": 5.080185125390162,
-      "grad_norm": 0.6844714283943176,
-      "learning_rate": 0.0002956297812735696,
-      "loss": 3.2976,
+      "epoch": 5.0889487870619945,
+      "grad_norm": 0.6744373440742493,
+      "learning_rate": 0.0002951041554236373,
+      "loss": 3.3215,
       "step": 47200
     },
     {
-      "epoch": 5.085566677429771,
-      "grad_norm": 0.7447516322135925,
-      "learning_rate": 0.00029530654024350825,
-      "loss": 3.289,
+      "epoch": 5.09433962264151,
+      "grad_norm": 0.66701740026474,
+      "learning_rate": 0.00029478035617916886,
+      "loss": 3.311,
       "step": 47250
     },
     {
-      "epoch": 5.090948229469379,
-      "grad_norm": 0.6593098044395447,
-      "learning_rate": 0.0002949832992134468,
-      "loss": 3.3324,
+      "epoch": 5.099730458221024,
+      "grad_norm": 0.6690517067909241,
+      "learning_rate": 0.00029445655693470047,
+      "loss": 3.3242,
       "step": 47300
     },
     {
-      "epoch": 5.096329781508987,
-      "grad_norm": 0.7182676196098328,
-      "learning_rate": 0.0002946600581833854,
-      "loss": 3.318,
+      "epoch": 5.105121293800539,
+      "grad_norm": 0.6964263916015625,
+      "learning_rate": 0.000294132757690232,
+      "loss": 3.3309,
       "step": 47350
     },
     {
-      "epoch": 5.101711333548596,
-      "grad_norm": 0.7529067993164062,
-      "learning_rate": 0.000294336817153324,
-      "loss": 3.2998,
+      "epoch": 5.110512129380054,
+      "grad_norm": 0.655750572681427,
+      "learning_rate": 0.0002938089584457636,
+      "loss": 3.3394,
       "step": 47400
     },
     {
-      "epoch": 5.107092885588203,
-      "grad_norm": 0.6834985613822937,
-      "learning_rate": 0.0002940135761232626,
-      "loss": 3.2992,
+      "epoch": 5.115902964959568,
+      "grad_norm": 0.7024949193000793,
+      "learning_rate": 0.00029348515920129517,
+      "loss": 3.3273,
       "step": 47450
     },
     {
-      "epoch": 5.112474437627812,
-      "grad_norm": 0.6932069659233093,
-      "learning_rate": 0.0002936903350932011,
-      "loss": 3.311,
+      "epoch": 5.121293800539084,
+      "grad_norm": 0.6735443472862244,
+      "learning_rate": 0.0002931613599568267,
+      "loss": 3.3326,
       "step": 47500
     },
     {
-      "epoch": 5.1178559896674205,
-      "grad_norm": 0.6838076710700989,
-      "learning_rate": 0.0002933670940631397,
-      "loss": 3.3151,
+      "epoch": 5.126684636118599,
+      "grad_norm": 0.699121356010437,
+      "learning_rate": 0.00029283756071235833,
+      "loss": 3.3294,
       "step": 47550
     },
     {
-      "epoch": 5.123237541707028,
-      "grad_norm": 0.7405902147293091,
-      "learning_rate": 0.0002930438530330783,
-      "loss": 3.3102,
+      "epoch": 5.132075471698113,
+      "grad_norm": 0.7013923525810242,
+      "learning_rate": 0.0002925137614678899,
+      "loss": 3.3299,
       "step": 47600
     },
     {
-      "epoch": 5.128619093746637,
-      "grad_norm": 0.7105470299720764,
-      "learning_rate": 0.0002927206120030169,
-      "loss": 3.3144,
+      "epoch": 5.137466307277628,
+      "grad_norm": 0.7405589818954468,
+      "learning_rate": 0.00029218996222342143,
+      "loss": 3.3206,
       "step": 47650
     },
     {
-      "epoch": 5.134000645786244,
-      "grad_norm": 0.6866734623908997,
-      "learning_rate": 0.0002923973709729555,
-      "loss": 3.3115,
+      "epoch": 5.142857142857143,
+      "grad_norm": 0.7323573231697083,
+      "learning_rate": 0.00029186616297895303,
+      "loss": 3.3202,
       "step": 47700
     },
     {
-      "epoch": 5.139382197825853,
-      "grad_norm": 0.6866410970687866,
-      "learning_rate": 0.00029207412994289403,
-      "loss": 3.3254,
+      "epoch": 5.1482479784366575,
+      "grad_norm": 0.6563040018081665,
+      "learning_rate": 0.0002915423637344846,
+      "loss": 3.3177,
       "step": 47750
     },
     {
-      "epoch": 5.1447637498654615,
-      "grad_norm": 0.6954620480537415,
-      "learning_rate": 0.0002917508889128327,
-      "loss": 3.3203,
+      "epoch": 5.153638814016173,
+      "grad_norm": 0.6860113143920898,
+      "learning_rate": 0.00029121856449001613,
+      "loss": 3.3356,
       "step": 47800
     },
     {
-      "epoch": 5.150145301905069,
-      "grad_norm": 0.7614784836769104,
-      "learning_rate": 0.0002914276478827712,
-      "loss": 3.313,
+      "epoch": 5.159029649595688,
+      "grad_norm": 0.7111574411392212,
+      "learning_rate": 0.00029089476524554774,
+      "loss": 3.308,
       "step": 47850
     },
     {
-      "epoch": 5.155526853944678,
-      "grad_norm": 0.6733461618423462,
-      "learning_rate": 0.0002911044068527098,
-      "loss": 3.3156,
+      "epoch": 5.164420485175202,
+      "grad_norm": 0.7348825931549072,
+      "learning_rate": 0.0002905709660010793,
+      "loss": 3.3461,
       "step": 47900
     },
     {
-      "epoch": 5.160908405984286,
-      "grad_norm": 0.7127929925918579,
-      "learning_rate": 0.0002907811658226484,
-      "loss": 3.3352,
+      "epoch": 5.169811320754717,
+      "grad_norm": 0.6920403838157654,
+      "learning_rate": 0.0002902471667566109,
+      "loss": 3.3288,
       "step": 47950
     },
     {
-      "epoch": 5.166289958023894,
-      "grad_norm": 0.7314544320106506,
-      "learning_rate": 0.000290457924792587,
-      "loss": 3.3121,
+      "epoch": 5.175202156334231,
+      "grad_norm": 0.6864628195762634,
+      "learning_rate": 0.00028992336751214245,
+      "loss": 3.3207,
       "step": 48000
     },
     {
-      "epoch": 5.166289958023894,
-      "eval_accuracy": 0.38084075821896624,
-      "eval_loss": 3.4111480712890625,
-      "eval_runtime": 183.9837,
-      "eval_samples_per_second": 97.895,
-      "eval_steps_per_second": 6.12,
+      "epoch": 5.175202156334231,
+      "eval_accuracy": 0.38056999522253254,
+      "eval_loss": 3.415780782699585,
+      "eval_runtime": 153.2292,
+      "eval_samples_per_second": 117.543,
+      "eval_steps_per_second": 7.348,
       "step": 48000
     },
     {
-      "epoch": 5.1716715100635025,
-      "grad_norm": 0.7012684941291809,
-      "learning_rate": 0.00029013468376252555,
-      "loss": 3.2963,
+      "epoch": 5.180592991913747,
+      "grad_norm": 0.6818411350250244,
+      "learning_rate": 0.00028959956826767405,
+      "loss": 3.3342,
       "step": 48050
     },
     {
-      "epoch": 5.17705306210311,
-      "grad_norm": 0.7914406657218933,
-      "learning_rate": 0.00028981144273246414,
-      "loss": 3.3173,
+      "epoch": 5.185983827493262,
+      "grad_norm": 0.6941342353820801,
+      "learning_rate": 0.00028927576902320555,
+      "loss": 3.3245,
       "step": 48100
     },
     {
-      "epoch": 5.182434614142719,
-      "grad_norm": 0.7475384473800659,
-      "learning_rate": 0.00028948820170240274,
-      "loss": 3.3295,
+      "epoch": 5.191374663072776,
+      "grad_norm": 0.7263535857200623,
+      "learning_rate": 0.00028895196977873715,
+      "loss": 3.3369,
       "step": 48150
     },
     {
-      "epoch": 5.187816166182327,
-      "grad_norm": 0.7128666043281555,
-      "learning_rate": 0.00028916496067234133,
-      "loss": 3.3328,
+      "epoch": 5.196765498652291,
+      "grad_norm": 0.6868119835853577,
+      "learning_rate": 0.0002886281705342687,
+      "loss": 3.3372,
       "step": 48200
     },
     {
-      "epoch": 5.193197718221935,
-      "grad_norm": 0.729526162147522,
-      "learning_rate": 0.0002888417196422799,
-      "loss": 3.3085,
+      "epoch": 5.202156334231806,
+      "grad_norm": 0.7449672818183899,
+      "learning_rate": 0.0002883043712898003,
+      "loss": 3.3312,
       "step": 48250
     },
     {
-      "epoch": 5.198579270261543,
-      "grad_norm": 0.7725980281829834,
-      "learning_rate": 0.00028851847861221847,
-      "loss": 3.3311,
+      "epoch": 5.2075471698113205,
+      "grad_norm": 0.7471989393234253,
+      "learning_rate": 0.00028798057204533186,
+      "loss": 3.3307,
       "step": 48300
     },
     {
-      "epoch": 5.203960822301152,
-      "grad_norm": 0.7038251757621765,
-      "learning_rate": 0.00028819523758215706,
-      "loss": 3.2954,
+      "epoch": 5.212938005390836,
+      "grad_norm": 0.6657915115356445,
+      "learning_rate": 0.00028765677280086346,
+      "loss": 3.3325,
       "step": 48350
     },
     {
-      "epoch": 5.20934237434076,
-      "grad_norm": 0.71957927942276,
-      "learning_rate": 0.00028787199655209566,
-      "loss": 3.3161,
+      "epoch": 5.218328840970351,
+      "grad_norm": 0.6448196172714233,
+      "learning_rate": 0.000287332973556395,
+      "loss": 3.3441,
       "step": 48400
     },
     {
-      "epoch": 5.214723926380368,
-      "grad_norm": 0.7028657793998718,
-      "learning_rate": 0.00028754875552203425,
-      "loss": 3.3225,
+      "epoch": 5.223719676549865,
+      "grad_norm": 0.7628692984580994,
+      "learning_rate": 0.0002870091743119266,
+      "loss": 3.3334,
       "step": 48450
     },
     {
-      "epoch": 5.220105478419977,
-      "grad_norm": 0.7714187502861023,
-      "learning_rate": 0.0002872255144919728,
-      "loss": 3.3132,
+      "epoch": 5.22911051212938,
+      "grad_norm": 0.7033239006996155,
+      "learning_rate": 0.00028668537506745817,
+      "loss": 3.3239,
       "step": 48500
     },
     {
-      "epoch": 5.225487030459584,
-      "grad_norm": 0.7329321503639221,
-      "learning_rate": 0.00028690227346191144,
-      "loss": 3.3355,
+      "epoch": 5.234501347708895,
+      "grad_norm": 0.6857210993766785,
+      "learning_rate": 0.0002863615758229897,
+      "loss": 3.3327,
       "step": 48550
     },
     {
-      "epoch": 5.230868582499193,
-      "grad_norm": 0.7220073938369751,
-      "learning_rate": 0.00028657903243185,
-      "loss": 3.312,
+      "epoch": 5.2398921832884096,
+      "grad_norm": 0.6825659871101379,
+      "learning_rate": 0.00028603777657852127,
+      "loss": 3.355,
       "step": 48600
     },
     {
-      "epoch": 5.236250134538801,
-      "grad_norm": 0.7374827861785889,
-      "learning_rate": 0.0002862557914017886,
-      "loss": 3.315,
+      "epoch": 5.245283018867925,
+      "grad_norm": 0.6725428700447083,
+      "learning_rate": 0.0002857139773340529,
+      "loss": 3.3372,
       "step": 48650
     },
     {
-      "epoch": 5.241631686578409,
-      "grad_norm": 0.7140796780586243,
-      "learning_rate": 0.00028593255037172717,
-      "loss": 3.328,
+      "epoch": 5.250673854447439,
+      "grad_norm": 0.7097809910774231,
+      "learning_rate": 0.0002853901780895844,
+      "loss": 3.3378,
       "step": 48700
     },
     {
-      "epoch": 5.247013238618018,
-      "grad_norm": 0.7001251578330994,
-      "learning_rate": 0.00028560930934166576,
-      "loss": 3.3156,
+      "epoch": 5.256064690026954,
+      "grad_norm": 0.7428537607192993,
+      "learning_rate": 0.00028506637884511603,
+      "loss": 3.3436,
       "step": 48750
     },
     {
-      "epoch": 5.252394790657625,
-      "grad_norm": 0.7350326776504517,
-      "learning_rate": 0.00028528606831160436,
-      "loss": 3.3208,
+      "epoch": 5.261455525606469,
+      "grad_norm": 0.6823304891586304,
+      "learning_rate": 0.0002847425796006476,
+      "loss": 3.3318,
       "step": 48800
     },
     {
-      "epoch": 5.257776342697234,
-      "grad_norm": 0.7280195951461792,
-      "learning_rate": 0.0002849628272815429,
-      "loss": 3.3354,
+      "epoch": 5.2668463611859835,
+      "grad_norm": 0.6988386511802673,
+      "learning_rate": 0.00028441878035617913,
+      "loss": 3.3262,
       "step": 48850
     },
     {
-      "epoch": 5.2631578947368425,
-      "grad_norm": 0.7969920635223389,
-      "learning_rate": 0.0002846395862514815,
-      "loss": 3.3245,
+      "epoch": 5.272237196765499,
+      "grad_norm": 0.7257696986198425,
+      "learning_rate": 0.00028409498111171073,
+      "loss": 3.3393,
       "step": 48900
     },
     {
-      "epoch": 5.26853944677645,
-      "grad_norm": 0.7405096888542175,
-      "learning_rate": 0.0002843163452214201,
-      "loss": 3.3069,
+      "epoch": 5.277628032345014,
+      "grad_norm": 0.7293081283569336,
+      "learning_rate": 0.0002837711818672423,
+      "loss": 3.3407,
       "step": 48950
     },
     {
-      "epoch": 5.273920998816059,
-      "grad_norm": 0.7196093797683716,
-      "learning_rate": 0.0002839931041913587,
-      "loss": 3.3511,
+      "epoch": 5.283018867924528,
+      "grad_norm": 0.7140598893165588,
+      "learning_rate": 0.00028344738262277384,
+      "loss": 3.3454,
       "step": 49000
     },
     {
-      "epoch": 5.273920998816059,
-      "eval_accuracy": 0.3811787773529571,
-      "eval_loss": 3.4072370529174805,
-      "eval_runtime": 183.7737,
-      "eval_samples_per_second": 98.006,
-      "eval_steps_per_second": 6.127,
+      "epoch": 5.283018867924528,
+      "eval_accuracy": 0.3809314833802913,
+      "eval_loss": 3.4116175174713135,
+      "eval_runtime": 153.1934,
+      "eval_samples_per_second": 117.57,
+      "eval_steps_per_second": 7.35,
       "step": 49000
     },
     {
-      "epoch": 5.279302550855666,
-      "grad_norm": 0.718604326248169,
-      "learning_rate": 0.0002836698631612972,
-      "loss": 3.3135,
+      "epoch": 5.288409703504043,
+      "grad_norm": 0.7210100293159485,
+      "learning_rate": 0.00028312358337830544,
+      "loss": 3.343,
       "step": 49050
     },
     {
-      "epoch": 5.284684102895275,
-      "grad_norm": 0.6941205859184265,
-      "learning_rate": 0.00028334662213123587,
-      "loss": 3.3297,
+      "epoch": 5.293800539083558,
+      "grad_norm": 0.6977286338806152,
+      "learning_rate": 0.000282799784133837,
+      "loss": 3.356,
       "step": 49100
     },
     {
-      "epoch": 5.2900656549348835,
-      "grad_norm": 0.7106788754463196,
-      "learning_rate": 0.0002830233811011744,
-      "loss": 3.3291,
+      "epoch": 5.2991913746630726,
+      "grad_norm": 0.6902194619178772,
+      "learning_rate": 0.00028247598488936854,
+      "loss": 3.3505,
       "step": 49150
     },
     {
-      "epoch": 5.295447206974491,
-      "grad_norm": 0.730547308921814,
-      "learning_rate": 0.000282700140071113,
-      "loss": 3.3177,
+      "epoch": 5.304582210242588,
+      "grad_norm": 0.7183405756950378,
+      "learning_rate": 0.00028215218564490015,
+      "loss": 3.3579,
       "step": 49200
     },
     {
-      "epoch": 5.3008287590141,
-      "grad_norm": 0.7090842723846436,
-      "learning_rate": 0.0002823768990410516,
-      "loss": 3.3246,
+      "epoch": 5.309973045822103,
+      "grad_norm": 0.7117700576782227,
+      "learning_rate": 0.0002818283864004317,
+      "loss": 3.337,
       "step": 49250
     },
     {
-      "epoch": 5.306210311053708,
-      "grad_norm": 0.7581778168678284,
-      "learning_rate": 0.00028205365801099014,
-      "loss": 3.3492,
+      "epoch": 5.315363881401617,
+      "grad_norm": 0.7029486894607544,
+      "learning_rate": 0.0002815045871559633,
+      "loss": 3.34,
       "step": 49300
     },
     {
-      "epoch": 5.311591863093316,
-      "grad_norm": 0.6873319149017334,
-      "learning_rate": 0.00028173041698092874,
-      "loss": 3.3251,
+      "epoch": 5.320754716981132,
+      "grad_norm": 0.7335899472236633,
+      "learning_rate": 0.00028118078791149485,
+      "loss": 3.341,
       "step": 49350
     },
     {
-      "epoch": 5.316973415132924,
-      "grad_norm": 0.7274459004402161,
-      "learning_rate": 0.00028140717595086733,
-      "loss": 3.318,
+      "epoch": 5.3261455525606465,
+      "grad_norm": 0.6861764192581177,
+      "learning_rate": 0.0002808569886670264,
+      "loss": 3.3348,
       "step": 49400
     },
     {
-      "epoch": 5.322354967172533,
-      "grad_norm": 0.719395101070404,
-      "learning_rate": 0.0002810839349208059,
-      "loss": 3.3157,
+      "epoch": 5.331536388140162,
+      "grad_norm": 0.6637265682220459,
+      "learning_rate": 0.00028053318942255795,
+      "loss": 3.3154,
       "step": 49450
     },
     {
-      "epoch": 5.327736519212141,
-      "grad_norm": 0.6984511017799377,
-      "learning_rate": 0.00028076069389074447,
-      "loss": 3.3224,
+      "epoch": 5.336927223719677,
+      "grad_norm": 0.7047399878501892,
+      "learning_rate": 0.00028020939017808956,
+      "loss": 3.3478,
       "step": 49500
     },
     {
-      "epoch": 5.333118071251749,
-      "grad_norm": 0.7140973210334778,
-      "learning_rate": 0.0002804374528606831,
-      "loss": 3.315,
+      "epoch": 5.342318059299191,
+      "grad_norm": 0.7020964026451111,
+      "learning_rate": 0.0002798855909336211,
+      "loss": 3.3443,
       "step": 49550
     },
     {
-      "epoch": 5.338499623291357,
-      "grad_norm": 0.7510828375816345,
-      "learning_rate": 0.00028011421183062166,
-      "loss": 3.3413,
+      "epoch": 5.347708894878706,
+      "grad_norm": 0.7507948875427246,
+      "learning_rate": 0.0002795617916891527,
+      "loss": 3.3356,
       "step": 49600
     },
     {
-      "epoch": 5.343881175330965,
-      "grad_norm": 0.707831859588623,
-      "learning_rate": 0.00027979097080056025,
-      "loss": 3.3314,
+      "epoch": 5.353099730458221,
+      "grad_norm": 0.6839717030525208,
+      "learning_rate": 0.00027923799244468426,
+      "loss": 3.3403,
       "step": 49650
     },
     {
-      "epoch": 5.349262727370574,
-      "grad_norm": 0.7078862190246582,
-      "learning_rate": 0.00027946772977049885,
-      "loss": 3.3174,
+      "epoch": 5.3584905660377355,
+      "grad_norm": 0.7505576610565186,
+      "learning_rate": 0.00027891419320021587,
+      "loss": 3.3238,
       "step": 49700
     },
     {
-      "epoch": 5.354644279410182,
-      "grad_norm": 0.7447099685668945,
-      "learning_rate": 0.00027914448874043744,
-      "loss": 3.3202,
+      "epoch": 5.363881401617251,
+      "grad_norm": 0.6769701242446899,
+      "learning_rate": 0.0002785903939557474,
+      "loss": 3.3639,
       "step": 49750
     },
     {
-      "epoch": 5.36002583144979,
-      "grad_norm": 0.7227188348770142,
-      "learning_rate": 0.00027882124771037603,
-      "loss": 3.3289,
+      "epoch": 5.369272237196766,
+      "grad_norm": 0.7084974050521851,
+      "learning_rate": 0.00027826659471127897,
+      "loss": 3.3399,
       "step": 49800
     },
     {
-      "epoch": 5.365407383489399,
-      "grad_norm": 0.7393759489059448,
-      "learning_rate": 0.0002784980066803146,
-      "loss": 3.3294,
+      "epoch": 5.37466307277628,
+      "grad_norm": 0.7358812093734741,
+      "learning_rate": 0.0002779427954668105,
+      "loss": 3.3481,
       "step": 49850
     },
     {
-      "epoch": 5.370788935529006,
-      "grad_norm": 0.6780804991722107,
-      "learning_rate": 0.00027817476565025317,
-      "loss": 3.3377,
+      "epoch": 5.380053908355795,
+      "grad_norm": 0.7199405431747437,
+      "learning_rate": 0.0002776189962223421,
+      "loss": 3.3398,
       "step": 49900
     },
     {
-      "epoch": 5.376170487568615,
-      "grad_norm": 0.7282770276069641,
-      "learning_rate": 0.00027785152462019176,
-      "loss": 3.3212,
+      "epoch": 5.38544474393531,
+      "grad_norm": 0.6487457752227783,
+      "learning_rate": 0.0002772951969778737,
+      "loss": 3.3542,
       "step": 49950
     },
     {
-      "epoch": 5.3815520396082235,
-      "grad_norm": 0.7298556566238403,
-      "learning_rate": 0.00027752828359013036,
-      "loss": 3.3288,
+      "epoch": 5.390835579514825,
+      "grad_norm": 0.6717367172241211,
+      "learning_rate": 0.0002769713977334053,
+      "loss": 3.3477,
       "step": 50000
     },
     {
-      "epoch": 5.3815520396082235,
-      "eval_accuracy": 0.38193728316278996,
-      "eval_loss": 3.4022934436798096,
-      "eval_runtime": 183.8008,
-      "eval_samples_per_second": 97.992,
-      "eval_steps_per_second": 6.126,
+      "epoch": 5.390835579514825,
+      "eval_accuracy": 0.3812627660352377,
+      "eval_loss": 3.410015344619751,
+      "eval_runtime": 153.327,
+      "eval_samples_per_second": 117.468,
+      "eval_steps_per_second": 7.344,
       "step": 50000
     },
     {
-      "epoch": 5.386933591647831,
-      "grad_norm": 0.7297439575195312,
-      "learning_rate": 0.0002772050425600689,
-      "loss": 3.333,
+      "epoch": 5.39622641509434,
+      "grad_norm": 0.7287888526916504,
+      "learning_rate": 0.00027664759848893683,
+      "loss": 3.3304,
       "step": 50050
     },
     {
-      "epoch": 5.39231514368744,
-      "grad_norm": 0.6795482039451599,
-      "learning_rate": 0.00027688180153000755,
-      "loss": 3.3283,
+      "epoch": 5.401617250673855,
+      "grad_norm": 0.7144545912742615,
+      "learning_rate": 0.00027632379924446844,
+      "loss": 3.3559,
       "step": 50100
     },
     {
-      "epoch": 5.397696695727047,
-      "grad_norm": 0.739975094795227,
-      "learning_rate": 0.0002765585604999461,
-      "loss": 3.3342,
+      "epoch": 5.407008086253369,
+      "grad_norm": 0.7062088847160339,
+      "learning_rate": 0.000276,
+      "loss": 3.3335,
       "step": 50150
     },
     {
-      "epoch": 5.403078247766656,
-      "grad_norm": 0.6876736283302307,
-      "learning_rate": 0.0002762353194698847,
-      "loss": 3.3451,
+      "epoch": 5.412398921832884,
+      "grad_norm": 0.7320288419723511,
+      "learning_rate": 0.00027567620075553154,
+      "loss": 3.3392,
       "step": 50200
     },
     {
-      "epoch": 5.4084597998062645,
-      "grad_norm": 0.7301447987556458,
-      "learning_rate": 0.0002759120784398233,
-      "loss": 3.3469,
+      "epoch": 5.4177897574123985,
+      "grad_norm": 0.7237269282341003,
+      "learning_rate": 0.0002753524015110631,
+      "loss": 3.3611,
       "step": 50250
     },
     {
-      "epoch": 5.413841351845872,
-      "grad_norm": 0.7510170936584473,
-      "learning_rate": 0.00027558883740976187,
-      "loss": 3.3233,
+      "epoch": 5.423180592991914,
+      "grad_norm": 0.6973865628242493,
+      "learning_rate": 0.0002750286022665947,
+      "loss": 3.3235,
       "step": 50300
     },
     {
-      "epoch": 5.419222903885481,
-      "grad_norm": 0.6457237601280212,
-      "learning_rate": 0.0002752655963797004,
-      "loss": 3.3344,
+      "epoch": 5.428571428571429,
+      "grad_norm": 0.697854220867157,
+      "learning_rate": 0.00027470480302212624,
+      "loss": 3.3396,
       "step": 50350
     },
     {
-      "epoch": 5.424604455925088,
-      "grad_norm": 0.7377134561538696,
-      "learning_rate": 0.000274942355349639,
-      "loss": 3.3385,
+      "epoch": 5.433962264150943,
+      "grad_norm": 0.6931610703468323,
+      "learning_rate": 0.00027438100377765785,
+      "loss": 3.3535,
       "step": 50400
     },
     {
-      "epoch": 5.429986007964697,
-      "grad_norm": 0.6777702569961548,
-      "learning_rate": 0.0002746191143195776,
-      "loss": 3.3399,
+      "epoch": 5.439353099730458,
+      "grad_norm": 0.6895565986633301,
+      "learning_rate": 0.0002740636805180788,
+      "loss": 3.3397,
       "step": 50450
     },
     {
-      "epoch": 5.435367560004305,
-      "grad_norm": 0.7350349426269531,
-      "learning_rate": 0.0002742958732895162,
-      "loss": 3.321,
+      "epoch": 5.444743935309973,
+      "grad_norm": 0.7018980979919434,
+      "learning_rate": 0.00027373988127361035,
+      "loss": 3.3515,
       "step": 50500
     },
     {
-      "epoch": 5.440749112043913,
-      "grad_norm": 0.6970446109771729,
-      "learning_rate": 0.0002739726322594548,
-      "loss": 3.3101,
+      "epoch": 5.450134770889488,
+      "grad_norm": 0.7320551872253418,
+      "learning_rate": 0.0002734160820291419,
+      "loss": 3.3494,
       "step": 50550
     },
     {
-      "epoch": 5.446130664083522,
-      "grad_norm": 0.7606780529022217,
-      "learning_rate": 0.00027364939122939333,
-      "loss": 3.3383,
+      "epoch": 5.455525606469003,
+      "grad_norm": 0.6875684261322021,
+      "learning_rate": 0.00027309228278467345,
+      "loss": 3.336,
       "step": 50600
     },
     {
-      "epoch": 5.45151221612313,
-      "grad_norm": 0.6950878500938416,
-      "learning_rate": 0.0002733261501993319,
-      "loss": 3.3227,
+      "epoch": 5.460916442048518,
+      "grad_norm": 0.696300745010376,
+      "learning_rate": 0.00027276848354020506,
+      "loss": 3.3484,
       "step": 50650
     },
     {
-      "epoch": 5.456893768162738,
-      "grad_norm": 0.7689592838287354,
-      "learning_rate": 0.0002730029091692705,
-      "loss": 3.3313,
+      "epoch": 5.466307277628032,
+      "grad_norm": 0.6981241703033447,
+      "learning_rate": 0.0002724446842957366,
+      "loss": 3.3623,
       "step": 50700
     },
     {
-      "epoch": 5.462275320202346,
-      "grad_norm": 0.7425930500030518,
-      "learning_rate": 0.0002726796681392091,
-      "loss": 3.3305,
+      "epoch": 5.471698113207547,
+      "grad_norm": 0.7438769936561584,
+      "learning_rate": 0.00027212736103615756,
+      "loss": 3.3679,
       "step": 50750
     },
     {
-      "epoch": 5.467656872241955,
-      "grad_norm": 0.7347174286842346,
-      "learning_rate": 0.0002723564271091477,
-      "loss": 3.3144,
+      "epoch": 5.4770889487870615,
+      "grad_norm": 0.7072102427482605,
+      "learning_rate": 0.00027180356179168916,
+      "loss": 3.3473,
       "step": 50800
     },
     {
-      "epoch": 5.473038424281563,
-      "grad_norm": 0.6965786218643188,
-      "learning_rate": 0.0002720331860790863,
-      "loss": 3.3344,
+      "epoch": 5.482479784366577,
+      "grad_norm": 0.6954070925712585,
+      "learning_rate": 0.0002714797625472207,
+      "loss": 3.3479,
       "step": 50850
     },
     {
-      "epoch": 5.478419976321171,
-      "grad_norm": 0.6878877878189087,
-      "learning_rate": 0.00027170994504902485,
-      "loss": 3.3463,
+      "epoch": 5.487870619946092,
+      "grad_norm": 0.697226881980896,
+      "learning_rate": 0.00027115596330275226,
+      "loss": 3.36,
       "step": 50900
     },
     {
-      "epoch": 5.483801528360779,
-      "grad_norm": 0.6694966554641724,
-      "learning_rate": 0.00027138670401896344,
-      "loss": 3.3245,
+      "epoch": 5.493261455525606,
+      "grad_norm": 0.6898880004882812,
+      "learning_rate": 0.0002708321640582838,
+      "loss": 3.3476,
       "step": 50950
     },
     {
-      "epoch": 5.489183080400387,
-      "grad_norm": 0.6729331016540527,
-      "learning_rate": 0.00027106346298890204,
-      "loss": 3.3275,
+      "epoch": 5.498652291105121,
+      "grad_norm": 0.6771557927131653,
+      "learning_rate": 0.0002705083648138154,
+      "loss": 3.3489,
       "step": 51000
     },
     {
-      "epoch": 5.489183080400387,
-      "eval_accuracy": 0.3819804183592523,
-      "eval_loss": 3.398364305496216,
-      "eval_runtime": 183.9315,
-      "eval_samples_per_second": 97.922,
-      "eval_steps_per_second": 6.122,
+      "epoch": 5.498652291105121,
+      "eval_accuracy": 0.3818572059844707,
+      "eval_loss": 3.4045839309692383,
+      "eval_runtime": 153.3785,
+      "eval_samples_per_second": 117.428,
+      "eval_steps_per_second": 7.341,
       "step": 51000
     },
     {
-      "epoch": 5.494564632439996,
-      "grad_norm": 0.7123387455940247,
-      "learning_rate": 0.00027074022195884063,
-      "loss": 3.3206,
+      "epoch": 5.504043126684636,
+      "grad_norm": 0.7411109209060669,
+      "learning_rate": 0.00027018456556934697,
+      "loss": 3.3397,
       "step": 51050
     },
     {
-      "epoch": 5.499946184479604,
-      "grad_norm": 0.6952930092811584,
-      "learning_rate": 0.0002704169809287792,
-      "loss": 3.3254,
+      "epoch": 5.509433962264151,
+      "grad_norm": 0.6906704902648926,
+      "learning_rate": 0.0002698607663248786,
+      "loss": 3.3323,
       "step": 51100
     },
     {
-      "epoch": 5.505327736519212,
-      "grad_norm": 0.69454425573349,
-      "learning_rate": 0.00027009373989871776,
-      "loss": 3.3371,
+      "epoch": 5.514824797843666,
+      "grad_norm": 0.7282754182815552,
+      "learning_rate": 0.0002695369670804101,
+      "loss": 3.3322,
       "step": 51150
     },
     {
-      "epoch": 5.510709288558821,
-      "grad_norm": 0.7247110605239868,
-      "learning_rate": 0.00026977049886865636,
-      "loss": 3.3252,
+      "epoch": 5.520215633423181,
+      "grad_norm": 0.7394979596138,
+      "learning_rate": 0.0002692131678359417,
+      "loss": 3.3337,
       "step": 51200
     },
     {
-      "epoch": 5.516090840598428,
-      "grad_norm": 0.6885068416595459,
-      "learning_rate": 0.00026944725783859495,
-      "loss": 3.3338,
+      "epoch": 5.525606469002695,
+      "grad_norm": 0.7136691212654114,
+      "learning_rate": 0.0002688893685914733,
+      "loss": 3.3614,
       "step": 51250
     },
     {
-      "epoch": 5.521472392638037,
-      "grad_norm": 0.7206288576126099,
-      "learning_rate": 0.00026912401680853355,
-      "loss": 3.3419,
+      "epoch": 5.53099730458221,
+      "grad_norm": 0.7183184027671814,
+      "learning_rate": 0.00026856556934700483,
+      "loss": 3.3535,
       "step": 51300
     },
     {
-      "epoch": 5.5268539446776455,
-      "grad_norm": 0.7450361847877502,
-      "learning_rate": 0.0002688007757784721,
-      "loss": 3.3212,
+      "epoch": 5.536388140161725,
+      "grad_norm": 0.7160298824310303,
+      "learning_rate": 0.0002682417701025364,
+      "loss": 3.3543,
       "step": 51350
     },
     {
-      "epoch": 5.532235496717253,
-      "grad_norm": 0.7636796236038208,
-      "learning_rate": 0.00026847753474841074,
-      "loss": 3.3248,
+      "epoch": 5.54177897574124,
+      "grad_norm": 0.706554651260376,
+      "learning_rate": 0.000267917970858068,
+      "loss": 3.3302,
       "step": 51400
     },
     {
-      "epoch": 5.537617048756862,
-      "grad_norm": Infinity,
-      "learning_rate": 0.0002681607585389505,
-      "loss": 3.3149,
+      "epoch": 5.547169811320755,
+      "grad_norm": 0.680901825428009,
+      "learning_rate": 0.00026759417161359954,
+      "loss": 3.3353,
       "step": 51450
     },
     {
-      "epoch": 5.542998600796469,
-      "grad_norm": 0.7319815754890442,
-      "learning_rate": 0.0002678375175088891,
-      "loss": 3.3363,
+      "epoch": 5.55256064690027,
+      "grad_norm": 0.7205814123153687,
+      "learning_rate": 0.0002672703723691311,
+      "loss": 3.3771,
       "step": 51500
     },
     {
-      "epoch": 5.548380152836078,
-      "grad_norm": 0.7388057708740234,
-      "learning_rate": 0.0002675142764788277,
-      "loss": 3.3453,
+      "epoch": 5.557951482479784,
+      "grad_norm": 0.6591745615005493,
+      "learning_rate": 0.0002669465731246627,
+      "loss": 3.3358,
       "step": 51550
     },
     {
-      "epoch": 5.553761704875686,
-      "grad_norm": 0.7642236351966858,
-      "learning_rate": 0.0002671910354487663,
-      "loss": 3.3118,
+      "epoch": 5.563342318059299,
+      "grad_norm": 0.6978381872177124,
+      "learning_rate": 0.00026662277388019424,
+      "loss": 3.3521,
       "step": 51600
     },
     {
-      "epoch": 5.559143256915294,
-      "grad_norm": 0.710657000541687,
-      "learning_rate": 0.00026686779441870487,
-      "loss": 3.3403,
+      "epoch": 5.568733153638814,
+      "grad_norm": 0.6810121536254883,
+      "learning_rate": 0.00026629897463572585,
+      "loss": 3.3365,
       "step": 51650
     },
     {
-      "epoch": 5.564524808954903,
-      "grad_norm": 0.7277242541313171,
-      "learning_rate": 0.0002665445533886434,
-      "loss": 3.3367,
+      "epoch": 5.574123989218329,
+      "grad_norm": 0.744707465171814,
+      "learning_rate": 0.0002659751753912574,
+      "loss": 3.3401,
       "step": 51700
     },
     {
-      "epoch": 5.569906360994511,
-      "grad_norm": 0.7145871520042419,
-      "learning_rate": 0.00026622131235858206,
-      "loss": 3.3262,
+      "epoch": 5.579514824797844,
+      "grad_norm": 0.6696263551712036,
+      "learning_rate": 0.00026565137614678895,
+      "loss": 3.3544,
       "step": 51750
     },
     {
-      "epoch": 5.575287913034119,
-      "grad_norm": 0.6839075088500977,
-      "learning_rate": 0.0002658980713285206,
-      "loss": 3.3251,
+      "epoch": 5.584905660377358,
+      "grad_norm": 0.6839249134063721,
+      "learning_rate": 0.0002653275769023205,
+      "loss": 3.3267,
       "step": 51800
     },
     {
-      "epoch": 5.580669465073727,
-      "grad_norm": 0.730114758014679,
-      "learning_rate": 0.0002655748302984592,
-      "loss": 3.333,
+      "epoch": 5.590296495956873,
+      "grad_norm": 0.712758481502533,
+      "learning_rate": 0.0002650037776578521,
+      "loss": 3.3266,
       "step": 51850
     },
     {
-      "epoch": 5.586051017113336,
-      "grad_norm": 0.7460644841194153,
-      "learning_rate": 0.0002652515892683978,
-      "loss": 3.3323,
+      "epoch": 5.595687331536388,
+      "grad_norm": 0.6807816028594971,
+      "learning_rate": 0.00026467997841338366,
+      "loss": 3.3785,
       "step": 51900
     },
     {
-      "epoch": 5.591432569152944,
-      "grad_norm": 0.7102095484733582,
-      "learning_rate": 0.0002649283482383364,
-      "loss": 3.3091,
+      "epoch": 5.601078167115903,
+      "grad_norm": 0.7055612802505493,
+      "learning_rate": 0.00026435617916891526,
+      "loss": 3.3434,
       "step": 51950
     },
     {
-      "epoch": 5.596814121192552,
-      "grad_norm": 0.7392861247062683,
-      "learning_rate": 0.0002646051072082749,
-      "loss": 3.3413,
+      "epoch": 5.606469002695418,
+      "grad_norm": 0.6885956525802612,
+      "learning_rate": 0.0002640323799244468,
+      "loss": 3.3582,
       "step": 52000
     },
     {
-      "epoch": 5.596814121192552,
-      "eval_accuracy": 0.38263146646303664,
-      "eval_loss": 3.3929498195648193,
-      "eval_runtime": 183.8249,
-      "eval_samples_per_second": 97.979,
-      "eval_steps_per_second": 6.125,
+      "epoch": 5.606469002695418,
+      "eval_accuracy": 0.38233397485620063,
+      "eval_loss": 3.4000039100646973,
+      "eval_runtime": 153.3166,
+      "eval_samples_per_second": 117.476,
+      "eval_steps_per_second": 7.344,
       "step": 52000
     },
     {
-      "epoch": 5.60219567323216,
-      "grad_norm": 0.7184988856315613,
-      "learning_rate": 0.0002642818661782135,
-      "loss": 3.3487,
+      "epoch": 5.611859838274933,
+      "grad_norm": 0.7244799137115479,
+      "learning_rate": 0.0002637085806799784,
+      "loss": 3.3321,
       "step": 52050
     },
     {
-      "epoch": 5.607577225271768,
-      "grad_norm": 0.7137402296066284,
-      "learning_rate": 0.0002639586251481521,
-      "loss": 3.3352,
+      "epoch": 5.617250673854447,
+      "grad_norm": 0.6627570390701294,
+      "learning_rate": 0.00026338478143550997,
+      "loss": 3.3552,
       "step": 52100
     },
     {
-      "epoch": 5.612958777311377,
-      "grad_norm": 0.7470986247062683,
-      "learning_rate": 0.0002636353841180907,
-      "loss": 3.3295,
+      "epoch": 5.622641509433962,
+      "grad_norm": 0.7210434079170227,
+      "learning_rate": 0.00026306098219104157,
+      "loss": 3.3257,
       "step": 52150
     },
     {
-      "epoch": 5.618340329350985,
-      "grad_norm": 0.7585472464561462,
-      "learning_rate": 0.0002633121430880293,
-      "loss": 3.3312,
+      "epoch": 5.628032345013477,
+      "grad_norm": 0.7402634024620056,
+      "learning_rate": 0.00026273718294657307,
+      "loss": 3.3258,
       "step": 52200
     },
     {
-      "epoch": 5.623721881390593,
-      "grad_norm": 0.7137346863746643,
-      "learning_rate": 0.00026298890205796784,
-      "loss": 3.3217,
+      "epoch": 5.633423180592992,
+      "grad_norm": 0.7102649807929993,
+      "learning_rate": 0.00026241338370210467,
+      "loss": 3.3439,
       "step": 52250
     },
     {
-      "epoch": 5.629103433430201,
-      "grad_norm": 0.7068004608154297,
-      "learning_rate": 0.00026266566102790644,
-      "loss": 3.3382,
+      "epoch": 5.638814016172507,
+      "grad_norm": 0.7167055606842041,
+      "learning_rate": 0.0002620895844576362,
+      "loss": 3.3612,
       "step": 52300
     },
     {
-      "epoch": 5.634484985469809,
-      "grad_norm": 0.7711502909660339,
-      "learning_rate": 0.00026234241999784503,
-      "loss": 3.3433,
+      "epoch": 5.644204851752022,
+      "grad_norm": 0.6946225166320801,
+      "learning_rate": 0.00026176578521316783,
+      "loss": 3.3291,
       "step": 52350
     },
     {
-      "epoch": 5.639866537509418,
-      "grad_norm": 0.7147432565689087,
-      "learning_rate": 0.0002620191789677836,
-      "loss": 3.3363,
+      "epoch": 5.649595687331536,
+      "grad_norm": 0.6670322418212891,
+      "learning_rate": 0.0002614419859686994,
+      "loss": 3.3259,
       "step": 52400
     },
     {
-      "epoch": 5.645248089549026,
-      "grad_norm": 0.6968666315078735,
-      "learning_rate": 0.00026169593793772217,
-      "loss": 3.3578,
+      "epoch": 5.654986522911051,
+      "grad_norm": 0.8273860216140747,
+      "learning_rate": 0.000261118186724231,
+      "loss": 3.336,
       "step": 52450
     },
     {
-      "epoch": 5.650629641588634,
-      "grad_norm": 0.7339818477630615,
-      "learning_rate": 0.0002613726969076608,
-      "loss": 3.3169,
+      "epoch": 5.660377358490566,
+      "grad_norm": 0.7513616681098938,
+      "learning_rate": 0.00026079438747976253,
+      "loss": 3.3527,
       "step": 52500
     },
     {
-      "epoch": 5.656011193628243,
-      "grad_norm": 0.7108085751533508,
-      "learning_rate": 0.00026104945587759936,
-      "loss": 3.3453,
+      "epoch": 5.665768194070081,
+      "grad_norm": 0.6907859444618225,
+      "learning_rate": 0.0002604705882352941,
+      "loss": 3.3436,
       "step": 52550
     },
     {
-      "epoch": 5.66139274566785,
-      "grad_norm": 0.7057158350944519,
-      "learning_rate": 0.00026072621484753795,
-      "loss": 3.3161,
+      "epoch": 5.671159029649596,
+      "grad_norm": 0.7823171615600586,
+      "learning_rate": 0.00026014678899082563,
+      "loss": 3.3461,
       "step": 52600
     },
     {
-      "epoch": 5.666774297707459,
-      "grad_norm": 0.7036101222038269,
-      "learning_rate": 0.00026040297381747655,
-      "loss": 3.3225,
+      "epoch": 5.67654986522911,
+      "grad_norm": 0.7309443950653076,
+      "learning_rate": 0.00025982298974635724,
+      "loss": 3.3597,
       "step": 52650
     },
     {
-      "epoch": 5.672155849747067,
-      "grad_norm": 0.7487311363220215,
-      "learning_rate": 0.00026007973278741514,
-      "loss": 3.3361,
+      "epoch": 5.681940700808625,
+      "grad_norm": 0.7320877313613892,
+      "learning_rate": 0.0002594991905018888,
+      "loss": 3.3605,
       "step": 52700
     },
     {
-      "epoch": 5.677537401786675,
-      "grad_norm": 0.7337835431098938,
-      "learning_rate": 0.00025975649175735373,
-      "loss": 3.3436,
+      "epoch": 5.6873315363881405,
+      "grad_norm": 0.7658337354660034,
+      "learning_rate": 0.0002591753912574204,
+      "loss": 3.3537,
       "step": 52750
     },
     {
-      "epoch": 5.682918953826284,
-      "grad_norm": 0.7318044900894165,
-      "learning_rate": 0.0002594332507272923,
-      "loss": 3.322,
+      "epoch": 5.692722371967655,
+      "grad_norm": 0.7092781662940979,
+      "learning_rate": 0.00025885159201295195,
+      "loss": 3.3318,
       "step": 52800
     },
     {
-      "epoch": 5.688300505865891,
-      "grad_norm": 0.7219624519348145,
-      "learning_rate": 0.00025911000969723087,
-      "loss": 3.3476,
+      "epoch": 5.69811320754717,
+      "grad_norm": 0.7133784294128418,
+      "learning_rate": 0.0002585277927684835,
+      "loss": 3.3384,
       "step": 52850
     },
     {
-      "epoch": 5.6936820579055,
-      "grad_norm": 0.7013553977012634,
-      "learning_rate": 0.00025878676866716946,
-      "loss": 3.323,
+      "epoch": 5.703504043126685,
+      "grad_norm": 0.701946496963501,
+      "learning_rate": 0.0002582039935240151,
+      "loss": 3.3591,
       "step": 52900
     },
     {
-      "epoch": 5.699063609945108,
-      "grad_norm": 0.684723436832428,
-      "learning_rate": 0.00025846352763710806,
-      "loss": 3.3382,
+      "epoch": 5.708894878706199,
+      "grad_norm": 0.7318477034568787,
+      "learning_rate": 0.00025788019427954665,
+      "loss": 3.3551,
       "step": 52950
     },
     {
-      "epoch": 5.704445161984716,
-      "grad_norm": 0.6875833868980408,
-      "learning_rate": 0.0002581402866070466,
-      "loss": 3.3336,
+      "epoch": 5.714285714285714,
+      "grad_norm": 0.6669468283653259,
+      "learning_rate": 0.00025755639503507826,
+      "loss": 3.3495,
       "step": 53000
     },
     {
-      "epoch": 5.704445161984716,
-      "eval_accuracy": 0.38308096345992515,
-      "eval_loss": 3.389054298400879,
-      "eval_runtime": 183.7986,
-      "eval_samples_per_second": 97.993,
-      "eval_steps_per_second": 6.126,
+      "epoch": 5.714285714285714,
+      "eval_accuracy": 0.38246772656110023,
+      "eval_loss": 3.39532470703125,
+      "eval_runtime": 153.3515,
+      "eval_samples_per_second": 117.449,
+      "eval_steps_per_second": 7.343,
       "step": 53000
     },
     {
-      "epoch": 5.709826714024325,
-      "grad_norm": 0.6966539621353149,
-      "learning_rate": 0.00025781704557698525,
-      "loss": 3.3223,
+      "epoch": 5.719676549865229,
+      "grad_norm": 0.6891975402832031,
+      "learning_rate": 0.0002572325957906098,
+      "loss": 3.3322,
       "step": 53050
     },
     {
-      "epoch": 5.715208266063933,
-      "grad_norm": 0.7413732409477234,
-      "learning_rate": 0.0002574938045469238,
-      "loss": 3.3276,
+      "epoch": 5.725067385444744,
+      "grad_norm": 0.707899808883667,
+      "learning_rate": 0.00025690879654614136,
+      "loss": 3.3557,
       "step": 53100
     },
     {
-      "epoch": 5.720589818103541,
-      "grad_norm": 0.692258894443512,
-      "learning_rate": 0.0002571705635168624,
-      "loss": 3.3501,
+      "epoch": 5.730458221024259,
+      "grad_norm": 0.7230016589164734,
+      "learning_rate": 0.0002565849973016729,
+      "loss": 3.3529,
       "step": 53150
     },
     {
-      "epoch": 5.725971370143149,
-      "grad_norm": 0.7227716445922852,
-      "learning_rate": 0.000256847322486801,
-      "loss": 3.3343,
+      "epoch": 5.735849056603773,
+      "grad_norm": 0.721045970916748,
+      "learning_rate": 0.0002562611980572045,
+      "loss": 3.3504,
       "step": 53200
     },
     {
-      "epoch": 5.731352922182758,
-      "grad_norm": 0.7071229815483093,
-      "learning_rate": 0.0002565240814567396,
-      "loss": 3.3293,
+      "epoch": 5.741239892183288,
+      "grad_norm": 0.7427518367767334,
+      "learning_rate": 0.00025593739881273606,
+      "loss": 3.3423,
       "step": 53250
     },
     {
-      "epoch": 5.736734474222366,
-      "grad_norm": 0.7656566500663757,
-      "learning_rate": 0.0002562008404266781,
-      "loss": 3.3302,
+      "epoch": 5.7466307277628035,
+      "grad_norm": 0.69859778881073,
+      "learning_rate": 0.00025561359956826767,
+      "loss": 3.3649,
       "step": 53300
     },
     {
-      "epoch": 5.742116026261974,
-      "grad_norm": 0.7478346824645996,
-      "learning_rate": 0.0002558775993966167,
-      "loss": 3.3538,
+      "epoch": 5.752021563342318,
+      "grad_norm": 0.6876549124717712,
+      "learning_rate": 0.0002552898003237992,
+      "loss": 3.3288,
       "step": 53350
     },
     {
-      "epoch": 5.747497578301582,
-      "grad_norm": 0.7838046550750732,
-      "learning_rate": 0.0002555543583665553,
-      "loss": 3.3278,
+      "epoch": 5.757412398921833,
+      "grad_norm": 0.6735378503799438,
+      "learning_rate": 0.0002549660010793308,
+      "loss": 3.3493,
       "step": 53400
     },
     {
-      "epoch": 5.75287913034119,
-      "grad_norm": 0.6677360534667969,
-      "learning_rate": 0.0002552311173364939,
-      "loss": 3.3324,
+      "epoch": 5.762803234501348,
+      "grad_norm": 0.7670485973358154,
+      "learning_rate": 0.0002546422018348624,
+      "loss": 3.3376,
       "step": 53450
     },
     {
-      "epoch": 5.758260682380799,
-      "grad_norm": 0.7196292877197266,
-      "learning_rate": 0.0002549078763064325,
-      "loss": 3.3446,
+      "epoch": 5.768194070080862,
+      "grad_norm": 0.7104763388633728,
+      "learning_rate": 0.0002543184025903939,
+      "loss": 3.3515,
       "step": 53500
     },
     {
-      "epoch": 5.763642234420407,
-      "grad_norm": 0.7231963276863098,
-      "learning_rate": 0.0002545911000969723,
-      "loss": 3.3411,
+      "epoch": 5.773584905660377,
+      "grad_norm": 0.6887285113334656,
+      "learning_rate": 0.0002539946033459255,
+      "loss": 3.3348,
       "step": 53550
     },
     {
-      "epoch": 5.769023786460015,
-      "grad_norm": 0.705787181854248,
-      "learning_rate": 0.0002542678590669109,
-      "loss": 3.3174,
+      "epoch": 5.7789757412398925,
+      "grad_norm": 0.7925183176994324,
+      "learning_rate": 0.0002536708041014571,
+      "loss": 3.3319,
       "step": 53600
     },
     {
-      "epoch": 5.774405338499624,
-      "grad_norm": 0.6919664740562439,
-      "learning_rate": 0.00025394461803684943,
-      "loss": 3.3319,
+      "epoch": 5.784366576819407,
+      "grad_norm": 0.721466600894928,
+      "learning_rate": 0.00025334700485698863,
+      "loss": 3.3286,
       "step": 53650
     },
     {
-      "epoch": 5.779786890539231,
-      "grad_norm": 0.734150767326355,
-      "learning_rate": 0.00025362137700678803,
-      "loss": 3.3385,
+      "epoch": 5.789757412398922,
+      "grad_norm": 0.734126091003418,
+      "learning_rate": 0.00025302320561252023,
+      "loss": 3.3406,
       "step": 53700
     },
     {
-      "epoch": 5.78516844257884,
-      "grad_norm": 0.7613967061042786,
-      "learning_rate": 0.0002532981359767266,
-      "loss": 3.3387,
+      "epoch": 5.795148247978437,
+      "grad_norm": 0.6946778893470764,
+      "learning_rate": 0.0002526994063680518,
+      "loss": 3.3467,
       "step": 53750
     },
     {
-      "epoch": 5.790549994618448,
-      "grad_norm": 0.6935803294181824,
-      "learning_rate": 0.0002529748949466652,
-      "loss": 3.3416,
+      "epoch": 5.800539083557951,
+      "grad_norm": 0.7649127840995789,
+      "learning_rate": 0.0002523756071235834,
+      "loss": 3.3546,
       "step": 53800
     },
     {
-      "epoch": 5.795931546658056,
-      "grad_norm": 0.6819015741348267,
-      "learning_rate": 0.0002526516539166038,
-      "loss": 3.3241,
+      "epoch": 5.8059299191374665,
+      "grad_norm": 0.6551430225372314,
+      "learning_rate": 0.00025205180787911494,
+      "loss": 3.3392,
       "step": 53850
     },
     {
-      "epoch": 5.801313098697665,
-      "grad_norm": 0.7101120352745056,
-      "learning_rate": 0.00025232841288654235,
-      "loss": 3.3491,
+      "epoch": 5.811320754716981,
+      "grad_norm": 0.6965331435203552,
+      "learning_rate": 0.0002517280086346465,
+      "loss": 3.3416,
       "step": 53900
     },
     {
-      "epoch": 5.806694650737272,
-      "grad_norm": 0.6915525197982788,
-      "learning_rate": 0.00025200517185648095,
-      "loss": 3.3275,
+      "epoch": 5.816711590296496,
+      "grad_norm": 0.6845802068710327,
+      "learning_rate": 0.00025140420939017804,
+      "loss": 3.3422,
       "step": 53950
     },
     {
-      "epoch": 5.812076202776881,
-      "grad_norm": 0.7256600856781006,
-      "learning_rate": 0.00025168193082641954,
-      "loss": 3.3563,
+      "epoch": 5.822102425876011,
+      "grad_norm": 0.657482922077179,
+      "learning_rate": 0.00025108041014570965,
+      "loss": 3.3373,
       "step": 54000
     },
     {
-      "epoch": 5.812076202776881,
-      "eval_accuracy": 0.3837210376220403,
-      "eval_loss": 3.3823492527008057,
-      "eval_runtime": 183.7717,
-      "eval_samples_per_second": 98.007,
-      "eval_steps_per_second": 6.127,
+      "epoch": 5.822102425876011,
+      "eval_accuracy": 0.38292765423521896,
+      "eval_loss": 3.3896946907043457,
+      "eval_runtime": 153.3975,
+      "eval_samples_per_second": 117.414,
+      "eval_steps_per_second": 7.34,
       "step": 54000
     },
     {
-      "epoch": 5.817457754816489,
-      "grad_norm": 0.731121838092804,
-      "learning_rate": 0.00025135868979635814,
-      "loss": 3.3254,
+      "epoch": 5.827493261455525,
+      "grad_norm": 0.6870075464248657,
+      "learning_rate": 0.0002507566109012412,
+      "loss": 3.357,
       "step": 54050
     },
     {
-      "epoch": 5.822839306856097,
-      "grad_norm": 0.6886929273605347,
-      "learning_rate": 0.0002510354487662967,
-      "loss": 3.3291,
+      "epoch": 5.83288409703504,
+      "grad_norm": 0.8188000321388245,
+      "learning_rate": 0.0002504328116567728,
+      "loss": 3.3418,
       "step": 54100
     },
     {
-      "epoch": 5.828220858895706,
-      "grad_norm": 0.739599883556366,
-      "learning_rate": 0.0002507122077362353,
-      "loss": 3.3272,
+      "epoch": 5.8382749326145555,
+      "grad_norm": 0.7052551507949829,
+      "learning_rate": 0.00025010901241230435,
+      "loss": 3.3633,
       "step": 54150
     },
     {
-      "epoch": 5.833602410935313,
-      "grad_norm": 0.775676429271698,
-      "learning_rate": 0.00025038896670617387,
-      "loss": 3.3317,
+      "epoch": 5.84366576819407,
+      "grad_norm": 0.7384065389633179,
+      "learning_rate": 0.0002497852131678359,
+      "loss": 3.3659,
       "step": 54200
     },
     {
-      "epoch": 5.838983962974922,
-      "grad_norm": 0.7169755101203918,
-      "learning_rate": 0.00025006572567611246,
-      "loss": 3.3305,
+      "epoch": 5.849056603773585,
+      "grad_norm": 0.6645802855491638,
+      "learning_rate": 0.0002494614139233675,
+      "loss": 3.3401,
       "step": 54250
     },
     {
-      "epoch": 5.84436551501453,
-      "grad_norm": 0.7075217366218567,
-      "learning_rate": 0.00024974248464605106,
-      "loss": 3.34,
+      "epoch": 5.8544474393531,
+      "grad_norm": 0.7286192178726196,
+      "learning_rate": 0.00024913761467889906,
+      "loss": 3.3503,
       "step": 54300
     },
     {
-      "epoch": 5.849747067054138,
-      "grad_norm": 0.688876748085022,
-      "learning_rate": 0.00024941924361598965,
-      "loss": 3.3261,
+      "epoch": 5.859838274932614,
+      "grad_norm": 0.7371515035629272,
+      "learning_rate": 0.0002488138154344306,
+      "loss": 3.3539,
       "step": 54350
     },
     {
-      "epoch": 5.855128619093747,
-      "grad_norm": 0.728689968585968,
-      "learning_rate": 0.00024909600258592825,
-      "loss": 3.3239,
+      "epoch": 5.8652291105121295,
+      "grad_norm": 0.6525185704231262,
+      "learning_rate": 0.0002484900161899622,
+      "loss": 3.3482,
       "step": 54400
     },
     {
-      "epoch": 5.860510171133355,
-      "grad_norm": 0.7593002915382385,
-      "learning_rate": 0.0002487727615558668,
-      "loss": 3.3464,
+      "epoch": 5.870619946091644,
+      "grad_norm": 0.6968188285827637,
+      "learning_rate": 0.00024816621694549376,
+      "loss": 3.3362,
       "step": 54450
     },
     {
-      "epoch": 5.865891723172963,
-      "grad_norm": 0.7504733204841614,
-      "learning_rate": 0.0002484495205258054,
-      "loss": 3.3405,
+      "epoch": 5.876010781671159,
+      "grad_norm": 0.7565712928771973,
+      "learning_rate": 0.00024784241770102537,
+      "loss": 3.3489,
       "step": 54500
     },
     {
-      "epoch": 5.871273275212571,
-      "grad_norm": 0.6948011517524719,
-      "learning_rate": 0.000248126279495744,
-      "loss": 3.3271,
+      "epoch": 5.881401617250674,
+      "grad_norm": 0.7541555762290955,
+      "learning_rate": 0.0002475186184565569,
+      "loss": 3.3484,
       "step": 54550
     },
     {
-      "epoch": 5.87665482725218,
-      "grad_norm": 0.7349565625190735,
-      "learning_rate": 0.00024780303846568257,
-      "loss": 3.3311,
+      "epoch": 5.886792452830189,
+      "grad_norm": 0.7156844139099121,
+      "learning_rate": 0.00024719481921208847,
+      "loss": 3.3483,
       "step": 54600
     },
     {
-      "epoch": 5.882036379291788,
-      "grad_norm": 0.7617678642272949,
-      "learning_rate": 0.0002474797974356211,
-      "loss": 3.3301,
+      "epoch": 5.892183288409703,
+      "grad_norm": 0.7071260213851929,
+      "learning_rate": 0.0002468710199676201,
+      "loss": 3.3543,
       "step": 54650
     },
     {
-      "epoch": 5.887417931331396,
-      "grad_norm": 0.7161452174186707,
-      "learning_rate": 0.00024715655640555976,
-      "loss": 3.3385,
+      "epoch": 5.8975741239892185,
+      "grad_norm": 0.7143062949180603,
+      "learning_rate": 0.0002465472207231516,
+      "loss": 3.3421,
       "step": 54700
     },
     {
-      "epoch": 5.892799483371004,
-      "grad_norm": 0.7551568746566772,
-      "learning_rate": 0.0002468333153754983,
-      "loss": 3.3271,
+      "epoch": 5.902964959568733,
+      "grad_norm": 0.754609227180481,
+      "learning_rate": 0.0002462234214786832,
+      "loss": 3.3439,
       "step": 54750
     },
     {
-      "epoch": 5.898181035410612,
-      "grad_norm": 0.686907172203064,
-      "learning_rate": 0.0002465100743454369,
-      "loss": 3.3208,
+      "epoch": 5.908355795148248,
+      "grad_norm": 0.794658899307251,
+      "learning_rate": 0.00024590609821910413,
+      "loss": 3.3406,
       "step": 54800
     },
     {
-      "epoch": 5.903562587450221,
-      "grad_norm": 0.7247705459594727,
-      "learning_rate": 0.0002461868333153755,
-      "loss": 3.3356,
+      "epoch": 5.913746630727763,
+      "grad_norm": 0.6644478440284729,
+      "learning_rate": 0.0002455822989746357,
+      "loss": 3.3427,
       "step": 54850
     },
     {
-      "epoch": 5.9089441394898286,
-      "grad_norm": 0.7339025139808655,
-      "learning_rate": 0.0002458635922853141,
-      "loss": 3.3208,
+      "epoch": 5.919137466307277,
+      "grad_norm": 0.7333596348762512,
+      "learning_rate": 0.0002452584997301673,
+      "loss": 3.3447,
       "step": 54900
     },
     {
-      "epoch": 5.914325691529437,
-      "grad_norm": 0.7308744192123413,
-      "learning_rate": 0.0002455403512552526,
-      "loss": 3.3289,
+      "epoch": 5.9245283018867925,
+      "grad_norm": 0.8418944478034973,
+      "learning_rate": 0.00024493470048569883,
+      "loss": 3.351,
       "step": 54950
     },
     {
-      "epoch": 5.919707243569046,
-      "grad_norm": 0.7355237007141113,
-      "learning_rate": 0.0002452171102251912,
-      "loss": 3.3361,
+      "epoch": 5.929919137466308,
+      "grad_norm": 0.699027419090271,
+      "learning_rate": 0.00024461090124123044,
+      "loss": 3.358,
       "step": 55000
     },
     {
-      "epoch": 5.919707243569046,
-      "eval_accuracy": 0.3839904967838202,
-      "eval_loss": 3.3800570964813232,
-      "eval_runtime": 183.8526,
-      "eval_samples_per_second": 97.964,
-      "eval_steps_per_second": 6.124,
+      "epoch": 5.929919137466308,
+      "eval_accuracy": 0.38380622148608684,
+      "eval_loss": 3.386404037475586,
+      "eval_runtime": 153.4051,
+      "eval_samples_per_second": 117.408,
+      "eval_steps_per_second": 7.34,
       "step": 55000
     },
     {
-      "epoch": 5.925088795608653,
-      "grad_norm": 0.7621373534202576,
-      "learning_rate": 0.0002448938691951298,
-      "loss": 3.3415,
+      "epoch": 5.935309973045822,
+      "grad_norm": 0.7251960039138794,
+      "learning_rate": 0.000244287101996762,
+      "loss": 3.3634,
       "step": 55050
     },
     {
-      "epoch": 5.930470347648262,
-      "grad_norm": 0.7478846311569214,
-      "learning_rate": 0.0002445706281650684,
-      "loss": 3.3356,
+      "epoch": 5.940700808625337,
+      "grad_norm": 0.7481929659843445,
+      "learning_rate": 0.00024396330275229354,
+      "loss": 3.3521,
       "step": 55100
     },
     {
-      "epoch": 5.93585189968787,
-      "grad_norm": 0.7545806765556335,
-      "learning_rate": 0.000244247387135007,
-      "loss": 3.321,
+      "epoch": 5.946091644204852,
+      "grad_norm": 0.7272557616233826,
+      "learning_rate": 0.00024363950350782512,
+      "loss": 3.348,
       "step": 55150
     },
     {
-      "epoch": 5.941233451727478,
-      "grad_norm": 0.7420586347579956,
-      "learning_rate": 0.00024392414610494557,
-      "loss": 3.3258,
+      "epoch": 5.951482479784366,
+      "grad_norm": 0.7121164202690125,
+      "learning_rate": 0.0002433157042633567,
+      "loss": 3.3567,
       "step": 55200
     },
     {
-      "epoch": 5.946615003767087,
-      "grad_norm": 0.7375722527503967,
-      "learning_rate": 0.00024360090507488414,
-      "loss": 3.3479,
+      "epoch": 5.9568733153638815,
+      "grad_norm": 0.7712081074714661,
+      "learning_rate": 0.00024299190501888827,
+      "loss": 3.3611,
       "step": 55250
     },
     {
-      "epoch": 5.951996555806694,
-      "grad_norm": 0.7203102111816406,
-      "learning_rate": 0.00024327766404482273,
-      "loss": 3.3365,
+      "epoch": 5.962264150943396,
+      "grad_norm": 0.7049492597579956,
+      "learning_rate": 0.00024266810577441985,
+      "loss": 3.3282,
       "step": 55300
     },
     {
-      "epoch": 5.957378107846303,
-      "grad_norm": 0.7424737215042114,
-      "learning_rate": 0.0002429544230147613,
-      "loss": 3.3271,
+      "epoch": 5.967654986522911,
+      "grad_norm": 0.7910096049308777,
+      "learning_rate": 0.00024234430652995143,
+      "loss": 3.3379,
       "step": 55350
     },
     {
-      "epoch": 5.962759659885911,
-      "grad_norm": 0.7237606644630432,
-      "learning_rate": 0.00024263118198469992,
-      "loss": 3.3408,
+      "epoch": 5.973045822102426,
+      "grad_norm": 0.6977636814117432,
+      "learning_rate": 0.00024202050728548298,
+      "loss": 3.3394,
       "step": 55400
     },
     {
-      "epoch": 5.968141211925519,
-      "grad_norm": 0.7651504874229431,
-      "learning_rate": 0.0002423079409546385,
-      "loss": 3.3406,
+      "epoch": 5.97843665768194,
+      "grad_norm": 0.7519221305847168,
+      "learning_rate": 0.00024169670804101456,
+      "loss": 3.3323,
       "step": 55450
     },
     {
-      "epoch": 5.973522763965128,
-      "grad_norm": 0.7724640965461731,
-      "learning_rate": 0.00024198469992457706,
-      "loss": 3.3367,
+      "epoch": 5.9838274932614555,
+      "grad_norm": 0.6827930808067322,
+      "learning_rate": 0.0002413729087965461,
+      "loss": 3.3518,
       "step": 55500
     },
     {
-      "epoch": 5.978904316004736,
-      "grad_norm": 0.760128915309906,
-      "learning_rate": 0.00024166145889451568,
-      "loss": 3.3295,
+      "epoch": 5.989218328840971,
+      "grad_norm": 0.7058985233306885,
+      "learning_rate": 0.00024104910955207768,
+      "loss": 3.3455,
       "step": 55550
     },
     {
-      "epoch": 5.984285868044344,
-      "grad_norm": 0.7127272486686707,
-      "learning_rate": 0.00024133821786445425,
-      "loss": 3.3253,
+      "epoch": 5.994609164420485,
+      "grad_norm": 0.733955442905426,
+      "learning_rate": 0.00024072531030760926,
+      "loss": 3.3257,
       "step": 55600
     },
     {
-      "epoch": 5.989667420083952,
-      "grad_norm": 0.7185876965522766,
-      "learning_rate": 0.0002410149768343928,
-      "loss": 3.3302,
+      "epoch": 6.0,
+      "grad_norm": 1.6491297483444214,
+      "learning_rate": 0.00024040151106314084,
+      "loss": 3.3396,
       "step": 55650
     },
     {
-      "epoch": 5.995048972123561,
-      "grad_norm": 0.738452672958374,
-      "learning_rate": 0.0002406917358043314,
-      "loss": 3.3362,
+      "epoch": 6.005390835579515,
+      "grad_norm": 0.7548044919967651,
+      "learning_rate": 0.0002400777118186724,
+      "loss": 3.2445,
       "step": 55700
     },
     {
-      "epoch": 6.000430524163169,
-      "grad_norm": 0.7108159065246582,
-      "learning_rate": 0.00024036849477427,
-      "loss": 3.3403,
+      "epoch": 6.010781671159029,
+      "grad_norm": 0.6957346200942993,
+      "learning_rate": 0.00023975391257420397,
+      "loss": 3.2518,
       "step": 55750
     },
     {
-      "epoch": 6.005812076202777,
-      "grad_norm": 0.807044506072998,
-      "learning_rate": 0.00024004525374420857,
-      "loss": 3.2482,
+      "epoch": 6.0161725067385445,
+      "grad_norm": 0.7332606911659241,
+      "learning_rate": 0.00023943011332973555,
+      "loss": 3.2655,
       "step": 55800
     },
     {
-      "epoch": 6.011193628242385,
-      "grad_norm": 0.7339485883712769,
-      "learning_rate": 0.00023972201271414716,
-      "loss": 3.2305,
+      "epoch": 6.02156334231806,
+      "grad_norm": 0.7256237864494324,
+      "learning_rate": 0.00023910631408526712,
+      "loss": 3.2536,
       "step": 55850
     },
     {
-      "epoch": 6.016575180281993,
-      "grad_norm": 0.7454125881195068,
-      "learning_rate": 0.00023939877168408573,
-      "loss": 3.2307,
+      "epoch": 6.026954177897574,
+      "grad_norm": 0.7203710675239563,
+      "learning_rate": 0.0002387825148407987,
+      "loss": 3.2648,
       "step": 55900
     },
     {
-      "epoch": 6.021956732321602,
-      "grad_norm": 0.7582858800888062,
-      "learning_rate": 0.00023907553065402433,
-      "loss": 3.2511,
+      "epoch": 6.032345013477089,
+      "grad_norm": 0.7059783339500427,
+      "learning_rate": 0.00023845871559633025,
+      "loss": 3.2589,
       "step": 55950
     },
     {
-      "epoch": 6.0273382843612096,
-      "grad_norm": 0.7501933574676514,
-      "learning_rate": 0.00023875228962396292,
-      "loss": 3.24,
+      "epoch": 6.037735849056604,
+      "grad_norm": 0.7440136671066284,
+      "learning_rate": 0.0002381349163518618,
+      "loss": 3.2566,
       "step": 56000
     },
     {
-      "epoch": 6.0273382843612096,
-      "eval_accuracy": 0.3841858546761106,
-      "eval_loss": 3.383694887161255,
-      "eval_runtime": 183.6814,
-      "eval_samples_per_second": 98.056,
-      "eval_steps_per_second": 6.13,
+      "epoch": 6.037735849056604,
+      "eval_accuracy": 0.38372147223359154,
+      "eval_loss": 3.388826608657837,
+      "eval_runtime": 153.1333,
+      "eval_samples_per_second": 117.616,
+      "eval_steps_per_second": 7.353,
       "step": 56000
     },
     {
-      "epoch": 6.032719836400818,
-      "grad_norm": 0.7172769904136658,
-      "learning_rate": 0.0002384290485939015,
-      "loss": 3.2432,
+      "epoch": 6.0431266846361185,
+      "grad_norm": 0.7758867740631104,
+      "learning_rate": 0.00023781111710739338,
+      "loss": 3.2568,
       "step": 56050
     },
     {
-      "epoch": 6.038101388440427,
-      "grad_norm": 0.7134289145469666,
-      "learning_rate": 0.00023810580756384006,
-      "loss": 3.2445,
+      "epoch": 6.048517520215634,
+      "grad_norm": 0.6822553277015686,
+      "learning_rate": 0.00023749379384781433,
+      "loss": 3.2771,
       "step": 56100
     },
     {
-      "epoch": 6.043482940480034,
-      "grad_norm": 0.712856650352478,
-      "learning_rate": 0.00023778256653377868,
-      "loss": 3.2649,
+      "epoch": 6.053908355795148,
+      "grad_norm": 0.6748834252357483,
+      "learning_rate": 0.0002371699946033459,
+      "loss": 3.2533,
       "step": 56150
     },
     {
-      "epoch": 6.048864492519643,
-      "grad_norm": 0.7082148194313049,
-      "learning_rate": 0.00023745932550371725,
-      "loss": 3.2509,
+      "epoch": 6.059299191374663,
+      "grad_norm": 0.7004907131195068,
+      "learning_rate": 0.0002368461953588775,
+      "loss": 3.2603,
       "step": 56200
     },
     {
-      "epoch": 6.0542460445592505,
-      "grad_norm": 0.7556239366531372,
-      "learning_rate": 0.00023713608447365584,
-      "loss": 3.2433,
+      "epoch": 6.064690026954178,
+      "grad_norm": 0.7573196291923523,
+      "learning_rate": 0.00023652239611440904,
+      "loss": 3.2784,
       "step": 56250
     },
     {
-      "epoch": 6.059627596598859,
-      "grad_norm": 0.6958671808242798,
-      "learning_rate": 0.00023681284344359444,
-      "loss": 3.2532,
+      "epoch": 6.070080862533692,
+      "grad_norm": 0.7477830052375793,
+      "learning_rate": 0.00023619859686994062,
+      "loss": 3.2598,
       "step": 56300
     },
     {
-      "epoch": 6.065009148638468,
-      "grad_norm": 0.7431511282920837,
-      "learning_rate": 0.000236489602413533,
-      "loss": 3.2572,
+      "epoch": 6.0754716981132075,
+      "grad_norm": 0.7486846446990967,
+      "learning_rate": 0.00023587479762547217,
+      "loss": 3.2725,
       "step": 56350
     },
     {
-      "epoch": 6.070390700678075,
-      "grad_norm": 0.7136642336845398,
-      "learning_rate": 0.0002361663613834716,
-      "loss": 3.2564,
+      "epoch": 6.080862533692723,
+      "grad_norm": 0.6958188414573669,
+      "learning_rate": 0.00023555099838100374,
+      "loss": 3.2749,
       "step": 56400
     },
     {
-      "epoch": 6.075772252717684,
-      "grad_norm": 0.7659745216369629,
-      "learning_rate": 0.00023584312035341017,
-      "loss": 3.2662,
+      "epoch": 6.086253369272237,
+      "grad_norm": 0.7112399935722351,
+      "learning_rate": 0.00023522719913653532,
+      "loss": 3.2648,
       "step": 56450
     },
     {
-      "epoch": 6.081153804757292,
-      "grad_norm": 0.7569080591201782,
-      "learning_rate": 0.00023551987932334876,
-      "loss": 3.2515,
+      "epoch": 6.091644204851752,
+      "grad_norm": 0.7368870377540588,
+      "learning_rate": 0.0002349033998920669,
+      "loss": 3.2749,
       "step": 56500
     },
     {
-      "epoch": 6.0865353567969,
-      "grad_norm": 0.8000427484512329,
-      "learning_rate": 0.00023519663829328735,
-      "loss": 3.2482,
+      "epoch": 6.097035040431267,
+      "grad_norm": 0.7018360495567322,
+      "learning_rate": 0.00023457960064759848,
+      "loss": 3.294,
       "step": 56550
     },
     {
-      "epoch": 6.091916908836509,
-      "grad_norm": 0.7419618964195251,
-      "learning_rate": 0.00023487339726322592,
-      "loss": 3.2459,
+      "epoch": 6.1024258760107815,
+      "grad_norm": 0.7728846669197083,
+      "learning_rate": 0.00023425580140313005,
+      "loss": 3.292,
       "step": 56600
     },
     {
-      "epoch": 6.097298460876116,
-      "grad_norm": 0.7625570893287659,
-      "learning_rate": 0.0002345501562331645,
-      "loss": 3.2512,
+      "epoch": 6.107816711590297,
+      "grad_norm": 0.7805960774421692,
+      "learning_rate": 0.00023393200215866163,
+      "loss": 3.2664,
       "step": 56650
     },
     {
-      "epoch": 6.102680012915725,
-      "grad_norm": 0.7880632281303406,
-      "learning_rate": 0.0002342269152031031,
-      "loss": 3.2357,
+      "epoch": 6.113207547169812,
+      "grad_norm": 0.7405673265457153,
+      "learning_rate": 0.00023360820291419316,
+      "loss": 3.2861,
       "step": 56700
     },
     {
-      "epoch": 6.108061564955333,
-      "grad_norm": 0.690700352191925,
-      "learning_rate": 0.00023390367417304168,
-      "loss": 3.2478,
+      "epoch": 6.118598382749326,
+      "grad_norm": 0.6923520565032959,
+      "learning_rate": 0.00023328440366972473,
+      "loss": 3.2744,
       "step": 56750
     },
     {
-      "epoch": 6.113443116994941,
-      "grad_norm": 0.8001773953437805,
-      "learning_rate": 0.00023358043314298025,
-      "loss": 3.2671,
+      "epoch": 6.123989218328841,
+      "grad_norm": 0.7368645668029785,
+      "learning_rate": 0.0002329606044252563,
+      "loss": 3.2656,
       "step": 56800
     },
     {
-      "epoch": 6.11882466903455,
-      "grad_norm": 0.7541936039924622,
-      "learning_rate": 0.00023325719211291887,
-      "loss": 3.2729,
+      "epoch": 6.129380053908355,
+      "grad_norm": 0.7301667928695679,
+      "learning_rate": 0.0002326368051807879,
+      "loss": 3.2659,
       "step": 56850
     },
     {
-      "epoch": 6.124206221074158,
-      "grad_norm": 0.7895681262016296,
-      "learning_rate": 0.00023293395108285744,
-      "loss": 3.2802,
+      "epoch": 6.1347708894878705,
+      "grad_norm": 0.7414416670799255,
+      "learning_rate": 0.00023231300593631947,
+      "loss": 3.2686,
       "step": 56900
     },
     {
-      "epoch": 6.129587773113766,
-      "grad_norm": 0.6663275361061096,
-      "learning_rate": 0.000232610710052796,
-      "loss": 3.2491,
+      "epoch": 6.140161725067386,
+      "grad_norm": 0.7112758159637451,
+      "learning_rate": 0.00023198920669185104,
+      "loss": 3.2797,
       "step": 56950
     },
     {
-      "epoch": 6.134969325153374,
-      "grad_norm": 0.7453574538230896,
-      "learning_rate": 0.0002322874690227346,
-      "loss": 3.2642,
+      "epoch": 6.1455525606469,
+      "grad_norm": 0.7612385749816895,
+      "learning_rate": 0.00023166540744738262,
+      "loss": 3.2756,
       "step": 57000
     },
     {
-      "epoch": 6.134969325153374,
-      "eval_accuracy": 0.38437719241155194,
-      "eval_loss": 3.382831335067749,
-      "eval_runtime": 183.8296,
-      "eval_samples_per_second": 97.977,
-      "eval_steps_per_second": 6.125,
+      "epoch": 6.1455525606469,
+      "eval_accuracy": 0.3838165435104292,
+      "eval_loss": 3.3862874507904053,
+      "eval_runtime": 153.3399,
+      "eval_samples_per_second": 117.458,
+      "eval_steps_per_second": 7.343,
       "step": 57000
     },
     {
-      "epoch": 6.140350877192983,
-      "grad_norm": 0.8501230478286743,
-      "learning_rate": 0.0002319642279926732,
-      "loss": 3.2478,
+      "epoch": 6.150943396226415,
+      "grad_norm": 0.7069201469421387,
+      "learning_rate": 0.0002313416082029142,
+      "loss": 3.2859,
       "step": 57050
     },
     {
-      "epoch": 6.1457324292325906,
-      "grad_norm": 0.762480616569519,
-      "learning_rate": 0.00023164098696261176,
-      "loss": 3.2614,
+      "epoch": 6.15633423180593,
+      "grad_norm": 0.761911153793335,
+      "learning_rate": 0.00023101780895844572,
+      "loss": 3.2684,
       "step": 57100
     },
     {
-      "epoch": 6.151113981272199,
-      "grad_norm": 0.7873736619949341,
-      "learning_rate": 0.00023131774593255036,
-      "loss": 3.2723,
+      "epoch": 6.1617250673854445,
+      "grad_norm": 0.7780871391296387,
+      "learning_rate": 0.0002306940097139773,
+      "loss": 3.2768,
       "step": 57150
     },
     {
-      "epoch": 6.156495533311807,
-      "grad_norm": 0.7489500045776367,
-      "learning_rate": 0.00023099450490248892,
-      "loss": 3.2794,
+      "epoch": 6.16711590296496,
+      "grad_norm": 0.7419090270996094,
+      "learning_rate": 0.00023037021046950888,
+      "loss": 3.2715,
       "step": 57200
     },
     {
-      "epoch": 6.161877085351415,
-      "grad_norm": 0.7635547518730164,
-      "learning_rate": 0.00023067126387242754,
-      "loss": 3.2536,
+      "epoch": 6.172506738544475,
+      "grad_norm": 0.7420386075973511,
+      "learning_rate": 0.00023004641122504046,
+      "loss": 3.2741,
       "step": 57250
     },
     {
-      "epoch": 6.167258637391024,
-      "grad_norm": 0.717904806137085,
-      "learning_rate": 0.0002303480228423661,
-      "loss": 3.2496,
+      "epoch": 6.177897574123989,
+      "grad_norm": 0.7601462006568909,
+      "learning_rate": 0.00022972261198057203,
+      "loss": 3.2751,
       "step": 57300
     },
     {
-      "epoch": 6.1726401894306315,
-      "grad_norm": 0.7651631832122803,
-      "learning_rate": 0.00023002478181230468,
-      "loss": 3.262,
+      "epoch": 6.183288409703504,
+      "grad_norm": 0.7667286992073059,
+      "learning_rate": 0.0002293988127361036,
+      "loss": 3.283,
       "step": 57350
     },
     {
-      "epoch": 6.17802174147024,
-      "grad_norm": 0.7424883246421814,
-      "learning_rate": 0.00022970154078224327,
-      "loss": 3.2717,
+      "epoch": 6.188679245283019,
+      "grad_norm": 0.7277544736862183,
+      "learning_rate": 0.00022907501349163516,
+      "loss": 3.2872,
       "step": 57400
     },
     {
-      "epoch": 6.183403293509849,
-      "grad_norm": 0.8002263903617859,
-      "learning_rate": 0.00022937829975218187,
-      "loss": 3.2373,
+      "epoch": 6.1940700808625335,
+      "grad_norm": 0.7943758368492126,
+      "learning_rate": 0.00022875121424716674,
+      "loss": 3.3029,
       "step": 57450
     },
     {
-      "epoch": 6.188784845549456,
-      "grad_norm": 0.7660368084907532,
-      "learning_rate": 0.00022905505872212044,
-      "loss": 3.2678,
+      "epoch": 6.199460916442049,
+      "grad_norm": 0.7654854655265808,
+      "learning_rate": 0.00022842741500269832,
+      "loss": 3.2552,
       "step": 57500
     },
     {
-      "epoch": 6.194166397589065,
-      "grad_norm": 0.7510610222816467,
-      "learning_rate": 0.00022873828251266024,
-      "loss": 3.2635,
+      "epoch": 6.204851752021563,
+      "grad_norm": 0.7684722542762756,
+      "learning_rate": 0.00022810361575822987,
+      "loss": 3.3059,
       "step": 57550
     },
     {
-      "epoch": 6.1995479496286725,
-      "grad_norm": 0.7793570160865784,
-      "learning_rate": 0.00022841504148259884,
-      "loss": 3.2656,
+      "epoch": 6.210242587601078,
+      "grad_norm": 0.743293285369873,
+      "learning_rate": 0.00022777981651376145,
+      "loss": 3.2768,
       "step": 57600
     },
     {
-      "epoch": 6.204929501668281,
-      "grad_norm": 0.7874563336372375,
-      "learning_rate": 0.00022809180045253743,
-      "loss": 3.2774,
+      "epoch": 6.215633423180593,
+      "grad_norm": 0.7013061046600342,
+      "learning_rate": 0.00022745601726929302,
+      "loss": 3.2797,
       "step": 57650
     },
     {
-      "epoch": 6.21031105370789,
-      "grad_norm": 0.716946005821228,
-      "learning_rate": 0.000227768559422476,
-      "loss": 3.2682,
+      "epoch": 6.2210242587601075,
+      "grad_norm": 0.7203818559646606,
+      "learning_rate": 0.00022713221802482457,
+      "loss": 3.2986,
       "step": 57700
     },
     {
-      "epoch": 6.215692605747497,
-      "grad_norm": 0.8085541725158691,
-      "learning_rate": 0.00022744531839241457,
-      "loss": 3.2489,
+      "epoch": 6.226415094339623,
+      "grad_norm": 0.7142075300216675,
+      "learning_rate": 0.00022680841878035615,
+      "loss": 3.2848,
       "step": 57750
     },
     {
-      "epoch": 6.221074157787106,
-      "grad_norm": 0.726571261882782,
-      "learning_rate": 0.0002271220773623532,
-      "loss": 3.2653,
+      "epoch": 6.231805929919138,
+      "grad_norm": 0.7651163339614868,
+      "learning_rate": 0.00022648461953588773,
+      "loss": 3.2898,
       "step": 57800
     },
     {
-      "epoch": 6.226455709826714,
-      "grad_norm": 0.7799760103225708,
-      "learning_rate": 0.00022680530115289297,
-      "loss": 3.2521,
+      "epoch": 6.237196765498652,
+      "grad_norm": 0.7103593349456787,
+      "learning_rate": 0.0002261608202914193,
+      "loss": 3.2896,
       "step": 57850
     },
     {
-      "epoch": 6.231837261866322,
-      "grad_norm": 0.7260156869888306,
-      "learning_rate": 0.0002264820601228316,
-      "loss": 3.2614,
+      "epoch": 6.242587601078167,
+      "grad_norm": 0.7322688102722168,
+      "learning_rate": 0.00022583702104695088,
+      "loss": 3.2891,
       "step": 57900
     },
     {
-      "epoch": 6.237218813905931,
-      "grad_norm": 0.8837427496910095,
-      "learning_rate": 0.00022615881909277016,
-      "loss": 3.2624,
+      "epoch": 6.247978436657682,
+      "grad_norm": 0.7510743141174316,
+      "learning_rate": 0.00022551322180248246,
+      "loss": 3.2734,
       "step": 57950
     },
     {
-      "epoch": 6.242600365945538,
-      "grad_norm": 0.7557101845741272,
-      "learning_rate": 0.00022583557806270875,
-      "loss": 3.2945,
+      "epoch": 6.2533692722371965,
+      "grad_norm": 0.7027929425239563,
+      "learning_rate": 0.00022518942255801399,
+      "loss": 3.2921,
       "step": 58000
     },
     {
-      "epoch": 6.242600365945538,
-      "eval_accuracy": 0.38504931917557234,
-      "eval_loss": 3.3780667781829834,
-      "eval_runtime": 183.8217,
-      "eval_samples_per_second": 97.981,
-      "eval_steps_per_second": 6.126,
+      "epoch": 6.2533692722371965,
+      "eval_accuracy": 0.3843459003798614,
+      "eval_loss": 3.3857808113098145,
+      "eval_runtime": 153.2962,
+      "eval_samples_per_second": 117.491,
+      "eval_steps_per_second": 7.345,
       "step": 58000
     },
     {
-      "epoch": 6.247981917985147,
-      "grad_norm": 0.7460571527481079,
-      "learning_rate": 0.00022551233703264732,
-      "loss": 3.2675,
+      "epoch": 6.258760107816712,
+      "grad_norm": 0.7400386929512024,
+      "learning_rate": 0.00022486562331354556,
+      "loss": 3.2814,
       "step": 58050
     },
     {
-      "epoch": 6.253363470024755,
-      "grad_norm": 0.7291666865348816,
-      "learning_rate": 0.0002251890960025859,
-      "loss": 3.2653,
+      "epoch": 6.264150943396227,
+      "grad_norm": 0.7593062520027161,
+      "learning_rate": 0.00022454182406907714,
+      "loss": 3.2893,
       "step": 58100
     },
     {
-      "epoch": 6.258745022064363,
-      "grad_norm": 0.7323653101921082,
-      "learning_rate": 0.0002248658549725245,
-      "loss": 3.2855,
+      "epoch": 6.269541778975741,
+      "grad_norm": 0.6829662919044495,
+      "learning_rate": 0.00022421802482460872,
+      "loss": 3.2827,
       "step": 58150
     },
     {
-      "epoch": 6.264126574103972,
-      "grad_norm": 0.7868167757987976,
-      "learning_rate": 0.00022454261394246308,
-      "loss": 3.2578,
+      "epoch": 6.274932614555256,
+      "grad_norm": 0.7365722060203552,
+      "learning_rate": 0.0002238942255801403,
+      "loss": 3.2806,
       "step": 58200
     },
     {
-      "epoch": 6.26950812614358,
-      "grad_norm": 0.7138116359710693,
-      "learning_rate": 0.00022421937291240164,
-      "loss": 3.2721,
+      "epoch": 6.280323450134771,
+      "grad_norm": 0.7851418852806091,
+      "learning_rate": 0.00022357042633567187,
+      "loss": 3.2877,
       "step": 58250
     },
     {
-      "epoch": 6.274889678183188,
-      "grad_norm": 0.7772981524467468,
-      "learning_rate": 0.00022389613188234027,
-      "loss": 3.2814,
+      "epoch": 6.285714285714286,
+      "grad_norm": 0.7480865120887756,
+      "learning_rate": 0.00022324662709120345,
+      "loss": 3.2732,
       "step": 58300
     },
     {
-      "epoch": 6.280271230222796,
-      "grad_norm": 0.7344262003898621,
-      "learning_rate": 0.00022357289085227883,
-      "loss": 3.2816,
+      "epoch": 6.291105121293801,
+      "grad_norm": 0.7707905173301697,
+      "learning_rate": 0.00022292930383162435,
+      "loss": 3.3035,
       "step": 58350
     },
     {
-      "epoch": 6.285652782262405,
-      "grad_norm": 0.7527721524238586,
-      "learning_rate": 0.0002232496498222174,
-      "loss": 3.2623,
+      "epoch": 6.296495956873315,
+      "grad_norm": 0.7680032253265381,
+      "learning_rate": 0.00022260550458715593,
+      "loss": 3.2821,
       "step": 58400
     },
     {
-      "epoch": 6.2910343343020125,
-      "grad_norm": 0.7556814551353455,
-      "learning_rate": 0.000222926408792156,
-      "loss": 3.2787,
+      "epoch": 6.30188679245283,
+      "grad_norm": 0.7762503027915955,
+      "learning_rate": 0.0002222817053426875,
+      "loss": 3.2847,
       "step": 58450
     },
     {
-      "epoch": 6.296415886341621,
-      "grad_norm": 0.7294765710830688,
-      "learning_rate": 0.0002226031677620946,
-      "loss": 3.2802,
+      "epoch": 6.307277628032345,
+      "grad_norm": 0.804146409034729,
+      "learning_rate": 0.00022195790609821908,
+      "loss": 3.2738,
       "step": 58500
     },
     {
-      "epoch": 6.301797438381229,
-      "grad_norm": 0.769719123840332,
-      "learning_rate": 0.00022227992673203316,
-      "loss": 3.2752,
+      "epoch": 6.3126684636118595,
+      "grad_norm": 0.7564888000488281,
+      "learning_rate": 0.00022163410685375066,
+      "loss": 3.2868,
       "step": 58550
     },
     {
-      "epoch": 6.307178990420837,
-      "grad_norm": 0.7557373046875,
-      "learning_rate": 0.00022195668570197175,
-      "loss": 3.2653,
+      "epoch": 6.318059299191375,
+      "grad_norm": 0.724154531955719,
+      "learning_rate": 0.00022131030760928224,
+      "loss": 3.302,
       "step": 58600
     },
     {
-      "epoch": 6.312560542460446,
-      "grad_norm": 0.7294456362724304,
-      "learning_rate": 0.00022163344467191032,
-      "loss": 3.258,
+      "epoch": 6.32345013477089,
+      "grad_norm": 0.713821291923523,
+      "learning_rate": 0.00022098650836481382,
+      "loss": 3.2788,
       "step": 58650
     },
     {
-      "epoch": 6.3179420945000535,
-      "grad_norm": 0.7464698553085327,
-      "learning_rate": 0.00022131020364184891,
-      "loss": 3.28,
+      "epoch": 6.328840970350404,
+      "grad_norm": 0.7591390013694763,
+      "learning_rate": 0.0002206627091203454,
+      "loss": 3.2936,
       "step": 58700
     },
     {
-      "epoch": 6.323323646539662,
-      "grad_norm": 0.7696951031684875,
-      "learning_rate": 0.0002209869626117875,
-      "loss": 3.2793,
+      "epoch": 6.334231805929919,
+      "grad_norm": 0.7434984445571899,
+      "learning_rate": 0.00022033890987587692,
+      "loss": 3.2973,
       "step": 58750
     },
     {
-      "epoch": 6.328705198579271,
-      "grad_norm": 0.7079215049743652,
-      "learning_rate": 0.00022066372158172608,
-      "loss": 3.2689,
+      "epoch": 6.339622641509434,
+      "grad_norm": 0.7890250086784363,
+      "learning_rate": 0.0002200151106314085,
+      "loss": 3.2954,
       "step": 58800
     },
     {
-      "epoch": 6.334086750618878,
-      "grad_norm": 0.7504850029945374,
-      "learning_rate": 0.00022034048055166464,
-      "loss": 3.2808,
+      "epoch": 6.345013477088949,
+      "grad_norm": 0.7133108377456665,
+      "learning_rate": 0.00021969131138694007,
+      "loss": 3.2804,
       "step": 58850
     },
     {
-      "epoch": 6.339468302658487,
-      "grad_norm": 0.7906299233436584,
-      "learning_rate": 0.00022001723952160327,
-      "loss": 3.2807,
+      "epoch": 6.350404312668464,
+      "grad_norm": 0.7732018232345581,
+      "learning_rate": 0.00021936751214247165,
+      "loss": 3.2827,
       "step": 58900
     },
     {
-      "epoch": 6.344849854698095,
-      "grad_norm": 0.7144827842712402,
-      "learning_rate": 0.00021969399849154183,
-      "loss": 3.2866,
+      "epoch": 6.355795148247978,
+      "grad_norm": 0.7213762998580933,
+      "learning_rate": 0.00021904371289800323,
+      "loss": 3.2841,
       "step": 58950
     },
     {
-      "epoch": 6.350231406737703,
-      "grad_norm": 0.7381623983383179,
-      "learning_rate": 0.00021937075746148043,
-      "loss": 3.2761,
+      "epoch": 6.361185983827493,
+      "grad_norm": 0.7363037467002869,
+      "learning_rate": 0.0002187199136535348,
+      "loss": 3.2778,
       "step": 59000
     },
     {
-      "epoch": 6.350231406737703,
-      "eval_accuracy": 0.3850140069870326,
-      "eval_loss": 3.3768343925476074,
-      "eval_runtime": 183.8969,
-      "eval_samples_per_second": 97.941,
-      "eval_steps_per_second": 6.123,
+      "epoch": 6.361185983827493,
+      "eval_accuracy": 0.3847574775189026,
+      "eval_loss": 3.379918336868286,
+      "eval_runtime": 153.4501,
+      "eval_samples_per_second": 117.374,
+      "eval_steps_per_second": 7.338,
       "step": 59000
     },
     {
-      "epoch": 6.355612958777312,
-      "grad_norm": 0.7797145843505859,
-      "learning_rate": 0.00021904751643141902,
-      "loss": 3.2798,
+      "epoch": 6.366576819407008,
+      "grad_norm": 0.7245897054672241,
+      "learning_rate": 0.00021839611440906638,
+      "loss": 3.2853,
       "step": 59050
     },
     {
-      "epoch": 6.360994510816919,
-      "grad_norm": 0.7724855542182922,
-      "learning_rate": 0.0002187242754013576,
-      "loss": 3.2848,
+      "epoch": 6.3719676549865225,
+      "grad_norm": 0.7317908406257629,
+      "learning_rate": 0.00021807231516459793,
+      "loss": 3.2821,
       "step": 59100
     },
     {
-      "epoch": 6.366376062856528,
-      "grad_norm": 0.7848725318908691,
-      "learning_rate": 0.00021840103437129619,
-      "loss": 3.2671,
+      "epoch": 6.377358490566038,
+      "grad_norm": 0.7486169934272766,
+      "learning_rate": 0.00021774851592012948,
+      "loss": 3.2919,
       "step": 59150
     },
     {
-      "epoch": 6.371757614896136,
-      "grad_norm": 0.714504599571228,
-      "learning_rate": 0.00021807779334123475,
-      "loss": 3.2908,
+      "epoch": 6.382749326145553,
+      "grad_norm": 0.7614594101905823,
+      "learning_rate": 0.00021742471667566106,
+      "loss": 3.2956,
       "step": 59200
     },
     {
-      "epoch": 6.377139166935744,
-      "grad_norm": 0.7425883412361145,
-      "learning_rate": 0.00021775455231117335,
-      "loss": 3.2784,
+      "epoch": 6.388140161725067,
+      "grad_norm": 0.8346837162971497,
+      "learning_rate": 0.00021710091743119264,
+      "loss": 3.2876,
       "step": 59250
     },
     {
-      "epoch": 6.382520718975353,
-      "grad_norm": 0.7488074898719788,
-      "learning_rate": 0.00021743131128111194,
-      "loss": 3.2756,
+      "epoch": 6.393530997304582,
+      "grad_norm": 0.7456716299057007,
+      "learning_rate": 0.00021677711818672422,
+      "loss": 3.3071,
       "step": 59300
     },
     {
-      "epoch": 6.387902271014961,
-      "grad_norm": 0.7575358152389526,
-      "learning_rate": 0.0002171080702510505,
-      "loss": 3.2867,
+      "epoch": 6.398921832884097,
+      "grad_norm": 0.7255779504776001,
+      "learning_rate": 0.0002164533189422558,
+      "loss": 3.2834,
       "step": 59350
     },
     {
-      "epoch": 6.393283823054569,
-      "grad_norm": 0.7843925356864929,
-      "learning_rate": 0.00021678482922098908,
-      "loss": 3.2735,
+      "epoch": 6.404312668463612,
+      "grad_norm": 0.6992196440696716,
+      "learning_rate": 0.00021612951969778734,
+      "loss": 3.2824,
       "step": 59400
     },
     {
-      "epoch": 6.398665375094177,
-      "grad_norm": 0.7430239915847778,
-      "learning_rate": 0.0002164615881909277,
-      "loss": 3.2835,
+      "epoch": 6.409703504043127,
+      "grad_norm": 0.7387642860412598,
+      "learning_rate": 0.00021580572045331892,
+      "loss": 3.2958,
       "step": 59450
     },
     {
-      "epoch": 6.404046927133785,
-      "grad_norm": 0.7865639328956604,
-      "learning_rate": 0.00021613834716086627,
-      "loss": 3.2894,
+      "epoch": 6.415094339622642,
+      "grad_norm": 0.7684344053268433,
+      "learning_rate": 0.0002154819212088505,
+      "loss": 3.2861,
       "step": 59500
     },
     {
-      "epoch": 6.4094284791733935,
-      "grad_norm": 0.7577503323554993,
-      "learning_rate": 0.00021581510613080483,
-      "loss": 3.2722,
+      "epoch": 6.420485175202156,
+      "grad_norm": 0.7793638110160828,
+      "learning_rate": 0.00021515812196438208,
+      "loss": 3.276,
       "step": 59550
     },
     {
-      "epoch": 6.414810031213002,
-      "grad_norm": 0.7468580603599548,
-      "learning_rate": 0.00021549186510074346,
-      "loss": 3.2759,
+      "epoch": 6.425876010781671,
+      "grad_norm": 0.761384129524231,
+      "learning_rate": 0.00021483432271991363,
+      "loss": 3.2853,
       "step": 59600
     },
     {
-      "epoch": 6.42019158325261,
-      "grad_norm": 0.783909261226654,
-      "learning_rate": 0.00021516862407068202,
-      "loss": 3.2793,
+      "epoch": 6.431266846361186,
+      "grad_norm": 0.7257922887802124,
+      "learning_rate": 0.0002145105234754452,
+      "loss": 3.2927,
       "step": 59650
     },
     {
-      "epoch": 6.425573135292218,
-      "grad_norm": 0.7352930307388306,
-      "learning_rate": 0.0002148453830406206,
-      "loss": 3.2804,
+      "epoch": 6.436657681940701,
+      "grad_norm": 0.7598720192909241,
+      "learning_rate": 0.00021418672423097676,
+      "loss": 3.2849,
       "step": 59700
     },
     {
-      "epoch": 6.430954687331827,
-      "grad_norm": 0.797244131565094,
-      "learning_rate": 0.00021452214201055919,
-      "loss": 3.2838,
+      "epoch": 6.442048517520216,
+      "grad_norm": 0.7282331585884094,
+      "learning_rate": 0.0002138694009713977,
+      "loss": 3.2975,
       "step": 59750
     },
     {
-      "epoch": 6.4363362393714345,
-      "grad_norm": 0.7799109816551208,
-      "learning_rate": 0.00021419890098049778,
-      "loss": 3.2784,
+      "epoch": 6.44743935309973,
+      "grad_norm": 0.7626290917396545,
+      "learning_rate": 0.00021354560172692929,
+      "loss": 3.2865,
       "step": 59800
     },
     {
-      "epoch": 6.441717791411043,
-      "grad_norm": 0.81449294090271,
-      "learning_rate": 0.00021387565995043638,
-      "loss": 3.2781,
+      "epoch": 6.452830188679245,
+      "grad_norm": 0.7948201298713684,
+      "learning_rate": 0.00021322180248246086,
+      "loss": 3.295,
       "step": 59850
     },
     {
-      "epoch": 6.447099343450651,
-      "grad_norm": 0.7860665917396545,
-      "learning_rate": 0.00021355241892037494,
-      "loss": 3.2764,
+      "epoch": 6.45822102425876,
+      "grad_norm": 0.758529543876648,
+      "learning_rate": 0.00021289800323799241,
+      "loss": 3.2821,
       "step": 59900
     },
     {
-      "epoch": 6.452480895490259,
-      "grad_norm": 0.7320060729980469,
-      "learning_rate": 0.0002132291778903135,
-      "loss": 3.2871,
+      "epoch": 6.463611859838275,
+      "grad_norm": 0.7200835943222046,
+      "learning_rate": 0.000212574203993524,
+      "loss": 3.298,
       "step": 59950
     },
     {
-      "epoch": 6.457862447529868,
-      "grad_norm": 0.7849714756011963,
-      "learning_rate": 0.00021290593686025213,
-      "loss": 3.2841,
+      "epoch": 6.46900269541779,
+      "grad_norm": 0.7930318713188171,
+      "learning_rate": 0.00021225040474905557,
+      "loss": 3.2944,
       "step": 60000
     },
     {
-      "epoch": 6.457862447529868,
-      "eval_accuracy": 0.3855985595234745,
-      "eval_loss": 3.3706517219543457,
-      "eval_runtime": 183.5345,
-      "eval_samples_per_second": 98.134,
-      "eval_steps_per_second": 6.135,
+      "epoch": 6.46900269541779,
+      "eval_accuracy": 0.3847094529424886,
+      "eval_loss": 3.3771378993988037,
+      "eval_runtime": 153.4052,
+      "eval_samples_per_second": 117.408,
+      "eval_steps_per_second": 7.34,
       "step": 60000
     }
   ],
   "logging_steps": 50,
-  "max_steps": 92910,
+  "max_steps": 92750,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 10,
   "save_steps": 10000,
@@ -8966,7 +8966,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 5.01650914148352e+17,
+  "total_flos": 5.016415076352e+17,
   "train_batch_size": 32,
   "trial_name": null,
   "trial_params": null